diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1suw4pGn-qQ.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1suw4pGn-qQ.pt new file mode 100644 index 0000000000000000000000000000000000000000..cdf285e5404f827c84ffee4a4bc539bc8eb72d7c --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1suw4pGn-qQ.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:403a59b86f5bca0b19e56e001452f56c7be5897a79dc453555e0905ec8fc5a01 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1swY162Ccfo.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1swY162Ccfo.pt new file mode 100644 index 0000000000000000000000000000000000000000..5bc41a318e17b7ad9eea189a0151bb939feb4bb5 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1swY162Ccfo.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5db7a905456561d0f0996948d7aee49f8adeee01b74db65e64495dc05a1a240d +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1t7G8T4V-sA.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1t7G8T4V-sA.pt new file mode 100644 index 0000000000000000000000000000000000000000..c3bfac8d5c40f3550299fff2950696dc6736b6ae --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1t7G8T4V-sA.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:775be9e628935275cf0877cc3924cac7e0733b0564ea514b8bfeffe1c4635b07 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1tE-9m6T0zo.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1tE-9m6T0zo.pt new file mode 100644 index 0000000000000000000000000000000000000000..8547a99060eeeafdf4bd40c6ceb18125b356ad8c --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1tE-9m6T0zo.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d3fe0f69f6921f14b60d5fd6e1e160197b9a17aee951b681afa938db731aaa16 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1tGI3j_TENw.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1tGI3j_TENw.pt new file mode 100644 index 0000000000000000000000000000000000000000..f534cf0772189ce118b83950734cd86d5990fbaf --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1tGI3j_TENw.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f58f9d0d2638c5ddee5e997d862338c590cad4767ba4ad09a5475cc2e61ded93 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1tRbDjpQktE.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1tRbDjpQktE.pt new file mode 100644 index 0000000000000000000000000000000000000000..d22640bd6ae619c285ff6fb5f809265b04deae4a --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1tRbDjpQktE.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:123d0af09da17a862d89d56ed840a9d769f8ac0427226c525953e785cfd3ab25 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1tTYu2eQnjk.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1tTYu2eQnjk.pt new file mode 100644 index 0000000000000000000000000000000000000000..5dc86595a27d49e01f70a441f6d9b9ddf2f074a6 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1tTYu2eQnjk.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:752574b81d76e11df8afb3f360ced50c40cf2ced7e87aef1462b761049d4abf6 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1tW0Gm6fuXM.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1tW0Gm6fuXM.pt new file mode 100644 index 0000000000000000000000000000000000000000..214389bde8ef9bdacc6c8b28b796c02c883a1ffd --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1tW0Gm6fuXM.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:57d6074b8c4a5e3dbe3369ad07acf6e1af2af27e105b5d404c305e2a9c19b112 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1t_EjzYSTPM.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1t_EjzYSTPM.pt new file mode 100644 index 0000000000000000000000000000000000000000..365ae7e1c4d23ce9e5226ff69e19f7fbb8d52f6b --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1t_EjzYSTPM.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:131f4dc56585acd3b869c2e98beaf7589f0e40b909600cc4dadfb3ad018b9478 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1t_orCH_v64.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1t_orCH_v64.pt new file mode 100644 index 0000000000000000000000000000000000000000..f35fd7a2e0a9b62c51cce849d6b5a9229fab7f52 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1t_orCH_v64.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a038bc25d421820ce2d6dc57138d0b6dd970cc4af2313c580ea5b0e441ce53c3 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1tdaxCqqiPQ.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1tdaxCqqiPQ.pt new file mode 100644 index 0000000000000000000000000000000000000000..a5c3cf9639c1e2e33c27e81661627feb7c4e839c --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1tdaxCqqiPQ.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b54a3ff8a8649896c66bc59065c9daea21a8f5d1b8436f6d8e253f539330f465 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1thEVGY3Fk8.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1thEVGY3Fk8.pt new file mode 100644 index 0000000000000000000000000000000000000000..c7eec4ac5cca0fd319eb24b7d1a10b670e4fbd3c --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1thEVGY3Fk8.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:823a5286d4f2d37756cbd27561244f9dd929353a5b4acd2cd798b4abb2042d1d +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1tiT7X3eyAY.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1tiT7X3eyAY.pt new file mode 100644 index 0000000000000000000000000000000000000000..cf912aac4eb2e5372ad63a682157864ec059f7b1 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1tiT7X3eyAY.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6a007c3793f7919cd27ddfa5d5637cb71f8a26b7110d6301160c0126fbd41d8e +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1tp-3HH0650.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1tp-3HH0650.pt new file mode 100644 index 0000000000000000000000000000000000000000..0336d68f12ba92be56b8dfd3f23ee457c1cb0bfc --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1tp-3HH0650.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4f2b1009708c3793f634f138fa12c7d0897b7098cd148aeb621cab9c66910519 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1tqcotHXJE4.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1tqcotHXJE4.pt new file mode 100644 index 0000000000000000000000000000000000000000..718cba1cd47fde3abf0002acb7f92994fa1677d7 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1tqcotHXJE4.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c417f591bf65692e9529c9269260d5fc792f185858be337945fac26afcf0c558 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1twTSBTEt3A.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1twTSBTEt3A.pt new file mode 100644 index 0000000000000000000000000000000000000000..f65e57164480cfe409fc934c62140c95598a565b --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1twTSBTEt3A.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:20fe5aa5f9e928282b3a86ff36f1a1c8dbe87d788ef22b1ac4a4a48441440541 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1u0wLxUaXLU.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1u0wLxUaXLU.pt new file mode 100644 index 0000000000000000000000000000000000000000..8915a80bfc85f5cf51e7718860a5de83a6dec8f3 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1u0wLxUaXLU.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b884f5760bb692dbdb3af14f6df655e60f624727a73e4e6c77b2143978cd994f +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1u1dvbC9Hv0.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1u1dvbC9Hv0.pt new file mode 100644 index 0000000000000000000000000000000000000000..e37dc689cbe7cff03ee75f6d581e1767f4c81324 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1u1dvbC9Hv0.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0ddd311a80a04f8386a2226d219f43ade78b79fdcee5def517b84473e5d8e48e +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1u32q87JJhw.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1u32q87JJhw.pt new file mode 100644 index 0000000000000000000000000000000000000000..ce4b1356cdb62d2dc5e11a2f55b3718fb59efe36 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1u32q87JJhw.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:590678ba7462e389f229b8ff9e0dfbbdeb271c7bf6e773900220497a6ceeba29 +size 17041736 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1uFMu6cMkdI.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1uFMu6cMkdI.pt new file mode 100644 index 0000000000000000000000000000000000000000..dfdf2dd0c3369730b6552c2a807f0b1de248cda1 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1uFMu6cMkdI.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d23d35169b01aebe865b6e13e70525d099c952314748c8867b70e7d1f2f19d99 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1uPYxxReSOA.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1uPYxxReSOA.pt new file mode 100644 index 0000000000000000000000000000000000000000..74ce12810a32eb0e6a81a82b8e4186c3fa24f72d --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1uPYxxReSOA.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a88df3bb8b220d2304ca11998015f6d2a15f3be0d41f67941dca123fc75d5c44 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1uSO5axfbPo.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1uSO5axfbPo.pt new file mode 100644 index 0000000000000000000000000000000000000000..3629468612276922b8e0a78a50737b1449c7bedf --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1uSO5axfbPo.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ca0b2f46ea3c127236e9999a3e9aa7e07bb10dec935cc1f4e629a3f609355d57 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1uU-yCOlBGg.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1uU-yCOlBGg.pt new file mode 100644 index 0000000000000000000000000000000000000000..dad1099f7b42d8f458a905f168374a3b2511a9d5 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1uU-yCOlBGg.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ad9a39e85fcff85dd731682d19524faee6790444eb3a9a7e079c4c967d6b11e6 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1ubOwsFkldM.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1ubOwsFkldM.pt new file mode 100644 index 0000000000000000000000000000000000000000..e8bd9247a35f5a4b5a7f09b31bad1499d0384016 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1ubOwsFkldM.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:33a01a8be4b0f34a4b7fc7b13d470ae817af8ade031229209869738c4c84e650 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1ueKtORsR9o.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1ueKtORsR9o.pt new file mode 100644 index 0000000000000000000000000000000000000000..ce165914d00d87b3e9052eed4e23dbb2ec96481c --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1ueKtORsR9o.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a10e407e81de5b9f6749099cd578eadcb390b07fa6bdf4f3a21297740ec132df +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1ukwH-dWMzg.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1ukwH-dWMzg.pt new file mode 100644 index 0000000000000000000000000000000000000000..e4826ec66720806abe83a9b39649f2087b0460c5 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1ukwH-dWMzg.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:11dfa29bff358e9ba1524ba94c7388d96d2179ce502463c6361aacd016681c0e +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1ur8-Cgjmns.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1ur8-Cgjmns.pt new file mode 100644 index 0000000000000000000000000000000000000000..1ac5bc3cf5e5feb8c24a4d2a9abdd39078204e72 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1ur8-Cgjmns.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d12933be065537b26d208b92e254b4a9a0d2c5befdac95f14f0ee09cd86f7de0 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1uzG8d78HPs.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1uzG8d78HPs.pt new file mode 100644 index 0000000000000000000000000000000000000000..3102c3ec1a1a239530f79844a9f0d8fb2ebd75ac --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1uzG8d78HPs.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:86046390e02b3dbb4c2b406d12c7d3f093071659c024dca4b44ffeed023e71b2 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1v4w3RPHNkw.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1v4w3RPHNkw.pt new file mode 100644 index 0000000000000000000000000000000000000000..a80278182fb1024054968c2d15a8e358eeeea4fa --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1v4w3RPHNkw.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1670ef2953ce333d79de831d69ec4c3d98844ae6e3668af1ef9a5ff0e68df187 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1v8BHC2iHeQ.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1v8BHC2iHeQ.pt new file mode 100644 index 0000000000000000000000000000000000000000..86d93fb101d2f50533dce3e3ce1ae14caf639f64 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1v8BHC2iHeQ.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:45a5ead0ee22b26f1ad1fbdcbab56ca4697747ed00ba6a14c00f5feb004fe1ba +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1vAr2N5GBUI.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1vAr2N5GBUI.pt new file mode 100644 index 0000000000000000000000000000000000000000..9b58614eb471b6495dc326dc84a402b498f7e5f9 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1vAr2N5GBUI.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f701112b13628a9054cd7737a6b315a92d1cfbcce4663d0e8b7aaf2b40d94d9d +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1vBnMKG4iOE.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1vBnMKG4iOE.pt new file mode 100644 index 0000000000000000000000000000000000000000..d09d7715c8dbd9a62abc7a337fc2b547b234fd4b --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1vBnMKG4iOE.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c6582d220ec9ccbff8014b1ee7ddb4974e49cca3c2c66b4448de5525de1aac18 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1vK2l8a8vqI.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1vK2l8a8vqI.pt new file mode 100644 index 0000000000000000000000000000000000000000..bc7ce2e75832e214efe2f4f498b5b6887deddf2e --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1vK2l8a8vqI.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c5005f0c73e50f1af3f1870763aa460d68132af95908b17522ab5430bcdde14b +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1vOTPD_CRjo.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1vOTPD_CRjo.pt new file mode 100644 index 0000000000000000000000000000000000000000..12edf9dd983cf49caf5adffdb62d976deb591a47 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1vOTPD_CRjo.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:efc4481b4d19701728cdfe5a1ac7c3fc78fe0ed0ea836a179ff7e10cce9fdf96 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1vOy30Hpgl0.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1vOy30Hpgl0.pt new file mode 100644 index 0000000000000000000000000000000000000000..7453bfe98ce3ae9455fa3b3e09290b5656f83100 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1vOy30Hpgl0.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2ff361ce3b2712feef5e9f27105c5391f1599753bc9648b63dac7be9e55b215c +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1vQ5B_7OpeA.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1vQ5B_7OpeA.pt new file mode 100644 index 0000000000000000000000000000000000000000..2c9a14d97287d53d3ebbc510b520dc3ec855fbe3 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1vQ5B_7OpeA.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:94da6afef9fa0c9439cb70c3a04fc0f24787acae56fc00526c47cab63c9336bf +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1vYQkUya38w.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1vYQkUya38w.pt new file mode 100644 index 0000000000000000000000000000000000000000..a95eff72503b2f5bb19d90df2ecf6a469b90865b --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1vYQkUya38w.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e8ed97f76e957322a7c8616c3b94209ea174a3470cfec0efbc6adfe8f4bc5997 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1vaHRGfSOeo.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1vaHRGfSOeo.pt new file mode 100644 index 0000000000000000000000000000000000000000..d4b0d9c3dd6f1e5d6638b04d66e62ec550e70c48 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1vaHRGfSOeo.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:63aedee807738137d58fc09009b91f3c28dba0628e80c1277688460740859d9e +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1vbB8jAzvU8.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1vbB8jAzvU8.pt new file mode 100644 index 0000000000000000000000000000000000000000..296d54d70ae910508dd5b377d906ce4521ede870 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1vbB8jAzvU8.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:96306911cb4d2238ac1ac839e7c3633cbad4239485ff22c04d29557f661e941e +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1vck7NnASb0.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1vck7NnASb0.pt new file mode 100644 index 0000000000000000000000000000000000000000..8540e7e9c02b01654fb574c1a21a5b4bf189caf5 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1vck7NnASb0.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:51adbef5a41388fc154d80b6e053e59e33348f4f65c9ffaa052d9658611b6b26 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1vgIed-R_gA.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1vgIed-R_gA.pt new file mode 100644 index 0000000000000000000000000000000000000000..9ba5d117b1ec20f2c3626a8f20984b6124be3e0d --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1vgIed-R_gA.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ddcd240e16a401e90b9f2fff845e4a16f64a24b55df53b3d618fcd743c4e0f7f +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1vk9b5Ef1GE.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1vk9b5Ef1GE.pt new file mode 100644 index 0000000000000000000000000000000000000000..5aa1632a695a6947fe647f4d10ddd09473d04541 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1vk9b5Ef1GE.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5239333bc6429a7064c98f256fd666673fdd3d2c2e4e5469e6a4d190ad29e779 +size 17303880 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1vwuwRQMYMA.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1vwuwRQMYMA.pt new file mode 100644 index 0000000000000000000000000000000000000000..b31383d8fc2d502868bb4062c692b0d8a7aa14f1 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1vwuwRQMYMA.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f176c23a647132a5c532737f84d05936dee903a3615c88c50216d2017438a359 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1w8zHaktw7M.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1w8zHaktw7M.pt new file mode 100644 index 0000000000000000000000000000000000000000..e819562bcc206855158f0e112e0ceac26705757f --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1w8zHaktw7M.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3ed5ff3ee31d136ebe099fd5ff6adb4825485159532f3baebc5f62bde170255f +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1wfHPkAjopA.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1wfHPkAjopA.pt new file mode 100644 index 0000000000000000000000000000000000000000..b38c4cca36cf8e061427fa10cffe823aaee22ec0 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1wfHPkAjopA.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4741eae2fd3504bd0005015fdc6b07e62a9bf5643e9b8e26e2f67b7956d68e14 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1wg8s3zjFjQ.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1wg8s3zjFjQ.pt new file mode 100644 index 0000000000000000000000000000000000000000..056bd60d202ad5bddcc45462db72a50256edc2cb --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1wg8s3zjFjQ.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ed1fb42ca556edb2fbb8d6f15224552234547332552c7a183c69fb9bddd883b5 +size 19663176 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1whCBBKHvu0.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1whCBBKHvu0.pt new file mode 100644 index 0000000000000000000000000000000000000000..055fb716abda452136c5422013005cb990d013f2 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1whCBBKHvu0.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4eaa75da49ae98758d33346aedbbe058dfa6cb1f7544dc680e2321317b6c18b2 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1wk23U6wBvQ.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1wk23U6wBvQ.pt new file mode 100644 index 0000000000000000000000000000000000000000..e03ddc4ced37ba72ae2b239589cedaa78a65e7a6 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1wk23U6wBvQ.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:93c20d1d8d17c7585a707d9479a5648c3cd90f25dec76bfc508e41c3e9891d04 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1wmay-wOigs.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1wmay-wOigs.pt new file mode 100644 index 0000000000000000000000000000000000000000..2cc2083e45901fd50fcf0e8d8fe9aacd2c7b3125 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1wmay-wOigs.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ce3d3976baa98e1311d7228a7c160bedd91d8a9688d1e810822865b4c4090f0d +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1wszv9Oj_Ao.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1wszv9Oj_Ao.pt new file mode 100644 index 0000000000000000000000000000000000000000..70975cf4ca702c75fb35f9408954f4b3924a8da8 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1wszv9Oj_Ao.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:52837c60ec82febab4013b98c271d3d470f56c012513ac664e9be21d2039bd68 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1wv0LE2BRPE.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1wv0LE2BRPE.pt new file mode 100644 index 0000000000000000000000000000000000000000..a76cd93a7b5044fdf0c0acc570eea0c82930387c --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1wv0LE2BRPE.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f05ce67afd8ae5316c6df62c32430cc6504e4c656bd7ae5d37a2e4bb5e8e63d2 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1wyB1r5vkjM.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1wyB1r5vkjM.pt new file mode 100644 index 0000000000000000000000000000000000000000..5fbc814dfaa8c43f0fec6b5e87a5598a69faf79a --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1wyB1r5vkjM.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f1aec79f0cf74a6798b5c1e7f65ec11d685d022c2cc024cbd85c388fcee9bd3f +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1wzX83kZs2I.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1wzX83kZs2I.pt new file mode 100644 index 0000000000000000000000000000000000000000..053531c502daffb4199cb476d893d535dedcef5f --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1wzX83kZs2I.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:683159cc9a961edb543466d9d8765d686dea36fa44602341c2eea7ab2e080e96 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1x6D91-1m-E.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1x6D91-1m-E.pt new file mode 100644 index 0000000000000000000000000000000000000000..38d7d360911670c08bb0630ad5ed5435cbce9249 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1x6D91-1m-E.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0aed6e0cdf91eb46dca6a3ff0df8394833acb814fef9fd4f8ee55839f30ed693 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1x9oWD29t00.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1x9oWD29t00.pt new file mode 100644 index 0000000000000000000000000000000000000000..805f122b7aa06e84e230fc4d1f4e09d7796edc3c --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1x9oWD29t00.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a1d45e41c28ed615509bfc59aa5b2f7fd767aa2458b124337cb3b661145b7630 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1xAMgcQEhyo.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1xAMgcQEhyo.pt new file mode 100644 index 0000000000000000000000000000000000000000..04cade154825b53757ccb8477d9fcd85bdef8363 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1xAMgcQEhyo.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:db6b9fe934faecd78f7182510eb27ea90bcca7e023efcbba9d4a413fa3698f6b +size 18352456 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1xKWZLgBwH8.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1xKWZLgBwH8.pt new file mode 100644 index 0000000000000000000000000000000000000000..df5ff2608b3e703911521e94bc8fe9afa004e9de --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1xKWZLgBwH8.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f9ebe268921489ffb1c2977356b4cbcff144b4e636a932b34c7de3ff65f13e9a +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1xLM9K8_K2c.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1xLM9K8_K2c.pt new file mode 100644 index 0000000000000000000000000000000000000000..5748fcf337b6104fa694f325d40cf6523c1ec3a6 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1xLM9K8_K2c.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6f17c069d4c19e8282de8427561a91a11717850281b52b35f70b9d73aed0b0c2 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1xPS8Ssx5YA.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1xPS8Ssx5YA.pt new file mode 100644 index 0000000000000000000000000000000000000000..db9854ba5c811b37bc45df59e3de3f22738b8c65 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1xPS8Ssx5YA.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0b2b740a89cd2fe427db60190ee043e5aaefff93c93c2bb751b42abce407b343 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1xZqtT5llT8.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1xZqtT5llT8.pt new file mode 100644 index 0000000000000000000000000000000000000000..1f944bbfd29d6fba86198935eeac4b066a8a96a3 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1xZqtT5llT8.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f339f5b990f41c6c6c35292ce1e0665c7d42be4d99deb0b2a92d5ca6656fddc6 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1xgHT39UQ6k.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1xgHT39UQ6k.pt new file mode 100644 index 0000000000000000000000000000000000000000..78f306e75e7542800c61795bd9fc0b5e9d86bb60 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1xgHT39UQ6k.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f2581d37758597d32363c2e82a01f4eb60ac76b468c2781c94941820bd581c42 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1xglX9nxxVc.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1xglX9nxxVc.pt new file mode 100644 index 0000000000000000000000000000000000000000..a8bae44f96215f45006759a9a69f2a9a8981c4cb --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1xglX9nxxVc.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:de630293d139c72117564b9318eed535473dbcff5df694281c5017583661e31b +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1xh_dcqt6vQ.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1xh_dcqt6vQ.pt new file mode 100644 index 0000000000000000000000000000000000000000..bd52c132d01402670cda30ecc02098bedb5cc4b7 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1xh_dcqt6vQ.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0b13887952f8cc36974fb0376193fabb35c5dfc2e2e9b18d0a637c3652b0d259 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1xko8Hi8zdw.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1xko8Hi8zdw.pt new file mode 100644 index 0000000000000000000000000000000000000000..c12959af83322a4b5a14c635bc0e877997483efa --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1xko8Hi8zdw.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ba60b93ed7b0f613a7b3c9c3b6f1b808ac119d7a3af5b453e6e6104ab196aabc +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1xrr6kBl2I4.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1xrr6kBl2I4.pt new file mode 100644 index 0000000000000000000000000000000000000000..515d09c8628bef3d714cfce2e3c10337486d4f76 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1xrr6kBl2I4.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c6f504e9182977c880047e685e8153f4c02096d99613b5c5b1ed73ae67c09f7b +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1xwNFK2sMpQ.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1xwNFK2sMpQ.pt new file mode 100644 index 0000000000000000000000000000000000000000..80ee0676eabd53eaa8f0cf18ef7d5bab72b9f20f --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1xwNFK2sMpQ.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1c0462bfa704d3fcbe8787d78bab59db162dcb5808e6f8a8811f0c84df3a4ac9 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1yBeic0Mo94.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1yBeic0Mo94.pt new file mode 100644 index 0000000000000000000000000000000000000000..3c6db45170f776c6707f9f951620ae24bd3fbdcd --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1yBeic0Mo94.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f6594704f895b0e0b4760eb4d6bb052ce6808fdc073911a2d00f584ff2b7deb2 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1yCqJku_gB0.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1yCqJku_gB0.pt new file mode 100644 index 0000000000000000000000000000000000000000..d1cdb04ae5f676fc68033e46f0b97803ed32d0b6 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1yCqJku_gB0.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:be687026ef58ce8170c4e128110da200f6b589e4d27b2ef5e5b41b425e812dfb +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1yT0hxplVBg.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1yT0hxplVBg.pt new file mode 100644 index 0000000000000000000000000000000000000000..408abaf7579b2d71a2c9c4211689fd150317b342 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1yT0hxplVBg.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:02e438ed08292705cd7dc0d1daafaddd43d63168c7f7f46d264f7c9c8d0a1dec +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1yZgQgJ9PhI.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1yZgQgJ9PhI.pt new file mode 100644 index 0000000000000000000000000000000000000000..b6a63c381e30630111cc33160ecea28bc04ae551 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1yZgQgJ9PhI.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9f5011953e0334659141d228b33d35f87fbf4567149d3c9b74a0060e0b9cbb5b +size 17041736 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1yesk5z2jgc.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1yesk5z2jgc.pt new file mode 100644 index 0000000000000000000000000000000000000000..5c82b7a92619eeac730136dbfce8f0ad3a19fc62 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1yesk5z2jgc.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5ffdb6cfc5d58c494007313d4015d65f9d11ee83f9a0737baede348d93fb91f7 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1yhiZ51Ti3g.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1yhiZ51Ti3g.pt new file mode 100644 index 0000000000000000000000000000000000000000..965b6ed9035d9e373f9f33395a48f0900fa58b93 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1yhiZ51Ti3g.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:91ecf9e9d01442041a0a49909a7c9cdb4f59f2003a3d41650fdcba0abb1ca807 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1yloDi-74Hk.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1yloDi-74Hk.pt new file mode 100644 index 0000000000000000000000000000000000000000..2ac5f8803bb1dffd59bcbdbf835e0865863d00e2 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1yloDi-74Hk.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:163a3fda31de95a5c35276f5fa81ce1c23e86611bf9a613517accb805601a790 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1ymCNfec90k.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1ymCNfec90k.pt new file mode 100644 index 0000000000000000000000000000000000000000..9e552952e7b18f9a5785e211a6911f365404a44e --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1ymCNfec90k.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b0a9a3b65941476ecd03711ce455553e419b2594f6bda0b53a90938d122201e5 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1ynKi1m5GIQ.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1ynKi1m5GIQ.pt new file mode 100644 index 0000000000000000000000000000000000000000..72ab4ec557d79ddd54287845e79a8b9ee7ededcf --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1ynKi1m5GIQ.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b1b17fee39cf425ceea4b61913025ea0caea9dd4d08ae4aa3008c01fa54eff5f +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1yq5gqjGQ74.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1yq5gqjGQ74.pt new file mode 100644 index 0000000000000000000000000000000000000000..60a0796215a04b341cfa050d2537780fbf756029 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1yq5gqjGQ74.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1c0321791dfef153a15f920bb3ed1214636f3e0698b2735020058f018a56ebf1 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1yyJiLnOcEQ.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1yyJiLnOcEQ.pt new file mode 100644 index 0000000000000000000000000000000000000000..687d26b80a73c5ad2bcb358ca86ae8a7e9e3e52c --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1yyJiLnOcEQ.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1e6cb6fff5ff529dd51cd9d7bc4abaec07cd3c47f8ba02f5884e77803b638c40 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1yyPoX_p8T4.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1yyPoX_p8T4.pt new file mode 100644 index 0000000000000000000000000000000000000000..8546d6835cf648df9ff6c67b1d5ea840c37cd14d --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1yyPoX_p8T4.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2e8370e2153d8dbf078bef12f2a854373a4d7d1571af8dc4cc83fd8c6ddc92a8 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1z5s5F2Zeu0.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1z5s5F2Zeu0.pt new file mode 100644 index 0000000000000000000000000000000000000000..4f8fb1e811cd699edf4222417097b95ff756bb39 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1z5s5F2Zeu0.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:07010c776adf536d5be44adc53acac82988f303ab3baecefcbead6c9e9fc4965 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1z9qBwll-UM.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1z9qBwll-UM.pt new file mode 100644 index 0000000000000000000000000000000000000000..e87b7372cad3da09c34a93409dfe82eebb1cdd7e --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1z9qBwll-UM.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0eebaa9b73cb4a514979a192ede51dc65d3934ccc7437b1f9410280b2d93f973 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1zGLrqL37Qc.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1zGLrqL37Qc.pt new file mode 100644 index 0000000000000000000000000000000000000000..d4cb7a996ef080cc7cf363d3203d22678a5f6419 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1zGLrqL37Qc.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6a62be3b660e670bcc530858be454fa6f13e4435634d9bc68b1a50bbba16180f +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1zIAm9hDfv4.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1zIAm9hDfv4.pt new file mode 100644 index 0000000000000000000000000000000000000000..4c66b9b68db24d0384dd520809dae43eed53a94f --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1zIAm9hDfv4.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:88a3d1e36e058c67c9bac538dfcde5dc3c9049f70284b332bc3e4ec4e6670b81 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1zMkzxW9xCI.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1zMkzxW9xCI.pt new file mode 100644 index 0000000000000000000000000000000000000000..d7ad85f7646ff7f85e22d2abe6d225d0bbf90ca5 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1zMkzxW9xCI.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6475d77d0dc5ae1bcd82fc9e787a667003c363d02eb2e561189c7d03f044769f +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1zOEK_FEsno.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1zOEK_FEsno.pt new file mode 100644 index 0000000000000000000000000000000000000000..492420000c480938dbb5500abddfc34026e12d4a --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1zOEK_FEsno.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ba669d72e220704afda5143681314ed181b45c85559274af31a4072d17ed6bc0 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1zP4SVpAHfM.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1zP4SVpAHfM.pt new file mode 100644 index 0000000000000000000000000000000000000000..f81053251bd15046c7c296c44ed924c372d89389 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1zP4SVpAHfM.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:466cc1efddf16ef6742bf6410b0610dfd9b3c7fe6d7ade34236874cdd371956e +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1zRLVGCe--I.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1zRLVGCe--I.pt new file mode 100644 index 0000000000000000000000000000000000000000..43ab5bf93e4df339feb66a93b991e73c6cfcf9af --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1zRLVGCe--I.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cacd601e34adf5784adbfa11e3d57fff842a3fe348d4d7d2264934594b6aed7a +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1zTTA2yYXEc.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1zTTA2yYXEc.pt new file mode 100644 index 0000000000000000000000000000000000000000..e1c7fac8a35d4730a4bcc4a7e8956398ccf2de8a --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1zTTA2yYXEc.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fa8a05d1dc9a3a8e146c540deb71b51557a6cb3273c081bac14418ce7bf12fdd +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1zYQADx1Y2E.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1zYQADx1Y2E.pt new file mode 100644 index 0000000000000000000000000000000000000000..14693147e8a5387f1d3d143e3d6df3b8299ab3bf --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1zYQADx1Y2E.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:634cd3bd274d6bf98ad0b1d010e3926de2512632e584a995e0bb6ebbf8932a9f +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1zcqLY8OkmY.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1zcqLY8OkmY.pt new file mode 100644 index 0000000000000000000000000000000000000000..90b7a9409ff6f3d063924af124d8ce16b5987ab3 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1zcqLY8OkmY.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d00013e3f65dfbc62ef6a9bf18e0bd435ba26d2030ba2b8dbc0858667cc521a4 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1zgKJnirLsQ.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1zgKJnirLsQ.pt new file mode 100644 index 0000000000000000000000000000000000000000..4ddd4069691dec98bac626bec9d32d3a507fc606 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1zgKJnirLsQ.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0d58a7b70375a3cc4f873275de051b7f5c4cc8184855f2aaf69517ec320cccca +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1zhT5EJgWJs.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1zhT5EJgWJs.pt new file mode 100644 index 0000000000000000000000000000000000000000..ad473a774e55d3d22fd7d6e209ce013910b28fb3 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1zhT5EJgWJs.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3f49bef4eb95ddac88afd2a090e39599cf46e4914d04c6f5e1a2f8a2dd913739 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1zhUTuxUgXQ.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1zhUTuxUgXQ.pt new file mode 100644 index 0000000000000000000000000000000000000000..005e5f186ac1fb8f88a83eaee84f073193cbfeeb --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1zhUTuxUgXQ.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bf099bacc8354cd6bfa47530d436e8d9eda94c81edab342add26cc1f52ca93a9 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1zjmrbDlslc.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1zjmrbDlslc.pt new file mode 100644 index 0000000000000000000000000000000000000000..7ca0e81a59195702c6d7135f2f87199973acec15 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1zjmrbDlslc.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:054014129330d48bab03d1ef65fe0c6a0efeaa27b50cc0caae26740110b5fc83 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1zmn8jaUAKg.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1zmn8jaUAKg.pt new file mode 100644 index 0000000000000000000000000000000000000000..d79997600f40864f857a91cb21ecf3fd2fa9328d --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1zmn8jaUAKg.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e8a2a75b159179f92a6628c4e3b5471ca4c7edaf5cbfc6f99c4a1982d30415ef +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1zowWCrgh-I.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1zowWCrgh-I.pt new file mode 100644 index 0000000000000000000000000000000000000000..421c0ddf8ae305f5d3d7c5ee879773eea797be62 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1zowWCrgh-I.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a3917a45d2b862eff476492da9085eec97ca6b51e15b3cab8c090140d933a59b +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1ztutQH-nM8.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1ztutQH-nM8.pt new file mode 100644 index 0000000000000000000000000000000000000000..83ce8ec912600a13f1fcc368c17304203ee9ad1c --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_1ztutQH-nM8.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bd27c679f03a7c88551e50b35fbc5f1ce3d24a925aedb8060ba708b77fb3565c +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2-4MbQ_Db24.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2-4MbQ_Db24.pt new file mode 100644 index 0000000000000000000000000000000000000000..b66bcac45357babe02cb25551ace68b07bea5dd5 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2-4MbQ_Db24.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d446c7ffc79baa4a10860f3dd96263771b5a25cbe77f87533d1080a7c44a9558 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2-6-r-Q1BgI.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2-6-r-Q1BgI.pt new file mode 100644 index 0000000000000000000000000000000000000000..b69c60bec75aefcd5015730c3ce2db240085918c --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2-6-r-Q1BgI.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dc26c7c99bd4b6ed51e4dea05ac5a654cbf60866cc46b15475282cfd8016e3e5 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2-69LXQYdCw.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2-69LXQYdCw.pt new file mode 100644 index 0000000000000000000000000000000000000000..42acbc0ad2b8202534a9e6f4825c82cac555013a --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2-69LXQYdCw.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f91fd7c6f248bea96986ec9a500f5be8ad478463c617cfda4a05d8706d9a036d +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2-EGbMf5Fhs.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2-EGbMf5Fhs.pt new file mode 100644 index 0000000000000000000000000000000000000000..6948a00bf694e3893f4d22affb41efcec0b608a6 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2-EGbMf5Fhs.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:85b587249ae7ebfc03881009653c49c0bcb6628296e4205c2fb8b802d88366f9 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2-NEWr7hCzU.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2-NEWr7hCzU.pt new file mode 100644 index 0000000000000000000000000000000000000000..60ad20177aecf19d50f9770abe235e07f08b5af2 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2-NEWr7hCzU.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4ae03e8f0b38aafde80bed211cc711049394ab35bf926fad3c4acfa77a366866 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2-PqZzplIRM.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2-PqZzplIRM.pt new file mode 100644 index 0000000000000000000000000000000000000000..d35629d08bca68faf0ab034e88904920cef226f4 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2-PqZzplIRM.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:82f8ebc622752718c36766adde3c9cd334b31e629a5b1b58616e0b043e6de5a1 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2-UBnEihrRI.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2-UBnEihrRI.pt new file mode 100644 index 0000000000000000000000000000000000000000..93fcb9471afe34a2050a98a3a5201b11ee4ac3e2 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2-UBnEihrRI.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bb61d7b28ca36e8ea527b973a009865e46a8b66293876f2f10b6971a39df5e68 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2-ZTJ3uuKwE.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2-ZTJ3uuKwE.pt new file mode 100644 index 0000000000000000000000000000000000000000..5a5f39b00e08b5fbad3f5a1f6cc7fe6283ecdfa0 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2-ZTJ3uuKwE.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:951623bd851c34232641bbcdd5748bbddb6203cd722b9806b84b6e82da8056a4 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2-ZoZkMrZlk.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2-ZoZkMrZlk.pt new file mode 100644 index 0000000000000000000000000000000000000000..6caea581732dbf32b3cd43ab57baed9f2a7d7380 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2-ZoZkMrZlk.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3c604e1d1a036ce3536ceac90097c8183ba3bc866c7b072e43f76ee119602525 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2-aJWBmp2yY.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2-aJWBmp2yY.pt new file mode 100644 index 0000000000000000000000000000000000000000..6113dd80671bb42eaca7faf79e6cae9c2c4a9f22 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2-aJWBmp2yY.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cb22eca2365f086984d3f643c5660b961e3d6fced129e237513efb8d45840748 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2-cmbposlHQ.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2-cmbposlHQ.pt new file mode 100644 index 0000000000000000000000000000000000000000..415a9fdddd7251fe35cea7fee900bb00d7ce3ab7 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2-cmbposlHQ.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0d86d0f07c1cb5786e526132b4fe07815239fc8980e7d1d7a864dfb6543f9a06 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2-cq9RTxZM0.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2-cq9RTxZM0.pt new file mode 100644 index 0000000000000000000000000000000000000000..9919132fbeece87d4c97b4f6b6689883f6f2ad4d --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2-cq9RTxZM0.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:005952ad920ffcf249cd0b2014363fed859d55dc550b40a1c7d5b4e0eb2a1460 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2-odQ_eltMA.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2-odQ_eltMA.pt new file mode 100644 index 0000000000000000000000000000000000000000..88e2df7a9f72168ecf5eb16e0f533402db8fa23f --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2-odQ_eltMA.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b4774c6acab4bd352a5f54a2ae65ce3c81adc55e5c4dc197453acfba06df0e46 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2-sG-fjGBUo.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2-sG-fjGBUo.pt new file mode 100644 index 0000000000000000000000000000000000000000..f365ab1d81a89cc4d8d61d4e14eda05d23b839dd --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2-sG-fjGBUo.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2314ce2ed63924223537d5dab20cd982c8db6925b26e83ca30f1500619c23562 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2-xudqSB4bU.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2-xudqSB4bU.pt new file mode 100644 index 0000000000000000000000000000000000000000..252c28f67bd31c05abfccb204fee8fda210288af --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2-xudqSB4bU.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:28cce33b9153d67a038f8b5919f37260c9d6d05e53b17168058e6fd12110b5a5 +size 18352456 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2039D4DYL-g.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2039D4DYL-g.pt new file mode 100644 index 0000000000000000000000000000000000000000..ec55ba53f85ac3aa38529aa238188351aeb288bb --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2039D4DYL-g.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c2b21830bf4f40f52ec17bab8545077527ed25ca42aff6e554544ab57378f526 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_205J2GhGYDA.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_205J2GhGYDA.pt new file mode 100644 index 0000000000000000000000000000000000000000..d9254cb3deb2da42fd18eeaa63061b9b93074730 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_205J2GhGYDA.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:249f536b4fdf8d6a08eff4bf7c418fcaedf9b3021bc5d603f0583373f87230c7 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_20975BDe4uA.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_20975BDe4uA.pt new file mode 100644 index 0000000000000000000000000000000000000000..4f28c915cddec7300af1b112b6fd6111cead049d --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_20975BDe4uA.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0492638e32b171de64e26f01821993e4d30ce660f5c0852b7430579abe8073e1 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_20AO8tQkIe8.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_20AO8tQkIe8.pt new file mode 100644 index 0000000000000000000000000000000000000000..82c2a246675a8aaa926ab5413061e304ddd8a3ea --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_20AO8tQkIe8.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:306c46c987b13f03298f597f866e470f70d8e5f2bdfec678cf4fda501727ca2d +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_20GVFFPl0-4.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_20GVFFPl0-4.pt new file mode 100644 index 0000000000000000000000000000000000000000..a4ab8e9988e1e371ab6115d1850e2170ab5583c4 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_20GVFFPl0-4.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8d15930ec0233d49f87fc21bb32a0e1182162378f4c7dc40e0701df522037deb +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_20GyVLit3Bs.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_20GyVLit3Bs.pt new file mode 100644 index 0000000000000000000000000000000000000000..598a5b56b8a26855be42e315f58a7b2f6a9883e3 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_20GyVLit3Bs.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a2a9c69533bcc340e7d0b72fdb469459243e38fb0d938b2788010ea6bebb7f62 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_20JSg_xsYBI.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_20JSg_xsYBI.pt new file mode 100644 index 0000000000000000000000000000000000000000..f2e6282615b1d0bc840c6dff303ab0103b07d5d5 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_20JSg_xsYBI.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f411906dcba07d15168a0776f7a32b85aa330ce7185e6d40abfcbb911359fbe7 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_20KrhzYe2b8.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_20KrhzYe2b8.pt new file mode 100644 index 0000000000000000000000000000000000000000..9f953d43b73bf7096008b685f1c35d4cd58c966b --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_20KrhzYe2b8.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2a78cf2ab1e3f993deb39d85be8dee98b78fd50e55700a7e8107bfc91f8fdb26 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_20PByseyko4.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_20PByseyko4.pt new file mode 100644 index 0000000000000000000000000000000000000000..2f390be03b16bb8937aec447379939a86b93607d --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_20PByseyko4.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1961c0b86b580fe615dfd7807990624f335fbcf8be11433037b83bcd1ae69511 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_20S15L7LHuc.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_20S15L7LHuc.pt new file mode 100644 index 0000000000000000000000000000000000000000..33ac1ff679cb340367e1ec914aa91545f492c1aa --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_20S15L7LHuc.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4e13f7e5c2df276184b42aacff967ba19db1ea43a223cdf46afa5f797dbfac66 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_20SCsRvEFcc.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_20SCsRvEFcc.pt new file mode 100644 index 0000000000000000000000000000000000000000..082591cd8bb21454265e837024e1d08c32f01464 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_20SCsRvEFcc.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5b6e9e55b3d57dc3299ba3a0963bfe65b5cb0286e8d8659385473f6585999af2 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_20WND2-aMTQ.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_20WND2-aMTQ.pt new file mode 100644 index 0000000000000000000000000000000000000000..35323dd64229f34eda23ebaea230c6d54d3b03e4 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_20WND2-aMTQ.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3dc0e72fbbad650d61f17615f758830c1e74a8689789984adef993c10f93b4a4 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_20Y7YoawrVo.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_20Y7YoawrVo.pt new file mode 100644 index 0000000000000000000000000000000000000000..da10829d8af0c3bcba5b5257703a9dce8e5ac0f7 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_20Y7YoawrVo.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8e5ea3520abd23834734d80e27fc7b5d294744cedea18d82c4d04658f1d4d672 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_20k4lBPWI3Y.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_20k4lBPWI3Y.pt new file mode 100644 index 0000000000000000000000000000000000000000..d40691aaee64b26dcbd8da4d16e588366fb39a7b --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_20k4lBPWI3Y.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3d885d46deb315058664adfa854d0281750acff349a60e9a8bbd5c31058a9f54 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_20oUY0vcEYs.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_20oUY0vcEYs.pt new file mode 100644 index 0000000000000000000000000000000000000000..3ede792f8c1deffebcdcaa67bf7845ebbef6c7cf --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_20oUY0vcEYs.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b49adbd715eac4f39318d8d21aad654d6ac61f83caeae470e13019a136edf0b8 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_20tXJCVncuE.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_20tXJCVncuE.pt new file mode 100644 index 0000000000000000000000000000000000000000..fc0455936ad99248e5c9350165d8e94357499ae0 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_20tXJCVncuE.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9832a5a86eaddc17bc33c248ac33e2cd2d02abc8f89af0548a917b8f442d2bca +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_20vOElMs5qc.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_20vOElMs5qc.pt new file mode 100644 index 0000000000000000000000000000000000000000..bea4d562799821368c431eb7e4fc5b2342c2f02b --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_20vOElMs5qc.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:99fa5f99b0ae737009a09512491077797e864060744c6729b799c9edff995c2e +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_20x_uagms1c.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_20x_uagms1c.pt new file mode 100644 index 0000000000000000000000000000000000000000..684ee6feeb751daee4e08f47709d61b75d9ef724 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_20x_uagms1c.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:92b1724be1e17735d2c157e5ec486afad828dd17f59fa4dfbf2733f448581c1e +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_211gsGIkywg.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_211gsGIkywg.pt new file mode 100644 index 0000000000000000000000000000000000000000..e31eef6031f40b8f06642ad67cb4deaae1cdc63d --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_211gsGIkywg.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a352f20f63d366b07624f71051d842adbfb2609dfbdfea8a7ea852761e415d20 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_216vFe2fjqY.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_216vFe2fjqY.pt new file mode 100644 index 0000000000000000000000000000000000000000..2938b6082038bff8aee2ba53e199baea43188059 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_216vFe2fjqY.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4fc2a10d8e225fbf5d63b9c2b4d9d9862f8b3057f9ec63a54bc1ebc8999aa8f4 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2180EATjNWw.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2180EATjNWw.pt new file mode 100644 index 0000000000000000000000000000000000000000..0bc76340ad41c5c4f76594137a7598dc89478477 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2180EATjNWw.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0521687621fa69991388020603c1edab07700459e3c68943080fbb0d2e315401 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_21Cn17I8Nok.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_21Cn17I8Nok.pt new file mode 100644 index 0000000000000000000000000000000000000000..6802ababa69f443235d24009a863a62454bbbe43 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_21Cn17I8Nok.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:61a0264cb4e7f449fc01cc7fd6d8f0ffba4837855743db2adcfa9723403f87dd +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_21G8jgCkBOM.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_21G8jgCkBOM.pt new file mode 100644 index 0000000000000000000000000000000000000000..6c5bac37e36add3ae2814fd31555a1b41dc987fe --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_21G8jgCkBOM.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3f35b9f9b30aef47aa04c207d0e385358da1973cd84756deee964ef681f9879d +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_21JhkWvguBo.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_21JhkWvguBo.pt new file mode 100644 index 0000000000000000000000000000000000000000..f2501aeef53927b6764c9b4f29cb1b032f627ed7 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_21JhkWvguBo.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:856a37d3546ba238e7ee36801aee9757246d0e0da5042705025636b5accdc958 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_21ORdxqANYc.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_21ORdxqANYc.pt new file mode 100644 index 0000000000000000000000000000000000000000..feabc9a8aa7c1d91a2607370779854dc5ee70cbc --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_21ORdxqANYc.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2d2d4d790fc764ed5e61876cc337a01a623cb4dd6b2930b79a72fd9618c142a7 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_21PAwFsAHdg.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_21PAwFsAHdg.pt new file mode 100644 index 0000000000000000000000000000000000000000..17f2f043320e8c0a78f6f3070aee7fd44a04deaf --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_21PAwFsAHdg.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:48e03adf3529d95f36e79224b8100dde0585465e82a49f7265a110adc379d575 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_21VWqkhpRpM.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_21VWqkhpRpM.pt new file mode 100644 index 0000000000000000000000000000000000000000..4799fdaa5cbaa43789b643c900c328b316fbce6a --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_21VWqkhpRpM.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:29479521f3d367d727fb4f28284f9acac780ccef2cb7de8e5860db5289881468 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_21Y-G0gmh20.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_21Y-G0gmh20.pt new file mode 100644 index 0000000000000000000000000000000000000000..2a3fc9d0570e212c06e66897da17eaaac923d1ea --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_21Y-G0gmh20.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f6523c2d6841c4db0ab8a4622f28bdf518925aabaea0ab743ef4027f9d00b6f4 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_21c1PI4Bkvc.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_21c1PI4Bkvc.pt new file mode 100644 index 0000000000000000000000000000000000000000..886d80736b17860e026d79bfb2ca88826dbfb2cb --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_21c1PI4Bkvc.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6f910ef9b8fc36282f38715811bfeb64ebf3ad9da33ec89bc0b9a87590450af2 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_21dIuCipLPU.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_21dIuCipLPU.pt new file mode 100644 index 0000000000000000000000000000000000000000..7e9e273e45c0e438b78ed4980952b3cba6cf5345 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_21dIuCipLPU.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9284835b549995184b3f9b10846ef01c5f21f8674c4d9e4f1a59338b00f10e18 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_21mQoktagqY.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_21mQoktagqY.pt new file mode 100644 index 0000000000000000000000000000000000000000..9480fa9f9e2afe22a6e2e026d9bbf46793155406 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_21mQoktagqY.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ae9a8878b50ca3cc7bf3c1ae14fd49e671504752816be2736a6bc53ce71577d3 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_21pZP7IZCMw.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_21pZP7IZCMw.pt new file mode 100644 index 0000000000000000000000000000000000000000..eb62fe95284b79bc1fce0a815c8114e3c4309b5c --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_21pZP7IZCMw.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cac4a80d7dbc218e9384ddd408d0b41cf4867e15a5370a08ce94e2331f1578f4 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_21tQ_7ySX2s.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_21tQ_7ySX2s.pt new file mode 100644 index 0000000000000000000000000000000000000000..0be027f89d9bfe755cdfb7931ca9dff219ae70e4 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_21tQ_7ySX2s.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:501faae5e70231e0d79be2257da1785d0a1f527517b114caa71fc7ffa5a4ff45 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_21xWwk2jIe8.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_21xWwk2jIe8.pt new file mode 100644 index 0000000000000000000000000000000000000000..6dd8a3e7f9a587edb67745c5afcc962a3cfca009 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_21xWwk2jIe8.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:40ee09866ac9b5ab2790f09e40df30edb8076dc7e2ec95da488c0a9917bdc563 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_21zUq2Aakmw.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_21zUq2Aakmw.pt new file mode 100644 index 0000000000000000000000000000000000000000..c9f7085c84bbfedb72ff593c2540432437b9127d --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_21zUq2Aakmw.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7d7038f90199380a496161581b9695fec20745eef5102db440d219829a429199 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2261QFwLPaI.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2261QFwLPaI.pt new file mode 100644 index 0000000000000000000000000000000000000000..4f3eb01854d3adf82821ee2853047c650115f06b --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2261QFwLPaI.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f945fc0a2c89e2c30cb6c0312d6103116a4349fff82b466be0d36093a030559e +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_229Um1PQI_U.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_229Um1PQI_U.pt new file mode 100644 index 0000000000000000000000000000000000000000..1f513bd405cb8adaac91980e66b8d0ab2ed67165 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_229Um1PQI_U.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0003229e0cae54c176c75459ecc475df2629ac473def8f76626f5f54fffdafc5 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_22B_ZTouQps.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_22B_ZTouQps.pt new file mode 100644 index 0000000000000000000000000000000000000000..c7561651573065802c7b49284c5a2582c1c78e4c --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_22B_ZTouQps.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6ac6ab1b069dd997a3642b0c59ab3edad95c2e87a743acac02a71268301cd41a +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_22CRyC1a8rI.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_22CRyC1a8rI.pt new file mode 100644 index 0000000000000000000000000000000000000000..e6664bd2e6a26e14b853fe31dace14eb12cb84d2 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_22CRyC1a8rI.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9a04a73fe1fc0c6afe83a2565801ed19081032c92cd589e177417062f7906e00 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_22DOFMJgAqg.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_22DOFMJgAqg.pt new file mode 100644 index 0000000000000000000000000000000000000000..c9a00b98aae0b682d2bdd8da3fb8fa649ed7b66f --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_22DOFMJgAqg.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3998fc208b113f85395f00b34f5f137e7ec149ebb1242d65902ace4bc30c259d +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_22E3BNpPouQ.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_22E3BNpPouQ.pt new file mode 100644 index 0000000000000000000000000000000000000000..2f64b22b2edfbdbfe738fa66c66ab6c6a9a4fb42 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_22E3BNpPouQ.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:68d2084e9c7ed70bc8ebbba56cc5859c982210b9e905da4e3691b9d33d531ffa +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_22JC1TUYhUE.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_22JC1TUYhUE.pt new file mode 100644 index 0000000000000000000000000000000000000000..bd3d8d608cf4b0a5955409835968a6f80fda2da9 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_22JC1TUYhUE.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:85de90b8b4bf2932605a890e0761fe0c2dcda95cbfa1258c40ee4ee95b548f43 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_22MFEfkrVNo.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_22MFEfkrVNo.pt new file mode 100644 index 0000000000000000000000000000000000000000..40d4c31eb1bca79b4c9095805d742189d930b159 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_22MFEfkrVNo.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0a0e392eb490a77ccae3221db046bcb773eb8df33cb84f8ec820fb6a2f7264e6 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_22Mdclceq8k.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_22Mdclceq8k.pt new file mode 100644 index 0000000000000000000000000000000000000000..70e12174d56384eb728bf2bc3e5e840b6357abe1 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_22Mdclceq8k.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2d3c69c8f5511c0f7064f78bcaab20f821120e0299eedd4f85dd0bb262a8b609 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_22VcQypwSLQ.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_22VcQypwSLQ.pt new file mode 100644 index 0000000000000000000000000000000000000000..5af09bf2b584a1c372d91d39eb468fba08b1e07d --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_22VcQypwSLQ.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:13d9a49eb55b9fe0b35250e8e74b138cea72d725e7b95e803ecc35604ac8fc7a +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_22d6i9B90bI.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_22d6i9B90bI.pt new file mode 100644 index 0000000000000000000000000000000000000000..d03ea908b98980832f6f1447b501872d4214a7d8 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_22d6i9B90bI.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:96cab0c66b3c48262e2d1f9722c0f4e85762ab432bf24a7829852b7fccb9139e +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_22la-qbtUmQ.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_22la-qbtUmQ.pt new file mode 100644 index 0000000000000000000000000000000000000000..c986f9a28bc58b08004d26d64eedd4d195b3f218 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_22la-qbtUmQ.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3faa11e17895dbdf29a9b3d1c9f90e5a74e11d4c44e7a4b287f794726b802a5b +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_22wvCNY_hYg.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_22wvCNY_hYg.pt new file mode 100644 index 0000000000000000000000000000000000000000..5f97976d548452fd91fa83dec064506165c6ef51 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_22wvCNY_hYg.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3f6c096bb3008fb6b2b7266308bfe2c0a3b24073738408392d70a8288578c23d +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_235KO--Fjko.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_235KO--Fjko.pt new file mode 100644 index 0000000000000000000000000000000000000000..bf430a10ca1a8d04fb3e475825e51d8e0016ec09 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_235KO--Fjko.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9774ac0ab2b8159ca41219ea9e8a64859ed48b3a5896ed4a73f0f94b984517bb +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_237tBybYeSk.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_237tBybYeSk.pt new file mode 100644 index 0000000000000000000000000000000000000000..e6a4989630bcf37c5d58e74d0edb49f92932188a --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_237tBybYeSk.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b23127967ef2f722cc0f1a4d168c6075ac52c4717eaa01d7d0328834075aea85 +size 18352456 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_239g8hf9Qzk.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_239g8hf9Qzk.pt new file mode 100644 index 0000000000000000000000000000000000000000..6d7d9069137f5950e4162a40bed879369408382d --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_239g8hf9Qzk.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a88776d2fd5e5ff7fdb71de61c90cbf2e3883e6b23960a4f9ff6572cda68f0b0 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23A5LFoo3WA.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23A5LFoo3WA.pt new file mode 100644 index 0000000000000000000000000000000000000000..12be3c7a3cb1c06b57e2761d5462cf00b57b828e --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23A5LFoo3WA.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e61574a1cd62e41a8b742f9132b732390583b50f2f5a8c2d5e78957d4804fb84 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23GhV8bEHjo.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23GhV8bEHjo.pt new file mode 100644 index 0000000000000000000000000000000000000000..40f68383fc41952065ff8bab337601a696304ae3 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23GhV8bEHjo.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dc577ac36109bcfac43ffd6a71e0d763c9d109c7616c38beb5b4746c16adf237 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23JzJObSnZY.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23JzJObSnZY.pt new file mode 100644 index 0000000000000000000000000000000000000000..f1b630ace08ef2a81ecc829110d48b585826f1fc --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23JzJObSnZY.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:73e884b4cc8befce7a29b83fa48fad8d3575f72a414431bd6603f6272575fd14 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23KCuNmrtho.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23KCuNmrtho.pt new file mode 100644 index 0000000000000000000000000000000000000000..2cac98089f5b01b11e836ea5b89670e34787a263 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23KCuNmrtho.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:27604e0f5ee9709a56b4f4ae2724e75d670871c506eb7a0417316e2f1fa6b2ee +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23LLJOtDn60.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23LLJOtDn60.pt new file mode 100644 index 0000000000000000000000000000000000000000..2a3297156dbc9cece146f7795b7a606dd85355a3 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23LLJOtDn60.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:931bb68f0a51272edf70c512be9c19ea349720dde39b878988a7b56038328747 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23M7KmevtYs.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23M7KmevtYs.pt new file mode 100644 index 0000000000000000000000000000000000000000..9c7b4442dc32a18e49823af9753845bc32768b23 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23M7KmevtYs.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:050d9efeaf55d3d00977bb968fac728062c1133468c6bd7b985e52371d5b219c +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23Oo2kkAge0.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23Oo2kkAge0.pt new file mode 100644 index 0000000000000000000000000000000000000000..32c46c2b2d404b89d1c2d21e294923151124c30d --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23Oo2kkAge0.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c2f4f80ee1c4646c9b69e4c56a87da562e0a4bb5ff699d751be29c2ffb9f249e +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23RHY86L5Hs.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23RHY86L5Hs.pt new file mode 100644 index 0000000000000000000000000000000000000000..f038a621e0c74ecc9988f6c20e3a63ae5e47c8f5 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23RHY86L5Hs.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d89e4735aa68edcf8b1f61e796a03b227c13c3014b21515c273f6436b17666a9 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23Ruxan0b5w.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23Ruxan0b5w.pt new file mode 100644 index 0000000000000000000000000000000000000000..3674e8788fb5ab5f10b08ed2ea1b55dc4dadcb59 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23Ruxan0b5w.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7bbc94e5ffdd50af3cb99b1ce11fad87f1772e640b881c89e96685599421bd45 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23Yh2ITsVmA.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23Yh2ITsVmA.pt new file mode 100644 index 0000000000000000000000000000000000000000..d572753666840adea69338eff5bf13208b541a24 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23Yh2ITsVmA.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b2fa21565f403ee309f98a60da4aed8a322a7548e92418a63e9f8c2cdcf63ab5 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23ZNlRWFPUs.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23ZNlRWFPUs.pt new file mode 100644 index 0000000000000000000000000000000000000000..5ec780846b5b18c0dba4153207e962ec6e5862f9 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23ZNlRWFPUs.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d18114fd3ed11aa7ecf5aaf4f2b6c759306cad146e7e13500394a1cb182d6e95 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23d5qf7jSLE.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23d5qf7jSLE.pt new file mode 100644 index 0000000000000000000000000000000000000000..11f7edb742c4c425dc8831b121ec2827a9b641c5 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23d5qf7jSLE.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:61de477bfde7f2104d6d03d6eff642b69eedaeef5befcde128ac4cb0ca564b67 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23hbTQLp5E8.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23hbTQLp5E8.pt new file mode 100644 index 0000000000000000000000000000000000000000..e292e5603686f0d6bf83fba619cc6cbb03433263 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23hbTQLp5E8.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c0eb8e199c9d7e19a59b20f3f2a935bcf3e0eb37bca0c972c1d382c2062236da +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23mEJHw68uE.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23mEJHw68uE.pt new file mode 100644 index 0000000000000000000000000000000000000000..b06f24e0a90c22748180e479734a1b5df47b30c0 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23mEJHw68uE.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e842fa30128d94b57132176d70609800ec2ee01aea0a8eaedb8713a2dfeb8fff +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23nJIAPsPkg.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23nJIAPsPkg.pt new file mode 100644 index 0000000000000000000000000000000000000000..2dd59ee36dbc0505392c7cc7e9f75e1d27299522 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23nJIAPsPkg.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f80b350774e1e772a1d266a3ec435c4c9505cf0bca701e763ca8efa7524f62ca +size 18352456 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23nSsBlh7IQ.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23nSsBlh7IQ.pt new file mode 100644 index 0000000000000000000000000000000000000000..fbf3d671caf546795f158549dd73972cd0bcd47b --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23nSsBlh7IQ.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e5765220b30c24dbdc1c10c67c97dade4ec15783bc7c4f75348a0fbdd8ad29b5 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23nsaKZFjqo.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23nsaKZFjqo.pt new file mode 100644 index 0000000000000000000000000000000000000000..932a25877a7651c4de2681554770c57aae2fb112 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23nsaKZFjqo.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1d90a4a9fb8f696f4acb00b24e54dcac5866324ba39ad12139beed370979f954 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23p8NmAXgrE.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23p8NmAXgrE.pt new file mode 100644 index 0000000000000000000000000000000000000000..5a8c248bbf87684773de6233d8ac20a4e5d0a181 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23p8NmAXgrE.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0444a7afa8da19870513fbe46ca6b6bf5cd4f9fd2f14f0c087005e91dd9692a8 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23tdTPJwhLk.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23tdTPJwhLk.pt new file mode 100644 index 0000000000000000000000000000000000000000..7767bdd5a341798e214a1a96fd23dfcb978d592b --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23tdTPJwhLk.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:80d4fccf80d49ebac950640d383eac88b60c86e7c5c5144b6ffcb0d58101a05a +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23uLZbqBuB8.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23uLZbqBuB8.pt new file mode 100644 index 0000000000000000000000000000000000000000..fa09afff956d3fbd475c94132c56e286b66f1bb1 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23uLZbqBuB8.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9d02fb20e7e151a59eb02cb820b9ba1ba4bb5c089a00f9e3e9747d624005a9b9 +size 17041736 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23vjb2OGay4.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23vjb2OGay4.pt new file mode 100644 index 0000000000000000000000000000000000000000..b9e63195867660e1636357753c79d80e8857fd79 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23vjb2OGay4.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1d5b2c71d999be3a04d74f561a06dc7c009c68f0b1c35e1c813e2c54b07289a5 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23ycQ-r1zCc.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23ycQ-r1zCc.pt new file mode 100644 index 0000000000000000000000000000000000000000..5f67ed008d4445ed23e750d384be585ee5429d54 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23ycQ-r1zCc.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0e353075a24feec595719f60ddf300fe68ada142834e6277ed1d2089b0323642 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23z2ZmhZ6c4.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23z2ZmhZ6c4.pt new file mode 100644 index 0000000000000000000000000000000000000000..3d69cdf3cc04693b9de0d2a9050dcb06580495c9 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_23z2ZmhZ6c4.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8111e909b8cacba352958d067840aa07a7e7315f35b43580fc59f436b24e9dfa +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2439g6JbKgo.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2439g6JbKgo.pt new file mode 100644 index 0000000000000000000000000000000000000000..28350e2d4cb00eee0607d2d3c657e2ed74f7b8d3 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2439g6JbKgo.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:75c5b77a4f6b2624414c28de754390618139af0bdafc7b4948a8e173a610d249 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_243yRBDOWFA.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_243yRBDOWFA.pt new file mode 100644 index 0000000000000000000000000000000000000000..fd80125c0950677fe3d06cb9007fd4e67ad4e82f --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_243yRBDOWFA.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:71fbd12e341a135f9e585b7d796d5bca91feddd6a67c07354f711dc3a34f7e43 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_246z8o8tjSc.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_246z8o8tjSc.pt new file mode 100644 index 0000000000000000000000000000000000000000..8242b4ab0470da956533d506780675148c49004d --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_246z8o8tjSc.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:386988e1b17a3719594a44a1bec284769cfeba7b66d02380ae4fd1328f808981 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_24DJTAFHrro.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_24DJTAFHrro.pt new file mode 100644 index 0000000000000000000000000000000000000000..585dbf1a01c59dcf1d6080b87df8b369e11bdfbf --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_24DJTAFHrro.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f9e513c2a00a37e1084b0edcb360ed44f2a2eccbe79c3bdf1ca466deee4c72ee +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_24JjTSQX3m4.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_24JjTSQX3m4.pt new file mode 100644 index 0000000000000000000000000000000000000000..75209a0ccd55ccce72211e4e1e43217829903694 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_24JjTSQX3m4.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7c2a4451fbb25611c31d0ce3d9e2ea33793eed4b20d6eb6fef043333ea23d4c1 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_24LtcBzoIPE.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_24LtcBzoIPE.pt new file mode 100644 index 0000000000000000000000000000000000000000..ee6021a1cfcf7e7fa2c8ef1d4ae76d1795f6da3a --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_24LtcBzoIPE.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:05f32f1e6d656446d22e6c0dad87c2fa7197f80564ce7a833077f5367a64f7af +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_24O1dx0KqFg.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_24O1dx0KqFg.pt new file mode 100644 index 0000000000000000000000000000000000000000..54b5939bb1628984cfbb9306a95acb17f9cdd0ed --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_24O1dx0KqFg.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cbf5c4f58335fba582c8e549d42e24aa9489ebd341874b414ceaa2d645cdda95 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_24TNzh53FUc.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_24TNzh53FUc.pt new file mode 100644 index 0000000000000000000000000000000000000000..7a6e91b11b2cc37098e24224451139ed806e1c30 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_24TNzh53FUc.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fdcfac543cbc511d8cb86e0d79fcf56758a7aadefa8fca213f7489d6179aa2da +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_24ZWbpr67wQ.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_24ZWbpr67wQ.pt new file mode 100644 index 0000000000000000000000000000000000000000..d090f0289465cb4d5317b6434dd5532f08d4b8b4 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_24ZWbpr67wQ.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:510beb5a0c45476d80678cff27897e3cb6b479e1d11536f4339559a2ac780346 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_24ZgeaPlKow.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_24ZgeaPlKow.pt new file mode 100644 index 0000000000000000000000000000000000000000..8bf4711045088170bd4c090ee8747782d1643430 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_24ZgeaPlKow.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:21e1741aad4817c7e5e5c55dbd25966a16ed38073803171e0d918c1081a9f155 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_24_EVgEZtdY.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_24_EVgEZtdY.pt new file mode 100644 index 0000000000000000000000000000000000000000..7cdaabc5fc20a2918a1258d5a22220a9edd79bda --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_24_EVgEZtdY.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0111c8b0240fbe2cd75342c3cb1415d64b4a822a6a79976255311d5ff278996e +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_24g9a1ckS5U.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_24g9a1ckS5U.pt new file mode 100644 index 0000000000000000000000000000000000000000..60994ac99a7dcedd9263b995809165c995967d52 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_24g9a1ckS5U.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6717c92f7fd8213099b9ec227e01e9c2b3e930845678fb52f1f51379c1dacc8d +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_24hB3u1idkU.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_24hB3u1idkU.pt new file mode 100644 index 0000000000000000000000000000000000000000..6047b19b21bd7a730572b027ffd07f4dea23ce39 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_24hB3u1idkU.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e5c830f44e8684fb57a06662d2b062b24bf5e4356e18468e0d22b062ebc500b6 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_24oEdNJAMXk.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_24oEdNJAMXk.pt new file mode 100644 index 0000000000000000000000000000000000000000..d485bacb0422f2fa1620033a3c6d216ae68bc24d --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_24oEdNJAMXk.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e9ceb526a9016b3f83f5575a604104dd5861c2a77f473972b3b536a1df61d55c +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_24qANfmfLys.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_24qANfmfLys.pt new file mode 100644 index 0000000000000000000000000000000000000000..fdc46039070dc11801762dc6b693b1e638fc7b63 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_24qANfmfLys.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d7abcb2929be8dc725b871767103afafd50230b2ed81112aae776747c2440481 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_24sMw-CRDd8.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_24sMw-CRDd8.pt new file mode 100644 index 0000000000000000000000000000000000000000..349b39a0d110a45bdf2cde1a2bcc67e511e678d0 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_24sMw-CRDd8.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:180012e541de14ae65fda0c90d3f949193d05f6e0496a99d6a9ad22894fdb39d +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_24sPFefYyco.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_24sPFefYyco.pt new file mode 100644 index 0000000000000000000000000000000000000000..f9f0624996cdfdd0ea1a3f941287868f4682292e --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_24sPFefYyco.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a6dacaaee23e9fc39b6b741697f48880082d727ec0dd12c40fe76ce79fd3f22e +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_24sbaGMb0ko.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_24sbaGMb0ko.pt new file mode 100644 index 0000000000000000000000000000000000000000..e1997c9a9cf8ca50710c4e5bef88466499ebeb57 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_24sbaGMb0ko.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:311bedc1a2e017b32942ce83aad37638a38fad3992a0ce5796084a1b7bf4f55b +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_24zr0oLIgbc.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_24zr0oLIgbc.pt new file mode 100644 index 0000000000000000000000000000000000000000..a86d66f5d84e3cea141754f162c6c5d3918316ff --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_24zr0oLIgbc.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b100c44637c3533009bc065a89e8ddafca45fcd24cf9a75940ad06003bfb58bb +size 17303880 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25-YOLdTz34.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25-YOLdTz34.pt new file mode 100644 index 0000000000000000000000000000000000000000..1d421ceb97e1c35cceb5a429ddb06d41f867cefa --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25-YOLdTz34.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:141e052528480a2783ef3282020e6dc69353715007e605e5512b508cbc9769c2 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_256oK89hxsE.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_256oK89hxsE.pt new file mode 100644 index 0000000000000000000000000000000000000000..2b8884665b5b7bda3f1e99f404a837bfe3667e85 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_256oK89hxsE.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:00d54e81a96d56cad5b60a3f9ae07f92e2eed54475780069cff4096b07ac4c07 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25Cfopx286s.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25Cfopx286s.pt new file mode 100644 index 0000000000000000000000000000000000000000..1ad3ba2ebd825c35c600fa77fb559fbfffc5f369 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25Cfopx286s.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b3c5bfa79c4ca2b1056158553ddaee65ffd72c715e136243d57722c9ba14a4c1 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25CrxMAqGpk.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25CrxMAqGpk.pt new file mode 100644 index 0000000000000000000000000000000000000000..31c759d23b0aab538207ca8e86e951fd4dad5efa --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25CrxMAqGpk.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e77dc21fafded9a2b888eaf8caba3a585e83322e874d9518acfb2d480269c0a2 +size 18352456 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25EZHfgM_EY.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25EZHfgM_EY.pt new file mode 100644 index 0000000000000000000000000000000000000000..0aed6bb3ca8b7058299e2533252b2696ecca45e7 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25EZHfgM_EY.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:80b5cb95b3056a038647416ad441ece82341b4fadd90d24732bd5d57e15164a2 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25HO5OSub7s.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25HO5OSub7s.pt new file mode 100644 index 0000000000000000000000000000000000000000..e795ea21d31e346950a9542d2816f1663e969dca --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25HO5OSub7s.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f11f4cc88d1b182f4e0372a84de8195b594bbd26df599fb16873b813204df7bf +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25QZ1TX051I.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25QZ1TX051I.pt new file mode 100644 index 0000000000000000000000000000000000000000..39214b8a9c5001d20ce8cca6f4892de6570eb910 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25QZ1TX051I.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:708de2c63826c180c61995b8311b3260e99e702b8dc82fd5fdc9ab7b45dcbc77 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25SfOnyUXuk.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25SfOnyUXuk.pt new file mode 100644 index 0000000000000000000000000000000000000000..378c964ac7b509112810e1e655903436d9b85e93 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25SfOnyUXuk.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bcf82c2ab54268b8a031d2755f629da65314ce9e124a9abebee61e725ad0375d +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25XNOJbeNtE.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25XNOJbeNtE.pt new file mode 100644 index 0000000000000000000000000000000000000000..b0a3d84f1e65e5566b9ed26ac0b9c4307e50c18c --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25XNOJbeNtE.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3214f90e1ccdac29f14d1a729ab3384b2ebeab4294f074b612faa875ffacc15a +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25ZW4FXQyEM.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25ZW4FXQyEM.pt new file mode 100644 index 0000000000000000000000000000000000000000..4619c8db4c030dc8e6908a7d18acef8ae5cb9790 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25ZW4FXQyEM.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:af3654f806211644df88d90a465eaadf2c6f7957d6de18037b08f735bc14cf12 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25_7XLn6OTo.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25_7XLn6OTo.pt new file mode 100644 index 0000000000000000000000000000000000000000..0a644f295d2a1ad86810ac2053470169ebbd4401 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25_7XLn6OTo.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:568e9b1a60ef27193fe4962b62b9d3e40514ab42e09a3c77358d0ad11ab7ade1 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25aKDmcK8fo.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25aKDmcK8fo.pt new file mode 100644 index 0000000000000000000000000000000000000000..42a1f0139fee770815acc19d5bdc29b5d0f45cbd --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25aKDmcK8fo.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8a423bf65ea9e32df93895f9a7a8fb1ce4f6c914d8449eaa979076fd92ee1397 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25fEETTTuQQ.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25fEETTTuQQ.pt new file mode 100644 index 0000000000000000000000000000000000000000..6eb4cd9f6a6d73a85d1a11fc47d1054d5de71c0c --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25fEETTTuQQ.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7c6ad900e0bfff9d956fa02d54b18639a6cac265a5cc3a687791c86562845fa6 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25foQA84718.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25foQA84718.pt new file mode 100644 index 0000000000000000000000000000000000000000..e6316051f3db05839f1ca27fdb075204a28a4d59 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25foQA84718.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d6ea6d989d325b8928c4c19362ee2f7dee06cdaa5bea9756be79116ea7993b21 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25hM7PRuaD4.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25hM7PRuaD4.pt new file mode 100644 index 0000000000000000000000000000000000000000..c7a540c42b3e43109b5044ae36bd0c923b4977ba --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25hM7PRuaD4.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:32c8f9563067c7c1b1bdfe50b4bbf13d54a63ade23e8cd1b94956553ec5b49e1 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25iqzwl5bfE.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25iqzwl5bfE.pt new file mode 100644 index 0000000000000000000000000000000000000000..ebf943148697db168f562369eebc646bf9a7eabe --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25iqzwl5bfE.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cf2a777ce3312197bf28439d849e89229db34dea7226805236cabe66dc377e3b +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25mv2bhmgII.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25mv2bhmgII.pt new file mode 100644 index 0000000000000000000000000000000000000000..0e8d812c750e92243c3e9c9c2ec4e96d33502246 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25mv2bhmgII.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a358bef4613bb46c89a0d160744056b5eb0142bd24e4883d89270d4e91515617 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25rcrTHWh8A.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25rcrTHWh8A.pt new file mode 100644 index 0000000000000000000000000000000000000000..37a9408133ebf051da89cca96c942001cb2c7b23 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25rcrTHWh8A.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:463d1c6324a6bd634e66a72fa1eb21ec1b26b69a737092363dd294ac1654e4c6 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25tzXh5bmO8.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25tzXh5bmO8.pt new file mode 100644 index 0000000000000000000000000000000000000000..6d07338575c4eef53da4081f9d7a4bd8af1e95cc --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25tzXh5bmO8.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8198b65d1eb6e463aa7faea9f2e282c9d1ecc6b8934a31607c2346830ff80466 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25vV1L6Lz7s.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25vV1L6Lz7s.pt new file mode 100644 index 0000000000000000000000000000000000000000..7167de6bf2bf127895a3bc368ecd0f8511ade9af --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25vV1L6Lz7s.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:65b9b6cef85d61372f9cb7de9da231a344ea1c6d8d4eefceeebea32629f5ca1c +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25x1yielnTk.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25x1yielnTk.pt new file mode 100644 index 0000000000000000000000000000000000000000..644b2f6512a9b12900b235d1b44b2658a0cad138 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25x1yielnTk.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a1be3de792955d08dc34c9de08d21d2d72f8fae05135a4e2ca48a24472173f8d +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25zTALIp2Ng.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25zTALIp2Ng.pt new file mode 100644 index 0000000000000000000000000000000000000000..8236a24755975de725e6e235a8be96303b438924 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_25zTALIp2Ng.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3e56e8a57130bcc317fbb30a09477571cdae8a097eaa82fcfb848cf52b4570ee +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2626HaXs410.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2626HaXs410.pt new file mode 100644 index 0000000000000000000000000000000000000000..edf806a0ba3665b0a1de65caaaaaf9fdeedec7d1 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2626HaXs410.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2338893ffd29222034507149f27db551aaa377e245e89160d9a8c3b78dc56019 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2633nIzzBas.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2633nIzzBas.pt new file mode 100644 index 0000000000000000000000000000000000000000..4935582a5cad2054e541424737645d768fa866d0 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2633nIzzBas.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0565aecd32a6a4c5f2a8ab375bb75a028c2e095dfef1cebb934ee82ede3353a8 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_268mypKSkvw.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_268mypKSkvw.pt new file mode 100644 index 0000000000000000000000000000000000000000..2c2ecda4b1570d901814aa5566e230b367271bec --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_268mypKSkvw.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:288fc8b30746e63deeef79cda117c794cecefb0d63f2dcd63e5e7dd8750623ee +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_268y2an96JA.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_268y2an96JA.pt new file mode 100644 index 0000000000000000000000000000000000000000..4923afef31e9be2d6b690af19fb5e5045cfb7d96 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_268y2an96JA.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6ba1446d310de27d0e0a294fea141f6f14fa2f1232d0aff810630e2604424487 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_26DcihcNKSQ.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_26DcihcNKSQ.pt new file mode 100644 index 0000000000000000000000000000000000000000..db401f550c6808d9dfd3f7430d65f91499ed244d --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_26DcihcNKSQ.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d696b4b0ccd42a55e5a9f6127986c6853a06177c7ff94a61dd79f5d242de0428 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_26Fn_FauZ04.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_26Fn_FauZ04.pt new file mode 100644 index 0000000000000000000000000000000000000000..9ddfd94a6e3886c171daab4ce51c883a26d9e8bd --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_26Fn_FauZ04.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:48e38eb52a0a6d41b01b4daddfe765835b27ff98faf12f8546dfa2372dcee29a +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_26N6AWJcCWQ.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_26N6AWJcCWQ.pt new file mode 100644 index 0000000000000000000000000000000000000000..b59b0d4fd4927d2488ed9c1d1bd79f992036c8df --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_26N6AWJcCWQ.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b594afdf92f97f38e06163c59d413784604f653f85bda1f6e78ed06bb3556a48 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_26NXa658j2M.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_26NXa658j2M.pt new file mode 100644 index 0000000000000000000000000000000000000000..04b555e4b249b7a6cd06f903d03cc15a3e9bc8c6 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_26NXa658j2M.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:82eb3185987d4a645b33d365e9a4194ba1ec838ca910fbaf5aa13562b66fd7a3 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_26O9-IOTZtg.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_26O9-IOTZtg.pt new file mode 100644 index 0000000000000000000000000000000000000000..aab4353d5cd10682e0187824a831a85403f61ead --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_26O9-IOTZtg.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:57b20fe260f107fab208f9d2172bcc8ae8cf8b3956a876ccd66ca49df42663aa +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_26T6rxoh66c.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_26T6rxoh66c.pt new file mode 100644 index 0000000000000000000000000000000000000000..c32ad681168f9e2efd9f8c717415ec0a8f444c20 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_26T6rxoh66c.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:422cdc3383babac0b2f3cdc45b91ea32039525b512d894213e6d76af7818a868 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_26Ueir4b41g.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_26Ueir4b41g.pt new file mode 100644 index 0000000000000000000000000000000000000000..f94791461ec9bac8e49ad927e72871015e9f27d1 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_26Ueir4b41g.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c4d34564bb5ac7ff8ff917a4e4aefed1e371112b3b76e38d448d5defcf32bd14 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_26XjA_tcTzk.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_26XjA_tcTzk.pt new file mode 100644 index 0000000000000000000000000000000000000000..b11900562adc30c5bbc29bfb52874bece3973838 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_26XjA_tcTzk.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:514940d650a04153570f8e985ccbf501a95a4e1a368fa1916c4a6e37a7dd58cb +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_26edEZGiEGw.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_26edEZGiEGw.pt new file mode 100644 index 0000000000000000000000000000000000000000..ea5dfe544801b3e83ccbc79ce7d8bc4cd889a09c --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_26edEZGiEGw.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b2bca8b286f672da40650a158f2698ca9df9250b1265f04aa0e7e0e959104a8a +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_26gjC4bwqzE.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_26gjC4bwqzE.pt new file mode 100644 index 0000000000000000000000000000000000000000..b988f11f4b113a6f48a42235311a78429cab1504 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_26gjC4bwqzE.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:04830f00553e777ffd7f96cbb1d43f4089fdfb42d586387a3c79c82fdaa86f46 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_26jE3mbj1uw.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_26jE3mbj1uw.pt new file mode 100644 index 0000000000000000000000000000000000000000..378c6c9845ec9d689a4ff804349b1827e67bf218 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_26jE3mbj1uw.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cc73cbebf0c5c056047826ae73d01ead8b53ba9c45f215c316815d2896d42515 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_26uHZoZNP0Q.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_26uHZoZNP0Q.pt new file mode 100644 index 0000000000000000000000000000000000000000..f4842a05c29077e5840fcae4ac5b1c4d9fe93197 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_26uHZoZNP0Q.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:967bb4b659455d6971372f7de8bc5d30fc08d79c7af36d1e388cdd76fe530355 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_26ubOKI7cRQ.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_26ubOKI7cRQ.pt new file mode 100644 index 0000000000000000000000000000000000000000..75bde8ac1f3bab0cb9eb017e6d6d07c267c694f9 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_26ubOKI7cRQ.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:03456bb2d6dc3da326565e8ad03541eea691e92c6fae53afccd02bf0b728be3f +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_26w1U1SmIq0.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_26w1U1SmIq0.pt new file mode 100644 index 0000000000000000000000000000000000000000..e22cbd5f2b03d3a6055d7c398071154996661b2f --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_26w1U1SmIq0.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5c52df579fab92bde0b8c067038799dececc4095b5eddacdc785222872d95d3a +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_271bmvoygY4.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_271bmvoygY4.pt new file mode 100644 index 0000000000000000000000000000000000000000..775badae3872175851a4dc3289177c52f783c10a --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_271bmvoygY4.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6d13b70e553f2778d592929566c2bc0e695a574fb8ae573238855a94b404175b +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_271oVaOT00c.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_271oVaOT00c.pt new file mode 100644 index 0000000000000000000000000000000000000000..31e413a0eefe4a718da860c4d1f3a42baff48372 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_271oVaOT00c.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:29a5ef70af3dcfd856025b972c9e7b06c6b45c32f6d457668b445aca7ac7c687 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2749gmDqtWk.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2749gmDqtWk.pt new file mode 100644 index 0000000000000000000000000000000000000000..bd18f57d359be8f5cd833f57ceffb6d44053d731 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2749gmDqtWk.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5a9a282889bb163a218c2fa956b0a16e933c14efb2d504194bee479bcbebb135 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_277GGLmTgNg.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_277GGLmTgNg.pt new file mode 100644 index 0000000000000000000000000000000000000000..568544c6a8825da0248c5d8177173f7d9c92a914 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_277GGLmTgNg.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e55839a8968e1954121c7767fc0079b3742b107272193938694427d426fb9ef5 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_27AqyRw7WuA.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_27AqyRw7WuA.pt new file mode 100644 index 0000000000000000000000000000000000000000..94cfb2175481e8fc3b1f79806ae5d7f244475056 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_27AqyRw7WuA.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3167c19610f7d8298ee2ff1321f9ca54250c922232154dde76407121a451f45b +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_27Hm-N6q7eY.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_27Hm-N6q7eY.pt new file mode 100644 index 0000000000000000000000000000000000000000..62684b8d99e24c2528ff358d5f7be2ff9ffeab32 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_27Hm-N6q7eY.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c6179700c120bfb5c0b6ae248d6d2a6b4fc54e65e9506088fd3fda5fdc47fe19 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_27IGPmP5vd4.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_27IGPmP5vd4.pt new file mode 100644 index 0000000000000000000000000000000000000000..4032e67a4282970b4d68985e91a8aee0316cb8a7 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_27IGPmP5vd4.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7696feba50a4e4dcea2ca1264536cd9c5a0204139a8f21999b75bfe7fec940eb +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_27JV5arc7Hs.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_27JV5arc7Hs.pt new file mode 100644 index 0000000000000000000000000000000000000000..0d876791cb78e075c3a144dc73b0da4c2236c4e4 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_27JV5arc7Hs.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:921cbadecb0679e3d59c47ce80289f4e75f171960b38ac1b58d90e2545dd7b05 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_27M7k1s55QM.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_27M7k1s55QM.pt new file mode 100644 index 0000000000000000000000000000000000000000..b836a493a159181d538be21a0558382345727c6c --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_27M7k1s55QM.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ea88614f1d131dea5cae1e3fe6b2126243eb052fbead983a0e7f737fc322eb8f +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_27VYRaAO_Is.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_27VYRaAO_Is.pt new file mode 100644 index 0000000000000000000000000000000000000000..554a879a73d3136042d089d14088104f03df02c4 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_27VYRaAO_Is.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:434e0f35ea43a868ed34e21ce645137b48dee4cd448f2e274904234676de2179 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_27XcRUPQU6w.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_27XcRUPQU6w.pt new file mode 100644 index 0000000000000000000000000000000000000000..7ac4d915c3b5df2c3b3a4fb194f1cf9fb9828012 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_27XcRUPQU6w.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b30b5cb1b3b26f04bfc13d7e36938053d579c0763dbc1aafe997fae9ef8effc7 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_27rBj_w87wE.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_27rBj_w87wE.pt new file mode 100644 index 0000000000000000000000000000000000000000..e38b8a9494a665c41a0641cafed62dcd1ac629bc --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_27rBj_w87wE.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4d1915669feb8a5c1797d63b3d64ff1e8b25162beaf794645cf553888fac153e +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_27uYMjFZCcM.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_27uYMjFZCcM.pt new file mode 100644 index 0000000000000000000000000000000000000000..f1d9d021fcf66a4c5027273584d1c2727d113fb5 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_27uYMjFZCcM.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:03cb2c79e9961c3cd508789d0f2f5c0c94d101f99ea21b736bbce84f03bc50bd +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_27vx4S7CI8s.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_27vx4S7CI8s.pt new file mode 100644 index 0000000000000000000000000000000000000000..8daf313498b16ac5c6a99f48191bda347efda540 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_27vx4S7CI8s.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:596f01e0f273e6592d12ef5e75109890b02d230260a719c1deb2c14b5a662592 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_27yqWssROGk.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_27yqWssROGk.pt new file mode 100644 index 0000000000000000000000000000000000000000..81960af83a9c068004cd69d88a136949da9a2023 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_27yqWssROGk.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d259afbe9e3e1d03a555f3cc3acac615fdcd6298a5f660fc83b1f024f41ce215 +size 17041736 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_280nOgCznz8.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_280nOgCznz8.pt new file mode 100644 index 0000000000000000000000000000000000000000..cadf42579e3013bd4bfa86be0ea8a58b2529a3d4 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_280nOgCznz8.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2b4fb1468ad0088ee010825d22de07612bee005edc4cd94c2f1fe862f8dfcac8 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_286tq_XPQkg.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_286tq_XPQkg.pt new file mode 100644 index 0000000000000000000000000000000000000000..05b091789216d5ae4da1a4a051963a5b557cbe62 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_286tq_XPQkg.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ba94b4e8347f464a3e5651a12591359475d4969e380b477a3b36b33eda8b5192 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_287V4zCVED8.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_287V4zCVED8.pt new file mode 100644 index 0000000000000000000000000000000000000000..edb3993f5a751b0e374c1e72238d71046c3d2e18 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_287V4zCVED8.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:600657a9f38576a252169e76a2122146f16117d9828b8b8dbe519b79ac5f4844 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_28A2zEk-67c.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_28A2zEk-67c.pt new file mode 100644 index 0000000000000000000000000000000000000000..f5b553d2f642d204204fd074862226283b80e611 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_28A2zEk-67c.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5cd6af83c18793e15954427201d60b157783b86e14450e15729886970df4f7fa +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_28K8jEYtYd8.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_28K8jEYtYd8.pt new file mode 100644 index 0000000000000000000000000000000000000000..9ef55be7aab88a7f66cd83507344ebedaa144b00 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_28K8jEYtYd8.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b82e033874b2322c9c1f2e7fe5caee39fec57768dfea2297efce33b636c61f16 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_28MPyyGA01k.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_28MPyyGA01k.pt new file mode 100644 index 0000000000000000000000000000000000000000..0188b8ff1a6e7bff6e3ffde2fe038b1139bdb057 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_28MPyyGA01k.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0aa4c0a83fd5236fc5bfa6e6368b0969a3b8fc1140830d0e8b530f2278c0bb80 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_28Qb3txcuRM.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_28Qb3txcuRM.pt new file mode 100644 index 0000000000000000000000000000000000000000..032e6dfaf3fb034ab68086a9f3e060a4eaef168e --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_28Qb3txcuRM.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fd6dd1bf03e2512df8e0b411c7989ffaa1146239212488709dc802292323d5fa +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_28SjRklpcmE.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_28SjRklpcmE.pt new file mode 100644 index 0000000000000000000000000000000000000000..742547806d71b2ba4913679b9006b191dff675fc --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_28SjRklpcmE.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3edf4babf8b594e1b911629392ba456196afda6f87b2bc18113e65ffda4da6fe +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_28rRjdGVkdk.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_28rRjdGVkdk.pt new file mode 100644 index 0000000000000000000000000000000000000000..f344f5cd8385d05179e2d4148686739e4d4e9ec7 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_28rRjdGVkdk.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:97636cd3c0240d318305c3cc51395a331490993f8c45815ad9efc5f8ea922ab4 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_28x0wmPPZC4.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_28x0wmPPZC4.pt new file mode 100644 index 0000000000000000000000000000000000000000..d456c7d97529d2b8d32447ea9d22fcf8211acbb6 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_28x0wmPPZC4.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:feec5404bfd1e1cc4a0f51087ef56287386e15099ee4aa4fb51ccf1d19750ebc +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_28xfVHhWdmI.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_28xfVHhWdmI.pt new file mode 100644 index 0000000000000000000000000000000000000000..bceea0643ec4f756f53e0d8ddcdf18c0f9d5798e --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_28xfVHhWdmI.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c762b6e5ec339942dd8b2a5e487d2ba781ca2c4c8d85a335e447a58f8899119b +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_28zbr7JjZb0.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_28zbr7JjZb0.pt new file mode 100644 index 0000000000000000000000000000000000000000..0bf9ab7a855cdd6cb1f77faa3ea17c860fea1cfb --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_28zbr7JjZb0.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:71a16cf8c3329aa1c8acaec4999f4fe9d5965fcd05708183796ec949e4c4e704 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_290Ma1tbi-4.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_290Ma1tbi-4.pt new file mode 100644 index 0000000000000000000000000000000000000000..539ef0ccf8b6ca6d0f3e25c5efc2552a7ed5bd5d --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_290Ma1tbi-4.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e78a47558a894f2ccc18cbff4efdba413844430a714a26c29c75c1496d61c35a +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_290TOEynE9s.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_290TOEynE9s.pt new file mode 100644 index 0000000000000000000000000000000000000000..6761f8bd36cadb1fcc66c141af400e81122d836c --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_290TOEynE9s.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:52dedfcdbdc00c73d3b5128eb4467efdb59905b0cc4fd9e807af681030aacd93 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_291BYNIWVTc.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_291BYNIWVTc.pt new file mode 100644 index 0000000000000000000000000000000000000000..01c8aa897d33f2bf205285e3a719275af65e13c9 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_291BYNIWVTc.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cd11fe88b44e355129ebbd9a27a908ee4e75e2789e39065dfee016d5a6f8016f +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_296uybTIzLQ.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_296uybTIzLQ.pt new file mode 100644 index 0000000000000000000000000000000000000000..8c927b8326b6e88108f24cb310f7ee207a9e6e70 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_296uybTIzLQ.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6348915929cf15309d85b6c837da340f404c5b7df8fc16405b32a14c13299172 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_299U_B9cCyQ.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_299U_B9cCyQ.pt new file mode 100644 index 0000000000000000000000000000000000000000..9e4504c52d48c0e19fb8f1a739873e2e58146dfd --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_299U_B9cCyQ.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:72c9a45943dac5dd1d48d01e2e58475c8120abd309966d0fff715bb115cdd3cc +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_29RHfes6E_E.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_29RHfes6E_E.pt new file mode 100644 index 0000000000000000000000000000000000000000..e26b112200747e7ad9eaa8ab0f58f6dd50ca1493 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_29RHfes6E_E.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8eb9a5606f94fd2517aee51bdee5763a3071158791032d556359d63e81477300 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_29Y5bJbIMGo.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_29Y5bJbIMGo.pt new file mode 100644 index 0000000000000000000000000000000000000000..075a525d83e656c7e52806056f734a058449a721 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_29Y5bJbIMGo.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:088cf468d20bc44f6e766349af378f68191cabc6351b12aa68f6500764256f89 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_29YB-wF-D_s.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_29YB-wF-D_s.pt new file mode 100644 index 0000000000000000000000000000000000000000..68d79dbf84ba4cfc22a2c599992a9c894c942d2d --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_29YB-wF-D_s.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:519ce0125dfe7dcc73512b9b46aae848870c9e246a73a1df41f694414b4cdb24 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_29cjEWBSeWE.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_29cjEWBSeWE.pt new file mode 100644 index 0000000000000000000000000000000000000000..82902a8371a6cea5887194f327f8943e6946e09a --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_29cjEWBSeWE.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:653dfcedca3d77a061419adc0e481ab11a1b41cb6093c61fbaab0d7344637a92 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_29dGnShXyDk.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_29dGnShXyDk.pt new file mode 100644 index 0000000000000000000000000000000000000000..0ccdd81e40d0133b1bbf06d8d1a7a047150af95c --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_29dGnShXyDk.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:beaa3e0719325df7fd735b40d958e9ef097da1c45cf89a855f55ada49d68243f +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_29eUPxSwyHc.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_29eUPxSwyHc.pt new file mode 100644 index 0000000000000000000000000000000000000000..d26b96cf6de424c7f354e1b1e68a029b236e7cf0 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_29eUPxSwyHc.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:298af82c8bd7327b3cbf4c0d9275e29d72699e5194c556fa01b7b502d63143d2 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_29gzMDEUkH8.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_29gzMDEUkH8.pt new file mode 100644 index 0000000000000000000000000000000000000000..336b8811c42e88525578efe3f3150745e6dccd1a --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_29gzMDEUkH8.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:38572b6589d95ae74ebeba8f5dab80cb20e6fb6e647296d25bb5ad9fb769aa6f +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_29jV5xMOqhA.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_29jV5xMOqhA.pt new file mode 100644 index 0000000000000000000000000000000000000000..c16daad225c5506a2a0838fa133ebb09b34faf57 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_29jV5xMOqhA.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:def72b037ae2547df4ead19604f791331a2cc2ad1982158d878136425eacf6e0 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_29lfNH3Ky_8.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_29lfNH3Ky_8.pt new file mode 100644 index 0000000000000000000000000000000000000000..a21839e076645a5137046f48195fa6bd259fa358 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_29lfNH3Ky_8.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ba993dba3583c133b9cf67d5b13cf6c01032f4e7ad5b3a5440261afcb1941310 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_29ly9mR0frc.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_29ly9mR0frc.pt new file mode 100644 index 0000000000000000000000000000000000000000..15d389a3cabc18fb1db57ca2884eb41d8fd81b6a --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_29ly9mR0frc.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a0f7f2c825c6200dc808fc3a6c14aefd147c319cc63f13e8f4e0a74047e2e987 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_29nho363JHQ.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_29nho363JHQ.pt new file mode 100644 index 0000000000000000000000000000000000000000..c81a58295a6948476689c9458d52e3087a340561 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_29nho363JHQ.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0731eff01710782e8d15686936c23f657d57554b83fa73a6ea44374fe6c2b225 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_29pxGi8PWZI.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_29pxGi8PWZI.pt new file mode 100644 index 0000000000000000000000000000000000000000..27ce95e17e273c93b1ff45d2c60060580dd7535e --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_29pxGi8PWZI.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d3074104a92ab9c51deb076cf3b48106388fe5c58abf38524b24e9210548942d +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_29sELhjff6E.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_29sELhjff6E.pt new file mode 100644 index 0000000000000000000000000000000000000000..da3610361f8a9c3868c91cf3770146723c02e916 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_29sELhjff6E.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cd7efd3381a3af7b80f052867e48445c7542b37bbaf3e8d40d4d23acaa25adb7 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_29ujH4IFRTQ.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_29ujH4IFRTQ.pt new file mode 100644 index 0000000000000000000000000000000000000000..4900c69628144af627953c490de7808afd2ac20f --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_29ujH4IFRTQ.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3a9ae7f283d87501c8c7a5ef50b2eb445f555384ba994885c8a2db49044193b4 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_29yQOXOrGRU.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_29yQOXOrGRU.pt new file mode 100644 index 0000000000000000000000000000000000000000..32030aa4eecf3fb4a31f95014e08087107fa3cba --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_29yQOXOrGRU.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ecc134dc0d31c9d27a98a7f6c6fbbd02f5e3528e22db05d0878f58c85ad0447c +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_29zSotQoruY.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_29zSotQoruY.pt new file mode 100644 index 0000000000000000000000000000000000000000..6773a995249a4210c9674bcb185d98316048ac41 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_29zSotQoruY.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a970a879bfcf3363dacbcf6818f9d757c190641f2fe568030483675e55f92016 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2A5HHAenC_Y.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2A5HHAenC_Y.pt new file mode 100644 index 0000000000000000000000000000000000000000..66a719781479c977dffc9263cedca0c4eecee2f9 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2A5HHAenC_Y.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0aefe1a16369cb26dd5d29b72465384a46facec8c1221c9789e5d04b5a8d1903 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2A9KvaR1paY.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2A9KvaR1paY.pt new file mode 100644 index 0000000000000000000000000000000000000000..b509127bf638535553caf20975e6b2660ee276ef --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2A9KvaR1paY.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9e996a16f98f0968c95e19d6611f421211202d6d94bbe6d73de3e8608cf7ff40 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2AJ79pnmU9o.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2AJ79pnmU9o.pt new file mode 100644 index 0000000000000000000000000000000000000000..484d97176a80aa0c85b8d08eaed6e0811bbf3f28 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2AJ79pnmU9o.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d7bfae55c86c622e3220236d87f9232013235274f35b8ac61b7eb25f199704f5 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2ANfVRgKbNE.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2ANfVRgKbNE.pt new file mode 100644 index 0000000000000000000000000000000000000000..e066234c89350fd514310cf3eacc53ab87296cce --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2ANfVRgKbNE.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:084a7a66c87f7aa5e62b2a0fabf3a2df38c0220bba60ead79a5d78b128055d68 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2APv0DtStro.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2APv0DtStro.pt new file mode 100644 index 0000000000000000000000000000000000000000..f81591d75f3d3fdd2c856ecdcee52ce44b0efff8 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2APv0DtStro.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:50014a3b14fe29f7f9ed94a4cfca8eda7e09bfb04ec220230a92b23efd685c66 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2AXOjuhyFMg.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2AXOjuhyFMg.pt new file mode 100644 index 0000000000000000000000000000000000000000..b02d3b248f94abf215bb2cfd71f3a9ccce5521fd --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2AXOjuhyFMg.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:91f619d8f1f3ac79905302f2ee7be77de402a5c0579ae06b20e42b987e8960f3 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2AZ7HQ4YAlI.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2AZ7HQ4YAlI.pt new file mode 100644 index 0000000000000000000000000000000000000000..66e7ea3e19fac7d3b65811f875c25ce2c1267f0b --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2AZ7HQ4YAlI.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9e7ec98b2f3d115e379a3a9f45f4b2d6f3867ed7249617650f75809835faf2f6 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2AZNcYS2OD4.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2AZNcYS2OD4.pt new file mode 100644 index 0000000000000000000000000000000000000000..fa1497dbc7b670eb53fc31943a06ac2e6039a6da --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2AZNcYS2OD4.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3ab26c773b24c14a3917e6b4ab5577bcb73046b5c4a7887986fcd4f13f76f8b1 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2AeRp38iq-4.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2AeRp38iq-4.pt new file mode 100644 index 0000000000000000000000000000000000000000..066e5eb8e6d28bfb356d7e7060ac08b191911ecc --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2AeRp38iq-4.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:83f7cda1914a18b847bbfb77b278bc499aef0bfcc74b0f317cc7a97070e4e558 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2AgyRHY4xug.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2AgyRHY4xug.pt new file mode 100644 index 0000000000000000000000000000000000000000..b35c1b568fa09a3c7e72c4ebe5aa0ebb3f4aab26 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2AgyRHY4xug.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b2de374eca90dd18fdb5b6bcbf06a9c3de5b89e1295ce55a2c1caac64252fa02 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Aq4v7VEFrM.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Aq4v7VEFrM.pt new file mode 100644 index 0000000000000000000000000000000000000000..0797fefb0e8cef8a3215e023b6bcfa1d1941e744 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Aq4v7VEFrM.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c93113e31adb323d57975f8341187f25f0156b49011f198d98b41bd39830c966 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Ay6S-8-ToE.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Ay6S-8-ToE.pt new file mode 100644 index 0000000000000000000000000000000000000000..c355c9bdee7ac33efa0491df93dc843ac7be5cc9 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Ay6S-8-ToE.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ea0f9ccd8b1d79fe501ca2d320c717c71c87f8d3f33f7c8a25af04617a27b1cc +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2B-Y0y8kawg.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2B-Y0y8kawg.pt new file mode 100644 index 0000000000000000000000000000000000000000..76894ff3d8dd74f446c0e649960325f39fbf3f7f --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2B-Y0y8kawg.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6b3aebb762e3fcd9130d202abfaf1bb27224e4e9f7611827bd1c924a3cd229b1 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2B22081hU9s.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2B22081hU9s.pt new file mode 100644 index 0000000000000000000000000000000000000000..233f14add7df93d94d06d7dcfcc01801943d5070 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2B22081hU9s.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cef1170c0bf4c8b493b22aaccd8ea0375c4821331c26503056f144173df40443 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BAHqVb_PMc.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BAHqVb_PMc.pt new file mode 100644 index 0000000000000000000000000000000000000000..4342bec28097932b2d96f9902da3280d8a6f00dd --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BAHqVb_PMc.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d1c8d5a162a2dda47c17e492f80fae045d32f68fa2a150568963cef8b369503b +size 17828168 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BBIcmxwgUw.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BBIcmxwgUw.pt new file mode 100644 index 0000000000000000000000000000000000000000..c51c865ec8f4cfcb23eacf74e171785fb01c6061 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BBIcmxwgUw.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5905c471c66376fce733f014fb19ea084b6aca4800310fb4822d163dfc5e5060 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BCAVIcBOiI.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BCAVIcBOiI.pt new file mode 100644 index 0000000000000000000000000000000000000000..aefb5f5d00be9baa8c5368d6612b22ef9e2b3746 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BCAVIcBOiI.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e3b6905ce06f5217717e6b9b80d4ba935aee9ba7d9d33312fc522cd5fda855d8 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BE3Re_2o3w.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BE3Re_2o3w.pt new file mode 100644 index 0000000000000000000000000000000000000000..52fbbe287ece8aae79d57e7cca6857881971b652 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BE3Re_2o3w.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5d6b3f8fc1b360109d3bcff5789ab86d5cdbcf0613a43cd9b2f8a17d2d9deac2 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BEdrsbFHjo.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BEdrsbFHjo.pt new file mode 100644 index 0000000000000000000000000000000000000000..59e39ff35fe8a07d847692b869b492d3e18ee814 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BEdrsbFHjo.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ea74b401a72dddf7fbff3caf2061a120f8e0ff0204bbcd6638c5f47d1b8ea90c +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BF-LIsbGTE.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BF-LIsbGTE.pt new file mode 100644 index 0000000000000000000000000000000000000000..78021b9b49fe60bd99ad9d749d03bfc692ec26ed --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BF-LIsbGTE.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9f3fd16d391859b72215e5a6eeaa6eb143e3d647cf5b2d1f802c533636372c69 +size 17041736 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BFmSWh2bCI.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BFmSWh2bCI.pt new file mode 100644 index 0000000000000000000000000000000000000000..1003ba099af4c04a889125ec5518139b729cd774 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BFmSWh2bCI.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cf197a0e79124756813981517cea22b751c87b1584b9f687d399ea24244d1f30 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BIOHrRUnIg.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BIOHrRUnIg.pt new file mode 100644 index 0000000000000000000000000000000000000000..468bb3476c294195b49090777759e9a89a632b5f --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BIOHrRUnIg.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7da653e21ce62af7ddf4a5db823097ea4bf622d49b4d5b9e5b8767cb62937c26 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BOifavvyFo.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BOifavvyFo.pt new file mode 100644 index 0000000000000000000000000000000000000000..11a41ec2c70f2f84935d4c1a8ce1405b5d46f71c --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BOifavvyFo.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:00a838aa16a46656d1b065174276f730dd050e5d600cdc9f117d7f4253d1113f +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BSbnUUgfP4.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BSbnUUgfP4.pt new file mode 100644 index 0000000000000000000000000000000000000000..d63d1743ed223bdc33040895a58c3dbde7814129 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BSbnUUgfP4.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f24a1f86616066c2dee681587e0990be5456d8f7983c6d859d7336a7bc597e35 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BYRRiU200E.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BYRRiU200E.pt new file mode 100644 index 0000000000000000000000000000000000000000..9833caf661ad93e5189cf73f282f7f49f2050e3c --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BYRRiU200E.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cbd6959a9811a6cb9f4029c0320e68ce1a138cc3481fe8239ddfcf691784e8f4 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2B_BYTFIidg.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2B_BYTFIidg.pt new file mode 100644 index 0000000000000000000000000000000000000000..896b606d5800eb752e65e3809098581c9a029055 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2B_BYTFIidg.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a12a725ded5b936124f25dfd3966dc778c0e7a2904e97a21100952884ac0b780 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BdCdcbCd0o.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BdCdcbCd0o.pt new file mode 100644 index 0000000000000000000000000000000000000000..238d1c1fa355b5c8cd73feaa41a867f78bd629a5 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BdCdcbCd0o.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1941f73cbc0a2436681a398b5a4e3047360b4787c816d76e930c3aff1a05dd8b +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BetBK0D2vY.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BetBK0D2vY.pt new file mode 100644 index 0000000000000000000000000000000000000000..63d477e4b7731c6e45958a448295c48f5cfc2536 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BetBK0D2vY.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:de355cd50abe9d4052c7fcfc1c0dc20489e2bd1c20c0fe7ace5b4eeafd72a434 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Bi3BU0UTsg.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Bi3BU0UTsg.pt new file mode 100644 index 0000000000000000000000000000000000000000..c043e3a23b978dda6d4a1dbe8f60ea6eaf259f26 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Bi3BU0UTsg.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a28be9595f0e68d75a19855d1300c4036c53509ebe2ed7665c52aaceb7956f12 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BkBSMHoLsY.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BkBSMHoLsY.pt new file mode 100644 index 0000000000000000000000000000000000000000..8d8e786995bd67bb0ee6754585fea81048cdac6b --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BkBSMHoLsY.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c728b75373500d8049ac355bb02a715d28546c19c8d23576f2353ab1d54719b3 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BmY2jIblPU.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BmY2jIblPU.pt new file mode 100644 index 0000000000000000000000000000000000000000..cd831a95da7bea11d2a8c37d2d5f7074ee93d15e --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BmY2jIblPU.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a62c18467926c278618eee3182986342acd4ddeba13521cf6ced3796d79e0856 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Bnb_2kFM34.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Bnb_2kFM34.pt new file mode 100644 index 0000000000000000000000000000000000000000..4f1df1851724606ced044b9341bf48806204347a --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Bnb_2kFM34.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2b199e6fe5c9b5316817784e9781a1f6d96a28d3ae398442312a49895cf36d11 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Boh8FViBRo.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Boh8FViBRo.pt new file mode 100644 index 0000000000000000000000000000000000000000..1edc168539c6e4dbd3426c3665a720ab5ff34032 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Boh8FViBRo.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:39a3f8299733c92cc44fc2c147f9ac4b78da336f9f8e6cfe5f381815997745b0 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BpIir_29qU.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BpIir_29qU.pt new file mode 100644 index 0000000000000000000000000000000000000000..841a94c9daf711d75e7eccc330d18114fad9fe26 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BpIir_29qU.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b81929c333e7517bbf8f1294fc092518f23766d275dadd642a6f10511efc5f38 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BrFrU87d1I.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BrFrU87d1I.pt new file mode 100644 index 0000000000000000000000000000000000000000..c8495ea5f31d4a0070d66f778af44e0e53ce15e9 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BrFrU87d1I.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:aa9a7cbdca5054b17044ff04464172abe4556ecae457f26dfe93777984912301 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BtjUMx0y2I.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BtjUMx0y2I.pt new file mode 100644 index 0000000000000000000000000000000000000000..837c32738271f00b9550c5a11711d9af76e1ef40 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BtjUMx0y2I.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e64341a77bb86b9bbce0b4a90bef85fa0faabae0133bc084dc149b38c19a03a8 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Bwg60swtfk.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Bwg60swtfk.pt new file mode 100644 index 0000000000000000000000000000000000000000..0dbbef9924ad77afc8a45478a393c8c7b0e16e94 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Bwg60swtfk.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0c74703fd90a03533f4e557f2615bfc83b0015c7809a6ca91056acd7c2cd3caa +size 20056392 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BwmNrSYtkM.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BwmNrSYtkM.pt new file mode 100644 index 0000000000000000000000000000000000000000..f5446aa53e67872fe4e9e398c6e4543cfadc05bd --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2BwmNrSYtkM.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:da22d7320b2ee32b74586560603fa42974c234ebd855f09f009bbb37430eae3a +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2C6Q_MfzdP0.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2C6Q_MfzdP0.pt new file mode 100644 index 0000000000000000000000000000000000000000..efab927a3ce9d7b7acb3159f8388324a67e3ce9e --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2C6Q_MfzdP0.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:88d84f79d9ffd04b4058874f058b7bdabbe0140f5bacae65873f80042a7940b6 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2C6Srco5tCU.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2C6Srco5tCU.pt new file mode 100644 index 0000000000000000000000000000000000000000..c18d1d09d35e239ef84da579c4d76fb2b7d82262 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2C6Srco5tCU.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:60f82d399e690a532b16ded6922d72cb6aef55a50e2191648ba1d6626e78f58f +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2CD3K6IkZpA.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2CD3K6IkZpA.pt new file mode 100644 index 0000000000000000000000000000000000000000..e46e9812b54558341964a8d5bc6eb7a45663d9c4 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2CD3K6IkZpA.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7ec052ed61ba5e99a44dd9358bb382470b1c0cb06d55983ee1df10a67e651872 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2CF2FOygX98.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2CF2FOygX98.pt new file mode 100644 index 0000000000000000000000000000000000000000..fde1b50bd389a092868f6e532494d68987a9f079 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2CF2FOygX98.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3f3697c5c1e09c72f1b8fb322fa2bf8fe37139f5a2e3cb3a9fc9ad6f27ebb27d +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2CJvsKt09F0.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2CJvsKt09F0.pt new file mode 100644 index 0000000000000000000000000000000000000000..36a2e92841725831b3146b1a2b013ed09d1883e7 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2CJvsKt09F0.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:aad59bc93f85d6b163cfb0ebc0fd4a8d630d88c4e9d0cc794c37821ed50249ef +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2COwAJP-Na4.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2COwAJP-Na4.pt new file mode 100644 index 0000000000000000000000000000000000000000..e721a57e640a7a6a2c58a838ef770716575a32a0 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2COwAJP-Na4.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:92b289beaf6234adcabdb180ca8a0fd1ef0e123944adc901faf6c7aeae7600d1 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2CVURQdFILk.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2CVURQdFILk.pt new file mode 100644 index 0000000000000000000000000000000000000000..0d08608a05ba9cc1867b6a50651be976adbc6edc --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2CVURQdFILk.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5a368ddb345b10ec97f712abef8428ed4642df0987036014055ef2a976890aa9 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2CWA0weNSpk.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2CWA0weNSpk.pt new file mode 100644 index 0000000000000000000000000000000000000000..fd963f2b8bc5132081bdb135338cec5c596db14b --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2CWA0weNSpk.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f86459e1ffacb66bd6b5c9aec987d9ecd9de997a2c03a5108f349ac265717a38 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2CWcFRUN0tM.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2CWcFRUN0tM.pt new file mode 100644 index 0000000000000000000000000000000000000000..bb0e8e893f542aa0534fd7d3ef1947fa5fefbe45 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2CWcFRUN0tM.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4844db5969b6bb5b93ed3cf0b338f521462d0fa8e96e432659d512daf2f2089e +size 17041736 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2CXlv__igdQ.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2CXlv__igdQ.pt new file mode 100644 index 0000000000000000000000000000000000000000..3581f31366ca194d5248f193f42763b9622b7250 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2CXlv__igdQ.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5034fea946c7ebbc7961cfd3170da7a3dbbfddc4530584bb6c95a739b60ce576 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2CfesigSsA0.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2CfesigSsA0.pt new file mode 100644 index 0000000000000000000000000000000000000000..14ddf5801b56aec3a843534695fd121d9764188a --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2CfesigSsA0.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d4a8a69384878976d0764b84609e6b3d7f77ddcf63c70edb5027f2e172206b56 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Cyu5MJ-JZk.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Cyu5MJ-JZk.pt new file mode 100644 index 0000000000000000000000000000000000000000..4a5f76a014a92d1434eeef09bfd190400290d014 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Cyu5MJ-JZk.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c5e0e27fee45e1dd5642117442ff74636fdcfb3f66616683305bc68c464d835a +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2D6Jazj-zlQ.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2D6Jazj-zlQ.pt new file mode 100644 index 0000000000000000000000000000000000000000..af0b2755b960e7627f64177fded2ed82bc723090 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2D6Jazj-zlQ.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ed675ae73ea8d22b3f5f913621246433a8e64ce77f5d9368c3a8504113f2329e +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2DGga-uIsZ8.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2DGga-uIsZ8.pt new file mode 100644 index 0000000000000000000000000000000000000000..a2279d57d3219cda6e4f2e2e37122c5d90b33e6f --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2DGga-uIsZ8.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c9463c4fe9cb8802a5000caadbab13288b56dfa1145ce8ea09141c18cdd84aa0 +size 20056392 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2DQVLq8KWfo.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2DQVLq8KWfo.pt new file mode 100644 index 0000000000000000000000000000000000000000..2197d1473f08258b5079dedd7a844a809785df7b --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2DQVLq8KWfo.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:22bed245ed260d44a8411af25593809c52044d78576ed62669363100e2b06603 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2DTSXr0oVNU.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2DTSXr0oVNU.pt new file mode 100644 index 0000000000000000000000000000000000000000..ac7f7a46e77d7203f605eaa5ddbb403c799cc952 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2DTSXr0oVNU.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6c0cac95c22dfab53a1aee1325063223ba82262dd74c107203585a3487adf5b3 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2DWK4SFmFR0.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2DWK4SFmFR0.pt new file mode 100644 index 0000000000000000000000000000000000000000..6f14793235c8287e016ae806acd69d7007c369db --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2DWK4SFmFR0.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b49e947796ea263d8ab8ad75806c6813bbc7fdefd661e45ed65c728f9cbc0367 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2DZkqZySZc8.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2DZkqZySZc8.pt new file mode 100644 index 0000000000000000000000000000000000000000..ebb425f49fa70fbf8eed2d76d369771ef084cf76 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2DZkqZySZc8.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f828014c4bc6a9a7b512ad17f41c0fdfc2faeab014168942ce0e7eabc30477b0 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2DbDd9X47b8.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2DbDd9X47b8.pt new file mode 100644 index 0000000000000000000000000000000000000000..28e0eafed4a3138c24c183169afe4123f91cd8f7 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2DbDd9X47b8.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:883cecff59ff04748e94bd18ad82eceb4e413b56698fefb12229047c15734dc5 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Dd7Ua-Vvik.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Dd7Ua-Vvik.pt new file mode 100644 index 0000000000000000000000000000000000000000..a27f4a8ada5f0d2a265fbe084a294d9ecd9397b7 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Dd7Ua-Vvik.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a577a3d0f9f37a6706411fb9daf1584f7c7f851059f3421cf1e1aa7b04687caf +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2DslZKZbDOE.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2DslZKZbDOE.pt new file mode 100644 index 0000000000000000000000000000000000000000..030f26b2563d81b56941905bfac3e19910fcde02 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2DslZKZbDOE.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:888a9a3b0569cd77c67ad486cc168423f393861115bffcd26292d7014d043585 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2DvBnDHRyEU.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2DvBnDHRyEU.pt new file mode 100644 index 0000000000000000000000000000000000000000..7b89f6df5f794f34847c4aba099c0c3910abee5b --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2DvBnDHRyEU.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d4aac9dd24a3a554e2f81505d9aa095d3c5de5ca2a05a06a0c9c982d5a286f57 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2DvK6-ZvE04.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2DvK6-ZvE04.pt new file mode 100644 index 0000000000000000000000000000000000000000..a1a70d72e429f00ca499f887e57ee6a78adecea0 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2DvK6-ZvE04.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1ac8c5eed655fa93addfdef34b83c109bab036adb7c28bf8d5296cf0d9b34d4a +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Dxwk2Q5ETs.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Dxwk2Q5ETs.pt new file mode 100644 index 0000000000000000000000000000000000000000..55ce6be47d9bf703cd00722f9e24459ad5fb34a4 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Dxwk2Q5ETs.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:16f7c15113223dcd5b5b76ae2a103ad72f634b6c38b1948e3a0dc6563ddef026 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2E0W9p_LHUo.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2E0W9p_LHUo.pt new file mode 100644 index 0000000000000000000000000000000000000000..f5c08693ac505578233d05c3feda0303ab79adae --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2E0W9p_LHUo.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b6063fb8b19f09692d1b5da5de8136d741f01dfad8fa0dec56969c0745520667 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2E1eYG5UoUA.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2E1eYG5UoUA.pt new file mode 100644 index 0000000000000000000000000000000000000000..50127ed2acd420affedc32f827e343eb1d3d5a60 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2E1eYG5UoUA.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a42448f4a378d16d342b3af78422b63fd8dfb6cb8949be6d20a537b217274141 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2E4aJ-Zpmv0.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2E4aJ-Zpmv0.pt new file mode 100644 index 0000000000000000000000000000000000000000..013282cad1822e5f50bdd5878d0594f09a3975aa --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2E4aJ-Zpmv0.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2b23e9adf71b938adcc19b21c219483b9f23ccf1946f86f33ca0525fb0c3f4c6 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2E6-wdow-aQ.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2E6-wdow-aQ.pt new file mode 100644 index 0000000000000000000000000000000000000000..e93fdafcfdf8789a50fda9488f2f695c7a1037ea --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2E6-wdow-aQ.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1684591690d27ec0eefdeab16ba1cce2db00b6a26393cfaf2a74635b66d91666 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2E6VxKdmq7Y.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2E6VxKdmq7Y.pt new file mode 100644 index 0000000000000000000000000000000000000000..47a2d9b91defd311083f00ffed45da3c3e1b6ec0 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2E6VxKdmq7Y.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4744430bfc393e955557df8a69e032709f4f6a15d522e6203f975bdd11b74a4b +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2E7nUzxpSBY.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2E7nUzxpSBY.pt new file mode 100644 index 0000000000000000000000000000000000000000..a1dd08dd0f2cb9d236b1c129813928811270b951 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2E7nUzxpSBY.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:46308cf8d5e804ba3c8ec80aefa2f4a4c94e297b2a829d5330ca2b6afa116792 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2EABJVJjvcs.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2EABJVJjvcs.pt new file mode 100644 index 0000000000000000000000000000000000000000..13d447ef53c0e3a747975c37b51d3d43e72ebe6f --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2EABJVJjvcs.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9ff361202e6035917a8f9df26dd6874549fecab305fd023926e13e44ec448e58 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2EEgqUNr4zc.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2EEgqUNr4zc.pt new file mode 100644 index 0000000000000000000000000000000000000000..d1f80dd73eaa7a66c773d16d9c1beb5a9bea8d31 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2EEgqUNr4zc.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2ca463504943f349d58e7e8fc08eb4e15315b3e4f33c927468fbefb9b430c08e +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2EOWcOsXQEA.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2EOWcOsXQEA.pt new file mode 100644 index 0000000000000000000000000000000000000000..a6a935b222afb50e3e9c433416e1c9cf5190bd55 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2EOWcOsXQEA.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fe28b5038c72efd6c60bddbc4c10f6ac2c1e292af6798d3dba4fe94f84b9d47b +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2EYHOFd5KG0.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2EYHOFd5KG0.pt new file mode 100644 index 0000000000000000000000000000000000000000..093700cd4cb3a785316902693e23591726e580ca --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2EYHOFd5KG0.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8aee25ce24bfb2bd9a966e6244e043963c2b5d1bfb351de44c0bb7db9d3278ff +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2EYdGbW_3Dw.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2EYdGbW_3Dw.pt new file mode 100644 index 0000000000000000000000000000000000000000..6c384adb195f498fc543e6a1c60425de97453f16 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2EYdGbW_3Dw.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:773b1aac58b594f0587b74d5f430a0c848382531f5cac224aeba0c4ccc78b33b +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Eba2ZWRJJw.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Eba2ZWRJJw.pt new file mode 100644 index 0000000000000000000000000000000000000000..9de318a0b9b1a70fc852de79013f48f38c59a5d4 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Eba2ZWRJJw.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4ef9ae2b0fc9b3134d83c812069d1c8cefcf158727f1c6a6e13e3d39f23b12cb +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2EbvXMGA1qk.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2EbvXMGA1qk.pt new file mode 100644 index 0000000000000000000000000000000000000000..5aafd77dc139e1db340660b068d04a369a2ac1f5 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2EbvXMGA1qk.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e1c8c056d0dcac54f196655910ffd8edd98889529cc6fa4c324683c0be381e2d +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2EddTaA3-sg.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2EddTaA3-sg.pt new file mode 100644 index 0000000000000000000000000000000000000000..8f30f9f662e6be758b18963bd2528e37bc42a6f6 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2EddTaA3-sg.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:02de852abd8db231c9b667fa14fe0dd022a631ef8800b434b58bb5b80474cad9 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Eglw6OlnUc.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Eglw6OlnUc.pt new file mode 100644 index 0000000000000000000000000000000000000000..3eaea3e0df187a5951bc5691640c9889612cb3a9 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Eglw6OlnUc.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9674a88970f4285d5db324d764749ba38e6d5435401edfb6d217d5d7165c8012 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2EjY-7xvGuM.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2EjY-7xvGuM.pt new file mode 100644 index 0000000000000000000000000000000000000000..36073e31d85ee3753f3a71d143e46dab80305f1f --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2EjY-7xvGuM.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:32200c010145d3502a50936e30185e75255f28b3d077fd0bcfad42da95763ef5 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2EjYUy0bzt8.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2EjYUy0bzt8.pt new file mode 100644 index 0000000000000000000000000000000000000000..fa0fe95989aed8f29390088c55f5dcd1dd146155 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2EjYUy0bzt8.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3c7877b34458b60bb6d6442ebcfcc1547b37d1cea1f42a3f2c01f1327284b83c +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2EsoYXMkaCU.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2EsoYXMkaCU.pt new file mode 100644 index 0000000000000000000000000000000000000000..1c29d2cecc4de1ba299071cdd48d90c8cd8d56d3 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2EsoYXMkaCU.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0b109aee66ff3839de071fe9418a769d88517d684ce9c0e3c4a45cad2dbabcae +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Eud-8ksGE4.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Eud-8ksGE4.pt new file mode 100644 index 0000000000000000000000000000000000000000..7aa9c7bc5430cdef36863e7bd1ff61c2c30b98a5 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Eud-8ksGE4.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:274fb0a4d886d287045537a83d6242cbc46e0beabdf4e64c9f642909cb18e735 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2FPmCAIuAxo.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2FPmCAIuAxo.pt new file mode 100644 index 0000000000000000000000000000000000000000..2b757745501240eab60c0cb54e609b75ea9bee39 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2FPmCAIuAxo.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f3175a61d191e36d0540015b63b12772d298b0fd4277957e5e4c6a4816494a76 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2FTfX4OWK6g.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2FTfX4OWK6g.pt new file mode 100644 index 0000000000000000000000000000000000000000..9ae28ffcb971b26c2b798b34386d3e5bc5a1f671 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2FTfX4OWK6g.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:002952600801c6bd501b18f17cf84baa30ffad9ad62065a1acd22c201b8a4660 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2FTzSa5KAyw.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2FTzSa5KAyw.pt new file mode 100644 index 0000000000000000000000000000000000000000..eb7d44dbd844e1ead0f47cb46fc9a03186e46529 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2FTzSa5KAyw.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9186a0273325dce217c93a12315c1b44e783ad5cedf98a4ca621c762e664481d +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2F_x3chk9_w.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2F_x3chk9_w.pt new file mode 100644 index 0000000000000000000000000000000000000000..eca9c24d60f7ff0c99c6fe3e56ea6371a2fbdab8 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2F_x3chk9_w.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:43ffcd215547e6f7437fe047d6599aa11392d3abf9f04d85357dd04fe50a1740 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2FaRMhA70RQ.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2FaRMhA70RQ.pt new file mode 100644 index 0000000000000000000000000000000000000000..91c951f9d86b27efcdcb37d503ceafcc0630d39f --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2FaRMhA70RQ.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b84f3d12a525b54ca19b7a649ac02e2f12d1de8a8313551d3c5667e3b6e9e089 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2FdIWLGeR2o.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2FdIWLGeR2o.pt new file mode 100644 index 0000000000000000000000000000000000000000..29a2ee520a3561439281def6149b413184a640e5 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2FdIWLGeR2o.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bd24d1ca567df89239c799341837693bdbd7b2ef4581e8e787dbed7089392a15 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2FgcNjOdXgA.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2FgcNjOdXgA.pt new file mode 100644 index 0000000000000000000000000000000000000000..624f5d38591675f3ef8ce2501bf4c5471337d218 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2FgcNjOdXgA.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b03966917defe61a887501914702528e3985f62dbf84bb4bc40ee89918e27cc6 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Fmy6xZlvB8.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Fmy6xZlvB8.pt new file mode 100644 index 0000000000000000000000000000000000000000..be31c82a5060e30f652a0cf03adb2d3b08a9635f --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Fmy6xZlvB8.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:14b622bfc2839b7b5e9800666c2e925447fea23ffa5139cefd585321173e9f68 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Fth30oQI9U.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Fth30oQI9U.pt new file mode 100644 index 0000000000000000000000000000000000000000..ca69785e46fbc64bda684fd8eacc097b24c79efe --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Fth30oQI9U.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6a70880a39d6880e67ac96886f5dcf4b83a859dedafd17cb023aaa8db17ad28c +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Ftz9n9A33c.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Ftz9n9A33c.pt new file mode 100644 index 0000000000000000000000000000000000000000..d6c5e844c8e179227afa703e3952c3b73bc78cef --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Ftz9n9A33c.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:689c19ec5014e1f6aae1e7f901ef101cdfa7a122d043643a065e12ee1b6c6d3a +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2FuyA92fg44.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2FuyA92fg44.pt new file mode 100644 index 0000000000000000000000000000000000000000..a8d315bc05b1b3a0faea328c000586b2b68b91be --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2FuyA92fg44.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f291e7297a66bf1fdc21ec6bed9abf31fbbc1331a1bec0662bd0b57c00a9c09b +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2G40CDxdoPU.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2G40CDxdoPU.pt new file mode 100644 index 0000000000000000000000000000000000000000..a96550c6c20d421851ffff6cae5d2e8eb55dbeee --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2G40CDxdoPU.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:abdc10730dcea30971dedf67716545a264f833d58c2f12ce16f434ba12f85825 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2G4HAbjW1pg.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2G4HAbjW1pg.pt new file mode 100644 index 0000000000000000000000000000000000000000..0bcb59d2647bc2fa9f64171fcf27f2fae8845f02 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2G4HAbjW1pg.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fb2fdd0c320d6226cddd1568dcba1804a0a8d6866e35c9566f88911a4085c6b4 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2G7vudRsiB8.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2G7vudRsiB8.pt new file mode 100644 index 0000000000000000000000000000000000000000..848d0538ee0189fed75ad84047bfc8d91b63a6c9 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2G7vudRsiB8.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:42476ef9fa9dab7f2b79d76688f570440d6946d2ea3012d31d9902d140ff791a +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2GEfqpALqIE.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2GEfqpALqIE.pt new file mode 100644 index 0000000000000000000000000000000000000000..57d9c75e0162f920bcf9a6817cc8baa66bb53440 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2GEfqpALqIE.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:03e9082b31cf82ad5e4b278d7bf27e74c32062d24720bda09b1158d7e98b80a7 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2GEtXqbYjNc.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2GEtXqbYjNc.pt new file mode 100644 index 0000000000000000000000000000000000000000..c28609c6783eaa74662292398fd6e7dc14e537bf --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2GEtXqbYjNc.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:23de46939b5fd0404d093d533833694679df26749e48b504614809bfa3f193dc +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2GJHBRQWfWs.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2GJHBRQWfWs.pt new file mode 100644 index 0000000000000000000000000000000000000000..08ef3b837fe31dbcc85e2af9e311c96a24c03a39 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2GJHBRQWfWs.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4c6c15b52009bcc6c81d6ddc9935f57d6c8d65623c80fff26b533edf8cad0213 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2GNLe7zPaHM.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2GNLe7zPaHM.pt new file mode 100644 index 0000000000000000000000000000000000000000..5bef6d6e644a484b8318dd1856503745769e79d0 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2GNLe7zPaHM.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:315156aa00c5b05efbe8bda69cf67b831814ce71a317b5b94dfba5dfb970f37e +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2GN_LRDClsM.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2GN_LRDClsM.pt new file mode 100644 index 0000000000000000000000000000000000000000..e218e029c5f6db9d78ee06de889663c5b0c13390 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2GN_LRDClsM.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ab8c31e66d0bdecd5c56cd41bc273a49450d7ad9e5147d6b2b3afca2125d95b4 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2GVWyRT2GgE.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2GVWyRT2GgE.pt new file mode 100644 index 0000000000000000000000000000000000000000..bb4f72907e222adfc9ca1f0b75c38029bf83767c --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2GVWyRT2GgE.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cd9048a48dc9d7f1152a20c45803fad26306a3a5e91f0e3bda37e03d0cee1d42 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2GWN79-zG9c.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2GWN79-zG9c.pt new file mode 100644 index 0000000000000000000000000000000000000000..960a6600d513ed70d774d9bc0eb1115666b249ff --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2GWN79-zG9c.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3ed07173f15a298944e0b462318a7228dc750b0bba899f5063ffca4fe1575e67 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2GYCjgMBF6A.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2GYCjgMBF6A.pt new file mode 100644 index 0000000000000000000000000000000000000000..64eae538d52ff8bf188e20b53ec09310a3d632a0 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2GYCjgMBF6A.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d75bf9383da326ef4d4b465c7d04f3c1b7b8af23f938c060152b850c070f91cf +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Ga5JnwBOzE.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Ga5JnwBOzE.pt new file mode 100644 index 0000000000000000000000000000000000000000..31e156045d57faa094f962417349d594c4500a29 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Ga5JnwBOzE.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e46e0a7ae662af0440a845e19cb6218419995580f28f22086176fed592796a06 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Gb4rGoAe5M.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Gb4rGoAe5M.pt new file mode 100644 index 0000000000000000000000000000000000000000..752976c986507a2a21dd9720844fe8ed2777a510 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Gb4rGoAe5M.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9101fd73c7d1f1fe503c94199150d6046c41fb7256f840ecf292d79c47636a4e +size 19663176 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2GdO7TJhK1Q.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2GdO7TJhK1Q.pt new file mode 100644 index 0000000000000000000000000000000000000000..c94cbe44992a7f186bf51cacab7b89aaa0d27fc7 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2GdO7TJhK1Q.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f81f903c24994e68a77f9b923bb54eac9e45f803c6cb979d035bd1852a31878e +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2GhFifLhxBI.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2GhFifLhxBI.pt new file mode 100644 index 0000000000000000000000000000000000000000..b43317b8f655d63cfee47a72dc3f814649b93555 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2GhFifLhxBI.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1b6b7ad130ebce16fc0f521464e832d48147e74b896f8b8e5912b422d4563dce +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2GhZT0gfSpc.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2GhZT0gfSpc.pt new file mode 100644 index 0000000000000000000000000000000000000000..fd7a07e8a7ce9c4c1fd1385bb834a387f0f6bf18 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2GhZT0gfSpc.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:90f4d250bb6ad469c3a9401821ce998cce02863e0108dcfb6b82ddcca9e1492b +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2GkmgmjjVSc.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2GkmgmjjVSc.pt new file mode 100644 index 0000000000000000000000000000000000000000..72caf1ed729ec851b818d8eab1a04cb9c13f8186 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2GkmgmjjVSc.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b7671215fc539742e2a3ac130015f09f4d11634f064cf1995269af7f7b4b7691 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2GmcP4nWf5w.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2GmcP4nWf5w.pt new file mode 100644 index 0000000000000000000000000000000000000000..f27e766e7a5d1e7c52bf203b7e3d8d2ffc51315f --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2GmcP4nWf5w.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dc9e7367fa4ae0cf646b02b3bc8f5928c5763a5a3095e7c059674de48dc882c4 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Guyw-JM_Cg.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Guyw-JM_Cg.pt new file mode 100644 index 0000000000000000000000000000000000000000..1e61b53c02da29f53e4947f12462604a1f4adb2a --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Guyw-JM_Cg.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:861c93ce1f21755eeb08661f8dab1d34a7f9220f21a0ebfc0418aec49efe69b5 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2HACvL-BXpc.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2HACvL-BXpc.pt new file mode 100644 index 0000000000000000000000000000000000000000..3518c5ee4d9b9a60f74acc080d38038417da96fb --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2HACvL-BXpc.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:78ce1d14e879806eb046519eab67624664a759abe495ee02fc90d09bd060c502 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2HEUIKJix3M.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2HEUIKJix3M.pt new file mode 100644 index 0000000000000000000000000000000000000000..54d4465689967a158352e341b0be598ddfade123 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2HEUIKJix3M.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a5f9ab83236c2ce13f4541c419ebebcc81d0099a3abff664a84244d1f468b800 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2HKOTRCTX7k.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2HKOTRCTX7k.pt new file mode 100644 index 0000000000000000000000000000000000000000..01e576c0ca767371bc0fc958b534a98e2ca82d98 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2HKOTRCTX7k.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0b98a61b9d363557d593e98f74fc25e730cd14da6e47a2dd736ad02fb3928f8e +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2HNAOsPsrdw.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2HNAOsPsrdw.pt new file mode 100644 index 0000000000000000000000000000000000000000..2c6b8151b92223b839c23d95209de930e785e119 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2HNAOsPsrdw.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:28b915b5912d6acf424b6b3ffbcf4e2c41da69c0c1aa39f584a67aa7ef296774 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2HOrYUN5IGU.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2HOrYUN5IGU.pt new file mode 100644 index 0000000000000000000000000000000000000000..478817b368492f52e16e8fe1f8639576c04ff761 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2HOrYUN5IGU.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:029d511c29507d9c872b47cd491427566dca540cdef4f3ceafb6cccc55f7007c +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2HSHQM8JoiU.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2HSHQM8JoiU.pt new file mode 100644 index 0000000000000000000000000000000000000000..509c719b229278550f7381fd03a39568e74d74ed --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2HSHQM8JoiU.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6e919958a7077fb054b06e7a11b72456b6f73e464f857941fb85a3e969204e14 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2HTqILqRnHA.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2HTqILqRnHA.pt new file mode 100644 index 0000000000000000000000000000000000000000..556f1ff78689dd6f5f09aff84ec304174b9b9403 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2HTqILqRnHA.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8ac9661625c57e63bca5652de676e546d522954be20346f9c76899ea4df67c7e +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2HUTpNOeLnk.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2HUTpNOeLnk.pt new file mode 100644 index 0000000000000000000000000000000000000000..9c7e766cc35522ddf13e6e5b18eeb6c82642a9c6 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2HUTpNOeLnk.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:17c8460e9a2f3bee9e2b2a6923e41fe417ea380f88e384cace39882292b20d64 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2HYaZMRX_dU.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2HYaZMRX_dU.pt new file mode 100644 index 0000000000000000000000000000000000000000..a2b8852a24f57789ba0597f9a93ba9102ece6a84 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2HYaZMRX_dU.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4e075496d4d28447ea116925a7987f86051143499c2030821e782e1153769770 +size 19663176 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2HddOBemzZE.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2HddOBemzZE.pt new file mode 100644 index 0000000000000000000000000000000000000000..5ac57d54197a6aaed97c52cc5cef04fd0ad0821e --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2HddOBemzZE.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:27f49fc4befbc3b074c65d743df1a2eace908a9b6979f406ed15ceb3c1d44524 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2HhdLYbPp1Q.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2HhdLYbPp1Q.pt new file mode 100644 index 0000000000000000000000000000000000000000..d8fe56677c1dc5976d2c9f27cb8dab97ab7d1d82 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2HhdLYbPp1Q.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1353fa73a1ee6f2849353298da5ec800921cf42733791726bfe817f36d886500 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2HkPglnbheM.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2HkPglnbheM.pt new file mode 100644 index 0000000000000000000000000000000000000000..b6d8f878a0dbb7671f86633c607870d8b388f069 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2HkPglnbheM.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cc79728a801176b12e938c53d536c71cb0b453dbe641ba5e9a591055d468fc2f +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Hqs2M1nV6g.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Hqs2M1nV6g.pt new file mode 100644 index 0000000000000000000000000000000000000000..1fc9442001d6b3c7124e489077beb8df4644f469 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Hqs2M1nV6g.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5391dfc69b62fada6e3eab59dcc711cf595bbb96f370bf197b0fc8d5e616f042 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2HrcE7-oSoA.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2HrcE7-oSoA.pt new file mode 100644 index 0000000000000000000000000000000000000000..815aa75347680161fd9a23e703010263a9d6fdb5 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2HrcE7-oSoA.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1206e8a0bcecf7f8b341e9670499410525029cec9d5cfba9f8bff62964dc49c3 +size 18352456 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Hvrc7hrSVk.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Hvrc7hrSVk.pt new file mode 100644 index 0000000000000000000000000000000000000000..9be91d55eb4dcc4e9d8d010ad045168445bebbbe --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Hvrc7hrSVk.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:09d2703170f6a979e04c87539f89195d1542067fe9240aa29fdfebf72d4b7b98 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Hyr2z1nHN4.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Hyr2z1nHN4.pt new file mode 100644 index 0000000000000000000000000000000000000000..2e018598708cf9defaae39cb72e1faa8176b54a8 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Hyr2z1nHN4.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6b0961ad142a570260bfa7543189e2b14b5a85657faa7065859b44f6ea8c321f +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2I4NaCHANlY.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2I4NaCHANlY.pt new file mode 100644 index 0000000000000000000000000000000000000000..1fde8a6a576988629e0261d00a6bce099984b073 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2I4NaCHANlY.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cd4082e58021a79cc85ada3a2a846ec03f00bdbb1d3faee295322b5d60e74512 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2IDFmxtIDAk.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2IDFmxtIDAk.pt new file mode 100644 index 0000000000000000000000000000000000000000..1a2342255c163a5e0f0c95465b06daddb5ef0bc8 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2IDFmxtIDAk.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:df7e2306219d23422236a213ff40788ddeda45ea7a93bce69f36009e89b84b72 +size 17697096 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2IIr_tvrbNI.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2IIr_tvrbNI.pt new file mode 100644 index 0000000000000000000000000000000000000000..6ac42a12d5af5f3b3e94adb36582f413c6dde172 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2IIr_tvrbNI.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7b5d9f9d29f79499f39cb39798719e5ac96eaac443d317ff4193fd0ae569480a +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2IJGv6odCTk.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2IJGv6odCTk.pt new file mode 100644 index 0000000000000000000000000000000000000000..b42483a1f6cf5d382249758320b07cd2782cf98a --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2IJGv6odCTk.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:67de58acf2cca73c56c30b303b0b7afbaf7a9a4676c9e82d1a8fff6464d78053 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2IT8Kkr5P3A.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2IT8Kkr5P3A.pt new file mode 100644 index 0000000000000000000000000000000000000000..17d5b68e2da6a9c1f625498ce9d6d2320928952e --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2IT8Kkr5P3A.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a6a3c625e91ab037498581d277fdeb2a09c98f4a5e292d5ffd1a2d053db01d7e +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2ITySxuNx0w.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2ITySxuNx0w.pt new file mode 100644 index 0000000000000000000000000000000000000000..fea2c95c3906b7db040fadef366d5804b59dc289 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2ITySxuNx0w.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3b01852dec570c2c8d702563b67c2101b98d2e0de2c60d2084b05f6aa0af9c53 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2IUL1ozTMt8.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2IUL1ozTMt8.pt new file mode 100644 index 0000000000000000000000000000000000000000..11fb00eab4ecfd5acdcac026c6c152ff31610ff5 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2IUL1ozTMt8.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ad8ecdc0ea4f27db453351f5d75f1ed07d3c6f1fdb090eae11f9514ae0101e5f +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2IVMYamXQtM.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2IVMYamXQtM.pt new file mode 100644 index 0000000000000000000000000000000000000000..268cd00575c56d5c84351538157e20a78cc94494 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2IVMYamXQtM.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5cc3db62cd2cce41ba9274c6c4b83cffa453fb19f1a11b398776a38ea92559e4 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2IdyD4eWKU0.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2IdyD4eWKU0.pt new file mode 100644 index 0000000000000000000000000000000000000000..d13d14c635c709e863a7a4685b0050c61576127e --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2IdyD4eWKU0.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0fd4e3ef81fc37b6258bd0b07c58a018ddc2772da701a55205ac5deed70abcf2 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2IxrfBj--dU.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2IxrfBj--dU.pt new file mode 100644 index 0000000000000000000000000000000000000000..efb77f664b8f7867e78244be830550ed6be8424b --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2IxrfBj--dU.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9a06e06b0fc999722dfc088afa103062fa91d68348a0b79ddd41d4823ec3c3ec +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2J0McIz2Zp4.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2J0McIz2Zp4.pt new file mode 100644 index 0000000000000000000000000000000000000000..b1ba6aa656eb92f5ac57f9ca3b0167faf5ce61a8 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2J0McIz2Zp4.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0297a3c0e8b4002254d0c8c11bfab2686b5f313286d9f965baa4e885959e1d19 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2J1DJ1OGBEk.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2J1DJ1OGBEk.pt new file mode 100644 index 0000000000000000000000000000000000000000..a2bb226447b8e9d42dc93bd6c8a1e33436fb7a98 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2J1DJ1OGBEk.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5f6a2505ad667ed3ebb2da9cf653704623addceac735b3e706b4625d2fa99de0 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2J6hF3aHyxs.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2J6hF3aHyxs.pt new file mode 100644 index 0000000000000000000000000000000000000000..cd4b1219fa36c4b2a211c763936dc28c2cef109c --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2J6hF3aHyxs.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6afdfb582950ff557ba38f8095f0bba39cbcd328bf71da9d84be25d10b22846d +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2J8B68HLgME.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2J8B68HLgME.pt new file mode 100644 index 0000000000000000000000000000000000000000..270f91bb1ff99eb16e18519ca371abd134cc2d03 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2J8B68HLgME.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8cda8065161672a1bd71524ddeb91379473db1a74b17f6318701ed44a0042fe8 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2J8jZNDVEzU.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2J8jZNDVEzU.pt new file mode 100644 index 0000000000000000000000000000000000000000..8c85966e4d3a9909b28c1aebcbae6eea92a7727d --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2J8jZNDVEzU.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1e7f87b47016581c2af2e247487f89262dc8f5259a7211af6da8f243ed838a7e +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2JIrMD9mQzs.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2JIrMD9mQzs.pt new file mode 100644 index 0000000000000000000000000000000000000000..29372d142562fe9934f2f40573c9d8dcaa42d2a1 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2JIrMD9mQzs.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5f6dbd775f42f2a21a6d5a06fdce6e4632f53c404f9fa21035b58c20ccb71f25 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2JIwhGACbdA.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2JIwhGACbdA.pt new file mode 100644 index 0000000000000000000000000000000000000000..81898a2832969748d5983c2ea81c7e3f7cc26a06 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2JIwhGACbdA.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5a6f413abbbeef2e5a4ffbb9925677fe8d0b13def46ccb6105aacfccd21139c2 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2JLt2Wf6jNY.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2JLt2Wf6jNY.pt new file mode 100644 index 0000000000000000000000000000000000000000..9761c6a5e4b73819b5a1525dca9f6b2e38e79a66 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2JLt2Wf6jNY.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:103758de2dca44639abf86957327f0915fe8e21f73a07ac370d521d60819b8fd +size 17041736 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2JOTdp69l-E.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2JOTdp69l-E.pt new file mode 100644 index 0000000000000000000000000000000000000000..0956005d9d53340666c3fe064632a21c082e8c82 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2JOTdp69l-E.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7f2e09a43696f20ba2a4cf6807b1d7faf98a2b9cc8d1f7f2e76d5fa69ba26cd4 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2JTvHBbegco.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2JTvHBbegco.pt new file mode 100644 index 0000000000000000000000000000000000000000..dc5aa1be75f8b07cfb58785cbf273e8086a7f6ed --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2JTvHBbegco.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:addacb013df6704563a6c078341f8d4eca4f7c61573bedf021399b91e32e7d40 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2JWGNN8-KA4.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2JWGNN8-KA4.pt new file mode 100644 index 0000000000000000000000000000000000000000..61955a0287b7e0852d7c5ba73028f50d8ef88da7 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2JWGNN8-KA4.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ffe7d9c5a39ba5fc165fcf11e3e849c53591a834727d61fc3aa038eaa820caff +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2JXYQs2EjOw.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2JXYQs2EjOw.pt new file mode 100644 index 0000000000000000000000000000000000000000..ab157358d1c1c1ba16c55b334c88025704e1fb18 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2JXYQs2EjOw.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1a82fdb908b5f34f9a7ab7d30c455fe470f892e077fe676b86ba60e3688be41f +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2JXZUXudKsA.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2JXZUXudKsA.pt new file mode 100644 index 0000000000000000000000000000000000000000..b1c6ec67db5fc3161df721852cc9a9a1959833ef --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2JXZUXudKsA.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e9fc2d88170a0af32d50c4b329a1e3ff35b9fe35892c331ae4063a5ddede0e4c +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2JgcYiVSBbE.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2JgcYiVSBbE.pt new file mode 100644 index 0000000000000000000000000000000000000000..17543ab726c19473ea3988f10e550a80a949adac --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2JgcYiVSBbE.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5029a17f0ff6e6f80e1d212800bcfae420e889964380b22409f5034640c6536f +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2JjzWr4ZDRo.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2JjzWr4ZDRo.pt new file mode 100644 index 0000000000000000000000000000000000000000..81436a0022d721b36917094e0ca4f49df9053350 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2JjzWr4ZDRo.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f07247e83801498a37e0f2ca7d6f321d56b857a403d23790783f4634225b60da +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2JmqLdVPqTE.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2JmqLdVPqTE.pt new file mode 100644 index 0000000000000000000000000000000000000000..5b4dd9ec73c8f8a80da69b945801dfd851c6f2a5 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2JmqLdVPqTE.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:696556717297c042705e10d01392f10a43d9ee5b454ce9d3c55d6f0151b8c38a +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Js1Svp8aJk.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Js1Svp8aJk.pt new file mode 100644 index 0000000000000000000000000000000000000000..907f8ffb766bd75b710c0b1c0802f804fce0d2e0 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Js1Svp8aJk.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a92208bcf4a117898f868dc67bb2bdf0acc66b2f2d89de8db0eb5590c62f3d98 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Jsq25WuOOs.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Jsq25WuOOs.pt new file mode 100644 index 0000000000000000000000000000000000000000..7f5e1bbc920be370a7637ac42cf2b5fb50a59261 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Jsq25WuOOs.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:201aff860acbe3b88b2cb1625ed9cad0cb0a0a9096240d47405e6fea7452dc1f +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2JwB4BSBPZg.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2JwB4BSBPZg.pt new file mode 100644 index 0000000000000000000000000000000000000000..6eeb3a12eff914d3254040ba89428c79ce468380 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2JwB4BSBPZg.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4548f3cfc393cee3f4844efb9e66cde673e823ea041e1418063e7d05592743f8 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2K26TIp2DJ4.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2K26TIp2DJ4.pt new file mode 100644 index 0000000000000000000000000000000000000000..cfd4cb6bafa4e4f45d2bf3520841d2546e5015f7 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2K26TIp2DJ4.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3397977d755d22241801ff32cf89875b81ec7637a0e1e4b42859430c062dd111 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2K2gP7hAFXU.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2K2gP7hAFXU.pt new file mode 100644 index 0000000000000000000000000000000000000000..c5a737b32eb118c842432f626ec3915ac5fbdc1b --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2K2gP7hAFXU.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:644f55d59018cb30c94957db8991fb4ed7140e9b3d24ee799aff9486b9f31e3f +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2KDahjNJOpg.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2KDahjNJOpg.pt new file mode 100644 index 0000000000000000000000000000000000000000..57fdd80c2976ad3c4ad4f69911cc5b1ce7af6ca7 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2KDahjNJOpg.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:506bc88384a48e4599eed40aa44a198a1ff40ee618f88a58771281384802b009 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Kbt_fgi614.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Kbt_fgi614.pt new file mode 100644 index 0000000000000000000000000000000000000000..1cfe6b9d3fb40fb49039e1d88ea3e93cdda72567 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Kbt_fgi614.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:717553deb087a597f1633145684f67dc530ec842401ee7c1094b0815abe12847 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Kk9tlQjPRI.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Kk9tlQjPRI.pt new file mode 100644 index 0000000000000000000000000000000000000000..8b6d634af016bc37521c73f854d241d41aa172ac --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Kk9tlQjPRI.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:612ca2fb181289998ea9e532519380df204a6a38e2009a0b6ad1e8439c13cd7b +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2KkpNPKGKv0.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2KkpNPKGKv0.pt new file mode 100644 index 0000000000000000000000000000000000000000..e0bf74393d763dc08ee48d49f95bb283946bcede --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2KkpNPKGKv0.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:30b360b9702a0c6ed8261a2cfd8f1699f4e1381772583558ffecb84db2849014 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Knra9hsJjc.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Knra9hsJjc.pt new file mode 100644 index 0000000000000000000000000000000000000000..c1a827f743677483fe8324a5f12fedc95de58a74 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Knra9hsJjc.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e4279854f2436391b21ca40980e601b4aefbbb0b4b53ef5bc7517b5f0714bb21 +size 17303880 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2KqoESqE_1A.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2KqoESqE_1A.pt new file mode 100644 index 0000000000000000000000000000000000000000..791554b9fb30ef4a8c3a5ee1300c0ad2924c49d0 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2KqoESqE_1A.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3a49111481a52239de4382ef46659ae9f88ad2c2528beec038d1bc86dcbfe36c +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2KsGQWq8vaU.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2KsGQWq8vaU.pt new file mode 100644 index 0000000000000000000000000000000000000000..7ad1e0c36f58736f7e5f856dd5bd44fb37d4bf43 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2KsGQWq8vaU.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9a37c50b7930ac75a8171fd2bf194ac58bff9d175eaed3c57ca9990ee25a8432 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2KwQJjsjN-A.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2KwQJjsjN-A.pt new file mode 100644 index 0000000000000000000000000000000000000000..332a435bfb0ac142cb70f1189278fb0897e9e820 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2KwQJjsjN-A.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:12a1fdd391db0f5d3d38de4c9fcd61adf955301121c37e87ee8d990a8df245b8 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2L5665vzHXA.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2L5665vzHXA.pt new file mode 100644 index 0000000000000000000000000000000000000000..b54b76b59e2f90cb9d56e8c464c1fb2dfa9dae16 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2L5665vzHXA.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1ddf1cf64e57cdcb3849f256d828c8c667d523a22e7151dccf67c8fbf3a4df56 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2LDxq9CNUH4.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2LDxq9CNUH4.pt new file mode 100644 index 0000000000000000000000000000000000000000..3c9f1a3a9f24387c449bacf276d4f5fe24271b34 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2LDxq9CNUH4.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4350fa0ffa226117ed5e8511e557fa7c16a1cb371c1d9c7cbc1fdefcb69d08a9 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2LEo4KltqU8.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2LEo4KltqU8.pt new file mode 100644 index 0000000000000000000000000000000000000000..2e6404f68a81ac31e62c5f500f5d1213e9dd2065 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2LEo4KltqU8.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:17177ee70c79b57b794c3a20c32d31684324ecf3a60b2bf37e7cea8261bf5f35 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2LF2l2A2a-g.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2LF2l2A2a-g.pt new file mode 100644 index 0000000000000000000000000000000000000000..45a689d723ad153916a47fba53fe7278a04fe230 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2LF2l2A2a-g.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:92dc023c078bd33eadbee3a8325b0fb9a418f77c2b6b9154814dcfdd7cc3a59e +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2LFGYdNnmdw.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2LFGYdNnmdw.pt new file mode 100644 index 0000000000000000000000000000000000000000..2a4e710da044a43daa8b298d42a214e17500525f --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2LFGYdNnmdw.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d64f2d680d17019ab3b7d7ddfe042fd45d409160d38daac3c940e2ab6705e8a9 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2LUh3PX11Ak.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2LUh3PX11Ak.pt new file mode 100644 index 0000000000000000000000000000000000000000..bb7bd60345c7fbd1c75a5a6fd1d073063e3bbf36 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2LUh3PX11Ak.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b0123f92d38d7d50c76ca6a0d9dc23c6491844b91390c218ef7dbff3062f9f03 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2LVa-TzRdQ0.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2LVa-TzRdQ0.pt new file mode 100644 index 0000000000000000000000000000000000000000..cc3e619246d4cb040f9dd994ff024af7b7257c43 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2LVa-TzRdQ0.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8a94accaaea88d137b51fd269fe5393d9831092d8750dc466585fea7080c0127 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2LVm0rd02bg.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2LVm0rd02bg.pt new file mode 100644 index 0000000000000000000000000000000000000000..808ce1acc7835825a8df2e601f44e10e9438b8a4 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2LVm0rd02bg.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cfbcfa11ed0e1bd24194995284e6a28e52f72612886fb5855f2c411b24a962c9 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2LZSnUIcxDQ.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2LZSnUIcxDQ.pt new file mode 100644 index 0000000000000000000000000000000000000000..25f2a7d84f6d9a13c6a3da11d9abf2d7d4ec4853 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2LZSnUIcxDQ.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d55d201d622182e3435eedde4247d83369301bf4c75c333d2c020259f99b8e52 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Lcf6KAhQqI.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Lcf6KAhQqI.pt new file mode 100644 index 0000000000000000000000000000000000000000..6b055edd4d614ed18e9db7e13cdd08aaeffb2f67 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Lcf6KAhQqI.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:febc9d75ce08b141939d63767de3fe479bf2b64cbc5c86e41f89dfd24431c18f +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Ljf-D361Hs.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Ljf-D361Hs.pt new file mode 100644 index 0000000000000000000000000000000000000000..c0360a8831f7b6fb24e9e70cef2741e13d1ec0ab --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Ljf-D361Hs.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e1a3aea42f8d5fa9f220240cc2748818753e9761201904161848757db839cab1 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Luex5Ho_ac.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Luex5Ho_ac.pt new file mode 100644 index 0000000000000000000000000000000000000000..792cd88b2814c0d260951ac9621e4d26b3bef294 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Luex5Ho_ac.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3320865ca36d0d3e05c4530b1a068827da6e554432ecf5e53d617eef68f705ee +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2LvJ3UDXkvE.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2LvJ3UDXkvE.pt new file mode 100644 index 0000000000000000000000000000000000000000..53b7952d8709f38b2485621a198e6284396eb9be --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2LvJ3UDXkvE.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a59c1ea1459c1139a83cff04b70e35d2f5899debf990862fd489555b712896b7 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2LvWFA02VmQ.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2LvWFA02VmQ.pt new file mode 100644 index 0000000000000000000000000000000000000000..e1fa41379c947839e6269ec49c5b0dfb128e8f5d --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2LvWFA02VmQ.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5192b2ab8f0f5121feef429c43f4599bdc527b05bf063de3bf242c20647f43ae +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2LwPnsHAqFs.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2LwPnsHAqFs.pt new file mode 100644 index 0000000000000000000000000000000000000000..2461eb31e813baf86dc0fdafc180dcd40a8a21ee --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2LwPnsHAqFs.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:afc03b65f3b342edfeae3eb9b49ff717cdb9192d1cbe3d593779b724327a5f5d +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Lwb9qXCfJE.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Lwb9qXCfJE.pt new file mode 100644 index 0000000000000000000000000000000000000000..c8035f4d4b6e08f312cd0bd3f9eb8a9c62712a3d --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Lwb9qXCfJE.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c42b1f1cfe248acd730e5e777ed62d90adf8e0450b3abd911b9fc7ea0be574a5 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2M14DJfnlno.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2M14DJfnlno.pt new file mode 100644 index 0000000000000000000000000000000000000000..b96b69c110793fbe2b3cfefbad6770bd5e26ef59 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2M14DJfnlno.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f5986b988e3171ddad13f5e0e3048aa0b9ec6032e6c24325e521dc56294f420e +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2M7BBXNnj0s.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2M7BBXNnj0s.pt new file mode 100644 index 0000000000000000000000000000000000000000..a300164ae73f281985c4081a9eb650bb9713280b --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2M7BBXNnj0s.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6120f4d2a4e8f95c866495943ad09f22df0d343925fc67ffc6d546be4eac4211 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2M8yhj5vF04.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2M8yhj5vF04.pt new file mode 100644 index 0000000000000000000000000000000000000000..1d466161c18aaa084ebe8cdd01935f971fb91bb9 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2M8yhj5vF04.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7e750cba1c39aab089b36717427b7ad7c2f21e8f51365fe2b45c947541ab81a6 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2M9TrJXOnnw.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2M9TrJXOnnw.pt new file mode 100644 index 0000000000000000000000000000000000000000..0a46cb7f2d92ea58b0a8370cbcbd3aaa7f610f6d --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2M9TrJXOnnw.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:47c9dfbbcba3b858aae78523328702f366cea8fad778405720e45c74e42913e4 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2MCuMCIkq9g.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2MCuMCIkq9g.pt new file mode 100644 index 0000000000000000000000000000000000000000..0ea4d8b33f520f95f8e9cda1f07e694ffdca8881 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2MCuMCIkq9g.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cb4bb24a00289ca5ed4131925dad70a9758283f97c51f4cf3fb8295c1c044117 +size 18352456 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2MIGvfld4HE.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2MIGvfld4HE.pt new file mode 100644 index 0000000000000000000000000000000000000000..8936d297b31165a1b4883648ca87e8d7937976d4 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2MIGvfld4HE.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:895d05a42f8a077c1c06d52d36331e4167833aa2bbf1ea740d348f105bba0a47 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2MMxDUknegE.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2MMxDUknegE.pt new file mode 100644 index 0000000000000000000000000000000000000000..c91fa682ccfe6b2a126baefa1c18f38628e0466a --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2MMxDUknegE.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fb6a4481183a5aa4a1820640420a09fefc45cabd533d64f31a1c9cfc0699aa21 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2MQehIh2x3Q.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2MQehIh2x3Q.pt new file mode 100644 index 0000000000000000000000000000000000000000..04f6548fcdad42ee7504cd1a3ea70199dd4be4c6 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2MQehIh2x3Q.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bd2c2ae71b6a32b9dde295661eb56ad02e2e8b22519390fd4ae48b66e54bbf39 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2MTT8N7lVr0.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2MTT8N7lVr0.pt new file mode 100644 index 0000000000000000000000000000000000000000..0c47bc8efdf30854cfc07192465597a4bc7738b2 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2MTT8N7lVr0.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4df56843f28fd343e2aa0a85a08a34a2b49720538aaba0f6b6d4d31eac644a11 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2MV4zuxohlk.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2MV4zuxohlk.pt new file mode 100644 index 0000000000000000000000000000000000000000..b6fec651f4fa69723e1123ad38853c684f1ad492 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2MV4zuxohlk.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0b114e28adcbb8f74d60d6302dfe21395b0aee52cf686c835ebf8a5b9975b854 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2MdhBmFfl_M.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2MdhBmFfl_M.pt new file mode 100644 index 0000000000000000000000000000000000000000..e7e752281e4715652aa8df17bcc43c71334e4e44 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2MdhBmFfl_M.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:548b60a627de34e8363a4501f90a071a9ffe7833aeab0e68c9ce72ac3d5bd12e +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2MfMw6UQ-iU.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2MfMw6UQ-iU.pt new file mode 100644 index 0000000000000000000000000000000000000000..2128a634516fe3b3de094cb95d615881f37e1b75 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2MfMw6UQ-iU.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e98c411c5558919d2eb803e12d2e1e9167ede084197e2c35c92a21bac50ea686 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Mfb-iSiy_o.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Mfb-iSiy_o.pt new file mode 100644 index 0000000000000000000000000000000000000000..e2a257fbc812c0d4a00ad9fdc1e00582b9af174d --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Mfb-iSiy_o.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1bc5014304783197857182c195cfe989c289626332ea9ff63df5843c27ede2c5 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2MzTiHWBGiA.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2MzTiHWBGiA.pt new file mode 100644 index 0000000000000000000000000000000000000000..4aa03758343b3fc3404acdb93f394f8c8718b765 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2MzTiHWBGiA.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e1ccc98db4652da07b914d5b8d962fbe7f080a1a502b5432eed60494bc9e4f4c +size 18352456 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2N1i6FVA44Y.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2N1i6FVA44Y.pt new file mode 100644 index 0000000000000000000000000000000000000000..b424765b3babfece6dd51250a6980dd4203df3f0 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2N1i6FVA44Y.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:76b0e790bd243b2032f8f3d8306548f40ced769e734ce576ce9c8eb6550ee909 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2N4DpttuYoE.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2N4DpttuYoE.pt new file mode 100644 index 0000000000000000000000000000000000000000..999151c8b6c17ecc6edce49ccce34a9ad798f7b8 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2N4DpttuYoE.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:25db1ae91f4e4d7206268ac859774407553720320baeee68aa17581ba7664603 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2N6STtW8n5Q.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2N6STtW8n5Q.pt new file mode 100644 index 0000000000000000000000000000000000000000..669bebfe51553f7352459b8a7d26e7c365599b85 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2N6STtW8n5Q.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9299a14d03d97833d3a0c01dc604edc5436c90c5dca357d455d44e8ac48a9d04 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2NEHWunj5i4.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2NEHWunj5i4.pt new file mode 100644 index 0000000000000000000000000000000000000000..a761d44d57e675ebbf0be0a1a944975b76a2b83b --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2NEHWunj5i4.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6a82bd7bb1f22cf06adec4a199356a93aea51b8897af2c880c32e17e88a772f7 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2NK6eOnUnmk.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2NK6eOnUnmk.pt new file mode 100644 index 0000000000000000000000000000000000000000..715f9a407d45b6403e8d631029fbfd80ff4f477b --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2NK6eOnUnmk.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cf44a94299180f5b1c436a44a74871dadcc7f82ab2f2b90808131f9e6ec0413a +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2NLDCDSwAFI.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2NLDCDSwAFI.pt new file mode 100644 index 0000000000000000000000000000000000000000..fea2fe008685929d820e09ed06dae8f5b10401ab --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2NLDCDSwAFI.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2cde888304aa54454bdb6075d63d25e78e746243b55732b5a4f33fdd0c2d38bf +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2NPcKimxBvM.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2NPcKimxBvM.pt new file mode 100644 index 0000000000000000000000000000000000000000..43c4fcff309755174e231f1c2c3b198ed7906c8b --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2NPcKimxBvM.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bea8e2355686589aa18283e41c608373caed151ba5f42ad76ba3c354cf1e73ec +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2NQ2-QxSocg.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2NQ2-QxSocg.pt new file mode 100644 index 0000000000000000000000000000000000000000..7b85196fce8b0ce0fcf8dab8a086a8624fe26c64 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2NQ2-QxSocg.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c33fbfe3297921b8bf1b10714f699e94502e3d9adcd30e596f7c5ab095daf4ef +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2NQkFvVr1UY.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2NQkFvVr1UY.pt new file mode 100644 index 0000000000000000000000000000000000000000..e239ffee0a13f0f819075e4bee38860d12597d44 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2NQkFvVr1UY.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bc36ade08249d0896b9266ec9c2f3d4278c0940d6989373f614fc9a3348d7e6a +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2NXpcLjgsyg.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2NXpcLjgsyg.pt new file mode 100644 index 0000000000000000000000000000000000000000..b5994a358562d37802c4655368055574da2b1532 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2NXpcLjgsyg.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fb4235daacc86077eaea896235a3cd339fe18c60ad7854c0306e4e81ce4d076b +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2NilN3nv2BI.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2NilN3nv2BI.pt new file mode 100644 index 0000000000000000000000000000000000000000..a89d90f961aa4fb5a767a427cde440bac54e6035 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2NilN3nv2BI.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:84a14a770f9edbe8ffb07772e0bf86816159d1f9002aca2fff5bf06a6a0e3b41 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2NpNE7sRwPg.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2NpNE7sRwPg.pt new file mode 100644 index 0000000000000000000000000000000000000000..2728cb7c0a3eda9d8234a91244e5c1019f2edc56 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2NpNE7sRwPg.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cdd7a61be0d26d7f34a5c0f3a6a4f3cf61b1e30c81dffe01d294ddecf6864db0 +size 18352456 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2NsYg2K6px8.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2NsYg2K6px8.pt new file mode 100644 index 0000000000000000000000000000000000000000..0530deb4465c7fdbcccdf601fdd808dc29227a08 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2NsYg2K6px8.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5779b3f097742c57693c53c977b8f6dac3311265f2c696062d2beae72a977264 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2NwmFbBof6s.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2NwmFbBof6s.pt new file mode 100644 index 0000000000000000000000000000000000000000..d3132613d6479cc58cfc988a631332d017925585 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2NwmFbBof6s.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:45e65f8721cd2f8380d6c604bfd8404bb311739edb957f29eec78e0d25fd460a +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Nz7FvsIhOI.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Nz7FvsIhOI.pt new file mode 100644 index 0000000000000000000000000000000000000000..72f227cce67b4810767ad5e75c052c4567b5ea8b --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Nz7FvsIhOI.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5e0846367d82aa7c84775a596821647a6171905bad70b657b0bd141e150a339b +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2OQYaTIGBts.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2OQYaTIGBts.pt new file mode 100644 index 0000000000000000000000000000000000000000..08aa017a7b1b30e1e6e9ae58fd798b7d9c9efb14 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2OQYaTIGBts.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9e2087da287e714333485081a8aec59804b3a601705cf60d62d4619e07d097d0 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2OXlwUPK7e8.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2OXlwUPK7e8.pt new file mode 100644 index 0000000000000000000000000000000000000000..a8bfec41bcb5f884b254ec1665c4533bb550b671 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2OXlwUPK7e8.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:48c5f5e0ec5c250cc43d5b35dbcf983c981704f9974587c3a957d623d8f58972 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2OY-iA1Ckso.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2OY-iA1Ckso.pt new file mode 100644 index 0000000000000000000000000000000000000000..e788e1cce84aa876553263d7b5737732796b3e12 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2OY-iA1Ckso.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:42cf89c108bf54ac13b24cec727267ac903808326c55e812eb67aaf188ced93d +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2OcLybYOQ9Q.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2OcLybYOQ9Q.pt new file mode 100644 index 0000000000000000000000000000000000000000..a88e6a8e937b4dd39ad3e71aab2e489a5a2d4100 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2OcLybYOQ9Q.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:da170b4c1123873deae62cf0e91d49c843bb5018cf5fe754635600301ea296d8 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2OeS1gAuLk4.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2OeS1gAuLk4.pt new file mode 100644 index 0000000000000000000000000000000000000000..f01f116362c7b60e4774db6a2b39e9a256f8c511 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2OeS1gAuLk4.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:def83628f2a34ceb26a236059aeaf974ca578da5ddd2729cab3db79db98860ed +size 17697096 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2OjKhklVsJ0.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2OjKhklVsJ0.pt new file mode 100644 index 0000000000000000000000000000000000000000..a9e11b94c768d266ae5c9d1d0aa94fc650aea916 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2OjKhklVsJ0.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:223aaf65987a902ebf04d4c1e42c33d1ad5d2bc401c306c2b8ddc0533acd721e +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Ok8H1dbYdQ.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Ok8H1dbYdQ.pt new file mode 100644 index 0000000000000000000000000000000000000000..b808878973fff121b6f6285fd7c098739c504eaa --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Ok8H1dbYdQ.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8e5541bedebb4343213b04c18d16ff005c32c02c49408a9836416a8d9846537f +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2OlTCgmt-u0.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2OlTCgmt-u0.pt new file mode 100644 index 0000000000000000000000000000000000000000..e5bb6f6c6f5b10e0702ee2c7d35576dbf2b7aec6 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2OlTCgmt-u0.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:81c913e84c86ee78b23fef92d5bfff915d123639c6df4403bcbb59fe2be7ef3d +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2OptAm-lDUU.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2OptAm-lDUU.pt new file mode 100644 index 0000000000000000000000000000000000000000..7b88eb938db3d5fe90fe9c0fd8562a02c6c15259 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2OptAm-lDUU.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ebc113274783f4e7217952ee351f6236c2aab347fc5d3f6e042038d95323ccf1 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2OrHXZB2wcM.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2OrHXZB2wcM.pt new file mode 100644 index 0000000000000000000000000000000000000000..c1eb0239a540e2af274b5b93861bc94ad4a79bf4 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2OrHXZB2wcM.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:461aa9c2b956152c186e8d15893fe7b772bef964905824f09ce25dfdd8445dd4 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Ot1CKGOsvg.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Ot1CKGOsvg.pt new file mode 100644 index 0000000000000000000000000000000000000000..ac21d6f0f85daa39a97a342c6f897c63f12c576e --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Ot1CKGOsvg.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:82ac3bea4ec4a89cd55764b6b20be4217c35c9a8a216a5dc2fa159dbf0e27edb +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2OuMiZW1gZs.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2OuMiZW1gZs.pt new file mode 100644 index 0000000000000000000000000000000000000000..c273411f5de9e5dd325c830a4f886de4de35ca59 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2OuMiZW1gZs.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b8805d1513ae1313ca6b9d73b6f5a8c2dd66ef02655045f347686b076ff66a70 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2OwjsS_MCJU.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2OwjsS_MCJU.pt new file mode 100644 index 0000000000000000000000000000000000000000..b437f5055142a91a5385ff21718f32d9043f256d --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2OwjsS_MCJU.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a75203be8fd9130bb95df7b32d0295dd192e1367e2262762cc61bea6b8e7df38 +size 17041736 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2OyYya7hBBA.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2OyYya7hBBA.pt new file mode 100644 index 0000000000000000000000000000000000000000..f2320cbad5035de6935d260f64bcfae10680799d --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2OyYya7hBBA.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3f4edd1fb8301452326b429bdd269ee9b12055553f336ee54461d0771f6e9b96 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Oyy8IZW4NY.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Oyy8IZW4NY.pt new file mode 100644 index 0000000000000000000000000000000000000000..537d947bc57810eafd6e5e0a997b8ef4477297a1 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Oyy8IZW4NY.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1ac980fc190adb1d43fef341b10ad212b0a8a32c2082adbac3a92c9f8318603b +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2P2rtuk5WJU.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2P2rtuk5WJU.pt new file mode 100644 index 0000000000000000000000000000000000000000..8b74cd9386d0728308f467248561b0030878ec45 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2P2rtuk5WJU.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7c6f5dcf68af8cf844426870aea1255398d445b548f6897d27393516b2f5c429 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2P4pAtzlp6Q.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2P4pAtzlp6Q.pt new file mode 100644 index 0000000000000000000000000000000000000000..cf6c0ef9ed40d3c28836bbc9e5e2f36026506267 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2P4pAtzlp6Q.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ef7e016280663ae25ba9099db66d18cc43b1d694f26160c4f440fd50f719b2ab +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2PDxdpiYQMQ.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2PDxdpiYQMQ.pt new file mode 100644 index 0000000000000000000000000000000000000000..2a18a6f2269d72b00ed97578c579303c059f5e75 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2PDxdpiYQMQ.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3457e5c7bacfd2d1782ba138367ca2586efe7bcdc9762814b091702450a0d494 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2PELy8cUGOA.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2PELy8cUGOA.pt new file mode 100644 index 0000000000000000000000000000000000000000..3854a82e4a967c70de93dd2540153c96f8e51bbd --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2PELy8cUGOA.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ac04fd3685d49fc871696ade001eea963a0d56197772b5807a0def03868b8040 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2PKXLTSKXHk.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2PKXLTSKXHk.pt new file mode 100644 index 0000000000000000000000000000000000000000..52158cd5b4c988c79198d2fb5b5ec68e921e8ee7 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2PKXLTSKXHk.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4727b07bec3748036f1ba06425a58d4dfb3cd3c500848518b85d991222cbaa73 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2PM-TTP_FKA.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2PM-TTP_FKA.pt new file mode 100644 index 0000000000000000000000000000000000000000..879c0a70db7783d5fa478152f5c176578f77e982 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2PM-TTP_FKA.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4c8c99e1d5281b10c7aadc773a0565da8c093c323b1f12159c940a1c5f867777 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2PO2hUKL8_4.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2PO2hUKL8_4.pt new file mode 100644 index 0000000000000000000000000000000000000000..6733a9229550ea6dda3eda1b883cbb191dda7bf5 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2PO2hUKL8_4.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:317a962a9b6d283df07d611c1065ee59c89390e1506269a86180dedc0ab19a3a +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2PQP_cJikyI.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2PQP_cJikyI.pt new file mode 100644 index 0000000000000000000000000000000000000000..0cec3336c59b4cd917017efdfef26237c72c25bd --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2PQP_cJikyI.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:393c23849571818b13f8312e8978cc85fda6b5821e09fb696e8c7bc0ed709746 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2PWi3I6IIcM.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2PWi3I6IIcM.pt new file mode 100644 index 0000000000000000000000000000000000000000..37bcdf2ffb041bfd4199cbcde4399c1578fcd96f --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2PWi3I6IIcM.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2a54435b6ad5f0e660120240a2cbb9c2c5864dc95e4245ad067e082af2c6ec37 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Pa6ZjXEPG4.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Pa6ZjXEPG4.pt new file mode 100644 index 0000000000000000000000000000000000000000..ed0f0775793146dae071012e44854ff2e623843d --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Pa6ZjXEPG4.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0fc34b7cd23e18bbc7344abc5548c186832d915c72f71f268fc3bac081299e2c +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2PhfqsgIKfA.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2PhfqsgIKfA.pt new file mode 100644 index 0000000000000000000000000000000000000000..3e7f7d5a837bba4c619067d4ec75c1e99f0224f1 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2PhfqsgIKfA.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6bd28d344059cff087c5fbfebbe070dfad7162c436ccc12f109e54497c808bf2 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2PqETZCjJY0.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2PqETZCjJY0.pt new file mode 100644 index 0000000000000000000000000000000000000000..d42b25afc1f72fd902169e1046ebb1324e94eed0 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2PqETZCjJY0.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e2f18aa39d967121dad618a1fa9041e2b90ad574240d1a33ddee6fbe49e297a7 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Ps8HLoq1C0.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Ps8HLoq1C0.pt new file mode 100644 index 0000000000000000000000000000000000000000..8f9e7994629f947cfcf4e1e33fd441690e93fc18 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Ps8HLoq1C0.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0094a6874f29d44c5de9d785e9ee58e0b0f873dbfeb36e2d3e9cd28cdefbed25 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Q1E7yM1NAI.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Q1E7yM1NAI.pt new file mode 100644 index 0000000000000000000000000000000000000000..cbc44e9d1ea644bb6449a43ba7d0e516e69a28a4 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Q1E7yM1NAI.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:42b25ada481cfdca3a88e4fe8b8b2207e080e33848bd1568a2e57edd2df2c788 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Q1EB0SZOeM.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Q1EB0SZOeM.pt new file mode 100644 index 0000000000000000000000000000000000000000..04cf6afef797473f8774174c9e7a7bc3d6b0d271 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Q1EB0SZOeM.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:69ebc0c006e4712dad607a89cf446874a6c176cc8f88961c73789cada7ab6812 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Q481kYvCNk.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Q481kYvCNk.pt new file mode 100644 index 0000000000000000000000000000000000000000..5b7f4ea609d7b5f7f29a7c168eeeab98e9f62b3e --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Q481kYvCNk.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dd33e914e0ceb829f87b714f0993a9d0212d0a59cf5eaf90b6ca4a7cdd395f34 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Q4i8oAEKqk.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Q4i8oAEKqk.pt new file mode 100644 index 0000000000000000000000000000000000000000..b8b49596d137b835f1c92f388a38288dfe5c49e0 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Q4i8oAEKqk.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c6bbd53177de9fcbe06a8b7bca82348c24bed544ce9ac91dafdfc8ad2d8472ec +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Q7Jx91hdGQ.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Q7Jx91hdGQ.pt new file mode 100644 index 0000000000000000000000000000000000000000..1dfc683c64f2db1eafe2b7cb3d3f91a2cff091c4 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Q7Jx91hdGQ.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9c069a5d5b70a1413e00b96ba71dbe5aec7dbefab8e67b88162ae44b002e52d3 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Q7u78vrmdQ.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Q7u78vrmdQ.pt new file mode 100644 index 0000000000000000000000000000000000000000..9319d64fa29c577703475ee6b335f8a61e3c819b --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Q7u78vrmdQ.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3d412e15573cfc4c3b12d48c688e79976b373d619edbca23d45cbf086dc8669f +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Q8Se9BDNT0.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Q8Se9BDNT0.pt new file mode 100644 index 0000000000000000000000000000000000000000..7b13d3e8d4ee90da53d5c0bd932a985a68899bb2 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Q8Se9BDNT0.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1a8b51dfdad7b2f268c7e423e100d16f3a5d7bcc6d806197c379a1ea32d28b4f +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Q9dVNoYHps.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Q9dVNoYHps.pt new file mode 100644 index 0000000000000000000000000000000000000000..9152abbbeb03b55825905cf0ea78b15dec6c3483 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Q9dVNoYHps.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:546be5e1c054a3e73c026ca12fad01af4245d8f1cf59de13730a225f36b00c24 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2QOM-byMB2E.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2QOM-byMB2E.pt new file mode 100644 index 0000000000000000000000000000000000000000..cf0dc7398bab0dbe119c1d2d1bdaaf71a19190b8 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2QOM-byMB2E.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:975f7897e940d16473943c2cf296e91addf6754ca155d4e728ea4e837c718e74 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2QPBxQKs_Xs.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2QPBxQKs_Xs.pt new file mode 100644 index 0000000000000000000000000000000000000000..ffbd5096baae75d51fdd307b330b33dd7bed383c --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2QPBxQKs_Xs.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e16da673c26f67717a7780fc07930aa4b5273b579f6319892923b353a3aa55d9 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2QXO1MCFKx4.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2QXO1MCFKx4.pt new file mode 100644 index 0000000000000000000000000000000000000000..8e3f9b4bce345355507f29e9df688cdd07430a80 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2QXO1MCFKx4.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:024782c36d96fc7faeb0119e6b58decafd0192011a6f871b28945c8b89d56d6a +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2QZqmBeBqNk.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2QZqmBeBqNk.pt new file mode 100644 index 0000000000000000000000000000000000000000..1d034818b4eeacd3e8270f77ba81387a2820374b --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2QZqmBeBqNk.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:99ce0e4f194b72324802c4ec6d5f1c2c21296fa693e5755dc0b61a24816c4704 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Qc-YrHP3-o.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Qc-YrHP3-o.pt new file mode 100644 index 0000000000000000000000000000000000000000..51ef985a566148576fcaef14dc95c8abb54d280f --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Qc-YrHP3-o.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:113eb936eee98e995c1ec270906324af84bfefe1b1b1626df86b2144cb898c30 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Qc1Skcs-b4.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Qc1Skcs-b4.pt new file mode 100644 index 0000000000000000000000000000000000000000..c140fe759e19e5e70a0dfc632c7e73735fb50650 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Qc1Skcs-b4.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:230e136b1d49474fad194cd0c9c2b1f4c313be199e8075951e299d33102dd237 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Qc76Y6eAzY.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Qc76Y6eAzY.pt new file mode 100644 index 0000000000000000000000000000000000000000..72a546cfc053dee5249858496637273d4ca18bd3 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Qc76Y6eAzY.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ae1b154a6cf0fce928f84c71f0bfd41a6bfa7087a690dbf75537cff1e26a059b +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2QfbiVWG9K0.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2QfbiVWG9K0.pt new file mode 100644 index 0000000000000000000000000000000000000000..e5a19d959dc0cb2662852bfaa7f30423c9f5f72e --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2QfbiVWG9K0.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ee12fbd396198ffb20c72b4ae7bea3ca61521ec4e01769e0388902cf607f028e +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2QgFU_HM0wQ.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2QgFU_HM0wQ.pt new file mode 100644 index 0000000000000000000000000000000000000000..ab381fd9dffc976f069cdeaf7b9021adf3366a2b --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2QgFU_HM0wQ.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7b295a3d87f48394cefbe7cd724841c0871d0ec30ac1acf7a8b1928dd0091038 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2QslDsbRPI8.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2QslDsbRPI8.pt new file mode 100644 index 0000000000000000000000000000000000000000..75eee2a8260fc5ceb8e73bc3edca729831d181b4 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2QslDsbRPI8.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:80376382399ad14d08e472b0150cf11f113eb0d466f3ed82d18980fda13ddc37 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2QysDr9DDL8.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2QysDr9DDL8.pt new file mode 100644 index 0000000000000000000000000000000000000000..0c3e5ada2257e70ad478c516404221066cbc83e6 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2QysDr9DDL8.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0d3d8f1e5ef67176a2adcfd6e2ef6ca734132c4558b08446be326f9ea564c5c5 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Qzx_5X9OgI.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Qzx_5X9OgI.pt new file mode 100644 index 0000000000000000000000000000000000000000..3cc7c90b28a6f617858f970a3d75b1aa1977598c --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Qzx_5X9OgI.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:051398166f14015409a7f0728944aa184ecc6eeb3af9c9486ec71dd81d533578 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2R-b0UmKVcA.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2R-b0UmKVcA.pt new file mode 100644 index 0000000000000000000000000000000000000000..1a5a07e5b5637c429274283b2af04fcf25d60949 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2R-b0UmKVcA.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b58fff347f814900cd2e2628cd746979ef85c78ce19b61d34b97f45d9ff6393d +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2R0HBv3h_F0.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2R0HBv3h_F0.pt new file mode 100644 index 0000000000000000000000000000000000000000..485fc3d67c4fea07e416638b3d27ebeff08d8661 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2R0HBv3h_F0.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2a1fb9e9fe4635ecd3de22adf43b630019dc5471ade1bbc97fa6f2ae3bf2ff67 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2R1CnV4kcq0.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2R1CnV4kcq0.pt new file mode 100644 index 0000000000000000000000000000000000000000..08ddcb47ac1a84fde7f2aa0b3284f00ba2e57c00 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2R1CnV4kcq0.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dc31dcb344f90897f27787198efb625d2538c61df4fb65199b02d303bbec2357 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2R1Jtkclrng.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2R1Jtkclrng.pt new file mode 100644 index 0000000000000000000000000000000000000000..0fe776c05add15a8a0b82a755e2858414376e447 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2R1Jtkclrng.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a8c6ecb5f3303deaff8d59d78c9cf2ad30b8cf29fc1af78efa6689ae7cd4a19d +size 18352456 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2R6-wXbx9ws.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2R6-wXbx9ws.pt new file mode 100644 index 0000000000000000000000000000000000000000..b45026d429c07e09b87db666ee71f8b0f1d3ee33 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2R6-wXbx9ws.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5578bf60ab067d88ee56f614f676b15f52a874d1b9c5939a18bcd73948d18d2a +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2R7NNdw2qZU.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2R7NNdw2qZU.pt new file mode 100644 index 0000000000000000000000000000000000000000..9e61119049953358ac4bb61b8ece2b3401dac0f8 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2R7NNdw2qZU.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e26281d3aabf8fefb32d1b4119eab81fd2b870916827ae11b6dd745691bfa04f +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2R83X7FtDEs.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2R83X7FtDEs.pt new file mode 100644 index 0000000000000000000000000000000000000000..7dd5e7b60e9f6c89f215b2dfb0e6e83ecf29e462 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2R83X7FtDEs.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:797785b01260531d9c08e4c108bb53ba247ce174847aea118d3a2771c18739e1 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2R8jT_lC6_Y.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2R8jT_lC6_Y.pt new file mode 100644 index 0000000000000000000000000000000000000000..ebc8f1a87b1794c33b0ee408f54e66cbd92fe5b9 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2R8jT_lC6_Y.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6ac3db989f556c38b905af6c8b5f465ee9f8aabe394f7ad9e596037d926e33d2 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2RBc2vxdm3s.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2RBc2vxdm3s.pt new file mode 100644 index 0000000000000000000000000000000000000000..f18e74af141164d36311683c2434c60e122fb11b --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2RBc2vxdm3s.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5d839ea118e4d09a1dbfad971ce2eb01b98da91702f85361d3214dbbb0010761 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2RJVCM4sgk8.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2RJVCM4sgk8.pt new file mode 100644 index 0000000000000000000000000000000000000000..7abf5dd7630b44786ab69e5ea8459b544fe27e0f --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2RJVCM4sgk8.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:67e5505932f301a092c306a136a4898dfb82ced45a780f335fe34c8cad721c24 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2RUBMpH3JhM.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2RUBMpH3JhM.pt new file mode 100644 index 0000000000000000000000000000000000000000..e06f8f4fdb8b6f82df3f22b1f7024188f050ad6c --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2RUBMpH3JhM.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4244d75d9b405ee5e494efdac03d2a37310963124a50a48a494a7b6dab9f303a +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2R_aOdkDJZo.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2R_aOdkDJZo.pt new file mode 100644 index 0000000000000000000000000000000000000000..3e46c389194296152509f26ceb90d8594a929b46 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2R_aOdkDJZo.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:931ab2cb5cf7acfc47ba2edc1603fa755f5e9261307255ff3c3b59209ff20132 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Ra2p1hap94.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Ra2p1hap94.pt new file mode 100644 index 0000000000000000000000000000000000000000..b402b49d1252024770754d953724d57ce065e8a4 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Ra2p1hap94.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6b8103156deefa4a9624e14d7a4c4a08d79029f83418e66193a281944fb713f5 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Re5L5yGH6g.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Re5L5yGH6g.pt new file mode 100644 index 0000000000000000000000000000000000000000..63d69c03081a1d1ced5a1ff5208c66cd2a5eaaf2 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Re5L5yGH6g.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9de815b78995d97d97e3538d9587c7198d1e286626735e771967ab05069dd316 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Ri80LQAmBA.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Ri80LQAmBA.pt new file mode 100644 index 0000000000000000000000000000000000000000..6d29b3d95685664e96211d039145c4af29e20f75 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Ri80LQAmBA.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:83b9ebbee130f0019050be45cc09ed406a74097493b346dbe6d1991cc85726ce +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2RlPUQoO5Sw.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2RlPUQoO5Sw.pt new file mode 100644 index 0000000000000000000000000000000000000000..a2f0a8f7cffd0380b3555d9e497f6c671be208ce --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2RlPUQoO5Sw.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:aca9a4cd5d02fc70ca0590cfee5c70a8045903c2284a56d551dfa65900d88f56 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Rl_W6F-3nc.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Rl_W6F-3nc.pt new file mode 100644 index 0000000000000000000000000000000000000000..aa37dad7bdd6200fa454cd5111391887cbed190e --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Rl_W6F-3nc.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a9ecddf221bb191456f0463ca9eb581c1a88f41037bc7d0f30ff603f5dc13440 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2RlcDBcBTm0.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2RlcDBcBTm0.pt new file mode 100644 index 0000000000000000000000000000000000000000..b0106c87eca58565c72ed9d948426a62862da726 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2RlcDBcBTm0.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9e16273ba88f3c3ec40110ed55fab9a9d5cce0125573c85d443f33028b00e9fc +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Rrgoua-Ns0.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Rrgoua-Ns0.pt new file mode 100644 index 0000000000000000000000000000000000000000..2ab92bf795f668a93a2fecc92a249065bdf920aa --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Rrgoua-Ns0.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0a78d2c37f494553235c1a4e8068e1d145484362e3fb6f95f4d141a52d7e57a5 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2RwZbl_G1yA.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2RwZbl_G1yA.pt new file mode 100644 index 0000000000000000000000000000000000000000..84c42357ae1f9428f76f84600398c73e48785510 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2RwZbl_G1yA.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1b503db61d334dd5e236033795080aaa4bb4d8dfafd496e9755d8f0adb034593 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2SHmFOAuBCY.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2SHmFOAuBCY.pt new file mode 100644 index 0000000000000000000000000000000000000000..b7d4b196521443f5dffdf61539a9f5b7b9a88795 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2SHmFOAuBCY.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b7a74b51085e2abed8a53d2e8a454a31f58cfb30068dd612f7a37319c6a3e8cc +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2SJQPh-FJ6c.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2SJQPh-FJ6c.pt new file mode 100644 index 0000000000000000000000000000000000000000..527714acfee12185615c3cf44f0a7691ac3ef053 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2SJQPh-FJ6c.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b7a86bac4558b3e854967a4593e07ff6fa74f3ae928cc827670b22b7a8b26abd +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2SMC8DiaZ78.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2SMC8DiaZ78.pt new file mode 100644 index 0000000000000000000000000000000000000000..687be8d677cc4d395738686dd2a85e0537dd211c --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2SMC8DiaZ78.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cc927544546285598729546791b1c39078a8ab61b68ee73462ea1fd8b1ad057c +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2SR_V8or3B0.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2SR_V8or3B0.pt new file mode 100644 index 0000000000000000000000000000000000000000..79228dbe27e67386d9e1e5601d70444ca6536240 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2SR_V8or3B0.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0a88c89ff430302d1b0afbdbc95dce6614f8aa583425ef7ecf06c9ddac7c78ed +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2SXPTIB5Lus.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2SXPTIB5Lus.pt new file mode 100644 index 0000000000000000000000000000000000000000..d529ca4dcc9be5a41269f76c5bb871fa6bab5026 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2SXPTIB5Lus.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:404e4dc6d2d63bff2aac584b5c7ececf4a27eec03b7cf7f451c91071a15134de +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2SZXZjahe7A.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2SZXZjahe7A.pt new file mode 100644 index 0000000000000000000000000000000000000000..7a99bd3589cb341e7dfd28b25b612fb21d3b5d8b --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2SZXZjahe7A.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:46cdbefb72d90d4a01895c0aae8579661c03256f6f64d1e5069f5a2de14860ac +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2SbM-bJVZSM.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2SbM-bJVZSM.pt new file mode 100644 index 0000000000000000000000000000000000000000..b6180950017c3d32e9fea3a31c3abcb8aebac339 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2SbM-bJVZSM.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4b258b50659382face1739eea85baedb79a363edf3f99a4ac309d04d213007af +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2SdxfLFtuFs.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2SdxfLFtuFs.pt new file mode 100644 index 0000000000000000000000000000000000000000..335738b22d4827c93f625058b77b03f3799694f1 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2SdxfLFtuFs.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4a08e1a1c984e92a7f95762b8c83af11ae72cd464fb7a2ec84b699fa572bf205 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2ShQ8X2gpyk.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2ShQ8X2gpyk.pt new file mode 100644 index 0000000000000000000000000000000000000000..42d123cda5f81ce3aee12e1221bad4a37d6e597d --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2ShQ8X2gpyk.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c6e377a7e1d343f088af731edcd28e98f0e450998f37fe1d6edbe73fafd98113 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2SlzXF1RcJk.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2SlzXF1RcJk.pt new file mode 100644 index 0000000000000000000000000000000000000000..47d3068070dc7954879c6dc241058b1538a57379 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2SlzXF1RcJk.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b9b792205104e383ad8c9980d5708c22686f11bd0cc9dde383794f70b3dd9cd9 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Smn9ZxbiRo.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Smn9ZxbiRo.pt new file mode 100644 index 0000000000000000000000000000000000000000..f19faa6c54021494f3f522978041598514bd6a71 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Smn9ZxbiRo.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:34df2f904316f1aca56973da45669e9e10cba8f004670c9ebdb2a6da96f5ce4f +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Sq6POM1tDo.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Sq6POM1tDo.pt new file mode 100644 index 0000000000000000000000000000000000000000..c76af7fd2542fee9f67d2106980f77bb7160eecd --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Sq6POM1tDo.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:94c0d148093109c17f8957948697d9f06916069be388db0742b840aaf11e2ba2 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2SvXEzs2uhU.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2SvXEzs2uhU.pt new file mode 100644 index 0000000000000000000000000000000000000000..2c8154f1248bca73efd266390e96d11cb5cc4b37 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2SvXEzs2uhU.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6899b72e053eed52d169a28693a75f0dd7a5350a36c20732092e8854e7f81a2c +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2SxuMywIOlI.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2SxuMywIOlI.pt new file mode 100644 index 0000000000000000000000000000000000000000..2a855683a87244b52871152f36af7f59f48b45c5 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2SxuMywIOlI.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3940dc1692487d8c29ead798a19120d985760dd8c8876d7d3c23d963cf1d6e44 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Sz23PAzAaU.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Sz23PAzAaU.pt new file mode 100644 index 0000000000000000000000000000000000000000..7582ce15f99d8654f28cc3d98ea2436df4be1e27 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Sz23PAzAaU.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:68aa9c42fea8d22393e822443211e22d73b8ecd2ff2cb5a86204e6bd61cd5967 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2T1XWfrRr7Q.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2T1XWfrRr7Q.pt new file mode 100644 index 0000000000000000000000000000000000000000..c772c966c80ea2a54acee14753eb531345416a7f --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2T1XWfrRr7Q.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f70a7ba732681b6a014e9fa87197f393f9dd75d5b85ebf232c30a2c389624438 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2T1ZozvdtUM.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2T1ZozvdtUM.pt new file mode 100644 index 0000000000000000000000000000000000000000..c20760b9d510b8bb06ba181f241a2f8f30644c7c --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2T1ZozvdtUM.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9c5a797fd3e078200fa2c4fa1ddec51e0924c28e6a93ca40217636415dd5078e +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2T2Mtm90MOY.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2T2Mtm90MOY.pt new file mode 100644 index 0000000000000000000000000000000000000000..54d323d9e08361e8c0da216b244fbba06ac7033d --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2T2Mtm90MOY.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2e9c722ea52ff33e8afd237e97e98f149429abd2e3c0a946cf7de0c4c33aaa6f +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2T8mloM2ZEk.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2T8mloM2ZEk.pt new file mode 100644 index 0000000000000000000000000000000000000000..64ce37a319d05e9dba30a92ba97979f7610081f9 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2T8mloM2ZEk.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d2ab920261d59cc18961db5d2bb92ff8bea11a908dcb15349b35db96eae55099 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2TKV5J9AfZE.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2TKV5J9AfZE.pt new file mode 100644 index 0000000000000000000000000000000000000000..2e66f849e3d1d7a362a7189599d0e0d338b9127f --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2TKV5J9AfZE.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:997baa2f9cfd4f5c5cb9324857b64fdaae6086d8f18cc17ae3eb88fa50ec2b3e +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2TKmUsUKstM.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2TKmUsUKstM.pt new file mode 100644 index 0000000000000000000000000000000000000000..cc8399c1af397f7c80c3739de28f8cffcb4ea0a1 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2TKmUsUKstM.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:958080503f509ecd4ee9091d6dcdb5b8b742c4fad81d6db6bc39af09059db532 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2TLDDITjziw.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2TLDDITjziw.pt new file mode 100644 index 0000000000000000000000000000000000000000..492bb8cf2b321f9a079dc7bbd105deed6f728176 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2TLDDITjziw.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:21fbe95ae65190caa884695897d9372c88ef1fd520024d8964196bf3ec330acb +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2TNTcPLxKMc.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2TNTcPLxKMc.pt new file mode 100644 index 0000000000000000000000000000000000000000..a66f25ca46547a2d08a1c6dd60b319c71f4d429f --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2TNTcPLxKMc.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:29e4820e51e6357d0d62c065118c55b9fe2235db7cad370079770935796d9435 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2TX7Eozn8cA.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2TX7Eozn8cA.pt new file mode 100644 index 0000000000000000000000000000000000000000..59bb684e10472f9cd5d5b70691e794e00d90c9a4 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2TX7Eozn8cA.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1fac72a804c19a90993481a05589b3dcac085833467d15e0d64a7f55a592ceff +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2TcPRLe-voU.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2TcPRLe-voU.pt new file mode 100644 index 0000000000000000000000000000000000000000..4db063df9fb8c8c96496f30fa5438e288e74545e --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2TcPRLe-voU.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:58499b0249673f6f611a946e352f1095ef585e7762ada32897498dc4a6124f24 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2ThSytr4T_A.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2ThSytr4T_A.pt new file mode 100644 index 0000000000000000000000000000000000000000..8752eb0b8031185659b72be03de0c2012e710316 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2ThSytr4T_A.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1f0065105728c8834f135330909e7041bc157acbad33c2b04a86715028f8766c +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2TjtFar4tjk.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2TjtFar4tjk.pt new file mode 100644 index 0000000000000000000000000000000000000000..93871534ec33e9e278e1d3c2cc97e0bace7216ba --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2TjtFar4tjk.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:54af4796c3fc21e0a6a9cf501b0061545cfa5b0303c19965b9cae55ece2a57e3 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2TqkjlOxEvg.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2TqkjlOxEvg.pt new file mode 100644 index 0000000000000000000000000000000000000000..1fa85ec5a6c5679ef7ad5670d0421a061e25cb18 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2TqkjlOxEvg.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0bf80f13796c6d03b150b2eef294e6b3ce7575649ecc9ba5f4d4a7a8258fd14c +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2TrCmyT7RwA.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2TrCmyT7RwA.pt new file mode 100644 index 0000000000000000000000000000000000000000..3d913d731ef7eb2ff4eec34f4b2d2def3e25ebb6 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2TrCmyT7RwA.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f9574db1fc9ff141c55cd9b9940126f3389dc05b714b61fc2a24770ef9fd3fbd +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2U5jiO3ep1A.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2U5jiO3ep1A.pt new file mode 100644 index 0000000000000000000000000000000000000000..83978231ad5b3d09260cd80816c73b5077c71ae7 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2U5jiO3ep1A.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3b0ae36d8665d96676e5ebdbbfbe1ec28e4f2c55d1593f6225a10122c8684e23 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2U7eq1pY2XA.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2U7eq1pY2XA.pt new file mode 100644 index 0000000000000000000000000000000000000000..e3d37634ec0ce9725b43e7f522c4515dafb05f69 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2U7eq1pY2XA.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e0d712b94df60b762d87460b5773298ce352fa1a36136c7bf28f8d77038cb01d +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2U9f09VZYUQ.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2U9f09VZYUQ.pt new file mode 100644 index 0000000000000000000000000000000000000000..721d3d7678e076c63cded7b6bef2dcf91bbd4679 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2U9f09VZYUQ.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7f373dfb7ba24a5da88c6d4d2b95ba4aab52f35293c107731b88f9f3a04d83ba +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2ULnE6P91Vw.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2ULnE6P91Vw.pt new file mode 100644 index 0000000000000000000000000000000000000000..ebe76d5e91dcf66402efc25056848549bdddd3fe --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2ULnE6P91Vw.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:83597a11145bde0cf95886acfbc7a7f6449d246228c9b3205512edc5aa937d4e +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2USY6q9oEhk.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2USY6q9oEhk.pt new file mode 100644 index 0000000000000000000000000000000000000000..5bb5379926a31d0f6228a4f58ed702d2b26a36ff --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2USY6q9oEhk.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:38cc44a0dee0eee59920f639cfef482a789f0be1d519edb0c3bbfce6f2f16fec +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2UUIxqen-kI.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2UUIxqen-kI.pt new file mode 100644 index 0000000000000000000000000000000000000000..457e5200351a588b34ce8a78dc5cc9e486580936 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2UUIxqen-kI.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:61958a8590efb6b3b80ef5fa97db1b573c75e9c49f6f867e96d06b23f5fa5a8f +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2UZZfLjvRLY.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2UZZfLjvRLY.pt new file mode 100644 index 0000000000000000000000000000000000000000..0b311629baba6ff017f982941689e19df870cbea --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2UZZfLjvRLY.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ba90cca01553a857171ec7e29f028eaaed0f185834c9e07c9bcf889b077c4f3c +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Ua3rk8FZNQ.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Ua3rk8FZNQ.pt new file mode 100644 index 0000000000000000000000000000000000000000..422ffa5f921da2a0e805704c749400023df233c2 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Ua3rk8FZNQ.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0f2785178e774d0c6136365ab2ac5a1aa4b06f35243502c05af115f62c33213a +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Up0LBAX688.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Up0LBAX688.pt new file mode 100644 index 0000000000000000000000000000000000000000..9f7a223e8e0ed50229ba77f33bdc8a0269317b4d --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Up0LBAX688.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:901babfed7f3ac7d8156a88d51553b765219b25462ffe8b143906969aca52fd0 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2UtO94tCKH4.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2UtO94tCKH4.pt new file mode 100644 index 0000000000000000000000000000000000000000..36d818a2b96d2927df59f29d65e9381c33ae557c --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2UtO94tCKH4.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2911aa41d9e7aa911aa73dd90908c56b0d2ae980add95632ccf8336aa95c1943 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2V-gNIW6cog.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2V-gNIW6cog.pt new file mode 100644 index 0000000000000000000000000000000000000000..3be2b57b7276281f78516f0989f8d4aa09eca19a --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2V-gNIW6cog.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:339cfc23694e1666bd9d40845f707a2c2e945a367375c961eca89018f98fec19 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2VE_sa4YrRQ.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2VE_sa4YrRQ.pt new file mode 100644 index 0000000000000000000000000000000000000000..65dd0ab8d465e847ba68f8d38b1ff144fff5543f --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2VE_sa4YrRQ.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4067198b994c56aac6fa0a2153b5765c21f1495a6b088881ebff6f9b210b2aaf +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2VGFczjfI0c.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2VGFczjfI0c.pt new file mode 100644 index 0000000000000000000000000000000000000000..67fcaa022ab04a133ad59f1d93423f769764c396 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2VGFczjfI0c.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:34be563d4fd008733c5495743a3659bea7812dfcfa2112ee66e7e0859acb51bc +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2VPgM2sr8qA.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2VPgM2sr8qA.pt new file mode 100644 index 0000000000000000000000000000000000000000..1b1d4ca928b855c0d90b88306ac526f9efb56787 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2VPgM2sr8qA.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a0c6ddc9a1cec5abf12e4ed45b3090d6aec000e22c2299713b90e474b24de201 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2VTyUW9kDnI.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2VTyUW9kDnI.pt new file mode 100644 index 0000000000000000000000000000000000000000..c91235fef037bfe5e7ea32f2804800290456fd4b --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2VTyUW9kDnI.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:72a0ea3300c5d85a66a3af8e32434e5071b53b80d6369272023df6c2cf14a79f +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2VY0mpyAXXw.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2VY0mpyAXXw.pt new file mode 100644 index 0000000000000000000000000000000000000000..5f3705f736fc3f5b6f709dbae6c11edee39759fc --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2VY0mpyAXXw.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6491c8b556afa8208403a52c3c569912a5076f85b586996e0e6ccfa0b16fc042 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2VYxcD062D8.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2VYxcD062D8.pt new file mode 100644 index 0000000000000000000000000000000000000000..1183182d688aad771d118e400cfaf6cedfa029f6 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2VYxcD062D8.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e202b8451162cead39bf498a6a2b38f527e260af3580d591717cecad976beb86 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2VeXkO_4JmU.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2VeXkO_4JmU.pt new file mode 100644 index 0000000000000000000000000000000000000000..201667e99b4eb6ec855fe55e5bb8a50a086eb15b --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2VeXkO_4JmU.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:994e7391ae5027e611b6215d5fcb94c6128ed7280044e6f4104c36b6f2e24827 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2VmTcIeTs4s.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2VmTcIeTs4s.pt new file mode 100644 index 0000000000000000000000000000000000000000..737df1cf30969bd08759a85fe43c1a22e5b02bc1 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2VmTcIeTs4s.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6673dd15bad2064562170ac2f75045c69d51bd5579f3a3ec4e5cce63e64ecc7f +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2VtOemBXsL8.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2VtOemBXsL8.pt new file mode 100644 index 0000000000000000000000000000000000000000..1cf6a1ef6273379d58627ae4b143693b99caf14c --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2VtOemBXsL8.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e0b8963abfb8465422a808e85965185da7c1e74ff3fb57bb5976ea0aeaa2b0dc +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2W0lI4VSEsk.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2W0lI4VSEsk.pt new file mode 100644 index 0000000000000000000000000000000000000000..bc9ce96d281824eddd4d32cd1e84a13866a1a5a9 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2W0lI4VSEsk.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a5f6480bdcc5c87a00eb1d8970dac0e776a16351907941fdb34b57b009f0cd96 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2W0vHcJtXBY.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2W0vHcJtXBY.pt new file mode 100644 index 0000000000000000000000000000000000000000..6df897a665c49ac381eccaf905d68fed7a8472ba --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2W0vHcJtXBY.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:57a1eae20d48a998467c8158be0a3d234524d3b80e6502a0850d7a2b0ea0b907 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2W34sIjX46M.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2W34sIjX46M.pt new file mode 100644 index 0000000000000000000000000000000000000000..a92406bc93ac18c5f5871a2dfd9a17e9fad0d27b --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2W34sIjX46M.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3ea735d7d10c99726c201bc8a10f031238dec973f7cfe86eaa6b983f4b378400 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2W3foW3GFPU.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2W3foW3GFPU.pt new file mode 100644 index 0000000000000000000000000000000000000000..6a92a26a1a7f0b2d841d3bc5dd7b0557122c1ac9 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2W3foW3GFPU.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a5524da412ddbad37a7026cd4cb6a957a7433b52e4c105da1e7933a75febb131 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2WE1rDGE1Tk.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2WE1rDGE1Tk.pt new file mode 100644 index 0000000000000000000000000000000000000000..ab4504f408194581c33f52f4dd12f1df93a88496 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2WE1rDGE1Tk.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:75d76b092111adedc3e0d4b3412efce3763e23e4cb585ad272e1407cf720724c +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2WE3gmuG_18.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2WE3gmuG_18.pt new file mode 100644 index 0000000000000000000000000000000000000000..b0c2f641d060917c799bc4e667f1c32c7861ca32 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2WE3gmuG_18.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:884981d8404f61a49fe22af12a0b38f98f0e8bf68eac99b06abe36f109ff49df +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2WEf6lFnuWk.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2WEf6lFnuWk.pt new file mode 100644 index 0000000000000000000000000000000000000000..c78d114e7dc8ca390e27612dcad0f71767da0313 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2WEf6lFnuWk.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:94f03072a7cf20f5bcfd852df1602587c88f4053d1b80e43cf0a0fd2b1541a5d +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2WGs8YXmjVQ.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2WGs8YXmjVQ.pt new file mode 100644 index 0000000000000000000000000000000000000000..3fa5c2a8fc8a2cefa1fd0856f517e9503c9cc705 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2WGs8YXmjVQ.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b91d715967ef9c41babb3044cea994910253f157ccfadd9559fee0be37923972 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2WH5Y6CP4dY.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2WH5Y6CP4dY.pt new file mode 100644 index 0000000000000000000000000000000000000000..9e27617dc6af729a0ba34a91174236ef4b9747c4 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2WH5Y6CP4dY.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7d3bdb365e349fa61657bb98b86341729c93f8a0b9751aa20a2903759ea2a39c +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2WVccq9cHhM.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2WVccq9cHhM.pt new file mode 100644 index 0000000000000000000000000000000000000000..1c37505cf1dcf5b208c0466cc7606b130bc274ae --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2WVccq9cHhM.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:91cfcee8565ed596b7987a4749021412e4a3ca86616ede172704bc33509169a0 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2WWDrSgmj20.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2WWDrSgmj20.pt new file mode 100644 index 0000000000000000000000000000000000000000..4f388858d2aef8c10ed68fc3c185f71646472aae --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2WWDrSgmj20.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8cdb7b2e5c6eb149c11f751fbf6f84b9cb921c2c8682e1cff073b33112f9dc6a +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2WZ6E89502M.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2WZ6E89502M.pt new file mode 100644 index 0000000000000000000000000000000000000000..7e98896d8d2827207fe488944d80047b0de29c09 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2WZ6E89502M.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ec961a81b1c1ff43e65631644c7cd0ec07efabc9f96a90e4f56ef55820078762 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Wcg4OeHsbA.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Wcg4OeHsbA.pt new file mode 100644 index 0000000000000000000000000000000000000000..d442151e8532dccfbc0177fd1c3847b5bb129fce --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Wcg4OeHsbA.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c0e058b00a51cf4c81d193a397d8179bfb8cbf86ae734b78827634ee01036d71 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Wd4-06knUI.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Wd4-06knUI.pt new file mode 100644 index 0000000000000000000000000000000000000000..94102acd9e1e6901a1542b67dbf2751ad63ed90b --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Wd4-06knUI.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a8bdd3dcb589f427fa437634ca73bd73f33cfa7ae91697e002367ceb35e0ae41 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2WgcnMpR00M.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2WgcnMpR00M.pt new file mode 100644 index 0000000000000000000000000000000000000000..9cc4089fceed8fb19a569a9e7142ca9bd76b2532 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2WgcnMpR00M.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:86fd1550be3130c55aaf2a33fde15af4996247f4cd18c2ca7164b7eaa94b23e4 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Wk2FocOTwk.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Wk2FocOTwk.pt new file mode 100644 index 0000000000000000000000000000000000000000..7bf5bf585e69651a354312ab547bc3231e313ca0 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Wk2FocOTwk.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:16bc83bfa82dcfa7c4777e3f9000ccc51e417825987b725f0fb346741a703ebb +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2WlNtZenipc.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2WlNtZenipc.pt new file mode 100644 index 0000000000000000000000000000000000000000..edb2b47ee45961e4566a97fe780f1ad562e4bb02 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2WlNtZenipc.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ec50b67204d5611fe8159899011459eeaa98e65a56e3ee34e5a9da17172d4c5b +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2WpfKAM8Cdo.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2WpfKAM8Cdo.pt new file mode 100644 index 0000000000000000000000000000000000000000..98c10ddd1ccdd5539232d3717b42ebcfde469ecc --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2WpfKAM8Cdo.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:972af4dc0943e6b9f4d3ed38fb63ac446fc7b0201656d47cbec41752fbed90a3 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Wr9Mq1cJxQ.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Wr9Mq1cJxQ.pt new file mode 100644 index 0000000000000000000000000000000000000000..1d48ff3375a19f3d3b45ce330331118449ab28fc --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Wr9Mq1cJxQ.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8ae236e91d08632a59078b6376e8ea27820bc7a29be41a4bdfcb2489606ae8ad +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2WswHvCDs_Y.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2WswHvCDs_Y.pt new file mode 100644 index 0000000000000000000000000000000000000000..5101783a52d08583a48650c83db19948b576437e --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2WswHvCDs_Y.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bd9bd38a492b80b4edc875b7c7c9c64cf1f6bd24fc0ac3f60e9b4b5045b3371d +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Wx2wokRoWw.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Wx2wokRoWw.pt new file mode 100644 index 0000000000000000000000000000000000000000..3932b8a0e1f70d0e9ea6f9c816ed71fb7ed9c2f9 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Wx2wokRoWw.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:eeab9c2223f2fd4a415fcd1f0c64fe564840b68e4cebe3bc5b6e39adea05e471 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2WyN8ByLXlY.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2WyN8ByLXlY.pt new file mode 100644 index 0000000000000000000000000000000000000000..a522061705036c57441426c26cfc175ae9802276 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2WyN8ByLXlY.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:89346bf9dbe8954dfa7a5808ab82d3886d00d559d675962a8f58321de2028b77 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2WyuCu309OI.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2WyuCu309OI.pt new file mode 100644 index 0000000000000000000000000000000000000000..afaa4a6ef8af0b8438f7a3d26152edff4cc93542 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2WyuCu309OI.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f88d7294a962e3bafa594513be97fa9c0e653fc8a9af1e54110698230609de29 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2WzJgI1GmbU.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2WzJgI1GmbU.pt new file mode 100644 index 0000000000000000000000000000000000000000..42e0b5e108f46f37a295a8834769f56f0838bcc8 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2WzJgI1GmbU.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8b73e5bca16931894dff40c30289740770bda8dfb9581fb6dddadd3b188eec49 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2X171fPIJsg.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2X171fPIJsg.pt new file mode 100644 index 0000000000000000000000000000000000000000..ef9d3023c8fd9de5d1a8ab35169ac1a78786ae1d --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2X171fPIJsg.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bfb680ab9beebd862380365f071d745b5ab6227571f23c0f3bf694339c5840b2 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2X7a5h--rK0.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2X7a5h--rK0.pt new file mode 100644 index 0000000000000000000000000000000000000000..e5731ebbcee7467213b049cc8ec60a508d67900d --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2X7a5h--rK0.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:945ddb13c4be299167522a1ad7d8cc88ac2b959d6bec6d477b1a90ab46041bd4 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2X8USZGTHuU.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2X8USZGTHuU.pt new file mode 100644 index 0000000000000000000000000000000000000000..d4016876b58416beb8dede94b1e763fdd021e9a9 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2X8USZGTHuU.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2f6072ae92e9c7ec7d3eca27d53d3e51ab03ed7ebed06f1c970b88f0f531f2d6 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2X9Pus8KsDU.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2X9Pus8KsDU.pt new file mode 100644 index 0000000000000000000000000000000000000000..1811329ba7f90b8acf97f48635e9a75877ea2bf6 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2X9Pus8KsDU.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:59fb78222ac0462f076c80274c061921c5a3435c12d4745afa1e86e8708b48f8 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2XBBDb01tTA.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2XBBDb01tTA.pt new file mode 100644 index 0000000000000000000000000000000000000000..9b2463e681c915d5a2bc552f4549077feabc6347 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2XBBDb01tTA.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f0e91c849a17e0d0d234fe2ada5560e48022656254b6cb95c1e364ac5b77ccaf +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2XHdotGfuG4.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2XHdotGfuG4.pt new file mode 100644 index 0000000000000000000000000000000000000000..65c35f02b5b5828df5e5d5071231b54e4b4d535b --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2XHdotGfuG4.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a0d056cbb78ffe2d453208fe3ba1a4f03b6db120f510fcf82ad4d88d6c7f5168 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2XJLJMm18_k.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2XJLJMm18_k.pt new file mode 100644 index 0000000000000000000000000000000000000000..5a5deec8298b4bf592264872dcac9eb76b8828a9 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2XJLJMm18_k.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4a4b9c26048e08e8e0d2f5d15e96b38f1166119f96fd85a7767566f3c719f85c +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2XOak3lVnmE.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2XOak3lVnmE.pt new file mode 100644 index 0000000000000000000000000000000000000000..86d1f36089a36a9fe811736fc49a17d1a22ae03d --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2XOak3lVnmE.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c31c31cf6785f519f70c0ff83b1abedf2c73ede11c26354752d2f6f02eb92851 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2XQpyiUaGUs.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2XQpyiUaGUs.pt new file mode 100644 index 0000000000000000000000000000000000000000..82049dc97f30ed6540089153ba63143f2b0db53a --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2XQpyiUaGUs.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:542bdd7e43369626842f4775c3565e3ca34999dd784f3f9c06b50a73196451ce +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2XWSjcCKPSA.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2XWSjcCKPSA.pt new file mode 100644 index 0000000000000000000000000000000000000000..d780f01963c6ce4be4d6b9bb03938f7d85ef7f60 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2XWSjcCKPSA.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d107c83e5f0bf2240a77ac553d3a1d96b3c0280f190e5c607bc6f6cccba5632f +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2XeEh04O-3Q.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2XeEh04O-3Q.pt new file mode 100644 index 0000000000000000000000000000000000000000..0993d5b7f02689ce4da40c638abf44b76721a076 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2XeEh04O-3Q.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:08cc26156198cd53579a30c9a68464d11c8bc68d1c6e6dceb524afc371d42855 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2XgZoBaEeEs.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2XgZoBaEeEs.pt new file mode 100644 index 0000000000000000000000000000000000000000..90b2606ec24bb218071fb5d6cc5b288b4323636f --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2XgZoBaEeEs.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:10af6ab82869136deb03a5256e714fdad271bbe90451600fbaf4b317b1257005 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2XiNFUvedqs.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2XiNFUvedqs.pt new file mode 100644 index 0000000000000000000000000000000000000000..8e890f1116b83505d5abb1721a2e52aa707f2029 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2XiNFUvedqs.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:72c86af7709e67c550a76488761fba5b8df00c3d17718a81f0bd836364e599d5 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2XlR_7swaxY.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2XlR_7swaxY.pt new file mode 100644 index 0000000000000000000000000000000000000000..a0e7235aacc492e3a6bd32fe563b1f80ef290247 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2XlR_7swaxY.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6c9a2f3b9922bd13c549b1568d8938a33f8b1d3cd8bfcc8e74be61b37c77c311 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2XoPTwVXLC8.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2XoPTwVXLC8.pt new file mode 100644 index 0000000000000000000000000000000000000000..b3fb0fdc0615eaccf8f1de015952d47c4a85c43d --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2XoPTwVXLC8.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e773abe82db4facf6c99d9fd3a9955904cba84068e5c84ce5f82b388268e47e7 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Xps25nYB6Q.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Xps25nYB6Q.pt new file mode 100644 index 0000000000000000000000000000000000000000..744650adf537efb90a4bec786b34bc6605dbc247 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Xps25nYB6Q.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:841b511edbf749a90043e11ec1041fabc194d19961a38ae67eb8ab19435f2a18 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2XqY9cGh3pY.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2XqY9cGh3pY.pt new file mode 100644 index 0000000000000000000000000000000000000000..7a4aca3e9817ec1c0ee9721a85bda5ba5326afbe --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2XqY9cGh3pY.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8eaa67fe68129c093aaf1fe54ead276b0fc65ac3cbc6b2e8f2b765b2a51351ce +size 18352456 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Xr_0OFaSjw.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Xr_0OFaSjw.pt new file mode 100644 index 0000000000000000000000000000000000000000..8cf8d91e07936f3b8966d9d98ecd098a4ec6c071 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Xr_0OFaSjw.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fde5a6f04c701da8369046ce4a48a3ebf87532cea3cc2a900d69ad70c07e2ffd +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2XuzKMQX14M.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2XuzKMQX14M.pt new file mode 100644 index 0000000000000000000000000000000000000000..6cfaa6b1c9a81884d26a016ba59bff595417b76c --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2XuzKMQX14M.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:36211bb17474b81b08032765d6608c288f8e4e476dfd0915a5726fdd070b7a2c +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2XviNVagsPg.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2XviNVagsPg.pt new file mode 100644 index 0000000000000000000000000000000000000000..ccd99747fde3c2cdb2884b0483f037215adf8ae2 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2XviNVagsPg.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:81a80b85a6d8ea366c3d65dcdf5dbb1da28c7767b9f9f97c8ec75e6bbe21a4a5 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Y-yNSwQ0TA.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Y-yNSwQ0TA.pt new file mode 100644 index 0000000000000000000000000000000000000000..c82f4a1ecbaa2ef8c0ba07fe31f7864acb117d7f --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Y-yNSwQ0TA.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8f4e98e116c9f4eb6bad45c522c61fa5afba837c42923c8abb2ef556aeb16f89 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Y0veQNiW_o.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Y0veQNiW_o.pt new file mode 100644 index 0000000000000000000000000000000000000000..889237f714050c2d56fd3b184d2423823254e716 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Y0veQNiW_o.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:06f39891643a3bd5426d8870697ba474c8e6d0f7e805e30d29973c64db6eb7d9 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Y3kIvUVO6o.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Y3kIvUVO6o.pt new file mode 100644 index 0000000000000000000000000000000000000000..b7952f4c6b4a0b11c80237b696bba2fa8638ac5d --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Y3kIvUVO6o.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8e16ddd5e27bdb228371eb224f746e82d7c4d8e1cead70d5fafc833adeb2cde3 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Y3mi2gVBmg.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Y3mi2gVBmg.pt new file mode 100644 index 0000000000000000000000000000000000000000..6d7bbe322c4efc082f0174ea405aae2e8727a360 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Y3mi2gVBmg.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:18270310857befa8d01691e7c08d3b7345ef5c67dcd10149419daf627ff92ea7 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2YBA3jTpDZE.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2YBA3jTpDZE.pt new file mode 100644 index 0000000000000000000000000000000000000000..6be74fe57bce5d3e8b55e63f8eacd901cd3b47d8 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2YBA3jTpDZE.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1513318c3858cb47d8ae3d6e55b42645bd0dbebd0b2a481eb87beab9873bea88 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2YBtVTVnIAc.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2YBtVTVnIAc.pt new file mode 100644 index 0000000000000000000000000000000000000000..620505a710e8fd9fe06b4264eb1d5f8993ca7ffb --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2YBtVTVnIAc.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b9504b6956c1e1b8eec685c1ab07fe9166a05d2deff1bc238ec0d4c306e84640 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2YD0f8HD0M0.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2YD0f8HD0M0.pt new file mode 100644 index 0000000000000000000000000000000000000000..d2220b666f896a9ac91553235713a2591c5220bc --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2YD0f8HD0M0.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b9dce1ff9ed283f6112c01aff2cf45e581a1f2a9c15ed64173d50bcf8dc5e543 +size 18352456 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2YDEr5h-epk.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2YDEr5h-epk.pt new file mode 100644 index 0000000000000000000000000000000000000000..48f6153bbea99ad0f37b67198f230931f4b00a13 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2YDEr5h-epk.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6c46275c72181ca426bbb6b82e8c8c7404d4db13e9fe3fe9f97df40d74877136 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2YFOuJ3XXpg.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2YFOuJ3XXpg.pt new file mode 100644 index 0000000000000000000000000000000000000000..2982e84d9225ea6a274211dff4744d37adbba799 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2YFOuJ3XXpg.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d310b6c53a702702f968f1277ed1433c83debaf5e475bf9c27da5db80712a7aa +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2YMPfLwMcZg.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2YMPfLwMcZg.pt new file mode 100644 index 0000000000000000000000000000000000000000..4f4c186171a2e408b904dbc55aa195611822b43d --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2YMPfLwMcZg.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b532ac443874f6750985fd310bebb4a0e6f5c66ef261052e576d5c9e804d6528 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2YO5UsE2OfU.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2YO5UsE2OfU.pt new file mode 100644 index 0000000000000000000000000000000000000000..e465364d7396cac7db5b4809d4a9fb50276d652e --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2YO5UsE2OfU.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a9b446d1d38025576ab24fd2d36a231716b739142523b5a7a4924e1923e97383 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2YOzWgwPUCc.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2YOzWgwPUCc.pt new file mode 100644 index 0000000000000000000000000000000000000000..fc85ead17ff62edbc7bdcd60418c66739bdbf18c --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2YOzWgwPUCc.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e9cd1a1b0a48dd9530bf4df65424a48137cd9feea9d9673498fd6b81e52bb98f +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2YWiZYZBDwI.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2YWiZYZBDwI.pt new file mode 100644 index 0000000000000000000000000000000000000000..c716565ea4c573b40da5be90982006aa3c0b9902 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2YWiZYZBDwI.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ff4e8acb7ef8be716004f31e4bf1bb7771b5eb37e6d83d01fe4b87554319d4e6 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2YZMKETZ1B8.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2YZMKETZ1B8.pt new file mode 100644 index 0000000000000000000000000000000000000000..7c297b3ca23d8b806672cf297c494485820f72c2 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2YZMKETZ1B8.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d87dde7df992b968ed2a26d1f04ce847cef9d9aa6d6989fd1640075892247e41 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Y_u5YuF2SU.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Y_u5YuF2SU.pt new file mode 100644 index 0000000000000000000000000000000000000000..c98393cda08b3ca7274ab20aceeef9913c9ae09a --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Y_u5YuF2SU.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d1fe486c65d4487d0d40a8b4d3c15351b31ff1defb1254b3c25572c27d791273 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2YbqO0yNjhQ.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2YbqO0yNjhQ.pt new file mode 100644 index 0000000000000000000000000000000000000000..bdf8ccf4b8c70bb4c895e4851e12ddf943033db8 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2YbqO0yNjhQ.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:137f7f452ad90cdc5b788e9ed3018bffe5c014fd7329f09ff74ce2afc97c229a +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Yhhj-jK8Pk.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Yhhj-jK8Pk.pt new file mode 100644 index 0000000000000000000000000000000000000000..59284c506bc0293924109881329933a153e22476 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Yhhj-jK8Pk.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f6f0c3b7bc2e1be7068f9c05e21476e3a884f4da8afd186633503ce0c7765ba9 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Yi4HgR3dHI.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Yi4HgR3dHI.pt new file mode 100644 index 0000000000000000000000000000000000000000..b5bc7400c6e6a3edcfa5175e88f03f6124b4f8ed --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Yi4HgR3dHI.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2950d276bbffef0ac60326d31f60b1b518e4c3cca6eff835b9f2b28d0b455e57 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2YnujWKIxhM.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2YnujWKIxhM.pt new file mode 100644 index 0000000000000000000000000000000000000000..042f7893d56a4e22efc7996bc0d7ea6b74d5e4cc --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2YnujWKIxhM.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b537b797e6180f719e0a396ca11d49936f96569beb6817ccb098f662202c9006 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Yobn4PutMs.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Yobn4PutMs.pt new file mode 100644 index 0000000000000000000000000000000000000000..79cb0e4f0ca65afd9886ec267723fc0e02a7d43f --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Yobn4PutMs.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:652c65147c443af99a68883793d13d6358b917c7c1b976e2996624687b730a45 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2YwDUhwP1gY.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2YwDUhwP1gY.pt new file mode 100644 index 0000000000000000000000000000000000000000..a6e8b5ce3fc6141bb5e67ce0ba1f61e95a27c3b5 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2YwDUhwP1gY.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:eec0ec7bbbbb2687799351cabaf979bbd62a559bba7bf13f7b3149d41f55f991 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2YybCUir4mc.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2YybCUir4mc.pt new file mode 100644 index 0000000000000000000000000000000000000000..683281db2791f39292ced5748b7d058948ac6d21 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2YybCUir4mc.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7a3b908dc8ae4e31b68a08c9ff3bdac5b18de8ecd69804868a1d73d54ec5a1c4 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Z-aiERtRoo.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Z-aiERtRoo.pt new file mode 100644 index 0000000000000000000000000000000000000000..05557f95a467d7bd1cdbb84566d5a1d534c27656 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Z-aiERtRoo.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:aeaaf7c5543f62d0e85e645e05898e735732cdfe867dadb50c7420d3971f8191 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Z7y1muZsdU.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Z7y1muZsdU.pt new file mode 100644 index 0000000000000000000000000000000000000000..51ec920d868d5caa5a5ba3a7b4ea8cc088622468 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Z7y1muZsdU.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cad65a2801a95a3c9f9c53076af11c89b0a50cd288440b7723b734edffef7f68 +size 17041736 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Z9Ueiagd04.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Z9Ueiagd04.pt new file mode 100644 index 0000000000000000000000000000000000000000..301a01d315709aa6312cae9c19e3e4dbb22e8900 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Z9Ueiagd04.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c18e4990357113df6f50f6d720dd6fdb879e78c073fd5fb56adb2b8086703cd2 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2ZGSAZJmxdM.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2ZGSAZJmxdM.pt new file mode 100644 index 0000000000000000000000000000000000000000..e4a35eb8865589142ec1b7e82f5d9507f38d7a97 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2ZGSAZJmxdM.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1e378b3540fe64d65f2ba574bbb699e1e33a35221c7053bfe04d1978d27c9452 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2ZHcM33Zm10.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2ZHcM33Zm10.pt new file mode 100644 index 0000000000000000000000000000000000000000..508b7f575e3d0a8ca132c78b9a7dbbd77815647d --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2ZHcM33Zm10.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:77e587f3938ef894e291377689c432f1ef36826da3a07331b9fb8372fa5273db +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2ZNeDK9rqWs.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2ZNeDK9rqWs.pt new file mode 100644 index 0000000000000000000000000000000000000000..7a37187d9196877c6177d98e1610cea1c2d08fc2 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2ZNeDK9rqWs.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:163fc8680ca484685a08ce880cb319c88ffd3c20b8c3d93619920d4775c7333a +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2ZQT8phATYU.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2ZQT8phATYU.pt new file mode 100644 index 0000000000000000000000000000000000000000..e429ea67c4d69bc598b4f2ed21aa480c6d969471 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2ZQT8phATYU.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cffde082cbbf8511677a76a15002a26459bda4e8f03a732322f3924bba53333e +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2ZVB6SA_O6o.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2ZVB6SA_O6o.pt new file mode 100644 index 0000000000000000000000000000000000000000..8a33365d2b70241fb70dff63d72b2c7f8abe32e6 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2ZVB6SA_O6o.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b12f3a6ed3b4a95d23d37aacd93991c49a21126c0d33f8b4fc2a694f686f3b45 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2ZWGC0abUw8.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2ZWGC0abUw8.pt new file mode 100644 index 0000000000000000000000000000000000000000..fa4b0d8e3dee12f15527bcb4d94a9e8f31a446f3 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2ZWGC0abUw8.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:abbb3a86bdc3e655ece4a9e6673da14f61a6178c4ff7b1f2637b4accb713b374 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Zdhke9Lsmc.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Zdhke9Lsmc.pt new file mode 100644 index 0000000000000000000000000000000000000000..ed4f820417c2afd950f8117390f7a33d5b6793a8 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Zdhke9Lsmc.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6844121aece3c93ed514772a490519fa962f081f140ef15175d9efd5c84abb83 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2ZefngWxXoE.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2ZefngWxXoE.pt new file mode 100644 index 0000000000000000000000000000000000000000..84339fef5bb1cc01c4952fe00880006749ba2d3f --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2ZefngWxXoE.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c0421114fdb86299da1b51246b41fae34d1280727239b2c676ef44951e5a6da3 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2ZfHNCeMPVk.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2ZfHNCeMPVk.pt new file mode 100644 index 0000000000000000000000000000000000000000..5b27445c34e1f1f4195a7a792ea061852f85feac --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2ZfHNCeMPVk.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4cc56d0cead54be22659c6eebe8f21b0c57c19503d2d603b940623894091beca +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2ZfWs96xhSI.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2ZfWs96xhSI.pt new file mode 100644 index 0000000000000000000000000000000000000000..7165ecc0d953286325ed24745f6b510974ce9019 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2ZfWs96xhSI.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e6a4eaf114c56a743fb7cdc744450043c6615444a96761554b3504b3e50a9e62 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Zj4BLIB5pQ.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Zj4BLIB5pQ.pt new file mode 100644 index 0000000000000000000000000000000000000000..f9916f03ee89c58a9de7d15ff29602e8c855e5de --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Zj4BLIB5pQ.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7cb7284c3ea74cacdbd82eac129818ba46493c926711503c0cce926fbef2b34d +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Zx7-AwiG9Q.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Zx7-AwiG9Q.pt new file mode 100644 index 0000000000000000000000000000000000000000..5166e4d7be8324157138e323870923cbe417ce5a --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2Zx7-AwiG9Q.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5ad3e7c993e04945cd389d853f508b5ef03600d650e1ccb4ec3bf34ec55de0d5 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2_-qIR2Lb9k.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2_-qIR2Lb9k.pt new file mode 100644 index 0000000000000000000000000000000000000000..6568cd05278d2ad31960da39df6eceed016a504f --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2_-qIR2Lb9k.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f67f85fdacacb4fce13e304a5b2c44c4bc61d742960245af56273c2a6f5efe89 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2_5zbct2Chs.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2_5zbct2Chs.pt new file mode 100644 index 0000000000000000000000000000000000000000..e103749b4a986fc20181c45626fda83a8a5ff843 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2_5zbct2Chs.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ebef40766d59435d7cd916f54731f6e50f01735e939c2d145080ad5ffcd10a7f +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2_9kbOzvm68.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2_9kbOzvm68.pt new file mode 100644 index 0000000000000000000000000000000000000000..53f90c91dd39df63bb69984879246217ea1870a6 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2_9kbOzvm68.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:811d326743c2f5c94929e5ccb41261c88ad7b00d63758649fa34376efaf7cbbc +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2_Bxkw4lZB8.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2_Bxkw4lZB8.pt new file mode 100644 index 0000000000000000000000000000000000000000..13ebd45b326f8f21878b5ba9697ec2ff25e37507 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2_Bxkw4lZB8.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:aaeada27b416441ec41c6a80da0aaca8a1a0b208ac6879cef687fad4f66caef9 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2_CNgOLBJWQ.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2_CNgOLBJWQ.pt new file mode 100644 index 0000000000000000000000000000000000000000..44335755ee6399ed62e68bd6c1ced29474978205 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2_CNgOLBJWQ.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3951eccbcf640fa0b19b2be2bbc25d89310c3d04b0a786ef342719376b6ecacf +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2_Em5gVYJ1g.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2_Em5gVYJ1g.pt new file mode 100644 index 0000000000000000000000000000000000000000..d94cd32042b537dc10a0d2c09b34c669ccd0b9ef --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2_Em5gVYJ1g.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:679792ed994a39f691c37051c50dff4507226c24069cce40719400a7aa705a44 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2_JFXfahI_s.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2_JFXfahI_s.pt new file mode 100644 index 0000000000000000000000000000000000000000..a6fb3b7f5c8c4b5bc163ebed15ffb68fc6d9576c --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2_JFXfahI_s.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9477af5e6a3fd63a5fc2ae2731c4c22a564681797db4980a451bbe243e3a8ed7 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2_YUnVrETws.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2_YUnVrETws.pt new file mode 100644 index 0000000000000000000000000000000000000000..7d1b2077a66d51da8c74fd9f5fef8dbf2048ff9d --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2_YUnVrETws.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e7161fc4d86dfbe5da37ff5224125d798785ffc39cf529691e9d3cd6a10a50c9 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2_ZI3yNNLtc.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2_ZI3yNNLtc.pt new file mode 100644 index 0000000000000000000000000000000000000000..3cca160e228150f09232f0165557f6183128497c --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2_ZI3yNNLtc.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e3bc698091973da918ffaece5043353957564294475cfdf5f4b855a769823dbf +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2_aIzISDda0.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2_aIzISDda0.pt new file mode 100644 index 0000000000000000000000000000000000000000..5be1b8a8c3c6ccddf4f6a1f4586737436729b20d --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2_aIzISDda0.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0d7ee8415a1b658d7b49666ff807dbe91438d4290657b9d05424910fb0e55011 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2_h_KpRjX4k.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2_h_KpRjX4k.pt new file mode 100644 index 0000000000000000000000000000000000000000..5afe2cfdd78f12228cc851ad7b828d1c2288a3c1 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2_h_KpRjX4k.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:aaff993e546ce5de8a84f5c3ce6f40417d7705202c71c03a04f45256560ff8d0 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2a1oeFpkKAw.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2a1oeFpkKAw.pt new file mode 100644 index 0000000000000000000000000000000000000000..31de5d4ae1e1e519b6385b877af753beb7ae6345 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2a1oeFpkKAw.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ebe2ba36f7433f06d65d23e86df289ad5e2a022745efafdf5d1c8781b3c2bf12 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2a3tFwKTbS8.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2a3tFwKTbS8.pt new file mode 100644 index 0000000000000000000000000000000000000000..2e51b912da4050a7dab6025adc98e24e3133c9c0 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2a3tFwKTbS8.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7def5626123400c3014843a2a92449a1614497736cff0b3149d751f24644d913 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2a62KoO--y8.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2a62KoO--y8.pt new file mode 100644 index 0000000000000000000000000000000000000000..e7fb7dc19cd2582f6535ea758487cd9c03530613 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2a62KoO--y8.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1223679e0a73bd75dc8f775a579737fbd1424fc300eba98eb812fbe227a268e1 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2a9oll96i38.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2a9oll96i38.pt new file mode 100644 index 0000000000000000000000000000000000000000..0b60df8a8c23172e48908e3287a7b5af32c96024 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2a9oll96i38.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:accc24efa0645f6e9c196cc815c4860b31781e40aa6cb789e35e9d6aa48d55ba +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2aCfb4sx-iY.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2aCfb4sx-iY.pt new file mode 100644 index 0000000000000000000000000000000000000000..53006a1a1e96fdedd4487707d3f06c4a06bd87ac --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2aCfb4sx-iY.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3cdd70b0adb40a17a3050eb5298fac56251914effea2ba288453413dcdc567a1 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2aJh1Li0ZTc.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2aJh1Li0ZTc.pt new file mode 100644 index 0000000000000000000000000000000000000000..0be64b463fc2521460cac16b1bf1c7b47cd4f9ae --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2aJh1Li0ZTc.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:79069d21b2bbc550d852eed153abb15568ee8b500b8c0e17e864e10c842e4fdb +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2aTYXG-H2FU.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2aTYXG-H2FU.pt new file mode 100644 index 0000000000000000000000000000000000000000..3931fcf742cea354e646a593339e6fe0c6c424ac --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2aTYXG-H2FU.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f044d1d01d6621d1282a9e211e52bc09c295f308f4dac792457552204841122e +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2aYOlw3USeM.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2aYOlw3USeM.pt new file mode 100644 index 0000000000000000000000000000000000000000..e278c49773b6b966263d876b9264f08bc3c1c1b8 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2aYOlw3USeM.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e5205584a22f6c1bfc62e4ee121dd2dd9a0d21f02310f5f42d28f9107b7ec746 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2agyUGe_uPc.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2agyUGe_uPc.pt new file mode 100644 index 0000000000000000000000000000000000000000..607285b2f97e7f16268088149d8ac689c28da59b --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2agyUGe_uPc.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2f16ebc30a77e6c16ef2a0110f5a1513d8793902ddc8c3e9c5eb7c613b693237 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2apuMEqXUnI.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2apuMEqXUnI.pt new file mode 100644 index 0000000000000000000000000000000000000000..0b708814fce9d2bab1a30b6d450e4afedda2f040 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2apuMEqXUnI.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:38b5d572bbdbeca0b77bf6436bc1e2ae0e4fa5866d376ea969be0f5b1824b97f +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2aquKFkp3PA.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2aquKFkp3PA.pt new file mode 100644 index 0000000000000000000000000000000000000000..cf6dc48139223a86ccb7d34da2c5e712ef418651 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2aquKFkp3PA.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bb1d6c80f383ae4aa748d57e798319c9ff817140b6fce1266556933b7713e78f +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2as_btlfZ6A.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2as_btlfZ6A.pt new file mode 100644 index 0000000000000000000000000000000000000000..1d829dc383e543a84c07e9a8a665cbfbf41a2b85 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2as_btlfZ6A.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9de95d3183e101992bcb121fc8c15b8c05f85b96be197ef2557e8dd41eabbabe +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2atJNkFYa-c.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2atJNkFYa-c.pt new file mode 100644 index 0000000000000000000000000000000000000000..bc9757da2412c5c0846e718849d30751f7b31c19 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2atJNkFYa-c.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1315c7705daf5bf43dc989c02feb61828705bf0bb83a65f0e8bb05481af2d0c4 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2au-gGMp0_Q.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2au-gGMp0_Q.pt new file mode 100644 index 0000000000000000000000000000000000000000..4eabe6bc68206e08a57188e38bfba9fe7f96b9b1 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2au-gGMp0_Q.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2dfbc00c111a4ea6a7b2e609c20991750b283ed94f8463ca7560c83dd0adefb1 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2auBr_lkpG0.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2auBr_lkpG0.pt new file mode 100644 index 0000000000000000000000000000000000000000..dd8d8d15450543e3df0f5b26607432026fdf7fe4 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2auBr_lkpG0.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d7b0a4813d6833b9992904a6800fac41dde61ddaa43fcddb3a7cb90aa6c18444 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2b-6z-lAeQY.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2b-6z-lAeQY.pt new file mode 100644 index 0000000000000000000000000000000000000000..dc5305cc26ffa36c5e89a812bb2bc7433dc754ee --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2b-6z-lAeQY.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c1cf671a22b73612536694a20d7cd6e9153a2e465a63e5f8c05aa1659f4b08e9 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2b-NCUTq5pM.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2b-NCUTq5pM.pt new file mode 100644 index 0000000000000000000000000000000000000000..f21cf182f86ad68380ce75a9dbecec3ea9abb821 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2b-NCUTq5pM.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:50ebcf7df45820a0bbdf18de5af7fb525ac36b29fb9c65c8593f35c4dff9d4a5 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2b31S7zqFn0.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2b31S7zqFn0.pt new file mode 100644 index 0000000000000000000000000000000000000000..50f2f9d31b8e31ecbd4208ad87b00e36d9d40a4e --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2b31S7zqFn0.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:041bac40a05187a4b01b0964c22215a257dc159f82088bd7fe8d6c02292e464d +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2b4h4nZYFho.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2b4h4nZYFho.pt new file mode 100644 index 0000000000000000000000000000000000000000..bc1d097275a86cd9b389663557782873050f055b --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2b4h4nZYFho.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e6211ebfc4417c557778be4406007f03723f2e839401d42df3570ea63bb58659 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2b5Dpvd60mk.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2b5Dpvd60mk.pt new file mode 100644 index 0000000000000000000000000000000000000000..a98fd723b13a274acdd0460210a501ce0f0d5f28 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2b5Dpvd60mk.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9987d6267454d88352e17b5c419cf59a7d710de2e78feb4e500e8a093efafcfe +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2bFg4-yT1XA.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2bFg4-yT1XA.pt new file mode 100644 index 0000000000000000000000000000000000000000..ed5eee0a02ca1e545e35790f87743127d44b9db0 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2bFg4-yT1XA.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f8c86169ed508d99b5a1808e0e2971af1558c22fb1df6aff267817977f8e8b2c +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2bFkkXxrYcI.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2bFkkXxrYcI.pt new file mode 100644 index 0000000000000000000000000000000000000000..074f7dd0816f60a02b50832c35935d741c4fcae7 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2bFkkXxrYcI.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ea8950985664c35a935cafc23819094566ef60bb081c87dee282a940a57c1c83 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2bHXB7B4Tl4.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2bHXB7B4Tl4.pt new file mode 100644 index 0000000000000000000000000000000000000000..c346b7148834d9f05f8ec4b9d61a699828dcbc82 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2bHXB7B4Tl4.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f6b9c433d63c648f83c1e8248a663959ae9ecff35d39ab665d17ca94f15219b9 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2bIdsZ4LgwU.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2bIdsZ4LgwU.pt new file mode 100644 index 0000000000000000000000000000000000000000..3533c4731daceba02a3371f63cdce9542802d449 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2bIdsZ4LgwU.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9a4ca0344837aa124ac831cbccd18ed841782a8050852fa252d988a7797796df +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2bKx9hFw3kI.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2bKx9hFw3kI.pt new file mode 100644 index 0000000000000000000000000000000000000000..cac97220e0d9b256c2423fe24099535eff7507af --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2bKx9hFw3kI.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:428701e93d1cdd1ece645807aa573bd3fe2f8d81262bd6a546b5048a8b3dd666 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2bMp4ldzJNk.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2bMp4ldzJNk.pt new file mode 100644 index 0000000000000000000000000000000000000000..62d51f38c0b535974bae3d2b6dbb23c5c4894f81 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2bMp4ldzJNk.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ea5f35c2e434d4219d38151b65d3f6ba93f9f478f720401817fcd15136c4ceb5 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2bN6LLvAV4k.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2bN6LLvAV4k.pt new file mode 100644 index 0000000000000000000000000000000000000000..c92684fb6518ed694e2a286bcc67137a3ff377c3 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2bN6LLvAV4k.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a40c7e42fed8603169807a0f3621ed2cbb8bdf5fad2832cc1bf3b771800852f0 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2bOOCHsPE8Y.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2bOOCHsPE8Y.pt new file mode 100644 index 0000000000000000000000000000000000000000..4402517d8349143d4b20754735c716fa4345ba9d --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2bOOCHsPE8Y.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cf9c638c40170d37a403ce3a8285835b64054b306aa2c3844c8aa54312cb0508 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2bOctFR1znM.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2bOctFR1znM.pt new file mode 100644 index 0000000000000000000000000000000000000000..264be9a4e1dd4cef389e8578ffcf66f553d187a6 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2bOctFR1znM.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:762a7582b5efbfa338483b7cb2b7aed47626fee01187836e382abea1c3bdb96d +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2bPZ8Hq5mWs.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2bPZ8Hq5mWs.pt new file mode 100644 index 0000000000000000000000000000000000000000..d022229d8181a2dd9665f31d79d8c099f2b01e0a --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2bPZ8Hq5mWs.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:14e08023f0bde91bbcc26d2a686092e45a461143da4ea5e7be8c641d0d54632e +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2bWR67pu86I.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2bWR67pu86I.pt new file mode 100644 index 0000000000000000000000000000000000000000..1a04d169726e9b5efb809477c5a9009df9169dd6 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2bWR67pu86I.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:18cb51e8c1634510f14d8810818ebfd28ad309133e9f5f73eea1e70e417e3065 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2bXcWeQAhSc.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2bXcWeQAhSc.pt new file mode 100644 index 0000000000000000000000000000000000000000..0350a4196015b85259b8564ac0e08ade1e745538 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2bXcWeQAhSc.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0b9f0309c8a279149c9eefb9ab07c4f61a351ca9d9d6a732a1ab6d5b2d9b8558 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2baVSKq8eWY.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2baVSKq8eWY.pt new file mode 100644 index 0000000000000000000000000000000000000000..489b8221d0f24415ab1c90d4a8c0710d6bc43252 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2baVSKq8eWY.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:02951c3d2cccdb59fd6f866a86d41243e083d90e368823a3c8d9a92d699aeca9 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2bbBGZCvE2s.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2bbBGZCvE2s.pt new file mode 100644 index 0000000000000000000000000000000000000000..e55a9ecec9951f1392c764cb9c400e9817f31dec --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2bbBGZCvE2s.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cf9f9a5b827243339a3c9c481b16d188fac5c09906e0b6c94c0edb67d5d46cdd +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2bce_3BEoOU.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2bce_3BEoOU.pt new file mode 100644 index 0000000000000000000000000000000000000000..adc2d6cb2518d211bd5bf14d88c55a7ca8aa59d4 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2bce_3BEoOU.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:175a1df07bdf162925ab302ee6e2507d5a90315dc7bef25f9b27324c3f349b7f +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2bglxLj8dsU.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2bglxLj8dsU.pt new file mode 100644 index 0000000000000000000000000000000000000000..ce138988b49f444d559bad6e26075329fd5815b9 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2bglxLj8dsU.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bfc95eb378287643adfd86804a10ef16f14bec059e2974336393364e9d360539 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2bmqTpObyqw.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2bmqTpObyqw.pt new file mode 100644 index 0000000000000000000000000000000000000000..97d4dca56094e647086f99389d7629e64a9d4af8 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2bmqTpObyqw.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:53ba6c52ee8c88888b6c1988bfe0d7136d4b752648cc390a10284be0f5ecb7fd +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2bo486U6wGs.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2bo486U6wGs.pt new file mode 100644 index 0000000000000000000000000000000000000000..2087a66ce89b20069b938b0879d3f1bb53f8ab7e --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2bo486U6wGs.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:643dec32f3149c91dd62ece979f6e243eed8c12a7b5a0c6fbfd7742232e57b2c +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2c0dBu0cx78.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2c0dBu0cx78.pt new file mode 100644 index 0000000000000000000000000000000000000000..e6f79bff7a9ff6d8a32103a999b5540caffb525f --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2c0dBu0cx78.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:65ef1c0cb75dac422e32966242cc69a1b04f17b55bc68472b6a1d2db5910e139 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2c1F97wGDc8.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2c1F97wGDc8.pt new file mode 100644 index 0000000000000000000000000000000000000000..a3291493f1c68ffa0e63a99c2ccd44f88dba43b1 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2c1F97wGDc8.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:79c90f5b80640a723237aafe6a42a74f4d945592466d2007fccbc1219aa286b3 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2cGj79rX6qw.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2cGj79rX6qw.pt new file mode 100644 index 0000000000000000000000000000000000000000..b39722a3461f3673ef6893670ef533e85cb84d9d --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2cGj79rX6qw.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0fa3e507195788262d268a507001476aa9679177e2623d5e4060d33b450d09a2 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2cMvPYwskbM.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2cMvPYwskbM.pt new file mode 100644 index 0000000000000000000000000000000000000000..f8d5369ef05b72dbce64dbac3f577912c7327f85 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2cMvPYwskbM.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:37039dbd31720adfd496fcc1ba7fb69eb98217768bd9872d690a3f855bbc3d47 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2cReF1sEyqU.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2cReF1sEyqU.pt new file mode 100644 index 0000000000000000000000000000000000000000..d6518a4ea77b610f756f3ab03565a74268754e4f --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2cReF1sEyqU.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c8c5ca3cf98a79f272a7188a2a6604fee9136e77dc1a698a09c3cabf10b1fbad +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2cSaHJYK7uY.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2cSaHJYK7uY.pt new file mode 100644 index 0000000000000000000000000000000000000000..978e11e27de7aceaa57562ac2983d329fbc6b712 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2cSaHJYK7uY.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b330ef3b4494efe4a008629984549ade5c27170168e5336dc73c7f9bbbb0a7a7 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2cW37meoXjs.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2cW37meoXjs.pt new file mode 100644 index 0000000000000000000000000000000000000000..ea03f6ad77d792b6b0ecbf970eff5d5d56d0d3e3 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2cW37meoXjs.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:98de43926b178ff86a94d860bb8c8259d79be6202ac69e67e600740dfa6b3f47 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2cWUo7vcb4k.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2cWUo7vcb4k.pt new file mode 100644 index 0000000000000000000000000000000000000000..b12e1dd5590f3ed7ac8cfc10b14405fdae640edb --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2cWUo7vcb4k.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bd3c78dc9701fa0905da1946d1526efd127001c873c8c33a4ac49c5f889b8c79 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2cWV4FScHwo.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2cWV4FScHwo.pt new file mode 100644 index 0000000000000000000000000000000000000000..2335a2ab2a89c9988cfd0703375800c21db61ff5 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2cWV4FScHwo.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8b32de72af08da4c391290aabf5e3f5f7f5b8ca9e32d6843306ab37a39dcee94 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2cZw98AkOjg.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2cZw98AkOjg.pt new file mode 100644 index 0000000000000000000000000000000000000000..7afc048a22021b03955ae48f74df81e64cece3c9 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2cZw98AkOjg.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:87ae0ba7c4060fcb2f897c64002c36aa2bdc37daac8fc9f1aed91df17f9db3ac +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2cfhKpB-lpk.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2cfhKpB-lpk.pt new file mode 100644 index 0000000000000000000000000000000000000000..59566b508e359b51240ad75d409434a62779bbfa --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2cfhKpB-lpk.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bb0d03cfe116813a7afff0bbc9b2b9f778ac18df2b6ed878babacfadf2412f24 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2ch1vXd1POA.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2ch1vXd1POA.pt new file mode 100644 index 0000000000000000000000000000000000000000..777f474030bae582f0580e4912fb82ebd6bd400a --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2ch1vXd1POA.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cf282b4be7c30d4839c81f784c11e7ce605242072a4f6566495a39b82720d45e +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2cs8TbMNYX4.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2cs8TbMNYX4.pt new file mode 100644 index 0000000000000000000000000000000000000000..dfdac1b94a190b3e4631495d00f75a47c767a68d --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2cs8TbMNYX4.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a36687db712c30f8e88c36bcfd4c67222a74d80578b58172d1084201750cd7b2 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2ctbaqT5KN8.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2ctbaqT5KN8.pt new file mode 100644 index 0000000000000000000000000000000000000000..125308383c5197192adf5110e597909610f20e3d --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2ctbaqT5KN8.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6cecab8a18a9d2d470ca27b96a3ad8c67daa571de2472185c419a6bb2bf78768 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2cwADEoQ-_A.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2cwADEoQ-_A.pt new file mode 100644 index 0000000000000000000000000000000000000000..c7c1e03a33d61cd42d3aff28ef99b91c436d02ec --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2cwADEoQ-_A.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bbaa8361c86774e43233f221844fb67aa40c21195257bf77209efcf6ff08bc3d +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2cyKKEWgOBA.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2cyKKEWgOBA.pt new file mode 100644 index 0000000000000000000000000000000000000000..7fc53a1e353157e5658f448f28e7619d89bfcf9f --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2cyKKEWgOBA.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:092967b85edbe6a37bf5ebaba6b3d24dd92a03fbe6c1fd9e4990ed2f0164a362 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2d09b9Bb7IM.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2d09b9Bb7IM.pt new file mode 100644 index 0000000000000000000000000000000000000000..eb4c1f674c0adfee518610d04ca0a3757cf6fce8 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2d09b9Bb7IM.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:379846269a1f7a9123660ff0358716271b82320cb1fe6d3da710afcd5278f8b1 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2dAhdkT8V5M.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2dAhdkT8V5M.pt new file mode 100644 index 0000000000000000000000000000000000000000..3ed3880d624e042207b60ff55d3784f27f3248bc --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2dAhdkT8V5M.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:783b8a869e55616baa3815b33313160600f67102e78ad8bb8eb714d977bd4735 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2dKFEsinqXQ.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2dKFEsinqXQ.pt new file mode 100644 index 0000000000000000000000000000000000000000..4b712f21d0115dd09f1dc37273c340522c9afe41 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2dKFEsinqXQ.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:92c5ca2ab3c1bde8758163d44a6e92ce4d9ffc6a9d43cbf14b8bd60163df35e2 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2dS2WajkfNI.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2dS2WajkfNI.pt new file mode 100644 index 0000000000000000000000000000000000000000..2f1cc173944a2f303b06fce5cdb077361ddbb8e1 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2dS2WajkfNI.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:128b6d5b8a0f1886dade5562b560cc2ad4ac998f17bb95b4396eeaf1254c822e +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2dUm_1MOnTE.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2dUm_1MOnTE.pt new file mode 100644 index 0000000000000000000000000000000000000000..679dd33b6eff9a361e067508b4cc664608aa46cd --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2dUm_1MOnTE.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d84a6a0727486e2653b5fac1805053d5e76100a9dc0ea01d6a8b9e7884a803a4 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2dbC0aUFp2c.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2dbC0aUFp2c.pt new file mode 100644 index 0000000000000000000000000000000000000000..9207a7a2422c81a3ba3fa212dfb7fdc25a5fd5ad --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2dbC0aUFp2c.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:13299339bb97884a08bc979b89b1793cbb3b695ceab24aa326aa958df3d5da58 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2deBNDJxDHw.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2deBNDJxDHw.pt new file mode 100644 index 0000000000000000000000000000000000000000..213329dd97681874f3870de845a3aa6fa0519229 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2deBNDJxDHw.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f5ae12a24dbd1473655ccb3c0d664fee9eb8227d5794adf4baf2dc4e5a40ae73 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2dgRoetfIBc.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2dgRoetfIBc.pt new file mode 100644 index 0000000000000000000000000000000000000000..54c97cd58f7f9381abc7d60470e84f0e5a09fe97 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2dgRoetfIBc.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bc24671cf7f1cc1e7f9a545d8974e21ed5cf2770e7963b0cbd36a4fbb16aded6 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2dhlcQhq37Y.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2dhlcQhq37Y.pt new file mode 100644 index 0000000000000000000000000000000000000000..ecda3bf58e65d13b8dc4f17a8f53314c292bcc13 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2dhlcQhq37Y.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:faae17cc0630bb7517993519cfdb54b2fc1d499a77271b2d1b980ea55571f750 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2dkgIRLJnGQ.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2dkgIRLJnGQ.pt new file mode 100644 index 0000000000000000000000000000000000000000..de85cae01a4775b4de902802ec181ac264272641 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2dkgIRLJnGQ.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:681140898b2e98796a930fb74dd912642c196a432f40e25cc049cf2ba8e250ec +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2dnCH0kAb1A.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2dnCH0kAb1A.pt new file mode 100644 index 0000000000000000000000000000000000000000..b9e8a0eb4719713ecada4be7df630d9adc79f80e --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2dnCH0kAb1A.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c0110b5a5c2014ab46344a1139b3208be58bcd67c243412d34ae62cbfc7756ff +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2do3Hb43O4c.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2do3Hb43O4c.pt new file mode 100644 index 0000000000000000000000000000000000000000..589662d52af72b34201c7117b15b547c63b0ba73 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2do3Hb43O4c.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:97bee90f957050c94c22144baeb11df714aea7aa873eff8cb63c8fbfe14e7e1c +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2dtma7PiQn4.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2dtma7PiQn4.pt new file mode 100644 index 0000000000000000000000000000000000000000..6c5f9fd591d89128f8250164e685970dcfcdbe71 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2dtma7PiQn4.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5c88a5f89b902fc65c7db7335c2bf93467acb2bba593080d11e72bad1706424d +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2dtmzqXQIaA.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2dtmzqXQIaA.pt new file mode 100644 index 0000000000000000000000000000000000000000..7daefab019b49f7e5572580b878a5eaabe365b82 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2dtmzqXQIaA.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:97d5cd729c512ebfd88f16ae1d1d7f543d954b1b38d24299facfd177ae48199c +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2duCicHUPdQ.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2duCicHUPdQ.pt new file mode 100644 index 0000000000000000000000000000000000000000..fd2e4f3a6c5994c4459dfc05b659c4be0da876b2 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2duCicHUPdQ.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8e73232badf7383f4d2a49cc557ddd5e2f8069de692b5cf0c625e584587ef30b +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2dvNAr9mksQ.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2dvNAr9mksQ.pt new file mode 100644 index 0000000000000000000000000000000000000000..389d8d65020023618bae8783aa7bd91963f7d68c --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2dvNAr9mksQ.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:917df330269bf99929becbb52d0e3d96c616388cb78fcc616e04e1d918adac03 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2e2Fi7wrKhQ.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2e2Fi7wrKhQ.pt new file mode 100644 index 0000000000000000000000000000000000000000..fd0b7c3e3e553189b0e53e297d3d3972bf312088 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2e2Fi7wrKhQ.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6a8fc859ffba069592e7dc832c0042d44ce11ab4a89c9e27d6f3a469b279c758 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2e50S3U_K8I.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2e50S3U_K8I.pt new file mode 100644 index 0000000000000000000000000000000000000000..c7bef5a87f7aa48a520174ac12e3ad73e3cb695c --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2e50S3U_K8I.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6b81b6e87f2d04f97a10b936ebb5f7c649c7067a1abffe83eb467f2746747c27 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2e63Tn1wl0A.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2e63Tn1wl0A.pt new file mode 100644 index 0000000000000000000000000000000000000000..f7ce5aaee208b76de7a5997c8c86f7e3cf2e8239 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2e63Tn1wl0A.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:03b48df1ab1831cfae8fa98d8848659182f5d509e803a5e2fab0b3108a84441c +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2e8RPtxeigM.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2e8RPtxeigM.pt new file mode 100644 index 0000000000000000000000000000000000000000..2f5c0a5656de5a0e09ee3b853aa3f46b0d4e6703 --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2e8RPtxeigM.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1c1ff68aaab1ba1f91206f12b43449f376cb57724592549cba1422ef60971236 +size 18876744 diff --git a/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2eANvJGuXGg.pt b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2eANvJGuXGg.pt new file mode 100644 index 0000000000000000000000000000000000000000..1e45ffae16fbb98549dad2c07707e0f21b4d7ace --- /dev/null +++ b/ADA/ADAEncoder/ADAEncoder/vit_tokens_v2/part-002/ytb_2eANvJGuXGg.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:17a662a9ce7a8f85e1b907cc647e8d76d632421f27bb8d34a38f262fc5cd0ee5 +size 18876744 diff --git a/ADA/_shards/small_files_summary.json b/ADA/_shards/small_files_summary.json new file mode 100644 index 0000000000000000000000000000000000000000..1bc02e702c38b3992cbefccd78a891cb4263e729 --- /dev/null +++ b/ADA/_shards/small_files_summary.json @@ -0,0 +1,60 @@ +{ + "source": "ADA", + "threshold_bytes": 16777216, + "shard_count": 4, + "shards": [ + { + "index": 0, + "payload_bytes": 35319850919, + "entries": 20781, + "filename": "small_files_lt16MiB_000_of_004.tar" + }, + { + "index": 1, + "payload_bytes": 35319850920, + "entries": 20647, + "filename": "small_files_lt16MiB_001_of_004.tar" + }, + { + "index": 2, + "payload_bytes": 35319850920, + "entries": 20647, + "filename": "small_files_lt16MiB_002_of_004.tar" + }, + { + "index": 3, + "payload_bytes": 35319850919, + "entries": 20647, + "filename": "small_files_lt16MiB_003_of_004.tar" + } + ], + "stats": { + "ada_large_files": 69664, + "ada_large_bytes": 5320554797449, + "ada_small_files": 82645, + "ada_small_bytes": 141279403678, + "ada_symlinks": 77, + "scale_files": 17530, + "scale_bytes": 1686535040349, + "scale_symlinks": 4, + "skipped_special": 0 + }, + "excluded_directory_names": [ + ".agents", + ".cache", + ".claude", + ".codex", + ".git", + ".hf_cache", + ".venv", + "__pycache__", + "hf_cache", + "node_modules", + "site-packages", + "venv" + ], + "excluded_file_names": [ + ".DS_Store", + ".env" + ] +} diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1256_magface_fake_quality_full.launcher.log b/ScaleDF_rebuttal/_host_logs/20260726_1256_magface_fake_quality_full.launcher.log new file mode 100644 index 0000000000000000000000000000000000000000..8f613d2bbd2c470651482279e13304e9a3fa552e --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1256_magface_fake_quality_full.launcher.log @@ -0,0 +1,96 @@ +tag=20260726_1256_magface_fake_quality_full +host=n116-041-181 +root=/dev/shm/scaledf_bind_storage/ScaleDF/train +base=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/analysis/magface_fake_quality_20260726_1256_magface_fake_quality_full +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/quality_eval/magface/pretrained/magface_iresnet50_official.pth +arch=iresnet50 batch_size=1024 workers=8 nproc=8 +2026-07-26T12:56:18Z shard_rows=500000 +2026-07-26T12:56:26Z shard_rows=1000000 +2026-07-26T12:56:33Z shard_rows=1500000 +2026-07-26T12:56:41Z shard_rows=2000000 +2026-07-26T12:56:49Z shard_rows=2500000 +2026-07-26T12:56:57Z shard_rows=3000000 +2026-07-26T12:57:05Z shard_rows=3500000 +2026-07-26T12:57:14Z shard_rows=4000000 +2026-07-26T12:57:22Z shard_rows=4500000 +2026-07-26T12:57:30Z shard_rows=5000000 +2026-07-26T12:57:38Z shard_rows=5500000 +2026-07-26T12:57:46Z shard_rows=6000000 +2026-07-26T12:57:54Z shard_rows=6500000 +2026-07-26T12:58:02Z shard_rows=7000000 +2026-07-26T12:58:10Z shard_rows=7500000 +2026-07-26T12:58:17Z shard_rows=8000000 +2026-07-26T12:58:25Z shard_rows=8500000 +2026-07-26T12:58:32Z shard_rows=9000000 +rows=9067198 methods=91 +/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/analysis/magface_fake_quality_20260726_1256_magface_fake_quality_full/shards/scaledf_fake_split_summary.tsv +W0726 12:58:48.460000 2127 torch/distributed/run.py:766] +W0726 12:58:48.460000 2127 torch/distributed/run.py:766] ***************************************** +W0726 12:58:48.460000 2127 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 12:58:48.460000 2127 torch/distributed/run.py:766] ***************************************** +2026-07-26T12:59:38Z rank=0/8 processed=200704/1133400 bad=0 ips=5288.1 +2026-07-26T12:59:39Z rank=3/8 processed=200704/1133400 bad=0 ips=5250.1 +2026-07-26T12:59:39Z rank=4/8 processed=200704/1133400 bad=0 ips=5216.5 +2026-07-26T12:59:39Z rank=6/8 processed=200704/1133399 bad=0 ips=5238.9 +2026-07-26T12:59:40Z rank=2/8 processed=200704/1133400 bad=0 ips=4999.1 +2026-07-26T12:59:41Z rank=1/8 processed=200704/1133400 bad=0 ips=5047.5 +2026-07-26T12:59:41Z rank=7/8 processed=200704/1133399 bad=0 ips=5033.3 +2026-07-26T12:59:41Z rank=5/8 processed=200704/1133400 bad=0 ips=5016.2 +2026-07-26T13:00:14Z rank=0/8 processed=400384/1133400 bad=0 ips=5393.8 +2026-07-26T13:00:15Z rank=4/8 processed=400384/1133400 bad=0 ips=5356.9 +2026-07-26T13:00:15Z rank=3/8 processed=400384/1133400 bad=0 ips=5357.9 +2026-07-26T13:00:16Z rank=6/8 processed=400384/1133399 bad=0 ips=5341.8 +2026-07-26T13:00:17Z rank=2/8 processed=400384/1133400 bad=0 ips=5212.3 +2026-07-26T13:00:17Z rank=5/8 processed=400384/1133400 bad=0 ips=5249.7 +2026-07-26T13:00:18Z rank=1/8 processed=400384/1133400 bad=0 ips=5200.5 +2026-07-26T13:00:19Z rank=7/8 processed=400384/1133399 bad=0 ips=5133.0 +2026-07-26T13:00:52Z rank=4/8 processed=600064/1133400 bad=0 ips=5372.3 +2026-07-26T13:00:53Z rank=3/8 processed=600064/1133400 bad=0 ips=5361.1 +2026-07-26T13:00:53Z rank=6/8 processed=600064/1133399 bad=0 ips=5364.9 +2026-07-26T13:00:53Z rank=0/8 processed=600064/1133400 bad=0 ips=5299.2 +2026-07-26T13:00:54Z rank=2/8 processed=600064/1133400 bad=0 ips=5281.6 +2026-07-26T13:00:54Z rank=7/8 processed=600064/1133399 bad=0 ips=5305.7 +2026-07-26T13:00:56Z rank=5/8 processed=600064/1133400 bad=0 ips=5231.8 +2026-07-26T13:00:57Z rank=1/8 processed=600064/1133400 bad=0 ips=5177.8 +2026-07-26T13:01:29Z rank=4/8 processed=800768/1133400 bad=0 ips=5393.4 +2026-07-26T13:01:29Z rank=6/8 processed=800768/1133399 bad=0 ips=5401.4 +2026-07-26T13:01:30Z rank=3/8 processed=800768/1133400 bad=0 ips=5376.1 +2026-07-26T13:01:31Z rank=0/8 processed=800768/1133400 bad=0 ips=5321.3 +2026-07-26T13:01:32Z rank=2/8 processed=800768/1133400 bad=0 ips=5287.4 +2026-07-26T13:01:32Z rank=7/8 processed=800768/1133399 bad=0 ips=5310.6 +2026-07-26T13:01:32Z rank=5/8 processed=800768/1133400 bad=0 ips=5296.4 +2026-07-26T13:01:33Z rank=1/8 processed=800768/1133400 bad=0 ips=5254.3 +2026-07-26T13:02:06Z rank=3/8 processed=1000448/1133400 bad=0 ips=5382.2 +2026-07-26T13:02:07Z rank=6/8 processed=1000448/1133399 bad=0 ips=5382.1 +2026-07-26T13:02:07Z rank=4/8 processed=1000448/1133400 bad=0 ips=5365.7 +2026-07-26T13:02:09Z rank=5/8 processed=1000448/1133400 bad=0 ips=5317.4 +2026-07-26T13:02:09Z rank=0/8 processed=1000448/1133400 bad=0 ips=5297.8 +2026-07-26T13:02:09Z rank=7/8 processed=1000448/1133399 bad=0 ips=5313.4 +2026-07-26T13:02:09Z rank=2/8 processed=1000448/1133400 bad=0 ips=5287.4 +2026-07-26T13:02:11Z rank=1/8 processed=1000448/1133400 bad=0 ips=5258.2 +2026-07-26T13:02:32Z rank=3 done processed=1133400 bad=0 elapsed=211.2s +2026-07-26T13:02:33Z rank=4 done processed=1133400 bad=0 elapsed=212.1s +2026-07-26T13:02:33Z rank=6 done processed=1133399 bad=0 elapsed=212.1s +2026-07-26T13:02:34Z rank=5 done processed=1133400 bad=0 elapsed=213.6s +2026-07-26T13:02:37Z rank=1 done processed=1133400 bad=0 elapsed=215.8s +2026-07-26T13:02:37Z rank=0 done processed=1133400 bad=0 elapsed=216.4s +2026-07-26T13:02:37Z rank=7 done processed=1133399 bad=0 elapsed=215.9s +2026-07-26T13:02:38Z rank=2 done processed=1133400 bad=0 elapsed=217.7s +{ + "n": 9067198, + "methods": 91, + "mean": 30.204923629760742, + "std": 2.2446415424346924, + "p01": 25.12969762802124, + "p05": 26.612859439849853, + "p10": 27.381268882751463, + "p25": 28.67769432067871, + "p50": 30.15822982788086, + "p75": 31.689302444458008, + "p90": 33.10612602233887, + "p95": 33.96895217895508, + "p99": 35.60852836608886 +} +/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/analysis/magface_fake_quality_20260726_1256_magface_fake_quality_full/summary/magface_fake_quality_method_quality_summary.tsv +done_utc=2026-07-26T13:03:11Z +summary=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/analysis/magface_fake_quality_20260726_1256_magface_fake_quality_full/summary diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1529_vith32_sync_from_n111-046-078/sync_manifest.tsv b/ScaleDF_rebuttal/_host_logs/20260726_1529_vith32_sync_from_n111-046-078/sync_manifest.tsv new file mode 100644 index 0000000000000000000000000000000000000000..69c86d4ed77a7789bfdbb40e658a32560faa9fcf --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1529_vith32_sync_from_n111-046-078/sync_manifest.tsv @@ -0,0 +1,4 @@ +label\tport\thost\tname +new1\t10913\tfdbd:dccd:cdc1:1368:0:2f5::\tn116-042-185 +new2\t11661\tfdbd:dccd:cdc1:1368:0:2e::\tn116-040-048 +new3\t11634\tfdbd:dccd:cdc1:1302:0:b9::\tn111-046-125 diff --git "a/ScaleDF_rebuttal/_host_logs/20260726_1529_vith32_sync_from_n111-046-078/sync_new1_tfdbd:dccd:cdc1:1368:0:2f5::\\tn116-042-185.log" "b/ScaleDF_rebuttal/_host_logs/20260726_1529_vith32_sync_from_n111-046-078/sync_new1_tfdbd:dccd:cdc1:1368:0:2f5::\\tn116-042-185.log" new file mode 100644 index 0000000000000000000000000000000000000000..99a26a7c0a80d232ad91d33561e28e4bb1e1c8a4 --- /dev/null +++ "b/ScaleDF_rebuttal/_host_logs/20260726_1529_vith32_sync_from_n111-046-078/sync_new1_tfdbd:dccd:cdc1:1368:0:2f5::\\tn116-042-185.log" @@ -0,0 +1,6 @@ +/home/tiger/.profile: line 29: /run/bin/env: No such file or directory +/home/tiger/.profile: line 31: /tmp/m9/bin/env: No such file or directory +started_utc=2026-07-26T15:29:36Z +source=n111-046-078 +dest=tfdbd:dccd:cdc1:1368:0:2f5::tn116-042-185 port= host=10913 +Bad port 'tiger@10913' diff --git "a/ScaleDF_rebuttal/_host_logs/20260726_1529_vith32_sync_from_n111-046-078/sync_new1_tfdbd:dccd:cdc1:1368:0:2f5::\\tn116-042-185.pid" "b/ScaleDF_rebuttal/_host_logs/20260726_1529_vith32_sync_from_n111-046-078/sync_new1_tfdbd:dccd:cdc1:1368:0:2f5::\\tn116-042-185.pid" new file mode 100644 index 0000000000000000000000000000000000000000..32ed469e84efd08f1e3e57c2bd061649e0800699 --- /dev/null +++ "b/ScaleDF_rebuttal/_host_logs/20260726_1529_vith32_sync_from_n111-046-078/sync_new1_tfdbd:dccd:cdc1:1368:0:2f5::\\tn116-042-185.pid" @@ -0,0 +1 @@ +1786 diff --git "a/ScaleDF_rebuttal/_host_logs/20260726_1529_vith32_sync_from_n111-046-078/sync_new2_tfdbd:dccd:cdc1:1368:0:2e::\\tn116-040-048.log" "b/ScaleDF_rebuttal/_host_logs/20260726_1529_vith32_sync_from_n111-046-078/sync_new2_tfdbd:dccd:cdc1:1368:0:2e::\\tn116-040-048.log" new file mode 100644 index 0000000000000000000000000000000000000000..c1eb1290e98f2ed273e511a767fab9dc4c081ff3 --- /dev/null +++ "b/ScaleDF_rebuttal/_host_logs/20260726_1529_vith32_sync_from_n111-046-078/sync_new2_tfdbd:dccd:cdc1:1368:0:2e::\\tn116-040-048.log" @@ -0,0 +1,6 @@ +/home/tiger/.profile: line 29: /run/bin/env: No such file or directory +/home/tiger/.profile: line 31: /tmp/m9/bin/env: No such file or directory +started_utc=2026-07-26T15:29:36Z +source=n111-046-078 +dest=tfdbd:dccd:cdc1:1368:0:2e::tn116-040-048 port= host=11661 +Bad port 'tiger@11661' diff --git "a/ScaleDF_rebuttal/_host_logs/20260726_1529_vith32_sync_from_n111-046-078/sync_new2_tfdbd:dccd:cdc1:1368:0:2e::\\tn116-040-048.pid" "b/ScaleDF_rebuttal/_host_logs/20260726_1529_vith32_sync_from_n111-046-078/sync_new2_tfdbd:dccd:cdc1:1368:0:2e::\\tn116-040-048.pid" new file mode 100644 index 0000000000000000000000000000000000000000..197a77e90c6bdbddd584bdfe22a19344f8922722 --- /dev/null +++ "b/ScaleDF_rebuttal/_host_logs/20260726_1529_vith32_sync_from_n111-046-078/sync_new2_tfdbd:dccd:cdc1:1368:0:2e::\\tn116-040-048.pid" @@ -0,0 +1 @@ +1788 diff --git "a/ScaleDF_rebuttal/_host_logs/20260726_1529_vith32_sync_from_n111-046-078/sync_new3_tfdbd:dccd:cdc1:1302:0:b9::\\tn111-046-125.log" "b/ScaleDF_rebuttal/_host_logs/20260726_1529_vith32_sync_from_n111-046-078/sync_new3_tfdbd:dccd:cdc1:1302:0:b9::\\tn111-046-125.log" new file mode 100644 index 0000000000000000000000000000000000000000..0fb66b36883ccf161b51f4a7feff8452478dbd5f --- /dev/null +++ "b/ScaleDF_rebuttal/_host_logs/20260726_1529_vith32_sync_from_n111-046-078/sync_new3_tfdbd:dccd:cdc1:1302:0:b9::\\tn111-046-125.log" @@ -0,0 +1,6 @@ +/home/tiger/.profile: line 29: /run/bin/env: No such file or directory +/home/tiger/.profile: line 31: /tmp/m9/bin/env: No such file or directory +started_utc=2026-07-26T15:29:36Z +source=n111-046-078 +dest=tfdbd:dccd:cdc1:1302:0:b9::tn111-046-125 port= host=11634 +Bad port 'tiger@11634' diff --git "a/ScaleDF_rebuttal/_host_logs/20260726_1529_vith32_sync_from_n111-046-078/sync_new3_tfdbd:dccd:cdc1:1302:0:b9::\\tn111-046-125.pid" "b/ScaleDF_rebuttal/_host_logs/20260726_1529_vith32_sync_from_n111-046-078/sync_new3_tfdbd:dccd:cdc1:1302:0:b9::\\tn111-046-125.pid" new file mode 100644 index 0000000000000000000000000000000000000000..81f2d5209ce419cd91c5c48d6a520567b07dbbd5 --- /dev/null +++ "b/ScaleDF_rebuttal/_host_logs/20260726_1529_vith32_sync_from_n111-046-078/sync_new3_tfdbd:dccd:cdc1:1302:0:b9::\\tn111-046-125.pid" @@ -0,0 +1 @@ +1792 diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1531_vith32_sync_from_n111-046-078/sync_manifest.tsv b/ScaleDF_rebuttal/_host_logs/20260726_1531_vith32_sync_from_n111-046-078/sync_manifest.tsv new file mode 100644 index 0000000000000000000000000000000000000000..d56da6ae48dcfcb343c48d296a5d606bf84060f4 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1531_vith32_sync_from_n111-046-078/sync_manifest.tsv @@ -0,0 +1,4 @@ +label port host name +new1 10913 fdbd:dccd:cdc1:1368:0:2f5:: n116-042-185 +new2 11661 fdbd:dccd:cdc1:1368:0:2e:: n116-040-048 +new3 11634 fdbd:dccd:cdc1:1302:0:b9:: n111-046-125 diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1531_vith32_sync_from_n111-046-078/sync_new1_n116-042-185.pid b/ScaleDF_rebuttal/_host_logs/20260726_1531_vith32_sync_from_n111-046-078/sync_new1_n116-042-185.pid new file mode 100644 index 0000000000000000000000000000000000000000..12ac2f5ef1f1b1eced732b35165fb748c97bd8c6 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1531_vith32_sync_from_n111-046-078/sync_new1_n116-042-185.pid @@ -0,0 +1 @@ +1833 diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1531_vith32_sync_from_n111-046-078/sync_new2_n116-040-048.pid b/ScaleDF_rebuttal/_host_logs/20260726_1531_vith32_sync_from_n111-046-078/sync_new2_n116-040-048.pid new file mode 100644 index 0000000000000000000000000000000000000000..20584201a1e7ec42480ca1fbe6cc692e60510008 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1531_vith32_sync_from_n111-046-078/sync_new2_n116-040-048.pid @@ -0,0 +1 @@ +1836 diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1531_vith32_sync_from_n111-046-078/sync_new3_n111-046-125.pid b/ScaleDF_rebuttal/_host_logs/20260726_1531_vith32_sync_from_n111-046-078/sync_new3_n111-046-125.pid new file mode 100644 index 0000000000000000000000000000000000000000..c5a86fa8bb561d8ec9b0f893ab9be4bf8e525099 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1531_vith32_sync_from_n111-046-078/sync_new3_n111-046-125.pid @@ -0,0 +1 @@ +1839 diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1548_backup_sync_from_n111-046-078/sync_backup1_n116-042-235.pid b/ScaleDF_rebuttal/_host_logs/20260726_1548_backup_sync_from_n111-046-078/sync_backup1_n116-042-235.pid new file mode 100644 index 0000000000000000000000000000000000000000..e64ac59301f60b62056b09017868a2f522dba4d0 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1548_backup_sync_from_n111-046-078/sync_backup1_n116-042-235.pid @@ -0,0 +1 @@ +2243 diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1548_backup_sync_from_n111-046-078/sync_backup2_n116-043-253.pid b/ScaleDF_rebuttal/_host_logs/20260726_1548_backup_sync_from_n111-046-078/sync_backup2_n116-043-253.pid new file mode 100644 index 0000000000000000000000000000000000000000..63a8c52ab62cf328354d1778475d4c747863d4d5 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1548_backup_sync_from_n111-046-078/sync_backup2_n116-043-253.pid @@ -0,0 +1 @@ +2246 diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1552_vith32_launch/launcher_rank0_n111-046-078.log b/ScaleDF_rebuttal/_host_logs/20260726_1552_vith32_launch/launcher_rank0_n111-046-078.log new file mode 100644 index 0000000000000000000000000000000000000000..ac3ed59dd319dfce8673dab4d30c4e048305641b --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1552_vith32_launch/launcher_rank0_n111-046-078.log @@ -0,0 +1,64 @@ +started_utc=2026-07-26T15:52:17Z +hostname=n111-046-078 +node_rank=0 +nnodes=4 +nproc_per_node=8 +world_size=32 +master_addr=10.111.46.78 +master_port=29647 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +main=main_cross_2direct_balance_realscale_perturb_v4_accum.py +model_name=deit_huge_patch14_LS +pretrain=/tmp/ScaleDF_repro_20260724/pretrained/deit_3_huge_224_21k_v1.pth +resume=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_24card_effbs3072_from_epoch7_n116-041-246_n116-042-186_n116-042-185_20260726_0736_vith24_from_epoch7/checkpoint.pth +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_32card_effbs2048_from_latest24_n111-046-078_n116-042-185_n116-040-048_n111-046-125_20260726_1552_vith32 +batch_per_gpu=64 +accum_iter=1 +effective_batch=2048 +eval_batch_per_gpu=64 +epochs=40 +W0726 15:52:31.291000 2336 torch/distributed/run.py:766] +W0726 15:52:31.291000 2336 torch/distributed/run.py:766] ***************************************** +W0726 15:52:31.291000 2336 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 15:52:31.291000 2336 torch/distributed/run.py:766] ***************************************** +Traceback (most recent call last): + File "/usr/local/bin/torchrun", line 7, in + sys.exit(main()) + ^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 355, in wrapper + return f(*args, **kwargs) + ^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 892, in main + run(args) + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 883, in run + elastic_launch( + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 139, in __call__ + return launch_agent(self._config, self._entrypoint, list(args)) + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 261, in launch_agent + result = agent.run() + ^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/metrics/api.py", line 138, in wrapper + result = f(*args, **kwargs) + ^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/agent/server/api.py", line 711, in run + result = self._invoke_run(role) + ^^^^^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/agent/server/api.py", line 864, in _invoke_run + self._initialize_workers(self._worker_group) + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/metrics/api.py", line 138, in wrapper + result = f(*args, **kwargs) + ^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/agent/server/api.py", line 683, in _initialize_workers + self._rendezvous(worker_group) + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/metrics/api.py", line 138, in wrapper + result = f(*args, **kwargs) + ^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/agent/server/api.py", line 500, in _rendezvous + rdzv_info = spec.rdzv_handler.next_rendezvous() + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/rendezvous/static_tcp_rendezvous.py", line 67, in next_rendezvous + self._store = TCPStore( # type: ignore[call-arg] + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ +torch.distributed.DistNetworkError: The server socket has failed to listen on any local network address. port: 29647, useIpv6: false, code: -98, name: EADDRINUSE, message: address already in use diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1552_vith32_launch/launcher_rank1_n116-042-185.log b/ScaleDF_rebuttal/_host_logs/20260726_1552_vith32_launch/launcher_rank1_n116-042-185.log new file mode 100644 index 0000000000000000000000000000000000000000..ac897cd1d32cddb253d14b4f4ac8cab7dd345058 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1552_vith32_launch/launcher_rank1_n116-042-185.log @@ -0,0 +1,25 @@ +started_utc=2026-07-26T15:52:17Z +hostname=n116-042-185 +node_rank=1 +nnodes=4 +nproc_per_node=8 +world_size=32 +master_addr=10.111.46.78 +master_port=29647 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +main=main_cross_2direct_balance_realscale_perturb_v4_accum.py +model_name=deit_huge_patch14_LS +pretrain=/tmp/ScaleDF_repro_20260724/pretrained/deit_3_huge_224_21k_v1.pth +resume=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_24card_effbs3072_from_epoch7_n116-041-246_n116-042-186_n116-042-185_20260726_0736_vith24_from_epoch7/checkpoint.pth +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_32card_effbs2048_from_latest24_n111-046-078_n116-042-185_n116-040-048_n111-046-125_20260726_1552_vith32 +batch_per_gpu=64 +accum_iter=1 +effective_batch=2048 +eval_batch_per_gpu=64 +epochs=40 +W0726 15:52:50.360000 1008 torch/distributed/run.py:766] +W0726 15:52:50.360000 1008 torch/distributed/run.py:766] ***************************************** +W0726 15:52:50.360000 1008 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 15:52:50.360000 1008 torch/distributed/run.py:766] ***************************************** +Terminated diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1552_vith32_launch/launcher_rank2_n116-040-048.log b/ScaleDF_rebuttal/_host_logs/20260726_1552_vith32_launch/launcher_rank2_n116-040-048.log new file mode 100644 index 0000000000000000000000000000000000000000..135a8079172606ee8581e95868400e350faa92d2 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1552_vith32_launch/launcher_rank2_n116-040-048.log @@ -0,0 +1,25 @@ +started_utc=2026-07-26T15:52:17Z +hostname=n116-040-048 +node_rank=2 +nnodes=4 +nproc_per_node=8 +world_size=32 +master_addr=10.111.46.78 +master_port=29647 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +main=main_cross_2direct_balance_realscale_perturb_v4_accum.py +model_name=deit_huge_patch14_LS +pretrain=/tmp/ScaleDF_repro_20260724/pretrained/deit_3_huge_224_21k_v1.pth +resume=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_24card_effbs3072_from_epoch7_n116-041-246_n116-042-186_n116-042-185_20260726_0736_vith24_from_epoch7/checkpoint.pth +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_32card_effbs2048_from_latest24_n111-046-078_n116-042-185_n116-040-048_n111-046-125_20260726_1552_vith32 +batch_per_gpu=64 +accum_iter=1 +effective_batch=2048 +eval_batch_per_gpu=64 +epochs=40 +W0726 15:52:43.550000 931 torch/distributed/run.py:766] +W0726 15:52:43.550000 931 torch/distributed/run.py:766] ***************************************** +W0726 15:52:43.550000 931 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 15:52:43.550000 931 torch/distributed/run.py:766] ***************************************** +Terminated diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1552_vith32_launch/launcher_rank3_n111-046-125.log b/ScaleDF_rebuttal/_host_logs/20260726_1552_vith32_launch/launcher_rank3_n111-046-125.log new file mode 100644 index 0000000000000000000000000000000000000000..66862938c72bd7e8526dce93a80bc54c9e729c8e --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1552_vith32_launch/launcher_rank3_n111-046-125.log @@ -0,0 +1,25 @@ +started_utc=2026-07-26T15:52:17Z +hostname=n111-046-125 +node_rank=3 +nnodes=4 +nproc_per_node=8 +world_size=32 +master_addr=10.111.46.78 +master_port=29647 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +main=main_cross_2direct_balance_realscale_perturb_v4_accum.py +model_name=deit_huge_patch14_LS +pretrain=/tmp/ScaleDF_repro_20260724/pretrained/deit_3_huge_224_21k_v1.pth +resume=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_24card_effbs3072_from_epoch7_n116-041-246_n116-042-186_n116-042-185_20260726_0736_vith24_from_epoch7/checkpoint.pth +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_32card_effbs2048_from_latest24_n111-046-078_n116-042-185_n116-040-048_n111-046-125_20260726_1552_vith32 +batch_per_gpu=64 +accum_iter=1 +effective_batch=2048 +eval_batch_per_gpu=64 +epochs=40 +W0726 15:52:48.112000 810 torch/distributed/run.py:766] +W0726 15:52:48.112000 810 torch/distributed/run.py:766] ***************************************** +W0726 15:52:48.112000 810 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 15:52:48.112000 810 torch/distributed/run.py:766] ***************************************** +Terminated diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1552_vith32_launch/output_root.txt b/ScaleDF_rebuttal/_host_logs/20260726_1552_vith32_launch/output_root.txt new file mode 100644 index 0000000000000000000000000000000000000000..18008f6886cc364aa68e451da7a4d9799b6d6b88 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1552_vith32_launch/output_root.txt @@ -0,0 +1 @@ +/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_32card_effbs2048_from_latest24_n111-046-078_n116-042-185_n116-040-048_n111-046-125_20260726_1552_vith32 diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1552_vith32_launch/pid_rank0_n111-046-078.txt b/ScaleDF_rebuttal/_host_logs/20260726_1552_vith32_launch/pid_rank0_n111-046-078.txt new file mode 100644 index 0000000000000000000000000000000000000000..677420767673de827fde8576e251542f5ff5deea --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1552_vith32_launch/pid_rank0_n111-046-078.txt @@ -0,0 +1 @@ +2329 diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1552_vith32_launch/pid_rank1_n116-042-185.txt b/ScaleDF_rebuttal/_host_logs/20260726_1552_vith32_launch/pid_rank1_n116-042-185.txt new file mode 100644 index 0000000000000000000000000000000000000000..dd11724042e28f013e4fee09715b24c1b9ab10d1 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1552_vith32_launch/pid_rank1_n116-042-185.txt @@ -0,0 +1 @@ +1001 diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1552_vith32_launch/pid_rank2_n116-040-048.txt b/ScaleDF_rebuttal/_host_logs/20260726_1552_vith32_launch/pid_rank2_n116-040-048.txt new file mode 100644 index 0000000000000000000000000000000000000000..0dac149173152e188ec6012aaced862b8947344a --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1552_vith32_launch/pid_rank2_n116-040-048.txt @@ -0,0 +1 @@ +924 diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1552_vith32_launch/pid_rank3_n111-046-125.txt b/ScaleDF_rebuttal/_host_logs/20260726_1552_vith32_launch/pid_rank3_n111-046-125.txt new file mode 100644 index 0000000000000000000000000000000000000000..25e1ae45fdb0d05d03c81ead248e94da23b07cc0 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1552_vith32_launch/pid_rank3_n111-046-125.txt @@ -0,0 +1 @@ +803 diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1552_vith32_launch/resume.txt b/ScaleDF_rebuttal/_host_logs/20260726_1552_vith32_launch/resume.txt new file mode 100644 index 0000000000000000000000000000000000000000..2158d35929432dee54a9499c09d7990967633387 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1552_vith32_launch/resume.txt @@ -0,0 +1 @@ +/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_24card_effbs3072_from_epoch7_n116-041-246_n116-042-186_n116-042-185_20260726_0736_vith24_from_epoch7/checkpoint.pth diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1552_vith32_launch/run_name.txt b/ScaleDF_rebuttal/_host_logs/20260726_1552_vith32_launch/run_name.txt new file mode 100644 index 0000000000000000000000000000000000000000..ef95b74cc73f6b0427bfb0ab5ee1a40434a87302 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1552_vith32_launch/run_name.txt @@ -0,0 +1 @@ +modelsize_vith_40epoch_32card_effbs2048_from_latest24_n111-046-078_n116-042-185_n116-040-048_n111-046-125_20260726_1552_vith32 diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1554_vith32_launch_port29789/launcher_rank0_n111-046-078.log b/ScaleDF_rebuttal/_host_logs/20260726_1554_vith32_launch_port29789/launcher_rank0_n111-046-078.log new file mode 100644 index 0000000000000000000000000000000000000000..9fe09817d791372f3e7b1af38421d6a953c30c13 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1554_vith32_launch_port29789/launcher_rank0_n111-046-078.log @@ -0,0 +1,64 @@ +started_utc=2026-07-26T15:54:25Z +hostname=n111-046-078 +node_rank=0 +nnodes=4 +nproc_per_node=8 +world_size=32 +master_addr=10.111.46.78 +master_port=29789 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +main=main_cross_2direct_balance_realscale_perturb_v4_accum.py +model_name=deit_huge_patch14_LS +pretrain=/tmp/ScaleDF_repro_20260724/pretrained/deit_3_huge_224_21k_v1.pth +resume=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_24card_effbs3072_from_epoch7_n116-041-246_n116-042-186_n116-042-185_20260726_0736_vith24_from_epoch7/checkpoint.pth +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_32card_effbs2048_from_latest24_n111-046-078_n116-042-185_n116-040-048_n111-046-125_20260726_1554_vith32_port29789 +batch_per_gpu=64 +accum_iter=1 +effective_batch=2048 +eval_batch_per_gpu=64 +epochs=40 +W0726 15:54:38.750000 2497 torch/distributed/run.py:766] +W0726 15:54:38.750000 2497 torch/distributed/run.py:766] ***************************************** +W0726 15:54:38.750000 2497 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 15:54:38.750000 2497 torch/distributed/run.py:766] ***************************************** +Traceback (most recent call last): + File "/usr/local/bin/torchrun", line 7, in + sys.exit(main()) + ^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 355, in wrapper + return f(*args, **kwargs) + ^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 892, in main + run(args) + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 883, in run + elastic_launch( + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 139, in __call__ + return launch_agent(self._config, self._entrypoint, list(args)) + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 261, in launch_agent + result = agent.run() + ^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/metrics/api.py", line 138, in wrapper + result = f(*args, **kwargs) + ^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/agent/server/api.py", line 711, in run + result = self._invoke_run(role) + ^^^^^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/agent/server/api.py", line 864, in _invoke_run + self._initialize_workers(self._worker_group) + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/metrics/api.py", line 138, in wrapper + result = f(*args, **kwargs) + ^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/agent/server/api.py", line 683, in _initialize_workers + self._rendezvous(worker_group) + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/metrics/api.py", line 138, in wrapper + result = f(*args, **kwargs) + ^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/agent/server/api.py", line 500, in _rendezvous + rdzv_info = spec.rdzv_handler.next_rendezvous() + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/rendezvous/static_tcp_rendezvous.py", line 67, in next_rendezvous + self._store = TCPStore( # type: ignore[call-arg] + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ +torch.distributed.DistNetworkError: The server socket has failed to listen on any local network address. port: 29789, useIpv6: false, code: -98, name: EADDRINUSE, message: address already in use diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1554_vith32_launch_port29789/launcher_rank1_n116-042-185.log b/ScaleDF_rebuttal/_host_logs/20260726_1554_vith32_launch_port29789/launcher_rank1_n116-042-185.log new file mode 100644 index 0000000000000000000000000000000000000000..64c7f9dfdecc5815a0127f94778817744250bbd6 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1554_vith32_launch_port29789/launcher_rank1_n116-042-185.log @@ -0,0 +1,25 @@ +started_utc=2026-07-26T15:54:25Z +hostname=n116-042-185 +node_rank=1 +nnodes=4 +nproc_per_node=8 +world_size=32 +master_addr=10.111.46.78 +master_port=29789 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +main=main_cross_2direct_balance_realscale_perturb_v4_accum.py +model_name=deit_huge_patch14_LS +pretrain=/tmp/ScaleDF_repro_20260724/pretrained/deit_3_huge_224_21k_v1.pth +resume=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_24card_effbs3072_from_epoch7_n116-041-246_n116-042-186_n116-042-185_20260726_0736_vith24_from_epoch7/checkpoint.pth +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_32card_effbs2048_from_latest24_n111-046-078_n116-042-185_n116-040-048_n111-046-125_20260726_1554_vith32_port29789 +batch_per_gpu=64 +accum_iter=1 +effective_batch=2048 +eval_batch_per_gpu=64 +epochs=40 +W0726 15:54:42.278000 1132 torch/distributed/run.py:766] +W0726 15:54:42.278000 1132 torch/distributed/run.py:766] ***************************************** +W0726 15:54:42.278000 1132 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 15:54:42.278000 1132 torch/distributed/run.py:766] ***************************************** +Terminated diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1554_vith32_launch_port29789/launcher_rank2_n116-040-048.log b/ScaleDF_rebuttal/_host_logs/20260726_1554_vith32_launch_port29789/launcher_rank2_n116-040-048.log new file mode 100644 index 0000000000000000000000000000000000000000..7f85255474fb47acb544d52bf777af61b28b6926 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1554_vith32_launch_port29789/launcher_rank2_n116-040-048.log @@ -0,0 +1,25 @@ +started_utc=2026-07-26T15:54:25Z +hostname=n116-040-048 +node_rank=2 +nnodes=4 +nproc_per_node=8 +world_size=32 +master_addr=10.111.46.78 +master_port=29789 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +main=main_cross_2direct_balance_realscale_perturb_v4_accum.py +model_name=deit_huge_patch14_LS +pretrain=/tmp/ScaleDF_repro_20260724/pretrained/deit_3_huge_224_21k_v1.pth +resume=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_24card_effbs3072_from_epoch7_n116-041-246_n116-042-186_n116-042-185_20260726_0736_vith24_from_epoch7/checkpoint.pth +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_32card_effbs2048_from_latest24_n111-046-078_n116-042-185_n116-040-048_n111-046-125_20260726_1554_vith32_port29789 +batch_per_gpu=64 +accum_iter=1 +effective_batch=2048 +eval_batch_per_gpu=64 +epochs=40 +W0726 15:54:39.211000 1055 torch/distributed/run.py:766] +W0726 15:54:39.211000 1055 torch/distributed/run.py:766] ***************************************** +W0726 15:54:39.211000 1055 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 15:54:39.211000 1055 torch/distributed/run.py:766] ***************************************** +Terminated diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1554_vith32_launch_port29789/launcher_rank3_n111-046-125.log b/ScaleDF_rebuttal/_host_logs/20260726_1554_vith32_launch_port29789/launcher_rank3_n111-046-125.log new file mode 100644 index 0000000000000000000000000000000000000000..20fffa1b9fdb260b546a4af75c03ebb23c143cff --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1554_vith32_launch_port29789/launcher_rank3_n111-046-125.log @@ -0,0 +1,25 @@ +started_utc=2026-07-26T15:54:25Z +hostname=n111-046-125 +node_rank=3 +nnodes=4 +nproc_per_node=8 +world_size=32 +master_addr=10.111.46.78 +master_port=29789 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +main=main_cross_2direct_balance_realscale_perturb_v4_accum.py +model_name=deit_huge_patch14_LS +pretrain=/tmp/ScaleDF_repro_20260724/pretrained/deit_3_huge_224_21k_v1.pth +resume=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_24card_effbs3072_from_epoch7_n116-041-246_n116-042-186_n116-042-185_20260726_0736_vith24_from_epoch7/checkpoint.pth +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_32card_effbs2048_from_latest24_n111-046-078_n116-042-185_n116-040-048_n111-046-125_20260726_1554_vith32_port29789 +batch_per_gpu=64 +accum_iter=1 +effective_batch=2048 +eval_batch_per_gpu=64 +epochs=40 +W0726 15:54:41.123000 934 torch/distributed/run.py:766] +W0726 15:54:41.123000 934 torch/distributed/run.py:766] ***************************************** +W0726 15:54:41.123000 934 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 15:54:41.123000 934 torch/distributed/run.py:766] ***************************************** +Terminated diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1554_vith32_launch_port29789/output_root.txt b/ScaleDF_rebuttal/_host_logs/20260726_1554_vith32_launch_port29789/output_root.txt new file mode 100644 index 0000000000000000000000000000000000000000..08fddf6267b7dc708661a849ef1705d05be87c7b --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1554_vith32_launch_port29789/output_root.txt @@ -0,0 +1 @@ +/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_32card_effbs2048_from_latest24_n111-046-078_n116-042-185_n116-040-048_n111-046-125_20260726_1554_vith32_port29789 diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1554_vith32_launch_port29789/pid_rank0_n111-046-078.txt b/ScaleDF_rebuttal/_host_logs/20260726_1554_vith32_launch_port29789/pid_rank0_n111-046-078.txt new file mode 100644 index 0000000000000000000000000000000000000000..697ade4bd557fe854376c9f9d27279dc037fd7f6 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1554_vith32_launch_port29789/pid_rank0_n111-046-078.txt @@ -0,0 +1 @@ +2489 diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1554_vith32_launch_port29789/pid_rank1_n116-042-185.txt b/ScaleDF_rebuttal/_host_logs/20260726_1554_vith32_launch_port29789/pid_rank1_n116-042-185.txt new file mode 100644 index 0000000000000000000000000000000000000000..4707d3c68d27eb4806e4c9034a05b5bc9152c66f --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1554_vith32_launch_port29789/pid_rank1_n116-042-185.txt @@ -0,0 +1 @@ +1124 diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1554_vith32_launch_port29789/pid_rank2_n116-040-048.txt b/ScaleDF_rebuttal/_host_logs/20260726_1554_vith32_launch_port29789/pid_rank2_n116-040-048.txt new file mode 100644 index 0000000000000000000000000000000000000000..0eb5aea3c9a213efc805764c92eb91ea0961a194 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1554_vith32_launch_port29789/pid_rank2_n116-040-048.txt @@ -0,0 +1 @@ +1047 diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1554_vith32_launch_port29789/pid_rank3_n111-046-125.txt b/ScaleDF_rebuttal/_host_logs/20260726_1554_vith32_launch_port29789/pid_rank3_n111-046-125.txt new file mode 100644 index 0000000000000000000000000000000000000000..cdf11a8dc67461393929387dc659e6c287b72852 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1554_vith32_launch_port29789/pid_rank3_n111-046-125.txt @@ -0,0 +1 @@ +926 diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1554_vith32_launch_port29789/resume.txt b/ScaleDF_rebuttal/_host_logs/20260726_1554_vith32_launch_port29789/resume.txt new file mode 100644 index 0000000000000000000000000000000000000000..2158d35929432dee54a9499c09d7990967633387 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1554_vith32_launch_port29789/resume.txt @@ -0,0 +1 @@ +/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_24card_effbs3072_from_epoch7_n116-041-246_n116-042-186_n116-042-185_20260726_0736_vith24_from_epoch7/checkpoint.pth diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1554_vith32_launch_port29789/run_name.txt b/ScaleDF_rebuttal/_host_logs/20260726_1554_vith32_launch_port29789/run_name.txt new file mode 100644 index 0000000000000000000000000000000000000000..25c7542aa214536af948c3a327946efa3be76ef0 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1554_vith32_launch_port29789/run_name.txt @@ -0,0 +1 @@ +modelsize_vith_40epoch_32card_effbs2048_from_latest24_n111-046-078_n116-042-185_n116-040-048_n111-046-125_20260726_1554_vith32_port29789 diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1556_vith32_launch_master_n116-042-185_port29877/launcher_rank0_n116-042-185.log b/ScaleDF_rebuttal/_host_logs/20260726_1556_vith32_launch_master_n116-042-185_port29877/launcher_rank0_n116-042-185.log new file mode 100644 index 0000000000000000000000000000000000000000..e0123c4c239f42fa53974307ab9043029eb84adb --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1556_vith32_launch_master_n116-042-185_port29877/launcher_rank0_n116-042-185.log @@ -0,0 +1,64 @@ +started_utc=2026-07-26T15:56:35Z +hostname=n116-042-185 +node_rank=0 +nnodes=4 +nproc_per_node=8 +world_size=32 +master_addr=10.116.42.185 +master_port=29877 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +main=main_cross_2direct_balance_realscale_perturb_v4_accum.py +model_name=deit_huge_patch14_LS +pretrain=/tmp/ScaleDF_repro_20260724/pretrained/deit_3_huge_224_21k_v1.pth +resume=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_24card_effbs3072_from_epoch7_n116-041-246_n116-042-186_n116-042-185_20260726_0736_vith24_from_epoch7/checkpoint.pth +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_32card_effbs2048_from_latest24_master_n116-042-185_20260726_1556_vith32_port29877 +batch_per_gpu=64 +accum_iter=1 +effective_batch=2048 +eval_batch_per_gpu=64 +epochs=40 +W0726 15:56:53.434000 1261 torch/distributed/run.py:766] +W0726 15:56:53.434000 1261 torch/distributed/run.py:766] ***************************************** +W0726 15:56:53.434000 1261 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 15:56:53.434000 1261 torch/distributed/run.py:766] ***************************************** +Traceback (most recent call last): + File "/usr/local/bin/torchrun", line 7, in + sys.exit(main()) + ^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 355, in wrapper + return f(*args, **kwargs) + ^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 892, in main + run(args) + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 883, in run + elastic_launch( + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 139, in __call__ + return launch_agent(self._config, self._entrypoint, list(args)) + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 261, in launch_agent + result = agent.run() + ^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/metrics/api.py", line 138, in wrapper + result = f(*args, **kwargs) + ^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/agent/server/api.py", line 711, in run + result = self._invoke_run(role) + ^^^^^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/agent/server/api.py", line 864, in _invoke_run + self._initialize_workers(self._worker_group) + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/metrics/api.py", line 138, in wrapper + result = f(*args, **kwargs) + ^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/agent/server/api.py", line 683, in _initialize_workers + self._rendezvous(worker_group) + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/metrics/api.py", line 138, in wrapper + result = f(*args, **kwargs) + ^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/agent/server/api.py", line 500, in _rendezvous + rdzv_info = spec.rdzv_handler.next_rendezvous() + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/rendezvous/static_tcp_rendezvous.py", line 67, in next_rendezvous + self._store = TCPStore( # type: ignore[call-arg] + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ +torch.distributed.DistNetworkError: The server socket has failed to listen on any local network address. port: 29877, useIpv6: false, code: -98, name: EADDRINUSE, message: address already in use diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1556_vith32_launch_master_n116-042-185_port29877/launcher_rank1_n111-046-078.log b/ScaleDF_rebuttal/_host_logs/20260726_1556_vith32_launch_master_n116-042-185_port29877/launcher_rank1_n111-046-078.log new file mode 100644 index 0000000000000000000000000000000000000000..914ec28491c67835f74a64c421db2d0545c0a404 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1556_vith32_launch_master_n116-042-185_port29877/launcher_rank1_n111-046-078.log @@ -0,0 +1,25 @@ +started_utc=2026-07-26T15:56:35Z +hostname=n111-046-078 +node_rank=1 +nnodes=4 +nproc_per_node=8 +world_size=32 +master_addr=10.116.42.185 +master_port=29877 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +main=main_cross_2direct_balance_realscale_perturb_v4_accum.py +model_name=deit_huge_patch14_LS +pretrain=/tmp/ScaleDF_repro_20260724/pretrained/deit_3_huge_224_21k_v1.pth +resume=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_24card_effbs3072_from_epoch7_n116-041-246_n116-042-186_n116-042-185_20260726_0736_vith24_from_epoch7/checkpoint.pth +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_32card_effbs2048_from_latest24_master_n116-042-185_20260726_1556_vith32_port29877 +batch_per_gpu=64 +accum_iter=1 +effective_batch=2048 +eval_batch_per_gpu=64 +epochs=40 +W0726 15:56:48.631000 2625 torch/distributed/run.py:766] +W0726 15:56:48.631000 2625 torch/distributed/run.py:766] ***************************************** +W0726 15:56:48.631000 2625 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 15:56:48.631000 2625 torch/distributed/run.py:766] ***************************************** +Terminated diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1556_vith32_launch_master_n116-042-185_port29877/launcher_rank2_n116-040-048.log b/ScaleDF_rebuttal/_host_logs/20260726_1556_vith32_launch_master_n116-042-185_port29877/launcher_rank2_n116-040-048.log new file mode 100644 index 0000000000000000000000000000000000000000..820cd4e1c87a3036c258cccd9ea8a130fee771d3 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1556_vith32_launch_master_n116-042-185_port29877/launcher_rank2_n116-040-048.log @@ -0,0 +1,25 @@ +started_utc=2026-07-26T15:56:35Z +hostname=n116-040-048 +node_rank=2 +nnodes=4 +nproc_per_node=8 +world_size=32 +master_addr=10.116.42.185 +master_port=29877 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +main=main_cross_2direct_balance_realscale_perturb_v4_accum.py +model_name=deit_huge_patch14_LS +pretrain=/tmp/ScaleDF_repro_20260724/pretrained/deit_3_huge_224_21k_v1.pth +resume=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_24card_effbs3072_from_epoch7_n116-041-246_n116-042-186_n116-042-185_20260726_0736_vith24_from_epoch7/checkpoint.pth +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_32card_effbs2048_from_latest24_master_n116-042-185_20260726_1556_vith32_port29877 +batch_per_gpu=64 +accum_iter=1 +effective_batch=2048 +eval_batch_per_gpu=64 +epochs=40 +W0726 15:56:48.663000 1183 torch/distributed/run.py:766] +W0726 15:56:48.663000 1183 torch/distributed/run.py:766] ***************************************** +W0726 15:56:48.663000 1183 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 15:56:48.663000 1183 torch/distributed/run.py:766] ***************************************** +Terminated diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1556_vith32_launch_master_n116-042-185_port29877/launcher_rank3_n111-046-125.log b/ScaleDF_rebuttal/_host_logs/20260726_1556_vith32_launch_master_n116-042-185_port29877/launcher_rank3_n111-046-125.log new file mode 100644 index 0000000000000000000000000000000000000000..dfdfb2c3efa99e7da9c25787525aee3dd0f4a2a0 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1556_vith32_launch_master_n116-042-185_port29877/launcher_rank3_n111-046-125.log @@ -0,0 +1,25 @@ +started_utc=2026-07-26T15:56:35Z +hostname=n111-046-125 +node_rank=3 +nnodes=4 +nproc_per_node=8 +world_size=32 +master_addr=10.116.42.185 +master_port=29877 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +main=main_cross_2direct_balance_realscale_perturb_v4_accum.py +model_name=deit_huge_patch14_LS +pretrain=/tmp/ScaleDF_repro_20260724/pretrained/deit_3_huge_224_21k_v1.pth +resume=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_24card_effbs3072_from_epoch7_n116-041-246_n116-042-186_n116-042-185_20260726_0736_vith24_from_epoch7/checkpoint.pth +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_32card_effbs2048_from_latest24_master_n116-042-185_20260726_1556_vith32_port29877 +batch_per_gpu=64 +accum_iter=1 +effective_batch=2048 +eval_batch_per_gpu=64 +epochs=40 +W0726 15:56:51.344000 1062 torch/distributed/run.py:766] +W0726 15:56:51.344000 1062 torch/distributed/run.py:766] ***************************************** +W0726 15:56:51.344000 1062 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 15:56:51.344000 1062 torch/distributed/run.py:766] ***************************************** +Terminated diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1556_vith32_launch_master_n116-042-185_port29877/output_root.txt b/ScaleDF_rebuttal/_host_logs/20260726_1556_vith32_launch_master_n116-042-185_port29877/output_root.txt new file mode 100644 index 0000000000000000000000000000000000000000..6a6d9b32449c757359ec5b7aad63e2cf55f2cbf5 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1556_vith32_launch_master_n116-042-185_port29877/output_root.txt @@ -0,0 +1 @@ +/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_32card_effbs2048_from_latest24_master_n116-042-185_20260726_1556_vith32_port29877 diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1556_vith32_launch_master_n116-042-185_port29877/pid_rank0_n116-042-185.txt b/ScaleDF_rebuttal/_host_logs/20260726_1556_vith32_launch_master_n116-042-185_port29877/pid_rank0_n116-042-185.txt new file mode 100644 index 0000000000000000000000000000000000000000..6c4b54876bdf8e2162e9f34bc4ca80c8d5feb8df --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1556_vith32_launch_master_n116-042-185_port29877/pid_rank0_n116-042-185.txt @@ -0,0 +1 @@ +1253 diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1556_vith32_launch_master_n116-042-185_port29877/pid_rank1_n111-046-078.txt b/ScaleDF_rebuttal/_host_logs/20260726_1556_vith32_launch_master_n116-042-185_port29877/pid_rank1_n111-046-078.txt new file mode 100644 index 0000000000000000000000000000000000000000..1da5938934e2761d6eed1e3e6c73d7a789d45e2b --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1556_vith32_launch_master_n116-042-185_port29877/pid_rank1_n111-046-078.txt @@ -0,0 +1 @@ +2617 diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1556_vith32_launch_master_n116-042-185_port29877/pid_rank2_n116-040-048.txt b/ScaleDF_rebuttal/_host_logs/20260726_1556_vith32_launch_master_n116-042-185_port29877/pid_rank2_n116-040-048.txt new file mode 100644 index 0000000000000000000000000000000000000000..1c509f5be8f34dea982707e444bb897ac7e4d0f9 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1556_vith32_launch_master_n116-042-185_port29877/pid_rank2_n116-040-048.txt @@ -0,0 +1 @@ +1175 diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1556_vith32_launch_master_n116-042-185_port29877/pid_rank3_n111-046-125.txt b/ScaleDF_rebuttal/_host_logs/20260726_1556_vith32_launch_master_n116-042-185_port29877/pid_rank3_n111-046-125.txt new file mode 100644 index 0000000000000000000000000000000000000000..628625b6bd60b54d04cd02e5c6a54f6786b8f657 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1556_vith32_launch_master_n116-042-185_port29877/pid_rank3_n111-046-125.txt @@ -0,0 +1 @@ +1054 diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1556_vith32_launch_master_n116-042-185_port29877/resume.txt b/ScaleDF_rebuttal/_host_logs/20260726_1556_vith32_launch_master_n116-042-185_port29877/resume.txt new file mode 100644 index 0000000000000000000000000000000000000000..2158d35929432dee54a9499c09d7990967633387 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1556_vith32_launch_master_n116-042-185_port29877/resume.txt @@ -0,0 +1 @@ +/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_24card_effbs3072_from_epoch7_n116-041-246_n116-042-186_n116-042-185_20260726_0736_vith24_from_epoch7/checkpoint.pth diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1556_vith32_launch_master_n116-042-185_port29877/run_name.txt b/ScaleDF_rebuttal/_host_logs/20260726_1556_vith32_launch_master_n116-042-185_port29877/run_name.txt new file mode 100644 index 0000000000000000000000000000000000000000..a50b92de92eb0ad9f4ec0aad405d4b7179d36fe9 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1556_vith32_launch_master_n116-042-185_port29877/run_name.txt @@ -0,0 +1 @@ +modelsize_vith_40epoch_32card_effbs2048_from_latest24_master_n116-042-185_20260726_1556_vith32_port29877 diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1559_vith32_launch_master_n116-042-185_port25003/launcher_rank0_n116-042-185.log b/ScaleDF_rebuttal/_host_logs/20260726_1559_vith32_launch_master_n116-042-185_port25003/launcher_rank0_n116-042-185.log new file mode 100644 index 0000000000000000000000000000000000000000..248371806bd54fd0507c91e4b99cbaa311b6d9da --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1559_vith32_launch_master_n116-042-185_port25003/launcher_rank0_n116-042-185.log @@ -0,0 +1,419 @@ +started_utc=2026-07-26T15:59:49Z +hostname=n116-042-185 +node_rank=0 +nnodes=4 +nproc_per_node=8 +world_size=32 +master_addr=10.116.42.185 +master_port=25003 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +main=main_cross_2direct_balance_realscale_perturb_v4_accum.py +model_name=deit_huge_patch14_LS +pretrain=/tmp/ScaleDF_repro_20260724/pretrained/deit_3_huge_224_21k_v1.pth +resume=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_24card_effbs3072_from_epoch7_n116-041-246_n116-042-186_n116-042-185_20260726_0736_vith24_from_epoch7/checkpoint.pth +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_32card_effbs2048_from_latest24_master_n116-042-185_20260726_1559_vith32_port25003 +batch_per_gpu=64 +accum_iter=1 +effective_batch=2048 +eval_batch_per_gpu=64 +epochs=40 +W0726 16:00:00.037000 1503 torch/distributed/run.py:766] +W0726 16:00:00.037000 1503 torch/distributed/run.py:766] ***************************************** +W0726 16:00:00.037000 1503 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 16:00:00.037000 1503 torch/distributed/run.py:766] ***************************************** +Traceback (most recent call last): +Traceback (most recent call last): +Traceback (most recent call last): + File "/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py", line 13, in + File "/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py", line 13, in +Traceback (most recent call last): + File "/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py", line 13, in + File "/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py", line 13, in +Traceback (most recent call last): +Traceback (most recent call last): + File "/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py", line 13, in + File "/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py", line 13, in +Traceback (most recent call last): + File "/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py", line 13, in +from timm.data import Mixupfrom timm.data import Mixupfrom timm.data import Mixupfrom timm.data import Mixup + + + + File "/home/tiger/.local/lib/python3.11/site-packages/timm/__init__.py", line 2, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/__init__.py", line 2, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/__init__.py", line 2, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/__init__.py", line 2, in + from timm.data import Mixupfrom timm.data import Mixup + +from timm.data import Mixup + File "/home/tiger/.local/lib/python3.11/site-packages/timm/__init__.py", line 2, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/__init__.py", line 2, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/__init__.py", line 2, in +Traceback (most recent call last): + File "/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py", line 13, in + from timm.data import Mixup + File "/home/tiger/.local/lib/python3.11/site-packages/timm/__init__.py", line 2, in + from .models import create_model, list_models, is_model, list_modules, model_entrypoint, \ from .models import create_model, list_models, is_model, list_modules, model_entrypoint, \ + +from .models import create_model, list_models, is_model, list_modules, model_entrypoint, \ + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/__init__.py", line 1, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/__init__.py", line 1, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/__init__.py", line 1, in + from .models import create_model, list_models, is_model, list_modules, model_entrypoint, \ + from .models import create_model, list_models, is_model, list_modules, model_entrypoint, \ + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/__init__.py", line 1, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/__init__.py", line 1, in +from .models import create_model, list_models, is_model, list_modules, model_entrypoint, \from .models import create_model, list_models, is_model, list_modules, model_entrypoint, \ + + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/__init__.py", line 1, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/__init__.py", line 1, in + from .models import create_model, list_models, is_model, list_modules, model_entrypoint, \ + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/__init__.py", line 1, in + from .byoanet import * + from .byoanet import * + from .byoanet import * +from .byoanet import * from .byoanet import * + + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/byoanet.py", line 15, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/byoanet.py", line 15, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/byoanet.py", line 15, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/byoanet.py", line 15, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/byoanet.py", line 15, in + from .byoanet import * + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/byoanet.py", line 15, in + from .byoanet import * + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/byoanet.py", line 15, in + from .byoanet import * + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/byoanet.py", line 15, in + from timm.data import IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STDfrom timm.data import IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STD + +from timm.data import IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STD + from timm.data import IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STD + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/__init__.py", line 5, in +from timm.data import IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STD File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/__init__.py", line 5, in + + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/__init__.py", line 5, in + from timm.data import IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STD +from timm.data import IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STD + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/__init__.py", line 5, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/__init__.py", line 5, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/__init__.py", line 5, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/__init__.py", line 5, in + from timm.data import IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STD + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/__init__.py", line 5, in + from .dataset import ImageDataset, IterableImageDataset, AugMixDatasetfrom .dataset import ImageDataset, IterableImageDataset, AugMixDataset + + from .dataset import ImageDataset, IterableImageDataset, AugMixDataset + from .dataset import ImageDataset, IterableImageDataset, AugMixDataset + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/dataset.py", line 12, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/dataset.py", line 12, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/dataset.py", line 12, in + from .dataset import ImageDataset, IterableImageDataset, AugMixDataset File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/dataset.py", line 12, in + +from .dataset import ImageDataset, IterableImageDataset, AugMixDataset +from .dataset import ImageDataset, IterableImageDataset, AugMixDataset + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/dataset.py", line 12, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/dataset.py", line 12, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/dataset.py", line 12, in + from .dataset import ImageDataset, IterableImageDataset, AugMixDataset + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/dataset.py", line 12, in + from .parsers import create_parserfrom .parsers import create_parser +from .parsers import create_parser + + from .parsers import create_parserfrom .parsers import create_parser + + from .parsers import create_parser File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/__init__.py", line 1, in + +from .parsers import create_parser File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/__init__.py", line 1, in + + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/__init__.py", line 1, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/__init__.py", line 1, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/__init__.py", line 1, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/__init__.py", line 1, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/__init__.py", line 1, in + from .parsers import create_parser + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/__init__.py", line 1, in + from .parser_factory import create_parserfrom .parser_factory import create_parser + + from .parser_factory import create_parserfrom .parser_factory import create_parser + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_factory.py", line 3, in + + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_factory.py", line 3, in + from .parser_factory import create_parser + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_factory.py", line 3, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_factory.py", line 3, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_factory.py", line 3, in +from .parser_factory import create_parser +from .parser_factory import create_parser + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_factory.py", line 3, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_factory.py", line 3, in + from .parser_factory import create_parser + from .parser_image_folder import ParserImageFolder + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_factory.py", line 3, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_image_folder.py", line 10, in + from .parser_image_folder import ParserImageFolder + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_image_folder.py", line 10, in + from .parser_image_folder import ParserImageFolder + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_image_folder.py", line 10, in + from timm.utils.misc import natural_key + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/__init__.py", line 13, in + from .parser_image_folder import ParserImageFolder + from .parser_image_folder import ParserImageFolder + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_image_folder.py", line 10, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_image_folder.py", line 10, in + from .parser_image_folder import ParserImageFolder + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_image_folder.py", line 10, in + from .parser_image_folder import ParserImageFolder + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_image_folder.py", line 10, in + from .parser_image_folder import ParserImageFolder + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_image_folder.py", line 10, in + from timm.utils.misc import natural_key + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/__init__.py", line 13, in + from timm.utils.misc import natural_key + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/__init__.py", line 13, in + from .summary import update_summary, get_outdir + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/summary.py", line 9, in + from timm.utils.misc import natural_key + from timm.utils.misc import natural_key + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/__init__.py", line 13, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/__init__.py", line 13, in + from timm.utils.misc import natural_key + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/__init__.py", line 13, in + from timm.utils.misc import natural_key + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/__init__.py", line 13, in + from timm.utils.misc import natural_key + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/__init__.py", line 13, in + from .summary import update_summary, get_outdir + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/summary.py", line 9, in + from .summary import update_summary, get_outdir + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/summary.py", line 9, in + from .summary import update_summary, get_outdir + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/summary.py", line 9, in + import wandb + File "/usr/local/lib/python3.11/dist-packages/wandb/__init__.py", line 35, in + from .summary import update_summary, get_outdir + from .summary import update_summary, get_outdir + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/summary.py", line 9, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/summary.py", line 9, in + from .summary import update_summary, get_outdir + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/summary.py", line 9, in + from .summary import update_summary, get_outdir + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/summary.py", line 9, in + import wandb + File "/usr/local/lib/python3.11/dist-packages/wandb/__init__.py", line 35, in + import wandb + File "/usr/local/lib/python3.11/dist-packages/wandb/__init__.py", line 35, in + import wandb + File "/usr/local/lib/python3.11/dist-packages/wandb/__init__.py", line 35, in + from wandb import sdk as wandb_sdk + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/__init__.py", line 9, in + import wandbimport wandb + + File "/usr/local/lib/python3.11/dist-packages/wandb/__init__.py", line 35, in + File "/usr/local/lib/python3.11/dist-packages/wandb/__init__.py", line 35, in + import wandb + File "/usr/local/lib/python3.11/dist-packages/wandb/__init__.py", line 35, in + import wandb + File "/usr/local/lib/python3.11/dist-packages/wandb/__init__.py", line 35, in + from wandb import sdk as wandb_sdk + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/__init__.py", line 9, in + from wandb import sdk as wandb_sdk + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/__init__.py", line 9, in + from wandb import sdk as wandb_sdk + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/__init__.py", line 9, in + from wandb import sdk as wandb_sdk + from wandb import sdk as wandb_sdk + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/__init__.py", line 9, in + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/__init__.py", line 9, in + from wandb import sdk as wandb_sdk + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/__init__.py", line 9, in + from .wandb_init import _attach, init # noqa: F401 + from .wandb_init import _attach, init # noqa: F401 + from .wandb_init import _attach, init # noqa: F401 + ^^^^^^ ^^ ^^ ^^ ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ ^^ +from wandb import sdk as wandb_sdk^^ + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/wandb_init.py", line 31, in +^ +^^ + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/wandb_init.py", line 31, in + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/__init__.py", line 9, in + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/wandb_init.py", line 31, in + from .wandb_init import _attach, init # noqa: F401 + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/wandb_init.py", line 31, in + from .wandb_init import _attach, init # noqa: F401 + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/wandb_init.py", line 31, in + from .wandb_init import _attach, init # noqa: F401 +from .wandb_init import _attach, init # noqa: F401 + ^^ ^ ^from .wandb_init import _attach, init # noqa: F401 +^from .backend.backend import Backend^from .backend.backend import Backend^ +^ +^ ^ File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/backend/backend.py", line 17, in +^ ^ File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/backend/backend.py", line 17, in +^ ^^ ^^ ^^^from .backend.backend import Backend^ +^^^^^^^^ File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/backend/backend.py", line 17, in +^^ ^^^^from .backend.backend import Backend^^^ +^^^^^^^^^^^ File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/backend/backend.py", line 17, in +^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ +^^^ File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/wandb_init.py", line 31, in + +^^^ File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/wandb_init.py", line 31, in +^^ + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/wandb_init.py", line 31, in + from .backend.backend import Backend + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/backend/backend.py", line 17, in + from ..interface.interface import InterfaceBase + from ..interface.interface import InterfaceBasefrom ..interface.interface import InterfaceBase + + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/interface/interface.py", line 34, in + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/interface/interface.py", line 34, in + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/interface/interface.py", line 34, in + from .backend.backend import Backend + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/backend/backend.py", line 17, in + from .backend.backend import Backend + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/backend/backend.py", line 17, in + from .backend.backend import Backend + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/backend/backend.py", line 17, in + from ..lib.mailbox import MailboxHandle + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 92, in + from ..interface.interface import InterfaceBase + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/interface/interface.py", line 34, in + from ..interface.interface import InterfaceBase + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/interface/interface.py", line 34, in + from ..interface.interface import InterfaceBase + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/interface/interface.py", line 34, in + from ..interface.interface import InterfaceBase + from ..interface.interface import InterfaceBase + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/interface/interface.py", line 34, in + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/interface/interface.py", line 34, in + from ..lib.mailbox import MailboxHandle + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 92, in + from ..lib.mailbox import MailboxHandle + from ..lib.mailbox import MailboxHandle + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 92, in + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 92, in + from ..lib.mailbox import MailboxHandle + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 92, in + from ..lib.mailbox import MailboxHandle + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 92, in + from ..lib.mailbox import MailboxHandle + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 92, in + from ..lib.mailbox import MailboxHandle + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 92, in + class _MailboxSlot: + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 93, in _MailboxSlot + class _MailboxSlot:class _MailboxSlot: + class _MailboxSlot: + class _MailboxSlot: File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 93, in _MailboxSlot + +class _MailboxSlot:class _MailboxSlot: File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 93, in _MailboxSlot + + + + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 93, in _MailboxSlot + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 93, in _MailboxSlot + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 93, in _MailboxSlot + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 93, in _MailboxSlot + class _MailboxSlot: + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 93, in _MailboxSlot + _result: Optional[pb.Result] + _result: Optional[pb.Result] + _result: Optional[pb.Result] + ^ _result: Optional[pb.Result]^ ^ + ^ ^ ^^ ^^^ ^^ ^ ^^ ^ + AttributeError^_result: Optional[pb.Result] ^ : + + module 'wandb.proto.wandb_internal_pb2' has no attribute 'Result' AttributeError : ^ + module 'wandb.proto.wandb_internal_pb2' has no attribute 'Result' ^ ^ ^ + ^ ^ ^ _result: Optional[pb.Result]^ + ^ + ^ AttributeError ^ : ^ ^ module 'wandb.proto.wandb_internal_pb2' has no attribute 'Result' ^ ^ + ^ ^ ^ + AttributeError : module 'wandb.proto.wandb_internal_pb2' has no attribute 'Result' + ^ ^ ^ ^ ^ ^ ^^^^^^ +^^^AttributeError^: ^^module 'wandb.proto.wandb_internal_pb2' has no attribute 'Result' +AttributeError +: module 'wandb.proto.wandb_internal_pb2' has no attribute 'Result' + _result: Optional[pb.Result] + _result: Optional[pb.Result] + ^^^^^ ^^ ^ ^ + AttributeError: module 'wandb.proto.wandb_internal_pb2' has no attribute 'Result' + ^^^^^^^^^ +AttributeError: module 'wandb.proto.wandb_internal_pb2' has no attribute 'Result' +W0726 16:00:35.028000 1503 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 1569 closing signal SIGTERM +E0726 16:00:35.060000 1503 torch/distributed/elastic/multiprocessing/api.py:874] failed (exitcode: 1) local_rank: 1 (pid: 1570) of binary: /usr/bin/python3 +Traceback (most recent call last): + File "/usr/local/bin/torchrun", line 7, in + sys.exit(main()) + ^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 355, in wrapper + return f(*args, **kwargs) + ^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 892, in main + run(args) + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 883, in run + elastic_launch( + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 139, in __call__ + return launch_agent(self._config, self._entrypoint, list(args)) + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 270, in launch_agent + raise ChildFailedError( +torch.distributed.elastic.multiprocessing.errors.ChildFailedError: +============================================================ +main_cross_2direct_balance_realscale_perturb_v4_accum.py FAILED +------------------------------------------------------------ +Failures: +[1]: + time : 2026-07-26_16:00:35 + host : n116-042-185.byted.org + rank : 2 (local_rank: 2) + exitcode : 1 (pid: 1571) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +[2]: + time : 2026-07-26_16:00:35 + host : n116-042-185.byted.org + rank : 3 (local_rank: 3) + exitcode : 1 (pid: 1572) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +[3]: + time : 2026-07-26_16:00:35 + host : n116-042-185.byted.org + rank : 4 (local_rank: 4) + exitcode : 1 (pid: 1573) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +[4]: + time : 2026-07-26_16:00:35 + host : n116-042-185.byted.org + rank : 5 (local_rank: 5) + exitcode : 1 (pid: 1574) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +[5]: + time : 2026-07-26_16:00:35 + host : n116-042-185.byted.org + rank : 6 (local_rank: 6) + exitcode : 1 (pid: 1575) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +[6]: + time : 2026-07-26_16:00:35 + host : n116-042-185.byted.org + rank : 7 (local_rank: 7) + exitcode : 1 (pid: 1576) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +------------------------------------------------------------ +Root Cause (first observed failure): +[0]: + time : 2026-07-26_16:00:35 + host : n116-042-185.byted.org + rank : 1 (local_rank: 1) + exitcode : 1 (pid: 1570) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +============================================================ diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1559_vith32_launch_master_n116-042-185_port25003/launcher_rank1_n111-046-078.log b/ScaleDF_rebuttal/_host_logs/20260726_1559_vith32_launch_master_n116-042-185_port25003/launcher_rank1_n111-046-078.log new file mode 100644 index 0000000000000000000000000000000000000000..1b1ac9901448ba67fc7acad3bb90f6f4a07238c4 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1559_vith32_launch_master_n116-042-185_port25003/launcher_rank1_n111-046-078.log @@ -0,0 +1,609 @@ +started_utc=2026-07-26T15:59:49Z +hostname=n111-046-078 +node_rank=1 +nnodes=4 +nproc_per_node=8 +world_size=32 +master_addr=10.116.42.185 +master_port=25003 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +main=main_cross_2direct_balance_realscale_perturb_v4_accum.py +model_name=deit_huge_patch14_LS +pretrain=/tmp/ScaleDF_repro_20260724/pretrained/deit_3_huge_224_21k_v1.pth +resume=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_24card_effbs3072_from_epoch7_n116-041-246_n116-042-186_n116-042-185_20260726_0736_vith24_from_epoch7/checkpoint.pth +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_32card_effbs2048_from_latest24_master_n116-042-185_20260726_1559_vith32_port25003 +batch_per_gpu=64 +accum_iter=1 +effective_batch=2048 +eval_batch_per_gpu=64 +epochs=40 +W0726 16:00:02.412000 2809 torch/distributed/run.py:766] +W0726 16:00:02.412000 2809 torch/distributed/run.py:766] ***************************************** +W0726 16:00:02.412000 2809 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 16:00:02.412000 2809 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 8): env:// +| distributed init (rank 15): env:// +[rank15]:[W726 16:00:31.282010636 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 15] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank8]:[W726 16:00:31.326449552 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 8] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 14): env:// +[rank14]:[W726 16:00:31.448833548 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 14] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 9): env:// +[rank9]:[W726 16:00:32.987884514 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 9] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 12): env:// +| distributed init (rank 13): env:// +| distributed init (rank 10): env:// +| distributed init (rank 11): env:// +[rank12]:[W726 16:00:32.127920929 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 12] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank13]:[W726 16:00:32.134819759 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 13] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank10]:[W726 16:00:32.139695018 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 10] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank11]:[W726 16:00:32.139792060 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 11] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank14]:[W726 16:00:35.037296703 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n111-046-078.byted.org]:46044, remote=[n116-042-185.byted.org]:25003): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7fac4e9785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7fac37ccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7fac37ccff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7fac37cd084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7fac37cca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fabf93c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7fac51e674a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7fac54a3e1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7fac54abe89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank14]:[W726 16:00:35.043628714 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 14] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank8]:[W726 16:00:35.037310838 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=71, addr=[n111-046-078.byted.org]:46024, remote=[n116-042-185.byted.org]:25003): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f1116b785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f11002cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7f11002cff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7f11002d084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7f11002ca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f10c19c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f111a3224a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f111cef91f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f111cf7989c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank8]:[W726 16:00:35.043706640 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 8] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank13]:[W726 16:00:35.037337339 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n111-046-078.byted.org]:46068, remote=[n116-042-185.byted.org]:25003): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7fc31f9785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7fc308ccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7fc308ccff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7fc308cd084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7fc308cca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fc2ca3c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7fc322e5f4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7fc325a361f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7fc325ab689c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank13]:[W726 16:00:35.044281871 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 13] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank15]:[W726 16:00:35.037307914 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n111-046-078.byted.org]:46040, remote=[n116-042-185.byted.org]:25003): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f50893785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f5072acdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7f5072acff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7f5072ad084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7f5072aca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f50341c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f508cadb4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f508f6b21f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f508f73289c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank15]:[W726 16:00:35.044330480 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 15] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank10]:[W726 16:00:35.037335838 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n111-046-078.byted.org]:46074, remote=[n116-042-185.byted.org]:25003): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f3f827785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f3f6becdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7f3f6becff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7f3f6bed084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7f3f6beca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f3f2d5c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f3f85f614a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f3f88b381f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f3f88bb889c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank10]:[W726 16:00:35.044368593 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 10] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank9]:[W726 16:00:35.037349372 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n111-046-078.byted.org]:46054, remote=[n116-042-185.byted.org]:25003): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7fde2c1785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7fde158cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7fde158cff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7fde158d084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7fde158ca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fddd6fc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7fde2f9334a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7fde3250a1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7fde3258a89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank11]:[W726 16:00:35.037335864 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n111-046-078.byted.org]:46084, remote=[n116-042-185.byted.org]:25003): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f7ed99785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f7ec30cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7f7ec30cff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7f7ec30d084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7f7ec30ca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f7e847c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f7edd0bd4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f7edfc941f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f7edfd1489c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank11]:[W726 16:00:35.044480166 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 11] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank9]:[W726 16:00:35.044488631 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 9] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank12]:[W726 16:00:35.037338450 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n111-046-078.byted.org]:46058, remote=[n116-042-185.byted.org]:25003): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7feec65785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7feeafccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7feeafccff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7feeafcd084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7feeafcca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fee713c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7feec9d314a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7feecc9081f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7feecc98889c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank12]:[W726 16:00:35.045647701 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 12] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank12]: Traceback (most recent call last): +[rank12]: File "/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py", line 551, in +[rank12]: main(args) +[rank12]: File "/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py", line 233, in main +[rank12]: utils.init_distributed_mode(args) +[rank12]: File "/tmp/ScaleDF_repro_20260724/code/deit/utils.py", line 237, in init_distributed_mode +[rank12]: torch.distributed.barrier() +[rank12]: File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/c10d_logger.py", line 81, in wrapper +[rank12]: return func(*args, **kwargs) +[rank12]: ^^^^^^^^^^^^^^^^^^^^^ +[rank12]: File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py", line 4635, in barrier +[rank12]: work = group.barrier(opts=opts) +[rank12]: ^^^^^^^^^^^^^^^^^^^^^^^^ +[rank12]: torch.distributed.DistBackendError: [12] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0', but store->get('0') got error: failed to recv, got 0 bytes +[rank12]: Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +[rank12]: frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7feec65785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +[rank12]: frame #1: + 0x5ba8bfe (0x7feeafccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank12]: frame #2: + 0x5bab100 (0x7feeafcd0100 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank12]: frame #3: c10d::TCPStore::doWait(c10::ArrayRef, std::allocator > >, std::chrono::duration >) + 0x28b (0x7feeafccacab in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank12]: frame #4: c10d::TCPStore::doGet(std::__cxx11::basic_string, std::allocator > const&) + 0x33 (0x7feeafccafa3 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank12]: frame #5: c10d::TCPStore::get(std::__cxx11::basic_string, std::allocator > const&) + 0xab (0x7feeafccc08b in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank12]: frame #6: c10d::PrefixStore::get(std::__cxx11::basic_string, std::allocator > const&) + 0x2f (0x7feeafc7a94f in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank12]: frame #7: c10d::PrefixStore::get(std::__cxx11::basic_string, std::allocator > const&) + 0x2f (0x7feeafc7a94f in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank12]: frame #8: c10d::PrefixStore::get(std::__cxx11::basic_string, std::allocator > const&) + 0x2f (0x7feeafc7a94f in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank12]: frame #9: c10d::ProcessGroupNCCL::broadcastUniqueNCCLID(ncclUniqueId*, bool, std::__cxx11::basic_string, std::allocator > const&, int) + 0x16d (0x7fee713c0e7d in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +[rank12]: frame #10: c10d::ProcessGroupNCCL::initNCCLComm(std::__cxx11::basic_string, std::allocator > const&, c10::Device&, c10d::OpType, int, bool) + 0x16bf (0x7fee713d0f6f in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +[rank12]: frame #11: + 0x11ecf83 (0x7fee713d5f83 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +[rank12]: frame #12: c10d::ProcessGroupNCCL::allreduce_impl(at::Tensor&, char const*, c10d::AllreduceOptions const&) + 0xee (0x7fee713d677e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +[rank12]: frame #13: c10d::ProcessGroupNCCL::barrier(c10d::BarrierOptions const&) + 0x683 (0x7fee713e63a3 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +[rank12]: frame #14: + 0x5b4363f (0x7feeafc6863f in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank12]: frame #15: + 0x5b52cfe (0x7feeafc77cfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank12]: frame #16: + 0x52125b0 (0x7feeaf3375b0 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank12]: frame #17: + 0x5b61a02 (0x7feeafc86a02 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank12]: frame #18: + 0x5b6293d (0x7feeafc8793d in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank12]: frame #19: + 0xc8c143 (0x7feebebda143 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_python.so) +[rank12]: frame #20: + 0x38a1ac (0x7feebe2d81ac in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_python.so) +[rank12]: frame #21: /usr/bin/python3() [0x547d56] +[rank12]: frame #22: _PyObject_MakeTpCall + 0x223 (0x5180e3 in /usr/bin/python3) +[rank12]: frame #23: _PyEval_EvalFrameDefault + 0x8f0 (0x52bb30 in /usr/bin/python3) +[rank12]: frame #24: _PyFunction_Vectorcall + 0x191 (0x55cb71 in /usr/bin/python3) +[rank12]: frame #25: _PyEval_EvalFrameDefault + 0x48cc (0x52fb0c in /usr/bin/python3) +[rank12]: frame #26: PyEval_EvalCode + 0xbb (0x5237fb in /usr/bin/python3) +[rank12]: frame #27: /usr/bin/python3() [0x648fe7] +[rank12]: frame #28: /usr/bin/python3() [0x64693f] +[rank12]: frame #29: /usr/bin/python3() [0x6529d0] +[rank12]: frame #30: _PyRun_SimpleFileObject + 0x18b (0x65271b in /usr/bin/python3) +[rank12]: frame #31: _PyRun_AnyFileObject + 0x44 (0x652544 in /usr/bin/python3) +[rank12]: frame #32: Py_RunMain + 0x20f (0x6512cf in /usr/bin/python3) +[rank12]: frame #33: Py_BytesMain + 0x27 (0x628fa7 in /usr/bin/python3) +[rank12]: frame #34: + 0x2724a (0x7feecc8a624a in /lib/x86_64-linux-gnu/libc.so.6) +[rank12]: frame #35: __libc_start_main + 0x85 (0x7feecc8a6305 in /lib/x86_64-linux-gnu/libc.so.6) +[rank12]: frame #36: _start + 0x21 (0x628e41 in /usr/bin/python3) +[rank12]: . This may indicate a possible application crash on rank 0 or a network set up issue. +[rank11]: Traceback (most recent call last): +[rank11]: File "/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py", line 551, in +[rank11]: main(args) +[rank11]: File "/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py", line 233, in main +[rank11]: utils.init_distributed_mode(args) +[rank11]: File "/tmp/ScaleDF_repro_20260724/code/deit/utils.py", line 237, in init_distributed_mode +[rank11]: torch.distributed.barrier() +[rank11]: File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/c10d_logger.py", line 81, in wrapper +[rank11]: return func(*args, **kwargs) +[rank11]: ^^^^^^^^^^^^^^^^^^^^^ +[rank11]: File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py", line 4635, in barrier +[rank11]: work = group.barrier(opts=opts) +[rank11]: ^^^^^^^^^^^^^^^^^^^^^^^^ +[rank11]: torch.distributed.DistBackendError: [11] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0', but store->get('0') got error: failed to recv, got 0 bytes +[rank11]: Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +[rank11]: frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f7ed99785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +[rank11]: frame #1: + 0x5ba8bfe (0x7f7ec30cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank11]: frame #2: + 0x5bab100 (0x7f7ec30d0100 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank11]: frame #3: c10d::TCPStore::doWait(c10::ArrayRef, std::allocator > >, std::chrono::duration >) + 0x28b (0x7f7ec30cacab in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank11]: frame #4: c10d::TCPStore::doGet(std::__cxx11::basic_string, std::allocator > const&) + 0x33 (0x7f7ec30cafa3 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank11]: frame #5: c10d::TCPStore::get(std::__cxx11::basic_string, std::allocator > const&) + 0xab (0x7f7ec30cc08b in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank11]: frame #6: c10d::PrefixStore::get(std::__cxx11::basic_string, std::allocator > const&) + 0x2f (0x7f7ec307a94f in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank11]: frame #7: c10d::PrefixStore::get(std::__cxx11::basic_string, std::allocator > const&) + 0x2f (0x7f7ec307a94f in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank11]: frame #8: c10d::PrefixStore::get(std::__cxx11::basic_string, std::allocator > const&) + 0x2f (0x7f7ec307a94f in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank11]: frame #9: c10d::ProcessGroupNCCL::broadcastUniqueNCCLID(ncclUniqueId*, bool, std::__cxx11::basic_string, std::allocator > const&, int) + 0x16d (0x7f7e847c0e7d in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +[rank11]: frame #10: c10d::ProcessGroupNCCL::initNCCLComm(std::__cxx11::basic_string, std::allocator > const&, c10::Device&, c10d::OpType, int, bool) + 0x16bf (0x7f7e847d0f6f in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +[rank11]: frame #11: + 0x11ecf83 (0x7f7e847d5f83 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +[rank11]: frame #12: c10d::ProcessGroupNCCL::allreduce_impl(at::Tensor&, char const*, c10d::AllreduceOptions const&) + 0xee (0x7f7e847d677e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +[rank11]: frame #13: c10d::ProcessGroupNCCL::barrier(c10d::BarrierOptions const&) + 0x683 (0x7f7e847e63a3 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +[rank11]: frame #14: + 0x5b4363f (0x7f7ec306863f in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank11]: frame #15: + 0x5b52cfe (0x7f7ec3077cfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank11]: frame #16: + 0x52125b0 (0x7f7ec27375b0 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank11]: frame #17: + 0x5b61a02 (0x7f7ec3086a02 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank11]: frame #18: + 0x5b6293d (0x7f7ec308793d in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank11]: frame #19: + 0xc8c143 (0x7f7ed1fda143 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_python.so) +[rank11]: frame #20: + 0x38a1ac (0x7f7ed16d81ac in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_python.so) +[rank11]: frame #21: /usr/bin/python3() [0x547d56] +[rank11]: frame #22: _PyObject_MakeTpCall + 0x223 (0x5180e3 in /usr/bin/python3) +[rank11]: frame #23: _PyEval_EvalFrameDefault + 0x8f0 (0x52bb30 in /usr/bin/python3) +[rank11]: frame #24: _PyFunction_Vectorcall + 0x191 (0x55cb71 in /usr/bin/python3) +[rank11]: frame #25: _PyEval_EvalFrameDefault + 0x48cc (0x52fb0c in /usr/bin/python3) +[rank11]: frame #26: PyEval_EvalCode + 0xbb (0x5237fb in /usr/bin/python3) +[rank11]: frame #27: /usr/bin/python3() [0x648fe7] +[rank11]: frame #28: /usr/bin/python3() [0x64693f] +[rank11]: frame #29: /usr/bin/python3() [0x6529d0] +[rank11]: frame #30: _PyRun_SimpleFileObject + 0x18b (0x65271b in /usr/bin/python3) +[rank11]: frame #31: _PyRun_AnyFileObject + 0x44 (0x652544 in /usr/bin/python3) +[rank11]: frame #32: Py_RunMain + 0x20f (0x6512cf in /usr/bin/python3) +[rank11]: frame #33: Py_BytesMain + 0x27 (0x628fa7 in /usr/bin/python3) +[rank11]: frame #34: + 0x2724a (0x7f7edfc3224a in /lib/x86_64-linux-gnu/libc.so.6) +[rank11]: frame #35: __libc_start_main + 0x85 (0x7f7edfc32305 in /lib/x86_64-linux-gnu/libc.so.6) +[rank11]: frame #36: _start + 0x21 (0x628e41 in /usr/bin/python3) +[rank11]: . This may indicate a possible application crash on rank 0 or a network set up issue. +[rank10]: Traceback (most recent call last): +[rank10]: File "/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py", line 551, in +[rank10]: main(args) +[rank10]: File "/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py", line 233, in main +[rank10]: utils.init_distributed_mode(args) +[rank10]: File "/tmp/ScaleDF_repro_20260724/code/deit/utils.py", line 237, in init_distributed_mode +[rank10]: torch.distributed.barrier() +[rank10]: File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/c10d_logger.py", line 81, in wrapper +[rank10]: return func(*args, **kwargs) +[rank10]: ^^^^^^^^^^^^^^^^^^^^^ +[rank10]: File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py", line 4635, in barrier +[rank10]: work = group.barrier(opts=opts) +[rank10]: ^^^^^^^^^^^^^^^^^^^^^^^^ +[rank10]: torch.distributed.DistBackendError: [10] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0', but store->get('0') got error: failed to recv, got 0 bytes +[rank10]: Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +[rank10]: frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f3f827785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +[rank10]: frame #1: + 0x5ba8bfe (0x7f3f6becdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank10]: frame #2: + 0x5bab100 (0x7f3f6bed0100 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank10]: frame #3: c10d::TCPStore::doWait(c10::ArrayRef, std::allocator > >, std::chrono::duration >) + 0x28b (0x7f3f6becacab in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank10]: frame #4: c10d::TCPStore::doGet(std::__cxx11::basic_string, std::allocator > const&) + 0x33 (0x7f3f6becafa3 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank10]: frame #5: c10d::TCPStore::get(std::__cxx11::basic_string, std::allocator > const&) + 0xab (0x7f3f6becc08b in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank10]: frame #6: c10d::PrefixStore::get(std::__cxx11::basic_string, std::allocator > const&) + 0x2f (0x7f3f6be7a94f in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank10]: frame #7: c10d::PrefixStore::get(std::__cxx11::basic_string, std::allocator > const&) + 0x2f (0x7f3f6be7a94f in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank10]: frame #8: c10d::PrefixStore::get(std::__cxx11::basic_string, std::allocator > const&) + 0x2f (0x7f3f6be7a94f in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank10]: frame #9: c10d::ProcessGroupNCCL::broadcastUniqueNCCLID(ncclUniqueId*, bool, std::__cxx11::basic_string, std::allocator > const&, int) + 0x16d (0x7f3f2d5c0e7d in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +[rank10]: frame #10: c10d::ProcessGroupNCCL::initNCCLComm(std::__cxx11::basic_string, std::allocator > const&, c10::Device&, c10d::OpType, int, bool) + 0x16bf (0x7f3f2d5d0f6f in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +[rank10]: frame #11: + 0x11ecf83 (0x7f3f2d5d5f83 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +[rank10]: frame #12: c10d::ProcessGroupNCCL::allreduce_impl(at::Tensor&, char const*, c10d::AllreduceOptions const&) + 0xee (0x7f3f2d5d677e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +[rank10]: frame #13: c10d::ProcessGroupNCCL::barrier(c10d::BarrierOptions const&) + 0x683 (0x7f3f2d5e63a3 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +[rank10]: frame #14: + 0x5b4363f (0x7f3f6be6863f in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank10]: frame #15: + 0x5b52cfe (0x7f3f6be77cfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank10]: frame #16: + 0x52125b0 (0x7f3f6b5375b0 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank10]: frame #17: + 0x5b61a02 (0x7f3f6be86a02 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank10]: frame #18: + 0x5b6293d (0x7f3f6be8793d in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank10]: frame #19: + 0xc8c143 (0x7f3f7adda143 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_python.so) +[rank10]: frame #20: + 0x38a1ac (0x7f3f7a4d81ac in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_python.so) +[rank10]: frame #21: /usr/bin/python3() [0x547d56] +[rank10]: frame #22: _PyObject_MakeTpCall + 0x223 (0x5180e3 in /usr/bin/python3) +[rank10]: frame #23: _PyEval_EvalFrameDefault + 0x8f0 (0x52bb30 in /usr/bin/python3) +[rank10]: frame #24: _PyFunction_Vectorcall + 0x191 (0x55cb71 in /usr/bin/python3) +[rank10]: frame #25: _PyEval_EvalFrameDefault + 0x48cc (0x52fb0c in /usr/bin/python3) +[rank10]: frame #26: PyEval_EvalCode + 0xbb (0x5237fb in /usr/bin/python3) +[rank10]: frame #27: /usr/bin/python3() [0x648fe7] +[rank10]: frame #28: /usr/bin/python3() [0x64693f] +[rank10]: frame #29: /usr/bin/python3() [0x6529d0] +[rank10]: frame #30: _PyRun_SimpleFileObject + 0x18b (0x65271b in /usr/bin/python3) +[rank10]: frame #31: _PyRun_AnyFileObject + 0x44 (0x652544 in /usr/bin/python3) +[rank10]: frame #32: Py_RunMain + 0x20f (0x6512cf in /usr/bin/python3) +[rank10]: frame #33: Py_BytesMain + 0x27 (0x628fa7 in /usr/bin/python3) +[rank10]: frame #34: + 0x2724a (0x7f3f88ad624a in /lib/x86_64-linux-gnu/libc.so.6) +[rank10]: frame #35: __libc_start_main + 0x85 (0x7f3f88ad6305 in /lib/x86_64-linux-gnu/libc.so.6) +[rank10]: frame #36: _start + 0x21 (0x628e41 in /usr/bin/python3) +[rank10]: . This may indicate a possible application crash on rank 0 or a network set up issue. +[rank9]: Traceback (most recent call last): +[rank9]: File "/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py", line 551, in +[rank9]: main(args) +[rank9]: File "/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py", line 233, in main +[rank9]: utils.init_distributed_mode(args) +[rank9]: File "/tmp/ScaleDF_repro_20260724/code/deit/utils.py", line 237, in init_distributed_mode +[rank9]: torch.distributed.barrier() +[rank9]: File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/c10d_logger.py", line 81, in wrapper +[rank9]: return func(*args, **kwargs) +[rank9]: ^^^^^^^^^^^^^^^^^^^^^ +[rank9]: File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py", line 4635, in barrier +[rank9]: work = group.barrier(opts=opts) +[rank9]: ^^^^^^^^^^^^^^^^^^^^^^^^ +[rank9]: torch.distributed.DistBackendError: [9] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0', but store->get('0') got error: failed to recv, got 0 bytes +[rank9]: Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +[rank9]: frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7fde2c1785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +[rank9]: frame #1: + 0x5ba8bfe (0x7fde158cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank9]: frame #2: + 0x5bab100 (0x7fde158d0100 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank9]: frame #3: c10d::TCPStore::doWait(c10::ArrayRef, std::allocator > >, std::chrono::duration >) + 0x28b (0x7fde158cacab in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank9]: frame #4: c10d::TCPStore::doGet(std::__cxx11::basic_string, std::allocator > const&) + 0x33 (0x7fde158cafa3 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank9]: frame #5: c10d::TCPStore::get(std::__cxx11::basic_string, std::allocator > const&) + 0xab (0x7fde158cc08b in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank9]: frame #6: c10d::PrefixStore::get(std::__cxx11::basic_string, std::allocator > const&) + 0x2f (0x7fde1587a94f in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank9]: frame #7: c10d::PrefixStore::get(std::__cxx11::basic_string, std::allocator > const&) + 0x2f (0x7fde1587a94f in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank9]: frame #8: c10d::PrefixStore::get(std::__cxx11::basic_string, std::allocator > const&) + 0x2f (0x7fde1587a94f in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank9]: frame #9: c10d::ProcessGroupNCCL::broadcastUniqueNCCLID(ncclUniqueId*, bool, std::__cxx11::basic_string, std::allocator > const&, int) + 0x16d (0x7fddd6fc0e7d in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +[rank9]: frame #10: c10d::ProcessGroupNCCL::initNCCLComm(std::__cxx11::basic_string, std::allocator > const&, c10::Device&, c10d::OpType, int, bool) + 0x16bf (0x7fddd6fd0f6f in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +[rank9]: frame #11: + 0x11ecf83 (0x7fddd6fd5f83 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +[rank9]: frame #12: c10d::ProcessGroupNCCL::allreduce_impl(at::Tensor&, char const*, c10d::AllreduceOptions const&) + 0xee (0x7fddd6fd677e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +[rank9]: frame #13: c10d::ProcessGroupNCCL::barrier(c10d::BarrierOptions const&) + 0x683 (0x7fddd6fe63a3 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +[rank9]: frame #14: + 0x5b4363f (0x7fde1586863f in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank9]: frame #15: + 0x5b52cfe (0x7fde15877cfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank9]: frame #16: + 0x52125b0 (0x7fde14f375b0 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank9]: frame #17: + 0x5b61a02 (0x7fde15886a02 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank9]: frame #18: + 0x5b6293d (0x7fde1588793d in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank9]: frame #19: + 0xc8c143 (0x7fde247da143 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_python.so) +[rank9]: frame #20: + 0x38a1ac (0x7fde23ed81ac in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_python.so) +[rank9]: frame #21: /usr/bin/python3() [0x547d56] +[rank9]: frame #22: _PyObject_MakeTpCall + 0x223 (0x5180e3 in /usr/bin/python3) +[rank9]: frame #23: _PyEval_EvalFrameDefault + 0x8f0 (0x52bb30 in /usr/bin/python3) +[rank9]: frame #24: _PyFunction_Vectorcall + 0x191 (0x55cb71 in /usr/bin/python3) +[rank9]: frame #25: _PyEval_EvalFrameDefault + 0x48cc (0x52fb0c in /usr/bin/python3) +[rank9]: frame #26: PyEval_EvalCode + 0xbb (0x5237fb in /usr/bin/python3) +[rank9]: frame #27: /usr/bin/python3() [0x648fe7] +[rank9]: frame #28: /usr/bin/python3() [0x64693f] +[rank9]: frame #29: /usr/bin/python3() [0x6529d0] +[rank9]: frame #30: _PyRun_SimpleFileObject + 0x18b (0x65271b in /usr/bin/python3) +[rank9]: frame #31: _PyRun_AnyFileObject + 0x44 (0x652544 in /usr/bin/python3) +[rank9]: frame #32: Py_RunMain + 0x20f (0x6512cf in /usr/bin/python3) +[rank9]: frame #33: Py_BytesMain + 0x27 (0x628fa7 in /usr/bin/python3) +[rank9]: frame #34: + 0x2724a (0x7fde324a824a in /lib/x86_64-linux-gnu/libc.so.6) +[rank9]: frame #35: __libc_start_main + 0x85 (0x7fde324a8305 in /lib/x86_64-linux-gnu/libc.so.6) +[rank9]: frame #36: _start + 0x21 (0x628e41 in /usr/bin/python3) +[rank9]: . This may indicate a possible application crash on rank 0 or a network set up issue. +[rank14]: Traceback (most recent call last): +[rank14]: File "/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py", line 551, in +[rank14]: main(args) +[rank14]: File "/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py", line 233, in main +[rank14]: utils.init_distributed_mode(args) +[rank14]: File "/tmp/ScaleDF_repro_20260724/code/deit/utils.py", line 237, in init_distributed_mode +[rank14]: torch.distributed.barrier() +[rank14]: File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/c10d_logger.py", line 81, in wrapper +[rank14]: return func(*args, **kwargs) +[rank14]: ^^^^^^^^^^^^^^^^^^^^^ +[rank14]: File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py", line 4635, in barrier +[rank14]: work = group.barrier(opts=opts) +[rank14]: ^^^^^^^^^^^^^^^^^^^^^^^^ +[rank14]: torch.distributed.DistBackendError: [14] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0', but store->get('0') got error: failed to recv, got 0 bytes +[rank14]: Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +[rank14]: frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7fac4e9785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +[rank14]: frame #1: + 0x5ba8bfe (0x7fac37ccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank14]: frame #2: + 0x5bab100 (0x7fac37cd0100 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank14]: frame #3: c10d::TCPStore::doWait(c10::ArrayRef, std::allocator > >, std::chrono::duration >) + 0x28b (0x7fac37ccacab in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank14]: frame #4: c10d::TCPStore::doGet(std::__cxx11::basic_string, std::allocator > const&) + 0x33 (0x7fac37ccafa3 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank14]: frame #5: c10d::TCPStore::get(std::__cxx11::basic_string, std::allocator > const&) + 0xab (0x7fac37ccc08b in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank14]: frame #6: c10d::PrefixStore::get(std::__cxx11::basic_string, std::allocator > const&) + 0x2f (0x7fac37c7a94f in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank14]: frame #7: c10d::PrefixStore::get(std::__cxx11::basic_string, std::allocator > const&) + 0x2f (0x7fac37c7a94f in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank14]: frame #8: c10d::PrefixStore::get(std::__cxx11::basic_string, std::allocator > const&) + 0x2f (0x7fac37c7a94f in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank14]: frame #9: c10d::ProcessGroupNCCL::broadcastUniqueNCCLID(ncclUniqueId*, bool, std::__cxx11::basic_string, std::allocator > const&, int) + 0x16d (0x7fabf93c0e7d in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +[rank14]: frame #10: c10d::ProcessGroupNCCL::initNCCLComm(std::__cxx11::basic_string, std::allocator > const&, c10::Device&, c10d::OpType, int, bool) + 0x16bf (0x7fabf93d0f6f in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +[rank14]: frame #11: + 0x11ecf83 (0x7fabf93d5f83 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +[rank14]: frame #12: c10d::ProcessGroupNCCL::allreduce_impl(at::Tensor&, char const*, c10d::AllreduceOptions const&) + 0xee (0x7fabf93d677e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +[rank14]: frame #13: c10d::ProcessGroupNCCL::barrier(c10d::BarrierOptions const&) + 0x683 (0x7fabf93e63a3 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +[rank14]: frame #14: + 0x5b4363f (0x7fac37c6863f in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank14]: frame #15: + 0x5b52cfe (0x7fac37c77cfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank14]: frame #16: + 0x52125b0 (0x7fac373375b0 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank14]: frame #17: + 0x5b61a02 (0x7fac37c86a02 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank14]: frame #18: + 0x5b6293d (0x7fac37c8793d in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank14]: frame #19: + 0xc8c143 (0x7fac46bda143 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_python.so) +[rank14]: frame #20: + 0x38a1ac (0x7fac462d81ac in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_python.so) +[rank14]: frame #21: /usr/bin/python3() [0x547d56] +[rank14]: frame #22: _PyObject_MakeTpCall + 0x223 (0x5180e3 in /usr/bin/python3) +[rank14]: frame #23: _PyEval_EvalFrameDefault + 0x8f0 (0x52bb30 in /usr/bin/python3) +[rank14]: frame #24: _PyFunction_Vectorcall + 0x191 (0x55cb71 in /usr/bin/python3) +[rank14]: frame #25: _PyEval_EvalFrameDefault + 0x48cc (0x52fb0c in /usr/bin/python3) +[rank14]: frame #26: PyEval_EvalCode + 0xbb (0x5237fb in /usr/bin/python3) +[rank14]: frame #27: /usr/bin/python3() [0x648fe7] +[rank14]: frame #28: /usr/bin/python3() [0x64693f] +[rank14]: frame #29: /usr/bin/python3() [0x6529d0] +[rank14]: frame #30: _PyRun_SimpleFileObject + 0x18b (0x65271b in /usr/bin/python3) +[rank14]: frame #31: _PyRun_AnyFileObject + 0x44 (0x652544 in /usr/bin/python3) +[rank14]: frame #32: Py_RunMain + 0x20f (0x6512cf in /usr/bin/python3) +[rank14]: frame #33: Py_BytesMain + 0x27 (0x628fa7 in /usr/bin/python3) +[rank14]: frame #34: + 0x2724a (0x7fac549dc24a in /lib/x86_64-linux-gnu/libc.so.6) +[rank14]: frame #35: __libc_start_main + 0x85 (0x7fac549dc305 in /lib/x86_64-linux-gnu/libc.so.6) +[rank14]: frame #36: _start + 0x21 (0x628e41 in /usr/bin/python3) +[rank14]: . This may indicate a possible application crash on rank 0 or a network set up issue. +[rank8]: Traceback (most recent call last): +[rank8]: File "/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py", line 551, in +[rank8]: main(args) +[rank8]: File "/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py", line 233, in main +[rank8]: utils.init_distributed_mode(args) +[rank8]: File "/tmp/ScaleDF_repro_20260724/code/deit/utils.py", line 237, in init_distributed_mode +[rank8]: torch.distributed.barrier() +[rank8]: File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/c10d_logger.py", line 81, in wrapper +[rank8]: return func(*args, **kwargs) +[rank8]: ^^^^^^^^^^^^^^^^^^^^^ +[rank8]: File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py", line 4635, in barrier +[rank8]: work = group.barrier(opts=opts) +[rank8]: ^^^^^^^^^^^^^^^^^^^^^^^^ +[rank8]: torch.distributed.DistBackendError: [8] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0', but store->get('0') got error: failed to recv, got 0 bytes +[rank8]: Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +[rank8]: frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f1116b785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +[rank8]: frame #1: + 0x5ba8bfe (0x7f11002cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank8]: frame #2: + 0x5bab100 (0x7f11002d0100 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank8]: frame #3: c10d::TCPStore::doWait(c10::ArrayRef, std::allocator > >, std::chrono::duration >) + 0x28b (0x7f11002cacab in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank8]: frame #4: c10d::TCPStore::doGet(std::__cxx11::basic_string, std::allocator > const&) + 0x33 (0x7f11002cafa3 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank8]: frame #5: c10d::TCPStore::get(std::__cxx11::basic_string, std::allocator > const&) + 0xab (0x7f11002cc08b in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank8]: frame #6: c10d::PrefixStore::get(std::__cxx11::basic_string, std::allocator > const&) + 0x2f (0x7f110027a94f in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank8]: frame #7: c10d::PrefixStore::get(std::__cxx11::basic_string, std::allocator > const&) + 0x2f (0x7f110027a94f in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank8]: frame #8: c10d::PrefixStore::get(std::__cxx11::basic_string, std::allocator > const&) + 0x2f (0x7f110027a94f in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank8]: frame #9: c10d::ProcessGroupNCCL::broadcastUniqueNCCLID(ncclUniqueId*, bool, std::__cxx11::basic_string, std::allocator > const&, int) + 0x16d (0x7f10c19c0e7d in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +[rank8]: frame #10: c10d::ProcessGroupNCCL::initNCCLComm(std::__cxx11::basic_string, std::allocator > const&, c10::Device&, c10d::OpType, int, bool) + 0x16bf (0x7f10c19d0f6f in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +[rank8]: frame #11: + 0x11ecf83 (0x7f10c19d5f83 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +[rank8]: frame #12: c10d::ProcessGroupNCCL::allreduce_impl(at::Tensor&, char const*, c10d::AllreduceOptions const&) + 0xee (0x7f10c19d677e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +[rank8]: frame #13: c10d::ProcessGroupNCCL::barrier(c10d::BarrierOptions const&) + 0x683 (0x7f10c19e63a3 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +[rank8]: frame #14: + 0x5b4363f (0x7f110026863f in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank8]: frame #15: + 0x5b52cfe (0x7f1100277cfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank8]: frame #16: + 0x52125b0 (0x7f10ff9375b0 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank8]: frame #17: + 0x5b61a02 (0x7f1100286a02 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank8]: frame #18: + 0x5b6293d (0x7f110028793d in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank8]: frame #19: + 0xc8c143 (0x7f110f1da143 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_python.so) +[rank8]: frame #20: + 0x38a1ac (0x7f110e8d81ac in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_python.so) +[rank8]: frame #21: /usr/bin/python3() [0x547d56] +[rank8]: frame #22: _PyObject_MakeTpCall + 0x223 (0x5180e3 in /usr/bin/python3) +[rank8]: frame #23: _PyEval_EvalFrameDefault + 0x8f0 (0x52bb30 in /usr/bin/python3) +[rank8]: frame #24: _PyFunction_Vectorcall + 0x191 (0x55cb71 in /usr/bin/python3) +[rank8]: frame #25: _PyEval_EvalFrameDefault + 0x48cc (0x52fb0c in /usr/bin/python3) +[rank8]: frame #26: PyEval_EvalCode + 0xbb (0x5237fb in /usr/bin/python3) +[rank8]: frame #27: /usr/bin/python3() [0x648fe7] +[rank8]: frame #28: /usr/bin/python3() [0x64693f] +[rank8]: frame #29: /usr/bin/python3() [0x6529d0] +[rank8]: frame #30: _PyRun_SimpleFileObject + 0x18b (0x65271b in /usr/bin/python3) +[rank8]: frame #31: _PyRun_AnyFileObject + 0x44 (0x652544 in /usr/bin/python3) +[rank8]: frame #32: Py_RunMain + 0x20f (0x6512cf in /usr/bin/python3) +[rank8]: frame #33: Py_BytesMain + 0x27 (0x628fa7 in /usr/bin/python3) +[rank8]: frame #34: + 0x2724a (0x7f111ce9724a in /lib/x86_64-linux-gnu/libc.so.6) +[rank8]: frame #35: __libc_start_main + 0x85 (0x7f111ce97305 in /lib/x86_64-linux-gnu/libc.so.6) +[rank8]: frame #36: _start + 0x21 (0x628e41 in /usr/bin/python3) +[rank8]: . This may indicate a possible application crash on rank 0 or a network set up issue. +[rank15]: Traceback (most recent call last): +[rank15]: File "/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py", line 551, in +[rank15]: main(args) +[rank15]: File "/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py", line 233, in main +[rank15]: utils.init_distributed_mode(args) +[rank15]: File "/tmp/ScaleDF_repro_20260724/code/deit/utils.py", line 237, in init_distributed_mode +[rank15]: torch.distributed.barrier() +[rank15]: File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/c10d_logger.py", line 81, in wrapper +[rank15]: return func(*args, **kwargs) +[rank15]: ^^^^^^^^^^^^^^^^^^^^^ +[rank15]: File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py", line 4635, in barrier +[rank15]: work = group.barrier(opts=opts) +[rank15]: ^^^^^^^^^^^^^^^^^^^^^^^^ +[rank15]: torch.distributed.DistBackendError: [15] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0', but store->get('0') got error: failed to recv, got 0 bytes +[rank15]: Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +[rank15]: frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f50893785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +[rank15]: frame #1: + 0x5ba8bfe (0x7f5072acdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank15]: frame #2: + 0x5bab100 (0x7f5072ad0100 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank15]: frame #3: c10d::TCPStore::doWait(c10::ArrayRef, std::allocator > >, std::chrono::duration >) + 0x28b (0x7f5072acacab in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank15]: frame #4: c10d::TCPStore::doGet(std::__cxx11::basic_string, std::allocator > const&) + 0x33 (0x7f5072acafa3 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank15]: frame #5: c10d::TCPStore::get(std::__cxx11::basic_string, std::allocator > const&) + 0xab (0x7f5072acc08b in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank15]: frame #6: c10d::PrefixStore::get(std::__cxx11::basic_string, std::allocator > const&) + 0x2f (0x7f5072a7a94f in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank15]: frame #7: c10d::PrefixStore::get(std::__cxx11::basic_string, std::allocator > const&) + 0x2f (0x7f5072a7a94f in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank15]: frame #8: c10d::PrefixStore::get(std::__cxx11::basic_string, std::allocator > const&) + 0x2f (0x7f5072a7a94f in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank15]: frame #9: c10d::ProcessGroupNCCL::broadcastUniqueNCCLID(ncclUniqueId*, bool, std::__cxx11::basic_string, std::allocator > const&, int) + 0x16d (0x7f50341c0e7d in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +[rank15]: frame #10: c10d::ProcessGroupNCCL::initNCCLComm(std::__cxx11::basic_string, std::allocator > const&, c10::Device&, c10d::OpType, int, bool) + 0x16bf (0x7f50341d0f6f in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +[rank15]: frame #11: + 0x11ecf83 (0x7f50341d5f83 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +[rank15]: frame #12: c10d::ProcessGroupNCCL::allreduce_impl(at::Tensor&, char const*, c10d::AllreduceOptions const&) + 0xee (0x7f50341d677e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +[rank15]: frame #13: c10d::ProcessGroupNCCL::barrier(c10d::BarrierOptions const&) + 0x683 (0x7f50341e63a3 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +[rank15]: frame #14: + 0x5b4363f (0x7f5072a6863f in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank15]: frame #15: + 0x5b52cfe (0x7f5072a77cfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank15]: frame #16: + 0x52125b0 (0x7f50721375b0 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank15]: frame #17: + 0x5b61a02 (0x7f5072a86a02 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank15]: frame #18: + 0x5b6293d (0x7f5072a8793d in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank15]: frame #19: + 0xc8c143 (0x7f50819da143 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_python.so) +[rank15]: frame #20: + 0x38a1ac (0x7f50810d81ac in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_python.so) +[rank15]: frame #21: /usr/bin/python3() [0x547d56] +[rank15]: frame #22: _PyObject_MakeTpCall + 0x223 (0x5180e3 in /usr/bin/python3) +[rank15]: frame #23: _PyEval_EvalFrameDefault + 0x8f0 (0x52bb30 in /usr/bin/python3) +[rank15]: frame #24: _PyFunction_Vectorcall + 0x191 (0x55cb71 in /usr/bin/python3) +[rank15]: frame #25: _PyEval_EvalFrameDefault + 0x48cc (0x52fb0c in /usr/bin/python3) +[rank15]: frame #26: PyEval_EvalCode + 0xbb (0x5237fb in /usr/bin/python3) +[rank15]: frame #27: /usr/bin/python3() [0x648fe7] +[rank15]: frame #28: /usr/bin/python3() [0x64693f] +[rank15]: frame #29: /usr/bin/python3() [0x6529d0] +[rank15]: frame #30: _PyRun_SimpleFileObject + 0x18b (0x65271b in /usr/bin/python3) +[rank15]: frame #31: _PyRun_AnyFileObject + 0x44 (0x652544 in /usr/bin/python3) +[rank15]: frame #32: Py_RunMain + 0x20f (0x6512cf in /usr/bin/python3) +[rank15]: frame #33: Py_BytesMain + 0x27 (0x628fa7 in /usr/bin/python3) +[rank15]: frame #34: + 0x2724a (0x7f508f65024a in /lib/x86_64-linux-gnu/libc.so.6) +[rank15]: frame #35: __libc_start_main + 0x85 (0x7f508f650305 in /lib/x86_64-linux-gnu/libc.so.6) +[rank15]: frame #36: _start + 0x21 (0x628e41 in /usr/bin/python3) +[rank15]: . This may indicate a possible application crash on rank 0 or a network set up issue. +[rank13]: Traceback (most recent call last): +[rank13]: File "/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py", line 551, in +[rank13]: main(args) +[rank13]: File "/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py", line 233, in main +[rank13]: utils.init_distributed_mode(args) +[rank13]: File "/tmp/ScaleDF_repro_20260724/code/deit/utils.py", line 237, in init_distributed_mode +[rank13]: torch.distributed.barrier() +[rank13]: File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/c10d_logger.py", line 81, in wrapper +[rank13]: return func(*args, **kwargs) +[rank13]: ^^^^^^^^^^^^^^^^^^^^^ +[rank13]: File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py", line 4635, in barrier +[rank13]: work = group.barrier(opts=opts) +[rank13]: ^^^^^^^^^^^^^^^^^^^^^^^^ +[rank13]: torch.distributed.DistBackendError: [13] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0', but store->get('0') got error: failed to recv, got 0 bytes +[rank13]: Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +[rank13]: frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7fc31f9785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +[rank13]: frame #1: + 0x5ba8bfe (0x7fc308ccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank13]: frame #2: + 0x5bab100 (0x7fc308cd0100 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank13]: frame #3: c10d::TCPStore::doWait(c10::ArrayRef, std::allocator > >, std::chrono::duration >) + 0x28b (0x7fc308ccacab in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank13]: frame #4: c10d::TCPStore::doGet(std::__cxx11::basic_string, std::allocator > const&) + 0x33 (0x7fc308ccafa3 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank13]: frame #5: c10d::TCPStore::get(std::__cxx11::basic_string, std::allocator > const&) + 0xab (0x7fc308ccc08b in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank13]: frame #6: c10d::PrefixStore::get(std::__cxx11::basic_string, std::allocator > const&) + 0x2f (0x7fc308c7a94f in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank13]: frame #7: c10d::PrefixStore::get(std::__cxx11::basic_string, std::allocator > const&) + 0x2f (0x7fc308c7a94f in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank13]: frame #8: c10d::PrefixStore::get(std::__cxx11::basic_string, std::allocator > const&) + 0x2f (0x7fc308c7a94f in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank13]: frame #9: c10d::ProcessGroupNCCL::broadcastUniqueNCCLID(ncclUniqueId*, bool, std::__cxx11::basic_string, std::allocator > const&, int) + 0x16d (0x7fc2ca3c0e7d in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +[rank13]: frame #10: c10d::ProcessGroupNCCL::initNCCLComm(std::__cxx11::basic_string, std::allocator > const&, c10::Device&, c10d::OpType, int, bool) + 0x16bf (0x7fc2ca3d0f6f in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +[rank13]: frame #11: + 0x11ecf83 (0x7fc2ca3d5f83 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +[rank13]: frame #12: c10d::ProcessGroupNCCL::allreduce_impl(at::Tensor&, char const*, c10d::AllreduceOptions const&) + 0xee (0x7fc2ca3d677e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +[rank13]: frame #13: c10d::ProcessGroupNCCL::barrier(c10d::BarrierOptions const&) + 0x683 (0x7fc2ca3e63a3 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +[rank13]: frame #14: + 0x5b4363f (0x7fc308c6863f in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank13]: frame #15: + 0x5b52cfe (0x7fc308c77cfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank13]: frame #16: + 0x52125b0 (0x7fc3083375b0 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank13]: frame #17: + 0x5b61a02 (0x7fc308c86a02 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank13]: frame #18: + 0x5b6293d (0x7fc308c8793d in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +[rank13]: frame #19: + 0xc8c143 (0x7fc317bda143 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_python.so) +[rank13]: frame #20: + 0x38a1ac (0x7fc3172d81ac in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_python.so) +[rank13]: frame #21: /usr/bin/python3() [0x547d56] +[rank13]: frame #22: _PyObject_MakeTpCall + 0x223 (0x5180e3 in /usr/bin/python3) +[rank13]: frame #23: _PyEval_EvalFrameDefault + 0x8f0 (0x52bb30 in /usr/bin/python3) +[rank13]: frame #24: _PyFunction_Vectorcall + 0x191 (0x55cb71 in /usr/bin/python3) +[rank13]: frame #25: _PyEval_EvalFrameDefault + 0x48cc (0x52fb0c in /usr/bin/python3) +[rank13]: frame #26: PyEval_EvalCode + 0xbb (0x5237fb in /usr/bin/python3) +[rank13]: frame #27: /usr/bin/python3() [0x648fe7] +[rank13]: frame #28: /usr/bin/python3() [0x64693f] +[rank13]: frame #29: /usr/bin/python3() [0x6529d0] +[rank13]: frame #30: _PyRun_SimpleFileObject + 0x18b (0x65271b in /usr/bin/python3) +[rank13]: frame #31: _PyRun_AnyFileObject + 0x44 (0x652544 in /usr/bin/python3) +[rank13]: frame #32: Py_RunMain + 0x20f (0x6512cf in /usr/bin/python3) +[rank13]: frame #33: Py_BytesMain + 0x27 (0x628fa7 in /usr/bin/python3) +[rank13]: frame #34: + 0x2724a (0x7fc3259d424a in /lib/x86_64-linux-gnu/libc.so.6) +[rank13]: frame #35: __libc_start_main + 0x85 (0x7fc3259d4305 in /lib/x86_64-linux-gnu/libc.so.6) +[rank13]: frame #36: _start + 0x21 (0x628e41 in /usr/bin/python3) +[rank13]: . This may indicate a possible application crash on rank 0 or a network set up issue. +[rank8]:[W726 16:00:35.644868549 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 16:00:36.332000 2809 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 2874 closing signal SIGTERM +W0726 16:00:36.335000 2809 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 2875 closing signal SIGTERM +W0726 16:00:36.335000 2809 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 2877 closing signal SIGTERM +W0726 16:00:36.335000 2809 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 2878 closing signal SIGTERM +W0726 16:00:36.336000 2809 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 2879 closing signal SIGTERM +W0726 16:00:36.336000 2809 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 2880 closing signal SIGTERM +W0726 16:00:36.336000 2809 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 2881 closing signal SIGTERM +E0726 16:00:37.742000 2809 torch/distributed/elastic/multiprocessing/api.py:874] failed (exitcode: 1) local_rank: 2 (pid: 2876) of binary: /usr/bin/python3 +Traceback (most recent call last): + File "/usr/local/bin/torchrun", line 7, in + sys.exit(main()) + ^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 355, in wrapper + return f(*args, **kwargs) + ^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 892, in main + run(args) + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 883, in run + elastic_launch( + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 139, in __call__ + return launch_agent(self._config, self._entrypoint, list(args)) + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 270, in launch_agent + raise ChildFailedError( +torch.distributed.elastic.multiprocessing.errors.ChildFailedError: +============================================================ +main_cross_2direct_balance_realscale_perturb_v4_accum.py FAILED +------------------------------------------------------------ +Failures: + +------------------------------------------------------------ +Root Cause (first observed failure): +[0]: + time : 2026-07-26_16:00:36 + host : n111-046-078.byted.org + rank : 10 (local_rank: 2) + exitcode : 1 (pid: 2876) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +============================================================ diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1559_vith32_launch_master_n116-042-185_port25003/launcher_rank2_n116-040-048.log b/ScaleDF_rebuttal/_host_logs/20260726_1559_vith32_launch_master_n116-042-185_port25003/launcher_rank2_n116-040-048.log new file mode 100644 index 0000000000000000000000000000000000000000..134a44f2c19b9074eaa7e6011b4011b742a1019a --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1559_vith32_launch_master_n116-042-185_port25003/launcher_rank2_n116-040-048.log @@ -0,0 +1,399 @@ +started_utc=2026-07-26T15:59:49Z +hostname=n116-040-048 +node_rank=2 +nnodes=4 +nproc_per_node=8 +world_size=32 +master_addr=10.116.42.185 +master_port=25003 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +main=main_cross_2direct_balance_realscale_perturb_v4_accum.py +model_name=deit_huge_patch14_LS +pretrain=/tmp/ScaleDF_repro_20260724/pretrained/deit_3_huge_224_21k_v1.pth +resume=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_24card_effbs3072_from_epoch7_n116-041-246_n116-042-186_n116-042-185_20260726_0736_vith24_from_epoch7/checkpoint.pth +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_32card_effbs2048_from_latest24_master_n116-042-185_20260726_1559_vith32_port25003 +batch_per_gpu=64 +accum_iter=1 +effective_batch=2048 +eval_batch_per_gpu=64 +epochs=40 +W0726 16:00:02.696000 1314 torch/distributed/run.py:766] +W0726 16:00:02.696000 1314 torch/distributed/run.py:766] ***************************************** +W0726 16:00:02.696000 1314 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 16:00:02.696000 1314 torch/distributed/run.py:766] ***************************************** +Traceback (most recent call last): + File "/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py", line 13, in + from timm.data import Mixup + File "/home/tiger/.local/lib/python3.11/site-packages/timm/__init__.py", line 2, in +Traceback (most recent call last): +Traceback (most recent call last): +Traceback (most recent call last): + File "/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py", line 13, in +Traceback (most recent call last): + File "/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py", line 13, in + File "/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py", line 13, in +Traceback (most recent call last): + File "/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py", line 13, in + File "/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py", line 13, in +Traceback (most recent call last): + from timm.data import Mixup File "/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py", line 13, in +from timm.data import Mixup + +from timm.data import Mixup + from timm.data import Mixupfrom timm.data import Mixup + + File "/home/tiger/.local/lib/python3.11/site-packages/timm/__init__.py", line 2, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/__init__.py", line 2, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/__init__.py", line 2, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/__init__.py", line 2, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/__init__.py", line 2, in + from timm.data import Mixup + File "/home/tiger/.local/lib/python3.11/site-packages/timm/__init__.py", line 2, in +Traceback (most recent call last): + File "/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py", line 13, in + from timm.data import Mixup + File "/home/tiger/.local/lib/python3.11/site-packages/timm/__init__.py", line 2, in + from .models import create_model, list_models, is_model, list_modules, model_entrypoint, \from .models import create_model, list_models, is_model, list_modules, model_entrypoint, \ + + from .models import create_model, list_models, is_model, list_modules, model_entrypoint, \ + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/__init__.py", line 1, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/__init__.py", line 1, in + from .models import create_model, list_models, is_model, list_modules, model_entrypoint, \ + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/__init__.py", line 1, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/__init__.py", line 1, in +from .models import create_model, list_models, is_model, list_modules, model_entrypoint, \from .models import create_model, list_models, is_model, list_modules, model_entrypoint, \ +from .models import create_model, list_models, is_model, list_modules, model_entrypoint, \ + + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/__init__.py", line 1, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/__init__.py", line 1, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/__init__.py", line 1, in + from .byoanet import * + from .byoanet import *from .byoanet import *from .byoanet import * + + +from .byoanet import *from .byoanet import * File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/byoanet.py", line 15, in + + + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/byoanet.py", line 15, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/byoanet.py", line 15, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/byoanet.py", line 15, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/byoanet.py", line 15, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/byoanet.py", line 15, in + from .models import create_model, list_models, is_model, list_modules, model_entrypoint, \ + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/__init__.py", line 1, in + from .byoanet import * + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/byoanet.py", line 15, in + from .byoanet import * + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/byoanet.py", line 15, in + from timm.data import IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STDfrom timm.data import IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STD + + from timm.data import IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STD + from timm.data import IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STDfrom timm.data import IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STDfrom timm.data import IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STD + + +from timm.data import IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STD + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/__init__.py", line 5, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/__init__.py", line 5, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/__init__.py", line 5, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/__init__.py", line 5, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/__init__.py", line 5, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/__init__.py", line 5, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/__init__.py", line 5, in + from timm.data import IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STD + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/__init__.py", line 5, in + from .dataset import ImageDataset, IterableImageDataset, AugMixDatasetfrom .dataset import ImageDataset, IterableImageDataset, AugMixDataset + +from .dataset import ImageDataset, IterableImageDataset, AugMixDataset from .dataset import ImageDataset, IterableImageDataset, AugMixDataset from .dataset import ImageDataset, IterableImageDataset, AugMixDatasetfrom .dataset import ImageDataset, IterableImageDataset, AugMixDataset + + + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/dataset.py", line 12, in +from .dataset import ImageDataset, IterableImageDataset, AugMixDataset File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/dataset.py", line 12, in + + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/dataset.py", line 12, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/dataset.py", line 12, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/dataset.py", line 12, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/dataset.py", line 12, in + + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/dataset.py", line 12, in + from .dataset import ImageDataset, IterableImageDataset, AugMixDataset + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/dataset.py", line 12, in + from .parsers import create_parser + from .parsers import create_parser + from .parsers import create_parserfrom .parsers import create_parser File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/__init__.py", line 1, in + + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/__init__.py", line 1, in + +from .parsers import create_parser from .parsers import create_parser +from .parsers import create_parser + + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/__init__.py", line 1, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/__init__.py", line 1, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/__init__.py", line 1, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/__init__.py", line 1, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/__init__.py", line 1, in + from .parsers import create_parser + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/__init__.py", line 1, in + from .parser_factory import create_parserfrom .parser_factory import create_parser + +from .parser_factory import create_parser + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_factory.py", line 3, in +from .parser_factory import create_parser File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_factory.py", line 3, in + + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_factory.py", line 3, in +from .parser_factory import create_parserfrom .parser_factory import create_parserfrom .parser_factory import create_parser + + + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_factory.py", line 3, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_factory.py", line 3, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_factory.py", line 3, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_factory.py", line 3, in + from .parser_factory import create_parser + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_factory.py", line 3, in + from .parser_image_folder import ParserImageFolderfrom .parser_image_folder import ParserImageFolder + + from .parser_image_folder import ParserImageFolder +from .parser_image_folder import ParserImageFolderfrom .parser_image_folder import ParserImageFolder File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_image_folder.py", line 10, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_image_folder.py", line 10, in + + +from .parser_image_folder import ParserImageFolderfrom .parser_image_folder import ParserImageFolder + + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_image_folder.py", line 10, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_image_folder.py", line 10, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_image_folder.py", line 10, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_image_folder.py", line 10, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_image_folder.py", line 10, in + from .parser_image_folder import ParserImageFolder + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_image_folder.py", line 10, in + from timm.utils.misc import natural_keyfrom timm.utils.misc import natural_key + + from timm.utils.misc import natural_key + from timm.utils.misc import natural_key + from timm.utils.misc import natural_key File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/__init__.py", line 13, in + + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/__init__.py", line 13, in + from timm.utils.misc import natural_key + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/__init__.py", line 13, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/__init__.py", line 13, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/__init__.py", line 13, in + from timm.utils.misc import natural_key File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/__init__.py", line 13, in + + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/__init__.py", line 13, in + from timm.utils.misc import natural_key + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/__init__.py", line 13, in + from .summary import update_summary, get_outdir + from .summary import update_summary, get_outdir +from .summary import update_summary, get_outdir + from .summary import update_summary, get_outdir + from .summary import update_summary, get_outdir + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/summary.py", line 9, in +from .summary import update_summary, get_outdir File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/summary.py", line 9, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/summary.py", line 9, in + +from .summary import update_summary, get_outdir + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/summary.py", line 9, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/summary.py", line 9, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/summary.py", line 9, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/summary.py", line 9, in + from .summary import update_summary, get_outdir + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/summary.py", line 9, in + import wandb + import wandb + import wandbimport wandb + + File "/usr/local/lib/python3.11/dist-packages/wandb/__init__.py", line 35, in + import wandb File "/usr/local/lib/python3.11/dist-packages/wandb/__init__.py", line 35, in + + import wandbimport wandb + File "/usr/local/lib/python3.11/dist-packages/wandb/__init__.py", line 35, in + + File "/usr/local/lib/python3.11/dist-packages/wandb/__init__.py", line 35, in + File "/usr/local/lib/python3.11/dist-packages/wandb/__init__.py", line 35, in + File "/usr/local/lib/python3.11/dist-packages/wandb/__init__.py", line 35, in + File "/usr/local/lib/python3.11/dist-packages/wandb/__init__.py", line 35, in + import wandb + File "/usr/local/lib/python3.11/dist-packages/wandb/__init__.py", line 35, in + from wandb import sdk as wandb_sdk + from wandb import sdk as wandb_sdkfrom wandb import sdk as wandb_sdk from wandb import sdk as wandb_sdk +from wandb import sdk as wandb_sdk +from wandb import sdk as wandb_sdk + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/__init__.py", line 9, in + + +from wandb import sdk as wandb_sdk + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/__init__.py", line 9, in + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/__init__.py", line 9, in + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/__init__.py", line 9, in + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/__init__.py", line 9, in + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/__init__.py", line 9, in + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/__init__.py", line 9, in + from wandb import sdk as wandb_sdk + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/__init__.py", line 9, in + from .wandb_init import _attach, init # noqa: F401 + from .wandb_init import _attach, init # noqa: F401 + from .wandb_init import _attach, init # noqa: F401 + from .wandb_init import _attach, init # noqa: F401 + from .wandb_init import _attach, init # noqa: F401 ^ + ^ ^from .wandb_init import _attach, init # noqa: F401from .wandb_init import _attach, init # noqa: F401^ ^ ^ + + ^ ^ ^ ^^^ ^^^^^ ^^^ ^^ ^^^ ^^^^^^ ^^^^^^^^^ ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ +^^^^^^^^^^^^^^^^ File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/wandb_init.py", line 31, in +^^^ +^^^^^^^^^^^ File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/wandb_init.py", line 31, in +^^^^ +^^^^ + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/wandb_init.py", line 31, in +^^^^^^^ File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/wandb_init.py", line 31, in +^^^^ +^ +^ File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/wandb_init.py", line 31, in +^ + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/wandb_init.py", line 31, in + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/wandb_init.py", line 31, in + from .wandb_init import _attach, init # noqa: F401 + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/wandb_init.py", line 31, in + from .backend.backend import Backendfrom .backend.backend import Backend + + from .backend.backend import Backend + from .backend.backend import Backend + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/backend/backend.py", line 17, in + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/backend/backend.py", line 17, in +from .backend.backend import Backend File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/backend/backend.py", line 17, in +from .backend.backend import Backend + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/backend/backend.py", line 17, in + +from .backend.backend import Backend + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/backend/backend.py", line 17, in + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/backend/backend.py", line 17, in + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/backend/backend.py", line 17, in + from .backend.backend import Backend + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/backend/backend.py", line 17, in + from ..interface.interface import InterfaceBasefrom ..interface.interface import InterfaceBase + +from ..interface.interface import InterfaceBase +from ..interface.interface import InterfaceBase + from ..interface.interface import InterfaceBase + from ..interface.interface import InterfaceBase + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/interface/interface.py", line 34, in + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/interface/interface.py", line 34, in + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/interface/interface.py", line 34, in + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/interface/interface.py", line 34, in + from ..interface.interface import InterfaceBase + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/interface/interface.py", line 34, in + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/interface/interface.py", line 34, in + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/interface/interface.py", line 34, in + from ..interface.interface import InterfaceBase + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/interface/interface.py", line 34, in + from ..lib.mailbox import MailboxHandle + from ..lib.mailbox import MailboxHandle +from ..lib.mailbox import MailboxHandle + from ..lib.mailbox import MailboxHandle + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 92, in +from ..lib.mailbox import MailboxHandle File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 92, in + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 92, in +from ..lib.mailbox import MailboxHandle + + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 92, in +from ..lib.mailbox import MailboxHandle + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 92, in + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 92, in + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 92, in + class _MailboxSlot:class _MailboxSlot: + +from ..lib.mailbox import MailboxHandle + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 93, in _MailboxSlot + class _MailboxSlot: + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 93, in _MailboxSlot + class _MailboxSlot: + class _MailboxSlot: + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 92, in + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 93, in _MailboxSlot + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 93, in _MailboxSlot +class _MailboxSlot: + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 93, in _MailboxSlot + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 93, in _MailboxSlot +class _MailboxSlot: + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 93, in _MailboxSlot + _result: Optional[pb.Result] + _result: Optional[pb.Result] + ^^^ ^ ^ ^ ^ ^ ^ _result: Optional[pb.Result] + + ^AttributeError^^: ^module 'wandb.proto.wandb_internal_pb2' has no attribute 'Result'^^^^^ + AttributeError : module 'wandb.proto.wandb_internal_pb2' has no attribute 'Result' + class _MailboxSlot: + + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 93, in _MailboxSlot + ^^^^^^^^^ +AttributeError: module 'wandb.proto.wandb_internal_pb2' has no attribute 'Result' + _result: Optional[pb.Result] + ^^^^^^^^^ +AttributeError: module 'wandb.proto.wandb_internal_pb2' has no attribute 'Result' + _result: Optional[pb.Result] + _result: Optional[pb.Result]_result: Optional[pb.Result]^^ +^ +^^^^^^ + _result: Optional[pb.Result] +AttributeError: module 'wandb.proto.wandb_internal_pb2' has no attribute 'Result' + ^ ^ ^ ^^ ^^^ ^^ ^^ ^^^^^ +^^^^^AttributeError^ +^: ^module 'wandb.proto.wandb_internal_pb2' has no attribute 'Result'^AttributeError^ +: +module 'wandb.proto.wandb_internal_pb2' has no attribute 'Result'AttributeError: +module 'wandb.proto.wandb_internal_pb2' has no attribute 'Result' +W0726 16:00:26.213000 1314 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 1379 closing signal SIGTERM +W0726 16:00:26.213000 1314 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 1383 closing signal SIGTERM +E0726 16:00:26.327000 1314 torch/distributed/elastic/multiprocessing/api.py:874] failed (exitcode: 1) local_rank: 3 (pid: 1382) of binary: /usr/bin/python3 +Traceback (most recent call last): + File "/usr/local/bin/torchrun", line 7, in + sys.exit(main()) + ^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 355, in wrapper + return f(*args, **kwargs) + ^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 892, in main + run(args) + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 883, in run + elastic_launch( + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 139, in __call__ + return launch_agent(self._config, self._entrypoint, list(args)) + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 270, in launch_agent + raise ChildFailedError( +torch.distributed.elastic.multiprocessing.errors.ChildFailedError: +============================================================ +main_cross_2direct_balance_realscale_perturb_v4_accum.py FAILED +------------------------------------------------------------ +Failures: +[1]: + time : 2026-07-26_16:00:26 + host : n116-040-048.byted.org + rank : 21 (local_rank: 5) + exitcode : 1 (pid: 1384) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +[2]: + time : 2026-07-26_16:00:26 + host : n116-040-048.byted.org + rank : 22 (local_rank: 6) + exitcode : 1 (pid: 1385) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +[3]: + time : 2026-07-26_16:00:26 + host : n116-040-048.byted.org + rank : 23 (local_rank: 7) + exitcode : 1 (pid: 1386) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +------------------------------------------------------------ +Root Cause (first observed failure): +[0]: + time : 2026-07-26_16:00:26 + host : n116-040-048.byted.org + rank : 19 (local_rank: 3) + exitcode : 1 (pid: 1382) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +============================================================ diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1559_vith32_launch_master_n116-042-185_port25003/launcher_rank3_n111-046-125.log b/ScaleDF_rebuttal/_host_logs/20260726_1559_vith32_launch_master_n116-042-185_port25003/launcher_rank3_n111-046-125.log new file mode 100644 index 0000000000000000000000000000000000000000..56a6c01483e600be2516c9be055aaec6d37385ae --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1559_vith32_launch_master_n116-042-185_port25003/launcher_rank3_n111-046-125.log @@ -0,0 +1,384 @@ +started_utc=2026-07-26T15:59:49Z +hostname=n111-046-125 +node_rank=3 +nnodes=4 +nproc_per_node=8 +world_size=32 +master_addr=10.116.42.185 +master_port=25003 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +main=main_cross_2direct_balance_realscale_perturb_v4_accum.py +model_name=deit_huge_patch14_LS +pretrain=/tmp/ScaleDF_repro_20260724/pretrained/deit_3_huge_224_21k_v1.pth +resume=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_24card_effbs3072_from_epoch7_n116-041-246_n116-042-186_n116-042-185_20260726_0736_vith24_from_epoch7/checkpoint.pth +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_32card_effbs2048_from_latest24_master_n116-042-185_20260726_1559_vith32_port25003 +batch_per_gpu=64 +accum_iter=1 +effective_batch=2048 +eval_batch_per_gpu=64 +epochs=40 +W0726 16:00:04.581000 1193 torch/distributed/run.py:766] +W0726 16:00:04.581000 1193 torch/distributed/run.py:766] ***************************************** +W0726 16:00:04.581000 1193 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 16:00:04.581000 1193 torch/distributed/run.py:766] ***************************************** +Traceback (most recent call last): +Traceback (most recent call last): +Traceback (most recent call last): +Traceback (most recent call last): + File "/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py", line 13, in + File "/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py", line 13, in + File "/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py", line 13, in + File "/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py", line 13, in +Traceback (most recent call last): +Traceback (most recent call last): + File "/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py", line 13, in + File "/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py", line 13, in +Traceback (most recent call last): + File "/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py", line 13, in + from timm.data import Mixupfrom timm.data import Mixupfrom timm.data import Mixup + + +from timm.data import Mixupfrom timm.data import Mixup + +from timm.data import Mixup + File "/home/tiger/.local/lib/python3.11/site-packages/timm/__init__.py", line 2, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/__init__.py", line 2, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/__init__.py", line 2, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/__init__.py", line 2, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/__init__.py", line 2, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/__init__.py", line 2, in + from timm.data import Mixup + File "/home/tiger/.local/lib/python3.11/site-packages/timm/__init__.py", line 2, in +Traceback (most recent call last): + File "/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py", line 13, in + from timm.data import Mixup + File "/home/tiger/.local/lib/python3.11/site-packages/timm/__init__.py", line 2, in + from .models import create_model, list_models, is_model, list_modules, model_entrypoint, \ + from .models import create_model, list_models, is_model, list_modules, model_entrypoint, \from .models import create_model, list_models, is_model, list_modules, model_entrypoint, \ + + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/__init__.py", line 1, in + from .models import create_model, list_models, is_model, list_modules, model_entrypoint, \from .models import create_model, list_models, is_model, list_modules, model_entrypoint, \ File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/__init__.py", line 1, in + + +from .models import create_model, list_models, is_model, list_modules, model_entrypoint, \ File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/__init__.py", line 1, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/__init__.py", line 1, in + + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/__init__.py", line 1, in +from .models import create_model, list_models, is_model, list_modules, model_entrypoint, \ File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/__init__.py", line 1, in + + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/__init__.py", line 1, in + from .models import create_model, list_models, is_model, list_modules, model_entrypoint, \ + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/__init__.py", line 1, in + from .byoanet import * +from .byoanet import * + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/byoanet.py", line 15, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/byoanet.py", line 15, in + from .byoanet import * + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/byoanet.py", line 15, in + from .byoanet import * + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/byoanet.py", line 15, in + from .byoanet import * + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/byoanet.py", line 15, in + from timm.data import IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STDfrom .byoanet import * + + from .byoanet import * + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/byoanet.py", line 15, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/byoanet.py", line 15, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/__init__.py", line 5, in +from .byoanet import * + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/byoanet.py", line 15, in + from timm.data import IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STD + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/__init__.py", line 5, in + from timm.data import IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STD + from timm.data import IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STD + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/__init__.py", line 5, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/__init__.py", line 5, in + from timm.data import IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STD + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/__init__.py", line 5, in + from timm.data import IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STD + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/__init__.py", line 5, in + from .dataset import ImageDataset, IterableImageDataset, AugMixDataset +from .dataset import ImageDataset, IterableImageDataset, AugMixDataset + from .dataset import ImageDataset, IterableImageDataset, AugMixDataset + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/dataset.py", line 12, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/dataset.py", line 12, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/dataset.py", line 12, in + from timm.data import IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STD + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/__init__.py", line 5, in + from timm.data import IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STD + from .dataset import ImageDataset, IterableImageDataset, AugMixDataset + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/__init__.py", line 5, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/dataset.py", line 12, in + from .dataset import ImageDataset, IterableImageDataset, AugMixDataset + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/dataset.py", line 12, in + from .parsers import create_parserfrom .parsers import create_parser + + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/__init__.py", line 1, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/__init__.py", line 1, in + from .dataset import ImageDataset, IterableImageDataset, AugMixDataset + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/dataset.py", line 12, in + from .dataset import ImageDataset, IterableImageDataset, AugMixDataset + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/dataset.py", line 12, in + from .parsers import create_parser + from .dataset import ImageDataset, IterableImageDataset, AugMixDataset + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/__init__.py", line 1, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/dataset.py", line 12, in + from .parsers import create_parser + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/__init__.py", line 1, in + from .parsers import create_parser + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/__init__.py", line 1, in + from .parsers import create_parser + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/__init__.py", line 1, in + from .parser_factory import create_parser + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_factory.py", line 3, in + from .parser_factory import create_parserfrom .parser_factory import create_parser + + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_factory.py", line 3, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_factory.py", line 3, in + from .parsers import create_parser + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/__init__.py", line 1, in + from .parsers import create_parser + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/__init__.py", line 1, in + from .parser_factory import create_parser + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_factory.py", line 3, in + from .parser_factory import create_parser + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_factory.py", line 3, in + from .parser_factory import create_parser + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_factory.py", line 3, in + from .parser_image_folder import ParserImageFolderfrom .parser_image_folder import ParserImageFolderfrom .parser_image_folder import ParserImageFolder + + + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_image_folder.py", line 10, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_image_folder.py", line 10, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_image_folder.py", line 10, in + from .parser_factory import create_parser + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_factory.py", line 3, in + from .parser_factory import create_parser + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_factory.py", line 3, in + from .parser_image_folder import ParserImageFolder + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_image_folder.py", line 10, in + from .parser_image_folder import ParserImageFolder + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_image_folder.py", line 10, in + from .parser_image_folder import ParserImageFolder + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_image_folder.py", line 10, in + from timm.utils.misc import natural_key + from timm.utils.misc import natural_keyfrom timm.utils.misc import natural_key + + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/__init__.py", line 13, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/__init__.py", line 13, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/__init__.py", line 13, in + from .parser_image_folder import ParserImageFolder + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_image_folder.py", line 10, in + from .parser_image_folder import ParserImageFolder + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_image_folder.py", line 10, in + from timm.utils.misc import natural_key + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/__init__.py", line 13, in + from timm.utils.misc import natural_key + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/__init__.py", line 13, in + from timm.utils.misc import natural_key + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/__init__.py", line 13, in + from .summary import update_summary, get_outdir + from .summary import update_summary, get_outdirfrom .summary import update_summary, get_outdir + + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/summary.py", line 9, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/summary.py", line 9, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/summary.py", line 9, in + from timm.utils.misc import natural_key + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/__init__.py", line 13, in + from timm.utils.misc import natural_key + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/__init__.py", line 13, in + from .summary import update_summary, get_outdir + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/summary.py", line 9, in + import wandbimport wandb + + from .summary import update_summary, get_outdir + File "/usr/local/lib/python3.11/dist-packages/wandb/__init__.py", line 35, in + File "/usr/local/lib/python3.11/dist-packages/wandb/__init__.py", line 35, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/summary.py", line 9, in + from .summary import update_summary, get_outdir + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/summary.py", line 9, in + from .summary import update_summary, get_outdir + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/summary.py", line 9, in + from .summary import update_summary, get_outdir + import wandb + File "/usr/local/lib/python3.11/dist-packages/wandb/__init__.py", line 35, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/summary.py", line 9, in + import wandb + File "/usr/local/lib/python3.11/dist-packages/wandb/__init__.py", line 35, in + from wandb import sdk as wandb_sdk + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/__init__.py", line 9, in + from wandb import sdk as wandb_sdk + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/__init__.py", line 9, in + import wandb + File "/usr/local/lib/python3.11/dist-packages/wandb/__init__.py", line 35, in + import wandb + import wandb + File "/usr/local/lib/python3.11/dist-packages/wandb/__init__.py", line 35, in + File "/usr/local/lib/python3.11/dist-packages/wandb/__init__.py", line 35, in + import wandb + File "/usr/local/lib/python3.11/dist-packages/wandb/__init__.py", line 35, in + from wandb import sdk as wandb_sdk + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/__init__.py", line 9, in + from wandb import sdk as wandb_sdk + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/__init__.py", line 9, in + from wandb import sdk as wandb_sdk + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/__init__.py", line 9, in + from wandb import sdk as wandb_sdk + from wandb import sdk as wandb_sdk + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/__init__.py", line 9, in + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/__init__.py", line 9, in + from .wandb_init import _attach, init # noqa: F401 + from .wandb_init import _attach, init # noqa: F401 + from .wandb_init import _attach, init # noqa: F401 + from .wandb_init import _attach, init # noqa: F401 +^^^^^^^^^^^^^^^^^^^^^ ^ ^ ^ ^^^^^^^^^^^^^^ + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/wandb_init.py", line 31, in + ^ ^ ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ +^^^^^^ File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/wandb_init.py", line 31, in +^^ +^^ + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/wandb_init.py", line 31, in + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/wandb_init.py", line 31, in + from wandb import sdk as wandb_sdk + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/__init__.py", line 9, in + from .wandb_init import _attach, init # noqa: F401 + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/wandb_init.py", line 31, in + from .wandb_init import _attach, init # noqa: F401 + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/wandb_init.py", line 31, in + from .wandb_init import _attach, init # noqa: F401 + ^^^^^^^^^^^^^^^^^^^ ^from .backend.backend import Backendfrom .backend.backend import Backend ^from .backend.backend import Backend^ + + +^^^^^^^ ^from .backend.backend import Backend^ +^ File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/backend/backend.py", line 17, in + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/backend/backend.py", line 17, in +^^ ^ File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/backend/backend.py", line 17, in +from .wandb_init import _attach, init # noqa: F401^^ + + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/backend/backend.py", line 17, in + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/wandb_init.py", line 31, in + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/wandb_init.py", line 31, in + from .backend.backend import Backend + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/backend/backend.py", line 17, in + from .backend.backend import Backend + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/backend/backend.py", line 17, in + from ..interface.interface import InterfaceBase + from ..interface.interface import InterfaceBase + from ..interface.interface import InterfaceBase + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/interface/interface.py", line 34, in +from ..interface.interface import InterfaceBase + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/interface/interface.py", line 34, in + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/interface/interface.py", line 34, in + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/interface/interface.py", line 34, in + from .backend.backend import Backend + from .backend.backend import Backend + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/backend/backend.py", line 17, in + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/backend/backend.py", line 17, in + from ..interface.interface import InterfaceBase + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/interface/interface.py", line 34, in + from ..interface.interface import InterfaceBase + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/interface/interface.py", line 34, in + from ..lib.mailbox import MailboxHandlefrom ..lib.mailbox import MailboxHandle + + from ..lib.mailbox import MailboxHandle + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 92, in + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 92, in + from ..lib.mailbox import MailboxHandle + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 92, in + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 92, in + from ..interface.interface import InterfaceBase + from ..interface.interface import InterfaceBase + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/interface/interface.py", line 34, in + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/interface/interface.py", line 34, in + from ..lib.mailbox import MailboxHandle + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 92, in + from ..lib.mailbox import MailboxHandle + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 92, in + class _MailboxSlot: + class _MailboxSlot: + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 93, in _MailboxSlot + class _MailboxSlot: + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 93, in _MailboxSlot + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 93, in _MailboxSlot + class _MailboxSlot: + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 93, in _MailboxSlot + from ..lib.mailbox import MailboxHandle + from ..lib.mailbox import MailboxHandle + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 92, in + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 92, in + class _MailboxSlot: + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 93, in _MailboxSlot + _result: Optional[pb.Result] + _result: Optional[pb.Result] + _result: Optional[pb.Result] + ^^ class _MailboxSlot: ^ + ^ ^^ ^ File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 93, in _MailboxSlot +^ ^^ ^^ class _MailboxSlot:^^ ^ + + ^ File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 93, in _MailboxSlot +AttributeError : ^module 'wandb.proto.wandb_internal_pb2' has no attribute 'Result' ^ + +^AttributeError^^: ^^module 'wandb.proto.wandb_internal_pb2' has no attribute 'Result'^^^^ + + _result: Optional[pb.Result]AttributeError +: module 'wandb.proto.wandb_internal_pb2' has no attribute 'Result' + class _MailboxSlot: + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 93, in _MailboxSlot + ^^^^^^^^^ +AttributeError: module 'wandb.proto.wandb_internal_pb2' has no attribute 'Result' + _result: Optional[pb.Result] + ^^^^^^^^^ + _result: Optional[pb.Result] +AttributeError: module 'wandb.proto.wandb_internal_pb2' has no attribute 'Result' + _result: Optional[pb.Result] + ^^^^^^^^^ + AttributeError : module 'wandb.proto.wandb_internal_pb2' has no attribute 'Result' + ^^^^^^^^^ +AttributeError: module 'wandb.proto.wandb_internal_pb2' has no attribute 'Result' + _result: Optional[pb.Result] + ^^^^^^^^^ +AttributeError: module 'wandb.proto.wandb_internal_pb2' has no attribute 'Result' +W0726 16:00:30.113000 1193 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 1259 closing signal SIGTERM +W0726 16:00:30.116000 1193 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 1260 closing signal SIGTERM +W0726 16:00:30.116000 1193 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 1261 closing signal SIGTERM +W0726 16:00:30.117000 1193 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 1262 closing signal SIGTERM +W0726 16:00:30.117000 1193 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 1263 closing signal SIGTERM +W0726 16:00:30.117000 1193 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 1264 closing signal SIGTERM +W0726 16:00:30.117000 1193 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 1265 closing signal SIGTERM +E0726 16:00:30.238000 1193 torch/distributed/elastic/multiprocessing/api.py:874] failed (exitcode: 1) local_rank: 0 (pid: 1258) of binary: /usr/bin/python3 +Traceback (most recent call last): + File "/usr/local/bin/torchrun", line 7, in + sys.exit(main()) + ^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 355, in wrapper + return f(*args, **kwargs) + ^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 892, in main + run(args) + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 883, in run + elastic_launch( + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 139, in __call__ + return launch_agent(self._config, self._entrypoint, list(args)) + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 270, in launch_agent + raise ChildFailedError( +torch.distributed.elastic.multiprocessing.errors.ChildFailedError: +============================================================ +main_cross_2direct_balance_realscale_perturb_v4_accum.py FAILED +------------------------------------------------------------ +Failures: + +------------------------------------------------------------ +Root Cause (first observed failure): +[0]: + time : 2026-07-26_16:00:30 + host : n111-046-125.byted.org + rank : 24 (local_rank: 0) + exitcode : 1 (pid: 1258) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +============================================================ diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1559_vith32_launch_master_n116-042-185_port25003/output_root.txt b/ScaleDF_rebuttal/_host_logs/20260726_1559_vith32_launch_master_n116-042-185_port25003/output_root.txt new file mode 100644 index 0000000000000000000000000000000000000000..fdda2b3ccee9f5ca9766a130c3f778ee3347b493 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1559_vith32_launch_master_n116-042-185_port25003/output_root.txt @@ -0,0 +1 @@ +/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_32card_effbs2048_from_latest24_master_n116-042-185_20260726_1559_vith32_port25003 diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1559_vith32_launch_master_n116-042-185_port25003/pid_rank0_n116-042-185.txt b/ScaleDF_rebuttal/_host_logs/20260726_1559_vith32_launch_master_n116-042-185_port25003/pid_rank0_n116-042-185.txt new file mode 100644 index 0000000000000000000000000000000000000000..5d030fe7d27f82cfb108ab489f7853e9c1e0983f --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1559_vith32_launch_master_n116-042-185_port25003/pid_rank0_n116-042-185.txt @@ -0,0 +1 @@ +1495 diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1559_vith32_launch_master_n116-042-185_port25003/pid_rank1_n111-046-078.txt b/ScaleDF_rebuttal/_host_logs/20260726_1559_vith32_launch_master_n116-042-185_port25003/pid_rank1_n111-046-078.txt new file mode 100644 index 0000000000000000000000000000000000000000..cd0c00e38b8918d29b260e6ebefdfd2141b40c1a --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1559_vith32_launch_master_n116-042-185_port25003/pid_rank1_n111-046-078.txt @@ -0,0 +1 @@ +2801 diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1559_vith32_launch_master_n116-042-185_port25003/pid_rank2_n116-040-048.txt b/ScaleDF_rebuttal/_host_logs/20260726_1559_vith32_launch_master_n116-042-185_port25003/pid_rank2_n116-040-048.txt new file mode 100644 index 0000000000000000000000000000000000000000..767efdbde5b13eefa493eee4dc09adc8947b2f85 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1559_vith32_launch_master_n116-042-185_port25003/pid_rank2_n116-040-048.txt @@ -0,0 +1 @@ +1306 diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1559_vith32_launch_master_n116-042-185_port25003/pid_rank3_n111-046-125.txt b/ScaleDF_rebuttal/_host_logs/20260726_1559_vith32_launch_master_n116-042-185_port25003/pid_rank3_n111-046-125.txt new file mode 100644 index 0000000000000000000000000000000000000000..73a426e80e5db4241e05405c11a9eed984b24db1 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1559_vith32_launch_master_n116-042-185_port25003/pid_rank3_n111-046-125.txt @@ -0,0 +1 @@ +1185 diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1559_vith32_launch_master_n116-042-185_port25003/resume.txt b/ScaleDF_rebuttal/_host_logs/20260726_1559_vith32_launch_master_n116-042-185_port25003/resume.txt new file mode 100644 index 0000000000000000000000000000000000000000..2158d35929432dee54a9499c09d7990967633387 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1559_vith32_launch_master_n116-042-185_port25003/resume.txt @@ -0,0 +1 @@ +/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_24card_effbs3072_from_epoch7_n116-041-246_n116-042-186_n116-042-185_20260726_0736_vith24_from_epoch7/checkpoint.pth diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1559_vith32_launch_master_n116-042-185_port25003/run_name.txt b/ScaleDF_rebuttal/_host_logs/20260726_1559_vith32_launch_master_n116-042-185_port25003/run_name.txt new file mode 100644 index 0000000000000000000000000000000000000000..809d603bddd08011a978a600ae3a67d6c4f0bb5e --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1559_vith32_launch_master_n116-042-185_port25003/run_name.txt @@ -0,0 +1 @@ +modelsize_vith_40epoch_32card_effbs2048_from_latest24_master_n116-042-185_20260726_1559_vith32_port25003 diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1603_vith32_launch_stub_master_n116-042-185_port25003/launcher_rank0_n116-042-185.log b/ScaleDF_rebuttal/_host_logs/20260726_1603_vith32_launch_stub_master_n116-042-185_port25003/launcher_rank0_n116-042-185.log new file mode 100644 index 0000000000000000000000000000000000000000..05f36df8dfaf1c66fa1911dc001422a1caa5a604 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1603_vith32_launch_stub_master_n116-042-185_port25003/launcher_rank0_n116-042-185.log @@ -0,0 +1,14538 @@ +started_utc=2026-07-26T16:03:56Z +hostname=n116-042-185 +node_rank=0 +nnodes=4 +nproc_per_node=8 +world_size=32 +master_addr=10.116.42.185 +master_port=25003 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +main=main_cross_2direct_balance_realscale_perturb_v4_accum.py +model_name=deit_huge_patch14_LS +pretrain=/tmp/ScaleDF_repro_20260724/pretrained/deit_3_huge_224_21k_v1.pth +resume=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_24card_effbs3072_from_epoch7_n116-041-246_n116-042-186_n116-042-185_20260726_0736_vith24_from_epoch7/checkpoint.pth +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_32card_effbs2048_from_latest24_master_n116-042-185_20260726_1603_vith32_stub_port25003 +batch_per_gpu=64 +accum_iter=1 +effective_batch=2048 +eval_batch_per_gpu=64 +epochs=40 +W0726 16:04:12.427000 1915 torch/distributed/run.py:766] +W0726 16:04:12.427000 1915 torch/distributed/run.py:766] ***************************************** +W0726 16:04:12.427000 1915 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 16:04:12.427000 1915 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +| distributed init (rank 3): env:// +| distributed init (rank 4): env:// +[rank4]:[W726 16:05:06.660922780 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W726 16:05:06.661304673 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 16:05:06.661335962 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +[rank6]:[W726 16:05:06.729545496 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +n116-042-185:1981:1981 [0] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-042-185:1981:1981 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-042-185:1981:1981 [0] NCCL INFO cudaDriverVersion 12060 +n116-042-185:1981:1981 [0] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-042-185:1981:1981 [0] NCCL INFO Comm config Blocking set to 1 +| distributed init (rank 7): env:// +[rank7]:[W726 16:05:06.968179621 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +n116-042-185:1985:1985 [4] NCCL INFO cudaDriverVersion 12060 +n116-042-185:1985:1985 [4] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-042-185:1985:1985 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-042-185:1985:1985 [4] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-042-185:1985:1985 [4] NCCL INFO Comm config Blocking set to 1 +n116-042-185:1987:1987 [6] NCCL INFO cudaDriverVersion 12060 +n116-042-185:1987:1987 [6] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-042-185:1987:1987 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-042-185:1987:1987 [6] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-042-185:1987:1987 [6] NCCL INFO Comm config Blocking set to 1 +n116-042-185:1984:1984 [3] NCCL INFO cudaDriverVersion 12060 +n116-042-185:1984:1984 [3] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-042-185:1984:1984 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-042-185:1984:1984 [3] NCCL INFO NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +n116-042-185:1984:1984 [3] NCCL INFO Comm config Blocking set to 1 +[rank5]:[W726 16:05:06.054052092 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +n116-042-185:1988:1988 [7] NCCL INFO cudaDriverVersion 12060 +n116-042-185:1988:1988 [7] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-042-185:1988:1988 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-042-185:1988:1988 [7] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-042-185:1988:1988 [7] NCCL INFO Comm config Blocking set to 1 +| distributed init (rank 1): env:// +[rank1]:[W726 16:05:06.100052526 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +n116-042-185:1986:1986 [5] NCCL INFO cudaDriverVersion 12060 +n116-042-185:1986:1986 [5] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-042-185:1986:1986 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-042-185:1986:1986 [5] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-042-185:1986:1986 [5] NCCL INFO Comm config Blocking set to 1 +| distributed init (rank 2): env:// +n116-042-185:1982:1982 [1] NCCL INFO cudaDriverVersion 12060 +n116-042-185:1982:1982 [1] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-042-185:1982:1982 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +[rank2]:[W726 16:05:06.131601504 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +n116-042-185:1982:1982 [1] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-042-185:1982:1982 [1] NCCL INFO Comm config Blocking set to 1 +n116-042-185:1983:1983 [2] NCCL INFO cudaDriverVersion 12060 +n116-042-185:1983:1983 [2] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-042-185:1983:1983 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-042-185:1983:1983 [2] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-042-185:1983:1983 [2] NCCL INFO Comm config Blocking set to 1 +n116-042-185:1987:2034 [6] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-042-185:1984:2037 [3] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-042-185:1985:2033 [4] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-042-185:1983:2045 [2] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-042-185:1981:2030 [0] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-042-185:1986:2041 [5] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-042-185:1982:2044 [1] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-042-185:1983:2045 [2] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-042-185:1984:2037 [3] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-042-185:1985:2033 [4] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-042-185:1987:2034 [6] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-042-185:1986:2041 [5] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-042-185:1981:2030 [0] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-042-185:1983:2045 [2] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-042-185:1985:2033 [4] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-042-185:1984:2037 [3] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-042-185:1986:2041 [5] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-042-185:1987:2034 [6] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-042-185:1983:2045 [2] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-042-185:1981:2030 [0] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-042-185:1984:2037 [3] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-042-185:1987:2034 [6] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-042-185:1985:2033 [4] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-042-185:1982:2044 [1] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-042-185:1986:2041 [5] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-042-185:1981:2030 [0] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-042-185:1982:2044 [1] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-042-185:1982:2044 [1] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-042-185:1983:2045 [2] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-042-185:1987:2034 [6] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-042-185:1986:2041 [5] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-042-185:1985:2033 [4] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-042-185:1983:2045 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-042-185:1987:2034 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-042-185:1986:2041 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-042-185:1985:2033 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-042-185:1987:2034 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-042-185:1983:2045 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-042-185:1986:2041 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-042-185:1984:2037 [3] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-042-185:1981:2030 [0] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-042-185:1987:2034 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-042-185:1983:2045 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-042-185:1986:2041 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-042-185:1987:2034 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-042-185:1982:2044 [1] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-042-185:1985:2033 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-042-185:1984:2037 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-042-185:1981:2030 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-042-185:1983:2045 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-042-185:1986:2041 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-042-185:1987:2034 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-042-185:1982:2044 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-042-185:1985:2033 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-042-185:1984:2037 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-042-185:1981:2030 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-042-185:1985:2033 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-042-185:1983:2045 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-042-185:1986:2041 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-042-185:1988:2038 [7] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-042-185:1982:2044 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-042-185:1984:2037 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-042-185:1985:2033 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-042-185:1982:2044 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-042-185:1984:2037 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-042-185:1981:2030 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-042-185:1982:2044 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-042-185:1984:2037 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-042-185:1981:2030 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-042-185:1982:2044 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-042-185:1981:2030 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-042-185:1988:2038 [7] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-042-185:1988:2038 [7] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-042-185:1988:2038 [7] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-042-185:1988:2038 [7] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-042-185:1988:2038 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-042-185:1988:2038 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-042-185:1988:2038 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-042-185:1988:2038 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-042-185:1988:2038 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-042-185:1981:2030 [0] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-042-185:1981:2030 [0] NCCL INFO Using network FasTrak +n116-042-185:1988:2038 [7] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-042-185:1988:2038 [7] NCCL INFO Using network FasTrak +n116-042-185:1983:2045 [2] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-042-185:1983:2045 [2] NCCL INFO Using network FasTrak +n116-042-185:1982:2044 [1] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-042-185:1982:2044 [1] NCCL INFO Using network FasTrak +n116-042-185:1984:2037 [3] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-042-185:1984:2037 [3] NCCL INFO Using network FasTrak +n116-042-185:1986:2041 [5] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-042-185:1986:2041 [5] NCCL INFO Using network FasTrak +n116-042-185:1985:2033 [4] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-042-185:1985:2033 [4] NCCL INFO Using network FasTrak +n116-042-185:1987:2034 [6] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-042-185:1987:2034 [6] NCCL INFO Using network FasTrak +n116-042-185:1981:2030 [0] NCCL INFO DMA-BUF is available on GPU device 0 +n116-042-185:1988:2038 [7] NCCL INFO DMA-BUF is available on GPU device 7 +n116-042-185:1981:2030 [0] NCCL INFO ncclCommInitRankConfig comm 0xc1bf960 rank 0 nranks 32 cudaDev 0 nvmlDev 0 busId 4000 commId 0x2a4d35ea441cbcba - Init START +n116-042-185:1983:2045 [2] NCCL INFO DMA-BUF is available on GPU device 2 +n116-042-185:1988:2038 [7] NCCL INFO ncclCommInitRankConfig comm 0xb3ad0e0 rank 7 nranks 32 cudaDev 7 nvmlDev 7 busId 8c000 commId 0x2a4d35ea441cbcba - Init START +n116-042-185:1986:2041 [5] NCCL INFO DMA-BUF is available on GPU device 5 +n116-042-185:1982:2044 [1] NCCL INFO DMA-BUF is available on GPU device 1 +n116-042-185:1983:2045 [2] NCCL INFO ncclCommInitRankConfig comm 0xc51c1b0 rank 2 nranks 32 cudaDev 2 nvmlDev 2 busId b000 commId 0x2a4d35ea441cbcba - Init START +n116-042-185:1985:2033 [4] NCCL INFO DMA-BUF is available on GPU device 4 +n116-042-185:1986:2041 [5] NCCL INFO ncclCommInitRankConfig comm 0xbf2f3e0 rank 5 nranks 32 cudaDev 5 nvmlDev 5 busId 85000 commId 0x2a4d35ea441cbcba - Init START +n116-042-185:1982:2044 [1] NCCL INFO ncclCommInitRankConfig comm 0xcdda240 rank 1 nranks 32 cudaDev 1 nvmlDev 1 busId 5000 commId 0x2a4d35ea441cbcba - Init START +n116-042-185:1984:2037 [3] NCCL INFO DMA-BUF is available on GPU device 3 +n116-042-185:1984:2037 [3] NCCL INFO ncclCommInitRankConfig comm 0xd0746a0 rank 3 nranks 32 cudaDev 3 nvmlDev 3 busId c000 commId 0x2a4d35ea441cbcba - Init START +n116-042-185:1985:2033 [4] NCCL INFO ncclCommInitRankConfig comm 0xc15b3b0 rank 4 nranks 32 cudaDev 4 nvmlDev 4 busId 84000 commId 0x2a4d35ea441cbcba - Init START +n116-042-185:1982:2044 [1] NCCL INFO RAS client listening socket at ::1<28028> +n116-042-185:1987:2034 [6] NCCL INFO DMA-BUF is available on GPU device 6 +n116-042-185:1984:2037 [3] NCCL INFO RAS client listening socket at ::1<28028> +n116-042-185:1983:2045 [2] NCCL INFO RAS client listening socket at ::1<28028> +n116-042-185:1985:2033 [4] NCCL INFO RAS client listening socket at ::1<28028> +n116-042-185:1987:2034 [6] NCCL INFO ncclCommInitRankConfig comm 0xc4fc250 rank 6 nranks 32 cudaDev 6 nvmlDev 6 busId 8b000 commId 0x2a4d35ea441cbcba - Init START +n116-042-185:1987:2034 [6] NCCL INFO RAS client listening socket at ::1<28028> +n116-042-185:1986:2041 [5] NCCL INFO RAS client listening socket at ::1<28028> +n116-042-185:1988:2038 [7] NCCL INFO RAS client listening socket at ::1<28028> +n116-042-185:1981:2030 [0] NCCL INFO RAS client listening socket at ::1<28028> +n116-042-185:1985:2033 [4] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-042-185:1985:2033 [4] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-042-185:1985:2033 [4] NCCL INFO Retrieving state for FasTrak +n116-042-185:1985:2033 [4] NCCL INFO Initialized state 0 for FasTrak +n116-042-185:1985:2033 [4] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-042-185:1985:2033 [4] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-042-185:1985:2033 [4] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-042-185:1985:2033 [4] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-042-185:1985:2033 [4] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-042-185:1985:2033 [4] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-042-185:1985:2033 [4] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-042-185:1985:2033 [4] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-042-185:1982:2044 [1] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-042-185:1982:2044 [1] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-042-185:1982:2044 [1] NCCL INFO Retrieving state for FasTrak +n116-042-185:1982:2044 [1] NCCL INFO Initialized state 0 for FasTrak +n116-042-185:1982:2044 [1] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-042-185:1982:2044 [1] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-042-185:1982:2044 [1] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-042-185:1982:2044 [1] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-042-185:1982:2044 [1] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-042-185:1982:2044 [1] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-042-185:1982:2044 [1] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-042-185:1982:2044 [1] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-042-185:1987:2034 [6] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-042-185:1987:2034 [6] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-042-185:1987:2034 [6] NCCL INFO Retrieving state for FasTrak +n116-042-185:1987:2034 [6] NCCL INFO Initialized state 0 for FasTrak +n116-042-185:1987:2034 [6] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-042-185:1987:2034 [6] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-042-185:1987:2034 [6] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-042-185:1987:2034 [6] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-042-185:1987:2034 [6] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-042-185:1984:2037 [3] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-042-185:1988:2038 [7] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-042-185:1984:2037 [3] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-042-185:1984:2037 [3] NCCL INFO Retrieving state for FasTrak +n116-042-185:1984:2037 [3] NCCL INFO Initialized state 0 for FasTrak +n116-042-185:1988:2038 [7] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-042-185:1988:2038 [7] NCCL INFO Retrieving state for FasTrak +n116-042-185:1988:2038 [7] NCCL INFO Initialized state 0 for FasTrak +n116-042-185:1987:2034 [6] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-042-185:1987:2034 [6] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-042-185:1984:2037 [3] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-042-185:1988:2038 [7] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-042-185:1987:2034 [6] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-042-185:1984:2037 [3] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-042-185:1988:2038 [7] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-042-185:1984:2037 [3] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-042-185:1983:2045 [2] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-042-185:1983:2045 [2] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-042-185:1983:2045 [2] NCCL INFO Retrieving state for FasTrak +n116-042-185:1983:2045 [2] NCCL INFO Initialized state 0 for FasTrak +n116-042-185:1984:2037 [3] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-042-185:1988:2038 [7] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-042-185:1983:2045 [2] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-042-185:1984:2037 [3] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-042-185:1988:2038 [7] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-042-185:1983:2045 [2] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-042-185:1984:2037 [3] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-042-185:1983:2045 [2] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-042-185:1986:2041 [5] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-042-185:1986:2041 [5] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-042-185:1986:2041 [5] NCCL INFO Retrieving state for FasTrak +n116-042-185:1986:2041 [5] NCCL INFO Initialized state 0 for FasTrak +n116-042-185:1984:2037 [3] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-042-185:1983:2045 [2] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-042-185:1988:2038 [7] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-042-185:1984:2037 [3] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-042-185:1988:2038 [7] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-042-185:1983:2045 [2] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-042-185:1986:2041 [5] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-042-185:1988:2038 [7] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-042-185:1983:2045 [2] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-042-185:1986:2041 [5] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-042-185:1988:2038 [7] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-042-185:1983:2045 [2] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-042-185:1986:2041 [5] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-042-185:1983:2045 [2] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-042-185:1986:2041 [5] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-042-185:1986:2041 [5] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-042-185:1986:2041 [5] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-042-185:1986:2041 [5] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-042-185:1986:2041 [5] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-042-185:1981:2030 [0] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-042-185:1981:2030 [0] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-042-185:1981:2030 [0] NCCL INFO Retrieving state for FasTrak +n116-042-185:1981:2030 [0] NCCL INFO Initialized state 0 for FasTrak +n116-042-185:1981:2030 [0] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-042-185:1981:2030 [0] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-042-185:1981:2030 [0] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-042-185:1981:2030 [0] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-042-185:1981:2030 [0] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-042-185:1981:2030 [0] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-042-185:1981:2030 [0] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-042-185:1981:2030 [0] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-042-185:1981:2030 [0] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-042-185:1988:2038 [7] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-042-185:1981:2030 [0] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-042-185:1981:2030 [0] NCCL INFO CPU/0-0 (1/1/3) +n116-042-185:1981:2030 [0] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-042-185:1981:2030 [0] NCCL INFO + PCI[24.0] - GPU/0-4000 (0) +n116-042-185:1981:2030 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1981:2030 [0] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-042-185:1981:2030 [0] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-042-185:1981:2030 [0] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-042-185:1981:2030 [0] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-042-185:1981:2030 [0] NCCL INFO + PCI[24.0] - GPU/0-5000 (1) +n116-042-185:1981:2030 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1981:2030 [0] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-042-185:1981:2030 [0] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-042-185:1981:2030 [0] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-042-185:1981:2030 [0] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-042-185:1981:2030 [0] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-042-185:1981:2030 [0] NCCL INFO + PCI[24.0] - GPU/0-b000 (2) +n116-042-185:1981:2030 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1981:2030 [0] NCCL INFO + PCI[24.0] - GPU/0-c000 (3) +n116-042-185:1981:2030 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1981:2030 [0] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-042-185:1981:2030 [0] NCCL INFO CPU/0-1 (1/1/3) +n116-042-185:1981:2030 [0] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-042-185:1981:2030 [0] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-042-185:1981:2030 [0] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-042-185:1981:2030 [0] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-042-185:1981:2030 [0] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-042-185:1981:2030 [0] NCCL INFO + PCI[24.0] - GPU/0-84000 (4) +n116-042-185:1981:2030 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1981:2030 [0] NCCL INFO + PCI[24.0] - GPU/0-85000 (5) +n116-042-185:1981:2030 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1981:2030 [0] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-042-185:1981:2030 [0] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-042-185:1981:2030 [0] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-042-185:1981:2030 [0] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-042-185:1981:2030 [0] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-042-185:1981:2030 [0] NCCL INFO + PCI[24.0] - GPU/0-8b000 (6) +n116-042-185:1981:2030 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1981:2030 [0] NCCL INFO + PCI[24.0] - GPU/0-8c000 (7) +n116-042-185:1981:2030 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1981:2030 [0] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-042-185:1981:2030 [0] NCCL INFO ========================================== +n116-042-185:1981:2030 [0] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-185:1985:2033 [4] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-042-185:1981:2030 [0] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-185:1981:2030 [0] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-185:1981:2030 [0] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-185:1981:2030 [0] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-185:1981:2030 [0] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-185:1981:2030 [0] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-042-185:1981:2030 [0] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-042-185:1981:2030 [0] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-185:1981:2030 [0] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-185:1981:2030 [0] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-185:1981:2030 [0] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-185:1981:2030 [0] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-042-185:1981:2030 [0] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-042-185:1981:2030 [0] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-042-185:1981:2030 [0] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-042-185:1981:2030 [0] NCCL INFO Setting affinity for GPU 0 to 0fffffff,fffffe00,00000000,000fffff,fffffffe +n116-042-185:1981:2030 [0] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-042-185:1987:2034 [6] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-042-185:1981:2030 [0] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-042-185:1984:2037 [3] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-042-185:1986:2041 [5] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-042-185:1983:2045 [2] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-042-185:1988:2038 [7] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-042-185:1988:2038 [7] NCCL INFO CPU/0-0 (1/1/3) +n116-042-185:1988:2038 [7] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-042-185:1988:2038 [7] NCCL INFO + PCI[24.0] - GPU/0-4000 (0) +n116-042-185:1988:2038 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1988:2038 [7] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-042-185:1988:2038 [7] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-042-185:1988:2038 [7] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-042-185:1988:2038 [7] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-042-185:1988:2038 [7] NCCL INFO + PCI[24.0] - GPU/0-5000 (1) +n116-042-185:1988:2038 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1988:2038 [7] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-042-185:1988:2038 [7] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-042-185:1988:2038 [7] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-042-185:1988:2038 [7] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-042-185:1988:2038 [7] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-042-185:1988:2038 [7] NCCL INFO + PCI[24.0] - GPU/0-b000 (2) +n116-042-185:1988:2038 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1988:2038 [7] NCCL INFO + PCI[24.0] - GPU/0-c000 (3) +n116-042-185:1988:2038 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1988:2038 [7] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-042-185:1988:2038 [7] NCCL INFO CPU/0-1 (1/1/3) +n116-042-185:1988:2038 [7] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-042-185:1988:2038 [7] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-042-185:1988:2038 [7] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-042-185:1988:2038 [7] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-042-185:1988:2038 [7] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-042-185:1988:2038 [7] NCCL INFO + PCI[24.0] - GPU/0-84000 (4) +n116-042-185:1988:2038 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1988:2038 [7] NCCL INFO + PCI[24.0] - GPU/0-85000 (5) +n116-042-185:1988:2038 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1988:2038 [7] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-042-185:1988:2038 [7] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-042-185:1988:2038 [7] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-042-185:1988:2038 [7] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-042-185:1988:2038 [7] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-042-185:1988:2038 [7] NCCL INFO + PCI[24.0] - GPU/0-8b000 (6) +n116-042-185:1988:2038 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1988:2038 [7] NCCL INFO + PCI[24.0] - GPU/0-8c000 (7) +n116-042-185:1988:2038 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1988:2038 [7] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-042-185:1988:2038 [7] NCCL INFO ========================================== +n116-042-185:1988:2038 [7] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-185:1988:2038 [7] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-185:1988:2038 [7] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-185:1988:2038 [7] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-185:1988:2038 [7] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-185:1988:2038 [7] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-185:1988:2038 [7] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-042-185:1988:2038 [7] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-042-185:1988:2038 [7] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-185:1988:2038 [7] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-185:1988:2038 [7] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-185:1988:2038 [7] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-185:1988:2038 [7] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-042-185:1988:2038 [7] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-042-185:1988:2038 [7] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-042-185:1988:2038 [7] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-042-185:1988:2038 [7] NCCL INFO Setting affinity for GPU 7 to 7fff,ffffffff,e0000000,0000007f,ffffffff,ffe00000,00000000 +n116-042-185:1988:2038 [7] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-042-185:1988:2038 [7] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-042-185:1985:2033 [4] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-042-185:1985:2033 [4] NCCL INFO CPU/0-0 (1/1/3) +n116-042-185:1985:2033 [4] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-042-185:1985:2033 [4] NCCL INFO + PCI[24.0] - GPU/0-4000 (0) +n116-042-185:1985:2033 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1985:2033 [4] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-042-185:1985:2033 [4] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-042-185:1985:2033 [4] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-042-185:1985:2033 [4] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-042-185:1985:2033 [4] NCCL INFO + PCI[24.0] - GPU/0-5000 (1) +n116-042-185:1985:2033 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1985:2033 [4] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-042-185:1985:2033 [4] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-042-185:1985:2033 [4] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-042-185:1985:2033 [4] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-042-185:1985:2033 [4] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-042-185:1985:2033 [4] NCCL INFO + PCI[24.0] - GPU/0-b000 (2) +n116-042-185:1985:2033 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1985:2033 [4] NCCL INFO + PCI[24.0] - GPU/0-c000 (3) +n116-042-185:1985:2033 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1985:2033 [4] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-042-185:1985:2033 [4] NCCL INFO CPU/0-1 (1/1/3) +n116-042-185:1985:2033 [4] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-042-185:1985:2033 [4] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-042-185:1985:2033 [4] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-042-185:1985:2033 [4] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-042-185:1985:2033 [4] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-042-185:1985:2033 [4] NCCL INFO + PCI[24.0] - GPU/0-84000 (4) +n116-042-185:1985:2033 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1985:2033 [4] NCCL INFO + PCI[24.0] - GPU/0-85000 (5) +n116-042-185:1985:2033 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1985:2033 [4] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-042-185:1985:2033 [4] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-042-185:1985:2033 [4] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-042-185:1985:2033 [4] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-042-185:1985:2033 [4] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-042-185:1985:2033 [4] NCCL INFO + PCI[24.0] - GPU/0-8b000 (6) +n116-042-185:1985:2033 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1985:2033 [4] NCCL INFO + PCI[24.0] - GPU/0-8c000 (7) +n116-042-185:1985:2033 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1985:2033 [4] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-042-185:1985:2033 [4] NCCL INFO ========================================== +n116-042-185:1985:2033 [4] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-185:1985:2033 [4] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-185:1985:2033 [4] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-185:1985:2033 [4] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-185:1985:2033 [4] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-185:1985:2033 [4] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-185:1985:2033 [4] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-042-185:1985:2033 [4] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-042-185:1985:2033 [4] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-185:1985:2033 [4] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-185:1985:2033 [4] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-185:1985:2033 [4] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-185:1985:2033 [4] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-042-185:1985:2033 [4] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-042-185:1985:2033 [4] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-042-185:1985:2033 [4] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-042-185:1985:2033 [4] NCCL INFO Setting affinity for GPU 4 to 7fff,ffffffff,e0000000,0000007f,ffffffff,ffe00000,00000000 +n116-042-185:1985:2033 [4] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-042-185:1984:2037 [3] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-042-185:1984:2037 [3] NCCL INFO CPU/0-0 (1/1/3) +n116-042-185:1984:2037 [3] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-042-185:1984:2037 [3] NCCL INFO + PCI[24.0] - GPU/0-4000 (0) +n116-042-185:1984:2037 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1984:2037 [3] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-042-185:1984:2037 [3] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-042-185:1984:2037 [3] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-042-185:1984:2037 [3] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-042-185:1984:2037 [3] NCCL INFO + PCI[24.0] - GPU/0-5000 (1) +n116-042-185:1984:2037 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1984:2037 [3] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-042-185:1984:2037 [3] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-042-185:1984:2037 [3] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-042-185:1984:2037 [3] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-042-185:1984:2037 [3] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-042-185:1985:2033 [4] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-042-185:1984:2037 [3] NCCL INFO + PCI[24.0] - GPU/0-b000 (2) +n116-042-185:1984:2037 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1984:2037 [3] NCCL INFO + PCI[24.0] - GPU/0-c000 (3) +n116-042-185:1984:2037 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1984:2037 [3] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-042-185:1984:2037 [3] NCCL INFO CPU/0-1 (1/1/3) +n116-042-185:1984:2037 [3] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-042-185:1984:2037 [3] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-042-185:1984:2037 [3] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-042-185:1984:2037 [3] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-042-185:1984:2037 [3] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-042-185:1984:2037 [3] NCCL INFO + PCI[24.0] - GPU/0-84000 (4) +n116-042-185:1984:2037 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1984:2037 [3] NCCL INFO + PCI[24.0] - GPU/0-85000 (5) +n116-042-185:1984:2037 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1984:2037 [3] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-042-185:1984:2037 [3] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-042-185:1984:2037 [3] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-042-185:1984:2037 [3] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-042-185:1984:2037 [3] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-042-185:1984:2037 [3] NCCL INFO + PCI[24.0] - GPU/0-8b000 (6) +n116-042-185:1984:2037 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1984:2037 [3] NCCL INFO + PCI[24.0] - GPU/0-8c000 (7) +n116-042-185:1984:2037 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1984:2037 [3] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-042-185:1984:2037 [3] NCCL INFO ========================================== +n116-042-185:1982:2044 [1] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-042-185:1984:2037 [3] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-185:1984:2037 [3] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-185:1984:2037 [3] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-185:1984:2037 [3] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-185:1984:2037 [3] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-185:1984:2037 [3] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-185:1984:2037 [3] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-042-185:1984:2037 [3] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-042-185:1984:2037 [3] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-185:1984:2037 [3] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-185:1984:2037 [3] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-185:1984:2037 [3] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-185:1984:2037 [3] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-042-185:1984:2037 [3] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-042-185:1984:2037 [3] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-042-185:1984:2037 [3] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-042-185:1984:2037 [3] NCCL INFO Setting affinity for GPU 3 to 0fffffff,fffffe00,00000000,000fffff,fffffffe +n116-042-185:1984:2037 [3] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-042-185:1983:2045 [2] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-042-185:1983:2045 [2] NCCL INFO CPU/0-0 (1/1/3) +n116-042-185:1983:2045 [2] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-042-185:1984:2037 [3] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-042-185:1983:2045 [2] NCCL INFO + PCI[24.0] - GPU/0-4000 (0) +n116-042-185:1983:2045 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1983:2045 [2] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-042-185:1983:2045 [2] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-042-185:1983:2045 [2] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-042-185:1983:2045 [2] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-042-185:1983:2045 [2] NCCL INFO + PCI[24.0] - GPU/0-5000 (1) +n116-042-185:1983:2045 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1983:2045 [2] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-042-185:1983:2045 [2] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-042-185:1983:2045 [2] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-042-185:1983:2045 [2] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-042-185:1983:2045 [2] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-042-185:1983:2045 [2] NCCL INFO + PCI[24.0] - GPU/0-b000 (2) +n116-042-185:1983:2045 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1983:2045 [2] NCCL INFO + PCI[24.0] - GPU/0-c000 (3) +n116-042-185:1983:2045 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1983:2045 [2] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-042-185:1983:2045 [2] NCCL INFO CPU/0-1 (1/1/3) +n116-042-185:1983:2045 [2] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-042-185:1983:2045 [2] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-042-185:1983:2045 [2] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-042-185:1983:2045 [2] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-042-185:1983:2045 [2] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-042-185:1983:2045 [2] NCCL INFO + PCI[24.0] - GPU/0-84000 (4) +n116-042-185:1983:2045 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1983:2045 [2] NCCL INFO + PCI[24.0] - GPU/0-85000 (5) +n116-042-185:1983:2045 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1983:2045 [2] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-042-185:1983:2045 [2] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-042-185:1983:2045 [2] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-042-185:1983:2045 [2] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-042-185:1983:2045 [2] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-042-185:1983:2045 [2] NCCL INFO + PCI[24.0] - GPU/0-8b000 (6) +n116-042-185:1983:2045 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1983:2045 [2] NCCL INFO + PCI[24.0] - GPU/0-8c000 (7) +n116-042-185:1983:2045 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1983:2045 [2] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-042-185:1983:2045 [2] NCCL INFO ========================================== +n116-042-185:1983:2045 [2] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-185:1983:2045 [2] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-185:1983:2045 [2] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-185:1983:2045 [2] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-185:1983:2045 [2] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-185:1983:2045 [2] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-185:1983:2045 [2] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-042-185:1983:2045 [2] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-042-185:1983:2045 [2] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-185:1983:2045 [2] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-185:1983:2045 [2] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-185:1983:2045 [2] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-185:1983:2045 [2] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-042-185:1983:2045 [2] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-042-185:1983:2045 [2] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-042-185:1983:2045 [2] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-042-185:1983:2045 [2] NCCL INFO Setting affinity for GPU 2 to 0fffffff,fffffe00,00000000,000fffff,fffffffe +n116-042-185:1986:2041 [5] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-042-185:1986:2041 [5] NCCL INFO CPU/0-0 (1/1/3) +n116-042-185:1986:2041 [5] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-042-185:1986:2041 [5] NCCL INFO + PCI[24.0] - GPU/0-4000 (0) +n116-042-185:1986:2041 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1986:2041 [5] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-042-185:1986:2041 [5] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-042-185:1986:2041 [5] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-042-185:1986:2041 [5] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-042-185:1986:2041 [5] NCCL INFO + PCI[24.0] - GPU/0-5000 (1) +n116-042-185:1986:2041 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1986:2041 [5] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-042-185:1986:2041 [5] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-042-185:1986:2041 [5] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-042-185:1986:2041 [5] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-042-185:1986:2041 [5] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-042-185:1986:2041 [5] NCCL INFO + PCI[24.0] - GPU/0-b000 (2) +n116-042-185:1983:2045 [2] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-042-185:1986:2041 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1986:2041 [5] NCCL INFO + PCI[24.0] - GPU/0-c000 (3) +n116-042-185:1986:2041 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1986:2041 [5] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-042-185:1986:2041 [5] NCCL INFO CPU/0-1 (1/1/3) +n116-042-185:1986:2041 [5] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-042-185:1986:2041 [5] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-042-185:1986:2041 [5] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-042-185:1987:2034 [6] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-042-185:1986:2041 [5] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-042-185:1986:2041 [5] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-042-185:1987:2034 [6] NCCL INFO CPU/0-0 (1/1/3) +n116-042-185:1987:2034 [6] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-042-185:1986:2041 [5] NCCL INFO + PCI[24.0] - GPU/0-84000 (4) +n116-042-185:1987:2034 [6] NCCL INFO + PCI[24.0] - GPU/0-4000 (0) +n116-042-185:1986:2041 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1987:2034 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1986:2041 [5] NCCL INFO + PCI[24.0] - GPU/0-85000 (5) +n116-042-185:1987:2034 [6] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-042-185:1986:2041 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1986:2041 [5] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-042-185:1987:2034 [6] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-042-185:1987:2034 [6] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-042-185:1986:2041 [5] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-042-185:1987:2034 [6] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-042-185:1987:2034 [6] NCCL INFO + PCI[24.0] - GPU/0-5000 (1) +n116-042-185:1986:2041 [5] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-042-185:1986:2041 [5] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-042-185:1987:2034 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1986:2041 [5] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-042-185:1987:2034 [6] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-042-185:1986:2041 [5] NCCL INFO + PCI[24.0] - GPU/0-8b000 (6) +n116-042-185:1987:2034 [6] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-042-185:1986:2041 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1987:2034 [6] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-042-185:1986:2041 [5] NCCL INFO + PCI[24.0] - GPU/0-8c000 (7) +n116-042-185:1983:2045 [2] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-042-185:1987:2034 [6] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-042-185:1986:2041 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1987:2034 [6] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-042-185:1987:2034 [6] NCCL INFO + PCI[24.0] - GPU/0-b000 (2) +n116-042-185:1986:2041 [5] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-042-185:1986:2041 [5] NCCL INFO ========================================== +n116-042-185:1987:2034 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1987:2034 [6] NCCL INFO + PCI[24.0] - GPU/0-c000 (3) +n116-042-185:1987:2034 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1987:2034 [6] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-042-185:1987:2034 [6] NCCL INFO CPU/0-1 (1/1/3) +n116-042-185:1987:2034 [6] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-042-185:1987:2034 [6] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-042-185:1987:2034 [6] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-042-185:1986:2041 [5] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-185:1987:2034 [6] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-042-185:1987:2034 [6] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-042-185:1987:2034 [6] NCCL INFO + PCI[24.0] - GPU/0-84000 (4) +n116-042-185:1987:2034 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1987:2034 [6] NCCL INFO + PCI[24.0] - GPU/0-85000 (5) +n116-042-185:1987:2034 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1986:2041 [5] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-185:1987:2034 [6] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-042-185:1987:2034 [6] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-042-185:1987:2034 [6] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-042-185:1987:2034 [6] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-042-185:1987:2034 [6] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-042-185:1986:2041 [5] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-185:1987:2034 [6] NCCL INFO + PCI[24.0] - GPU/0-8b000 (6) +n116-042-185:1987:2034 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1987:2034 [6] NCCL INFO + PCI[24.0] - GPU/0-8c000 (7) +n116-042-185:1987:2034 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1987:2034 [6] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-042-185:1987:2034 [6] NCCL INFO ========================================== +n116-042-185:1986:2041 [5] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-185:1987:2034 [6] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-185:1986:2041 [5] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-185:1986:2041 [5] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-185:1987:2034 [6] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-185:1986:2041 [5] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-042-185:1987:2034 [6] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-185:1987:2034 [6] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-185:1986:2041 [5] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-042-185:1986:2041 [5] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-185:1987:2034 [6] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-185:1987:2034 [6] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-185:1986:2041 [5] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-185:1986:2041 [5] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-185:1987:2034 [6] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-042-185:1986:2041 [5] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-185:1987:2034 [6] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-042-185:1987:2034 [6] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-185:1986:2041 [5] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-042-185:1986:2041 [5] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-042-185:1987:2034 [6] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-185:1986:2041 [5] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-042-185:1987:2034 [6] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-185:1987:2034 [6] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-185:1986:2041 [5] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-042-185:1986:2041 [5] NCCL INFO Setting affinity for GPU 5 to 7fff,ffffffff,e0000000,0000007f,ffffffff,ffe00000,00000000 +n116-042-185:1987:2034 [6] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-042-185:1986:2041 [5] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-042-185:1987:2034 [6] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-042-185:1987:2034 [6] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-042-185:1987:2034 [6] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-042-185:1986:2041 [5] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-042-185:1987:2034 [6] NCCL INFO Setting affinity for GPU 6 to 7fff,ffffffff,e0000000,0000007f,ffffffff,ffe00000,00000000 +n116-042-185:1987:2034 [6] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-042-185:1987:2034 [6] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-042-185:1982:2044 [1] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-042-185:1982:2044 [1] NCCL INFO CPU/0-0 (1/1/3) +n116-042-185:1982:2044 [1] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-042-185:1982:2044 [1] NCCL INFO + PCI[24.0] - GPU/0-4000 (0) +n116-042-185:1982:2044 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1982:2044 [1] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-042-185:1982:2044 [1] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-042-185:1982:2044 [1] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-042-185:1982:2044 [1] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-042-185:1982:2044 [1] NCCL INFO + PCI[24.0] - GPU/0-5000 (1) +n116-042-185:1982:2044 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1982:2044 [1] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-042-185:1982:2044 [1] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-042-185:1982:2044 [1] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-042-185:1982:2044 [1] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-042-185:1982:2044 [1] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-042-185:1982:2044 [1] NCCL INFO + PCI[24.0] - GPU/0-b000 (2) +n116-042-185:1982:2044 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1982:2044 [1] NCCL INFO + PCI[24.0] - GPU/0-c000 (3) +n116-042-185:1982:2044 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1982:2044 [1] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-042-185:1982:2044 [1] NCCL INFO CPU/0-1 (1/1/3) +n116-042-185:1982:2044 [1] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-042-185:1982:2044 [1] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-042-185:1982:2044 [1] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-042-185:1982:2044 [1] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-042-185:1982:2044 [1] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-042-185:1982:2044 [1] NCCL INFO + PCI[24.0] - GPU/0-84000 (4) +n116-042-185:1982:2044 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1982:2044 [1] NCCL INFO + PCI[24.0] - GPU/0-85000 (5) +n116-042-185:1982:2044 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1982:2044 [1] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-042-185:1982:2044 [1] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-042-185:1982:2044 [1] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-042-185:1982:2044 [1] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-042-185:1982:2044 [1] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-042-185:1982:2044 [1] NCCL INFO + PCI[24.0] - GPU/0-8b000 (6) +n116-042-185:1982:2044 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1982:2044 [1] NCCL INFO + PCI[24.0] - GPU/0-8c000 (7) +n116-042-185:1982:2044 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-185:1982:2044 [1] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-042-185:1982:2044 [1] NCCL INFO ========================================== +n116-042-185:1982:2044 [1] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-185:1982:2044 [1] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-185:1982:2044 [1] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-185:1982:2044 [1] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-185:1982:2044 [1] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-185:1982:2044 [1] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-185:1982:2044 [1] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-042-185:1982:2044 [1] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-042-185:1982:2044 [1] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-185:1982:2044 [1] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-185:1982:2044 [1] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-185:1982:2044 [1] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-185:1982:2044 [1] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-042-185:1982:2044 [1] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-042-185:1982:2044 [1] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-042-185:1982:2044 [1] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-042-185:1982:2044 [1] NCCL INFO Setting affinity for GPU 1 to 0fffffff,fffffe00,00000000,000fffff,fffffffe +n116-042-185:1982:2044 [1] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-042-185:1982:2044 [1] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-042-185:1981:2030 [0] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-042-185:1981:2030 [0] NCCL INFO 0 : NET/0-0 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 NET/0-0 +n116-042-185:1981:2030 [0] NCCL INFO 1 : NET/0-1 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 NET/0-1 +n116-042-185:1981:2030 [0] NCCL INFO 2 : NET/0-2 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 NET/0-2 +n116-042-185:1981:2030 [0] NCCL INFO 3 : NET/0-3 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 NET/0-3 +n116-042-185:1981:2030 [0] NCCL INFO 4 : NET/0-4 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 NET/0-4 +n116-042-185:1981:2030 [0] NCCL INFO 5 : NET/0-5 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 NET/0-5 +n116-042-185:1981:2030 [0] NCCL INFO 6 : NET/0-6 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 NET/0-6 +n116-042-185:1981:2030 [0] NCCL INFO 7 : NET/0-7 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 NET/0-7 +n116-042-185:1981:2030 [0] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-042-185:1981:2030 [0] NCCL INFO 0 : NET/0-0 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 NET/0-0 +n116-042-185:1981:2030 [0] NCCL INFO 1 : NET/0-1 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 NET/0-1 +n116-042-185:1981:2030 [0] NCCL INFO 2 : NET/0-2 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 NET/0-2 +n116-042-185:1981:2030 [0] NCCL INFO 3 : NET/0-3 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 NET/0-3 +n116-042-185:1981:2030 [0] NCCL INFO 4 : NET/0-4 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 NET/0-4 +n116-042-185:1981:2030 [0] NCCL INFO 5 : NET/0-5 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 NET/0-5 +n116-042-185:1981:2030 [0] NCCL INFO 6 : NET/0-6 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 NET/0-6 +n116-042-185:1981:2030 [0] NCCL INFO 7 : NET/0-7 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 NET/0-7 +n116-042-185:1985:2033 [4] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-042-185:1985:2033 [4] NCCL INFO 0 : NET/0-0 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 NET/0-0 +n116-042-185:1985:2033 [4] NCCL INFO 1 : NET/0-1 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 NET/0-1 +n116-042-185:1985:2033 [4] NCCL INFO 2 : NET/0-2 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 NET/0-2 +n116-042-185:1985:2033 [4] NCCL INFO 3 : NET/0-3 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 NET/0-3 +n116-042-185:1985:2033 [4] NCCL INFO 4 : NET/0-4 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 NET/0-4 +n116-042-185:1985:2033 [4] NCCL INFO 5 : NET/0-5 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 NET/0-5 +n116-042-185:1985:2033 [4] NCCL INFO 6 : NET/0-6 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 NET/0-6 +n116-042-185:1985:2033 [4] NCCL INFO 7 : NET/0-7 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 NET/0-7 +n116-042-185:1985:2033 [4] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-042-185:1985:2033 [4] NCCL INFO 0 : NET/0-0 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 NET/0-0 +n116-042-185:1985:2033 [4] NCCL INFO 1 : NET/0-1 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 NET/0-1 +n116-042-185:1985:2033 [4] NCCL INFO 2 : NET/0-2 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 NET/0-2 +n116-042-185:1985:2033 [4] NCCL INFO 3 : NET/0-3 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 NET/0-3 +n116-042-185:1985:2033 [4] NCCL INFO 4 : NET/0-4 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 NET/0-4 +n116-042-185:1985:2033 [4] NCCL INFO 5 : NET/0-5 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 NET/0-5 +n116-042-185:1985:2033 [4] NCCL INFO 6 : NET/0-6 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 NET/0-6 +n116-042-185:1985:2033 [4] NCCL INFO 7 : NET/0-7 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 NET/0-7 +n116-042-185:1988:2038 [7] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-042-185:1988:2038 [7] NCCL INFO 0 : NET/0-0 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 NET/0-0 +n116-042-185:1988:2038 [7] NCCL INFO 1 : NET/0-1 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 NET/0-1 +n116-042-185:1988:2038 [7] NCCL INFO 2 : NET/0-2 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 NET/0-2 +n116-042-185:1988:2038 [7] NCCL INFO 3 : NET/0-3 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 NET/0-3 +n116-042-185:1988:2038 [7] NCCL INFO 4 : NET/0-4 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 NET/0-4 +n116-042-185:1988:2038 [7] NCCL INFO 5 : NET/0-5 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 NET/0-5 +n116-042-185:1988:2038 [7] NCCL INFO 6 : NET/0-6 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 NET/0-6 +n116-042-185:1988:2038 [7] NCCL INFO 7 : NET/0-7 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 NET/0-7 +n116-042-185:1988:2038 [7] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-042-185:1988:2038 [7] NCCL INFO 0 : NET/0-0 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 NET/0-0 +n116-042-185:1988:2038 [7] NCCL INFO 1 : NET/0-1 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 NET/0-1 +n116-042-185:1988:2038 [7] NCCL INFO 2 : NET/0-2 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 NET/0-2 +n116-042-185:1988:2038 [7] NCCL INFO 3 : NET/0-3 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 NET/0-3 +n116-042-185:1988:2038 [7] NCCL INFO 4 : NET/0-4 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 NET/0-4 +n116-042-185:1988:2038 [7] NCCL INFO 5 : NET/0-5 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 NET/0-5 +n116-042-185:1988:2038 [7] NCCL INFO 6 : NET/0-6 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 NET/0-6 +n116-042-185:1988:2038 [7] NCCL INFO 7 : NET/0-7 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 NET/0-7 +n116-042-185:1982:2044 [1] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-042-185:1982:2044 [1] NCCL INFO 0 : NET/0-0 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 NET/0-0 +n116-042-185:1982:2044 [1] NCCL INFO 1 : NET/0-1 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 NET/0-1 +n116-042-185:1982:2044 [1] NCCL INFO 2 : NET/0-2 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 NET/0-2 +n116-042-185:1982:2044 [1] NCCL INFO 3 : NET/0-3 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 NET/0-3 +n116-042-185:1982:2044 [1] NCCL INFO 4 : NET/0-4 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 NET/0-4 +n116-042-185:1982:2044 [1] NCCL INFO 5 : NET/0-5 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 NET/0-5 +n116-042-185:1982:2044 [1] NCCL INFO 6 : NET/0-6 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 NET/0-6 +n116-042-185:1982:2044 [1] NCCL INFO 7 : NET/0-7 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 NET/0-7 +n116-042-185:1982:2044 [1] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-042-185:1982:2044 [1] NCCL INFO 0 : NET/0-0 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 NET/0-0 +n116-042-185:1982:2044 [1] NCCL INFO 1 : NET/0-1 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 NET/0-1 +n116-042-185:1982:2044 [1] NCCL INFO 2 : NET/0-2 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 NET/0-2 +n116-042-185:1982:2044 [1] NCCL INFO 3 : NET/0-3 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 NET/0-3 +n116-042-185:1982:2044 [1] NCCL INFO 4 : NET/0-4 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 NET/0-4 +n116-042-185:1982:2044 [1] NCCL INFO 5 : NET/0-5 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 NET/0-5 +n116-042-185:1982:2044 [1] NCCL INFO 6 : NET/0-6 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 NET/0-6 +n116-042-185:1982:2044 [1] NCCL INFO 7 : NET/0-7 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 NET/0-7 +n116-042-185:1984:2037 [3] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-042-185:1984:2037 [3] NCCL INFO 0 : NET/0-0 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 NET/0-0 +n116-042-185:1984:2037 [3] NCCL INFO 1 : NET/0-1 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 NET/0-1 +n116-042-185:1984:2037 [3] NCCL INFO 2 : NET/0-2 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 NET/0-2 +n116-042-185:1984:2037 [3] NCCL INFO 3 : NET/0-3 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 NET/0-3 +n116-042-185:1984:2037 [3] NCCL INFO 4 : NET/0-4 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 NET/0-4 +n116-042-185:1984:2037 [3] NCCL INFO 5 : NET/0-5 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 NET/0-5 +n116-042-185:1984:2037 [3] NCCL INFO 6 : NET/0-6 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 NET/0-6 +n116-042-185:1984:2037 [3] NCCL INFO 7 : NET/0-7 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 NET/0-7 +n116-042-185:1984:2037 [3] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-042-185:1984:2037 [3] NCCL INFO 0 : NET/0-0 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 NET/0-0 +n116-042-185:1984:2037 [3] NCCL INFO 1 : NET/0-1 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 NET/0-1 +n116-042-185:1984:2037 [3] NCCL INFO 2 : NET/0-2 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 NET/0-2 +n116-042-185:1984:2037 [3] NCCL INFO 3 : NET/0-3 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 NET/0-3 +n116-042-185:1984:2037 [3] NCCL INFO 4 : NET/0-4 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 NET/0-4 +n116-042-185:1984:2037 [3] NCCL INFO 5 : NET/0-5 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 NET/0-5 +n116-042-185:1984:2037 [3] NCCL INFO 6 : NET/0-6 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 NET/0-6 +n116-042-185:1984:2037 [3] NCCL INFO 7 : NET/0-7 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 NET/0-7 +n116-042-185:1983:2045 [2] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-042-185:1983:2045 [2] NCCL INFO 0 : NET/0-0 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 NET/0-0 +n116-042-185:1983:2045 [2] NCCL INFO 1 : NET/0-1 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 NET/0-1 +n116-042-185:1983:2045 [2] NCCL INFO 2 : NET/0-2 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 NET/0-2 +n116-042-185:1983:2045 [2] NCCL INFO 3 : NET/0-3 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 NET/0-3 +n116-042-185:1983:2045 [2] NCCL INFO 4 : NET/0-4 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 NET/0-4 +n116-042-185:1983:2045 [2] NCCL INFO 5 : NET/0-5 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 NET/0-5 +n116-042-185:1983:2045 [2] NCCL INFO 6 : NET/0-6 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 NET/0-6 +n116-042-185:1983:2045 [2] NCCL INFO 7 : NET/0-7 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 NET/0-7 +n116-042-185:1983:2045 [2] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-042-185:1983:2045 [2] NCCL INFO 0 : NET/0-0 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 NET/0-0 +n116-042-185:1983:2045 [2] NCCL INFO 1 : NET/0-1 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 NET/0-1 +n116-042-185:1983:2045 [2] NCCL INFO 2 : NET/0-2 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 NET/0-2 +n116-042-185:1983:2045 [2] NCCL INFO 3 : NET/0-3 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 NET/0-3 +n116-042-185:1983:2045 [2] NCCL INFO 4 : NET/0-4 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 NET/0-4 +n116-042-185:1983:2045 [2] NCCL INFO 5 : NET/0-5 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 NET/0-5 +n116-042-185:1983:2045 [2] NCCL INFO 6 : NET/0-6 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 NET/0-6 +n116-042-185:1983:2045 [2] NCCL INFO 7 : NET/0-7 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 NET/0-7 +n116-042-185:1987:2034 [6] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-042-185:1987:2034 [6] NCCL INFO 0 : NET/0-0 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 NET/0-0 +n116-042-185:1987:2034 [6] NCCL INFO 1 : NET/0-1 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 NET/0-1 +n116-042-185:1987:2034 [6] NCCL INFO 2 : NET/0-2 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 NET/0-2 +n116-042-185:1987:2034 [6] NCCL INFO 3 : NET/0-3 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 NET/0-3 +n116-042-185:1987:2034 [6] NCCL INFO 4 : NET/0-4 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 NET/0-4 +n116-042-185:1987:2034 [6] NCCL INFO 5 : NET/0-5 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 NET/0-5 +n116-042-185:1987:2034 [6] NCCL INFO 6 : NET/0-6 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 NET/0-6 +n116-042-185:1987:2034 [6] NCCL INFO 7 : NET/0-7 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 NET/0-7 +n116-042-185:1987:2034 [6] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-042-185:1987:2034 [6] NCCL INFO 0 : NET/0-0 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 NET/0-0 +n116-042-185:1987:2034 [6] NCCL INFO 1 : NET/0-1 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 NET/0-1 +n116-042-185:1987:2034 [6] NCCL INFO 2 : NET/0-2 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 NET/0-2 +n116-042-185:1987:2034 [6] NCCL INFO 3 : NET/0-3 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 NET/0-3 +n116-042-185:1987:2034 [6] NCCL INFO 4 : NET/0-4 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 NET/0-4 +n116-042-185:1987:2034 [6] NCCL INFO 5 : NET/0-5 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 NET/0-5 +n116-042-185:1987:2034 [6] NCCL INFO 6 : NET/0-6 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 NET/0-6 +n116-042-185:1987:2034 [6] NCCL INFO 7 : NET/0-7 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 NET/0-7 +n116-042-185:1986:2041 [5] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-042-185:1986:2041 [5] NCCL INFO 0 : NET/0-0 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 NET/0-0 +n116-042-185:1986:2041 [5] NCCL INFO 1 : NET/0-1 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 NET/0-1 +n116-042-185:1986:2041 [5] NCCL INFO 2 : NET/0-2 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 NET/0-2 +n116-042-185:1986:2041 [5] NCCL INFO 3 : NET/0-3 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 NET/0-3 +n116-042-185:1986:2041 [5] NCCL INFO 4 : NET/0-4 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 NET/0-4 +n116-042-185:1986:2041 [5] NCCL INFO 5 : NET/0-5 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 NET/0-5 +n116-042-185:1986:2041 [5] NCCL INFO 6 : NET/0-6 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 NET/0-6 +n116-042-185:1986:2041 [5] NCCL INFO 7 : NET/0-7 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 NET/0-7 +n116-042-185:1986:2041 [5] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-042-185:1986:2041 [5] NCCL INFO 0 : NET/0-0 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 NET/0-0 +n116-042-185:1986:2041 [5] NCCL INFO 1 : NET/0-1 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 NET/0-1 +n116-042-185:1986:2041 [5] NCCL INFO 2 : NET/0-2 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 NET/0-2 +n116-042-185:1986:2041 [5] NCCL INFO 3 : NET/0-3 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 NET/0-3 +n116-042-185:1986:2041 [5] NCCL INFO 4 : NET/0-4 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 NET/0-4 +n116-042-185:1986:2041 [5] NCCL INFO 5 : NET/0-5 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 NET/0-5 +n116-042-185:1986:2041 [5] NCCL INFO 6 : NET/0-6 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 NET/0-6 +n116-042-185:1986:2041 [5] NCCL INFO 7 : NET/0-7 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 NET/0-7 +n116-042-185:1988:2038 [7] NCCL INFO comm 0xb3ad0e0 rank 7 nRanks 32 nNodes 4 localRanks 8 localRank 7 MNNVL 0 +n116-042-185:1987:2034 [6] NCCL INFO comm 0xc4fc250 rank 6 nRanks 32 nNodes 4 localRanks 8 localRank 6 MNNVL 0 +n116-042-185:1988:2038 [7] NCCL INFO Tree 6 : 6 -> 7 -> 0/-1/-1 +n116-042-185:1988:2038 [7] NCCL INFO Tree 14 : 6 -> 7 -> 0/-1/-1 +n116-042-185:1986:2041 [5] NCCL INFO comm 0xbf2f3e0 rank 5 nRanks 32 nNodes 4 localRanks 8 localRank 5 MNNVL 0 +n116-042-185:1988:2038 [7] NCCL INFO Tree 7 : -1 -> 7 -> 6/23/-1 +n116-042-185:1988:2038 [7] NCCL INFO Tree 15 : 15 -> 7 -> 6/-1/-1 +n116-042-185:1988:2038 [7] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-042-185:1984:2037 [3] NCCL INFO comm 0xd0746a0 rank 3 nRanks 32 nNodes 4 localRanks 8 localRank 3 MNNVL 0 +n116-042-185:1985:2033 [4] NCCL INFO comm 0xc15b3b0 rank 4 nRanks 32 nNodes 4 localRanks 8 localRank 4 MNNVL 0 +n116-042-185:1983:2045 [2] NCCL INFO comm 0xc51c1b0 rank 2 nRanks 32 nNodes 4 localRanks 8 localRank 2 MNNVL 0 +n116-042-185:1986:2041 [5] NCCL INFO Tree 4 : 4 -> 5 -> 6/-1/-1 +n116-042-185:1987:2034 [6] NCCL INFO Tree 6 : -1 -> 6 -> 7/22/-1 +n116-042-185:1987:2034 [6] NCCL INFO Tree 14 : 14 -> 6 -> 7/-1/-1 +n116-042-185:1986:2041 [5] NCCL INFO Tree 12 : 4 -> 5 -> 6/-1/-1 +n116-042-185:1988:2038 [7] NCCL INFO Ring 00 : 0 -> 7 -> 6 +n116-042-185:1987:2034 [6] NCCL INFO Tree 7 : 7 -> 6 -> 5/-1/-1 +n116-042-185:1986:2041 [5] NCCL INFO Tree 5 : -1 -> 5 -> 4/21/-1 +n116-042-185:1988:2038 [7] NCCL INFO Ring 01 : 6 -> 7 -> 0 +n116-042-185:1987:2034 [6] NCCL INFO Tree 15 : 7 -> 6 -> 5/-1/-1 +n116-042-185:1985:2033 [4] NCCL INFO Tree 4 : -1 -> 4 -> 5/20/-1 +n116-042-185:1986:2041 [5] NCCL INFO Tree 13 : 13 -> 5 -> 4/-1/-1 +n116-042-185:1988:2038 [7] NCCL INFO Ring 02 : 0 -> 7 -> 6 +n116-042-185:1985:2033 [4] NCCL INFO Tree 12 : 12 -> 4 -> 5/-1/-1 +n116-042-185:1988:2038 [7] NCCL INFO Ring 03 : 6 -> 7 -> 0 +n116-042-185:1984:2037 [3] NCCL INFO Tree 2 : 2 -> 3 -> 4/-1/-1 +n116-042-185:1983:2045 [2] NCCL INFO Tree 2 : -1 -> 2 -> 3/18/-1 +n116-042-185:1987:2034 [6] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-042-185:1985:2033 [4] NCCL INFO Tree 5 : 5 -> 4 -> 3/-1/-1 +n116-042-185:1988:2038 [7] NCCL INFO Ring 04 : 0 -> 7 -> 6 +n116-042-185:1986:2041 [5] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-042-185:1984:2037 [3] NCCL INFO Tree 10 : 2 -> 3 -> 4/-1/-1 +n116-042-185:1985:2033 [4] NCCL INFO Tree 13 : 5 -> 4 -> 3/-1/-1 +n116-042-185:1983:2045 [2] NCCL INFO Tree 10 : 10 -> 2 -> 3/-1/-1 +n116-042-185:1988:2038 [7] NCCL INFO Ring 05 : 6 -> 7 -> 0 +n116-042-185:1982:2044 [1] NCCL INFO comm 0xcdda240 rank 1 nRanks 32 nNodes 4 localRanks 8 localRank 1 MNNVL 0 +n116-042-185:1984:2037 [3] NCCL INFO Tree 3 : -1 -> 3 -> 2/19/-1 +n116-042-185:1988:2038 [7] NCCL INFO Ring 06 : 0 -> 7 -> 14 +n116-042-185:1983:2045 [2] NCCL INFO Tree 3 : 3 -> 2 -> 1/-1/-1 +n116-042-185:1984:2037 [3] NCCL INFO Tree 11 : 11 -> 3 -> 2/-1/-1 +n116-042-185:1985:2033 [4] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-042-185:1983:2045 [2] NCCL INFO Tree 11 : 3 -> 2 -> 1/-1/-1 +n116-042-185:1988:2038 [7] NCCL INFO Ring 07 : 30 -> 7 -> 0 +n116-042-185:1988:2038 [7] NCCL INFO Ring 08 : 0 -> 7 -> 6 +n116-042-185:1987:2034 [6] NCCL INFO Ring 00 : 7 -> 6 -> 5 +n116-042-185:1988:2038 [7] NCCL INFO Ring 09 : 6 -> 7 -> 0 +n116-042-185:1983:2045 [2] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-042-185:1986:2041 [5] NCCL INFO Ring 00 : 6 -> 5 -> 4 +n116-042-185:1984:2037 [3] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-042-185:1988:2038 [7] NCCL INFO Ring 10 : 0 -> 7 -> 6 +n116-042-185:1981:2030 [0] NCCL INFO comm 0xc1bf960 rank 0 nRanks 32 nNodes 4 localRanks 8 localRank 0 MNNVL 0 +n116-042-185:1987:2034 [6] NCCL INFO Ring 01 : 5 -> 6 -> 7 +n116-042-185:1986:2041 [5] NCCL INFO Ring 01 : 4 -> 5 -> 6 +n116-042-185:1988:2038 [7] NCCL INFO Ring 11 : 6 -> 7 -> 0 +n116-042-185:1985:2033 [4] NCCL INFO Ring 00 : 5 -> 4 -> 3 +n116-042-185:1987:2034 [6] NCCL INFO Ring 02 : 7 -> 6 -> 5 +n116-042-185:1988:2038 [7] NCCL INFO Ring 12 : 0 -> 7 -> 6 +n116-042-185:1986:2041 [5] NCCL INFO Ring 02 : 6 -> 5 -> 4 +n116-042-185:1985:2033 [4] NCCL INFO Ring 01 : 3 -> 4 -> 5 +n116-042-185:1988:2038 [7] NCCL INFO Ring 13 : 6 -> 7 -> 0 +n116-042-185:1987:2034 [6] NCCL INFO Ring 03 : 5 -> 6 -> 7 +n116-042-185:1985:2033 [4] NCCL INFO Ring 02 : 5 -> 4 -> 3 +n116-042-185:1986:2041 [5] NCCL INFO Ring 03 : 4 -> 5 -> 6 +n116-042-185:1988:2038 [7] NCCL INFO Ring 14 : 0 -> 7 -> 14 +n116-042-185:1987:2034 [6] NCCL INFO Ring 04 : 7 -> 6 -> 5 +n116-042-185:1985:2033 [4] NCCL INFO Ring 03 : 3 -> 4 -> 5 +n116-042-185:1988:2038 [7] NCCL INFO Ring 15 : 30 -> 7 -> 0 +n116-042-185:1986:2041 [5] NCCL INFO Ring 04 : 6 -> 5 -> 12 +n116-042-185:1985:2033 [4] NCCL INFO Ring 04 : 29 -> 4 -> 3 +n116-042-185:1988:2038 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] 6/-1/-1->7->0 [2] 0/-1/-1->7->6 [3] 6/-1/-1->7->0 [4] 0/-1/-1->7->6 [5] 6/-1/-1->7->0 [6] 0/-1/-1->7->6 [7] 6/23/-1->7->-1 [8] -1/-1/-1->7->6 [9] 6/-1/-1->7->0 [10] 0/-1/-1->7->6 [11] 6/-1/-1->7->0 [12] 0/-1/-1->7->6 [13] 6/-1/-1->7->0 [14] 0/-1/-1->7->6 [15] 6/-1/-1->7->15 +n116-042-185:1982:2044 [1] NCCL INFO Tree 0 : 0 -> 1 -> 2/-1/-1 +n116-042-185:1984:2037 [3] NCCL INFO Ring 00 : 4 -> 3 -> 2 +n116-042-185:1987:2034 [6] NCCL INFO Ring 05 : 5 -> 6 -> 7 +n116-042-185:1982:2044 [1] NCCL INFO Tree 8 : 0 -> 1 -> 2/-1/-1 +n116-042-185:1983:2045 [2] NCCL INFO Ring 00 : 3 -> 2 -> 1 +n116-042-185:1986:2041 [5] NCCL INFO Ring 05 : 28 -> 5 -> 6 +n116-042-185:1985:2033 [4] NCCL INFO Ring 05 : 3 -> 4 -> 13 +n116-042-185:1985:2033 [4] NCCL INFO Ring 06 : 5 -> 4 -> 3 +n116-042-185:1986:2041 [5] NCCL INFO Ring 06 : 6 -> 5 -> 4 +n116-042-185:1981:2030 [0] NCCL INFO Tree 0 : -1 -> 0 -> 1/16/-1 +n116-042-185:1988:2038 [7] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-042-185:1984:2037 [3] NCCL INFO Ring 01 : 2 -> 3 -> 4 +n116-042-185:1987:2034 [6] NCCL INFO Ring 06 : 31 -> 6 -> 5 +n116-042-185:1982:2044 [1] NCCL INFO Tree 1 : -1 -> 1 -> 0/17/-1 +n116-042-185:1983:2045 [2] NCCL INFO Ring 01 : 1 -> 2 -> 3 +n116-042-185:1985:2033 [4] NCCL INFO Ring 07 : 3 -> 4 -> 5 +n116-042-185:1981:2030 [0] NCCL INFO Tree 8 : 8 -> 0 -> 1/-1/-1 +n116-042-185:1986:2041 [5] NCCL INFO Ring 07 : 4 -> 5 -> 6 +n116-042-185:1985:2033 [4] NCCL INFO Ring 08 : 5 -> 4 -> 3 +n116-042-185:1988:2038 [7] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-042-185:1984:2037 [3] NCCL INFO Ring 02 : 4 -> 3 -> 10 +n116-042-185:1987:2034 [6] NCCL INFO Ring 07 : 5 -> 6 -> 15 +n116-042-185:1987:2034 [6] NCCL INFO Ring 08 : 7 -> 6 -> 5 +n116-042-185:1982:2044 [1] NCCL INFO Tree 9 : 9 -> 1 -> 0/-1/-1 +n116-042-185:1983:2045 [2] NCCL INFO Ring 02 : 27 -> 2 -> 1 +n116-042-185:1981:2030 [0] NCCL INFO Tree 1 : 1 -> 0 -> 7/-1/-1 +n116-042-185:1985:2033 [4] NCCL INFO Ring 09 : 3 -> 4 -> 5 +n116-042-185:1988:2038 [7] NCCL INFO P2P Chunksize set to 524288 +n116-042-185:1986:2041 [5] NCCL INFO Ring 08 : 6 -> 5 -> 4 +n116-042-185:1985:2033 [4] NCCL INFO Ring 10 : 5 -> 4 -> 3 +n116-042-185:1984:2037 [3] NCCL INFO Ring 03 : 26 -> 3 -> 4 +n116-042-185:1987:2034 [6] NCCL INFO Ring 09 : 5 -> 6 -> 7 +n116-042-185:1986:2041 [5] NCCL INFO Ring 09 : 4 -> 5 -> 6 +n116-042-185:1984:2037 [3] NCCL INFO Ring 04 : 4 -> 3 -> 2 +n116-042-185:1987:2034 [6] NCCL INFO Ring 10 : 7 -> 6 -> 5 +n116-042-185:1983:2045 [2] NCCL INFO Ring 03 : 1 -> 2 -> 11 +n116-042-185:1981:2030 [0] NCCL INFO Tree 9 : 1 -> 0 -> 7/-1/-1 +n116-042-185:1985:2033 [4] NCCL INFO Ring 11 : 3 -> 4 -> 5 +n116-042-185:1986:2041 [5] NCCL INFO Ring 10 : 6 -> 5 -> 4 +n116-042-185:1984:2037 [3] NCCL INFO Ring 05 : 2 -> 3 -> 4 +n116-042-185:1982:2044 [1] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-042-185:1987:2034 [6] NCCL INFO Ring 11 : 5 -> 6 -> 7 +n116-042-185:1983:2045 [2] NCCL INFO Ring 04 : 3 -> 2 -> 1 +n116-042-185:1987:2034 [6] NCCL INFO Ring 12 : 7 -> 6 -> 5 +n116-042-185:1985:2033 [4] NCCL INFO Ring 12 : 29 -> 4 -> 3 +n116-042-185:1984:2037 [3] NCCL INFO Ring 06 : 4 -> 3 -> 2 +n116-042-185:1986:2041 [5] NCCL INFO Ring 11 : 4 -> 5 -> 6 +n116-042-185:1981:2030 [0] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-042-185:1983:2045 [2] NCCL INFO Ring 05 : 1 -> 2 -> 3 +n116-042-185:1987:2034 [6] NCCL INFO Ring 13 : 5 -> 6 -> 7 +n116-042-185:1985:2033 [4] NCCL INFO Ring 13 : 3 -> 4 -> 13 +n116-042-185:1984:2037 [3] NCCL INFO Ring 07 : 2 -> 3 -> 4 +n116-042-185:1986:2041 [5] NCCL INFO Ring 12 : 6 -> 5 -> 12 +n116-042-185:1983:2045 [2] NCCL INFO Ring 06 : 3 -> 2 -> 1 +n116-042-185:1987:2034 [6] NCCL INFO Ring 14 : 31 -> 6 -> 5 +n116-042-185:1985:2033 [4] NCCL INFO Ring 14 : 5 -> 4 -> 3 +n116-042-185:1984:2037 [3] NCCL INFO Ring 08 : 4 -> 3 -> 2 +n116-042-185:1981:2030 [0] NCCL INFO Channel 00/16 : 0 7 6 5 4 3 2 1 8 15 14 13 12 11 10 9 16 23 22 21 20 19 18 17 24 31 30 29 28 27 26 25 +n116-042-185:1987:2034 [6] NCCL INFO Ring 15 : 5 -> 6 -> 15 +n116-042-185:1983:2045 [2] NCCL INFO Ring 07 : 1 -> 2 -> 3 +n116-042-185:1986:2041 [5] NCCL INFO Ring 13 : 28 -> 5 -> 6 +n116-042-185:1982:2044 [1] NCCL INFO Ring 00 : 2 -> 1 -> 8 +n116-042-185:1985:2033 [4] NCCL INFO Ring 15 : 3 -> 4 -> 5 +n116-042-185:1984:2037 [3] NCCL INFO Ring 09 : 2 -> 3 -> 4 +n116-042-185:1985:2033 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 3/-1/-1->4->5 [2] 5/-1/-1->4->3 [3] -1/-1/-1->4->5 [4] 5/20/-1->4->-1 [5] 3/-1/-1->4->5 [6] 5/-1/-1->4->3 [7] 3/-1/-1->4->5 [8] 5/-1/-1->4->3 [9] 3/-1/-1->4->5 [10] 5/-1/-1->4->3 [11] -1/-1/-1->4->5 [12] 5/-1/-1->4->12 [13] 3/-1/-1->4->5 [14] 5/-1/-1->4->3 [15] 3/-1/-1->4->5 +n116-042-185:1987:2034 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 5/-1/-1->6->7 [2] 7/-1/-1->6->5 [3] 5/-1/-1->6->7 [4] 7/-1/-1->6->5 [5] -1/-1/-1->6->7 [6] 7/22/-1->6->-1 [7] 5/-1/-1->6->7 [8] 7/-1/-1->6->5 [9] 5/-1/-1->6->7 [10] 7/-1/-1->6->5 [11] 5/-1/-1->6->7 [12] 7/-1/-1->6->5 [13] -1/-1/-1->6->7 [14] 7/-1/-1->6->14 [15] 5/-1/-1->6->7 +n116-042-185:1983:2045 [2] NCCL INFO Ring 08 : 3 -> 2 -> 1 +n116-042-185:1981:2030 [0] NCCL INFO Channel 01/16 : 0 9 10 11 12 13 14 15 8 17 18 19 20 21 22 23 16 25 26 27 28 29 30 31 24 1 2 3 4 5 6 7 +n116-042-185:1987:2034 [6] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-042-185:1982:2044 [1] NCCL INFO Ring 01 : 24 -> 1 -> 2 +n116-042-185:1986:2041 [5] NCCL INFO Ring 14 : 6 -> 5 -> 4 +n116-042-185:1984:2037 [3] NCCL INFO Ring 10 : 4 -> 3 -> 10 +n116-042-185:1985:2033 [4] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-042-185:1983:2045 [2] NCCL INFO Ring 09 : 1 -> 2 -> 3 +n116-042-185:1987:2034 [6] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-042-185:1981:2030 [0] NCCL INFO Channel 02/16 : 0 7 6 5 4 3 10 9 8 15 14 13 12 11 18 17 16 23 22 21 20 19 26 25 24 31 30 29 28 27 2 1 +n116-042-185:1982:2044 [1] NCCL INFO Ring 02 : 2 -> 1 -> 0 +n116-042-185:1986:2041 [5] NCCL INFO Ring 15 : 4 -> 5 -> 6 +n116-042-185:1984:2037 [3] NCCL INFO Ring 11 : 26 -> 3 -> 4 +n116-042-185:1985:2033 [4] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-042-185:1983:2045 [2] NCCL INFO Ring 10 : 27 -> 2 -> 1 +n116-042-185:1987:2034 [6] NCCL INFO P2P Chunksize set to 524288 +n116-042-185:1982:2044 [1] NCCL INFO Ring 03 : 0 -> 1 -> 2 +n116-042-185:1986:2041 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 4/-1/-1->5->6 [2] 6/-1/-1->5->4 [3] 4/-1/-1->5->6 [4] 6/-1/-1->5->4 [5] 4/21/-1->5->-1 [6] -1/-1/-1->5->4 [7] 4/-1/-1->5->6 [8] 6/-1/-1->5->4 [9] 4/-1/-1->5->6 [10] 6/-1/-1->5->4 [11] 4/-1/-1->5->6 [12] 6/-1/-1->5->4 [13] 4/-1/-1->5->13 [14] -1/-1/-1->5->4 [15] 4/-1/-1->5->6 +n116-042-185:1981:2030 [0] NCCL INFO Channel 03/16 : 0 1 2 11 12 13 14 15 8 9 10 19 20 21 22 23 16 17 18 27 28 29 30 31 24 25 26 3 4 5 6 7 +n116-042-185:1984:2037 [3] NCCL INFO Ring 12 : 4 -> 3 -> 2 +n116-042-185:1985:2033 [4] NCCL INFO P2P Chunksize set to 524288 +n116-042-185:1984:2037 [3] NCCL INFO Ring 13 : 2 -> 3 -> 4 +n116-042-185:1983:2045 [2] NCCL INFO Ring 11 : 1 -> 2 -> 11 +n116-042-185:1982:2044 [1] NCCL INFO Ring 04 : 2 -> 1 -> 0 +n116-042-185:1983:2045 [2] NCCL INFO Ring 12 : 3 -> 2 -> 1 +n116-042-185:1986:2041 [5] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-042-185:1986:2041 [5] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-042-185:1984:2037 [3] NCCL INFO Ring 14 : 4 -> 3 -> 2 +n116-042-185:1981:2030 [0] NCCL INFO Channel 04/16 : 0 7 6 5 12 11 10 9 8 15 14 13 20 19 18 17 16 23 22 21 28 27 26 25 24 31 30 29 4 3 2 1 +n116-042-185:1982:2044 [1] NCCL INFO Ring 05 : 0 -> 1 -> 2 +n116-042-185:1983:2045 [2] NCCL INFO Ring 13 : 1 -> 2 -> 3 +n116-042-185:1986:2041 [5] NCCL INFO P2P Chunksize set to 524288 +n116-042-185:1984:2037 [3] NCCL INFO Ring 15 : 2 -> 3 -> 4 +n116-042-185:1982:2044 [1] NCCL INFO Ring 06 : 2 -> 1 -> 0 +n116-042-185:1983:2045 [2] NCCL INFO Ring 14 : 3 -> 2 -> 1 +n116-042-185:1984:2037 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 2/-1/-1->3->4 [2] 4/-1/-1->3->2 [3] 2/19/-1->3->-1 [4] -1/-1/-1->3->2 [5] 2/-1/-1->3->4 [6] 4/-1/-1->3->2 [7] 2/-1/-1->3->4 [8] 4/-1/-1->3->2 [9] 2/-1/-1->3->4 [10] 4/-1/-1->3->2 [11] 2/-1/-1->3->11 [12] -1/-1/-1->3->2 [13] 2/-1/-1->3->4 [14] 4/-1/-1->3->2 [15] 2/-1/-1->3->4 +n116-042-185:1982:2044 [1] NCCL INFO Ring 07 : 0 -> 1 -> 2 +n116-042-185:1981:2030 [0] NCCL INFO Channel 05/16 : 0 1 2 3 4 13 14 15 8 9 10 11 12 21 22 23 16 17 18 19 20 29 30 31 24 25 26 27 28 5 6 7 +n116-042-185:1983:2045 [2] NCCL INFO Ring 15 : 1 -> 2 -> 3 +n116-042-185:1982:2044 [1] NCCL INFO Ring 08 : 2 -> 1 -> 8 +n116-042-185:1984:2037 [3] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-042-185:1984:2037 [3] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-042-185:1983:2045 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] -1/-1/-1->2->3 [2] 3/18/-1->2->-1 [3] 1/-1/-1->2->3 [4] 3/-1/-1->2->1 [5] 1/-1/-1->2->3 [6] 3/-1/-1->2->1 [7] 1/-1/-1->2->3 [8] 3/-1/-1->2->1 [9] -1/-1/-1->2->3 [10] 3/-1/-1->2->10 [11] 1/-1/-1->2->3 [12] 3/-1/-1->2->1 [13] 1/-1/-1->2->3 [14] 3/-1/-1->2->1 [15] 1/-1/-1->2->3 +n116-042-185:1982:2044 [1] NCCL INFO Ring 09 : 24 -> 1 -> 2 +n116-042-185:1981:2030 [0] NCCL INFO Channel 06/16 : 0 7 14 13 12 11 10 9 8 15 22 21 20 19 18 17 16 23 30 29 28 27 26 25 24 31 6 5 4 3 2 1 +n116-042-185:1984:2037 [3] NCCL INFO P2P Chunksize set to 524288 +n116-042-185:1982:2044 [1] NCCL INFO Ring 10 : 2 -> 1 -> 0 +n116-042-185:1983:2045 [2] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-042-185:1982:2044 [1] NCCL INFO Ring 11 : 0 -> 1 -> 2 +n116-042-185:1983:2045 [2] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-042-185:1983:2045 [2] NCCL INFO P2P Chunksize set to 524288 +n116-042-185:1981:2030 [0] NCCL INFO Channel 07/16 : 0 1 2 3 4 5 6 15 8 9 10 11 12 13 14 23 16 17 18 19 20 21 22 31 24 25 26 27 28 29 30 7 +n116-042-185:1982:2044 [1] NCCL INFO Ring 12 : 2 -> 1 -> 0 +n116-042-185:1982:2044 [1] NCCL INFO Ring 13 : 0 -> 1 -> 2 +n116-042-185:1982:2044 [1] NCCL INFO Ring 14 : 2 -> 1 -> 0 +n116-042-185:1981:2030 [0] NCCL INFO Channel 08/16 : 0 7 6 5 4 3 2 1 8 15 14 13 12 11 10 9 16 23 22 21 20 19 18 17 24 31 30 29 28 27 26 25 +n116-042-185:1982:2044 [1] NCCL INFO Ring 15 : 0 -> 1 -> 2 +n116-042-185:1982:2044 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 0/17/-1->1->-1 [2] -1/-1/-1->1->0 [3] 0/-1/-1->1->2 [4] 2/-1/-1->1->0 [5] 0/-1/-1->1->2 [6] 2/-1/-1->1->0 [7] 0/-1/-1->1->2 [8] 2/-1/-1->1->0 [9] 0/-1/-1->1->9 [10] -1/-1/-1->1->0 [11] 0/-1/-1->1->2 [12] 2/-1/-1->1->0 [13] 0/-1/-1->1->2 [14] 2/-1/-1->1->0 [15] 0/-1/-1->1->2 +n116-042-185:1981:2030 [0] NCCL INFO Channel 09/16 : 0 9 10 11 12 13 14 15 8 17 18 19 20 21 22 23 16 25 26 27 28 29 30 31 24 1 2 3 4 5 6 7 +n116-042-185:1982:2044 [1] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-042-185:1982:2044 [1] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-042-185:1981:2030 [0] NCCL INFO Channel 10/16 : 0 7 6 5 4 3 10 9 8 15 14 13 12 11 18 17 16 23 22 21 20 19 26 25 24 31 30 29 28 27 2 1 +n116-042-185:1982:2044 [1] NCCL INFO P2P Chunksize set to 524288 +n116-042-185:1981:2030 [0] NCCL INFO Channel 11/16 : 0 1 2 11 12 13 14 15 8 9 10 19 20 21 22 23 16 17 18 27 28 29 30 31 24 25 26 3 4 5 6 7 +n116-042-185:1981:2030 [0] NCCL INFO Channel 12/16 : 0 7 6 5 12 11 10 9 8 15 14 13 20 19 18 17 16 23 22 21 28 27 26 25 24 31 30 29 4 3 2 1 +n116-042-185:1981:2030 [0] NCCL INFO Channel 13/16 : 0 1 2 3 4 13 14 15 8 9 10 11 12 21 22 23 16 17 18 19 20 29 30 31 24 25 26 27 28 5 6 7 +n116-042-185:1981:2030 [0] NCCL INFO Channel 14/16 : 0 7 14 13 12 11 10 9 8 15 22 21 20 19 18 17 16 23 30 29 28 27 26 25 24 31 6 5 4 3 2 1 +n116-042-185:1981:2030 [0] NCCL INFO Channel 15/16 : 0 1 2 3 4 5 6 15 8 9 10 11 12 13 14 23 16 17 18 19 20 21 22 31 24 25 26 27 28 29 30 7 +n116-042-185:1981:2030 [0] NCCL INFO Ring 00 : 25 -> 0 -> 7 +n116-042-185:1981:2030 [0] NCCL INFO Ring 01 : 7 -> 0 -> 9 +n116-042-185:1981:2030 [0] NCCL INFO Ring 02 : 1 -> 0 -> 7 +n116-042-185:1981:2030 [0] NCCL INFO Ring 03 : 7 -> 0 -> 1 +n116-042-185:1981:2030 [0] NCCL INFO Ring 04 : 1 -> 0 -> 7 +n116-042-185:1981:2030 [0] NCCL INFO Ring 05 : 7 -> 0 -> 1 +n116-042-185:1981:2030 [0] NCCL INFO Ring 06 : 1 -> 0 -> 7 +n116-042-185:1981:2030 [0] NCCL INFO Ring 07 : 7 -> 0 -> 1 +n116-042-185:1981:2030 [0] NCCL INFO Ring 08 : 25 -> 0 -> 7 +n116-042-185:1981:2030 [0] NCCL INFO Ring 09 : 7 -> 0 -> 9 +n116-042-185:1981:2030 [0] NCCL INFO Ring 10 : 1 -> 0 -> 7 +n116-042-185:1981:2030 [0] NCCL INFO Ring 11 : 7 -> 0 -> 1 +n116-042-185:1981:2030 [0] NCCL INFO Ring 12 : 1 -> 0 -> 7 +n116-042-185:1981:2030 [0] NCCL INFO Ring 13 : 7 -> 0 -> 1 +n116-042-185:1981:2030 [0] NCCL INFO Ring 14 : 1 -> 0 -> 7 +n116-042-185:1981:2030 [0] NCCL INFO Ring 15 : 7 -> 0 -> 1 +n116-042-185:1981:2030 [0] NCCL INFO Trees [0] 1/16/-1->0->-1 [1] 7/-1/-1->0->1 [2] 1/-1/-1->0->7 [3] 7/-1/-1->0->1 [4] 1/-1/-1->0->7 [5] 7/-1/-1->0->1 [6] 1/-1/-1->0->7 [7] -1/-1/-1->0->1 [8] 1/-1/-1->0->8 [9] 7/-1/-1->0->1 [10] 1/-1/-1->0->7 [11] 7/-1/-1->0->1 [12] 1/-1/-1->0->7 [13] 7/-1/-1->0->1 [14] 1/-1/-1->0->7 [15] -1/-1/-1->0->1 +n116-042-185:1981:2030 [0] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-042-185:1981:2030 [0] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-042-185:1981:2030 [0] NCCL INFO P2P Chunksize set to 524288 +n116-042-185:1981:2030 [0] NCCL INFO Check P2P Type intraNodeP2pSupport 1 directMode 0 +n116-042-185:1985:2079 [4] NCCL INFO [Proxy Service] Device 4 CPU core 170 +n116-042-185:1988:2078 [7] NCCL INFO [Proxy Service] Device 7 CPU core 60 +n116-042-185:1987:2092 [6] NCCL INFO [Proxy Service UDS] Device 6 CPU core 206 +n116-042-185:1985:2081 [4] NCCL INFO [Proxy Service UDS] Device 4 CPU core 74 +n116-042-185:1988:2080 [7] NCCL INFO [Proxy Service UDS] Device 7 CPU core 174 +n116-042-185:1984:2082 [3] NCCL INFO [Proxy Service] Device 3 CPU core 45 +n116-042-185:1982:2084 [1] NCCL INFO [Proxy Service] Device 1 CPU core 116 +n116-042-185:1983:2083 [2] NCCL INFO [Proxy Service] Device 2 CPU core 12 +n116-042-185:1982:2090 [1] NCCL INFO [Proxy Service UDS] Device 1 CPU core 19 +n116-042-185:1987:2085 [6] NCCL INFO [Proxy Service] Device 6 CPU core 194 +n116-042-185:1984:2087 [3] NCCL INFO [Proxy Service UDS] Device 3 CPU core 151 +n116-042-185:1981:2093 [0] NCCL INFO [Proxy Service UDS] Device 0 CPU core 154 +n116-042-185:1981:2088 [0] NCCL INFO [Proxy Service] Device 0 CPU core 47 +n116-042-185:1983:2089 [2] NCCL INFO [Proxy Service UDS] Device 2 CPU core 19 +n116-042-185:1986:2086 [5] NCCL INFO [Proxy Service] Device 5 CPU core 90 +n116-042-185:1986:2091 [5] NCCL INFO [Proxy Service UDS] Device 5 CPU core 81 +n116-042-185:1985:2033 [4] NCCL INFO NCCL_PROTO set by environment to Simple +n116-042-185:1985:2033 [4] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-042-185:1985:2033 [4] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512 +n116-042-185:1983:2045 [2] NCCL INFO NCCL_PROTO set by environment to Simple +n116-042-185:1985:2033 [4] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 2 p2p channels per peer +n116-042-185:1983:2045 [2] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-042-185:1983:2045 [2] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512 +n116-042-185:1983:2045 [2] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 2 p2p channels per peer +n116-042-185:1988:2038 [7] NCCL INFO NCCL_PROTO set by environment to Simple +n116-042-185:1988:2038 [7] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-042-185:1988:2038 [7] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512 +n116-042-185:1988:2038 [7] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 2 p2p channels per peer +n116-042-185:1984:2037 [3] NCCL INFO NCCL_PROTO set by environment to Simple +n116-042-185:1984:2037 [3] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-042-185:1984:2037 [3] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512 +n116-042-185:1984:2037 [3] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 2 p2p channels per peer +n116-042-185:1982:2044 [1] NCCL INFO NCCL_PROTO set by environment to Simple +n116-042-185:1982:2044 [1] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-042-185:1982:2044 [1] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512 +n116-042-185:1982:2044 [1] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 2 p2p channels per peer +n116-042-185:1981:2030 [0] NCCL INFO NCCL_PROTO set by environment to Simple +n116-042-185:1981:2030 [0] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-042-185:1981:2030 [0] NCCL INFO Enabled NCCL Func/Proto/Algo Matrix: + Function | LL LL128 Simple | Tree Ring CollNetDirect CollNetChain NVLS NVLSTree PAT + Broadcast | 0 0 1 | 1 1 0 0 0 0 0 + Reduce | 0 0 1 | 1 1 0 0 0 0 0 + AllGather | 0 0 1 | 1 1 0 0 0 0 0 +ReduceScatter | 0 0 1 | 1 1 0 0 0 0 0 + AllReduce | 0 0 1 | 1 1 0 0 0 0 0 + +n116-042-185:1981:2030 [0] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512 +n116-042-185:1981:2030 [0] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 2 p2p channels per peer +n116-042-185:1987:2034 [6] NCCL INFO NCCL_PROTO set by environment to Simple +n116-042-185:1987:2034 [6] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-042-185:1987:2034 [6] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512 +n116-042-185:1987:2034 [6] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 2 p2p channels per peer +n116-042-185:1986:2041 [5] NCCL INFO NCCL_PROTO set by environment to Simple +n116-042-185:1986:2041 [5] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-042-185:1986:2041 [5] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512 +n116-042-185:1986:2041 [5] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 2 p2p channels per peer +n116-042-185:1981:2030 [0] NCCL INFO CC Off, workFifoBytes 1048576 +n116-042-185:1988:2038 [7] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-042-185:1988:2038 [7] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-042-185:1988:2038 [7] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-042-185:1988:2038 [7] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-042-185:1988:2038 [7] NCCL INFO ncclCommInitRankConfig comm 0xb3ad0e0 rank 7 nranks 32 cudaDev 7 nvmlDev 7 busId 8c000 commId 0x2a4d35ea441cbcba - Init COMPLETE +n116-042-185:1988:2038 [7] NCCL INFO Init timings - ncclCommInitRankConfig: rank 7 nranks 32 total 6.56 (kernels 0.67, alloc 3.93, bootstrap 0.29, allgathers 0.18, topo 0.69, graphs 0.08, connections 0.04, rest 0.67) +n116-042-185:1982:2044 [1] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-042-185:1982:2044 [1] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-042-185:1982:2044 [1] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-042-185:1982:2044 [1] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-042-185:1982:2044 [1] NCCL INFO ncclCommInitRankConfig comm 0xcdda240 rank 1 nranks 32 cudaDev 1 nvmlDev 1 busId 5000 commId 0x2a4d35ea441cbcba - Init COMPLETE +n116-042-185:1982:2044 [1] NCCL INFO Init timings - ncclCommInitRankConfig: rank 1 nranks 32 total 6.50 (kernels 0.55, alloc 4.02, bootstrap 0.26, allgathers 0.18, topo 0.69, graphs 0.08, connections 0.04, rest 0.68) +n116-042-185:1982:2095 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/CUMEM +n116-042-185:1982:2095 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/CUMEM +n116-042-185:1982:2095 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/CUMEM +n116-042-185:1982:2095 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/CUMEM +n116-042-185:1982:2095 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/CUMEM +n116-042-185:1982:2095 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/CUMEM +n116-042-185:1982:2095 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/CUMEM +n116-042-185:1982:2095 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/CUMEM +n116-042-185:1982:2095 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/CUMEM +n116-042-185:1984:2037 [3] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-042-185:1984:2037 [3] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-042-185:1984:2037 [3] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-042-185:1984:2037 [3] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-042-185:1984:2037 [3] NCCL INFO ncclCommInitRankConfig comm 0xd0746a0 rank 3 nranks 32 cudaDev 3 nvmlDev 3 busId c000 commId 0x2a4d35ea441cbcba - Init COMPLETE +n116-042-185:1984:2037 [3] NCCL INFO Init timings - ncclCommInitRankConfig: rank 3 nranks 32 total 6.59 (kernels 0.63, alloc 4.03, bootstrap 0.26, allgathers 0.15, topo 0.69, graphs 0.12, connections 0.04, rest 0.69) +n116-042-185:1982:2095 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/CUMEM +n116-042-185:1982:2095 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/CUMEM +n116-042-185:1982:2095 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/CUMEM +n116-042-185:1981:2030 [0] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-042-185:1981:2030 [0] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-042-185:1981:2030 [0] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-042-185:1981:2030 [0] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-042-185:1985:2033 [4] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-042-185:1981:2030 [0] NCCL INFO ncclCommInitRankConfig comm 0xc1bf960 rank 0 nranks 32 cudaDev 0 nvmlDev 0 busId 4000 commId 0x2a4d35ea441cbcba - Init COMPLETE +n116-042-185:1981:2030 [0] NCCL INFO Init timings - ncclCommInitRankConfig: rank 0 nranks 32 total 6.71 (kernels 0.73, alloc 3.96, bootstrap 0.32, allgathers 0.19, topo 0.69, graphs 0.08, connections 0.04, rest 0.70) +n116-042-185:1985:2033 [4] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-042-185:1985:2033 [4] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-042-185:1985:2033 [4] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-042-185:1985:2033 [4] NCCL INFO ncclCommInitRankConfig comm 0xc15b3b0 rank 4 nranks 32 cudaDev 4 nvmlDev 4 busId 84000 commId 0x2a4d35ea441cbcba - Init COMPLETE +n116-042-185:1985:2033 [4] NCCL INFO Init timings - ncclCommInitRankConfig: rank 4 nranks 32 total 6.65 (kernels 0.67, alloc 4.03, bootstrap 0.26, allgathers 0.19, topo 0.69, graphs 0.08, connections 0.04, rest 0.70) +n116-042-185:1984:2096 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/CUMEM +n116-042-185:1984:2096 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/CUMEM +n116-042-185:1984:2096 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/CUMEM +n116-042-185:1984:2096 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/CUMEM +n116-042-185:1984:2096 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/CUMEM +n116-042-185:1984:2096 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/CUMEM +n116-042-185:1984:2096 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/CUMEM +n116-042-185:1984:2096 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/CUMEM +n116-042-185:1984:2096 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/CUMEM +n116-042-185:1984:2096 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/CUMEM +n116-042-185:1984:2096 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/CUMEM +n116-042-185:1984:2096 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/CUMEM +n116-042-185:1981:2097 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/CUMEM +n116-042-185:1983:2045 [2] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-042-185:1983:2045 [2] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-042-185:1983:2045 [2] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-042-185:1983:2045 [2] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-042-185:1983:2045 [2] NCCL INFO ncclCommInitRankConfig comm 0xc51c1b0 rank 2 nranks 32 cudaDev 2 nvmlDev 2 busId b000 commId 0x2a4d35ea441cbcba - Init COMPLETE +n116-042-185:1983:2045 [2] NCCL INFO Init timings - ncclCommInitRankConfig: rank 2 nranks 32 total 6.50 (kernels 0.52, alloc 4.01, bootstrap 0.27, allgathers 0.14, topo 0.69, graphs 0.12, connections 0.04, rest 0.71) +n116-042-185:1981:2097 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/CUMEM +n116-042-185:1981:2097 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/CUMEM +n116-042-185:1981:2097 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/CUMEM +n116-042-185:1981:2097 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/CUMEM +n116-042-185:1981:2097 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/CUMEM +n116-042-185:1981:2097 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/CUMEM +n116-042-185:1981:2097 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/CUMEM +n116-042-185:1981:2097 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/CUMEM +n116-042-185:1981:2097 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/CUMEM +n116-042-185:1981:2097 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/CUMEM +n116-042-185:1981:2097 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/CUMEM +n116-042-185:1985:2098 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/CUMEM +n116-042-185:1981:2097 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/CUMEM +n116-042-185:1985:2098 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/CUMEM +n116-042-185:1981:2097 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/CUMEM +n116-042-185:1985:2098 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/CUMEM +n116-042-185:1981:2097 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/CUMEM +n116-042-185:1986:2041 [5] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-042-185:1986:2041 [5] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-042-185:1986:2041 [5] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-042-185:1986:2041 [5] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-042-185:1986:2041 [5] NCCL INFO ncclCommInitRankConfig comm 0xbf2f3e0 rank 5 nranks 32 cudaDev 5 nvmlDev 5 busId 85000 commId 0x2a4d35ea441cbcba - Init COMPLETE +n116-042-185:1986:2041 [5] NCCL INFO Init timings - ncclCommInitRankConfig: rank 5 nranks 32 total 6.56 (kernels 0.61, alloc 3.99, bootstrap 0.26, allgathers 0.14, topo 0.69, graphs 0.12, connections 0.04, rest 0.71) +n116-042-185:1985:2098 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/CUMEM +n116-042-185:1981:2097 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/CUMEM +n116-042-185:1985:2098 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/CUMEM +n116-042-185:1985:2098 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/CUMEM +n116-042-185:1981:2097 [0] NCCL INFO Channel 01/0 : 0[0] -> 7[7] via P2P/CUMEM +n116-042-185:1985:2098 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/CUMEM +n116-042-185:1981:2097 [0] NCCL INFO Channel 02/0 : 0[0] -> 7[7] via P2P/CUMEM +n116-042-185:1985:2098 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/CUMEM +n116-042-185:1981:2097 [0] NCCL INFO Channel 03/0 : 0[0] -> 7[7] via P2P/CUMEM +n116-042-185:1985:2098 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/CUMEM +n116-042-185:1981:2097 [0] NCCL INFO Channel 04/0 : 0[0] -> 7[7] via P2P/CUMEM +n116-042-185:1985:2098 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/CUMEM +n116-042-185:1981:2097 [0] NCCL INFO Channel 05/0 : 0[0] -> 7[7] via P2P/CUMEM +n116-042-185:1985:2098 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/CUMEM +n116-042-185:1981:2097 [0] NCCL INFO Channel 06/0 : 0[0] -> 7[7] via P2P/CUMEM +n116-042-185:1985:2098 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/CUMEM +n116-042-185:1981:2097 [0] NCCL INFO Channel 09/0 : 0[0] -> 7[7] via P2P/CUMEM +n116-042-185:1985:2098 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/CUMEM +n116-042-185:1981:2097 [0] NCCL INFO Channel 10/0 : 0[0] -> 7[7] via P2P/CUMEM +n116-042-185:1985:2098 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/CUMEM +n116-042-185:1981:2097 [0] NCCL INFO Channel 11/0 : 0[0] -> 7[7] via P2P/CUMEM +n116-042-185:1985:2098 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/CUMEM +n116-042-185:1987:2034 [6] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-042-185:1987:2034 [6] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-042-185:1987:2034 [6] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-042-185:1987:2034 [6] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-042-185:1987:2034 [6] NCCL INFO ncclCommInitRankConfig comm 0xc4fc250 rank 6 nranks 32 cudaDev 6 nvmlDev 6 busId 8b000 commId 0x2a4d35ea441cbcba - Init COMPLETE +n116-042-185:1987:2034 [6] NCCL INFO Init timings - ncclCommInitRankConfig: rank 6 nranks 32 total 6.64 (kernels 0.71, alloc 3.98, bootstrap 0.25, allgathers 0.14, topo 0.69, graphs 0.12, connections 0.04, rest 0.71) +n116-042-185:1981:2097 [0] NCCL INFO Channel 12/0 : 0[0] -> 7[7] via P2P/CUMEM +n116-042-185:1985:2098 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/CUMEM +n116-042-185:1981:2097 [0] NCCL INFO Channel 13/0 : 0[0] -> 7[7] via P2P/CUMEM +n116-042-185:1981:2097 [0] NCCL INFO Channel 14/0 : 0[0] -> 7[7] via P2P/CUMEM +n116-042-185:1983:2099 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/CUMEM +n116-042-185:1983:2099 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/CUMEM +n116-042-185:1983:2099 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/CUMEM +n116-042-185:1983:2099 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/CUMEM +n116-042-185:1983:2099 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/CUMEM +n116-042-185:1983:2099 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/CUMEM +n116-042-185:1983:2099 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/CUMEM +n116-042-185:1983:2099 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/CUMEM +n116-042-185:1983:2099 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/CUMEM +n116-042-185:1983:2099 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/CUMEM +n116-042-185:1983:2099 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/CUMEM +n116-042-185:1983:2099 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/CUMEM +n116-042-185:1983:2099 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/CUMEM +n116-042-185:1983:2099 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/CUMEM +n116-042-185:1983:2099 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/CUMEM +n116-042-185:1983:2099 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/CUMEM +n116-042-185:1986:2100 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/CUMEM +n116-042-185:1986:2100 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/CUMEM +n116-042-185:1986:2100 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/CUMEM +n116-042-185:1986:2100 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/CUMEM +n116-042-185:1986:2100 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/CUMEM +n116-042-185:1986:2100 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/CUMEM +n116-042-185:1986:2100 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/CUMEM +n116-042-185:1986:2100 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/CUMEM +n116-042-185:1986:2100 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/CUMEM +n116-042-185:1986:2100 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/CUMEM +n116-042-185:1987:2101 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/CUMEM +n116-042-185:1986:2100 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/CUMEM +n116-042-185:1987:2101 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/CUMEM +n116-042-185:1986:2100 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/CUMEM +n116-042-185:1987:2101 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/CUMEM +n116-042-185:1987:2101 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/CUMEM +n116-042-185:1987:2101 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/CUMEM +n116-042-185:1987:2101 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/CUMEM +n116-042-185:1987:2101 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/CUMEM +n116-042-185:1987:2101 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/CUMEM +n116-042-185:1987:2101 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/CUMEM +n116-042-185:1987:2101 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/CUMEM +n116-042-185:1987:2101 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/CUMEM +n116-042-185:1987:2101 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/CUMEM +n116-042-185:1987:2101 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/CUMEM +n116-042-185:1987:2101 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/CUMEM +n116-042-185:1987:2101 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/CUMEM +n116-042-185:1987:2101 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/CUMEM +n116-042-185:1984:2102 [3] NCCL INFO [Proxy Progress] Device 3 CPU core 154 +n116-042-185:1984:2096 [3] NCCL INFO Channel 11/0 : 3[3] -> 11[3] [send] via NET/FasTrak/3/GDRDMA +n116-042-185:1984:2096 [3] NCCL INFO Channel 03/0 : 19[3] -> 3[3] [receive] via NET/FasTrak/3/GDRDMA +n116-042-185:1984:2096 [3] NCCL INFO Channel 03/0 : 3[3] -> 19[3] [send] via NET/FasTrak/3/GDRDMA +n116-042-185:1985:2103 [4] NCCL INFO [Proxy Progress] Device 4 CPU core 54 +n116-042-185:1985:2098 [4] NCCL INFO Channel 12/0 : 4[4] -> 12[4] [send] via NET/FasTrak/4/GDRDMA +n116-042-185:1982:2104 [1] NCCL INFO [Proxy Progress] Device 1 CPU core 16 +n116-042-185:1982:2095 [1] NCCL INFO Channel 09/0 : 1[1] -> 9[1] [send] via NET/FasTrak/1/GDRDMA +n116-042-185:1985:2098 [4] NCCL INFO Channel 04/0 : 20[4] -> 4[4] [receive] via NET/FasTrak/4/GDRDMA +n116-042-185:1985:2098 [4] NCCL INFO Channel 04/0 : 4[4] -> 20[4] [send] via NET/FasTrak/4/GDRDMA +n116-042-185:1983:2105 [2] NCCL INFO [Proxy Progress] Device 2 CPU core 19 +n116-042-185:1983:2099 [2] NCCL INFO Channel 10/0 : 2[2] -> 10[2] [send] via NET/FasTrak/2/GDRDMA +n116-042-185:1982:2095 [1] NCCL INFO Channel 01/0 : 17[1] -> 1[1] [receive] via NET/FasTrak/1/GDRDMA +n116-042-185:1982:2095 [1] NCCL INFO Channel 01/0 : 1[1] -> 17[1] [send] via NET/FasTrak/1/GDRDMA +n116-042-185:1983:2099 [2] NCCL INFO Channel 02/0 : 18[2] -> 2[2] [receive] via NET/FasTrak/2/GDRDMA +n116-042-185:1983:2099 [2] NCCL INFO Channel 02/0 : 2[2] -> 18[2] [send] via NET/FasTrak/2/GDRDMA +n116-042-185:1988:2106 [7] NCCL INFO [Proxy Progress] Device 7 CPU core 71 +n116-042-185:1988:2094 [7] NCCL INFO Channel 15/0 : 7[7] -> 15[7] [send] via NET/FasTrak/7/GDRDMA +n116-042-185:1981:2107 [0] NCCL INFO [Proxy Progress] Device 0 CPU core 21 +n116-042-185:1981:2097 [0] NCCL INFO Channel 08/0 : 0[0] -> 8[0] [send] via NET/FasTrak/0/GDRDMA +n116-042-185:1988:2094 [7] NCCL INFO Channel 07/0 : 23[7] -> 7[7] [receive] via NET/FasTrak/7/GDRDMA +n116-042-185:1986:2108 [5] NCCL INFO [Proxy Progress] Device 5 CPU core 93 +n116-042-185:1988:2094 [7] NCCL INFO Channel 07/0 : 7[7] -> 23[7] [send] via NET/FasTrak/7/GDRDMA +n116-042-185:1981:2097 [0] NCCL INFO Channel 00/0 : 16[0] -> 0[0] [receive] via NET/FasTrak/0/GDRDMA +n116-042-185:1987:2109 [6] NCCL INFO [Proxy Progress] Device 6 CPU core 74 +n116-042-185:1987:2101 [6] NCCL INFO Channel 14/0 : 6[6] -> 14[6] [send] via NET/FasTrak/6/GDRDMA +n116-042-185:1986:2100 [5] NCCL INFO Channel 13/0 : 5[5] -> 13[5] [send] via NET/FasTrak/5/GDRDMA +n116-042-185:1981:2097 [0] NCCL INFO Channel 00/0 : 0[0] -> 16[0] [send] via NET/FasTrak/0/GDRDMA +n116-042-185:1987:2101 [6] NCCL INFO Channel 06/0 : 22[6] -> 6[6] [receive] via NET/FasTrak/6/GDRDMA +n116-042-185:1987:2101 [6] NCCL INFO Channel 06/0 : 6[6] -> 22[6] [send] via NET/FasTrak/6/GDRDMA +n116-042-185:1986:2100 [5] NCCL INFO Channel 05/0 : 21[5] -> 5[5] [receive] via NET/FasTrak/5/GDRDMA +n116-042-185:1986:2100 [5] NCCL INFO Channel 05/0 : 5[5] -> 21[5] [send] via NET/FasTrak/5/GDRDMA +n116-042-185:1987:2101 [6] NCCL INFO Channel 14/0 : 14[6] -> 6[6] [receive] via NET/FasTrak/6/GDRDMA +n116-042-185:1982:2095 [1] NCCL INFO Channel 09/0 : 9[1] -> 1[1] [receive] via NET/FasTrak/1/GDRDMA +n116-042-185:1982:2095 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/CUMEM +n116-042-185:1982:2095 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/CUMEM +n116-042-185:1982:2095 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/CUMEM +n116-042-185:1982:2095 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/CUMEM +n116-042-185:1982:2095 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/CUMEM +n116-042-185:1982:2095 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/CUMEM +n116-042-185:1982:2095 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/CUMEM +n116-042-185:1982:2095 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/CUMEM +n116-042-185:1982:2095 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/CUMEM +n116-042-185:1982:2095 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/CUMEM +n116-042-185:1982:2095 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/CUMEM +n116-042-185:1982:2095 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/CUMEM +n116-042-185:1982:2095 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/CUMEM +n116-042-185:1982:2095 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/CUMEM +n116-042-185:1982:2095 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/CUMEM +n116-042-185:1982:2095 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/CUMEM +n116-042-185:1986:2100 [5] NCCL INFO Channel 13/0 : 13[5] -> 5[5] [receive] via NET/FasTrak/5/GDRDMA +n116-042-185:1986:2100 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/CUMEM +n116-042-185:1986:2100 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/CUMEM +n116-042-185:1986:2100 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/CUMEM +n116-042-185:1986:2100 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/CUMEM +n116-042-185:1986:2100 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/CUMEM +n116-042-185:1986:2100 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/CUMEM +n116-042-185:1986:2100 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/CUMEM +n116-042-185:1986:2100 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/CUMEM +n116-042-185:1986:2100 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/CUMEM +n116-042-185:1986:2100 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/CUMEM +n116-042-185:1986:2100 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/CUMEM +n116-042-185:1986:2100 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/CUMEM +n116-042-185:1986:2100 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/CUMEM +n116-042-185:1986:2100 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/CUMEM +n116-042-185:1986:2100 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/CUMEM +n116-042-185:1986:2100 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/CUMEM +n116-042-185:1983:2099 [2] NCCL INFO Channel 10/0 : 10[2] -> 2[2] [receive] via NET/FasTrak/2/GDRDMA +n116-042-185:1984:2096 [3] NCCL INFO Channel 11/0 : 11[3] -> 3[3] [receive] via NET/FasTrak/3/GDRDMA +n116-042-185:1985:2098 [4] NCCL INFO Channel 12/0 : 12[4] -> 4[4] [receive] via NET/FasTrak/4/GDRDMA +n116-042-185:1984:2096 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/CUMEM +n116-042-185:1984:2096 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/CUMEM +n116-042-185:1984:2096 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/CUMEM +n116-042-185:1983:2099 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/CUMEM +n116-042-185:1984:2096 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/CUMEM +n116-042-185:1983:2099 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/CUMEM +n116-042-185:1984:2096 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/CUMEM +n116-042-185:1985:2098 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/CUMEM +n116-042-185:1983:2099 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/CUMEM +n116-042-185:1984:2096 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/CUMEM +n116-042-185:1985:2098 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/CUMEM +n116-042-185:1983:2099 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/CUMEM +n116-042-185:1984:2096 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/CUMEM +n116-042-185:1985:2098 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/CUMEM +n116-042-185:1983:2099 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/CUMEM +n116-042-185:1984:2096 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/CUMEM +n116-042-185:1985:2098 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/CUMEM +n116-042-185:1983:2099 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/CUMEM +n116-042-185:1984:2096 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/CUMEM +n116-042-185:1985:2098 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/CUMEM +n116-042-185:1983:2099 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/CUMEM +n116-042-185:1984:2096 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/CUMEM +n116-042-185:1985:2098 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/CUMEM +n116-042-185:1983:2099 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/CUMEM +n116-042-185:1984:2096 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/CUMEM +n116-042-185:1985:2098 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/CUMEM +n116-042-185:1983:2099 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/CUMEM +n116-042-185:1984:2096 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/CUMEM +n116-042-185:1985:2098 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/CUMEM +n116-042-185:1983:2099 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/CUMEM +n116-042-185:1984:2096 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/CUMEM +n116-042-185:1985:2098 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/CUMEM +n116-042-185:1983:2099 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/CUMEM +n116-042-185:1984:2096 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/CUMEM +n116-042-185:1985:2098 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/CUMEM +n116-042-185:1983:2099 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/CUMEM +n116-042-185:1984:2096 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/CUMEM +n116-042-185:1985:2098 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/CUMEM +n116-042-185:1984:2096 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/CUMEM +n116-042-185:1985:2098 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/CUMEM +n116-042-185:1988:2094 [7] NCCL INFO Channel 15/0 : 15[7] -> 7[7] [receive] via NET/FasTrak/7/GDRDMA +n116-042-185:1988:2094 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/CUMEM +n116-042-185:1981:2097 [0] NCCL INFO Channel 08/0 : 8[0] -> 0[0] [receive] via NET/FasTrak/0/GDRDMA +n116-042-185:1988:2094 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/CUMEM +n116-042-185:1988:2094 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/CUMEM +n116-042-185:1988:2094 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/CUMEM +n116-042-185:1988:2094 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/CUMEM +n116-042-185:1988:2094 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/CUMEM +n116-042-185:1988:2094 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/CUMEM +n116-042-185:1988:2094 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/CUMEM +n116-042-185:1988:2094 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/CUMEM +n116-042-185:1988:2094 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/CUMEM +n116-042-185:1988:2094 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/CUMEM +n116-042-185:1988:2094 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/CUMEM +n116-042-185:1988:2094 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/CUMEM +n116-042-185:1988:2094 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/CUMEM +n116-042-185:1988:2094 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/CUMEM +n116-042-185:1988:2094 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/CUMEM +n116-042-185:1988:2094 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/CUMEM +n116-042-185:1988:2094 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/CUMEM +n116-042-185:1988:2094 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/CUMEM +n116-042-185:1988:2094 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/CUMEM +n116-042-185:1988:2094 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/CUMEM +n116-042-185:1988:2094 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/CUMEM +n116-042-185:1988:2094 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/CUMEM +n116-042-185:1987:2101 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/CUMEM +n116-042-185:1987:2101 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/CUMEM +n116-042-185:1988:2094 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/CUMEM +n116-042-185:1988:2094 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/CUMEM +n116-042-185:1987:2101 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/CUMEM +n116-042-185:1988:2094 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/CUMEM +n116-042-185:1987:2101 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/CUMEM +n116-042-185:1987:2101 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/CUMEM +n116-042-185:1988:2094 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/CUMEM +n116-042-185:1988:2094 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/CUMEM +n116-042-185:1987:2101 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/CUMEM +n116-042-185:1987:2101 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/CUMEM +n116-042-185:1987:2101 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/CUMEM +n116-042-185:1987:2101 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/CUMEM +n116-042-185:1987:2101 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/CUMEM +n116-042-185:1987:2101 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/CUMEM +n116-042-185:1987:2101 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/CUMEM +n116-042-185:1983:2099 [2] NCCL INFO Connected all trees +n116-042-185:1984:2096 [3] NCCL INFO Connected all trees +n116-042-185:1985:2098 [4] NCCL INFO Connected all trees +n116-042-185:1981:2097 [0] NCCL INFO Connected all trees +n116-042-185:1982:2095 [1] NCCL INFO Connected all trees +n116-042-185:1986:2100 [5] NCCL INFO Connected all trees +n116-042-185:1988:2094 [7] NCCL INFO Connected all trees +n116-042-185:1987:2101 [6] NCCL INFO Connected all trees +Namespace(batch_size=64, accum_iter=1, eval_batch_size=64, epochs=40, bce_loss=False, unscale_lr=True, select_models='scaledf_v4_train_fix.pkl', model='deit_huge_patch14_LS', input_size=224, drop=0.0, drop_path=0.2, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.1, sched='cosine', lr=1e-05, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-06, decay_epochs=15, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=False, train_mode=True, ThreeAugment=False, src=False, reprob=0.0, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/ScaleDF', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_32card_effbs2048_from_latest24_master_n116-042-185_20260726_1603_vith32_stub_port25003', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_24card_effbs3072_from_epoch7_n116-041-246_n116-042-186_n116-042-185_20260726_0736_vith24_from_epoch7/checkpoint.pth', start_epoch=0, eval=False, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=32, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_huge_patch14_LS +n116-042-185:1988:2132 [7] NCCL INFO Channel 06/0 : 7[7] -> 14[6] [send] via NET/FasTrak/6/GDRDMA +n116-042-185:1988:2132 [7] NCCL INFO Channel 14/0 : 7[7] -> 14[6] [send] via NET/FasTrak/6/GDRDMA +n116-042-185:1982:2134 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/CUMEM +n116-042-185:1982:2134 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/CUMEM +n116-042-185:1982:2134 [1] NCCL INFO Channel 00/0 : 1[1] -> 8[0] [send] via NET/FasTrak/0/GDRDMA +n116-042-185:1982:2134 [1] NCCL INFO Channel 08/0 : 1[1] -> 8[0] [send] via NET/FasTrak/0/GDRDMA +n116-042-185:1983:2133 [2] NCCL INFO Channel 02/0 : 27[3] -> 2[2] [receive] via NET/FasTrak/2/GDRDMA +n116-042-185:1983:2133 [2] NCCL INFO Channel 10/0 : 27[3] -> 2[2] [receive] via NET/FasTrak/2/GDRDMA +n116-042-185:1983:2133 [2] NCCL INFO Channel 03/0 : 2[2] -> 11[3] [send] via NET/FasTrak/3/GDRDMA +n116-042-185:1983:2133 [2] NCCL INFO Channel 11/0 : 2[2] -> 11[3] [send] via NET/FasTrak/3/GDRDMA +n116-042-185:1981:2135 [0] NCCL INFO Channel 00/0 : 25[1] -> 0[0] [receive] via NET/FasTrak/0/GDRDMA +n116-042-185:1981:2135 [0] NCCL INFO Channel 08/0 : 25[1] -> 0[0] [receive] via NET/FasTrak/0/GDRDMA +n116-042-185:1981:2135 [0] NCCL INFO Channel 00/0 : 0[0] -> 7[7] via P2P/CUMEM +n116-042-185:1981:2135 [0] NCCL INFO Channel 08/0 : 0[0] -> 7[7] via P2P/CUMEM +n116-042-185:1982:2134 [1] NCCL INFO Channel 01/0 : 24[0] -> 1[1] [receive] via NET/FasTrak/1/GDRDMA +n116-042-185:1982:2134 [1] NCCL INFO Channel 09/0 : 24[0] -> 1[1] [receive] via NET/FasTrak/1/GDRDMA +n116-042-185:1981:2135 [0] NCCL INFO Channel 01/0 : 0[0] -> 9[1] [send] via NET/FasTrak/1/GDRDMA +n116-042-185:1981:2135 [0] NCCL INFO Channel 09/0 : 0[0] -> 9[1] [send] via NET/FasTrak/1/GDRDMA +n116-042-185:1983:2133 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/CUMEM +n116-042-185:1983:2133 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/CUMEM +n116-042-185:1986:2138 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/CUMEM +n116-042-185:1986:2138 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/CUMEM +n116-042-185:1986:2138 [5] NCCL INFO Channel 04/0 : 5[5] -> 12[4] [send] via NET/FasTrak/4/GDRDMA +n116-042-185:1986:2138 [5] NCCL INFO Channel 12/0 : 5[5] -> 12[4] [send] via NET/FasTrak/4/GDRDMA +n116-042-185:1987:2131 [6] NCCL INFO Channel 06/0 : 31[7] -> 6[6] [receive] via NET/FasTrak/6/GDRDMA +n116-042-185:1987:2131 [6] NCCL INFO Channel 14/0 : 31[7] -> 6[6] [receive] via NET/FasTrak/6/GDRDMA +n116-042-185:1986:2138 [5] NCCL INFO Channel 05/0 : 28[4] -> 5[5] [receive] via NET/FasTrak/5/GDRDMA +n116-042-185:1987:2131 [6] NCCL INFO Channel 07/0 : 6[6] -> 15[7] [send] via NET/FasTrak/7/GDRDMA +n116-042-185:1987:2131 [6] NCCL INFO Channel 15/0 : 6[6] -> 15[7] [send] via NET/FasTrak/7/GDRDMA +n116-042-185:1986:2138 [5] NCCL INFO Channel 13/0 : 28[4] -> 5[5] [receive] via NET/FasTrak/5/GDRDMA +n116-042-185:1984:2139 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/CUMEM +n116-042-185:1984:2139 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/CUMEM +n116-042-185:1984:2139 [3] NCCL INFO Channel 02/0 : 3[3] -> 10[2] [send] via NET/FasTrak/2/GDRDMA +n116-042-185:1984:2139 [3] NCCL INFO Channel 10/0 : 3[3] -> 10[2] [send] via NET/FasTrak/2/GDRDMA +n116-042-185:1985:2140 [4] NCCL INFO Channel 04/0 : 29[5] -> 4[4] [receive] via NET/FasTrak/4/GDRDMA +n116-042-185:1985:2140 [4] NCCL INFO Channel 12/0 : 29[5] -> 4[4] [receive] via NET/FasTrak/4/GDRDMA +n116-042-185:1984:2139 [3] NCCL INFO Channel 03/0 : 26[2] -> 3[3] [receive] via NET/FasTrak/3/GDRDMA +n116-042-185:1985:2140 [4] NCCL INFO Channel 05/0 : 4[4] -> 13[5] [send] via NET/FasTrak/5/GDRDMA +n116-042-185:1985:2140 [4] NCCL INFO Channel 13/0 : 4[4] -> 13[5] [send] via NET/FasTrak/5/GDRDMA +n116-042-185:1984:2139 [3] NCCL INFO Channel 11/0 : 26[2] -> 3[3] [receive] via NET/FasTrak/3/GDRDMA +n116-042-185:1987:2131 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/CUMEM +n116-042-185:1987:2131 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/CUMEM +n116-042-185:1988:2132 [7] NCCL INFO Channel 07/0 : 30[6] -> 7[7] [receive] via NET/FasTrak/7/GDRDMA +n116-042-185:1988:2132 [7] NCCL INFO Channel 15/0 : 30[6] -> 7[7] [receive] via NET/FasTrak/7/GDRDMA +n116-042-185:1988:2132 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/CUMEM +n116-042-185:1988:2132 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/CUMEM +n116-042-185:1985:2140 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/CUMEM +n116-042-185:1985:2140 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/CUMEM +n116-042-185:1988:2132 [7] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-042-185:1987:2131 [6] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-042-185:1986:2138 [5] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-042-185:1985:2140 [4] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-042-185:1984:2139 [3] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-042-185:1983:2133 [2] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-042-185:1981:2135 [0] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-042-185:1982:2134 [1] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +number of params: 630848002 +Start training for 40 epochs +Epoch: [15] [ 0/7269] eta: 18:23:49 lr: 0.000007 loss: 0.6480 (0.6480) time: 9.1112 data: 4.2180 max mem: 52148 +Epoch: [15] [ 10/7269] eta: 2:31:56 lr: 0.000007 loss: 0.5896 (0.5861) time: 1.2559 data: 0.3836 max mem: 52148 +Epoch: [15] [ 20/7269] eta: 1:40:49 lr: 0.000007 loss: 0.5733 (0.5707) time: 0.4206 data: 0.0002 max mem: 52148 +Epoch: [15] [ 30/7269] eta: 1:22:38 lr: 0.000007 loss: 0.5364 (0.5559) time: 0.3709 data: 0.0002 max mem: 52148 +Epoch: [15] [ 40/7269] eta: 1:13:17 lr: 0.000007 loss: 0.5247 (0.5488) time: 0.3708 data: 0.0002 max mem: 52148 +Epoch: [15] [ 50/7269] eta: 1:07:36 lr: 0.000007 loss: 0.5443 (0.5533) time: 0.3711 data: 0.0002 max mem: 52148 +Epoch: [15] [ 60/7269] eta: 1:03:45 lr: 0.000007 loss: 0.5508 (0.5484) time: 0.3714 data: 0.0002 max mem: 52148 +Epoch: [15] [ 70/7269] eta: 1:00:58 lr: 0.000007 loss: 0.5145 (0.5462) time: 0.3713 data: 0.0002 max mem: 52148 +Epoch: [15] [ 80/7269] eta: 0:58:51 lr: 0.000007 loss: 0.5145 (0.5430) time: 0.3710 data: 0.0002 max mem: 52148 +Epoch: [15] [ 90/7269] eta: 0:57:11 lr: 0.000007 loss: 0.5332 (0.5424) time: 0.3709 data: 0.0001 max mem: 52148 +Epoch: [15] [ 100/7269] eta: 0:55:51 lr: 0.000007 loss: 0.5602 (0.5441) time: 0.3711 data: 0.0001 max mem: 52148 +Epoch: [15] [ 110/7269] eta: 0:54:44 lr: 0.000007 loss: 0.5602 (0.5446) time: 0.3713 data: 0.0001 max mem: 52148 +Epoch: [15] [ 120/7269] eta: 0:53:47 lr: 0.000007 loss: 0.5771 (0.5456) time: 0.3712 data: 0.0001 max mem: 52148 +Epoch: [15] [ 130/7269] eta: 0:52:59 lr: 0.000007 loss: 0.5620 (0.5450) time: 0.3713 data: 0.0002 max mem: 52148 +Epoch: [15] [ 140/7269] eta: 0:52:18 lr: 0.000007 loss: 0.5327 (0.5429) time: 0.3716 data: 0.0002 max mem: 52148 +Epoch: [15] [ 150/7269] eta: 0:51:41 lr: 0.000007 loss: 0.5235 (0.5415) time: 0.3716 data: 0.0002 max mem: 52148 +Epoch: [15] [ 160/7269] eta: 0:51:09 lr: 0.000007 loss: 0.5309 (0.5421) time: 0.3720 data: 0.0002 max mem: 52148 +Epoch: [15] [ 170/7269] eta: 0:50:39 lr: 0.000007 loss: 0.5387 (0.5418) time: 0.3718 data: 0.0002 max mem: 52148 +Epoch: [15] [ 180/7269] eta: 0:50:13 lr: 0.000007 loss: 0.5231 (0.5403) time: 0.3716 data: 0.0002 max mem: 52148 +Epoch: [15] [ 190/7269] eta: 0:49:49 lr: 0.000007 loss: 0.5207 (0.5404) time: 0.3715 data: 0.0002 max mem: 52148 +Epoch: [15] [ 200/7269] eta: 0:49:26 lr: 0.000007 loss: 0.5134 (0.5385) time: 0.3710 data: 0.0002 max mem: 52148 +Epoch: [15] [ 210/7269] eta: 0:49:06 lr: 0.000007 loss: 0.5166 (0.5373) time: 0.3719 data: 0.0002 max mem: 52148 +Epoch: [15] [ 220/7269] eta: 0:48:48 lr: 0.000007 loss: 0.5348 (0.5359) time: 0.3724 data: 0.0002 max mem: 52148 +Epoch: [15] [ 230/7269] eta: 0:48:30 lr: 0.000007 loss: 0.5010 (0.5345) time: 0.3720 data: 0.0002 max mem: 52148 +Epoch: [15] [ 240/7269] eta: 0:48:14 lr: 0.000007 loss: 0.5269 (0.5338) time: 0.3717 data: 0.0002 max mem: 52148 +Epoch: [15] [ 250/7269] eta: 0:47:59 lr: 0.000007 loss: 0.5204 (0.5324) time: 0.3714 data: 0.0002 max mem: 52148 +Epoch: [15] [ 260/7269] eta: 0:47:44 lr: 0.000007 loss: 0.4913 (0.5312) time: 0.3711 data: 0.0001 max mem: 52148 +Epoch: [15] [ 270/7269] eta: 0:47:30 lr: 0.000007 loss: 0.5101 (0.5305) time: 0.3711 data: 0.0001 max mem: 52148 +Epoch: [15] [ 280/7269] eta: 0:48:12 lr: 0.000007 loss: 0.5573 (0.5304) time: 0.4811 data: 0.0001 max mem: 52148 +Epoch: [15] [ 290/7269] eta: 0:47:57 lr: 0.000007 loss: 0.5074 (0.5304) time: 0.4812 data: 0.0002 max mem: 52148 +Epoch: [15] [ 300/7269] eta: 0:47:44 lr: 0.000007 loss: 0.5057 (0.5295) time: 0.3713 data: 0.0002 max mem: 52148 +Epoch: [15] [ 310/7269] eta: 0:47:59 lr: 0.000007 loss: 0.5430 (0.5299) time: 0.4342 data: 0.0001 max mem: 52148 +Epoch: [15] [ 320/7269] eta: 0:47:46 lr: 0.000007 loss: 0.5619 (0.5308) time: 0.4345 data: 0.0001 max mem: 52148 +Epoch: [15] [ 330/7269] eta: 0:47:33 lr: 0.000007 loss: 0.5517 (0.5306) time: 0.3714 data: 0.0001 max mem: 52148 +Epoch: [15] [ 340/7269] eta: 0:47:41 lr: 0.000007 loss: 0.5495 (0.5312) time: 0.4203 data: 0.0001 max mem: 52148 +Epoch: [15] [ 350/7269] eta: 0:47:28 lr: 0.000007 loss: 0.5457 (0.5312) time: 0.4205 data: 0.0001 max mem: 52148 +Epoch: [15] [ 360/7269] eta: 0:47:16 lr: 0.000007 loss: 0.5343 (0.5312) time: 0.3712 data: 0.0001 max mem: 52148 +Epoch: [15] [ 370/7269] eta: 0:47:29 lr: 0.000007 loss: 0.5500 (0.5312) time: 0.4344 data: 0.0001 max mem: 52148 +Epoch: [15] [ 380/7269] eta: 0:47:17 lr: 0.000007 loss: 0.5395 (0.5305) time: 0.4346 data: 0.0001 max mem: 52148 +Epoch: [15] [ 390/7269] eta: 0:47:06 lr: 0.000007 loss: 0.5241 (0.5299) time: 0.3720 data: 0.0001 max mem: 52148 +Epoch: [15] [ 400/7269] eta: 0:47:17 lr: 0.000007 loss: 0.5294 (0.5304) time: 0.4366 data: 0.0002 max mem: 52148 +Epoch: [15] [ 410/7269] eta: 0:47:06 lr: 0.000007 loss: 0.5396 (0.5307) time: 0.4362 data: 0.0002 max mem: 52148 +Epoch: [15] [ 420/7269] eta: 0:46:55 lr: 0.000007 loss: 0.5473 (0.5301) time: 0.3719 data: 0.0002 max mem: 52148 +Epoch: [15] [ 430/7269] eta: 0:47:01 lr: 0.000007 loss: 0.5421 (0.5308) time: 0.4213 data: 0.0001 max mem: 52148 +Epoch: [15] [ 440/7269] eta: 0:46:50 lr: 0.000007 loss: 0.5424 (0.5316) time: 0.4210 data: 0.0001 max mem: 52148 +Epoch: [15] [ 450/7269] eta: 0:46:59 lr: 0.000007 loss: 0.5263 (0.5311) time: 0.4344 data: 0.0002 max mem: 52148 +Epoch: [15] [ 460/7269] eta: 0:46:49 lr: 0.000007 loss: 0.4968 (0.5305) time: 0.4342 data: 0.0002 max mem: 52148 +Epoch: [15] [ 470/7269] eta: 0:46:58 lr: 0.000007 loss: 0.4797 (0.5294) time: 0.4368 data: 0.0002 max mem: 52148 +Epoch: [15] [ 480/7269] eta: 0:46:47 lr: 0.000007 loss: 0.5037 (0.5295) time: 0.4374 data: 0.0002 max mem: 52148 +Epoch: [15] [ 490/7269] eta: 0:46:37 lr: 0.000007 loss: 0.5292 (0.5299) time: 0.3716 data: 0.0001 max mem: 52148 +Epoch: [15] [ 500/7269] eta: 0:46:28 lr: 0.000007 loss: 0.5135 (0.5298) time: 0.3716 data: 0.0002 max mem: 52148 +Epoch: [15] [ 510/7269] eta: 0:46:19 lr: 0.000007 loss: 0.5535 (0.5303) time: 0.3726 data: 0.0002 max mem: 52148 +Epoch: [15] [ 520/7269] eta: 0:46:09 lr: 0.000007 loss: 0.5381 (0.5304) time: 0.3724 data: 0.0001 max mem: 52148 +Epoch: [15] [ 530/7269] eta: 0:46:00 lr: 0.000007 loss: 0.5238 (0.5296) time: 0.3717 data: 0.0002 max mem: 52148 +Epoch: [15] [ 540/7269] eta: 0:45:51 lr: 0.000007 loss: 0.5538 (0.5297) time: 0.3711 data: 0.0002 max mem: 52148 +Epoch: [15] [ 550/7269] eta: 0:45:43 lr: 0.000007 loss: 0.5589 (0.5297) time: 0.3711 data: 0.0002 max mem: 52148 +Epoch: [15] [ 560/7269] eta: 0:45:34 lr: 0.000007 loss: 0.5608 (0.5304) time: 0.3713 data: 0.0002 max mem: 52148 +Epoch: [15] [ 570/7269] eta: 0:45:26 lr: 0.000007 loss: 0.5725 (0.5309) time: 0.3718 data: 0.0002 max mem: 52148 +Epoch: [15] [ 580/7269] eta: 0:45:18 lr: 0.000007 loss: 0.5486 (0.5310) time: 0.3721 data: 0.0002 max mem: 52148 +Epoch: [15] [ 590/7269] eta: 0:45:10 lr: 0.000007 loss: 0.5486 (0.5312) time: 0.3717 data: 0.0001 max mem: 52148 +Epoch: [15] [ 600/7269] eta: 0:45:02 lr: 0.000007 loss: 0.5616 (0.5314) time: 0.3715 data: 0.0001 max mem: 52148 +Epoch: [15] [ 610/7269] eta: 0:44:54 lr: 0.000007 loss: 0.5501 (0.5315) time: 0.3716 data: 0.0002 max mem: 52148 +Epoch: [15] [ 620/7269] eta: 0:44:47 lr: 0.000007 loss: 0.5388 (0.5318) time: 0.3722 data: 0.0001 max mem: 52148 +Epoch: [15] [ 630/7269] eta: 0:44:39 lr: 0.000007 loss: 0.5562 (0.5320) time: 0.3720 data: 0.0002 max mem: 52148 +Epoch: [15] [ 640/7269] eta: 0:44:32 lr: 0.000007 loss: 0.5200 (0.5314) time: 0.3714 data: 0.0002 max mem: 52148 +Epoch: [15] [ 650/7269] eta: 0:44:25 lr: 0.000007 loss: 0.4954 (0.5312) time: 0.3717 data: 0.0002 max mem: 52148 +Epoch: [15] [ 660/7269] eta: 0:44:18 lr: 0.000007 loss: 0.4839 (0.5304) time: 0.3718 data: 0.0002 max mem: 52148 +Epoch: [15] [ 670/7269] eta: 0:44:11 lr: 0.000007 loss: 0.5507 (0.5313) time: 0.3714 data: 0.0001 max mem: 52148 +Epoch: [15] [ 680/7269] eta: 0:44:04 lr: 0.000007 loss: 0.5724 (0.5314) time: 0.3716 data: 0.0002 max mem: 52148 +Epoch: [15] [ 690/7269] eta: 0:43:57 lr: 0.000007 loss: 0.5183 (0.5311) time: 0.3718 data: 0.0002 max mem: 52148 +Epoch: [15] [ 700/7269] eta: 0:43:50 lr: 0.000007 loss: 0.5207 (0.5310) time: 0.3717 data: 0.0002 max mem: 52148 +Epoch: [15] [ 710/7269] eta: 0:43:43 lr: 0.000007 loss: 0.5312 (0.5307) time: 0.3719 data: 0.0001 max mem: 52148 +Epoch: [15] [ 720/7269] eta: 0:43:37 lr: 0.000007 loss: 0.5303 (0.5303) time: 0.3719 data: 0.0001 max mem: 52148 +Epoch: [15] [ 730/7269] eta: 0:43:30 lr: 0.000007 loss: 0.5401 (0.5307) time: 0.3718 data: 0.0001 max mem: 52148 +Epoch: [15] [ 740/7269] eta: 0:43:24 lr: 0.000007 loss: 0.5507 (0.5314) time: 0.3718 data: 0.0001 max mem: 52148 +Epoch: [15] [ 750/7269] eta: 0:43:18 lr: 0.000007 loss: 0.5507 (0.5312) time: 0.3715 data: 0.0001 max mem: 52148 +Epoch: [15] [ 760/7269] eta: 0:43:11 lr: 0.000007 loss: 0.5333 (0.5314) time: 0.3714 data: 0.0001 max mem: 52148 +Epoch: [15] [ 770/7269] eta: 0:43:05 lr: 0.000007 loss: 0.5535 (0.5316) time: 0.3717 data: 0.0002 max mem: 52148 +Epoch: [15] [ 780/7269] eta: 0:42:59 lr: 0.000007 loss: 0.5608 (0.5317) time: 0.3721 data: 0.0002 max mem: 52148 +Epoch: [15] [ 790/7269] eta: 0:42:53 lr: 0.000007 loss: 0.5553 (0.5315) time: 0.3721 data: 0.0001 max mem: 52148 +Epoch: [15] [ 800/7269] eta: 0:42:47 lr: 0.000007 loss: 0.5111 (0.5316) time: 0.3720 data: 0.0002 max mem: 52148 +Epoch: [15] [ 810/7269] eta: 0:42:41 lr: 0.000007 loss: 0.4890 (0.5310) time: 0.3723 data: 0.0002 max mem: 52148 +Epoch: [15] [ 820/7269] eta: 0:42:35 lr: 0.000007 loss: 0.5369 (0.5311) time: 0.3727 data: 0.0001 max mem: 52148 +Epoch: [15] [ 830/7269] eta: 0:42:39 lr: 0.000007 loss: 0.5502 (0.5306) time: 0.4337 data: 0.0001 max mem: 52148 +Epoch: [15] [ 840/7269] eta: 0:42:49 lr: 0.000007 loss: 0.4763 (0.5303) time: 0.5388 data: 0.0001 max mem: 52148 +Epoch: [15] [ 850/7269] eta: 0:42:50 lr: 0.000007 loss: 0.5076 (0.5299) time: 0.5226 data: 0.0001 max mem: 52148 +Epoch: [15] [ 860/7269] eta: 0:42:53 lr: 0.000007 loss: 0.5329 (0.5300) time: 0.4796 data: 0.0002 max mem: 52148 +Epoch: [15] [ 870/7269] eta: 0:42:54 lr: 0.000007 loss: 0.5439 (0.5301) time: 0.4859 data: 0.0001 max mem: 52148 +Epoch: [15] [ 880/7269] eta: 0:42:48 lr: 0.000007 loss: 0.5178 (0.5298) time: 0.4238 data: 0.0001 max mem: 52148 +Epoch: [15] [ 890/7269] eta: 0:42:55 lr: 0.000007 loss: 0.5178 (0.5294) time: 0.4655 data: 0.0002 max mem: 52148 +Epoch: [15] [ 900/7269] eta: 0:42:49 lr: 0.000007 loss: 0.5075 (0.5294) time: 0.4651 data: 0.0002 max mem: 52148 +Epoch: [15] [ 910/7269] eta: 0:42:51 lr: 0.000007 loss: 0.5154 (0.5295) time: 0.4326 data: 0.0001 max mem: 52148 +Epoch: [15] [ 920/7269] eta: 0:42:45 lr: 0.000007 loss: 0.5816 (0.5298) time: 0.4321 data: 0.0001 max mem: 52148 +Epoch: [15] [ 930/7269] eta: 0:42:47 lr: 0.000007 loss: 0.5434 (0.5301) time: 0.4331 data: 0.0001 max mem: 52148 +Epoch: [15] [ 940/7269] eta: 0:42:49 lr: 0.000007 loss: 0.5353 (0.5302) time: 0.4956 data: 0.0002 max mem: 52148 +Epoch: [15] [ 950/7269] eta: 0:42:43 lr: 0.000007 loss: 0.5005 (0.5297) time: 0.4341 data: 0.0001 max mem: 52148 +Epoch: [15] [ 960/7269] eta: 0:42:37 lr: 0.000007 loss: 0.4875 (0.5297) time: 0.3721 data: 0.0001 max mem: 52148 +Epoch: [15] [ 970/7269] eta: 0:42:30 lr: 0.000007 loss: 0.5513 (0.5300) time: 0.3728 data: 0.0001 max mem: 52148 +Epoch: [15] [ 980/7269] eta: 0:42:36 lr: 0.000007 loss: 0.5808 (0.5302) time: 0.4656 data: 0.0001 max mem: 52148 +Epoch: [15] [ 990/7269] eta: 0:42:30 lr: 0.000007 loss: 0.5689 (0.5302) time: 0.4646 data: 0.0001 max mem: 52148 +Epoch: [15] [1000/7269] eta: 0:42:24 lr: 0.000007 loss: 0.5340 (0.5302) time: 0.3718 data: 0.0002 max mem: 52148 +Epoch: [15] [1010/7269] eta: 0:42:18 lr: 0.000007 loss: 0.5386 (0.5304) time: 0.3721 data: 0.0002 max mem: 52148 +Epoch: [15] [1020/7269] eta: 0:42:11 lr: 0.000007 loss: 0.5621 (0.5307) time: 0.3715 data: 0.0001 max mem: 52148 +Epoch: [15] [1030/7269] eta: 0:42:05 lr: 0.000007 loss: 0.5240 (0.5303) time: 0.3714 data: 0.0002 max mem: 52148 +Epoch: [15] [1040/7269] eta: 0:41:59 lr: 0.000007 loss: 0.5081 (0.5306) time: 0.3716 data: 0.0002 max mem: 52148 +Epoch: [15] [1050/7269] eta: 0:41:53 lr: 0.000007 loss: 0.5356 (0.5305) time: 0.3714 data: 0.0001 max mem: 52148 +Epoch: [15] [1060/7269] eta: 0:41:53 lr: 0.000007 loss: 0.5288 (0.5307) time: 0.4162 data: 0.0002 max mem: 52148 +Epoch: [15] [1070/7269] eta: 0:41:47 lr: 0.000007 loss: 0.5290 (0.5306) time: 0.4162 data: 0.0002 max mem: 52148 +Epoch: [15] [1080/7269] eta: 0:41:41 lr: 0.000007 loss: 0.5278 (0.5305) time: 0.3713 data: 0.0002 max mem: 52148 +Epoch: [15] [1090/7269] eta: 0:41:35 lr: 0.000007 loss: 0.5195 (0.5304) time: 0.3717 data: 0.0002 max mem: 52148 +Epoch: [15] [1100/7269] eta: 0:41:29 lr: 0.000007 loss: 0.5195 (0.5303) time: 0.3714 data: 0.0002 max mem: 52148 +Epoch: [15] [1110/7269] eta: 0:41:23 lr: 0.000007 loss: 0.5434 (0.5304) time: 0.3711 data: 0.0002 max mem: 52148 +Epoch: [15] [1120/7269] eta: 0:41:17 lr: 0.000007 loss: 0.5512 (0.5304) time: 0.3717 data: 0.0002 max mem: 52148 +Epoch: [15] [1130/7269] eta: 0:41:12 lr: 0.000007 loss: 0.5579 (0.5307) time: 0.3717 data: 0.0001 max mem: 52148 +Epoch: [15] [1140/7269] eta: 0:41:06 lr: 0.000007 loss: 0.5222 (0.5304) time: 0.3717 data: 0.0001 max mem: 52148 +Epoch: [15] [1150/7269] eta: 0:41:00 lr: 0.000007 loss: 0.5491 (0.5306) time: 0.3720 data: 0.0001 max mem: 52148 +Epoch: [15] [1160/7269] eta: 0:40:55 lr: 0.000007 loss: 0.5534 (0.5305) time: 0.3714 data: 0.0002 max mem: 52148 +Epoch: [15] [1170/7269] eta: 0:40:49 lr: 0.000007 loss: 0.5534 (0.5305) time: 0.3711 data: 0.0002 max mem: 52148 +Epoch: [15] [1180/7269] eta: 0:40:43 lr: 0.000007 loss: 0.5655 (0.5309) time: 0.3712 data: 0.0002 max mem: 52148 +Epoch: [15] [1190/7269] eta: 0:40:38 lr: 0.000007 loss: 0.5643 (0.5311) time: 0.3710 data: 0.0001 max mem: 52148 +Epoch: [15] [1200/7269] eta: 0:40:32 lr: 0.000007 loss: 0.5456 (0.5311) time: 0.3714 data: 0.0001 max mem: 52148 +Epoch: [15] [1210/7269] eta: 0:40:27 lr: 0.000007 loss: 0.5563 (0.5312) time: 0.3715 data: 0.0001 max mem: 52148 +Epoch: [15] [1220/7269] eta: 0:40:26 lr: 0.000007 loss: 0.5393 (0.5311) time: 0.4171 data: 0.0002 max mem: 52148 +Epoch: [15] [1230/7269] eta: 0:40:20 lr: 0.000007 loss: 0.5137 (0.5310) time: 0.4173 data: 0.0002 max mem: 52148 +Epoch: [15] [1240/7269] eta: 0:40:15 lr: 0.000007 loss: 0.5249 (0.5310) time: 0.3714 data: 0.0001 max mem: 52148 +Epoch: [15] [1250/7269] eta: 0:40:10 lr: 0.000007 loss: 0.4990 (0.5308) time: 0.3711 data: 0.0001 max mem: 52148 +Epoch: [15] [1260/7269] eta: 0:40:04 lr: 0.000007 loss: 0.5206 (0.5311) time: 0.3714 data: 0.0001 max mem: 52148 +Epoch: [15] [1270/7269] eta: 0:39:59 lr: 0.000007 loss: 0.5534 (0.5312) time: 0.3717 data: 0.0001 max mem: 52148 +Epoch: [15] [1280/7269] eta: 0:39:54 lr: 0.000007 loss: 0.5550 (0.5317) time: 0.3715 data: 0.0001 max mem: 52148 +Epoch: [15] [1290/7269] eta: 0:39:48 lr: 0.000007 loss: 0.5690 (0.5318) time: 0.3719 data: 0.0001 max mem: 52148 +Epoch: [15] [1300/7269] eta: 0:39:53 lr: 0.000007 loss: 0.5297 (0.5319) time: 0.4776 data: 0.0001 max mem: 52148 +Epoch: [15] [1310/7269] eta: 0:39:52 lr: 0.000007 loss: 0.5206 (0.5318) time: 0.5237 data: 0.0002 max mem: 52148 +Epoch: [15] [1320/7269] eta: 0:39:46 lr: 0.000007 loss: 0.5010 (0.5316) time: 0.4179 data: 0.0002 max mem: 52148 +Epoch: [15] [1330/7269] eta: 0:39:41 lr: 0.000007 loss: 0.5010 (0.5316) time: 0.3720 data: 0.0002 max mem: 52148 +Epoch: [15] [1340/7269] eta: 0:39:36 lr: 0.000007 loss: 0.5569 (0.5316) time: 0.3725 data: 0.0002 max mem: 52148 +Epoch: [15] [1350/7269] eta: 0:39:30 lr: 0.000007 loss: 0.5533 (0.5315) time: 0.3719 data: 0.0001 max mem: 52148 +Epoch: [15] [1360/7269] eta: 0:39:34 lr: 0.000007 loss: 0.5533 (0.5318) time: 0.4767 data: 0.0001 max mem: 52148 +Epoch: [15] [1370/7269] eta: 0:39:29 lr: 0.000007 loss: 0.5404 (0.5318) time: 0.4773 data: 0.0001 max mem: 52148 +Epoch: [15] [1380/7269] eta: 0:39:24 lr: 0.000007 loss: 0.5289 (0.5318) time: 0.3721 data: 0.0002 max mem: 52148 +Epoch: [15] [1390/7269] eta: 0:39:18 lr: 0.000007 loss: 0.5288 (0.5318) time: 0.3717 data: 0.0002 max mem: 52148 +Epoch: [15] [1400/7269] eta: 0:39:18 lr: 0.000007 loss: 0.5335 (0.5317) time: 0.4330 data: 0.0002 max mem: 52148 +Epoch: [15] [1410/7269] eta: 0:39:18 lr: 0.000007 loss: 0.5510 (0.5319) time: 0.4921 data: 0.0002 max mem: 52148 +Epoch: [15] [1420/7269] eta: 0:39:13 lr: 0.000007 loss: 0.5523 (0.5318) time: 0.4314 data: 0.0001 max mem: 52148 +Epoch: [15] [1430/7269] eta: 0:39:07 lr: 0.000007 loss: 0.5440 (0.5320) time: 0.3723 data: 0.0001 max mem: 52148 +Epoch: [15] [1440/7269] eta: 0:39:07 lr: 0.000007 loss: 0.5403 (0.5319) time: 0.4348 data: 0.0002 max mem: 52148 +Epoch: [15] [1450/7269] eta: 0:39:02 lr: 0.000007 loss: 0.5403 (0.5320) time: 0.4348 data: 0.0002 max mem: 52148 +Epoch: [15] [1460/7269] eta: 0:38:57 lr: 0.000007 loss: 0.5605 (0.5322) time: 0.3719 data: 0.0001 max mem: 52148 +Epoch: [15] [1470/7269] eta: 0:38:51 lr: 0.000007 loss: 0.5605 (0.5321) time: 0.3720 data: 0.0002 max mem: 52148 +Epoch: [15] [1480/7269] eta: 0:38:46 lr: 0.000007 loss: 0.4969 (0.5319) time: 0.3721 data: 0.0001 max mem: 52148 +Epoch: [15] [1490/7269] eta: 0:38:41 lr: 0.000007 loss: 0.4939 (0.5315) time: 0.3724 data: 0.0001 max mem: 52148 +Epoch: [15] [1500/7269] eta: 0:38:36 lr: 0.000007 loss: 0.4951 (0.5314) time: 0.3722 data: 0.0002 max mem: 52148 +Epoch: [15] [1510/7269] eta: 0:38:31 lr: 0.000007 loss: 0.5262 (0.5313) time: 0.3714 data: 0.0002 max mem: 52148 +Epoch: [15] [1520/7269] eta: 0:38:29 lr: 0.000007 loss: 0.5513 (0.5316) time: 0.4178 data: 0.0001 max mem: 52148 +Epoch: [15] [1530/7269] eta: 0:38:27 lr: 0.000007 loss: 0.5513 (0.5315) time: 0.4639 data: 0.0001 max mem: 52148 +Epoch: [15] [1540/7269] eta: 0:38:30 lr: 0.000007 loss: 0.5474 (0.5316) time: 0.5237 data: 0.0001 max mem: 52148 +Epoch: [15] [1550/7269] eta: 0:38:32 lr: 0.000007 loss: 0.5530 (0.5315) time: 0.5708 data: 0.0001 max mem: 52148 +Epoch: [15] [1560/7269] eta: 0:38:27 lr: 0.000007 loss: 0.5286 (0.5315) time: 0.4648 data: 0.0001 max mem: 52148 +Epoch: [15] [1570/7269] eta: 0:38:21 lr: 0.000007 loss: 0.5222 (0.5313) time: 0.3717 data: 0.0001 max mem: 52148 +Epoch: [15] [1580/7269] eta: 0:38:21 lr: 0.000007 loss: 0.5239 (0.5311) time: 0.4324 data: 0.0001 max mem: 52148 +Epoch: [15] [1590/7269] eta: 0:38:19 lr: 0.000007 loss: 0.5239 (0.5312) time: 0.4787 data: 0.0001 max mem: 52148 +Epoch: [15] [1600/7269] eta: 0:38:13 lr: 0.000007 loss: 0.5527 (0.5314) time: 0.4180 data: 0.0001 max mem: 52148 +Epoch: [15] [1610/7269] eta: 0:38:08 lr: 0.000007 loss: 0.5527 (0.5314) time: 0.3716 data: 0.0001 max mem: 52148 +Epoch: [15] [1620/7269] eta: 0:38:10 lr: 0.000007 loss: 0.5243 (0.5313) time: 0.4764 data: 0.0001 max mem: 52148 +Epoch: [15] [1630/7269] eta: 0:38:05 lr: 0.000007 loss: 0.5224 (0.5313) time: 0.4759 data: 0.0001 max mem: 52148 +Epoch: [15] [1640/7269] eta: 0:38:00 lr: 0.000007 loss: 0.5244 (0.5313) time: 0.3713 data: 0.0002 max mem: 52148 +Epoch: [15] [1650/7269] eta: 0:37:55 lr: 0.000007 loss: 0.5161 (0.5308) time: 0.3717 data: 0.0002 max mem: 52148 +Epoch: [15] [1660/7269] eta: 0:37:50 lr: 0.000007 loss: 0.4314 (0.5305) time: 0.3717 data: 0.0002 max mem: 52148 +Epoch: [15] [1670/7269] eta: 0:37:48 lr: 0.000007 loss: 0.4910 (0.5306) time: 0.4317 data: 0.0001 max mem: 52148 +Epoch: [15] [1680/7269] eta: 0:37:47 lr: 0.000007 loss: 0.5425 (0.5306) time: 0.4928 data: 0.0001 max mem: 52148 +Epoch: [15] [1690/7269] eta: 0:37:48 lr: 0.000007 loss: 0.5159 (0.5306) time: 0.5242 data: 0.0002 max mem: 52148 +Epoch: [15] [1700/7269] eta: 0:37:43 lr: 0.000007 loss: 0.5565 (0.5308) time: 0.4631 data: 0.0002 max mem: 52148 +Epoch: [15] [1710/7269] eta: 0:37:38 lr: 0.000007 loss: 0.5533 (0.5308) time: 0.3726 data: 0.0002 max mem: 52148 +Epoch: [15] [1720/7269] eta: 0:37:33 lr: 0.000007 loss: 0.5069 (0.5306) time: 0.3731 data: 0.0002 max mem: 52148 +Epoch: [15] [1730/7269] eta: 0:37:27 lr: 0.000007 loss: 0.5532 (0.5308) time: 0.3722 data: 0.0002 max mem: 52148 +Epoch: [15] [1740/7269] eta: 0:37:22 lr: 0.000007 loss: 0.5352 (0.5307) time: 0.3718 data: 0.0001 max mem: 52148 +Epoch: [15] [1750/7269] eta: 0:37:17 lr: 0.000007 loss: 0.5062 (0.5305) time: 0.3717 data: 0.0001 max mem: 52148 +Epoch: [15] [1760/7269] eta: 0:37:12 lr: 0.000007 loss: 0.5032 (0.5304) time: 0.3714 data: 0.0002 max mem: 52148 +Epoch: [15] [1770/7269] eta: 0:37:10 lr: 0.000007 loss: 0.5338 (0.5305) time: 0.4175 data: 0.0001 max mem: 52148 +Epoch: [15] [1780/7269] eta: 0:37:14 lr: 0.000007 loss: 0.5338 (0.5304) time: 0.5719 data: 0.0001 max mem: 52148 +Epoch: [15] [1790/7269] eta: 0:37:09 lr: 0.000007 loss: 0.5227 (0.5303) time: 0.5261 data: 0.0001 max mem: 52148 +Epoch: [15] [1800/7269] eta: 0:37:07 lr: 0.000007 loss: 0.5640 (0.5305) time: 0.4172 data: 0.0001 max mem: 52148 +Epoch: [15] [1810/7269] eta: 0:37:02 lr: 0.000007 loss: 0.5489 (0.5304) time: 0.4172 data: 0.0001 max mem: 52148 +Epoch: [15] [1820/7269] eta: 0:36:56 lr: 0.000007 loss: 0.5257 (0.5305) time: 0.3722 data: 0.0001 max mem: 52148 +Epoch: [15] [1830/7269] eta: 0:36:54 lr: 0.000007 loss: 0.5258 (0.5305) time: 0.4189 data: 0.0002 max mem: 52148 +Epoch: [15] [1840/7269] eta: 0:36:49 lr: 0.000007 loss: 0.5414 (0.5306) time: 0.4187 data: 0.0002 max mem: 52148 +Epoch: [15] [1850/7269] eta: 0:36:44 lr: 0.000007 loss: 0.5389 (0.5305) time: 0.3719 data: 0.0001 max mem: 52148 +Epoch: [15] [1860/7269] eta: 0:36:45 lr: 0.000007 loss: 0.5389 (0.5305) time: 0.4772 data: 0.0001 max mem: 52148 +Epoch: [15] [1870/7269] eta: 0:36:40 lr: 0.000007 loss: 0.5273 (0.5303) time: 0.4772 data: 0.0002 max mem: 52148 +Epoch: [15] [1880/7269] eta: 0:36:35 lr: 0.000007 loss: 0.5397 (0.5306) time: 0.3725 data: 0.0002 max mem: 52148 +Epoch: [15] [1890/7269] eta: 0:36:33 lr: 0.000007 loss: 0.5618 (0.5305) time: 0.4209 data: 0.0002 max mem: 52148 +Epoch: [15] [1900/7269] eta: 0:36:27 lr: 0.000007 loss: 0.5271 (0.5305) time: 0.4203 data: 0.0002 max mem: 52148 +Epoch: [15] [1910/7269] eta: 0:36:22 lr: 0.000007 loss: 0.5442 (0.5306) time: 0.3723 data: 0.0002 max mem: 52148 +Epoch: [15] [1920/7269] eta: 0:36:21 lr: 0.000007 loss: 0.5037 (0.5302) time: 0.4351 data: 0.0002 max mem: 52148 +Epoch: [15] [1930/7269] eta: 0:36:16 lr: 0.000007 loss: 0.4876 (0.5302) time: 0.4345 data: 0.0002 max mem: 52148 +Epoch: [15] [1940/7269] eta: 0:36:11 lr: 0.000007 loss: 0.5142 (0.5301) time: 0.3717 data: 0.0002 max mem: 52148 +Epoch: [15] [1950/7269] eta: 0:36:12 lr: 0.000007 loss: 0.5122 (0.5299) time: 0.4841 data: 0.0002 max mem: 52148 +Epoch: [15] [1960/7269] eta: 0:36:10 lr: 0.000007 loss: 0.5138 (0.5298) time: 0.5461 data: 0.0002 max mem: 52148 +Epoch: [15] [1970/7269] eta: 0:36:05 lr: 0.000007 loss: 0.5138 (0.5297) time: 0.4338 data: 0.0002 max mem: 52148 +Epoch: [15] [1980/7269] eta: 0:36:05 lr: 0.000007 loss: 0.5189 (0.5296) time: 0.4606 data: 0.0002 max mem: 52148 +Epoch: [15] [1990/7269] eta: 0:36:00 lr: 0.000007 loss: 0.5524 (0.5296) time: 0.4613 data: 0.0002 max mem: 52148 +Epoch: [15] [2000/7269] eta: 0:35:54 lr: 0.000007 loss: 0.5238 (0.5296) time: 0.3728 data: 0.0002 max mem: 52148 +Epoch: [15] [2010/7269] eta: 0:35:52 lr: 0.000007 loss: 0.5427 (0.5297) time: 0.4193 data: 0.0002 max mem: 52148 +Epoch: [15] [2020/7269] eta: 0:35:47 lr: 0.000007 loss: 0.5427 (0.5296) time: 0.4192 data: 0.0002 max mem: 52148 +Epoch: [15] [2030/7269] eta: 0:35:42 lr: 0.000007 loss: 0.5507 (0.5298) time: 0.3720 data: 0.0001 max mem: 52148 +Epoch: [15] [2040/7269] eta: 0:35:39 lr: 0.000007 loss: 0.5666 (0.5299) time: 0.4173 data: 0.0002 max mem: 52148 +Epoch: [15] [2050/7269] eta: 0:35:34 lr: 0.000007 loss: 0.5245 (0.5298) time: 0.4176 data: 0.0002 max mem: 52148 +Epoch: [15] [2060/7269] eta: 0:35:29 lr: 0.000007 loss: 0.5245 (0.5299) time: 0.3728 data: 0.0001 max mem: 52148 +Epoch: [15] [2070/7269] eta: 0:35:26 lr: 0.000007 loss: 0.5419 (0.5299) time: 0.4185 data: 0.0002 max mem: 52148 +Epoch: [15] [2080/7269] eta: 0:35:21 lr: 0.000007 loss: 0.5329 (0.5299) time: 0.4180 data: 0.0002 max mem: 52148 +Epoch: [15] [2090/7269] eta: 0:35:19 lr: 0.000007 loss: 0.5509 (0.5300) time: 0.4171 data: 0.0001 max mem: 52148 +Epoch: [15] [2100/7269] eta: 0:35:14 lr: 0.000007 loss: 0.5319 (0.5299) time: 0.4171 data: 0.0002 max mem: 52148 +Epoch: [15] [2110/7269] eta: 0:35:16 lr: 0.000007 loss: 0.5313 (0.5300) time: 0.5291 data: 0.0002 max mem: 52148 +Epoch: [15] [2120/7269] eta: 0:35:11 lr: 0.000007 loss: 0.5342 (0.5299) time: 0.5289 data: 0.0002 max mem: 52148 +Epoch: [15] [2130/7269] eta: 0:35:11 lr: 0.000007 loss: 0.5221 (0.5300) time: 0.4802 data: 0.0002 max mem: 52148 +Epoch: [15] [2140/7269] eta: 0:35:09 lr: 0.000007 loss: 0.5387 (0.5300) time: 0.5414 data: 0.0001 max mem: 52148 +Epoch: [15] [2150/7269] eta: 0:35:09 lr: 0.000007 loss: 0.5427 (0.5301) time: 0.5402 data: 0.0002 max mem: 52148 +Epoch: [15] [2160/7269] eta: 0:35:09 lr: 0.000007 loss: 0.5586 (0.5302) time: 0.5694 data: 0.0002 max mem: 52148 +Epoch: [15] [2170/7269] eta: 0:35:03 lr: 0.000007 loss: 0.5630 (0.5303) time: 0.4622 data: 0.0002 max mem: 52148 +Epoch: [15] [2180/7269] eta: 0:34:58 lr: 0.000007 loss: 0.5258 (0.5302) time: 0.3723 data: 0.0001 max mem: 52148 +Epoch: [15] [2190/7269] eta: 0:34:53 lr: 0.000007 loss: 0.5331 (0.5303) time: 0.3723 data: 0.0001 max mem: 52148 +Epoch: [15] [2200/7269] eta: 0:34:48 lr: 0.000007 loss: 0.5551 (0.5302) time: 0.3722 data: 0.0002 max mem: 52148 +Epoch: [15] [2210/7269] eta: 0:34:43 lr: 0.000007 loss: 0.5272 (0.5301) time: 0.3719 data: 0.0002 max mem: 52148 +Epoch: [15] [2220/7269] eta: 0:34:38 lr: 0.000007 loss: 0.4996 (0.5301) time: 0.3717 data: 0.0002 max mem: 52148 +Epoch: [15] [2230/7269] eta: 0:34:33 lr: 0.000007 loss: 0.4875 (0.5299) time: 0.3725 data: 0.0002 max mem: 52148 +Epoch: [15] [2240/7269] eta: 0:34:28 lr: 0.000007 loss: 0.5035 (0.5299) time: 0.3726 data: 0.0001 max mem: 52148 +Epoch: [15] [2250/7269] eta: 0:34:23 lr: 0.000007 loss: 0.5128 (0.5298) time: 0.3724 data: 0.0002 max mem: 52148 +Epoch: [15] [2260/7269] eta: 0:34:18 lr: 0.000007 loss: 0.5030 (0.5297) time: 0.3720 data: 0.0002 max mem: 52148 +Epoch: [15] [2270/7269] eta: 0:34:16 lr: 0.000007 loss: 0.5228 (0.5298) time: 0.4312 data: 0.0002 max mem: 52148 +Epoch: [15] [2280/7269] eta: 0:34:13 lr: 0.000007 loss: 0.5353 (0.5298) time: 0.4792 data: 0.0002 max mem: 52148 +Epoch: [15] [2290/7269] eta: 0:34:08 lr: 0.000007 loss: 0.5353 (0.5299) time: 0.4201 data: 0.0001 max mem: 52148 +Epoch: [15] [2300/7269] eta: 0:34:03 lr: 0.000007 loss: 0.5049 (0.5297) time: 0.3727 data: 0.0001 max mem: 52148 +Epoch: [15] [2310/7269] eta: 0:34:01 lr: 0.000007 loss: 0.5004 (0.5297) time: 0.4330 data: 0.0001 max mem: 52148 +Epoch: [15] [2320/7269] eta: 0:33:58 lr: 0.000007 loss: 0.5552 (0.5298) time: 0.4795 data: 0.0002 max mem: 52148 +Epoch: [15] [2330/7269] eta: 0:33:55 lr: 0.000007 loss: 0.5739 (0.5300) time: 0.4653 data: 0.0002 max mem: 52148 +Epoch: [15] [2340/7269] eta: 0:33:50 lr: 0.000007 loss: 0.5528 (0.5300) time: 0.4188 data: 0.0002 max mem: 52148 +Epoch: [15] [2350/7269] eta: 0:33:45 lr: 0.000007 loss: 0.5203 (0.5299) time: 0.3725 data: 0.0001 max mem: 52148 +Epoch: [15] [2360/7269] eta: 0:33:42 lr: 0.000007 loss: 0.5174 (0.5298) time: 0.4205 data: 0.0001 max mem: 52148 +Epoch: [15] [2370/7269] eta: 0:33:40 lr: 0.000007 loss: 0.5179 (0.5298) time: 0.4811 data: 0.0001 max mem: 52148 +Epoch: [15] [2380/7269] eta: 0:33:37 lr: 0.000007 loss: 0.5423 (0.5299) time: 0.4803 data: 0.0001 max mem: 52148 +Epoch: [15] [2390/7269] eta: 0:33:33 lr: 0.000007 loss: 0.5224 (0.5298) time: 0.4661 data: 0.0001 max mem: 52148 +Epoch: [15] [2400/7269] eta: 0:33:29 lr: 0.000007 loss: 0.5074 (0.5297) time: 0.4188 data: 0.0001 max mem: 52148 +Epoch: [15] [2410/7269] eta: 0:33:24 lr: 0.000007 loss: 0.5679 (0.5299) time: 0.3723 data: 0.0001 max mem: 52148 +Epoch: [15] [2420/7269] eta: 0:33:21 lr: 0.000007 loss: 0.5679 (0.5298) time: 0.4322 data: 0.0001 max mem: 52148 +Epoch: [15] [2430/7269] eta: 0:33:16 lr: 0.000007 loss: 0.5110 (0.5298) time: 0.4321 data: 0.0002 max mem: 52148 +Epoch: [15] [2440/7269] eta: 0:33:11 lr: 0.000007 loss: 0.5352 (0.5299) time: 0.3722 data: 0.0001 max mem: 52148 +Epoch: [15] [2450/7269] eta: 0:33:09 lr: 0.000007 loss: 0.5276 (0.5299) time: 0.4343 data: 0.0001 max mem: 52148 +Epoch: [15] [2460/7269] eta: 0:33:04 lr: 0.000007 loss: 0.5208 (0.5300) time: 0.4348 data: 0.0002 max mem: 52148 +Epoch: [15] [2470/7269] eta: 0:33:01 lr: 0.000007 loss: 0.5269 (0.5298) time: 0.4344 data: 0.0002 max mem: 52148 +Epoch: [15] [2480/7269] eta: 0:32:58 lr: 0.000007 loss: 0.5149 (0.5297) time: 0.4820 data: 0.0001 max mem: 52148 +Epoch: [15] [2490/7269] eta: 0:32:53 lr: 0.000007 loss: 0.5298 (0.5298) time: 0.4203 data: 0.0001 max mem: 52148 +Epoch: [15] [2500/7269] eta: 0:32:48 lr: 0.000007 loss: 0.5296 (0.5298) time: 0.3721 data: 0.0001 max mem: 52148 +Epoch: [15] [2510/7269] eta: 0:32:44 lr: 0.000007 loss: 0.5014 (0.5298) time: 0.3719 data: 0.0001 max mem: 52148 +Epoch: [15] [2520/7269] eta: 0:32:39 lr: 0.000007 loss: 0.5123 (0.5298) time: 0.3719 data: 0.0001 max mem: 52148 +Epoch: [15] [2530/7269] eta: 0:32:35 lr: 0.000007 loss: 0.5236 (0.5298) time: 0.4166 data: 0.0001 max mem: 52148 +Epoch: [15] [2540/7269] eta: 0:32:31 lr: 0.000007 loss: 0.5283 (0.5297) time: 0.4172 data: 0.0001 max mem: 52148 +Epoch: [15] [2550/7269] eta: 0:32:28 lr: 0.000007 loss: 0.5227 (0.5298) time: 0.4362 data: 0.0001 max mem: 52148 +Epoch: [15] [2560/7269] eta: 0:32:25 lr: 0.000007 loss: 0.5619 (0.5300) time: 0.4801 data: 0.0001 max mem: 52148 +Epoch: [15] [2570/7269] eta: 0:32:20 lr: 0.000007 loss: 0.5619 (0.5301) time: 0.4167 data: 0.0001 max mem: 52148 +Epoch: [15] [2580/7269] eta: 0:32:17 lr: 0.000007 loss: 0.5427 (0.5301) time: 0.4200 data: 0.0001 max mem: 52148 +Epoch: [15] [2590/7269] eta: 0:32:14 lr: 0.000007 loss: 0.5256 (0.5301) time: 0.4651 data: 0.0002 max mem: 52148 +Epoch: [15] [2600/7269] eta: 0:32:09 lr: 0.000007 loss: 0.5321 (0.5302) time: 0.4175 data: 0.0002 max mem: 52148 +Epoch: [15] [2610/7269] eta: 0:32:04 lr: 0.000007 loss: 0.5241 (0.5301) time: 0.3718 data: 0.0002 max mem: 52148 +Epoch: [15] [2620/7269] eta: 0:32:03 lr: 0.000007 loss: 0.5193 (0.5300) time: 0.4770 data: 0.0002 max mem: 52148 +Epoch: [15] [2630/7269] eta: 0:32:01 lr: 0.000007 loss: 0.5037 (0.5300) time: 0.5713 data: 0.0002 max mem: 52148 +Epoch: [15] [2640/7269] eta: 0:31:56 lr: 0.000007 loss: 0.5147 (0.5301) time: 0.4659 data: 0.0002 max mem: 52148 +Epoch: [15] [2650/7269] eta: 0:31:54 lr: 0.000007 loss: 0.5570 (0.5301) time: 0.4343 data: 0.0001 max mem: 52148 +Epoch: [15] [2660/7269] eta: 0:31:49 lr: 0.000007 loss: 0.5566 (0.5301) time: 0.4352 data: 0.0002 max mem: 52148 +Epoch: [15] [2670/7269] eta: 0:31:46 lr: 0.000007 loss: 0.5517 (0.5302) time: 0.4343 data: 0.0001 max mem: 52148 +Epoch: [15] [2680/7269] eta: 0:31:41 lr: 0.000007 loss: 0.5496 (0.5302) time: 0.4331 data: 0.0001 max mem: 52148 +Epoch: [15] [2690/7269] eta: 0:31:38 lr: 0.000007 loss: 0.5812 (0.5303) time: 0.4181 data: 0.0002 max mem: 52148 +Epoch: [15] [2700/7269] eta: 0:31:33 lr: 0.000007 loss: 0.5446 (0.5304) time: 0.4186 data: 0.0002 max mem: 52148 +Epoch: [15] [2710/7269] eta: 0:31:28 lr: 0.000007 loss: 0.5384 (0.5305) time: 0.3726 data: 0.0002 max mem: 52148 +Epoch: [15] [2720/7269] eta: 0:31:23 lr: 0.000007 loss: 0.5434 (0.5305) time: 0.3722 data: 0.0002 max mem: 52148 +Epoch: [15] [2730/7269] eta: 0:31:20 lr: 0.000007 loss: 0.5642 (0.5307) time: 0.4315 data: 0.0001 max mem: 52148 +Epoch: [15] [2740/7269] eta: 0:31:17 lr: 0.000007 loss: 0.5254 (0.5305) time: 0.4772 data: 0.0001 max mem: 52148 +Epoch: [15] [2750/7269] eta: 0:31:12 lr: 0.000007 loss: 0.5186 (0.5305) time: 0.4175 data: 0.0001 max mem: 52148 +Epoch: [15] [2760/7269] eta: 0:31:07 lr: 0.000007 loss: 0.5146 (0.5304) time: 0.3722 data: 0.0001 max mem: 52148 +Epoch: [15] [2770/7269] eta: 0:31:04 lr: 0.000007 loss: 0.5133 (0.5305) time: 0.4173 data: 0.0001 max mem: 52148 +Epoch: [15] [2780/7269] eta: 0:30:59 lr: 0.000007 loss: 0.5550 (0.5305) time: 0.4169 data: 0.0001 max mem: 52148 +Epoch: [15] [2790/7269] eta: 0:30:56 lr: 0.000007 loss: 0.4913 (0.5304) time: 0.4344 data: 0.0002 max mem: 52148 +Epoch: [15] [2800/7269] eta: 0:30:53 lr: 0.000007 loss: 0.4997 (0.5303) time: 0.4953 data: 0.0002 max mem: 52148 +Epoch: [15] [2810/7269] eta: 0:30:50 lr: 0.000007 loss: 0.5398 (0.5304) time: 0.4785 data: 0.0002 max mem: 52148 +Epoch: [15] [2820/7269] eta: 0:30:45 lr: 0.000007 loss: 0.5444 (0.5304) time: 0.4182 data: 0.0002 max mem: 52148 +Epoch: [15] [2830/7269] eta: 0:30:42 lr: 0.000007 loss: 0.5182 (0.5303) time: 0.4326 data: 0.0002 max mem: 52148 +Epoch: [15] [2840/7269] eta: 0:30:40 lr: 0.000007 loss: 0.5198 (0.5303) time: 0.5237 data: 0.0001 max mem: 52148 +Epoch: [15] [2850/7269] eta: 0:30:37 lr: 0.000007 loss: 0.5061 (0.5303) time: 0.5109 data: 0.0001 max mem: 52148 +Epoch: [15] [2860/7269] eta: 0:30:32 lr: 0.000007 loss: 0.4772 (0.5300) time: 0.4199 data: 0.0002 max mem: 52148 +Epoch: [15] [2870/7269] eta: 0:30:27 lr: 0.000007 loss: 0.4772 (0.5300) time: 0.3726 data: 0.0002 max mem: 52148 +Epoch: [15] [2880/7269] eta: 0:30:24 lr: 0.000007 loss: 0.5258 (0.5300) time: 0.4340 data: 0.0001 max mem: 52148 +Epoch: [15] [2890/7269] eta: 0:30:21 lr: 0.000007 loss: 0.5242 (0.5299) time: 0.4788 data: 0.0001 max mem: 52148 +Epoch: [15] [2900/7269] eta: 0:30:16 lr: 0.000007 loss: 0.5253 (0.5301) time: 0.4169 data: 0.0002 max mem: 52148 +Epoch: [15] [2910/7269] eta: 0:30:11 lr: 0.000007 loss: 0.5646 (0.5300) time: 0.3724 data: 0.0002 max mem: 52148 +Epoch: [15] [2920/7269] eta: 0:30:08 lr: 0.000007 loss: 0.5087 (0.5300) time: 0.4186 data: 0.0002 max mem: 52148 +Epoch: [15] [2930/7269] eta: 0:30:03 lr: 0.000007 loss: 0.5365 (0.5301) time: 0.4187 data: 0.0002 max mem: 52148 +Epoch: [15] [2940/7269] eta: 0:30:01 lr: 0.000007 loss: 0.5523 (0.5302) time: 0.4797 data: 0.0002 max mem: 52148 +Epoch: [15] [2950/7269] eta: 0:29:57 lr: 0.000007 loss: 0.5424 (0.5301) time: 0.4798 data: 0.0002 max mem: 52148 +Epoch: [15] [2960/7269] eta: 0:29:54 lr: 0.000007 loss: 0.5108 (0.5300) time: 0.4329 data: 0.0001 max mem: 52148 +Epoch: [15] [2970/7269] eta: 0:29:51 lr: 0.000007 loss: 0.5278 (0.5300) time: 0.4924 data: 0.0001 max mem: 52148 +Epoch: [15] [2980/7269] eta: 0:29:46 lr: 0.000007 loss: 0.5374 (0.5300) time: 0.4321 data: 0.0002 max mem: 52148 +Epoch: [15] [2990/7269] eta: 0:29:41 lr: 0.000007 loss: 0.5160 (0.5299) time: 0.3723 data: 0.0002 max mem: 52148 +Epoch: [15] [3000/7269] eta: 0:29:36 lr: 0.000007 loss: 0.5160 (0.5300) time: 0.3722 data: 0.0002 max mem: 52148 +Epoch: [15] [3010/7269] eta: 0:29:33 lr: 0.000007 loss: 0.5394 (0.5300) time: 0.4200 data: 0.0002 max mem: 52148 +Epoch: [15] [3020/7269] eta: 0:29:29 lr: 0.000007 loss: 0.5448 (0.5300) time: 0.4648 data: 0.0001 max mem: 52148 +Epoch: [15] [3030/7269] eta: 0:29:24 lr: 0.000007 loss: 0.5225 (0.5299) time: 0.4168 data: 0.0001 max mem: 52148 +Epoch: [15] [3040/7269] eta: 0:29:21 lr: 0.000007 loss: 0.4988 (0.5300) time: 0.4342 data: 0.0001 max mem: 52148 +Epoch: [15] [3050/7269] eta: 0:29:17 lr: 0.000007 loss: 0.5331 (0.5299) time: 0.4341 data: 0.0001 max mem: 52148 +Epoch: [15] [3060/7269] eta: 0:29:12 lr: 0.000007 loss: 0.5432 (0.5299) time: 0.3722 data: 0.0001 max mem: 52148 +Epoch: [15] [3070/7269] eta: 0:29:07 lr: 0.000007 loss: 0.5291 (0.5300) time: 0.3729 data: 0.0002 max mem: 52148 +Epoch: [15] [3080/7269] eta: 0:29:04 lr: 0.000007 loss: 0.5244 (0.5299) time: 0.4181 data: 0.0002 max mem: 52148 +Epoch: [15] [3090/7269] eta: 0:29:00 lr: 0.000007 loss: 0.5210 (0.5299) time: 0.4661 data: 0.0001 max mem: 52148 +Epoch: [15] [3100/7269] eta: 0:28:55 lr: 0.000007 loss: 0.5290 (0.5299) time: 0.4209 data: 0.0001 max mem: 52148 +Epoch: [15] [3110/7269] eta: 0:28:51 lr: 0.000007 loss: 0.5431 (0.5299) time: 0.3725 data: 0.0002 max mem: 52148 +Epoch: [15] [3120/7269] eta: 0:28:47 lr: 0.000007 loss: 0.5431 (0.5299) time: 0.4349 data: 0.0002 max mem: 52148 +Epoch: [15] [3130/7269] eta: 0:28:43 lr: 0.000007 loss: 0.5337 (0.5299) time: 0.4348 data: 0.0002 max mem: 52148 +Epoch: [15] [3140/7269] eta: 0:28:38 lr: 0.000007 loss: 0.5126 (0.5298) time: 0.3718 data: 0.0001 max mem: 52148 +Epoch: [15] [3150/7269] eta: 0:28:33 lr: 0.000007 loss: 0.5090 (0.5297) time: 0.3719 data: 0.0001 max mem: 52148 +Epoch: [15] [3160/7269] eta: 0:28:31 lr: 0.000007 loss: 0.5080 (0.5297) time: 0.4648 data: 0.0001 max mem: 52148 +Epoch: [15] [3170/7269] eta: 0:28:26 lr: 0.000007 loss: 0.5447 (0.5298) time: 0.4650 data: 0.0001 max mem: 52148 +Epoch: [15] [3180/7269] eta: 0:28:21 lr: 0.000007 loss: 0.5447 (0.5297) time: 0.3725 data: 0.0001 max mem: 52148 +Epoch: [15] [3190/7269] eta: 0:28:17 lr: 0.000007 loss: 0.5263 (0.5297) time: 0.3726 data: 0.0002 max mem: 52148 +Epoch: [15] [3200/7269] eta: 0:28:12 lr: 0.000007 loss: 0.5410 (0.5297) time: 0.3726 data: 0.0002 max mem: 52148 +Epoch: [15] [3210/7269] eta: 0:28:07 lr: 0.000007 loss: 0.5420 (0.5297) time: 0.3724 data: 0.0001 max mem: 52148 +Epoch: [15] [3220/7269] eta: 0:28:03 lr: 0.000007 loss: 0.5699 (0.5298) time: 0.3727 data: 0.0002 max mem: 52148 +Epoch: [15] [3230/7269] eta: 0:27:58 lr: 0.000007 loss: 0.5432 (0.5298) time: 0.3728 data: 0.0002 max mem: 52148 +Epoch: [15] [3240/7269] eta: 0:27:53 lr: 0.000007 loss: 0.5555 (0.5300) time: 0.3727 data: 0.0001 max mem: 52148 +Epoch: [15] [3250/7269] eta: 0:27:50 lr: 0.000007 loss: 0.5500 (0.5299) time: 0.4170 data: 0.0002 max mem: 52148 +Epoch: [15] [3260/7269] eta: 0:27:49 lr: 0.000007 loss: 0.5267 (0.5300) time: 0.5709 data: 0.0002 max mem: 52148 +Epoch: [15] [3270/7269] eta: 0:27:44 lr: 0.000007 loss: 0.5397 (0.5300) time: 0.5262 data: 0.0002 max mem: 52148 +Epoch: [15] [3280/7269] eta: 0:27:39 lr: 0.000007 loss: 0.5012 (0.5299) time: 0.3718 data: 0.0002 max mem: 52148 +Epoch: [15] [3290/7269] eta: 0:27:36 lr: 0.000007 loss: 0.5169 (0.5300) time: 0.4311 data: 0.0002 max mem: 52148 +Epoch: [15] [3300/7269] eta: 0:27:32 lr: 0.000007 loss: 0.5497 (0.5300) time: 0.4768 data: 0.0002 max mem: 52148 +Epoch: [15] [3310/7269] eta: 0:27:28 lr: 0.000007 loss: 0.5618 (0.5301) time: 0.4177 data: 0.0001 max mem: 52148 +Epoch: [15] [3320/7269] eta: 0:27:24 lr: 0.000007 loss: 0.5457 (0.5301) time: 0.4168 data: 0.0001 max mem: 52148 +Epoch: [15] [3330/7269] eta: 0:27:19 lr: 0.000007 loss: 0.5117 (0.5300) time: 0.4173 data: 0.0001 max mem: 52148 +Epoch: [15] [3340/7269] eta: 0:27:15 lr: 0.000007 loss: 0.4623 (0.5299) time: 0.3728 data: 0.0001 max mem: 52148 +Epoch: [15] [3350/7269] eta: 0:27:12 lr: 0.000007 loss: 0.5478 (0.5300) time: 0.4337 data: 0.0002 max mem: 52148 +Epoch: [15] [3360/7269] eta: 0:27:07 lr: 0.000007 loss: 0.5349 (0.5299) time: 0.4336 data: 0.0002 max mem: 52148 +Epoch: [15] [3370/7269] eta: 0:27:02 lr: 0.000007 loss: 0.5155 (0.5299) time: 0.3719 data: 0.0002 max mem: 52148 +Epoch: [15] [3380/7269] eta: 0:27:00 lr: 0.000007 loss: 0.5332 (0.5300) time: 0.4642 data: 0.0001 max mem: 52148 +Epoch: [15] [3390/7269] eta: 0:26:56 lr: 0.000007 loss: 0.5523 (0.5300) time: 0.5099 data: 0.0001 max mem: 52148 +Epoch: [15] [3400/7269] eta: 0:26:51 lr: 0.000007 loss: 0.5311 (0.5300) time: 0.4176 data: 0.0002 max mem: 52148 +Epoch: [15] [3410/7269] eta: 0:26:49 lr: 0.000007 loss: 0.5279 (0.5300) time: 0.4667 data: 0.0002 max mem: 52148 +Epoch: [15] [3420/7269] eta: 0:26:46 lr: 0.000007 loss: 0.5466 (0.5300) time: 0.5285 data: 0.0001 max mem: 52148 +Epoch: [15] [3430/7269] eta: 0:26:42 lr: 0.000007 loss: 0.5162 (0.5299) time: 0.4937 data: 0.0001 max mem: 52148 +Epoch: [15] [3440/7269] eta: 0:26:37 lr: 0.000007 loss: 0.5036 (0.5299) time: 0.4318 data: 0.0001 max mem: 52148 +Epoch: [15] [3450/7269] eta: 0:26:33 lr: 0.000007 loss: 0.5072 (0.5299) time: 0.3725 data: 0.0002 max mem: 52148 +Epoch: [15] [3460/7269] eta: 0:26:28 lr: 0.000007 loss: 0.5436 (0.5299) time: 0.3730 data: 0.0002 max mem: 52148 +Epoch: [15] [3470/7269] eta: 0:26:25 lr: 0.000007 loss: 0.5112 (0.5299) time: 0.4201 data: 0.0001 max mem: 52148 +Epoch: [15] [3480/7269] eta: 0:26:22 lr: 0.000007 loss: 0.5018 (0.5298) time: 0.5291 data: 0.0001 max mem: 52148 +Epoch: [15] [3490/7269] eta: 0:26:18 lr: 0.000007 loss: 0.5082 (0.5298) time: 0.4816 data: 0.0001 max mem: 52148 +Epoch: [15] [3500/7269] eta: 0:26:14 lr: 0.000007 loss: 0.5225 (0.5298) time: 0.4198 data: 0.0001 max mem: 52148 +Epoch: [15] [3510/7269] eta: 0:26:09 lr: 0.000007 loss: 0.5391 (0.5298) time: 0.4199 data: 0.0001 max mem: 52148 +Epoch: [15] [3520/7269] eta: 0:26:07 lr: 0.000007 loss: 0.5234 (0.5298) time: 0.4787 data: 0.0001 max mem: 52148 +Epoch: [15] [3530/7269] eta: 0:26:02 lr: 0.000007 loss: 0.5318 (0.5298) time: 0.4786 data: 0.0001 max mem: 52148 +Epoch: [15] [3540/7269] eta: 0:25:58 lr: 0.000007 loss: 0.5281 (0.5298) time: 0.3721 data: 0.0001 max mem: 52148 +Epoch: [15] [3550/7269] eta: 0:25:54 lr: 0.000007 loss: 0.5036 (0.5297) time: 0.4194 data: 0.0002 max mem: 52148 +Epoch: [15] [3560/7269] eta: 0:25:50 lr: 0.000007 loss: 0.5474 (0.5297) time: 0.4660 data: 0.0002 max mem: 52148 +Epoch: [15] [3570/7269] eta: 0:25:46 lr: 0.000007 loss: 0.5679 (0.5298) time: 0.4186 data: 0.0001 max mem: 52148 +Epoch: [15] [3580/7269] eta: 0:25:42 lr: 0.000007 loss: 0.5549 (0.5299) time: 0.4321 data: 0.0001 max mem: 52148 +Epoch: [15] [3590/7269] eta: 0:25:39 lr: 0.000007 loss: 0.5549 (0.5299) time: 0.4829 data: 0.0001 max mem: 52148 +Epoch: [15] [3600/7269] eta: 0:25:34 lr: 0.000007 loss: 0.5367 (0.5299) time: 0.4227 data: 0.0001 max mem: 52148 +Epoch: [15] [3610/7269] eta: 0:25:29 lr: 0.000007 loss: 0.5328 (0.5298) time: 0.3724 data: 0.0001 max mem: 52148 +Epoch: [15] [3620/7269] eta: 0:25:25 lr: 0.000007 loss: 0.5424 (0.5299) time: 0.3725 data: 0.0001 max mem: 52148 +Epoch: [15] [3630/7269] eta: 0:25:20 lr: 0.000007 loss: 0.5424 (0.5300) time: 0.3721 data: 0.0001 max mem: 52148 +Epoch: [15] [3640/7269] eta: 0:25:15 lr: 0.000007 loss: 0.5337 (0.5299) time: 0.3723 data: 0.0001 max mem: 52148 +Epoch: [15] [3650/7269] eta: 0:25:11 lr: 0.000007 loss: 0.5445 (0.5300) time: 0.3726 data: 0.0001 max mem: 52148 +Epoch: [15] [3660/7269] eta: 0:25:07 lr: 0.000007 loss: 0.5445 (0.5300) time: 0.4180 data: 0.0001 max mem: 52148 +Epoch: [15] [3670/7269] eta: 0:25:02 lr: 0.000007 loss: 0.5633 (0.5300) time: 0.4175 data: 0.0001 max mem: 52148 +Epoch: [15] [3680/7269] eta: 0:24:58 lr: 0.000007 loss: 0.5382 (0.5300) time: 0.3732 data: 0.0001 max mem: 52148 +Epoch: [15] [3690/7269] eta: 0:24:53 lr: 0.000007 loss: 0.5375 (0.5300) time: 0.3738 data: 0.0001 max mem: 52148 +Epoch: [15] [3700/7269] eta: 0:24:49 lr: 0.000007 loss: 0.5384 (0.5300) time: 0.3727 data: 0.0001 max mem: 52148 +Epoch: [15] [3710/7269] eta: 0:24:44 lr: 0.000007 loss: 0.5408 (0.5301) time: 0.3726 data: 0.0001 max mem: 52148 +Epoch: [15] [3720/7269] eta: 0:24:41 lr: 0.000007 loss: 0.5421 (0.5301) time: 0.4364 data: 0.0002 max mem: 52148 +Epoch: [15] [3730/7269] eta: 0:24:36 lr: 0.000007 loss: 0.5416 (0.5301) time: 0.4364 data: 0.0002 max mem: 52148 +Epoch: [15] [3740/7269] eta: 0:24:32 lr: 0.000007 loss: 0.5585 (0.5302) time: 0.4179 data: 0.0002 max mem: 52148 +Epoch: [15] [3750/7269] eta: 0:24:28 lr: 0.000007 loss: 0.5487 (0.5302) time: 0.4176 data: 0.0002 max mem: 52148 +Epoch: [15] [3760/7269] eta: 0:24:23 lr: 0.000007 loss: 0.4951 (0.5302) time: 0.3720 data: 0.0001 max mem: 52148 +Epoch: [15] [3770/7269] eta: 0:24:18 lr: 0.000007 loss: 0.5651 (0.5302) time: 0.3718 data: 0.0001 max mem: 52148 +Epoch: [15] [3780/7269] eta: 0:24:14 lr: 0.000007 loss: 0.5533 (0.5302) time: 0.3717 data: 0.0001 max mem: 52148 +Epoch: [15] [3790/7269] eta: 0:24:11 lr: 0.000007 loss: 0.5316 (0.5302) time: 0.4794 data: 0.0001 max mem: 52148 +Epoch: [15] [3800/7269] eta: 0:24:09 lr: 0.000007 loss: 0.5202 (0.5302) time: 0.6020 data: 0.0002 max mem: 52148 +Epoch: [15] [3810/7269] eta: 0:24:05 lr: 0.000007 loss: 0.5202 (0.5302) time: 0.5579 data: 0.0002 max mem: 52148 +Epoch: [15] [3820/7269] eta: 0:24:01 lr: 0.000007 loss: 0.5292 (0.5301) time: 0.4355 data: 0.0001 max mem: 52148 +Epoch: [15] [3830/7269] eta: 0:23:57 lr: 0.000007 loss: 0.5460 (0.5301) time: 0.4181 data: 0.0002 max mem: 52148 +Epoch: [15] [3840/7269] eta: 0:23:53 lr: 0.000007 loss: 0.5421 (0.5301) time: 0.4652 data: 0.0002 max mem: 52148 +Epoch: [15] [3850/7269] eta: 0:23:49 lr: 0.000007 loss: 0.5421 (0.5301) time: 0.4192 data: 0.0002 max mem: 52148 +Epoch: [15] [3860/7269] eta: 0:23:45 lr: 0.000007 loss: 0.5472 (0.5300) time: 0.4186 data: 0.0002 max mem: 52148 +Epoch: [15] [3870/7269] eta: 0:23:40 lr: 0.000007 loss: 0.5333 (0.5301) time: 0.4187 data: 0.0001 max mem: 52148 +Epoch: [15] [3880/7269] eta: 0:23:37 lr: 0.000007 loss: 0.5333 (0.5301) time: 0.4330 data: 0.0001 max mem: 52148 +Epoch: [15] [3890/7269] eta: 0:23:34 lr: 0.000007 loss: 0.5326 (0.5301) time: 0.5266 data: 0.0001 max mem: 52148 +Epoch: [15] [3900/7269] eta: 0:23:29 lr: 0.000007 loss: 0.5322 (0.5301) time: 0.4657 data: 0.0002 max mem: 52148 +Epoch: [15] [3910/7269] eta: 0:23:26 lr: 0.000007 loss: 0.5287 (0.5301) time: 0.4339 data: 0.0001 max mem: 52148 +Epoch: [15] [3920/7269] eta: 0:23:21 lr: 0.000007 loss: 0.5329 (0.5301) time: 0.4340 data: 0.0001 max mem: 52148 +Epoch: [15] [3930/7269] eta: 0:23:17 lr: 0.000007 loss: 0.5164 (0.5300) time: 0.3721 data: 0.0001 max mem: 52148 +Epoch: [15] [3940/7269] eta: 0:23:13 lr: 0.000007 loss: 0.4970 (0.5300) time: 0.4328 data: 0.0001 max mem: 52148 +Epoch: [15] [3950/7269] eta: 0:23:09 lr: 0.000007 loss: 0.5192 (0.5299) time: 0.4329 data: 0.0001 max mem: 52148 +Epoch: [15] [3960/7269] eta: 0:23:05 lr: 0.000007 loss: 0.5276 (0.5300) time: 0.4177 data: 0.0001 max mem: 52148 +Epoch: [15] [3970/7269] eta: 0:23:00 lr: 0.000007 loss: 0.5434 (0.5300) time: 0.4175 data: 0.0001 max mem: 52148 +Epoch: [15] [3980/7269] eta: 0:22:56 lr: 0.000007 loss: 0.5504 (0.5301) time: 0.3717 data: 0.0001 max mem: 52148 +Epoch: [15] [3990/7269] eta: 0:22:52 lr: 0.000007 loss: 0.5225 (0.5301) time: 0.4189 data: 0.0001 max mem: 52148 +Epoch: [15] [4000/7269] eta: 0:22:48 lr: 0.000007 loss: 0.5225 (0.5300) time: 0.4649 data: 0.0001 max mem: 52148 +Epoch: [15] [4010/7269] eta: 0:22:43 lr: 0.000007 loss: 0.5274 (0.5300) time: 0.4182 data: 0.0001 max mem: 52148 +Epoch: [15] [4020/7269] eta: 0:22:40 lr: 0.000007 loss: 0.5274 (0.5300) time: 0.4647 data: 0.0001 max mem: 52148 +Epoch: [15] [4030/7269] eta: 0:22:37 lr: 0.000007 loss: 0.5210 (0.5300) time: 0.5132 data: 0.0001 max mem: 52148 +Epoch: [15] [4040/7269] eta: 0:22:32 lr: 0.000007 loss: 0.5285 (0.5300) time: 0.4210 data: 0.0001 max mem: 52148 +Epoch: [15] [4050/7269] eta: 0:22:29 lr: 0.000007 loss: 0.5605 (0.5301) time: 0.4809 data: 0.0001 max mem: 52148 +Epoch: [15] [4060/7269] eta: 0:22:26 lr: 0.000007 loss: 0.5699 (0.5301) time: 0.5434 data: 0.0001 max mem: 52148 +Epoch: [15] [4070/7269] eta: 0:22:21 lr: 0.000007 loss: 0.4807 (0.5300) time: 0.4347 data: 0.0001 max mem: 52148 +Epoch: [15] [4080/7269] eta: 0:22:16 lr: 0.000007 loss: 0.5189 (0.5300) time: 0.3726 data: 0.0001 max mem: 52148 +Epoch: [15] [4090/7269] eta: 0:22:13 lr: 0.000007 loss: 0.4827 (0.5299) time: 0.4211 data: 0.0001 max mem: 52148 +Epoch: [15] [4100/7269] eta: 0:22:08 lr: 0.000007 loss: 0.4706 (0.5298) time: 0.4212 data: 0.0001 max mem: 52148 +Epoch: [15] [4110/7269] eta: 0:22:04 lr: 0.000007 loss: 0.5181 (0.5298) time: 0.4207 data: 0.0001 max mem: 52148 +Epoch: [15] [4120/7269] eta: 0:22:00 lr: 0.000007 loss: 0.5185 (0.5297) time: 0.4207 data: 0.0001 max mem: 52148 +Epoch: [15] [4130/7269] eta: 0:21:55 lr: 0.000007 loss: 0.5394 (0.5298) time: 0.3730 data: 0.0002 max mem: 52148 +Epoch: [15] [4140/7269] eta: 0:21:51 lr: 0.000007 loss: 0.5455 (0.5297) time: 0.3729 data: 0.0002 max mem: 52148 +Epoch: [15] [4150/7269] eta: 0:21:46 lr: 0.000007 loss: 0.5060 (0.5297) time: 0.3724 data: 0.0001 max mem: 52148 +Epoch: [15] [4160/7269] eta: 0:21:42 lr: 0.000007 loss: 0.5141 (0.5297) time: 0.3719 data: 0.0001 max mem: 52148 +Epoch: [15] [4170/7269] eta: 0:21:37 lr: 0.000007 loss: 0.5630 (0.5298) time: 0.3721 data: 0.0001 max mem: 52148 +Epoch: [15] [4180/7269] eta: 0:21:33 lr: 0.000007 loss: 0.5691 (0.5299) time: 0.3724 data: 0.0002 max mem: 52148 +Epoch: [15] [4190/7269] eta: 0:21:28 lr: 0.000007 loss: 0.5669 (0.5299) time: 0.3720 data: 0.0002 max mem: 52148 +Epoch: [15] [4200/7269] eta: 0:21:25 lr: 0.000007 loss: 0.5155 (0.5299) time: 0.4789 data: 0.0001 max mem: 52148 +Epoch: [15] [4210/7269] eta: 0:21:22 lr: 0.000007 loss: 0.5520 (0.5299) time: 0.5843 data: 0.0002 max mem: 52148 +Epoch: [15] [4220/7269] eta: 0:21:18 lr: 0.000007 loss: 0.5520 (0.5299) time: 0.5236 data: 0.0002 max mem: 52148 +Epoch: [15] [4230/7269] eta: 0:21:14 lr: 0.000007 loss: 0.5250 (0.5298) time: 0.4185 data: 0.0001 max mem: 52148 +Epoch: [15] [4240/7269] eta: 0:21:09 lr: 0.000007 loss: 0.5187 (0.5298) time: 0.3729 data: 0.0001 max mem: 52148 +Epoch: [15] [4250/7269] eta: 0:21:05 lr: 0.000007 loss: 0.5280 (0.5299) time: 0.4336 data: 0.0002 max mem: 52148 +Epoch: [15] [4260/7269] eta: 0:21:02 lr: 0.000007 loss: 0.5552 (0.5299) time: 0.4954 data: 0.0002 max mem: 52148 +Epoch: [15] [4270/7269] eta: 0:20:57 lr: 0.000007 loss: 0.5061 (0.5298) time: 0.4344 data: 0.0002 max mem: 52148 +Epoch: [15] [4280/7269] eta: 0:20:53 lr: 0.000007 loss: 0.5222 (0.5298) time: 0.3717 data: 0.0001 max mem: 52148 +Epoch: [15] [4290/7269] eta: 0:20:48 lr: 0.000007 loss: 0.5629 (0.5298) time: 0.3714 data: 0.0001 max mem: 52148 +Epoch: [15] [4300/7269] eta: 0:20:44 lr: 0.000007 loss: 0.5442 (0.5298) time: 0.3720 data: 0.0001 max mem: 52148 +Epoch: [15] [4310/7269] eta: 0:20:39 lr: 0.000007 loss: 0.5314 (0.5299) time: 0.3731 data: 0.0001 max mem: 52148 +Epoch: [15] [4320/7269] eta: 0:20:36 lr: 0.000007 loss: 0.5845 (0.5300) time: 0.4841 data: 0.0002 max mem: 52148 +Epoch: [15] [4330/7269] eta: 0:20:33 lr: 0.000007 loss: 0.5487 (0.5299) time: 0.5463 data: 0.0002 max mem: 52148 +Epoch: [15] [4340/7269] eta: 0:20:29 lr: 0.000007 loss: 0.5315 (0.5299) time: 0.4812 data: 0.0001 max mem: 52148 +Epoch: [15] [4350/7269] eta: 0:20:24 lr: 0.000007 loss: 0.4918 (0.5298) time: 0.4185 data: 0.0001 max mem: 52148 +Epoch: [15] [4360/7269] eta: 0:20:21 lr: 0.000007 loss: 0.4967 (0.5298) time: 0.4648 data: 0.0002 max mem: 52148 +Epoch: [15] [4370/7269] eta: 0:20:16 lr: 0.000007 loss: 0.5354 (0.5298) time: 0.4647 data: 0.0002 max mem: 52148 +Epoch: [15] [4380/7269] eta: 0:20:12 lr: 0.000007 loss: 0.5550 (0.5298) time: 0.3725 data: 0.0002 max mem: 52148 +Epoch: [15] [4390/7269] eta: 0:20:07 lr: 0.000007 loss: 0.5070 (0.5298) time: 0.3724 data: 0.0002 max mem: 52148 +Epoch: [15] [4400/7269] eta: 0:20:04 lr: 0.000007 loss: 0.5394 (0.5299) time: 0.4321 data: 0.0001 max mem: 52148 +Epoch: [15] [4410/7269] eta: 0:20:01 lr: 0.000007 loss: 0.5574 (0.5299) time: 0.5826 data: 0.0001 max mem: 52148 +Epoch: [15] [4420/7269] eta: 0:19:57 lr: 0.000007 loss: 0.5508 (0.5299) time: 0.5230 data: 0.0002 max mem: 52148 +Epoch: [15] [4430/7269] eta: 0:19:53 lr: 0.000007 loss: 0.5311 (0.5298) time: 0.4209 data: 0.0002 max mem: 52148 +Epoch: [15] [4440/7269] eta: 0:19:48 lr: 0.000007 loss: 0.5579 (0.5299) time: 0.4206 data: 0.0002 max mem: 52148 +Epoch: [15] [4450/7269] eta: 0:19:44 lr: 0.000007 loss: 0.5804 (0.5300) time: 0.3722 data: 0.0002 max mem: 52148 +Epoch: [15] [4460/7269] eta: 0:19:40 lr: 0.000007 loss: 0.4938 (0.5299) time: 0.4182 data: 0.0002 max mem: 52148 +Epoch: [15] [4470/7269] eta: 0:19:35 lr: 0.000007 loss: 0.5580 (0.5299) time: 0.4185 data: 0.0002 max mem: 52148 +Epoch: [15] [4480/7269] eta: 0:19:32 lr: 0.000007 loss: 0.5677 (0.5300) time: 0.4610 data: 0.0002 max mem: 52148 +Epoch: [15] [4490/7269] eta: 0:19:27 lr: 0.000007 loss: 0.5272 (0.5299) time: 0.4607 data: 0.0002 max mem: 52148 +Epoch: [15] [4500/7269] eta: 0:19:24 lr: 0.000007 loss: 0.5081 (0.5299) time: 0.4952 data: 0.0002 max mem: 52148 +Epoch: [15] [4510/7269] eta: 0:19:20 lr: 0.000007 loss: 0.5264 (0.5300) time: 0.4948 data: 0.0002 max mem: 52148 +Epoch: [15] [4520/7269] eta: 0:19:15 lr: 0.000007 loss: 0.5582 (0.5299) time: 0.3721 data: 0.0002 max mem: 52148 +Epoch: [15] [4530/7269] eta: 0:19:11 lr: 0.000007 loss: 0.5402 (0.5299) time: 0.4198 data: 0.0002 max mem: 52148 +Epoch: [15] [4540/7269] eta: 0:19:07 lr: 0.000007 loss: 0.5510 (0.5300) time: 0.4194 data: 0.0001 max mem: 52148 +Epoch: [15] [4550/7269] eta: 0:19:02 lr: 0.000007 loss: 0.5337 (0.5300) time: 0.3719 data: 0.0001 max mem: 52148 +Epoch: [15] [4560/7269] eta: 0:18:58 lr: 0.000007 loss: 0.5337 (0.5300) time: 0.4181 data: 0.0001 max mem: 52148 +Epoch: [15] [4570/7269] eta: 0:18:54 lr: 0.000007 loss: 0.5619 (0.5300) time: 0.4180 data: 0.0001 max mem: 52148 +Epoch: [15] [4580/7269] eta: 0:18:50 lr: 0.000007 loss: 0.5619 (0.5301) time: 0.4194 data: 0.0001 max mem: 52148 +Epoch: [15] [4590/7269] eta: 0:18:46 lr: 0.000007 loss: 0.5523 (0.5301) time: 0.4198 data: 0.0001 max mem: 52148 +Epoch: [15] [4600/7269] eta: 0:18:41 lr: 0.000007 loss: 0.5349 (0.5301) time: 0.3727 data: 0.0002 max mem: 52148 +Epoch: [15] [4610/7269] eta: 0:18:37 lr: 0.000007 loss: 0.5161 (0.5301) time: 0.3729 data: 0.0002 max mem: 52148 +Epoch: [15] [4620/7269] eta: 0:18:32 lr: 0.000007 loss: 0.5206 (0.5301) time: 0.3726 data: 0.0001 max mem: 52148 +Epoch: [15] [4630/7269] eta: 0:18:28 lr: 0.000007 loss: 0.5625 (0.5301) time: 0.3724 data: 0.0002 max mem: 52148 +Epoch: [15] [4640/7269] eta: 0:18:23 lr: 0.000007 loss: 0.5593 (0.5302) time: 0.3722 data: 0.0001 max mem: 52148 +Epoch: [15] [4650/7269] eta: 0:18:19 lr: 0.000007 loss: 0.5364 (0.5302) time: 0.3722 data: 0.0001 max mem: 52148 +Epoch: [15] [4660/7269] eta: 0:18:15 lr: 0.000007 loss: 0.5412 (0.5302) time: 0.4190 data: 0.0001 max mem: 52148 +Epoch: [15] [4670/7269] eta: 0:18:12 lr: 0.000007 loss: 0.5667 (0.5303) time: 0.5287 data: 0.0001 max mem: 52148 +Epoch: [15] [4680/7269] eta: 0:18:08 lr: 0.000007 loss: 0.5723 (0.5303) time: 0.5421 data: 0.0001 max mem: 52148 +Epoch: [15] [4690/7269] eta: 0:18:04 lr: 0.000007 loss: 0.5495 (0.5303) time: 0.4928 data: 0.0001 max mem: 52148 +Epoch: [15] [4700/7269] eta: 0:17:59 lr: 0.000007 loss: 0.5495 (0.5304) time: 0.4331 data: 0.0001 max mem: 52148 +Epoch: [15] [4710/7269] eta: 0:17:56 lr: 0.000007 loss: 0.5401 (0.5304) time: 0.4196 data: 0.0001 max mem: 52148 +Epoch: [15] [4720/7269] eta: 0:17:51 lr: 0.000007 loss: 0.5571 (0.5305) time: 0.4191 data: 0.0001 max mem: 52148 +Epoch: [15] [4730/7269] eta: 0:17:47 lr: 0.000007 loss: 0.5571 (0.5304) time: 0.4357 data: 0.0001 max mem: 52148 +Epoch: [15] [4740/7269] eta: 0:17:43 lr: 0.000007 loss: 0.5251 (0.5305) time: 0.4366 data: 0.0001 max mem: 52148 +Epoch: [15] [4750/7269] eta: 0:17:39 lr: 0.000007 loss: 0.5660 (0.5305) time: 0.4179 data: 0.0001 max mem: 52148 +Epoch: [15] [4760/7269] eta: 0:17:35 lr: 0.000007 loss: 0.5262 (0.5304) time: 0.4645 data: 0.0001 max mem: 52148 +Epoch: [15] [4770/7269] eta: 0:17:30 lr: 0.000007 loss: 0.5186 (0.5304) time: 0.4203 data: 0.0001 max mem: 52148 +Epoch: [15] [4780/7269] eta: 0:17:26 lr: 0.000007 loss: 0.5304 (0.5304) time: 0.3728 data: 0.0002 max mem: 52148 +Epoch: [15] [4790/7269] eta: 0:17:22 lr: 0.000007 loss: 0.5350 (0.5304) time: 0.3726 data: 0.0001 max mem: 52148 +Epoch: [15] [4800/7269] eta: 0:17:18 lr: 0.000007 loss: 0.5366 (0.5304) time: 0.4329 data: 0.0001 max mem: 52148 +Epoch: [15] [4810/7269] eta: 0:17:14 lr: 0.000007 loss: 0.5122 (0.5304) time: 0.4774 data: 0.0002 max mem: 52148 +Epoch: [15] [4820/7269] eta: 0:17:10 lr: 0.000007 loss: 0.5270 (0.5304) time: 0.4615 data: 0.0002 max mem: 52148 +Epoch: [15] [4830/7269] eta: 0:17:06 lr: 0.000007 loss: 0.5270 (0.5303) time: 0.4804 data: 0.0001 max mem: 52148 +Epoch: [15] [4840/7269] eta: 0:17:02 lr: 0.000007 loss: 0.5283 (0.5304) time: 0.5000 data: 0.0001 max mem: 52148 +Epoch: [15] [4850/7269] eta: 0:16:59 lr: 0.000007 loss: 0.5454 (0.5303) time: 0.5430 data: 0.0001 max mem: 52148 +Epoch: [15] [4860/7269] eta: 0:16:55 lr: 0.000007 loss: 0.4911 (0.5302) time: 0.5241 data: 0.0001 max mem: 52148 +Epoch: [15] [4870/7269] eta: 0:16:51 lr: 0.000007 loss: 0.4682 (0.5302) time: 0.4672 data: 0.0001 max mem: 52148 +Epoch: [15] [4880/7269] eta: 0:16:46 lr: 0.000007 loss: 0.4910 (0.5301) time: 0.4223 data: 0.0001 max mem: 52148 +Epoch: [15] [4890/7269] eta: 0:16:42 lr: 0.000007 loss: 0.5181 (0.5301) time: 0.3726 data: 0.0001 max mem: 52148 +Epoch: [15] [4900/7269] eta: 0:16:38 lr: 0.000007 loss: 0.4995 (0.5300) time: 0.4779 data: 0.0002 max mem: 52148 +Epoch: [15] [4910/7269] eta: 0:16:34 lr: 0.000007 loss: 0.5455 (0.5300) time: 0.5221 data: 0.0002 max mem: 52148 +Epoch: [15] [4920/7269] eta: 0:16:30 lr: 0.000007 loss: 0.5484 (0.5301) time: 0.4175 data: 0.0001 max mem: 52148 +Epoch: [15] [4930/7269] eta: 0:16:26 lr: 0.000007 loss: 0.5416 (0.5301) time: 0.4215 data: 0.0001 max mem: 52148 +Epoch: [15] [4940/7269] eta: 0:16:21 lr: 0.000007 loss: 0.5386 (0.5301) time: 0.4214 data: 0.0001 max mem: 52148 +Epoch: [15] [4950/7269] eta: 0:16:17 lr: 0.000007 loss: 0.5429 (0.5301) time: 0.3730 data: 0.0001 max mem: 52148 +Epoch: [15] [4960/7269] eta: 0:16:13 lr: 0.000007 loss: 0.5359 (0.5301) time: 0.3723 data: 0.0002 max mem: 52148 +Epoch: [15] [4970/7269] eta: 0:16:08 lr: 0.000007 loss: 0.5148 (0.5300) time: 0.3725 data: 0.0002 max mem: 52148 +Epoch: [15] [4980/7269] eta: 0:16:04 lr: 0.000007 loss: 0.5450 (0.5302) time: 0.3725 data: 0.0001 max mem: 52148 +Epoch: [15] [4990/7269] eta: 0:16:00 lr: 0.000007 loss: 0.5750 (0.5301) time: 0.4335 data: 0.0001 max mem: 52148 +Epoch: [15] [5000/7269] eta: 0:15:55 lr: 0.000007 loss: 0.5466 (0.5302) time: 0.4336 data: 0.0001 max mem: 52148 +Epoch: [15] [5010/7269] eta: 0:15:51 lr: 0.000007 loss: 0.5416 (0.5302) time: 0.4183 data: 0.0001 max mem: 52148 +Epoch: [15] [5020/7269] eta: 0:15:47 lr: 0.000007 loss: 0.5447 (0.5302) time: 0.4637 data: 0.0001 max mem: 52148 +Epoch: [15] [5030/7269] eta: 0:15:44 lr: 0.000007 loss: 0.5618 (0.5302) time: 0.5257 data: 0.0001 max mem: 52148 +Epoch: [15] [5040/7269] eta: 0:15:39 lr: 0.000007 loss: 0.5356 (0.5302) time: 0.4805 data: 0.0001 max mem: 52148 +Epoch: [15] [5050/7269] eta: 0:15:35 lr: 0.000007 loss: 0.5311 (0.5301) time: 0.3727 data: 0.0001 max mem: 52148 +Epoch: [15] [5060/7269] eta: 0:15:31 lr: 0.000007 loss: 0.5281 (0.5302) time: 0.3723 data: 0.0001 max mem: 52148 +Epoch: [15] [5070/7269] eta: 0:15:27 lr: 0.000007 loss: 0.5160 (0.5301) time: 0.4339 data: 0.0001 max mem: 52148 +Epoch: [15] [5080/7269] eta: 0:15:23 lr: 0.000007 loss: 0.4943 (0.5301) time: 0.4819 data: 0.0002 max mem: 52148 +Epoch: [15] [5090/7269] eta: 0:15:19 lr: 0.000007 loss: 0.5524 (0.5301) time: 0.4687 data: 0.0002 max mem: 52148 +Epoch: [15] [5100/7269] eta: 0:15:14 lr: 0.000007 loss: 0.5359 (0.5300) time: 0.4211 data: 0.0001 max mem: 52148 +Epoch: [15] [5110/7269] eta: 0:15:10 lr: 0.000007 loss: 0.4900 (0.5300) time: 0.3725 data: 0.0001 max mem: 52148 +Epoch: [15] [5120/7269] eta: 0:15:05 lr: 0.000007 loss: 0.5094 (0.5300) time: 0.3725 data: 0.0001 max mem: 52148 +Epoch: [15] [5130/7269] eta: 0:15:01 lr: 0.000007 loss: 0.5334 (0.5300) time: 0.3725 data: 0.0001 max mem: 52148 +Epoch: [15] [5140/7269] eta: 0:14:57 lr: 0.000007 loss: 0.5365 (0.5301) time: 0.3723 data: 0.0001 max mem: 52148 +Epoch: [15] [5150/7269] eta: 0:14:53 lr: 0.000007 loss: 0.5455 (0.5301) time: 0.4185 data: 0.0001 max mem: 52148 +Epoch: [15] [5160/7269] eta: 0:14:48 lr: 0.000007 loss: 0.5455 (0.5301) time: 0.4648 data: 0.0001 max mem: 52148 +Epoch: [15] [5170/7269] eta: 0:14:44 lr: 0.000007 loss: 0.5472 (0.5301) time: 0.4188 data: 0.0001 max mem: 52148 +Epoch: [15] [5180/7269] eta: 0:14:40 lr: 0.000007 loss: 0.5751 (0.5301) time: 0.4188 data: 0.0001 max mem: 52148 +Epoch: [15] [5190/7269] eta: 0:14:36 lr: 0.000007 loss: 0.5379 (0.5301) time: 0.4648 data: 0.0001 max mem: 52148 +Epoch: [15] [5200/7269] eta: 0:14:32 lr: 0.000007 loss: 0.5240 (0.5301) time: 0.4187 data: 0.0002 max mem: 52148 +Epoch: [15] [5210/7269] eta: 0:14:27 lr: 0.000007 loss: 0.5485 (0.5302) time: 0.3726 data: 0.0002 max mem: 52148 +Epoch: [15] [5220/7269] eta: 0:14:24 lr: 0.000007 loss: 0.5485 (0.5302) time: 0.4964 data: 0.0002 max mem: 52148 +Epoch: [15] [5230/7269] eta: 0:14:19 lr: 0.000007 loss: 0.5518 (0.5302) time: 0.4964 data: 0.0002 max mem: 52148 +Epoch: [15] [5240/7269] eta: 0:14:15 lr: 0.000007 loss: 0.5600 (0.5303) time: 0.3723 data: 0.0001 max mem: 52148 +Epoch: [15] [5250/7269] eta: 0:14:10 lr: 0.000007 loss: 0.5716 (0.5303) time: 0.3724 data: 0.0002 max mem: 52148 +Epoch: [15] [5260/7269] eta: 0:14:06 lr: 0.000007 loss: 0.5586 (0.5303) time: 0.3729 data: 0.0002 max mem: 52148 +Epoch: [15] [5270/7269] eta: 0:14:02 lr: 0.000007 loss: 0.5378 (0.5303) time: 0.4216 data: 0.0001 max mem: 52148 +Epoch: [15] [5280/7269] eta: 0:13:58 lr: 0.000007 loss: 0.5468 (0.5303) time: 0.4218 data: 0.0002 max mem: 52148 +Epoch: [15] [5290/7269] eta: 0:13:54 lr: 0.000007 loss: 0.5438 (0.5303) time: 0.4167 data: 0.0002 max mem: 52148 +Epoch: [15] [5300/7269] eta: 0:13:50 lr: 0.000007 loss: 0.5694 (0.5304) time: 0.4617 data: 0.0002 max mem: 52148 +Epoch: [15] [5310/7269] eta: 0:13:46 lr: 0.000007 loss: 0.5618 (0.5304) time: 0.4768 data: 0.0002 max mem: 52148 +Epoch: [15] [5320/7269] eta: 0:13:42 lr: 0.000007 loss: 0.4968 (0.5303) time: 0.4921 data: 0.0002 max mem: 52148 +Epoch: [15] [5330/7269] eta: 0:13:38 lr: 0.000007 loss: 0.5039 (0.5303) time: 0.4963 data: 0.0002 max mem: 52148 +Epoch: [15] [5340/7269] eta: 0:13:33 lr: 0.000007 loss: 0.5219 (0.5304) time: 0.4363 data: 0.0002 max mem: 52148 +Epoch: [15] [5350/7269] eta: 0:13:29 lr: 0.000007 loss: 0.5451 (0.5304) time: 0.3729 data: 0.0002 max mem: 52148 +Epoch: [15] [5360/7269] eta: 0:13:25 lr: 0.000007 loss: 0.5451 (0.5303) time: 0.4351 data: 0.0002 max mem: 52148 +Epoch: [15] [5370/7269] eta: 0:13:21 lr: 0.000007 loss: 0.5030 (0.5303) time: 0.4358 data: 0.0002 max mem: 52148 +Epoch: [15] [5380/7269] eta: 0:13:16 lr: 0.000007 loss: 0.5370 (0.5303) time: 0.3728 data: 0.0002 max mem: 52148 +Epoch: [15] [5390/7269] eta: 0:13:12 lr: 0.000007 loss: 0.5232 (0.5302) time: 0.4172 data: 0.0002 max mem: 52148 +Epoch: [15] [5400/7269] eta: 0:13:08 lr: 0.000007 loss: 0.5387 (0.5303) time: 0.4801 data: 0.0002 max mem: 52148 +Epoch: [15] [5410/7269] eta: 0:13:04 lr: 0.000007 loss: 0.5535 (0.5303) time: 0.4804 data: 0.0002 max mem: 52148 +Epoch: [15] [5420/7269] eta: 0:13:00 lr: 0.000007 loss: 0.5535 (0.5304) time: 0.5162 data: 0.0002 max mem: 52148 +Epoch: [15] [5430/7269] eta: 0:12:56 lr: 0.000007 loss: 0.5558 (0.5303) time: 0.4711 data: 0.0002 max mem: 52148 +Epoch: [15] [5440/7269] eta: 0:12:52 lr: 0.000007 loss: 0.5106 (0.5303) time: 0.3731 data: 0.0002 max mem: 52148 +Epoch: [15] [5450/7269] eta: 0:12:48 lr: 0.000007 loss: 0.5637 (0.5304) time: 0.4350 data: 0.0002 max mem: 52148 +Epoch: [15] [5460/7269] eta: 0:12:44 lr: 0.000007 loss: 0.5637 (0.5304) time: 0.4993 data: 0.0002 max mem: 52148 +Epoch: [15] [5470/7269] eta: 0:12:39 lr: 0.000007 loss: 0.5555 (0.5304) time: 0.4380 data: 0.0002 max mem: 52148 +Epoch: [15] [5480/7269] eta: 0:12:35 lr: 0.000007 loss: 0.5505 (0.5304) time: 0.4185 data: 0.0002 max mem: 52148 +Epoch: [15] [5490/7269] eta: 0:12:31 lr: 0.000007 loss: 0.5279 (0.5304) time: 0.4783 data: 0.0002 max mem: 52148 +Epoch: [15] [5500/7269] eta: 0:12:27 lr: 0.000007 loss: 0.5282 (0.5304) time: 0.4773 data: 0.0002 max mem: 52148 +Epoch: [15] [5510/7269] eta: 0:12:23 lr: 0.000007 loss: 0.5282 (0.5305) time: 0.4635 data: 0.0002 max mem: 52148 +Epoch: [15] [5520/7269] eta: 0:12:19 lr: 0.000007 loss: 0.5358 (0.5304) time: 0.4193 data: 0.0002 max mem: 52148 +Epoch: [15] [5530/7269] eta: 0:12:14 lr: 0.000007 loss: 0.5231 (0.5304) time: 0.3721 data: 0.0002 max mem: 52148 +Epoch: [15] [5540/7269] eta: 0:12:10 lr: 0.000007 loss: 0.5305 (0.5304) time: 0.4168 data: 0.0002 max mem: 52148 +Epoch: [15] [5550/7269] eta: 0:12:06 lr: 0.000007 loss: 0.4982 (0.5303) time: 0.4174 data: 0.0002 max mem: 52148 +Epoch: [15] [5560/7269] eta: 0:12:02 lr: 0.000007 loss: 0.4982 (0.5303) time: 0.4344 data: 0.0002 max mem: 52148 +Epoch: [15] [5570/7269] eta: 0:11:57 lr: 0.000007 loss: 0.5460 (0.5303) time: 0.4345 data: 0.0002 max mem: 52148 +Epoch: [15] [5580/7269] eta: 0:11:53 lr: 0.000007 loss: 0.5402 (0.5303) time: 0.3731 data: 0.0002 max mem: 52148 +Epoch: [15] [5590/7269] eta: 0:11:49 lr: 0.000007 loss: 0.5402 (0.5303) time: 0.4341 data: 0.0002 max mem: 52148 +Epoch: [15] [5600/7269] eta: 0:11:45 lr: 0.000007 loss: 0.5237 (0.5302) time: 0.4344 data: 0.0002 max mem: 52148 +Epoch: [15] [5610/7269] eta: 0:11:41 lr: 0.000007 loss: 0.5217 (0.5302) time: 0.4213 data: 0.0002 max mem: 52148 +Epoch: [15] [5620/7269] eta: 0:11:36 lr: 0.000007 loss: 0.5321 (0.5302) time: 0.4210 data: 0.0002 max mem: 52148 +Epoch: [15] [5630/7269] eta: 0:11:32 lr: 0.000007 loss: 0.5532 (0.5303) time: 0.3728 data: 0.0002 max mem: 52148 +Epoch: [15] [5640/7269] eta: 0:11:28 lr: 0.000007 loss: 0.5691 (0.5303) time: 0.4182 data: 0.0002 max mem: 52148 +Epoch: [15] [5650/7269] eta: 0:11:24 lr: 0.000007 loss: 0.5253 (0.5303) time: 0.4631 data: 0.0002 max mem: 52148 +Epoch: [15] [5660/7269] eta: 0:11:20 lr: 0.000007 loss: 0.5406 (0.5303) time: 0.4808 data: 0.0002 max mem: 52148 +Epoch: [15] [5670/7269] eta: 0:11:15 lr: 0.000007 loss: 0.5473 (0.5303) time: 0.4353 data: 0.0002 max mem: 52148 +Epoch: [15] [5680/7269] eta: 0:11:11 lr: 0.000007 loss: 0.5208 (0.5303) time: 0.3721 data: 0.0002 max mem: 52148 +Epoch: [15] [5690/7269] eta: 0:11:06 lr: 0.000007 loss: 0.5208 (0.5303) time: 0.3725 data: 0.0002 max mem: 52148 +Epoch: [15] [5700/7269] eta: 0:11:02 lr: 0.000007 loss: 0.5699 (0.5304) time: 0.3728 data: 0.0002 max mem: 52148 +Epoch: [15] [5710/7269] eta: 0:10:58 lr: 0.000007 loss: 0.5596 (0.5303) time: 0.3731 data: 0.0002 max mem: 52148 +Epoch: [15] [5720/7269] eta: 0:10:53 lr: 0.000007 loss: 0.5344 (0.5304) time: 0.3727 data: 0.0002 max mem: 52148 +Epoch: [15] [5730/7269] eta: 0:10:49 lr: 0.000007 loss: 0.4987 (0.5303) time: 0.3725 data: 0.0002 max mem: 52148 +Epoch: [15] [5740/7269] eta: 0:10:45 lr: 0.000007 loss: 0.4949 (0.5303) time: 0.3725 data: 0.0002 max mem: 52148 +Epoch: [15] [5750/7269] eta: 0:10:40 lr: 0.000007 loss: 0.5281 (0.5303) time: 0.3722 data: 0.0002 max mem: 52148 +Epoch: [15] [5760/7269] eta: 0:10:36 lr: 0.000007 loss: 0.5441 (0.5303) time: 0.3723 data: 0.0002 max mem: 52148 +Epoch: [15] [5770/7269] eta: 0:10:32 lr: 0.000007 loss: 0.5184 (0.5303) time: 0.3724 data: 0.0002 max mem: 52148 +Epoch: [15] [5780/7269] eta: 0:10:28 lr: 0.000007 loss: 0.5076 (0.5302) time: 0.4330 data: 0.0002 max mem: 52148 +Epoch: [15] [5790/7269] eta: 0:10:23 lr: 0.000007 loss: 0.5399 (0.5303) time: 0.4328 data: 0.0002 max mem: 52148 +Epoch: [15] [5800/7269] eta: 0:10:19 lr: 0.000007 loss: 0.5272 (0.5303) time: 0.4844 data: 0.0002 max mem: 52148 +Epoch: [15] [5810/7269] eta: 0:10:16 lr: 0.000007 loss: 0.5272 (0.5303) time: 0.5918 data: 0.0002 max mem: 52148 +Epoch: [15] [5820/7269] eta: 0:10:11 lr: 0.000007 loss: 0.5365 (0.5303) time: 0.4797 data: 0.0002 max mem: 52148 +Epoch: [15] [5830/7269] eta: 0:10:08 lr: 0.000007 loss: 0.5365 (0.5304) time: 0.4801 data: 0.0002 max mem: 52148 +Epoch: [15] [5840/7269] eta: 0:10:03 lr: 0.000007 loss: 0.5256 (0.5304) time: 0.4803 data: 0.0002 max mem: 52148 +Epoch: [15] [5850/7269] eta: 0:09:59 lr: 0.000007 loss: 0.5008 (0.5304) time: 0.4187 data: 0.0002 max mem: 52148 +Epoch: [15] [5860/7269] eta: 0:09:55 lr: 0.000007 loss: 0.5512 (0.5304) time: 0.4186 data: 0.0002 max mem: 52148 +Epoch: [15] [5870/7269] eta: 0:09:50 lr: 0.000007 loss: 0.5639 (0.5305) time: 0.3731 data: 0.0002 max mem: 52148 +Epoch: [15] [5880/7269] eta: 0:09:46 lr: 0.000007 loss: 0.5686 (0.5305) time: 0.4214 data: 0.0002 max mem: 52148 +Epoch: [15] [5890/7269] eta: 0:09:42 lr: 0.000007 loss: 0.5511 (0.5306) time: 0.4693 data: 0.0002 max mem: 52148 +Epoch: [15] [5900/7269] eta: 0:09:38 lr: 0.000007 loss: 0.5157 (0.5305) time: 0.4215 data: 0.0002 max mem: 52148 +Epoch: [15] [5910/7269] eta: 0:09:33 lr: 0.000007 loss: 0.5028 (0.5306) time: 0.3727 data: 0.0002 max mem: 52148 +Epoch: [15] [5920/7269] eta: 0:09:29 lr: 0.000007 loss: 0.5736 (0.5306) time: 0.4344 data: 0.0002 max mem: 52148 +Epoch: [15] [5930/7269] eta: 0:09:25 lr: 0.000007 loss: 0.5590 (0.5306) time: 0.4834 data: 0.0002 max mem: 52148 +Epoch: [15] [5940/7269] eta: 0:09:21 lr: 0.000007 loss: 0.5590 (0.5306) time: 0.4667 data: 0.0002 max mem: 52148 +Epoch: [15] [5950/7269] eta: 0:09:17 lr: 0.000007 loss: 0.5467 (0.5306) time: 0.4687 data: 0.0002 max mem: 52148 +Epoch: [15] [5960/7269] eta: 0:09:13 lr: 0.000007 loss: 0.5245 (0.5305) time: 0.4699 data: 0.0002 max mem: 52148 +Epoch: [15] [5970/7269] eta: 0:09:09 lr: 0.000007 loss: 0.5245 (0.5306) time: 0.4192 data: 0.0002 max mem: 52148 +Epoch: [15] [5980/7269] eta: 0:09:05 lr: 0.000007 loss: 0.5758 (0.5307) time: 0.4367 data: 0.0002 max mem: 52148 +Epoch: [15] [5990/7269] eta: 0:09:00 lr: 0.000007 loss: 0.5494 (0.5307) time: 0.4973 data: 0.0002 max mem: 52148 +Epoch: [15] [6000/7269] eta: 0:08:56 lr: 0.000007 loss: 0.5494 (0.5306) time: 0.4327 data: 0.0002 max mem: 52148 +Epoch: [15] [6010/7269] eta: 0:08:52 lr: 0.000007 loss: 0.5268 (0.5306) time: 0.4159 data: 0.0002 max mem: 52148 +Epoch: [15] [6020/7269] eta: 0:08:48 lr: 0.000007 loss: 0.5194 (0.5306) time: 0.4164 data: 0.0002 max mem: 52148 +Epoch: [15] [6030/7269] eta: 0:08:43 lr: 0.000007 loss: 0.5194 (0.5306) time: 0.3728 data: 0.0002 max mem: 52148 +Epoch: [15] [6040/7269] eta: 0:08:39 lr: 0.000007 loss: 0.5066 (0.5306) time: 0.3724 data: 0.0002 max mem: 52148 +Epoch: [15] [6050/7269] eta: 0:08:35 lr: 0.000007 loss: 0.5194 (0.5305) time: 0.4329 data: 0.0002 max mem: 52148 +Epoch: [15] [6060/7269] eta: 0:08:31 lr: 0.000007 loss: 0.5281 (0.5305) time: 0.4781 data: 0.0002 max mem: 52148 +Epoch: [15] [6070/7269] eta: 0:08:26 lr: 0.000007 loss: 0.5511 (0.5306) time: 0.4178 data: 0.0002 max mem: 52148 +Epoch: [15] [6080/7269] eta: 0:08:22 lr: 0.000007 loss: 0.5559 (0.5306) time: 0.3726 data: 0.0002 max mem: 52148 +Epoch: [15] [6090/7269] eta: 0:08:18 lr: 0.000007 loss: 0.5538 (0.5306) time: 0.4668 data: 0.0002 max mem: 52148 +Epoch: [15] [6100/7269] eta: 0:08:14 lr: 0.000007 loss: 0.5186 (0.5305) time: 0.5576 data: 0.0002 max mem: 52148 +Epoch: [15] [6110/7269] eta: 0:08:10 lr: 0.000007 loss: 0.4863 (0.5305) time: 0.4632 data: 0.0002 max mem: 52148 +Epoch: [15] [6120/7269] eta: 0:08:05 lr: 0.000007 loss: 0.5541 (0.5305) time: 0.3724 data: 0.0002 max mem: 52148 +Epoch: [15] [6130/7269] eta: 0:08:01 lr: 0.000007 loss: 0.5485 (0.5305) time: 0.3721 data: 0.0002 max mem: 52148 +Epoch: [15] [6140/7269] eta: 0:07:57 lr: 0.000007 loss: 0.5022 (0.5305) time: 0.3720 data: 0.0002 max mem: 52148 +Epoch: [15] [6150/7269] eta: 0:07:53 lr: 0.000007 loss: 0.5044 (0.5305) time: 0.3721 data: 0.0002 max mem: 52148 +Epoch: [15] [6160/7269] eta: 0:07:48 lr: 0.000007 loss: 0.5170 (0.5305) time: 0.3730 data: 0.0002 max mem: 52148 +Epoch: [15] [6170/7269] eta: 0:07:44 lr: 0.000007 loss: 0.5170 (0.5305) time: 0.3732 data: 0.0002 max mem: 52148 +Epoch: [15] [6180/7269] eta: 0:07:40 lr: 0.000007 loss: 0.5586 (0.5306) time: 0.3723 data: 0.0002 max mem: 52148 +Epoch: [15] [6190/7269] eta: 0:07:35 lr: 0.000007 loss: 0.5586 (0.5305) time: 0.3722 data: 0.0002 max mem: 52148 +Epoch: [15] [6200/7269] eta: 0:07:31 lr: 0.000007 loss: 0.5276 (0.5305) time: 0.3723 data: 0.0002 max mem: 52148 +Epoch: [15] [6210/7269] eta: 0:07:27 lr: 0.000007 loss: 0.4993 (0.5305) time: 0.4361 data: 0.0002 max mem: 52148 +Epoch: [15] [6220/7269] eta: 0:07:23 lr: 0.000007 loss: 0.4971 (0.5305) time: 0.4988 data: 0.0002 max mem: 52148 +Epoch: [15] [6230/7269] eta: 0:07:18 lr: 0.000007 loss: 0.5122 (0.5305) time: 0.4348 data: 0.0002 max mem: 52148 +Epoch: [15] [6240/7269] eta: 0:07:14 lr: 0.000007 loss: 0.5484 (0.5305) time: 0.3723 data: 0.0002 max mem: 52148 +Epoch: [15] [6250/7269] eta: 0:07:10 lr: 0.000007 loss: 0.5584 (0.5306) time: 0.3724 data: 0.0002 max mem: 52148 +Epoch: [15] [6260/7269] eta: 0:07:06 lr: 0.000007 loss: 0.5545 (0.5306) time: 0.3720 data: 0.0002 max mem: 52148 +Epoch: [15] [6270/7269] eta: 0:07:02 lr: 0.000007 loss: 0.5649 (0.5306) time: 0.4805 data: 0.0002 max mem: 52148 +Epoch: [15] [6280/7269] eta: 0:06:57 lr: 0.000007 loss: 0.5631 (0.5306) time: 0.5406 data: 0.0002 max mem: 52148 +Epoch: [15] [6290/7269] eta: 0:06:53 lr: 0.000007 loss: 0.5519 (0.5307) time: 0.4329 data: 0.0002 max mem: 52148 +Epoch: [15] [6300/7269] eta: 0:06:49 lr: 0.000007 loss: 0.5402 (0.5307) time: 0.4222 data: 0.0002 max mem: 52148 +Epoch: [15] [6310/7269] eta: 0:06:45 lr: 0.000007 loss: 0.5524 (0.5308) time: 0.5273 data: 0.0002 max mem: 52148 +Epoch: [15] [6320/7269] eta: 0:06:41 lr: 0.000007 loss: 0.5632 (0.5307) time: 0.4779 data: 0.0002 max mem: 52148 +Epoch: [15] [6330/7269] eta: 0:06:37 lr: 0.000007 loss: 0.5352 (0.5307) time: 0.4633 data: 0.0002 max mem: 52148 +Epoch: [15] [6340/7269] eta: 0:06:32 lr: 0.000007 loss: 0.5430 (0.5308) time: 0.4634 data: 0.0002 max mem: 52148 +Epoch: [15] [6350/7269] eta: 0:06:28 lr: 0.000007 loss: 0.5492 (0.5308) time: 0.3730 data: 0.0002 max mem: 52148 +Epoch: [15] [6360/7269] eta: 0:06:24 lr: 0.000007 loss: 0.5525 (0.5308) time: 0.3732 data: 0.0002 max mem: 52148 +Epoch: [15] [6370/7269] eta: 0:06:19 lr: 0.000007 loss: 0.5449 (0.5308) time: 0.3723 data: 0.0002 max mem: 52148 +Epoch: [15] [6380/7269] eta: 0:06:15 lr: 0.000007 loss: 0.5482 (0.5308) time: 0.3728 data: 0.0002 max mem: 52148 +Epoch: [15] [6390/7269] eta: 0:06:11 lr: 0.000007 loss: 0.5432 (0.5309) time: 0.3729 data: 0.0002 max mem: 52148 +Epoch: [15] [6400/7269] eta: 0:06:07 lr: 0.000007 loss: 0.5178 (0.5308) time: 0.5280 data: 0.0002 max mem: 52148 +Epoch: [15] [6410/7269] eta: 0:06:03 lr: 0.000007 loss: 0.5178 (0.5308) time: 0.5880 data: 0.0002 max mem: 52148 +Epoch: [15] [6420/7269] eta: 0:05:59 lr: 0.000007 loss: 0.5551 (0.5309) time: 0.5413 data: 0.0002 max mem: 52148 +Epoch: [15] [6430/7269] eta: 0:05:55 lr: 0.000007 loss: 0.5510 (0.5308) time: 0.4814 data: 0.0002 max mem: 52148 +Epoch: [15] [6440/7269] eta: 0:05:50 lr: 0.000007 loss: 0.5353 (0.5309) time: 0.4345 data: 0.0002 max mem: 52148 +Epoch: [15] [6450/7269] eta: 0:05:46 lr: 0.000007 loss: 0.5094 (0.5308) time: 0.4347 data: 0.0002 max mem: 52148 +Epoch: [15] [6460/7269] eta: 0:05:42 lr: 0.000007 loss: 0.4973 (0.5308) time: 0.4336 data: 0.0002 max mem: 52148 +Epoch: [15] [6470/7269] eta: 0:05:38 lr: 0.000007 loss: 0.5211 (0.5308) time: 0.4788 data: 0.0002 max mem: 52148 +Epoch: [15] [6480/7269] eta: 0:05:34 lr: 0.000007 loss: 0.5277 (0.5307) time: 0.4175 data: 0.0002 max mem: 52148 +Epoch: [15] [6490/7269] eta: 0:05:29 lr: 0.000007 loss: 0.5336 (0.5307) time: 0.4313 data: 0.0002 max mem: 52148 +Epoch: [15] [6500/7269] eta: 0:05:25 lr: 0.000007 loss: 0.5336 (0.5307) time: 0.4769 data: 0.0002 max mem: 52148 +Epoch: [15] [6510/7269] eta: 0:05:21 lr: 0.000007 loss: 0.5317 (0.5307) time: 0.4177 data: 0.0002 max mem: 52148 +Epoch: [15] [6520/7269] eta: 0:05:17 lr: 0.000007 loss: 0.5317 (0.5307) time: 0.3722 data: 0.0002 max mem: 52148 +Epoch: [15] [6530/7269] eta: 0:05:12 lr: 0.000007 loss: 0.5313 (0.5307) time: 0.3727 data: 0.0002 max mem: 52148 +Epoch: [15] [6540/7269] eta: 0:05:08 lr: 0.000007 loss: 0.5338 (0.5307) time: 0.3729 data: 0.0002 max mem: 52148 +Epoch: [15] [6550/7269] eta: 0:05:04 lr: 0.000007 loss: 0.5500 (0.5307) time: 0.5258 data: 0.0002 max mem: 52148 +Epoch: [15] [6560/7269] eta: 0:05:00 lr: 0.000007 loss: 0.5248 (0.5307) time: 0.5259 data: 0.0002 max mem: 52148 +Epoch: [15] [6570/7269] eta: 0:04:55 lr: 0.000007 loss: 0.5092 (0.5307) time: 0.3726 data: 0.0002 max mem: 52148 +Epoch: [15] [6580/7269] eta: 0:04:51 lr: 0.000007 loss: 0.5319 (0.5307) time: 0.3723 data: 0.0002 max mem: 52148 +Epoch: [15] [6590/7269] eta: 0:04:47 lr: 0.000007 loss: 0.5264 (0.5308) time: 0.3729 data: 0.0002 max mem: 52148 +Epoch: [15] [6600/7269] eta: 0:04:43 lr: 0.000007 loss: 0.5348 (0.5308) time: 0.3728 data: 0.0002 max mem: 52148 +Epoch: [15] [6610/7269] eta: 0:04:38 lr: 0.000007 loss: 0.5221 (0.5307) time: 0.3721 data: 0.0002 max mem: 52148 +Epoch: [15] [6620/7269] eta: 0:04:34 lr: 0.000007 loss: 0.5449 (0.5308) time: 0.3722 data: 0.0002 max mem: 52148 +Epoch: [15] [6630/7269] eta: 0:04:30 lr: 0.000007 loss: 0.5489 (0.5308) time: 0.3720 data: 0.0002 max mem: 52148 +Epoch: [15] [6640/7269] eta: 0:04:25 lr: 0.000007 loss: 0.5514 (0.5308) time: 0.3719 data: 0.0002 max mem: 52148 +Epoch: [15] [6650/7269] eta: 0:04:21 lr: 0.000007 loss: 0.5399 (0.5308) time: 0.4616 data: 0.0001 max mem: 52148 +Epoch: [15] [6660/7269] eta: 0:04:17 lr: 0.000007 loss: 0.5215 (0.5308) time: 0.4619 data: 0.0002 max mem: 52148 +Epoch: [15] [6670/7269] eta: 0:04:13 lr: 0.000007 loss: 0.5329 (0.5308) time: 0.4182 data: 0.0002 max mem: 52148 +Epoch: [15] [6680/7269] eta: 0:04:09 lr: 0.000007 loss: 0.4842 (0.5308) time: 0.4802 data: 0.0001 max mem: 52148 +Epoch: [15] [6690/7269] eta: 0:04:04 lr: 0.000007 loss: 0.4988 (0.5307) time: 0.4345 data: 0.0001 max mem: 52148 +Epoch: [15] [6700/7269] eta: 0:04:00 lr: 0.000007 loss: 0.5175 (0.5308) time: 0.3720 data: 0.0002 max mem: 52148 +Epoch: [15] [6710/7269] eta: 0:03:56 lr: 0.000007 loss: 0.5509 (0.5308) time: 0.3718 data: 0.0002 max mem: 52148 +Epoch: [15] [6720/7269] eta: 0:03:52 lr: 0.000007 loss: 0.5431 (0.5308) time: 0.3721 data: 0.0001 max mem: 52148 +Epoch: [15] [6730/7269] eta: 0:03:47 lr: 0.000007 loss: 0.5380 (0.5308) time: 0.4223 data: 0.0001 max mem: 52148 +Epoch: [15] [6740/7269] eta: 0:03:43 lr: 0.000007 loss: 0.5327 (0.5308) time: 0.4685 data: 0.0001 max mem: 52148 +Epoch: [15] [6750/7269] eta: 0:03:39 lr: 0.000007 loss: 0.4748 (0.5307) time: 0.4681 data: 0.0002 max mem: 52148 +Epoch: [15] [6760/7269] eta: 0:03:35 lr: 0.000007 loss: 0.4938 (0.5307) time: 0.4214 data: 0.0002 max mem: 52148 +Epoch: [15] [6770/7269] eta: 0:03:31 lr: 0.000007 loss: 0.5075 (0.5307) time: 0.3716 data: 0.0002 max mem: 52148 +Epoch: [15] [6780/7269] eta: 0:03:26 lr: 0.000007 loss: 0.5509 (0.5307) time: 0.3719 data: 0.0001 max mem: 52148 +Epoch: [15] [6790/7269] eta: 0:03:22 lr: 0.000007 loss: 0.5529 (0.5307) time: 0.4202 data: 0.0002 max mem: 52148 +Epoch: [15] [6800/7269] eta: 0:03:18 lr: 0.000007 loss: 0.5195 (0.5307) time: 0.4700 data: 0.0001 max mem: 52148 +Epoch: [15] [6810/7269] eta: 0:03:14 lr: 0.000007 loss: 0.5107 (0.5307) time: 0.4219 data: 0.0002 max mem: 52148 +Epoch: [15] [6820/7269] eta: 0:03:09 lr: 0.000007 loss: 0.5336 (0.5307) time: 0.3721 data: 0.0002 max mem: 52148 +Epoch: [15] [6830/7269] eta: 0:03:05 lr: 0.000007 loss: 0.5330 (0.5307) time: 0.4358 data: 0.0002 max mem: 52148 +Epoch: [15] [6840/7269] eta: 0:03:01 lr: 0.000007 loss: 0.5330 (0.5307) time: 0.4823 data: 0.0001 max mem: 52148 +Epoch: [15] [6850/7269] eta: 0:02:57 lr: 0.000007 loss: 0.5544 (0.5306) time: 0.4182 data: 0.0001 max mem: 52148 +Epoch: [15] [6860/7269] eta: 0:02:52 lr: 0.000007 loss: 0.5054 (0.5306) time: 0.3717 data: 0.0001 max mem: 52148 +Epoch: [15] [6870/7269] eta: 0:02:48 lr: 0.000007 loss: 0.4867 (0.5306) time: 0.3720 data: 0.0002 max mem: 52148 +Epoch: [15] [6880/7269] eta: 0:02:44 lr: 0.000007 loss: 0.4867 (0.5306) time: 0.5527 data: 0.0002 max mem: 52148 +Epoch: [15] [6890/7269] eta: 0:02:40 lr: 0.000007 loss: 0.5500 (0.5306) time: 0.5524 data: 0.0001 max mem: 52148 +Epoch: [15] [6900/7269] eta: 0:02:36 lr: 0.000007 loss: 0.5500 (0.5305) time: 0.3718 data: 0.0002 max mem: 52148 +Epoch: [15] [6910/7269] eta: 0:02:31 lr: 0.000007 loss: 0.5210 (0.5305) time: 0.4170 data: 0.0002 max mem: 52148 +Epoch: [15] [6920/7269] eta: 0:02:27 lr: 0.000007 loss: 0.5328 (0.5305) time: 0.4676 data: 0.0001 max mem: 52148 +Epoch: [15] [6930/7269] eta: 0:02:23 lr: 0.000007 loss: 0.5394 (0.5306) time: 0.4679 data: 0.0002 max mem: 52148 +Epoch: [15] [6940/7269] eta: 0:02:19 lr: 0.000007 loss: 0.5214 (0.5305) time: 0.4171 data: 0.0002 max mem: 52148 +Epoch: [15] [6950/7269] eta: 0:02:14 lr: 0.000007 loss: 0.5090 (0.5305) time: 0.3719 data: 0.0001 max mem: 52148 +Epoch: [15] [6960/7269] eta: 0:02:10 lr: 0.000007 loss: 0.5090 (0.5305) time: 0.3730 data: 0.0002 max mem: 52148 +Epoch: [15] [6970/7269] eta: 0:02:06 lr: 0.000007 loss: 0.5145 (0.5305) time: 0.4636 data: 0.0002 max mem: 52148 +Epoch: [15] [6980/7269] eta: 0:02:02 lr: 0.000007 loss: 0.5384 (0.5305) time: 0.4627 data: 0.0002 max mem: 52148 +Epoch: [15] [6990/7269] eta: 0:01:58 lr: 0.000007 loss: 0.5384 (0.5305) time: 0.3721 data: 0.0002 max mem: 52148 +Epoch: [15] [7000/7269] eta: 0:01:53 lr: 0.000007 loss: 0.5142 (0.5304) time: 0.3719 data: 0.0001 max mem: 52148 +Epoch: [15] [7010/7269] eta: 0:01:49 lr: 0.000007 loss: 0.4992 (0.5304) time: 0.4650 data: 0.0001 max mem: 52148 +Epoch: [15] [7020/7269] eta: 0:01:45 lr: 0.000007 loss: 0.5131 (0.5304) time: 0.5248 data: 0.0002 max mem: 52148 +Epoch: [15] [7030/7269] eta: 0:01:41 lr: 0.000007 loss: 0.5442 (0.5304) time: 0.4321 data: 0.0002 max mem: 52148 +Epoch: [15] [7040/7269] eta: 0:01:36 lr: 0.000007 loss: 0.5442 (0.5304) time: 0.3724 data: 0.0002 max mem: 52148 +Epoch: [15] [7050/7269] eta: 0:01:32 lr: 0.000007 loss: 0.5434 (0.5304) time: 0.3723 data: 0.0001 max mem: 52148 +Epoch: [15] [7060/7269] eta: 0:01:28 lr: 0.000007 loss: 0.5434 (0.5303) time: 0.3722 data: 0.0002 max mem: 52148 +Epoch: [15] [7070/7269] eta: 0:01:24 lr: 0.000007 loss: 0.5351 (0.5303) time: 0.4316 data: 0.0002 max mem: 52148 +Epoch: [15] [7080/7269] eta: 0:01:19 lr: 0.000007 loss: 0.4975 (0.5303) time: 0.4316 data: 0.0001 max mem: 52148 +Epoch: [15] [7090/7269] eta: 0:01:15 lr: 0.000007 loss: 0.5055 (0.5303) time: 0.4190 data: 0.0002 max mem: 52148 +Epoch: [15] [7100/7269] eta: 0:01:11 lr: 0.000007 loss: 0.5458 (0.5303) time: 0.4654 data: 0.0002 max mem: 52148 +Epoch: [15] [7110/7269] eta: 0:01:07 lr: 0.000007 loss: 0.5183 (0.5302) time: 0.4181 data: 0.0001 max mem: 52148 +Epoch: [15] [7120/7269] eta: 0:01:03 lr: 0.000007 loss: 0.4883 (0.5303) time: 0.4622 data: 0.0001 max mem: 52148 +Epoch: [15] [7130/7269] eta: 0:00:58 lr: 0.000007 loss: 0.5274 (0.5303) time: 0.5666 data: 0.0001 max mem: 52148 +Epoch: [15] [7140/7269] eta: 0:00:54 lr: 0.000007 loss: 0.5365 (0.5303) time: 0.4760 data: 0.0001 max mem: 52148 +Epoch: [15] [7150/7269] eta: 0:00:50 lr: 0.000007 loss: 0.5604 (0.5303) time: 0.4826 data: 0.0001 max mem: 52148 +Epoch: [15] [7160/7269] eta: 0:00:46 lr: 0.000007 loss: 0.5653 (0.5304) time: 0.4830 data: 0.0001 max mem: 52148 +Epoch: [15] [7170/7269] eta: 0:00:41 lr: 0.000007 loss: 0.5705 (0.5304) time: 0.3721 data: 0.0001 max mem: 52148 +Epoch: [15] [7180/7269] eta: 0:00:37 lr: 0.000007 loss: 0.5359 (0.5304) time: 0.4226 data: 0.0001 max mem: 52148 +Epoch: [15] [7190/7269] eta: 0:00:33 lr: 0.000007 loss: 0.5059 (0.5303) time: 0.4228 data: 0.0002 max mem: 52148 +Epoch: [15] [7200/7269] eta: 0:00:29 lr: 0.000007 loss: 0.5455 (0.5303) time: 0.3721 data: 0.0002 max mem: 52148 +Epoch: [15] [7210/7269] eta: 0:00:24 lr: 0.000007 loss: 0.4952 (0.5303) time: 0.4207 data: 0.0002 max mem: 52148 +Epoch: [15] [7220/7269] eta: 0:00:20 lr: 0.000007 loss: 0.4952 (0.5303) time: 0.4703 data: 0.0001 max mem: 52148 +Epoch: [15] [7230/7269] eta: 0:00:16 lr: 0.000007 loss: 0.5334 (0.5303) time: 0.4217 data: 0.0001 max mem: 52148 +Epoch: [15] [7240/7269] eta: 0:00:12 lr: 0.000007 loss: 0.5163 (0.5303) time: 0.3720 data: 0.0001 max mem: 52148 +Epoch: [15] [7250/7269] eta: 0:00:08 lr: 0.000007 loss: 0.5179 (0.5303) time: 0.3745 data: 0.0024 max mem: 52148 +Epoch: [15] [7260/7269] eta: 0:00:03 lr: 0.000007 loss: 0.5364 (0.5303) time: 0.4634 data: 0.0024 max mem: 52148 +Epoch: [15] [7268/7269] eta: 0:00:00 lr: 0.000007 loss: 0.5883 (0.5304) time: 0.4629 data: 0.0023 max mem: 52148 +Epoch: [15] Total time: 0:51:21 (0.4239 s / it) +Averaged stats: lr: 0.000007 loss: 0.5883 (0.5299) +Test: [ 0/1403] eta: 1:21:33 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1008 (0.1008) acc1: 98.4375 (98.4375) acc5: 100.0000 (100.0000) time: 3.4880 data: 3.2109 max mem: 52148 +Test: [ 10/1403] eta: 0:09:24 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1580 (0.1682) acc1: 98.4375 (96.5909) acc5: 100.0000 (100.0000) time: 0.4049 data: 0.2922 max mem: 52148 +Test: [ 20/1403] eta: 0:05:56 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1446 (0.1524) acc1: 98.4375 (96.8750) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52148 +Test: [ 30/1403] eta: 0:04:42 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1010 (0.1579) acc1: 98.4375 (96.1190) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52148 +Test: [ 40/1403] eta: 0:04:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1191 (0.1531) acc1: 100.0000 (96.6463) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52148 +Test: [ 50/1403] eta: 0:03:40 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1099 (0.1698) acc1: 100.0000 (95.5270) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52148 +Test: [ 60/1403] eta: 0:03:24 acc1_pos: 89.062 acc5_pos: 100.000 loss: 0.1310 (0.1676) acc1: 100.0000 (96.0297) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52148 +Test: [ 70/1403] eta: 0:03:12 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.1501 (0.1709) acc1: 98.4375 (95.6206) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52148 +Test: [ 80/1403] eta: 0:03:03 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.1621 (0.1786) acc1: 95.3125 (95.2739) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52148 +Test: [ 90/1403] eta: 0:02:55 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1808 (0.1969) acc1: 95.3125 (94.3681) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52148 +Test: [ 100/1403] eta: 0:02:49 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1904 (0.1991) acc1: 95.3125 (94.3379) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52148 +Test: [ 110/1403] eta: 0:02:44 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1658 (0.1978) acc1: 96.8750 (94.3412) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52148 +Test: [ 120/1403] eta: 0:02:39 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0961 (0.1888) acc1: 98.4375 (94.7314) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52148 +Test: [ 130/1403] eta: 0:02:35 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0957 (0.1825) acc1: 98.4375 (94.9785) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52148 +Test: [ 140/1403] eta: 0:02:32 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0902 (0.1753) acc1: 100.0000 (95.3236) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52148 +Test: [ 150/1403] eta: 0:02:29 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0830 (0.1695) acc1: 100.0000 (95.5608) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52148 +Test: [ 160/1403] eta: 0:02:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0838 (0.1647) acc1: 98.4375 (95.7589) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52148 +Test: [ 170/1403] eta: 0:02:23 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0782 (0.1600) acc1: 100.0000 (95.9704) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52148 +Test: [ 180/1403] eta: 0:02:20 acc1_pos: 84.375 acc5_pos: 100.000 loss: 0.0760 (0.1648) acc1: 100.0000 (95.6751) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52148 +Test: [ 190/1403] eta: 0:02:18 acc1_pos: 71.875 acc5_pos: 100.000 loss: 0.3175 (0.1781) acc1: 87.5000 (94.8707) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52148 +Test: [ 200/1403] eta: 0:02:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3362 (0.1920) acc1: 84.3750 (94.1465) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52148 +Test: [ 210/1403] eta: 0:02:14 acc1_pos: 75.000 acc5_pos: 100.000 loss: 0.3536 (0.2057) acc1: 84.3750 (93.4020) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52148 +Test: [ 220/1403] eta: 0:02:12 acc1_pos: 84.375 acc5_pos: 100.000 loss: 0.4037 (0.2151) acc1: 79.6875 (92.9087) acc5: 100.0000 (100.0000) time: 0.0967 data: 0.0003 max mem: 52148 +Test: [ 230/1403] eta: 0:02:10 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2443 (0.2100) acc1: 90.6250 (93.1345) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52148 +Test: [ 240/1403] eta: 0:02:08 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0723 (0.2042) acc1: 100.0000 (93.4129) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52148 +Test: [ 250/1403] eta: 0:02:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0728 (0.1992) acc1: 100.0000 (93.6504) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52148 +Test: [ 260/1403] eta: 0:02:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0734 (0.1946) acc1: 100.0000 (93.8757) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52148 +Test: [ 270/1403] eta: 0:02:03 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0723 (0.1902) acc1: 100.0000 (94.0844) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52148 +Test: [ 280/1403] eta: 0:02:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0686 (0.1859) acc1: 100.0000 (94.2894) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52148 +Test: [ 290/1403] eta: 0:02:00 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0656 (0.1819) acc1: 100.0000 (94.4749) acc5: 100.0000 (100.0000) time: 0.0967 data: 0.0003 max mem: 52148 +Test: [ 300/1403] eta: 0:01:58 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0656 (0.1782) acc1: 100.0000 (94.6532) acc5: 100.0000 (100.0000) time: 0.0972 data: 0.0003 max mem: 52148 +Test: [ 310/1403] eta: 0:01:57 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0701 (0.1750) acc1: 100.0000 (94.8151) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0974 data: 0.0003 max mem: 52148 +Test: [ 320/1403] eta: 0:01:56 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0944 (0.1726) acc1: 100.0000 (94.9766) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52148 +Test: [ 330/1403] eta: 0:01:54 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0988 (0.1705) acc1: 100.0000 (95.1284) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52148 +Test: [ 340/1403] eta: 0:01:53 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0988 (0.1683) acc1: 100.0000 (95.2667) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52148 +Test: [ 350/1403] eta: 0:01:51 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0988 (0.1664) acc1: 100.0000 (95.4015) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52148 +Test: [ 360/1403] eta: 0:01:50 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1014 (0.1646) acc1: 100.0000 (95.5203) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52148 +Test: [ 370/1403] eta: 0:01:49 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0993 (0.1628) acc1: 100.0000 (95.6410) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52148 +Test: [ 380/1403] eta: 0:01:47 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0967 (0.1611) acc1: 100.0000 (95.7554) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52148 +Test: [ 390/1403] eta: 0:01:46 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0980 (0.1614) acc1: 100.0000 (95.7561) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52148 +Test: [ 400/1403] eta: 0:01:45 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2412 (0.1655) acc1: 89.0625 (95.5151) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52148 +Test: [ 410/1403] eta: 0:01:44 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3261 (0.1690) acc1: 87.5000 (95.3163) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52148 +Test: [ 420/1403] eta: 0:01:42 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2565 (0.1707) acc1: 89.0625 (95.1900) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52148 +Test: [ 430/1403] eta: 0:01:41 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2565 (0.1740) acc1: 89.0625 (95.0189) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52148 +Test: [ 440/1403] eta: 0:01:40 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3287 (0.1777) acc1: 87.5000 (94.8271) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52148 +Test: [ 450/1403] eta: 0:01:39 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2918 (0.1800) acc1: 89.0625 (94.7027) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52148 +Test: [ 460/1403] eta: 0:01:38 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2856 (0.1826) acc1: 89.0625 (94.5533) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52148 +Test: [ 470/1403] eta: 0:01:36 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2485 (0.1816) acc1: 90.6250 (94.5827) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52148 +Test: [ 480/1403] eta: 0:01:35 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0600 (0.1791) acc1: 100.0000 (94.6953) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52148 +Test: [ 490/1403] eta: 0:01:34 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0600 (0.1767) acc1: 100.0000 (94.8002) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52148 +Test: [ 500/1403] eta: 0:01:33 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0611 (0.1744) acc1: 100.0000 (94.9039) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52148 +Test: [ 510/1403] eta: 0:01:32 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0611 (0.1722) acc1: 100.0000 (95.0037) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52148 +Test: [ 520/1403] eta: 0:01:31 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0643 (0.1702) acc1: 100.0000 (95.0996) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52148 +Test: [ 530/1403] eta: 0:01:29 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0590 (0.1681) acc1: 100.0000 (95.1919) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52148 +Test: [ 540/1403] eta: 0:01:28 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0585 (0.1661) acc1: 100.0000 (95.2807) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52148 +Test: [ 550/1403] eta: 0:01:27 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0595 (0.1647) acc1: 100.0000 (95.3437) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52148 +Test: [ 560/1403] eta: 0:01:26 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0911 (0.1637) acc1: 98.4375 (95.3877) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52148 +Test: [ 570/1403] eta: 0:01:25 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0911 (0.1624) acc1: 98.4375 (95.4548) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52148 +Test: [ 580/1403] eta: 0:01:24 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0911 (0.1616) acc1: 100.0000 (95.4927) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52148 +Test: [ 590/1403] eta: 0:01:23 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1081 (0.1611) acc1: 96.8750 (95.5081) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52148 +Test: [ 600/1403] eta: 0:01:22 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1237 (0.1614) acc1: 96.8750 (95.4711) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52148 +Test: [ 610/1403] eta: 0:01:21 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1319 (0.1608) acc1: 95.3125 (95.4864) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52148 +Test: [ 620/1403] eta: 0:01:19 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1378 (0.1615) acc1: 93.7500 (95.4081) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52148 +Test: [ 630/1403] eta: 0:01:18 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1826 (0.1616) acc1: 92.1875 (95.3942) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52148 +Test: [ 640/1403] eta: 0:01:17 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1412 (0.1613) acc1: 95.3125 (95.4027) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52148 +Test: [ 650/1403] eta: 0:01:16 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1404 (0.1609) acc1: 96.8750 (95.4181) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52148 +Test: [ 660/1403] eta: 0:01:15 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1078 (0.1598) acc1: 98.4375 (95.4756) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52148 +Test: [ 670/1403] eta: 0:01:14 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0746 (0.1585) acc1: 100.0000 (95.5360) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52148 +Test: [ 680/1403] eta: 0:01:13 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0868 (0.1609) acc1: 98.4375 (95.3676) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52148 +Test: [ 690/1403] eta: 0:01:12 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3763 (0.1664) acc1: 79.6875 (95.0366) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52148 +Test: [ 700/1403] eta: 0:01:11 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2767 (0.1670) acc1: 85.9375 (95.0004) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52148 +Test: [ 710/1403] eta: 0:01:10 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3891 (0.1722) acc1: 84.3750 (94.7082) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52148 +Test: [ 720/1403] eta: 0:01:09 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3891 (0.1723) acc1: 79.6875 (94.6992) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52148 +Test: [ 730/1403] eta: 0:01:08 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0660 (0.1710) acc1: 100.0000 (94.7589) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52148 +Test: [ 740/1403] eta: 0:01:07 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0743 (0.1700) acc1: 100.0000 (94.8106) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52148 +Test: [ 750/1403] eta: 0:01:06 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1198 (0.1719) acc1: 96.8750 (94.6883) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52148 +Test: [ 760/1403] eta: 0:01:04 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3721 (0.1762) acc1: 79.6875 (94.3968) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52148 +Test: [ 770/1403] eta: 0:01:03 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.5366 (0.1814) acc1: 70.3125 (94.0722) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52148 +Test: [ 780/1403] eta: 0:01:02 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.4081 (0.1833) acc1: 79.6875 (93.9601) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52148 +Test: [ 790/1403] eta: 0:01:01 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2855 (0.1850) acc1: 85.9375 (93.8725) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52148 +Test: [ 800/1403] eta: 0:01:00 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2161 (0.1851) acc1: 92.1875 (93.8670) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0970 data: 0.0003 max mem: 52148 +Test: [ 810/1403] eta: 0:00:59 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2281 (0.1877) acc1: 90.6250 (93.7384) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52148 +Test: [ 820/1403] eta: 0:00:58 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.5730 (0.1959) acc1: 73.4375 (93.3846) acc5: 100.0000 (100.0000) acc1_neg: 40.625 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52148 +Test: [ 830/1403] eta: 0:00:57 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.6245 (0.1980) acc1: 73.4375 (93.2743) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52148 +Test: [ 840/1403] eta: 0:00:56 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.4260 (0.2052) acc1: 75.0000 (92.9604) acc5: 100.0000 (100.0000) acc1_neg: 53.125 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52148 +Test: [ 850/1403] eta: 0:00:55 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.9657 (0.2154) acc1: 56.2500 (92.5143) acc5: 100.0000 (100.0000) acc1_neg: 51.562 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52148 +Test: [ 860/1403] eta: 0:00:54 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.9229 (0.2186) acc1: 57.8125 (92.3726) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52148 +Test: [ 870/1403] eta: 0:00:53 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1484 (0.2175) acc1: 95.3125 (92.4171) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52148 +Test: [ 880/1403] eta: 0:00:52 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1101 (0.2164) acc1: 96.8750 (92.4677) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52148 +Test: [ 890/1403] eta: 0:00:51 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0857 (0.2149) acc1: 98.4375 (92.5452) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52148 +Test: [ 900/1403] eta: 0:00:50 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0766 (0.2133) acc1: 100.0000 (92.6245) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52148 +Test: [ 910/1403] eta: 0:00:49 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0790 (0.2126) acc1: 98.4375 (92.6403) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52148 +Test: [ 920/1403] eta: 0:00:48 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1302 (0.2118) acc1: 95.3125 (92.6795) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52148 +Test: [ 930/1403] eta: 0:00:47 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1255 (0.2108) acc1: 96.8750 (92.7262) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52148 +Test: [ 940/1403] eta: 0:00:46 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1344 (0.2115) acc1: 93.7500 (92.6674) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52148 +Test: [ 950/1403] eta: 0:00:45 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3196 (0.2133) acc1: 82.8125 (92.5506) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.0968 data: 0.0004 max mem: 52148 +Test: [ 960/1403] eta: 0:00:44 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3572 (0.2147) acc1: 82.8125 (92.4720) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52148 +Test: [ 970/1403] eta: 0:00:43 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3501 (0.2164) acc1: 84.3750 (92.3774) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52148 +Test: [ 980/1403] eta: 0:00:42 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3544 (0.2183) acc1: 82.8125 (92.2560) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52148 +Test: [ 990/1403] eta: 0:00:41 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3526 (0.2196) acc1: 82.8125 (92.1670) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52148 +Test: [1000/1403] eta: 0:00:40 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3526 (0.2212) acc1: 82.8125 (92.0782) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52148 +Test: [1010/1403] eta: 0:00:39 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3444 (0.2225) acc1: 82.8125 (92.0005) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52148 +Test: [1020/1403] eta: 0:00:38 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.4646 (0.2286) acc1: 78.1250 (91.6611) acc5: 100.0000 (100.0000) acc1_neg: 28.125 acc5_neg: 100.000 time: 0.0970 data: 0.0004 max mem: 52148 +Test: [1030/1403] eta: 0:00:37 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.5767 (0.2307) acc1: 64.0625 (91.5222) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0969 data: 0.0004 max mem: 52148 +Test: [1040/1403] eta: 0:00:36 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2306 (0.2304) acc1: 90.6250 (91.5376) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52148 +Test: [1050/1403] eta: 0:00:35 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1706 (0.2299) acc1: 95.3125 (91.5705) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52148 +Test: [1060/1403] eta: 0:00:34 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1339 (0.2289) acc1: 96.8750 (91.6279) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52148 +Test: [1070/1403] eta: 0:00:33 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1352 (0.2294) acc1: 96.8750 (91.6098) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52148 +Test: [1080/1403] eta: 0:00:32 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.4043 (0.2316) acc1: 82.8125 (91.4937) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52148 +Test: [1090/1403] eta: 0:00:31 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.4694 (0.2338) acc1: 79.6875 (91.3869) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52148 +Test: [1100/1403] eta: 0:00:30 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.4125 (0.2346) acc1: 82.8125 (91.3545) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52148 +Test: [1110/1403] eta: 0:00:29 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3046 (0.2353) acc1: 87.5000 (91.3268) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52148 +Test: [1120/1403] eta: 0:00:28 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3041 (0.2359) acc1: 89.0625 (91.2954) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52148 +Test: [1130/1403] eta: 0:00:27 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3071 (0.2369) acc1: 85.9375 (91.2398) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52148 +Test: [1140/1403] eta: 0:00:26 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2713 (0.2371) acc1: 89.0625 (91.2275) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52148 +Test: [1150/1403] eta: 0:00:25 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2715 (0.2377) acc1: 87.5000 (91.1911) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52148 +Test: [1160/1403] eta: 0:00:24 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3062 (0.2384) acc1: 87.5000 (91.1620) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52148 +Test: [1170/1403] eta: 0:00:23 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2942 (0.2383) acc1: 89.0625 (91.1627) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52148 +Test: [1180/1403] eta: 0:00:22 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0671 (0.2369) acc1: 100.0000 (91.2376) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52148 +Test: [1190/1403] eta: 0:00:21 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0663 (0.2355) acc1: 100.0000 (91.3111) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52148 +Test: [1200/1403] eta: 0:00:20 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0663 (0.2340) acc1: 100.0000 (91.3835) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52148 +Test: [1210/1403] eta: 0:00:19 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0658 (0.2327) acc1: 100.0000 (91.4546) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52148 +Test: [1220/1403] eta: 0:00:18 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0656 (0.2313) acc1: 100.0000 (91.5246) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52148 +Test: [1230/1403] eta: 0:00:17 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0668 (0.2300) acc1: 100.0000 (91.5935) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52148 +Test: [1240/1403] eta: 0:00:16 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0663 (0.2286) acc1: 100.0000 (91.6612) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52148 +Test: [1250/1403] eta: 0:00:15 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0659 (0.2274) acc1: 100.0000 (91.7279) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52148 +Test: [1260/1403] eta: 0:00:14 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0687 (0.2261) acc1: 100.0000 (91.7935) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52148 +Test: [1270/1403] eta: 0:00:13 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0701 (0.2249) acc1: 100.0000 (91.8580) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52148 +Test: [1280/1403] eta: 0:00:12 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0699 (0.2237) acc1: 100.0000 (91.9216) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52148 +Test: [1290/1403] eta: 0:00:11 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0696 (0.2225) acc1: 100.0000 (91.9842) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52148 +Test: [1300/1403] eta: 0:00:10 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0704 (0.2213) acc1: 100.0000 (92.0458) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0002 max mem: 52148 +Test: [1310/1403] eta: 0:00:09 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0704 (0.2202) acc1: 100.0000 (92.1065) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0002 max mem: 52148 +Test: [1320/1403] eta: 0:00:08 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0690 (0.2190) acc1: 100.0000 (92.1662) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0002 max mem: 52148 +Test: [1330/1403] eta: 0:00:07 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0690 (0.2179) acc1: 100.0000 (92.2192) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0002 max mem: 52148 +Test: [1340/1403] eta: 0:00:06 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0677 (0.2168) acc1: 100.0000 (92.2737) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52148 +Test: [1350/1403] eta: 0:00:05 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0676 (0.2157) acc1: 100.0000 (92.3263) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52148 +Test: [1360/1403] eta: 0:00:04 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0664 (0.2147) acc1: 100.0000 (92.3827) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52148 +Test: [1370/1403] eta: 0:00:03 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0664 (0.2136) acc1: 100.0000 (92.4348) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0002 max mem: 52148 +Test: [1380/1403] eta: 0:00:02 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0656 (0.2126) acc1: 100.0000 (92.4873) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0963 data: 0.0002 max mem: 52148 +Test: [1390/1403] eta: 0:00:01 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0653 (0.2115) acc1: 100.0000 (92.5413) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0002 max mem: 52148 +Test: [1400/1403] eta: 0:00:00 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0657 (0.2106) acc1: 100.0000 (92.5901) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0001 max mem: 52148 +Test: [1402/1403] eta: 0:00:00 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0657 (0.2103) acc1: 100.0000 (92.5955) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0990 data: 0.0001 max mem: 52148 +Test: Total time: 0:02:19 (0.0995 s / it) +* Overall Acc@1 92.595 Acc@5 100.000 + target == real Acc@1 94.769 Acc@5 100.000 + target == fake Acc@1 91.983 Acc@5 100.000 + loss 0.210 +Accuracy of the network on the 89729 test images: 92.6% +Max accuracy: 92.60% +Epoch: [16] [ 0/7269] eta: 11:39:02 lr: 0.000007 loss: 0.4480 (0.4480) time: 5.7701 data: 5.3649 max mem: 52149 +Epoch: [16] [ 10/7269] eta: 1:44:52 lr: 0.000007 loss: 0.5606 (0.5406) time: 0.8668 data: 0.4879 max mem: 52150 +Epoch: [16] [ 20/7269] eta: 1:36:07 lr: 0.000007 loss: 0.5386 (0.5324) time: 0.5468 data: 0.0001 max mem: 52150 +Epoch: [16] [ 30/7269] eta: 1:19:32 lr: 0.000007 loss: 0.5210 (0.5260) time: 0.5451 data: 0.0002 max mem: 52150 +Epoch: [16] [ 40/7269] eta: 1:10:58 lr: 0.000007 loss: 0.5032 (0.5268) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [16] [ 50/7269] eta: 1:07:58 lr: 0.000007 loss: 0.5032 (0.5256) time: 0.4187 data: 0.0002 max mem: 52150 +Epoch: [16] [ 60/7269] eta: 1:04:05 lr: 0.000007 loss: 0.5074 (0.5233) time: 0.4193 data: 0.0002 max mem: 52150 +Epoch: [16] [ 70/7269] eta: 1:02:57 lr: 0.000007 loss: 0.5360 (0.5241) time: 0.4224 data: 0.0002 max mem: 52150 +Epoch: [16] [ 80/7269] eta: 1:00:36 lr: 0.000007 loss: 0.5463 (0.5263) time: 0.4219 data: 0.0002 max mem: 52150 +Epoch: [16] [ 90/7269] eta: 1:00:02 lr: 0.000007 loss: 0.5463 (0.5263) time: 0.4201 data: 0.0002 max mem: 52150 +Epoch: [16] [ 100/7269] eta: 0:59:52 lr: 0.000007 loss: 0.5427 (0.5275) time: 0.4821 data: 0.0002 max mem: 52150 +Epoch: [16] [ 110/7269] eta: 0:59:46 lr: 0.000007 loss: 0.5427 (0.5278) time: 0.4975 data: 0.0002 max mem: 52150 +Epoch: [16] [ 120/7269] eta: 0:58:25 lr: 0.000007 loss: 0.5537 (0.5279) time: 0.4356 data: 0.0002 max mem: 52150 +Epoch: [16] [ 130/7269] eta: 0:58:22 lr: 0.000007 loss: 0.5160 (0.5247) time: 0.4331 data: 0.0002 max mem: 52150 +Epoch: [16] [ 140/7269] eta: 0:57:17 lr: 0.000007 loss: 0.5104 (0.5232) time: 0.4328 data: 0.0002 max mem: 52150 +Epoch: [16] [ 150/7269] eta: 0:57:20 lr: 0.000007 loss: 0.5471 (0.5255) time: 0.4349 data: 0.0002 max mem: 52150 +Epoch: [16] [ 160/7269] eta: 0:56:26 lr: 0.000007 loss: 0.5471 (0.5261) time: 0.4354 data: 0.0002 max mem: 52150 +Epoch: [16] [ 170/7269] eta: 0:55:38 lr: 0.000007 loss: 0.5393 (0.5272) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [16] [ 180/7269] eta: 0:54:54 lr: 0.000007 loss: 0.5368 (0.5253) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [16] [ 190/7269] eta: 0:55:39 lr: 0.000007 loss: 0.4828 (0.5248) time: 0.4837 data: 0.0002 max mem: 52150 +Epoch: [16] [ 200/7269] eta: 0:54:59 lr: 0.000007 loss: 0.5544 (0.5273) time: 0.4838 data: 0.0002 max mem: 52150 +Epoch: [16] [ 210/7269] eta: 0:54:54 lr: 0.000007 loss: 0.5603 (0.5268) time: 0.4196 data: 0.0001 max mem: 52150 +Epoch: [16] [ 220/7269] eta: 0:54:19 lr: 0.000007 loss: 0.5240 (0.5262) time: 0.4197 data: 0.0002 max mem: 52150 +Epoch: [16] [ 230/7269] eta: 0:53:47 lr: 0.000007 loss: 0.5284 (0.5258) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [16] [ 240/7269] eta: 0:53:17 lr: 0.000007 loss: 0.5015 (0.5253) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [16] [ 250/7269] eta: 0:53:16 lr: 0.000007 loss: 0.5175 (0.5259) time: 0.4189 data: 0.0002 max mem: 52150 +Epoch: [16] [ 260/7269] eta: 0:52:49 lr: 0.000007 loss: 0.5585 (0.5261) time: 0.4191 data: 0.0002 max mem: 52150 +Epoch: [16] [ 270/7269] eta: 0:52:56 lr: 0.000007 loss: 0.5343 (0.5255) time: 0.4342 data: 0.0002 max mem: 52150 +Epoch: [16] [ 280/7269] eta: 0:52:31 lr: 0.000007 loss: 0.5245 (0.5259) time: 0.4345 data: 0.0002 max mem: 52150 +Epoch: [16] [ 290/7269] eta: 0:52:37 lr: 0.000007 loss: 0.5106 (0.5248) time: 0.4340 data: 0.0002 max mem: 52150 +Epoch: [16] [ 300/7269] eta: 0:52:14 lr: 0.000007 loss: 0.4969 (0.5242) time: 0.4335 data: 0.0002 max mem: 52150 +Epoch: [16] [ 310/7269] eta: 0:51:52 lr: 0.000007 loss: 0.5364 (0.5250) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [16] [ 320/7269] eta: 0:51:52 lr: 0.000007 loss: 0.5458 (0.5248) time: 0.4188 data: 0.0002 max mem: 52150 +Epoch: [16] [ 330/7269] eta: 0:51:57 lr: 0.000007 loss: 0.5297 (0.5243) time: 0.4802 data: 0.0002 max mem: 52150 +Epoch: [16] [ 340/7269] eta: 0:51:37 lr: 0.000007 loss: 0.5329 (0.5250) time: 0.4335 data: 0.0002 max mem: 52150 +Epoch: [16] [ 350/7269] eta: 0:51:18 lr: 0.000007 loss: 0.5334 (0.5242) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [16] [ 360/7269] eta: 0:51:18 lr: 0.000007 loss: 0.5334 (0.5244) time: 0.4204 data: 0.0002 max mem: 52150 +Epoch: [16] [ 370/7269] eta: 0:51:00 lr: 0.000007 loss: 0.5420 (0.5250) time: 0.4214 data: 0.0002 max mem: 52150 +Epoch: [16] [ 380/7269] eta: 0:51:22 lr: 0.000007 loss: 0.5438 (0.5255) time: 0.4828 data: 0.0002 max mem: 52150 +Epoch: [16] [ 390/7269] eta: 0:51:48 lr: 0.000007 loss: 0.5407 (0.5255) time: 0.6062 data: 0.0002 max mem: 52150 +Epoch: [16] [ 400/7269] eta: 0:51:51 lr: 0.000007 loss: 0.5479 (0.5263) time: 0.5582 data: 0.0002 max mem: 52150 +Epoch: [16] [ 410/7269] eta: 0:51:54 lr: 0.000007 loss: 0.5316 (0.5254) time: 0.4958 data: 0.0002 max mem: 52150 +Epoch: [16] [ 420/7269] eta: 0:51:56 lr: 0.000007 loss: 0.5043 (0.5257) time: 0.4963 data: 0.0002 max mem: 52150 +Epoch: [16] [ 430/7269] eta: 0:51:39 lr: 0.000007 loss: 0.5646 (0.5265) time: 0.4342 data: 0.0002 max mem: 52150 +Epoch: [16] [ 440/7269] eta: 0:51:56 lr: 0.000007 loss: 0.5624 (0.5273) time: 0.4821 data: 0.0002 max mem: 52150 +Epoch: [16] [ 450/7269] eta: 0:51:38 lr: 0.000007 loss: 0.5179 (0.5258) time: 0.4823 data: 0.0002 max mem: 52150 +Epoch: [16] [ 460/7269] eta: 0:51:22 lr: 0.000007 loss: 0.4718 (0.5256) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [16] [ 470/7269] eta: 0:51:06 lr: 0.000007 loss: 0.5102 (0.5253) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [16] [ 480/7269] eta: 0:51:20 lr: 0.000007 loss: 0.4919 (0.5250) time: 0.4788 data: 0.0001 max mem: 52150 +Epoch: [16] [ 490/7269] eta: 0:51:04 lr: 0.000007 loss: 0.5293 (0.5253) time: 0.4780 data: 0.0002 max mem: 52150 +Epoch: [16] [ 500/7269] eta: 0:50:49 lr: 0.000007 loss: 0.5332 (0.5251) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [16] [ 510/7269] eta: 0:50:46 lr: 0.000007 loss: 0.5324 (0.5251) time: 0.4170 data: 0.0002 max mem: 52150 +Epoch: [16] [ 520/7269] eta: 0:51:03 lr: 0.000007 loss: 0.5478 (0.5252) time: 0.5424 data: 0.0002 max mem: 52150 +Epoch: [16] [ 530/7269] eta: 0:51:00 lr: 0.000007 loss: 0.5304 (0.5246) time: 0.5413 data: 0.0002 max mem: 52150 +Epoch: [16] [ 540/7269] eta: 0:50:45 lr: 0.000007 loss: 0.5304 (0.5251) time: 0.4158 data: 0.0002 max mem: 52150 +Epoch: [16] [ 550/7269] eta: 0:50:31 lr: 0.000007 loss: 0.5431 (0.5251) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [16] [ 560/7269] eta: 0:50:16 lr: 0.000007 loss: 0.5307 (0.5250) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [16] [ 570/7269] eta: 0:50:03 lr: 0.000007 loss: 0.5173 (0.5245) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [16] [ 580/7269] eta: 0:50:01 lr: 0.000007 loss: 0.5569 (0.5252) time: 0.4233 data: 0.0002 max mem: 52150 +Epoch: [16] [ 590/7269] eta: 0:49:48 lr: 0.000007 loss: 0.5569 (0.5249) time: 0.4234 data: 0.0002 max mem: 52150 +Epoch: [16] [ 600/7269] eta: 0:49:35 lr: 0.000007 loss: 0.5608 (0.5257) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [16] [ 610/7269] eta: 0:49:23 lr: 0.000007 loss: 0.5608 (0.5256) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [16] [ 620/7269] eta: 0:49:21 lr: 0.000007 loss: 0.5435 (0.5259) time: 0.4198 data: 0.0002 max mem: 52150 +Epoch: [16] [ 630/7269] eta: 0:49:18 lr: 0.000007 loss: 0.5697 (0.5264) time: 0.4648 data: 0.0002 max mem: 52150 +Epoch: [16] [ 640/7269] eta: 0:49:19 lr: 0.000007 loss: 0.5636 (0.5268) time: 0.4779 data: 0.0002 max mem: 52150 +Epoch: [16] [ 650/7269] eta: 0:49:07 lr: 0.000007 loss: 0.5158 (0.5265) time: 0.4336 data: 0.0002 max mem: 52150 +Epoch: [16] [ 660/7269] eta: 0:49:07 lr: 0.000007 loss: 0.5127 (0.5263) time: 0.4339 data: 0.0002 max mem: 52150 +Epoch: [16] [ 670/7269] eta: 0:48:55 lr: 0.000007 loss: 0.5573 (0.5266) time: 0.4336 data: 0.0002 max mem: 52150 +Epoch: [16] [ 680/7269] eta: 0:48:44 lr: 0.000007 loss: 0.5428 (0.5270) time: 0.3721 data: 0.0001 max mem: 52150 +Epoch: [16] [ 690/7269] eta: 0:48:33 lr: 0.000007 loss: 0.5428 (0.5269) time: 0.3726 data: 0.0001 max mem: 52150 +Epoch: [16] [ 700/7269] eta: 0:48:41 lr: 0.000007 loss: 0.5159 (0.5267) time: 0.4776 data: 0.0002 max mem: 52150 +Epoch: [16] [ 710/7269] eta: 0:48:30 lr: 0.000007 loss: 0.5183 (0.5268) time: 0.4771 data: 0.0002 max mem: 52150 +Epoch: [16] [ 720/7269] eta: 0:48:19 lr: 0.000007 loss: 0.5233 (0.5269) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [16] [ 730/7269] eta: 0:48:09 lr: 0.000007 loss: 0.5194 (0.5265) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [16] [ 740/7269] eta: 0:47:58 lr: 0.000007 loss: 0.5057 (0.5264) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [16] [ 750/7269] eta: 0:47:55 lr: 0.000007 loss: 0.5280 (0.5267) time: 0.4172 data: 0.0002 max mem: 52150 +Epoch: [16] [ 760/7269] eta: 0:47:45 lr: 0.000007 loss: 0.5226 (0.5265) time: 0.4170 data: 0.0002 max mem: 52150 +Epoch: [16] [ 770/7269] eta: 0:47:56 lr: 0.000007 loss: 0.5146 (0.5262) time: 0.4941 data: 0.0002 max mem: 52150 +Epoch: [16] [ 780/7269] eta: 0:47:45 lr: 0.000007 loss: 0.5092 (0.5256) time: 0.4935 data: 0.0001 max mem: 52150 +Epoch: [16] [ 790/7269] eta: 0:47:35 lr: 0.000007 loss: 0.4832 (0.5255) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [16] [ 800/7269] eta: 0:47:25 lr: 0.000007 loss: 0.5379 (0.5260) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [16] [ 810/7269] eta: 0:47:25 lr: 0.000007 loss: 0.5514 (0.5262) time: 0.4324 data: 0.0002 max mem: 52150 +Epoch: [16] [ 820/7269] eta: 0:47:15 lr: 0.000007 loss: 0.5428 (0.5264) time: 0.4322 data: 0.0002 max mem: 52150 +Epoch: [16] [ 830/7269] eta: 0:47:15 lr: 0.000007 loss: 0.5313 (0.5264) time: 0.4331 data: 0.0002 max mem: 52150 +Epoch: [16] [ 840/7269] eta: 0:47:05 lr: 0.000007 loss: 0.5435 (0.5266) time: 0.4337 data: 0.0002 max mem: 52150 +Epoch: [16] [ 850/7269] eta: 0:47:03 lr: 0.000007 loss: 0.5449 (0.5267) time: 0.4183 data: 0.0001 max mem: 52150 +Epoch: [16] [ 860/7269] eta: 0:47:02 lr: 0.000007 loss: 0.5585 (0.5272) time: 0.4780 data: 0.0002 max mem: 52150 +Epoch: [16] [ 870/7269] eta: 0:47:02 lr: 0.000007 loss: 0.5320 (0.5270) time: 0.4927 data: 0.0002 max mem: 52150 +Epoch: [16] [ 880/7269] eta: 0:46:52 lr: 0.000007 loss: 0.5289 (0.5272) time: 0.4326 data: 0.0002 max mem: 52150 +Epoch: [16] [ 890/7269] eta: 0:46:43 lr: 0.000007 loss: 0.5284 (0.5268) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [16] [ 900/7269] eta: 0:46:47 lr: 0.000007 loss: 0.4882 (0.5265) time: 0.4622 data: 0.0002 max mem: 52150 +Epoch: [16] [ 910/7269] eta: 0:46:37 lr: 0.000007 loss: 0.5107 (0.5265) time: 0.4620 data: 0.0002 max mem: 52150 +Epoch: [16] [ 920/7269] eta: 0:46:28 lr: 0.000007 loss: 0.5235 (0.5266) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [16] [ 930/7269] eta: 0:46:19 lr: 0.000007 loss: 0.5360 (0.5267) time: 0.3724 data: 0.0001 max mem: 52150 +Epoch: [16] [ 940/7269] eta: 0:46:11 lr: 0.000007 loss: 0.5360 (0.5268) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [16] [ 950/7269] eta: 0:46:10 lr: 0.000007 loss: 0.5308 (0.5268) time: 0.4311 data: 0.0001 max mem: 52150 +Epoch: [16] [ 960/7269] eta: 0:46:07 lr: 0.000007 loss: 0.5308 (0.5269) time: 0.4772 data: 0.0002 max mem: 52150 +Epoch: [16] [ 970/7269] eta: 0:45:58 lr: 0.000007 loss: 0.5231 (0.5267) time: 0.4180 data: 0.0002 max mem: 52150 +Epoch: [16] [ 980/7269] eta: 0:45:57 lr: 0.000007 loss: 0.5182 (0.5269) time: 0.4316 data: 0.0001 max mem: 52150 +Epoch: [16] [ 990/7269] eta: 0:45:49 lr: 0.000007 loss: 0.5811 (0.5273) time: 0.4318 data: 0.0002 max mem: 52150 +Epoch: [16] [1000/7269] eta: 0:45:55 lr: 0.000007 loss: 0.5684 (0.5275) time: 0.4910 data: 0.0002 max mem: 52150 +Epoch: [16] [1010/7269] eta: 0:45:47 lr: 0.000007 loss: 0.5509 (0.5276) time: 0.4911 data: 0.0001 max mem: 52150 +Epoch: [16] [1020/7269] eta: 0:45:38 lr: 0.000007 loss: 0.5415 (0.5279) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [16] [1030/7269] eta: 0:45:47 lr: 0.000007 loss: 0.5545 (0.5281) time: 0.5193 data: 0.0002 max mem: 52150 +Epoch: [16] [1040/7269] eta: 0:45:39 lr: 0.000007 loss: 0.4990 (0.5278) time: 0.5193 data: 0.0002 max mem: 52150 +Epoch: [16] [1050/7269] eta: 0:45:38 lr: 0.000007 loss: 0.4965 (0.5277) time: 0.4337 data: 0.0001 max mem: 52150 +Epoch: [16] [1060/7269] eta: 0:45:29 lr: 0.000007 loss: 0.5371 (0.5277) time: 0.4338 data: 0.0002 max mem: 52150 +Epoch: [16] [1070/7269] eta: 0:45:34 lr: 0.000007 loss: 0.5272 (0.5276) time: 0.4796 data: 0.0002 max mem: 52150 +Epoch: [16] [1080/7269] eta: 0:45:31 lr: 0.000007 loss: 0.4941 (0.5274) time: 0.5256 data: 0.0002 max mem: 52150 +Epoch: [16] [1090/7269] eta: 0:45:22 lr: 0.000007 loss: 0.5150 (0.5274) time: 0.4182 data: 0.0002 max mem: 52150 +Epoch: [16] [1100/7269] eta: 0:45:19 lr: 0.000007 loss: 0.5220 (0.5273) time: 0.4175 data: 0.0001 max mem: 52150 +Epoch: [16] [1110/7269] eta: 0:45:11 lr: 0.000007 loss: 0.5442 (0.5276) time: 0.4181 data: 0.0001 max mem: 52150 +Epoch: [16] [1120/7269] eta: 0:45:03 lr: 0.000007 loss: 0.5392 (0.5278) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [16] [1130/7269] eta: 0:44:55 lr: 0.000007 loss: 0.5251 (0.5278) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [16] [1140/7269] eta: 0:44:51 lr: 0.000007 loss: 0.5265 (0.5279) time: 0.4156 data: 0.0002 max mem: 52150 +Epoch: [16] [1150/7269] eta: 0:44:43 lr: 0.000007 loss: 0.5051 (0.5276) time: 0.4155 data: 0.0002 max mem: 52150 +Epoch: [16] [1160/7269] eta: 0:44:35 lr: 0.000007 loss: 0.5440 (0.5279) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [16] [1170/7269] eta: 0:44:39 lr: 0.000007 loss: 0.5467 (0.5280) time: 0.4802 data: 0.0002 max mem: 52150 +Epoch: [16] [1180/7269] eta: 0:44:47 lr: 0.000007 loss: 0.5730 (0.5285) time: 0.6389 data: 0.0002 max mem: 52150 +Epoch: [16] [1190/7269] eta: 0:44:44 lr: 0.000007 loss: 0.5730 (0.5287) time: 0.5759 data: 0.0002 max mem: 52150 +Epoch: [16] [1200/7269] eta: 0:44:36 lr: 0.000007 loss: 0.5424 (0.5287) time: 0.4179 data: 0.0002 max mem: 52150 +Epoch: [16] [1210/7269] eta: 0:44:34 lr: 0.000007 loss: 0.4958 (0.5283) time: 0.4326 data: 0.0002 max mem: 52150 +Epoch: [16] [1220/7269] eta: 0:44:26 lr: 0.000007 loss: 0.4965 (0.5282) time: 0.4324 data: 0.0002 max mem: 52150 +Epoch: [16] [1230/7269] eta: 0:44:19 lr: 0.000007 loss: 0.5113 (0.5283) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [16] [1240/7269] eta: 0:44:11 lr: 0.000007 loss: 0.5736 (0.5287) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [16] [1250/7269] eta: 0:44:03 lr: 0.000007 loss: 0.5300 (0.5284) time: 0.3727 data: 0.0002 max mem: 52150 +Epoch: [16] [1260/7269] eta: 0:44:06 lr: 0.000007 loss: 0.5314 (0.5285) time: 0.4785 data: 0.0002 max mem: 52150 +Epoch: [16] [1270/7269] eta: 0:43:58 lr: 0.000007 loss: 0.5352 (0.5285) time: 0.4780 data: 0.0002 max mem: 52150 +Epoch: [16] [1280/7269] eta: 0:43:56 lr: 0.000007 loss: 0.5193 (0.5286) time: 0.4334 data: 0.0002 max mem: 52150 +Epoch: [16] [1290/7269] eta: 0:43:49 lr: 0.000007 loss: 0.5343 (0.5285) time: 0.4334 data: 0.0002 max mem: 52150 +Epoch: [16] [1300/7269] eta: 0:43:41 lr: 0.000007 loss: 0.5174 (0.5284) time: 0.3731 data: 0.0002 max mem: 52150 +Epoch: [16] [1310/7269] eta: 0:43:34 lr: 0.000007 loss: 0.5094 (0.5283) time: 0.3728 data: 0.0001 max mem: 52150 +Epoch: [16] [1320/7269] eta: 0:43:27 lr: 0.000007 loss: 0.5094 (0.5282) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [16] [1330/7269] eta: 0:43:19 lr: 0.000007 loss: 0.5487 (0.5283) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [16] [1340/7269] eta: 0:43:12 lr: 0.000007 loss: 0.5676 (0.5286) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [16] [1350/7269] eta: 0:43:10 lr: 0.000007 loss: 0.5484 (0.5287) time: 0.4319 data: 0.0002 max mem: 52150 +Epoch: [16] [1360/7269] eta: 0:43:08 lr: 0.000007 loss: 0.5287 (0.5287) time: 0.4932 data: 0.0002 max mem: 52150 +Epoch: [16] [1370/7269] eta: 0:43:06 lr: 0.000007 loss: 0.5385 (0.5286) time: 0.4956 data: 0.0001 max mem: 52150 +Epoch: [16] [1380/7269] eta: 0:43:04 lr: 0.000007 loss: 0.5470 (0.5286) time: 0.4949 data: 0.0002 max mem: 52150 +Epoch: [16] [1390/7269] eta: 0:42:57 lr: 0.000007 loss: 0.5046 (0.5284) time: 0.4325 data: 0.0002 max mem: 52150 +Epoch: [16] [1400/7269] eta: 0:42:50 lr: 0.000007 loss: 0.5250 (0.5284) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [16] [1410/7269] eta: 0:42:43 lr: 0.000007 loss: 0.5307 (0.5284) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [16] [1420/7269] eta: 0:42:36 lr: 0.000007 loss: 0.5253 (0.5284) time: 0.3723 data: 0.0001 max mem: 52150 +Epoch: [16] [1430/7269] eta: 0:42:29 lr: 0.000007 loss: 0.5228 (0.5283) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [16] [1440/7269] eta: 0:42:26 lr: 0.000007 loss: 0.5285 (0.5286) time: 0.4322 data: 0.0002 max mem: 52150 +Epoch: [16] [1450/7269] eta: 0:42:20 lr: 0.000007 loss: 0.5436 (0.5286) time: 0.4329 data: 0.0002 max mem: 52150 +Epoch: [16] [1460/7269] eta: 0:42:20 lr: 0.000007 loss: 0.5250 (0.5286) time: 0.4648 data: 0.0001 max mem: 52150 +Epoch: [16] [1470/7269] eta: 0:42:13 lr: 0.000007 loss: 0.5534 (0.5289) time: 0.4643 data: 0.0001 max mem: 52150 +Epoch: [16] [1480/7269] eta: 0:42:11 lr: 0.000007 loss: 0.5725 (0.5291) time: 0.4303 data: 0.0001 max mem: 52150 +Epoch: [16] [1490/7269] eta: 0:42:04 lr: 0.000007 loss: 0.5597 (0.5294) time: 0.4302 data: 0.0001 max mem: 52150 +Epoch: [16] [1500/7269] eta: 0:42:00 lr: 0.000007 loss: 0.5596 (0.5294) time: 0.4175 data: 0.0002 max mem: 52150 +Epoch: [16] [1510/7269] eta: 0:41:54 lr: 0.000007 loss: 0.5466 (0.5295) time: 0.4179 data: 0.0002 max mem: 52150 +Epoch: [16] [1520/7269] eta: 0:41:51 lr: 0.000007 loss: 0.5483 (0.5295) time: 0.4336 data: 0.0002 max mem: 52150 +Epoch: [16] [1530/7269] eta: 0:41:49 lr: 0.000007 loss: 0.5592 (0.5294) time: 0.4947 data: 0.0002 max mem: 52150 +Epoch: [16] [1540/7269] eta: 0:41:50 lr: 0.000007 loss: 0.5226 (0.5293) time: 0.5402 data: 0.0002 max mem: 52150 +Epoch: [16] [1550/7269] eta: 0:41:43 lr: 0.000007 loss: 0.5309 (0.5294) time: 0.4787 data: 0.0002 max mem: 52150 +Epoch: [16] [1560/7269] eta: 0:41:37 lr: 0.000007 loss: 0.5394 (0.5296) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [16] [1570/7269] eta: 0:41:30 lr: 0.000007 loss: 0.5439 (0.5296) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [16] [1580/7269] eta: 0:41:32 lr: 0.000007 loss: 0.5461 (0.5297) time: 0.4938 data: 0.0002 max mem: 52150 +Epoch: [16] [1590/7269] eta: 0:41:29 lr: 0.000007 loss: 0.5456 (0.5297) time: 0.5422 data: 0.0001 max mem: 52150 +Epoch: [16] [1600/7269] eta: 0:41:26 lr: 0.000007 loss: 0.5483 (0.5299) time: 0.4812 data: 0.0001 max mem: 52150 +Epoch: [16] [1610/7269] eta: 0:41:24 lr: 0.000007 loss: 0.5540 (0.5298) time: 0.4922 data: 0.0001 max mem: 52150 +Epoch: [16] [1620/7269] eta: 0:41:17 lr: 0.000007 loss: 0.5484 (0.5297) time: 0.4310 data: 0.0001 max mem: 52150 +Epoch: [16] [1630/7269] eta: 0:41:14 lr: 0.000007 loss: 0.4829 (0.5295) time: 0.4306 data: 0.0001 max mem: 52150 +Epoch: [16] [1640/7269] eta: 0:41:08 lr: 0.000007 loss: 0.5340 (0.5296) time: 0.4306 data: 0.0002 max mem: 52150 +Epoch: [16] [1650/7269] eta: 0:41:01 lr: 0.000007 loss: 0.5327 (0.5295) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [16] [1660/7269] eta: 0:40:58 lr: 0.000007 loss: 0.5470 (0.5298) time: 0.4325 data: 0.0002 max mem: 52150 +Epoch: [16] [1670/7269] eta: 0:40:59 lr: 0.000007 loss: 0.5470 (0.5296) time: 0.5387 data: 0.0002 max mem: 52150 +Epoch: [16] [1680/7269] eta: 0:40:52 lr: 0.000007 loss: 0.4967 (0.5295) time: 0.4782 data: 0.0002 max mem: 52150 +Epoch: [16] [1690/7269] eta: 0:40:46 lr: 0.000007 loss: 0.5118 (0.5295) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [16] [1700/7269] eta: 0:40:39 lr: 0.000007 loss: 0.5710 (0.5299) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [16] [1710/7269] eta: 0:40:33 lr: 0.000007 loss: 0.5272 (0.5297) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [16] [1720/7269] eta: 0:40:29 lr: 0.000007 loss: 0.5272 (0.5299) time: 0.4180 data: 0.0002 max mem: 52150 +Epoch: [16] [1730/7269] eta: 0:40:26 lr: 0.000007 loss: 0.5252 (0.5298) time: 0.4674 data: 0.0001 max mem: 52150 +Epoch: [16] [1740/7269] eta: 0:40:22 lr: 0.000007 loss: 0.4965 (0.5295) time: 0.4684 data: 0.0001 max mem: 52150 +Epoch: [16] [1750/7269] eta: 0:40:20 lr: 0.000007 loss: 0.5254 (0.5295) time: 0.4801 data: 0.0002 max mem: 52150 +Epoch: [16] [1760/7269] eta: 0:40:13 lr: 0.000007 loss: 0.5473 (0.5295) time: 0.4337 data: 0.0002 max mem: 52150 +Epoch: [16] [1770/7269] eta: 0:40:07 lr: 0.000007 loss: 0.5500 (0.5295) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [16] [1780/7269] eta: 0:40:00 lr: 0.000007 loss: 0.5202 (0.5294) time: 0.3728 data: 0.0002 max mem: 52150 +Epoch: [16] [1790/7269] eta: 0:39:57 lr: 0.000007 loss: 0.4977 (0.5294) time: 0.4189 data: 0.0002 max mem: 52150 +Epoch: [16] [1800/7269] eta: 0:39:51 lr: 0.000007 loss: 0.5450 (0.5294) time: 0.4185 data: 0.0001 max mem: 52150 +Epoch: [16] [1810/7269] eta: 0:39:44 lr: 0.000007 loss: 0.5483 (0.5295) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [16] [1820/7269] eta: 0:39:42 lr: 0.000007 loss: 0.5360 (0.5295) time: 0.4326 data: 0.0002 max mem: 52150 +Epoch: [16] [1830/7269] eta: 0:39:35 lr: 0.000007 loss: 0.5205 (0.5294) time: 0.4333 data: 0.0002 max mem: 52150 +Epoch: [16] [1840/7269] eta: 0:39:32 lr: 0.000007 loss: 0.5544 (0.5297) time: 0.4217 data: 0.0002 max mem: 52150 +Epoch: [16] [1850/7269] eta: 0:39:26 lr: 0.000007 loss: 0.5544 (0.5295) time: 0.4207 data: 0.0001 max mem: 52150 +Epoch: [16] [1860/7269] eta: 0:39:19 lr: 0.000007 loss: 0.5199 (0.5295) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [16] [1870/7269] eta: 0:39:16 lr: 0.000007 loss: 0.5271 (0.5296) time: 0.4161 data: 0.0002 max mem: 52150 +Epoch: [16] [1880/7269] eta: 0:39:09 lr: 0.000007 loss: 0.5494 (0.5295) time: 0.4158 data: 0.0002 max mem: 52150 +Epoch: [16] [1890/7269] eta: 0:39:03 lr: 0.000007 loss: 0.5448 (0.5295) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [16] [1900/7269] eta: 0:39:01 lr: 0.000007 loss: 0.5448 (0.5295) time: 0.4334 data: 0.0002 max mem: 52150 +Epoch: [16] [1910/7269] eta: 0:38:54 lr: 0.000007 loss: 0.5505 (0.5295) time: 0.4328 data: 0.0002 max mem: 52150 +Epoch: [16] [1920/7269] eta: 0:38:52 lr: 0.000007 loss: 0.5505 (0.5295) time: 0.4330 data: 0.0002 max mem: 52150 +Epoch: [16] [1930/7269] eta: 0:38:46 lr: 0.000007 loss: 0.5608 (0.5295) time: 0.4330 data: 0.0001 max mem: 52150 +Epoch: [16] [1940/7269] eta: 0:38:42 lr: 0.000007 loss: 0.5665 (0.5294) time: 0.4187 data: 0.0001 max mem: 52150 +Epoch: [16] [1950/7269] eta: 0:38:38 lr: 0.000007 loss: 0.5747 (0.5296) time: 0.4634 data: 0.0001 max mem: 52150 +Epoch: [16] [1960/7269] eta: 0:38:32 lr: 0.000007 loss: 0.5243 (0.5295) time: 0.4167 data: 0.0002 max mem: 52150 +Epoch: [16] [1970/7269] eta: 0:38:30 lr: 0.000007 loss: 0.4978 (0.5296) time: 0.4332 data: 0.0002 max mem: 52150 +Epoch: [16] [1980/7269] eta: 0:38:23 lr: 0.000007 loss: 0.5248 (0.5294) time: 0.4330 data: 0.0002 max mem: 52150 +Epoch: [16] [1990/7269] eta: 0:38:20 lr: 0.000007 loss: 0.5189 (0.5295) time: 0.4182 data: 0.0002 max mem: 52150 +Epoch: [16] [2000/7269] eta: 0:38:17 lr: 0.000007 loss: 0.5469 (0.5296) time: 0.4813 data: 0.0001 max mem: 52150 +Epoch: [16] [2010/7269] eta: 0:38:11 lr: 0.000007 loss: 0.5525 (0.5296) time: 0.4349 data: 0.0002 max mem: 52150 +Epoch: [16] [2020/7269] eta: 0:38:05 lr: 0.000007 loss: 0.5318 (0.5296) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [16] [2030/7269] eta: 0:38:01 lr: 0.000007 loss: 0.5187 (0.5294) time: 0.4182 data: 0.0002 max mem: 52150 +Epoch: [16] [2040/7269] eta: 0:37:55 lr: 0.000007 loss: 0.5187 (0.5294) time: 0.4183 data: 0.0002 max mem: 52150 +Epoch: [16] [2050/7269] eta: 0:37:52 lr: 0.000007 loss: 0.5369 (0.5293) time: 0.4233 data: 0.0002 max mem: 52150 +Epoch: [16] [2060/7269] eta: 0:37:49 lr: 0.000007 loss: 0.5157 (0.5293) time: 0.4719 data: 0.0002 max mem: 52150 +Epoch: [16] [2070/7269] eta: 0:37:46 lr: 0.000007 loss: 0.5157 (0.5293) time: 0.4804 data: 0.0001 max mem: 52150 +Epoch: [16] [2080/7269] eta: 0:37:40 lr: 0.000007 loss: 0.4876 (0.5293) time: 0.4314 data: 0.0002 max mem: 52150 +Epoch: [16] [2090/7269] eta: 0:37:34 lr: 0.000007 loss: 0.5331 (0.5294) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [16] [2100/7269] eta: 0:37:33 lr: 0.000007 loss: 0.5331 (0.5294) time: 0.4836 data: 0.0001 max mem: 52150 +Epoch: [16] [2110/7269] eta: 0:37:30 lr: 0.000007 loss: 0.5242 (0.5294) time: 0.5440 data: 0.0002 max mem: 52150 +Epoch: [16] [2120/7269] eta: 0:37:25 lr: 0.000007 loss: 0.5316 (0.5294) time: 0.4330 data: 0.0002 max mem: 52150 +Epoch: [16] [2130/7269] eta: 0:37:27 lr: 0.000007 loss: 0.5461 (0.5295) time: 0.5529 data: 0.0001 max mem: 52150 +Epoch: [16] [2140/7269] eta: 0:37:24 lr: 0.000007 loss: 0.5469 (0.5297) time: 0.6116 data: 0.0001 max mem: 52150 +Epoch: [16] [2150/7269] eta: 0:37:18 lr: 0.000007 loss: 0.5625 (0.5298) time: 0.4315 data: 0.0001 max mem: 52150 +Epoch: [16] [2160/7269] eta: 0:37:15 lr: 0.000007 loss: 0.5578 (0.5299) time: 0.4208 data: 0.0001 max mem: 52150 +Epoch: [16] [2170/7269] eta: 0:37:12 lr: 0.000007 loss: 0.5549 (0.5299) time: 0.4812 data: 0.0001 max mem: 52150 +Epoch: [16] [2180/7269] eta: 0:37:08 lr: 0.000007 loss: 0.5462 (0.5300) time: 0.4788 data: 0.0001 max mem: 52150 +Epoch: [16] [2190/7269] eta: 0:37:05 lr: 0.000007 loss: 0.5202 (0.5299) time: 0.4784 data: 0.0001 max mem: 52150 +Epoch: [16] [2200/7269] eta: 0:37:02 lr: 0.000007 loss: 0.5302 (0.5300) time: 0.4925 data: 0.0002 max mem: 52150 +Epoch: [16] [2210/7269] eta: 0:36:56 lr: 0.000007 loss: 0.5593 (0.5302) time: 0.4323 data: 0.0001 max mem: 52150 +Epoch: [16] [2220/7269] eta: 0:36:50 lr: 0.000007 loss: 0.5593 (0.5304) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [16] [2230/7269] eta: 0:36:44 lr: 0.000007 loss: 0.5511 (0.5303) time: 0.3721 data: 0.0001 max mem: 52150 +Epoch: [16] [2240/7269] eta: 0:36:40 lr: 0.000007 loss: 0.4641 (0.5300) time: 0.4175 data: 0.0002 max mem: 52150 +Epoch: [16] [2250/7269] eta: 0:36:34 lr: 0.000007 loss: 0.4775 (0.5299) time: 0.4173 data: 0.0002 max mem: 52150 +Epoch: [16] [2260/7269] eta: 0:36:30 lr: 0.000007 loss: 0.5562 (0.5299) time: 0.4175 data: 0.0001 max mem: 52150 +Epoch: [16] [2270/7269] eta: 0:36:25 lr: 0.000007 loss: 0.5586 (0.5301) time: 0.4176 data: 0.0001 max mem: 52150 +Epoch: [16] [2280/7269] eta: 0:36:21 lr: 0.000007 loss: 0.5338 (0.5300) time: 0.4207 data: 0.0002 max mem: 52150 +Epoch: [16] [2290/7269] eta: 0:36:15 lr: 0.000007 loss: 0.5349 (0.5301) time: 0.4209 data: 0.0002 max mem: 52150 +Epoch: [16] [2300/7269] eta: 0:36:09 lr: 0.000007 loss: 0.5414 (0.5302) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [16] [2310/7269] eta: 0:36:08 lr: 0.000007 loss: 0.5405 (0.5302) time: 0.4615 data: 0.0002 max mem: 52150 +Epoch: [16] [2320/7269] eta: 0:36:02 lr: 0.000007 loss: 0.5419 (0.5302) time: 0.4619 data: 0.0002 max mem: 52150 +Epoch: [16] [2330/7269] eta: 0:35:59 lr: 0.000007 loss: 0.5043 (0.5301) time: 0.4334 data: 0.0001 max mem: 52150 +Epoch: [16] [2340/7269] eta: 0:35:53 lr: 0.000007 loss: 0.4973 (0.5301) time: 0.4327 data: 0.0001 max mem: 52150 +Epoch: [16] [2350/7269] eta: 0:35:47 lr: 0.000007 loss: 0.5302 (0.5300) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [16] [2360/7269] eta: 0:35:41 lr: 0.000007 loss: 0.5557 (0.5302) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [16] [2370/7269] eta: 0:35:36 lr: 0.000007 loss: 0.5413 (0.5300) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [16] [2380/7269] eta: 0:35:30 lr: 0.000007 loss: 0.5035 (0.5300) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [16] [2390/7269] eta: 0:35:24 lr: 0.000007 loss: 0.5393 (0.5301) time: 0.3728 data: 0.0002 max mem: 52150 +Epoch: [16] [2400/7269] eta: 0:35:21 lr: 0.000007 loss: 0.5523 (0.5303) time: 0.4184 data: 0.0002 max mem: 52150 +Epoch: [16] [2410/7269] eta: 0:35:17 lr: 0.000007 loss: 0.5600 (0.5304) time: 0.4780 data: 0.0002 max mem: 52150 +Epoch: [16] [2420/7269] eta: 0:35:14 lr: 0.000007 loss: 0.5600 (0.5305) time: 0.4785 data: 0.0002 max mem: 52150 +Epoch: [16] [2430/7269] eta: 0:35:08 lr: 0.000007 loss: 0.5569 (0.5306) time: 0.4190 data: 0.0002 max mem: 52150 +Epoch: [16] [2440/7269] eta: 0:35:02 lr: 0.000007 loss: 0.5444 (0.5306) time: 0.3730 data: 0.0002 max mem: 52150 +Epoch: [16] [2450/7269] eta: 0:34:57 lr: 0.000007 loss: 0.5280 (0.5305) time: 0.3727 data: 0.0002 max mem: 52150 +Epoch: [16] [2460/7269] eta: 0:34:53 lr: 0.000007 loss: 0.5239 (0.5304) time: 0.4226 data: 0.0002 max mem: 52150 +Epoch: [16] [2470/7269] eta: 0:34:50 lr: 0.000007 loss: 0.5214 (0.5304) time: 0.4835 data: 0.0002 max mem: 52150 +Epoch: [16] [2480/7269] eta: 0:34:44 lr: 0.000007 loss: 0.5435 (0.5304) time: 0.4332 data: 0.0002 max mem: 52150 +Epoch: [16] [2490/7269] eta: 0:34:39 lr: 0.000007 loss: 0.5482 (0.5305) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [16] [2500/7269] eta: 0:34:35 lr: 0.000007 loss: 0.5062 (0.5304) time: 0.4201 data: 0.0002 max mem: 52150 +Epoch: [16] [2510/7269] eta: 0:34:32 lr: 0.000007 loss: 0.5288 (0.5304) time: 0.4704 data: 0.0001 max mem: 52150 +Epoch: [16] [2520/7269] eta: 0:34:28 lr: 0.000007 loss: 0.5398 (0.5305) time: 0.4677 data: 0.0001 max mem: 52150 +Epoch: [16] [2530/7269] eta: 0:34:24 lr: 0.000007 loss: 0.5684 (0.5305) time: 0.4653 data: 0.0001 max mem: 52150 +Epoch: [16] [2540/7269] eta: 0:34:22 lr: 0.000007 loss: 0.5335 (0.5304) time: 0.5287 data: 0.0001 max mem: 52150 +Epoch: [16] [2550/7269] eta: 0:34:17 lr: 0.000007 loss: 0.5335 (0.5305) time: 0.4808 data: 0.0002 max mem: 52150 +Epoch: [16] [2560/7269] eta: 0:34:14 lr: 0.000007 loss: 0.5340 (0.5303) time: 0.4312 data: 0.0002 max mem: 52150 +Epoch: [16] [2570/7269] eta: 0:34:08 lr: 0.000007 loss: 0.5157 (0.5303) time: 0.4316 data: 0.0002 max mem: 52150 +Epoch: [16] [2580/7269] eta: 0:34:03 lr: 0.000007 loss: 0.5043 (0.5302) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [16] [2590/7269] eta: 0:33:59 lr: 0.000007 loss: 0.5043 (0.5302) time: 0.4160 data: 0.0002 max mem: 52150 +Epoch: [16] [2600/7269] eta: 0:33:57 lr: 0.000007 loss: 0.5358 (0.5302) time: 0.5366 data: 0.0001 max mem: 52150 +Epoch: [16] [2610/7269] eta: 0:33:52 lr: 0.000007 loss: 0.5199 (0.5301) time: 0.4925 data: 0.0002 max mem: 52150 +Epoch: [16] [2620/7269] eta: 0:33:49 lr: 0.000007 loss: 0.5285 (0.5301) time: 0.4325 data: 0.0002 max mem: 52150 +Epoch: [16] [2630/7269] eta: 0:33:43 lr: 0.000007 loss: 0.5285 (0.5300) time: 0.4322 data: 0.0002 max mem: 52150 +Epoch: [16] [2640/7269] eta: 0:33:38 lr: 0.000007 loss: 0.5148 (0.5300) time: 0.3712 data: 0.0002 max mem: 52150 +Epoch: [16] [2650/7269] eta: 0:33:36 lr: 0.000007 loss: 0.5386 (0.5300) time: 0.4787 data: 0.0002 max mem: 52150 +Epoch: [16] [2660/7269] eta: 0:33:30 lr: 0.000007 loss: 0.5161 (0.5299) time: 0.4789 data: 0.0002 max mem: 52150 +Epoch: [16] [2670/7269] eta: 0:33:27 lr: 0.000007 loss: 0.5161 (0.5300) time: 0.4331 data: 0.0002 max mem: 52150 +Epoch: [16] [2680/7269] eta: 0:33:21 lr: 0.000007 loss: 0.5509 (0.5301) time: 0.4330 data: 0.0002 max mem: 52150 +Epoch: [16] [2690/7269] eta: 0:33:19 lr: 0.000007 loss: 0.5306 (0.5299) time: 0.4687 data: 0.0002 max mem: 52150 +Epoch: [16] [2700/7269] eta: 0:33:16 lr: 0.000007 loss: 0.4876 (0.5299) time: 0.5299 data: 0.0003 max mem: 52150 +Epoch: [16] [2710/7269] eta: 0:33:10 lr: 0.000007 loss: 0.5223 (0.5299) time: 0.4328 data: 0.0003 max mem: 52150 +Epoch: [16] [2720/7269] eta: 0:33:05 lr: 0.000007 loss: 0.5419 (0.5299) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [16] [2730/7269] eta: 0:33:00 lr: 0.000007 loss: 0.5253 (0.5298) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [16] [2740/7269] eta: 0:32:54 lr: 0.000007 loss: 0.5253 (0.5299) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [16] [2750/7269] eta: 0:32:49 lr: 0.000007 loss: 0.5632 (0.5300) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [16] [2760/7269] eta: 0:32:43 lr: 0.000007 loss: 0.5698 (0.5301) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [16] [2770/7269] eta: 0:32:38 lr: 0.000007 loss: 0.5789 (0.5303) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [16] [2780/7269] eta: 0:32:33 lr: 0.000007 loss: 0.5547 (0.5303) time: 0.3722 data: 0.0001 max mem: 52150 +Epoch: [16] [2790/7269] eta: 0:32:29 lr: 0.000007 loss: 0.5344 (0.5304) time: 0.4324 data: 0.0002 max mem: 52150 +Epoch: [16] [2800/7269] eta: 0:32:24 lr: 0.000007 loss: 0.5485 (0.5304) time: 0.4321 data: 0.0002 max mem: 52150 +Epoch: [16] [2810/7269] eta: 0:32:20 lr: 0.000007 loss: 0.5418 (0.5305) time: 0.4174 data: 0.0002 max mem: 52150 +Epoch: [16] [2820/7269] eta: 0:32:14 lr: 0.000007 loss: 0.5291 (0.5305) time: 0.4181 data: 0.0002 max mem: 52150 +Epoch: [16] [2830/7269] eta: 0:32:09 lr: 0.000007 loss: 0.5291 (0.5305) time: 0.3728 data: 0.0002 max mem: 52150 +Epoch: [16] [2840/7269] eta: 0:32:04 lr: 0.000007 loss: 0.5456 (0.5305) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [16] [2850/7269] eta: 0:31:59 lr: 0.000007 loss: 0.5568 (0.5306) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [16] [2860/7269] eta: 0:31:53 lr: 0.000007 loss: 0.5536 (0.5306) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [16] [2870/7269] eta: 0:31:51 lr: 0.000007 loss: 0.5159 (0.5305) time: 0.4770 data: 0.0002 max mem: 52150 +Epoch: [16] [2880/7269] eta: 0:31:47 lr: 0.000007 loss: 0.4999 (0.5305) time: 0.5243 data: 0.0002 max mem: 52150 +Epoch: [16] [2890/7269] eta: 0:31:44 lr: 0.000007 loss: 0.4999 (0.5304) time: 0.4791 data: 0.0002 max mem: 52150 +Epoch: [16] [2900/7269] eta: 0:31:38 lr: 0.000007 loss: 0.5203 (0.5304) time: 0.4318 data: 0.0002 max mem: 52150 +Epoch: [16] [2910/7269] eta: 0:31:35 lr: 0.000007 loss: 0.5309 (0.5305) time: 0.4180 data: 0.0002 max mem: 52150 +Epoch: [16] [2920/7269] eta: 0:31:32 lr: 0.000007 loss: 0.5543 (0.5304) time: 0.5229 data: 0.0002 max mem: 52150 +Epoch: [16] [2930/7269] eta: 0:31:28 lr: 0.000007 loss: 0.5543 (0.5305) time: 0.5225 data: 0.0001 max mem: 52150 +Epoch: [16] [2940/7269] eta: 0:31:23 lr: 0.000007 loss: 0.5461 (0.5305) time: 0.4174 data: 0.0002 max mem: 52150 +Epoch: [16] [2950/7269] eta: 0:31:21 lr: 0.000007 loss: 0.5203 (0.5304) time: 0.4779 data: 0.0002 max mem: 52150 +Epoch: [16] [2960/7269] eta: 0:31:16 lr: 0.000007 loss: 0.4893 (0.5302) time: 0.4781 data: 0.0001 max mem: 52150 +Epoch: [16] [2970/7269] eta: 0:31:12 lr: 0.000007 loss: 0.4890 (0.5301) time: 0.4193 data: 0.0001 max mem: 52150 +Epoch: [16] [2980/7269] eta: 0:31:08 lr: 0.000007 loss: 0.5371 (0.5302) time: 0.4812 data: 0.0001 max mem: 52150 +Epoch: [16] [2990/7269] eta: 0:31:04 lr: 0.000007 loss: 0.5465 (0.5302) time: 0.4796 data: 0.0001 max mem: 52150 +Epoch: [16] [3000/7269] eta: 0:31:01 lr: 0.000007 loss: 0.5237 (0.5302) time: 0.4786 data: 0.0001 max mem: 52150 +Epoch: [16] [3010/7269] eta: 0:30:56 lr: 0.000007 loss: 0.5180 (0.5302) time: 0.4326 data: 0.0001 max mem: 52150 +Epoch: [16] [3020/7269] eta: 0:30:52 lr: 0.000007 loss: 0.5189 (0.5302) time: 0.4180 data: 0.0002 max mem: 52150 +Epoch: [16] [3030/7269] eta: 0:30:48 lr: 0.000007 loss: 0.5338 (0.5302) time: 0.4674 data: 0.0002 max mem: 52150 +Epoch: [16] [3040/7269] eta: 0:30:44 lr: 0.000007 loss: 0.5392 (0.5302) time: 0.4798 data: 0.0002 max mem: 52150 +Epoch: [16] [3050/7269] eta: 0:30:40 lr: 0.000007 loss: 0.5162 (0.5301) time: 0.4785 data: 0.0002 max mem: 52150 +Epoch: [16] [3060/7269] eta: 0:30:35 lr: 0.000007 loss: 0.5162 (0.5301) time: 0.4197 data: 0.0001 max mem: 52150 +Epoch: [16] [3070/7269] eta: 0:30:31 lr: 0.000007 loss: 0.5404 (0.5301) time: 0.4245 data: 0.0001 max mem: 52150 +Epoch: [16] [3080/7269] eta: 0:30:29 lr: 0.000007 loss: 0.5459 (0.5302) time: 0.5308 data: 0.0001 max mem: 52150 +Epoch: [16] [3090/7269] eta: 0:30:24 lr: 0.000007 loss: 0.5526 (0.5302) time: 0.4780 data: 0.0002 max mem: 52150 +Epoch: [16] [3100/7269] eta: 0:30:18 lr: 0.000007 loss: 0.5526 (0.5302) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [16] [3110/7269] eta: 0:30:13 lr: 0.000007 loss: 0.5175 (0.5301) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [16] [3120/7269] eta: 0:30:11 lr: 0.000007 loss: 0.4973 (0.5300) time: 0.4814 data: 0.0001 max mem: 52150 +Epoch: [16] [3130/7269] eta: 0:30:07 lr: 0.000007 loss: 0.5282 (0.5301) time: 0.5401 data: 0.0002 max mem: 52150 +Epoch: [16] [3140/7269] eta: 0:30:04 lr: 0.000007 loss: 0.5398 (0.5301) time: 0.4910 data: 0.0002 max mem: 52150 +Epoch: [16] [3150/7269] eta: 0:29:58 lr: 0.000007 loss: 0.5311 (0.5301) time: 0.4322 data: 0.0002 max mem: 52150 +Epoch: [16] [3160/7269] eta: 0:29:53 lr: 0.000007 loss: 0.5154 (0.5300) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [16] [3170/7269] eta: 0:29:48 lr: 0.000007 loss: 0.5114 (0.5300) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [16] [3180/7269] eta: 0:29:43 lr: 0.000007 loss: 0.5241 (0.5300) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [16] [3190/7269] eta: 0:29:38 lr: 0.000007 loss: 0.5172 (0.5300) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [16] [3200/7269] eta: 0:29:32 lr: 0.000007 loss: 0.5531 (0.5300) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [16] [3210/7269] eta: 0:29:27 lr: 0.000007 loss: 0.5611 (0.5301) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [16] [3220/7269] eta: 0:29:22 lr: 0.000007 loss: 0.5361 (0.5300) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [16] [3230/7269] eta: 0:29:17 lr: 0.000007 loss: 0.5351 (0.5300) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [16] [3240/7269] eta: 0:29:12 lr: 0.000007 loss: 0.5366 (0.5301) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [16] [3250/7269] eta: 0:29:07 lr: 0.000007 loss: 0.5325 (0.5300) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [16] [3260/7269] eta: 0:29:03 lr: 0.000007 loss: 0.5060 (0.5301) time: 0.4330 data: 0.0002 max mem: 52150 +Epoch: [16] [3270/7269] eta: 0:28:58 lr: 0.000007 loss: 0.5032 (0.5299) time: 0.4330 data: 0.0002 max mem: 52150 +Epoch: [16] [3280/7269] eta: 0:28:54 lr: 0.000007 loss: 0.4972 (0.5299) time: 0.4173 data: 0.0002 max mem: 52150 +Epoch: [16] [3290/7269] eta: 0:28:49 lr: 0.000007 loss: 0.5338 (0.5299) time: 0.4172 data: 0.0002 max mem: 52150 +Epoch: [16] [3300/7269] eta: 0:28:44 lr: 0.000007 loss: 0.4913 (0.5297) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [16] [3310/7269] eta: 0:28:41 lr: 0.000007 loss: 0.4810 (0.5297) time: 0.4645 data: 0.0002 max mem: 52150 +Epoch: [16] [3320/7269] eta: 0:28:36 lr: 0.000007 loss: 0.5031 (0.5296) time: 0.4648 data: 0.0002 max mem: 52150 +Epoch: [16] [3330/7269] eta: 0:28:31 lr: 0.000007 loss: 0.5018 (0.5296) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [16] [3340/7269] eta: 0:28:27 lr: 0.000007 loss: 0.5433 (0.5296) time: 0.4308 data: 0.0002 max mem: 52150 +Epoch: [16] [3350/7269] eta: 0:28:23 lr: 0.000007 loss: 0.5456 (0.5296) time: 0.4931 data: 0.0002 max mem: 52150 +Epoch: [16] [3360/7269] eta: 0:28:18 lr: 0.000007 loss: 0.5202 (0.5296) time: 0.4340 data: 0.0002 max mem: 52150 +Epoch: [16] [3370/7269] eta: 0:28:15 lr: 0.000007 loss: 0.5365 (0.5296) time: 0.4327 data: 0.0001 max mem: 52150 +Epoch: [16] [3380/7269] eta: 0:28:09 lr: 0.000007 loss: 0.5192 (0.5295) time: 0.4325 data: 0.0002 max mem: 52150 +Epoch: [16] [3390/7269] eta: 0:28:04 lr: 0.000007 loss: 0.5077 (0.5295) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [16] [3400/7269] eta: 0:28:00 lr: 0.000007 loss: 0.5175 (0.5296) time: 0.4174 data: 0.0002 max mem: 52150 +Epoch: [16] [3410/7269] eta: 0:27:56 lr: 0.000007 loss: 0.5408 (0.5296) time: 0.4627 data: 0.0002 max mem: 52150 +Epoch: [16] [3420/7269] eta: 0:27:51 lr: 0.000007 loss: 0.5408 (0.5296) time: 0.4170 data: 0.0002 max mem: 52150 +Epoch: [16] [3430/7269] eta: 0:27:46 lr: 0.000007 loss: 0.5253 (0.5296) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [16] [3440/7269] eta: 0:27:43 lr: 0.000007 loss: 0.5279 (0.5296) time: 0.4336 data: 0.0002 max mem: 52150 +Epoch: [16] [3450/7269] eta: 0:27:38 lr: 0.000007 loss: 0.5508 (0.5296) time: 0.4331 data: 0.0002 max mem: 52150 +Epoch: [16] [3460/7269] eta: 0:27:33 lr: 0.000007 loss: 0.5332 (0.5296) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [16] [3470/7269] eta: 0:27:28 lr: 0.000007 loss: 0.5570 (0.5296) time: 0.3723 data: 0.0001 max mem: 52150 +Epoch: [16] [3480/7269] eta: 0:27:22 lr: 0.000007 loss: 0.5570 (0.5296) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [16] [3490/7269] eta: 0:27:17 lr: 0.000007 loss: 0.5284 (0.5296) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [16] [3500/7269] eta: 0:27:14 lr: 0.000007 loss: 0.4908 (0.5296) time: 0.4331 data: 0.0001 max mem: 52150 +Epoch: [16] [3510/7269] eta: 0:27:09 lr: 0.000007 loss: 0.5190 (0.5295) time: 0.4330 data: 0.0002 max mem: 52150 +Epoch: [16] [3520/7269] eta: 0:27:05 lr: 0.000007 loss: 0.5286 (0.5296) time: 0.4199 data: 0.0001 max mem: 52150 +Epoch: [16] [3530/7269] eta: 0:27:01 lr: 0.000007 loss: 0.5353 (0.5295) time: 0.4681 data: 0.0001 max mem: 52150 +Epoch: [16] [3540/7269] eta: 0:26:58 lr: 0.000007 loss: 0.5353 (0.5296) time: 0.5146 data: 0.0001 max mem: 52150 +Epoch: [16] [3550/7269] eta: 0:26:54 lr: 0.000007 loss: 0.5458 (0.5295) time: 0.5112 data: 0.0001 max mem: 52150 +Epoch: [16] [3560/7269] eta: 0:26:49 lr: 0.000007 loss: 0.5185 (0.5295) time: 0.4165 data: 0.0002 max mem: 52150 +Epoch: [16] [3570/7269] eta: 0:26:45 lr: 0.000007 loss: 0.5288 (0.5295) time: 0.4326 data: 0.0002 max mem: 52150 +Epoch: [16] [3580/7269] eta: 0:26:41 lr: 0.000007 loss: 0.5461 (0.5296) time: 0.4952 data: 0.0002 max mem: 52150 +Epoch: [16] [3590/7269] eta: 0:26:37 lr: 0.000007 loss: 0.5491 (0.5296) time: 0.4801 data: 0.0002 max mem: 52150 +Epoch: [16] [3600/7269] eta: 0:26:32 lr: 0.000007 loss: 0.5645 (0.5297) time: 0.4172 data: 0.0002 max mem: 52150 +Epoch: [16] [3610/7269] eta: 0:26:31 lr: 0.000007 loss: 0.5500 (0.5297) time: 0.5374 data: 0.0002 max mem: 52150 +Epoch: [16] [3620/7269] eta: 0:26:26 lr: 0.000007 loss: 0.5617 (0.5298) time: 0.5376 data: 0.0002 max mem: 52150 +Epoch: [16] [3630/7269] eta: 0:26:22 lr: 0.000007 loss: 0.5359 (0.5296) time: 0.4322 data: 0.0002 max mem: 52150 +Epoch: [16] [3640/7269] eta: 0:26:19 lr: 0.000007 loss: 0.5213 (0.5297) time: 0.5378 data: 0.0002 max mem: 52150 +Epoch: [16] [3650/7269] eta: 0:26:14 lr: 0.000007 loss: 0.5411 (0.5296) time: 0.4771 data: 0.0002 max mem: 52150 +Epoch: [16] [3660/7269] eta: 0:26:09 lr: 0.000007 loss: 0.5178 (0.5296) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [16] [3670/7269] eta: 0:26:04 lr: 0.000007 loss: 0.5178 (0.5295) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [16] [3680/7269] eta: 0:25:59 lr: 0.000007 loss: 0.5351 (0.5296) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [16] [3690/7269] eta: 0:25:54 lr: 0.000007 loss: 0.5610 (0.5295) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [16] [3700/7269] eta: 0:25:49 lr: 0.000007 loss: 0.5384 (0.5296) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [16] [3710/7269] eta: 0:25:44 lr: 0.000007 loss: 0.5498 (0.5297) time: 0.3711 data: 0.0002 max mem: 52150 +Epoch: [16] [3720/7269] eta: 0:25:39 lr: 0.000007 loss: 0.5482 (0.5296) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [16] [3730/7269] eta: 0:25:35 lr: 0.000007 loss: 0.5301 (0.5296) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [16] [3740/7269] eta: 0:25:30 lr: 0.000007 loss: 0.5579 (0.5297) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [16] [3750/7269] eta: 0:25:25 lr: 0.000007 loss: 0.5779 (0.5298) time: 0.3711 data: 0.0002 max mem: 52150 +Epoch: [16] [3760/7269] eta: 0:25:22 lr: 0.000007 loss: 0.5337 (0.5298) time: 0.4786 data: 0.0002 max mem: 52150 +Epoch: [16] [3770/7269] eta: 0:25:17 lr: 0.000007 loss: 0.5082 (0.5297) time: 0.4792 data: 0.0001 max mem: 52150 +Epoch: [16] [3780/7269] eta: 0:25:13 lr: 0.000007 loss: 0.5434 (0.5298) time: 0.4172 data: 0.0001 max mem: 52150 +Epoch: [16] [3790/7269] eta: 0:25:08 lr: 0.000007 loss: 0.5297 (0.5298) time: 0.4173 data: 0.0002 max mem: 52150 +Epoch: [16] [3800/7269] eta: 0:25:03 lr: 0.000007 loss: 0.5296 (0.5298) time: 0.3721 data: 0.0001 max mem: 52150 +Epoch: [16] [3810/7269] eta: 0:24:58 lr: 0.000007 loss: 0.5296 (0.5298) time: 0.3722 data: 0.0001 max mem: 52150 +Epoch: [16] [3820/7269] eta: 0:24:53 lr: 0.000007 loss: 0.5188 (0.5297) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [16] [3830/7269] eta: 0:24:48 lr: 0.000007 loss: 0.5364 (0.5299) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [16] [3840/7269] eta: 0:24:44 lr: 0.000007 loss: 0.5364 (0.5298) time: 0.4197 data: 0.0002 max mem: 52150 +Epoch: [16] [3850/7269] eta: 0:24:41 lr: 0.000007 loss: 0.5224 (0.5298) time: 0.4802 data: 0.0001 max mem: 52150 +Epoch: [16] [3860/7269] eta: 0:24:36 lr: 0.000007 loss: 0.5467 (0.5299) time: 0.4325 data: 0.0001 max mem: 52150 +Epoch: [16] [3870/7269] eta: 0:24:33 lr: 0.000007 loss: 0.5364 (0.5298) time: 0.4827 data: 0.0001 max mem: 52150 +Epoch: [16] [3880/7269] eta: 0:24:28 lr: 0.000007 loss: 0.5534 (0.5299) time: 0.4824 data: 0.0001 max mem: 52150 +Epoch: [16] [3890/7269] eta: 0:24:23 lr: 0.000007 loss: 0.5305 (0.5299) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [16] [3900/7269] eta: 0:24:20 lr: 0.000007 loss: 0.5185 (0.5298) time: 0.4792 data: 0.0001 max mem: 52150 +Epoch: [16] [3910/7269] eta: 0:24:15 lr: 0.000007 loss: 0.5178 (0.5298) time: 0.4790 data: 0.0001 max mem: 52150 +Epoch: [16] [3920/7269] eta: 0:24:11 lr: 0.000007 loss: 0.5080 (0.5297) time: 0.4172 data: 0.0002 max mem: 52150 +Epoch: [16] [3930/7269] eta: 0:24:06 lr: 0.000007 loss: 0.5222 (0.5298) time: 0.4172 data: 0.0002 max mem: 52150 +Epoch: [16] [3940/7269] eta: 0:24:02 lr: 0.000007 loss: 0.5297 (0.5297) time: 0.4329 data: 0.0001 max mem: 52150 +Epoch: [16] [3950/7269] eta: 0:23:58 lr: 0.000007 loss: 0.5463 (0.5297) time: 0.4329 data: 0.0002 max mem: 52150 +Epoch: [16] [3960/7269] eta: 0:23:54 lr: 0.000007 loss: 0.5463 (0.5297) time: 0.4317 data: 0.0002 max mem: 52150 +Epoch: [16] [3970/7269] eta: 0:23:49 lr: 0.000007 loss: 0.5247 (0.5296) time: 0.4321 data: 0.0002 max mem: 52150 +Epoch: [16] [3980/7269] eta: 0:23:45 lr: 0.000007 loss: 0.5272 (0.5296) time: 0.4320 data: 0.0002 max mem: 52150 +Epoch: [16] [3990/7269] eta: 0:23:40 lr: 0.000007 loss: 0.5330 (0.5296) time: 0.4316 data: 0.0001 max mem: 52150 +Epoch: [16] [4000/7269] eta: 0:23:35 lr: 0.000007 loss: 0.5445 (0.5296) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [16] [4010/7269] eta: 0:23:32 lr: 0.000007 loss: 0.5138 (0.5295) time: 0.4644 data: 0.0002 max mem: 52150 +Epoch: [16] [4020/7269] eta: 0:23:27 lr: 0.000007 loss: 0.5138 (0.5295) time: 0.4643 data: 0.0001 max mem: 52150 +Epoch: [16] [4030/7269] eta: 0:23:24 lr: 0.000007 loss: 0.5323 (0.5296) time: 0.4776 data: 0.0001 max mem: 52150 +Epoch: [16] [4040/7269] eta: 0:23:19 lr: 0.000007 loss: 0.5347 (0.5296) time: 0.4777 data: 0.0001 max mem: 52150 +Epoch: [16] [4050/7269] eta: 0:23:14 lr: 0.000007 loss: 0.5340 (0.5296) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [16] [4060/7269] eta: 0:23:10 lr: 0.000007 loss: 0.5333 (0.5297) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [16] [4070/7269] eta: 0:23:05 lr: 0.000007 loss: 0.5397 (0.5296) time: 0.3714 data: 0.0001 max mem: 52150 +Epoch: [16] [4080/7269] eta: 0:23:01 lr: 0.000007 loss: 0.5538 (0.5298) time: 0.4316 data: 0.0002 max mem: 52150 +Epoch: [16] [4090/7269] eta: 0:22:57 lr: 0.000007 loss: 0.5955 (0.5298) time: 0.4836 data: 0.0002 max mem: 52150 +Epoch: [16] [4100/7269] eta: 0:22:53 lr: 0.000007 loss: 0.5421 (0.5299) time: 0.4872 data: 0.0001 max mem: 52150 +Epoch: [16] [4110/7269] eta: 0:22:48 lr: 0.000007 loss: 0.5475 (0.5298) time: 0.4353 data: 0.0002 max mem: 52150 +Epoch: [16] [4120/7269] eta: 0:22:43 lr: 0.000007 loss: 0.5472 (0.5298) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [16] [4130/7269] eta: 0:22:39 lr: 0.000007 loss: 0.5543 (0.5299) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [16] [4140/7269] eta: 0:22:34 lr: 0.000007 loss: 0.5355 (0.5298) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [16] [4150/7269] eta: 0:22:31 lr: 0.000007 loss: 0.5651 (0.5299) time: 0.4775 data: 0.0001 max mem: 52150 +Epoch: [16] [4160/7269] eta: 0:22:27 lr: 0.000007 loss: 0.5729 (0.5299) time: 0.5224 data: 0.0002 max mem: 52150 +Epoch: [16] [4170/7269] eta: 0:22:22 lr: 0.000007 loss: 0.5020 (0.5299) time: 0.4167 data: 0.0002 max mem: 52150 +Epoch: [16] [4180/7269] eta: 0:22:17 lr: 0.000007 loss: 0.5251 (0.5299) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [16] [4190/7269] eta: 0:22:13 lr: 0.000007 loss: 0.5586 (0.5299) time: 0.4169 data: 0.0002 max mem: 52150 +Epoch: [16] [4200/7269] eta: 0:22:08 lr: 0.000007 loss: 0.5661 (0.5300) time: 0.4178 data: 0.0002 max mem: 52150 +Epoch: [16] [4210/7269] eta: 0:22:03 lr: 0.000007 loss: 0.5556 (0.5300) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [16] [4220/7269] eta: 0:21:59 lr: 0.000007 loss: 0.5591 (0.5301) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [16] [4230/7269] eta: 0:21:55 lr: 0.000007 loss: 0.5736 (0.5302) time: 0.4326 data: 0.0002 max mem: 52150 +Epoch: [16] [4240/7269] eta: 0:21:50 lr: 0.000007 loss: 0.5736 (0.5303) time: 0.4329 data: 0.0002 max mem: 52150 +Epoch: [16] [4250/7269] eta: 0:21:45 lr: 0.000007 loss: 0.5551 (0.5303) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [16] [4260/7269] eta: 0:21:41 lr: 0.000007 loss: 0.5291 (0.5303) time: 0.4315 data: 0.0002 max mem: 52150 +Epoch: [16] [4270/7269] eta: 0:21:36 lr: 0.000007 loss: 0.5291 (0.5303) time: 0.4316 data: 0.0002 max mem: 52150 +Epoch: [16] [4280/7269] eta: 0:21:33 lr: 0.000007 loss: 0.5462 (0.5303) time: 0.4646 data: 0.0001 max mem: 52150 +Epoch: [16] [4290/7269] eta: 0:21:28 lr: 0.000007 loss: 0.5384 (0.5302) time: 0.4647 data: 0.0001 max mem: 52150 +Epoch: [16] [4300/7269] eta: 0:21:24 lr: 0.000007 loss: 0.5106 (0.5302) time: 0.4227 data: 0.0001 max mem: 52150 +Epoch: [16] [4310/7269] eta: 0:21:20 lr: 0.000007 loss: 0.5331 (0.5302) time: 0.4831 data: 0.0002 max mem: 52150 +Epoch: [16] [4320/7269] eta: 0:21:16 lr: 0.000007 loss: 0.5011 (0.5302) time: 0.4319 data: 0.0002 max mem: 52150 +Epoch: [16] [4330/7269] eta: 0:21:11 lr: 0.000007 loss: 0.5306 (0.5302) time: 0.4206 data: 0.0002 max mem: 52150 +Epoch: [16] [4340/7269] eta: 0:21:08 lr: 0.000007 loss: 0.5123 (0.5302) time: 0.4814 data: 0.0002 max mem: 52150 +Epoch: [16] [4350/7269] eta: 0:21:03 lr: 0.000007 loss: 0.5404 (0.5302) time: 0.4322 data: 0.0002 max mem: 52150 +Epoch: [16] [4360/7269] eta: 0:20:59 lr: 0.000007 loss: 0.5382 (0.5301) time: 0.4314 data: 0.0002 max mem: 52150 +Epoch: [16] [4370/7269] eta: 0:20:54 lr: 0.000007 loss: 0.5382 (0.5302) time: 0.4313 data: 0.0002 max mem: 52150 +Epoch: [16] [4380/7269] eta: 0:20:51 lr: 0.000007 loss: 0.5433 (0.5302) time: 0.4845 data: 0.0002 max mem: 52150 +Epoch: [16] [4390/7269] eta: 0:20:47 lr: 0.000007 loss: 0.5416 (0.5302) time: 0.5862 data: 0.0002 max mem: 52150 +Epoch: [16] [4400/7269] eta: 0:20:44 lr: 0.000007 loss: 0.5484 (0.5302) time: 0.5342 data: 0.0002 max mem: 52150 +Epoch: [16] [4410/7269] eta: 0:20:39 lr: 0.000007 loss: 0.5465 (0.5302) time: 0.4328 data: 0.0002 max mem: 52150 +Epoch: [16] [4420/7269] eta: 0:20:34 lr: 0.000007 loss: 0.5316 (0.5302) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [16] [4430/7269] eta: 0:20:30 lr: 0.000007 loss: 0.5353 (0.5302) time: 0.4179 data: 0.0002 max mem: 52150 +Epoch: [16] [4440/7269] eta: 0:20:25 lr: 0.000007 loss: 0.5340 (0.5302) time: 0.4175 data: 0.0002 max mem: 52150 +Epoch: [16] [4450/7269] eta: 0:20:20 lr: 0.000007 loss: 0.5263 (0.5302) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [16] [4460/7269] eta: 0:20:16 lr: 0.000007 loss: 0.4770 (0.5301) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [16] [4470/7269] eta: 0:20:12 lr: 0.000007 loss: 0.4770 (0.5301) time: 0.4312 data: 0.0002 max mem: 52150 +Epoch: [16] [4480/7269] eta: 0:20:07 lr: 0.000007 loss: 0.5194 (0.5300) time: 0.4312 data: 0.0002 max mem: 52150 +Epoch: [16] [4490/7269] eta: 0:20:02 lr: 0.000007 loss: 0.5134 (0.5300) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [16] [4500/7269] eta: 0:19:58 lr: 0.000007 loss: 0.5299 (0.5300) time: 0.3714 data: 0.0001 max mem: 52150 +Epoch: [16] [4510/7269] eta: 0:19:54 lr: 0.000007 loss: 0.5452 (0.5300) time: 0.4217 data: 0.0002 max mem: 52150 +Epoch: [16] [4520/7269] eta: 0:19:49 lr: 0.000007 loss: 0.5410 (0.5301) time: 0.4216 data: 0.0002 max mem: 52150 +Epoch: [16] [4530/7269] eta: 0:19:44 lr: 0.000007 loss: 0.5375 (0.5300) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [16] [4540/7269] eta: 0:19:39 lr: 0.000007 loss: 0.5138 (0.5301) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [16] [4550/7269] eta: 0:19:35 lr: 0.000007 loss: 0.5449 (0.5301) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [16] [4560/7269] eta: 0:19:30 lr: 0.000007 loss: 0.5316 (0.5301) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [16] [4570/7269] eta: 0:19:25 lr: 0.000007 loss: 0.5316 (0.5301) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [16] [4580/7269] eta: 0:19:21 lr: 0.000007 loss: 0.5317 (0.5301) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [16] [4590/7269] eta: 0:19:17 lr: 0.000007 loss: 0.5078 (0.5300) time: 0.4308 data: 0.0002 max mem: 52150 +Epoch: [16] [4600/7269] eta: 0:19:12 lr: 0.000007 loss: 0.5031 (0.5300) time: 0.4309 data: 0.0001 max mem: 52150 +Epoch: [16] [4610/7269] eta: 0:19:08 lr: 0.000007 loss: 0.5578 (0.5300) time: 0.4318 data: 0.0001 max mem: 52150 +Epoch: [16] [4620/7269] eta: 0:19:04 lr: 0.000007 loss: 0.5075 (0.5299) time: 0.4766 data: 0.0001 max mem: 52150 +Epoch: [16] [4630/7269] eta: 0:19:00 lr: 0.000007 loss: 0.5582 (0.5300) time: 0.4622 data: 0.0001 max mem: 52150 +Epoch: [16] [4640/7269] eta: 0:18:55 lr: 0.000007 loss: 0.5527 (0.5300) time: 0.4169 data: 0.0002 max mem: 52150 +Epoch: [16] [4650/7269] eta: 0:18:52 lr: 0.000007 loss: 0.5262 (0.5300) time: 0.4767 data: 0.0001 max mem: 52150 +Epoch: [16] [4660/7269] eta: 0:18:47 lr: 0.000007 loss: 0.5437 (0.5300) time: 0.4776 data: 0.0002 max mem: 52150 +Epoch: [16] [4670/7269] eta: 0:18:42 lr: 0.000007 loss: 0.5437 (0.5299) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [16] [4680/7269] eta: 0:18:38 lr: 0.000007 loss: 0.5432 (0.5299) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [16] [4690/7269] eta: 0:18:34 lr: 0.000007 loss: 0.5306 (0.5299) time: 0.4323 data: 0.0002 max mem: 52150 +Epoch: [16] [4700/7269] eta: 0:18:30 lr: 0.000007 loss: 0.5104 (0.5299) time: 0.4781 data: 0.0001 max mem: 52150 +Epoch: [16] [4710/7269] eta: 0:18:25 lr: 0.000007 loss: 0.5306 (0.5299) time: 0.4180 data: 0.0001 max mem: 52150 +Epoch: [16] [4720/7269] eta: 0:18:21 lr: 0.000007 loss: 0.5395 (0.5299) time: 0.4182 data: 0.0002 max mem: 52150 +Epoch: [16] [4730/7269] eta: 0:18:16 lr: 0.000007 loss: 0.5246 (0.5298) time: 0.4179 data: 0.0002 max mem: 52150 +Epoch: [16] [4740/7269] eta: 0:18:12 lr: 0.000007 loss: 0.5289 (0.5298) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [16] [4750/7269] eta: 0:18:07 lr: 0.000007 loss: 0.5490 (0.5299) time: 0.4186 data: 0.0002 max mem: 52150 +Epoch: [16] [4760/7269] eta: 0:18:03 lr: 0.000007 loss: 0.5444 (0.5298) time: 0.4780 data: 0.0002 max mem: 52150 +Epoch: [16] [4770/7269] eta: 0:18:00 lr: 0.000007 loss: 0.5421 (0.5299) time: 0.5286 data: 0.0002 max mem: 52150 +Epoch: [16] [4780/7269] eta: 0:17:56 lr: 0.000007 loss: 0.5421 (0.5299) time: 0.5181 data: 0.0002 max mem: 52150 +Epoch: [16] [4790/7269] eta: 0:17:51 lr: 0.000007 loss: 0.5398 (0.5299) time: 0.4209 data: 0.0002 max mem: 52150 +Epoch: [16] [4800/7269] eta: 0:17:46 lr: 0.000007 loss: 0.5530 (0.5300) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [16] [4810/7269] eta: 0:17:42 lr: 0.000007 loss: 0.5636 (0.5300) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [16] [4820/7269] eta: 0:17:37 lr: 0.000007 loss: 0.5040 (0.5299) time: 0.3733 data: 0.0002 max mem: 52150 +Epoch: [16] [4830/7269] eta: 0:17:34 lr: 0.000007 loss: 0.4913 (0.5298) time: 0.4850 data: 0.0002 max mem: 52150 +Epoch: [16] [4840/7269] eta: 0:17:30 lr: 0.000007 loss: 0.5062 (0.5298) time: 0.5317 data: 0.0001 max mem: 52150 +Epoch: [16] [4850/7269] eta: 0:17:25 lr: 0.000007 loss: 0.5166 (0.5298) time: 0.4679 data: 0.0001 max mem: 52150 +Epoch: [16] [4860/7269] eta: 0:17:21 lr: 0.000007 loss: 0.4831 (0.5297) time: 0.4664 data: 0.0001 max mem: 52150 +Epoch: [16] [4870/7269] eta: 0:17:17 lr: 0.000007 loss: 0.4821 (0.5296) time: 0.4177 data: 0.0001 max mem: 52150 +Epoch: [16] [4880/7269] eta: 0:17:12 lr: 0.000007 loss: 0.5250 (0.5296) time: 0.4207 data: 0.0001 max mem: 52150 +Epoch: [16] [4890/7269] eta: 0:17:08 lr: 0.000007 loss: 0.5396 (0.5296) time: 0.4206 data: 0.0001 max mem: 52150 +Epoch: [16] [4900/7269] eta: 0:17:03 lr: 0.000007 loss: 0.5470 (0.5296) time: 0.3713 data: 0.0001 max mem: 52150 +Epoch: [16] [4910/7269] eta: 0:16:59 lr: 0.000007 loss: 0.5201 (0.5296) time: 0.3714 data: 0.0001 max mem: 52150 +Epoch: [16] [4920/7269] eta: 0:16:54 lr: 0.000007 loss: 0.4951 (0.5295) time: 0.4200 data: 0.0001 max mem: 52150 +Epoch: [16] [4930/7269] eta: 0:16:50 lr: 0.000007 loss: 0.4951 (0.5295) time: 0.4198 data: 0.0002 max mem: 52150 +Epoch: [16] [4940/7269] eta: 0:16:45 lr: 0.000007 loss: 0.5046 (0.5294) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [16] [4950/7269] eta: 0:16:41 lr: 0.000007 loss: 0.5111 (0.5294) time: 0.4319 data: 0.0002 max mem: 52150 +Epoch: [16] [4960/7269] eta: 0:16:37 lr: 0.000007 loss: 0.5293 (0.5294) time: 0.4315 data: 0.0002 max mem: 52150 +Epoch: [16] [4970/7269] eta: 0:16:32 lr: 0.000007 loss: 0.5362 (0.5294) time: 0.4191 data: 0.0001 max mem: 52150 +Epoch: [16] [4980/7269] eta: 0:16:30 lr: 0.000007 loss: 0.5421 (0.5294) time: 0.6062 data: 0.0002 max mem: 52150 +Epoch: [16] [4990/7269] eta: 0:16:27 lr: 0.000007 loss: 0.5421 (0.5294) time: 0.7399 data: 0.0002 max mem: 52150 +Epoch: [16] [5000/7269] eta: 0:16:22 lr: 0.000007 loss: 0.5542 (0.5294) time: 0.5530 data: 0.0001 max mem: 52150 +Epoch: [16] [5010/7269] eta: 0:16:17 lr: 0.000007 loss: 0.5197 (0.5294) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [16] [5020/7269] eta: 0:16:13 lr: 0.000007 loss: 0.5197 (0.5294) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [16] [5030/7269] eta: 0:16:08 lr: 0.000007 loss: 0.5507 (0.5295) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [16] [5040/7269] eta: 0:16:04 lr: 0.000007 loss: 0.5369 (0.5295) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [16] [5050/7269] eta: 0:15:59 lr: 0.000007 loss: 0.5174 (0.5295) time: 0.4209 data: 0.0002 max mem: 52150 +Epoch: [16] [5060/7269] eta: 0:15:55 lr: 0.000007 loss: 0.5518 (0.5295) time: 0.4203 data: 0.0002 max mem: 52150 +Epoch: [16] [5070/7269] eta: 0:15:50 lr: 0.000007 loss: 0.5518 (0.5294) time: 0.3712 data: 0.0001 max mem: 52150 +Epoch: [16] [5080/7269] eta: 0:15:46 lr: 0.000007 loss: 0.5600 (0.5295) time: 0.3713 data: 0.0001 max mem: 52150 +Epoch: [16] [5090/7269] eta: 0:15:41 lr: 0.000007 loss: 0.5269 (0.5294) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [16] [5100/7269] eta: 0:15:37 lr: 0.000007 loss: 0.4993 (0.5293) time: 0.4322 data: 0.0001 max mem: 52150 +Epoch: [16] [5110/7269] eta: 0:15:33 lr: 0.000007 loss: 0.5153 (0.5293) time: 0.5371 data: 0.0001 max mem: 52150 +Epoch: [16] [5120/7269] eta: 0:15:29 lr: 0.000007 loss: 0.5423 (0.5293) time: 0.4766 data: 0.0001 max mem: 52150 +Epoch: [16] [5130/7269] eta: 0:15:24 lr: 0.000007 loss: 0.5306 (0.5293) time: 0.3711 data: 0.0002 max mem: 52150 +Epoch: [16] [5140/7269] eta: 0:15:20 lr: 0.000007 loss: 0.5357 (0.5294) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [16] [5150/7269] eta: 0:15:15 lr: 0.000007 loss: 0.5532 (0.5293) time: 0.3726 data: 0.0001 max mem: 52150 +Epoch: [16] [5160/7269] eta: 0:15:10 lr: 0.000007 loss: 0.5246 (0.5293) time: 0.3724 data: 0.0001 max mem: 52150 +Epoch: [16] [5170/7269] eta: 0:15:06 lr: 0.000007 loss: 0.5015 (0.5292) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [16] [5180/7269] eta: 0:15:02 lr: 0.000007 loss: 0.5088 (0.5292) time: 0.4174 data: 0.0001 max mem: 52150 +Epoch: [16] [5190/7269] eta: 0:14:58 lr: 0.000007 loss: 0.5378 (0.5292) time: 0.5096 data: 0.0001 max mem: 52150 +Epoch: [16] [5200/7269] eta: 0:14:53 lr: 0.000007 loss: 0.5657 (0.5293) time: 0.4644 data: 0.0001 max mem: 52150 +Epoch: [16] [5210/7269] eta: 0:14:49 lr: 0.000007 loss: 0.5460 (0.5292) time: 0.3722 data: 0.0001 max mem: 52150 +Epoch: [16] [5220/7269] eta: 0:14:44 lr: 0.000007 loss: 0.5231 (0.5292) time: 0.3722 data: 0.0001 max mem: 52150 +Epoch: [16] [5230/7269] eta: 0:14:40 lr: 0.000007 loss: 0.5254 (0.5292) time: 0.4172 data: 0.0002 max mem: 52150 +Epoch: [16] [5240/7269] eta: 0:14:35 lr: 0.000007 loss: 0.5431 (0.5293) time: 0.4172 data: 0.0002 max mem: 52150 +Epoch: [16] [5250/7269] eta: 0:14:31 lr: 0.000007 loss: 0.5583 (0.5293) time: 0.4317 data: 0.0001 max mem: 52150 +Epoch: [16] [5260/7269] eta: 0:14:27 lr: 0.000007 loss: 0.5640 (0.5294) time: 0.4319 data: 0.0001 max mem: 52150 +Epoch: [16] [5270/7269] eta: 0:14:22 lr: 0.000007 loss: 0.5121 (0.5293) time: 0.3724 data: 0.0001 max mem: 52150 +Epoch: [16] [5280/7269] eta: 0:14:19 lr: 0.000007 loss: 0.4940 (0.5293) time: 0.5137 data: 0.0002 max mem: 52150 +Epoch: [16] [5290/7269] eta: 0:14:15 lr: 0.000007 loss: 0.5470 (0.5293) time: 0.5598 data: 0.0001 max mem: 52150 +Epoch: [16] [5300/7269] eta: 0:14:10 lr: 0.000007 loss: 0.5241 (0.5293) time: 0.4177 data: 0.0001 max mem: 52150 +Epoch: [16] [5310/7269] eta: 0:14:06 lr: 0.000007 loss: 0.5241 (0.5293) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [16] [5320/7269] eta: 0:14:01 lr: 0.000007 loss: 0.5694 (0.5293) time: 0.4224 data: 0.0001 max mem: 52150 +Epoch: [16] [5330/7269] eta: 0:13:57 lr: 0.000007 loss: 0.5372 (0.5293) time: 0.4827 data: 0.0001 max mem: 52150 +Epoch: [16] [5340/7269] eta: 0:13:53 lr: 0.000007 loss: 0.5260 (0.5293) time: 0.4328 data: 0.0002 max mem: 52150 +Epoch: [16] [5350/7269] eta: 0:13:49 lr: 0.000007 loss: 0.5651 (0.5294) time: 0.4223 data: 0.0002 max mem: 52150 +Epoch: [16] [5360/7269] eta: 0:13:44 lr: 0.000007 loss: 0.5528 (0.5294) time: 0.4225 data: 0.0002 max mem: 52150 +Epoch: [16] [5370/7269] eta: 0:13:40 lr: 0.000007 loss: 0.5292 (0.5293) time: 0.4335 data: 0.0002 max mem: 52150 +Epoch: [16] [5380/7269] eta: 0:13:36 lr: 0.000007 loss: 0.5269 (0.5293) time: 0.4956 data: 0.0002 max mem: 52150 +Epoch: [16] [5390/7269] eta: 0:13:31 lr: 0.000007 loss: 0.5269 (0.5293) time: 0.4346 data: 0.0002 max mem: 52150 +Epoch: [16] [5400/7269] eta: 0:13:27 lr: 0.000007 loss: 0.5308 (0.5293) time: 0.3727 data: 0.0002 max mem: 52150 +Epoch: [16] [5410/7269] eta: 0:13:22 lr: 0.000007 loss: 0.5572 (0.5294) time: 0.3729 data: 0.0002 max mem: 52150 +Epoch: [16] [5420/7269] eta: 0:13:18 lr: 0.000007 loss: 0.5755 (0.5295) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [16] [5430/7269] eta: 0:13:13 lr: 0.000007 loss: 0.5534 (0.5295) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [16] [5440/7269] eta: 0:13:09 lr: 0.000007 loss: 0.5525 (0.5295) time: 0.4628 data: 0.0002 max mem: 52150 +Epoch: [16] [5450/7269] eta: 0:13:05 lr: 0.000007 loss: 0.5525 (0.5295) time: 0.4633 data: 0.0001 max mem: 52150 +Epoch: [16] [5460/7269] eta: 0:13:01 lr: 0.000007 loss: 0.5690 (0.5296) time: 0.4807 data: 0.0001 max mem: 52150 +Epoch: [16] [5470/7269] eta: 0:12:56 lr: 0.000007 loss: 0.5236 (0.5295) time: 0.4809 data: 0.0001 max mem: 52150 +Epoch: [16] [5480/7269] eta: 0:12:53 lr: 0.000007 loss: 0.4939 (0.5295) time: 0.4958 data: 0.0001 max mem: 52150 +Epoch: [16] [5490/7269] eta: 0:12:49 lr: 0.000007 loss: 0.4939 (0.5295) time: 0.5433 data: 0.0001 max mem: 52150 +Epoch: [16] [5500/7269] eta: 0:12:44 lr: 0.000007 loss: 0.5464 (0.5295) time: 0.4199 data: 0.0001 max mem: 52150 +Epoch: [16] [5510/7269] eta: 0:12:40 lr: 0.000007 loss: 0.5702 (0.5295) time: 0.4230 data: 0.0002 max mem: 52150 +Epoch: [16] [5520/7269] eta: 0:12:35 lr: 0.000007 loss: 0.4858 (0.5294) time: 0.4230 data: 0.0002 max mem: 52150 +Epoch: [16] [5530/7269] eta: 0:12:31 lr: 0.000007 loss: 0.5162 (0.5294) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [16] [5540/7269] eta: 0:12:26 lr: 0.000007 loss: 0.5353 (0.5294) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [16] [5550/7269] eta: 0:12:22 lr: 0.000007 loss: 0.5308 (0.5295) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [16] [5560/7269] eta: 0:12:18 lr: 0.000007 loss: 0.5276 (0.5295) time: 0.4952 data: 0.0002 max mem: 52150 +Epoch: [16] [5570/7269] eta: 0:12:14 lr: 0.000007 loss: 0.5132 (0.5294) time: 0.4953 data: 0.0001 max mem: 52150 +Epoch: [16] [5580/7269] eta: 0:12:09 lr: 0.000007 loss: 0.5253 (0.5295) time: 0.3726 data: 0.0001 max mem: 52150 +Epoch: [16] [5590/7269] eta: 0:12:05 lr: 0.000007 loss: 0.5253 (0.5295) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [16] [5600/7269] eta: 0:12:00 lr: 0.000007 loss: 0.5090 (0.5295) time: 0.3714 data: 0.0001 max mem: 52150 +Epoch: [16] [5610/7269] eta: 0:11:56 lr: 0.000007 loss: 0.5168 (0.5295) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [16] [5620/7269] eta: 0:11:51 lr: 0.000007 loss: 0.5362 (0.5295) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [16] [5630/7269] eta: 0:11:47 lr: 0.000007 loss: 0.5464 (0.5295) time: 0.4211 data: 0.0002 max mem: 52150 +Epoch: [16] [5640/7269] eta: 0:11:43 lr: 0.000007 loss: 0.5603 (0.5295) time: 0.4828 data: 0.0001 max mem: 52150 +Epoch: [16] [5650/7269] eta: 0:11:38 lr: 0.000007 loss: 0.5603 (0.5296) time: 0.4340 data: 0.0002 max mem: 52150 +Epoch: [16] [5660/7269] eta: 0:11:34 lr: 0.000007 loss: 0.5451 (0.5296) time: 0.3730 data: 0.0002 max mem: 52150 +Epoch: [16] [5670/7269] eta: 0:11:29 lr: 0.000007 loss: 0.5291 (0.5296) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [16] [5680/7269] eta: 0:11:25 lr: 0.000007 loss: 0.5477 (0.5295) time: 0.4172 data: 0.0001 max mem: 52150 +Epoch: [16] [5690/7269] eta: 0:11:21 lr: 0.000007 loss: 0.5284 (0.5295) time: 0.4623 data: 0.0001 max mem: 52150 +Epoch: [16] [5700/7269] eta: 0:11:16 lr: 0.000007 loss: 0.5342 (0.5296) time: 0.4165 data: 0.0002 max mem: 52150 +Epoch: [16] [5710/7269] eta: 0:11:12 lr: 0.000007 loss: 0.5420 (0.5296) time: 0.4165 data: 0.0001 max mem: 52150 +Epoch: [16] [5720/7269] eta: 0:11:08 lr: 0.000007 loss: 0.4902 (0.5295) time: 0.4624 data: 0.0001 max mem: 52150 +Epoch: [16] [5730/7269] eta: 0:11:03 lr: 0.000007 loss: 0.4893 (0.5294) time: 0.4176 data: 0.0002 max mem: 52150 +Epoch: [16] [5740/7269] eta: 0:10:59 lr: 0.000007 loss: 0.5421 (0.5295) time: 0.4329 data: 0.0001 max mem: 52150 +Epoch: [16] [5750/7269] eta: 0:10:55 lr: 0.000007 loss: 0.5578 (0.5295) time: 0.4330 data: 0.0001 max mem: 52150 +Epoch: [16] [5760/7269] eta: 0:10:51 lr: 0.000007 loss: 0.5683 (0.5295) time: 0.4199 data: 0.0001 max mem: 52150 +Epoch: [16] [5770/7269] eta: 0:10:46 lr: 0.000007 loss: 0.5669 (0.5296) time: 0.4673 data: 0.0001 max mem: 52150 +Epoch: [16] [5780/7269] eta: 0:10:42 lr: 0.000007 loss: 0.5753 (0.5296) time: 0.4198 data: 0.0001 max mem: 52150 +Epoch: [16] [5790/7269] eta: 0:10:38 lr: 0.000007 loss: 0.5275 (0.5296) time: 0.4780 data: 0.0001 max mem: 52150 +Epoch: [16] [5800/7269] eta: 0:10:34 lr: 0.000007 loss: 0.4946 (0.5295) time: 0.4777 data: 0.0001 max mem: 52150 +Epoch: [16] [5810/7269] eta: 0:10:29 lr: 0.000007 loss: 0.5016 (0.5295) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [16] [5820/7269] eta: 0:10:25 lr: 0.000007 loss: 0.5371 (0.5296) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [16] [5830/7269] eta: 0:10:21 lr: 0.000007 loss: 0.5649 (0.5296) time: 0.4901 data: 0.0002 max mem: 52150 +Epoch: [16] [5840/7269] eta: 0:10:17 lr: 0.000007 loss: 0.5473 (0.5296) time: 0.5531 data: 0.0001 max mem: 52150 +Epoch: [16] [5850/7269] eta: 0:10:13 lr: 0.000007 loss: 0.5194 (0.5296) time: 0.5437 data: 0.0001 max mem: 52150 +Epoch: [16] [5860/7269] eta: 0:10:08 lr: 0.000007 loss: 0.5106 (0.5296) time: 0.5267 data: 0.0002 max mem: 52150 +Epoch: [16] [5870/7269] eta: 0:10:04 lr: 0.000007 loss: 0.5515 (0.5297) time: 0.4179 data: 0.0002 max mem: 52150 +Epoch: [16] [5880/7269] eta: 0:09:59 lr: 0.000007 loss: 0.5543 (0.5297) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [16] [5890/7269] eta: 0:09:55 lr: 0.000007 loss: 0.5470 (0.5296) time: 0.3722 data: 0.0001 max mem: 52150 +Epoch: [16] [5900/7269] eta: 0:09:51 lr: 0.000007 loss: 0.5271 (0.5296) time: 0.4190 data: 0.0001 max mem: 52150 +Epoch: [16] [5910/7269] eta: 0:09:46 lr: 0.000007 loss: 0.5230 (0.5296) time: 0.4645 data: 0.0001 max mem: 52150 +Epoch: [16] [5920/7269] eta: 0:09:42 lr: 0.000007 loss: 0.5109 (0.5296) time: 0.4178 data: 0.0001 max mem: 52150 +Epoch: [16] [5930/7269] eta: 0:09:38 lr: 0.000007 loss: 0.5280 (0.5296) time: 0.4333 data: 0.0001 max mem: 52150 +Epoch: [16] [5940/7269] eta: 0:09:33 lr: 0.000007 loss: 0.5298 (0.5296) time: 0.4332 data: 0.0002 max mem: 52150 +Epoch: [16] [5950/7269] eta: 0:09:29 lr: 0.000007 loss: 0.5602 (0.5297) time: 0.4813 data: 0.0002 max mem: 52150 +Epoch: [16] [5960/7269] eta: 0:09:25 lr: 0.000007 loss: 0.5602 (0.5297) time: 0.4809 data: 0.0002 max mem: 52150 +Epoch: [16] [5970/7269] eta: 0:09:21 lr: 0.000007 loss: 0.5396 (0.5297) time: 0.4661 data: 0.0002 max mem: 52150 +Epoch: [16] [5980/7269] eta: 0:09:17 lr: 0.000007 loss: 0.5206 (0.5296) time: 0.5258 data: 0.0002 max mem: 52150 +Epoch: [16] [5990/7269] eta: 0:09:12 lr: 0.000007 loss: 0.5328 (0.5296) time: 0.4315 data: 0.0002 max mem: 52150 +Epoch: [16] [6000/7269] eta: 0:09:08 lr: 0.000007 loss: 0.5297 (0.5296) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [16] [6010/7269] eta: 0:09:03 lr: 0.000007 loss: 0.4983 (0.5295) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [16] [6020/7269] eta: 0:08:59 lr: 0.000007 loss: 0.5185 (0.5296) time: 0.4345 data: 0.0002 max mem: 52150 +Epoch: [16] [6030/7269] eta: 0:08:55 lr: 0.000007 loss: 0.5489 (0.5296) time: 0.4964 data: 0.0002 max mem: 52150 +Epoch: [16] [6040/7269] eta: 0:08:51 lr: 0.000007 loss: 0.5128 (0.5295) time: 0.4949 data: 0.0002 max mem: 52150 +Epoch: [16] [6050/7269] eta: 0:08:46 lr: 0.000007 loss: 0.4606 (0.5295) time: 0.4326 data: 0.0002 max mem: 52150 +Epoch: [16] [6060/7269] eta: 0:08:42 lr: 0.000007 loss: 0.5215 (0.5295) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [16] [6070/7269] eta: 0:08:38 lr: 0.000007 loss: 0.5296 (0.5295) time: 0.4209 data: 0.0002 max mem: 52150 +Epoch: [16] [6080/7269] eta: 0:08:33 lr: 0.000007 loss: 0.5553 (0.5295) time: 0.4216 data: 0.0002 max mem: 52150 +Epoch: [16] [6090/7269] eta: 0:08:29 lr: 0.000007 loss: 0.5513 (0.5295) time: 0.3740 data: 0.0002 max mem: 52150 +Epoch: [16] [6100/7269] eta: 0:08:24 lr: 0.000007 loss: 0.5592 (0.5295) time: 0.3735 data: 0.0002 max mem: 52150 +Epoch: [16] [6110/7269] eta: 0:08:20 lr: 0.000007 loss: 0.5749 (0.5296) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [16] [6120/7269] eta: 0:08:16 lr: 0.000007 loss: 0.5616 (0.5296) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [16] [6130/7269] eta: 0:08:11 lr: 0.000007 loss: 0.4979 (0.5296) time: 0.4206 data: 0.0002 max mem: 52150 +Epoch: [16] [6140/7269] eta: 0:08:07 lr: 0.000007 loss: 0.4703 (0.5295) time: 0.4209 data: 0.0002 max mem: 52150 +Epoch: [16] [6150/7269] eta: 0:08:02 lr: 0.000007 loss: 0.4790 (0.5295) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [16] [6160/7269] eta: 0:07:58 lr: 0.000007 loss: 0.5326 (0.5295) time: 0.4238 data: 0.0002 max mem: 52150 +Epoch: [16] [6170/7269] eta: 0:07:54 lr: 0.000007 loss: 0.5546 (0.5296) time: 0.4837 data: 0.0002 max mem: 52150 +Epoch: [16] [6180/7269] eta: 0:07:50 lr: 0.000007 loss: 0.5560 (0.5296) time: 0.4318 data: 0.0002 max mem: 52150 +Epoch: [16] [6190/7269] eta: 0:07:45 lr: 0.000007 loss: 0.5259 (0.5296) time: 0.3731 data: 0.0002 max mem: 52150 +Epoch: [16] [6200/7269] eta: 0:07:41 lr: 0.000007 loss: 0.5368 (0.5296) time: 0.4339 data: 0.0002 max mem: 52150 +Epoch: [16] [6210/7269] eta: 0:07:37 lr: 0.000007 loss: 0.5406 (0.5295) time: 0.4333 data: 0.0002 max mem: 52150 +Epoch: [16] [6220/7269] eta: 0:07:32 lr: 0.000007 loss: 0.5264 (0.5296) time: 0.3729 data: 0.0002 max mem: 52150 +Epoch: [16] [6230/7269] eta: 0:07:28 lr: 0.000007 loss: 0.5402 (0.5295) time: 0.4191 data: 0.0002 max mem: 52150 +Epoch: [16] [6240/7269] eta: 0:07:24 lr: 0.000007 loss: 0.5004 (0.5295) time: 0.4647 data: 0.0002 max mem: 52150 +Epoch: [16] [6250/7269] eta: 0:07:19 lr: 0.000007 loss: 0.4941 (0.5294) time: 0.4181 data: 0.0002 max mem: 52150 +Epoch: [16] [6260/7269] eta: 0:07:15 lr: 0.000007 loss: 0.5108 (0.5294) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [16] [6270/7269] eta: 0:07:10 lr: 0.000007 loss: 0.5182 (0.5294) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [16] [6280/7269] eta: 0:07:06 lr: 0.000007 loss: 0.5192 (0.5293) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [16] [6290/7269] eta: 0:07:02 lr: 0.000007 loss: 0.5356 (0.5293) time: 0.5247 data: 0.0002 max mem: 52150 +Epoch: [16] [6300/7269] eta: 0:06:58 lr: 0.000007 loss: 0.5249 (0.5293) time: 0.5694 data: 0.0002 max mem: 52150 +Epoch: [16] [6310/7269] eta: 0:06:53 lr: 0.000007 loss: 0.4977 (0.5293) time: 0.4177 data: 0.0002 max mem: 52150 +Epoch: [16] [6320/7269] eta: 0:06:49 lr: 0.000007 loss: 0.5233 (0.5293) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [16] [6330/7269] eta: 0:06:45 lr: 0.000007 loss: 0.5059 (0.5293) time: 0.4843 data: 0.0002 max mem: 52150 +Epoch: [16] [6340/7269] eta: 0:06:41 lr: 0.000007 loss: 0.4970 (0.5292) time: 0.6503 data: 0.0002 max mem: 52150 +Epoch: [16] [6350/7269] eta: 0:06:37 lr: 0.000007 loss: 0.4882 (0.5291) time: 0.5380 data: 0.0002 max mem: 52150 +Epoch: [16] [6360/7269] eta: 0:06:32 lr: 0.000007 loss: 0.5270 (0.5292) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [16] [6370/7269] eta: 0:06:28 lr: 0.000007 loss: 0.5270 (0.5291) time: 0.4191 data: 0.0002 max mem: 52150 +Epoch: [16] [6380/7269] eta: 0:06:24 lr: 0.000007 loss: 0.4546 (0.5290) time: 0.4811 data: 0.0002 max mem: 52150 +Epoch: [16] [6390/7269] eta: 0:06:19 lr: 0.000007 loss: 0.4479 (0.5290) time: 0.4956 data: 0.0002 max mem: 52150 +Epoch: [16] [6400/7269] eta: 0:06:15 lr: 0.000007 loss: 0.4931 (0.5289) time: 0.4339 data: 0.0002 max mem: 52150 +Epoch: [16] [6410/7269] eta: 0:06:11 lr: 0.000007 loss: 0.5304 (0.5290) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [16] [6420/7269] eta: 0:06:06 lr: 0.000007 loss: 0.5353 (0.5290) time: 0.4188 data: 0.0002 max mem: 52150 +Epoch: [16] [6430/7269] eta: 0:06:02 lr: 0.000007 loss: 0.5316 (0.5290) time: 0.4184 data: 0.0002 max mem: 52150 +Epoch: [16] [6440/7269] eta: 0:05:58 lr: 0.000007 loss: 0.5316 (0.5289) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [16] [6450/7269] eta: 0:05:53 lr: 0.000007 loss: 0.5240 (0.5289) time: 0.4177 data: 0.0002 max mem: 52150 +Epoch: [16] [6460/7269] eta: 0:05:49 lr: 0.000007 loss: 0.5270 (0.5289) time: 0.4782 data: 0.0002 max mem: 52150 +Epoch: [16] [6470/7269] eta: 0:05:45 lr: 0.000007 loss: 0.5384 (0.5289) time: 0.4321 data: 0.0002 max mem: 52150 +Epoch: [16] [6480/7269] eta: 0:05:40 lr: 0.000007 loss: 0.5315 (0.5289) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [16] [6490/7269] eta: 0:05:36 lr: 0.000007 loss: 0.5357 (0.5290) time: 0.4228 data: 0.0002 max mem: 52150 +Epoch: [16] [6500/7269] eta: 0:05:32 lr: 0.000007 loss: 0.5350 (0.5289) time: 0.4684 data: 0.0002 max mem: 52150 +Epoch: [16] [6510/7269] eta: 0:05:27 lr: 0.000007 loss: 0.5662 (0.5290) time: 0.4184 data: 0.0002 max mem: 52150 +Epoch: [16] [6520/7269] eta: 0:05:23 lr: 0.000007 loss: 0.5620 (0.5290) time: 0.3727 data: 0.0002 max mem: 52150 +Epoch: [16] [6530/7269] eta: 0:05:19 lr: 0.000007 loss: 0.5193 (0.5290) time: 0.4922 data: 0.0002 max mem: 52150 +Epoch: [16] [6540/7269] eta: 0:05:14 lr: 0.000007 loss: 0.5201 (0.5289) time: 0.4921 data: 0.0002 max mem: 52150 +Epoch: [16] [6550/7269] eta: 0:05:10 lr: 0.000007 loss: 0.5280 (0.5290) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [16] [6560/7269] eta: 0:05:06 lr: 0.000007 loss: 0.5280 (0.5289) time: 0.4188 data: 0.0002 max mem: 52150 +Epoch: [16] [6570/7269] eta: 0:05:01 lr: 0.000007 loss: 0.5237 (0.5289) time: 0.4189 data: 0.0002 max mem: 52150 +Epoch: [16] [6580/7269] eta: 0:04:57 lr: 0.000007 loss: 0.5308 (0.5289) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [16] [6590/7269] eta: 0:04:53 lr: 0.000007 loss: 0.5456 (0.5289) time: 0.4311 data: 0.0002 max mem: 52150 +Epoch: [16] [6600/7269] eta: 0:04:48 lr: 0.000007 loss: 0.5601 (0.5289) time: 0.4796 data: 0.0002 max mem: 52150 +Epoch: [16] [6610/7269] eta: 0:04:44 lr: 0.000007 loss: 0.5285 (0.5289) time: 0.4205 data: 0.0002 max mem: 52150 +Epoch: [16] [6620/7269] eta: 0:04:40 lr: 0.000007 loss: 0.5232 (0.5289) time: 0.4814 data: 0.0002 max mem: 52150 +Epoch: [16] [6630/7269] eta: 0:04:36 lr: 0.000007 loss: 0.5504 (0.5289) time: 0.4816 data: 0.0002 max mem: 52150 +Epoch: [16] [6640/7269] eta: 0:04:31 lr: 0.000007 loss: 0.5471 (0.5289) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [16] [6650/7269] eta: 0:04:27 lr: 0.000007 loss: 0.5337 (0.5289) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [16] [6660/7269] eta: 0:04:22 lr: 0.000007 loss: 0.5282 (0.5289) time: 0.4355 data: 0.0002 max mem: 52150 +Epoch: [16] [6670/7269] eta: 0:04:18 lr: 0.000007 loss: 0.5129 (0.5289) time: 0.4994 data: 0.0002 max mem: 52150 +Epoch: [16] [6680/7269] eta: 0:04:14 lr: 0.000007 loss: 0.4949 (0.5288) time: 0.4363 data: 0.0002 max mem: 52150 +Epoch: [16] [6690/7269] eta: 0:04:10 lr: 0.000007 loss: 0.4949 (0.5288) time: 0.4220 data: 0.0002 max mem: 52150 +Epoch: [16] [6700/7269] eta: 0:04:05 lr: 0.000007 loss: 0.5322 (0.5288) time: 0.4221 data: 0.0002 max mem: 52150 +Epoch: [16] [6710/7269] eta: 0:04:01 lr: 0.000007 loss: 0.5295 (0.5288) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [16] [6720/7269] eta: 0:03:57 lr: 0.000007 loss: 0.5372 (0.5288) time: 0.4828 data: 0.0002 max mem: 52150 +Epoch: [16] [6730/7269] eta: 0:03:52 lr: 0.000007 loss: 0.5427 (0.5289) time: 0.4824 data: 0.0002 max mem: 52150 +Epoch: [16] [6740/7269] eta: 0:03:48 lr: 0.000007 loss: 0.5649 (0.5289) time: 0.4171 data: 0.0002 max mem: 52150 +Epoch: [16] [6750/7269] eta: 0:03:44 lr: 0.000007 loss: 0.5461 (0.5289) time: 0.4663 data: 0.0002 max mem: 52150 +Epoch: [16] [6760/7269] eta: 0:03:39 lr: 0.000007 loss: 0.5259 (0.5289) time: 0.4214 data: 0.0002 max mem: 52150 +Epoch: [16] [6770/7269] eta: 0:03:35 lr: 0.000007 loss: 0.5259 (0.5289) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [16] [6780/7269] eta: 0:03:31 lr: 0.000007 loss: 0.5368 (0.5289) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [16] [6790/7269] eta: 0:03:26 lr: 0.000007 loss: 0.5387 (0.5288) time: 0.4167 data: 0.0002 max mem: 52150 +Epoch: [16] [6800/7269] eta: 0:03:22 lr: 0.000007 loss: 0.5539 (0.5289) time: 0.4172 data: 0.0002 max mem: 52150 +Epoch: [16] [6810/7269] eta: 0:03:18 lr: 0.000007 loss: 0.5519 (0.5289) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [16] [6820/7269] eta: 0:03:13 lr: 0.000007 loss: 0.5519 (0.5290) time: 0.4327 data: 0.0002 max mem: 52150 +Epoch: [16] [6830/7269] eta: 0:03:09 lr: 0.000007 loss: 0.5618 (0.5290) time: 0.4941 data: 0.0002 max mem: 52150 +Epoch: [16] [6840/7269] eta: 0:03:05 lr: 0.000007 loss: 0.5274 (0.5289) time: 0.6608 data: 0.0002 max mem: 52150 +Epoch: [16] [6850/7269] eta: 0:03:01 lr: 0.000007 loss: 0.4712 (0.5289) time: 0.5992 data: 0.0002 max mem: 52150 +Epoch: [16] [6860/7269] eta: 0:02:56 lr: 0.000007 loss: 0.5408 (0.5289) time: 0.4208 data: 0.0002 max mem: 52150 +Epoch: [16] [6870/7269] eta: 0:02:52 lr: 0.000007 loss: 0.5286 (0.5289) time: 0.4207 data: 0.0002 max mem: 52150 +Epoch: [16] [6880/7269] eta: 0:02:48 lr: 0.000007 loss: 0.5373 (0.5289) time: 0.4181 data: 0.0002 max mem: 52150 +Epoch: [16] [6890/7269] eta: 0:02:43 lr: 0.000007 loss: 0.5419 (0.5289) time: 0.4183 data: 0.0002 max mem: 52150 +Epoch: [16] [6900/7269] eta: 0:02:39 lr: 0.000007 loss: 0.5210 (0.5289) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [16] [6910/7269] eta: 0:02:35 lr: 0.000007 loss: 0.5185 (0.5289) time: 0.3727 data: 0.0002 max mem: 52150 +Epoch: [16] [6920/7269] eta: 0:02:30 lr: 0.000007 loss: 0.5494 (0.5289) time: 0.4314 data: 0.0002 max mem: 52150 +Epoch: [16] [6930/7269] eta: 0:02:26 lr: 0.000007 loss: 0.5486 (0.5289) time: 0.4770 data: 0.0002 max mem: 52150 +Epoch: [16] [6940/7269] eta: 0:02:22 lr: 0.000007 loss: 0.5098 (0.5288) time: 0.4777 data: 0.0002 max mem: 52150 +Epoch: [16] [6950/7269] eta: 0:02:17 lr: 0.000007 loss: 0.4585 (0.5288) time: 0.4317 data: 0.0002 max mem: 52150 +Epoch: [16] [6960/7269] eta: 0:02:13 lr: 0.000007 loss: 0.5096 (0.5287) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [16] [6970/7269] eta: 0:02:09 lr: 0.000007 loss: 0.5425 (0.5288) time: 0.4193 data: 0.0002 max mem: 52150 +Epoch: [16] [6980/7269] eta: 0:02:04 lr: 0.000007 loss: 0.5524 (0.5288) time: 0.4190 data: 0.0002 max mem: 52150 +Epoch: [16] [6990/7269] eta: 0:02:00 lr: 0.000007 loss: 0.5524 (0.5288) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [16] [7000/7269] eta: 0:01:56 lr: 0.000007 loss: 0.5102 (0.5287) time: 0.4358 data: 0.0002 max mem: 52150 +Epoch: [16] [7010/7269] eta: 0:01:51 lr: 0.000007 loss: 0.4991 (0.5287) time: 0.4357 data: 0.0002 max mem: 52150 +Epoch: [16] [7020/7269] eta: 0:01:47 lr: 0.000007 loss: 0.5228 (0.5287) time: 0.4188 data: 0.0002 max mem: 52150 +Epoch: [16] [7030/7269] eta: 0:01:43 lr: 0.000007 loss: 0.5321 (0.5288) time: 0.4199 data: 0.0002 max mem: 52150 +Epoch: [16] [7040/7269] eta: 0:01:38 lr: 0.000007 loss: 0.5283 (0.5287) time: 0.3731 data: 0.0002 max mem: 52150 +Epoch: [16] [7050/7269] eta: 0:01:34 lr: 0.000007 loss: 0.5283 (0.5288) time: 0.4331 data: 0.0002 max mem: 52150 +Epoch: [16] [7060/7269] eta: 0:01:30 lr: 0.000007 loss: 0.5343 (0.5287) time: 0.4331 data: 0.0002 max mem: 52150 +Epoch: [16] [7070/7269] eta: 0:01:25 lr: 0.000007 loss: 0.5327 (0.5287) time: 0.4212 data: 0.0002 max mem: 52150 +Epoch: [16] [7080/7269] eta: 0:01:21 lr: 0.000007 loss: 0.5324 (0.5287) time: 0.4819 data: 0.0002 max mem: 52150 +Epoch: [16] [7090/7269] eta: 0:01:17 lr: 0.000007 loss: 0.5189 (0.5287) time: 0.4326 data: 0.0002 max mem: 52150 +Epoch: [16] [7100/7269] eta: 0:01:13 lr: 0.000007 loss: 0.5440 (0.5288) time: 0.4331 data: 0.0002 max mem: 52150 +Epoch: [16] [7110/7269] eta: 0:01:08 lr: 0.000007 loss: 0.5630 (0.5288) time: 0.4336 data: 0.0002 max mem: 52150 +Epoch: [16] [7120/7269] eta: 0:01:04 lr: 0.000007 loss: 0.5502 (0.5288) time: 0.4779 data: 0.0002 max mem: 52150 +Epoch: [16] [7130/7269] eta: 0:01:00 lr: 0.000007 loss: 0.5650 (0.5289) time: 0.5879 data: 0.0002 max mem: 52150 +Epoch: [16] [7140/7269] eta: 0:00:55 lr: 0.000007 loss: 0.5650 (0.5289) time: 0.4831 data: 0.0002 max mem: 52150 +Epoch: [16] [7150/7269] eta: 0:00:51 lr: 0.000007 loss: 0.5249 (0.5289) time: 0.3727 data: 0.0002 max mem: 52150 +Epoch: [16] [7160/7269] eta: 0:00:47 lr: 0.000007 loss: 0.5215 (0.5289) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [16] [7170/7269] eta: 0:00:42 lr: 0.000007 loss: 0.5384 (0.5289) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [16] [7180/7269] eta: 0:00:38 lr: 0.000007 loss: 0.5578 (0.5289) time: 0.4329 data: 0.0002 max mem: 52150 +Epoch: [16] [7190/7269] eta: 0:00:34 lr: 0.000007 loss: 0.5578 (0.5289) time: 0.4330 data: 0.0002 max mem: 52150 +Epoch: [16] [7200/7269] eta: 0:00:29 lr: 0.000007 loss: 0.5574 (0.5289) time: 0.4313 data: 0.0002 max mem: 52150 +Epoch: [16] [7210/7269] eta: 0:00:25 lr: 0.000007 loss: 0.5510 (0.5289) time: 0.4799 data: 0.0002 max mem: 52150 +Epoch: [16] [7220/7269] eta: 0:00:21 lr: 0.000007 loss: 0.5510 (0.5290) time: 0.4205 data: 0.0002 max mem: 52150 +Epoch: [16] [7230/7269] eta: 0:00:16 lr: 0.000007 loss: 0.5539 (0.5290) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [16] [7240/7269] eta: 0:00:12 lr: 0.000007 loss: 0.5210 (0.5290) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [16] [7250/7269] eta: 0:00:08 lr: 0.000007 loss: 0.5334 (0.5290) time: 0.3755 data: 0.0025 max mem: 52150 +Epoch: [16] [7260/7269] eta: 0:00:03 lr: 0.000007 loss: 0.5714 (0.5291) time: 0.3746 data: 0.0024 max mem: 52150 +Epoch: [16] [7268/7269] eta: 0:00:00 lr: 0.000007 loss: 0.5714 (0.5291) time: 0.3736 data: 0.0024 max mem: 52150 +Epoch: [16] Total time: 0:52:20 (0.4321 s / it) +Averaged stats: lr: 0.000007 loss: 0.5714 (0.5284) +Test: [ 0/1403] eta: 1:18:46 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1267 (0.1267) acc1: 98.4375 (98.4375) acc5: 100.0000 (100.0000) time: 3.3687 data: 3.2719 max mem: 52150 +Test: [ 10/1403] eta: 0:09:08 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1713 (0.2103) acc1: 98.4375 (93.4659) acc5: 100.0000 (100.0000) time: 0.3939 data: 0.2977 max mem: 52150 +Test: [ 20/1403] eta: 0:05:48 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1653 (0.1917) acc1: 98.4375 (94.0476) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 30/1403] eta: 0:04:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1146 (0.1884) acc1: 98.4375 (93.7500) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 40/1403] eta: 0:04:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1298 (0.1860) acc1: 98.4375 (94.3598) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 50/1403] eta: 0:03:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1398 (0.2153) acc1: 98.4375 (92.7390) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 60/1403] eta: 0:03:21 acc1_pos: 90.625 acc5_pos: 100.000 loss: 0.1553 (0.2111) acc1: 98.4375 (93.4170) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 70/1403] eta: 0:03:10 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.1519 (0.2120) acc1: 96.8750 (93.2879) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 80/1403] eta: 0:03:01 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1630 (0.2178) acc1: 95.3125 (92.9784) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 90/1403] eta: 0:02:53 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2042 (0.2294) acc1: 93.7500 (92.2905) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 100/1403] eta: 0:02:47 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1924 (0.2281) acc1: 95.3125 (92.5124) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 110/1403] eta: 0:02:42 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1554 (0.2221) acc1: 96.8750 (92.8491) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 120/1403] eta: 0:02:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0850 (0.2105) acc1: 98.4375 (93.3884) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 130/1403] eta: 0:02:34 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0820 (0.2020) acc1: 98.4375 (93.7858) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 140/1403] eta: 0:02:31 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0816 (0.1930) acc1: 100.0000 (94.2265) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 150/1403] eta: 0:02:28 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0744 (0.1856) acc1: 100.0000 (94.5675) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 160/1403] eta: 0:02:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0748 (0.1793) acc1: 100.0000 (94.8564) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 170/1403] eta: 0:02:22 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0741 (0.1736) acc1: 100.0000 (95.1298) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 180/1403] eta: 0:02:20 acc1_pos: 81.250 acc5_pos: 100.000 loss: 0.0741 (0.1805) acc1: 100.0000 (94.7169) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 190/1403] eta: 0:02:17 acc1_pos: 70.312 acc5_pos: 100.000 loss: 0.3947 (0.1976) acc1: 81.2500 (93.6846) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 200/1403] eta: 0:02:15 acc1_pos: 92.188 acc5_pos: 100.000 loss: 0.3947 (0.2145) acc1: 82.8125 (92.7627) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 210/1403] eta: 0:02:13 acc1_pos: 70.312 acc5_pos: 100.000 loss: 0.4075 (0.2313) acc1: 81.2500 (91.8172) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 220/1403] eta: 0:02:11 acc1_pos: 78.125 acc5_pos: 100.000 loss: 0.5264 (0.2445) acc1: 71.8750 (91.1270) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 230/1403] eta: 0:02:09 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.3051 (0.2385) acc1: 89.0625 (91.4232) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 240/1403] eta: 0:02:08 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0767 (0.2318) acc1: 100.0000 (91.7661) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 250/1403] eta: 0:02:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0765 (0.2259) acc1: 100.0000 (92.0692) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 260/1403] eta: 0:02:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0747 (0.2204) acc1: 100.0000 (92.3551) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 270/1403] eta: 0:02:02 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0748 (0.2153) acc1: 100.0000 (92.6084) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 280/1403] eta: 0:02:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0728 (0.2101) acc1: 100.0000 (92.8659) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 290/1403] eta: 0:01:59 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0687 (0.2053) acc1: 100.0000 (93.1003) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 300/1403] eta: 0:01:58 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0681 (0.2009) acc1: 100.0000 (93.3243) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 310/1403] eta: 0:01:56 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0709 (0.1970) acc1: 100.0000 (93.5239) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 320/1403] eta: 0:01:55 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0870 (0.1936) acc1: 100.0000 (93.7257) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 330/1403] eta: 0:01:54 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0894 (0.1905) acc1: 100.0000 (93.9152) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 340/1403] eta: 0:01:52 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0890 (0.1875) acc1: 100.0000 (94.0937) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 350/1403] eta: 0:01:51 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0886 (0.1847) acc1: 100.0000 (94.2619) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 360/1403] eta: 0:01:50 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0894 (0.1822) acc1: 100.0000 (94.4209) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 370/1403] eta: 0:01:48 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0890 (0.1796) acc1: 100.0000 (94.5713) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 380/1403] eta: 0:01:47 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0880 (0.1773) acc1: 100.0000 (94.7137) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 390/1403] eta: 0:01:46 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0894 (0.1766) acc1: 100.0000 (94.7490) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 400/1403] eta: 0:01:44 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1966 (0.1788) acc1: 92.1875 (94.6189) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 410/1403] eta: 0:01:43 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2425 (0.1803) acc1: 89.0625 (94.5217) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 420/1403] eta: 0:01:42 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2000 (0.1804) acc1: 92.1875 (94.4960) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 430/1403] eta: 0:01:41 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2033 (0.1819) acc1: 92.1875 (94.4062) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 440/1403] eta: 0:01:40 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2684 (0.1840) acc1: 89.0625 (94.2885) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 450/1403] eta: 0:01:38 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2526 (0.1851) acc1: 90.6250 (94.2385) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 460/1403] eta: 0:01:37 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2450 (0.1863) acc1: 90.6250 (94.1601) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 470/1403] eta: 0:01:36 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1880 (0.1849) acc1: 93.7500 (94.2178) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [ 480/1403] eta: 0:01:35 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0585 (0.1823) acc1: 100.0000 (94.3380) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 490/1403] eta: 0:01:34 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0586 (0.1798) acc1: 100.0000 (94.4501) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 500/1403] eta: 0:01:33 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0592 (0.1774) acc1: 100.0000 (94.5609) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 510/1403] eta: 0:01:31 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0593 (0.1751) acc1: 100.0000 (94.6673) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 520/1403] eta: 0:01:30 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0622 (0.1729) acc1: 100.0000 (94.7697) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 530/1403] eta: 0:01:29 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0565 (0.1707) acc1: 100.0000 (94.8682) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 540/1403] eta: 0:01:28 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0560 (0.1686) acc1: 100.0000 (94.9630) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 550/1403] eta: 0:01:27 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0566 (0.1671) acc1: 100.0000 (95.0318) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 560/1403] eta: 0:01:26 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0849 (0.1658) acc1: 98.4375 (95.0925) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 570/1403] eta: 0:01:25 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0849 (0.1644) acc1: 98.4375 (95.1675) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 580/1403] eta: 0:01:24 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0796 (0.1638) acc1: 100.0000 (95.1995) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 590/1403] eta: 0:01:22 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1153 (0.1635) acc1: 96.8750 (95.2041) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 600/1403] eta: 0:01:21 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1237 (0.1634) acc1: 95.3125 (95.1825) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0973 data: 0.0003 max mem: 52150 +Test: [ 610/1403] eta: 0:01:20 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1135 (0.1625) acc1: 95.3125 (95.2204) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0971 data: 0.0003 max mem: 52150 +Test: [ 620/1403] eta: 0:01:19 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1245 (0.1633) acc1: 95.3125 (95.1464) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 630/1403] eta: 0:01:18 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1654 (0.1632) acc1: 93.7500 (95.1416) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 640/1403] eta: 0:01:17 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1282 (0.1627) acc1: 95.3125 (95.1565) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 650/1403] eta: 0:01:16 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1147 (0.1618) acc1: 96.8750 (95.2093) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 660/1403] eta: 0:01:15 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0795 (0.1605) acc1: 100.0000 (95.2770) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 670/1403] eta: 0:01:14 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0690 (0.1591) acc1: 100.0000 (95.3381) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 680/1403] eta: 0:01:13 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0904 (0.1596) acc1: 98.4375 (95.3125) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [ 690/1403] eta: 0:01:12 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2152 (0.1617) acc1: 92.1875 (95.2085) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 700/1403] eta: 0:01:11 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1504 (0.1619) acc1: 96.8750 (95.2055) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 710/1403] eta: 0:01:10 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3294 (0.1662) acc1: 84.3750 (94.9609) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 720/1403] eta: 0:01:09 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3294 (0.1661) acc1: 84.3750 (94.9593) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 730/1403] eta: 0:01:07 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0641 (0.1649) acc1: 100.0000 (95.0175) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 740/1403] eta: 0:01:06 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0708 (0.1639) acc1: 100.0000 (95.0679) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 750/1403] eta: 0:01:05 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1100 (0.1653) acc1: 98.4375 (94.9879) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 760/1403] eta: 0:01:04 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2966 (0.1684) acc1: 85.9375 (94.7766) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 770/1403] eta: 0:01:03 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.4984 (0.1732) acc1: 75.0000 (94.4998) acc5: 100.0000 (100.0000) acc1_neg: 68.750 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 780/1403] eta: 0:01:02 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.4429 (0.1755) acc1: 81.2500 (94.3742) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 790/1403] eta: 0:01:01 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2973 (0.1770) acc1: 87.5000 (94.3031) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 800/1403] eta: 0:01:00 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2025 (0.1772) acc1: 92.1875 (94.2981) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 810/1403] eta: 0:00:59 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2167 (0.1804) acc1: 89.0625 (94.1450) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 820/1403] eta: 0:00:58 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.6476 (0.1895) acc1: 71.8750 (93.7576) acc5: 100.0000 (100.0000) acc1_neg: 40.625 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 830/1403] eta: 0:00:57 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.6344 (0.1915) acc1: 71.8750 (93.6597) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 840/1403] eta: 0:00:56 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.4401 (0.1990) acc1: 76.5625 (93.3357) acc5: 100.0000 (100.0000) acc1_neg: 50.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 850/1403] eta: 0:00:55 acc1_pos: 94.118 acc5_pos: 100.000 loss: 1.0268 (0.2097) acc1: 56.2500 (92.8907) acc5: 100.0000 (100.0000) acc1_neg: 53.125 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 860/1403] eta: 0:00:54 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.9373 (0.2129) acc1: 59.3750 (92.7610) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 870/1403] eta: 0:00:53 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1357 (0.2118) acc1: 95.3125 (92.8136) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 880/1403] eta: 0:00:52 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1012 (0.2106) acc1: 98.4375 (92.8685) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 890/1403] eta: 0:00:51 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0789 (0.2091) acc1: 100.0000 (92.9433) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 900/1403] eta: 0:00:50 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0738 (0.2076) acc1: 100.0000 (93.0182) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 910/1403] eta: 0:00:49 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0826 (0.2068) acc1: 98.4375 (93.0502) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [ 920/1403] eta: 0:00:48 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1023 (0.2058) acc1: 96.8750 (93.0934) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 930/1403] eta: 0:00:47 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1106 (0.2048) acc1: 96.8750 (93.1391) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 940/1403] eta: 0:00:46 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1213 (0.2049) acc1: 95.3125 (93.1290) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 950/1403] eta: 0:00:45 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2337 (0.2058) acc1: 87.5000 (93.0599) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 960/1403] eta: 0:00:44 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2566 (0.2065) acc1: 87.5000 (93.0183) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 970/1403] eta: 0:00:43 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2579 (0.2074) acc1: 89.0625 (92.9535) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 980/1403] eta: 0:00:42 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2646 (0.2083) acc1: 85.9375 (92.8915) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 990/1403] eta: 0:00:41 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2591 (0.2089) acc1: 89.0625 (92.8623) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1000/1403] eta: 0:00:40 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2573 (0.2096) acc1: 89.0625 (92.8197) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1010/1403] eta: 0:00:39 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2604 (0.2103) acc1: 89.0625 (92.7794) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1020/1403] eta: 0:00:38 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3438 (0.2151) acc1: 81.2500 (92.4875) acc5: 100.0000 (100.0000) acc1_neg: 34.375 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1030/1403] eta: 0:00:37 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.5214 (0.2163) acc1: 73.4375 (92.4012) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1040/1403] eta: 0:00:36 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2116 (0.2159) acc1: 90.6250 (92.4201) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1050/1403] eta: 0:00:35 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1601 (0.2156) acc1: 95.3125 (92.4417) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1060/1403] eta: 0:00:34 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1280 (0.2145) acc1: 96.8750 (92.4968) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1070/1403] eta: 0:00:33 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1220 (0.2151) acc1: 98.4375 (92.4720) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1080/1403] eta: 0:00:32 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.4173 (0.2175) acc1: 81.2500 (92.3537) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1090/1403] eta: 0:00:31 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.4601 (0.2198) acc1: 81.2500 (92.2362) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1100/1403] eta: 0:00:30 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3464 (0.2203) acc1: 82.8125 (92.2130) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1110/1403] eta: 0:00:29 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2719 (0.2207) acc1: 89.0625 (92.1945) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1120/1403] eta: 0:00:28 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2612 (0.2211) acc1: 89.0625 (92.1777) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1130/1403] eta: 0:00:27 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2597 (0.2218) acc1: 89.0625 (92.1433) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1140/1403] eta: 0:00:26 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2510 (0.2219) acc1: 90.6250 (92.1300) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1150/1403] eta: 0:00:25 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2350 (0.2222) acc1: 90.6250 (92.1074) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52150 +Test: [1160/1403] eta: 0:00:24 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2544 (0.2226) acc1: 90.6250 (92.0919) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1170/1403] eta: 0:00:23 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2558 (0.2224) acc1: 90.6250 (92.0994) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52150 +Test: [1180/1403] eta: 0:00:22 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0656 (0.2210) acc1: 100.0000 (92.1663) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52150 +Test: [1190/1403] eta: 0:00:21 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0649 (0.2197) acc1: 100.0000 (92.2321) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52150 +Test: [1200/1403] eta: 0:00:20 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0650 (0.2184) acc1: 100.0000 (92.2968) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0002 max mem: 52150 +Test: [1210/1403] eta: 0:00:19 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0649 (0.2172) acc1: 100.0000 (92.3604) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52150 +Test: [1220/1403] eta: 0:00:18 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0644 (0.2159) acc1: 100.0000 (92.4230) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52150 +Test: [1230/1403] eta: 0:00:17 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0656 (0.2147) acc1: 100.0000 (92.4845) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52150 +Test: [1240/1403] eta: 0:00:16 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0654 (0.2135) acc1: 100.0000 (92.5451) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0002 max mem: 52150 +Test: [1250/1403] eta: 0:00:15 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0647 (0.2123) acc1: 100.0000 (92.6047) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52150 +Test: [1260/1403] eta: 0:00:14 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0650 (0.2111) acc1: 100.0000 (92.6633) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52150 +Test: [1270/1403] eta: 0:00:13 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0661 (0.2100) acc1: 100.0000 (92.7210) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0002 max mem: 52150 +Test: [1280/1403] eta: 0:00:12 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0658 (0.2089) acc1: 100.0000 (92.7779) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52150 +Test: [1290/1403] eta: 0:00:11 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0654 (0.2078) acc1: 100.0000 (92.8338) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52150 +Test: [1300/1403] eta: 0:00:10 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0657 (0.2067) acc1: 100.0000 (92.8889) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52150 +Test: [1310/1403] eta: 0:00:09 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0662 (0.2056) acc1: 100.0000 (92.9431) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52150 +Test: [1320/1403] eta: 0:00:08 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0648 (0.2045) acc1: 100.0000 (92.9965) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52150 +Test: [1330/1403] eta: 0:00:07 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0644 (0.2035) acc1: 100.0000 (93.0480) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52150 +Test: [1340/1403] eta: 0:00:06 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0624 (0.2024) acc1: 100.0000 (93.0998) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52150 +Test: [1350/1403] eta: 0:00:05 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0617 (0.2014) acc1: 100.0000 (93.1486) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0002 max mem: 52150 +Test: [1360/1403] eta: 0:00:04 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0616 (0.2004) acc1: 100.0000 (93.1989) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0002 max mem: 52150 +Test: [1370/1403] eta: 0:00:03 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0622 (0.1994) acc1: 100.0000 (93.2474) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52150 +Test: [1380/1403] eta: 0:00:02 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0617 (0.1984) acc1: 100.0000 (93.2963) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0002 max mem: 52150 +Test: [1390/1403] eta: 0:00:01 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0610 (0.1974) acc1: 100.0000 (93.3445) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0002 max mem: 52150 +Test: [1400/1403] eta: 0:00:00 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0610 (0.1965) acc1: 100.0000 (93.3909) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0001 max mem: 52150 +Test: [1402/1403] eta: 0:00:00 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0613 (0.1963) acc1: 100.0000 (93.3957) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0920 data: 0.0001 max mem: 52150 +Test: Total time: 0:02:19 (0.0992 s / it) +* Overall Acc@1 93.396 Acc@5 100.000 + target == real Acc@1 93.466 Acc@5 100.000 + target == fake Acc@1 93.376 Acc@5 100.000 + loss 0.196 +Accuracy of the network on the 89729 test images: 93.4% +Max accuracy: 93.40% +Epoch: [17] [ 0/7269] eta: 9:04:12 lr: 0.000007 loss: 0.6026 (0.6026) time: 4.4920 data: 4.1257 max mem: 52150 +Epoch: [17] [ 10/7269] eta: 1:30:40 lr: 0.000007 loss: 0.5414 (0.5196) time: 0.7494 data: 0.3752 max mem: 52150 +Epoch: [17] [ 20/7269] eta: 1:08:46 lr: 0.000007 loss: 0.5474 (0.5356) time: 0.3732 data: 0.0001 max mem: 52150 +Epoch: [17] [ 30/7269] eta: 1:08:38 lr: 0.000007 loss: 0.5501 (0.5288) time: 0.4696 data: 0.0001 max mem: 52150 +Epoch: [17] [ 40/7269] eta: 1:02:43 lr: 0.000007 loss: 0.5413 (0.5330) time: 0.4696 data: 0.0001 max mem: 52150 +Epoch: [17] [ 50/7269] eta: 0:59:07 lr: 0.000007 loss: 0.5502 (0.5330) time: 0.3712 data: 0.0002 max mem: 52150 +Epoch: [17] [ 60/7269] eta: 1:03:33 lr: 0.000007 loss: 0.5505 (0.5358) time: 0.5459 data: 0.0002 max mem: 52150 +Epoch: [17] [ 70/7269] eta: 1:00:48 lr: 0.000007 loss: 0.5349 (0.5380) time: 0.5460 data: 0.0002 max mem: 52150 +Epoch: [17] [ 80/7269] eta: 1:00:31 lr: 0.000007 loss: 0.5519 (0.5416) time: 0.4328 data: 0.0002 max mem: 52150 +Epoch: [17] [ 90/7269] eta: 1:00:17 lr: 0.000007 loss: 0.5673 (0.5428) time: 0.4940 data: 0.0001 max mem: 52150 +Epoch: [17] [ 100/7269] eta: 0:58:39 lr: 0.000007 loss: 0.5064 (0.5357) time: 0.4333 data: 0.0001 max mem: 52150 +Epoch: [17] [ 110/7269] eta: 0:59:44 lr: 0.000007 loss: 0.4897 (0.5333) time: 0.4859 data: 0.0002 max mem: 52150 +Epoch: [17] [ 120/7269] eta: 0:58:23 lr: 0.000007 loss: 0.5571 (0.5362) time: 0.4852 data: 0.0002 max mem: 52150 +Epoch: [17] [ 130/7269] eta: 0:57:13 lr: 0.000007 loss: 0.5369 (0.5330) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [17] [ 140/7269] eta: 0:56:13 lr: 0.000007 loss: 0.5053 (0.5326) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [17] [ 150/7269] eta: 0:57:53 lr: 0.000007 loss: 0.5161 (0.5307) time: 0.5329 data: 0.0002 max mem: 52150 +Epoch: [17] [ 160/7269] eta: 0:56:56 lr: 0.000007 loss: 0.4741 (0.5277) time: 0.5331 data: 0.0001 max mem: 52150 +Epoch: [17] [ 170/7269] eta: 0:56:06 lr: 0.000007 loss: 0.5085 (0.5289) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [17] [ 180/7269] eta: 0:56:10 lr: 0.000007 loss: 0.5266 (0.5279) time: 0.4333 data: 0.0002 max mem: 52150 +Epoch: [17] [ 190/7269] eta: 0:55:26 lr: 0.000007 loss: 0.5081 (0.5274) time: 0.4330 data: 0.0001 max mem: 52150 +Epoch: [17] [ 200/7269] eta: 0:54:47 lr: 0.000007 loss: 0.5312 (0.5272) time: 0.3710 data: 0.0002 max mem: 52150 +Epoch: [17] [ 210/7269] eta: 0:54:11 lr: 0.000007 loss: 0.5201 (0.5257) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [17] [ 220/7269] eta: 0:54:18 lr: 0.000007 loss: 0.5201 (0.5256) time: 0.4348 data: 0.0002 max mem: 52150 +Epoch: [17] [ 230/7269] eta: 0:53:46 lr: 0.000007 loss: 0.5269 (0.5250) time: 0.4348 data: 0.0001 max mem: 52150 +Epoch: [17] [ 240/7269] eta: 0:53:16 lr: 0.000007 loss: 0.4985 (0.5243) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [17] [ 250/7269] eta: 0:53:15 lr: 0.000007 loss: 0.5233 (0.5242) time: 0.4188 data: 0.0001 max mem: 52150 +Epoch: [17] [ 260/7269] eta: 0:53:21 lr: 0.000007 loss: 0.5391 (0.5259) time: 0.4799 data: 0.0001 max mem: 52150 +Epoch: [17] [ 270/7269] eta: 0:53:27 lr: 0.000007 loss: 0.5488 (0.5264) time: 0.4958 data: 0.0002 max mem: 52150 +Epoch: [17] [ 280/7269] eta: 0:53:01 lr: 0.000007 loss: 0.5269 (0.5260) time: 0.4350 data: 0.0002 max mem: 52150 +Epoch: [17] [ 290/7269] eta: 0:52:36 lr: 0.000007 loss: 0.5162 (0.5262) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [17] [ 300/7269] eta: 0:52:13 lr: 0.000007 loss: 0.5550 (0.5270) time: 0.3711 data: 0.0002 max mem: 52150 +Epoch: [17] [ 310/7269] eta: 0:52:34 lr: 0.000007 loss: 0.5548 (0.5276) time: 0.4688 data: 0.0002 max mem: 52150 +Epoch: [17] [ 320/7269] eta: 0:53:00 lr: 0.000007 loss: 0.5515 (0.5271) time: 0.5810 data: 0.0001 max mem: 52150 +Epoch: [17] [ 330/7269] eta: 0:53:24 lr: 0.000007 loss: 0.5042 (0.5271) time: 0.5936 data: 0.0001 max mem: 52150 +Epoch: [17] [ 340/7269] eta: 0:53:01 lr: 0.000007 loss: 0.5402 (0.5274) time: 0.4818 data: 0.0002 max mem: 52150 +Epoch: [17] [ 350/7269] eta: 0:53:29 lr: 0.000007 loss: 0.5335 (0.5271) time: 0.4968 data: 0.0002 max mem: 52150 +Epoch: [17] [ 360/7269] eta: 0:53:06 lr: 0.000007 loss: 0.5271 (0.5267) time: 0.4969 data: 0.0001 max mem: 52150 +Epoch: [17] [ 370/7269] eta: 0:53:08 lr: 0.000007 loss: 0.5502 (0.5273) time: 0.4333 data: 0.0001 max mem: 52150 +Epoch: [17] [ 380/7269] eta: 0:53:10 lr: 0.000007 loss: 0.5426 (0.5272) time: 0.4970 data: 0.0001 max mem: 52150 +Epoch: [17] [ 390/7269] eta: 0:52:50 lr: 0.000007 loss: 0.5130 (0.5268) time: 0.4362 data: 0.0002 max mem: 52150 +Epoch: [17] [ 400/7269] eta: 0:53:08 lr: 0.000007 loss: 0.5378 (0.5268) time: 0.4849 data: 0.0002 max mem: 52150 +Epoch: [17] [ 410/7269] eta: 0:52:48 lr: 0.000007 loss: 0.5391 (0.5266) time: 0.4846 data: 0.0002 max mem: 52150 +Epoch: [17] [ 420/7269] eta: 0:52:29 lr: 0.000007 loss: 0.5390 (0.5268) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [17] [ 430/7269] eta: 0:52:31 lr: 0.000007 loss: 0.5197 (0.5263) time: 0.4354 data: 0.0002 max mem: 52150 +Epoch: [17] [ 440/7269] eta: 0:52:12 lr: 0.000007 loss: 0.5188 (0.5267) time: 0.4355 data: 0.0002 max mem: 52150 +Epoch: [17] [ 450/7269] eta: 0:51:55 lr: 0.000007 loss: 0.5544 (0.5270) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [17] [ 460/7269] eta: 0:51:38 lr: 0.000007 loss: 0.5549 (0.5273) time: 0.3725 data: 0.0001 max mem: 52150 +Epoch: [17] [ 470/7269] eta: 0:51:39 lr: 0.000007 loss: 0.5666 (0.5279) time: 0.4344 data: 0.0001 max mem: 52150 +Epoch: [17] [ 480/7269] eta: 0:51:23 lr: 0.000007 loss: 0.5703 (0.5279) time: 0.4341 data: 0.0001 max mem: 52150 +Epoch: [17] [ 490/7269] eta: 0:51:20 lr: 0.000007 loss: 0.5462 (0.5280) time: 0.4185 data: 0.0002 max mem: 52150 +Epoch: [17] [ 500/7269] eta: 0:51:04 lr: 0.000007 loss: 0.5203 (0.5275) time: 0.4185 data: 0.0002 max mem: 52150 +Epoch: [17] [ 510/7269] eta: 0:50:49 lr: 0.000007 loss: 0.5225 (0.5279) time: 0.3729 data: 0.0002 max mem: 52150 +Epoch: [17] [ 520/7269] eta: 0:50:34 lr: 0.000007 loss: 0.5225 (0.5278) time: 0.3732 data: 0.0002 max mem: 52150 +Epoch: [17] [ 530/7269] eta: 0:50:48 lr: 0.000007 loss: 0.5170 (0.5269) time: 0.4811 data: 0.0002 max mem: 52150 +Epoch: [17] [ 540/7269] eta: 0:50:33 lr: 0.000007 loss: 0.5155 (0.5273) time: 0.4808 data: 0.0001 max mem: 52150 +Epoch: [17] [ 550/7269] eta: 0:50:48 lr: 0.000007 loss: 0.5155 (0.5268) time: 0.4905 data: 0.0001 max mem: 52150 +Epoch: [17] [ 560/7269] eta: 0:50:34 lr: 0.000007 loss: 0.5287 (0.5274) time: 0.4905 data: 0.0002 max mem: 52150 +Epoch: [17] [ 570/7269] eta: 0:50:30 lr: 0.000007 loss: 0.5420 (0.5274) time: 0.4182 data: 0.0002 max mem: 52150 +Epoch: [17] [ 580/7269] eta: 0:50:28 lr: 0.000007 loss: 0.5271 (0.5273) time: 0.4690 data: 0.0001 max mem: 52150 +Epoch: [17] [ 590/7269] eta: 0:50:28 lr: 0.000007 loss: 0.5374 (0.5272) time: 0.4829 data: 0.0002 max mem: 52150 +Epoch: [17] [ 600/7269] eta: 0:50:15 lr: 0.000007 loss: 0.5276 (0.5271) time: 0.4319 data: 0.0002 max mem: 52150 +Epoch: [17] [ 610/7269] eta: 0:50:11 lr: 0.000007 loss: 0.5301 (0.5274) time: 0.4175 data: 0.0002 max mem: 52150 +Epoch: [17] [ 620/7269] eta: 0:50:11 lr: 0.000007 loss: 0.5509 (0.5277) time: 0.4784 data: 0.0002 max mem: 52150 +Epoch: [17] [ 630/7269] eta: 0:49:58 lr: 0.000007 loss: 0.5509 (0.5278) time: 0.4325 data: 0.0002 max mem: 52150 +Epoch: [17] [ 640/7269] eta: 0:49:45 lr: 0.000007 loss: 0.5308 (0.5278) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [17] [ 650/7269] eta: 0:49:33 lr: 0.000007 loss: 0.5374 (0.5280) time: 0.3712 data: 0.0002 max mem: 52150 +Epoch: [17] [ 660/7269] eta: 0:49:31 lr: 0.000007 loss: 0.5448 (0.5279) time: 0.4238 data: 0.0002 max mem: 52150 +Epoch: [17] [ 670/7269] eta: 0:49:28 lr: 0.000007 loss: 0.4965 (0.5275) time: 0.4701 data: 0.0002 max mem: 52150 +Epoch: [17] [ 680/7269] eta: 0:49:16 lr: 0.000007 loss: 0.5105 (0.5276) time: 0.4176 data: 0.0002 max mem: 52150 +Epoch: [17] [ 690/7269] eta: 0:49:04 lr: 0.000007 loss: 0.5405 (0.5279) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [17] [ 700/7269] eta: 0:48:53 lr: 0.000007 loss: 0.5266 (0.5277) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [17] [ 710/7269] eta: 0:48:51 lr: 0.000007 loss: 0.5517 (0.5283) time: 0.4235 data: 0.0002 max mem: 52150 +Epoch: [17] [ 720/7269] eta: 0:48:50 lr: 0.000007 loss: 0.5517 (0.5278) time: 0.4843 data: 0.0001 max mem: 52150 +Epoch: [17] [ 730/7269] eta: 0:48:39 lr: 0.000007 loss: 0.5195 (0.5280) time: 0.4330 data: 0.0001 max mem: 52150 +Epoch: [17] [ 740/7269] eta: 0:48:28 lr: 0.000007 loss: 0.5507 (0.5284) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [17] [ 750/7269] eta: 0:48:17 lr: 0.000007 loss: 0.5335 (0.5285) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [17] [ 760/7269] eta: 0:48:07 lr: 0.000007 loss: 0.5317 (0.5285) time: 0.3730 data: 0.0002 max mem: 52150 +Epoch: [17] [ 770/7269] eta: 0:47:56 lr: 0.000007 loss: 0.5303 (0.5280) time: 0.3727 data: 0.0002 max mem: 52150 +Epoch: [17] [ 780/7269] eta: 0:47:53 lr: 0.000007 loss: 0.5408 (0.5284) time: 0.4157 data: 0.0001 max mem: 52150 +Epoch: [17] [ 790/7269] eta: 0:47:43 lr: 0.000007 loss: 0.5487 (0.5283) time: 0.4158 data: 0.0002 max mem: 52150 +Epoch: [17] [ 800/7269] eta: 0:47:48 lr: 0.000007 loss: 0.5165 (0.5280) time: 0.4657 data: 0.0002 max mem: 52150 +Epoch: [17] [ 810/7269] eta: 0:47:38 lr: 0.000007 loss: 0.4910 (0.5275) time: 0.4658 data: 0.0002 max mem: 52150 +Epoch: [17] [ 820/7269] eta: 0:47:38 lr: 0.000007 loss: 0.5449 (0.5280) time: 0.4326 data: 0.0002 max mem: 52150 +Epoch: [17] [ 830/7269] eta: 0:47:35 lr: 0.000007 loss: 0.5449 (0.5278) time: 0.4782 data: 0.0002 max mem: 52150 +Epoch: [17] [ 840/7269] eta: 0:47:34 lr: 0.000007 loss: 0.5181 (0.5278) time: 0.4793 data: 0.0002 max mem: 52150 +Epoch: [17] [ 850/7269] eta: 0:47:34 lr: 0.000007 loss: 0.5216 (0.5278) time: 0.4984 data: 0.0002 max mem: 52150 +Epoch: [17] [ 860/7269] eta: 0:47:31 lr: 0.000007 loss: 0.4988 (0.5270) time: 0.4825 data: 0.0002 max mem: 52150 +Epoch: [17] [ 870/7269] eta: 0:47:21 lr: 0.000007 loss: 0.4504 (0.5267) time: 0.4177 data: 0.0002 max mem: 52150 +Epoch: [17] [ 880/7269] eta: 0:47:12 lr: 0.000007 loss: 0.5455 (0.5269) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [17] [ 890/7269] eta: 0:47:20 lr: 0.000007 loss: 0.5440 (0.5266) time: 0.4961 data: 0.0002 max mem: 52150 +Epoch: [17] [ 900/7269] eta: 0:47:10 lr: 0.000007 loss: 0.5008 (0.5265) time: 0.4957 data: 0.0002 max mem: 52150 +Epoch: [17] [ 910/7269] eta: 0:47:01 lr: 0.000007 loss: 0.5310 (0.5265) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [17] [ 920/7269] eta: 0:46:51 lr: 0.000007 loss: 0.5370 (0.5264) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [17] [ 930/7269] eta: 0:46:42 lr: 0.000007 loss: 0.5165 (0.5261) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [17] [ 940/7269] eta: 0:46:33 lr: 0.000007 loss: 0.5165 (0.5260) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [17] [ 950/7269] eta: 0:46:24 lr: 0.000007 loss: 0.5157 (0.5261) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [17] [ 960/7269] eta: 0:46:15 lr: 0.000007 loss: 0.5557 (0.5265) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [17] [ 970/7269] eta: 0:46:06 lr: 0.000007 loss: 0.5412 (0.5264) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [17] [ 980/7269] eta: 0:45:57 lr: 0.000007 loss: 0.5427 (0.5265) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [17] [ 990/7269] eta: 0:45:55 lr: 0.000007 loss: 0.5533 (0.5267) time: 0.4179 data: 0.0002 max mem: 52150 +Epoch: [17] [1000/7269] eta: 0:45:46 lr: 0.000007 loss: 0.5710 (0.5268) time: 0.4178 data: 0.0002 max mem: 52150 +Epoch: [17] [1010/7269] eta: 0:45:38 lr: 0.000007 loss: 0.5645 (0.5270) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [17] [1020/7269] eta: 0:45:29 lr: 0.000007 loss: 0.5645 (0.5270) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [17] [1030/7269] eta: 0:45:27 lr: 0.000007 loss: 0.5803 (0.5275) time: 0.4238 data: 0.0002 max mem: 52150 +Epoch: [17] [1040/7269] eta: 0:45:19 lr: 0.000007 loss: 0.5742 (0.5277) time: 0.4237 data: 0.0002 max mem: 52150 +Epoch: [17] [1050/7269] eta: 0:45:24 lr: 0.000007 loss: 0.5103 (0.5275) time: 0.4840 data: 0.0002 max mem: 52150 +Epoch: [17] [1060/7269] eta: 0:45:23 lr: 0.000007 loss: 0.4973 (0.5271) time: 0.5450 data: 0.0002 max mem: 52150 +Epoch: [17] [1070/7269] eta: 0:45:15 lr: 0.000007 loss: 0.5219 (0.5272) time: 0.4331 data: 0.0002 max mem: 52150 +Epoch: [17] [1080/7269] eta: 0:45:13 lr: 0.000007 loss: 0.5666 (0.5276) time: 0.4320 data: 0.0002 max mem: 52150 +Epoch: [17] [1090/7269] eta: 0:45:12 lr: 0.000007 loss: 0.5571 (0.5277) time: 0.4923 data: 0.0002 max mem: 52150 +Epoch: [17] [1100/7269] eta: 0:45:04 lr: 0.000007 loss: 0.5344 (0.5276) time: 0.4320 data: 0.0002 max mem: 52150 +Epoch: [17] [1110/7269] eta: 0:45:03 lr: 0.000007 loss: 0.4958 (0.5273) time: 0.4328 data: 0.0002 max mem: 52150 +Epoch: [17] [1120/7269] eta: 0:45:00 lr: 0.000007 loss: 0.4968 (0.5274) time: 0.4820 data: 0.0002 max mem: 52150 +Epoch: [17] [1130/7269] eta: 0:44:58 lr: 0.000007 loss: 0.5431 (0.5275) time: 0.4815 data: 0.0002 max mem: 52150 +Epoch: [17] [1140/7269] eta: 0:44:50 lr: 0.000007 loss: 0.5431 (0.5274) time: 0.4332 data: 0.0002 max mem: 52150 +Epoch: [17] [1150/7269] eta: 0:44:43 lr: 0.000007 loss: 0.5152 (0.5271) time: 0.3727 data: 0.0002 max mem: 52150 +Epoch: [17] [1160/7269] eta: 0:44:35 lr: 0.000007 loss: 0.5205 (0.5271) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [17] [1170/7269] eta: 0:44:33 lr: 0.000007 loss: 0.5426 (0.5272) time: 0.4331 data: 0.0001 max mem: 52150 +Epoch: [17] [1180/7269] eta: 0:44:32 lr: 0.000007 loss: 0.5287 (0.5272) time: 0.4929 data: 0.0002 max mem: 52150 +Epoch: [17] [1190/7269] eta: 0:44:24 lr: 0.000007 loss: 0.5176 (0.5270) time: 0.4323 data: 0.0002 max mem: 52150 +Epoch: [17] [1200/7269] eta: 0:44:22 lr: 0.000007 loss: 0.5176 (0.5270) time: 0.4332 data: 0.0002 max mem: 52150 +Epoch: [17] [1210/7269] eta: 0:44:19 lr: 0.000007 loss: 0.5297 (0.5269) time: 0.4822 data: 0.0001 max mem: 52150 +Epoch: [17] [1220/7269] eta: 0:44:16 lr: 0.000007 loss: 0.5272 (0.5271) time: 0.4688 data: 0.0002 max mem: 52150 +Epoch: [17] [1230/7269] eta: 0:44:13 lr: 0.000007 loss: 0.5272 (0.5269) time: 0.4668 data: 0.0002 max mem: 52150 +Epoch: [17] [1240/7269] eta: 0:44:06 lr: 0.000007 loss: 0.5160 (0.5269) time: 0.4191 data: 0.0002 max mem: 52150 +Epoch: [17] [1250/7269] eta: 0:43:58 lr: 0.000007 loss: 0.5160 (0.5269) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [17] [1260/7269] eta: 0:43:51 lr: 0.000007 loss: 0.5339 (0.5268) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [17] [1270/7269] eta: 0:43:43 lr: 0.000007 loss: 0.5339 (0.5268) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [17] [1280/7269] eta: 0:43:36 lr: 0.000007 loss: 0.5390 (0.5271) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [17] [1290/7269] eta: 0:43:33 lr: 0.000007 loss: 0.5147 (0.5270) time: 0.4225 data: 0.0002 max mem: 52150 +Epoch: [17] [1300/7269] eta: 0:43:26 lr: 0.000007 loss: 0.5147 (0.5271) time: 0.4227 data: 0.0002 max mem: 52150 +Epoch: [17] [1310/7269] eta: 0:43:24 lr: 0.000007 loss: 0.5583 (0.5272) time: 0.4312 data: 0.0002 max mem: 52150 +Epoch: [17] [1320/7269] eta: 0:43:27 lr: 0.000007 loss: 0.5298 (0.5273) time: 0.5539 data: 0.0002 max mem: 52150 +Epoch: [17] [1330/7269] eta: 0:43:20 lr: 0.000007 loss: 0.5343 (0.5273) time: 0.4944 data: 0.0002 max mem: 52150 +Epoch: [17] [1340/7269] eta: 0:43:17 lr: 0.000007 loss: 0.5469 (0.5275) time: 0.4191 data: 0.0002 max mem: 52150 +Epoch: [17] [1350/7269] eta: 0:43:18 lr: 0.000007 loss: 0.5469 (0.5277) time: 0.5152 data: 0.0002 max mem: 52150 +Epoch: [17] [1360/7269] eta: 0:43:11 lr: 0.000007 loss: 0.5438 (0.5276) time: 0.4685 data: 0.0002 max mem: 52150 +Epoch: [17] [1370/7269] eta: 0:43:04 lr: 0.000007 loss: 0.5493 (0.5276) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [17] [1380/7269] eta: 0:43:04 lr: 0.000007 loss: 0.5250 (0.5275) time: 0.4643 data: 0.0002 max mem: 52150 +Epoch: [17] [1390/7269] eta: 0:42:57 lr: 0.000007 loss: 0.4874 (0.5272) time: 0.4643 data: 0.0002 max mem: 52150 +Epoch: [17] [1400/7269] eta: 0:42:50 lr: 0.000007 loss: 0.4951 (0.5272) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [17] [1410/7269] eta: 0:42:48 lr: 0.000007 loss: 0.5351 (0.5273) time: 0.4351 data: 0.0002 max mem: 52150 +Epoch: [17] [1420/7269] eta: 0:42:41 lr: 0.000007 loss: 0.5099 (0.5272) time: 0.4350 data: 0.0002 max mem: 52150 +Epoch: [17] [1430/7269] eta: 0:42:34 lr: 0.000007 loss: 0.5080 (0.5272) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [17] [1440/7269] eta: 0:42:31 lr: 0.000007 loss: 0.5203 (0.5271) time: 0.4178 data: 0.0002 max mem: 52150 +Epoch: [17] [1450/7269] eta: 0:42:28 lr: 0.000007 loss: 0.4855 (0.5267) time: 0.4784 data: 0.0002 max mem: 52150 +Epoch: [17] [1460/7269] eta: 0:42:21 lr: 0.000007 loss: 0.5078 (0.5268) time: 0.4328 data: 0.0002 max mem: 52150 +Epoch: [17] [1470/7269] eta: 0:42:19 lr: 0.000007 loss: 0.5364 (0.5269) time: 0.4329 data: 0.0002 max mem: 52150 +Epoch: [17] [1480/7269] eta: 0:42:16 lr: 0.000007 loss: 0.5327 (0.5267) time: 0.4850 data: 0.0002 max mem: 52150 +Epoch: [17] [1490/7269] eta: 0:42:13 lr: 0.000007 loss: 0.4706 (0.5264) time: 0.4703 data: 0.0002 max mem: 52150 +Epoch: [17] [1500/7269] eta: 0:42:06 lr: 0.000007 loss: 0.5279 (0.5265) time: 0.4178 data: 0.0002 max mem: 52150 +Epoch: [17] [1510/7269] eta: 0:41:59 lr: 0.000007 loss: 0.5139 (0.5264) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [17] [1520/7269] eta: 0:41:56 lr: 0.000007 loss: 0.5133 (0.5262) time: 0.4187 data: 0.0002 max mem: 52150 +Epoch: [17] [1530/7269] eta: 0:41:49 lr: 0.000007 loss: 0.5146 (0.5261) time: 0.4183 data: 0.0002 max mem: 52150 +Epoch: [17] [1540/7269] eta: 0:41:42 lr: 0.000007 loss: 0.5374 (0.5264) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [17] [1550/7269] eta: 0:41:44 lr: 0.000007 loss: 0.5521 (0.5264) time: 0.4931 data: 0.0002 max mem: 52150 +Epoch: [17] [1560/7269] eta: 0:41:37 lr: 0.000007 loss: 0.5494 (0.5262) time: 0.4926 data: 0.0001 max mem: 52150 +Epoch: [17] [1570/7269] eta: 0:41:31 lr: 0.000007 loss: 0.5016 (0.5262) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [17] [1580/7269] eta: 0:41:33 lr: 0.000007 loss: 0.5115 (0.5262) time: 0.4940 data: 0.0002 max mem: 52150 +Epoch: [17] [1590/7269] eta: 0:41:26 lr: 0.000007 loss: 0.5115 (0.5261) time: 0.4944 data: 0.0002 max mem: 52150 +Epoch: [17] [1600/7269] eta: 0:41:19 lr: 0.000007 loss: 0.5258 (0.5263) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [17] [1610/7269] eta: 0:41:13 lr: 0.000007 loss: 0.5702 (0.5265) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [17] [1620/7269] eta: 0:41:06 lr: 0.000007 loss: 0.5630 (0.5267) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [17] [1630/7269] eta: 0:41:08 lr: 0.000007 loss: 0.5335 (0.5265) time: 0.4927 data: 0.0002 max mem: 52150 +Epoch: [17] [1640/7269] eta: 0:41:01 lr: 0.000007 loss: 0.4882 (0.5262) time: 0.4929 data: 0.0002 max mem: 52150 +Epoch: [17] [1650/7269] eta: 0:40:55 lr: 0.000007 loss: 0.5053 (0.5262) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [17] [1660/7269] eta: 0:40:48 lr: 0.000007 loss: 0.5473 (0.5264) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [17] [1670/7269] eta: 0:40:46 lr: 0.000007 loss: 0.5473 (0.5267) time: 0.4328 data: 0.0002 max mem: 52150 +Epoch: [17] [1680/7269] eta: 0:40:43 lr: 0.000007 loss: 0.5424 (0.5266) time: 0.4923 data: 0.0002 max mem: 52150 +Epoch: [17] [1690/7269] eta: 0:40:40 lr: 0.000007 loss: 0.5340 (0.5266) time: 0.4794 data: 0.0002 max mem: 52150 +Epoch: [17] [1700/7269] eta: 0:40:36 lr: 0.000007 loss: 0.5277 (0.5265) time: 0.4654 data: 0.0001 max mem: 52150 +Epoch: [17] [1710/7269] eta: 0:40:30 lr: 0.000007 loss: 0.5240 (0.5265) time: 0.4175 data: 0.0002 max mem: 52150 +Epoch: [17] [1720/7269] eta: 0:40:23 lr: 0.000007 loss: 0.5529 (0.5266) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [17] [1730/7269] eta: 0:40:21 lr: 0.000007 loss: 0.5529 (0.5266) time: 0.4352 data: 0.0002 max mem: 52150 +Epoch: [17] [1740/7269] eta: 0:40:14 lr: 0.000007 loss: 0.5301 (0.5265) time: 0.4348 data: 0.0002 max mem: 52150 +Epoch: [17] [1750/7269] eta: 0:40:08 lr: 0.000007 loss: 0.5444 (0.5267) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [17] [1760/7269] eta: 0:40:04 lr: 0.000007 loss: 0.5229 (0.5265) time: 0.4184 data: 0.0002 max mem: 52150 +Epoch: [17] [1770/7269] eta: 0:39:58 lr: 0.000007 loss: 0.5213 (0.5265) time: 0.4192 data: 0.0001 max mem: 52150 +Epoch: [17] [1780/7269] eta: 0:39:55 lr: 0.000007 loss: 0.5330 (0.5264) time: 0.4319 data: 0.0002 max mem: 52150 +Epoch: [17] [1790/7269] eta: 0:39:53 lr: 0.000007 loss: 0.5146 (0.5265) time: 0.4904 data: 0.0002 max mem: 52150 +Epoch: [17] [1800/7269] eta: 0:39:49 lr: 0.000007 loss: 0.5146 (0.5263) time: 0.4761 data: 0.0002 max mem: 52150 +Epoch: [17] [1810/7269] eta: 0:39:43 lr: 0.000007 loss: 0.4831 (0.5262) time: 0.4170 data: 0.0002 max mem: 52150 +Epoch: [17] [1820/7269] eta: 0:39:39 lr: 0.000007 loss: 0.4947 (0.5261) time: 0.4196 data: 0.0001 max mem: 52150 +Epoch: [17] [1830/7269] eta: 0:39:39 lr: 0.000007 loss: 0.5247 (0.5259) time: 0.5245 data: 0.0002 max mem: 52150 +Epoch: [17] [1840/7269] eta: 0:39:36 lr: 0.000007 loss: 0.5303 (0.5260) time: 0.5227 data: 0.0002 max mem: 52150 +Epoch: [17] [1850/7269] eta: 0:39:29 lr: 0.000007 loss: 0.5594 (0.5260) time: 0.4174 data: 0.0002 max mem: 52150 +Epoch: [17] [1860/7269] eta: 0:39:26 lr: 0.000007 loss: 0.5644 (0.5262) time: 0.4167 data: 0.0001 max mem: 52150 +Epoch: [17] [1870/7269] eta: 0:39:19 lr: 0.000007 loss: 0.5352 (0.5262) time: 0.4167 data: 0.0001 max mem: 52150 +Epoch: [17] [1880/7269] eta: 0:39:13 lr: 0.000007 loss: 0.5301 (0.5261) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [17] [1890/7269] eta: 0:39:07 lr: 0.000007 loss: 0.5301 (0.5262) time: 0.3723 data: 0.0003 max mem: 52150 +Epoch: [17] [1900/7269] eta: 0:39:01 lr: 0.000007 loss: 0.5332 (0.5262) time: 0.3725 data: 0.0003 max mem: 52150 +Epoch: [17] [1910/7269] eta: 0:38:55 lr: 0.000007 loss: 0.5356 (0.5262) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [17] [1920/7269] eta: 0:38:48 lr: 0.000007 loss: 0.5190 (0.5261) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [17] [1930/7269] eta: 0:38:45 lr: 0.000007 loss: 0.5028 (0.5260) time: 0.4203 data: 0.0001 max mem: 52150 +Epoch: [17] [1940/7269] eta: 0:38:39 lr: 0.000007 loss: 0.5148 (0.5260) time: 0.4204 data: 0.0001 max mem: 52150 +Epoch: [17] [1950/7269] eta: 0:38:36 lr: 0.000007 loss: 0.5246 (0.5260) time: 0.4345 data: 0.0001 max mem: 52150 +Epoch: [17] [1960/7269] eta: 0:38:33 lr: 0.000007 loss: 0.5199 (0.5260) time: 0.4797 data: 0.0002 max mem: 52150 +Epoch: [17] [1970/7269] eta: 0:38:30 lr: 0.000007 loss: 0.5388 (0.5262) time: 0.4792 data: 0.0002 max mem: 52150 +Epoch: [17] [1980/7269] eta: 0:38:26 lr: 0.000007 loss: 0.5388 (0.5261) time: 0.4824 data: 0.0002 max mem: 52150 +Epoch: [17] [1990/7269] eta: 0:38:20 lr: 0.000007 loss: 0.4757 (0.5259) time: 0.4201 data: 0.0002 max mem: 52150 +Epoch: [17] [2000/7269] eta: 0:38:14 lr: 0.000007 loss: 0.5094 (0.5259) time: 0.3711 data: 0.0002 max mem: 52150 +Epoch: [17] [2010/7269] eta: 0:38:08 lr: 0.000007 loss: 0.5157 (0.5258) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [17] [2020/7269] eta: 0:38:02 lr: 0.000007 loss: 0.4608 (0.5255) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [17] [2030/7269] eta: 0:37:56 lr: 0.000007 loss: 0.4804 (0.5255) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [17] [2040/7269] eta: 0:37:56 lr: 0.000007 loss: 0.5468 (0.5257) time: 0.4914 data: 0.0002 max mem: 52150 +Epoch: [17] [2050/7269] eta: 0:37:54 lr: 0.000007 loss: 0.5550 (0.5258) time: 0.5521 data: 0.0002 max mem: 52150 +Epoch: [17] [2060/7269] eta: 0:37:48 lr: 0.000007 loss: 0.5118 (0.5256) time: 0.4329 data: 0.0002 max mem: 52150 +Epoch: [17] [2070/7269] eta: 0:37:44 lr: 0.000007 loss: 0.5327 (0.5257) time: 0.4222 data: 0.0001 max mem: 52150 +Epoch: [17] [2080/7269] eta: 0:37:44 lr: 0.000007 loss: 0.5361 (0.5256) time: 0.5306 data: 0.0002 max mem: 52150 +Epoch: [17] [2090/7269] eta: 0:37:41 lr: 0.000007 loss: 0.5093 (0.5256) time: 0.5423 data: 0.0002 max mem: 52150 +Epoch: [17] [2100/7269] eta: 0:37:38 lr: 0.000007 loss: 0.5110 (0.5256) time: 0.4937 data: 0.0002 max mem: 52150 +Epoch: [17] [2110/7269] eta: 0:37:35 lr: 0.000007 loss: 0.5331 (0.5256) time: 0.4919 data: 0.0002 max mem: 52150 +Epoch: [17] [2120/7269] eta: 0:37:32 lr: 0.000007 loss: 0.5331 (0.5257) time: 0.4913 data: 0.0002 max mem: 52150 +Epoch: [17] [2130/7269] eta: 0:37:26 lr: 0.000007 loss: 0.5587 (0.5257) time: 0.4317 data: 0.0002 max mem: 52150 +Epoch: [17] [2140/7269] eta: 0:37:25 lr: 0.000007 loss: 0.5587 (0.5258) time: 0.4731 data: 0.0002 max mem: 52150 +Epoch: [17] [2150/7269] eta: 0:37:21 lr: 0.000007 loss: 0.5425 (0.5258) time: 0.5192 data: 0.0002 max mem: 52150 +Epoch: [17] [2160/7269] eta: 0:37:15 lr: 0.000007 loss: 0.5474 (0.5259) time: 0.4183 data: 0.0001 max mem: 52150 +Epoch: [17] [2170/7269] eta: 0:37:09 lr: 0.000007 loss: 0.5474 (0.5259) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [17] [2180/7269] eta: 0:37:05 lr: 0.000007 loss: 0.5159 (0.5259) time: 0.4176 data: 0.0002 max mem: 52150 +Epoch: [17] [2190/7269] eta: 0:36:59 lr: 0.000007 loss: 0.5156 (0.5258) time: 0.4179 data: 0.0002 max mem: 52150 +Epoch: [17] [2200/7269] eta: 0:36:53 lr: 0.000007 loss: 0.5185 (0.5258) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [17] [2210/7269] eta: 0:36:50 lr: 0.000007 loss: 0.5450 (0.5260) time: 0.4209 data: 0.0002 max mem: 52150 +Epoch: [17] [2220/7269] eta: 0:36:47 lr: 0.000007 loss: 0.5781 (0.5262) time: 0.4827 data: 0.0002 max mem: 52150 +Epoch: [17] [2230/7269] eta: 0:36:44 lr: 0.000007 loss: 0.5717 (0.5263) time: 0.4957 data: 0.0002 max mem: 52150 +Epoch: [17] [2240/7269] eta: 0:36:38 lr: 0.000007 loss: 0.5424 (0.5264) time: 0.4337 data: 0.0002 max mem: 52150 +Epoch: [17] [2250/7269] eta: 0:36:32 lr: 0.000007 loss: 0.5219 (0.5262) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [17] [2260/7269] eta: 0:36:26 lr: 0.000007 loss: 0.4878 (0.5261) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [17] [2270/7269] eta: 0:36:23 lr: 0.000007 loss: 0.5212 (0.5261) time: 0.4331 data: 0.0002 max mem: 52150 +Epoch: [17] [2280/7269] eta: 0:36:20 lr: 0.000007 loss: 0.5132 (0.5259) time: 0.4945 data: 0.0002 max mem: 52150 +Epoch: [17] [2290/7269] eta: 0:36:19 lr: 0.000007 loss: 0.5263 (0.5260) time: 0.5365 data: 0.0002 max mem: 52150 +Epoch: [17] [2300/7269] eta: 0:36:13 lr: 0.000007 loss: 0.5366 (0.5261) time: 0.4755 data: 0.0002 max mem: 52150 +Epoch: [17] [2310/7269] eta: 0:36:07 lr: 0.000007 loss: 0.5640 (0.5262) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [17] [2320/7269] eta: 0:36:01 lr: 0.000007 loss: 0.5578 (0.5261) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [17] [2330/7269] eta: 0:35:56 lr: 0.000007 loss: 0.5524 (0.5262) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [17] [2340/7269] eta: 0:35:50 lr: 0.000007 loss: 0.5479 (0.5262) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [17] [2350/7269] eta: 0:35:44 lr: 0.000007 loss: 0.5330 (0.5262) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [17] [2360/7269] eta: 0:35:39 lr: 0.000007 loss: 0.5270 (0.5261) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [17] [2370/7269] eta: 0:35:33 lr: 0.000007 loss: 0.4935 (0.5259) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [17] [2380/7269] eta: 0:35:27 lr: 0.000007 loss: 0.5182 (0.5259) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [17] [2390/7269] eta: 0:35:24 lr: 0.000007 loss: 0.5182 (0.5259) time: 0.4331 data: 0.0002 max mem: 52150 +Epoch: [17] [2400/7269] eta: 0:35:18 lr: 0.000007 loss: 0.5165 (0.5258) time: 0.4328 data: 0.0002 max mem: 52150 +Epoch: [17] [2410/7269] eta: 0:35:18 lr: 0.000007 loss: 0.5353 (0.5259) time: 0.5113 data: 0.0002 max mem: 52150 +Epoch: [17] [2420/7269] eta: 0:35:15 lr: 0.000007 loss: 0.5515 (0.5259) time: 0.5579 data: 0.0002 max mem: 52150 +Epoch: [17] [2430/7269] eta: 0:35:11 lr: 0.000007 loss: 0.5239 (0.5259) time: 0.4631 data: 0.0002 max mem: 52150 +Epoch: [17] [2440/7269] eta: 0:35:05 lr: 0.000007 loss: 0.5390 (0.5261) time: 0.4172 data: 0.0002 max mem: 52150 +Epoch: [17] [2450/7269] eta: 0:35:00 lr: 0.000007 loss: 0.5345 (0.5259) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [17] [2460/7269] eta: 0:34:54 lr: 0.000007 loss: 0.5152 (0.5260) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [17] [2470/7269] eta: 0:34:48 lr: 0.000007 loss: 0.5382 (0.5260) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [17] [2480/7269] eta: 0:34:46 lr: 0.000007 loss: 0.4935 (0.5258) time: 0.4679 data: 0.0002 max mem: 52150 +Epoch: [17] [2490/7269] eta: 0:34:41 lr: 0.000007 loss: 0.4855 (0.5258) time: 0.4678 data: 0.0002 max mem: 52150 +Epoch: [17] [2500/7269] eta: 0:34:37 lr: 0.000007 loss: 0.4886 (0.5258) time: 0.4174 data: 0.0002 max mem: 52150 +Epoch: [17] [2510/7269] eta: 0:34:34 lr: 0.000007 loss: 0.5074 (0.5257) time: 0.4816 data: 0.0002 max mem: 52150 +Epoch: [17] [2520/7269] eta: 0:34:28 lr: 0.000007 loss: 0.5074 (0.5257) time: 0.4361 data: 0.0002 max mem: 52150 +Epoch: [17] [2530/7269] eta: 0:34:25 lr: 0.000007 loss: 0.5299 (0.5256) time: 0.4339 data: 0.0002 max mem: 52150 +Epoch: [17] [2540/7269] eta: 0:34:20 lr: 0.000007 loss: 0.5479 (0.5257) time: 0.4339 data: 0.0002 max mem: 52150 +Epoch: [17] [2550/7269] eta: 0:34:16 lr: 0.000007 loss: 0.5593 (0.5258) time: 0.4320 data: 0.0002 max mem: 52150 +Epoch: [17] [2560/7269] eta: 0:34:11 lr: 0.000007 loss: 0.5477 (0.5258) time: 0.4320 data: 0.0002 max mem: 52150 +Epoch: [17] [2570/7269] eta: 0:34:05 lr: 0.000007 loss: 0.5364 (0.5258) time: 0.3727 data: 0.0002 max mem: 52150 +Epoch: [17] [2580/7269] eta: 0:34:00 lr: 0.000007 loss: 0.5480 (0.5259) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [17] [2590/7269] eta: 0:33:56 lr: 0.000007 loss: 0.5209 (0.5259) time: 0.4312 data: 0.0002 max mem: 52150 +Epoch: [17] [2600/7269] eta: 0:33:51 lr: 0.000007 loss: 0.5209 (0.5260) time: 0.4315 data: 0.0002 max mem: 52150 +Epoch: [17] [2610/7269] eta: 0:33:47 lr: 0.000007 loss: 0.5602 (0.5260) time: 0.4237 data: 0.0002 max mem: 52150 +Epoch: [17] [2620/7269] eta: 0:33:42 lr: 0.000007 loss: 0.5602 (0.5261) time: 0.4239 data: 0.0002 max mem: 52150 +Epoch: [17] [2630/7269] eta: 0:33:38 lr: 0.000007 loss: 0.5470 (0.5261) time: 0.4323 data: 0.0002 max mem: 52150 +Epoch: [17] [2640/7269] eta: 0:33:35 lr: 0.000007 loss: 0.5350 (0.5261) time: 0.4781 data: 0.0002 max mem: 52150 +Epoch: [17] [2650/7269] eta: 0:33:29 lr: 0.000007 loss: 0.4988 (0.5260) time: 0.4177 data: 0.0002 max mem: 52150 +Epoch: [17] [2660/7269] eta: 0:33:24 lr: 0.000007 loss: 0.4997 (0.5260) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [17] [2670/7269] eta: 0:33:20 lr: 0.000007 loss: 0.5327 (0.5260) time: 0.4360 data: 0.0002 max mem: 52150 +Epoch: [17] [2680/7269] eta: 0:33:19 lr: 0.000007 loss: 0.5359 (0.5261) time: 0.5569 data: 0.0002 max mem: 52150 +Epoch: [17] [2690/7269] eta: 0:33:14 lr: 0.000007 loss: 0.5246 (0.5260) time: 0.4929 data: 0.0002 max mem: 52150 +Epoch: [17] [2700/7269] eta: 0:33:08 lr: 0.000007 loss: 0.5062 (0.5260) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [17] [2710/7269] eta: 0:33:03 lr: 0.000007 loss: 0.5368 (0.5261) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [17] [2720/7269] eta: 0:32:57 lr: 0.000007 loss: 0.5368 (0.5260) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [17] [2730/7269] eta: 0:32:56 lr: 0.000007 loss: 0.5254 (0.5260) time: 0.4838 data: 0.0002 max mem: 52150 +Epoch: [17] [2740/7269] eta: 0:32:52 lr: 0.000007 loss: 0.5180 (0.5259) time: 0.5450 data: 0.0002 max mem: 52150 +Epoch: [17] [2750/7269] eta: 0:32:50 lr: 0.000007 loss: 0.5119 (0.5259) time: 0.5296 data: 0.0001 max mem: 52150 +Epoch: [17] [2760/7269] eta: 0:32:45 lr: 0.000007 loss: 0.5325 (0.5259) time: 0.4689 data: 0.0001 max mem: 52150 +Epoch: [17] [2770/7269] eta: 0:32:39 lr: 0.000007 loss: 0.5343 (0.5259) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [17] [2780/7269] eta: 0:32:37 lr: 0.000007 loss: 0.5339 (0.5259) time: 0.4754 data: 0.0001 max mem: 52150 +Epoch: [17] [2790/7269] eta: 0:32:34 lr: 0.000007 loss: 0.5415 (0.5260) time: 0.5382 data: 0.0001 max mem: 52150 +Epoch: [17] [2800/7269] eta: 0:32:29 lr: 0.000007 loss: 0.5561 (0.5260) time: 0.4341 data: 0.0002 max mem: 52150 +Epoch: [17] [2810/7269] eta: 0:32:23 lr: 0.000007 loss: 0.5333 (0.5260) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [17] [2820/7269] eta: 0:32:18 lr: 0.000007 loss: 0.5333 (0.5261) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [17] [2830/7269] eta: 0:32:14 lr: 0.000007 loss: 0.5436 (0.5260) time: 0.4332 data: 0.0002 max mem: 52150 +Epoch: [17] [2840/7269] eta: 0:32:09 lr: 0.000007 loss: 0.4975 (0.5259) time: 0.4330 data: 0.0002 max mem: 52150 +Epoch: [17] [2850/7269] eta: 0:32:04 lr: 0.000007 loss: 0.4685 (0.5258) time: 0.3714 data: 0.0001 max mem: 52150 +Epoch: [17] [2860/7269] eta: 0:32:00 lr: 0.000007 loss: 0.5272 (0.5259) time: 0.4316 data: 0.0002 max mem: 52150 +Epoch: [17] [2870/7269] eta: 0:31:59 lr: 0.000007 loss: 0.5381 (0.5260) time: 0.5567 data: 0.0002 max mem: 52150 +Epoch: [17] [2880/7269] eta: 0:31:53 lr: 0.000007 loss: 0.5285 (0.5260) time: 0.4967 data: 0.0002 max mem: 52150 +Epoch: [17] [2890/7269] eta: 0:31:48 lr: 0.000007 loss: 0.5285 (0.5261) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [17] [2900/7269] eta: 0:31:45 lr: 0.000007 loss: 0.5268 (0.5260) time: 0.4343 data: 0.0002 max mem: 52150 +Epoch: [17] [2910/7269] eta: 0:31:39 lr: 0.000007 loss: 0.5110 (0.5259) time: 0.4341 data: 0.0002 max mem: 52150 +Epoch: [17] [2920/7269] eta: 0:31:34 lr: 0.000007 loss: 0.4691 (0.5258) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [17] [2930/7269] eta: 0:31:30 lr: 0.000007 loss: 0.5100 (0.5258) time: 0.4328 data: 0.0002 max mem: 52150 +Epoch: [17] [2940/7269] eta: 0:31:25 lr: 0.000007 loss: 0.5115 (0.5257) time: 0.4336 data: 0.0001 max mem: 52150 +Epoch: [17] [2950/7269] eta: 0:31:20 lr: 0.000007 loss: 0.5507 (0.5257) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [17] [2960/7269] eta: 0:31:16 lr: 0.000007 loss: 0.5412 (0.5257) time: 0.4218 data: 0.0002 max mem: 52150 +Epoch: [17] [2970/7269] eta: 0:31:11 lr: 0.000007 loss: 0.5184 (0.5256) time: 0.4219 data: 0.0002 max mem: 52150 +Epoch: [17] [2980/7269] eta: 0:31:07 lr: 0.000007 loss: 0.5338 (0.5257) time: 0.4193 data: 0.0002 max mem: 52150 +Epoch: [17] [2990/7269] eta: 0:31:04 lr: 0.000007 loss: 0.5456 (0.5257) time: 0.5096 data: 0.0001 max mem: 52150 +Epoch: [17] [3000/7269] eta: 0:30:59 lr: 0.000007 loss: 0.5456 (0.5258) time: 0.4619 data: 0.0001 max mem: 52150 +Epoch: [17] [3010/7269] eta: 0:30:54 lr: 0.000007 loss: 0.5496 (0.5258) time: 0.3724 data: 0.0001 max mem: 52150 +Epoch: [17] [3020/7269] eta: 0:30:50 lr: 0.000007 loss: 0.5077 (0.5257) time: 0.4335 data: 0.0001 max mem: 52150 +Epoch: [17] [3030/7269] eta: 0:30:46 lr: 0.000007 loss: 0.5077 (0.5257) time: 0.4793 data: 0.0002 max mem: 52150 +Epoch: [17] [3040/7269] eta: 0:30:41 lr: 0.000007 loss: 0.5545 (0.5257) time: 0.4184 data: 0.0002 max mem: 52150 +Epoch: [17] [3050/7269] eta: 0:30:36 lr: 0.000007 loss: 0.5313 (0.5258) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [17] [3060/7269] eta: 0:30:31 lr: 0.000007 loss: 0.5273 (0.5258) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [17] [3070/7269] eta: 0:30:27 lr: 0.000007 loss: 0.5435 (0.5259) time: 0.4212 data: 0.0002 max mem: 52150 +Epoch: [17] [3080/7269] eta: 0:30:21 lr: 0.000007 loss: 0.5442 (0.5259) time: 0.4208 data: 0.0002 max mem: 52150 +Epoch: [17] [3090/7269] eta: 0:30:18 lr: 0.000007 loss: 0.5387 (0.5259) time: 0.4343 data: 0.0002 max mem: 52150 +Epoch: [17] [3100/7269] eta: 0:30:13 lr: 0.000007 loss: 0.5368 (0.5259) time: 0.4343 data: 0.0002 max mem: 52150 +Epoch: [17] [3110/7269] eta: 0:30:09 lr: 0.000007 loss: 0.5309 (0.5259) time: 0.4366 data: 0.0002 max mem: 52150 +Epoch: [17] [3120/7269] eta: 0:30:04 lr: 0.000007 loss: 0.5138 (0.5259) time: 0.4369 data: 0.0002 max mem: 52150 +Epoch: [17] [3130/7269] eta: 0:29:59 lr: 0.000007 loss: 0.5139 (0.5258) time: 0.3727 data: 0.0002 max mem: 52150 +Epoch: [17] [3140/7269] eta: 0:29:54 lr: 0.000007 loss: 0.5335 (0.5259) time: 0.3727 data: 0.0002 max mem: 52150 +Epoch: [17] [3150/7269] eta: 0:29:49 lr: 0.000007 loss: 0.5332 (0.5257) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [17] [3160/7269] eta: 0:29:45 lr: 0.000007 loss: 0.4529 (0.5256) time: 0.4196 data: 0.0002 max mem: 52150 +Epoch: [17] [3170/7269] eta: 0:29:42 lr: 0.000007 loss: 0.5250 (0.5257) time: 0.5215 data: 0.0002 max mem: 52150 +Epoch: [17] [3180/7269] eta: 0:29:38 lr: 0.000007 loss: 0.5698 (0.5257) time: 0.5205 data: 0.0002 max mem: 52150 +Epoch: [17] [3190/7269] eta: 0:29:36 lr: 0.000007 loss: 0.4900 (0.5256) time: 0.5177 data: 0.0002 max mem: 52150 +Epoch: [17] [3200/7269] eta: 0:29:30 lr: 0.000007 loss: 0.4900 (0.5255) time: 0.4714 data: 0.0001 max mem: 52150 +Epoch: [17] [3210/7269] eta: 0:29:25 lr: 0.000007 loss: 0.5531 (0.5256) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [17] [3220/7269] eta: 0:29:20 lr: 0.000007 loss: 0.5558 (0.5256) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [17] [3230/7269] eta: 0:29:19 lr: 0.000007 loss: 0.5284 (0.5256) time: 0.5314 data: 0.0001 max mem: 52150 +Epoch: [17] [3240/7269] eta: 0:29:14 lr: 0.000007 loss: 0.5157 (0.5255) time: 0.5312 data: 0.0001 max mem: 52150 +Epoch: [17] [3250/7269] eta: 0:29:10 lr: 0.000007 loss: 0.5121 (0.5254) time: 0.4206 data: 0.0002 max mem: 52150 +Epoch: [17] [3260/7269] eta: 0:29:06 lr: 0.000007 loss: 0.4988 (0.5254) time: 0.4799 data: 0.0002 max mem: 52150 +Epoch: [17] [3270/7269] eta: 0:29:01 lr: 0.000007 loss: 0.5068 (0.5254) time: 0.4313 data: 0.0002 max mem: 52150 +Epoch: [17] [3280/7269] eta: 0:28:56 lr: 0.000007 loss: 0.4949 (0.5253) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [17] [3290/7269] eta: 0:28:52 lr: 0.000007 loss: 0.4843 (0.5253) time: 0.4338 data: 0.0002 max mem: 52150 +Epoch: [17] [3300/7269] eta: 0:28:50 lr: 0.000007 loss: 0.5161 (0.5253) time: 0.5400 data: 0.0002 max mem: 52150 +Epoch: [17] [3310/7269] eta: 0:28:46 lr: 0.000007 loss: 0.5317 (0.5253) time: 0.5412 data: 0.0001 max mem: 52150 +Epoch: [17] [3320/7269] eta: 0:28:41 lr: 0.000007 loss: 0.5285 (0.5253) time: 0.4354 data: 0.0002 max mem: 52150 +Epoch: [17] [3330/7269] eta: 0:28:36 lr: 0.000007 loss: 0.5285 (0.5252) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [17] [3340/7269] eta: 0:28:32 lr: 0.000007 loss: 0.5367 (0.5252) time: 0.4329 data: 0.0002 max mem: 52150 +Epoch: [17] [3350/7269] eta: 0:28:27 lr: 0.000007 loss: 0.5367 (0.5252) time: 0.4331 data: 0.0002 max mem: 52150 +Epoch: [17] [3360/7269] eta: 0:28:22 lr: 0.000007 loss: 0.5446 (0.5252) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [17] [3370/7269] eta: 0:28:17 lr: 0.000007 loss: 0.5446 (0.5252) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [17] [3380/7269] eta: 0:28:13 lr: 0.000007 loss: 0.5544 (0.5253) time: 0.4315 data: 0.0001 max mem: 52150 +Epoch: [17] [3390/7269] eta: 0:28:09 lr: 0.000007 loss: 0.5657 (0.5254) time: 0.4926 data: 0.0002 max mem: 52150 +Epoch: [17] [3400/7269] eta: 0:28:04 lr: 0.000007 loss: 0.5218 (0.5253) time: 0.4317 data: 0.0002 max mem: 52150 +Epoch: [17] [3410/7269] eta: 0:27:59 lr: 0.000007 loss: 0.5173 (0.5253) time: 0.3710 data: 0.0002 max mem: 52150 +Epoch: [17] [3420/7269] eta: 0:27:54 lr: 0.000007 loss: 0.5173 (0.5252) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [17] [3430/7269] eta: 0:27:51 lr: 0.000007 loss: 0.5350 (0.5253) time: 0.4329 data: 0.0002 max mem: 52150 +Epoch: [17] [3440/7269] eta: 0:27:45 lr: 0.000007 loss: 0.5423 (0.5253) time: 0.4324 data: 0.0002 max mem: 52150 +Epoch: [17] [3450/7269] eta: 0:27:42 lr: 0.000007 loss: 0.5399 (0.5252) time: 0.4313 data: 0.0002 max mem: 52150 +Epoch: [17] [3460/7269] eta: 0:27:37 lr: 0.000007 loss: 0.5053 (0.5252) time: 0.4314 data: 0.0002 max mem: 52150 +Epoch: [17] [3470/7269] eta: 0:27:32 lr: 0.000007 loss: 0.5362 (0.5252) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [17] [3480/7269] eta: 0:27:27 lr: 0.000007 loss: 0.5485 (0.5253) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [17] [3490/7269] eta: 0:27:24 lr: 0.000007 loss: 0.5186 (0.5252) time: 0.4833 data: 0.0001 max mem: 52150 +Epoch: [17] [3500/7269] eta: 0:27:19 lr: 0.000007 loss: 0.4975 (0.5252) time: 0.4830 data: 0.0001 max mem: 52150 +Epoch: [17] [3510/7269] eta: 0:27:14 lr: 0.000007 loss: 0.5334 (0.5252) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [17] [3520/7269] eta: 0:27:12 lr: 0.000007 loss: 0.5376 (0.5252) time: 0.5266 data: 0.0001 max mem: 52150 +Epoch: [17] [3530/7269] eta: 0:27:07 lr: 0.000007 loss: 0.5256 (0.5251) time: 0.5262 data: 0.0001 max mem: 52150 +Epoch: [17] [3540/7269] eta: 0:27:02 lr: 0.000007 loss: 0.5403 (0.5252) time: 0.3714 data: 0.0001 max mem: 52150 +Epoch: [17] [3550/7269] eta: 0:26:57 lr: 0.000007 loss: 0.5654 (0.5253) time: 0.3721 data: 0.0001 max mem: 52150 +Epoch: [17] [3560/7269] eta: 0:26:52 lr: 0.000007 loss: 0.5715 (0.5254) time: 0.3723 data: 0.0001 max mem: 52150 +Epoch: [17] [3570/7269] eta: 0:26:47 lr: 0.000007 loss: 0.5653 (0.5255) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [17] [3580/7269] eta: 0:26:43 lr: 0.000007 loss: 0.5361 (0.5255) time: 0.4339 data: 0.0001 max mem: 52150 +Epoch: [17] [3590/7269] eta: 0:26:39 lr: 0.000007 loss: 0.5254 (0.5255) time: 0.4836 data: 0.0002 max mem: 52150 +Epoch: [17] [3600/7269] eta: 0:26:37 lr: 0.000007 loss: 0.5334 (0.5255) time: 0.5424 data: 0.0002 max mem: 52150 +Epoch: [17] [3610/7269] eta: 0:26:33 lr: 0.000007 loss: 0.5331 (0.5255) time: 0.5388 data: 0.0001 max mem: 52150 +Epoch: [17] [3620/7269] eta: 0:26:28 lr: 0.000007 loss: 0.5261 (0.5255) time: 0.4176 data: 0.0001 max mem: 52150 +Epoch: [17] [3630/7269] eta: 0:26:23 lr: 0.000007 loss: 0.5123 (0.5255) time: 0.3714 data: 0.0001 max mem: 52150 +Epoch: [17] [3640/7269] eta: 0:26:18 lr: 0.000007 loss: 0.5242 (0.5255) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [17] [3650/7269] eta: 0:26:13 lr: 0.000007 loss: 0.5365 (0.5256) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [17] [3660/7269] eta: 0:26:09 lr: 0.000007 loss: 0.5492 (0.5256) time: 0.4194 data: 0.0002 max mem: 52150 +Epoch: [17] [3670/7269] eta: 0:26:04 lr: 0.000007 loss: 0.5047 (0.5256) time: 0.4199 data: 0.0002 max mem: 52150 +Epoch: [17] [3680/7269] eta: 0:26:00 lr: 0.000007 loss: 0.5275 (0.5257) time: 0.4204 data: 0.0002 max mem: 52150 +Epoch: [17] [3690/7269] eta: 0:25:56 lr: 0.000007 loss: 0.5698 (0.5257) time: 0.4679 data: 0.0001 max mem: 52150 +Epoch: [17] [3700/7269] eta: 0:25:51 lr: 0.000007 loss: 0.5741 (0.5259) time: 0.4199 data: 0.0001 max mem: 52150 +Epoch: [17] [3710/7269] eta: 0:25:46 lr: 0.000007 loss: 0.5601 (0.5259) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [17] [3720/7269] eta: 0:25:42 lr: 0.000007 loss: 0.5497 (0.5260) time: 0.4349 data: 0.0002 max mem: 52150 +Epoch: [17] [3730/7269] eta: 0:25:38 lr: 0.000007 loss: 0.5497 (0.5261) time: 0.4984 data: 0.0002 max mem: 52150 +Epoch: [17] [3740/7269] eta: 0:25:34 lr: 0.000007 loss: 0.5538 (0.5261) time: 0.4844 data: 0.0002 max mem: 52150 +Epoch: [17] [3750/7269] eta: 0:25:31 lr: 0.000007 loss: 0.5631 (0.5262) time: 0.4813 data: 0.0002 max mem: 52150 +Epoch: [17] [3760/7269] eta: 0:25:27 lr: 0.000007 loss: 0.5631 (0.5263) time: 0.4927 data: 0.0002 max mem: 52150 +Epoch: [17] [3770/7269] eta: 0:25:22 lr: 0.000007 loss: 0.5077 (0.5262) time: 0.4324 data: 0.0002 max mem: 52150 +Epoch: [17] [3780/7269] eta: 0:25:17 lr: 0.000007 loss: 0.5203 (0.5263) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [17] [3790/7269] eta: 0:25:14 lr: 0.000007 loss: 0.5340 (0.5263) time: 0.4814 data: 0.0002 max mem: 52150 +Epoch: [17] [3800/7269] eta: 0:25:09 lr: 0.000007 loss: 0.5242 (0.5263) time: 0.4807 data: 0.0002 max mem: 52150 +Epoch: [17] [3810/7269] eta: 0:25:05 lr: 0.000007 loss: 0.5177 (0.5263) time: 0.4169 data: 0.0002 max mem: 52150 +Epoch: [17] [3820/7269] eta: 0:25:00 lr: 0.000007 loss: 0.5146 (0.5263) time: 0.4179 data: 0.0002 max mem: 52150 +Epoch: [17] [3830/7269] eta: 0:24:55 lr: 0.000007 loss: 0.5172 (0.5263) time: 0.3730 data: 0.0002 max mem: 52150 +Epoch: [17] [3840/7269] eta: 0:24:52 lr: 0.000007 loss: 0.5270 (0.5263) time: 0.4935 data: 0.0002 max mem: 52150 +Epoch: [17] [3850/7269] eta: 0:24:48 lr: 0.000007 loss: 0.5389 (0.5263) time: 0.4931 data: 0.0002 max mem: 52150 +Epoch: [17] [3860/7269] eta: 0:24:43 lr: 0.000007 loss: 0.5564 (0.5263) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [17] [3870/7269] eta: 0:24:39 lr: 0.000007 loss: 0.5441 (0.5263) time: 0.4354 data: 0.0002 max mem: 52150 +Epoch: [17] [3880/7269] eta: 0:24:34 lr: 0.000007 loss: 0.5418 (0.5263) time: 0.4359 data: 0.0002 max mem: 52150 +Epoch: [17] [3890/7269] eta: 0:24:30 lr: 0.000007 loss: 0.5303 (0.5263) time: 0.4211 data: 0.0002 max mem: 52150 +Epoch: [17] [3900/7269] eta: 0:24:26 lr: 0.000007 loss: 0.5303 (0.5262) time: 0.4808 data: 0.0001 max mem: 52150 +Epoch: [17] [3910/7269] eta: 0:24:21 lr: 0.000007 loss: 0.5345 (0.5262) time: 0.4316 data: 0.0001 max mem: 52150 +Epoch: [17] [3920/7269] eta: 0:24:17 lr: 0.000007 loss: 0.5040 (0.5261) time: 0.4217 data: 0.0001 max mem: 52150 +Epoch: [17] [3930/7269] eta: 0:24:13 lr: 0.000007 loss: 0.5009 (0.5260) time: 0.4690 data: 0.0001 max mem: 52150 +Epoch: [17] [3940/7269] eta: 0:24:09 lr: 0.000007 loss: 0.5375 (0.5261) time: 0.4666 data: 0.0002 max mem: 52150 +Epoch: [17] [3950/7269] eta: 0:24:04 lr: 0.000007 loss: 0.5492 (0.5261) time: 0.4191 data: 0.0002 max mem: 52150 +Epoch: [17] [3960/7269] eta: 0:24:01 lr: 0.000007 loss: 0.5171 (0.5261) time: 0.4938 data: 0.0002 max mem: 52150 +Epoch: [17] [3970/7269] eta: 0:23:56 lr: 0.000007 loss: 0.5152 (0.5261) time: 0.4940 data: 0.0002 max mem: 52150 +Epoch: [17] [3980/7269] eta: 0:23:53 lr: 0.000007 loss: 0.5226 (0.5262) time: 0.4794 data: 0.0001 max mem: 52150 +Epoch: [17] [3990/7269] eta: 0:23:48 lr: 0.000007 loss: 0.5298 (0.5261) time: 0.4798 data: 0.0001 max mem: 52150 +Epoch: [17] [4000/7269] eta: 0:23:43 lr: 0.000007 loss: 0.4909 (0.5260) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [17] [4010/7269] eta: 0:23:39 lr: 0.000007 loss: 0.4866 (0.5259) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [17] [4020/7269] eta: 0:23:34 lr: 0.000007 loss: 0.5044 (0.5260) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [17] [4030/7269] eta: 0:23:29 lr: 0.000007 loss: 0.5327 (0.5259) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [17] [4040/7269] eta: 0:23:24 lr: 0.000007 loss: 0.5225 (0.5259) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [17] [4050/7269] eta: 0:23:20 lr: 0.000007 loss: 0.5269 (0.5259) time: 0.4179 data: 0.0002 max mem: 52150 +Epoch: [17] [4060/7269] eta: 0:23:17 lr: 0.000007 loss: 0.5115 (0.5259) time: 0.5279 data: 0.0002 max mem: 52150 +Epoch: [17] [4070/7269] eta: 0:23:13 lr: 0.000007 loss: 0.5271 (0.5259) time: 0.5279 data: 0.0001 max mem: 52150 +Epoch: [17] [4080/7269] eta: 0:23:08 lr: 0.000007 loss: 0.5342 (0.5259) time: 0.4177 data: 0.0002 max mem: 52150 +Epoch: [17] [4090/7269] eta: 0:23:03 lr: 0.000007 loss: 0.5102 (0.5257) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [17] [4100/7269] eta: 0:22:58 lr: 0.000007 loss: 0.5362 (0.5258) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [17] [4110/7269] eta: 0:22:53 lr: 0.000007 loss: 0.5498 (0.5259) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [17] [4120/7269] eta: 0:22:48 lr: 0.000007 loss: 0.5506 (0.5260) time: 0.3728 data: 0.0002 max mem: 52150 +Epoch: [17] [4130/7269] eta: 0:22:44 lr: 0.000007 loss: 0.5189 (0.5260) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [17] [4140/7269] eta: 0:22:39 lr: 0.000007 loss: 0.5136 (0.5259) time: 0.4194 data: 0.0001 max mem: 52150 +Epoch: [17] [4150/7269] eta: 0:22:35 lr: 0.000007 loss: 0.5253 (0.5260) time: 0.4193 data: 0.0001 max mem: 52150 +Epoch: [17] [4160/7269] eta: 0:22:31 lr: 0.000007 loss: 0.5527 (0.5260) time: 0.4349 data: 0.0002 max mem: 52150 +Epoch: [17] [4170/7269] eta: 0:22:26 lr: 0.000007 loss: 0.5381 (0.5260) time: 0.4350 data: 0.0002 max mem: 52150 +Epoch: [17] [4180/7269] eta: 0:22:21 lr: 0.000007 loss: 0.4826 (0.5259) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [17] [4190/7269] eta: 0:22:17 lr: 0.000007 loss: 0.5367 (0.5259) time: 0.4367 data: 0.0001 max mem: 52150 +Epoch: [17] [4200/7269] eta: 0:22:13 lr: 0.000007 loss: 0.5367 (0.5259) time: 0.4863 data: 0.0001 max mem: 52150 +Epoch: [17] [4210/7269] eta: 0:22:08 lr: 0.000007 loss: 0.4755 (0.5259) time: 0.4213 data: 0.0001 max mem: 52150 +Epoch: [17] [4220/7269] eta: 0:22:05 lr: 0.000007 loss: 0.5117 (0.5259) time: 0.4318 data: 0.0001 max mem: 52150 +Epoch: [17] [4230/7269] eta: 0:22:00 lr: 0.000007 loss: 0.5187 (0.5259) time: 0.4315 data: 0.0001 max mem: 52150 +Epoch: [17] [4240/7269] eta: 0:21:55 lr: 0.000007 loss: 0.5358 (0.5259) time: 0.3713 data: 0.0001 max mem: 52150 +Epoch: [17] [4250/7269] eta: 0:21:51 lr: 0.000007 loss: 0.5362 (0.5260) time: 0.4169 data: 0.0001 max mem: 52150 +Epoch: [17] [4260/7269] eta: 0:21:46 lr: 0.000007 loss: 0.5336 (0.5259) time: 0.4169 data: 0.0001 max mem: 52150 +Epoch: [17] [4270/7269] eta: 0:21:41 lr: 0.000007 loss: 0.5510 (0.5260) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [17] [4280/7269] eta: 0:21:36 lr: 0.000007 loss: 0.5381 (0.5260) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [17] [4290/7269] eta: 0:21:32 lr: 0.000007 loss: 0.5299 (0.5260) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [17] [4300/7269] eta: 0:21:28 lr: 0.000007 loss: 0.5299 (0.5259) time: 0.4219 data: 0.0001 max mem: 52150 +Epoch: [17] [4310/7269] eta: 0:21:24 lr: 0.000007 loss: 0.4928 (0.5258) time: 0.4816 data: 0.0002 max mem: 52150 +Epoch: [17] [4320/7269] eta: 0:21:19 lr: 0.000007 loss: 0.4928 (0.5258) time: 0.4317 data: 0.0002 max mem: 52150 +Epoch: [17] [4330/7269] eta: 0:21:16 lr: 0.000007 loss: 0.5362 (0.5258) time: 0.4988 data: 0.0002 max mem: 52150 +Epoch: [17] [4340/7269] eta: 0:21:12 lr: 0.000007 loss: 0.5595 (0.5259) time: 0.5459 data: 0.0002 max mem: 52150 +Epoch: [17] [4350/7269] eta: 0:21:08 lr: 0.000007 loss: 0.5393 (0.5259) time: 0.5180 data: 0.0002 max mem: 52150 +Epoch: [17] [4360/7269] eta: 0:21:04 lr: 0.000007 loss: 0.5242 (0.5259) time: 0.5308 data: 0.0002 max mem: 52150 +Epoch: [17] [4370/7269] eta: 0:21:00 lr: 0.000007 loss: 0.5374 (0.5259) time: 0.4818 data: 0.0001 max mem: 52150 +Epoch: [17] [4380/7269] eta: 0:20:55 lr: 0.000007 loss: 0.5320 (0.5258) time: 0.4215 data: 0.0001 max mem: 52150 +Epoch: [17] [4390/7269] eta: 0:20:51 lr: 0.000007 loss: 0.5345 (0.5259) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [17] [4400/7269] eta: 0:20:47 lr: 0.000007 loss: 0.5326 (0.5258) time: 0.4355 data: 0.0002 max mem: 52150 +Epoch: [17] [4410/7269] eta: 0:20:43 lr: 0.000007 loss: 0.5200 (0.5258) time: 0.4965 data: 0.0002 max mem: 52150 +Epoch: [17] [4420/7269] eta: 0:20:39 lr: 0.000007 loss: 0.5279 (0.5258) time: 0.4806 data: 0.0002 max mem: 52150 +Epoch: [17] [4430/7269] eta: 0:20:35 lr: 0.000007 loss: 0.5239 (0.5258) time: 0.4813 data: 0.0002 max mem: 52150 +Epoch: [17] [4440/7269] eta: 0:20:31 lr: 0.000007 loss: 0.5591 (0.5259) time: 0.4825 data: 0.0002 max mem: 52150 +Epoch: [17] [4450/7269] eta: 0:20:27 lr: 0.000007 loss: 0.5591 (0.5259) time: 0.5471 data: 0.0001 max mem: 52150 +Epoch: [17] [4460/7269] eta: 0:20:23 lr: 0.000007 loss: 0.5024 (0.5259) time: 0.4990 data: 0.0002 max mem: 52150 +Epoch: [17] [4470/7269] eta: 0:20:19 lr: 0.000007 loss: 0.5417 (0.5259) time: 0.4201 data: 0.0002 max mem: 52150 +Epoch: [17] [4480/7269] eta: 0:20:14 lr: 0.000007 loss: 0.5494 (0.5258) time: 0.4200 data: 0.0001 max mem: 52150 +Epoch: [17] [4490/7269] eta: 0:20:09 lr: 0.000007 loss: 0.5559 (0.5259) time: 0.3721 data: 0.0001 max mem: 52150 +Epoch: [17] [4500/7269] eta: 0:20:04 lr: 0.000007 loss: 0.5244 (0.5258) time: 0.3712 data: 0.0001 max mem: 52150 +Epoch: [17] [4510/7269] eta: 0:20:00 lr: 0.000007 loss: 0.4673 (0.5258) time: 0.3712 data: 0.0002 max mem: 52150 +Epoch: [17] [4520/7269] eta: 0:19:56 lr: 0.000007 loss: 0.5328 (0.5258) time: 0.4949 data: 0.0002 max mem: 52150 +Epoch: [17] [4530/7269] eta: 0:19:52 lr: 0.000007 loss: 0.5466 (0.5259) time: 0.4944 data: 0.0002 max mem: 52150 +Epoch: [17] [4540/7269] eta: 0:19:47 lr: 0.000007 loss: 0.5635 (0.5259) time: 0.3710 data: 0.0002 max mem: 52150 +Epoch: [17] [4550/7269] eta: 0:19:43 lr: 0.000007 loss: 0.5416 (0.5260) time: 0.4171 data: 0.0002 max mem: 52150 +Epoch: [17] [4560/7269] eta: 0:19:38 lr: 0.000007 loss: 0.5333 (0.5260) time: 0.4663 data: 0.0001 max mem: 52150 +Epoch: [17] [4570/7269] eta: 0:19:34 lr: 0.000007 loss: 0.5363 (0.5261) time: 0.4732 data: 0.0002 max mem: 52150 +Epoch: [17] [4580/7269] eta: 0:19:30 lr: 0.000007 loss: 0.5478 (0.5260) time: 0.4240 data: 0.0002 max mem: 52150 +Epoch: [17] [4590/7269] eta: 0:19:25 lr: 0.000007 loss: 0.5178 (0.5261) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [17] [4600/7269] eta: 0:19:20 lr: 0.000007 loss: 0.5101 (0.5259) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [17] [4610/7269] eta: 0:19:16 lr: 0.000007 loss: 0.4913 (0.5259) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [17] [4620/7269] eta: 0:19:11 lr: 0.000007 loss: 0.5509 (0.5260) time: 0.4322 data: 0.0001 max mem: 52150 +Epoch: [17] [4630/7269] eta: 0:19:07 lr: 0.000007 loss: 0.5421 (0.5260) time: 0.4327 data: 0.0002 max mem: 52150 +Epoch: [17] [4640/7269] eta: 0:19:02 lr: 0.000007 loss: 0.5276 (0.5260) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [17] [4650/7269] eta: 0:18:58 lr: 0.000007 loss: 0.5652 (0.5261) time: 0.4371 data: 0.0002 max mem: 52150 +Epoch: [17] [4660/7269] eta: 0:18:53 lr: 0.000007 loss: 0.5382 (0.5260) time: 0.4373 data: 0.0002 max mem: 52150 +Epoch: [17] [4670/7269] eta: 0:18:49 lr: 0.000007 loss: 0.5324 (0.5261) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [17] [4680/7269] eta: 0:18:44 lr: 0.000007 loss: 0.5398 (0.5261) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [17] [4690/7269] eta: 0:18:40 lr: 0.000007 loss: 0.5398 (0.5261) time: 0.4351 data: 0.0001 max mem: 52150 +Epoch: [17] [4700/7269] eta: 0:18:35 lr: 0.000007 loss: 0.5402 (0.5261) time: 0.4347 data: 0.0002 max mem: 52150 +Epoch: [17] [4710/7269] eta: 0:18:31 lr: 0.000007 loss: 0.5402 (0.5262) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [17] [4720/7269] eta: 0:18:26 lr: 0.000007 loss: 0.5065 (0.5261) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [17] [4730/7269] eta: 0:18:21 lr: 0.000007 loss: 0.4978 (0.5260) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [17] [4740/7269] eta: 0:18:17 lr: 0.000007 loss: 0.5169 (0.5260) time: 0.4353 data: 0.0002 max mem: 52150 +Epoch: [17] [4750/7269] eta: 0:18:13 lr: 0.000007 loss: 0.4826 (0.5259) time: 0.4347 data: 0.0002 max mem: 52150 +Epoch: [17] [4760/7269] eta: 0:18:09 lr: 0.000007 loss: 0.5155 (0.5259) time: 0.4322 data: 0.0002 max mem: 52150 +Epoch: [17] [4770/7269] eta: 0:18:04 lr: 0.000007 loss: 0.5364 (0.5259) time: 0.4805 data: 0.0002 max mem: 52150 +Epoch: [17] [4780/7269] eta: 0:18:00 lr: 0.000007 loss: 0.5364 (0.5259) time: 0.4196 data: 0.0002 max mem: 52150 +Epoch: [17] [4790/7269] eta: 0:17:56 lr: 0.000007 loss: 0.5141 (0.5259) time: 0.4355 data: 0.0002 max mem: 52150 +Epoch: [17] [4800/7269] eta: 0:17:51 lr: 0.000007 loss: 0.5301 (0.5259) time: 0.4360 data: 0.0002 max mem: 52150 +Epoch: [17] [4810/7269] eta: 0:17:47 lr: 0.000007 loss: 0.5360 (0.5259) time: 0.4224 data: 0.0002 max mem: 52150 +Epoch: [17] [4820/7269] eta: 0:17:42 lr: 0.000007 loss: 0.5387 (0.5259) time: 0.4227 data: 0.0002 max mem: 52150 +Epoch: [17] [4830/7269] eta: 0:17:38 lr: 0.000007 loss: 0.5226 (0.5259) time: 0.4185 data: 0.0002 max mem: 52150 +Epoch: [17] [4840/7269] eta: 0:17:34 lr: 0.000007 loss: 0.5093 (0.5258) time: 0.4668 data: 0.0002 max mem: 52150 +Epoch: [17] [4850/7269] eta: 0:17:30 lr: 0.000007 loss: 0.5098 (0.5258) time: 0.4720 data: 0.0002 max mem: 52150 +Epoch: [17] [4860/7269] eta: 0:17:25 lr: 0.000007 loss: 0.4885 (0.5257) time: 0.4229 data: 0.0002 max mem: 52150 +Epoch: [17] [4870/7269] eta: 0:17:21 lr: 0.000007 loss: 0.5008 (0.5257) time: 0.4198 data: 0.0002 max mem: 52150 +Epoch: [17] [4880/7269] eta: 0:17:18 lr: 0.000007 loss: 0.5535 (0.5258) time: 0.5422 data: 0.0001 max mem: 52150 +Epoch: [17] [4890/7269] eta: 0:17:13 lr: 0.000007 loss: 0.5244 (0.5258) time: 0.4944 data: 0.0002 max mem: 52150 +Epoch: [17] [4900/7269] eta: 0:17:08 lr: 0.000007 loss: 0.5244 (0.5258) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [17] [4910/7269] eta: 0:17:04 lr: 0.000007 loss: 0.5380 (0.5258) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [17] [4920/7269] eta: 0:16:59 lr: 0.000007 loss: 0.5599 (0.5259) time: 0.4226 data: 0.0002 max mem: 52150 +Epoch: [17] [4930/7269] eta: 0:16:55 lr: 0.000007 loss: 0.5735 (0.5259) time: 0.4831 data: 0.0002 max mem: 52150 +Epoch: [17] [4940/7269] eta: 0:16:52 lr: 0.000007 loss: 0.5348 (0.5259) time: 0.5439 data: 0.0002 max mem: 52150 +Epoch: [17] [4950/7269] eta: 0:16:47 lr: 0.000007 loss: 0.5449 (0.5259) time: 0.4835 data: 0.0002 max mem: 52150 +Epoch: [17] [4960/7269] eta: 0:16:43 lr: 0.000007 loss: 0.5291 (0.5259) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [17] [4970/7269] eta: 0:16:38 lr: 0.000007 loss: 0.5206 (0.5258) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [17] [4980/7269] eta: 0:16:34 lr: 0.000007 loss: 0.5219 (0.5258) time: 0.4948 data: 0.0002 max mem: 52150 +Epoch: [17] [4990/7269] eta: 0:16:30 lr: 0.000007 loss: 0.5332 (0.5259) time: 0.5416 data: 0.0002 max mem: 52150 +Epoch: [17] [5000/7269] eta: 0:16:26 lr: 0.000007 loss: 0.5380 (0.5259) time: 0.4186 data: 0.0002 max mem: 52150 +Epoch: [17] [5010/7269] eta: 0:16:21 lr: 0.000007 loss: 0.5380 (0.5259) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [17] [5020/7269] eta: 0:16:18 lr: 0.000007 loss: 0.4956 (0.5258) time: 0.5601 data: 0.0002 max mem: 52150 +Epoch: [17] [5030/7269] eta: 0:16:14 lr: 0.000007 loss: 0.5224 (0.5259) time: 0.6077 data: 0.0002 max mem: 52150 +Epoch: [17] [5040/7269] eta: 0:16:10 lr: 0.000007 loss: 0.5279 (0.5259) time: 0.4678 data: 0.0002 max mem: 52150 +Epoch: [17] [5050/7269] eta: 0:16:05 lr: 0.000007 loss: 0.5276 (0.5259) time: 0.4691 data: 0.0002 max mem: 52150 +Epoch: [17] [5060/7269] eta: 0:16:01 lr: 0.000007 loss: 0.5255 (0.5259) time: 0.4209 data: 0.0001 max mem: 52150 +Epoch: [17] [5070/7269] eta: 0:15:56 lr: 0.000007 loss: 0.5194 (0.5259) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [17] [5080/7269] eta: 0:15:52 lr: 0.000007 loss: 0.5150 (0.5259) time: 0.3723 data: 0.0001 max mem: 52150 +Epoch: [17] [5090/7269] eta: 0:15:48 lr: 0.000007 loss: 0.5441 (0.5259) time: 0.4689 data: 0.0002 max mem: 52150 +Epoch: [17] [5100/7269] eta: 0:15:43 lr: 0.000007 loss: 0.5350 (0.5259) time: 0.4686 data: 0.0002 max mem: 52150 +Epoch: [17] [5110/7269] eta: 0:15:38 lr: 0.000007 loss: 0.5162 (0.5258) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [17] [5120/7269] eta: 0:15:34 lr: 0.000007 loss: 0.4858 (0.5258) time: 0.4318 data: 0.0002 max mem: 52150 +Epoch: [17] [5130/7269] eta: 0:15:30 lr: 0.000007 loss: 0.4858 (0.5257) time: 0.4314 data: 0.0002 max mem: 52150 +Epoch: [17] [5140/7269] eta: 0:15:26 lr: 0.000007 loss: 0.5454 (0.5257) time: 0.4330 data: 0.0002 max mem: 52150 +Epoch: [17] [5150/7269] eta: 0:15:21 lr: 0.000007 loss: 0.5541 (0.5258) time: 0.4787 data: 0.0002 max mem: 52150 +Epoch: [17] [5160/7269] eta: 0:15:17 lr: 0.000007 loss: 0.5524 (0.5258) time: 0.4178 data: 0.0002 max mem: 52150 +Epoch: [17] [5170/7269] eta: 0:15:12 lr: 0.000007 loss: 0.5353 (0.5258) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [17] [5180/7269] eta: 0:15:08 lr: 0.000007 loss: 0.5076 (0.5257) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [17] [5190/7269] eta: 0:15:03 lr: 0.000007 loss: 0.5185 (0.5258) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [17] [5200/7269] eta: 0:14:58 lr: 0.000007 loss: 0.5450 (0.5258) time: 0.3710 data: 0.0002 max mem: 52150 +Epoch: [17] [5210/7269] eta: 0:14:54 lr: 0.000007 loss: 0.5264 (0.5258) time: 0.4341 data: 0.0001 max mem: 52150 +Epoch: [17] [5220/7269] eta: 0:14:50 lr: 0.000007 loss: 0.5205 (0.5258) time: 0.4807 data: 0.0001 max mem: 52150 +Epoch: [17] [5230/7269] eta: 0:14:46 lr: 0.000007 loss: 0.5457 (0.5258) time: 0.4813 data: 0.0002 max mem: 52150 +Epoch: [17] [5240/7269] eta: 0:14:41 lr: 0.000007 loss: 0.5387 (0.5258) time: 0.4347 data: 0.0002 max mem: 52150 +Epoch: [17] [5250/7269] eta: 0:14:37 lr: 0.000007 loss: 0.4661 (0.5257) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [17] [5260/7269] eta: 0:14:33 lr: 0.000007 loss: 0.4743 (0.5256) time: 0.4355 data: 0.0002 max mem: 52150 +Epoch: [17] [5270/7269] eta: 0:14:28 lr: 0.000007 loss: 0.5317 (0.5256) time: 0.4835 data: 0.0002 max mem: 52150 +Epoch: [17] [5280/7269] eta: 0:14:24 lr: 0.000007 loss: 0.4944 (0.5256) time: 0.4846 data: 0.0002 max mem: 52150 +Epoch: [17] [5290/7269] eta: 0:14:20 lr: 0.000007 loss: 0.4882 (0.5255) time: 0.4839 data: 0.0002 max mem: 52150 +Epoch: [17] [5300/7269] eta: 0:14:16 lr: 0.000007 loss: 0.5230 (0.5255) time: 0.4683 data: 0.0002 max mem: 52150 +Epoch: [17] [5310/7269] eta: 0:14:11 lr: 0.000007 loss: 0.5367 (0.5255) time: 0.4207 data: 0.0001 max mem: 52150 +Epoch: [17] [5320/7269] eta: 0:14:07 lr: 0.000007 loss: 0.5452 (0.5255) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [17] [5330/7269] eta: 0:14:02 lr: 0.000007 loss: 0.5366 (0.5255) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [17] [5340/7269] eta: 0:13:58 lr: 0.000007 loss: 0.5226 (0.5255) time: 0.4369 data: 0.0002 max mem: 52150 +Epoch: [17] [5350/7269] eta: 0:13:54 lr: 0.000007 loss: 0.5806 (0.5256) time: 0.4979 data: 0.0002 max mem: 52150 +Epoch: [17] [5360/7269] eta: 0:13:49 lr: 0.000007 loss: 0.5807 (0.5257) time: 0.4329 data: 0.0002 max mem: 52150 +Epoch: [17] [5370/7269] eta: 0:13:45 lr: 0.000007 loss: 0.5193 (0.5256) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [17] [5380/7269] eta: 0:13:41 lr: 0.000007 loss: 0.5127 (0.5256) time: 0.4347 data: 0.0002 max mem: 52150 +Epoch: [17] [5390/7269] eta: 0:13:36 lr: 0.000007 loss: 0.5335 (0.5256) time: 0.4825 data: 0.0002 max mem: 52150 +Epoch: [17] [5400/7269] eta: 0:13:32 lr: 0.000007 loss: 0.5381 (0.5256) time: 0.4202 data: 0.0002 max mem: 52150 +Epoch: [17] [5410/7269] eta: 0:13:27 lr: 0.000007 loss: 0.5417 (0.5256) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [17] [5420/7269] eta: 0:13:23 lr: 0.000007 loss: 0.5460 (0.5257) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [17] [5430/7269] eta: 0:13:18 lr: 0.000007 loss: 0.5516 (0.5257) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [17] [5440/7269] eta: 0:13:14 lr: 0.000007 loss: 0.5592 (0.5257) time: 0.4773 data: 0.0002 max mem: 52150 +Epoch: [17] [5450/7269] eta: 0:13:10 lr: 0.000007 loss: 0.5200 (0.5257) time: 0.5401 data: 0.0002 max mem: 52150 +Epoch: [17] [5460/7269] eta: 0:13:06 lr: 0.000007 loss: 0.5141 (0.5256) time: 0.4955 data: 0.0002 max mem: 52150 +Epoch: [17] [5470/7269] eta: 0:13:02 lr: 0.000007 loss: 0.5137 (0.5257) time: 0.4771 data: 0.0002 max mem: 52150 +Epoch: [17] [5480/7269] eta: 0:12:58 lr: 0.000007 loss: 0.5625 (0.5257) time: 0.4815 data: 0.0002 max mem: 52150 +Epoch: [17] [5490/7269] eta: 0:12:53 lr: 0.000007 loss: 0.5645 (0.5257) time: 0.4373 data: 0.0002 max mem: 52150 +Epoch: [17] [5500/7269] eta: 0:12:49 lr: 0.000007 loss: 0.5432 (0.5258) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [17] [5510/7269] eta: 0:12:44 lr: 0.000007 loss: 0.5432 (0.5258) time: 0.4360 data: 0.0002 max mem: 52150 +Epoch: [17] [5520/7269] eta: 0:12:40 lr: 0.000007 loss: 0.5460 (0.5258) time: 0.4357 data: 0.0002 max mem: 52150 +Epoch: [17] [5530/7269] eta: 0:12:35 lr: 0.000007 loss: 0.5436 (0.5259) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [17] [5540/7269] eta: 0:12:31 lr: 0.000007 loss: 0.5436 (0.5259) time: 0.4312 data: 0.0002 max mem: 52150 +Epoch: [17] [5550/7269] eta: 0:12:27 lr: 0.000007 loss: 0.5398 (0.5259) time: 0.4309 data: 0.0002 max mem: 52150 +Epoch: [17] [5560/7269] eta: 0:12:22 lr: 0.000007 loss: 0.5398 (0.5259) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [17] [5570/7269] eta: 0:12:18 lr: 0.000007 loss: 0.5317 (0.5259) time: 0.4305 data: 0.0002 max mem: 52150 +Epoch: [17] [5580/7269] eta: 0:12:13 lr: 0.000007 loss: 0.5067 (0.5259) time: 0.4303 data: 0.0002 max mem: 52150 +Epoch: [17] [5590/7269] eta: 0:12:09 lr: 0.000007 loss: 0.5067 (0.5259) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [17] [5600/7269] eta: 0:12:04 lr: 0.000007 loss: 0.5161 (0.5259) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [17] [5610/7269] eta: 0:12:00 lr: 0.000007 loss: 0.5161 (0.5258) time: 0.4178 data: 0.0001 max mem: 52150 +Epoch: [17] [5620/7269] eta: 0:11:56 lr: 0.000007 loss: 0.5027 (0.5258) time: 0.4179 data: 0.0002 max mem: 52150 +Epoch: [17] [5630/7269] eta: 0:11:51 lr: 0.000007 loss: 0.5021 (0.5258) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [17] [5640/7269] eta: 0:11:47 lr: 0.000007 loss: 0.5146 (0.5258) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [17] [5650/7269] eta: 0:11:42 lr: 0.000007 loss: 0.4811 (0.5257) time: 0.4328 data: 0.0002 max mem: 52150 +Epoch: [17] [5660/7269] eta: 0:11:38 lr: 0.000007 loss: 0.4830 (0.5257) time: 0.4326 data: 0.0002 max mem: 52150 +Epoch: [17] [5670/7269] eta: 0:11:33 lr: 0.000007 loss: 0.5165 (0.5257) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [17] [5680/7269] eta: 0:11:29 lr: 0.000007 loss: 0.5526 (0.5257) time: 0.4798 data: 0.0002 max mem: 52150 +Epoch: [17] [5690/7269] eta: 0:11:25 lr: 0.000007 loss: 0.5506 (0.5257) time: 0.5295 data: 0.0002 max mem: 52150 +Epoch: [17] [5700/7269] eta: 0:11:21 lr: 0.000007 loss: 0.5292 (0.5257) time: 0.4712 data: 0.0002 max mem: 52150 +Epoch: [17] [5710/7269] eta: 0:11:16 lr: 0.000007 loss: 0.5356 (0.5257) time: 0.4212 data: 0.0002 max mem: 52150 +Epoch: [17] [5720/7269] eta: 0:11:12 lr: 0.000007 loss: 0.5316 (0.5257) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [17] [5730/7269] eta: 0:11:08 lr: 0.000007 loss: 0.4936 (0.5256) time: 0.4351 data: 0.0002 max mem: 52150 +Epoch: [17] [5740/7269] eta: 0:11:03 lr: 0.000007 loss: 0.5185 (0.5256) time: 0.4846 data: 0.0002 max mem: 52150 +Epoch: [17] [5750/7269] eta: 0:10:59 lr: 0.000007 loss: 0.5410 (0.5256) time: 0.4849 data: 0.0002 max mem: 52150 +Epoch: [17] [5760/7269] eta: 0:10:55 lr: 0.000007 loss: 0.5567 (0.5257) time: 0.4849 data: 0.0001 max mem: 52150 +Epoch: [17] [5770/7269] eta: 0:10:51 lr: 0.000007 loss: 0.5663 (0.5257) time: 0.4841 data: 0.0001 max mem: 52150 +Epoch: [17] [5780/7269] eta: 0:10:48 lr: 0.000007 loss: 0.5472 (0.5257) time: 0.6557 data: 0.0002 max mem: 52150 +Epoch: [17] [5790/7269] eta: 0:10:43 lr: 0.000007 loss: 0.5293 (0.5257) time: 0.5932 data: 0.0002 max mem: 52150 +Epoch: [17] [5800/7269] eta: 0:10:38 lr: 0.000007 loss: 0.5347 (0.5258) time: 0.3728 data: 0.0002 max mem: 52150 +Epoch: [17] [5810/7269] eta: 0:10:34 lr: 0.000007 loss: 0.5547 (0.5257) time: 0.3728 data: 0.0002 max mem: 52150 +Epoch: [17] [5820/7269] eta: 0:10:30 lr: 0.000007 loss: 0.4665 (0.5257) time: 0.4327 data: 0.0002 max mem: 52150 +Epoch: [17] [5830/7269] eta: 0:10:25 lr: 0.000007 loss: 0.4971 (0.5257) time: 0.4328 data: 0.0002 max mem: 52150 +Epoch: [17] [5840/7269] eta: 0:10:21 lr: 0.000007 loss: 0.5360 (0.5257) time: 0.3727 data: 0.0001 max mem: 52150 +Epoch: [17] [5850/7269] eta: 0:10:16 lr: 0.000007 loss: 0.5552 (0.5257) time: 0.3723 data: 0.0001 max mem: 52150 +Epoch: [17] [5860/7269] eta: 0:10:12 lr: 0.000007 loss: 0.5422 (0.5257) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [17] [5870/7269] eta: 0:10:08 lr: 0.000007 loss: 0.5097 (0.5256) time: 0.4320 data: 0.0002 max mem: 52150 +Epoch: [17] [5880/7269] eta: 0:10:03 lr: 0.000007 loss: 0.5100 (0.5257) time: 0.4320 data: 0.0002 max mem: 52150 +Epoch: [17] [5890/7269] eta: 0:09:59 lr: 0.000007 loss: 0.5439 (0.5257) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [17] [5900/7269] eta: 0:09:54 lr: 0.000007 loss: 0.5254 (0.5257) time: 0.4316 data: 0.0002 max mem: 52150 +Epoch: [17] [5910/7269] eta: 0:09:50 lr: 0.000007 loss: 0.4869 (0.5256) time: 0.5391 data: 0.0001 max mem: 52150 +Epoch: [17] [5920/7269] eta: 0:09:46 lr: 0.000007 loss: 0.4958 (0.5256) time: 0.4788 data: 0.0002 max mem: 52150 +Epoch: [17] [5930/7269] eta: 0:09:41 lr: 0.000007 loss: 0.5219 (0.5256) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [17] [5940/7269] eta: 0:09:38 lr: 0.000007 loss: 0.5269 (0.5256) time: 0.5314 data: 0.0002 max mem: 52150 +Epoch: [17] [5950/7269] eta: 0:09:33 lr: 0.000007 loss: 0.5416 (0.5257) time: 0.5318 data: 0.0002 max mem: 52150 +Epoch: [17] [5960/7269] eta: 0:09:29 lr: 0.000007 loss: 0.5416 (0.5257) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [17] [5970/7269] eta: 0:09:24 lr: 0.000007 loss: 0.5556 (0.5257) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [17] [5980/7269] eta: 0:09:20 lr: 0.000007 loss: 0.5297 (0.5257) time: 0.4312 data: 0.0001 max mem: 52150 +Epoch: [17] [5990/7269] eta: 0:09:16 lr: 0.000007 loss: 0.5297 (0.5257) time: 0.4778 data: 0.0001 max mem: 52150 +Epoch: [17] [6000/7269] eta: 0:09:11 lr: 0.000007 loss: 0.5675 (0.5258) time: 0.4186 data: 0.0002 max mem: 52150 +Epoch: [17] [6010/7269] eta: 0:09:07 lr: 0.000007 loss: 0.5271 (0.5258) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [17] [6020/7269] eta: 0:09:02 lr: 0.000007 loss: 0.4907 (0.5258) time: 0.4320 data: 0.0002 max mem: 52150 +Epoch: [17] [6030/7269] eta: 0:08:58 lr: 0.000007 loss: 0.5485 (0.5259) time: 0.4326 data: 0.0002 max mem: 52150 +Epoch: [17] [6040/7269] eta: 0:08:54 lr: 0.000007 loss: 0.5572 (0.5259) time: 0.4365 data: 0.0002 max mem: 52150 +Epoch: [17] [6050/7269] eta: 0:08:49 lr: 0.000007 loss: 0.5587 (0.5259) time: 0.4361 data: 0.0001 max mem: 52150 +Epoch: [17] [6060/7269] eta: 0:08:45 lr: 0.000007 loss: 0.5561 (0.5260) time: 0.3735 data: 0.0001 max mem: 52150 +Epoch: [17] [6070/7269] eta: 0:08:40 lr: 0.000007 loss: 0.5432 (0.5260) time: 0.3737 data: 0.0001 max mem: 52150 +Epoch: [17] [6080/7269] eta: 0:08:36 lr: 0.000007 loss: 0.5552 (0.5260) time: 0.4809 data: 0.0002 max mem: 52150 +Epoch: [17] [6090/7269] eta: 0:08:32 lr: 0.000007 loss: 0.5172 (0.5260) time: 0.4802 data: 0.0002 max mem: 52150 +Epoch: [17] [6100/7269] eta: 0:08:27 lr: 0.000007 loss: 0.4988 (0.5260) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [17] [6110/7269] eta: 0:08:23 lr: 0.000007 loss: 0.5300 (0.5260) time: 0.4179 data: 0.0001 max mem: 52150 +Epoch: [17] [6120/7269] eta: 0:08:19 lr: 0.000007 loss: 0.5502 (0.5261) time: 0.4174 data: 0.0002 max mem: 52150 +Epoch: [17] [6130/7269] eta: 0:08:14 lr: 0.000007 loss: 0.5502 (0.5261) time: 0.4323 data: 0.0002 max mem: 52150 +Epoch: [17] [6140/7269] eta: 0:08:10 lr: 0.000007 loss: 0.5276 (0.5261) time: 0.4328 data: 0.0001 max mem: 52150 +Epoch: [17] [6150/7269] eta: 0:08:05 lr: 0.000007 loss: 0.5092 (0.5260) time: 0.3722 data: 0.0001 max mem: 52150 +Epoch: [17] [6160/7269] eta: 0:08:01 lr: 0.000007 loss: 0.5202 (0.5261) time: 0.4169 data: 0.0002 max mem: 52150 +Epoch: [17] [6170/7269] eta: 0:07:57 lr: 0.000007 loss: 0.5225 (0.5260) time: 0.4168 data: 0.0002 max mem: 52150 +Epoch: [17] [6180/7269] eta: 0:07:52 lr: 0.000007 loss: 0.5191 (0.5261) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [17] [6190/7269] eta: 0:07:48 lr: 0.000007 loss: 0.5504 (0.5261) time: 0.4355 data: 0.0002 max mem: 52150 +Epoch: [17] [6200/7269] eta: 0:07:44 lr: 0.000007 loss: 0.5158 (0.5261) time: 0.4350 data: 0.0002 max mem: 52150 +Epoch: [17] [6210/7269] eta: 0:07:39 lr: 0.000007 loss: 0.5287 (0.5261) time: 0.4179 data: 0.0002 max mem: 52150 +Epoch: [17] [6220/7269] eta: 0:07:35 lr: 0.000007 loss: 0.5287 (0.5261) time: 0.4183 data: 0.0002 max mem: 52150 +Epoch: [17] [6230/7269] eta: 0:07:31 lr: 0.000007 loss: 0.5234 (0.5261) time: 0.5456 data: 0.0002 max mem: 52150 +Epoch: [17] [6240/7269] eta: 0:07:27 lr: 0.000007 loss: 0.5413 (0.5261) time: 0.6084 data: 0.0002 max mem: 52150 +Epoch: [17] [6250/7269] eta: 0:07:23 lr: 0.000007 loss: 0.5563 (0.5261) time: 0.5341 data: 0.0002 max mem: 52150 +Epoch: [17] [6260/7269] eta: 0:07:18 lr: 0.000007 loss: 0.5450 (0.5261) time: 0.4715 data: 0.0002 max mem: 52150 +Epoch: [17] [6270/7269] eta: 0:07:14 lr: 0.000007 loss: 0.5226 (0.5260) time: 0.4206 data: 0.0002 max mem: 52150 +Epoch: [17] [6280/7269] eta: 0:07:09 lr: 0.000007 loss: 0.5455 (0.5261) time: 0.4197 data: 0.0002 max mem: 52150 +Epoch: [17] [6290/7269] eta: 0:07:05 lr: 0.000007 loss: 0.5627 (0.5261) time: 0.4196 data: 0.0002 max mem: 52150 +Epoch: [17] [6300/7269] eta: 0:07:01 lr: 0.000007 loss: 0.5627 (0.5262) time: 0.4198 data: 0.0001 max mem: 52150 +Epoch: [17] [6310/7269] eta: 0:06:56 lr: 0.000007 loss: 0.5465 (0.5262) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [17] [6320/7269] eta: 0:06:52 lr: 0.000007 loss: 0.5041 (0.5261) time: 0.4320 data: 0.0002 max mem: 52150 +Epoch: [17] [6330/7269] eta: 0:06:48 lr: 0.000007 loss: 0.4943 (0.5261) time: 0.4314 data: 0.0002 max mem: 52150 +Epoch: [17] [6340/7269] eta: 0:06:43 lr: 0.000007 loss: 0.5352 (0.5261) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [17] [6350/7269] eta: 0:06:39 lr: 0.000007 loss: 0.5155 (0.5261) time: 0.4319 data: 0.0002 max mem: 52150 +Epoch: [17] [6360/7269] eta: 0:06:35 lr: 0.000007 loss: 0.5155 (0.5261) time: 0.4922 data: 0.0002 max mem: 52150 +Epoch: [17] [6370/7269] eta: 0:06:30 lr: 0.000007 loss: 0.5429 (0.5261) time: 0.4960 data: 0.0002 max mem: 52150 +Epoch: [17] [6380/7269] eta: 0:06:26 lr: 0.000007 loss: 0.5429 (0.5261) time: 0.4354 data: 0.0002 max mem: 52150 +Epoch: [17] [6390/7269] eta: 0:06:22 lr: 0.000007 loss: 0.5527 (0.5262) time: 0.4626 data: 0.0002 max mem: 52150 +Epoch: [17] [6400/7269] eta: 0:06:17 lr: 0.000007 loss: 0.5527 (0.5261) time: 0.4626 data: 0.0002 max mem: 52150 +Epoch: [17] [6410/7269] eta: 0:06:13 lr: 0.000007 loss: 0.4818 (0.5261) time: 0.4828 data: 0.0002 max mem: 52150 +Epoch: [17] [6420/7269] eta: 0:06:09 lr: 0.000007 loss: 0.5314 (0.5261) time: 0.4830 data: 0.0002 max mem: 52150 +Epoch: [17] [6430/7269] eta: 0:06:04 lr: 0.000007 loss: 0.5497 (0.5261) time: 0.4343 data: 0.0001 max mem: 52150 +Epoch: [17] [6440/7269] eta: 0:06:00 lr: 0.000007 loss: 0.5128 (0.5261) time: 0.4340 data: 0.0001 max mem: 52150 +Epoch: [17] [6450/7269] eta: 0:05:56 lr: 0.000007 loss: 0.4943 (0.5261) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [17] [6460/7269] eta: 0:05:51 lr: 0.000007 loss: 0.5253 (0.5261) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [17] [6470/7269] eta: 0:05:47 lr: 0.000007 loss: 0.5333 (0.5261) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [17] [6480/7269] eta: 0:05:42 lr: 0.000007 loss: 0.5254 (0.5260) time: 0.4347 data: 0.0002 max mem: 52150 +Epoch: [17] [6490/7269] eta: 0:05:38 lr: 0.000007 loss: 0.5124 (0.5260) time: 0.4825 data: 0.0002 max mem: 52150 +Epoch: [17] [6500/7269] eta: 0:05:34 lr: 0.000007 loss: 0.5210 (0.5260) time: 0.4710 data: 0.0001 max mem: 52150 +Epoch: [17] [6510/7269] eta: 0:05:30 lr: 0.000007 loss: 0.5204 (0.5260) time: 0.4670 data: 0.0001 max mem: 52150 +Epoch: [17] [6520/7269] eta: 0:05:25 lr: 0.000007 loss: 0.5028 (0.5259) time: 0.4160 data: 0.0001 max mem: 52150 +Epoch: [17] [6530/7269] eta: 0:05:21 lr: 0.000007 loss: 0.5028 (0.5259) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [17] [6540/7269] eta: 0:05:16 lr: 0.000007 loss: 0.5058 (0.5259) time: 0.4221 data: 0.0001 max mem: 52150 +Epoch: [17] [6550/7269] eta: 0:05:12 lr: 0.000007 loss: 0.5058 (0.5259) time: 0.4220 data: 0.0002 max mem: 52150 +Epoch: [17] [6560/7269] eta: 0:05:08 lr: 0.000007 loss: 0.5375 (0.5259) time: 0.4344 data: 0.0002 max mem: 52150 +Epoch: [17] [6570/7269] eta: 0:05:03 lr: 0.000007 loss: 0.5354 (0.5259) time: 0.4831 data: 0.0001 max mem: 52150 +Epoch: [17] [6580/7269] eta: 0:04:59 lr: 0.000007 loss: 0.5237 (0.5259) time: 0.4202 data: 0.0002 max mem: 52150 +Epoch: [17] [6590/7269] eta: 0:04:55 lr: 0.000007 loss: 0.4997 (0.5258) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [17] [6600/7269] eta: 0:04:50 lr: 0.000007 loss: 0.4934 (0.5258) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [17] [6610/7269] eta: 0:04:46 lr: 0.000007 loss: 0.5295 (0.5258) time: 0.4178 data: 0.0002 max mem: 52150 +Epoch: [17] [6620/7269] eta: 0:04:41 lr: 0.000007 loss: 0.5513 (0.5259) time: 0.4175 data: 0.0002 max mem: 52150 +Epoch: [17] [6630/7269] eta: 0:04:37 lr: 0.000007 loss: 0.5304 (0.5258) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [17] [6640/7269] eta: 0:04:33 lr: 0.000007 loss: 0.5142 (0.5258) time: 0.4175 data: 0.0002 max mem: 52150 +Epoch: [17] [6650/7269] eta: 0:04:28 lr: 0.000007 loss: 0.5445 (0.5259) time: 0.4177 data: 0.0002 max mem: 52150 +Epoch: [17] [6660/7269] eta: 0:04:24 lr: 0.000007 loss: 0.5735 (0.5259) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [17] [6670/7269] eta: 0:04:20 lr: 0.000007 loss: 0.5426 (0.5259) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [17] [6680/7269] eta: 0:04:15 lr: 0.000007 loss: 0.5377 (0.5259) time: 0.4320 data: 0.0002 max mem: 52150 +Epoch: [17] [6690/7269] eta: 0:04:11 lr: 0.000007 loss: 0.5416 (0.5259) time: 0.4941 data: 0.0002 max mem: 52150 +Epoch: [17] [6700/7269] eta: 0:04:07 lr: 0.000007 loss: 0.5177 (0.5259) time: 0.4338 data: 0.0002 max mem: 52150 +Epoch: [17] [6710/7269] eta: 0:04:02 lr: 0.000007 loss: 0.5405 (0.5260) time: 0.4343 data: 0.0002 max mem: 52150 +Epoch: [17] [6720/7269] eta: 0:03:58 lr: 0.000007 loss: 0.5478 (0.5260) time: 0.4345 data: 0.0001 max mem: 52150 +Epoch: [17] [6730/7269] eta: 0:03:53 lr: 0.000007 loss: 0.5239 (0.5259) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [17] [6740/7269] eta: 0:03:49 lr: 0.000007 loss: 0.5238 (0.5260) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [17] [6750/7269] eta: 0:03:45 lr: 0.000007 loss: 0.5438 (0.5260) time: 0.4828 data: 0.0001 max mem: 52150 +Epoch: [17] [6760/7269] eta: 0:03:40 lr: 0.000007 loss: 0.5419 (0.5260) time: 0.4828 data: 0.0001 max mem: 52150 +Epoch: [17] [6770/7269] eta: 0:03:36 lr: 0.000007 loss: 0.5036 (0.5259) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [17] [6780/7269] eta: 0:03:32 lr: 0.000007 loss: 0.4682 (0.5259) time: 0.4236 data: 0.0002 max mem: 52150 +Epoch: [17] [6790/7269] eta: 0:03:27 lr: 0.000007 loss: 0.5279 (0.5259) time: 0.4236 data: 0.0002 max mem: 52150 +Epoch: [17] [6800/7269] eta: 0:03:23 lr: 0.000007 loss: 0.5404 (0.5259) time: 0.4332 data: 0.0002 max mem: 52150 +Epoch: [17] [6810/7269] eta: 0:03:19 lr: 0.000007 loss: 0.5278 (0.5259) time: 0.6519 data: 0.0002 max mem: 52150 +Epoch: [17] [6820/7269] eta: 0:03:15 lr: 0.000007 loss: 0.5371 (0.5259) time: 0.6522 data: 0.0002 max mem: 52150 +Epoch: [17] [6830/7269] eta: 0:03:10 lr: 0.000007 loss: 0.5557 (0.5259) time: 0.4338 data: 0.0002 max mem: 52150 +Epoch: [17] [6840/7269] eta: 0:03:06 lr: 0.000007 loss: 0.5437 (0.5259) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [17] [6850/7269] eta: 0:03:02 lr: 0.000007 loss: 0.5205 (0.5259) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [17] [6860/7269] eta: 0:02:57 lr: 0.000007 loss: 0.5068 (0.5260) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [17] [6870/7269] eta: 0:02:53 lr: 0.000007 loss: 0.4926 (0.5259) time: 0.4845 data: 0.0001 max mem: 52150 +Epoch: [17] [6880/7269] eta: 0:02:49 lr: 0.000007 loss: 0.4923 (0.5259) time: 0.5457 data: 0.0001 max mem: 52150 +Epoch: [17] [6890/7269] eta: 0:02:44 lr: 0.000007 loss: 0.5310 (0.5259) time: 0.4335 data: 0.0002 max mem: 52150 +Epoch: [17] [6900/7269] eta: 0:02:40 lr: 0.000007 loss: 0.5089 (0.5258) time: 0.3722 data: 0.0001 max mem: 52150 +Epoch: [17] [6910/7269] eta: 0:02:36 lr: 0.000007 loss: 0.5089 (0.5258) time: 0.4341 data: 0.0001 max mem: 52150 +Epoch: [17] [6920/7269] eta: 0:02:31 lr: 0.000007 loss: 0.5045 (0.5258) time: 0.4337 data: 0.0001 max mem: 52150 +Epoch: [17] [6930/7269] eta: 0:02:27 lr: 0.000007 loss: 0.5336 (0.5258) time: 0.4348 data: 0.0002 max mem: 52150 +Epoch: [17] [6940/7269] eta: 0:02:22 lr: 0.000007 loss: 0.5148 (0.5258) time: 0.4345 data: 0.0001 max mem: 52150 +Epoch: [17] [6950/7269] eta: 0:02:18 lr: 0.000007 loss: 0.5148 (0.5258) time: 0.4301 data: 0.0001 max mem: 52150 +Epoch: [17] [6960/7269] eta: 0:02:14 lr: 0.000007 loss: 0.5007 (0.5258) time: 0.4310 data: 0.0002 max mem: 52150 +Epoch: [17] [6970/7269] eta: 0:02:10 lr: 0.000007 loss: 0.4975 (0.5258) time: 0.5118 data: 0.0002 max mem: 52150 +Epoch: [17] [6980/7269] eta: 0:02:05 lr: 0.000007 loss: 0.5153 (0.5257) time: 0.5112 data: 0.0002 max mem: 52150 +Epoch: [17] [6990/7269] eta: 0:02:01 lr: 0.000007 loss: 0.5284 (0.5258) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [17] [7000/7269] eta: 0:01:56 lr: 0.000007 loss: 0.5284 (0.5257) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [17] [7010/7269] eta: 0:01:52 lr: 0.000007 loss: 0.4810 (0.5257) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [17] [7020/7269] eta: 0:01:48 lr: 0.000007 loss: 0.4810 (0.5256) time: 0.4231 data: 0.0001 max mem: 52150 +Epoch: [17] [7030/7269] eta: 0:01:43 lr: 0.000007 loss: 0.5330 (0.5257) time: 0.4705 data: 0.0001 max mem: 52150 +Epoch: [17] [7040/7269] eta: 0:01:39 lr: 0.000007 loss: 0.5539 (0.5257) time: 0.4194 data: 0.0002 max mem: 52150 +Epoch: [17] [7050/7269] eta: 0:01:35 lr: 0.000007 loss: 0.5443 (0.5257) time: 0.4228 data: 0.0002 max mem: 52150 +Epoch: [17] [7060/7269] eta: 0:01:30 lr: 0.000007 loss: 0.5241 (0.5256) time: 0.4849 data: 0.0002 max mem: 52150 +Epoch: [17] [7070/7269] eta: 0:01:26 lr: 0.000007 loss: 0.5241 (0.5256) time: 0.4965 data: 0.0002 max mem: 52150 +Epoch: [17] [7080/7269] eta: 0:01:22 lr: 0.000007 loss: 0.4994 (0.5256) time: 0.4343 data: 0.0001 max mem: 52150 +Epoch: [17] [7090/7269] eta: 0:01:17 lr: 0.000007 loss: 0.5102 (0.5256) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [17] [7100/7269] eta: 0:01:13 lr: 0.000007 loss: 0.5471 (0.5256) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [17] [7110/7269] eta: 0:01:09 lr: 0.000007 loss: 0.5432 (0.5256) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [17] [7120/7269] eta: 0:01:04 lr: 0.000007 loss: 0.5330 (0.5256) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [17] [7130/7269] eta: 0:01:00 lr: 0.000007 loss: 0.5220 (0.5256) time: 0.3729 data: 0.0001 max mem: 52150 +Epoch: [17] [7140/7269] eta: 0:00:55 lr: 0.000007 loss: 0.5163 (0.5256) time: 0.3727 data: 0.0001 max mem: 52150 +Epoch: [17] [7150/7269] eta: 0:00:51 lr: 0.000007 loss: 0.5298 (0.5256) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [17] [7160/7269] eta: 0:00:47 lr: 0.000007 loss: 0.5416 (0.5256) time: 0.4206 data: 0.0002 max mem: 52150 +Epoch: [17] [7170/7269] eta: 0:00:42 lr: 0.000007 loss: 0.5200 (0.5256) time: 0.4216 data: 0.0001 max mem: 52150 +Epoch: [17] [7180/7269] eta: 0:00:38 lr: 0.000007 loss: 0.5246 (0.5256) time: 0.3731 data: 0.0001 max mem: 52150 +Epoch: [17] [7190/7269] eta: 0:00:34 lr: 0.000007 loss: 0.5266 (0.5256) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [17] [7200/7269] eta: 0:00:29 lr: 0.000007 loss: 0.5209 (0.5256) time: 0.3708 data: 0.0002 max mem: 52150 +Epoch: [17] [7210/7269] eta: 0:00:25 lr: 0.000007 loss: 0.5209 (0.5256) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [17] [7220/7269] eta: 0:00:21 lr: 0.000007 loss: 0.5210 (0.5256) time: 0.5409 data: 0.0001 max mem: 52150 +Epoch: [17] [7230/7269] eta: 0:00:16 lr: 0.000007 loss: 0.5503 (0.5256) time: 0.5874 data: 0.0001 max mem: 52150 +Epoch: [17] [7240/7269] eta: 0:00:12 lr: 0.000007 loss: 0.5409 (0.5256) time: 0.4184 data: 0.0002 max mem: 52150 +Epoch: [17] [7250/7269] eta: 0:00:08 lr: 0.000007 loss: 0.5194 (0.5256) time: 0.4383 data: 0.0023 max mem: 52150 +Epoch: [17] [7260/7269] eta: 0:00:03 lr: 0.000007 loss: 0.5194 (0.5255) time: 0.4869 data: 0.0023 max mem: 52150 +Epoch: [17] [7268/7269] eta: 0:00:00 lr: 0.000007 loss: 0.5429 (0.5255) time: 0.5292 data: 0.0022 max mem: 52150 +Epoch: [17] Total time: 0:52:38 (0.4345 s / it) +Averaged stats: lr: 0.000007 loss: 0.5429 (0.5268) +Test: [ 0/1403] eta: 1:18:38 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0945 (0.0945) acc1: 98.4375 (98.4375) acc5: 100.0000 (100.0000) time: 3.3635 data: 3.2667 max mem: 52150 +Test: [ 10/1403] eta: 0:09:08 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1260 (0.1419) acc1: 98.4375 (97.7273) acc5: 100.0000 (100.0000) time: 0.3934 data: 0.2973 max mem: 52150 +Test: [ 20/1403] eta: 0:05:48 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1083 (0.1353) acc1: 98.4375 (97.1726) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 30/1403] eta: 0:04:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0792 (0.1358) acc1: 98.4375 (96.7742) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 40/1403] eta: 0:04:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0934 (0.1334) acc1: 100.0000 (97.2180) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 50/1403] eta: 0:03:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1055 (0.1512) acc1: 100.0000 (96.1703) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 60/1403] eta: 0:03:21 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.1233 (0.1476) acc1: 100.0000 (96.6701) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 70/1403] eta: 0:03:09 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1188 (0.1477) acc1: 100.0000 (96.5449) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 80/1403] eta: 0:03:00 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1233 (0.1556) acc1: 98.4375 (96.1998) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 90/1403] eta: 0:02:53 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1596 (0.1713) acc1: 95.3125 (95.3468) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 100/1403] eta: 0:02:47 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1543 (0.1706) acc1: 96.8750 (95.4517) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 110/1403] eta: 0:02:42 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1230 (0.1680) acc1: 98.4375 (95.5800) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 120/1403] eta: 0:02:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0838 (0.1607) acc1: 98.4375 (95.9065) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 130/1403] eta: 0:02:34 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.0845 (0.1558) acc1: 98.4375 (96.0878) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 140/1403] eta: 0:02:31 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0771 (0.1499) acc1: 100.0000 (96.3652) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 150/1403] eta: 0:02:27 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0733 (0.1452) acc1: 100.0000 (96.5749) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 160/1403] eta: 0:02:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0735 (0.1416) acc1: 100.0000 (96.7197) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 170/1403] eta: 0:02:22 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0737 (0.1381) acc1: 100.0000 (96.8841) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 180/1403] eta: 0:02:20 acc1_pos: 89.062 acc5_pos: 100.000 loss: 0.0750 (0.1422) acc1: 100.0000 (96.7196) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 190/1403] eta: 0:02:17 acc1_pos: 89.062 acc5_pos: 100.000 loss: 0.2518 (0.1513) acc1: 90.6250 (96.2287) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 200/1403] eta: 0:02:15 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3019 (0.1627) acc1: 87.5000 (95.6157) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 210/1403] eta: 0:02:13 acc1_pos: 82.812 acc5_pos: 100.000 loss: 0.3090 (0.1725) acc1: 85.9375 (95.0607) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 220/1403] eta: 0:02:11 acc1_pos: 89.062 acc5_pos: 100.000 loss: 0.3524 (0.1807) acc1: 84.3750 (94.6620) acc5: 100.0000 (100.0000) time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 230/1403] eta: 0:02:09 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.2053 (0.1772) acc1: 90.6250 (94.8120) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 240/1403] eta: 0:02:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0781 (0.1729) acc1: 100.0000 (95.0207) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 250/1403] eta: 0:02:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0719 (0.1693) acc1: 100.0000 (95.1942) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 260/1403] eta: 0:02:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0743 (0.1659) acc1: 100.0000 (95.3604) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 270/1403] eta: 0:02:02 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0747 (0.1626) acc1: 100.0000 (95.5085) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 280/1403] eta: 0:02:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0710 (0.1594) acc1: 100.0000 (95.6628) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 290/1403] eta: 0:01:59 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0685 (0.1564) acc1: 100.0000 (95.8065) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 300/1403] eta: 0:01:58 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0680 (0.1536) acc1: 100.0000 (95.9406) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 310/1403] eta: 0:01:56 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0717 (0.1514) acc1: 100.0000 (96.0561) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 320/1403] eta: 0:01:55 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1048 (0.1501) acc1: 100.0000 (96.1789) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 330/1403] eta: 0:01:54 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1121 (0.1490) acc1: 100.0000 (96.2944) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 340/1403] eta: 0:01:52 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1121 (0.1478) acc1: 100.0000 (96.4030) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0004 max mem: 52150 +Test: [ 350/1403] eta: 0:01:51 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1090 (0.1468) acc1: 100.0000 (96.5055) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 360/1403] eta: 0:01:50 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1128 (0.1459) acc1: 100.0000 (96.5937) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 370/1403] eta: 0:01:48 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1123 (0.1449) acc1: 100.0000 (96.6855) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0004 max mem: 52150 +Test: [ 380/1403] eta: 0:01:47 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1081 (0.1440) acc1: 100.0000 (96.7725) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 390/1403] eta: 0:01:46 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1104 (0.1452) acc1: 100.0000 (96.7351) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 400/1403] eta: 0:01:44 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2639 (0.1497) acc1: 87.5000 (96.4853) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 410/1403] eta: 0:01:43 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3156 (0.1537) acc1: 85.9375 (96.2439) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 420/1403] eta: 0:01:42 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2713 (0.1560) acc1: 89.0625 (96.1067) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 430/1403] eta: 0:01:41 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2472 (0.1595) acc1: 89.0625 (95.9252) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 440/1403] eta: 0:01:40 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3020 (0.1636) acc1: 87.5000 (95.7093) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 450/1403] eta: 0:01:38 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2922 (0.1664) acc1: 87.5000 (95.5585) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 460/1403] eta: 0:01:37 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2810 (0.1695) acc1: 89.0625 (95.3905) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 470/1403] eta: 0:01:36 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2406 (0.1689) acc1: 90.6250 (95.4087) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 480/1403] eta: 0:01:35 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0630 (0.1667) acc1: 100.0000 (95.5042) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 490/1403] eta: 0:01:34 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0636 (0.1646) acc1: 100.0000 (95.5957) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 500/1403] eta: 0:01:33 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0644 (0.1626) acc1: 100.0000 (95.6836) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 510/1403] eta: 0:01:31 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0644 (0.1607) acc1: 100.0000 (95.7681) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 520/1403] eta: 0:01:30 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0698 (0.1590) acc1: 100.0000 (95.8493) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 530/1403] eta: 0:01:29 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0624 (0.1572) acc1: 100.0000 (95.9275) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 540/1403] eta: 0:01:28 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0617 (0.1554) acc1: 100.0000 (96.0028) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 550/1403] eta: 0:01:27 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0625 (0.1542) acc1: 100.0000 (96.0583) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 560/1403] eta: 0:01:26 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0809 (0.1531) acc1: 98.4375 (96.1063) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 570/1403] eta: 0:01:25 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0930 (0.1522) acc1: 98.4375 (96.1416) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 580/1403] eta: 0:01:24 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1077 (0.1519) acc1: 98.4375 (96.1543) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 590/1403] eta: 0:01:22 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1207 (0.1517) acc1: 96.8750 (96.1479) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 600/1403] eta: 0:01:21 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1303 (0.1522) acc1: 95.3125 (96.1002) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 610/1403] eta: 0:01:20 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1352 (0.1518) acc1: 95.3125 (96.1104) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 620/1403] eta: 0:01:19 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1476 (0.1532) acc1: 95.3125 (95.9994) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 630/1403] eta: 0:01:18 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1785 (0.1533) acc1: 93.7500 (95.9860) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 640/1403] eta: 0:01:17 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1360 (0.1530) acc1: 96.8750 (95.9950) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 650/1403] eta: 0:01:16 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1360 (0.1530) acc1: 96.8750 (95.9917) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 660/1403] eta: 0:01:15 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1204 (0.1521) acc1: 98.4375 (96.0382) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 670/1403] eta: 0:01:14 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0816 (0.1511) acc1: 100.0000 (96.0879) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 680/1403] eta: 0:01:13 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0896 (0.1532) acc1: 98.4375 (95.9412) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 690/1403] eta: 0:01:12 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3482 (0.1584) acc1: 82.8125 (95.6178) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 700/1403] eta: 0:01:11 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3076 (0.1590) acc1: 84.3750 (95.5777) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 710/1403] eta: 0:01:10 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3805 (0.1639) acc1: 81.2500 (95.3037) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 720/1403] eta: 0:01:09 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3805 (0.1641) acc1: 79.6875 (95.2908) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 730/1403] eta: 0:01:07 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0712 (0.1630) acc1: 100.0000 (95.3424) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 740/1403] eta: 0:01:06 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0801 (0.1622) acc1: 100.0000 (95.3863) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 750/1403] eta: 0:01:05 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1244 (0.1650) acc1: 96.8750 (95.2064) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 760/1403] eta: 0:01:04 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.4447 (0.1706) acc1: 78.1250 (94.8628) acc5: 100.0000 (100.0000) acc1_neg: 70.312 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 770/1403] eta: 0:01:03 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.6357 (0.1771) acc1: 65.6250 (94.4593) acc5: 100.0000 (100.0000) acc1_neg: 53.125 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 780/1403] eta: 0:01:02 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.5457 (0.1802) acc1: 70.3125 (94.2762) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 790/1403] eta: 0:01:01 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3594 (0.1822) acc1: 81.2500 (94.1727) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 800/1403] eta: 0:01:00 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2259 (0.1826) acc1: 92.1875 (94.1596) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 810/1403] eta: 0:00:59 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2401 (0.1858) acc1: 90.6250 (94.0159) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 820/1403] eta: 0:00:58 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.6557 (0.1946) acc1: 71.8750 (93.6377) acc5: 100.0000 (100.0000) acc1_neg: 42.188 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 830/1403] eta: 0:00:57 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.6786 (0.1972) acc1: 68.7500 (93.5112) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 840/1403] eta: 0:00:56 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.4882 (0.2045) acc1: 75.0000 (93.1908) acc5: 100.0000 (100.0000) acc1_neg: 51.562 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 850/1403] eta: 0:00:55 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.9559 (0.2144) acc1: 56.2500 (92.7677) acc5: 100.0000 (100.0000) acc1_neg: 56.250 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 860/1403] eta: 0:00:54 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.9309 (0.2175) acc1: 57.8125 (92.6267) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 870/1403] eta: 0:00:53 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1373 (0.2164) acc1: 96.8750 (92.6737) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 880/1403] eta: 0:00:52 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1077 (0.2153) acc1: 98.4375 (92.7302) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 890/1403] eta: 0:00:51 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0983 (0.2139) acc1: 98.4375 (92.7995) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 900/1403] eta: 0:00:50 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0830 (0.2125) acc1: 100.0000 (92.8742) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 910/1403] eta: 0:00:49 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0869 (0.2118) acc1: 98.4375 (92.8976) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 920/1403] eta: 0:00:48 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1409 (0.2111) acc1: 95.3125 (92.9289) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 930/1403] eta: 0:00:47 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1338 (0.2103) acc1: 95.3125 (92.9578) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 940/1403] eta: 0:00:46 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1514 (0.2113) acc1: 93.7500 (92.8949) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 950/1403] eta: 0:00:45 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3594 (0.2136) acc1: 84.3750 (92.7609) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 960/1403] eta: 0:00:44 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.4001 (0.2155) acc1: 79.6875 (92.6558) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 970/1403] eta: 0:00:43 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3800 (0.2174) acc1: 81.2500 (92.5528) acc5: 100.0000 (100.0000) acc1_neg: 70.312 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 980/1403] eta: 0:00:42 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3786 (0.2195) acc1: 81.2500 (92.4169) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 990/1403] eta: 0:00:41 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3745 (0.2211) acc1: 81.2500 (92.3294) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [1000/1403] eta: 0:00:40 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3834 (0.2229) acc1: 82.8125 (92.2250) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1010/1403] eta: 0:00:39 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.4092 (0.2247) acc1: 81.2500 (92.1149) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1020/1403] eta: 0:00:38 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.5320 (0.2321) acc1: 73.4375 (91.7161) acc5: 100.0000 (100.0000) acc1_neg: 28.125 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1030/1403] eta: 0:00:37 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.7371 (0.2347) acc1: 53.1250 (91.5570) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1040/1403] eta: 0:00:36 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2398 (0.2347) acc1: 89.0625 (91.5406) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1050/1403] eta: 0:00:35 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1949 (0.2342) acc1: 93.7500 (91.5705) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1060/1403] eta: 0:00:34 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1633 (0.2334) acc1: 95.3125 (91.6161) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1070/1403] eta: 0:00:33 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1870 (0.2340) acc1: 95.3125 (91.5952) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1080/1403] eta: 0:00:32 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3826 (0.2358) acc1: 84.3750 (91.5024) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1090/1403] eta: 0:00:31 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.4697 (0.2379) acc1: 81.2500 (91.4055) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1100/1403] eta: 0:00:30 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.4467 (0.2392) acc1: 81.2500 (91.3403) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1110/1403] eta: 0:00:29 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3682 (0.2404) acc1: 85.9375 (91.2930) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1120/1403] eta: 0:00:28 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3680 (0.2415) acc1: 85.9375 (91.2480) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1130/1403] eta: 0:00:27 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3685 (0.2431) acc1: 82.8125 (91.1610) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1140/1403] eta: 0:00:26 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3542 (0.2439) acc1: 82.8125 (91.1248) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1150/1403] eta: 0:00:25 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3542 (0.2451) acc1: 85.9375 (91.0635) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1160/1403] eta: 0:00:24 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3824 (0.2463) acc1: 82.8125 (91.0032) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52150 +Test: [1170/1403] eta: 0:00:23 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3706 (0.2465) acc1: 84.3750 (90.9973) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1180/1403] eta: 0:00:22 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0730 (0.2450) acc1: 100.0000 (91.0735) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1190/1403] eta: 0:00:21 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0719 (0.2435) acc1: 100.0000 (91.1485) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1200/1403] eta: 0:00:20 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0722 (0.2421) acc1: 100.0000 (91.2222) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1210/1403] eta: 0:00:19 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0718 (0.2407) acc1: 100.0000 (91.2946) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1220/1403] eta: 0:00:18 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0713 (0.2393) acc1: 100.0000 (91.3659) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1230/1403] eta: 0:00:17 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0724 (0.2380) acc1: 100.0000 (91.4361) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1240/1403] eta: 0:00:16 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0718 (0.2366) acc1: 100.0000 (91.5051) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52150 +Test: [1250/1403] eta: 0:00:15 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0719 (0.2353) acc1: 100.0000 (91.5730) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52150 +Test: [1260/1403] eta: 0:00:14 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0754 (0.2341) acc1: 100.0000 (91.6398) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1270/1403] eta: 0:00:13 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0772 (0.2329) acc1: 100.0000 (91.7056) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52150 +Test: [1280/1403] eta: 0:00:12 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0765 (0.2316) acc1: 100.0000 (91.7703) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52150 +Test: [1290/1403] eta: 0:00:11 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0765 (0.2304) acc1: 100.0000 (91.8341) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52150 +Test: [1300/1403] eta: 0:00:10 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0771 (0.2293) acc1: 100.0000 (91.8969) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1310/1403] eta: 0:00:09 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0773 (0.2281) acc1: 100.0000 (91.9587) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52150 +Test: [1320/1403] eta: 0:00:08 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0755 (0.2269) acc1: 100.0000 (92.0195) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52150 +Test: [1330/1403] eta: 0:00:07 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0760 (0.2258) acc1: 100.0000 (92.0760) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1340/1403] eta: 0:00:06 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0718 (0.2247) acc1: 100.0000 (92.1351) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0002 max mem: 52150 +Test: [1350/1403] eta: 0:00:05 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0702 (0.2235) acc1: 100.0000 (92.1921) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1360/1403] eta: 0:00:04 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0699 (0.2224) acc1: 100.0000 (92.2495) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52150 +Test: [1370/1403] eta: 0:00:03 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0717 (0.2214) acc1: 100.0000 (92.3037) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52150 +Test: [1380/1403] eta: 0:00:02 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0691 (0.2203) acc1: 100.0000 (92.3595) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52150 +Test: [1390/1403] eta: 0:00:01 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0691 (0.2192) acc1: 100.0000 (92.4144) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0002 max mem: 52150 +Test: [1400/1403] eta: 0:00:00 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0696 (0.2182) acc1: 100.0000 (92.4652) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0001 max mem: 52150 +Test: [1402/1403] eta: 0:00:00 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0706 (0.2180) acc1: 100.0000 (92.4707) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0922 data: 0.0001 max mem: 52150 +Test: Total time: 0:02:19 (0.0992 s / it) +* Overall Acc@1 92.471 Acc@5 100.000 + target == real Acc@1 96.021 Acc@5 100.000 + target == fake Acc@1 91.470 Acc@5 100.000 + loss 0.218 +Accuracy of the network on the 89729 test images: 92.5% +Max accuracy: 93.40% +Epoch: [18] [ 0/7269] eta: 11:29:07 lr: 0.000007 loss: 0.4733 (0.4733) time: 5.6882 data: 4.8796 max mem: 52150 +Epoch: [18] [ 10/7269] eta: 1:57:52 lr: 0.000007 loss: 0.5768 (0.5658) time: 0.9743 data: 0.4438 max mem: 52150 +Epoch: [18] [ 20/7269] eta: 1:23:01 lr: 0.000007 loss: 0.5890 (0.5749) time: 0.4371 data: 0.0002 max mem: 52150 +Epoch: [18] [ 30/7269] eta: 1:10:38 lr: 0.000007 loss: 0.5421 (0.5516) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [18] [ 40/7269] eta: 1:07:08 lr: 0.000007 loss: 0.5361 (0.5525) time: 0.4207 data: 0.0002 max mem: 52150 +Epoch: [18] [ 50/7269] eta: 1:02:40 lr: 0.000007 loss: 0.5497 (0.5404) time: 0.4207 data: 0.0002 max mem: 52150 +Epoch: [18] [ 60/7269] eta: 0:59:40 lr: 0.000007 loss: 0.5331 (0.5412) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [18] [ 70/7269] eta: 0:59:12 lr: 0.000007 loss: 0.5379 (0.5422) time: 0.4234 data: 0.0002 max mem: 52150 +Epoch: [18] [ 80/7269] eta: 0:58:48 lr: 0.000007 loss: 0.5142 (0.5340) time: 0.4732 data: 0.0002 max mem: 52150 +Epoch: [18] [ 90/7269] eta: 0:59:39 lr: 0.000007 loss: 0.5125 (0.5366) time: 0.5172 data: 0.0002 max mem: 52150 +Epoch: [18] [ 100/7269] eta: 0:59:15 lr: 0.000007 loss: 0.5378 (0.5352) time: 0.5169 data: 0.0002 max mem: 52150 +Epoch: [18] [ 110/7269] eta: 0:58:56 lr: 0.000007 loss: 0.5170 (0.5329) time: 0.4728 data: 0.0002 max mem: 52150 +Epoch: [18] [ 120/7269] eta: 0:57:39 lr: 0.000007 loss: 0.5376 (0.5336) time: 0.4232 data: 0.0002 max mem: 52150 +Epoch: [18] [ 130/7269] eta: 0:58:32 lr: 0.000007 loss: 0.5376 (0.5319) time: 0.4806 data: 0.0002 max mem: 52150 +Epoch: [18] [ 140/7269] eta: 0:57:26 lr: 0.000007 loss: 0.5147 (0.5301) time: 0.4801 data: 0.0002 max mem: 52150 +Epoch: [18] [ 150/7269] eta: 0:56:28 lr: 0.000007 loss: 0.5041 (0.5288) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [18] [ 160/7269] eta: 0:57:16 lr: 0.000007 loss: 0.4791 (0.5267) time: 0.4833 data: 0.0002 max mem: 52150 +Epoch: [18] [ 170/7269] eta: 0:56:25 lr: 0.000007 loss: 0.5525 (0.5286) time: 0.4835 data: 0.0002 max mem: 52150 +Epoch: [18] [ 180/7269] eta: 0:56:17 lr: 0.000007 loss: 0.5649 (0.5294) time: 0.4208 data: 0.0002 max mem: 52150 +Epoch: [18] [ 190/7269] eta: 0:56:20 lr: 0.000007 loss: 0.5539 (0.5296) time: 0.4828 data: 0.0002 max mem: 52150 +Epoch: [18] [ 200/7269] eta: 0:55:38 lr: 0.000007 loss: 0.5620 (0.5315) time: 0.4341 data: 0.0002 max mem: 52150 +Epoch: [18] [ 210/7269] eta: 0:54:59 lr: 0.000007 loss: 0.5490 (0.5311) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [18] [ 220/7269] eta: 0:54:24 lr: 0.000007 loss: 0.5129 (0.5294) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [18] [ 230/7269] eta: 0:53:52 lr: 0.000007 loss: 0.4989 (0.5279) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [18] [ 240/7269] eta: 0:53:50 lr: 0.000007 loss: 0.5283 (0.5293) time: 0.4199 data: 0.0002 max mem: 52150 +Epoch: [18] [ 250/7269] eta: 0:53:21 lr: 0.000007 loss: 0.5416 (0.5293) time: 0.4195 data: 0.0002 max mem: 52150 +Epoch: [18] [ 260/7269] eta: 0:53:19 lr: 0.000007 loss: 0.5362 (0.5298) time: 0.4196 data: 0.0002 max mem: 52150 +Epoch: [18] [ 270/7269] eta: 0:53:58 lr: 0.000007 loss: 0.5289 (0.5283) time: 0.5453 data: 0.0002 max mem: 52150 +Epoch: [18] [ 280/7269] eta: 0:54:01 lr: 0.000007 loss: 0.5136 (0.5279) time: 0.5596 data: 0.0002 max mem: 52150 +Epoch: [18] [ 290/7269] eta: 0:53:57 lr: 0.000007 loss: 0.5136 (0.5275) time: 0.4809 data: 0.0002 max mem: 52150 +Epoch: [18] [ 300/7269] eta: 0:54:01 lr: 0.000007 loss: 0.5169 (0.5280) time: 0.4825 data: 0.0002 max mem: 52150 +Epoch: [18] [ 310/7269] eta: 0:53:35 lr: 0.000007 loss: 0.5028 (0.5265) time: 0.4353 data: 0.0002 max mem: 52150 +Epoch: [18] [ 320/7269] eta: 0:53:11 lr: 0.000007 loss: 0.4859 (0.5260) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [18] [ 330/7269] eta: 0:52:48 lr: 0.000007 loss: 0.5104 (0.5262) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [18] [ 340/7269] eta: 0:52:27 lr: 0.000007 loss: 0.5212 (0.5261) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [18] [ 350/7269] eta: 0:52:06 lr: 0.000007 loss: 0.5399 (0.5262) time: 0.3730 data: 0.0002 max mem: 52150 +Epoch: [18] [ 360/7269] eta: 0:51:46 lr: 0.000007 loss: 0.5201 (0.5258) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [18] [ 370/7269] eta: 0:51:27 lr: 0.000007 loss: 0.5033 (0.5261) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [18] [ 380/7269] eta: 0:51:32 lr: 0.000007 loss: 0.5184 (0.5266) time: 0.4345 data: 0.0002 max mem: 52150 +Epoch: [18] [ 390/7269] eta: 0:51:14 lr: 0.000007 loss: 0.4992 (0.5253) time: 0.4345 data: 0.0002 max mem: 52150 +Epoch: [18] [ 400/7269] eta: 0:51:39 lr: 0.000007 loss: 0.5107 (0.5256) time: 0.4966 data: 0.0002 max mem: 52150 +Epoch: [18] [ 410/7269] eta: 0:51:21 lr: 0.000007 loss: 0.5360 (0.5251) time: 0.4968 data: 0.0002 max mem: 52150 +Epoch: [18] [ 420/7269] eta: 0:51:24 lr: 0.000007 loss: 0.5459 (0.5256) time: 0.4335 data: 0.0002 max mem: 52150 +Epoch: [18] [ 430/7269] eta: 0:51:07 lr: 0.000007 loss: 0.5368 (0.5251) time: 0.4335 data: 0.0002 max mem: 52150 +Epoch: [18] [ 440/7269] eta: 0:50:51 lr: 0.000007 loss: 0.4972 (0.5251) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [18] [ 450/7269] eta: 0:50:54 lr: 0.000007 loss: 0.5418 (0.5259) time: 0.4336 data: 0.0002 max mem: 52150 +Epoch: [18] [ 460/7269] eta: 0:50:38 lr: 0.000007 loss: 0.5671 (0.5270) time: 0.4339 data: 0.0002 max mem: 52150 +Epoch: [18] [ 470/7269] eta: 0:50:23 lr: 0.000007 loss: 0.5757 (0.5282) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [18] [ 480/7269] eta: 0:50:09 lr: 0.000007 loss: 0.5602 (0.5277) time: 0.3728 data: 0.0002 max mem: 52150 +Epoch: [18] [ 490/7269] eta: 0:49:54 lr: 0.000007 loss: 0.4947 (0.5271) time: 0.3728 data: 0.0002 max mem: 52150 +Epoch: [18] [ 500/7269] eta: 0:49:41 lr: 0.000007 loss: 0.4910 (0.5270) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [18] [ 510/7269] eta: 0:49:56 lr: 0.000007 loss: 0.5386 (0.5269) time: 0.4824 data: 0.0002 max mem: 52150 +Epoch: [18] [ 520/7269] eta: 0:49:43 lr: 0.000007 loss: 0.5422 (0.5277) time: 0.4823 data: 0.0002 max mem: 52150 +Epoch: [18] [ 530/7269] eta: 0:49:29 lr: 0.000007 loss: 0.5611 (0.5282) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [18] [ 540/7269] eta: 0:49:16 lr: 0.000007 loss: 0.5547 (0.5285) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [18] [ 550/7269] eta: 0:49:04 lr: 0.000007 loss: 0.5151 (0.5282) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [18] [ 560/7269] eta: 0:48:51 lr: 0.000007 loss: 0.5107 (0.5283) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [18] [ 570/7269] eta: 0:48:39 lr: 0.000007 loss: 0.5402 (0.5283) time: 0.3712 data: 0.0002 max mem: 52150 +Epoch: [18] [ 580/7269] eta: 0:48:28 lr: 0.000007 loss: 0.5332 (0.5286) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [18] [ 590/7269] eta: 0:48:16 lr: 0.000007 loss: 0.5288 (0.5285) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [18] [ 600/7269] eta: 0:48:05 lr: 0.000007 loss: 0.5379 (0.5286) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [18] [ 610/7269] eta: 0:47:54 lr: 0.000007 loss: 0.5457 (0.5289) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [18] [ 620/7269] eta: 0:47:43 lr: 0.000007 loss: 0.5249 (0.5285) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [18] [ 630/7269] eta: 0:47:46 lr: 0.000007 loss: 0.5490 (0.5293) time: 0.4362 data: 0.0002 max mem: 52150 +Epoch: [18] [ 640/7269] eta: 0:47:49 lr: 0.000007 loss: 0.5748 (0.5297) time: 0.4997 data: 0.0002 max mem: 52150 +Epoch: [18] [ 650/7269] eta: 0:47:51 lr: 0.000007 loss: 0.5625 (0.5297) time: 0.4959 data: 0.0002 max mem: 52150 +Epoch: [18] [ 660/7269] eta: 0:47:50 lr: 0.000007 loss: 0.5274 (0.5299) time: 0.4801 data: 0.0002 max mem: 52150 +Epoch: [18] [ 670/7269] eta: 0:47:39 lr: 0.000007 loss: 0.5274 (0.5293) time: 0.4194 data: 0.0002 max mem: 52150 +Epoch: [18] [ 680/7269] eta: 0:47:46 lr: 0.000007 loss: 0.5155 (0.5292) time: 0.4614 data: 0.0002 max mem: 52150 +Epoch: [18] [ 690/7269] eta: 0:47:45 lr: 0.000007 loss: 0.5144 (0.5292) time: 0.5073 data: 0.0001 max mem: 52150 +Epoch: [18] [ 700/7269] eta: 0:47:34 lr: 0.000007 loss: 0.5273 (0.5295) time: 0.4173 data: 0.0002 max mem: 52150 +Epoch: [18] [ 710/7269] eta: 0:47:24 lr: 0.000007 loss: 0.5423 (0.5301) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [18] [ 720/7269] eta: 0:47:14 lr: 0.000007 loss: 0.5395 (0.5299) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [18] [ 730/7269] eta: 0:47:15 lr: 0.000007 loss: 0.5173 (0.5296) time: 0.4329 data: 0.0002 max mem: 52150 +Epoch: [18] [ 740/7269] eta: 0:47:16 lr: 0.000007 loss: 0.4992 (0.5290) time: 0.4942 data: 0.0002 max mem: 52150 +Epoch: [18] [ 750/7269] eta: 0:47:17 lr: 0.000007 loss: 0.5264 (0.5294) time: 0.4940 data: 0.0002 max mem: 52150 +Epoch: [18] [ 760/7269] eta: 0:47:07 lr: 0.000007 loss: 0.5173 (0.5287) time: 0.4327 data: 0.0002 max mem: 52150 +Epoch: [18] [ 770/7269] eta: 0:46:58 lr: 0.000007 loss: 0.4995 (0.5289) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [18] [ 780/7269] eta: 0:46:48 lr: 0.000007 loss: 0.5602 (0.5289) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [18] [ 790/7269] eta: 0:46:46 lr: 0.000007 loss: 0.5024 (0.5283) time: 0.4185 data: 0.0002 max mem: 52150 +Epoch: [18] [ 800/7269] eta: 0:46:37 lr: 0.000007 loss: 0.5226 (0.5284) time: 0.4188 data: 0.0002 max mem: 52150 +Epoch: [18] [ 810/7269] eta: 0:46:38 lr: 0.000007 loss: 0.5291 (0.5286) time: 0.4370 data: 0.0002 max mem: 52150 +Epoch: [18] [ 820/7269] eta: 0:46:37 lr: 0.000007 loss: 0.5606 (0.5289) time: 0.4838 data: 0.0002 max mem: 52150 +Epoch: [18] [ 830/7269] eta: 0:46:46 lr: 0.000007 loss: 0.5509 (0.5290) time: 0.5408 data: 0.0002 max mem: 52150 +Epoch: [18] [ 840/7269] eta: 0:46:44 lr: 0.000007 loss: 0.5502 (0.5292) time: 0.5406 data: 0.0002 max mem: 52150 +Epoch: [18] [ 850/7269] eta: 0:46:35 lr: 0.000007 loss: 0.5513 (0.5294) time: 0.4189 data: 0.0002 max mem: 52150 +Epoch: [18] [ 860/7269] eta: 0:46:34 lr: 0.000007 loss: 0.5308 (0.5292) time: 0.4242 data: 0.0002 max mem: 52150 +Epoch: [18] [ 870/7269] eta: 0:46:34 lr: 0.000007 loss: 0.4912 (0.5285) time: 0.4877 data: 0.0002 max mem: 52150 +Epoch: [18] [ 880/7269] eta: 0:46:25 lr: 0.000007 loss: 0.4872 (0.5281) time: 0.4359 data: 0.0002 max mem: 52150 +Epoch: [18] [ 890/7269] eta: 0:46:23 lr: 0.000007 loss: 0.4672 (0.5275) time: 0.4192 data: 0.0002 max mem: 52150 +Epoch: [18] [ 900/7269] eta: 0:46:14 lr: 0.000007 loss: 0.5199 (0.5276) time: 0.4191 data: 0.0002 max mem: 52150 +Epoch: [18] [ 910/7269] eta: 0:46:05 lr: 0.000007 loss: 0.5255 (0.5274) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [18] [ 920/7269] eta: 0:45:56 lr: 0.000007 loss: 0.4910 (0.5270) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [18] [ 930/7269] eta: 0:45:55 lr: 0.000007 loss: 0.5139 (0.5270) time: 0.4235 data: 0.0002 max mem: 52150 +Epoch: [18] [ 940/7269] eta: 0:46:01 lr: 0.000007 loss: 0.5539 (0.5269) time: 0.5320 data: 0.0002 max mem: 52150 +Epoch: [18] [ 950/7269] eta: 0:45:52 lr: 0.000007 loss: 0.5560 (0.5271) time: 0.4799 data: 0.0002 max mem: 52150 +Epoch: [18] [ 960/7269] eta: 0:45:44 lr: 0.000007 loss: 0.5580 (0.5274) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [18] [ 970/7269] eta: 0:45:51 lr: 0.000007 loss: 0.5234 (0.5270) time: 0.4910 data: 0.0002 max mem: 52150 +Epoch: [18] [ 980/7269] eta: 0:45:48 lr: 0.000007 loss: 0.5110 (0.5272) time: 0.5404 data: 0.0002 max mem: 52150 +Epoch: [18] [ 990/7269] eta: 0:45:40 lr: 0.000007 loss: 0.5407 (0.5273) time: 0.4213 data: 0.0002 max mem: 52150 +Epoch: [18] [1000/7269] eta: 0:45:32 lr: 0.000007 loss: 0.5517 (0.5276) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [18] [1010/7269] eta: 0:45:31 lr: 0.000007 loss: 0.5427 (0.5274) time: 0.4339 data: 0.0001 max mem: 52150 +Epoch: [18] [1020/7269] eta: 0:45:23 lr: 0.000007 loss: 0.4918 (0.5271) time: 0.4332 data: 0.0002 max mem: 52150 +Epoch: [18] [1030/7269] eta: 0:45:14 lr: 0.000007 loss: 0.5065 (0.5271) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [18] [1040/7269] eta: 0:45:06 lr: 0.000007 loss: 0.5421 (0.5273) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [18] [1050/7269] eta: 0:44:58 lr: 0.000007 loss: 0.5600 (0.5273) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [18] [1060/7269] eta: 0:44:50 lr: 0.000007 loss: 0.5432 (0.5272) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [18] [1070/7269] eta: 0:44:42 lr: 0.000007 loss: 0.5141 (0.5269) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [18] [1080/7269] eta: 0:44:35 lr: 0.000007 loss: 0.5088 (0.5268) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [18] [1090/7269] eta: 0:44:39 lr: 0.000007 loss: 0.5439 (0.5268) time: 0.4803 data: 0.0002 max mem: 52150 +Epoch: [18] [1100/7269] eta: 0:44:38 lr: 0.000007 loss: 0.5283 (0.5268) time: 0.5407 data: 0.0001 max mem: 52150 +Epoch: [18] [1110/7269] eta: 0:44:37 lr: 0.000007 loss: 0.5283 (0.5270) time: 0.4962 data: 0.0001 max mem: 52150 +Epoch: [18] [1120/7269] eta: 0:44:36 lr: 0.000007 loss: 0.5553 (0.5270) time: 0.4956 data: 0.0002 max mem: 52150 +Epoch: [18] [1130/7269] eta: 0:44:33 lr: 0.000007 loss: 0.5523 (0.5271) time: 0.4780 data: 0.0002 max mem: 52150 +Epoch: [18] [1140/7269] eta: 0:44:26 lr: 0.000007 loss: 0.5605 (0.5272) time: 0.4175 data: 0.0002 max mem: 52150 +Epoch: [18] [1150/7269] eta: 0:44:23 lr: 0.000007 loss: 0.5773 (0.5274) time: 0.4195 data: 0.0001 max mem: 52150 +Epoch: [18] [1160/7269] eta: 0:44:21 lr: 0.000007 loss: 0.5475 (0.5275) time: 0.4711 data: 0.0001 max mem: 52150 +Epoch: [18] [1170/7269] eta: 0:44:13 lr: 0.000007 loss: 0.5070 (0.5272) time: 0.4231 data: 0.0001 max mem: 52150 +Epoch: [18] [1180/7269] eta: 0:44:05 lr: 0.000007 loss: 0.5070 (0.5272) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [18] [1190/7269] eta: 0:44:02 lr: 0.000007 loss: 0.5283 (0.5271) time: 0.4180 data: 0.0002 max mem: 52150 +Epoch: [18] [1200/7269] eta: 0:43:55 lr: 0.000007 loss: 0.5447 (0.5273) time: 0.4182 data: 0.0002 max mem: 52150 +Epoch: [18] [1210/7269] eta: 0:43:54 lr: 0.000007 loss: 0.5412 (0.5273) time: 0.4325 data: 0.0002 max mem: 52150 +Epoch: [18] [1220/7269] eta: 0:43:57 lr: 0.000007 loss: 0.5366 (0.5272) time: 0.5401 data: 0.0002 max mem: 52150 +Epoch: [18] [1230/7269] eta: 0:43:54 lr: 0.000007 loss: 0.5220 (0.5270) time: 0.5250 data: 0.0002 max mem: 52150 +Epoch: [18] [1240/7269] eta: 0:43:46 lr: 0.000007 loss: 0.4943 (0.5269) time: 0.4173 data: 0.0002 max mem: 52150 +Epoch: [18] [1250/7269] eta: 0:43:45 lr: 0.000007 loss: 0.4942 (0.5266) time: 0.4320 data: 0.0002 max mem: 52150 +Epoch: [18] [1260/7269] eta: 0:43:37 lr: 0.000007 loss: 0.5293 (0.5266) time: 0.4322 data: 0.0002 max mem: 52150 +Epoch: [18] [1270/7269] eta: 0:43:30 lr: 0.000007 loss: 0.5472 (0.5267) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [18] [1280/7269] eta: 0:43:23 lr: 0.000007 loss: 0.5509 (0.5267) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [18] [1290/7269] eta: 0:43:15 lr: 0.000007 loss: 0.5279 (0.5265) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [18] [1300/7269] eta: 0:43:14 lr: 0.000007 loss: 0.5279 (0.5267) time: 0.4330 data: 0.0001 max mem: 52150 +Epoch: [18] [1310/7269] eta: 0:43:11 lr: 0.000007 loss: 0.5297 (0.5265) time: 0.4784 data: 0.0001 max mem: 52150 +Epoch: [18] [1320/7269] eta: 0:43:03 lr: 0.000007 loss: 0.5193 (0.5265) time: 0.4179 data: 0.0002 max mem: 52150 +Epoch: [18] [1330/7269] eta: 0:43:02 lr: 0.000007 loss: 0.5246 (0.5263) time: 0.4329 data: 0.0002 max mem: 52150 +Epoch: [18] [1340/7269] eta: 0:43:00 lr: 0.000007 loss: 0.5139 (0.5263) time: 0.4962 data: 0.0002 max mem: 52150 +Epoch: [18] [1350/7269] eta: 0:42:53 lr: 0.000007 loss: 0.5258 (0.5264) time: 0.4359 data: 0.0002 max mem: 52150 +Epoch: [18] [1360/7269] eta: 0:42:50 lr: 0.000007 loss: 0.5160 (0.5261) time: 0.4178 data: 0.0002 max mem: 52150 +Epoch: [18] [1370/7269] eta: 0:42:43 lr: 0.000007 loss: 0.4857 (0.5259) time: 0.4179 data: 0.0002 max mem: 52150 +Epoch: [18] [1380/7269] eta: 0:42:36 lr: 0.000007 loss: 0.5025 (0.5258) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [18] [1390/7269] eta: 0:42:33 lr: 0.000007 loss: 0.5065 (0.5259) time: 0.4211 data: 0.0002 max mem: 52150 +Epoch: [18] [1400/7269] eta: 0:42:34 lr: 0.000007 loss: 0.5318 (0.5257) time: 0.5136 data: 0.0002 max mem: 52150 +Epoch: [18] [1410/7269] eta: 0:42:31 lr: 0.000007 loss: 0.5139 (0.5258) time: 0.5132 data: 0.0002 max mem: 52150 +Epoch: [18] [1420/7269] eta: 0:42:24 lr: 0.000007 loss: 0.5200 (0.5258) time: 0.4215 data: 0.0002 max mem: 52150 +Epoch: [18] [1430/7269] eta: 0:42:17 lr: 0.000007 loss: 0.5336 (0.5259) time: 0.3730 data: 0.0002 max mem: 52150 +Epoch: [18] [1440/7269] eta: 0:42:10 lr: 0.000007 loss: 0.5233 (0.5257) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [18] [1450/7269] eta: 0:42:12 lr: 0.000007 loss: 0.5284 (0.5256) time: 0.4807 data: 0.0002 max mem: 52150 +Epoch: [18] [1460/7269] eta: 0:42:05 lr: 0.000007 loss: 0.5481 (0.5259) time: 0.4809 data: 0.0002 max mem: 52150 +Epoch: [18] [1470/7269] eta: 0:41:59 lr: 0.000007 loss: 0.5534 (0.5259) time: 0.3728 data: 0.0002 max mem: 52150 +Epoch: [18] [1480/7269] eta: 0:41:52 lr: 0.000007 loss: 0.5160 (0.5257) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [18] [1490/7269] eta: 0:41:45 lr: 0.000007 loss: 0.5260 (0.5256) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [18] [1500/7269] eta: 0:41:43 lr: 0.000007 loss: 0.5351 (0.5256) time: 0.4323 data: 0.0002 max mem: 52150 +Epoch: [18] [1510/7269] eta: 0:41:36 lr: 0.000007 loss: 0.5385 (0.5259) time: 0.4332 data: 0.0002 max mem: 52150 +Epoch: [18] [1520/7269] eta: 0:41:30 lr: 0.000007 loss: 0.5360 (0.5258) time: 0.3729 data: 0.0002 max mem: 52150 +Epoch: [18] [1530/7269] eta: 0:41:23 lr: 0.000007 loss: 0.4910 (0.5257) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [18] [1540/7269] eta: 0:41:16 lr: 0.000007 loss: 0.5359 (0.5256) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [18] [1550/7269] eta: 0:41:14 lr: 0.000007 loss: 0.5359 (0.5257) time: 0.4318 data: 0.0002 max mem: 52150 +Epoch: [18] [1560/7269] eta: 0:41:12 lr: 0.000007 loss: 0.5188 (0.5256) time: 0.4948 data: 0.0002 max mem: 52150 +Epoch: [18] [1570/7269] eta: 0:41:14 lr: 0.000007 loss: 0.5145 (0.5255) time: 0.5425 data: 0.0002 max mem: 52150 +Epoch: [18] [1580/7269] eta: 0:41:07 lr: 0.000007 loss: 0.5317 (0.5254) time: 0.4794 data: 0.0002 max mem: 52150 +Epoch: [18] [1590/7269] eta: 0:41:08 lr: 0.000007 loss: 0.5426 (0.5253) time: 0.4786 data: 0.0002 max mem: 52150 +Epoch: [18] [1600/7269] eta: 0:41:02 lr: 0.000007 loss: 0.4916 (0.5251) time: 0.4786 data: 0.0002 max mem: 52150 +Epoch: [18] [1610/7269] eta: 0:40:55 lr: 0.000007 loss: 0.4645 (0.5248) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [18] [1620/7269] eta: 0:40:49 lr: 0.000007 loss: 0.4763 (0.5247) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [18] [1630/7269] eta: 0:40:46 lr: 0.000007 loss: 0.5495 (0.5249) time: 0.4221 data: 0.0002 max mem: 52150 +Epoch: [18] [1640/7269] eta: 0:40:39 lr: 0.000007 loss: 0.5450 (0.5248) time: 0.4224 data: 0.0002 max mem: 52150 +Epoch: [18] [1650/7269] eta: 0:40:33 lr: 0.000007 loss: 0.5450 (0.5250) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [18] [1660/7269] eta: 0:40:26 lr: 0.000007 loss: 0.5483 (0.5250) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [18] [1670/7269] eta: 0:40:20 lr: 0.000007 loss: 0.5165 (0.5250) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [18] [1680/7269] eta: 0:40:17 lr: 0.000007 loss: 0.5167 (0.5249) time: 0.4169 data: 0.0002 max mem: 52150 +Epoch: [18] [1690/7269] eta: 0:40:14 lr: 0.000007 loss: 0.5015 (0.5247) time: 0.4779 data: 0.0002 max mem: 52150 +Epoch: [18] [1700/7269] eta: 0:40:11 lr: 0.000007 loss: 0.5074 (0.5246) time: 0.4777 data: 0.0002 max mem: 52150 +Epoch: [18] [1710/7269] eta: 0:40:13 lr: 0.000007 loss: 0.5351 (0.5248) time: 0.5413 data: 0.0002 max mem: 52150 +Epoch: [18] [1720/7269] eta: 0:40:13 lr: 0.000007 loss: 0.5351 (0.5247) time: 0.6045 data: 0.0002 max mem: 52150 +Epoch: [18] [1730/7269] eta: 0:40:07 lr: 0.000007 loss: 0.5407 (0.5249) time: 0.4803 data: 0.0002 max mem: 52150 +Epoch: [18] [1740/7269] eta: 0:40:05 lr: 0.000007 loss: 0.5407 (0.5249) time: 0.4350 data: 0.0002 max mem: 52150 +Epoch: [18] [1750/7269] eta: 0:39:58 lr: 0.000007 loss: 0.5324 (0.5250) time: 0.4356 data: 0.0002 max mem: 52150 +Epoch: [18] [1760/7269] eta: 0:39:52 lr: 0.000007 loss: 0.5597 (0.5252) time: 0.3731 data: 0.0002 max mem: 52150 +Epoch: [18] [1770/7269] eta: 0:39:46 lr: 0.000007 loss: 0.5490 (0.5253) time: 0.3733 data: 0.0002 max mem: 52150 +Epoch: [18] [1780/7269] eta: 0:39:43 lr: 0.000007 loss: 0.5147 (0.5252) time: 0.4223 data: 0.0002 max mem: 52150 +Epoch: [18] [1790/7269] eta: 0:39:36 lr: 0.000007 loss: 0.5326 (0.5253) time: 0.4211 data: 0.0002 max mem: 52150 +Epoch: [18] [1800/7269] eta: 0:39:34 lr: 0.000007 loss: 0.5455 (0.5254) time: 0.4319 data: 0.0002 max mem: 52150 +Epoch: [18] [1810/7269] eta: 0:39:30 lr: 0.000007 loss: 0.5437 (0.5255) time: 0.4799 data: 0.0002 max mem: 52150 +Epoch: [18] [1820/7269] eta: 0:39:24 lr: 0.000007 loss: 0.5437 (0.5257) time: 0.4194 data: 0.0002 max mem: 52150 +Epoch: [18] [1830/7269] eta: 0:39:22 lr: 0.000007 loss: 0.5396 (0.5257) time: 0.4308 data: 0.0002 max mem: 52150 +Epoch: [18] [1840/7269] eta: 0:39:15 lr: 0.000007 loss: 0.5033 (0.5255) time: 0.4310 data: 0.0002 max mem: 52150 +Epoch: [18] [1850/7269] eta: 0:39:12 lr: 0.000007 loss: 0.4999 (0.5253) time: 0.4156 data: 0.0002 max mem: 52150 +Epoch: [18] [1860/7269] eta: 0:39:06 lr: 0.000007 loss: 0.5308 (0.5254) time: 0.4156 data: 0.0002 max mem: 52150 +Epoch: [18] [1870/7269] eta: 0:39:00 lr: 0.000007 loss: 0.5715 (0.5256) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [18] [1880/7269] eta: 0:38:56 lr: 0.000007 loss: 0.5354 (0.5255) time: 0.4171 data: 0.0002 max mem: 52150 +Epoch: [18] [1890/7269] eta: 0:38:50 lr: 0.000007 loss: 0.4986 (0.5256) time: 0.4170 data: 0.0002 max mem: 52150 +Epoch: [18] [1900/7269] eta: 0:38:47 lr: 0.000007 loss: 0.5122 (0.5256) time: 0.4196 data: 0.0002 max mem: 52150 +Epoch: [18] [1910/7269] eta: 0:38:44 lr: 0.000007 loss: 0.5344 (0.5258) time: 0.4799 data: 0.0002 max mem: 52150 +Epoch: [18] [1920/7269] eta: 0:38:41 lr: 0.000007 loss: 0.5611 (0.5257) time: 0.4911 data: 0.0002 max mem: 52150 +Epoch: [18] [1930/7269] eta: 0:38:38 lr: 0.000007 loss: 0.5460 (0.5258) time: 0.4775 data: 0.0002 max mem: 52150 +Epoch: [18] [1940/7269] eta: 0:38:32 lr: 0.000007 loss: 0.5325 (0.5257) time: 0.4188 data: 0.0002 max mem: 52150 +Epoch: [18] [1950/7269] eta: 0:38:26 lr: 0.000007 loss: 0.4827 (0.5255) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [18] [1960/7269] eta: 0:38:20 lr: 0.000007 loss: 0.5031 (0.5257) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [18] [1970/7269] eta: 0:38:14 lr: 0.000007 loss: 0.5226 (0.5257) time: 0.3729 data: 0.0002 max mem: 52150 +Epoch: [18] [1980/7269] eta: 0:38:08 lr: 0.000007 loss: 0.5197 (0.5255) time: 0.3733 data: 0.0002 max mem: 52150 +Epoch: [18] [1990/7269] eta: 0:38:04 lr: 0.000007 loss: 0.5142 (0.5255) time: 0.4198 data: 0.0002 max mem: 52150 +Epoch: [18] [2000/7269] eta: 0:37:58 lr: 0.000007 loss: 0.5296 (0.5255) time: 0.4187 data: 0.0002 max mem: 52150 +Epoch: [18] [2010/7269] eta: 0:37:55 lr: 0.000007 loss: 0.5300 (0.5254) time: 0.4224 data: 0.0002 max mem: 52150 +Epoch: [18] [2020/7269] eta: 0:37:49 lr: 0.000007 loss: 0.5300 (0.5253) time: 0.4223 data: 0.0002 max mem: 52150 +Epoch: [18] [2030/7269] eta: 0:37:46 lr: 0.000007 loss: 0.5740 (0.5256) time: 0.4171 data: 0.0002 max mem: 52150 +Epoch: [18] [2040/7269] eta: 0:37:45 lr: 0.000007 loss: 0.5308 (0.5253) time: 0.5225 data: 0.0002 max mem: 52150 +Epoch: [18] [2050/7269] eta: 0:37:45 lr: 0.000007 loss: 0.4982 (0.5254) time: 0.5842 data: 0.0002 max mem: 52150 +Epoch: [18] [2060/7269] eta: 0:37:42 lr: 0.000007 loss: 0.5482 (0.5256) time: 0.5396 data: 0.0002 max mem: 52150 +Epoch: [18] [2070/7269] eta: 0:37:36 lr: 0.000007 loss: 0.5415 (0.5255) time: 0.4320 data: 0.0002 max mem: 52150 +Epoch: [18] [2080/7269] eta: 0:37:30 lr: 0.000007 loss: 0.5291 (0.5255) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [18] [2090/7269] eta: 0:37:24 lr: 0.000007 loss: 0.5421 (0.5255) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [18] [2100/7269] eta: 0:37:21 lr: 0.000007 loss: 0.5421 (0.5255) time: 0.4229 data: 0.0002 max mem: 52150 +Epoch: [18] [2110/7269] eta: 0:37:15 lr: 0.000007 loss: 0.5331 (0.5255) time: 0.4231 data: 0.0002 max mem: 52150 +Epoch: [18] [2120/7269] eta: 0:37:09 lr: 0.000007 loss: 0.5024 (0.5254) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [18] [2130/7269] eta: 0:37:06 lr: 0.000007 loss: 0.5260 (0.5255) time: 0.4324 data: 0.0002 max mem: 52150 +Epoch: [18] [2140/7269] eta: 0:37:01 lr: 0.000007 loss: 0.5244 (0.5255) time: 0.4328 data: 0.0002 max mem: 52150 +Epoch: [18] [2150/7269] eta: 0:36:57 lr: 0.000007 loss: 0.5244 (0.5255) time: 0.4188 data: 0.0002 max mem: 52150 +Epoch: [18] [2160/7269] eta: 0:36:56 lr: 0.000007 loss: 0.5257 (0.5255) time: 0.5122 data: 0.0002 max mem: 52150 +Epoch: [18] [2170/7269] eta: 0:36:50 lr: 0.000007 loss: 0.5257 (0.5254) time: 0.4659 data: 0.0002 max mem: 52150 +Epoch: [18] [2180/7269] eta: 0:36:46 lr: 0.000007 loss: 0.5314 (0.5254) time: 0.4173 data: 0.0002 max mem: 52150 +Epoch: [18] [2190/7269] eta: 0:36:40 lr: 0.000007 loss: 0.5244 (0.5254) time: 0.4171 data: 0.0002 max mem: 52150 +Epoch: [18] [2200/7269] eta: 0:36:35 lr: 0.000007 loss: 0.5519 (0.5255) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [18] [2210/7269] eta: 0:36:32 lr: 0.000007 loss: 0.5504 (0.5255) time: 0.4314 data: 0.0002 max mem: 52150 +Epoch: [18] [2220/7269] eta: 0:36:30 lr: 0.000007 loss: 0.5118 (0.5254) time: 0.5220 data: 0.0002 max mem: 52150 +Epoch: [18] [2230/7269] eta: 0:36:27 lr: 0.000007 loss: 0.4890 (0.5253) time: 0.5251 data: 0.0001 max mem: 52150 +Epoch: [18] [2240/7269] eta: 0:36:24 lr: 0.000007 loss: 0.5200 (0.5254) time: 0.4954 data: 0.0002 max mem: 52150 +Epoch: [18] [2250/7269] eta: 0:36:18 lr: 0.000007 loss: 0.5442 (0.5254) time: 0.4328 data: 0.0002 max mem: 52150 +Epoch: [18] [2260/7269] eta: 0:36:15 lr: 0.000007 loss: 0.5442 (0.5254) time: 0.4321 data: 0.0002 max mem: 52150 +Epoch: [18] [2270/7269] eta: 0:36:10 lr: 0.000007 loss: 0.5414 (0.5254) time: 0.4326 data: 0.0002 max mem: 52150 +Epoch: [18] [2280/7269] eta: 0:36:04 lr: 0.000007 loss: 0.5304 (0.5254) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [18] [2290/7269] eta: 0:35:58 lr: 0.000007 loss: 0.5535 (0.5257) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [18] [2300/7269] eta: 0:35:53 lr: 0.000007 loss: 0.5504 (0.5256) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [18] [2310/7269] eta: 0:35:50 lr: 0.000007 loss: 0.5103 (0.5256) time: 0.4349 data: 0.0002 max mem: 52150 +Epoch: [18] [2320/7269] eta: 0:35:44 lr: 0.000007 loss: 0.5103 (0.5254) time: 0.4343 data: 0.0001 max mem: 52150 +Epoch: [18] [2330/7269] eta: 0:35:38 lr: 0.000007 loss: 0.5092 (0.5256) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [18] [2340/7269] eta: 0:35:33 lr: 0.000007 loss: 0.5634 (0.5257) time: 0.3712 data: 0.0002 max mem: 52150 +Epoch: [18] [2350/7269] eta: 0:35:29 lr: 0.000007 loss: 0.5369 (0.5256) time: 0.4184 data: 0.0002 max mem: 52150 +Epoch: [18] [2360/7269] eta: 0:35:28 lr: 0.000007 loss: 0.5468 (0.5257) time: 0.5257 data: 0.0002 max mem: 52150 +Epoch: [18] [2370/7269] eta: 0:35:22 lr: 0.000007 loss: 0.5243 (0.5255) time: 0.4790 data: 0.0002 max mem: 52150 +Epoch: [18] [2380/7269] eta: 0:35:19 lr: 0.000007 loss: 0.5152 (0.5254) time: 0.4332 data: 0.0002 max mem: 52150 +Epoch: [18] [2390/7269] eta: 0:35:16 lr: 0.000007 loss: 0.5227 (0.5254) time: 0.4800 data: 0.0001 max mem: 52150 +Epoch: [18] [2400/7269] eta: 0:35:10 lr: 0.000007 loss: 0.5338 (0.5255) time: 0.4190 data: 0.0002 max mem: 52150 +Epoch: [18] [2410/7269] eta: 0:35:04 lr: 0.000007 loss: 0.5338 (0.5254) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [18] [2420/7269] eta: 0:34:59 lr: 0.000007 loss: 0.5138 (0.5253) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [18] [2430/7269] eta: 0:34:58 lr: 0.000007 loss: 0.5138 (0.5252) time: 0.4936 data: 0.0002 max mem: 52150 +Epoch: [18] [2440/7269] eta: 0:34:57 lr: 0.000007 loss: 0.5375 (0.5252) time: 0.6041 data: 0.0002 max mem: 52150 +Epoch: [18] [2450/7269] eta: 0:34:51 lr: 0.000007 loss: 0.5239 (0.5251) time: 0.4824 data: 0.0002 max mem: 52150 +Epoch: [18] [2460/7269] eta: 0:34:46 lr: 0.000007 loss: 0.5504 (0.5252) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [18] [2470/7269] eta: 0:34:40 lr: 0.000007 loss: 0.5504 (0.5253) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [18] [2480/7269] eta: 0:34:35 lr: 0.000007 loss: 0.5278 (0.5253) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [18] [2490/7269] eta: 0:34:29 lr: 0.000007 loss: 0.5192 (0.5252) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [18] [2500/7269] eta: 0:34:24 lr: 0.000007 loss: 0.5430 (0.5255) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [18] [2510/7269] eta: 0:34:18 lr: 0.000007 loss: 0.5627 (0.5255) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [18] [2520/7269] eta: 0:34:15 lr: 0.000007 loss: 0.5410 (0.5255) time: 0.4321 data: 0.0002 max mem: 52150 +Epoch: [18] [2530/7269] eta: 0:34:13 lr: 0.000007 loss: 0.5410 (0.5255) time: 0.5270 data: 0.0002 max mem: 52150 +Epoch: [18] [2540/7269] eta: 0:34:08 lr: 0.000007 loss: 0.5622 (0.5256) time: 0.4663 data: 0.0002 max mem: 52150 +Epoch: [18] [2550/7269] eta: 0:34:06 lr: 0.000007 loss: 0.5253 (0.5254) time: 0.4812 data: 0.0002 max mem: 52150 +Epoch: [18] [2560/7269] eta: 0:34:02 lr: 0.000007 loss: 0.5083 (0.5254) time: 0.5303 data: 0.0002 max mem: 52150 +Epoch: [18] [2570/7269] eta: 0:33:59 lr: 0.000007 loss: 0.5186 (0.5253) time: 0.4666 data: 0.0002 max mem: 52150 +Epoch: [18] [2580/7269] eta: 0:33:55 lr: 0.000007 loss: 0.4819 (0.5252) time: 0.4666 data: 0.0002 max mem: 52150 +Epoch: [18] [2590/7269] eta: 0:33:50 lr: 0.000007 loss: 0.4900 (0.5253) time: 0.4212 data: 0.0002 max mem: 52150 +Epoch: [18] [2600/7269] eta: 0:33:44 lr: 0.000007 loss: 0.5444 (0.5253) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [18] [2610/7269] eta: 0:33:39 lr: 0.000007 loss: 0.5212 (0.5253) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [18] [2620/7269] eta: 0:33:37 lr: 0.000007 loss: 0.5155 (0.5252) time: 0.4778 data: 0.0002 max mem: 52150 +Epoch: [18] [2630/7269] eta: 0:33:34 lr: 0.000007 loss: 0.4917 (0.5250) time: 0.5389 data: 0.0001 max mem: 52150 +Epoch: [18] [2640/7269] eta: 0:33:28 lr: 0.000007 loss: 0.4917 (0.5249) time: 0.4335 data: 0.0002 max mem: 52150 +Epoch: [18] [2650/7269] eta: 0:33:25 lr: 0.000007 loss: 0.4914 (0.5248) time: 0.4202 data: 0.0002 max mem: 52150 +Epoch: [18] [2660/7269] eta: 0:33:19 lr: 0.000007 loss: 0.5147 (0.5247) time: 0.4196 data: 0.0002 max mem: 52150 +Epoch: [18] [2670/7269] eta: 0:33:15 lr: 0.000007 loss: 0.5302 (0.5248) time: 0.4199 data: 0.0002 max mem: 52150 +Epoch: [18] [2680/7269] eta: 0:33:12 lr: 0.000007 loss: 0.5472 (0.5248) time: 0.4665 data: 0.0002 max mem: 52150 +Epoch: [18] [2690/7269] eta: 0:33:06 lr: 0.000007 loss: 0.5409 (0.5249) time: 0.4181 data: 0.0002 max mem: 52150 +Epoch: [18] [2700/7269] eta: 0:33:04 lr: 0.000007 loss: 0.5409 (0.5249) time: 0.4658 data: 0.0002 max mem: 52150 +Epoch: [18] [2710/7269] eta: 0:32:59 lr: 0.000007 loss: 0.5256 (0.5249) time: 0.4654 data: 0.0002 max mem: 52150 +Epoch: [18] [2720/7269] eta: 0:32:53 lr: 0.000007 loss: 0.5335 (0.5249) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [18] [2730/7269] eta: 0:32:49 lr: 0.000007 loss: 0.5367 (0.5250) time: 0.4167 data: 0.0002 max mem: 52150 +Epoch: [18] [2740/7269] eta: 0:32:44 lr: 0.000007 loss: 0.5685 (0.5252) time: 0.4169 data: 0.0002 max mem: 52150 +Epoch: [18] [2750/7269] eta: 0:32:39 lr: 0.000007 loss: 0.5549 (0.5252) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [18] [2760/7269] eta: 0:32:33 lr: 0.000007 loss: 0.5189 (0.5251) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [18] [2770/7269] eta: 0:32:30 lr: 0.000007 loss: 0.5531 (0.5253) time: 0.4322 data: 0.0002 max mem: 52150 +Epoch: [18] [2780/7269] eta: 0:32:25 lr: 0.000007 loss: 0.5325 (0.5253) time: 0.4325 data: 0.0002 max mem: 52150 +Epoch: [18] [2790/7269] eta: 0:32:19 lr: 0.000007 loss: 0.5142 (0.5252) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [18] [2800/7269] eta: 0:32:14 lr: 0.000007 loss: 0.5093 (0.5252) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [18] [2810/7269] eta: 0:32:11 lr: 0.000007 loss: 0.5083 (0.5252) time: 0.4361 data: 0.0002 max mem: 52150 +Epoch: [18] [2820/7269] eta: 0:32:11 lr: 0.000007 loss: 0.5316 (0.5253) time: 0.6097 data: 0.0002 max mem: 52150 +Epoch: [18] [2830/7269] eta: 0:32:07 lr: 0.000007 loss: 0.5295 (0.5252) time: 0.5909 data: 0.0002 max mem: 52150 +Epoch: [18] [2840/7269] eta: 0:32:02 lr: 0.000007 loss: 0.5357 (0.5253) time: 0.4173 data: 0.0002 max mem: 52150 +Epoch: [18] [2850/7269] eta: 0:31:56 lr: 0.000007 loss: 0.5405 (0.5253) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [18] [2860/7269] eta: 0:31:51 lr: 0.000007 loss: 0.5405 (0.5254) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [18] [2870/7269] eta: 0:31:47 lr: 0.000007 loss: 0.5277 (0.5254) time: 0.4249 data: 0.0002 max mem: 52150 +Epoch: [18] [2880/7269] eta: 0:31:42 lr: 0.000007 loss: 0.5068 (0.5254) time: 0.4247 data: 0.0002 max mem: 52150 +Epoch: [18] [2890/7269] eta: 0:31:38 lr: 0.000007 loss: 0.5368 (0.5254) time: 0.4173 data: 0.0002 max mem: 52150 +Epoch: [18] [2900/7269] eta: 0:31:33 lr: 0.000007 loss: 0.5262 (0.5254) time: 0.4173 data: 0.0002 max mem: 52150 +Epoch: [18] [2910/7269] eta: 0:31:28 lr: 0.000007 loss: 0.5118 (0.5253) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [18] [2920/7269] eta: 0:31:22 lr: 0.000007 loss: 0.5209 (0.5253) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [18] [2930/7269] eta: 0:31:17 lr: 0.000007 loss: 0.5618 (0.5254) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [18] [2940/7269] eta: 0:31:12 lr: 0.000007 loss: 0.5837 (0.5256) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [18] [2950/7269] eta: 0:31:07 lr: 0.000007 loss: 0.5825 (0.5256) time: 0.3731 data: 0.0002 max mem: 52150 +Epoch: [18] [2960/7269] eta: 0:31:03 lr: 0.000007 loss: 0.5384 (0.5257) time: 0.4328 data: 0.0002 max mem: 52150 +Epoch: [18] [2970/7269] eta: 0:31:00 lr: 0.000007 loss: 0.5644 (0.5258) time: 0.4954 data: 0.0002 max mem: 52150 +Epoch: [18] [2980/7269] eta: 0:30:55 lr: 0.000007 loss: 0.4916 (0.5255) time: 0.4351 data: 0.0002 max mem: 52150 +Epoch: [18] [2990/7269] eta: 0:30:50 lr: 0.000007 loss: 0.4911 (0.5255) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [18] [3000/7269] eta: 0:30:44 lr: 0.000007 loss: 0.5451 (0.5255) time: 0.3711 data: 0.0001 max mem: 52150 +Epoch: [18] [3010/7269] eta: 0:30:42 lr: 0.000007 loss: 0.5249 (0.5255) time: 0.4792 data: 0.0002 max mem: 52150 +Epoch: [18] [3020/7269] eta: 0:30:39 lr: 0.000007 loss: 0.5249 (0.5255) time: 0.5423 data: 0.0002 max mem: 52150 +Epoch: [18] [3030/7269] eta: 0:30:35 lr: 0.000007 loss: 0.5471 (0.5255) time: 0.4828 data: 0.0002 max mem: 52150 +Epoch: [18] [3040/7269] eta: 0:30:31 lr: 0.000007 loss: 0.5431 (0.5255) time: 0.4645 data: 0.0002 max mem: 52150 +Epoch: [18] [3050/7269] eta: 0:30:27 lr: 0.000007 loss: 0.5421 (0.5256) time: 0.4762 data: 0.0002 max mem: 52150 +Epoch: [18] [3060/7269] eta: 0:30:22 lr: 0.000007 loss: 0.5405 (0.5256) time: 0.4319 data: 0.0002 max mem: 52150 +Epoch: [18] [3070/7269] eta: 0:30:19 lr: 0.000007 loss: 0.5301 (0.5257) time: 0.4324 data: 0.0002 max mem: 52150 +Epoch: [18] [3080/7269] eta: 0:30:14 lr: 0.000007 loss: 0.5379 (0.5257) time: 0.4326 data: 0.0002 max mem: 52150 +Epoch: [18] [3090/7269] eta: 0:30:12 lr: 0.000007 loss: 0.5281 (0.5256) time: 0.4855 data: 0.0002 max mem: 52150 +Epoch: [18] [3100/7269] eta: 0:30:06 lr: 0.000007 loss: 0.5281 (0.5256) time: 0.4856 data: 0.0002 max mem: 52150 +Epoch: [18] [3110/7269] eta: 0:30:01 lr: 0.000007 loss: 0.5043 (0.5255) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [18] [3120/7269] eta: 0:29:58 lr: 0.000007 loss: 0.5018 (0.5255) time: 0.4361 data: 0.0002 max mem: 52150 +Epoch: [18] [3130/7269] eta: 0:29:54 lr: 0.000007 loss: 0.5433 (0.5255) time: 0.4844 data: 0.0002 max mem: 52150 +Epoch: [18] [3140/7269] eta: 0:29:49 lr: 0.000007 loss: 0.5433 (0.5254) time: 0.4199 data: 0.0002 max mem: 52150 +Epoch: [18] [3150/7269] eta: 0:29:44 lr: 0.000007 loss: 0.5330 (0.5255) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [18] [3160/7269] eta: 0:29:40 lr: 0.000007 loss: 0.5330 (0.5256) time: 0.4164 data: 0.0002 max mem: 52150 +Epoch: [18] [3170/7269] eta: 0:29:37 lr: 0.000007 loss: 0.5727 (0.5257) time: 0.5062 data: 0.0002 max mem: 52150 +Epoch: [18] [3180/7269] eta: 0:29:32 lr: 0.000007 loss: 0.5304 (0.5256) time: 0.4611 data: 0.0002 max mem: 52150 +Epoch: [18] [3190/7269] eta: 0:29:28 lr: 0.000007 loss: 0.5095 (0.5256) time: 0.4333 data: 0.0002 max mem: 52150 +Epoch: [18] [3200/7269] eta: 0:29:24 lr: 0.000007 loss: 0.5198 (0.5256) time: 0.4782 data: 0.0002 max mem: 52150 +Epoch: [18] [3210/7269] eta: 0:29:19 lr: 0.000007 loss: 0.5423 (0.5256) time: 0.4173 data: 0.0002 max mem: 52150 +Epoch: [18] [3220/7269] eta: 0:29:14 lr: 0.000007 loss: 0.5005 (0.5254) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [18] [3230/7269] eta: 0:29:09 lr: 0.000007 loss: 0.5005 (0.5254) time: 0.3714 data: 0.0001 max mem: 52150 +Epoch: [18] [3240/7269] eta: 0:29:05 lr: 0.000007 loss: 0.5126 (0.5254) time: 0.4321 data: 0.0001 max mem: 52150 +Epoch: [18] [3250/7269] eta: 0:29:00 lr: 0.000007 loss: 0.5007 (0.5253) time: 0.4326 data: 0.0002 max mem: 52150 +Epoch: [18] [3260/7269] eta: 0:28:55 lr: 0.000007 loss: 0.5265 (0.5253) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [18] [3270/7269] eta: 0:28:53 lr: 0.000007 loss: 0.4965 (0.5252) time: 0.4939 data: 0.0002 max mem: 52150 +Epoch: [18] [3280/7269] eta: 0:28:48 lr: 0.000007 loss: 0.5000 (0.5253) time: 0.4942 data: 0.0002 max mem: 52150 +Epoch: [18] [3290/7269] eta: 0:28:44 lr: 0.000007 loss: 0.5350 (0.5253) time: 0.4231 data: 0.0002 max mem: 52150 +Epoch: [18] [3300/7269] eta: 0:28:39 lr: 0.000007 loss: 0.5370 (0.5254) time: 0.4228 data: 0.0001 max mem: 52150 +Epoch: [18] [3310/7269] eta: 0:28:34 lr: 0.000007 loss: 0.4877 (0.5252) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [18] [3320/7269] eta: 0:28:30 lr: 0.000007 loss: 0.4646 (0.5251) time: 0.4198 data: 0.0002 max mem: 52150 +Epoch: [18] [3330/7269] eta: 0:28:26 lr: 0.000007 loss: 0.5140 (0.5251) time: 0.4829 data: 0.0001 max mem: 52150 +Epoch: [18] [3340/7269] eta: 0:28:21 lr: 0.000007 loss: 0.5140 (0.5250) time: 0.4351 data: 0.0002 max mem: 52150 +Epoch: [18] [3350/7269] eta: 0:28:18 lr: 0.000007 loss: 0.4831 (0.5250) time: 0.4209 data: 0.0002 max mem: 52150 +Epoch: [18] [3360/7269] eta: 0:28:12 lr: 0.000007 loss: 0.5202 (0.5250) time: 0.4210 data: 0.0002 max mem: 52150 +Epoch: [18] [3370/7269] eta: 0:28:07 lr: 0.000007 loss: 0.5476 (0.5250) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [18] [3380/7269] eta: 0:28:02 lr: 0.000007 loss: 0.5301 (0.5249) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [18] [3390/7269] eta: 0:27:57 lr: 0.000007 loss: 0.5073 (0.5249) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [18] [3400/7269] eta: 0:27:52 lr: 0.000007 loss: 0.5591 (0.5251) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [18] [3410/7269] eta: 0:27:48 lr: 0.000007 loss: 0.5960 (0.5252) time: 0.4174 data: 0.0002 max mem: 52150 +Epoch: [18] [3420/7269] eta: 0:27:43 lr: 0.000007 loss: 0.5687 (0.5253) time: 0.4171 data: 0.0002 max mem: 52150 +Epoch: [18] [3430/7269] eta: 0:27:40 lr: 0.000007 loss: 0.5449 (0.5254) time: 0.4314 data: 0.0002 max mem: 52150 +Epoch: [18] [3440/7269] eta: 0:27:37 lr: 0.000007 loss: 0.5601 (0.5254) time: 0.5226 data: 0.0002 max mem: 52150 +Epoch: [18] [3450/7269] eta: 0:27:33 lr: 0.000007 loss: 0.5188 (0.5254) time: 0.5108 data: 0.0002 max mem: 52150 +Epoch: [18] [3460/7269] eta: 0:27:28 lr: 0.000007 loss: 0.5188 (0.5254) time: 0.4191 data: 0.0002 max mem: 52150 +Epoch: [18] [3470/7269] eta: 0:27:23 lr: 0.000007 loss: 0.5274 (0.5254) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [18] [3480/7269] eta: 0:27:19 lr: 0.000007 loss: 0.5274 (0.5254) time: 0.4314 data: 0.0002 max mem: 52150 +Epoch: [18] [3490/7269] eta: 0:27:14 lr: 0.000007 loss: 0.5518 (0.5255) time: 0.4314 data: 0.0002 max mem: 52150 +Epoch: [18] [3500/7269] eta: 0:27:09 lr: 0.000007 loss: 0.5508 (0.5255) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [18] [3510/7269] eta: 0:27:05 lr: 0.000007 loss: 0.4883 (0.5254) time: 0.4318 data: 0.0002 max mem: 52150 +Epoch: [18] [3520/7269] eta: 0:27:03 lr: 0.000007 loss: 0.5028 (0.5254) time: 0.5460 data: 0.0002 max mem: 52150 +Epoch: [18] [3530/7269] eta: 0:26:58 lr: 0.000007 loss: 0.5249 (0.5254) time: 0.4859 data: 0.0002 max mem: 52150 +Epoch: [18] [3540/7269] eta: 0:26:54 lr: 0.000007 loss: 0.5303 (0.5254) time: 0.4322 data: 0.0002 max mem: 52150 +Epoch: [18] [3550/7269] eta: 0:26:49 lr: 0.000007 loss: 0.5397 (0.5254) time: 0.4322 data: 0.0002 max mem: 52150 +Epoch: [18] [3560/7269] eta: 0:26:45 lr: 0.000007 loss: 0.5457 (0.5254) time: 0.4202 data: 0.0002 max mem: 52150 +Epoch: [18] [3570/7269] eta: 0:26:40 lr: 0.000007 loss: 0.5632 (0.5254) time: 0.4203 data: 0.0002 max mem: 52150 +Epoch: [18] [3580/7269] eta: 0:26:35 lr: 0.000007 loss: 0.5601 (0.5254) time: 0.3727 data: 0.0002 max mem: 52150 +Epoch: [18] [3590/7269] eta: 0:26:30 lr: 0.000007 loss: 0.5334 (0.5254) time: 0.3733 data: 0.0002 max mem: 52150 +Epoch: [18] [3600/7269] eta: 0:26:27 lr: 0.000007 loss: 0.5334 (0.5254) time: 0.4214 data: 0.0002 max mem: 52150 +Epoch: [18] [3610/7269] eta: 0:26:22 lr: 0.000007 loss: 0.5375 (0.5253) time: 0.4212 data: 0.0002 max mem: 52150 +Epoch: [18] [3620/7269] eta: 0:26:17 lr: 0.000007 loss: 0.5124 (0.5253) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [18] [3630/7269] eta: 0:26:13 lr: 0.000007 loss: 0.5309 (0.5254) time: 0.4169 data: 0.0002 max mem: 52150 +Epoch: [18] [3640/7269] eta: 0:26:08 lr: 0.000007 loss: 0.5731 (0.5255) time: 0.4170 data: 0.0002 max mem: 52150 +Epoch: [18] [3650/7269] eta: 0:26:03 lr: 0.000007 loss: 0.5711 (0.5255) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [18] [3660/7269] eta: 0:25:58 lr: 0.000007 loss: 0.5092 (0.5254) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [18] [3670/7269] eta: 0:25:54 lr: 0.000007 loss: 0.5040 (0.5253) time: 0.4333 data: 0.0002 max mem: 52150 +Epoch: [18] [3680/7269] eta: 0:25:50 lr: 0.000007 loss: 0.5429 (0.5253) time: 0.4813 data: 0.0002 max mem: 52150 +Epoch: [18] [3690/7269] eta: 0:25:45 lr: 0.000007 loss: 0.5572 (0.5253) time: 0.4199 data: 0.0002 max mem: 52150 +Epoch: [18] [3700/7269] eta: 0:25:41 lr: 0.000007 loss: 0.5022 (0.5252) time: 0.4162 data: 0.0002 max mem: 52150 +Epoch: [18] [3710/7269] eta: 0:25:36 lr: 0.000007 loss: 0.5002 (0.5251) time: 0.4163 data: 0.0002 max mem: 52150 +Epoch: [18] [3720/7269] eta: 0:25:32 lr: 0.000007 loss: 0.4586 (0.5250) time: 0.4181 data: 0.0002 max mem: 52150 +Epoch: [18] [3730/7269] eta: 0:25:29 lr: 0.000007 loss: 0.4680 (0.5249) time: 0.4784 data: 0.0002 max mem: 52150 +Epoch: [18] [3740/7269] eta: 0:25:25 lr: 0.000007 loss: 0.5297 (0.5249) time: 0.4836 data: 0.0002 max mem: 52150 +Epoch: [18] [3750/7269] eta: 0:25:21 lr: 0.000007 loss: 0.5402 (0.5249) time: 0.4851 data: 0.0002 max mem: 52150 +Epoch: [18] [3760/7269] eta: 0:25:16 lr: 0.000007 loss: 0.5262 (0.5248) time: 0.4336 data: 0.0002 max mem: 52150 +Epoch: [18] [3770/7269] eta: 0:25:13 lr: 0.000007 loss: 0.5217 (0.5248) time: 0.4820 data: 0.0002 max mem: 52150 +Epoch: [18] [3780/7269] eta: 0:25:08 lr: 0.000007 loss: 0.5294 (0.5249) time: 0.4823 data: 0.0002 max mem: 52150 +Epoch: [18] [3790/7269] eta: 0:25:04 lr: 0.000007 loss: 0.5327 (0.5249) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [18] [3800/7269] eta: 0:25:00 lr: 0.000007 loss: 0.5318 (0.5249) time: 0.4344 data: 0.0002 max mem: 52150 +Epoch: [18] [3810/7269] eta: 0:24:56 lr: 0.000007 loss: 0.5563 (0.5250) time: 0.4803 data: 0.0002 max mem: 52150 +Epoch: [18] [3820/7269] eta: 0:24:52 lr: 0.000007 loss: 0.5337 (0.5250) time: 0.4643 data: 0.0002 max mem: 52150 +Epoch: [18] [3830/7269] eta: 0:24:47 lr: 0.000007 loss: 0.5164 (0.5250) time: 0.4185 data: 0.0002 max mem: 52150 +Epoch: [18] [3840/7269] eta: 0:24:42 lr: 0.000007 loss: 0.5218 (0.5249) time: 0.3725 data: 0.0001 max mem: 52150 +Epoch: [18] [3850/7269] eta: 0:24:37 lr: 0.000007 loss: 0.5218 (0.5249) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [18] [3860/7269] eta: 0:24:32 lr: 0.000007 loss: 0.5302 (0.5249) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [18] [3870/7269] eta: 0:24:27 lr: 0.000007 loss: 0.5226 (0.5248) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [18] [3880/7269] eta: 0:24:23 lr: 0.000007 loss: 0.5175 (0.5248) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [18] [3890/7269] eta: 0:24:19 lr: 0.000007 loss: 0.5125 (0.5247) time: 0.4674 data: 0.0002 max mem: 52150 +Epoch: [18] [3900/7269] eta: 0:24:15 lr: 0.000007 loss: 0.5162 (0.5248) time: 0.4673 data: 0.0002 max mem: 52150 +Epoch: [18] [3910/7269] eta: 0:24:11 lr: 0.000007 loss: 0.5462 (0.5248) time: 0.4665 data: 0.0002 max mem: 52150 +Epoch: [18] [3920/7269] eta: 0:24:06 lr: 0.000007 loss: 0.5414 (0.5249) time: 0.4662 data: 0.0002 max mem: 52150 +Epoch: [18] [3930/7269] eta: 0:24:02 lr: 0.000007 loss: 0.5366 (0.5249) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [18] [3940/7269] eta: 0:23:58 lr: 0.000007 loss: 0.5152 (0.5248) time: 0.4240 data: 0.0002 max mem: 52150 +Epoch: [18] [3950/7269] eta: 0:23:54 lr: 0.000007 loss: 0.5244 (0.5248) time: 0.4723 data: 0.0001 max mem: 52150 +Epoch: [18] [3960/7269] eta: 0:23:50 lr: 0.000007 loss: 0.4968 (0.5247) time: 0.4643 data: 0.0002 max mem: 52150 +Epoch: [18] [3970/7269] eta: 0:23:46 lr: 0.000007 loss: 0.4676 (0.5247) time: 0.4623 data: 0.0002 max mem: 52150 +Epoch: [18] [3980/7269] eta: 0:23:42 lr: 0.000007 loss: 0.4955 (0.5246) time: 0.4813 data: 0.0002 max mem: 52150 +Epoch: [18] [3990/7269] eta: 0:23:38 lr: 0.000007 loss: 0.5363 (0.5246) time: 0.4824 data: 0.0002 max mem: 52150 +Epoch: [18] [4000/7269] eta: 0:23:33 lr: 0.000007 loss: 0.5503 (0.5247) time: 0.4195 data: 0.0002 max mem: 52150 +Epoch: [18] [4010/7269] eta: 0:23:28 lr: 0.000007 loss: 0.5500 (0.5246) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [18] [4020/7269] eta: 0:23:23 lr: 0.000007 loss: 0.5246 (0.5246) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [18] [4030/7269] eta: 0:23:19 lr: 0.000007 loss: 0.5084 (0.5246) time: 0.4200 data: 0.0002 max mem: 52150 +Epoch: [18] [4040/7269] eta: 0:23:14 lr: 0.000007 loss: 0.5180 (0.5246) time: 0.4205 data: 0.0002 max mem: 52150 +Epoch: [18] [4050/7269] eta: 0:23:10 lr: 0.000007 loss: 0.5587 (0.5247) time: 0.3729 data: 0.0002 max mem: 52150 +Epoch: [18] [4060/7269] eta: 0:23:05 lr: 0.000007 loss: 0.5742 (0.5248) time: 0.3735 data: 0.0002 max mem: 52150 +Epoch: [18] [4070/7269] eta: 0:23:02 lr: 0.000007 loss: 0.5514 (0.5248) time: 0.4783 data: 0.0001 max mem: 52150 +Epoch: [18] [4080/7269] eta: 0:22:58 lr: 0.000007 loss: 0.4785 (0.5246) time: 0.5260 data: 0.0001 max mem: 52150 +Epoch: [18] [4090/7269] eta: 0:22:53 lr: 0.000007 loss: 0.4853 (0.5246) time: 0.4203 data: 0.0001 max mem: 52150 +Epoch: [18] [4100/7269] eta: 0:22:48 lr: 0.000007 loss: 0.5189 (0.5245) time: 0.3726 data: 0.0001 max mem: 52150 +Epoch: [18] [4110/7269] eta: 0:22:44 lr: 0.000007 loss: 0.5040 (0.5245) time: 0.4193 data: 0.0002 max mem: 52150 +Epoch: [18] [4120/7269] eta: 0:22:39 lr: 0.000007 loss: 0.5040 (0.5245) time: 0.4183 data: 0.0002 max mem: 52150 +Epoch: [18] [4130/7269] eta: 0:22:35 lr: 0.000007 loss: 0.5248 (0.5245) time: 0.4313 data: 0.0002 max mem: 52150 +Epoch: [18] [4140/7269] eta: 0:22:32 lr: 0.000007 loss: 0.5248 (0.5245) time: 0.4942 data: 0.0001 max mem: 52150 +Epoch: [18] [4150/7269] eta: 0:22:27 lr: 0.000007 loss: 0.5101 (0.5244) time: 0.4348 data: 0.0001 max mem: 52150 +Epoch: [18] [4160/7269] eta: 0:22:22 lr: 0.000007 loss: 0.5153 (0.5244) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [18] [4170/7269] eta: 0:22:19 lr: 0.000007 loss: 0.5317 (0.5244) time: 0.4945 data: 0.0002 max mem: 52150 +Epoch: [18] [4180/7269] eta: 0:22:14 lr: 0.000007 loss: 0.5281 (0.5244) time: 0.4951 data: 0.0002 max mem: 52150 +Epoch: [18] [4190/7269] eta: 0:22:11 lr: 0.000007 loss: 0.5131 (0.5244) time: 0.4784 data: 0.0002 max mem: 52150 +Epoch: [18] [4200/7269] eta: 0:22:06 lr: 0.000007 loss: 0.5131 (0.5243) time: 0.4780 data: 0.0002 max mem: 52150 +Epoch: [18] [4210/7269] eta: 0:22:02 lr: 0.000007 loss: 0.5227 (0.5243) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [18] [4220/7269] eta: 0:21:57 lr: 0.000007 loss: 0.5237 (0.5243) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [18] [4230/7269] eta: 0:21:53 lr: 0.000007 loss: 0.5254 (0.5243) time: 0.4205 data: 0.0002 max mem: 52150 +Epoch: [18] [4240/7269] eta: 0:21:49 lr: 0.000007 loss: 0.4974 (0.5242) time: 0.4810 data: 0.0002 max mem: 52150 +Epoch: [18] [4250/7269] eta: 0:21:44 lr: 0.000007 loss: 0.4759 (0.5242) time: 0.4330 data: 0.0002 max mem: 52150 +Epoch: [18] [4260/7269] eta: 0:21:39 lr: 0.000007 loss: 0.4901 (0.5241) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [18] [4270/7269] eta: 0:21:35 lr: 0.000007 loss: 0.5272 (0.5241) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [18] [4280/7269] eta: 0:21:31 lr: 0.000007 loss: 0.5272 (0.5241) time: 0.4321 data: 0.0002 max mem: 52150 +Epoch: [18] [4290/7269] eta: 0:21:26 lr: 0.000007 loss: 0.5272 (0.5241) time: 0.4322 data: 0.0001 max mem: 52150 +Epoch: [18] [4300/7269] eta: 0:21:22 lr: 0.000007 loss: 0.5272 (0.5241) time: 0.4175 data: 0.0001 max mem: 52150 +Epoch: [18] [4310/7269] eta: 0:21:18 lr: 0.000007 loss: 0.5116 (0.5240) time: 0.4633 data: 0.0002 max mem: 52150 +Epoch: [18] [4320/7269] eta: 0:21:13 lr: 0.000007 loss: 0.5123 (0.5240) time: 0.4176 data: 0.0002 max mem: 52150 +Epoch: [18] [4330/7269] eta: 0:21:09 lr: 0.000007 loss: 0.5211 (0.5240) time: 0.4321 data: 0.0002 max mem: 52150 +Epoch: [18] [4340/7269] eta: 0:21:05 lr: 0.000007 loss: 0.4697 (0.5239) time: 0.4324 data: 0.0002 max mem: 52150 +Epoch: [18] [4350/7269] eta: 0:21:00 lr: 0.000007 loss: 0.4697 (0.5239) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [18] [4360/7269] eta: 0:20:55 lr: 0.000007 loss: 0.5274 (0.5239) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [18] [4370/7269] eta: 0:20:50 lr: 0.000007 loss: 0.5398 (0.5240) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [18] [4380/7269] eta: 0:20:46 lr: 0.000007 loss: 0.5549 (0.5240) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [18] [4390/7269] eta: 0:20:41 lr: 0.000007 loss: 0.5355 (0.5240) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [18] [4400/7269] eta: 0:20:38 lr: 0.000007 loss: 0.5223 (0.5240) time: 0.4676 data: 0.0002 max mem: 52150 +Epoch: [18] [4410/7269] eta: 0:20:33 lr: 0.000007 loss: 0.5075 (0.5240) time: 0.4678 data: 0.0002 max mem: 52150 +Epoch: [18] [4420/7269] eta: 0:20:29 lr: 0.000007 loss: 0.4961 (0.5239) time: 0.4173 data: 0.0002 max mem: 52150 +Epoch: [18] [4430/7269] eta: 0:20:25 lr: 0.000007 loss: 0.4961 (0.5239) time: 0.4658 data: 0.0002 max mem: 52150 +Epoch: [18] [4440/7269] eta: 0:20:20 lr: 0.000007 loss: 0.5265 (0.5239) time: 0.4203 data: 0.0002 max mem: 52150 +Epoch: [18] [4450/7269] eta: 0:20:15 lr: 0.000007 loss: 0.5547 (0.5240) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [18] [4460/7269] eta: 0:20:11 lr: 0.000007 loss: 0.5638 (0.5241) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [18] [4470/7269] eta: 0:20:06 lr: 0.000007 loss: 0.5547 (0.5242) time: 0.4195 data: 0.0002 max mem: 52150 +Epoch: [18] [4480/7269] eta: 0:20:02 lr: 0.000007 loss: 0.5408 (0.5242) time: 0.4195 data: 0.0002 max mem: 52150 +Epoch: [18] [4490/7269] eta: 0:19:58 lr: 0.000007 loss: 0.5063 (0.5241) time: 0.4349 data: 0.0002 max mem: 52150 +Epoch: [18] [4500/7269] eta: 0:19:53 lr: 0.000007 loss: 0.5278 (0.5242) time: 0.4349 data: 0.0002 max mem: 52150 +Epoch: [18] [4510/7269] eta: 0:19:49 lr: 0.000007 loss: 0.5349 (0.5241) time: 0.4178 data: 0.0002 max mem: 52150 +Epoch: [18] [4520/7269] eta: 0:19:44 lr: 0.000007 loss: 0.5349 (0.5242) time: 0.4178 data: 0.0002 max mem: 52150 +Epoch: [18] [4530/7269] eta: 0:19:40 lr: 0.000007 loss: 0.5437 (0.5242) time: 0.4152 data: 0.0002 max mem: 52150 +Epoch: [18] [4540/7269] eta: 0:19:36 lr: 0.000007 loss: 0.5324 (0.5243) time: 0.4606 data: 0.0002 max mem: 52150 +Epoch: [18] [4550/7269] eta: 0:19:31 lr: 0.000007 loss: 0.5108 (0.5242) time: 0.4167 data: 0.0002 max mem: 52150 +Epoch: [18] [4560/7269] eta: 0:19:28 lr: 0.000007 loss: 0.5028 (0.5242) time: 0.4352 data: 0.0002 max mem: 52150 +Epoch: [18] [4570/7269] eta: 0:19:24 lr: 0.000007 loss: 0.5407 (0.5242) time: 0.4860 data: 0.0002 max mem: 52150 +Epoch: [18] [4580/7269] eta: 0:19:20 lr: 0.000007 loss: 0.5476 (0.5242) time: 0.5181 data: 0.0002 max mem: 52150 +Epoch: [18] [4590/7269] eta: 0:19:15 lr: 0.000007 loss: 0.5149 (0.5242) time: 0.4672 data: 0.0002 max mem: 52150 +Epoch: [18] [4600/7269] eta: 0:19:11 lr: 0.000007 loss: 0.4919 (0.5241) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [18] [4610/7269] eta: 0:19:06 lr: 0.000007 loss: 0.5369 (0.5242) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [18] [4620/7269] eta: 0:19:01 lr: 0.000007 loss: 0.5355 (0.5241) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [18] [4630/7269] eta: 0:18:57 lr: 0.000007 loss: 0.5457 (0.5242) time: 0.4359 data: 0.0002 max mem: 52150 +Epoch: [18] [4640/7269] eta: 0:18:53 lr: 0.000007 loss: 0.5403 (0.5242) time: 0.4354 data: 0.0002 max mem: 52150 +Epoch: [18] [4650/7269] eta: 0:18:49 lr: 0.000007 loss: 0.5376 (0.5242) time: 0.4204 data: 0.0002 max mem: 52150 +Epoch: [18] [4660/7269] eta: 0:18:45 lr: 0.000007 loss: 0.5181 (0.5243) time: 0.5264 data: 0.0002 max mem: 52150 +Epoch: [18] [4670/7269] eta: 0:18:41 lr: 0.000007 loss: 0.4868 (0.5241) time: 0.4777 data: 0.0002 max mem: 52150 +Epoch: [18] [4680/7269] eta: 0:18:37 lr: 0.000007 loss: 0.4912 (0.5241) time: 0.4862 data: 0.0002 max mem: 52150 +Epoch: [18] [4690/7269] eta: 0:18:33 lr: 0.000007 loss: 0.5098 (0.5241) time: 0.4861 data: 0.0002 max mem: 52150 +Epoch: [18] [4700/7269] eta: 0:18:28 lr: 0.000007 loss: 0.5386 (0.5241) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [18] [4710/7269] eta: 0:18:24 lr: 0.000007 loss: 0.5068 (0.5240) time: 0.4326 data: 0.0002 max mem: 52150 +Epoch: [18] [4720/7269] eta: 0:18:19 lr: 0.000007 loss: 0.5169 (0.5241) time: 0.4332 data: 0.0002 max mem: 52150 +Epoch: [18] [4730/7269] eta: 0:18:17 lr: 0.000007 loss: 0.5427 (0.5241) time: 0.5552 data: 0.0002 max mem: 52150 +Epoch: [18] [4740/7269] eta: 0:18:14 lr: 0.000007 loss: 0.5643 (0.5243) time: 0.7046 data: 0.0002 max mem: 52150 +Epoch: [18] [4750/7269] eta: 0:18:09 lr: 0.000007 loss: 0.5477 (0.5242) time: 0.5219 data: 0.0002 max mem: 52150 +Epoch: [18] [4760/7269] eta: 0:18:04 lr: 0.000007 loss: 0.5222 (0.5242) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [18] [4770/7269] eta: 0:18:00 lr: 0.000007 loss: 0.5354 (0.5242) time: 0.4180 data: 0.0002 max mem: 52150 +Epoch: [18] [4780/7269] eta: 0:17:55 lr: 0.000007 loss: 0.5507 (0.5243) time: 0.4181 data: 0.0002 max mem: 52150 +Epoch: [18] [4790/7269] eta: 0:17:51 lr: 0.000007 loss: 0.5361 (0.5242) time: 0.3710 data: 0.0002 max mem: 52150 +Epoch: [18] [4800/7269] eta: 0:17:46 lr: 0.000007 loss: 0.5189 (0.5242) time: 0.3707 data: 0.0002 max mem: 52150 +Epoch: [18] [4810/7269] eta: 0:17:42 lr: 0.000007 loss: 0.5175 (0.5242) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [18] [4820/7269] eta: 0:17:37 lr: 0.000007 loss: 0.5043 (0.5241) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [18] [4830/7269] eta: 0:17:33 lr: 0.000007 loss: 0.4978 (0.5241) time: 0.4177 data: 0.0002 max mem: 52150 +Epoch: [18] [4840/7269] eta: 0:17:28 lr: 0.000007 loss: 0.4978 (0.5241) time: 0.4176 data: 0.0002 max mem: 52150 +Epoch: [18] [4850/7269] eta: 0:17:24 lr: 0.000007 loss: 0.5282 (0.5242) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [18] [4860/7269] eta: 0:17:19 lr: 0.000007 loss: 0.5688 (0.5242) time: 0.4202 data: 0.0002 max mem: 52150 +Epoch: [18] [4870/7269] eta: 0:17:15 lr: 0.000007 loss: 0.5404 (0.5242) time: 0.4206 data: 0.0002 max mem: 52150 +Epoch: [18] [4880/7269] eta: 0:17:10 lr: 0.000007 loss: 0.5300 (0.5243) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [18] [4890/7269] eta: 0:17:06 lr: 0.000007 loss: 0.5365 (0.5243) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [18] [4900/7269] eta: 0:17:01 lr: 0.000007 loss: 0.5416 (0.5244) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [18] [4910/7269] eta: 0:16:56 lr: 0.000007 loss: 0.5416 (0.5243) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [18] [4920/7269] eta: 0:16:52 lr: 0.000007 loss: 0.5446 (0.5244) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [18] [4930/7269] eta: 0:16:47 lr: 0.000007 loss: 0.5457 (0.5244) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [18] [4940/7269] eta: 0:16:43 lr: 0.000007 loss: 0.4959 (0.5244) time: 0.3728 data: 0.0002 max mem: 52150 +Epoch: [18] [4950/7269] eta: 0:16:38 lr: 0.000007 loss: 0.4959 (0.5243) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [18] [4960/7269] eta: 0:16:33 lr: 0.000007 loss: 0.5057 (0.5243) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [18] [4970/7269] eta: 0:16:29 lr: 0.000007 loss: 0.5182 (0.5243) time: 0.4162 data: 0.0002 max mem: 52150 +Epoch: [18] [4980/7269] eta: 0:16:25 lr: 0.000007 loss: 0.5176 (0.5243) time: 0.4648 data: 0.0002 max mem: 52150 +Epoch: [18] [4990/7269] eta: 0:16:21 lr: 0.000007 loss: 0.5267 (0.5243) time: 0.4202 data: 0.0002 max mem: 52150 +Epoch: [18] [5000/7269] eta: 0:16:17 lr: 0.000007 loss: 0.5437 (0.5244) time: 0.4317 data: 0.0002 max mem: 52150 +Epoch: [18] [5010/7269] eta: 0:16:12 lr: 0.000007 loss: 0.5327 (0.5243) time: 0.4326 data: 0.0002 max mem: 52150 +Epoch: [18] [5020/7269] eta: 0:16:07 lr: 0.000007 loss: 0.5498 (0.5244) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [18] [5030/7269] eta: 0:16:03 lr: 0.000007 loss: 0.5498 (0.5244) time: 0.4159 data: 0.0002 max mem: 52150 +Epoch: [18] [5040/7269] eta: 0:15:59 lr: 0.000007 loss: 0.5103 (0.5244) time: 0.4647 data: 0.0002 max mem: 52150 +Epoch: [18] [5050/7269] eta: 0:15:55 lr: 0.000007 loss: 0.5554 (0.5244) time: 0.4202 data: 0.0002 max mem: 52150 +Epoch: [18] [5060/7269] eta: 0:15:50 lr: 0.000007 loss: 0.5522 (0.5244) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [18] [5070/7269] eta: 0:15:45 lr: 0.000007 loss: 0.5362 (0.5244) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [18] [5080/7269] eta: 0:15:42 lr: 0.000007 loss: 0.5455 (0.5244) time: 0.4828 data: 0.0002 max mem: 52150 +Epoch: [18] [5090/7269] eta: 0:15:37 lr: 0.000007 loss: 0.5470 (0.5244) time: 0.4829 data: 0.0002 max mem: 52150 +Epoch: [18] [5100/7269] eta: 0:15:34 lr: 0.000007 loss: 0.5474 (0.5245) time: 0.4964 data: 0.0002 max mem: 52150 +Epoch: [18] [5110/7269] eta: 0:15:30 lr: 0.000007 loss: 0.5305 (0.5244) time: 0.5455 data: 0.0002 max mem: 52150 +Epoch: [18] [5120/7269] eta: 0:15:26 lr: 0.000007 loss: 0.5098 (0.5244) time: 0.4671 data: 0.0002 max mem: 52150 +Epoch: [18] [5130/7269] eta: 0:15:21 lr: 0.000007 loss: 0.5098 (0.5244) time: 0.4649 data: 0.0002 max mem: 52150 +Epoch: [18] [5140/7269] eta: 0:15:17 lr: 0.000007 loss: 0.5544 (0.5245) time: 0.4189 data: 0.0002 max mem: 52150 +Epoch: [18] [5150/7269] eta: 0:15:13 lr: 0.000007 loss: 0.5706 (0.5246) time: 0.4202 data: 0.0002 max mem: 52150 +Epoch: [18] [5160/7269] eta: 0:15:08 lr: 0.000007 loss: 0.5602 (0.5246) time: 0.4197 data: 0.0002 max mem: 52150 +Epoch: [18] [5170/7269] eta: 0:15:04 lr: 0.000007 loss: 0.5109 (0.5246) time: 0.4339 data: 0.0002 max mem: 52150 +Epoch: [18] [5180/7269] eta: 0:14:59 lr: 0.000007 loss: 0.5179 (0.5246) time: 0.4346 data: 0.0002 max mem: 52150 +Epoch: [18] [5190/7269] eta: 0:14:57 lr: 0.000007 loss: 0.5265 (0.5246) time: 0.5905 data: 0.0002 max mem: 52150 +Epoch: [18] [5200/7269] eta: 0:14:52 lr: 0.000007 loss: 0.5265 (0.5245) time: 0.6349 data: 0.0001 max mem: 52150 +Epoch: [18] [5210/7269] eta: 0:14:48 lr: 0.000007 loss: 0.5099 (0.5245) time: 0.4166 data: 0.0002 max mem: 52150 +Epoch: [18] [5220/7269] eta: 0:14:44 lr: 0.000007 loss: 0.5275 (0.5246) time: 0.4198 data: 0.0002 max mem: 52150 +Epoch: [18] [5230/7269] eta: 0:14:40 lr: 0.000007 loss: 0.5265 (0.5245) time: 0.4803 data: 0.0002 max mem: 52150 +Epoch: [18] [5240/7269] eta: 0:14:36 lr: 0.000007 loss: 0.5221 (0.5246) time: 0.4796 data: 0.0002 max mem: 52150 +Epoch: [18] [5250/7269] eta: 0:14:31 lr: 0.000007 loss: 0.5445 (0.5246) time: 0.4182 data: 0.0002 max mem: 52150 +Epoch: [18] [5260/7269] eta: 0:14:27 lr: 0.000007 loss: 0.5332 (0.5246) time: 0.4661 data: 0.0002 max mem: 52150 +Epoch: [18] [5270/7269] eta: 0:14:23 lr: 0.000007 loss: 0.5178 (0.5246) time: 0.4668 data: 0.0002 max mem: 52150 +Epoch: [18] [5280/7269] eta: 0:14:18 lr: 0.000007 loss: 0.5381 (0.5246) time: 0.4180 data: 0.0002 max mem: 52150 +Epoch: [18] [5290/7269] eta: 0:14:14 lr: 0.000007 loss: 0.5613 (0.5246) time: 0.4183 data: 0.0002 max mem: 52150 +Epoch: [18] [5300/7269] eta: 0:14:09 lr: 0.000007 loss: 0.5424 (0.5246) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [18] [5310/7269] eta: 0:14:05 lr: 0.000007 loss: 0.5632 (0.5247) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [18] [5320/7269] eta: 0:14:00 lr: 0.000007 loss: 0.5690 (0.5247) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [18] [5330/7269] eta: 0:13:56 lr: 0.000007 loss: 0.5177 (0.5247) time: 0.4172 data: 0.0002 max mem: 52150 +Epoch: [18] [5340/7269] eta: 0:13:52 lr: 0.000007 loss: 0.5297 (0.5247) time: 0.4657 data: 0.0002 max mem: 52150 +Epoch: [18] [5350/7269] eta: 0:13:48 lr: 0.000007 loss: 0.5411 (0.5247) time: 0.4811 data: 0.0002 max mem: 52150 +Epoch: [18] [5360/7269] eta: 0:13:43 lr: 0.000007 loss: 0.5584 (0.5248) time: 0.4326 data: 0.0002 max mem: 52150 +Epoch: [18] [5370/7269] eta: 0:13:39 lr: 0.000007 loss: 0.5438 (0.5248) time: 0.4196 data: 0.0002 max mem: 52150 +Epoch: [18] [5380/7269] eta: 0:13:35 lr: 0.000007 loss: 0.5204 (0.5248) time: 0.4201 data: 0.0002 max mem: 52150 +Epoch: [18] [5390/7269] eta: 0:13:30 lr: 0.000007 loss: 0.5418 (0.5248) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [18] [5400/7269] eta: 0:13:26 lr: 0.000007 loss: 0.5418 (0.5248) time: 0.4332 data: 0.0002 max mem: 52150 +Epoch: [18] [5410/7269] eta: 0:13:22 lr: 0.000007 loss: 0.5217 (0.5248) time: 0.4781 data: 0.0002 max mem: 52150 +Epoch: [18] [5420/7269] eta: 0:13:18 lr: 0.000007 loss: 0.4862 (0.5247) time: 0.4792 data: 0.0002 max mem: 52150 +Epoch: [18] [5430/7269] eta: 0:13:13 lr: 0.000007 loss: 0.5119 (0.5247) time: 0.4339 data: 0.0002 max mem: 52150 +Epoch: [18] [5440/7269] eta: 0:13:09 lr: 0.000007 loss: 0.5169 (0.5247) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [18] [5450/7269] eta: 0:13:04 lr: 0.000007 loss: 0.5346 (0.5247) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [18] [5460/7269] eta: 0:13:00 lr: 0.000007 loss: 0.5386 (0.5248) time: 0.3711 data: 0.0002 max mem: 52150 +Epoch: [18] [5470/7269] eta: 0:12:55 lr: 0.000007 loss: 0.5470 (0.5248) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [18] [5480/7269] eta: 0:12:51 lr: 0.000007 loss: 0.5345 (0.5248) time: 0.3733 data: 0.0002 max mem: 52150 +Epoch: [18] [5490/7269] eta: 0:12:46 lr: 0.000007 loss: 0.5257 (0.5248) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [18] [5500/7269] eta: 0:12:42 lr: 0.000007 loss: 0.5219 (0.5248) time: 0.4178 data: 0.0002 max mem: 52150 +Epoch: [18] [5510/7269] eta: 0:12:37 lr: 0.000007 loss: 0.5390 (0.5248) time: 0.4176 data: 0.0002 max mem: 52150 +Epoch: [18] [5520/7269] eta: 0:12:33 lr: 0.000007 loss: 0.5615 (0.5248) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [18] [5530/7269] eta: 0:12:28 lr: 0.000007 loss: 0.5364 (0.5248) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [18] [5540/7269] eta: 0:12:24 lr: 0.000007 loss: 0.5321 (0.5249) time: 0.4227 data: 0.0002 max mem: 52150 +Epoch: [18] [5550/7269] eta: 0:12:20 lr: 0.000007 loss: 0.5321 (0.5248) time: 0.4825 data: 0.0002 max mem: 52150 +Epoch: [18] [5560/7269] eta: 0:12:16 lr: 0.000007 loss: 0.5363 (0.5248) time: 0.4318 data: 0.0002 max mem: 52150 +Epoch: [18] [5570/7269] eta: 0:12:11 lr: 0.000007 loss: 0.5481 (0.5248) time: 0.4200 data: 0.0002 max mem: 52150 +Epoch: [18] [5580/7269] eta: 0:12:07 lr: 0.000007 loss: 0.5317 (0.5248) time: 0.5122 data: 0.0002 max mem: 52150 +Epoch: [18] [5590/7269] eta: 0:12:03 lr: 0.000007 loss: 0.5275 (0.5249) time: 0.5090 data: 0.0002 max mem: 52150 +Epoch: [18] [5600/7269] eta: 0:11:59 lr: 0.000007 loss: 0.5231 (0.5248) time: 0.4166 data: 0.0002 max mem: 52150 +Epoch: [18] [5610/7269] eta: 0:11:54 lr: 0.000007 loss: 0.5159 (0.5248) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [18] [5620/7269] eta: 0:11:50 lr: 0.000007 loss: 0.5390 (0.5248) time: 0.4164 data: 0.0002 max mem: 52150 +Epoch: [18] [5630/7269] eta: 0:11:46 lr: 0.000007 loss: 0.5314 (0.5248) time: 0.4169 data: 0.0002 max mem: 52150 +Epoch: [18] [5640/7269] eta: 0:11:41 lr: 0.000007 loss: 0.4844 (0.5248) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [18] [5650/7269] eta: 0:11:37 lr: 0.000007 loss: 0.5046 (0.5248) time: 0.4814 data: 0.0002 max mem: 52150 +Epoch: [18] [5660/7269] eta: 0:11:33 lr: 0.000007 loss: 0.5115 (0.5248) time: 0.4814 data: 0.0002 max mem: 52150 +Epoch: [18] [5670/7269] eta: 0:11:28 lr: 0.000007 loss: 0.5115 (0.5248) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [18] [5680/7269] eta: 0:11:24 lr: 0.000007 loss: 0.5188 (0.5248) time: 0.4341 data: 0.0002 max mem: 52150 +Epoch: [18] [5690/7269] eta: 0:11:20 lr: 0.000007 loss: 0.5478 (0.5248) time: 0.4934 data: 0.0002 max mem: 52150 +Epoch: [18] [5700/7269] eta: 0:11:16 lr: 0.000007 loss: 0.5226 (0.5248) time: 0.4780 data: 0.0002 max mem: 52150 +Epoch: [18] [5710/7269] eta: 0:11:11 lr: 0.000007 loss: 0.5059 (0.5248) time: 0.4186 data: 0.0002 max mem: 52150 +Epoch: [18] [5720/7269] eta: 0:11:07 lr: 0.000007 loss: 0.5599 (0.5248) time: 0.4641 data: 0.0002 max mem: 52150 +Epoch: [18] [5730/7269] eta: 0:11:03 lr: 0.000007 loss: 0.5390 (0.5248) time: 0.5082 data: 0.0002 max mem: 52150 +Epoch: [18] [5740/7269] eta: 0:10:59 lr: 0.000007 loss: 0.5652 (0.5249) time: 0.4163 data: 0.0002 max mem: 52150 +Epoch: [18] [5750/7269] eta: 0:10:54 lr: 0.000007 loss: 0.5649 (0.5249) time: 0.4211 data: 0.0002 max mem: 52150 +Epoch: [18] [5760/7269] eta: 0:10:50 lr: 0.000007 loss: 0.5500 (0.5249) time: 0.4210 data: 0.0002 max mem: 52150 +Epoch: [18] [5770/7269] eta: 0:10:46 lr: 0.000007 loss: 0.5089 (0.5249) time: 0.4160 data: 0.0001 max mem: 52150 +Epoch: [18] [5780/7269] eta: 0:10:42 lr: 0.000007 loss: 0.5365 (0.5249) time: 0.4644 data: 0.0001 max mem: 52150 +Epoch: [18] [5790/7269] eta: 0:10:37 lr: 0.000007 loss: 0.5422 (0.5249) time: 0.4695 data: 0.0002 max mem: 52150 +Epoch: [18] [5800/7269] eta: 0:10:33 lr: 0.000007 loss: 0.5554 (0.5249) time: 0.4212 data: 0.0002 max mem: 52150 +Epoch: [18] [5810/7269] eta: 0:10:29 lr: 0.000007 loss: 0.5347 (0.5249) time: 0.4344 data: 0.0002 max mem: 52150 +Epoch: [18] [5820/7269] eta: 0:10:24 lr: 0.000007 loss: 0.5117 (0.5249) time: 0.4343 data: 0.0002 max mem: 52150 +Epoch: [18] [5830/7269] eta: 0:10:20 lr: 0.000007 loss: 0.5117 (0.5249) time: 0.4780 data: 0.0002 max mem: 52150 +Epoch: [18] [5840/7269] eta: 0:10:16 lr: 0.000007 loss: 0.5122 (0.5249) time: 0.4783 data: 0.0002 max mem: 52150 +Epoch: [18] [5850/7269] eta: 0:10:11 lr: 0.000007 loss: 0.5245 (0.5249) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [18] [5860/7269] eta: 0:10:07 lr: 0.000007 loss: 0.5489 (0.5249) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [18] [5870/7269] eta: 0:10:03 lr: 0.000007 loss: 0.5271 (0.5248) time: 0.4305 data: 0.0002 max mem: 52150 +Epoch: [18] [5880/7269] eta: 0:09:58 lr: 0.000007 loss: 0.5377 (0.5249) time: 0.4309 data: 0.0002 max mem: 52150 +Epoch: [18] [5890/7269] eta: 0:09:54 lr: 0.000007 loss: 0.5623 (0.5249) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [18] [5900/7269] eta: 0:09:49 lr: 0.000007 loss: 0.5387 (0.5249) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [18] [5910/7269] eta: 0:09:45 lr: 0.000007 loss: 0.5317 (0.5249) time: 0.4168 data: 0.0002 max mem: 52150 +Epoch: [18] [5920/7269] eta: 0:09:41 lr: 0.000007 loss: 0.5212 (0.5249) time: 0.4176 data: 0.0002 max mem: 52150 +Epoch: [18] [5930/7269] eta: 0:09:37 lr: 0.000007 loss: 0.5027 (0.5248) time: 0.4377 data: 0.0002 max mem: 52150 +Epoch: [18] [5940/7269] eta: 0:09:32 lr: 0.000007 loss: 0.4878 (0.5248) time: 0.4979 data: 0.0002 max mem: 52150 +Epoch: [18] [5950/7269] eta: 0:09:28 lr: 0.000007 loss: 0.5179 (0.5248) time: 0.4323 data: 0.0002 max mem: 52150 +Epoch: [18] [5960/7269] eta: 0:09:24 lr: 0.000007 loss: 0.5080 (0.5247) time: 0.4175 data: 0.0001 max mem: 52150 +Epoch: [18] [5970/7269] eta: 0:09:20 lr: 0.000007 loss: 0.5322 (0.5248) time: 0.4779 data: 0.0001 max mem: 52150 +Epoch: [18] [5980/7269] eta: 0:09:15 lr: 0.000007 loss: 0.5322 (0.5248) time: 0.4318 data: 0.0001 max mem: 52150 +Epoch: [18] [5990/7269] eta: 0:09:11 lr: 0.000007 loss: 0.5161 (0.5248) time: 0.3714 data: 0.0001 max mem: 52150 +Epoch: [18] [6000/7269] eta: 0:09:06 lr: 0.000007 loss: 0.5350 (0.5248) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [18] [6010/7269] eta: 0:09:02 lr: 0.000007 loss: 0.5662 (0.5249) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [18] [6020/7269] eta: 0:08:57 lr: 0.000007 loss: 0.5433 (0.5249) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [18] [6030/7269] eta: 0:08:53 lr: 0.000007 loss: 0.5360 (0.5249) time: 0.4790 data: 0.0002 max mem: 52150 +Epoch: [18] [6040/7269] eta: 0:08:49 lr: 0.000007 loss: 0.5360 (0.5249) time: 0.4789 data: 0.0002 max mem: 52150 +Epoch: [18] [6050/7269] eta: 0:08:45 lr: 0.000007 loss: 0.5485 (0.5250) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [18] [6060/7269] eta: 0:08:40 lr: 0.000007 loss: 0.5824 (0.5251) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [18] [6070/7269] eta: 0:08:36 lr: 0.000007 loss: 0.5744 (0.5251) time: 0.4805 data: 0.0002 max mem: 52150 +Epoch: [18] [6080/7269] eta: 0:08:32 lr: 0.000007 loss: 0.5166 (0.5250) time: 0.6348 data: 0.0002 max mem: 52150 +Epoch: [18] [6090/7269] eta: 0:08:28 lr: 0.000007 loss: 0.5168 (0.5250) time: 0.5262 data: 0.0002 max mem: 52150 +Epoch: [18] [6100/7269] eta: 0:08:23 lr: 0.000007 loss: 0.5040 (0.5249) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [18] [6110/7269] eta: 0:08:19 lr: 0.000007 loss: 0.5003 (0.5249) time: 0.4184 data: 0.0001 max mem: 52150 +Epoch: [18] [6120/7269] eta: 0:08:15 lr: 0.000007 loss: 0.5041 (0.5249) time: 0.4807 data: 0.0001 max mem: 52150 +Epoch: [18] [6130/7269] eta: 0:08:11 lr: 0.000007 loss: 0.5032 (0.5249) time: 0.4336 data: 0.0001 max mem: 52150 +Epoch: [18] [6140/7269] eta: 0:08:06 lr: 0.000007 loss: 0.5052 (0.5248) time: 0.3711 data: 0.0001 max mem: 52150 +Epoch: [18] [6150/7269] eta: 0:08:02 lr: 0.000007 loss: 0.5414 (0.5249) time: 0.3712 data: 0.0002 max mem: 52150 +Epoch: [18] [6160/7269] eta: 0:07:57 lr: 0.000007 loss: 0.5671 (0.5249) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [18] [6170/7269] eta: 0:07:53 lr: 0.000007 loss: 0.5586 (0.5249) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [18] [6180/7269] eta: 0:07:49 lr: 0.000007 loss: 0.5144 (0.5249) time: 0.4193 data: 0.0002 max mem: 52150 +Epoch: [18] [6190/7269] eta: 0:07:44 lr: 0.000007 loss: 0.4932 (0.5249) time: 0.4193 data: 0.0001 max mem: 52150 +Epoch: [18] [6200/7269] eta: 0:07:40 lr: 0.000007 loss: 0.5353 (0.5249) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [18] [6210/7269] eta: 0:07:35 lr: 0.000007 loss: 0.5172 (0.5249) time: 0.3728 data: 0.0002 max mem: 52150 +Epoch: [18] [6220/7269] eta: 0:07:31 lr: 0.000007 loss: 0.4836 (0.5248) time: 0.4652 data: 0.0002 max mem: 52150 +Epoch: [18] [6230/7269] eta: 0:07:27 lr: 0.000007 loss: 0.5001 (0.5248) time: 0.4648 data: 0.0002 max mem: 52150 +Epoch: [18] [6240/7269] eta: 0:07:23 lr: 0.000007 loss: 0.5049 (0.5248) time: 0.4177 data: 0.0002 max mem: 52150 +Epoch: [18] [6250/7269] eta: 0:07:19 lr: 0.000007 loss: 0.5029 (0.5248) time: 0.5164 data: 0.0002 max mem: 52150 +Epoch: [18] [6260/7269] eta: 0:07:14 lr: 0.000007 loss: 0.5086 (0.5248) time: 0.4713 data: 0.0002 max mem: 52150 +Epoch: [18] [6270/7269] eta: 0:07:10 lr: 0.000007 loss: 0.5375 (0.5248) time: 0.4180 data: 0.0002 max mem: 52150 +Epoch: [18] [6280/7269] eta: 0:07:06 lr: 0.000007 loss: 0.5481 (0.5248) time: 0.4172 data: 0.0002 max mem: 52150 +Epoch: [18] [6290/7269] eta: 0:07:02 lr: 0.000007 loss: 0.5424 (0.5248) time: 0.5400 data: 0.0002 max mem: 52150 +Epoch: [18] [6300/7269] eta: 0:06:57 lr: 0.000007 loss: 0.5397 (0.5248) time: 0.5873 data: 0.0001 max mem: 52150 +Epoch: [18] [6310/7269] eta: 0:06:53 lr: 0.000007 loss: 0.5406 (0.5248) time: 0.4824 data: 0.0002 max mem: 52150 +Epoch: [18] [6320/7269] eta: 0:06:49 lr: 0.000007 loss: 0.5134 (0.5248) time: 0.4355 data: 0.0002 max mem: 52150 +Epoch: [18] [6330/7269] eta: 0:06:44 lr: 0.000007 loss: 0.4684 (0.5247) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [18] [6340/7269] eta: 0:06:40 lr: 0.000007 loss: 0.5104 (0.5247) time: 0.4201 data: 0.0002 max mem: 52150 +Epoch: [18] [6350/7269] eta: 0:06:36 lr: 0.000007 loss: 0.5332 (0.5247) time: 0.4203 data: 0.0002 max mem: 52150 +Epoch: [18] [6360/7269] eta: 0:06:31 lr: 0.000007 loss: 0.5332 (0.5247) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [18] [6370/7269] eta: 0:06:27 lr: 0.000007 loss: 0.5253 (0.5247) time: 0.3711 data: 0.0001 max mem: 52150 +Epoch: [18] [6380/7269] eta: 0:06:23 lr: 0.000007 loss: 0.5056 (0.5247) time: 0.3712 data: 0.0002 max mem: 52150 +Epoch: [18] [6390/7269] eta: 0:06:18 lr: 0.000007 loss: 0.5087 (0.5247) time: 0.4177 data: 0.0002 max mem: 52150 +Epoch: [18] [6400/7269] eta: 0:06:14 lr: 0.000007 loss: 0.5111 (0.5247) time: 0.4799 data: 0.0002 max mem: 52150 +Epoch: [18] [6410/7269] eta: 0:06:10 lr: 0.000007 loss: 0.4997 (0.5247) time: 0.5535 data: 0.0002 max mem: 52150 +Epoch: [18] [6420/7269] eta: 0:06:06 lr: 0.000007 loss: 0.4997 (0.5247) time: 0.5379 data: 0.0002 max mem: 52150 +Epoch: [18] [6430/7269] eta: 0:06:01 lr: 0.000007 loss: 0.5236 (0.5247) time: 0.4627 data: 0.0002 max mem: 52150 +Epoch: [18] [6440/7269] eta: 0:05:57 lr: 0.000007 loss: 0.5100 (0.5246) time: 0.4759 data: 0.0001 max mem: 52150 +Epoch: [18] [6450/7269] eta: 0:05:53 lr: 0.000007 loss: 0.5325 (0.5246) time: 0.4313 data: 0.0001 max mem: 52150 +Epoch: [18] [6460/7269] eta: 0:05:48 lr: 0.000007 loss: 0.5481 (0.5247) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [18] [6470/7269] eta: 0:05:44 lr: 0.000007 loss: 0.5400 (0.5247) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [18] [6480/7269] eta: 0:05:40 lr: 0.000007 loss: 0.4888 (0.5246) time: 0.4349 data: 0.0002 max mem: 52150 +Epoch: [18] [6490/7269] eta: 0:05:36 lr: 0.000007 loss: 0.4891 (0.5246) time: 0.4809 data: 0.0002 max mem: 52150 +Epoch: [18] [6500/7269] eta: 0:05:31 lr: 0.000007 loss: 0.5334 (0.5246) time: 0.4173 data: 0.0002 max mem: 52150 +Epoch: [18] [6510/7269] eta: 0:05:27 lr: 0.000007 loss: 0.5334 (0.5246) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [18] [6520/7269] eta: 0:05:22 lr: 0.000007 loss: 0.5397 (0.5246) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [18] [6530/7269] eta: 0:05:18 lr: 0.000007 loss: 0.5105 (0.5246) time: 0.4339 data: 0.0002 max mem: 52150 +Epoch: [18] [6540/7269] eta: 0:05:14 lr: 0.000007 loss: 0.4974 (0.5245) time: 0.4336 data: 0.0002 max mem: 52150 +Epoch: [18] [6550/7269] eta: 0:05:09 lr: 0.000007 loss: 0.5076 (0.5245) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [18] [6560/7269] eta: 0:05:05 lr: 0.000007 loss: 0.5399 (0.5246) time: 0.4766 data: 0.0001 max mem: 52150 +Epoch: [18] [6570/7269] eta: 0:05:01 lr: 0.000007 loss: 0.5352 (0.5246) time: 0.5249 data: 0.0002 max mem: 52150 +Epoch: [18] [6580/7269] eta: 0:04:57 lr: 0.000007 loss: 0.5320 (0.5245) time: 0.4202 data: 0.0002 max mem: 52150 +Epoch: [18] [6590/7269] eta: 0:04:52 lr: 0.000007 loss: 0.4811 (0.5245) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [18] [6600/7269] eta: 0:04:48 lr: 0.000007 loss: 0.4811 (0.5245) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [18] [6610/7269] eta: 0:04:44 lr: 0.000007 loss: 0.5221 (0.5245) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [18] [6620/7269] eta: 0:04:39 lr: 0.000007 loss: 0.5221 (0.5245) time: 0.4188 data: 0.0002 max mem: 52150 +Epoch: [18] [6630/7269] eta: 0:04:35 lr: 0.000007 loss: 0.5025 (0.5244) time: 0.4188 data: 0.0002 max mem: 52150 +Epoch: [18] [6640/7269] eta: 0:04:31 lr: 0.000007 loss: 0.5192 (0.5244) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [18] [6650/7269] eta: 0:04:26 lr: 0.000007 loss: 0.5214 (0.5244) time: 0.4807 data: 0.0002 max mem: 52150 +Epoch: [18] [6660/7269] eta: 0:04:22 lr: 0.000007 loss: 0.5119 (0.5244) time: 0.4809 data: 0.0002 max mem: 52150 +Epoch: [18] [6670/7269] eta: 0:04:18 lr: 0.000007 loss: 0.5149 (0.5244) time: 0.3721 data: 0.0001 max mem: 52150 +Epoch: [18] [6680/7269] eta: 0:04:13 lr: 0.000007 loss: 0.5245 (0.5244) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [18] [6690/7269] eta: 0:04:09 lr: 0.000007 loss: 0.5422 (0.5244) time: 0.3712 data: 0.0002 max mem: 52150 +Epoch: [18] [6700/7269] eta: 0:04:05 lr: 0.000007 loss: 0.5123 (0.5244) time: 0.4162 data: 0.0001 max mem: 52150 +Epoch: [18] [6710/7269] eta: 0:04:00 lr: 0.000007 loss: 0.5123 (0.5244) time: 0.5109 data: 0.0002 max mem: 52150 +Epoch: [18] [6720/7269] eta: 0:03:56 lr: 0.000007 loss: 0.5313 (0.5244) time: 0.5284 data: 0.0001 max mem: 52150 +Epoch: [18] [6730/7269] eta: 0:03:52 lr: 0.000007 loss: 0.5342 (0.5244) time: 0.4825 data: 0.0001 max mem: 52150 +Epoch: [18] [6740/7269] eta: 0:03:48 lr: 0.000007 loss: 0.5246 (0.5244) time: 0.4203 data: 0.0001 max mem: 52150 +Epoch: [18] [6750/7269] eta: 0:03:43 lr: 0.000007 loss: 0.5124 (0.5244) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [18] [6760/7269] eta: 0:03:39 lr: 0.000007 loss: 0.5407 (0.5245) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [18] [6770/7269] eta: 0:03:34 lr: 0.000007 loss: 0.5396 (0.5244) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [18] [6780/7269] eta: 0:03:30 lr: 0.000007 loss: 0.5423 (0.5245) time: 0.4931 data: 0.0002 max mem: 52150 +Epoch: [18] [6790/7269] eta: 0:03:26 lr: 0.000007 loss: 0.5509 (0.5245) time: 0.5420 data: 0.0002 max mem: 52150 +Epoch: [18] [6800/7269] eta: 0:03:22 lr: 0.000007 loss: 0.5231 (0.5245) time: 0.4206 data: 0.0002 max mem: 52150 +Epoch: [18] [6810/7269] eta: 0:03:17 lr: 0.000007 loss: 0.5194 (0.5245) time: 0.4354 data: 0.0002 max mem: 52150 +Epoch: [18] [6820/7269] eta: 0:03:13 lr: 0.000007 loss: 0.5545 (0.5246) time: 0.4816 data: 0.0001 max mem: 52150 +Epoch: [18] [6830/7269] eta: 0:03:09 lr: 0.000007 loss: 0.5625 (0.5246) time: 0.5144 data: 0.0002 max mem: 52150 +Epoch: [18] [6840/7269] eta: 0:03:05 lr: 0.000007 loss: 0.5393 (0.5246) time: 0.4681 data: 0.0002 max mem: 52150 +Epoch: [18] [6850/7269] eta: 0:03:00 lr: 0.000007 loss: 0.5323 (0.5246) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [18] [6860/7269] eta: 0:02:56 lr: 0.000007 loss: 0.5275 (0.5245) time: 0.4170 data: 0.0002 max mem: 52150 +Epoch: [18] [6870/7269] eta: 0:02:52 lr: 0.000007 loss: 0.4993 (0.5245) time: 0.4817 data: 0.0002 max mem: 52150 +Epoch: [18] [6880/7269] eta: 0:02:47 lr: 0.000007 loss: 0.5278 (0.5246) time: 0.4365 data: 0.0002 max mem: 52150 +Epoch: [18] [6890/7269] eta: 0:02:43 lr: 0.000007 loss: 0.5561 (0.5246) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [18] [6900/7269] eta: 0:02:39 lr: 0.000007 loss: 0.5561 (0.5246) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [18] [6910/7269] eta: 0:02:34 lr: 0.000007 loss: 0.5355 (0.5246) time: 0.4184 data: 0.0001 max mem: 52150 +Epoch: [18] [6920/7269] eta: 0:02:30 lr: 0.000007 loss: 0.5355 (0.5247) time: 0.4626 data: 0.0002 max mem: 52150 +Epoch: [18] [6930/7269] eta: 0:02:26 lr: 0.000007 loss: 0.5489 (0.5247) time: 0.4615 data: 0.0002 max mem: 52150 +Epoch: [18] [6940/7269] eta: 0:02:21 lr: 0.000007 loss: 0.5176 (0.5247) time: 0.5110 data: 0.0001 max mem: 52150 +Epoch: [18] [6950/7269] eta: 0:02:17 lr: 0.000007 loss: 0.5274 (0.5247) time: 0.4655 data: 0.0001 max mem: 52150 +Epoch: [18] [6960/7269] eta: 0:02:13 lr: 0.000007 loss: 0.5238 (0.5247) time: 0.4199 data: 0.0001 max mem: 52150 +Epoch: [18] [6970/7269] eta: 0:02:09 lr: 0.000007 loss: 0.4893 (0.5246) time: 0.4809 data: 0.0002 max mem: 52150 +Epoch: [18] [6980/7269] eta: 0:02:04 lr: 0.000007 loss: 0.4954 (0.5246) time: 0.4329 data: 0.0002 max mem: 52150 +Epoch: [18] [6990/7269] eta: 0:02:00 lr: 0.000007 loss: 0.5136 (0.5246) time: 0.4336 data: 0.0002 max mem: 52150 +Epoch: [18] [7000/7269] eta: 0:01:56 lr: 0.000007 loss: 0.5182 (0.5246) time: 0.4335 data: 0.0002 max mem: 52150 +Epoch: [18] [7010/7269] eta: 0:01:51 lr: 0.000007 loss: 0.5408 (0.5246) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [18] [7020/7269] eta: 0:01:47 lr: 0.000007 loss: 0.5327 (0.5246) time: 0.4309 data: 0.0002 max mem: 52150 +Epoch: [18] [7030/7269] eta: 0:01:43 lr: 0.000007 loss: 0.5252 (0.5245) time: 0.4308 data: 0.0002 max mem: 52150 +Epoch: [18] [7040/7269] eta: 0:01:38 lr: 0.000007 loss: 0.5076 (0.5245) time: 0.4171 data: 0.0002 max mem: 52150 +Epoch: [18] [7050/7269] eta: 0:01:34 lr: 0.000007 loss: 0.5076 (0.5245) time: 0.4175 data: 0.0002 max mem: 52150 +Epoch: [18] [7060/7269] eta: 0:01:30 lr: 0.000007 loss: 0.5443 (0.5246) time: 0.4389 data: 0.0002 max mem: 52150 +Epoch: [18] [7070/7269] eta: 0:01:25 lr: 0.000007 loss: 0.5380 (0.5246) time: 0.4390 data: 0.0002 max mem: 52150 +Epoch: [18] [7080/7269] eta: 0:01:21 lr: 0.000007 loss: 0.5037 (0.5245) time: 0.3727 data: 0.0002 max mem: 52150 +Epoch: [18] [7090/7269] eta: 0:01:17 lr: 0.000007 loss: 0.4928 (0.5245) time: 0.3727 data: 0.0002 max mem: 52150 +Epoch: [18] [7100/7269] eta: 0:01:12 lr: 0.000007 loss: 0.5221 (0.5245) time: 0.4188 data: 0.0002 max mem: 52150 +Epoch: [18] [7110/7269] eta: 0:01:08 lr: 0.000007 loss: 0.5261 (0.5245) time: 0.4191 data: 0.0002 max mem: 52150 +Epoch: [18] [7120/7269] eta: 0:01:04 lr: 0.000007 loss: 0.5153 (0.5245) time: 0.4830 data: 0.0002 max mem: 52150 +Epoch: [18] [7130/7269] eta: 0:00:59 lr: 0.000007 loss: 0.5127 (0.5245) time: 0.4824 data: 0.0001 max mem: 52150 +Epoch: [18] [7140/7269] eta: 0:00:55 lr: 0.000007 loss: 0.4742 (0.5244) time: 0.3714 data: 0.0001 max mem: 52150 +Epoch: [18] [7150/7269] eta: 0:00:51 lr: 0.000007 loss: 0.4877 (0.5243) time: 0.3723 data: 0.0001 max mem: 52150 +Epoch: [18] [7160/7269] eta: 0:00:46 lr: 0.000007 loss: 0.5129 (0.5243) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [18] [7170/7269] eta: 0:00:42 lr: 0.000007 loss: 0.5457 (0.5244) time: 0.4156 data: 0.0002 max mem: 52150 +Epoch: [18] [7180/7269] eta: 0:00:38 lr: 0.000007 loss: 0.5457 (0.5244) time: 0.4160 data: 0.0002 max mem: 52150 +Epoch: [18] [7190/7269] eta: 0:00:34 lr: 0.000007 loss: 0.5444 (0.5244) time: 0.4648 data: 0.0001 max mem: 52150 +Epoch: [18] [7200/7269] eta: 0:00:29 lr: 0.000007 loss: 0.5115 (0.5244) time: 0.5798 data: 0.0002 max mem: 52150 +Epoch: [18] [7210/7269] eta: 0:00:25 lr: 0.000007 loss: 0.5285 (0.5244) time: 0.4872 data: 0.0002 max mem: 52150 +Epoch: [18] [7220/7269] eta: 0:00:21 lr: 0.000007 loss: 0.5300 (0.5244) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [18] [7230/7269] eta: 0:00:16 lr: 0.000007 loss: 0.5366 (0.5244) time: 0.4794 data: 0.0002 max mem: 52150 +Epoch: [18] [7240/7269] eta: 0:00:12 lr: 0.000007 loss: 0.5366 (0.5244) time: 0.4798 data: 0.0002 max mem: 52150 +Epoch: [18] [7250/7269] eta: 0:00:08 lr: 0.000007 loss: 0.5530 (0.5244) time: 0.3756 data: 0.0030 max mem: 52150 +Epoch: [18] [7260/7269] eta: 0:00:03 lr: 0.000007 loss: 0.5530 (0.5244) time: 0.4337 data: 0.0030 max mem: 52150 +Epoch: [18] [7268/7269] eta: 0:00:00 lr: 0.000007 loss: 0.5436 (0.5244) time: 0.4331 data: 0.0029 max mem: 52150 +Epoch: [18] Total time: 0:52:17 (0.4317 s / it) +Averaged stats: lr: 0.000007 loss: 0.5436 (0.5249) +Test: [ 0/1403] eta: 1:19:29 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0799 (0.0799) acc1: 100.0000 (100.0000) acc5: 100.0000 (100.0000) time: 3.3995 data: 3.3024 max mem: 52150 +Test: [ 10/1403] eta: 0:09:12 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1124 (0.1296) acc1: 100.0000 (98.4375) acc5: 100.0000 (100.0000) time: 0.3967 data: 0.3005 max mem: 52150 +Test: [ 20/1403] eta: 0:05:51 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1074 (0.1219) acc1: 98.4375 (98.0655) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 30/1403] eta: 0:04:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0789 (0.1212) acc1: 98.4375 (97.7823) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 40/1403] eta: 0:04:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0881 (0.1201) acc1: 100.0000 (97.9421) acc5: 100.0000 (100.0000) time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 50/1403] eta: 0:03:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0881 (0.1324) acc1: 100.0000 (97.3346) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 60/1403] eta: 0:03:22 acc1_pos: 92.188 acc5_pos: 100.000 loss: 0.0964 (0.1310) acc1: 100.0000 (97.6178) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 70/1403] eta: 0:03:10 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0989 (0.1323) acc1: 100.0000 (97.4252) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 80/1403] eta: 0:03:01 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1174 (0.1373) acc1: 98.4375 (97.2222) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 90/1403] eta: 0:02:54 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1399 (0.1526) acc1: 95.3125 (96.3427) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 100/1403] eta: 0:02:48 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1418 (0.1522) acc1: 96.8750 (96.4418) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 110/1403] eta: 0:02:43 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1146 (0.1506) acc1: 98.4375 (96.5372) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 120/1403] eta: 0:02:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0837 (0.1449) acc1: 100.0000 (96.8104) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 130/1403] eta: 0:02:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0826 (0.1413) acc1: 100.0000 (96.9466) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 140/1403] eta: 0:02:31 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0796 (0.1367) acc1: 100.0000 (97.1631) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 150/1403] eta: 0:02:28 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0766 (0.1330) acc1: 100.0000 (97.3406) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 160/1403] eta: 0:02:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0768 (0.1302) acc1: 100.0000 (97.4573) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 170/1403] eta: 0:02:22 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0786 (0.1276) acc1: 100.0000 (97.5786) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 180/1403] eta: 0:02:20 acc1_pos: 90.625 acc5_pos: 100.000 loss: 0.0814 (0.1319) acc1: 100.0000 (97.3757) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0004 max mem: 52150 +Test: [ 190/1403] eta: 0:02:18 acc1_pos: 84.375 acc5_pos: 100.000 loss: 0.2519 (0.1405) acc1: 90.6250 (96.8750) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0004 max mem: 52150 +Test: [ 200/1403] eta: 0:02:15 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.2461 (0.1526) acc1: 90.6250 (96.2920) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 210/1403] eta: 0:02:13 acc1_pos: 85.938 acc5_pos: 100.000 loss: 0.2878 (0.1642) acc1: 85.9375 (95.6531) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 220/1403] eta: 0:02:11 acc1_pos: 89.062 acc5_pos: 100.000 loss: 0.3535 (0.1741) acc1: 85.9375 (95.1782) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 230/1403] eta: 0:02:10 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2403 (0.1709) acc1: 90.6250 (95.3328) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 240/1403] eta: 0:02:08 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0798 (0.1671) acc1: 100.0000 (95.5135) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 250/1403] eta: 0:02:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0775 (0.1639) acc1: 100.0000 (95.6673) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 260/1403] eta: 0:02:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0800 (0.1609) acc1: 100.0000 (95.8094) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 270/1403] eta: 0:02:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0807 (0.1580) acc1: 100.0000 (95.9525) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 280/1403] eta: 0:02:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0790 (0.1552) acc1: 100.0000 (96.0910) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 290/1403] eta: 0:02:00 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0754 (0.1525) acc1: 100.0000 (96.2199) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 300/1403] eta: 0:01:58 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0743 (0.1501) acc1: 100.0000 (96.3403) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 310/1403] eta: 0:01:57 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0776 (0.1481) acc1: 100.0000 (96.4480) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 320/1403] eta: 0:01:55 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1034 (0.1468) acc1: 100.0000 (96.5586) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 330/1403] eta: 0:01:54 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1095 (0.1457) acc1: 100.0000 (96.6626) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 340/1403] eta: 0:01:52 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1088 (0.1445) acc1: 100.0000 (96.7604) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 350/1403] eta: 0:01:51 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1088 (0.1435) acc1: 100.0000 (96.8527) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 360/1403] eta: 0:01:50 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1100 (0.1427) acc1: 100.0000 (96.9313) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 370/1403] eta: 0:01:48 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1070 (0.1417) acc1: 100.0000 (97.0140) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 380/1403] eta: 0:01:47 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1058 (0.1408) acc1: 100.0000 (97.0924) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 390/1403] eta: 0:01:46 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1071 (0.1426) acc1: 100.0000 (96.9989) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 400/1403] eta: 0:01:45 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3066 (0.1483) acc1: 85.9375 (96.6802) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 410/1403] eta: 0:01:43 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3473 (0.1532) acc1: 85.9375 (96.4074) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 420/1403] eta: 0:01:42 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2967 (0.1563) acc1: 87.5000 (96.2218) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 430/1403] eta: 0:01:41 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2853 (0.1606) acc1: 89.0625 (96.0049) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 440/1403] eta: 0:01:40 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3616 (0.1656) acc1: 85.9375 (95.7483) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 450/1403] eta: 0:01:38 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3506 (0.1690) acc1: 85.9375 (95.5758) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 460/1403] eta: 0:01:37 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3159 (0.1729) acc1: 87.5000 (95.3600) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 470/1403] eta: 0:01:36 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3006 (0.1724) acc1: 87.5000 (95.3656) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 480/1403] eta: 0:01:35 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0636 (0.1701) acc1: 100.0000 (95.4619) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 490/1403] eta: 0:01:34 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0636 (0.1680) acc1: 100.0000 (95.5544) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 500/1403] eta: 0:01:33 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0648 (0.1659) acc1: 100.0000 (95.6431) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 510/1403] eta: 0:01:32 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0648 (0.1640) acc1: 100.0000 (95.7284) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 520/1403] eta: 0:01:30 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0690 (0.1622) acc1: 100.0000 (95.8103) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 530/1403] eta: 0:01:29 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0614 (0.1603) acc1: 100.0000 (95.8892) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 540/1403] eta: 0:01:28 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0610 (0.1585) acc1: 100.0000 (95.9652) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 550/1403] eta: 0:01:27 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0619 (0.1573) acc1: 100.0000 (96.0129) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 560/1403] eta: 0:01:26 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0883 (0.1564) acc1: 98.4375 (96.0506) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 570/1403] eta: 0:01:25 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1018 (0.1555) acc1: 98.4375 (96.0760) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 580/1403] eta: 0:01:24 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1067 (0.1552) acc1: 98.4375 (96.0816) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 590/1403] eta: 0:01:23 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1268 (0.1550) acc1: 96.8750 (96.0739) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 600/1403] eta: 0:01:21 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1545 (0.1559) acc1: 93.7500 (96.0067) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 610/1403] eta: 0:01:20 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1545 (0.1557) acc1: 93.7500 (95.9953) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 620/1403] eta: 0:01:19 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1661 (0.1574) acc1: 93.7500 (95.8711) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 630/1403] eta: 0:01:18 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1959 (0.1575) acc1: 92.1875 (95.8474) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 640/1403] eta: 0:01:17 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1352 (0.1573) acc1: 95.3125 (95.8561) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 650/1403] eta: 0:01:16 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1450 (0.1573) acc1: 95.3125 (95.8405) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 660/1403] eta: 0:01:15 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1262 (0.1564) acc1: 96.8750 (95.8845) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 670/1403] eta: 0:01:14 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0769 (0.1552) acc1: 100.0000 (95.9366) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 680/1403] eta: 0:01:13 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0904 (0.1583) acc1: 98.4375 (95.7507) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 690/1403] eta: 0:01:12 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.4327 (0.1647) acc1: 79.6875 (95.3894) acc5: 100.0000 (100.0000) acc1_neg: 70.312 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [ 700/1403] eta: 0:01:11 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2811 (0.1655) acc1: 84.3750 (95.3437) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [ 710/1403] eta: 0:01:10 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3682 (0.1707) acc1: 84.3750 (95.0576) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 720/1403] eta: 0:01:09 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3682 (0.1709) acc1: 81.2500 (95.0351) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 730/1403] eta: 0:01:08 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0744 (0.1698) acc1: 100.0000 (95.0881) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 740/1403] eta: 0:01:06 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0792 (0.1689) acc1: 98.4375 (95.1248) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 750/1403] eta: 0:01:05 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1245 (0.1716) acc1: 96.8750 (94.9630) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 760/1403] eta: 0:01:04 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.4564 (0.1775) acc1: 71.8750 (94.6103) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 770/1403] eta: 0:01:03 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.6890 (0.1844) acc1: 62.5000 (94.1736) acc5: 100.0000 (100.0000) acc1_neg: 60.938 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 780/1403] eta: 0:01:02 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.6061 (0.1883) acc1: 67.1875 (93.9461) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 790/1403] eta: 0:01:01 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3827 (0.1908) acc1: 81.2500 (93.8231) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 800/1403] eta: 0:01:00 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2356 (0.1913) acc1: 90.6250 (93.8027) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 810/1403] eta: 0:00:59 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2356 (0.1945) acc1: 87.5000 (93.6460) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 820/1403] eta: 0:00:58 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.6381 (0.2033) acc1: 70.3125 (93.2571) acc5: 100.0000 (100.0000) acc1_neg: 40.625 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 830/1403] eta: 0:00:57 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.7291 (0.2063) acc1: 67.1875 (93.1107) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 840/1403] eta: 0:00:56 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.5349 (0.2140) acc1: 70.3125 (92.7783) acc5: 100.0000 (100.0000) acc1_neg: 50.000 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [ 850/1403] eta: 0:00:55 acc1_pos: 94.118 acc5_pos: 100.000 loss: 1.0235 (0.2247) acc1: 54.6875 (92.3087) acc5: 100.0000 (100.0000) acc1_neg: 50.000 acc5_neg: 100.000 time: 0.0969 data: 0.0004 max mem: 52150 +Test: [ 860/1403] eta: 0:00:54 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.9875 (0.2280) acc1: 56.2500 (92.1585) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 870/1403] eta: 0:00:53 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1528 (0.2269) acc1: 93.7500 (92.2036) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 880/1403] eta: 0:00:52 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1073 (0.2257) acc1: 96.8750 (92.2638) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 890/1403] eta: 0:00:51 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1073 (0.2243) acc1: 98.4375 (92.3295) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 900/1403] eta: 0:00:50 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0865 (0.2228) acc1: 98.4375 (92.4077) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 910/1403] eta: 0:00:49 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0913 (0.2222) acc1: 98.4375 (92.4208) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 920/1403] eta: 0:00:48 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1574 (0.2217) acc1: 93.7500 (92.4420) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 930/1403] eta: 0:00:47 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1554 (0.2210) acc1: 95.3125 (92.4678) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 940/1403] eta: 0:00:46 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1890 (0.2219) acc1: 93.7500 (92.4067) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 950/1403] eta: 0:00:45 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3363 (0.2238) acc1: 81.2500 (92.2910) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 960/1403] eta: 0:00:44 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3843 (0.2253) acc1: 81.2500 (92.2070) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 970/1403] eta: 0:00:43 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3736 (0.2270) acc1: 82.8125 (92.1119) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 980/1403] eta: 0:00:42 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3575 (0.2287) acc1: 81.2500 (91.9884) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 990/1403] eta: 0:00:41 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3400 (0.2299) acc1: 82.8125 (91.9210) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1000/1403] eta: 0:00:40 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3548 (0.2315) acc1: 84.3750 (91.8254) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1010/1403] eta: 0:00:39 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3876 (0.2330) acc1: 82.8125 (91.7455) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1020/1403] eta: 0:00:38 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.4941 (0.2403) acc1: 78.1250 (91.3473) acc5: 100.0000 (100.0000) acc1_neg: 26.562 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1030/1403] eta: 0:00:37 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.6646 (0.2431) acc1: 53.1250 (91.1782) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1040/1403] eta: 0:00:36 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2698 (0.2430) acc1: 87.5000 (91.1623) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1050/1403] eta: 0:00:35 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1969 (0.2425) acc1: 93.7500 (91.1974) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1060/1403] eta: 0:00:34 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1487 (0.2416) acc1: 95.3125 (91.2450) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1070/1403] eta: 0:00:33 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1867 (0.2422) acc1: 95.3125 (91.2202) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1080/1403] eta: 0:00:32 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3802 (0.2442) acc1: 84.3750 (91.1121) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52150 +Test: [1090/1403] eta: 0:00:31 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.4567 (0.2462) acc1: 81.2500 (91.0146) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52150 +Test: [1100/1403] eta: 0:00:30 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3981 (0.2469) acc1: 84.3750 (90.9840) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52150 +Test: [1110/1403] eta: 0:00:29 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3082 (0.2475) acc1: 87.5000 (90.9541) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52150 +Test: [1120/1403] eta: 0:00:28 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3177 (0.2481) acc1: 87.5000 (90.9219) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52150 +Test: [1130/1403] eta: 0:00:27 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3216 (0.2490) acc1: 85.9375 (90.8681) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1140/1403] eta: 0:00:26 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2947 (0.2493) acc1: 87.5000 (90.8578) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1150/1403] eta: 0:00:25 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2921 (0.2500) acc1: 87.5000 (90.8110) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1160/1403] eta: 0:00:24 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3156 (0.2506) acc1: 85.9375 (90.7811) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52150 +Test: [1170/1403] eta: 0:00:23 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3103 (0.2505) acc1: 87.5000 (90.7891) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52150 +Test: [1180/1403] eta: 0:00:22 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0723 (0.2489) acc1: 100.0000 (90.8671) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52150 +Test: [1190/1403] eta: 0:00:21 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0717 (0.2475) acc1: 100.0000 (90.9438) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52150 +Test: [1200/1403] eta: 0:00:20 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0718 (0.2460) acc1: 100.0000 (91.0192) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52150 +Test: [1210/1403] eta: 0:00:19 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0712 (0.2446) acc1: 100.0000 (91.0934) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52150 +Test: [1220/1403] eta: 0:00:18 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0707 (0.2431) acc1: 100.0000 (91.1663) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52150 +Test: [1230/1403] eta: 0:00:17 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0717 (0.2417) acc1: 100.0000 (91.2381) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52150 +Test: [1240/1403] eta: 0:00:16 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0713 (0.2404) acc1: 100.0000 (91.3087) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52150 +Test: [1250/1403] eta: 0:00:15 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0713 (0.2390) acc1: 100.0000 (91.3781) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52150 +Test: [1260/1403] eta: 0:00:14 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0746 (0.2377) acc1: 100.0000 (91.4465) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52150 +Test: [1270/1403] eta: 0:00:13 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0749 (0.2365) acc1: 100.0000 (91.5138) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52150 +Test: [1280/1403] eta: 0:00:12 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0749 (0.2352) acc1: 100.0000 (91.5801) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52150 +Test: [1290/1403] eta: 0:00:11 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0750 (0.2340) acc1: 100.0000 (91.6453) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52150 +Test: [1300/1403] eta: 0:00:10 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0755 (0.2328) acc1: 100.0000 (91.7095) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52150 +Test: [1310/1403] eta: 0:00:09 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0755 (0.2316) acc1: 100.0000 (91.7727) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52150 +Test: [1320/1403] eta: 0:00:08 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0751 (0.2304) acc1: 100.0000 (91.8350) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52150 +Test: [1330/1403] eta: 0:00:07 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0752 (0.2293) acc1: 100.0000 (91.8905) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52150 +Test: [1340/1403] eta: 0:00:06 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0771 (0.2282) acc1: 100.0000 (91.9463) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52150 +Test: [1350/1403] eta: 0:00:05 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0767 (0.2271) acc1: 100.0000 (92.0001) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52150 +Test: [1360/1403] eta: 0:00:04 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0765 (0.2260) acc1: 100.0000 (92.0566) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52150 +Test: [1370/1403] eta: 0:00:03 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0773 (0.2250) acc1: 100.0000 (92.1066) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52150 +Test: [1380/1403] eta: 0:00:02 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0779 (0.2239) acc1: 100.0000 (92.1570) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52150 +Test: [1390/1403] eta: 0:00:01 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0767 (0.2229) acc1: 100.0000 (92.2111) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0002 max mem: 52150 +Test: [1400/1403] eta: 0:00:00 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0752 (0.2219) acc1: 100.0000 (92.2611) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0001 max mem: 52150 +Test: [1402/1403] eta: 0:00:00 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0753 (0.2217) acc1: 100.0000 (92.2667) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0922 data: 0.0001 max mem: 52150 +Test: Total time: 0:02:19 (0.0995 s / it) +* Overall Acc@1 92.267 Acc@5 100.000 + target == real Acc@1 96.416 Acc@5 100.000 + target == fake Acc@1 91.097 Acc@5 100.000 + loss 0.222 +Accuracy of the network on the 89729 test images: 92.3% +Max accuracy: 93.40% +Epoch: [19] [ 0/7269] eta: 12:07:48 lr: 0.000006 loss: 0.5354 (0.5354) time: 6.0075 data: 4.8098 max mem: 52150 +Epoch: [19] [ 10/7269] eta: 1:47:30 lr: 0.000006 loss: 0.5675 (0.5473) time: 0.8886 data: 0.4374 max mem: 52150 +Epoch: [19] [ 20/7269] eta: 1:17:34 lr: 0.000006 loss: 0.5316 (0.5341) time: 0.3738 data: 0.0001 max mem: 52150 +Epoch: [19] [ 30/7269] eta: 1:06:56 lr: 0.000006 loss: 0.5239 (0.5285) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [19] [ 40/7269] eta: 1:05:06 lr: 0.000006 loss: 0.5329 (0.5264) time: 0.4336 data: 0.0002 max mem: 52150 +Epoch: [19] [ 50/7269] eta: 1:01:01 lr: 0.000006 loss: 0.5368 (0.5201) time: 0.4335 data: 0.0002 max mem: 52150 +Epoch: [19] [ 60/7269] eta: 1:02:41 lr: 0.000006 loss: 0.5276 (0.5184) time: 0.4838 data: 0.0001 max mem: 52150 +Epoch: [19] [ 70/7269] eta: 1:00:04 lr: 0.000006 loss: 0.5276 (0.5229) time: 0.4837 data: 0.0002 max mem: 52150 +Epoch: [19] [ 80/7269] eta: 0:58:04 lr: 0.000006 loss: 0.5233 (0.5226) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [19] [ 90/7269] eta: 0:56:30 lr: 0.000006 loss: 0.4998 (0.5209) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [19] [ 100/7269] eta: 0:55:14 lr: 0.000006 loss: 0.4998 (0.5192) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [19] [ 110/7269] eta: 0:54:11 lr: 0.000006 loss: 0.5040 (0.5195) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [19] [ 120/7269] eta: 0:54:16 lr: 0.000006 loss: 0.5041 (0.5172) time: 0.4212 data: 0.0001 max mem: 52150 +Epoch: [19] [ 130/7269] eta: 0:54:20 lr: 0.000006 loss: 0.5460 (0.5204) time: 0.4704 data: 0.0002 max mem: 52150 +Epoch: [19] [ 140/7269] eta: 0:53:31 lr: 0.000006 loss: 0.5542 (0.5196) time: 0.4201 data: 0.0002 max mem: 52150 +Epoch: [19] [ 150/7269] eta: 0:52:50 lr: 0.000006 loss: 0.5236 (0.5190) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [19] [ 160/7269] eta: 0:53:37 lr: 0.000006 loss: 0.5539 (0.5208) time: 0.4673 data: 0.0002 max mem: 52150 +Epoch: [19] [ 170/7269] eta: 0:55:36 lr: 0.000006 loss: 0.5316 (0.5197) time: 0.6556 data: 0.0001 max mem: 52150 +Epoch: [19] [ 180/7269] eta: 0:54:52 lr: 0.000006 loss: 0.5229 (0.5213) time: 0.5598 data: 0.0002 max mem: 52150 +Epoch: [19] [ 190/7269] eta: 0:55:36 lr: 0.000006 loss: 0.5226 (0.5200) time: 0.4835 data: 0.0002 max mem: 52150 +Epoch: [19] [ 200/7269] eta: 0:54:57 lr: 0.000006 loss: 0.5165 (0.5204) time: 0.4840 data: 0.0002 max mem: 52150 +Epoch: [19] [ 210/7269] eta: 0:54:52 lr: 0.000006 loss: 0.5244 (0.5216) time: 0.4193 data: 0.0001 max mem: 52150 +Epoch: [19] [ 220/7269] eta: 0:54:57 lr: 0.000006 loss: 0.5364 (0.5234) time: 0.4819 data: 0.0001 max mem: 52150 +Epoch: [19] [ 230/7269] eta: 0:54:54 lr: 0.000006 loss: 0.5479 (0.5228) time: 0.4844 data: 0.0001 max mem: 52150 +Epoch: [19] [ 240/7269] eta: 0:54:57 lr: 0.000006 loss: 0.5058 (0.5231) time: 0.4841 data: 0.0001 max mem: 52150 +Epoch: [19] [ 250/7269] eta: 0:54:53 lr: 0.000006 loss: 0.5058 (0.5226) time: 0.4838 data: 0.0001 max mem: 52150 +Epoch: [19] [ 260/7269] eta: 0:54:22 lr: 0.000006 loss: 0.5288 (0.5242) time: 0.4214 data: 0.0002 max mem: 52150 +Epoch: [19] [ 270/7269] eta: 0:53:54 lr: 0.000006 loss: 0.5651 (0.5258) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [19] [ 280/7269] eta: 0:53:26 lr: 0.000006 loss: 0.5575 (0.5255) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [19] [ 290/7269] eta: 0:53:31 lr: 0.000006 loss: 0.5074 (0.5249) time: 0.4352 data: 0.0002 max mem: 52150 +Epoch: [19] [ 300/7269] eta: 0:53:06 lr: 0.000006 loss: 0.5164 (0.5260) time: 0.4353 data: 0.0002 max mem: 52150 +Epoch: [19] [ 310/7269] eta: 0:53:04 lr: 0.000006 loss: 0.5128 (0.5250) time: 0.4192 data: 0.0001 max mem: 52150 +Epoch: [19] [ 320/7269] eta: 0:53:02 lr: 0.000006 loss: 0.5237 (0.5258) time: 0.4689 data: 0.0002 max mem: 52150 +Epoch: [19] [ 330/7269] eta: 0:53:06 lr: 0.000006 loss: 0.5299 (0.5252) time: 0.4847 data: 0.0002 max mem: 52150 +Epoch: [19] [ 340/7269] eta: 0:53:09 lr: 0.000006 loss: 0.5341 (0.5255) time: 0.4967 data: 0.0001 max mem: 52150 +Epoch: [19] [ 350/7269] eta: 0:53:11 lr: 0.000006 loss: 0.5341 (0.5251) time: 0.4959 data: 0.0001 max mem: 52150 +Epoch: [19] [ 360/7269] eta: 0:52:50 lr: 0.000006 loss: 0.5281 (0.5252) time: 0.4344 data: 0.0001 max mem: 52150 +Epoch: [19] [ 370/7269] eta: 0:52:29 lr: 0.000006 loss: 0.5373 (0.5250) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [19] [ 380/7269] eta: 0:52:09 lr: 0.000006 loss: 0.5284 (0.5247) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [19] [ 390/7269] eta: 0:52:07 lr: 0.000006 loss: 0.5464 (0.5254) time: 0.4194 data: 0.0001 max mem: 52150 +Epoch: [19] [ 400/7269] eta: 0:51:48 lr: 0.000006 loss: 0.5368 (0.5254) time: 0.4194 data: 0.0001 max mem: 52150 +Epoch: [19] [ 410/7269] eta: 0:51:46 lr: 0.000006 loss: 0.5230 (0.5255) time: 0.4186 data: 0.0001 max mem: 52150 +Epoch: [19] [ 420/7269] eta: 0:52:05 lr: 0.000006 loss: 0.5554 (0.5261) time: 0.5342 data: 0.0001 max mem: 52150 +Epoch: [19] [ 430/7269] eta: 0:51:47 lr: 0.000006 loss: 0.5532 (0.5267) time: 0.4872 data: 0.0001 max mem: 52150 +Epoch: [19] [ 440/7269] eta: 0:51:30 lr: 0.000006 loss: 0.5532 (0.5270) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [19] [ 450/7269] eta: 0:51:50 lr: 0.000006 loss: 0.5117 (0.5263) time: 0.4941 data: 0.0001 max mem: 52150 +Epoch: [19] [ 460/7269] eta: 0:51:33 lr: 0.000006 loss: 0.5135 (0.5265) time: 0.4943 data: 0.0001 max mem: 52150 +Epoch: [19] [ 470/7269] eta: 0:51:17 lr: 0.000006 loss: 0.5503 (0.5267) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [19] [ 480/7269] eta: 0:51:01 lr: 0.000006 loss: 0.5233 (0.5263) time: 0.3711 data: 0.0001 max mem: 52150 +Epoch: [19] [ 490/7269] eta: 0:51:02 lr: 0.000006 loss: 0.5086 (0.5263) time: 0.4331 data: 0.0001 max mem: 52150 +Epoch: [19] [ 500/7269] eta: 0:50:47 lr: 0.000006 loss: 0.5304 (0.5265) time: 0.4333 data: 0.0002 max mem: 52150 +Epoch: [19] [ 510/7269] eta: 0:50:32 lr: 0.000006 loss: 0.5253 (0.5259) time: 0.3712 data: 0.0001 max mem: 52150 +Epoch: [19] [ 520/7269] eta: 0:50:33 lr: 0.000006 loss: 0.5154 (0.5257) time: 0.4307 data: 0.0001 max mem: 52150 +Epoch: [19] [ 530/7269] eta: 0:50:30 lr: 0.000006 loss: 0.5117 (0.5256) time: 0.4777 data: 0.0001 max mem: 52150 +Epoch: [19] [ 540/7269] eta: 0:50:16 lr: 0.000006 loss: 0.5028 (0.5255) time: 0.4183 data: 0.0001 max mem: 52150 +Epoch: [19] [ 550/7269] eta: 0:50:02 lr: 0.000006 loss: 0.4954 (0.5249) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [19] [ 560/7269] eta: 0:49:49 lr: 0.000006 loss: 0.5198 (0.5255) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [19] [ 570/7269] eta: 0:49:36 lr: 0.000006 loss: 0.5223 (0.5251) time: 0.3713 data: 0.0001 max mem: 52150 +Epoch: [19] [ 580/7269] eta: 0:49:23 lr: 0.000006 loss: 0.5078 (0.5250) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [19] [ 590/7269] eta: 0:49:10 lr: 0.000006 loss: 0.5335 (0.5252) time: 0.3712 data: 0.0002 max mem: 52150 +Epoch: [19] [ 600/7269] eta: 0:48:58 lr: 0.000006 loss: 0.5479 (0.5252) time: 0.3714 data: 0.0001 max mem: 52150 +Epoch: [19] [ 610/7269] eta: 0:48:46 lr: 0.000006 loss: 0.5444 (0.5259) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [19] [ 620/7269] eta: 0:48:34 lr: 0.000006 loss: 0.5594 (0.5259) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [19] [ 630/7269] eta: 0:48:34 lr: 0.000006 loss: 0.5249 (0.5258) time: 0.4217 data: 0.0002 max mem: 52150 +Epoch: [19] [ 640/7269] eta: 0:48:35 lr: 0.000006 loss: 0.5330 (0.5257) time: 0.4831 data: 0.0002 max mem: 52150 +Epoch: [19] [ 650/7269] eta: 0:48:36 lr: 0.000006 loss: 0.5333 (0.5258) time: 0.4942 data: 0.0002 max mem: 52150 +Epoch: [19] [ 660/7269] eta: 0:48:34 lr: 0.000006 loss: 0.5239 (0.5254) time: 0.4809 data: 0.0002 max mem: 52150 +Epoch: [19] [ 670/7269] eta: 0:48:32 lr: 0.000006 loss: 0.5050 (0.5251) time: 0.4656 data: 0.0002 max mem: 52150 +Epoch: [19] [ 680/7269] eta: 0:48:21 lr: 0.000006 loss: 0.5263 (0.5252) time: 0.4182 data: 0.0001 max mem: 52150 +Epoch: [19] [ 690/7269] eta: 0:48:22 lr: 0.000006 loss: 0.5275 (0.5253) time: 0.4332 data: 0.0002 max mem: 52150 +Epoch: [19] [ 700/7269] eta: 0:48:22 lr: 0.000006 loss: 0.5282 (0.5251) time: 0.4939 data: 0.0001 max mem: 52150 +Epoch: [19] [ 710/7269] eta: 0:48:23 lr: 0.000006 loss: 0.5241 (0.5248) time: 0.4933 data: 0.0002 max mem: 52150 +Epoch: [19] [ 720/7269] eta: 0:48:12 lr: 0.000006 loss: 0.5396 (0.5253) time: 0.4320 data: 0.0002 max mem: 52150 +Epoch: [19] [ 730/7269] eta: 0:48:01 lr: 0.000006 loss: 0.5562 (0.5255) time: 0.3714 data: 0.0001 max mem: 52150 +Epoch: [19] [ 740/7269] eta: 0:47:51 lr: 0.000006 loss: 0.5298 (0.5253) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [19] [ 750/7269] eta: 0:47:59 lr: 0.000006 loss: 0.5221 (0.5252) time: 0.4789 data: 0.0001 max mem: 52150 +Epoch: [19] [ 760/7269] eta: 0:48:05 lr: 0.000006 loss: 0.5438 (0.5254) time: 0.5773 data: 0.0001 max mem: 52150 +Epoch: [19] [ 770/7269] eta: 0:47:55 lr: 0.000006 loss: 0.5367 (0.5253) time: 0.4713 data: 0.0001 max mem: 52150 +Epoch: [19] [ 780/7269] eta: 0:47:45 lr: 0.000006 loss: 0.5188 (0.5250) time: 0.3727 data: 0.0001 max mem: 52150 +Epoch: [19] [ 790/7269] eta: 0:47:52 lr: 0.000006 loss: 0.4932 (0.5248) time: 0.4758 data: 0.0001 max mem: 52150 +Epoch: [19] [ 800/7269] eta: 0:47:49 lr: 0.000006 loss: 0.4932 (0.5247) time: 0.5224 data: 0.0001 max mem: 52150 +Epoch: [19] [ 810/7269] eta: 0:47:39 lr: 0.000006 loss: 0.4731 (0.5243) time: 0.4184 data: 0.0001 max mem: 52150 +Epoch: [19] [ 820/7269] eta: 0:47:29 lr: 0.000006 loss: 0.5242 (0.5241) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [19] [ 830/7269] eta: 0:47:28 lr: 0.000006 loss: 0.5318 (0.5241) time: 0.4330 data: 0.0002 max mem: 52150 +Epoch: [19] [ 840/7269] eta: 0:47:19 lr: 0.000006 loss: 0.5318 (0.5239) time: 0.4330 data: 0.0002 max mem: 52150 +Epoch: [19] [ 850/7269] eta: 0:47:25 lr: 0.000006 loss: 0.5262 (0.5239) time: 0.4769 data: 0.0002 max mem: 52150 +Epoch: [19] [ 860/7269] eta: 0:47:15 lr: 0.000006 loss: 0.5292 (0.5241) time: 0.4779 data: 0.0002 max mem: 52150 +Epoch: [19] [ 870/7269] eta: 0:47:12 lr: 0.000006 loss: 0.5549 (0.5245) time: 0.4174 data: 0.0002 max mem: 52150 +Epoch: [19] [ 880/7269] eta: 0:47:16 lr: 0.000006 loss: 0.5587 (0.5249) time: 0.5077 data: 0.0002 max mem: 52150 +Epoch: [19] [ 890/7269] eta: 0:47:13 lr: 0.000006 loss: 0.5732 (0.5252) time: 0.5081 data: 0.0002 max mem: 52150 +Epoch: [19] [ 900/7269] eta: 0:47:03 lr: 0.000006 loss: 0.5597 (0.5255) time: 0.4173 data: 0.0002 max mem: 52150 +Epoch: [19] [ 910/7269] eta: 0:46:54 lr: 0.000006 loss: 0.5235 (0.5251) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [19] [ 920/7269] eta: 0:46:44 lr: 0.000006 loss: 0.5555 (0.5258) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [19] [ 930/7269] eta: 0:46:35 lr: 0.000006 loss: 0.5654 (0.5259) time: 0.3713 data: 0.0001 max mem: 52150 +Epoch: [19] [ 940/7269] eta: 0:46:34 lr: 0.000006 loss: 0.5426 (0.5262) time: 0.4312 data: 0.0001 max mem: 52150 +Epoch: [19] [ 950/7269] eta: 0:46:31 lr: 0.000006 loss: 0.5273 (0.5259) time: 0.4775 data: 0.0001 max mem: 52150 +Epoch: [19] [ 960/7269] eta: 0:46:22 lr: 0.000006 loss: 0.5029 (0.5256) time: 0.4178 data: 0.0001 max mem: 52150 +Epoch: [19] [ 970/7269] eta: 0:46:21 lr: 0.000006 loss: 0.5263 (0.5258) time: 0.4342 data: 0.0002 max mem: 52150 +Epoch: [19] [ 980/7269] eta: 0:46:12 lr: 0.000006 loss: 0.5324 (0.5258) time: 0.4341 data: 0.0002 max mem: 52150 +Epoch: [19] [ 990/7269] eta: 0:46:11 lr: 0.000006 loss: 0.5104 (0.5258) time: 0.4321 data: 0.0002 max mem: 52150 +Epoch: [19] [1000/7269] eta: 0:46:10 lr: 0.000006 loss: 0.5068 (0.5260) time: 0.4931 data: 0.0002 max mem: 52150 +Epoch: [19] [1010/7269] eta: 0:46:07 lr: 0.000006 loss: 0.5680 (0.5260) time: 0.4778 data: 0.0002 max mem: 52150 +Epoch: [19] [1020/7269] eta: 0:46:06 lr: 0.000006 loss: 0.5680 (0.5260) time: 0.4775 data: 0.0002 max mem: 52150 +Epoch: [19] [1030/7269] eta: 0:45:57 lr: 0.000006 loss: 0.5353 (0.5261) time: 0.4325 data: 0.0002 max mem: 52150 +Epoch: [19] [1040/7269] eta: 0:45:48 lr: 0.000006 loss: 0.5353 (0.5261) time: 0.3722 data: 0.0001 max mem: 52150 +Epoch: [19] [1050/7269] eta: 0:45:45 lr: 0.000006 loss: 0.5357 (0.5261) time: 0.4193 data: 0.0001 max mem: 52150 +Epoch: [19] [1060/7269] eta: 0:45:37 lr: 0.000006 loss: 0.5357 (0.5262) time: 0.4190 data: 0.0001 max mem: 52150 +Epoch: [19] [1070/7269] eta: 0:45:29 lr: 0.000006 loss: 0.5356 (0.5262) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [19] [1080/7269] eta: 0:45:20 lr: 0.000006 loss: 0.5560 (0.5264) time: 0.3723 data: 0.0001 max mem: 52150 +Epoch: [19] [1090/7269] eta: 0:45:19 lr: 0.000006 loss: 0.5567 (0.5264) time: 0.4335 data: 0.0002 max mem: 52150 +Epoch: [19] [1100/7269] eta: 0:45:11 lr: 0.000006 loss: 0.5584 (0.5265) time: 0.4334 data: 0.0002 max mem: 52150 +Epoch: [19] [1110/7269] eta: 0:45:08 lr: 0.000006 loss: 0.5320 (0.5266) time: 0.4229 data: 0.0001 max mem: 52150 +Epoch: [19] [1120/7269] eta: 0:45:00 lr: 0.000006 loss: 0.5331 (0.5268) time: 0.4229 data: 0.0002 max mem: 52150 +Epoch: [19] [1130/7269] eta: 0:44:52 lr: 0.000006 loss: 0.5331 (0.5268) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [19] [1140/7269] eta: 0:44:44 lr: 0.000006 loss: 0.5054 (0.5266) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [19] [1150/7269] eta: 0:44:41 lr: 0.000006 loss: 0.4678 (0.5260) time: 0.4203 data: 0.0001 max mem: 52150 +Epoch: [19] [1160/7269] eta: 0:44:40 lr: 0.000006 loss: 0.4731 (0.5260) time: 0.4792 data: 0.0001 max mem: 52150 +Epoch: [19] [1170/7269] eta: 0:44:38 lr: 0.000006 loss: 0.5467 (0.5261) time: 0.4919 data: 0.0001 max mem: 52150 +Epoch: [19] [1180/7269] eta: 0:44:30 lr: 0.000006 loss: 0.5467 (0.5262) time: 0.4336 data: 0.0001 max mem: 52150 +Epoch: [19] [1190/7269] eta: 0:44:27 lr: 0.000006 loss: 0.5233 (0.5261) time: 0.4188 data: 0.0001 max mem: 52150 +Epoch: [19] [1200/7269] eta: 0:44:24 lr: 0.000006 loss: 0.5293 (0.5261) time: 0.4633 data: 0.0001 max mem: 52150 +Epoch: [19] [1210/7269] eta: 0:44:16 lr: 0.000006 loss: 0.5060 (0.5257) time: 0.4168 data: 0.0002 max mem: 52150 +Epoch: [19] [1220/7269] eta: 0:44:20 lr: 0.000006 loss: 0.5052 (0.5255) time: 0.4812 data: 0.0002 max mem: 52150 +Epoch: [19] [1230/7269] eta: 0:44:12 lr: 0.000006 loss: 0.5411 (0.5255) time: 0.4812 data: 0.0002 max mem: 52150 +Epoch: [19] [1240/7269] eta: 0:44:04 lr: 0.000006 loss: 0.5434 (0.5254) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [19] [1250/7269] eta: 0:44:01 lr: 0.000006 loss: 0.5366 (0.5257) time: 0.4181 data: 0.0002 max mem: 52150 +Epoch: [19] [1260/7269] eta: 0:43:53 lr: 0.000006 loss: 0.5353 (0.5255) time: 0.4184 data: 0.0002 max mem: 52150 +Epoch: [19] [1270/7269] eta: 0:43:52 lr: 0.000006 loss: 0.5015 (0.5252) time: 0.4330 data: 0.0002 max mem: 52150 +Epoch: [19] [1280/7269] eta: 0:43:44 lr: 0.000006 loss: 0.5169 (0.5253) time: 0.4327 data: 0.0002 max mem: 52150 +Epoch: [19] [1290/7269] eta: 0:43:48 lr: 0.000006 loss: 0.5359 (0.5252) time: 0.4928 data: 0.0001 max mem: 52150 +Epoch: [19] [1300/7269] eta: 0:43:40 lr: 0.000006 loss: 0.5103 (0.5251) time: 0.4933 data: 0.0001 max mem: 52150 +Epoch: [19] [1310/7269] eta: 0:43:33 lr: 0.000006 loss: 0.5277 (0.5252) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [19] [1320/7269] eta: 0:43:30 lr: 0.000006 loss: 0.5420 (0.5253) time: 0.4219 data: 0.0001 max mem: 52150 +Epoch: [19] [1330/7269] eta: 0:43:23 lr: 0.000006 loss: 0.4907 (0.5249) time: 0.4223 data: 0.0001 max mem: 52150 +Epoch: [19] [1340/7269] eta: 0:43:21 lr: 0.000006 loss: 0.4773 (0.5249) time: 0.4341 data: 0.0002 max mem: 52150 +Epoch: [19] [1350/7269] eta: 0:43:14 lr: 0.000006 loss: 0.5397 (0.5249) time: 0.4338 data: 0.0001 max mem: 52150 +Epoch: [19] [1360/7269] eta: 0:43:12 lr: 0.000006 loss: 0.5393 (0.5249) time: 0.4324 data: 0.0001 max mem: 52150 +Epoch: [19] [1370/7269] eta: 0:43:04 lr: 0.000006 loss: 0.5270 (0.5248) time: 0.4326 data: 0.0002 max mem: 52150 +Epoch: [19] [1380/7269] eta: 0:42:57 lr: 0.000006 loss: 0.5082 (0.5247) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [19] [1390/7269] eta: 0:42:54 lr: 0.000006 loss: 0.5137 (0.5247) time: 0.4183 data: 0.0001 max mem: 52150 +Epoch: [19] [1400/7269] eta: 0:42:47 lr: 0.000006 loss: 0.5425 (0.5248) time: 0.4185 data: 0.0001 max mem: 52150 +Epoch: [19] [1410/7269] eta: 0:42:40 lr: 0.000006 loss: 0.5425 (0.5250) time: 0.3723 data: 0.0001 max mem: 52150 +Epoch: [19] [1420/7269] eta: 0:42:33 lr: 0.000006 loss: 0.5449 (0.5251) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [19] [1430/7269] eta: 0:42:31 lr: 0.000006 loss: 0.5390 (0.5250) time: 0.4339 data: 0.0002 max mem: 52150 +Epoch: [19] [1440/7269] eta: 0:42:24 lr: 0.000006 loss: 0.5026 (0.5249) time: 0.4334 data: 0.0002 max mem: 52150 +Epoch: [19] [1450/7269] eta: 0:42:17 lr: 0.000006 loss: 0.4917 (0.5245) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [19] [1460/7269] eta: 0:42:15 lr: 0.000006 loss: 0.4807 (0.5243) time: 0.4323 data: 0.0002 max mem: 52150 +Epoch: [19] [1470/7269] eta: 0:42:16 lr: 0.000006 loss: 0.4879 (0.5243) time: 0.5366 data: 0.0002 max mem: 52150 +Epoch: [19] [1480/7269] eta: 0:42:09 lr: 0.000006 loss: 0.5165 (0.5244) time: 0.4766 data: 0.0002 max mem: 52150 +Epoch: [19] [1490/7269] eta: 0:42:10 lr: 0.000006 loss: 0.5480 (0.5243) time: 0.4770 data: 0.0002 max mem: 52150 +Epoch: [19] [1500/7269] eta: 0:42:07 lr: 0.000006 loss: 0.4873 (0.5241) time: 0.5218 data: 0.0001 max mem: 52150 +Epoch: [19] [1510/7269] eta: 0:42:03 lr: 0.000006 loss: 0.4930 (0.5240) time: 0.4628 data: 0.0001 max mem: 52150 +Epoch: [19] [1520/7269] eta: 0:42:00 lr: 0.000006 loss: 0.5225 (0.5241) time: 0.4628 data: 0.0002 max mem: 52150 +Epoch: [19] [1530/7269] eta: 0:41:53 lr: 0.000006 loss: 0.5401 (0.5242) time: 0.4170 data: 0.0002 max mem: 52150 +Epoch: [19] [1540/7269] eta: 0:41:46 lr: 0.000006 loss: 0.5256 (0.5241) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [19] [1550/7269] eta: 0:41:39 lr: 0.000006 loss: 0.5234 (0.5242) time: 0.3728 data: 0.0002 max mem: 52150 +Epoch: [19] [1560/7269] eta: 0:41:33 lr: 0.000006 loss: 0.5296 (0.5242) time: 0.3729 data: 0.0002 max mem: 52150 +Epoch: [19] [1570/7269] eta: 0:41:30 lr: 0.000006 loss: 0.5319 (0.5243) time: 0.4208 data: 0.0001 max mem: 52150 +Epoch: [19] [1580/7269] eta: 0:41:23 lr: 0.000006 loss: 0.5405 (0.5244) time: 0.4198 data: 0.0001 max mem: 52150 +Epoch: [19] [1590/7269] eta: 0:41:16 lr: 0.000006 loss: 0.5405 (0.5244) time: 0.3707 data: 0.0001 max mem: 52150 +Epoch: [19] [1600/7269] eta: 0:41:13 lr: 0.000006 loss: 0.5386 (0.5244) time: 0.4175 data: 0.0002 max mem: 52150 +Epoch: [19] [1610/7269] eta: 0:41:10 lr: 0.000006 loss: 0.4896 (0.5240) time: 0.4783 data: 0.0002 max mem: 52150 +Epoch: [19] [1620/7269] eta: 0:41:04 lr: 0.000006 loss: 0.4896 (0.5239) time: 0.4329 data: 0.0001 max mem: 52150 +Epoch: [19] [1630/7269] eta: 0:40:57 lr: 0.000006 loss: 0.5270 (0.5240) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [19] [1640/7269] eta: 0:40:51 lr: 0.000006 loss: 0.5612 (0.5242) time: 0.3713 data: 0.0001 max mem: 52150 +Epoch: [19] [1650/7269] eta: 0:40:44 lr: 0.000006 loss: 0.5572 (0.5243) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [19] [1660/7269] eta: 0:40:45 lr: 0.000006 loss: 0.5284 (0.5241) time: 0.4780 data: 0.0001 max mem: 52150 +Epoch: [19] [1670/7269] eta: 0:40:41 lr: 0.000006 loss: 0.5363 (0.5241) time: 0.5229 data: 0.0002 max mem: 52150 +Epoch: [19] [1680/7269] eta: 0:40:35 lr: 0.000006 loss: 0.5365 (0.5241) time: 0.4167 data: 0.0002 max mem: 52150 +Epoch: [19] [1690/7269] eta: 0:40:28 lr: 0.000006 loss: 0.5433 (0.5242) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [19] [1700/7269] eta: 0:40:22 lr: 0.000006 loss: 0.5205 (0.5242) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [19] [1710/7269] eta: 0:40:15 lr: 0.000006 loss: 0.5205 (0.5242) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [19] [1720/7269] eta: 0:40:09 lr: 0.000006 loss: 0.5428 (0.5244) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [19] [1730/7269] eta: 0:40:07 lr: 0.000006 loss: 0.5297 (0.5245) time: 0.4335 data: 0.0002 max mem: 52150 +Epoch: [19] [1740/7269] eta: 0:40:00 lr: 0.000006 loss: 0.5448 (0.5247) time: 0.4333 data: 0.0002 max mem: 52150 +Epoch: [19] [1750/7269] eta: 0:39:54 lr: 0.000006 loss: 0.5448 (0.5249) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [19] [1760/7269] eta: 0:39:48 lr: 0.000006 loss: 0.5406 (0.5248) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [19] [1770/7269] eta: 0:39:45 lr: 0.000006 loss: 0.5302 (0.5248) time: 0.4335 data: 0.0002 max mem: 52150 +Epoch: [19] [1780/7269] eta: 0:39:42 lr: 0.000006 loss: 0.5326 (0.5249) time: 0.4828 data: 0.0001 max mem: 52150 +Epoch: [19] [1790/7269] eta: 0:39:40 lr: 0.000006 loss: 0.5522 (0.5251) time: 0.4820 data: 0.0001 max mem: 52150 +Epoch: [19] [1800/7269] eta: 0:39:33 lr: 0.000006 loss: 0.5566 (0.5251) time: 0.4325 data: 0.0001 max mem: 52150 +Epoch: [19] [1810/7269] eta: 0:39:33 lr: 0.000006 loss: 0.5314 (0.5252) time: 0.4677 data: 0.0001 max mem: 52150 +Epoch: [19] [1820/7269] eta: 0:39:27 lr: 0.000006 loss: 0.5139 (0.5251) time: 0.4684 data: 0.0001 max mem: 52150 +Epoch: [19] [1830/7269] eta: 0:39:23 lr: 0.000006 loss: 0.5004 (0.5250) time: 0.4182 data: 0.0001 max mem: 52150 +Epoch: [19] [1840/7269] eta: 0:39:20 lr: 0.000006 loss: 0.5394 (0.5251) time: 0.4664 data: 0.0002 max mem: 52150 +Epoch: [19] [1850/7269] eta: 0:39:16 lr: 0.000006 loss: 0.5570 (0.5251) time: 0.4660 data: 0.0002 max mem: 52150 +Epoch: [19] [1860/7269] eta: 0:39:10 lr: 0.000006 loss: 0.5378 (0.5251) time: 0.4175 data: 0.0001 max mem: 52150 +Epoch: [19] [1870/7269] eta: 0:39:10 lr: 0.000006 loss: 0.5345 (0.5250) time: 0.4795 data: 0.0001 max mem: 52150 +Epoch: [19] [1880/7269] eta: 0:39:04 lr: 0.000006 loss: 0.5217 (0.5249) time: 0.4792 data: 0.0001 max mem: 52150 +Epoch: [19] [1890/7269] eta: 0:38:58 lr: 0.000006 loss: 0.5377 (0.5249) time: 0.3713 data: 0.0001 max mem: 52150 +Epoch: [19] [1900/7269] eta: 0:39:01 lr: 0.000006 loss: 0.5374 (0.5248) time: 0.5313 data: 0.0001 max mem: 52150 +Epoch: [19] [1910/7269] eta: 0:38:55 lr: 0.000006 loss: 0.5379 (0.5249) time: 0.5315 data: 0.0001 max mem: 52150 +Epoch: [19] [1920/7269] eta: 0:38:48 lr: 0.000006 loss: 0.5288 (0.5248) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [19] [1930/7269] eta: 0:38:42 lr: 0.000006 loss: 0.5175 (0.5248) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [19] [1940/7269] eta: 0:38:36 lr: 0.000006 loss: 0.5121 (0.5247) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [19] [1950/7269] eta: 0:38:30 lr: 0.000006 loss: 0.5012 (0.5246) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [19] [1960/7269] eta: 0:38:24 lr: 0.000006 loss: 0.5119 (0.5246) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [19] [1970/7269] eta: 0:38:24 lr: 0.000006 loss: 0.5546 (0.5248) time: 0.4772 data: 0.0001 max mem: 52150 +Epoch: [19] [1980/7269] eta: 0:38:23 lr: 0.000006 loss: 0.5672 (0.5248) time: 0.5837 data: 0.0001 max mem: 52150 +Epoch: [19] [1990/7269] eta: 0:38:21 lr: 0.000006 loss: 0.5480 (0.5249) time: 0.5411 data: 0.0001 max mem: 52150 +Epoch: [19] [2000/7269] eta: 0:38:17 lr: 0.000006 loss: 0.5368 (0.5248) time: 0.4799 data: 0.0001 max mem: 52150 +Epoch: [19] [2010/7269] eta: 0:38:13 lr: 0.000006 loss: 0.5230 (0.5249) time: 0.4630 data: 0.0002 max mem: 52150 +Epoch: [19] [2020/7269] eta: 0:38:07 lr: 0.000006 loss: 0.5348 (0.5251) time: 0.4181 data: 0.0002 max mem: 52150 +Epoch: [19] [2030/7269] eta: 0:38:01 lr: 0.000006 loss: 0.5828 (0.5254) time: 0.3726 data: 0.0001 max mem: 52150 +Epoch: [19] [2040/7269] eta: 0:38:01 lr: 0.000006 loss: 0.5707 (0.5255) time: 0.4779 data: 0.0001 max mem: 52150 +Epoch: [19] [2050/7269] eta: 0:37:55 lr: 0.000006 loss: 0.5365 (0.5256) time: 0.4776 data: 0.0001 max mem: 52150 +Epoch: [19] [2060/7269] eta: 0:37:51 lr: 0.000006 loss: 0.5456 (0.5257) time: 0.4166 data: 0.0001 max mem: 52150 +Epoch: [19] [2070/7269] eta: 0:37:48 lr: 0.000006 loss: 0.5613 (0.5257) time: 0.4760 data: 0.0001 max mem: 52150 +Epoch: [19] [2080/7269] eta: 0:37:42 lr: 0.000006 loss: 0.5664 (0.5260) time: 0.4312 data: 0.0001 max mem: 52150 +Epoch: [19] [2090/7269] eta: 0:37:36 lr: 0.000006 loss: 0.5179 (0.5258) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [19] [2100/7269] eta: 0:37:33 lr: 0.000006 loss: 0.4998 (0.5258) time: 0.4198 data: 0.0002 max mem: 52150 +Epoch: [19] [2110/7269] eta: 0:37:27 lr: 0.000006 loss: 0.5175 (0.5257) time: 0.4200 data: 0.0002 max mem: 52150 +Epoch: [19] [2120/7269] eta: 0:37:21 lr: 0.000006 loss: 0.4702 (0.5254) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [19] [2130/7269] eta: 0:37:15 lr: 0.000006 loss: 0.4665 (0.5254) time: 0.3724 data: 0.0001 max mem: 52150 +Epoch: [19] [2140/7269] eta: 0:37:09 lr: 0.000006 loss: 0.5134 (0.5253) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [19] [2150/7269] eta: 0:37:05 lr: 0.000006 loss: 0.5134 (0.5253) time: 0.4184 data: 0.0002 max mem: 52150 +Epoch: [19] [2160/7269] eta: 0:37:00 lr: 0.000006 loss: 0.5413 (0.5253) time: 0.4178 data: 0.0001 max mem: 52150 +Epoch: [19] [2170/7269] eta: 0:36:54 lr: 0.000006 loss: 0.5452 (0.5253) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [19] [2180/7269] eta: 0:36:50 lr: 0.000006 loss: 0.5435 (0.5255) time: 0.4196 data: 0.0001 max mem: 52150 +Epoch: [19] [2190/7269] eta: 0:36:44 lr: 0.000006 loss: 0.5554 (0.5256) time: 0.4198 data: 0.0001 max mem: 52150 +Epoch: [19] [2200/7269] eta: 0:36:42 lr: 0.000006 loss: 0.5257 (0.5256) time: 0.4333 data: 0.0002 max mem: 52150 +Epoch: [19] [2210/7269] eta: 0:36:36 lr: 0.000006 loss: 0.5257 (0.5257) time: 0.4335 data: 0.0002 max mem: 52150 +Epoch: [19] [2220/7269] eta: 0:36:33 lr: 0.000006 loss: 0.5144 (0.5255) time: 0.4333 data: 0.0001 max mem: 52150 +Epoch: [19] [2230/7269] eta: 0:36:29 lr: 0.000006 loss: 0.5024 (0.5255) time: 0.4784 data: 0.0001 max mem: 52150 +Epoch: [19] [2240/7269] eta: 0:36:23 lr: 0.000006 loss: 0.5323 (0.5255) time: 0.4173 data: 0.0001 max mem: 52150 +Epoch: [19] [2250/7269] eta: 0:36:18 lr: 0.000006 loss: 0.5233 (0.5254) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [19] [2260/7269] eta: 0:36:12 lr: 0.000006 loss: 0.5251 (0.5255) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [19] [2270/7269] eta: 0:36:06 lr: 0.000006 loss: 0.5451 (0.5255) time: 0.3722 data: 0.0001 max mem: 52150 +Epoch: [19] [2280/7269] eta: 0:36:03 lr: 0.000006 loss: 0.5456 (0.5255) time: 0.4327 data: 0.0002 max mem: 52150 +Epoch: [19] [2290/7269] eta: 0:35:57 lr: 0.000006 loss: 0.5515 (0.5256) time: 0.4323 data: 0.0002 max mem: 52150 +Epoch: [19] [2300/7269] eta: 0:35:54 lr: 0.000006 loss: 0.5240 (0.5256) time: 0.4323 data: 0.0002 max mem: 52150 +Epoch: [19] [2310/7269] eta: 0:35:49 lr: 0.000006 loss: 0.5278 (0.5256) time: 0.4327 data: 0.0002 max mem: 52150 +Epoch: [19] [2320/7269] eta: 0:35:45 lr: 0.000006 loss: 0.5278 (0.5255) time: 0.4194 data: 0.0002 max mem: 52150 +Epoch: [19] [2330/7269] eta: 0:35:42 lr: 0.000006 loss: 0.5066 (0.5255) time: 0.4806 data: 0.0002 max mem: 52150 +Epoch: [19] [2340/7269] eta: 0:35:38 lr: 0.000006 loss: 0.5234 (0.5255) time: 0.4789 data: 0.0001 max mem: 52150 +Epoch: [19] [2350/7269] eta: 0:35:33 lr: 0.000006 loss: 0.5369 (0.5256) time: 0.4176 data: 0.0001 max mem: 52150 +Epoch: [19] [2360/7269] eta: 0:35:29 lr: 0.000006 loss: 0.5496 (0.5257) time: 0.4187 data: 0.0001 max mem: 52150 +Epoch: [19] [2370/7269] eta: 0:35:25 lr: 0.000006 loss: 0.5212 (0.5256) time: 0.4679 data: 0.0002 max mem: 52150 +Epoch: [19] [2380/7269] eta: 0:35:20 lr: 0.000006 loss: 0.5103 (0.5255) time: 0.4214 data: 0.0002 max mem: 52150 +Epoch: [19] [2390/7269] eta: 0:35:19 lr: 0.000006 loss: 0.5194 (0.5256) time: 0.4831 data: 0.0002 max mem: 52150 +Epoch: [19] [2400/7269] eta: 0:35:13 lr: 0.000006 loss: 0.5468 (0.5257) time: 0.4831 data: 0.0002 max mem: 52150 +Epoch: [19] [2410/7269] eta: 0:35:08 lr: 0.000006 loss: 0.5373 (0.5257) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [19] [2420/7269] eta: 0:35:04 lr: 0.000006 loss: 0.5269 (0.5258) time: 0.4198 data: 0.0002 max mem: 52150 +Epoch: [19] [2430/7269] eta: 0:35:00 lr: 0.000006 loss: 0.5371 (0.5258) time: 0.4688 data: 0.0002 max mem: 52150 +Epoch: [19] [2440/7269] eta: 0:34:55 lr: 0.000006 loss: 0.5385 (0.5258) time: 0.4205 data: 0.0002 max mem: 52150 +Epoch: [19] [2450/7269] eta: 0:34:49 lr: 0.000006 loss: 0.5364 (0.5258) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [19] [2460/7269] eta: 0:34:44 lr: 0.000006 loss: 0.5255 (0.5258) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [19] [2470/7269] eta: 0:34:40 lr: 0.000006 loss: 0.5039 (0.5257) time: 0.4326 data: 0.0002 max mem: 52150 +Epoch: [19] [2480/7269] eta: 0:34:37 lr: 0.000006 loss: 0.4822 (0.5255) time: 0.4775 data: 0.0001 max mem: 52150 +Epoch: [19] [2490/7269] eta: 0:34:33 lr: 0.000006 loss: 0.5039 (0.5254) time: 0.4618 data: 0.0001 max mem: 52150 +Epoch: [19] [2500/7269] eta: 0:34:29 lr: 0.000006 loss: 0.5438 (0.5255) time: 0.4623 data: 0.0002 max mem: 52150 +Epoch: [19] [2510/7269] eta: 0:34:27 lr: 0.000006 loss: 0.5530 (0.5255) time: 0.5145 data: 0.0002 max mem: 52150 +Epoch: [19] [2520/7269] eta: 0:34:22 lr: 0.000006 loss: 0.5472 (0.5255) time: 0.4684 data: 0.0002 max mem: 52150 +Epoch: [19] [2530/7269] eta: 0:34:16 lr: 0.000006 loss: 0.4986 (0.5254) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [19] [2540/7269] eta: 0:34:11 lr: 0.000006 loss: 0.5057 (0.5255) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [19] [2550/7269] eta: 0:34:05 lr: 0.000006 loss: 0.5282 (0.5254) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [19] [2560/7269] eta: 0:34:02 lr: 0.000006 loss: 0.5170 (0.5254) time: 0.4304 data: 0.0002 max mem: 52150 +Epoch: [19] [2570/7269] eta: 0:33:59 lr: 0.000006 loss: 0.5377 (0.5254) time: 0.4920 data: 0.0002 max mem: 52150 +Epoch: [19] [2580/7269] eta: 0:33:57 lr: 0.000006 loss: 0.5544 (0.5254) time: 0.5289 data: 0.0002 max mem: 52150 +Epoch: [19] [2590/7269] eta: 0:33:51 lr: 0.000006 loss: 0.5424 (0.5255) time: 0.4669 data: 0.0002 max mem: 52150 +Epoch: [19] [2600/7269] eta: 0:33:47 lr: 0.000006 loss: 0.5708 (0.5256) time: 0.4161 data: 0.0002 max mem: 52150 +Epoch: [19] [2610/7269] eta: 0:33:44 lr: 0.000006 loss: 0.5708 (0.5257) time: 0.4767 data: 0.0002 max mem: 52150 +Epoch: [19] [2620/7269] eta: 0:33:40 lr: 0.000006 loss: 0.5475 (0.5258) time: 0.4810 data: 0.0002 max mem: 52150 +Epoch: [19] [2630/7269] eta: 0:33:36 lr: 0.000006 loss: 0.5594 (0.5259) time: 0.4658 data: 0.0001 max mem: 52150 +Epoch: [19] [2640/7269] eta: 0:33:31 lr: 0.000006 loss: 0.5594 (0.5258) time: 0.4170 data: 0.0001 max mem: 52150 +Epoch: [19] [2650/7269] eta: 0:33:26 lr: 0.000006 loss: 0.5642 (0.5259) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [19] [2660/7269] eta: 0:33:20 lr: 0.000006 loss: 0.4983 (0.5256) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [19] [2670/7269] eta: 0:33:15 lr: 0.000006 loss: 0.5261 (0.5257) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [19] [2680/7269] eta: 0:33:09 lr: 0.000006 loss: 0.5393 (0.5257) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [19] [2690/7269] eta: 0:33:06 lr: 0.000006 loss: 0.5412 (0.5259) time: 0.4321 data: 0.0002 max mem: 52150 +Epoch: [19] [2700/7269] eta: 0:33:01 lr: 0.000006 loss: 0.5854 (0.5260) time: 0.4326 data: 0.0002 max mem: 52150 +Epoch: [19] [2710/7269] eta: 0:32:57 lr: 0.000006 loss: 0.5554 (0.5260) time: 0.4170 data: 0.0001 max mem: 52150 +Epoch: [19] [2720/7269] eta: 0:32:51 lr: 0.000006 loss: 0.5472 (0.5261) time: 0.4166 data: 0.0001 max mem: 52150 +Epoch: [19] [2730/7269] eta: 0:32:47 lr: 0.000006 loss: 0.5472 (0.5261) time: 0.4159 data: 0.0002 max mem: 52150 +Epoch: [19] [2740/7269] eta: 0:32:42 lr: 0.000006 loss: 0.5496 (0.5262) time: 0.4162 data: 0.0002 max mem: 52150 +Epoch: [19] [2750/7269] eta: 0:32:40 lr: 0.000006 loss: 0.5496 (0.5261) time: 0.4782 data: 0.0002 max mem: 52150 +Epoch: [19] [2760/7269] eta: 0:32:35 lr: 0.000006 loss: 0.5188 (0.5261) time: 0.4774 data: 0.0002 max mem: 52150 +Epoch: [19] [2770/7269] eta: 0:32:32 lr: 0.000006 loss: 0.5451 (0.5261) time: 0.4332 data: 0.0001 max mem: 52150 +Epoch: [19] [2780/7269] eta: 0:32:26 lr: 0.000006 loss: 0.4700 (0.5258) time: 0.4337 data: 0.0001 max mem: 52150 +Epoch: [19] [2790/7269] eta: 0:32:21 lr: 0.000006 loss: 0.5012 (0.5260) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [19] [2800/7269] eta: 0:32:17 lr: 0.000006 loss: 0.5429 (0.5260) time: 0.4308 data: 0.0002 max mem: 52150 +Epoch: [19] [2810/7269] eta: 0:32:12 lr: 0.000006 loss: 0.5055 (0.5259) time: 0.4310 data: 0.0002 max mem: 52150 +Epoch: [19] [2820/7269] eta: 0:32:07 lr: 0.000006 loss: 0.4985 (0.5259) time: 0.3725 data: 0.0001 max mem: 52150 +Epoch: [19] [2830/7269] eta: 0:32:02 lr: 0.000006 loss: 0.5022 (0.5259) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [19] [2840/7269] eta: 0:31:56 lr: 0.000006 loss: 0.5448 (0.5260) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [19] [2850/7269] eta: 0:31:54 lr: 0.000006 loss: 0.5448 (0.5261) time: 0.4779 data: 0.0001 max mem: 52150 +Epoch: [19] [2860/7269] eta: 0:31:53 lr: 0.000006 loss: 0.5301 (0.5260) time: 0.5997 data: 0.0002 max mem: 52150 +Epoch: [19] [2870/7269] eta: 0:31:47 lr: 0.000006 loss: 0.5503 (0.5261) time: 0.4938 data: 0.0002 max mem: 52150 +Epoch: [19] [2880/7269] eta: 0:31:44 lr: 0.000006 loss: 0.5520 (0.5261) time: 0.4180 data: 0.0001 max mem: 52150 +Epoch: [19] [2890/7269] eta: 0:31:40 lr: 0.000006 loss: 0.4920 (0.5260) time: 0.4802 data: 0.0001 max mem: 52150 +Epoch: [19] [2900/7269] eta: 0:31:35 lr: 0.000006 loss: 0.4952 (0.5259) time: 0.4337 data: 0.0001 max mem: 52150 +Epoch: [19] [2910/7269] eta: 0:31:30 lr: 0.000006 loss: 0.4941 (0.5258) time: 0.3728 data: 0.0001 max mem: 52150 +Epoch: [19] [2920/7269] eta: 0:31:26 lr: 0.000006 loss: 0.5163 (0.5258) time: 0.4231 data: 0.0001 max mem: 52150 +Epoch: [19] [2930/7269] eta: 0:31:21 lr: 0.000006 loss: 0.5163 (0.5258) time: 0.4216 data: 0.0002 max mem: 52150 +Epoch: [19] [2940/7269] eta: 0:31:15 lr: 0.000006 loss: 0.5031 (0.5257) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [19] [2950/7269] eta: 0:31:12 lr: 0.000006 loss: 0.5395 (0.5257) time: 0.4168 data: 0.0002 max mem: 52150 +Epoch: [19] [2960/7269] eta: 0:31:06 lr: 0.000006 loss: 0.5077 (0.5256) time: 0.4167 data: 0.0001 max mem: 52150 +Epoch: [19] [2970/7269] eta: 0:31:02 lr: 0.000006 loss: 0.5029 (0.5256) time: 0.4173 data: 0.0002 max mem: 52150 +Epoch: [19] [2980/7269] eta: 0:30:57 lr: 0.000006 loss: 0.4969 (0.5255) time: 0.4177 data: 0.0002 max mem: 52150 +Epoch: [19] [2990/7269] eta: 0:30:52 lr: 0.000006 loss: 0.5095 (0.5256) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [19] [3000/7269] eta: 0:30:50 lr: 0.000006 loss: 0.5450 (0.5257) time: 0.4686 data: 0.0002 max mem: 52150 +Epoch: [19] [3010/7269] eta: 0:30:44 lr: 0.000006 loss: 0.5228 (0.5257) time: 0.4689 data: 0.0001 max mem: 52150 +Epoch: [19] [3020/7269] eta: 0:30:39 lr: 0.000006 loss: 0.5106 (0.5256) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [19] [3030/7269] eta: 0:30:36 lr: 0.000006 loss: 0.5264 (0.5256) time: 0.4314 data: 0.0002 max mem: 52150 +Epoch: [19] [3040/7269] eta: 0:30:34 lr: 0.000006 loss: 0.5507 (0.5257) time: 0.5534 data: 0.0002 max mem: 52150 +Epoch: [19] [3050/7269] eta: 0:30:29 lr: 0.000006 loss: 0.5725 (0.5259) time: 0.4936 data: 0.0002 max mem: 52150 +Epoch: [19] [3060/7269] eta: 0:30:26 lr: 0.000006 loss: 0.5580 (0.5259) time: 0.4680 data: 0.0002 max mem: 52150 +Epoch: [19] [3070/7269] eta: 0:30:24 lr: 0.000006 loss: 0.5426 (0.5259) time: 0.5736 data: 0.0002 max mem: 52150 +Epoch: [19] [3080/7269] eta: 0:30:19 lr: 0.000006 loss: 0.5697 (0.5260) time: 0.4779 data: 0.0002 max mem: 52150 +Epoch: [19] [3090/7269] eta: 0:30:16 lr: 0.000006 loss: 0.5424 (0.5260) time: 0.4791 data: 0.0002 max mem: 52150 +Epoch: [19] [3100/7269] eta: 0:30:11 lr: 0.000006 loss: 0.5398 (0.5260) time: 0.4787 data: 0.0002 max mem: 52150 +Epoch: [19] [3110/7269] eta: 0:30:06 lr: 0.000006 loss: 0.5380 (0.5260) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [19] [3120/7269] eta: 0:30:02 lr: 0.000006 loss: 0.5138 (0.5260) time: 0.4212 data: 0.0002 max mem: 52150 +Epoch: [19] [3130/7269] eta: 0:29:57 lr: 0.000006 loss: 0.5376 (0.5260) time: 0.4207 data: 0.0002 max mem: 52150 +Epoch: [19] [3140/7269] eta: 0:29:52 lr: 0.000006 loss: 0.5264 (0.5260) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [19] [3150/7269] eta: 0:29:47 lr: 0.000006 loss: 0.5302 (0.5260) time: 0.3722 data: 0.0001 max mem: 52150 +Epoch: [19] [3160/7269] eta: 0:29:41 lr: 0.000006 loss: 0.5302 (0.5260) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [19] [3170/7269] eta: 0:29:39 lr: 0.000006 loss: 0.5343 (0.5260) time: 0.4826 data: 0.0001 max mem: 52150 +Epoch: [19] [3180/7269] eta: 0:29:35 lr: 0.000006 loss: 0.5397 (0.5260) time: 0.5278 data: 0.0001 max mem: 52150 +Epoch: [19] [3190/7269] eta: 0:29:31 lr: 0.000006 loss: 0.5200 (0.5260) time: 0.4618 data: 0.0002 max mem: 52150 +Epoch: [19] [3200/7269] eta: 0:29:27 lr: 0.000006 loss: 0.5612 (0.5261) time: 0.4622 data: 0.0002 max mem: 52150 +Epoch: [19] [3210/7269] eta: 0:29:24 lr: 0.000006 loss: 0.5126 (0.5260) time: 0.4788 data: 0.0001 max mem: 52150 +Epoch: [19] [3220/7269] eta: 0:29:18 lr: 0.000006 loss: 0.4995 (0.5260) time: 0.4326 data: 0.0001 max mem: 52150 +Epoch: [19] [3230/7269] eta: 0:29:13 lr: 0.000006 loss: 0.5119 (0.5260) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [19] [3240/7269] eta: 0:29:09 lr: 0.000006 loss: 0.5460 (0.5261) time: 0.4219 data: 0.0001 max mem: 52150 +Epoch: [19] [3250/7269] eta: 0:29:04 lr: 0.000006 loss: 0.5578 (0.5261) time: 0.4218 data: 0.0001 max mem: 52150 +Epoch: [19] [3260/7269] eta: 0:28:59 lr: 0.000006 loss: 0.5550 (0.5262) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [19] [3270/7269] eta: 0:28:54 lr: 0.000006 loss: 0.5720 (0.5263) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [19] [3280/7269] eta: 0:28:51 lr: 0.000006 loss: 0.5439 (0.5263) time: 0.4330 data: 0.0002 max mem: 52150 +Epoch: [19] [3290/7269] eta: 0:28:45 lr: 0.000006 loss: 0.5189 (0.5262) time: 0.4330 data: 0.0002 max mem: 52150 +Epoch: [19] [3300/7269] eta: 0:28:41 lr: 0.000006 loss: 0.5035 (0.5262) time: 0.4189 data: 0.0001 max mem: 52150 +Epoch: [19] [3310/7269] eta: 0:28:36 lr: 0.000006 loss: 0.5411 (0.5263) time: 0.4188 data: 0.0002 max mem: 52150 +Epoch: [19] [3320/7269] eta: 0:28:34 lr: 0.000006 loss: 0.5410 (0.5263) time: 0.4929 data: 0.0002 max mem: 52150 +Epoch: [19] [3330/7269] eta: 0:28:29 lr: 0.000006 loss: 0.5213 (0.5263) time: 0.4932 data: 0.0001 max mem: 52150 +Epoch: [19] [3340/7269] eta: 0:28:24 lr: 0.000006 loss: 0.5164 (0.5262) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [19] [3350/7269] eta: 0:28:19 lr: 0.000006 loss: 0.4995 (0.5262) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [19] [3360/7269] eta: 0:28:15 lr: 0.000006 loss: 0.5062 (0.5261) time: 0.4360 data: 0.0002 max mem: 52150 +Epoch: [19] [3370/7269] eta: 0:28:10 lr: 0.000006 loss: 0.5401 (0.5262) time: 0.4355 data: 0.0002 max mem: 52150 +Epoch: [19] [3380/7269] eta: 0:28:08 lr: 0.000006 loss: 0.5363 (0.5260) time: 0.4787 data: 0.0002 max mem: 52150 +Epoch: [19] [3390/7269] eta: 0:28:03 lr: 0.000006 loss: 0.4902 (0.5261) time: 0.4782 data: 0.0002 max mem: 52150 +Epoch: [19] [3400/7269] eta: 0:27:58 lr: 0.000006 loss: 0.5542 (0.5261) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [19] [3410/7269] eta: 0:27:54 lr: 0.000006 loss: 0.5148 (0.5260) time: 0.4334 data: 0.0001 max mem: 52150 +Epoch: [19] [3420/7269] eta: 0:27:50 lr: 0.000006 loss: 0.5082 (0.5260) time: 0.4805 data: 0.0001 max mem: 52150 +Epoch: [19] [3430/7269] eta: 0:27:45 lr: 0.000006 loss: 0.5425 (0.5261) time: 0.4189 data: 0.0001 max mem: 52150 +Epoch: [19] [3440/7269] eta: 0:27:40 lr: 0.000006 loss: 0.5081 (0.5260) time: 0.3721 data: 0.0001 max mem: 52150 +Epoch: [19] [3450/7269] eta: 0:27:35 lr: 0.000006 loss: 0.5114 (0.5260) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [19] [3460/7269] eta: 0:27:30 lr: 0.000006 loss: 0.5135 (0.5259) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [19] [3470/7269] eta: 0:27:25 lr: 0.000006 loss: 0.5135 (0.5258) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [19] [3480/7269] eta: 0:27:21 lr: 0.000006 loss: 0.5266 (0.5258) time: 0.4174 data: 0.0002 max mem: 52150 +Epoch: [19] [3490/7269] eta: 0:27:16 lr: 0.000006 loss: 0.5498 (0.5259) time: 0.4176 data: 0.0002 max mem: 52150 +Epoch: [19] [3500/7269] eta: 0:27:11 lr: 0.000006 loss: 0.5561 (0.5259) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [19] [3510/7269] eta: 0:27:07 lr: 0.000006 loss: 0.5199 (0.5258) time: 0.4310 data: 0.0002 max mem: 52150 +Epoch: [19] [3520/7269] eta: 0:27:03 lr: 0.000006 loss: 0.5263 (0.5258) time: 0.4816 data: 0.0002 max mem: 52150 +Epoch: [19] [3530/7269] eta: 0:26:58 lr: 0.000006 loss: 0.5355 (0.5258) time: 0.4224 data: 0.0002 max mem: 52150 +Epoch: [19] [3540/7269] eta: 0:26:53 lr: 0.000006 loss: 0.5162 (0.5257) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [19] [3550/7269] eta: 0:26:48 lr: 0.000006 loss: 0.5157 (0.5257) time: 0.3727 data: 0.0002 max mem: 52150 +Epoch: [19] [3560/7269] eta: 0:26:43 lr: 0.000006 loss: 0.5421 (0.5257) time: 0.3732 data: 0.0002 max mem: 52150 +Epoch: [19] [3570/7269] eta: 0:26:41 lr: 0.000006 loss: 0.5462 (0.5257) time: 0.4815 data: 0.0002 max mem: 52150 +Epoch: [19] [3580/7269] eta: 0:26:37 lr: 0.000006 loss: 0.5417 (0.5257) time: 0.5298 data: 0.0002 max mem: 52150 +Epoch: [19] [3590/7269] eta: 0:26:32 lr: 0.000006 loss: 0.5157 (0.5257) time: 0.4206 data: 0.0002 max mem: 52150 +Epoch: [19] [3600/7269] eta: 0:26:27 lr: 0.000006 loss: 0.5292 (0.5257) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [19] [3610/7269] eta: 0:26:24 lr: 0.000006 loss: 0.5337 (0.5257) time: 0.4790 data: 0.0002 max mem: 52150 +Epoch: [19] [3620/7269] eta: 0:26:19 lr: 0.000006 loss: 0.5337 (0.5257) time: 0.4791 data: 0.0002 max mem: 52150 +Epoch: [19] [3630/7269] eta: 0:26:16 lr: 0.000006 loss: 0.4880 (0.5257) time: 0.4326 data: 0.0002 max mem: 52150 +Epoch: [19] [3640/7269] eta: 0:26:12 lr: 0.000006 loss: 0.5315 (0.5257) time: 0.4799 data: 0.0002 max mem: 52150 +Epoch: [19] [3650/7269] eta: 0:26:07 lr: 0.000006 loss: 0.5490 (0.5258) time: 0.4644 data: 0.0002 max mem: 52150 +Epoch: [19] [3660/7269] eta: 0:26:03 lr: 0.000006 loss: 0.5549 (0.5258) time: 0.4169 data: 0.0001 max mem: 52150 +Epoch: [19] [3670/7269] eta: 0:26:00 lr: 0.000006 loss: 0.5115 (0.5258) time: 0.4935 data: 0.0001 max mem: 52150 +Epoch: [19] [3680/7269] eta: 0:25:55 lr: 0.000006 loss: 0.4997 (0.5257) time: 0.4940 data: 0.0001 max mem: 52150 +Epoch: [19] [3690/7269] eta: 0:25:50 lr: 0.000006 loss: 0.5332 (0.5257) time: 0.3723 data: 0.0001 max mem: 52150 +Epoch: [19] [3700/7269] eta: 0:25:45 lr: 0.000006 loss: 0.5471 (0.5257) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [19] [3710/7269] eta: 0:25:40 lr: 0.000006 loss: 0.5215 (0.5257) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [19] [3720/7269] eta: 0:25:39 lr: 0.000006 loss: 0.5470 (0.5258) time: 0.5723 data: 0.0001 max mem: 52150 +Epoch: [19] [3730/7269] eta: 0:25:34 lr: 0.000006 loss: 0.5247 (0.5257) time: 0.5724 data: 0.0001 max mem: 52150 +Epoch: [19] [3740/7269] eta: 0:25:30 lr: 0.000006 loss: 0.5271 (0.5257) time: 0.4176 data: 0.0002 max mem: 52150 +Epoch: [19] [3750/7269] eta: 0:25:26 lr: 0.000006 loss: 0.5560 (0.5258) time: 0.4784 data: 0.0002 max mem: 52150 +Epoch: [19] [3760/7269] eta: 0:25:24 lr: 0.000006 loss: 0.5560 (0.5259) time: 0.5451 data: 0.0001 max mem: 52150 +Epoch: [19] [3770/7269] eta: 0:25:19 lr: 0.000006 loss: 0.5381 (0.5259) time: 0.4843 data: 0.0002 max mem: 52150 +Epoch: [19] [3780/7269] eta: 0:25:14 lr: 0.000006 loss: 0.5773 (0.5260) time: 0.3727 data: 0.0001 max mem: 52150 +Epoch: [19] [3790/7269] eta: 0:25:09 lr: 0.000006 loss: 0.5637 (0.5261) time: 0.3727 data: 0.0002 max mem: 52150 +Epoch: [19] [3800/7269] eta: 0:25:06 lr: 0.000006 loss: 0.5508 (0.5260) time: 0.4670 data: 0.0002 max mem: 52150 +Epoch: [19] [3810/7269] eta: 0:25:02 lr: 0.000006 loss: 0.5036 (0.5260) time: 0.5278 data: 0.0001 max mem: 52150 +Epoch: [19] [3820/7269] eta: 0:24:57 lr: 0.000006 loss: 0.5036 (0.5259) time: 0.4327 data: 0.0001 max mem: 52150 +Epoch: [19] [3830/7269] eta: 0:24:52 lr: 0.000006 loss: 0.5410 (0.5259) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [19] [3840/7269] eta: 0:24:47 lr: 0.000006 loss: 0.5476 (0.5260) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [19] [3850/7269] eta: 0:24:42 lr: 0.000006 loss: 0.5482 (0.5260) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [19] [3860/7269] eta: 0:24:38 lr: 0.000006 loss: 0.5292 (0.5260) time: 0.4171 data: 0.0002 max mem: 52150 +Epoch: [19] [3870/7269] eta: 0:24:33 lr: 0.000006 loss: 0.5321 (0.5260) time: 0.4173 data: 0.0001 max mem: 52150 +Epoch: [19] [3880/7269] eta: 0:24:30 lr: 0.000006 loss: 0.5151 (0.5259) time: 0.4316 data: 0.0001 max mem: 52150 +Epoch: [19] [3890/7269] eta: 0:24:25 lr: 0.000006 loss: 0.4824 (0.5258) time: 0.4317 data: 0.0001 max mem: 52150 +Epoch: [19] [3900/7269] eta: 0:24:20 lr: 0.000006 loss: 0.5033 (0.5258) time: 0.3722 data: 0.0001 max mem: 52150 +Epoch: [19] [3910/7269] eta: 0:24:15 lr: 0.000006 loss: 0.5013 (0.5257) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [19] [3920/7269] eta: 0:24:11 lr: 0.000006 loss: 0.5204 (0.5257) time: 0.4202 data: 0.0001 max mem: 52150 +Epoch: [19] [3930/7269] eta: 0:24:07 lr: 0.000006 loss: 0.5449 (0.5258) time: 0.4664 data: 0.0001 max mem: 52150 +Epoch: [19] [3940/7269] eta: 0:24:03 lr: 0.000006 loss: 0.5445 (0.5258) time: 0.4797 data: 0.0002 max mem: 52150 +Epoch: [19] [3950/7269] eta: 0:23:58 lr: 0.000006 loss: 0.5562 (0.5259) time: 0.4336 data: 0.0002 max mem: 52150 +Epoch: [19] [3960/7269] eta: 0:23:54 lr: 0.000006 loss: 0.5497 (0.5259) time: 0.4201 data: 0.0001 max mem: 52150 +Epoch: [19] [3970/7269] eta: 0:23:49 lr: 0.000006 loss: 0.5093 (0.5259) time: 0.4206 data: 0.0001 max mem: 52150 +Epoch: [19] [3980/7269] eta: 0:23:44 lr: 0.000006 loss: 0.5261 (0.5260) time: 0.3726 data: 0.0001 max mem: 52150 +Epoch: [19] [3990/7269] eta: 0:23:40 lr: 0.000006 loss: 0.5435 (0.5260) time: 0.3724 data: 0.0001 max mem: 52150 +Epoch: [19] [4000/7269] eta: 0:23:35 lr: 0.000006 loss: 0.5587 (0.5261) time: 0.3727 data: 0.0001 max mem: 52150 +Epoch: [19] [4010/7269] eta: 0:23:31 lr: 0.000006 loss: 0.5533 (0.5261) time: 0.4221 data: 0.0001 max mem: 52150 +Epoch: [19] [4020/7269] eta: 0:23:26 lr: 0.000006 loss: 0.5438 (0.5261) time: 0.4221 data: 0.0002 max mem: 52150 +Epoch: [19] [4030/7269] eta: 0:23:22 lr: 0.000006 loss: 0.5192 (0.5260) time: 0.4350 data: 0.0002 max mem: 52150 +Epoch: [19] [4040/7269] eta: 0:23:17 lr: 0.000006 loss: 0.5297 (0.5261) time: 0.4342 data: 0.0001 max mem: 52150 +Epoch: [19] [4050/7269] eta: 0:23:13 lr: 0.000006 loss: 0.5512 (0.5261) time: 0.4307 data: 0.0001 max mem: 52150 +Epoch: [19] [4060/7269] eta: 0:23:09 lr: 0.000006 loss: 0.5237 (0.5261) time: 0.4308 data: 0.0002 max mem: 52150 +Epoch: [19] [4070/7269] eta: 0:23:04 lr: 0.000006 loss: 0.5202 (0.5260) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [19] [4080/7269] eta: 0:23:00 lr: 0.000006 loss: 0.5218 (0.5260) time: 0.4201 data: 0.0002 max mem: 52150 +Epoch: [19] [4090/7269] eta: 0:22:58 lr: 0.000006 loss: 0.5356 (0.5260) time: 0.6022 data: 0.0002 max mem: 52150 +Epoch: [19] [4100/7269] eta: 0:22:54 lr: 0.000006 loss: 0.5309 (0.5259) time: 0.6034 data: 0.0002 max mem: 52150 +Epoch: [19] [4110/7269] eta: 0:22:49 lr: 0.000006 loss: 0.5193 (0.5259) time: 0.4212 data: 0.0002 max mem: 52150 +Epoch: [19] [4120/7269] eta: 0:22:44 lr: 0.000006 loss: 0.5262 (0.5259) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [19] [4130/7269] eta: 0:22:40 lr: 0.000006 loss: 0.5426 (0.5260) time: 0.4324 data: 0.0002 max mem: 52150 +Epoch: [19] [4140/7269] eta: 0:22:35 lr: 0.000006 loss: 0.5336 (0.5259) time: 0.4322 data: 0.0002 max mem: 52150 +Epoch: [19] [4150/7269] eta: 0:22:31 lr: 0.000006 loss: 0.4960 (0.5258) time: 0.4178 data: 0.0002 max mem: 52150 +Epoch: [19] [4160/7269] eta: 0:22:27 lr: 0.000006 loss: 0.5448 (0.5258) time: 0.4177 data: 0.0002 max mem: 52150 +Epoch: [19] [4170/7269] eta: 0:22:22 lr: 0.000006 loss: 0.5557 (0.5259) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [19] [4180/7269] eta: 0:22:18 lr: 0.000006 loss: 0.5371 (0.5258) time: 0.4152 data: 0.0002 max mem: 52150 +Epoch: [19] [4190/7269] eta: 0:22:14 lr: 0.000006 loss: 0.5332 (0.5259) time: 0.5060 data: 0.0002 max mem: 52150 +Epoch: [19] [4200/7269] eta: 0:22:09 lr: 0.000006 loss: 0.5422 (0.5259) time: 0.4624 data: 0.0002 max mem: 52150 +Epoch: [19] [4210/7269] eta: 0:22:05 lr: 0.000006 loss: 0.5310 (0.5259) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [19] [4220/7269] eta: 0:22:00 lr: 0.000006 loss: 0.5338 (0.5259) time: 0.4166 data: 0.0002 max mem: 52150 +Epoch: [19] [4230/7269] eta: 0:21:56 lr: 0.000006 loss: 0.5533 (0.5260) time: 0.4165 data: 0.0002 max mem: 52150 +Epoch: [19] [4240/7269] eta: 0:21:51 lr: 0.000006 loss: 0.5466 (0.5260) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [19] [4250/7269] eta: 0:21:46 lr: 0.000006 loss: 0.5468 (0.5260) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [19] [4260/7269] eta: 0:21:43 lr: 0.000006 loss: 0.5468 (0.5260) time: 0.4673 data: 0.0002 max mem: 52150 +Epoch: [19] [4270/7269] eta: 0:21:38 lr: 0.000006 loss: 0.5088 (0.5260) time: 0.4673 data: 0.0002 max mem: 52150 +Epoch: [19] [4280/7269] eta: 0:21:34 lr: 0.000006 loss: 0.5341 (0.5260) time: 0.4227 data: 0.0002 max mem: 52150 +Epoch: [19] [4290/7269] eta: 0:21:30 lr: 0.000006 loss: 0.4927 (0.5259) time: 0.4710 data: 0.0002 max mem: 52150 +Epoch: [19] [4300/7269] eta: 0:21:25 lr: 0.000006 loss: 0.5006 (0.5259) time: 0.4196 data: 0.0002 max mem: 52150 +Epoch: [19] [4310/7269] eta: 0:21:20 lr: 0.000006 loss: 0.5607 (0.5260) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [19] [4320/7269] eta: 0:21:16 lr: 0.000006 loss: 0.5619 (0.5260) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [19] [4330/7269] eta: 0:21:11 lr: 0.000006 loss: 0.4908 (0.5260) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [19] [4340/7269] eta: 0:21:09 lr: 0.000006 loss: 0.4908 (0.5259) time: 0.5546 data: 0.0002 max mem: 52150 +Epoch: [19] [4350/7269] eta: 0:21:04 lr: 0.000006 loss: 0.5179 (0.5259) time: 0.5547 data: 0.0002 max mem: 52150 +Epoch: [19] [4360/7269] eta: 0:20:59 lr: 0.000006 loss: 0.4920 (0.5257) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [19] [4370/7269] eta: 0:20:54 lr: 0.000006 loss: 0.4920 (0.5258) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [19] [4380/7269] eta: 0:20:50 lr: 0.000006 loss: 0.4911 (0.5257) time: 0.4182 data: 0.0002 max mem: 52150 +Epoch: [19] [4390/7269] eta: 0:20:46 lr: 0.000006 loss: 0.5036 (0.5257) time: 0.4184 data: 0.0002 max mem: 52150 +Epoch: [19] [4400/7269] eta: 0:20:41 lr: 0.000006 loss: 0.5240 (0.5258) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [19] [4410/7269] eta: 0:20:37 lr: 0.000006 loss: 0.5241 (0.5258) time: 0.4344 data: 0.0002 max mem: 52150 +Epoch: [19] [4420/7269] eta: 0:20:32 lr: 0.000006 loss: 0.5241 (0.5258) time: 0.4343 data: 0.0002 max mem: 52150 +Epoch: [19] [4430/7269] eta: 0:20:28 lr: 0.000006 loss: 0.5364 (0.5259) time: 0.4350 data: 0.0002 max mem: 52150 +Epoch: [19] [4440/7269] eta: 0:20:25 lr: 0.000006 loss: 0.5524 (0.5259) time: 0.5405 data: 0.0002 max mem: 52150 +Epoch: [19] [4450/7269] eta: 0:20:20 lr: 0.000006 loss: 0.5356 (0.5259) time: 0.4773 data: 0.0002 max mem: 52150 +Epoch: [19] [4460/7269] eta: 0:20:15 lr: 0.000006 loss: 0.5347 (0.5259) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [19] [4470/7269] eta: 0:20:11 lr: 0.000006 loss: 0.5440 (0.5260) time: 0.4325 data: 0.0002 max mem: 52150 +Epoch: [19] [4480/7269] eta: 0:20:07 lr: 0.000006 loss: 0.5305 (0.5259) time: 0.4793 data: 0.0002 max mem: 52150 +Epoch: [19] [4490/7269] eta: 0:20:03 lr: 0.000006 loss: 0.5252 (0.5260) time: 0.4815 data: 0.0002 max mem: 52150 +Epoch: [19] [4500/7269] eta: 0:19:59 lr: 0.000006 loss: 0.5674 (0.5260) time: 0.4346 data: 0.0002 max mem: 52150 +Epoch: [19] [4510/7269] eta: 0:19:55 lr: 0.000006 loss: 0.5466 (0.5261) time: 0.4235 data: 0.0002 max mem: 52150 +Epoch: [19] [4520/7269] eta: 0:19:50 lr: 0.000006 loss: 0.5531 (0.5261) time: 0.4234 data: 0.0002 max mem: 52150 +Epoch: [19] [4530/7269] eta: 0:19:45 lr: 0.000006 loss: 0.5499 (0.5261) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [19] [4540/7269] eta: 0:19:41 lr: 0.000006 loss: 0.5499 (0.5262) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [19] [4550/7269] eta: 0:19:36 lr: 0.000006 loss: 0.5575 (0.5262) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [19] [4560/7269] eta: 0:19:31 lr: 0.000006 loss: 0.5113 (0.5262) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [19] [4570/7269] eta: 0:19:27 lr: 0.000006 loss: 0.4891 (0.5261) time: 0.4182 data: 0.0002 max mem: 52150 +Epoch: [19] [4580/7269] eta: 0:19:22 lr: 0.000006 loss: 0.5190 (0.5260) time: 0.4187 data: 0.0002 max mem: 52150 +Epoch: [19] [4590/7269] eta: 0:19:18 lr: 0.000006 loss: 0.5053 (0.5260) time: 0.4210 data: 0.0002 max mem: 52150 +Epoch: [19] [4600/7269] eta: 0:19:14 lr: 0.000006 loss: 0.5053 (0.5260) time: 0.4204 data: 0.0002 max mem: 52150 +Epoch: [19] [4610/7269] eta: 0:19:10 lr: 0.000006 loss: 0.5336 (0.5260) time: 0.4315 data: 0.0002 max mem: 52150 +Epoch: [19] [4620/7269] eta: 0:19:05 lr: 0.000006 loss: 0.5284 (0.5260) time: 0.4316 data: 0.0002 max mem: 52150 +Epoch: [19] [4630/7269] eta: 0:19:00 lr: 0.000006 loss: 0.5239 (0.5259) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [19] [4640/7269] eta: 0:18:56 lr: 0.000006 loss: 0.5232 (0.5259) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [19] [4650/7269] eta: 0:18:51 lr: 0.000006 loss: 0.4746 (0.5258) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [19] [4660/7269] eta: 0:18:47 lr: 0.000006 loss: 0.4746 (0.5258) time: 0.4204 data: 0.0002 max mem: 52150 +Epoch: [19] [4670/7269] eta: 0:18:43 lr: 0.000006 loss: 0.5464 (0.5259) time: 0.5294 data: 0.0002 max mem: 52150 +Epoch: [19] [4680/7269] eta: 0:18:40 lr: 0.000006 loss: 0.5312 (0.5259) time: 0.5864 data: 0.0002 max mem: 52150 +Epoch: [19] [4690/7269] eta: 0:18:37 lr: 0.000006 loss: 0.4975 (0.5258) time: 0.6000 data: 0.0002 max mem: 52150 +Epoch: [19] [4700/7269] eta: 0:18:32 lr: 0.000006 loss: 0.5345 (0.5258) time: 0.4946 data: 0.0002 max mem: 52150 +Epoch: [19] [4710/7269] eta: 0:18:28 lr: 0.000006 loss: 0.5100 (0.5258) time: 0.4177 data: 0.0002 max mem: 52150 +Epoch: [19] [4720/7269] eta: 0:18:24 lr: 0.000006 loss: 0.5100 (0.5257) time: 0.4627 data: 0.0002 max mem: 52150 +Epoch: [19] [4730/7269] eta: 0:18:19 lr: 0.000006 loss: 0.5517 (0.5258) time: 0.4166 data: 0.0002 max mem: 52150 +Epoch: [19] [4740/7269] eta: 0:18:15 lr: 0.000006 loss: 0.5363 (0.5258) time: 0.4164 data: 0.0002 max mem: 52150 +Epoch: [19] [4750/7269] eta: 0:18:11 lr: 0.000006 loss: 0.5185 (0.5258) time: 0.4655 data: 0.0002 max mem: 52150 +Epoch: [19] [4760/7269] eta: 0:18:07 lr: 0.000006 loss: 0.5305 (0.5258) time: 0.4845 data: 0.0002 max mem: 52150 +Epoch: [19] [4770/7269] eta: 0:18:03 lr: 0.000006 loss: 0.5268 (0.5258) time: 0.5263 data: 0.0002 max mem: 52150 +Epoch: [19] [4780/7269] eta: 0:17:58 lr: 0.000006 loss: 0.5215 (0.5258) time: 0.4627 data: 0.0002 max mem: 52150 +Epoch: [19] [4790/7269] eta: 0:17:54 lr: 0.000006 loss: 0.5585 (0.5258) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [19] [4800/7269] eta: 0:17:50 lr: 0.000006 loss: 0.5718 (0.5259) time: 0.4334 data: 0.0002 max mem: 52150 +Epoch: [19] [4810/7269] eta: 0:17:45 lr: 0.000006 loss: 0.5141 (0.5258) time: 0.4328 data: 0.0002 max mem: 52150 +Epoch: [19] [4820/7269] eta: 0:17:40 lr: 0.000006 loss: 0.5089 (0.5258) time: 0.3711 data: 0.0002 max mem: 52150 +Epoch: [19] [4830/7269] eta: 0:17:36 lr: 0.000006 loss: 0.5092 (0.5258) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [19] [4840/7269] eta: 0:17:32 lr: 0.000006 loss: 0.5068 (0.5257) time: 0.4326 data: 0.0002 max mem: 52150 +Epoch: [19] [4850/7269] eta: 0:17:27 lr: 0.000006 loss: 0.4830 (0.5257) time: 0.4322 data: 0.0002 max mem: 52150 +Epoch: [19] [4860/7269] eta: 0:17:23 lr: 0.000006 loss: 0.5125 (0.5257) time: 0.4174 data: 0.0002 max mem: 52150 +Epoch: [19] [4870/7269] eta: 0:17:18 lr: 0.000006 loss: 0.5094 (0.5257) time: 0.4172 data: 0.0002 max mem: 52150 +Epoch: [19] [4880/7269] eta: 0:17:14 lr: 0.000006 loss: 0.4854 (0.5256) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [19] [4890/7269] eta: 0:17:09 lr: 0.000006 loss: 0.5164 (0.5256) time: 0.4206 data: 0.0002 max mem: 52150 +Epoch: [19] [4900/7269] eta: 0:17:05 lr: 0.000006 loss: 0.5339 (0.5256) time: 0.4838 data: 0.0002 max mem: 52150 +Epoch: [19] [4910/7269] eta: 0:17:01 lr: 0.000006 loss: 0.5283 (0.5256) time: 0.4348 data: 0.0002 max mem: 52150 +Epoch: [19] [4920/7269] eta: 0:16:56 lr: 0.000006 loss: 0.5283 (0.5256) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [19] [4930/7269] eta: 0:16:52 lr: 0.000006 loss: 0.5497 (0.5256) time: 0.4205 data: 0.0002 max mem: 52150 +Epoch: [19] [4940/7269] eta: 0:16:48 lr: 0.000006 loss: 0.5074 (0.5256) time: 0.4814 data: 0.0002 max mem: 52150 +Epoch: [19] [4950/7269] eta: 0:16:44 lr: 0.000006 loss: 0.5062 (0.5255) time: 0.4933 data: 0.0002 max mem: 52150 +Epoch: [19] [4960/7269] eta: 0:16:39 lr: 0.000006 loss: 0.5029 (0.5255) time: 0.4324 data: 0.0002 max mem: 52150 +Epoch: [19] [4970/7269] eta: 0:16:35 lr: 0.000006 loss: 0.5395 (0.5255) time: 0.4338 data: 0.0002 max mem: 52150 +Epoch: [19] [4980/7269] eta: 0:16:31 lr: 0.000006 loss: 0.5234 (0.5255) time: 0.4336 data: 0.0002 max mem: 52150 +Epoch: [19] [4990/7269] eta: 0:16:27 lr: 0.000006 loss: 0.5243 (0.5255) time: 0.4334 data: 0.0002 max mem: 52150 +Epoch: [19] [5000/7269] eta: 0:16:22 lr: 0.000006 loss: 0.5342 (0.5255) time: 0.4814 data: 0.0002 max mem: 52150 +Epoch: [19] [5010/7269] eta: 0:16:18 lr: 0.000006 loss: 0.5033 (0.5254) time: 0.4655 data: 0.0002 max mem: 52150 +Epoch: [19] [5020/7269] eta: 0:16:14 lr: 0.000006 loss: 0.5080 (0.5254) time: 0.4183 data: 0.0002 max mem: 52150 +Epoch: [19] [5030/7269] eta: 0:16:09 lr: 0.000006 loss: 0.5384 (0.5254) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [19] [5040/7269] eta: 0:16:04 lr: 0.000006 loss: 0.5107 (0.5254) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [19] [5050/7269] eta: 0:16:00 lr: 0.000006 loss: 0.5077 (0.5254) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [19] [5060/7269] eta: 0:15:56 lr: 0.000006 loss: 0.5503 (0.5254) time: 0.4205 data: 0.0002 max mem: 52150 +Epoch: [19] [5070/7269] eta: 0:15:52 lr: 0.000006 loss: 0.5465 (0.5254) time: 0.4809 data: 0.0002 max mem: 52150 +Epoch: [19] [5080/7269] eta: 0:15:47 lr: 0.000006 loss: 0.5187 (0.5254) time: 0.4326 data: 0.0002 max mem: 52150 +Epoch: [19] [5090/7269] eta: 0:15:42 lr: 0.000006 loss: 0.5822 (0.5255) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [19] [5100/7269] eta: 0:15:38 lr: 0.000006 loss: 0.4893 (0.5253) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [19] [5110/7269] eta: 0:15:33 lr: 0.000006 loss: 0.5132 (0.5254) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [19] [5120/7269] eta: 0:15:29 lr: 0.000006 loss: 0.5375 (0.5254) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [19] [5130/7269] eta: 0:15:25 lr: 0.000006 loss: 0.4858 (0.5253) time: 0.4799 data: 0.0002 max mem: 52150 +Epoch: [19] [5140/7269] eta: 0:15:20 lr: 0.000006 loss: 0.4858 (0.5252) time: 0.4803 data: 0.0002 max mem: 52150 +Epoch: [19] [5150/7269] eta: 0:15:16 lr: 0.000006 loss: 0.4863 (0.5251) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [19] [5160/7269] eta: 0:15:12 lr: 0.000006 loss: 0.5441 (0.5252) time: 0.4186 data: 0.0002 max mem: 52150 +Epoch: [19] [5170/7269] eta: 0:15:08 lr: 0.000006 loss: 0.5615 (0.5252) time: 0.4796 data: 0.0002 max mem: 52150 +Epoch: [19] [5180/7269] eta: 0:15:03 lr: 0.000006 loss: 0.5580 (0.5252) time: 0.4326 data: 0.0002 max mem: 52150 +Epoch: [19] [5190/7269] eta: 0:14:59 lr: 0.000006 loss: 0.4964 (0.5251) time: 0.4687 data: 0.0002 max mem: 52150 +Epoch: [19] [5200/7269] eta: 0:14:55 lr: 0.000006 loss: 0.5242 (0.5252) time: 0.4688 data: 0.0002 max mem: 52150 +Epoch: [19] [5210/7269] eta: 0:14:50 lr: 0.000006 loss: 0.5485 (0.5252) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [19] [5220/7269] eta: 0:14:46 lr: 0.000006 loss: 0.5132 (0.5252) time: 0.4339 data: 0.0002 max mem: 52150 +Epoch: [19] [5230/7269] eta: 0:14:41 lr: 0.000006 loss: 0.5397 (0.5252) time: 0.4347 data: 0.0002 max mem: 52150 +Epoch: [19] [5240/7269] eta: 0:14:37 lr: 0.000006 loss: 0.5397 (0.5252) time: 0.4170 data: 0.0002 max mem: 52150 +Epoch: [19] [5250/7269] eta: 0:14:33 lr: 0.000006 loss: 0.5226 (0.5253) time: 0.4165 data: 0.0002 max mem: 52150 +Epoch: [19] [5260/7269] eta: 0:14:29 lr: 0.000006 loss: 0.5389 (0.5253) time: 0.4939 data: 0.0002 max mem: 52150 +Epoch: [19] [5270/7269] eta: 0:14:24 lr: 0.000006 loss: 0.4791 (0.5251) time: 0.4935 data: 0.0002 max mem: 52150 +Epoch: [19] [5280/7269] eta: 0:14:20 lr: 0.000006 loss: 0.4935 (0.5252) time: 0.4191 data: 0.0002 max mem: 52150 +Epoch: [19] [5290/7269] eta: 0:14:16 lr: 0.000006 loss: 0.5490 (0.5253) time: 0.4652 data: 0.0002 max mem: 52150 +Epoch: [19] [5300/7269] eta: 0:14:11 lr: 0.000006 loss: 0.5630 (0.5253) time: 0.4179 data: 0.0002 max mem: 52150 +Epoch: [19] [5310/7269] eta: 0:14:07 lr: 0.000006 loss: 0.5375 (0.5253) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [19] [5320/7269] eta: 0:14:02 lr: 0.000006 loss: 0.5233 (0.5253) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [19] [5330/7269] eta: 0:13:58 lr: 0.000006 loss: 0.5420 (0.5254) time: 0.4315 data: 0.0002 max mem: 52150 +Epoch: [19] [5340/7269] eta: 0:13:54 lr: 0.000006 loss: 0.5481 (0.5254) time: 0.4318 data: 0.0002 max mem: 52150 +Epoch: [19] [5350/7269] eta: 0:13:50 lr: 0.000006 loss: 0.4941 (0.5253) time: 0.4222 data: 0.0002 max mem: 52150 +Epoch: [19] [5360/7269] eta: 0:13:45 lr: 0.000006 loss: 0.4827 (0.5253) time: 0.4704 data: 0.0002 max mem: 52150 +Epoch: [19] [5370/7269] eta: 0:13:41 lr: 0.000006 loss: 0.5219 (0.5253) time: 0.4204 data: 0.0002 max mem: 52150 +Epoch: [19] [5380/7269] eta: 0:13:36 lr: 0.000006 loss: 0.5225 (0.5253) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [19] [5390/7269] eta: 0:13:32 lr: 0.000006 loss: 0.5225 (0.5253) time: 0.4201 data: 0.0002 max mem: 52150 +Epoch: [19] [5400/7269] eta: 0:13:28 lr: 0.000006 loss: 0.5199 (0.5253) time: 0.4202 data: 0.0002 max mem: 52150 +Epoch: [19] [5410/7269] eta: 0:13:23 lr: 0.000006 loss: 0.5468 (0.5253) time: 0.4327 data: 0.0002 max mem: 52150 +Epoch: [19] [5420/7269] eta: 0:13:19 lr: 0.000006 loss: 0.5666 (0.5253) time: 0.4962 data: 0.0002 max mem: 52150 +Epoch: [19] [5430/7269] eta: 0:13:15 lr: 0.000006 loss: 0.5588 (0.5253) time: 0.4984 data: 0.0002 max mem: 52150 +Epoch: [19] [5440/7269] eta: 0:13:11 lr: 0.000006 loss: 0.5226 (0.5253) time: 0.4355 data: 0.0002 max mem: 52150 +Epoch: [19] [5450/7269] eta: 0:13:06 lr: 0.000006 loss: 0.5107 (0.5252) time: 0.3730 data: 0.0002 max mem: 52150 +Epoch: [19] [5460/7269] eta: 0:13:02 lr: 0.000006 loss: 0.5221 (0.5252) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [19] [5470/7269] eta: 0:12:58 lr: 0.000006 loss: 0.5024 (0.5251) time: 0.5363 data: 0.0002 max mem: 52150 +Epoch: [19] [5480/7269] eta: 0:12:54 lr: 0.000006 loss: 0.5187 (0.5252) time: 0.5980 data: 0.0002 max mem: 52150 +Epoch: [19] [5490/7269] eta: 0:12:50 lr: 0.000006 loss: 0.5383 (0.5251) time: 0.4342 data: 0.0002 max mem: 52150 +Epoch: [19] [5500/7269] eta: 0:12:45 lr: 0.000006 loss: 0.5383 (0.5252) time: 0.4169 data: 0.0002 max mem: 52150 +Epoch: [19] [5510/7269] eta: 0:12:41 lr: 0.000006 loss: 0.5444 (0.5252) time: 0.4167 data: 0.0002 max mem: 52150 +Epoch: [19] [5520/7269] eta: 0:12:36 lr: 0.000006 loss: 0.5214 (0.5251) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [19] [5530/7269] eta: 0:12:32 lr: 0.000006 loss: 0.5066 (0.5252) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [19] [5540/7269] eta: 0:12:27 lr: 0.000006 loss: 0.5066 (0.5251) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [19] [5550/7269] eta: 0:12:23 lr: 0.000006 loss: 0.4585 (0.5251) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [19] [5560/7269] eta: 0:12:18 lr: 0.000006 loss: 0.5356 (0.5251) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [19] [5570/7269] eta: 0:12:14 lr: 0.000006 loss: 0.4933 (0.5250) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [19] [5580/7269] eta: 0:12:09 lr: 0.000006 loss: 0.5088 (0.5250) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [19] [5590/7269] eta: 0:12:05 lr: 0.000006 loss: 0.5452 (0.5250) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [19] [5600/7269] eta: 0:12:00 lr: 0.000006 loss: 0.5373 (0.5250) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [19] [5610/7269] eta: 0:11:56 lr: 0.000006 loss: 0.5453 (0.5251) time: 0.4341 data: 0.0002 max mem: 52150 +Epoch: [19] [5620/7269] eta: 0:11:52 lr: 0.000006 loss: 0.5620 (0.5252) time: 0.4817 data: 0.0002 max mem: 52150 +Epoch: [19] [5630/7269] eta: 0:11:48 lr: 0.000006 loss: 0.5620 (0.5252) time: 0.5130 data: 0.0002 max mem: 52150 +Epoch: [19] [5640/7269] eta: 0:11:43 lr: 0.000006 loss: 0.5638 (0.5253) time: 0.4657 data: 0.0002 max mem: 52150 +Epoch: [19] [5650/7269] eta: 0:11:39 lr: 0.000006 loss: 0.5737 (0.5253) time: 0.4177 data: 0.0002 max mem: 52150 +Epoch: [19] [5660/7269] eta: 0:11:35 lr: 0.000006 loss: 0.5361 (0.5253) time: 0.4176 data: 0.0002 max mem: 52150 +Epoch: [19] [5670/7269] eta: 0:11:30 lr: 0.000006 loss: 0.5345 (0.5253) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [19] [5680/7269] eta: 0:11:26 lr: 0.000006 loss: 0.5345 (0.5253) time: 0.4346 data: 0.0002 max mem: 52150 +Epoch: [19] [5690/7269] eta: 0:11:22 lr: 0.000006 loss: 0.5249 (0.5253) time: 0.4959 data: 0.0002 max mem: 52150 +Epoch: [19] [5700/7269] eta: 0:11:18 lr: 0.000006 loss: 0.5222 (0.5253) time: 0.4789 data: 0.0002 max mem: 52150 +Epoch: [19] [5710/7269] eta: 0:11:14 lr: 0.000006 loss: 0.5176 (0.5253) time: 0.4792 data: 0.0002 max mem: 52150 +Epoch: [19] [5720/7269] eta: 0:11:09 lr: 0.000006 loss: 0.5176 (0.5253) time: 0.4824 data: 0.0002 max mem: 52150 +Epoch: [19] [5730/7269] eta: 0:11:05 lr: 0.000006 loss: 0.5088 (0.5252) time: 0.4209 data: 0.0002 max mem: 52150 +Epoch: [19] [5740/7269] eta: 0:11:00 lr: 0.000006 loss: 0.5252 (0.5252) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [19] [5750/7269] eta: 0:10:56 lr: 0.000006 loss: 0.5413 (0.5252) time: 0.4353 data: 0.0002 max mem: 52150 +Epoch: [19] [5760/7269] eta: 0:10:52 lr: 0.000006 loss: 0.5411 (0.5252) time: 0.4350 data: 0.0002 max mem: 52150 +Epoch: [19] [5770/7269] eta: 0:10:47 lr: 0.000006 loss: 0.5305 (0.5253) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [19] [5780/7269] eta: 0:10:43 lr: 0.000006 loss: 0.5362 (0.5253) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [19] [5790/7269] eta: 0:10:38 lr: 0.000006 loss: 0.5362 (0.5253) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [19] [5800/7269] eta: 0:10:34 lr: 0.000006 loss: 0.5520 (0.5254) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [19] [5810/7269] eta: 0:10:30 lr: 0.000006 loss: 0.5418 (0.5253) time: 0.4334 data: 0.0002 max mem: 52150 +Epoch: [19] [5820/7269] eta: 0:10:25 lr: 0.000006 loss: 0.4849 (0.5253) time: 0.4972 data: 0.0002 max mem: 52150 +Epoch: [19] [5830/7269] eta: 0:10:21 lr: 0.000006 loss: 0.5228 (0.5253) time: 0.4977 data: 0.0002 max mem: 52150 +Epoch: [19] [5840/7269] eta: 0:10:17 lr: 0.000006 loss: 0.5163 (0.5253) time: 0.4342 data: 0.0002 max mem: 52150 +Epoch: [19] [5850/7269] eta: 0:10:13 lr: 0.000006 loss: 0.5172 (0.5253) time: 0.4206 data: 0.0002 max mem: 52150 +Epoch: [19] [5860/7269] eta: 0:10:08 lr: 0.000006 loss: 0.5376 (0.5253) time: 0.4201 data: 0.0002 max mem: 52150 +Epoch: [19] [5870/7269] eta: 0:10:04 lr: 0.000006 loss: 0.5368 (0.5254) time: 0.4782 data: 0.0002 max mem: 52150 +Epoch: [19] [5880/7269] eta: 0:10:00 lr: 0.000006 loss: 0.5231 (0.5254) time: 0.4787 data: 0.0002 max mem: 52150 +Epoch: [19] [5890/7269] eta: 0:09:55 lr: 0.000006 loss: 0.5317 (0.5254) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [19] [5900/7269] eta: 0:09:51 lr: 0.000006 loss: 0.5381 (0.5254) time: 0.4935 data: 0.0002 max mem: 52150 +Epoch: [19] [5910/7269] eta: 0:09:47 lr: 0.000006 loss: 0.5202 (0.5254) time: 0.5411 data: 0.0002 max mem: 52150 +Epoch: [19] [5920/7269] eta: 0:09:43 lr: 0.000006 loss: 0.4755 (0.5253) time: 0.4196 data: 0.0002 max mem: 52150 +Epoch: [19] [5930/7269] eta: 0:09:38 lr: 0.000006 loss: 0.5361 (0.5254) time: 0.3727 data: 0.0002 max mem: 52150 +Epoch: [19] [5940/7269] eta: 0:09:34 lr: 0.000006 loss: 0.5072 (0.5253) time: 0.4332 data: 0.0002 max mem: 52150 +Epoch: [19] [5950/7269] eta: 0:09:30 lr: 0.000006 loss: 0.4939 (0.5253) time: 0.4826 data: 0.0002 max mem: 52150 +Epoch: [19] [5960/7269] eta: 0:09:26 lr: 0.000006 loss: 0.5180 (0.5252) time: 0.4661 data: 0.0002 max mem: 52150 +Epoch: [19] [5970/7269] eta: 0:09:21 lr: 0.000006 loss: 0.5188 (0.5252) time: 0.4642 data: 0.0002 max mem: 52150 +Epoch: [19] [5980/7269] eta: 0:09:17 lr: 0.000006 loss: 0.5206 (0.5252) time: 0.5429 data: 0.0002 max mem: 52150 +Epoch: [19] [5990/7269] eta: 0:09:13 lr: 0.000006 loss: 0.4942 (0.5252) time: 0.4952 data: 0.0002 max mem: 52150 +Epoch: [19] [6000/7269] eta: 0:09:08 lr: 0.000006 loss: 0.5248 (0.5252) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [19] [6010/7269] eta: 0:09:04 lr: 0.000006 loss: 0.5420 (0.5252) time: 0.4333 data: 0.0002 max mem: 52150 +Epoch: [19] [6020/7269] eta: 0:09:00 lr: 0.000006 loss: 0.5420 (0.5252) time: 0.4325 data: 0.0002 max mem: 52150 +Epoch: [19] [6030/7269] eta: 0:08:56 lr: 0.000006 loss: 0.5293 (0.5252) time: 0.4168 data: 0.0002 max mem: 52150 +Epoch: [19] [6040/7269] eta: 0:08:51 lr: 0.000006 loss: 0.5506 (0.5252) time: 0.4184 data: 0.0002 max mem: 52150 +Epoch: [19] [6050/7269] eta: 0:08:47 lr: 0.000006 loss: 0.5536 (0.5252) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [19] [6060/7269] eta: 0:08:42 lr: 0.000006 loss: 0.5456 (0.5253) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [19] [6070/7269] eta: 0:08:38 lr: 0.000006 loss: 0.5456 (0.5253) time: 0.4195 data: 0.0002 max mem: 52150 +Epoch: [19] [6080/7269] eta: 0:08:33 lr: 0.000006 loss: 0.5405 (0.5253) time: 0.4191 data: 0.0002 max mem: 52150 +Epoch: [19] [6090/7269] eta: 0:08:29 lr: 0.000006 loss: 0.5280 (0.5253) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [19] [6100/7269] eta: 0:08:25 lr: 0.000006 loss: 0.5226 (0.5253) time: 0.4797 data: 0.0002 max mem: 52150 +Epoch: [19] [6110/7269] eta: 0:08:21 lr: 0.000006 loss: 0.5226 (0.5253) time: 0.4796 data: 0.0002 max mem: 52150 +Epoch: [19] [6120/7269] eta: 0:08:16 lr: 0.000006 loss: 0.5089 (0.5253) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [19] [6130/7269] eta: 0:08:12 lr: 0.000006 loss: 0.5322 (0.5253) time: 0.4330 data: 0.0002 max mem: 52150 +Epoch: [19] [6140/7269] eta: 0:08:08 lr: 0.000006 loss: 0.5532 (0.5253) time: 0.4327 data: 0.0002 max mem: 52150 +Epoch: [19] [6150/7269] eta: 0:08:03 lr: 0.000006 loss: 0.5297 (0.5253) time: 0.4201 data: 0.0002 max mem: 52150 +Epoch: [19] [6160/7269] eta: 0:07:59 lr: 0.000006 loss: 0.5293 (0.5254) time: 0.4200 data: 0.0002 max mem: 52150 +Epoch: [19] [6170/7269] eta: 0:07:54 lr: 0.000006 loss: 0.5371 (0.5254) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [19] [6180/7269] eta: 0:07:50 lr: 0.000006 loss: 0.5652 (0.5254) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [19] [6190/7269] eta: 0:07:46 lr: 0.000006 loss: 0.5396 (0.5254) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [19] [6200/7269] eta: 0:07:41 lr: 0.000006 loss: 0.5099 (0.5254) time: 0.4195 data: 0.0002 max mem: 52150 +Epoch: [19] [6210/7269] eta: 0:07:37 lr: 0.000006 loss: 0.4819 (0.5254) time: 0.4800 data: 0.0002 max mem: 52150 +Epoch: [19] [6220/7269] eta: 0:07:33 lr: 0.000006 loss: 0.4819 (0.5253) time: 0.4804 data: 0.0002 max mem: 52150 +Epoch: [19] [6230/7269] eta: 0:07:29 lr: 0.000006 loss: 0.4827 (0.5253) time: 0.4827 data: 0.0002 max mem: 52150 +Epoch: [19] [6240/7269] eta: 0:07:24 lr: 0.000006 loss: 0.5210 (0.5254) time: 0.4345 data: 0.0002 max mem: 52150 +Epoch: [19] [6250/7269] eta: 0:07:20 lr: 0.000006 loss: 0.5178 (0.5253) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [19] [6260/7269] eta: 0:07:16 lr: 0.000006 loss: 0.4964 (0.5253) time: 0.4853 data: 0.0002 max mem: 52150 +Epoch: [19] [6270/7269] eta: 0:07:11 lr: 0.000006 loss: 0.5045 (0.5253) time: 0.4854 data: 0.0002 max mem: 52150 +Epoch: [19] [6280/7269] eta: 0:07:07 lr: 0.000006 loss: 0.5128 (0.5253) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [19] [6290/7269] eta: 0:07:02 lr: 0.000006 loss: 0.5054 (0.5253) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [19] [6300/7269] eta: 0:06:58 lr: 0.000006 loss: 0.5037 (0.5252) time: 0.4177 data: 0.0002 max mem: 52150 +Epoch: [19] [6310/7269] eta: 0:06:54 lr: 0.000006 loss: 0.5092 (0.5253) time: 0.4645 data: 0.0002 max mem: 52150 +Epoch: [19] [6320/7269] eta: 0:06:49 lr: 0.000006 loss: 0.4870 (0.5252) time: 0.4187 data: 0.0002 max mem: 52150 +Epoch: [19] [6330/7269] eta: 0:06:45 lr: 0.000006 loss: 0.4870 (0.5251) time: 0.4351 data: 0.0002 max mem: 52150 +Epoch: [19] [6340/7269] eta: 0:06:41 lr: 0.000006 loss: 0.5334 (0.5252) time: 0.4799 data: 0.0002 max mem: 52150 +Epoch: [19] [6350/7269] eta: 0:06:37 lr: 0.000006 loss: 0.5374 (0.5252) time: 0.4761 data: 0.0002 max mem: 52150 +Epoch: [19] [6360/7269] eta: 0:06:32 lr: 0.000006 loss: 0.5479 (0.5252) time: 0.4777 data: 0.0002 max mem: 52150 +Epoch: [19] [6370/7269] eta: 0:06:28 lr: 0.000006 loss: 0.5487 (0.5252) time: 0.4772 data: 0.0002 max mem: 52150 +Epoch: [19] [6380/7269] eta: 0:06:24 lr: 0.000006 loss: 0.5261 (0.5252) time: 0.4306 data: 0.0002 max mem: 52150 +Epoch: [19] [6390/7269] eta: 0:06:20 lr: 0.000006 loss: 0.5572 (0.5252) time: 0.4375 data: 0.0002 max mem: 52150 +Epoch: [19] [6400/7269] eta: 0:06:15 lr: 0.000006 loss: 0.5545 (0.5252) time: 0.4983 data: 0.0002 max mem: 52150 +Epoch: [19] [6410/7269] eta: 0:06:11 lr: 0.000006 loss: 0.5232 (0.5253) time: 0.4919 data: 0.0002 max mem: 52150 +Epoch: [19] [6420/7269] eta: 0:06:07 lr: 0.000006 loss: 0.5441 (0.5253) time: 0.4314 data: 0.0002 max mem: 52150 +Epoch: [19] [6430/7269] eta: 0:06:02 lr: 0.000006 loss: 0.5357 (0.5253) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [19] [6440/7269] eta: 0:05:58 lr: 0.000006 loss: 0.5542 (0.5253) time: 0.4792 data: 0.0002 max mem: 52150 +Epoch: [19] [6450/7269] eta: 0:05:54 lr: 0.000006 loss: 0.5550 (0.5253) time: 0.5253 data: 0.0002 max mem: 52150 +Epoch: [19] [6460/7269] eta: 0:05:50 lr: 0.000006 loss: 0.4806 (0.5252) time: 0.4804 data: 0.0002 max mem: 52150 +Epoch: [19] [6470/7269] eta: 0:05:45 lr: 0.000006 loss: 0.5313 (0.5252) time: 0.4345 data: 0.0002 max mem: 52150 +Epoch: [19] [6480/7269] eta: 0:05:41 lr: 0.000006 loss: 0.5437 (0.5252) time: 0.4327 data: 0.0002 max mem: 52150 +Epoch: [19] [6490/7269] eta: 0:05:37 lr: 0.000006 loss: 0.5220 (0.5252) time: 0.5251 data: 0.0002 max mem: 52150 +Epoch: [19] [6500/7269] eta: 0:05:32 lr: 0.000006 loss: 0.5169 (0.5252) time: 0.4645 data: 0.0002 max mem: 52150 +Epoch: [19] [6510/7269] eta: 0:05:28 lr: 0.000006 loss: 0.5302 (0.5252) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [19] [6520/7269] eta: 0:05:24 lr: 0.000006 loss: 0.5343 (0.5253) time: 0.4221 data: 0.0002 max mem: 52150 +Epoch: [19] [6530/7269] eta: 0:05:19 lr: 0.000006 loss: 0.5186 (0.5253) time: 0.4220 data: 0.0002 max mem: 52150 +Epoch: [19] [6540/7269] eta: 0:05:15 lr: 0.000006 loss: 0.5337 (0.5253) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [19] [6550/7269] eta: 0:05:11 lr: 0.000006 loss: 0.5372 (0.5253) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [19] [6560/7269] eta: 0:05:06 lr: 0.000006 loss: 0.5354 (0.5253) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [19] [6570/7269] eta: 0:05:02 lr: 0.000006 loss: 0.5026 (0.5253) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [19] [6580/7269] eta: 0:04:57 lr: 0.000006 loss: 0.4862 (0.5252) time: 0.4321 data: 0.0002 max mem: 52150 +Epoch: [19] [6590/7269] eta: 0:04:53 lr: 0.000006 loss: 0.4909 (0.5252) time: 0.5418 data: 0.0002 max mem: 52150 +Epoch: [19] [6600/7269] eta: 0:04:49 lr: 0.000006 loss: 0.4909 (0.5252) time: 0.4829 data: 0.0002 max mem: 52150 +Epoch: [19] [6610/7269] eta: 0:04:45 lr: 0.000006 loss: 0.5095 (0.5252) time: 0.3731 data: 0.0002 max mem: 52150 +Epoch: [19] [6620/7269] eta: 0:04:40 lr: 0.000006 loss: 0.5042 (0.5251) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [19] [6630/7269] eta: 0:04:36 lr: 0.000006 loss: 0.5437 (0.5252) time: 0.4969 data: 0.0002 max mem: 52150 +Epoch: [19] [6640/7269] eta: 0:04:32 lr: 0.000006 loss: 0.5175 (0.5251) time: 0.4967 data: 0.0002 max mem: 52150 +Epoch: [19] [6650/7269] eta: 0:04:27 lr: 0.000006 loss: 0.4774 (0.5250) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [19] [6660/7269] eta: 0:04:23 lr: 0.000006 loss: 0.5198 (0.5250) time: 0.3727 data: 0.0002 max mem: 52150 +Epoch: [19] [6670/7269] eta: 0:04:19 lr: 0.000006 loss: 0.5358 (0.5251) time: 0.4181 data: 0.0002 max mem: 52150 +Epoch: [19] [6680/7269] eta: 0:04:14 lr: 0.000006 loss: 0.5416 (0.5251) time: 0.4171 data: 0.0002 max mem: 52150 +Epoch: [19] [6690/7269] eta: 0:04:10 lr: 0.000006 loss: 0.4979 (0.5250) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [19] [6700/7269] eta: 0:04:06 lr: 0.000006 loss: 0.4838 (0.5250) time: 0.4175 data: 0.0002 max mem: 52150 +Epoch: [19] [6710/7269] eta: 0:04:01 lr: 0.000006 loss: 0.5291 (0.5250) time: 0.4800 data: 0.0002 max mem: 52150 +Epoch: [19] [6720/7269] eta: 0:03:57 lr: 0.000006 loss: 0.5251 (0.5250) time: 0.4836 data: 0.0002 max mem: 52150 +Epoch: [19] [6730/7269] eta: 0:03:53 lr: 0.000006 loss: 0.5543 (0.5251) time: 0.4682 data: 0.0002 max mem: 52150 +Epoch: [19] [6740/7269] eta: 0:03:48 lr: 0.000006 loss: 0.5520 (0.5251) time: 0.4191 data: 0.0002 max mem: 52150 +Epoch: [19] [6750/7269] eta: 0:03:44 lr: 0.000006 loss: 0.5074 (0.5250) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [19] [6760/7269] eta: 0:03:40 lr: 0.000006 loss: 0.5091 (0.5250) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [19] [6770/7269] eta: 0:03:35 lr: 0.000006 loss: 0.5404 (0.5251) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [19] [6780/7269] eta: 0:03:31 lr: 0.000006 loss: 0.5347 (0.5250) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [19] [6790/7269] eta: 0:03:26 lr: 0.000006 loss: 0.4941 (0.5250) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [19] [6800/7269] eta: 0:03:22 lr: 0.000006 loss: 0.5058 (0.5250) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [19] [6810/7269] eta: 0:03:18 lr: 0.000006 loss: 0.4991 (0.5249) time: 0.4823 data: 0.0002 max mem: 52150 +Epoch: [19] [6820/7269] eta: 0:03:14 lr: 0.000006 loss: 0.4918 (0.5249) time: 0.4821 data: 0.0002 max mem: 52150 +Epoch: [19] [6830/7269] eta: 0:03:09 lr: 0.000006 loss: 0.5318 (0.5249) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [19] [6840/7269] eta: 0:03:05 lr: 0.000006 loss: 0.5157 (0.5248) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [19] [6850/7269] eta: 0:03:00 lr: 0.000006 loss: 0.5135 (0.5249) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [19] [6860/7269] eta: 0:02:56 lr: 0.000006 loss: 0.5279 (0.5248) time: 0.4329 data: 0.0002 max mem: 52150 +Epoch: [19] [6870/7269] eta: 0:02:52 lr: 0.000006 loss: 0.5300 (0.5249) time: 0.4327 data: 0.0002 max mem: 52150 +Epoch: [19] [6880/7269] eta: 0:02:47 lr: 0.000006 loss: 0.5436 (0.5249) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [19] [6890/7269] eta: 0:02:43 lr: 0.000006 loss: 0.5475 (0.5249) time: 0.4318 data: 0.0002 max mem: 52150 +Epoch: [19] [6900/7269] eta: 0:02:39 lr: 0.000006 loss: 0.5060 (0.5248) time: 0.4314 data: 0.0002 max mem: 52150 +Epoch: [19] [6910/7269] eta: 0:02:35 lr: 0.000006 loss: 0.4547 (0.5248) time: 0.4311 data: 0.0002 max mem: 52150 +Epoch: [19] [6920/7269] eta: 0:02:30 lr: 0.000006 loss: 0.4972 (0.5247) time: 0.5879 data: 0.0002 max mem: 52150 +Epoch: [19] [6930/7269] eta: 0:02:26 lr: 0.000006 loss: 0.5079 (0.5247) time: 0.5283 data: 0.0002 max mem: 52150 +Epoch: [19] [6940/7269] eta: 0:02:22 lr: 0.000006 loss: 0.5278 (0.5247) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [19] [6950/7269] eta: 0:02:17 lr: 0.000006 loss: 0.5106 (0.5247) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [19] [6960/7269] eta: 0:02:13 lr: 0.000006 loss: 0.5299 (0.5247) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [19] [6970/7269] eta: 0:02:09 lr: 0.000006 loss: 0.5437 (0.5247) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [19] [6980/7269] eta: 0:02:04 lr: 0.000006 loss: 0.5267 (0.5247) time: 0.4363 data: 0.0002 max mem: 52150 +Epoch: [19] [6990/7269] eta: 0:02:00 lr: 0.000006 loss: 0.5208 (0.5247) time: 0.5440 data: 0.0002 max mem: 52150 +Epoch: [19] [7000/7269] eta: 0:01:56 lr: 0.000006 loss: 0.5208 (0.5247) time: 0.4796 data: 0.0002 max mem: 52150 +Epoch: [19] [7010/7269] eta: 0:01:51 lr: 0.000006 loss: 0.4791 (0.5246) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [19] [7020/7269] eta: 0:01:47 lr: 0.000006 loss: 0.5364 (0.5246) time: 0.4363 data: 0.0002 max mem: 52150 +Epoch: [19] [7030/7269] eta: 0:01:43 lr: 0.000006 loss: 0.5317 (0.5246) time: 0.4843 data: 0.0002 max mem: 52150 +Epoch: [19] [7040/7269] eta: 0:01:38 lr: 0.000006 loss: 0.4967 (0.5245) time: 0.4198 data: 0.0002 max mem: 52150 +Epoch: [19] [7050/7269] eta: 0:01:34 lr: 0.000006 loss: 0.5517 (0.5245) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [19] [7060/7269] eta: 0:01:30 lr: 0.000006 loss: 0.5006 (0.5245) time: 0.4646 data: 0.0002 max mem: 52150 +Epoch: [19] [7070/7269] eta: 0:01:26 lr: 0.000006 loss: 0.5006 (0.5245) time: 0.5902 data: 0.0002 max mem: 52150 +Epoch: [19] [7080/7269] eta: 0:01:21 lr: 0.000006 loss: 0.5315 (0.5245) time: 0.4976 data: 0.0002 max mem: 52150 +Epoch: [19] [7090/7269] eta: 0:01:17 lr: 0.000006 loss: 0.5470 (0.5245) time: 0.4803 data: 0.0002 max mem: 52150 +Epoch: [19] [7100/7269] eta: 0:01:13 lr: 0.000006 loss: 0.5142 (0.5245) time: 0.5284 data: 0.0002 max mem: 52150 +Epoch: [19] [7110/7269] eta: 0:01:08 lr: 0.000006 loss: 0.5026 (0.5245) time: 0.4211 data: 0.0002 max mem: 52150 +Epoch: [19] [7120/7269] eta: 0:01:04 lr: 0.000006 loss: 0.5198 (0.5245) time: 0.3728 data: 0.0002 max mem: 52150 +Epoch: [19] [7130/7269] eta: 0:01:00 lr: 0.000006 loss: 0.5218 (0.5245) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [19] [7140/7269] eta: 0:00:55 lr: 0.000006 loss: 0.5311 (0.5245) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [19] [7150/7269] eta: 0:00:51 lr: 0.000006 loss: 0.5311 (0.5245) time: 0.4164 data: 0.0002 max mem: 52150 +Epoch: [19] [7160/7269] eta: 0:00:47 lr: 0.000006 loss: 0.5241 (0.5245) time: 0.4168 data: 0.0002 max mem: 52150 +Epoch: [19] [7170/7269] eta: 0:00:42 lr: 0.000006 loss: 0.5114 (0.5246) time: 0.5287 data: 0.0002 max mem: 52150 +Epoch: [19] [7180/7269] eta: 0:00:38 lr: 0.000006 loss: 0.5185 (0.5245) time: 0.5286 data: 0.0002 max mem: 52150 +Epoch: [19] [7190/7269] eta: 0:00:34 lr: 0.000006 loss: 0.5246 (0.5246) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [19] [7200/7269] eta: 0:00:29 lr: 0.000006 loss: 0.5603 (0.5246) time: 0.4196 data: 0.0002 max mem: 52150 +Epoch: [19] [7210/7269] eta: 0:00:25 lr: 0.000006 loss: 0.5534 (0.5247) time: 0.4190 data: 0.0002 max mem: 52150 +Epoch: [19] [7220/7269] eta: 0:00:21 lr: 0.000006 loss: 0.5142 (0.5247) time: 0.4314 data: 0.0002 max mem: 52150 +Epoch: [19] [7230/7269] eta: 0:00:16 lr: 0.000006 loss: 0.5079 (0.5246) time: 0.4317 data: 0.0002 max mem: 52150 +Epoch: [19] [7240/7269] eta: 0:00:12 lr: 0.000006 loss: 0.4954 (0.5246) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [19] [7250/7269] eta: 0:00:08 lr: 0.000006 loss: 0.5026 (0.5246) time: 0.4232 data: 0.0032 max mem: 52150 +Epoch: [19] [7260/7269] eta: 0:00:03 lr: 0.000006 loss: 0.5035 (0.5245) time: 0.4820 data: 0.0032 max mem: 52150 +Epoch: [19] [7268/7269] eta: 0:00:00 lr: 0.000006 loss: 0.5035 (0.5245) time: 0.4805 data: 0.0032 max mem: 52150 +Epoch: [19] Total time: 0:52:26 (0.4329 s / it) +Averaged stats: lr: 0.000006 loss: 0.5035 (0.5235) +Test: [ 0/1403] eta: 1:17:50 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0884 (0.0884) acc1: 100.0000 (100.0000) acc5: 100.0000 (100.0000) time: 3.3286 data: 3.2314 max mem: 52150 +Test: [ 10/1403] eta: 0:09:03 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1336 (0.1616) acc1: 98.4375 (96.1648) acc5: 100.0000 (100.0000) time: 0.3903 data: 0.2940 max mem: 52150 +Test: [ 20/1403] eta: 0:05:46 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1272 (0.1512) acc1: 98.4375 (96.2798) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 30/1403] eta: 0:04:35 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0842 (0.1467) acc1: 100.0000 (96.2198) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 40/1403] eta: 0:03:59 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0980 (0.1474) acc1: 100.0000 (96.4558) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 50/1403] eta: 0:03:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1110 (0.1663) acc1: 100.0000 (95.3738) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 60/1403] eta: 0:03:20 acc1_pos: 89.062 acc5_pos: 100.000 loss: 0.1242 (0.1630) acc1: 100.0000 (95.7992) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 70/1403] eta: 0:03:09 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.1242 (0.1634) acc1: 98.4375 (95.6646) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 80/1403] eta: 0:03:00 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1304 (0.1707) acc1: 96.8750 (95.2739) acc5: 100.0000 (100.0000) time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 90/1403] eta: 0:02:53 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1738 (0.1880) acc1: 95.3125 (94.4540) acc5: 100.0000 (100.0000) time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 100/1403] eta: 0:02:47 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1493 (0.1877) acc1: 95.3125 (94.5390) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 110/1403] eta: 0:02:42 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1429 (0.1855) acc1: 98.4375 (94.6791) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 120/1403] eta: 0:02:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0780 (0.1766) acc1: 100.0000 (95.0930) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 130/1403] eta: 0:02:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0761 (0.1706) acc1: 100.0000 (95.3602) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 140/1403] eta: 0:02:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0776 (0.1638) acc1: 100.0000 (95.6893) acc5: 100.0000 (100.0000) time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 150/1403] eta: 0:02:27 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0727 (0.1582) acc1: 100.0000 (95.9644) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 160/1403] eta: 0:02:24 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0749 (0.1536) acc1: 100.0000 (96.1762) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 170/1403] eta: 0:02:22 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0749 (0.1494) acc1: 100.0000 (96.3724) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 180/1403] eta: 0:02:19 acc1_pos: 85.938 acc5_pos: 100.000 loss: 0.0754 (0.1544) acc1: 100.0000 (96.1153) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 190/1403] eta: 0:02:17 acc1_pos: 89.062 acc5_pos: 100.000 loss: 0.3027 (0.1648) acc1: 89.0625 (95.5661) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 200/1403] eta: 0:02:15 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.3027 (0.1774) acc1: 85.9375 (94.9471) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 210/1403] eta: 0:02:13 acc1_pos: 76.562 acc5_pos: 100.000 loss: 0.3300 (0.1882) acc1: 85.9375 (94.3350) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 220/1403] eta: 0:02:11 acc1_pos: 82.812 acc5_pos: 100.000 loss: 0.3955 (0.1986) acc1: 84.3750 (93.8490) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 230/1403] eta: 0:02:09 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2666 (0.1943) acc1: 90.6250 (94.0611) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 240/1403] eta: 0:02:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0755 (0.1894) acc1: 100.0000 (94.2881) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 250/1403] eta: 0:02:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0745 (0.1851) acc1: 100.0000 (94.4908) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 260/1403] eta: 0:02:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0745 (0.1811) acc1: 100.0000 (94.6839) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 270/1403] eta: 0:02:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0752 (0.1773) acc1: 100.0000 (94.8570) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 280/1403] eta: 0:02:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0733 (0.1736) acc1: 100.0000 (95.0345) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 290/1403] eta: 0:01:59 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0684 (0.1701) acc1: 100.0000 (95.1944) acc5: 100.0000 (100.0000) time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 300/1403] eta: 0:01:58 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0683 (0.1668) acc1: 100.0000 (95.3488) acc5: 100.0000 (100.0000) time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 310/1403] eta: 0:01:56 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0709 (0.1640) acc1: 100.0000 (95.4883) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 320/1403] eta: 0:01:55 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0880 (0.1617) acc1: 100.0000 (95.6289) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 330/1403] eta: 0:01:54 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0919 (0.1597) acc1: 100.0000 (95.7610) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 340/1403] eta: 0:01:52 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0905 (0.1576) acc1: 100.0000 (95.8853) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 350/1403] eta: 0:01:51 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0894 (0.1557) acc1: 100.0000 (96.0025) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [ 360/1403] eta: 0:01:50 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0922 (0.1540) acc1: 100.0000 (96.1089) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 370/1403] eta: 0:01:48 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0900 (0.1522) acc1: 100.0000 (96.2138) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 380/1403] eta: 0:01:47 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0895 (0.1506) acc1: 100.0000 (96.3132) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 390/1403] eta: 0:01:46 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0909 (0.1512) acc1: 100.0000 (96.2796) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [ 400/1403] eta: 0:01:44 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2557 (0.1555) acc1: 89.0625 (96.0450) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 410/1403] eta: 0:01:43 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2999 (0.1591) acc1: 87.5000 (95.8371) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 420/1403] eta: 0:01:42 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2581 (0.1609) acc1: 89.0625 (95.7282) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 430/1403] eta: 0:01:41 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2509 (0.1642) acc1: 90.6250 (95.5481) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 440/1403] eta: 0:01:40 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3135 (0.1680) acc1: 87.5000 (95.3373) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 450/1403] eta: 0:01:38 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2743 (0.1705) acc1: 89.0625 (95.2155) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 460/1403] eta: 0:01:37 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2632 (0.1732) acc1: 89.0625 (95.0583) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 470/1403] eta: 0:01:36 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2459 (0.1723) acc1: 90.6250 (95.0902) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 480/1403] eta: 0:01:35 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0581 (0.1699) acc1: 100.0000 (95.1923) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 490/1403] eta: 0:01:34 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0583 (0.1677) acc1: 100.0000 (95.2870) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 500/1403] eta: 0:01:33 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0582 (0.1655) acc1: 100.0000 (95.3811) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 510/1403] eta: 0:01:31 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0582 (0.1634) acc1: 100.0000 (95.4715) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 520/1403] eta: 0:01:30 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0619 (0.1615) acc1: 100.0000 (95.5584) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 530/1403] eta: 0:01:29 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0562 (0.1595) acc1: 100.0000 (95.6421) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 540/1403] eta: 0:01:28 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0558 (0.1576) acc1: 100.0000 (95.7226) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0971 data: 0.0003 max mem: 52150 +Test: [ 550/1403] eta: 0:01:27 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0565 (0.1563) acc1: 100.0000 (95.7832) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0972 data: 0.0003 max mem: 52150 +Test: [ 560/1403] eta: 0:01:26 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0818 (0.1552) acc1: 98.4375 (95.8333) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [ 570/1403] eta: 0:01:25 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0863 (0.1540) acc1: 98.4375 (95.8871) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 580/1403] eta: 0:01:24 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0863 (0.1533) acc1: 98.4375 (95.9149) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 590/1403] eta: 0:01:23 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1021 (0.1528) acc1: 98.4375 (95.9179) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 600/1403] eta: 0:01:21 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1229 (0.1531) acc1: 95.3125 (95.8715) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 610/1403] eta: 0:01:20 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1229 (0.1525) acc1: 95.3125 (95.8777) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 620/1403] eta: 0:01:19 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1288 (0.1533) acc1: 95.3125 (95.7855) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 630/1403] eta: 0:01:18 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1436 (0.1532) acc1: 95.3125 (95.7855) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0968 data: 0.0004 max mem: 52150 +Test: [ 640/1403] eta: 0:01:17 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1193 (0.1528) acc1: 96.8750 (95.8073) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0968 data: 0.0004 max mem: 52150 +Test: [ 650/1403] eta: 0:01:16 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1191 (0.1523) acc1: 96.8750 (95.8309) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 660/1403] eta: 0:01:15 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0954 (0.1512) acc1: 100.0000 (95.8845) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [ 670/1403] eta: 0:01:14 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0669 (0.1500) acc1: 100.0000 (95.9389) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [ 680/1403] eta: 0:01:13 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0824 (0.1522) acc1: 96.8750 (95.7852) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 690/1403] eta: 0:01:12 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3646 (0.1571) acc1: 84.3750 (95.4911) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 700/1403] eta: 0:01:11 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2206 (0.1575) acc1: 85.9375 (95.4708) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 710/1403] eta: 0:01:10 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3212 (0.1616) acc1: 85.9375 (95.2444) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 720/1403] eta: 0:01:09 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3212 (0.1616) acc1: 85.9375 (95.2367) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 730/1403] eta: 0:01:08 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0658 (0.1604) acc1: 100.0000 (95.2890) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 740/1403] eta: 0:01:07 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0728 (0.1595) acc1: 100.0000 (95.3336) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 750/1403] eta: 0:01:05 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1082 (0.1611) acc1: 96.8750 (95.2272) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 760/1403] eta: 0:01:04 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3383 (0.1655) acc1: 81.2500 (94.9368) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [ 770/1403] eta: 0:01:03 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.5027 (0.1703) acc1: 73.4375 (94.6458) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [ 780/1403] eta: 0:01:02 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.4104 (0.1727) acc1: 78.1250 (94.5102) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 790/1403] eta: 0:01:01 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3186 (0.1752) acc1: 84.3750 (94.3841) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 800/1403] eta: 0:01:00 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2254 (0.1757) acc1: 92.1875 (94.3586) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 810/1403] eta: 0:00:59 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2254 (0.1790) acc1: 87.5000 (94.2008) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [ 820/1403] eta: 0:00:58 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.6383 (0.1881) acc1: 71.8750 (93.8166) acc5: 100.0000 (100.0000) acc1_neg: 39.062 acc5_neg: 100.000 time: 0.0970 data: 0.0003 max mem: 52150 +Test: [ 830/1403] eta: 0:00:57 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.7017 (0.1908) acc1: 70.3125 (93.6767) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 840/1403] eta: 0:00:56 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.4888 (0.1989) acc1: 71.8750 (93.3301) acc5: 100.0000 (100.0000) acc1_neg: 48.438 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 850/1403] eta: 0:00:55 acc1_pos: 94.118 acc5_pos: 100.000 loss: 1.0631 (0.2101) acc1: 54.6875 (92.8595) acc5: 100.0000 (100.0000) acc1_neg: 50.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 860/1403] eta: 0:00:54 acc1_pos: 94.118 acc5_pos: 100.000 loss: 1.0354 (0.2136) acc1: 56.2500 (92.7065) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 870/1403] eta: 0:00:53 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1378 (0.2125) acc1: 95.3125 (92.7544) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 880/1403] eta: 0:00:52 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0977 (0.2113) acc1: 98.4375 (92.8136) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 890/1403] eta: 0:00:51 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0859 (0.2099) acc1: 98.4375 (92.8837) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [ 900/1403] eta: 0:00:50 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0722 (0.2083) acc1: 100.0000 (92.9592) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [ 910/1403] eta: 0:00:49 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0740 (0.2075) acc1: 98.4375 (92.9885) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 920/1403] eta: 0:00:48 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1142 (0.2066) acc1: 96.8750 (93.0239) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 930/1403] eta: 0:00:47 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1036 (0.2056) acc1: 96.8750 (93.0770) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 940/1403] eta: 0:00:46 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1263 (0.2059) acc1: 96.8750 (93.0460) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [ 950/1403] eta: 0:00:45 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2760 (0.2071) acc1: 89.0625 (92.9696) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [ 960/1403] eta: 0:00:44 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2869 (0.2079) acc1: 85.9375 (92.9159) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [ 970/1403] eta: 0:00:43 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2770 (0.2091) acc1: 89.0625 (92.8585) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [ 980/1403] eta: 0:00:42 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2901 (0.2101) acc1: 85.9375 (92.7848) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [ 990/1403] eta: 0:00:41 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2782 (0.2108) acc1: 87.5000 (92.7456) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [1000/1403] eta: 0:00:40 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2782 (0.2119) acc1: 87.5000 (92.6886) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [1010/1403] eta: 0:00:39 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3022 (0.2129) acc1: 87.5000 (92.6326) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [1020/1403] eta: 0:00:38 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.4343 (0.2195) acc1: 79.6875 (92.2717) acc5: 100.0000 (100.0000) acc1_neg: 32.812 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [1030/1403] eta: 0:00:37 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.5719 (0.2215) acc1: 59.3750 (92.1451) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [1040/1403] eta: 0:00:36 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2124 (0.2213) acc1: 90.6250 (92.1530) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [1050/1403] eta: 0:00:35 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1660 (0.2208) acc1: 93.7500 (92.1771) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [1060/1403] eta: 0:00:34 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1258 (0.2200) acc1: 95.3125 (92.2243) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0972 data: 0.0003 max mem: 52150 +Test: [1070/1403] eta: 0:00:33 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.1538 (0.2204) acc1: 95.3125 (92.2167) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0972 data: 0.0003 max mem: 52150 +Test: [1080/1403] eta: 0:00:32 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3631 (0.2221) acc1: 85.9375 (92.1268) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [1090/1403] eta: 0:00:31 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.4004 (0.2239) acc1: 81.2500 (92.0328) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1100/1403] eta: 0:00:30 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.3492 (0.2244) acc1: 84.3750 (92.0058) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [1110/1403] eta: 0:00:29 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2730 (0.2248) acc1: 89.0625 (91.9906) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0971 data: 0.0003 max mem: 52150 +Test: [1120/1403] eta: 0:00:28 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2730 (0.2253) acc1: 89.0625 (91.9659) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0972 data: 0.0003 max mem: 52150 +Test: [1130/1403] eta: 0:00:27 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2970 (0.2260) acc1: 87.5000 (91.9264) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0970 data: 0.0003 max mem: 52150 +Test: [1140/1403] eta: 0:00:26 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2431 (0.2261) acc1: 90.6250 (91.9191) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [1150/1403] eta: 0:00:25 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2435 (0.2265) acc1: 90.6250 (91.8861) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1160/1403] eta: 0:00:24 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2728 (0.2270) acc1: 89.0625 (91.8618) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0967 data: 0.0002 max mem: 52150 +Test: [1170/1403] eta: 0:00:23 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.2692 (0.2268) acc1: 89.0625 (91.8726) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1180/1403] eta: 0:00:22 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0648 (0.2255) acc1: 100.0000 (91.9414) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1190/1403] eta: 0:00:21 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0639 (0.2241) acc1: 100.0000 (92.0091) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0002 max mem: 52150 +Test: [1200/1403] eta: 0:00:20 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0637 (0.2228) acc1: 100.0000 (92.0756) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0002 max mem: 52150 +Test: [1210/1403] eta: 0:00:19 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0635 (0.2215) acc1: 100.0000 (92.1411) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0002 max mem: 52150 +Test: [1220/1403] eta: 0:00:18 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0633 (0.2202) acc1: 100.0000 (92.2054) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1230/1403] eta: 0:00:17 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0640 (0.2189) acc1: 100.0000 (92.2687) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1240/1403] eta: 0:00:16 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0640 (0.2176) acc1: 100.0000 (92.3310) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0002 max mem: 52150 +Test: [1250/1403] eta: 0:00:15 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0637 (0.2164) acc1: 100.0000 (92.3923) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0002 max mem: 52150 +Test: [1260/1403] eta: 0:00:14 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0655 (0.2152) acc1: 100.0000 (92.4527) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1270/1403] eta: 0:00:13 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0662 (0.2141) acc1: 100.0000 (92.5120) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1280/1403] eta: 0:00:12 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0662 (0.2129) acc1: 100.0000 (92.5705) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1290/1403] eta: 0:00:11 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0659 (0.2118) acc1: 100.0000 (92.6280) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1300/1403] eta: 0:00:10 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0659 (0.2107) acc1: 100.0000 (92.6847) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1310/1403] eta: 0:00:09 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0662 (0.2096) acc1: 100.0000 (92.7405) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52150 +Test: [1320/1403] eta: 0:00:08 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0653 (0.2085) acc1: 100.0000 (92.7955) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52150 +Test: [1330/1403] eta: 0:00:07 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0653 (0.2074) acc1: 100.0000 (92.8484) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1340/1403] eta: 0:00:06 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0622 (0.2063) acc1: 100.0000 (92.9006) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1350/1403] eta: 0:00:05 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0621 (0.2053) acc1: 100.0000 (92.9520) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1360/1403] eta: 0:00:04 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0615 (0.2042) acc1: 100.0000 (93.0038) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1370/1403] eta: 0:00:03 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0625 (0.2032) acc1: 100.0000 (93.0525) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1380/1403] eta: 0:00:02 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0611 (0.2022) acc1: 100.0000 (93.1028) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1390/1403] eta: 0:00:01 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0604 (0.2012) acc1: 100.0000 (93.1524) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52150 +Test: [1400/1403] eta: 0:00:00 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0608 (0.2003) acc1: 100.0000 (93.1991) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0001 max mem: 52150 +Test: [1402/1403] eta: 0:00:00 acc1_pos: 94.118 acc5_pos: 100.000 loss: 0.0615 (0.2001) acc1: 100.0000 (93.2040) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0923 data: 0.0001 max mem: 52150 +Test: Total time: 0:02:19 (0.0994 s / it) +* Overall Acc@1 93.204 Acc@5 100.000 + target == real Acc@1 95.448 Acc@5 100.000 + target == fake Acc@1 92.571 Acc@5 100.000 + loss 0.200 +Accuracy of the network on the 89729 test images: 93.2% +Max accuracy: 93.40% +Epoch: [20] [ 0/7269] eta: 12:10:25 lr: 0.000006 loss: 0.6456 (0.6456) time: 6.0292 data: 5.6461 max mem: 52150 +Epoch: [20] [ 10/7269] eta: 1:47:24 lr: 0.000006 loss: 0.5798 (0.5588) time: 0.8878 data: 0.5134 max mem: 52150 +Epoch: [20] [ 20/7269] eta: 1:17:35 lr: 0.000006 loss: 0.5439 (0.5500) time: 0.3729 data: 0.0001 max mem: 52150 +Epoch: [20] [ 30/7269] eta: 1:06:57 lr: 0.000006 loss: 0.5240 (0.5443) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [20] [ 40/7269] eta: 1:01:27 lr: 0.000006 loss: 0.5226 (0.5342) time: 0.3714 data: 0.0001 max mem: 52150 +Epoch: [20] [ 50/7269] eta: 0:58:06 lr: 0.000006 loss: 0.5056 (0.5278) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [20] [ 60/7269] eta: 0:55:49 lr: 0.000006 loss: 0.5156 (0.5280) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [20] [ 70/7269] eta: 0:54:10 lr: 0.000006 loss: 0.5403 (0.5253) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [20] [ 80/7269] eta: 0:54:19 lr: 0.000006 loss: 0.5302 (0.5269) time: 0.4192 data: 0.0002 max mem: 52150 +Epoch: [20] [ 90/7269] eta: 0:53:10 lr: 0.000006 loss: 0.5410 (0.5287) time: 0.4193 data: 0.0002 max mem: 52150 +Epoch: [20] [ 100/7269] eta: 0:53:22 lr: 0.000006 loss: 0.5558 (0.5303) time: 0.4193 data: 0.0002 max mem: 52150 +Epoch: [20] [ 110/7269] eta: 0:53:31 lr: 0.000006 loss: 0.5367 (0.5280) time: 0.4677 data: 0.0002 max mem: 52150 +Epoch: [20] [ 120/7269] eta: 0:52:41 lr: 0.000006 loss: 0.5143 (0.5283) time: 0.4195 data: 0.0002 max mem: 52150 +Epoch: [20] [ 130/7269] eta: 0:51:58 lr: 0.000006 loss: 0.5009 (0.5234) time: 0.3711 data: 0.0002 max mem: 52150 +Epoch: [20] [ 140/7269] eta: 0:51:20 lr: 0.000006 loss: 0.5009 (0.5225) time: 0.3711 data: 0.0002 max mem: 52150 +Epoch: [20] [ 150/7269] eta: 0:51:34 lr: 0.000006 loss: 0.5343 (0.5230) time: 0.4214 data: 0.0002 max mem: 52150 +Epoch: [20] [ 160/7269] eta: 0:51:45 lr: 0.000006 loss: 0.5264 (0.5224) time: 0.4701 data: 0.0002 max mem: 52150 +Epoch: [20] [ 170/7269] eta: 0:51:55 lr: 0.000006 loss: 0.5264 (0.5235) time: 0.4705 data: 0.0002 max mem: 52150 +Epoch: [20] [ 180/7269] eta: 0:51:24 lr: 0.000006 loss: 0.5692 (0.5261) time: 0.4218 data: 0.0002 max mem: 52150 +Epoch: [20] [ 190/7269] eta: 0:50:56 lr: 0.000006 loss: 0.5426 (0.5238) time: 0.3712 data: 0.0002 max mem: 52150 +Epoch: [20] [ 200/7269] eta: 0:52:25 lr: 0.000006 loss: 0.5382 (0.5248) time: 0.5336 data: 0.0002 max mem: 52150 +Epoch: [20] [ 210/7269] eta: 0:51:56 lr: 0.000006 loss: 0.5382 (0.5247) time: 0.5334 data: 0.0001 max mem: 52150 +Epoch: [20] [ 220/7269] eta: 0:53:03 lr: 0.000006 loss: 0.5151 (0.5235) time: 0.5185 data: 0.0002 max mem: 52150 +Epoch: [20] [ 230/7269] eta: 0:52:34 lr: 0.000006 loss: 0.5382 (0.5242) time: 0.5185 data: 0.0002 max mem: 52150 +Epoch: [20] [ 240/7269] eta: 0:52:07 lr: 0.000006 loss: 0.5382 (0.5237) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [20] [ 250/7269] eta: 0:52:09 lr: 0.000006 loss: 0.5308 (0.5235) time: 0.4194 data: 0.0002 max mem: 52150 +Epoch: [20] [ 260/7269] eta: 0:52:51 lr: 0.000006 loss: 0.5349 (0.5242) time: 0.5433 data: 0.0002 max mem: 52150 +Epoch: [20] [ 270/7269] eta: 0:52:25 lr: 0.000006 loss: 0.5444 (0.5242) time: 0.4952 data: 0.0002 max mem: 52150 +Epoch: [20] [ 280/7269] eta: 0:52:02 lr: 0.000006 loss: 0.5132 (0.5235) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [20] [ 290/7269] eta: 0:52:02 lr: 0.000006 loss: 0.4720 (0.5220) time: 0.4201 data: 0.0002 max mem: 52150 +Epoch: [20] [ 300/7269] eta: 0:52:03 lr: 0.000006 loss: 0.4917 (0.5219) time: 0.4684 data: 0.0002 max mem: 52150 +Epoch: [20] [ 310/7269] eta: 0:51:41 lr: 0.000006 loss: 0.5532 (0.5230) time: 0.4204 data: 0.0002 max mem: 52150 +Epoch: [20] [ 320/7269] eta: 0:51:21 lr: 0.000006 loss: 0.5458 (0.5235) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [20] [ 330/7269] eta: 0:51:28 lr: 0.000006 loss: 0.5137 (0.5227) time: 0.4348 data: 0.0002 max mem: 52150 +Epoch: [20] [ 340/7269] eta: 0:51:08 lr: 0.000006 loss: 0.4874 (0.5223) time: 0.4346 data: 0.0002 max mem: 52150 +Epoch: [20] [ 350/7269] eta: 0:50:50 lr: 0.000006 loss: 0.5122 (0.5223) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [20] [ 360/7269] eta: 0:50:32 lr: 0.000006 loss: 0.5241 (0.5223) time: 0.3712 data: 0.0002 max mem: 52150 +Epoch: [20] [ 370/7269] eta: 0:50:38 lr: 0.000006 loss: 0.5130 (0.5217) time: 0.4338 data: 0.0002 max mem: 52150 +Epoch: [20] [ 380/7269] eta: 0:50:21 lr: 0.000006 loss: 0.5092 (0.5223) time: 0.4340 data: 0.0002 max mem: 52150 +Epoch: [20] [ 390/7269] eta: 0:50:27 lr: 0.000006 loss: 0.5092 (0.5213) time: 0.4345 data: 0.0002 max mem: 52150 +Epoch: [20] [ 400/7269] eta: 0:50:28 lr: 0.000006 loss: 0.5021 (0.5215) time: 0.4837 data: 0.0002 max mem: 52150 +Epoch: [20] [ 410/7269] eta: 0:50:49 lr: 0.000006 loss: 0.5347 (0.5226) time: 0.5298 data: 0.0001 max mem: 52150 +Epoch: [20] [ 420/7269] eta: 0:50:32 lr: 0.000006 loss: 0.5714 (0.5232) time: 0.4805 data: 0.0002 max mem: 52150 +Epoch: [20] [ 430/7269] eta: 0:50:16 lr: 0.000006 loss: 0.5295 (0.5231) time: 0.3712 data: 0.0002 max mem: 52150 +Epoch: [20] [ 440/7269] eta: 0:50:01 lr: 0.000006 loss: 0.4906 (0.5224) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [20] [ 450/7269] eta: 0:50:01 lr: 0.000006 loss: 0.5283 (0.5221) time: 0.4199 data: 0.0002 max mem: 52150 +Epoch: [20] [ 460/7269] eta: 0:50:01 lr: 0.000006 loss: 0.5243 (0.5219) time: 0.4686 data: 0.0001 max mem: 52150 +Epoch: [20] [ 470/7269] eta: 0:50:00 lr: 0.000006 loss: 0.5243 (0.5217) time: 0.4667 data: 0.0001 max mem: 52150 +Epoch: [20] [ 480/7269] eta: 0:49:46 lr: 0.000006 loss: 0.5274 (0.5218) time: 0.4181 data: 0.0001 max mem: 52150 +Epoch: [20] [ 490/7269] eta: 0:49:50 lr: 0.000006 loss: 0.4771 (0.5209) time: 0.4354 data: 0.0002 max mem: 52150 +Epoch: [20] [ 500/7269] eta: 0:49:36 lr: 0.000006 loss: 0.5493 (0.5219) time: 0.4351 data: 0.0002 max mem: 52150 +Epoch: [20] [ 510/7269] eta: 0:49:22 lr: 0.000006 loss: 0.5585 (0.5213) time: 0.3708 data: 0.0002 max mem: 52150 +Epoch: [20] [ 520/7269] eta: 0:49:21 lr: 0.000006 loss: 0.5372 (0.5219) time: 0.4183 data: 0.0002 max mem: 52150 +Epoch: [20] [ 530/7269] eta: 0:49:08 lr: 0.000006 loss: 0.5372 (0.5217) time: 0.4186 data: 0.0002 max mem: 52150 +Epoch: [20] [ 540/7269] eta: 0:48:56 lr: 0.000006 loss: 0.5065 (0.5218) time: 0.3711 data: 0.0002 max mem: 52150 +Epoch: [20] [ 550/7269] eta: 0:48:43 lr: 0.000006 loss: 0.5299 (0.5223) time: 0.3710 data: 0.0002 max mem: 52150 +Epoch: [20] [ 560/7269] eta: 0:48:31 lr: 0.000006 loss: 0.5299 (0.5222) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [20] [ 570/7269] eta: 0:48:20 lr: 0.000006 loss: 0.5138 (0.5223) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [20] [ 580/7269] eta: 0:48:08 lr: 0.000006 loss: 0.5204 (0.5219) time: 0.3711 data: 0.0001 max mem: 52150 +Epoch: [20] [ 590/7269] eta: 0:48:07 lr: 0.000006 loss: 0.5340 (0.5221) time: 0.4166 data: 0.0001 max mem: 52150 +Epoch: [20] [ 600/7269] eta: 0:48:07 lr: 0.000006 loss: 0.5407 (0.5218) time: 0.4665 data: 0.0001 max mem: 52150 +Epoch: [20] [ 610/7269] eta: 0:48:10 lr: 0.000006 loss: 0.5160 (0.5217) time: 0.4838 data: 0.0001 max mem: 52150 +Epoch: [20] [ 620/7269] eta: 0:48:09 lr: 0.000006 loss: 0.5166 (0.5222) time: 0.4797 data: 0.0001 max mem: 52150 +Epoch: [20] [ 630/7269] eta: 0:47:58 lr: 0.000006 loss: 0.5120 (0.5217) time: 0.4173 data: 0.0002 max mem: 52150 +Epoch: [20] [ 640/7269] eta: 0:47:47 lr: 0.000006 loss: 0.5256 (0.5220) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [20] [ 650/7269] eta: 0:47:46 lr: 0.000006 loss: 0.5555 (0.5224) time: 0.4188 data: 0.0002 max mem: 52150 +Epoch: [20] [ 660/7269] eta: 0:47:45 lr: 0.000006 loss: 0.5244 (0.5225) time: 0.4636 data: 0.0002 max mem: 52150 +Epoch: [20] [ 670/7269] eta: 0:47:34 lr: 0.000006 loss: 0.5046 (0.5221) time: 0.4170 data: 0.0002 max mem: 52150 +Epoch: [20] [ 680/7269] eta: 0:47:24 lr: 0.000006 loss: 0.5132 (0.5218) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [20] [ 690/7269] eta: 0:47:14 lr: 0.000006 loss: 0.5290 (0.5216) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [20] [ 700/7269] eta: 0:47:21 lr: 0.000006 loss: 0.5081 (0.5215) time: 0.4634 data: 0.0001 max mem: 52150 +Epoch: [20] [ 710/7269] eta: 0:47:11 lr: 0.000006 loss: 0.5040 (0.5211) time: 0.4635 data: 0.0002 max mem: 52150 +Epoch: [20] [ 720/7269] eta: 0:47:10 lr: 0.000006 loss: 0.5100 (0.5209) time: 0.4182 data: 0.0002 max mem: 52150 +Epoch: [20] [ 730/7269] eta: 0:47:00 lr: 0.000006 loss: 0.5290 (0.5211) time: 0.4184 data: 0.0001 max mem: 52150 +Epoch: [20] [ 740/7269] eta: 0:46:59 lr: 0.000006 loss: 0.5491 (0.5211) time: 0.4189 data: 0.0001 max mem: 52150 +Epoch: [20] [ 750/7269] eta: 0:46:50 lr: 0.000006 loss: 0.5144 (0.5210) time: 0.4195 data: 0.0002 max mem: 52150 +Epoch: [20] [ 760/7269] eta: 0:46:40 lr: 0.000006 loss: 0.5144 (0.5209) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [20] [ 770/7269] eta: 0:46:39 lr: 0.000006 loss: 0.5261 (0.5211) time: 0.4172 data: 0.0002 max mem: 52150 +Epoch: [20] [ 780/7269] eta: 0:46:49 lr: 0.000006 loss: 0.5244 (0.5207) time: 0.5370 data: 0.0002 max mem: 52150 +Epoch: [20] [ 790/7269] eta: 0:46:40 lr: 0.000006 loss: 0.5377 (0.5212) time: 0.4915 data: 0.0001 max mem: 52150 +Epoch: [20] [ 800/7269] eta: 0:46:38 lr: 0.000006 loss: 0.5534 (0.5211) time: 0.4173 data: 0.0001 max mem: 52150 +Epoch: [20] [ 810/7269] eta: 0:46:29 lr: 0.000006 loss: 0.5015 (0.5209) time: 0.4171 data: 0.0001 max mem: 52150 +Epoch: [20] [ 820/7269] eta: 0:46:27 lr: 0.000006 loss: 0.5082 (0.5209) time: 0.4169 data: 0.0002 max mem: 52150 +Epoch: [20] [ 830/7269] eta: 0:46:25 lr: 0.000006 loss: 0.5318 (0.5212) time: 0.4609 data: 0.0002 max mem: 52150 +Epoch: [20] [ 840/7269] eta: 0:46:16 lr: 0.000006 loss: 0.5277 (0.5210) time: 0.4157 data: 0.0002 max mem: 52150 +Epoch: [20] [ 850/7269] eta: 0:46:07 lr: 0.000006 loss: 0.4822 (0.5206) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [20] [ 860/7269] eta: 0:46:07 lr: 0.000006 loss: 0.4732 (0.5204) time: 0.4314 data: 0.0002 max mem: 52150 +Epoch: [20] [ 870/7269] eta: 0:45:58 lr: 0.000006 loss: 0.5119 (0.5202) time: 0.4316 data: 0.0002 max mem: 52150 +Epoch: [20] [ 880/7269] eta: 0:45:58 lr: 0.000006 loss: 0.5220 (0.5202) time: 0.4311 data: 0.0001 max mem: 52150 +Epoch: [20] [ 890/7269] eta: 0:45:56 lr: 0.000006 loss: 0.5298 (0.5204) time: 0.4762 data: 0.0001 max mem: 52150 +Epoch: [20] [ 900/7269] eta: 0:45:48 lr: 0.000006 loss: 0.5306 (0.5206) time: 0.4168 data: 0.0001 max mem: 52150 +Epoch: [20] [ 910/7269] eta: 0:45:45 lr: 0.000006 loss: 0.5344 (0.5205) time: 0.4163 data: 0.0001 max mem: 52150 +Epoch: [20] [ 920/7269] eta: 0:45:37 lr: 0.000006 loss: 0.5193 (0.5203) time: 0.4159 data: 0.0001 max mem: 52150 +Epoch: [20] [ 930/7269] eta: 0:45:29 lr: 0.000006 loss: 0.5128 (0.5200) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [20] [ 940/7269] eta: 0:45:27 lr: 0.000006 loss: 0.5523 (0.5206) time: 0.4184 data: 0.0001 max mem: 52150 +Epoch: [20] [ 950/7269] eta: 0:45:18 lr: 0.000006 loss: 0.5318 (0.5206) time: 0.4181 data: 0.0001 max mem: 52150 +Epoch: [20] [ 960/7269] eta: 0:45:18 lr: 0.000006 loss: 0.5042 (0.5205) time: 0.4343 data: 0.0001 max mem: 52150 +Epoch: [20] [ 970/7269] eta: 0:45:10 lr: 0.000006 loss: 0.5169 (0.5210) time: 0.4340 data: 0.0001 max mem: 52150 +Epoch: [20] [ 980/7269] eta: 0:45:08 lr: 0.000006 loss: 0.5460 (0.5212) time: 0.4171 data: 0.0001 max mem: 52150 +Epoch: [20] [ 990/7269] eta: 0:45:19 lr: 0.000006 loss: 0.5406 (0.5211) time: 0.5700 data: 0.0001 max mem: 52150 +Epoch: [20] [1000/7269] eta: 0:45:11 lr: 0.000006 loss: 0.5399 (0.5211) time: 0.5241 data: 0.0001 max mem: 52150 +Epoch: [20] [1010/7269] eta: 0:45:10 lr: 0.000006 loss: 0.5524 (0.5214) time: 0.4311 data: 0.0001 max mem: 52150 +Epoch: [20] [1020/7269] eta: 0:45:02 lr: 0.000006 loss: 0.5684 (0.5215) time: 0.4316 data: 0.0002 max mem: 52150 +Epoch: [20] [1030/7269] eta: 0:44:54 lr: 0.000006 loss: 0.5257 (0.5215) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [20] [1040/7269] eta: 0:44:54 lr: 0.000006 loss: 0.5163 (0.5216) time: 0.4320 data: 0.0002 max mem: 52150 +Epoch: [20] [1050/7269] eta: 0:44:46 lr: 0.000006 loss: 0.5562 (0.5218) time: 0.4324 data: 0.0002 max mem: 52150 +Epoch: [20] [1060/7269] eta: 0:44:38 lr: 0.000006 loss: 0.5229 (0.5217) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [20] [1070/7269] eta: 0:44:30 lr: 0.000006 loss: 0.5180 (0.5220) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [20] [1080/7269] eta: 0:44:22 lr: 0.000006 loss: 0.5540 (0.5223) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [20] [1090/7269] eta: 0:44:20 lr: 0.000006 loss: 0.5260 (0.5219) time: 0.4168 data: 0.0001 max mem: 52150 +Epoch: [20] [1100/7269] eta: 0:44:24 lr: 0.000006 loss: 0.4917 (0.5220) time: 0.5237 data: 0.0001 max mem: 52150 +Epoch: [20] [1110/7269] eta: 0:44:17 lr: 0.000006 loss: 0.5056 (0.5219) time: 0.4783 data: 0.0001 max mem: 52150 +Epoch: [20] [1120/7269] eta: 0:44:09 lr: 0.000006 loss: 0.5198 (0.5220) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [20] [1130/7269] eta: 0:44:02 lr: 0.000006 loss: 0.5347 (0.5218) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [20] [1140/7269] eta: 0:43:54 lr: 0.000006 loss: 0.5334 (0.5218) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [20] [1150/7269] eta: 0:43:52 lr: 0.000006 loss: 0.4783 (0.5214) time: 0.4170 data: 0.0002 max mem: 52150 +Epoch: [20] [1160/7269] eta: 0:43:44 lr: 0.000006 loss: 0.4783 (0.5214) time: 0.4170 data: 0.0001 max mem: 52150 +Epoch: [20] [1170/7269] eta: 0:43:37 lr: 0.000006 loss: 0.5453 (0.5215) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [20] [1180/7269] eta: 0:43:34 lr: 0.000006 loss: 0.5258 (0.5215) time: 0.4164 data: 0.0001 max mem: 52150 +Epoch: [20] [1190/7269] eta: 0:43:42 lr: 0.000006 loss: 0.5268 (0.5216) time: 0.5656 data: 0.0001 max mem: 52150 +Epoch: [20] [1200/7269] eta: 0:43:35 lr: 0.000006 loss: 0.5362 (0.5215) time: 0.5208 data: 0.0001 max mem: 52150 +Epoch: [20] [1210/7269] eta: 0:43:28 lr: 0.000006 loss: 0.5255 (0.5213) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [20] [1220/7269] eta: 0:43:20 lr: 0.000006 loss: 0.5146 (0.5212) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [20] [1230/7269] eta: 0:43:13 lr: 0.000006 loss: 0.4883 (0.5209) time: 0.3726 data: 0.0001 max mem: 52150 +Epoch: [20] [1240/7269] eta: 0:43:06 lr: 0.000006 loss: 0.5032 (0.5209) time: 0.3721 data: 0.0001 max mem: 52150 +Epoch: [20] [1250/7269] eta: 0:43:05 lr: 0.000006 loss: 0.5445 (0.5209) time: 0.4333 data: 0.0002 max mem: 52150 +Epoch: [20] [1260/7269] eta: 0:42:58 lr: 0.000006 loss: 0.5445 (0.5209) time: 0.4332 data: 0.0002 max mem: 52150 +Epoch: [20] [1270/7269] eta: 0:42:55 lr: 0.000006 loss: 0.5255 (0.5208) time: 0.4172 data: 0.0001 max mem: 52150 +Epoch: [20] [1280/7269] eta: 0:42:53 lr: 0.000006 loss: 0.5214 (0.5208) time: 0.4632 data: 0.0001 max mem: 52150 +Epoch: [20] [1290/7269] eta: 0:42:50 lr: 0.000006 loss: 0.5339 (0.5211) time: 0.4633 data: 0.0001 max mem: 52150 +Epoch: [20] [1300/7269] eta: 0:42:51 lr: 0.000006 loss: 0.5339 (0.5211) time: 0.5098 data: 0.0002 max mem: 52150 +Epoch: [20] [1310/7269] eta: 0:42:44 lr: 0.000006 loss: 0.5433 (0.5214) time: 0.4645 data: 0.0002 max mem: 52150 +Epoch: [20] [1320/7269] eta: 0:42:42 lr: 0.000006 loss: 0.5402 (0.5214) time: 0.4173 data: 0.0001 max mem: 52150 +Epoch: [20] [1330/7269] eta: 0:42:39 lr: 0.000006 loss: 0.5134 (0.5214) time: 0.4642 data: 0.0001 max mem: 52150 +Epoch: [20] [1340/7269] eta: 0:42:42 lr: 0.000006 loss: 0.5118 (0.5212) time: 0.5281 data: 0.0002 max mem: 52150 +Epoch: [20] [1350/7269] eta: 0:42:39 lr: 0.000006 loss: 0.5198 (0.5213) time: 0.5260 data: 0.0002 max mem: 52150 +Epoch: [20] [1360/7269] eta: 0:42:32 lr: 0.000006 loss: 0.5269 (0.5215) time: 0.4171 data: 0.0002 max mem: 52150 +Epoch: [20] [1370/7269] eta: 0:42:25 lr: 0.000006 loss: 0.5418 (0.5216) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [20] [1380/7269] eta: 0:42:27 lr: 0.000006 loss: 0.5418 (0.5216) time: 0.4779 data: 0.0002 max mem: 52150 +Epoch: [20] [1390/7269] eta: 0:42:20 lr: 0.000006 loss: 0.5438 (0.5218) time: 0.4779 data: 0.0002 max mem: 52150 +Epoch: [20] [1400/7269] eta: 0:42:13 lr: 0.000006 loss: 0.5138 (0.5215) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [20] [1410/7269] eta: 0:42:06 lr: 0.000006 loss: 0.4633 (0.5212) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [20] [1420/7269] eta: 0:42:00 lr: 0.000006 loss: 0.5322 (0.5214) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [20] [1430/7269] eta: 0:41:53 lr: 0.000006 loss: 0.5510 (0.5214) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [20] [1440/7269] eta: 0:41:46 lr: 0.000006 loss: 0.4845 (0.5212) time: 0.3725 data: 0.0001 max mem: 52150 +Epoch: [20] [1450/7269] eta: 0:41:47 lr: 0.000006 loss: 0.5322 (0.5213) time: 0.4652 data: 0.0001 max mem: 52150 +Epoch: [20] [1460/7269] eta: 0:41:40 lr: 0.000006 loss: 0.5340 (0.5213) time: 0.4644 data: 0.0001 max mem: 52150 +Epoch: [20] [1470/7269] eta: 0:41:34 lr: 0.000006 loss: 0.5324 (0.5212) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [20] [1480/7269] eta: 0:41:31 lr: 0.000006 loss: 0.5251 (0.5211) time: 0.4188 data: 0.0002 max mem: 52150 +Epoch: [20] [1490/7269] eta: 0:41:29 lr: 0.000006 loss: 0.5065 (0.5210) time: 0.4784 data: 0.0002 max mem: 52150 +Epoch: [20] [1500/7269] eta: 0:41:26 lr: 0.000006 loss: 0.5126 (0.5210) time: 0.4787 data: 0.0001 max mem: 52150 +Epoch: [20] [1510/7269] eta: 0:41:19 lr: 0.000006 loss: 0.5367 (0.5212) time: 0.4197 data: 0.0002 max mem: 52150 +Epoch: [20] [1520/7269] eta: 0:41:17 lr: 0.000006 loss: 0.5766 (0.5212) time: 0.4326 data: 0.0002 max mem: 52150 +Epoch: [20] [1530/7269] eta: 0:41:11 lr: 0.000006 loss: 0.5298 (0.5212) time: 0.4318 data: 0.0001 max mem: 52150 +Epoch: [20] [1540/7269] eta: 0:41:08 lr: 0.000006 loss: 0.5484 (0.5214) time: 0.4168 data: 0.0002 max mem: 52150 +Epoch: [20] [1550/7269] eta: 0:41:09 lr: 0.000006 loss: 0.5665 (0.5215) time: 0.5251 data: 0.0002 max mem: 52150 +Epoch: [20] [1560/7269] eta: 0:41:07 lr: 0.000006 loss: 0.5371 (0.5214) time: 0.5398 data: 0.0001 max mem: 52150 +Epoch: [20] [1570/7269] eta: 0:41:01 lr: 0.000006 loss: 0.5230 (0.5213) time: 0.4315 data: 0.0001 max mem: 52150 +Epoch: [20] [1580/7269] eta: 0:40:57 lr: 0.000006 loss: 0.5254 (0.5214) time: 0.4159 data: 0.0001 max mem: 52150 +Epoch: [20] [1590/7269] eta: 0:40:51 lr: 0.000006 loss: 0.5133 (0.5212) time: 0.4163 data: 0.0001 max mem: 52150 +Epoch: [20] [1600/7269] eta: 0:40:44 lr: 0.000006 loss: 0.5133 (0.5212) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [20] [1610/7269] eta: 0:40:38 lr: 0.000006 loss: 0.4995 (0.5209) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [20] [1620/7269] eta: 0:40:32 lr: 0.000006 loss: 0.5438 (0.5212) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [20] [1630/7269] eta: 0:40:25 lr: 0.000006 loss: 0.5490 (0.5213) time: 0.3710 data: 0.0001 max mem: 52150 +Epoch: [20] [1640/7269] eta: 0:40:19 lr: 0.000006 loss: 0.5102 (0.5212) time: 0.3712 data: 0.0001 max mem: 52150 +Epoch: [20] [1650/7269] eta: 0:40:16 lr: 0.000006 loss: 0.5493 (0.5213) time: 0.4170 data: 0.0001 max mem: 52150 +Epoch: [20] [1660/7269] eta: 0:40:14 lr: 0.000006 loss: 0.5160 (0.5211) time: 0.4827 data: 0.0002 max mem: 52150 +Epoch: [20] [1670/7269] eta: 0:40:08 lr: 0.000006 loss: 0.5069 (0.5209) time: 0.4377 data: 0.0002 max mem: 52150 +Epoch: [20] [1680/7269] eta: 0:40:01 lr: 0.000006 loss: 0.5123 (0.5209) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [20] [1690/7269] eta: 0:39:55 lr: 0.000006 loss: 0.5140 (0.5209) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [20] [1700/7269] eta: 0:39:49 lr: 0.000006 loss: 0.5274 (0.5209) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [20] [1710/7269] eta: 0:39:43 lr: 0.000006 loss: 0.5375 (0.5209) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [20] [1720/7269] eta: 0:39:37 lr: 0.000006 loss: 0.5234 (0.5210) time: 0.3711 data: 0.0002 max mem: 52150 +Epoch: [20] [1730/7269] eta: 0:39:34 lr: 0.000006 loss: 0.5165 (0.5211) time: 0.4169 data: 0.0002 max mem: 52150 +Epoch: [20] [1740/7269] eta: 0:39:30 lr: 0.000006 loss: 0.5153 (0.5210) time: 0.4624 data: 0.0002 max mem: 52150 +Epoch: [20] [1750/7269] eta: 0:39:24 lr: 0.000006 loss: 0.5378 (0.5212) time: 0.4167 data: 0.0002 max mem: 52150 +Epoch: [20] [1760/7269] eta: 0:39:18 lr: 0.000006 loss: 0.5651 (0.5215) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [20] [1770/7269] eta: 0:39:19 lr: 0.000006 loss: 0.5465 (0.5215) time: 0.4817 data: 0.0002 max mem: 52150 +Epoch: [20] [1780/7269] eta: 0:39:13 lr: 0.000006 loss: 0.5420 (0.5216) time: 0.4818 data: 0.0001 max mem: 52150 +Epoch: [20] [1790/7269] eta: 0:39:10 lr: 0.000006 loss: 0.5490 (0.5216) time: 0.4194 data: 0.0001 max mem: 52150 +Epoch: [20] [1800/7269] eta: 0:39:04 lr: 0.000006 loss: 0.5390 (0.5216) time: 0.4193 data: 0.0001 max mem: 52150 +Epoch: [20] [1810/7269] eta: 0:39:01 lr: 0.000006 loss: 0.5436 (0.5218) time: 0.4179 data: 0.0002 max mem: 52150 +Epoch: [20] [1820/7269] eta: 0:38:55 lr: 0.000006 loss: 0.5323 (0.5218) time: 0.4180 data: 0.0002 max mem: 52150 +Epoch: [20] [1830/7269] eta: 0:38:52 lr: 0.000006 loss: 0.5008 (0.5217) time: 0.4331 data: 0.0001 max mem: 52150 +Epoch: [20] [1840/7269] eta: 0:38:50 lr: 0.000006 loss: 0.5035 (0.5216) time: 0.4960 data: 0.0002 max mem: 52150 +Epoch: [20] [1850/7269] eta: 0:38:44 lr: 0.000006 loss: 0.5306 (0.5216) time: 0.4350 data: 0.0002 max mem: 52150 +Epoch: [20] [1860/7269] eta: 0:38:38 lr: 0.000006 loss: 0.5487 (0.5217) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [20] [1870/7269] eta: 0:38:35 lr: 0.000006 loss: 0.5471 (0.5218) time: 0.4222 data: 0.0001 max mem: 52150 +Epoch: [20] [1880/7269] eta: 0:38:32 lr: 0.000006 loss: 0.5247 (0.5216) time: 0.4813 data: 0.0001 max mem: 52150 +Epoch: [20] [1890/7269] eta: 0:38:29 lr: 0.000006 loss: 0.5228 (0.5216) time: 0.4753 data: 0.0001 max mem: 52150 +Epoch: [20] [1900/7269] eta: 0:38:23 lr: 0.000006 loss: 0.5203 (0.5216) time: 0.4163 data: 0.0001 max mem: 52150 +Epoch: [20] [1910/7269] eta: 0:38:17 lr: 0.000006 loss: 0.5065 (0.5216) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [20] [1920/7269] eta: 0:38:11 lr: 0.000006 loss: 0.5057 (0.5216) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [20] [1930/7269] eta: 0:38:05 lr: 0.000006 loss: 0.5285 (0.5216) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [20] [1940/7269] eta: 0:38:02 lr: 0.000006 loss: 0.5242 (0.5216) time: 0.4174 data: 0.0002 max mem: 52150 +Epoch: [20] [1950/7269] eta: 0:37:56 lr: 0.000006 loss: 0.5175 (0.5217) time: 0.4172 data: 0.0001 max mem: 52150 +Epoch: [20] [1960/7269] eta: 0:37:53 lr: 0.000006 loss: 0.4872 (0.5215) time: 0.4172 data: 0.0001 max mem: 52150 +Epoch: [20] [1970/7269] eta: 0:37:52 lr: 0.000006 loss: 0.4934 (0.5216) time: 0.5111 data: 0.0001 max mem: 52150 +Epoch: [20] [1980/7269] eta: 0:37:49 lr: 0.000006 loss: 0.5044 (0.5214) time: 0.5096 data: 0.0001 max mem: 52150 +Epoch: [20] [1990/7269] eta: 0:37:43 lr: 0.000006 loss: 0.5013 (0.5213) time: 0.4157 data: 0.0002 max mem: 52150 +Epoch: [20] [2000/7269] eta: 0:37:43 lr: 0.000006 loss: 0.5013 (0.5212) time: 0.4771 data: 0.0002 max mem: 52150 +Epoch: [20] [2010/7269] eta: 0:37:37 lr: 0.000006 loss: 0.5371 (0.5212) time: 0.4781 data: 0.0001 max mem: 52150 +Epoch: [20] [2020/7269] eta: 0:37:39 lr: 0.000006 loss: 0.5546 (0.5212) time: 0.5257 data: 0.0001 max mem: 52150 +Epoch: [20] [2030/7269] eta: 0:37:33 lr: 0.000006 loss: 0.5350 (0.5213) time: 0.5248 data: 0.0002 max mem: 52150 +Epoch: [20] [2040/7269] eta: 0:37:28 lr: 0.000006 loss: 0.5302 (0.5212) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [20] [2050/7269] eta: 0:37:22 lr: 0.000006 loss: 0.5291 (0.5213) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [20] [2060/7269] eta: 0:37:18 lr: 0.000006 loss: 0.5119 (0.5211) time: 0.4178 data: 0.0002 max mem: 52150 +Epoch: [20] [2070/7269] eta: 0:37:18 lr: 0.000006 loss: 0.4973 (0.5211) time: 0.5267 data: 0.0001 max mem: 52150 +Epoch: [20] [2080/7269] eta: 0:37:12 lr: 0.000006 loss: 0.5244 (0.5211) time: 0.4814 data: 0.0002 max mem: 52150 +Epoch: [20] [2090/7269] eta: 0:37:07 lr: 0.000006 loss: 0.5410 (0.5212) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [20] [2100/7269] eta: 0:37:03 lr: 0.000006 loss: 0.5456 (0.5211) time: 0.4178 data: 0.0002 max mem: 52150 +Epoch: [20] [2110/7269] eta: 0:36:57 lr: 0.000006 loss: 0.4891 (0.5209) time: 0.4181 data: 0.0002 max mem: 52150 +Epoch: [20] [2120/7269] eta: 0:36:52 lr: 0.000006 loss: 0.5040 (0.5208) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [20] [2130/7269] eta: 0:36:49 lr: 0.000006 loss: 0.5345 (0.5209) time: 0.4309 data: 0.0001 max mem: 52150 +Epoch: [20] [2140/7269] eta: 0:36:45 lr: 0.000006 loss: 0.5028 (0.5207) time: 0.4775 data: 0.0001 max mem: 52150 +Epoch: [20] [2150/7269] eta: 0:36:43 lr: 0.000006 loss: 0.4854 (0.5207) time: 0.4806 data: 0.0001 max mem: 52150 +Epoch: [20] [2160/7269] eta: 0:36:40 lr: 0.000006 loss: 0.5279 (0.5208) time: 0.4977 data: 0.0002 max mem: 52150 +Epoch: [20] [2170/7269] eta: 0:36:37 lr: 0.000006 loss: 0.5639 (0.5210) time: 0.4972 data: 0.0002 max mem: 52150 +Epoch: [20] [2180/7269] eta: 0:36:31 lr: 0.000006 loss: 0.5644 (0.5210) time: 0.4333 data: 0.0002 max mem: 52150 +Epoch: [20] [2190/7269] eta: 0:36:26 lr: 0.000006 loss: 0.5462 (0.5211) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [20] [2200/7269] eta: 0:36:20 lr: 0.000006 loss: 0.5411 (0.5210) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [20] [2210/7269] eta: 0:36:15 lr: 0.000006 loss: 0.5086 (0.5210) time: 0.3727 data: 0.0002 max mem: 52150 +Epoch: [20] [2220/7269] eta: 0:36:09 lr: 0.000006 loss: 0.5086 (0.5209) time: 0.3727 data: 0.0002 max mem: 52150 +Epoch: [20] [2230/7269] eta: 0:36:03 lr: 0.000006 loss: 0.5202 (0.5210) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [20] [2240/7269] eta: 0:36:00 lr: 0.000006 loss: 0.5340 (0.5210) time: 0.4174 data: 0.0002 max mem: 52150 +Epoch: [20] [2250/7269] eta: 0:35:54 lr: 0.000006 loss: 0.5169 (0.5209) time: 0.4168 data: 0.0002 max mem: 52150 +Epoch: [20] [2260/7269] eta: 0:35:49 lr: 0.000006 loss: 0.5269 (0.5210) time: 0.3712 data: 0.0002 max mem: 52150 +Epoch: [20] [2270/7269] eta: 0:35:45 lr: 0.000006 loss: 0.5414 (0.5209) time: 0.4185 data: 0.0002 max mem: 52150 +Epoch: [20] [2280/7269] eta: 0:35:40 lr: 0.000006 loss: 0.5207 (0.5210) time: 0.4187 data: 0.0002 max mem: 52150 +Epoch: [20] [2290/7269] eta: 0:35:34 lr: 0.000006 loss: 0.5276 (0.5209) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [20] [2300/7269] eta: 0:35:31 lr: 0.000006 loss: 0.5276 (0.5209) time: 0.4338 data: 0.0001 max mem: 52150 +Epoch: [20] [2310/7269] eta: 0:35:26 lr: 0.000006 loss: 0.5288 (0.5209) time: 0.4339 data: 0.0001 max mem: 52150 +Epoch: [20] [2320/7269] eta: 0:35:25 lr: 0.000006 loss: 0.5196 (0.5209) time: 0.4822 data: 0.0002 max mem: 52150 +Epoch: [20] [2330/7269] eta: 0:35:19 lr: 0.000006 loss: 0.5127 (0.5209) time: 0.4820 data: 0.0002 max mem: 52150 +Epoch: [20] [2340/7269] eta: 0:35:14 lr: 0.000006 loss: 0.5127 (0.5208) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [20] [2350/7269] eta: 0:35:08 lr: 0.000006 loss: 0.4959 (0.5207) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [20] [2360/7269] eta: 0:35:03 lr: 0.000006 loss: 0.5151 (0.5208) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [20] [2370/7269] eta: 0:34:57 lr: 0.000006 loss: 0.5660 (0.5209) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [20] [2380/7269] eta: 0:34:54 lr: 0.000006 loss: 0.5658 (0.5208) time: 0.4161 data: 0.0002 max mem: 52150 +Epoch: [20] [2390/7269] eta: 0:34:50 lr: 0.000006 loss: 0.5567 (0.5209) time: 0.4628 data: 0.0002 max mem: 52150 +Epoch: [20] [2400/7269] eta: 0:34:45 lr: 0.000006 loss: 0.5582 (0.5210) time: 0.4185 data: 0.0001 max mem: 52150 +Epoch: [20] [2410/7269] eta: 0:34:39 lr: 0.000006 loss: 0.5402 (0.5210) time: 0.3722 data: 0.0001 max mem: 52150 +Epoch: [20] [2420/7269] eta: 0:34:34 lr: 0.000006 loss: 0.5131 (0.5209) time: 0.3729 data: 0.0002 max mem: 52150 +Epoch: [20] [2430/7269] eta: 0:34:31 lr: 0.000006 loss: 0.4961 (0.5208) time: 0.4373 data: 0.0002 max mem: 52150 +Epoch: [20] [2440/7269] eta: 0:34:26 lr: 0.000006 loss: 0.5031 (0.5207) time: 0.4365 data: 0.0002 max mem: 52150 +Epoch: [20] [2450/7269] eta: 0:34:22 lr: 0.000006 loss: 0.5324 (0.5208) time: 0.4174 data: 0.0001 max mem: 52150 +Epoch: [20] [2460/7269] eta: 0:34:22 lr: 0.000006 loss: 0.5419 (0.5207) time: 0.5389 data: 0.0001 max mem: 52150 +Epoch: [20] [2470/7269] eta: 0:34:18 lr: 0.000006 loss: 0.4762 (0.5204) time: 0.5398 data: 0.0002 max mem: 52150 +Epoch: [20] [2480/7269] eta: 0:34:15 lr: 0.000006 loss: 0.4477 (0.5203) time: 0.4781 data: 0.0002 max mem: 52150 +Epoch: [20] [2490/7269] eta: 0:34:11 lr: 0.000006 loss: 0.4811 (0.5202) time: 0.4789 data: 0.0002 max mem: 52150 +Epoch: [20] [2500/7269] eta: 0:34:06 lr: 0.000006 loss: 0.4919 (0.5201) time: 0.4194 data: 0.0002 max mem: 52150 +Epoch: [20] [2510/7269] eta: 0:34:02 lr: 0.000006 loss: 0.5123 (0.5200) time: 0.4183 data: 0.0002 max mem: 52150 +Epoch: [20] [2520/7269] eta: 0:33:57 lr: 0.000006 loss: 0.5064 (0.5201) time: 0.4184 data: 0.0002 max mem: 52150 +Epoch: [20] [2530/7269] eta: 0:33:54 lr: 0.000006 loss: 0.5565 (0.5202) time: 0.4353 data: 0.0002 max mem: 52150 +Epoch: [20] [2540/7269] eta: 0:33:51 lr: 0.000006 loss: 0.5473 (0.5203) time: 0.4957 data: 0.0002 max mem: 52150 +Epoch: [20] [2550/7269] eta: 0:33:46 lr: 0.000006 loss: 0.5447 (0.5203) time: 0.4323 data: 0.0002 max mem: 52150 +Epoch: [20] [2560/7269] eta: 0:33:43 lr: 0.000006 loss: 0.5388 (0.5204) time: 0.4625 data: 0.0002 max mem: 52150 +Epoch: [20] [2570/7269] eta: 0:33:38 lr: 0.000006 loss: 0.5282 (0.5204) time: 0.4629 data: 0.0002 max mem: 52150 +Epoch: [20] [2580/7269] eta: 0:33:35 lr: 0.000006 loss: 0.4950 (0.5202) time: 0.4189 data: 0.0002 max mem: 52150 +Epoch: [20] [2590/7269] eta: 0:33:31 lr: 0.000006 loss: 0.5115 (0.5203) time: 0.4800 data: 0.0002 max mem: 52150 +Epoch: [20] [2600/7269] eta: 0:33:30 lr: 0.000006 loss: 0.5301 (0.5203) time: 0.5519 data: 0.0001 max mem: 52150 +Epoch: [20] [2610/7269] eta: 0:33:31 lr: 0.000006 loss: 0.5523 (0.5204) time: 0.6604 data: 0.0002 max mem: 52150 +Epoch: [20] [2620/7269] eta: 0:33:26 lr: 0.000006 loss: 0.5356 (0.5203) time: 0.5418 data: 0.0002 max mem: 52150 +Epoch: [20] [2630/7269] eta: 0:33:20 lr: 0.000006 loss: 0.4808 (0.5203) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [20] [2640/7269] eta: 0:33:15 lr: 0.000006 loss: 0.5460 (0.5204) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [20] [2650/7269] eta: 0:33:12 lr: 0.000006 loss: 0.5398 (0.5204) time: 0.4318 data: 0.0002 max mem: 52150 +Epoch: [20] [2660/7269] eta: 0:33:08 lr: 0.000006 loss: 0.5123 (0.5204) time: 0.4765 data: 0.0002 max mem: 52150 +Epoch: [20] [2670/7269] eta: 0:33:03 lr: 0.000006 loss: 0.4797 (0.5202) time: 0.4162 data: 0.0002 max mem: 52150 +Epoch: [20] [2680/7269] eta: 0:32:57 lr: 0.000006 loss: 0.4964 (0.5202) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [20] [2690/7269] eta: 0:32:52 lr: 0.000006 loss: 0.5257 (0.5203) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [20] [2700/7269] eta: 0:32:48 lr: 0.000006 loss: 0.5532 (0.5204) time: 0.4168 data: 0.0002 max mem: 52150 +Epoch: [20] [2710/7269] eta: 0:32:46 lr: 0.000006 loss: 0.5532 (0.5203) time: 0.5236 data: 0.0002 max mem: 52150 +Epoch: [20] [2720/7269] eta: 0:32:43 lr: 0.000006 loss: 0.5312 (0.5205) time: 0.5423 data: 0.0002 max mem: 52150 +Epoch: [20] [2730/7269] eta: 0:32:38 lr: 0.000006 loss: 0.5514 (0.5205) time: 0.4357 data: 0.0002 max mem: 52150 +Epoch: [20] [2740/7269] eta: 0:32:33 lr: 0.000006 loss: 0.5400 (0.5206) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [20] [2750/7269] eta: 0:32:27 lr: 0.000006 loss: 0.5383 (0.5206) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [20] [2760/7269] eta: 0:32:22 lr: 0.000006 loss: 0.5383 (0.5207) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [20] [2770/7269] eta: 0:32:17 lr: 0.000006 loss: 0.5780 (0.5208) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [20] [2780/7269] eta: 0:32:12 lr: 0.000006 loss: 0.5369 (0.5208) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [20] [2790/7269] eta: 0:32:06 lr: 0.000006 loss: 0.4866 (0.5206) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [20] [2800/7269] eta: 0:32:01 lr: 0.000006 loss: 0.4644 (0.5206) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [20] [2810/7269] eta: 0:31:58 lr: 0.000006 loss: 0.5140 (0.5205) time: 0.4353 data: 0.0001 max mem: 52150 +Epoch: [20] [2820/7269] eta: 0:31:54 lr: 0.000006 loss: 0.5047 (0.5204) time: 0.4816 data: 0.0001 max mem: 52150 +Epoch: [20] [2830/7269] eta: 0:31:50 lr: 0.000006 loss: 0.5047 (0.5204) time: 0.4646 data: 0.0002 max mem: 52150 +Epoch: [20] [2840/7269] eta: 0:31:45 lr: 0.000006 loss: 0.5366 (0.5205) time: 0.4180 data: 0.0002 max mem: 52150 +Epoch: [20] [2850/7269] eta: 0:31:40 lr: 0.000006 loss: 0.5691 (0.5206) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [20] [2860/7269] eta: 0:31:36 lr: 0.000006 loss: 0.5356 (0.5206) time: 0.4175 data: 0.0002 max mem: 52150 +Epoch: [20] [2870/7269] eta: 0:31:32 lr: 0.000006 loss: 0.5367 (0.5207) time: 0.4641 data: 0.0001 max mem: 52150 +Epoch: [20] [2880/7269] eta: 0:31:27 lr: 0.000006 loss: 0.5459 (0.5207) time: 0.4182 data: 0.0001 max mem: 52150 +Epoch: [20] [2890/7269] eta: 0:31:22 lr: 0.000006 loss: 0.5508 (0.5207) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [20] [2900/7269] eta: 0:31:18 lr: 0.000006 loss: 0.5370 (0.5206) time: 0.4185 data: 0.0001 max mem: 52150 +Epoch: [20] [2910/7269] eta: 0:31:13 lr: 0.000006 loss: 0.5164 (0.5207) time: 0.4183 data: 0.0001 max mem: 52150 +Epoch: [20] [2920/7269] eta: 0:31:08 lr: 0.000006 loss: 0.5680 (0.5208) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [20] [2930/7269] eta: 0:31:05 lr: 0.000006 loss: 0.5338 (0.5208) time: 0.4353 data: 0.0001 max mem: 52150 +Epoch: [20] [2940/7269] eta: 0:31:01 lr: 0.000006 loss: 0.4942 (0.5207) time: 0.4954 data: 0.0002 max mem: 52150 +Epoch: [20] [2950/7269] eta: 0:30:56 lr: 0.000006 loss: 0.4628 (0.5205) time: 0.4323 data: 0.0002 max mem: 52150 +Epoch: [20] [2960/7269] eta: 0:30:51 lr: 0.000006 loss: 0.4870 (0.5206) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [20] [2970/7269] eta: 0:30:46 lr: 0.000006 loss: 0.5054 (0.5206) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [20] [2980/7269] eta: 0:30:41 lr: 0.000006 loss: 0.5422 (0.5206) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [20] [2990/7269] eta: 0:30:40 lr: 0.000006 loss: 0.5441 (0.5205) time: 0.5407 data: 0.0002 max mem: 52150 +Epoch: [20] [3000/7269] eta: 0:30:35 lr: 0.000006 loss: 0.4991 (0.5205) time: 0.5407 data: 0.0001 max mem: 52150 +Epoch: [20] [3010/7269] eta: 0:30:30 lr: 0.000006 loss: 0.5318 (0.5205) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [20] [3020/7269] eta: 0:30:27 lr: 0.000006 loss: 0.5318 (0.5206) time: 0.4330 data: 0.0001 max mem: 52150 +Epoch: [20] [3030/7269] eta: 0:30:25 lr: 0.000006 loss: 0.5227 (0.5205) time: 0.5439 data: 0.0002 max mem: 52150 +Epoch: [20] [3040/7269] eta: 0:30:20 lr: 0.000006 loss: 0.5179 (0.5205) time: 0.4834 data: 0.0002 max mem: 52150 +Epoch: [20] [3050/7269] eta: 0:30:16 lr: 0.000006 loss: 0.5535 (0.5206) time: 0.4169 data: 0.0001 max mem: 52150 +Epoch: [20] [3060/7269] eta: 0:30:11 lr: 0.000006 loss: 0.5862 (0.5208) time: 0.4166 data: 0.0001 max mem: 52150 +Epoch: [20] [3070/7269] eta: 0:30:11 lr: 0.000006 loss: 0.5261 (0.5206) time: 0.5580 data: 0.0001 max mem: 52150 +Epoch: [20] [3080/7269] eta: 0:30:09 lr: 0.000006 loss: 0.5261 (0.5208) time: 0.6795 data: 0.0001 max mem: 52150 +Epoch: [20] [3090/7269] eta: 0:30:04 lr: 0.000006 loss: 0.5479 (0.5208) time: 0.4937 data: 0.0002 max mem: 52150 +Epoch: [20] [3100/7269] eta: 0:29:59 lr: 0.000006 loss: 0.5437 (0.5208) time: 0.3737 data: 0.0002 max mem: 52150 +Epoch: [20] [3110/7269] eta: 0:29:55 lr: 0.000006 loss: 0.5419 (0.5208) time: 0.4199 data: 0.0001 max mem: 52150 +Epoch: [20] [3120/7269] eta: 0:29:50 lr: 0.000006 loss: 0.5332 (0.5209) time: 0.4179 data: 0.0001 max mem: 52150 +Epoch: [20] [3130/7269] eta: 0:29:46 lr: 0.000006 loss: 0.5220 (0.5209) time: 0.4191 data: 0.0001 max mem: 52150 +Epoch: [20] [3140/7269] eta: 0:29:41 lr: 0.000006 loss: 0.5288 (0.5210) time: 0.4193 data: 0.0001 max mem: 52150 +Epoch: [20] [3150/7269] eta: 0:29:36 lr: 0.000006 loss: 0.5009 (0.5209) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [20] [3160/7269] eta: 0:29:30 lr: 0.000006 loss: 0.5283 (0.5210) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [20] [3170/7269] eta: 0:29:27 lr: 0.000006 loss: 0.5452 (0.5211) time: 0.4378 data: 0.0002 max mem: 52150 +Epoch: [20] [3180/7269] eta: 0:29:22 lr: 0.000006 loss: 0.5661 (0.5211) time: 0.4369 data: 0.0002 max mem: 52150 +Epoch: [20] [3190/7269] eta: 0:29:17 lr: 0.000006 loss: 0.5327 (0.5211) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [20] [3200/7269] eta: 0:29:13 lr: 0.000006 loss: 0.5020 (0.5211) time: 0.4343 data: 0.0002 max mem: 52150 +Epoch: [20] [3210/7269] eta: 0:29:08 lr: 0.000006 loss: 0.5029 (0.5210) time: 0.4349 data: 0.0002 max mem: 52150 +Epoch: [20] [3220/7269] eta: 0:29:03 lr: 0.000006 loss: 0.5029 (0.5210) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [20] [3230/7269] eta: 0:29:00 lr: 0.000006 loss: 0.5702 (0.5212) time: 0.4353 data: 0.0002 max mem: 52150 +Epoch: [20] [3240/7269] eta: 0:28:56 lr: 0.000006 loss: 0.5708 (0.5212) time: 0.4820 data: 0.0002 max mem: 52150 +Epoch: [20] [3250/7269] eta: 0:28:51 lr: 0.000006 loss: 0.5564 (0.5214) time: 0.4197 data: 0.0002 max mem: 52150 +Epoch: [20] [3260/7269] eta: 0:28:46 lr: 0.000006 loss: 0.5449 (0.5213) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [20] [3270/7269] eta: 0:28:43 lr: 0.000006 loss: 0.5357 (0.5213) time: 0.4355 data: 0.0002 max mem: 52150 +Epoch: [20] [3280/7269] eta: 0:28:39 lr: 0.000006 loss: 0.5466 (0.5214) time: 0.5009 data: 0.0001 max mem: 52150 +Epoch: [20] [3290/7269] eta: 0:28:35 lr: 0.000006 loss: 0.5441 (0.5215) time: 0.4829 data: 0.0001 max mem: 52150 +Epoch: [20] [3300/7269] eta: 0:28:30 lr: 0.000006 loss: 0.5390 (0.5216) time: 0.4181 data: 0.0001 max mem: 52150 +Epoch: [20] [3310/7269] eta: 0:28:25 lr: 0.000006 loss: 0.5585 (0.5216) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [20] [3320/7269] eta: 0:28:23 lr: 0.000006 loss: 0.5548 (0.5216) time: 0.4814 data: 0.0002 max mem: 52150 +Epoch: [20] [3330/7269] eta: 0:28:19 lr: 0.000006 loss: 0.5503 (0.5217) time: 0.5264 data: 0.0002 max mem: 52150 +Epoch: [20] [3340/7269] eta: 0:28:15 lr: 0.000006 loss: 0.5489 (0.5217) time: 0.4678 data: 0.0002 max mem: 52150 +Epoch: [20] [3350/7269] eta: 0:28:11 lr: 0.000006 loss: 0.5349 (0.5218) time: 0.4828 data: 0.0001 max mem: 52150 +Epoch: [20] [3360/7269] eta: 0:28:07 lr: 0.000006 loss: 0.5153 (0.5217) time: 0.4770 data: 0.0001 max mem: 52150 +Epoch: [20] [3370/7269] eta: 0:28:02 lr: 0.000006 loss: 0.5153 (0.5217) time: 0.4171 data: 0.0002 max mem: 52150 +Epoch: [20] [3380/7269] eta: 0:27:57 lr: 0.000006 loss: 0.5611 (0.5218) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [20] [3390/7269] eta: 0:27:52 lr: 0.000006 loss: 0.5192 (0.5218) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [20] [3400/7269] eta: 0:27:49 lr: 0.000006 loss: 0.5346 (0.5219) time: 0.4350 data: 0.0002 max mem: 52150 +Epoch: [20] [3410/7269] eta: 0:27:44 lr: 0.000006 loss: 0.5646 (0.5219) time: 0.4348 data: 0.0002 max mem: 52150 +Epoch: [20] [3420/7269] eta: 0:27:40 lr: 0.000006 loss: 0.5196 (0.5219) time: 0.4167 data: 0.0002 max mem: 52150 +Epoch: [20] [3430/7269] eta: 0:27:35 lr: 0.000006 loss: 0.5077 (0.5218) time: 0.4170 data: 0.0002 max mem: 52150 +Epoch: [20] [3440/7269] eta: 0:27:30 lr: 0.000006 loss: 0.5018 (0.5217) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [20] [3450/7269] eta: 0:27:25 lr: 0.000006 loss: 0.5036 (0.5217) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [20] [3460/7269] eta: 0:27:20 lr: 0.000006 loss: 0.5115 (0.5217) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [20] [3470/7269] eta: 0:27:15 lr: 0.000006 loss: 0.5398 (0.5217) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [20] [3480/7269] eta: 0:27:11 lr: 0.000006 loss: 0.5274 (0.5217) time: 0.4310 data: 0.0002 max mem: 52150 +Epoch: [20] [3490/7269] eta: 0:27:09 lr: 0.000006 loss: 0.5274 (0.5217) time: 0.5384 data: 0.0002 max mem: 52150 +Epoch: [20] [3500/7269] eta: 0:27:04 lr: 0.000006 loss: 0.5221 (0.5216) time: 0.4798 data: 0.0002 max mem: 52150 +Epoch: [20] [3510/7269] eta: 0:26:59 lr: 0.000006 loss: 0.5242 (0.5217) time: 0.3730 data: 0.0002 max mem: 52150 +Epoch: [20] [3520/7269] eta: 0:26:54 lr: 0.000006 loss: 0.5168 (0.5217) time: 0.3729 data: 0.0002 max mem: 52150 +Epoch: [20] [3530/7269] eta: 0:26:50 lr: 0.000006 loss: 0.5168 (0.5217) time: 0.4359 data: 0.0002 max mem: 52150 +Epoch: [20] [3540/7269] eta: 0:26:47 lr: 0.000006 loss: 0.5066 (0.5216) time: 0.4956 data: 0.0002 max mem: 52150 +Epoch: [20] [3550/7269] eta: 0:26:44 lr: 0.000006 loss: 0.4866 (0.5215) time: 0.5408 data: 0.0001 max mem: 52150 +Epoch: [20] [3560/7269] eta: 0:26:39 lr: 0.000006 loss: 0.5079 (0.5215) time: 0.4806 data: 0.0002 max mem: 52150 +Epoch: [20] [3570/7269] eta: 0:26:36 lr: 0.000006 loss: 0.5270 (0.5215) time: 0.4349 data: 0.0002 max mem: 52150 +Epoch: [20] [3580/7269] eta: 0:26:32 lr: 0.000006 loss: 0.5150 (0.5215) time: 0.4949 data: 0.0002 max mem: 52150 +Epoch: [20] [3590/7269] eta: 0:26:28 lr: 0.000006 loss: 0.5100 (0.5214) time: 0.4776 data: 0.0002 max mem: 52150 +Epoch: [20] [3600/7269] eta: 0:26:23 lr: 0.000006 loss: 0.5462 (0.5216) time: 0.4176 data: 0.0002 max mem: 52150 +Epoch: [20] [3610/7269] eta: 0:26:18 lr: 0.000006 loss: 0.5599 (0.5216) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [20] [3620/7269] eta: 0:26:13 lr: 0.000006 loss: 0.5283 (0.5215) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [20] [3630/7269] eta: 0:26:08 lr: 0.000006 loss: 0.4546 (0.5215) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [20] [3640/7269] eta: 0:26:05 lr: 0.000006 loss: 0.5212 (0.5215) time: 0.4346 data: 0.0002 max mem: 52150 +Epoch: [20] [3650/7269] eta: 0:26:01 lr: 0.000006 loss: 0.5480 (0.5216) time: 0.4805 data: 0.0002 max mem: 52150 +Epoch: [20] [3660/7269] eta: 0:25:56 lr: 0.000006 loss: 0.5372 (0.5216) time: 0.4179 data: 0.0001 max mem: 52150 +Epoch: [20] [3670/7269] eta: 0:25:51 lr: 0.000006 loss: 0.5139 (0.5216) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [20] [3680/7269] eta: 0:25:46 lr: 0.000006 loss: 0.4968 (0.5215) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [20] [3690/7269] eta: 0:25:43 lr: 0.000006 loss: 0.4615 (0.5214) time: 0.4793 data: 0.0002 max mem: 52150 +Epoch: [20] [3700/7269] eta: 0:25:38 lr: 0.000006 loss: 0.4720 (0.5213) time: 0.4793 data: 0.0002 max mem: 52150 +Epoch: [20] [3710/7269] eta: 0:25:35 lr: 0.000006 loss: 0.5274 (0.5213) time: 0.4625 data: 0.0002 max mem: 52150 +Epoch: [20] [3720/7269] eta: 0:25:30 lr: 0.000006 loss: 0.5398 (0.5214) time: 0.4628 data: 0.0002 max mem: 52150 +Epoch: [20] [3730/7269] eta: 0:25:25 lr: 0.000006 loss: 0.5450 (0.5215) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [20] [3740/7269] eta: 0:25:20 lr: 0.000006 loss: 0.5364 (0.5214) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [20] [3750/7269] eta: 0:25:16 lr: 0.000006 loss: 0.5378 (0.5215) time: 0.3731 data: 0.0002 max mem: 52150 +Epoch: [20] [3760/7269] eta: 0:25:11 lr: 0.000006 loss: 0.5325 (0.5214) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [20] [3770/7269] eta: 0:25:06 lr: 0.000006 loss: 0.5160 (0.5215) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [20] [3780/7269] eta: 0:25:02 lr: 0.000006 loss: 0.5160 (0.5214) time: 0.4355 data: 0.0002 max mem: 52150 +Epoch: [20] [3790/7269] eta: 0:24:58 lr: 0.000006 loss: 0.5091 (0.5214) time: 0.4812 data: 0.0002 max mem: 52150 +Epoch: [20] [3800/7269] eta: 0:24:57 lr: 0.000006 loss: 0.5452 (0.5215) time: 0.6219 data: 0.0002 max mem: 52150 +Epoch: [20] [3810/7269] eta: 0:24:53 lr: 0.000006 loss: 0.5466 (0.5215) time: 0.6345 data: 0.0002 max mem: 52150 +Epoch: [20] [3820/7269] eta: 0:24:48 lr: 0.000006 loss: 0.5281 (0.5214) time: 0.4305 data: 0.0002 max mem: 52150 +Epoch: [20] [3830/7269] eta: 0:24:44 lr: 0.000006 loss: 0.5281 (0.5215) time: 0.3711 data: 0.0002 max mem: 52150 +Epoch: [20] [3840/7269] eta: 0:24:40 lr: 0.000006 loss: 0.5204 (0.5214) time: 0.4316 data: 0.0002 max mem: 52150 +Epoch: [20] [3850/7269] eta: 0:24:35 lr: 0.000006 loss: 0.4863 (0.5214) time: 0.4317 data: 0.0002 max mem: 52150 +Epoch: [20] [3860/7269] eta: 0:24:31 lr: 0.000006 loss: 0.5350 (0.5215) time: 0.4325 data: 0.0002 max mem: 52150 +Epoch: [20] [3870/7269] eta: 0:24:26 lr: 0.000006 loss: 0.5174 (0.5214) time: 0.4327 data: 0.0002 max mem: 52150 +Epoch: [20] [3880/7269] eta: 0:24:21 lr: 0.000006 loss: 0.5174 (0.5214) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [20] [3890/7269] eta: 0:24:17 lr: 0.000006 loss: 0.5244 (0.5214) time: 0.4166 data: 0.0002 max mem: 52150 +Epoch: [20] [3900/7269] eta: 0:24:13 lr: 0.000006 loss: 0.5060 (0.5213) time: 0.4168 data: 0.0002 max mem: 52150 +Epoch: [20] [3910/7269] eta: 0:24:09 lr: 0.000006 loss: 0.5275 (0.5214) time: 0.4624 data: 0.0002 max mem: 52150 +Epoch: [20] [3920/7269] eta: 0:24:04 lr: 0.000006 loss: 0.5463 (0.5215) time: 0.4626 data: 0.0002 max mem: 52150 +Epoch: [20] [3930/7269] eta: 0:24:01 lr: 0.000006 loss: 0.5490 (0.5216) time: 0.4340 data: 0.0002 max mem: 52150 +Epoch: [20] [3940/7269] eta: 0:23:56 lr: 0.000006 loss: 0.5490 (0.5217) time: 0.4335 data: 0.0002 max mem: 52150 +Epoch: [20] [3950/7269] eta: 0:23:51 lr: 0.000006 loss: 0.5676 (0.5218) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [20] [3960/7269] eta: 0:23:47 lr: 0.000006 loss: 0.5452 (0.5218) time: 0.4204 data: 0.0002 max mem: 52150 +Epoch: [20] [3970/7269] eta: 0:23:42 lr: 0.000006 loss: 0.5084 (0.5218) time: 0.4199 data: 0.0002 max mem: 52150 +Epoch: [20] [3980/7269] eta: 0:23:37 lr: 0.000006 loss: 0.5084 (0.5217) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [20] [3990/7269] eta: 0:23:33 lr: 0.000006 loss: 0.5137 (0.5216) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [20] [4000/7269] eta: 0:23:28 lr: 0.000006 loss: 0.5047 (0.5216) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [20] [4010/7269] eta: 0:23:24 lr: 0.000006 loss: 0.5206 (0.5216) time: 0.4339 data: 0.0001 max mem: 52150 +Epoch: [20] [4020/7269] eta: 0:23:19 lr: 0.000006 loss: 0.4983 (0.5215) time: 0.4342 data: 0.0002 max mem: 52150 +Epoch: [20] [4030/7269] eta: 0:23:15 lr: 0.000006 loss: 0.5056 (0.5215) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [20] [4040/7269] eta: 0:23:10 lr: 0.000006 loss: 0.5377 (0.5216) time: 0.4192 data: 0.0002 max mem: 52150 +Epoch: [20] [4050/7269] eta: 0:23:06 lr: 0.000006 loss: 0.5414 (0.5215) time: 0.4194 data: 0.0002 max mem: 52150 +Epoch: [20] [4060/7269] eta: 0:23:01 lr: 0.000006 loss: 0.4944 (0.5214) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [20] [4070/7269] eta: 0:22:56 lr: 0.000006 loss: 0.5034 (0.5214) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [20] [4080/7269] eta: 0:22:52 lr: 0.000006 loss: 0.5094 (0.5214) time: 0.4352 data: 0.0002 max mem: 52150 +Epoch: [20] [4090/7269] eta: 0:22:48 lr: 0.000006 loss: 0.5216 (0.5214) time: 0.4349 data: 0.0002 max mem: 52150 +Epoch: [20] [4100/7269] eta: 0:22:44 lr: 0.000006 loss: 0.5145 (0.5214) time: 0.4172 data: 0.0002 max mem: 52150 +Epoch: [20] [4110/7269] eta: 0:22:40 lr: 0.000006 loss: 0.5584 (0.5215) time: 0.4788 data: 0.0002 max mem: 52150 +Epoch: [20] [4120/7269] eta: 0:22:38 lr: 0.000006 loss: 0.5775 (0.5216) time: 0.6047 data: 0.0002 max mem: 52150 +Epoch: [20] [4130/7269] eta: 0:22:33 lr: 0.000006 loss: 0.5418 (0.5216) time: 0.5436 data: 0.0001 max mem: 52150 +Epoch: [20] [4140/7269] eta: 0:22:28 lr: 0.000006 loss: 0.5107 (0.5217) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [20] [4150/7269] eta: 0:22:24 lr: 0.000006 loss: 0.5107 (0.5216) time: 0.4336 data: 0.0002 max mem: 52150 +Epoch: [20] [4160/7269] eta: 0:22:20 lr: 0.000006 loss: 0.4923 (0.5216) time: 0.4963 data: 0.0001 max mem: 52150 +Epoch: [20] [4170/7269] eta: 0:22:16 lr: 0.000006 loss: 0.5161 (0.5216) time: 0.4800 data: 0.0003 max mem: 52150 +Epoch: [20] [4180/7269] eta: 0:22:12 lr: 0.000006 loss: 0.5418 (0.5217) time: 0.4172 data: 0.0003 max mem: 52150 +Epoch: [20] [4190/7269] eta: 0:22:07 lr: 0.000006 loss: 0.5410 (0.5217) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [20] [4200/7269] eta: 0:22:03 lr: 0.000006 loss: 0.5129 (0.5216) time: 0.4320 data: 0.0002 max mem: 52150 +Epoch: [20] [4210/7269] eta: 0:22:00 lr: 0.000006 loss: 0.5121 (0.5216) time: 0.5425 data: 0.0002 max mem: 52150 +Epoch: [20] [4220/7269] eta: 0:21:56 lr: 0.000006 loss: 0.5208 (0.5216) time: 0.5280 data: 0.0002 max mem: 52150 +Epoch: [20] [4230/7269] eta: 0:21:52 lr: 0.000006 loss: 0.5430 (0.5216) time: 0.4635 data: 0.0002 max mem: 52150 +Epoch: [20] [4240/7269] eta: 0:21:48 lr: 0.000006 loss: 0.5348 (0.5216) time: 0.4802 data: 0.0002 max mem: 52150 +Epoch: [20] [4250/7269] eta: 0:21:44 lr: 0.000006 loss: 0.5348 (0.5216) time: 0.4975 data: 0.0002 max mem: 52150 +Epoch: [20] [4260/7269] eta: 0:21:40 lr: 0.000006 loss: 0.5148 (0.5216) time: 0.4798 data: 0.0002 max mem: 52150 +Epoch: [20] [4270/7269] eta: 0:21:35 lr: 0.000006 loss: 0.5335 (0.5216) time: 0.4169 data: 0.0002 max mem: 52150 +Epoch: [20] [4280/7269] eta: 0:21:30 lr: 0.000006 loss: 0.5362 (0.5215) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [20] [4290/7269] eta: 0:21:27 lr: 0.000006 loss: 0.5322 (0.5215) time: 0.4354 data: 0.0002 max mem: 52150 +Epoch: [20] [4300/7269] eta: 0:21:23 lr: 0.000006 loss: 0.5394 (0.5215) time: 0.5552 data: 0.0001 max mem: 52150 +Epoch: [20] [4310/7269] eta: 0:21:19 lr: 0.000006 loss: 0.5149 (0.5216) time: 0.5387 data: 0.0001 max mem: 52150 +Epoch: [20] [4320/7269] eta: 0:21:15 lr: 0.000006 loss: 0.5149 (0.5216) time: 0.4192 data: 0.0002 max mem: 52150 +Epoch: [20] [4330/7269] eta: 0:21:10 lr: 0.000006 loss: 0.5064 (0.5216) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [20] [4340/7269] eta: 0:21:06 lr: 0.000006 loss: 0.4913 (0.5215) time: 0.4255 data: 0.0002 max mem: 52150 +Epoch: [20] [4350/7269] eta: 0:21:01 lr: 0.000006 loss: 0.4882 (0.5215) time: 0.4253 data: 0.0002 max mem: 52150 +Epoch: [20] [4360/7269] eta: 0:20:56 lr: 0.000006 loss: 0.4978 (0.5214) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [20] [4370/7269] eta: 0:20:52 lr: 0.000006 loss: 0.5183 (0.5214) time: 0.4183 data: 0.0002 max mem: 52150 +Epoch: [20] [4380/7269] eta: 0:20:48 lr: 0.000006 loss: 0.5446 (0.5215) time: 0.4644 data: 0.0002 max mem: 52150 +Epoch: [20] [4390/7269] eta: 0:20:44 lr: 0.000006 loss: 0.5325 (0.5214) time: 0.4178 data: 0.0002 max mem: 52150 +Epoch: [20] [4400/7269] eta: 0:20:39 lr: 0.000006 loss: 0.4933 (0.5214) time: 0.3727 data: 0.0001 max mem: 52150 +Epoch: [20] [4410/7269] eta: 0:20:34 lr: 0.000006 loss: 0.5284 (0.5214) time: 0.3727 data: 0.0001 max mem: 52150 +Epoch: [20] [4420/7269] eta: 0:20:29 lr: 0.000006 loss: 0.5284 (0.5215) time: 0.3712 data: 0.0001 max mem: 52150 +Epoch: [20] [4430/7269] eta: 0:20:25 lr: 0.000006 loss: 0.5007 (0.5214) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [20] [4440/7269] eta: 0:20:21 lr: 0.000006 loss: 0.4882 (0.5214) time: 0.4181 data: 0.0001 max mem: 52150 +Epoch: [20] [4450/7269] eta: 0:20:16 lr: 0.000006 loss: 0.4991 (0.5213) time: 0.4641 data: 0.0001 max mem: 52150 +Epoch: [20] [4460/7269] eta: 0:20:12 lr: 0.000006 loss: 0.5059 (0.5213) time: 0.4178 data: 0.0002 max mem: 52150 +Epoch: [20] [4470/7269] eta: 0:20:08 lr: 0.000006 loss: 0.5335 (0.5213) time: 0.4324 data: 0.0002 max mem: 52150 +Epoch: [20] [4480/7269] eta: 0:20:03 lr: 0.000006 loss: 0.5183 (0.5212) time: 0.4327 data: 0.0002 max mem: 52150 +Epoch: [20] [4490/7269] eta: 0:19:58 lr: 0.000006 loss: 0.4948 (0.5212) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [20] [4500/7269] eta: 0:19:54 lr: 0.000006 loss: 0.5076 (0.5212) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [20] [4510/7269] eta: 0:19:49 lr: 0.000006 loss: 0.4959 (0.5211) time: 0.3722 data: 0.0001 max mem: 52150 +Epoch: [20] [4520/7269] eta: 0:19:44 lr: 0.000006 loss: 0.5232 (0.5212) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [20] [4530/7269] eta: 0:19:40 lr: 0.000006 loss: 0.5510 (0.5212) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [20] [4540/7269] eta: 0:19:36 lr: 0.000006 loss: 0.5157 (0.5212) time: 0.4198 data: 0.0002 max mem: 52150 +Epoch: [20] [4550/7269] eta: 0:19:31 lr: 0.000006 loss: 0.4993 (0.5211) time: 0.4196 data: 0.0002 max mem: 52150 +Epoch: [20] [4560/7269] eta: 0:19:26 lr: 0.000006 loss: 0.5200 (0.5212) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [20] [4570/7269] eta: 0:19:22 lr: 0.000006 loss: 0.5319 (0.5212) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [20] [4580/7269] eta: 0:19:18 lr: 0.000006 loss: 0.5219 (0.5211) time: 0.4797 data: 0.0001 max mem: 52150 +Epoch: [20] [4590/7269] eta: 0:19:14 lr: 0.000006 loss: 0.4984 (0.5211) time: 0.5247 data: 0.0002 max mem: 52150 +Epoch: [20] [4600/7269] eta: 0:19:10 lr: 0.000006 loss: 0.4984 (0.5211) time: 0.4802 data: 0.0002 max mem: 52150 +Epoch: [20] [4610/7269] eta: 0:19:06 lr: 0.000006 loss: 0.5186 (0.5210) time: 0.4361 data: 0.0002 max mem: 52150 +Epoch: [20] [4620/7269] eta: 0:19:02 lr: 0.000006 loss: 0.5143 (0.5210) time: 0.4199 data: 0.0001 max mem: 52150 +Epoch: [20] [4630/7269] eta: 0:18:58 lr: 0.000006 loss: 0.5211 (0.5210) time: 0.5461 data: 0.0001 max mem: 52150 +Epoch: [20] [4640/7269] eta: 0:18:54 lr: 0.000006 loss: 0.5431 (0.5210) time: 0.4987 data: 0.0002 max mem: 52150 +Epoch: [20] [4650/7269] eta: 0:18:49 lr: 0.000006 loss: 0.5609 (0.5211) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [20] [4660/7269] eta: 0:18:44 lr: 0.000006 loss: 0.5675 (0.5211) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [20] [4670/7269] eta: 0:18:40 lr: 0.000006 loss: 0.5536 (0.5212) time: 0.4314 data: 0.0002 max mem: 52150 +Epoch: [20] [4680/7269] eta: 0:18:36 lr: 0.000006 loss: 0.5205 (0.5211) time: 0.4776 data: 0.0001 max mem: 52150 +Epoch: [20] [4690/7269] eta: 0:18:32 lr: 0.000006 loss: 0.5266 (0.5212) time: 0.4179 data: 0.0001 max mem: 52150 +Epoch: [20] [4700/7269] eta: 0:18:28 lr: 0.000006 loss: 0.5481 (0.5212) time: 0.4814 data: 0.0001 max mem: 52150 +Epoch: [20] [4710/7269] eta: 0:18:24 lr: 0.000006 loss: 0.5384 (0.5211) time: 0.5440 data: 0.0002 max mem: 52150 +Epoch: [20] [4720/7269] eta: 0:18:20 lr: 0.000006 loss: 0.5415 (0.5212) time: 0.4343 data: 0.0002 max mem: 52150 +Epoch: [20] [4730/7269] eta: 0:18:16 lr: 0.000006 loss: 0.5443 (0.5212) time: 0.4656 data: 0.0001 max mem: 52150 +Epoch: [20] [4740/7269] eta: 0:18:12 lr: 0.000006 loss: 0.5443 (0.5212) time: 0.5117 data: 0.0001 max mem: 52150 +Epoch: [20] [4750/7269] eta: 0:18:07 lr: 0.000006 loss: 0.5443 (0.5213) time: 0.4174 data: 0.0001 max mem: 52150 +Epoch: [20] [4760/7269] eta: 0:18:04 lr: 0.000006 loss: 0.5441 (0.5213) time: 0.4961 data: 0.0001 max mem: 52150 +Epoch: [20] [4770/7269] eta: 0:17:59 lr: 0.000006 loss: 0.5526 (0.5213) time: 0.4968 data: 0.0001 max mem: 52150 +Epoch: [20] [4780/7269] eta: 0:17:55 lr: 0.000006 loss: 0.5705 (0.5214) time: 0.4352 data: 0.0001 max mem: 52150 +Epoch: [20] [4790/7269] eta: 0:17:51 lr: 0.000006 loss: 0.5531 (0.5214) time: 0.4354 data: 0.0001 max mem: 52150 +Epoch: [20] [4800/7269] eta: 0:17:47 lr: 0.000006 loss: 0.5645 (0.5215) time: 0.4349 data: 0.0002 max mem: 52150 +Epoch: [20] [4810/7269] eta: 0:17:42 lr: 0.000006 loss: 0.5465 (0.5215) time: 0.4345 data: 0.0002 max mem: 52150 +Epoch: [20] [4820/7269] eta: 0:17:38 lr: 0.000006 loss: 0.5055 (0.5214) time: 0.4330 data: 0.0002 max mem: 52150 +Epoch: [20] [4830/7269] eta: 0:17:33 lr: 0.000006 loss: 0.5210 (0.5215) time: 0.4328 data: 0.0002 max mem: 52150 +Epoch: [20] [4840/7269] eta: 0:17:29 lr: 0.000006 loss: 0.5188 (0.5214) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [20] [4850/7269] eta: 0:17:24 lr: 0.000006 loss: 0.5188 (0.5215) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [20] [4860/7269] eta: 0:17:20 lr: 0.000006 loss: 0.5310 (0.5214) time: 0.4309 data: 0.0002 max mem: 52150 +Epoch: [20] [4870/7269] eta: 0:17:15 lr: 0.000006 loss: 0.5155 (0.5214) time: 0.4309 data: 0.0001 max mem: 52150 +Epoch: [20] [4880/7269] eta: 0:17:11 lr: 0.000006 loss: 0.4954 (0.5214) time: 0.4320 data: 0.0001 max mem: 52150 +Epoch: [20] [4890/7269] eta: 0:17:07 lr: 0.000006 loss: 0.5000 (0.5213) time: 0.4322 data: 0.0001 max mem: 52150 +Epoch: [20] [4900/7269] eta: 0:17:02 lr: 0.000006 loss: 0.4930 (0.5213) time: 0.3721 data: 0.0001 max mem: 52150 +Epoch: [20] [4910/7269] eta: 0:16:58 lr: 0.000006 loss: 0.4813 (0.5213) time: 0.4177 data: 0.0002 max mem: 52150 +Epoch: [20] [4920/7269] eta: 0:16:53 lr: 0.000006 loss: 0.5347 (0.5212) time: 0.4173 data: 0.0002 max mem: 52150 +Epoch: [20] [4930/7269] eta: 0:16:50 lr: 0.000006 loss: 0.5347 (0.5213) time: 0.4787 data: 0.0001 max mem: 52150 +Epoch: [20] [4940/7269] eta: 0:16:45 lr: 0.000006 loss: 0.5699 (0.5214) time: 0.4790 data: 0.0001 max mem: 52150 +Epoch: [20] [4950/7269] eta: 0:16:41 lr: 0.000006 loss: 0.5767 (0.5215) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [20] [4960/7269] eta: 0:16:36 lr: 0.000006 loss: 0.5337 (0.5215) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [20] [4970/7269] eta: 0:16:32 lr: 0.000006 loss: 0.5067 (0.5216) time: 0.4169 data: 0.0001 max mem: 52150 +Epoch: [20] [4980/7269] eta: 0:16:28 lr: 0.000006 loss: 0.5390 (0.5216) time: 0.4627 data: 0.0001 max mem: 52150 +Epoch: [20] [4990/7269] eta: 0:16:23 lr: 0.000006 loss: 0.5390 (0.5215) time: 0.4172 data: 0.0001 max mem: 52150 +Epoch: [20] [5000/7269] eta: 0:16:19 lr: 0.000006 loss: 0.4870 (0.5215) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [20] [5010/7269] eta: 0:16:15 lr: 0.000006 loss: 0.4881 (0.5215) time: 0.5168 data: 0.0001 max mem: 52150 +Epoch: [20] [5020/7269] eta: 0:16:11 lr: 0.000006 loss: 0.5528 (0.5215) time: 0.5157 data: 0.0002 max mem: 52150 +Epoch: [20] [5030/7269] eta: 0:16:06 lr: 0.000006 loss: 0.5676 (0.5215) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [20] [5040/7269] eta: 0:16:02 lr: 0.000006 loss: 0.5538 (0.5215) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [20] [5050/7269] eta: 0:15:57 lr: 0.000006 loss: 0.5453 (0.5216) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [20] [5060/7269] eta: 0:15:52 lr: 0.000006 loss: 0.5266 (0.5216) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [20] [5070/7269] eta: 0:15:48 lr: 0.000006 loss: 0.5127 (0.5216) time: 0.4337 data: 0.0002 max mem: 52150 +Epoch: [20] [5080/7269] eta: 0:15:44 lr: 0.000006 loss: 0.5143 (0.5216) time: 0.4784 data: 0.0001 max mem: 52150 +Epoch: [20] [5090/7269] eta: 0:15:40 lr: 0.000006 loss: 0.5039 (0.5216) time: 0.4166 data: 0.0002 max mem: 52150 +Epoch: [20] [5100/7269] eta: 0:15:35 lr: 0.000006 loss: 0.5304 (0.5216) time: 0.4217 data: 0.0002 max mem: 52150 +Epoch: [20] [5110/7269] eta: 0:15:31 lr: 0.000006 loss: 0.5489 (0.5216) time: 0.4220 data: 0.0002 max mem: 52150 +Epoch: [20] [5120/7269] eta: 0:15:26 lr: 0.000006 loss: 0.5047 (0.5215) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [20] [5130/7269] eta: 0:15:22 lr: 0.000006 loss: 0.5205 (0.5215) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [20] [5140/7269] eta: 0:15:17 lr: 0.000006 loss: 0.5205 (0.5215) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [20] [5150/7269] eta: 0:15:13 lr: 0.000006 loss: 0.5165 (0.5214) time: 0.4349 data: 0.0002 max mem: 52150 +Epoch: [20] [5160/7269] eta: 0:15:09 lr: 0.000006 loss: 0.5165 (0.5214) time: 0.4346 data: 0.0002 max mem: 52150 +Epoch: [20] [5170/7269] eta: 0:15:04 lr: 0.000006 loss: 0.5245 (0.5214) time: 0.4182 data: 0.0002 max mem: 52150 +Epoch: [20] [5180/7269] eta: 0:15:00 lr: 0.000006 loss: 0.5305 (0.5214) time: 0.4183 data: 0.0002 max mem: 52150 +Epoch: [20] [5190/7269] eta: 0:14:56 lr: 0.000006 loss: 0.5305 (0.5214) time: 0.5095 data: 0.0002 max mem: 52150 +Epoch: [20] [5200/7269] eta: 0:14:52 lr: 0.000006 loss: 0.5269 (0.5214) time: 0.5100 data: 0.0001 max mem: 52150 +Epoch: [20] [5210/7269] eta: 0:14:49 lr: 0.000006 loss: 0.5013 (0.5213) time: 0.5266 data: 0.0002 max mem: 52150 +Epoch: [20] [5220/7269] eta: 0:14:44 lr: 0.000006 loss: 0.5401 (0.5214) time: 0.5868 data: 0.0002 max mem: 52150 +Epoch: [20] [5230/7269] eta: 0:14:40 lr: 0.000006 loss: 0.5499 (0.5215) time: 0.4784 data: 0.0001 max mem: 52150 +Epoch: [20] [5240/7269] eta: 0:14:36 lr: 0.000006 loss: 0.5506 (0.5215) time: 0.4172 data: 0.0001 max mem: 52150 +Epoch: [20] [5250/7269] eta: 0:14:31 lr: 0.000006 loss: 0.5490 (0.5215) time: 0.3710 data: 0.0001 max mem: 52150 +Epoch: [20] [5260/7269] eta: 0:14:27 lr: 0.000006 loss: 0.5091 (0.5214) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [20] [5270/7269] eta: 0:14:22 lr: 0.000006 loss: 0.4945 (0.5214) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [20] [5280/7269] eta: 0:14:18 lr: 0.000006 loss: 0.5449 (0.5214) time: 0.4202 data: 0.0002 max mem: 52150 +Epoch: [20] [5290/7269] eta: 0:14:14 lr: 0.000006 loss: 0.5463 (0.5214) time: 0.4792 data: 0.0002 max mem: 52150 +Epoch: [20] [5300/7269] eta: 0:14:10 lr: 0.000006 loss: 0.5029 (0.5214) time: 0.4927 data: 0.0002 max mem: 52150 +Epoch: [20] [5310/7269] eta: 0:14:06 lr: 0.000006 loss: 0.5062 (0.5214) time: 0.5593 data: 0.0002 max mem: 52150 +Epoch: [20] [5320/7269] eta: 0:14:02 lr: 0.000006 loss: 0.5097 (0.5214) time: 0.4981 data: 0.0001 max mem: 52150 +Epoch: [20] [5330/7269] eta: 0:13:57 lr: 0.000006 loss: 0.5580 (0.5214) time: 0.3725 data: 0.0001 max mem: 52150 +Epoch: [20] [5340/7269] eta: 0:13:53 lr: 0.000006 loss: 0.5391 (0.5214) time: 0.4308 data: 0.0001 max mem: 52150 +Epoch: [20] [5350/7269] eta: 0:13:49 lr: 0.000006 loss: 0.5118 (0.5214) time: 0.5519 data: 0.0001 max mem: 52150 +Epoch: [20] [5360/7269] eta: 0:13:45 lr: 0.000006 loss: 0.5118 (0.5214) time: 0.5564 data: 0.0001 max mem: 52150 +Epoch: [20] [5370/7269] eta: 0:13:41 lr: 0.000006 loss: 0.5065 (0.5213) time: 0.4353 data: 0.0001 max mem: 52150 +Epoch: [20] [5380/7269] eta: 0:13:36 lr: 0.000006 loss: 0.5186 (0.5213) time: 0.3714 data: 0.0001 max mem: 52150 +Epoch: [20] [5390/7269] eta: 0:13:32 lr: 0.000006 loss: 0.5186 (0.5213) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [20] [5400/7269] eta: 0:13:27 lr: 0.000006 loss: 0.5269 (0.5213) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [20] [5410/7269] eta: 0:13:22 lr: 0.000006 loss: 0.5081 (0.5212) time: 0.3727 data: 0.0002 max mem: 52150 +Epoch: [20] [5420/7269] eta: 0:13:18 lr: 0.000006 loss: 0.5118 (0.5212) time: 0.3725 data: 0.0001 max mem: 52150 +Epoch: [20] [5430/7269] eta: 0:13:14 lr: 0.000006 loss: 0.5255 (0.5212) time: 0.4321 data: 0.0002 max mem: 52150 +Epoch: [20] [5440/7269] eta: 0:13:09 lr: 0.000006 loss: 0.5126 (0.5212) time: 0.4322 data: 0.0002 max mem: 52150 +Epoch: [20] [5450/7269] eta: 0:13:05 lr: 0.000006 loss: 0.5284 (0.5212) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [20] [5460/7269] eta: 0:13:00 lr: 0.000006 loss: 0.5146 (0.5211) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [20] [5470/7269] eta: 0:12:56 lr: 0.000006 loss: 0.4708 (0.5211) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [20] [5480/7269] eta: 0:12:52 lr: 0.000006 loss: 0.5029 (0.5211) time: 0.4226 data: 0.0001 max mem: 52150 +Epoch: [20] [5490/7269] eta: 0:12:47 lr: 0.000006 loss: 0.5055 (0.5211) time: 0.4221 data: 0.0001 max mem: 52150 +Epoch: [20] [5500/7269] eta: 0:12:43 lr: 0.000006 loss: 0.5055 (0.5211) time: 0.3714 data: 0.0001 max mem: 52150 +Epoch: [20] [5510/7269] eta: 0:12:38 lr: 0.000006 loss: 0.5317 (0.5211) time: 0.4176 data: 0.0001 max mem: 52150 +Epoch: [20] [5520/7269] eta: 0:12:34 lr: 0.000006 loss: 0.5288 (0.5210) time: 0.4178 data: 0.0001 max mem: 52150 +Epoch: [20] [5530/7269] eta: 0:12:29 lr: 0.000006 loss: 0.5083 (0.5210) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [20] [5540/7269] eta: 0:12:25 lr: 0.000006 loss: 0.5293 (0.5210) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [20] [5550/7269] eta: 0:12:20 lr: 0.000006 loss: 0.5293 (0.5210) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [20] [5560/7269] eta: 0:12:16 lr: 0.000006 loss: 0.5398 (0.5210) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [20] [5570/7269] eta: 0:12:12 lr: 0.000006 loss: 0.5421 (0.5210) time: 0.4720 data: 0.0001 max mem: 52150 +Epoch: [20] [5580/7269] eta: 0:12:08 lr: 0.000006 loss: 0.5246 (0.5210) time: 0.4725 data: 0.0001 max mem: 52150 +Epoch: [20] [5590/7269] eta: 0:12:03 lr: 0.000006 loss: 0.5182 (0.5210) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [20] [5600/7269] eta: 0:11:59 lr: 0.000006 loss: 0.5178 (0.5210) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [20] [5610/7269] eta: 0:11:54 lr: 0.000006 loss: 0.5244 (0.5210) time: 0.4175 data: 0.0001 max mem: 52150 +Epoch: [20] [5620/7269] eta: 0:11:50 lr: 0.000006 loss: 0.5592 (0.5211) time: 0.4800 data: 0.0001 max mem: 52150 +Epoch: [20] [5630/7269] eta: 0:11:46 lr: 0.000006 loss: 0.5460 (0.5210) time: 0.4340 data: 0.0001 max mem: 52150 +Epoch: [20] [5640/7269] eta: 0:11:41 lr: 0.000006 loss: 0.5429 (0.5211) time: 0.3712 data: 0.0001 max mem: 52150 +Epoch: [20] [5650/7269] eta: 0:11:37 lr: 0.000006 loss: 0.5606 (0.5211) time: 0.4179 data: 0.0001 max mem: 52150 +Epoch: [20] [5660/7269] eta: 0:11:33 lr: 0.000006 loss: 0.5530 (0.5211) time: 0.4177 data: 0.0001 max mem: 52150 +Epoch: [20] [5670/7269] eta: 0:11:28 lr: 0.000006 loss: 0.4991 (0.5211) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [20] [5680/7269] eta: 0:11:24 lr: 0.000006 loss: 0.5096 (0.5211) time: 0.5111 data: 0.0002 max mem: 52150 +Epoch: [20] [5690/7269] eta: 0:11:21 lr: 0.000006 loss: 0.5305 (0.5211) time: 0.6201 data: 0.0002 max mem: 52150 +Epoch: [20] [5700/7269] eta: 0:11:16 lr: 0.000006 loss: 0.5459 (0.5212) time: 0.5253 data: 0.0001 max mem: 52150 +Epoch: [20] [5710/7269] eta: 0:11:12 lr: 0.000006 loss: 0.5365 (0.5212) time: 0.4165 data: 0.0001 max mem: 52150 +Epoch: [20] [5720/7269] eta: 0:11:07 lr: 0.000006 loss: 0.5170 (0.5212) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [20] [5730/7269] eta: 0:11:03 lr: 0.000006 loss: 0.4965 (0.5212) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [20] [5740/7269] eta: 0:10:58 lr: 0.000006 loss: 0.4975 (0.5211) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [20] [5750/7269] eta: 0:10:54 lr: 0.000006 loss: 0.5114 (0.5211) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [20] [5760/7269] eta: 0:10:49 lr: 0.000006 loss: 0.5114 (0.5211) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [20] [5770/7269] eta: 0:10:45 lr: 0.000006 loss: 0.5237 (0.5211) time: 0.4636 data: 0.0001 max mem: 52150 +Epoch: [20] [5780/7269] eta: 0:10:41 lr: 0.000006 loss: 0.5138 (0.5211) time: 0.5090 data: 0.0002 max mem: 52150 +Epoch: [20] [5790/7269] eta: 0:10:37 lr: 0.000006 loss: 0.5270 (0.5211) time: 0.4173 data: 0.0002 max mem: 52150 +Epoch: [20] [5800/7269] eta: 0:10:33 lr: 0.000006 loss: 0.5304 (0.5211) time: 0.4352 data: 0.0002 max mem: 52150 +Epoch: [20] [5810/7269] eta: 0:10:29 lr: 0.000006 loss: 0.4876 (0.5211) time: 0.4980 data: 0.0001 max mem: 52150 +Epoch: [20] [5820/7269] eta: 0:10:25 lr: 0.000006 loss: 0.4876 (0.5210) time: 0.5264 data: 0.0002 max mem: 52150 +Epoch: [20] [5830/7269] eta: 0:10:20 lr: 0.000006 loss: 0.5004 (0.5211) time: 0.4642 data: 0.0002 max mem: 52150 +Epoch: [20] [5840/7269] eta: 0:10:16 lr: 0.000006 loss: 0.5416 (0.5211) time: 0.3730 data: 0.0002 max mem: 52150 +Epoch: [20] [5850/7269] eta: 0:10:11 lr: 0.000006 loss: 0.5417 (0.5211) time: 0.4186 data: 0.0001 max mem: 52150 +Epoch: [20] [5860/7269] eta: 0:10:07 lr: 0.000006 loss: 0.5417 (0.5212) time: 0.4813 data: 0.0001 max mem: 52150 +Epoch: [20] [5870/7269] eta: 0:10:03 lr: 0.000006 loss: 0.5235 (0.5211) time: 0.4351 data: 0.0002 max mem: 52150 +Epoch: [20] [5880/7269] eta: 0:09:58 lr: 0.000006 loss: 0.5033 (0.5211) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [20] [5890/7269] eta: 0:09:54 lr: 0.000006 loss: 0.5478 (0.5211) time: 0.4332 data: 0.0001 max mem: 52150 +Epoch: [20] [5900/7269] eta: 0:09:50 lr: 0.000006 loss: 0.5440 (0.5211) time: 0.4952 data: 0.0002 max mem: 52150 +Epoch: [20] [5910/7269] eta: 0:09:46 lr: 0.000006 loss: 0.5049 (0.5211) time: 0.4333 data: 0.0002 max mem: 52150 +Epoch: [20] [5920/7269] eta: 0:09:41 lr: 0.000006 loss: 0.5241 (0.5211) time: 0.3721 data: 0.0001 max mem: 52150 +Epoch: [20] [5930/7269] eta: 0:09:37 lr: 0.000006 loss: 0.5133 (0.5210) time: 0.4365 data: 0.0002 max mem: 52150 +Epoch: [20] [5940/7269] eta: 0:09:33 lr: 0.000006 loss: 0.5147 (0.5211) time: 0.4949 data: 0.0002 max mem: 52150 +Epoch: [20] [5950/7269] eta: 0:09:28 lr: 0.000006 loss: 0.5074 (0.5210) time: 0.4308 data: 0.0001 max mem: 52150 +Epoch: [20] [5960/7269] eta: 0:09:24 lr: 0.000006 loss: 0.5074 (0.5210) time: 0.4319 data: 0.0002 max mem: 52150 +Epoch: [20] [5970/7269] eta: 0:09:20 lr: 0.000006 loss: 0.5348 (0.5210) time: 0.4943 data: 0.0002 max mem: 52150 +Epoch: [20] [5980/7269] eta: 0:09:16 lr: 0.000006 loss: 0.5373 (0.5211) time: 0.4943 data: 0.0001 max mem: 52150 +Epoch: [20] [5990/7269] eta: 0:09:11 lr: 0.000006 loss: 0.5397 (0.5211) time: 0.4317 data: 0.0001 max mem: 52150 +Epoch: [20] [6000/7269] eta: 0:09:07 lr: 0.000006 loss: 0.5495 (0.5212) time: 0.4172 data: 0.0001 max mem: 52150 +Epoch: [20] [6010/7269] eta: 0:09:03 lr: 0.000006 loss: 0.5245 (0.5211) time: 0.4177 data: 0.0002 max mem: 52150 +Epoch: [20] [6020/7269] eta: 0:08:58 lr: 0.000006 loss: 0.5180 (0.5211) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [20] [6030/7269] eta: 0:08:54 lr: 0.000006 loss: 0.5136 (0.5211) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [20] [6040/7269] eta: 0:08:49 lr: 0.000006 loss: 0.5117 (0.5210) time: 0.3723 data: 0.0001 max mem: 52150 +Epoch: [20] [6050/7269] eta: 0:08:45 lr: 0.000006 loss: 0.5158 (0.5211) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [20] [6060/7269] eta: 0:08:41 lr: 0.000006 loss: 0.5246 (0.5210) time: 0.4713 data: 0.0001 max mem: 52150 +Epoch: [20] [6070/7269] eta: 0:08:36 lr: 0.000006 loss: 0.5067 (0.5210) time: 0.4718 data: 0.0001 max mem: 52150 +Epoch: [20] [6080/7269] eta: 0:08:32 lr: 0.000006 loss: 0.5193 (0.5210) time: 0.3722 data: 0.0001 max mem: 52150 +Epoch: [20] [6090/7269] eta: 0:08:28 lr: 0.000006 loss: 0.5385 (0.5210) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [20] [6100/7269] eta: 0:08:23 lr: 0.000006 loss: 0.5137 (0.5210) time: 0.4332 data: 0.0001 max mem: 52150 +Epoch: [20] [6110/7269] eta: 0:08:19 lr: 0.000006 loss: 0.5137 (0.5210) time: 0.4958 data: 0.0001 max mem: 52150 +Epoch: [20] [6120/7269] eta: 0:08:15 lr: 0.000006 loss: 0.5422 (0.5210) time: 0.4789 data: 0.0001 max mem: 52150 +Epoch: [20] [6130/7269] eta: 0:08:11 lr: 0.000006 loss: 0.5418 (0.5210) time: 0.4160 data: 0.0002 max mem: 52150 +Epoch: [20] [6140/7269] eta: 0:08:06 lr: 0.000006 loss: 0.5502 (0.5210) time: 0.4159 data: 0.0002 max mem: 52150 +Epoch: [20] [6150/7269] eta: 0:08:02 lr: 0.000006 loss: 0.5712 (0.5211) time: 0.4781 data: 0.0002 max mem: 52150 +Epoch: [20] [6160/7269] eta: 0:07:58 lr: 0.000006 loss: 0.5705 (0.5211) time: 0.4980 data: 0.0002 max mem: 52150 +Epoch: [20] [6170/7269] eta: 0:07:54 lr: 0.000006 loss: 0.5450 (0.5211) time: 0.4846 data: 0.0001 max mem: 52150 +Epoch: [20] [6180/7269] eta: 0:07:49 lr: 0.000006 loss: 0.5225 (0.5212) time: 0.4209 data: 0.0001 max mem: 52150 +Epoch: [20] [6190/7269] eta: 0:07:45 lr: 0.000006 loss: 0.5382 (0.5212) time: 0.3722 data: 0.0001 max mem: 52150 +Epoch: [20] [6200/7269] eta: 0:07:40 lr: 0.000006 loss: 0.5445 (0.5212) time: 0.3722 data: 0.0001 max mem: 52150 +Epoch: [20] [6210/7269] eta: 0:07:36 lr: 0.000006 loss: 0.5219 (0.5212) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [20] [6220/7269] eta: 0:07:32 lr: 0.000006 loss: 0.4659 (0.5211) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [20] [6230/7269] eta: 0:07:27 lr: 0.000006 loss: 0.4880 (0.5211) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [20] [6240/7269] eta: 0:07:23 lr: 0.000006 loss: 0.5550 (0.5211) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [20] [6250/7269] eta: 0:07:18 lr: 0.000006 loss: 0.5146 (0.5211) time: 0.3727 data: 0.0002 max mem: 52150 +Epoch: [20] [6260/7269] eta: 0:07:14 lr: 0.000006 loss: 0.4808 (0.5210) time: 0.4172 data: 0.0002 max mem: 52150 +Epoch: [20] [6270/7269] eta: 0:07:10 lr: 0.000006 loss: 0.4790 (0.5210) time: 0.4172 data: 0.0002 max mem: 52150 +Epoch: [20] [6280/7269] eta: 0:07:05 lr: 0.000006 loss: 0.4990 (0.5209) time: 0.3729 data: 0.0002 max mem: 52150 +Epoch: [20] [6290/7269] eta: 0:07:01 lr: 0.000006 loss: 0.5145 (0.5209) time: 0.4183 data: 0.0002 max mem: 52150 +Epoch: [20] [6300/7269] eta: 0:06:57 lr: 0.000006 loss: 0.5530 (0.5210) time: 0.4179 data: 0.0001 max mem: 52150 +Epoch: [20] [6310/7269] eta: 0:06:52 lr: 0.000006 loss: 0.5421 (0.5210) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [20] [6320/7269] eta: 0:06:48 lr: 0.000006 loss: 0.5383 (0.5210) time: 0.4365 data: 0.0002 max mem: 52150 +Epoch: [20] [6330/7269] eta: 0:06:44 lr: 0.000006 loss: 0.5431 (0.5210) time: 0.4368 data: 0.0002 max mem: 52150 +Epoch: [20] [6340/7269] eta: 0:06:40 lr: 0.000006 loss: 0.5146 (0.5209) time: 0.4814 data: 0.0002 max mem: 52150 +Epoch: [20] [6350/7269] eta: 0:06:35 lr: 0.000006 loss: 0.4948 (0.5209) time: 0.5286 data: 0.0002 max mem: 52150 +Epoch: [20] [6360/7269] eta: 0:06:31 lr: 0.000006 loss: 0.5044 (0.5209) time: 0.5095 data: 0.0002 max mem: 52150 +Epoch: [20] [6370/7269] eta: 0:06:27 lr: 0.000006 loss: 0.5240 (0.5208) time: 0.5238 data: 0.0002 max mem: 52150 +Epoch: [20] [6380/7269] eta: 0:06:23 lr: 0.000006 loss: 0.5030 (0.5208) time: 0.4788 data: 0.0001 max mem: 52150 +Epoch: [20] [6390/7269] eta: 0:06:18 lr: 0.000006 loss: 0.5163 (0.5208) time: 0.4176 data: 0.0001 max mem: 52150 +Epoch: [20] [6400/7269] eta: 0:06:14 lr: 0.000006 loss: 0.5513 (0.5209) time: 0.3734 data: 0.0001 max mem: 52150 +Epoch: [20] [6410/7269] eta: 0:06:10 lr: 0.000006 loss: 0.5314 (0.5209) time: 0.4344 data: 0.0002 max mem: 52150 +Epoch: [20] [6420/7269] eta: 0:06:05 lr: 0.000006 loss: 0.5126 (0.5209) time: 0.4334 data: 0.0002 max mem: 52150 +Epoch: [20] [6430/7269] eta: 0:06:01 lr: 0.000006 loss: 0.5485 (0.5208) time: 0.4340 data: 0.0002 max mem: 52150 +Epoch: [20] [6440/7269] eta: 0:05:57 lr: 0.000006 loss: 0.4982 (0.5208) time: 0.4787 data: 0.0001 max mem: 52150 +Epoch: [20] [6450/7269] eta: 0:05:53 lr: 0.000006 loss: 0.4897 (0.5208) time: 0.4798 data: 0.0001 max mem: 52150 +Epoch: [20] [6460/7269] eta: 0:05:48 lr: 0.000006 loss: 0.5314 (0.5208) time: 0.4355 data: 0.0001 max mem: 52150 +Epoch: [20] [6470/7269] eta: 0:05:44 lr: 0.000006 loss: 0.5314 (0.5207) time: 0.3721 data: 0.0001 max mem: 52150 +Epoch: [20] [6480/7269] eta: 0:05:39 lr: 0.000006 loss: 0.5424 (0.5208) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [20] [6490/7269] eta: 0:05:35 lr: 0.000006 loss: 0.5443 (0.5208) time: 0.3723 data: 0.0001 max mem: 52150 +Epoch: [20] [6500/7269] eta: 0:05:31 lr: 0.000006 loss: 0.5004 (0.5208) time: 0.4686 data: 0.0001 max mem: 52150 +Epoch: [20] [6510/7269] eta: 0:05:27 lr: 0.000006 loss: 0.5225 (0.5208) time: 0.5147 data: 0.0001 max mem: 52150 +Epoch: [20] [6520/7269] eta: 0:05:22 lr: 0.000006 loss: 0.5380 (0.5208) time: 0.4636 data: 0.0001 max mem: 52150 +Epoch: [20] [6530/7269] eta: 0:05:18 lr: 0.000006 loss: 0.5493 (0.5209) time: 0.4170 data: 0.0002 max mem: 52150 +Epoch: [20] [6540/7269] eta: 0:05:14 lr: 0.000006 loss: 0.5570 (0.5208) time: 0.4321 data: 0.0002 max mem: 52150 +Epoch: [20] [6550/7269] eta: 0:05:10 lr: 0.000006 loss: 0.4756 (0.5207) time: 0.4958 data: 0.0001 max mem: 52150 +Epoch: [20] [6560/7269] eta: 0:05:05 lr: 0.000006 loss: 0.4904 (0.5208) time: 0.4358 data: 0.0001 max mem: 52150 +Epoch: [20] [6570/7269] eta: 0:05:01 lr: 0.000006 loss: 0.5395 (0.5208) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [20] [6580/7269] eta: 0:04:56 lr: 0.000006 loss: 0.5237 (0.5208) time: 0.3726 data: 0.0001 max mem: 52150 +Epoch: [20] [6590/7269] eta: 0:04:52 lr: 0.000006 loss: 0.5055 (0.5207) time: 0.4970 data: 0.0002 max mem: 52150 +Epoch: [20] [6600/7269] eta: 0:04:48 lr: 0.000006 loss: 0.4634 (0.5207) time: 0.4963 data: 0.0002 max mem: 52150 +Epoch: [20] [6610/7269] eta: 0:04:44 lr: 0.000006 loss: 0.4854 (0.5207) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [20] [6620/7269] eta: 0:04:39 lr: 0.000006 loss: 0.5103 (0.5206) time: 0.4670 data: 0.0002 max mem: 52150 +Epoch: [20] [6630/7269] eta: 0:04:35 lr: 0.000006 loss: 0.4944 (0.5206) time: 0.4670 data: 0.0002 max mem: 52150 +Epoch: [20] [6640/7269] eta: 0:04:31 lr: 0.000006 loss: 0.4944 (0.5206) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [20] [6650/7269] eta: 0:04:26 lr: 0.000006 loss: 0.5320 (0.5205) time: 0.4188 data: 0.0001 max mem: 52150 +Epoch: [20] [6660/7269] eta: 0:04:22 lr: 0.000006 loss: 0.5140 (0.5205) time: 0.4190 data: 0.0001 max mem: 52150 +Epoch: [20] [6670/7269] eta: 0:04:18 lr: 0.000006 loss: 0.5078 (0.5205) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [20] [6680/7269] eta: 0:04:13 lr: 0.000006 loss: 0.5030 (0.5205) time: 0.4320 data: 0.0002 max mem: 52150 +Epoch: [20] [6690/7269] eta: 0:04:09 lr: 0.000006 loss: 0.5406 (0.5205) time: 0.4321 data: 0.0001 max mem: 52150 +Epoch: [20] [6700/7269] eta: 0:04:05 lr: 0.000006 loss: 0.5508 (0.5205) time: 0.4203 data: 0.0001 max mem: 52150 +Epoch: [20] [6710/7269] eta: 0:04:00 lr: 0.000006 loss: 0.5310 (0.5205) time: 0.4199 data: 0.0002 max mem: 52150 +Epoch: [20] [6720/7269] eta: 0:03:56 lr: 0.000006 loss: 0.5317 (0.5205) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [20] [6730/7269] eta: 0:03:52 lr: 0.000006 loss: 0.5478 (0.5206) time: 0.3732 data: 0.0001 max mem: 52150 +Epoch: [20] [6740/7269] eta: 0:03:47 lr: 0.000006 loss: 0.5479 (0.5206) time: 0.4811 data: 0.0002 max mem: 52150 +Epoch: [20] [6750/7269] eta: 0:03:43 lr: 0.000006 loss: 0.5543 (0.5206) time: 0.4807 data: 0.0002 max mem: 52150 +Epoch: [20] [6760/7269] eta: 0:03:39 lr: 0.000006 loss: 0.5493 (0.5206) time: 0.4168 data: 0.0002 max mem: 52150 +Epoch: [20] [6770/7269] eta: 0:03:35 lr: 0.000006 loss: 0.5309 (0.5206) time: 0.4810 data: 0.0002 max mem: 52150 +Epoch: [20] [6780/7269] eta: 0:03:30 lr: 0.000006 loss: 0.5205 (0.5206) time: 0.4367 data: 0.0002 max mem: 52150 +Epoch: [20] [6790/7269] eta: 0:03:26 lr: 0.000006 loss: 0.5182 (0.5206) time: 0.3726 data: 0.0001 max mem: 52150 +Epoch: [20] [6800/7269] eta: 0:03:22 lr: 0.000006 loss: 0.5182 (0.5206) time: 0.3727 data: 0.0001 max mem: 52150 +Epoch: [20] [6810/7269] eta: 0:03:17 lr: 0.000006 loss: 0.5116 (0.5206) time: 0.4366 data: 0.0001 max mem: 52150 +Epoch: [20] [6820/7269] eta: 0:03:13 lr: 0.000006 loss: 0.5116 (0.5206) time: 0.4361 data: 0.0001 max mem: 52150 +Epoch: [20] [6830/7269] eta: 0:03:09 lr: 0.000006 loss: 0.5678 (0.5206) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [20] [6840/7269] eta: 0:03:04 lr: 0.000006 loss: 0.5741 (0.5206) time: 0.4180 data: 0.0001 max mem: 52150 +Epoch: [20] [6850/7269] eta: 0:03:00 lr: 0.000006 loss: 0.5387 (0.5206) time: 0.4181 data: 0.0002 max mem: 52150 +Epoch: [20] [6860/7269] eta: 0:02:56 lr: 0.000006 loss: 0.5348 (0.5207) time: 0.4170 data: 0.0002 max mem: 52150 +Epoch: [20] [6870/7269] eta: 0:02:51 lr: 0.000006 loss: 0.5507 (0.5207) time: 0.4172 data: 0.0001 max mem: 52150 +Epoch: [20] [6880/7269] eta: 0:02:47 lr: 0.000006 loss: 0.5453 (0.5207) time: 0.3727 data: 0.0002 max mem: 52150 +Epoch: [20] [6890/7269] eta: 0:02:43 lr: 0.000006 loss: 0.5476 (0.5207) time: 0.4172 data: 0.0002 max mem: 52150 +Epoch: [20] [6900/7269] eta: 0:02:38 lr: 0.000006 loss: 0.5641 (0.5208) time: 0.4794 data: 0.0001 max mem: 52150 +Epoch: [20] [6910/7269] eta: 0:02:34 lr: 0.000006 loss: 0.5524 (0.5208) time: 0.4348 data: 0.0001 max mem: 52150 +Epoch: [20] [6920/7269] eta: 0:02:30 lr: 0.000006 loss: 0.5367 (0.5208) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [20] [6930/7269] eta: 0:02:25 lr: 0.000006 loss: 0.5320 (0.5208) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [20] [6940/7269] eta: 0:02:21 lr: 0.000006 loss: 0.5320 (0.5209) time: 0.4170 data: 0.0002 max mem: 52150 +Epoch: [20] [6950/7269] eta: 0:02:17 lr: 0.000006 loss: 0.5488 (0.5209) time: 0.4172 data: 0.0002 max mem: 52150 +Epoch: [20] [6960/7269] eta: 0:02:13 lr: 0.000006 loss: 0.5208 (0.5208) time: 0.4197 data: 0.0002 max mem: 52150 +Epoch: [20] [6970/7269] eta: 0:02:08 lr: 0.000006 loss: 0.4916 (0.5208) time: 0.5099 data: 0.0002 max mem: 52150 +Epoch: [20] [6980/7269] eta: 0:02:04 lr: 0.000006 loss: 0.5068 (0.5208) time: 0.4623 data: 0.0002 max mem: 52150 +Epoch: [20] [6990/7269] eta: 0:02:00 lr: 0.000006 loss: 0.5235 (0.5208) time: 0.4345 data: 0.0002 max mem: 52150 +Epoch: [20] [7000/7269] eta: 0:01:55 lr: 0.000006 loss: 0.5415 (0.5208) time: 0.4344 data: 0.0002 max mem: 52150 +Epoch: [20] [7010/7269] eta: 0:01:51 lr: 0.000006 loss: 0.5159 (0.5208) time: 0.4329 data: 0.0001 max mem: 52150 +Epoch: [20] [7020/7269] eta: 0:01:47 lr: 0.000006 loss: 0.5159 (0.5208) time: 0.6055 data: 0.0001 max mem: 52150 +Epoch: [20] [7030/7269] eta: 0:01:43 lr: 0.000006 loss: 0.5239 (0.5208) time: 0.5916 data: 0.0001 max mem: 52150 +Epoch: [20] [7040/7269] eta: 0:01:38 lr: 0.000006 loss: 0.5277 (0.5208) time: 0.4823 data: 0.0001 max mem: 52150 +Epoch: [20] [7050/7269] eta: 0:01:34 lr: 0.000006 loss: 0.5329 (0.5208) time: 0.4835 data: 0.0001 max mem: 52150 +Epoch: [20] [7060/7269] eta: 0:01:30 lr: 0.000006 loss: 0.5339 (0.5208) time: 0.4663 data: 0.0001 max mem: 52150 +Epoch: [20] [7070/7269] eta: 0:01:25 lr: 0.000006 loss: 0.5332 (0.5208) time: 0.4179 data: 0.0001 max mem: 52150 +Epoch: [20] [7080/7269] eta: 0:01:21 lr: 0.000006 loss: 0.5397 (0.5209) time: 0.4352 data: 0.0001 max mem: 52150 +Epoch: [20] [7090/7269] eta: 0:01:17 lr: 0.000006 loss: 0.5580 (0.5209) time: 0.4354 data: 0.0001 max mem: 52150 +Epoch: [20] [7100/7269] eta: 0:01:12 lr: 0.000006 loss: 0.5415 (0.5209) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [20] [7110/7269] eta: 0:01:08 lr: 0.000006 loss: 0.5143 (0.5209) time: 0.4180 data: 0.0002 max mem: 52150 +Epoch: [20] [7120/7269] eta: 0:01:04 lr: 0.000006 loss: 0.5202 (0.5209) time: 0.4174 data: 0.0002 max mem: 52150 +Epoch: [20] [7130/7269] eta: 0:00:59 lr: 0.000006 loss: 0.5226 (0.5209) time: 0.4187 data: 0.0001 max mem: 52150 +Epoch: [20] [7140/7269] eta: 0:00:55 lr: 0.000006 loss: 0.5095 (0.5209) time: 0.4800 data: 0.0001 max mem: 52150 +Epoch: [20] [7150/7269] eta: 0:00:51 lr: 0.000006 loss: 0.5068 (0.5209) time: 0.4962 data: 0.0001 max mem: 52150 +Epoch: [20] [7160/7269] eta: 0:00:47 lr: 0.000006 loss: 0.5201 (0.5209) time: 0.4353 data: 0.0002 max mem: 52150 +Epoch: [20] [7170/7269] eta: 0:00:42 lr: 0.000006 loss: 0.5226 (0.5209) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [20] [7180/7269] eta: 0:00:38 lr: 0.000006 loss: 0.5349 (0.5209) time: 0.3727 data: 0.0002 max mem: 52150 +Epoch: [20] [7190/7269] eta: 0:00:34 lr: 0.000006 loss: 0.4975 (0.5209) time: 0.3731 data: 0.0002 max mem: 52150 +Epoch: [20] [7200/7269] eta: 0:00:29 lr: 0.000006 loss: 0.5130 (0.5209) time: 0.4207 data: 0.0001 max mem: 52150 +Epoch: [20] [7210/7269] eta: 0:00:25 lr: 0.000006 loss: 0.5264 (0.5209) time: 0.4813 data: 0.0001 max mem: 52150 +Epoch: [20] [7220/7269] eta: 0:00:21 lr: 0.000006 loss: 0.4992 (0.5209) time: 0.4330 data: 0.0001 max mem: 52150 +Epoch: [20] [7230/7269] eta: 0:00:16 lr: 0.000006 loss: 0.4811 (0.5208) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [20] [7240/7269] eta: 0:00:12 lr: 0.000006 loss: 0.5263 (0.5208) time: 0.3722 data: 0.0001 max mem: 52150 +Epoch: [20] [7250/7269] eta: 0:00:08 lr: 0.000006 loss: 0.5055 (0.5208) time: 0.3751 data: 0.0038 max mem: 52150 +Epoch: [20] [7260/7269] eta: 0:00:03 lr: 0.000006 loss: 0.5221 (0.5209) time: 0.3748 data: 0.0038 max mem: 52150 +Epoch: [20] [7268/7269] eta: 0:00:00 lr: 0.000006 loss: 0.5544 (0.5209) time: 0.3745 data: 0.0038 max mem: 52150 +Epoch: [20] Total time: 0:52:13 (0.4311 s / it) +Averaged stats: lr: 0.000006 loss: 0.5544 (0.5223) +Test: [ 0/1403] eta: 1:17:11 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0913 (0.0913) acc1: 100.0000 (100.0000) acc5: 100.0000 (100.0000) time: 3.3014 data: 3.2044 max mem: 52150 +Test: [ 10/1403] eta: 0:09:00 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1494 (0.1843) acc1: 98.4375 (95.0284) acc5: 100.0000 (100.0000) time: 0.3878 data: 0.2916 max mem: 52150 +Test: [ 20/1403] eta: 0:05:44 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1305 (0.1705) acc1: 98.4375 (95.5357) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 30/1403] eta: 0:04:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0953 (0.1594) acc1: 100.0000 (96.0181) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 40/1403] eta: 0:03:58 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1022 (0.1572) acc1: 100.0000 (96.3796) acc5: 100.0000 (100.0000) time: 0.0967 data: 0.0004 max mem: 52150 +Test: [ 50/1403] eta: 0:03:35 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1086 (0.1756) acc1: 100.0000 (95.1287) acc5: 100.0000 (100.0000) time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 60/1403] eta: 0:03:20 acc1_pos: 90.625 acc5_pos: 100.000 loss: 0.1105 (0.1685) acc1: 100.0000 (95.6967) acc5: 100.0000 (100.0000) time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 70/1403] eta: 0:03:08 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.1322 (0.1696) acc1: 98.4375 (95.5326) acc5: 100.0000 (100.0000) time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 80/1403] eta: 0:03:00 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1407 (0.1748) acc1: 96.8750 (95.2160) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 90/1403] eta: 0:02:53 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1788 (0.1908) acc1: 95.3125 (94.4540) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 100/1403] eta: 0:02:47 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1784 (0.1893) acc1: 95.3125 (94.5854) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 110/1403] eta: 0:02:42 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1330 (0.1857) acc1: 98.4375 (94.7776) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 120/1403] eta: 0:02:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0798 (0.1769) acc1: 100.0000 (95.2092) acc5: 100.0000 (100.0000) time: 0.0969 data: 0.0003 max mem: 52150 +Test: [ 130/1403] eta: 0:02:34 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0804 (0.1707) acc1: 100.0000 (95.4676) acc5: 100.0000 (100.0000) time: 0.0970 data: 0.0003 max mem: 52150 +Test: [ 140/1403] eta: 0:02:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0771 (0.1639) acc1: 100.0000 (95.7890) acc5: 100.0000 (100.0000) time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 150/1403] eta: 0:02:27 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0744 (0.1581) acc1: 100.0000 (96.0575) acc5: 100.0000 (100.0000) time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 160/1403] eta: 0:02:24 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0747 (0.1535) acc1: 100.0000 (96.2636) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 170/1403] eta: 0:02:22 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0767 (0.1497) acc1: 100.0000 (96.4547) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 180/1403] eta: 0:02:19 acc1_pos: 79.688 acc5_pos: 100.000 loss: 0.0845 (0.1549) acc1: 100.0000 (96.1758) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 190/1403] eta: 0:02:17 acc1_pos: 89.062 acc5_pos: 100.000 loss: 0.2742 (0.1649) acc1: 89.0625 (95.6397) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 200/1403] eta: 0:02:15 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.2742 (0.1790) acc1: 89.0625 (94.9316) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 210/1403] eta: 0:02:13 acc1_pos: 79.688 acc5_pos: 100.000 loss: 0.3683 (0.1913) acc1: 84.3750 (94.2758) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 220/1403] eta: 0:02:11 acc1_pos: 82.812 acc5_pos: 100.000 loss: 0.3683 (0.2018) acc1: 82.8125 (93.7429) acc5: 100.0000 (100.0000) time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 230/1403] eta: 0:02:09 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2481 (0.1976) acc1: 90.6250 (93.9597) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 240/1403] eta: 0:02:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0781 (0.1928) acc1: 100.0000 (94.1974) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 250/1403] eta: 0:02:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0790 (0.1886) acc1: 100.0000 (94.4099) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 260/1403] eta: 0:02:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0795 (0.1847) acc1: 100.0000 (94.6061) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 270/1403] eta: 0:02:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0785 (0.1809) acc1: 100.0000 (94.7821) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 280/1403] eta: 0:02:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0781 (0.1772) acc1: 100.0000 (94.9622) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 290/1403] eta: 0:01:59 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0755 (0.1737) acc1: 100.0000 (95.1299) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 300/1403] eta: 0:01:58 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0736 (0.1706) acc1: 100.0000 (95.2865) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 310/1403] eta: 0:01:56 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0787 (0.1677) acc1: 100.0000 (95.4381) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 320/1403] eta: 0:01:55 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0942 (0.1655) acc1: 100.0000 (95.5802) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 330/1403] eta: 0:01:54 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0997 (0.1636) acc1: 100.0000 (95.7137) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 340/1403] eta: 0:01:52 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0986 (0.1616) acc1: 100.0000 (95.8394) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 350/1403] eta: 0:01:51 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0975 (0.1599) acc1: 100.0000 (95.9580) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 360/1403] eta: 0:01:50 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0998 (0.1583) acc1: 100.0000 (96.0656) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 370/1403] eta: 0:01:48 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0991 (0.1566) acc1: 100.0000 (96.1717) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 380/1403] eta: 0:01:47 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0973 (0.1551) acc1: 100.0000 (96.2721) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 390/1403] eta: 0:01:46 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0982 (0.1558) acc1: 100.0000 (96.2276) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 400/1403] eta: 0:01:44 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2467 (0.1602) acc1: 87.5000 (95.9944) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 410/1403] eta: 0:01:43 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3172 (0.1639) acc1: 87.5000 (95.7801) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 420/1403] eta: 0:01:42 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2779 (0.1658) acc1: 89.0625 (95.6725) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 430/1403] eta: 0:01:41 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2669 (0.1692) acc1: 89.0625 (95.4938) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 440/1403] eta: 0:01:40 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3193 (0.1733) acc1: 87.5000 (95.2948) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 450/1403] eta: 0:01:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3044 (0.1759) acc1: 87.5000 (95.1601) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 460/1403] eta: 0:01:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2902 (0.1787) acc1: 89.0625 (95.0142) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 470/1403] eta: 0:01:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2494 (0.1778) acc1: 90.6250 (95.0372) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 480/1403] eta: 0:01:35 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0620 (0.1754) acc1: 100.0000 (95.1403) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 490/1403] eta: 0:01:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0623 (0.1731) acc1: 100.0000 (95.2393) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 500/1403] eta: 0:01:33 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0624 (0.1709) acc1: 100.0000 (95.3343) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 510/1403] eta: 0:01:31 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0625 (0.1688) acc1: 100.0000 (95.4256) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 520/1403] eta: 0:01:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0657 (0.1669) acc1: 100.0000 (95.5134) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 530/1403] eta: 0:01:29 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0597 (0.1649) acc1: 100.0000 (95.5979) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 540/1403] eta: 0:01:28 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0594 (0.1629) acc1: 100.0000 (95.6793) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 550/1403] eta: 0:01:27 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0599 (0.1615) acc1: 100.0000 (95.7379) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 560/1403] eta: 0:01:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0830 (0.1604) acc1: 98.4375 (95.7888) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 570/1403] eta: 0:01:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0878 (0.1592) acc1: 98.4375 (95.8461) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 580/1403] eta: 0:01:24 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0878 (0.1585) acc1: 98.4375 (95.8665) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 590/1403] eta: 0:01:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1154 (0.1582) acc1: 96.8750 (95.8677) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 600/1403] eta: 0:01:21 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1289 (0.1583) acc1: 95.3125 (95.8247) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 610/1403] eta: 0:01:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1289 (0.1577) acc1: 95.3125 (95.8393) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 620/1403] eta: 0:01:19 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1348 (0.1587) acc1: 95.3125 (95.7428) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 630/1403] eta: 0:01:18 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1752 (0.1588) acc1: 92.1875 (95.7260) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 640/1403] eta: 0:01:17 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1359 (0.1585) acc1: 95.3125 (95.7293) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0969 data: 0.0004 max mem: 52150 +Test: [ 650/1403] eta: 0:01:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1359 (0.1582) acc1: 95.3125 (95.7349) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0970 data: 0.0004 max mem: 52150 +Test: [ 660/1403] eta: 0:01:15 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0994 (0.1572) acc1: 98.4375 (95.7876) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 670/1403] eta: 0:01:14 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0742 (0.1559) acc1: 100.0000 (95.8411) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 680/1403] eta: 0:01:13 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0864 (0.1575) acc1: 98.4375 (95.7347) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 690/1403] eta: 0:01:12 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3206 (0.1623) acc1: 85.9375 (95.4482) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 700/1403] eta: 0:01:11 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1993 (0.1626) acc1: 89.0625 (95.4329) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 710/1403] eta: 0:01:10 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3259 (0.1672) acc1: 85.9375 (95.1850) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 720/1403] eta: 0:01:09 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3259 (0.1672) acc1: 85.9375 (95.1825) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 730/1403] eta: 0:01:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0691 (0.1660) acc1: 100.0000 (95.2313) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 740/1403] eta: 0:01:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0743 (0.1651) acc1: 100.0000 (95.2767) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 750/1403] eta: 0:01:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1144 (0.1667) acc1: 96.8750 (95.1835) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 760/1403] eta: 0:01:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3611 (0.1703) acc1: 82.8125 (94.9573) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 770/1403] eta: 0:01:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5311 (0.1754) acc1: 70.3125 (94.6417) acc5: 100.0000 (100.0000) acc1_neg: 70.312 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 780/1403] eta: 0:01:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4429 (0.1781) acc1: 75.0000 (94.4842) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 790/1403] eta: 0:01:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3185 (0.1805) acc1: 82.8125 (94.3624) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 800/1403] eta: 0:01:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2259 (0.1808) acc1: 90.6250 (94.3450) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 810/1403] eta: 0:00:59 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2259 (0.1839) acc1: 89.0625 (94.1951) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 820/1403] eta: 0:00:58 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6548 (0.1929) acc1: 70.3125 (93.8109) acc5: 100.0000 (100.0000) acc1_neg: 42.188 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 830/1403] eta: 0:00:57 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.7274 (0.1958) acc1: 70.3125 (93.6710) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 840/1403] eta: 0:00:56 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4964 (0.2037) acc1: 70.3125 (93.3394) acc5: 100.0000 (100.0000) acc1_neg: 50.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 850/1403] eta: 0:00:55 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.0449 (0.2146) acc1: 56.2500 (92.8815) acc5: 100.0000 (100.0000) acc1_neg: 57.812 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 860/1403] eta: 0:00:54 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.9816 (0.2180) acc1: 57.8125 (92.7374) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 870/1403] eta: 0:00:53 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1357 (0.2169) acc1: 95.3125 (92.7831) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 880/1403] eta: 0:00:52 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1019 (0.2157) acc1: 98.4375 (92.8384) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 890/1403] eta: 0:00:51 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0937 (0.2143) acc1: 98.4375 (92.9082) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 900/1403] eta: 0:00:50 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0756 (0.2127) acc1: 100.0000 (92.9818) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 910/1403] eta: 0:00:49 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0819 (0.2118) acc1: 98.4375 (93.0245) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 920/1403] eta: 0:00:48 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1134 (0.2108) acc1: 98.4375 (93.0646) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 930/1403] eta: 0:00:47 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1008 (0.2097) acc1: 98.4375 (93.1156) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 940/1403] eta: 0:00:46 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1376 (0.2098) acc1: 96.8750 (93.1024) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 950/1403] eta: 0:00:45 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2246 (0.2105) acc1: 89.0625 (93.0419) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 960/1403] eta: 0:00:44 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2636 (0.2110) acc1: 89.0625 (92.9988) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 970/1403] eta: 0:00:43 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2397 (0.2117) acc1: 90.6250 (92.9535) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 980/1403] eta: 0:00:42 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2652 (0.2125) acc1: 89.0625 (92.8995) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 990/1403] eta: 0:00:41 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2457 (0.2128) acc1: 89.0625 (92.8734) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1000/1403] eta: 0:00:40 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2263 (0.2134) acc1: 89.0625 (92.8384) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1010/1403] eta: 0:00:39 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2642 (0.2140) acc1: 89.0625 (92.7980) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1020/1403] eta: 0:00:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4183 (0.2202) acc1: 81.2500 (92.4492) acc5: 100.0000 (100.0000) acc1_neg: 32.812 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1030/1403] eta: 0:00:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5618 (0.2224) acc1: 60.9375 (92.3133) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1040/1403] eta: 0:00:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2412 (0.2223) acc1: 90.6250 (92.3076) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1050/1403] eta: 0:00:35 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1869 (0.2220) acc1: 93.7500 (92.3287) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1060/1403] eta: 0:00:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1605 (0.2213) acc1: 95.3125 (92.3627) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1070/1403] eta: 0:00:33 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1693 (0.2219) acc1: 95.3125 (92.3319) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1080/1403] eta: 0:00:32 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3969 (0.2240) acc1: 82.8125 (92.2294) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1090/1403] eta: 0:00:31 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4432 (0.2261) acc1: 81.2500 (92.1231) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1100/1403] eta: 0:00:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3245 (0.2263) acc1: 85.9375 (92.1109) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1110/1403] eta: 0:00:29 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2501 (0.2264) acc1: 90.6250 (92.1073) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1120/1403] eta: 0:00:28 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2501 (0.2266) acc1: 92.1875 (92.0941) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1130/1403] eta: 0:00:27 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2509 (0.2270) acc1: 90.6250 (92.0701) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1140/1403] eta: 0:00:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2289 (0.2269) acc1: 90.6250 (92.0670) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1150/1403] eta: 0:00:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2196 (0.2270) acc1: 92.1875 (92.0613) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1160/1403] eta: 0:00:24 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2284 (0.2273) acc1: 90.6250 (92.0502) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1170/1403] eta: 0:00:23 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2284 (0.2269) acc1: 90.6250 (92.0607) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1180/1403] eta: 0:00:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0687 (0.2256) acc1: 100.0000 (92.1280) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1190/1403] eta: 0:00:21 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0681 (0.2243) acc1: 100.0000 (92.1941) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52150 +Test: [1200/1403] eta: 0:00:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0681 (0.2230) acc1: 100.0000 (92.2591) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1210/1403] eta: 0:00:19 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0674 (0.2217) acc1: 100.0000 (92.3230) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52150 +Test: [1220/1403] eta: 0:00:18 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0670 (0.2204) acc1: 100.0000 (92.3859) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52150 +Test: [1230/1403] eta: 0:00:17 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0677 (0.2192) acc1: 100.0000 (92.4477) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1240/1403] eta: 0:00:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0677 (0.2180) acc1: 100.0000 (92.5086) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1250/1403] eta: 0:00:15 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0678 (0.2168) acc1: 100.0000 (92.5684) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1260/1403] eta: 0:00:14 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0711 (0.2156) acc1: 100.0000 (92.6274) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1270/1403] eta: 0:00:13 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0723 (0.2145) acc1: 100.0000 (92.6854) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1280/1403] eta: 0:00:12 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0721 (0.2134) acc1: 100.0000 (92.7425) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52150 +Test: [1290/1403] eta: 0:00:11 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0717 (0.2123) acc1: 100.0000 (92.7987) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52150 +Test: [1300/1403] eta: 0:00:10 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0722 (0.2112) acc1: 100.0000 (92.8541) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0004 max mem: 52150 +Test: [1310/1403] eta: 0:00:09 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0722 (0.2102) acc1: 100.0000 (92.9086) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0004 max mem: 52150 +Test: [1320/1403] eta: 0:00:08 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0709 (0.2091) acc1: 100.0000 (92.9622) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1330/1403] eta: 0:00:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0712 (0.2081) acc1: 100.0000 (93.0116) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1340/1403] eta: 0:00:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0675 (0.2070) acc1: 100.0000 (93.0614) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1350/1403] eta: 0:00:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0653 (0.2060) acc1: 100.0000 (93.1116) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1360/1403] eta: 0:00:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0651 (0.2050) acc1: 100.0000 (93.1610) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1370/1403] eta: 0:00:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0657 (0.2040) acc1: 100.0000 (93.2087) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1380/1403] eta: 0:00:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0639 (0.2030) acc1: 100.0000 (93.2567) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1390/1403] eta: 0:00:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0634 (0.2020) acc1: 100.0000 (93.3041) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0002 max mem: 52150 +Test: [1400/1403] eta: 0:00:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0648 (0.2011) acc1: 100.0000 (93.3496) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0001 max mem: 52150 +Test: [1402/1403] eta: 0:00:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0648 (0.2009) acc1: 100.0000 (93.3544) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0924 data: 0.0001 max mem: 52150 +Test: Total time: 0:02:19 (0.0993 s / it) +* Overall Acc@1 93.354 Acc@5 100.000 + target == real Acc@1 95.398 Acc@5 100.000 + target == fake Acc@1 92.779 Acc@5 100.000 + loss 0.201 +Accuracy of the network on the 89729 test images: 93.4% +Max accuracy: 93.40% +Epoch: [21] [ 0/7269] eta: 11:32:03 lr: 0.000005 loss: 0.4628 (0.4628) time: 5.7124 data: 4.9485 max mem: 52150 +Epoch: [21] [ 10/7269] eta: 1:55:04 lr: 0.000005 loss: 0.4628 (0.4814) time: 0.9512 data: 0.4500 max mem: 52150 +Epoch: [21] [ 20/7269] eta: 1:28:49 lr: 0.000005 loss: 0.5096 (0.5089) time: 0.4864 data: 0.0002 max mem: 52150 +Epoch: [21] [ 30/7269] eta: 1:14:35 lr: 0.000005 loss: 0.5381 (0.5107) time: 0.4352 data: 0.0002 max mem: 52150 +Epoch: [21] [ 40/7269] eta: 1:10:07 lr: 0.000005 loss: 0.5244 (0.5171) time: 0.4212 data: 0.0002 max mem: 52150 +Epoch: [21] [ 50/7269] eta: 1:05:03 lr: 0.000005 loss: 0.5244 (0.5122) time: 0.4206 data: 0.0002 max mem: 52150 +Epoch: [21] [ 60/7269] eta: 1:03:30 lr: 0.000005 loss: 0.5263 (0.5140) time: 0.4190 data: 0.0002 max mem: 52150 +Epoch: [21] [ 70/7269] eta: 1:00:46 lr: 0.000005 loss: 0.5263 (0.5153) time: 0.4193 data: 0.0002 max mem: 52150 +Epoch: [21] [ 80/7269] eta: 0:58:41 lr: 0.000005 loss: 0.5350 (0.5177) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [21] [ 90/7269] eta: 0:58:19 lr: 0.000005 loss: 0.5454 (0.5172) time: 0.4195 data: 0.0001 max mem: 52150 +Epoch: [21] [ 100/7269] eta: 0:58:22 lr: 0.000005 loss: 0.5173 (0.5190) time: 0.4834 data: 0.0001 max mem: 52150 +Epoch: [21] [ 110/7269] eta: 0:57:02 lr: 0.000005 loss: 0.5358 (0.5203) time: 0.4355 data: 0.0002 max mem: 52150 +Epoch: [21] [ 120/7269] eta: 0:56:54 lr: 0.000005 loss: 0.5080 (0.5170) time: 0.4221 data: 0.0002 max mem: 52150 +Epoch: [21] [ 130/7269] eta: 0:57:03 lr: 0.000005 loss: 0.5080 (0.5184) time: 0.4880 data: 0.0002 max mem: 52150 +Epoch: [21] [ 140/7269] eta: 0:56:53 lr: 0.000005 loss: 0.5339 (0.5200) time: 0.4865 data: 0.0002 max mem: 52150 +Epoch: [21] [ 150/7269] eta: 0:56:57 lr: 0.000005 loss: 0.5140 (0.5178) time: 0.4837 data: 0.0002 max mem: 52150 +Epoch: [21] [ 160/7269] eta: 0:56:05 lr: 0.000005 loss: 0.4843 (0.5153) time: 0.4346 data: 0.0002 max mem: 52150 +Epoch: [21] [ 170/7269] eta: 0:55:18 lr: 0.000005 loss: 0.5208 (0.5163) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [21] [ 180/7269] eta: 0:54:36 lr: 0.000005 loss: 0.5276 (0.5159) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [21] [ 190/7269] eta: 0:54:33 lr: 0.000005 loss: 0.5045 (0.5139) time: 0.4195 data: 0.0002 max mem: 52150 +Epoch: [21] [ 200/7269] eta: 0:55:06 lr: 0.000005 loss: 0.4982 (0.5140) time: 0.5181 data: 0.0002 max mem: 52150 +Epoch: [21] [ 210/7269] eta: 0:54:29 lr: 0.000005 loss: 0.5438 (0.5155) time: 0.4704 data: 0.0001 max mem: 52150 +Epoch: [21] [ 220/7269] eta: 0:54:26 lr: 0.000005 loss: 0.5265 (0.5158) time: 0.4203 data: 0.0002 max mem: 52150 +Epoch: [21] [ 230/7269] eta: 0:53:53 lr: 0.000005 loss: 0.5004 (0.5152) time: 0.4200 data: 0.0002 max mem: 52150 +Epoch: [21] [ 240/7269] eta: 0:53:23 lr: 0.000005 loss: 0.5235 (0.5164) time: 0.3714 data: 0.0001 max mem: 52150 +Epoch: [21] [ 250/7269] eta: 0:53:22 lr: 0.000005 loss: 0.5354 (0.5155) time: 0.4194 data: 0.0002 max mem: 52150 +Epoch: [21] [ 260/7269] eta: 0:53:21 lr: 0.000005 loss: 0.5204 (0.5156) time: 0.4695 data: 0.0002 max mem: 52150 +Epoch: [21] [ 270/7269] eta: 0:52:55 lr: 0.000005 loss: 0.5204 (0.5158) time: 0.4219 data: 0.0002 max mem: 52150 +Epoch: [21] [ 280/7269] eta: 0:52:30 lr: 0.000005 loss: 0.5201 (0.5147) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [21] [ 290/7269] eta: 0:52:59 lr: 0.000005 loss: 0.5314 (0.5149) time: 0.4815 data: 0.0002 max mem: 52150 +Epoch: [21] [ 300/7269] eta: 0:52:35 lr: 0.000005 loss: 0.5557 (0.5155) time: 0.4819 data: 0.0002 max mem: 52150 +Epoch: [21] [ 310/7269] eta: 0:52:41 lr: 0.000005 loss: 0.5333 (0.5164) time: 0.4346 data: 0.0002 max mem: 52150 +Epoch: [21] [ 320/7269] eta: 0:52:46 lr: 0.000005 loss: 0.5333 (0.5166) time: 0.4976 data: 0.0002 max mem: 52150 +Epoch: [21] [ 330/7269] eta: 0:52:23 lr: 0.000005 loss: 0.5247 (0.5165) time: 0.4348 data: 0.0001 max mem: 52150 +Epoch: [21] [ 340/7269] eta: 0:52:22 lr: 0.000005 loss: 0.5187 (0.5168) time: 0.4196 data: 0.0002 max mem: 52150 +Epoch: [21] [ 350/7269] eta: 0:52:25 lr: 0.000005 loss: 0.5102 (0.5166) time: 0.4809 data: 0.0002 max mem: 52150 +Epoch: [21] [ 360/7269] eta: 0:52:05 lr: 0.000005 loss: 0.5097 (0.5168) time: 0.4332 data: 0.0001 max mem: 52150 +Epoch: [21] [ 370/7269] eta: 0:52:22 lr: 0.000005 loss: 0.5393 (0.5173) time: 0.4708 data: 0.0001 max mem: 52150 +Epoch: [21] [ 380/7269] eta: 0:52:02 lr: 0.000005 loss: 0.5513 (0.5176) time: 0.4700 data: 0.0001 max mem: 52150 +Epoch: [21] [ 390/7269] eta: 0:52:28 lr: 0.000005 loss: 0.5335 (0.5181) time: 0.4968 data: 0.0001 max mem: 52150 +Epoch: [21] [ 400/7269] eta: 0:53:02 lr: 0.000005 loss: 0.4967 (0.5174) time: 0.6549 data: 0.0001 max mem: 52150 +Epoch: [21] [ 410/7269] eta: 0:52:42 lr: 0.000005 loss: 0.4935 (0.5171) time: 0.5294 data: 0.0001 max mem: 52150 +Epoch: [21] [ 420/7269] eta: 0:52:23 lr: 0.000005 loss: 0.5370 (0.5179) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [21] [ 430/7269] eta: 0:52:05 lr: 0.000005 loss: 0.5326 (0.5177) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [21] [ 440/7269] eta: 0:51:47 lr: 0.000005 loss: 0.5366 (0.5183) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [21] [ 450/7269] eta: 0:51:45 lr: 0.000005 loss: 0.5366 (0.5180) time: 0.4205 data: 0.0002 max mem: 52150 +Epoch: [21] [ 460/7269] eta: 0:51:28 lr: 0.000005 loss: 0.5252 (0.5183) time: 0.4209 data: 0.0002 max mem: 52150 +Epoch: [21] [ 470/7269] eta: 0:51:24 lr: 0.000005 loss: 0.5449 (0.5188) time: 0.4156 data: 0.0001 max mem: 52150 +Epoch: [21] [ 480/7269] eta: 0:51:08 lr: 0.000005 loss: 0.5219 (0.5183) time: 0.4158 data: 0.0001 max mem: 52150 +Epoch: [21] [ 490/7269] eta: 0:50:53 lr: 0.000005 loss: 0.5366 (0.5189) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [21] [ 500/7269] eta: 0:50:54 lr: 0.000005 loss: 0.5532 (0.5198) time: 0.4342 data: 0.0001 max mem: 52150 +Epoch: [21] [ 510/7269] eta: 0:50:39 lr: 0.000005 loss: 0.5514 (0.5204) time: 0.4343 data: 0.0002 max mem: 52150 +Epoch: [21] [ 520/7269] eta: 0:50:50 lr: 0.000005 loss: 0.5391 (0.5204) time: 0.4684 data: 0.0002 max mem: 52150 +Epoch: [21] [ 530/7269] eta: 0:50:35 lr: 0.000005 loss: 0.5147 (0.5206) time: 0.4685 data: 0.0001 max mem: 52150 +Epoch: [21] [ 540/7269] eta: 0:50:21 lr: 0.000005 loss: 0.5563 (0.5210) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [21] [ 550/7269] eta: 0:50:18 lr: 0.000005 loss: 0.5447 (0.5213) time: 0.4177 data: 0.0002 max mem: 52150 +Epoch: [21] [ 560/7269] eta: 0:50:19 lr: 0.000005 loss: 0.5190 (0.5212) time: 0.4793 data: 0.0002 max mem: 52150 +Epoch: [21] [ 570/7269] eta: 0:50:05 lr: 0.000005 loss: 0.5557 (0.5214) time: 0.4327 data: 0.0002 max mem: 52150 +Epoch: [21] [ 580/7269] eta: 0:49:52 lr: 0.000005 loss: 0.5568 (0.5214) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [21] [ 590/7269] eta: 0:49:39 lr: 0.000005 loss: 0.5395 (0.5214) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [21] [ 600/7269] eta: 0:49:36 lr: 0.000005 loss: 0.5113 (0.5207) time: 0.4157 data: 0.0002 max mem: 52150 +Epoch: [21] [ 610/7269] eta: 0:49:23 lr: 0.000005 loss: 0.5221 (0.5207) time: 0.4154 data: 0.0001 max mem: 52150 +Epoch: [21] [ 620/7269] eta: 0:49:24 lr: 0.000005 loss: 0.5304 (0.5209) time: 0.4325 data: 0.0001 max mem: 52150 +Epoch: [21] [ 630/7269] eta: 0:49:24 lr: 0.000005 loss: 0.5362 (0.5211) time: 0.4931 data: 0.0002 max mem: 52150 +Epoch: [21] [ 640/7269] eta: 0:49:12 lr: 0.000005 loss: 0.5563 (0.5215) time: 0.4324 data: 0.0002 max mem: 52150 +Epoch: [21] [ 650/7269] eta: 0:49:01 lr: 0.000005 loss: 0.5347 (0.5213) time: 0.3730 data: 0.0002 max mem: 52150 +Epoch: [21] [ 660/7269] eta: 0:48:49 lr: 0.000005 loss: 0.5188 (0.5211) time: 0.3729 data: 0.0001 max mem: 52150 +Epoch: [21] [ 670/7269] eta: 0:48:38 lr: 0.000005 loss: 0.5188 (0.5207) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [21] [ 680/7269] eta: 0:48:26 lr: 0.000005 loss: 0.5176 (0.5207) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [21] [ 690/7269] eta: 0:48:24 lr: 0.000005 loss: 0.5318 (0.5209) time: 0.4171 data: 0.0002 max mem: 52150 +Epoch: [21] [ 700/7269] eta: 0:48:13 lr: 0.000005 loss: 0.5282 (0.5211) time: 0.4173 data: 0.0001 max mem: 52150 +Epoch: [21] [ 710/7269] eta: 0:48:11 lr: 0.000005 loss: 0.5119 (0.5208) time: 0.4178 data: 0.0001 max mem: 52150 +Epoch: [21] [ 720/7269] eta: 0:48:00 lr: 0.000005 loss: 0.5223 (0.5211) time: 0.4179 data: 0.0002 max mem: 52150 +Epoch: [21] [ 730/7269] eta: 0:47:50 lr: 0.000005 loss: 0.5223 (0.5208) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [21] [ 740/7269] eta: 0:47:39 lr: 0.000005 loss: 0.5330 (0.5213) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [21] [ 750/7269] eta: 0:47:37 lr: 0.000005 loss: 0.5624 (0.5216) time: 0.4162 data: 0.0002 max mem: 52150 +Epoch: [21] [ 760/7269] eta: 0:47:27 lr: 0.000005 loss: 0.5270 (0.5216) time: 0.4165 data: 0.0002 max mem: 52150 +Epoch: [21] [ 770/7269] eta: 0:47:17 lr: 0.000005 loss: 0.5347 (0.5219) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [21] [ 780/7269] eta: 0:47:17 lr: 0.000005 loss: 0.5347 (0.5218) time: 0.4327 data: 0.0001 max mem: 52150 +Epoch: [21] [ 790/7269] eta: 0:47:07 lr: 0.000005 loss: 0.5135 (0.5217) time: 0.4326 data: 0.0001 max mem: 52150 +Epoch: [21] [ 800/7269] eta: 0:47:05 lr: 0.000005 loss: 0.4957 (0.5213) time: 0.4181 data: 0.0001 max mem: 52150 +Epoch: [21] [ 810/7269] eta: 0:46:56 lr: 0.000005 loss: 0.4889 (0.5212) time: 0.4180 data: 0.0002 max mem: 52150 +Epoch: [21] [ 820/7269] eta: 0:46:46 lr: 0.000005 loss: 0.5252 (0.5213) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [21] [ 830/7269] eta: 0:46:37 lr: 0.000005 loss: 0.5343 (0.5213) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [21] [ 840/7269] eta: 0:46:35 lr: 0.000005 loss: 0.4950 (0.5208) time: 0.4182 data: 0.0002 max mem: 52150 +Epoch: [21] [ 850/7269] eta: 0:46:33 lr: 0.000005 loss: 0.5028 (0.5210) time: 0.4648 data: 0.0002 max mem: 52150 +Epoch: [21] [ 860/7269] eta: 0:46:24 lr: 0.000005 loss: 0.5324 (0.5211) time: 0.4186 data: 0.0002 max mem: 52150 +Epoch: [21] [ 870/7269] eta: 0:46:31 lr: 0.000005 loss: 0.5292 (0.5210) time: 0.4811 data: 0.0002 max mem: 52150 +Epoch: [21] [ 880/7269] eta: 0:46:29 lr: 0.000005 loss: 0.5415 (0.5212) time: 0.5269 data: 0.0002 max mem: 52150 +Epoch: [21] [ 890/7269] eta: 0:46:20 lr: 0.000005 loss: 0.5373 (0.5213) time: 0.4174 data: 0.0002 max mem: 52150 +Epoch: [21] [ 900/7269] eta: 0:46:19 lr: 0.000005 loss: 0.5277 (0.5212) time: 0.4313 data: 0.0001 max mem: 52150 +Epoch: [21] [ 910/7269] eta: 0:46:17 lr: 0.000005 loss: 0.4933 (0.5208) time: 0.4790 data: 0.0001 max mem: 52150 +Epoch: [21] [ 920/7269] eta: 0:46:23 lr: 0.000005 loss: 0.4893 (0.5205) time: 0.5297 data: 0.0002 max mem: 52150 +Epoch: [21] [ 930/7269] eta: 0:46:21 lr: 0.000005 loss: 0.5097 (0.5206) time: 0.5283 data: 0.0002 max mem: 52150 +Epoch: [21] [ 940/7269] eta: 0:46:12 lr: 0.000005 loss: 0.5168 (0.5206) time: 0.4181 data: 0.0001 max mem: 52150 +Epoch: [21] [ 950/7269] eta: 0:46:09 lr: 0.000005 loss: 0.5098 (0.5204) time: 0.4171 data: 0.0002 max mem: 52150 +Epoch: [21] [ 960/7269] eta: 0:46:00 lr: 0.000005 loss: 0.5125 (0.5203) time: 0.4170 data: 0.0002 max mem: 52150 +Epoch: [21] [ 970/7269] eta: 0:45:52 lr: 0.000005 loss: 0.5516 (0.5203) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [21] [ 980/7269] eta: 0:45:43 lr: 0.000005 loss: 0.5532 (0.5204) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [21] [ 990/7269] eta: 0:45:35 lr: 0.000005 loss: 0.5287 (0.5204) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [21] [1000/7269] eta: 0:45:34 lr: 0.000005 loss: 0.5338 (0.5206) time: 0.4313 data: 0.0002 max mem: 52150 +Epoch: [21] [1010/7269] eta: 0:45:31 lr: 0.000005 loss: 0.5128 (0.5205) time: 0.4757 data: 0.0001 max mem: 52150 +Epoch: [21] [1020/7269] eta: 0:45:23 lr: 0.000005 loss: 0.4952 (0.5202) time: 0.4158 data: 0.0001 max mem: 52150 +Epoch: [21] [1030/7269] eta: 0:45:31 lr: 0.000005 loss: 0.5128 (0.5203) time: 0.5077 data: 0.0002 max mem: 52150 +Epoch: [21] [1040/7269] eta: 0:45:22 lr: 0.000005 loss: 0.5520 (0.5207) time: 0.5076 data: 0.0002 max mem: 52150 +Epoch: [21] [1050/7269] eta: 0:45:14 lr: 0.000005 loss: 0.5193 (0.5206) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [21] [1060/7269] eta: 0:45:06 lr: 0.000005 loss: 0.5005 (0.5204) time: 0.3712 data: 0.0002 max mem: 52150 +Epoch: [21] [1070/7269] eta: 0:45:03 lr: 0.000005 loss: 0.5205 (0.5204) time: 0.4166 data: 0.0002 max mem: 52150 +Epoch: [21] [1080/7269] eta: 0:45:02 lr: 0.000005 loss: 0.5312 (0.5205) time: 0.4774 data: 0.0002 max mem: 52150 +Epoch: [21] [1090/7269] eta: 0:45:15 lr: 0.000005 loss: 0.5362 (0.5206) time: 0.6150 data: 0.0001 max mem: 52150 +Epoch: [21] [1100/7269] eta: 0:45:07 lr: 0.000005 loss: 0.5585 (0.5209) time: 0.5545 data: 0.0001 max mem: 52150 +Epoch: [21] [1110/7269] eta: 0:44:58 lr: 0.000005 loss: 0.5570 (0.5211) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [21] [1120/7269] eta: 0:44:50 lr: 0.000005 loss: 0.5509 (0.5212) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [21] [1130/7269] eta: 0:44:42 lr: 0.000005 loss: 0.5386 (0.5213) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [21] [1140/7269] eta: 0:44:41 lr: 0.000005 loss: 0.5311 (0.5212) time: 0.4344 data: 0.0002 max mem: 52150 +Epoch: [21] [1150/7269] eta: 0:44:33 lr: 0.000005 loss: 0.5068 (0.5211) time: 0.4347 data: 0.0002 max mem: 52150 +Epoch: [21] [1160/7269] eta: 0:44:26 lr: 0.000005 loss: 0.5194 (0.5211) time: 0.3743 data: 0.0003 max mem: 52150 +Epoch: [21] [1170/7269] eta: 0:44:23 lr: 0.000005 loss: 0.5343 (0.5212) time: 0.4183 data: 0.0002 max mem: 52150 +Epoch: [21] [1180/7269] eta: 0:44:15 lr: 0.000005 loss: 0.5260 (0.5212) time: 0.4159 data: 0.0002 max mem: 52150 +Epoch: [21] [1190/7269] eta: 0:44:07 lr: 0.000005 loss: 0.5269 (0.5213) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [21] [1200/7269] eta: 0:44:00 lr: 0.000005 loss: 0.5269 (0.5210) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [21] [1210/7269] eta: 0:44:01 lr: 0.000005 loss: 0.5035 (0.5211) time: 0.4628 data: 0.0001 max mem: 52150 +Epoch: [21] [1220/7269] eta: 0:43:54 lr: 0.000005 loss: 0.5214 (0.5211) time: 0.4622 data: 0.0002 max mem: 52150 +Epoch: [21] [1230/7269] eta: 0:43:46 lr: 0.000005 loss: 0.5074 (0.5209) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [21] [1240/7269] eta: 0:43:45 lr: 0.000005 loss: 0.5074 (0.5209) time: 0.4330 data: 0.0002 max mem: 52150 +Epoch: [21] [1250/7269] eta: 0:43:37 lr: 0.000005 loss: 0.5212 (0.5208) time: 0.4333 data: 0.0002 max mem: 52150 +Epoch: [21] [1260/7269] eta: 0:43:30 lr: 0.000005 loss: 0.5069 (0.5205) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [21] [1270/7269] eta: 0:43:23 lr: 0.000005 loss: 0.5071 (0.5205) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [21] [1280/7269] eta: 0:43:20 lr: 0.000005 loss: 0.5319 (0.5207) time: 0.4198 data: 0.0002 max mem: 52150 +Epoch: [21] [1290/7269] eta: 0:43:17 lr: 0.000005 loss: 0.5362 (0.5205) time: 0.4655 data: 0.0002 max mem: 52150 +Epoch: [21] [1300/7269] eta: 0:43:20 lr: 0.000005 loss: 0.5354 (0.5207) time: 0.5233 data: 0.0002 max mem: 52150 +Epoch: [21] [1310/7269] eta: 0:43:16 lr: 0.000005 loss: 0.5324 (0.5208) time: 0.5239 data: 0.0002 max mem: 52150 +Epoch: [21] [1320/7269] eta: 0:43:09 lr: 0.000005 loss: 0.4837 (0.5205) time: 0.4182 data: 0.0002 max mem: 52150 +Epoch: [21] [1330/7269] eta: 0:43:08 lr: 0.000005 loss: 0.5043 (0.5206) time: 0.4333 data: 0.0002 max mem: 52150 +Epoch: [21] [1340/7269] eta: 0:43:00 lr: 0.000005 loss: 0.5491 (0.5209) time: 0.4334 data: 0.0002 max mem: 52150 +Epoch: [21] [1350/7269] eta: 0:42:53 lr: 0.000005 loss: 0.5343 (0.5208) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [21] [1360/7269] eta: 0:42:46 lr: 0.000005 loss: 0.5362 (0.5210) time: 0.3709 data: 0.0002 max mem: 52150 +Epoch: [21] [1370/7269] eta: 0:42:39 lr: 0.000005 loss: 0.5194 (0.5209) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [21] [1380/7269] eta: 0:42:32 lr: 0.000005 loss: 0.5038 (0.5209) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [21] [1390/7269] eta: 0:42:30 lr: 0.000005 loss: 0.5297 (0.5211) time: 0.4309 data: 0.0001 max mem: 52150 +Epoch: [21] [1400/7269] eta: 0:42:32 lr: 0.000005 loss: 0.5297 (0.5211) time: 0.5420 data: 0.0001 max mem: 52150 +Epoch: [21] [1410/7269] eta: 0:42:25 lr: 0.000005 loss: 0.5198 (0.5211) time: 0.4825 data: 0.0001 max mem: 52150 +Epoch: [21] [1420/7269] eta: 0:42:19 lr: 0.000005 loss: 0.5254 (0.5211) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [21] [1430/7269] eta: 0:42:12 lr: 0.000005 loss: 0.5266 (0.5211) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [21] [1440/7269] eta: 0:42:12 lr: 0.000005 loss: 0.5139 (0.5210) time: 0.4652 data: 0.0002 max mem: 52150 +Epoch: [21] [1450/7269] eta: 0:42:06 lr: 0.000005 loss: 0.5039 (0.5209) time: 0.4650 data: 0.0002 max mem: 52150 +Epoch: [21] [1460/7269] eta: 0:41:59 lr: 0.000005 loss: 0.5234 (0.5210) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [21] [1470/7269] eta: 0:41:57 lr: 0.000005 loss: 0.5506 (0.5213) time: 0.4339 data: 0.0001 max mem: 52150 +Epoch: [21] [1480/7269] eta: 0:41:54 lr: 0.000005 loss: 0.5486 (0.5216) time: 0.4791 data: 0.0001 max mem: 52150 +Epoch: [21] [1490/7269] eta: 0:41:47 lr: 0.000005 loss: 0.5444 (0.5216) time: 0.4175 data: 0.0002 max mem: 52150 +Epoch: [21] [1500/7269] eta: 0:41:45 lr: 0.000005 loss: 0.5446 (0.5217) time: 0.4332 data: 0.0002 max mem: 52150 +Epoch: [21] [1510/7269] eta: 0:41:43 lr: 0.000005 loss: 0.5232 (0.5216) time: 0.4939 data: 0.0002 max mem: 52150 +Epoch: [21] [1520/7269] eta: 0:41:36 lr: 0.000005 loss: 0.5446 (0.5218) time: 0.4324 data: 0.0002 max mem: 52150 +Epoch: [21] [1530/7269] eta: 0:41:29 lr: 0.000005 loss: 0.5449 (0.5219) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [21] [1540/7269] eta: 0:41:26 lr: 0.000005 loss: 0.5049 (0.5219) time: 0.4172 data: 0.0002 max mem: 52150 +Epoch: [21] [1550/7269] eta: 0:41:23 lr: 0.000005 loss: 0.5095 (0.5217) time: 0.4627 data: 0.0002 max mem: 52150 +Epoch: [21] [1560/7269] eta: 0:41:16 lr: 0.000005 loss: 0.5212 (0.5218) time: 0.4179 data: 0.0002 max mem: 52150 +Epoch: [21] [1570/7269] eta: 0:41:10 lr: 0.000005 loss: 0.5356 (0.5220) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [21] [1580/7269] eta: 0:41:03 lr: 0.000005 loss: 0.5354 (0.5218) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [21] [1590/7269] eta: 0:41:00 lr: 0.000005 loss: 0.5084 (0.5217) time: 0.4234 data: 0.0002 max mem: 52150 +Epoch: [21] [1600/7269] eta: 0:40:54 lr: 0.000005 loss: 0.5084 (0.5217) time: 0.4235 data: 0.0001 max mem: 52150 +Epoch: [21] [1610/7269] eta: 0:40:52 lr: 0.000005 loss: 0.5060 (0.5218) time: 0.4340 data: 0.0001 max mem: 52150 +Epoch: [21] [1620/7269] eta: 0:40:45 lr: 0.000005 loss: 0.5133 (0.5218) time: 0.4344 data: 0.0001 max mem: 52150 +Epoch: [21] [1630/7269] eta: 0:40:46 lr: 0.000005 loss: 0.5126 (0.5217) time: 0.4821 data: 0.0002 max mem: 52150 +Epoch: [21] [1640/7269] eta: 0:40:43 lr: 0.000005 loss: 0.5122 (0.5217) time: 0.5276 data: 0.0002 max mem: 52150 +Epoch: [21] [1650/7269] eta: 0:40:37 lr: 0.000005 loss: 0.5386 (0.5218) time: 0.4179 data: 0.0001 max mem: 52150 +Epoch: [21] [1660/7269] eta: 0:40:38 lr: 0.000005 loss: 0.5334 (0.5217) time: 0.4943 data: 0.0002 max mem: 52150 +Epoch: [21] [1670/7269] eta: 0:40:36 lr: 0.000005 loss: 0.5425 (0.5219) time: 0.5546 data: 0.0001 max mem: 52150 +Epoch: [21] [1680/7269] eta: 0:40:30 lr: 0.000005 loss: 0.5592 (0.5220) time: 0.4326 data: 0.0001 max mem: 52150 +Epoch: [21] [1690/7269] eta: 0:40:30 lr: 0.000005 loss: 0.5531 (0.5222) time: 0.4815 data: 0.0001 max mem: 52150 +Epoch: [21] [1700/7269] eta: 0:40:24 lr: 0.000005 loss: 0.5451 (0.5222) time: 0.4814 data: 0.0002 max mem: 52150 +Epoch: [21] [1710/7269] eta: 0:40:21 lr: 0.000005 loss: 0.5048 (0.5219) time: 0.4326 data: 0.0002 max mem: 52150 +Epoch: [21] [1720/7269] eta: 0:40:15 lr: 0.000005 loss: 0.4997 (0.5218) time: 0.4325 data: 0.0001 max mem: 52150 +Epoch: [21] [1730/7269] eta: 0:40:09 lr: 0.000005 loss: 0.5273 (0.5219) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [21] [1740/7269] eta: 0:40:02 lr: 0.000005 loss: 0.4979 (0.5217) time: 0.3728 data: 0.0002 max mem: 52150 +Epoch: [21] [1750/7269] eta: 0:40:02 lr: 0.000005 loss: 0.4926 (0.5217) time: 0.4617 data: 0.0002 max mem: 52150 +Epoch: [21] [1760/7269] eta: 0:39:55 lr: 0.000005 loss: 0.5147 (0.5217) time: 0.4609 data: 0.0002 max mem: 52150 +Epoch: [21] [1770/7269] eta: 0:39:49 lr: 0.000005 loss: 0.5059 (0.5215) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [21] [1780/7269] eta: 0:39:49 lr: 0.000005 loss: 0.5386 (0.5217) time: 0.4677 data: 0.0002 max mem: 52150 +Epoch: [21] [1790/7269] eta: 0:39:42 lr: 0.000005 loss: 0.5538 (0.5218) time: 0.4677 data: 0.0002 max mem: 52150 +Epoch: [21] [1800/7269] eta: 0:39:42 lr: 0.000005 loss: 0.5209 (0.5217) time: 0.4628 data: 0.0002 max mem: 52150 +Epoch: [21] [1810/7269] eta: 0:39:35 lr: 0.000005 loss: 0.5132 (0.5217) time: 0.4638 data: 0.0001 max mem: 52150 +Epoch: [21] [1820/7269] eta: 0:39:33 lr: 0.000005 loss: 0.5587 (0.5219) time: 0.4354 data: 0.0001 max mem: 52150 +Epoch: [21] [1830/7269] eta: 0:39:27 lr: 0.000005 loss: 0.5472 (0.5219) time: 0.4346 data: 0.0001 max mem: 52150 +Epoch: [21] [1840/7269] eta: 0:39:20 lr: 0.000005 loss: 0.5383 (0.5218) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [21] [1850/7269] eta: 0:39:14 lr: 0.000005 loss: 0.5345 (0.5218) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [21] [1860/7269] eta: 0:39:08 lr: 0.000005 loss: 0.5066 (0.5217) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [21] [1870/7269] eta: 0:39:02 lr: 0.000005 loss: 0.5138 (0.5219) time: 0.3707 data: 0.0002 max mem: 52150 +Epoch: [21] [1880/7269] eta: 0:38:56 lr: 0.000005 loss: 0.5404 (0.5217) time: 0.3708 data: 0.0002 max mem: 52150 +Epoch: [21] [1890/7269] eta: 0:38:53 lr: 0.000005 loss: 0.5157 (0.5216) time: 0.4361 data: 0.0001 max mem: 52150 +Epoch: [21] [1900/7269] eta: 0:38:47 lr: 0.000005 loss: 0.5122 (0.5215) time: 0.4371 data: 0.0001 max mem: 52150 +Epoch: [21] [1910/7269] eta: 0:38:41 lr: 0.000005 loss: 0.4946 (0.5216) time: 0.3730 data: 0.0002 max mem: 52150 +Epoch: [21] [1920/7269] eta: 0:38:38 lr: 0.000005 loss: 0.4917 (0.5214) time: 0.4192 data: 0.0002 max mem: 52150 +Epoch: [21] [1930/7269] eta: 0:38:34 lr: 0.000005 loss: 0.5033 (0.5214) time: 0.4628 data: 0.0002 max mem: 52150 +Epoch: [21] [1940/7269] eta: 0:38:31 lr: 0.000005 loss: 0.5345 (0.5214) time: 0.4656 data: 0.0001 max mem: 52150 +Epoch: [21] [1950/7269] eta: 0:38:31 lr: 0.000005 loss: 0.5365 (0.5215) time: 0.5313 data: 0.0001 max mem: 52150 +Epoch: [21] [1960/7269] eta: 0:38:27 lr: 0.000005 loss: 0.5016 (0.5213) time: 0.5280 data: 0.0002 max mem: 52150 +Epoch: [21] [1970/7269] eta: 0:38:24 lr: 0.000005 loss: 0.4900 (0.5212) time: 0.4776 data: 0.0001 max mem: 52150 +Epoch: [21] [1980/7269] eta: 0:38:18 lr: 0.000005 loss: 0.5353 (0.5214) time: 0.4311 data: 0.0001 max mem: 52150 +Epoch: [21] [1990/7269] eta: 0:38:12 lr: 0.000005 loss: 0.5353 (0.5214) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [21] [2000/7269] eta: 0:38:06 lr: 0.000005 loss: 0.5203 (0.5214) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [21] [2010/7269] eta: 0:38:00 lr: 0.000005 loss: 0.5223 (0.5213) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [21] [2020/7269] eta: 0:37:54 lr: 0.000005 loss: 0.5044 (0.5212) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [21] [2030/7269] eta: 0:37:51 lr: 0.000005 loss: 0.4952 (0.5211) time: 0.4186 data: 0.0002 max mem: 52150 +Epoch: [21] [2040/7269] eta: 0:37:45 lr: 0.000005 loss: 0.5173 (0.5211) time: 0.4187 data: 0.0002 max mem: 52150 +Epoch: [21] [2050/7269] eta: 0:37:39 lr: 0.000005 loss: 0.5224 (0.5212) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [21] [2060/7269] eta: 0:37:36 lr: 0.000005 loss: 0.5514 (0.5213) time: 0.4194 data: 0.0002 max mem: 52150 +Epoch: [21] [2070/7269] eta: 0:37:30 lr: 0.000005 loss: 0.5369 (0.5214) time: 0.4190 data: 0.0002 max mem: 52150 +Epoch: [21] [2080/7269] eta: 0:37:24 lr: 0.000005 loss: 0.5378 (0.5215) time: 0.3714 data: 0.0001 max mem: 52150 +Epoch: [21] [2090/7269] eta: 0:37:23 lr: 0.000005 loss: 0.5562 (0.5216) time: 0.4758 data: 0.0001 max mem: 52150 +Epoch: [21] [2100/7269] eta: 0:37:23 lr: 0.000005 loss: 0.5309 (0.5216) time: 0.5842 data: 0.0001 max mem: 52150 +Epoch: [21] [2110/7269] eta: 0:37:17 lr: 0.000005 loss: 0.5309 (0.5218) time: 0.4803 data: 0.0001 max mem: 52150 +Epoch: [21] [2120/7269] eta: 0:37:13 lr: 0.000005 loss: 0.5155 (0.5216) time: 0.4176 data: 0.0001 max mem: 52150 +Epoch: [21] [2130/7269] eta: 0:37:07 lr: 0.000005 loss: 0.4883 (0.5216) time: 0.4170 data: 0.0001 max mem: 52150 +Epoch: [21] [2140/7269] eta: 0:37:02 lr: 0.000005 loss: 0.5120 (0.5216) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [21] [2150/7269] eta: 0:36:56 lr: 0.000005 loss: 0.5130 (0.5216) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [21] [2160/7269] eta: 0:36:50 lr: 0.000005 loss: 0.5294 (0.5216) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [21] [2170/7269] eta: 0:36:50 lr: 0.000005 loss: 0.5440 (0.5218) time: 0.4917 data: 0.0002 max mem: 52150 +Epoch: [21] [2180/7269] eta: 0:36:44 lr: 0.000005 loss: 0.5368 (0.5217) time: 0.4923 data: 0.0001 max mem: 52150 +Epoch: [21] [2190/7269] eta: 0:36:41 lr: 0.000005 loss: 0.4985 (0.5214) time: 0.4344 data: 0.0002 max mem: 52150 +Epoch: [21] [2200/7269] eta: 0:36:36 lr: 0.000005 loss: 0.5175 (0.5214) time: 0.4339 data: 0.0002 max mem: 52150 +Epoch: [21] [2210/7269] eta: 0:36:32 lr: 0.000005 loss: 0.5177 (0.5214) time: 0.4169 data: 0.0002 max mem: 52150 +Epoch: [21] [2220/7269] eta: 0:36:26 lr: 0.000005 loss: 0.5392 (0.5215) time: 0.4175 data: 0.0002 max mem: 52150 +Epoch: [21] [2230/7269] eta: 0:36:20 lr: 0.000005 loss: 0.4655 (0.5212) time: 0.3722 data: 0.0001 max mem: 52150 +Epoch: [21] [2240/7269] eta: 0:36:17 lr: 0.000005 loss: 0.4538 (0.5212) time: 0.4327 data: 0.0001 max mem: 52150 +Epoch: [21] [2250/7269] eta: 0:36:12 lr: 0.000005 loss: 0.5251 (0.5210) time: 0.4325 data: 0.0001 max mem: 52150 +Epoch: [21] [2260/7269] eta: 0:36:08 lr: 0.000005 loss: 0.4674 (0.5209) time: 0.4200 data: 0.0001 max mem: 52150 +Epoch: [21] [2270/7269] eta: 0:36:07 lr: 0.000005 loss: 0.5395 (0.5210) time: 0.5300 data: 0.0001 max mem: 52150 +Epoch: [21] [2280/7269] eta: 0:36:04 lr: 0.000005 loss: 0.5395 (0.5209) time: 0.5273 data: 0.0002 max mem: 52150 +Epoch: [21] [2290/7269] eta: 0:36:01 lr: 0.000005 loss: 0.5374 (0.5210) time: 0.4776 data: 0.0002 max mem: 52150 +Epoch: [21] [2300/7269] eta: 0:35:58 lr: 0.000005 loss: 0.5374 (0.5210) time: 0.4960 data: 0.0002 max mem: 52150 +Epoch: [21] [2310/7269] eta: 0:35:52 lr: 0.000005 loss: 0.5409 (0.5211) time: 0.4362 data: 0.0001 max mem: 52150 +Epoch: [21] [2320/7269] eta: 0:35:48 lr: 0.000005 loss: 0.5739 (0.5212) time: 0.4171 data: 0.0001 max mem: 52150 +Epoch: [21] [2330/7269] eta: 0:35:43 lr: 0.000005 loss: 0.5870 (0.5214) time: 0.4168 data: 0.0002 max mem: 52150 +Epoch: [21] [2340/7269] eta: 0:35:41 lr: 0.000005 loss: 0.5453 (0.5213) time: 0.4664 data: 0.0002 max mem: 52150 +Epoch: [21] [2350/7269] eta: 0:35:37 lr: 0.000005 loss: 0.5332 (0.5214) time: 0.5119 data: 0.0002 max mem: 52150 +Epoch: [21] [2360/7269] eta: 0:35:34 lr: 0.000005 loss: 0.5304 (0.5215) time: 0.4660 data: 0.0002 max mem: 52150 +Epoch: [21] [2370/7269] eta: 0:35:28 lr: 0.000005 loss: 0.5199 (0.5214) time: 0.4203 data: 0.0002 max mem: 52150 +Epoch: [21] [2380/7269] eta: 0:35:22 lr: 0.000005 loss: 0.5442 (0.5214) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [21] [2390/7269] eta: 0:35:17 lr: 0.000005 loss: 0.5443 (0.5216) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [21] [2400/7269] eta: 0:35:11 lr: 0.000005 loss: 0.5444 (0.5215) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [21] [2410/7269] eta: 0:35:05 lr: 0.000005 loss: 0.5427 (0.5216) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [21] [2420/7269] eta: 0:35:00 lr: 0.000005 loss: 0.5286 (0.5217) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [21] [2430/7269] eta: 0:34:54 lr: 0.000005 loss: 0.5064 (0.5215) time: 0.3713 data: 0.0001 max mem: 52150 +Epoch: [21] [2440/7269] eta: 0:34:49 lr: 0.000005 loss: 0.4873 (0.5215) time: 0.3714 data: 0.0001 max mem: 52150 +Epoch: [21] [2450/7269] eta: 0:34:43 lr: 0.000005 loss: 0.5155 (0.5215) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [21] [2460/7269] eta: 0:34:40 lr: 0.000005 loss: 0.5112 (0.5215) time: 0.4170 data: 0.0002 max mem: 52150 +Epoch: [21] [2470/7269] eta: 0:34:34 lr: 0.000005 loss: 0.5154 (0.5214) time: 0.4169 data: 0.0001 max mem: 52150 +Epoch: [21] [2480/7269] eta: 0:34:29 lr: 0.000005 loss: 0.5154 (0.5214) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [21] [2490/7269] eta: 0:34:25 lr: 0.000005 loss: 0.5164 (0.5213) time: 0.4180 data: 0.0001 max mem: 52150 +Epoch: [21] [2500/7269] eta: 0:34:26 lr: 0.000005 loss: 0.5392 (0.5214) time: 0.5902 data: 0.0001 max mem: 52150 +Epoch: [21] [2510/7269] eta: 0:34:23 lr: 0.000005 loss: 0.5452 (0.5214) time: 0.6063 data: 0.0001 max mem: 52150 +Epoch: [21] [2520/7269] eta: 0:34:19 lr: 0.000005 loss: 0.5164 (0.5214) time: 0.4807 data: 0.0001 max mem: 52150 +Epoch: [21] [2530/7269] eta: 0:34:14 lr: 0.000005 loss: 0.5472 (0.5216) time: 0.4184 data: 0.0001 max mem: 52150 +Epoch: [21] [2540/7269] eta: 0:34:10 lr: 0.000005 loss: 0.5505 (0.5217) time: 0.4351 data: 0.0001 max mem: 52150 +Epoch: [21] [2550/7269] eta: 0:34:05 lr: 0.000005 loss: 0.5406 (0.5216) time: 0.4350 data: 0.0002 max mem: 52150 +Epoch: [21] [2560/7269] eta: 0:34:03 lr: 0.000005 loss: 0.5419 (0.5218) time: 0.4772 data: 0.0002 max mem: 52150 +Epoch: [21] [2570/7269] eta: 0:33:58 lr: 0.000005 loss: 0.5355 (0.5217) time: 0.4770 data: 0.0002 max mem: 52150 +Epoch: [21] [2580/7269] eta: 0:33:52 lr: 0.000005 loss: 0.5358 (0.5219) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [21] [2590/7269] eta: 0:33:49 lr: 0.000005 loss: 0.5358 (0.5218) time: 0.4323 data: 0.0002 max mem: 52150 +Epoch: [21] [2600/7269] eta: 0:33:44 lr: 0.000005 loss: 0.5020 (0.5218) time: 0.4319 data: 0.0002 max mem: 52150 +Epoch: [21] [2610/7269] eta: 0:33:38 lr: 0.000005 loss: 0.5379 (0.5218) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [21] [2620/7269] eta: 0:33:36 lr: 0.000005 loss: 0.5381 (0.5218) time: 0.4622 data: 0.0001 max mem: 52150 +Epoch: [21] [2630/7269] eta: 0:33:31 lr: 0.000005 loss: 0.5120 (0.5217) time: 0.4626 data: 0.0001 max mem: 52150 +Epoch: [21] [2640/7269] eta: 0:33:25 lr: 0.000005 loss: 0.5016 (0.5217) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [21] [2650/7269] eta: 0:33:21 lr: 0.000005 loss: 0.5272 (0.5217) time: 0.4202 data: 0.0002 max mem: 52150 +Epoch: [21] [2660/7269] eta: 0:33:18 lr: 0.000005 loss: 0.5084 (0.5216) time: 0.4645 data: 0.0001 max mem: 52150 +Epoch: [21] [2670/7269] eta: 0:33:12 lr: 0.000005 loss: 0.5317 (0.5218) time: 0.4158 data: 0.0002 max mem: 52150 +Epoch: [21] [2680/7269] eta: 0:33:09 lr: 0.000005 loss: 0.5767 (0.5219) time: 0.4321 data: 0.0002 max mem: 52150 +Epoch: [21] [2690/7269] eta: 0:33:08 lr: 0.000005 loss: 0.5498 (0.5218) time: 0.5540 data: 0.0001 max mem: 52150 +Epoch: [21] [2700/7269] eta: 0:33:02 lr: 0.000005 loss: 0.5295 (0.5220) time: 0.4934 data: 0.0001 max mem: 52150 +Epoch: [21] [2710/7269] eta: 0:32:57 lr: 0.000005 loss: 0.5309 (0.5219) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [21] [2720/7269] eta: 0:32:51 lr: 0.000005 loss: 0.5467 (0.5220) time: 0.3714 data: 0.0001 max mem: 52150 +Epoch: [21] [2730/7269] eta: 0:32:46 lr: 0.000005 loss: 0.5507 (0.5221) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [21] [2740/7269] eta: 0:32:42 lr: 0.000005 loss: 0.5037 (0.5220) time: 0.4174 data: 0.0001 max mem: 52150 +Epoch: [21] [2750/7269] eta: 0:32:37 lr: 0.000005 loss: 0.5518 (0.5222) time: 0.4171 data: 0.0001 max mem: 52150 +Epoch: [21] [2760/7269] eta: 0:32:34 lr: 0.000005 loss: 0.5518 (0.5222) time: 0.4319 data: 0.0001 max mem: 52150 +Epoch: [21] [2770/7269] eta: 0:32:32 lr: 0.000005 loss: 0.5289 (0.5222) time: 0.5600 data: 0.0001 max mem: 52150 +Epoch: [21] [2780/7269] eta: 0:32:27 lr: 0.000005 loss: 0.5260 (0.5222) time: 0.5000 data: 0.0001 max mem: 52150 +Epoch: [21] [2790/7269] eta: 0:32:22 lr: 0.000005 loss: 0.5260 (0.5222) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [21] [2800/7269] eta: 0:32:19 lr: 0.000005 loss: 0.5395 (0.5222) time: 0.4669 data: 0.0001 max mem: 52150 +Epoch: [21] [2810/7269] eta: 0:32:15 lr: 0.000005 loss: 0.5319 (0.5222) time: 0.5103 data: 0.0001 max mem: 52150 +Epoch: [21] [2820/7269] eta: 0:32:12 lr: 0.000005 loss: 0.5099 (0.5222) time: 0.4764 data: 0.0001 max mem: 52150 +Epoch: [21] [2830/7269] eta: 0:32:07 lr: 0.000005 loss: 0.5099 (0.5220) time: 0.4329 data: 0.0001 max mem: 52150 +Epoch: [21] [2840/7269] eta: 0:32:01 lr: 0.000005 loss: 0.5162 (0.5221) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [21] [2850/7269] eta: 0:31:56 lr: 0.000005 loss: 0.5490 (0.5221) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [21] [2860/7269] eta: 0:31:51 lr: 0.000005 loss: 0.5481 (0.5222) time: 0.3712 data: 0.0002 max mem: 52150 +Epoch: [21] [2870/7269] eta: 0:31:46 lr: 0.000005 loss: 0.5020 (0.5221) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [21] [2880/7269] eta: 0:31:40 lr: 0.000005 loss: 0.4606 (0.5219) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [21] [2890/7269] eta: 0:31:35 lr: 0.000005 loss: 0.5307 (0.5220) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [21] [2900/7269] eta: 0:31:30 lr: 0.000005 loss: 0.5191 (0.5219) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [21] [2910/7269] eta: 0:31:24 lr: 0.000005 loss: 0.5034 (0.5218) time: 0.3712 data: 0.0001 max mem: 52150 +Epoch: [21] [2920/7269] eta: 0:31:19 lr: 0.000005 loss: 0.5039 (0.5218) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [21] [2930/7269] eta: 0:31:14 lr: 0.000005 loss: 0.5039 (0.5217) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [21] [2940/7269] eta: 0:31:09 lr: 0.000005 loss: 0.5498 (0.5218) time: 0.3714 data: 0.0001 max mem: 52150 +Epoch: [21] [2950/7269] eta: 0:31:05 lr: 0.000005 loss: 0.5364 (0.5217) time: 0.4202 data: 0.0001 max mem: 52150 +Epoch: [21] [2960/7269] eta: 0:31:02 lr: 0.000005 loss: 0.5364 (0.5218) time: 0.4855 data: 0.0001 max mem: 52150 +Epoch: [21] [2970/7269] eta: 0:30:57 lr: 0.000005 loss: 0.5264 (0.5218) time: 0.4372 data: 0.0001 max mem: 52150 +Epoch: [21] [2980/7269] eta: 0:30:53 lr: 0.000005 loss: 0.5137 (0.5218) time: 0.4340 data: 0.0002 max mem: 52150 +Epoch: [21] [2990/7269] eta: 0:30:48 lr: 0.000005 loss: 0.5149 (0.5218) time: 0.4336 data: 0.0002 max mem: 52150 +Epoch: [21] [3000/7269] eta: 0:30:45 lr: 0.000005 loss: 0.5487 (0.5218) time: 0.4651 data: 0.0002 max mem: 52150 +Epoch: [21] [3010/7269] eta: 0:30:40 lr: 0.000005 loss: 0.5152 (0.5218) time: 0.4648 data: 0.0001 max mem: 52150 +Epoch: [21] [3020/7269] eta: 0:30:35 lr: 0.000005 loss: 0.5167 (0.5217) time: 0.3711 data: 0.0002 max mem: 52150 +Epoch: [21] [3030/7269] eta: 0:30:31 lr: 0.000005 loss: 0.5511 (0.5219) time: 0.4166 data: 0.0002 max mem: 52150 +Epoch: [21] [3040/7269] eta: 0:30:28 lr: 0.000005 loss: 0.5434 (0.5219) time: 0.4777 data: 0.0001 max mem: 52150 +Epoch: [21] [3050/7269] eta: 0:30:23 lr: 0.000005 loss: 0.5312 (0.5218) time: 0.4320 data: 0.0001 max mem: 52150 +Epoch: [21] [3060/7269] eta: 0:30:17 lr: 0.000005 loss: 0.5392 (0.5219) time: 0.3708 data: 0.0002 max mem: 52150 +Epoch: [21] [3070/7269] eta: 0:30:14 lr: 0.000005 loss: 0.5243 (0.5218) time: 0.4180 data: 0.0002 max mem: 52150 +Epoch: [21] [3080/7269] eta: 0:30:10 lr: 0.000005 loss: 0.5243 (0.5218) time: 0.4640 data: 0.0001 max mem: 52150 +Epoch: [21] [3090/7269] eta: 0:30:04 lr: 0.000005 loss: 0.5024 (0.5217) time: 0.4173 data: 0.0001 max mem: 52150 +Epoch: [21] [3100/7269] eta: 0:30:02 lr: 0.000005 loss: 0.5276 (0.5218) time: 0.4663 data: 0.0001 max mem: 52150 +Epoch: [21] [3110/7269] eta: 0:29:58 lr: 0.000005 loss: 0.5378 (0.5217) time: 0.5112 data: 0.0002 max mem: 52150 +Epoch: [21] [3120/7269] eta: 0:29:54 lr: 0.000005 loss: 0.4871 (0.5216) time: 0.4627 data: 0.0002 max mem: 52150 +Epoch: [21] [3130/7269] eta: 0:29:49 lr: 0.000005 loss: 0.4796 (0.5216) time: 0.4175 data: 0.0001 max mem: 52150 +Epoch: [21] [3140/7269] eta: 0:29:44 lr: 0.000005 loss: 0.4796 (0.5215) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [21] [3150/7269] eta: 0:29:42 lr: 0.000005 loss: 0.5261 (0.5216) time: 0.4828 data: 0.0002 max mem: 52150 +Epoch: [21] [3160/7269] eta: 0:29:38 lr: 0.000005 loss: 0.5544 (0.5217) time: 0.5282 data: 0.0002 max mem: 52150 +Epoch: [21] [3170/7269] eta: 0:29:33 lr: 0.000005 loss: 0.5392 (0.5218) time: 0.4173 data: 0.0002 max mem: 52150 +Epoch: [21] [3180/7269] eta: 0:29:27 lr: 0.000005 loss: 0.5485 (0.5219) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [21] [3190/7269] eta: 0:29:25 lr: 0.000005 loss: 0.5357 (0.5218) time: 0.4925 data: 0.0001 max mem: 52150 +Epoch: [21] [3200/7269] eta: 0:29:20 lr: 0.000005 loss: 0.5228 (0.5218) time: 0.4931 data: 0.0002 max mem: 52150 +Epoch: [21] [3210/7269] eta: 0:29:15 lr: 0.000005 loss: 0.5286 (0.5219) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [21] [3220/7269] eta: 0:29:10 lr: 0.000005 loss: 0.5172 (0.5219) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [21] [3230/7269] eta: 0:29:07 lr: 0.000005 loss: 0.5171 (0.5219) time: 0.4314 data: 0.0002 max mem: 52150 +Epoch: [21] [3240/7269] eta: 0:29:01 lr: 0.000005 loss: 0.5133 (0.5219) time: 0.4311 data: 0.0002 max mem: 52150 +Epoch: [21] [3250/7269] eta: 0:28:56 lr: 0.000005 loss: 0.5134 (0.5219) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [21] [3260/7269] eta: 0:28:51 lr: 0.000005 loss: 0.5430 (0.5219) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [21] [3270/7269] eta: 0:28:46 lr: 0.000005 loss: 0.5430 (0.5219) time: 0.3711 data: 0.0001 max mem: 52150 +Epoch: [21] [3280/7269] eta: 0:28:42 lr: 0.000005 loss: 0.5314 (0.5219) time: 0.4176 data: 0.0002 max mem: 52150 +Epoch: [21] [3290/7269] eta: 0:28:37 lr: 0.000005 loss: 0.5127 (0.5219) time: 0.4177 data: 0.0001 max mem: 52150 +Epoch: [21] [3300/7269] eta: 0:28:32 lr: 0.000005 loss: 0.5081 (0.5218) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [21] [3310/7269] eta: 0:28:27 lr: 0.000005 loss: 0.5122 (0.5219) time: 0.3714 data: 0.0001 max mem: 52150 +Epoch: [21] [3320/7269] eta: 0:28:22 lr: 0.000005 loss: 0.5374 (0.5219) time: 0.3710 data: 0.0001 max mem: 52150 +Epoch: [21] [3330/7269] eta: 0:28:20 lr: 0.000005 loss: 0.5501 (0.5220) time: 0.4911 data: 0.0002 max mem: 52150 +Epoch: [21] [3340/7269] eta: 0:28:15 lr: 0.000005 loss: 0.5524 (0.5221) time: 0.4917 data: 0.0002 max mem: 52150 +Epoch: [21] [3350/7269] eta: 0:28:10 lr: 0.000005 loss: 0.5410 (0.5221) time: 0.3723 data: 0.0001 max mem: 52150 +Epoch: [21] [3360/7269] eta: 0:28:06 lr: 0.000005 loss: 0.5123 (0.5220) time: 0.4232 data: 0.0001 max mem: 52150 +Epoch: [21] [3370/7269] eta: 0:28:03 lr: 0.000005 loss: 0.4936 (0.5219) time: 0.4830 data: 0.0001 max mem: 52150 +Epoch: [21] [3380/7269] eta: 0:28:02 lr: 0.000005 loss: 0.4887 (0.5218) time: 0.6379 data: 0.0001 max mem: 52150 +Epoch: [21] [3390/7269] eta: 0:27:59 lr: 0.000005 loss: 0.5438 (0.5220) time: 0.6665 data: 0.0001 max mem: 52150 +Epoch: [21] [3400/7269] eta: 0:27:54 lr: 0.000005 loss: 0.5692 (0.5220) time: 0.4608 data: 0.0001 max mem: 52150 +Epoch: [21] [3410/7269] eta: 0:27:50 lr: 0.000005 loss: 0.5288 (0.5221) time: 0.4193 data: 0.0001 max mem: 52150 +Epoch: [21] [3420/7269] eta: 0:27:45 lr: 0.000005 loss: 0.5393 (0.5221) time: 0.4187 data: 0.0001 max mem: 52150 +Epoch: [21] [3430/7269] eta: 0:27:40 lr: 0.000005 loss: 0.5393 (0.5222) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [21] [3440/7269] eta: 0:27:35 lr: 0.000005 loss: 0.5139 (0.5222) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [21] [3450/7269] eta: 0:27:30 lr: 0.000005 loss: 0.5041 (0.5221) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [21] [3460/7269] eta: 0:27:27 lr: 0.000005 loss: 0.5152 (0.5220) time: 0.4349 data: 0.0001 max mem: 52150 +Epoch: [21] [3470/7269] eta: 0:27:23 lr: 0.000005 loss: 0.5169 (0.5220) time: 0.4792 data: 0.0001 max mem: 52150 +Epoch: [21] [3480/7269] eta: 0:27:19 lr: 0.000005 loss: 0.5347 (0.5220) time: 0.4771 data: 0.0001 max mem: 52150 +Epoch: [21] [3490/7269] eta: 0:27:16 lr: 0.000005 loss: 0.5377 (0.5220) time: 0.5236 data: 0.0001 max mem: 52150 +Epoch: [21] [3500/7269] eta: 0:27:12 lr: 0.000005 loss: 0.5381 (0.5221) time: 0.5072 data: 0.0001 max mem: 52150 +Epoch: [21] [3510/7269] eta: 0:27:07 lr: 0.000005 loss: 0.5294 (0.5220) time: 0.4163 data: 0.0001 max mem: 52150 +Epoch: [21] [3520/7269] eta: 0:27:02 lr: 0.000005 loss: 0.5173 (0.5220) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [21] [3530/7269] eta: 0:26:57 lr: 0.000005 loss: 0.5268 (0.5221) time: 0.3721 data: 0.0001 max mem: 52150 +Epoch: [21] [3540/7269] eta: 0:26:52 lr: 0.000005 loss: 0.5369 (0.5221) time: 0.3733 data: 0.0001 max mem: 52150 +Epoch: [21] [3550/7269] eta: 0:26:49 lr: 0.000005 loss: 0.5369 (0.5221) time: 0.4709 data: 0.0001 max mem: 52150 +Epoch: [21] [3560/7269] eta: 0:26:44 lr: 0.000005 loss: 0.5092 (0.5220) time: 0.4700 data: 0.0001 max mem: 52150 +Epoch: [21] [3570/7269] eta: 0:26:40 lr: 0.000005 loss: 0.5164 (0.5221) time: 0.4177 data: 0.0001 max mem: 52150 +Epoch: [21] [3580/7269] eta: 0:26:35 lr: 0.000005 loss: 0.5099 (0.5219) time: 0.4171 data: 0.0001 max mem: 52150 +Epoch: [21] [3590/7269] eta: 0:26:30 lr: 0.000005 loss: 0.4964 (0.5219) time: 0.3713 data: 0.0001 max mem: 52150 +Epoch: [21] [3600/7269] eta: 0:26:26 lr: 0.000005 loss: 0.5418 (0.5219) time: 0.4173 data: 0.0002 max mem: 52150 +Epoch: [21] [3610/7269] eta: 0:26:21 lr: 0.000005 loss: 0.5494 (0.5220) time: 0.4171 data: 0.0002 max mem: 52150 +Epoch: [21] [3620/7269] eta: 0:26:16 lr: 0.000005 loss: 0.5494 (0.5220) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [21] [3630/7269] eta: 0:26:11 lr: 0.000005 loss: 0.5457 (0.5220) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [21] [3640/7269] eta: 0:26:07 lr: 0.000005 loss: 0.5299 (0.5220) time: 0.4172 data: 0.0001 max mem: 52150 +Epoch: [21] [3650/7269] eta: 0:26:04 lr: 0.000005 loss: 0.5316 (0.5221) time: 0.4792 data: 0.0001 max mem: 52150 +Epoch: [21] [3660/7269] eta: 0:26:00 lr: 0.000005 loss: 0.5430 (0.5221) time: 0.4812 data: 0.0001 max mem: 52150 +Epoch: [21] [3670/7269] eta: 0:25:55 lr: 0.000005 loss: 0.5139 (0.5221) time: 0.4192 data: 0.0002 max mem: 52150 +Epoch: [21] [3680/7269] eta: 0:25:50 lr: 0.000005 loss: 0.5407 (0.5222) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [21] [3690/7269] eta: 0:25:45 lr: 0.000005 loss: 0.5419 (0.5221) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [21] [3700/7269] eta: 0:25:41 lr: 0.000005 loss: 0.5128 (0.5221) time: 0.4175 data: 0.0002 max mem: 52150 +Epoch: [21] [3710/7269] eta: 0:25:36 lr: 0.000005 loss: 0.5501 (0.5221) time: 0.4170 data: 0.0001 max mem: 52150 +Epoch: [21] [3720/7269] eta: 0:25:33 lr: 0.000005 loss: 0.5382 (0.5220) time: 0.4602 data: 0.0002 max mem: 52150 +Epoch: [21] [3730/7269] eta: 0:25:28 lr: 0.000005 loss: 0.5305 (0.5221) time: 0.4600 data: 0.0002 max mem: 52150 +Epoch: [21] [3740/7269] eta: 0:25:24 lr: 0.000005 loss: 0.5612 (0.5221) time: 0.4315 data: 0.0001 max mem: 52150 +Epoch: [21] [3750/7269] eta: 0:25:19 lr: 0.000005 loss: 0.5358 (0.5221) time: 0.4314 data: 0.0002 max mem: 52150 +Epoch: [21] [3760/7269] eta: 0:25:14 lr: 0.000005 loss: 0.5054 (0.5220) time: 0.3711 data: 0.0001 max mem: 52150 +Epoch: [21] [3770/7269] eta: 0:25:10 lr: 0.000005 loss: 0.5086 (0.5221) time: 0.4189 data: 0.0001 max mem: 52150 +Epoch: [21] [3780/7269] eta: 0:25:06 lr: 0.000005 loss: 0.5164 (0.5221) time: 0.4193 data: 0.0001 max mem: 52150 +Epoch: [21] [3790/7269] eta: 0:25:02 lr: 0.000005 loss: 0.5139 (0.5220) time: 0.4170 data: 0.0001 max mem: 52150 +Epoch: [21] [3800/7269] eta: 0:24:59 lr: 0.000005 loss: 0.5645 (0.5222) time: 0.5378 data: 0.0001 max mem: 52150 +Epoch: [21] [3810/7269] eta: 0:24:54 lr: 0.000005 loss: 0.5701 (0.5223) time: 0.4923 data: 0.0001 max mem: 52150 +Epoch: [21] [3820/7269] eta: 0:24:49 lr: 0.000005 loss: 0.5484 (0.5223) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [21] [3830/7269] eta: 0:24:44 lr: 0.000005 loss: 0.5325 (0.5223) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [21] [3840/7269] eta: 0:24:41 lr: 0.000005 loss: 0.5110 (0.5222) time: 0.4312 data: 0.0002 max mem: 52150 +Epoch: [21] [3850/7269] eta: 0:24:36 lr: 0.000005 loss: 0.5020 (0.5221) time: 0.4309 data: 0.0002 max mem: 52150 +Epoch: [21] [3860/7269] eta: 0:24:33 lr: 0.000005 loss: 0.5287 (0.5222) time: 0.4759 data: 0.0001 max mem: 52150 +Epoch: [21] [3870/7269] eta: 0:24:30 lr: 0.000005 loss: 0.5657 (0.5222) time: 0.5831 data: 0.0001 max mem: 52150 +Epoch: [21] [3880/7269] eta: 0:24:25 lr: 0.000005 loss: 0.5363 (0.5223) time: 0.4788 data: 0.0001 max mem: 52150 +Epoch: [21] [3890/7269] eta: 0:24:21 lr: 0.000005 loss: 0.5039 (0.5222) time: 0.4186 data: 0.0002 max mem: 52150 +Epoch: [21] [3900/7269] eta: 0:24:17 lr: 0.000005 loss: 0.4832 (0.5221) time: 0.4680 data: 0.0002 max mem: 52150 +Epoch: [21] [3910/7269] eta: 0:24:13 lr: 0.000005 loss: 0.5430 (0.5223) time: 0.4662 data: 0.0002 max mem: 52150 +Epoch: [21] [3920/7269] eta: 0:24:09 lr: 0.000005 loss: 0.5699 (0.5224) time: 0.4665 data: 0.0001 max mem: 52150 +Epoch: [21] [3930/7269] eta: 0:24:04 lr: 0.000005 loss: 0.5692 (0.5224) time: 0.4214 data: 0.0001 max mem: 52150 +Epoch: [21] [3940/7269] eta: 0:23:59 lr: 0.000005 loss: 0.5197 (0.5224) time: 0.3713 data: 0.0001 max mem: 52150 +Epoch: [21] [3950/7269] eta: 0:23:55 lr: 0.000005 loss: 0.5314 (0.5224) time: 0.4364 data: 0.0001 max mem: 52150 +Epoch: [21] [3960/7269] eta: 0:23:52 lr: 0.000005 loss: 0.5165 (0.5223) time: 0.5427 data: 0.0001 max mem: 52150 +Epoch: [21] [3970/7269] eta: 0:23:47 lr: 0.000005 loss: 0.5253 (0.5223) time: 0.4777 data: 0.0002 max mem: 52150 +Epoch: [21] [3980/7269] eta: 0:23:43 lr: 0.000005 loss: 0.5301 (0.5223) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [21] [3990/7269] eta: 0:23:38 lr: 0.000005 loss: 0.5449 (0.5224) time: 0.3725 data: 0.0001 max mem: 52150 +Epoch: [21] [4000/7269] eta: 0:23:33 lr: 0.000005 loss: 0.5353 (0.5224) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [21] [4010/7269] eta: 0:23:28 lr: 0.000005 loss: 0.5154 (0.5224) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [21] [4020/7269] eta: 0:23:23 lr: 0.000005 loss: 0.5255 (0.5224) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [21] [4030/7269] eta: 0:23:18 lr: 0.000005 loss: 0.5349 (0.5224) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [21] [4040/7269] eta: 0:23:14 lr: 0.000005 loss: 0.5329 (0.5223) time: 0.4175 data: 0.0002 max mem: 52150 +Epoch: [21] [4050/7269] eta: 0:23:12 lr: 0.000005 loss: 0.5169 (0.5224) time: 0.5417 data: 0.0002 max mem: 52150 +Epoch: [21] [4060/7269] eta: 0:23:07 lr: 0.000005 loss: 0.5256 (0.5224) time: 0.4964 data: 0.0001 max mem: 52150 +Epoch: [21] [4070/7269] eta: 0:23:02 lr: 0.000005 loss: 0.4957 (0.5223) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [21] [4080/7269] eta: 0:22:57 lr: 0.000005 loss: 0.5372 (0.5224) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [21] [4090/7269] eta: 0:22:52 lr: 0.000005 loss: 0.5488 (0.5224) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [21] [4100/7269] eta: 0:22:48 lr: 0.000005 loss: 0.5355 (0.5224) time: 0.4167 data: 0.0001 max mem: 52150 +Epoch: [21] [4110/7269] eta: 0:22:44 lr: 0.000005 loss: 0.5282 (0.5224) time: 0.4169 data: 0.0001 max mem: 52150 +Epoch: [21] [4120/7269] eta: 0:22:40 lr: 0.000005 loss: 0.5228 (0.5224) time: 0.4326 data: 0.0001 max mem: 52150 +Epoch: [21] [4130/7269] eta: 0:22:36 lr: 0.000005 loss: 0.5615 (0.5225) time: 0.4771 data: 0.0001 max mem: 52150 +Epoch: [21] [4140/7269] eta: 0:22:31 lr: 0.000005 loss: 0.5728 (0.5227) time: 0.4164 data: 0.0001 max mem: 52150 +Epoch: [21] [4150/7269] eta: 0:22:27 lr: 0.000005 loss: 0.5531 (0.5227) time: 0.4212 data: 0.0001 max mem: 52150 +Epoch: [21] [4160/7269] eta: 0:22:22 lr: 0.000005 loss: 0.5417 (0.5227) time: 0.4210 data: 0.0001 max mem: 52150 +Epoch: [21] [4170/7269] eta: 0:22:19 lr: 0.000005 loss: 0.5453 (0.5228) time: 0.5102 data: 0.0002 max mem: 52150 +Epoch: [21] [4180/7269] eta: 0:22:15 lr: 0.000005 loss: 0.5453 (0.5228) time: 0.5720 data: 0.0001 max mem: 52150 +Epoch: [21] [4190/7269] eta: 0:22:11 lr: 0.000005 loss: 0.5455 (0.5228) time: 0.4799 data: 0.0001 max mem: 52150 +Epoch: [21] [4200/7269] eta: 0:22:07 lr: 0.000005 loss: 0.5417 (0.5228) time: 0.4182 data: 0.0001 max mem: 52150 +Epoch: [21] [4210/7269] eta: 0:22:02 lr: 0.000005 loss: 0.5522 (0.5229) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [21] [4220/7269] eta: 0:21:57 lr: 0.000005 loss: 0.5203 (0.5229) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [21] [4230/7269] eta: 0:21:52 lr: 0.000005 loss: 0.4948 (0.5228) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [21] [4240/7269] eta: 0:21:48 lr: 0.000005 loss: 0.5269 (0.5228) time: 0.3727 data: 0.0001 max mem: 52150 +Epoch: [21] [4250/7269] eta: 0:21:43 lr: 0.000005 loss: 0.5259 (0.5228) time: 0.3729 data: 0.0001 max mem: 52150 +Epoch: [21] [4260/7269] eta: 0:21:39 lr: 0.000005 loss: 0.5259 (0.5229) time: 0.4317 data: 0.0001 max mem: 52150 +Epoch: [21] [4270/7269] eta: 0:21:34 lr: 0.000005 loss: 0.5498 (0.5229) time: 0.4311 data: 0.0002 max mem: 52150 +Epoch: [21] [4280/7269] eta: 0:21:29 lr: 0.000005 loss: 0.5463 (0.5229) time: 0.3709 data: 0.0002 max mem: 52150 +Epoch: [21] [4290/7269] eta: 0:21:25 lr: 0.000005 loss: 0.5544 (0.5230) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [21] [4300/7269] eta: 0:21:21 lr: 0.000005 loss: 0.5535 (0.5230) time: 0.4307 data: 0.0001 max mem: 52150 +Epoch: [21] [4310/7269] eta: 0:21:17 lr: 0.000005 loss: 0.5131 (0.5230) time: 0.4906 data: 0.0001 max mem: 52150 +Epoch: [21] [4320/7269] eta: 0:21:12 lr: 0.000005 loss: 0.5002 (0.5229) time: 0.4330 data: 0.0001 max mem: 52150 +Epoch: [21] [4330/7269] eta: 0:21:08 lr: 0.000005 loss: 0.5268 (0.5230) time: 0.4181 data: 0.0002 max mem: 52150 +Epoch: [21] [4340/7269] eta: 0:21:04 lr: 0.000005 loss: 0.5166 (0.5229) time: 0.4610 data: 0.0001 max mem: 52150 +Epoch: [21] [4350/7269] eta: 0:20:59 lr: 0.000005 loss: 0.5119 (0.5230) time: 0.4154 data: 0.0001 max mem: 52150 +Epoch: [21] [4360/7269] eta: 0:20:55 lr: 0.000005 loss: 0.5080 (0.5229) time: 0.4359 data: 0.0002 max mem: 52150 +Epoch: [21] [4370/7269] eta: 0:20:51 lr: 0.000005 loss: 0.4613 (0.5228) time: 0.4809 data: 0.0002 max mem: 52150 +Epoch: [21] [4380/7269] eta: 0:20:47 lr: 0.000005 loss: 0.5308 (0.5229) time: 0.4162 data: 0.0002 max mem: 52150 +Epoch: [21] [4390/7269] eta: 0:20:42 lr: 0.000005 loss: 0.5392 (0.5229) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [21] [4400/7269] eta: 0:20:37 lr: 0.000005 loss: 0.5521 (0.5229) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [21] [4410/7269] eta: 0:20:33 lr: 0.000005 loss: 0.5395 (0.5229) time: 0.4351 data: 0.0001 max mem: 52150 +Epoch: [21] [4420/7269] eta: 0:20:30 lr: 0.000005 loss: 0.5064 (0.5228) time: 0.5566 data: 0.0001 max mem: 52150 +Epoch: [21] [4430/7269] eta: 0:20:26 lr: 0.000005 loss: 0.5102 (0.5228) time: 0.5409 data: 0.0001 max mem: 52150 +Epoch: [21] [4440/7269] eta: 0:20:21 lr: 0.000005 loss: 0.5314 (0.5228) time: 0.4193 data: 0.0001 max mem: 52150 +Epoch: [21] [4450/7269] eta: 0:20:17 lr: 0.000005 loss: 0.5588 (0.5229) time: 0.3722 data: 0.0001 max mem: 52150 +Epoch: [21] [4460/7269] eta: 0:20:12 lr: 0.000005 loss: 0.5099 (0.5228) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [21] [4470/7269] eta: 0:20:08 lr: 0.000005 loss: 0.5074 (0.5228) time: 0.4195 data: 0.0001 max mem: 52150 +Epoch: [21] [4480/7269] eta: 0:20:04 lr: 0.000005 loss: 0.4967 (0.5228) time: 0.4654 data: 0.0001 max mem: 52150 +Epoch: [21] [4490/7269] eta: 0:19:59 lr: 0.000005 loss: 0.4996 (0.5227) time: 0.4172 data: 0.0001 max mem: 52150 +Epoch: [21] [4500/7269] eta: 0:19:55 lr: 0.000005 loss: 0.5300 (0.5227) time: 0.4179 data: 0.0002 max mem: 52150 +Epoch: [21] [4510/7269] eta: 0:19:51 lr: 0.000005 loss: 0.5300 (0.5227) time: 0.4809 data: 0.0002 max mem: 52150 +Epoch: [21] [4520/7269] eta: 0:19:46 lr: 0.000005 loss: 0.5334 (0.5227) time: 0.4349 data: 0.0002 max mem: 52150 +Epoch: [21] [4530/7269] eta: 0:19:42 lr: 0.000005 loss: 0.5334 (0.5227) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [21] [4540/7269] eta: 0:19:37 lr: 0.000005 loss: 0.5490 (0.5228) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [21] [4550/7269] eta: 0:19:33 lr: 0.000005 loss: 0.5387 (0.5228) time: 0.4163 data: 0.0001 max mem: 52150 +Epoch: [21] [4560/7269] eta: 0:19:28 lr: 0.000005 loss: 0.5197 (0.5227) time: 0.4163 data: 0.0001 max mem: 52150 +Epoch: [21] [4570/7269] eta: 0:19:23 lr: 0.000005 loss: 0.5197 (0.5227) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [21] [4580/7269] eta: 0:19:19 lr: 0.000005 loss: 0.5463 (0.5227) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [21] [4590/7269] eta: 0:19:16 lr: 0.000005 loss: 0.5015 (0.5226) time: 0.4921 data: 0.0001 max mem: 52150 +Epoch: [21] [4600/7269] eta: 0:19:11 lr: 0.000005 loss: 0.5015 (0.5226) time: 0.4924 data: 0.0001 max mem: 52150 +Epoch: [21] [4610/7269] eta: 0:19:07 lr: 0.000005 loss: 0.5320 (0.5226) time: 0.4194 data: 0.0002 max mem: 52150 +Epoch: [21] [4620/7269] eta: 0:19:02 lr: 0.000005 loss: 0.5398 (0.5227) time: 0.4195 data: 0.0002 max mem: 52150 +Epoch: [21] [4630/7269] eta: 0:18:58 lr: 0.000005 loss: 0.5560 (0.5228) time: 0.4204 data: 0.0002 max mem: 52150 +Epoch: [21] [4640/7269] eta: 0:18:54 lr: 0.000005 loss: 0.5606 (0.5228) time: 0.4672 data: 0.0002 max mem: 52150 +Epoch: [21] [4650/7269] eta: 0:18:49 lr: 0.000005 loss: 0.5388 (0.5228) time: 0.4189 data: 0.0001 max mem: 52150 +Epoch: [21] [4660/7269] eta: 0:18:45 lr: 0.000005 loss: 0.5302 (0.5228) time: 0.4209 data: 0.0002 max mem: 52150 +Epoch: [21] [4670/7269] eta: 0:18:41 lr: 0.000005 loss: 0.5474 (0.5229) time: 0.4655 data: 0.0002 max mem: 52150 +Epoch: [21] [4680/7269] eta: 0:18:36 lr: 0.000005 loss: 0.5454 (0.5229) time: 0.4164 data: 0.0002 max mem: 52150 +Epoch: [21] [4690/7269] eta: 0:18:32 lr: 0.000005 loss: 0.4640 (0.5227) time: 0.4166 data: 0.0001 max mem: 52150 +Epoch: [21] [4700/7269] eta: 0:18:28 lr: 0.000005 loss: 0.4472 (0.5227) time: 0.4169 data: 0.0001 max mem: 52150 +Epoch: [21] [4710/7269] eta: 0:18:23 lr: 0.000005 loss: 0.4646 (0.5226) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [21] [4720/7269] eta: 0:18:19 lr: 0.000005 loss: 0.5372 (0.5226) time: 0.4321 data: 0.0001 max mem: 52150 +Epoch: [21] [4730/7269] eta: 0:18:14 lr: 0.000005 loss: 0.5522 (0.5227) time: 0.4313 data: 0.0001 max mem: 52150 +Epoch: [21] [4740/7269] eta: 0:18:10 lr: 0.000005 loss: 0.5373 (0.5226) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [21] [4750/7269] eta: 0:18:06 lr: 0.000005 loss: 0.5352 (0.5226) time: 0.4180 data: 0.0001 max mem: 52150 +Epoch: [21] [4760/7269] eta: 0:18:01 lr: 0.000005 loss: 0.5491 (0.5227) time: 0.4178 data: 0.0001 max mem: 52150 +Epoch: [21] [4770/7269] eta: 0:17:56 lr: 0.000005 loss: 0.5565 (0.5226) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [21] [4780/7269] eta: 0:17:52 lr: 0.000005 loss: 0.5522 (0.5227) time: 0.3714 data: 0.0001 max mem: 52150 +Epoch: [21] [4790/7269] eta: 0:17:47 lr: 0.000005 loss: 0.5368 (0.5227) time: 0.3713 data: 0.0001 max mem: 52150 +Epoch: [21] [4800/7269] eta: 0:17:44 lr: 0.000005 loss: 0.5172 (0.5227) time: 0.4802 data: 0.0001 max mem: 52150 +Epoch: [21] [4810/7269] eta: 0:17:39 lr: 0.000005 loss: 0.5308 (0.5227) time: 0.5262 data: 0.0001 max mem: 52150 +Epoch: [21] [4820/7269] eta: 0:17:35 lr: 0.000005 loss: 0.5416 (0.5227) time: 0.4798 data: 0.0001 max mem: 52150 +Epoch: [21] [4830/7269] eta: 0:17:31 lr: 0.000005 loss: 0.5492 (0.5228) time: 0.4966 data: 0.0002 max mem: 52150 +Epoch: [21] [4840/7269] eta: 0:17:27 lr: 0.000005 loss: 0.5598 (0.5228) time: 0.4798 data: 0.0002 max mem: 52150 +Epoch: [21] [4850/7269] eta: 0:17:23 lr: 0.000005 loss: 0.5404 (0.5228) time: 0.4621 data: 0.0002 max mem: 52150 +Epoch: [21] [4860/7269] eta: 0:17:19 lr: 0.000005 loss: 0.5404 (0.5228) time: 0.4168 data: 0.0002 max mem: 52150 +Epoch: [21] [4870/7269] eta: 0:17:15 lr: 0.000005 loss: 0.5674 (0.5228) time: 0.4317 data: 0.0002 max mem: 52150 +Epoch: [21] [4880/7269] eta: 0:17:10 lr: 0.000005 loss: 0.5567 (0.5228) time: 0.4316 data: 0.0001 max mem: 52150 +Epoch: [21] [4890/7269] eta: 0:17:06 lr: 0.000005 loss: 0.5288 (0.5228) time: 0.4765 data: 0.0001 max mem: 52150 +Epoch: [21] [4900/7269] eta: 0:17:02 lr: 0.000005 loss: 0.5186 (0.5229) time: 0.5364 data: 0.0001 max mem: 52150 +Epoch: [21] [4910/7269] eta: 0:16:58 lr: 0.000005 loss: 0.5161 (0.5228) time: 0.4319 data: 0.0002 max mem: 52150 +Epoch: [21] [4920/7269] eta: 0:16:53 lr: 0.000005 loss: 0.5199 (0.5228) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [21] [4930/7269] eta: 0:16:49 lr: 0.000005 loss: 0.5228 (0.5228) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [21] [4940/7269] eta: 0:16:45 lr: 0.000005 loss: 0.5179 (0.5227) time: 0.4690 data: 0.0002 max mem: 52150 +Epoch: [21] [4950/7269] eta: 0:16:41 lr: 0.000005 loss: 0.5111 (0.5227) time: 0.5144 data: 0.0002 max mem: 52150 +Epoch: [21] [4960/7269] eta: 0:16:36 lr: 0.000005 loss: 0.5190 (0.5227) time: 0.4167 data: 0.0001 max mem: 52150 +Epoch: [21] [4970/7269] eta: 0:16:31 lr: 0.000005 loss: 0.5051 (0.5227) time: 0.3713 data: 0.0001 max mem: 52150 +Epoch: [21] [4980/7269] eta: 0:16:27 lr: 0.000005 loss: 0.5365 (0.5227) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [21] [4990/7269] eta: 0:16:22 lr: 0.000005 loss: 0.5306 (0.5227) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [21] [5000/7269] eta: 0:16:18 lr: 0.000005 loss: 0.4910 (0.5226) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [21] [5010/7269] eta: 0:16:13 lr: 0.000005 loss: 0.5270 (0.5227) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [21] [5020/7269] eta: 0:16:09 lr: 0.000005 loss: 0.5596 (0.5227) time: 0.4351 data: 0.0001 max mem: 52150 +Epoch: [21] [5030/7269] eta: 0:16:05 lr: 0.000005 loss: 0.5184 (0.5226) time: 0.4348 data: 0.0002 max mem: 52150 +Epoch: [21] [5040/7269] eta: 0:16:00 lr: 0.000005 loss: 0.5063 (0.5226) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [21] [5050/7269] eta: 0:15:55 lr: 0.000005 loss: 0.5111 (0.5226) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [21] [5060/7269] eta: 0:15:51 lr: 0.000005 loss: 0.5263 (0.5226) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [21] [5070/7269] eta: 0:15:46 lr: 0.000005 loss: 0.5406 (0.5227) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [21] [5080/7269] eta: 0:15:42 lr: 0.000005 loss: 0.5307 (0.5226) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [21] [5090/7269] eta: 0:15:38 lr: 0.000005 loss: 0.5082 (0.5226) time: 0.4199 data: 0.0001 max mem: 52150 +Epoch: [21] [5100/7269] eta: 0:15:33 lr: 0.000005 loss: 0.4985 (0.5225) time: 0.4651 data: 0.0002 max mem: 52150 +Epoch: [21] [5110/7269] eta: 0:15:29 lr: 0.000005 loss: 0.5069 (0.5225) time: 0.4771 data: 0.0002 max mem: 52150 +Epoch: [21] [5120/7269] eta: 0:15:25 lr: 0.000005 loss: 0.5429 (0.5225) time: 0.4319 data: 0.0002 max mem: 52150 +Epoch: [21] [5130/7269] eta: 0:15:21 lr: 0.000005 loss: 0.4973 (0.5225) time: 0.4194 data: 0.0002 max mem: 52150 +Epoch: [21] [5140/7269] eta: 0:15:16 lr: 0.000005 loss: 0.5624 (0.5226) time: 0.4642 data: 0.0001 max mem: 52150 +Epoch: [21] [5150/7269] eta: 0:15:12 lr: 0.000005 loss: 0.5624 (0.5226) time: 0.4608 data: 0.0001 max mem: 52150 +Epoch: [21] [5160/7269] eta: 0:15:08 lr: 0.000005 loss: 0.5005 (0.5225) time: 0.4160 data: 0.0001 max mem: 52150 +Epoch: [21] [5170/7269] eta: 0:15:03 lr: 0.000005 loss: 0.5005 (0.5225) time: 0.3713 data: 0.0001 max mem: 52150 +Epoch: [21] [5180/7269] eta: 0:15:00 lr: 0.000005 loss: 0.5376 (0.5225) time: 0.4771 data: 0.0001 max mem: 52150 +Epoch: [21] [5190/7269] eta: 0:14:55 lr: 0.000005 loss: 0.5291 (0.5225) time: 0.4778 data: 0.0001 max mem: 52150 +Epoch: [21] [5200/7269] eta: 0:14:50 lr: 0.000005 loss: 0.5506 (0.5226) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [21] [5210/7269] eta: 0:14:46 lr: 0.000005 loss: 0.5318 (0.5226) time: 0.3712 data: 0.0001 max mem: 52150 +Epoch: [21] [5220/7269] eta: 0:14:42 lr: 0.000005 loss: 0.4957 (0.5225) time: 0.4165 data: 0.0002 max mem: 52150 +Epoch: [21] [5230/7269] eta: 0:14:38 lr: 0.000005 loss: 0.4923 (0.5224) time: 0.4757 data: 0.0002 max mem: 52150 +Epoch: [21] [5240/7269] eta: 0:14:33 lr: 0.000005 loss: 0.4858 (0.5223) time: 0.4303 data: 0.0001 max mem: 52150 +Epoch: [21] [5250/7269] eta: 0:14:29 lr: 0.000005 loss: 0.4995 (0.5223) time: 0.4335 data: 0.0002 max mem: 52150 +Epoch: [21] [5260/7269] eta: 0:14:25 lr: 0.000005 loss: 0.5230 (0.5223) time: 0.4795 data: 0.0001 max mem: 52150 +Epoch: [21] [5270/7269] eta: 0:14:20 lr: 0.000005 loss: 0.5250 (0.5224) time: 0.4173 data: 0.0002 max mem: 52150 +Epoch: [21] [5280/7269] eta: 0:14:16 lr: 0.000005 loss: 0.5910 (0.5225) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [21] [5290/7269] eta: 0:14:12 lr: 0.000005 loss: 0.5659 (0.5225) time: 0.4344 data: 0.0002 max mem: 52150 +Epoch: [21] [5300/7269] eta: 0:14:08 lr: 0.000005 loss: 0.5476 (0.5226) time: 0.5625 data: 0.0002 max mem: 52150 +Epoch: [21] [5310/7269] eta: 0:14:04 lr: 0.000005 loss: 0.5321 (0.5226) time: 0.5004 data: 0.0002 max mem: 52150 +Epoch: [21] [5320/7269] eta: 0:13:59 lr: 0.000005 loss: 0.5280 (0.5226) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [21] [5330/7269] eta: 0:13:55 lr: 0.000005 loss: 0.5288 (0.5225) time: 0.4934 data: 0.0002 max mem: 52150 +Epoch: [21] [5340/7269] eta: 0:13:51 lr: 0.000005 loss: 0.5053 (0.5225) time: 0.5589 data: 0.0002 max mem: 52150 +Epoch: [21] [5350/7269] eta: 0:13:47 lr: 0.000005 loss: 0.5221 (0.5225) time: 0.4382 data: 0.0002 max mem: 52150 +Epoch: [21] [5360/7269] eta: 0:13:43 lr: 0.000005 loss: 0.5315 (0.5225) time: 0.4170 data: 0.0001 max mem: 52150 +Epoch: [21] [5370/7269] eta: 0:13:39 lr: 0.000005 loss: 0.5369 (0.5225) time: 0.5368 data: 0.0002 max mem: 52150 +Epoch: [21] [5380/7269] eta: 0:13:35 lr: 0.000005 loss: 0.5420 (0.5225) time: 0.5957 data: 0.0002 max mem: 52150 +Epoch: [21] [5390/7269] eta: 0:13:31 lr: 0.000005 loss: 0.4992 (0.5225) time: 0.4760 data: 0.0001 max mem: 52150 +Epoch: [21] [5400/7269] eta: 0:13:26 lr: 0.000005 loss: 0.5077 (0.5225) time: 0.3728 data: 0.0002 max mem: 52150 +Epoch: [21] [5410/7269] eta: 0:13:22 lr: 0.000005 loss: 0.5568 (0.5225) time: 0.4167 data: 0.0002 max mem: 52150 +Epoch: [21] [5420/7269] eta: 0:13:18 lr: 0.000005 loss: 0.4784 (0.5225) time: 0.4800 data: 0.0002 max mem: 52150 +Epoch: [21] [5430/7269] eta: 0:13:13 lr: 0.000005 loss: 0.4784 (0.5225) time: 0.4354 data: 0.0001 max mem: 52150 +Epoch: [21] [5440/7269] eta: 0:13:09 lr: 0.000005 loss: 0.5344 (0.5225) time: 0.4181 data: 0.0001 max mem: 52150 +Epoch: [21] [5450/7269] eta: 0:13:05 lr: 0.000005 loss: 0.5400 (0.5225) time: 0.4789 data: 0.0001 max mem: 52150 +Epoch: [21] [5460/7269] eta: 0:13:01 lr: 0.000005 loss: 0.5187 (0.5225) time: 0.4331 data: 0.0002 max mem: 52150 +Epoch: [21] [5470/7269] eta: 0:12:56 lr: 0.000005 loss: 0.5158 (0.5225) time: 0.3732 data: 0.0002 max mem: 52150 +Epoch: [21] [5480/7269] eta: 0:12:52 lr: 0.000005 loss: 0.5475 (0.5226) time: 0.3738 data: 0.0002 max mem: 52150 +Epoch: [21] [5490/7269] eta: 0:12:47 lr: 0.000005 loss: 0.5475 (0.5226) time: 0.3734 data: 0.0002 max mem: 52150 +Epoch: [21] [5500/7269] eta: 0:12:43 lr: 0.000005 loss: 0.5276 (0.5227) time: 0.3729 data: 0.0001 max mem: 52150 +Epoch: [21] [5510/7269] eta: 0:12:38 lr: 0.000005 loss: 0.5243 (0.5226) time: 0.3725 data: 0.0001 max mem: 52150 +Epoch: [21] [5520/7269] eta: 0:12:34 lr: 0.000005 loss: 0.5289 (0.5226) time: 0.4356 data: 0.0001 max mem: 52150 +Epoch: [21] [5530/7269] eta: 0:12:29 lr: 0.000005 loss: 0.4963 (0.5225) time: 0.4355 data: 0.0001 max mem: 52150 +Epoch: [21] [5540/7269] eta: 0:12:25 lr: 0.000005 loss: 0.4988 (0.5225) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [21] [5550/7269] eta: 0:12:20 lr: 0.000005 loss: 0.5388 (0.5225) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [21] [5560/7269] eta: 0:12:16 lr: 0.000005 loss: 0.5288 (0.5226) time: 0.4177 data: 0.0001 max mem: 52150 +Epoch: [21] [5570/7269] eta: 0:12:12 lr: 0.000005 loss: 0.5276 (0.5226) time: 0.4177 data: 0.0002 max mem: 52150 +Epoch: [21] [5580/7269] eta: 0:12:08 lr: 0.000005 loss: 0.5500 (0.5226) time: 0.4338 data: 0.0002 max mem: 52150 +Epoch: [21] [5590/7269] eta: 0:12:03 lr: 0.000005 loss: 0.5500 (0.5226) time: 0.4830 data: 0.0002 max mem: 52150 +Epoch: [21] [5600/7269] eta: 0:11:59 lr: 0.000005 loss: 0.5526 (0.5227) time: 0.4223 data: 0.0002 max mem: 52150 +Epoch: [21] [5610/7269] eta: 0:11:54 lr: 0.000005 loss: 0.5528 (0.5227) time: 0.3733 data: 0.0002 max mem: 52150 +Epoch: [21] [5620/7269] eta: 0:11:50 lr: 0.000005 loss: 0.5360 (0.5227) time: 0.3735 data: 0.0002 max mem: 52150 +Epoch: [21] [5630/7269] eta: 0:11:46 lr: 0.000005 loss: 0.4846 (0.5226) time: 0.4381 data: 0.0002 max mem: 52150 +Epoch: [21] [5640/7269] eta: 0:11:41 lr: 0.000005 loss: 0.5079 (0.5226) time: 0.4369 data: 0.0002 max mem: 52150 +Epoch: [21] [5650/7269] eta: 0:11:37 lr: 0.000005 loss: 0.5182 (0.5226) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [21] [5660/7269] eta: 0:11:32 lr: 0.000005 loss: 0.5204 (0.5227) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [21] [5670/7269] eta: 0:11:28 lr: 0.000005 loss: 0.5209 (0.5226) time: 0.4183 data: 0.0002 max mem: 52150 +Epoch: [21] [5680/7269] eta: 0:11:24 lr: 0.000005 loss: 0.5234 (0.5227) time: 0.4180 data: 0.0002 max mem: 52150 +Epoch: [21] [5690/7269] eta: 0:11:19 lr: 0.000005 loss: 0.5299 (0.5227) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [21] [5700/7269] eta: 0:11:15 lr: 0.000005 loss: 0.4935 (0.5226) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [21] [5710/7269] eta: 0:11:11 lr: 0.000005 loss: 0.4864 (0.5226) time: 0.4191 data: 0.0001 max mem: 52150 +Epoch: [21] [5720/7269] eta: 0:11:06 lr: 0.000005 loss: 0.5300 (0.5227) time: 0.4646 data: 0.0001 max mem: 52150 +Epoch: [21] [5730/7269] eta: 0:11:02 lr: 0.000005 loss: 0.5406 (0.5227) time: 0.4777 data: 0.0002 max mem: 52150 +Epoch: [21] [5740/7269] eta: 0:10:58 lr: 0.000005 loss: 0.5303 (0.5226) time: 0.4320 data: 0.0002 max mem: 52150 +Epoch: [21] [5750/7269] eta: 0:10:53 lr: 0.000005 loss: 0.5435 (0.5227) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [21] [5760/7269] eta: 0:10:50 lr: 0.000005 loss: 0.5637 (0.5228) time: 0.5407 data: 0.0001 max mem: 52150 +Epoch: [21] [5770/7269] eta: 0:10:45 lr: 0.000005 loss: 0.5563 (0.5228) time: 0.5408 data: 0.0002 max mem: 52150 +Epoch: [21] [5780/7269] eta: 0:10:41 lr: 0.000005 loss: 0.5527 (0.5229) time: 0.4312 data: 0.0002 max mem: 52150 +Epoch: [21] [5790/7269] eta: 0:10:37 lr: 0.000005 loss: 0.5605 (0.5230) time: 0.4315 data: 0.0002 max mem: 52150 +Epoch: [21] [5800/7269] eta: 0:10:33 lr: 0.000005 loss: 0.5442 (0.5229) time: 0.4842 data: 0.0002 max mem: 52150 +Epoch: [21] [5810/7269] eta: 0:10:29 lr: 0.000005 loss: 0.4947 (0.5229) time: 0.6110 data: 0.0001 max mem: 52150 +Epoch: [21] [5820/7269] eta: 0:10:25 lr: 0.000005 loss: 0.5056 (0.5229) time: 0.4991 data: 0.0003 max mem: 52150 +Epoch: [21] [5830/7269] eta: 0:10:20 lr: 0.000005 loss: 0.5108 (0.5229) time: 0.4320 data: 0.0003 max mem: 52150 +Epoch: [21] [5840/7269] eta: 0:10:16 lr: 0.000005 loss: 0.5211 (0.5229) time: 0.5445 data: 0.0001 max mem: 52150 +Epoch: [21] [5850/7269] eta: 0:10:12 lr: 0.000005 loss: 0.5508 (0.5230) time: 0.5321 data: 0.0002 max mem: 52150 +Epoch: [21] [5860/7269] eta: 0:10:08 lr: 0.000005 loss: 0.5353 (0.5230) time: 0.5108 data: 0.0001 max mem: 52150 +Epoch: [21] [5870/7269] eta: 0:10:04 lr: 0.000005 loss: 0.5047 (0.5229) time: 0.4634 data: 0.0001 max mem: 52150 +Epoch: [21] [5880/7269] eta: 0:09:59 lr: 0.000005 loss: 0.4859 (0.5229) time: 0.4182 data: 0.0001 max mem: 52150 +Epoch: [21] [5890/7269] eta: 0:09:55 lr: 0.000005 loss: 0.4876 (0.5229) time: 0.4182 data: 0.0002 max mem: 52150 +Epoch: [21] [5900/7269] eta: 0:09:51 lr: 0.000005 loss: 0.5319 (0.5229) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [21] [5910/7269] eta: 0:09:46 lr: 0.000005 loss: 0.5319 (0.5229) time: 0.4171 data: 0.0002 max mem: 52150 +Epoch: [21] [5920/7269] eta: 0:09:42 lr: 0.000005 loss: 0.5191 (0.5229) time: 0.4171 data: 0.0002 max mem: 52150 +Epoch: [21] [5930/7269] eta: 0:09:38 lr: 0.000005 loss: 0.5191 (0.5229) time: 0.4353 data: 0.0002 max mem: 52150 +Epoch: [21] [5940/7269] eta: 0:09:33 lr: 0.000005 loss: 0.5572 (0.5229) time: 0.4849 data: 0.0001 max mem: 52150 +Epoch: [21] [5950/7269] eta: 0:09:29 lr: 0.000005 loss: 0.4986 (0.5228) time: 0.4218 data: 0.0002 max mem: 52150 +Epoch: [21] [5960/7269] eta: 0:09:25 lr: 0.000005 loss: 0.5012 (0.5228) time: 0.4321 data: 0.0002 max mem: 52150 +Epoch: [21] [5970/7269] eta: 0:09:21 lr: 0.000005 loss: 0.5207 (0.5228) time: 0.4770 data: 0.0001 max mem: 52150 +Epoch: [21] [5980/7269] eta: 0:09:16 lr: 0.000005 loss: 0.5252 (0.5228) time: 0.4170 data: 0.0001 max mem: 52150 +Epoch: [21] [5990/7269] eta: 0:09:12 lr: 0.000005 loss: 0.5252 (0.5228) time: 0.4670 data: 0.0001 max mem: 52150 +Epoch: [21] [6000/7269] eta: 0:09:08 lr: 0.000005 loss: 0.5485 (0.5229) time: 0.4670 data: 0.0001 max mem: 52150 +Epoch: [21] [6010/7269] eta: 0:09:03 lr: 0.000005 loss: 0.5485 (0.5228) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [21] [6020/7269] eta: 0:08:59 lr: 0.000005 loss: 0.4925 (0.5228) time: 0.4190 data: 0.0001 max mem: 52150 +Epoch: [21] [6030/7269] eta: 0:08:55 lr: 0.000005 loss: 0.5296 (0.5228) time: 0.4835 data: 0.0001 max mem: 52150 +Epoch: [21] [6040/7269] eta: 0:08:51 lr: 0.000005 loss: 0.5455 (0.5228) time: 0.4962 data: 0.0001 max mem: 52150 +Epoch: [21] [6050/7269] eta: 0:08:46 lr: 0.000005 loss: 0.5287 (0.5228) time: 0.4329 data: 0.0001 max mem: 52150 +Epoch: [21] [6060/7269] eta: 0:08:42 lr: 0.000005 loss: 0.5118 (0.5228) time: 0.3730 data: 0.0002 max mem: 52150 +Epoch: [21] [6070/7269] eta: 0:08:37 lr: 0.000005 loss: 0.5149 (0.5228) time: 0.3727 data: 0.0001 max mem: 52150 +Epoch: [21] [6080/7269] eta: 0:08:33 lr: 0.000005 loss: 0.5149 (0.5228) time: 0.3725 data: 0.0001 max mem: 52150 +Epoch: [21] [6090/7269] eta: 0:08:28 lr: 0.000005 loss: 0.5186 (0.5228) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [21] [6100/7269] eta: 0:08:24 lr: 0.000005 loss: 0.5027 (0.5227) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [21] [6110/7269] eta: 0:08:20 lr: 0.000005 loss: 0.4856 (0.5227) time: 0.3727 data: 0.0002 max mem: 52150 +Epoch: [21] [6120/7269] eta: 0:08:15 lr: 0.000005 loss: 0.5262 (0.5227) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [21] [6130/7269] eta: 0:08:11 lr: 0.000005 loss: 0.5424 (0.5227) time: 0.4357 data: 0.0002 max mem: 52150 +Epoch: [21] [6140/7269] eta: 0:08:06 lr: 0.000005 loss: 0.5058 (0.5227) time: 0.4353 data: 0.0001 max mem: 52150 +Epoch: [21] [6150/7269] eta: 0:08:02 lr: 0.000005 loss: 0.5031 (0.5227) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [21] [6160/7269] eta: 0:07:58 lr: 0.000005 loss: 0.5278 (0.5227) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [21] [6170/7269] eta: 0:07:53 lr: 0.000005 loss: 0.5581 (0.5228) time: 0.4175 data: 0.0002 max mem: 52150 +Epoch: [21] [6180/7269] eta: 0:07:49 lr: 0.000005 loss: 0.5273 (0.5228) time: 0.4179 data: 0.0002 max mem: 52150 +Epoch: [21] [6190/7269] eta: 0:07:45 lr: 0.000005 loss: 0.5073 (0.5228) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [21] [6200/7269] eta: 0:07:40 lr: 0.000005 loss: 0.5483 (0.5228) time: 0.3721 data: 0.0001 max mem: 52150 +Epoch: [21] [6210/7269] eta: 0:07:36 lr: 0.000005 loss: 0.5483 (0.5228) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [21] [6220/7269] eta: 0:07:31 lr: 0.000005 loss: 0.5335 (0.5228) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [21] [6230/7269] eta: 0:07:27 lr: 0.000005 loss: 0.5483 (0.5229) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [21] [6240/7269] eta: 0:07:23 lr: 0.000005 loss: 0.5483 (0.5228) time: 0.3727 data: 0.0002 max mem: 52150 +Epoch: [21] [6250/7269] eta: 0:07:18 lr: 0.000005 loss: 0.5097 (0.5228) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [21] [6260/7269] eta: 0:07:14 lr: 0.000005 loss: 0.5214 (0.5228) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [21] [6270/7269] eta: 0:07:10 lr: 0.000005 loss: 0.5252 (0.5228) time: 0.4336 data: 0.0001 max mem: 52150 +Epoch: [21] [6280/7269] eta: 0:07:05 lr: 0.000005 loss: 0.5273 (0.5228) time: 0.4330 data: 0.0002 max mem: 52150 +Epoch: [21] [6290/7269] eta: 0:07:01 lr: 0.000005 loss: 0.5625 (0.5229) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [21] [6300/7269] eta: 0:06:57 lr: 0.000005 loss: 0.5397 (0.5229) time: 0.4334 data: 0.0002 max mem: 52150 +Epoch: [21] [6310/7269] eta: 0:06:52 lr: 0.000005 loss: 0.5088 (0.5229) time: 0.4328 data: 0.0002 max mem: 52150 +Epoch: [21] [6320/7269] eta: 0:06:48 lr: 0.000005 loss: 0.5888 (0.5230) time: 0.4179 data: 0.0001 max mem: 52150 +Epoch: [21] [6330/7269] eta: 0:06:44 lr: 0.000005 loss: 0.5780 (0.5231) time: 0.5109 data: 0.0001 max mem: 52150 +Epoch: [21] [6340/7269] eta: 0:06:40 lr: 0.000005 loss: 0.5235 (0.5230) time: 0.5102 data: 0.0001 max mem: 52150 +Epoch: [21] [6350/7269] eta: 0:06:35 lr: 0.000005 loss: 0.5228 (0.5231) time: 0.4171 data: 0.0002 max mem: 52150 +Epoch: [21] [6360/7269] eta: 0:06:31 lr: 0.000005 loss: 0.5040 (0.5230) time: 0.4324 data: 0.0002 max mem: 52150 +Epoch: [21] [6370/7269] eta: 0:06:27 lr: 0.000005 loss: 0.5010 (0.5231) time: 0.4328 data: 0.0002 max mem: 52150 +Epoch: [21] [6380/7269] eta: 0:06:22 lr: 0.000005 loss: 0.5083 (0.5230) time: 0.4814 data: 0.0002 max mem: 52150 +Epoch: [21] [6390/7269] eta: 0:06:18 lr: 0.000005 loss: 0.5057 (0.5230) time: 0.5453 data: 0.0001 max mem: 52150 +Epoch: [21] [6400/7269] eta: 0:06:14 lr: 0.000005 loss: 0.4885 (0.5230) time: 0.5962 data: 0.0001 max mem: 52150 +Epoch: [21] [6410/7269] eta: 0:06:10 lr: 0.000005 loss: 0.5149 (0.5230) time: 0.5925 data: 0.0002 max mem: 52150 +Epoch: [21] [6420/7269] eta: 0:06:06 lr: 0.000005 loss: 0.5292 (0.5229) time: 0.4927 data: 0.0002 max mem: 52150 +Epoch: [21] [6430/7269] eta: 0:06:01 lr: 0.000005 loss: 0.5175 (0.5229) time: 0.4323 data: 0.0002 max mem: 52150 +Epoch: [21] [6440/7269] eta: 0:05:57 lr: 0.000005 loss: 0.5135 (0.5229) time: 0.4209 data: 0.0002 max mem: 52150 +Epoch: [21] [6450/7269] eta: 0:05:53 lr: 0.000005 loss: 0.5265 (0.5229) time: 0.4668 data: 0.0002 max mem: 52150 +Epoch: [21] [6460/7269] eta: 0:05:48 lr: 0.000005 loss: 0.5233 (0.5229) time: 0.4184 data: 0.0001 max mem: 52150 +Epoch: [21] [6470/7269] eta: 0:05:44 lr: 0.000005 loss: 0.5165 (0.5228) time: 0.3724 data: 0.0001 max mem: 52150 +Epoch: [21] [6480/7269] eta: 0:05:40 lr: 0.000005 loss: 0.5272 (0.5228) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [21] [6490/7269] eta: 0:05:35 lr: 0.000005 loss: 0.5236 (0.5228) time: 0.3722 data: 0.0001 max mem: 52150 +Epoch: [21] [6500/7269] eta: 0:05:31 lr: 0.000005 loss: 0.4811 (0.5227) time: 0.3722 data: 0.0001 max mem: 52150 +Epoch: [21] [6510/7269] eta: 0:05:27 lr: 0.000005 loss: 0.5128 (0.5227) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [21] [6520/7269] eta: 0:05:23 lr: 0.000005 loss: 0.5167 (0.5227) time: 0.5293 data: 0.0001 max mem: 52150 +Epoch: [21] [6530/7269] eta: 0:05:18 lr: 0.000005 loss: 0.5103 (0.5227) time: 0.6382 data: 0.0002 max mem: 52150 +Epoch: [21] [6540/7269] eta: 0:05:14 lr: 0.000005 loss: 0.5103 (0.5227) time: 0.5457 data: 0.0002 max mem: 52150 +Epoch: [21] [6550/7269] eta: 0:05:10 lr: 0.000005 loss: 0.5166 (0.5227) time: 0.4843 data: 0.0001 max mem: 52150 +Epoch: [21] [6560/7269] eta: 0:05:06 lr: 0.000005 loss: 0.5243 (0.5226) time: 0.4641 data: 0.0001 max mem: 52150 +Epoch: [21] [6570/7269] eta: 0:05:01 lr: 0.000005 loss: 0.5261 (0.5226) time: 0.4166 data: 0.0001 max mem: 52150 +Epoch: [21] [6580/7269] eta: 0:04:57 lr: 0.000005 loss: 0.5261 (0.5226) time: 0.3725 data: 0.0001 max mem: 52150 +Epoch: [21] [6590/7269] eta: 0:04:53 lr: 0.000005 loss: 0.5321 (0.5226) time: 0.4187 data: 0.0001 max mem: 52150 +Epoch: [21] [6600/7269] eta: 0:04:48 lr: 0.000005 loss: 0.5351 (0.5226) time: 0.4182 data: 0.0001 max mem: 52150 +Epoch: [21] [6610/7269] eta: 0:04:44 lr: 0.000005 loss: 0.5338 (0.5227) time: 0.3722 data: 0.0001 max mem: 52150 +Epoch: [21] [6620/7269] eta: 0:04:39 lr: 0.000005 loss: 0.5049 (0.5226) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [21] [6630/7269] eta: 0:04:35 lr: 0.000005 loss: 0.5222 (0.5226) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [21] [6640/7269] eta: 0:04:31 lr: 0.000005 loss: 0.5322 (0.5226) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [21] [6650/7269] eta: 0:04:26 lr: 0.000005 loss: 0.5204 (0.5226) time: 0.4330 data: 0.0001 max mem: 52150 +Epoch: [21] [6660/7269] eta: 0:04:22 lr: 0.000005 loss: 0.5671 (0.5227) time: 0.4336 data: 0.0001 max mem: 52150 +Epoch: [21] [6670/7269] eta: 0:04:18 lr: 0.000005 loss: 0.5759 (0.5227) time: 0.4361 data: 0.0001 max mem: 52150 +Epoch: [21] [6680/7269] eta: 0:04:14 lr: 0.000005 loss: 0.5076 (0.5227) time: 0.4804 data: 0.0001 max mem: 52150 +Epoch: [21] [6690/7269] eta: 0:04:09 lr: 0.000005 loss: 0.4850 (0.5226) time: 0.4176 data: 0.0002 max mem: 52150 +Epoch: [21] [6700/7269] eta: 0:04:05 lr: 0.000005 loss: 0.4736 (0.5226) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [21] [6710/7269] eta: 0:04:00 lr: 0.000005 loss: 0.4546 (0.5225) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [21] [6720/7269] eta: 0:03:56 lr: 0.000005 loss: 0.4963 (0.5226) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [21] [6730/7269] eta: 0:03:52 lr: 0.000005 loss: 0.5662 (0.5227) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [21] [6740/7269] eta: 0:03:47 lr: 0.000005 loss: 0.5662 (0.5227) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [21] [6750/7269] eta: 0:03:43 lr: 0.000005 loss: 0.5062 (0.5226) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [21] [6760/7269] eta: 0:03:39 lr: 0.000005 loss: 0.5161 (0.5226) time: 0.3725 data: 0.0001 max mem: 52150 +Epoch: [21] [6770/7269] eta: 0:03:34 lr: 0.000005 loss: 0.5546 (0.5227) time: 0.4324 data: 0.0001 max mem: 52150 +Epoch: [21] [6780/7269] eta: 0:03:30 lr: 0.000005 loss: 0.5538 (0.5227) time: 0.4326 data: 0.0001 max mem: 52150 +Epoch: [21] [6790/7269] eta: 0:03:26 lr: 0.000005 loss: 0.5383 (0.5228) time: 0.3723 data: 0.0001 max mem: 52150 +Epoch: [21] [6800/7269] eta: 0:03:21 lr: 0.000005 loss: 0.5462 (0.5228) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [21] [6810/7269] eta: 0:03:17 lr: 0.000005 loss: 0.5317 (0.5228) time: 0.3726 data: 0.0001 max mem: 52150 +Epoch: [21] [6820/7269] eta: 0:03:13 lr: 0.000005 loss: 0.5294 (0.5228) time: 0.4324 data: 0.0002 max mem: 52150 +Epoch: [21] [6830/7269] eta: 0:03:08 lr: 0.000005 loss: 0.5316 (0.5229) time: 0.4326 data: 0.0001 max mem: 52150 +Epoch: [21] [6840/7269] eta: 0:03:04 lr: 0.000005 loss: 0.5383 (0.5229) time: 0.4185 data: 0.0001 max mem: 52150 +Epoch: [21] [6850/7269] eta: 0:03:00 lr: 0.000005 loss: 0.5494 (0.5230) time: 0.4175 data: 0.0002 max mem: 52150 +Epoch: [21] [6860/7269] eta: 0:02:55 lr: 0.000005 loss: 0.5749 (0.5230) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [21] [6870/7269] eta: 0:02:51 lr: 0.000005 loss: 0.5444 (0.5230) time: 0.4777 data: 0.0001 max mem: 52150 +Epoch: [21] [6880/7269] eta: 0:02:47 lr: 0.000005 loss: 0.5444 (0.5230) time: 0.5237 data: 0.0001 max mem: 52150 +Epoch: [21] [6890/7269] eta: 0:02:43 lr: 0.000005 loss: 0.5530 (0.5231) time: 0.4181 data: 0.0001 max mem: 52150 +Epoch: [21] [6900/7269] eta: 0:02:38 lr: 0.000005 loss: 0.5552 (0.5231) time: 0.4193 data: 0.0001 max mem: 52150 +Epoch: [21] [6910/7269] eta: 0:02:34 lr: 0.000005 loss: 0.5131 (0.5231) time: 0.4198 data: 0.0001 max mem: 52150 +Epoch: [21] [6920/7269] eta: 0:02:30 lr: 0.000005 loss: 0.5089 (0.5231) time: 0.4316 data: 0.0001 max mem: 52150 +Epoch: [21] [6930/7269] eta: 0:02:25 lr: 0.000005 loss: 0.5208 (0.5231) time: 0.4313 data: 0.0001 max mem: 52150 +Epoch: [21] [6940/7269] eta: 0:02:21 lr: 0.000005 loss: 0.5611 (0.5231) time: 0.4171 data: 0.0001 max mem: 52150 +Epoch: [21] [6950/7269] eta: 0:02:17 lr: 0.000005 loss: 0.5266 (0.5231) time: 0.5405 data: 0.0001 max mem: 52150 +Epoch: [21] [6960/7269] eta: 0:02:13 lr: 0.000005 loss: 0.5397 (0.5232) time: 0.5590 data: 0.0001 max mem: 52150 +Epoch: [21] [6970/7269] eta: 0:02:08 lr: 0.000005 loss: 0.5652 (0.5232) time: 0.5307 data: 0.0001 max mem: 52150 +Epoch: [21] [6980/7269] eta: 0:02:04 lr: 0.000005 loss: 0.5346 (0.5231) time: 0.5144 data: 0.0001 max mem: 52150 +Epoch: [21] [6990/7269] eta: 0:02:00 lr: 0.000005 loss: 0.5086 (0.5231) time: 0.4653 data: 0.0001 max mem: 52150 +Epoch: [21] [7000/7269] eta: 0:01:55 lr: 0.000005 loss: 0.4829 (0.5231) time: 0.4630 data: 0.0001 max mem: 52150 +Epoch: [21] [7010/7269] eta: 0:01:51 lr: 0.000005 loss: 0.4952 (0.5230) time: 0.4765 data: 0.0001 max mem: 52150 +Epoch: [21] [7020/7269] eta: 0:01:47 lr: 0.000005 loss: 0.4861 (0.5230) time: 0.4770 data: 0.0001 max mem: 52150 +Epoch: [21] [7030/7269] eta: 0:01:43 lr: 0.000005 loss: 0.5421 (0.5230) time: 0.4628 data: 0.0001 max mem: 52150 +Epoch: [21] [7040/7269] eta: 0:01:38 lr: 0.000005 loss: 0.5266 (0.5230) time: 0.4815 data: 0.0002 max mem: 52150 +Epoch: [21] [7050/7269] eta: 0:01:34 lr: 0.000005 loss: 0.5035 (0.5230) time: 0.4819 data: 0.0002 max mem: 52150 +Epoch: [21] [7060/7269] eta: 0:01:30 lr: 0.000005 loss: 0.5256 (0.5230) time: 0.4779 data: 0.0002 max mem: 52150 +Epoch: [21] [7070/7269] eta: 0:01:25 lr: 0.000005 loss: 0.5600 (0.5230) time: 0.4324 data: 0.0002 max mem: 52150 +Epoch: [21] [7080/7269] eta: 0:01:21 lr: 0.000005 loss: 0.5480 (0.5230) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [21] [7090/7269] eta: 0:01:17 lr: 0.000005 loss: 0.5354 (0.5231) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [21] [7100/7269] eta: 0:01:12 lr: 0.000005 loss: 0.5500 (0.5231) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [21] [7110/7269] eta: 0:01:08 lr: 0.000005 loss: 0.5528 (0.5231) time: 0.3712 data: 0.0002 max mem: 52150 +Epoch: [21] [7120/7269] eta: 0:01:04 lr: 0.000005 loss: 0.5449 (0.5231) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [21] [7130/7269] eta: 0:00:59 lr: 0.000005 loss: 0.5382 (0.5231) time: 0.4349 data: 0.0001 max mem: 52150 +Epoch: [21] [7140/7269] eta: 0:00:55 lr: 0.000005 loss: 0.5306 (0.5231) time: 0.5459 data: 0.0001 max mem: 52150 +Epoch: [21] [7150/7269] eta: 0:00:51 lr: 0.000005 loss: 0.4592 (0.5230) time: 0.4827 data: 0.0001 max mem: 52150 +Epoch: [21] [7160/7269] eta: 0:00:47 lr: 0.000005 loss: 0.4640 (0.5230) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [21] [7170/7269] eta: 0:00:42 lr: 0.000005 loss: 0.4668 (0.5229) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [21] [7180/7269] eta: 0:00:38 lr: 0.000005 loss: 0.4934 (0.5229) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [21] [7190/7269] eta: 0:00:34 lr: 0.000005 loss: 0.5023 (0.5229) time: 0.4211 data: 0.0001 max mem: 52150 +Epoch: [21] [7200/7269] eta: 0:00:29 lr: 0.000005 loss: 0.5308 (0.5229) time: 0.4209 data: 0.0001 max mem: 52150 +Epoch: [21] [7210/7269] eta: 0:00:25 lr: 0.000005 loss: 0.5308 (0.5229) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [21] [7220/7269] eta: 0:00:21 lr: 0.000005 loss: 0.5384 (0.5229) time: 0.3721 data: 0.0001 max mem: 52150 +Epoch: [21] [7230/7269] eta: 0:00:16 lr: 0.000005 loss: 0.5342 (0.5229) time: 0.4334 data: 0.0001 max mem: 52150 +Epoch: [21] [7240/7269] eta: 0:00:12 lr: 0.000005 loss: 0.5319 (0.5230) time: 0.4327 data: 0.0001 max mem: 52150 +Epoch: [21] [7250/7269] eta: 0:00:08 lr: 0.000005 loss: 0.5684 (0.5230) time: 0.3753 data: 0.0032 max mem: 52150 +Epoch: [21] [7260/7269] eta: 0:00:03 lr: 0.000005 loss: 0.5526 (0.5230) time: 0.3753 data: 0.0031 max mem: 52150 +Epoch: [21] [7268/7269] eta: 0:00:00 lr: 0.000005 loss: 0.5518 (0.5230) time: 0.3745 data: 0.0031 max mem: 52150 +Epoch: [21] Total time: 0:52:13 (0.4311 s / it) +Averaged stats: lr: 0.000005 loss: 0.5518 (0.5209) +Test: [ 0/1403] eta: 1:17:11 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1070 (0.1070) acc1: 98.4375 (98.4375) acc5: 100.0000 (100.0000) time: 3.3008 data: 3.2034 max mem: 52150 +Test: [ 10/1403] eta: 0:09:00 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1519 (0.1965) acc1: 96.8750 (93.8920) acc5: 100.0000 (100.0000) time: 0.3878 data: 0.2915 max mem: 52150 +Test: [ 20/1403] eta: 0:05:44 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1449 (0.1811) acc1: 96.8750 (94.4196) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 30/1403] eta: 0:04:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0897 (0.1673) acc1: 98.4375 (94.9093) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 40/1403] eta: 0:03:58 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1036 (0.1720) acc1: 98.4375 (94.9695) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 50/1403] eta: 0:03:35 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1273 (0.1975) acc1: 98.4375 (93.5968) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 60/1403] eta: 0:03:20 acc1_pos: 90.625 acc5_pos: 100.000 loss: 0.1426 (0.1884) acc1: 100.0000 (94.4160) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 70/1403] eta: 0:03:08 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.1448 (0.1895) acc1: 98.4375 (94.2342) acc5: 100.0000 (100.0000) time: 0.0963 data: 0.0003 max mem: 52150 +Test: [ 80/1403] eta: 0:02:59 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1547 (0.1913) acc1: 95.3125 (94.1744) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 90/1403] eta: 0:02:52 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1679 (0.2048) acc1: 95.3125 (93.4066) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 100/1403] eta: 0:02:47 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1679 (0.2025) acc1: 95.3125 (93.6572) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 110/1403] eta: 0:02:42 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1645 (0.1993) acc1: 98.4375 (93.9330) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 120/1403] eta: 0:02:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0875 (0.1896) acc1: 98.4375 (94.3957) acc5: 100.0000 (100.0000) time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 130/1403] eta: 0:02:33 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0886 (0.1827) acc1: 100.0000 (94.7281) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 140/1403] eta: 0:02:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0788 (0.1751) acc1: 100.0000 (95.1020) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 150/1403] eta: 0:02:27 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0763 (0.1690) acc1: 100.0000 (95.3953) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 160/1403] eta: 0:02:24 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0770 (0.1641) acc1: 100.0000 (95.6231) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 170/1403] eta: 0:02:22 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0787 (0.1595) acc1: 100.0000 (95.8516) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 180/1403] eta: 0:02:19 acc1_pos: 87.500 acc5_pos: 100.000 loss: 0.0793 (0.1635) acc1: 100.0000 (95.6578) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 190/1403] eta: 0:02:17 acc1_pos: 87.500 acc5_pos: 100.000 loss: 0.3011 (0.1740) acc1: 87.5000 (95.0753) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 200/1403] eta: 0:02:15 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.3203 (0.1843) acc1: 85.9375 (94.5818) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 210/1403] eta: 0:02:13 acc1_pos: 75.000 acc5_pos: 100.000 loss: 0.3259 (0.1955) acc1: 84.3750 (93.9055) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 220/1403] eta: 0:02:11 acc1_pos: 87.500 acc5_pos: 100.000 loss: 0.3858 (0.2042) acc1: 82.8125 (93.4813) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 230/1403] eta: 0:02:09 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2416 (0.1999) acc1: 92.1875 (93.7094) acc5: 100.0000 (100.0000) time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 240/1403] eta: 0:02:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0777 (0.1949) acc1: 100.0000 (93.9575) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0004 max mem: 52150 +Test: [ 250/1403] eta: 0:02:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0769 (0.1905) acc1: 100.0000 (94.1795) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 260/1403] eta: 0:02:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0785 (0.1864) acc1: 100.0000 (94.3846) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 270/1403] eta: 0:02:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0787 (0.1826) acc1: 100.0000 (94.5745) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 280/1403] eta: 0:02:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0783 (0.1788) acc1: 100.0000 (94.7620) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 290/1403] eta: 0:01:59 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0759 (0.1754) acc1: 100.0000 (94.9366) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 300/1403] eta: 0:01:58 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0753 (0.1721) acc1: 100.0000 (95.0945) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 310/1403] eta: 0:01:56 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0796 (0.1694) acc1: 100.0000 (95.2472) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 320/1403] eta: 0:01:55 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1038 (0.1675) acc1: 100.0000 (95.3952) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 330/1403] eta: 0:01:53 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1104 (0.1659) acc1: 100.0000 (95.5344) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 340/1403] eta: 0:01:52 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1104 (0.1641) acc1: 100.0000 (95.6653) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 350/1403] eta: 0:01:51 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1077 (0.1626) acc1: 100.0000 (95.7888) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 360/1403] eta: 0:01:49 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1112 (0.1613) acc1: 100.0000 (95.8968) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 370/1403] eta: 0:01:48 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1101 (0.1598) acc1: 100.0000 (96.0074) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 380/1403] eta: 0:01:47 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1081 (0.1585) acc1: 100.0000 (96.1122) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 390/1403] eta: 0:01:46 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1100 (0.1591) acc1: 100.0000 (96.0758) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 400/1403] eta: 0:01:44 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2636 (0.1632) acc1: 89.0625 (95.8580) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [ 410/1403] eta: 0:01:43 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2982 (0.1666) acc1: 87.5000 (95.6585) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [ 420/1403] eta: 0:01:42 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2627 (0.1686) acc1: 89.0625 (95.5278) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [ 430/1403] eta: 0:01:41 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2589 (0.1715) acc1: 89.0625 (95.3596) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [ 440/1403] eta: 0:01:40 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3178 (0.1752) acc1: 87.5000 (95.1779) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0970 data: 0.0003 max mem: 52150 +Test: [ 450/1403] eta: 0:01:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2902 (0.1776) acc1: 89.0625 (95.0561) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [ 460/1403] eta: 0:01:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2863 (0.1804) acc1: 89.0625 (94.8854) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 470/1403] eta: 0:01:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2496 (0.1795) acc1: 90.6250 (94.9144) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 480/1403] eta: 0:01:35 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0646 (0.1771) acc1: 100.0000 (95.0201) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 490/1403] eta: 0:01:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0648 (0.1749) acc1: 100.0000 (95.1216) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 500/1403] eta: 0:01:33 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0648 (0.1727) acc1: 100.0000 (95.2189) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 510/1403] eta: 0:01:31 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0650 (0.1706) acc1: 100.0000 (95.3125) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 520/1403] eta: 0:01:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0700 (0.1687) acc1: 100.0000 (95.4025) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 530/1403] eta: 0:01:29 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0600 (0.1667) acc1: 100.0000 (95.4891) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 540/1403] eta: 0:01:28 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0596 (0.1647) acc1: 100.0000 (95.5724) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 550/1403] eta: 0:01:27 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0604 (0.1632) acc1: 100.0000 (95.6329) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 560/1403] eta: 0:01:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0840 (0.1620) acc1: 98.4375 (95.6913) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 570/1403] eta: 0:01:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0858 (0.1608) acc1: 98.4375 (95.7503) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [ 580/1403] eta: 0:01:24 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0894 (0.1602) acc1: 98.4375 (95.7724) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [ 590/1403] eta: 0:01:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1193 (0.1598) acc1: 96.8750 (95.7699) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [ 600/1403] eta: 0:01:21 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1306 (0.1599) acc1: 93.7500 (95.7259) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 610/1403] eta: 0:01:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1314 (0.1593) acc1: 95.3125 (95.7421) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 620/1403] eta: 0:01:19 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1335 (0.1601) acc1: 95.3125 (95.6622) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 630/1403] eta: 0:01:18 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1560 (0.1599) acc1: 93.7500 (95.6592) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 640/1403] eta: 0:01:17 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1198 (0.1595) acc1: 95.3125 (95.6708) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 650/1403] eta: 0:01:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1198 (0.1589) acc1: 96.8750 (95.7013) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 660/1403] eta: 0:01:15 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0916 (0.1577) acc1: 100.0000 (95.7593) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 670/1403] eta: 0:01:14 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0743 (0.1565) acc1: 100.0000 (95.8132) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 680/1403] eta: 0:01:13 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0900 (0.1580) acc1: 96.8750 (95.6980) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 690/1403] eta: 0:01:12 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2839 (0.1621) acc1: 87.5000 (95.4640) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 700/1403] eta: 0:01:11 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1930 (0.1623) acc1: 89.0625 (95.4440) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 710/1403] eta: 0:01:10 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3337 (0.1672) acc1: 82.8125 (95.1894) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 720/1403] eta: 0:01:09 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3337 (0.1672) acc1: 81.2500 (95.1890) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 730/1403] eta: 0:01:08 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0682 (0.1660) acc1: 100.0000 (95.2398) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 740/1403] eta: 0:01:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0781 (0.1652) acc1: 98.4375 (95.2830) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0972 data: 0.0003 max mem: 52150 +Test: [ 750/1403] eta: 0:01:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1218 (0.1669) acc1: 96.8750 (95.1669) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0976 data: 0.0003 max mem: 52150 +Test: [ 760/1403] eta: 0:01:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3336 (0.1706) acc1: 81.2500 (94.9203) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.0974 data: 0.0003 max mem: 52150 +Test: [ 770/1403] eta: 0:01:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5380 (0.1758) acc1: 68.7500 (94.5951) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.0973 data: 0.0003 max mem: 52150 +Test: [ 780/1403] eta: 0:01:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4446 (0.1784) acc1: 75.0000 (94.4382) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0974 data: 0.0003 max mem: 52150 +Test: [ 790/1403] eta: 0:01:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3134 (0.1806) acc1: 82.8125 (94.3229) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.0971 data: 0.0003 max mem: 52150 +Test: [ 800/1403] eta: 0:01:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2086 (0.1809) acc1: 92.1875 (94.3137) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 810/1403] eta: 0:00:59 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2233 (0.1840) acc1: 89.0625 (94.1719) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 820/1403] eta: 0:00:58 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6445 (0.1931) acc1: 71.8750 (93.7805) acc5: 100.0000 (100.0000) acc1_neg: 42.188 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 830/1403] eta: 0:00:57 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6867 (0.1957) acc1: 68.7500 (93.6485) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 840/1403] eta: 0:00:56 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5130 (0.2036) acc1: 73.4375 (93.3171) acc5: 100.0000 (100.0000) acc1_neg: 50.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 850/1403] eta: 0:00:55 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.0240 (0.2143) acc1: 56.2500 (92.8705) acc5: 100.0000 (100.0000) acc1_neg: 56.250 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 860/1403] eta: 0:00:54 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.9937 (0.2176) acc1: 57.8125 (92.7283) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0970 data: 0.0003 max mem: 52150 +Test: [ 870/1403] eta: 0:00:53 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1299 (0.2165) acc1: 95.3125 (92.7777) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [ 880/1403] eta: 0:00:52 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0974 (0.2152) acc1: 98.4375 (92.8348) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 890/1403] eta: 0:00:51 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0823 (0.2138) acc1: 98.4375 (92.9065) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [ 900/1403] eta: 0:00:50 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0789 (0.2122) acc1: 100.0000 (92.9818) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0973 data: 0.0003 max mem: 52150 +Test: [ 910/1403] eta: 0:00:49 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0803 (0.2114) acc1: 98.4375 (93.0142) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0976 data: 0.0003 max mem: 52150 +Test: [ 920/1403] eta: 0:00:48 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1207 (0.2105) acc1: 96.8750 (93.0442) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0976 data: 0.0003 max mem: 52150 +Test: [ 930/1403] eta: 0:00:47 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1107 (0.2094) acc1: 96.8750 (93.1005) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0974 data: 0.0003 max mem: 52150 +Test: [ 940/1403] eta: 0:00:46 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1276 (0.2096) acc1: 96.8750 (93.0825) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0975 data: 0.0003 max mem: 52150 +Test: [ 950/1403] eta: 0:00:45 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2420 (0.2105) acc1: 87.5000 (93.0090) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.0978 data: 0.0003 max mem: 52150 +Test: [ 960/1403] eta: 0:00:44 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2839 (0.2112) acc1: 85.9375 (92.9582) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0975 data: 0.0003 max mem: 52150 +Test: [ 970/1403] eta: 0:00:43 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2453 (0.2121) acc1: 87.5000 (92.8988) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.0970 data: 0.0003 max mem: 52150 +Test: [ 980/1403] eta: 0:00:42 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2711 (0.2129) acc1: 87.5000 (92.8421) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [ 990/1403] eta: 0:00:41 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2529 (0.2134) acc1: 89.0625 (92.8134) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [1000/1403] eta: 0:00:40 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2432 (0.2142) acc1: 90.6250 (92.7713) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [1010/1403] eta: 0:00:39 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2738 (0.2150) acc1: 87.5000 (92.7222) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [1020/1403] eta: 0:00:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4248 (0.2208) acc1: 81.2500 (92.3880) acc5: 100.0000 (100.0000) acc1_neg: 35.938 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1030/1403] eta: 0:00:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4285 (0.2225) acc1: 60.9375 (92.2709) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1040/1403] eta: 0:00:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2383 (0.2222) acc1: 89.0625 (92.2821) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1050/1403] eta: 0:00:35 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1796 (0.2219) acc1: 93.7500 (92.3049) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1060/1403] eta: 0:00:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1449 (0.2211) acc1: 96.8750 (92.3495) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1070/1403] eta: 0:00:33 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1449 (0.2217) acc1: 96.8750 (92.3319) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1080/1403] eta: 0:00:32 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3980 (0.2237) acc1: 84.3750 (92.2280) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [1090/1403] eta: 0:00:31 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4448 (0.2258) acc1: 79.6875 (92.1245) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [1100/1403] eta: 0:00:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3591 (0.2264) acc1: 84.3750 (92.0953) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1110/1403] eta: 0:00:29 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2927 (0.2269) acc1: 89.0625 (92.0694) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1120/1403] eta: 0:00:28 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2927 (0.2275) acc1: 89.0625 (92.0398) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1130/1403] eta: 0:00:27 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2891 (0.2283) acc1: 89.0625 (91.9927) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1140/1403] eta: 0:00:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2648 (0.2285) acc1: 89.0625 (91.9780) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1150/1403] eta: 0:00:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2632 (0.2290) acc1: 89.0625 (91.9472) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1160/1403] eta: 0:00:24 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2855 (0.2296) acc1: 89.0625 (91.9237) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1170/1403] eta: 0:00:23 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2815 (0.2295) acc1: 89.0625 (91.9273) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1180/1403] eta: 0:00:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0714 (0.2282) acc1: 100.0000 (91.9957) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1190/1403] eta: 0:00:21 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0705 (0.2269) acc1: 100.0000 (92.0629) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1200/1403] eta: 0:00:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0707 (0.2256) acc1: 100.0000 (92.1290) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1210/1403] eta: 0:00:19 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0699 (0.2243) acc1: 100.0000 (92.1940) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0002 max mem: 52150 +Test: [1220/1403] eta: 0:00:18 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0699 (0.2230) acc1: 100.0000 (92.2579) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1230/1403] eta: 0:00:17 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0711 (0.2218) acc1: 100.0000 (92.3208) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1240/1403] eta: 0:00:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0707 (0.2206) acc1: 100.0000 (92.3827) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1250/1403] eta: 0:00:15 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0704 (0.2194) acc1: 100.0000 (92.4435) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1260/1403] eta: 0:00:14 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0742 (0.2182) acc1: 100.0000 (92.5035) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1270/1403] eta: 0:00:13 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0751 (0.2171) acc1: 100.0000 (92.5625) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1280/1403] eta: 0:00:12 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0750 (0.2160) acc1: 100.0000 (92.6205) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1290/1403] eta: 0:00:11 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0745 (0.2149) acc1: 100.0000 (92.6777) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1300/1403] eta: 0:00:10 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0754 (0.2139) acc1: 100.0000 (92.7340) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1310/1403] eta: 0:00:09 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0753 (0.2128) acc1: 100.0000 (92.7894) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1320/1403] eta: 0:00:08 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0739 (0.2117) acc1: 100.0000 (92.8440) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1330/1403] eta: 0:00:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0731 (0.2107) acc1: 100.0000 (92.8966) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1340/1403] eta: 0:00:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0673 (0.2096) acc1: 100.0000 (92.9484) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1350/1403] eta: 0:00:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0662 (0.2086) acc1: 100.0000 (92.9982) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1360/1403] eta: 0:00:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0658 (0.2075) acc1: 100.0000 (93.0497) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1370/1403] eta: 0:00:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0654 (0.2066) acc1: 100.0000 (93.0981) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1380/1403] eta: 0:00:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0650 (0.2055) acc1: 100.0000 (93.1481) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1390/1403] eta: 0:00:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0648 (0.2045) acc1: 100.0000 (93.1973) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52150 +Test: [1400/1403] eta: 0:00:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0656 (0.2036) acc1: 100.0000 (93.2437) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0001 max mem: 52150 +Test: [1402/1403] eta: 0:00:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0665 (0.2034) acc1: 100.0000 (93.2486) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0923 data: 0.0001 max mem: 52150 +Test: Total time: 0:02:19 (0.0994 s / it) +* Overall Acc@1 93.249 Acc@5 100.000 + target == real Acc@1 95.205 Acc@5 100.000 + target == fake Acc@1 92.697 Acc@5 100.000 + loss 0.203 +Accuracy of the network on the 89729 test images: 93.2% +Max accuracy: 93.40% +Epoch: [22] [ 0/7269] eta: 11:24:59 lr: 0.000005 loss: 0.5483 (0.5483) time: 5.6540 data: 5.2424 max mem: 52150 +Epoch: [22] [ 10/7269] eta: 1:54:50 lr: 0.000005 loss: 0.5603 (0.5210) time: 0.9492 data: 0.4767 max mem: 52150 +Epoch: [22] [ 20/7269] eta: 1:28:32 lr: 0.000005 loss: 0.5682 (0.5406) time: 0.4868 data: 0.0002 max mem: 52150 +Epoch: [22] [ 30/7269] eta: 1:14:23 lr: 0.000005 loss: 0.5331 (0.5369) time: 0.4337 data: 0.0001 max mem: 52150 +Epoch: [22] [ 40/7269] eta: 1:07:05 lr: 0.000005 loss: 0.5299 (0.5379) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [22] [ 50/7269] eta: 1:02:37 lr: 0.000005 loss: 0.5299 (0.5364) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [22] [ 60/7269] eta: 1:01:27 lr: 0.000005 loss: 0.5382 (0.5391) time: 0.4182 data: 0.0002 max mem: 52150 +Epoch: [22] [ 70/7269] eta: 1:00:36 lr: 0.000005 loss: 0.5346 (0.5376) time: 0.4658 data: 0.0002 max mem: 52150 +Epoch: [22] [ 80/7269] eta: 0:58:32 lr: 0.000005 loss: 0.5319 (0.5380) time: 0.4188 data: 0.0002 max mem: 52150 +Epoch: [22] [ 90/7269] eta: 0:56:56 lr: 0.000005 loss: 0.5279 (0.5372) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [22] [ 100/7269] eta: 0:58:33 lr: 0.000005 loss: 0.5279 (0.5336) time: 0.4961 data: 0.0002 max mem: 52150 +Epoch: [22] [ 110/7269] eta: 0:58:33 lr: 0.000005 loss: 0.5114 (0.5287) time: 0.5584 data: 0.0002 max mem: 52150 +Epoch: [22] [ 120/7269] eta: 0:57:17 lr: 0.000005 loss: 0.5039 (0.5277) time: 0.4343 data: 0.0002 max mem: 52150 +Epoch: [22] [ 130/7269] eta: 0:56:13 lr: 0.000005 loss: 0.5284 (0.5258) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [22] [ 140/7269] eta: 0:55:17 lr: 0.000005 loss: 0.5260 (0.5250) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [22] [ 150/7269] eta: 0:54:28 lr: 0.000005 loss: 0.5308 (0.5261) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [22] [ 160/7269] eta: 0:54:27 lr: 0.000005 loss: 0.5471 (0.5288) time: 0.4188 data: 0.0002 max mem: 52150 +Epoch: [22] [ 170/7269] eta: 0:54:27 lr: 0.000005 loss: 0.5366 (0.5270) time: 0.4695 data: 0.0002 max mem: 52150 +Epoch: [22] [ 180/7269] eta: 0:53:48 lr: 0.000005 loss: 0.5020 (0.5272) time: 0.4220 data: 0.0002 max mem: 52150 +Epoch: [22] [ 190/7269] eta: 0:54:01 lr: 0.000005 loss: 0.4965 (0.5245) time: 0.4371 data: 0.0002 max mem: 52150 +Epoch: [22] [ 200/7269] eta: 0:53:26 lr: 0.000005 loss: 0.5247 (0.5242) time: 0.4370 data: 0.0002 max mem: 52150 +Epoch: [22] [ 210/7269] eta: 0:53:59 lr: 0.000005 loss: 0.5411 (0.5248) time: 0.4693 data: 0.0002 max mem: 52150 +Epoch: [22] [ 220/7269] eta: 0:53:59 lr: 0.000005 loss: 0.5284 (0.5238) time: 0.5196 data: 0.0002 max mem: 52150 +Epoch: [22] [ 230/7269] eta: 0:53:57 lr: 0.000005 loss: 0.5284 (0.5254) time: 0.4697 data: 0.0002 max mem: 52150 +Epoch: [22] [ 240/7269] eta: 0:53:26 lr: 0.000005 loss: 0.5322 (0.5241) time: 0.4195 data: 0.0002 max mem: 52150 +Epoch: [22] [ 250/7269] eta: 0:53:33 lr: 0.000005 loss: 0.5007 (0.5231) time: 0.4345 data: 0.0002 max mem: 52150 +Epoch: [22] [ 260/7269] eta: 0:53:06 lr: 0.000005 loss: 0.4812 (0.5221) time: 0.4348 data: 0.0002 max mem: 52150 +Epoch: [22] [ 270/7269] eta: 0:53:37 lr: 0.000005 loss: 0.5144 (0.5216) time: 0.4827 data: 0.0001 max mem: 52150 +Epoch: [22] [ 280/7269] eta: 0:53:11 lr: 0.000005 loss: 0.5523 (0.5230) time: 0.4825 data: 0.0002 max mem: 52150 +Epoch: [22] [ 290/7269] eta: 0:52:46 lr: 0.000005 loss: 0.5523 (0.5233) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [22] [ 300/7269] eta: 0:52:22 lr: 0.000005 loss: 0.5184 (0.5228) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [22] [ 310/7269] eta: 0:52:00 lr: 0.000005 loss: 0.5103 (0.5230) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [22] [ 320/7269] eta: 0:52:27 lr: 0.000005 loss: 0.5113 (0.5222) time: 0.4814 data: 0.0002 max mem: 52150 +Epoch: [22] [ 330/7269] eta: 0:52:05 lr: 0.000005 loss: 0.5113 (0.5212) time: 0.4817 data: 0.0002 max mem: 52150 +Epoch: [22] [ 340/7269] eta: 0:52:50 lr: 0.000005 loss: 0.5244 (0.5211) time: 0.5335 data: 0.0002 max mem: 52150 +Epoch: [22] [ 350/7269] eta: 0:52:47 lr: 0.000005 loss: 0.5015 (0.5190) time: 0.5799 data: 0.0001 max mem: 52150 +Epoch: [22] [ 360/7269] eta: 0:52:26 lr: 0.000005 loss: 0.4590 (0.5184) time: 0.4188 data: 0.0001 max mem: 52150 +Epoch: [22] [ 370/7269] eta: 0:52:24 lr: 0.000005 loss: 0.4909 (0.5178) time: 0.4201 data: 0.0001 max mem: 52150 +Epoch: [22] [ 380/7269] eta: 0:52:23 lr: 0.000005 loss: 0.5142 (0.5181) time: 0.4704 data: 0.0002 max mem: 52150 +Epoch: [22] [ 390/7269] eta: 0:52:03 lr: 0.000005 loss: 0.5198 (0.5178) time: 0.4218 data: 0.0001 max mem: 52150 +Epoch: [22] [ 400/7269] eta: 0:52:18 lr: 0.000005 loss: 0.5056 (0.5176) time: 0.4705 data: 0.0001 max mem: 52150 +Epoch: [22] [ 410/7269] eta: 0:52:00 lr: 0.000005 loss: 0.5238 (0.5182) time: 0.4716 data: 0.0002 max mem: 52150 +Epoch: [22] [ 420/7269] eta: 0:51:42 lr: 0.000005 loss: 0.5472 (0.5184) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [22] [ 430/7269] eta: 0:51:40 lr: 0.000005 loss: 0.5364 (0.5183) time: 0.4227 data: 0.0002 max mem: 52150 +Epoch: [22] [ 440/7269] eta: 0:51:23 lr: 0.000005 loss: 0.4893 (0.5173) time: 0.4229 data: 0.0002 max mem: 52150 +Epoch: [22] [ 450/7269] eta: 0:51:07 lr: 0.000005 loss: 0.5125 (0.5179) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [22] [ 460/7269] eta: 0:51:04 lr: 0.000005 loss: 0.5448 (0.5184) time: 0.4186 data: 0.0001 max mem: 52150 +Epoch: [22] [ 470/7269] eta: 0:50:49 lr: 0.000005 loss: 0.5121 (0.5180) time: 0.4190 data: 0.0001 max mem: 52150 +Epoch: [22] [ 480/7269] eta: 0:50:34 lr: 0.000005 loss: 0.5150 (0.5183) time: 0.3723 data: 0.0001 max mem: 52150 +Epoch: [22] [ 490/7269] eta: 0:50:44 lr: 0.000005 loss: 0.5337 (0.5186) time: 0.4650 data: 0.0002 max mem: 52150 +Epoch: [22] [ 500/7269] eta: 0:50:29 lr: 0.000005 loss: 0.5354 (0.5185) time: 0.4645 data: 0.0002 max mem: 52150 +Epoch: [22] [ 510/7269] eta: 0:50:15 lr: 0.000005 loss: 0.5316 (0.5189) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [22] [ 520/7269] eta: 0:50:14 lr: 0.000005 loss: 0.5219 (0.5193) time: 0.4229 data: 0.0002 max mem: 52150 +Epoch: [22] [ 530/7269] eta: 0:50:00 lr: 0.000005 loss: 0.5457 (0.5196) time: 0.4233 data: 0.0002 max mem: 52150 +Epoch: [22] [ 540/7269] eta: 0:49:47 lr: 0.000005 loss: 0.5549 (0.5205) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [22] [ 550/7269] eta: 0:49:33 lr: 0.000005 loss: 0.5466 (0.5204) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [22] [ 560/7269] eta: 0:49:21 lr: 0.000005 loss: 0.5243 (0.5209) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [22] [ 570/7269] eta: 0:49:37 lr: 0.000005 loss: 0.5140 (0.5208) time: 0.4967 data: 0.0002 max mem: 52150 +Epoch: [22] [ 580/7269] eta: 0:49:38 lr: 0.000005 loss: 0.5130 (0.5210) time: 0.5573 data: 0.0002 max mem: 52150 +Epoch: [22] [ 590/7269] eta: 0:49:26 lr: 0.000005 loss: 0.5018 (0.5205) time: 0.4335 data: 0.0002 max mem: 52150 +Epoch: [22] [ 600/7269] eta: 0:49:26 lr: 0.000005 loss: 0.4735 (0.5196) time: 0.4306 data: 0.0002 max mem: 52150 +Epoch: [22] [ 610/7269] eta: 0:49:14 lr: 0.000005 loss: 0.4992 (0.5198) time: 0.4304 data: 0.0001 max mem: 52150 +Epoch: [22] [ 620/7269] eta: 0:49:02 lr: 0.000005 loss: 0.5394 (0.5199) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [22] [ 630/7269] eta: 0:48:59 lr: 0.000005 loss: 0.5181 (0.5200) time: 0.4177 data: 0.0001 max mem: 52150 +Epoch: [22] [ 640/7269] eta: 0:48:48 lr: 0.000005 loss: 0.5291 (0.5201) time: 0.4175 data: 0.0002 max mem: 52150 +Epoch: [22] [ 650/7269] eta: 0:48:48 lr: 0.000005 loss: 0.5528 (0.5206) time: 0.4309 data: 0.0002 max mem: 52150 +Epoch: [22] [ 660/7269] eta: 0:48:46 lr: 0.000005 loss: 0.5591 (0.5207) time: 0.4773 data: 0.0002 max mem: 52150 +Epoch: [22] [ 670/7269] eta: 0:48:35 lr: 0.000005 loss: 0.5067 (0.5204) time: 0.4188 data: 0.0001 max mem: 52150 +Epoch: [22] [ 680/7269] eta: 0:48:35 lr: 0.000005 loss: 0.5213 (0.5205) time: 0.4343 data: 0.0002 max mem: 52150 +Epoch: [22] [ 690/7269] eta: 0:48:36 lr: 0.000005 loss: 0.5105 (0.5199) time: 0.4950 data: 0.0002 max mem: 52150 +Epoch: [22] [ 700/7269] eta: 0:48:25 lr: 0.000005 loss: 0.5105 (0.5201) time: 0.4339 data: 0.0002 max mem: 52150 +Epoch: [22] [ 710/7269] eta: 0:48:34 lr: 0.000005 loss: 0.5333 (0.5202) time: 0.4807 data: 0.0002 max mem: 52150 +Epoch: [22] [ 720/7269] eta: 0:48:34 lr: 0.000005 loss: 0.5450 (0.5203) time: 0.5404 data: 0.0002 max mem: 52150 +Epoch: [22] [ 730/7269] eta: 0:48:23 lr: 0.000005 loss: 0.5324 (0.5205) time: 0.4322 data: 0.0002 max mem: 52150 +Epoch: [22] [ 740/7269] eta: 0:48:12 lr: 0.000005 loss: 0.5252 (0.5205) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [22] [ 750/7269] eta: 0:48:10 lr: 0.000005 loss: 0.5234 (0.5204) time: 0.4193 data: 0.0002 max mem: 52150 +Epoch: [22] [ 760/7269] eta: 0:47:59 lr: 0.000005 loss: 0.5344 (0.5210) time: 0.4198 data: 0.0002 max mem: 52150 +Epoch: [22] [ 770/7269] eta: 0:47:49 lr: 0.000005 loss: 0.5344 (0.5207) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [22] [ 780/7269] eta: 0:47:39 lr: 0.000005 loss: 0.5051 (0.5206) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [22] [ 790/7269] eta: 0:47:29 lr: 0.000005 loss: 0.5147 (0.5207) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [22] [ 800/7269] eta: 0:47:19 lr: 0.000005 loss: 0.5190 (0.5209) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [22] [ 810/7269] eta: 0:47:09 lr: 0.000005 loss: 0.5217 (0.5207) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [22] [ 820/7269] eta: 0:47:09 lr: 0.000005 loss: 0.4988 (0.5204) time: 0.4315 data: 0.0001 max mem: 52150 +Epoch: [22] [ 830/7269] eta: 0:47:06 lr: 0.000005 loss: 0.4869 (0.5203) time: 0.4780 data: 0.0002 max mem: 52150 +Epoch: [22] [ 840/7269] eta: 0:47:11 lr: 0.000005 loss: 0.4869 (0.5199) time: 0.5093 data: 0.0002 max mem: 52150 +Epoch: [22] [ 850/7269] eta: 0:47:01 lr: 0.000005 loss: 0.4691 (0.5198) time: 0.4630 data: 0.0002 max mem: 52150 +Epoch: [22] [ 860/7269] eta: 0:46:52 lr: 0.000005 loss: 0.5363 (0.5202) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [22] [ 870/7269] eta: 0:46:43 lr: 0.000005 loss: 0.5363 (0.5202) time: 0.3726 data: 0.0001 max mem: 52150 +Epoch: [22] [ 880/7269] eta: 0:46:33 lr: 0.000005 loss: 0.4952 (0.5198) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [22] [ 890/7269] eta: 0:46:24 lr: 0.000005 loss: 0.4928 (0.5196) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [22] [ 900/7269] eta: 0:46:15 lr: 0.000005 loss: 0.5151 (0.5195) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [22] [ 910/7269] eta: 0:46:15 lr: 0.000005 loss: 0.5106 (0.5193) time: 0.4321 data: 0.0002 max mem: 52150 +Epoch: [22] [ 920/7269] eta: 0:46:21 lr: 0.000005 loss: 0.5095 (0.5194) time: 0.5428 data: 0.0002 max mem: 52150 +Epoch: [22] [ 930/7269] eta: 0:46:13 lr: 0.000005 loss: 0.5095 (0.5191) time: 0.4831 data: 0.0002 max mem: 52150 +Epoch: [22] [ 940/7269] eta: 0:46:10 lr: 0.000005 loss: 0.5142 (0.5191) time: 0.4183 data: 0.0002 max mem: 52150 +Epoch: [22] [ 950/7269] eta: 0:46:09 lr: 0.000005 loss: 0.5276 (0.5189) time: 0.4786 data: 0.0002 max mem: 52150 +Epoch: [22] [ 960/7269] eta: 0:46:07 lr: 0.000005 loss: 0.4962 (0.5190) time: 0.4794 data: 0.0002 max mem: 52150 +Epoch: [22] [ 970/7269] eta: 0:46:04 lr: 0.000005 loss: 0.4962 (0.5189) time: 0.4632 data: 0.0002 max mem: 52150 +Epoch: [22] [ 980/7269] eta: 0:46:07 lr: 0.000005 loss: 0.5275 (0.5192) time: 0.5108 data: 0.0002 max mem: 52150 +Epoch: [22] [ 990/7269] eta: 0:45:58 lr: 0.000005 loss: 0.5340 (0.5192) time: 0.4662 data: 0.0002 max mem: 52150 +Epoch: [22] [1000/7269] eta: 0:45:56 lr: 0.000005 loss: 0.5088 (0.5192) time: 0.4195 data: 0.0001 max mem: 52150 +Epoch: [22] [1010/7269] eta: 0:45:47 lr: 0.000005 loss: 0.5501 (0.5196) time: 0.4195 data: 0.0001 max mem: 52150 +Epoch: [22] [1020/7269] eta: 0:45:39 lr: 0.000005 loss: 0.5511 (0.5196) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [22] [1030/7269] eta: 0:45:30 lr: 0.000005 loss: 0.5314 (0.5199) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [22] [1040/7269] eta: 0:45:22 lr: 0.000005 loss: 0.5401 (0.5199) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [22] [1050/7269] eta: 0:45:26 lr: 0.000005 loss: 0.5415 (0.5203) time: 0.4797 data: 0.0002 max mem: 52150 +Epoch: [22] [1060/7269] eta: 0:45:25 lr: 0.000005 loss: 0.5451 (0.5203) time: 0.5400 data: 0.0002 max mem: 52150 +Epoch: [22] [1070/7269] eta: 0:45:17 lr: 0.000005 loss: 0.5270 (0.5205) time: 0.4320 data: 0.0002 max mem: 52150 +Epoch: [22] [1080/7269] eta: 0:45:09 lr: 0.000005 loss: 0.5386 (0.5206) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [22] [1090/7269] eta: 0:45:01 lr: 0.000005 loss: 0.5304 (0.5205) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [22] [1100/7269] eta: 0:44:53 lr: 0.000005 loss: 0.4975 (0.5202) time: 0.3722 data: 0.0001 max mem: 52150 +Epoch: [22] [1110/7269] eta: 0:44:45 lr: 0.000005 loss: 0.5016 (0.5202) time: 0.3721 data: 0.0001 max mem: 52150 +Epoch: [22] [1120/7269] eta: 0:44:37 lr: 0.000005 loss: 0.5214 (0.5203) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [22] [1130/7269] eta: 0:44:42 lr: 0.000005 loss: 0.5285 (0.5204) time: 0.4939 data: 0.0001 max mem: 52150 +Epoch: [22] [1140/7269] eta: 0:44:34 lr: 0.000005 loss: 0.5285 (0.5205) time: 0.4940 data: 0.0001 max mem: 52150 +Epoch: [22] [1150/7269] eta: 0:44:33 lr: 0.000005 loss: 0.4993 (0.5202) time: 0.4340 data: 0.0001 max mem: 52150 +Epoch: [22] [1160/7269] eta: 0:44:26 lr: 0.000005 loss: 0.5287 (0.5207) time: 0.4345 data: 0.0001 max mem: 52150 +Epoch: [22] [1170/7269] eta: 0:44:24 lr: 0.000005 loss: 0.5493 (0.5203) time: 0.4314 data: 0.0001 max mem: 52150 +Epoch: [22] [1180/7269] eta: 0:44:16 lr: 0.000005 loss: 0.4734 (0.5201) time: 0.4313 data: 0.0002 max mem: 52150 +Epoch: [22] [1190/7269] eta: 0:44:09 lr: 0.000005 loss: 0.4970 (0.5203) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [22] [1200/7269] eta: 0:44:01 lr: 0.000005 loss: 0.5416 (0.5203) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [22] [1210/7269] eta: 0:43:54 lr: 0.000005 loss: 0.5416 (0.5205) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [22] [1220/7269] eta: 0:43:46 lr: 0.000005 loss: 0.5289 (0.5203) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [22] [1230/7269] eta: 0:43:48 lr: 0.000005 loss: 0.5240 (0.5203) time: 0.4700 data: 0.0002 max mem: 52150 +Epoch: [22] [1240/7269] eta: 0:43:41 lr: 0.000005 loss: 0.5303 (0.5206) time: 0.4691 data: 0.0001 max mem: 52150 +Epoch: [22] [1250/7269] eta: 0:43:33 lr: 0.000005 loss: 0.5393 (0.5209) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [22] [1260/7269] eta: 0:43:26 lr: 0.000005 loss: 0.5412 (0.5210) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [22] [1270/7269] eta: 0:43:19 lr: 0.000005 loss: 0.5191 (0.5211) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [22] [1280/7269] eta: 0:43:16 lr: 0.000005 loss: 0.5280 (0.5214) time: 0.4161 data: 0.0001 max mem: 52150 +Epoch: [22] [1290/7269] eta: 0:43:13 lr: 0.000005 loss: 0.5323 (0.5213) time: 0.4626 data: 0.0002 max mem: 52150 +Epoch: [22] [1300/7269] eta: 0:43:06 lr: 0.000005 loss: 0.5123 (0.5213) time: 0.4184 data: 0.0002 max mem: 52150 +Epoch: [22] [1310/7269] eta: 0:43:03 lr: 0.000005 loss: 0.5141 (0.5213) time: 0.4193 data: 0.0001 max mem: 52150 +Epoch: [22] [1320/7269] eta: 0:43:00 lr: 0.000005 loss: 0.5076 (0.5211) time: 0.4676 data: 0.0002 max mem: 52150 +Epoch: [22] [1330/7269] eta: 0:42:53 lr: 0.000005 loss: 0.5076 (0.5210) time: 0.4206 data: 0.0001 max mem: 52150 +Epoch: [22] [1340/7269] eta: 0:42:46 lr: 0.000005 loss: 0.5494 (0.5212) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [22] [1350/7269] eta: 0:42:43 lr: 0.000005 loss: 0.5504 (0.5214) time: 0.4159 data: 0.0001 max mem: 52150 +Epoch: [22] [1360/7269] eta: 0:42:36 lr: 0.000005 loss: 0.5333 (0.5215) time: 0.4167 data: 0.0002 max mem: 52150 +Epoch: [22] [1370/7269] eta: 0:42:33 lr: 0.000005 loss: 0.5525 (0.5218) time: 0.4190 data: 0.0002 max mem: 52150 +Epoch: [22] [1380/7269] eta: 0:42:30 lr: 0.000005 loss: 0.5525 (0.5215) time: 0.4665 data: 0.0002 max mem: 52150 +Epoch: [22] [1390/7269] eta: 0:42:23 lr: 0.000005 loss: 0.5059 (0.5217) time: 0.4202 data: 0.0001 max mem: 52150 +Epoch: [22] [1400/7269] eta: 0:42:20 lr: 0.000005 loss: 0.5263 (0.5218) time: 0.4163 data: 0.0002 max mem: 52150 +Epoch: [22] [1410/7269] eta: 0:42:17 lr: 0.000005 loss: 0.5308 (0.5218) time: 0.4626 data: 0.0002 max mem: 52150 +Epoch: [22] [1420/7269] eta: 0:42:19 lr: 0.000005 loss: 0.5308 (0.5218) time: 0.5240 data: 0.0001 max mem: 52150 +Epoch: [22] [1430/7269] eta: 0:42:12 lr: 0.000005 loss: 0.5448 (0.5220) time: 0.4775 data: 0.0001 max mem: 52150 +Epoch: [22] [1440/7269] eta: 0:42:09 lr: 0.000005 loss: 0.5187 (0.5219) time: 0.4184 data: 0.0001 max mem: 52150 +Epoch: [22] [1450/7269] eta: 0:42:06 lr: 0.000005 loss: 0.5292 (0.5220) time: 0.4648 data: 0.0001 max mem: 52150 +Epoch: [22] [1460/7269] eta: 0:41:59 lr: 0.000005 loss: 0.5241 (0.5215) time: 0.4180 data: 0.0002 max mem: 52150 +Epoch: [22] [1470/7269] eta: 0:41:56 lr: 0.000005 loss: 0.4568 (0.5213) time: 0.4187 data: 0.0002 max mem: 52150 +Epoch: [22] [1480/7269] eta: 0:41:53 lr: 0.000005 loss: 0.5053 (0.5214) time: 0.4643 data: 0.0002 max mem: 52150 +Epoch: [22] [1490/7269] eta: 0:41:46 lr: 0.000005 loss: 0.5599 (0.5217) time: 0.4172 data: 0.0001 max mem: 52150 +Epoch: [22] [1500/7269] eta: 0:41:44 lr: 0.000005 loss: 0.5439 (0.5217) time: 0.4320 data: 0.0001 max mem: 52150 +Epoch: [22] [1510/7269] eta: 0:41:37 lr: 0.000005 loss: 0.5198 (0.5218) time: 0.4322 data: 0.0001 max mem: 52150 +Epoch: [22] [1520/7269] eta: 0:41:34 lr: 0.000005 loss: 0.5690 (0.5220) time: 0.4186 data: 0.0001 max mem: 52150 +Epoch: [22] [1530/7269] eta: 0:41:27 lr: 0.000005 loss: 0.5491 (0.5221) time: 0.4189 data: 0.0001 max mem: 52150 +Epoch: [22] [1540/7269] eta: 0:41:28 lr: 0.000005 loss: 0.5451 (0.5221) time: 0.4649 data: 0.0001 max mem: 52150 +Epoch: [22] [1550/7269] eta: 0:41:25 lr: 0.000005 loss: 0.5233 (0.5219) time: 0.5127 data: 0.0002 max mem: 52150 +Epoch: [22] [1560/7269] eta: 0:41:18 lr: 0.000005 loss: 0.5241 (0.5221) time: 0.4199 data: 0.0002 max mem: 52150 +Epoch: [22] [1570/7269] eta: 0:41:16 lr: 0.000005 loss: 0.5270 (0.5219) time: 0.4313 data: 0.0001 max mem: 52150 +Epoch: [22] [1580/7269] eta: 0:41:13 lr: 0.000005 loss: 0.5270 (0.5221) time: 0.4929 data: 0.0001 max mem: 52150 +Epoch: [22] [1590/7269] eta: 0:41:11 lr: 0.000005 loss: 0.5457 (0.5221) time: 0.4929 data: 0.0001 max mem: 52150 +Epoch: [22] [1600/7269] eta: 0:41:09 lr: 0.000005 loss: 0.5422 (0.5219) time: 0.4965 data: 0.0002 max mem: 52150 +Epoch: [22] [1610/7269] eta: 0:41:03 lr: 0.000005 loss: 0.4916 (0.5218) time: 0.4371 data: 0.0001 max mem: 52150 +Epoch: [22] [1620/7269] eta: 0:40:56 lr: 0.000005 loss: 0.5176 (0.5218) time: 0.3721 data: 0.0001 max mem: 52150 +Epoch: [22] [1630/7269] eta: 0:40:49 lr: 0.000005 loss: 0.5409 (0.5219) time: 0.3722 data: 0.0001 max mem: 52150 +Epoch: [22] [1640/7269] eta: 0:40:51 lr: 0.000005 loss: 0.5206 (0.5218) time: 0.4937 data: 0.0002 max mem: 52150 +Epoch: [22] [1650/7269] eta: 0:40:45 lr: 0.000005 loss: 0.5125 (0.5218) time: 0.4936 data: 0.0002 max mem: 52150 +Epoch: [22] [1660/7269] eta: 0:40:38 lr: 0.000005 loss: 0.5310 (0.5220) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [22] [1670/7269] eta: 0:40:32 lr: 0.000005 loss: 0.5444 (0.5221) time: 0.3721 data: 0.0001 max mem: 52150 +Epoch: [22] [1680/7269] eta: 0:40:25 lr: 0.000005 loss: 0.5397 (0.5220) time: 0.3721 data: 0.0001 max mem: 52150 +Epoch: [22] [1690/7269] eta: 0:40:22 lr: 0.000005 loss: 0.5363 (0.5221) time: 0.4204 data: 0.0001 max mem: 52150 +Epoch: [22] [1700/7269] eta: 0:40:22 lr: 0.000005 loss: 0.5461 (0.5224) time: 0.5136 data: 0.0001 max mem: 52150 +Epoch: [22] [1710/7269] eta: 0:40:16 lr: 0.000005 loss: 0.5461 (0.5224) time: 0.4654 data: 0.0001 max mem: 52150 +Epoch: [22] [1720/7269] eta: 0:40:09 lr: 0.000005 loss: 0.5228 (0.5225) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [22] [1730/7269] eta: 0:40:03 lr: 0.000005 loss: 0.5313 (0.5226) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [22] [1740/7269] eta: 0:39:57 lr: 0.000005 loss: 0.5255 (0.5225) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [22] [1750/7269] eta: 0:39:50 lr: 0.000005 loss: 0.4965 (0.5222) time: 0.3724 data: 0.0001 max mem: 52150 +Epoch: [22] [1760/7269] eta: 0:39:47 lr: 0.000005 loss: 0.4771 (0.5222) time: 0.4170 data: 0.0002 max mem: 52150 +Epoch: [22] [1770/7269] eta: 0:39:41 lr: 0.000005 loss: 0.4771 (0.5220) time: 0.4167 data: 0.0002 max mem: 52150 +Epoch: [22] [1780/7269] eta: 0:39:37 lr: 0.000005 loss: 0.4622 (0.5218) time: 0.4190 data: 0.0002 max mem: 52150 +Epoch: [22] [1790/7269] eta: 0:39:35 lr: 0.000005 loss: 0.4817 (0.5215) time: 0.4801 data: 0.0002 max mem: 52150 +Epoch: [22] [1800/7269] eta: 0:39:29 lr: 0.000005 loss: 0.4908 (0.5215) time: 0.4327 data: 0.0002 max mem: 52150 +Epoch: [22] [1810/7269] eta: 0:39:25 lr: 0.000005 loss: 0.5217 (0.5215) time: 0.4163 data: 0.0002 max mem: 52150 +Epoch: [22] [1820/7269] eta: 0:39:19 lr: 0.000005 loss: 0.5313 (0.5215) time: 0.4164 data: 0.0002 max mem: 52150 +Epoch: [22] [1830/7269] eta: 0:39:13 lr: 0.000005 loss: 0.5142 (0.5215) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [22] [1840/7269] eta: 0:39:09 lr: 0.000005 loss: 0.5504 (0.5217) time: 0.4179 data: 0.0002 max mem: 52150 +Epoch: [22] [1850/7269] eta: 0:39:03 lr: 0.000005 loss: 0.5504 (0.5216) time: 0.4188 data: 0.0002 max mem: 52150 +Epoch: [22] [1860/7269] eta: 0:38:57 lr: 0.000005 loss: 0.5177 (0.5214) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [22] [1870/7269] eta: 0:38:57 lr: 0.000005 loss: 0.5047 (0.5214) time: 0.4638 data: 0.0002 max mem: 52150 +Epoch: [22] [1880/7269] eta: 0:38:56 lr: 0.000005 loss: 0.5333 (0.5215) time: 0.5669 data: 0.0002 max mem: 52150 +Epoch: [22] [1890/7269] eta: 0:38:50 lr: 0.000005 loss: 0.5196 (0.5213) time: 0.4749 data: 0.0002 max mem: 52150 +Epoch: [22] [1900/7269] eta: 0:38:44 lr: 0.000005 loss: 0.4880 (0.5212) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [22] [1910/7269] eta: 0:38:38 lr: 0.000005 loss: 0.4960 (0.5212) time: 0.3713 data: 0.0001 max mem: 52150 +Epoch: [22] [1920/7269] eta: 0:38:37 lr: 0.000005 loss: 0.5121 (0.5211) time: 0.4626 data: 0.0001 max mem: 52150 +Epoch: [22] [1930/7269] eta: 0:38:34 lr: 0.000005 loss: 0.5071 (0.5212) time: 0.5071 data: 0.0002 max mem: 52150 +Epoch: [22] [1940/7269] eta: 0:38:28 lr: 0.000005 loss: 0.5020 (0.5210) time: 0.4160 data: 0.0002 max mem: 52150 +Epoch: [22] [1950/7269] eta: 0:38:22 lr: 0.000005 loss: 0.5111 (0.5211) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [22] [1960/7269] eta: 0:38:18 lr: 0.000005 loss: 0.5136 (0.5210) time: 0.4187 data: 0.0001 max mem: 52150 +Epoch: [22] [1970/7269] eta: 0:38:15 lr: 0.000005 loss: 0.5032 (0.5210) time: 0.4799 data: 0.0001 max mem: 52150 +Epoch: [22] [1980/7269] eta: 0:38:14 lr: 0.000005 loss: 0.5215 (0.5210) time: 0.5250 data: 0.0002 max mem: 52150 +Epoch: [22] [1990/7269] eta: 0:38:08 lr: 0.000005 loss: 0.5104 (0.5210) time: 0.4643 data: 0.0002 max mem: 52150 +Epoch: [22] [2000/7269] eta: 0:38:02 lr: 0.000005 loss: 0.5307 (0.5211) time: 0.3721 data: 0.0001 max mem: 52150 +Epoch: [22] [2010/7269] eta: 0:37:56 lr: 0.000005 loss: 0.5307 (0.5211) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [22] [2020/7269] eta: 0:37:55 lr: 0.000005 loss: 0.4872 (0.5208) time: 0.4618 data: 0.0001 max mem: 52150 +Epoch: [22] [2030/7269] eta: 0:37:49 lr: 0.000005 loss: 0.4573 (0.5207) time: 0.4615 data: 0.0001 max mem: 52150 +Epoch: [22] [2040/7269] eta: 0:37:46 lr: 0.000005 loss: 0.5347 (0.5208) time: 0.4321 data: 0.0002 max mem: 52150 +Epoch: [22] [2050/7269] eta: 0:37:41 lr: 0.000005 loss: 0.5414 (0.5209) time: 0.4327 data: 0.0002 max mem: 52150 +Epoch: [22] [2060/7269] eta: 0:37:37 lr: 0.000005 loss: 0.5445 (0.5210) time: 0.4188 data: 0.0001 max mem: 52150 +Epoch: [22] [2070/7269] eta: 0:37:34 lr: 0.000005 loss: 0.5475 (0.5209) time: 0.4646 data: 0.0001 max mem: 52150 +Epoch: [22] [2080/7269] eta: 0:37:28 lr: 0.000005 loss: 0.4582 (0.5206) time: 0.4177 data: 0.0002 max mem: 52150 +Epoch: [22] [2090/7269] eta: 0:37:22 lr: 0.000005 loss: 0.4582 (0.5204) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [22] [2100/7269] eta: 0:37:16 lr: 0.000005 loss: 0.5325 (0.5205) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [22] [2110/7269] eta: 0:37:13 lr: 0.000005 loss: 0.5325 (0.5205) time: 0.4317 data: 0.0002 max mem: 52150 +Epoch: [22] [2120/7269] eta: 0:37:10 lr: 0.000005 loss: 0.5318 (0.5206) time: 0.4924 data: 0.0002 max mem: 52150 +Epoch: [22] [2130/7269] eta: 0:37:04 lr: 0.000005 loss: 0.5276 (0.5206) time: 0.4333 data: 0.0002 max mem: 52150 +Epoch: [22] [2140/7269] eta: 0:36:59 lr: 0.000005 loss: 0.5276 (0.5207) time: 0.3728 data: 0.0001 max mem: 52150 +Epoch: [22] [2150/7269] eta: 0:36:53 lr: 0.000005 loss: 0.5509 (0.5208) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [22] [2160/7269] eta: 0:36:47 lr: 0.000005 loss: 0.5507 (0.5208) time: 0.3713 data: 0.0001 max mem: 52150 +Epoch: [22] [2170/7269] eta: 0:36:44 lr: 0.000005 loss: 0.5475 (0.5209) time: 0.4312 data: 0.0002 max mem: 52150 +Epoch: [22] [2180/7269] eta: 0:36:38 lr: 0.000005 loss: 0.5352 (0.5211) time: 0.4319 data: 0.0002 max mem: 52150 +Epoch: [22] [2190/7269] eta: 0:36:36 lr: 0.000005 loss: 0.5288 (0.5210) time: 0.4347 data: 0.0001 max mem: 52150 +Epoch: [22] [2200/7269] eta: 0:36:30 lr: 0.000005 loss: 0.5116 (0.5211) time: 0.4344 data: 0.0001 max mem: 52150 +Epoch: [22] [2210/7269] eta: 0:36:27 lr: 0.000005 loss: 0.5116 (0.5209) time: 0.4334 data: 0.0002 max mem: 52150 +Epoch: [22] [2220/7269] eta: 0:36:21 lr: 0.000005 loss: 0.5060 (0.5210) time: 0.4334 data: 0.0002 max mem: 52150 +Epoch: [22] [2230/7269] eta: 0:36:16 lr: 0.000005 loss: 0.5372 (0.5210) time: 0.3713 data: 0.0001 max mem: 52150 +Epoch: [22] [2240/7269] eta: 0:36:13 lr: 0.000005 loss: 0.5541 (0.5211) time: 0.4322 data: 0.0001 max mem: 52150 +Epoch: [22] [2250/7269] eta: 0:36:13 lr: 0.000005 loss: 0.5541 (0.5210) time: 0.5565 data: 0.0002 max mem: 52150 +Epoch: [22] [2260/7269] eta: 0:36:07 lr: 0.000005 loss: 0.5288 (0.5210) time: 0.4956 data: 0.0002 max mem: 52150 +Epoch: [22] [2270/7269] eta: 0:36:01 lr: 0.000005 loss: 0.5351 (0.5212) time: 0.3724 data: 0.0001 max mem: 52150 +Epoch: [22] [2280/7269] eta: 0:35:56 lr: 0.000005 loss: 0.5260 (0.5212) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [22] [2290/7269] eta: 0:35:52 lr: 0.000005 loss: 0.5107 (0.5211) time: 0.4188 data: 0.0002 max mem: 52150 +Epoch: [22] [2300/7269] eta: 0:35:46 lr: 0.000005 loss: 0.5229 (0.5211) time: 0.4195 data: 0.0002 max mem: 52150 +Epoch: [22] [2310/7269] eta: 0:35:41 lr: 0.000005 loss: 0.5229 (0.5212) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [22] [2320/7269] eta: 0:35:35 lr: 0.000005 loss: 0.5072 (0.5210) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [22] [2330/7269] eta: 0:35:34 lr: 0.000005 loss: 0.4492 (0.5209) time: 0.4688 data: 0.0002 max mem: 52150 +Epoch: [22] [2340/7269] eta: 0:35:33 lr: 0.000005 loss: 0.5544 (0.5211) time: 0.5763 data: 0.0001 max mem: 52150 +Epoch: [22] [2350/7269] eta: 0:35:27 lr: 0.000005 loss: 0.5544 (0.5211) time: 0.4794 data: 0.0001 max mem: 52150 +Epoch: [22] [2360/7269] eta: 0:35:23 lr: 0.000005 loss: 0.5311 (0.5210) time: 0.4167 data: 0.0001 max mem: 52150 +Epoch: [22] [2370/7269] eta: 0:35:20 lr: 0.000005 loss: 0.4747 (0.5209) time: 0.4628 data: 0.0002 max mem: 52150 +Epoch: [22] [2380/7269] eta: 0:35:14 lr: 0.000005 loss: 0.4801 (0.5209) time: 0.4179 data: 0.0002 max mem: 52150 +Epoch: [22] [2390/7269] eta: 0:35:08 lr: 0.000005 loss: 0.5306 (0.5209) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [22] [2400/7269] eta: 0:35:03 lr: 0.000005 loss: 0.5394 (0.5210) time: 0.3714 data: 0.0001 max mem: 52150 +Epoch: [22] [2410/7269] eta: 0:35:00 lr: 0.000005 loss: 0.5439 (0.5210) time: 0.4359 data: 0.0001 max mem: 52150 +Epoch: [22] [2420/7269] eta: 0:34:58 lr: 0.000005 loss: 0.5318 (0.5210) time: 0.5297 data: 0.0001 max mem: 52150 +Epoch: [22] [2430/7269] eta: 0:34:57 lr: 0.000005 loss: 0.5318 (0.5210) time: 0.5708 data: 0.0001 max mem: 52150 +Epoch: [22] [2440/7269] eta: 0:34:51 lr: 0.000005 loss: 0.4842 (0.5208) time: 0.4771 data: 0.0002 max mem: 52150 +Epoch: [22] [2450/7269] eta: 0:34:46 lr: 0.000005 loss: 0.4776 (0.5207) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [22] [2460/7269] eta: 0:34:42 lr: 0.000005 loss: 0.5247 (0.5208) time: 0.4160 data: 0.0001 max mem: 52150 +Epoch: [22] [2470/7269] eta: 0:34:36 lr: 0.000005 loss: 0.5231 (0.5208) time: 0.4160 data: 0.0001 max mem: 52150 +Epoch: [22] [2480/7269] eta: 0:34:33 lr: 0.000005 loss: 0.5581 (0.5208) time: 0.4188 data: 0.0001 max mem: 52150 +Epoch: [22] [2490/7269] eta: 0:34:29 lr: 0.000005 loss: 0.5581 (0.5208) time: 0.4663 data: 0.0001 max mem: 52150 +Epoch: [22] [2500/7269] eta: 0:34:24 lr: 0.000005 loss: 0.4679 (0.5206) time: 0.4187 data: 0.0002 max mem: 52150 +Epoch: [22] [2510/7269] eta: 0:34:21 lr: 0.000005 loss: 0.4729 (0.5204) time: 0.4623 data: 0.0002 max mem: 52150 +Epoch: [22] [2520/7269] eta: 0:34:20 lr: 0.000005 loss: 0.4933 (0.5204) time: 0.5727 data: 0.0002 max mem: 52150 +Epoch: [22] [2530/7269] eta: 0:34:16 lr: 0.000005 loss: 0.5219 (0.5204) time: 0.5266 data: 0.0001 max mem: 52150 +Epoch: [22] [2540/7269] eta: 0:34:11 lr: 0.000005 loss: 0.5219 (0.5204) time: 0.4169 data: 0.0001 max mem: 52150 +Epoch: [22] [2550/7269] eta: 0:34:05 lr: 0.000005 loss: 0.5285 (0.5204) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [22] [2560/7269] eta: 0:34:02 lr: 0.000005 loss: 0.5349 (0.5204) time: 0.4179 data: 0.0001 max mem: 52150 +Epoch: [22] [2570/7269] eta: 0:33:56 lr: 0.000005 loss: 0.5172 (0.5203) time: 0.4180 data: 0.0002 max mem: 52150 +Epoch: [22] [2580/7269] eta: 0:33:51 lr: 0.000005 loss: 0.4878 (0.5202) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [22] [2590/7269] eta: 0:33:45 lr: 0.000005 loss: 0.4854 (0.5201) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [22] [2600/7269] eta: 0:33:40 lr: 0.000005 loss: 0.4892 (0.5201) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [22] [2610/7269] eta: 0:33:34 lr: 0.000005 loss: 0.5083 (0.5202) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [22] [2620/7269] eta: 0:33:29 lr: 0.000005 loss: 0.5421 (0.5202) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [22] [2630/7269] eta: 0:33:24 lr: 0.000005 loss: 0.5378 (0.5202) time: 0.3712 data: 0.0002 max mem: 52150 +Epoch: [22] [2640/7269] eta: 0:33:18 lr: 0.000005 loss: 0.5349 (0.5203) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [22] [2650/7269] eta: 0:33:15 lr: 0.000005 loss: 0.5643 (0.5205) time: 0.4327 data: 0.0002 max mem: 52150 +Epoch: [22] [2660/7269] eta: 0:33:10 lr: 0.000005 loss: 0.5469 (0.5205) time: 0.4327 data: 0.0002 max mem: 52150 +Epoch: [22] [2670/7269] eta: 0:33:04 lr: 0.000005 loss: 0.5174 (0.5205) time: 0.3727 data: 0.0002 max mem: 52150 +Epoch: [22] [2680/7269] eta: 0:33:03 lr: 0.000005 loss: 0.5229 (0.5206) time: 0.4933 data: 0.0002 max mem: 52150 +Epoch: [22] [2690/7269] eta: 0:32:58 lr: 0.000005 loss: 0.5287 (0.5205) time: 0.4928 data: 0.0001 max mem: 52150 +Epoch: [22] [2700/7269] eta: 0:32:52 lr: 0.000005 loss: 0.5468 (0.5207) time: 0.3713 data: 0.0001 max mem: 52150 +Epoch: [22] [2710/7269] eta: 0:32:47 lr: 0.000005 loss: 0.5446 (0.5207) time: 0.3709 data: 0.0002 max mem: 52150 +Epoch: [22] [2720/7269] eta: 0:32:42 lr: 0.000005 loss: 0.5373 (0.5208) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [22] [2730/7269] eta: 0:32:38 lr: 0.000005 loss: 0.5510 (0.5208) time: 0.4349 data: 0.0002 max mem: 52150 +Epoch: [22] [2740/7269] eta: 0:32:33 lr: 0.000005 loss: 0.4651 (0.5207) time: 0.4347 data: 0.0002 max mem: 52150 +Epoch: [22] [2750/7269] eta: 0:32:28 lr: 0.000005 loss: 0.5033 (0.5206) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [22] [2760/7269] eta: 0:32:25 lr: 0.000005 loss: 0.4838 (0.5204) time: 0.4336 data: 0.0001 max mem: 52150 +Epoch: [22] [2770/7269] eta: 0:32:19 lr: 0.000005 loss: 0.5218 (0.5204) time: 0.4336 data: 0.0001 max mem: 52150 +Epoch: [22] [2780/7269] eta: 0:32:16 lr: 0.000005 loss: 0.5207 (0.5202) time: 0.4335 data: 0.0001 max mem: 52150 +Epoch: [22] [2790/7269] eta: 0:32:12 lr: 0.000005 loss: 0.4757 (0.5201) time: 0.4785 data: 0.0001 max mem: 52150 +Epoch: [22] [2800/7269] eta: 0:32:07 lr: 0.000005 loss: 0.5150 (0.5201) time: 0.4166 data: 0.0002 max mem: 52150 +Epoch: [22] [2810/7269] eta: 0:32:03 lr: 0.000005 loss: 0.5494 (0.5202) time: 0.4160 data: 0.0002 max mem: 52150 +Epoch: [22] [2820/7269] eta: 0:31:58 lr: 0.000005 loss: 0.5465 (0.5202) time: 0.4164 data: 0.0002 max mem: 52150 +Epoch: [22] [2830/7269] eta: 0:31:53 lr: 0.000005 loss: 0.5223 (0.5202) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [22] [2840/7269] eta: 0:31:49 lr: 0.000005 loss: 0.5161 (0.5201) time: 0.4308 data: 0.0001 max mem: 52150 +Epoch: [22] [2850/7269] eta: 0:31:44 lr: 0.000005 loss: 0.5211 (0.5201) time: 0.4305 data: 0.0002 max mem: 52150 +Epoch: [22] [2860/7269] eta: 0:31:41 lr: 0.000005 loss: 0.5381 (0.5202) time: 0.4318 data: 0.0002 max mem: 52150 +Epoch: [22] [2870/7269] eta: 0:31:35 lr: 0.000005 loss: 0.5360 (0.5201) time: 0.4318 data: 0.0002 max mem: 52150 +Epoch: [22] [2880/7269] eta: 0:31:30 lr: 0.000005 loss: 0.5341 (0.5201) time: 0.3713 data: 0.0001 max mem: 52150 +Epoch: [22] [2890/7269] eta: 0:31:27 lr: 0.000005 loss: 0.5343 (0.5200) time: 0.4315 data: 0.0001 max mem: 52150 +Epoch: [22] [2900/7269] eta: 0:31:22 lr: 0.000005 loss: 0.5348 (0.5200) time: 0.4314 data: 0.0001 max mem: 52150 +Epoch: [22] [2910/7269] eta: 0:31:18 lr: 0.000005 loss: 0.5348 (0.5200) time: 0.4188 data: 0.0002 max mem: 52150 +Epoch: [22] [2920/7269] eta: 0:31:18 lr: 0.000005 loss: 0.4815 (0.5199) time: 0.5892 data: 0.0001 max mem: 52150 +Epoch: [22] [2930/7269] eta: 0:31:14 lr: 0.000005 loss: 0.5329 (0.5199) time: 0.6002 data: 0.0001 max mem: 52150 +Epoch: [22] [2940/7269] eta: 0:31:09 lr: 0.000005 loss: 0.5466 (0.5200) time: 0.4302 data: 0.0002 max mem: 52150 +Epoch: [22] [2950/7269] eta: 0:31:04 lr: 0.000005 loss: 0.5466 (0.5201) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [22] [2960/7269] eta: 0:31:00 lr: 0.000005 loss: 0.5621 (0.5202) time: 0.4174 data: 0.0001 max mem: 52150 +Epoch: [22] [2970/7269] eta: 0:30:57 lr: 0.000005 loss: 0.5438 (0.5202) time: 0.5085 data: 0.0001 max mem: 52150 +Epoch: [22] [2980/7269] eta: 0:30:53 lr: 0.000005 loss: 0.5220 (0.5201) time: 0.5086 data: 0.0001 max mem: 52150 +Epoch: [22] [2990/7269] eta: 0:30:48 lr: 0.000005 loss: 0.5220 (0.5201) time: 0.4169 data: 0.0001 max mem: 52150 +Epoch: [22] [3000/7269] eta: 0:30:47 lr: 0.000005 loss: 0.5429 (0.5202) time: 0.5102 data: 0.0001 max mem: 52150 +Epoch: [22] [3010/7269] eta: 0:30:45 lr: 0.000005 loss: 0.5422 (0.5201) time: 0.6184 data: 0.0001 max mem: 52150 +Epoch: [22] [3020/7269] eta: 0:30:40 lr: 0.000005 loss: 0.5096 (0.5201) time: 0.4799 data: 0.0001 max mem: 52150 +Epoch: [22] [3030/7269] eta: 0:30:36 lr: 0.000005 loss: 0.5184 (0.5201) time: 0.4185 data: 0.0001 max mem: 52150 +Epoch: [22] [3040/7269] eta: 0:30:31 lr: 0.000005 loss: 0.4924 (0.5200) time: 0.4178 data: 0.0002 max mem: 52150 +Epoch: [22] [3050/7269] eta: 0:30:25 lr: 0.000005 loss: 0.5210 (0.5200) time: 0.3712 data: 0.0002 max mem: 52150 +Epoch: [22] [3060/7269] eta: 0:30:22 lr: 0.000005 loss: 0.5497 (0.5202) time: 0.4166 data: 0.0002 max mem: 52150 +Epoch: [22] [3070/7269] eta: 0:30:18 lr: 0.000005 loss: 0.5435 (0.5202) time: 0.4779 data: 0.0002 max mem: 52150 +Epoch: [22] [3080/7269] eta: 0:30:13 lr: 0.000005 loss: 0.4888 (0.5202) time: 0.4325 data: 0.0002 max mem: 52150 +Epoch: [22] [3090/7269] eta: 0:30:10 lr: 0.000005 loss: 0.5021 (0.5201) time: 0.4603 data: 0.0002 max mem: 52150 +Epoch: [22] [3100/7269] eta: 0:30:05 lr: 0.000005 loss: 0.5021 (0.5200) time: 0.4607 data: 0.0002 max mem: 52150 +Epoch: [22] [3110/7269] eta: 0:30:00 lr: 0.000005 loss: 0.5418 (0.5201) time: 0.3727 data: 0.0002 max mem: 52150 +Epoch: [22] [3120/7269] eta: 0:29:55 lr: 0.000005 loss: 0.5150 (0.5200) time: 0.3729 data: 0.0002 max mem: 52150 +Epoch: [22] [3130/7269] eta: 0:29:50 lr: 0.000005 loss: 0.4732 (0.5199) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [22] [3140/7269] eta: 0:29:44 lr: 0.000005 loss: 0.4963 (0.5199) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [22] [3150/7269] eta: 0:29:39 lr: 0.000005 loss: 0.5399 (0.5200) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [22] [3160/7269] eta: 0:29:34 lr: 0.000005 loss: 0.5399 (0.5200) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [22] [3170/7269] eta: 0:29:29 lr: 0.000005 loss: 0.4977 (0.5200) time: 0.3713 data: 0.0001 max mem: 52150 +Epoch: [22] [3180/7269] eta: 0:29:24 lr: 0.000005 loss: 0.5463 (0.5201) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [22] [3190/7269] eta: 0:29:21 lr: 0.000005 loss: 0.5463 (0.5201) time: 0.4334 data: 0.0001 max mem: 52150 +Epoch: [22] [3200/7269] eta: 0:29:15 lr: 0.000005 loss: 0.5230 (0.5202) time: 0.4333 data: 0.0001 max mem: 52150 +Epoch: [22] [3210/7269] eta: 0:29:12 lr: 0.000005 loss: 0.5671 (0.5204) time: 0.4316 data: 0.0001 max mem: 52150 +Epoch: [22] [3220/7269] eta: 0:29:07 lr: 0.000005 loss: 0.5085 (0.5202) time: 0.4313 data: 0.0001 max mem: 52150 +Epoch: [22] [3230/7269] eta: 0:29:03 lr: 0.000005 loss: 0.4805 (0.5202) time: 0.4320 data: 0.0002 max mem: 52150 +Epoch: [22] [3240/7269] eta: 0:28:58 lr: 0.000005 loss: 0.4773 (0.5201) time: 0.4316 data: 0.0002 max mem: 52150 +Epoch: [22] [3250/7269] eta: 0:28:55 lr: 0.000005 loss: 0.5050 (0.5200) time: 0.4322 data: 0.0001 max mem: 52150 +Epoch: [22] [3260/7269] eta: 0:28:50 lr: 0.000005 loss: 0.5180 (0.5201) time: 0.4330 data: 0.0001 max mem: 52150 +Epoch: [22] [3270/7269] eta: 0:28:45 lr: 0.000005 loss: 0.5083 (0.5200) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [22] [3280/7269] eta: 0:28:39 lr: 0.000005 loss: 0.4860 (0.5199) time: 0.3712 data: 0.0002 max mem: 52150 +Epoch: [22] [3290/7269] eta: 0:28:34 lr: 0.000005 loss: 0.5222 (0.5199) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [22] [3300/7269] eta: 0:28:31 lr: 0.000005 loss: 0.5368 (0.5199) time: 0.4350 data: 0.0001 max mem: 52150 +Epoch: [22] [3310/7269] eta: 0:28:26 lr: 0.000005 loss: 0.5368 (0.5200) time: 0.4344 data: 0.0001 max mem: 52150 +Epoch: [22] [3320/7269] eta: 0:28:22 lr: 0.000005 loss: 0.5371 (0.5201) time: 0.4304 data: 0.0001 max mem: 52150 +Epoch: [22] [3330/7269] eta: 0:28:18 lr: 0.000005 loss: 0.5275 (0.5200) time: 0.4761 data: 0.0001 max mem: 52150 +Epoch: [22] [3340/7269] eta: 0:28:13 lr: 0.000005 loss: 0.5270 (0.5200) time: 0.4164 data: 0.0001 max mem: 52150 +Epoch: [22] [3350/7269] eta: 0:28:08 lr: 0.000005 loss: 0.5363 (0.5200) time: 0.3708 data: 0.0002 max mem: 52150 +Epoch: [22] [3360/7269] eta: 0:28:05 lr: 0.000005 loss: 0.5354 (0.5200) time: 0.4318 data: 0.0002 max mem: 52150 +Epoch: [22] [3370/7269] eta: 0:28:00 lr: 0.000005 loss: 0.5206 (0.5200) time: 0.4319 data: 0.0002 max mem: 52150 +Epoch: [22] [3380/7269] eta: 0:27:56 lr: 0.000005 loss: 0.5247 (0.5200) time: 0.4340 data: 0.0001 max mem: 52150 +Epoch: [22] [3390/7269] eta: 0:27:52 lr: 0.000005 loss: 0.5514 (0.5202) time: 0.4799 data: 0.0002 max mem: 52150 +Epoch: [22] [3400/7269] eta: 0:27:48 lr: 0.000005 loss: 0.5514 (0.5202) time: 0.4657 data: 0.0002 max mem: 52150 +Epoch: [22] [3410/7269] eta: 0:27:43 lr: 0.000005 loss: 0.5155 (0.5202) time: 0.4200 data: 0.0001 max mem: 52150 +Epoch: [22] [3420/7269] eta: 0:27:38 lr: 0.000005 loss: 0.5155 (0.5201) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [22] [3430/7269] eta: 0:27:34 lr: 0.000005 loss: 0.5325 (0.5202) time: 0.4175 data: 0.0001 max mem: 52150 +Epoch: [22] [3440/7269] eta: 0:27:30 lr: 0.000005 loss: 0.5439 (0.5202) time: 0.4630 data: 0.0001 max mem: 52150 +Epoch: [22] [3450/7269] eta: 0:27:27 lr: 0.000005 loss: 0.5389 (0.5202) time: 0.4640 data: 0.0002 max mem: 52150 +Epoch: [22] [3460/7269] eta: 0:27:23 lr: 0.000005 loss: 0.4908 (0.5202) time: 0.4806 data: 0.0002 max mem: 52150 +Epoch: [22] [3470/7269] eta: 0:27:21 lr: 0.000005 loss: 0.4704 (0.5200) time: 0.5925 data: 0.0002 max mem: 52150 +Epoch: [22] [3480/7269] eta: 0:27:16 lr: 0.000005 loss: 0.4877 (0.5200) time: 0.5300 data: 0.0002 max mem: 52150 +Epoch: [22] [3490/7269] eta: 0:27:12 lr: 0.000005 loss: 0.4877 (0.5198) time: 0.4177 data: 0.0001 max mem: 52150 +Epoch: [22] [3500/7269] eta: 0:27:08 lr: 0.000005 loss: 0.5325 (0.5200) time: 0.4622 data: 0.0001 max mem: 52150 +Epoch: [22] [3510/7269] eta: 0:27:07 lr: 0.000005 loss: 0.5588 (0.5201) time: 0.5592 data: 0.0002 max mem: 52150 +Epoch: [22] [3520/7269] eta: 0:27:02 lr: 0.000005 loss: 0.5209 (0.5200) time: 0.5150 data: 0.0002 max mem: 52150 +Epoch: [22] [3530/7269] eta: 0:26:59 lr: 0.000005 loss: 0.4649 (0.5199) time: 0.4779 data: 0.0001 max mem: 52150 +Epoch: [22] [3540/7269] eta: 0:26:56 lr: 0.000005 loss: 0.5117 (0.5199) time: 0.5678 data: 0.0002 max mem: 52150 +Epoch: [22] [3550/7269] eta: 0:26:53 lr: 0.000005 loss: 0.5267 (0.5200) time: 0.5696 data: 0.0002 max mem: 52150 +Epoch: [22] [3560/7269] eta: 0:26:48 lr: 0.000005 loss: 0.5267 (0.5200) time: 0.4795 data: 0.0002 max mem: 52150 +Epoch: [22] [3570/7269] eta: 0:26:43 lr: 0.000005 loss: 0.4948 (0.5199) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [22] [3580/7269] eta: 0:26:38 lr: 0.000005 loss: 0.4948 (0.5199) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [22] [3590/7269] eta: 0:26:33 lr: 0.000005 loss: 0.5329 (0.5200) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [22] [3600/7269] eta: 0:26:28 lr: 0.000005 loss: 0.5272 (0.5200) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [22] [3610/7269] eta: 0:26:23 lr: 0.000005 loss: 0.5235 (0.5199) time: 0.3714 data: 0.0001 max mem: 52150 +Epoch: [22] [3620/7269] eta: 0:26:18 lr: 0.000005 loss: 0.5178 (0.5200) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [22] [3630/7269] eta: 0:26:13 lr: 0.000005 loss: 0.5346 (0.5200) time: 0.3710 data: 0.0001 max mem: 52150 +Epoch: [22] [3640/7269] eta: 0:26:08 lr: 0.000005 loss: 0.5107 (0.5199) time: 0.3709 data: 0.0002 max mem: 52150 +Epoch: [22] [3650/7269] eta: 0:26:04 lr: 0.000005 loss: 0.5059 (0.5199) time: 0.4160 data: 0.0002 max mem: 52150 +Epoch: [22] [3660/7269] eta: 0:25:59 lr: 0.000005 loss: 0.4905 (0.5199) time: 0.4162 data: 0.0001 max mem: 52150 +Epoch: [22] [3670/7269] eta: 0:25:56 lr: 0.000005 loss: 0.5412 (0.5199) time: 0.4323 data: 0.0002 max mem: 52150 +Epoch: [22] [3680/7269] eta: 0:25:51 lr: 0.000005 loss: 0.5553 (0.5200) time: 0.4323 data: 0.0001 max mem: 52150 +Epoch: [22] [3690/7269] eta: 0:25:46 lr: 0.000005 loss: 0.5513 (0.5201) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [22] [3700/7269] eta: 0:25:41 lr: 0.000005 loss: 0.5369 (0.5201) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [22] [3710/7269] eta: 0:25:36 lr: 0.000005 loss: 0.5353 (0.5201) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [22] [3720/7269] eta: 0:25:31 lr: 0.000005 loss: 0.5427 (0.5201) time: 0.3711 data: 0.0001 max mem: 52150 +Epoch: [22] [3730/7269] eta: 0:25:26 lr: 0.000005 loss: 0.5464 (0.5201) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [22] [3740/7269] eta: 0:25:21 lr: 0.000005 loss: 0.5296 (0.5201) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [22] [3750/7269] eta: 0:25:17 lr: 0.000005 loss: 0.5296 (0.5202) time: 0.3714 data: 0.0001 max mem: 52150 +Epoch: [22] [3760/7269] eta: 0:25:13 lr: 0.000005 loss: 0.5291 (0.5202) time: 0.4320 data: 0.0001 max mem: 52150 +Epoch: [22] [3770/7269] eta: 0:25:08 lr: 0.000005 loss: 0.5145 (0.5201) time: 0.4321 data: 0.0002 max mem: 52150 +Epoch: [22] [3780/7269] eta: 0:25:04 lr: 0.000005 loss: 0.5302 (0.5202) time: 0.4159 data: 0.0002 max mem: 52150 +Epoch: [22] [3790/7269] eta: 0:24:59 lr: 0.000005 loss: 0.5302 (0.5201) time: 0.4160 data: 0.0001 max mem: 52150 +Epoch: [22] [3800/7269] eta: 0:24:54 lr: 0.000005 loss: 0.4975 (0.5201) time: 0.3713 data: 0.0001 max mem: 52150 +Epoch: [22] [3810/7269] eta: 0:24:49 lr: 0.000005 loss: 0.5267 (0.5201) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [22] [3820/7269] eta: 0:24:46 lr: 0.000005 loss: 0.4874 (0.5199) time: 0.4317 data: 0.0002 max mem: 52150 +Epoch: [22] [3830/7269] eta: 0:24:42 lr: 0.000005 loss: 0.4836 (0.5198) time: 0.4812 data: 0.0001 max mem: 52150 +Epoch: [22] [3840/7269] eta: 0:24:37 lr: 0.000005 loss: 0.4766 (0.5198) time: 0.4215 data: 0.0001 max mem: 52150 +Epoch: [22] [3850/7269] eta: 0:24:33 lr: 0.000005 loss: 0.5009 (0.5197) time: 0.4339 data: 0.0001 max mem: 52150 +Epoch: [22] [3860/7269] eta: 0:24:28 lr: 0.000005 loss: 0.5024 (0.5197) time: 0.4340 data: 0.0002 max mem: 52150 +Epoch: [22] [3870/7269] eta: 0:24:24 lr: 0.000005 loss: 0.4935 (0.5197) time: 0.4174 data: 0.0002 max mem: 52150 +Epoch: [22] [3880/7269] eta: 0:24:19 lr: 0.000005 loss: 0.4976 (0.5196) time: 0.4176 data: 0.0002 max mem: 52150 +Epoch: [22] [3890/7269] eta: 0:24:15 lr: 0.000005 loss: 0.5182 (0.5196) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [22] [3900/7269] eta: 0:24:11 lr: 0.000005 loss: 0.5203 (0.5196) time: 0.4328 data: 0.0001 max mem: 52150 +Epoch: [22] [3910/7269] eta: 0:24:06 lr: 0.000005 loss: 0.5182 (0.5195) time: 0.4326 data: 0.0001 max mem: 52150 +Epoch: [22] [3920/7269] eta: 0:24:03 lr: 0.000005 loss: 0.5331 (0.5195) time: 0.4638 data: 0.0001 max mem: 52150 +Epoch: [22] [3930/7269] eta: 0:24:00 lr: 0.000005 loss: 0.5371 (0.5195) time: 0.5841 data: 0.0002 max mem: 52150 +Epoch: [22] [3940/7269] eta: 0:23:56 lr: 0.000005 loss: 0.5254 (0.5195) time: 0.5361 data: 0.0002 max mem: 52150 +Epoch: [22] [3950/7269] eta: 0:23:51 lr: 0.000005 loss: 0.5401 (0.5195) time: 0.4155 data: 0.0002 max mem: 52150 +Epoch: [22] [3960/7269] eta: 0:23:48 lr: 0.000005 loss: 0.5339 (0.5195) time: 0.4817 data: 0.0001 max mem: 52150 +Epoch: [22] [3970/7269] eta: 0:23:45 lr: 0.000005 loss: 0.5339 (0.5195) time: 0.5742 data: 0.0001 max mem: 52150 +Epoch: [22] [3980/7269] eta: 0:23:40 lr: 0.000005 loss: 0.5371 (0.5195) time: 0.4640 data: 0.0002 max mem: 52150 +Epoch: [22] [3990/7269] eta: 0:23:35 lr: 0.000005 loss: 0.5371 (0.5196) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [22] [4000/7269] eta: 0:23:31 lr: 0.000005 loss: 0.5474 (0.5196) time: 0.4175 data: 0.0001 max mem: 52150 +Epoch: [22] [4010/7269] eta: 0:23:26 lr: 0.000005 loss: 0.5358 (0.5196) time: 0.4175 data: 0.0002 max mem: 52150 +Epoch: [22] [4020/7269] eta: 0:23:23 lr: 0.000005 loss: 0.5429 (0.5197) time: 0.4803 data: 0.0002 max mem: 52150 +Epoch: [22] [4030/7269] eta: 0:23:18 lr: 0.000005 loss: 0.5492 (0.5198) time: 0.4802 data: 0.0002 max mem: 52150 +Epoch: [22] [4040/7269] eta: 0:23:14 lr: 0.000005 loss: 0.5169 (0.5198) time: 0.3728 data: 0.0002 max mem: 52150 +Epoch: [22] [4050/7269] eta: 0:23:09 lr: 0.000005 loss: 0.5169 (0.5197) time: 0.3734 data: 0.0002 max mem: 52150 +Epoch: [22] [4060/7269] eta: 0:23:04 lr: 0.000005 loss: 0.5423 (0.5198) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [22] [4070/7269] eta: 0:22:59 lr: 0.000005 loss: 0.5370 (0.5197) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [22] [4080/7269] eta: 0:22:56 lr: 0.000005 loss: 0.5184 (0.5198) time: 0.4321 data: 0.0002 max mem: 52150 +Epoch: [22] [4090/7269] eta: 0:22:51 lr: 0.000005 loss: 0.5092 (0.5196) time: 0.4322 data: 0.0001 max mem: 52150 +Epoch: [22] [4100/7269] eta: 0:22:46 lr: 0.000005 loss: 0.5268 (0.5197) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [22] [4110/7269] eta: 0:22:42 lr: 0.000005 loss: 0.5285 (0.5196) time: 0.4174 data: 0.0002 max mem: 52150 +Epoch: [22] [4120/7269] eta: 0:22:37 lr: 0.000005 loss: 0.5285 (0.5197) time: 0.4172 data: 0.0001 max mem: 52150 +Epoch: [22] [4130/7269] eta: 0:22:32 lr: 0.000005 loss: 0.5130 (0.5196) time: 0.3713 data: 0.0001 max mem: 52150 +Epoch: [22] [4140/7269] eta: 0:22:28 lr: 0.000005 loss: 0.5021 (0.5196) time: 0.3712 data: 0.0001 max mem: 52150 +Epoch: [22] [4150/7269] eta: 0:22:23 lr: 0.000005 loss: 0.5312 (0.5197) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [22] [4160/7269] eta: 0:22:18 lr: 0.000005 loss: 0.5224 (0.5196) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [22] [4170/7269] eta: 0:22:13 lr: 0.000005 loss: 0.5435 (0.5196) time: 0.3711 data: 0.0001 max mem: 52150 +Epoch: [22] [4180/7269] eta: 0:22:09 lr: 0.000005 loss: 0.5502 (0.5196) time: 0.4171 data: 0.0001 max mem: 52150 +Epoch: [22] [4190/7269] eta: 0:22:05 lr: 0.000005 loss: 0.5333 (0.5197) time: 0.4637 data: 0.0001 max mem: 52150 +Epoch: [22] [4200/7269] eta: 0:22:00 lr: 0.000005 loss: 0.5305 (0.5197) time: 0.4182 data: 0.0002 max mem: 52150 +Epoch: [22] [4210/7269] eta: 0:21:56 lr: 0.000005 loss: 0.5304 (0.5198) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [22] [4220/7269] eta: 0:21:51 lr: 0.000005 loss: 0.5304 (0.5198) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [22] [4230/7269] eta: 0:21:47 lr: 0.000005 loss: 0.5188 (0.5197) time: 0.4318 data: 0.0001 max mem: 52150 +Epoch: [22] [4240/7269] eta: 0:21:42 lr: 0.000005 loss: 0.5346 (0.5198) time: 0.4320 data: 0.0001 max mem: 52150 +Epoch: [22] [4250/7269] eta: 0:21:38 lr: 0.000005 loss: 0.5520 (0.5198) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [22] [4260/7269] eta: 0:21:33 lr: 0.000005 loss: 0.5239 (0.5198) time: 0.3712 data: 0.0002 max mem: 52150 +Epoch: [22] [4270/7269] eta: 0:21:28 lr: 0.000005 loss: 0.4967 (0.5197) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [22] [4280/7269] eta: 0:21:24 lr: 0.000005 loss: 0.4992 (0.5197) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [22] [4290/7269] eta: 0:21:20 lr: 0.000005 loss: 0.5507 (0.5198) time: 0.4156 data: 0.0001 max mem: 52150 +Epoch: [22] [4300/7269] eta: 0:21:17 lr: 0.000005 loss: 0.5109 (0.5197) time: 0.5451 data: 0.0001 max mem: 52150 +Epoch: [22] [4310/7269] eta: 0:21:12 lr: 0.000005 loss: 0.5064 (0.5197) time: 0.5011 data: 0.0002 max mem: 52150 +Epoch: [22] [4320/7269] eta: 0:21:08 lr: 0.000005 loss: 0.5414 (0.5198) time: 0.4353 data: 0.0002 max mem: 52150 +Epoch: [22] [4330/7269] eta: 0:21:03 lr: 0.000005 loss: 0.5264 (0.5197) time: 0.4348 data: 0.0002 max mem: 52150 +Epoch: [22] [4340/7269] eta: 0:20:59 lr: 0.000005 loss: 0.5133 (0.5197) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [22] [4350/7269] eta: 0:20:54 lr: 0.000005 loss: 0.5380 (0.5197) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [22] [4360/7269] eta: 0:20:51 lr: 0.000005 loss: 0.4992 (0.5197) time: 0.4634 data: 0.0002 max mem: 52150 +Epoch: [22] [4370/7269] eta: 0:20:47 lr: 0.000005 loss: 0.5099 (0.5197) time: 0.5229 data: 0.0002 max mem: 52150 +Epoch: [22] [4380/7269] eta: 0:20:43 lr: 0.000005 loss: 0.5099 (0.5196) time: 0.4779 data: 0.0001 max mem: 52150 +Epoch: [22] [4390/7269] eta: 0:20:39 lr: 0.000005 loss: 0.4944 (0.5196) time: 0.4810 data: 0.0002 max mem: 52150 +Epoch: [22] [4400/7269] eta: 0:20:34 lr: 0.000005 loss: 0.5142 (0.5196) time: 0.4339 data: 0.0002 max mem: 52150 +Epoch: [22] [4410/7269] eta: 0:20:30 lr: 0.000005 loss: 0.5176 (0.5196) time: 0.4165 data: 0.0001 max mem: 52150 +Epoch: [22] [4420/7269] eta: 0:20:27 lr: 0.000005 loss: 0.5006 (0.5196) time: 0.5256 data: 0.0001 max mem: 52150 +Epoch: [22] [4430/7269] eta: 0:20:23 lr: 0.000005 loss: 0.5305 (0.5197) time: 0.5419 data: 0.0001 max mem: 52150 +Epoch: [22] [4440/7269] eta: 0:20:18 lr: 0.000005 loss: 0.5319 (0.5196) time: 0.4333 data: 0.0001 max mem: 52150 +Epoch: [22] [4450/7269] eta: 0:20:13 lr: 0.000005 loss: 0.4971 (0.5196) time: 0.3714 data: 0.0001 max mem: 52150 +Epoch: [22] [4460/7269] eta: 0:20:09 lr: 0.000005 loss: 0.5096 (0.5196) time: 0.4310 data: 0.0001 max mem: 52150 +Epoch: [22] [4470/7269] eta: 0:20:05 lr: 0.000005 loss: 0.5363 (0.5196) time: 0.4317 data: 0.0001 max mem: 52150 +Epoch: [22] [4480/7269] eta: 0:20:01 lr: 0.000005 loss: 0.5373 (0.5197) time: 0.4342 data: 0.0001 max mem: 52150 +Epoch: [22] [4490/7269] eta: 0:19:58 lr: 0.000005 loss: 0.5380 (0.5197) time: 0.5425 data: 0.0001 max mem: 52150 +Epoch: [22] [4500/7269] eta: 0:19:53 lr: 0.000005 loss: 0.5264 (0.5197) time: 0.4802 data: 0.0001 max mem: 52150 +Epoch: [22] [4510/7269] eta: 0:19:49 lr: 0.000005 loss: 0.5072 (0.5197) time: 0.4343 data: 0.0002 max mem: 52150 +Epoch: [22] [4520/7269] eta: 0:19:44 lr: 0.000005 loss: 0.5072 (0.5197) time: 0.4342 data: 0.0002 max mem: 52150 +Epoch: [22] [4530/7269] eta: 0:19:40 lr: 0.000005 loss: 0.4960 (0.5197) time: 0.3712 data: 0.0001 max mem: 52150 +Epoch: [22] [4540/7269] eta: 0:19:36 lr: 0.000005 loss: 0.5380 (0.5197) time: 0.4928 data: 0.0001 max mem: 52150 +Epoch: [22] [4550/7269] eta: 0:19:32 lr: 0.000005 loss: 0.5252 (0.5197) time: 0.5539 data: 0.0002 max mem: 52150 +Epoch: [22] [4560/7269] eta: 0:19:28 lr: 0.000005 loss: 0.5287 (0.5198) time: 0.4802 data: 0.0002 max mem: 52150 +Epoch: [22] [4570/7269] eta: 0:19:24 lr: 0.000005 loss: 0.5322 (0.5198) time: 0.4189 data: 0.0001 max mem: 52150 +Epoch: [22] [4580/7269] eta: 0:19:20 lr: 0.000005 loss: 0.5322 (0.5198) time: 0.4178 data: 0.0001 max mem: 52150 +Epoch: [22] [4590/7269] eta: 0:19:15 lr: 0.000005 loss: 0.5513 (0.5198) time: 0.4178 data: 0.0001 max mem: 52150 +Epoch: [22] [4600/7269] eta: 0:19:10 lr: 0.000005 loss: 0.5371 (0.5199) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [22] [4610/7269] eta: 0:19:06 lr: 0.000005 loss: 0.5439 (0.5199) time: 0.4311 data: 0.0002 max mem: 52150 +Epoch: [22] [4620/7269] eta: 0:19:02 lr: 0.000005 loss: 0.5232 (0.5198) time: 0.4759 data: 0.0002 max mem: 52150 +Epoch: [22] [4630/7269] eta: 0:18:57 lr: 0.000005 loss: 0.5269 (0.5199) time: 0.4170 data: 0.0002 max mem: 52150 +Epoch: [22] [4640/7269] eta: 0:18:53 lr: 0.000005 loss: 0.5787 (0.5199) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [22] [4650/7269] eta: 0:18:49 lr: 0.000005 loss: 0.5659 (0.5200) time: 0.4631 data: 0.0002 max mem: 52150 +Epoch: [22] [4660/7269] eta: 0:18:45 lr: 0.000005 loss: 0.5253 (0.5199) time: 0.4630 data: 0.0001 max mem: 52150 +Epoch: [22] [4670/7269] eta: 0:18:40 lr: 0.000005 loss: 0.5177 (0.5199) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [22] [4680/7269] eta: 0:18:36 lr: 0.000005 loss: 0.5451 (0.5200) time: 0.4197 data: 0.0001 max mem: 52150 +Epoch: [22] [4690/7269] eta: 0:18:31 lr: 0.000005 loss: 0.5437 (0.5199) time: 0.4194 data: 0.0001 max mem: 52150 +Epoch: [22] [4700/7269] eta: 0:18:27 lr: 0.000005 loss: 0.5240 (0.5200) time: 0.3709 data: 0.0001 max mem: 52150 +Epoch: [22] [4710/7269] eta: 0:18:22 lr: 0.000005 loss: 0.5321 (0.5200) time: 0.3714 data: 0.0001 max mem: 52150 +Epoch: [22] [4720/7269] eta: 0:18:17 lr: 0.000005 loss: 0.5560 (0.5200) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [22] [4730/7269] eta: 0:18:13 lr: 0.000005 loss: 0.5366 (0.5200) time: 0.4174 data: 0.0001 max mem: 52150 +Epoch: [22] [4740/7269] eta: 0:18:09 lr: 0.000005 loss: 0.5353 (0.5201) time: 0.4171 data: 0.0001 max mem: 52150 +Epoch: [22] [4750/7269] eta: 0:18:04 lr: 0.000005 loss: 0.5389 (0.5202) time: 0.3709 data: 0.0002 max mem: 52150 +Epoch: [22] [4760/7269] eta: 0:18:00 lr: 0.000005 loss: 0.5389 (0.5201) time: 0.4165 data: 0.0002 max mem: 52150 +Epoch: [22] [4770/7269] eta: 0:17:55 lr: 0.000005 loss: 0.5010 (0.5201) time: 0.4165 data: 0.0001 max mem: 52150 +Epoch: [22] [4780/7269] eta: 0:17:51 lr: 0.000005 loss: 0.4857 (0.5200) time: 0.3709 data: 0.0002 max mem: 52150 +Epoch: [22] [4790/7269] eta: 0:17:47 lr: 0.000005 loss: 0.4857 (0.5200) time: 0.4355 data: 0.0001 max mem: 52150 +Epoch: [22] [4800/7269] eta: 0:17:43 lr: 0.000005 loss: 0.5258 (0.5200) time: 0.4984 data: 0.0001 max mem: 52150 +Epoch: [22] [4810/7269] eta: 0:17:38 lr: 0.000005 loss: 0.5258 (0.5200) time: 0.4337 data: 0.0001 max mem: 52150 +Epoch: [22] [4820/7269] eta: 0:17:35 lr: 0.000005 loss: 0.5619 (0.5201) time: 0.4822 data: 0.0001 max mem: 52150 +Epoch: [22] [4830/7269] eta: 0:17:30 lr: 0.000005 loss: 0.5168 (0.5200) time: 0.4829 data: 0.0001 max mem: 52150 +Epoch: [22] [4840/7269] eta: 0:17:25 lr: 0.000005 loss: 0.4670 (0.5200) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [22] [4850/7269] eta: 0:17:22 lr: 0.000005 loss: 0.5083 (0.5200) time: 0.4630 data: 0.0001 max mem: 52150 +Epoch: [22] [4860/7269] eta: 0:17:18 lr: 0.000005 loss: 0.5234 (0.5200) time: 0.5258 data: 0.0001 max mem: 52150 +Epoch: [22] [4870/7269] eta: 0:17:14 lr: 0.000005 loss: 0.5272 (0.5200) time: 0.4970 data: 0.0002 max mem: 52150 +Epoch: [22] [4880/7269] eta: 0:17:09 lr: 0.000005 loss: 0.5290 (0.5200) time: 0.4345 data: 0.0002 max mem: 52150 +Epoch: [22] [4890/7269] eta: 0:17:04 lr: 0.000005 loss: 0.4878 (0.5199) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [22] [4900/7269] eta: 0:17:00 lr: 0.000005 loss: 0.5001 (0.5199) time: 0.4329 data: 0.0002 max mem: 52150 +Epoch: [22] [4910/7269] eta: 0:16:56 lr: 0.000005 loss: 0.5499 (0.5200) time: 0.4793 data: 0.0002 max mem: 52150 +Epoch: [22] [4920/7269] eta: 0:16:52 lr: 0.000005 loss: 0.5169 (0.5199) time: 0.4179 data: 0.0001 max mem: 52150 +Epoch: [22] [4930/7269] eta: 0:16:48 lr: 0.000005 loss: 0.5082 (0.5200) time: 0.4328 data: 0.0002 max mem: 52150 +Epoch: [22] [4940/7269] eta: 0:16:43 lr: 0.000005 loss: 0.5145 (0.5199) time: 0.4329 data: 0.0002 max mem: 52150 +Epoch: [22] [4950/7269] eta: 0:16:40 lr: 0.000005 loss: 0.4771 (0.5199) time: 0.5397 data: 0.0001 max mem: 52150 +Epoch: [22] [4960/7269] eta: 0:16:36 lr: 0.000005 loss: 0.5298 (0.5199) time: 0.5396 data: 0.0001 max mem: 52150 +Epoch: [22] [4970/7269] eta: 0:16:31 lr: 0.000005 loss: 0.5298 (0.5198) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [22] [4980/7269] eta: 0:16:26 lr: 0.000005 loss: 0.5506 (0.5199) time: 0.3725 data: 0.0001 max mem: 52150 +Epoch: [22] [4990/7269] eta: 0:16:22 lr: 0.000005 loss: 0.5544 (0.5199) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [22] [5000/7269] eta: 0:16:18 lr: 0.000005 loss: 0.5195 (0.5199) time: 0.4328 data: 0.0002 max mem: 52150 +Epoch: [22] [5010/7269] eta: 0:16:13 lr: 0.000005 loss: 0.4976 (0.5198) time: 0.4335 data: 0.0002 max mem: 52150 +Epoch: [22] [5020/7269] eta: 0:16:09 lr: 0.000005 loss: 0.4810 (0.5198) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [22] [5030/7269] eta: 0:16:05 lr: 0.000005 loss: 0.4917 (0.5198) time: 0.4340 data: 0.0001 max mem: 52150 +Epoch: [22] [5040/7269] eta: 0:16:01 lr: 0.000005 loss: 0.5293 (0.5198) time: 0.4954 data: 0.0001 max mem: 52150 +Epoch: [22] [5050/7269] eta: 0:15:56 lr: 0.000005 loss: 0.5312 (0.5199) time: 0.4331 data: 0.0001 max mem: 52150 +Epoch: [22] [5060/7269] eta: 0:15:51 lr: 0.000005 loss: 0.5342 (0.5198) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [22] [5070/7269] eta: 0:15:48 lr: 0.000005 loss: 0.5323 (0.5199) time: 0.4798 data: 0.0001 max mem: 52150 +Epoch: [22] [5080/7269] eta: 0:15:44 lr: 0.000005 loss: 0.5526 (0.5199) time: 0.5403 data: 0.0001 max mem: 52150 +Epoch: [22] [5090/7269] eta: 0:15:40 lr: 0.000005 loss: 0.5447 (0.5199) time: 0.4968 data: 0.0001 max mem: 52150 +Epoch: [22] [5100/7269] eta: 0:15:35 lr: 0.000005 loss: 0.5286 (0.5199) time: 0.4368 data: 0.0001 max mem: 52150 +Epoch: [22] [5110/7269] eta: 0:15:31 lr: 0.000005 loss: 0.5448 (0.5200) time: 0.3722 data: 0.0001 max mem: 52150 +Epoch: [22] [5120/7269] eta: 0:15:27 lr: 0.000005 loss: 0.5358 (0.5199) time: 0.4824 data: 0.0002 max mem: 52150 +Epoch: [22] [5130/7269] eta: 0:15:23 lr: 0.000005 loss: 0.5225 (0.5200) time: 0.5415 data: 0.0002 max mem: 52150 +Epoch: [22] [5140/7269] eta: 0:15:18 lr: 0.000005 loss: 0.5114 (0.5199) time: 0.4313 data: 0.0001 max mem: 52150 +Epoch: [22] [5150/7269] eta: 0:15:14 lr: 0.000005 loss: 0.5023 (0.5199) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [22] [5160/7269] eta: 0:15:09 lr: 0.000005 loss: 0.5307 (0.5199) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [22] [5170/7269] eta: 0:15:05 lr: 0.000005 loss: 0.5196 (0.5199) time: 0.3725 data: 0.0001 max mem: 52150 +Epoch: [22] [5180/7269] eta: 0:15:00 lr: 0.000005 loss: 0.5501 (0.5200) time: 0.4186 data: 0.0001 max mem: 52150 +Epoch: [22] [5190/7269] eta: 0:14:56 lr: 0.000005 loss: 0.5518 (0.5200) time: 0.4182 data: 0.0001 max mem: 52150 +Epoch: [22] [5200/7269] eta: 0:14:52 lr: 0.000005 loss: 0.4932 (0.5199) time: 0.4159 data: 0.0002 max mem: 52150 +Epoch: [22] [5210/7269] eta: 0:14:47 lr: 0.000005 loss: 0.4844 (0.5199) time: 0.4160 data: 0.0002 max mem: 52150 +Epoch: [22] [5220/7269] eta: 0:14:43 lr: 0.000005 loss: 0.5042 (0.5199) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [22] [5230/7269] eta: 0:14:38 lr: 0.000005 loss: 0.5180 (0.5199) time: 0.4202 data: 0.0001 max mem: 52150 +Epoch: [22] [5240/7269] eta: 0:14:34 lr: 0.000005 loss: 0.5002 (0.5198) time: 0.4205 data: 0.0001 max mem: 52150 +Epoch: [22] [5250/7269] eta: 0:14:29 lr: 0.000005 loss: 0.5002 (0.5198) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [22] [5260/7269] eta: 0:14:25 lr: 0.000005 loss: 0.5262 (0.5198) time: 0.4159 data: 0.0002 max mem: 52150 +Epoch: [22] [5270/7269] eta: 0:14:21 lr: 0.000005 loss: 0.5274 (0.5198) time: 0.4804 data: 0.0001 max mem: 52150 +Epoch: [22] [5280/7269] eta: 0:14:17 lr: 0.000005 loss: 0.4912 (0.5198) time: 0.4361 data: 0.0001 max mem: 52150 +Epoch: [22] [5290/7269] eta: 0:14:13 lr: 0.000005 loss: 0.5081 (0.5198) time: 0.4343 data: 0.0001 max mem: 52150 +Epoch: [22] [5300/7269] eta: 0:14:08 lr: 0.000005 loss: 0.4994 (0.5197) time: 0.4340 data: 0.0002 max mem: 52150 +Epoch: [22] [5310/7269] eta: 0:14:03 lr: 0.000005 loss: 0.5171 (0.5198) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [22] [5320/7269] eta: 0:13:59 lr: 0.000005 loss: 0.5313 (0.5198) time: 0.4181 data: 0.0002 max mem: 52150 +Epoch: [22] [5330/7269] eta: 0:13:55 lr: 0.000005 loss: 0.4993 (0.5198) time: 0.4184 data: 0.0002 max mem: 52150 +Epoch: [22] [5340/7269] eta: 0:13:51 lr: 0.000005 loss: 0.5044 (0.5198) time: 0.4160 data: 0.0002 max mem: 52150 +Epoch: [22] [5350/7269] eta: 0:13:47 lr: 0.000005 loss: 0.5046 (0.5197) time: 0.4804 data: 0.0002 max mem: 52150 +Epoch: [22] [5360/7269] eta: 0:13:42 lr: 0.000005 loss: 0.5398 (0.5198) time: 0.4991 data: 0.0002 max mem: 52150 +Epoch: [22] [5370/7269] eta: 0:13:38 lr: 0.000005 loss: 0.5348 (0.5198) time: 0.4351 data: 0.0003 max mem: 52150 +Epoch: [22] [5380/7269] eta: 0:13:34 lr: 0.000005 loss: 0.5150 (0.5198) time: 0.4165 data: 0.0003 max mem: 52150 +Epoch: [22] [5390/7269] eta: 0:13:30 lr: 0.000005 loss: 0.5486 (0.5199) time: 0.4634 data: 0.0002 max mem: 52150 +Epoch: [22] [5400/7269] eta: 0:13:25 lr: 0.000005 loss: 0.5302 (0.5198) time: 0.4195 data: 0.0001 max mem: 52150 +Epoch: [22] [5410/7269] eta: 0:13:21 lr: 0.000005 loss: 0.5025 (0.5198) time: 0.4175 data: 0.0001 max mem: 52150 +Epoch: [22] [5420/7269] eta: 0:13:17 lr: 0.000005 loss: 0.5249 (0.5198) time: 0.4793 data: 0.0001 max mem: 52150 +Epoch: [22] [5430/7269] eta: 0:13:12 lr: 0.000005 loss: 0.5363 (0.5198) time: 0.4339 data: 0.0001 max mem: 52150 +Epoch: [22] [5440/7269] eta: 0:13:08 lr: 0.000005 loss: 0.5627 (0.5199) time: 0.4213 data: 0.0001 max mem: 52150 +Epoch: [22] [5450/7269] eta: 0:13:04 lr: 0.000005 loss: 0.5512 (0.5200) time: 0.4217 data: 0.0001 max mem: 52150 +Epoch: [22] [5460/7269] eta: 0:12:59 lr: 0.000005 loss: 0.5264 (0.5199) time: 0.3723 data: 0.0001 max mem: 52150 +Epoch: [22] [5470/7269] eta: 0:12:56 lr: 0.000005 loss: 0.5067 (0.5199) time: 0.5272 data: 0.0001 max mem: 52150 +Epoch: [22] [5480/7269] eta: 0:12:51 lr: 0.000005 loss: 0.5170 (0.5199) time: 0.5269 data: 0.0001 max mem: 52150 +Epoch: [22] [5490/7269] eta: 0:12:47 lr: 0.000005 loss: 0.5170 (0.5199) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [22] [5500/7269] eta: 0:12:42 lr: 0.000005 loss: 0.5591 (0.5200) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [22] [5510/7269] eta: 0:12:38 lr: 0.000005 loss: 0.5591 (0.5200) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [22] [5520/7269] eta: 0:12:33 lr: 0.000005 loss: 0.5549 (0.5201) time: 0.4322 data: 0.0001 max mem: 52150 +Epoch: [22] [5530/7269] eta: 0:12:29 lr: 0.000005 loss: 0.5320 (0.5200) time: 0.4328 data: 0.0001 max mem: 52150 +Epoch: [22] [5540/7269] eta: 0:12:25 lr: 0.000005 loss: 0.5343 (0.5201) time: 0.5387 data: 0.0001 max mem: 52150 +Epoch: [22] [5550/7269] eta: 0:12:21 lr: 0.000005 loss: 0.5343 (0.5201) time: 0.5839 data: 0.0002 max mem: 52150 +Epoch: [22] [5560/7269] eta: 0:12:17 lr: 0.000005 loss: 0.5238 (0.5201) time: 0.4177 data: 0.0002 max mem: 52150 +Epoch: [22] [5570/7269] eta: 0:12:13 lr: 0.000005 loss: 0.5317 (0.5200) time: 0.4344 data: 0.0001 max mem: 52150 +Epoch: [22] [5580/7269] eta: 0:12:09 lr: 0.000005 loss: 0.5208 (0.5200) time: 0.4983 data: 0.0001 max mem: 52150 +Epoch: [22] [5590/7269] eta: 0:12:04 lr: 0.000005 loss: 0.5088 (0.5200) time: 0.5008 data: 0.0002 max mem: 52150 +Epoch: [22] [5600/7269] eta: 0:12:00 lr: 0.000005 loss: 0.5412 (0.5200) time: 0.4361 data: 0.0002 max mem: 52150 +Epoch: [22] [5610/7269] eta: 0:11:56 lr: 0.000005 loss: 0.5416 (0.5200) time: 0.4318 data: 0.0002 max mem: 52150 +Epoch: [22] [5620/7269] eta: 0:11:51 lr: 0.000005 loss: 0.5267 (0.5200) time: 0.4326 data: 0.0001 max mem: 52150 +Epoch: [22] [5630/7269] eta: 0:11:47 lr: 0.000005 loss: 0.5295 (0.5200) time: 0.4338 data: 0.0001 max mem: 52150 +Epoch: [22] [5640/7269] eta: 0:11:43 lr: 0.000005 loss: 0.5465 (0.5201) time: 0.4330 data: 0.0002 max mem: 52150 +Epoch: [22] [5650/7269] eta: 0:11:38 lr: 0.000005 loss: 0.5465 (0.5201) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [22] [5660/7269] eta: 0:11:34 lr: 0.000005 loss: 0.5159 (0.5200) time: 0.4167 data: 0.0001 max mem: 52150 +Epoch: [22] [5670/7269] eta: 0:11:29 lr: 0.000005 loss: 0.5482 (0.5201) time: 0.4165 data: 0.0002 max mem: 52150 +Epoch: [22] [5680/7269] eta: 0:11:25 lr: 0.000005 loss: 0.5408 (0.5201) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [22] [5690/7269] eta: 0:11:21 lr: 0.000005 loss: 0.5089 (0.5200) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [22] [5700/7269] eta: 0:11:16 lr: 0.000005 loss: 0.5089 (0.5200) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [22] [5710/7269] eta: 0:11:12 lr: 0.000005 loss: 0.5271 (0.5200) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [22] [5720/7269] eta: 0:11:07 lr: 0.000005 loss: 0.5504 (0.5200) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [22] [5730/7269] eta: 0:11:03 lr: 0.000005 loss: 0.5358 (0.5200) time: 0.4321 data: 0.0002 max mem: 52150 +Epoch: [22] [5740/7269] eta: 0:10:59 lr: 0.000005 loss: 0.5358 (0.5199) time: 0.4959 data: 0.0002 max mem: 52150 +Epoch: [22] [5750/7269] eta: 0:10:55 lr: 0.000005 loss: 0.5332 (0.5199) time: 0.4843 data: 0.0001 max mem: 52150 +Epoch: [22] [5760/7269] eta: 0:10:50 lr: 0.000005 loss: 0.5302 (0.5199) time: 0.4207 data: 0.0001 max mem: 52150 +Epoch: [22] [5770/7269] eta: 0:10:46 lr: 0.000005 loss: 0.5312 (0.5199) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [22] [5780/7269] eta: 0:10:41 lr: 0.000005 loss: 0.5268 (0.5199) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [22] [5790/7269] eta: 0:10:37 lr: 0.000005 loss: 0.5175 (0.5199) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [22] [5800/7269] eta: 0:10:33 lr: 0.000005 loss: 0.5271 (0.5199) time: 0.4202 data: 0.0002 max mem: 52150 +Epoch: [22] [5810/7269] eta: 0:10:28 lr: 0.000005 loss: 0.5266 (0.5199) time: 0.4201 data: 0.0002 max mem: 52150 +Epoch: [22] [5820/7269] eta: 0:10:24 lr: 0.000005 loss: 0.5185 (0.5199) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [22] [5830/7269] eta: 0:10:19 lr: 0.000005 loss: 0.5450 (0.5199) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [22] [5840/7269] eta: 0:10:15 lr: 0.000005 loss: 0.5450 (0.5199) time: 0.4335 data: 0.0001 max mem: 52150 +Epoch: [22] [5850/7269] eta: 0:10:11 lr: 0.000005 loss: 0.5474 (0.5200) time: 0.4785 data: 0.0001 max mem: 52150 +Epoch: [22] [5860/7269] eta: 0:10:06 lr: 0.000005 loss: 0.5266 (0.5200) time: 0.4172 data: 0.0001 max mem: 52150 +Epoch: [22] [5870/7269] eta: 0:10:02 lr: 0.000005 loss: 0.5062 (0.5199) time: 0.4353 data: 0.0002 max mem: 52150 +Epoch: [22] [5880/7269] eta: 0:09:58 lr: 0.000005 loss: 0.5040 (0.5199) time: 0.4346 data: 0.0002 max mem: 52150 +Epoch: [22] [5890/7269] eta: 0:09:54 lr: 0.000005 loss: 0.5287 (0.5199) time: 0.4801 data: 0.0001 max mem: 52150 +Epoch: [22] [5900/7269] eta: 0:09:50 lr: 0.000005 loss: 0.5391 (0.5199) time: 0.5265 data: 0.0001 max mem: 52150 +Epoch: [22] [5910/7269] eta: 0:09:45 lr: 0.000005 loss: 0.5391 (0.5200) time: 0.4821 data: 0.0001 max mem: 52150 +Epoch: [22] [5920/7269] eta: 0:09:41 lr: 0.000005 loss: 0.5262 (0.5199) time: 0.4363 data: 0.0002 max mem: 52150 +Epoch: [22] [5930/7269] eta: 0:09:37 lr: 0.000005 loss: 0.5152 (0.5199) time: 0.4179 data: 0.0002 max mem: 52150 +Epoch: [22] [5940/7269] eta: 0:09:32 lr: 0.000005 loss: 0.5283 (0.5199) time: 0.4175 data: 0.0002 max mem: 52150 +Epoch: [22] [5950/7269] eta: 0:09:28 lr: 0.000005 loss: 0.5301 (0.5199) time: 0.3721 data: 0.0001 max mem: 52150 +Epoch: [22] [5960/7269] eta: 0:09:23 lr: 0.000005 loss: 0.5405 (0.5199) time: 0.3725 data: 0.0001 max mem: 52150 +Epoch: [22] [5970/7269] eta: 0:09:19 lr: 0.000005 loss: 0.5280 (0.5199) time: 0.4657 data: 0.0002 max mem: 52150 +Epoch: [22] [5980/7269] eta: 0:09:15 lr: 0.000005 loss: 0.5226 (0.5199) time: 0.4652 data: 0.0002 max mem: 52150 +Epoch: [22] [5990/7269] eta: 0:09:11 lr: 0.000005 loss: 0.4882 (0.5198) time: 0.4364 data: 0.0002 max mem: 52150 +Epoch: [22] [6000/7269] eta: 0:09:06 lr: 0.000005 loss: 0.4909 (0.5198) time: 0.4366 data: 0.0001 max mem: 52150 +Epoch: [22] [6010/7269] eta: 0:09:02 lr: 0.000005 loss: 0.5128 (0.5198) time: 0.4793 data: 0.0001 max mem: 52150 +Epoch: [22] [6020/7269] eta: 0:08:58 lr: 0.000005 loss: 0.4917 (0.5198) time: 0.4796 data: 0.0002 max mem: 52150 +Epoch: [22] [6030/7269] eta: 0:08:54 lr: 0.000005 loss: 0.5191 (0.5198) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [22] [6040/7269] eta: 0:08:49 lr: 0.000005 loss: 0.5168 (0.5198) time: 0.4358 data: 0.0002 max mem: 52150 +Epoch: [22] [6050/7269] eta: 0:08:45 lr: 0.000005 loss: 0.4799 (0.5198) time: 0.4972 data: 0.0002 max mem: 52150 +Epoch: [22] [6060/7269] eta: 0:08:41 lr: 0.000005 loss: 0.5044 (0.5198) time: 0.4340 data: 0.0002 max mem: 52150 +Epoch: [22] [6070/7269] eta: 0:08:36 lr: 0.000005 loss: 0.5073 (0.5197) time: 0.3732 data: 0.0002 max mem: 52150 +Epoch: [22] [6080/7269] eta: 0:08:32 lr: 0.000005 loss: 0.5098 (0.5197) time: 0.4189 data: 0.0001 max mem: 52150 +Epoch: [22] [6090/7269] eta: 0:08:28 lr: 0.000005 loss: 0.5289 (0.5197) time: 0.4621 data: 0.0002 max mem: 52150 +Epoch: [22] [6100/7269] eta: 0:08:23 lr: 0.000005 loss: 0.5422 (0.5197) time: 0.4160 data: 0.0001 max mem: 52150 +Epoch: [22] [6110/7269] eta: 0:08:19 lr: 0.000005 loss: 0.5376 (0.5198) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [22] [6120/7269] eta: 0:08:15 lr: 0.000005 loss: 0.5341 (0.5198) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [22] [6130/7269] eta: 0:08:10 lr: 0.000005 loss: 0.5341 (0.5197) time: 0.3723 data: 0.0001 max mem: 52150 +Epoch: [22] [6140/7269] eta: 0:08:06 lr: 0.000005 loss: 0.5122 (0.5197) time: 0.4317 data: 0.0001 max mem: 52150 +Epoch: [22] [6150/7269] eta: 0:08:02 lr: 0.000005 loss: 0.5105 (0.5197) time: 0.4313 data: 0.0001 max mem: 52150 +Epoch: [22] [6160/7269] eta: 0:07:57 lr: 0.000005 loss: 0.4807 (0.5196) time: 0.4169 data: 0.0001 max mem: 52150 +Epoch: [22] [6170/7269] eta: 0:07:53 lr: 0.000005 loss: 0.5050 (0.5197) time: 0.5101 data: 0.0001 max mem: 52150 +Epoch: [22] [6180/7269] eta: 0:07:49 lr: 0.000005 loss: 0.5316 (0.5197) time: 0.4645 data: 0.0002 max mem: 52150 +Epoch: [22] [6190/7269] eta: 0:07:45 lr: 0.000005 loss: 0.5317 (0.5197) time: 0.4936 data: 0.0002 max mem: 52150 +Epoch: [22] [6200/7269] eta: 0:07:40 lr: 0.000005 loss: 0.4931 (0.5196) time: 0.4942 data: 0.0001 max mem: 52150 +Epoch: [22] [6210/7269] eta: 0:07:36 lr: 0.000005 loss: 0.4960 (0.5196) time: 0.3733 data: 0.0002 max mem: 52150 +Epoch: [22] [6220/7269] eta: 0:07:32 lr: 0.000005 loss: 0.5183 (0.5196) time: 0.3727 data: 0.0001 max mem: 52150 +Epoch: [22] [6230/7269] eta: 0:07:27 lr: 0.000005 loss: 0.5219 (0.5196) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [22] [6240/7269] eta: 0:07:23 lr: 0.000005 loss: 0.5273 (0.5196) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [22] [6250/7269] eta: 0:07:18 lr: 0.000005 loss: 0.5152 (0.5196) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [22] [6260/7269] eta: 0:07:14 lr: 0.000005 loss: 0.5017 (0.5196) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [22] [6270/7269] eta: 0:07:10 lr: 0.000005 loss: 0.5214 (0.5195) time: 0.4705 data: 0.0002 max mem: 52150 +Epoch: [22] [6280/7269] eta: 0:07:05 lr: 0.000005 loss: 0.5322 (0.5196) time: 0.4702 data: 0.0002 max mem: 52150 +Epoch: [22] [6290/7269] eta: 0:07:01 lr: 0.000005 loss: 0.5597 (0.5196) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [22] [6300/7269] eta: 0:06:57 lr: 0.000005 loss: 0.5040 (0.5195) time: 0.4328 data: 0.0002 max mem: 52150 +Epoch: [22] [6310/7269] eta: 0:06:52 lr: 0.000005 loss: 0.4690 (0.5195) time: 0.4328 data: 0.0002 max mem: 52150 +Epoch: [22] [6320/7269] eta: 0:06:48 lr: 0.000005 loss: 0.5134 (0.5195) time: 0.4342 data: 0.0002 max mem: 52150 +Epoch: [22] [6330/7269] eta: 0:06:44 lr: 0.000005 loss: 0.5192 (0.5195) time: 0.4813 data: 0.0002 max mem: 52150 +Epoch: [22] [6340/7269] eta: 0:06:40 lr: 0.000005 loss: 0.5149 (0.5195) time: 0.4187 data: 0.0001 max mem: 52150 +Epoch: [22] [6350/7269] eta: 0:06:35 lr: 0.000005 loss: 0.5318 (0.5195) time: 0.4198 data: 0.0001 max mem: 52150 +Epoch: [22] [6360/7269] eta: 0:06:31 lr: 0.000005 loss: 0.4533 (0.5194) time: 0.4835 data: 0.0001 max mem: 52150 +Epoch: [22] [6370/7269] eta: 0:06:27 lr: 0.000005 loss: 0.4533 (0.5194) time: 0.6106 data: 0.0001 max mem: 52150 +Epoch: [22] [6380/7269] eta: 0:06:23 lr: 0.000005 loss: 0.5372 (0.5195) time: 0.6705 data: 0.0002 max mem: 52150 +Epoch: [22] [6390/7269] eta: 0:06:19 lr: 0.000005 loss: 0.5280 (0.5195) time: 0.4953 data: 0.0002 max mem: 52150 +Epoch: [22] [6400/7269] eta: 0:06:14 lr: 0.000005 loss: 0.5042 (0.5194) time: 0.3722 data: 0.0001 max mem: 52150 +Epoch: [22] [6410/7269] eta: 0:06:10 lr: 0.000005 loss: 0.4909 (0.5193) time: 0.3725 data: 0.0001 max mem: 52150 +Epoch: [22] [6420/7269] eta: 0:06:06 lr: 0.000005 loss: 0.5405 (0.5194) time: 0.3728 data: 0.0002 max mem: 52150 +Epoch: [22] [6430/7269] eta: 0:06:01 lr: 0.000005 loss: 0.5629 (0.5195) time: 0.4178 data: 0.0002 max mem: 52150 +Epoch: [22] [6440/7269] eta: 0:05:57 lr: 0.000005 loss: 0.5590 (0.5195) time: 0.4795 data: 0.0002 max mem: 52150 +Epoch: [22] [6450/7269] eta: 0:05:53 lr: 0.000005 loss: 0.5509 (0.5195) time: 0.4815 data: 0.0002 max mem: 52150 +Epoch: [22] [6460/7269] eta: 0:05:48 lr: 0.000005 loss: 0.5478 (0.5195) time: 0.4188 data: 0.0001 max mem: 52150 +Epoch: [22] [6470/7269] eta: 0:05:44 lr: 0.000005 loss: 0.5108 (0.5194) time: 0.4169 data: 0.0002 max mem: 52150 +Epoch: [22] [6480/7269] eta: 0:05:40 lr: 0.000005 loss: 0.4955 (0.5194) time: 0.4168 data: 0.0002 max mem: 52150 +Epoch: [22] [6490/7269] eta: 0:05:35 lr: 0.000005 loss: 0.5205 (0.5194) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [22] [6500/7269] eta: 0:05:31 lr: 0.000005 loss: 0.5245 (0.5194) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [22] [6510/7269] eta: 0:05:27 lr: 0.000005 loss: 0.5392 (0.5194) time: 0.4187 data: 0.0001 max mem: 52150 +Epoch: [22] [6520/7269] eta: 0:05:22 lr: 0.000005 loss: 0.5217 (0.5194) time: 0.4186 data: 0.0001 max mem: 52150 +Epoch: [22] [6530/7269] eta: 0:05:18 lr: 0.000005 loss: 0.5348 (0.5195) time: 0.4815 data: 0.0001 max mem: 52150 +Epoch: [22] [6540/7269] eta: 0:05:14 lr: 0.000005 loss: 0.5403 (0.5194) time: 0.5441 data: 0.0001 max mem: 52150 +Epoch: [22] [6550/7269] eta: 0:05:10 lr: 0.000005 loss: 0.4954 (0.5194) time: 0.4803 data: 0.0001 max mem: 52150 +Epoch: [22] [6560/7269] eta: 0:05:05 lr: 0.000005 loss: 0.4998 (0.5194) time: 0.4807 data: 0.0001 max mem: 52150 +Epoch: [22] [6570/7269] eta: 0:05:01 lr: 0.000005 loss: 0.5163 (0.5194) time: 0.4345 data: 0.0001 max mem: 52150 +Epoch: [22] [6580/7269] eta: 0:04:57 lr: 0.000005 loss: 0.5163 (0.5194) time: 0.3714 data: 0.0001 max mem: 52150 +Epoch: [22] [6590/7269] eta: 0:04:52 lr: 0.000005 loss: 0.5043 (0.5194) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [22] [6600/7269] eta: 0:04:48 lr: 0.000005 loss: 0.5163 (0.5193) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [22] [6610/7269] eta: 0:04:44 lr: 0.000005 loss: 0.5162 (0.5193) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [22] [6620/7269] eta: 0:04:39 lr: 0.000005 loss: 0.4840 (0.5193) time: 0.3728 data: 0.0002 max mem: 52150 +Epoch: [22] [6630/7269] eta: 0:04:35 lr: 0.000005 loss: 0.5402 (0.5193) time: 0.4211 data: 0.0001 max mem: 52150 +Epoch: [22] [6640/7269] eta: 0:04:31 lr: 0.000005 loss: 0.5457 (0.5194) time: 0.4204 data: 0.0001 max mem: 52150 +Epoch: [22] [6650/7269] eta: 0:04:26 lr: 0.000005 loss: 0.5457 (0.5194) time: 0.4315 data: 0.0001 max mem: 52150 +Epoch: [22] [6660/7269] eta: 0:04:22 lr: 0.000005 loss: 0.5233 (0.5194) time: 0.4317 data: 0.0001 max mem: 52150 +Epoch: [22] [6670/7269] eta: 0:04:18 lr: 0.000005 loss: 0.4919 (0.5193) time: 0.4818 data: 0.0001 max mem: 52150 +Epoch: [22] [6680/7269] eta: 0:04:13 lr: 0.000005 loss: 0.4851 (0.5193) time: 0.4822 data: 0.0002 max mem: 52150 +Epoch: [22] [6690/7269] eta: 0:04:09 lr: 0.000005 loss: 0.5338 (0.5193) time: 0.3721 data: 0.0001 max mem: 52150 +Epoch: [22] [6700/7269] eta: 0:04:05 lr: 0.000005 loss: 0.5393 (0.5194) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [22] [6710/7269] eta: 0:04:00 lr: 0.000005 loss: 0.5462 (0.5194) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [22] [6720/7269] eta: 0:03:56 lr: 0.000005 loss: 0.5219 (0.5194) time: 0.4196 data: 0.0001 max mem: 52150 +Epoch: [22] [6730/7269] eta: 0:03:52 lr: 0.000005 loss: 0.5219 (0.5194) time: 0.4806 data: 0.0002 max mem: 52150 +Epoch: [22] [6740/7269] eta: 0:03:47 lr: 0.000005 loss: 0.5648 (0.5195) time: 0.4325 data: 0.0002 max mem: 52150 +Epoch: [22] [6750/7269] eta: 0:03:43 lr: 0.000005 loss: 0.5492 (0.5195) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [22] [6760/7269] eta: 0:03:39 lr: 0.000005 loss: 0.5179 (0.5194) time: 0.4178 data: 0.0001 max mem: 52150 +Epoch: [22] [6770/7269] eta: 0:03:35 lr: 0.000005 loss: 0.5175 (0.5194) time: 0.4782 data: 0.0001 max mem: 52150 +Epoch: [22] [6780/7269] eta: 0:03:30 lr: 0.000005 loss: 0.5019 (0.5194) time: 0.4322 data: 0.0002 max mem: 52150 +Epoch: [22] [6790/7269] eta: 0:03:26 lr: 0.000005 loss: 0.5019 (0.5194) time: 0.4632 data: 0.0002 max mem: 52150 +Epoch: [22] [6800/7269] eta: 0:03:22 lr: 0.000005 loss: 0.5287 (0.5194) time: 0.4631 data: 0.0001 max mem: 52150 +Epoch: [22] [6810/7269] eta: 0:03:17 lr: 0.000005 loss: 0.5138 (0.5194) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [22] [6820/7269] eta: 0:03:13 lr: 0.000005 loss: 0.5366 (0.5194) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [22] [6830/7269] eta: 0:03:09 lr: 0.000005 loss: 0.5241 (0.5194) time: 0.4189 data: 0.0001 max mem: 52150 +Epoch: [22] [6840/7269] eta: 0:03:04 lr: 0.000005 loss: 0.5241 (0.5194) time: 0.4823 data: 0.0001 max mem: 52150 +Epoch: [22] [6850/7269] eta: 0:03:00 lr: 0.000005 loss: 0.5300 (0.5194) time: 0.4354 data: 0.0001 max mem: 52150 +Epoch: [22] [6860/7269] eta: 0:02:56 lr: 0.000005 loss: 0.5383 (0.5194) time: 0.4337 data: 0.0001 max mem: 52150 +Epoch: [22] [6870/7269] eta: 0:02:51 lr: 0.000005 loss: 0.5377 (0.5194) time: 0.4335 data: 0.0001 max mem: 52150 +Epoch: [22] [6880/7269] eta: 0:02:47 lr: 0.000005 loss: 0.5336 (0.5194) time: 0.4218 data: 0.0002 max mem: 52150 +Epoch: [22] [6890/7269] eta: 0:02:43 lr: 0.000005 loss: 0.5417 (0.5194) time: 0.4224 data: 0.0002 max mem: 52150 +Epoch: [22] [6900/7269] eta: 0:02:38 lr: 0.000005 loss: 0.5344 (0.5194) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [22] [6910/7269] eta: 0:02:34 lr: 0.000005 loss: 0.5225 (0.5194) time: 0.4338 data: 0.0001 max mem: 52150 +Epoch: [22] [6920/7269] eta: 0:02:30 lr: 0.000005 loss: 0.5404 (0.5195) time: 0.4949 data: 0.0001 max mem: 52150 +Epoch: [22] [6930/7269] eta: 0:02:26 lr: 0.000005 loss: 0.5115 (0.5194) time: 0.4324 data: 0.0001 max mem: 52150 +Epoch: [22] [6940/7269] eta: 0:02:21 lr: 0.000005 loss: 0.5084 (0.5194) time: 0.4802 data: 0.0001 max mem: 52150 +Epoch: [22] [6950/7269] eta: 0:02:17 lr: 0.000005 loss: 0.5359 (0.5194) time: 0.4813 data: 0.0002 max mem: 52150 +Epoch: [22] [6960/7269] eta: 0:02:13 lr: 0.000005 loss: 0.5359 (0.5194) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [22] [6970/7269] eta: 0:02:08 lr: 0.000005 loss: 0.5353 (0.5194) time: 0.4174 data: 0.0002 max mem: 52150 +Epoch: [22] [6980/7269] eta: 0:02:04 lr: 0.000005 loss: 0.5130 (0.5193) time: 0.4619 data: 0.0002 max mem: 52150 +Epoch: [22] [6990/7269] eta: 0:02:00 lr: 0.000005 loss: 0.4601 (0.5192) time: 0.4160 data: 0.0002 max mem: 52150 +Epoch: [22] [7000/7269] eta: 0:01:56 lr: 0.000005 loss: 0.4573 (0.5192) time: 0.5617 data: 0.0002 max mem: 52150 +Epoch: [22] [7010/7269] eta: 0:01:51 lr: 0.000005 loss: 0.5157 (0.5192) time: 0.5616 data: 0.0002 max mem: 52150 +Epoch: [22] [7020/7269] eta: 0:01:47 lr: 0.000005 loss: 0.5213 (0.5192) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [22] [7030/7269] eta: 0:01:43 lr: 0.000005 loss: 0.5242 (0.5192) time: 0.3722 data: 0.0001 max mem: 52150 +Epoch: [22] [7040/7269] eta: 0:01:38 lr: 0.000005 loss: 0.5289 (0.5192) time: 0.4323 data: 0.0002 max mem: 52150 +Epoch: [22] [7050/7269] eta: 0:01:34 lr: 0.000005 loss: 0.5226 (0.5192) time: 0.4320 data: 0.0002 max mem: 52150 +Epoch: [22] [7060/7269] eta: 0:01:30 lr: 0.000005 loss: 0.5365 (0.5192) time: 0.4189 data: 0.0002 max mem: 52150 +Epoch: [22] [7070/7269] eta: 0:01:25 lr: 0.000005 loss: 0.5406 (0.5192) time: 0.4189 data: 0.0001 max mem: 52150 +Epoch: [22] [7080/7269] eta: 0:01:21 lr: 0.000005 loss: 0.5157 (0.5192) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [22] [7090/7269] eta: 0:01:17 lr: 0.000005 loss: 0.5091 (0.5192) time: 0.4980 data: 0.0001 max mem: 52150 +Epoch: [22] [7100/7269] eta: 0:01:12 lr: 0.000005 loss: 0.4930 (0.5191) time: 0.5447 data: 0.0001 max mem: 52150 +Epoch: [22] [7110/7269] eta: 0:01:08 lr: 0.000005 loss: 0.4876 (0.5191) time: 0.4844 data: 0.0002 max mem: 52150 +Epoch: [22] [7120/7269] eta: 0:01:04 lr: 0.000005 loss: 0.5040 (0.5191) time: 0.4876 data: 0.0001 max mem: 52150 +Epoch: [22] [7130/7269] eta: 0:00:59 lr: 0.000005 loss: 0.5189 (0.5190) time: 0.4227 data: 0.0001 max mem: 52150 +Epoch: [22] [7140/7269] eta: 0:00:55 lr: 0.000005 loss: 0.4898 (0.5190) time: 0.4365 data: 0.0002 max mem: 52150 +Epoch: [22] [7150/7269] eta: 0:00:51 lr: 0.000005 loss: 0.5398 (0.5191) time: 0.4359 data: 0.0002 max mem: 52150 +Epoch: [22] [7160/7269] eta: 0:00:47 lr: 0.000005 loss: 0.5538 (0.5191) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [22] [7170/7269] eta: 0:00:42 lr: 0.000005 loss: 0.5538 (0.5192) time: 0.3721 data: 0.0001 max mem: 52150 +Epoch: [22] [7180/7269] eta: 0:00:38 lr: 0.000005 loss: 0.5448 (0.5192) time: 0.4200 data: 0.0002 max mem: 52150 +Epoch: [22] [7190/7269] eta: 0:00:34 lr: 0.000005 loss: 0.5134 (0.5192) time: 0.4202 data: 0.0002 max mem: 52150 +Epoch: [22] [7200/7269] eta: 0:00:29 lr: 0.000005 loss: 0.4868 (0.5192) time: 0.4313 data: 0.0002 max mem: 52150 +Epoch: [22] [7210/7269] eta: 0:00:25 lr: 0.000005 loss: 0.4857 (0.5191) time: 0.4307 data: 0.0002 max mem: 52150 +Epoch: [22] [7220/7269] eta: 0:00:21 lr: 0.000005 loss: 0.4939 (0.5192) time: 0.4162 data: 0.0002 max mem: 52150 +Epoch: [22] [7230/7269] eta: 0:00:16 lr: 0.000005 loss: 0.5433 (0.5192) time: 0.4167 data: 0.0002 max mem: 52150 +Epoch: [22] [7240/7269] eta: 0:00:12 lr: 0.000005 loss: 0.5099 (0.5192) time: 0.4349 data: 0.0002 max mem: 52150 +Epoch: [22] [7250/7269] eta: 0:00:08 lr: 0.000005 loss: 0.5348 (0.5192) time: 0.4853 data: 0.0026 max mem: 52150 +Epoch: [22] [7260/7269] eta: 0:00:03 lr: 0.000005 loss: 0.5373 (0.5192) time: 0.4222 data: 0.0026 max mem: 52150 +Epoch: [22] [7268/7269] eta: 0:00:00 lr: 0.000005 loss: 0.4846 (0.5191) time: 0.4204 data: 0.0026 max mem: 52150 +Epoch: [22] Total time: 0:52:17 (0.4316 s / it) +Averaged stats: lr: 0.000005 loss: 0.4846 (0.5198) +Test: [ 0/1403] eta: 1:17:27 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0793 (0.0793) acc1: 100.0000 (100.0000) acc5: 100.0000 (100.0000) time: 3.3127 data: 3.2139 max mem: 52150 +Test: [ 10/1403] eta: 0:09:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1092 (0.1385) acc1: 98.4375 (97.0170) acc5: 100.0000 (100.0000) time: 0.3888 data: 0.2925 max mem: 52150 +Test: [ 20/1403] eta: 0:05:45 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1062 (0.1304) acc1: 98.4375 (97.0982) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 30/1403] eta: 0:04:35 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0745 (0.1217) acc1: 100.0000 (97.4294) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 40/1403] eta: 0:03:58 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0837 (0.1250) acc1: 100.0000 (97.4085) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 50/1403] eta: 0:03:35 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0994 (0.1428) acc1: 100.0000 (96.2929) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 60/1403] eta: 0:03:20 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1007 (0.1399) acc1: 100.0000 (96.6701) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 70/1403] eta: 0:03:08 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.1167 (0.1402) acc1: 98.4375 (96.5229) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 80/1403] eta: 0:03:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1154 (0.1417) acc1: 98.4375 (96.4506) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 90/1403] eta: 0:02:53 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1172 (0.1543) acc1: 96.8750 (95.7589) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 100/1403] eta: 0:02:47 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1172 (0.1521) acc1: 96.8750 (95.9158) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 110/1403] eta: 0:02:42 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1017 (0.1493) acc1: 98.4375 (96.0867) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 120/1403] eta: 0:02:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0759 (0.1432) acc1: 100.0000 (96.3843) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 130/1403] eta: 0:02:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0757 (0.1392) acc1: 100.0000 (96.5768) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 140/1403] eta: 0:02:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0710 (0.1343) acc1: 100.0000 (96.8196) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 150/1403] eta: 0:02:27 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0704 (0.1304) acc1: 100.0000 (96.9992) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 160/1403] eta: 0:02:24 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0720 (0.1276) acc1: 100.0000 (97.1176) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 170/1403] eta: 0:02:22 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0720 (0.1248) acc1: 100.0000 (97.2588) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 180/1403] eta: 0:02:19 acc1_pos: 89.062 acc5_pos: 100.000 loss: 0.0733 (0.1306) acc1: 100.0000 (96.9700) acc5: 100.0000 (100.0000) time: 0.0963 data: 0.0003 max mem: 52150 +Test: [ 190/1403] eta: 0:02:17 acc1_pos: 87.500 acc5_pos: 100.000 loss: 0.3087 (0.1417) acc1: 89.0625 (96.3351) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 200/1403] eta: 0:02:15 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.3193 (0.1540) acc1: 87.5000 (95.7400) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 210/1403] eta: 0:02:13 acc1_pos: 75.000 acc5_pos: 100.000 loss: 0.3333 (0.1662) acc1: 85.9375 (95.0459) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 220/1403] eta: 0:02:11 acc1_pos: 87.500 acc5_pos: 100.000 loss: 0.3555 (0.1765) acc1: 82.8125 (94.5206) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 230/1403] eta: 0:02:09 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2550 (0.1731) acc1: 89.0625 (94.7037) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 240/1403] eta: 0:02:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0703 (0.1688) acc1: 100.0000 (94.9170) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 250/1403] eta: 0:02:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0675 (0.1651) acc1: 100.0000 (95.0946) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 260/1403] eta: 0:02:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0683 (0.1617) acc1: 100.0000 (95.2646) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 270/1403] eta: 0:02:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0691 (0.1584) acc1: 100.0000 (95.4163) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 280/1403] eta: 0:02:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0691 (0.1552) acc1: 100.0000 (95.5738) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 290/1403] eta: 0:01:59 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0646 (0.1522) acc1: 100.0000 (95.7206) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 300/1403] eta: 0:01:58 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0645 (0.1495) acc1: 100.0000 (95.8524) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 310/1403] eta: 0:01:56 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0698 (0.1471) acc1: 100.0000 (95.9857) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 320/1403] eta: 0:01:55 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0951 (0.1456) acc1: 100.0000 (96.1108) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 330/1403] eta: 0:01:53 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1018 (0.1444) acc1: 100.0000 (96.2236) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 340/1403] eta: 0:01:52 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1018 (0.1431) acc1: 100.0000 (96.3343) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 350/1403] eta: 0:01:51 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0989 (0.1418) acc1: 100.0000 (96.4387) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 360/1403] eta: 0:01:49 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1011 (0.1408) acc1: 100.0000 (96.5287) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 370/1403] eta: 0:01:48 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0999 (0.1397) acc1: 100.0000 (96.6223) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 380/1403] eta: 0:01:47 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0991 (0.1387) acc1: 100.0000 (96.7110) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 390/1403] eta: 0:01:46 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0998 (0.1398) acc1: 100.0000 (96.6592) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 400/1403] eta: 0:01:44 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2573 (0.1445) acc1: 89.0625 (96.3996) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 410/1403] eta: 0:01:43 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3010 (0.1485) acc1: 85.9375 (96.1603) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 420/1403] eta: 0:01:42 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2737 (0.1510) acc1: 89.0625 (96.0214) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 430/1403] eta: 0:01:41 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2669 (0.1545) acc1: 90.6250 (95.8563) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 440/1403] eta: 0:01:39 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3281 (0.1588) acc1: 87.5000 (95.6562) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 450/1403] eta: 0:01:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2911 (0.1617) acc1: 87.5000 (95.5030) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 460/1403] eta: 0:01:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2898 (0.1650) acc1: 87.5000 (95.3227) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 470/1403] eta: 0:01:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2590 (0.1645) acc1: 90.6250 (95.3424) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 480/1403] eta: 0:01:35 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0609 (0.1623) acc1: 100.0000 (95.4392) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 490/1403] eta: 0:01:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0618 (0.1603) acc1: 100.0000 (95.5321) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 500/1403] eta: 0:01:32 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0618 (0.1583) acc1: 100.0000 (95.6213) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 510/1403] eta: 0:01:31 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0617 (0.1565) acc1: 100.0000 (95.7069) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 520/1403] eta: 0:01:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0643 (0.1548) acc1: 100.0000 (95.7893) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0004 max mem: 52150 +Test: [ 530/1403] eta: 0:01:29 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0588 (0.1529) acc1: 100.0000 (95.8686) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 540/1403] eta: 0:01:28 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0585 (0.1512) acc1: 100.0000 (95.9450) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 550/1403] eta: 0:01:27 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0590 (0.1500) acc1: 100.0000 (95.9988) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 560/1403] eta: 0:01:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0854 (0.1491) acc1: 98.4375 (96.0394) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 570/1403] eta: 0:01:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0902 (0.1481) acc1: 98.4375 (96.0869) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 580/1403] eta: 0:01:24 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0949 (0.1477) acc1: 98.4375 (96.1032) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 590/1403] eta: 0:01:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1252 (0.1477) acc1: 96.8750 (96.0898) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 600/1403] eta: 0:01:21 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1427 (0.1486) acc1: 93.7500 (96.0197) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 610/1403] eta: 0:01:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1437 (0.1483) acc1: 93.7500 (96.0234) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 620/1403] eta: 0:01:19 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1559 (0.1502) acc1: 93.7500 (95.9013) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 630/1403] eta: 0:01:18 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1705 (0.1502) acc1: 92.1875 (95.8944) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 640/1403] eta: 0:01:17 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1247 (0.1500) acc1: 95.3125 (95.9024) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 650/1403] eta: 0:01:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1213 (0.1496) acc1: 96.8750 (95.9197) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 660/1403] eta: 0:01:15 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0964 (0.1486) acc1: 98.4375 (95.9720) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 670/1403] eta: 0:01:14 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0701 (0.1475) acc1: 100.0000 (96.0204) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 680/1403] eta: 0:01:13 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0869 (0.1501) acc1: 96.8750 (95.8677) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 690/1403] eta: 0:01:12 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3772 (0.1554) acc1: 82.8125 (95.5816) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 700/1403] eta: 0:01:11 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2244 (0.1559) acc1: 87.5000 (95.5555) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 710/1403] eta: 0:01:10 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3171 (0.1605) acc1: 87.5000 (95.3213) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 720/1403] eta: 0:01:08 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3171 (0.1605) acc1: 84.3750 (95.3147) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 730/1403] eta: 0:01:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0690 (0.1595) acc1: 100.0000 (95.3638) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 740/1403] eta: 0:01:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0757 (0.1586) acc1: 98.4375 (95.4053) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 750/1403] eta: 0:01:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1071 (0.1615) acc1: 96.8750 (95.2418) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 760/1403] eta: 0:01:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4497 (0.1668) acc1: 76.5625 (94.9121) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 770/1403] eta: 0:01:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.7216 (0.1741) acc1: 59.3750 (94.4755) acc5: 100.0000 (100.0000) acc1_neg: 59.375 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 780/1403] eta: 0:01:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6331 (0.1786) acc1: 64.0625 (94.2181) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 790/1403] eta: 0:01:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3809 (0.1815) acc1: 79.6875 (94.0740) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 800/1403] eta: 0:01:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2463 (0.1821) acc1: 87.5000 (94.0543) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 810/1403] eta: 0:00:59 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2417 (0.1856) acc1: 87.5000 (93.8906) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 820/1403] eta: 0:00:58 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6760 (0.1951) acc1: 70.3125 (93.4969) acc5: 100.0000 (100.0000) acc1_neg: 40.625 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 830/1403] eta: 0:00:57 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.7160 (0.1979) acc1: 67.1875 (93.3645) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 840/1403] eta: 0:00:56 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5082 (0.2063) acc1: 73.4375 (93.0273) acc5: 100.0000 (100.0000) acc1_neg: 50.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 850/1403] eta: 0:00:55 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.1373 (0.2179) acc1: 54.6875 (92.5510) acc5: 100.0000 (100.0000) acc1_neg: 50.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 860/1403] eta: 0:00:54 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.0619 (0.2216) acc1: 56.2500 (92.3980) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 870/1403] eta: 0:00:53 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1464 (0.2206) acc1: 95.3125 (92.4404) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 880/1403] eta: 0:00:52 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1046 (0.2194) acc1: 96.8750 (92.4979) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 890/1403] eta: 0:00:51 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0973 (0.2179) acc1: 98.4375 (92.5680) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 900/1403] eta: 0:00:50 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0777 (0.2163) acc1: 100.0000 (92.6471) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 910/1403] eta: 0:00:49 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0812 (0.2155) acc1: 98.4375 (92.6815) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 920/1403] eta: 0:00:48 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1434 (0.2147) acc1: 95.3125 (92.7083) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 930/1403] eta: 0:00:47 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1179 (0.2139) acc1: 96.8750 (92.7497) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 940/1403] eta: 0:00:46 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1640 (0.2142) acc1: 95.3125 (92.7089) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 950/1403] eta: 0:00:45 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3053 (0.2156) acc1: 84.3750 (92.6213) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.0971 data: 0.0003 max mem: 52150 +Test: [ 960/1403] eta: 0:00:44 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3139 (0.2165) acc1: 85.9375 (92.5745) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0971 data: 0.0003 max mem: 52150 +Test: [ 970/1403] eta: 0:00:43 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2765 (0.2178) acc1: 89.0625 (92.5158) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 980/1403] eta: 0:00:42 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3074 (0.2190) acc1: 85.9375 (92.4312) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 990/1403] eta: 0:00:41 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3021 (0.2197) acc1: 85.9375 (92.3814) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1000/1403] eta: 0:00:40 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2867 (0.2208) acc1: 87.5000 (92.3280) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1010/1403] eta: 0:00:39 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3229 (0.2220) acc1: 85.9375 (92.2617) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1020/1403] eta: 0:00:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4775 (0.2292) acc1: 81.2500 (91.8830) acc5: 100.0000 (100.0000) acc1_neg: 31.250 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1030/1403] eta: 0:00:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6336 (0.2316) acc1: 57.8125 (91.7435) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1040/1403] eta: 0:00:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2316 (0.2316) acc1: 87.5000 (91.7312) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1050/1403] eta: 0:00:35 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2050 (0.2312) acc1: 92.1875 (91.7578) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1060/1403] eta: 0:00:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1487 (0.2303) acc1: 95.3125 (91.8061) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1070/1403] eta: 0:00:33 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1487 (0.2307) acc1: 95.3125 (91.7951) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1080/1403] eta: 0:00:32 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3766 (0.2326) acc1: 84.3750 (91.7018) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1090/1403] eta: 0:00:31 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4298 (0.2345) acc1: 81.2500 (91.6132) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1100/1403] eta: 0:00:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3393 (0.2347) acc1: 85.9375 (91.6014) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1110/1403] eta: 0:00:29 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2541 (0.2350) acc1: 90.6250 (91.5968) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1120/1403] eta: 0:00:28 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2567 (0.2352) acc1: 90.6250 (91.5812) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1130/1403] eta: 0:00:27 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2572 (0.2356) acc1: 89.0625 (91.5644) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1140/1403] eta: 0:00:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2361 (0.2354) acc1: 90.6250 (91.5658) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0966 data: 0.0002 max mem: 52150 +Test: [1150/1403] eta: 0:00:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2310 (0.2355) acc1: 92.1875 (91.5644) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0966 data: 0.0002 max mem: 52150 +Test: [1160/1403] eta: 0:00:24 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2546 (0.2358) acc1: 90.6250 (91.5509) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0966 data: 0.0002 max mem: 52150 +Test: [1170/1403] eta: 0:00:23 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2533 (0.2355) acc1: 90.6250 (91.5617) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1180/1403] eta: 0:00:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0695 (0.2341) acc1: 100.0000 (91.6331) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52150 +Test: [1190/1403] eta: 0:00:21 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0685 (0.2327) acc1: 100.0000 (91.7034) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1200/1403] eta: 0:00:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0684 (0.2313) acc1: 100.0000 (91.7725) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1210/1403] eta: 0:00:19 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0679 (0.2300) acc1: 100.0000 (91.8404) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1220/1403] eta: 0:00:18 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0679 (0.2287) acc1: 100.0000 (91.9072) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1230/1403] eta: 0:00:17 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0688 (0.2274) acc1: 100.0000 (91.9730) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1240/1403] eta: 0:00:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0687 (0.2261) acc1: 100.0000 (92.0377) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1250/1403] eta: 0:00:15 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0683 (0.2248) acc1: 100.0000 (92.1013) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1260/1403] eta: 0:00:14 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0713 (0.2236) acc1: 100.0000 (92.1640) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1270/1403] eta: 0:00:13 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0728 (0.2224) acc1: 100.0000 (92.2256) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1280/1403] eta: 0:00:12 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0728 (0.2213) acc1: 100.0000 (92.2863) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1290/1403] eta: 0:00:11 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0720 (0.2201) acc1: 100.0000 (92.3460) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52150 +Test: [1300/1403] eta: 0:00:10 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0730 (0.2190) acc1: 100.0000 (92.4049) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1310/1403] eta: 0:00:09 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0729 (0.2179) acc1: 100.0000 (92.4628) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1320/1403] eta: 0:00:08 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0714 (0.2168) acc1: 100.0000 (92.5199) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52150 +Test: [1330/1403] eta: 0:00:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0711 (0.2157) acc1: 100.0000 (92.5737) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52150 +Test: [1340/1403] eta: 0:00:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0660 (0.2146) acc1: 100.0000 (92.6279) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1350/1403] eta: 0:00:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0649 (0.2135) acc1: 100.0000 (92.6802) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1360/1403] eta: 0:00:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0646 (0.2124) acc1: 100.0000 (92.7340) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52150 +Test: [1370/1403] eta: 0:00:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0651 (0.2114) acc1: 100.0000 (92.7847) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1380/1403] eta: 0:00:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0641 (0.2103) acc1: 100.0000 (92.8369) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1390/1403] eta: 0:00:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0636 (0.2093) acc1: 100.0000 (92.8884) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52150 +Test: [1400/1403] eta: 0:00:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0647 (0.2083) acc1: 100.0000 (92.9370) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0001 max mem: 52150 +Test: [1402/1403] eta: 0:00:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0653 (0.2081) acc1: 100.0000 (92.9421) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0922 data: 0.0001 max mem: 52150 +Test: Total time: 0:02:19 (0.0992 s / it) +* Overall Acc@1 92.942 Acc@5 100.000 + target == real Acc@1 95.950 Acc@5 100.000 + target == fake Acc@1 92.094 Acc@5 100.000 + loss 0.208 +Accuracy of the network on the 89729 test images: 92.9% +Max accuracy: 93.40% +Epoch: [23] [ 0/7269] eta: 11:53:00 lr: 0.000005 loss: 0.5449 (0.5449) time: 5.8854 data: 5.5164 max mem: 52150 +Epoch: [23] [ 10/7269] eta: 1:46:07 lr: 0.000005 loss: 0.5185 (0.4999) time: 0.8771 data: 0.5017 max mem: 52150 +Epoch: [23] [ 20/7269] eta: 1:16:52 lr: 0.000005 loss: 0.5131 (0.5007) time: 0.3739 data: 0.0002 max mem: 52150 +Epoch: [23] [ 30/7269] eta: 1:06:28 lr: 0.000005 loss: 0.5233 (0.5015) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [23] [ 40/7269] eta: 1:01:07 lr: 0.000005 loss: 0.5297 (0.5134) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [23] [ 50/7269] eta: 0:57:51 lr: 0.000005 loss: 0.5440 (0.5203) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [23] [ 60/7269] eta: 0:55:38 lr: 0.000005 loss: 0.5223 (0.5178) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [23] [ 70/7269] eta: 0:56:08 lr: 0.000005 loss: 0.4976 (0.5173) time: 0.4345 data: 0.0002 max mem: 52150 +Epoch: [23] [ 80/7269] eta: 0:56:29 lr: 0.000005 loss: 0.4987 (0.5174) time: 0.4969 data: 0.0002 max mem: 52150 +Epoch: [23] [ 90/7269] eta: 0:55:05 lr: 0.000005 loss: 0.5260 (0.5173) time: 0.4342 data: 0.0001 max mem: 52150 +Epoch: [23] [ 100/7269] eta: 0:53:57 lr: 0.000005 loss: 0.5260 (0.5162) time: 0.3712 data: 0.0002 max mem: 52150 +Epoch: [23] [ 110/7269] eta: 0:53:01 lr: 0.000005 loss: 0.5005 (0.5163) time: 0.3712 data: 0.0002 max mem: 52150 +Epoch: [23] [ 120/7269] eta: 0:54:23 lr: 0.000005 loss: 0.5545 (0.5178) time: 0.4809 data: 0.0002 max mem: 52150 +Epoch: [23] [ 130/7269] eta: 0:54:24 lr: 0.000005 loss: 0.5567 (0.5198) time: 0.5285 data: 0.0002 max mem: 52150 +Epoch: [23] [ 140/7269] eta: 0:54:24 lr: 0.000005 loss: 0.5046 (0.5165) time: 0.4670 data: 0.0002 max mem: 52150 +Epoch: [23] [ 150/7269] eta: 0:54:37 lr: 0.000005 loss: 0.5103 (0.5181) time: 0.4812 data: 0.0002 max mem: 52150 +Epoch: [23] [ 160/7269] eta: 0:53:53 lr: 0.000005 loss: 0.5182 (0.5183) time: 0.4333 data: 0.0002 max mem: 52150 +Epoch: [23] [ 170/7269] eta: 0:54:06 lr: 0.000005 loss: 0.5418 (0.5198) time: 0.4336 data: 0.0002 max mem: 52150 +Epoch: [23] [ 180/7269] eta: 0:53:28 lr: 0.000005 loss: 0.5221 (0.5195) time: 0.4336 data: 0.0002 max mem: 52150 +Epoch: [23] [ 190/7269] eta: 0:54:18 lr: 0.000005 loss: 0.5142 (0.5195) time: 0.4861 data: 0.0001 max mem: 52150 +Epoch: [23] [ 200/7269] eta: 0:53:42 lr: 0.000005 loss: 0.5090 (0.5189) time: 0.4861 data: 0.0001 max mem: 52150 +Epoch: [23] [ 210/7269] eta: 0:53:09 lr: 0.000005 loss: 0.4744 (0.5165) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [23] [ 220/7269] eta: 0:52:39 lr: 0.000005 loss: 0.4846 (0.5157) time: 0.3710 data: 0.0002 max mem: 52150 +Epoch: [23] [ 230/7269] eta: 0:53:27 lr: 0.000005 loss: 0.5047 (0.5158) time: 0.4966 data: 0.0002 max mem: 52150 +Epoch: [23] [ 240/7269] eta: 0:53:35 lr: 0.000005 loss: 0.5196 (0.5151) time: 0.5597 data: 0.0002 max mem: 52150 +Epoch: [23] [ 250/7269] eta: 0:53:34 lr: 0.000005 loss: 0.5044 (0.5138) time: 0.4838 data: 0.0002 max mem: 52150 +Epoch: [23] [ 260/7269] eta: 0:53:33 lr: 0.000005 loss: 0.4996 (0.5126) time: 0.4712 data: 0.0001 max mem: 52150 +Epoch: [23] [ 270/7269] eta: 0:53:06 lr: 0.000005 loss: 0.4996 (0.5122) time: 0.4219 data: 0.0001 max mem: 52150 +Epoch: [23] [ 280/7269] eta: 0:53:04 lr: 0.000005 loss: 0.5433 (0.5148) time: 0.4184 data: 0.0002 max mem: 52150 +Epoch: [23] [ 290/7269] eta: 0:52:39 lr: 0.000005 loss: 0.5687 (0.5161) time: 0.4182 data: 0.0002 max mem: 52150 +Epoch: [23] [ 300/7269] eta: 0:52:45 lr: 0.000005 loss: 0.5355 (0.5151) time: 0.4337 data: 0.0002 max mem: 52150 +Epoch: [23] [ 310/7269] eta: 0:53:11 lr: 0.000005 loss: 0.5281 (0.5158) time: 0.5444 data: 0.0002 max mem: 52150 +Epoch: [23] [ 320/7269] eta: 0:52:48 lr: 0.000005 loss: 0.5462 (0.5170) time: 0.4827 data: 0.0002 max mem: 52150 +Epoch: [23] [ 330/7269] eta: 0:52:26 lr: 0.000005 loss: 0.5366 (0.5170) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [23] [ 340/7269] eta: 0:52:05 lr: 0.000005 loss: 0.5142 (0.5178) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [23] [ 350/7269] eta: 0:51:44 lr: 0.000005 loss: 0.5368 (0.5188) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [23] [ 360/7269] eta: 0:51:25 lr: 0.000005 loss: 0.5177 (0.5189) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [23] [ 370/7269] eta: 0:51:07 lr: 0.000005 loss: 0.5146 (0.5186) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [23] [ 380/7269] eta: 0:50:49 lr: 0.000005 loss: 0.5318 (0.5184) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [23] [ 390/7269] eta: 0:50:55 lr: 0.000005 loss: 0.5456 (0.5184) time: 0.4355 data: 0.0001 max mem: 52150 +Epoch: [23] [ 400/7269] eta: 0:50:55 lr: 0.000005 loss: 0.5113 (0.5176) time: 0.4865 data: 0.0001 max mem: 52150 +Epoch: [23] [ 410/7269] eta: 0:50:55 lr: 0.000005 loss: 0.4919 (0.5167) time: 0.4713 data: 0.0002 max mem: 52150 +Epoch: [23] [ 420/7269] eta: 0:51:12 lr: 0.000005 loss: 0.4727 (0.5163) time: 0.5230 data: 0.0002 max mem: 52150 +Epoch: [23] [ 430/7269] eta: 0:51:15 lr: 0.000005 loss: 0.4952 (0.5154) time: 0.5353 data: 0.0002 max mem: 52150 +Epoch: [23] [ 440/7269] eta: 0:51:17 lr: 0.000005 loss: 0.5019 (0.5154) time: 0.4945 data: 0.0002 max mem: 52150 +Epoch: [23] [ 450/7269] eta: 0:51:00 lr: 0.000005 loss: 0.4969 (0.5148) time: 0.4328 data: 0.0002 max mem: 52150 +Epoch: [23] [ 460/7269] eta: 0:50:45 lr: 0.000005 loss: 0.5344 (0.5155) time: 0.3712 data: 0.0002 max mem: 52150 +Epoch: [23] [ 470/7269] eta: 0:50:29 lr: 0.000005 loss: 0.5508 (0.5148) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [23] [ 480/7269] eta: 0:50:28 lr: 0.000005 loss: 0.4842 (0.5143) time: 0.4197 data: 0.0001 max mem: 52150 +Epoch: [23] [ 490/7269] eta: 0:50:13 lr: 0.000005 loss: 0.5101 (0.5141) time: 0.4195 data: 0.0002 max mem: 52150 +Epoch: [23] [ 500/7269] eta: 0:49:59 lr: 0.000005 loss: 0.5225 (0.5141) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [23] [ 510/7269] eta: 0:49:45 lr: 0.000005 loss: 0.5081 (0.5140) time: 0.3711 data: 0.0002 max mem: 52150 +Epoch: [23] [ 520/7269] eta: 0:49:31 lr: 0.000005 loss: 0.5280 (0.5145) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [23] [ 530/7269] eta: 0:49:18 lr: 0.000005 loss: 0.5543 (0.5151) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [23] [ 540/7269] eta: 0:49:18 lr: 0.000005 loss: 0.5543 (0.5150) time: 0.4203 data: 0.0001 max mem: 52150 +Epoch: [23] [ 550/7269] eta: 0:49:05 lr: 0.000005 loss: 0.5340 (0.5151) time: 0.4201 data: 0.0001 max mem: 52150 +Epoch: [23] [ 560/7269] eta: 0:48:52 lr: 0.000005 loss: 0.4998 (0.5147) time: 0.3712 data: 0.0002 max mem: 52150 +Epoch: [23] [ 570/7269] eta: 0:48:40 lr: 0.000005 loss: 0.4998 (0.5148) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [23] [ 580/7269] eta: 0:48:29 lr: 0.000005 loss: 0.5309 (0.5147) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [23] [ 590/7269] eta: 0:48:17 lr: 0.000005 loss: 0.4973 (0.5144) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [23] [ 600/7269] eta: 0:48:29 lr: 0.000005 loss: 0.5336 (0.5149) time: 0.4764 data: 0.0002 max mem: 52150 +Epoch: [23] [ 610/7269] eta: 0:48:18 lr: 0.000005 loss: 0.5469 (0.5154) time: 0.4762 data: 0.0002 max mem: 52150 +Epoch: [23] [ 620/7269] eta: 0:48:07 lr: 0.000005 loss: 0.5245 (0.5155) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [23] [ 630/7269] eta: 0:48:16 lr: 0.000005 loss: 0.5331 (0.5162) time: 0.4687 data: 0.0002 max mem: 52150 +Epoch: [23] [ 640/7269] eta: 0:48:05 lr: 0.000005 loss: 0.5428 (0.5165) time: 0.4683 data: 0.0002 max mem: 52150 +Epoch: [23] [ 650/7269] eta: 0:48:12 lr: 0.000005 loss: 0.5422 (0.5166) time: 0.4612 data: 0.0002 max mem: 52150 +Epoch: [23] [ 660/7269] eta: 0:48:01 lr: 0.000005 loss: 0.5054 (0.5167) time: 0.4613 data: 0.0002 max mem: 52150 +Epoch: [23] [ 670/7269] eta: 0:48:03 lr: 0.000005 loss: 0.4908 (0.5161) time: 0.4330 data: 0.0002 max mem: 52150 +Epoch: [23] [ 680/7269] eta: 0:47:52 lr: 0.000005 loss: 0.4972 (0.5160) time: 0.4333 data: 0.0002 max mem: 52150 +Epoch: [23] [ 690/7269] eta: 0:47:42 lr: 0.000005 loss: 0.4972 (0.5155) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [23] [ 700/7269] eta: 0:47:43 lr: 0.000005 loss: 0.4929 (0.5152) time: 0.4327 data: 0.0002 max mem: 52150 +Epoch: [23] [ 710/7269] eta: 0:47:43 lr: 0.000005 loss: 0.5037 (0.5153) time: 0.4922 data: 0.0002 max mem: 52150 +Epoch: [23] [ 720/7269] eta: 0:47:41 lr: 0.000005 loss: 0.5428 (0.5157) time: 0.4763 data: 0.0002 max mem: 52150 +Epoch: [23] [ 730/7269] eta: 0:47:47 lr: 0.000005 loss: 0.5392 (0.5157) time: 0.5072 data: 0.0002 max mem: 52150 +Epoch: [23] [ 740/7269] eta: 0:47:37 lr: 0.000005 loss: 0.5243 (0.5160) time: 0.4618 data: 0.0002 max mem: 52150 +Epoch: [23] [ 750/7269] eta: 0:47:27 lr: 0.000005 loss: 0.4980 (0.5157) time: 0.3712 data: 0.0001 max mem: 52150 +Epoch: [23] [ 760/7269] eta: 0:47:26 lr: 0.000005 loss: 0.4929 (0.5157) time: 0.4222 data: 0.0002 max mem: 52150 +Epoch: [23] [ 770/7269] eta: 0:47:16 lr: 0.000005 loss: 0.5327 (0.5159) time: 0.4228 data: 0.0002 max mem: 52150 +Epoch: [23] [ 780/7269] eta: 0:47:06 lr: 0.000005 loss: 0.5411 (0.5163) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [23] [ 790/7269] eta: 0:46:57 lr: 0.000005 loss: 0.5411 (0.5162) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [23] [ 800/7269] eta: 0:47:04 lr: 0.000005 loss: 0.5432 (0.5166) time: 0.4784 data: 0.0002 max mem: 52150 +Epoch: [23] [ 810/7269] eta: 0:47:10 lr: 0.000005 loss: 0.5647 (0.5171) time: 0.5718 data: 0.0002 max mem: 52150 +Epoch: [23] [ 820/7269] eta: 0:47:09 lr: 0.000005 loss: 0.5416 (0.5172) time: 0.5251 data: 0.0002 max mem: 52150 +Epoch: [23] [ 830/7269] eta: 0:47:00 lr: 0.000005 loss: 0.5207 (0.5173) time: 0.4316 data: 0.0002 max mem: 52150 +Epoch: [23] [ 840/7269] eta: 0:47:00 lr: 0.000005 loss: 0.5570 (0.5178) time: 0.4317 data: 0.0002 max mem: 52150 +Epoch: [23] [ 850/7269] eta: 0:46:50 lr: 0.000005 loss: 0.5570 (0.5178) time: 0.4316 data: 0.0002 max mem: 52150 +Epoch: [23] [ 860/7269] eta: 0:46:41 lr: 0.000005 loss: 0.5371 (0.5177) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [23] [ 870/7269] eta: 0:46:47 lr: 0.000005 loss: 0.4968 (0.5174) time: 0.4767 data: 0.0002 max mem: 52150 +Epoch: [23] [ 880/7269] eta: 0:46:38 lr: 0.000005 loss: 0.4836 (0.5172) time: 0.4763 data: 0.0002 max mem: 52150 +Epoch: [23] [ 890/7269] eta: 0:46:35 lr: 0.000005 loss: 0.5169 (0.5173) time: 0.4184 data: 0.0002 max mem: 52150 +Epoch: [23] [ 900/7269] eta: 0:46:26 lr: 0.000005 loss: 0.4940 (0.5167) time: 0.4183 data: 0.0002 max mem: 52150 +Epoch: [23] [ 910/7269] eta: 0:46:34 lr: 0.000005 loss: 0.4768 (0.5167) time: 0.4920 data: 0.0002 max mem: 52150 +Epoch: [23] [ 920/7269] eta: 0:46:25 lr: 0.000005 loss: 0.5120 (0.5166) time: 0.4921 data: 0.0002 max mem: 52150 +Epoch: [23] [ 930/7269] eta: 0:46:16 lr: 0.000005 loss: 0.5564 (0.5172) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [23] [ 940/7269] eta: 0:46:14 lr: 0.000005 loss: 0.5681 (0.5170) time: 0.4189 data: 0.0002 max mem: 52150 +Epoch: [23] [ 950/7269] eta: 0:46:05 lr: 0.000005 loss: 0.4734 (0.5167) time: 0.4187 data: 0.0002 max mem: 52150 +Epoch: [23] [ 960/7269] eta: 0:45:56 lr: 0.000005 loss: 0.5281 (0.5168) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [23] [ 970/7269] eta: 0:45:53 lr: 0.000005 loss: 0.5407 (0.5171) time: 0.4179 data: 0.0002 max mem: 52150 +Epoch: [23] [ 980/7269] eta: 0:45:53 lr: 0.000005 loss: 0.5432 (0.5172) time: 0.4784 data: 0.0002 max mem: 52150 +Epoch: [23] [ 990/7269] eta: 0:45:44 lr: 0.000005 loss: 0.5219 (0.5170) time: 0.4323 data: 0.0002 max mem: 52150 +Epoch: [23] [1000/7269] eta: 0:45:41 lr: 0.000005 loss: 0.4975 (0.5170) time: 0.4169 data: 0.0002 max mem: 52150 +Epoch: [23] [1010/7269] eta: 0:45:33 lr: 0.000005 loss: 0.5267 (0.5173) time: 0.4167 data: 0.0002 max mem: 52150 +Epoch: [23] [1020/7269] eta: 0:45:24 lr: 0.000005 loss: 0.5361 (0.5174) time: 0.3711 data: 0.0002 max mem: 52150 +Epoch: [23] [1030/7269] eta: 0:45:16 lr: 0.000005 loss: 0.5062 (0.5173) time: 0.3711 data: 0.0002 max mem: 52150 +Epoch: [23] [1040/7269] eta: 0:45:08 lr: 0.000005 loss: 0.5013 (0.5174) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [23] [1050/7269] eta: 0:45:00 lr: 0.000005 loss: 0.5557 (0.5175) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [23] [1060/7269] eta: 0:44:52 lr: 0.000005 loss: 0.5589 (0.5177) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [23] [1070/7269] eta: 0:44:44 lr: 0.000005 loss: 0.5379 (0.5179) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [23] [1080/7269] eta: 0:44:36 lr: 0.000005 loss: 0.5347 (0.5179) time: 0.3711 data: 0.0002 max mem: 52150 +Epoch: [23] [1090/7269] eta: 0:44:28 lr: 0.000005 loss: 0.5250 (0.5181) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [23] [1100/7269] eta: 0:44:33 lr: 0.000005 loss: 0.5007 (0.5176) time: 0.4852 data: 0.0002 max mem: 52150 +Epoch: [23] [1110/7269] eta: 0:44:31 lr: 0.000005 loss: 0.5007 (0.5177) time: 0.5312 data: 0.0002 max mem: 52150 +Epoch: [23] [1120/7269] eta: 0:44:23 lr: 0.000005 loss: 0.4744 (0.5173) time: 0.4178 data: 0.0002 max mem: 52150 +Epoch: [23] [1130/7269] eta: 0:44:15 lr: 0.000005 loss: 0.4624 (0.5170) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [23] [1140/7269] eta: 0:44:08 lr: 0.000005 loss: 0.4910 (0.5169) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [23] [1150/7269] eta: 0:44:06 lr: 0.000005 loss: 0.5209 (0.5168) time: 0.4221 data: 0.0002 max mem: 52150 +Epoch: [23] [1160/7269] eta: 0:43:58 lr: 0.000005 loss: 0.5269 (0.5168) time: 0.4221 data: 0.0002 max mem: 52150 +Epoch: [23] [1170/7269] eta: 0:43:51 lr: 0.000005 loss: 0.5333 (0.5172) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [23] [1180/7269] eta: 0:43:48 lr: 0.000005 loss: 0.5505 (0.5172) time: 0.4231 data: 0.0002 max mem: 52150 +Epoch: [23] [1190/7269] eta: 0:43:41 lr: 0.000005 loss: 0.5262 (0.5173) time: 0.4233 data: 0.0002 max mem: 52150 +Epoch: [23] [1200/7269] eta: 0:43:38 lr: 0.000005 loss: 0.5319 (0.5175) time: 0.4169 data: 0.0002 max mem: 52150 +Epoch: [23] [1210/7269] eta: 0:43:36 lr: 0.000005 loss: 0.5375 (0.5178) time: 0.4627 data: 0.0002 max mem: 52150 +Epoch: [23] [1220/7269] eta: 0:43:33 lr: 0.000005 loss: 0.5374 (0.5178) time: 0.4628 data: 0.0002 max mem: 52150 +Epoch: [23] [1230/7269] eta: 0:43:30 lr: 0.000005 loss: 0.5196 (0.5179) time: 0.4651 data: 0.0002 max mem: 52150 +Epoch: [23] [1240/7269] eta: 0:43:28 lr: 0.000005 loss: 0.5168 (0.5179) time: 0.4694 data: 0.0002 max mem: 52150 +Epoch: [23] [1250/7269] eta: 0:43:26 lr: 0.000005 loss: 0.5168 (0.5179) time: 0.4819 data: 0.0002 max mem: 52150 +Epoch: [23] [1260/7269] eta: 0:43:25 lr: 0.000005 loss: 0.5289 (0.5181) time: 0.4934 data: 0.0002 max mem: 52150 +Epoch: [23] [1270/7269] eta: 0:43:22 lr: 0.000005 loss: 0.5302 (0.5183) time: 0.4785 data: 0.0002 max mem: 52150 +Epoch: [23] [1280/7269] eta: 0:43:15 lr: 0.000005 loss: 0.5302 (0.5184) time: 0.4175 data: 0.0002 max mem: 52150 +Epoch: [23] [1290/7269] eta: 0:43:08 lr: 0.000005 loss: 0.5446 (0.5186) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [23] [1300/7269] eta: 0:43:00 lr: 0.000005 loss: 0.5466 (0.5185) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [23] [1310/7269] eta: 0:42:53 lr: 0.000005 loss: 0.5249 (0.5181) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [23] [1320/7269] eta: 0:43:00 lr: 0.000005 loss: 0.5249 (0.5180) time: 0.5253 data: 0.0002 max mem: 52150 +Epoch: [23] [1330/7269] eta: 0:43:04 lr: 0.000005 loss: 0.5414 (0.5182) time: 0.6458 data: 0.0002 max mem: 52150 +Epoch: [23] [1340/7269] eta: 0:42:57 lr: 0.000005 loss: 0.5398 (0.5182) time: 0.4926 data: 0.0002 max mem: 52150 +Epoch: [23] [1350/7269] eta: 0:42:50 lr: 0.000005 loss: 0.5429 (0.5184) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [23] [1360/7269] eta: 0:42:47 lr: 0.000005 loss: 0.5449 (0.5185) time: 0.4181 data: 0.0002 max mem: 52150 +Epoch: [23] [1370/7269] eta: 0:42:49 lr: 0.000005 loss: 0.5241 (0.5186) time: 0.5235 data: 0.0002 max mem: 52150 +Epoch: [23] [1380/7269] eta: 0:42:42 lr: 0.000005 loss: 0.5192 (0.5188) time: 0.4772 data: 0.0002 max mem: 52150 +Epoch: [23] [1390/7269] eta: 0:42:35 lr: 0.000005 loss: 0.4968 (0.5185) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [23] [1400/7269] eta: 0:42:38 lr: 0.000005 loss: 0.5097 (0.5186) time: 0.4954 data: 0.0002 max mem: 52150 +Epoch: [23] [1410/7269] eta: 0:42:35 lr: 0.000005 loss: 0.5333 (0.5187) time: 0.5412 data: 0.0002 max mem: 52150 +Epoch: [23] [1420/7269] eta: 0:42:33 lr: 0.000005 loss: 0.5221 (0.5186) time: 0.4777 data: 0.0002 max mem: 52150 +Epoch: [23] [1430/7269] eta: 0:42:26 lr: 0.000005 loss: 0.5119 (0.5186) time: 0.4323 data: 0.0002 max mem: 52150 +Epoch: [23] [1440/7269] eta: 0:42:19 lr: 0.000005 loss: 0.5032 (0.5184) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [23] [1450/7269] eta: 0:42:15 lr: 0.000005 loss: 0.5161 (0.5184) time: 0.4170 data: 0.0002 max mem: 52150 +Epoch: [23] [1460/7269] eta: 0:42:08 lr: 0.000005 loss: 0.5358 (0.5182) time: 0.4165 data: 0.0002 max mem: 52150 +Epoch: [23] [1470/7269] eta: 0:42:02 lr: 0.000005 loss: 0.5167 (0.5183) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [23] [1480/7269] eta: 0:41:55 lr: 0.000005 loss: 0.5565 (0.5185) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [23] [1490/7269] eta: 0:41:48 lr: 0.000005 loss: 0.5524 (0.5184) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [23] [1500/7269] eta: 0:41:45 lr: 0.000005 loss: 0.5045 (0.5184) time: 0.4181 data: 0.0002 max mem: 52150 +Epoch: [23] [1510/7269] eta: 0:41:42 lr: 0.000005 loss: 0.5050 (0.5183) time: 0.4690 data: 0.0002 max mem: 52150 +Epoch: [23] [1520/7269] eta: 0:41:35 lr: 0.000005 loss: 0.5125 (0.5182) time: 0.4225 data: 0.0002 max mem: 52150 +Epoch: [23] [1530/7269] eta: 0:41:29 lr: 0.000005 loss: 0.5032 (0.5181) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [23] [1540/7269] eta: 0:41:22 lr: 0.000005 loss: 0.4748 (0.5178) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [23] [1550/7269] eta: 0:41:19 lr: 0.000005 loss: 0.4595 (0.5177) time: 0.4220 data: 0.0002 max mem: 52150 +Epoch: [23] [1560/7269] eta: 0:41:12 lr: 0.000005 loss: 0.4850 (0.5175) time: 0.4218 data: 0.0002 max mem: 52150 +Epoch: [23] [1570/7269] eta: 0:41:06 lr: 0.000005 loss: 0.5207 (0.5175) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [23] [1580/7269] eta: 0:41:06 lr: 0.000005 loss: 0.5075 (0.5174) time: 0.4617 data: 0.0002 max mem: 52150 +Epoch: [23] [1590/7269] eta: 0:40:59 lr: 0.000005 loss: 0.4823 (0.5173) time: 0.4623 data: 0.0002 max mem: 52150 +Epoch: [23] [1600/7269] eta: 0:40:53 lr: 0.000005 loss: 0.5324 (0.5175) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [23] [1610/7269] eta: 0:40:54 lr: 0.000005 loss: 0.5438 (0.5174) time: 0.4812 data: 0.0002 max mem: 52150 +Epoch: [23] [1620/7269] eta: 0:40:48 lr: 0.000005 loss: 0.5168 (0.5174) time: 0.4817 data: 0.0002 max mem: 52150 +Epoch: [23] [1630/7269] eta: 0:40:44 lr: 0.000005 loss: 0.4914 (0.5173) time: 0.4198 data: 0.0002 max mem: 52150 +Epoch: [23] [1640/7269] eta: 0:40:41 lr: 0.000005 loss: 0.5053 (0.5172) time: 0.4653 data: 0.0002 max mem: 52150 +Epoch: [23] [1650/7269] eta: 0:40:38 lr: 0.000005 loss: 0.5518 (0.5174) time: 0.4675 data: 0.0002 max mem: 52150 +Epoch: [23] [1660/7269] eta: 0:40:32 lr: 0.000005 loss: 0.5518 (0.5176) time: 0.4214 data: 0.0002 max mem: 52150 +Epoch: [23] [1670/7269] eta: 0:40:25 lr: 0.000005 loss: 0.5383 (0.5176) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [23] [1680/7269] eta: 0:40:19 lr: 0.000005 loss: 0.5418 (0.5176) time: 0.3711 data: 0.0002 max mem: 52150 +Epoch: [23] [1690/7269] eta: 0:40:12 lr: 0.000005 loss: 0.5296 (0.5175) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [23] [1700/7269] eta: 0:40:06 lr: 0.000005 loss: 0.4860 (0.5172) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [23] [1710/7269] eta: 0:40:03 lr: 0.000005 loss: 0.5348 (0.5173) time: 0.4220 data: 0.0002 max mem: 52150 +Epoch: [23] [1720/7269] eta: 0:39:57 lr: 0.000005 loss: 0.5581 (0.5175) time: 0.4227 data: 0.0002 max mem: 52150 +Epoch: [23] [1730/7269] eta: 0:39:51 lr: 0.000005 loss: 0.5366 (0.5174) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [23] [1740/7269] eta: 0:39:47 lr: 0.000005 loss: 0.5148 (0.5174) time: 0.4203 data: 0.0002 max mem: 52150 +Epoch: [23] [1750/7269] eta: 0:39:41 lr: 0.000005 loss: 0.5291 (0.5175) time: 0.4200 data: 0.0002 max mem: 52150 +Epoch: [23] [1760/7269] eta: 0:39:38 lr: 0.000005 loss: 0.5065 (0.5172) time: 0.4173 data: 0.0002 max mem: 52150 +Epoch: [23] [1770/7269] eta: 0:39:32 lr: 0.000005 loss: 0.4891 (0.5172) time: 0.4177 data: 0.0002 max mem: 52150 +Epoch: [23] [1780/7269] eta: 0:39:26 lr: 0.000005 loss: 0.4845 (0.5170) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [23] [1790/7269] eta: 0:39:22 lr: 0.000005 loss: 0.4884 (0.5172) time: 0.4159 data: 0.0002 max mem: 52150 +Epoch: [23] [1800/7269] eta: 0:39:16 lr: 0.000005 loss: 0.5457 (0.5172) time: 0.4164 data: 0.0002 max mem: 52150 +Epoch: [23] [1810/7269] eta: 0:39:14 lr: 0.000005 loss: 0.5278 (0.5171) time: 0.4343 data: 0.0002 max mem: 52150 +Epoch: [23] [1820/7269] eta: 0:39:08 lr: 0.000005 loss: 0.5436 (0.5174) time: 0.4342 data: 0.0002 max mem: 52150 +Epoch: [23] [1830/7269] eta: 0:39:02 lr: 0.000005 loss: 0.5436 (0.5174) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [23] [1840/7269] eta: 0:38:55 lr: 0.000005 loss: 0.4996 (0.5173) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [23] [1850/7269] eta: 0:38:56 lr: 0.000005 loss: 0.5557 (0.5176) time: 0.4792 data: 0.0002 max mem: 52150 +Epoch: [23] [1860/7269] eta: 0:38:57 lr: 0.000005 loss: 0.5563 (0.5175) time: 0.5993 data: 0.0002 max mem: 52150 +Epoch: [23] [1870/7269] eta: 0:38:51 lr: 0.000005 loss: 0.5348 (0.5175) time: 0.4917 data: 0.0002 max mem: 52150 +Epoch: [23] [1880/7269] eta: 0:38:45 lr: 0.000005 loss: 0.5244 (0.5175) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [23] [1890/7269] eta: 0:38:39 lr: 0.000005 loss: 0.5257 (0.5176) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [23] [1900/7269] eta: 0:38:33 lr: 0.000005 loss: 0.5510 (0.5178) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [23] [1910/7269] eta: 0:38:27 lr: 0.000005 loss: 0.5510 (0.5178) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [23] [1920/7269] eta: 0:38:23 lr: 0.000005 loss: 0.5000 (0.5176) time: 0.4201 data: 0.0002 max mem: 52150 +Epoch: [23] [1930/7269] eta: 0:38:21 lr: 0.000005 loss: 0.4977 (0.5177) time: 0.4802 data: 0.0002 max mem: 52150 +Epoch: [23] [1940/7269] eta: 0:38:21 lr: 0.000005 loss: 0.5327 (0.5178) time: 0.5367 data: 0.0002 max mem: 52150 +Epoch: [23] [1950/7269] eta: 0:38:15 lr: 0.000005 loss: 0.5483 (0.5179) time: 0.4762 data: 0.0002 max mem: 52150 +Epoch: [23] [1960/7269] eta: 0:38:15 lr: 0.000005 loss: 0.5494 (0.5180) time: 0.4797 data: 0.0002 max mem: 52150 +Epoch: [23] [1970/7269] eta: 0:38:09 lr: 0.000005 loss: 0.5366 (0.5181) time: 0.4794 data: 0.0002 max mem: 52150 +Epoch: [23] [1980/7269] eta: 0:38:06 lr: 0.000005 loss: 0.5649 (0.5183) time: 0.4248 data: 0.0002 max mem: 52150 +Epoch: [23] [1990/7269] eta: 0:38:00 lr: 0.000005 loss: 0.5483 (0.5183) time: 0.4245 data: 0.0002 max mem: 52150 +Epoch: [23] [2000/7269] eta: 0:37:54 lr: 0.000005 loss: 0.5106 (0.5183) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [23] [2010/7269] eta: 0:37:48 lr: 0.000005 loss: 0.4938 (0.5183) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [23] [2020/7269] eta: 0:37:48 lr: 0.000005 loss: 0.5035 (0.5181) time: 0.4953 data: 0.0002 max mem: 52150 +Epoch: [23] [2030/7269] eta: 0:37:42 lr: 0.000005 loss: 0.5128 (0.5182) time: 0.4955 data: 0.0002 max mem: 52150 +Epoch: [23] [2040/7269] eta: 0:37:37 lr: 0.000005 loss: 0.5599 (0.5184) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [23] [2050/7269] eta: 0:37:33 lr: 0.000005 loss: 0.5374 (0.5183) time: 0.4174 data: 0.0002 max mem: 52150 +Epoch: [23] [2060/7269] eta: 0:37:27 lr: 0.000005 loss: 0.5405 (0.5185) time: 0.4175 data: 0.0002 max mem: 52150 +Epoch: [23] [2070/7269] eta: 0:37:24 lr: 0.000005 loss: 0.5389 (0.5184) time: 0.4172 data: 0.0002 max mem: 52150 +Epoch: [23] [2080/7269] eta: 0:37:18 lr: 0.000005 loss: 0.4972 (0.5185) time: 0.4170 data: 0.0002 max mem: 52150 +Epoch: [23] [2090/7269] eta: 0:37:12 lr: 0.000005 loss: 0.5216 (0.5185) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [23] [2100/7269] eta: 0:37:06 lr: 0.000005 loss: 0.4983 (0.5184) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [23] [2110/7269] eta: 0:37:03 lr: 0.000005 loss: 0.5241 (0.5184) time: 0.4175 data: 0.0002 max mem: 52150 +Epoch: [23] [2120/7269] eta: 0:36:57 lr: 0.000005 loss: 0.5458 (0.5185) time: 0.4172 data: 0.0002 max mem: 52150 +Epoch: [23] [2130/7269] eta: 0:36:54 lr: 0.000005 loss: 0.5248 (0.5183) time: 0.4352 data: 0.0002 max mem: 52150 +Epoch: [23] [2140/7269] eta: 0:36:51 lr: 0.000005 loss: 0.5250 (0.5184) time: 0.4833 data: 0.0002 max mem: 52150 +Epoch: [23] [2150/7269] eta: 0:36:47 lr: 0.000005 loss: 0.5493 (0.5184) time: 0.4655 data: 0.0002 max mem: 52150 +Epoch: [23] [2160/7269] eta: 0:36:42 lr: 0.000005 loss: 0.5296 (0.5184) time: 0.4177 data: 0.0002 max mem: 52150 +Epoch: [23] [2170/7269] eta: 0:36:41 lr: 0.000005 loss: 0.5182 (0.5184) time: 0.4853 data: 0.0002 max mem: 52150 +Epoch: [23] [2180/7269] eta: 0:36:36 lr: 0.000005 loss: 0.5182 (0.5184) time: 0.4855 data: 0.0002 max mem: 52150 +Epoch: [23] [2190/7269] eta: 0:36:30 lr: 0.000005 loss: 0.5325 (0.5184) time: 0.3732 data: 0.0002 max mem: 52150 +Epoch: [23] [2200/7269] eta: 0:36:26 lr: 0.000005 loss: 0.5315 (0.5184) time: 0.4187 data: 0.0002 max mem: 52150 +Epoch: [23] [2210/7269] eta: 0:36:23 lr: 0.000005 loss: 0.5315 (0.5184) time: 0.4670 data: 0.0002 max mem: 52150 +Epoch: [23] [2220/7269] eta: 0:36:19 lr: 0.000005 loss: 0.5474 (0.5186) time: 0.4666 data: 0.0002 max mem: 52150 +Epoch: [23] [2230/7269] eta: 0:36:14 lr: 0.000005 loss: 0.5025 (0.5184) time: 0.4176 data: 0.0002 max mem: 52150 +Epoch: [23] [2240/7269] eta: 0:36:08 lr: 0.000005 loss: 0.4848 (0.5184) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [23] [2250/7269] eta: 0:36:02 lr: 0.000005 loss: 0.5172 (0.5185) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [23] [2260/7269] eta: 0:35:59 lr: 0.000005 loss: 0.5229 (0.5184) time: 0.4178 data: 0.0002 max mem: 52150 +Epoch: [23] [2270/7269] eta: 0:35:55 lr: 0.000005 loss: 0.5073 (0.5184) time: 0.4667 data: 0.0002 max mem: 52150 +Epoch: [23] [2280/7269] eta: 0:35:50 lr: 0.000005 loss: 0.5466 (0.5187) time: 0.4208 data: 0.0002 max mem: 52150 +Epoch: [23] [2290/7269] eta: 0:35:44 lr: 0.000005 loss: 0.5586 (0.5186) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [23] [2300/7269] eta: 0:35:39 lr: 0.000005 loss: 0.5406 (0.5186) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [23] [2310/7269] eta: 0:35:33 lr: 0.000005 loss: 0.5341 (0.5186) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [23] [2320/7269] eta: 0:35:28 lr: 0.000005 loss: 0.5341 (0.5187) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [23] [2330/7269] eta: 0:35:22 lr: 0.000005 loss: 0.5145 (0.5184) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [23] [2340/7269] eta: 0:35:16 lr: 0.000005 loss: 0.4672 (0.5184) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [23] [2350/7269] eta: 0:35:11 lr: 0.000005 loss: 0.4891 (0.5183) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [23] [2360/7269] eta: 0:35:07 lr: 0.000005 loss: 0.5013 (0.5184) time: 0.4203 data: 0.0002 max mem: 52150 +Epoch: [23] [2370/7269] eta: 0:35:02 lr: 0.000005 loss: 0.5268 (0.5184) time: 0.4199 data: 0.0002 max mem: 52150 +Epoch: [23] [2380/7269] eta: 0:35:01 lr: 0.000005 loss: 0.5293 (0.5185) time: 0.4829 data: 0.0002 max mem: 52150 +Epoch: [23] [2390/7269] eta: 0:34:58 lr: 0.000005 loss: 0.5382 (0.5186) time: 0.5432 data: 0.0002 max mem: 52150 +Epoch: [23] [2400/7269] eta: 0:34:55 lr: 0.000005 loss: 0.5285 (0.5187) time: 0.4920 data: 0.0002 max mem: 52150 +Epoch: [23] [2410/7269] eta: 0:34:51 lr: 0.000005 loss: 0.5285 (0.5188) time: 0.4787 data: 0.0002 max mem: 52150 +Epoch: [23] [2420/7269] eta: 0:34:48 lr: 0.000005 loss: 0.4975 (0.5186) time: 0.4785 data: 0.0002 max mem: 52150 +Epoch: [23] [2430/7269] eta: 0:34:43 lr: 0.000005 loss: 0.4906 (0.5186) time: 0.4312 data: 0.0002 max mem: 52150 +Epoch: [23] [2440/7269] eta: 0:34:37 lr: 0.000005 loss: 0.5626 (0.5187) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [23] [2450/7269] eta: 0:34:34 lr: 0.000005 loss: 0.5453 (0.5187) time: 0.4215 data: 0.0002 max mem: 52150 +Epoch: [23] [2460/7269] eta: 0:34:28 lr: 0.000005 loss: 0.5187 (0.5187) time: 0.4220 data: 0.0002 max mem: 52150 +Epoch: [23] [2470/7269] eta: 0:34:23 lr: 0.000005 loss: 0.5262 (0.5188) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [23] [2480/7269] eta: 0:34:20 lr: 0.000005 loss: 0.5192 (0.5187) time: 0.4321 data: 0.0002 max mem: 52150 +Epoch: [23] [2490/7269] eta: 0:34:15 lr: 0.000005 loss: 0.5089 (0.5188) time: 0.4319 data: 0.0002 max mem: 52150 +Epoch: [23] [2500/7269] eta: 0:34:11 lr: 0.000005 loss: 0.5165 (0.5188) time: 0.4219 data: 0.0002 max mem: 52150 +Epoch: [23] [2510/7269] eta: 0:34:06 lr: 0.000005 loss: 0.5197 (0.5188) time: 0.4224 data: 0.0002 max mem: 52150 +Epoch: [23] [2520/7269] eta: 0:34:00 lr: 0.000005 loss: 0.5426 (0.5188) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [23] [2530/7269] eta: 0:33:55 lr: 0.000005 loss: 0.5282 (0.5189) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [23] [2540/7269] eta: 0:33:49 lr: 0.000005 loss: 0.5248 (0.5188) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [23] [2550/7269] eta: 0:33:44 lr: 0.000005 loss: 0.5129 (0.5188) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [23] [2560/7269] eta: 0:33:39 lr: 0.000005 loss: 0.5129 (0.5188) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [23] [2570/7269] eta: 0:33:33 lr: 0.000005 loss: 0.5159 (0.5188) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [23] [2580/7269] eta: 0:33:31 lr: 0.000005 loss: 0.5637 (0.5189) time: 0.4635 data: 0.0002 max mem: 52150 +Epoch: [23] [2590/7269] eta: 0:33:30 lr: 0.000005 loss: 0.5685 (0.5191) time: 0.5728 data: 0.0002 max mem: 52150 +Epoch: [23] [2600/7269] eta: 0:33:25 lr: 0.000005 loss: 0.5756 (0.5193) time: 0.4815 data: 0.0002 max mem: 52150 +Epoch: [23] [2610/7269] eta: 0:33:25 lr: 0.000005 loss: 0.5328 (0.5192) time: 0.5238 data: 0.0002 max mem: 52150 +Epoch: [23] [2620/7269] eta: 0:33:21 lr: 0.000005 loss: 0.4873 (0.5191) time: 0.5703 data: 0.0002 max mem: 52150 +Epoch: [23] [2630/7269] eta: 0:33:18 lr: 0.000005 loss: 0.4798 (0.5190) time: 0.4814 data: 0.0002 max mem: 52150 +Epoch: [23] [2640/7269] eta: 0:33:13 lr: 0.000005 loss: 0.5078 (0.5189) time: 0.4349 data: 0.0002 max mem: 52150 +Epoch: [23] [2650/7269] eta: 0:33:07 lr: 0.000005 loss: 0.5300 (0.5191) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [23] [2660/7269] eta: 0:33:02 lr: 0.000005 loss: 0.5300 (0.5190) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [23] [2670/7269] eta: 0:32:58 lr: 0.000005 loss: 0.5341 (0.5192) time: 0.4185 data: 0.0002 max mem: 52150 +Epoch: [23] [2680/7269] eta: 0:32:55 lr: 0.000005 loss: 0.5697 (0.5193) time: 0.4828 data: 0.0002 max mem: 52150 +Epoch: [23] [2690/7269] eta: 0:32:52 lr: 0.000005 loss: 0.5493 (0.5194) time: 0.4822 data: 0.0001 max mem: 52150 +Epoch: [23] [2700/7269] eta: 0:32:48 lr: 0.000005 loss: 0.5433 (0.5194) time: 0.4791 data: 0.0001 max mem: 52150 +Epoch: [23] [2710/7269] eta: 0:32:43 lr: 0.000005 loss: 0.5440 (0.5194) time: 0.4337 data: 0.0001 max mem: 52150 +Epoch: [23] [2720/7269] eta: 0:32:41 lr: 0.000005 loss: 0.5292 (0.5194) time: 0.4702 data: 0.0002 max mem: 52150 +Epoch: [23] [2730/7269] eta: 0:32:39 lr: 0.000005 loss: 0.5362 (0.5195) time: 0.5618 data: 0.0002 max mem: 52150 +Epoch: [23] [2740/7269] eta: 0:32:33 lr: 0.000005 loss: 0.5125 (0.5193) time: 0.4638 data: 0.0002 max mem: 52150 +Epoch: [23] [2750/7269] eta: 0:32:28 lr: 0.000005 loss: 0.4982 (0.5193) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [23] [2760/7269] eta: 0:32:23 lr: 0.000005 loss: 0.5356 (0.5193) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [23] [2770/7269] eta: 0:32:19 lr: 0.000005 loss: 0.5466 (0.5194) time: 0.4221 data: 0.0002 max mem: 52150 +Epoch: [23] [2780/7269] eta: 0:32:16 lr: 0.000005 loss: 0.5372 (0.5194) time: 0.4856 data: 0.0002 max mem: 52150 +Epoch: [23] [2790/7269] eta: 0:32:11 lr: 0.000005 loss: 0.5041 (0.5193) time: 0.4354 data: 0.0002 max mem: 52150 +Epoch: [23] [2800/7269] eta: 0:32:06 lr: 0.000005 loss: 0.5324 (0.5194) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [23] [2810/7269] eta: 0:32:02 lr: 0.000005 loss: 0.5324 (0.5194) time: 0.4204 data: 0.0002 max mem: 52150 +Epoch: [23] [2820/7269] eta: 0:31:57 lr: 0.000005 loss: 0.5014 (0.5194) time: 0.4201 data: 0.0002 max mem: 52150 +Epoch: [23] [2830/7269] eta: 0:31:53 lr: 0.000005 loss: 0.5504 (0.5194) time: 0.4359 data: 0.0002 max mem: 52150 +Epoch: [23] [2840/7269] eta: 0:31:50 lr: 0.000005 loss: 0.5562 (0.5196) time: 0.4846 data: 0.0002 max mem: 52150 +Epoch: [23] [2850/7269] eta: 0:31:44 lr: 0.000005 loss: 0.5314 (0.5196) time: 0.4206 data: 0.0002 max mem: 52150 +Epoch: [23] [2860/7269] eta: 0:31:39 lr: 0.000005 loss: 0.4979 (0.5195) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [23] [2870/7269] eta: 0:31:34 lr: 0.000005 loss: 0.4554 (0.5193) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [23] [2880/7269] eta: 0:31:29 lr: 0.000005 loss: 0.5279 (0.5195) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [23] [2890/7269] eta: 0:31:27 lr: 0.000005 loss: 0.5651 (0.5197) time: 0.4793 data: 0.0002 max mem: 52150 +Epoch: [23] [2900/7269] eta: 0:31:22 lr: 0.000005 loss: 0.5264 (0.5196) time: 0.4789 data: 0.0002 max mem: 52150 +Epoch: [23] [2910/7269] eta: 0:31:16 lr: 0.000005 loss: 0.4809 (0.5196) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [23] [2920/7269] eta: 0:31:13 lr: 0.000005 loss: 0.4809 (0.5195) time: 0.4206 data: 0.0002 max mem: 52150 +Epoch: [23] [2930/7269] eta: 0:31:08 lr: 0.000005 loss: 0.4976 (0.5194) time: 0.4215 data: 0.0002 max mem: 52150 +Epoch: [23] [2940/7269] eta: 0:31:02 lr: 0.000005 loss: 0.5019 (0.5193) time: 0.3738 data: 0.0002 max mem: 52150 +Epoch: [23] [2950/7269] eta: 0:30:57 lr: 0.000005 loss: 0.4921 (0.5193) time: 0.3729 data: 0.0002 max mem: 52150 +Epoch: [23] [2960/7269] eta: 0:30:54 lr: 0.000005 loss: 0.4921 (0.5192) time: 0.4323 data: 0.0002 max mem: 52150 +Epoch: [23] [2970/7269] eta: 0:30:49 lr: 0.000005 loss: 0.5550 (0.5194) time: 0.4324 data: 0.0002 max mem: 52150 +Epoch: [23] [2980/7269] eta: 0:30:44 lr: 0.000005 loss: 0.5677 (0.5194) time: 0.3729 data: 0.0002 max mem: 52150 +Epoch: [23] [2990/7269] eta: 0:30:40 lr: 0.000005 loss: 0.5272 (0.5193) time: 0.4200 data: 0.0002 max mem: 52150 +Epoch: [23] [3000/7269] eta: 0:30:35 lr: 0.000005 loss: 0.5272 (0.5194) time: 0.4195 data: 0.0002 max mem: 52150 +Epoch: [23] [3010/7269] eta: 0:30:31 lr: 0.000005 loss: 0.5201 (0.5194) time: 0.4315 data: 0.0002 max mem: 52150 +Epoch: [23] [3020/7269] eta: 0:30:31 lr: 0.000005 loss: 0.5058 (0.5192) time: 0.5897 data: 0.0002 max mem: 52150 +Epoch: [23] [3030/7269] eta: 0:30:25 lr: 0.000005 loss: 0.5236 (0.5193) time: 0.5305 data: 0.0002 max mem: 52150 +Epoch: [23] [3040/7269] eta: 0:30:20 lr: 0.000005 loss: 0.5198 (0.5191) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [23] [3050/7269] eta: 0:30:17 lr: 0.000005 loss: 0.4720 (0.5191) time: 0.4200 data: 0.0002 max mem: 52150 +Epoch: [23] [3060/7269] eta: 0:30:13 lr: 0.000005 loss: 0.5270 (0.5191) time: 0.4846 data: 0.0002 max mem: 52150 +Epoch: [23] [3070/7269] eta: 0:30:09 lr: 0.000005 loss: 0.5625 (0.5192) time: 0.4835 data: 0.0002 max mem: 52150 +Epoch: [23] [3080/7269] eta: 0:30:06 lr: 0.000005 loss: 0.5364 (0.5192) time: 0.4650 data: 0.0002 max mem: 52150 +Epoch: [23] [3090/7269] eta: 0:30:02 lr: 0.000005 loss: 0.5258 (0.5193) time: 0.4641 data: 0.0002 max mem: 52150 +Epoch: [23] [3100/7269] eta: 0:29:58 lr: 0.000005 loss: 0.5277 (0.5193) time: 0.4635 data: 0.0002 max mem: 52150 +Epoch: [23] [3110/7269] eta: 0:29:53 lr: 0.000005 loss: 0.5479 (0.5194) time: 0.4177 data: 0.0002 max mem: 52150 +Epoch: [23] [3120/7269] eta: 0:29:48 lr: 0.000005 loss: 0.5461 (0.5194) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [23] [3130/7269] eta: 0:29:44 lr: 0.000005 loss: 0.5152 (0.5194) time: 0.4163 data: 0.0002 max mem: 52150 +Epoch: [23] [3140/7269] eta: 0:29:39 lr: 0.000005 loss: 0.5119 (0.5193) time: 0.4168 data: 0.0002 max mem: 52150 +Epoch: [23] [3150/7269] eta: 0:29:33 lr: 0.000005 loss: 0.5294 (0.5194) time: 0.3722 data: 0.0001 max mem: 52150 +Epoch: [23] [3160/7269] eta: 0:29:28 lr: 0.000005 loss: 0.5294 (0.5194) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [23] [3170/7269] eta: 0:29:27 lr: 0.000005 loss: 0.5263 (0.5194) time: 0.4959 data: 0.0002 max mem: 52150 +Epoch: [23] [3180/7269] eta: 0:29:21 lr: 0.000005 loss: 0.5166 (0.5194) time: 0.4957 data: 0.0002 max mem: 52150 +Epoch: [23] [3190/7269] eta: 0:29:16 lr: 0.000005 loss: 0.5174 (0.5194) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [23] [3200/7269] eta: 0:29:13 lr: 0.000005 loss: 0.5383 (0.5195) time: 0.4191 data: 0.0002 max mem: 52150 +Epoch: [23] [3210/7269] eta: 0:29:09 lr: 0.000005 loss: 0.5416 (0.5194) time: 0.4676 data: 0.0002 max mem: 52150 +Epoch: [23] [3220/7269] eta: 0:29:05 lr: 0.000005 loss: 0.5166 (0.5194) time: 0.4817 data: 0.0002 max mem: 52150 +Epoch: [23] [3230/7269] eta: 0:29:00 lr: 0.000005 loss: 0.5166 (0.5194) time: 0.4329 data: 0.0002 max mem: 52150 +Epoch: [23] [3240/7269] eta: 0:28:55 lr: 0.000005 loss: 0.5156 (0.5193) time: 0.3724 data: 0.0001 max mem: 52150 +Epoch: [23] [3250/7269] eta: 0:28:50 lr: 0.000005 loss: 0.5252 (0.5193) time: 0.3722 data: 0.0001 max mem: 52150 +Epoch: [23] [3260/7269] eta: 0:28:47 lr: 0.000005 loss: 0.5212 (0.5193) time: 0.4340 data: 0.0002 max mem: 52150 +Epoch: [23] [3270/7269] eta: 0:28:42 lr: 0.000005 loss: 0.4960 (0.5192) time: 0.4343 data: 0.0002 max mem: 52150 +Epoch: [23] [3280/7269] eta: 0:28:39 lr: 0.000005 loss: 0.5056 (0.5192) time: 0.4818 data: 0.0002 max mem: 52150 +Epoch: [23] [3290/7269] eta: 0:28:36 lr: 0.000005 loss: 0.5056 (0.5191) time: 0.5717 data: 0.0002 max mem: 52150 +Epoch: [23] [3300/7269] eta: 0:28:31 lr: 0.000005 loss: 0.4949 (0.5191) time: 0.4628 data: 0.0002 max mem: 52150 +Epoch: [23] [3310/7269] eta: 0:28:26 lr: 0.000005 loss: 0.5016 (0.5190) time: 0.3730 data: 0.0002 max mem: 52150 +Epoch: [23] [3320/7269] eta: 0:28:24 lr: 0.000005 loss: 0.5176 (0.5190) time: 0.4849 data: 0.0002 max mem: 52150 +Epoch: [23] [3330/7269] eta: 0:28:19 lr: 0.000005 loss: 0.5267 (0.5191) time: 0.4851 data: 0.0002 max mem: 52150 +Epoch: [23] [3340/7269] eta: 0:28:14 lr: 0.000005 loss: 0.5136 (0.5190) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [23] [3350/7269] eta: 0:28:09 lr: 0.000005 loss: 0.5407 (0.5191) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [23] [3360/7269] eta: 0:28:04 lr: 0.000005 loss: 0.5319 (0.5190) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [23] [3370/7269] eta: 0:27:59 lr: 0.000005 loss: 0.5312 (0.5191) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [23] [3380/7269] eta: 0:27:54 lr: 0.000005 loss: 0.5326 (0.5191) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [23] [3390/7269] eta: 0:27:50 lr: 0.000005 loss: 0.5406 (0.5191) time: 0.4185 data: 0.0002 max mem: 52150 +Epoch: [23] [3400/7269] eta: 0:27:48 lr: 0.000005 loss: 0.5110 (0.5190) time: 0.5245 data: 0.0002 max mem: 52150 +Epoch: [23] [3410/7269] eta: 0:27:43 lr: 0.000005 loss: 0.5110 (0.5190) time: 0.4777 data: 0.0002 max mem: 52150 +Epoch: [23] [3420/7269] eta: 0:27:41 lr: 0.000005 loss: 0.5452 (0.5191) time: 0.5154 data: 0.0002 max mem: 52150 +Epoch: [23] [3430/7269] eta: 0:27:36 lr: 0.000005 loss: 0.5306 (0.5191) time: 0.5155 data: 0.0002 max mem: 52150 +Epoch: [23] [3440/7269] eta: 0:27:32 lr: 0.000005 loss: 0.5312 (0.5191) time: 0.4318 data: 0.0002 max mem: 52150 +Epoch: [23] [3450/7269] eta: 0:27:27 lr: 0.000005 loss: 0.4827 (0.5189) time: 0.4319 data: 0.0002 max mem: 52150 +Epoch: [23] [3460/7269] eta: 0:27:23 lr: 0.000005 loss: 0.4836 (0.5189) time: 0.4229 data: 0.0002 max mem: 52150 +Epoch: [23] [3470/7269] eta: 0:27:20 lr: 0.000005 loss: 0.5069 (0.5189) time: 0.4839 data: 0.0002 max mem: 52150 +Epoch: [23] [3480/7269] eta: 0:27:15 lr: 0.000005 loss: 0.5061 (0.5189) time: 0.4332 data: 0.0002 max mem: 52150 +Epoch: [23] [3490/7269] eta: 0:27:10 lr: 0.000005 loss: 0.5093 (0.5189) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [23] [3500/7269] eta: 0:27:05 lr: 0.000005 loss: 0.4984 (0.5189) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [23] [3510/7269] eta: 0:27:00 lr: 0.000005 loss: 0.5278 (0.5191) time: 0.3729 data: 0.0002 max mem: 52150 +Epoch: [23] [3520/7269] eta: 0:26:56 lr: 0.000005 loss: 0.5413 (0.5191) time: 0.4218 data: 0.0002 max mem: 52150 +Epoch: [23] [3530/7269] eta: 0:26:51 lr: 0.000005 loss: 0.5211 (0.5192) time: 0.4211 data: 0.0002 max mem: 52150 +Epoch: [23] [3540/7269] eta: 0:26:47 lr: 0.000005 loss: 0.5211 (0.5192) time: 0.4194 data: 0.0002 max mem: 52150 +Epoch: [23] [3550/7269] eta: 0:26:43 lr: 0.000005 loss: 0.5326 (0.5193) time: 0.4805 data: 0.0002 max mem: 52150 +Epoch: [23] [3560/7269] eta: 0:26:39 lr: 0.000005 loss: 0.5326 (0.5192) time: 0.4331 data: 0.0002 max mem: 52150 +Epoch: [23] [3570/7269] eta: 0:26:35 lr: 0.000005 loss: 0.5117 (0.5192) time: 0.4175 data: 0.0002 max mem: 52150 +Epoch: [23] [3580/7269] eta: 0:26:30 lr: 0.000005 loss: 0.5444 (0.5193) time: 0.4179 data: 0.0002 max mem: 52150 +Epoch: [23] [3590/7269] eta: 0:26:25 lr: 0.000005 loss: 0.5392 (0.5193) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [23] [3600/7269] eta: 0:26:22 lr: 0.000005 loss: 0.5290 (0.5194) time: 0.4664 data: 0.0002 max mem: 52150 +Epoch: [23] [3610/7269] eta: 0:26:17 lr: 0.000005 loss: 0.5227 (0.5193) time: 0.4666 data: 0.0002 max mem: 52150 +Epoch: [23] [3620/7269] eta: 0:26:13 lr: 0.000005 loss: 0.4755 (0.5191) time: 0.4174 data: 0.0002 max mem: 52150 +Epoch: [23] [3630/7269] eta: 0:26:08 lr: 0.000005 loss: 0.4696 (0.5191) time: 0.4170 data: 0.0002 max mem: 52150 +Epoch: [23] [3640/7269] eta: 0:26:03 lr: 0.000005 loss: 0.5465 (0.5192) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [23] [3650/7269] eta: 0:25:58 lr: 0.000005 loss: 0.5265 (0.5191) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [23] [3660/7269] eta: 0:25:53 lr: 0.000005 loss: 0.5191 (0.5191) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [23] [3670/7269] eta: 0:25:51 lr: 0.000005 loss: 0.5279 (0.5192) time: 0.4953 data: 0.0002 max mem: 52150 +Epoch: [23] [3680/7269] eta: 0:25:46 lr: 0.000005 loss: 0.5208 (0.5190) time: 0.4954 data: 0.0002 max mem: 52150 +Epoch: [23] [3690/7269] eta: 0:25:41 lr: 0.000005 loss: 0.4819 (0.5190) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [23] [3700/7269] eta: 0:25:37 lr: 0.000005 loss: 0.5293 (0.5190) time: 0.4182 data: 0.0002 max mem: 52150 +Epoch: [23] [3710/7269] eta: 0:25:32 lr: 0.000005 loss: 0.5282 (0.5190) time: 0.4175 data: 0.0002 max mem: 52150 +Epoch: [23] [3720/7269] eta: 0:25:27 lr: 0.000005 loss: 0.4806 (0.5189) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [23] [3730/7269] eta: 0:25:24 lr: 0.000005 loss: 0.4914 (0.5189) time: 0.4356 data: 0.0002 max mem: 52150 +Epoch: [23] [3740/7269] eta: 0:25:20 lr: 0.000005 loss: 0.5288 (0.5190) time: 0.4978 data: 0.0002 max mem: 52150 +Epoch: [23] [3750/7269] eta: 0:25:16 lr: 0.000005 loss: 0.5243 (0.5189) time: 0.4803 data: 0.0002 max mem: 52150 +Epoch: [23] [3760/7269] eta: 0:25:11 lr: 0.000005 loss: 0.5076 (0.5189) time: 0.4171 data: 0.0002 max mem: 52150 +Epoch: [23] [3770/7269] eta: 0:25:06 lr: 0.000005 loss: 0.5508 (0.5191) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [23] [3780/7269] eta: 0:25:05 lr: 0.000005 loss: 0.5677 (0.5191) time: 0.5567 data: 0.0002 max mem: 52150 +Epoch: [23] [3790/7269] eta: 0:25:01 lr: 0.000005 loss: 0.5781 (0.5192) time: 0.6198 data: 0.0002 max mem: 52150 +Epoch: [23] [3800/7269] eta: 0:24:59 lr: 0.000005 loss: 0.5683 (0.5192) time: 0.5580 data: 0.0002 max mem: 52150 +Epoch: [23] [3810/7269] eta: 0:24:54 lr: 0.000005 loss: 0.4954 (0.5192) time: 0.4945 data: 0.0002 max mem: 52150 +Epoch: [23] [3820/7269] eta: 0:24:49 lr: 0.000005 loss: 0.5152 (0.5192) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [23] [3830/7269] eta: 0:24:44 lr: 0.000005 loss: 0.5075 (0.5191) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [23] [3840/7269] eta: 0:24:40 lr: 0.000005 loss: 0.5086 (0.5192) time: 0.4188 data: 0.0002 max mem: 52150 +Epoch: [23] [3850/7269] eta: 0:24:35 lr: 0.000005 loss: 0.5184 (0.5192) time: 0.4193 data: 0.0002 max mem: 52150 +Epoch: [23] [3860/7269] eta: 0:24:30 lr: 0.000005 loss: 0.5477 (0.5192) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [23] [3870/7269] eta: 0:24:26 lr: 0.000005 loss: 0.5527 (0.5192) time: 0.4190 data: 0.0002 max mem: 52150 +Epoch: [23] [3880/7269] eta: 0:24:22 lr: 0.000005 loss: 0.5316 (0.5192) time: 0.4787 data: 0.0002 max mem: 52150 +Epoch: [23] [3890/7269] eta: 0:24:18 lr: 0.000005 loss: 0.5213 (0.5192) time: 0.4811 data: 0.0002 max mem: 52150 +Epoch: [23] [3900/7269] eta: 0:24:14 lr: 0.000005 loss: 0.5577 (0.5193) time: 0.4213 data: 0.0002 max mem: 52150 +Epoch: [23] [3910/7269] eta: 0:24:11 lr: 0.000005 loss: 0.5629 (0.5193) time: 0.4807 data: 0.0002 max mem: 52150 +Epoch: [23] [3920/7269] eta: 0:24:06 lr: 0.000005 loss: 0.5153 (0.5193) time: 0.4810 data: 0.0002 max mem: 52150 +Epoch: [23] [3930/7269] eta: 0:24:02 lr: 0.000005 loss: 0.5100 (0.5192) time: 0.4208 data: 0.0002 max mem: 52150 +Epoch: [23] [3940/7269] eta: 0:23:57 lr: 0.000005 loss: 0.4867 (0.5192) time: 0.4204 data: 0.0002 max mem: 52150 +Epoch: [23] [3950/7269] eta: 0:23:52 lr: 0.000005 loss: 0.5362 (0.5192) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [23] [3960/7269] eta: 0:23:47 lr: 0.000005 loss: 0.5012 (0.5191) time: 0.3728 data: 0.0002 max mem: 52150 +Epoch: [23] [3970/7269] eta: 0:23:43 lr: 0.000005 loss: 0.5012 (0.5192) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [23] [3980/7269] eta: 0:23:38 lr: 0.000005 loss: 0.5093 (0.5191) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [23] [3990/7269] eta: 0:23:34 lr: 0.000005 loss: 0.4981 (0.5190) time: 0.4174 data: 0.0002 max mem: 52150 +Epoch: [23] [4000/7269] eta: 0:23:29 lr: 0.000005 loss: 0.5311 (0.5191) time: 0.4175 data: 0.0002 max mem: 52150 +Epoch: [23] [4010/7269] eta: 0:23:26 lr: 0.000005 loss: 0.5370 (0.5191) time: 0.4831 data: 0.0002 max mem: 52150 +Epoch: [23] [4020/7269] eta: 0:23:21 lr: 0.000005 loss: 0.5326 (0.5191) time: 0.4828 data: 0.0002 max mem: 52150 +Epoch: [23] [4030/7269] eta: 0:23:17 lr: 0.000005 loss: 0.5326 (0.5192) time: 0.4170 data: 0.0002 max mem: 52150 +Epoch: [23] [4040/7269] eta: 0:23:12 lr: 0.000005 loss: 0.5346 (0.5192) time: 0.4177 data: 0.0002 max mem: 52150 +Epoch: [23] [4050/7269] eta: 0:23:07 lr: 0.000005 loss: 0.5207 (0.5192) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [23] [4060/7269] eta: 0:23:03 lr: 0.000005 loss: 0.5207 (0.5191) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [23] [4070/7269] eta: 0:23:00 lr: 0.000005 loss: 0.5305 (0.5191) time: 0.4791 data: 0.0002 max mem: 52150 +Epoch: [23] [4080/7269] eta: 0:22:55 lr: 0.000005 loss: 0.5252 (0.5191) time: 0.4790 data: 0.0002 max mem: 52150 +Epoch: [23] [4090/7269] eta: 0:22:52 lr: 0.000005 loss: 0.5015 (0.5190) time: 0.4676 data: 0.0002 max mem: 52150 +Epoch: [23] [4100/7269] eta: 0:22:47 lr: 0.000005 loss: 0.5075 (0.5190) time: 0.4676 data: 0.0002 max mem: 52150 +Epoch: [23] [4110/7269] eta: 0:22:42 lr: 0.000005 loss: 0.4815 (0.5189) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [23] [4120/7269] eta: 0:22:37 lr: 0.000005 loss: 0.4794 (0.5189) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [23] [4130/7269] eta: 0:22:32 lr: 0.000005 loss: 0.4891 (0.5188) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [23] [4140/7269] eta: 0:22:28 lr: 0.000005 loss: 0.5304 (0.5189) time: 0.4172 data: 0.0002 max mem: 52150 +Epoch: [23] [4150/7269] eta: 0:22:24 lr: 0.000005 loss: 0.5411 (0.5189) time: 0.4177 data: 0.0002 max mem: 52150 +Epoch: [23] [4160/7269] eta: 0:22:19 lr: 0.000005 loss: 0.5300 (0.5189) time: 0.3728 data: 0.0002 max mem: 52150 +Epoch: [23] [4170/7269] eta: 0:22:15 lr: 0.000005 loss: 0.5339 (0.5189) time: 0.4318 data: 0.0002 max mem: 52150 +Epoch: [23] [4180/7269] eta: 0:22:10 lr: 0.000005 loss: 0.5437 (0.5190) time: 0.4315 data: 0.0002 max mem: 52150 +Epoch: [23] [4190/7269] eta: 0:22:06 lr: 0.000005 loss: 0.5338 (0.5189) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [23] [4200/7269] eta: 0:22:01 lr: 0.000005 loss: 0.4963 (0.5188) time: 0.3727 data: 0.0001 max mem: 52150 +Epoch: [23] [4210/7269] eta: 0:21:58 lr: 0.000005 loss: 0.5159 (0.5189) time: 0.4812 data: 0.0002 max mem: 52150 +Epoch: [23] [4220/7269] eta: 0:21:55 lr: 0.000005 loss: 0.5153 (0.5189) time: 0.5929 data: 0.0002 max mem: 52150 +Epoch: [23] [4230/7269] eta: 0:21:50 lr: 0.000005 loss: 0.5043 (0.5188) time: 0.4839 data: 0.0001 max mem: 52150 +Epoch: [23] [4240/7269] eta: 0:21:45 lr: 0.000005 loss: 0.4987 (0.5188) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [23] [4250/7269] eta: 0:21:40 lr: 0.000005 loss: 0.5023 (0.5189) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [23] [4260/7269] eta: 0:21:36 lr: 0.000005 loss: 0.4981 (0.5187) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [23] [4270/7269] eta: 0:21:32 lr: 0.000005 loss: 0.5019 (0.5187) time: 0.4173 data: 0.0002 max mem: 52150 +Epoch: [23] [4280/7269] eta: 0:21:28 lr: 0.000005 loss: 0.5203 (0.5187) time: 0.4773 data: 0.0002 max mem: 52150 +Epoch: [23] [4290/7269] eta: 0:21:24 lr: 0.000005 loss: 0.5199 (0.5187) time: 0.4935 data: 0.0002 max mem: 52150 +Epoch: [23] [4300/7269] eta: 0:21:20 lr: 0.000005 loss: 0.5393 (0.5188) time: 0.4844 data: 0.0002 max mem: 52150 +Epoch: [23] [4310/7269] eta: 0:21:16 lr: 0.000005 loss: 0.5544 (0.5188) time: 0.4833 data: 0.0002 max mem: 52150 +Epoch: [23] [4320/7269] eta: 0:21:11 lr: 0.000005 loss: 0.5328 (0.5188) time: 0.4324 data: 0.0002 max mem: 52150 +Epoch: [23] [4330/7269] eta: 0:21:07 lr: 0.000005 loss: 0.5094 (0.5187) time: 0.4186 data: 0.0002 max mem: 52150 +Epoch: [23] [4340/7269] eta: 0:21:03 lr: 0.000005 loss: 0.4780 (0.5187) time: 0.4825 data: 0.0002 max mem: 52150 +Epoch: [23] [4350/7269] eta: 0:20:59 lr: 0.000005 loss: 0.5277 (0.5187) time: 0.4847 data: 0.0002 max mem: 52150 +Epoch: [23] [4360/7269] eta: 0:20:54 lr: 0.000005 loss: 0.5277 (0.5187) time: 0.4212 data: 0.0002 max mem: 52150 +Epoch: [23] [4370/7269] eta: 0:20:50 lr: 0.000005 loss: 0.5010 (0.5186) time: 0.3731 data: 0.0001 max mem: 52150 +Epoch: [23] [4380/7269] eta: 0:20:45 lr: 0.000005 loss: 0.5107 (0.5185) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [23] [4390/7269] eta: 0:20:40 lr: 0.000005 loss: 0.5619 (0.5186) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [23] [4400/7269] eta: 0:20:36 lr: 0.000005 loss: 0.5619 (0.5186) time: 0.4323 data: 0.0002 max mem: 52150 +Epoch: [23] [4410/7269] eta: 0:20:32 lr: 0.000005 loss: 0.5127 (0.5187) time: 0.4322 data: 0.0002 max mem: 52150 +Epoch: [23] [4420/7269] eta: 0:20:28 lr: 0.000005 loss: 0.5103 (0.5187) time: 0.4355 data: 0.0002 max mem: 52150 +Epoch: [23] [4430/7269] eta: 0:20:24 lr: 0.000005 loss: 0.5020 (0.5186) time: 0.4854 data: 0.0002 max mem: 52150 +Epoch: [23] [4440/7269] eta: 0:20:19 lr: 0.000005 loss: 0.4629 (0.5186) time: 0.4213 data: 0.0002 max mem: 52150 +Epoch: [23] [4450/7269] eta: 0:20:14 lr: 0.000005 loss: 0.5038 (0.5185) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [23] [4460/7269] eta: 0:20:10 lr: 0.000005 loss: 0.5295 (0.5187) time: 0.4166 data: 0.0001 max mem: 52150 +Epoch: [23] [4470/7269] eta: 0:20:06 lr: 0.000005 loss: 0.5162 (0.5185) time: 0.4169 data: 0.0002 max mem: 52150 +Epoch: [23] [4480/7269] eta: 0:20:01 lr: 0.000005 loss: 0.5067 (0.5185) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [23] [4490/7269] eta: 0:19:57 lr: 0.000005 loss: 0.5332 (0.5185) time: 0.4171 data: 0.0002 max mem: 52150 +Epoch: [23] [4500/7269] eta: 0:19:53 lr: 0.000005 loss: 0.5205 (0.5185) time: 0.4621 data: 0.0002 max mem: 52150 +Epoch: [23] [4510/7269] eta: 0:19:49 lr: 0.000005 loss: 0.5086 (0.5185) time: 0.4618 data: 0.0002 max mem: 52150 +Epoch: [23] [4520/7269] eta: 0:19:45 lr: 0.000005 loss: 0.5419 (0.5186) time: 0.4778 data: 0.0002 max mem: 52150 +Epoch: [23] [4530/7269] eta: 0:19:40 lr: 0.000005 loss: 0.5421 (0.5185) time: 0.4327 data: 0.0002 max mem: 52150 +Epoch: [23] [4540/7269] eta: 0:19:36 lr: 0.000005 loss: 0.5223 (0.5186) time: 0.4319 data: 0.0002 max mem: 52150 +Epoch: [23] [4550/7269] eta: 0:19:31 lr: 0.000005 loss: 0.5223 (0.5186) time: 0.4323 data: 0.0002 max mem: 52150 +Epoch: [23] [4560/7269] eta: 0:19:27 lr: 0.000005 loss: 0.5299 (0.5187) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [23] [4570/7269] eta: 0:19:22 lr: 0.000005 loss: 0.5306 (0.5186) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [23] [4580/7269] eta: 0:19:18 lr: 0.000005 loss: 0.5306 (0.5187) time: 0.4185 data: 0.0002 max mem: 52150 +Epoch: [23] [4590/7269] eta: 0:19:15 lr: 0.000005 loss: 0.5270 (0.5187) time: 0.5288 data: 0.0002 max mem: 52150 +Epoch: [23] [4600/7269] eta: 0:19:11 lr: 0.000005 loss: 0.5082 (0.5186) time: 0.5326 data: 0.0001 max mem: 52150 +Epoch: [23] [4610/7269] eta: 0:19:06 lr: 0.000005 loss: 0.4971 (0.5186) time: 0.4222 data: 0.0002 max mem: 52150 +Epoch: [23] [4620/7269] eta: 0:19:01 lr: 0.000005 loss: 0.5232 (0.5186) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [23] [4630/7269] eta: 0:18:57 lr: 0.000005 loss: 0.5278 (0.5186) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [23] [4640/7269] eta: 0:18:52 lr: 0.000005 loss: 0.5484 (0.5186) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [23] [4650/7269] eta: 0:18:48 lr: 0.000005 loss: 0.5264 (0.5186) time: 0.4182 data: 0.0002 max mem: 52150 +Epoch: [23] [4660/7269] eta: 0:18:44 lr: 0.000005 loss: 0.5264 (0.5186) time: 0.4779 data: 0.0002 max mem: 52150 +Epoch: [23] [4670/7269] eta: 0:18:39 lr: 0.000005 loss: 0.5606 (0.5187) time: 0.4317 data: 0.0002 max mem: 52150 +Epoch: [23] [4680/7269] eta: 0:18:36 lr: 0.000005 loss: 0.5534 (0.5186) time: 0.5419 data: 0.0002 max mem: 52150 +Epoch: [23] [4690/7269] eta: 0:18:32 lr: 0.000005 loss: 0.4960 (0.5186) time: 0.5415 data: 0.0002 max mem: 52150 +Epoch: [23] [4700/7269] eta: 0:18:27 lr: 0.000005 loss: 0.4672 (0.5185) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [23] [4710/7269] eta: 0:18:23 lr: 0.000005 loss: 0.5433 (0.5186) time: 0.4215 data: 0.0002 max mem: 52150 +Epoch: [23] [4720/7269] eta: 0:18:18 lr: 0.000005 loss: 0.5509 (0.5186) time: 0.4210 data: 0.0002 max mem: 52150 +Epoch: [23] [4730/7269] eta: 0:18:14 lr: 0.000005 loss: 0.5183 (0.5186) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [23] [4740/7269] eta: 0:18:10 lr: 0.000005 loss: 0.5029 (0.5186) time: 0.4188 data: 0.0002 max mem: 52150 +Epoch: [23] [4750/7269] eta: 0:18:05 lr: 0.000005 loss: 0.5335 (0.5187) time: 0.4188 data: 0.0002 max mem: 52150 +Epoch: [23] [4760/7269] eta: 0:18:01 lr: 0.000005 loss: 0.5335 (0.5187) time: 0.4185 data: 0.0002 max mem: 52150 +Epoch: [23] [4770/7269] eta: 0:17:57 lr: 0.000005 loss: 0.5326 (0.5187) time: 0.4800 data: 0.0002 max mem: 52150 +Epoch: [23] [4780/7269] eta: 0:17:52 lr: 0.000005 loss: 0.5270 (0.5187) time: 0.4332 data: 0.0002 max mem: 52150 +Epoch: [23] [4790/7269] eta: 0:17:48 lr: 0.000005 loss: 0.5159 (0.5187) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [23] [4800/7269] eta: 0:17:44 lr: 0.000005 loss: 0.5159 (0.5187) time: 0.4188 data: 0.0002 max mem: 52150 +Epoch: [23] [4810/7269] eta: 0:17:40 lr: 0.000005 loss: 0.5243 (0.5187) time: 0.4674 data: 0.0002 max mem: 52150 +Epoch: [23] [4820/7269] eta: 0:17:35 lr: 0.000005 loss: 0.5381 (0.5188) time: 0.4207 data: 0.0002 max mem: 52150 +Epoch: [23] [4830/7269] eta: 0:17:30 lr: 0.000005 loss: 0.5655 (0.5188) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [23] [4840/7269] eta: 0:17:26 lr: 0.000005 loss: 0.5378 (0.5188) time: 0.4190 data: 0.0002 max mem: 52150 +Epoch: [23] [4850/7269] eta: 0:17:22 lr: 0.000005 loss: 0.5193 (0.5188) time: 0.4191 data: 0.0002 max mem: 52150 +Epoch: [23] [4860/7269] eta: 0:17:17 lr: 0.000005 loss: 0.5314 (0.5188) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [23] [4870/7269] eta: 0:17:12 lr: 0.000005 loss: 0.5286 (0.5188) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [23] [4880/7269] eta: 0:17:08 lr: 0.000005 loss: 0.5050 (0.5188) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [23] [4890/7269] eta: 0:17:04 lr: 0.000005 loss: 0.4915 (0.5188) time: 0.4316 data: 0.0002 max mem: 52150 +Epoch: [23] [4900/7269] eta: 0:17:00 lr: 0.000005 loss: 0.4960 (0.5187) time: 0.4770 data: 0.0002 max mem: 52150 +Epoch: [23] [4910/7269] eta: 0:16:55 lr: 0.000005 loss: 0.5160 (0.5188) time: 0.4170 data: 0.0002 max mem: 52150 +Epoch: [23] [4920/7269] eta: 0:16:51 lr: 0.000005 loss: 0.5294 (0.5187) time: 0.4338 data: 0.0002 max mem: 52150 +Epoch: [23] [4930/7269] eta: 0:16:46 lr: 0.000005 loss: 0.5218 (0.5188) time: 0.4340 data: 0.0002 max mem: 52150 +Epoch: [23] [4940/7269] eta: 0:16:43 lr: 0.000005 loss: 0.5218 (0.5188) time: 0.4614 data: 0.0002 max mem: 52150 +Epoch: [23] [4950/7269] eta: 0:16:38 lr: 0.000005 loss: 0.5279 (0.5188) time: 0.4616 data: 0.0002 max mem: 52150 +Epoch: [23] [4960/7269] eta: 0:16:34 lr: 0.000005 loss: 0.5292 (0.5188) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [23] [4970/7269] eta: 0:16:29 lr: 0.000005 loss: 0.5391 (0.5188) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [23] [4980/7269] eta: 0:16:24 lr: 0.000005 loss: 0.5391 (0.5189) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [23] [4990/7269] eta: 0:16:20 lr: 0.000005 loss: 0.5017 (0.5188) time: 0.4352 data: 0.0002 max mem: 52150 +Epoch: [23] [5000/7269] eta: 0:16:16 lr: 0.000005 loss: 0.4674 (0.5187) time: 0.4959 data: 0.0002 max mem: 52150 +Epoch: [23] [5010/7269] eta: 0:16:12 lr: 0.000005 loss: 0.5140 (0.5187) time: 0.4331 data: 0.0002 max mem: 52150 +Epoch: [23] [5020/7269] eta: 0:16:08 lr: 0.000005 loss: 0.5202 (0.5187) time: 0.4245 data: 0.0002 max mem: 52150 +Epoch: [23] [5030/7269] eta: 0:16:03 lr: 0.000005 loss: 0.5441 (0.5187) time: 0.4246 data: 0.0002 max mem: 52150 +Epoch: [23] [5040/7269] eta: 0:15:59 lr: 0.000005 loss: 0.5426 (0.5187) time: 0.4233 data: 0.0002 max mem: 52150 +Epoch: [23] [5050/7269] eta: 0:15:55 lr: 0.000005 loss: 0.4793 (0.5186) time: 0.4229 data: 0.0002 max mem: 52150 +Epoch: [23] [5060/7269] eta: 0:15:50 lr: 0.000005 loss: 0.5127 (0.5187) time: 0.4312 data: 0.0001 max mem: 52150 +Epoch: [23] [5070/7269] eta: 0:15:46 lr: 0.000005 loss: 0.5215 (0.5187) time: 0.4319 data: 0.0002 max mem: 52150 +Epoch: [23] [5080/7269] eta: 0:15:42 lr: 0.000005 loss: 0.5253 (0.5187) time: 0.4204 data: 0.0002 max mem: 52150 +Epoch: [23] [5090/7269] eta: 0:15:37 lr: 0.000005 loss: 0.5233 (0.5187) time: 0.4198 data: 0.0002 max mem: 52150 +Epoch: [23] [5100/7269] eta: 0:15:33 lr: 0.000005 loss: 0.4960 (0.5186) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [23] [5110/7269] eta: 0:15:28 lr: 0.000005 loss: 0.5291 (0.5186) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [23] [5120/7269] eta: 0:15:24 lr: 0.000005 loss: 0.5441 (0.5187) time: 0.4204 data: 0.0002 max mem: 52150 +Epoch: [23] [5130/7269] eta: 0:15:20 lr: 0.000005 loss: 0.5372 (0.5187) time: 0.5096 data: 0.0002 max mem: 52150 +Epoch: [23] [5140/7269] eta: 0:15:16 lr: 0.000005 loss: 0.5059 (0.5186) time: 0.5220 data: 0.0002 max mem: 52150 +Epoch: [23] [5150/7269] eta: 0:15:12 lr: 0.000005 loss: 0.5040 (0.5186) time: 0.4325 data: 0.0002 max mem: 52150 +Epoch: [23] [5160/7269] eta: 0:15:07 lr: 0.000005 loss: 0.5214 (0.5186) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [23] [5170/7269] eta: 0:15:03 lr: 0.000005 loss: 0.5527 (0.5187) time: 0.4649 data: 0.0002 max mem: 52150 +Epoch: [23] [5180/7269] eta: 0:14:59 lr: 0.000005 loss: 0.5642 (0.5188) time: 0.4648 data: 0.0002 max mem: 52150 +Epoch: [23] [5190/7269] eta: 0:14:55 lr: 0.000005 loss: 0.5661 (0.5188) time: 0.4205 data: 0.0002 max mem: 52150 +Epoch: [23] [5200/7269] eta: 0:14:50 lr: 0.000005 loss: 0.5572 (0.5188) time: 0.4203 data: 0.0002 max mem: 52150 +Epoch: [23] [5210/7269] eta: 0:14:46 lr: 0.000005 loss: 0.5427 (0.5188) time: 0.4322 data: 0.0002 max mem: 52150 +Epoch: [23] [5220/7269] eta: 0:14:42 lr: 0.000005 loss: 0.5194 (0.5188) time: 0.4773 data: 0.0002 max mem: 52150 +Epoch: [23] [5230/7269] eta: 0:14:38 lr: 0.000005 loss: 0.5121 (0.5188) time: 0.4635 data: 0.0002 max mem: 52150 +Epoch: [23] [5240/7269] eta: 0:14:33 lr: 0.000005 loss: 0.5239 (0.5189) time: 0.4661 data: 0.0002 max mem: 52150 +Epoch: [23] [5250/7269] eta: 0:14:29 lr: 0.000005 loss: 0.5091 (0.5189) time: 0.4801 data: 0.0002 max mem: 52150 +Epoch: [23] [5260/7269] eta: 0:14:25 lr: 0.000005 loss: 0.4743 (0.5188) time: 0.4324 data: 0.0002 max mem: 52150 +Epoch: [23] [5270/7269] eta: 0:14:20 lr: 0.000005 loss: 0.5637 (0.5189) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [23] [5280/7269] eta: 0:14:16 lr: 0.000005 loss: 0.5655 (0.5189) time: 0.4345 data: 0.0002 max mem: 52150 +Epoch: [23] [5290/7269] eta: 0:14:12 lr: 0.000005 loss: 0.5409 (0.5190) time: 0.4341 data: 0.0002 max mem: 52150 +Epoch: [23] [5300/7269] eta: 0:14:08 lr: 0.000005 loss: 0.5409 (0.5190) time: 0.4187 data: 0.0002 max mem: 52150 +Epoch: [23] [5310/7269] eta: 0:14:03 lr: 0.000005 loss: 0.5418 (0.5190) time: 0.4189 data: 0.0002 max mem: 52150 +Epoch: [23] [5320/7269] eta: 0:13:59 lr: 0.000005 loss: 0.5135 (0.5189) time: 0.4192 data: 0.0002 max mem: 52150 +Epoch: [23] [5330/7269] eta: 0:13:54 lr: 0.000005 loss: 0.5173 (0.5189) time: 0.4191 data: 0.0002 max mem: 52150 +Epoch: [23] [5340/7269] eta: 0:13:51 lr: 0.000005 loss: 0.5173 (0.5189) time: 0.4664 data: 0.0002 max mem: 52150 +Epoch: [23] [5350/7269] eta: 0:13:46 lr: 0.000005 loss: 0.5198 (0.5189) time: 0.4668 data: 0.0002 max mem: 52150 +Epoch: [23] [5360/7269] eta: 0:13:42 lr: 0.000005 loss: 0.5384 (0.5189) time: 0.4335 data: 0.0002 max mem: 52150 +Epoch: [23] [5370/7269] eta: 0:13:38 lr: 0.000005 loss: 0.5228 (0.5189) time: 0.5265 data: 0.0002 max mem: 52150 +Epoch: [23] [5380/7269] eta: 0:13:34 lr: 0.000005 loss: 0.4767 (0.5188) time: 0.4651 data: 0.0002 max mem: 52150 +Epoch: [23] [5390/7269] eta: 0:13:29 lr: 0.000005 loss: 0.5129 (0.5189) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [23] [5400/7269] eta: 0:13:25 lr: 0.000005 loss: 0.5501 (0.5189) time: 0.4345 data: 0.0002 max mem: 52150 +Epoch: [23] [5410/7269] eta: 0:13:20 lr: 0.000005 loss: 0.5429 (0.5189) time: 0.4340 data: 0.0002 max mem: 52150 +Epoch: [23] [5420/7269] eta: 0:13:16 lr: 0.000005 loss: 0.5710 (0.5190) time: 0.4320 data: 0.0002 max mem: 52150 +Epoch: [23] [5430/7269] eta: 0:13:12 lr: 0.000005 loss: 0.5298 (0.5190) time: 0.4322 data: 0.0002 max mem: 52150 +Epoch: [23] [5440/7269] eta: 0:13:07 lr: 0.000005 loss: 0.5079 (0.5190) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [23] [5450/7269] eta: 0:13:03 lr: 0.000005 loss: 0.5202 (0.5189) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [23] [5460/7269] eta: 0:12:58 lr: 0.000005 loss: 0.5211 (0.5189) time: 0.3730 data: 0.0002 max mem: 52150 +Epoch: [23] [5470/7269] eta: 0:12:55 lr: 0.000005 loss: 0.5227 (0.5189) time: 0.4848 data: 0.0002 max mem: 52150 +Epoch: [23] [5480/7269] eta: 0:12:50 lr: 0.000005 loss: 0.5250 (0.5189) time: 0.5329 data: 0.0002 max mem: 52150 +Epoch: [23] [5490/7269] eta: 0:12:46 lr: 0.000005 loss: 0.5082 (0.5189) time: 0.4811 data: 0.0002 max mem: 52150 +Epoch: [23] [5500/7269] eta: 0:12:42 lr: 0.000005 loss: 0.5201 (0.5189) time: 0.4327 data: 0.0002 max mem: 52150 +Epoch: [23] [5510/7269] eta: 0:12:38 lr: 0.000005 loss: 0.5138 (0.5189) time: 0.4215 data: 0.0002 max mem: 52150 +Epoch: [23] [5520/7269] eta: 0:12:33 lr: 0.000005 loss: 0.5343 (0.5190) time: 0.4212 data: 0.0002 max mem: 52150 +Epoch: [23] [5530/7269] eta: 0:12:29 lr: 0.000005 loss: 0.5280 (0.5190) time: 0.4167 data: 0.0002 max mem: 52150 +Epoch: [23] [5540/7269] eta: 0:12:24 lr: 0.000005 loss: 0.5270 (0.5190) time: 0.4171 data: 0.0002 max mem: 52150 +Epoch: [23] [5550/7269] eta: 0:12:20 lr: 0.000005 loss: 0.5509 (0.5190) time: 0.4210 data: 0.0002 max mem: 52150 +Epoch: [23] [5560/7269] eta: 0:12:16 lr: 0.000005 loss: 0.5356 (0.5189) time: 0.4208 data: 0.0002 max mem: 52150 +Epoch: [23] [5570/7269] eta: 0:12:11 lr: 0.000005 loss: 0.5066 (0.5189) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [23] [5580/7269] eta: 0:12:07 lr: 0.000005 loss: 0.5424 (0.5189) time: 0.4352 data: 0.0002 max mem: 52150 +Epoch: [23] [5590/7269] eta: 0:12:03 lr: 0.000005 loss: 0.5424 (0.5190) time: 0.4344 data: 0.0002 max mem: 52150 +Epoch: [23] [5600/7269] eta: 0:11:58 lr: 0.000005 loss: 0.5563 (0.5190) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [23] [5610/7269] eta: 0:11:54 lr: 0.000005 loss: 0.5551 (0.5190) time: 0.4774 data: 0.0001 max mem: 52150 +Epoch: [23] [5620/7269] eta: 0:11:50 lr: 0.000005 loss: 0.5468 (0.5191) time: 0.4776 data: 0.0002 max mem: 52150 +Epoch: [23] [5630/7269] eta: 0:11:45 lr: 0.000005 loss: 0.5419 (0.5191) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [23] [5640/7269] eta: 0:11:41 lr: 0.000005 loss: 0.5335 (0.5191) time: 0.3727 data: 0.0002 max mem: 52150 +Epoch: [23] [5650/7269] eta: 0:11:36 lr: 0.000005 loss: 0.5335 (0.5191) time: 0.3727 data: 0.0002 max mem: 52150 +Epoch: [23] [5660/7269] eta: 0:11:32 lr: 0.000005 loss: 0.5376 (0.5192) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [23] [5670/7269] eta: 0:11:28 lr: 0.000005 loss: 0.5629 (0.5192) time: 0.4212 data: 0.0002 max mem: 52150 +Epoch: [23] [5680/7269] eta: 0:11:24 lr: 0.000005 loss: 0.5446 (0.5192) time: 0.4680 data: 0.0001 max mem: 52150 +Epoch: [23] [5690/7269] eta: 0:11:19 lr: 0.000005 loss: 0.4962 (0.5191) time: 0.4668 data: 0.0002 max mem: 52150 +Epoch: [23] [5700/7269] eta: 0:11:15 lr: 0.000005 loss: 0.5064 (0.5192) time: 0.4202 data: 0.0002 max mem: 52150 +Epoch: [23] [5710/7269] eta: 0:11:10 lr: 0.000005 loss: 0.5402 (0.5191) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [23] [5720/7269] eta: 0:11:06 lr: 0.000005 loss: 0.5135 (0.5191) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [23] [5730/7269] eta: 0:11:01 lr: 0.000005 loss: 0.5241 (0.5191) time: 0.3728 data: 0.0002 max mem: 52150 +Epoch: [23] [5740/7269] eta: 0:10:57 lr: 0.000005 loss: 0.5232 (0.5191) time: 0.4173 data: 0.0002 max mem: 52150 +Epoch: [23] [5750/7269] eta: 0:10:53 lr: 0.000005 loss: 0.5111 (0.5192) time: 0.5403 data: 0.0002 max mem: 52150 +Epoch: [23] [5760/7269] eta: 0:10:49 lr: 0.000005 loss: 0.5222 (0.5192) time: 0.5416 data: 0.0002 max mem: 52150 +Epoch: [23] [5770/7269] eta: 0:10:45 lr: 0.000005 loss: 0.5267 (0.5192) time: 0.4184 data: 0.0002 max mem: 52150 +Epoch: [23] [5780/7269] eta: 0:10:41 lr: 0.000005 loss: 0.5267 (0.5192) time: 0.4670 data: 0.0002 max mem: 52150 +Epoch: [23] [5790/7269] eta: 0:10:36 lr: 0.000005 loss: 0.4957 (0.5192) time: 0.4666 data: 0.0002 max mem: 52150 +Epoch: [23] [5800/7269] eta: 0:10:32 lr: 0.000005 loss: 0.5060 (0.5192) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [23] [5810/7269] eta: 0:10:27 lr: 0.000005 loss: 0.5275 (0.5192) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [23] [5820/7269] eta: 0:10:23 lr: 0.000005 loss: 0.5550 (0.5192) time: 0.4233 data: 0.0002 max mem: 52150 +Epoch: [23] [5830/7269] eta: 0:10:19 lr: 0.000005 loss: 0.5453 (0.5192) time: 0.4735 data: 0.0002 max mem: 52150 +Epoch: [23] [5840/7269] eta: 0:10:15 lr: 0.000005 loss: 0.5065 (0.5192) time: 0.5156 data: 0.0002 max mem: 52150 +Epoch: [23] [5850/7269] eta: 0:10:11 lr: 0.000005 loss: 0.5060 (0.5192) time: 0.4656 data: 0.0002 max mem: 52150 +Epoch: [23] [5860/7269] eta: 0:10:06 lr: 0.000005 loss: 0.5099 (0.5192) time: 0.4366 data: 0.0002 max mem: 52150 +Epoch: [23] [5870/7269] eta: 0:10:02 lr: 0.000005 loss: 0.5447 (0.5192) time: 0.4362 data: 0.0002 max mem: 52150 +Epoch: [23] [5880/7269] eta: 0:09:58 lr: 0.000005 loss: 0.5411 (0.5192) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [23] [5890/7269] eta: 0:09:53 lr: 0.000005 loss: 0.5069 (0.5192) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [23] [5900/7269] eta: 0:09:49 lr: 0.000005 loss: 0.5069 (0.5191) time: 0.4780 data: 0.0002 max mem: 52150 +Epoch: [23] [5910/7269] eta: 0:09:45 lr: 0.000005 loss: 0.5352 (0.5192) time: 0.5228 data: 0.0002 max mem: 52150 +Epoch: [23] [5920/7269] eta: 0:09:41 lr: 0.000005 loss: 0.5301 (0.5192) time: 0.4174 data: 0.0002 max mem: 52150 +Epoch: [23] [5930/7269] eta: 0:09:36 lr: 0.000005 loss: 0.5212 (0.5193) time: 0.4343 data: 0.0002 max mem: 52150 +Epoch: [23] [5940/7269] eta: 0:09:33 lr: 0.000005 loss: 0.4986 (0.5192) time: 0.6025 data: 0.0002 max mem: 52150 +Epoch: [23] [5950/7269] eta: 0:09:28 lr: 0.000005 loss: 0.5119 (0.5192) time: 0.5402 data: 0.0002 max mem: 52150 +Epoch: [23] [5960/7269] eta: 0:09:24 lr: 0.000005 loss: 0.5208 (0.5192) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [23] [5970/7269] eta: 0:09:19 lr: 0.000005 loss: 0.5030 (0.5191) time: 0.3729 data: 0.0002 max mem: 52150 +Epoch: [23] [5980/7269] eta: 0:09:15 lr: 0.000005 loss: 0.4929 (0.5191) time: 0.3727 data: 0.0002 max mem: 52150 +Epoch: [23] [5990/7269] eta: 0:09:11 lr: 0.000005 loss: 0.5058 (0.5191) time: 0.4215 data: 0.0001 max mem: 52150 +Epoch: [23] [6000/7269] eta: 0:09:06 lr: 0.000005 loss: 0.5333 (0.5192) time: 0.4654 data: 0.0002 max mem: 52150 +Epoch: [23] [6010/7269] eta: 0:09:02 lr: 0.000005 loss: 0.4884 (0.5191) time: 0.4163 data: 0.0002 max mem: 52150 +Epoch: [23] [6020/7269] eta: 0:08:58 lr: 0.000005 loss: 0.4747 (0.5191) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [23] [6030/7269] eta: 0:08:53 lr: 0.000005 loss: 0.5232 (0.5191) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [23] [6040/7269] eta: 0:08:50 lr: 0.000005 loss: 0.5223 (0.5191) time: 0.5932 data: 0.0001 max mem: 52150 +Epoch: [23] [6050/7269] eta: 0:08:45 lr: 0.000005 loss: 0.5189 (0.5191) time: 0.5932 data: 0.0001 max mem: 52150 +Epoch: [23] [6060/7269] eta: 0:08:41 lr: 0.000005 loss: 0.4949 (0.5191) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [23] [6070/7269] eta: 0:08:37 lr: 0.000005 loss: 0.5091 (0.5191) time: 0.4320 data: 0.0002 max mem: 52150 +Epoch: [23] [6080/7269] eta: 0:08:32 lr: 0.000005 loss: 0.5233 (0.5191) time: 0.4319 data: 0.0002 max mem: 52150 +Epoch: [23] [6090/7269] eta: 0:08:28 lr: 0.000005 loss: 0.5273 (0.5191) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [23] [6100/7269] eta: 0:08:23 lr: 0.000005 loss: 0.5273 (0.5191) time: 0.3730 data: 0.0002 max mem: 52150 +Epoch: [23] [6110/7269] eta: 0:08:19 lr: 0.000005 loss: 0.5025 (0.5191) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [23] [6120/7269] eta: 0:08:14 lr: 0.000005 loss: 0.4979 (0.5190) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [23] [6130/7269] eta: 0:08:10 lr: 0.000005 loss: 0.4837 (0.5190) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [23] [6140/7269] eta: 0:08:06 lr: 0.000005 loss: 0.5485 (0.5190) time: 0.4210 data: 0.0002 max mem: 52150 +Epoch: [23] [6150/7269] eta: 0:08:01 lr: 0.000005 loss: 0.5369 (0.5190) time: 0.4213 data: 0.0002 max mem: 52150 +Epoch: [23] [6160/7269] eta: 0:07:57 lr: 0.000005 loss: 0.5221 (0.5190) time: 0.4189 data: 0.0002 max mem: 52150 +Epoch: [23] [6170/7269] eta: 0:07:53 lr: 0.000005 loss: 0.5123 (0.5190) time: 0.4825 data: 0.0002 max mem: 52150 +Epoch: [23] [6180/7269] eta: 0:07:49 lr: 0.000005 loss: 0.5274 (0.5190) time: 0.4362 data: 0.0002 max mem: 52150 +Epoch: [23] [6190/7269] eta: 0:07:44 lr: 0.000005 loss: 0.4983 (0.5190) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [23] [6200/7269] eta: 0:07:40 lr: 0.000005 loss: 0.4913 (0.5189) time: 0.4204 data: 0.0001 max mem: 52150 +Epoch: [23] [6210/7269] eta: 0:07:36 lr: 0.000005 loss: 0.5056 (0.5189) time: 0.4835 data: 0.0001 max mem: 52150 +Epoch: [23] [6220/7269] eta: 0:07:32 lr: 0.000005 loss: 0.5056 (0.5189) time: 0.5436 data: 0.0002 max mem: 52150 +Epoch: [23] [6230/7269] eta: 0:07:27 lr: 0.000005 loss: 0.5263 (0.5189) time: 0.4804 data: 0.0002 max mem: 52150 +Epoch: [23] [6240/7269] eta: 0:07:23 lr: 0.000005 loss: 0.5263 (0.5189) time: 0.4344 data: 0.0002 max mem: 52150 +Epoch: [23] [6250/7269] eta: 0:07:19 lr: 0.000005 loss: 0.5227 (0.5189) time: 0.4955 data: 0.0002 max mem: 52150 +Epoch: [23] [6260/7269] eta: 0:07:14 lr: 0.000005 loss: 0.5221 (0.5189) time: 0.4330 data: 0.0002 max mem: 52150 +Epoch: [23] [6270/7269] eta: 0:07:10 lr: 0.000005 loss: 0.5317 (0.5190) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [23] [6280/7269] eta: 0:07:06 lr: 0.000005 loss: 0.5182 (0.5189) time: 0.3730 data: 0.0002 max mem: 52150 +Epoch: [23] [6290/7269] eta: 0:07:01 lr: 0.000005 loss: 0.4740 (0.5189) time: 0.3731 data: 0.0002 max mem: 52150 +Epoch: [23] [6300/7269] eta: 0:06:57 lr: 0.000005 loss: 0.5254 (0.5189) time: 0.4172 data: 0.0002 max mem: 52150 +Epoch: [23] [6310/7269] eta: 0:06:53 lr: 0.000005 loss: 0.5548 (0.5189) time: 0.4169 data: 0.0002 max mem: 52150 +Epoch: [23] [6320/7269] eta: 0:06:48 lr: 0.000005 loss: 0.5374 (0.5189) time: 0.4177 data: 0.0002 max mem: 52150 +Epoch: [23] [6330/7269] eta: 0:06:44 lr: 0.000005 loss: 0.5478 (0.5190) time: 0.4174 data: 0.0002 max mem: 52150 +Epoch: [23] [6340/7269] eta: 0:06:40 lr: 0.000005 loss: 0.5497 (0.5190) time: 0.4200 data: 0.0002 max mem: 52150 +Epoch: [23] [6350/7269] eta: 0:06:35 lr: 0.000005 loss: 0.4958 (0.5189) time: 0.4208 data: 0.0002 max mem: 52150 +Epoch: [23] [6360/7269] eta: 0:06:31 lr: 0.000005 loss: 0.4798 (0.5189) time: 0.4360 data: 0.0002 max mem: 52150 +Epoch: [23] [6370/7269] eta: 0:06:27 lr: 0.000005 loss: 0.4951 (0.5189) time: 0.5758 data: 0.0002 max mem: 52150 +Epoch: [23] [6380/7269] eta: 0:06:23 lr: 0.000005 loss: 0.5194 (0.5189) time: 0.5129 data: 0.0002 max mem: 52150 +Epoch: [23] [6390/7269] eta: 0:06:18 lr: 0.000005 loss: 0.5074 (0.5189) time: 0.4323 data: 0.0002 max mem: 52150 +Epoch: [23] [6400/7269] eta: 0:06:14 lr: 0.000005 loss: 0.5074 (0.5189) time: 0.5383 data: 0.0002 max mem: 52150 +Epoch: [23] [6410/7269] eta: 0:06:10 lr: 0.000005 loss: 0.5124 (0.5189) time: 0.4787 data: 0.0002 max mem: 52150 +Epoch: [23] [6420/7269] eta: 0:06:06 lr: 0.000005 loss: 0.5541 (0.5189) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [23] [6430/7269] eta: 0:06:01 lr: 0.000005 loss: 0.5292 (0.5189) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [23] [6440/7269] eta: 0:05:57 lr: 0.000005 loss: 0.5265 (0.5189) time: 0.4192 data: 0.0001 max mem: 52150 +Epoch: [23] [6450/7269] eta: 0:05:53 lr: 0.000005 loss: 0.5286 (0.5190) time: 0.4680 data: 0.0002 max mem: 52150 +Epoch: [23] [6460/7269] eta: 0:05:48 lr: 0.000005 loss: 0.5644 (0.5190) time: 0.4813 data: 0.0002 max mem: 52150 +Epoch: [23] [6470/7269] eta: 0:05:44 lr: 0.000005 loss: 0.5711 (0.5191) time: 0.4325 data: 0.0002 max mem: 52150 +Epoch: [23] [6480/7269] eta: 0:05:40 lr: 0.000005 loss: 0.5337 (0.5191) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [23] [6490/7269] eta: 0:05:35 lr: 0.000005 loss: 0.5064 (0.5190) time: 0.4339 data: 0.0001 max mem: 52150 +Epoch: [23] [6500/7269] eta: 0:05:31 lr: 0.000005 loss: 0.5133 (0.5190) time: 0.4336 data: 0.0002 max mem: 52150 +Epoch: [23] [6510/7269] eta: 0:05:27 lr: 0.000005 loss: 0.5133 (0.5190) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [23] [6520/7269] eta: 0:05:22 lr: 0.000005 loss: 0.4746 (0.5190) time: 0.4215 data: 0.0002 max mem: 52150 +Epoch: [23] [6530/7269] eta: 0:05:18 lr: 0.000005 loss: 0.5264 (0.5190) time: 0.4214 data: 0.0001 max mem: 52150 +Epoch: [23] [6540/7269] eta: 0:05:14 lr: 0.000005 loss: 0.5450 (0.5190) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [23] [6550/7269] eta: 0:05:09 lr: 0.000005 loss: 0.5427 (0.5190) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [23] [6560/7269] eta: 0:05:05 lr: 0.000005 loss: 0.5220 (0.5190) time: 0.3731 data: 0.0002 max mem: 52150 +Epoch: [23] [6570/7269] eta: 0:05:01 lr: 0.000005 loss: 0.5220 (0.5190) time: 0.3728 data: 0.0002 max mem: 52150 +Epoch: [23] [6580/7269] eta: 0:04:56 lr: 0.000005 loss: 0.4897 (0.5189) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [23] [6590/7269] eta: 0:04:52 lr: 0.000005 loss: 0.5189 (0.5190) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [23] [6600/7269] eta: 0:04:47 lr: 0.000005 loss: 0.5334 (0.5190) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [23] [6610/7269] eta: 0:04:43 lr: 0.000005 loss: 0.5071 (0.5189) time: 0.4189 data: 0.0002 max mem: 52150 +Epoch: [23] [6620/7269] eta: 0:04:39 lr: 0.000005 loss: 0.5007 (0.5189) time: 0.4655 data: 0.0002 max mem: 52150 +Epoch: [23] [6630/7269] eta: 0:04:35 lr: 0.000005 loss: 0.5176 (0.5189) time: 0.4187 data: 0.0002 max mem: 52150 +Epoch: [23] [6640/7269] eta: 0:04:30 lr: 0.000005 loss: 0.5183 (0.5189) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [23] [6650/7269] eta: 0:04:26 lr: 0.000005 loss: 0.5029 (0.5188) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [23] [6660/7269] eta: 0:04:21 lr: 0.000005 loss: 0.4646 (0.5187) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [23] [6670/7269] eta: 0:04:17 lr: 0.000005 loss: 0.4904 (0.5187) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [23] [6680/7269] eta: 0:04:13 lr: 0.000005 loss: 0.4915 (0.5187) time: 0.4338 data: 0.0001 max mem: 52150 +Epoch: [23] [6690/7269] eta: 0:04:09 lr: 0.000005 loss: 0.5105 (0.5187) time: 0.4816 data: 0.0001 max mem: 52150 +Epoch: [23] [6700/7269] eta: 0:04:04 lr: 0.000005 loss: 0.5304 (0.5187) time: 0.4206 data: 0.0002 max mem: 52150 +Epoch: [23] [6710/7269] eta: 0:04:00 lr: 0.000005 loss: 0.5357 (0.5187) time: 0.4338 data: 0.0002 max mem: 52150 +Epoch: [23] [6720/7269] eta: 0:03:56 lr: 0.000005 loss: 0.5237 (0.5187) time: 0.4328 data: 0.0002 max mem: 52150 +Epoch: [23] [6730/7269] eta: 0:03:51 lr: 0.000005 loss: 0.5145 (0.5188) time: 0.4165 data: 0.0002 max mem: 52150 +Epoch: [23] [6740/7269] eta: 0:03:47 lr: 0.000005 loss: 0.5051 (0.5188) time: 0.4170 data: 0.0002 max mem: 52150 +Epoch: [23] [6750/7269] eta: 0:03:43 lr: 0.000005 loss: 0.5203 (0.5188) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [23] [6760/7269] eta: 0:03:38 lr: 0.000005 loss: 0.5182 (0.5188) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [23] [6770/7269] eta: 0:03:34 lr: 0.000005 loss: 0.5055 (0.5188) time: 0.4320 data: 0.0002 max mem: 52150 +Epoch: [23] [6780/7269] eta: 0:03:30 lr: 0.000005 loss: 0.5357 (0.5188) time: 0.6204 data: 0.0002 max mem: 52150 +Epoch: [23] [6790/7269] eta: 0:03:26 lr: 0.000005 loss: 0.5138 (0.5188) time: 0.6075 data: 0.0002 max mem: 52150 +Epoch: [23] [6800/7269] eta: 0:03:22 lr: 0.000005 loss: 0.5096 (0.5188) time: 0.5291 data: 0.0002 max mem: 52150 +Epoch: [23] [6810/7269] eta: 0:03:17 lr: 0.000005 loss: 0.5133 (0.5187) time: 0.5270 data: 0.0002 max mem: 52150 +Epoch: [23] [6820/7269] eta: 0:03:13 lr: 0.000005 loss: 0.5086 (0.5187) time: 0.4174 data: 0.0002 max mem: 52150 +Epoch: [23] [6830/7269] eta: 0:03:09 lr: 0.000005 loss: 0.5245 (0.5187) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [23] [6840/7269] eta: 0:03:04 lr: 0.000005 loss: 0.5376 (0.5188) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [23] [6850/7269] eta: 0:03:00 lr: 0.000005 loss: 0.5130 (0.5188) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [23] [6860/7269] eta: 0:02:56 lr: 0.000005 loss: 0.4835 (0.5187) time: 0.4179 data: 0.0002 max mem: 52150 +Epoch: [23] [6870/7269] eta: 0:02:51 lr: 0.000005 loss: 0.5333 (0.5188) time: 0.4181 data: 0.0002 max mem: 52150 +Epoch: [23] [6880/7269] eta: 0:02:47 lr: 0.000005 loss: 0.5338 (0.5188) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [23] [6890/7269] eta: 0:02:43 lr: 0.000005 loss: 0.5285 (0.5188) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [23] [6900/7269] eta: 0:02:38 lr: 0.000005 loss: 0.5290 (0.5188) time: 0.4215 data: 0.0002 max mem: 52150 +Epoch: [23] [6910/7269] eta: 0:02:34 lr: 0.000005 loss: 0.5089 (0.5188) time: 0.4817 data: 0.0002 max mem: 52150 +Epoch: [23] [6920/7269] eta: 0:02:30 lr: 0.000005 loss: 0.5023 (0.5188) time: 0.4324 data: 0.0002 max mem: 52150 +Epoch: [23] [6930/7269] eta: 0:02:25 lr: 0.000005 loss: 0.5576 (0.5188) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [23] [6940/7269] eta: 0:02:21 lr: 0.000005 loss: 0.5606 (0.5188) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [23] [6950/7269] eta: 0:02:17 lr: 0.000005 loss: 0.5226 (0.5188) time: 0.4796 data: 0.0002 max mem: 52150 +Epoch: [23] [6960/7269] eta: 0:02:12 lr: 0.000005 loss: 0.5138 (0.5188) time: 0.4794 data: 0.0002 max mem: 52150 +Epoch: [23] [6970/7269] eta: 0:02:08 lr: 0.000005 loss: 0.5359 (0.5188) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [23] [6980/7269] eta: 0:02:04 lr: 0.000005 loss: 0.5098 (0.5188) time: 0.4673 data: 0.0002 max mem: 52150 +Epoch: [23] [6990/7269] eta: 0:02:00 lr: 0.000005 loss: 0.5098 (0.5188) time: 0.5164 data: 0.0002 max mem: 52150 +Epoch: [23] [7000/7269] eta: 0:01:55 lr: 0.000005 loss: 0.5173 (0.5188) time: 0.5435 data: 0.0002 max mem: 52150 +Epoch: [23] [7010/7269] eta: 0:01:51 lr: 0.000005 loss: 0.5262 (0.5188) time: 0.4948 data: 0.0002 max mem: 52150 +Epoch: [23] [7020/7269] eta: 0:01:47 lr: 0.000005 loss: 0.5460 (0.5188) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [23] [7030/7269] eta: 0:01:42 lr: 0.000005 loss: 0.5460 (0.5189) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [23] [7040/7269] eta: 0:01:38 lr: 0.000005 loss: 0.5440 (0.5189) time: 0.4319 data: 0.0002 max mem: 52150 +Epoch: [23] [7050/7269] eta: 0:01:34 lr: 0.000005 loss: 0.5305 (0.5189) time: 0.5322 data: 0.0002 max mem: 52150 +Epoch: [23] [7060/7269] eta: 0:01:30 lr: 0.000005 loss: 0.5305 (0.5190) time: 0.4720 data: 0.0002 max mem: 52150 +Epoch: [23] [7070/7269] eta: 0:01:25 lr: 0.000005 loss: 0.5129 (0.5189) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [23] [7080/7269] eta: 0:01:21 lr: 0.000005 loss: 0.5104 (0.5189) time: 0.4187 data: 0.0002 max mem: 52150 +Epoch: [23] [7090/7269] eta: 0:01:17 lr: 0.000005 loss: 0.5187 (0.5189) time: 0.4183 data: 0.0002 max mem: 52150 +Epoch: [23] [7100/7269] eta: 0:01:12 lr: 0.000005 loss: 0.5029 (0.5189) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [23] [7110/7269] eta: 0:01:08 lr: 0.000005 loss: 0.5029 (0.5189) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [23] [7120/7269] eta: 0:01:04 lr: 0.000005 loss: 0.5142 (0.5189) time: 0.4193 data: 0.0002 max mem: 52150 +Epoch: [23] [7130/7269] eta: 0:00:59 lr: 0.000005 loss: 0.5494 (0.5189) time: 0.4195 data: 0.0002 max mem: 52150 +Epoch: [23] [7140/7269] eta: 0:00:55 lr: 0.000005 loss: 0.5347 (0.5189) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [23] [7150/7269] eta: 0:00:51 lr: 0.000005 loss: 0.4551 (0.5188) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [23] [7160/7269] eta: 0:00:46 lr: 0.000005 loss: 0.4609 (0.5188) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [23] [7170/7269] eta: 0:00:42 lr: 0.000005 loss: 0.5174 (0.5188) time: 0.3727 data: 0.0002 max mem: 52150 +Epoch: [23] [7180/7269] eta: 0:00:38 lr: 0.000005 loss: 0.5174 (0.5188) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [23] [7190/7269] eta: 0:00:33 lr: 0.000005 loss: 0.4954 (0.5187) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [23] [7200/7269] eta: 0:00:29 lr: 0.000005 loss: 0.5073 (0.5187) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [23] [7210/7269] eta: 0:00:25 lr: 0.000005 loss: 0.5117 (0.5187) time: 0.4325 data: 0.0002 max mem: 52150 +Epoch: [23] [7220/7269] eta: 0:00:21 lr: 0.000005 loss: 0.5117 (0.5187) time: 0.4318 data: 0.0002 max mem: 52150 +Epoch: [23] [7230/7269] eta: 0:00:16 lr: 0.000005 loss: 0.5278 (0.5187) time: 0.4781 data: 0.0002 max mem: 52150 +Epoch: [23] [7240/7269] eta: 0:00:12 lr: 0.000005 loss: 0.5244 (0.5186) time: 0.4783 data: 0.0002 max mem: 52150 +Epoch: [23] [7250/7269] eta: 0:00:08 lr: 0.000005 loss: 0.5140 (0.5186) time: 0.3754 data: 0.0033 max mem: 52150 +Epoch: [23] [7260/7269] eta: 0:00:03 lr: 0.000005 loss: 0.5358 (0.5187) time: 0.4787 data: 0.0032 max mem: 52150 +Epoch: [23] [7268/7269] eta: 0:00:00 lr: 0.000005 loss: 0.5421 (0.5187) time: 0.5224 data: 0.0032 max mem: 52150 +Epoch: [23] Total time: 0:52:09 (0.4306 s / it) +Averaged stats: lr: 0.000005 loss: 0.5421 (0.5192) +Test: [ 0/1403] eta: 1:18:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0924 (0.0924) acc1: 100.0000 (100.0000) acc5: 100.0000 (100.0000) time: 3.3615 data: 3.2638 max mem: 52150 +Test: [ 10/1403] eta: 0:09:07 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1523 (0.1970) acc1: 98.4375 (94.4602) acc5: 100.0000 (100.0000) time: 0.3933 data: 0.2970 max mem: 52150 +Test: [ 20/1403] eta: 0:05:48 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1433 (0.1762) acc1: 98.4375 (95.0149) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 30/1403] eta: 0:04:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1001 (0.1640) acc1: 98.4375 (95.4637) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 40/1403] eta: 0:04:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1135 (0.1670) acc1: 98.4375 (95.6174) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 50/1403] eta: 0:03:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1265 (0.1877) acc1: 98.4375 (94.3934) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 60/1403] eta: 0:03:21 acc1_pos: 89.062 acc5_pos: 100.000 loss: 0.1350 (0.1828) acc1: 100.0000 (94.9795) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 70/1403] eta: 0:03:10 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.1391 (0.1832) acc1: 98.4375 (94.8283) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 80/1403] eta: 0:03:01 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.1391 (0.1886) acc1: 96.8750 (94.6566) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 90/1403] eta: 0:02:53 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.1868 (0.2060) acc1: 95.3125 (93.8015) acc5: 100.0000 (100.0000) time: 0.0963 data: 0.0003 max mem: 52150 +Test: [ 100/1403] eta: 0:02:47 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.1913 (0.2045) acc1: 95.3125 (93.9975) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 110/1403] eta: 0:02:42 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1636 (0.2024) acc1: 96.8750 (94.0878) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 120/1403] eta: 0:02:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0836 (0.1923) acc1: 100.0000 (94.5506) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 130/1403] eta: 0:02:34 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0857 (0.1853) acc1: 100.0000 (94.8593) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 140/1403] eta: 0:02:31 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0768 (0.1774) acc1: 100.0000 (95.2238) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 150/1403] eta: 0:02:28 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0757 (0.1712) acc1: 100.0000 (95.4884) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 160/1403] eta: 0:02:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0787 (0.1663) acc1: 100.0000 (95.6813) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0004 max mem: 52150 +Test: [ 170/1403] eta: 0:02:22 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0790 (0.1616) acc1: 100.0000 (95.9064) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 180/1403] eta: 0:02:20 acc1_pos: 81.250 acc5_pos: 100.000 loss: 0.0799 (0.1677) acc1: 100.0000 (95.5801) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 190/1403] eta: 0:02:17 acc1_pos: 81.250 acc5_pos: 100.000 loss: 0.3513 (0.1807) acc1: 84.3750 (94.8217) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 200/1403] eta: 0:02:15 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.3513 (0.1965) acc1: 84.3750 (93.9988) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 210/1403] eta: 0:02:13 acc1_pos: 73.438 acc5_pos: 100.000 loss: 0.4287 (0.2115) acc1: 78.1250 (93.1798) acc5: 100.0000 (100.0000) time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 220/1403] eta: 0:02:11 acc1_pos: 84.375 acc5_pos: 100.000 loss: 0.4623 (0.2236) acc1: 78.1250 (92.5693) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 230/1403] eta: 0:02:09 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.3406 (0.2187) acc1: 87.5000 (92.8166) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 240/1403] eta: 0:02:08 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0816 (0.2131) acc1: 100.0000 (93.0952) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 250/1403] eta: 0:02:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0789 (0.2081) acc1: 100.0000 (93.3454) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 260/1403] eta: 0:02:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0794 (0.2034) acc1: 100.0000 (93.5824) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 270/1403] eta: 0:02:03 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0800 (0.1990) acc1: 100.0000 (93.7961) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 280/1403] eta: 0:02:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0810 (0.1948) acc1: 100.0000 (94.0113) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 290/1403] eta: 0:01:59 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0738 (0.1907) acc1: 100.0000 (94.2118) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 300/1403] eta: 0:01:58 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0730 (0.1870) acc1: 100.0000 (94.3937) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 310/1403] eta: 0:01:56 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0811 (0.1837) acc1: 100.0000 (94.5639) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 320/1403] eta: 0:01:55 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0957 (0.1811) acc1: 100.0000 (94.7333) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 330/1403] eta: 0:01:54 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1015 (0.1788) acc1: 100.0000 (94.8924) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 340/1403] eta: 0:01:52 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1015 (0.1764) acc1: 100.0000 (95.0422) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 350/1403] eta: 0:01:51 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0985 (0.1742) acc1: 100.0000 (95.1834) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 360/1403] eta: 0:01:50 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1019 (0.1723) acc1: 100.0000 (95.3125) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 370/1403] eta: 0:01:48 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1003 (0.1703) acc1: 100.0000 (95.4388) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 380/1403] eta: 0:01:47 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0984 (0.1685) acc1: 100.0000 (95.5586) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 390/1403] eta: 0:01:46 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0999 (0.1682) acc1: 100.0000 (95.5683) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 400/1403] eta: 0:01:44 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2020 (0.1705) acc1: 92.1875 (95.4138) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 410/1403] eta: 0:01:43 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2363 (0.1720) acc1: 90.6250 (95.3163) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 420/1403] eta: 0:01:42 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1999 (0.1724) acc1: 92.1875 (95.2717) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 430/1403] eta: 0:01:41 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1911 (0.1740) acc1: 93.7500 (95.1711) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 440/1403] eta: 0:01:40 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2704 (0.1763) acc1: 89.0625 (95.0503) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 450/1403] eta: 0:01:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2390 (0.1774) acc1: 89.0625 (94.9834) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 460/1403] eta: 0:01:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2113 (0.1786) acc1: 92.1875 (94.8990) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 470/1403] eta: 0:01:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1856 (0.1772) acc1: 93.7500 (94.9476) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 480/1403] eta: 0:01:35 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0618 (0.1748) acc1: 100.0000 (95.0526) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 490/1403] eta: 0:01:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0622 (0.1725) acc1: 100.0000 (95.1534) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 500/1403] eta: 0:01:33 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0623 (0.1703) acc1: 100.0000 (95.2501) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 510/1403] eta: 0:01:31 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0627 (0.1683) acc1: 100.0000 (95.3431) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 520/1403] eta: 0:01:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0658 (0.1664) acc1: 100.0000 (95.4325) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 530/1403] eta: 0:01:29 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0594 (0.1643) acc1: 100.0000 (95.5185) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 540/1403] eta: 0:01:28 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0588 (0.1624) acc1: 100.0000 (95.6013) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 550/1403] eta: 0:01:27 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0596 (0.1608) acc1: 100.0000 (95.6726) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 560/1403] eta: 0:01:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0726 (0.1594) acc1: 100.0000 (95.7331) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 570/1403] eta: 0:01:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0753 (0.1580) acc1: 100.0000 (95.7968) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 580/1403] eta: 0:01:24 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0765 (0.1571) acc1: 100.0000 (95.8342) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 590/1403] eta: 0:01:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1049 (0.1565) acc1: 98.4375 (95.8571) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 600/1403] eta: 0:01:21 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1164 (0.1564) acc1: 95.3125 (95.8351) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 610/1403] eta: 0:01:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1137 (0.1555) acc1: 95.3125 (95.8674) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 620/1403] eta: 0:01:19 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1176 (0.1560) acc1: 95.3125 (95.8057) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 630/1403] eta: 0:01:18 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1315 (0.1555) acc1: 95.3125 (95.8127) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 640/1403] eta: 0:01:17 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1128 (0.1548) acc1: 96.8750 (95.8439) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 650/1403] eta: 0:01:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0975 (0.1539) acc1: 98.4375 (95.8933) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 660/1403] eta: 0:01:15 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0762 (0.1527) acc1: 100.0000 (95.9507) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 670/1403] eta: 0:01:14 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0670 (0.1515) acc1: 100.0000 (96.0088) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 680/1403] eta: 0:01:13 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0731 (0.1520) acc1: 98.4375 (95.9710) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 690/1403] eta: 0:01:12 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1971 (0.1550) acc1: 92.1875 (95.7896) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 700/1403] eta: 0:01:11 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1383 (0.1549) acc1: 95.3125 (95.7962) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 710/1403] eta: 0:01:10 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2145 (0.1579) acc1: 90.6250 (95.6355) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 720/1403] eta: 0:01:09 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2145 (0.1576) acc1: 90.6250 (95.6462) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 730/1403] eta: 0:01:08 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0649 (0.1565) acc1: 100.0000 (95.6951) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 740/1403] eta: 0:01:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0688 (0.1555) acc1: 100.0000 (95.7384) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 750/1403] eta: 0:01:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0885 (0.1566) acc1: 98.4375 (95.6787) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 760/1403] eta: 0:01:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3022 (0.1594) acc1: 87.5000 (95.4932) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 770/1403] eta: 0:01:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4185 (0.1632) acc1: 81.2500 (95.2720) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 780/1403] eta: 0:01:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3849 (0.1653) acc1: 82.8125 (95.1404) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 790/1403] eta: 0:01:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3030 (0.1672) acc1: 85.9375 (95.0419) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 800/1403] eta: 0:01:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1912 (0.1674) acc1: 93.7500 (95.0336) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 810/1403] eta: 0:00:59 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2114 (0.1702) acc1: 92.1875 (94.9041) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 820/1403] eta: 0:00:58 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6119 (0.1787) acc1: 71.8750 (94.5303) acc5: 100.0000 (100.0000) acc1_neg: 43.750 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 830/1403] eta: 0:00:57 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6200 (0.1809) acc1: 71.8750 (94.4269) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 840/1403] eta: 0:00:56 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4530 (0.1876) acc1: 78.1250 (94.1346) acc5: 100.0000 (100.0000) acc1_neg: 56.250 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 850/1403] eta: 0:00:55 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.8798 (0.1969) acc1: 64.0625 (93.7408) acc5: 100.0000 (100.0000) acc1_neg: 64.062 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 860/1403] eta: 0:00:54 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.8055 (0.1997) acc1: 65.6250 (93.6139) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 870/1403] eta: 0:00:53 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1150 (0.1985) acc1: 96.8750 (93.6693) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 880/1403] eta: 0:00:52 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0830 (0.1973) acc1: 98.4375 (93.7252) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 890/1403] eta: 0:00:51 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0763 (0.1960) acc1: 100.0000 (93.7886) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 900/1403] eta: 0:00:50 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0703 (0.1946) acc1: 100.0000 (93.8558) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 910/1403] eta: 0:00:49 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0729 (0.1936) acc1: 100.0000 (93.8924) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 920/1403] eta: 0:00:48 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0953 (0.1927) acc1: 98.4375 (93.9366) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 930/1403] eta: 0:00:47 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0919 (0.1916) acc1: 98.4375 (93.9917) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 940/1403] eta: 0:00:46 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1054 (0.1914) acc1: 98.4375 (93.9908) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 950/1403] eta: 0:00:45 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1879 (0.1920) acc1: 89.0625 (93.9488) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 960/1403] eta: 0:00:44 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2338 (0.1924) acc1: 89.0625 (93.9175) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 970/1403] eta: 0:00:43 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2161 (0.1930) acc1: 92.1875 (93.8852) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 980/1403] eta: 0:00:42 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2324 (0.1935) acc1: 90.6250 (93.8456) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 990/1403] eta: 0:00:41 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2199 (0.1938) acc1: 92.1875 (93.8273) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [1000/1403] eta: 0:00:40 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2030 (0.1942) acc1: 92.1875 (93.7968) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [1010/1403] eta: 0:00:39 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2370 (0.1948) acc1: 90.6250 (93.7577) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [1020/1403] eta: 0:00:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3709 (0.2000) acc1: 84.3750 (93.4669) acc5: 100.0000 (100.0000) acc1_neg: 34.375 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [1030/1403] eta: 0:00:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4037 (0.2016) acc1: 73.4375 (93.3590) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [1040/1403] eta: 0:00:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1828 (0.2012) acc1: 92.1875 (93.3718) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [1050/1403] eta: 0:00:35 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1578 (0.2008) acc1: 95.3125 (93.3902) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [1060/1403] eta: 0:00:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1128 (0.2000) acc1: 98.4375 (93.4334) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [1070/1403] eta: 0:00:33 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1220 (0.2002) acc1: 96.8750 (93.4276) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [1080/1403] eta: 0:00:32 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2889 (0.2013) acc1: 87.5000 (93.3525) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [1090/1403] eta: 0:00:31 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3205 (0.2025) acc1: 85.9375 (93.2831) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0970 data: 0.0003 max mem: 52150 +Test: [1100/1403] eta: 0:00:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2927 (0.2029) acc1: 87.5000 (93.2632) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0970 data: 0.0003 max mem: 52150 +Test: [1110/1403] eta: 0:00:29 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2440 (0.2032) acc1: 90.6250 (93.2479) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [1120/1403] eta: 0:00:28 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2440 (0.2036) acc1: 90.6250 (93.2259) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1130/1403] eta: 0:00:27 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2521 (0.2042) acc1: 90.6250 (93.2002) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [1140/1403] eta: 0:00:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2203 (0.2042) acc1: 92.1875 (93.1954) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [1150/1403] eta: 0:00:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2116 (0.2044) acc1: 92.1875 (93.1771) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [1160/1403] eta: 0:00:24 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2318 (0.2048) acc1: 90.6250 (93.1619) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1170/1403] eta: 0:00:23 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2301 (0.2046) acc1: 90.6250 (93.1656) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0002 max mem: 52150 +Test: [1180/1403] eta: 0:00:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0692 (0.2035) acc1: 100.0000 (93.2234) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0002 max mem: 52150 +Test: [1190/1403] eta: 0:00:21 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0685 (0.2024) acc1: 100.0000 (93.2803) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0002 max mem: 52150 +Test: [1200/1403] eta: 0:00:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0685 (0.2012) acc1: 100.0000 (93.3363) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1210/1403] eta: 0:00:19 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0678 (0.2001) acc1: 100.0000 (93.3913) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1220/1403] eta: 0:00:18 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0677 (0.1991) acc1: 100.0000 (93.4454) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0002 max mem: 52150 +Test: [1230/1403] eta: 0:00:17 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0683 (0.1980) acc1: 100.0000 (93.4987) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1240/1403] eta: 0:00:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0683 (0.1969) acc1: 100.0000 (93.5511) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1250/1403] eta: 0:00:15 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0684 (0.1959) acc1: 100.0000 (93.6026) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1260/1403] eta: 0:00:14 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0723 (0.1950) acc1: 100.0000 (93.6534) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1270/1403] eta: 0:00:13 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0735 (0.1940) acc1: 100.0000 (93.7033) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1280/1403] eta: 0:00:12 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0733 (0.1931) acc1: 100.0000 (93.7524) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1290/1403] eta: 0:00:11 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0728 (0.1921) acc1: 100.0000 (93.8008) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1300/1403] eta: 0:00:10 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0739 (0.1912) acc1: 100.0000 (93.8485) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1310/1403] eta: 0:00:09 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0739 (0.1903) acc1: 100.0000 (93.8954) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1320/1403] eta: 0:00:08 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0723 (0.1895) acc1: 100.0000 (93.9416) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1330/1403] eta: 0:00:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0712 (0.1885) acc1: 100.0000 (93.9860) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1340/1403] eta: 0:00:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0647 (0.1876) acc1: 100.0000 (94.0308) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1350/1403] eta: 0:00:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0638 (0.1867) acc1: 100.0000 (94.0727) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1360/1403] eta: 0:00:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0637 (0.1858) acc1: 100.0000 (94.1162) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1370/1403] eta: 0:00:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0635 (0.1850) acc1: 100.0000 (94.1569) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1380/1403] eta: 0:00:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0628 (0.1841) acc1: 100.0000 (94.1992) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1390/1403] eta: 0:00:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0630 (0.1832) acc1: 100.0000 (94.2409) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52150 +Test: [1400/1403] eta: 0:00:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0632 (0.1824) acc1: 100.0000 (94.2798) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0001 max mem: 52150 +Test: [1402/1403] eta: 0:00:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0634 (0.1822) acc1: 100.0000 (94.2839) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0923 data: 0.0001 max mem: 52150 +Test: Total time: 0:02:19 (0.0993 s / it) +* Overall Acc@1 94.284 Acc@5 100.000 + target == real Acc@1 94.516 Acc@5 100.000 + target == fake Acc@1 94.219 Acc@5 100.000 + loss 0.182 +Accuracy of the network on the 89729 test images: 94.3% +Max accuracy: 94.28% +Epoch: [24] [ 0/7269] eta: 8:54:07 lr: 0.000004 loss: 0.6201 (0.6201) time: 4.4088 data: 4.0424 max mem: 52150 +Epoch: [24] [ 10/7269] eta: 1:29:46 lr: 0.000004 loss: 0.5753 (0.5571) time: 0.7421 data: 0.3677 max mem: 52150 +Epoch: [24] [ 20/7269] eta: 1:28:26 lr: 0.000004 loss: 0.5644 (0.5494) time: 0.5482 data: 0.0002 max mem: 52150 +Epoch: [24] [ 30/7269] eta: 1:19:14 lr: 0.000004 loss: 0.5261 (0.5405) time: 0.6098 data: 0.0001 max mem: 52150 +Epoch: [24] [ 40/7269] eta: 1:14:22 lr: 0.000004 loss: 0.5261 (0.5387) time: 0.4969 data: 0.0002 max mem: 52150 +Epoch: [24] [ 50/7269] eta: 1:08:30 lr: 0.000004 loss: 0.5499 (0.5340) time: 0.4339 data: 0.0002 max mem: 52150 +Epoch: [24] [ 60/7269] eta: 1:06:19 lr: 0.000004 loss: 0.5390 (0.5334) time: 0.4182 data: 0.0002 max mem: 52150 +Epoch: [24] [ 70/7269] eta: 1:03:10 lr: 0.000004 loss: 0.5431 (0.5331) time: 0.4175 data: 0.0002 max mem: 52150 +Epoch: [24] [ 80/7269] eta: 1:00:47 lr: 0.000004 loss: 0.5431 (0.5335) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [24] [ 90/7269] eta: 1:00:12 lr: 0.000004 loss: 0.5490 (0.5358) time: 0.4203 data: 0.0002 max mem: 52150 +Epoch: [24] [ 100/7269] eta: 1:00:00 lr: 0.000004 loss: 0.5490 (0.5333) time: 0.4813 data: 0.0002 max mem: 52150 +Epoch: [24] [ 110/7269] eta: 0:59:32 lr: 0.000004 loss: 0.5293 (0.5335) time: 0.4800 data: 0.0002 max mem: 52150 +Epoch: [24] [ 120/7269] eta: 1:00:27 lr: 0.000004 loss: 0.5651 (0.5361) time: 0.5338 data: 0.0002 max mem: 52150 +Epoch: [24] [ 130/7269] eta: 0:59:08 lr: 0.000004 loss: 0.5372 (0.5340) time: 0.4862 data: 0.0002 max mem: 52150 +Epoch: [24] [ 140/7269] eta: 0:58:00 lr: 0.000004 loss: 0.5201 (0.5339) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [24] [ 150/7269] eta: 0:57:00 lr: 0.000004 loss: 0.5274 (0.5340) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [24] [ 160/7269] eta: 0:56:49 lr: 0.000004 loss: 0.5233 (0.5323) time: 0.4191 data: 0.0002 max mem: 52150 +Epoch: [24] [ 170/7269] eta: 0:55:59 lr: 0.000004 loss: 0.5090 (0.5306) time: 0.4190 data: 0.0002 max mem: 52150 +Epoch: [24] [ 180/7269] eta: 0:55:15 lr: 0.000004 loss: 0.5142 (0.5292) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [24] [ 190/7269] eta: 0:56:09 lr: 0.000004 loss: 0.5428 (0.5299) time: 0.4990 data: 0.0001 max mem: 52150 +Epoch: [24] [ 200/7269] eta: 0:56:12 lr: 0.000004 loss: 0.5361 (0.5284) time: 0.5618 data: 0.0001 max mem: 52150 +Epoch: [24] [ 210/7269] eta: 0:55:32 lr: 0.000004 loss: 0.5087 (0.5271) time: 0.4345 data: 0.0001 max mem: 52150 +Epoch: [24] [ 220/7269] eta: 0:54:55 lr: 0.000004 loss: 0.4825 (0.5252) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [24] [ 230/7269] eta: 0:54:21 lr: 0.000004 loss: 0.4919 (0.5233) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [24] [ 240/7269] eta: 0:53:50 lr: 0.000004 loss: 0.4919 (0.5216) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [24] [ 250/7269] eta: 0:53:55 lr: 0.000004 loss: 0.5371 (0.5231) time: 0.4345 data: 0.0001 max mem: 52150 +Epoch: [24] [ 260/7269] eta: 0:53:53 lr: 0.000004 loss: 0.5371 (0.5229) time: 0.4821 data: 0.0002 max mem: 52150 +Epoch: [24] [ 270/7269] eta: 0:53:25 lr: 0.000004 loss: 0.5014 (0.5222) time: 0.4194 data: 0.0002 max mem: 52150 +Epoch: [24] [ 280/7269] eta: 0:52:59 lr: 0.000004 loss: 0.4936 (0.5214) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [24] [ 290/7269] eta: 0:52:34 lr: 0.000004 loss: 0.5209 (0.5212) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [24] [ 300/7269] eta: 0:52:11 lr: 0.000004 loss: 0.5494 (0.5214) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [24] [ 310/7269] eta: 0:51:49 lr: 0.000004 loss: 0.5329 (0.5208) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [24] [ 320/7269] eta: 0:51:56 lr: 0.000004 loss: 0.5067 (0.5210) time: 0.4352 data: 0.0002 max mem: 52150 +Epoch: [24] [ 330/7269] eta: 0:51:35 lr: 0.000004 loss: 0.5199 (0.5204) time: 0.4352 data: 0.0002 max mem: 52150 +Epoch: [24] [ 340/7269] eta: 0:51:41 lr: 0.000004 loss: 0.5329 (0.5204) time: 0.4326 data: 0.0001 max mem: 52150 +Epoch: [24] [ 350/7269] eta: 0:51:40 lr: 0.000004 loss: 0.5450 (0.5206) time: 0.4817 data: 0.0002 max mem: 52150 +Epoch: [24] [ 360/7269] eta: 0:51:21 lr: 0.000004 loss: 0.5227 (0.5187) time: 0.4203 data: 0.0002 max mem: 52150 +Epoch: [24] [ 370/7269] eta: 0:51:03 lr: 0.000004 loss: 0.5430 (0.5199) time: 0.3711 data: 0.0001 max mem: 52150 +Epoch: [24] [ 380/7269] eta: 0:50:45 lr: 0.000004 loss: 0.5474 (0.5206) time: 0.3711 data: 0.0001 max mem: 52150 +Epoch: [24] [ 390/7269] eta: 0:50:28 lr: 0.000004 loss: 0.5361 (0.5204) time: 0.3711 data: 0.0001 max mem: 52150 +Epoch: [24] [ 400/7269] eta: 0:50:12 lr: 0.000004 loss: 0.4872 (0.5190) time: 0.3713 data: 0.0001 max mem: 52150 +Epoch: [24] [ 410/7269] eta: 0:49:56 lr: 0.000004 loss: 0.4879 (0.5183) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [24] [ 420/7269] eta: 0:49:41 lr: 0.000004 loss: 0.5051 (0.5185) time: 0.3712 data: 0.0001 max mem: 52150 +Epoch: [24] [ 430/7269] eta: 0:49:27 lr: 0.000004 loss: 0.5500 (0.5190) time: 0.3708 data: 0.0002 max mem: 52150 +Epoch: [24] [ 440/7269] eta: 0:49:13 lr: 0.000004 loss: 0.5429 (0.5192) time: 0.3710 data: 0.0002 max mem: 52150 +Epoch: [24] [ 450/7269] eta: 0:49:32 lr: 0.000004 loss: 0.5372 (0.5197) time: 0.4810 data: 0.0002 max mem: 52150 +Epoch: [24] [ 460/7269] eta: 0:49:19 lr: 0.000004 loss: 0.5292 (0.5197) time: 0.4815 data: 0.0001 max mem: 52150 +Epoch: [24] [ 470/7269] eta: 0:49:05 lr: 0.000004 loss: 0.5156 (0.5198) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [24] [ 480/7269] eta: 0:49:09 lr: 0.000004 loss: 0.5146 (0.5195) time: 0.4327 data: 0.0002 max mem: 52150 +Epoch: [24] [ 490/7269] eta: 0:49:25 lr: 0.000004 loss: 0.5431 (0.5201) time: 0.5384 data: 0.0002 max mem: 52150 +Epoch: [24] [ 500/7269] eta: 0:49:12 lr: 0.000004 loss: 0.5555 (0.5206) time: 0.4773 data: 0.0002 max mem: 52150 +Epoch: [24] [ 510/7269] eta: 0:48:59 lr: 0.000004 loss: 0.5485 (0.5207) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [24] [ 520/7269] eta: 0:48:59 lr: 0.000004 loss: 0.5314 (0.5208) time: 0.4195 data: 0.0001 max mem: 52150 +Epoch: [24] [ 530/7269] eta: 0:48:47 lr: 0.000004 loss: 0.5261 (0.5206) time: 0.4195 data: 0.0002 max mem: 52150 +Epoch: [24] [ 540/7269] eta: 0:48:51 lr: 0.000004 loss: 0.5013 (0.5205) time: 0.4363 data: 0.0002 max mem: 52150 +Epoch: [24] [ 550/7269] eta: 0:48:39 lr: 0.000004 loss: 0.5013 (0.5199) time: 0.4366 data: 0.0001 max mem: 52150 +Epoch: [24] [ 560/7269] eta: 0:48:27 lr: 0.000004 loss: 0.4933 (0.5196) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [24] [ 570/7269] eta: 0:48:15 lr: 0.000004 loss: 0.4984 (0.5196) time: 0.3711 data: 0.0001 max mem: 52150 +Epoch: [24] [ 580/7269] eta: 0:48:29 lr: 0.000004 loss: 0.5134 (0.5193) time: 0.4793 data: 0.0001 max mem: 52150 +Epoch: [24] [ 590/7269] eta: 0:48:31 lr: 0.000004 loss: 0.5164 (0.5194) time: 0.5423 data: 0.0002 max mem: 52150 +Epoch: [24] [ 600/7269] eta: 0:48:20 lr: 0.000004 loss: 0.5156 (0.5189) time: 0.4343 data: 0.0002 max mem: 52150 +Epoch: [24] [ 610/7269] eta: 0:48:19 lr: 0.000004 loss: 0.5144 (0.5194) time: 0.4198 data: 0.0002 max mem: 52150 +Epoch: [24] [ 620/7269] eta: 0:48:08 lr: 0.000004 loss: 0.5406 (0.5195) time: 0.4198 data: 0.0002 max mem: 52150 +Epoch: [24] [ 630/7269] eta: 0:48:10 lr: 0.000004 loss: 0.5406 (0.5199) time: 0.4345 data: 0.0002 max mem: 52150 +Epoch: [24] [ 640/7269] eta: 0:48:09 lr: 0.000004 loss: 0.5534 (0.5198) time: 0.4806 data: 0.0001 max mem: 52150 +Epoch: [24] [ 650/7269] eta: 0:48:11 lr: 0.000004 loss: 0.5277 (0.5199) time: 0.4821 data: 0.0001 max mem: 52150 +Epoch: [24] [ 660/7269] eta: 0:48:30 lr: 0.000004 loss: 0.5469 (0.5206) time: 0.5882 data: 0.0001 max mem: 52150 +Epoch: [24] [ 670/7269] eta: 0:48:31 lr: 0.000004 loss: 0.5449 (0.5209) time: 0.5830 data: 0.0001 max mem: 52150 +Epoch: [24] [ 680/7269] eta: 0:48:20 lr: 0.000004 loss: 0.5437 (0.5215) time: 0.4307 data: 0.0002 max mem: 52150 +Epoch: [24] [ 690/7269] eta: 0:48:17 lr: 0.000004 loss: 0.5566 (0.5221) time: 0.4148 data: 0.0001 max mem: 52150 +Epoch: [24] [ 700/7269] eta: 0:48:06 lr: 0.000004 loss: 0.5250 (0.5217) time: 0.4154 data: 0.0001 max mem: 52150 +Epoch: [24] [ 710/7269] eta: 0:48:07 lr: 0.000004 loss: 0.5014 (0.5219) time: 0.4320 data: 0.0001 max mem: 52150 +Epoch: [24] [ 720/7269] eta: 0:48:05 lr: 0.000004 loss: 0.5359 (0.5221) time: 0.4826 data: 0.0001 max mem: 52150 +Epoch: [24] [ 730/7269] eta: 0:48:06 lr: 0.000004 loss: 0.5183 (0.5222) time: 0.4846 data: 0.0001 max mem: 52150 +Epoch: [24] [ 740/7269] eta: 0:48:04 lr: 0.000004 loss: 0.5390 (0.5226) time: 0.4806 data: 0.0002 max mem: 52150 +Epoch: [24] [ 750/7269] eta: 0:47:53 lr: 0.000004 loss: 0.5545 (0.5226) time: 0.4187 data: 0.0002 max mem: 52150 +Epoch: [24] [ 760/7269] eta: 0:47:43 lr: 0.000004 loss: 0.5525 (0.5225) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [24] [ 770/7269] eta: 0:47:41 lr: 0.000004 loss: 0.5194 (0.5223) time: 0.4197 data: 0.0001 max mem: 52150 +Epoch: [24] [ 780/7269] eta: 0:47:40 lr: 0.000004 loss: 0.5299 (0.5227) time: 0.4794 data: 0.0001 max mem: 52150 +Epoch: [24] [ 790/7269] eta: 0:47:30 lr: 0.000004 loss: 0.5471 (0.5229) time: 0.4309 data: 0.0002 max mem: 52150 +Epoch: [24] [ 800/7269] eta: 0:47:31 lr: 0.000004 loss: 0.5245 (0.5226) time: 0.4381 data: 0.0001 max mem: 52150 +Epoch: [24] [ 810/7269] eta: 0:47:21 lr: 0.000004 loss: 0.5284 (0.5228) time: 0.4385 data: 0.0002 max mem: 52150 +Epoch: [24] [ 820/7269] eta: 0:47:21 lr: 0.000004 loss: 0.5587 (0.5231) time: 0.4306 data: 0.0002 max mem: 52150 +Epoch: [24] [ 830/7269] eta: 0:47:11 lr: 0.000004 loss: 0.5261 (0.5230) time: 0.4304 data: 0.0001 max mem: 52150 +Epoch: [24] [ 840/7269] eta: 0:47:01 lr: 0.000004 loss: 0.5143 (0.5228) time: 0.3714 data: 0.0001 max mem: 52150 +Epoch: [24] [ 850/7269] eta: 0:46:52 lr: 0.000004 loss: 0.5143 (0.5225) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [24] [ 860/7269] eta: 0:46:58 lr: 0.000004 loss: 0.5034 (0.5222) time: 0.4765 data: 0.0002 max mem: 52150 +Epoch: [24] [ 870/7269] eta: 0:46:49 lr: 0.000004 loss: 0.5004 (0.5218) time: 0.4761 data: 0.0002 max mem: 52150 +Epoch: [24] [ 880/7269] eta: 0:46:39 lr: 0.000004 loss: 0.5087 (0.5216) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [24] [ 890/7269] eta: 0:46:30 lr: 0.000004 loss: 0.5112 (0.5214) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [24] [ 900/7269] eta: 0:46:21 lr: 0.000004 loss: 0.5112 (0.5210) time: 0.3722 data: 0.0001 max mem: 52150 +Epoch: [24] [ 910/7269] eta: 0:46:12 lr: 0.000004 loss: 0.4957 (0.5207) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [24] [ 920/7269] eta: 0:46:04 lr: 0.000004 loss: 0.5002 (0.5205) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [24] [ 930/7269] eta: 0:46:01 lr: 0.000004 loss: 0.5381 (0.5208) time: 0.4172 data: 0.0001 max mem: 52150 +Epoch: [24] [ 940/7269] eta: 0:45:53 lr: 0.000004 loss: 0.5404 (0.5209) time: 0.4171 data: 0.0001 max mem: 52150 +Epoch: [24] [ 950/7269] eta: 0:45:50 lr: 0.000004 loss: 0.5649 (0.5216) time: 0.4167 data: 0.0002 max mem: 52150 +Epoch: [24] [ 960/7269] eta: 0:45:41 lr: 0.000004 loss: 0.5564 (0.5213) time: 0.4169 data: 0.0002 max mem: 52150 +Epoch: [24] [ 970/7269] eta: 0:45:33 lr: 0.000004 loss: 0.5104 (0.5212) time: 0.3723 data: 0.0001 max mem: 52150 +Epoch: [24] [ 980/7269] eta: 0:45:25 lr: 0.000004 loss: 0.5309 (0.5213) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [24] [ 990/7269] eta: 0:45:16 lr: 0.000004 loss: 0.5285 (0.5213) time: 0.3712 data: 0.0001 max mem: 52150 +Epoch: [24] [1000/7269] eta: 0:45:08 lr: 0.000004 loss: 0.5285 (0.5212) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [24] [1010/7269] eta: 0:45:00 lr: 0.000004 loss: 0.5358 (0.5211) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [24] [1020/7269] eta: 0:45:00 lr: 0.000004 loss: 0.5278 (0.5211) time: 0.4316 data: 0.0001 max mem: 52150 +Epoch: [24] [1030/7269] eta: 0:44:57 lr: 0.000004 loss: 0.4657 (0.5203) time: 0.4772 data: 0.0001 max mem: 52150 +Epoch: [24] [1040/7269] eta: 0:44:56 lr: 0.000004 loss: 0.4949 (0.5206) time: 0.4782 data: 0.0001 max mem: 52150 +Epoch: [24] [1050/7269] eta: 0:44:49 lr: 0.000004 loss: 0.5290 (0.5202) time: 0.4328 data: 0.0002 max mem: 52150 +Epoch: [24] [1060/7269] eta: 0:44:48 lr: 0.000004 loss: 0.5075 (0.5203) time: 0.4332 data: 0.0002 max mem: 52150 +Epoch: [24] [1070/7269] eta: 0:44:47 lr: 0.000004 loss: 0.5124 (0.5203) time: 0.4926 data: 0.0001 max mem: 52150 +Epoch: [24] [1080/7269] eta: 0:44:39 lr: 0.000004 loss: 0.5124 (0.5201) time: 0.4311 data: 0.0001 max mem: 52150 +Epoch: [24] [1090/7269] eta: 0:44:31 lr: 0.000004 loss: 0.5106 (0.5200) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [24] [1100/7269] eta: 0:44:30 lr: 0.000004 loss: 0.5307 (0.5202) time: 0.4317 data: 0.0001 max mem: 52150 +Epoch: [24] [1110/7269] eta: 0:44:28 lr: 0.000004 loss: 0.5442 (0.5205) time: 0.4799 data: 0.0002 max mem: 52150 +Epoch: [24] [1120/7269] eta: 0:44:32 lr: 0.000004 loss: 0.5595 (0.5205) time: 0.5258 data: 0.0002 max mem: 52150 +Epoch: [24] [1130/7269] eta: 0:44:36 lr: 0.000004 loss: 0.4621 (0.5202) time: 0.5840 data: 0.0002 max mem: 52150 +Epoch: [24] [1140/7269] eta: 0:44:28 lr: 0.000004 loss: 0.5050 (0.5203) time: 0.4780 data: 0.0002 max mem: 52150 +Epoch: [24] [1150/7269] eta: 0:44:20 lr: 0.000004 loss: 0.5280 (0.5203) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [24] [1160/7269] eta: 0:44:17 lr: 0.000004 loss: 0.5510 (0.5205) time: 0.4178 data: 0.0001 max mem: 52150 +Epoch: [24] [1170/7269] eta: 0:44:09 lr: 0.000004 loss: 0.5557 (0.5205) time: 0.4172 data: 0.0001 max mem: 52150 +Epoch: [24] [1180/7269] eta: 0:44:02 lr: 0.000004 loss: 0.5355 (0.5205) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [24] [1190/7269] eta: 0:43:54 lr: 0.000004 loss: 0.4787 (0.5202) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [24] [1200/7269] eta: 0:43:52 lr: 0.000004 loss: 0.4744 (0.5199) time: 0.4188 data: 0.0001 max mem: 52150 +Epoch: [24] [1210/7269] eta: 0:43:56 lr: 0.000004 loss: 0.4992 (0.5199) time: 0.5390 data: 0.0001 max mem: 52150 +Epoch: [24] [1220/7269] eta: 0:43:53 lr: 0.000004 loss: 0.5035 (0.5196) time: 0.5387 data: 0.0001 max mem: 52150 +Epoch: [24] [1230/7269] eta: 0:43:57 lr: 0.000004 loss: 0.5212 (0.5195) time: 0.5269 data: 0.0001 max mem: 52150 +Epoch: [24] [1240/7269] eta: 0:43:49 lr: 0.000004 loss: 0.5345 (0.5196) time: 0.4801 data: 0.0002 max mem: 52150 +Epoch: [24] [1250/7269] eta: 0:43:42 lr: 0.000004 loss: 0.5396 (0.5196) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [24] [1260/7269] eta: 0:43:34 lr: 0.000004 loss: 0.5396 (0.5196) time: 0.3713 data: 0.0001 max mem: 52150 +Epoch: [24] [1270/7269] eta: 0:43:27 lr: 0.000004 loss: 0.5285 (0.5196) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [24] [1280/7269] eta: 0:43:20 lr: 0.000004 loss: 0.5285 (0.5196) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [24] [1290/7269] eta: 0:43:24 lr: 0.000004 loss: 0.5499 (0.5197) time: 0.4934 data: 0.0001 max mem: 52150 +Epoch: [24] [1300/7269] eta: 0:43:26 lr: 0.000004 loss: 0.5569 (0.5201) time: 0.6047 data: 0.0002 max mem: 52150 +Epoch: [24] [1310/7269] eta: 0:43:24 lr: 0.000004 loss: 0.5569 (0.5201) time: 0.5348 data: 0.0001 max mem: 52150 +Epoch: [24] [1320/7269] eta: 0:43:17 lr: 0.000004 loss: 0.5217 (0.5200) time: 0.4235 data: 0.0001 max mem: 52150 +Epoch: [24] [1330/7269] eta: 0:43:09 lr: 0.000004 loss: 0.5266 (0.5201) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [24] [1340/7269] eta: 0:43:02 lr: 0.000004 loss: 0.5354 (0.5201) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [24] [1350/7269] eta: 0:42:55 lr: 0.000004 loss: 0.5313 (0.5202) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [24] [1360/7269] eta: 0:42:48 lr: 0.000004 loss: 0.5313 (0.5202) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [24] [1370/7269] eta: 0:42:41 lr: 0.000004 loss: 0.5323 (0.5203) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [24] [1380/7269] eta: 0:42:34 lr: 0.000004 loss: 0.5262 (0.5203) time: 0.3714 data: 0.0001 max mem: 52150 +Epoch: [24] [1390/7269] eta: 0:42:34 lr: 0.000004 loss: 0.5333 (0.5205) time: 0.4611 data: 0.0001 max mem: 52150 +Epoch: [24] [1400/7269] eta: 0:42:27 lr: 0.000004 loss: 0.5668 (0.5206) time: 0.4619 data: 0.0001 max mem: 52150 +Epoch: [24] [1410/7269] eta: 0:42:26 lr: 0.000004 loss: 0.5562 (0.5208) time: 0.4335 data: 0.0001 max mem: 52150 +Epoch: [24] [1420/7269] eta: 0:42:19 lr: 0.000004 loss: 0.5422 (0.5210) time: 0.4328 data: 0.0001 max mem: 52150 +Epoch: [24] [1430/7269] eta: 0:42:12 lr: 0.000004 loss: 0.5429 (0.5211) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [24] [1440/7269] eta: 0:42:05 lr: 0.000004 loss: 0.5223 (0.5211) time: 0.3723 data: 0.0001 max mem: 52150 +Epoch: [24] [1450/7269] eta: 0:41:58 lr: 0.000004 loss: 0.4972 (0.5209) time: 0.3725 data: 0.0001 max mem: 52150 +Epoch: [24] [1460/7269] eta: 0:41:51 lr: 0.000004 loss: 0.5002 (0.5206) time: 0.3722 data: 0.0001 max mem: 52150 +Epoch: [24] [1470/7269] eta: 0:41:45 lr: 0.000004 loss: 0.5335 (0.5206) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [24] [1480/7269] eta: 0:41:38 lr: 0.000004 loss: 0.5002 (0.5203) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [24] [1490/7269] eta: 0:41:31 lr: 0.000004 loss: 0.4887 (0.5202) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [24] [1500/7269] eta: 0:41:25 lr: 0.000004 loss: 0.5087 (0.5203) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [24] [1510/7269] eta: 0:41:18 lr: 0.000004 loss: 0.5198 (0.5203) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [24] [1520/7269] eta: 0:41:15 lr: 0.000004 loss: 0.5337 (0.5204) time: 0.4185 data: 0.0001 max mem: 52150 +Epoch: [24] [1530/7269] eta: 0:41:13 lr: 0.000004 loss: 0.5258 (0.5205) time: 0.4790 data: 0.0001 max mem: 52150 +Epoch: [24] [1540/7269] eta: 0:41:07 lr: 0.000004 loss: 0.5187 (0.5205) time: 0.4320 data: 0.0001 max mem: 52150 +Epoch: [24] [1550/7269] eta: 0:41:05 lr: 0.000004 loss: 0.5404 (0.5204) time: 0.4312 data: 0.0001 max mem: 52150 +Epoch: [24] [1560/7269] eta: 0:41:07 lr: 0.000004 loss: 0.5195 (0.5204) time: 0.5531 data: 0.0001 max mem: 52150 +Epoch: [24] [1570/7269] eta: 0:41:05 lr: 0.000004 loss: 0.5201 (0.5206) time: 0.5553 data: 0.0002 max mem: 52150 +Epoch: [24] [1580/7269] eta: 0:41:02 lr: 0.000004 loss: 0.5239 (0.5205) time: 0.4799 data: 0.0002 max mem: 52150 +Epoch: [24] [1590/7269] eta: 0:41:02 lr: 0.000004 loss: 0.4962 (0.5204) time: 0.5091 data: 0.0002 max mem: 52150 +Epoch: [24] [1600/7269] eta: 0:40:55 lr: 0.000004 loss: 0.5035 (0.5204) time: 0.4628 data: 0.0001 max mem: 52150 +Epoch: [24] [1610/7269] eta: 0:40:49 lr: 0.000004 loss: 0.5270 (0.5203) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [24] [1620/7269] eta: 0:40:51 lr: 0.000004 loss: 0.5270 (0.5203) time: 0.4905 data: 0.0001 max mem: 52150 +Epoch: [24] [1630/7269] eta: 0:40:48 lr: 0.000004 loss: 0.5299 (0.5204) time: 0.5503 data: 0.0002 max mem: 52150 +Epoch: [24] [1640/7269] eta: 0:40:42 lr: 0.000004 loss: 0.5235 (0.5203) time: 0.4313 data: 0.0002 max mem: 52150 +Epoch: [24] [1650/7269] eta: 0:40:35 lr: 0.000004 loss: 0.5171 (0.5203) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [24] [1660/7269] eta: 0:40:33 lr: 0.000004 loss: 0.5465 (0.5203) time: 0.4338 data: 0.0002 max mem: 52150 +Epoch: [24] [1670/7269] eta: 0:40:34 lr: 0.000004 loss: 0.5334 (0.5202) time: 0.5379 data: 0.0002 max mem: 52150 +Epoch: [24] [1680/7269] eta: 0:40:31 lr: 0.000004 loss: 0.5272 (0.5203) time: 0.5276 data: 0.0001 max mem: 52150 +Epoch: [24] [1690/7269] eta: 0:40:27 lr: 0.000004 loss: 0.5081 (0.5200) time: 0.4686 data: 0.0001 max mem: 52150 +Epoch: [24] [1700/7269] eta: 0:40:24 lr: 0.000004 loss: 0.4530 (0.5198) time: 0.4620 data: 0.0001 max mem: 52150 +Epoch: [24] [1710/7269] eta: 0:40:17 lr: 0.000004 loss: 0.5309 (0.5198) time: 0.4174 data: 0.0002 max mem: 52150 +Epoch: [24] [1720/7269] eta: 0:40:11 lr: 0.000004 loss: 0.5317 (0.5197) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [24] [1730/7269] eta: 0:40:05 lr: 0.000004 loss: 0.5517 (0.5198) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [24] [1740/7269] eta: 0:39:58 lr: 0.000004 loss: 0.5345 (0.5198) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [24] [1750/7269] eta: 0:39:55 lr: 0.000004 loss: 0.5095 (0.5198) time: 0.4243 data: 0.0001 max mem: 52150 +Epoch: [24] [1760/7269] eta: 0:39:49 lr: 0.000004 loss: 0.5378 (0.5199) time: 0.4245 data: 0.0001 max mem: 52150 +Epoch: [24] [1770/7269] eta: 0:39:43 lr: 0.000004 loss: 0.5378 (0.5199) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [24] [1780/7269] eta: 0:39:40 lr: 0.000004 loss: 0.4652 (0.5196) time: 0.4350 data: 0.0001 max mem: 52150 +Epoch: [24] [1790/7269] eta: 0:39:34 lr: 0.000004 loss: 0.4645 (0.5195) time: 0.4351 data: 0.0001 max mem: 52150 +Epoch: [24] [1800/7269] eta: 0:39:31 lr: 0.000004 loss: 0.5219 (0.5197) time: 0.4204 data: 0.0001 max mem: 52150 +Epoch: [24] [1810/7269] eta: 0:39:28 lr: 0.000004 loss: 0.5409 (0.5195) time: 0.4810 data: 0.0001 max mem: 52150 +Epoch: [24] [1820/7269] eta: 0:39:26 lr: 0.000004 loss: 0.4946 (0.5195) time: 0.4958 data: 0.0001 max mem: 52150 +Epoch: [24] [1830/7269] eta: 0:39:20 lr: 0.000004 loss: 0.5066 (0.5192) time: 0.4353 data: 0.0001 max mem: 52150 +Epoch: [24] [1840/7269] eta: 0:39:16 lr: 0.000004 loss: 0.5066 (0.5191) time: 0.4176 data: 0.0001 max mem: 52150 +Epoch: [24] [1850/7269] eta: 0:39:10 lr: 0.000004 loss: 0.5109 (0.5190) time: 0.4173 data: 0.0001 max mem: 52150 +Epoch: [24] [1860/7269] eta: 0:39:04 lr: 0.000004 loss: 0.5109 (0.5190) time: 0.3724 data: 0.0001 max mem: 52150 +Epoch: [24] [1870/7269] eta: 0:39:01 lr: 0.000004 loss: 0.5409 (0.5191) time: 0.4311 data: 0.0002 max mem: 52150 +Epoch: [24] [1880/7269] eta: 0:38:55 lr: 0.000004 loss: 0.5495 (0.5191) time: 0.4306 data: 0.0002 max mem: 52150 +Epoch: [24] [1890/7269] eta: 0:38:52 lr: 0.000004 loss: 0.5119 (0.5190) time: 0.4169 data: 0.0001 max mem: 52150 +Epoch: [24] [1900/7269] eta: 0:38:49 lr: 0.000004 loss: 0.4931 (0.5188) time: 0.4774 data: 0.0002 max mem: 52150 +Epoch: [24] [1910/7269] eta: 0:38:43 lr: 0.000004 loss: 0.4931 (0.5187) time: 0.4323 data: 0.0002 max mem: 52150 +Epoch: [24] [1920/7269] eta: 0:38:37 lr: 0.000004 loss: 0.5384 (0.5189) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [24] [1930/7269] eta: 0:38:34 lr: 0.000004 loss: 0.5563 (0.5189) time: 0.4248 data: 0.0001 max mem: 52150 +Epoch: [24] [1940/7269] eta: 0:38:28 lr: 0.000004 loss: 0.5359 (0.5189) time: 0.4247 data: 0.0001 max mem: 52150 +Epoch: [24] [1950/7269] eta: 0:38:24 lr: 0.000004 loss: 0.5259 (0.5188) time: 0.4174 data: 0.0001 max mem: 52150 +Epoch: [24] [1960/7269] eta: 0:38:18 lr: 0.000004 loss: 0.5132 (0.5187) time: 0.4174 data: 0.0001 max mem: 52150 +Epoch: [24] [1970/7269] eta: 0:38:12 lr: 0.000004 loss: 0.5072 (0.5188) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [24] [1980/7269] eta: 0:38:06 lr: 0.000004 loss: 0.5009 (0.5187) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [24] [1990/7269] eta: 0:38:00 lr: 0.000004 loss: 0.5421 (0.5188) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [24] [2000/7269] eta: 0:37:55 lr: 0.000004 loss: 0.5461 (0.5188) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [24] [2010/7269] eta: 0:37:49 lr: 0.000004 loss: 0.5194 (0.5187) time: 0.3727 data: 0.0001 max mem: 52150 +Epoch: [24] [2020/7269] eta: 0:37:46 lr: 0.000004 loss: 0.5072 (0.5187) time: 0.4334 data: 0.0002 max mem: 52150 +Epoch: [24] [2030/7269] eta: 0:37:40 lr: 0.000004 loss: 0.5099 (0.5186) time: 0.4329 data: 0.0002 max mem: 52150 +Epoch: [24] [2040/7269] eta: 0:37:34 lr: 0.000004 loss: 0.5080 (0.5186) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [24] [2050/7269] eta: 0:37:28 lr: 0.000004 loss: 0.5080 (0.5186) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [24] [2060/7269] eta: 0:37:23 lr: 0.000004 loss: 0.4800 (0.5184) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [24] [2070/7269] eta: 0:37:17 lr: 0.000004 loss: 0.5064 (0.5183) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [24] [2080/7269] eta: 0:37:11 lr: 0.000004 loss: 0.5345 (0.5183) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [24] [2090/7269] eta: 0:37:08 lr: 0.000004 loss: 0.5184 (0.5182) time: 0.4315 data: 0.0001 max mem: 52150 +Epoch: [24] [2100/7269] eta: 0:37:10 lr: 0.000004 loss: 0.5087 (0.5181) time: 0.5877 data: 0.0001 max mem: 52150 +Epoch: [24] [2110/7269] eta: 0:37:04 lr: 0.000004 loss: 0.4975 (0.5179) time: 0.5275 data: 0.0002 max mem: 52150 +Epoch: [24] [2120/7269] eta: 0:36:59 lr: 0.000004 loss: 0.4741 (0.5179) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [24] [2130/7269] eta: 0:36:53 lr: 0.000004 loss: 0.4876 (0.5177) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [24] [2140/7269] eta: 0:36:47 lr: 0.000004 loss: 0.4916 (0.5176) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [24] [2150/7269] eta: 0:36:44 lr: 0.000004 loss: 0.5143 (0.5177) time: 0.4201 data: 0.0002 max mem: 52150 +Epoch: [24] [2160/7269] eta: 0:36:38 lr: 0.000004 loss: 0.5315 (0.5177) time: 0.4201 data: 0.0002 max mem: 52150 +Epoch: [24] [2170/7269] eta: 0:36:40 lr: 0.000004 loss: 0.5298 (0.5178) time: 0.5381 data: 0.0002 max mem: 52150 +Epoch: [24] [2180/7269] eta: 0:36:44 lr: 0.000004 loss: 0.5111 (0.5176) time: 0.7352 data: 0.0001 max mem: 52150 +Epoch: [24] [2190/7269] eta: 0:36:40 lr: 0.000004 loss: 0.5030 (0.5176) time: 0.6163 data: 0.0001 max mem: 52150 +Epoch: [24] [2200/7269] eta: 0:36:37 lr: 0.000004 loss: 0.5287 (0.5177) time: 0.4655 data: 0.0001 max mem: 52150 +Epoch: [24] [2210/7269] eta: 0:36:31 lr: 0.000004 loss: 0.5486 (0.5178) time: 0.4181 data: 0.0002 max mem: 52150 +Epoch: [24] [2220/7269] eta: 0:36:25 lr: 0.000004 loss: 0.5438 (0.5178) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [24] [2230/7269] eta: 0:36:22 lr: 0.000004 loss: 0.5053 (0.5177) time: 0.4201 data: 0.0001 max mem: 52150 +Epoch: [24] [2240/7269] eta: 0:36:18 lr: 0.000004 loss: 0.5406 (0.5178) time: 0.4660 data: 0.0001 max mem: 52150 +Epoch: [24] [2250/7269] eta: 0:36:12 lr: 0.000004 loss: 0.5651 (0.5179) time: 0.4178 data: 0.0002 max mem: 52150 +Epoch: [24] [2260/7269] eta: 0:36:07 lr: 0.000004 loss: 0.5508 (0.5179) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [24] [2270/7269] eta: 0:36:05 lr: 0.000004 loss: 0.5366 (0.5179) time: 0.4704 data: 0.0001 max mem: 52150 +Epoch: [24] [2280/7269] eta: 0:36:00 lr: 0.000004 loss: 0.5184 (0.5179) time: 0.4708 data: 0.0001 max mem: 52150 +Epoch: [24] [2290/7269] eta: 0:35:57 lr: 0.000004 loss: 0.5666 (0.5180) time: 0.4373 data: 0.0001 max mem: 52150 +Epoch: [24] [2300/7269] eta: 0:35:53 lr: 0.000004 loss: 0.5345 (0.5179) time: 0.4811 data: 0.0001 max mem: 52150 +Epoch: [24] [2310/7269] eta: 0:35:50 lr: 0.000004 loss: 0.5049 (0.5179) time: 0.4755 data: 0.0001 max mem: 52150 +Epoch: [24] [2320/7269] eta: 0:35:46 lr: 0.000004 loss: 0.5214 (0.5179) time: 0.4818 data: 0.0001 max mem: 52150 +Epoch: [24] [2330/7269] eta: 0:35:41 lr: 0.000004 loss: 0.5237 (0.5179) time: 0.4220 data: 0.0001 max mem: 52150 +Epoch: [24] [2340/7269] eta: 0:35:35 lr: 0.000004 loss: 0.5371 (0.5179) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [24] [2350/7269] eta: 0:35:32 lr: 0.000004 loss: 0.5266 (0.5178) time: 0.4206 data: 0.0002 max mem: 52150 +Epoch: [24] [2360/7269] eta: 0:35:28 lr: 0.000004 loss: 0.4932 (0.5177) time: 0.4814 data: 0.0001 max mem: 52150 +Epoch: [24] [2370/7269] eta: 0:35:23 lr: 0.000004 loss: 0.4881 (0.5177) time: 0.4328 data: 0.0001 max mem: 52150 +Epoch: [24] [2380/7269] eta: 0:35:20 lr: 0.000004 loss: 0.5508 (0.5178) time: 0.4353 data: 0.0001 max mem: 52150 +Epoch: [24] [2390/7269] eta: 0:35:16 lr: 0.000004 loss: 0.5508 (0.5179) time: 0.4850 data: 0.0001 max mem: 52150 +Epoch: [24] [2400/7269] eta: 0:35:11 lr: 0.000004 loss: 0.5315 (0.5179) time: 0.4217 data: 0.0001 max mem: 52150 +Epoch: [24] [2410/7269] eta: 0:35:07 lr: 0.000004 loss: 0.4791 (0.5178) time: 0.4179 data: 0.0002 max mem: 52150 +Epoch: [24] [2420/7269] eta: 0:35:03 lr: 0.000004 loss: 0.5222 (0.5178) time: 0.4641 data: 0.0001 max mem: 52150 +Epoch: [24] [2430/7269] eta: 0:34:58 lr: 0.000004 loss: 0.5110 (0.5177) time: 0.4183 data: 0.0001 max mem: 52150 +Epoch: [24] [2440/7269] eta: 0:34:52 lr: 0.000004 loss: 0.4802 (0.5176) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [24] [2450/7269] eta: 0:34:47 lr: 0.000004 loss: 0.5091 (0.5175) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [24] [2460/7269] eta: 0:34:41 lr: 0.000004 loss: 0.5293 (0.5177) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [24] [2470/7269] eta: 0:34:36 lr: 0.000004 loss: 0.5360 (0.5177) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [24] [2480/7269] eta: 0:34:30 lr: 0.000004 loss: 0.4999 (0.5178) time: 0.3725 data: 0.0001 max mem: 52150 +Epoch: [24] [2490/7269] eta: 0:34:27 lr: 0.000004 loss: 0.4953 (0.5175) time: 0.4226 data: 0.0001 max mem: 52150 +Epoch: [24] [2500/7269] eta: 0:34:21 lr: 0.000004 loss: 0.4889 (0.5175) time: 0.4221 data: 0.0001 max mem: 52150 +Epoch: [24] [2510/7269] eta: 0:34:16 lr: 0.000004 loss: 0.5083 (0.5174) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [24] [2520/7269] eta: 0:34:10 lr: 0.000004 loss: 0.5083 (0.5173) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [24] [2530/7269] eta: 0:34:05 lr: 0.000004 loss: 0.4988 (0.5173) time: 0.3714 data: 0.0001 max mem: 52150 +Epoch: [24] [2540/7269] eta: 0:33:59 lr: 0.000004 loss: 0.5146 (0.5172) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [24] [2550/7269] eta: 0:33:54 lr: 0.000004 loss: 0.5273 (0.5172) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [24] [2560/7269] eta: 0:33:51 lr: 0.000004 loss: 0.5167 (0.5171) time: 0.4315 data: 0.0001 max mem: 52150 +Epoch: [24] [2570/7269] eta: 0:33:45 lr: 0.000004 loss: 0.5163 (0.5171) time: 0.4315 data: 0.0001 max mem: 52150 +Epoch: [24] [2580/7269] eta: 0:33:42 lr: 0.000004 loss: 0.5184 (0.5171) time: 0.4330 data: 0.0002 max mem: 52150 +Epoch: [24] [2590/7269] eta: 0:33:37 lr: 0.000004 loss: 0.5061 (0.5169) time: 0.4329 data: 0.0002 max mem: 52150 +Epoch: [24] [2600/7269] eta: 0:33:33 lr: 0.000004 loss: 0.5006 (0.5169) time: 0.4326 data: 0.0002 max mem: 52150 +Epoch: [24] [2610/7269] eta: 0:33:31 lr: 0.000004 loss: 0.5312 (0.5170) time: 0.5247 data: 0.0002 max mem: 52150 +Epoch: [24] [2620/7269] eta: 0:33:26 lr: 0.000004 loss: 0.5249 (0.5170) time: 0.4639 data: 0.0001 max mem: 52150 +Epoch: [24] [2630/7269] eta: 0:33:24 lr: 0.000004 loss: 0.4965 (0.5169) time: 0.4765 data: 0.0001 max mem: 52150 +Epoch: [24] [2640/7269] eta: 0:33:22 lr: 0.000004 loss: 0.4958 (0.5168) time: 0.5678 data: 0.0002 max mem: 52150 +Epoch: [24] [2650/7269] eta: 0:33:17 lr: 0.000004 loss: 0.4989 (0.5169) time: 0.4632 data: 0.0002 max mem: 52150 +Epoch: [24] [2660/7269] eta: 0:33:14 lr: 0.000004 loss: 0.5001 (0.5168) time: 0.4326 data: 0.0001 max mem: 52150 +Epoch: [24] [2670/7269] eta: 0:33:10 lr: 0.000004 loss: 0.5022 (0.5168) time: 0.4798 data: 0.0001 max mem: 52150 +Epoch: [24] [2680/7269] eta: 0:33:06 lr: 0.000004 loss: 0.5294 (0.5168) time: 0.4792 data: 0.0001 max mem: 52150 +Epoch: [24] [2690/7269] eta: 0:33:01 lr: 0.000004 loss: 0.5100 (0.5169) time: 0.4323 data: 0.0001 max mem: 52150 +Epoch: [24] [2700/7269] eta: 0:32:57 lr: 0.000004 loss: 0.4946 (0.5168) time: 0.4191 data: 0.0001 max mem: 52150 +Epoch: [24] [2710/7269] eta: 0:32:52 lr: 0.000004 loss: 0.5235 (0.5168) time: 0.4189 data: 0.0002 max mem: 52150 +Epoch: [24] [2720/7269] eta: 0:32:47 lr: 0.000004 loss: 0.5235 (0.5168) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [24] [2730/7269] eta: 0:32:41 lr: 0.000004 loss: 0.5324 (0.5168) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [24] [2740/7269] eta: 0:32:36 lr: 0.000004 loss: 0.5064 (0.5167) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [24] [2750/7269] eta: 0:32:35 lr: 0.000004 loss: 0.5131 (0.5168) time: 0.4945 data: 0.0002 max mem: 52150 +Epoch: [24] [2760/7269] eta: 0:32:31 lr: 0.000004 loss: 0.5698 (0.5171) time: 0.5411 data: 0.0002 max mem: 52150 +Epoch: [24] [2770/7269] eta: 0:32:28 lr: 0.000004 loss: 0.5255 (0.5171) time: 0.4785 data: 0.0002 max mem: 52150 +Epoch: [24] [2780/7269] eta: 0:32:24 lr: 0.000004 loss: 0.5236 (0.5172) time: 0.4797 data: 0.0002 max mem: 52150 +Epoch: [24] [2790/7269] eta: 0:32:18 lr: 0.000004 loss: 0.5682 (0.5173) time: 0.4200 data: 0.0002 max mem: 52150 +Epoch: [24] [2800/7269] eta: 0:32:13 lr: 0.000004 loss: 0.5413 (0.5173) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [24] [2810/7269] eta: 0:32:10 lr: 0.000004 loss: 0.5413 (0.5173) time: 0.4238 data: 0.0001 max mem: 52150 +Epoch: [24] [2820/7269] eta: 0:32:06 lr: 0.000004 loss: 0.5243 (0.5173) time: 0.4726 data: 0.0001 max mem: 52150 +Epoch: [24] [2830/7269] eta: 0:32:02 lr: 0.000004 loss: 0.5271 (0.5175) time: 0.4816 data: 0.0001 max mem: 52150 +Epoch: [24] [2840/7269] eta: 0:31:59 lr: 0.000004 loss: 0.5579 (0.5175) time: 0.4961 data: 0.0002 max mem: 52150 +Epoch: [24] [2850/7269] eta: 0:31:55 lr: 0.000004 loss: 0.5348 (0.5176) time: 0.4817 data: 0.0002 max mem: 52150 +Epoch: [24] [2860/7269] eta: 0:31:50 lr: 0.000004 loss: 0.4872 (0.5174) time: 0.4182 data: 0.0002 max mem: 52150 +Epoch: [24] [2870/7269] eta: 0:31:46 lr: 0.000004 loss: 0.4989 (0.5174) time: 0.4205 data: 0.0002 max mem: 52150 +Epoch: [24] [2880/7269] eta: 0:31:41 lr: 0.000004 loss: 0.5390 (0.5175) time: 0.4204 data: 0.0002 max mem: 52150 +Epoch: [24] [2890/7269] eta: 0:31:36 lr: 0.000004 loss: 0.5142 (0.5175) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [24] [2900/7269] eta: 0:31:30 lr: 0.000004 loss: 0.5134 (0.5175) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [24] [2910/7269] eta: 0:31:25 lr: 0.000004 loss: 0.5591 (0.5176) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [24] [2920/7269] eta: 0:31:21 lr: 0.000004 loss: 0.5448 (0.5177) time: 0.4183 data: 0.0001 max mem: 52150 +Epoch: [24] [2930/7269] eta: 0:31:16 lr: 0.000004 loss: 0.5299 (0.5177) time: 0.4183 data: 0.0001 max mem: 52150 +Epoch: [24] [2940/7269] eta: 0:31:11 lr: 0.000004 loss: 0.5295 (0.5178) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [24] [2950/7269] eta: 0:31:06 lr: 0.000004 loss: 0.5491 (0.5179) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [24] [2960/7269] eta: 0:31:00 lr: 0.000004 loss: 0.5555 (0.5181) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [24] [2970/7269] eta: 0:30:57 lr: 0.000004 loss: 0.5643 (0.5181) time: 0.4206 data: 0.0002 max mem: 52150 +Epoch: [24] [2980/7269] eta: 0:30:51 lr: 0.000004 loss: 0.5261 (0.5182) time: 0.4204 data: 0.0001 max mem: 52150 +Epoch: [24] [2990/7269] eta: 0:30:46 lr: 0.000004 loss: 0.5337 (0.5183) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [24] [3000/7269] eta: 0:30:41 lr: 0.000004 loss: 0.4995 (0.5182) time: 0.3723 data: 0.0001 max mem: 52150 +Epoch: [24] [3010/7269] eta: 0:30:36 lr: 0.000004 loss: 0.5022 (0.5183) time: 0.3723 data: 0.0001 max mem: 52150 +Epoch: [24] [3020/7269] eta: 0:30:31 lr: 0.000004 loss: 0.5418 (0.5184) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [24] [3030/7269] eta: 0:30:26 lr: 0.000004 loss: 0.5467 (0.5185) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [24] [3040/7269] eta: 0:30:21 lr: 0.000004 loss: 0.5428 (0.5185) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [24] [3050/7269] eta: 0:30:15 lr: 0.000004 loss: 0.5485 (0.5186) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [24] [3060/7269] eta: 0:30:10 lr: 0.000004 loss: 0.5282 (0.5186) time: 0.3721 data: 0.0001 max mem: 52150 +Epoch: [24] [3070/7269] eta: 0:30:07 lr: 0.000004 loss: 0.5282 (0.5186) time: 0.4324 data: 0.0001 max mem: 52150 +Epoch: [24] [3080/7269] eta: 0:30:03 lr: 0.000004 loss: 0.5194 (0.5186) time: 0.4782 data: 0.0001 max mem: 52150 +Epoch: [24] [3090/7269] eta: 0:29:58 lr: 0.000004 loss: 0.5130 (0.5185) time: 0.4176 data: 0.0001 max mem: 52150 +Epoch: [24] [3100/7269] eta: 0:29:54 lr: 0.000004 loss: 0.5053 (0.5185) time: 0.4178 data: 0.0001 max mem: 52150 +Epoch: [24] [3110/7269] eta: 0:29:49 lr: 0.000004 loss: 0.5377 (0.5185) time: 0.4180 data: 0.0001 max mem: 52150 +Epoch: [24] [3120/7269] eta: 0:29:44 lr: 0.000004 loss: 0.5256 (0.5184) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [24] [3130/7269] eta: 0:29:39 lr: 0.000004 loss: 0.4952 (0.5184) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [24] [3140/7269] eta: 0:29:34 lr: 0.000004 loss: 0.5152 (0.5185) time: 0.3731 data: 0.0002 max mem: 52150 +Epoch: [24] [3150/7269] eta: 0:29:33 lr: 0.000004 loss: 0.5526 (0.5186) time: 0.5269 data: 0.0002 max mem: 52150 +Epoch: [24] [3160/7269] eta: 0:29:29 lr: 0.000004 loss: 0.4969 (0.5185) time: 0.5767 data: 0.0001 max mem: 52150 +Epoch: [24] [3170/7269] eta: 0:29:24 lr: 0.000004 loss: 0.5132 (0.5186) time: 0.4232 data: 0.0001 max mem: 52150 +Epoch: [24] [3180/7269] eta: 0:29:20 lr: 0.000004 loss: 0.5533 (0.5186) time: 0.4316 data: 0.0001 max mem: 52150 +Epoch: [24] [3190/7269] eta: 0:29:15 lr: 0.000004 loss: 0.5198 (0.5186) time: 0.4309 data: 0.0001 max mem: 52150 +Epoch: [24] [3200/7269] eta: 0:29:13 lr: 0.000004 loss: 0.5258 (0.5186) time: 0.4655 data: 0.0001 max mem: 52150 +Epoch: [24] [3210/7269] eta: 0:29:10 lr: 0.000004 loss: 0.5262 (0.5187) time: 0.5699 data: 0.0001 max mem: 52150 +Epoch: [24] [3220/7269] eta: 0:29:08 lr: 0.000004 loss: 0.4954 (0.5186) time: 0.5810 data: 0.0001 max mem: 52150 +Epoch: [24] [3230/7269] eta: 0:29:03 lr: 0.000004 loss: 0.4924 (0.5186) time: 0.4767 data: 0.0001 max mem: 52150 +Epoch: [24] [3240/7269] eta: 0:28:58 lr: 0.000004 loss: 0.5438 (0.5187) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [24] [3250/7269] eta: 0:28:55 lr: 0.000004 loss: 0.5474 (0.5187) time: 0.4794 data: 0.0002 max mem: 52150 +Epoch: [24] [3260/7269] eta: 0:28:50 lr: 0.000004 loss: 0.5156 (0.5186) time: 0.4793 data: 0.0001 max mem: 52150 +Epoch: [24] [3270/7269] eta: 0:28:45 lr: 0.000004 loss: 0.5358 (0.5186) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [24] [3280/7269] eta: 0:28:40 lr: 0.000004 loss: 0.5314 (0.5185) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [24] [3290/7269] eta: 0:28:35 lr: 0.000004 loss: 0.5314 (0.5185) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [24] [3300/7269] eta: 0:28:32 lr: 0.000004 loss: 0.5620 (0.5187) time: 0.4313 data: 0.0001 max mem: 52150 +Epoch: [24] [3310/7269] eta: 0:28:29 lr: 0.000004 loss: 0.5533 (0.5187) time: 0.5271 data: 0.0001 max mem: 52150 +Epoch: [24] [3320/7269] eta: 0:28:24 lr: 0.000004 loss: 0.5097 (0.5187) time: 0.4676 data: 0.0001 max mem: 52150 +Epoch: [24] [3330/7269] eta: 0:28:20 lr: 0.000004 loss: 0.5039 (0.5187) time: 0.4182 data: 0.0002 max mem: 52150 +Epoch: [24] [3340/7269] eta: 0:28:17 lr: 0.000004 loss: 0.4873 (0.5186) time: 0.5241 data: 0.0001 max mem: 52150 +Epoch: [24] [3350/7269] eta: 0:28:13 lr: 0.000004 loss: 0.4963 (0.5185) time: 0.5235 data: 0.0001 max mem: 52150 +Epoch: [24] [3360/7269] eta: 0:28:08 lr: 0.000004 loss: 0.4963 (0.5184) time: 0.4178 data: 0.0001 max mem: 52150 +Epoch: [24] [3370/7269] eta: 0:28:05 lr: 0.000004 loss: 0.4804 (0.5184) time: 0.4211 data: 0.0001 max mem: 52150 +Epoch: [24] [3380/7269] eta: 0:28:00 lr: 0.000004 loss: 0.5251 (0.5184) time: 0.4211 data: 0.0001 max mem: 52150 +Epoch: [24] [3390/7269] eta: 0:27:55 lr: 0.000004 loss: 0.5109 (0.5183) time: 0.3725 data: 0.0001 max mem: 52150 +Epoch: [24] [3400/7269] eta: 0:27:51 lr: 0.000004 loss: 0.5149 (0.5183) time: 0.4358 data: 0.0001 max mem: 52150 +Epoch: [24] [3410/7269] eta: 0:27:46 lr: 0.000004 loss: 0.5329 (0.5184) time: 0.4361 data: 0.0001 max mem: 52150 +Epoch: [24] [3420/7269] eta: 0:27:42 lr: 0.000004 loss: 0.5443 (0.5184) time: 0.4208 data: 0.0001 max mem: 52150 +Epoch: [24] [3430/7269] eta: 0:27:37 lr: 0.000004 loss: 0.5302 (0.5184) time: 0.4199 data: 0.0002 max mem: 52150 +Epoch: [24] [3440/7269] eta: 0:27:33 lr: 0.000004 loss: 0.5174 (0.5183) time: 0.4316 data: 0.0002 max mem: 52150 +Epoch: [24] [3450/7269] eta: 0:27:32 lr: 0.000004 loss: 0.5240 (0.5183) time: 0.5755 data: 0.0001 max mem: 52150 +Epoch: [24] [3460/7269] eta: 0:27:27 lr: 0.000004 loss: 0.5214 (0.5183) time: 0.5159 data: 0.0001 max mem: 52150 +Epoch: [24] [3470/7269] eta: 0:27:22 lr: 0.000004 loss: 0.5337 (0.5184) time: 0.3723 data: 0.0001 max mem: 52150 +Epoch: [24] [3480/7269] eta: 0:27:17 lr: 0.000004 loss: 0.5230 (0.5184) time: 0.3725 data: 0.0001 max mem: 52150 +Epoch: [24] [3490/7269] eta: 0:27:13 lr: 0.000004 loss: 0.5125 (0.5183) time: 0.4204 data: 0.0001 max mem: 52150 +Epoch: [24] [3500/7269] eta: 0:27:08 lr: 0.000004 loss: 0.5238 (0.5183) time: 0.4204 data: 0.0001 max mem: 52150 +Epoch: [24] [3510/7269] eta: 0:27:03 lr: 0.000004 loss: 0.5231 (0.5183) time: 0.3724 data: 0.0001 max mem: 52150 +Epoch: [24] [3520/7269] eta: 0:26:58 lr: 0.000004 loss: 0.5231 (0.5184) time: 0.3723 data: 0.0001 max mem: 52150 +Epoch: [24] [3530/7269] eta: 0:26:53 lr: 0.000004 loss: 0.5217 (0.5184) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [24] [3540/7269] eta: 0:26:48 lr: 0.000004 loss: 0.5003 (0.5184) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [24] [3550/7269] eta: 0:26:43 lr: 0.000004 loss: 0.5451 (0.5185) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [24] [3560/7269] eta: 0:26:38 lr: 0.000004 loss: 0.5461 (0.5186) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [24] [3570/7269] eta: 0:26:34 lr: 0.000004 loss: 0.5095 (0.5185) time: 0.4319 data: 0.0001 max mem: 52150 +Epoch: [24] [3580/7269] eta: 0:26:30 lr: 0.000004 loss: 0.5054 (0.5186) time: 0.4322 data: 0.0001 max mem: 52150 +Epoch: [24] [3590/7269] eta: 0:26:25 lr: 0.000004 loss: 0.5285 (0.5186) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [24] [3600/7269] eta: 0:26:21 lr: 0.000004 loss: 0.5221 (0.5185) time: 0.4168 data: 0.0002 max mem: 52150 +Epoch: [24] [3610/7269] eta: 0:26:16 lr: 0.000004 loss: 0.5152 (0.5185) time: 0.4171 data: 0.0001 max mem: 52150 +Epoch: [24] [3620/7269] eta: 0:26:12 lr: 0.000004 loss: 0.4914 (0.5184) time: 0.4341 data: 0.0002 max mem: 52150 +Epoch: [24] [3630/7269] eta: 0:26:08 lr: 0.000004 loss: 0.5053 (0.5185) time: 0.4792 data: 0.0002 max mem: 52150 +Epoch: [24] [3640/7269] eta: 0:26:04 lr: 0.000004 loss: 0.5596 (0.5187) time: 0.4630 data: 0.0001 max mem: 52150 +Epoch: [24] [3650/7269] eta: 0:25:59 lr: 0.000004 loss: 0.5450 (0.5186) time: 0.4174 data: 0.0001 max mem: 52150 +Epoch: [24] [3660/7269] eta: 0:25:54 lr: 0.000004 loss: 0.4780 (0.5186) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [24] [3670/7269] eta: 0:25:49 lr: 0.000004 loss: 0.4780 (0.5184) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [24] [3680/7269] eta: 0:25:44 lr: 0.000004 loss: 0.4999 (0.5184) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [24] [3690/7269] eta: 0:25:40 lr: 0.000004 loss: 0.5175 (0.5184) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [24] [3700/7269] eta: 0:25:36 lr: 0.000004 loss: 0.5052 (0.5184) time: 0.4211 data: 0.0002 max mem: 52150 +Epoch: [24] [3710/7269] eta: 0:25:31 lr: 0.000004 loss: 0.5078 (0.5184) time: 0.4212 data: 0.0002 max mem: 52150 +Epoch: [24] [3720/7269] eta: 0:25:27 lr: 0.000004 loss: 0.5259 (0.5183) time: 0.4169 data: 0.0001 max mem: 52150 +Epoch: [24] [3730/7269] eta: 0:25:23 lr: 0.000004 loss: 0.5259 (0.5183) time: 0.4628 data: 0.0002 max mem: 52150 +Epoch: [24] [3740/7269] eta: 0:25:20 lr: 0.000004 loss: 0.5094 (0.5183) time: 0.5258 data: 0.0002 max mem: 52150 +Epoch: [24] [3750/7269] eta: 0:25:15 lr: 0.000004 loss: 0.5094 (0.5183) time: 0.4796 data: 0.0001 max mem: 52150 +Epoch: [24] [3760/7269] eta: 0:25:10 lr: 0.000004 loss: 0.5542 (0.5184) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [24] [3770/7269] eta: 0:25:05 lr: 0.000004 loss: 0.5085 (0.5183) time: 0.3713 data: 0.0001 max mem: 52150 +Epoch: [24] [3780/7269] eta: 0:25:03 lr: 0.000004 loss: 0.4828 (0.5183) time: 0.4827 data: 0.0001 max mem: 52150 +Epoch: [24] [3790/7269] eta: 0:24:58 lr: 0.000004 loss: 0.4840 (0.5183) time: 0.4829 data: 0.0001 max mem: 52150 +Epoch: [24] [3800/7269] eta: 0:24:53 lr: 0.000004 loss: 0.5337 (0.5183) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [24] [3810/7269] eta: 0:24:48 lr: 0.000004 loss: 0.5430 (0.5183) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [24] [3820/7269] eta: 0:24:43 lr: 0.000004 loss: 0.5014 (0.5182) time: 0.3711 data: 0.0001 max mem: 52150 +Epoch: [24] [3830/7269] eta: 0:24:38 lr: 0.000004 loss: 0.4907 (0.5181) time: 0.3711 data: 0.0001 max mem: 52150 +Epoch: [24] [3840/7269] eta: 0:24:35 lr: 0.000004 loss: 0.4909 (0.5181) time: 0.4323 data: 0.0002 max mem: 52150 +Epoch: [24] [3850/7269] eta: 0:24:30 lr: 0.000004 loss: 0.5160 (0.5181) time: 0.4323 data: 0.0002 max mem: 52150 +Epoch: [24] [3860/7269] eta: 0:24:26 lr: 0.000004 loss: 0.5160 (0.5181) time: 0.4302 data: 0.0002 max mem: 52150 +Epoch: [24] [3870/7269] eta: 0:24:22 lr: 0.000004 loss: 0.4998 (0.5181) time: 0.4793 data: 0.0002 max mem: 52150 +Epoch: [24] [3880/7269] eta: 0:24:19 lr: 0.000004 loss: 0.5351 (0.5181) time: 0.5281 data: 0.0002 max mem: 52150 +Epoch: [24] [3890/7269] eta: 0:24:16 lr: 0.000004 loss: 0.5318 (0.5181) time: 0.5866 data: 0.0001 max mem: 52150 +Epoch: [24] [3900/7269] eta: 0:24:11 lr: 0.000004 loss: 0.4977 (0.5180) time: 0.4787 data: 0.0002 max mem: 52150 +Epoch: [24] [3910/7269] eta: 0:24:08 lr: 0.000004 loss: 0.5210 (0.5180) time: 0.4683 data: 0.0002 max mem: 52150 +Epoch: [24] [3920/7269] eta: 0:24:05 lr: 0.000004 loss: 0.5272 (0.5180) time: 0.5593 data: 0.0002 max mem: 52150 +Epoch: [24] [3930/7269] eta: 0:24:01 lr: 0.000004 loss: 0.5293 (0.5181) time: 0.5067 data: 0.0002 max mem: 52150 +Epoch: [24] [3940/7269] eta: 0:23:56 lr: 0.000004 loss: 0.5554 (0.5182) time: 0.4158 data: 0.0002 max mem: 52150 +Epoch: [24] [3950/7269] eta: 0:23:51 lr: 0.000004 loss: 0.5554 (0.5182) time: 0.3713 data: 0.0001 max mem: 52150 +Epoch: [24] [3960/7269] eta: 0:23:46 lr: 0.000004 loss: 0.5146 (0.5182) time: 0.3712 data: 0.0001 max mem: 52150 +Epoch: [24] [3970/7269] eta: 0:23:43 lr: 0.000004 loss: 0.5298 (0.5182) time: 0.4329 data: 0.0001 max mem: 52150 +Epoch: [24] [3980/7269] eta: 0:23:38 lr: 0.000004 loss: 0.5384 (0.5182) time: 0.4333 data: 0.0001 max mem: 52150 +Epoch: [24] [3990/7269] eta: 0:23:34 lr: 0.000004 loss: 0.4488 (0.5181) time: 0.4191 data: 0.0001 max mem: 52150 +Epoch: [24] [4000/7269] eta: 0:23:30 lr: 0.000004 loss: 0.5449 (0.5182) time: 0.4821 data: 0.0001 max mem: 52150 +Epoch: [24] [4010/7269] eta: 0:23:25 lr: 0.000004 loss: 0.5449 (0.5182) time: 0.4346 data: 0.0002 max mem: 52150 +Epoch: [24] [4020/7269] eta: 0:23:20 lr: 0.000004 loss: 0.5207 (0.5184) time: 0.3713 data: 0.0001 max mem: 52150 +Epoch: [24] [4030/7269] eta: 0:23:16 lr: 0.000004 loss: 0.5889 (0.5185) time: 0.4215 data: 0.0001 max mem: 52150 +Epoch: [24] [4040/7269] eta: 0:23:12 lr: 0.000004 loss: 0.5411 (0.5184) time: 0.4701 data: 0.0001 max mem: 52150 +Epoch: [24] [4050/7269] eta: 0:23:08 lr: 0.000004 loss: 0.5102 (0.5184) time: 0.4201 data: 0.0002 max mem: 52150 +Epoch: [24] [4060/7269] eta: 0:23:03 lr: 0.000004 loss: 0.4975 (0.5183) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [24] [4070/7269] eta: 0:22:58 lr: 0.000004 loss: 0.5395 (0.5183) time: 0.3722 data: 0.0001 max mem: 52150 +Epoch: [24] [4080/7269] eta: 0:22:55 lr: 0.000004 loss: 0.5292 (0.5182) time: 0.4838 data: 0.0002 max mem: 52150 +Epoch: [24] [4090/7269] eta: 0:22:51 lr: 0.000004 loss: 0.5191 (0.5182) time: 0.5291 data: 0.0002 max mem: 52150 +Epoch: [24] [4100/7269] eta: 0:22:46 lr: 0.000004 loss: 0.5049 (0.5181) time: 0.4174 data: 0.0001 max mem: 52150 +Epoch: [24] [4110/7269] eta: 0:22:41 lr: 0.000004 loss: 0.5026 (0.5181) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [24] [4120/7269] eta: 0:22:37 lr: 0.000004 loss: 0.5219 (0.5182) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [24] [4130/7269] eta: 0:22:32 lr: 0.000004 loss: 0.5330 (0.5182) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [24] [4140/7269] eta: 0:22:27 lr: 0.000004 loss: 0.4970 (0.5181) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [24] [4150/7269] eta: 0:22:22 lr: 0.000004 loss: 0.4711 (0.5180) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [24] [4160/7269] eta: 0:22:18 lr: 0.000004 loss: 0.4772 (0.5179) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [24] [4170/7269] eta: 0:22:14 lr: 0.000004 loss: 0.4845 (0.5179) time: 0.4196 data: 0.0002 max mem: 52150 +Epoch: [24] [4180/7269] eta: 0:22:09 lr: 0.000004 loss: 0.4845 (0.5179) time: 0.4200 data: 0.0002 max mem: 52150 +Epoch: [24] [4190/7269] eta: 0:22:05 lr: 0.000004 loss: 0.4961 (0.5179) time: 0.4186 data: 0.0001 max mem: 52150 +Epoch: [24] [4200/7269] eta: 0:22:00 lr: 0.000004 loss: 0.5275 (0.5179) time: 0.4185 data: 0.0001 max mem: 52150 +Epoch: [24] [4210/7269] eta: 0:21:55 lr: 0.000004 loss: 0.5200 (0.5178) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [24] [4220/7269] eta: 0:21:51 lr: 0.000004 loss: 0.4944 (0.5178) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [24] [4230/7269] eta: 0:21:46 lr: 0.000004 loss: 0.4867 (0.5176) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [24] [4240/7269] eta: 0:21:43 lr: 0.000004 loss: 0.4637 (0.5176) time: 0.4810 data: 0.0002 max mem: 52150 +Epoch: [24] [4250/7269] eta: 0:21:38 lr: 0.000004 loss: 0.5215 (0.5176) time: 0.4811 data: 0.0002 max mem: 52150 +Epoch: [24] [4260/7269] eta: 0:21:33 lr: 0.000004 loss: 0.5271 (0.5176) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [24] [4270/7269] eta: 0:21:29 lr: 0.000004 loss: 0.5348 (0.5177) time: 0.4159 data: 0.0001 max mem: 52150 +Epoch: [24] [4280/7269] eta: 0:21:25 lr: 0.000004 loss: 0.5385 (0.5177) time: 0.4771 data: 0.0002 max mem: 52150 +Epoch: [24] [4290/7269] eta: 0:21:21 lr: 0.000004 loss: 0.5500 (0.5179) time: 0.4795 data: 0.0002 max mem: 52150 +Epoch: [24] [4300/7269] eta: 0:21:17 lr: 0.000004 loss: 0.5586 (0.5178) time: 0.4635 data: 0.0002 max mem: 52150 +Epoch: [24] [4310/7269] eta: 0:21:13 lr: 0.000004 loss: 0.4994 (0.5178) time: 0.4660 data: 0.0002 max mem: 52150 +Epoch: [24] [4320/7269] eta: 0:21:09 lr: 0.000004 loss: 0.5384 (0.5179) time: 0.4206 data: 0.0001 max mem: 52150 +Epoch: [24] [4330/7269] eta: 0:21:05 lr: 0.000004 loss: 0.5656 (0.5180) time: 0.4323 data: 0.0002 max mem: 52150 +Epoch: [24] [4340/7269] eta: 0:21:01 lr: 0.000004 loss: 0.5562 (0.5180) time: 0.4804 data: 0.0002 max mem: 52150 +Epoch: [24] [4350/7269] eta: 0:20:56 lr: 0.000004 loss: 0.5387 (0.5180) time: 0.4197 data: 0.0001 max mem: 52150 +Epoch: [24] [4360/7269] eta: 0:20:51 lr: 0.000004 loss: 0.5345 (0.5179) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [24] [4370/7269] eta: 0:20:47 lr: 0.000004 loss: 0.5209 (0.5179) time: 0.4322 data: 0.0001 max mem: 52150 +Epoch: [24] [4380/7269] eta: 0:20:43 lr: 0.000004 loss: 0.5229 (0.5180) time: 0.4319 data: 0.0001 max mem: 52150 +Epoch: [24] [4390/7269] eta: 0:20:38 lr: 0.000004 loss: 0.4976 (0.5179) time: 0.3712 data: 0.0001 max mem: 52150 +Epoch: [24] [4400/7269] eta: 0:20:34 lr: 0.000004 loss: 0.5010 (0.5179) time: 0.4322 data: 0.0002 max mem: 52150 +Epoch: [24] [4410/7269] eta: 0:20:31 lr: 0.000004 loss: 0.5299 (0.5178) time: 0.5244 data: 0.0002 max mem: 52150 +Epoch: [24] [4420/7269] eta: 0:20:27 lr: 0.000004 loss: 0.5396 (0.5179) time: 0.5233 data: 0.0002 max mem: 52150 +Epoch: [24] [4430/7269] eta: 0:20:22 lr: 0.000004 loss: 0.5458 (0.5180) time: 0.4314 data: 0.0002 max mem: 52150 +Epoch: [24] [4440/7269] eta: 0:20:18 lr: 0.000004 loss: 0.5398 (0.5180) time: 0.4187 data: 0.0001 max mem: 52150 +Epoch: [24] [4450/7269] eta: 0:20:13 lr: 0.000004 loss: 0.5031 (0.5179) time: 0.4183 data: 0.0001 max mem: 52150 +Epoch: [24] [4460/7269] eta: 0:20:09 lr: 0.000004 loss: 0.4736 (0.5178) time: 0.4332 data: 0.0001 max mem: 52150 +Epoch: [24] [4470/7269] eta: 0:20:05 lr: 0.000004 loss: 0.5144 (0.5178) time: 0.4332 data: 0.0001 max mem: 52150 +Epoch: [24] [4480/7269] eta: 0:20:01 lr: 0.000004 loss: 0.5365 (0.5179) time: 0.4201 data: 0.0002 max mem: 52150 +Epoch: [24] [4490/7269] eta: 0:19:57 lr: 0.000004 loss: 0.5358 (0.5179) time: 0.4807 data: 0.0002 max mem: 52150 +Epoch: [24] [4500/7269] eta: 0:19:53 lr: 0.000004 loss: 0.5358 (0.5179) time: 0.4919 data: 0.0001 max mem: 52150 +Epoch: [24] [4510/7269] eta: 0:19:49 lr: 0.000004 loss: 0.5313 (0.5179) time: 0.4767 data: 0.0001 max mem: 52150 +Epoch: [24] [4520/7269] eta: 0:19:44 lr: 0.000004 loss: 0.5306 (0.5180) time: 0.4165 data: 0.0001 max mem: 52150 +Epoch: [24] [4530/7269] eta: 0:19:40 lr: 0.000004 loss: 0.5221 (0.5180) time: 0.4623 data: 0.0001 max mem: 52150 +Epoch: [24] [4540/7269] eta: 0:19:36 lr: 0.000004 loss: 0.5070 (0.5179) time: 0.5257 data: 0.0001 max mem: 52150 +Epoch: [24] [4550/7269] eta: 0:19:32 lr: 0.000004 loss: 0.5139 (0.5179) time: 0.4837 data: 0.0001 max mem: 52150 +Epoch: [24] [4560/7269] eta: 0:19:28 lr: 0.000004 loss: 0.4892 (0.5179) time: 0.4206 data: 0.0001 max mem: 52150 +Epoch: [24] [4570/7269] eta: 0:19:23 lr: 0.000004 loss: 0.4787 (0.5178) time: 0.3714 data: 0.0001 max mem: 52150 +Epoch: [24] [4580/7269] eta: 0:19:18 lr: 0.000004 loss: 0.4953 (0.5178) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [24] [4590/7269] eta: 0:19:14 lr: 0.000004 loss: 0.5123 (0.5178) time: 0.3721 data: 0.0001 max mem: 52150 +Epoch: [24] [4600/7269] eta: 0:19:09 lr: 0.000004 loss: 0.4927 (0.5177) time: 0.3722 data: 0.0001 max mem: 52150 +Epoch: [24] [4610/7269] eta: 0:19:05 lr: 0.000004 loss: 0.5139 (0.5177) time: 0.4194 data: 0.0001 max mem: 52150 +Epoch: [24] [4620/7269] eta: 0:19:01 lr: 0.000004 loss: 0.5269 (0.5177) time: 0.4645 data: 0.0002 max mem: 52150 +Epoch: [24] [4630/7269] eta: 0:18:56 lr: 0.000004 loss: 0.5042 (0.5177) time: 0.4170 data: 0.0002 max mem: 52150 +Epoch: [24] [4640/7269] eta: 0:18:52 lr: 0.000004 loss: 0.5226 (0.5177) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [24] [4650/7269] eta: 0:18:47 lr: 0.000004 loss: 0.5295 (0.5176) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [24] [4660/7269] eta: 0:18:44 lr: 0.000004 loss: 0.5290 (0.5177) time: 0.4812 data: 0.0002 max mem: 52150 +Epoch: [24] [4670/7269] eta: 0:18:39 lr: 0.000004 loss: 0.5394 (0.5177) time: 0.4805 data: 0.0002 max mem: 52150 +Epoch: [24] [4680/7269] eta: 0:18:35 lr: 0.000004 loss: 0.4537 (0.5175) time: 0.4194 data: 0.0002 max mem: 52150 +Epoch: [24] [4690/7269] eta: 0:18:30 lr: 0.000004 loss: 0.4593 (0.5175) time: 0.4197 data: 0.0001 max mem: 52150 +Epoch: [24] [4700/7269] eta: 0:18:26 lr: 0.000004 loss: 0.5268 (0.5175) time: 0.4336 data: 0.0001 max mem: 52150 +Epoch: [24] [4710/7269] eta: 0:18:22 lr: 0.000004 loss: 0.5127 (0.5175) time: 0.4332 data: 0.0001 max mem: 52150 +Epoch: [24] [4720/7269] eta: 0:18:17 lr: 0.000004 loss: 0.5142 (0.5175) time: 0.4182 data: 0.0002 max mem: 52150 +Epoch: [24] [4730/7269] eta: 0:18:13 lr: 0.000004 loss: 0.5142 (0.5175) time: 0.4186 data: 0.0002 max mem: 52150 +Epoch: [24] [4740/7269] eta: 0:18:09 lr: 0.000004 loss: 0.5434 (0.5176) time: 0.4186 data: 0.0001 max mem: 52150 +Epoch: [24] [4750/7269] eta: 0:18:04 lr: 0.000004 loss: 0.5351 (0.5176) time: 0.4185 data: 0.0001 max mem: 52150 +Epoch: [24] [4760/7269] eta: 0:18:00 lr: 0.000004 loss: 0.5163 (0.5176) time: 0.4320 data: 0.0001 max mem: 52150 +Epoch: [24] [4770/7269] eta: 0:17:55 lr: 0.000004 loss: 0.4987 (0.5176) time: 0.4325 data: 0.0001 max mem: 52150 +Epoch: [24] [4780/7269] eta: 0:17:51 lr: 0.000004 loss: 0.5427 (0.5176) time: 0.4191 data: 0.0001 max mem: 52150 +Epoch: [24] [4790/7269] eta: 0:17:48 lr: 0.000004 loss: 0.5532 (0.5176) time: 0.5399 data: 0.0001 max mem: 52150 +Epoch: [24] [4800/7269] eta: 0:17:44 lr: 0.000004 loss: 0.5143 (0.5176) time: 0.5548 data: 0.0001 max mem: 52150 +Epoch: [24] [4810/7269] eta: 0:17:40 lr: 0.000004 loss: 0.4965 (0.5176) time: 0.4803 data: 0.0001 max mem: 52150 +Epoch: [24] [4820/7269] eta: 0:17:35 lr: 0.000004 loss: 0.5087 (0.5176) time: 0.4189 data: 0.0001 max mem: 52150 +Epoch: [24] [4830/7269] eta: 0:17:31 lr: 0.000004 loss: 0.5043 (0.5175) time: 0.3729 data: 0.0002 max mem: 52150 +Epoch: [24] [4840/7269] eta: 0:17:26 lr: 0.000004 loss: 0.5104 (0.5176) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [24] [4850/7269] eta: 0:17:21 lr: 0.000004 loss: 0.5405 (0.5176) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [24] [4860/7269] eta: 0:17:17 lr: 0.000004 loss: 0.5145 (0.5175) time: 0.4318 data: 0.0001 max mem: 52150 +Epoch: [24] [4870/7269] eta: 0:17:13 lr: 0.000004 loss: 0.5115 (0.5176) time: 0.4319 data: 0.0001 max mem: 52150 +Epoch: [24] [4880/7269] eta: 0:17:09 lr: 0.000004 loss: 0.5472 (0.5176) time: 0.4209 data: 0.0001 max mem: 52150 +Epoch: [24] [4890/7269] eta: 0:17:04 lr: 0.000004 loss: 0.5231 (0.5176) time: 0.4208 data: 0.0001 max mem: 52150 +Epoch: [24] [4900/7269] eta: 0:17:00 lr: 0.000004 loss: 0.4851 (0.5176) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [24] [4910/7269] eta: 0:16:55 lr: 0.000004 loss: 0.5164 (0.5176) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [24] [4920/7269] eta: 0:16:51 lr: 0.000004 loss: 0.5323 (0.5176) time: 0.4318 data: 0.0001 max mem: 52150 +Epoch: [24] [4930/7269] eta: 0:16:46 lr: 0.000004 loss: 0.5381 (0.5176) time: 0.4314 data: 0.0001 max mem: 52150 +Epoch: [24] [4940/7269] eta: 0:16:42 lr: 0.000004 loss: 0.5393 (0.5176) time: 0.4224 data: 0.0001 max mem: 52150 +Epoch: [24] [4950/7269] eta: 0:16:38 lr: 0.000004 loss: 0.5453 (0.5176) time: 0.4230 data: 0.0001 max mem: 52150 +Epoch: [24] [4960/7269] eta: 0:16:34 lr: 0.000004 loss: 0.5254 (0.5177) time: 0.4320 data: 0.0002 max mem: 52150 +Epoch: [24] [4970/7269] eta: 0:16:30 lr: 0.000004 loss: 0.5391 (0.5177) time: 0.5407 data: 0.0002 max mem: 52150 +Epoch: [24] [4980/7269] eta: 0:16:26 lr: 0.000004 loss: 0.5420 (0.5178) time: 0.5320 data: 0.0001 max mem: 52150 +Epoch: [24] [4990/7269] eta: 0:16:21 lr: 0.000004 loss: 0.5266 (0.5178) time: 0.4229 data: 0.0001 max mem: 52150 +Epoch: [24] [5000/7269] eta: 0:16:18 lr: 0.000004 loss: 0.5266 (0.5178) time: 0.4629 data: 0.0001 max mem: 52150 +Epoch: [24] [5010/7269] eta: 0:16:14 lr: 0.000004 loss: 0.5200 (0.5178) time: 0.5267 data: 0.0001 max mem: 52150 +Epoch: [24] [5020/7269] eta: 0:16:10 lr: 0.000004 loss: 0.5107 (0.5178) time: 0.4954 data: 0.0001 max mem: 52150 +Epoch: [24] [5030/7269] eta: 0:16:05 lr: 0.000004 loss: 0.5231 (0.5179) time: 0.4315 data: 0.0001 max mem: 52150 +Epoch: [24] [5040/7269] eta: 0:16:00 lr: 0.000004 loss: 0.5231 (0.5178) time: 0.3714 data: 0.0001 max mem: 52150 +Epoch: [24] [5050/7269] eta: 0:15:56 lr: 0.000004 loss: 0.5290 (0.5178) time: 0.4355 data: 0.0001 max mem: 52150 +Epoch: [24] [5060/7269] eta: 0:15:52 lr: 0.000004 loss: 0.5290 (0.5178) time: 0.4358 data: 0.0001 max mem: 52150 +Epoch: [24] [5070/7269] eta: 0:15:47 lr: 0.000004 loss: 0.5159 (0.5178) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [24] [5080/7269] eta: 0:15:43 lr: 0.000004 loss: 0.5214 (0.5178) time: 0.4166 data: 0.0001 max mem: 52150 +Epoch: [24] [5090/7269] eta: 0:15:39 lr: 0.000004 loss: 0.5229 (0.5178) time: 0.4168 data: 0.0001 max mem: 52150 +Epoch: [24] [5100/7269] eta: 0:15:34 lr: 0.000004 loss: 0.5287 (0.5178) time: 0.4174 data: 0.0001 max mem: 52150 +Epoch: [24] [5110/7269] eta: 0:15:30 lr: 0.000004 loss: 0.5317 (0.5178) time: 0.4179 data: 0.0002 max mem: 52150 +Epoch: [24] [5120/7269] eta: 0:15:26 lr: 0.000004 loss: 0.5328 (0.5179) time: 0.4323 data: 0.0002 max mem: 52150 +Epoch: [24] [5130/7269] eta: 0:15:21 lr: 0.000004 loss: 0.5383 (0.5179) time: 0.4319 data: 0.0002 max mem: 52150 +Epoch: [24] [5140/7269] eta: 0:15:17 lr: 0.000004 loss: 0.5238 (0.5179) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [24] [5150/7269] eta: 0:15:12 lr: 0.000004 loss: 0.5332 (0.5179) time: 0.4169 data: 0.0001 max mem: 52150 +Epoch: [24] [5160/7269] eta: 0:15:08 lr: 0.000004 loss: 0.5284 (0.5178) time: 0.4168 data: 0.0001 max mem: 52150 +Epoch: [24] [5170/7269] eta: 0:15:03 lr: 0.000004 loss: 0.5107 (0.5178) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [24] [5180/7269] eta: 0:15:00 lr: 0.000004 loss: 0.5501 (0.5179) time: 0.4644 data: 0.0002 max mem: 52150 +Epoch: [24] [5190/7269] eta: 0:14:56 lr: 0.000004 loss: 0.5402 (0.5178) time: 0.5606 data: 0.0002 max mem: 52150 +Epoch: [24] [5200/7269] eta: 0:14:51 lr: 0.000004 loss: 0.5123 (0.5178) time: 0.4675 data: 0.0002 max mem: 52150 +Epoch: [24] [5210/7269] eta: 0:14:47 lr: 0.000004 loss: 0.5143 (0.5178) time: 0.3711 data: 0.0001 max mem: 52150 +Epoch: [24] [5220/7269] eta: 0:14:42 lr: 0.000004 loss: 0.5105 (0.5178) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [24] [5230/7269] eta: 0:14:38 lr: 0.000004 loss: 0.5105 (0.5178) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [24] [5240/7269] eta: 0:14:33 lr: 0.000004 loss: 0.5173 (0.5178) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [24] [5250/7269] eta: 0:14:29 lr: 0.000004 loss: 0.5173 (0.5178) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [24] [5260/7269] eta: 0:14:24 lr: 0.000004 loss: 0.5319 (0.5179) time: 0.4206 data: 0.0001 max mem: 52150 +Epoch: [24] [5270/7269] eta: 0:14:20 lr: 0.000004 loss: 0.5371 (0.5178) time: 0.4203 data: 0.0001 max mem: 52150 +Epoch: [24] [5280/7269] eta: 0:14:16 lr: 0.000004 loss: 0.5438 (0.5178) time: 0.4361 data: 0.0001 max mem: 52150 +Epoch: [24] [5290/7269] eta: 0:14:11 lr: 0.000004 loss: 0.5550 (0.5179) time: 0.4363 data: 0.0001 max mem: 52150 +Epoch: [24] [5300/7269] eta: 0:14:07 lr: 0.000004 loss: 0.5550 (0.5179) time: 0.4321 data: 0.0001 max mem: 52150 +Epoch: [24] [5310/7269] eta: 0:14:03 lr: 0.000004 loss: 0.5447 (0.5179) time: 0.4788 data: 0.0001 max mem: 52150 +Epoch: [24] [5320/7269] eta: 0:13:59 lr: 0.000004 loss: 0.5373 (0.5179) time: 0.4642 data: 0.0001 max mem: 52150 +Epoch: [24] [5330/7269] eta: 0:13:55 lr: 0.000004 loss: 0.4902 (0.5179) time: 0.4781 data: 0.0002 max mem: 52150 +Epoch: [24] [5340/7269] eta: 0:13:51 lr: 0.000004 loss: 0.4882 (0.5178) time: 0.4932 data: 0.0002 max mem: 52150 +Epoch: [24] [5350/7269] eta: 0:13:47 lr: 0.000004 loss: 0.4997 (0.5178) time: 0.5252 data: 0.0002 max mem: 52150 +Epoch: [24] [5360/7269] eta: 0:13:43 lr: 0.000004 loss: 0.5285 (0.5178) time: 0.5261 data: 0.0002 max mem: 52150 +Epoch: [24] [5370/7269] eta: 0:13:39 lr: 0.000004 loss: 0.5106 (0.5178) time: 0.4786 data: 0.0001 max mem: 52150 +Epoch: [24] [5380/7269] eta: 0:13:34 lr: 0.000004 loss: 0.4855 (0.5177) time: 0.4168 data: 0.0001 max mem: 52150 +Epoch: [24] [5390/7269] eta: 0:13:30 lr: 0.000004 loss: 0.4953 (0.5177) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [24] [5400/7269] eta: 0:13:26 lr: 0.000004 loss: 0.4911 (0.5177) time: 0.5509 data: 0.0002 max mem: 52150 +Epoch: [24] [5410/7269] eta: 0:13:22 lr: 0.000004 loss: 0.5009 (0.5177) time: 0.5510 data: 0.0001 max mem: 52150 +Epoch: [24] [5420/7269] eta: 0:13:17 lr: 0.000004 loss: 0.5009 (0.5177) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [24] [5430/7269] eta: 0:13:13 lr: 0.000004 loss: 0.5135 (0.5177) time: 0.3711 data: 0.0001 max mem: 52150 +Epoch: [24] [5440/7269] eta: 0:13:08 lr: 0.000004 loss: 0.5573 (0.5178) time: 0.3712 data: 0.0001 max mem: 52150 +Epoch: [24] [5450/7269] eta: 0:13:04 lr: 0.000004 loss: 0.5573 (0.5178) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [24] [5460/7269] eta: 0:12:59 lr: 0.000004 loss: 0.4813 (0.5177) time: 0.4180 data: 0.0002 max mem: 52150 +Epoch: [24] [5470/7269] eta: 0:12:55 lr: 0.000004 loss: 0.5280 (0.5178) time: 0.4622 data: 0.0001 max mem: 52150 +Epoch: [24] [5480/7269] eta: 0:12:51 lr: 0.000004 loss: 0.5469 (0.5178) time: 0.4160 data: 0.0001 max mem: 52150 +Epoch: [24] [5490/7269] eta: 0:12:46 lr: 0.000004 loss: 0.5236 (0.5178) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [24] [5500/7269] eta: 0:12:42 lr: 0.000004 loss: 0.5236 (0.5177) time: 0.4309 data: 0.0002 max mem: 52150 +Epoch: [24] [5510/7269] eta: 0:12:38 lr: 0.000004 loss: 0.5222 (0.5177) time: 0.4310 data: 0.0001 max mem: 52150 +Epoch: [24] [5520/7269] eta: 0:12:33 lr: 0.000004 loss: 0.5258 (0.5177) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [24] [5530/7269] eta: 0:12:29 lr: 0.000004 loss: 0.5617 (0.5178) time: 0.4197 data: 0.0002 max mem: 52150 +Epoch: [24] [5540/7269] eta: 0:12:24 lr: 0.000004 loss: 0.5621 (0.5179) time: 0.4197 data: 0.0001 max mem: 52150 +Epoch: [24] [5550/7269] eta: 0:12:20 lr: 0.000004 loss: 0.5621 (0.5179) time: 0.4191 data: 0.0001 max mem: 52150 +Epoch: [24] [5560/7269] eta: 0:12:16 lr: 0.000004 loss: 0.5278 (0.5179) time: 0.4191 data: 0.0002 max mem: 52150 +Epoch: [24] [5570/7269] eta: 0:12:12 lr: 0.000004 loss: 0.5101 (0.5179) time: 0.4168 data: 0.0002 max mem: 52150 +Epoch: [24] [5580/7269] eta: 0:12:07 lr: 0.000004 loss: 0.5177 (0.5178) time: 0.4168 data: 0.0002 max mem: 52150 +Epoch: [24] [5590/7269] eta: 0:12:03 lr: 0.000004 loss: 0.5177 (0.5178) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [24] [5600/7269] eta: 0:11:58 lr: 0.000004 loss: 0.5307 (0.5179) time: 0.3722 data: 0.0001 max mem: 52150 +Epoch: [24] [5610/7269] eta: 0:11:54 lr: 0.000004 loss: 0.5293 (0.5179) time: 0.4942 data: 0.0001 max mem: 52150 +Epoch: [24] [5620/7269] eta: 0:11:50 lr: 0.000004 loss: 0.5262 (0.5179) time: 0.6011 data: 0.0002 max mem: 52150 +Epoch: [24] [5630/7269] eta: 0:11:46 lr: 0.000004 loss: 0.5514 (0.5180) time: 0.5241 data: 0.0002 max mem: 52150 +Epoch: [24] [5640/7269] eta: 0:11:42 lr: 0.000004 loss: 0.5375 (0.5180) time: 0.4801 data: 0.0002 max mem: 52150 +Epoch: [24] [5650/7269] eta: 0:11:38 lr: 0.000004 loss: 0.4999 (0.5180) time: 0.4827 data: 0.0001 max mem: 52150 +Epoch: [24] [5660/7269] eta: 0:11:33 lr: 0.000004 loss: 0.4925 (0.5179) time: 0.4200 data: 0.0002 max mem: 52150 +Epoch: [24] [5670/7269] eta: 0:11:29 lr: 0.000004 loss: 0.4602 (0.5179) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [24] [5680/7269] eta: 0:11:24 lr: 0.000004 loss: 0.5140 (0.5179) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [24] [5690/7269] eta: 0:11:20 lr: 0.000004 loss: 0.5038 (0.5178) time: 0.4318 data: 0.0001 max mem: 52150 +Epoch: [24] [5700/7269] eta: 0:11:16 lr: 0.000004 loss: 0.5038 (0.5179) time: 0.4775 data: 0.0001 max mem: 52150 +Epoch: [24] [5710/7269] eta: 0:11:12 lr: 0.000004 loss: 0.5510 (0.5179) time: 0.4179 data: 0.0002 max mem: 52150 +Epoch: [24] [5720/7269] eta: 0:11:07 lr: 0.000004 loss: 0.5093 (0.5178) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [24] [5730/7269] eta: 0:11:03 lr: 0.000004 loss: 0.5044 (0.5178) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [24] [5740/7269] eta: 0:10:58 lr: 0.000004 loss: 0.5174 (0.5179) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [24] [5750/7269] eta: 0:10:54 lr: 0.000004 loss: 0.5292 (0.5179) time: 0.4211 data: 0.0002 max mem: 52150 +Epoch: [24] [5760/7269] eta: 0:10:50 lr: 0.000004 loss: 0.5005 (0.5178) time: 0.4209 data: 0.0001 max mem: 52150 +Epoch: [24] [5770/7269] eta: 0:10:46 lr: 0.000004 loss: 0.5085 (0.5178) time: 0.4618 data: 0.0002 max mem: 52150 +Epoch: [24] [5780/7269] eta: 0:10:41 lr: 0.000004 loss: 0.4790 (0.5178) time: 0.4617 data: 0.0002 max mem: 52150 +Epoch: [24] [5790/7269] eta: 0:10:37 lr: 0.000004 loss: 0.4892 (0.5178) time: 0.4182 data: 0.0001 max mem: 52150 +Epoch: [24] [5800/7269] eta: 0:10:33 lr: 0.000004 loss: 0.5339 (0.5178) time: 0.4795 data: 0.0001 max mem: 52150 +Epoch: [24] [5810/7269] eta: 0:10:28 lr: 0.000004 loss: 0.5311 (0.5178) time: 0.4325 data: 0.0001 max mem: 52150 +Epoch: [24] [5820/7269] eta: 0:10:24 lr: 0.000004 loss: 0.5311 (0.5178) time: 0.4197 data: 0.0001 max mem: 52150 +Epoch: [24] [5830/7269] eta: 0:10:20 lr: 0.000004 loss: 0.5605 (0.5178) time: 0.4202 data: 0.0001 max mem: 52150 +Epoch: [24] [5840/7269] eta: 0:10:15 lr: 0.000004 loss: 0.5149 (0.5178) time: 0.4345 data: 0.0001 max mem: 52150 +Epoch: [24] [5850/7269] eta: 0:10:11 lr: 0.000004 loss: 0.5256 (0.5178) time: 0.4342 data: 0.0001 max mem: 52150 +Epoch: [24] [5860/7269] eta: 0:10:07 lr: 0.000004 loss: 0.5418 (0.5178) time: 0.4332 data: 0.0001 max mem: 52150 +Epoch: [24] [5870/7269] eta: 0:10:03 lr: 0.000004 loss: 0.4812 (0.5177) time: 0.4924 data: 0.0001 max mem: 52150 +Epoch: [24] [5880/7269] eta: 0:09:58 lr: 0.000004 loss: 0.4750 (0.5177) time: 0.4308 data: 0.0001 max mem: 52150 +Epoch: [24] [5890/7269] eta: 0:09:54 lr: 0.000004 loss: 0.5249 (0.5177) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [24] [5900/7269] eta: 0:09:49 lr: 0.000004 loss: 0.5514 (0.5178) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [24] [5910/7269] eta: 0:09:45 lr: 0.000004 loss: 0.5409 (0.5177) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [24] [5920/7269] eta: 0:09:41 lr: 0.000004 loss: 0.5114 (0.5177) time: 0.4176 data: 0.0001 max mem: 52150 +Epoch: [24] [5930/7269] eta: 0:09:36 lr: 0.000004 loss: 0.5117 (0.5176) time: 0.4649 data: 0.0001 max mem: 52150 +Epoch: [24] [5940/7269] eta: 0:09:32 lr: 0.000004 loss: 0.5290 (0.5176) time: 0.4789 data: 0.0001 max mem: 52150 +Epoch: [24] [5950/7269] eta: 0:09:28 lr: 0.000004 loss: 0.5215 (0.5176) time: 0.4317 data: 0.0001 max mem: 52150 +Epoch: [24] [5960/7269] eta: 0:09:23 lr: 0.000004 loss: 0.5350 (0.5177) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [24] [5970/7269] eta: 0:09:19 lr: 0.000004 loss: 0.5508 (0.5177) time: 0.4309 data: 0.0002 max mem: 52150 +Epoch: [24] [5980/7269] eta: 0:09:15 lr: 0.000004 loss: 0.5249 (0.5177) time: 0.4313 data: 0.0001 max mem: 52150 +Epoch: [24] [5990/7269] eta: 0:09:11 lr: 0.000004 loss: 0.5529 (0.5178) time: 0.4610 data: 0.0002 max mem: 52150 +Epoch: [24] [6000/7269] eta: 0:09:06 lr: 0.000004 loss: 0.4901 (0.5177) time: 0.4605 data: 0.0002 max mem: 52150 +Epoch: [24] [6010/7269] eta: 0:09:02 lr: 0.000004 loss: 0.5214 (0.5177) time: 0.4300 data: 0.0002 max mem: 52150 +Epoch: [24] [6020/7269] eta: 0:08:58 lr: 0.000004 loss: 0.5304 (0.5178) time: 0.4919 data: 0.0002 max mem: 52150 +Epoch: [24] [6030/7269] eta: 0:08:53 lr: 0.000004 loss: 0.5316 (0.5178) time: 0.4332 data: 0.0001 max mem: 52150 +Epoch: [24] [6040/7269] eta: 0:08:49 lr: 0.000004 loss: 0.5349 (0.5178) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [24] [6050/7269] eta: 0:08:45 lr: 0.000004 loss: 0.5356 (0.5178) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [24] [6060/7269] eta: 0:08:40 lr: 0.000004 loss: 0.5354 (0.5178) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [24] [6070/7269] eta: 0:08:36 lr: 0.000004 loss: 0.5318 (0.5178) time: 0.4350 data: 0.0002 max mem: 52150 +Epoch: [24] [6080/7269] eta: 0:08:32 lr: 0.000004 loss: 0.5038 (0.5178) time: 0.5436 data: 0.0002 max mem: 52150 +Epoch: [24] [6090/7269] eta: 0:08:28 lr: 0.000004 loss: 0.4629 (0.5177) time: 0.4799 data: 0.0002 max mem: 52150 +Epoch: [24] [6100/7269] eta: 0:08:23 lr: 0.000004 loss: 0.4629 (0.5177) time: 0.3726 data: 0.0001 max mem: 52150 +Epoch: [24] [6110/7269] eta: 0:08:19 lr: 0.000004 loss: 0.4699 (0.5176) time: 0.3728 data: 0.0001 max mem: 52150 +Epoch: [24] [6120/7269] eta: 0:08:15 lr: 0.000004 loss: 0.5061 (0.5176) time: 0.4355 data: 0.0001 max mem: 52150 +Epoch: [24] [6130/7269] eta: 0:08:10 lr: 0.000004 loss: 0.5327 (0.5176) time: 0.4351 data: 0.0001 max mem: 52150 +Epoch: [24] [6140/7269] eta: 0:08:06 lr: 0.000004 loss: 0.5058 (0.5175) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [24] [6150/7269] eta: 0:08:01 lr: 0.000004 loss: 0.5060 (0.5175) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [24] [6160/7269] eta: 0:07:57 lr: 0.000004 loss: 0.5183 (0.5175) time: 0.4178 data: 0.0001 max mem: 52150 +Epoch: [24] [6170/7269] eta: 0:07:53 lr: 0.000004 loss: 0.5299 (0.5175) time: 0.5233 data: 0.0001 max mem: 52150 +Epoch: [24] [6180/7269] eta: 0:07:49 lr: 0.000004 loss: 0.5470 (0.5175) time: 0.4769 data: 0.0001 max mem: 52150 +Epoch: [24] [6190/7269] eta: 0:07:44 lr: 0.000004 loss: 0.5026 (0.5175) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [24] [6200/7269] eta: 0:07:40 lr: 0.000004 loss: 0.5345 (0.5176) time: 0.4174 data: 0.0002 max mem: 52150 +Epoch: [24] [6210/7269] eta: 0:07:36 lr: 0.000004 loss: 0.5540 (0.5176) time: 0.4179 data: 0.0001 max mem: 52150 +Epoch: [24] [6220/7269] eta: 0:07:31 lr: 0.000004 loss: 0.5430 (0.5176) time: 0.4176 data: 0.0001 max mem: 52150 +Epoch: [24] [6230/7269] eta: 0:07:27 lr: 0.000004 loss: 0.5192 (0.5176) time: 0.4174 data: 0.0001 max mem: 52150 +Epoch: [24] [6240/7269] eta: 0:07:23 lr: 0.000004 loss: 0.5135 (0.5176) time: 0.4325 data: 0.0001 max mem: 52150 +Epoch: [24] [6250/7269] eta: 0:07:18 lr: 0.000004 loss: 0.5338 (0.5176) time: 0.4782 data: 0.0001 max mem: 52150 +Epoch: [24] [6260/7269] eta: 0:07:14 lr: 0.000004 loss: 0.5338 (0.5176) time: 0.4628 data: 0.0001 max mem: 52150 +Epoch: [24] [6270/7269] eta: 0:07:10 lr: 0.000004 loss: 0.5451 (0.5177) time: 0.4174 data: 0.0002 max mem: 52150 +Epoch: [24] [6280/7269] eta: 0:07:05 lr: 0.000004 loss: 0.5493 (0.5177) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [24] [6290/7269] eta: 0:07:01 lr: 0.000004 loss: 0.5318 (0.5178) time: 0.3713 data: 0.0001 max mem: 52150 +Epoch: [24] [6300/7269] eta: 0:06:57 lr: 0.000004 loss: 0.5387 (0.5178) time: 0.4183 data: 0.0001 max mem: 52150 +Epoch: [24] [6310/7269] eta: 0:06:52 lr: 0.000004 loss: 0.5430 (0.5178) time: 0.4185 data: 0.0001 max mem: 52150 +Epoch: [24] [6320/7269] eta: 0:06:48 lr: 0.000004 loss: 0.5285 (0.5178) time: 0.4342 data: 0.0001 max mem: 52150 +Epoch: [24] [6330/7269] eta: 0:06:44 lr: 0.000004 loss: 0.5434 (0.5178) time: 0.4970 data: 0.0001 max mem: 52150 +Epoch: [24] [6340/7269] eta: 0:06:40 lr: 0.000004 loss: 0.5506 (0.5179) time: 0.4346 data: 0.0001 max mem: 52150 +Epoch: [24] [6350/7269] eta: 0:06:35 lr: 0.000004 loss: 0.5049 (0.5179) time: 0.4334 data: 0.0001 max mem: 52150 +Epoch: [24] [6360/7269] eta: 0:06:31 lr: 0.000004 loss: 0.5283 (0.5179) time: 0.4337 data: 0.0001 max mem: 52150 +Epoch: [24] [6370/7269] eta: 0:06:27 lr: 0.000004 loss: 0.4887 (0.5178) time: 0.3721 data: 0.0001 max mem: 52150 +Epoch: [24] [6380/7269] eta: 0:06:22 lr: 0.000004 loss: 0.4777 (0.5178) time: 0.4207 data: 0.0002 max mem: 52150 +Epoch: [24] [6390/7269] eta: 0:06:18 lr: 0.000004 loss: 0.5140 (0.5178) time: 0.4669 data: 0.0002 max mem: 52150 +Epoch: [24] [6400/7269] eta: 0:06:14 lr: 0.000004 loss: 0.5140 (0.5178) time: 0.6191 data: 0.0001 max mem: 52150 +Epoch: [24] [6410/7269] eta: 0:06:10 lr: 0.000004 loss: 0.5112 (0.5178) time: 0.5720 data: 0.0001 max mem: 52150 +Epoch: [24] [6420/7269] eta: 0:06:05 lr: 0.000004 loss: 0.5218 (0.5178) time: 0.3708 data: 0.0001 max mem: 52150 +Epoch: [24] [6430/7269] eta: 0:06:01 lr: 0.000004 loss: 0.5218 (0.5178) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [24] [6440/7269] eta: 0:05:57 lr: 0.000004 loss: 0.5076 (0.5178) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [24] [6450/7269] eta: 0:05:52 lr: 0.000004 loss: 0.5004 (0.5178) time: 0.3712 data: 0.0001 max mem: 52150 +Epoch: [24] [6460/7269] eta: 0:05:48 lr: 0.000004 loss: 0.5004 (0.5178) time: 0.4205 data: 0.0001 max mem: 52150 +Epoch: [24] [6470/7269] eta: 0:05:44 lr: 0.000004 loss: 0.4873 (0.5177) time: 0.4205 data: 0.0001 max mem: 52150 +Epoch: [24] [6480/7269] eta: 0:05:39 lr: 0.000004 loss: 0.5052 (0.5177) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [24] [6490/7269] eta: 0:05:35 lr: 0.000004 loss: 0.5163 (0.5177) time: 0.4930 data: 0.0001 max mem: 52150 +Epoch: [24] [6500/7269] eta: 0:05:31 lr: 0.000004 loss: 0.5381 (0.5177) time: 0.4927 data: 0.0001 max mem: 52150 +Epoch: [24] [6510/7269] eta: 0:05:26 lr: 0.000004 loss: 0.5060 (0.5176) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [24] [6520/7269] eta: 0:05:22 lr: 0.000004 loss: 0.5266 (0.5177) time: 0.4823 data: 0.0002 max mem: 52150 +Epoch: [24] [6530/7269] eta: 0:05:18 lr: 0.000004 loss: 0.5543 (0.5177) time: 0.4824 data: 0.0002 max mem: 52150 +Epoch: [24] [6540/7269] eta: 0:05:14 lr: 0.000004 loss: 0.5446 (0.5177) time: 0.3712 data: 0.0002 max mem: 52150 +Epoch: [24] [6550/7269] eta: 0:05:09 lr: 0.000004 loss: 0.5446 (0.5177) time: 0.4334 data: 0.0002 max mem: 52150 +Epoch: [24] [6560/7269] eta: 0:05:05 lr: 0.000004 loss: 0.5556 (0.5178) time: 0.4339 data: 0.0002 max mem: 52150 +Epoch: [24] [6570/7269] eta: 0:05:01 lr: 0.000004 loss: 0.5425 (0.5178) time: 0.4321 data: 0.0002 max mem: 52150 +Epoch: [24] [6580/7269] eta: 0:04:56 lr: 0.000004 loss: 0.5601 (0.5178) time: 0.4320 data: 0.0002 max mem: 52150 +Epoch: [24] [6590/7269] eta: 0:04:52 lr: 0.000004 loss: 0.5510 (0.5178) time: 0.4344 data: 0.0001 max mem: 52150 +Epoch: [24] [6600/7269] eta: 0:04:48 lr: 0.000004 loss: 0.5194 (0.5178) time: 0.4346 data: 0.0001 max mem: 52150 +Epoch: [24] [6610/7269] eta: 0:04:43 lr: 0.000004 loss: 0.5013 (0.5178) time: 0.4180 data: 0.0001 max mem: 52150 +Epoch: [24] [6620/7269] eta: 0:04:39 lr: 0.000004 loss: 0.5013 (0.5178) time: 0.4771 data: 0.0001 max mem: 52150 +Epoch: [24] [6630/7269] eta: 0:04:35 lr: 0.000004 loss: 0.5300 (0.5178) time: 0.4314 data: 0.0001 max mem: 52150 +Epoch: [24] [6640/7269] eta: 0:04:31 lr: 0.000004 loss: 0.4679 (0.5177) time: 0.4325 data: 0.0002 max mem: 52150 +Epoch: [24] [6650/7269] eta: 0:04:26 lr: 0.000004 loss: 0.5041 (0.5177) time: 0.4772 data: 0.0001 max mem: 52150 +Epoch: [24] [6660/7269] eta: 0:04:22 lr: 0.000004 loss: 0.5416 (0.5177) time: 0.4769 data: 0.0001 max mem: 52150 +Epoch: [24] [6670/7269] eta: 0:04:18 lr: 0.000004 loss: 0.5509 (0.5178) time: 0.4768 data: 0.0001 max mem: 52150 +Epoch: [24] [6680/7269] eta: 0:04:13 lr: 0.000004 loss: 0.5509 (0.5178) time: 0.4165 data: 0.0001 max mem: 52150 +Epoch: [24] [6690/7269] eta: 0:04:09 lr: 0.000004 loss: 0.5679 (0.5178) time: 0.4846 data: 0.0001 max mem: 52150 +Epoch: [24] [6700/7269] eta: 0:04:05 lr: 0.000004 loss: 0.5410 (0.5178) time: 0.4846 data: 0.0002 max mem: 52150 +Epoch: [24] [6710/7269] eta: 0:04:00 lr: 0.000004 loss: 0.5180 (0.5178) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [24] [6720/7269] eta: 0:03:56 lr: 0.000004 loss: 0.4870 (0.5178) time: 0.4160 data: 0.0002 max mem: 52150 +Epoch: [24] [6730/7269] eta: 0:03:52 lr: 0.000004 loss: 0.5228 (0.5178) time: 0.4160 data: 0.0002 max mem: 52150 +Epoch: [24] [6740/7269] eta: 0:03:47 lr: 0.000004 loss: 0.5108 (0.5177) time: 0.3714 data: 0.0001 max mem: 52150 +Epoch: [24] [6750/7269] eta: 0:03:43 lr: 0.000004 loss: 0.5025 (0.5177) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [24] [6760/7269] eta: 0:03:39 lr: 0.000004 loss: 0.5178 (0.5177) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [24] [6770/7269] eta: 0:03:34 lr: 0.000004 loss: 0.4767 (0.5176) time: 0.3714 data: 0.0001 max mem: 52150 +Epoch: [24] [6780/7269] eta: 0:03:30 lr: 0.000004 loss: 0.4584 (0.5176) time: 0.4323 data: 0.0001 max mem: 52150 +Epoch: [24] [6790/7269] eta: 0:03:26 lr: 0.000004 loss: 0.5185 (0.5176) time: 0.4778 data: 0.0001 max mem: 52150 +Epoch: [24] [6800/7269] eta: 0:03:22 lr: 0.000004 loss: 0.5372 (0.5177) time: 0.4171 data: 0.0001 max mem: 52150 +Epoch: [24] [6810/7269] eta: 0:03:17 lr: 0.000004 loss: 0.5408 (0.5177) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [24] [6820/7269] eta: 0:03:13 lr: 0.000004 loss: 0.5342 (0.5177) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [24] [6830/7269] eta: 0:03:09 lr: 0.000004 loss: 0.4884 (0.5176) time: 0.4950 data: 0.0001 max mem: 52150 +Epoch: [24] [6840/7269] eta: 0:03:04 lr: 0.000004 loss: 0.5150 (0.5176) time: 0.4957 data: 0.0001 max mem: 52150 +Epoch: [24] [6850/7269] eta: 0:03:00 lr: 0.000004 loss: 0.5564 (0.5177) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [24] [6860/7269] eta: 0:02:56 lr: 0.000004 loss: 0.5336 (0.5177) time: 0.4652 data: 0.0001 max mem: 52150 +Epoch: [24] [6870/7269] eta: 0:02:51 lr: 0.000004 loss: 0.5377 (0.5177) time: 0.5259 data: 0.0002 max mem: 52150 +Epoch: [24] [6880/7269] eta: 0:02:47 lr: 0.000004 loss: 0.5377 (0.5177) time: 0.4323 data: 0.0002 max mem: 52150 +Epoch: [24] [6890/7269] eta: 0:02:43 lr: 0.000004 loss: 0.5150 (0.5177) time: 0.3726 data: 0.0001 max mem: 52150 +Epoch: [24] [6900/7269] eta: 0:02:38 lr: 0.000004 loss: 0.4981 (0.5177) time: 0.3726 data: 0.0001 max mem: 52150 +Epoch: [24] [6910/7269] eta: 0:02:34 lr: 0.000004 loss: 0.4832 (0.5176) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [24] [6920/7269] eta: 0:02:30 lr: 0.000004 loss: 0.5096 (0.5176) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [24] [6930/7269] eta: 0:02:25 lr: 0.000004 loss: 0.5168 (0.5176) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [24] [6940/7269] eta: 0:02:21 lr: 0.000004 loss: 0.5150 (0.5176) time: 0.3713 data: 0.0001 max mem: 52150 +Epoch: [24] [6950/7269] eta: 0:02:17 lr: 0.000004 loss: 0.5034 (0.5175) time: 0.3713 data: 0.0001 max mem: 52150 +Epoch: [24] [6960/7269] eta: 0:02:12 lr: 0.000004 loss: 0.5147 (0.5175) time: 0.4342 data: 0.0001 max mem: 52150 +Epoch: [24] [6970/7269] eta: 0:02:08 lr: 0.000004 loss: 0.5477 (0.5176) time: 0.5411 data: 0.0001 max mem: 52150 +Epoch: [24] [6980/7269] eta: 0:02:04 lr: 0.000004 loss: 0.5622 (0.5176) time: 0.4784 data: 0.0002 max mem: 52150 +Epoch: [24] [6990/7269] eta: 0:02:00 lr: 0.000004 loss: 0.5395 (0.5176) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [24] [7000/7269] eta: 0:01:55 lr: 0.000004 loss: 0.5395 (0.5176) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [24] [7010/7269] eta: 0:01:51 lr: 0.000004 loss: 0.5136 (0.5176) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [24] [7020/7269] eta: 0:01:47 lr: 0.000004 loss: 0.5196 (0.5176) time: 0.4182 data: 0.0001 max mem: 52150 +Epoch: [24] [7030/7269] eta: 0:01:42 lr: 0.000004 loss: 0.5622 (0.5176) time: 0.4645 data: 0.0002 max mem: 52150 +Epoch: [24] [7040/7269] eta: 0:01:38 lr: 0.000004 loss: 0.5714 (0.5176) time: 0.4681 data: 0.0002 max mem: 52150 +Epoch: [24] [7050/7269] eta: 0:01:34 lr: 0.000004 loss: 0.5069 (0.5176) time: 0.5465 data: 0.0001 max mem: 52150 +Epoch: [24] [7060/7269] eta: 0:01:29 lr: 0.000004 loss: 0.4783 (0.5175) time: 0.4965 data: 0.0002 max mem: 52150 +Epoch: [24] [7070/7269] eta: 0:01:25 lr: 0.000004 loss: 0.4883 (0.5175) time: 0.4207 data: 0.0002 max mem: 52150 +Epoch: [24] [7080/7269] eta: 0:01:21 lr: 0.000004 loss: 0.4954 (0.5175) time: 0.4200 data: 0.0002 max mem: 52150 +Epoch: [24] [7090/7269] eta: 0:01:17 lr: 0.000004 loss: 0.5515 (0.5175) time: 0.4312 data: 0.0001 max mem: 52150 +Epoch: [24] [7100/7269] eta: 0:01:12 lr: 0.000004 loss: 0.4888 (0.5174) time: 0.4313 data: 0.0001 max mem: 52150 +Epoch: [24] [7110/7269] eta: 0:01:08 lr: 0.000004 loss: 0.4888 (0.5174) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [24] [7120/7269] eta: 0:01:04 lr: 0.000004 loss: 0.4794 (0.5174) time: 0.4319 data: 0.0002 max mem: 52150 +Epoch: [24] [7130/7269] eta: 0:00:59 lr: 0.000004 loss: 0.5114 (0.5174) time: 0.4314 data: 0.0002 max mem: 52150 +Epoch: [24] [7140/7269] eta: 0:00:55 lr: 0.000004 loss: 0.5263 (0.5173) time: 0.4308 data: 0.0001 max mem: 52150 +Epoch: [24] [7150/7269] eta: 0:00:51 lr: 0.000004 loss: 0.4521 (0.5173) time: 0.6034 data: 0.0001 max mem: 52150 +Epoch: [24] [7160/7269] eta: 0:00:46 lr: 0.000004 loss: 0.4528 (0.5173) time: 0.5440 data: 0.0001 max mem: 52150 +Epoch: [24] [7170/7269] eta: 0:00:42 lr: 0.000004 loss: 0.5094 (0.5173) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [24] [7180/7269] eta: 0:00:38 lr: 0.000004 loss: 0.4963 (0.5172) time: 0.3712 data: 0.0001 max mem: 52150 +Epoch: [24] [7190/7269] eta: 0:00:34 lr: 0.000004 loss: 0.4883 (0.5172) time: 0.3714 data: 0.0001 max mem: 52150 +Epoch: [24] [7200/7269] eta: 0:00:29 lr: 0.000004 loss: 0.5005 (0.5171) time: 0.4773 data: 0.0001 max mem: 52150 +Epoch: [24] [7210/7269] eta: 0:00:25 lr: 0.000004 loss: 0.5104 (0.5171) time: 0.4776 data: 0.0001 max mem: 52150 +Epoch: [24] [7220/7269] eta: 0:00:21 lr: 0.000004 loss: 0.5330 (0.5172) time: 0.4620 data: 0.0001 max mem: 52150 +Epoch: [24] [7230/7269] eta: 0:00:16 lr: 0.000004 loss: 0.5666 (0.5172) time: 0.4620 data: 0.0001 max mem: 52150 +Epoch: [24] [7240/7269] eta: 0:00:12 lr: 0.000004 loss: 0.5583 (0.5172) time: 0.3722 data: 0.0001 max mem: 52150 +Epoch: [24] [7250/7269] eta: 0:00:08 lr: 0.000004 loss: 0.4713 (0.5172) time: 0.4226 data: 0.0031 max mem: 52150 +Epoch: [24] [7260/7269] eta: 0:00:03 lr: 0.000004 loss: 0.5251 (0.5172) time: 0.4218 data: 0.0030 max mem: 52150 +Epoch: [24] [7268/7269] eta: 0:00:00 lr: 0.000004 loss: 0.5306 (0.5172) time: 0.4183 data: 0.0030 max mem: 52150 +Epoch: [24] Total time: 0:52:14 (0.4312 s / it) +Averaged stats: lr: 0.000004 loss: 0.5306 (0.5179) +Test: [ 0/1403] eta: 1:17:53 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0843 (0.0843) acc1: 100.0000 (100.0000) acc5: 100.0000 (100.0000) time: 3.3309 data: 3.2338 max mem: 52150 +Test: [ 10/1403] eta: 0:09:03 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1512 (0.1747) acc1: 98.4375 (95.5966) acc5: 100.0000 (100.0000) time: 0.3905 data: 0.2943 max mem: 52150 +Test: [ 20/1403] eta: 0:05:46 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1373 (0.1607) acc1: 98.4375 (95.8333) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 30/1403] eta: 0:04:35 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0910 (0.1467) acc1: 100.0000 (96.3710) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 40/1403] eta: 0:03:59 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0973 (0.1477) acc1: 100.0000 (96.6463) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 50/1403] eta: 0:03:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1207 (0.1604) acc1: 98.4375 (95.7721) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 60/1403] eta: 0:03:20 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1334 (0.1578) acc1: 100.0000 (96.2090) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 70/1403] eta: 0:03:09 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.1358 (0.1596) acc1: 98.4375 (96.0167) acc5: 100.0000 (100.0000) time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 80/1403] eta: 0:03:00 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.1358 (0.1682) acc1: 96.8750 (95.5633) acc5: 100.0000 (100.0000) time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 90/1403] eta: 0:02:53 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1820 (0.1819) acc1: 93.7500 (94.8317) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 100/1403] eta: 0:02:47 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1531 (0.1787) acc1: 96.8750 (95.0804) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 110/1403] eta: 0:02:42 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1326 (0.1770) acc1: 98.4375 (95.1999) acc5: 100.0000 (100.0000) time: 0.0963 data: 0.0003 max mem: 52150 +Test: [ 120/1403] eta: 0:02:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0817 (0.1691) acc1: 100.0000 (95.5837) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 130/1403] eta: 0:02:34 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0817 (0.1639) acc1: 100.0000 (95.7896) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 140/1403] eta: 0:02:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0779 (0.1577) acc1: 100.0000 (96.0882) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 150/1403] eta: 0:02:27 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0761 (0.1528) acc1: 100.0000 (96.3162) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 160/1403] eta: 0:02:24 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0849 (0.1488) acc1: 100.0000 (96.4674) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 170/1403] eta: 0:02:22 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0849 (0.1455) acc1: 100.0000 (96.6466) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 180/1403] eta: 0:02:19 acc1_pos: 89.062 acc5_pos: 100.000 loss: 0.0896 (0.1497) acc1: 100.0000 (96.4606) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 190/1403] eta: 0:02:17 acc1_pos: 90.625 acc5_pos: 100.000 loss: 0.3018 (0.1587) acc1: 89.0625 (95.9751) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 200/1403] eta: 0:02:15 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.3077 (0.1708) acc1: 89.0625 (95.3980) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 210/1403] eta: 0:02:13 acc1_pos: 78.125 acc5_pos: 100.000 loss: 0.3421 (0.1823) acc1: 85.9375 (94.7867) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 220/1403] eta: 0:02:11 acc1_pos: 87.500 acc5_pos: 100.000 loss: 0.3493 (0.1916) acc1: 85.9375 (94.3085) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 230/1403] eta: 0:02:09 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2670 (0.1881) acc1: 90.6250 (94.5076) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 240/1403] eta: 0:02:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0834 (0.1839) acc1: 100.0000 (94.7225) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 250/1403] eta: 0:02:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0847 (0.1803) acc1: 100.0000 (94.9141) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 260/1403] eta: 0:02:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0857 (0.1769) acc1: 100.0000 (95.0910) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 270/1403] eta: 0:02:02 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0867 (0.1737) acc1: 100.0000 (95.2433) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 280/1403] eta: 0:02:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0883 (0.1706) acc1: 100.0000 (95.4070) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 290/1403] eta: 0:01:59 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0815 (0.1676) acc1: 100.0000 (95.5595) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 300/1403] eta: 0:01:58 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0795 (0.1648) acc1: 100.0000 (95.7018) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 310/1403] eta: 0:01:56 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0866 (0.1624) acc1: 100.0000 (95.8400) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 320/1403] eta: 0:01:55 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1017 (0.1607) acc1: 100.0000 (95.9696) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 330/1403] eta: 0:01:54 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1099 (0.1592) acc1: 100.0000 (96.0914) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 340/1403] eta: 0:01:52 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1099 (0.1577) acc1: 100.0000 (96.2060) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 350/1403] eta: 0:01:51 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1056 (0.1563) acc1: 100.0000 (96.3141) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 360/1403] eta: 0:01:50 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1104 (0.1551) acc1: 100.0000 (96.4119) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 370/1403] eta: 0:01:48 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1082 (0.1538) acc1: 100.0000 (96.5086) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 380/1403] eta: 0:01:47 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1071 (0.1526) acc1: 100.0000 (96.6002) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 390/1403] eta: 0:01:46 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1082 (0.1531) acc1: 100.0000 (96.5673) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 400/1403] eta: 0:01:44 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2232 (0.1567) acc1: 90.6250 (96.3490) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 410/1403] eta: 0:01:43 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2699 (0.1595) acc1: 89.0625 (96.1945) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 420/1403] eta: 0:01:42 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2405 (0.1609) acc1: 90.6250 (96.1030) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 430/1403] eta: 0:01:41 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2289 (0.1635) acc1: 90.6250 (95.9723) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0970 data: 0.0004 max mem: 52150 +Test: [ 440/1403] eta: 0:01:40 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2849 (0.1668) acc1: 89.0625 (95.8085) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0970 data: 0.0004 max mem: 52150 +Test: [ 450/1403] eta: 0:01:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2746 (0.1688) acc1: 89.0625 (95.6971) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 460/1403] eta: 0:01:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2509 (0.1710) acc1: 90.6250 (95.5701) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 470/1403] eta: 0:01:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2202 (0.1701) acc1: 92.1875 (95.5945) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 480/1403] eta: 0:01:35 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0631 (0.1679) acc1: 100.0000 (95.6861) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 490/1403] eta: 0:01:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0631 (0.1658) acc1: 100.0000 (95.7739) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 500/1403] eta: 0:01:33 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0639 (0.1637) acc1: 100.0000 (95.8583) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 510/1403] eta: 0:01:31 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0642 (0.1618) acc1: 100.0000 (95.9393) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 520/1403] eta: 0:01:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0685 (0.1601) acc1: 100.0000 (96.0173) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 530/1403] eta: 0:01:29 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0598 (0.1582) acc1: 100.0000 (96.0923) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 540/1403] eta: 0:01:28 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0590 (0.1563) acc1: 100.0000 (96.1645) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 550/1403] eta: 0:01:27 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0603 (0.1550) acc1: 100.0000 (96.2199) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 560/1403] eta: 0:01:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0763 (0.1538) acc1: 100.0000 (96.2734) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 570/1403] eta: 0:01:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0815 (0.1526) acc1: 100.0000 (96.3250) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 580/1403] eta: 0:01:24 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0856 (0.1519) acc1: 100.0000 (96.3533) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 590/1403] eta: 0:01:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1063 (0.1515) acc1: 98.4375 (96.3621) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 600/1403] eta: 0:01:21 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1235 (0.1516) acc1: 96.8750 (96.3212) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0972 data: 0.0003 max mem: 52150 +Test: [ 610/1403] eta: 0:01:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1263 (0.1511) acc1: 95.3125 (96.3329) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0971 data: 0.0003 max mem: 52150 +Test: [ 620/1403] eta: 0:01:19 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1333 (0.1521) acc1: 95.3125 (96.2334) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0970 data: 0.0003 max mem: 52150 +Test: [ 630/1403] eta: 0:01:18 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1472 (0.1518) acc1: 93.7500 (96.2312) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0971 data: 0.0003 max mem: 52150 +Test: [ 640/1403] eta: 0:01:17 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1250 (0.1514) acc1: 96.8750 (96.2534) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0973 data: 0.0003 max mem: 52150 +Test: [ 650/1403] eta: 0:01:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1187 (0.1509) acc1: 96.8750 (96.2726) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0973 data: 0.0003 max mem: 52150 +Test: [ 660/1403] eta: 0:01:15 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0926 (0.1499) acc1: 100.0000 (96.3219) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0974 data: 0.0003 max mem: 52150 +Test: [ 670/1403] eta: 0:01:14 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0694 (0.1487) acc1: 100.0000 (96.3767) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0975 data: 0.0003 max mem: 52150 +Test: [ 680/1403] eta: 0:01:13 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0784 (0.1502) acc1: 100.0000 (96.2807) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0970 data: 0.0003 max mem: 52150 +Test: [ 690/1403] eta: 0:01:12 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2794 (0.1549) acc1: 87.5000 (95.9954) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0970 data: 0.0003 max mem: 52150 +Test: [ 700/1403] eta: 0:01:11 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2226 (0.1553) acc1: 89.0625 (95.9700) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.0973 data: 0.0003 max mem: 52150 +Test: [ 710/1403] eta: 0:01:10 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3597 (0.1606) acc1: 84.3750 (95.6927) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0970 data: 0.0003 max mem: 52150 +Test: [ 720/1403] eta: 0:01:09 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3597 (0.1607) acc1: 79.6875 (95.6722) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 730/1403] eta: 0:01:08 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0666 (0.1595) acc1: 100.0000 (95.7250) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 740/1403] eta: 0:01:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0724 (0.1586) acc1: 100.0000 (95.7680) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 750/1403] eta: 0:01:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1062 (0.1601) acc1: 98.4375 (95.6828) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 760/1403] eta: 0:01:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3410 (0.1643) acc1: 81.2500 (95.4028) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 770/1403] eta: 0:01:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4985 (0.1690) acc1: 73.4375 (95.1078) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 780/1403] eta: 0:01:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4390 (0.1716) acc1: 76.5625 (94.9464) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 790/1403] eta: 0:01:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3054 (0.1736) acc1: 84.3750 (94.8345) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 800/1403] eta: 0:01:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2102 (0.1739) acc1: 92.1875 (94.8248) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 810/1403] eta: 0:00:59 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2198 (0.1768) acc1: 89.0625 (94.6883) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 820/1403] eta: 0:00:58 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6191 (0.1855) acc1: 71.8750 (94.3038) acc5: 100.0000 (100.0000) acc1_neg: 42.188 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 830/1403] eta: 0:00:57 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6558 (0.1878) acc1: 71.8750 (94.1862) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 840/1403] eta: 0:00:56 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4817 (0.1951) acc1: 73.4375 (93.8652) acc5: 100.0000 (100.0000) acc1_neg: 50.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 850/1403] eta: 0:00:55 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.9611 (0.2051) acc1: 59.3750 (93.4342) acc5: 100.0000 (100.0000) acc1_neg: 57.812 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 860/1403] eta: 0:00:54 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.8990 (0.2082) acc1: 60.9375 (93.2891) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 870/1403] eta: 0:00:53 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1294 (0.2071) acc1: 95.3125 (93.3392) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 880/1403] eta: 0:00:52 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0937 (0.2059) acc1: 98.4375 (93.3935) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 890/1403] eta: 0:00:51 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0824 (0.2046) acc1: 98.4375 (93.4571) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 900/1403] eta: 0:00:50 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0745 (0.2031) acc1: 100.0000 (93.5263) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 910/1403] eta: 0:00:49 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0766 (0.2022) acc1: 100.0000 (93.5716) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 920/1403] eta: 0:00:48 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1103 (0.2013) acc1: 96.8750 (93.6024) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 930/1403] eta: 0:00:47 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1032 (0.2002) acc1: 98.4375 (93.6560) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 940/1403] eta: 0:00:46 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1164 (0.2005) acc1: 96.8750 (93.6255) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 950/1403] eta: 0:00:45 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2659 (0.2017) acc1: 89.0625 (93.5561) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 960/1403] eta: 0:00:44 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2966 (0.2024) acc1: 87.5000 (93.5110) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 970/1403] eta: 0:00:43 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2538 (0.2034) acc1: 89.0625 (93.4507) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 980/1403] eta: 0:00:42 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2660 (0.2044) acc1: 89.0625 (93.3916) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 990/1403] eta: 0:00:41 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2660 (0.2049) acc1: 89.0625 (93.3495) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [1000/1403] eta: 0:00:40 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2577 (0.2059) acc1: 90.6250 (93.2958) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [1010/1403] eta: 0:00:39 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2978 (0.2069) acc1: 87.5000 (93.2369) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1020/1403] eta: 0:00:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4364 (0.2130) acc1: 82.8125 (92.8976) acc5: 100.0000 (100.0000) acc1_neg: 29.688 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1030/1403] eta: 0:00:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5078 (0.2151) acc1: 62.5000 (92.7634) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [1040/1403] eta: 0:00:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2242 (0.2149) acc1: 89.0625 (92.7594) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [1050/1403] eta: 0:00:35 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1643 (0.2144) acc1: 93.7500 (92.7807) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1060/1403] eta: 0:00:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1291 (0.2137) acc1: 96.8750 (92.8178) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1070/1403] eta: 0:00:33 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1525 (0.2140) acc1: 95.3125 (92.8017) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1080/1403] eta: 0:00:32 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3668 (0.2157) acc1: 85.9375 (92.7107) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1090/1403] eta: 0:00:31 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4064 (0.2174) acc1: 82.8125 (92.6172) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1100/1403] eta: 0:00:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3210 (0.2178) acc1: 85.9375 (92.5920) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1110/1403] eta: 0:00:29 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2520 (0.2181) acc1: 90.6250 (92.5827) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0966 data: 0.0002 max mem: 52150 +Test: [1120/1403] eta: 0:00:28 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2528 (0.2184) acc1: 90.6250 (92.5652) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0966 data: 0.0002 max mem: 52150 +Test: [1130/1403] eta: 0:00:27 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2625 (0.2189) acc1: 90.6250 (92.5343) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1140/1403] eta: 0:00:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2242 (0.2189) acc1: 92.1875 (92.5326) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1150/1403] eta: 0:00:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2240 (0.2191) acc1: 92.1875 (92.5228) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1160/1403] eta: 0:00:24 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2392 (0.2195) acc1: 90.6250 (92.5051) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0966 data: 0.0002 max mem: 52150 +Test: [1170/1403] eta: 0:00:23 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2365 (0.2192) acc1: 90.6250 (92.5117) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0002 max mem: 52150 +Test: [1180/1403] eta: 0:00:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0717 (0.2180) acc1: 100.0000 (92.5751) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1190/1403] eta: 0:00:21 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0716 (0.2167) acc1: 100.0000 (92.6375) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1200/1403] eta: 0:00:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0715 (0.2155) acc1: 100.0000 (92.6988) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1210/1403] eta: 0:00:19 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0702 (0.2143) acc1: 100.0000 (92.7591) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1220/1403] eta: 0:00:18 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0694 (0.2131) acc1: 100.0000 (92.8184) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1230/1403] eta: 0:00:17 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0706 (0.2120) acc1: 100.0000 (92.8767) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1240/1403] eta: 0:00:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0706 (0.2109) acc1: 100.0000 (92.9341) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1250/1403] eta: 0:00:15 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0709 (0.2098) acc1: 100.0000 (92.9906) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1260/1403] eta: 0:00:14 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0772 (0.2087) acc1: 100.0000 (93.0462) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1270/1403] eta: 0:00:13 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0782 (0.2077) acc1: 100.0000 (93.1009) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1280/1403] eta: 0:00:12 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0778 (0.2067) acc1: 100.0000 (93.1548) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1290/1403] eta: 0:00:11 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0775 (0.2057) acc1: 100.0000 (93.2078) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1300/1403] eta: 0:00:10 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0784 (0.2047) acc1: 100.0000 (93.2600) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1310/1403] eta: 0:00:09 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0788 (0.2038) acc1: 100.0000 (93.3114) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0002 max mem: 52150 +Test: [1320/1403] eta: 0:00:08 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0765 (0.2028) acc1: 100.0000 (93.3620) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1330/1403] eta: 0:00:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0765 (0.2018) acc1: 100.0000 (93.4096) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1340/1403] eta: 0:00:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0678 (0.2008) acc1: 100.0000 (93.4564) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1350/1403] eta: 0:00:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0663 (0.1999) acc1: 100.0000 (93.5013) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1360/1403] eta: 0:00:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0669 (0.1989) acc1: 100.0000 (93.5491) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1370/1403] eta: 0:00:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0669 (0.1980) acc1: 100.0000 (93.5927) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1380/1403] eta: 0:00:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0660 (0.1971) acc1: 100.0000 (93.6380) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1390/1403] eta: 0:00:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0656 (0.1961) acc1: 100.0000 (93.6837) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0002 max mem: 52150 +Test: [1400/1403] eta: 0:00:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0663 (0.1953) acc1: 100.0000 (93.7255) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0001 max mem: 52150 +Test: [1402/1403] eta: 0:00:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0669 (0.1951) acc1: 100.0000 (93.7300) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0922 data: 0.0001 max mem: 52150 +Test: Total time: 0:02:19 (0.0994 s / it) +* Overall Acc@1 93.730 Acc@5 100.000 + target == real Acc@1 95.803 Acc@5 100.000 + target == fake Acc@1 93.146 Acc@5 100.000 + loss 0.195 +Accuracy of the network on the 89729 test images: 93.7% +Max accuracy: 94.28% +Epoch: [25] [ 0/7269] eta: 11:30:30 lr: 0.000004 loss: 0.5308 (0.5308) time: 5.6997 data: 4.8555 max mem: 52150 +Epoch: [25] [ 10/7269] eta: 1:54:55 lr: 0.000004 loss: 0.5740 (0.5534) time: 0.9499 data: 0.4416 max mem: 52150 +Epoch: [25] [ 20/7269] eta: 1:21:31 lr: 0.000004 loss: 0.5624 (0.5411) time: 0.4235 data: 0.0002 max mem: 52150 +Epoch: [25] [ 30/7269] eta: 1:09:38 lr: 0.000004 loss: 0.5332 (0.5385) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [25] [ 40/7269] eta: 1:03:29 lr: 0.000004 loss: 0.5332 (0.5362) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [25] [ 50/7269] eta: 0:59:44 lr: 0.000004 loss: 0.5478 (0.5334) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [25] [ 60/7269] eta: 1:04:01 lr: 0.000004 loss: 0.5486 (0.5266) time: 0.5450 data: 0.0002 max mem: 52150 +Epoch: [25] [ 70/7269] eta: 1:01:12 lr: 0.000004 loss: 0.5223 (0.5235) time: 0.5448 data: 0.0002 max mem: 52150 +Epoch: [25] [ 80/7269] eta: 0:59:04 lr: 0.000004 loss: 0.5223 (0.5226) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [25] [ 90/7269] eta: 0:57:23 lr: 0.000004 loss: 0.5082 (0.5194) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [25] [ 100/7269] eta: 0:56:01 lr: 0.000004 loss: 0.4858 (0.5145) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [25] [ 110/7269] eta: 0:54:54 lr: 0.000004 loss: 0.5031 (0.5185) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [25] [ 120/7269] eta: 0:53:56 lr: 0.000004 loss: 0.4900 (0.5154) time: 0.3709 data: 0.0002 max mem: 52150 +Epoch: [25] [ 130/7269] eta: 0:53:07 lr: 0.000004 loss: 0.5234 (0.5175) time: 0.3712 data: 0.0002 max mem: 52150 +Epoch: [25] [ 140/7269] eta: 0:54:04 lr: 0.000004 loss: 0.5345 (0.5172) time: 0.4696 data: 0.0002 max mem: 52150 +Epoch: [25] [ 150/7269] eta: 0:53:20 lr: 0.000004 loss: 0.5578 (0.5204) time: 0.4696 data: 0.0002 max mem: 52150 +Epoch: [25] [ 160/7269] eta: 0:52:41 lr: 0.000004 loss: 0.5348 (0.5209) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [25] [ 170/7269] eta: 0:52:06 lr: 0.000004 loss: 0.5114 (0.5200) time: 0.3711 data: 0.0002 max mem: 52150 +Epoch: [25] [ 180/7269] eta: 0:51:35 lr: 0.000004 loss: 0.5114 (0.5207) time: 0.3711 data: 0.0002 max mem: 52150 +Epoch: [25] [ 190/7269] eta: 0:51:44 lr: 0.000004 loss: 0.5241 (0.5208) time: 0.4222 data: 0.0001 max mem: 52150 +Epoch: [25] [ 200/7269] eta: 0:51:49 lr: 0.000004 loss: 0.5566 (0.5228) time: 0.4693 data: 0.0002 max mem: 52150 +Epoch: [25] [ 210/7269] eta: 0:51:21 lr: 0.000004 loss: 0.5566 (0.5226) time: 0.4184 data: 0.0002 max mem: 52150 +Epoch: [25] [ 220/7269] eta: 0:51:28 lr: 0.000004 loss: 0.5319 (0.5232) time: 0.4206 data: 0.0002 max mem: 52150 +Epoch: [25] [ 230/7269] eta: 0:51:41 lr: 0.000004 loss: 0.5139 (0.5222) time: 0.4827 data: 0.0002 max mem: 52150 +Epoch: [25] [ 240/7269] eta: 0:51:45 lr: 0.000004 loss: 0.5384 (0.5224) time: 0.4826 data: 0.0002 max mem: 52150 +Epoch: [25] [ 250/7269] eta: 0:51:21 lr: 0.000004 loss: 0.5388 (0.5225) time: 0.4210 data: 0.0002 max mem: 52150 +Epoch: [25] [ 260/7269] eta: 0:50:59 lr: 0.000004 loss: 0.5518 (0.5231) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [25] [ 270/7269] eta: 0:51:09 lr: 0.000004 loss: 0.5475 (0.5229) time: 0.4335 data: 0.0002 max mem: 52150 +Epoch: [25] [ 280/7269] eta: 0:51:12 lr: 0.000004 loss: 0.5295 (0.5223) time: 0.4804 data: 0.0001 max mem: 52150 +Epoch: [25] [ 290/7269] eta: 0:50:51 lr: 0.000004 loss: 0.5295 (0.5221) time: 0.4184 data: 0.0001 max mem: 52150 +Epoch: [25] [ 300/7269] eta: 0:51:00 lr: 0.000004 loss: 0.5454 (0.5232) time: 0.4324 data: 0.0002 max mem: 52150 +Epoch: [25] [ 310/7269] eta: 0:50:40 lr: 0.000004 loss: 0.5305 (0.5234) time: 0.4328 data: 0.0002 max mem: 52150 +Epoch: [25] [ 320/7269] eta: 0:50:22 lr: 0.000004 loss: 0.5131 (0.5228) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [25] [ 330/7269] eta: 0:50:04 lr: 0.000004 loss: 0.5105 (0.5227) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [25] [ 340/7269] eta: 0:50:13 lr: 0.000004 loss: 0.5214 (0.5225) time: 0.4352 data: 0.0002 max mem: 52150 +Epoch: [25] [ 350/7269] eta: 0:49:56 lr: 0.000004 loss: 0.5331 (0.5227) time: 0.4353 data: 0.0002 max mem: 52150 +Epoch: [25] [ 360/7269] eta: 0:50:04 lr: 0.000004 loss: 0.5396 (0.5227) time: 0.4334 data: 0.0002 max mem: 52150 +Epoch: [25] [ 370/7269] eta: 0:50:11 lr: 0.000004 loss: 0.5037 (0.5226) time: 0.4956 data: 0.0002 max mem: 52150 +Epoch: [25] [ 380/7269] eta: 0:50:13 lr: 0.000004 loss: 0.5366 (0.5232) time: 0.4838 data: 0.0002 max mem: 52150 +Epoch: [25] [ 390/7269] eta: 0:50:42 lr: 0.000004 loss: 0.5218 (0.5226) time: 0.5508 data: 0.0002 max mem: 52150 +Epoch: [25] [ 400/7269] eta: 0:50:47 lr: 0.000004 loss: 0.5056 (0.5227) time: 0.5637 data: 0.0001 max mem: 52150 +Epoch: [25] [ 410/7269] eta: 0:50:31 lr: 0.000004 loss: 0.5416 (0.5231) time: 0.4343 data: 0.0001 max mem: 52150 +Epoch: [25] [ 420/7269] eta: 0:50:32 lr: 0.000004 loss: 0.5152 (0.5228) time: 0.4231 data: 0.0002 max mem: 52150 +Epoch: [25] [ 430/7269] eta: 0:50:16 lr: 0.000004 loss: 0.5184 (0.5232) time: 0.4233 data: 0.0002 max mem: 52150 +Epoch: [25] [ 440/7269] eta: 0:50:20 lr: 0.000004 loss: 0.5569 (0.5241) time: 0.4325 data: 0.0002 max mem: 52150 +Epoch: [25] [ 450/7269] eta: 0:50:19 lr: 0.000004 loss: 0.5647 (0.5248) time: 0.4806 data: 0.0002 max mem: 52150 +Epoch: [25] [ 460/7269] eta: 0:50:04 lr: 0.000004 loss: 0.5272 (0.5245) time: 0.4198 data: 0.0002 max mem: 52150 +Epoch: [25] [ 470/7269] eta: 0:49:50 lr: 0.000004 loss: 0.5124 (0.5248) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [25] [ 480/7269] eta: 0:49:36 lr: 0.000004 loss: 0.5124 (0.5249) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [25] [ 490/7269] eta: 0:49:40 lr: 0.000004 loss: 0.5104 (0.5245) time: 0.4354 data: 0.0002 max mem: 52150 +Epoch: [25] [ 500/7269] eta: 0:49:38 lr: 0.000004 loss: 0.5089 (0.5237) time: 0.4804 data: 0.0002 max mem: 52150 +Epoch: [25] [ 510/7269] eta: 0:49:25 lr: 0.000004 loss: 0.5475 (0.5243) time: 0.4168 data: 0.0002 max mem: 52150 +Epoch: [25] [ 520/7269] eta: 0:49:12 lr: 0.000004 loss: 0.5544 (0.5246) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [25] [ 530/7269] eta: 0:48:59 lr: 0.000004 loss: 0.5482 (0.5248) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [25] [ 540/7269] eta: 0:48:47 lr: 0.000004 loss: 0.5443 (0.5252) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [25] [ 550/7269] eta: 0:48:47 lr: 0.000004 loss: 0.5217 (0.5245) time: 0.4223 data: 0.0002 max mem: 52150 +Epoch: [25] [ 560/7269] eta: 0:48:35 lr: 0.000004 loss: 0.5064 (0.5245) time: 0.4220 data: 0.0002 max mem: 52150 +Epoch: [25] [ 570/7269] eta: 0:48:23 lr: 0.000004 loss: 0.5156 (0.5241) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [25] [ 580/7269] eta: 0:48:41 lr: 0.000004 loss: 0.5454 (0.5246) time: 0.4977 data: 0.0001 max mem: 52150 +Epoch: [25] [ 590/7269] eta: 0:48:29 lr: 0.000004 loss: 0.5457 (0.5249) time: 0.4972 data: 0.0002 max mem: 52150 +Epoch: [25] [ 600/7269] eta: 0:48:18 lr: 0.000004 loss: 0.5041 (0.5246) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [25] [ 610/7269] eta: 0:48:07 lr: 0.000004 loss: 0.5026 (0.5246) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [25] [ 620/7269] eta: 0:47:56 lr: 0.000004 loss: 0.5461 (0.5248) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [25] [ 630/7269] eta: 0:47:45 lr: 0.000004 loss: 0.5190 (0.5242) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [25] [ 640/7269] eta: 0:47:47 lr: 0.000004 loss: 0.4896 (0.5236) time: 0.4327 data: 0.0002 max mem: 52150 +Epoch: [25] [ 650/7269] eta: 0:47:46 lr: 0.000004 loss: 0.5180 (0.5231) time: 0.4781 data: 0.0002 max mem: 52150 +Epoch: [25] [ 660/7269] eta: 0:47:35 lr: 0.000004 loss: 0.5113 (0.5224) time: 0.4176 data: 0.0002 max mem: 52150 +Epoch: [25] [ 670/7269] eta: 0:47:37 lr: 0.000004 loss: 0.5113 (0.5227) time: 0.4340 data: 0.0002 max mem: 52150 +Epoch: [25] [ 680/7269] eta: 0:47:36 lr: 0.000004 loss: 0.5406 (0.5228) time: 0.4795 data: 0.0002 max mem: 52150 +Epoch: [25] [ 690/7269] eta: 0:47:37 lr: 0.000004 loss: 0.5370 (0.5230) time: 0.4780 data: 0.0001 max mem: 52150 +Epoch: [25] [ 700/7269] eta: 0:47:44 lr: 0.000004 loss: 0.5182 (0.5229) time: 0.5256 data: 0.0001 max mem: 52150 +Epoch: [25] [ 710/7269] eta: 0:47:46 lr: 0.000004 loss: 0.5238 (0.5229) time: 0.5292 data: 0.0002 max mem: 52150 +Epoch: [25] [ 720/7269] eta: 0:47:44 lr: 0.000004 loss: 0.5357 (0.5226) time: 0.4813 data: 0.0002 max mem: 52150 +Epoch: [25] [ 730/7269] eta: 0:47:34 lr: 0.000004 loss: 0.5314 (0.5228) time: 0.4176 data: 0.0002 max mem: 52150 +Epoch: [25] [ 740/7269] eta: 0:47:32 lr: 0.000004 loss: 0.5314 (0.5227) time: 0.4178 data: 0.0002 max mem: 52150 +Epoch: [25] [ 750/7269] eta: 0:47:22 lr: 0.000004 loss: 0.5127 (0.5227) time: 0.4181 data: 0.0002 max mem: 52150 +Epoch: [25] [ 760/7269] eta: 0:47:12 lr: 0.000004 loss: 0.5090 (0.5223) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [25] [ 770/7269] eta: 0:47:22 lr: 0.000004 loss: 0.5077 (0.5226) time: 0.4906 data: 0.0002 max mem: 52150 +Epoch: [25] [ 780/7269] eta: 0:47:12 lr: 0.000004 loss: 0.5077 (0.5225) time: 0.4912 data: 0.0002 max mem: 52150 +Epoch: [25] [ 790/7269] eta: 0:47:03 lr: 0.000004 loss: 0.5154 (0.5229) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [25] [ 800/7269] eta: 0:46:53 lr: 0.000004 loss: 0.5696 (0.5232) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [25] [ 810/7269] eta: 0:46:52 lr: 0.000004 loss: 0.5438 (0.5229) time: 0.4193 data: 0.0002 max mem: 52150 +Epoch: [25] [ 820/7269] eta: 0:46:42 lr: 0.000004 loss: 0.5262 (0.5228) time: 0.4192 data: 0.0002 max mem: 52150 +Epoch: [25] [ 830/7269] eta: 0:46:33 lr: 0.000004 loss: 0.5195 (0.5225) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [25] [ 840/7269] eta: 0:46:40 lr: 0.000004 loss: 0.5145 (0.5224) time: 0.4772 data: 0.0002 max mem: 52150 +Epoch: [25] [ 850/7269] eta: 0:46:31 lr: 0.000004 loss: 0.5145 (0.5224) time: 0.4764 data: 0.0001 max mem: 52150 +Epoch: [25] [ 860/7269] eta: 0:46:31 lr: 0.000004 loss: 0.4959 (0.5220) time: 0.4320 data: 0.0002 max mem: 52150 +Epoch: [25] [ 870/7269] eta: 0:46:22 lr: 0.000004 loss: 0.5045 (0.5216) time: 0.4323 data: 0.0002 max mem: 52150 +Epoch: [25] [ 880/7269] eta: 0:46:13 lr: 0.000004 loss: 0.5319 (0.5219) time: 0.3727 data: 0.0002 max mem: 52150 +Epoch: [25] [ 890/7269] eta: 0:46:11 lr: 0.000004 loss: 0.5386 (0.5218) time: 0.4179 data: 0.0002 max mem: 52150 +Epoch: [25] [ 900/7269] eta: 0:46:08 lr: 0.000004 loss: 0.5386 (0.5219) time: 0.4636 data: 0.0001 max mem: 52150 +Epoch: [25] [ 910/7269] eta: 0:46:00 lr: 0.000004 loss: 0.5078 (0.5214) time: 0.4189 data: 0.0001 max mem: 52150 +Epoch: [25] [ 920/7269] eta: 0:45:59 lr: 0.000004 loss: 0.5266 (0.5218) time: 0.4323 data: 0.0001 max mem: 52150 +Epoch: [25] [ 930/7269] eta: 0:45:51 lr: 0.000004 loss: 0.5530 (0.5219) time: 0.4319 data: 0.0001 max mem: 52150 +Epoch: [25] [ 940/7269] eta: 0:45:50 lr: 0.000004 loss: 0.5330 (0.5219) time: 0.4321 data: 0.0001 max mem: 52150 +Epoch: [25] [ 950/7269] eta: 0:45:48 lr: 0.000004 loss: 0.5338 (0.5219) time: 0.4819 data: 0.0002 max mem: 52150 +Epoch: [25] [ 960/7269] eta: 0:45:46 lr: 0.000004 loss: 0.5338 (0.5219) time: 0.4664 data: 0.0002 max mem: 52150 +Epoch: [25] [ 970/7269] eta: 0:45:37 lr: 0.000004 loss: 0.4854 (0.5214) time: 0.4163 data: 0.0002 max mem: 52150 +Epoch: [25] [ 980/7269] eta: 0:45:35 lr: 0.000004 loss: 0.4972 (0.5217) time: 0.4185 data: 0.0002 max mem: 52150 +Epoch: [25] [ 990/7269] eta: 0:45:34 lr: 0.000004 loss: 0.5450 (0.5217) time: 0.4784 data: 0.0002 max mem: 52150 +Epoch: [25] [1000/7269] eta: 0:45:26 lr: 0.000004 loss: 0.5576 (0.5218) time: 0.4316 data: 0.0002 max mem: 52150 +Epoch: [25] [1010/7269] eta: 0:45:18 lr: 0.000004 loss: 0.5271 (0.5215) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [25] [1020/7269] eta: 0:45:30 lr: 0.000004 loss: 0.4890 (0.5211) time: 0.5397 data: 0.0002 max mem: 52150 +Epoch: [25] [1030/7269] eta: 0:45:22 lr: 0.000004 loss: 0.4918 (0.5209) time: 0.5392 data: 0.0002 max mem: 52150 +Epoch: [25] [1040/7269] eta: 0:45:13 lr: 0.000004 loss: 0.5099 (0.5208) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [25] [1050/7269] eta: 0:45:05 lr: 0.000004 loss: 0.5355 (0.5209) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [25] [1060/7269] eta: 0:44:57 lr: 0.000004 loss: 0.5318 (0.5209) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [25] [1070/7269] eta: 0:44:49 lr: 0.000004 loss: 0.5318 (0.5209) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [25] [1080/7269] eta: 0:44:41 lr: 0.000004 loss: 0.5278 (0.5208) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [25] [1090/7269] eta: 0:44:34 lr: 0.000004 loss: 0.5055 (0.5208) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [25] [1100/7269] eta: 0:44:26 lr: 0.000004 loss: 0.5273 (0.5209) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [25] [1110/7269] eta: 0:44:24 lr: 0.000004 loss: 0.5273 (0.5209) time: 0.4229 data: 0.0002 max mem: 52150 +Epoch: [25] [1120/7269] eta: 0:44:16 lr: 0.000004 loss: 0.5119 (0.5207) time: 0.4231 data: 0.0002 max mem: 52150 +Epoch: [25] [1130/7269] eta: 0:44:14 lr: 0.000004 loss: 0.5215 (0.5208) time: 0.4179 data: 0.0002 max mem: 52150 +Epoch: [25] [1140/7269] eta: 0:44:12 lr: 0.000004 loss: 0.5066 (0.5203) time: 0.4778 data: 0.0002 max mem: 52150 +Epoch: [25] [1150/7269] eta: 0:44:05 lr: 0.000004 loss: 0.5066 (0.5204) time: 0.4319 data: 0.0001 max mem: 52150 +Epoch: [25] [1160/7269] eta: 0:44:13 lr: 0.000004 loss: 0.5160 (0.5204) time: 0.5221 data: 0.0001 max mem: 52150 +Epoch: [25] [1170/7269] eta: 0:44:06 lr: 0.000004 loss: 0.5034 (0.5203) time: 0.5223 data: 0.0002 max mem: 52150 +Epoch: [25] [1180/7269] eta: 0:44:04 lr: 0.000004 loss: 0.5292 (0.5202) time: 0.4323 data: 0.0002 max mem: 52150 +Epoch: [25] [1190/7269] eta: 0:44:01 lr: 0.000004 loss: 0.5181 (0.5201) time: 0.4772 data: 0.0002 max mem: 52150 +Epoch: [25] [1200/7269] eta: 0:43:59 lr: 0.000004 loss: 0.4732 (0.5197) time: 0.4642 data: 0.0002 max mem: 52150 +Epoch: [25] [1210/7269] eta: 0:44:08 lr: 0.000004 loss: 0.4732 (0.5196) time: 0.5840 data: 0.0002 max mem: 52150 +Epoch: [25] [1220/7269] eta: 0:44:00 lr: 0.000004 loss: 0.5184 (0.5197) time: 0.5373 data: 0.0002 max mem: 52150 +Epoch: [25] [1230/7269] eta: 0:43:58 lr: 0.000004 loss: 0.5349 (0.5199) time: 0.4327 data: 0.0001 max mem: 52150 +Epoch: [25] [1240/7269] eta: 0:43:57 lr: 0.000004 loss: 0.5495 (0.5199) time: 0.4926 data: 0.0001 max mem: 52150 +Epoch: [25] [1250/7269] eta: 0:43:49 lr: 0.000004 loss: 0.5035 (0.5196) time: 0.4321 data: 0.0002 max mem: 52150 +Epoch: [25] [1260/7269] eta: 0:43:42 lr: 0.000004 loss: 0.4930 (0.5194) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [25] [1270/7269] eta: 0:43:34 lr: 0.000004 loss: 0.5372 (0.5194) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [25] [1280/7269] eta: 0:43:27 lr: 0.000004 loss: 0.5480 (0.5195) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [25] [1290/7269] eta: 0:43:20 lr: 0.000004 loss: 0.5391 (0.5195) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [25] [1300/7269] eta: 0:43:17 lr: 0.000004 loss: 0.5213 (0.5195) time: 0.4169 data: 0.0001 max mem: 52150 +Epoch: [25] [1310/7269] eta: 0:43:14 lr: 0.000004 loss: 0.5310 (0.5197) time: 0.4632 data: 0.0001 max mem: 52150 +Epoch: [25] [1320/7269] eta: 0:43:06 lr: 0.000004 loss: 0.5356 (0.5196) time: 0.4192 data: 0.0001 max mem: 52150 +Epoch: [25] [1330/7269] eta: 0:42:59 lr: 0.000004 loss: 0.5416 (0.5197) time: 0.3728 data: 0.0001 max mem: 52150 +Epoch: [25] [1340/7269] eta: 0:42:52 lr: 0.000004 loss: 0.5471 (0.5200) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [25] [1350/7269] eta: 0:42:49 lr: 0.000004 loss: 0.5365 (0.5199) time: 0.4181 data: 0.0002 max mem: 52150 +Epoch: [25] [1360/7269] eta: 0:42:42 lr: 0.000004 loss: 0.5158 (0.5200) time: 0.4181 data: 0.0002 max mem: 52150 +Epoch: [25] [1370/7269] eta: 0:42:35 lr: 0.000004 loss: 0.5158 (0.5198) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [25] [1380/7269] eta: 0:42:28 lr: 0.000004 loss: 0.5127 (0.5199) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [25] [1390/7269] eta: 0:42:26 lr: 0.000004 loss: 0.5379 (0.5199) time: 0.4314 data: 0.0002 max mem: 52150 +Epoch: [25] [1400/7269] eta: 0:42:19 lr: 0.000004 loss: 0.5501 (0.5200) time: 0.4312 data: 0.0002 max mem: 52150 +Epoch: [25] [1410/7269] eta: 0:42:12 lr: 0.000004 loss: 0.5221 (0.5200) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [25] [1420/7269] eta: 0:42:06 lr: 0.000004 loss: 0.5198 (0.5201) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [25] [1430/7269] eta: 0:41:59 lr: 0.000004 loss: 0.5427 (0.5203) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [25] [1440/7269] eta: 0:42:01 lr: 0.000004 loss: 0.5427 (0.5204) time: 0.4797 data: 0.0002 max mem: 52150 +Epoch: [25] [1450/7269] eta: 0:41:59 lr: 0.000004 loss: 0.4998 (0.5202) time: 0.5407 data: 0.0002 max mem: 52150 +Epoch: [25] [1460/7269] eta: 0:41:52 lr: 0.000004 loss: 0.4998 (0.5200) time: 0.4327 data: 0.0002 max mem: 52150 +Epoch: [25] [1470/7269] eta: 0:41:49 lr: 0.000004 loss: 0.5258 (0.5202) time: 0.4162 data: 0.0002 max mem: 52150 +Epoch: [25] [1480/7269] eta: 0:41:42 lr: 0.000004 loss: 0.5244 (0.5202) time: 0.4166 data: 0.0002 max mem: 52150 +Epoch: [25] [1490/7269] eta: 0:41:36 lr: 0.000004 loss: 0.5220 (0.5201) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [25] [1500/7269] eta: 0:41:34 lr: 0.000004 loss: 0.4849 (0.5199) time: 0.4327 data: 0.0002 max mem: 52150 +Epoch: [25] [1510/7269] eta: 0:41:27 lr: 0.000004 loss: 0.4921 (0.5198) time: 0.4327 data: 0.0002 max mem: 52150 +Epoch: [25] [1520/7269] eta: 0:41:21 lr: 0.000004 loss: 0.5006 (0.5194) time: 0.3723 data: 0.0001 max mem: 52150 +Epoch: [25] [1530/7269] eta: 0:41:22 lr: 0.000004 loss: 0.4681 (0.5191) time: 0.4822 data: 0.0002 max mem: 52150 +Epoch: [25] [1540/7269] eta: 0:41:19 lr: 0.000004 loss: 0.4447 (0.5190) time: 0.5262 data: 0.0002 max mem: 52150 +Epoch: [25] [1550/7269] eta: 0:41:17 lr: 0.000004 loss: 0.4930 (0.5188) time: 0.4776 data: 0.0002 max mem: 52150 +Epoch: [25] [1560/7269] eta: 0:41:10 lr: 0.000004 loss: 0.4973 (0.5189) time: 0.4333 data: 0.0002 max mem: 52150 +Epoch: [25] [1570/7269] eta: 0:41:04 lr: 0.000004 loss: 0.5035 (0.5188) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [25] [1580/7269] eta: 0:40:57 lr: 0.000004 loss: 0.5330 (0.5190) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [25] [1590/7269] eta: 0:40:51 lr: 0.000004 loss: 0.5545 (0.5191) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [25] [1600/7269] eta: 0:40:48 lr: 0.000004 loss: 0.5310 (0.5190) time: 0.4170 data: 0.0001 max mem: 52150 +Epoch: [25] [1610/7269] eta: 0:40:45 lr: 0.000004 loss: 0.5018 (0.5190) time: 0.4777 data: 0.0001 max mem: 52150 +Epoch: [25] [1620/7269] eta: 0:40:39 lr: 0.000004 loss: 0.5477 (0.5192) time: 0.4324 data: 0.0001 max mem: 52150 +Epoch: [25] [1630/7269] eta: 0:40:33 lr: 0.000004 loss: 0.5511 (0.5192) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [25] [1640/7269] eta: 0:40:29 lr: 0.000004 loss: 0.5450 (0.5194) time: 0.4180 data: 0.0002 max mem: 52150 +Epoch: [25] [1650/7269] eta: 0:40:30 lr: 0.000004 loss: 0.5635 (0.5197) time: 0.5237 data: 0.0001 max mem: 52150 +Epoch: [25] [1660/7269] eta: 0:40:24 lr: 0.000004 loss: 0.5460 (0.5197) time: 0.4774 data: 0.0001 max mem: 52150 +Epoch: [25] [1670/7269] eta: 0:40:18 lr: 0.000004 loss: 0.5031 (0.5197) time: 0.3724 data: 0.0001 max mem: 52150 +Epoch: [25] [1680/7269] eta: 0:40:18 lr: 0.000004 loss: 0.5025 (0.5196) time: 0.4810 data: 0.0002 max mem: 52150 +Epoch: [25] [1690/7269] eta: 0:40:16 lr: 0.000004 loss: 0.5122 (0.5196) time: 0.5447 data: 0.0002 max mem: 52150 +Epoch: [25] [1700/7269] eta: 0:40:10 lr: 0.000004 loss: 0.5226 (0.5197) time: 0.4359 data: 0.0001 max mem: 52150 +Epoch: [25] [1710/7269] eta: 0:40:08 lr: 0.000004 loss: 0.5079 (0.5196) time: 0.4327 data: 0.0002 max mem: 52150 +Epoch: [25] [1720/7269] eta: 0:40:04 lr: 0.000004 loss: 0.5218 (0.5195) time: 0.4784 data: 0.0002 max mem: 52150 +Epoch: [25] [1730/7269] eta: 0:39:58 lr: 0.000004 loss: 0.5107 (0.5194) time: 0.4172 data: 0.0002 max mem: 52150 +Epoch: [25] [1740/7269] eta: 0:39:55 lr: 0.000004 loss: 0.5107 (0.5194) time: 0.4220 data: 0.0002 max mem: 52150 +Epoch: [25] [1750/7269] eta: 0:39:52 lr: 0.000004 loss: 0.5176 (0.5194) time: 0.4812 data: 0.0002 max mem: 52150 +Epoch: [25] [1760/7269] eta: 0:39:50 lr: 0.000004 loss: 0.5132 (0.5192) time: 0.4901 data: 0.0002 max mem: 52150 +Epoch: [25] [1770/7269] eta: 0:39:47 lr: 0.000004 loss: 0.4945 (0.5191) time: 0.4791 data: 0.0002 max mem: 52150 +Epoch: [25] [1780/7269] eta: 0:39:40 lr: 0.000004 loss: 0.5237 (0.5193) time: 0.4205 data: 0.0002 max mem: 52150 +Epoch: [25] [1790/7269] eta: 0:39:34 lr: 0.000004 loss: 0.5291 (0.5193) time: 0.3727 data: 0.0002 max mem: 52150 +Epoch: [25] [1800/7269] eta: 0:39:32 lr: 0.000004 loss: 0.5052 (0.5190) time: 0.4326 data: 0.0001 max mem: 52150 +Epoch: [25] [1810/7269] eta: 0:39:25 lr: 0.000004 loss: 0.4791 (0.5190) time: 0.4327 data: 0.0002 max mem: 52150 +Epoch: [25] [1820/7269] eta: 0:39:23 lr: 0.000004 loss: 0.5123 (0.5189) time: 0.4313 data: 0.0002 max mem: 52150 +Epoch: [25] [1830/7269] eta: 0:39:19 lr: 0.000004 loss: 0.5127 (0.5189) time: 0.4781 data: 0.0002 max mem: 52150 +Epoch: [25] [1840/7269] eta: 0:39:13 lr: 0.000004 loss: 0.4954 (0.5188) time: 0.4188 data: 0.0001 max mem: 52150 +Epoch: [25] [1850/7269] eta: 0:39:07 lr: 0.000004 loss: 0.4890 (0.5187) time: 0.3728 data: 0.0002 max mem: 52150 +Epoch: [25] [1860/7269] eta: 0:39:04 lr: 0.000004 loss: 0.4890 (0.5186) time: 0.4176 data: 0.0002 max mem: 52150 +Epoch: [25] [1870/7269] eta: 0:38:57 lr: 0.000004 loss: 0.4902 (0.5185) time: 0.4168 data: 0.0001 max mem: 52150 +Epoch: [25] [1880/7269] eta: 0:38:51 lr: 0.000004 loss: 0.5252 (0.5186) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [25] [1890/7269] eta: 0:38:45 lr: 0.000004 loss: 0.5408 (0.5186) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [25] [1900/7269] eta: 0:38:42 lr: 0.000004 loss: 0.5408 (0.5186) time: 0.4191 data: 0.0002 max mem: 52150 +Epoch: [25] [1910/7269] eta: 0:38:36 lr: 0.000004 loss: 0.5567 (0.5188) time: 0.4198 data: 0.0002 max mem: 52150 +Epoch: [25] [1920/7269] eta: 0:38:30 lr: 0.000004 loss: 0.5597 (0.5188) time: 0.3728 data: 0.0001 max mem: 52150 +Epoch: [25] [1930/7269] eta: 0:38:24 lr: 0.000004 loss: 0.4829 (0.5185) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [25] [1940/7269] eta: 0:38:18 lr: 0.000004 loss: 0.4813 (0.5184) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [25] [1950/7269] eta: 0:38:12 lr: 0.000004 loss: 0.5135 (0.5186) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [25] [1960/7269] eta: 0:38:06 lr: 0.000004 loss: 0.5585 (0.5187) time: 0.3723 data: 0.0001 max mem: 52150 +Epoch: [25] [1970/7269] eta: 0:38:04 lr: 0.000004 loss: 0.5577 (0.5188) time: 0.4339 data: 0.0001 max mem: 52150 +Epoch: [25] [1980/7269] eta: 0:37:58 lr: 0.000004 loss: 0.5303 (0.5189) time: 0.4338 data: 0.0002 max mem: 52150 +Epoch: [25] [1990/7269] eta: 0:38:01 lr: 0.000004 loss: 0.5238 (0.5189) time: 0.5399 data: 0.0002 max mem: 52150 +Epoch: [25] [2000/7269] eta: 0:37:58 lr: 0.000004 loss: 0.5215 (0.5189) time: 0.6003 data: 0.0001 max mem: 52150 +Epoch: [25] [2010/7269] eta: 0:37:52 lr: 0.000004 loss: 0.5194 (0.5189) time: 0.4321 data: 0.0001 max mem: 52150 +Epoch: [25] [2020/7269] eta: 0:37:49 lr: 0.000004 loss: 0.5194 (0.5189) time: 0.4323 data: 0.0002 max mem: 52150 +Epoch: [25] [2030/7269] eta: 0:37:44 lr: 0.000004 loss: 0.5437 (0.5190) time: 0.4324 data: 0.0002 max mem: 52150 +Epoch: [25] [2040/7269] eta: 0:37:38 lr: 0.000004 loss: 0.5534 (0.5192) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [25] [2050/7269] eta: 0:37:35 lr: 0.000004 loss: 0.5534 (0.5193) time: 0.4333 data: 0.0002 max mem: 52150 +Epoch: [25] [2060/7269] eta: 0:37:29 lr: 0.000004 loss: 0.5531 (0.5193) time: 0.4327 data: 0.0002 max mem: 52150 +Epoch: [25] [2070/7269] eta: 0:37:23 lr: 0.000004 loss: 0.5659 (0.5195) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [25] [2080/7269] eta: 0:37:20 lr: 0.000004 loss: 0.5518 (0.5194) time: 0.4175 data: 0.0002 max mem: 52150 +Epoch: [25] [2090/7269] eta: 0:37:16 lr: 0.000004 loss: 0.5095 (0.5193) time: 0.4625 data: 0.0001 max mem: 52150 +Epoch: [25] [2100/7269] eta: 0:37:10 lr: 0.000004 loss: 0.5222 (0.5194) time: 0.4172 data: 0.0002 max mem: 52150 +Epoch: [25] [2110/7269] eta: 0:37:05 lr: 0.000004 loss: 0.5395 (0.5195) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [25] [2120/7269] eta: 0:37:04 lr: 0.000004 loss: 0.5357 (0.5196) time: 0.4788 data: 0.0001 max mem: 52150 +Epoch: [25] [2130/7269] eta: 0:36:58 lr: 0.000004 loss: 0.5320 (0.5196) time: 0.4785 data: 0.0002 max mem: 52150 +Epoch: [25] [2140/7269] eta: 0:36:53 lr: 0.000004 loss: 0.5164 (0.5195) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [25] [2150/7269] eta: 0:36:47 lr: 0.000004 loss: 0.5074 (0.5195) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [25] [2160/7269] eta: 0:36:44 lr: 0.000004 loss: 0.5355 (0.5196) time: 0.4326 data: 0.0001 max mem: 52150 +Epoch: [25] [2170/7269] eta: 0:36:43 lr: 0.000004 loss: 0.5355 (0.5194) time: 0.5294 data: 0.0002 max mem: 52150 +Epoch: [25] [2180/7269] eta: 0:36:37 lr: 0.000004 loss: 0.5092 (0.5195) time: 0.4688 data: 0.0002 max mem: 52150 +Epoch: [25] [2190/7269] eta: 0:36:31 lr: 0.000004 loss: 0.4914 (0.5193) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [25] [2200/7269] eta: 0:36:28 lr: 0.000004 loss: 0.5091 (0.5194) time: 0.4208 data: 0.0002 max mem: 52150 +Epoch: [25] [2210/7269] eta: 0:36:22 lr: 0.000004 loss: 0.5308 (0.5195) time: 0.4209 data: 0.0002 max mem: 52150 +Epoch: [25] [2220/7269] eta: 0:36:17 lr: 0.000004 loss: 0.5504 (0.5195) time: 0.3730 data: 0.0002 max mem: 52150 +Epoch: [25] [2230/7269] eta: 0:36:11 lr: 0.000004 loss: 0.5333 (0.5196) time: 0.3727 data: 0.0002 max mem: 52150 +Epoch: [25] [2240/7269] eta: 0:36:13 lr: 0.000004 loss: 0.5002 (0.5195) time: 0.5377 data: 0.0002 max mem: 52150 +Epoch: [25] [2250/7269] eta: 0:36:07 lr: 0.000004 loss: 0.5128 (0.5196) time: 0.5376 data: 0.0001 max mem: 52150 +Epoch: [25] [2260/7269] eta: 0:36:04 lr: 0.000004 loss: 0.5343 (0.5195) time: 0.4176 data: 0.0002 max mem: 52150 +Epoch: [25] [2270/7269] eta: 0:35:58 lr: 0.000004 loss: 0.5283 (0.5195) time: 0.4173 data: 0.0001 max mem: 52150 +Epoch: [25] [2280/7269] eta: 0:35:57 lr: 0.000004 loss: 0.5128 (0.5194) time: 0.4828 data: 0.0002 max mem: 52150 +Epoch: [25] [2290/7269] eta: 0:35:51 lr: 0.000004 loss: 0.5122 (0.5194) time: 0.4831 data: 0.0002 max mem: 52150 +Epoch: [25] [2300/7269] eta: 0:35:51 lr: 0.000004 loss: 0.5293 (0.5193) time: 0.4796 data: 0.0002 max mem: 52150 +Epoch: [25] [2310/7269] eta: 0:35:50 lr: 0.000004 loss: 0.5292 (0.5194) time: 0.5994 data: 0.0001 max mem: 52150 +Epoch: [25] [2320/7269] eta: 0:35:44 lr: 0.000004 loss: 0.5163 (0.5193) time: 0.4915 data: 0.0001 max mem: 52150 +Epoch: [25] [2330/7269] eta: 0:35:43 lr: 0.000004 loss: 0.5163 (0.5193) time: 0.4638 data: 0.0001 max mem: 52150 +Epoch: [25] [2340/7269] eta: 0:35:37 lr: 0.000004 loss: 0.5119 (0.5193) time: 0.4649 data: 0.0001 max mem: 52150 +Epoch: [25] [2350/7269] eta: 0:35:31 lr: 0.000004 loss: 0.5222 (0.5194) time: 0.3729 data: 0.0001 max mem: 52150 +Epoch: [25] [2360/7269] eta: 0:35:26 lr: 0.000004 loss: 0.5652 (0.5195) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [25] [2370/7269] eta: 0:35:20 lr: 0.000004 loss: 0.5652 (0.5195) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [25] [2380/7269] eta: 0:35:17 lr: 0.000004 loss: 0.4853 (0.5194) time: 0.4186 data: 0.0001 max mem: 52150 +Epoch: [25] [2390/7269] eta: 0:35:11 lr: 0.000004 loss: 0.4785 (0.5192) time: 0.4186 data: 0.0001 max mem: 52150 +Epoch: [25] [2400/7269] eta: 0:35:07 lr: 0.000004 loss: 0.5231 (0.5193) time: 0.4168 data: 0.0002 max mem: 52150 +Epoch: [25] [2410/7269] eta: 0:35:04 lr: 0.000004 loss: 0.5277 (0.5193) time: 0.4763 data: 0.0002 max mem: 52150 +Epoch: [25] [2420/7269] eta: 0:34:58 lr: 0.000004 loss: 0.5237 (0.5194) time: 0.4314 data: 0.0001 max mem: 52150 +Epoch: [25] [2430/7269] eta: 0:34:55 lr: 0.000004 loss: 0.5456 (0.5194) time: 0.4331 data: 0.0001 max mem: 52150 +Epoch: [25] [2440/7269] eta: 0:34:55 lr: 0.000004 loss: 0.5534 (0.5195) time: 0.5526 data: 0.0002 max mem: 52150 +Epoch: [25] [2450/7269] eta: 0:34:51 lr: 0.000004 loss: 0.5229 (0.5195) time: 0.5525 data: 0.0001 max mem: 52150 +Epoch: [25] [2460/7269] eta: 0:34:46 lr: 0.000004 loss: 0.5175 (0.5194) time: 0.4325 data: 0.0001 max mem: 52150 +Epoch: [25] [2470/7269] eta: 0:34:40 lr: 0.000004 loss: 0.5042 (0.5193) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [25] [2480/7269] eta: 0:34:35 lr: 0.000004 loss: 0.5042 (0.5193) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [25] [2490/7269] eta: 0:34:32 lr: 0.000004 loss: 0.5199 (0.5193) time: 0.4332 data: 0.0002 max mem: 52150 +Epoch: [25] [2500/7269] eta: 0:34:26 lr: 0.000004 loss: 0.5199 (0.5193) time: 0.4327 data: 0.0002 max mem: 52150 +Epoch: [25] [2510/7269] eta: 0:34:22 lr: 0.000004 loss: 0.5505 (0.5194) time: 0.4180 data: 0.0002 max mem: 52150 +Epoch: [25] [2520/7269] eta: 0:34:17 lr: 0.000004 loss: 0.5276 (0.5193) time: 0.4181 data: 0.0002 max mem: 52150 +Epoch: [25] [2530/7269] eta: 0:34:11 lr: 0.000004 loss: 0.5020 (0.5193) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [25] [2540/7269] eta: 0:34:06 lr: 0.000004 loss: 0.5325 (0.5194) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [25] [2550/7269] eta: 0:34:03 lr: 0.000004 loss: 0.5307 (0.5193) time: 0.4321 data: 0.0002 max mem: 52150 +Epoch: [25] [2560/7269] eta: 0:33:57 lr: 0.000004 loss: 0.5154 (0.5193) time: 0.4327 data: 0.0002 max mem: 52150 +Epoch: [25] [2570/7269] eta: 0:33:54 lr: 0.000004 loss: 0.4972 (0.5193) time: 0.4325 data: 0.0002 max mem: 52150 +Epoch: [25] [2580/7269] eta: 0:33:50 lr: 0.000004 loss: 0.5060 (0.5193) time: 0.4792 data: 0.0002 max mem: 52150 +Epoch: [25] [2590/7269] eta: 0:33:45 lr: 0.000004 loss: 0.5060 (0.5193) time: 0.4194 data: 0.0002 max mem: 52150 +Epoch: [25] [2600/7269] eta: 0:33:39 lr: 0.000004 loss: 0.5031 (0.5192) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [25] [2610/7269] eta: 0:33:34 lr: 0.000004 loss: 0.5335 (0.5193) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [25] [2620/7269] eta: 0:33:30 lr: 0.000004 loss: 0.5360 (0.5195) time: 0.4177 data: 0.0002 max mem: 52150 +Epoch: [25] [2630/7269] eta: 0:33:25 lr: 0.000004 loss: 0.5140 (0.5193) time: 0.4175 data: 0.0002 max mem: 52150 +Epoch: [25] [2640/7269] eta: 0:33:21 lr: 0.000004 loss: 0.5095 (0.5194) time: 0.4170 data: 0.0002 max mem: 52150 +Epoch: [25] [2650/7269] eta: 0:33:16 lr: 0.000004 loss: 0.5251 (0.5194) time: 0.4175 data: 0.0001 max mem: 52150 +Epoch: [25] [2660/7269] eta: 0:33:12 lr: 0.000004 loss: 0.5238 (0.5194) time: 0.4220 data: 0.0002 max mem: 52150 +Epoch: [25] [2670/7269] eta: 0:33:09 lr: 0.000004 loss: 0.5166 (0.5193) time: 0.4806 data: 0.0002 max mem: 52150 +Epoch: [25] [2680/7269] eta: 0:33:05 lr: 0.000004 loss: 0.4744 (0.5192) time: 0.4919 data: 0.0002 max mem: 52150 +Epoch: [25] [2690/7269] eta: 0:33:00 lr: 0.000004 loss: 0.5260 (0.5192) time: 0.4331 data: 0.0001 max mem: 52150 +Epoch: [25] [2700/7269] eta: 0:32:55 lr: 0.000004 loss: 0.4703 (0.5190) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [25] [2710/7269] eta: 0:32:52 lr: 0.000004 loss: 0.5030 (0.5191) time: 0.4350 data: 0.0002 max mem: 52150 +Epoch: [25] [2720/7269] eta: 0:32:48 lr: 0.000004 loss: 0.5548 (0.5191) time: 0.4817 data: 0.0002 max mem: 52150 +Epoch: [25] [2730/7269] eta: 0:32:44 lr: 0.000004 loss: 0.5410 (0.5192) time: 0.4780 data: 0.0002 max mem: 52150 +Epoch: [25] [2740/7269] eta: 0:32:41 lr: 0.000004 loss: 0.5280 (0.5192) time: 0.4943 data: 0.0002 max mem: 52150 +Epoch: [25] [2750/7269] eta: 0:32:36 lr: 0.000004 loss: 0.5144 (0.5192) time: 0.4352 data: 0.0002 max mem: 52150 +Epoch: [25] [2760/7269] eta: 0:32:32 lr: 0.000004 loss: 0.5144 (0.5192) time: 0.4175 data: 0.0002 max mem: 52150 +Epoch: [25] [2770/7269] eta: 0:32:33 lr: 0.000004 loss: 0.5315 (0.5191) time: 0.6029 data: 0.0002 max mem: 52150 +Epoch: [25] [2780/7269] eta: 0:32:29 lr: 0.000004 loss: 0.5410 (0.5192) time: 0.6094 data: 0.0002 max mem: 52150 +Epoch: [25] [2790/7269] eta: 0:32:28 lr: 0.000004 loss: 0.5478 (0.5192) time: 0.5737 data: 0.0002 max mem: 52150 +Epoch: [25] [2800/7269] eta: 0:32:23 lr: 0.000004 loss: 0.5215 (0.5192) time: 0.5216 data: 0.0002 max mem: 52150 +Epoch: [25] [2810/7269] eta: 0:32:18 lr: 0.000004 loss: 0.5215 (0.5192) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [25] [2820/7269] eta: 0:32:12 lr: 0.000004 loss: 0.5386 (0.5192) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [25] [2830/7269] eta: 0:32:07 lr: 0.000004 loss: 0.5518 (0.5194) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [25] [2840/7269] eta: 0:32:02 lr: 0.000004 loss: 0.5447 (0.5193) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [25] [2850/7269] eta: 0:31:58 lr: 0.000004 loss: 0.5175 (0.5193) time: 0.4332 data: 0.0001 max mem: 52150 +Epoch: [25] [2860/7269] eta: 0:31:54 lr: 0.000004 loss: 0.4992 (0.5192) time: 0.4783 data: 0.0001 max mem: 52150 +Epoch: [25] [2870/7269] eta: 0:31:49 lr: 0.000004 loss: 0.4929 (0.5192) time: 0.4166 data: 0.0001 max mem: 52150 +Epoch: [25] [2880/7269] eta: 0:31:46 lr: 0.000004 loss: 0.5153 (0.5192) time: 0.4307 data: 0.0001 max mem: 52150 +Epoch: [25] [2890/7269] eta: 0:31:40 lr: 0.000004 loss: 0.5389 (0.5192) time: 0.4312 data: 0.0001 max mem: 52150 +Epoch: [25] [2900/7269] eta: 0:31:35 lr: 0.000004 loss: 0.5482 (0.5193) time: 0.3724 data: 0.0001 max mem: 52150 +Epoch: [25] [2910/7269] eta: 0:31:32 lr: 0.000004 loss: 0.5476 (0.5193) time: 0.4312 data: 0.0001 max mem: 52150 +Epoch: [25] [2920/7269] eta: 0:31:28 lr: 0.000004 loss: 0.5368 (0.5193) time: 0.4806 data: 0.0001 max mem: 52150 +Epoch: [25] [2930/7269] eta: 0:31:23 lr: 0.000004 loss: 0.5156 (0.5193) time: 0.4216 data: 0.0001 max mem: 52150 +Epoch: [25] [2940/7269] eta: 0:31:17 lr: 0.000004 loss: 0.5275 (0.5193) time: 0.3722 data: 0.0001 max mem: 52150 +Epoch: [25] [2950/7269] eta: 0:31:13 lr: 0.000004 loss: 0.5345 (0.5193) time: 0.4184 data: 0.0002 max mem: 52150 +Epoch: [25] [2960/7269] eta: 0:31:08 lr: 0.000004 loss: 0.5403 (0.5194) time: 0.4183 data: 0.0002 max mem: 52150 +Epoch: [25] [2970/7269] eta: 0:31:03 lr: 0.000004 loss: 0.5503 (0.5195) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [25] [2980/7269] eta: 0:30:59 lr: 0.000004 loss: 0.5384 (0.5195) time: 0.4326 data: 0.0002 max mem: 52150 +Epoch: [25] [2990/7269] eta: 0:30:54 lr: 0.000004 loss: 0.5331 (0.5195) time: 0.4322 data: 0.0001 max mem: 52150 +Epoch: [25] [3000/7269] eta: 0:30:49 lr: 0.000004 loss: 0.5331 (0.5195) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [25] [3010/7269] eta: 0:30:47 lr: 0.000004 loss: 0.5265 (0.5195) time: 0.4790 data: 0.0002 max mem: 52150 +Epoch: [25] [3020/7269] eta: 0:30:42 lr: 0.000004 loss: 0.5507 (0.5196) time: 0.4789 data: 0.0002 max mem: 52150 +Epoch: [25] [3030/7269] eta: 0:30:36 lr: 0.000004 loss: 0.5425 (0.5196) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [25] [3040/7269] eta: 0:30:31 lr: 0.000004 loss: 0.5156 (0.5196) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [25] [3050/7269] eta: 0:30:27 lr: 0.000004 loss: 0.5298 (0.5197) time: 0.4182 data: 0.0002 max mem: 52150 +Epoch: [25] [3060/7269] eta: 0:30:22 lr: 0.000004 loss: 0.5298 (0.5197) time: 0.4185 data: 0.0002 max mem: 52150 +Epoch: [25] [3070/7269] eta: 0:30:17 lr: 0.000004 loss: 0.5123 (0.5197) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [25] [3080/7269] eta: 0:30:14 lr: 0.000004 loss: 0.5172 (0.5198) time: 0.4319 data: 0.0002 max mem: 52150 +Epoch: [25] [3090/7269] eta: 0:30:10 lr: 0.000004 loss: 0.5400 (0.5198) time: 0.4931 data: 0.0002 max mem: 52150 +Epoch: [25] [3100/7269] eta: 0:30:07 lr: 0.000004 loss: 0.5381 (0.5198) time: 0.4960 data: 0.0001 max mem: 52150 +Epoch: [25] [3110/7269] eta: 0:30:01 lr: 0.000004 loss: 0.5461 (0.5199) time: 0.4349 data: 0.0001 max mem: 52150 +Epoch: [25] [3120/7269] eta: 0:29:57 lr: 0.000004 loss: 0.5447 (0.5198) time: 0.4183 data: 0.0002 max mem: 52150 +Epoch: [25] [3130/7269] eta: 0:29:54 lr: 0.000004 loss: 0.4872 (0.5197) time: 0.4783 data: 0.0002 max mem: 52150 +Epoch: [25] [3140/7269] eta: 0:29:49 lr: 0.000004 loss: 0.5006 (0.5197) time: 0.4322 data: 0.0002 max mem: 52150 +Epoch: [25] [3150/7269] eta: 0:29:44 lr: 0.000004 loss: 0.5174 (0.5198) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [25] [3160/7269] eta: 0:29:39 lr: 0.000004 loss: 0.5588 (0.5199) time: 0.3712 data: 0.0002 max mem: 52150 +Epoch: [25] [3170/7269] eta: 0:29:35 lr: 0.000004 loss: 0.5340 (0.5198) time: 0.4213 data: 0.0001 max mem: 52150 +Epoch: [25] [3180/7269] eta: 0:29:31 lr: 0.000004 loss: 0.5229 (0.5198) time: 0.4668 data: 0.0002 max mem: 52150 +Epoch: [25] [3190/7269] eta: 0:29:26 lr: 0.000004 loss: 0.5419 (0.5198) time: 0.4173 data: 0.0002 max mem: 52150 +Epoch: [25] [3200/7269] eta: 0:29:20 lr: 0.000004 loss: 0.5311 (0.5198) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [25] [3210/7269] eta: 0:29:19 lr: 0.000004 loss: 0.5172 (0.5197) time: 0.4963 data: 0.0001 max mem: 52150 +Epoch: [25] [3220/7269] eta: 0:29:13 lr: 0.000004 loss: 0.5172 (0.5198) time: 0.4971 data: 0.0001 max mem: 52150 +Epoch: [25] [3230/7269] eta: 0:29:08 lr: 0.000004 loss: 0.5086 (0.5198) time: 0.3727 data: 0.0001 max mem: 52150 +Epoch: [25] [3240/7269] eta: 0:29:03 lr: 0.000004 loss: 0.4855 (0.5197) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [25] [3250/7269] eta: 0:28:58 lr: 0.000004 loss: 0.4819 (0.5197) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [25] [3260/7269] eta: 0:28:55 lr: 0.000004 loss: 0.5213 (0.5198) time: 0.4667 data: 0.0002 max mem: 52150 +Epoch: [25] [3270/7269] eta: 0:28:52 lr: 0.000004 loss: 0.5203 (0.5197) time: 0.5253 data: 0.0001 max mem: 52150 +Epoch: [25] [3280/7269] eta: 0:28:47 lr: 0.000004 loss: 0.5095 (0.5197) time: 0.4311 data: 0.0002 max mem: 52150 +Epoch: [25] [3290/7269] eta: 0:28:42 lr: 0.000004 loss: 0.5084 (0.5196) time: 0.3732 data: 0.0002 max mem: 52150 +Epoch: [25] [3300/7269] eta: 0:28:37 lr: 0.000004 loss: 0.5070 (0.5196) time: 0.3734 data: 0.0002 max mem: 52150 +Epoch: [25] [3310/7269] eta: 0:28:33 lr: 0.000004 loss: 0.4967 (0.5195) time: 0.4327 data: 0.0002 max mem: 52150 +Epoch: [25] [3320/7269] eta: 0:28:28 lr: 0.000004 loss: 0.5000 (0.5195) time: 0.4323 data: 0.0002 max mem: 52150 +Epoch: [25] [3330/7269] eta: 0:28:26 lr: 0.000004 loss: 0.5089 (0.5195) time: 0.4956 data: 0.0002 max mem: 52150 +Epoch: [25] [3340/7269] eta: 0:28:23 lr: 0.000004 loss: 0.5265 (0.5195) time: 0.5874 data: 0.0002 max mem: 52150 +Epoch: [25] [3350/7269] eta: 0:28:19 lr: 0.000004 loss: 0.5105 (0.5194) time: 0.5273 data: 0.0001 max mem: 52150 +Epoch: [25] [3360/7269] eta: 0:28:14 lr: 0.000004 loss: 0.5100 (0.5194) time: 0.4354 data: 0.0002 max mem: 52150 +Epoch: [25] [3370/7269] eta: 0:28:09 lr: 0.000004 loss: 0.5102 (0.5194) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [25] [3380/7269] eta: 0:28:05 lr: 0.000004 loss: 0.5101 (0.5193) time: 0.4177 data: 0.0002 max mem: 52150 +Epoch: [25] [3390/7269] eta: 0:28:03 lr: 0.000004 loss: 0.5089 (0.5193) time: 0.5384 data: 0.0002 max mem: 52150 +Epoch: [25] [3400/7269] eta: 0:28:00 lr: 0.000004 loss: 0.5436 (0.5194) time: 0.5901 data: 0.0002 max mem: 52150 +Epoch: [25] [3410/7269] eta: 0:27:57 lr: 0.000004 loss: 0.5436 (0.5193) time: 0.5350 data: 0.0002 max mem: 52150 +Epoch: [25] [3420/7269] eta: 0:27:52 lr: 0.000004 loss: 0.5223 (0.5194) time: 0.4381 data: 0.0002 max mem: 52150 +Epoch: [25] [3430/7269] eta: 0:27:47 lr: 0.000004 loss: 0.5303 (0.5194) time: 0.3723 data: 0.0001 max mem: 52150 +Epoch: [25] [3440/7269] eta: 0:27:42 lr: 0.000004 loss: 0.5579 (0.5195) time: 0.3721 data: 0.0001 max mem: 52150 +Epoch: [25] [3450/7269] eta: 0:27:36 lr: 0.000004 loss: 0.5450 (0.5196) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [25] [3460/7269] eta: 0:27:32 lr: 0.000004 loss: 0.5068 (0.5195) time: 0.4175 data: 0.0002 max mem: 52150 +Epoch: [25] [3470/7269] eta: 0:27:27 lr: 0.000004 loss: 0.4889 (0.5194) time: 0.4177 data: 0.0002 max mem: 52150 +Epoch: [25] [3480/7269] eta: 0:27:23 lr: 0.000004 loss: 0.5131 (0.5194) time: 0.4178 data: 0.0002 max mem: 52150 +Epoch: [25] [3490/7269] eta: 0:27:18 lr: 0.000004 loss: 0.5202 (0.5195) time: 0.4176 data: 0.0001 max mem: 52150 +Epoch: [25] [3500/7269] eta: 0:27:15 lr: 0.000004 loss: 0.5318 (0.5195) time: 0.4318 data: 0.0001 max mem: 52150 +Epoch: [25] [3510/7269] eta: 0:27:10 lr: 0.000004 loss: 0.5392 (0.5195) time: 0.4327 data: 0.0002 max mem: 52150 +Epoch: [25] [3520/7269] eta: 0:27:05 lr: 0.000004 loss: 0.5127 (0.5194) time: 0.3736 data: 0.0002 max mem: 52150 +Epoch: [25] [3530/7269] eta: 0:27:01 lr: 0.000004 loss: 0.5127 (0.5194) time: 0.4319 data: 0.0002 max mem: 52150 +Epoch: [25] [3540/7269] eta: 0:26:56 lr: 0.000004 loss: 0.5415 (0.5195) time: 0.4317 data: 0.0002 max mem: 52150 +Epoch: [25] [3550/7269] eta: 0:26:51 lr: 0.000004 loss: 0.5296 (0.5195) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [25] [3560/7269] eta: 0:26:46 lr: 0.000004 loss: 0.5235 (0.5195) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [25] [3570/7269] eta: 0:26:42 lr: 0.000004 loss: 0.5235 (0.5195) time: 0.4325 data: 0.0001 max mem: 52150 +Epoch: [25] [3580/7269] eta: 0:26:37 lr: 0.000004 loss: 0.4977 (0.5194) time: 0.4324 data: 0.0001 max mem: 52150 +Epoch: [25] [3590/7269] eta: 0:26:32 lr: 0.000004 loss: 0.5011 (0.5194) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [25] [3600/7269] eta: 0:26:28 lr: 0.000004 loss: 0.4873 (0.5193) time: 0.3727 data: 0.0002 max mem: 52150 +Epoch: [25] [3610/7269] eta: 0:26:23 lr: 0.000004 loss: 0.4682 (0.5192) time: 0.3742 data: 0.0002 max mem: 52150 +Epoch: [25] [3620/7269] eta: 0:26:19 lr: 0.000004 loss: 0.5351 (0.5193) time: 0.4347 data: 0.0002 max mem: 52150 +Epoch: [25] [3630/7269] eta: 0:26:15 lr: 0.000004 loss: 0.5414 (0.5192) time: 0.4786 data: 0.0002 max mem: 52150 +Epoch: [25] [3640/7269] eta: 0:26:10 lr: 0.000004 loss: 0.5379 (0.5192) time: 0.4172 data: 0.0002 max mem: 52150 +Epoch: [25] [3650/7269] eta: 0:26:06 lr: 0.000004 loss: 0.5069 (0.5191) time: 0.4341 data: 0.0002 max mem: 52150 +Epoch: [25] [3660/7269] eta: 0:26:06 lr: 0.000004 loss: 0.5195 (0.5192) time: 0.6485 data: 0.0002 max mem: 52150 +Epoch: [25] [3670/7269] eta: 0:26:01 lr: 0.000004 loss: 0.5199 (0.5191) time: 0.5865 data: 0.0001 max mem: 52150 +Epoch: [25] [3680/7269] eta: 0:25:57 lr: 0.000004 loss: 0.5013 (0.5190) time: 0.4335 data: 0.0002 max mem: 52150 +Epoch: [25] [3690/7269] eta: 0:25:52 lr: 0.000004 loss: 0.5030 (0.5190) time: 0.4333 data: 0.0002 max mem: 52150 +Epoch: [25] [3700/7269] eta: 0:25:48 lr: 0.000004 loss: 0.4928 (0.5189) time: 0.4209 data: 0.0002 max mem: 52150 +Epoch: [25] [3710/7269] eta: 0:25:43 lr: 0.000004 loss: 0.5232 (0.5190) time: 0.4210 data: 0.0002 max mem: 52150 +Epoch: [25] [3720/7269] eta: 0:25:38 lr: 0.000004 loss: 0.5569 (0.5190) time: 0.3722 data: 0.0001 max mem: 52150 +Epoch: [25] [3730/7269] eta: 0:25:34 lr: 0.000004 loss: 0.5342 (0.5189) time: 0.4353 data: 0.0002 max mem: 52150 +Epoch: [25] [3740/7269] eta: 0:25:29 lr: 0.000004 loss: 0.5367 (0.5190) time: 0.4354 data: 0.0001 max mem: 52150 +Epoch: [25] [3750/7269] eta: 0:25:25 lr: 0.000004 loss: 0.5367 (0.5190) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [25] [3760/7269] eta: 0:25:20 lr: 0.000004 loss: 0.5342 (0.5190) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [25] [3770/7269] eta: 0:25:15 lr: 0.000004 loss: 0.5213 (0.5190) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [25] [3780/7269] eta: 0:25:11 lr: 0.000004 loss: 0.4898 (0.5189) time: 0.4344 data: 0.0001 max mem: 52150 +Epoch: [25] [3790/7269] eta: 0:25:08 lr: 0.000004 loss: 0.4989 (0.5189) time: 0.5467 data: 0.0001 max mem: 52150 +Epoch: [25] [3800/7269] eta: 0:25:03 lr: 0.000004 loss: 0.5399 (0.5189) time: 0.4840 data: 0.0002 max mem: 52150 +Epoch: [25] [3810/7269] eta: 0:24:58 lr: 0.000004 loss: 0.5399 (0.5189) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [25] [3820/7269] eta: 0:24:54 lr: 0.000004 loss: 0.5309 (0.5189) time: 0.4227 data: 0.0002 max mem: 52150 +Epoch: [25] [3830/7269] eta: 0:24:49 lr: 0.000004 loss: 0.5014 (0.5189) time: 0.4235 data: 0.0002 max mem: 52150 +Epoch: [25] [3840/7269] eta: 0:24:45 lr: 0.000004 loss: 0.5046 (0.5189) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [25] [3850/7269] eta: 0:24:40 lr: 0.000004 loss: 0.5453 (0.5189) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [25] [3860/7269] eta: 0:24:36 lr: 0.000004 loss: 0.5395 (0.5190) time: 0.4357 data: 0.0001 max mem: 52150 +Epoch: [25] [3870/7269] eta: 0:24:31 lr: 0.000004 loss: 0.5323 (0.5189) time: 0.4360 data: 0.0002 max mem: 52150 +Epoch: [25] [3880/7269] eta: 0:24:26 lr: 0.000004 loss: 0.5058 (0.5189) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [25] [3890/7269] eta: 0:24:21 lr: 0.000004 loss: 0.5110 (0.5189) time: 0.3713 data: 0.0001 max mem: 52150 +Epoch: [25] [3900/7269] eta: 0:24:18 lr: 0.000004 loss: 0.5644 (0.5190) time: 0.4687 data: 0.0001 max mem: 52150 +Epoch: [25] [3910/7269] eta: 0:24:13 lr: 0.000004 loss: 0.5644 (0.5189) time: 0.4688 data: 0.0001 max mem: 52150 +Epoch: [25] [3920/7269] eta: 0:24:11 lr: 0.000004 loss: 0.4947 (0.5188) time: 0.4939 data: 0.0001 max mem: 52150 +Epoch: [25] [3930/7269] eta: 0:24:07 lr: 0.000004 loss: 0.5177 (0.5189) time: 0.5591 data: 0.0001 max mem: 52150 +Epoch: [25] [3940/7269] eta: 0:24:04 lr: 0.000004 loss: 0.5261 (0.5189) time: 0.5460 data: 0.0002 max mem: 52150 +Epoch: [25] [3950/7269] eta: 0:24:00 lr: 0.000004 loss: 0.5378 (0.5189) time: 0.5283 data: 0.0002 max mem: 52150 +Epoch: [25] [3960/7269] eta: 0:23:57 lr: 0.000004 loss: 0.5013 (0.5188) time: 0.5393 data: 0.0001 max mem: 52150 +Epoch: [25] [3970/7269] eta: 0:23:53 lr: 0.000004 loss: 0.5071 (0.5188) time: 0.5515 data: 0.0002 max mem: 52150 +Epoch: [25] [3980/7269] eta: 0:23:49 lr: 0.000004 loss: 0.5133 (0.5187) time: 0.4909 data: 0.0002 max mem: 52150 +Epoch: [25] [3990/7269] eta: 0:23:45 lr: 0.000004 loss: 0.5254 (0.5188) time: 0.4774 data: 0.0002 max mem: 52150 +Epoch: [25] [4000/7269] eta: 0:23:40 lr: 0.000004 loss: 0.5519 (0.5189) time: 0.4178 data: 0.0002 max mem: 52150 +Epoch: [25] [4010/7269] eta: 0:23:35 lr: 0.000004 loss: 0.5349 (0.5188) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [25] [4020/7269] eta: 0:23:30 lr: 0.000004 loss: 0.4891 (0.5188) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [25] [4030/7269] eta: 0:23:26 lr: 0.000004 loss: 0.5226 (0.5188) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [25] [4040/7269] eta: 0:23:21 lr: 0.000004 loss: 0.5226 (0.5188) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [25] [4050/7269] eta: 0:23:16 lr: 0.000004 loss: 0.5105 (0.5187) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [25] [4060/7269] eta: 0:23:11 lr: 0.000004 loss: 0.5058 (0.5187) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [25] [4070/7269] eta: 0:23:06 lr: 0.000004 loss: 0.5032 (0.5187) time: 0.3722 data: 0.0001 max mem: 52150 +Epoch: [25] [4080/7269] eta: 0:23:02 lr: 0.000004 loss: 0.5193 (0.5187) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [25] [4090/7269] eta: 0:22:57 lr: 0.000004 loss: 0.5193 (0.5187) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [25] [4100/7269] eta: 0:22:52 lr: 0.000004 loss: 0.5158 (0.5188) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [25] [4110/7269] eta: 0:22:47 lr: 0.000004 loss: 0.5158 (0.5187) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [25] [4120/7269] eta: 0:22:46 lr: 0.000004 loss: 0.5266 (0.5188) time: 0.5897 data: 0.0002 max mem: 52150 +Epoch: [25] [4130/7269] eta: 0:22:41 lr: 0.000004 loss: 0.5225 (0.5187) time: 0.5893 data: 0.0002 max mem: 52150 +Epoch: [25] [4140/7269] eta: 0:22:37 lr: 0.000004 loss: 0.5225 (0.5187) time: 0.4311 data: 0.0001 max mem: 52150 +Epoch: [25] [4150/7269] eta: 0:22:32 lr: 0.000004 loss: 0.5561 (0.5188) time: 0.4310 data: 0.0001 max mem: 52150 +Epoch: [25] [4160/7269] eta: 0:22:28 lr: 0.000004 loss: 0.5201 (0.5188) time: 0.4194 data: 0.0002 max mem: 52150 +Epoch: [25] [4170/7269] eta: 0:22:23 lr: 0.000004 loss: 0.5285 (0.5188) time: 0.4192 data: 0.0002 max mem: 52150 +Epoch: [25] [4180/7269] eta: 0:22:19 lr: 0.000004 loss: 0.5350 (0.5188) time: 0.4339 data: 0.0001 max mem: 52150 +Epoch: [25] [4190/7269] eta: 0:22:15 lr: 0.000004 loss: 0.5620 (0.5189) time: 0.4795 data: 0.0001 max mem: 52150 +Epoch: [25] [4200/7269] eta: 0:22:10 lr: 0.000004 loss: 0.5225 (0.5188) time: 0.4173 data: 0.0001 max mem: 52150 +Epoch: [25] [4210/7269] eta: 0:22:06 lr: 0.000004 loss: 0.5081 (0.5188) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [25] [4220/7269] eta: 0:22:01 lr: 0.000004 loss: 0.5105 (0.5188) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [25] [4230/7269] eta: 0:21:56 lr: 0.000004 loss: 0.5312 (0.5188) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [25] [4240/7269] eta: 0:21:52 lr: 0.000004 loss: 0.5423 (0.5189) time: 0.4217 data: 0.0002 max mem: 52150 +Epoch: [25] [4250/7269] eta: 0:21:47 lr: 0.000004 loss: 0.5426 (0.5189) time: 0.4218 data: 0.0002 max mem: 52150 +Epoch: [25] [4260/7269] eta: 0:21:43 lr: 0.000004 loss: 0.5249 (0.5189) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [25] [4270/7269] eta: 0:21:38 lr: 0.000004 loss: 0.5243 (0.5189) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [25] [4280/7269] eta: 0:21:34 lr: 0.000004 loss: 0.5040 (0.5189) time: 0.4778 data: 0.0002 max mem: 52150 +Epoch: [25] [4290/7269] eta: 0:21:30 lr: 0.000004 loss: 0.4962 (0.5187) time: 0.4778 data: 0.0002 max mem: 52150 +Epoch: [25] [4300/7269] eta: 0:21:25 lr: 0.000004 loss: 0.5142 (0.5187) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [25] [4310/7269] eta: 0:21:20 lr: 0.000004 loss: 0.5142 (0.5187) time: 0.3729 data: 0.0002 max mem: 52150 +Epoch: [25] [4320/7269] eta: 0:21:17 lr: 0.000004 loss: 0.5086 (0.5187) time: 0.4972 data: 0.0002 max mem: 52150 +Epoch: [25] [4330/7269] eta: 0:21:13 lr: 0.000004 loss: 0.5511 (0.5187) time: 0.5588 data: 0.0002 max mem: 52150 +Epoch: [25] [4340/7269] eta: 0:21:09 lr: 0.000004 loss: 0.5385 (0.5187) time: 0.4345 data: 0.0002 max mem: 52150 +Epoch: [25] [4350/7269] eta: 0:21:04 lr: 0.000004 loss: 0.5145 (0.5187) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [25] [4360/7269] eta: 0:21:00 lr: 0.000004 loss: 0.4838 (0.5186) time: 0.4175 data: 0.0001 max mem: 52150 +Epoch: [25] [4370/7269] eta: 0:20:55 lr: 0.000004 loss: 0.5166 (0.5186) time: 0.4173 data: 0.0001 max mem: 52150 +Epoch: [25] [4380/7269] eta: 0:20:52 lr: 0.000004 loss: 0.5468 (0.5186) time: 0.5395 data: 0.0001 max mem: 52150 +Epoch: [25] [4390/7269] eta: 0:20:48 lr: 0.000004 loss: 0.5528 (0.5187) time: 0.5401 data: 0.0001 max mem: 52150 +Epoch: [25] [4400/7269] eta: 0:20:44 lr: 0.000004 loss: 0.5320 (0.5187) time: 0.4185 data: 0.0001 max mem: 52150 +Epoch: [25] [4410/7269] eta: 0:20:40 lr: 0.000004 loss: 0.5089 (0.5187) time: 0.4810 data: 0.0001 max mem: 52150 +Epoch: [25] [4420/7269] eta: 0:20:36 lr: 0.000004 loss: 0.4968 (0.5186) time: 0.5432 data: 0.0001 max mem: 52150 +Epoch: [25] [4430/7269] eta: 0:20:32 lr: 0.000004 loss: 0.4801 (0.5187) time: 0.4803 data: 0.0001 max mem: 52150 +Epoch: [25] [4440/7269] eta: 0:20:27 lr: 0.000004 loss: 0.4998 (0.5186) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [25] [4450/7269] eta: 0:20:22 lr: 0.000004 loss: 0.5287 (0.5185) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [25] [4460/7269] eta: 0:20:17 lr: 0.000004 loss: 0.5418 (0.5186) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [25] [4470/7269] eta: 0:20:13 lr: 0.000004 loss: 0.5521 (0.5186) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [25] [4480/7269] eta: 0:20:09 lr: 0.000004 loss: 0.5512 (0.5187) time: 0.4866 data: 0.0002 max mem: 52150 +Epoch: [25] [4490/7269] eta: 0:20:05 lr: 0.000004 loss: 0.5527 (0.5188) time: 0.5466 data: 0.0002 max mem: 52150 +Epoch: [25] [4500/7269] eta: 0:20:01 lr: 0.000004 loss: 0.5264 (0.5187) time: 0.4320 data: 0.0002 max mem: 52150 +Epoch: [25] [4510/7269] eta: 0:19:56 lr: 0.000004 loss: 0.5155 (0.5187) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [25] [4520/7269] eta: 0:19:53 lr: 0.000004 loss: 0.5760 (0.5188) time: 0.4822 data: 0.0002 max mem: 52150 +Epoch: [25] [4530/7269] eta: 0:19:48 lr: 0.000004 loss: 0.5557 (0.5188) time: 0.4827 data: 0.0001 max mem: 52150 +Epoch: [25] [4540/7269] eta: 0:19:44 lr: 0.000004 loss: 0.5274 (0.5188) time: 0.4186 data: 0.0001 max mem: 52150 +Epoch: [25] [4550/7269] eta: 0:19:39 lr: 0.000004 loss: 0.5257 (0.5188) time: 0.4190 data: 0.0002 max mem: 52150 +Epoch: [25] [4560/7269] eta: 0:19:34 lr: 0.000004 loss: 0.5180 (0.5188) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [25] [4570/7269] eta: 0:19:30 lr: 0.000004 loss: 0.5399 (0.5188) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [25] [4580/7269] eta: 0:19:25 lr: 0.000004 loss: 0.5206 (0.5188) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [25] [4590/7269] eta: 0:19:20 lr: 0.000004 loss: 0.5271 (0.5188) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [25] [4600/7269] eta: 0:19:16 lr: 0.000004 loss: 0.5168 (0.5187) time: 0.4325 data: 0.0001 max mem: 52150 +Epoch: [25] [4610/7269] eta: 0:19:11 lr: 0.000004 loss: 0.5344 (0.5188) time: 0.4321 data: 0.0002 max mem: 52150 +Epoch: [25] [4620/7269] eta: 0:19:07 lr: 0.000004 loss: 0.5411 (0.5187) time: 0.4317 data: 0.0001 max mem: 52150 +Epoch: [25] [4630/7269] eta: 0:19:05 lr: 0.000004 loss: 0.4967 (0.5187) time: 0.5966 data: 0.0001 max mem: 52150 +Epoch: [25] [4640/7269] eta: 0:19:00 lr: 0.000004 loss: 0.5331 (0.5188) time: 0.5364 data: 0.0001 max mem: 52150 +Epoch: [25] [4650/7269] eta: 0:18:55 lr: 0.000004 loss: 0.5331 (0.5187) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [25] [4660/7269] eta: 0:18:51 lr: 0.000004 loss: 0.5360 (0.5188) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [25] [4670/7269] eta: 0:18:46 lr: 0.000004 loss: 0.5627 (0.5189) time: 0.3724 data: 0.0001 max mem: 52150 +Epoch: [25] [4680/7269] eta: 0:18:42 lr: 0.000004 loss: 0.5343 (0.5188) time: 0.4174 data: 0.0001 max mem: 52150 +Epoch: [25] [4690/7269] eta: 0:18:37 lr: 0.000004 loss: 0.5177 (0.5188) time: 0.4177 data: 0.0001 max mem: 52150 +Epoch: [25] [4700/7269] eta: 0:18:33 lr: 0.000004 loss: 0.5325 (0.5188) time: 0.4187 data: 0.0001 max mem: 52150 +Epoch: [25] [4710/7269] eta: 0:18:29 lr: 0.000004 loss: 0.5510 (0.5189) time: 0.4672 data: 0.0001 max mem: 52150 +Epoch: [25] [4720/7269] eta: 0:18:25 lr: 0.000004 loss: 0.5544 (0.5189) time: 0.4809 data: 0.0002 max mem: 52150 +Epoch: [25] [4730/7269] eta: 0:18:20 lr: 0.000004 loss: 0.5463 (0.5189) time: 0.4320 data: 0.0002 max mem: 52150 +Epoch: [25] [4740/7269] eta: 0:18:16 lr: 0.000004 loss: 0.5026 (0.5189) time: 0.4318 data: 0.0001 max mem: 52150 +Epoch: [25] [4750/7269] eta: 0:18:11 lr: 0.000004 loss: 0.4709 (0.5188) time: 0.4318 data: 0.0001 max mem: 52150 +Epoch: [25] [4760/7269] eta: 0:18:07 lr: 0.000004 loss: 0.4946 (0.5188) time: 0.4175 data: 0.0002 max mem: 52150 +Epoch: [25] [4770/7269] eta: 0:18:03 lr: 0.000004 loss: 0.5426 (0.5188) time: 0.4174 data: 0.0002 max mem: 52150 +Epoch: [25] [4780/7269] eta: 0:17:58 lr: 0.000004 loss: 0.5420 (0.5188) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [25] [4790/7269] eta: 0:17:53 lr: 0.000004 loss: 0.5326 (0.5189) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [25] [4800/7269] eta: 0:17:50 lr: 0.000004 loss: 0.5087 (0.5188) time: 0.4832 data: 0.0001 max mem: 52150 +Epoch: [25] [4810/7269] eta: 0:17:45 lr: 0.000004 loss: 0.5087 (0.5188) time: 0.4832 data: 0.0001 max mem: 52150 +Epoch: [25] [4820/7269] eta: 0:17:40 lr: 0.000004 loss: 0.4915 (0.5188) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [25] [4830/7269] eta: 0:17:36 lr: 0.000004 loss: 0.4993 (0.5187) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [25] [4840/7269] eta: 0:17:32 lr: 0.000004 loss: 0.5001 (0.5187) time: 0.4172 data: 0.0002 max mem: 52150 +Epoch: [25] [4850/7269] eta: 0:17:28 lr: 0.000004 loss: 0.4883 (0.5187) time: 0.4664 data: 0.0002 max mem: 52150 +Epoch: [25] [4860/7269] eta: 0:17:23 lr: 0.000004 loss: 0.4823 (0.5186) time: 0.4822 data: 0.0002 max mem: 52150 +Epoch: [25] [4870/7269] eta: 0:17:19 lr: 0.000004 loss: 0.5053 (0.5186) time: 0.4330 data: 0.0001 max mem: 52150 +Epoch: [25] [4880/7269] eta: 0:17:15 lr: 0.000004 loss: 0.5238 (0.5187) time: 0.4304 data: 0.0001 max mem: 52150 +Epoch: [25] [4890/7269] eta: 0:17:11 lr: 0.000004 loss: 0.5226 (0.5186) time: 0.4766 data: 0.0001 max mem: 52150 +Epoch: [25] [4900/7269] eta: 0:17:06 lr: 0.000004 loss: 0.5226 (0.5187) time: 0.4178 data: 0.0001 max mem: 52150 +Epoch: [25] [4910/7269] eta: 0:17:01 lr: 0.000004 loss: 0.5258 (0.5186) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [25] [4920/7269] eta: 0:16:57 lr: 0.000004 loss: 0.5140 (0.5186) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [25] [4930/7269] eta: 0:16:53 lr: 0.000004 loss: 0.5379 (0.5186) time: 0.4227 data: 0.0002 max mem: 52150 +Epoch: [25] [4940/7269] eta: 0:16:48 lr: 0.000004 loss: 0.5078 (0.5187) time: 0.4742 data: 0.0002 max mem: 52150 +Epoch: [25] [4950/7269] eta: 0:16:45 lr: 0.000004 loss: 0.5078 (0.5186) time: 0.5354 data: 0.0001 max mem: 52150 +Epoch: [25] [4960/7269] eta: 0:16:41 lr: 0.000004 loss: 0.5143 (0.5186) time: 0.5298 data: 0.0001 max mem: 52150 +Epoch: [25] [4970/7269] eta: 0:16:36 lr: 0.000004 loss: 0.5023 (0.5185) time: 0.4179 data: 0.0002 max mem: 52150 +Epoch: [25] [4980/7269] eta: 0:16:32 lr: 0.000004 loss: 0.4858 (0.5185) time: 0.4360 data: 0.0002 max mem: 52150 +Epoch: [25] [4990/7269] eta: 0:16:28 lr: 0.000004 loss: 0.5230 (0.5186) time: 0.4833 data: 0.0001 max mem: 52150 +Epoch: [25] [5000/7269] eta: 0:16:23 lr: 0.000004 loss: 0.5033 (0.5185) time: 0.4192 data: 0.0001 max mem: 52150 +Epoch: [25] [5010/7269] eta: 0:16:19 lr: 0.000004 loss: 0.4571 (0.5185) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [25] [5020/7269] eta: 0:16:15 lr: 0.000004 loss: 0.4571 (0.5183) time: 0.4315 data: 0.0002 max mem: 52150 +Epoch: [25] [5030/7269] eta: 0:16:10 lr: 0.000004 loss: 0.5106 (0.5183) time: 0.4316 data: 0.0002 max mem: 52150 +Epoch: [25] [5040/7269] eta: 0:16:05 lr: 0.000004 loss: 0.5283 (0.5183) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [25] [5050/7269] eta: 0:16:02 lr: 0.000004 loss: 0.5283 (0.5183) time: 0.4853 data: 0.0001 max mem: 52150 +Epoch: [25] [5060/7269] eta: 0:15:57 lr: 0.000004 loss: 0.5046 (0.5182) time: 0.4854 data: 0.0001 max mem: 52150 +Epoch: [25] [5070/7269] eta: 0:15:53 lr: 0.000004 loss: 0.5192 (0.5183) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [25] [5080/7269] eta: 0:15:48 lr: 0.000004 loss: 0.5525 (0.5183) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [25] [5090/7269] eta: 0:15:44 lr: 0.000004 loss: 0.5389 (0.5183) time: 0.4183 data: 0.0001 max mem: 52150 +Epoch: [25] [5100/7269] eta: 0:15:39 lr: 0.000004 loss: 0.5083 (0.5183) time: 0.4183 data: 0.0001 max mem: 52150 +Epoch: [25] [5110/7269] eta: 0:15:35 lr: 0.000004 loss: 0.5303 (0.5183) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [25] [5120/7269] eta: 0:15:30 lr: 0.000004 loss: 0.5303 (0.5183) time: 0.4310 data: 0.0001 max mem: 52150 +Epoch: [25] [5130/7269] eta: 0:15:26 lr: 0.000004 loss: 0.5157 (0.5183) time: 0.4751 data: 0.0002 max mem: 52150 +Epoch: [25] [5140/7269] eta: 0:15:22 lr: 0.000004 loss: 0.5085 (0.5183) time: 0.4698 data: 0.0002 max mem: 52150 +Epoch: [25] [5150/7269] eta: 0:15:17 lr: 0.000004 loss: 0.4919 (0.5182) time: 0.4254 data: 0.0002 max mem: 52150 +Epoch: [25] [5160/7269] eta: 0:15:13 lr: 0.000004 loss: 0.5004 (0.5182) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [25] [5170/7269] eta: 0:15:08 lr: 0.000004 loss: 0.5218 (0.5182) time: 0.3713 data: 0.0001 max mem: 52150 +Epoch: [25] [5180/7269] eta: 0:15:04 lr: 0.000004 loss: 0.5307 (0.5183) time: 0.4189 data: 0.0001 max mem: 52150 +Epoch: [25] [5190/7269] eta: 0:15:00 lr: 0.000004 loss: 0.5409 (0.5183) time: 0.4780 data: 0.0002 max mem: 52150 +Epoch: [25] [5200/7269] eta: 0:14:55 lr: 0.000004 loss: 0.5293 (0.5183) time: 0.4308 data: 0.0002 max mem: 52150 +Epoch: [25] [5210/7269] eta: 0:14:51 lr: 0.000004 loss: 0.5068 (0.5182) time: 0.4325 data: 0.0001 max mem: 52150 +Epoch: [25] [5220/7269] eta: 0:14:47 lr: 0.000004 loss: 0.5145 (0.5182) time: 0.4326 data: 0.0001 max mem: 52150 +Epoch: [25] [5230/7269] eta: 0:14:42 lr: 0.000004 loss: 0.5310 (0.5182) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [25] [5240/7269] eta: 0:14:38 lr: 0.000004 loss: 0.5532 (0.5183) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [25] [5250/7269] eta: 0:14:34 lr: 0.000004 loss: 0.5517 (0.5183) time: 0.4350 data: 0.0002 max mem: 52150 +Epoch: [25] [5260/7269] eta: 0:14:29 lr: 0.000004 loss: 0.5398 (0.5183) time: 0.4946 data: 0.0002 max mem: 52150 +Epoch: [25] [5270/7269] eta: 0:14:26 lr: 0.000004 loss: 0.5307 (0.5183) time: 0.5417 data: 0.0002 max mem: 52150 +Epoch: [25] [5280/7269] eta: 0:14:22 lr: 0.000004 loss: 0.5179 (0.5183) time: 0.5413 data: 0.0002 max mem: 52150 +Epoch: [25] [5290/7269] eta: 0:14:17 lr: 0.000004 loss: 0.5334 (0.5183) time: 0.4772 data: 0.0001 max mem: 52150 +Epoch: [25] [5300/7269] eta: 0:14:13 lr: 0.000004 loss: 0.5246 (0.5183) time: 0.4178 data: 0.0001 max mem: 52150 +Epoch: [25] [5310/7269] eta: 0:14:09 lr: 0.000004 loss: 0.5000 (0.5182) time: 0.4180 data: 0.0001 max mem: 52150 +Epoch: [25] [5320/7269] eta: 0:14:05 lr: 0.000004 loss: 0.4975 (0.5182) time: 0.5137 data: 0.0002 max mem: 52150 +Epoch: [25] [5330/7269] eta: 0:14:01 lr: 0.000004 loss: 0.5174 (0.5182) time: 0.5275 data: 0.0002 max mem: 52150 +Epoch: [25] [5340/7269] eta: 0:13:56 lr: 0.000004 loss: 0.4785 (0.5181) time: 0.4317 data: 0.0002 max mem: 52150 +Epoch: [25] [5350/7269] eta: 0:13:52 lr: 0.000004 loss: 0.4877 (0.5182) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [25] [5360/7269] eta: 0:13:47 lr: 0.000004 loss: 0.5079 (0.5181) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [25] [5370/7269] eta: 0:13:43 lr: 0.000004 loss: 0.5203 (0.5182) time: 0.4777 data: 0.0001 max mem: 52150 +Epoch: [25] [5380/7269] eta: 0:13:39 lr: 0.000004 loss: 0.5397 (0.5182) time: 0.4786 data: 0.0002 max mem: 52150 +Epoch: [25] [5390/7269] eta: 0:13:35 lr: 0.000004 loss: 0.5182 (0.5182) time: 0.4716 data: 0.0001 max mem: 52150 +Epoch: [25] [5400/7269] eta: 0:13:30 lr: 0.000004 loss: 0.5181 (0.5182) time: 0.4712 data: 0.0001 max mem: 52150 +Epoch: [25] [5410/7269] eta: 0:13:26 lr: 0.000004 loss: 0.5524 (0.5182) time: 0.3725 data: 0.0001 max mem: 52150 +Epoch: [25] [5420/7269] eta: 0:13:21 lr: 0.000004 loss: 0.5167 (0.5182) time: 0.3723 data: 0.0001 max mem: 52150 +Epoch: [25] [5430/7269] eta: 0:13:17 lr: 0.000004 loss: 0.5022 (0.5182) time: 0.3714 data: 0.0001 max mem: 52150 +Epoch: [25] [5440/7269] eta: 0:13:12 lr: 0.000004 loss: 0.5443 (0.5182) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [25] [5450/7269] eta: 0:13:07 lr: 0.000004 loss: 0.5443 (0.5183) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [25] [5460/7269] eta: 0:13:03 lr: 0.000004 loss: 0.5525 (0.5183) time: 0.4350 data: 0.0001 max mem: 52150 +Epoch: [25] [5470/7269] eta: 0:12:59 lr: 0.000004 loss: 0.5417 (0.5183) time: 0.4349 data: 0.0001 max mem: 52150 +Epoch: [25] [5480/7269] eta: 0:12:55 lr: 0.000004 loss: 0.5417 (0.5184) time: 0.4309 data: 0.0001 max mem: 52150 +Epoch: [25] [5490/7269] eta: 0:12:50 lr: 0.000004 loss: 0.5386 (0.5184) time: 0.4772 data: 0.0001 max mem: 52150 +Epoch: [25] [5500/7269] eta: 0:12:46 lr: 0.000004 loss: 0.5239 (0.5184) time: 0.4180 data: 0.0001 max mem: 52150 +Epoch: [25] [5510/7269] eta: 0:12:41 lr: 0.000004 loss: 0.5141 (0.5184) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [25] [5520/7269] eta: 0:12:37 lr: 0.000004 loss: 0.4971 (0.5183) time: 0.4347 data: 0.0002 max mem: 52150 +Epoch: [25] [5530/7269] eta: 0:12:33 lr: 0.000004 loss: 0.5289 (0.5183) time: 0.4343 data: 0.0002 max mem: 52150 +Epoch: [25] [5540/7269] eta: 0:12:28 lr: 0.000004 loss: 0.5477 (0.5184) time: 0.4178 data: 0.0002 max mem: 52150 +Epoch: [25] [5550/7269] eta: 0:12:24 lr: 0.000004 loss: 0.5533 (0.5184) time: 0.4181 data: 0.0001 max mem: 52150 +Epoch: [25] [5560/7269] eta: 0:12:19 lr: 0.000004 loss: 0.5143 (0.5184) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [25] [5570/7269] eta: 0:12:15 lr: 0.000004 loss: 0.4935 (0.5184) time: 0.4326 data: 0.0001 max mem: 52150 +Epoch: [25] [5580/7269] eta: 0:12:11 lr: 0.000004 loss: 0.5354 (0.5184) time: 0.5329 data: 0.0001 max mem: 52150 +Epoch: [25] [5590/7269] eta: 0:12:07 lr: 0.000004 loss: 0.5048 (0.5183) time: 0.4720 data: 0.0001 max mem: 52150 +Epoch: [25] [5600/7269] eta: 0:12:03 lr: 0.000004 loss: 0.4788 (0.5184) time: 0.4214 data: 0.0001 max mem: 52150 +Epoch: [25] [5610/7269] eta: 0:11:58 lr: 0.000004 loss: 0.5430 (0.5184) time: 0.4214 data: 0.0002 max mem: 52150 +Epoch: [25] [5620/7269] eta: 0:11:54 lr: 0.000004 loss: 0.5334 (0.5183) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [25] [5630/7269] eta: 0:11:49 lr: 0.000004 loss: 0.5210 (0.5184) time: 0.4208 data: 0.0001 max mem: 52150 +Epoch: [25] [5640/7269] eta: 0:11:45 lr: 0.000004 loss: 0.5529 (0.5184) time: 0.4656 data: 0.0001 max mem: 52150 +Epoch: [25] [5650/7269] eta: 0:11:41 lr: 0.000004 loss: 0.5121 (0.5184) time: 0.4623 data: 0.0001 max mem: 52150 +Epoch: [25] [5660/7269] eta: 0:11:37 lr: 0.000004 loss: 0.5023 (0.5183) time: 0.4772 data: 0.0001 max mem: 52150 +Epoch: [25] [5670/7269] eta: 0:11:32 lr: 0.000004 loss: 0.5121 (0.5183) time: 0.4315 data: 0.0001 max mem: 52150 +Epoch: [25] [5680/7269] eta: 0:11:28 lr: 0.000004 loss: 0.5254 (0.5183) time: 0.4952 data: 0.0001 max mem: 52150 +Epoch: [25] [5690/7269] eta: 0:11:24 lr: 0.000004 loss: 0.5254 (0.5184) time: 0.4957 data: 0.0001 max mem: 52150 +Epoch: [25] [5700/7269] eta: 0:11:19 lr: 0.000004 loss: 0.4902 (0.5183) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [25] [5710/7269] eta: 0:11:15 lr: 0.000004 loss: 0.4781 (0.5183) time: 0.4357 data: 0.0001 max mem: 52150 +Epoch: [25] [5720/7269] eta: 0:11:11 lr: 0.000004 loss: 0.5097 (0.5183) time: 0.4359 data: 0.0002 max mem: 52150 +Epoch: [25] [5730/7269] eta: 0:11:07 lr: 0.000004 loss: 0.5029 (0.5182) time: 0.4330 data: 0.0002 max mem: 52150 +Epoch: [25] [5740/7269] eta: 0:11:03 lr: 0.000004 loss: 0.4751 (0.5182) time: 0.5965 data: 0.0001 max mem: 52150 +Epoch: [25] [5750/7269] eta: 0:10:58 lr: 0.000004 loss: 0.4687 (0.5181) time: 0.5358 data: 0.0001 max mem: 52150 +Epoch: [25] [5760/7269] eta: 0:10:54 lr: 0.000004 loss: 0.5127 (0.5182) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [25] [5770/7269] eta: 0:10:49 lr: 0.000004 loss: 0.5322 (0.5182) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [25] [5780/7269] eta: 0:10:45 lr: 0.000004 loss: 0.5283 (0.5182) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [25] [5790/7269] eta: 0:10:40 lr: 0.000004 loss: 0.5283 (0.5182) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [25] [5800/7269] eta: 0:10:36 lr: 0.000004 loss: 0.5244 (0.5182) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [25] [5810/7269] eta: 0:10:31 lr: 0.000004 loss: 0.5356 (0.5182) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [25] [5820/7269] eta: 0:10:27 lr: 0.000004 loss: 0.5514 (0.5182) time: 0.3712 data: 0.0002 max mem: 52150 +Epoch: [25] [5830/7269] eta: 0:10:23 lr: 0.000004 loss: 0.5428 (0.5182) time: 0.4311 data: 0.0001 max mem: 52150 +Epoch: [25] [5840/7269] eta: 0:10:18 lr: 0.000004 loss: 0.5357 (0.5183) time: 0.4319 data: 0.0001 max mem: 52150 +Epoch: [25] [5850/7269] eta: 0:10:14 lr: 0.000004 loss: 0.5053 (0.5182) time: 0.4341 data: 0.0001 max mem: 52150 +Epoch: [25] [5860/7269] eta: 0:10:10 lr: 0.000004 loss: 0.5010 (0.5182) time: 0.4798 data: 0.0001 max mem: 52150 +Epoch: [25] [5870/7269] eta: 0:10:06 lr: 0.000004 loss: 0.5044 (0.5182) time: 0.4625 data: 0.0002 max mem: 52150 +Epoch: [25] [5880/7269] eta: 0:10:01 lr: 0.000004 loss: 0.4968 (0.5182) time: 0.4165 data: 0.0002 max mem: 52150 +Epoch: [25] [5890/7269] eta: 0:09:57 lr: 0.000004 loss: 0.5216 (0.5182) time: 0.3721 data: 0.0001 max mem: 52150 +Epoch: [25] [5900/7269] eta: 0:09:52 lr: 0.000004 loss: 0.5415 (0.5182) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [25] [5910/7269] eta: 0:09:48 lr: 0.000004 loss: 0.5446 (0.5182) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [25] [5920/7269] eta: 0:09:43 lr: 0.000004 loss: 0.5141 (0.5182) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [25] [5930/7269] eta: 0:09:39 lr: 0.000004 loss: 0.5045 (0.5182) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [25] [5940/7269] eta: 0:09:34 lr: 0.000004 loss: 0.4941 (0.5181) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [25] [5950/7269] eta: 0:09:30 lr: 0.000004 loss: 0.5135 (0.5181) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [25] [5960/7269] eta: 0:09:26 lr: 0.000004 loss: 0.5135 (0.5181) time: 0.4197 data: 0.0002 max mem: 52150 +Epoch: [25] [5970/7269] eta: 0:09:21 lr: 0.000004 loss: 0.5238 (0.5181) time: 0.4689 data: 0.0001 max mem: 52150 +Epoch: [25] [5980/7269] eta: 0:09:17 lr: 0.000004 loss: 0.5438 (0.5182) time: 0.5305 data: 0.0001 max mem: 52150 +Epoch: [25] [5990/7269] eta: 0:09:13 lr: 0.000004 loss: 0.5438 (0.5182) time: 0.4825 data: 0.0001 max mem: 52150 +Epoch: [25] [6000/7269] eta: 0:09:09 lr: 0.000004 loss: 0.5174 (0.5182) time: 0.3728 data: 0.0001 max mem: 52150 +Epoch: [25] [6010/7269] eta: 0:09:04 lr: 0.000004 loss: 0.4926 (0.5182) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [25] [6020/7269] eta: 0:09:00 lr: 0.000004 loss: 0.4917 (0.5181) time: 0.4181 data: 0.0002 max mem: 52150 +Epoch: [25] [6030/7269] eta: 0:08:56 lr: 0.000004 loss: 0.5017 (0.5182) time: 0.5275 data: 0.0001 max mem: 52150 +Epoch: [25] [6040/7269] eta: 0:08:52 lr: 0.000004 loss: 0.5464 (0.5182) time: 0.5308 data: 0.0001 max mem: 52150 +Epoch: [25] [6050/7269] eta: 0:08:47 lr: 0.000004 loss: 0.5354 (0.5182) time: 0.4210 data: 0.0002 max mem: 52150 +Epoch: [25] [6060/7269] eta: 0:08:43 lr: 0.000004 loss: 0.5405 (0.5182) time: 0.4169 data: 0.0002 max mem: 52150 +Epoch: [25] [6070/7269] eta: 0:08:39 lr: 0.000004 loss: 0.5420 (0.5182) time: 0.4636 data: 0.0002 max mem: 52150 +Epoch: [25] [6080/7269] eta: 0:08:34 lr: 0.000004 loss: 0.5279 (0.5182) time: 0.4192 data: 0.0002 max mem: 52150 +Epoch: [25] [6090/7269] eta: 0:08:30 lr: 0.000004 loss: 0.5505 (0.5182) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [25] [6100/7269] eta: 0:08:25 lr: 0.000004 loss: 0.4743 (0.5181) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [25] [6110/7269] eta: 0:08:21 lr: 0.000004 loss: 0.4741 (0.5180) time: 0.3723 data: 0.0001 max mem: 52150 +Epoch: [25] [6120/7269] eta: 0:08:17 lr: 0.000004 loss: 0.5125 (0.5180) time: 0.4862 data: 0.0001 max mem: 52150 +Epoch: [25] [6130/7269] eta: 0:08:13 lr: 0.000004 loss: 0.5450 (0.5180) time: 0.5352 data: 0.0001 max mem: 52150 +Epoch: [25] [6140/7269] eta: 0:08:08 lr: 0.000004 loss: 0.5153 (0.5180) time: 0.4850 data: 0.0001 max mem: 52150 +Epoch: [25] [6150/7269] eta: 0:08:04 lr: 0.000004 loss: 0.5153 (0.5180) time: 0.4353 data: 0.0002 max mem: 52150 +Epoch: [25] [6160/7269] eta: 0:08:00 lr: 0.000004 loss: 0.5214 (0.5180) time: 0.4312 data: 0.0001 max mem: 52150 +Epoch: [25] [6170/7269] eta: 0:07:55 lr: 0.000004 loss: 0.5214 (0.5180) time: 0.4312 data: 0.0001 max mem: 52150 +Epoch: [25] [6180/7269] eta: 0:07:51 lr: 0.000004 loss: 0.5025 (0.5180) time: 0.4316 data: 0.0002 max mem: 52150 +Epoch: [25] [6190/7269] eta: 0:07:47 lr: 0.000004 loss: 0.5281 (0.5180) time: 0.4320 data: 0.0002 max mem: 52150 +Epoch: [25] [6200/7269] eta: 0:07:42 lr: 0.000004 loss: 0.5238 (0.5180) time: 0.4327 data: 0.0002 max mem: 52150 +Epoch: [25] [6210/7269] eta: 0:07:38 lr: 0.000004 loss: 0.5074 (0.5180) time: 0.4322 data: 0.0001 max mem: 52150 +Epoch: [25] [6220/7269] eta: 0:07:34 lr: 0.000004 loss: 0.5074 (0.5180) time: 0.4312 data: 0.0001 max mem: 52150 +Epoch: [25] [6230/7269] eta: 0:07:29 lr: 0.000004 loss: 0.5238 (0.5180) time: 0.4317 data: 0.0002 max mem: 52150 +Epoch: [25] [6240/7269] eta: 0:07:25 lr: 0.000004 loss: 0.5237 (0.5180) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [25] [6250/7269] eta: 0:07:20 lr: 0.000004 loss: 0.4973 (0.5180) time: 0.3728 data: 0.0001 max mem: 52150 +Epoch: [25] [6260/7269] eta: 0:07:16 lr: 0.000004 loss: 0.5037 (0.5180) time: 0.4978 data: 0.0001 max mem: 52150 +Epoch: [25] [6270/7269] eta: 0:07:12 lr: 0.000004 loss: 0.5113 (0.5180) time: 0.5442 data: 0.0001 max mem: 52150 +Epoch: [25] [6280/7269] eta: 0:07:08 lr: 0.000004 loss: 0.4761 (0.5179) time: 0.4190 data: 0.0002 max mem: 52150 +Epoch: [25] [6290/7269] eta: 0:07:03 lr: 0.000004 loss: 0.4733 (0.5179) time: 0.4309 data: 0.0001 max mem: 52150 +Epoch: [25] [6300/7269] eta: 0:06:59 lr: 0.000004 loss: 0.4982 (0.5179) time: 0.4311 data: 0.0001 max mem: 52150 +Epoch: [25] [6310/7269] eta: 0:06:55 lr: 0.000004 loss: 0.4982 (0.5178) time: 0.4964 data: 0.0001 max mem: 52150 +Epoch: [25] [6320/7269] eta: 0:06:51 lr: 0.000004 loss: 0.4943 (0.5178) time: 0.4957 data: 0.0002 max mem: 52150 +Epoch: [25] [6330/7269] eta: 0:06:46 lr: 0.000004 loss: 0.5129 (0.5178) time: 0.4650 data: 0.0002 max mem: 52150 +Epoch: [25] [6340/7269] eta: 0:06:42 lr: 0.000004 loss: 0.5539 (0.5178) time: 0.4653 data: 0.0001 max mem: 52150 +Epoch: [25] [6350/7269] eta: 0:06:38 lr: 0.000004 loss: 0.5334 (0.5178) time: 0.4923 data: 0.0001 max mem: 52150 +Epoch: [25] [6360/7269] eta: 0:06:33 lr: 0.000004 loss: 0.5333 (0.5178) time: 0.4923 data: 0.0001 max mem: 52150 +Epoch: [25] [6370/7269] eta: 0:06:29 lr: 0.000004 loss: 0.5453 (0.5179) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [25] [6380/7269] eta: 0:06:25 lr: 0.000004 loss: 0.5453 (0.5179) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [25] [6390/7269] eta: 0:06:20 lr: 0.000004 loss: 0.4904 (0.5178) time: 0.3725 data: 0.0001 max mem: 52150 +Epoch: [25] [6400/7269] eta: 0:06:16 lr: 0.000004 loss: 0.4540 (0.5178) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [25] [6410/7269] eta: 0:06:11 lr: 0.000004 loss: 0.4926 (0.5177) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [25] [6420/7269] eta: 0:06:07 lr: 0.000004 loss: 0.5032 (0.5177) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [25] [6430/7269] eta: 0:06:03 lr: 0.000004 loss: 0.4981 (0.5176) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [25] [6440/7269] eta: 0:05:58 lr: 0.000004 loss: 0.5110 (0.5176) time: 0.4661 data: 0.0002 max mem: 52150 +Epoch: [25] [6450/7269] eta: 0:05:54 lr: 0.000004 loss: 0.5151 (0.5176) time: 0.5137 data: 0.0001 max mem: 52150 +Epoch: [25] [6460/7269] eta: 0:05:50 lr: 0.000004 loss: 0.5303 (0.5176) time: 0.4688 data: 0.0001 max mem: 52150 +Epoch: [25] [6470/7269] eta: 0:05:45 lr: 0.000004 loss: 0.5303 (0.5176) time: 0.4211 data: 0.0001 max mem: 52150 +Epoch: [25] [6480/7269] eta: 0:05:41 lr: 0.000004 loss: 0.5333 (0.5176) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [25] [6490/7269] eta: 0:05:37 lr: 0.000004 loss: 0.5381 (0.5177) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [25] [6500/7269] eta: 0:05:32 lr: 0.000004 loss: 0.5381 (0.5177) time: 0.4209 data: 0.0001 max mem: 52150 +Epoch: [25] [6510/7269] eta: 0:05:28 lr: 0.000004 loss: 0.5389 (0.5177) time: 0.4212 data: 0.0001 max mem: 52150 +Epoch: [25] [6520/7269] eta: 0:05:24 lr: 0.000004 loss: 0.5478 (0.5177) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [25] [6530/7269] eta: 0:05:19 lr: 0.000004 loss: 0.5394 (0.5178) time: 0.4179 data: 0.0002 max mem: 52150 +Epoch: [25] [6540/7269] eta: 0:05:15 lr: 0.000004 loss: 0.5417 (0.5179) time: 0.4819 data: 0.0002 max mem: 52150 +Epoch: [25] [6550/7269] eta: 0:05:11 lr: 0.000004 loss: 0.5417 (0.5179) time: 0.4357 data: 0.0002 max mem: 52150 +Epoch: [25] [6560/7269] eta: 0:05:06 lr: 0.000004 loss: 0.5420 (0.5179) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [25] [6570/7269] eta: 0:05:02 lr: 0.000004 loss: 0.5420 (0.5179) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [25] [6580/7269] eta: 0:04:58 lr: 0.000004 loss: 0.5327 (0.5179) time: 0.4350 data: 0.0002 max mem: 52150 +Epoch: [25] [6590/7269] eta: 0:04:53 lr: 0.000004 loss: 0.5003 (0.5179) time: 0.5428 data: 0.0002 max mem: 52150 +Epoch: [25] [6600/7269] eta: 0:04:49 lr: 0.000004 loss: 0.4976 (0.5179) time: 0.4802 data: 0.0001 max mem: 52150 +Epoch: [25] [6610/7269] eta: 0:04:45 lr: 0.000004 loss: 0.4976 (0.5179) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [25] [6620/7269] eta: 0:04:40 lr: 0.000004 loss: 0.5194 (0.5179) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [25] [6630/7269] eta: 0:04:36 lr: 0.000004 loss: 0.5330 (0.5179) time: 0.4306 data: 0.0002 max mem: 52150 +Epoch: [25] [6640/7269] eta: 0:04:32 lr: 0.000004 loss: 0.5288 (0.5179) time: 0.4307 data: 0.0002 max mem: 52150 +Epoch: [25] [6650/7269] eta: 0:04:27 lr: 0.000004 loss: 0.5288 (0.5179) time: 0.4323 data: 0.0002 max mem: 52150 +Epoch: [25] [6660/7269] eta: 0:04:23 lr: 0.000004 loss: 0.5124 (0.5179) time: 0.4322 data: 0.0002 max mem: 52150 +Epoch: [25] [6670/7269] eta: 0:04:19 lr: 0.000004 loss: 0.5445 (0.5180) time: 0.4199 data: 0.0002 max mem: 52150 +Epoch: [25] [6680/7269] eta: 0:04:14 lr: 0.000004 loss: 0.5541 (0.5180) time: 0.4798 data: 0.0001 max mem: 52150 +Epoch: [25] [6690/7269] eta: 0:04:10 lr: 0.000004 loss: 0.5256 (0.5180) time: 0.4935 data: 0.0001 max mem: 52150 +Epoch: [25] [6700/7269] eta: 0:04:06 lr: 0.000004 loss: 0.4909 (0.5180) time: 0.4336 data: 0.0002 max mem: 52150 +Epoch: [25] [6710/7269] eta: 0:04:01 lr: 0.000004 loss: 0.4767 (0.5179) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [25] [6720/7269] eta: 0:03:57 lr: 0.000004 loss: 0.4767 (0.5179) time: 0.4353 data: 0.0001 max mem: 52150 +Epoch: [25] [6730/7269] eta: 0:03:53 lr: 0.000004 loss: 0.5023 (0.5179) time: 0.4352 data: 0.0002 max mem: 52150 +Epoch: [25] [6740/7269] eta: 0:03:48 lr: 0.000004 loss: 0.5584 (0.5180) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [25] [6750/7269] eta: 0:03:44 lr: 0.000004 loss: 0.5564 (0.5179) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [25] [6760/7269] eta: 0:03:40 lr: 0.000004 loss: 0.5390 (0.5180) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [25] [6770/7269] eta: 0:03:35 lr: 0.000004 loss: 0.5390 (0.5180) time: 0.4785 data: 0.0002 max mem: 52150 +Epoch: [25] [6780/7269] eta: 0:03:31 lr: 0.000004 loss: 0.5362 (0.5180) time: 0.5406 data: 0.0001 max mem: 52150 +Epoch: [25] [6790/7269] eta: 0:03:27 lr: 0.000004 loss: 0.5334 (0.5180) time: 0.4339 data: 0.0001 max mem: 52150 +Epoch: [25] [6800/7269] eta: 0:03:22 lr: 0.000004 loss: 0.5556 (0.5181) time: 0.3725 data: 0.0001 max mem: 52150 +Epoch: [25] [6810/7269] eta: 0:03:18 lr: 0.000004 loss: 0.5398 (0.5180) time: 0.3728 data: 0.0002 max mem: 52150 +Epoch: [25] [6820/7269] eta: 0:03:14 lr: 0.000004 loss: 0.5105 (0.5180) time: 0.4337 data: 0.0002 max mem: 52150 +Epoch: [25] [6830/7269] eta: 0:03:09 lr: 0.000004 loss: 0.5432 (0.5181) time: 0.4332 data: 0.0002 max mem: 52150 +Epoch: [25] [6840/7269] eta: 0:03:05 lr: 0.000004 loss: 0.5372 (0.5180) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [25] [6850/7269] eta: 0:03:01 lr: 0.000004 loss: 0.4684 (0.5180) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [25] [6860/7269] eta: 0:02:56 lr: 0.000004 loss: 0.4952 (0.5180) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [25] [6870/7269] eta: 0:02:52 lr: 0.000004 loss: 0.5185 (0.5180) time: 0.4347 data: 0.0002 max mem: 52150 +Epoch: [25] [6880/7269] eta: 0:02:48 lr: 0.000004 loss: 0.5288 (0.5180) time: 0.4349 data: 0.0002 max mem: 52150 +Epoch: [25] [6890/7269] eta: 0:02:43 lr: 0.000004 loss: 0.5251 (0.5180) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [25] [6900/7269] eta: 0:02:39 lr: 0.000004 loss: 0.4949 (0.5179) time: 0.4682 data: 0.0001 max mem: 52150 +Epoch: [25] [6910/7269] eta: 0:02:35 lr: 0.000004 loss: 0.4961 (0.5179) time: 0.6238 data: 0.0001 max mem: 52150 +Epoch: [25] [6920/7269] eta: 0:02:30 lr: 0.000004 loss: 0.5147 (0.5179) time: 0.5755 data: 0.0002 max mem: 52150 +Epoch: [25] [6930/7269] eta: 0:02:26 lr: 0.000004 loss: 0.5400 (0.5179) time: 0.5189 data: 0.0002 max mem: 52150 +Epoch: [25] [6940/7269] eta: 0:02:22 lr: 0.000004 loss: 0.5384 (0.5179) time: 0.5154 data: 0.0001 max mem: 52150 +Epoch: [25] [6950/7269] eta: 0:02:18 lr: 0.000004 loss: 0.5042 (0.5179) time: 0.4158 data: 0.0002 max mem: 52150 +Epoch: [25] [6960/7269] eta: 0:02:13 lr: 0.000004 loss: 0.5097 (0.5179) time: 0.4339 data: 0.0002 max mem: 52150 +Epoch: [25] [6970/7269] eta: 0:02:09 lr: 0.000004 loss: 0.5276 (0.5178) time: 0.4792 data: 0.0002 max mem: 52150 +Epoch: [25] [6980/7269] eta: 0:02:05 lr: 0.000004 loss: 0.4971 (0.5178) time: 0.4169 data: 0.0001 max mem: 52150 +Epoch: [25] [6990/7269] eta: 0:02:00 lr: 0.000004 loss: 0.4911 (0.5178) time: 0.3722 data: 0.0001 max mem: 52150 +Epoch: [25] [7000/7269] eta: 0:01:56 lr: 0.000004 loss: 0.5189 (0.5178) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [25] [7010/7269] eta: 0:01:52 lr: 0.000004 loss: 0.5189 (0.5178) time: 0.4342 data: 0.0001 max mem: 52150 +Epoch: [25] [7020/7269] eta: 0:01:47 lr: 0.000004 loss: 0.4971 (0.5177) time: 0.5396 data: 0.0001 max mem: 52150 +Epoch: [25] [7030/7269] eta: 0:01:43 lr: 0.000004 loss: 0.5089 (0.5177) time: 0.4774 data: 0.0001 max mem: 52150 +Epoch: [25] [7040/7269] eta: 0:01:39 lr: 0.000004 loss: 0.5098 (0.5177) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [25] [7050/7269] eta: 0:01:34 lr: 0.000004 loss: 0.5098 (0.5177) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [25] [7060/7269] eta: 0:01:30 lr: 0.000004 loss: 0.5049 (0.5177) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [25] [7070/7269] eta: 0:01:26 lr: 0.000004 loss: 0.4931 (0.5177) time: 0.4323 data: 0.0001 max mem: 52150 +Epoch: [25] [7080/7269] eta: 0:01:21 lr: 0.000004 loss: 0.4888 (0.5177) time: 0.4319 data: 0.0002 max mem: 52150 +Epoch: [25] [7090/7269] eta: 0:01:17 lr: 0.000004 loss: 0.4888 (0.5177) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [25] [7100/7269] eta: 0:01:13 lr: 0.000004 loss: 0.5205 (0.5177) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [25] [7110/7269] eta: 0:01:08 lr: 0.000004 loss: 0.5205 (0.5177) time: 0.4183 data: 0.0002 max mem: 52150 +Epoch: [25] [7120/7269] eta: 0:01:04 lr: 0.000004 loss: 0.5649 (0.5177) time: 0.4780 data: 0.0002 max mem: 52150 +Epoch: [25] [7130/7269] eta: 0:01:00 lr: 0.000004 loss: 0.5651 (0.5177) time: 0.4311 data: 0.0001 max mem: 52150 +Epoch: [25] [7140/7269] eta: 0:00:55 lr: 0.000004 loss: 0.5184 (0.5177) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [25] [7150/7269] eta: 0:00:51 lr: 0.000004 loss: 0.4935 (0.5177) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [25] [7160/7269] eta: 0:00:47 lr: 0.000004 loss: 0.5162 (0.5177) time: 0.4356 data: 0.0001 max mem: 52150 +Epoch: [25] [7170/7269] eta: 0:00:42 lr: 0.000004 loss: 0.5556 (0.5178) time: 0.4358 data: 0.0002 max mem: 52150 +Epoch: [25] [7180/7269] eta: 0:00:38 lr: 0.000004 loss: 0.5510 (0.5177) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [25] [7190/7269] eta: 0:00:34 lr: 0.000004 loss: 0.4844 (0.5177) time: 0.4338 data: 0.0002 max mem: 52150 +Epoch: [25] [7200/7269] eta: 0:00:29 lr: 0.000004 loss: 0.5265 (0.5177) time: 0.4826 data: 0.0002 max mem: 52150 +Epoch: [25] [7210/7269] eta: 0:00:25 lr: 0.000004 loss: 0.5265 (0.5178) time: 0.4202 data: 0.0002 max mem: 52150 +Epoch: [25] [7220/7269] eta: 0:00:21 lr: 0.000004 loss: 0.5224 (0.5177) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [25] [7230/7269] eta: 0:00:16 lr: 0.000004 loss: 0.4950 (0.5177) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [25] [7240/7269] eta: 0:00:12 lr: 0.000004 loss: 0.5206 (0.5178) time: 0.4299 data: 0.0002 max mem: 52150 +Epoch: [25] [7250/7269] eta: 0:00:08 lr: 0.000004 loss: 0.5316 (0.5178) time: 0.4952 data: 0.0038 max mem: 52150 +Epoch: [25] [7260/7269] eta: 0:00:03 lr: 0.000004 loss: 0.5349 (0.5178) time: 0.4976 data: 0.0037 max mem: 52150 +Epoch: [25] [7268/7269] eta: 0:00:00 lr: 0.000004 loss: 0.5202 (0.5178) time: 0.4359 data: 0.0037 max mem: 52150 +Epoch: [25] Total time: 0:52:25 (0.4327 s / it) +Averaged stats: lr: 0.000004 loss: 0.5202 (0.5171) +Test: [ 0/1403] eta: 1:18:12 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0811 (0.0811) acc1: 100.0000 (100.0000) acc5: 100.0000 (100.0000) time: 3.3445 data: 3.2469 max mem: 52150 +Test: [ 10/1403] eta: 0:09:05 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1435 (0.1650) acc1: 98.4375 (95.1705) acc5: 100.0000 (100.0000) time: 0.3917 data: 0.2955 max mem: 52150 +Test: [ 20/1403] eta: 0:05:47 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1214 (0.1522) acc1: 98.4375 (95.7589) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0004 max mem: 52150 +Test: [ 30/1403] eta: 0:04:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0885 (0.1427) acc1: 98.4375 (96.3206) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0004 max mem: 52150 +Test: [ 40/1403] eta: 0:03:59 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0906 (0.1394) acc1: 100.0000 (96.7607) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0004 max mem: 52150 +Test: [ 50/1403] eta: 0:03:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1105 (0.1472) acc1: 100.0000 (96.5686) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 60/1403] eta: 0:03:21 acc1_pos: 93.750 acc5_pos: 100.000 loss: 0.1110 (0.1450) acc1: 100.0000 (96.8750) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 70/1403] eta: 0:03:09 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.1282 (0.1504) acc1: 98.4375 (96.5449) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 80/1403] eta: 0:03:00 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.1609 (0.1570) acc1: 96.8750 (96.1613) acc5: 100.0000 (100.0000) time: 0.0963 data: 0.0003 max mem: 52150 +Test: [ 90/1403] eta: 0:02:53 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1617 (0.1726) acc1: 95.3125 (95.3297) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 100/1403] eta: 0:02:47 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1531 (0.1710) acc1: 96.8750 (95.4981) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 110/1403] eta: 0:02:42 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1253 (0.1690) acc1: 98.4375 (95.6081) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 120/1403] eta: 0:02:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0756 (0.1612) acc1: 100.0000 (95.9711) acc5: 100.0000 (100.0000) time: 0.0963 data: 0.0003 max mem: 52150 +Test: [ 130/1403] eta: 0:02:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0766 (0.1557) acc1: 100.0000 (96.1951) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 140/1403] eta: 0:02:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0733 (0.1498) acc1: 100.0000 (96.4650) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 150/1403] eta: 0:02:27 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0719 (0.1448) acc1: 100.0000 (96.6784) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 160/1403] eta: 0:02:24 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0734 (0.1407) acc1: 100.0000 (96.8653) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 170/1403] eta: 0:02:22 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0740 (0.1373) acc1: 100.0000 (97.0212) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0004 max mem: 52150 +Test: [ 180/1403] eta: 0:02:19 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.0747 (0.1395) acc1: 100.0000 (96.9700) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0004 max mem: 52150 +Test: [ 190/1403] eta: 0:02:17 acc1_pos: 93.750 acc5_pos: 100.000 loss: 0.2018 (0.1446) acc1: 96.8750 (96.6868) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 200/1403] eta: 0:02:15 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.2008 (0.1531) acc1: 93.7500 (96.3231) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 210/1403] eta: 0:02:13 acc1_pos: 84.375 acc5_pos: 100.000 loss: 0.2482 (0.1607) acc1: 90.6250 (95.9049) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 220/1403] eta: 0:02:11 acc1_pos: 93.750 acc5_pos: 100.000 loss: 0.2806 (0.1672) acc1: 92.1875 (95.6519) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 230/1403] eta: 0:02:09 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2077 (0.1638) acc1: 93.7500 (95.7860) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 240/1403] eta: 0:02:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0715 (0.1602) acc1: 100.0000 (95.9479) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 250/1403] eta: 0:02:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0715 (0.1569) acc1: 100.0000 (96.0906) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 260/1403] eta: 0:02:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0720 (0.1538) acc1: 100.0000 (96.2225) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 270/1403] eta: 0:02:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0727 (0.1510) acc1: 100.0000 (96.3446) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 280/1403] eta: 0:02:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0715 (0.1482) acc1: 100.0000 (96.4691) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 290/1403] eta: 0:01:59 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0690 (0.1455) acc1: 100.0000 (96.5851) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 300/1403] eta: 0:01:58 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0683 (0.1430) acc1: 100.0000 (96.6933) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 310/1403] eta: 0:01:56 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0699 (0.1411) acc1: 100.0000 (96.7996) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 320/1403] eta: 0:01:55 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1030 (0.1400) acc1: 100.0000 (96.8993) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 330/1403] eta: 0:01:54 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1103 (0.1392) acc1: 100.0000 (96.9883) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 340/1403] eta: 0:01:52 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1103 (0.1383) acc1: 100.0000 (97.0766) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 350/1403] eta: 0:01:51 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1073 (0.1375) acc1: 100.0000 (97.1599) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 360/1403] eta: 0:01:50 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1123 (0.1369) acc1: 100.0000 (97.2299) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 370/1403] eta: 0:01:48 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1100 (0.1361) acc1: 100.0000 (97.3046) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 380/1403] eta: 0:01:47 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1081 (0.1354) acc1: 100.0000 (97.3753) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 390/1403] eta: 0:01:46 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1097 (0.1369) acc1: 100.0000 (97.2866) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 400/1403] eta: 0:01:44 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2521 (0.1417) acc1: 89.0625 (97.0309) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 410/1403] eta: 0:01:43 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3114 (0.1459) acc1: 87.5000 (96.7914) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 420/1403] eta: 0:01:42 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2756 (0.1484) acc1: 89.0625 (96.6486) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 430/1403] eta: 0:01:41 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2488 (0.1521) acc1: 90.6250 (96.4617) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 440/1403] eta: 0:01:40 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3109 (0.1563) acc1: 87.5000 (96.2585) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 450/1403] eta: 0:01:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2904 (0.1593) acc1: 89.0625 (96.1024) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 460/1403] eta: 0:01:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2784 (0.1625) acc1: 89.0625 (95.9260) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 470/1403] eta: 0:01:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2470 (0.1620) acc1: 90.6250 (95.9329) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [ 480/1403] eta: 0:01:35 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0650 (0.1600) acc1: 100.0000 (96.0174) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [ 490/1403] eta: 0:01:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0653 (0.1581) acc1: 100.0000 (96.0985) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 500/1403] eta: 0:01:33 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0658 (0.1563) acc1: 100.0000 (96.1764) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [ 510/1403] eta: 0:01:31 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0658 (0.1545) acc1: 100.0000 (96.2512) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 520/1403] eta: 0:01:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0710 (0.1530) acc1: 100.0000 (96.3232) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 530/1403] eta: 0:01:29 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0608 (0.1512) acc1: 100.0000 (96.3924) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 540/1403] eta: 0:01:28 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0605 (0.1496) acc1: 100.0000 (96.4591) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 550/1403] eta: 0:01:27 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0615 (0.1484) acc1: 100.0000 (96.5064) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 560/1403] eta: 0:01:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0834 (0.1474) acc1: 98.4375 (96.5408) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 570/1403] eta: 0:01:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0858 (0.1464) acc1: 98.4375 (96.5849) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 580/1403] eta: 0:01:24 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0912 (0.1460) acc1: 98.4375 (96.5980) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 590/1403] eta: 0:01:23 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1161 (0.1458) acc1: 98.4375 (96.5895) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 600/1403] eta: 0:01:21 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1373 (0.1463) acc1: 93.7500 (96.5396) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 610/1403] eta: 0:01:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1433 (0.1461) acc1: 93.7500 (96.5323) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 620/1403] eta: 0:01:19 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1481 (0.1474) acc1: 93.7500 (96.4221) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 630/1403] eta: 0:01:18 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1542 (0.1474) acc1: 95.3125 (96.4169) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0967 data: 0.0004 max mem: 52150 +Test: [ 640/1403] eta: 0:01:17 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1279 (0.1471) acc1: 96.8750 (96.4338) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 650/1403] eta: 0:01:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1248 (0.1469) acc1: 96.8750 (96.4334) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 660/1403] eta: 0:01:15 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0997 (0.1460) acc1: 98.4375 (96.4731) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [ 670/1403] eta: 0:01:14 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0731 (0.1450) acc1: 100.0000 (96.5211) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 680/1403] eta: 0:01:13 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0872 (0.1469) acc1: 98.4375 (96.3840) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 690/1403] eta: 0:01:12 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3029 (0.1523) acc1: 85.9375 (96.0791) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 700/1403] eta: 0:01:11 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2163 (0.1529) acc1: 89.0625 (96.0458) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 710/1403] eta: 0:01:10 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4039 (0.1588) acc1: 79.6875 (95.7322) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 720/1403] eta: 0:01:09 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4289 (0.1591) acc1: 78.1250 (95.7091) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 730/1403] eta: 0:01:08 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0694 (0.1581) acc1: 100.0000 (95.7571) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 740/1403] eta: 0:01:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0799 (0.1574) acc1: 100.0000 (95.7933) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 750/1403] eta: 0:01:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1214 (0.1592) acc1: 96.8750 (95.6870) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 760/1403] eta: 0:01:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3981 (0.1641) acc1: 81.2500 (95.3782) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 770/1403] eta: 0:01:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5565 (0.1695) acc1: 68.7500 (95.0511) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 780/1403] eta: 0:01:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5185 (0.1727) acc1: 71.8750 (94.8564) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 790/1403] eta: 0:01:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3475 (0.1755) acc1: 78.1250 (94.7021) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.0969 data: 0.0004 max mem: 52150 +Test: [ 800/1403] eta: 0:01:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2463 (0.1760) acc1: 87.5000 (94.6805) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0969 data: 0.0004 max mem: 52150 +Test: [ 810/1403] eta: 0:00:59 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2534 (0.1793) acc1: 87.5000 (94.5245) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 820/1403] eta: 0:00:58 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6559 (0.1885) acc1: 71.8750 (94.1382) acc5: 100.0000 (100.0000) acc1_neg: 40.625 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [ 830/1403] eta: 0:00:57 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.7582 (0.1918) acc1: 67.1875 (93.9700) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 840/1403] eta: 0:00:56 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5346 (0.1999) acc1: 68.7500 (93.6162) acc5: 100.0000 (100.0000) acc1_neg: 50.000 acc5_neg: 100.000 time: 0.0971 data: 0.0003 max mem: 52150 +Test: [ 850/1403] eta: 0:00:55 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.0624 (0.2109) acc1: 54.6875 (93.1496) acc5: 100.0000 (100.0000) acc1_neg: 53.125 acc5_neg: 100.000 time: 0.0972 data: 0.0003 max mem: 52150 +Test: [ 860/1403] eta: 0:00:54 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.0104 (0.2143) acc1: 56.2500 (92.9969) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 870/1403] eta: 0:00:53 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1313 (0.2131) acc1: 95.3125 (93.0450) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 880/1403] eta: 0:00:52 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0983 (0.2119) acc1: 98.4375 (93.1009) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 890/1403] eta: 0:00:51 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0860 (0.2105) acc1: 98.4375 (93.1713) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 900/1403] eta: 0:00:50 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0756 (0.2090) acc1: 100.0000 (93.2436) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 910/1403] eta: 0:00:49 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0825 (0.2080) acc1: 100.0000 (93.2869) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 920/1403] eta: 0:00:48 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1087 (0.2071) acc1: 96.8750 (93.3242) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 930/1403] eta: 0:00:47 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1033 (0.2060) acc1: 96.8750 (93.3774) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [ 940/1403] eta: 0:00:46 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1205 (0.2061) acc1: 96.8750 (93.3565) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [ 950/1403] eta: 0:00:45 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2560 (0.2069) acc1: 89.0625 (93.2916) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 960/1403] eta: 0:00:44 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2801 (0.2075) acc1: 87.5000 (93.2443) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 970/1403] eta: 0:00:43 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2580 (0.2084) acc1: 89.0625 (93.1916) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 980/1403] eta: 0:00:42 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2693 (0.2093) acc1: 85.9375 (93.1177) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 990/1403] eta: 0:00:41 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2501 (0.2097) acc1: 89.0625 (93.0909) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [1000/1403] eta: 0:00:40 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2359 (0.2104) acc1: 92.1875 (93.0538) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [1010/1403] eta: 0:00:39 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2637 (0.2112) acc1: 89.0625 (93.0051) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [1020/1403] eta: 0:00:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4267 (0.2180) acc1: 79.6875 (92.6267) acc5: 100.0000 (100.0000) acc1_neg: 29.688 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [1030/1403] eta: 0:00:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6246 (0.2204) acc1: 59.3750 (92.4815) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [1040/1403] eta: 0:00:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2388 (0.2203) acc1: 89.0625 (92.4772) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [1050/1403] eta: 0:00:35 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2058 (0.2201) acc1: 93.7500 (92.4923) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [1060/1403] eta: 0:00:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1559 (0.2195) acc1: 95.3125 (92.5247) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [1070/1403] eta: 0:00:33 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1700 (0.2201) acc1: 95.3125 (92.4910) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1080/1403] eta: 0:00:32 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4005 (0.2222) acc1: 82.8125 (92.3884) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1090/1403] eta: 0:00:31 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4637 (0.2244) acc1: 82.8125 (92.2835) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1100/1403] eta: 0:00:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3740 (0.2252) acc1: 84.3750 (92.2443) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1110/1403] eta: 0:00:29 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2990 (0.2258) acc1: 87.5000 (92.2156) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1120/1403] eta: 0:00:28 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2994 (0.2265) acc1: 89.0625 (92.1763) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1130/1403] eta: 0:00:27 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3113 (0.2274) acc1: 87.5000 (92.1253) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1140/1403] eta: 0:00:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2769 (0.2277) acc1: 89.0625 (92.1136) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1150/1403] eta: 0:00:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2722 (0.2283) acc1: 89.0625 (92.0775) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1160/1403] eta: 0:00:24 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3023 (0.2290) acc1: 87.5000 (92.0435) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1170/1403] eta: 0:00:23 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2782 (0.2290) acc1: 89.0625 (92.0434) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1180/1403] eta: 0:00:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0718 (0.2276) acc1: 100.0000 (92.1108) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1190/1403] eta: 0:00:21 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0710 (0.2263) acc1: 100.0000 (92.1770) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1200/1403] eta: 0:00:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0716 (0.2250) acc1: 100.0000 (92.2421) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1210/1403] eta: 0:00:19 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0705 (0.2238) acc1: 100.0000 (92.3062) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1220/1403] eta: 0:00:18 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0704 (0.2225) acc1: 100.0000 (92.3692) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1230/1403] eta: 0:00:17 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0714 (0.2213) acc1: 100.0000 (92.4312) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1240/1403] eta: 0:00:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0713 (0.2201) acc1: 100.0000 (92.4922) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1250/1403] eta: 0:00:15 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0716 (0.2189) acc1: 100.0000 (92.5522) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1260/1403] eta: 0:00:14 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0771 (0.2178) acc1: 100.0000 (92.6113) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52150 +Test: [1270/1403] eta: 0:00:13 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0781 (0.2167) acc1: 100.0000 (92.6694) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1280/1403] eta: 0:00:12 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0780 (0.2156) acc1: 100.0000 (92.7266) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1290/1403] eta: 0:00:11 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0774 (0.2146) acc1: 100.0000 (92.7830) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1300/1403] eta: 0:00:10 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0783 (0.2135) acc1: 100.0000 (92.8384) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1310/1403] eta: 0:00:09 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0785 (0.2125) acc1: 100.0000 (92.8931) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52150 +Test: [1320/1403] eta: 0:00:08 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0771 (0.2115) acc1: 100.0000 (92.9469) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52150 +Test: [1330/1403] eta: 0:00:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0765 (0.2105) acc1: 100.0000 (92.9963) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1340/1403] eta: 0:00:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0739 (0.2095) acc1: 100.0000 (93.0439) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1350/1403] eta: 0:00:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0739 (0.2085) acc1: 100.0000 (93.0908) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1360/1403] eta: 0:00:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0736 (0.2075) acc1: 100.0000 (93.1381) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1370/1403] eta: 0:00:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0744 (0.2066) acc1: 100.0000 (93.1847) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1380/1403] eta: 0:00:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0723 (0.2057) acc1: 100.0000 (93.2307) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1390/1403] eta: 0:00:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0708 (0.2047) acc1: 100.0000 (93.2793) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1400/1403] eta: 0:00:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0724 (0.2039) acc1: 100.0000 (93.3240) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0001 max mem: 52150 +Test: [1402/1403] eta: 0:00:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0732 (0.2037) acc1: 100.0000 (93.3277) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0922 data: 0.0001 max mem: 52150 +Test: Total time: 0:02:19 (0.0994 s / it) +* Overall Acc@1 93.328 Acc@5 100.000 + target == real Acc@1 96.771 Acc@5 100.000 + target == fake Acc@1 92.357 Acc@5 100.000 + loss 0.204 +Accuracy of the network on the 89729 test images: 93.3% +Max accuracy: 94.28% +Epoch: [26] [ 0/7269] eta: 12:33:59 lr: 0.000004 loss: 0.5745 (0.5745) time: 6.2236 data: 4.4335 max mem: 52150 +Epoch: [26] [ 10/7269] eta: 2:16:48 lr: 0.000004 loss: 0.5745 (0.5576) time: 1.1309 data: 0.4032 max mem: 52150 +Epoch: [26] [ 20/7269] eta: 1:52:40 lr: 0.000004 loss: 0.5584 (0.5438) time: 0.6681 data: 0.0002 max mem: 52150 +Epoch: [26] [ 30/7269] eta: 1:30:40 lr: 0.000004 loss: 0.5155 (0.5268) time: 0.5430 data: 0.0002 max mem: 52150 +Epoch: [26] [ 40/7269] eta: 1:19:23 lr: 0.000004 loss: 0.5171 (0.5237) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [26] [ 50/7269] eta: 1:12:30 lr: 0.000004 loss: 0.5226 (0.5206) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [26] [ 60/7269] eta: 1:07:51 lr: 0.000004 loss: 0.5250 (0.5230) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [26] [ 70/7269] eta: 1:06:14 lr: 0.000004 loss: 0.5163 (0.5184) time: 0.4231 data: 0.0002 max mem: 52150 +Epoch: [26] [ 80/7269] eta: 1:05:21 lr: 0.000004 loss: 0.4719 (0.5114) time: 0.4868 data: 0.0002 max mem: 52150 +Epoch: [26] [ 90/7269] eta: 1:02:59 lr: 0.000004 loss: 0.4822 (0.5103) time: 0.4355 data: 0.0002 max mem: 52150 +Epoch: [26] [ 100/7269] eta: 1:03:39 lr: 0.000004 loss: 0.4961 (0.5080) time: 0.4813 data: 0.0002 max mem: 52150 +Epoch: [26] [ 110/7269] eta: 1:05:52 lr: 0.000004 loss: 0.4656 (0.5028) time: 0.6691 data: 0.0002 max mem: 52150 +Epoch: [26] [ 120/7269] eta: 1:04:00 lr: 0.000004 loss: 0.4981 (0.5058) time: 0.5594 data: 0.0002 max mem: 52150 +Epoch: [26] [ 130/7269] eta: 1:02:25 lr: 0.000004 loss: 0.5359 (0.5095) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [26] [ 140/7269] eta: 1:01:02 lr: 0.000004 loss: 0.4970 (0.5062) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [26] [ 150/7269] eta: 0:59:50 lr: 0.000004 loss: 0.4488 (0.5077) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [26] [ 160/7269] eta: 1:00:13 lr: 0.000004 loss: 0.5093 (0.5076) time: 0.4698 data: 0.0002 max mem: 52150 +Epoch: [26] [ 170/7269] eta: 0:59:11 lr: 0.000004 loss: 0.5320 (0.5073) time: 0.4702 data: 0.0002 max mem: 52150 +Epoch: [26] [ 180/7269] eta: 0:58:16 lr: 0.000004 loss: 0.5269 (0.5062) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [26] [ 190/7269] eta: 0:58:12 lr: 0.000004 loss: 0.4845 (0.5063) time: 0.4338 data: 0.0002 max mem: 52150 +Epoch: [26] [ 200/7269] eta: 0:57:25 lr: 0.000004 loss: 0.5211 (0.5079) time: 0.4337 data: 0.0002 max mem: 52150 +Epoch: [26] [ 210/7269] eta: 0:56:41 lr: 0.000004 loss: 0.5096 (0.5080) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [26] [ 220/7269] eta: 0:56:01 lr: 0.000004 loss: 0.4911 (0.5070) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [26] [ 230/7269] eta: 0:55:24 lr: 0.000004 loss: 0.5031 (0.5071) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [26] [ 240/7269] eta: 0:55:46 lr: 0.000004 loss: 0.5178 (0.5069) time: 0.4674 data: 0.0002 max mem: 52150 +Epoch: [26] [ 250/7269] eta: 0:55:48 lr: 0.000004 loss: 0.5184 (0.5078) time: 0.5306 data: 0.0002 max mem: 52150 +Epoch: [26] [ 260/7269] eta: 0:55:15 lr: 0.000004 loss: 0.5253 (0.5081) time: 0.4351 data: 0.0002 max mem: 52150 +Epoch: [26] [ 270/7269] eta: 0:55:10 lr: 0.000004 loss: 0.5483 (0.5086) time: 0.4222 data: 0.0002 max mem: 52150 +Epoch: [26] [ 280/7269] eta: 0:54:40 lr: 0.000004 loss: 0.5411 (0.5093) time: 0.4221 data: 0.0002 max mem: 52150 +Epoch: [26] [ 290/7269] eta: 0:55:06 lr: 0.000004 loss: 0.5551 (0.5106) time: 0.4852 data: 0.0002 max mem: 52150 +Epoch: [26] [ 300/7269] eta: 0:55:00 lr: 0.000004 loss: 0.5330 (0.5111) time: 0.5336 data: 0.0002 max mem: 52150 +Epoch: [26] [ 310/7269] eta: 0:54:33 lr: 0.000004 loss: 0.5316 (0.5108) time: 0.4206 data: 0.0002 max mem: 52150 +Epoch: [26] [ 320/7269] eta: 0:54:07 lr: 0.000004 loss: 0.4993 (0.5112) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [26] [ 330/7269] eta: 0:53:42 lr: 0.000004 loss: 0.5152 (0.5114) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [26] [ 340/7269] eta: 0:53:19 lr: 0.000004 loss: 0.5305 (0.5124) time: 0.3727 data: 0.0002 max mem: 52150 +Epoch: [26] [ 350/7269] eta: 0:53:42 lr: 0.000004 loss: 0.5176 (0.5125) time: 0.4869 data: 0.0002 max mem: 52150 +Epoch: [26] [ 360/7269] eta: 0:53:19 lr: 0.000004 loss: 0.5070 (0.5128) time: 0.4871 data: 0.0002 max mem: 52150 +Epoch: [26] [ 370/7269] eta: 0:52:58 lr: 0.000004 loss: 0.5342 (0.5138) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [26] [ 380/7269] eta: 0:52:37 lr: 0.000004 loss: 0.5408 (0.5144) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [26] [ 390/7269] eta: 0:52:17 lr: 0.000004 loss: 0.5207 (0.5139) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [26] [ 400/7269] eta: 0:51:59 lr: 0.000004 loss: 0.5137 (0.5135) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [26] [ 410/7269] eta: 0:52:01 lr: 0.000004 loss: 0.4914 (0.5130) time: 0.4348 data: 0.0002 max mem: 52150 +Epoch: [26] [ 420/7269] eta: 0:51:43 lr: 0.000004 loss: 0.4907 (0.5129) time: 0.4336 data: 0.0002 max mem: 52150 +Epoch: [26] [ 430/7269] eta: 0:51:26 lr: 0.000004 loss: 0.5103 (0.5130) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [26] [ 440/7269] eta: 0:51:28 lr: 0.000004 loss: 0.4956 (0.5127) time: 0.4354 data: 0.0002 max mem: 52150 +Epoch: [26] [ 450/7269] eta: 0:51:12 lr: 0.000004 loss: 0.5096 (0.5132) time: 0.4354 data: 0.0002 max mem: 52150 +Epoch: [26] [ 460/7269] eta: 0:51:14 lr: 0.000004 loss: 0.5096 (0.5130) time: 0.4348 data: 0.0002 max mem: 52150 +Epoch: [26] [ 470/7269] eta: 0:50:58 lr: 0.000004 loss: 0.5070 (0.5131) time: 0.4348 data: 0.0002 max mem: 52150 +Epoch: [26] [ 480/7269] eta: 0:51:10 lr: 0.000004 loss: 0.5245 (0.5128) time: 0.4686 data: 0.0002 max mem: 52150 +Epoch: [26] [ 490/7269] eta: 0:50:54 lr: 0.000004 loss: 0.5245 (0.5128) time: 0.4681 data: 0.0002 max mem: 52150 +Epoch: [26] [ 500/7269] eta: 0:51:12 lr: 0.000004 loss: 0.5301 (0.5126) time: 0.4947 data: 0.0002 max mem: 52150 +Epoch: [26] [ 510/7269] eta: 0:50:57 lr: 0.000004 loss: 0.5138 (0.5125) time: 0.4951 data: 0.0002 max mem: 52150 +Epoch: [26] [ 520/7269] eta: 0:50:42 lr: 0.000004 loss: 0.5035 (0.5116) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [26] [ 530/7269] eta: 0:50:56 lr: 0.000004 loss: 0.5035 (0.5112) time: 0.4841 data: 0.0002 max mem: 52150 +Epoch: [26] [ 540/7269] eta: 0:50:41 lr: 0.000004 loss: 0.5209 (0.5117) time: 0.4844 data: 0.0002 max mem: 52150 +Epoch: [26] [ 550/7269] eta: 0:50:41 lr: 0.000004 loss: 0.5434 (0.5119) time: 0.4309 data: 0.0002 max mem: 52150 +Epoch: [26] [ 560/7269] eta: 0:50:27 lr: 0.000004 loss: 0.5336 (0.5124) time: 0.4307 data: 0.0002 max mem: 52150 +Epoch: [26] [ 570/7269] eta: 0:50:27 lr: 0.000004 loss: 0.5465 (0.5128) time: 0.4317 data: 0.0002 max mem: 52150 +Epoch: [26] [ 580/7269] eta: 0:50:14 lr: 0.000004 loss: 0.4882 (0.5123) time: 0.4318 data: 0.0002 max mem: 52150 +Epoch: [26] [ 590/7269] eta: 0:50:00 lr: 0.000004 loss: 0.4841 (0.5120) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [26] [ 600/7269] eta: 0:49:47 lr: 0.000004 loss: 0.5127 (0.5126) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [26] [ 610/7269] eta: 0:49:35 lr: 0.000004 loss: 0.5197 (0.5124) time: 0.3729 data: 0.0002 max mem: 52150 +Epoch: [26] [ 620/7269] eta: 0:49:32 lr: 0.000004 loss: 0.5516 (0.5131) time: 0.4182 data: 0.0002 max mem: 52150 +Epoch: [26] [ 630/7269] eta: 0:49:32 lr: 0.000004 loss: 0.5516 (0.5135) time: 0.4776 data: 0.0002 max mem: 52150 +Epoch: [26] [ 640/7269] eta: 0:49:42 lr: 0.000004 loss: 0.5370 (0.5135) time: 0.5409 data: 0.0002 max mem: 52150 +Epoch: [26] [ 650/7269] eta: 0:49:30 lr: 0.000004 loss: 0.5313 (0.5134) time: 0.4809 data: 0.0002 max mem: 52150 +Epoch: [26] [ 660/7269] eta: 0:49:30 lr: 0.000004 loss: 0.5332 (0.5139) time: 0.4347 data: 0.0002 max mem: 52150 +Epoch: [26] [ 670/7269] eta: 0:49:18 lr: 0.000004 loss: 0.5490 (0.5139) time: 0.4343 data: 0.0002 max mem: 52150 +Epoch: [26] [ 680/7269] eta: 0:49:30 lr: 0.000004 loss: 0.5186 (0.5137) time: 0.4932 data: 0.0002 max mem: 52150 +Epoch: [26] [ 690/7269] eta: 0:49:18 lr: 0.000004 loss: 0.5237 (0.5139) time: 0.4934 data: 0.0002 max mem: 52150 +Epoch: [26] [ 700/7269] eta: 0:49:06 lr: 0.000004 loss: 0.5667 (0.5145) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [26] [ 710/7269] eta: 0:48:55 lr: 0.000004 loss: 0.5232 (0.5143) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [26] [ 720/7269] eta: 0:48:43 lr: 0.000004 loss: 0.5114 (0.5145) time: 0.3728 data: 0.0002 max mem: 52150 +Epoch: [26] [ 730/7269] eta: 0:48:32 lr: 0.000004 loss: 0.5065 (0.5141) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [26] [ 740/7269] eta: 0:48:29 lr: 0.000004 loss: 0.4973 (0.5142) time: 0.4157 data: 0.0002 max mem: 52150 +Epoch: [26] [ 750/7269] eta: 0:48:18 lr: 0.000004 loss: 0.5164 (0.5141) time: 0.4159 data: 0.0002 max mem: 52150 +Epoch: [26] [ 760/7269] eta: 0:48:18 lr: 0.000004 loss: 0.5041 (0.5138) time: 0.4353 data: 0.0002 max mem: 52150 +Epoch: [26] [ 770/7269] eta: 0:48:08 lr: 0.000004 loss: 0.4979 (0.5136) time: 0.4349 data: 0.0002 max mem: 52150 +Epoch: [26] [ 780/7269] eta: 0:47:57 lr: 0.000004 loss: 0.5085 (0.5136) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [26] [ 790/7269] eta: 0:47:47 lr: 0.000004 loss: 0.5328 (0.5138) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [26] [ 800/7269] eta: 0:47:37 lr: 0.000004 loss: 0.5172 (0.5135) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [26] [ 810/7269] eta: 0:47:27 lr: 0.000004 loss: 0.4772 (0.5133) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [26] [ 820/7269] eta: 0:47:26 lr: 0.000004 loss: 0.4855 (0.5132) time: 0.4314 data: 0.0002 max mem: 52150 +Epoch: [26] [ 830/7269] eta: 0:47:26 lr: 0.000004 loss: 0.4964 (0.5132) time: 0.4920 data: 0.0002 max mem: 52150 +Epoch: [26] [ 840/7269] eta: 0:47:23 lr: 0.000004 loss: 0.5207 (0.5132) time: 0.4809 data: 0.0002 max mem: 52150 +Epoch: [26] [ 850/7269] eta: 0:47:30 lr: 0.000004 loss: 0.5295 (0.5134) time: 0.5255 data: 0.0002 max mem: 52150 +Epoch: [26] [ 860/7269] eta: 0:47:26 lr: 0.000004 loss: 0.5295 (0.5134) time: 0.5229 data: 0.0002 max mem: 52150 +Epoch: [26] [ 870/7269] eta: 0:47:17 lr: 0.000004 loss: 0.4987 (0.5134) time: 0.4171 data: 0.0002 max mem: 52150 +Epoch: [26] [ 880/7269] eta: 0:47:07 lr: 0.000004 loss: 0.5266 (0.5138) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [26] [ 890/7269] eta: 0:47:06 lr: 0.000004 loss: 0.5272 (0.5140) time: 0.4327 data: 0.0002 max mem: 52150 +Epoch: [26] [ 900/7269] eta: 0:46:57 lr: 0.000004 loss: 0.5603 (0.5142) time: 0.4327 data: 0.0002 max mem: 52150 +Epoch: [26] [ 910/7269] eta: 0:46:54 lr: 0.000004 loss: 0.5565 (0.5145) time: 0.4175 data: 0.0002 max mem: 52150 +Epoch: [26] [ 920/7269] eta: 0:46:59 lr: 0.000004 loss: 0.5042 (0.5141) time: 0.5222 data: 0.0002 max mem: 52150 +Epoch: [26] [ 930/7269] eta: 0:46:50 lr: 0.000004 loss: 0.5051 (0.5146) time: 0.4770 data: 0.0002 max mem: 52150 +Epoch: [26] [ 940/7269] eta: 0:46:55 lr: 0.000004 loss: 0.5314 (0.5145) time: 0.4785 data: 0.0002 max mem: 52150 +Epoch: [26] [ 950/7269] eta: 0:46:45 lr: 0.000004 loss: 0.5153 (0.5147) time: 0.4785 data: 0.0002 max mem: 52150 +Epoch: [26] [ 960/7269] eta: 0:46:43 lr: 0.000004 loss: 0.5374 (0.5149) time: 0.4219 data: 0.0002 max mem: 52150 +Epoch: [26] [ 970/7269] eta: 0:46:34 lr: 0.000004 loss: 0.5184 (0.5150) time: 0.4219 data: 0.0002 max mem: 52150 +Epoch: [26] [ 980/7269] eta: 0:46:25 lr: 0.000004 loss: 0.5184 (0.5150) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [26] [ 990/7269] eta: 0:46:31 lr: 0.000004 loss: 0.5254 (0.5151) time: 0.4925 data: 0.0002 max mem: 52150 +Epoch: [26] [1000/7269] eta: 0:46:22 lr: 0.000004 loss: 0.5218 (0.5154) time: 0.4926 data: 0.0002 max mem: 52150 +Epoch: [26] [1010/7269] eta: 0:46:13 lr: 0.000004 loss: 0.5331 (0.5155) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [26] [1020/7269] eta: 0:46:04 lr: 0.000004 loss: 0.5494 (0.5158) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [26] [1030/7269] eta: 0:46:03 lr: 0.000004 loss: 0.5344 (0.5157) time: 0.4314 data: 0.0002 max mem: 52150 +Epoch: [26] [1040/7269] eta: 0:46:01 lr: 0.000004 loss: 0.5344 (0.5161) time: 0.4918 data: 0.0002 max mem: 52150 +Epoch: [26] [1050/7269] eta: 0:45:58 lr: 0.000004 loss: 0.5557 (0.5165) time: 0.4771 data: 0.0002 max mem: 52150 +Epoch: [26] [1060/7269] eta: 0:45:57 lr: 0.000004 loss: 0.5420 (0.5167) time: 0.4775 data: 0.0002 max mem: 52150 +Epoch: [26] [1070/7269] eta: 0:46:07 lr: 0.000004 loss: 0.4922 (0.5163) time: 0.6010 data: 0.0002 max mem: 52150 +Epoch: [26] [1080/7269] eta: 0:45:59 lr: 0.000004 loss: 0.5126 (0.5163) time: 0.5405 data: 0.0002 max mem: 52150 +Epoch: [26] [1090/7269] eta: 0:45:50 lr: 0.000004 loss: 0.5274 (0.5161) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [26] [1100/7269] eta: 0:45:42 lr: 0.000004 loss: 0.4985 (0.5159) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [26] [1110/7269] eta: 0:45:33 lr: 0.000004 loss: 0.5060 (0.5159) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [26] [1120/7269] eta: 0:45:30 lr: 0.000004 loss: 0.5382 (0.5162) time: 0.4172 data: 0.0002 max mem: 52150 +Epoch: [26] [1130/7269] eta: 0:45:21 lr: 0.000004 loss: 0.5257 (0.5159) time: 0.4174 data: 0.0002 max mem: 52150 +Epoch: [26] [1140/7269] eta: 0:45:13 lr: 0.000004 loss: 0.5081 (0.5160) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [26] [1150/7269] eta: 0:45:11 lr: 0.000004 loss: 0.5313 (0.5161) time: 0.4334 data: 0.0002 max mem: 52150 +Epoch: [26] [1160/7269] eta: 0:45:03 lr: 0.000004 loss: 0.5024 (0.5158) time: 0.4328 data: 0.0002 max mem: 52150 +Epoch: [26] [1170/7269] eta: 0:44:55 lr: 0.000004 loss: 0.5179 (0.5161) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [26] [1180/7269] eta: 0:44:47 lr: 0.000004 loss: 0.5354 (0.5161) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [26] [1190/7269] eta: 0:44:39 lr: 0.000004 loss: 0.5040 (0.5160) time: 0.3728 data: 0.0002 max mem: 52150 +Epoch: [26] [1200/7269] eta: 0:44:31 lr: 0.000004 loss: 0.5343 (0.5161) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [26] [1210/7269] eta: 0:44:35 lr: 0.000004 loss: 0.5134 (0.5159) time: 0.4839 data: 0.0002 max mem: 52150 +Epoch: [26] [1220/7269] eta: 0:44:27 lr: 0.000004 loss: 0.5242 (0.5160) time: 0.4837 data: 0.0002 max mem: 52150 +Epoch: [26] [1230/7269] eta: 0:44:19 lr: 0.000004 loss: 0.5388 (0.5161) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [26] [1240/7269] eta: 0:44:11 lr: 0.000004 loss: 0.5147 (0.5159) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [26] [1250/7269] eta: 0:44:04 lr: 0.000004 loss: 0.5029 (0.5158) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [26] [1260/7269] eta: 0:43:56 lr: 0.000004 loss: 0.5102 (0.5156) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [26] [1270/7269] eta: 0:43:48 lr: 0.000004 loss: 0.5096 (0.5154) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [26] [1280/7269] eta: 0:43:47 lr: 0.000004 loss: 0.5278 (0.5154) time: 0.4363 data: 0.0002 max mem: 52150 +Epoch: [26] [1290/7269] eta: 0:43:39 lr: 0.000004 loss: 0.5277 (0.5155) time: 0.4361 data: 0.0002 max mem: 52150 +Epoch: [26] [1300/7269] eta: 0:43:37 lr: 0.000004 loss: 0.5277 (0.5157) time: 0.4301 data: 0.0002 max mem: 52150 +Epoch: [26] [1310/7269] eta: 0:43:30 lr: 0.000004 loss: 0.5547 (0.5156) time: 0.4300 data: 0.0002 max mem: 52150 +Epoch: [26] [1320/7269] eta: 0:43:31 lr: 0.000004 loss: 0.4759 (0.5152) time: 0.4662 data: 0.0002 max mem: 52150 +Epoch: [26] [1330/7269] eta: 0:43:39 lr: 0.000004 loss: 0.5029 (0.5153) time: 0.6344 data: 0.0002 max mem: 52150 +Epoch: [26] [1340/7269] eta: 0:43:35 lr: 0.000004 loss: 0.5387 (0.5152) time: 0.5862 data: 0.0002 max mem: 52150 +Epoch: [26] [1350/7269] eta: 0:43:33 lr: 0.000004 loss: 0.4998 (0.5152) time: 0.4790 data: 0.0002 max mem: 52150 +Epoch: [26] [1360/7269] eta: 0:43:26 lr: 0.000004 loss: 0.5101 (0.5151) time: 0.4322 data: 0.0002 max mem: 52150 +Epoch: [26] [1370/7269] eta: 0:43:22 lr: 0.000004 loss: 0.5101 (0.5149) time: 0.4167 data: 0.0002 max mem: 52150 +Epoch: [26] [1380/7269] eta: 0:43:20 lr: 0.000004 loss: 0.5067 (0.5150) time: 0.4768 data: 0.0002 max mem: 52150 +Epoch: [26] [1390/7269] eta: 0:43:13 lr: 0.000004 loss: 0.5134 (0.5148) time: 0.4319 data: 0.0002 max mem: 52150 +Epoch: [26] [1400/7269] eta: 0:43:05 lr: 0.000004 loss: 0.5134 (0.5149) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [26] [1410/7269] eta: 0:43:06 lr: 0.000004 loss: 0.5421 (0.5151) time: 0.4629 data: 0.0002 max mem: 52150 +Epoch: [26] [1420/7269] eta: 0:43:02 lr: 0.000004 loss: 0.5421 (0.5152) time: 0.5122 data: 0.0002 max mem: 52150 +Epoch: [26] [1430/7269] eta: 0:42:55 lr: 0.000004 loss: 0.5283 (0.5151) time: 0.4213 data: 0.0002 max mem: 52150 +Epoch: [26] [1440/7269] eta: 0:42:48 lr: 0.000004 loss: 0.5173 (0.5151) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [26] [1450/7269] eta: 0:42:41 lr: 0.000004 loss: 0.4948 (0.5148) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [26] [1460/7269] eta: 0:42:34 lr: 0.000004 loss: 0.4761 (0.5147) time: 0.3711 data: 0.0002 max mem: 52150 +Epoch: [26] [1470/7269] eta: 0:42:26 lr: 0.000004 loss: 0.5022 (0.5147) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [26] [1480/7269] eta: 0:42:19 lr: 0.000004 loss: 0.5192 (0.5146) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [26] [1490/7269] eta: 0:42:12 lr: 0.000004 loss: 0.5192 (0.5148) time: 0.3712 data: 0.0002 max mem: 52150 +Epoch: [26] [1500/7269] eta: 0:42:05 lr: 0.000004 loss: 0.5458 (0.5149) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [26] [1510/7269] eta: 0:41:59 lr: 0.000004 loss: 0.5551 (0.5151) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [26] [1520/7269] eta: 0:41:55 lr: 0.000004 loss: 0.5344 (0.5149) time: 0.4190 data: 0.0002 max mem: 52150 +Epoch: [26] [1530/7269] eta: 0:41:53 lr: 0.000004 loss: 0.5080 (0.5149) time: 0.4788 data: 0.0002 max mem: 52150 +Epoch: [26] [1540/7269] eta: 0:41:51 lr: 0.000004 loss: 0.5070 (0.5146) time: 0.4916 data: 0.0002 max mem: 52150 +Epoch: [26] [1550/7269] eta: 0:41:48 lr: 0.000004 loss: 0.4945 (0.5147) time: 0.4924 data: 0.0002 max mem: 52150 +Epoch: [26] [1560/7269] eta: 0:41:41 lr: 0.000004 loss: 0.5388 (0.5148) time: 0.4328 data: 0.0002 max mem: 52150 +Epoch: [26] [1570/7269] eta: 0:41:35 lr: 0.000004 loss: 0.5392 (0.5148) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [26] [1580/7269] eta: 0:41:32 lr: 0.000004 loss: 0.5063 (0.5147) time: 0.4339 data: 0.0002 max mem: 52150 +Epoch: [26] [1590/7269] eta: 0:41:26 lr: 0.000004 loss: 0.4968 (0.5144) time: 0.4341 data: 0.0002 max mem: 52150 +Epoch: [26] [1600/7269] eta: 0:41:26 lr: 0.000004 loss: 0.4834 (0.5144) time: 0.4784 data: 0.0002 max mem: 52150 +Epoch: [26] [1610/7269] eta: 0:41:27 lr: 0.000004 loss: 0.5493 (0.5147) time: 0.5836 data: 0.0002 max mem: 52150 +Epoch: [26] [1620/7269] eta: 0:41:20 lr: 0.000004 loss: 0.5445 (0.5145) time: 0.4779 data: 0.0002 max mem: 52150 +Epoch: [26] [1630/7269] eta: 0:41:18 lr: 0.000004 loss: 0.4906 (0.5144) time: 0.4312 data: 0.0002 max mem: 52150 +Epoch: [26] [1640/7269] eta: 0:41:11 lr: 0.000004 loss: 0.5014 (0.5145) time: 0.4310 data: 0.0002 max mem: 52150 +Epoch: [26] [1650/7269] eta: 0:41:04 lr: 0.000004 loss: 0.5257 (0.5145) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [26] [1660/7269] eta: 0:41:02 lr: 0.000004 loss: 0.5257 (0.5146) time: 0.4324 data: 0.0002 max mem: 52150 +Epoch: [26] [1670/7269] eta: 0:40:55 lr: 0.000004 loss: 0.5342 (0.5147) time: 0.4326 data: 0.0002 max mem: 52150 +Epoch: [26] [1680/7269] eta: 0:40:52 lr: 0.000004 loss: 0.5326 (0.5146) time: 0.4223 data: 0.0002 max mem: 52150 +Epoch: [26] [1690/7269] eta: 0:40:45 lr: 0.000004 loss: 0.4956 (0.5145) time: 0.4224 data: 0.0002 max mem: 52150 +Epoch: [26] [1700/7269] eta: 0:40:39 lr: 0.000004 loss: 0.5200 (0.5147) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [26] [1710/7269] eta: 0:40:36 lr: 0.000004 loss: 0.5824 (0.5151) time: 0.4243 data: 0.0002 max mem: 52150 +Epoch: [26] [1720/7269] eta: 0:40:29 lr: 0.000004 loss: 0.5593 (0.5153) time: 0.4250 data: 0.0002 max mem: 52150 +Epoch: [26] [1730/7269] eta: 0:40:23 lr: 0.000004 loss: 0.5298 (0.5150) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [26] [1740/7269] eta: 0:40:16 lr: 0.000004 loss: 0.4967 (0.5150) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [26] [1750/7269] eta: 0:40:10 lr: 0.000004 loss: 0.5245 (0.5151) time: 0.3727 data: 0.0002 max mem: 52150 +Epoch: [26] [1760/7269] eta: 0:40:03 lr: 0.000004 loss: 0.5073 (0.5151) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [26] [1770/7269] eta: 0:39:57 lr: 0.000004 loss: 0.5100 (0.5151) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [26] [1780/7269] eta: 0:39:51 lr: 0.000004 loss: 0.5103 (0.5152) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [26] [1790/7269] eta: 0:39:54 lr: 0.000004 loss: 0.4975 (0.5152) time: 0.5303 data: 0.0002 max mem: 52150 +Epoch: [26] [1800/7269] eta: 0:39:55 lr: 0.000004 loss: 0.4805 (0.5150) time: 0.6554 data: 0.0002 max mem: 52150 +Epoch: [26] [1810/7269] eta: 0:39:49 lr: 0.000004 loss: 0.4742 (0.5149) time: 0.4971 data: 0.0002 max mem: 52150 +Epoch: [26] [1820/7269] eta: 0:39:43 lr: 0.000004 loss: 0.5242 (0.5149) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [26] [1830/7269] eta: 0:39:36 lr: 0.000004 loss: 0.5292 (0.5149) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [26] [1840/7269] eta: 0:39:33 lr: 0.000004 loss: 0.5290 (0.5150) time: 0.4187 data: 0.0002 max mem: 52150 +Epoch: [26] [1850/7269] eta: 0:39:26 lr: 0.000004 loss: 0.5314 (0.5151) time: 0.4187 data: 0.0002 max mem: 52150 +Epoch: [26] [1860/7269] eta: 0:39:24 lr: 0.000004 loss: 0.5655 (0.5152) time: 0.4331 data: 0.0002 max mem: 52150 +Epoch: [26] [1870/7269] eta: 0:39:20 lr: 0.000004 loss: 0.5399 (0.5153) time: 0.4788 data: 0.0002 max mem: 52150 +Epoch: [26] [1880/7269] eta: 0:39:21 lr: 0.000004 loss: 0.5334 (0.5152) time: 0.5386 data: 0.0002 max mem: 52150 +Epoch: [26] [1890/7269] eta: 0:39:15 lr: 0.000004 loss: 0.5476 (0.5154) time: 0.4929 data: 0.0002 max mem: 52150 +Epoch: [26] [1900/7269] eta: 0:39:08 lr: 0.000004 loss: 0.5205 (0.5154) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [26] [1910/7269] eta: 0:39:02 lr: 0.000004 loss: 0.5225 (0.5156) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [26] [1920/7269] eta: 0:38:56 lr: 0.000004 loss: 0.5225 (0.5154) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [26] [1930/7269] eta: 0:38:52 lr: 0.000004 loss: 0.4556 (0.5151) time: 0.4175 data: 0.0002 max mem: 52150 +Epoch: [26] [1940/7269] eta: 0:38:46 lr: 0.000004 loss: 0.5171 (0.5153) time: 0.4173 data: 0.0002 max mem: 52150 +Epoch: [26] [1950/7269] eta: 0:38:43 lr: 0.000004 loss: 0.5548 (0.5154) time: 0.4335 data: 0.0002 max mem: 52150 +Epoch: [26] [1960/7269] eta: 0:38:37 lr: 0.000004 loss: 0.5083 (0.5154) time: 0.4331 data: 0.0002 max mem: 52150 +Epoch: [26] [1970/7269] eta: 0:38:34 lr: 0.000004 loss: 0.5063 (0.5153) time: 0.4324 data: 0.0002 max mem: 52150 +Epoch: [26] [1980/7269] eta: 0:38:28 lr: 0.000004 loss: 0.4923 (0.5153) time: 0.4328 data: 0.0002 max mem: 52150 +Epoch: [26] [1990/7269] eta: 0:38:22 lr: 0.000004 loss: 0.4976 (0.5153) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [26] [2000/7269] eta: 0:38:16 lr: 0.000004 loss: 0.5495 (0.5155) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [26] [2010/7269] eta: 0:38:13 lr: 0.000004 loss: 0.5420 (0.5155) time: 0.4181 data: 0.0002 max mem: 52150 +Epoch: [26] [2020/7269] eta: 0:38:09 lr: 0.000004 loss: 0.5408 (0.5157) time: 0.4619 data: 0.0002 max mem: 52150 +Epoch: [26] [2030/7269] eta: 0:38:06 lr: 0.000004 loss: 0.5447 (0.5158) time: 0.4743 data: 0.0002 max mem: 52150 +Epoch: [26] [2040/7269] eta: 0:38:03 lr: 0.000004 loss: 0.5447 (0.5160) time: 0.4906 data: 0.0002 max mem: 52150 +Epoch: [26] [2050/7269] eta: 0:37:57 lr: 0.000004 loss: 0.5279 (0.5160) time: 0.4313 data: 0.0002 max mem: 52150 +Epoch: [26] [2060/7269] eta: 0:37:56 lr: 0.000004 loss: 0.5199 (0.5161) time: 0.4778 data: 0.0002 max mem: 52150 +Epoch: [26] [2070/7269] eta: 0:37:56 lr: 0.000004 loss: 0.5533 (0.5162) time: 0.5985 data: 0.0002 max mem: 52150 +Epoch: [26] [2080/7269] eta: 0:37:50 lr: 0.000004 loss: 0.5385 (0.5163) time: 0.4924 data: 0.0002 max mem: 52150 +Epoch: [26] [2090/7269] eta: 0:37:47 lr: 0.000004 loss: 0.5385 (0.5162) time: 0.4213 data: 0.0002 max mem: 52150 +Epoch: [26] [2100/7269] eta: 0:37:41 lr: 0.000004 loss: 0.5268 (0.5163) time: 0.4209 data: 0.0002 max mem: 52150 +Epoch: [26] [2110/7269] eta: 0:37:35 lr: 0.000004 loss: 0.5542 (0.5165) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [26] [2120/7269] eta: 0:37:29 lr: 0.000004 loss: 0.5568 (0.5166) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [26] [2130/7269] eta: 0:37:29 lr: 0.000004 loss: 0.5568 (0.5167) time: 0.4936 data: 0.0002 max mem: 52150 +Epoch: [26] [2140/7269] eta: 0:37:25 lr: 0.000004 loss: 0.5099 (0.5167) time: 0.5417 data: 0.0002 max mem: 52150 +Epoch: [26] [2150/7269] eta: 0:37:19 lr: 0.000004 loss: 0.5099 (0.5168) time: 0.4205 data: 0.0002 max mem: 52150 +Epoch: [26] [2160/7269] eta: 0:37:16 lr: 0.000004 loss: 0.5716 (0.5169) time: 0.4332 data: 0.0002 max mem: 52150 +Epoch: [26] [2170/7269] eta: 0:37:10 lr: 0.000004 loss: 0.5362 (0.5169) time: 0.4328 data: 0.0002 max mem: 52150 +Epoch: [26] [2180/7269] eta: 0:37:04 lr: 0.000004 loss: 0.5362 (0.5169) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [26] [2190/7269] eta: 0:37:03 lr: 0.000004 loss: 0.5093 (0.5169) time: 0.4772 data: 0.0002 max mem: 52150 +Epoch: [26] [2200/7269] eta: 0:36:57 lr: 0.000004 loss: 0.4865 (0.5166) time: 0.4768 data: 0.0002 max mem: 52150 +Epoch: [26] [2210/7269] eta: 0:36:56 lr: 0.000004 loss: 0.4865 (0.5167) time: 0.4778 data: 0.0002 max mem: 52150 +Epoch: [26] [2220/7269] eta: 0:36:50 lr: 0.000004 loss: 0.5358 (0.5166) time: 0.4783 data: 0.0002 max mem: 52150 +Epoch: [26] [2230/7269] eta: 0:36:45 lr: 0.000004 loss: 0.5136 (0.5165) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [26] [2240/7269] eta: 0:36:39 lr: 0.000004 loss: 0.5056 (0.5166) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [26] [2250/7269] eta: 0:36:33 lr: 0.000004 loss: 0.4931 (0.5165) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [26] [2260/7269] eta: 0:36:27 lr: 0.000004 loss: 0.4911 (0.5164) time: 0.3712 data: 0.0002 max mem: 52150 +Epoch: [26] [2270/7269] eta: 0:36:21 lr: 0.000004 loss: 0.5175 (0.5164) time: 0.3711 data: 0.0002 max mem: 52150 +Epoch: [26] [2280/7269] eta: 0:36:15 lr: 0.000004 loss: 0.5210 (0.5166) time: 0.3712 data: 0.0002 max mem: 52150 +Epoch: [26] [2290/7269] eta: 0:36:10 lr: 0.000004 loss: 0.4954 (0.5165) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [26] [2300/7269] eta: 0:36:11 lr: 0.000004 loss: 0.5066 (0.5164) time: 0.5462 data: 0.0002 max mem: 52150 +Epoch: [26] [2310/7269] eta: 0:36:06 lr: 0.000004 loss: 0.5112 (0.5164) time: 0.5458 data: 0.0002 max mem: 52150 +Epoch: [26] [2320/7269] eta: 0:36:00 lr: 0.000004 loss: 0.5229 (0.5163) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [26] [2330/7269] eta: 0:35:59 lr: 0.000004 loss: 0.5261 (0.5163) time: 0.4818 data: 0.0002 max mem: 52150 +Epoch: [26] [2340/7269] eta: 0:35:53 lr: 0.000004 loss: 0.5308 (0.5164) time: 0.4819 data: 0.0002 max mem: 52150 +Epoch: [26] [2350/7269] eta: 0:35:47 lr: 0.000004 loss: 0.5308 (0.5164) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [26] [2360/7269] eta: 0:35:44 lr: 0.000004 loss: 0.4980 (0.5163) time: 0.4184 data: 0.0002 max mem: 52150 +Epoch: [26] [2370/7269] eta: 0:35:38 lr: 0.000004 loss: 0.5069 (0.5163) time: 0.4188 data: 0.0002 max mem: 52150 +Epoch: [26] [2380/7269] eta: 0:35:32 lr: 0.000004 loss: 0.5085 (0.5162) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [26] [2390/7269] eta: 0:35:28 lr: 0.000004 loss: 0.5205 (0.5163) time: 0.4203 data: 0.0002 max mem: 52150 +Epoch: [26] [2400/7269] eta: 0:35:25 lr: 0.000004 loss: 0.5405 (0.5162) time: 0.4801 data: 0.0002 max mem: 52150 +Epoch: [26] [2410/7269] eta: 0:35:24 lr: 0.000004 loss: 0.4923 (0.5162) time: 0.5382 data: 0.0002 max mem: 52150 +Epoch: [26] [2420/7269] eta: 0:35:18 lr: 0.000004 loss: 0.5290 (0.5163) time: 0.4783 data: 0.0002 max mem: 52150 +Epoch: [26] [2430/7269] eta: 0:35:15 lr: 0.000004 loss: 0.5362 (0.5163) time: 0.4325 data: 0.0002 max mem: 52150 +Epoch: [26] [2440/7269] eta: 0:35:09 lr: 0.000004 loss: 0.5158 (0.5161) time: 0.4323 data: 0.0002 max mem: 52150 +Epoch: [26] [2450/7269] eta: 0:35:04 lr: 0.000004 loss: 0.4888 (0.5161) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [26] [2460/7269] eta: 0:34:58 lr: 0.000004 loss: 0.5375 (0.5163) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [26] [2470/7269] eta: 0:34:52 lr: 0.000004 loss: 0.5375 (0.5163) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [26] [2480/7269] eta: 0:34:47 lr: 0.000004 loss: 0.5305 (0.5163) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [26] [2490/7269] eta: 0:34:43 lr: 0.000004 loss: 0.5382 (0.5163) time: 0.4180 data: 0.0002 max mem: 52150 +Epoch: [26] [2500/7269] eta: 0:34:37 lr: 0.000004 loss: 0.5425 (0.5164) time: 0.4173 data: 0.0002 max mem: 52150 +Epoch: [26] [2510/7269] eta: 0:34:34 lr: 0.000004 loss: 0.5307 (0.5164) time: 0.4184 data: 0.0002 max mem: 52150 +Epoch: [26] [2520/7269] eta: 0:34:32 lr: 0.000004 loss: 0.5307 (0.5164) time: 0.5206 data: 0.0002 max mem: 52150 +Epoch: [26] [2530/7269] eta: 0:34:26 lr: 0.000004 loss: 0.4884 (0.5163) time: 0.4743 data: 0.0002 max mem: 52150 +Epoch: [26] [2540/7269] eta: 0:34:22 lr: 0.000004 loss: 0.5178 (0.5163) time: 0.4190 data: 0.0002 max mem: 52150 +Epoch: [26] [2550/7269] eta: 0:34:19 lr: 0.000004 loss: 0.5360 (0.5164) time: 0.4691 data: 0.0002 max mem: 52150 +Epoch: [26] [2560/7269] eta: 0:34:15 lr: 0.000004 loss: 0.5324 (0.5164) time: 0.4679 data: 0.0002 max mem: 52150 +Epoch: [26] [2570/7269] eta: 0:34:12 lr: 0.000004 loss: 0.5315 (0.5163) time: 0.4777 data: 0.0002 max mem: 52150 +Epoch: [26] [2580/7269] eta: 0:34:06 lr: 0.000004 loss: 0.5110 (0.5163) time: 0.4320 data: 0.0002 max mem: 52150 +Epoch: [26] [2590/7269] eta: 0:34:06 lr: 0.000004 loss: 0.5653 (0.5166) time: 0.5294 data: 0.0002 max mem: 52150 +Epoch: [26] [2600/7269] eta: 0:34:01 lr: 0.000004 loss: 0.5574 (0.5166) time: 0.5295 data: 0.0002 max mem: 52150 +Epoch: [26] [2610/7269] eta: 0:33:57 lr: 0.000004 loss: 0.4926 (0.5165) time: 0.4314 data: 0.0002 max mem: 52150 +Epoch: [26] [2620/7269] eta: 0:33:54 lr: 0.000004 loss: 0.4900 (0.5165) time: 0.4915 data: 0.0002 max mem: 52150 +Epoch: [26] [2630/7269] eta: 0:33:50 lr: 0.000004 loss: 0.5363 (0.5166) time: 0.4806 data: 0.0002 max mem: 52150 +Epoch: [26] [2640/7269] eta: 0:33:50 lr: 0.000004 loss: 0.5642 (0.5167) time: 0.5896 data: 0.0002 max mem: 52150 +Epoch: [26] [2650/7269] eta: 0:33:45 lr: 0.000004 loss: 0.5685 (0.5168) time: 0.5409 data: 0.0002 max mem: 52150 +Epoch: [26] [2660/7269] eta: 0:33:41 lr: 0.000004 loss: 0.5650 (0.5170) time: 0.4172 data: 0.0002 max mem: 52150 +Epoch: [26] [2670/7269] eta: 0:33:35 lr: 0.000004 loss: 0.5352 (0.5169) time: 0.4169 data: 0.0002 max mem: 52150 +Epoch: [26] [2680/7269] eta: 0:33:30 lr: 0.000004 loss: 0.5247 (0.5169) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [26] [2690/7269] eta: 0:33:24 lr: 0.000004 loss: 0.5272 (0.5170) time: 0.3712 data: 0.0002 max mem: 52150 +Epoch: [26] [2700/7269] eta: 0:33:19 lr: 0.000004 loss: 0.5287 (0.5169) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [26] [2710/7269] eta: 0:33:13 lr: 0.000004 loss: 0.4915 (0.5167) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [26] [2720/7269] eta: 0:33:08 lr: 0.000004 loss: 0.5163 (0.5168) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [26] [2730/7269] eta: 0:33:04 lr: 0.000004 loss: 0.5278 (0.5168) time: 0.4179 data: 0.0002 max mem: 52150 +Epoch: [26] [2740/7269] eta: 0:32:58 lr: 0.000004 loss: 0.5311 (0.5169) time: 0.4183 data: 0.0002 max mem: 52150 +Epoch: [26] [2750/7269] eta: 0:32:53 lr: 0.000004 loss: 0.5611 (0.5169) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [26] [2760/7269] eta: 0:32:48 lr: 0.000004 loss: 0.5198 (0.5169) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [26] [2770/7269] eta: 0:32:44 lr: 0.000004 loss: 0.5212 (0.5169) time: 0.4305 data: 0.0002 max mem: 52150 +Epoch: [26] [2780/7269] eta: 0:32:39 lr: 0.000004 loss: 0.5072 (0.5168) time: 0.4309 data: 0.0002 max mem: 52150 +Epoch: [26] [2790/7269] eta: 0:32:33 lr: 0.000004 loss: 0.4952 (0.5167) time: 0.3728 data: 0.0002 max mem: 52150 +Epoch: [26] [2800/7269] eta: 0:32:30 lr: 0.000004 loss: 0.5203 (0.5169) time: 0.4335 data: 0.0002 max mem: 52150 +Epoch: [26] [2810/7269] eta: 0:32:24 lr: 0.000004 loss: 0.5486 (0.5169) time: 0.4328 data: 0.0002 max mem: 52150 +Epoch: [26] [2820/7269] eta: 0:32:19 lr: 0.000004 loss: 0.5230 (0.5169) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [26] [2830/7269] eta: 0:32:14 lr: 0.000004 loss: 0.5057 (0.5167) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [26] [2840/7269] eta: 0:32:08 lr: 0.000004 loss: 0.5023 (0.5167) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [26] [2850/7269] eta: 0:32:03 lr: 0.000004 loss: 0.5220 (0.5167) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [26] [2860/7269] eta: 0:31:59 lr: 0.000004 loss: 0.5199 (0.5165) time: 0.4174 data: 0.0002 max mem: 52150 +Epoch: [26] [2870/7269] eta: 0:31:56 lr: 0.000004 loss: 0.5384 (0.5167) time: 0.4788 data: 0.0002 max mem: 52150 +Epoch: [26] [2880/7269] eta: 0:31:50 lr: 0.000004 loss: 0.5501 (0.5167) time: 0.4334 data: 0.0002 max mem: 52150 +Epoch: [26] [2890/7269] eta: 0:31:45 lr: 0.000004 loss: 0.5564 (0.5168) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [26] [2900/7269] eta: 0:31:40 lr: 0.000004 loss: 0.5579 (0.5170) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [26] [2910/7269] eta: 0:31:36 lr: 0.000004 loss: 0.5284 (0.5169) time: 0.4326 data: 0.0002 max mem: 52150 +Epoch: [26] [2920/7269] eta: 0:31:34 lr: 0.000004 loss: 0.5356 (0.5170) time: 0.5448 data: 0.0002 max mem: 52150 +Epoch: [26] [2930/7269] eta: 0:31:30 lr: 0.000004 loss: 0.5345 (0.5169) time: 0.5292 data: 0.0002 max mem: 52150 +Epoch: [26] [2940/7269] eta: 0:31:25 lr: 0.000004 loss: 0.5345 (0.5169) time: 0.4168 data: 0.0002 max mem: 52150 +Epoch: [26] [2950/7269] eta: 0:31:20 lr: 0.000004 loss: 0.5419 (0.5170) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [26] [2960/7269] eta: 0:31:16 lr: 0.000004 loss: 0.5042 (0.5168) time: 0.4185 data: 0.0003 max mem: 52150 +Epoch: [26] [2970/7269] eta: 0:31:12 lr: 0.000004 loss: 0.5243 (0.5170) time: 0.4649 data: 0.0003 max mem: 52150 +Epoch: [26] [2980/7269] eta: 0:31:07 lr: 0.000004 loss: 0.5467 (0.5170) time: 0.4179 data: 0.0002 max mem: 52150 +Epoch: [26] [2990/7269] eta: 0:31:05 lr: 0.000004 loss: 0.5147 (0.5170) time: 0.4912 data: 0.0002 max mem: 52150 +Epoch: [26] [3000/7269] eta: 0:30:59 lr: 0.000004 loss: 0.4913 (0.5169) time: 0.4911 data: 0.0002 max mem: 52150 +Epoch: [26] [3010/7269] eta: 0:30:56 lr: 0.000004 loss: 0.5023 (0.5170) time: 0.4314 data: 0.0002 max mem: 52150 +Epoch: [26] [3020/7269] eta: 0:30:52 lr: 0.000004 loss: 0.5163 (0.5169) time: 0.4787 data: 0.0002 max mem: 52150 +Epoch: [26] [3030/7269] eta: 0:30:48 lr: 0.000004 loss: 0.5043 (0.5168) time: 0.4657 data: 0.0002 max mem: 52150 +Epoch: [26] [3040/7269] eta: 0:30:44 lr: 0.000004 loss: 0.4778 (0.5168) time: 0.4665 data: 0.0002 max mem: 52150 +Epoch: [26] [3050/7269] eta: 0:30:39 lr: 0.000004 loss: 0.5027 (0.5167) time: 0.4208 data: 0.0002 max mem: 52150 +Epoch: [26] [3060/7269] eta: 0:30:35 lr: 0.000004 loss: 0.5329 (0.5168) time: 0.4218 data: 0.0002 max mem: 52150 +Epoch: [26] [3070/7269] eta: 0:30:30 lr: 0.000004 loss: 0.5329 (0.5167) time: 0.4206 data: 0.0002 max mem: 52150 +Epoch: [26] [3080/7269] eta: 0:30:26 lr: 0.000004 loss: 0.5042 (0.5167) time: 0.4318 data: 0.0002 max mem: 52150 +Epoch: [26] [3090/7269] eta: 0:30:21 lr: 0.000004 loss: 0.5097 (0.5167) time: 0.4319 data: 0.0002 max mem: 52150 +Epoch: [26] [3100/7269] eta: 0:30:19 lr: 0.000004 loss: 0.4888 (0.5165) time: 0.4775 data: 0.0002 max mem: 52150 +Epoch: [26] [3110/7269] eta: 0:30:13 lr: 0.000004 loss: 0.4982 (0.5165) time: 0.4780 data: 0.0002 max mem: 52150 +Epoch: [26] [3120/7269] eta: 0:30:11 lr: 0.000004 loss: 0.5386 (0.5166) time: 0.4771 data: 0.0002 max mem: 52150 +Epoch: [26] [3130/7269] eta: 0:30:07 lr: 0.000004 loss: 0.5506 (0.5167) time: 0.5222 data: 0.0002 max mem: 52150 +Epoch: [26] [3140/7269] eta: 0:30:02 lr: 0.000004 loss: 0.5374 (0.5166) time: 0.4169 data: 0.0002 max mem: 52150 +Epoch: [26] [3150/7269] eta: 0:29:59 lr: 0.000004 loss: 0.4972 (0.5166) time: 0.4616 data: 0.0002 max mem: 52150 +Epoch: [26] [3160/7269] eta: 0:29:55 lr: 0.000004 loss: 0.5290 (0.5167) time: 0.5221 data: 0.0002 max mem: 52150 +Epoch: [26] [3170/7269] eta: 0:29:50 lr: 0.000004 loss: 0.5339 (0.5167) time: 0.4322 data: 0.0002 max mem: 52150 +Epoch: [26] [3180/7269] eta: 0:29:45 lr: 0.000004 loss: 0.5500 (0.5168) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [26] [3190/7269] eta: 0:29:41 lr: 0.000004 loss: 0.5251 (0.5168) time: 0.4346 data: 0.0002 max mem: 52150 +Epoch: [26] [3200/7269] eta: 0:29:38 lr: 0.000004 loss: 0.5217 (0.5168) time: 0.5264 data: 0.0002 max mem: 52150 +Epoch: [26] [3210/7269] eta: 0:29:33 lr: 0.000004 loss: 0.5456 (0.5168) time: 0.4645 data: 0.0002 max mem: 52150 +Epoch: [26] [3220/7269] eta: 0:29:28 lr: 0.000004 loss: 0.4999 (0.5166) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [26] [3230/7269] eta: 0:29:23 lr: 0.000004 loss: 0.4999 (0.5167) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [26] [3240/7269] eta: 0:29:18 lr: 0.000004 loss: 0.5561 (0.5167) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [26] [3250/7269] eta: 0:29:12 lr: 0.000004 loss: 0.5770 (0.5169) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [26] [3260/7269] eta: 0:29:08 lr: 0.000004 loss: 0.5545 (0.5169) time: 0.4199 data: 0.0002 max mem: 52150 +Epoch: [26] [3270/7269] eta: 0:29:03 lr: 0.000004 loss: 0.5188 (0.5169) time: 0.4196 data: 0.0002 max mem: 52150 +Epoch: [26] [3280/7269] eta: 0:28:58 lr: 0.000004 loss: 0.5294 (0.5170) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [26] [3290/7269] eta: 0:28:53 lr: 0.000004 loss: 0.5349 (0.5170) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [26] [3300/7269] eta: 0:28:48 lr: 0.000004 loss: 0.5106 (0.5169) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [26] [3310/7269] eta: 0:28:43 lr: 0.000004 loss: 0.5291 (0.5170) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [26] [3320/7269] eta: 0:28:38 lr: 0.000004 loss: 0.5426 (0.5171) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [26] [3330/7269] eta: 0:28:34 lr: 0.000004 loss: 0.5426 (0.5171) time: 0.4188 data: 0.0002 max mem: 52150 +Epoch: [26] [3340/7269] eta: 0:28:29 lr: 0.000004 loss: 0.5105 (0.5170) time: 0.4187 data: 0.0002 max mem: 52150 +Epoch: [26] [3350/7269] eta: 0:28:25 lr: 0.000004 loss: 0.5286 (0.5170) time: 0.4323 data: 0.0002 max mem: 52150 +Epoch: [26] [3360/7269] eta: 0:28:20 lr: 0.000004 loss: 0.5210 (0.5170) time: 0.4324 data: 0.0002 max mem: 52150 +Epoch: [26] [3370/7269] eta: 0:28:16 lr: 0.000004 loss: 0.4903 (0.5169) time: 0.4344 data: 0.0002 max mem: 52150 +Epoch: [26] [3380/7269] eta: 0:28:11 lr: 0.000004 loss: 0.5053 (0.5169) time: 0.4341 data: 0.0002 max mem: 52150 +Epoch: [26] [3390/7269] eta: 0:28:06 lr: 0.000004 loss: 0.5635 (0.5170) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [26] [3400/7269] eta: 0:28:01 lr: 0.000004 loss: 0.5283 (0.5170) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [26] [3410/7269] eta: 0:27:56 lr: 0.000004 loss: 0.5111 (0.5169) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [26] [3420/7269] eta: 0:27:51 lr: 0.000004 loss: 0.5126 (0.5169) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [26] [3430/7269] eta: 0:27:47 lr: 0.000004 loss: 0.5126 (0.5168) time: 0.4366 data: 0.0002 max mem: 52150 +Epoch: [26] [3440/7269] eta: 0:27:42 lr: 0.000004 loss: 0.5338 (0.5168) time: 0.4365 data: 0.0002 max mem: 52150 +Epoch: [26] [3450/7269] eta: 0:27:37 lr: 0.000004 loss: 0.5291 (0.5168) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [26] [3460/7269] eta: 0:27:36 lr: 0.000004 loss: 0.5194 (0.5168) time: 0.5407 data: 0.0002 max mem: 52150 +Epoch: [26] [3470/7269] eta: 0:27:31 lr: 0.000004 loss: 0.5003 (0.5167) time: 0.5410 data: 0.0002 max mem: 52150 +Epoch: [26] [3480/7269] eta: 0:27:28 lr: 0.000004 loss: 0.4711 (0.5166) time: 0.4844 data: 0.0002 max mem: 52150 +Epoch: [26] [3490/7269] eta: 0:27:23 lr: 0.000004 loss: 0.5214 (0.5166) time: 0.4844 data: 0.0002 max mem: 52150 +Epoch: [26] [3500/7269] eta: 0:27:19 lr: 0.000004 loss: 0.5321 (0.5167) time: 0.4323 data: 0.0002 max mem: 52150 +Epoch: [26] [3510/7269] eta: 0:27:14 lr: 0.000004 loss: 0.5276 (0.5167) time: 0.4320 data: 0.0002 max mem: 52150 +Epoch: [26] [3520/7269] eta: 0:27:11 lr: 0.000004 loss: 0.5243 (0.5167) time: 0.4379 data: 0.0002 max mem: 52150 +Epoch: [26] [3530/7269] eta: 0:27:06 lr: 0.000004 loss: 0.5320 (0.5168) time: 0.4389 data: 0.0002 max mem: 52150 +Epoch: [26] [3540/7269] eta: 0:27:01 lr: 0.000004 loss: 0.5251 (0.5168) time: 0.3729 data: 0.0002 max mem: 52150 +Epoch: [26] [3550/7269] eta: 0:26:57 lr: 0.000004 loss: 0.5139 (0.5167) time: 0.4181 data: 0.0002 max mem: 52150 +Epoch: [26] [3560/7269] eta: 0:26:53 lr: 0.000004 loss: 0.4880 (0.5167) time: 0.4655 data: 0.0002 max mem: 52150 +Epoch: [26] [3570/7269] eta: 0:26:49 lr: 0.000004 loss: 0.5120 (0.5167) time: 0.4696 data: 0.0002 max mem: 52150 +Epoch: [26] [3580/7269] eta: 0:26:45 lr: 0.000004 loss: 0.5120 (0.5166) time: 0.4680 data: 0.0002 max mem: 52150 +Epoch: [26] [3590/7269] eta: 0:26:40 lr: 0.000004 loss: 0.4887 (0.5166) time: 0.4178 data: 0.0002 max mem: 52150 +Epoch: [26] [3600/7269] eta: 0:26:36 lr: 0.000004 loss: 0.5059 (0.5166) time: 0.4200 data: 0.0002 max mem: 52150 +Epoch: [26] [3610/7269] eta: 0:26:34 lr: 0.000004 loss: 0.5229 (0.5166) time: 0.5865 data: 0.0002 max mem: 52150 +Epoch: [26] [3620/7269] eta: 0:26:30 lr: 0.000004 loss: 0.5615 (0.5168) time: 0.5853 data: 0.0002 max mem: 52150 +Epoch: [26] [3630/7269] eta: 0:26:25 lr: 0.000004 loss: 0.5547 (0.5167) time: 0.4194 data: 0.0002 max mem: 52150 +Epoch: [26] [3640/7269] eta: 0:26:20 lr: 0.000004 loss: 0.5207 (0.5168) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [26] [3650/7269] eta: 0:26:15 lr: 0.000004 loss: 0.5584 (0.5169) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [26] [3660/7269] eta: 0:26:13 lr: 0.000004 loss: 0.5579 (0.5170) time: 0.5413 data: 0.0002 max mem: 52150 +Epoch: [26] [3670/7269] eta: 0:26:08 lr: 0.000004 loss: 0.5557 (0.5170) time: 0.5416 data: 0.0002 max mem: 52150 +Epoch: [26] [3680/7269] eta: 0:26:04 lr: 0.000004 loss: 0.4815 (0.5169) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [26] [3690/7269] eta: 0:25:59 lr: 0.000004 loss: 0.4814 (0.5169) time: 0.4202 data: 0.0002 max mem: 52150 +Epoch: [26] [3700/7269] eta: 0:25:54 lr: 0.000004 loss: 0.5048 (0.5169) time: 0.4206 data: 0.0002 max mem: 52150 +Epoch: [26] [3710/7269] eta: 0:25:51 lr: 0.000004 loss: 0.5579 (0.5170) time: 0.4324 data: 0.0002 max mem: 52150 +Epoch: [26] [3720/7269] eta: 0:25:46 lr: 0.000004 loss: 0.5432 (0.5171) time: 0.4323 data: 0.0002 max mem: 52150 +Epoch: [26] [3730/7269] eta: 0:25:43 lr: 0.000004 loss: 0.5139 (0.5170) time: 0.4655 data: 0.0002 max mem: 52150 +Epoch: [26] [3740/7269] eta: 0:25:38 lr: 0.000004 loss: 0.5139 (0.5170) time: 0.4654 data: 0.0002 max mem: 52150 +Epoch: [26] [3750/7269] eta: 0:25:33 lr: 0.000004 loss: 0.5357 (0.5170) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [26] [3760/7269] eta: 0:25:28 lr: 0.000004 loss: 0.5088 (0.5169) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [26] [3770/7269] eta: 0:25:23 lr: 0.000004 loss: 0.4908 (0.5169) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [26] [3780/7269] eta: 0:25:18 lr: 0.000004 loss: 0.5038 (0.5169) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [26] [3790/7269] eta: 0:25:13 lr: 0.000004 loss: 0.5264 (0.5169) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [26] [3800/7269] eta: 0:25:11 lr: 0.000004 loss: 0.5644 (0.5170) time: 0.5378 data: 0.0002 max mem: 52150 +Epoch: [26] [3810/7269] eta: 0:25:06 lr: 0.000004 loss: 0.5245 (0.5170) time: 0.5378 data: 0.0002 max mem: 52150 +Epoch: [26] [3820/7269] eta: 0:25:02 lr: 0.000004 loss: 0.5203 (0.5171) time: 0.4166 data: 0.0002 max mem: 52150 +Epoch: [26] [3830/7269] eta: 0:24:57 lr: 0.000004 loss: 0.5481 (0.5171) time: 0.4162 data: 0.0002 max mem: 52150 +Epoch: [26] [3840/7269] eta: 0:24:52 lr: 0.000004 loss: 0.5481 (0.5171) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [26] [3850/7269] eta: 0:24:47 lr: 0.000004 loss: 0.5360 (0.5171) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [26] [3860/7269] eta: 0:24:42 lr: 0.000004 loss: 0.5370 (0.5172) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [26] [3870/7269] eta: 0:24:37 lr: 0.000004 loss: 0.5442 (0.5172) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [26] [3880/7269] eta: 0:24:32 lr: 0.000004 loss: 0.5388 (0.5172) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [26] [3890/7269] eta: 0:24:28 lr: 0.000004 loss: 0.5130 (0.5172) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [26] [3900/7269] eta: 0:24:23 lr: 0.000004 loss: 0.5130 (0.5172) time: 0.4172 data: 0.0002 max mem: 52150 +Epoch: [26] [3910/7269] eta: 0:24:19 lr: 0.000004 loss: 0.4553 (0.5172) time: 0.4171 data: 0.0002 max mem: 52150 +Epoch: [26] [3920/7269] eta: 0:24:14 lr: 0.000004 loss: 0.5429 (0.5172) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [26] [3930/7269] eta: 0:24:09 lr: 0.000004 loss: 0.5538 (0.5173) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [26] [3940/7269] eta: 0:24:04 lr: 0.000004 loss: 0.5326 (0.5173) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [26] [3950/7269] eta: 0:24:00 lr: 0.000004 loss: 0.5003 (0.5172) time: 0.4344 data: 0.0002 max mem: 52150 +Epoch: [26] [3960/7269] eta: 0:23:57 lr: 0.000004 loss: 0.4723 (0.5171) time: 0.5562 data: 0.0002 max mem: 52150 +Epoch: [26] [3970/7269] eta: 0:23:52 lr: 0.000004 loss: 0.4945 (0.5171) time: 0.4936 data: 0.0002 max mem: 52150 +Epoch: [26] [3980/7269] eta: 0:23:48 lr: 0.000004 loss: 0.5358 (0.5172) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [26] [3990/7269] eta: 0:23:43 lr: 0.000004 loss: 0.5182 (0.5171) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [26] [4000/7269] eta: 0:23:38 lr: 0.000004 loss: 0.4915 (0.5171) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [26] [4010/7269] eta: 0:23:34 lr: 0.000004 loss: 0.5118 (0.5171) time: 0.4356 data: 0.0002 max mem: 52150 +Epoch: [26] [4020/7269] eta: 0:23:30 lr: 0.000004 loss: 0.4927 (0.5170) time: 0.4841 data: 0.0002 max mem: 52150 +Epoch: [26] [4030/7269] eta: 0:23:25 lr: 0.000004 loss: 0.4998 (0.5170) time: 0.4208 data: 0.0002 max mem: 52150 +Epoch: [26] [4040/7269] eta: 0:23:21 lr: 0.000004 loss: 0.4998 (0.5170) time: 0.4222 data: 0.0002 max mem: 52150 +Epoch: [26] [4050/7269] eta: 0:23:16 lr: 0.000004 loss: 0.5258 (0.5170) time: 0.4213 data: 0.0002 max mem: 52150 +Epoch: [26] [4060/7269] eta: 0:23:11 lr: 0.000004 loss: 0.5742 (0.5171) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [26] [4070/7269] eta: 0:23:09 lr: 0.000004 loss: 0.5552 (0.5171) time: 0.5285 data: 0.0002 max mem: 52150 +Epoch: [26] [4080/7269] eta: 0:23:06 lr: 0.000004 loss: 0.5272 (0.5171) time: 0.6191 data: 0.0002 max mem: 52150 +Epoch: [26] [4090/7269] eta: 0:23:01 lr: 0.000004 loss: 0.5448 (0.5171) time: 0.4623 data: 0.0002 max mem: 52150 +Epoch: [26] [4100/7269] eta: 0:22:58 lr: 0.000004 loss: 0.5087 (0.5171) time: 0.4868 data: 0.0002 max mem: 52150 +Epoch: [26] [4110/7269] eta: 0:22:54 lr: 0.000004 loss: 0.4872 (0.5171) time: 0.5473 data: 0.0002 max mem: 52150 +Epoch: [26] [4120/7269] eta: 0:22:50 lr: 0.000004 loss: 0.5237 (0.5171) time: 0.4802 data: 0.0002 max mem: 52150 +Epoch: [26] [4130/7269] eta: 0:22:46 lr: 0.000004 loss: 0.5481 (0.5171) time: 0.4798 data: 0.0002 max mem: 52150 +Epoch: [26] [4140/7269] eta: 0:22:41 lr: 0.000004 loss: 0.5511 (0.5172) time: 0.4320 data: 0.0002 max mem: 52150 +Epoch: [26] [4150/7269] eta: 0:22:38 lr: 0.000004 loss: 0.5288 (0.5172) time: 0.4785 data: 0.0002 max mem: 52150 +Epoch: [26] [4160/7269] eta: 0:22:34 lr: 0.000004 loss: 0.4937 (0.5172) time: 0.5268 data: 0.0002 max mem: 52150 +Epoch: [26] [4170/7269] eta: 0:22:30 lr: 0.000004 loss: 0.5324 (0.5172) time: 0.4842 data: 0.0002 max mem: 52150 +Epoch: [26] [4180/7269] eta: 0:22:25 lr: 0.000004 loss: 0.5324 (0.5172) time: 0.4365 data: 0.0002 max mem: 52150 +Epoch: [26] [4190/7269] eta: 0:22:20 lr: 0.000004 loss: 0.4968 (0.5171) time: 0.3727 data: 0.0002 max mem: 52150 +Epoch: [26] [4200/7269] eta: 0:22:15 lr: 0.000004 loss: 0.5135 (0.5172) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [26] [4210/7269] eta: 0:22:11 lr: 0.000004 loss: 0.5214 (0.5172) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [26] [4220/7269] eta: 0:22:06 lr: 0.000004 loss: 0.5367 (0.5172) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [26] [4230/7269] eta: 0:22:01 lr: 0.000004 loss: 0.5222 (0.5172) time: 0.3728 data: 0.0002 max mem: 52150 +Epoch: [26] [4240/7269] eta: 0:21:57 lr: 0.000004 loss: 0.4863 (0.5171) time: 0.4315 data: 0.0002 max mem: 52150 +Epoch: [26] [4250/7269] eta: 0:21:53 lr: 0.000004 loss: 0.5017 (0.5171) time: 0.4930 data: 0.0002 max mem: 52150 +Epoch: [26] [4260/7269] eta: 0:21:49 lr: 0.000004 loss: 0.5202 (0.5171) time: 0.4931 data: 0.0002 max mem: 52150 +Epoch: [26] [4270/7269] eta: 0:21:44 lr: 0.000004 loss: 0.5202 (0.5172) time: 0.4311 data: 0.0002 max mem: 52150 +Epoch: [26] [4280/7269] eta: 0:21:40 lr: 0.000004 loss: 0.5590 (0.5173) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [26] [4290/7269] eta: 0:21:35 lr: 0.000004 loss: 0.5234 (0.5172) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [26] [4300/7269] eta: 0:21:30 lr: 0.000004 loss: 0.5062 (0.5172) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [26] [4310/7269] eta: 0:21:25 lr: 0.000004 loss: 0.5357 (0.5173) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [26] [4320/7269] eta: 0:21:20 lr: 0.000004 loss: 0.5369 (0.5173) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [26] [4330/7269] eta: 0:21:16 lr: 0.000004 loss: 0.5257 (0.5173) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [26] [4340/7269] eta: 0:21:11 lr: 0.000004 loss: 0.5257 (0.5173) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [26] [4350/7269] eta: 0:21:06 lr: 0.000004 loss: 0.5221 (0.5173) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [26] [4360/7269] eta: 0:21:02 lr: 0.000004 loss: 0.4918 (0.5173) time: 0.4180 data: 0.0002 max mem: 52150 +Epoch: [26] [4370/7269] eta: 0:20:57 lr: 0.000004 loss: 0.5549 (0.5174) time: 0.4187 data: 0.0002 max mem: 52150 +Epoch: [26] [4380/7269] eta: 0:20:53 lr: 0.000004 loss: 0.5609 (0.5174) time: 0.3729 data: 0.0002 max mem: 52150 +Epoch: [26] [4390/7269] eta: 0:20:48 lr: 0.000004 loss: 0.5200 (0.5174) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [26] [4400/7269] eta: 0:20:44 lr: 0.000004 loss: 0.5069 (0.5173) time: 0.4349 data: 0.0002 max mem: 52150 +Epoch: [26] [4410/7269] eta: 0:20:40 lr: 0.000004 loss: 0.5126 (0.5174) time: 0.4860 data: 0.0002 max mem: 52150 +Epoch: [26] [4420/7269] eta: 0:20:36 lr: 0.000004 loss: 0.5445 (0.5173) time: 0.4820 data: 0.0002 max mem: 52150 +Epoch: [26] [4430/7269] eta: 0:20:31 lr: 0.000004 loss: 0.5083 (0.5173) time: 0.4311 data: 0.0002 max mem: 52150 +Epoch: [26] [4440/7269] eta: 0:20:26 lr: 0.000004 loss: 0.5270 (0.5173) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [26] [4450/7269] eta: 0:20:22 lr: 0.000004 loss: 0.5404 (0.5173) time: 0.3728 data: 0.0002 max mem: 52150 +Epoch: [26] [4460/7269] eta: 0:20:18 lr: 0.000004 loss: 0.5614 (0.5174) time: 0.4342 data: 0.0002 max mem: 52150 +Epoch: [26] [4470/7269] eta: 0:20:13 lr: 0.000004 loss: 0.5614 (0.5174) time: 0.4332 data: 0.0002 max mem: 52150 +Epoch: [26] [4480/7269] eta: 0:20:09 lr: 0.000004 loss: 0.4810 (0.5173) time: 0.4171 data: 0.0002 max mem: 52150 +Epoch: [26] [4490/7269] eta: 0:20:04 lr: 0.000004 loss: 0.4806 (0.5172) time: 0.4184 data: 0.0002 max mem: 52150 +Epoch: [26] [4500/7269] eta: 0:19:59 lr: 0.000004 loss: 0.4880 (0.5172) time: 0.3729 data: 0.0002 max mem: 52150 +Epoch: [26] [4510/7269] eta: 0:19:55 lr: 0.000004 loss: 0.5345 (0.5172) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [26] [4520/7269] eta: 0:19:50 lr: 0.000004 loss: 0.5114 (0.5173) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [26] [4530/7269] eta: 0:19:46 lr: 0.000004 loss: 0.5329 (0.5173) time: 0.4177 data: 0.0002 max mem: 52150 +Epoch: [26] [4540/7269] eta: 0:19:41 lr: 0.000004 loss: 0.5479 (0.5174) time: 0.4175 data: 0.0002 max mem: 52150 +Epoch: [26] [4550/7269] eta: 0:19:37 lr: 0.000004 loss: 0.5628 (0.5175) time: 0.4313 data: 0.0002 max mem: 52150 +Epoch: [26] [4560/7269] eta: 0:19:32 lr: 0.000004 loss: 0.5289 (0.5174) time: 0.4312 data: 0.0002 max mem: 52150 +Epoch: [26] [4570/7269] eta: 0:19:28 lr: 0.000004 loss: 0.4973 (0.5173) time: 0.4177 data: 0.0002 max mem: 52150 +Epoch: [26] [4580/7269] eta: 0:19:24 lr: 0.000004 loss: 0.4974 (0.5173) time: 0.4183 data: 0.0002 max mem: 52150 +Epoch: [26] [4590/7269] eta: 0:19:19 lr: 0.000004 loss: 0.5139 (0.5173) time: 0.3727 data: 0.0002 max mem: 52150 +Epoch: [26] [4600/7269] eta: 0:19:14 lr: 0.000004 loss: 0.5139 (0.5173) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [26] [4610/7269] eta: 0:19:10 lr: 0.000004 loss: 0.5289 (0.5173) time: 0.4322 data: 0.0002 max mem: 52150 +Epoch: [26] [4620/7269] eta: 0:19:08 lr: 0.000004 loss: 0.5539 (0.5174) time: 0.6500 data: 0.0002 max mem: 52150 +Epoch: [26] [4630/7269] eta: 0:19:04 lr: 0.000004 loss: 0.5518 (0.5174) time: 0.6536 data: 0.0002 max mem: 52150 +Epoch: [26] [4640/7269] eta: 0:19:01 lr: 0.000004 loss: 0.5466 (0.5174) time: 0.5320 data: 0.0002 max mem: 52150 +Epoch: [26] [4650/7269] eta: 0:18:57 lr: 0.000004 loss: 0.5333 (0.5174) time: 0.5824 data: 0.0002 max mem: 52150 +Epoch: [26] [4660/7269] eta: 0:18:52 lr: 0.000004 loss: 0.5278 (0.5174) time: 0.4866 data: 0.0002 max mem: 52150 +Epoch: [26] [4670/7269] eta: 0:18:48 lr: 0.000004 loss: 0.4802 (0.5174) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [26] [4680/7269] eta: 0:18:44 lr: 0.000004 loss: 0.5129 (0.5174) time: 0.4826 data: 0.0002 max mem: 52150 +Epoch: [26] [4690/7269] eta: 0:18:40 lr: 0.000004 loss: 0.5253 (0.5174) time: 0.5324 data: 0.0002 max mem: 52150 +Epoch: [26] [4700/7269] eta: 0:18:37 lr: 0.000004 loss: 0.5387 (0.5174) time: 0.5181 data: 0.0002 max mem: 52150 +Epoch: [26] [4710/7269] eta: 0:18:33 lr: 0.000004 loss: 0.5387 (0.5174) time: 0.5313 data: 0.0002 max mem: 52150 +Epoch: [26] [4720/7269] eta: 0:18:28 lr: 0.000004 loss: 0.5617 (0.5176) time: 0.4347 data: 0.0002 max mem: 52150 +Epoch: [26] [4730/7269] eta: 0:18:24 lr: 0.000004 loss: 0.5701 (0.5176) time: 0.4954 data: 0.0002 max mem: 52150 +Epoch: [26] [4740/7269] eta: 0:18:20 lr: 0.000004 loss: 0.5237 (0.5176) time: 0.5552 data: 0.0002 max mem: 52150 +Epoch: [26] [4750/7269] eta: 0:18:16 lr: 0.000004 loss: 0.5164 (0.5176) time: 0.4316 data: 0.0002 max mem: 52150 +Epoch: [26] [4760/7269] eta: 0:18:12 lr: 0.000004 loss: 0.5534 (0.5177) time: 0.4183 data: 0.0002 max mem: 52150 +Epoch: [26] [4770/7269] eta: 0:18:07 lr: 0.000004 loss: 0.5459 (0.5177) time: 0.4187 data: 0.0002 max mem: 52150 +Epoch: [26] [4780/7269] eta: 0:18:02 lr: 0.000004 loss: 0.5057 (0.5177) time: 0.3731 data: 0.0002 max mem: 52150 +Epoch: [26] [4790/7269] eta: 0:17:57 lr: 0.000004 loss: 0.5095 (0.5176) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [26] [4800/7269] eta: 0:17:53 lr: 0.000004 loss: 0.5116 (0.5177) time: 0.4178 data: 0.0002 max mem: 52150 +Epoch: [26] [4810/7269] eta: 0:17:49 lr: 0.000004 loss: 0.5156 (0.5176) time: 0.4172 data: 0.0002 max mem: 52150 +Epoch: [26] [4820/7269] eta: 0:17:44 lr: 0.000004 loss: 0.5020 (0.5176) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [26] [4830/7269] eta: 0:17:40 lr: 0.000004 loss: 0.5020 (0.5175) time: 0.4321 data: 0.0002 max mem: 52150 +Epoch: [26] [4840/7269] eta: 0:17:35 lr: 0.000004 loss: 0.4874 (0.5175) time: 0.4323 data: 0.0002 max mem: 52150 +Epoch: [26] [4850/7269] eta: 0:17:31 lr: 0.000004 loss: 0.5412 (0.5176) time: 0.3729 data: 0.0002 max mem: 52150 +Epoch: [26] [4860/7269] eta: 0:17:26 lr: 0.000004 loss: 0.5570 (0.5176) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [26] [4870/7269] eta: 0:17:22 lr: 0.000004 loss: 0.5537 (0.5176) time: 0.4192 data: 0.0002 max mem: 52150 +Epoch: [26] [4880/7269] eta: 0:17:17 lr: 0.000004 loss: 0.5433 (0.5176) time: 0.4193 data: 0.0002 max mem: 52150 +Epoch: [26] [4890/7269] eta: 0:17:13 lr: 0.000004 loss: 0.5497 (0.5177) time: 0.4327 data: 0.0002 max mem: 52150 +Epoch: [26] [4900/7269] eta: 0:17:08 lr: 0.000004 loss: 0.5432 (0.5177) time: 0.4324 data: 0.0002 max mem: 52150 +Epoch: [26] [4910/7269] eta: 0:17:04 lr: 0.000004 loss: 0.5155 (0.5177) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [26] [4920/7269] eta: 0:17:00 lr: 0.000004 loss: 0.5076 (0.5177) time: 0.4332 data: 0.0002 max mem: 52150 +Epoch: [26] [4930/7269] eta: 0:16:56 lr: 0.000004 loss: 0.5295 (0.5177) time: 0.4953 data: 0.0002 max mem: 52150 +Epoch: [26] [4940/7269] eta: 0:16:51 lr: 0.000004 loss: 0.5081 (0.5177) time: 0.4342 data: 0.0002 max mem: 52150 +Epoch: [26] [4950/7269] eta: 0:16:47 lr: 0.000004 loss: 0.5180 (0.5177) time: 0.4175 data: 0.0002 max mem: 52150 +Epoch: [26] [4960/7269] eta: 0:16:42 lr: 0.000004 loss: 0.5348 (0.5178) time: 0.4177 data: 0.0002 max mem: 52150 +Epoch: [26] [4970/7269] eta: 0:16:38 lr: 0.000004 loss: 0.5348 (0.5179) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [26] [4980/7269] eta: 0:16:34 lr: 0.000004 loss: 0.5403 (0.5179) time: 0.4778 data: 0.0002 max mem: 52150 +Epoch: [26] [4990/7269] eta: 0:16:29 lr: 0.000004 loss: 0.5350 (0.5179) time: 0.4777 data: 0.0002 max mem: 52150 +Epoch: [26] [5000/7269] eta: 0:16:25 lr: 0.000004 loss: 0.5518 (0.5179) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [26] [5010/7269] eta: 0:16:20 lr: 0.000004 loss: 0.5518 (0.5179) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [26] [5020/7269] eta: 0:16:15 lr: 0.000004 loss: 0.5487 (0.5180) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [26] [5030/7269] eta: 0:16:11 lr: 0.000004 loss: 0.5248 (0.5179) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [26] [5040/7269] eta: 0:16:06 lr: 0.000004 loss: 0.5185 (0.5179) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [26] [5050/7269] eta: 0:16:02 lr: 0.000004 loss: 0.5379 (0.5179) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [26] [5060/7269] eta: 0:15:57 lr: 0.000004 loss: 0.4810 (0.5178) time: 0.4180 data: 0.0002 max mem: 52150 +Epoch: [26] [5070/7269] eta: 0:15:53 lr: 0.000004 loss: 0.4977 (0.5178) time: 0.4179 data: 0.0002 max mem: 52150 +Epoch: [26] [5080/7269] eta: 0:15:48 lr: 0.000004 loss: 0.5291 (0.5179) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [26] [5090/7269] eta: 0:15:44 lr: 0.000004 loss: 0.5291 (0.5178) time: 0.4795 data: 0.0002 max mem: 52150 +Epoch: [26] [5100/7269] eta: 0:15:40 lr: 0.000004 loss: 0.5066 (0.5178) time: 0.4798 data: 0.0002 max mem: 52150 +Epoch: [26] [5110/7269] eta: 0:15:36 lr: 0.000004 loss: 0.4977 (0.5177) time: 0.4211 data: 0.0002 max mem: 52150 +Epoch: [26] [5120/7269] eta: 0:15:32 lr: 0.000004 loss: 0.4977 (0.5177) time: 0.4843 data: 0.0002 max mem: 52150 +Epoch: [26] [5130/7269] eta: 0:15:28 lr: 0.000004 loss: 0.4835 (0.5176) time: 0.4952 data: 0.0002 max mem: 52150 +Epoch: [26] [5140/7269] eta: 0:15:23 lr: 0.000004 loss: 0.5104 (0.5177) time: 0.4318 data: 0.0002 max mem: 52150 +Epoch: [26] [5150/7269] eta: 0:15:19 lr: 0.000004 loss: 0.5399 (0.5177) time: 0.4822 data: 0.0002 max mem: 52150 +Epoch: [26] [5160/7269] eta: 0:15:15 lr: 0.000004 loss: 0.5368 (0.5177) time: 0.4824 data: 0.0002 max mem: 52150 +Epoch: [26] [5170/7269] eta: 0:15:10 lr: 0.000004 loss: 0.5061 (0.5177) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [26] [5180/7269] eta: 0:15:05 lr: 0.000004 loss: 0.5262 (0.5177) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [26] [5190/7269] eta: 0:15:02 lr: 0.000004 loss: 0.5631 (0.5178) time: 0.4962 data: 0.0002 max mem: 52150 +Epoch: [26] [5200/7269] eta: 0:14:58 lr: 0.000004 loss: 0.5257 (0.5178) time: 0.6222 data: 0.0002 max mem: 52150 +Epoch: [26] [5210/7269] eta: 0:14:54 lr: 0.000004 loss: 0.5306 (0.5178) time: 0.4984 data: 0.0002 max mem: 52150 +Epoch: [26] [5220/7269] eta: 0:14:50 lr: 0.000004 loss: 0.5482 (0.5178) time: 0.4853 data: 0.0002 max mem: 52150 +Epoch: [26] [5230/7269] eta: 0:14:45 lr: 0.000004 loss: 0.5295 (0.5178) time: 0.4855 data: 0.0002 max mem: 52150 +Epoch: [26] [5240/7269] eta: 0:14:41 lr: 0.000004 loss: 0.5098 (0.5178) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [26] [5250/7269] eta: 0:14:37 lr: 0.000004 loss: 0.5098 (0.5178) time: 0.4326 data: 0.0002 max mem: 52150 +Epoch: [26] [5260/7269] eta: 0:14:33 lr: 0.000004 loss: 0.5345 (0.5178) time: 0.4951 data: 0.0002 max mem: 52150 +Epoch: [26] [5270/7269] eta: 0:14:28 lr: 0.000004 loss: 0.5305 (0.5178) time: 0.4784 data: 0.0002 max mem: 52150 +Epoch: [26] [5280/7269] eta: 0:14:24 lr: 0.000004 loss: 0.5279 (0.5178) time: 0.4161 data: 0.0002 max mem: 52150 +Epoch: [26] [5290/7269] eta: 0:14:20 lr: 0.000004 loss: 0.4999 (0.5178) time: 0.4196 data: 0.0002 max mem: 52150 +Epoch: [26] [5300/7269] eta: 0:14:15 lr: 0.000004 loss: 0.4896 (0.5177) time: 0.4794 data: 0.0002 max mem: 52150 +Epoch: [26] [5310/7269] eta: 0:14:11 lr: 0.000004 loss: 0.4849 (0.5177) time: 0.4312 data: 0.0002 max mem: 52150 +Epoch: [26] [5320/7269] eta: 0:14:06 lr: 0.000004 loss: 0.5028 (0.5176) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [26] [5330/7269] eta: 0:14:02 lr: 0.000004 loss: 0.5330 (0.5176) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [26] [5340/7269] eta: 0:13:57 lr: 0.000004 loss: 0.5281 (0.5176) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [26] [5350/7269] eta: 0:13:53 lr: 0.000004 loss: 0.4901 (0.5176) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [26] [5360/7269] eta: 0:13:48 lr: 0.000004 loss: 0.4959 (0.5176) time: 0.4186 data: 0.0002 max mem: 52150 +Epoch: [26] [5370/7269] eta: 0:13:44 lr: 0.000004 loss: 0.5381 (0.5176) time: 0.4186 data: 0.0002 max mem: 52150 +Epoch: [26] [5380/7269] eta: 0:13:40 lr: 0.000004 loss: 0.5384 (0.5176) time: 0.5247 data: 0.0002 max mem: 52150 +Epoch: [26] [5390/7269] eta: 0:13:36 lr: 0.000004 loss: 0.5311 (0.5176) time: 0.5738 data: 0.0002 max mem: 52150 +Epoch: [26] [5400/7269] eta: 0:13:31 lr: 0.000004 loss: 0.5176 (0.5176) time: 0.4211 data: 0.0002 max mem: 52150 +Epoch: [26] [5410/7269] eta: 0:13:27 lr: 0.000004 loss: 0.5221 (0.5177) time: 0.4349 data: 0.0002 max mem: 52150 +Epoch: [26] [5420/7269] eta: 0:13:23 lr: 0.000004 loss: 0.4894 (0.5176) time: 0.4350 data: 0.0002 max mem: 52150 +Epoch: [26] [5430/7269] eta: 0:13:18 lr: 0.000004 loss: 0.4950 (0.5176) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [26] [5440/7269] eta: 0:13:14 lr: 0.000004 loss: 0.5232 (0.5176) time: 0.4167 data: 0.0002 max mem: 52150 +Epoch: [26] [5450/7269] eta: 0:13:09 lr: 0.000004 loss: 0.5106 (0.5176) time: 0.4168 data: 0.0002 max mem: 52150 +Epoch: [26] [5460/7269] eta: 0:13:05 lr: 0.000004 loss: 0.5032 (0.5175) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [26] [5470/7269] eta: 0:13:00 lr: 0.000004 loss: 0.5032 (0.5175) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [26] [5480/7269] eta: 0:12:56 lr: 0.000004 loss: 0.5440 (0.5176) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [26] [5490/7269] eta: 0:12:51 lr: 0.000004 loss: 0.5642 (0.5176) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [26] [5500/7269] eta: 0:12:47 lr: 0.000004 loss: 0.5595 (0.5177) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [26] [5510/7269] eta: 0:12:43 lr: 0.000004 loss: 0.5595 (0.5177) time: 0.4820 data: 0.0002 max mem: 52150 +Epoch: [26] [5520/7269] eta: 0:12:39 lr: 0.000004 loss: 0.5197 (0.5177) time: 0.5434 data: 0.0002 max mem: 52150 +Epoch: [26] [5530/7269] eta: 0:12:35 lr: 0.000004 loss: 0.5107 (0.5177) time: 0.4812 data: 0.0002 max mem: 52150 +Epoch: [26] [5540/7269] eta: 0:12:30 lr: 0.000004 loss: 0.5149 (0.5177) time: 0.4196 data: 0.0002 max mem: 52150 +Epoch: [26] [5550/7269] eta: 0:12:25 lr: 0.000004 loss: 0.4897 (0.5176) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [26] [5560/7269] eta: 0:12:21 lr: 0.000004 loss: 0.5120 (0.5176) time: 0.4344 data: 0.0002 max mem: 52150 +Epoch: [26] [5570/7269] eta: 0:12:17 lr: 0.000004 loss: 0.5476 (0.5176) time: 0.4338 data: 0.0002 max mem: 52150 +Epoch: [26] [5580/7269] eta: 0:12:13 lr: 0.000004 loss: 0.5363 (0.5176) time: 0.4181 data: 0.0002 max mem: 52150 +Epoch: [26] [5590/7269] eta: 0:12:08 lr: 0.000004 loss: 0.4964 (0.5176) time: 0.4688 data: 0.0002 max mem: 52150 +Epoch: [26] [5600/7269] eta: 0:12:04 lr: 0.000004 loss: 0.5250 (0.5176) time: 0.4235 data: 0.0002 max mem: 52150 +Epoch: [26] [5610/7269] eta: 0:12:00 lr: 0.000004 loss: 0.5565 (0.5177) time: 0.4686 data: 0.0002 max mem: 52150 +Epoch: [26] [5620/7269] eta: 0:11:55 lr: 0.000004 loss: 0.5424 (0.5177) time: 0.4677 data: 0.0002 max mem: 52150 +Epoch: [26] [5630/7269] eta: 0:11:51 lr: 0.000004 loss: 0.5168 (0.5177) time: 0.4346 data: 0.0002 max mem: 52150 +Epoch: [26] [5640/7269] eta: 0:11:47 lr: 0.000004 loss: 0.5171 (0.5177) time: 0.4816 data: 0.0002 max mem: 52150 +Epoch: [26] [5650/7269] eta: 0:11:42 lr: 0.000004 loss: 0.5178 (0.5177) time: 0.4191 data: 0.0002 max mem: 52150 +Epoch: [26] [5660/7269] eta: 0:11:38 lr: 0.000004 loss: 0.5178 (0.5177) time: 0.4340 data: 0.0002 max mem: 52150 +Epoch: [26] [5670/7269] eta: 0:11:34 lr: 0.000004 loss: 0.5280 (0.5177) time: 0.4951 data: 0.0002 max mem: 52150 +Epoch: [26] [5680/7269] eta: 0:11:30 lr: 0.000004 loss: 0.5271 (0.5177) time: 0.4819 data: 0.0002 max mem: 52150 +Epoch: [26] [5690/7269] eta: 0:11:26 lr: 0.000004 loss: 0.5329 (0.5177) time: 0.5339 data: 0.0002 max mem: 52150 +Epoch: [26] [5700/7269] eta: 0:11:22 lr: 0.000004 loss: 0.5329 (0.5178) time: 0.5366 data: 0.0002 max mem: 52150 +Epoch: [26] [5710/7269] eta: 0:11:17 lr: 0.000004 loss: 0.5236 (0.5177) time: 0.4865 data: 0.0002 max mem: 52150 +Epoch: [26] [5720/7269] eta: 0:11:13 lr: 0.000004 loss: 0.4840 (0.5177) time: 0.4954 data: 0.0002 max mem: 52150 +Epoch: [26] [5730/7269] eta: 0:11:09 lr: 0.000004 loss: 0.5073 (0.5177) time: 0.4325 data: 0.0002 max mem: 52150 +Epoch: [26] [5740/7269] eta: 0:11:04 lr: 0.000004 loss: 0.5257 (0.5177) time: 0.3727 data: 0.0002 max mem: 52150 +Epoch: [26] [5750/7269] eta: 0:11:00 lr: 0.000004 loss: 0.5260 (0.5177) time: 0.4214 data: 0.0002 max mem: 52150 +Epoch: [26] [5760/7269] eta: 0:10:55 lr: 0.000004 loss: 0.5348 (0.5177) time: 0.4207 data: 0.0002 max mem: 52150 +Epoch: [26] [5770/7269] eta: 0:10:51 lr: 0.000004 loss: 0.5481 (0.5177) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [26] [5780/7269] eta: 0:10:47 lr: 0.000004 loss: 0.5435 (0.5178) time: 0.4325 data: 0.0002 max mem: 52150 +Epoch: [26] [5790/7269] eta: 0:10:42 lr: 0.000004 loss: 0.5296 (0.5178) time: 0.4332 data: 0.0002 max mem: 52150 +Epoch: [26] [5800/7269] eta: 0:10:38 lr: 0.000004 loss: 0.5443 (0.5178) time: 0.4193 data: 0.0002 max mem: 52150 +Epoch: [26] [5810/7269] eta: 0:10:34 lr: 0.000004 loss: 0.5406 (0.5178) time: 0.4830 data: 0.0002 max mem: 52150 +Epoch: [26] [5820/7269] eta: 0:10:29 lr: 0.000004 loss: 0.5376 (0.5179) time: 0.4359 data: 0.0002 max mem: 52150 +Epoch: [26] [5830/7269] eta: 0:10:25 lr: 0.000004 loss: 0.5238 (0.5178) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [26] [5840/7269] eta: 0:10:21 lr: 0.000004 loss: 0.4839 (0.5178) time: 0.4180 data: 0.0002 max mem: 52150 +Epoch: [26] [5850/7269] eta: 0:10:16 lr: 0.000004 loss: 0.4839 (0.5177) time: 0.4177 data: 0.0002 max mem: 52150 +Epoch: [26] [5860/7269] eta: 0:10:12 lr: 0.000004 loss: 0.5052 (0.5177) time: 0.4336 data: 0.0002 max mem: 52150 +Epoch: [26] [5870/7269] eta: 0:10:08 lr: 0.000004 loss: 0.5165 (0.5177) time: 0.4816 data: 0.0002 max mem: 52150 +Epoch: [26] [5880/7269] eta: 0:10:03 lr: 0.000004 loss: 0.5101 (0.5177) time: 0.4204 data: 0.0002 max mem: 52150 +Epoch: [26] [5890/7269] eta: 0:09:59 lr: 0.000004 loss: 0.5137 (0.5177) time: 0.4323 data: 0.0002 max mem: 52150 +Epoch: [26] [5900/7269] eta: 0:09:54 lr: 0.000004 loss: 0.5573 (0.5178) time: 0.4319 data: 0.0002 max mem: 52150 +Epoch: [26] [5910/7269] eta: 0:09:50 lr: 0.000004 loss: 0.5599 (0.5179) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [26] [5920/7269] eta: 0:09:45 lr: 0.000004 loss: 0.5442 (0.5178) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [26] [5930/7269] eta: 0:09:41 lr: 0.000004 loss: 0.5261 (0.5178) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [26] [5940/7269] eta: 0:09:37 lr: 0.000004 loss: 0.5174 (0.5178) time: 0.4836 data: 0.0002 max mem: 52150 +Epoch: [26] [5950/7269] eta: 0:09:32 lr: 0.000004 loss: 0.5174 (0.5178) time: 0.4835 data: 0.0002 max mem: 52150 +Epoch: [26] [5960/7269] eta: 0:09:28 lr: 0.000004 loss: 0.5182 (0.5178) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [26] [5970/7269] eta: 0:09:24 lr: 0.000004 loss: 0.5089 (0.5178) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [26] [5980/7269] eta: 0:09:19 lr: 0.000004 loss: 0.5131 (0.5178) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [26] [5990/7269] eta: 0:09:15 lr: 0.000004 loss: 0.5352 (0.5178) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [26] [6000/7269] eta: 0:09:10 lr: 0.000004 loss: 0.5441 (0.5179) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [26] [6010/7269] eta: 0:09:06 lr: 0.000004 loss: 0.5253 (0.5178) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [26] [6020/7269] eta: 0:09:01 lr: 0.000004 loss: 0.5076 (0.5179) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [26] [6030/7269] eta: 0:08:57 lr: 0.000004 loss: 0.5342 (0.5179) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [26] [6040/7269] eta: 0:08:52 lr: 0.000004 loss: 0.5394 (0.5179) time: 0.4165 data: 0.0002 max mem: 52150 +Epoch: [26] [6050/7269] eta: 0:08:48 lr: 0.000004 loss: 0.5363 (0.5179) time: 0.5086 data: 0.0002 max mem: 52150 +Epoch: [26] [6060/7269] eta: 0:08:44 lr: 0.000004 loss: 0.5203 (0.5179) time: 0.4636 data: 0.0002 max mem: 52150 +Epoch: [26] [6070/7269] eta: 0:08:39 lr: 0.000004 loss: 0.5041 (0.5178) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [26] [6080/7269] eta: 0:08:35 lr: 0.000004 loss: 0.5103 (0.5179) time: 0.4797 data: 0.0002 max mem: 52150 +Epoch: [26] [6090/7269] eta: 0:08:31 lr: 0.000004 loss: 0.5029 (0.5178) time: 0.6038 data: 0.0002 max mem: 52150 +Epoch: [26] [6100/7269] eta: 0:08:27 lr: 0.000004 loss: 0.4793 (0.5177) time: 0.4962 data: 0.0002 max mem: 52150 +Epoch: [26] [6110/7269] eta: 0:08:23 lr: 0.000004 loss: 0.4872 (0.5177) time: 0.4336 data: 0.0002 max mem: 52150 +Epoch: [26] [6120/7269] eta: 0:08:18 lr: 0.000004 loss: 0.5001 (0.5177) time: 0.4335 data: 0.0002 max mem: 52150 +Epoch: [26] [6130/7269] eta: 0:08:14 lr: 0.000004 loss: 0.5182 (0.5177) time: 0.4332 data: 0.0002 max mem: 52150 +Epoch: [26] [6140/7269] eta: 0:08:10 lr: 0.000004 loss: 0.5306 (0.5177) time: 0.5436 data: 0.0002 max mem: 52150 +Epoch: [26] [6150/7269] eta: 0:08:06 lr: 0.000004 loss: 0.5159 (0.5177) time: 0.4820 data: 0.0002 max mem: 52150 +Epoch: [26] [6160/7269] eta: 0:08:01 lr: 0.000004 loss: 0.5067 (0.5177) time: 0.4193 data: 0.0002 max mem: 52150 +Epoch: [26] [6170/7269] eta: 0:07:57 lr: 0.000004 loss: 0.5194 (0.5177) time: 0.4821 data: 0.0002 max mem: 52150 +Epoch: [26] [6180/7269] eta: 0:07:53 lr: 0.000004 loss: 0.5202 (0.5177) time: 0.4343 data: 0.0002 max mem: 52150 +Epoch: [26] [6190/7269] eta: 0:07:48 lr: 0.000004 loss: 0.4766 (0.5176) time: 0.4206 data: 0.0002 max mem: 52150 +Epoch: [26] [6200/7269] eta: 0:07:44 lr: 0.000004 loss: 0.5169 (0.5177) time: 0.4206 data: 0.0002 max mem: 52150 +Epoch: [26] [6210/7269] eta: 0:07:40 lr: 0.000004 loss: 0.5171 (0.5176) time: 0.4344 data: 0.0002 max mem: 52150 +Epoch: [26] [6220/7269] eta: 0:07:35 lr: 0.000004 loss: 0.5299 (0.5177) time: 0.4343 data: 0.0002 max mem: 52150 +Epoch: [26] [6230/7269] eta: 0:07:31 lr: 0.000004 loss: 0.5299 (0.5176) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [26] [6240/7269] eta: 0:07:26 lr: 0.000004 loss: 0.5155 (0.5176) time: 0.4316 data: 0.0002 max mem: 52150 +Epoch: [26] [6250/7269] eta: 0:07:22 lr: 0.000004 loss: 0.5259 (0.5176) time: 0.5538 data: 0.0002 max mem: 52150 +Epoch: [26] [6260/7269] eta: 0:07:18 lr: 0.000004 loss: 0.5406 (0.5176) time: 0.5402 data: 0.0002 max mem: 52150 +Epoch: [26] [6270/7269] eta: 0:07:14 lr: 0.000004 loss: 0.5088 (0.5176) time: 0.4642 data: 0.0002 max mem: 52150 +Epoch: [26] [6280/7269] eta: 0:07:09 lr: 0.000004 loss: 0.5088 (0.5175) time: 0.4176 data: 0.0002 max mem: 52150 +Epoch: [26] [6290/7269] eta: 0:07:05 lr: 0.000004 loss: 0.5013 (0.5175) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [26] [6300/7269] eta: 0:07:00 lr: 0.000004 loss: 0.5120 (0.5176) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [26] [6310/7269] eta: 0:06:56 lr: 0.000004 loss: 0.5226 (0.5175) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [26] [6320/7269] eta: 0:06:52 lr: 0.000004 loss: 0.4911 (0.5175) time: 0.4373 data: 0.0002 max mem: 52150 +Epoch: [26] [6330/7269] eta: 0:06:48 lr: 0.000004 loss: 0.4911 (0.5175) time: 0.4990 data: 0.0002 max mem: 52150 +Epoch: [26] [6340/7269] eta: 0:06:43 lr: 0.000004 loss: 0.5398 (0.5175) time: 0.4335 data: 0.0002 max mem: 52150 +Epoch: [26] [6350/7269] eta: 0:06:39 lr: 0.000004 loss: 0.5576 (0.5176) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [26] [6360/7269] eta: 0:06:34 lr: 0.000004 loss: 0.5424 (0.5176) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [26] [6370/7269] eta: 0:06:30 lr: 0.000004 loss: 0.5372 (0.5177) time: 0.4214 data: 0.0002 max mem: 52150 +Epoch: [26] [6380/7269] eta: 0:06:25 lr: 0.000004 loss: 0.5376 (0.5177) time: 0.4219 data: 0.0002 max mem: 52150 +Epoch: [26] [6390/7269] eta: 0:06:21 lr: 0.000004 loss: 0.5271 (0.5177) time: 0.4337 data: 0.0002 max mem: 52150 +Epoch: [26] [6400/7269] eta: 0:06:17 lr: 0.000004 loss: 0.4827 (0.5176) time: 0.4934 data: 0.0002 max mem: 52150 +Epoch: [26] [6410/7269] eta: 0:06:13 lr: 0.000004 loss: 0.5206 (0.5176) time: 0.4771 data: 0.0002 max mem: 52150 +Epoch: [26] [6420/7269] eta: 0:06:08 lr: 0.000004 loss: 0.5362 (0.5176) time: 0.4174 data: 0.0002 max mem: 52150 +Epoch: [26] [6430/7269] eta: 0:06:04 lr: 0.000004 loss: 0.5357 (0.5176) time: 0.3731 data: 0.0002 max mem: 52150 +Epoch: [26] [6440/7269] eta: 0:05:59 lr: 0.000004 loss: 0.5490 (0.5176) time: 0.3728 data: 0.0002 max mem: 52150 +Epoch: [26] [6450/7269] eta: 0:05:55 lr: 0.000004 loss: 0.5188 (0.5176) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [26] [6460/7269] eta: 0:05:51 lr: 0.000004 loss: 0.4975 (0.5176) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [26] [6470/7269] eta: 0:05:46 lr: 0.000004 loss: 0.5177 (0.5176) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [26] [6480/7269] eta: 0:05:42 lr: 0.000004 loss: 0.5335 (0.5176) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [26] [6490/7269] eta: 0:05:37 lr: 0.000004 loss: 0.5486 (0.5176) time: 0.4207 data: 0.0002 max mem: 52150 +Epoch: [26] [6500/7269] eta: 0:05:33 lr: 0.000004 loss: 0.5345 (0.5176) time: 0.4201 data: 0.0002 max mem: 52150 +Epoch: [26] [6510/7269] eta: 0:05:29 lr: 0.000004 loss: 0.5338 (0.5177) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [26] [6520/7269] eta: 0:05:24 lr: 0.000004 loss: 0.5351 (0.5177) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [26] [6530/7269] eta: 0:05:20 lr: 0.000004 loss: 0.5279 (0.5176) time: 0.3728 data: 0.0002 max mem: 52150 +Epoch: [26] [6540/7269] eta: 0:05:15 lr: 0.000004 loss: 0.4654 (0.5175) time: 0.4166 data: 0.0002 max mem: 52150 +Epoch: [26] [6550/7269] eta: 0:05:11 lr: 0.000004 loss: 0.4938 (0.5175) time: 0.4770 data: 0.0002 max mem: 52150 +Epoch: [26] [6560/7269] eta: 0:05:07 lr: 0.000004 loss: 0.5432 (0.5175) time: 0.4937 data: 0.0002 max mem: 52150 +Epoch: [26] [6570/7269] eta: 0:05:03 lr: 0.000004 loss: 0.5302 (0.5175) time: 0.4338 data: 0.0002 max mem: 52150 +Epoch: [26] [6580/7269] eta: 0:04:58 lr: 0.000004 loss: 0.5172 (0.5175) time: 0.4243 data: 0.0002 max mem: 52150 +Epoch: [26] [6590/7269] eta: 0:04:54 lr: 0.000004 loss: 0.5757 (0.5176) time: 0.5169 data: 0.0002 max mem: 52150 +Epoch: [26] [6600/7269] eta: 0:04:50 lr: 0.000004 loss: 0.5479 (0.5176) time: 0.4650 data: 0.0002 max mem: 52150 +Epoch: [26] [6610/7269] eta: 0:04:45 lr: 0.000004 loss: 0.4654 (0.5175) time: 0.4359 data: 0.0002 max mem: 52150 +Epoch: [26] [6620/7269] eta: 0:04:41 lr: 0.000004 loss: 0.4917 (0.5175) time: 0.4837 data: 0.0002 max mem: 52150 +Epoch: [26] [6630/7269] eta: 0:04:37 lr: 0.000004 loss: 0.5175 (0.5175) time: 0.4799 data: 0.0002 max mem: 52150 +Epoch: [26] [6640/7269] eta: 0:04:32 lr: 0.000004 loss: 0.5175 (0.5175) time: 0.4325 data: 0.0002 max mem: 52150 +Epoch: [26] [6650/7269] eta: 0:04:28 lr: 0.000004 loss: 0.5187 (0.5175) time: 0.4672 data: 0.0002 max mem: 52150 +Epoch: [26] [6660/7269] eta: 0:04:24 lr: 0.000004 loss: 0.5340 (0.5175) time: 0.4670 data: 0.0002 max mem: 52150 +Epoch: [26] [6670/7269] eta: 0:04:19 lr: 0.000004 loss: 0.5160 (0.5175) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [26] [6680/7269] eta: 0:04:15 lr: 0.000004 loss: 0.5224 (0.5175) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [26] [6690/7269] eta: 0:04:11 lr: 0.000004 loss: 0.5459 (0.5174) time: 0.4667 data: 0.0002 max mem: 52150 +Epoch: [26] [6700/7269] eta: 0:04:06 lr: 0.000004 loss: 0.5423 (0.5175) time: 0.4664 data: 0.0002 max mem: 52150 +Epoch: [26] [6710/7269] eta: 0:04:02 lr: 0.000004 loss: 0.5278 (0.5175) time: 0.4341 data: 0.0002 max mem: 52150 +Epoch: [26] [6720/7269] eta: 0:03:58 lr: 0.000004 loss: 0.5140 (0.5174) time: 0.4963 data: 0.0002 max mem: 52150 +Epoch: [26] [6730/7269] eta: 0:03:53 lr: 0.000004 loss: 0.5145 (0.5174) time: 0.4817 data: 0.0002 max mem: 52150 +Epoch: [26] [6740/7269] eta: 0:03:49 lr: 0.000004 loss: 0.4870 (0.5173) time: 0.4200 data: 0.0002 max mem: 52150 +Epoch: [26] [6750/7269] eta: 0:03:45 lr: 0.000004 loss: 0.5036 (0.5174) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [26] [6760/7269] eta: 0:03:40 lr: 0.000004 loss: 0.5357 (0.5174) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [26] [6770/7269] eta: 0:03:36 lr: 0.000004 loss: 0.5225 (0.5173) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [26] [6780/7269] eta: 0:03:32 lr: 0.000004 loss: 0.4827 (0.5173) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [26] [6790/7269] eta: 0:03:27 lr: 0.000004 loss: 0.5096 (0.5173) time: 0.4832 data: 0.0002 max mem: 52150 +Epoch: [26] [6800/7269] eta: 0:03:23 lr: 0.000004 loss: 0.5204 (0.5173) time: 0.4831 data: 0.0002 max mem: 52150 +Epoch: [26] [6810/7269] eta: 0:03:19 lr: 0.000004 loss: 0.5145 (0.5173) time: 0.4318 data: 0.0002 max mem: 52150 +Epoch: [26] [6820/7269] eta: 0:03:14 lr: 0.000004 loss: 0.4854 (0.5172) time: 0.4943 data: 0.0002 max mem: 52150 +Epoch: [26] [6830/7269] eta: 0:03:10 lr: 0.000004 loss: 0.5460 (0.5173) time: 0.4346 data: 0.0002 max mem: 52150 +Epoch: [26] [6840/7269] eta: 0:03:06 lr: 0.000004 loss: 0.5286 (0.5173) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [26] [6850/7269] eta: 0:03:01 lr: 0.000004 loss: 0.5033 (0.5172) time: 0.4192 data: 0.0002 max mem: 52150 +Epoch: [26] [6860/7269] eta: 0:02:57 lr: 0.000004 loss: 0.5222 (0.5172) time: 0.4186 data: 0.0002 max mem: 52150 +Epoch: [26] [6870/7269] eta: 0:02:53 lr: 0.000004 loss: 0.5406 (0.5173) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [26] [6880/7269] eta: 0:02:48 lr: 0.000004 loss: 0.5679 (0.5174) time: 0.4817 data: 0.0002 max mem: 52150 +Epoch: [26] [6890/7269] eta: 0:02:44 lr: 0.000004 loss: 0.5523 (0.5174) time: 0.5430 data: 0.0002 max mem: 52150 +Epoch: [26] [6900/7269] eta: 0:02:40 lr: 0.000004 loss: 0.4946 (0.5173) time: 0.4818 data: 0.0002 max mem: 52150 +Epoch: [26] [6910/7269] eta: 0:02:35 lr: 0.000004 loss: 0.5026 (0.5173) time: 0.4211 data: 0.0002 max mem: 52150 +Epoch: [26] [6920/7269] eta: 0:02:31 lr: 0.000004 loss: 0.5596 (0.5173) time: 0.4324 data: 0.0002 max mem: 52150 +Epoch: [26] [6930/7269] eta: 0:02:27 lr: 0.000004 loss: 0.5456 (0.5174) time: 0.4315 data: 0.0002 max mem: 52150 +Epoch: [26] [6940/7269] eta: 0:02:22 lr: 0.000004 loss: 0.5456 (0.5174) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [26] [6950/7269] eta: 0:02:18 lr: 0.000004 loss: 0.5437 (0.5175) time: 0.4775 data: 0.0002 max mem: 52150 +Epoch: [26] [6960/7269] eta: 0:02:14 lr: 0.000004 loss: 0.5343 (0.5175) time: 0.4776 data: 0.0002 max mem: 52150 +Epoch: [26] [6970/7269] eta: 0:02:09 lr: 0.000004 loss: 0.5401 (0.5175) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [26] [6980/7269] eta: 0:02:05 lr: 0.000004 loss: 0.5606 (0.5176) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [26] [6990/7269] eta: 0:02:01 lr: 0.000004 loss: 0.5376 (0.5176) time: 0.4335 data: 0.0002 max mem: 52150 +Epoch: [26] [7000/7269] eta: 0:01:56 lr: 0.000004 loss: 0.5076 (0.5175) time: 0.4334 data: 0.0002 max mem: 52150 +Epoch: [26] [7010/7269] eta: 0:01:52 lr: 0.000004 loss: 0.4784 (0.5175) time: 0.3727 data: 0.0002 max mem: 52150 +Epoch: [26] [7020/7269] eta: 0:01:48 lr: 0.000004 loss: 0.5258 (0.5175) time: 0.4180 data: 0.0002 max mem: 52150 +Epoch: [26] [7030/7269] eta: 0:01:43 lr: 0.000004 loss: 0.5308 (0.5175) time: 0.4172 data: 0.0002 max mem: 52150 +Epoch: [26] [7040/7269] eta: 0:01:39 lr: 0.000004 loss: 0.5247 (0.5175) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [26] [7050/7269] eta: 0:01:34 lr: 0.000004 loss: 0.4741 (0.5174) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [26] [7060/7269] eta: 0:01:30 lr: 0.000004 loss: 0.4657 (0.5174) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [26] [7070/7269] eta: 0:01:26 lr: 0.000004 loss: 0.5129 (0.5174) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [26] [7080/7269] eta: 0:01:21 lr: 0.000004 loss: 0.5137 (0.5174) time: 0.4927 data: 0.0002 max mem: 52150 +Epoch: [26] [7090/7269] eta: 0:01:17 lr: 0.000004 loss: 0.5634 (0.5174) time: 0.5552 data: 0.0002 max mem: 52150 +Epoch: [26] [7100/7269] eta: 0:01:13 lr: 0.000004 loss: 0.5493 (0.5174) time: 0.4342 data: 0.0002 max mem: 52150 +Epoch: [26] [7110/7269] eta: 0:01:08 lr: 0.000004 loss: 0.5163 (0.5174) time: 0.4790 data: 0.0002 max mem: 52150 +Epoch: [26] [7120/7269] eta: 0:01:04 lr: 0.000004 loss: 0.5166 (0.5174) time: 0.5772 data: 0.0002 max mem: 52150 +Epoch: [26] [7130/7269] eta: 0:01:00 lr: 0.000004 loss: 0.5047 (0.5174) time: 0.4696 data: 0.0002 max mem: 52150 +Epoch: [26] [7140/7269] eta: 0:00:55 lr: 0.000004 loss: 0.5270 (0.5174) time: 0.4172 data: 0.0002 max mem: 52150 +Epoch: [26] [7150/7269] eta: 0:00:51 lr: 0.000004 loss: 0.5248 (0.5174) time: 0.5270 data: 0.0002 max mem: 52150 +Epoch: [26] [7160/7269] eta: 0:00:47 lr: 0.000004 loss: 0.5381 (0.5174) time: 0.5303 data: 0.0002 max mem: 52150 +Epoch: [26] [7170/7269] eta: 0:00:42 lr: 0.000004 loss: 0.5451 (0.5174) time: 0.4687 data: 0.0002 max mem: 52150 +Epoch: [26] [7180/7269] eta: 0:00:38 lr: 0.000004 loss: 0.5257 (0.5174) time: 0.4822 data: 0.0002 max mem: 52150 +Epoch: [26] [7190/7269] eta: 0:00:34 lr: 0.000004 loss: 0.5333 (0.5174) time: 0.4339 data: 0.0002 max mem: 52150 +Epoch: [26] [7200/7269] eta: 0:00:29 lr: 0.000004 loss: 0.5007 (0.5174) time: 0.4337 data: 0.0002 max mem: 52150 +Epoch: [26] [7210/7269] eta: 0:00:25 lr: 0.000004 loss: 0.4970 (0.5173) time: 0.4952 data: 0.0002 max mem: 52150 +Epoch: [26] [7220/7269] eta: 0:00:21 lr: 0.000004 loss: 0.5119 (0.5173) time: 0.4327 data: 0.0002 max mem: 52150 +Epoch: [26] [7230/7269] eta: 0:00:16 lr: 0.000004 loss: 0.5188 (0.5174) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [26] [7240/7269] eta: 0:00:12 lr: 0.000004 loss: 0.5367 (0.5174) time: 0.4181 data: 0.0002 max mem: 52150 +Epoch: [26] [7250/7269] eta: 0:00:08 lr: 0.000004 loss: 0.4880 (0.5173) time: 0.4221 data: 0.0036 max mem: 52150 +Epoch: [26] [7260/7269] eta: 0:00:03 lr: 0.000004 loss: 0.4764 (0.5173) time: 0.3748 data: 0.0036 max mem: 52150 +Epoch: [26] [7268/7269] eta: 0:00:00 lr: 0.000004 loss: 0.5100 (0.5173) time: 0.4315 data: 0.0036 max mem: 52150 +Epoch: [26] Total time: 0:52:39 (0.4346 s / it) +Averaged stats: lr: 0.000004 loss: 0.5100 (0.5166) +Test: [ 0/1403] eta: 1:14:51 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1143 (0.1143) acc1: 98.4375 (98.4375) acc5: 100.0000 (100.0000) time: 3.2012 data: 3.1040 max mem: 52150 +Test: [ 10/1403] eta: 0:08:47 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.2077 (0.2423) acc1: 95.3125 (91.4773) acc5: 100.0000 (100.0000) time: 0.3788 data: 0.2825 max mem: 52150 +Test: [ 20/1403] eta: 0:05:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1591 (0.2189) acc1: 95.3125 (92.5595) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 30/1403] eta: 0:04:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1068 (0.2010) acc1: 98.4375 (93.3972) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 40/1403] eta: 0:03:54 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1164 (0.1954) acc1: 98.4375 (94.0168) acc5: 100.0000 (100.0000) time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 50/1403] eta: 0:03:33 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1214 (0.2066) acc1: 98.4375 (93.2904) acc5: 100.0000 (100.0000) time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 60/1403] eta: 0:03:18 acc1_pos: 92.188 acc5_pos: 100.000 loss: 0.1230 (0.1991) acc1: 100.0000 (93.9037) acc5: 100.0000 (100.0000) time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 70/1403] eta: 0:03:07 acc1_pos: 92.188 acc5_pos: 100.000 loss: 0.1725 (0.2057) acc1: 95.3125 (93.5519) acc5: 100.0000 (100.0000) time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 80/1403] eta: 0:02:58 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.2182 (0.2135) acc1: 93.7500 (93.0941) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 90/1403] eta: 0:02:51 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.2032 (0.2327) acc1: 93.7500 (92.2047) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 100/1403] eta: 0:02:45 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1977 (0.2309) acc1: 95.3125 (92.3886) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 110/1403] eta: 0:02:41 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1637 (0.2275) acc1: 96.8750 (92.5676) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 120/1403] eta: 0:02:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0871 (0.2155) acc1: 98.4375 (93.1431) acc5: 100.0000 (100.0000) time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 130/1403] eta: 0:02:33 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0841 (0.2073) acc1: 98.4375 (93.5115) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 140/1403] eta: 0:02:29 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0789 (0.1979) acc1: 100.0000 (93.9716) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 150/1403] eta: 0:02:26 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0733 (0.1903) acc1: 100.0000 (94.3295) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 160/1403] eta: 0:02:24 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0769 (0.1839) acc1: 100.0000 (94.6234) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 170/1403] eta: 0:02:21 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0784 (0.1782) acc1: 100.0000 (94.9105) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 180/1403] eta: 0:02:19 acc1_pos: 90.625 acc5_pos: 100.000 loss: 0.0805 (0.1803) acc1: 100.0000 (94.8291) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 190/1403] eta: 0:02:16 acc1_pos: 90.625 acc5_pos: 100.000 loss: 0.2832 (0.1886) acc1: 90.6250 (94.3717) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 200/1403] eta: 0:02:14 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.2832 (0.1984) acc1: 87.5000 (93.9288) acc5: 100.0000 (100.0000) time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 210/1403] eta: 0:02:12 acc1_pos: 73.438 acc5_pos: 100.000 loss: 0.3367 (0.2082) acc1: 85.9375 (93.3723) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 220/1403] eta: 0:02:10 acc1_pos: 87.500 acc5_pos: 100.000 loss: 0.3430 (0.2160) acc1: 87.5000 (93.0076) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 230/1403] eta: 0:02:09 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2673 (0.2111) acc1: 92.1875 (93.2495) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 240/1403] eta: 0:02:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0768 (0.2056) acc1: 100.0000 (93.5166) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 250/1403] eta: 0:02:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0759 (0.2008) acc1: 100.0000 (93.7562) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 260/1403] eta: 0:02:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0758 (0.1963) acc1: 100.0000 (93.9775) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 270/1403] eta: 0:02:02 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0778 (0.1921) acc1: 100.0000 (94.1709) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 280/1403] eta: 0:02:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0779 (0.1880) acc1: 100.0000 (94.3728) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 290/1403] eta: 0:01:59 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0714 (0.1841) acc1: 100.0000 (94.5608) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 300/1403] eta: 0:01:57 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0708 (0.1805) acc1: 100.0000 (94.7363) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 310/1403] eta: 0:01:56 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0742 (0.1774) acc1: 100.0000 (94.9005) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 320/1403] eta: 0:01:55 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0930 (0.1748) acc1: 100.0000 (95.0594) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 330/1403] eta: 0:01:53 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0988 (0.1726) acc1: 100.0000 (95.2086) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 340/1403] eta: 0:01:52 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0988 (0.1703) acc1: 100.0000 (95.3492) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0004 max mem: 52150 +Test: [ 350/1403] eta: 0:01:51 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0961 (0.1682) acc1: 100.0000 (95.4817) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 360/1403] eta: 0:01:49 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0988 (0.1664) acc1: 100.0000 (95.6025) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 370/1403] eta: 0:01:48 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0977 (0.1645) acc1: 100.0000 (95.7210) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 380/1403] eta: 0:01:47 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0965 (0.1628) acc1: 100.0000 (95.8333) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 390/1403] eta: 0:01:45 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0975 (0.1625) acc1: 100.0000 (95.8360) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 400/1403] eta: 0:01:44 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2036 (0.1647) acc1: 92.1875 (95.7177) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 410/1403] eta: 0:01:43 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2374 (0.1663) acc1: 92.1875 (95.6356) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 420/1403] eta: 0:01:42 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1984 (0.1666) acc1: 93.7500 (95.6020) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 430/1403] eta: 0:01:41 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1946 (0.1682) acc1: 93.7500 (95.5083) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 440/1403] eta: 0:01:39 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2439 (0.1705) acc1: 90.6250 (95.3940) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 450/1403] eta: 0:01:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2342 (0.1718) acc1: 90.6250 (95.3160) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 460/1403] eta: 0:01:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2068 (0.1730) acc1: 92.1875 (95.2312) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 470/1403] eta: 0:01:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1817 (0.1718) acc1: 93.7500 (95.2760) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 480/1403] eta: 0:01:35 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0624 (0.1695) acc1: 100.0000 (95.3742) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0004 max mem: 52150 +Test: [ 490/1403] eta: 0:01:33 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0626 (0.1673) acc1: 100.0000 (95.4684) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0004 max mem: 52150 +Test: [ 500/1403] eta: 0:01:32 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0628 (0.1652) acc1: 100.0000 (95.5589) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 510/1403] eta: 0:01:31 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0629 (0.1633) acc1: 100.0000 (95.6458) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0004 max mem: 52150 +Test: [ 520/1403] eta: 0:01:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0661 (0.1615) acc1: 100.0000 (95.7294) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 530/1403] eta: 0:01:29 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0595 (0.1595) acc1: 100.0000 (95.8098) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 540/1403] eta: 0:01:28 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0589 (0.1577) acc1: 100.0000 (95.8872) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 550/1403] eta: 0:01:27 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0598 (0.1563) acc1: 100.0000 (95.9505) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 560/1403] eta: 0:01:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0726 (0.1549) acc1: 100.0000 (96.0116) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 570/1403] eta: 0:01:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0763 (0.1537) acc1: 100.0000 (96.0705) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 580/1403] eta: 0:01:23 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0808 (0.1529) acc1: 100.0000 (96.1005) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 590/1403] eta: 0:01:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1010 (0.1522) acc1: 98.4375 (96.1242) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0968 data: 0.0004 max mem: 52150 +Test: [ 600/1403] eta: 0:01:21 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1173 (0.1522) acc1: 96.8750 (96.1080) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 610/1403] eta: 0:01:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1173 (0.1514) acc1: 96.8750 (96.1257) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 620/1403] eta: 0:01:19 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1210 (0.1519) acc1: 95.3125 (96.0648) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 630/1403] eta: 0:01:18 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1472 (0.1517) acc1: 95.3125 (96.0628) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 640/1403] eta: 0:01:17 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1278 (0.1512) acc1: 96.8750 (96.0828) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 650/1403] eta: 0:01:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1238 (0.1509) acc1: 96.8750 (96.0974) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 660/1403] eta: 0:01:15 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0980 (0.1498) acc1: 98.4375 (96.1422) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 670/1403] eta: 0:01:14 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0663 (0.1486) acc1: 100.0000 (96.1974) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 680/1403] eta: 0:01:13 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0744 (0.1495) acc1: 98.4375 (96.1431) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [ 690/1403] eta: 0:01:12 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2294 (0.1534) acc1: 90.6250 (95.9185) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 700/1403] eta: 0:01:11 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1712 (0.1532) acc1: 95.3125 (95.9255) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 710/1403] eta: 0:01:10 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2137 (0.1564) acc1: 90.6250 (95.7542) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 720/1403] eta: 0:01:08 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2137 (0.1561) acc1: 89.0625 (95.7633) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 730/1403] eta: 0:01:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0668 (0.1550) acc1: 100.0000 (95.8191) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 740/1403] eta: 0:01:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0721 (0.1541) acc1: 100.0000 (95.8629) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 750/1403] eta: 0:01:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0883 (0.1550) acc1: 98.4375 (95.8098) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 760/1403] eta: 0:01:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2929 (0.1581) acc1: 87.5000 (95.6184) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 770/1403] eta: 0:01:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4008 (0.1616) acc1: 79.6875 (95.3956) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 780/1403] eta: 0:01:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3767 (0.1638) acc1: 81.2500 (95.2645) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 790/1403] eta: 0:01:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2887 (0.1660) acc1: 87.5000 (95.1584) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 800/1403] eta: 0:01:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1980 (0.1664) acc1: 93.7500 (95.1408) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 810/1403] eta: 0:00:59 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2142 (0.1691) acc1: 90.6250 (95.0216) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 820/1403] eta: 0:00:58 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5935 (0.1776) acc1: 73.4375 (94.6578) acc5: 100.0000 (100.0000) acc1_neg: 43.750 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 830/1403] eta: 0:00:57 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6429 (0.1798) acc1: 71.8750 (94.5491) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 840/1403] eta: 0:00:56 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4630 (0.1872) acc1: 76.5625 (94.2256) acc5: 100.0000 (100.0000) acc1_neg: 51.562 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 850/1403] eta: 0:00:55 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.9601 (0.1974) acc1: 59.3750 (93.8069) acc5: 100.0000 (100.0000) acc1_neg: 59.375 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 860/1403] eta: 0:00:54 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.8986 (0.2006) acc1: 62.5000 (93.6702) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 870/1403] eta: 0:00:53 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1130 (0.1994) acc1: 96.8750 (93.7249) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 880/1403] eta: 0:00:52 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0858 (0.1982) acc1: 98.4375 (93.7766) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [ 890/1403] eta: 0:00:51 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0764 (0.1969) acc1: 100.0000 (93.8394) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [ 900/1403] eta: 0:00:50 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0699 (0.1954) acc1: 100.0000 (93.9061) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 910/1403] eta: 0:00:49 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0734 (0.1945) acc1: 100.0000 (93.9524) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 920/1403] eta: 0:00:48 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0933 (0.1934) acc1: 98.4375 (94.0011) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 930/1403] eta: 0:00:47 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0891 (0.1923) acc1: 98.4375 (94.0555) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 940/1403] eta: 0:00:46 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1088 (0.1922) acc1: 98.4375 (94.0522) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 950/1403] eta: 0:00:45 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1927 (0.1927) acc1: 89.0625 (94.0080) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 960/1403] eta: 0:00:44 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2223 (0.1930) acc1: 89.0625 (93.9841) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 970/1403] eta: 0:00:43 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2084 (0.1935) acc1: 93.7500 (93.9560) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 980/1403] eta: 0:00:42 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2212 (0.1940) acc1: 92.1875 (93.9204) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 990/1403] eta: 0:00:41 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2037 (0.1941) acc1: 93.7500 (93.9203) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1000/1403] eta: 0:00:40 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1987 (0.1944) acc1: 93.7500 (93.9014) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1010/1403] eta: 0:00:39 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2117 (0.1949) acc1: 90.6250 (93.8659) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1020/1403] eta: 0:00:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3616 (0.2005) acc1: 82.8125 (93.5556) acc5: 100.0000 (100.0000) acc1_neg: 35.938 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1030/1403] eta: 0:00:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4419 (0.2019) acc1: 65.6250 (93.4711) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1040/1403] eta: 0:00:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1827 (0.2015) acc1: 93.7500 (93.4858) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1050/1403] eta: 0:00:35 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1328 (0.2010) acc1: 95.3125 (93.5062) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [1060/1403] eta: 0:00:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1122 (0.2002) acc1: 96.8750 (93.5424) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [1070/1403] eta: 0:00:33 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1363 (0.2006) acc1: 95.3125 (93.5268) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1080/1403] eta: 0:00:32 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3165 (0.2021) acc1: 87.5000 (93.4465) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1090/1403] eta: 0:00:31 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3635 (0.2036) acc1: 84.3750 (93.3676) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0970 data: 0.0003 max mem: 52150 +Test: [1100/1403] eta: 0:00:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3064 (0.2040) acc1: 85.9375 (93.3413) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0970 data: 0.0003 max mem: 52150 +Test: [1110/1403] eta: 0:00:29 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2364 (0.2044) acc1: 90.6250 (93.3281) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1120/1403] eta: 0:00:28 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2537 (0.2047) acc1: 90.6250 (93.3082) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1130/1403] eta: 0:00:27 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2538 (0.2053) acc1: 90.6250 (93.2775) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0966 data: 0.0002 max mem: 52150 +Test: [1140/1403] eta: 0:00:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2299 (0.2054) acc1: 92.1875 (93.2693) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1150/1403] eta: 0:00:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2172 (0.2057) acc1: 92.1875 (93.2491) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0966 data: 0.0002 max mem: 52150 +Test: [1160/1403] eta: 0:00:24 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2385 (0.2062) acc1: 90.6250 (93.2292) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1170/1403] eta: 0:00:23 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2375 (0.2060) acc1: 90.6250 (93.2323) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1180/1403] eta: 0:00:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0671 (0.2048) acc1: 100.0000 (93.2896) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1190/1403] eta: 0:00:21 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0658 (0.2036) acc1: 100.0000 (93.3459) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0002 max mem: 52150 +Test: [1200/1403] eta: 0:00:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0658 (0.2025) acc1: 100.0000 (93.4013) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1210/1403] eta: 0:00:19 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0655 (0.2014) acc1: 100.0000 (93.4558) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1220/1403] eta: 0:00:18 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0652 (0.2003) acc1: 100.0000 (93.5094) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1230/1403] eta: 0:00:17 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0660 (0.1992) acc1: 100.0000 (93.5621) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1240/1403] eta: 0:00:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0660 (0.1981) acc1: 100.0000 (93.6140) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1250/1403] eta: 0:00:15 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0661 (0.1971) acc1: 100.0000 (93.6651) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1260/1403] eta: 0:00:14 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0704 (0.1961) acc1: 100.0000 (93.7153) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1270/1403] eta: 0:00:13 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0719 (0.1951) acc1: 100.0000 (93.7648) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52150 +Test: [1280/1403] eta: 0:00:12 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0715 (0.1941) acc1: 100.0000 (93.8134) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52150 +Test: [1290/1403] eta: 0:00:11 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0712 (0.1932) acc1: 100.0000 (93.8613) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1300/1403] eta: 0:00:10 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0713 (0.1923) acc1: 100.0000 (93.9085) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1310/1403] eta: 0:00:09 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0721 (0.1913) acc1: 100.0000 (93.9550) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52150 +Test: [1320/1403] eta: 0:00:08 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0705 (0.1904) acc1: 100.0000 (94.0008) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1330/1403] eta: 0:00:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0703 (0.1895) acc1: 100.0000 (94.0423) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1340/1403] eta: 0:00:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0689 (0.1887) acc1: 100.0000 (94.0832) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1350/1403] eta: 0:00:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0689 (0.1878) acc1: 100.0000 (94.1224) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1360/1403] eta: 0:00:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0680 (0.1870) acc1: 100.0000 (94.1644) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52150 +Test: [1370/1403] eta: 0:00:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0699 (0.1862) acc1: 100.0000 (94.2047) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52150 +Test: [1380/1403] eta: 0:00:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0692 (0.1853) acc1: 100.0000 (94.2444) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1390/1403] eta: 0:00:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0670 (0.1845) acc1: 100.0000 (94.2847) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0002 max mem: 52150 +Test: [1400/1403] eta: 0:00:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0677 (0.1837) acc1: 100.0000 (94.3221) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0001 max mem: 52150 +Test: [1402/1403] eta: 0:00:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0677 (0.1836) acc1: 100.0000 (94.3262) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0922 data: 0.0001 max mem: 52150 +Test: Total time: 0:02:19 (0.0992 s / it) +* Overall Acc@1 94.326 Acc@5 100.000 + target == real Acc@1 94.855 Acc@5 100.000 + target == fake Acc@1 94.177 Acc@5 100.000 + loss 0.184 +Accuracy of the network on the 89729 test images: 94.3% +Max accuracy: 94.33% +Epoch: [27] [ 0/7269] eta: 9:06:29 lr: 0.000003 loss: 0.5214 (0.5214) time: 4.5108 data: 4.1277 max mem: 52150 +Epoch: [27] [ 10/7269] eta: 1:41:22 lr: 0.000003 loss: 0.5311 (0.5350) time: 0.8380 data: 0.3754 max mem: 52150 +Epoch: [27] [ 20/7269] eta: 1:14:21 lr: 0.000003 loss: 0.5439 (0.5406) time: 0.4207 data: 0.0002 max mem: 52150 +Epoch: [27] [ 30/7269] eta: 1:18:24 lr: 0.000003 loss: 0.5306 (0.5348) time: 0.5465 data: 0.0002 max mem: 52150 +Epoch: [27] [ 40/7269] eta: 1:13:48 lr: 0.000003 loss: 0.5286 (0.5349) time: 0.6097 data: 0.0002 max mem: 52150 +Epoch: [27] [ 50/7269] eta: 1:08:02 lr: 0.000003 loss: 0.5248 (0.5264) time: 0.4348 data: 0.0002 max mem: 52150 +Epoch: [27] [ 60/7269] eta: 1:04:07 lr: 0.000003 loss: 0.5248 (0.5284) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [27] [ 70/7269] eta: 1:01:18 lr: 0.000003 loss: 0.5152 (0.5214) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [27] [ 80/7269] eta: 0:59:09 lr: 0.000003 loss: 0.5152 (0.5214) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [27] [ 90/7269] eta: 0:57:27 lr: 0.000003 loss: 0.5311 (0.5228) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [27] [ 100/7269] eta: 0:57:16 lr: 0.000003 loss: 0.5311 (0.5225) time: 0.4217 data: 0.0002 max mem: 52150 +Epoch: [27] [ 110/7269] eta: 0:56:02 lr: 0.000003 loss: 0.5354 (0.5231) time: 0.4214 data: 0.0001 max mem: 52150 +Epoch: [27] [ 120/7269] eta: 0:54:59 lr: 0.000003 loss: 0.5354 (0.5217) time: 0.3711 data: 0.0002 max mem: 52150 +Epoch: [27] [ 130/7269] eta: 0:55:12 lr: 0.000003 loss: 0.5380 (0.5231) time: 0.4323 data: 0.0002 max mem: 52150 +Epoch: [27] [ 140/7269] eta: 0:55:24 lr: 0.000003 loss: 0.5472 (0.5243) time: 0.4954 data: 0.0002 max mem: 52150 +Epoch: [27] [ 150/7269] eta: 0:54:35 lr: 0.000003 loss: 0.5346 (0.5246) time: 0.4345 data: 0.0002 max mem: 52150 +Epoch: [27] [ 160/7269] eta: 0:55:42 lr: 0.000003 loss: 0.4907 (0.5234) time: 0.4979 data: 0.0002 max mem: 52150 +Epoch: [27] [ 170/7269] eta: 0:54:57 lr: 0.000003 loss: 0.5130 (0.5230) time: 0.4984 data: 0.0001 max mem: 52150 +Epoch: [27] [ 180/7269] eta: 0:54:16 lr: 0.000003 loss: 0.5146 (0.5216) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [27] [ 190/7269] eta: 0:53:39 lr: 0.000003 loss: 0.5146 (0.5230) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [27] [ 200/7269] eta: 0:53:49 lr: 0.000003 loss: 0.5341 (0.5232) time: 0.4348 data: 0.0002 max mem: 52150 +Epoch: [27] [ 210/7269] eta: 0:54:21 lr: 0.000003 loss: 0.5091 (0.5216) time: 0.5302 data: 0.0002 max mem: 52150 +Epoch: [27] [ 220/7269] eta: 0:53:47 lr: 0.000003 loss: 0.5101 (0.5215) time: 0.4674 data: 0.0002 max mem: 52150 +Epoch: [27] [ 230/7269] eta: 0:53:16 lr: 0.000003 loss: 0.5102 (0.5207) time: 0.3714 data: 0.0001 max mem: 52150 +Epoch: [27] [ 240/7269] eta: 0:52:47 lr: 0.000003 loss: 0.5062 (0.5205) time: 0.3710 data: 0.0001 max mem: 52150 +Epoch: [27] [ 250/7269] eta: 0:52:21 lr: 0.000003 loss: 0.4998 (0.5191) time: 0.3709 data: 0.0002 max mem: 52150 +Epoch: [27] [ 260/7269] eta: 0:53:04 lr: 0.000003 loss: 0.5017 (0.5195) time: 0.4990 data: 0.0002 max mem: 52150 +Epoch: [27] [ 270/7269] eta: 0:52:38 lr: 0.000003 loss: 0.5407 (0.5201) time: 0.4991 data: 0.0002 max mem: 52150 +Epoch: [27] [ 280/7269] eta: 0:53:10 lr: 0.000003 loss: 0.5117 (0.5183) time: 0.4835 data: 0.0002 max mem: 52150 +Epoch: [27] [ 290/7269] eta: 0:52:44 lr: 0.000003 loss: 0.5045 (0.5188) time: 0.4830 data: 0.0002 max mem: 52150 +Epoch: [27] [ 300/7269] eta: 0:52:21 lr: 0.000003 loss: 0.5147 (0.5184) time: 0.3712 data: 0.0002 max mem: 52150 +Epoch: [27] [ 310/7269] eta: 0:51:58 lr: 0.000003 loss: 0.5129 (0.5182) time: 0.3712 data: 0.0002 max mem: 52150 +Epoch: [27] [ 320/7269] eta: 0:51:37 lr: 0.000003 loss: 0.5165 (0.5179) time: 0.3708 data: 0.0002 max mem: 52150 +Epoch: [27] [ 330/7269] eta: 0:51:17 lr: 0.000003 loss: 0.5281 (0.5185) time: 0.3705 data: 0.0002 max mem: 52150 +Epoch: [27] [ 340/7269] eta: 0:50:58 lr: 0.000003 loss: 0.5361 (0.5188) time: 0.3706 data: 0.0002 max mem: 52150 +Epoch: [27] [ 350/7269] eta: 0:51:24 lr: 0.000003 loss: 0.5447 (0.5198) time: 0.4837 data: 0.0002 max mem: 52150 +Epoch: [27] [ 360/7269] eta: 0:51:30 lr: 0.000003 loss: 0.5502 (0.5197) time: 0.5478 data: 0.0002 max mem: 52150 +Epoch: [27] [ 370/7269] eta: 0:51:11 lr: 0.000003 loss: 0.5394 (0.5198) time: 0.4353 data: 0.0002 max mem: 52150 +Epoch: [27] [ 380/7269] eta: 0:51:16 lr: 0.000003 loss: 0.5335 (0.5194) time: 0.4344 data: 0.0002 max mem: 52150 +Epoch: [27] [ 390/7269] eta: 0:51:42 lr: 0.000003 loss: 0.5205 (0.5193) time: 0.5572 data: 0.0001 max mem: 52150 +Epoch: [27] [ 400/7269] eta: 0:51:24 lr: 0.000003 loss: 0.5229 (0.5201) time: 0.4946 data: 0.0002 max mem: 52150 +Epoch: [27] [ 410/7269] eta: 0:51:44 lr: 0.000003 loss: 0.5064 (0.5195) time: 0.4839 data: 0.0002 max mem: 52150 +Epoch: [27] [ 420/7269] eta: 0:51:26 lr: 0.000003 loss: 0.5016 (0.5189) time: 0.4836 data: 0.0002 max mem: 52150 +Epoch: [27] [ 430/7269] eta: 0:51:29 lr: 0.000003 loss: 0.5320 (0.5187) time: 0.4341 data: 0.0002 max mem: 52150 +Epoch: [27] [ 440/7269] eta: 0:51:12 lr: 0.000003 loss: 0.5071 (0.5181) time: 0.4347 data: 0.0002 max mem: 52150 +Epoch: [27] [ 450/7269] eta: 0:51:14 lr: 0.000003 loss: 0.4977 (0.5180) time: 0.4345 data: 0.0001 max mem: 52150 +Epoch: [27] [ 460/7269] eta: 0:51:13 lr: 0.000003 loss: 0.4965 (0.5174) time: 0.4846 data: 0.0001 max mem: 52150 +Epoch: [27] [ 470/7269] eta: 0:51:16 lr: 0.000003 loss: 0.5349 (0.5179) time: 0.4868 data: 0.0001 max mem: 52150 +Epoch: [27] [ 480/7269] eta: 0:51:00 lr: 0.000003 loss: 0.4957 (0.5175) time: 0.4360 data: 0.0002 max mem: 52150 +Epoch: [27] [ 490/7269] eta: 0:50:57 lr: 0.000003 loss: 0.4957 (0.5180) time: 0.4188 data: 0.0002 max mem: 52150 +Epoch: [27] [ 500/7269] eta: 0:50:42 lr: 0.000003 loss: 0.5369 (0.5180) time: 0.4189 data: 0.0002 max mem: 52150 +Epoch: [27] [ 510/7269] eta: 0:50:27 lr: 0.000003 loss: 0.4995 (0.5179) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [27] [ 520/7269] eta: 0:50:25 lr: 0.000003 loss: 0.4995 (0.5180) time: 0.4164 data: 0.0002 max mem: 52150 +Epoch: [27] [ 530/7269] eta: 0:50:10 lr: 0.000003 loss: 0.5278 (0.5184) time: 0.4166 data: 0.0002 max mem: 52150 +Epoch: [27] [ 540/7269] eta: 0:49:57 lr: 0.000003 loss: 0.5575 (0.5191) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [27] [ 550/7269] eta: 0:49:43 lr: 0.000003 loss: 0.5472 (0.5186) time: 0.3712 data: 0.0002 max mem: 52150 +Epoch: [27] [ 560/7269] eta: 0:49:44 lr: 0.000003 loss: 0.5422 (0.5186) time: 0.4308 data: 0.0002 max mem: 52150 +Epoch: [27] [ 570/7269] eta: 0:49:31 lr: 0.000003 loss: 0.5678 (0.5192) time: 0.4308 data: 0.0002 max mem: 52150 +Epoch: [27] [ 580/7269] eta: 0:49:44 lr: 0.000003 loss: 0.5318 (0.5190) time: 0.4836 data: 0.0002 max mem: 52150 +Epoch: [27] [ 590/7269] eta: 0:49:43 lr: 0.000003 loss: 0.5309 (0.5192) time: 0.5337 data: 0.0001 max mem: 52150 +Epoch: [27] [ 600/7269] eta: 0:49:30 lr: 0.000003 loss: 0.4977 (0.5188) time: 0.4213 data: 0.0001 max mem: 52150 +Epoch: [27] [ 610/7269] eta: 0:49:17 lr: 0.000003 loss: 0.4990 (0.5188) time: 0.3711 data: 0.0002 max mem: 52150 +Epoch: [27] [ 620/7269] eta: 0:49:05 lr: 0.000003 loss: 0.5011 (0.5183) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [27] [ 630/7269] eta: 0:49:13 lr: 0.000003 loss: 0.5143 (0.5183) time: 0.4673 data: 0.0002 max mem: 52150 +Epoch: [27] [ 640/7269] eta: 0:49:01 lr: 0.000003 loss: 0.5143 (0.5182) time: 0.4669 data: 0.0002 max mem: 52150 +Epoch: [27] [ 650/7269] eta: 0:49:02 lr: 0.000003 loss: 0.5196 (0.5185) time: 0.4323 data: 0.0002 max mem: 52150 +Epoch: [27] [ 660/7269] eta: 0:48:50 lr: 0.000003 loss: 0.5507 (0.5183) time: 0.4320 data: 0.0002 max mem: 52150 +Epoch: [27] [ 670/7269] eta: 0:48:50 lr: 0.000003 loss: 0.5172 (0.5181) time: 0.4310 data: 0.0002 max mem: 52150 +Epoch: [27] [ 680/7269] eta: 0:48:39 lr: 0.000003 loss: 0.5203 (0.5182) time: 0.4318 data: 0.0002 max mem: 52150 +Epoch: [27] [ 690/7269] eta: 0:48:28 lr: 0.000003 loss: 0.5203 (0.5180) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [27] [ 700/7269] eta: 0:48:28 lr: 0.000003 loss: 0.5337 (0.5183) time: 0.4327 data: 0.0001 max mem: 52150 +Epoch: [27] [ 710/7269] eta: 0:48:28 lr: 0.000003 loss: 0.5392 (0.5182) time: 0.4932 data: 0.0002 max mem: 52150 +Epoch: [27] [ 720/7269] eta: 0:48:17 lr: 0.000003 loss: 0.5034 (0.5178) time: 0.4324 data: 0.0002 max mem: 52150 +Epoch: [27] [ 730/7269] eta: 0:48:06 lr: 0.000003 loss: 0.5429 (0.5184) time: 0.3713 data: 0.0001 max mem: 52150 +Epoch: [27] [ 740/7269] eta: 0:48:05 lr: 0.000003 loss: 0.5458 (0.5179) time: 0.4220 data: 0.0001 max mem: 52150 +Epoch: [27] [ 750/7269] eta: 0:47:54 lr: 0.000003 loss: 0.5128 (0.5180) time: 0.4226 data: 0.0002 max mem: 52150 +Epoch: [27] [ 760/7269] eta: 0:47:44 lr: 0.000003 loss: 0.5364 (0.5183) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [27] [ 770/7269] eta: 0:47:34 lr: 0.000003 loss: 0.5394 (0.5188) time: 0.3710 data: 0.0002 max mem: 52150 +Epoch: [27] [ 780/7269] eta: 0:47:34 lr: 0.000003 loss: 0.5120 (0.5181) time: 0.4318 data: 0.0002 max mem: 52150 +Epoch: [27] [ 790/7269] eta: 0:47:24 lr: 0.000003 loss: 0.5018 (0.5180) time: 0.4318 data: 0.0002 max mem: 52150 +Epoch: [27] [ 800/7269] eta: 0:47:14 lr: 0.000003 loss: 0.5091 (0.5176) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [27] [ 810/7269] eta: 0:47:14 lr: 0.000003 loss: 0.5199 (0.5176) time: 0.4321 data: 0.0002 max mem: 52150 +Epoch: [27] [ 820/7269] eta: 0:47:11 lr: 0.000003 loss: 0.5240 (0.5177) time: 0.4774 data: 0.0002 max mem: 52150 +Epoch: [27] [ 830/7269] eta: 0:47:02 lr: 0.000003 loss: 0.4888 (0.5172) time: 0.4172 data: 0.0002 max mem: 52150 +Epoch: [27] [ 840/7269] eta: 0:46:52 lr: 0.000003 loss: 0.4888 (0.5174) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [27] [ 850/7269] eta: 0:47:01 lr: 0.000003 loss: 0.5463 (0.5171) time: 0.4927 data: 0.0002 max mem: 52150 +Epoch: [27] [ 860/7269] eta: 0:46:52 lr: 0.000003 loss: 0.4900 (0.5168) time: 0.4920 data: 0.0002 max mem: 52150 +Epoch: [27] [ 870/7269] eta: 0:47:09 lr: 0.000003 loss: 0.5122 (0.5167) time: 0.5520 data: 0.0002 max mem: 52150 +Epoch: [27] [ 880/7269] eta: 0:46:59 lr: 0.000003 loss: 0.5238 (0.5166) time: 0.5524 data: 0.0002 max mem: 52150 +Epoch: [27] [ 890/7269] eta: 0:46:50 lr: 0.000003 loss: 0.5157 (0.5167) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [27] [ 900/7269] eta: 0:46:58 lr: 0.000003 loss: 0.5095 (0.5163) time: 0.4921 data: 0.0001 max mem: 52150 +Epoch: [27] [ 910/7269] eta: 0:46:55 lr: 0.000003 loss: 0.4554 (0.5160) time: 0.5377 data: 0.0002 max mem: 52150 +Epoch: [27] [ 920/7269] eta: 0:46:52 lr: 0.000003 loss: 0.4693 (0.5158) time: 0.4640 data: 0.0002 max mem: 52150 +Epoch: [27] [ 930/7269] eta: 0:46:51 lr: 0.000003 loss: 0.5207 (0.5160) time: 0.4776 data: 0.0002 max mem: 52150 +Epoch: [27] [ 940/7269] eta: 0:46:41 lr: 0.000003 loss: 0.5028 (0.5157) time: 0.4305 data: 0.0002 max mem: 52150 +Epoch: [27] [ 950/7269] eta: 0:46:32 lr: 0.000003 loss: 0.4867 (0.5155) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [27] [ 960/7269] eta: 0:46:31 lr: 0.000003 loss: 0.5264 (0.5156) time: 0.4319 data: 0.0002 max mem: 52150 +Epoch: [27] [ 970/7269] eta: 0:46:22 lr: 0.000003 loss: 0.5392 (0.5159) time: 0.4317 data: 0.0002 max mem: 52150 +Epoch: [27] [ 980/7269] eta: 0:46:21 lr: 0.000003 loss: 0.5132 (0.5156) time: 0.4322 data: 0.0002 max mem: 52150 +Epoch: [27] [ 990/7269] eta: 0:46:12 lr: 0.000003 loss: 0.5132 (0.5158) time: 0.4319 data: 0.0002 max mem: 52150 +Epoch: [27] [1000/7269] eta: 0:46:03 lr: 0.000003 loss: 0.4698 (0.5151) time: 0.3711 data: 0.0002 max mem: 52150 +Epoch: [27] [1010/7269] eta: 0:45:54 lr: 0.000003 loss: 0.4867 (0.5152) time: 0.3712 data: 0.0002 max mem: 52150 +Epoch: [27] [1020/7269] eta: 0:45:46 lr: 0.000003 loss: 0.5115 (0.5152) time: 0.3712 data: 0.0002 max mem: 52150 +Epoch: [27] [1030/7269] eta: 0:45:50 lr: 0.000003 loss: 0.5132 (0.5152) time: 0.4762 data: 0.0002 max mem: 52150 +Epoch: [27] [1040/7269] eta: 0:45:47 lr: 0.000003 loss: 0.5624 (0.5156) time: 0.5259 data: 0.0002 max mem: 52150 +Epoch: [27] [1050/7269] eta: 0:45:39 lr: 0.000003 loss: 0.5624 (0.5156) time: 0.4209 data: 0.0001 max mem: 52150 +Epoch: [27] [1060/7269] eta: 0:45:30 lr: 0.000003 loss: 0.5550 (0.5158) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [27] [1070/7269] eta: 0:45:22 lr: 0.000003 loss: 0.5102 (0.5154) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [27] [1080/7269] eta: 0:45:14 lr: 0.000003 loss: 0.5102 (0.5155) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [27] [1090/7269] eta: 0:45:05 lr: 0.000003 loss: 0.5487 (0.5158) time: 0.3709 data: 0.0002 max mem: 52150 +Epoch: [27] [1100/7269] eta: 0:45:02 lr: 0.000003 loss: 0.5573 (0.5159) time: 0.4159 data: 0.0002 max mem: 52150 +Epoch: [27] [1110/7269] eta: 0:44:54 lr: 0.000003 loss: 0.4735 (0.5154) time: 0.4162 data: 0.0001 max mem: 52150 +Epoch: [27] [1120/7269] eta: 0:44:46 lr: 0.000003 loss: 0.4973 (0.5154) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [27] [1130/7269] eta: 0:44:49 lr: 0.000003 loss: 0.5479 (0.5158) time: 0.4657 data: 0.0002 max mem: 52150 +Epoch: [27] [1140/7269] eta: 0:44:41 lr: 0.000003 loss: 0.5452 (0.5159) time: 0.4662 data: 0.0002 max mem: 52150 +Epoch: [27] [1150/7269] eta: 0:44:38 lr: 0.000003 loss: 0.5072 (0.5159) time: 0.4190 data: 0.0001 max mem: 52150 +Epoch: [27] [1160/7269] eta: 0:44:30 lr: 0.000003 loss: 0.5181 (0.5160) time: 0.4181 data: 0.0002 max mem: 52150 +Epoch: [27] [1170/7269] eta: 0:44:28 lr: 0.000003 loss: 0.5286 (0.5162) time: 0.4309 data: 0.0002 max mem: 52150 +Epoch: [27] [1180/7269] eta: 0:44:21 lr: 0.000003 loss: 0.5286 (0.5164) time: 0.4310 data: 0.0002 max mem: 52150 +Epoch: [27] [1190/7269] eta: 0:44:13 lr: 0.000003 loss: 0.5237 (0.5162) time: 0.3710 data: 0.0002 max mem: 52150 +Epoch: [27] [1200/7269] eta: 0:44:05 lr: 0.000003 loss: 0.5371 (0.5162) time: 0.3711 data: 0.0002 max mem: 52150 +Epoch: [27] [1210/7269] eta: 0:43:58 lr: 0.000003 loss: 0.5376 (0.5164) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [27] [1220/7269] eta: 0:43:50 lr: 0.000003 loss: 0.5363 (0.5163) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [27] [1230/7269] eta: 0:43:43 lr: 0.000003 loss: 0.5488 (0.5165) time: 0.3709 data: 0.0002 max mem: 52150 +Epoch: [27] [1240/7269] eta: 0:43:41 lr: 0.000003 loss: 0.5488 (0.5165) time: 0.4315 data: 0.0001 max mem: 52150 +Epoch: [27] [1250/7269] eta: 0:43:34 lr: 0.000003 loss: 0.5513 (0.5168) time: 0.4316 data: 0.0002 max mem: 52150 +Epoch: [27] [1260/7269] eta: 0:43:26 lr: 0.000003 loss: 0.5513 (0.5170) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [27] [1270/7269] eta: 0:43:25 lr: 0.000003 loss: 0.5131 (0.5171) time: 0.4313 data: 0.0002 max mem: 52150 +Epoch: [27] [1280/7269] eta: 0:43:36 lr: 0.000003 loss: 0.5196 (0.5172) time: 0.6354 data: 0.0002 max mem: 52150 +Epoch: [27] [1290/7269] eta: 0:43:29 lr: 0.000003 loss: 0.5165 (0.5171) time: 0.5755 data: 0.0002 max mem: 52150 +Epoch: [27] [1300/7269] eta: 0:43:22 lr: 0.000003 loss: 0.5101 (0.5171) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [27] [1310/7269] eta: 0:43:20 lr: 0.000003 loss: 0.5101 (0.5169) time: 0.4315 data: 0.0002 max mem: 52150 +Epoch: [27] [1320/7269] eta: 0:43:18 lr: 0.000003 loss: 0.5303 (0.5169) time: 0.4941 data: 0.0002 max mem: 52150 +Epoch: [27] [1330/7269] eta: 0:43:11 lr: 0.000003 loss: 0.5303 (0.5169) time: 0.4345 data: 0.0002 max mem: 52150 +Epoch: [27] [1340/7269] eta: 0:43:09 lr: 0.000003 loss: 0.4892 (0.5167) time: 0.4315 data: 0.0002 max mem: 52150 +Epoch: [27] [1350/7269] eta: 0:43:02 lr: 0.000003 loss: 0.4870 (0.5166) time: 0.4305 data: 0.0002 max mem: 52150 +Epoch: [27] [1360/7269] eta: 0:43:00 lr: 0.000003 loss: 0.5423 (0.5169) time: 0.4307 data: 0.0002 max mem: 52150 +Epoch: [27] [1370/7269] eta: 0:42:53 lr: 0.000003 loss: 0.5725 (0.5172) time: 0.4309 data: 0.0002 max mem: 52150 +Epoch: [27] [1380/7269] eta: 0:42:58 lr: 0.000003 loss: 0.5313 (0.5170) time: 0.5224 data: 0.0001 max mem: 52150 +Epoch: [27] [1390/7269] eta: 0:42:51 lr: 0.000003 loss: 0.5229 (0.5172) time: 0.5223 data: 0.0001 max mem: 52150 +Epoch: [27] [1400/7269] eta: 0:42:44 lr: 0.000003 loss: 0.5340 (0.5173) time: 0.3711 data: 0.0001 max mem: 52150 +Epoch: [27] [1410/7269] eta: 0:42:37 lr: 0.000003 loss: 0.5282 (0.5172) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [27] [1420/7269] eta: 0:42:30 lr: 0.000003 loss: 0.5258 (0.5173) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [27] [1430/7269] eta: 0:42:27 lr: 0.000003 loss: 0.5530 (0.5176) time: 0.4218 data: 0.0001 max mem: 52150 +Epoch: [27] [1440/7269] eta: 0:42:25 lr: 0.000003 loss: 0.5530 (0.5176) time: 0.4831 data: 0.0001 max mem: 52150 +Epoch: [27] [1450/7269] eta: 0:42:18 lr: 0.000003 loss: 0.5467 (0.5178) time: 0.4328 data: 0.0002 max mem: 52150 +Epoch: [27] [1460/7269] eta: 0:42:16 lr: 0.000003 loss: 0.5391 (0.5177) time: 0.4309 data: 0.0002 max mem: 52150 +Epoch: [27] [1470/7269] eta: 0:42:09 lr: 0.000003 loss: 0.5460 (0.5180) time: 0.4310 data: 0.0002 max mem: 52150 +Epoch: [27] [1480/7269] eta: 0:42:02 lr: 0.000003 loss: 0.5239 (0.5178) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [27] [1490/7269] eta: 0:41:55 lr: 0.000003 loss: 0.5172 (0.5178) time: 0.3711 data: 0.0002 max mem: 52150 +Epoch: [27] [1500/7269] eta: 0:41:58 lr: 0.000003 loss: 0.5329 (0.5178) time: 0.4922 data: 0.0002 max mem: 52150 +Epoch: [27] [1510/7269] eta: 0:41:55 lr: 0.000003 loss: 0.5234 (0.5176) time: 0.5554 data: 0.0002 max mem: 52150 +Epoch: [27] [1520/7269] eta: 0:41:49 lr: 0.000003 loss: 0.5021 (0.5175) time: 0.4347 data: 0.0002 max mem: 52150 +Epoch: [27] [1530/7269] eta: 0:41:42 lr: 0.000003 loss: 0.5141 (0.5175) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [27] [1540/7269] eta: 0:41:38 lr: 0.000003 loss: 0.5344 (0.5176) time: 0.4161 data: 0.0002 max mem: 52150 +Epoch: [27] [1550/7269] eta: 0:41:36 lr: 0.000003 loss: 0.4900 (0.5173) time: 0.4777 data: 0.0002 max mem: 52150 +Epoch: [27] [1560/7269] eta: 0:41:29 lr: 0.000003 loss: 0.4756 (0.5173) time: 0.4327 data: 0.0002 max mem: 52150 +Epoch: [27] [1570/7269] eta: 0:41:23 lr: 0.000003 loss: 0.5342 (0.5173) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [27] [1580/7269] eta: 0:41:16 lr: 0.000003 loss: 0.4992 (0.5173) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [27] [1590/7269] eta: 0:41:13 lr: 0.000003 loss: 0.5252 (0.5174) time: 0.4196 data: 0.0002 max mem: 52150 +Epoch: [27] [1600/7269] eta: 0:41:06 lr: 0.000003 loss: 0.5252 (0.5175) time: 0.4201 data: 0.0002 max mem: 52150 +Epoch: [27] [1610/7269] eta: 0:41:03 lr: 0.000003 loss: 0.5448 (0.5176) time: 0.4171 data: 0.0002 max mem: 52150 +Epoch: [27] [1620/7269] eta: 0:40:56 lr: 0.000003 loss: 0.5086 (0.5175) time: 0.4163 data: 0.0002 max mem: 52150 +Epoch: [27] [1630/7269] eta: 0:40:58 lr: 0.000003 loss: 0.5014 (0.5174) time: 0.4932 data: 0.0002 max mem: 52150 +Epoch: [27] [1640/7269] eta: 0:40:52 lr: 0.000003 loss: 0.5258 (0.5176) time: 0.4938 data: 0.0002 max mem: 52150 +Epoch: [27] [1650/7269] eta: 0:40:45 lr: 0.000003 loss: 0.5311 (0.5177) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [27] [1660/7269] eta: 0:40:39 lr: 0.000003 loss: 0.5352 (0.5178) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [27] [1670/7269] eta: 0:40:32 lr: 0.000003 loss: 0.5253 (0.5177) time: 0.3710 data: 0.0002 max mem: 52150 +Epoch: [27] [1680/7269] eta: 0:40:26 lr: 0.000003 loss: 0.5399 (0.5179) time: 0.3710 data: 0.0002 max mem: 52150 +Epoch: [27] [1690/7269] eta: 0:40:19 lr: 0.000003 loss: 0.5646 (0.5179) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [27] [1700/7269] eta: 0:40:13 lr: 0.000003 loss: 0.5364 (0.5179) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [27] [1710/7269] eta: 0:40:10 lr: 0.000003 loss: 0.5225 (0.5179) time: 0.4175 data: 0.0002 max mem: 52150 +Epoch: [27] [1720/7269] eta: 0:40:06 lr: 0.000003 loss: 0.5281 (0.5178) time: 0.4679 data: 0.0002 max mem: 52150 +Epoch: [27] [1730/7269] eta: 0:40:00 lr: 0.000003 loss: 0.5477 (0.5179) time: 0.4214 data: 0.0002 max mem: 52150 +Epoch: [27] [1740/7269] eta: 0:39:57 lr: 0.000003 loss: 0.5254 (0.5178) time: 0.4195 data: 0.0002 max mem: 52150 +Epoch: [27] [1750/7269] eta: 0:39:54 lr: 0.000003 loss: 0.5254 (0.5180) time: 0.4810 data: 0.0002 max mem: 52150 +Epoch: [27] [1760/7269] eta: 0:39:51 lr: 0.000003 loss: 0.5241 (0.5178) time: 0.4782 data: 0.0002 max mem: 52150 +Epoch: [27] [1770/7269] eta: 0:39:48 lr: 0.000003 loss: 0.5361 (0.5181) time: 0.4778 data: 0.0002 max mem: 52150 +Epoch: [27] [1780/7269] eta: 0:39:42 lr: 0.000003 loss: 0.5689 (0.5181) time: 0.4319 data: 0.0002 max mem: 52150 +Epoch: [27] [1790/7269] eta: 0:39:43 lr: 0.000003 loss: 0.5099 (0.5177) time: 0.4800 data: 0.0002 max mem: 52150 +Epoch: [27] [1800/7269] eta: 0:39:36 lr: 0.000003 loss: 0.4837 (0.5177) time: 0.4801 data: 0.0002 max mem: 52150 +Epoch: [27] [1810/7269] eta: 0:39:34 lr: 0.000003 loss: 0.5108 (0.5176) time: 0.4316 data: 0.0002 max mem: 52150 +Epoch: [27] [1820/7269] eta: 0:39:27 lr: 0.000003 loss: 0.4910 (0.5175) time: 0.4321 data: 0.0002 max mem: 52150 +Epoch: [27] [1830/7269] eta: 0:39:21 lr: 0.000003 loss: 0.4910 (0.5173) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [27] [1840/7269] eta: 0:39:19 lr: 0.000003 loss: 0.5031 (0.5172) time: 0.4322 data: 0.0002 max mem: 52150 +Epoch: [27] [1850/7269] eta: 0:39:15 lr: 0.000003 loss: 0.5034 (0.5172) time: 0.4781 data: 0.0002 max mem: 52150 +Epoch: [27] [1860/7269] eta: 0:39:09 lr: 0.000003 loss: 0.5130 (0.5170) time: 0.4172 data: 0.0002 max mem: 52150 +Epoch: [27] [1870/7269] eta: 0:39:06 lr: 0.000003 loss: 0.5130 (0.5169) time: 0.4324 data: 0.0002 max mem: 52150 +Epoch: [27] [1880/7269] eta: 0:39:00 lr: 0.000003 loss: 0.4767 (0.5169) time: 0.4328 data: 0.0002 max mem: 52150 +Epoch: [27] [1890/7269] eta: 0:39:01 lr: 0.000003 loss: 0.5056 (0.5168) time: 0.4923 data: 0.0002 max mem: 52150 +Epoch: [27] [1900/7269] eta: 0:38:55 lr: 0.000003 loss: 0.5332 (0.5170) time: 0.4924 data: 0.0002 max mem: 52150 +Epoch: [27] [1910/7269] eta: 0:38:49 lr: 0.000003 loss: 0.5332 (0.5169) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [27] [1920/7269] eta: 0:38:46 lr: 0.000003 loss: 0.5110 (0.5170) time: 0.4331 data: 0.0002 max mem: 52150 +Epoch: [27] [1930/7269] eta: 0:38:46 lr: 0.000003 loss: 0.5007 (0.5169) time: 0.5367 data: 0.0002 max mem: 52150 +Epoch: [27] [1940/7269] eta: 0:38:39 lr: 0.000003 loss: 0.4998 (0.5166) time: 0.4758 data: 0.0002 max mem: 52150 +Epoch: [27] [1950/7269] eta: 0:38:33 lr: 0.000003 loss: 0.5105 (0.5165) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [27] [1960/7269] eta: 0:38:27 lr: 0.000003 loss: 0.5259 (0.5165) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [27] [1970/7269] eta: 0:38:21 lr: 0.000003 loss: 0.5148 (0.5164) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [27] [1980/7269] eta: 0:38:15 lr: 0.000003 loss: 0.5006 (0.5163) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [27] [1990/7269] eta: 0:38:12 lr: 0.000003 loss: 0.5104 (0.5164) time: 0.4192 data: 0.0002 max mem: 52150 +Epoch: [27] [2000/7269] eta: 0:38:08 lr: 0.000003 loss: 0.5060 (0.5164) time: 0.4648 data: 0.0002 max mem: 52150 +Epoch: [27] [2010/7269] eta: 0:38:05 lr: 0.000003 loss: 0.4951 (0.5162) time: 0.4679 data: 0.0002 max mem: 52150 +Epoch: [27] [2020/7269] eta: 0:37:59 lr: 0.000003 loss: 0.4783 (0.5162) time: 0.4218 data: 0.0002 max mem: 52150 +Epoch: [27] [2030/7269] eta: 0:37:53 lr: 0.000003 loss: 0.5464 (0.5163) time: 0.3710 data: 0.0002 max mem: 52150 +Epoch: [27] [2040/7269] eta: 0:37:53 lr: 0.000003 loss: 0.5522 (0.5162) time: 0.4921 data: 0.0002 max mem: 52150 +Epoch: [27] [2050/7269] eta: 0:37:50 lr: 0.000003 loss: 0.5372 (0.5162) time: 0.5414 data: 0.0002 max mem: 52150 +Epoch: [27] [2060/7269] eta: 0:37:52 lr: 0.000003 loss: 0.5291 (0.5163) time: 0.5889 data: 0.0002 max mem: 52150 +Epoch: [27] [2070/7269] eta: 0:37:46 lr: 0.000003 loss: 0.5394 (0.5165) time: 0.5396 data: 0.0002 max mem: 52150 +Epoch: [27] [2080/7269] eta: 0:37:40 lr: 0.000003 loss: 0.5394 (0.5165) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [27] [2090/7269] eta: 0:37:37 lr: 0.000003 loss: 0.5258 (0.5165) time: 0.4331 data: 0.0002 max mem: 52150 +Epoch: [27] [2100/7269] eta: 0:37:31 lr: 0.000003 loss: 0.5151 (0.5165) time: 0.4338 data: 0.0002 max mem: 52150 +Epoch: [27] [2110/7269] eta: 0:37:26 lr: 0.000003 loss: 0.5151 (0.5165) time: 0.3728 data: 0.0002 max mem: 52150 +Epoch: [27] [2120/7269] eta: 0:37:20 lr: 0.000003 loss: 0.5274 (0.5166) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [27] [2130/7269] eta: 0:37:14 lr: 0.000003 loss: 0.5210 (0.5165) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [27] [2140/7269] eta: 0:37:08 lr: 0.000003 loss: 0.5235 (0.5166) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [27] [2150/7269] eta: 0:37:02 lr: 0.000003 loss: 0.5584 (0.5165) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [27] [2160/7269] eta: 0:36:56 lr: 0.000003 loss: 0.5222 (0.5166) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [27] [2170/7269] eta: 0:36:53 lr: 0.000003 loss: 0.5119 (0.5164) time: 0.4169 data: 0.0002 max mem: 52150 +Epoch: [27] [2180/7269] eta: 0:36:49 lr: 0.000003 loss: 0.5239 (0.5165) time: 0.4670 data: 0.0002 max mem: 52150 +Epoch: [27] [2190/7269] eta: 0:36:43 lr: 0.000003 loss: 0.5312 (0.5166) time: 0.4220 data: 0.0002 max mem: 52150 +Epoch: [27] [2200/7269] eta: 0:36:40 lr: 0.000003 loss: 0.5367 (0.5167) time: 0.4199 data: 0.0002 max mem: 52150 +Epoch: [27] [2210/7269] eta: 0:36:39 lr: 0.000003 loss: 0.5445 (0.5167) time: 0.5255 data: 0.0002 max mem: 52150 +Epoch: [27] [2220/7269] eta: 0:36:33 lr: 0.000003 loss: 0.5243 (0.5168) time: 0.4775 data: 0.0002 max mem: 52150 +Epoch: [27] [2230/7269] eta: 0:36:31 lr: 0.000003 loss: 0.5336 (0.5170) time: 0.4630 data: 0.0002 max mem: 52150 +Epoch: [27] [2240/7269] eta: 0:36:26 lr: 0.000003 loss: 0.5323 (0.5168) time: 0.4632 data: 0.0002 max mem: 52150 +Epoch: [27] [2250/7269] eta: 0:36:23 lr: 0.000003 loss: 0.5254 (0.5169) time: 0.4321 data: 0.0002 max mem: 52150 +Epoch: [27] [2260/7269] eta: 0:36:17 lr: 0.000003 loss: 0.5342 (0.5170) time: 0.4325 data: 0.0002 max mem: 52150 +Epoch: [27] [2270/7269] eta: 0:36:16 lr: 0.000003 loss: 0.4867 (0.5169) time: 0.4843 data: 0.0002 max mem: 52150 +Epoch: [27] [2280/7269] eta: 0:36:10 lr: 0.000003 loss: 0.4977 (0.5169) time: 0.4836 data: 0.0002 max mem: 52150 +Epoch: [27] [2290/7269] eta: 0:36:05 lr: 0.000003 loss: 0.5152 (0.5169) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [27] [2300/7269] eta: 0:35:59 lr: 0.000003 loss: 0.5097 (0.5168) time: 0.3711 data: 0.0002 max mem: 52150 +Epoch: [27] [2310/7269] eta: 0:35:58 lr: 0.000003 loss: 0.5146 (0.5168) time: 0.4949 data: 0.0002 max mem: 52150 +Epoch: [27] [2320/7269] eta: 0:35:53 lr: 0.000003 loss: 0.5026 (0.5167) time: 0.4951 data: 0.0002 max mem: 52150 +Epoch: [27] [2330/7269] eta: 0:35:47 lr: 0.000003 loss: 0.5187 (0.5168) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [27] [2340/7269] eta: 0:35:41 lr: 0.000003 loss: 0.5136 (0.5167) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [27] [2350/7269] eta: 0:35:36 lr: 0.000003 loss: 0.5136 (0.5167) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [27] [2360/7269] eta: 0:35:30 lr: 0.000003 loss: 0.5243 (0.5167) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [27] [2370/7269] eta: 0:35:24 lr: 0.000003 loss: 0.5414 (0.5169) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [27] [2380/7269] eta: 0:35:19 lr: 0.000003 loss: 0.5393 (0.5169) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [27] [2390/7269] eta: 0:35:13 lr: 0.000003 loss: 0.5496 (0.5171) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [27] [2400/7269] eta: 0:35:10 lr: 0.000003 loss: 0.5570 (0.5170) time: 0.4170 data: 0.0002 max mem: 52150 +Epoch: [27] [2410/7269] eta: 0:35:04 lr: 0.000003 loss: 0.5250 (0.5170) time: 0.4168 data: 0.0002 max mem: 52150 +Epoch: [27] [2420/7269] eta: 0:34:58 lr: 0.000003 loss: 0.5284 (0.5171) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [27] [2430/7269] eta: 0:34:53 lr: 0.000003 loss: 0.5531 (0.5171) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [27] [2440/7269] eta: 0:34:47 lr: 0.000003 loss: 0.5403 (0.5172) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [27] [2450/7269] eta: 0:34:44 lr: 0.000003 loss: 0.5391 (0.5172) time: 0.4309 data: 0.0002 max mem: 52150 +Epoch: [27] [2460/7269] eta: 0:34:40 lr: 0.000003 loss: 0.5247 (0.5172) time: 0.4759 data: 0.0002 max mem: 52150 +Epoch: [27] [2470/7269] eta: 0:34:35 lr: 0.000003 loss: 0.5241 (0.5171) time: 0.4169 data: 0.0002 max mem: 52150 +Epoch: [27] [2480/7269] eta: 0:34:29 lr: 0.000003 loss: 0.4960 (0.5171) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [27] [2490/7269] eta: 0:34:26 lr: 0.000003 loss: 0.4960 (0.5171) time: 0.4173 data: 0.0002 max mem: 52150 +Epoch: [27] [2500/7269] eta: 0:34:20 lr: 0.000003 loss: 0.5165 (0.5170) time: 0.4170 data: 0.0002 max mem: 52150 +Epoch: [27] [2510/7269] eta: 0:34:21 lr: 0.000003 loss: 0.5213 (0.5170) time: 0.5314 data: 0.0002 max mem: 52150 +Epoch: [27] [2520/7269] eta: 0:34:19 lr: 0.000003 loss: 0.5462 (0.5171) time: 0.6391 data: 0.0002 max mem: 52150 +Epoch: [27] [2530/7269] eta: 0:34:16 lr: 0.000003 loss: 0.5462 (0.5171) time: 0.5396 data: 0.0002 max mem: 52150 +Epoch: [27] [2540/7269] eta: 0:34:12 lr: 0.000003 loss: 0.5744 (0.5175) time: 0.4781 data: 0.0002 max mem: 52150 +Epoch: [27] [2550/7269] eta: 0:34:09 lr: 0.000003 loss: 0.5670 (0.5174) time: 0.4656 data: 0.0002 max mem: 52150 +Epoch: [27] [2560/7269] eta: 0:34:03 lr: 0.000003 loss: 0.5330 (0.5174) time: 0.4202 data: 0.0002 max mem: 52150 +Epoch: [27] [2570/7269] eta: 0:33:58 lr: 0.000003 loss: 0.5445 (0.5174) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [27] [2580/7269] eta: 0:33:54 lr: 0.000003 loss: 0.5445 (0.5175) time: 0.4216 data: 0.0002 max mem: 52150 +Epoch: [27] [2590/7269] eta: 0:33:49 lr: 0.000003 loss: 0.5314 (0.5175) time: 0.4218 data: 0.0002 max mem: 52150 +Epoch: [27] [2600/7269] eta: 0:33:45 lr: 0.000003 loss: 0.5213 (0.5175) time: 0.4171 data: 0.0002 max mem: 52150 +Epoch: [27] [2610/7269] eta: 0:33:39 lr: 0.000003 loss: 0.5399 (0.5176) time: 0.4169 data: 0.0002 max mem: 52150 +Epoch: [27] [2620/7269] eta: 0:33:34 lr: 0.000003 loss: 0.5583 (0.5176) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [27] [2630/7269] eta: 0:33:28 lr: 0.000003 loss: 0.5145 (0.5175) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [27] [2640/7269] eta: 0:33:23 lr: 0.000003 loss: 0.5041 (0.5174) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [27] [2650/7269] eta: 0:33:18 lr: 0.000003 loss: 0.5006 (0.5173) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [27] [2660/7269] eta: 0:33:12 lr: 0.000003 loss: 0.5084 (0.5173) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [27] [2670/7269] eta: 0:33:10 lr: 0.000003 loss: 0.5152 (0.5173) time: 0.4711 data: 0.0002 max mem: 52150 +Epoch: [27] [2680/7269] eta: 0:33:05 lr: 0.000003 loss: 0.5290 (0.5174) time: 0.4712 data: 0.0002 max mem: 52150 +Epoch: [27] [2690/7269] eta: 0:33:00 lr: 0.000003 loss: 0.5589 (0.5174) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [27] [2700/7269] eta: 0:32:56 lr: 0.000003 loss: 0.5192 (0.5173) time: 0.4169 data: 0.0002 max mem: 52150 +Epoch: [27] [2710/7269] eta: 0:32:50 lr: 0.000003 loss: 0.5126 (0.5173) time: 0.4166 data: 0.0002 max mem: 52150 +Epoch: [27] [2720/7269] eta: 0:32:49 lr: 0.000003 loss: 0.5212 (0.5173) time: 0.4857 data: 0.0002 max mem: 52150 +Epoch: [27] [2730/7269] eta: 0:32:47 lr: 0.000003 loss: 0.5210 (0.5172) time: 0.5814 data: 0.0002 max mem: 52150 +Epoch: [27] [2740/7269] eta: 0:32:43 lr: 0.000003 loss: 0.5405 (0.5173) time: 0.5161 data: 0.0002 max mem: 52150 +Epoch: [27] [2750/7269] eta: 0:32:38 lr: 0.000003 loss: 0.5705 (0.5174) time: 0.4206 data: 0.0002 max mem: 52150 +Epoch: [27] [2760/7269] eta: 0:32:32 lr: 0.000003 loss: 0.4836 (0.5173) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [27] [2770/7269] eta: 0:32:29 lr: 0.000003 loss: 0.4706 (0.5173) time: 0.4321 data: 0.0002 max mem: 52150 +Epoch: [27] [2780/7269] eta: 0:32:27 lr: 0.000003 loss: 0.4980 (0.5172) time: 0.5527 data: 0.0002 max mem: 52150 +Epoch: [27] [2790/7269] eta: 0:32:22 lr: 0.000003 loss: 0.4682 (0.5170) time: 0.4922 data: 0.0002 max mem: 52150 +Epoch: [27] [2800/7269] eta: 0:32:17 lr: 0.000003 loss: 0.5135 (0.5170) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [27] [2810/7269] eta: 0:32:11 lr: 0.000003 loss: 0.5135 (0.5170) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [27] [2820/7269] eta: 0:32:06 lr: 0.000003 loss: 0.5074 (0.5170) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [27] [2830/7269] eta: 0:32:03 lr: 0.000003 loss: 0.5155 (0.5169) time: 0.4313 data: 0.0002 max mem: 52150 +Epoch: [27] [2840/7269] eta: 0:31:57 lr: 0.000003 loss: 0.5048 (0.5169) time: 0.4313 data: 0.0002 max mem: 52150 +Epoch: [27] [2850/7269] eta: 0:31:52 lr: 0.000003 loss: 0.5331 (0.5170) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [27] [2860/7269] eta: 0:31:47 lr: 0.000003 loss: 0.5347 (0.5171) time: 0.3711 data: 0.0002 max mem: 52150 +Epoch: [27] [2870/7269] eta: 0:31:42 lr: 0.000003 loss: 0.5247 (0.5170) time: 0.3709 data: 0.0002 max mem: 52150 +Epoch: [27] [2880/7269] eta: 0:31:36 lr: 0.000003 loss: 0.4853 (0.5170) time: 0.3710 data: 0.0002 max mem: 52150 +Epoch: [27] [2890/7269] eta: 0:31:32 lr: 0.000003 loss: 0.5138 (0.5171) time: 0.4161 data: 0.0002 max mem: 52150 +Epoch: [27] [2900/7269] eta: 0:31:27 lr: 0.000003 loss: 0.5363 (0.5171) time: 0.4166 data: 0.0002 max mem: 52150 +Epoch: [27] [2910/7269] eta: 0:31:23 lr: 0.000003 loss: 0.4989 (0.5171) time: 0.4214 data: 0.0002 max mem: 52150 +Epoch: [27] [2920/7269] eta: 0:31:20 lr: 0.000003 loss: 0.4989 (0.5170) time: 0.4654 data: 0.0002 max mem: 52150 +Epoch: [27] [2930/7269] eta: 0:31:14 lr: 0.000003 loss: 0.4626 (0.5169) time: 0.4161 data: 0.0002 max mem: 52150 +Epoch: [27] [2940/7269] eta: 0:31:11 lr: 0.000003 loss: 0.4875 (0.5168) time: 0.4329 data: 0.0002 max mem: 52150 +Epoch: [27] [2950/7269] eta: 0:31:08 lr: 0.000003 loss: 0.5292 (0.5170) time: 0.5245 data: 0.0002 max mem: 52150 +Epoch: [27] [2960/7269] eta: 0:31:03 lr: 0.000003 loss: 0.5394 (0.5170) time: 0.4640 data: 0.0002 max mem: 52150 +Epoch: [27] [2970/7269] eta: 0:30:58 lr: 0.000003 loss: 0.5422 (0.5171) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [27] [2980/7269] eta: 0:30:55 lr: 0.000003 loss: 0.5357 (0.5171) time: 0.4350 data: 0.0002 max mem: 52150 +Epoch: [27] [2990/7269] eta: 0:30:51 lr: 0.000003 loss: 0.5031 (0.5170) time: 0.4988 data: 0.0002 max mem: 52150 +Epoch: [27] [3000/7269] eta: 0:30:46 lr: 0.000003 loss: 0.4876 (0.5170) time: 0.4360 data: 0.0002 max mem: 52150 +Epoch: [27] [3010/7269] eta: 0:30:41 lr: 0.000003 loss: 0.5460 (0.5171) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [27] [3020/7269] eta: 0:30:38 lr: 0.000003 loss: 0.5609 (0.5171) time: 0.4688 data: 0.0002 max mem: 52150 +Epoch: [27] [3030/7269] eta: 0:30:33 lr: 0.000003 loss: 0.5323 (0.5171) time: 0.4691 data: 0.0002 max mem: 52150 +Epoch: [27] [3040/7269] eta: 0:30:30 lr: 0.000003 loss: 0.5323 (0.5171) time: 0.4223 data: 0.0002 max mem: 52150 +Epoch: [27] [3050/7269] eta: 0:30:24 lr: 0.000003 loss: 0.5501 (0.5172) time: 0.4218 data: 0.0002 max mem: 52150 +Epoch: [27] [3060/7269] eta: 0:30:19 lr: 0.000003 loss: 0.5606 (0.5172) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [27] [3070/7269] eta: 0:30:16 lr: 0.000003 loss: 0.4928 (0.5172) time: 0.4317 data: 0.0002 max mem: 52150 +Epoch: [27] [3080/7269] eta: 0:30:11 lr: 0.000003 loss: 0.4928 (0.5171) time: 0.4314 data: 0.0002 max mem: 52150 +Epoch: [27] [3090/7269] eta: 0:30:07 lr: 0.000003 loss: 0.5437 (0.5172) time: 0.4310 data: 0.0002 max mem: 52150 +Epoch: [27] [3100/7269] eta: 0:30:03 lr: 0.000003 loss: 0.5328 (0.5171) time: 0.4769 data: 0.0002 max mem: 52150 +Epoch: [27] [3110/7269] eta: 0:29:58 lr: 0.000003 loss: 0.4737 (0.5171) time: 0.4174 data: 0.0002 max mem: 52150 +Epoch: [27] [3120/7269] eta: 0:29:53 lr: 0.000003 loss: 0.5242 (0.5171) time: 0.3727 data: 0.0002 max mem: 52150 +Epoch: [27] [3130/7269] eta: 0:29:49 lr: 0.000003 loss: 0.4883 (0.5170) time: 0.4210 data: 0.0002 max mem: 52150 +Epoch: [27] [3140/7269] eta: 0:29:44 lr: 0.000003 loss: 0.4883 (0.5170) time: 0.4204 data: 0.0002 max mem: 52150 +Epoch: [27] [3150/7269] eta: 0:29:39 lr: 0.000003 loss: 0.5265 (0.5170) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [27] [3160/7269] eta: 0:29:35 lr: 0.000003 loss: 0.5442 (0.5171) time: 0.4223 data: 0.0002 max mem: 52150 +Epoch: [27] [3170/7269] eta: 0:29:30 lr: 0.000003 loss: 0.5071 (0.5170) time: 0.4228 data: 0.0002 max mem: 52150 +Epoch: [27] [3180/7269] eta: 0:29:25 lr: 0.000003 loss: 0.5255 (0.5171) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [27] [3190/7269] eta: 0:29:22 lr: 0.000003 loss: 0.5519 (0.5172) time: 0.4664 data: 0.0002 max mem: 52150 +Epoch: [27] [3200/7269] eta: 0:29:19 lr: 0.000003 loss: 0.5381 (0.5172) time: 0.5302 data: 0.0002 max mem: 52150 +Epoch: [27] [3210/7269] eta: 0:29:14 lr: 0.000003 loss: 0.5373 (0.5172) time: 0.4355 data: 0.0002 max mem: 52150 +Epoch: [27] [3220/7269] eta: 0:29:08 lr: 0.000003 loss: 0.5607 (0.5173) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [27] [3230/7269] eta: 0:29:03 lr: 0.000003 loss: 0.5234 (0.5172) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [27] [3240/7269] eta: 0:29:00 lr: 0.000003 loss: 0.5234 (0.5173) time: 0.4322 data: 0.0002 max mem: 52150 +Epoch: [27] [3250/7269] eta: 0:28:56 lr: 0.000003 loss: 0.5306 (0.5173) time: 0.4831 data: 0.0002 max mem: 52150 +Epoch: [27] [3260/7269] eta: 0:28:54 lr: 0.000003 loss: 0.5083 (0.5173) time: 0.5305 data: 0.0002 max mem: 52150 +Epoch: [27] [3270/7269] eta: 0:28:49 lr: 0.000003 loss: 0.5224 (0.5173) time: 0.4803 data: 0.0002 max mem: 52150 +Epoch: [27] [3280/7269] eta: 0:28:45 lr: 0.000003 loss: 0.5522 (0.5174) time: 0.4206 data: 0.0002 max mem: 52150 +Epoch: [27] [3290/7269] eta: 0:28:40 lr: 0.000003 loss: 0.5241 (0.5172) time: 0.4200 data: 0.0002 max mem: 52150 +Epoch: [27] [3300/7269] eta: 0:28:37 lr: 0.000003 loss: 0.5241 (0.5173) time: 0.4909 data: 0.0002 max mem: 52150 +Epoch: [27] [3310/7269] eta: 0:28:34 lr: 0.000003 loss: 0.5557 (0.5173) time: 0.5518 data: 0.0002 max mem: 52150 +Epoch: [27] [3320/7269] eta: 0:28:29 lr: 0.000003 loss: 0.5161 (0.5173) time: 0.4324 data: 0.0002 max mem: 52150 +Epoch: [27] [3330/7269] eta: 0:28:24 lr: 0.000003 loss: 0.5294 (0.5174) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [27] [3340/7269] eta: 0:28:20 lr: 0.000003 loss: 0.5282 (0.5173) time: 0.4178 data: 0.0002 max mem: 52150 +Epoch: [27] [3350/7269] eta: 0:28:16 lr: 0.000003 loss: 0.4826 (0.5172) time: 0.4781 data: 0.0002 max mem: 52150 +Epoch: [27] [3360/7269] eta: 0:28:11 lr: 0.000003 loss: 0.4826 (0.5171) time: 0.4319 data: 0.0002 max mem: 52150 +Epoch: [27] [3370/7269] eta: 0:28:06 lr: 0.000003 loss: 0.4773 (0.5170) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [27] [3380/7269] eta: 0:28:02 lr: 0.000003 loss: 0.4902 (0.5169) time: 0.4199 data: 0.0002 max mem: 52150 +Epoch: [27] [3390/7269] eta: 0:27:58 lr: 0.000003 loss: 0.4829 (0.5169) time: 0.4670 data: 0.0002 max mem: 52150 +Epoch: [27] [3400/7269] eta: 0:27:53 lr: 0.000003 loss: 0.4972 (0.5169) time: 0.4188 data: 0.0002 max mem: 52150 +Epoch: [27] [3410/7269] eta: 0:27:51 lr: 0.000003 loss: 0.4982 (0.5168) time: 0.4803 data: 0.0002 max mem: 52150 +Epoch: [27] [3420/7269] eta: 0:27:46 lr: 0.000003 loss: 0.4982 (0.5168) time: 0.4810 data: 0.0002 max mem: 52150 +Epoch: [27] [3430/7269] eta: 0:27:41 lr: 0.000003 loss: 0.5087 (0.5168) time: 0.3727 data: 0.0002 max mem: 52150 +Epoch: [27] [3440/7269] eta: 0:27:37 lr: 0.000003 loss: 0.4927 (0.5166) time: 0.4188 data: 0.0002 max mem: 52150 +Epoch: [27] [3450/7269] eta: 0:27:34 lr: 0.000003 loss: 0.4746 (0.5166) time: 0.5249 data: 0.0002 max mem: 52150 +Epoch: [27] [3460/7269] eta: 0:27:29 lr: 0.000003 loss: 0.5090 (0.5166) time: 0.4779 data: 0.0002 max mem: 52150 +Epoch: [27] [3470/7269] eta: 0:27:24 lr: 0.000003 loss: 0.5506 (0.5168) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [27] [3480/7269] eta: 0:27:19 lr: 0.000003 loss: 0.5692 (0.5169) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [27] [3490/7269] eta: 0:27:14 lr: 0.000003 loss: 0.5563 (0.5170) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [27] [3500/7269] eta: 0:27:09 lr: 0.000003 loss: 0.5313 (0.5170) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [27] [3510/7269] eta: 0:27:05 lr: 0.000003 loss: 0.4999 (0.5169) time: 0.4214 data: 0.0002 max mem: 52150 +Epoch: [27] [3520/7269] eta: 0:27:01 lr: 0.000003 loss: 0.4881 (0.5168) time: 0.4817 data: 0.0002 max mem: 52150 +Epoch: [27] [3530/7269] eta: 0:26:56 lr: 0.000003 loss: 0.5163 (0.5169) time: 0.4321 data: 0.0002 max mem: 52150 +Epoch: [27] [3540/7269] eta: 0:26:53 lr: 0.000003 loss: 0.5072 (0.5168) time: 0.4193 data: 0.0002 max mem: 52150 +Epoch: [27] [3550/7269] eta: 0:26:48 lr: 0.000003 loss: 0.5393 (0.5168) time: 0.4194 data: 0.0002 max mem: 52150 +Epoch: [27] [3560/7269] eta: 0:26:43 lr: 0.000003 loss: 0.5426 (0.5168) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [27] [3570/7269] eta: 0:26:38 lr: 0.000003 loss: 0.5139 (0.5167) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [27] [3580/7269] eta: 0:26:33 lr: 0.000003 loss: 0.5211 (0.5168) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [27] [3590/7269] eta: 0:26:28 lr: 0.000003 loss: 0.5212 (0.5167) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [27] [3600/7269] eta: 0:26:23 lr: 0.000003 loss: 0.5218 (0.5168) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [27] [3610/7269] eta: 0:26:19 lr: 0.000003 loss: 0.5342 (0.5167) time: 0.4208 data: 0.0002 max mem: 52150 +Epoch: [27] [3620/7269] eta: 0:26:15 lr: 0.000003 loss: 0.5383 (0.5167) time: 0.4691 data: 0.0002 max mem: 52150 +Epoch: [27] [3630/7269] eta: 0:26:11 lr: 0.000003 loss: 0.5538 (0.5168) time: 0.4649 data: 0.0002 max mem: 52150 +Epoch: [27] [3640/7269] eta: 0:26:06 lr: 0.000003 loss: 0.5395 (0.5168) time: 0.4166 data: 0.0002 max mem: 52150 +Epoch: [27] [3650/7269] eta: 0:26:01 lr: 0.000003 loss: 0.5568 (0.5170) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [27] [3660/7269] eta: 0:25:56 lr: 0.000003 loss: 0.5766 (0.5171) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [27] [3670/7269] eta: 0:25:51 lr: 0.000003 loss: 0.5529 (0.5170) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [27] [3680/7269] eta: 0:25:46 lr: 0.000003 loss: 0.4701 (0.5169) time: 0.3712 data: 0.0002 max mem: 52150 +Epoch: [27] [3690/7269] eta: 0:25:42 lr: 0.000003 loss: 0.5110 (0.5170) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [27] [3700/7269] eta: 0:25:38 lr: 0.000003 loss: 0.5354 (0.5170) time: 0.4196 data: 0.0002 max mem: 52150 +Epoch: [27] [3710/7269] eta: 0:25:33 lr: 0.000003 loss: 0.5144 (0.5169) time: 0.4199 data: 0.0002 max mem: 52150 +Epoch: [27] [3720/7269] eta: 0:25:30 lr: 0.000003 loss: 0.5067 (0.5169) time: 0.4782 data: 0.0002 max mem: 52150 +Epoch: [27] [3730/7269] eta: 0:25:25 lr: 0.000003 loss: 0.5319 (0.5170) time: 0.4777 data: 0.0002 max mem: 52150 +Epoch: [27] [3740/7269] eta: 0:25:21 lr: 0.000003 loss: 0.5275 (0.5170) time: 0.4194 data: 0.0002 max mem: 52150 +Epoch: [27] [3750/7269] eta: 0:25:16 lr: 0.000003 loss: 0.4988 (0.5169) time: 0.4192 data: 0.0002 max mem: 52150 +Epoch: [27] [3760/7269] eta: 0:25:11 lr: 0.000003 loss: 0.5196 (0.5170) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [27] [3770/7269] eta: 0:25:10 lr: 0.000003 loss: 0.5269 (0.5170) time: 0.5512 data: 0.0002 max mem: 52150 +Epoch: [27] [3780/7269] eta: 0:25:05 lr: 0.000003 loss: 0.5028 (0.5169) time: 0.5511 data: 0.0002 max mem: 52150 +Epoch: [27] [3790/7269] eta: 0:25:00 lr: 0.000003 loss: 0.4667 (0.5167) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [27] [3800/7269] eta: 0:24:56 lr: 0.000003 loss: 0.4797 (0.5167) time: 0.4177 data: 0.0002 max mem: 52150 +Epoch: [27] [3810/7269] eta: 0:24:51 lr: 0.000003 loss: 0.5160 (0.5167) time: 0.4175 data: 0.0002 max mem: 52150 +Epoch: [27] [3820/7269] eta: 0:24:47 lr: 0.000003 loss: 0.5477 (0.5167) time: 0.4312 data: 0.0002 max mem: 52150 +Epoch: [27] [3830/7269] eta: 0:24:43 lr: 0.000003 loss: 0.5455 (0.5168) time: 0.4807 data: 0.0002 max mem: 52150 +Epoch: [27] [3840/7269] eta: 0:24:39 lr: 0.000003 loss: 0.5436 (0.5167) time: 0.4206 data: 0.0002 max mem: 52150 +Epoch: [27] [3850/7269] eta: 0:24:35 lr: 0.000003 loss: 0.5298 (0.5167) time: 0.4324 data: 0.0002 max mem: 52150 +Epoch: [27] [3860/7269] eta: 0:24:30 lr: 0.000003 loss: 0.5115 (0.5167) time: 0.4326 data: 0.0002 max mem: 52150 +Epoch: [27] [3870/7269] eta: 0:24:26 lr: 0.000003 loss: 0.5049 (0.5167) time: 0.4337 data: 0.0002 max mem: 52150 +Epoch: [27] [3880/7269] eta: 0:24:23 lr: 0.000003 loss: 0.5355 (0.5168) time: 0.5521 data: 0.0002 max mem: 52150 +Epoch: [27] [3890/7269] eta: 0:24:19 lr: 0.000003 loss: 0.5228 (0.5167) time: 0.4899 data: 0.0002 max mem: 52150 +Epoch: [27] [3900/7269] eta: 0:24:15 lr: 0.000003 loss: 0.5088 (0.5168) time: 0.4180 data: 0.0002 max mem: 52150 +Epoch: [27] [3910/7269] eta: 0:24:10 lr: 0.000003 loss: 0.5482 (0.5168) time: 0.4179 data: 0.0002 max mem: 52150 +Epoch: [27] [3920/7269] eta: 0:24:06 lr: 0.000003 loss: 0.5491 (0.5169) time: 0.4168 data: 0.0002 max mem: 52150 +Epoch: [27] [3930/7269] eta: 0:24:02 lr: 0.000003 loss: 0.5491 (0.5169) time: 0.4645 data: 0.0002 max mem: 52150 +Epoch: [27] [3940/7269] eta: 0:23:57 lr: 0.000003 loss: 0.5541 (0.5170) time: 0.4187 data: 0.0002 max mem: 52150 +Epoch: [27] [3950/7269] eta: 0:23:52 lr: 0.000003 loss: 0.5682 (0.5171) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [27] [3960/7269] eta: 0:23:47 lr: 0.000003 loss: 0.5324 (0.5169) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [27] [3970/7269] eta: 0:23:45 lr: 0.000003 loss: 0.4773 (0.5170) time: 0.5428 data: 0.0002 max mem: 52150 +Epoch: [27] [3980/7269] eta: 0:23:41 lr: 0.000003 loss: 0.5222 (0.5169) time: 0.5914 data: 0.0002 max mem: 52150 +Epoch: [27] [3990/7269] eta: 0:23:37 lr: 0.000003 loss: 0.5366 (0.5169) time: 0.4823 data: 0.0002 max mem: 52150 +Epoch: [27] [4000/7269] eta: 0:23:34 lr: 0.000003 loss: 0.5343 (0.5169) time: 0.4973 data: 0.0002 max mem: 52150 +Epoch: [27] [4010/7269] eta: 0:23:29 lr: 0.000003 loss: 0.5100 (0.5169) time: 0.4353 data: 0.0002 max mem: 52150 +Epoch: [27] [4020/7269] eta: 0:23:24 lr: 0.000003 loss: 0.5082 (0.5169) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [27] [4030/7269] eta: 0:23:19 lr: 0.000003 loss: 0.5450 (0.5169) time: 0.3711 data: 0.0002 max mem: 52150 +Epoch: [27] [4040/7269] eta: 0:23:14 lr: 0.000003 loss: 0.5405 (0.5170) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [27] [4050/7269] eta: 0:23:10 lr: 0.000003 loss: 0.5405 (0.5169) time: 0.4212 data: 0.0002 max mem: 52150 +Epoch: [27] [4060/7269] eta: 0:23:06 lr: 0.000003 loss: 0.5293 (0.5169) time: 0.4704 data: 0.0002 max mem: 52150 +Epoch: [27] [4070/7269] eta: 0:23:02 lr: 0.000003 loss: 0.5296 (0.5169) time: 0.4671 data: 0.0002 max mem: 52150 +Epoch: [27] [4080/7269] eta: 0:22:58 lr: 0.000003 loss: 0.5352 (0.5170) time: 0.4646 data: 0.0002 max mem: 52150 +Epoch: [27] [4090/7269] eta: 0:22:53 lr: 0.000003 loss: 0.5170 (0.5170) time: 0.4184 data: 0.0002 max mem: 52150 +Epoch: [27] [4100/7269] eta: 0:22:49 lr: 0.000003 loss: 0.5043 (0.5170) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [27] [4110/7269] eta: 0:22:44 lr: 0.000003 loss: 0.4926 (0.5169) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [27] [4120/7269] eta: 0:22:40 lr: 0.000003 loss: 0.4928 (0.5168) time: 0.4335 data: 0.0002 max mem: 52150 +Epoch: [27] [4130/7269] eta: 0:22:35 lr: 0.000003 loss: 0.5103 (0.5168) time: 0.4332 data: 0.0002 max mem: 52150 +Epoch: [27] [4140/7269] eta: 0:22:30 lr: 0.000003 loss: 0.4972 (0.5167) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [27] [4150/7269] eta: 0:22:26 lr: 0.000003 loss: 0.5210 (0.5168) time: 0.4204 data: 0.0002 max mem: 52150 +Epoch: [27] [4160/7269] eta: 0:22:22 lr: 0.000003 loss: 0.5159 (0.5167) time: 0.4207 data: 0.0002 max mem: 52150 +Epoch: [27] [4170/7269] eta: 0:22:17 lr: 0.000003 loss: 0.4762 (0.5166) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [27] [4180/7269] eta: 0:22:12 lr: 0.000003 loss: 0.5048 (0.5167) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [27] [4190/7269] eta: 0:22:07 lr: 0.000003 loss: 0.4880 (0.5166) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [27] [4200/7269] eta: 0:22:03 lr: 0.000003 loss: 0.4806 (0.5166) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [27] [4210/7269] eta: 0:21:58 lr: 0.000003 loss: 0.5048 (0.5166) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [27] [4220/7269] eta: 0:21:53 lr: 0.000003 loss: 0.5201 (0.5166) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [27] [4230/7269] eta: 0:21:49 lr: 0.000003 loss: 0.5375 (0.5166) time: 0.4200 data: 0.0002 max mem: 52150 +Epoch: [27] [4240/7269] eta: 0:21:44 lr: 0.000003 loss: 0.5201 (0.5165) time: 0.4202 data: 0.0002 max mem: 52150 +Epoch: [27] [4250/7269] eta: 0:21:40 lr: 0.000003 loss: 0.5334 (0.5166) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [27] [4260/7269] eta: 0:21:35 lr: 0.000003 loss: 0.5334 (0.5166) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [27] [4270/7269] eta: 0:21:31 lr: 0.000003 loss: 0.5148 (0.5166) time: 0.4187 data: 0.0002 max mem: 52150 +Epoch: [27] [4280/7269] eta: 0:21:26 lr: 0.000003 loss: 0.5236 (0.5165) time: 0.4186 data: 0.0002 max mem: 52150 +Epoch: [27] [4290/7269] eta: 0:21:22 lr: 0.000003 loss: 0.5236 (0.5165) time: 0.4311 data: 0.0002 max mem: 52150 +Epoch: [27] [4300/7269] eta: 0:21:18 lr: 0.000003 loss: 0.5172 (0.5165) time: 0.4310 data: 0.0002 max mem: 52150 +Epoch: [27] [4310/7269] eta: 0:21:13 lr: 0.000003 loss: 0.5290 (0.5166) time: 0.4162 data: 0.0002 max mem: 52150 +Epoch: [27] [4320/7269] eta: 0:21:10 lr: 0.000003 loss: 0.4928 (0.5165) time: 0.4779 data: 0.0002 max mem: 52150 +Epoch: [27] [4330/7269] eta: 0:21:05 lr: 0.000003 loss: 0.4928 (0.5165) time: 0.4334 data: 0.0002 max mem: 52150 +Epoch: [27] [4340/7269] eta: 0:21:02 lr: 0.000003 loss: 0.5070 (0.5165) time: 0.4793 data: 0.0002 max mem: 52150 +Epoch: [27] [4350/7269] eta: 0:20:58 lr: 0.000003 loss: 0.5543 (0.5166) time: 0.5845 data: 0.0002 max mem: 52150 +Epoch: [27] [4360/7269] eta: 0:20:54 lr: 0.000003 loss: 0.5381 (0.5165) time: 0.4777 data: 0.0002 max mem: 52150 +Epoch: [27] [4370/7269] eta: 0:20:50 lr: 0.000003 loss: 0.5315 (0.5166) time: 0.4187 data: 0.0002 max mem: 52150 +Epoch: [27] [4380/7269] eta: 0:20:46 lr: 0.000003 loss: 0.4892 (0.5165) time: 0.5387 data: 0.0002 max mem: 52150 +Epoch: [27] [4390/7269] eta: 0:20:42 lr: 0.000003 loss: 0.4878 (0.5166) time: 0.4923 data: 0.0002 max mem: 52150 +Epoch: [27] [4400/7269] eta: 0:20:37 lr: 0.000003 loss: 0.4862 (0.5165) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [27] [4410/7269] eta: 0:20:33 lr: 0.000003 loss: 0.4812 (0.5165) time: 0.4342 data: 0.0002 max mem: 52150 +Epoch: [27] [4420/7269] eta: 0:20:28 lr: 0.000003 loss: 0.5158 (0.5165) time: 0.4339 data: 0.0002 max mem: 52150 +Epoch: [27] [4430/7269] eta: 0:20:24 lr: 0.000003 loss: 0.5274 (0.5165) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [27] [4440/7269] eta: 0:20:20 lr: 0.000003 loss: 0.5457 (0.5165) time: 0.4792 data: 0.0002 max mem: 52150 +Epoch: [27] [4450/7269] eta: 0:20:16 lr: 0.000003 loss: 0.5244 (0.5165) time: 0.5247 data: 0.0002 max mem: 52150 +Epoch: [27] [4460/7269] eta: 0:20:12 lr: 0.000003 loss: 0.5244 (0.5165) time: 0.4169 data: 0.0002 max mem: 52150 +Epoch: [27] [4470/7269] eta: 0:20:07 lr: 0.000003 loss: 0.5319 (0.5165) time: 0.4201 data: 0.0002 max mem: 52150 +Epoch: [27] [4480/7269] eta: 0:20:03 lr: 0.000003 loss: 0.5050 (0.5165) time: 0.4204 data: 0.0002 max mem: 52150 +Epoch: [27] [4490/7269] eta: 0:19:58 lr: 0.000003 loss: 0.5261 (0.5166) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [27] [4500/7269] eta: 0:19:54 lr: 0.000003 loss: 0.5104 (0.5165) time: 0.4329 data: 0.0002 max mem: 52150 +Epoch: [27] [4510/7269] eta: 0:19:50 lr: 0.000003 loss: 0.4901 (0.5165) time: 0.4796 data: 0.0002 max mem: 52150 +Epoch: [27] [4520/7269] eta: 0:19:46 lr: 0.000003 loss: 0.4787 (0.5164) time: 0.4672 data: 0.0002 max mem: 52150 +Epoch: [27] [4530/7269] eta: 0:19:41 lr: 0.000003 loss: 0.5164 (0.5164) time: 0.4206 data: 0.0002 max mem: 52150 +Epoch: [27] [4540/7269] eta: 0:19:37 lr: 0.000003 loss: 0.5164 (0.5164) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [27] [4550/7269] eta: 0:19:32 lr: 0.000003 loss: 0.4955 (0.5163) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [27] [4560/7269] eta: 0:19:27 lr: 0.000003 loss: 0.5152 (0.5163) time: 0.3727 data: 0.0002 max mem: 52150 +Epoch: [27] [4570/7269] eta: 0:19:24 lr: 0.000003 loss: 0.5152 (0.5162) time: 0.4940 data: 0.0002 max mem: 52150 +Epoch: [27] [4580/7269] eta: 0:19:20 lr: 0.000003 loss: 0.5047 (0.5162) time: 0.5434 data: 0.0002 max mem: 52150 +Epoch: [27] [4590/7269] eta: 0:19:16 lr: 0.000003 loss: 0.5023 (0.5162) time: 0.4690 data: 0.0002 max mem: 52150 +Epoch: [27] [4600/7269] eta: 0:19:12 lr: 0.000003 loss: 0.4928 (0.5161) time: 0.4798 data: 0.0002 max mem: 52150 +Epoch: [27] [4610/7269] eta: 0:19:09 lr: 0.000003 loss: 0.5194 (0.5161) time: 0.5455 data: 0.0002 max mem: 52150 +Epoch: [27] [4620/7269] eta: 0:19:04 lr: 0.000003 loss: 0.4943 (0.5160) time: 0.4846 data: 0.0002 max mem: 52150 +Epoch: [27] [4630/7269] eta: 0:19:00 lr: 0.000003 loss: 0.4896 (0.5160) time: 0.4370 data: 0.0002 max mem: 52150 +Epoch: [27] [4640/7269] eta: 0:18:56 lr: 0.000003 loss: 0.4845 (0.5159) time: 0.5006 data: 0.0002 max mem: 52150 +Epoch: [27] [4650/7269] eta: 0:18:52 lr: 0.000003 loss: 0.5199 (0.5159) time: 0.4856 data: 0.0002 max mem: 52150 +Epoch: [27] [4660/7269] eta: 0:18:47 lr: 0.000003 loss: 0.5170 (0.5159) time: 0.4220 data: 0.0002 max mem: 52150 +Epoch: [27] [4670/7269] eta: 0:18:43 lr: 0.000003 loss: 0.5126 (0.5158) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [27] [4680/7269] eta: 0:18:38 lr: 0.000003 loss: 0.5038 (0.5158) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [27] [4690/7269] eta: 0:18:33 lr: 0.000003 loss: 0.4925 (0.5158) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [27] [4700/7269] eta: 0:18:29 lr: 0.000003 loss: 0.4724 (0.5157) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [27] [4710/7269] eta: 0:18:24 lr: 0.000003 loss: 0.4756 (0.5157) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [27] [4720/7269] eta: 0:18:19 lr: 0.000003 loss: 0.5074 (0.5157) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [27] [4730/7269] eta: 0:18:16 lr: 0.000003 loss: 0.5214 (0.5157) time: 0.4733 data: 0.0002 max mem: 52150 +Epoch: [27] [4740/7269] eta: 0:18:11 lr: 0.000003 loss: 0.5516 (0.5157) time: 0.4734 data: 0.0002 max mem: 52150 +Epoch: [27] [4750/7269] eta: 0:18:07 lr: 0.000003 loss: 0.5391 (0.5158) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [27] [4760/7269] eta: 0:18:02 lr: 0.000003 loss: 0.5293 (0.5158) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [27] [4770/7269] eta: 0:17:57 lr: 0.000003 loss: 0.5382 (0.5158) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [27] [4780/7269] eta: 0:17:54 lr: 0.000003 loss: 0.5286 (0.5158) time: 0.4926 data: 0.0002 max mem: 52150 +Epoch: [27] [4790/7269] eta: 0:17:50 lr: 0.000003 loss: 0.5286 (0.5158) time: 0.5548 data: 0.0002 max mem: 52150 +Epoch: [27] [4800/7269] eta: 0:17:45 lr: 0.000003 loss: 0.4746 (0.5157) time: 0.4338 data: 0.0002 max mem: 52150 +Epoch: [27] [4810/7269] eta: 0:17:41 lr: 0.000003 loss: 0.4698 (0.5157) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [27] [4820/7269] eta: 0:17:36 lr: 0.000003 loss: 0.5261 (0.5157) time: 0.3727 data: 0.0002 max mem: 52150 +Epoch: [27] [4830/7269] eta: 0:17:32 lr: 0.000003 loss: 0.5015 (0.5157) time: 0.4179 data: 0.0002 max mem: 52150 +Epoch: [27] [4840/7269] eta: 0:17:27 lr: 0.000003 loss: 0.5015 (0.5157) time: 0.4170 data: 0.0002 max mem: 52150 +Epoch: [27] [4850/7269] eta: 0:17:23 lr: 0.000003 loss: 0.5429 (0.5157) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [27] [4860/7269] eta: 0:17:18 lr: 0.000003 loss: 0.5498 (0.5158) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [27] [4870/7269] eta: 0:17:14 lr: 0.000003 loss: 0.5057 (0.5157) time: 0.4319 data: 0.0002 max mem: 52150 +Epoch: [27] [4880/7269] eta: 0:17:10 lr: 0.000003 loss: 0.4966 (0.5157) time: 0.4972 data: 0.0002 max mem: 52150 +Epoch: [27] [4890/7269] eta: 0:17:06 lr: 0.000003 loss: 0.4961 (0.5156) time: 0.4371 data: 0.0002 max mem: 52150 +Epoch: [27] [4900/7269] eta: 0:17:02 lr: 0.000003 loss: 0.4961 (0.5156) time: 0.4651 data: 0.0002 max mem: 52150 +Epoch: [27] [4910/7269] eta: 0:16:57 lr: 0.000003 loss: 0.5032 (0.5156) time: 0.4661 data: 0.0002 max mem: 52150 +Epoch: [27] [4920/7269] eta: 0:16:53 lr: 0.000003 loss: 0.5092 (0.5156) time: 0.4203 data: 0.0002 max mem: 52150 +Epoch: [27] [4930/7269] eta: 0:16:50 lr: 0.000003 loss: 0.5187 (0.5156) time: 0.5271 data: 0.0002 max mem: 52150 +Epoch: [27] [4940/7269] eta: 0:16:45 lr: 0.000003 loss: 0.5397 (0.5157) time: 0.5392 data: 0.0002 max mem: 52150 +Epoch: [27] [4950/7269] eta: 0:16:41 lr: 0.000003 loss: 0.4977 (0.5156) time: 0.4316 data: 0.0002 max mem: 52150 +Epoch: [27] [4960/7269] eta: 0:16:36 lr: 0.000003 loss: 0.4961 (0.5156) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [27] [4970/7269] eta: 0:16:32 lr: 0.000003 loss: 0.5303 (0.5156) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [27] [4980/7269] eta: 0:16:28 lr: 0.000003 loss: 0.5448 (0.5157) time: 0.4316 data: 0.0002 max mem: 52150 +Epoch: [27] [4990/7269] eta: 0:16:24 lr: 0.000003 loss: 0.5309 (0.5156) time: 0.4919 data: 0.0002 max mem: 52150 +Epoch: [27] [5000/7269] eta: 0:16:20 lr: 0.000003 loss: 0.5070 (0.5156) time: 0.4913 data: 0.0002 max mem: 52150 +Epoch: [27] [5010/7269] eta: 0:16:16 lr: 0.000003 loss: 0.5195 (0.5156) time: 0.4900 data: 0.0002 max mem: 52150 +Epoch: [27] [5020/7269] eta: 0:16:11 lr: 0.000003 loss: 0.5267 (0.5156) time: 0.4307 data: 0.0002 max mem: 52150 +Epoch: [27] [5030/7269] eta: 0:16:06 lr: 0.000003 loss: 0.5267 (0.5157) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [27] [5040/7269] eta: 0:16:03 lr: 0.000003 loss: 0.5328 (0.5157) time: 0.4794 data: 0.0002 max mem: 52150 +Epoch: [27] [5050/7269] eta: 0:15:59 lr: 0.000003 loss: 0.5445 (0.5158) time: 0.5425 data: 0.0002 max mem: 52150 +Epoch: [27] [5060/7269] eta: 0:15:55 lr: 0.000003 loss: 0.5120 (0.5157) time: 0.4946 data: 0.0002 max mem: 52150 +Epoch: [27] [5070/7269] eta: 0:15:50 lr: 0.000003 loss: 0.4776 (0.5157) time: 0.4319 data: 0.0002 max mem: 52150 +Epoch: [27] [5080/7269] eta: 0:15:46 lr: 0.000003 loss: 0.5172 (0.5157) time: 0.4328 data: 0.0002 max mem: 52150 +Epoch: [27] [5090/7269] eta: 0:15:42 lr: 0.000003 loss: 0.5207 (0.5157) time: 0.4842 data: 0.0002 max mem: 52150 +Epoch: [27] [5100/7269] eta: 0:15:38 lr: 0.000003 loss: 0.5308 (0.5156) time: 0.4865 data: 0.0002 max mem: 52150 +Epoch: [27] [5110/7269] eta: 0:15:34 lr: 0.000003 loss: 0.5199 (0.5157) time: 0.5441 data: 0.0002 max mem: 52150 +Epoch: [27] [5120/7269] eta: 0:15:30 lr: 0.000003 loss: 0.5290 (0.5157) time: 0.4811 data: 0.0002 max mem: 52150 +Epoch: [27] [5130/7269] eta: 0:15:25 lr: 0.000003 loss: 0.5131 (0.5157) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [27] [5140/7269] eta: 0:15:20 lr: 0.000003 loss: 0.5099 (0.5156) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [27] [5150/7269] eta: 0:15:16 lr: 0.000003 loss: 0.4591 (0.5156) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [27] [5160/7269] eta: 0:15:11 lr: 0.000003 loss: 0.4893 (0.5156) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [27] [5170/7269] eta: 0:15:07 lr: 0.000003 loss: 0.5254 (0.5156) time: 0.4174 data: 0.0002 max mem: 52150 +Epoch: [27] [5180/7269] eta: 0:15:02 lr: 0.000003 loss: 0.5296 (0.5156) time: 0.4167 data: 0.0002 max mem: 52150 +Epoch: [27] [5190/7269] eta: 0:14:58 lr: 0.000003 loss: 0.5050 (0.5155) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [27] [5200/7269] eta: 0:14:54 lr: 0.000003 loss: 0.4970 (0.5155) time: 0.4208 data: 0.0002 max mem: 52150 +Epoch: [27] [5210/7269] eta: 0:14:50 lr: 0.000003 loss: 0.4970 (0.5155) time: 0.4844 data: 0.0002 max mem: 52150 +Epoch: [27] [5220/7269] eta: 0:14:45 lr: 0.000003 loss: 0.5162 (0.5155) time: 0.4358 data: 0.0002 max mem: 52150 +Epoch: [27] [5230/7269] eta: 0:14:41 lr: 0.000003 loss: 0.5200 (0.5155) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [27] [5240/7269] eta: 0:14:36 lr: 0.000003 loss: 0.5203 (0.5155) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [27] [5250/7269] eta: 0:14:31 lr: 0.000003 loss: 0.5156 (0.5155) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [27] [5260/7269] eta: 0:14:27 lr: 0.000003 loss: 0.5201 (0.5155) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [27] [5270/7269] eta: 0:14:22 lr: 0.000003 loss: 0.5201 (0.5154) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [27] [5280/7269] eta: 0:14:18 lr: 0.000003 loss: 0.5047 (0.5154) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [27] [5290/7269] eta: 0:14:13 lr: 0.000003 loss: 0.5193 (0.5155) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [27] [5300/7269] eta: 0:14:09 lr: 0.000003 loss: 0.4980 (0.5154) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [27] [5310/7269] eta: 0:14:04 lr: 0.000003 loss: 0.4867 (0.5154) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [27] [5320/7269] eta: 0:14:00 lr: 0.000003 loss: 0.4882 (0.5153) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [27] [5330/7269] eta: 0:13:55 lr: 0.000003 loss: 0.5102 (0.5153) time: 0.4176 data: 0.0002 max mem: 52150 +Epoch: [27] [5340/7269] eta: 0:13:51 lr: 0.000003 loss: 0.5236 (0.5154) time: 0.4803 data: 0.0002 max mem: 52150 +Epoch: [27] [5350/7269] eta: 0:13:47 lr: 0.000003 loss: 0.5378 (0.5154) time: 0.4349 data: 0.0002 max mem: 52150 +Epoch: [27] [5360/7269] eta: 0:13:43 lr: 0.000003 loss: 0.5119 (0.5154) time: 0.4162 data: 0.0002 max mem: 52150 +Epoch: [27] [5370/7269] eta: 0:13:39 lr: 0.000003 loss: 0.5050 (0.5153) time: 0.4771 data: 0.0002 max mem: 52150 +Epoch: [27] [5380/7269] eta: 0:13:34 lr: 0.000003 loss: 0.4766 (0.5153) time: 0.4814 data: 0.0002 max mem: 52150 +Epoch: [27] [5390/7269] eta: 0:13:30 lr: 0.000003 loss: 0.5093 (0.5152) time: 0.4712 data: 0.0002 max mem: 52150 +Epoch: [27] [5400/7269] eta: 0:13:26 lr: 0.000003 loss: 0.5195 (0.5152) time: 0.4229 data: 0.0002 max mem: 52150 +Epoch: [27] [5410/7269] eta: 0:13:22 lr: 0.000003 loss: 0.5266 (0.5153) time: 0.4829 data: 0.0002 max mem: 52150 +Epoch: [27] [5420/7269] eta: 0:13:17 lr: 0.000003 loss: 0.5266 (0.5153) time: 0.4832 data: 0.0002 max mem: 52150 +Epoch: [27] [5430/7269] eta: 0:13:13 lr: 0.000003 loss: 0.5013 (0.5152) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [27] [5440/7269] eta: 0:13:09 lr: 0.000003 loss: 0.5363 (0.5153) time: 0.4314 data: 0.0002 max mem: 52150 +Epoch: [27] [5450/7269] eta: 0:13:04 lr: 0.000003 loss: 0.5482 (0.5154) time: 0.4316 data: 0.0002 max mem: 52150 +Epoch: [27] [5460/7269] eta: 0:13:00 lr: 0.000003 loss: 0.5166 (0.5154) time: 0.4242 data: 0.0002 max mem: 52150 +Epoch: [27] [5470/7269] eta: 0:12:56 lr: 0.000003 loss: 0.5423 (0.5154) time: 0.4842 data: 0.0002 max mem: 52150 +Epoch: [27] [5480/7269] eta: 0:12:52 lr: 0.000003 loss: 0.5385 (0.5154) time: 0.5563 data: 0.0002 max mem: 52150 +Epoch: [27] [5490/7269] eta: 0:12:48 lr: 0.000003 loss: 0.5140 (0.5155) time: 0.4960 data: 0.0002 max mem: 52150 +Epoch: [27] [5500/7269] eta: 0:12:43 lr: 0.000003 loss: 0.5140 (0.5155) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [27] [5510/7269] eta: 0:12:39 lr: 0.000003 loss: 0.5098 (0.5155) time: 0.4185 data: 0.0002 max mem: 52150 +Epoch: [27] [5520/7269] eta: 0:12:35 lr: 0.000003 loss: 0.5291 (0.5155) time: 0.4174 data: 0.0002 max mem: 52150 +Epoch: [27] [5530/7269] eta: 0:12:30 lr: 0.000003 loss: 0.5233 (0.5155) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [27] [5540/7269] eta: 0:12:27 lr: 0.000003 loss: 0.5219 (0.5155) time: 0.5511 data: 0.0002 max mem: 52150 +Epoch: [27] [5550/7269] eta: 0:12:22 lr: 0.000003 loss: 0.5263 (0.5155) time: 0.6004 data: 0.0002 max mem: 52150 +Epoch: [27] [5560/7269] eta: 0:12:18 lr: 0.000003 loss: 0.5272 (0.5154) time: 0.4211 data: 0.0002 max mem: 52150 +Epoch: [27] [5570/7269] eta: 0:12:13 lr: 0.000003 loss: 0.5336 (0.5154) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [27] [5580/7269] eta: 0:12:09 lr: 0.000003 loss: 0.5146 (0.5154) time: 0.3729 data: 0.0002 max mem: 52150 +Epoch: [27] [5590/7269] eta: 0:12:05 lr: 0.000003 loss: 0.4987 (0.5154) time: 0.4214 data: 0.0002 max mem: 52150 +Epoch: [27] [5600/7269] eta: 0:12:00 lr: 0.000003 loss: 0.4940 (0.5154) time: 0.4215 data: 0.0002 max mem: 52150 +Epoch: [27] [5610/7269] eta: 0:11:56 lr: 0.000003 loss: 0.4940 (0.5153) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [27] [5620/7269] eta: 0:11:52 lr: 0.000003 loss: 0.4982 (0.5153) time: 0.4314 data: 0.0002 max mem: 52150 +Epoch: [27] [5630/7269] eta: 0:11:47 lr: 0.000003 loss: 0.5268 (0.5153) time: 0.4315 data: 0.0002 max mem: 52150 +Epoch: [27] [5640/7269] eta: 0:11:43 lr: 0.000003 loss: 0.5112 (0.5152) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [27] [5650/7269] eta: 0:11:38 lr: 0.000003 loss: 0.5006 (0.5152) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [27] [5660/7269] eta: 0:11:34 lr: 0.000003 loss: 0.5151 (0.5152) time: 0.4199 data: 0.0002 max mem: 52150 +Epoch: [27] [5670/7269] eta: 0:11:29 lr: 0.000003 loss: 0.5343 (0.5152) time: 0.4197 data: 0.0002 max mem: 52150 +Epoch: [27] [5680/7269] eta: 0:11:25 lr: 0.000003 loss: 0.5174 (0.5152) time: 0.4343 data: 0.0002 max mem: 52150 +Epoch: [27] [5690/7269] eta: 0:11:21 lr: 0.000003 loss: 0.5038 (0.5152) time: 0.4938 data: 0.0002 max mem: 52150 +Epoch: [27] [5700/7269] eta: 0:11:17 lr: 0.000003 loss: 0.4998 (0.5151) time: 0.4313 data: 0.0002 max mem: 52150 +Epoch: [27] [5710/7269] eta: 0:11:12 lr: 0.000003 loss: 0.4821 (0.5151) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [27] [5720/7269] eta: 0:11:08 lr: 0.000003 loss: 0.5161 (0.5152) time: 0.4195 data: 0.0002 max mem: 52150 +Epoch: [27] [5730/7269] eta: 0:11:04 lr: 0.000003 loss: 0.5522 (0.5152) time: 0.4680 data: 0.0002 max mem: 52150 +Epoch: [27] [5740/7269] eta: 0:10:59 lr: 0.000003 loss: 0.5481 (0.5153) time: 0.4203 data: 0.0002 max mem: 52150 +Epoch: [27] [5750/7269] eta: 0:10:55 lr: 0.000003 loss: 0.5323 (0.5152) time: 0.4339 data: 0.0002 max mem: 52150 +Epoch: [27] [5760/7269] eta: 0:10:51 lr: 0.000003 loss: 0.5047 (0.5152) time: 0.4339 data: 0.0002 max mem: 52150 +Epoch: [27] [5770/7269] eta: 0:10:46 lr: 0.000003 loss: 0.5370 (0.5153) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [27] [5780/7269] eta: 0:10:42 lr: 0.000003 loss: 0.5529 (0.5153) time: 0.4205 data: 0.0002 max mem: 52150 +Epoch: [27] [5790/7269] eta: 0:10:38 lr: 0.000003 loss: 0.5529 (0.5154) time: 0.4201 data: 0.0002 max mem: 52150 +Epoch: [27] [5800/7269] eta: 0:10:33 lr: 0.000003 loss: 0.5769 (0.5154) time: 0.4157 data: 0.0002 max mem: 52150 +Epoch: [27] [5810/7269] eta: 0:10:29 lr: 0.000003 loss: 0.5247 (0.5154) time: 0.4157 data: 0.0002 max mem: 52150 +Epoch: [27] [5820/7269] eta: 0:10:24 lr: 0.000003 loss: 0.5133 (0.5154) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [27] [5830/7269] eta: 0:10:20 lr: 0.000003 loss: 0.5149 (0.5154) time: 0.4936 data: 0.0002 max mem: 52150 +Epoch: [27] [5840/7269] eta: 0:10:16 lr: 0.000003 loss: 0.4848 (0.5153) time: 0.4935 data: 0.0002 max mem: 52150 +Epoch: [27] [5850/7269] eta: 0:10:12 lr: 0.000003 loss: 0.4596 (0.5153) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [27] [5860/7269] eta: 0:10:07 lr: 0.000003 loss: 0.5116 (0.5153) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [27] [5870/7269] eta: 0:10:03 lr: 0.000003 loss: 0.5348 (0.5154) time: 0.4656 data: 0.0002 max mem: 52150 +Epoch: [27] [5880/7269] eta: 0:09:59 lr: 0.000003 loss: 0.5751 (0.5154) time: 0.5152 data: 0.0002 max mem: 52150 +Epoch: [27] [5890/7269] eta: 0:09:54 lr: 0.000003 loss: 0.5682 (0.5155) time: 0.4209 data: 0.0002 max mem: 52150 +Epoch: [27] [5900/7269] eta: 0:09:50 lr: 0.000003 loss: 0.5382 (0.5155) time: 0.4325 data: 0.0002 max mem: 52150 +Epoch: [27] [5910/7269] eta: 0:09:46 lr: 0.000003 loss: 0.5382 (0.5155) time: 0.4967 data: 0.0001 max mem: 52150 +Epoch: [27] [5920/7269] eta: 0:09:42 lr: 0.000003 loss: 0.5349 (0.5155) time: 0.4356 data: 0.0002 max mem: 52150 +Epoch: [27] [5930/7269] eta: 0:09:37 lr: 0.000003 loss: 0.5154 (0.5155) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [27] [5940/7269] eta: 0:09:33 lr: 0.000003 loss: 0.5189 (0.5155) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [27] [5950/7269] eta: 0:09:29 lr: 0.000003 loss: 0.5247 (0.5155) time: 0.4980 data: 0.0002 max mem: 52150 +Epoch: [27] [5960/7269] eta: 0:09:24 lr: 0.000003 loss: 0.4775 (0.5154) time: 0.4977 data: 0.0002 max mem: 52150 +Epoch: [27] [5970/7269] eta: 0:09:20 lr: 0.000003 loss: 0.4761 (0.5154) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [27] [5980/7269] eta: 0:09:16 lr: 0.000003 loss: 0.5073 (0.5154) time: 0.4355 data: 0.0002 max mem: 52150 +Epoch: [27] [5990/7269] eta: 0:09:11 lr: 0.000003 loss: 0.5085 (0.5154) time: 0.4359 data: 0.0002 max mem: 52150 +Epoch: [27] [6000/7269] eta: 0:09:07 lr: 0.000003 loss: 0.4999 (0.5153) time: 0.4358 data: 0.0002 max mem: 52150 +Epoch: [27] [6010/7269] eta: 0:09:03 lr: 0.000003 loss: 0.4890 (0.5153) time: 0.4357 data: 0.0002 max mem: 52150 +Epoch: [27] [6020/7269] eta: 0:08:59 lr: 0.000003 loss: 0.4932 (0.5153) time: 0.4187 data: 0.0002 max mem: 52150 +Epoch: [27] [6030/7269] eta: 0:08:54 lr: 0.000003 loss: 0.5343 (0.5154) time: 0.4189 data: 0.0002 max mem: 52150 +Epoch: [27] [6040/7269] eta: 0:08:50 lr: 0.000003 loss: 0.5343 (0.5153) time: 0.4335 data: 0.0002 max mem: 52150 +Epoch: [27] [6050/7269] eta: 0:08:45 lr: 0.000003 loss: 0.5216 (0.5153) time: 0.4329 data: 0.0002 max mem: 52150 +Epoch: [27] [6060/7269] eta: 0:08:41 lr: 0.000003 loss: 0.5483 (0.5154) time: 0.4330 data: 0.0002 max mem: 52150 +Epoch: [27] [6070/7269] eta: 0:08:37 lr: 0.000003 loss: 0.5193 (0.5153) time: 0.4331 data: 0.0002 max mem: 52150 +Epoch: [27] [6080/7269] eta: 0:08:33 lr: 0.000003 loss: 0.4905 (0.5153) time: 0.5960 data: 0.0002 max mem: 52150 +Epoch: [27] [6090/7269] eta: 0:08:29 lr: 0.000003 loss: 0.5204 (0.5152) time: 0.6580 data: 0.0002 max mem: 52150 +Epoch: [27] [6100/7269] eta: 0:08:25 lr: 0.000003 loss: 0.5000 (0.5152) time: 0.4336 data: 0.0002 max mem: 52150 +Epoch: [27] [6110/7269] eta: 0:08:20 lr: 0.000003 loss: 0.5000 (0.5153) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [27] [6120/7269] eta: 0:08:16 lr: 0.000003 loss: 0.5013 (0.5152) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [27] [6130/7269] eta: 0:08:12 lr: 0.000003 loss: 0.5088 (0.5152) time: 0.4182 data: 0.0002 max mem: 52150 +Epoch: [27] [6140/7269] eta: 0:08:07 lr: 0.000003 loss: 0.5092 (0.5152) time: 0.4183 data: 0.0001 max mem: 52150 +Epoch: [27] [6150/7269] eta: 0:08:03 lr: 0.000003 loss: 0.4836 (0.5152) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [27] [6160/7269] eta: 0:07:58 lr: 0.000003 loss: 0.4938 (0.5152) time: 0.4234 data: 0.0002 max mem: 52150 +Epoch: [27] [6170/7269] eta: 0:07:54 lr: 0.000003 loss: 0.4853 (0.5151) time: 0.4696 data: 0.0002 max mem: 52150 +Epoch: [27] [6180/7269] eta: 0:07:50 lr: 0.000003 loss: 0.4853 (0.5151) time: 0.4801 data: 0.0002 max mem: 52150 +Epoch: [27] [6190/7269] eta: 0:07:46 lr: 0.000003 loss: 0.5054 (0.5151) time: 0.4327 data: 0.0002 max mem: 52150 +Epoch: [27] [6200/7269] eta: 0:07:41 lr: 0.000003 loss: 0.5054 (0.5150) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [27] [6210/7269] eta: 0:07:37 lr: 0.000003 loss: 0.5274 (0.5151) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [27] [6220/7269] eta: 0:07:32 lr: 0.000003 loss: 0.5274 (0.5151) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [27] [6230/7269] eta: 0:07:28 lr: 0.000003 loss: 0.5402 (0.5152) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [27] [6240/7269] eta: 0:07:23 lr: 0.000003 loss: 0.5440 (0.5152) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [27] [6250/7269] eta: 0:07:19 lr: 0.000003 loss: 0.5440 (0.5152) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [27] [6260/7269] eta: 0:07:15 lr: 0.000003 loss: 0.5601 (0.5152) time: 0.4939 data: 0.0002 max mem: 52150 +Epoch: [27] [6270/7269] eta: 0:07:11 lr: 0.000003 loss: 0.5196 (0.5153) time: 0.4941 data: 0.0002 max mem: 52150 +Epoch: [27] [6280/7269] eta: 0:07:06 lr: 0.000003 loss: 0.5110 (0.5152) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [27] [6290/7269] eta: 0:07:02 lr: 0.000003 loss: 0.4891 (0.5152) time: 0.4321 data: 0.0002 max mem: 52150 +Epoch: [27] [6300/7269] eta: 0:06:58 lr: 0.000003 loss: 0.4993 (0.5153) time: 0.4326 data: 0.0001 max mem: 52150 +Epoch: [27] [6310/7269] eta: 0:06:53 lr: 0.000003 loss: 0.5352 (0.5152) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [27] [6320/7269] eta: 0:06:49 lr: 0.000003 loss: 0.5139 (0.5152) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [27] [6330/7269] eta: 0:06:45 lr: 0.000003 loss: 0.4971 (0.5152) time: 0.4949 data: 0.0001 max mem: 52150 +Epoch: [27] [6340/7269] eta: 0:06:40 lr: 0.000003 loss: 0.5022 (0.5151) time: 0.4951 data: 0.0002 max mem: 52150 +Epoch: [27] [6350/7269] eta: 0:06:36 lr: 0.000003 loss: 0.4965 (0.5151) time: 0.4677 data: 0.0002 max mem: 52150 +Epoch: [27] [6360/7269] eta: 0:06:32 lr: 0.000003 loss: 0.4965 (0.5150) time: 0.5174 data: 0.0002 max mem: 52150 +Epoch: [27] [6370/7269] eta: 0:06:28 lr: 0.000003 loss: 0.4853 (0.5150) time: 0.4818 data: 0.0002 max mem: 52150 +Epoch: [27] [6380/7269] eta: 0:06:23 lr: 0.000003 loss: 0.5018 (0.5150) time: 0.4317 data: 0.0002 max mem: 52150 +Epoch: [27] [6390/7269] eta: 0:06:19 lr: 0.000003 loss: 0.5207 (0.5150) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [27] [6400/7269] eta: 0:06:15 lr: 0.000003 loss: 0.5000 (0.5150) time: 0.4211 data: 0.0002 max mem: 52150 +Epoch: [27] [6410/7269] eta: 0:06:10 lr: 0.000003 loss: 0.5074 (0.5150) time: 0.4219 data: 0.0002 max mem: 52150 +Epoch: [27] [6420/7269] eta: 0:06:06 lr: 0.000003 loss: 0.5076 (0.5150) time: 0.4336 data: 0.0002 max mem: 52150 +Epoch: [27] [6430/7269] eta: 0:06:02 lr: 0.000003 loss: 0.5056 (0.5150) time: 0.4329 data: 0.0002 max mem: 52150 +Epoch: [27] [6440/7269] eta: 0:05:57 lr: 0.000003 loss: 0.5448 (0.5151) time: 0.4806 data: 0.0002 max mem: 52150 +Epoch: [27] [6450/7269] eta: 0:05:53 lr: 0.000003 loss: 0.5416 (0.5151) time: 0.4804 data: 0.0002 max mem: 52150 +Epoch: [27] [6460/7269] eta: 0:05:49 lr: 0.000003 loss: 0.5183 (0.5151) time: 0.4345 data: 0.0002 max mem: 52150 +Epoch: [27] [6470/7269] eta: 0:05:45 lr: 0.000003 loss: 0.5072 (0.5151) time: 0.5604 data: 0.0001 max mem: 52150 +Epoch: [27] [6480/7269] eta: 0:05:40 lr: 0.000003 loss: 0.5227 (0.5151) time: 0.5439 data: 0.0001 max mem: 52150 +Epoch: [27] [6490/7269] eta: 0:05:36 lr: 0.000003 loss: 0.5217 (0.5151) time: 0.4816 data: 0.0001 max mem: 52150 +Epoch: [27] [6500/7269] eta: 0:05:32 lr: 0.000003 loss: 0.4893 (0.5150) time: 0.4352 data: 0.0001 max mem: 52150 +Epoch: [27] [6510/7269] eta: 0:05:27 lr: 0.000003 loss: 0.4951 (0.5150) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [27] [6520/7269] eta: 0:05:23 lr: 0.000003 loss: 0.5361 (0.5151) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [27] [6530/7269] eta: 0:05:19 lr: 0.000003 loss: 0.5178 (0.5150) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [27] [6540/7269] eta: 0:05:14 lr: 0.000003 loss: 0.4929 (0.5150) time: 0.4313 data: 0.0002 max mem: 52150 +Epoch: [27] [6550/7269] eta: 0:05:10 lr: 0.000003 loss: 0.5039 (0.5150) time: 0.4904 data: 0.0002 max mem: 52150 +Epoch: [27] [6560/7269] eta: 0:05:06 lr: 0.000003 loss: 0.5293 (0.5150) time: 0.4311 data: 0.0002 max mem: 52150 +Epoch: [27] [6570/7269] eta: 0:05:01 lr: 0.000003 loss: 0.5352 (0.5151) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [27] [6580/7269] eta: 0:04:57 lr: 0.000003 loss: 0.5252 (0.5150) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [27] [6590/7269] eta: 0:04:53 lr: 0.000003 loss: 0.5293 (0.5151) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [27] [6600/7269] eta: 0:04:48 lr: 0.000003 loss: 0.5677 (0.5152) time: 0.4188 data: 0.0001 max mem: 52150 +Epoch: [27] [6610/7269] eta: 0:04:44 lr: 0.000003 loss: 0.5479 (0.5152) time: 0.4824 data: 0.0001 max mem: 52150 +Epoch: [27] [6620/7269] eta: 0:04:40 lr: 0.000003 loss: 0.5074 (0.5151) time: 0.4839 data: 0.0001 max mem: 52150 +Epoch: [27] [6630/7269] eta: 0:04:36 lr: 0.000003 loss: 0.5118 (0.5151) time: 0.5306 data: 0.0002 max mem: 52150 +Epoch: [27] [6640/7269] eta: 0:04:31 lr: 0.000003 loss: 0.5471 (0.5152) time: 0.4819 data: 0.0002 max mem: 52150 +Epoch: [27] [6650/7269] eta: 0:04:27 lr: 0.000003 loss: 0.5428 (0.5152) time: 0.4337 data: 0.0002 max mem: 52150 +Epoch: [27] [6660/7269] eta: 0:04:23 lr: 0.000003 loss: 0.5038 (0.5152) time: 0.4333 data: 0.0002 max mem: 52150 +Epoch: [27] [6670/7269] eta: 0:04:18 lr: 0.000003 loss: 0.4818 (0.5150) time: 0.3712 data: 0.0002 max mem: 52150 +Epoch: [27] [6680/7269] eta: 0:04:14 lr: 0.000003 loss: 0.4709 (0.5150) time: 0.4310 data: 0.0002 max mem: 52150 +Epoch: [27] [6690/7269] eta: 0:04:10 lr: 0.000003 loss: 0.5403 (0.5151) time: 0.4317 data: 0.0002 max mem: 52150 +Epoch: [27] [6700/7269] eta: 0:04:05 lr: 0.000003 loss: 0.5213 (0.5150) time: 0.3724 data: 0.0001 max mem: 52150 +Epoch: [27] [6710/7269] eta: 0:04:01 lr: 0.000003 loss: 0.5054 (0.5150) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [27] [6720/7269] eta: 0:03:57 lr: 0.000003 loss: 0.5054 (0.5150) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [27] [6730/7269] eta: 0:03:52 lr: 0.000003 loss: 0.5236 (0.5150) time: 0.4339 data: 0.0001 max mem: 52150 +Epoch: [27] [6740/7269] eta: 0:03:48 lr: 0.000003 loss: 0.5316 (0.5151) time: 0.4948 data: 0.0001 max mem: 52150 +Epoch: [27] [6750/7269] eta: 0:03:44 lr: 0.000003 loss: 0.5260 (0.5151) time: 0.4328 data: 0.0002 max mem: 52150 +Epoch: [27] [6760/7269] eta: 0:03:39 lr: 0.000003 loss: 0.5005 (0.5151) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [27] [6770/7269] eta: 0:03:35 lr: 0.000003 loss: 0.5197 (0.5151) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [27] [6780/7269] eta: 0:03:31 lr: 0.000003 loss: 0.5048 (0.5151) time: 0.4208 data: 0.0001 max mem: 52150 +Epoch: [27] [6790/7269] eta: 0:03:26 lr: 0.000003 loss: 0.5307 (0.5151) time: 0.4680 data: 0.0001 max mem: 52150 +Epoch: [27] [6800/7269] eta: 0:03:22 lr: 0.000003 loss: 0.5468 (0.5151) time: 0.4651 data: 0.0001 max mem: 52150 +Epoch: [27] [6810/7269] eta: 0:03:18 lr: 0.000003 loss: 0.5240 (0.5151) time: 0.4681 data: 0.0001 max mem: 52150 +Epoch: [27] [6820/7269] eta: 0:03:13 lr: 0.000003 loss: 0.5184 (0.5151) time: 0.4706 data: 0.0001 max mem: 52150 +Epoch: [27] [6830/7269] eta: 0:03:09 lr: 0.000003 loss: 0.5223 (0.5151) time: 0.4208 data: 0.0002 max mem: 52150 +Epoch: [27] [6840/7269] eta: 0:03:05 lr: 0.000003 loss: 0.5223 (0.5151) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [27] [6850/7269] eta: 0:03:00 lr: 0.000003 loss: 0.5392 (0.5151) time: 0.4774 data: 0.0002 max mem: 52150 +Epoch: [27] [6860/7269] eta: 0:02:56 lr: 0.000003 loss: 0.5135 (0.5151) time: 0.4770 data: 0.0002 max mem: 52150 +Epoch: [27] [6870/7269] eta: 0:02:52 lr: 0.000003 loss: 0.5052 (0.5151) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [27] [6880/7269] eta: 0:02:47 lr: 0.000003 loss: 0.5083 (0.5151) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [27] [6890/7269] eta: 0:02:43 lr: 0.000003 loss: 0.5522 (0.5152) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [27] [6900/7269] eta: 0:02:39 lr: 0.000003 loss: 0.5263 (0.5151) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [27] [6910/7269] eta: 0:02:34 lr: 0.000003 loss: 0.4957 (0.5151) time: 0.4206 data: 0.0001 max mem: 52150 +Epoch: [27] [6920/7269] eta: 0:02:30 lr: 0.000003 loss: 0.5349 (0.5151) time: 0.5317 data: 0.0002 max mem: 52150 +Epoch: [27] [6930/7269] eta: 0:02:26 lr: 0.000003 loss: 0.5293 (0.5151) time: 0.5317 data: 0.0002 max mem: 52150 +Epoch: [27] [6940/7269] eta: 0:02:22 lr: 0.000003 loss: 0.5291 (0.5151) time: 0.4206 data: 0.0002 max mem: 52150 +Epoch: [27] [6950/7269] eta: 0:02:17 lr: 0.000003 loss: 0.5301 (0.5151) time: 0.4358 data: 0.0002 max mem: 52150 +Epoch: [27] [6960/7269] eta: 0:02:13 lr: 0.000003 loss: 0.5195 (0.5152) time: 0.4356 data: 0.0002 max mem: 52150 +Epoch: [27] [6970/7269] eta: 0:02:09 lr: 0.000003 loss: 0.5148 (0.5151) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [27] [6980/7269] eta: 0:02:04 lr: 0.000003 loss: 0.5347 (0.5152) time: 0.4362 data: 0.0002 max mem: 52150 +Epoch: [27] [6990/7269] eta: 0:02:00 lr: 0.000003 loss: 0.5592 (0.5152) time: 0.4959 data: 0.0002 max mem: 52150 +Epoch: [27] [7000/7269] eta: 0:01:56 lr: 0.000003 loss: 0.5658 (0.5153) time: 0.4322 data: 0.0001 max mem: 52150 +Epoch: [27] [7010/7269] eta: 0:01:51 lr: 0.000003 loss: 0.5259 (0.5153) time: 0.3721 data: 0.0001 max mem: 52150 +Epoch: [27] [7020/7269] eta: 0:01:47 lr: 0.000003 loss: 0.5243 (0.5153) time: 0.4347 data: 0.0001 max mem: 52150 +Epoch: [27] [7030/7269] eta: 0:01:43 lr: 0.000003 loss: 0.5280 (0.5153) time: 0.4348 data: 0.0001 max mem: 52150 +Epoch: [27] [7040/7269] eta: 0:01:38 lr: 0.000003 loss: 0.5396 (0.5153) time: 0.3724 data: 0.0001 max mem: 52150 +Epoch: [27] [7050/7269] eta: 0:01:34 lr: 0.000003 loss: 0.5261 (0.5153) time: 0.4360 data: 0.0001 max mem: 52150 +Epoch: [27] [7060/7269] eta: 0:01:30 lr: 0.000003 loss: 0.5261 (0.5153) time: 0.4355 data: 0.0001 max mem: 52150 +Epoch: [27] [7070/7269] eta: 0:01:25 lr: 0.000003 loss: 0.5302 (0.5154) time: 0.4235 data: 0.0001 max mem: 52150 +Epoch: [27] [7080/7269] eta: 0:01:21 lr: 0.000003 loss: 0.5234 (0.5153) time: 0.4751 data: 0.0002 max mem: 52150 +Epoch: [27] [7090/7269] eta: 0:01:17 lr: 0.000003 loss: 0.5039 (0.5153) time: 0.4702 data: 0.0002 max mem: 52150 +Epoch: [27] [7100/7269] eta: 0:01:12 lr: 0.000003 loss: 0.5076 (0.5153) time: 0.4190 data: 0.0002 max mem: 52150 +Epoch: [27] [7110/7269] eta: 0:01:08 lr: 0.000003 loss: 0.5070 (0.5153) time: 0.4326 data: 0.0001 max mem: 52150 +Epoch: [27] [7120/7269] eta: 0:01:04 lr: 0.000003 loss: 0.4940 (0.5153) time: 0.4329 data: 0.0001 max mem: 52150 +Epoch: [27] [7130/7269] eta: 0:01:00 lr: 0.000003 loss: 0.5272 (0.5153) time: 0.3725 data: 0.0001 max mem: 52150 +Epoch: [27] [7140/7269] eta: 0:00:55 lr: 0.000003 loss: 0.5424 (0.5154) time: 0.4925 data: 0.0001 max mem: 52150 +Epoch: [27] [7150/7269] eta: 0:00:51 lr: 0.000003 loss: 0.5107 (0.5153) time: 0.4924 data: 0.0001 max mem: 52150 +Epoch: [27] [7160/7269] eta: 0:00:47 lr: 0.000003 loss: 0.5015 (0.5153) time: 0.4346 data: 0.0001 max mem: 52150 +Epoch: [27] [7170/7269] eta: 0:00:42 lr: 0.000003 loss: 0.5452 (0.5154) time: 0.4347 data: 0.0002 max mem: 52150 +Epoch: [27] [7180/7269] eta: 0:00:38 lr: 0.000003 loss: 0.5397 (0.5154) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [27] [7190/7269] eta: 0:00:34 lr: 0.000003 loss: 0.5209 (0.5154) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [27] [7200/7269] eta: 0:00:29 lr: 0.000003 loss: 0.5296 (0.5154) time: 0.4191 data: 0.0002 max mem: 52150 +Epoch: [27] [7210/7269] eta: 0:00:25 lr: 0.000003 loss: 0.5387 (0.5154) time: 0.4199 data: 0.0001 max mem: 52150 +Epoch: [27] [7220/7269] eta: 0:00:21 lr: 0.000003 loss: 0.5387 (0.5154) time: 0.4235 data: 0.0002 max mem: 52150 +Epoch: [27] [7230/7269] eta: 0:00:16 lr: 0.000003 loss: 0.5415 (0.5154) time: 0.4860 data: 0.0002 max mem: 52150 +Epoch: [27] [7240/7269] eta: 0:00:12 lr: 0.000003 loss: 0.5521 (0.5155) time: 0.4354 data: 0.0001 max mem: 52150 +Epoch: [27] [7250/7269] eta: 0:00:08 lr: 0.000003 loss: 0.5326 (0.5155) time: 0.3772 data: 0.0041 max mem: 52150 +Epoch: [27] [7260/7269] eta: 0:00:03 lr: 0.000003 loss: 0.5326 (0.5155) time: 0.4193 data: 0.0040 max mem: 52150 +Epoch: [27] [7268/7269] eta: 0:00:00 lr: 0.000003 loss: 0.5345 (0.5155) time: 0.4656 data: 0.0040 max mem: 52150 +Epoch: [27] Total time: 0:52:21 (0.4322 s / it) +Averaged stats: lr: 0.000003 loss: 0.5345 (0.5158) +Test: [ 0/1403] eta: 1:17:23 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0842 (0.0842) acc1: 100.0000 (100.0000) acc5: 100.0000 (100.0000) time: 3.3098 data: 3.2128 max mem: 52150 +Test: [ 10/1403] eta: 0:09:01 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1529 (0.1906) acc1: 98.4375 (94.4602) acc5: 100.0000 (100.0000) time: 0.3885 data: 0.2923 max mem: 52150 +Test: [ 20/1403] eta: 0:05:44 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1294 (0.1714) acc1: 98.4375 (95.0149) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 30/1403] eta: 0:04:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0846 (0.1553) acc1: 100.0000 (95.7157) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 40/1403] eta: 0:03:58 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0915 (0.1508) acc1: 100.0000 (96.2652) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 50/1403] eta: 0:03:35 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1064 (0.1609) acc1: 100.0000 (95.6495) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 60/1403] eta: 0:03:20 acc1_pos: 90.625 acc5_pos: 100.000 loss: 0.1100 (0.1567) acc1: 100.0000 (96.0297) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 70/1403] eta: 0:03:08 acc1_pos: 93.750 acc5_pos: 100.000 loss: 0.1269 (0.1623) acc1: 98.4375 (95.7526) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 80/1403] eta: 0:03:00 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.1548 (0.1708) acc1: 95.3125 (95.2932) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 90/1403] eta: 0:02:53 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.1814 (0.1918) acc1: 93.7500 (94.4025) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 100/1403] eta: 0:02:47 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1655 (0.1888) acc1: 95.3125 (94.6009) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 110/1403] eta: 0:02:42 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1274 (0.1853) acc1: 98.4375 (94.7917) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 120/1403] eta: 0:02:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0735 (0.1761) acc1: 100.0000 (95.2092) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 130/1403] eta: 0:02:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0735 (0.1692) acc1: 100.0000 (95.5153) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 140/1403] eta: 0:02:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0702 (0.1622) acc1: 100.0000 (95.8333) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 150/1403] eta: 0:02:27 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0690 (0.1562) acc1: 100.0000 (96.0886) acc5: 100.0000 (100.0000) time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 160/1403] eta: 0:02:24 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0709 (0.1511) acc1: 100.0000 (96.3218) acc5: 100.0000 (100.0000) time: 0.0969 data: 0.0003 max mem: 52150 +Test: [ 170/1403] eta: 0:02:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0716 (0.1471) acc1: 100.0000 (96.5186) acc5: 100.0000 (100.0000) time: 0.0969 data: 0.0003 max mem: 52150 +Test: [ 180/1403] eta: 0:02:19 acc1_pos: 92.188 acc5_pos: 100.000 loss: 0.0771 (0.1489) acc1: 100.0000 (96.4434) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 190/1403] eta: 0:02:17 acc1_pos: 92.188 acc5_pos: 100.000 loss: 0.1929 (0.1541) acc1: 93.7500 (96.1469) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 200/1403] eta: 0:02:15 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1929 (0.1633) acc1: 92.1875 (95.7167) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 210/1403] eta: 0:02:13 acc1_pos: 82.812 acc5_pos: 100.000 loss: 0.2419 (0.1710) acc1: 89.0625 (95.3199) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 220/1403] eta: 0:02:11 acc1_pos: 90.625 acc5_pos: 100.000 loss: 0.2717 (0.1775) acc1: 90.6250 (95.0297) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 230/1403] eta: 0:02:09 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.2140 (0.1738) acc1: 93.7500 (95.1907) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 240/1403] eta: 0:02:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0713 (0.1696) acc1: 100.0000 (95.3773) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 250/1403] eta: 0:02:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0714 (0.1660) acc1: 100.0000 (95.5428) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 260/1403] eta: 0:02:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0721 (0.1627) acc1: 100.0000 (95.6956) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 270/1403] eta: 0:02:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0733 (0.1596) acc1: 100.0000 (95.8314) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 280/1403] eta: 0:02:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0732 (0.1565) acc1: 100.0000 (95.9742) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 290/1403] eta: 0:01:59 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0694 (0.1536) acc1: 100.0000 (96.1072) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 300/1403] eta: 0:01:58 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0685 (0.1509) acc1: 100.0000 (96.2313) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 310/1403] eta: 0:01:56 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0707 (0.1486) acc1: 100.0000 (96.3525) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 320/1403] eta: 0:01:55 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0966 (0.1471) acc1: 100.0000 (96.4661) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 330/1403] eta: 0:01:53 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1031 (0.1458) acc1: 100.0000 (96.5729) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 340/1403] eta: 0:01:52 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1023 (0.1445) acc1: 100.0000 (96.6734) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 350/1403] eta: 0:01:51 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1011 (0.1433) acc1: 100.0000 (96.7682) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 360/1403] eta: 0:01:49 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1040 (0.1423) acc1: 100.0000 (96.8534) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 370/1403] eta: 0:01:48 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1033 (0.1412) acc1: 100.0000 (96.9382) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 380/1403] eta: 0:01:47 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1009 (0.1402) acc1: 100.0000 (97.0185) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 390/1403] eta: 0:01:46 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1012 (0.1413) acc1: 100.0000 (96.9629) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 400/1403] eta: 0:01:44 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2435 (0.1456) acc1: 90.6250 (96.7269) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 410/1403] eta: 0:01:43 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2882 (0.1493) acc1: 87.5000 (96.5138) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 420/1403] eta: 0:01:42 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2541 (0.1513) acc1: 89.0625 (96.3814) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 430/1403] eta: 0:01:41 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2416 (0.1544) acc1: 90.6250 (96.2297) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 440/1403] eta: 0:01:40 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3108 (0.1582) acc1: 89.0625 (96.0565) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 450/1403] eta: 0:01:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2772 (0.1609) acc1: 89.0625 (95.9223) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 460/1403] eta: 0:01:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2624 (0.1637) acc1: 90.6250 (95.7599) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 470/1403] eta: 0:01:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2287 (0.1630) acc1: 92.1875 (95.7836) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 480/1403] eta: 0:01:35 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0629 (0.1610) acc1: 100.0000 (95.8712) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 490/1403] eta: 0:01:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0630 (0.1590) acc1: 100.0000 (95.9553) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 500/1403] eta: 0:01:32 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0638 (0.1571) acc1: 100.0000 (96.0361) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 510/1403] eta: 0:01:31 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0643 (0.1553) acc1: 100.0000 (96.1136) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 520/1403] eta: 0:01:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0682 (0.1537) acc1: 100.0000 (96.1882) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 530/1403] eta: 0:01:29 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0605 (0.1519) acc1: 100.0000 (96.2600) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 540/1403] eta: 0:01:28 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0600 (0.1502) acc1: 100.0000 (96.3291) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 550/1403] eta: 0:01:27 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0610 (0.1491) acc1: 100.0000 (96.3674) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 560/1403] eta: 0:01:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0872 (0.1483) acc1: 98.4375 (96.3959) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 570/1403] eta: 0:01:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0924 (0.1474) acc1: 98.4375 (96.4399) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [ 580/1403] eta: 0:01:24 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0979 (0.1470) acc1: 98.4375 (96.4474) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0974 data: 0.0003 max mem: 52150 +Test: [ 590/1403] eta: 0:01:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1207 (0.1470) acc1: 96.8750 (96.4361) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0974 data: 0.0003 max mem: 52150 +Test: [ 600/1403] eta: 0:01:21 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1369 (0.1472) acc1: 95.3125 (96.3836) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0972 data: 0.0003 max mem: 52150 +Test: [ 610/1403] eta: 0:01:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1228 (0.1467) acc1: 95.3125 (96.3866) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0971 data: 0.0003 max mem: 52150 +Test: [ 620/1403] eta: 0:01:19 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1288 (0.1477) acc1: 95.3125 (96.3013) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [ 630/1403] eta: 0:01:18 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1531 (0.1478) acc1: 93.7500 (96.2782) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0973 data: 0.0003 max mem: 52150 +Test: [ 640/1403] eta: 0:01:17 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1336 (0.1476) acc1: 95.3125 (96.2900) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0973 data: 0.0003 max mem: 52150 +Test: [ 650/1403] eta: 0:01:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1313 (0.1473) acc1: 96.8750 (96.2990) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [ 660/1403] eta: 0:01:15 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1011 (0.1464) acc1: 98.4375 (96.3408) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [ 670/1403] eta: 0:01:14 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0695 (0.1453) acc1: 100.0000 (96.3883) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0971 data: 0.0003 max mem: 52150 +Test: [ 680/1403] eta: 0:01:13 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0836 (0.1465) acc1: 98.4375 (96.3083) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0973 data: 0.0003 max mem: 52150 +Test: [ 690/1403] eta: 0:01:12 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2385 (0.1508) acc1: 89.0625 (96.0519) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0971 data: 0.0003 max mem: 52150 +Test: [ 700/1403] eta: 0:01:11 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1671 (0.1511) acc1: 92.1875 (96.0325) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 710/1403] eta: 0:01:10 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3269 (0.1564) acc1: 82.8125 (95.7630) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 720/1403] eta: 0:01:09 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3269 (0.1566) acc1: 81.2500 (95.7459) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 730/1403] eta: 0:01:08 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0662 (0.1556) acc1: 100.0000 (95.7892) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 740/1403] eta: 0:01:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0746 (0.1548) acc1: 98.4375 (95.8186) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 750/1403] eta: 0:01:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1120 (0.1562) acc1: 96.8750 (95.7432) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 760/1403] eta: 0:01:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3180 (0.1599) acc1: 82.8125 (95.5137) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 770/1403] eta: 0:01:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4419 (0.1641) acc1: 78.1250 (95.2618) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 780/1403] eta: 0:01:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4311 (0.1667) acc1: 79.6875 (95.1124) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0970 data: 0.0003 max mem: 52150 +Test: [ 790/1403] eta: 0:01:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3072 (0.1695) acc1: 82.8125 (94.9708) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 800/1403] eta: 0:01:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2332 (0.1700) acc1: 89.0625 (94.9516) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0970 data: 0.0003 max mem: 52150 +Test: [ 810/1403] eta: 0:00:59 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2485 (0.1733) acc1: 89.0625 (94.7904) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.0974 data: 0.0003 max mem: 52150 +Test: [ 820/1403] eta: 0:00:58 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6642 (0.1826) acc1: 70.3125 (94.3990) acc5: 100.0000 (100.0000) acc1_neg: 40.625 acc5_neg: 100.000 time: 0.0977 data: 0.0003 max mem: 52150 +Test: [ 830/1403] eta: 0:00:57 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.7757 (0.1862) acc1: 68.7500 (94.2313) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0974 data: 0.0003 max mem: 52150 +Test: [ 840/1403] eta: 0:00:56 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5528 (0.1949) acc1: 68.7500 (93.8726) acc5: 100.0000 (100.0000) acc1_neg: 50.000 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [ 850/1403] eta: 0:00:55 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.1057 (0.2065) acc1: 54.6875 (93.3846) acc5: 100.0000 (100.0000) acc1_neg: 53.125 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 860/1403] eta: 0:00:54 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.0820 (0.2104) acc1: 54.6875 (93.2219) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 870/1403] eta: 0:00:53 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1397 (0.2094) acc1: 95.3125 (93.2621) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0970 data: 0.0003 max mem: 52150 +Test: [ 880/1403] eta: 0:00:52 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0984 (0.2082) acc1: 96.8750 (93.3102) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [ 890/1403] eta: 0:00:51 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0895 (0.2068) acc1: 98.4375 (93.3747) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 900/1403] eta: 0:00:50 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0742 (0.2054) acc1: 100.0000 (93.4430) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 910/1403] eta: 0:00:49 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0787 (0.2043) acc1: 98.4375 (93.4876) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 920/1403] eta: 0:00:48 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0952 (0.2033) acc1: 98.4375 (93.5295) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 930/1403] eta: 0:00:47 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0901 (0.2021) acc1: 98.4375 (93.5872) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 940/1403] eta: 0:00:46 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1103 (0.2021) acc1: 98.4375 (93.5690) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 950/1403] eta: 0:00:45 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2227 (0.2029) acc1: 89.0625 (93.5150) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 960/1403] eta: 0:00:44 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2728 (0.2034) acc1: 87.5000 (93.4752) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 970/1403] eta: 0:00:43 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2272 (0.2042) acc1: 90.6250 (93.4346) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 980/1403] eta: 0:00:42 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2483 (0.2050) acc1: 89.0625 (93.3725) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 990/1403] eta: 0:00:41 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2351 (0.2053) acc1: 90.6250 (93.3527) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [1000/1403] eta: 0:00:40 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2282 (0.2060) acc1: 90.6250 (93.3129) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1010/1403] eta: 0:00:39 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2595 (0.2067) acc1: 90.6250 (93.2755) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1020/1403] eta: 0:00:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3989 (0.2129) acc1: 82.8125 (92.9358) acc5: 100.0000 (100.0000) acc1_neg: 31.250 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [1030/1403] eta: 0:00:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5056 (0.2147) acc1: 62.5000 (92.8255) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1040/1403] eta: 0:00:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2330 (0.2146) acc1: 90.6250 (92.8269) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [1050/1403] eta: 0:00:35 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1732 (0.2143) acc1: 93.7500 (92.8446) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [1060/1403] eta: 0:00:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1401 (0.2137) acc1: 95.3125 (92.8723) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1070/1403] eta: 0:00:33 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1671 (0.2143) acc1: 93.7500 (92.8411) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0971 data: 0.0003 max mem: 52150 +Test: [1080/1403] eta: 0:00:32 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3873 (0.2163) acc1: 84.3750 (92.7397) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0971 data: 0.0003 max mem: 52150 +Test: [1090/1403] eta: 0:00:31 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4277 (0.2184) acc1: 81.2500 (92.6401) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1100/1403] eta: 0:00:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3737 (0.2189) acc1: 84.3750 (92.6161) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1110/1403] eta: 0:00:29 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2683 (0.2193) acc1: 90.6250 (92.5968) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0966 data: 0.0002 max mem: 52150 +Test: [1120/1403] eta: 0:00:28 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2705 (0.2198) acc1: 90.6250 (92.5736) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0966 data: 0.0002 max mem: 52150 +Test: [1130/1403] eta: 0:00:27 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2724 (0.2205) acc1: 89.0625 (92.5398) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1140/1403] eta: 0:00:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2399 (0.2207) acc1: 90.6250 (92.5257) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1150/1403] eta: 0:00:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2360 (0.2210) acc1: 90.6250 (92.5038) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0966 data: 0.0002 max mem: 52150 +Test: [1160/1403] eta: 0:00:24 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2619 (0.2216) acc1: 90.6250 (92.4809) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1170/1403] eta: 0:00:23 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2619 (0.2215) acc1: 90.6250 (92.4837) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0002 max mem: 52150 +Test: [1180/1403] eta: 0:00:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0695 (0.2202) acc1: 100.0000 (92.5474) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1190/1403] eta: 0:00:21 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0682 (0.2189) acc1: 100.0000 (92.6099) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1200/1403] eta: 0:00:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0684 (0.2176) acc1: 100.0000 (92.6715) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1210/1403] eta: 0:00:19 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0677 (0.2164) acc1: 100.0000 (92.7320) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52150 +Test: [1220/1403] eta: 0:00:18 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0676 (0.2152) acc1: 100.0000 (92.7915) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52150 +Test: [1230/1403] eta: 0:00:17 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0681 (0.2140) acc1: 100.0000 (92.8501) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52150 +Test: [1240/1403] eta: 0:00:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0682 (0.2128) acc1: 100.0000 (92.9077) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52150 +Test: [1250/1403] eta: 0:00:15 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0683 (0.2117) acc1: 100.0000 (92.9644) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52150 +Test: [1260/1403] eta: 0:00:14 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0743 (0.2106) acc1: 100.0000 (93.0202) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52150 +Test: [1270/1403] eta: 0:00:13 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0753 (0.2096) acc1: 100.0000 (93.0751) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52150 +Test: [1280/1403] eta: 0:00:12 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0747 (0.2085) acc1: 100.0000 (93.1291) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52150 +Test: [1290/1403] eta: 0:00:11 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0747 (0.2075) acc1: 100.0000 (93.1824) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52150 +Test: [1300/1403] eta: 0:00:10 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0751 (0.2065) acc1: 100.0000 (93.2348) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52150 +Test: [1310/1403] eta: 0:00:09 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0755 (0.2055) acc1: 100.0000 (93.2864) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0002 max mem: 52150 +Test: [1320/1403] eta: 0:00:08 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0738 (0.2045) acc1: 100.0000 (93.3372) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52150 +Test: [1330/1403] eta: 0:00:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0738 (0.2035) acc1: 100.0000 (93.3849) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1340/1403] eta: 0:00:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0689 (0.2025) acc1: 100.0000 (93.4319) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52150 +Test: [1350/1403] eta: 0:00:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0689 (0.2015) acc1: 100.0000 (93.4771) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52150 +Test: [1360/1403] eta: 0:00:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0668 (0.2006) acc1: 100.0000 (93.5250) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52150 +Test: [1370/1403] eta: 0:00:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0692 (0.1997) acc1: 100.0000 (93.5688) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0002 max mem: 52150 +Test: [1380/1403] eta: 0:00:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0690 (0.1987) acc1: 100.0000 (93.6142) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0963 data: 0.0002 max mem: 52150 +Test: [1390/1403] eta: 0:00:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0655 (0.1978) acc1: 100.0000 (93.6590) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0002 max mem: 52150 +Test: [1400/1403] eta: 0:00:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0667 (0.1969) acc1: 100.0000 (93.6998) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0001 max mem: 52150 +Test: [1402/1403] eta: 0:00:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0668 (0.1967) acc1: 100.0000 (93.7044) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0922 data: 0.0001 max mem: 52150 +Test: Total time: 0:02:19 (0.0993 s / it) +* Overall Acc@1 93.704 Acc@5 100.000 + target == real Acc@1 96.320 Acc@5 100.000 + target == fake Acc@1 92.967 Acc@5 100.000 + loss 0.197 +Accuracy of the network on the 89729 test images: 93.7% +Max accuracy: 94.33% +Epoch: [28] [ 0/7269] eta: 11:17:49 lr: 0.000003 loss: 0.3375 (0.3375) time: 5.5950 data: 5.2105 max mem: 52150 +Epoch: [28] [ 10/7269] eta: 1:42:49 lr: 0.000003 loss: 0.5320 (0.5011) time: 0.8499 data: 0.4738 max mem: 52150 +Epoch: [28] [ 20/7269] eta: 1:15:09 lr: 0.000003 loss: 0.5320 (0.5139) time: 0.3734 data: 0.0001 max mem: 52150 +Epoch: [28] [ 30/7269] eta: 1:05:19 lr: 0.000003 loss: 0.5279 (0.5117) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [28] [ 40/7269] eta: 1:00:15 lr: 0.000003 loss: 0.5116 (0.5084) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [28] [ 50/7269] eta: 0:59:30 lr: 0.000003 loss: 0.4888 (0.5081) time: 0.4222 data: 0.0001 max mem: 52150 +Epoch: [28] [ 60/7269] eta: 0:58:52 lr: 0.000003 loss: 0.5174 (0.5092) time: 0.4694 data: 0.0002 max mem: 52150 +Epoch: [28] [ 70/7269] eta: 0:56:47 lr: 0.000003 loss: 0.5159 (0.5045) time: 0.4191 data: 0.0002 max mem: 52150 +Epoch: [28] [ 80/7269] eta: 0:56:40 lr: 0.000003 loss: 0.4585 (0.5013) time: 0.4210 data: 0.0002 max mem: 52150 +Epoch: [28] [ 90/7269] eta: 0:55:14 lr: 0.000003 loss: 0.4862 (0.5022) time: 0.4206 data: 0.0002 max mem: 52150 +Epoch: [28] [ 100/7269] eta: 0:58:12 lr: 0.000003 loss: 0.5297 (0.5045) time: 0.5448 data: 0.0002 max mem: 52150 +Epoch: [28] [ 110/7269] eta: 0:56:53 lr: 0.000003 loss: 0.5288 (0.5076) time: 0.5450 data: 0.0002 max mem: 52150 +Epoch: [28] [ 120/7269] eta: 0:57:00 lr: 0.000003 loss: 0.5406 (0.5120) time: 0.4341 data: 0.0001 max mem: 52150 +Epoch: [28] [ 130/7269] eta: 0:55:57 lr: 0.000003 loss: 0.5447 (0.5123) time: 0.4341 data: 0.0002 max mem: 52150 +Epoch: [28] [ 140/7269] eta: 0:55:02 lr: 0.000003 loss: 0.5203 (0.5116) time: 0.3711 data: 0.0002 max mem: 52150 +Epoch: [28] [ 150/7269] eta: 0:55:01 lr: 0.000003 loss: 0.5257 (0.5131) time: 0.4212 data: 0.0002 max mem: 52150 +Epoch: [28] [ 160/7269] eta: 0:54:15 lr: 0.000003 loss: 0.5433 (0.5120) time: 0.4209 data: 0.0001 max mem: 52150 +Epoch: [28] [ 170/7269] eta: 0:53:35 lr: 0.000003 loss: 0.5384 (0.5137) time: 0.3709 data: 0.0002 max mem: 52150 +Epoch: [28] [ 180/7269] eta: 0:53:47 lr: 0.000003 loss: 0.5333 (0.5133) time: 0.4333 data: 0.0002 max mem: 52150 +Epoch: [28] [ 190/7269] eta: 0:53:11 lr: 0.000003 loss: 0.4796 (0.5110) time: 0.4335 data: 0.0002 max mem: 52150 +Epoch: [28] [ 200/7269] eta: 0:53:14 lr: 0.000003 loss: 0.5133 (0.5128) time: 0.4212 data: 0.0002 max mem: 52150 +Epoch: [28] [ 210/7269] eta: 0:53:14 lr: 0.000003 loss: 0.5577 (0.5151) time: 0.4686 data: 0.0002 max mem: 52150 +Epoch: [28] [ 220/7269] eta: 0:53:15 lr: 0.000003 loss: 0.5300 (0.5155) time: 0.4678 data: 0.0002 max mem: 52150 +Epoch: [28] [ 230/7269] eta: 0:53:24 lr: 0.000003 loss: 0.5093 (0.5138) time: 0.4833 data: 0.0002 max mem: 52150 +Epoch: [28] [ 240/7269] eta: 0:52:55 lr: 0.000003 loss: 0.5049 (0.5138) time: 0.4347 data: 0.0002 max mem: 52150 +Epoch: [28] [ 250/7269] eta: 0:52:28 lr: 0.000003 loss: 0.5131 (0.5137) time: 0.3712 data: 0.0001 max mem: 52150 +Epoch: [28] [ 260/7269] eta: 0:52:29 lr: 0.000003 loss: 0.5298 (0.5140) time: 0.4197 data: 0.0002 max mem: 52150 +Epoch: [28] [ 270/7269] eta: 0:52:04 lr: 0.000003 loss: 0.5298 (0.5131) time: 0.4199 data: 0.0002 max mem: 52150 +Epoch: [28] [ 280/7269] eta: 0:52:12 lr: 0.000003 loss: 0.5334 (0.5145) time: 0.4332 data: 0.0002 max mem: 52150 +Epoch: [28] [ 290/7269] eta: 0:52:18 lr: 0.000003 loss: 0.5582 (0.5160) time: 0.4940 data: 0.0002 max mem: 52150 +Epoch: [28] [ 300/7269] eta: 0:51:55 lr: 0.000003 loss: 0.5370 (0.5158) time: 0.4325 data: 0.0002 max mem: 52150 +Epoch: [28] [ 310/7269] eta: 0:51:34 lr: 0.000003 loss: 0.5116 (0.5153) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [28] [ 320/7269] eta: 0:51:14 lr: 0.000003 loss: 0.5106 (0.5148) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [28] [ 330/7269] eta: 0:51:20 lr: 0.000003 loss: 0.4870 (0.5146) time: 0.4330 data: 0.0002 max mem: 52150 +Epoch: [28] [ 340/7269] eta: 0:51:01 lr: 0.000003 loss: 0.5312 (0.5151) time: 0.4329 data: 0.0002 max mem: 52150 +Epoch: [28] [ 350/7269] eta: 0:51:22 lr: 0.000003 loss: 0.5818 (0.5166) time: 0.4705 data: 0.0002 max mem: 52150 +Epoch: [28] [ 360/7269] eta: 0:51:03 lr: 0.000003 loss: 0.5387 (0.5162) time: 0.4704 data: 0.0002 max mem: 52150 +Epoch: [28] [ 370/7269] eta: 0:50:45 lr: 0.000003 loss: 0.5256 (0.5166) time: 0.3710 data: 0.0002 max mem: 52150 +Epoch: [28] [ 380/7269] eta: 0:50:50 lr: 0.000003 loss: 0.5546 (0.5179) time: 0.4321 data: 0.0002 max mem: 52150 +Epoch: [28] [ 390/7269] eta: 0:50:51 lr: 0.000003 loss: 0.5174 (0.5177) time: 0.4819 data: 0.0002 max mem: 52150 +Epoch: [28] [ 400/7269] eta: 0:50:34 lr: 0.000003 loss: 0.5171 (0.5181) time: 0.4212 data: 0.0002 max mem: 52150 +Epoch: [28] [ 410/7269] eta: 0:50:18 lr: 0.000003 loss: 0.5158 (0.5176) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [28] [ 420/7269] eta: 0:50:38 lr: 0.000003 loss: 0.4827 (0.5166) time: 0.4808 data: 0.0002 max mem: 52150 +Epoch: [28] [ 430/7269] eta: 0:50:22 lr: 0.000003 loss: 0.5053 (0.5172) time: 0.4808 data: 0.0002 max mem: 52150 +Epoch: [28] [ 440/7269] eta: 0:50:07 lr: 0.000003 loss: 0.5357 (0.5170) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [28] [ 450/7269] eta: 0:50:26 lr: 0.000003 loss: 0.5219 (0.5171) time: 0.4827 data: 0.0002 max mem: 52150 +Epoch: [28] [ 460/7269] eta: 0:50:11 lr: 0.000003 loss: 0.5445 (0.5177) time: 0.4828 data: 0.0002 max mem: 52150 +Epoch: [28] [ 470/7269] eta: 0:49:56 lr: 0.000003 loss: 0.5216 (0.5168) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [28] [ 480/7269] eta: 0:49:55 lr: 0.000003 loss: 0.4697 (0.5165) time: 0.4191 data: 0.0001 max mem: 52150 +Epoch: [28] [ 490/7269] eta: 0:49:41 lr: 0.000003 loss: 0.5191 (0.5164) time: 0.4189 data: 0.0002 max mem: 52150 +Epoch: [28] [ 500/7269] eta: 0:49:28 lr: 0.000003 loss: 0.5323 (0.5169) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [28] [ 510/7269] eta: 0:49:14 lr: 0.000003 loss: 0.5335 (0.5173) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [28] [ 520/7269] eta: 0:49:01 lr: 0.000003 loss: 0.5313 (0.5175) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [28] [ 530/7269] eta: 0:49:00 lr: 0.000003 loss: 0.5147 (0.5168) time: 0.4160 data: 0.0002 max mem: 52150 +Epoch: [28] [ 540/7269] eta: 0:48:48 lr: 0.000003 loss: 0.4573 (0.5162) time: 0.4156 data: 0.0001 max mem: 52150 +Epoch: [28] [ 550/7269] eta: 0:48:36 lr: 0.000003 loss: 0.5204 (0.5166) time: 0.3711 data: 0.0001 max mem: 52150 +Epoch: [28] [ 560/7269] eta: 0:48:35 lr: 0.000003 loss: 0.5449 (0.5170) time: 0.4169 data: 0.0002 max mem: 52150 +Epoch: [28] [ 570/7269] eta: 0:48:34 lr: 0.000003 loss: 0.5179 (0.5164) time: 0.4651 data: 0.0002 max mem: 52150 +Epoch: [28] [ 580/7269] eta: 0:48:34 lr: 0.000003 loss: 0.4864 (0.5161) time: 0.4682 data: 0.0002 max mem: 52150 +Epoch: [28] [ 590/7269] eta: 0:48:50 lr: 0.000003 loss: 0.5178 (0.5161) time: 0.5450 data: 0.0002 max mem: 52150 +Epoch: [28] [ 600/7269] eta: 0:49:03 lr: 0.000003 loss: 0.4990 (0.5158) time: 0.6075 data: 0.0002 max mem: 52150 +Epoch: [28] [ 610/7269] eta: 0:48:51 lr: 0.000003 loss: 0.5154 (0.5162) time: 0.4827 data: 0.0002 max mem: 52150 +Epoch: [28] [ 620/7269] eta: 0:48:39 lr: 0.000003 loss: 0.5154 (0.5160) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [28] [ 630/7269] eta: 0:48:48 lr: 0.000003 loss: 0.5159 (0.5159) time: 0.4685 data: 0.0002 max mem: 52150 +Epoch: [28] [ 640/7269] eta: 0:48:37 lr: 0.000003 loss: 0.5159 (0.5158) time: 0.4680 data: 0.0002 max mem: 52150 +Epoch: [28] [ 650/7269] eta: 0:48:25 lr: 0.000003 loss: 0.5409 (0.5163) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [28] [ 660/7269] eta: 0:48:14 lr: 0.000003 loss: 0.5050 (0.5159) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [28] [ 670/7269] eta: 0:48:03 lr: 0.000003 loss: 0.5095 (0.5164) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [28] [ 680/7269] eta: 0:47:53 lr: 0.000003 loss: 0.5513 (0.5168) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [28] [ 690/7269] eta: 0:47:54 lr: 0.000003 loss: 0.5370 (0.5168) time: 0.4324 data: 0.0002 max mem: 52150 +Epoch: [28] [ 700/7269] eta: 0:47:43 lr: 0.000003 loss: 0.5364 (0.5170) time: 0.4326 data: 0.0002 max mem: 52150 +Epoch: [28] [ 710/7269] eta: 0:47:33 lr: 0.000003 loss: 0.5451 (0.5173) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [28] [ 720/7269] eta: 0:47:40 lr: 0.000003 loss: 0.5319 (0.5172) time: 0.4652 data: 0.0002 max mem: 52150 +Epoch: [28] [ 730/7269] eta: 0:47:30 lr: 0.000003 loss: 0.5009 (0.5168) time: 0.4654 data: 0.0002 max mem: 52150 +Epoch: [28] [ 740/7269] eta: 0:47:39 lr: 0.000003 loss: 0.5093 (0.5167) time: 0.4794 data: 0.0002 max mem: 52150 +Epoch: [28] [ 750/7269] eta: 0:47:39 lr: 0.000003 loss: 0.5161 (0.5164) time: 0.5406 data: 0.0002 max mem: 52150 +Epoch: [28] [ 760/7269] eta: 0:47:29 lr: 0.000003 loss: 0.4689 (0.5154) time: 0.4333 data: 0.0002 max mem: 52150 +Epoch: [28] [ 770/7269] eta: 0:47:29 lr: 0.000003 loss: 0.4942 (0.5158) time: 0.4308 data: 0.0002 max mem: 52150 +Epoch: [28] [ 780/7269] eta: 0:47:26 lr: 0.000003 loss: 0.5256 (0.5156) time: 0.4752 data: 0.0001 max mem: 52150 +Epoch: [28] [ 790/7269] eta: 0:47:17 lr: 0.000003 loss: 0.4777 (0.5153) time: 0.4165 data: 0.0001 max mem: 52150 +Epoch: [28] [ 800/7269] eta: 0:47:14 lr: 0.000003 loss: 0.4801 (0.5153) time: 0.4186 data: 0.0001 max mem: 52150 +Epoch: [28] [ 810/7269] eta: 0:47:05 lr: 0.000003 loss: 0.5204 (0.5153) time: 0.4183 data: 0.0002 max mem: 52150 +Epoch: [28] [ 820/7269] eta: 0:47:03 lr: 0.000003 loss: 0.5289 (0.5153) time: 0.4224 data: 0.0002 max mem: 52150 +Epoch: [28] [ 830/7269] eta: 0:46:54 lr: 0.000003 loss: 0.4922 (0.5146) time: 0.4229 data: 0.0002 max mem: 52150 +Epoch: [28] [ 840/7269] eta: 0:46:44 lr: 0.000003 loss: 0.5182 (0.5150) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [28] [ 850/7269] eta: 0:46:42 lr: 0.000003 loss: 0.5423 (0.5151) time: 0.4187 data: 0.0002 max mem: 52150 +Epoch: [28] [ 860/7269] eta: 0:46:33 lr: 0.000003 loss: 0.5241 (0.5150) time: 0.4188 data: 0.0002 max mem: 52150 +Epoch: [28] [ 870/7269] eta: 0:46:39 lr: 0.000003 loss: 0.5133 (0.5148) time: 0.4770 data: 0.0002 max mem: 52150 +Epoch: [28] [ 880/7269] eta: 0:46:30 lr: 0.000003 loss: 0.5078 (0.5150) time: 0.4770 data: 0.0002 max mem: 52150 +Epoch: [28] [ 890/7269] eta: 0:46:21 lr: 0.000003 loss: 0.5066 (0.5148) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [28] [ 900/7269] eta: 0:46:21 lr: 0.000003 loss: 0.5196 (0.5151) time: 0.4324 data: 0.0002 max mem: 52150 +Epoch: [28] [ 910/7269] eta: 0:46:12 lr: 0.000003 loss: 0.5352 (0.5149) time: 0.4325 data: 0.0002 max mem: 52150 +Epoch: [28] [ 920/7269] eta: 0:46:10 lr: 0.000003 loss: 0.4706 (0.5146) time: 0.4178 data: 0.0001 max mem: 52150 +Epoch: [28] [ 930/7269] eta: 0:46:07 lr: 0.000003 loss: 0.4879 (0.5149) time: 0.4633 data: 0.0001 max mem: 52150 +Epoch: [28] [ 940/7269] eta: 0:45:58 lr: 0.000003 loss: 0.5345 (0.5150) time: 0.4171 data: 0.0001 max mem: 52150 +Epoch: [28] [ 950/7269] eta: 0:45:50 lr: 0.000003 loss: 0.4829 (0.5143) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [28] [ 960/7269] eta: 0:45:41 lr: 0.000003 loss: 0.4804 (0.5143) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [28] [ 970/7269] eta: 0:45:51 lr: 0.000003 loss: 0.4837 (0.5138) time: 0.5103 data: 0.0002 max mem: 52150 +Epoch: [28] [ 980/7269] eta: 0:45:42 lr: 0.000003 loss: 0.5184 (0.5140) time: 0.5112 data: 0.0002 max mem: 52150 +Epoch: [28] [ 990/7269] eta: 0:45:34 lr: 0.000003 loss: 0.5267 (0.5139) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [28] [1000/7269] eta: 0:45:25 lr: 0.000003 loss: 0.5291 (0.5140) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [28] [1010/7269] eta: 0:45:17 lr: 0.000003 loss: 0.5459 (0.5141) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [28] [1020/7269] eta: 0:45:17 lr: 0.000003 loss: 0.5483 (0.5145) time: 0.4334 data: 0.0002 max mem: 52150 +Epoch: [28] [1030/7269] eta: 0:45:08 lr: 0.000003 loss: 0.5003 (0.5140) time: 0.4333 data: 0.0002 max mem: 52150 +Epoch: [28] [1040/7269] eta: 0:45:00 lr: 0.000003 loss: 0.4915 (0.5139) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [28] [1050/7269] eta: 0:44:52 lr: 0.000003 loss: 0.5168 (0.5139) time: 0.3711 data: 0.0002 max mem: 52150 +Epoch: [28] [1060/7269] eta: 0:44:57 lr: 0.000003 loss: 0.5173 (0.5138) time: 0.4795 data: 0.0002 max mem: 52150 +Epoch: [28] [1070/7269] eta: 0:44:56 lr: 0.000003 loss: 0.5346 (0.5142) time: 0.5398 data: 0.0002 max mem: 52150 +Epoch: [28] [1080/7269] eta: 0:45:00 lr: 0.000003 loss: 0.5248 (0.5140) time: 0.5372 data: 0.0002 max mem: 52150 +Epoch: [28] [1090/7269] eta: 0:44:52 lr: 0.000003 loss: 0.5248 (0.5142) time: 0.4768 data: 0.0002 max mem: 52150 +Epoch: [28] [1100/7269] eta: 0:44:44 lr: 0.000003 loss: 0.5256 (0.5140) time: 0.3712 data: 0.0002 max mem: 52150 +Epoch: [28] [1110/7269] eta: 0:44:48 lr: 0.000003 loss: 0.5252 (0.5141) time: 0.4778 data: 0.0002 max mem: 52150 +Epoch: [28] [1120/7269] eta: 0:44:45 lr: 0.000003 loss: 0.5204 (0.5140) time: 0.5244 data: 0.0002 max mem: 52150 +Epoch: [28] [1130/7269] eta: 0:44:37 lr: 0.000003 loss: 0.5376 (0.5143) time: 0.4181 data: 0.0002 max mem: 52150 +Epoch: [28] [1140/7269] eta: 0:44:29 lr: 0.000003 loss: 0.5469 (0.5141) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [28] [1150/7269] eta: 0:44:26 lr: 0.000003 loss: 0.4772 (0.5138) time: 0.4172 data: 0.0002 max mem: 52150 +Epoch: [28] [1160/7269] eta: 0:44:25 lr: 0.000003 loss: 0.4992 (0.5137) time: 0.4796 data: 0.0002 max mem: 52150 +Epoch: [28] [1170/7269] eta: 0:44:18 lr: 0.000003 loss: 0.5196 (0.5139) time: 0.4343 data: 0.0002 max mem: 52150 +Epoch: [28] [1180/7269] eta: 0:44:10 lr: 0.000003 loss: 0.5179 (0.5137) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [28] [1190/7269] eta: 0:44:02 lr: 0.000003 loss: 0.5320 (0.5138) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [28] [1200/7269] eta: 0:43:59 lr: 0.000003 loss: 0.5386 (0.5137) time: 0.4169 data: 0.0002 max mem: 52150 +Epoch: [28] [1210/7269] eta: 0:43:58 lr: 0.000003 loss: 0.5586 (0.5141) time: 0.4776 data: 0.0002 max mem: 52150 +Epoch: [28] [1220/7269] eta: 0:43:50 lr: 0.000003 loss: 0.5413 (0.5140) time: 0.4323 data: 0.0002 max mem: 52150 +Epoch: [28] [1230/7269] eta: 0:43:43 lr: 0.000003 loss: 0.5310 (0.5143) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [28] [1240/7269] eta: 0:43:36 lr: 0.000003 loss: 0.5470 (0.5145) time: 0.3710 data: 0.0002 max mem: 52150 +Epoch: [28] [1250/7269] eta: 0:43:28 lr: 0.000003 loss: 0.5104 (0.5143) time: 0.3709 data: 0.0002 max mem: 52150 +Epoch: [28] [1260/7269] eta: 0:43:21 lr: 0.000003 loss: 0.5450 (0.5147) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [28] [1270/7269] eta: 0:43:24 lr: 0.000003 loss: 0.5364 (0.5144) time: 0.4790 data: 0.0002 max mem: 52150 +Epoch: [28] [1280/7269] eta: 0:43:21 lr: 0.000003 loss: 0.5124 (0.5146) time: 0.5235 data: 0.0002 max mem: 52150 +Epoch: [28] [1290/7269] eta: 0:43:18 lr: 0.000003 loss: 0.5279 (0.5147) time: 0.4651 data: 0.0002 max mem: 52150 +Epoch: [28] [1300/7269] eta: 0:43:15 lr: 0.000003 loss: 0.5331 (0.5148) time: 0.4666 data: 0.0002 max mem: 52150 +Epoch: [28] [1310/7269] eta: 0:43:14 lr: 0.000003 loss: 0.5337 (0.5148) time: 0.4821 data: 0.0002 max mem: 52150 +Epoch: [28] [1320/7269] eta: 0:43:06 lr: 0.000003 loss: 0.5164 (0.5147) time: 0.4360 data: 0.0002 max mem: 52150 +Epoch: [28] [1330/7269] eta: 0:42:59 lr: 0.000003 loss: 0.5115 (0.5146) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [28] [1340/7269] eta: 0:42:52 lr: 0.000003 loss: 0.5320 (0.5149) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [28] [1350/7269] eta: 0:42:51 lr: 0.000003 loss: 0.5320 (0.5149) time: 0.4341 data: 0.0002 max mem: 52150 +Epoch: [28] [1360/7269] eta: 0:42:43 lr: 0.000003 loss: 0.5240 (0.5150) time: 0.4342 data: 0.0002 max mem: 52150 +Epoch: [28] [1370/7269] eta: 0:42:40 lr: 0.000003 loss: 0.4928 (0.5146) time: 0.4180 data: 0.0002 max mem: 52150 +Epoch: [28] [1380/7269] eta: 0:42:33 lr: 0.000003 loss: 0.4965 (0.5148) time: 0.4180 data: 0.0002 max mem: 52150 +Epoch: [28] [1390/7269] eta: 0:42:34 lr: 0.000003 loss: 0.5311 (0.5150) time: 0.4632 data: 0.0002 max mem: 52150 +Epoch: [28] [1400/7269] eta: 0:42:27 lr: 0.000003 loss: 0.5101 (0.5148) time: 0.4635 data: 0.0002 max mem: 52150 +Epoch: [28] [1410/7269] eta: 0:42:20 lr: 0.000003 loss: 0.4928 (0.5150) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [28] [1420/7269] eta: 0:42:13 lr: 0.000003 loss: 0.5134 (0.5149) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [28] [1430/7269] eta: 0:42:11 lr: 0.000003 loss: 0.5289 (0.5151) time: 0.4309 data: 0.0002 max mem: 52150 +Epoch: [28] [1440/7269] eta: 0:42:09 lr: 0.000003 loss: 0.5032 (0.5149) time: 0.4911 data: 0.0002 max mem: 52150 +Epoch: [28] [1450/7269] eta: 0:42:03 lr: 0.000003 loss: 0.4856 (0.5147) time: 0.4315 data: 0.0002 max mem: 52150 +Epoch: [28] [1460/7269] eta: 0:41:56 lr: 0.000003 loss: 0.4853 (0.5145) time: 0.3712 data: 0.0002 max mem: 52150 +Epoch: [28] [1470/7269] eta: 0:41:49 lr: 0.000003 loss: 0.4990 (0.5146) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [28] [1480/7269] eta: 0:41:42 lr: 0.000003 loss: 0.5389 (0.5146) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [28] [1490/7269] eta: 0:41:36 lr: 0.000003 loss: 0.5102 (0.5144) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [28] [1500/7269] eta: 0:41:36 lr: 0.000003 loss: 0.4603 (0.5142) time: 0.4631 data: 0.0002 max mem: 52150 +Epoch: [28] [1510/7269] eta: 0:41:33 lr: 0.000003 loss: 0.4839 (0.5141) time: 0.5095 data: 0.0002 max mem: 52150 +Epoch: [28] [1520/7269] eta: 0:41:30 lr: 0.000003 loss: 0.4743 (0.5138) time: 0.4648 data: 0.0002 max mem: 52150 +Epoch: [28] [1530/7269] eta: 0:41:23 lr: 0.000003 loss: 0.4920 (0.5137) time: 0.4184 data: 0.0002 max mem: 52150 +Epoch: [28] [1540/7269] eta: 0:41:20 lr: 0.000003 loss: 0.5281 (0.5138) time: 0.4187 data: 0.0001 max mem: 52150 +Epoch: [28] [1550/7269] eta: 0:41:18 lr: 0.000003 loss: 0.5373 (0.5139) time: 0.4800 data: 0.0002 max mem: 52150 +Epoch: [28] [1560/7269] eta: 0:41:11 lr: 0.000003 loss: 0.5247 (0.5139) time: 0.4327 data: 0.0002 max mem: 52150 +Epoch: [28] [1570/7269] eta: 0:41:09 lr: 0.000003 loss: 0.4903 (0.5137) time: 0.4306 data: 0.0002 max mem: 52150 +Epoch: [28] [1580/7269] eta: 0:41:07 lr: 0.000003 loss: 0.5115 (0.5138) time: 0.4925 data: 0.0002 max mem: 52150 +Epoch: [28] [1590/7269] eta: 0:41:00 lr: 0.000003 loss: 0.5149 (0.5137) time: 0.4333 data: 0.0002 max mem: 52150 +Epoch: [28] [1600/7269] eta: 0:40:54 lr: 0.000003 loss: 0.5265 (0.5137) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [28] [1610/7269] eta: 0:40:51 lr: 0.000003 loss: 0.5089 (0.5136) time: 0.4171 data: 0.0002 max mem: 52150 +Epoch: [28] [1620/7269] eta: 0:40:44 lr: 0.000003 loss: 0.5089 (0.5137) time: 0.4172 data: 0.0002 max mem: 52150 +Epoch: [28] [1630/7269] eta: 0:40:42 lr: 0.000003 loss: 0.5413 (0.5138) time: 0.4342 data: 0.0002 max mem: 52150 +Epoch: [28] [1640/7269] eta: 0:40:36 lr: 0.000003 loss: 0.5413 (0.5139) time: 0.4342 data: 0.0002 max mem: 52150 +Epoch: [28] [1650/7269] eta: 0:40:29 lr: 0.000003 loss: 0.5413 (0.5139) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [28] [1660/7269] eta: 0:40:30 lr: 0.000003 loss: 0.5413 (0.5141) time: 0.4796 data: 0.0002 max mem: 52150 +Epoch: [28] [1670/7269] eta: 0:40:28 lr: 0.000003 loss: 0.5329 (0.5140) time: 0.5436 data: 0.0002 max mem: 52150 +Epoch: [28] [1680/7269] eta: 0:40:22 lr: 0.000003 loss: 0.5115 (0.5139) time: 0.4363 data: 0.0002 max mem: 52150 +Epoch: [28] [1690/7269] eta: 0:40:15 lr: 0.000003 loss: 0.5241 (0.5140) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [28] [1700/7269] eta: 0:40:09 lr: 0.000003 loss: 0.5367 (0.5140) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [28] [1710/7269] eta: 0:40:03 lr: 0.000003 loss: 0.4738 (0.5136) time: 0.3712 data: 0.0002 max mem: 52150 +Epoch: [28] [1720/7269] eta: 0:39:56 lr: 0.000003 loss: 0.4738 (0.5136) time: 0.3710 data: 0.0002 max mem: 52150 +Epoch: [28] [1730/7269] eta: 0:39:54 lr: 0.000003 loss: 0.5055 (0.5136) time: 0.4318 data: 0.0002 max mem: 52150 +Epoch: [28] [1740/7269] eta: 0:39:51 lr: 0.000003 loss: 0.5055 (0.5136) time: 0.4790 data: 0.0002 max mem: 52150 +Epoch: [28] [1750/7269] eta: 0:39:44 lr: 0.000003 loss: 0.5034 (0.5136) time: 0.4187 data: 0.0002 max mem: 52150 +Epoch: [28] [1760/7269] eta: 0:39:38 lr: 0.000003 loss: 0.5021 (0.5135) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [28] [1770/7269] eta: 0:39:32 lr: 0.000003 loss: 0.5021 (0.5133) time: 0.3712 data: 0.0002 max mem: 52150 +Epoch: [28] [1780/7269] eta: 0:39:29 lr: 0.000003 loss: 0.5207 (0.5135) time: 0.4307 data: 0.0002 max mem: 52150 +Epoch: [28] [1790/7269] eta: 0:39:26 lr: 0.000003 loss: 0.5223 (0.5134) time: 0.4766 data: 0.0002 max mem: 52150 +Epoch: [28] [1800/7269] eta: 0:39:23 lr: 0.000003 loss: 0.5047 (0.5133) time: 0.4645 data: 0.0002 max mem: 52150 +Epoch: [28] [1810/7269] eta: 0:39:17 lr: 0.000003 loss: 0.5128 (0.5134) time: 0.4189 data: 0.0002 max mem: 52150 +Epoch: [28] [1820/7269] eta: 0:39:11 lr: 0.000003 loss: 0.4994 (0.5133) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [28] [1830/7269] eta: 0:39:10 lr: 0.000003 loss: 0.4989 (0.5131) time: 0.4626 data: 0.0002 max mem: 52150 +Epoch: [28] [1840/7269] eta: 0:39:04 lr: 0.000003 loss: 0.4996 (0.5132) time: 0.4624 data: 0.0001 max mem: 52150 +Epoch: [28] [1850/7269] eta: 0:39:04 lr: 0.000003 loss: 0.5107 (0.5130) time: 0.4773 data: 0.0001 max mem: 52150 +Epoch: [28] [1860/7269] eta: 0:39:01 lr: 0.000003 loss: 0.5092 (0.5129) time: 0.5407 data: 0.0001 max mem: 52150 +Epoch: [28] [1870/7269] eta: 0:38:55 lr: 0.000003 loss: 0.5296 (0.5130) time: 0.4352 data: 0.0002 max mem: 52150 +Epoch: [28] [1880/7269] eta: 0:38:56 lr: 0.000003 loss: 0.5296 (0.5130) time: 0.4940 data: 0.0002 max mem: 52150 +Epoch: [28] [1890/7269] eta: 0:38:50 lr: 0.000003 loss: 0.5130 (0.5132) time: 0.4933 data: 0.0001 max mem: 52150 +Epoch: [28] [1900/7269] eta: 0:38:44 lr: 0.000003 loss: 0.5449 (0.5134) time: 0.3712 data: 0.0002 max mem: 52150 +Epoch: [28] [1910/7269] eta: 0:38:41 lr: 0.000003 loss: 0.5299 (0.5133) time: 0.4304 data: 0.0002 max mem: 52150 +Epoch: [28] [1920/7269] eta: 0:38:41 lr: 0.000003 loss: 0.5245 (0.5135) time: 0.5389 data: 0.0002 max mem: 52150 +Epoch: [28] [1930/7269] eta: 0:38:35 lr: 0.000003 loss: 0.5297 (0.5137) time: 0.4804 data: 0.0002 max mem: 52150 +Epoch: [28] [1940/7269] eta: 0:38:33 lr: 0.000003 loss: 0.5363 (0.5137) time: 0.4328 data: 0.0002 max mem: 52150 +Epoch: [28] [1950/7269] eta: 0:38:27 lr: 0.000003 loss: 0.5194 (0.5136) time: 0.4329 data: 0.0002 max mem: 52150 +Epoch: [28] [1960/7269] eta: 0:38:21 lr: 0.000003 loss: 0.5165 (0.5137) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [28] [1970/7269] eta: 0:38:15 lr: 0.000003 loss: 0.5559 (0.5141) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [28] [1980/7269] eta: 0:38:11 lr: 0.000003 loss: 0.5456 (0.5140) time: 0.4165 data: 0.0002 max mem: 52150 +Epoch: [28] [1990/7269] eta: 0:38:07 lr: 0.000003 loss: 0.5199 (0.5141) time: 0.4627 data: 0.0002 max mem: 52150 +Epoch: [28] [2000/7269] eta: 0:38:04 lr: 0.000003 loss: 0.5455 (0.5141) time: 0.4634 data: 0.0001 max mem: 52150 +Epoch: [28] [2010/7269] eta: 0:38:01 lr: 0.000003 loss: 0.5344 (0.5140) time: 0.4764 data: 0.0002 max mem: 52150 +Epoch: [28] [2020/7269] eta: 0:37:57 lr: 0.000003 loss: 0.5491 (0.5141) time: 0.4768 data: 0.0002 max mem: 52150 +Epoch: [28] [2030/7269] eta: 0:37:54 lr: 0.000003 loss: 0.5252 (0.5140) time: 0.4639 data: 0.0002 max mem: 52150 +Epoch: [28] [2040/7269] eta: 0:37:48 lr: 0.000003 loss: 0.5308 (0.5142) time: 0.4180 data: 0.0002 max mem: 52150 +Epoch: [28] [2050/7269] eta: 0:37:42 lr: 0.000003 loss: 0.5382 (0.5142) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [28] [2060/7269] eta: 0:37:39 lr: 0.000003 loss: 0.5279 (0.5142) time: 0.4321 data: 0.0002 max mem: 52150 +Epoch: [28] [2070/7269] eta: 0:37:33 lr: 0.000003 loss: 0.5279 (0.5144) time: 0.4320 data: 0.0002 max mem: 52150 +Epoch: [28] [2080/7269] eta: 0:37:27 lr: 0.000003 loss: 0.5245 (0.5143) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [28] [2090/7269] eta: 0:37:22 lr: 0.000003 loss: 0.5297 (0.5145) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [28] [2100/7269] eta: 0:37:16 lr: 0.000003 loss: 0.5613 (0.5146) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [28] [2110/7269] eta: 0:37:13 lr: 0.000003 loss: 0.5552 (0.5148) time: 0.4346 data: 0.0002 max mem: 52150 +Epoch: [28] [2120/7269] eta: 0:37:10 lr: 0.000003 loss: 0.5047 (0.5147) time: 0.4944 data: 0.0002 max mem: 52150 +Epoch: [28] [2130/7269] eta: 0:37:04 lr: 0.000003 loss: 0.5270 (0.5147) time: 0.4315 data: 0.0002 max mem: 52150 +Epoch: [28] [2140/7269] eta: 0:36:59 lr: 0.000003 loss: 0.5388 (0.5148) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [28] [2150/7269] eta: 0:36:53 lr: 0.000003 loss: 0.5111 (0.5147) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [28] [2160/7269] eta: 0:36:49 lr: 0.000003 loss: 0.5001 (0.5146) time: 0.4187 data: 0.0002 max mem: 52150 +Epoch: [28] [2170/7269] eta: 0:36:43 lr: 0.000003 loss: 0.5366 (0.5148) time: 0.4187 data: 0.0002 max mem: 52150 +Epoch: [28] [2180/7269] eta: 0:36:38 lr: 0.000003 loss: 0.5301 (0.5148) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [28] [2190/7269] eta: 0:36:32 lr: 0.000003 loss: 0.5238 (0.5148) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [28] [2200/7269] eta: 0:36:26 lr: 0.000003 loss: 0.5193 (0.5148) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [28] [2210/7269] eta: 0:36:21 lr: 0.000003 loss: 0.5361 (0.5150) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [28] [2220/7269] eta: 0:36:15 lr: 0.000003 loss: 0.5570 (0.5150) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [28] [2230/7269] eta: 0:36:09 lr: 0.000003 loss: 0.5057 (0.5149) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [28] [2240/7269] eta: 0:36:06 lr: 0.000003 loss: 0.5257 (0.5152) time: 0.4318 data: 0.0002 max mem: 52150 +Epoch: [28] [2250/7269] eta: 0:36:06 lr: 0.000003 loss: 0.5084 (0.5151) time: 0.5370 data: 0.0002 max mem: 52150 +Epoch: [28] [2260/7269] eta: 0:36:00 lr: 0.000003 loss: 0.5084 (0.5152) time: 0.4776 data: 0.0002 max mem: 52150 +Epoch: [28] [2270/7269] eta: 0:35:54 lr: 0.000003 loss: 0.5226 (0.5151) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [28] [2280/7269] eta: 0:35:53 lr: 0.000003 loss: 0.5112 (0.5151) time: 0.4804 data: 0.0002 max mem: 52150 +Epoch: [28] [2290/7269] eta: 0:35:50 lr: 0.000003 loss: 0.5409 (0.5152) time: 0.5258 data: 0.0002 max mem: 52150 +Epoch: [28] [2300/7269] eta: 0:35:46 lr: 0.000003 loss: 0.5097 (0.5152) time: 0.4612 data: 0.0001 max mem: 52150 +Epoch: [28] [2310/7269] eta: 0:35:40 lr: 0.000003 loss: 0.4932 (0.5152) time: 0.4161 data: 0.0002 max mem: 52150 +Epoch: [28] [2320/7269] eta: 0:35:37 lr: 0.000003 loss: 0.5274 (0.5153) time: 0.4162 data: 0.0002 max mem: 52150 +Epoch: [28] [2330/7269] eta: 0:35:31 lr: 0.000003 loss: 0.5477 (0.5154) time: 0.4160 data: 0.0002 max mem: 52150 +Epoch: [28] [2340/7269] eta: 0:35:26 lr: 0.000003 loss: 0.5288 (0.5154) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [28] [2350/7269] eta: 0:35:25 lr: 0.000003 loss: 0.5115 (0.5152) time: 0.4962 data: 0.0002 max mem: 52150 +Epoch: [28] [2360/7269] eta: 0:35:20 lr: 0.000003 loss: 0.4474 (0.5150) time: 0.4957 data: 0.0002 max mem: 52150 +Epoch: [28] [2370/7269] eta: 0:35:19 lr: 0.000003 loss: 0.4670 (0.5149) time: 0.4804 data: 0.0002 max mem: 52150 +Epoch: [28] [2380/7269] eta: 0:35:16 lr: 0.000003 loss: 0.5184 (0.5151) time: 0.5441 data: 0.0002 max mem: 52150 +Epoch: [28] [2390/7269] eta: 0:35:10 lr: 0.000003 loss: 0.5425 (0.5151) time: 0.4353 data: 0.0001 max mem: 52150 +Epoch: [28] [2400/7269] eta: 0:35:04 lr: 0.000003 loss: 0.5061 (0.5150) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [28] [2410/7269] eta: 0:35:01 lr: 0.000003 loss: 0.5144 (0.5150) time: 0.4181 data: 0.0002 max mem: 52150 +Epoch: [28] [2420/7269] eta: 0:34:57 lr: 0.000003 loss: 0.5260 (0.5150) time: 0.4633 data: 0.0002 max mem: 52150 +Epoch: [28] [2430/7269] eta: 0:34:54 lr: 0.000003 loss: 0.5260 (0.5152) time: 0.4725 data: 0.0002 max mem: 52150 +Epoch: [28] [2440/7269] eta: 0:34:48 lr: 0.000003 loss: 0.5233 (0.5152) time: 0.4264 data: 0.0002 max mem: 52150 +Epoch: [28] [2450/7269] eta: 0:34:43 lr: 0.000003 loss: 0.5085 (0.5151) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [28] [2460/7269] eta: 0:34:42 lr: 0.000003 loss: 0.5214 (0.5152) time: 0.4945 data: 0.0002 max mem: 52150 +Epoch: [28] [2470/7269] eta: 0:34:36 lr: 0.000003 loss: 0.5046 (0.5151) time: 0.4942 data: 0.0002 max mem: 52150 +Epoch: [28] [2480/7269] eta: 0:34:33 lr: 0.000003 loss: 0.4743 (0.5149) time: 0.4170 data: 0.0002 max mem: 52150 +Epoch: [28] [2490/7269] eta: 0:34:29 lr: 0.000003 loss: 0.4902 (0.5149) time: 0.4623 data: 0.0002 max mem: 52150 +Epoch: [28] [2500/7269] eta: 0:34:31 lr: 0.000003 loss: 0.4902 (0.5148) time: 0.6193 data: 0.0002 max mem: 52150 +Epoch: [28] [2510/7269] eta: 0:34:26 lr: 0.000003 loss: 0.5282 (0.5148) time: 0.5741 data: 0.0002 max mem: 52150 +Epoch: [28] [2520/7269] eta: 0:34:20 lr: 0.000003 loss: 0.5298 (0.5148) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [28] [2530/7269] eta: 0:34:15 lr: 0.000003 loss: 0.5298 (0.5149) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [28] [2540/7269] eta: 0:34:09 lr: 0.000003 loss: 0.5029 (0.5149) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [28] [2550/7269] eta: 0:34:06 lr: 0.000003 loss: 0.5015 (0.5148) time: 0.4316 data: 0.0002 max mem: 52150 +Epoch: [28] [2560/7269] eta: 0:34:00 lr: 0.000003 loss: 0.5428 (0.5149) time: 0.4312 data: 0.0002 max mem: 52150 +Epoch: [28] [2570/7269] eta: 0:33:55 lr: 0.000003 loss: 0.5525 (0.5151) time: 0.3712 data: 0.0002 max mem: 52150 +Epoch: [28] [2580/7269] eta: 0:33:49 lr: 0.000003 loss: 0.5286 (0.5150) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [28] [2590/7269] eta: 0:33:44 lr: 0.000003 loss: 0.5095 (0.5150) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [28] [2600/7269] eta: 0:33:39 lr: 0.000003 loss: 0.5179 (0.5150) time: 0.3712 data: 0.0001 max mem: 52150 +Epoch: [28] [2610/7269] eta: 0:33:33 lr: 0.000003 loss: 0.5330 (0.5151) time: 0.3712 data: 0.0001 max mem: 52150 +Epoch: [28] [2620/7269] eta: 0:33:28 lr: 0.000003 loss: 0.5325 (0.5151) time: 0.3711 data: 0.0001 max mem: 52150 +Epoch: [28] [2630/7269] eta: 0:33:22 lr: 0.000003 loss: 0.5122 (0.5150) time: 0.3714 data: 0.0001 max mem: 52150 +Epoch: [28] [2640/7269] eta: 0:33:19 lr: 0.000003 loss: 0.5188 (0.5151) time: 0.4183 data: 0.0001 max mem: 52150 +Epoch: [28] [2650/7269] eta: 0:33:15 lr: 0.000003 loss: 0.5350 (0.5151) time: 0.4702 data: 0.0001 max mem: 52150 +Epoch: [28] [2660/7269] eta: 0:33:10 lr: 0.000003 loss: 0.5146 (0.5150) time: 0.4235 data: 0.0002 max mem: 52150 +Epoch: [28] [2670/7269] eta: 0:33:04 lr: 0.000003 loss: 0.5146 (0.5152) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [28] [2680/7269] eta: 0:32:59 lr: 0.000003 loss: 0.5411 (0.5152) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [28] [2690/7269] eta: 0:32:57 lr: 0.000003 loss: 0.5386 (0.5152) time: 0.4777 data: 0.0002 max mem: 52150 +Epoch: [28] [2700/7269] eta: 0:32:52 lr: 0.000003 loss: 0.5107 (0.5151) time: 0.4774 data: 0.0002 max mem: 52150 +Epoch: [28] [2710/7269] eta: 0:32:47 lr: 0.000003 loss: 0.4993 (0.5150) time: 0.3712 data: 0.0002 max mem: 52150 +Epoch: [28] [2720/7269] eta: 0:32:43 lr: 0.000003 loss: 0.5080 (0.5150) time: 0.4178 data: 0.0002 max mem: 52150 +Epoch: [28] [2730/7269] eta: 0:32:38 lr: 0.000003 loss: 0.5382 (0.5150) time: 0.4176 data: 0.0001 max mem: 52150 +Epoch: [28] [2740/7269] eta: 0:32:38 lr: 0.000003 loss: 0.5399 (0.5151) time: 0.5426 data: 0.0001 max mem: 52150 +Epoch: [28] [2750/7269] eta: 0:32:33 lr: 0.000003 loss: 0.5399 (0.5151) time: 0.5432 data: 0.0002 max mem: 52150 +Epoch: [28] [2760/7269] eta: 0:32:29 lr: 0.000003 loss: 0.5215 (0.5151) time: 0.4356 data: 0.0002 max mem: 52150 +Epoch: [28] [2770/7269] eta: 0:32:24 lr: 0.000003 loss: 0.5215 (0.5150) time: 0.4356 data: 0.0002 max mem: 52150 +Epoch: [28] [2780/7269] eta: 0:32:19 lr: 0.000003 loss: 0.5286 (0.5151) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [28] [2790/7269] eta: 0:32:15 lr: 0.000003 loss: 0.5286 (0.5150) time: 0.4169 data: 0.0002 max mem: 52150 +Epoch: [28] [2800/7269] eta: 0:32:10 lr: 0.000003 loss: 0.5352 (0.5152) time: 0.4163 data: 0.0002 max mem: 52150 +Epoch: [28] [2810/7269] eta: 0:32:08 lr: 0.000003 loss: 0.5627 (0.5153) time: 0.4955 data: 0.0002 max mem: 52150 +Epoch: [28] [2820/7269] eta: 0:32:03 lr: 0.000003 loss: 0.5081 (0.5152) time: 0.4960 data: 0.0002 max mem: 52150 +Epoch: [28] [2830/7269] eta: 0:32:00 lr: 0.000003 loss: 0.4856 (0.5152) time: 0.4364 data: 0.0002 max mem: 52150 +Epoch: [28] [2840/7269] eta: 0:31:54 lr: 0.000003 loss: 0.5410 (0.5152) time: 0.4362 data: 0.0002 max mem: 52150 +Epoch: [28] [2850/7269] eta: 0:31:51 lr: 0.000003 loss: 0.5233 (0.5153) time: 0.4209 data: 0.0002 max mem: 52150 +Epoch: [28] [2860/7269] eta: 0:31:45 lr: 0.000003 loss: 0.5515 (0.5154) time: 0.4209 data: 0.0002 max mem: 52150 +Epoch: [28] [2870/7269] eta: 0:31:40 lr: 0.000003 loss: 0.5515 (0.5154) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [28] [2880/7269] eta: 0:31:36 lr: 0.000003 loss: 0.5404 (0.5154) time: 0.4176 data: 0.0002 max mem: 52150 +Epoch: [28] [2890/7269] eta: 0:31:31 lr: 0.000003 loss: 0.5409 (0.5154) time: 0.4182 data: 0.0002 max mem: 52150 +Epoch: [28] [2900/7269] eta: 0:31:26 lr: 0.000003 loss: 0.5637 (0.5157) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [28] [2910/7269] eta: 0:31:21 lr: 0.000003 loss: 0.5220 (0.5156) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [28] [2920/7269] eta: 0:31:15 lr: 0.000003 loss: 0.4654 (0.5155) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [28] [2930/7269] eta: 0:31:10 lr: 0.000003 loss: 0.4860 (0.5154) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [28] [2940/7269] eta: 0:31:05 lr: 0.000003 loss: 0.5178 (0.5155) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [28] [2950/7269] eta: 0:31:01 lr: 0.000003 loss: 0.5321 (0.5155) time: 0.4173 data: 0.0001 max mem: 52150 +Epoch: [28] [2960/7269] eta: 0:30:57 lr: 0.000003 loss: 0.5133 (0.5154) time: 0.4621 data: 0.0002 max mem: 52150 +Epoch: [28] [2970/7269] eta: 0:30:52 lr: 0.000003 loss: 0.5023 (0.5153) time: 0.4167 data: 0.0002 max mem: 52150 +Epoch: [28] [2980/7269] eta: 0:30:47 lr: 0.000003 loss: 0.5251 (0.5154) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [28] [2990/7269] eta: 0:30:43 lr: 0.000003 loss: 0.5473 (0.5154) time: 0.4187 data: 0.0001 max mem: 52150 +Epoch: [28] [3000/7269] eta: 0:30:40 lr: 0.000003 loss: 0.4992 (0.5152) time: 0.4790 data: 0.0001 max mem: 52150 +Epoch: [28] [3010/7269] eta: 0:30:36 lr: 0.000003 loss: 0.4992 (0.5152) time: 0.4918 data: 0.0001 max mem: 52150 +Epoch: [28] [3020/7269] eta: 0:30:31 lr: 0.000003 loss: 0.4971 (0.5151) time: 0.4308 data: 0.0002 max mem: 52150 +Epoch: [28] [3030/7269] eta: 0:30:26 lr: 0.000003 loss: 0.4915 (0.5150) time: 0.3712 data: 0.0002 max mem: 52150 +Epoch: [28] [3040/7269] eta: 0:30:21 lr: 0.000003 loss: 0.5270 (0.5151) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [28] [3050/7269] eta: 0:30:16 lr: 0.000003 loss: 0.5364 (0.5151) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [28] [3060/7269] eta: 0:30:11 lr: 0.000003 loss: 0.5369 (0.5152) time: 0.3728 data: 0.0002 max mem: 52150 +Epoch: [28] [3070/7269] eta: 0:30:06 lr: 0.000003 loss: 0.5264 (0.5151) time: 0.3721 data: 0.0001 max mem: 52150 +Epoch: [28] [3080/7269] eta: 0:30:01 lr: 0.000003 loss: 0.5200 (0.5152) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [28] [3090/7269] eta: 0:29:55 lr: 0.000003 loss: 0.5306 (0.5152) time: 0.3722 data: 0.0001 max mem: 52150 +Epoch: [28] [3100/7269] eta: 0:29:50 lr: 0.000003 loss: 0.5515 (0.5154) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [28] [3110/7269] eta: 0:29:47 lr: 0.000003 loss: 0.5277 (0.5153) time: 0.4218 data: 0.0002 max mem: 52150 +Epoch: [28] [3120/7269] eta: 0:29:44 lr: 0.000003 loss: 0.5219 (0.5154) time: 0.5297 data: 0.0002 max mem: 52150 +Epoch: [28] [3130/7269] eta: 0:29:41 lr: 0.000003 loss: 0.5066 (0.5153) time: 0.5298 data: 0.0002 max mem: 52150 +Epoch: [28] [3140/7269] eta: 0:29:37 lr: 0.000003 loss: 0.4791 (0.5151) time: 0.4680 data: 0.0002 max mem: 52150 +Epoch: [28] [3150/7269] eta: 0:29:33 lr: 0.000003 loss: 0.4572 (0.5150) time: 0.4634 data: 0.0001 max mem: 52150 +Epoch: [28] [3160/7269] eta: 0:29:31 lr: 0.000003 loss: 0.5337 (0.5151) time: 0.5231 data: 0.0001 max mem: 52150 +Epoch: [28] [3170/7269] eta: 0:29:26 lr: 0.000003 loss: 0.5335 (0.5151) time: 0.4784 data: 0.0002 max mem: 52150 +Epoch: [28] [3180/7269] eta: 0:29:21 lr: 0.000003 loss: 0.5378 (0.5153) time: 0.3731 data: 0.0002 max mem: 52150 +Epoch: [28] [3190/7269] eta: 0:29:17 lr: 0.000003 loss: 0.5349 (0.5152) time: 0.4165 data: 0.0002 max mem: 52150 +Epoch: [28] [3200/7269] eta: 0:29:14 lr: 0.000003 loss: 0.5157 (0.5153) time: 0.5248 data: 0.0001 max mem: 52150 +Epoch: [28] [3210/7269] eta: 0:29:09 lr: 0.000003 loss: 0.5056 (0.5153) time: 0.4807 data: 0.0001 max mem: 52150 +Epoch: [28] [3220/7269] eta: 0:29:04 lr: 0.000003 loss: 0.5076 (0.5153) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [28] [3230/7269] eta: 0:29:01 lr: 0.000003 loss: 0.5201 (0.5152) time: 0.4362 data: 0.0001 max mem: 52150 +Epoch: [28] [3240/7269] eta: 0:28:56 lr: 0.000003 loss: 0.5417 (0.5153) time: 0.4364 data: 0.0001 max mem: 52150 +Epoch: [28] [3250/7269] eta: 0:28:54 lr: 0.000003 loss: 0.5443 (0.5153) time: 0.4947 data: 0.0001 max mem: 52150 +Epoch: [28] [3260/7269] eta: 0:28:49 lr: 0.000003 loss: 0.4996 (0.5152) time: 0.4949 data: 0.0001 max mem: 52150 +Epoch: [28] [3270/7269] eta: 0:28:48 lr: 0.000003 loss: 0.4349 (0.5150) time: 0.5403 data: 0.0002 max mem: 52150 +Epoch: [28] [3280/7269] eta: 0:28:43 lr: 0.000003 loss: 0.4691 (0.5150) time: 0.5395 data: 0.0002 max mem: 52150 +Epoch: [28] [3290/7269] eta: 0:28:38 lr: 0.000003 loss: 0.5165 (0.5151) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [28] [3300/7269] eta: 0:28:33 lr: 0.000003 loss: 0.5451 (0.5150) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [28] [3310/7269] eta: 0:28:28 lr: 0.000003 loss: 0.5214 (0.5150) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [28] [3320/7269] eta: 0:28:23 lr: 0.000003 loss: 0.5214 (0.5150) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [28] [3330/7269] eta: 0:28:20 lr: 0.000003 loss: 0.5060 (0.5150) time: 0.4820 data: 0.0002 max mem: 52150 +Epoch: [28] [3340/7269] eta: 0:28:17 lr: 0.000003 loss: 0.4824 (0.5148) time: 0.5445 data: 0.0002 max mem: 52150 +Epoch: [28] [3350/7269] eta: 0:28:14 lr: 0.000003 loss: 0.4920 (0.5149) time: 0.5449 data: 0.0002 max mem: 52150 +Epoch: [28] [3360/7269] eta: 0:28:09 lr: 0.000003 loss: 0.5481 (0.5150) time: 0.4820 data: 0.0002 max mem: 52150 +Epoch: [28] [3370/7269] eta: 0:28:04 lr: 0.000003 loss: 0.5538 (0.5150) time: 0.3712 data: 0.0002 max mem: 52150 +Epoch: [28] [3380/7269] eta: 0:27:59 lr: 0.000003 loss: 0.5236 (0.5150) time: 0.3724 data: 0.0001 max mem: 52150 +Epoch: [28] [3390/7269] eta: 0:27:54 lr: 0.000003 loss: 0.5082 (0.5149) time: 0.3730 data: 0.0001 max mem: 52150 +Epoch: [28] [3400/7269] eta: 0:27:50 lr: 0.000003 loss: 0.4751 (0.5148) time: 0.4349 data: 0.0001 max mem: 52150 +Epoch: [28] [3410/7269] eta: 0:27:45 lr: 0.000003 loss: 0.4758 (0.5147) time: 0.4350 data: 0.0001 max mem: 52150 +Epoch: [28] [3420/7269] eta: 0:27:42 lr: 0.000003 loss: 0.4825 (0.5146) time: 0.4179 data: 0.0002 max mem: 52150 +Epoch: [28] [3430/7269] eta: 0:27:38 lr: 0.000003 loss: 0.5308 (0.5147) time: 0.4629 data: 0.0002 max mem: 52150 +Epoch: [28] [3440/7269] eta: 0:27:33 lr: 0.000003 loss: 0.5323 (0.5147) time: 0.4170 data: 0.0001 max mem: 52150 +Epoch: [28] [3450/7269] eta: 0:27:28 lr: 0.000003 loss: 0.5371 (0.5147) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [28] [3460/7269] eta: 0:27:23 lr: 0.000003 loss: 0.5313 (0.5147) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [28] [3470/7269] eta: 0:27:19 lr: 0.000003 loss: 0.5102 (0.5148) time: 0.4181 data: 0.0001 max mem: 52150 +Epoch: [28] [3480/7269] eta: 0:27:14 lr: 0.000003 loss: 0.5059 (0.5147) time: 0.4182 data: 0.0002 max mem: 52150 +Epoch: [28] [3490/7269] eta: 0:27:09 lr: 0.000003 loss: 0.5215 (0.5147) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [28] [3500/7269] eta: 0:27:05 lr: 0.000003 loss: 0.5576 (0.5149) time: 0.4322 data: 0.0002 max mem: 52150 +Epoch: [28] [3510/7269] eta: 0:27:00 lr: 0.000003 loss: 0.5127 (0.5148) time: 0.4316 data: 0.0001 max mem: 52150 +Epoch: [28] [3520/7269] eta: 0:26:55 lr: 0.000003 loss: 0.4985 (0.5148) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [28] [3530/7269] eta: 0:26:50 lr: 0.000003 loss: 0.5057 (0.5147) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [28] [3540/7269] eta: 0:26:46 lr: 0.000003 loss: 0.5339 (0.5148) time: 0.4168 data: 0.0001 max mem: 52150 +Epoch: [28] [3550/7269] eta: 0:26:41 lr: 0.000003 loss: 0.5248 (0.5147) time: 0.4165 data: 0.0002 max mem: 52150 +Epoch: [28] [3560/7269] eta: 0:26:36 lr: 0.000003 loss: 0.4880 (0.5146) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [28] [3570/7269] eta: 0:26:32 lr: 0.000003 loss: 0.5029 (0.5147) time: 0.4183 data: 0.0002 max mem: 52150 +Epoch: [28] [3580/7269] eta: 0:26:30 lr: 0.000003 loss: 0.5473 (0.5148) time: 0.5292 data: 0.0002 max mem: 52150 +Epoch: [28] [3590/7269] eta: 0:26:25 lr: 0.000003 loss: 0.5506 (0.5148) time: 0.4825 data: 0.0001 max mem: 52150 +Epoch: [28] [3600/7269] eta: 0:26:21 lr: 0.000003 loss: 0.4777 (0.5147) time: 0.4164 data: 0.0001 max mem: 52150 +Epoch: [28] [3610/7269] eta: 0:26:16 lr: 0.000003 loss: 0.4635 (0.5147) time: 0.4166 data: 0.0001 max mem: 52150 +Epoch: [28] [3620/7269] eta: 0:26:11 lr: 0.000003 loss: 0.5300 (0.5148) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [28] [3630/7269] eta: 0:26:06 lr: 0.000003 loss: 0.5015 (0.5147) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [28] [3640/7269] eta: 0:26:02 lr: 0.000003 loss: 0.5033 (0.5146) time: 0.4181 data: 0.0002 max mem: 52150 +Epoch: [28] [3650/7269] eta: 0:25:57 lr: 0.000003 loss: 0.5044 (0.5145) time: 0.4179 data: 0.0001 max mem: 52150 +Epoch: [28] [3660/7269] eta: 0:25:52 lr: 0.000003 loss: 0.5086 (0.5146) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [28] [3670/7269] eta: 0:25:48 lr: 0.000003 loss: 0.5466 (0.5146) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [28] [3680/7269] eta: 0:25:43 lr: 0.000003 loss: 0.5048 (0.5145) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [28] [3690/7269] eta: 0:25:39 lr: 0.000003 loss: 0.4597 (0.5144) time: 0.4358 data: 0.0001 max mem: 52150 +Epoch: [28] [3700/7269] eta: 0:25:34 lr: 0.000003 loss: 0.4985 (0.5144) time: 0.4352 data: 0.0001 max mem: 52150 +Epoch: [28] [3710/7269] eta: 0:25:29 lr: 0.000003 loss: 0.4665 (0.5143) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [28] [3720/7269] eta: 0:25:26 lr: 0.000003 loss: 0.5135 (0.5143) time: 0.4322 data: 0.0001 max mem: 52150 +Epoch: [28] [3730/7269] eta: 0:25:24 lr: 0.000003 loss: 0.5238 (0.5143) time: 0.5975 data: 0.0001 max mem: 52150 +Epoch: [28] [3740/7269] eta: 0:25:20 lr: 0.000003 loss: 0.5428 (0.5143) time: 0.5822 data: 0.0001 max mem: 52150 +Epoch: [28] [3750/7269] eta: 0:25:15 lr: 0.000003 loss: 0.5453 (0.5143) time: 0.4169 data: 0.0001 max mem: 52150 +Epoch: [28] [3760/7269] eta: 0:25:10 lr: 0.000003 loss: 0.5624 (0.5144) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [28] [3770/7269] eta: 0:25:05 lr: 0.000003 loss: 0.5341 (0.5144) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [28] [3780/7269] eta: 0:25:01 lr: 0.000003 loss: 0.5232 (0.5144) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [28] [3790/7269] eta: 0:24:56 lr: 0.000003 loss: 0.5553 (0.5145) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [28] [3800/7269] eta: 0:24:54 lr: 0.000003 loss: 0.5529 (0.5145) time: 0.5297 data: 0.0001 max mem: 52150 +Epoch: [28] [3810/7269] eta: 0:24:51 lr: 0.000003 loss: 0.5499 (0.5146) time: 0.6268 data: 0.0002 max mem: 52150 +Epoch: [28] [3820/7269] eta: 0:24:46 lr: 0.000003 loss: 0.5425 (0.5145) time: 0.4690 data: 0.0002 max mem: 52150 +Epoch: [28] [3830/7269] eta: 0:24:42 lr: 0.000003 loss: 0.5425 (0.5146) time: 0.4165 data: 0.0002 max mem: 52150 +Epoch: [28] [3840/7269] eta: 0:24:38 lr: 0.000003 loss: 0.5301 (0.5147) time: 0.4644 data: 0.0002 max mem: 52150 +Epoch: [28] [3850/7269] eta: 0:24:34 lr: 0.000003 loss: 0.5301 (0.5146) time: 0.4658 data: 0.0002 max mem: 52150 +Epoch: [28] [3860/7269] eta: 0:24:29 lr: 0.000003 loss: 0.5178 (0.5146) time: 0.4175 data: 0.0002 max mem: 52150 +Epoch: [28] [3870/7269] eta: 0:24:25 lr: 0.000003 loss: 0.5609 (0.5147) time: 0.4314 data: 0.0002 max mem: 52150 +Epoch: [28] [3880/7269] eta: 0:24:21 lr: 0.000003 loss: 0.5671 (0.5149) time: 0.4965 data: 0.0002 max mem: 52150 +Epoch: [28] [3890/7269] eta: 0:24:17 lr: 0.000003 loss: 0.5535 (0.5148) time: 0.4364 data: 0.0001 max mem: 52150 +Epoch: [28] [3900/7269] eta: 0:24:13 lr: 0.000003 loss: 0.5088 (0.5148) time: 0.4309 data: 0.0001 max mem: 52150 +Epoch: [28] [3910/7269] eta: 0:24:10 lr: 0.000003 loss: 0.4748 (0.5148) time: 0.5415 data: 0.0001 max mem: 52150 +Epoch: [28] [3920/7269] eta: 0:24:05 lr: 0.000003 loss: 0.4656 (0.5147) time: 0.4833 data: 0.0002 max mem: 52150 +Epoch: [28] [3930/7269] eta: 0:24:01 lr: 0.000003 loss: 0.5140 (0.5147) time: 0.4379 data: 0.0002 max mem: 52150 +Epoch: [28] [3940/7269] eta: 0:23:57 lr: 0.000003 loss: 0.5500 (0.5149) time: 0.4370 data: 0.0002 max mem: 52150 +Epoch: [28] [3950/7269] eta: 0:23:52 lr: 0.000003 loss: 0.5707 (0.5150) time: 0.4185 data: 0.0002 max mem: 52150 +Epoch: [28] [3960/7269] eta: 0:23:49 lr: 0.000003 loss: 0.5266 (0.5150) time: 0.4803 data: 0.0002 max mem: 52150 +Epoch: [28] [3970/7269] eta: 0:23:45 lr: 0.000003 loss: 0.5266 (0.5150) time: 0.4938 data: 0.0002 max mem: 52150 +Epoch: [28] [3980/7269] eta: 0:23:40 lr: 0.000003 loss: 0.5343 (0.5150) time: 0.4314 data: 0.0002 max mem: 52150 +Epoch: [28] [3990/7269] eta: 0:23:35 lr: 0.000003 loss: 0.5256 (0.5151) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [28] [4000/7269] eta: 0:23:30 lr: 0.000003 loss: 0.5243 (0.5151) time: 0.3721 data: 0.0001 max mem: 52150 +Epoch: [28] [4010/7269] eta: 0:23:26 lr: 0.000003 loss: 0.5255 (0.5151) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [28] [4020/7269] eta: 0:23:21 lr: 0.000003 loss: 0.5378 (0.5152) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [28] [4030/7269] eta: 0:23:17 lr: 0.000003 loss: 0.5543 (0.5152) time: 0.4187 data: 0.0002 max mem: 52150 +Epoch: [28] [4040/7269] eta: 0:23:13 lr: 0.000003 loss: 0.5218 (0.5152) time: 0.4820 data: 0.0001 max mem: 52150 +Epoch: [28] [4050/7269] eta: 0:23:08 lr: 0.000003 loss: 0.5422 (0.5153) time: 0.4355 data: 0.0001 max mem: 52150 +Epoch: [28] [4060/7269] eta: 0:23:04 lr: 0.000003 loss: 0.5499 (0.5154) time: 0.4326 data: 0.0001 max mem: 52150 +Epoch: [28] [4070/7269] eta: 0:23:00 lr: 0.000003 loss: 0.5452 (0.5154) time: 0.4323 data: 0.0001 max mem: 52150 +Epoch: [28] [4080/7269] eta: 0:22:56 lr: 0.000003 loss: 0.5282 (0.5154) time: 0.4165 data: 0.0002 max mem: 52150 +Epoch: [28] [4090/7269] eta: 0:22:51 lr: 0.000003 loss: 0.4937 (0.5153) time: 0.4612 data: 0.0002 max mem: 52150 +Epoch: [28] [4100/7269] eta: 0:22:47 lr: 0.000003 loss: 0.5229 (0.5154) time: 0.4166 data: 0.0002 max mem: 52150 +Epoch: [28] [4110/7269] eta: 0:22:44 lr: 0.000003 loss: 0.5464 (0.5154) time: 0.4796 data: 0.0002 max mem: 52150 +Epoch: [28] [4120/7269] eta: 0:22:39 lr: 0.000003 loss: 0.5213 (0.5154) time: 0.4797 data: 0.0002 max mem: 52150 +Epoch: [28] [4130/7269] eta: 0:22:34 lr: 0.000003 loss: 0.5232 (0.5155) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [28] [4140/7269] eta: 0:22:29 lr: 0.000003 loss: 0.5249 (0.5156) time: 0.3714 data: 0.0001 max mem: 52150 +Epoch: [28] [4150/7269] eta: 0:22:25 lr: 0.000003 loss: 0.5249 (0.5156) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [28] [4160/7269] eta: 0:22:20 lr: 0.000003 loss: 0.4926 (0.5156) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [28] [4170/7269] eta: 0:22:15 lr: 0.000003 loss: 0.4894 (0.5156) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [28] [4180/7269] eta: 0:22:11 lr: 0.000003 loss: 0.5196 (0.5156) time: 0.4356 data: 0.0001 max mem: 52150 +Epoch: [28] [4190/7269] eta: 0:22:06 lr: 0.000003 loss: 0.5341 (0.5156) time: 0.4352 data: 0.0001 max mem: 52150 +Epoch: [28] [4200/7269] eta: 0:22:02 lr: 0.000003 loss: 0.5434 (0.5156) time: 0.3712 data: 0.0002 max mem: 52150 +Epoch: [28] [4210/7269] eta: 0:21:57 lr: 0.000003 loss: 0.5373 (0.5156) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [28] [4220/7269] eta: 0:21:52 lr: 0.000003 loss: 0.5373 (0.5156) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [28] [4230/7269] eta: 0:21:48 lr: 0.000003 loss: 0.4981 (0.5155) time: 0.4184 data: 0.0001 max mem: 52150 +Epoch: [28] [4240/7269] eta: 0:21:43 lr: 0.000003 loss: 0.4981 (0.5155) time: 0.4182 data: 0.0002 max mem: 52150 +Epoch: [28] [4250/7269] eta: 0:21:39 lr: 0.000003 loss: 0.4985 (0.5155) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [28] [4260/7269] eta: 0:21:34 lr: 0.000003 loss: 0.5365 (0.5156) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [28] [4270/7269] eta: 0:21:30 lr: 0.000003 loss: 0.5191 (0.5155) time: 0.4176 data: 0.0002 max mem: 52150 +Epoch: [28] [4280/7269] eta: 0:21:26 lr: 0.000003 loss: 0.5014 (0.5155) time: 0.4647 data: 0.0002 max mem: 52150 +Epoch: [28] [4290/7269] eta: 0:21:22 lr: 0.000003 loss: 0.5462 (0.5156) time: 0.5100 data: 0.0001 max mem: 52150 +Epoch: [28] [4300/7269] eta: 0:21:18 lr: 0.000003 loss: 0.5172 (0.5155) time: 0.4629 data: 0.0001 max mem: 52150 +Epoch: [28] [4310/7269] eta: 0:21:13 lr: 0.000003 loss: 0.5308 (0.5157) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [28] [4320/7269] eta: 0:21:08 lr: 0.000003 loss: 0.5416 (0.5156) time: 0.3728 data: 0.0001 max mem: 52150 +Epoch: [28] [4330/7269] eta: 0:21:04 lr: 0.000003 loss: 0.4964 (0.5155) time: 0.4215 data: 0.0001 max mem: 52150 +Epoch: [28] [4340/7269] eta: 0:21:00 lr: 0.000003 loss: 0.5127 (0.5156) time: 0.4205 data: 0.0002 max mem: 52150 +Epoch: [28] [4350/7269] eta: 0:20:56 lr: 0.000003 loss: 0.5090 (0.5156) time: 0.4169 data: 0.0002 max mem: 52150 +Epoch: [28] [4360/7269] eta: 0:20:52 lr: 0.000003 loss: 0.5040 (0.5156) time: 0.5280 data: 0.0001 max mem: 52150 +Epoch: [28] [4370/7269] eta: 0:20:49 lr: 0.000003 loss: 0.5370 (0.5156) time: 0.6034 data: 0.0001 max mem: 52150 +Epoch: [28] [4380/7269] eta: 0:20:44 lr: 0.000003 loss: 0.5370 (0.5157) time: 0.4926 data: 0.0001 max mem: 52150 +Epoch: [28] [4390/7269] eta: 0:20:40 lr: 0.000003 loss: 0.5309 (0.5157) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [28] [4400/7269] eta: 0:20:36 lr: 0.000003 loss: 0.5087 (0.5157) time: 0.4320 data: 0.0001 max mem: 52150 +Epoch: [28] [4410/7269] eta: 0:20:33 lr: 0.000003 loss: 0.5206 (0.5157) time: 0.6003 data: 0.0002 max mem: 52150 +Epoch: [28] [4420/7269] eta: 0:20:29 lr: 0.000003 loss: 0.5224 (0.5157) time: 0.5404 data: 0.0002 max mem: 52150 +Epoch: [28] [4430/7269] eta: 0:20:24 lr: 0.000003 loss: 0.5231 (0.5157) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [28] [4440/7269] eta: 0:20:19 lr: 0.000003 loss: 0.5266 (0.5157) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [28] [4450/7269] eta: 0:20:16 lr: 0.000003 loss: 0.5271 (0.5157) time: 0.4762 data: 0.0002 max mem: 52150 +Epoch: [28] [4460/7269] eta: 0:20:13 lr: 0.000003 loss: 0.5271 (0.5158) time: 0.5913 data: 0.0002 max mem: 52150 +Epoch: [28] [4470/7269] eta: 0:20:10 lr: 0.000003 loss: 0.5334 (0.5158) time: 0.6081 data: 0.0002 max mem: 52150 +Epoch: [28] [4480/7269] eta: 0:20:06 lr: 0.000003 loss: 0.5383 (0.5158) time: 0.5586 data: 0.0002 max mem: 52150 +Epoch: [28] [4490/7269] eta: 0:20:01 lr: 0.000003 loss: 0.5214 (0.5157) time: 0.4373 data: 0.0002 max mem: 52150 +Epoch: [28] [4500/7269] eta: 0:19:56 lr: 0.000003 loss: 0.5318 (0.5158) time: 0.3722 data: 0.0001 max mem: 52150 +Epoch: [28] [4510/7269] eta: 0:19:52 lr: 0.000003 loss: 0.5546 (0.5159) time: 0.3725 data: 0.0001 max mem: 52150 +Epoch: [28] [4520/7269] eta: 0:19:48 lr: 0.000003 loss: 0.5133 (0.5159) time: 0.4889 data: 0.0001 max mem: 52150 +Epoch: [28] [4530/7269] eta: 0:19:44 lr: 0.000003 loss: 0.5407 (0.5160) time: 0.5378 data: 0.0001 max mem: 52150 +Epoch: [28] [4540/7269] eta: 0:19:40 lr: 0.000003 loss: 0.5394 (0.5160) time: 0.4215 data: 0.0001 max mem: 52150 +Epoch: [28] [4550/7269] eta: 0:19:35 lr: 0.000003 loss: 0.5204 (0.5160) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [28] [4560/7269] eta: 0:19:31 lr: 0.000003 loss: 0.5119 (0.5160) time: 0.4180 data: 0.0002 max mem: 52150 +Epoch: [28] [4570/7269] eta: 0:19:26 lr: 0.000003 loss: 0.4958 (0.5159) time: 0.4175 data: 0.0002 max mem: 52150 +Epoch: [28] [4580/7269] eta: 0:19:22 lr: 0.000003 loss: 0.4923 (0.5159) time: 0.4167 data: 0.0002 max mem: 52150 +Epoch: [28] [4590/7269] eta: 0:19:17 lr: 0.000003 loss: 0.4786 (0.5158) time: 0.4165 data: 0.0002 max mem: 52150 +Epoch: [28] [4600/7269] eta: 0:19:13 lr: 0.000003 loss: 0.5235 (0.5159) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [28] [4610/7269] eta: 0:19:08 lr: 0.000003 loss: 0.5555 (0.5159) time: 0.3722 data: 0.0001 max mem: 52150 +Epoch: [28] [4620/7269] eta: 0:19:03 lr: 0.000003 loss: 0.5133 (0.5158) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [28] [4630/7269] eta: 0:18:59 lr: 0.000003 loss: 0.5295 (0.5159) time: 0.4345 data: 0.0002 max mem: 52150 +Epoch: [28] [4640/7269] eta: 0:18:55 lr: 0.000003 loss: 0.5559 (0.5160) time: 0.4347 data: 0.0002 max mem: 52150 +Epoch: [28] [4650/7269] eta: 0:18:50 lr: 0.000003 loss: 0.4772 (0.5159) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [28] [4660/7269] eta: 0:18:45 lr: 0.000003 loss: 0.4978 (0.5159) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [28] [4670/7269] eta: 0:18:41 lr: 0.000003 loss: 0.5221 (0.5160) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [28] [4680/7269] eta: 0:18:36 lr: 0.000003 loss: 0.5060 (0.5159) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [28] [4690/7269] eta: 0:18:32 lr: 0.000003 loss: 0.5060 (0.5159) time: 0.4167 data: 0.0001 max mem: 52150 +Epoch: [28] [4700/7269] eta: 0:18:27 lr: 0.000003 loss: 0.5213 (0.5159) time: 0.4166 data: 0.0002 max mem: 52150 +Epoch: [28] [4710/7269] eta: 0:18:23 lr: 0.000003 loss: 0.5067 (0.5159) time: 0.3711 data: 0.0002 max mem: 52150 +Epoch: [28] [4720/7269] eta: 0:18:18 lr: 0.000003 loss: 0.5067 (0.5159) time: 0.3710 data: 0.0002 max mem: 52150 +Epoch: [28] [4730/7269] eta: 0:18:13 lr: 0.000003 loss: 0.4893 (0.5158) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [28] [4740/7269] eta: 0:18:09 lr: 0.000003 loss: 0.5308 (0.5158) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [28] [4750/7269] eta: 0:18:04 lr: 0.000003 loss: 0.5560 (0.5159) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [28] [4760/7269] eta: 0:17:59 lr: 0.000003 loss: 0.5560 (0.5160) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [28] [4770/7269] eta: 0:17:55 lr: 0.000003 loss: 0.5645 (0.5161) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [28] [4780/7269] eta: 0:17:50 lr: 0.000003 loss: 0.5063 (0.5160) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [28] [4790/7269] eta: 0:17:47 lr: 0.000003 loss: 0.5104 (0.5160) time: 0.4820 data: 0.0002 max mem: 52150 +Epoch: [28] [4800/7269] eta: 0:17:43 lr: 0.000003 loss: 0.5110 (0.5161) time: 0.5441 data: 0.0001 max mem: 52150 +Epoch: [28] [4810/7269] eta: 0:17:38 lr: 0.000003 loss: 0.5443 (0.5161) time: 0.4338 data: 0.0001 max mem: 52150 +Epoch: [28] [4820/7269] eta: 0:17:34 lr: 0.000003 loss: 0.5443 (0.5161) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [28] [4830/7269] eta: 0:17:30 lr: 0.000003 loss: 0.5255 (0.5161) time: 0.4785 data: 0.0001 max mem: 52150 +Epoch: [28] [4840/7269] eta: 0:17:25 lr: 0.000003 loss: 0.5167 (0.5160) time: 0.4788 data: 0.0001 max mem: 52150 +Epoch: [28] [4850/7269] eta: 0:17:21 lr: 0.000003 loss: 0.5316 (0.5161) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [28] [4860/7269] eta: 0:17:17 lr: 0.000003 loss: 0.5336 (0.5161) time: 0.4335 data: 0.0002 max mem: 52150 +Epoch: [28] [4870/7269] eta: 0:17:13 lr: 0.000003 loss: 0.5413 (0.5161) time: 0.4786 data: 0.0002 max mem: 52150 +Epoch: [28] [4880/7269] eta: 0:17:08 lr: 0.000003 loss: 0.5580 (0.5162) time: 0.4172 data: 0.0001 max mem: 52150 +Epoch: [28] [4890/7269] eta: 0:17:04 lr: 0.000003 loss: 0.5631 (0.5163) time: 0.3714 data: 0.0001 max mem: 52150 +Epoch: [28] [4900/7269] eta: 0:16:59 lr: 0.000003 loss: 0.5447 (0.5163) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [28] [4910/7269] eta: 0:16:55 lr: 0.000003 loss: 0.5441 (0.5164) time: 0.4159 data: 0.0001 max mem: 52150 +Epoch: [28] [4920/7269] eta: 0:16:51 lr: 0.000003 loss: 0.5414 (0.5164) time: 0.4760 data: 0.0001 max mem: 52150 +Epoch: [28] [4930/7269] eta: 0:16:47 lr: 0.000003 loss: 0.5342 (0.5164) time: 0.4943 data: 0.0002 max mem: 52150 +Epoch: [28] [4940/7269] eta: 0:16:43 lr: 0.000003 loss: 0.5137 (0.5164) time: 0.4969 data: 0.0002 max mem: 52150 +Epoch: [28] [4950/7269] eta: 0:16:40 lr: 0.000003 loss: 0.5671 (0.5166) time: 0.6113 data: 0.0002 max mem: 52150 +Epoch: [28] [4960/7269] eta: 0:16:36 lr: 0.000003 loss: 0.5818 (0.5167) time: 0.6615 data: 0.0001 max mem: 52150 +Epoch: [28] [4970/7269] eta: 0:16:32 lr: 0.000003 loss: 0.5520 (0.5167) time: 0.5301 data: 0.0001 max mem: 52150 +Epoch: [28] [4980/7269] eta: 0:16:29 lr: 0.000003 loss: 0.5124 (0.5166) time: 0.5400 data: 0.0001 max mem: 52150 +Epoch: [28] [4990/7269] eta: 0:16:24 lr: 0.000003 loss: 0.5124 (0.5166) time: 0.4943 data: 0.0001 max mem: 52150 +Epoch: [28] [5000/7269] eta: 0:16:19 lr: 0.000003 loss: 0.5469 (0.5166) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [28] [5010/7269] eta: 0:16:15 lr: 0.000003 loss: 0.5279 (0.5166) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [28] [5020/7269] eta: 0:16:11 lr: 0.000003 loss: 0.5446 (0.5167) time: 0.4793 data: 0.0001 max mem: 52150 +Epoch: [28] [5030/7269] eta: 0:16:08 lr: 0.000003 loss: 0.5417 (0.5167) time: 0.6389 data: 0.0001 max mem: 52150 +Epoch: [28] [5040/7269] eta: 0:16:04 lr: 0.000003 loss: 0.5152 (0.5167) time: 0.5314 data: 0.0001 max mem: 52150 +Epoch: [28] [5050/7269] eta: 0:15:59 lr: 0.000003 loss: 0.5152 (0.5167) time: 0.4352 data: 0.0001 max mem: 52150 +Epoch: [28] [5060/7269] eta: 0:15:55 lr: 0.000003 loss: 0.5054 (0.5166) time: 0.4355 data: 0.0001 max mem: 52150 +Epoch: [28] [5070/7269] eta: 0:15:50 lr: 0.000003 loss: 0.5209 (0.5166) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [28] [5080/7269] eta: 0:15:46 lr: 0.000003 loss: 0.5028 (0.5166) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [28] [5090/7269] eta: 0:15:42 lr: 0.000003 loss: 0.5279 (0.5167) time: 0.4181 data: 0.0002 max mem: 52150 +Epoch: [28] [5100/7269] eta: 0:15:37 lr: 0.000003 loss: 0.5279 (0.5166) time: 0.4180 data: 0.0002 max mem: 52150 +Epoch: [28] [5110/7269] eta: 0:15:32 lr: 0.000003 loss: 0.4706 (0.5165) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [28] [5120/7269] eta: 0:15:29 lr: 0.000003 loss: 0.5365 (0.5165) time: 0.4788 data: 0.0002 max mem: 52150 +Epoch: [28] [5130/7269] eta: 0:15:24 lr: 0.000003 loss: 0.5219 (0.5164) time: 0.5248 data: 0.0002 max mem: 52150 +Epoch: [28] [5140/7269] eta: 0:15:20 lr: 0.000003 loss: 0.5300 (0.5165) time: 0.4782 data: 0.0002 max mem: 52150 +Epoch: [28] [5150/7269] eta: 0:15:16 lr: 0.000003 loss: 0.5515 (0.5166) time: 0.4323 data: 0.0001 max mem: 52150 +Epoch: [28] [5160/7269] eta: 0:15:11 lr: 0.000003 loss: 0.5408 (0.5165) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [28] [5170/7269] eta: 0:15:07 lr: 0.000003 loss: 0.5150 (0.5165) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [28] [5180/7269] eta: 0:15:02 lr: 0.000003 loss: 0.5150 (0.5166) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [28] [5190/7269] eta: 0:14:58 lr: 0.000003 loss: 0.5193 (0.5165) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [28] [5200/7269] eta: 0:14:53 lr: 0.000003 loss: 0.4930 (0.5164) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [28] [5210/7269] eta: 0:14:48 lr: 0.000003 loss: 0.4843 (0.5164) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [28] [5220/7269] eta: 0:14:44 lr: 0.000003 loss: 0.4898 (0.5164) time: 0.3722 data: 0.0001 max mem: 52150 +Epoch: [28] [5230/7269] eta: 0:14:39 lr: 0.000003 loss: 0.5027 (0.5164) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [28] [5240/7269] eta: 0:14:35 lr: 0.000003 loss: 0.5222 (0.5164) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [28] [5250/7269] eta: 0:14:30 lr: 0.000003 loss: 0.5251 (0.5164) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [28] [5260/7269] eta: 0:14:26 lr: 0.000003 loss: 0.5046 (0.5164) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [28] [5270/7269] eta: 0:14:22 lr: 0.000003 loss: 0.5007 (0.5164) time: 0.4339 data: 0.0001 max mem: 52150 +Epoch: [28] [5280/7269] eta: 0:14:17 lr: 0.000003 loss: 0.5070 (0.5163) time: 0.4346 data: 0.0002 max mem: 52150 +Epoch: [28] [5290/7269] eta: 0:14:13 lr: 0.000003 loss: 0.5365 (0.5164) time: 0.4178 data: 0.0001 max mem: 52150 +Epoch: [28] [5300/7269] eta: 0:14:08 lr: 0.000003 loss: 0.5158 (0.5163) time: 0.4174 data: 0.0001 max mem: 52150 +Epoch: [28] [5310/7269] eta: 0:14:04 lr: 0.000003 loss: 0.5139 (0.5163) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [28] [5320/7269] eta: 0:14:00 lr: 0.000003 loss: 0.5462 (0.5163) time: 0.4311 data: 0.0002 max mem: 52150 +Epoch: [28] [5330/7269] eta: 0:13:55 lr: 0.000003 loss: 0.5549 (0.5164) time: 0.4316 data: 0.0002 max mem: 52150 +Epoch: [28] [5340/7269] eta: 0:13:51 lr: 0.000003 loss: 0.5489 (0.5165) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [28] [5350/7269] eta: 0:13:46 lr: 0.000003 loss: 0.5407 (0.5165) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [28] [5360/7269] eta: 0:13:42 lr: 0.000003 loss: 0.5404 (0.5165) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [28] [5370/7269] eta: 0:13:37 lr: 0.000003 loss: 0.5404 (0.5165) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [28] [5380/7269] eta: 0:13:33 lr: 0.000003 loss: 0.5318 (0.5165) time: 0.4314 data: 0.0002 max mem: 52150 +Epoch: [28] [5390/7269] eta: 0:13:29 lr: 0.000003 loss: 0.5272 (0.5166) time: 0.4313 data: 0.0001 max mem: 52150 +Epoch: [28] [5400/7269] eta: 0:13:24 lr: 0.000003 loss: 0.5272 (0.5166) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [28] [5410/7269] eta: 0:13:20 lr: 0.000003 loss: 0.5177 (0.5166) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [28] [5420/7269] eta: 0:13:15 lr: 0.000003 loss: 0.5080 (0.5166) time: 0.4309 data: 0.0002 max mem: 52150 +Epoch: [28] [5430/7269] eta: 0:13:12 lr: 0.000003 loss: 0.5080 (0.5166) time: 0.5395 data: 0.0002 max mem: 52150 +Epoch: [28] [5440/7269] eta: 0:13:07 lr: 0.000003 loss: 0.4937 (0.5165) time: 0.4806 data: 0.0002 max mem: 52150 +Epoch: [28] [5450/7269] eta: 0:13:03 lr: 0.000003 loss: 0.4937 (0.5165) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [28] [5460/7269] eta: 0:12:59 lr: 0.000003 loss: 0.5443 (0.5166) time: 0.4839 data: 0.0002 max mem: 52150 +Epoch: [28] [5470/7269] eta: 0:12:54 lr: 0.000003 loss: 0.5436 (0.5166) time: 0.4838 data: 0.0002 max mem: 52150 +Epoch: [28] [5480/7269] eta: 0:12:50 lr: 0.000003 loss: 0.5436 (0.5167) time: 0.3712 data: 0.0002 max mem: 52150 +Epoch: [28] [5490/7269] eta: 0:12:46 lr: 0.000003 loss: 0.5329 (0.5167) time: 0.4164 data: 0.0002 max mem: 52150 +Epoch: [28] [5500/7269] eta: 0:12:42 lr: 0.000003 loss: 0.5257 (0.5167) time: 0.5921 data: 0.0002 max mem: 52150 +Epoch: [28] [5510/7269] eta: 0:12:39 lr: 0.000003 loss: 0.5225 (0.5167) time: 0.7003 data: 0.0002 max mem: 52150 +Epoch: [28] [5520/7269] eta: 0:12:35 lr: 0.000003 loss: 0.5419 (0.5167) time: 0.5882 data: 0.0001 max mem: 52150 +Epoch: [28] [5530/7269] eta: 0:12:30 lr: 0.000003 loss: 0.5419 (0.5167) time: 0.4350 data: 0.0001 max mem: 52150 +Epoch: [28] [5540/7269] eta: 0:12:26 lr: 0.000003 loss: 0.5086 (0.5167) time: 0.4335 data: 0.0001 max mem: 52150 +Epoch: [28] [5550/7269] eta: 0:12:22 lr: 0.000003 loss: 0.5381 (0.5167) time: 0.4336 data: 0.0001 max mem: 52150 +Epoch: [28] [5560/7269] eta: 0:12:17 lr: 0.000003 loss: 0.5077 (0.5167) time: 0.4339 data: 0.0002 max mem: 52150 +Epoch: [28] [5570/7269] eta: 0:12:14 lr: 0.000003 loss: 0.4844 (0.5166) time: 0.6041 data: 0.0002 max mem: 52150 +Epoch: [28] [5580/7269] eta: 0:12:10 lr: 0.000003 loss: 0.5125 (0.5166) time: 0.5425 data: 0.0002 max mem: 52150 +Epoch: [28] [5590/7269] eta: 0:12:05 lr: 0.000003 loss: 0.4904 (0.5166) time: 0.3727 data: 0.0002 max mem: 52150 +Epoch: [28] [5600/7269] eta: 0:12:01 lr: 0.000003 loss: 0.4779 (0.5165) time: 0.4171 data: 0.0002 max mem: 52150 +Epoch: [28] [5610/7269] eta: 0:11:57 lr: 0.000003 loss: 0.5106 (0.5165) time: 0.4674 data: 0.0002 max mem: 52150 +Epoch: [28] [5620/7269] eta: 0:11:52 lr: 0.000003 loss: 0.5210 (0.5165) time: 0.4674 data: 0.0002 max mem: 52150 +Epoch: [28] [5630/7269] eta: 0:11:48 lr: 0.000003 loss: 0.5206 (0.5165) time: 0.4167 data: 0.0002 max mem: 52150 +Epoch: [28] [5640/7269] eta: 0:11:44 lr: 0.000003 loss: 0.5206 (0.5166) time: 0.4317 data: 0.0002 max mem: 52150 +Epoch: [28] [5650/7269] eta: 0:11:39 lr: 0.000003 loss: 0.5095 (0.5165) time: 0.4318 data: 0.0002 max mem: 52150 +Epoch: [28] [5660/7269] eta: 0:11:35 lr: 0.000003 loss: 0.4904 (0.5165) time: 0.4328 data: 0.0002 max mem: 52150 +Epoch: [28] [5670/7269] eta: 0:11:31 lr: 0.000003 loss: 0.5217 (0.5165) time: 0.4922 data: 0.0002 max mem: 52150 +Epoch: [28] [5680/7269] eta: 0:11:26 lr: 0.000003 loss: 0.5338 (0.5166) time: 0.4312 data: 0.0001 max mem: 52150 +Epoch: [28] [5690/7269] eta: 0:11:22 lr: 0.000003 loss: 0.5645 (0.5165) time: 0.4207 data: 0.0002 max mem: 52150 +Epoch: [28] [5700/7269] eta: 0:11:18 lr: 0.000003 loss: 0.5035 (0.5165) time: 0.4679 data: 0.0002 max mem: 52150 +Epoch: [28] [5710/7269] eta: 0:11:13 lr: 0.000003 loss: 0.5419 (0.5166) time: 0.4191 data: 0.0002 max mem: 52150 +Epoch: [28] [5720/7269] eta: 0:11:09 lr: 0.000003 loss: 0.5356 (0.5166) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [28] [5730/7269] eta: 0:11:05 lr: 0.000003 loss: 0.5312 (0.5166) time: 0.4202 data: 0.0002 max mem: 52150 +Epoch: [28] [5740/7269] eta: 0:11:01 lr: 0.000003 loss: 0.4999 (0.5165) time: 0.4697 data: 0.0002 max mem: 52150 +Epoch: [28] [5750/7269] eta: 0:10:56 lr: 0.000003 loss: 0.5012 (0.5166) time: 0.4212 data: 0.0002 max mem: 52150 +Epoch: [28] [5760/7269] eta: 0:10:52 lr: 0.000003 loss: 0.5652 (0.5167) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [28] [5770/7269] eta: 0:10:47 lr: 0.000003 loss: 0.5363 (0.5167) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [28] [5780/7269] eta: 0:10:43 lr: 0.000003 loss: 0.5362 (0.5167) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [28] [5790/7269] eta: 0:10:38 lr: 0.000003 loss: 0.5362 (0.5167) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [28] [5800/7269] eta: 0:10:34 lr: 0.000003 loss: 0.5209 (0.5166) time: 0.3728 data: 0.0002 max mem: 52150 +Epoch: [28] [5810/7269] eta: 0:10:29 lr: 0.000003 loss: 0.4741 (0.5166) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [28] [5820/7269] eta: 0:10:25 lr: 0.000003 loss: 0.5326 (0.5166) time: 0.4323 data: 0.0002 max mem: 52150 +Epoch: [28] [5830/7269] eta: 0:10:21 lr: 0.000003 loss: 0.5456 (0.5166) time: 0.4938 data: 0.0002 max mem: 52150 +Epoch: [28] [5840/7269] eta: 0:10:17 lr: 0.000003 loss: 0.5097 (0.5166) time: 0.4817 data: 0.0001 max mem: 52150 +Epoch: [28] [5850/7269] eta: 0:10:12 lr: 0.000003 loss: 0.4978 (0.5166) time: 0.4798 data: 0.0002 max mem: 52150 +Epoch: [28] [5860/7269] eta: 0:10:08 lr: 0.000003 loss: 0.5039 (0.5165) time: 0.4323 data: 0.0002 max mem: 52150 +Epoch: [28] [5870/7269] eta: 0:10:04 lr: 0.000003 loss: 0.5147 (0.5165) time: 0.3729 data: 0.0002 max mem: 52150 +Epoch: [28] [5880/7269] eta: 0:09:59 lr: 0.000003 loss: 0.5427 (0.5166) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [28] [5890/7269] eta: 0:09:55 lr: 0.000003 loss: 0.5154 (0.5165) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [28] [5900/7269] eta: 0:09:50 lr: 0.000003 loss: 0.4883 (0.5165) time: 0.4342 data: 0.0002 max mem: 52150 +Epoch: [28] [5910/7269] eta: 0:09:46 lr: 0.000003 loss: 0.5182 (0.5166) time: 0.4348 data: 0.0002 max mem: 52150 +Epoch: [28] [5920/7269] eta: 0:09:42 lr: 0.000003 loss: 0.5039 (0.5165) time: 0.4183 data: 0.0002 max mem: 52150 +Epoch: [28] [5930/7269] eta: 0:09:37 lr: 0.000003 loss: 0.4925 (0.5165) time: 0.4177 data: 0.0002 max mem: 52150 +Epoch: [28] [5940/7269] eta: 0:09:33 lr: 0.000003 loss: 0.5088 (0.5165) time: 0.4207 data: 0.0002 max mem: 52150 +Epoch: [28] [5950/7269] eta: 0:09:29 lr: 0.000003 loss: 0.5129 (0.5164) time: 0.4206 data: 0.0002 max mem: 52150 +Epoch: [28] [5960/7269] eta: 0:09:24 lr: 0.000003 loss: 0.5129 (0.5164) time: 0.4304 data: 0.0002 max mem: 52150 +Epoch: [28] [5970/7269] eta: 0:09:20 lr: 0.000003 loss: 0.5195 (0.5164) time: 0.5379 data: 0.0002 max mem: 52150 +Epoch: [28] [5980/7269] eta: 0:09:16 lr: 0.000003 loss: 0.5259 (0.5164) time: 0.5433 data: 0.0001 max mem: 52150 +Epoch: [28] [5990/7269] eta: 0:09:12 lr: 0.000003 loss: 0.5004 (0.5163) time: 0.4363 data: 0.0002 max mem: 52150 +Epoch: [28] [6000/7269] eta: 0:09:08 lr: 0.000003 loss: 0.5004 (0.5163) time: 0.4999 data: 0.0002 max mem: 52150 +Epoch: [28] [6010/7269] eta: 0:09:04 lr: 0.000003 loss: 0.5329 (0.5163) time: 0.4998 data: 0.0002 max mem: 52150 +Epoch: [28] [6020/7269] eta: 0:08:59 lr: 0.000003 loss: 0.5072 (0.5163) time: 0.4336 data: 0.0002 max mem: 52150 +Epoch: [28] [6030/7269] eta: 0:08:55 lr: 0.000003 loss: 0.5111 (0.5163) time: 0.5423 data: 0.0002 max mem: 52150 +Epoch: [28] [6040/7269] eta: 0:08:51 lr: 0.000003 loss: 0.5301 (0.5163) time: 0.4808 data: 0.0002 max mem: 52150 +Epoch: [28] [6050/7269] eta: 0:08:46 lr: 0.000003 loss: 0.4990 (0.5162) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [28] [6060/7269] eta: 0:08:42 lr: 0.000003 loss: 0.5038 (0.5163) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [28] [6070/7269] eta: 0:08:38 lr: 0.000003 loss: 0.5185 (0.5162) time: 0.4806 data: 0.0001 max mem: 52150 +Epoch: [28] [6080/7269] eta: 0:08:34 lr: 0.000003 loss: 0.5145 (0.5162) time: 0.4801 data: 0.0002 max mem: 52150 +Epoch: [28] [6090/7269] eta: 0:08:29 lr: 0.000003 loss: 0.5172 (0.5162) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [28] [6100/7269] eta: 0:08:25 lr: 0.000003 loss: 0.5365 (0.5163) time: 0.4820 data: 0.0002 max mem: 52150 +Epoch: [28] [6110/7269] eta: 0:08:21 lr: 0.000003 loss: 0.5217 (0.5162) time: 0.4813 data: 0.0002 max mem: 52150 +Epoch: [28] [6120/7269] eta: 0:08:16 lr: 0.000003 loss: 0.4732 (0.5162) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [28] [6130/7269] eta: 0:08:12 lr: 0.000003 loss: 0.4736 (0.5162) time: 0.4313 data: 0.0002 max mem: 52150 +Epoch: [28] [6140/7269] eta: 0:08:08 lr: 0.000003 loss: 0.5156 (0.5162) time: 0.4311 data: 0.0002 max mem: 52150 +Epoch: [28] [6150/7269] eta: 0:08:03 lr: 0.000003 loss: 0.5337 (0.5162) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [28] [6160/7269] eta: 0:07:59 lr: 0.000003 loss: 0.5416 (0.5162) time: 0.4193 data: 0.0002 max mem: 52150 +Epoch: [28] [6170/7269] eta: 0:07:54 lr: 0.000003 loss: 0.5482 (0.5163) time: 0.4197 data: 0.0002 max mem: 52150 +Epoch: [28] [6180/7269] eta: 0:07:50 lr: 0.000003 loss: 0.5505 (0.5163) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [28] [6190/7269] eta: 0:07:46 lr: 0.000003 loss: 0.5139 (0.5162) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [28] [6200/7269] eta: 0:07:42 lr: 0.000003 loss: 0.4938 (0.5162) time: 0.4807 data: 0.0002 max mem: 52150 +Epoch: [28] [6210/7269] eta: 0:07:37 lr: 0.000003 loss: 0.5030 (0.5162) time: 0.5268 data: 0.0001 max mem: 52150 +Epoch: [28] [6220/7269] eta: 0:07:33 lr: 0.000003 loss: 0.5191 (0.5162) time: 0.4190 data: 0.0001 max mem: 52150 +Epoch: [28] [6230/7269] eta: 0:07:29 lr: 0.000003 loss: 0.5274 (0.5162) time: 0.4214 data: 0.0001 max mem: 52150 +Epoch: [28] [6240/7269] eta: 0:07:25 lr: 0.000003 loss: 0.5030 (0.5162) time: 0.5288 data: 0.0002 max mem: 52150 +Epoch: [28] [6250/7269] eta: 0:07:20 lr: 0.000003 loss: 0.4798 (0.5162) time: 0.4805 data: 0.0002 max mem: 52150 +Epoch: [28] [6260/7269] eta: 0:07:16 lr: 0.000003 loss: 0.4798 (0.5161) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [28] [6270/7269] eta: 0:07:11 lr: 0.000003 loss: 0.5338 (0.5162) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [28] [6280/7269] eta: 0:07:07 lr: 0.000003 loss: 0.5292 (0.5162) time: 0.4351 data: 0.0002 max mem: 52150 +Epoch: [28] [6290/7269] eta: 0:07:03 lr: 0.000003 loss: 0.5292 (0.5162) time: 0.4828 data: 0.0002 max mem: 52150 +Epoch: [28] [6300/7269] eta: 0:06:58 lr: 0.000003 loss: 0.5500 (0.5162) time: 0.4198 data: 0.0001 max mem: 52150 +Epoch: [28] [6310/7269] eta: 0:06:54 lr: 0.000003 loss: 0.5447 (0.5162) time: 0.4182 data: 0.0002 max mem: 52150 +Epoch: [28] [6320/7269] eta: 0:06:50 lr: 0.000003 loss: 0.5005 (0.5161) time: 0.4181 data: 0.0002 max mem: 52150 +Epoch: [28] [6330/7269] eta: 0:06:45 lr: 0.000003 loss: 0.5225 (0.5162) time: 0.4316 data: 0.0002 max mem: 52150 +Epoch: [28] [6340/7269] eta: 0:06:41 lr: 0.000003 loss: 0.5225 (0.5161) time: 0.4324 data: 0.0002 max mem: 52150 +Epoch: [28] [6350/7269] eta: 0:06:37 lr: 0.000003 loss: 0.5044 (0.5161) time: 0.3730 data: 0.0002 max mem: 52150 +Epoch: [28] [6360/7269] eta: 0:06:32 lr: 0.000003 loss: 0.5195 (0.5161) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [28] [6370/7269] eta: 0:06:28 lr: 0.000003 loss: 0.5195 (0.5161) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [28] [6380/7269] eta: 0:06:23 lr: 0.000003 loss: 0.5494 (0.5162) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [28] [6390/7269] eta: 0:06:19 lr: 0.000003 loss: 0.5453 (0.5161) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [28] [6400/7269] eta: 0:06:15 lr: 0.000003 loss: 0.4897 (0.5161) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [28] [6410/7269] eta: 0:06:10 lr: 0.000003 loss: 0.4897 (0.5160) time: 0.4358 data: 0.0002 max mem: 52150 +Epoch: [28] [6420/7269] eta: 0:06:06 lr: 0.000003 loss: 0.5113 (0.5160) time: 0.4819 data: 0.0002 max mem: 52150 +Epoch: [28] [6430/7269] eta: 0:06:02 lr: 0.000003 loss: 0.5244 (0.5160) time: 0.4787 data: 0.0002 max mem: 52150 +Epoch: [28] [6440/7269] eta: 0:05:57 lr: 0.000003 loss: 0.5206 (0.5160) time: 0.4332 data: 0.0002 max mem: 52150 +Epoch: [28] [6450/7269] eta: 0:05:53 lr: 0.000003 loss: 0.5146 (0.5159) time: 0.4353 data: 0.0001 max mem: 52150 +Epoch: [28] [6460/7269] eta: 0:05:49 lr: 0.000003 loss: 0.5034 (0.5159) time: 0.4946 data: 0.0001 max mem: 52150 +Epoch: [28] [6470/7269] eta: 0:05:45 lr: 0.000003 loss: 0.4952 (0.5159) time: 0.4823 data: 0.0001 max mem: 52150 +Epoch: [28] [6480/7269] eta: 0:05:40 lr: 0.000003 loss: 0.4936 (0.5158) time: 0.4683 data: 0.0001 max mem: 52150 +Epoch: [28] [6490/7269] eta: 0:05:36 lr: 0.000003 loss: 0.5073 (0.5159) time: 0.4181 data: 0.0002 max mem: 52150 +Epoch: [28] [6500/7269] eta: 0:05:32 lr: 0.000003 loss: 0.5538 (0.5159) time: 0.4956 data: 0.0002 max mem: 52150 +Epoch: [28] [6510/7269] eta: 0:05:28 lr: 0.000003 loss: 0.5538 (0.5159) time: 0.4957 data: 0.0002 max mem: 52150 +Epoch: [28] [6520/7269] eta: 0:05:23 lr: 0.000003 loss: 0.5161 (0.5160) time: 0.4368 data: 0.0002 max mem: 52150 +Epoch: [28] [6530/7269] eta: 0:05:19 lr: 0.000003 loss: 0.5126 (0.5159) time: 0.4985 data: 0.0001 max mem: 52150 +Epoch: [28] [6540/7269] eta: 0:05:15 lr: 0.000003 loss: 0.5147 (0.5159) time: 0.4338 data: 0.0001 max mem: 52150 +Epoch: [28] [6550/7269] eta: 0:05:10 lr: 0.000003 loss: 0.5211 (0.5159) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [28] [6560/7269] eta: 0:05:06 lr: 0.000003 loss: 0.5205 (0.5159) time: 0.4346 data: 0.0001 max mem: 52150 +Epoch: [28] [6570/7269] eta: 0:05:02 lr: 0.000003 loss: 0.5362 (0.5160) time: 0.4350 data: 0.0002 max mem: 52150 +Epoch: [28] [6580/7269] eta: 0:04:57 lr: 0.000003 loss: 0.5251 (0.5159) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [28] [6590/7269] eta: 0:04:53 lr: 0.000003 loss: 0.4815 (0.5159) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [28] [6600/7269] eta: 0:04:49 lr: 0.000003 loss: 0.5329 (0.5159) time: 0.4774 data: 0.0002 max mem: 52150 +Epoch: [28] [6610/7269] eta: 0:04:44 lr: 0.000003 loss: 0.5345 (0.5159) time: 0.5250 data: 0.0002 max mem: 52150 +Epoch: [28] [6620/7269] eta: 0:04:40 lr: 0.000003 loss: 0.5259 (0.5159) time: 0.4818 data: 0.0001 max mem: 52150 +Epoch: [28] [6630/7269] eta: 0:04:36 lr: 0.000003 loss: 0.4261 (0.5158) time: 0.4343 data: 0.0001 max mem: 52150 +Epoch: [28] [6640/7269] eta: 0:04:31 lr: 0.000003 loss: 0.4483 (0.5158) time: 0.3722 data: 0.0001 max mem: 52150 +Epoch: [28] [6650/7269] eta: 0:04:27 lr: 0.000003 loss: 0.5235 (0.5158) time: 0.3722 data: 0.0001 max mem: 52150 +Epoch: [28] [6660/7269] eta: 0:04:23 lr: 0.000003 loss: 0.5243 (0.5158) time: 0.4771 data: 0.0002 max mem: 52150 +Epoch: [28] [6670/7269] eta: 0:04:19 lr: 0.000003 loss: 0.5462 (0.5158) time: 0.5228 data: 0.0001 max mem: 52150 +Epoch: [28] [6680/7269] eta: 0:04:14 lr: 0.000003 loss: 0.5462 (0.5159) time: 0.4806 data: 0.0001 max mem: 52150 +Epoch: [28] [6690/7269] eta: 0:04:10 lr: 0.000003 loss: 0.5175 (0.5158) time: 0.4353 data: 0.0002 max mem: 52150 +Epoch: [28] [6700/7269] eta: 0:04:06 lr: 0.000003 loss: 0.4322 (0.5157) time: 0.4257 data: 0.0002 max mem: 52150 +Epoch: [28] [6710/7269] eta: 0:04:01 lr: 0.000003 loss: 0.4674 (0.5157) time: 0.4251 data: 0.0002 max mem: 52150 +Epoch: [28] [6720/7269] eta: 0:03:57 lr: 0.000003 loss: 0.5164 (0.5157) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [28] [6730/7269] eta: 0:03:53 lr: 0.000003 loss: 0.5487 (0.5157) time: 0.4328 data: 0.0001 max mem: 52150 +Epoch: [28] [6740/7269] eta: 0:03:48 lr: 0.000003 loss: 0.5240 (0.5157) time: 0.4806 data: 0.0001 max mem: 52150 +Epoch: [28] [6750/7269] eta: 0:03:44 lr: 0.000003 loss: 0.5302 (0.5157) time: 0.4210 data: 0.0002 max mem: 52150 +Epoch: [28] [6760/7269] eta: 0:03:40 lr: 0.000003 loss: 0.5320 (0.5157) time: 0.3728 data: 0.0002 max mem: 52150 +Epoch: [28] [6770/7269] eta: 0:03:35 lr: 0.000003 loss: 0.5229 (0.5157) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [28] [6780/7269] eta: 0:03:31 lr: 0.000003 loss: 0.5223 (0.5157) time: 0.4721 data: 0.0002 max mem: 52150 +Epoch: [28] [6790/7269] eta: 0:03:27 lr: 0.000003 loss: 0.5492 (0.5158) time: 0.5176 data: 0.0002 max mem: 52150 +Epoch: [28] [6800/7269] eta: 0:03:22 lr: 0.000003 loss: 0.5594 (0.5158) time: 0.4634 data: 0.0002 max mem: 52150 +Epoch: [28] [6810/7269] eta: 0:03:18 lr: 0.000003 loss: 0.5318 (0.5158) time: 0.4179 data: 0.0001 max mem: 52150 +Epoch: [28] [6820/7269] eta: 0:03:14 lr: 0.000003 loss: 0.5402 (0.5159) time: 0.3729 data: 0.0002 max mem: 52150 +Epoch: [28] [6830/7269] eta: 0:03:09 lr: 0.000003 loss: 0.5618 (0.5160) time: 0.4389 data: 0.0001 max mem: 52150 +Epoch: [28] [6840/7269] eta: 0:03:05 lr: 0.000003 loss: 0.5228 (0.5159) time: 0.4384 data: 0.0002 max mem: 52150 +Epoch: [28] [6850/7269] eta: 0:03:01 lr: 0.000003 loss: 0.5046 (0.5159) time: 0.4642 data: 0.0001 max mem: 52150 +Epoch: [28] [6860/7269] eta: 0:02:56 lr: 0.000003 loss: 0.5366 (0.5159) time: 0.4636 data: 0.0002 max mem: 52150 +Epoch: [28] [6870/7269] eta: 0:02:52 lr: 0.000003 loss: 0.5028 (0.5159) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [28] [6880/7269] eta: 0:02:48 lr: 0.000003 loss: 0.5028 (0.5159) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [28] [6890/7269] eta: 0:02:43 lr: 0.000003 loss: 0.5137 (0.5159) time: 0.3729 data: 0.0002 max mem: 52150 +Epoch: [28] [6900/7269] eta: 0:02:39 lr: 0.000003 loss: 0.4827 (0.5158) time: 0.3725 data: 0.0001 max mem: 52150 +Epoch: [28] [6910/7269] eta: 0:02:35 lr: 0.000003 loss: 0.5435 (0.5159) time: 0.4177 data: 0.0001 max mem: 52150 +Epoch: [28] [6920/7269] eta: 0:02:30 lr: 0.000003 loss: 0.5435 (0.5159) time: 0.4770 data: 0.0001 max mem: 52150 +Epoch: [28] [6930/7269] eta: 0:02:26 lr: 0.000003 loss: 0.5213 (0.5159) time: 0.4316 data: 0.0001 max mem: 52150 +Epoch: [28] [6940/7269] eta: 0:02:22 lr: 0.000003 loss: 0.5085 (0.5158) time: 0.3724 data: 0.0001 max mem: 52150 +Epoch: [28] [6950/7269] eta: 0:02:17 lr: 0.000003 loss: 0.4788 (0.5158) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [28] [6960/7269] eta: 0:02:13 lr: 0.000003 loss: 0.5150 (0.5158) time: 0.4778 data: 0.0002 max mem: 52150 +Epoch: [28] [6970/7269] eta: 0:02:09 lr: 0.000003 loss: 0.5150 (0.5158) time: 0.4778 data: 0.0002 max mem: 52150 +Epoch: [28] [6980/7269] eta: 0:02:04 lr: 0.000003 loss: 0.5228 (0.5158) time: 0.4950 data: 0.0001 max mem: 52150 +Epoch: [28] [6990/7269] eta: 0:02:00 lr: 0.000003 loss: 0.5391 (0.5158) time: 0.4948 data: 0.0001 max mem: 52150 +Epoch: [28] [7000/7269] eta: 0:01:56 lr: 0.000003 loss: 0.5576 (0.5159) time: 0.4353 data: 0.0001 max mem: 52150 +Epoch: [28] [7010/7269] eta: 0:01:51 lr: 0.000003 loss: 0.5580 (0.5159) time: 0.4349 data: 0.0001 max mem: 52150 +Epoch: [28] [7020/7269] eta: 0:01:47 lr: 0.000003 loss: 0.5666 (0.5160) time: 0.4363 data: 0.0001 max mem: 52150 +Epoch: [28] [7030/7269] eta: 0:01:43 lr: 0.000003 loss: 0.5240 (0.5159) time: 0.4371 data: 0.0001 max mem: 52150 +Epoch: [28] [7040/7269] eta: 0:01:38 lr: 0.000003 loss: 0.5039 (0.5159) time: 0.3728 data: 0.0001 max mem: 52150 +Epoch: [28] [7050/7269] eta: 0:01:34 lr: 0.000003 loss: 0.5433 (0.5159) time: 0.4369 data: 0.0002 max mem: 52150 +Epoch: [28] [7060/7269] eta: 0:01:30 lr: 0.000003 loss: 0.5054 (0.5159) time: 0.4816 data: 0.0001 max mem: 52150 +Epoch: [28] [7070/7269] eta: 0:01:26 lr: 0.000003 loss: 0.5086 (0.5159) time: 0.4798 data: 0.0002 max mem: 52150 +Epoch: [28] [7080/7269] eta: 0:01:21 lr: 0.000003 loss: 0.4938 (0.5159) time: 0.4348 data: 0.0002 max mem: 52150 +Epoch: [28] [7090/7269] eta: 0:01:17 lr: 0.000003 loss: 0.4733 (0.5158) time: 0.4204 data: 0.0002 max mem: 52150 +Epoch: [28] [7100/7269] eta: 0:01:13 lr: 0.000003 loss: 0.5076 (0.5158) time: 0.4204 data: 0.0002 max mem: 52150 +Epoch: [28] [7110/7269] eta: 0:01:08 lr: 0.000003 loss: 0.5214 (0.5158) time: 0.4197 data: 0.0002 max mem: 52150 +Epoch: [28] [7120/7269] eta: 0:01:04 lr: 0.000003 loss: 0.4927 (0.5158) time: 0.4200 data: 0.0002 max mem: 52150 +Epoch: [28] [7130/7269] eta: 0:01:00 lr: 0.000003 loss: 0.5472 (0.5158) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [28] [7140/7269] eta: 0:00:55 lr: 0.000003 loss: 0.5171 (0.5158) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [28] [7150/7269] eta: 0:00:51 lr: 0.000003 loss: 0.5081 (0.5158) time: 0.3721 data: 0.0001 max mem: 52150 +Epoch: [28] [7160/7269] eta: 0:00:47 lr: 0.000003 loss: 0.4976 (0.5158) time: 0.5944 data: 0.0002 max mem: 52150 +Epoch: [28] [7170/7269] eta: 0:00:42 lr: 0.000003 loss: 0.5121 (0.5158) time: 0.5944 data: 0.0002 max mem: 52150 +Epoch: [28] [7180/7269] eta: 0:00:38 lr: 0.000003 loss: 0.5070 (0.5158) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [28] [7190/7269] eta: 0:00:34 lr: 0.000003 loss: 0.5263 (0.5158) time: 0.4197 data: 0.0002 max mem: 52150 +Epoch: [28] [7200/7269] eta: 0:00:29 lr: 0.000003 loss: 0.5455 (0.5158) time: 0.4668 data: 0.0002 max mem: 52150 +Epoch: [28] [7210/7269] eta: 0:00:25 lr: 0.000003 loss: 0.5187 (0.5158) time: 0.4199 data: 0.0002 max mem: 52150 +Epoch: [28] [7220/7269] eta: 0:00:21 lr: 0.000003 loss: 0.5052 (0.5158) time: 0.3728 data: 0.0001 max mem: 52150 +Epoch: [28] [7230/7269] eta: 0:00:16 lr: 0.000003 loss: 0.5330 (0.5158) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [28] [7240/7269] eta: 0:00:12 lr: 0.000003 loss: 0.5557 (0.5158) time: 0.3724 data: 0.0001 max mem: 52150 +Epoch: [28] [7250/7269] eta: 0:00:08 lr: 0.000003 loss: 0.5313 (0.5158) time: 0.4272 data: 0.0034 max mem: 52150 +Epoch: [28] [7260/7269] eta: 0:00:03 lr: 0.000003 loss: 0.5361 (0.5158) time: 0.5333 data: 0.0034 max mem: 52150 +Epoch: [28] [7268/7269] eta: 0:00:00 lr: 0.000003 loss: 0.5116 (0.5158) time: 0.4812 data: 0.0034 max mem: 52150 +Epoch: [28] Total time: 0:52:25 (0.4328 s / it) +Averaged stats: lr: 0.000003 loss: 0.5116 (0.5152) +Test: [ 0/1403] eta: 1:16:42 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0930 (0.0930) acc1: 100.0000 (100.0000) acc5: 100.0000 (100.0000) time: 3.2808 data: 3.1825 max mem: 52150 +Test: [ 10/1403] eta: 0:08:57 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1843 (0.1912) acc1: 95.3125 (94.0341) acc5: 100.0000 (100.0000) time: 0.3861 data: 0.2897 max mem: 52150 +Test: [ 20/1403] eta: 0:05:43 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1421 (0.1728) acc1: 95.3125 (94.8661) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0004 max mem: 52150 +Test: [ 30/1403] eta: 0:04:33 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0905 (0.1600) acc1: 98.4375 (95.4133) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 40/1403] eta: 0:03:57 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0981 (0.1557) acc1: 98.4375 (96.0366) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 50/1403] eta: 0:03:35 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1074 (0.1678) acc1: 98.4375 (95.4350) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 60/1403] eta: 0:03:19 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1195 (0.1647) acc1: 100.0000 (95.6967) acc5: 100.0000 (100.0000) time: 0.0963 data: 0.0003 max mem: 52150 +Test: [ 70/1403] eta: 0:03:08 acc1_pos: 90.625 acc5_pos: 100.000 loss: 0.1283 (0.1706) acc1: 98.4375 (95.3565) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 80/1403] eta: 0:02:59 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.1806 (0.1756) acc1: 93.7500 (95.0810) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 90/1403] eta: 0:02:52 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1814 (0.1913) acc1: 93.7500 (94.1964) acc5: 100.0000 (100.0000) time: 0.0963 data: 0.0003 max mem: 52150 +Test: [ 100/1403] eta: 0:02:46 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1604 (0.1883) acc1: 95.3125 (94.3843) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 110/1403] eta: 0:02:41 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1256 (0.1855) acc1: 98.4375 (94.5805) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 120/1403] eta: 0:02:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0764 (0.1763) acc1: 100.0000 (95.0284) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 130/1403] eta: 0:02:33 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0756 (0.1698) acc1: 100.0000 (95.3244) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 140/1403] eta: 0:02:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0722 (0.1628) acc1: 100.0000 (95.6560) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 150/1403] eta: 0:02:27 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0715 (0.1570) acc1: 100.0000 (95.9230) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 160/1403] eta: 0:02:24 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0731 (0.1521) acc1: 100.0000 (96.1568) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 170/1403] eta: 0:02:21 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0770 (0.1482) acc1: 100.0000 (96.3633) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 180/1403] eta: 0:02:19 acc1_pos: 90.625 acc5_pos: 100.000 loss: 0.0837 (0.1511) acc1: 100.0000 (96.2621) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52150 +Test: [ 190/1403] eta: 0:02:17 acc1_pos: 92.188 acc5_pos: 100.000 loss: 0.2338 (0.1581) acc1: 93.7500 (95.8770) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 200/1403] eta: 0:02:15 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.2338 (0.1680) acc1: 92.1875 (95.3980) acc5: 100.0000 (100.0000) time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 210/1403] eta: 0:02:13 acc1_pos: 75.000 acc5_pos: 100.000 loss: 0.2940 (0.1777) acc1: 87.5000 (94.8534) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 220/1403] eta: 0:02:11 acc1_pos: 89.062 acc5_pos: 100.000 loss: 0.3170 (0.1856) acc1: 87.5000 (94.4853) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 230/1403] eta: 0:02:09 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2516 (0.1818) acc1: 92.1875 (94.6699) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 240/1403] eta: 0:02:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0768 (0.1776) acc1: 100.0000 (94.8781) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 250/1403] eta: 0:02:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0790 (0.1739) acc1: 100.0000 (95.0635) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 260/1403] eta: 0:02:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0812 (0.1705) acc1: 100.0000 (95.2347) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 270/1403] eta: 0:02:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0812 (0.1674) acc1: 100.0000 (95.3875) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 280/1403] eta: 0:02:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0802 (0.1642) acc1: 100.0000 (95.5460) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 290/1403] eta: 0:01:59 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0745 (0.1613) acc1: 100.0000 (95.6937) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 300/1403] eta: 0:01:58 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0737 (0.1585) acc1: 100.0000 (95.8316) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 52150 +Test: [ 310/1403] eta: 0:01:56 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0782 (0.1561) acc1: 100.0000 (95.9606) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 320/1403] eta: 0:01:55 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0928 (0.1542) acc1: 100.0000 (96.0864) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 330/1403] eta: 0:01:53 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0995 (0.1527) acc1: 100.0000 (96.2000) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 340/1403] eta: 0:01:52 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0993 (0.1510) acc1: 100.0000 (96.3114) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 350/1403] eta: 0:01:51 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0976 (0.1495) acc1: 100.0000 (96.4165) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 360/1403] eta: 0:01:49 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1009 (0.1483) acc1: 100.0000 (96.5071) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 370/1403] eta: 0:01:48 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0990 (0.1469) acc1: 100.0000 (96.6012) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 380/1403] eta: 0:01:47 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0980 (0.1457) acc1: 100.0000 (96.6905) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 390/1403] eta: 0:01:46 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0987 (0.1466) acc1: 100.0000 (96.6352) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 400/1403] eta: 0:01:44 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2342 (0.1508) acc1: 89.0625 (96.4113) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 410/1403] eta: 0:01:43 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2996 (0.1544) acc1: 87.5000 (96.1983) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 420/1403] eta: 0:01:42 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2553 (0.1564) acc1: 89.0625 (96.0733) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 430/1403] eta: 0:01:41 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2507 (0.1595) acc1: 89.0625 (95.9215) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 440/1403] eta: 0:01:39 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3000 (0.1633) acc1: 89.0625 (95.7448) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 450/1403] eta: 0:01:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2742 (0.1659) acc1: 89.0625 (95.6035) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 460/1403] eta: 0:01:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2600 (0.1687) acc1: 89.0625 (95.4515) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 470/1403] eta: 0:01:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2281 (0.1680) acc1: 90.6250 (95.4651) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 480/1403] eta: 0:01:35 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0619 (0.1657) acc1: 100.0000 (95.5594) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 490/1403] eta: 0:01:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0619 (0.1636) acc1: 100.0000 (95.6498) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 500/1403] eta: 0:01:32 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0620 (0.1616) acc1: 100.0000 (95.7367) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 510/1403] eta: 0:01:31 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0625 (0.1597) acc1: 100.0000 (95.8201) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 520/1403] eta: 0:01:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0671 (0.1580) acc1: 100.0000 (95.9003) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 530/1403] eta: 0:01:29 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0590 (0.1561) acc1: 100.0000 (95.9775) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 540/1403] eta: 0:01:28 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0585 (0.1543) acc1: 100.0000 (96.0519) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 550/1403] eta: 0:01:27 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0595 (0.1532) acc1: 100.0000 (96.0980) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 560/1403] eta: 0:01:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0911 (0.1523) acc1: 98.4375 (96.1286) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 570/1403] eta: 0:01:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0927 (0.1513) acc1: 98.4375 (96.1772) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 580/1403] eta: 0:01:23 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0951 (0.1508) acc1: 98.4375 (96.1946) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 590/1403] eta: 0:01:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1201 (0.1506) acc1: 96.8750 (96.1902) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 600/1403] eta: 0:01:21 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1335 (0.1510) acc1: 95.3125 (96.1392) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 610/1403] eta: 0:01:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1335 (0.1505) acc1: 95.3125 (96.1436) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 620/1403] eta: 0:01:19 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1414 (0.1515) acc1: 95.3125 (96.0447) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 630/1403] eta: 0:01:18 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1529 (0.1514) acc1: 93.7500 (96.0356) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 640/1403] eta: 0:01:17 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1295 (0.1511) acc1: 96.8750 (96.0511) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 650/1403] eta: 0:01:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1205 (0.1506) acc1: 96.8750 (96.0733) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 660/1403] eta: 0:01:15 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0935 (0.1495) acc1: 98.4375 (96.1233) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 670/1403] eta: 0:01:14 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0664 (0.1483) acc1: 100.0000 (96.1787) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 680/1403] eta: 0:01:13 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0776 (0.1497) acc1: 98.4375 (96.0857) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 690/1403] eta: 0:01:12 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2574 (0.1540) acc1: 89.0625 (95.8416) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 700/1403] eta: 0:01:11 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1779 (0.1543) acc1: 92.1875 (95.8229) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [ 710/1403] eta: 0:01:10 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3368 (0.1591) acc1: 84.3750 (95.5740) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [ 720/1403] eta: 0:01:08 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3368 (0.1592) acc1: 82.8125 (95.5639) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [ 730/1403] eta: 0:01:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0678 (0.1582) acc1: 100.0000 (95.6117) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 740/1403] eta: 0:01:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0760 (0.1573) acc1: 100.0000 (95.6541) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 750/1403] eta: 0:01:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1087 (0.1588) acc1: 98.4375 (95.5767) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 760/1403] eta: 0:01:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3377 (0.1625) acc1: 87.5000 (95.3330) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 770/1403] eta: 0:01:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4757 (0.1675) acc1: 73.4375 (95.0268) acc5: 100.0000 (100.0000) acc1_neg: 68.750 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 780/1403] eta: 0:01:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4796 (0.1706) acc1: 75.0000 (94.8524) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 790/1403] eta: 0:01:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3314 (0.1733) acc1: 79.6875 (94.7140) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 800/1403] eta: 0:01:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2418 (0.1737) acc1: 89.0625 (94.6941) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 810/1403] eta: 0:00:59 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2458 (0.1771) acc1: 87.5000 (94.5399) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 820/1403] eta: 0:00:58 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6621 (0.1865) acc1: 71.8750 (94.1535) acc5: 100.0000 (100.0000) acc1_neg: 39.062 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 830/1403] eta: 0:00:57 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.7236 (0.1894) acc1: 68.7500 (94.0132) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 840/1403] eta: 0:00:56 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5198 (0.1981) acc1: 68.7500 (93.6441) acc5: 100.0000 (100.0000) acc1_neg: 48.438 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 850/1403] eta: 0:00:55 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.1301 (0.2098) acc1: 54.6875 (93.1569) acc5: 100.0000 (100.0000) acc1_neg: 50.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 860/1403] eta: 0:00:54 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.0679 (0.2136) acc1: 54.6875 (92.9932) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 870/1403] eta: 0:00:53 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1399 (0.2126) acc1: 95.3125 (93.0342) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 880/1403] eta: 0:00:52 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1014 (0.2114) acc1: 98.4375 (93.0885) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 890/1403] eta: 0:00:51 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0914 (0.2099) acc1: 98.4375 (93.1573) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 900/1403] eta: 0:00:50 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0721 (0.2084) acc1: 100.0000 (93.2297) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 910/1403] eta: 0:00:49 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0784 (0.2074) acc1: 98.4375 (93.2663) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 920/1403] eta: 0:00:48 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0973 (0.2064) acc1: 98.4375 (93.3072) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [ 930/1403] eta: 0:00:47 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0929 (0.2052) acc1: 98.4375 (93.3673) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0970 data: 0.0003 max mem: 52150 +Test: [ 940/1403] eta: 0:00:46 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1096 (0.2052) acc1: 98.4375 (93.3482) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 950/1403] eta: 0:00:45 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2379 (0.2060) acc1: 89.0625 (93.2916) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 960/1403] eta: 0:00:44 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2637 (0.2065) acc1: 87.5000 (93.2557) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 970/1403] eta: 0:00:43 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2330 (0.2073) acc1: 90.6250 (93.2093) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [ 980/1403] eta: 0:00:42 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2703 (0.2081) acc1: 89.0625 (93.1607) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [ 990/1403] eta: 0:00:41 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2463 (0.2084) acc1: 89.0625 (93.1477) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [1000/1403] eta: 0:00:40 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2365 (0.2090) acc1: 90.6250 (93.1131) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1010/1403] eta: 0:00:39 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2589 (0.2097) acc1: 90.6250 (93.0762) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [1020/1403] eta: 0:00:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3973 (0.2158) acc1: 82.8125 (92.7507) acc5: 100.0000 (100.0000) acc1_neg: 35.938 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [1030/1403] eta: 0:00:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5687 (0.2174) acc1: 65.6250 (92.6619) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1040/1403] eta: 0:00:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2128 (0.2172) acc1: 92.1875 (92.6678) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1050/1403] eta: 0:00:35 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1826 (0.2169) acc1: 95.3125 (92.6870) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1060/1403] eta: 0:00:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1472 (0.2161) acc1: 96.8750 (92.7265) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [1070/1403] eta: 0:00:33 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1472 (0.2166) acc1: 96.8750 (92.7098) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [1080/1403] eta: 0:00:32 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3918 (0.2186) acc1: 84.3750 (92.6081) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52150 +Test: [1090/1403] eta: 0:00:31 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4335 (0.2206) acc1: 79.6875 (92.4983) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [1100/1403] eta: 0:00:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3626 (0.2211) acc1: 84.3750 (92.4699) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1110/1403] eta: 0:00:29 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2667 (0.2215) acc1: 90.6250 (92.4547) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1120/1403] eta: 0:00:28 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2675 (0.2220) acc1: 90.6250 (92.4231) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0967 data: 0.0003 max mem: 52150 +Test: [1130/1403] eta: 0:00:27 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2795 (0.2227) acc1: 89.0625 (92.3878) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52150 +Test: [1140/1403] eta: 0:00:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2396 (0.2228) acc1: 90.6250 (92.3778) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1150/1403] eta: 0:00:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2396 (0.2232) acc1: 90.6250 (92.3558) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52150 +Test: [1160/1403] eta: 0:00:24 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2715 (0.2237) acc1: 90.6250 (92.3355) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0967 data: 0.0002 max mem: 52150 +Test: [1170/1403] eta: 0:00:23 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2694 (0.2236) acc1: 89.0625 (92.3423) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1180/1403] eta: 0:00:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0677 (0.2222) acc1: 100.0000 (92.4071) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1190/1403] eta: 0:00:21 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0671 (0.2209) acc1: 100.0000 (92.4709) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1200/1403] eta: 0:00:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0671 (0.2197) acc1: 100.0000 (92.5336) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52150 +Test: [1210/1403] eta: 0:00:19 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0667 (0.2184) acc1: 100.0000 (92.5952) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52150 +Test: [1220/1403] eta: 0:00:18 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0664 (0.2172) acc1: 100.0000 (92.6559) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1230/1403] eta: 0:00:17 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0669 (0.2160) acc1: 100.0000 (92.7155) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1240/1403] eta: 0:00:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0669 (0.2147) acc1: 100.0000 (92.7742) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52150 +Test: [1250/1403] eta: 0:00:15 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0672 (0.2136) acc1: 100.0000 (92.8320) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52150 +Test: [1260/1403] eta: 0:00:14 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0712 (0.2125) acc1: 100.0000 (92.8888) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52150 +Test: [1270/1403] eta: 0:00:13 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0721 (0.2114) acc1: 100.0000 (92.9448) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52150 +Test: [1280/1403] eta: 0:00:12 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0717 (0.2103) acc1: 100.0000 (92.9999) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52150 +Test: [1290/1403] eta: 0:00:11 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0717 (0.2092) acc1: 100.0000 (93.0541) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52150 +Test: [1300/1403] eta: 0:00:10 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0721 (0.2082) acc1: 100.0000 (93.1075) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1310/1403] eta: 0:00:09 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0726 (0.2071) acc1: 100.0000 (93.1600) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0002 max mem: 52150 +Test: [1320/1403] eta: 0:00:08 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0712 (0.2061) acc1: 100.0000 (93.2118) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52150 +Test: [1330/1403] eta: 0:00:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0712 (0.2051) acc1: 100.0000 (93.2616) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52150 +Test: [1340/1403] eta: 0:00:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0662 (0.2040) acc1: 100.0000 (93.3107) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52150 +Test: [1350/1403] eta: 0:00:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0650 (0.2030) acc1: 100.0000 (93.3579) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52150 +Test: [1360/1403] eta: 0:00:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0649 (0.2020) acc1: 100.0000 (93.4067) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52150 +Test: [1370/1403] eta: 0:00:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0649 (0.2011) acc1: 100.0000 (93.4525) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52150 +Test: [1380/1403] eta: 0:00:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0645 (0.2001) acc1: 100.0000 (93.4988) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52150 +Test: [1390/1403] eta: 0:00:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0644 (0.1991) acc1: 100.0000 (93.5456) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0002 max mem: 52150 +Test: [1400/1403] eta: 0:00:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0648 (0.1982) acc1: 100.0000 (93.5894) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0001 max mem: 52150 +Test: [1402/1403] eta: 0:00:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0651 (0.1980) acc1: 100.0000 (93.5940) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0922 data: 0.0001 max mem: 52150 +Test: Total time: 0:02:19 (0.0992 s / it) +* Overall Acc@1 93.594 Acc@5 100.000 + target == real Acc@1 95.925 Acc@5 100.000 + target == fake Acc@1 92.937 Acc@5 100.000 + loss 0.198 +Accuracy of the network on the 89729 test images: 93.6% +Max accuracy: 94.33% +Epoch: [29] [ 0/7269] eta: 12:12:37 lr: 0.000003 loss: 0.4357 (0.4357) time: 6.0472 data: 5.2237 max mem: 52150 +Epoch: [29] [ 10/7269] eta: 1:47:55 lr: 0.000003 loss: 0.5269 (0.5188) time: 0.8921 data: 0.4750 max mem: 52150 +Epoch: [29] [ 20/7269] eta: 1:25:00 lr: 0.000003 loss: 0.5497 (0.5338) time: 0.4365 data: 0.0002 max mem: 52150 +Epoch: [29] [ 30/7269] eta: 1:15:54 lr: 0.000003 loss: 0.5349 (0.5295) time: 0.4846 data: 0.0002 max mem: 52150 +Epoch: [29] [ 40/7269] eta: 1:08:13 lr: 0.000003 loss: 0.4976 (0.5185) time: 0.4220 data: 0.0002 max mem: 52150 +Epoch: [29] [ 50/7269] eta: 1:06:28 lr: 0.000003 loss: 0.4773 (0.5177) time: 0.4336 data: 0.0002 max mem: 52150 +Epoch: [29] [ 60/7269] eta: 1:02:49 lr: 0.000003 loss: 0.5423 (0.5200) time: 0.4341 data: 0.0002 max mem: 52150 +Epoch: [29] [ 70/7269] eta: 1:00:10 lr: 0.000003 loss: 0.5121 (0.5160) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [29] [ 80/7269] eta: 0:59:41 lr: 0.000003 loss: 0.5199 (0.5179) time: 0.4229 data: 0.0002 max mem: 52150 +Epoch: [29] [ 90/7269] eta: 0:57:56 lr: 0.000003 loss: 0.5152 (0.5139) time: 0.4232 data: 0.0002 max mem: 52150 +Epoch: [29] [ 100/7269] eta: 0:57:59 lr: 0.000003 loss: 0.4876 (0.5119) time: 0.4332 data: 0.0002 max mem: 52150 +Epoch: [29] [ 110/7269] eta: 0:57:42 lr: 0.000003 loss: 0.5226 (0.5143) time: 0.4803 data: 0.0002 max mem: 52150 +Epoch: [29] [ 120/7269] eta: 0:57:43 lr: 0.000003 loss: 0.5348 (0.5175) time: 0.4803 data: 0.0002 max mem: 52150 +Epoch: [29] [ 130/7269] eta: 0:56:37 lr: 0.000003 loss: 0.5339 (0.5167) time: 0.4328 data: 0.0002 max mem: 52150 +Epoch: [29] [ 140/7269] eta: 0:55:39 lr: 0.000003 loss: 0.5058 (0.5149) time: 0.3709 data: 0.0002 max mem: 52150 +Epoch: [29] [ 150/7269] eta: 0:55:36 lr: 0.000003 loss: 0.5287 (0.5171) time: 0.4219 data: 0.0001 max mem: 52150 +Epoch: [29] [ 160/7269] eta: 0:55:31 lr: 0.000003 loss: 0.5349 (0.5163) time: 0.4701 data: 0.0001 max mem: 52150 +Epoch: [29] [ 170/7269] eta: 0:55:27 lr: 0.000003 loss: 0.5327 (0.5174) time: 0.4683 data: 0.0001 max mem: 52150 +Epoch: [29] [ 180/7269] eta: 0:54:44 lr: 0.000003 loss: 0.5501 (0.5183) time: 0.4202 data: 0.0001 max mem: 52150 +Epoch: [29] [ 190/7269] eta: 0:54:05 lr: 0.000003 loss: 0.5501 (0.5182) time: 0.3713 data: 0.0001 max mem: 52150 +Epoch: [29] [ 200/7269] eta: 0:54:48 lr: 0.000003 loss: 0.5553 (0.5205) time: 0.4832 data: 0.0001 max mem: 52150 +Epoch: [29] [ 210/7269] eta: 0:54:13 lr: 0.000003 loss: 0.5479 (0.5205) time: 0.4835 data: 0.0001 max mem: 52150 +Epoch: [29] [ 220/7269] eta: 0:53:40 lr: 0.000003 loss: 0.5158 (0.5206) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [29] [ 230/7269] eta: 0:53:48 lr: 0.000003 loss: 0.5232 (0.5205) time: 0.4354 data: 0.0001 max mem: 52150 +Epoch: [29] [ 240/7269] eta: 0:53:18 lr: 0.000003 loss: 0.5301 (0.5208) time: 0.4346 data: 0.0001 max mem: 52150 +Epoch: [29] [ 250/7269] eta: 0:52:50 lr: 0.000003 loss: 0.5137 (0.5210) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [29] [ 260/7269] eta: 0:52:24 lr: 0.000003 loss: 0.5137 (0.5192) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [29] [ 270/7269] eta: 0:52:26 lr: 0.000003 loss: 0.4889 (0.5181) time: 0.4235 data: 0.0002 max mem: 52150 +Epoch: [29] [ 280/7269] eta: 0:52:02 lr: 0.000003 loss: 0.5173 (0.5185) time: 0.4231 data: 0.0001 max mem: 52150 +Epoch: [29] [ 290/7269] eta: 0:51:40 lr: 0.000003 loss: 0.5173 (0.5185) time: 0.3719 data: 0.0001 max mem: 52150 +Epoch: [29] [ 300/7269] eta: 0:52:35 lr: 0.000003 loss: 0.5030 (0.5185) time: 0.5367 data: 0.0001 max mem: 52150 +Epoch: [29] [ 310/7269] eta: 0:52:35 lr: 0.000003 loss: 0.4884 (0.5181) time: 0.5873 data: 0.0001 max mem: 52150 +Epoch: [29] [ 320/7269] eta: 0:52:13 lr: 0.000003 loss: 0.5039 (0.5179) time: 0.4227 data: 0.0002 max mem: 52150 +Epoch: [29] [ 330/7269] eta: 0:51:52 lr: 0.000003 loss: 0.4987 (0.5168) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [29] [ 340/7269] eta: 0:52:18 lr: 0.000003 loss: 0.5284 (0.5174) time: 0.4854 data: 0.0001 max mem: 52150 +Epoch: [29] [ 350/7269] eta: 0:52:22 lr: 0.000003 loss: 0.4983 (0.5159) time: 0.5488 data: 0.0001 max mem: 52150 +Epoch: [29] [ 360/7269] eta: 0:52:02 lr: 0.000003 loss: 0.4943 (0.5155) time: 0.4351 data: 0.0001 max mem: 52150 +Epoch: [29] [ 370/7269] eta: 0:51:43 lr: 0.000003 loss: 0.5147 (0.5161) time: 0.3724 data: 0.0001 max mem: 52150 +Epoch: [29] [ 380/7269] eta: 0:51:24 lr: 0.000003 loss: 0.5470 (0.5172) time: 0.3716 data: 0.0001 max mem: 52150 +Epoch: [29] [ 390/7269] eta: 0:51:24 lr: 0.000003 loss: 0.5418 (0.5171) time: 0.4216 data: 0.0002 max mem: 52150 +Epoch: [29] [ 400/7269] eta: 0:51:06 lr: 0.000003 loss: 0.5277 (0.5174) time: 0.4223 data: 0.0002 max mem: 52150 +Epoch: [29] [ 410/7269] eta: 0:51:10 lr: 0.000003 loss: 0.5451 (0.5180) time: 0.4339 data: 0.0002 max mem: 52150 +Epoch: [29] [ 420/7269] eta: 0:51:09 lr: 0.000003 loss: 0.5472 (0.5174) time: 0.4827 data: 0.0002 max mem: 52150 +Epoch: [29] [ 430/7269] eta: 0:50:52 lr: 0.000003 loss: 0.5467 (0.5179) time: 0.4199 data: 0.0002 max mem: 52150 +Epoch: [29] [ 440/7269] eta: 0:50:51 lr: 0.000003 loss: 0.5260 (0.5173) time: 0.4182 data: 0.0001 max mem: 52150 +Epoch: [29] [ 450/7269] eta: 0:50:49 lr: 0.000003 loss: 0.5165 (0.5174) time: 0.4660 data: 0.0001 max mem: 52150 +Epoch: [29] [ 460/7269] eta: 0:50:34 lr: 0.000003 loss: 0.5349 (0.5176) time: 0.4190 data: 0.0001 max mem: 52150 +Epoch: [29] [ 470/7269] eta: 0:50:19 lr: 0.000003 loss: 0.5284 (0.5176) time: 0.3715 data: 0.0001 max mem: 52150 +Epoch: [29] [ 480/7269] eta: 0:50:18 lr: 0.000003 loss: 0.5037 (0.5170) time: 0.4227 data: 0.0002 max mem: 52150 +Epoch: [29] [ 490/7269] eta: 0:50:04 lr: 0.000003 loss: 0.4927 (0.5168) time: 0.4227 data: 0.0002 max mem: 52150 +Epoch: [29] [ 500/7269] eta: 0:49:50 lr: 0.000003 loss: 0.5194 (0.5170) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [29] [ 510/7269] eta: 0:50:18 lr: 0.000003 loss: 0.5137 (0.5169) time: 0.5311 data: 0.0002 max mem: 52150 +Epoch: [29] [ 520/7269] eta: 0:50:17 lr: 0.000003 loss: 0.5137 (0.5170) time: 0.5812 data: 0.0002 max mem: 52150 +Epoch: [29] [ 530/7269] eta: 0:50:15 lr: 0.000003 loss: 0.5319 (0.5173) time: 0.4680 data: 0.0001 max mem: 52150 +Epoch: [29] [ 540/7269] eta: 0:50:01 lr: 0.000003 loss: 0.5319 (0.5174) time: 0.4179 data: 0.0001 max mem: 52150 +Epoch: [29] [ 550/7269] eta: 0:49:47 lr: 0.000003 loss: 0.5468 (0.5178) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [29] [ 560/7269] eta: 0:49:34 lr: 0.000003 loss: 0.5154 (0.5177) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [29] [ 570/7269] eta: 0:49:50 lr: 0.000003 loss: 0.4968 (0.5174) time: 0.4945 data: 0.0002 max mem: 52150 +Epoch: [29] [ 580/7269] eta: 0:50:02 lr: 0.000003 loss: 0.5350 (0.5181) time: 0.6017 data: 0.0002 max mem: 52150 +Epoch: [29] [ 590/7269] eta: 0:49:48 lr: 0.000003 loss: 0.5247 (0.5176) time: 0.4786 data: 0.0002 max mem: 52150 +Epoch: [29] [ 600/7269] eta: 0:49:36 lr: 0.000003 loss: 0.5060 (0.5178) time: 0.3712 data: 0.0001 max mem: 52150 +Epoch: [29] [ 610/7269] eta: 0:49:33 lr: 0.000003 loss: 0.5072 (0.5176) time: 0.4167 data: 0.0002 max mem: 52150 +Epoch: [29] [ 620/7269] eta: 0:49:20 lr: 0.000003 loss: 0.5266 (0.5181) time: 0.4168 data: 0.0002 max mem: 52150 +Epoch: [29] [ 630/7269] eta: 0:49:08 lr: 0.000003 loss: 0.5133 (0.5173) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [29] [ 640/7269] eta: 0:48:56 lr: 0.000003 loss: 0.4438 (0.5163) time: 0.3720 data: 0.0001 max mem: 52150 +Epoch: [29] [ 650/7269] eta: 0:48:45 lr: 0.000003 loss: 0.4858 (0.5165) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [29] [ 660/7269] eta: 0:48:45 lr: 0.000003 loss: 0.5383 (0.5165) time: 0.4318 data: 0.0001 max mem: 52150 +Epoch: [29] [ 670/7269] eta: 0:48:55 lr: 0.000003 loss: 0.4984 (0.5159) time: 0.5417 data: 0.0001 max mem: 52150 +Epoch: [29] [ 680/7269] eta: 0:48:44 lr: 0.000003 loss: 0.4888 (0.5157) time: 0.4821 data: 0.0002 max mem: 52150 +Epoch: [29] [ 690/7269] eta: 0:48:41 lr: 0.000003 loss: 0.4888 (0.5151) time: 0.4172 data: 0.0002 max mem: 52150 +Epoch: [29] [ 700/7269] eta: 0:48:38 lr: 0.000003 loss: 0.4829 (0.5154) time: 0.4623 data: 0.0002 max mem: 52150 +Epoch: [29] [ 710/7269] eta: 0:48:38 lr: 0.000003 loss: 0.5143 (0.5158) time: 0.4771 data: 0.0001 max mem: 52150 +Epoch: [29] [ 720/7269] eta: 0:48:27 lr: 0.000003 loss: 0.5290 (0.5162) time: 0.4319 data: 0.0001 max mem: 52150 +Epoch: [29] [ 730/7269] eta: 0:48:35 lr: 0.000003 loss: 0.5269 (0.5162) time: 0.4782 data: 0.0002 max mem: 52150 +Epoch: [29] [ 740/7269] eta: 0:48:25 lr: 0.000003 loss: 0.5201 (0.5164) time: 0.4785 data: 0.0002 max mem: 52150 +Epoch: [29] [ 750/7269] eta: 0:48:14 lr: 0.000003 loss: 0.5141 (0.5164) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [29] [ 760/7269] eta: 0:48:03 lr: 0.000003 loss: 0.5193 (0.5166) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [29] [ 770/7269] eta: 0:48:00 lr: 0.000003 loss: 0.5372 (0.5168) time: 0.4159 data: 0.0002 max mem: 52150 +Epoch: [29] [ 780/7269] eta: 0:47:57 lr: 0.000003 loss: 0.5231 (0.5167) time: 0.4613 data: 0.0001 max mem: 52150 +Epoch: [29] [ 790/7269] eta: 0:47:47 lr: 0.000003 loss: 0.5231 (0.5168) time: 0.4174 data: 0.0001 max mem: 52150 +Epoch: [29] [ 800/7269] eta: 0:47:37 lr: 0.000003 loss: 0.5271 (0.5171) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [29] [ 810/7269] eta: 0:47:34 lr: 0.000003 loss: 0.5164 (0.5171) time: 0.4179 data: 0.0001 max mem: 52150 +Epoch: [29] [ 820/7269] eta: 0:47:32 lr: 0.000003 loss: 0.5324 (0.5173) time: 0.4645 data: 0.0001 max mem: 52150 +Epoch: [29] [ 830/7269] eta: 0:47:41 lr: 0.000003 loss: 0.5283 (0.5172) time: 0.5411 data: 0.0002 max mem: 52150 +Epoch: [29] [ 840/7269] eta: 0:47:31 lr: 0.000003 loss: 0.5030 (0.5169) time: 0.4944 data: 0.0002 max mem: 52150 +Epoch: [29] [ 850/7269] eta: 0:47:21 lr: 0.000003 loss: 0.4942 (0.5169) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [29] [ 860/7269] eta: 0:47:11 lr: 0.000003 loss: 0.5309 (0.5173) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [29] [ 870/7269] eta: 0:47:02 lr: 0.000003 loss: 0.5343 (0.5170) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [29] [ 880/7269] eta: 0:46:52 lr: 0.000003 loss: 0.4981 (0.5171) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [29] [ 890/7269] eta: 0:46:52 lr: 0.000003 loss: 0.5430 (0.5173) time: 0.4341 data: 0.0001 max mem: 52150 +Epoch: [29] [ 900/7269] eta: 0:46:43 lr: 0.000003 loss: 0.5023 (0.5170) time: 0.4341 data: 0.0001 max mem: 52150 +Epoch: [29] [ 910/7269] eta: 0:46:33 lr: 0.000003 loss: 0.5047 (0.5172) time: 0.3717 data: 0.0001 max mem: 52150 +Epoch: [29] [ 920/7269] eta: 0:46:31 lr: 0.000003 loss: 0.5212 (0.5173) time: 0.4172 data: 0.0001 max mem: 52150 +Epoch: [29] [ 930/7269] eta: 0:46:30 lr: 0.000003 loss: 0.5406 (0.5176) time: 0.4764 data: 0.0001 max mem: 52150 +Epoch: [29] [ 940/7269] eta: 0:46:29 lr: 0.000003 loss: 0.5327 (0.5175) time: 0.4946 data: 0.0001 max mem: 52150 +Epoch: [29] [ 950/7269] eta: 0:46:20 lr: 0.000003 loss: 0.5211 (0.5176) time: 0.4354 data: 0.0001 max mem: 52150 +Epoch: [29] [ 960/7269] eta: 0:46:17 lr: 0.000003 loss: 0.5355 (0.5179) time: 0.4159 data: 0.0001 max mem: 52150 +Epoch: [29] [ 970/7269] eta: 0:46:08 lr: 0.000003 loss: 0.5539 (0.5182) time: 0.4160 data: 0.0001 max mem: 52150 +Epoch: [29] [ 980/7269] eta: 0:46:05 lr: 0.000003 loss: 0.5274 (0.5180) time: 0.4184 data: 0.0001 max mem: 52150 +Epoch: [29] [ 990/7269] eta: 0:45:57 lr: 0.000003 loss: 0.5161 (0.5179) time: 0.4186 data: 0.0001 max mem: 52150 +Epoch: [29] [1000/7269] eta: 0:45:56 lr: 0.000003 loss: 0.5322 (0.5180) time: 0.4324 data: 0.0001 max mem: 52150 +Epoch: [29] [1010/7269] eta: 0:45:47 lr: 0.000003 loss: 0.5210 (0.5179) time: 0.4328 data: 0.0002 max mem: 52150 +Epoch: [29] [1020/7269] eta: 0:45:39 lr: 0.000003 loss: 0.5105 (0.5177) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [29] [1030/7269] eta: 0:45:36 lr: 0.000003 loss: 0.5141 (0.5177) time: 0.4171 data: 0.0001 max mem: 52150 +Epoch: [29] [1040/7269] eta: 0:45:33 lr: 0.000003 loss: 0.5186 (0.5176) time: 0.4620 data: 0.0002 max mem: 52150 +Epoch: [29] [1050/7269] eta: 0:45:24 lr: 0.000003 loss: 0.5186 (0.5176) time: 0.4168 data: 0.0002 max mem: 52150 +Epoch: [29] [1060/7269] eta: 0:45:23 lr: 0.000003 loss: 0.5209 (0.5176) time: 0.4302 data: 0.0002 max mem: 52150 +Epoch: [29] [1070/7269] eta: 0:45:15 lr: 0.000003 loss: 0.5213 (0.5178) time: 0.4303 data: 0.0002 max mem: 52150 +Epoch: [29] [1080/7269] eta: 0:45:07 lr: 0.000003 loss: 0.5609 (0.5182) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [29] [1090/7269] eta: 0:45:05 lr: 0.000003 loss: 0.5412 (0.5182) time: 0.4322 data: 0.0002 max mem: 52150 +Epoch: [29] [1100/7269] eta: 0:45:04 lr: 0.000003 loss: 0.5332 (0.5184) time: 0.4923 data: 0.0002 max mem: 52150 +Epoch: [29] [1110/7269] eta: 0:44:56 lr: 0.000003 loss: 0.5371 (0.5185) time: 0.4321 data: 0.0001 max mem: 52150 +Epoch: [29] [1120/7269] eta: 0:44:54 lr: 0.000003 loss: 0.5371 (0.5185) time: 0.4316 data: 0.0002 max mem: 52150 +Epoch: [29] [1130/7269] eta: 0:44:46 lr: 0.000003 loss: 0.5102 (0.5183) time: 0.4313 data: 0.0001 max mem: 52150 +Epoch: [29] [1140/7269] eta: 0:44:39 lr: 0.000003 loss: 0.5300 (0.5184) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [29] [1150/7269] eta: 0:44:31 lr: 0.000003 loss: 0.5432 (0.5184) time: 0.3726 data: 0.0001 max mem: 52150 +Epoch: [29] [1160/7269] eta: 0:44:23 lr: 0.000003 loss: 0.5618 (0.5186) time: 0.3726 data: 0.0001 max mem: 52150 +Epoch: [29] [1170/7269] eta: 0:44:22 lr: 0.000003 loss: 0.5220 (0.5187) time: 0.4355 data: 0.0001 max mem: 52150 +Epoch: [29] [1180/7269] eta: 0:44:14 lr: 0.000003 loss: 0.5296 (0.5190) time: 0.4354 data: 0.0001 max mem: 52150 +Epoch: [29] [1190/7269] eta: 0:44:13 lr: 0.000003 loss: 0.5098 (0.5188) time: 0.4317 data: 0.0001 max mem: 52150 +Epoch: [29] [1200/7269] eta: 0:44:10 lr: 0.000003 loss: 0.5075 (0.5185) time: 0.4768 data: 0.0001 max mem: 52150 +Epoch: [29] [1210/7269] eta: 0:44:07 lr: 0.000003 loss: 0.5126 (0.5185) time: 0.4621 data: 0.0002 max mem: 52150 +Epoch: [29] [1220/7269] eta: 0:44:04 lr: 0.000003 loss: 0.5276 (0.5187) time: 0.4630 data: 0.0002 max mem: 52150 +Epoch: [29] [1230/7269] eta: 0:44:01 lr: 0.000003 loss: 0.5473 (0.5188) time: 0.4648 data: 0.0001 max mem: 52150 +Epoch: [29] [1240/7269] eta: 0:43:53 lr: 0.000003 loss: 0.5251 (0.5184) time: 0.4187 data: 0.0001 max mem: 52150 +Epoch: [29] [1250/7269] eta: 0:43:46 lr: 0.000003 loss: 0.5184 (0.5184) time: 0.3718 data: 0.0001 max mem: 52150 +Epoch: [29] [1260/7269] eta: 0:43:38 lr: 0.000003 loss: 0.5184 (0.5182) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [29] [1270/7269] eta: 0:43:31 lr: 0.000003 loss: 0.5251 (0.5182) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [29] [1280/7269] eta: 0:43:23 lr: 0.000003 loss: 0.5388 (0.5181) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [29] [1290/7269] eta: 0:43:22 lr: 0.000003 loss: 0.4871 (0.5180) time: 0.4326 data: 0.0002 max mem: 52150 +Epoch: [29] [1300/7269] eta: 0:43:23 lr: 0.000003 loss: 0.5004 (0.5181) time: 0.5237 data: 0.0002 max mem: 52150 +Epoch: [29] [1310/7269] eta: 0:43:20 lr: 0.000003 loss: 0.5387 (0.5180) time: 0.5087 data: 0.0002 max mem: 52150 +Epoch: [29] [1320/7269] eta: 0:43:13 lr: 0.000003 loss: 0.5186 (0.5181) time: 0.4181 data: 0.0002 max mem: 52150 +Epoch: [29] [1330/7269] eta: 0:43:09 lr: 0.000003 loss: 0.5084 (0.5178) time: 0.4166 data: 0.0002 max mem: 52150 +Epoch: [29] [1340/7269] eta: 0:43:02 lr: 0.000003 loss: 0.5213 (0.5178) time: 0.4161 data: 0.0002 max mem: 52150 +Epoch: [29] [1350/7269] eta: 0:43:08 lr: 0.000003 loss: 0.5409 (0.5178) time: 0.5247 data: 0.0002 max mem: 52150 +Epoch: [29] [1360/7269] eta: 0:43:01 lr: 0.000003 loss: 0.5260 (0.5177) time: 0.5252 data: 0.0002 max mem: 52150 +Epoch: [29] [1370/7269] eta: 0:42:54 lr: 0.000003 loss: 0.5107 (0.5177) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [29] [1380/7269] eta: 0:42:56 lr: 0.000003 loss: 0.5107 (0.5176) time: 0.4796 data: 0.0002 max mem: 52150 +Epoch: [29] [1390/7269] eta: 0:42:49 lr: 0.000003 loss: 0.5419 (0.5177) time: 0.4799 data: 0.0002 max mem: 52150 +Epoch: [29] [1400/7269] eta: 0:42:42 lr: 0.000003 loss: 0.5259 (0.5176) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [29] [1410/7269] eta: 0:42:35 lr: 0.000003 loss: 0.5017 (0.5173) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [29] [1420/7269] eta: 0:42:33 lr: 0.000003 loss: 0.5365 (0.5176) time: 0.4320 data: 0.0002 max mem: 52150 +Epoch: [29] [1430/7269] eta: 0:42:31 lr: 0.000003 loss: 0.5365 (0.5176) time: 0.4909 data: 0.0002 max mem: 52150 +Epoch: [29] [1440/7269] eta: 0:42:24 lr: 0.000003 loss: 0.5205 (0.5175) time: 0.4310 data: 0.0002 max mem: 52150 +Epoch: [29] [1450/7269] eta: 0:42:17 lr: 0.000003 loss: 0.5205 (0.5175) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [29] [1460/7269] eta: 0:42:10 lr: 0.000003 loss: 0.5121 (0.5175) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [29] [1470/7269] eta: 0:42:03 lr: 0.000003 loss: 0.5121 (0.5175) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [29] [1480/7269] eta: 0:41:56 lr: 0.000003 loss: 0.5128 (0.5173) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [29] [1490/7269] eta: 0:41:53 lr: 0.000003 loss: 0.4802 (0.5170) time: 0.4183 data: 0.0002 max mem: 52150 +Epoch: [29] [1500/7269] eta: 0:41:46 lr: 0.000003 loss: 0.4880 (0.5170) time: 0.4187 data: 0.0002 max mem: 52150 +Epoch: [29] [1510/7269] eta: 0:41:39 lr: 0.000003 loss: 0.5365 (0.5171) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [29] [1520/7269] eta: 0:41:42 lr: 0.000003 loss: 0.5478 (0.5173) time: 0.4918 data: 0.0002 max mem: 52150 +Epoch: [29] [1530/7269] eta: 0:41:35 lr: 0.000003 loss: 0.5356 (0.5170) time: 0.4920 data: 0.0002 max mem: 52150 +Epoch: [29] [1540/7269] eta: 0:41:28 lr: 0.000003 loss: 0.4861 (0.5168) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [29] [1550/7269] eta: 0:41:22 lr: 0.000003 loss: 0.4878 (0.5166) time: 0.3729 data: 0.0002 max mem: 52150 +Epoch: [29] [1560/7269] eta: 0:41:20 lr: 0.000003 loss: 0.4915 (0.5167) time: 0.4331 data: 0.0002 max mem: 52150 +Epoch: [29] [1570/7269] eta: 0:41:13 lr: 0.000003 loss: 0.5109 (0.5166) time: 0.4325 data: 0.0002 max mem: 52150 +Epoch: [29] [1580/7269] eta: 0:41:10 lr: 0.000003 loss: 0.5459 (0.5168) time: 0.4177 data: 0.0002 max mem: 52150 +Epoch: [29] [1590/7269] eta: 0:41:03 lr: 0.000003 loss: 0.5043 (0.5166) time: 0.4176 data: 0.0002 max mem: 52150 +Epoch: [29] [1600/7269] eta: 0:41:01 lr: 0.000003 loss: 0.4815 (0.5164) time: 0.4328 data: 0.0002 max mem: 52150 +Epoch: [29] [1610/7269] eta: 0:40:54 lr: 0.000003 loss: 0.5047 (0.5166) time: 0.4326 data: 0.0002 max mem: 52150 +Epoch: [29] [1620/7269] eta: 0:40:51 lr: 0.000003 loss: 0.5209 (0.5165) time: 0.4170 data: 0.0002 max mem: 52150 +Epoch: [29] [1630/7269] eta: 0:40:49 lr: 0.000003 loss: 0.5286 (0.5167) time: 0.4794 data: 0.0002 max mem: 52150 +Epoch: [29] [1640/7269] eta: 0:40:42 lr: 0.000003 loss: 0.5331 (0.5167) time: 0.4344 data: 0.0002 max mem: 52150 +Epoch: [29] [1650/7269] eta: 0:40:43 lr: 0.000003 loss: 0.5414 (0.5169) time: 0.4785 data: 0.0002 max mem: 52150 +Epoch: [29] [1660/7269] eta: 0:40:37 lr: 0.000003 loss: 0.5566 (0.5171) time: 0.4781 data: 0.0002 max mem: 52150 +Epoch: [29] [1670/7269] eta: 0:40:36 lr: 0.000003 loss: 0.5550 (0.5172) time: 0.4631 data: 0.0002 max mem: 52150 +Epoch: [29] [1680/7269] eta: 0:40:30 lr: 0.000003 loss: 0.4957 (0.5168) time: 0.4631 data: 0.0002 max mem: 52150 +Epoch: [29] [1690/7269] eta: 0:40:23 lr: 0.000003 loss: 0.4964 (0.5169) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [29] [1700/7269] eta: 0:40:17 lr: 0.000003 loss: 0.5147 (0.5168) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [29] [1710/7269] eta: 0:40:11 lr: 0.000003 loss: 0.5059 (0.5168) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [29] [1720/7269] eta: 0:40:07 lr: 0.000003 loss: 0.5421 (0.5170) time: 0.4178 data: 0.0002 max mem: 52150 +Epoch: [29] [1730/7269] eta: 0:40:01 lr: 0.000003 loss: 0.5150 (0.5168) time: 0.4178 data: 0.0002 max mem: 52150 +Epoch: [29] [1740/7269] eta: 0:39:58 lr: 0.000003 loss: 0.5022 (0.5167) time: 0.4240 data: 0.0002 max mem: 52150 +Epoch: [29] [1750/7269] eta: 0:39:52 lr: 0.000003 loss: 0.5185 (0.5169) time: 0.4240 data: 0.0002 max mem: 52150 +Epoch: [29] [1760/7269] eta: 0:39:45 lr: 0.000003 loss: 0.5388 (0.5168) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [29] [1770/7269] eta: 0:39:46 lr: 0.000003 loss: 0.5227 (0.5169) time: 0.4800 data: 0.0002 max mem: 52150 +Epoch: [29] [1780/7269] eta: 0:39:43 lr: 0.000003 loss: 0.5172 (0.5167) time: 0.5400 data: 0.0002 max mem: 52150 +Epoch: [29] [1790/7269] eta: 0:39:47 lr: 0.000003 loss: 0.4722 (0.5164) time: 0.5856 data: 0.0002 max mem: 52150 +Epoch: [29] [1800/7269] eta: 0:39:43 lr: 0.000003 loss: 0.4929 (0.5164) time: 0.5710 data: 0.0002 max mem: 52150 +Epoch: [29] [1810/7269] eta: 0:39:40 lr: 0.000003 loss: 0.5471 (0.5167) time: 0.4772 data: 0.0002 max mem: 52150 +Epoch: [29] [1820/7269] eta: 0:39:37 lr: 0.000003 loss: 0.5191 (0.5163) time: 0.4784 data: 0.0002 max mem: 52150 +Epoch: [29] [1830/7269] eta: 0:39:34 lr: 0.000003 loss: 0.5040 (0.5163) time: 0.4782 data: 0.0002 max mem: 52150 +Epoch: [29] [1840/7269] eta: 0:39:34 lr: 0.000003 loss: 0.5349 (0.5162) time: 0.5397 data: 0.0002 max mem: 52150 +Epoch: [29] [1850/7269] eta: 0:39:28 lr: 0.000003 loss: 0.5181 (0.5162) time: 0.4799 data: 0.0002 max mem: 52150 +Epoch: [29] [1860/7269] eta: 0:39:22 lr: 0.000003 loss: 0.5172 (0.5161) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [29] [1870/7269] eta: 0:39:15 lr: 0.000003 loss: 0.4964 (0.5160) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [29] [1880/7269] eta: 0:39:09 lr: 0.000003 loss: 0.5090 (0.5160) time: 0.3726 data: 0.0002 max mem: 52150 +Epoch: [29] [1890/7269] eta: 0:39:09 lr: 0.000003 loss: 0.5191 (0.5162) time: 0.4849 data: 0.0002 max mem: 52150 +Epoch: [29] [1900/7269] eta: 0:39:03 lr: 0.000003 loss: 0.4992 (0.5160) time: 0.4852 data: 0.0002 max mem: 52150 +Epoch: [29] [1910/7269] eta: 0:39:00 lr: 0.000003 loss: 0.4858 (0.5160) time: 0.4168 data: 0.0002 max mem: 52150 +Epoch: [29] [1920/7269] eta: 0:38:53 lr: 0.000003 loss: 0.5328 (0.5160) time: 0.4161 data: 0.0002 max mem: 52150 +Epoch: [29] [1930/7269] eta: 0:38:47 lr: 0.000003 loss: 0.5445 (0.5161) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [29] [1940/7269] eta: 0:38:41 lr: 0.000003 loss: 0.5093 (0.5160) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [29] [1950/7269] eta: 0:38:35 lr: 0.000003 loss: 0.5303 (0.5160) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [29] [1960/7269] eta: 0:38:29 lr: 0.000003 loss: 0.5344 (0.5160) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [29] [1970/7269] eta: 0:38:23 lr: 0.000003 loss: 0.4912 (0.5159) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [29] [1980/7269] eta: 0:38:17 lr: 0.000003 loss: 0.4618 (0.5157) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [29] [1990/7269] eta: 0:38:11 lr: 0.000003 loss: 0.4736 (0.5158) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [29] [2000/7269] eta: 0:38:05 lr: 0.000003 loss: 0.5199 (0.5157) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [29] [2010/7269] eta: 0:37:59 lr: 0.000003 loss: 0.5039 (0.5157) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [29] [2020/7269] eta: 0:37:53 lr: 0.000003 loss: 0.5039 (0.5156) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [29] [2030/7269] eta: 0:37:47 lr: 0.000003 loss: 0.5271 (0.5157) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [29] [2040/7269] eta: 0:37:41 lr: 0.000003 loss: 0.5271 (0.5156) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [29] [2050/7269] eta: 0:37:38 lr: 0.000003 loss: 0.5424 (0.5157) time: 0.4322 data: 0.0002 max mem: 52150 +Epoch: [29] [2060/7269] eta: 0:37:33 lr: 0.000003 loss: 0.5366 (0.5156) time: 0.4322 data: 0.0002 max mem: 52150 +Epoch: [29] [2070/7269] eta: 0:37:27 lr: 0.000003 loss: 0.5100 (0.5156) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [29] [2080/7269] eta: 0:37:21 lr: 0.000003 loss: 0.5081 (0.5156) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [29] [2090/7269] eta: 0:37:15 lr: 0.000003 loss: 0.5065 (0.5155) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [29] [2100/7269] eta: 0:37:11 lr: 0.000003 loss: 0.5140 (0.5155) time: 0.4170 data: 0.0002 max mem: 52150 +Epoch: [29] [2110/7269] eta: 0:37:09 lr: 0.000003 loss: 0.5116 (0.5155) time: 0.4770 data: 0.0002 max mem: 52150 +Epoch: [29] [2120/7269] eta: 0:37:05 lr: 0.000003 loss: 0.5319 (0.5156) time: 0.4791 data: 0.0002 max mem: 52150 +Epoch: [29] [2130/7269] eta: 0:36:59 lr: 0.000003 loss: 0.5230 (0.5156) time: 0.4186 data: 0.0002 max mem: 52150 +Epoch: [29] [2140/7269] eta: 0:36:54 lr: 0.000003 loss: 0.5230 (0.5156) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [29] [2150/7269] eta: 0:36:50 lr: 0.000003 loss: 0.5321 (0.5156) time: 0.4174 data: 0.0002 max mem: 52150 +Epoch: [29] [2160/7269] eta: 0:36:44 lr: 0.000003 loss: 0.5435 (0.5157) time: 0.4178 data: 0.0002 max mem: 52150 +Epoch: [29] [2170/7269] eta: 0:36:39 lr: 0.000003 loss: 0.5559 (0.5159) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [29] [2180/7269] eta: 0:36:35 lr: 0.000003 loss: 0.5559 (0.5160) time: 0.4165 data: 0.0002 max mem: 52150 +Epoch: [29] [2190/7269] eta: 0:36:29 lr: 0.000003 loss: 0.5395 (0.5160) time: 0.4166 data: 0.0002 max mem: 52150 +Epoch: [29] [2200/7269] eta: 0:36:24 lr: 0.000003 loss: 0.5238 (0.5159) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [29] [2210/7269] eta: 0:36:20 lr: 0.000003 loss: 0.5238 (0.5160) time: 0.4219 data: 0.0002 max mem: 52150 +Epoch: [29] [2220/7269] eta: 0:36:17 lr: 0.000003 loss: 0.5360 (0.5160) time: 0.4827 data: 0.0002 max mem: 52150 +Epoch: [29] [2230/7269] eta: 0:36:14 lr: 0.000003 loss: 0.5065 (0.5160) time: 0.4948 data: 0.0002 max mem: 52150 +Epoch: [29] [2240/7269] eta: 0:36:11 lr: 0.000003 loss: 0.5065 (0.5160) time: 0.4807 data: 0.0002 max mem: 52150 +Epoch: [29] [2250/7269] eta: 0:36:11 lr: 0.000003 loss: 0.5042 (0.5160) time: 0.5429 data: 0.0002 max mem: 52150 +Epoch: [29] [2260/7269] eta: 0:36:09 lr: 0.000003 loss: 0.5445 (0.5161) time: 0.5880 data: 0.0002 max mem: 52150 +Epoch: [29] [2270/7269] eta: 0:36:04 lr: 0.000003 loss: 0.5397 (0.5160) time: 0.4635 data: 0.0002 max mem: 52150 +Epoch: [29] [2280/7269] eta: 0:35:58 lr: 0.000003 loss: 0.5397 (0.5161) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [29] [2290/7269] eta: 0:35:52 lr: 0.000003 loss: 0.5266 (0.5162) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [29] [2300/7269] eta: 0:35:49 lr: 0.000003 loss: 0.5266 (0.5162) time: 0.4177 data: 0.0002 max mem: 52150 +Epoch: [29] [2310/7269] eta: 0:35:46 lr: 0.000003 loss: 0.5409 (0.5162) time: 0.4790 data: 0.0002 max mem: 52150 +Epoch: [29] [2320/7269] eta: 0:35:40 lr: 0.000003 loss: 0.4733 (0.5161) time: 0.4338 data: 0.0002 max mem: 52150 +Epoch: [29] [2330/7269] eta: 0:35:42 lr: 0.000003 loss: 0.4733 (0.5160) time: 0.5588 data: 0.0002 max mem: 52150 +Epoch: [29] [2340/7269] eta: 0:35:37 lr: 0.000003 loss: 0.4765 (0.5160) time: 0.5590 data: 0.0002 max mem: 52150 +Epoch: [29] [2350/7269] eta: 0:35:36 lr: 0.000003 loss: 0.5142 (0.5159) time: 0.4845 data: 0.0002 max mem: 52150 +Epoch: [29] [2360/7269] eta: 0:35:32 lr: 0.000003 loss: 0.5241 (0.5160) time: 0.5304 data: 0.0002 max mem: 52150 +Epoch: [29] [2370/7269] eta: 0:35:26 lr: 0.000003 loss: 0.5142 (0.5160) time: 0.4178 data: 0.0002 max mem: 52150 +Epoch: [29] [2380/7269] eta: 0:35:26 lr: 0.000003 loss: 0.5087 (0.5159) time: 0.5076 data: 0.0002 max mem: 52150 +Epoch: [29] [2390/7269] eta: 0:35:21 lr: 0.000003 loss: 0.4748 (0.5157) time: 0.5084 data: 0.0002 max mem: 52150 +Epoch: [29] [2400/7269] eta: 0:35:15 lr: 0.000003 loss: 0.5044 (0.5157) time: 0.3729 data: 0.0002 max mem: 52150 +Epoch: [29] [2410/7269] eta: 0:35:09 lr: 0.000003 loss: 0.5239 (0.5157) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [29] [2420/7269] eta: 0:35:04 lr: 0.000003 loss: 0.5239 (0.5158) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [29] [2430/7269] eta: 0:35:00 lr: 0.000003 loss: 0.5179 (0.5157) time: 0.4178 data: 0.0002 max mem: 52150 +Epoch: [29] [2440/7269] eta: 0:34:55 lr: 0.000003 loss: 0.5179 (0.5158) time: 0.4180 data: 0.0002 max mem: 52150 +Epoch: [29] [2450/7269] eta: 0:34:51 lr: 0.000003 loss: 0.5059 (0.5158) time: 0.4212 data: 0.0002 max mem: 52150 +Epoch: [29] [2460/7269] eta: 0:34:48 lr: 0.000003 loss: 0.5059 (0.5158) time: 0.4814 data: 0.0002 max mem: 52150 +Epoch: [29] [2470/7269] eta: 0:34:42 lr: 0.000003 loss: 0.5171 (0.5158) time: 0.4326 data: 0.0002 max mem: 52150 +Epoch: [29] [2480/7269] eta: 0:34:37 lr: 0.000003 loss: 0.5218 (0.5158) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [29] [2490/7269] eta: 0:34:31 lr: 0.000003 loss: 0.5362 (0.5159) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [29] [2500/7269] eta: 0:34:26 lr: 0.000003 loss: 0.5211 (0.5158) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [29] [2510/7269] eta: 0:34:22 lr: 0.000003 loss: 0.4649 (0.5158) time: 0.4319 data: 0.0002 max mem: 52150 +Epoch: [29] [2520/7269] eta: 0:34:17 lr: 0.000003 loss: 0.4775 (0.5157) time: 0.4319 data: 0.0002 max mem: 52150 +Epoch: [29] [2530/7269] eta: 0:34:11 lr: 0.000003 loss: 0.4811 (0.5157) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [29] [2540/7269] eta: 0:34:06 lr: 0.000003 loss: 0.4851 (0.5156) time: 0.3728 data: 0.0002 max mem: 52150 +Epoch: [29] [2550/7269] eta: 0:34:01 lr: 0.000003 loss: 0.5103 (0.5156) time: 0.3732 data: 0.0002 max mem: 52150 +Epoch: [29] [2560/7269] eta: 0:33:57 lr: 0.000003 loss: 0.5299 (0.5158) time: 0.4177 data: 0.0002 max mem: 52150 +Epoch: [29] [2570/7269] eta: 0:33:54 lr: 0.000003 loss: 0.5299 (0.5157) time: 0.4772 data: 0.0002 max mem: 52150 +Epoch: [29] [2580/7269] eta: 0:33:50 lr: 0.000003 loss: 0.5017 (0.5157) time: 0.4762 data: 0.0002 max mem: 52150 +Epoch: [29] [2590/7269] eta: 0:33:44 lr: 0.000003 loss: 0.4988 (0.5156) time: 0.4165 data: 0.0002 max mem: 52150 +Epoch: [29] [2600/7269] eta: 0:33:39 lr: 0.000003 loss: 0.4710 (0.5154) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [29] [2610/7269] eta: 0:33:34 lr: 0.000003 loss: 0.5055 (0.5155) time: 0.3738 data: 0.0002 max mem: 52150 +Epoch: [29] [2620/7269] eta: 0:33:30 lr: 0.000003 loss: 0.5148 (0.5155) time: 0.4205 data: 0.0002 max mem: 52150 +Epoch: [29] [2630/7269] eta: 0:33:24 lr: 0.000003 loss: 0.5148 (0.5155) time: 0.4193 data: 0.0002 max mem: 52150 +Epoch: [29] [2640/7269] eta: 0:33:19 lr: 0.000003 loss: 0.5327 (0.5155) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [29] [2650/7269] eta: 0:33:17 lr: 0.000003 loss: 0.5280 (0.5154) time: 0.4638 data: 0.0002 max mem: 52150 +Epoch: [29] [2660/7269] eta: 0:33:12 lr: 0.000003 loss: 0.5301 (0.5154) time: 0.4637 data: 0.0002 max mem: 52150 +Epoch: [29] [2670/7269] eta: 0:33:06 lr: 0.000003 loss: 0.5301 (0.5153) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [29] [2680/7269] eta: 0:33:01 lr: 0.000003 loss: 0.5322 (0.5155) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [29] [2690/7269] eta: 0:32:57 lr: 0.000003 loss: 0.5407 (0.5155) time: 0.4190 data: 0.0002 max mem: 52150 +Epoch: [29] [2700/7269] eta: 0:32:53 lr: 0.000003 loss: 0.4938 (0.5154) time: 0.4634 data: 0.0002 max mem: 52150 +Epoch: [29] [2710/7269] eta: 0:32:48 lr: 0.000003 loss: 0.5069 (0.5154) time: 0.4169 data: 0.0002 max mem: 52150 +Epoch: [29] [2720/7269] eta: 0:32:43 lr: 0.000003 loss: 0.4955 (0.5154) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [29] [2730/7269] eta: 0:32:37 lr: 0.000003 loss: 0.5153 (0.5155) time: 0.3723 data: 0.0002 max mem: 52150 +Epoch: [29] [2740/7269] eta: 0:32:32 lr: 0.000003 loss: 0.5233 (0.5154) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [29] [2750/7269] eta: 0:32:27 lr: 0.000003 loss: 0.4991 (0.5154) time: 0.3727 data: 0.0002 max mem: 52150 +Epoch: [29] [2760/7269] eta: 0:32:22 lr: 0.000003 loss: 0.5323 (0.5154) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [29] [2770/7269] eta: 0:32:16 lr: 0.000003 loss: 0.5323 (0.5154) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [29] [2780/7269] eta: 0:32:11 lr: 0.000003 loss: 0.5045 (0.5153) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [29] [2790/7269] eta: 0:32:06 lr: 0.000003 loss: 0.5002 (0.5152) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [29] [2800/7269] eta: 0:32:03 lr: 0.000003 loss: 0.4989 (0.5151) time: 0.4350 data: 0.0002 max mem: 52150 +Epoch: [29] [2810/7269] eta: 0:31:57 lr: 0.000003 loss: 0.4886 (0.5151) time: 0.4360 data: 0.0002 max mem: 52150 +Epoch: [29] [2820/7269] eta: 0:31:54 lr: 0.000003 loss: 0.4899 (0.5151) time: 0.4333 data: 0.0002 max mem: 52150 +Epoch: [29] [2830/7269] eta: 0:31:53 lr: 0.000003 loss: 0.4902 (0.5151) time: 0.5546 data: 0.0002 max mem: 52150 +Epoch: [29] [2840/7269] eta: 0:31:51 lr: 0.000003 loss: 0.5071 (0.5150) time: 0.6032 data: 0.0002 max mem: 52150 +Epoch: [29] [2850/7269] eta: 0:31:47 lr: 0.000003 loss: 0.5318 (0.5152) time: 0.5270 data: 0.0002 max mem: 52150 +Epoch: [29] [2860/7269] eta: 0:31:42 lr: 0.000003 loss: 0.5235 (0.5152) time: 0.4177 data: 0.0002 max mem: 52150 +Epoch: [29] [2870/7269] eta: 0:31:37 lr: 0.000003 loss: 0.4897 (0.5150) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [29] [2880/7269] eta: 0:31:31 lr: 0.000003 loss: 0.4333 (0.5149) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [29] [2890/7269] eta: 0:31:27 lr: 0.000003 loss: 0.4742 (0.5149) time: 0.4170 data: 0.0002 max mem: 52150 +Epoch: [29] [2900/7269] eta: 0:31:22 lr: 0.000003 loss: 0.5029 (0.5149) time: 0.4171 data: 0.0002 max mem: 52150 +Epoch: [29] [2910/7269] eta: 0:31:17 lr: 0.000003 loss: 0.5321 (0.5149) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [29] [2920/7269] eta: 0:31:13 lr: 0.000003 loss: 0.5321 (0.5150) time: 0.4186 data: 0.0002 max mem: 52150 +Epoch: [29] [2930/7269] eta: 0:31:10 lr: 0.000003 loss: 0.5276 (0.5149) time: 0.4811 data: 0.0002 max mem: 52150 +Epoch: [29] [2940/7269] eta: 0:31:06 lr: 0.000003 loss: 0.5663 (0.5151) time: 0.4837 data: 0.0002 max mem: 52150 +Epoch: [29] [2950/7269] eta: 0:31:01 lr: 0.000003 loss: 0.5431 (0.5151) time: 0.4208 data: 0.0002 max mem: 52150 +Epoch: [29] [2960/7269] eta: 0:30:59 lr: 0.000003 loss: 0.4896 (0.5151) time: 0.4926 data: 0.0002 max mem: 52150 +Epoch: [29] [2970/7269] eta: 0:30:56 lr: 0.000003 loss: 0.5145 (0.5150) time: 0.5587 data: 0.0002 max mem: 52150 +Epoch: [29] [2980/7269] eta: 0:30:52 lr: 0.000003 loss: 0.5355 (0.5151) time: 0.4825 data: 0.0002 max mem: 52150 +Epoch: [29] [2990/7269] eta: 0:30:47 lr: 0.000003 loss: 0.5317 (0.5151) time: 0.4169 data: 0.0002 max mem: 52150 +Epoch: [29] [3000/7269] eta: 0:30:42 lr: 0.000003 loss: 0.5007 (0.5150) time: 0.3727 data: 0.0002 max mem: 52150 +Epoch: [29] [3010/7269] eta: 0:30:37 lr: 0.000003 loss: 0.5092 (0.5150) time: 0.3728 data: 0.0002 max mem: 52150 +Epoch: [29] [3020/7269] eta: 0:30:33 lr: 0.000003 loss: 0.5286 (0.5151) time: 0.4188 data: 0.0002 max mem: 52150 +Epoch: [29] [3030/7269] eta: 0:30:29 lr: 0.000003 loss: 0.5135 (0.5151) time: 0.4802 data: 0.0002 max mem: 52150 +Epoch: [29] [3040/7269] eta: 0:30:26 lr: 0.000003 loss: 0.5391 (0.5152) time: 0.4968 data: 0.0002 max mem: 52150 +Epoch: [29] [3050/7269] eta: 0:30:21 lr: 0.000003 loss: 0.5363 (0.5153) time: 0.4358 data: 0.0002 max mem: 52150 +Epoch: [29] [3060/7269] eta: 0:30:18 lr: 0.000003 loss: 0.5208 (0.5152) time: 0.4645 data: 0.0002 max mem: 52150 +Epoch: [29] [3070/7269] eta: 0:30:17 lr: 0.000003 loss: 0.5196 (0.5151) time: 0.5852 data: 0.0002 max mem: 52150 +Epoch: [29] [3080/7269] eta: 0:30:14 lr: 0.000003 loss: 0.5281 (0.5151) time: 0.6034 data: 0.0002 max mem: 52150 +Epoch: [29] [3090/7269] eta: 0:30:09 lr: 0.000003 loss: 0.5402 (0.5151) time: 0.4820 data: 0.0002 max mem: 52150 +Epoch: [29] [3100/7269] eta: 0:30:04 lr: 0.000003 loss: 0.4644 (0.5149) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [29] [3110/7269] eta: 0:29:59 lr: 0.000003 loss: 0.4644 (0.5148) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [29] [3120/7269] eta: 0:29:54 lr: 0.000003 loss: 0.5603 (0.5150) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [29] [3130/7269] eta: 0:29:50 lr: 0.000003 loss: 0.5463 (0.5150) time: 0.4347 data: 0.0002 max mem: 52150 +Epoch: [29] [3140/7269] eta: 0:29:46 lr: 0.000003 loss: 0.5251 (0.5150) time: 0.4806 data: 0.0002 max mem: 52150 +Epoch: [29] [3150/7269] eta: 0:29:41 lr: 0.000003 loss: 0.5011 (0.5150) time: 0.4185 data: 0.0002 max mem: 52150 +Epoch: [29] [3160/7269] eta: 0:29:36 lr: 0.000003 loss: 0.5150 (0.5150) time: 0.3731 data: 0.0002 max mem: 52150 +Epoch: [29] [3170/7269] eta: 0:29:33 lr: 0.000003 loss: 0.5150 (0.5149) time: 0.4658 data: 0.0002 max mem: 52150 +Epoch: [29] [3180/7269] eta: 0:29:28 lr: 0.000003 loss: 0.5228 (0.5149) time: 0.4653 data: 0.0002 max mem: 52150 +Epoch: [29] [3190/7269] eta: 0:29:24 lr: 0.000003 loss: 0.5112 (0.5148) time: 0.4189 data: 0.0002 max mem: 52150 +Epoch: [29] [3200/7269] eta: 0:29:19 lr: 0.000003 loss: 0.4994 (0.5148) time: 0.4190 data: 0.0002 max mem: 52150 +Epoch: [29] [3210/7269] eta: 0:29:14 lr: 0.000003 loss: 0.5168 (0.5148) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [29] [3220/7269] eta: 0:29:11 lr: 0.000003 loss: 0.5249 (0.5148) time: 0.4257 data: 0.0002 max mem: 52150 +Epoch: [29] [3230/7269] eta: 0:29:05 lr: 0.000003 loss: 0.5470 (0.5150) time: 0.4255 data: 0.0002 max mem: 52150 +Epoch: [29] [3240/7269] eta: 0:29:00 lr: 0.000003 loss: 0.5427 (0.5150) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [29] [3250/7269] eta: 0:28:55 lr: 0.000003 loss: 0.5234 (0.5150) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [29] [3260/7269] eta: 0:28:50 lr: 0.000003 loss: 0.5248 (0.5150) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [29] [3270/7269] eta: 0:28:45 lr: 0.000003 loss: 0.5193 (0.5149) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [29] [3280/7269] eta: 0:28:42 lr: 0.000003 loss: 0.5219 (0.5150) time: 0.4342 data: 0.0002 max mem: 52150 +Epoch: [29] [3290/7269] eta: 0:28:38 lr: 0.000003 loss: 0.5419 (0.5149) time: 0.4796 data: 0.0002 max mem: 52150 +Epoch: [29] [3300/7269] eta: 0:28:33 lr: 0.000003 loss: 0.4915 (0.5149) time: 0.4174 data: 0.0002 max mem: 52150 +Epoch: [29] [3310/7269] eta: 0:28:28 lr: 0.000003 loss: 0.4999 (0.5149) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [29] [3320/7269] eta: 0:28:24 lr: 0.000003 loss: 0.4999 (0.5147) time: 0.4174 data: 0.0002 max mem: 52150 +Epoch: [29] [3330/7269] eta: 0:28:19 lr: 0.000003 loss: 0.4936 (0.5147) time: 0.4179 data: 0.0002 max mem: 52150 +Epoch: [29] [3340/7269] eta: 0:28:14 lr: 0.000003 loss: 0.5629 (0.5148) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [29] [3350/7269] eta: 0:28:12 lr: 0.000003 loss: 0.5284 (0.5147) time: 0.4955 data: 0.0002 max mem: 52150 +Epoch: [29] [3360/7269] eta: 0:28:07 lr: 0.000003 loss: 0.5405 (0.5149) time: 0.4960 data: 0.0002 max mem: 52150 +Epoch: [29] [3370/7269] eta: 0:28:02 lr: 0.000003 loss: 0.5432 (0.5148) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [29] [3380/7269] eta: 0:27:57 lr: 0.000003 loss: 0.5244 (0.5149) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [29] [3390/7269] eta: 0:27:52 lr: 0.000003 loss: 0.5137 (0.5148) time: 0.3735 data: 0.0002 max mem: 52150 +Epoch: [29] [3400/7269] eta: 0:27:48 lr: 0.000003 loss: 0.5038 (0.5148) time: 0.4188 data: 0.0002 max mem: 52150 +Epoch: [29] [3410/7269] eta: 0:27:43 lr: 0.000003 loss: 0.5081 (0.5148) time: 0.4181 data: 0.0002 max mem: 52150 +Epoch: [29] [3420/7269] eta: 0:27:40 lr: 0.000003 loss: 0.5071 (0.5147) time: 0.4724 data: 0.0002 max mem: 52150 +Epoch: [29] [3430/7269] eta: 0:27:35 lr: 0.000003 loss: 0.5071 (0.5146) time: 0.4709 data: 0.0002 max mem: 52150 +Epoch: [29] [3440/7269] eta: 0:27:30 lr: 0.000003 loss: 0.5192 (0.5146) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [29] [3450/7269] eta: 0:27:28 lr: 0.000003 loss: 0.5630 (0.5148) time: 0.4989 data: 0.0002 max mem: 52150 +Epoch: [29] [3460/7269] eta: 0:27:24 lr: 0.000003 loss: 0.5547 (0.5147) time: 0.5431 data: 0.0002 max mem: 52150 +Epoch: [29] [3470/7269] eta: 0:27:19 lr: 0.000003 loss: 0.5304 (0.5148) time: 0.4170 data: 0.0002 max mem: 52150 +Epoch: [29] [3480/7269] eta: 0:27:14 lr: 0.000003 loss: 0.5041 (0.5147) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [29] [3490/7269] eta: 0:27:10 lr: 0.000003 loss: 0.5041 (0.5147) time: 0.4191 data: 0.0002 max mem: 52150 +Epoch: [29] [3500/7269] eta: 0:27:06 lr: 0.000003 loss: 0.5227 (0.5146) time: 0.4670 data: 0.0002 max mem: 52150 +Epoch: [29] [3510/7269] eta: 0:27:03 lr: 0.000003 loss: 0.5145 (0.5146) time: 0.4836 data: 0.0002 max mem: 52150 +Epoch: [29] [3520/7269] eta: 0:26:58 lr: 0.000003 loss: 0.5122 (0.5146) time: 0.4358 data: 0.0002 max mem: 52150 +Epoch: [29] [3530/7269] eta: 0:26:53 lr: 0.000003 loss: 0.4804 (0.5145) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [29] [3540/7269] eta: 0:26:52 lr: 0.000003 loss: 0.4858 (0.5144) time: 0.5603 data: 0.0002 max mem: 52150 +Epoch: [29] [3550/7269] eta: 0:26:47 lr: 0.000003 loss: 0.4819 (0.5144) time: 0.5614 data: 0.0002 max mem: 52150 +Epoch: [29] [3560/7269] eta: 0:26:42 lr: 0.000003 loss: 0.4739 (0.5144) time: 0.3729 data: 0.0002 max mem: 52150 +Epoch: [29] [3570/7269] eta: 0:26:37 lr: 0.000003 loss: 0.5447 (0.5145) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [29] [3580/7269] eta: 0:26:32 lr: 0.000003 loss: 0.5447 (0.5144) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [29] [3590/7269] eta: 0:26:27 lr: 0.000003 loss: 0.5031 (0.5144) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [29] [3600/7269] eta: 0:26:22 lr: 0.000003 loss: 0.5031 (0.5144) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [29] [3610/7269] eta: 0:26:17 lr: 0.000003 loss: 0.5353 (0.5144) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [29] [3620/7269] eta: 0:26:12 lr: 0.000003 loss: 0.5353 (0.5144) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [29] [3630/7269] eta: 0:26:09 lr: 0.000003 loss: 0.5327 (0.5145) time: 0.4816 data: 0.0002 max mem: 52150 +Epoch: [29] [3640/7269] eta: 0:26:05 lr: 0.000003 loss: 0.5188 (0.5145) time: 0.4816 data: 0.0002 max mem: 52150 +Epoch: [29] [3650/7269] eta: 0:26:00 lr: 0.000003 loss: 0.5034 (0.5145) time: 0.3739 data: 0.0002 max mem: 52150 +Epoch: [29] [3660/7269] eta: 0:25:56 lr: 0.000003 loss: 0.5337 (0.5145) time: 0.4186 data: 0.0002 max mem: 52150 +Epoch: [29] [3670/7269] eta: 0:25:53 lr: 0.000003 loss: 0.5395 (0.5146) time: 0.5238 data: 0.0002 max mem: 52150 +Epoch: [29] [3680/7269] eta: 0:25:48 lr: 0.000003 loss: 0.5395 (0.5146) time: 0.4783 data: 0.0002 max mem: 52150 +Epoch: [29] [3690/7269] eta: 0:25:44 lr: 0.000003 loss: 0.5142 (0.5146) time: 0.4360 data: 0.0002 max mem: 52150 +Epoch: [29] [3700/7269] eta: 0:25:40 lr: 0.000003 loss: 0.5142 (0.5146) time: 0.4830 data: 0.0002 max mem: 52150 +Epoch: [29] [3710/7269] eta: 0:25:37 lr: 0.000003 loss: 0.5326 (0.5146) time: 0.4777 data: 0.0002 max mem: 52150 +Epoch: [29] [3720/7269] eta: 0:25:33 lr: 0.000003 loss: 0.5474 (0.5147) time: 0.4915 data: 0.0002 max mem: 52150 +Epoch: [29] [3730/7269] eta: 0:25:28 lr: 0.000003 loss: 0.5299 (0.5147) time: 0.4325 data: 0.0002 max mem: 52150 +Epoch: [29] [3740/7269] eta: 0:25:23 lr: 0.000003 loss: 0.5078 (0.5147) time: 0.3725 data: 0.0002 max mem: 52150 +Epoch: [29] [3750/7269] eta: 0:25:18 lr: 0.000003 loss: 0.5239 (0.5147) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [29] [3760/7269] eta: 0:25:13 lr: 0.000003 loss: 0.5473 (0.5147) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [29] [3770/7269] eta: 0:25:08 lr: 0.000003 loss: 0.5191 (0.5147) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [29] [3780/7269] eta: 0:25:04 lr: 0.000003 loss: 0.5074 (0.5147) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [29] [3790/7269] eta: 0:25:00 lr: 0.000003 loss: 0.5298 (0.5148) time: 0.4321 data: 0.0002 max mem: 52150 +Epoch: [29] [3800/7269] eta: 0:24:56 lr: 0.000003 loss: 0.5298 (0.5148) time: 0.4778 data: 0.0002 max mem: 52150 +Epoch: [29] [3810/7269] eta: 0:24:52 lr: 0.000003 loss: 0.5120 (0.5148) time: 0.4794 data: 0.0002 max mem: 52150 +Epoch: [29] [3820/7269] eta: 0:24:47 lr: 0.000003 loss: 0.5340 (0.5148) time: 0.4336 data: 0.0002 max mem: 52150 +Epoch: [29] [3830/7269] eta: 0:24:42 lr: 0.000003 loss: 0.5175 (0.5147) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [29] [3840/7269] eta: 0:24:38 lr: 0.000003 loss: 0.5187 (0.5147) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [29] [3850/7269] eta: 0:24:34 lr: 0.000003 loss: 0.5390 (0.5148) time: 0.4202 data: 0.0002 max mem: 52150 +Epoch: [29] [3860/7269] eta: 0:24:29 lr: 0.000003 loss: 0.5167 (0.5148) time: 0.4201 data: 0.0002 max mem: 52150 +Epoch: [29] [3870/7269] eta: 0:24:24 lr: 0.000003 loss: 0.4912 (0.5147) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [29] [3880/7269] eta: 0:24:19 lr: 0.000003 loss: 0.4815 (0.5147) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [29] [3890/7269] eta: 0:24:14 lr: 0.000003 loss: 0.4939 (0.5147) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [29] [3900/7269] eta: 0:24:10 lr: 0.000003 loss: 0.5119 (0.5146) time: 0.4174 data: 0.0002 max mem: 52150 +Epoch: [29] [3910/7269] eta: 0:24:08 lr: 0.000003 loss: 0.5475 (0.5147) time: 0.5375 data: 0.0002 max mem: 52150 +Epoch: [29] [3920/7269] eta: 0:24:03 lr: 0.000003 loss: 0.5542 (0.5146) time: 0.4926 data: 0.0002 max mem: 52150 +Epoch: [29] [3930/7269] eta: 0:23:59 lr: 0.000003 loss: 0.4730 (0.5146) time: 0.4354 data: 0.0002 max mem: 52150 +Epoch: [29] [3940/7269] eta: 0:23:54 lr: 0.000003 loss: 0.5187 (0.5146) time: 0.4347 data: 0.0002 max mem: 52150 +Epoch: [29] [3950/7269] eta: 0:23:50 lr: 0.000003 loss: 0.5344 (0.5147) time: 0.4201 data: 0.0002 max mem: 52150 +Epoch: [29] [3960/7269] eta: 0:23:46 lr: 0.000003 loss: 0.5217 (0.5147) time: 0.4644 data: 0.0002 max mem: 52150 +Epoch: [29] [3970/7269] eta: 0:23:41 lr: 0.000003 loss: 0.5008 (0.5147) time: 0.4164 data: 0.0002 max mem: 52150 +Epoch: [29] [3980/7269] eta: 0:23:38 lr: 0.000003 loss: 0.5155 (0.5147) time: 0.4353 data: 0.0002 max mem: 52150 +Epoch: [29] [3990/7269] eta: 0:23:33 lr: 0.000003 loss: 0.5444 (0.5148) time: 0.4349 data: 0.0002 max mem: 52150 +Epoch: [29] [4000/7269] eta: 0:23:29 lr: 0.000003 loss: 0.5220 (0.5147) time: 0.4343 data: 0.0002 max mem: 52150 +Epoch: [29] [4010/7269] eta: 0:23:24 lr: 0.000003 loss: 0.5183 (0.5148) time: 0.4342 data: 0.0002 max mem: 52150 +Epoch: [29] [4020/7269] eta: 0:23:19 lr: 0.000003 loss: 0.5240 (0.5148) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [29] [4030/7269] eta: 0:23:16 lr: 0.000003 loss: 0.5240 (0.5148) time: 0.4798 data: 0.0002 max mem: 52150 +Epoch: [29] [4040/7269] eta: 0:23:12 lr: 0.000003 loss: 0.5177 (0.5148) time: 0.5285 data: 0.0002 max mem: 52150 +Epoch: [29] [4050/7269] eta: 0:23:08 lr: 0.000003 loss: 0.5432 (0.5148) time: 0.4672 data: 0.0002 max mem: 52150 +Epoch: [29] [4060/7269] eta: 0:23:03 lr: 0.000003 loss: 0.5548 (0.5149) time: 0.4180 data: 0.0002 max mem: 52150 +Epoch: [29] [4070/7269] eta: 0:22:59 lr: 0.000003 loss: 0.5261 (0.5149) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [29] [4080/7269] eta: 0:22:55 lr: 0.000003 loss: 0.5150 (0.5149) time: 0.4339 data: 0.0002 max mem: 52150 +Epoch: [29] [4090/7269] eta: 0:22:50 lr: 0.000003 loss: 0.5128 (0.5148) time: 0.4341 data: 0.0002 max mem: 52150 +Epoch: [29] [4100/7269] eta: 0:22:45 lr: 0.000003 loss: 0.5038 (0.5148) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [29] [4110/7269] eta: 0:22:41 lr: 0.000003 loss: 0.5431 (0.5149) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [29] [4120/7269] eta: 0:22:37 lr: 0.000003 loss: 0.5438 (0.5149) time: 0.4371 data: 0.0002 max mem: 52150 +Epoch: [29] [4130/7269] eta: 0:22:34 lr: 0.000003 loss: 0.5380 (0.5149) time: 0.5451 data: 0.0002 max mem: 52150 +Epoch: [29] [4140/7269] eta: 0:22:29 lr: 0.000003 loss: 0.5380 (0.5150) time: 0.4802 data: 0.0002 max mem: 52150 +Epoch: [29] [4150/7269] eta: 0:22:25 lr: 0.000003 loss: 0.5194 (0.5149) time: 0.4380 data: 0.0002 max mem: 52150 +Epoch: [29] [4160/7269] eta: 0:22:20 lr: 0.000003 loss: 0.5073 (0.5150) time: 0.4376 data: 0.0002 max mem: 52150 +Epoch: [29] [4170/7269] eta: 0:22:16 lr: 0.000003 loss: 0.5347 (0.5150) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [29] [4180/7269] eta: 0:22:13 lr: 0.000003 loss: 0.5293 (0.5150) time: 0.4948 data: 0.0002 max mem: 52150 +Epoch: [29] [4190/7269] eta: 0:22:09 lr: 0.000003 loss: 0.5059 (0.5150) time: 0.5561 data: 0.0002 max mem: 52150 +Epoch: [29] [4200/7269] eta: 0:22:04 lr: 0.000003 loss: 0.5180 (0.5150) time: 0.4329 data: 0.0002 max mem: 52150 +Epoch: [29] [4210/7269] eta: 0:21:59 lr: 0.000003 loss: 0.5190 (0.5150) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [29] [4220/7269] eta: 0:21:55 lr: 0.000003 loss: 0.5190 (0.5150) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [29] [4230/7269] eta: 0:21:50 lr: 0.000003 loss: 0.5228 (0.5150) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [29] [4240/7269] eta: 0:21:46 lr: 0.000003 loss: 0.5228 (0.5150) time: 0.4176 data: 0.0002 max mem: 52150 +Epoch: [29] [4250/7269] eta: 0:21:41 lr: 0.000003 loss: 0.5200 (0.5150) time: 0.4185 data: 0.0002 max mem: 52150 +Epoch: [29] [4260/7269] eta: 0:21:36 lr: 0.000003 loss: 0.5144 (0.5150) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [29] [4270/7269] eta: 0:21:33 lr: 0.000003 loss: 0.5168 (0.5149) time: 0.4655 data: 0.0002 max mem: 52150 +Epoch: [29] [4280/7269] eta: 0:21:28 lr: 0.000003 loss: 0.5168 (0.5150) time: 0.4656 data: 0.0002 max mem: 52150 +Epoch: [29] [4290/7269] eta: 0:21:24 lr: 0.000003 loss: 0.5367 (0.5149) time: 0.4350 data: 0.0002 max mem: 52150 +Epoch: [29] [4300/7269] eta: 0:21:20 lr: 0.000003 loss: 0.5367 (0.5149) time: 0.4351 data: 0.0002 max mem: 52150 +Epoch: [29] [4310/7269] eta: 0:21:16 lr: 0.000003 loss: 0.5113 (0.5149) time: 0.4192 data: 0.0002 max mem: 52150 +Epoch: [29] [4320/7269] eta: 0:21:11 lr: 0.000003 loss: 0.5080 (0.5149) time: 0.4188 data: 0.0002 max mem: 52150 +Epoch: [29] [4330/7269] eta: 0:21:09 lr: 0.000003 loss: 0.5339 (0.5149) time: 0.6077 data: 0.0002 max mem: 52150 +Epoch: [29] [4340/7269] eta: 0:21:05 lr: 0.000003 loss: 0.5395 (0.5150) time: 0.6080 data: 0.0001 max mem: 52150 +Epoch: [29] [4350/7269] eta: 0:21:00 lr: 0.000003 loss: 0.5327 (0.5149) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [29] [4360/7269] eta: 0:20:55 lr: 0.000003 loss: 0.5074 (0.5150) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [29] [4370/7269] eta: 0:20:50 lr: 0.000003 loss: 0.5451 (0.5149) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [29] [4380/7269] eta: 0:20:46 lr: 0.000003 loss: 0.4951 (0.5149) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [29] [4390/7269] eta: 0:20:42 lr: 0.000003 loss: 0.4951 (0.5148) time: 0.4191 data: 0.0002 max mem: 52150 +Epoch: [29] [4400/7269] eta: 0:20:38 lr: 0.000003 loss: 0.4978 (0.5148) time: 0.4822 data: 0.0002 max mem: 52150 +Epoch: [29] [4410/7269] eta: 0:20:33 lr: 0.000003 loss: 0.5111 (0.5148) time: 0.4347 data: 0.0002 max mem: 52150 +Epoch: [29] [4420/7269] eta: 0:20:29 lr: 0.000003 loss: 0.5374 (0.5148) time: 0.4178 data: 0.0002 max mem: 52150 +Epoch: [29] [4430/7269] eta: 0:20:25 lr: 0.000003 loss: 0.5357 (0.5148) time: 0.4823 data: 0.0002 max mem: 52150 +Epoch: [29] [4440/7269] eta: 0:20:20 lr: 0.000003 loss: 0.4988 (0.5147) time: 0.4355 data: 0.0002 max mem: 52150 +Epoch: [29] [4450/7269] eta: 0:20:16 lr: 0.000003 loss: 0.4988 (0.5147) time: 0.4168 data: 0.0002 max mem: 52150 +Epoch: [29] [4460/7269] eta: 0:20:12 lr: 0.000003 loss: 0.5444 (0.5147) time: 0.4174 data: 0.0002 max mem: 52150 +Epoch: [29] [4470/7269] eta: 0:20:07 lr: 0.000003 loss: 0.5297 (0.5147) time: 0.3721 data: 0.0002 max mem: 52150 +Epoch: [29] [4480/7269] eta: 0:20:02 lr: 0.000003 loss: 0.5407 (0.5148) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [29] [4490/7269] eta: 0:19:58 lr: 0.000003 loss: 0.5625 (0.5149) time: 0.4344 data: 0.0002 max mem: 52150 +Epoch: [29] [4500/7269] eta: 0:19:54 lr: 0.000003 loss: 0.5496 (0.5150) time: 0.4341 data: 0.0002 max mem: 52150 +Epoch: [29] [4510/7269] eta: 0:19:49 lr: 0.000003 loss: 0.4862 (0.5149) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [29] [4520/7269] eta: 0:19:45 lr: 0.000003 loss: 0.4796 (0.5148) time: 0.4337 data: 0.0002 max mem: 52150 +Epoch: [29] [4530/7269] eta: 0:19:41 lr: 0.000003 loss: 0.5071 (0.5148) time: 0.4801 data: 0.0002 max mem: 52150 +Epoch: [29] [4540/7269] eta: 0:19:37 lr: 0.000003 loss: 0.5235 (0.5148) time: 0.4775 data: 0.0002 max mem: 52150 +Epoch: [29] [4550/7269] eta: 0:19:33 lr: 0.000003 loss: 0.5008 (0.5147) time: 0.4904 data: 0.0002 max mem: 52150 +Epoch: [29] [4560/7269] eta: 0:19:28 lr: 0.000003 loss: 0.5283 (0.5148) time: 0.4305 data: 0.0002 max mem: 52150 +Epoch: [29] [4570/7269] eta: 0:19:24 lr: 0.000003 loss: 0.5179 (0.5148) time: 0.3710 data: 0.0002 max mem: 52150 +Epoch: [29] [4580/7269] eta: 0:19:20 lr: 0.000003 loss: 0.4869 (0.5147) time: 0.4205 data: 0.0002 max mem: 52150 +Epoch: [29] [4590/7269] eta: 0:19:15 lr: 0.000003 loss: 0.5058 (0.5147) time: 0.4205 data: 0.0002 max mem: 52150 +Epoch: [29] [4600/7269] eta: 0:19:11 lr: 0.000003 loss: 0.5267 (0.5148) time: 0.4163 data: 0.0002 max mem: 52150 +Epoch: [29] [4610/7269] eta: 0:19:06 lr: 0.000003 loss: 0.5405 (0.5148) time: 0.4166 data: 0.0002 max mem: 52150 +Epoch: [29] [4620/7269] eta: 0:19:02 lr: 0.000003 loss: 0.5333 (0.5148) time: 0.4193 data: 0.0002 max mem: 52150 +Epoch: [29] [4630/7269] eta: 0:18:57 lr: 0.000003 loss: 0.5231 (0.5149) time: 0.4194 data: 0.0002 max mem: 52150 +Epoch: [29] [4640/7269] eta: 0:18:53 lr: 0.000003 loss: 0.5257 (0.5149) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [29] [4650/7269] eta: 0:18:49 lr: 0.000003 loss: 0.5402 (0.5150) time: 0.4332 data: 0.0002 max mem: 52150 +Epoch: [29] [4660/7269] eta: 0:18:45 lr: 0.000003 loss: 0.4838 (0.5149) time: 0.4965 data: 0.0002 max mem: 52150 +Epoch: [29] [4670/7269] eta: 0:18:40 lr: 0.000003 loss: 0.4963 (0.5149) time: 0.4343 data: 0.0002 max mem: 52150 +Epoch: [29] [4680/7269] eta: 0:18:36 lr: 0.000003 loss: 0.5345 (0.5149) time: 0.4197 data: 0.0002 max mem: 52150 +Epoch: [29] [4690/7269] eta: 0:18:32 lr: 0.000003 loss: 0.5201 (0.5149) time: 0.4805 data: 0.0002 max mem: 52150 +Epoch: [29] [4700/7269] eta: 0:18:28 lr: 0.000003 loss: 0.5201 (0.5150) time: 0.4957 data: 0.0002 max mem: 52150 +Epoch: [29] [4710/7269] eta: 0:18:24 lr: 0.000003 loss: 0.4810 (0.5148) time: 0.4993 data: 0.0002 max mem: 52150 +Epoch: [29] [4720/7269] eta: 0:18:21 lr: 0.000003 loss: 0.5174 (0.5149) time: 0.5318 data: 0.0002 max mem: 52150 +Epoch: [29] [4730/7269] eta: 0:18:16 lr: 0.000003 loss: 0.5219 (0.5148) time: 0.4686 data: 0.0002 max mem: 52150 +Epoch: [29] [4740/7269] eta: 0:18:12 lr: 0.000003 loss: 0.5391 (0.5149) time: 0.4207 data: 0.0002 max mem: 52150 +Epoch: [29] [4750/7269] eta: 0:18:07 lr: 0.000003 loss: 0.5338 (0.5148) time: 0.4198 data: 0.0002 max mem: 52150 +Epoch: [29] [4760/7269] eta: 0:18:03 lr: 0.000003 loss: 0.5124 (0.5148) time: 0.3715 data: 0.0002 max mem: 52150 +Epoch: [29] [4770/7269] eta: 0:17:58 lr: 0.000003 loss: 0.5348 (0.5149) time: 0.3711 data: 0.0002 max mem: 52150 +Epoch: [29] [4780/7269] eta: 0:17:53 lr: 0.000003 loss: 0.5204 (0.5148) time: 0.3713 data: 0.0002 max mem: 52150 +Epoch: [29] [4790/7269] eta: 0:17:49 lr: 0.000003 loss: 0.5023 (0.5148) time: 0.4188 data: 0.0002 max mem: 52150 +Epoch: [29] [4800/7269] eta: 0:17:46 lr: 0.000003 loss: 0.5203 (0.5147) time: 0.5249 data: 0.0002 max mem: 52150 +Epoch: [29] [4810/7269] eta: 0:17:41 lr: 0.000003 loss: 0.4989 (0.5147) time: 0.4778 data: 0.0002 max mem: 52150 +Epoch: [29] [4820/7269] eta: 0:17:37 lr: 0.000003 loss: 0.5234 (0.5147) time: 0.4349 data: 0.0002 max mem: 52150 +Epoch: [29] [4830/7269] eta: 0:17:32 lr: 0.000003 loss: 0.5104 (0.5147) time: 0.4350 data: 0.0002 max mem: 52150 +Epoch: [29] [4840/7269] eta: 0:17:28 lr: 0.000003 loss: 0.5188 (0.5147) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [29] [4850/7269] eta: 0:17:23 lr: 0.000003 loss: 0.5267 (0.5147) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [29] [4860/7269] eta: 0:17:19 lr: 0.000003 loss: 0.5340 (0.5148) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [29] [4870/7269] eta: 0:17:14 lr: 0.000003 loss: 0.5409 (0.5148) time: 0.4209 data: 0.0002 max mem: 52150 +Epoch: [29] [4880/7269] eta: 0:17:10 lr: 0.000003 loss: 0.5070 (0.5148) time: 0.4208 data: 0.0002 max mem: 52150 +Epoch: [29] [4890/7269] eta: 0:17:05 lr: 0.000003 loss: 0.4881 (0.5148) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [29] [4900/7269] eta: 0:17:01 lr: 0.000003 loss: 0.5394 (0.5148) time: 0.4329 data: 0.0002 max mem: 52150 +Epoch: [29] [4910/7269] eta: 0:16:57 lr: 0.000003 loss: 0.5447 (0.5148) time: 0.4934 data: 0.0002 max mem: 52150 +Epoch: [29] [4920/7269] eta: 0:16:53 lr: 0.000003 loss: 0.5223 (0.5148) time: 0.4319 data: 0.0002 max mem: 52150 +Epoch: [29] [4930/7269] eta: 0:16:48 lr: 0.000003 loss: 0.4975 (0.5147) time: 0.4172 data: 0.0002 max mem: 52150 +Epoch: [29] [4940/7269] eta: 0:16:44 lr: 0.000003 loss: 0.5256 (0.5148) time: 0.4176 data: 0.0002 max mem: 52150 +Epoch: [29] [4950/7269] eta: 0:16:40 lr: 0.000003 loss: 0.5385 (0.5148) time: 0.4673 data: 0.0002 max mem: 52150 +Epoch: [29] [4960/7269] eta: 0:16:36 lr: 0.000003 loss: 0.5326 (0.5147) time: 0.5289 data: 0.0002 max mem: 52150 +Epoch: [29] [4970/7269] eta: 0:16:32 lr: 0.000003 loss: 0.5326 (0.5147) time: 0.4951 data: 0.0002 max mem: 52150 +Epoch: [29] [4980/7269] eta: 0:16:28 lr: 0.000003 loss: 0.5147 (0.5148) time: 0.4845 data: 0.0002 max mem: 52150 +Epoch: [29] [4990/7269] eta: 0:16:23 lr: 0.000003 loss: 0.5147 (0.5148) time: 0.4232 data: 0.0002 max mem: 52150 +Epoch: [29] [5000/7269] eta: 0:16:19 lr: 0.000003 loss: 0.5089 (0.5147) time: 0.3720 data: 0.0002 max mem: 52150 +Epoch: [29] [5010/7269] eta: 0:16:15 lr: 0.000003 loss: 0.5089 (0.5147) time: 0.4303 data: 0.0002 max mem: 52150 +Epoch: [29] [5020/7269] eta: 0:16:11 lr: 0.000003 loss: 0.5371 (0.5147) time: 0.4762 data: 0.0002 max mem: 52150 +Epoch: [29] [5030/7269] eta: 0:16:06 lr: 0.000003 loss: 0.5122 (0.5146) time: 0.4639 data: 0.0002 max mem: 52150 +Epoch: [29] [5040/7269] eta: 0:16:02 lr: 0.000003 loss: 0.5134 (0.5146) time: 0.4182 data: 0.0002 max mem: 52150 +Epoch: [29] [5050/7269] eta: 0:15:58 lr: 0.000003 loss: 0.5304 (0.5146) time: 0.4187 data: 0.0002 max mem: 52150 +Epoch: [29] [5060/7269] eta: 0:15:54 lr: 0.000003 loss: 0.5198 (0.5146) time: 0.4821 data: 0.0002 max mem: 52150 +Epoch: [29] [5070/7269] eta: 0:15:49 lr: 0.000003 loss: 0.5281 (0.5147) time: 0.4356 data: 0.0002 max mem: 52150 +Epoch: [29] [5080/7269] eta: 0:15:45 lr: 0.000003 loss: 0.5088 (0.5147) time: 0.4362 data: 0.0002 max mem: 52150 +Epoch: [29] [5090/7269] eta: 0:15:41 lr: 0.000003 loss: 0.5041 (0.5147) time: 0.4955 data: 0.0002 max mem: 52150 +Epoch: [29] [5100/7269] eta: 0:15:36 lr: 0.000003 loss: 0.5081 (0.5146) time: 0.4311 data: 0.0002 max mem: 52150 +Epoch: [29] [5110/7269] eta: 0:15:32 lr: 0.000003 loss: 0.5252 (0.5146) time: 0.4340 data: 0.0002 max mem: 52150 +Epoch: [29] [5120/7269] eta: 0:15:28 lr: 0.000003 loss: 0.5143 (0.5146) time: 0.4976 data: 0.0002 max mem: 52150 +Epoch: [29] [5130/7269] eta: 0:15:24 lr: 0.000003 loss: 0.5069 (0.5145) time: 0.4353 data: 0.0002 max mem: 52150 +Epoch: [29] [5140/7269] eta: 0:15:20 lr: 0.000003 loss: 0.5247 (0.5146) time: 0.4205 data: 0.0002 max mem: 52150 +Epoch: [29] [5150/7269] eta: 0:15:15 lr: 0.000003 loss: 0.5479 (0.5147) time: 0.4203 data: 0.0002 max mem: 52150 +Epoch: [29] [5160/7269] eta: 0:15:11 lr: 0.000003 loss: 0.5364 (0.5146) time: 0.4310 data: 0.0002 max mem: 52150 +Epoch: [29] [5170/7269] eta: 0:15:06 lr: 0.000003 loss: 0.4779 (0.5146) time: 0.4312 data: 0.0002 max mem: 52150 +Epoch: [29] [5180/7269] eta: 0:15:02 lr: 0.000003 loss: 0.4830 (0.5145) time: 0.4640 data: 0.0002 max mem: 52150 +Epoch: [29] [5190/7269] eta: 0:14:58 lr: 0.000003 loss: 0.5451 (0.5146) time: 0.4641 data: 0.0002 max mem: 52150 +Epoch: [29] [5200/7269] eta: 0:14:53 lr: 0.000003 loss: 0.5216 (0.5146) time: 0.3724 data: 0.0002 max mem: 52150 +Epoch: [29] [5210/7269] eta: 0:14:49 lr: 0.000003 loss: 0.5265 (0.5147) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [29] [5220/7269] eta: 0:14:44 lr: 0.000003 loss: 0.5265 (0.5146) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [29] [5230/7269] eta: 0:14:40 lr: 0.000003 loss: 0.4811 (0.5146) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [29] [5240/7269] eta: 0:14:36 lr: 0.000003 loss: 0.4739 (0.5146) time: 0.4173 data: 0.0002 max mem: 52150 +Epoch: [29] [5250/7269] eta: 0:14:31 lr: 0.000003 loss: 0.5377 (0.5146) time: 0.4172 data: 0.0002 max mem: 52150 +Epoch: [29] [5260/7269] eta: 0:14:27 lr: 0.000003 loss: 0.5609 (0.5147) time: 0.4764 data: 0.0002 max mem: 52150 +Epoch: [29] [5270/7269] eta: 0:14:23 lr: 0.000003 loss: 0.5590 (0.5147) time: 0.4763 data: 0.0002 max mem: 52150 +Epoch: [29] [5280/7269] eta: 0:14:19 lr: 0.000003 loss: 0.5011 (0.5147) time: 0.4368 data: 0.0002 max mem: 52150 +Epoch: [29] [5290/7269] eta: 0:14:15 lr: 0.000003 loss: 0.5202 (0.5146) time: 0.4992 data: 0.0002 max mem: 52150 +Epoch: [29] [5300/7269] eta: 0:14:10 lr: 0.000003 loss: 0.5204 (0.5147) time: 0.4337 data: 0.0002 max mem: 52150 +Epoch: [29] [5310/7269] eta: 0:14:05 lr: 0.000003 loss: 0.5470 (0.5147) time: 0.3717 data: 0.0002 max mem: 52150 +Epoch: [29] [5320/7269] eta: 0:14:01 lr: 0.000003 loss: 0.5658 (0.5148) time: 0.3722 data: 0.0002 max mem: 52150 +Epoch: [29] [5330/7269] eta: 0:13:56 lr: 0.000003 loss: 0.5436 (0.5148) time: 0.3719 data: 0.0002 max mem: 52150 +Epoch: [29] [5340/7269] eta: 0:13:52 lr: 0.000003 loss: 0.5239 (0.5148) time: 0.4187 data: 0.0002 max mem: 52150 +Epoch: [29] [5350/7269] eta: 0:13:48 lr: 0.000003 loss: 0.5032 (0.5147) time: 0.4677 data: 0.0002 max mem: 52150 +Epoch: [29] [5360/7269] eta: 0:13:43 lr: 0.000003 loss: 0.4920 (0.5147) time: 0.4207 data: 0.0002 max mem: 52150 +Epoch: [29] [5370/7269] eta: 0:13:40 lr: 0.000003 loss: 0.4888 (0.5147) time: 0.4822 data: 0.0002 max mem: 52150 +Epoch: [29] [5380/7269] eta: 0:13:36 lr: 0.000003 loss: 0.4791 (0.5146) time: 0.5872 data: 0.0002 max mem: 52150 +Epoch: [29] [5390/7269] eta: 0:13:32 lr: 0.000003 loss: 0.4881 (0.5146) time: 0.5419 data: 0.0002 max mem: 52150 +Epoch: [29] [5400/7269] eta: 0:13:27 lr: 0.000003 loss: 0.5181 (0.5146) time: 0.4371 data: 0.0002 max mem: 52150 +Epoch: [29] [5410/7269] eta: 0:13:23 lr: 0.000003 loss: 0.5113 (0.5146) time: 0.3716 data: 0.0002 max mem: 52150 +Epoch: [29] [5420/7269] eta: 0:13:19 lr: 0.000003 loss: 0.4916 (0.5145) time: 0.4330 data: 0.0002 max mem: 52150 +Epoch: [29] [5430/7269] eta: 0:13:15 lr: 0.000003 loss: 0.4914 (0.5144) time: 0.4966 data: 0.0002 max mem: 52150 +Epoch: [29] [5440/7269] eta: 0:13:10 lr: 0.000003 loss: 0.4831 (0.5144) time: 0.4831 data: 0.0002 max mem: 52150 +Epoch: [29] [5450/7269] eta: 0:13:06 lr: 0.000003 loss: 0.4831 (0.5143) time: 0.4195 data: 0.0002 max mem: 52150 +Epoch: [29] [5460/7269] eta: 0:13:01 lr: 0.000003 loss: 0.5327 (0.5144) time: 0.3714 data: 0.0002 max mem: 52150 +Epoch: [29] [5470/7269] eta: 0:12:57 lr: 0.000003 loss: 0.5370 (0.5144) time: 0.3718 data: 0.0002 max mem: 52150 +Epoch: [29] [5480/7269] eta: 0:12:52 lr: 0.000003 loss: 0.5389 (0.5145) time: 0.3727 data: 0.0002 max mem: 52150 +Epoch: [29] [5490/7269] eta: 0:12:48 lr: 0.000003 loss: 0.5575 (0.5145) time: 0.4639 data: 0.0002 max mem: 52150 +Epoch: [29] [5500/7269] eta: 0:12:44 lr: 0.000003 loss: 0.5465 (0.5145) time: 0.4629 data: 0.0002 max mem: 52150 +Epoch: [29] [5510/7269] eta: 0:12:40 lr: 0.000003 loss: 0.5051 (0.5145) time: 0.4168 data: 0.0002 max mem: 52150 +Epoch: [29] [5520/7269] eta: 0:12:36 lr: 0.000003 loss: 0.5051 (0.5145) time: 0.4769 data: 0.0002 max mem: 52150 +E0727 05:39:59.262000 1915 torch/distributed/elastic/multiprocessing/api.py:874] failed (exitcode: -15) local_rank: 0 (pid: 1981) of binary: /usr/bin/python3 +Traceback (most recent call last): + File "/usr/local/bin/torchrun", line 7, in + sys.exit(main()) + ^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 355, in wrapper + return f(*args, **kwargs) + ^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 892, in main + run(args) + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 883, in run + elastic_launch( + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 139, in __call__ + return launch_agent(self._config, self._entrypoint, list(args)) + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 270, in launch_agent + raise ChildFailedError( +torch.distributed.elastic.multiprocessing.errors.ChildFailedError: +============================================================ +main_cross_2direct_balance_realscale_perturb_v4_accum.py FAILED +------------------------------------------------------------ +Failures: +[1]: + time : 2026-07-27_05:39:59 + host : n116-042-185.byted.org + rank : 1 (local_rank: 1) + exitcode : -15 (pid: 1982) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 1982 +[2]: + time : 2026-07-27_05:39:59 + host : n116-042-185.byted.org + rank : 2 (local_rank: 2) + exitcode : -15 (pid: 1983) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 1983 +[3]: + time : 2026-07-27_05:39:59 + host : n116-042-185.byted.org + rank : 3 (local_rank: 3) + exitcode : -15 (pid: 1984) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 1984 +[4]: + time : 2026-07-27_05:39:59 + host : n116-042-185.byted.org + rank : 4 (local_rank: 4) + exitcode : -15 (pid: 1985) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 1985 +[5]: + time : 2026-07-27_05:39:59 + host : n116-042-185.byted.org + rank : 5 (local_rank: 5) + exitcode : -15 (pid: 1986) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 1986 +[6]: + time : 2026-07-27_05:39:59 + host : n116-042-185.byted.org + rank : 6 (local_rank: 6) + exitcode : -15 (pid: 1987) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 1987 +[7]: + time : 2026-07-27_05:39:59 + host : n116-042-185.byted.org + rank : 7 (local_rank: 7) + exitcode : -15 (pid: 1988) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 1988 +------------------------------------------------------------ +Root Cause (first observed failure): +[0]: + time : 2026-07-27_05:39:59 + host : n116-042-185.byted.org + rank : 0 (local_rank: 0) + exitcode : -15 (pid: 1981) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 1981 +============================================================ diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1603_vith32_launch_stub_master_n116-042-185_port25003/launcher_rank1_n111-046-078.log b/ScaleDF_rebuttal/_host_logs/20260726_1603_vith32_launch_stub_master_n116-042-185_port25003/launcher_rank1_n111-046-078.log new file mode 100644 index 0000000000000000000000000000000000000000..0186903dbc49f7fa46d90577dfd2e6c3f42593b3 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1603_vith32_launch_stub_master_n116-042-185_port25003/launcher_rank1_n111-046-078.log @@ -0,0 +1,1751 @@ +started_utc=2026-07-26T16:03:56Z +hostname=n111-046-078 +node_rank=1 +nnodes=4 +nproc_per_node=8 +world_size=32 +master_addr=10.116.42.185 +master_port=25003 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +main=main_cross_2direct_balance_realscale_perturb_v4_accum.py +model_name=deit_huge_patch14_LS +pretrain=/tmp/ScaleDF_repro_20260724/pretrained/deit_3_huge_224_21k_v1.pth +resume=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_24card_effbs3072_from_epoch7_n116-041-246_n116-042-186_n116-042-185_20260726_0736_vith24_from_epoch7/checkpoint.pth +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_32card_effbs2048_from_latest24_master_n116-042-185_20260726_1603_vith32_stub_port25003 +batch_per_gpu=64 +accum_iter=1 +effective_batch=2048 +eval_batch_per_gpu=64 +epochs=40 +W0726 16:04:10.208000 3248 torch/distributed/run.py:766] +W0726 16:04:10.208000 3248 torch/distributed/run.py:766] ***************************************** +W0726 16:04:10.208000 3248 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 16:04:10.208000 3248 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 8): env:// +[rank8]:[W726 16:04:42.770805197 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 8] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 15): env:// +[rank15]:[W726 16:04:43.095522859 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 15] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 14): env:// +[rank14]:[W726 16:04:43.547373002 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 14] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 9): env:// +[rank9]:[W726 16:04:43.582063027 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 9] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 13): env:// +[rank13]:[W726 16:04:43.702035413 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 13] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 11): env:// +| distributed init (rank 10): env:// +[rank10]:[W726 16:04:44.023779433 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 10] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank11]:[W726 16:04:44.023849220 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 11] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 12): env:// +[rank12]:[W726 16:04:44.104584431 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 12] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +n111-046-078:3320:3320 [7] NCCL INFO cudaDriverVersion 12060 +n111-046-078:3316:3316 [3] NCCL INFO cudaDriverVersion 12060 +n111-046-078:3320:3320 [7] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n111-046-078:3320:3320 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n111-046-078:3316:3316 [3] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n111-046-078:3316:3316 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n111-046-078:3316:3316 [3] NCCL INFO NCCL version 2.26.2+cuda12.2 +n111-046-078:3320:3320 [7] NCCL INFO NCCL version 2.26.2+cuda12.2 +n111-046-078:3316:3316 [3] NCCL INFO Comm config Blocking set to 1 +n111-046-078:3320:3320 [7] NCCL INFO Comm config Blocking set to 1 +n111-046-078:3319:3319 [6] NCCL INFO cudaDriverVersion 12060 +n111-046-078:3319:3319 [6] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n111-046-078:3319:3319 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n111-046-078:3319:3319 [6] NCCL INFO NCCL version 2.26.2+cuda12.2 +n111-046-078:3319:3319 [6] NCCL INFO Comm config Blocking set to 1 +n111-046-078:3313:3313 [0] NCCL INFO cudaDriverVersion 12060 +n111-046-078:3313:3313 [0] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n111-046-078:3313:3313 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n111-046-078:3313:3313 [0] NCCL INFO NCCL version 2.26.2+cuda12.2 +n111-046-078:3313:3313 [0] NCCL INFO Comm config Blocking set to 1 +n111-046-078:3314:3314 [1] NCCL INFO cudaDriverVersion 12060 +n111-046-078:3314:3314 [1] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n111-046-078:3314:3314 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n111-046-078:3314:3314 [1] NCCL INFO NCCL version 2.26.2+cuda12.2 +n111-046-078:3318:3318 [5] NCCL INFO cudaDriverVersion 12060 +n111-046-078:3318:3318 [5] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n111-046-078:3318:3318 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n111-046-078:3318:3318 [5] NCCL INFO NCCL version 2.26.2+cuda12.2 +n111-046-078:3317:3317 [4] NCCL INFO cudaDriverVersion 12060 +n111-046-078:3317:3317 [4] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n111-046-078:3317:3317 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n111-046-078:3317:3317 [4] NCCL INFO NCCL version 2.26.2+cuda12.2 +n111-046-078:3315:3315 [2] NCCL INFO cudaDriverVersion 12060 +n111-046-078:3315:3315 [2] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n111-046-078:3315:3315 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n111-046-078:3314:3314 [1] NCCL INFO Comm config Blocking set to 1 +n111-046-078:3318:3318 [5] NCCL INFO Comm config Blocking set to 1 +n111-046-078:3317:3317 [4] NCCL INFO Comm config Blocking set to 1 +n111-046-078:3315:3315 [2] NCCL INFO NCCL version 2.26.2+cuda12.2 +n111-046-078:3315:3315 [2] NCCL INFO Comm config Blocking set to 1 +n111-046-078:3316:3369 [3] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n111-046-078:3317:3375 [4] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n111-046-078:3315:3376 [2] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n111-046-078:3314:3373 [1] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n111-046-078:3319:3371 [6] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n111-046-078:3317:3375 [4] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n111-046-078:3317:3375 [4] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n111-046-078:3315:3376 [2] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n111-046-078:3316:3369 [3] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n111-046-078:3314:3373 [1] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n111-046-078:3317:3375 [4] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n111-046-078:3315:3376 [2] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n111-046-078:3319:3371 [6] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n111-046-078:3316:3369 [3] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n111-046-078:3315:3376 [2] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n111-046-078:3314:3373 [1] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n111-046-078:3316:3369 [3] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n111-046-078:3319:3371 [6] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n111-046-078:3314:3373 [1] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n111-046-078:3319:3371 [6] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n111-046-078:3315:3376 [2] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n111-046-078:3316:3369 [3] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n111-046-078:3319:3371 [6] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n111-046-078:3314:3373 [1] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n111-046-078:3315:3376 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n111-046-078:3316:3369 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n111-046-078:3317:3375 [4] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n111-046-078:3319:3371 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n111-046-078:3314:3373 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n111-046-078:3316:3369 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n111-046-078:3315:3376 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n111-046-078:3317:3375 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n111-046-078:3319:3371 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n111-046-078:3314:3373 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n111-046-078:3316:3369 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n111-046-078:3315:3376 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n111-046-078:3317:3375 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n111-046-078:3319:3371 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n111-046-078:3316:3369 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n111-046-078:3315:3376 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n111-046-078:3314:3373 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n111-046-078:3319:3371 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n111-046-078:3317:3375 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n111-046-078:3316:3369 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n111-046-078:3315:3376 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n111-046-078:3314:3373 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n111-046-078:3319:3371 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n111-046-078:3317:3375 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n111-046-078:3314:3373 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n111-046-078:3317:3375 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n111-046-078:3313:3372 [0] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n111-046-078:3320:3370 [7] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n111-046-078:3318:3374 [5] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n111-046-078:3313:3372 [0] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n111-046-078:3313:3372 [0] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n111-046-078:3313:3372 [0] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n111-046-078:3318:3374 [5] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n111-046-078:3318:3374 [5] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n111-046-078:3318:3374 [5] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n111-046-078:3320:3370 [7] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n111-046-078:3313:3372 [0] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n111-046-078:3320:3370 [7] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n111-046-078:3313:3372 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n111-046-078:3320:3370 [7] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n111-046-078:3313:3372 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n111-046-078:3313:3372 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n111-046-078:3313:3372 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n111-046-078:3318:3374 [5] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n111-046-078:3313:3372 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n111-046-078:3318:3374 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n111-046-078:3318:3374 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n111-046-078:3320:3370 [7] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n111-046-078:3318:3374 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n111-046-078:3320:3370 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n111-046-078:3318:3374 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n111-046-078:3320:3370 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n111-046-078:3318:3374 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n111-046-078:3320:3370 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n111-046-078:3320:3370 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n111-046-078:3320:3370 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n111-046-078:3314:3373 [1] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n111-046-078:3314:3373 [1] NCCL INFO Using network FasTrak +n111-046-078:3319:3371 [6] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n111-046-078:3319:3371 [6] NCCL INFO Using network FasTrak +n111-046-078:3317:3375 [4] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n111-046-078:3317:3375 [4] NCCL INFO Using network FasTrak +n111-046-078:3313:3372 [0] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n111-046-078:3313:3372 [0] NCCL INFO Using network FasTrak +n111-046-078:3316:3369 [3] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n111-046-078:3316:3369 [3] NCCL INFO Using network FasTrak +n111-046-078:3318:3374 [5] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n111-046-078:3318:3374 [5] NCCL INFO Using network FasTrak +n111-046-078:3320:3370 [7] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n111-046-078:3320:3370 [7] NCCL INFO Using network FasTrak +n111-046-078:3315:3376 [2] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n111-046-078:3315:3376 [2] NCCL INFO Using network FasTrak +n111-046-078:3314:3373 [1] NCCL INFO DMA-BUF is available on GPU device 1 +n111-046-078:3319:3371 [6] NCCL INFO DMA-BUF is available on GPU device 6 +n111-046-078:3314:3373 [1] NCCL INFO ncclCommInitRankConfig comm 0xc823c10 rank 9 nranks 32 cudaDev 1 nvmlDev 1 busId 5000 commId 0x2a4d35ea441cbcba - Init START +n111-046-078:3319:3371 [6] NCCL INFO ncclCommInitRankConfig comm 0xb9c26e0 rank 14 nranks 32 cudaDev 6 nvmlDev 6 busId 8b000 commId 0x2a4d35ea441cbcba - Init START +n111-046-078:3317:3375 [4] NCCL INFO DMA-BUF is available on GPU device 4 +n111-046-078:3317:3375 [4] NCCL INFO ncclCommInitRankConfig comm 0xc1a21c0 rank 12 nranks 32 cudaDev 4 nvmlDev 4 busId 84000 commId 0x2a4d35ea441cbcba - Init START +n111-046-078:3318:3374 [5] NCCL INFO DMA-BUF is available on GPU device 5 +n111-046-078:3320:3370 [7] NCCL INFO DMA-BUF is available on GPU device 7 +n111-046-078:3316:3369 [3] NCCL INFO DMA-BUF is available on GPU device 3 +n111-046-078:3313:3372 [0] NCCL INFO DMA-BUF is available on GPU device 0 +n111-046-078:3315:3376 [2] NCCL INFO DMA-BUF is available on GPU device 2 +n111-046-078:3318:3374 [5] NCCL INFO ncclCommInitRankConfig comm 0xb26c190 rank 13 nranks 32 cudaDev 5 nvmlDev 5 busId 85000 commId 0x2a4d35ea441cbcba - Init START +n111-046-078:3320:3370 [7] NCCL INFO ncclCommInitRankConfig comm 0xbdf0f50 rank 15 nranks 32 cudaDev 7 nvmlDev 7 busId 8c000 commId 0x2a4d35ea441cbcba - Init START +n111-046-078:3316:3369 [3] NCCL INFO ncclCommInitRankConfig comm 0xcc63a70 rank 11 nranks 32 cudaDev 3 nvmlDev 3 busId c000 commId 0x2a4d35ea441cbcba - Init START +n111-046-078:3313:3372 [0] NCCL INFO ncclCommInitRankConfig comm 0xba468c0 rank 8 nranks 32 cudaDev 0 nvmlDev 0 busId 4000 commId 0x2a4d35ea441cbcba - Init START +n111-046-078:3315:3376 [2] NCCL INFO ncclCommInitRankConfig comm 0xb4b5590 rank 10 nranks 32 cudaDev 2 nvmlDev 2 busId b000 commId 0x2a4d35ea441cbcba - Init START +n111-046-078:3318:3374 [5] NCCL INFO RAS client listening socket at ::1<28028> +n111-046-078:3317:3375 [4] NCCL INFO RAS client listening socket at ::1<28028> +n111-046-078:3319:3371 [6] NCCL INFO RAS client listening socket at ::1<28028> +n111-046-078:3320:3370 [7] NCCL INFO RAS client listening socket at ::1<28028> +n111-046-078:3315:3376 [2] NCCL INFO RAS client listening socket at ::1<28028> +n111-046-078:3314:3373 [1] NCCL INFO RAS client listening socket at ::1<28028> +n111-046-078:3316:3369 [3] NCCL INFO RAS client listening socket at ::1<28028> +n111-046-078:3313:3372 [0] NCCL INFO RAS client listening socket at ::1<28028> +n111-046-078:3320:3370 [7] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n111-046-078:3320:3370 [7] NCCL INFO TOPO/NET : Importing network plugins to topology +n111-046-078:3320:3370 [7] NCCL INFO Retrieving state for FasTrak +n111-046-078:3320:3370 [7] NCCL INFO Initialized state 0 for FasTrak +n111-046-078:3320:3370 [7] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n111-046-078:3320:3370 [7] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n111-046-078:3320:3370 [7] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n111-046-078:3319:3371 [6] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n111-046-078:3320:3370 [7] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n111-046-078:3319:3371 [6] NCCL INFO TOPO/NET : Importing network plugins to topology +n111-046-078:3319:3371 [6] NCCL INFO Retrieving state for FasTrak +n111-046-078:3319:3371 [6] NCCL INFO Initialized state 0 for FasTrak +n111-046-078:3320:3370 [7] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n111-046-078:3319:3371 [6] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n111-046-078:3320:3370 [7] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n111-046-078:3319:3371 [6] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n111-046-078:3320:3370 [7] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n111-046-078:3320:3370 [7] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n111-046-078:3319:3371 [6] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n111-046-078:3319:3371 [6] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n111-046-078:3319:3371 [6] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n111-046-078:3319:3371 [6] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n111-046-078:3319:3371 [6] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n111-046-078:3319:3371 [6] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n111-046-078:3316:3369 [3] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n111-046-078:3316:3369 [3] NCCL INFO TOPO/NET : Importing network plugins to topology +n111-046-078:3316:3369 [3] NCCL INFO Retrieving state for FasTrak +n111-046-078:3316:3369 [3] NCCL INFO Initialized state 0 for FasTrak +n111-046-078:3316:3369 [3] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n111-046-078:3316:3369 [3] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n111-046-078:3316:3369 [3] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n111-046-078:3316:3369 [3] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n111-046-078:3317:3375 [4] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n111-046-078:3317:3375 [4] NCCL INFO TOPO/NET : Importing network plugins to topology +n111-046-078:3317:3375 [4] NCCL INFO Retrieving state for FasTrak +n111-046-078:3317:3375 [4] NCCL INFO Initialized state 0 for FasTrak +n111-046-078:3316:3369 [3] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n111-046-078:3316:3369 [3] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n111-046-078:3317:3375 [4] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n111-046-078:3316:3369 [3] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n111-046-078:3317:3375 [4] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n111-046-078:3316:3369 [3] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n111-046-078:3317:3375 [4] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n111-046-078:3317:3375 [4] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n111-046-078:3318:3374 [5] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n111-046-078:3318:3374 [5] NCCL INFO TOPO/NET : Importing network plugins to topology +n111-046-078:3317:3375 [4] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n111-046-078:3318:3374 [5] NCCL INFO Retrieving state for FasTrak +n111-046-078:3318:3374 [5] NCCL INFO Initialized state 0 for FasTrak +n111-046-078:3317:3375 [4] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n111-046-078:3318:3374 [5] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n111-046-078:3317:3375 [4] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n111-046-078:3318:3374 [5] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n111-046-078:3317:3375 [4] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n111-046-078:3318:3374 [5] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n111-046-078:3315:3376 [2] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n111-046-078:3315:3376 [2] NCCL INFO TOPO/NET : Importing network plugins to topology +n111-046-078:3318:3374 [5] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n111-046-078:3315:3376 [2] NCCL INFO Retrieving state for FasTrak +n111-046-078:3315:3376 [2] NCCL INFO Initialized state 0 for FasTrak +n111-046-078:3318:3374 [5] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n111-046-078:3318:3374 [5] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n111-046-078:3315:3376 [2] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n111-046-078:3315:3376 [2] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n111-046-078:3318:3374 [5] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n111-046-078:3315:3376 [2] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n111-046-078:3318:3374 [5] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n111-046-078:3315:3376 [2] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n111-046-078:3314:3373 [1] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n111-046-078:3314:3373 [1] NCCL INFO TOPO/NET : Importing network plugins to topology +n111-046-078:3314:3373 [1] NCCL INFO Retrieving state for FasTrak +n111-046-078:3314:3373 [1] NCCL INFO Initialized state 0 for FasTrak +n111-046-078:3315:3376 [2] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n111-046-078:3314:3373 [1] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n111-046-078:3315:3376 [2] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n111-046-078:3314:3373 [1] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n111-046-078:3315:3376 [2] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n111-046-078:3315:3376 [2] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n111-046-078:3314:3373 [1] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n111-046-078:3314:3373 [1] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n111-046-078:3314:3373 [1] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n111-046-078:3314:3373 [1] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n111-046-078:3314:3373 [1] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n111-046-078:3314:3373 [1] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n111-046-078:3313:3372 [0] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n111-046-078:3313:3372 [0] NCCL INFO TOPO/NET : Importing network plugins to topology +n111-046-078:3313:3372 [0] NCCL INFO Retrieving state for FasTrak +n111-046-078:3313:3372 [0] NCCL INFO Initialized state 0 for FasTrak +n111-046-078:3313:3372 [0] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n111-046-078:3313:3372 [0] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n111-046-078:3313:3372 [0] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n111-046-078:3313:3372 [0] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n111-046-078:3313:3372 [0] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n111-046-078:3313:3372 [0] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n111-046-078:3313:3372 [0] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n111-046-078:3313:3372 [0] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n111-046-078:3314:3373 [1] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n111-046-078:3315:3376 [2] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n111-046-078:3313:3372 [0] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n111-046-078:3320:3370 [7] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n111-046-078:3318:3374 [5] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n111-046-078:3319:3371 [6] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n111-046-078:3317:3375 [4] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n111-046-078:3316:3369 [3] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n111-046-078:3315:3376 [2] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n111-046-078:3315:3376 [2] NCCL INFO CPU/0-0 (1/1/3) +n111-046-078:3315:3376 [2] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n111-046-078:3315:3376 [2] NCCL INFO + PCI[24.0] - GPU/0-4000 (8) +n111-046-078:3315:3376 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3315:3376 [2] NCCL INFO + PCI[24.0] - NIC/0-6000 +n111-046-078:3315:3376 [2] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n111-046-078:3315:3376 [2] NCCL INFO + PCI[24.0] - NIC/0-7000 +n111-046-078:3315:3376 [2] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n111-046-078:3315:3376 [2] NCCL INFO + PCI[24.0] - GPU/0-5000 (9) +n111-046-078:3315:3376 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3315:3376 [2] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n111-046-078:3315:3376 [2] NCCL INFO + PCI[24.0] - NIC/0-d000 +n111-046-078:3315:3376 [2] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n111-046-078:3315:3376 [2] NCCL INFO + PCI[24.0] - NIC/0-e000 +n111-046-078:3315:3376 [2] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n111-046-078:3315:3376 [2] NCCL INFO + PCI[24.0] - GPU/0-b000 (10) +n111-046-078:3315:3376 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3315:3376 [2] NCCL INFO + PCI[24.0] - GPU/0-c000 (11) +n111-046-078:3315:3376 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3315:3376 [2] NCCL INFO + SYS[22.0] - CPU/0-1 +n111-046-078:3315:3376 [2] NCCL INFO CPU/0-1 (1/1/3) +n111-046-078:3315:3376 [2] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n111-046-078:3315:3376 [2] NCCL INFO + PCI[24.0] - NIC/0-86000 +n111-046-078:3315:3376 [2] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n111-046-078:3315:3376 [2] NCCL INFO + PCI[24.0] - NIC/0-87000 +n111-046-078:3315:3376 [2] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n111-046-078:3315:3376 [2] NCCL INFO + PCI[24.0] - GPU/0-84000 (12) +n111-046-078:3315:3376 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3315:3376 [2] NCCL INFO + PCI[24.0] - GPU/0-85000 (13) +n111-046-078:3315:3376 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3315:3376 [2] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n111-046-078:3315:3376 [2] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n111-046-078:3315:3376 [2] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n111-046-078:3315:3376 [2] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n111-046-078:3315:3376 [2] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n111-046-078:3315:3376 [2] NCCL INFO + PCI[24.0] - GPU/0-8b000 (14) +n111-046-078:3315:3376 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3315:3376 [2] NCCL INFO + PCI[24.0] - GPU/0-8c000 (15) +n111-046-078:3315:3376 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3315:3376 [2] NCCL INFO + SYS[22.0] - CPU/0-0 +n111-046-078:3315:3376 [2] NCCL INFO ========================================== +n111-046-078:3315:3376 [2] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-078:3315:3376 [2] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-078:3315:3376 [2] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-078:3318:3374 [5] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n111-046-078:3315:3376 [2] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-078:3318:3374 [5] NCCL INFO CPU/0-0 (1/1/3) +n111-046-078:3318:3374 [5] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n111-046-078:3318:3374 [5] NCCL INFO + PCI[24.0] - GPU/0-4000 (8) +n111-046-078:3315:3376 [2] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-078:3318:3374 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3318:3374 [5] NCCL INFO + PCI[24.0] - NIC/0-6000 +n111-046-078:3315:3376 [2] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-078:3318:3374 [5] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n111-046-078:3318:3374 [5] NCCL INFO + PCI[24.0] - NIC/0-7000 +n111-046-078:3318:3374 [5] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n111-046-078:3318:3374 [5] NCCL INFO + PCI[24.0] - GPU/0-5000 (9) +n111-046-078:3318:3374 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3318:3374 [5] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n111-046-078:3318:3374 [5] NCCL INFO + PCI[24.0] - NIC/0-d000 +n111-046-078:3318:3374 [5] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n111-046-078:3315:3376 [2] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n111-046-078:3318:3374 [5] NCCL INFO + PCI[24.0] - NIC/0-e000 +n111-046-078:3318:3374 [5] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n111-046-078:3318:3374 [5] NCCL INFO + PCI[24.0] - GPU/0-b000 (10) +n111-046-078:3318:3374 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3318:3374 [5] NCCL INFO + PCI[24.0] - GPU/0-c000 (11) +n111-046-078:3318:3374 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3315:3376 [2] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n111-046-078:3318:3374 [5] NCCL INFO + SYS[22.0] - CPU/0-1 +n111-046-078:3318:3374 [5] NCCL INFO CPU/0-1 (1/1/3) +n111-046-078:3318:3374 [5] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n111-046-078:3318:3374 [5] NCCL INFO + PCI[24.0] - NIC/0-86000 +n111-046-078:3318:3374 [5] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n111-046-078:3318:3374 [5] NCCL INFO + PCI[24.0] - NIC/0-87000 +n111-046-078:3318:3374 [5] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n111-046-078:3315:3376 [2] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-078:3318:3374 [5] NCCL INFO + PCI[24.0] - GPU/0-84000 (12) +n111-046-078:3318:3374 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3318:3374 [5] NCCL INFO + PCI[24.0] - GPU/0-85000 (13) +n111-046-078:3318:3374 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3318:3374 [5] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n111-046-078:3318:3374 [5] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n111-046-078:3318:3374 [5] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n111-046-078:3315:3376 [2] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-078:3318:3374 [5] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n111-046-078:3318:3374 [5] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n111-046-078:3318:3374 [5] NCCL INFO + PCI[24.0] - GPU/0-8b000 (14) +n111-046-078:3318:3374 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3318:3374 [5] NCCL INFO + PCI[24.0] - GPU/0-8c000 (15) +n111-046-078:3318:3374 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3318:3374 [5] NCCL INFO + SYS[22.0] - CPU/0-0 +n111-046-078:3315:3376 [2] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-078:3318:3374 [5] NCCL INFO ========================================== +n111-046-078:3315:3376 [2] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-078:3318:3374 [5] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-078:3317:3375 [4] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n111-046-078:3317:3375 [4] NCCL INFO CPU/0-0 (1/1/3) +n111-046-078:3318:3374 [5] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-078:3317:3375 [4] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n111-046-078:3313:3372 [0] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n111-046-078:3317:3375 [4] NCCL INFO + PCI[24.0] - GPU/0-4000 (8) +n111-046-078:3315:3376 [2] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n111-046-078:3313:3372 [0] NCCL INFO CPU/0-0 (1/1/3) +n111-046-078:3319:3371 [6] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n111-046-078:3317:3375 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3318:3374 [5] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-078:3313:3372 [0] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n111-046-078:3319:3371 [6] NCCL INFO CPU/0-0 (1/1/3) +n111-046-078:3317:3375 [4] NCCL INFO + PCI[24.0] - NIC/0-6000 +n111-046-078:3313:3372 [0] NCCL INFO + PCI[24.0] - GPU/0-4000 (8) +n111-046-078:3315:3376 [2] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n111-046-078:3317:3375 [4] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n111-046-078:3319:3371 [6] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n111-046-078:3313:3372 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3317:3375 [4] NCCL INFO + PCI[24.0] - NIC/0-7000 +n111-046-078:3319:3371 [6] NCCL INFO + PCI[24.0] - GPU/0-4000 (8) +n111-046-078:3318:3374 [5] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-078:3317:3375 [4] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n111-046-078:3319:3371 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3315:3376 [2] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n111-046-078:3313:3372 [0] NCCL INFO + PCI[24.0] - NIC/0-6000 +n111-046-078:3317:3375 [4] NCCL INFO + PCI[24.0] - GPU/0-5000 (9) +n111-046-078:3319:3371 [6] NCCL INFO + PCI[24.0] - NIC/0-6000 +n111-046-078:3313:3372 [0] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n111-046-078:3317:3375 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3318:3374 [5] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-078:3317:3375 [4] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n111-046-078:3320:3370 [7] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n111-046-078:3319:3371 [6] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n111-046-078:3315:3376 [2] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n111-046-078:3313:3372 [0] NCCL INFO + PCI[24.0] - NIC/0-7000 +n111-046-078:3317:3375 [4] NCCL INFO + PCI[24.0] - NIC/0-d000 +n111-046-078:3320:3370 [7] NCCL INFO CPU/0-0 (1/1/3) +n111-046-078:3319:3371 [6] NCCL INFO + PCI[24.0] - NIC/0-7000 +n111-046-078:3313:3372 [0] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n111-046-078:3317:3375 [4] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n111-046-078:3318:3374 [5] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-078:3313:3372 [0] NCCL INFO + PCI[24.0] - GPU/0-5000 (9) +n111-046-078:3320:3370 [7] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n111-046-078:3319:3371 [6] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n111-046-078:3317:3375 [4] NCCL INFO + PCI[24.0] - NIC/0-e000 +n111-046-078:3313:3372 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3320:3370 [7] NCCL INFO + PCI[24.0] - GPU/0-4000 (8) +n111-046-078:3319:3371 [6] NCCL INFO + PCI[24.0] - GPU/0-5000 (9) +n111-046-078:3317:3375 [4] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n111-046-078:3320:3370 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3318:3374 [5] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n111-046-078:3315:3376 [2] NCCL INFO Setting affinity for GPU 2 to 0fffffff,fffffe00,00000000,000fffff,fffffffe +n111-046-078:3313:3372 [0] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n111-046-078:3317:3375 [4] NCCL INFO + PCI[24.0] - GPU/0-b000 (10) +n111-046-078:3319:3371 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3320:3370 [7] NCCL INFO + PCI[24.0] - NIC/0-6000 +n111-046-078:3313:3372 [0] NCCL INFO + PCI[24.0] - NIC/0-d000 +n111-046-078:3317:3375 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3319:3371 [6] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n111-046-078:3320:3370 [7] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n111-046-078:3315:3376 [2] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n111-046-078:3313:3372 [0] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n111-046-078:3319:3371 [6] NCCL INFO + PCI[24.0] - NIC/0-d000 +n111-046-078:3313:3372 [0] NCCL INFO + PCI[24.0] - NIC/0-e000 +n111-046-078:3317:3375 [4] NCCL INFO + PCI[24.0] - GPU/0-c000 (11) +n111-046-078:3320:3370 [7] NCCL INFO + PCI[24.0] - NIC/0-7000 +n111-046-078:3319:3371 [6] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n111-046-078:3318:3374 [5] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n111-046-078:3317:3375 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3313:3372 [0] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n111-046-078:3320:3370 [7] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n111-046-078:3317:3375 [4] NCCL INFO + SYS[22.0] - CPU/0-1 +n111-046-078:3319:3371 [6] NCCL INFO + PCI[24.0] - NIC/0-e000 +n111-046-078:3313:3372 [0] NCCL INFO + PCI[24.0] - GPU/0-b000 (10) +n111-046-078:3320:3370 [7] NCCL INFO + PCI[24.0] - GPU/0-5000 (9) +n111-046-078:3317:3375 [4] NCCL INFO CPU/0-1 (1/1/3) +n111-046-078:3319:3371 [6] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n111-046-078:3313:3372 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3318:3374 [5] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-078:3317:3375 [4] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n111-046-078:3320:3370 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3319:3371 [6] NCCL INFO + PCI[24.0] - GPU/0-b000 (10) +n111-046-078:3313:3372 [0] NCCL INFO + PCI[24.0] - GPU/0-c000 (11) +n111-046-078:3317:3375 [4] NCCL INFO + PCI[24.0] - NIC/0-86000 +n111-046-078:3320:3370 [7] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n111-046-078:3319:3371 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3313:3372 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3317:3375 [4] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n111-046-078:3318:3374 [5] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-078:3313:3372 [0] NCCL INFO + SYS[22.0] - CPU/0-1 +n111-046-078:3320:3370 [7] NCCL INFO + PCI[24.0] - NIC/0-d000 +n111-046-078:3319:3371 [6] NCCL INFO + PCI[24.0] - GPU/0-c000 (11) +n111-046-078:3320:3370 [7] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n111-046-078:3315:3376 [2] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n111-046-078:3317:3375 [4] NCCL INFO + PCI[24.0] - NIC/0-87000 +n111-046-078:3313:3372 [0] NCCL INFO CPU/0-1 (1/1/3) +n111-046-078:3319:3371 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3320:3370 [7] NCCL INFO + PCI[24.0] - NIC/0-e000 +n111-046-078:3318:3374 [5] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-078:3317:3375 [4] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n111-046-078:3313:3372 [0] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n111-046-078:3313:3372 [0] NCCL INFO + PCI[24.0] - NIC/0-86000 +n111-046-078:3319:3371 [6] NCCL INFO + SYS[22.0] - CPU/0-1 +n111-046-078:3320:3370 [7] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n111-046-078:3317:3375 [4] NCCL INFO + PCI[24.0] - GPU/0-84000 (12) +n111-046-078:3313:3372 [0] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n111-046-078:3319:3371 [6] NCCL INFO CPU/0-1 (1/1/3) +n111-046-078:3318:3374 [5] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-078:3320:3370 [7] NCCL INFO + PCI[24.0] - GPU/0-b000 (10) +n111-046-078:3317:3375 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3313:3372 [0] NCCL INFO + PCI[24.0] - NIC/0-87000 +n111-046-078:3319:3371 [6] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n111-046-078:3320:3370 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3317:3375 [4] NCCL INFO + PCI[24.0] - GPU/0-85000 (13) +n111-046-078:3319:3371 [6] NCCL INFO + PCI[24.0] - NIC/0-86000 +n111-046-078:3313:3372 [0] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n111-046-078:3320:3370 [7] NCCL INFO + PCI[24.0] - GPU/0-c000 (11) +n111-046-078:3317:3375 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3313:3372 [0] NCCL INFO + PCI[24.0] - GPU/0-84000 (12) +n111-046-078:3320:3370 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3317:3375 [4] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n111-046-078:3313:3372 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3319:3371 [6] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n111-046-078:3317:3375 [4] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n111-046-078:3318:3374 [5] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n111-046-078:3319:3371 [6] NCCL INFO + PCI[24.0] - NIC/0-87000 +n111-046-078:3319:3371 [6] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n111-046-078:3320:3370 [7] NCCL INFO + SYS[22.0] - CPU/0-1 +n111-046-078:3313:3372 [0] NCCL INFO + PCI[24.0] - GPU/0-85000 (13) +n111-046-078:3317:3375 [4] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n111-046-078:3319:3371 [6] NCCL INFO + PCI[24.0] - GPU/0-84000 (12) +n111-046-078:3313:3372 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3320:3370 [7] NCCL INFO CPU/0-1 (1/1/3) +n111-046-078:3317:3375 [4] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n111-046-078:3318:3374 [5] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n111-046-078:3319:3371 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3313:3372 [0] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n111-046-078:3320:3370 [7] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n111-046-078:3317:3375 [4] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n111-046-078:3320:3370 [7] NCCL INFO + PCI[24.0] - NIC/0-86000 +n111-046-078:3319:3371 [6] NCCL INFO + PCI[24.0] - GPU/0-85000 (13) +n111-046-078:3313:3372 [0] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n111-046-078:3317:3375 [4] NCCL INFO + PCI[24.0] - GPU/0-8b000 (14) +n111-046-078:3320:3370 [7] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n111-046-078:3319:3371 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3313:3372 [0] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n111-046-078:3320:3370 [7] NCCL INFO + PCI[24.0] - NIC/0-87000 +n111-046-078:3318:3374 [5] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n111-046-078:3317:3375 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3319:3371 [6] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n111-046-078:3313:3372 [0] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n111-046-078:3320:3370 [7] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n111-046-078:3317:3375 [4] NCCL INFO + PCI[24.0] - GPU/0-8c000 (15) +n111-046-078:3319:3371 [6] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n111-046-078:3313:3372 [0] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n111-046-078:3318:3374 [5] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n111-046-078:3320:3370 [7] NCCL INFO + PCI[24.0] - GPU/0-84000 (12) +n111-046-078:3317:3375 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3319:3371 [6] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n111-046-078:3313:3372 [0] NCCL INFO + PCI[24.0] - GPU/0-8b000 (14) +n111-046-078:3317:3375 [4] NCCL INFO + SYS[22.0] - CPU/0-0 +n111-046-078:3319:3371 [6] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n111-046-078:3319:3371 [6] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n111-046-078:3320:3370 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3313:3372 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3319:3371 [6] NCCL INFO + PCI[24.0] - GPU/0-8b000 (14) +n111-046-078:3317:3375 [4] NCCL INFO ========================================== +n111-046-078:3320:3370 [7] NCCL INFO + PCI[24.0] - GPU/0-85000 (13) +n111-046-078:3313:3372 [0] NCCL INFO + PCI[24.0] - GPU/0-8c000 (15) +n111-046-078:3319:3371 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3320:3370 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3319:3371 [6] NCCL INFO + PCI[24.0] - GPU/0-8c000 (15) +n111-046-078:3318:3374 [5] NCCL INFO Setting affinity for GPU 5 to 7fff,ffffffff,e0000000,0000007f,ffffffff,ffe00000,00000000 +n111-046-078:3320:3370 [7] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n111-046-078:3319:3371 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3313:3372 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3313:3372 [0] NCCL INFO + SYS[22.0] - CPU/0-0 +n111-046-078:3320:3370 [7] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n111-046-078:3317:3375 [4] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-078:3319:3371 [6] NCCL INFO + SYS[22.0] - CPU/0-0 +n111-046-078:3313:3372 [0] NCCL INFO ========================================== +n111-046-078:3319:3371 [6] NCCL INFO ========================================== +n111-046-078:3320:3370 [7] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n111-046-078:3318:3374 [5] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n111-046-078:3320:3370 [7] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n111-046-078:3320:3370 [7] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n111-046-078:3320:3370 [7] NCCL INFO + PCI[24.0] - GPU/0-8b000 (14) +n111-046-078:3317:3375 [4] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-078:3320:3370 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3313:3372 [0] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-078:3320:3370 [7] NCCL INFO + PCI[24.0] - GPU/0-8c000 (15) +n111-046-078:3319:3371 [6] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-078:3320:3370 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3320:3370 [7] NCCL INFO + SYS[22.0] - CPU/0-0 +n111-046-078:3320:3370 [7] NCCL INFO ========================================== +n111-046-078:3313:3372 [0] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-078:3319:3371 [6] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-078:3317:3375 [4] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-078:3320:3370 [7] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-078:3316:3369 [3] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n111-046-078:3313:3372 [0] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-078:3319:3371 [6] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-078:3317:3375 [4] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-078:3316:3369 [3] NCCL INFO CPU/0-0 (1/1/3) +n111-046-078:3316:3369 [3] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n111-046-078:3320:3370 [7] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-078:3313:3372 [0] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-078:3318:3374 [5] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n111-046-078:3316:3369 [3] NCCL INFO + PCI[24.0] - GPU/0-4000 (8) +n111-046-078:3317:3375 [4] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-078:3319:3371 [6] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-078:3316:3369 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3320:3370 [7] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-078:3316:3369 [3] NCCL INFO + PCI[24.0] - NIC/0-6000 +n111-046-078:3313:3372 [0] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-078:3316:3369 [3] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n111-046-078:3317:3375 [4] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-078:3319:3371 [6] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-078:3316:3369 [3] NCCL INFO + PCI[24.0] - NIC/0-7000 +n111-046-078:3320:3370 [7] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-078:3313:3372 [0] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-078:3316:3369 [3] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n111-046-078:3316:3369 [3] NCCL INFO + PCI[24.0] - GPU/0-5000 (9) +n111-046-078:3317:3375 [4] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n111-046-078:3316:3369 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3319:3371 [6] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-078:3320:3370 [7] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-078:3313:3372 [0] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n111-046-078:3316:3369 [3] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n111-046-078:3317:3375 [4] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n111-046-078:3316:3369 [3] NCCL INFO + PCI[24.0] - NIC/0-d000 +n111-046-078:3319:3371 [6] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n111-046-078:3320:3370 [7] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-078:3316:3369 [3] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n111-046-078:3313:3372 [0] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n111-046-078:3316:3369 [3] NCCL INFO + PCI[24.0] - NIC/0-e000 +n111-046-078:3317:3375 [4] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-078:3316:3369 [3] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n111-046-078:3316:3369 [3] NCCL INFO + PCI[24.0] - GPU/0-b000 (10) +n111-046-078:3320:3370 [7] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n111-046-078:3316:3369 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3313:3372 [0] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-078:3317:3375 [4] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-078:3316:3369 [3] NCCL INFO + PCI[24.0] - GPU/0-c000 (11) +n111-046-078:3316:3369 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3319:3371 [6] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n111-046-078:3316:3369 [3] NCCL INFO + SYS[22.0] - CPU/0-1 +n111-046-078:3320:3370 [7] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n111-046-078:3313:3372 [0] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-078:3317:3375 [4] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-078:3316:3369 [3] NCCL INFO CPU/0-1 (1/1/3) +n111-046-078:3319:3371 [6] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-078:3316:3369 [3] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n111-046-078:3316:3369 [3] NCCL INFO + PCI[24.0] - NIC/0-86000 +n111-046-078:3316:3369 [3] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n111-046-078:3320:3370 [7] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-078:3317:3375 [4] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-078:3319:3371 [6] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-078:3313:3372 [0] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-078:3316:3369 [3] NCCL INFO + PCI[24.0] - NIC/0-87000 +n111-046-078:3316:3369 [3] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n111-046-078:3316:3369 [3] NCCL INFO + PCI[24.0] - GPU/0-84000 (12) +n111-046-078:3320:3370 [7] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-078:3317:3375 [4] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n111-046-078:3316:3369 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3313:3372 [0] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-078:3319:3371 [6] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-078:3316:3369 [3] NCCL INFO + PCI[24.0] - GPU/0-85000 (13) +n111-046-078:3316:3369 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3316:3369 [3] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n111-046-078:3316:3369 [3] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n111-046-078:3317:3375 [4] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n111-046-078:3316:3369 [3] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n111-046-078:3313:3372 [0] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n111-046-078:3316:3369 [3] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n111-046-078:3319:3371 [6] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-078:3320:3370 [7] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-078:3316:3369 [3] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n111-046-078:3317:3375 [4] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n111-046-078:3316:3369 [3] NCCL INFO + PCI[24.0] - GPU/0-8b000 (14) +n111-046-078:3316:3369 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3313:3372 [0] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n111-046-078:3319:3371 [6] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n111-046-078:3316:3369 [3] NCCL INFO + PCI[24.0] - GPU/0-8c000 (15) +n111-046-078:3317:3375 [4] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n111-046-078:3320:3370 [7] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-078:3316:3369 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3316:3369 [3] NCCL INFO + SYS[22.0] - CPU/0-0 +n111-046-078:3313:3372 [0] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n111-046-078:3316:3369 [3] NCCL INFO ========================================== +n111-046-078:3319:3371 [6] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n111-046-078:3320:3370 [7] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n111-046-078:3317:3375 [4] NCCL INFO Setting affinity for GPU 4 to 7fff,ffffffff,e0000000,0000007f,ffffffff,ffe00000,00000000 +n111-046-078:3313:3372 [0] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n111-046-078:3319:3371 [6] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n111-046-078:3320:3370 [7] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n111-046-078:3317:3375 [4] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n111-046-078:3316:3369 [3] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-078:3319:3371 [6] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n111-046-078:3320:3370 [7] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n111-046-078:3313:3372 [0] NCCL INFO Setting affinity for GPU 0 to 0fffffff,fffffe00,00000000,000fffff,fffffffe +n111-046-078:3319:3371 [6] NCCL INFO Setting affinity for GPU 6 to 7fff,ffffffff,e0000000,0000007f,ffffffff,ffe00000,00000000 +n111-046-078:3320:3370 [7] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n111-046-078:3316:3369 [3] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-078:3319:3371 [6] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n111-046-078:3313:3372 [0] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n111-046-078:3320:3370 [7] NCCL INFO Setting affinity for GPU 7 to 7fff,ffffffff,e0000000,0000007f,ffffffff,ffe00000,00000000 +n111-046-078:3316:3369 [3] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-078:3317:3375 [4] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n111-046-078:3316:3369 [3] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-078:3313:3372 [0] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n111-046-078:3319:3371 [6] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n111-046-078:3316:3369 [3] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-078:3316:3369 [3] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-078:3320:3370 [7] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n111-046-078:3316:3369 [3] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n111-046-078:3316:3369 [3] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n111-046-078:3316:3369 [3] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-078:3320:3370 [7] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n111-046-078:3316:3369 [3] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-078:3316:3369 [3] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-078:3314:3373 [1] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n111-046-078:3314:3373 [1] NCCL INFO CPU/0-0 (1/1/3) +n111-046-078:3314:3373 [1] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n111-046-078:3316:3369 [3] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-078:3314:3373 [1] NCCL INFO + PCI[24.0] - GPU/0-4000 (8) +n111-046-078:3314:3373 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3314:3373 [1] NCCL INFO + PCI[24.0] - NIC/0-6000 +n111-046-078:3316:3369 [3] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n111-046-078:3314:3373 [1] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n111-046-078:3314:3373 [1] NCCL INFO + PCI[24.0] - NIC/0-7000 +n111-046-078:3314:3373 [1] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n111-046-078:3314:3373 [1] NCCL INFO + PCI[24.0] - GPU/0-5000 (9) +n111-046-078:3314:3373 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3314:3373 [1] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n111-046-078:3314:3373 [1] NCCL INFO + PCI[24.0] - NIC/0-d000 +n111-046-078:3316:3369 [3] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n111-046-078:3314:3373 [1] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n111-046-078:3314:3373 [1] NCCL INFO + PCI[24.0] - NIC/0-e000 +n111-046-078:3314:3373 [1] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n111-046-078:3314:3373 [1] NCCL INFO + PCI[24.0] - GPU/0-b000 (10) +n111-046-078:3314:3373 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3314:3373 [1] NCCL INFO + PCI[24.0] - GPU/0-c000 (11) +n111-046-078:3314:3373 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3316:3369 [3] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n111-046-078:3314:3373 [1] NCCL INFO + SYS[22.0] - CPU/0-1 +n111-046-078:3314:3373 [1] NCCL INFO CPU/0-1 (1/1/3) +n111-046-078:3314:3373 [1] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n111-046-078:3316:3369 [3] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n111-046-078:3314:3373 [1] NCCL INFO + PCI[24.0] - NIC/0-86000 +n111-046-078:3314:3373 [1] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n111-046-078:3314:3373 [1] NCCL INFO + PCI[24.0] - NIC/0-87000 +n111-046-078:3314:3373 [1] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n111-046-078:3314:3373 [1] NCCL INFO + PCI[24.0] - GPU/0-84000 (12) +n111-046-078:3316:3369 [3] NCCL INFO Setting affinity for GPU 3 to 0fffffff,fffffe00,00000000,000fffff,fffffffe +n111-046-078:3314:3373 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3314:3373 [1] NCCL INFO + PCI[24.0] - GPU/0-85000 (13) +n111-046-078:3314:3373 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3314:3373 [1] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n111-046-078:3314:3373 [1] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n111-046-078:3314:3373 [1] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n111-046-078:3314:3373 [1] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n111-046-078:3314:3373 [1] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n111-046-078:3314:3373 [1] NCCL INFO + PCI[24.0] - GPU/0-8b000 (14) +n111-046-078:3314:3373 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3314:3373 [1] NCCL INFO + PCI[24.0] - GPU/0-8c000 (15) +n111-046-078:3314:3373 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-078:3314:3373 [1] NCCL INFO + SYS[22.0] - CPU/0-0 +n111-046-078:3314:3373 [1] NCCL INFO ========================================== +n111-046-078:3314:3373 [1] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-078:3316:3369 [3] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n111-046-078:3314:3373 [1] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-078:3314:3373 [1] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-078:3314:3373 [1] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-078:3314:3373 [1] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-078:3316:3369 [3] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n111-046-078:3314:3373 [1] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-078:3314:3373 [1] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n111-046-078:3314:3373 [1] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n111-046-078:3314:3373 [1] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-078:3314:3373 [1] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-078:3314:3373 [1] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-078:3314:3373 [1] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-078:3314:3373 [1] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n111-046-078:3314:3373 [1] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n111-046-078:3314:3373 [1] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n111-046-078:3314:3373 [1] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n111-046-078:3314:3373 [1] NCCL INFO Setting affinity for GPU 1 to 0fffffff,fffffe00,00000000,000fffff,fffffffe +n111-046-078:3314:3373 [1] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n111-046-078:3314:3373 [1] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n111-046-078:3315:3376 [2] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n111-046-078:3315:3376 [2] NCCL INFO 0 : NET/0-0 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 NET/0-0 +n111-046-078:3315:3376 [2] NCCL INFO 1 : NET/0-1 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 NET/0-1 +n111-046-078:3315:3376 [2] NCCL INFO 2 : NET/0-2 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 NET/0-2 +n111-046-078:3315:3376 [2] NCCL INFO 3 : NET/0-3 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 NET/0-3 +n111-046-078:3315:3376 [2] NCCL INFO 4 : NET/0-4 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 NET/0-4 +n111-046-078:3315:3376 [2] NCCL INFO 5 : NET/0-5 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 NET/0-5 +n111-046-078:3315:3376 [2] NCCL INFO 6 : NET/0-6 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 NET/0-6 +n111-046-078:3315:3376 [2] NCCL INFO 7 : NET/0-7 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 NET/0-7 +n111-046-078:3315:3376 [2] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n111-046-078:3315:3376 [2] NCCL INFO 0 : NET/0-0 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 NET/0-0 +n111-046-078:3315:3376 [2] NCCL INFO 1 : NET/0-1 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 NET/0-1 +n111-046-078:3315:3376 [2] NCCL INFO 2 : NET/0-2 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 NET/0-2 +n111-046-078:3315:3376 [2] NCCL INFO 3 : NET/0-3 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 NET/0-3 +n111-046-078:3315:3376 [2] NCCL INFO 4 : NET/0-4 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 NET/0-4 +n111-046-078:3315:3376 [2] NCCL INFO 5 : NET/0-5 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 NET/0-5 +n111-046-078:3315:3376 [2] NCCL INFO 6 : NET/0-6 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 NET/0-6 +n111-046-078:3315:3376 [2] NCCL INFO 7 : NET/0-7 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 NET/0-7 +n111-046-078:3318:3374 [5] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n111-046-078:3318:3374 [5] NCCL INFO 0 : NET/0-0 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 NET/0-0 +n111-046-078:3318:3374 [5] NCCL INFO 1 : NET/0-1 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 NET/0-1 +n111-046-078:3318:3374 [5] NCCL INFO 2 : NET/0-2 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 NET/0-2 +n111-046-078:3318:3374 [5] NCCL INFO 3 : NET/0-3 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 NET/0-3 +n111-046-078:3318:3374 [5] NCCL INFO 4 : NET/0-4 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 NET/0-4 +n111-046-078:3318:3374 [5] NCCL INFO 5 : NET/0-5 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 NET/0-5 +n111-046-078:3318:3374 [5] NCCL INFO 6 : NET/0-6 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 NET/0-6 +n111-046-078:3318:3374 [5] NCCL INFO 7 : NET/0-7 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 NET/0-7 +n111-046-078:3319:3371 [6] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n111-046-078:3319:3371 [6] NCCL INFO 0 : NET/0-0 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 NET/0-0 +n111-046-078:3319:3371 [6] NCCL INFO 1 : NET/0-1 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 NET/0-1 +n111-046-078:3319:3371 [6] NCCL INFO 2 : NET/0-2 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 NET/0-2 +n111-046-078:3319:3371 [6] NCCL INFO 3 : NET/0-3 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 NET/0-3 +n111-046-078:3319:3371 [6] NCCL INFO 4 : NET/0-4 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 NET/0-4 +n111-046-078:3319:3371 [6] NCCL INFO 5 : NET/0-5 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 NET/0-5 +n111-046-078:3319:3371 [6] NCCL INFO 6 : NET/0-6 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 NET/0-6 +n111-046-078:3319:3371 [6] NCCL INFO 7 : NET/0-7 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 NET/0-7 +n111-046-078:3317:3375 [4] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n111-046-078:3317:3375 [4] NCCL INFO 0 : NET/0-0 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 NET/0-0 +n111-046-078:3317:3375 [4] NCCL INFO 1 : NET/0-1 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 NET/0-1 +n111-046-078:3317:3375 [4] NCCL INFO 2 : NET/0-2 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 NET/0-2 +n111-046-078:3317:3375 [4] NCCL INFO 3 : NET/0-3 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 NET/0-3 +n111-046-078:3317:3375 [4] NCCL INFO 4 : NET/0-4 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 NET/0-4 +n111-046-078:3317:3375 [4] NCCL INFO 5 : NET/0-5 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 NET/0-5 +n111-046-078:3317:3375 [4] NCCL INFO 6 : NET/0-6 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 NET/0-6 +n111-046-078:3317:3375 [4] NCCL INFO 7 : NET/0-7 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 NET/0-7 +n111-046-078:3318:3374 [5] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n111-046-078:3318:3374 [5] NCCL INFO 0 : NET/0-0 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 NET/0-0 +n111-046-078:3318:3374 [5] NCCL INFO 1 : NET/0-1 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 NET/0-1 +n111-046-078:3318:3374 [5] NCCL INFO 2 : NET/0-2 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 NET/0-2 +n111-046-078:3318:3374 [5] NCCL INFO 3 : NET/0-3 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 NET/0-3 +n111-046-078:3318:3374 [5] NCCL INFO 4 : NET/0-4 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 NET/0-4 +n111-046-078:3318:3374 [5] NCCL INFO 5 : NET/0-5 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 NET/0-5 +n111-046-078:3318:3374 [5] NCCL INFO 6 : NET/0-6 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 NET/0-6 +n111-046-078:3319:3371 [6] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n111-046-078:3318:3374 [5] NCCL INFO 7 : NET/0-7 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 NET/0-7 +n111-046-078:3319:3371 [6] NCCL INFO 0 : NET/0-0 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 NET/0-0 +n111-046-078:3319:3371 [6] NCCL INFO 1 : NET/0-1 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 NET/0-1 +n111-046-078:3319:3371 [6] NCCL INFO 2 : NET/0-2 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 NET/0-2 +n111-046-078:3319:3371 [6] NCCL INFO 3 : NET/0-3 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 NET/0-3 +n111-046-078:3319:3371 [6] NCCL INFO 4 : NET/0-4 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 NET/0-4 +n111-046-078:3319:3371 [6] NCCL INFO 5 : NET/0-5 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 NET/0-5 +n111-046-078:3319:3371 [6] NCCL INFO 6 : NET/0-6 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 NET/0-6 +n111-046-078:3320:3370 [7] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n111-046-078:3319:3371 [6] NCCL INFO 7 : NET/0-7 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 NET/0-7 +n111-046-078:3320:3370 [7] NCCL INFO 0 : NET/0-0 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 NET/0-0 +n111-046-078:3320:3370 [7] NCCL INFO 1 : NET/0-1 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 NET/0-1 +n111-046-078:3320:3370 [7] NCCL INFO 2 : NET/0-2 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 NET/0-2 +n111-046-078:3320:3370 [7] NCCL INFO 3 : NET/0-3 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 NET/0-3 +n111-046-078:3320:3370 [7] NCCL INFO 4 : NET/0-4 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 NET/0-4 +n111-046-078:3320:3370 [7] NCCL INFO 5 : NET/0-5 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 NET/0-5 +n111-046-078:3320:3370 [7] NCCL INFO 6 : NET/0-6 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 NET/0-6 +n111-046-078:3320:3370 [7] NCCL INFO 7 : NET/0-7 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 NET/0-7 +n111-046-078:3317:3375 [4] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n111-046-078:3317:3375 [4] NCCL INFO 0 : NET/0-0 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 NET/0-0 +n111-046-078:3317:3375 [4] NCCL INFO 1 : NET/0-1 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 NET/0-1 +n111-046-078:3317:3375 [4] NCCL INFO 2 : NET/0-2 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 NET/0-2 +n111-046-078:3317:3375 [4] NCCL INFO 3 : NET/0-3 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 NET/0-3 +n111-046-078:3313:3372 [0] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n111-046-078:3317:3375 [4] NCCL INFO 4 : NET/0-4 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 NET/0-4 +n111-046-078:3317:3375 [4] NCCL INFO 5 : NET/0-5 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 NET/0-5 +n111-046-078:3313:3372 [0] NCCL INFO 0 : NET/0-0 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 NET/0-0 +n111-046-078:3317:3375 [4] NCCL INFO 6 : NET/0-6 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 NET/0-6 +n111-046-078:3317:3375 [4] NCCL INFO 7 : NET/0-7 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 NET/0-7 +n111-046-078:3313:3372 [0] NCCL INFO 1 : NET/0-1 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 NET/0-1 +n111-046-078:3313:3372 [0] NCCL INFO 2 : NET/0-2 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 NET/0-2 +n111-046-078:3313:3372 [0] NCCL INFO 3 : NET/0-3 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 NET/0-3 +n111-046-078:3313:3372 [0] NCCL INFO 4 : NET/0-4 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 NET/0-4 +n111-046-078:3313:3372 [0] NCCL INFO 5 : NET/0-5 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 NET/0-5 +n111-046-078:3313:3372 [0] NCCL INFO 6 : NET/0-6 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 NET/0-6 +n111-046-078:3313:3372 [0] NCCL INFO 7 : NET/0-7 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 NET/0-7 +n111-046-078:3320:3370 [7] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n111-046-078:3320:3370 [7] NCCL INFO 0 : NET/0-0 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 NET/0-0 +n111-046-078:3320:3370 [7] NCCL INFO 1 : NET/0-1 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 NET/0-1 +n111-046-078:3320:3370 [7] NCCL INFO 2 : NET/0-2 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 NET/0-2 +n111-046-078:3320:3370 [7] NCCL INFO 3 : NET/0-3 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 NET/0-3 +n111-046-078:3320:3370 [7] NCCL INFO 4 : NET/0-4 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 NET/0-4 +n111-046-078:3320:3370 [7] NCCL INFO 5 : NET/0-5 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 NET/0-5 +n111-046-078:3320:3370 [7] NCCL INFO 6 : NET/0-6 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 NET/0-6 +n111-046-078:3320:3370 [7] NCCL INFO 7 : NET/0-7 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 NET/0-7 +n111-046-078:3313:3372 [0] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n111-046-078:3313:3372 [0] NCCL INFO 0 : NET/0-0 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 NET/0-0 +n111-046-078:3313:3372 [0] NCCL INFO 1 : NET/0-1 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 NET/0-1 +n111-046-078:3313:3372 [0] NCCL INFO 2 : NET/0-2 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 NET/0-2 +n111-046-078:3313:3372 [0] NCCL INFO 3 : NET/0-3 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 NET/0-3 +n111-046-078:3313:3372 [0] NCCL INFO 4 : NET/0-4 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 NET/0-4 +n111-046-078:3313:3372 [0] NCCL INFO 5 : NET/0-5 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 NET/0-5 +n111-046-078:3313:3372 [0] NCCL INFO 6 : NET/0-6 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 NET/0-6 +n111-046-078:3313:3372 [0] NCCL INFO 7 : NET/0-7 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 NET/0-7 +n111-046-078:3314:3373 [1] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n111-046-078:3314:3373 [1] NCCL INFO 0 : NET/0-0 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 NET/0-0 +n111-046-078:3314:3373 [1] NCCL INFO 1 : NET/0-1 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 NET/0-1 +n111-046-078:3314:3373 [1] NCCL INFO 2 : NET/0-2 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 NET/0-2 +n111-046-078:3314:3373 [1] NCCL INFO 3 : NET/0-3 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 NET/0-3 +n111-046-078:3314:3373 [1] NCCL INFO 4 : NET/0-4 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 NET/0-4 +n111-046-078:3314:3373 [1] NCCL INFO 5 : NET/0-5 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 NET/0-5 +n111-046-078:3314:3373 [1] NCCL INFO 6 : NET/0-6 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 NET/0-6 +n111-046-078:3314:3373 [1] NCCL INFO 7 : NET/0-7 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 NET/0-7 +n111-046-078:3314:3373 [1] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n111-046-078:3314:3373 [1] NCCL INFO 0 : NET/0-0 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 NET/0-0 +n111-046-078:3314:3373 [1] NCCL INFO 1 : NET/0-1 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 NET/0-1 +n111-046-078:3314:3373 [1] NCCL INFO 2 : NET/0-2 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 NET/0-2 +n111-046-078:3314:3373 [1] NCCL INFO 3 : NET/0-3 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 NET/0-3 +n111-046-078:3314:3373 [1] NCCL INFO 4 : NET/0-4 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 NET/0-4 +n111-046-078:3314:3373 [1] NCCL INFO 5 : NET/0-5 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 NET/0-5 +n111-046-078:3314:3373 [1] NCCL INFO 6 : NET/0-6 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 NET/0-6 +n111-046-078:3314:3373 [1] NCCL INFO 7 : NET/0-7 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 NET/0-7 +n111-046-078:3316:3369 [3] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n111-046-078:3316:3369 [3] NCCL INFO 0 : NET/0-0 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 NET/0-0 +n111-046-078:3316:3369 [3] NCCL INFO 1 : NET/0-1 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 NET/0-1 +n111-046-078:3316:3369 [3] NCCL INFO 2 : NET/0-2 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 NET/0-2 +n111-046-078:3316:3369 [3] NCCL INFO 3 : NET/0-3 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 NET/0-3 +n111-046-078:3316:3369 [3] NCCL INFO 4 : NET/0-4 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 NET/0-4 +n111-046-078:3316:3369 [3] NCCL INFO 5 : NET/0-5 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 NET/0-5 +n111-046-078:3316:3369 [3] NCCL INFO 6 : NET/0-6 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 NET/0-6 +n111-046-078:3316:3369 [3] NCCL INFO 7 : NET/0-7 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 NET/0-7 +n111-046-078:3316:3369 [3] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n111-046-078:3316:3369 [3] NCCL INFO 0 : NET/0-0 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 NET/0-0 +n111-046-078:3316:3369 [3] NCCL INFO 1 : NET/0-1 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 NET/0-1 +n111-046-078:3316:3369 [3] NCCL INFO 2 : NET/0-2 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 NET/0-2 +n111-046-078:3316:3369 [3] NCCL INFO 3 : NET/0-3 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 NET/0-3 +n111-046-078:3316:3369 [3] NCCL INFO 4 : NET/0-4 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 NET/0-4 +n111-046-078:3316:3369 [3] NCCL INFO 5 : NET/0-5 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 NET/0-5 +n111-046-078:3316:3369 [3] NCCL INFO 6 : NET/0-6 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 NET/0-6 +n111-046-078:3316:3369 [3] NCCL INFO 7 : NET/0-7 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 NET/0-7 +n111-046-078:3320:3370 [7] NCCL INFO comm 0xbdf0f50 rank 15 nRanks 32 nNodes 4 localRanks 8 localRank 7 MNNVL 0 +n111-046-078:3320:3370 [7] NCCL INFO Tree 6 : 14 -> 15 -> 8/-1/-1 +n111-046-078:3320:3370 [7] NCCL INFO Tree 14 : 14 -> 15 -> 8/22/-1 +n111-046-078:3320:3370 [7] NCCL INFO Tree 7 : 22 -> 15 -> 14/-1/-1 +n111-046-078:3320:3370 [7] NCCL INFO Tree 15 : 31 -> 15 -> 14/7/-1 +n111-046-078:3319:3371 [6] NCCL INFO comm 0xb9c26e0 rank 14 nRanks 32 nNodes 4 localRanks 8 localRank 6 MNNVL 0 +n111-046-078:3320:3370 [7] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n111-046-078:3318:3374 [5] NCCL INFO comm 0xb26c190 rank 13 nRanks 32 nNodes 4 localRanks 8 localRank 5 MNNVL 0 +n111-046-078:3320:3370 [7] NCCL INFO Ring 00 : 8 -> 15 -> 14 +n111-046-078:3320:3370 [7] NCCL INFO Ring 01 : 14 -> 15 -> 8 +n111-046-078:3320:3370 [7] NCCL INFO Ring 02 : 8 -> 15 -> 14 +n111-046-078:3320:3370 [7] NCCL INFO Ring 03 : 14 -> 15 -> 8 +n111-046-078:3320:3370 [7] NCCL INFO Ring 04 : 8 -> 15 -> 14 +n111-046-078:3320:3370 [7] NCCL INFO Ring 05 : 14 -> 15 -> 8 +n111-046-078:3320:3370 [7] NCCL INFO Ring 06 : 8 -> 15 -> 22 +n111-046-078:3318:3374 [5] NCCL INFO Tree 4 : 12 -> 13 -> 14/-1/-1 +n111-046-078:3320:3370 [7] NCCL INFO Ring 07 : 6 -> 15 -> 8 +n111-046-078:3319:3371 [6] NCCL INFO Tree 6 : 23 -> 14 -> 15/-1/-1 +n111-046-078:3320:3370 [7] NCCL INFO Ring 08 : 8 -> 15 -> 14 +n111-046-078:3318:3374 [5] NCCL INFO Tree 12 : 12 -> 13 -> 14/20/-1 +n111-046-078:3320:3370 [7] NCCL INFO Ring 09 : 14 -> 15 -> 8 +n111-046-078:3319:3371 [6] NCCL INFO Tree 14 : 30 -> 14 -> 15/6/-1 +n111-046-078:3320:3370 [7] NCCL INFO Ring 10 : 8 -> 15 -> 14 +n111-046-078:3319:3371 [6] NCCL INFO Tree 7 : 15 -> 14 -> 13/-1/-1 +n111-046-078:3320:3370 [7] NCCL INFO Ring 11 : 14 -> 15 -> 8 +n111-046-078:3318:3374 [5] NCCL INFO Tree 5 : 20 -> 13 -> 12/-1/-1 +n111-046-078:3319:3371 [6] NCCL INFO Tree 15 : 15 -> 14 -> 13/23/-1 +n111-046-078:3320:3370 [7] NCCL INFO Ring 12 : 8 -> 15 -> 14 +n111-046-078:3318:3374 [5] NCCL INFO Tree 13 : 29 -> 13 -> 12/5/-1 +n111-046-078:3320:3370 [7] NCCL INFO Ring 13 : 14 -> 15 -> 8 +n111-046-078:3320:3370 [7] NCCL INFO Ring 14 : 8 -> 15 -> 22 +n111-046-078:3320:3370 [7] NCCL INFO Ring 15 : 6 -> 15 -> 8 +n111-046-078:3320:3370 [7] NCCL INFO Trees [0] -1/-1/-1->15->14 [1] 14/-1/-1->15->8 [2] 8/-1/-1->15->14 [3] 14/-1/-1->15->8 [4] 8/-1/-1->15->14 [5] 14/-1/-1->15->8 [6] 8/-1/-1->15->14 [7] 14/-1/-1->15->22 [8] -1/-1/-1->15->14 [9] 14/-1/-1->15->8 [10] 8/-1/-1->15->14 [11] 14/-1/-1->15->8 [12] 8/-1/-1->15->14 [13] 14/-1/-1->15->8 [14] 8/22/-1->15->14 [15] 14/7/-1->15->31 +n111-046-078:3319:3371 [6] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n111-046-078:3318:3374 [5] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n111-046-078:3320:3370 [7] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n111-046-078:3320:3370 [7] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n111-046-078:3320:3370 [7] NCCL INFO P2P Chunksize set to 524288 +n111-046-078:3319:3371 [6] NCCL INFO Ring 00 : 15 -> 14 -> 13 +n111-046-078:3319:3371 [6] NCCL INFO Ring 01 : 13 -> 14 -> 15 +n111-046-078:3318:3374 [5] NCCL INFO Ring 00 : 14 -> 13 -> 12 +n111-046-078:3319:3371 [6] NCCL INFO Ring 02 : 15 -> 14 -> 13 +n111-046-078:3318:3374 [5] NCCL INFO Ring 01 : 12 -> 13 -> 14 +n111-046-078:3319:3371 [6] NCCL INFO Ring 03 : 13 -> 14 -> 15 +n111-046-078:3318:3374 [5] NCCL INFO Ring 02 : 14 -> 13 -> 12 +n111-046-078:3319:3371 [6] NCCL INFO Ring 04 : 15 -> 14 -> 13 +n111-046-078:3318:3374 [5] NCCL INFO Ring 03 : 12 -> 13 -> 14 +n111-046-078:3317:3375 [4] NCCL INFO comm 0xc1a21c0 rank 12 nRanks 32 nNodes 4 localRanks 8 localRank 4 MNNVL 0 +n111-046-078:3319:3371 [6] NCCL INFO Ring 05 : 13 -> 14 -> 15 +n111-046-078:3318:3374 [5] NCCL INFO Ring 04 : 14 -> 13 -> 20 +n111-046-078:3319:3371 [6] NCCL INFO Ring 06 : 7 -> 14 -> 13 +n111-046-078:3319:3371 [6] NCCL INFO Ring 07 : 13 -> 14 -> 23 +n111-046-078:3318:3374 [5] NCCL INFO Ring 05 : 4 -> 13 -> 14 +n111-046-078:3319:3371 [6] NCCL INFO Ring 08 : 15 -> 14 -> 13 +n111-046-078:3318:3374 [5] NCCL INFO Ring 06 : 14 -> 13 -> 12 +n111-046-078:3319:3371 [6] NCCL INFO Ring 09 : 13 -> 14 -> 15 +n111-046-078:3318:3374 [5] NCCL INFO Ring 07 : 12 -> 13 -> 14 +n111-046-078:3319:3371 [6] NCCL INFO Ring 10 : 15 -> 14 -> 13 +n111-046-078:3318:3374 [5] NCCL INFO Ring 08 : 14 -> 13 -> 12 +n111-046-078:3319:3371 [6] NCCL INFO Ring 11 : 13 -> 14 -> 15 +n111-046-078:3318:3374 [5] NCCL INFO Ring 09 : 12 -> 13 -> 14 +n111-046-078:3319:3371 [6] NCCL INFO Ring 12 : 15 -> 14 -> 13 +n111-046-078:3318:3374 [5] NCCL INFO Ring 10 : 14 -> 13 -> 12 +n111-046-078:3319:3371 [6] NCCL INFO Ring 13 : 13 -> 14 -> 15 +n111-046-078:3318:3374 [5] NCCL INFO Ring 11 : 12 -> 13 -> 14 +n111-046-078:3319:3371 [6] NCCL INFO Ring 14 : 7 -> 14 -> 13 +n111-046-078:3318:3374 [5] NCCL INFO Ring 12 : 14 -> 13 -> 20 +n111-046-078:3319:3371 [6] NCCL INFO Ring 15 : 13 -> 14 -> 23 +n111-046-078:3318:3374 [5] NCCL INFO Ring 13 : 4 -> 13 -> 14 +n111-046-078:3319:3371 [6] NCCL INFO Trees [0] 15/-1/-1->14->13 [1] 13/-1/-1->14->15 [2] 15/-1/-1->14->13 [3] 13/-1/-1->14->15 [4] 15/-1/-1->14->13 [5] -1/-1/-1->14->15 [6] 15/-1/-1->14->23 [7] 13/-1/-1->14->15 [8] 15/-1/-1->14->13 [9] 13/-1/-1->14->15 [10] 15/-1/-1->14->13 [11] 13/-1/-1->14->15 [12] 15/-1/-1->14->13 [13] -1/-1/-1->14->15 [14] 15/6/-1->14->30 [15] 13/23/-1->14->15 +n111-046-078:3318:3374 [5] NCCL INFO Ring 14 : 14 -> 13 -> 12 +n111-046-078:3318:3374 [5] NCCL INFO Ring 15 : 12 -> 13 -> 14 +n111-046-078:3319:3371 [6] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n111-046-078:3318:3374 [5] NCCL INFO Trees [0] 14/-1/-1->13->12 [1] 12/-1/-1->13->14 [2] 14/-1/-1->13->12 [3] 12/-1/-1->13->14 [4] 14/-1/-1->13->12 [5] 12/-1/-1->13->20 [6] -1/-1/-1->13->12 [7] 12/-1/-1->13->14 [8] 14/-1/-1->13->12 [9] 12/-1/-1->13->14 [10] 14/-1/-1->13->12 [11] 12/-1/-1->13->14 [12] 14/20/-1->13->12 [13] 12/5/-1->13->29 [14] -1/-1/-1->13->12 [15] 12/-1/-1->13->14 +n111-046-078:3319:3371 [6] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n111-046-078:3319:3371 [6] NCCL INFO P2P Chunksize set to 524288 +n111-046-078:3317:3375 [4] NCCL INFO Tree 4 : 21 -> 12 -> 13/-1/-1 +n111-046-078:3318:3374 [5] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n111-046-078:3317:3375 [4] NCCL INFO Tree 12 : 28 -> 12 -> 13/4/-1 +n111-046-078:3316:3369 [3] NCCL INFO comm 0xcc63a70 rank 11 nRanks 32 nNodes 4 localRanks 8 localRank 3 MNNVL 0 +n111-046-078:3318:3374 [5] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n111-046-078:3317:3375 [4] NCCL INFO Tree 5 : 13 -> 12 -> 11/-1/-1 +n111-046-078:3317:3375 [4] NCCL INFO Tree 13 : 13 -> 12 -> 11/21/-1 +n111-046-078:3318:3374 [5] NCCL INFO P2P Chunksize set to 524288 +n111-046-078:3317:3375 [4] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n111-046-078:3315:3376 [2] NCCL INFO comm 0xb4b5590 rank 10 nRanks 32 nNodes 4 localRanks 8 localRank 2 MNNVL 0 +n111-046-078:3317:3375 [4] NCCL INFO Ring 00 : 13 -> 12 -> 11 +n111-046-078:3317:3375 [4] NCCL INFO Ring 01 : 11 -> 12 -> 13 +n111-046-078:3317:3375 [4] NCCL INFO Ring 02 : 13 -> 12 -> 11 +n111-046-078:3316:3369 [3] NCCL INFO Tree 2 : 10 -> 11 -> 12/-1/-1 +n111-046-078:3317:3375 [4] NCCL INFO Ring 03 : 11 -> 12 -> 13 +n111-046-078:3315:3376 [2] NCCL INFO Tree 2 : 19 -> 10 -> 11/-1/-1 +n111-046-078:3317:3375 [4] NCCL INFO Ring 04 : 5 -> 12 -> 11 +n111-046-078:3316:3369 [3] NCCL INFO Tree 10 : 10 -> 11 -> 12/18/-1 +n111-046-078:3315:3376 [2] NCCL INFO Tree 10 : 26 -> 10 -> 11/2/-1 +n111-046-078:3317:3375 [4] NCCL INFO Ring 05 : 11 -> 12 -> 21 +n111-046-078:3315:3376 [2] NCCL INFO Tree 3 : 11 -> 10 -> 9/-1/-1 +n111-046-078:3316:3369 [3] NCCL INFO Tree 3 : 18 -> 11 -> 10/-1/-1 +n111-046-078:3317:3375 [4] NCCL INFO Ring 06 : 13 -> 12 -> 11 +n111-046-078:3315:3376 [2] NCCL INFO Tree 11 : 11 -> 10 -> 9/19/-1 +n111-046-078:3316:3369 [3] NCCL INFO Tree 11 : 27 -> 11 -> 10/3/-1 +n111-046-078:3317:3375 [4] NCCL INFO Ring 07 : 11 -> 12 -> 13 +n111-046-078:3317:3375 [4] NCCL INFO Ring 08 : 13 -> 12 -> 11 +n111-046-078:3317:3375 [4] NCCL INFO Ring 09 : 11 -> 12 -> 13 +n111-046-078:3315:3376 [2] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n111-046-078:3317:3375 [4] NCCL INFO Ring 10 : 13 -> 12 -> 11 +n111-046-078:3317:3375 [4] NCCL INFO Ring 11 : 11 -> 12 -> 13 +n111-046-078:3314:3373 [1] NCCL INFO comm 0xc823c10 rank 9 nRanks 32 nNodes 4 localRanks 8 localRank 1 MNNVL 0 +n111-046-078:3316:3369 [3] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n111-046-078:3317:3375 [4] NCCL INFO Ring 12 : 5 -> 12 -> 11 +n111-046-078:3317:3375 [4] NCCL INFO Ring 13 : 11 -> 12 -> 21 +n111-046-078:3317:3375 [4] NCCL INFO Ring 14 : 13 -> 12 -> 11 +n111-046-078:3317:3375 [4] NCCL INFO Ring 15 : 11 -> 12 -> 13 +n111-046-078:3317:3375 [4] NCCL INFO Trees [0] 13/-1/-1->12->11 [1] 11/-1/-1->12->13 [2] 13/-1/-1->12->11 [3] -1/-1/-1->12->13 [4] 13/-1/-1->12->21 [5] 11/-1/-1->12->13 [6] 13/-1/-1->12->11 [7] 11/-1/-1->12->13 [8] 13/-1/-1->12->11 [9] 11/-1/-1->12->13 [10] 13/-1/-1->12->11 [11] -1/-1/-1->12->13 [12] 13/4/-1->12->28 [13] 11/21/-1->12->13 [14] 13/-1/-1->12->11 [15] 11/-1/-1->12->13 +n111-046-078:3317:3375 [4] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n111-046-078:3315:3376 [2] NCCL INFO Ring 00 : 11 -> 10 -> 9 +n111-046-078:3315:3376 [2] NCCL INFO Ring 01 : 9 -> 10 -> 11 +n111-046-078:3315:3376 [2] NCCL INFO Ring 02 : 3 -> 10 -> 9 +n111-046-078:3317:3375 [4] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n111-046-078:3315:3376 [2] NCCL INFO Ring 03 : 9 -> 10 -> 19 +n111-046-078:3317:3375 [4] NCCL INFO P2P Chunksize set to 524288 +n111-046-078:3315:3376 [2] NCCL INFO Ring 04 : 11 -> 10 -> 9 +n111-046-078:3313:3372 [0] NCCL INFO comm 0xba468c0 rank 8 nRanks 32 nNodes 4 localRanks 8 localRank 0 MNNVL 0 +n111-046-078:3315:3376 [2] NCCL INFO Ring 05 : 9 -> 10 -> 11 +n111-046-078:3316:3369 [3] NCCL INFO Ring 00 : 12 -> 11 -> 10 +n111-046-078:3315:3376 [2] NCCL INFO Ring 06 : 11 -> 10 -> 9 +n111-046-078:3315:3376 [2] NCCL INFO Ring 07 : 9 -> 10 -> 11 +n111-046-078:3314:3373 [1] NCCL INFO Tree 0 : 8 -> 9 -> 10/-1/-1 +n111-046-078:3316:3369 [3] NCCL INFO Ring 01 : 10 -> 11 -> 12 +n111-046-078:3314:3373 [1] NCCL INFO Tree 8 : 8 -> 9 -> 10/16/-1 +n111-046-078:3315:3376 [2] NCCL INFO Ring 08 : 11 -> 10 -> 9 +n111-046-078:3316:3369 [3] NCCL INFO Ring 02 : 12 -> 11 -> 18 +n111-046-078:3315:3376 [2] NCCL INFO Ring 09 : 9 -> 10 -> 11 +n111-046-078:3314:3373 [1] NCCL INFO Tree 1 : 16 -> 9 -> 8/-1/-1 +n111-046-078:3316:3369 [3] NCCL INFO Ring 03 : 2 -> 11 -> 12 +n111-046-078:3315:3376 [2] NCCL INFO Ring 10 : 3 -> 10 -> 9 +n111-046-078:3314:3373 [1] NCCL INFO Tree 9 : 25 -> 9 -> 8/1/-1 +n111-046-078:3315:3376 [2] NCCL INFO Ring 11 : 9 -> 10 -> 19 +n111-046-078:3316:3369 [3] NCCL INFO Ring 04 : 12 -> 11 -> 10 +n111-046-078:3315:3376 [2] NCCL INFO Ring 12 : 11 -> 10 -> 9 +n111-046-078:3316:3369 [3] NCCL INFO Ring 05 : 10 -> 11 -> 12 +n111-046-078:3315:3376 [2] NCCL INFO Ring 13 : 9 -> 10 -> 11 +n111-046-078:3316:3369 [3] NCCL INFO Ring 06 : 12 -> 11 -> 10 +n111-046-078:3314:3373 [1] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n111-046-078:3315:3376 [2] NCCL INFO Ring 14 : 11 -> 10 -> 9 +n111-046-078:3316:3369 [3] NCCL INFO Ring 07 : 10 -> 11 -> 12 +n111-046-078:3315:3376 [2] NCCL INFO Ring 15 : 9 -> 10 -> 11 +n111-046-078:3313:3372 [0] NCCL INFO Tree 0 : 17 -> 8 -> 9/-1/-1 +n111-046-078:3316:3369 [3] NCCL INFO Ring 08 : 12 -> 11 -> 10 +n111-046-078:3315:3376 [2] NCCL INFO Trees [0] 11/-1/-1->10->9 [1] -1/-1/-1->10->11 [2] 11/-1/-1->10->19 [3] 9/-1/-1->10->11 [4] 11/-1/-1->10->9 [5] 9/-1/-1->10->11 [6] 11/-1/-1->10->9 [7] 9/-1/-1->10->11 [8] 11/-1/-1->10->9 [9] -1/-1/-1->10->11 [10] 11/2/-1->10->26 [11] 9/19/-1->10->11 [12] 11/-1/-1->10->9 [13] 9/-1/-1->10->11 [14] 11/-1/-1->10->9 [15] 9/-1/-1->10->11 +n111-046-078:3313:3372 [0] NCCL INFO Tree 8 : 24 -> 8 -> 9/0/-1 +n111-046-078:3316:3369 [3] NCCL INFO Ring 09 : 10 -> 11 -> 12 +n111-046-078:3313:3372 [0] NCCL INFO Tree 1 : 9 -> 8 -> 15/-1/-1 +n111-046-078:3313:3372 [0] NCCL INFO Tree 9 : 9 -> 8 -> 15/17/-1 +n111-046-078:3315:3376 [2] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n111-046-078:3316:3369 [3] NCCL INFO Ring 10 : 12 -> 11 -> 18 +n111-046-078:3315:3376 [2] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n111-046-078:3316:3369 [3] NCCL INFO Ring 11 : 2 -> 11 -> 12 +n111-046-078:3315:3376 [2] NCCL INFO P2P Chunksize set to 524288 +n111-046-078:3313:3372 [0] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n111-046-078:3314:3373 [1] NCCL INFO Ring 00 : 10 -> 9 -> 16 +n111-046-078:3316:3369 [3] NCCL INFO Ring 12 : 12 -> 11 -> 10 +n111-046-078:3314:3373 [1] NCCL INFO Ring 01 : 0 -> 9 -> 10 +n111-046-078:3316:3369 [3] NCCL INFO Ring 13 : 10 -> 11 -> 12 +n111-046-078:3314:3373 [1] NCCL INFO Ring 02 : 10 -> 9 -> 8 +n111-046-078:3316:3369 [3] NCCL INFO Ring 14 : 12 -> 11 -> 10 +n111-046-078:3314:3373 [1] NCCL INFO Ring 03 : 8 -> 9 -> 10 +n111-046-078:3316:3369 [3] NCCL INFO Ring 15 : 10 -> 11 -> 12 +n111-046-078:3316:3369 [3] NCCL INFO Trees [0] 12/-1/-1->11->10 [1] 10/-1/-1->11->12 [2] 12/-1/-1->11->10 [3] 10/-1/-1->11->18 [4] -1/-1/-1->11->10 [5] 10/-1/-1->11->12 [6] 12/-1/-1->11->10 [7] 10/-1/-1->11->12 [8] 12/-1/-1->11->10 [9] 10/-1/-1->11->12 [10] 12/18/-1->11->10 [11] 10/3/-1->11->27 [12] -1/-1/-1->11->10 [13] 10/-1/-1->11->12 [14] 12/-1/-1->11->10 [15] 10/-1/-1->11->12 +n111-046-078:3314:3373 [1] NCCL INFO Ring 04 : 10 -> 9 -> 8 +n111-046-078:3314:3373 [1] NCCL INFO Ring 05 : 8 -> 9 -> 10 +n111-046-078:3316:3369 [3] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n111-046-078:3313:3372 [0] NCCL INFO Ring 00 : 1 -> 8 -> 15 +n111-046-078:3314:3373 [1] NCCL INFO Ring 06 : 10 -> 9 -> 8 +n111-046-078:3316:3369 [3] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n111-046-078:3313:3372 [0] NCCL INFO Ring 01 : 15 -> 8 -> 17 +n111-046-078:3314:3373 [1] NCCL INFO Ring 07 : 8 -> 9 -> 10 +n111-046-078:3316:3369 [3] NCCL INFO P2P Chunksize set to 524288 +n111-046-078:3313:3372 [0] NCCL INFO Ring 02 : 9 -> 8 -> 15 +n111-046-078:3314:3373 [1] NCCL INFO Ring 08 : 10 -> 9 -> 16 +n111-046-078:3313:3372 [0] NCCL INFO Ring 03 : 15 -> 8 -> 9 +n111-046-078:3314:3373 [1] NCCL INFO Ring 09 : 0 -> 9 -> 10 +n111-046-078:3313:3372 [0] NCCL INFO Ring 04 : 9 -> 8 -> 15 +n111-046-078:3314:3373 [1] NCCL INFO Ring 10 : 10 -> 9 -> 8 +n111-046-078:3313:3372 [0] NCCL INFO Ring 05 : 15 -> 8 -> 9 +n111-046-078:3314:3373 [1] NCCL INFO Ring 11 : 8 -> 9 -> 10 +n111-046-078:3313:3372 [0] NCCL INFO Ring 06 : 9 -> 8 -> 15 +n111-046-078:3314:3373 [1] NCCL INFO Ring 12 : 10 -> 9 -> 8 +n111-046-078:3313:3372 [0] NCCL INFO Ring 07 : 15 -> 8 -> 9 +n111-046-078:3314:3373 [1] NCCL INFO Ring 13 : 8 -> 9 -> 10 +n111-046-078:3313:3372 [0] NCCL INFO Ring 08 : 1 -> 8 -> 15 +n111-046-078:3314:3373 [1] NCCL INFO Ring 14 : 10 -> 9 -> 8 +n111-046-078:3313:3372 [0] NCCL INFO Ring 09 : 15 -> 8 -> 17 +n111-046-078:3314:3373 [1] NCCL INFO Ring 15 : 8 -> 9 -> 10 +n111-046-078:3313:3372 [0] NCCL INFO Ring 10 : 9 -> 8 -> 15 +n111-046-078:3314:3373 [1] NCCL INFO Trees [0] 10/-1/-1->9->8 [1] 8/-1/-1->9->16 [2] -1/-1/-1->9->8 [3] 8/-1/-1->9->10 [4] 10/-1/-1->9->8 [5] 8/-1/-1->9->10 [6] 10/-1/-1->9->8 [7] 8/-1/-1->9->10 [8] 10/16/-1->9->8 [9] 8/1/-1->9->25 [10] -1/-1/-1->9->8 [11] 8/-1/-1->9->10 [12] 10/-1/-1->9->8 [13] 8/-1/-1->9->10 [14] 10/-1/-1->9->8 [15] 8/-1/-1->9->10 +n111-046-078:3313:3372 [0] NCCL INFO Ring 11 : 15 -> 8 -> 9 +n111-046-078:3314:3373 [1] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n111-046-078:3313:3372 [0] NCCL INFO Ring 12 : 9 -> 8 -> 15 +n111-046-078:3314:3373 [1] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n111-046-078:3313:3372 [0] NCCL INFO Ring 13 : 15 -> 8 -> 9 +n111-046-078:3314:3373 [1] NCCL INFO P2P Chunksize set to 524288 +n111-046-078:3313:3372 [0] NCCL INFO Ring 14 : 9 -> 8 -> 15 +n111-046-078:3313:3372 [0] NCCL INFO Ring 15 : 15 -> 8 -> 9 +n111-046-078:3313:3372 [0] NCCL INFO Trees [0] 9/-1/-1->8->17 [1] 15/-1/-1->8->9 [2] 9/-1/-1->8->15 [3] 15/-1/-1->8->9 [4] 9/-1/-1->8->15 [5] 15/-1/-1->8->9 [6] 9/-1/-1->8->15 [7] -1/-1/-1->8->9 [8] 9/0/-1->8->24 [9] 15/17/-1->8->9 [10] 9/-1/-1->8->15 [11] 15/-1/-1->8->9 [12] 9/-1/-1->8->15 [13] 15/-1/-1->8->9 [14] 9/-1/-1->8->15 [15] -1/-1/-1->8->9 +n111-046-078:3313:3372 [0] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n111-046-078:3313:3372 [0] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n111-046-078:3313:3372 [0] NCCL INFO P2P Chunksize set to 524288 +n111-046-078:3318:3411 [5] NCCL INFO [Proxy Service] Device 5 CPU core 79 +n111-046-078:3320:3409 [7] NCCL INFO [Proxy Service] Device 7 CPU core 205 +n111-046-078:3318:3414 [5] NCCL INFO [Proxy Service UDS] Device 5 CPU core 186 +n111-046-078:3320:3413 [7] NCCL INFO [Proxy Service UDS] Device 7 CPU core 78 +n111-046-078:3319:3410 [6] NCCL INFO [Proxy Service] Device 6 CPU core 78 +n111-046-078:3317:3416 [4] NCCL INFO [Proxy Service UDS] Device 4 CPU core 94 +n111-046-078:3319:3412 [6] NCCL INFO [Proxy Service UDS] Device 6 CPU core 157 +n111-046-078:3317:3415 [4] NCCL INFO [Proxy Service] Device 4 CPU core 188 +n111-046-078:3315:3417 [2] NCCL INFO [Proxy Service] Device 2 CPU core 13 +n111-046-078:3315:3418 [2] NCCL INFO [Proxy Service UDS] Device 2 CPU core 24 +n111-046-078:3314:3421 [1] NCCL INFO [Proxy Service UDS] Device 1 CPU core 18 +n111-046-078:3316:3424 [3] NCCL INFO [Proxy Service UDS] Device 3 CPU core 148 +n111-046-078:3313:3420 [0] NCCL INFO [Proxy Service] Device 0 CPU core 29 +n111-046-078:3314:3419 [1] NCCL INFO [Proxy Service] Device 1 CPU core 133 +n111-046-078:3316:3423 [3] NCCL INFO [Proxy Service] Device 3 CPU core 141 +n111-046-078:3313:3422 [0] NCCL INFO [Proxy Service UDS] Device 0 CPU core 131 +n111-046-078:3318:3374 [5] NCCL INFO NCCL_PROTO set by environment to Simple +n111-046-078:3318:3374 [5] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n111-046-078:3318:3374 [5] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512 +n111-046-078:3318:3374 [5] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 2 p2p channels per peer +n111-046-078:3313:3372 [0] NCCL INFO NCCL_PROTO set by environment to Simple +n111-046-078:3313:3372 [0] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n111-046-078:3313:3372 [0] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512 +n111-046-078:3313:3372 [0] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 2 p2p channels per peer +n111-046-078:3314:3373 [1] NCCL INFO NCCL_PROTO set by environment to Simple +n111-046-078:3314:3373 [1] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n111-046-078:3314:3373 [1] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512 +n111-046-078:3314:3373 [1] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 2 p2p channels per peer +n111-046-078:3315:3376 [2] NCCL INFO NCCL_PROTO set by environment to Simple +n111-046-078:3315:3376 [2] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n111-046-078:3315:3376 [2] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512 +n111-046-078:3315:3376 [2] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 2 p2p channels per peer +n111-046-078:3319:3371 [6] NCCL INFO NCCL_PROTO set by environment to Simple +n111-046-078:3319:3371 [6] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n111-046-078:3319:3371 [6] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512 +n111-046-078:3319:3371 [6] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 2 p2p channels per peer +n111-046-078:3316:3369 [3] NCCL INFO NCCL_PROTO set by environment to Simple +n111-046-078:3316:3369 [3] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n111-046-078:3316:3369 [3] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512 +n111-046-078:3316:3369 [3] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 2 p2p channels per peer +n111-046-078:3317:3375 [4] NCCL INFO NCCL_PROTO set by environment to Simple +n111-046-078:3317:3375 [4] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n111-046-078:3317:3375 [4] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512 +n111-046-078:3317:3375 [4] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 2 p2p channels per peer +n111-046-078:3320:3370 [7] NCCL INFO NCCL_PROTO set by environment to Simple +n111-046-078:3320:3370 [7] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n111-046-078:3320:3370 [7] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512 +n111-046-078:3320:3370 [7] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 2 p2p channels per peer +n111-046-078:3319:3371 [6] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n111-046-078:3319:3371 [6] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n111-046-078:3319:3371 [6] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n111-046-078:3319:3371 [6] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n111-046-078:3319:3371 [6] NCCL INFO ncclCommInitRankConfig comm 0xb9c26e0 rank 14 nranks 32 cudaDev 6 nvmlDev 6 busId 8b000 commId 0x2a4d35ea441cbcba - Init COMPLETE +n111-046-078:3319:3371 [6] NCCL INFO Init timings - ncclCommInitRankConfig: rank 14 nranks 32 total 6.55 (kernels 0.68, alloc 3.59, bootstrap 0.67, allgathers 0.21, topo 0.66, graphs 0.08, connections 0.05, rest 0.61) +n111-046-078:3314:3373 [1] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n111-046-078:3314:3373 [1] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n111-046-078:3314:3373 [1] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n111-046-078:3314:3373 [1] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n111-046-078:3314:3373 [1] NCCL INFO ncclCommInitRankConfig comm 0xc823c10 rank 9 nranks 32 cudaDev 1 nvmlDev 1 busId 5000 commId 0x2a4d35ea441cbcba - Init COMPLETE +n111-046-078:3314:3373 [1] NCCL INFO Init timings - ncclCommInitRankConfig: rank 9 nranks 32 total 6.53 (kernels 0.65, alloc 3.58, bootstrap 0.69, allgathers 0.19, topo 0.66, graphs 0.10, connections 0.03, rest 0.62) +n111-046-078:3317:3375 [4] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n111-046-078:3317:3375 [4] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n111-046-078:3317:3375 [4] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n111-046-078:3317:3375 [4] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n111-046-078:3317:3375 [4] NCCL INFO ncclCommInitRankConfig comm 0xc1a21c0 rank 12 nranks 32 cudaDev 4 nvmlDev 4 busId 84000 commId 0x2a4d35ea441cbcba - Init COMPLETE +n111-046-078:3317:3375 [4] NCCL INFO Init timings - ncclCommInitRankConfig: rank 12 nranks 32 total 6.54 (kernels 0.65, alloc 3.62, bootstrap 0.65, allgathers 0.21, topo 0.66, graphs 0.08, connections 0.05, rest 0.62) +n111-046-078:3319:3425 [6] NCCL INFO Channel 00/0 : 14[6] -> 15[7] via P2P/CUMEM +n111-046-078:3319:3425 [6] NCCL INFO Channel 01/0 : 14[6] -> 15[7] via P2P/CUMEM +n111-046-078:3314:3426 [1] NCCL INFO Channel 00/0 : 9[1] -> 10[2] via P2P/CUMEM +n111-046-078:3319:3425 [6] NCCL INFO Channel 02/0 : 14[6] -> 15[7] via P2P/CUMEM +n111-046-078:3314:3426 [1] NCCL INFO Channel 03/0 : 9[1] -> 10[2] via P2P/CUMEM +n111-046-078:3319:3425 [6] NCCL INFO Channel 03/0 : 14[6] -> 15[7] via P2P/CUMEM +n111-046-078:3314:3426 [1] NCCL INFO Channel 04/0 : 9[1] -> 10[2] via P2P/CUMEM +n111-046-078:3319:3425 [6] NCCL INFO Channel 04/0 : 14[6] -> 15[7] via P2P/CUMEM +n111-046-078:3314:3426 [1] NCCL INFO Channel 05/0 : 9[1] -> 10[2] via P2P/CUMEM +n111-046-078:3319:3425 [6] NCCL INFO Channel 05/0 : 14[6] -> 15[7] via P2P/CUMEM +n111-046-078:3314:3426 [1] NCCL INFO Channel 06/0 : 9[1] -> 10[2] via P2P/CUMEM +n111-046-078:3319:3425 [6] NCCL INFO Channel 06/0 : 14[6] -> 15[7] via P2P/CUMEM +n111-046-078:3314:3426 [1] NCCL INFO Channel 07/0 : 9[1] -> 10[2] via P2P/CUMEM +n111-046-078:3319:3425 [6] NCCL INFO Channel 07/0 : 14[6] -> 15[7] via P2P/CUMEM +n111-046-078:3314:3426 [1] NCCL INFO Channel 08/0 : 9[1] -> 10[2] via P2P/CUMEM +n111-046-078:3319:3425 [6] NCCL INFO Channel 08/0 : 14[6] -> 15[7] via P2P/CUMEM +n111-046-078:3314:3426 [1] NCCL INFO Channel 11/0 : 9[1] -> 10[2] via P2P/CUMEM +n111-046-078:3319:3425 [6] NCCL INFO Channel 09/0 : 14[6] -> 15[7] via P2P/CUMEM +n111-046-078:3314:3426 [1] NCCL INFO Channel 12/0 : 9[1] -> 10[2] via P2P/CUMEM +n111-046-078:3319:3425 [6] NCCL INFO Channel 10/0 : 14[6] -> 15[7] via P2P/CUMEM +n111-046-078:3314:3426 [1] NCCL INFO Channel 13/0 : 9[1] -> 10[2] via P2P/CUMEM +n111-046-078:3319:3425 [6] NCCL INFO Channel 11/0 : 14[6] -> 15[7] via P2P/CUMEM +n111-046-078:3314:3426 [1] NCCL INFO Channel 14/0 : 9[1] -> 10[2] via P2P/CUMEM +n111-046-078:3319:3425 [6] NCCL INFO Channel 12/0 : 14[6] -> 15[7] via P2P/CUMEM +n111-046-078:3314:3426 [1] NCCL INFO Channel 15/0 : 9[1] -> 10[2] via P2P/CUMEM +n111-046-078:3319:3425 [6] NCCL INFO Channel 13/0 : 14[6] -> 15[7] via P2P/CUMEM +n111-046-078:3319:3425 [6] NCCL INFO Channel 14/0 : 14[6] -> 15[7] via P2P/CUMEM +n111-046-078:3319:3425 [6] NCCL INFO Channel 15/0 : 14[6] -> 15[7] via P2P/CUMEM +n111-046-078:3316:3369 [3] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n111-046-078:3316:3369 [3] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n111-046-078:3316:3369 [3] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n111-046-078:3316:3369 [3] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n111-046-078:3316:3369 [3] NCCL INFO ncclCommInitRankConfig comm 0xcc63a70 rank 11 nranks 32 cudaDev 3 nvmlDev 3 busId c000 commId 0x2a4d35ea441cbcba - Init COMPLETE +n111-046-078:3316:3369 [3] NCCL INFO Init timings - ncclCommInitRankConfig: rank 11 nranks 32 total 6.59 (kernels 0.69, alloc 3.64, bootstrap 0.62, allgathers 0.17, topo 0.66, graphs 0.12, connections 0.05, rest 0.63) +n111-046-078:3317:3427 [4] NCCL INFO Channel 00/0 : 12[4] -> 13[5] via P2P/CUMEM +n111-046-078:3317:3427 [4] NCCL INFO Channel 01/0 : 12[4] -> 13[5] via P2P/CUMEM +n111-046-078:3317:3427 [4] NCCL INFO Channel 02/0 : 12[4] -> 13[5] via P2P/CUMEM +n111-046-078:3317:3427 [4] NCCL INFO Channel 03/0 : 12[4] -> 13[5] via P2P/CUMEM +n111-046-078:3317:3427 [4] NCCL INFO Channel 04/0 : 12[4] -> 13[5] via P2P/CUMEM +n111-046-078:3317:3427 [4] NCCL INFO Channel 05/0 : 12[4] -> 13[5] via P2P/CUMEM +n111-046-078:3317:3427 [4] NCCL INFO Channel 06/0 : 12[4] -> 13[5] via P2P/CUMEM +n111-046-078:3317:3427 [4] NCCL INFO Channel 07/0 : 12[4] -> 13[5] via P2P/CUMEM +n111-046-078:3317:3427 [4] NCCL INFO Channel 08/0 : 12[4] -> 13[5] via P2P/CUMEM +n111-046-078:3317:3427 [4] NCCL INFO Channel 09/0 : 12[4] -> 13[5] via P2P/CUMEM +n111-046-078:3317:3427 [4] NCCL INFO Channel 10/0 : 12[4] -> 13[5] via P2P/CUMEM +n111-046-078:3317:3427 [4] NCCL INFO Channel 11/0 : 12[4] -> 13[5] via P2P/CUMEM +n111-046-078:3317:3427 [4] NCCL INFO Channel 12/0 : 12[4] -> 13[5] via P2P/CUMEM +n111-046-078:3317:3427 [4] NCCL INFO Channel 13/0 : 12[4] -> 13[5] via P2P/CUMEM +n111-046-078:3317:3427 [4] NCCL INFO Channel 14/0 : 12[4] -> 13[5] via P2P/CUMEM +n111-046-078:3317:3427 [4] NCCL INFO Channel 15/0 : 12[4] -> 13[5] via P2P/CUMEM +n111-046-078:3318:3374 [5] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n111-046-078:3318:3374 [5] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n111-046-078:3318:3374 [5] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n111-046-078:3318:3374 [5] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n111-046-078:3318:3374 [5] NCCL INFO ncclCommInitRankConfig comm 0xb26c190 rank 13 nranks 32 cudaDev 5 nvmlDev 5 busId 85000 commId 0x2a4d35ea441cbcba - Init COMPLETE +n111-046-078:3318:3374 [5] NCCL INFO Init timings - ncclCommInitRankConfig: rank 13 nranks 32 total 6.55 (kernels 0.66, alloc 3.63, bootstrap 0.62, allgathers 0.21, topo 0.66, graphs 0.08, connections 0.03, rest 0.65) +n111-046-078:3316:3428 [3] NCCL INFO Channel 00/0 : 11[3] -> 12[4] via P2P/CUMEM +n111-046-078:3316:3428 [3] NCCL INFO Channel 01/0 : 11[3] -> 12[4] via P2P/CUMEM +n111-046-078:3316:3428 [3] NCCL INFO Channel 02/0 : 11[3] -> 12[4] via P2P/CUMEM +n111-046-078:3316:3428 [3] NCCL INFO Channel 05/0 : 11[3] -> 12[4] via P2P/CUMEM +n111-046-078:3316:3428 [3] NCCL INFO Channel 06/0 : 11[3] -> 12[4] via P2P/CUMEM +n111-046-078:3320:3370 [7] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n111-046-078:3320:3370 [7] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n111-046-078:3320:3370 [7] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n111-046-078:3320:3370 [7] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n111-046-078:3320:3370 [7] NCCL INFO ncclCommInitRankConfig comm 0xbdf0f50 rank 15 nranks 32 cudaDev 7 nvmlDev 7 busId 8c000 commId 0x2a4d35ea441cbcba - Init COMPLETE +n111-046-078:3320:3370 [7] NCCL INFO Init timings - ncclCommInitRankConfig: rank 15 nranks 32 total 6.60 (kernels 0.72, alloc 3.61, bootstrap 0.62, allgathers 0.21, topo 0.66, graphs 0.08, connections 0.05, rest 0.64) +n111-046-078:3316:3428 [3] NCCL INFO Channel 07/0 : 11[3] -> 12[4] via P2P/CUMEM +n111-046-078:3316:3428 [3] NCCL INFO Channel 08/0 : 11[3] -> 12[4] via P2P/CUMEM +n111-046-078:3316:3428 [3] NCCL INFO Channel 09/0 : 11[3] -> 12[4] via P2P/CUMEM +n111-046-078:3316:3428 [3] NCCL INFO Channel 10/0 : 11[3] -> 12[4] via P2P/CUMEM +n111-046-078:3316:3428 [3] NCCL INFO Channel 13/0 : 11[3] -> 12[4] via P2P/CUMEM +n111-046-078:3316:3428 [3] NCCL INFO Channel 14/0 : 11[3] -> 12[4] via P2P/CUMEM +n111-046-078:3316:3428 [3] NCCL INFO Channel 15/0 : 11[3] -> 12[4] via P2P/CUMEM +n111-046-078:3315:3376 [2] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n111-046-078:3315:3376 [2] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n111-046-078:3315:3376 [2] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n111-046-078:3315:3376 [2] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n111-046-078:3315:3376 [2] NCCL INFO ncclCommInitRankConfig comm 0xb4b5590 rank 10 nranks 32 cudaDev 2 nvmlDev 2 busId b000 commId 0x2a4d35ea441cbcba - Init COMPLETE +n111-046-078:3315:3376 [2] NCCL INFO Init timings - ncclCommInitRankConfig: rank 10 nranks 32 total 6.56 (kernels 0.64, alloc 3.65, bootstrap 0.62, allgathers 0.22, topo 0.66, graphs 0.08, connections 0.04, rest 0.65) +n111-046-078:3318:3429 [5] NCCL INFO Channel 00/0 : 13[5] -> 14[6] via P2P/CUMEM +n111-046-078:3318:3429 [5] NCCL INFO Channel 01/0 : 13[5] -> 14[6] via P2P/CUMEM +n111-046-078:3313:3372 [0] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n111-046-078:3313:3372 [0] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n111-046-078:3313:3372 [0] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n111-046-078:3313:3372 [0] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n111-046-078:3313:3372 [0] NCCL INFO ncclCommInitRankConfig comm 0xba468c0 rank 8 nranks 32 cudaDev 0 nvmlDev 0 busId 4000 commId 0x2a4d35ea441cbcba - Init COMPLETE +n111-046-078:3313:3372 [0] NCCL INFO Init timings - ncclCommInitRankConfig: rank 8 nranks 32 total 6.58 (kernels 0.67, alloc 3.63, bootstrap 0.62, allgathers 0.21, topo 0.66, graphs 0.08, connections 0.03, rest 0.66) +n111-046-078:3318:3429 [5] NCCL INFO Channel 02/0 : 13[5] -> 14[6] via P2P/CUMEM +n111-046-078:3318:3429 [5] NCCL INFO Channel 03/0 : 13[5] -> 14[6] via P2P/CUMEM +n111-046-078:3318:3429 [5] NCCL INFO Channel 04/0 : 13[5] -> 14[6] via P2P/CUMEM +n111-046-078:3318:3429 [5] NCCL INFO Channel 07/0 : 13[5] -> 14[6] via P2P/CUMEM +n111-046-078:3318:3429 [5] NCCL INFO Channel 08/0 : 13[5] -> 14[6] via P2P/CUMEM +n111-046-078:3318:3429 [5] NCCL INFO Channel 09/0 : 13[5] -> 14[6] via P2P/CUMEM +n111-046-078:3318:3429 [5] NCCL INFO Channel 10/0 : 13[5] -> 14[6] via P2P/CUMEM +n111-046-078:3318:3429 [5] NCCL INFO Channel 11/0 : 13[5] -> 14[6] via P2P/CUMEM +n111-046-078:3318:3429 [5] NCCL INFO Channel 12/0 : 13[5] -> 14[6] via P2P/CUMEM +n111-046-078:3318:3429 [5] NCCL INFO Channel 15/0 : 13[5] -> 14[6] via P2P/CUMEM +n111-046-078:3313:3432 [0] NCCL INFO Channel 00/0 : 8[0] -> 9[1] via P2P/CUMEM +n111-046-078:3313:3432 [0] NCCL INFO Channel 01/0 : 8[0] -> 9[1] via P2P/CUMEM +n111-046-078:3313:3432 [0] NCCL INFO Channel 02/0 : 8[0] -> 9[1] via P2P/CUMEM +n111-046-078:3315:3431 [2] NCCL INFO Channel 00/0 : 10[2] -> 11[3] via P2P/CUMEM +n111-046-078:3313:3432 [0] NCCL INFO Channel 03/0 : 8[0] -> 9[1] via P2P/CUMEM +n111-046-078:3315:3431 [2] NCCL INFO Channel 01/0 : 10[2] -> 11[3] via P2P/CUMEM +n111-046-078:3313:3432 [0] NCCL INFO Channel 04/0 : 8[0] -> 9[1] via P2P/CUMEM +n111-046-078:3315:3431 [2] NCCL INFO Channel 02/0 : 10[2] -> 11[3] via P2P/CUMEM +n111-046-078:3315:3431 [2] NCCL INFO Channel 03/0 : 10[2] -> 11[3] via P2P/CUMEM +n111-046-078:3313:3432 [0] NCCL INFO Channel 05/0 : 8[0] -> 9[1] via P2P/CUMEM +n111-046-078:3315:3431 [2] NCCL INFO Channel 04/0 : 10[2] -> 11[3] via P2P/CUMEM +n111-046-078:3313:3432 [0] NCCL INFO Channel 06/0 : 8[0] -> 9[1] via P2P/CUMEM +n111-046-078:3315:3431 [2] NCCL INFO Channel 05/0 : 10[2] -> 11[3] via P2P/CUMEM +n111-046-078:3313:3432 [0] NCCL INFO Channel 07/0 : 8[0] -> 9[1] via P2P/CUMEM +n111-046-078:3315:3431 [2] NCCL INFO Channel 06/0 : 10[2] -> 11[3] via P2P/CUMEM +n111-046-078:3313:3432 [0] NCCL INFO Channel 08/0 : 8[0] -> 9[1] via P2P/CUMEM +n111-046-078:3315:3431 [2] NCCL INFO Channel 07/0 : 10[2] -> 11[3] via P2P/CUMEM +n111-046-078:3313:3432 [0] NCCL INFO Channel 09/0 : 8[0] -> 9[1] via P2P/CUMEM +n111-046-078:3315:3431 [2] NCCL INFO Channel 08/0 : 10[2] -> 11[3] via P2P/CUMEM +n111-046-078:3313:3432 [0] NCCL INFO Channel 10/0 : 8[0] -> 9[1] via P2P/CUMEM +n111-046-078:3315:3431 [2] NCCL INFO Channel 09/0 : 10[2] -> 11[3] via P2P/CUMEM +n111-046-078:3315:3431 [2] NCCL INFO Channel 10/0 : 10[2] -> 11[3] via P2P/CUMEM +n111-046-078:3313:3432 [0] NCCL INFO Channel 11/0 : 8[0] -> 9[1] via P2P/CUMEM +n111-046-078:3315:3431 [2] NCCL INFO Channel 11/0 : 10[2] -> 11[3] via P2P/CUMEM +n111-046-078:3313:3432 [0] NCCL INFO Channel 12/0 : 8[0] -> 9[1] via P2P/CUMEM +n111-046-078:3315:3431 [2] NCCL INFO Channel 12/0 : 10[2] -> 11[3] via P2P/CUMEM +n111-046-078:3313:3432 [0] NCCL INFO Channel 13/0 : 8[0] -> 9[1] via P2P/CUMEM +n111-046-078:3315:3431 [2] NCCL INFO Channel 13/0 : 10[2] -> 11[3] via P2P/CUMEM +n111-046-078:3313:3432 [0] NCCL INFO Channel 14/0 : 8[0] -> 9[1] via P2P/CUMEM +n111-046-078:3315:3431 [2] NCCL INFO Channel 14/0 : 10[2] -> 11[3] via P2P/CUMEM +n111-046-078:3313:3432 [0] NCCL INFO Channel 15/0 : 8[0] -> 9[1] via P2P/CUMEM +n111-046-078:3315:3431 [2] NCCL INFO Channel 15/0 : 10[2] -> 11[3] via P2P/CUMEM +n111-046-078:3313:3432 [0] NCCL INFO Channel 01/0 : 8[0] -> 15[7] via P2P/CUMEM +n111-046-078:3313:3432 [0] NCCL INFO Channel 02/0 : 8[0] -> 15[7] via P2P/CUMEM +n111-046-078:3313:3432 [0] NCCL INFO Channel 03/0 : 8[0] -> 15[7] via P2P/CUMEM +n111-046-078:3313:3432 [0] NCCL INFO Channel 04/0 : 8[0] -> 15[7] via P2P/CUMEM +n111-046-078:3313:3432 [0] NCCL INFO Channel 05/0 : 8[0] -> 15[7] via P2P/CUMEM +n111-046-078:3313:3432 [0] NCCL INFO Channel 06/0 : 8[0] -> 15[7] via P2P/CUMEM +n111-046-078:3313:3432 [0] NCCL INFO Channel 09/0 : 8[0] -> 15[7] via P2P/CUMEM +n111-046-078:3313:3432 [0] NCCL INFO Channel 10/0 : 8[0] -> 15[7] via P2P/CUMEM +n111-046-078:3313:3432 [0] NCCL INFO Channel 11/0 : 8[0] -> 15[7] via P2P/CUMEM +n111-046-078:3313:3432 [0] NCCL INFO Channel 12/0 : 8[0] -> 15[7] via P2P/CUMEM +n111-046-078:3313:3432 [0] NCCL INFO Channel 13/0 : 8[0] -> 15[7] via P2P/CUMEM +n111-046-078:3313:3432 [0] NCCL INFO Channel 14/0 : 8[0] -> 15[7] via P2P/CUMEM +n111-046-078:3318:3433 [5] NCCL INFO [Proxy Progress] Device 5 CPU core 92 +n111-046-078:3318:3429 [5] NCCL INFO Channel 05/0 : 13[5] -> 20[4] [send] via NET/FasTrak/5/GDRDMA +n111-046-078:3318:3429 [5] NCCL INFO Channel 12/0 : 13[5] -> 20[4] [send] via NET/FasTrak/4/GDRDMA +n111-046-078:3317:3434 [4] NCCL INFO [Proxy Progress] Device 4 CPU core 166 +n111-046-078:3317:3427 [4] NCCL INFO Channel 12/0 : 4[4] -> 12[4] [receive] via NET/FasTrak/4/GDRDMA +n111-046-078:3318:3429 [5] NCCL INFO Channel 13/0 : 5[5] -> 13[5] [receive] via NET/FasTrak/5/GDRDMA +n111-046-078:3319:3435 [6] NCCL INFO [Proxy Progress] Device 6 CPU core 194 +n111-046-078:3320:3436 [7] NCCL INFO [Proxy Progress] Device 7 CPU core 171 +n111-046-078:3320:3430 [7] NCCL INFO Channel 07/0 : 15[7] -> 22[6] [send] via NET/FasTrak/7/GDRDMA +n111-046-078:3319:3425 [6] NCCL INFO Channel 14/0 : 6[6] -> 14[6] [receive] via NET/FasTrak/6/GDRDMA +n111-046-078:3320:3430 [7] NCCL INFO Channel 14/0 : 15[7] -> 22[6] [send] via NET/FasTrak/6/GDRDMA +n111-046-078:3315:3437 [2] NCCL INFO [Proxy Progress] Device 2 CPU core 28 +n111-046-078:3320:3430 [7] NCCL INFO Channel 15/0 : 7[7] -> 15[7] [receive] via NET/FasTrak/7/GDRDMA +n111-046-078:3316:3438 [3] NCCL INFO [Proxy Progress] Device 3 CPU core 47 +n111-046-078:3316:3428 [3] NCCL INFO Channel 03/0 : 11[3] -> 18[2] [send] via NET/FasTrak/3/GDRDMA +n111-046-078:3316:3428 [3] NCCL INFO Channel 10/0 : 11[3] -> 18[2] [send] via NET/FasTrak/2/GDRDMA +n111-046-078:3315:3431 [2] NCCL INFO Channel 10/0 : 2[2] -> 10[2] [receive] via NET/FasTrak/2/GDRDMA +n111-046-078:3316:3428 [3] NCCL INFO Channel 11/0 : 3[3] -> 11[3] [receive] via NET/FasTrak/3/GDRDMA +n111-046-078:3314:3439 [1] NCCL INFO [Proxy Progress] Device 1 CPU core 130 +n111-046-078:3314:3426 [1] NCCL INFO Channel 01/0 : 9[1] -> 16[0] [send] via NET/FasTrak/1/GDRDMA +n111-046-078:3314:3426 [1] NCCL INFO Channel 08/0 : 9[1] -> 16[0] [send] via NET/FasTrak/0/GDRDMA +n111-046-078:3314:3426 [1] NCCL INFO Channel 09/0 : 1[1] -> 9[1] [receive] via NET/FasTrak/1/GDRDMA +n111-046-078:3313:3440 [0] NCCL INFO [Proxy Progress] Device 0 CPU core 147 +n111-046-078:3313:3432 [0] NCCL INFO Channel 08/0 : 0[0] -> 8[0] [receive] via NET/FasTrak/0/GDRDMA +n111-046-078:3316:3428 [3] NCCL INFO Channel 11/0 : 27[3] -> 11[3] [receive] via NET/FasTrak/3/GDRDMA +n111-046-078:3316:3428 [3] NCCL INFO Channel 11/0 : 11[3] -> 27[3] [send] via NET/FasTrak/3/GDRDMA +n111-046-078:3317:3427 [4] NCCL INFO Channel 04/0 : 12[4] -> 21[5] [send] via NET/FasTrak/4/GDRDMA +n111-046-078:3317:3427 [4] NCCL INFO Channel 13/0 : 12[4] -> 21[5] [send] via NET/FasTrak/5/GDRDMA +n111-046-078:3315:3431 [2] NCCL INFO Channel 02/0 : 10[2] -> 19[3] [send] via NET/FasTrak/2/GDRDMA +n111-046-078:3315:3431 [2] NCCL INFO Channel 11/0 : 10[2] -> 19[3] [send] via NET/FasTrak/3/GDRDMA +n111-046-078:3314:3426 [1] NCCL INFO Channel 09/0 : 25[1] -> 9[1] [receive] via NET/FasTrak/1/GDRDMA +n111-046-078:3314:3426 [1] NCCL INFO Channel 09/0 : 9[1] -> 25[1] [send] via NET/FasTrak/1/GDRDMA +n111-046-078:3313:3432 [0] NCCL INFO Channel 00/0 : 8[0] -> 17[1] [send] via NET/FasTrak/0/GDRDMA +n111-046-078:3313:3432 [0] NCCL INFO Channel 09/0 : 8[0] -> 17[1] [send] via NET/FasTrak/1/GDRDMA +n111-046-078:3320:3430 [7] NCCL INFO Channel 15/0 : 31[7] -> 15[7] [receive] via NET/FasTrak/7/GDRDMA +n111-046-078:3320:3430 [7] NCCL INFO Channel 15/0 : 15[7] -> 31[7] [send] via NET/FasTrak/7/GDRDMA +n111-046-078:3319:3425 [6] NCCL INFO Channel 06/0 : 14[6] -> 23[7] [send] via NET/FasTrak/6/GDRDMA +n111-046-078:3319:3425 [6] NCCL INFO Channel 15/0 : 14[6] -> 23[7] [send] via NET/FasTrak/7/GDRDMA +n111-046-078:3318:3429 [5] NCCL INFO Channel 13/0 : 29[5] -> 13[5] [receive] via NET/FasTrak/5/GDRDMA +n111-046-078:3318:3429 [5] NCCL INFO Channel 13/0 : 13[5] -> 29[5] [send] via NET/FasTrak/5/GDRDMA +n111-046-078:3314:3426 [1] NCCL INFO Channel 09/0 : 9[1] -> 1[1] [send] via NET/FasTrak/1/GDRDMA +n111-046-078:3313:3432 [0] NCCL INFO Channel 08/0 : 24[0] -> 8[0] [receive] via NET/FasTrak/0/GDRDMA +n111-046-078:3313:3432 [0] NCCL INFO Channel 08/0 : 8[0] -> 24[0] [send] via NET/FasTrak/0/GDRDMA +n111-046-078:3314:3426 [1] NCCL INFO Channel 01/0 : 16[0] -> 9[1] [receive] via NET/FasTrak/1/GDRDMA +n111-046-078:3314:3426 [1] NCCL INFO Channel 08/0 : 16[0] -> 9[1] [receive] via NET/FasTrak/0/GDRDMA +n111-046-078:3318:3429 [5] NCCL INFO Channel 13/0 : 13[5] -> 5[5] [send] via NET/FasTrak/5/GDRDMA +n111-046-078:3317:3427 [4] NCCL INFO Channel 12/0 : 28[4] -> 12[4] [receive] via NET/FasTrak/4/GDRDMA +n111-046-078:3317:3427 [4] NCCL INFO Channel 12/0 : 12[4] -> 28[4] [send] via NET/FasTrak/4/GDRDMA +n111-046-078:3318:3429 [5] NCCL INFO Channel 05/0 : 20[4] -> 13[5] [receive] via NET/FasTrak/5/GDRDMA +n111-046-078:3318:3429 [5] NCCL INFO Channel 12/0 : 20[4] -> 13[5] [receive] via NET/FasTrak/4/GDRDMA +n111-046-078:3316:3428 [3] NCCL INFO Channel 11/0 : 11[3] -> 3[3] [send] via NET/FasTrak/3/GDRDMA +n111-046-078:3317:3427 [4] NCCL INFO Channel 04/0 : 21[5] -> 12[4] [receive] via NET/FasTrak/4/GDRDMA +n111-046-078:3315:3431 [2] NCCL INFO Channel 10/0 : 26[2] -> 10[2] [receive] via NET/FasTrak/2/GDRDMA +n111-046-078:3315:3431 [2] NCCL INFO Channel 10/0 : 10[2] -> 26[2] [send] via NET/FasTrak/2/GDRDMA +n111-046-078:3317:3427 [4] NCCL INFO Channel 13/0 : 21[5] -> 12[4] [receive] via NET/FasTrak/5/GDRDMA +n111-046-078:3315:3431 [2] NCCL INFO Channel 02/0 : 19[3] -> 10[2] [receive] via NET/FasTrak/2/GDRDMA +n111-046-078:3317:3427 [4] NCCL INFO Channel 12/0 : 12[4] -> 4[4] [send] via NET/FasTrak/4/GDRDMA +n111-046-078:3315:3431 [2] NCCL INFO Channel 11/0 : 19[3] -> 10[2] [receive] via NET/FasTrak/3/GDRDMA +n111-046-078:3315:3431 [2] NCCL INFO Channel 10/0 : 10[2] -> 2[2] [send] via NET/FasTrak/2/GDRDMA +n111-046-078:3318:3429 [5] NCCL INFO Channel 00/0 : 13[5] -> 12[4] via P2P/CUMEM +n111-046-078:3318:3429 [5] NCCL INFO Channel 01/0 : 13[5] -> 12[4] via P2P/CUMEM +n111-046-078:3316:3428 [3] NCCL INFO Channel 03/0 : 18[2] -> 11[3] [receive] via NET/FasTrak/3/GDRDMA +n111-046-078:3318:3429 [5] NCCL INFO Channel 02/0 : 13[5] -> 12[4] via P2P/CUMEM +n111-046-078:3318:3429 [5] NCCL INFO Channel 03/0 : 13[5] -> 12[4] via P2P/CUMEM +n111-046-078:3318:3429 [5] NCCL INFO Channel 04/0 : 13[5] -> 12[4] via P2P/CUMEM +n111-046-078:3318:3429 [5] NCCL INFO Channel 05/0 : 13[5] -> 12[4] via P2P/CUMEM +n111-046-078:3318:3429 [5] NCCL INFO Channel 06/0 : 13[5] -> 12[4] via P2P/CUMEM +n111-046-078:3316:3428 [3] NCCL INFO Channel 10/0 : 18[2] -> 11[3] [receive] via NET/FasTrak/2/GDRDMA +n111-046-078:3318:3429 [5] NCCL INFO Channel 07/0 : 13[5] -> 12[4] via P2P/CUMEM +n111-046-078:3318:3429 [5] NCCL INFO Channel 08/0 : 13[5] -> 12[4] via P2P/CUMEM +n111-046-078:3318:3429 [5] NCCL INFO Channel 09/0 : 13[5] -> 12[4] via P2P/CUMEM +n111-046-078:3318:3429 [5] NCCL INFO Channel 10/0 : 13[5] -> 12[4] via P2P/CUMEM +n111-046-078:3318:3429 [5] NCCL INFO Channel 11/0 : 13[5] -> 12[4] via P2P/CUMEM +n111-046-078:3318:3429 [5] NCCL INFO Channel 12/0 : 13[5] -> 12[4] via P2P/CUMEM +n111-046-078:3315:3431 [2] NCCL INFO Channel 00/0 : 10[2] -> 9[1] via P2P/CUMEM +n111-046-078:3318:3429 [5] NCCL INFO Channel 13/0 : 13[5] -> 12[4] via P2P/CUMEM +n111-046-078:3315:3431 [2] NCCL INFO Channel 03/0 : 10[2] -> 9[1] via P2P/CUMEM +n111-046-078:3318:3429 [5] NCCL INFO Channel 14/0 : 13[5] -> 12[4] via P2P/CUMEM +n111-046-078:3315:3431 [2] NCCL INFO Channel 04/0 : 10[2] -> 9[1] via P2P/CUMEM +n111-046-078:3318:3429 [5] NCCL INFO Channel 15/0 : 13[5] -> 12[4] via P2P/CUMEM +n111-046-078:3317:3427 [4] NCCL INFO Channel 00/0 : 12[4] -> 11[3] via P2P/CUMEM +n111-046-078:3315:3431 [2] NCCL INFO Channel 05/0 : 10[2] -> 9[1] via P2P/CUMEM +n111-046-078:3317:3427 [4] NCCL INFO Channel 01/0 : 12[4] -> 11[3] via P2P/CUMEM +n111-046-078:3315:3431 [2] NCCL INFO Channel 06/0 : 10[2] -> 9[1] via P2P/CUMEM +n111-046-078:3317:3427 [4] NCCL INFO Channel 02/0 : 12[4] -> 11[3] via P2P/CUMEM +n111-046-078:3315:3431 [2] NCCL INFO Channel 07/0 : 10[2] -> 9[1] via P2P/CUMEM +n111-046-078:3317:3427 [4] NCCL INFO Channel 05/0 : 12[4] -> 11[3] via P2P/CUMEM +n111-046-078:3315:3431 [2] NCCL INFO Channel 08/0 : 10[2] -> 9[1] via P2P/CUMEM +n111-046-078:3317:3427 [4] NCCL INFO Channel 06/0 : 12[4] -> 11[3] via P2P/CUMEM +n111-046-078:3315:3431 [2] NCCL INFO Channel 11/0 : 10[2] -> 9[1] via P2P/CUMEM +n111-046-078:3317:3427 [4] NCCL INFO Channel 07/0 : 12[4] -> 11[3] via P2P/CUMEM +n111-046-078:3316:3428 [3] NCCL INFO Channel 00/0 : 11[3] -> 10[2] via P2P/CUMEM +n111-046-078:3315:3431 [2] NCCL INFO Channel 12/0 : 10[2] -> 9[1] via P2P/CUMEM +n111-046-078:3317:3427 [4] NCCL INFO Channel 08/0 : 12[4] -> 11[3] via P2P/CUMEM +n111-046-078:3316:3428 [3] NCCL INFO Channel 01/0 : 11[3] -> 10[2] via P2P/CUMEM +n111-046-078:3315:3431 [2] NCCL INFO Channel 13/0 : 10[2] -> 9[1] via P2P/CUMEM +n111-046-078:3317:3427 [4] NCCL INFO Channel 09/0 : 12[4] -> 11[3] via P2P/CUMEM +n111-046-078:3316:3428 [3] NCCL INFO Channel 02/0 : 11[3] -> 10[2] via P2P/CUMEM +n111-046-078:3315:3431 [2] NCCL INFO Channel 14/0 : 10[2] -> 9[1] via P2P/CUMEM +n111-046-078:3317:3427 [4] NCCL INFO Channel 10/0 : 12[4] -> 11[3] via P2P/CUMEM +n111-046-078:3316:3428 [3] NCCL INFO Channel 03/0 : 11[3] -> 10[2] via P2P/CUMEM +n111-046-078:3315:3431 [2] NCCL INFO Channel 15/0 : 10[2] -> 9[1] via P2P/CUMEM +n111-046-078:3317:3427 [4] NCCL INFO Channel 13/0 : 12[4] -> 11[3] via P2P/CUMEM +n111-046-078:3316:3428 [3] NCCL INFO Channel 04/0 : 11[3] -> 10[2] via P2P/CUMEM +n111-046-078:3317:3427 [4] NCCL INFO Channel 14/0 : 12[4] -> 11[3] via P2P/CUMEM +n111-046-078:3316:3428 [3] NCCL INFO Channel 05/0 : 11[3] -> 10[2] via P2P/CUMEM +n111-046-078:3317:3427 [4] NCCL INFO Channel 15/0 : 12[4] -> 11[3] via P2P/CUMEM +n111-046-078:3316:3428 [3] NCCL INFO Channel 06/0 : 11[3] -> 10[2] via P2P/CUMEM +n111-046-078:3316:3428 [3] NCCL INFO Channel 07/0 : 11[3] -> 10[2] via P2P/CUMEM +n111-046-078:3316:3428 [3] NCCL INFO Channel 08/0 : 11[3] -> 10[2] via P2P/CUMEM +n111-046-078:3316:3428 [3] NCCL INFO Channel 09/0 : 11[3] -> 10[2] via P2P/CUMEM +n111-046-078:3316:3428 [3] NCCL INFO Channel 10/0 : 11[3] -> 10[2] via P2P/CUMEM +n111-046-078:3316:3428 [3] NCCL INFO Channel 11/0 : 11[3] -> 10[2] via P2P/CUMEM +n111-046-078:3316:3428 [3] NCCL INFO Channel 12/0 : 11[3] -> 10[2] via P2P/CUMEM +n111-046-078:3316:3428 [3] NCCL INFO Channel 13/0 : 11[3] -> 10[2] via P2P/CUMEM +n111-046-078:3316:3428 [3] NCCL INFO Channel 14/0 : 11[3] -> 10[2] via P2P/CUMEM +n111-046-078:3316:3428 [3] NCCL INFO Channel 15/0 : 11[3] -> 10[2] via P2P/CUMEM +n111-046-078:3320:3430 [7] NCCL INFO Channel 15/0 : 15[7] -> 7[7] [send] via NET/FasTrak/7/GDRDMA +n111-046-078:3313:3432 [0] NCCL INFO Channel 00/0 : 17[1] -> 8[0] [receive] via NET/FasTrak/0/GDRDMA +n111-046-078:3313:3432 [0] NCCL INFO Channel 09/0 : 17[1] -> 8[0] [receive] via NET/FasTrak/1/GDRDMA +n111-046-078:3313:3432 [0] NCCL INFO Channel 08/0 : 8[0] -> 0[0] [send] via NET/FasTrak/0/GDRDMA +n111-046-078:3319:3425 [6] NCCL INFO Channel 14/0 : 30[6] -> 14[6] [receive] via NET/FasTrak/6/GDRDMA +n111-046-078:3319:3425 [6] NCCL INFO Channel 14/0 : 14[6] -> 30[6] [send] via NET/FasTrak/6/GDRDMA +n111-046-078:3320:3430 [7] NCCL INFO Channel 07/0 : 22[6] -> 15[7] [receive] via NET/FasTrak/7/GDRDMA +n111-046-078:3319:3425 [6] NCCL INFO Channel 06/0 : 23[7] -> 14[6] [receive] via NET/FasTrak/6/GDRDMA +n111-046-078:3320:3430 [7] NCCL INFO Channel 14/0 : 22[6] -> 15[7] [receive] via NET/FasTrak/6/GDRDMA +n111-046-078:3320:3430 [7] NCCL INFO Channel 01/0 : 15[7] -> 8[0] via P2P/CUMEM +n111-046-078:3319:3425 [6] NCCL INFO Channel 15/0 : 23[7] -> 14[6] [receive] via NET/FasTrak/7/GDRDMA +n111-046-078:3319:3425 [6] NCCL INFO Channel 14/0 : 14[6] -> 6[6] [send] via NET/FasTrak/6/GDRDMA +n111-046-078:3320:3430 [7] NCCL INFO Channel 02/0 : 15[7] -> 8[0] via P2P/CUMEM +n111-046-078:3320:3430 [7] NCCL INFO Channel 03/0 : 15[7] -> 8[0] via P2P/CUMEM +n111-046-078:3320:3430 [7] NCCL INFO Channel 04/0 : 15[7] -> 8[0] via P2P/CUMEM +n111-046-078:3320:3430 [7] NCCL INFO Channel 05/0 : 15[7] -> 8[0] via P2P/CUMEM +n111-046-078:3320:3430 [7] NCCL INFO Channel 06/0 : 15[7] -> 8[0] via P2P/CUMEM +n111-046-078:3320:3430 [7] NCCL INFO Channel 09/0 : 15[7] -> 8[0] via P2P/CUMEM +n111-046-078:3320:3430 [7] NCCL INFO Channel 10/0 : 15[7] -> 8[0] via P2P/CUMEM +n111-046-078:3320:3430 [7] NCCL INFO Channel 11/0 : 15[7] -> 8[0] via P2P/CUMEM +n111-046-078:3320:3430 [7] NCCL INFO Channel 12/0 : 15[7] -> 8[0] via P2P/CUMEM +n111-046-078:3320:3430 [7] NCCL INFO Channel 13/0 : 15[7] -> 8[0] via P2P/CUMEM +n111-046-078:3320:3430 [7] NCCL INFO Channel 14/0 : 15[7] -> 8[0] via P2P/CUMEM +n111-046-078:3320:3430 [7] NCCL INFO Channel 00/0 : 15[7] -> 14[6] via P2P/CUMEM +n111-046-078:3319:3425 [6] NCCL INFO Channel 00/0 : 14[6] -> 13[5] via P2P/CUMEM +n111-046-078:3319:3425 [6] NCCL INFO Channel 01/0 : 14[6] -> 13[5] via P2P/CUMEM +n111-046-078:3320:3430 [7] NCCL INFO Channel 01/0 : 15[7] -> 14[6] via P2P/CUMEM +n111-046-078:3320:3430 [7] NCCL INFO Channel 02/0 : 15[7] -> 14[6] via P2P/CUMEM +n111-046-078:3320:3430 [7] NCCL INFO Channel 03/0 : 15[7] -> 14[6] via P2P/CUMEM +n111-046-078:3319:3425 [6] NCCL INFO Channel 02/0 : 14[6] -> 13[5] via P2P/CUMEM +n111-046-078:3319:3425 [6] NCCL INFO Channel 03/0 : 14[6] -> 13[5] via P2P/CUMEM +n111-046-078:3320:3430 [7] NCCL INFO Channel 04/0 : 15[7] -> 14[6] via P2P/CUMEM +n111-046-078:3319:3425 [6] NCCL INFO Channel 04/0 : 14[6] -> 13[5] via P2P/CUMEM +n111-046-078:3320:3430 [7] NCCL INFO Channel 05/0 : 15[7] -> 14[6] via P2P/CUMEM +n111-046-078:3319:3425 [6] NCCL INFO Channel 07/0 : 14[6] -> 13[5] via P2P/CUMEM +n111-046-078:3320:3430 [7] NCCL INFO Channel 06/0 : 15[7] -> 14[6] via P2P/CUMEM +n111-046-078:3320:3430 [7] NCCL INFO Channel 07/0 : 15[7] -> 14[6] via P2P/CUMEM +n111-046-078:3319:3425 [6] NCCL INFO Channel 08/0 : 14[6] -> 13[5] via P2P/CUMEM +n111-046-078:3320:3430 [7] NCCL INFO Channel 08/0 : 15[7] -> 14[6] via P2P/CUMEM +n111-046-078:3319:3425 [6] NCCL INFO Channel 09/0 : 14[6] -> 13[5] via P2P/CUMEM +n111-046-078:3320:3430 [7] NCCL INFO Channel 09/0 : 15[7] -> 14[6] via P2P/CUMEM +n111-046-078:3319:3425 [6] NCCL INFO Channel 10/0 : 14[6] -> 13[5] via P2P/CUMEM +n111-046-078:3319:3425 [6] NCCL INFO Channel 11/0 : 14[6] -> 13[5] via P2P/CUMEM +n111-046-078:3314:3426 [1] NCCL INFO Channel 00/0 : 9[1] -> 8[0] via P2P/CUMEM +n111-046-078:3319:3425 [6] NCCL INFO Channel 12/0 : 14[6] -> 13[5] via P2P/CUMEM +n111-046-078:3320:3430 [7] NCCL INFO Channel 10/0 : 15[7] -> 14[6] via P2P/CUMEM +n111-046-078:3314:3426 [1] NCCL INFO Channel 01/0 : 9[1] -> 8[0] via P2P/CUMEM +n111-046-078:3320:3430 [7] NCCL INFO Channel 11/0 : 15[7] -> 14[6] via P2P/CUMEM +n111-046-078:3319:3425 [6] NCCL INFO Channel 15/0 : 14[6] -> 13[5] via P2P/CUMEM +n111-046-078:3314:3426 [1] NCCL INFO Channel 02/0 : 9[1] -> 8[0] via P2P/CUMEM +n111-046-078:3320:3430 [7] NCCL INFO Channel 12/0 : 15[7] -> 14[6] via P2P/CUMEM +n111-046-078:3314:3426 [1] NCCL INFO Channel 03/0 : 9[1] -> 8[0] via P2P/CUMEM +n111-046-078:3320:3430 [7] NCCL INFO Channel 13/0 : 15[7] -> 14[6] via P2P/CUMEM +n111-046-078:3314:3426 [1] NCCL INFO Channel 04/0 : 9[1] -> 8[0] via P2P/CUMEM +n111-046-078:3320:3430 [7] NCCL INFO Channel 14/0 : 15[7] -> 14[6] via P2P/CUMEM +n111-046-078:3314:3426 [1] NCCL INFO Channel 05/0 : 9[1] -> 8[0] via P2P/CUMEM +n111-046-078:3314:3426 [1] NCCL INFO Channel 06/0 : 9[1] -> 8[0] via P2P/CUMEM +n111-046-078:3320:3430 [7] NCCL INFO Channel 15/0 : 15[7] -> 14[6] via P2P/CUMEM +n111-046-078:3314:3426 [1] NCCL INFO Channel 07/0 : 9[1] -> 8[0] via P2P/CUMEM +n111-046-078:3314:3426 [1] NCCL INFO Channel 08/0 : 9[1] -> 8[0] via P2P/CUMEM +n111-046-078:3314:3426 [1] NCCL INFO Channel 09/0 : 9[1] -> 8[0] via P2P/CUMEM +n111-046-078:3314:3426 [1] NCCL INFO Channel 10/0 : 9[1] -> 8[0] via P2P/CUMEM +n111-046-078:3314:3426 [1] NCCL INFO Channel 11/0 : 9[1] -> 8[0] via P2P/CUMEM +n111-046-078:3314:3426 [1] NCCL INFO Channel 12/0 : 9[1] -> 8[0] via P2P/CUMEM +n111-046-078:3314:3426 [1] NCCL INFO Channel 13/0 : 9[1] -> 8[0] via P2P/CUMEM +n111-046-078:3314:3426 [1] NCCL INFO Channel 14/0 : 9[1] -> 8[0] via P2P/CUMEM +n111-046-078:3314:3426 [1] NCCL INFO Channel 15/0 : 9[1] -> 8[0] via P2P/CUMEM +n111-046-078:3316:3428 [3] NCCL INFO Connected all trees +n111-046-078:3317:3427 [4] NCCL INFO Connected all trees +n111-046-078:3318:3429 [5] NCCL INFO Connected all trees +n111-046-078:3313:3432 [0] NCCL INFO Connected all trees +n111-046-078:3314:3426 [1] NCCL INFO Connected all trees +n111-046-078:3319:3425 [6] NCCL INFO Connected all trees +n111-046-078:3320:3430 [7] NCCL INFO Connected all trees +n111-046-078:3315:3431 [2] NCCL INFO Connected all trees +n111-046-078:3316:3471 [3] NCCL INFO Channel 03/0 : 11[3] -> 12[4] via P2P/CUMEM +n111-046-078:3316:3471 [3] NCCL INFO Channel 11/0 : 11[3] -> 12[4] via P2P/CUMEM +n111-046-078:3314:3472 [1] NCCL INFO Channel 01/0 : 9[1] -> 10[2] via P2P/CUMEM +n111-046-078:3314:3472 [1] NCCL INFO Channel 09/0 : 9[1] -> 10[2] via P2P/CUMEM +n111-046-078:3314:3472 [1] NCCL INFO Channel 00/0 : 9[1] -> 16[0] [send] via NET/FasTrak/0/GDRDMA +n111-046-078:3315:3470 [2] NCCL INFO Channel 02/0 : 3[3] -> 10[2] [receive] via NET/FasTrak/2/GDRDMA +n111-046-078:3315:3470 [2] NCCL INFO Channel 10/0 : 3[3] -> 10[2] [receive] via NET/FasTrak/2/GDRDMA +n111-046-078:3314:3472 [1] NCCL INFO Channel 01/0 : 0[0] -> 9[1] [receive] via NET/FasTrak/1/GDRDMA +n111-046-078:3314:3472 [1] NCCL INFO Channel 09/0 : 0[0] -> 9[1] [receive] via NET/FasTrak/1/GDRDMA +n111-046-078:3313:3473 [0] NCCL INFO Channel 00/0 : 1[1] -> 8[0] [receive] via NET/FasTrak/0/GDRDMA +n111-046-078:3313:3473 [0] NCCL INFO Channel 08/0 : 1[1] -> 8[0] [receive] via NET/FasTrak/0/GDRDMA +n111-046-078:3313:3473 [0] NCCL INFO Channel 00/0 : 8[0] -> 15[7] via P2P/CUMEM +n111-046-078:3313:3473 [0] NCCL INFO Channel 08/0 : 8[0] -> 15[7] via P2P/CUMEM +n111-046-078:3318:3474 [5] NCCL INFO Channel 05/0 : 13[5] -> 14[6] via P2P/CUMEM +n111-046-078:3318:3474 [5] NCCL INFO Channel 13/0 : 13[5] -> 14[6] via P2P/CUMEM +n111-046-078:3316:3471 [3] NCCL INFO Channel 02/0 : 11[3] -> 18[2] [send] via NET/FasTrak/2/GDRDMA +n111-046-078:3317:3475 [4] NCCL INFO Channel 04/0 : 5[5] -> 12[4] [receive] via NET/FasTrak/4/GDRDMA +n111-046-078:3316:3471 [3] NCCL INFO Channel 03/0 : 2[2] -> 11[3] [receive] via NET/FasTrak/3/GDRDMA +n111-046-078:3317:3475 [4] NCCL INFO Channel 12/0 : 5[5] -> 12[4] [receive] via NET/FasTrak/4/GDRDMA +n111-046-078:3316:3471 [3] NCCL INFO Channel 11/0 : 2[2] -> 11[3] [receive] via NET/FasTrak/3/GDRDMA +n111-046-078:3317:3475 [4] NCCL INFO Channel 05/0 : 12[4] -> 21[5] [send] via NET/FasTrak/5/GDRDMA +n111-046-078:3317:3475 [4] NCCL INFO Channel 04/0 : 12[4] -> 11[3] via P2P/CUMEM +n111-046-078:3317:3475 [4] NCCL INFO Channel 12/0 : 12[4] -> 11[3] via P2P/CUMEM +n111-046-078:3315:3470 [2] NCCL INFO Channel 03/0 : 10[2] -> 19[3] [send] via NET/FasTrak/3/GDRDMA +n111-046-078:3315:3470 [2] NCCL INFO Channel 02/0 : 10[2] -> 9[1] via P2P/CUMEM +n111-046-078:3315:3470 [2] NCCL INFO Channel 10/0 : 10[2] -> 9[1] via P2P/CUMEM +n111-046-078:3320:3476 [7] NCCL INFO Channel 06/0 : 15[7] -> 22[6] [send] via NET/FasTrak/6/GDRDMA +n111-046-078:3313:3473 [0] NCCL INFO Channel 01/0 : 8[0] -> 17[1] [send] via NET/FasTrak/1/GDRDMA +n111-046-078:3320:3476 [7] NCCL INFO Channel 07/0 : 6[6] -> 15[7] [receive] via NET/FasTrak/7/GDRDMA +n111-046-078:3320:3476 [7] NCCL INFO Channel 15/0 : 6[6] -> 15[7] [receive] via NET/FasTrak/7/GDRDMA +n111-046-078:3320:3476 [7] NCCL INFO Channel 07/0 : 15[7] -> 8[0] via P2P/CUMEM +n111-046-078:3320:3476 [7] NCCL INFO Channel 15/0 : 15[7] -> 8[0] via P2P/CUMEM +n111-046-078:3319:3477 [6] NCCL INFO Channel 06/0 : 7[7] -> 14[6] [receive] via NET/FasTrak/6/GDRDMA +n111-046-078:3319:3477 [6] NCCL INFO Channel 14/0 : 7[7] -> 14[6] [receive] via NET/FasTrak/6/GDRDMA +n111-046-078:3319:3477 [6] NCCL INFO Channel 07/0 : 14[6] -> 23[7] [send] via NET/FasTrak/7/GDRDMA +n111-046-078:3319:3477 [6] NCCL INFO Channel 06/0 : 14[6] -> 13[5] via P2P/CUMEM +n111-046-078:3319:3477 [6] NCCL INFO Channel 14/0 : 14[6] -> 13[5] via P2P/CUMEM +n111-046-078:3318:3474 [5] NCCL INFO Channel 04/0 : 13[5] -> 20[4] [send] via NET/FasTrak/4/GDRDMA +n111-046-078:3318:3474 [5] NCCL INFO Channel 05/0 : 4[4] -> 13[5] [receive] via NET/FasTrak/5/GDRDMA +n111-046-078:3318:3474 [5] NCCL INFO Channel 13/0 : 4[4] -> 13[5] [receive] via NET/FasTrak/5/GDRDMA +n111-046-078:3318:3474 [5] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n111-046-078:3317:3475 [4] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n111-046-078:3319:3477 [6] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n111-046-078:3320:3476 [7] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n111-046-078:3316:3471 [3] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n111-046-078:3315:3470 [2] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n111-046-078:3314:3472 [1] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n111-046-078:3313:3473 [0] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +[rank10]:[W727 05:39:59.336891743 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n111-046-078.byted.org]:46712, remote=[n116-042-185.byted.org]:25003): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f88225785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f880bccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7f880bccff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7f880bcd084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7f880bcca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f87cd3c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f8825d004a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f88288d71f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f882895789c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank10]:[W727 05:39:59.344659339 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 10] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank15]:[W727 05:39:59.444643595 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n111-046-078.byted.org]:46670, remote=[n116-042-185.byted.org]:25003): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f424ef785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f42386cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7f42386cff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7f42386d084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7f42386ca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f41f9dc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f42526fe4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f42552d51f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f425535589c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank15]:[W727 05:39:59.451621862 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 15] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank9]:[W727 05:39:59.494845403 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n111-046-078.byted.org]:46694, remote=[n116-042-185.byted.org]:25003): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7fd317d785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7fd3010cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7fd3010cff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7fd3010d084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7fd3010ca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fd2c27c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7fd31b2214a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7fd31ddf81f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7fd31de7889c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank9]:[W727 05:39:59.502078610 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 9] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank14]:[W727 05:39:59.586400397 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n111-046-078.byted.org]:46682, remote=[n116-042-185.byted.org]:25003): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7fd8b87785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7fd8a1ecdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7fd8a1ecff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7fd8a1ed084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7fd8a1eca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fd8635c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7fd8bbf144a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7fd8beaeb1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7fd8beb6b89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank14]:[W727 05:39:59.594321258 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 14] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank13]:[W727 05:39:59.642986997 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n111-046-078.byted.org]:46700, remote=[n116-042-185.byted.org]:25003): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7febcf1785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7febb84cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7febb84cff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7febb84d084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7febb84ca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7feb79bc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7febd26644a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7febd523b1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7febd52bb89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank13]:[W727 05:39:59.649823162 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 13] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank11]:[W727 05:39:59.749478556 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n111-046-078.byted.org]:46706, remote=[n116-042-185.byted.org]:25003): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f5149b785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f5132ecdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7f5132ecff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7f5132ed084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7f5132eca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f50f45c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f514d01f4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f514fbf61f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f514fc7689c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank11]:[W727 05:39:59.757565376 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 11] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +E0727 05:40:00.289000 3248 torch/distributed/elastic/multiprocessing/api.py:874] failed (exitcode: -15) local_rank: 0 (pid: 3313) of binary: /usr/bin/python3 +Traceback (most recent call last): + File "/usr/local/bin/torchrun", line 7, in + sys.exit(main()) + ^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 355, in wrapper + return f(*args, **kwargs) + ^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 892, in main + run(args) + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 883, in run + elastic_launch( + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 139, in __call__ + return launch_agent(self._config, self._entrypoint, list(args)) + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 270, in launch_agent + raise ChildFailedError( +torch.distributed.elastic.multiprocessing.errors.ChildFailedError: +============================================================ +main_cross_2direct_balance_realscale_perturb_v4_accum.py FAILED +------------------------------------------------------------ +Failures: +[1]: + time : 2026-07-27_05:40:00 + host : n111-046-078.byted.org + rank : 9 (local_rank: 1) + exitcode : -15 (pid: 3314) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 3314 +[2]: + time : 2026-07-27_05:40:00 + host : n111-046-078.byted.org + rank : 10 (local_rank: 2) + exitcode : -15 (pid: 3315) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 3315 +[3]: + time : 2026-07-27_05:40:00 + host : n111-046-078.byted.org + rank : 11 (local_rank: 3) + exitcode : -15 (pid: 3316) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 3316 +[4]: + time : 2026-07-27_05:40:00 + host : n111-046-078.byted.org + rank : 12 (local_rank: 4) + exitcode : -15 (pid: 3317) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 3317 +[5]: + time : 2026-07-27_05:40:00 + host : n111-046-078.byted.org + rank : 13 (local_rank: 5) + exitcode : -15 (pid: 3318) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 3318 +[6]: + time : 2026-07-27_05:40:00 + host : n111-046-078.byted.org + rank : 14 (local_rank: 6) + exitcode : -15 (pid: 3319) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 3319 +[7]: + time : 2026-07-27_05:40:00 + host : n111-046-078.byted.org + rank : 15 (local_rank: 7) + exitcode : -15 (pid: 3320) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 3320 +------------------------------------------------------------ +Root Cause (first observed failure): +[0]: + time : 2026-07-27_05:40:00 + host : n111-046-078.byted.org + rank : 8 (local_rank: 0) + exitcode : -15 (pid: 3313) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 3313 +============================================================ diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1603_vith32_launch_stub_master_n116-042-185_port25003/launcher_rank2_n116-040-048.log b/ScaleDF_rebuttal/_host_logs/20260726_1603_vith32_launch_stub_master_n116-042-185_port25003/launcher_rank2_n116-040-048.log new file mode 100644 index 0000000000000000000000000000000000000000..e7c89223698206f7bfda6ff17b730a433c45e7d7 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1603_vith32_launch_stub_master_n116-042-185_port25003/launcher_rank2_n116-040-048.log @@ -0,0 +1,1868 @@ +started_utc=2026-07-26T16:03:57Z +hostname=n116-040-048 +node_rank=2 +nnodes=4 +nproc_per_node=8 +world_size=32 +master_addr=10.116.42.185 +master_port=25003 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +main=main_cross_2direct_balance_realscale_perturb_v4_accum.py +model_name=deit_huge_patch14_LS +pretrain=/tmp/ScaleDF_repro_20260724/pretrained/deit_3_huge_224_21k_v1.pth +resume=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_24card_effbs3072_from_epoch7_n116-041-246_n116-042-186_n116-042-185_20260726_0736_vith24_from_epoch7/checkpoint.pth +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_32card_effbs2048_from_latest24_master_n116-042-185_20260726_1603_vith32_stub_port25003 +batch_per_gpu=64 +accum_iter=1 +effective_batch=2048 +eval_batch_per_gpu=64 +epochs=40 +W0726 16:04:10.307000 1714 torch/distributed/run.py:766] +W0726 16:04:10.307000 1714 torch/distributed/run.py:766] ***************************************** +W0726 16:04:10.307000 1714 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 16:04:10.307000 1714 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 16): env:// +[rank16]:[W726 16:04:57.173332685 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 16] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 22): env:// +| distributed init (rank 18): env:// +[rank22]:[W726 16:04:57.471532254 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 22] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank18]:[W726 16:04:57.475699518 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 18] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 23): env:// +| distributed init (rank 19): env:// +[rank23]:[W726 16:04:57.516725737 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 23] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank19]:[W726 16:04:57.527440933 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 19] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 21): env:// +[rank21]:[W726 16:04:57.555697932 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 21] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 20): env:// +| distributed init (rank 17): env:// +[rank20]:[W726 16:04:57.607706989 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 20] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank17]:[W726 16:04:57.607898491 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 17] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +n116-040-048:1782:1782 [3] NCCL INFO cudaDriverVersion 12060 +n116-040-048:1782:1782 [3] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-040-048:1782:1782 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-040-048:1782:1782 [3] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-040-048:1782:1782 [3] NCCL INFO Comm config Blocking set to 1 +n116-040-048:1784:1784 [5] NCCL INFO cudaDriverVersion 12060 +n116-040-048:1784:1784 [5] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-040-048:1784:1784 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-040-048:1784:1784 [5] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-040-048:1784:1784 [5] NCCL INFO Comm config Blocking set to 1 +n116-040-048:1783:1783 [4] NCCL INFO cudaDriverVersion 12060 +n116-040-048:1783:1783 [4] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-040-048:1783:1783 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-040-048:1783:1783 [4] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-040-048:1783:1783 [4] NCCL INFO Comm config Blocking set to 1 +n116-040-048:1785:1785 [6] NCCL INFO cudaDriverVersion 12060 +n116-040-048:1785:1785 [6] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-040-048:1785:1785 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-040-048:1780:1780 [1] NCCL INFO cudaDriverVersion 12060 +n116-040-048:1780:1780 [1] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-040-048:1780:1780 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-040-048:1785:1785 [6] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-040-048:1780:1780 [1] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-040-048:1786:1786 [7] NCCL INFO cudaDriverVersion 12060 +n116-040-048:1786:1786 [7] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-040-048:1786:1786 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-040-048:1786:1786 [7] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-040-048:1785:1785 [6] NCCL INFO Comm config Blocking set to 1 +n116-040-048:1780:1780 [1] NCCL INFO Comm config Blocking set to 1 +n116-040-048:1786:1786 [7] NCCL INFO Comm config Blocking set to 1 +n116-040-048:1779:1779 [0] NCCL INFO cudaDriverVersion 12060 +n116-040-048:1779:1779 [0] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-040-048:1779:1779 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-040-048:1779:1779 [0] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-040-048:1781:1781 [2] NCCL INFO cudaDriverVersion 12060 +n116-040-048:1781:1781 [2] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-040-048:1781:1781 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-040-048:1781:1781 [2] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-040-048:1779:1779 [0] NCCL INFO Comm config Blocking set to 1 +n116-040-048:1781:1781 [2] NCCL INFO Comm config Blocking set to 1 +n116-040-048:1783:1837 [4] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-040-048:1781:1842 [2] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-040-048:1782:1835 [3] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-040-048:1786:1840 [7] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-040-048:1782:1835 [3] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-040-048:1786:1840 [7] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-040-048:1781:1842 [2] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-040-048:1782:1835 [3] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-040-048:1783:1837 [4] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-040-048:1781:1842 [2] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-040-048:1786:1840 [7] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-040-048:1782:1835 [3] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-040-048:1781:1842 [2] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-040-048:1783:1837 [4] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-040-048:1786:1840 [7] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-040-048:1783:1837 [4] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-040-048:1781:1842 [2] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-040-048:1782:1835 [3] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-040-048:1786:1840 [7] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-040-048:1783:1837 [4] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-040-048:1781:1842 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-040-048:1786:1840 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-040-048:1782:1835 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-040-048:1783:1837 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-040-048:1786:1840 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-040-048:1781:1842 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-040-048:1783:1837 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-040-048:1779:1841 [0] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-040-048:1782:1835 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-040-048:1786:1840 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-040-048:1781:1842 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-040-048:1783:1837 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-040-048:1785:1838 [6] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-040-048:1783:1837 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-040-048:1782:1835 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-040-048:1781:1842 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-040-048:1786:1840 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-040-048:1783:1837 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-040-048:1782:1835 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-040-048:1781:1842 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-040-048:1786:1840 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-040-048:1782:1835 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-040-048:1779:1841 [0] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-040-048:1779:1841 [0] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-040-048:1779:1841 [0] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-040-048:1780:1839 [1] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-040-048:1785:1838 [6] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-040-048:1785:1838 [6] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-040-048:1779:1841 [0] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-040-048:1785:1838 [6] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-040-048:1779:1841 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-040-048:1779:1841 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-040-048:1779:1841 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-040-048:1779:1841 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-040-048:1784:1836 [5] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-040-048:1780:1839 [1] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-040-048:1779:1841 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-040-048:1785:1838 [6] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-040-048:1780:1839 [1] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-040-048:1785:1838 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-040-048:1780:1839 [1] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-040-048:1785:1838 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-040-048:1785:1838 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-040-048:1784:1836 [5] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-040-048:1785:1838 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-040-048:1780:1839 [1] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-040-048:1785:1838 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-040-048:1784:1836 [5] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-040-048:1780:1839 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-040-048:1784:1836 [5] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-040-048:1780:1839 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-040-048:1780:1839 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-040-048:1780:1839 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-040-048:1780:1839 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-040-048:1784:1836 [5] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-040-048:1784:1836 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-040-048:1784:1836 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-040-048:1784:1836 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-040-048:1784:1836 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-040-048:1784:1836 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-040-048:1781:1842 [2] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-040-048:1781:1842 [2] NCCL INFO Using network FasTrak +n116-040-048:1780:1839 [1] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-040-048:1780:1839 [1] NCCL INFO Using network FasTrak +n116-040-048:1783:1837 [4] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-040-048:1783:1837 [4] NCCL INFO Using network FasTrak +n116-040-048:1782:1835 [3] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-040-048:1782:1835 [3] NCCL INFO Using network FasTrak +n116-040-048:1786:1840 [7] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-040-048:1786:1840 [7] NCCL INFO Using network FasTrak +n116-040-048:1779:1841 [0] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-040-048:1779:1841 [0] NCCL INFO Using network FasTrak +n116-040-048:1785:1838 [6] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-040-048:1785:1838 [6] NCCL INFO Using network FasTrak +n116-040-048:1784:1836 [5] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-040-048:1784:1836 [5] NCCL INFO Using network FasTrak +n116-040-048:1781:1842 [2] NCCL INFO DMA-BUF is available on GPU device 2 +n116-040-048:1781:1842 [2] NCCL INFO ncclCommInitRankConfig comm 0xcd2af00 rank 18 nranks 32 cudaDev 2 nvmlDev 2 busId b000 commId 0x2a4d35ea441cbcba - Init START +n116-040-048:1786:1840 [7] NCCL INFO DMA-BUF is available on GPU device 7 +n116-040-048:1780:1839 [1] NCCL INFO DMA-BUF is available on GPU device 1 +n116-040-048:1783:1837 [4] NCCL INFO DMA-BUF is available on GPU device 4 +n116-040-048:1782:1835 [3] NCCL INFO DMA-BUF is available on GPU device 3 +n116-040-048:1779:1841 [0] NCCL INFO DMA-BUF is available on GPU device 0 +n116-040-048:1780:1839 [1] NCCL INFO ncclCommInitRankConfig comm 0xbfe3fc0 rank 17 nranks 32 cudaDev 1 nvmlDev 1 busId 5000 commId 0x2a4d35ea441cbcba - Init START +n116-040-048:1786:1840 [7] NCCL INFO ncclCommInitRankConfig comm 0xcc263e0 rank 23 nranks 32 cudaDev 7 nvmlDev 7 busId 8c000 commId 0x2a4d35ea441cbcba - Init START +n116-040-048:1779:1841 [0] NCCL INFO ncclCommInitRankConfig comm 0xb7cca60 rank 16 nranks 32 cudaDev 0 nvmlDev 0 busId 4000 commId 0x2a4d35ea441cbcba - Init START +n116-040-048:1785:1838 [6] NCCL INFO DMA-BUF is available on GPU device 6 +n116-040-048:1783:1837 [4] NCCL INFO ncclCommInitRankConfig comm 0xb926bd0 rank 20 nranks 32 cudaDev 4 nvmlDev 4 busId 84000 commId 0x2a4d35ea441cbcba - Init START +n116-040-048:1782:1835 [3] NCCL INFO ncclCommInitRankConfig comm 0xc150970 rank 19 nranks 32 cudaDev 3 nvmlDev 3 busId c000 commId 0x2a4d35ea441cbcba - Init START +n116-040-048:1784:1836 [5] NCCL INFO DMA-BUF is available on GPU device 5 +n116-040-048:1780:1839 [1] NCCL INFO RAS client listening socket at ::1<28028> +n116-040-048:1785:1838 [6] NCCL INFO ncclCommInitRankConfig comm 0xc067bd0 rank 22 nranks 32 cudaDev 6 nvmlDev 6 busId 8b000 commId 0x2a4d35ea441cbcba - Init START +n116-040-048:1782:1835 [3] NCCL INFO RAS client listening socket at ::1<28028> +n116-040-048:1781:1842 [2] NCCL INFO RAS client listening socket at ::1<28028> +n116-040-048:1784:1836 [5] NCCL INFO ncclCommInitRankConfig comm 0xc314c50 rank 21 nranks 32 cudaDev 5 nvmlDev 5 busId 85000 commId 0x2a4d35ea441cbcba - Init START +n116-040-048:1783:1837 [4] NCCL INFO RAS client listening socket at ::1<28028> +n116-040-048:1784:1836 [5] NCCL INFO RAS client listening socket at ::1<28028> +n116-040-048:1785:1838 [6] NCCL INFO RAS client listening socket at ::1<28028> +n116-040-048:1779:1841 [0] NCCL INFO RAS client listening socket at ::1<28028> +n116-040-048:1786:1840 [7] NCCL INFO RAS client listening socket at ::1<28028> +n116-040-048:1785:1838 [6] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-040-048:1785:1838 [6] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-040-048:1785:1838 [6] NCCL INFO Retrieving state for FasTrak +n116-040-048:1785:1838 [6] NCCL INFO Initialized state 0 for FasTrak +n116-040-048:1785:1838 [6] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-040-048:1785:1838 [6] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-040-048:1785:1838 [6] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-040-048:1785:1838 [6] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-040-048:1785:1838 [6] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-040-048:1785:1838 [6] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-040-048:1785:1838 [6] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-040-048:1785:1838 [6] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-040-048:1786:1840 [7] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-040-048:1786:1840 [7] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-040-048:1786:1840 [7] NCCL INFO Retrieving state for FasTrak +n116-040-048:1786:1840 [7] NCCL INFO Initialized state 0 for FasTrak +n116-040-048:1786:1840 [7] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-040-048:1786:1840 [7] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-040-048:1786:1840 [7] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-040-048:1786:1840 [7] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-040-048:1786:1840 [7] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-040-048:1786:1840 [7] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-040-048:1786:1840 [7] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-040-048:1786:1840 [7] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-040-048:1784:1836 [5] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-040-048:1784:1836 [5] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-040-048:1784:1836 [5] NCCL INFO Retrieving state for FasTrak +n116-040-048:1784:1836 [5] NCCL INFO Initialized state 0 for FasTrak +n116-040-048:1784:1836 [5] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-040-048:1784:1836 [5] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-040-048:1784:1836 [5] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-040-048:1784:1836 [5] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-040-048:1784:1836 [5] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-040-048:1784:1836 [5] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-040-048:1784:1836 [5] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-040-048:1783:1837 [4] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-040-048:1783:1837 [4] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-040-048:1783:1837 [4] NCCL INFO Retrieving state for FasTrak +n116-040-048:1783:1837 [4] NCCL INFO Initialized state 0 for FasTrak +n116-040-048:1784:1836 [5] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-040-048:1783:1837 [4] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-040-048:1783:1837 [4] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-040-048:1783:1837 [4] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-040-048:1782:1835 [3] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-040-048:1783:1837 [4] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-040-048:1782:1835 [3] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-040-048:1782:1835 [3] NCCL INFO Retrieving state for FasTrak +n116-040-048:1782:1835 [3] NCCL INFO Initialized state 0 for FasTrak +n116-040-048:1783:1837 [4] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-040-048:1782:1835 [3] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-040-048:1783:1837 [4] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-040-048:1782:1835 [3] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-040-048:1783:1837 [4] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-040-048:1782:1835 [3] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-040-048:1783:1837 [4] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-040-048:1782:1835 [3] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-040-048:1782:1835 [3] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-040-048:1782:1835 [3] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-040-048:1782:1835 [3] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-040-048:1782:1835 [3] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-040-048:1780:1839 [1] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-040-048:1780:1839 [1] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-040-048:1780:1839 [1] NCCL INFO Retrieving state for FasTrak +n116-040-048:1780:1839 [1] NCCL INFO Initialized state 0 for FasTrak +n116-040-048:1780:1839 [1] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-040-048:1780:1839 [1] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-040-048:1780:1839 [1] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-040-048:1780:1839 [1] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-040-048:1781:1842 [2] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-040-048:1781:1842 [2] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-040-048:1781:1842 [2] NCCL INFO Retrieving state for FasTrak +n116-040-048:1781:1842 [2] NCCL INFO Initialized state 0 for FasTrak +n116-040-048:1780:1839 [1] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-040-048:1780:1839 [1] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-040-048:1781:1842 [2] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-040-048:1780:1839 [1] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-040-048:1781:1842 [2] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-040-048:1780:1839 [1] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-040-048:1781:1842 [2] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-040-048:1781:1842 [2] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-040-048:1781:1842 [2] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-040-048:1781:1842 [2] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-040-048:1781:1842 [2] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-040-048:1781:1842 [2] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-040-048:1779:1841 [0] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-040-048:1779:1841 [0] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-040-048:1779:1841 [0] NCCL INFO Retrieving state for FasTrak +n116-040-048:1779:1841 [0] NCCL INFO Initialized state 0 for FasTrak +n116-040-048:1779:1841 [0] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-040-048:1779:1841 [0] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-040-048:1779:1841 [0] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-040-048:1779:1841 [0] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-040-048:1779:1841 [0] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-040-048:1779:1841 [0] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-040-048:1779:1841 [0] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-040-048:1779:1841 [0] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-040-048:1783:1837 [4] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-040-048:1784:1836 [5] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-040-048:1780:1839 [1] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-040-048:1782:1835 [3] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-040-048:1779:1841 [0] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-040-048:1785:1838 [6] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-040-048:1786:1840 [7] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-040-048:1781:1842 [2] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-040-048:1783:1837 [4] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-040-048:1783:1837 [4] NCCL INFO CPU/0-0 (1/1/3) +n116-040-048:1783:1837 [4] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-040-048:1783:1837 [4] NCCL INFO + PCI[24.0] - GPU/0-4000 (16) +n116-040-048:1783:1837 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1783:1837 [4] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-040-048:1783:1837 [4] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-040-048:1783:1837 [4] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-040-048:1783:1837 [4] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-040-048:1783:1837 [4] NCCL INFO + PCI[24.0] - GPU/0-5000 (17) +n116-040-048:1783:1837 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1783:1837 [4] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-040-048:1783:1837 [4] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-040-048:1783:1837 [4] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-040-048:1783:1837 [4] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-040-048:1783:1837 [4] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-040-048:1783:1837 [4] NCCL INFO + PCI[24.0] - GPU/0-b000 (18) +n116-040-048:1783:1837 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1783:1837 [4] NCCL INFO + PCI[24.0] - GPU/0-c000 (19) +n116-040-048:1783:1837 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1783:1837 [4] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-040-048:1783:1837 [4] NCCL INFO CPU/0-1 (1/1/3) +n116-040-048:1783:1837 [4] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-040-048:1783:1837 [4] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-040-048:1783:1837 [4] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-040-048:1783:1837 [4] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-040-048:1783:1837 [4] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-040-048:1783:1837 [4] NCCL INFO + PCI[24.0] - GPU/0-84000 (20) +n116-040-048:1783:1837 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1783:1837 [4] NCCL INFO + PCI[24.0] - GPU/0-85000 (21) +n116-040-048:1783:1837 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1783:1837 [4] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-040-048:1783:1837 [4] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-040-048:1783:1837 [4] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-040-048:1783:1837 [4] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-040-048:1783:1837 [4] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-040-048:1783:1837 [4] NCCL INFO + PCI[24.0] - GPU/0-8b000 (22) +n116-040-048:1783:1837 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1783:1837 [4] NCCL INFO + PCI[24.0] - GPU/0-8c000 (23) +n116-040-048:1783:1837 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1783:1837 [4] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-040-048:1783:1837 [4] NCCL INFO ========================================== +n116-040-048:1783:1837 [4] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-048:1783:1837 [4] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-048:1783:1837 [4] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-048:1783:1837 [4] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-048:1783:1837 [4] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-048:1783:1837 [4] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-048:1783:1837 [4] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-040-048:1782:1835 [3] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-040-048:1784:1836 [5] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-040-048:1783:1837 [4] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-040-048:1784:1836 [5] NCCL INFO CPU/0-0 (1/1/3) +n116-040-048:1784:1836 [5] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-040-048:1782:1835 [3] NCCL INFO CPU/0-0 (1/1/3) +n116-040-048:1784:1836 [5] NCCL INFO + PCI[24.0] - GPU/0-4000 (16) +n116-040-048:1784:1836 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1783:1837 [4] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-048:1782:1835 [3] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-040-048:1784:1836 [5] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-040-048:1782:1835 [3] NCCL INFO + PCI[24.0] - GPU/0-4000 (16) +n116-040-048:1784:1836 [5] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-040-048:1782:1835 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1782:1835 [3] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-040-048:1784:1836 [5] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-040-048:1784:1836 [5] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-040-048:1782:1835 [3] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-040-048:1783:1837 [4] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-048:1784:1836 [5] NCCL INFO + PCI[24.0] - GPU/0-5000 (17) +n116-040-048:1782:1835 [3] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-040-048:1784:1836 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1784:1836 [5] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-040-048:1782:1835 [3] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-040-048:1784:1836 [5] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-040-048:1782:1835 [3] NCCL INFO + PCI[24.0] - GPU/0-5000 (17) +n116-040-048:1784:1836 [5] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-040-048:1782:1835 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1784:1836 [5] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-040-048:1782:1835 [3] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-040-048:1783:1837 [4] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-048:1782:1835 [3] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-040-048:1784:1836 [5] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-040-048:1782:1835 [3] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-040-048:1784:1836 [5] NCCL INFO + PCI[24.0] - GPU/0-b000 (18) +n116-040-048:1782:1835 [3] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-040-048:1784:1836 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1784:1836 [5] NCCL INFO + PCI[24.0] - GPU/0-c000 (19) +n116-040-048:1782:1835 [3] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-040-048:1784:1836 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1782:1835 [3] NCCL INFO + PCI[24.0] - GPU/0-b000 (18) +n116-040-048:1783:1837 [4] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-048:1784:1836 [5] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-040-048:1782:1835 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1784:1836 [5] NCCL INFO CPU/0-1 (1/1/3) +n116-040-048:1782:1835 [3] NCCL INFO + PCI[24.0] - GPU/0-c000 (19) +n116-040-048:1784:1836 [5] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-040-048:1782:1835 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1784:1836 [5] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-040-048:1782:1835 [3] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-040-048:1784:1836 [5] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-040-048:1782:1835 [3] NCCL INFO CPU/0-1 (1/1/3) +n116-040-048:1784:1836 [5] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-040-048:1782:1835 [3] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-040-048:1783:1837 [4] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-040-048:1782:1835 [3] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-040-048:1784:1836 [5] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-040-048:1782:1835 [3] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-040-048:1784:1836 [5] NCCL INFO + PCI[24.0] - GPU/0-84000 (20) +n116-040-048:1782:1835 [3] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-040-048:1784:1836 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1782:1835 [3] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-040-048:1784:1836 [5] NCCL INFO + PCI[24.0] - GPU/0-85000 (21) +n116-040-048:1782:1835 [3] NCCL INFO + PCI[24.0] - GPU/0-84000 (20) +n116-040-048:1784:1836 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1782:1835 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1783:1837 [4] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-040-048:1784:1836 [5] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-040-048:1782:1835 [3] NCCL INFO + PCI[24.0] - GPU/0-85000 (21) +n116-040-048:1779:1841 [0] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-040-048:1784:1836 [5] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-040-048:1782:1835 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1784:1836 [5] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-040-048:1782:1835 [3] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-040-048:1779:1841 [0] NCCL INFO CPU/0-0 (1/1/3) +n116-040-048:1784:1836 [5] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-040-048:1782:1835 [3] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-040-048:1783:1837 [4] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-040-048:1779:1841 [0] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-040-048:1784:1836 [5] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-040-048:1782:1835 [3] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-040-048:1784:1836 [5] NCCL INFO + PCI[24.0] - GPU/0-8b000 (22) +n116-040-048:1779:1841 [0] NCCL INFO + PCI[24.0] - GPU/0-4000 (16) +n116-040-048:1782:1835 [3] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-040-048:1784:1836 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1783:1837 [4] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-040-048:1782:1835 [3] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-040-048:1779:1841 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1782:1835 [3] NCCL INFO + PCI[24.0] - GPU/0-8b000 (22) +n116-040-048:1784:1836 [5] NCCL INFO + PCI[24.0] - GPU/0-8c000 (23) +n116-040-048:1779:1841 [0] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-040-048:1782:1835 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1784:1836 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1779:1841 [0] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-040-048:1782:1835 [3] NCCL INFO + PCI[24.0] - GPU/0-8c000 (23) +n116-040-048:1784:1836 [5] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-040-048:1779:1841 [0] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-040-048:1782:1835 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1784:1836 [5] NCCL INFO ========================================== +n116-040-048:1782:1835 [3] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-040-048:1783:1837 [4] NCCL INFO Setting affinity for GPU 4 to 7fff,ffffffff,e0000000,0000007f,ffffffff,ffe00000,00000000 +n116-040-048:1779:1841 [0] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-040-048:1782:1835 [3] NCCL INFO ========================================== +n116-040-048:1779:1841 [0] NCCL INFO + PCI[24.0] - GPU/0-5000 (17) +n116-040-048:1779:1841 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1779:1841 [0] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-040-048:1784:1836 [5] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-048:1779:1841 [0] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-040-048:1779:1841 [0] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-040-048:1782:1835 [3] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-048:1779:1841 [0] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-040-048:1779:1841 [0] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-040-048:1779:1841 [0] NCCL INFO + PCI[24.0] - GPU/0-b000 (18) +n116-040-048:1784:1836 [5] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-048:1779:1841 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1782:1835 [3] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-048:1779:1841 [0] NCCL INFO + PCI[24.0] - GPU/0-c000 (19) +n116-040-048:1779:1841 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1784:1836 [5] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-048:1782:1835 [3] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-048:1779:1841 [0] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-040-048:1779:1841 [0] NCCL INFO CPU/0-1 (1/1/3) +n116-040-048:1784:1836 [5] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-048:1779:1841 [0] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-040-048:1779:1841 [0] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-040-048:1779:1841 [0] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-040-048:1782:1835 [3] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-048:1779:1841 [0] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-040-048:1779:1841 [0] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-040-048:1779:1841 [0] NCCL INFO + PCI[24.0] - GPU/0-84000 (20) +n116-040-048:1779:1841 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1784:1836 [5] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-048:1779:1841 [0] NCCL INFO + PCI[24.0] - GPU/0-85000 (21) +n116-040-048:1782:1835 [3] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-048:1779:1841 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1779:1841 [0] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-040-048:1779:1841 [0] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-040-048:1783:1837 [4] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-040-048:1779:1841 [0] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-040-048:1784:1836 [5] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-048:1779:1841 [0] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-040-048:1779:1841 [0] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-040-048:1779:1841 [0] NCCL INFO + PCI[24.0] - GPU/0-8b000 (22) +n116-040-048:1782:1835 [3] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-048:1779:1841 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1779:1841 [0] NCCL INFO + PCI[24.0] - GPU/0-8c000 (23) +n116-040-048:1779:1841 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1779:1841 [0] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-040-048:1779:1841 [0] NCCL INFO ========================================== +n116-040-048:1784:1836 [5] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-040-048:1785:1838 [6] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-040-048:1782:1835 [3] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-040-048:1785:1838 [6] NCCL INFO CPU/0-0 (1/1/3) +n116-040-048:1785:1838 [6] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-040-048:1784:1836 [5] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-040-048:1779:1841 [0] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-048:1785:1838 [6] NCCL INFO + PCI[24.0] - GPU/0-4000 (16) +n116-040-048:1785:1838 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1785:1838 [6] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-040-048:1785:1838 [6] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-040-048:1782:1835 [3] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-040-048:1785:1838 [6] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-040-048:1784:1836 [5] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-048:1785:1838 [6] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-040-048:1779:1841 [0] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-048:1785:1838 [6] NCCL INFO + PCI[24.0] - GPU/0-5000 (17) +n116-040-048:1785:1838 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1785:1838 [6] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-040-048:1785:1838 [6] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-040-048:1782:1835 [3] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-048:1785:1838 [6] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-040-048:1784:1836 [5] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-048:1785:1838 [6] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-040-048:1785:1838 [6] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-040-048:1779:1841 [0] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-048:1785:1838 [6] NCCL INFO + PCI[24.0] - GPU/0-b000 (18) +n116-040-048:1785:1838 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1785:1838 [6] NCCL INFO + PCI[24.0] - GPU/0-c000 (19) +n116-040-048:1782:1835 [3] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-048:1785:1838 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1786:1840 [7] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-040-048:1784:1836 [5] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-048:1786:1840 [7] NCCL INFO CPU/0-0 (1/1/3) +n116-040-048:1779:1841 [0] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-048:1785:1838 [6] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-040-048:1783:1837 [4] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-040-048:1782:1835 [3] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-048:1785:1838 [6] NCCL INFO CPU/0-1 (1/1/3) +n116-040-048:1786:1840 [7] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-040-048:1784:1836 [5] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-048:1785:1838 [6] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-040-048:1786:1840 [7] NCCL INFO + PCI[24.0] - GPU/0-4000 (16) +n116-040-048:1779:1841 [0] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-048:1785:1838 [6] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-040-048:1786:1840 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1785:1838 [6] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-040-048:1782:1835 [3] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-048:1785:1838 [6] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-040-048:1786:1840 [7] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-040-048:1785:1838 [6] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-040-048:1785:1838 [6] NCCL INFO + PCI[24.0] - GPU/0-84000 (20) +n116-040-048:1786:1840 [7] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-040-048:1784:1836 [5] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-040-048:1785:1838 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1786:1840 [7] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-040-048:1779:1841 [0] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-048:1785:1838 [6] NCCL INFO + PCI[24.0] - GPU/0-85000 (21) +n116-040-048:1782:1835 [3] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-040-048:1786:1840 [7] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-040-048:1785:1838 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1786:1840 [7] NCCL INFO + PCI[24.0] - GPU/0-5000 (17) +n116-040-048:1785:1838 [6] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-040-048:1786:1840 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1784:1836 [5] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-040-048:1785:1838 [6] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-040-048:1786:1840 [7] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-040-048:1779:1841 [0] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-040-048:1785:1838 [6] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-040-048:1786:1840 [7] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-040-048:1782:1835 [3] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-040-048:1785:1838 [6] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-040-048:1786:1840 [7] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-040-048:1785:1838 [6] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-040-048:1786:1840 [7] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-040-048:1785:1838 [6] NCCL INFO + PCI[24.0] - GPU/0-8b000 (22) +n116-040-048:1784:1836 [5] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-040-048:1779:1841 [0] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-040-048:1782:1835 [3] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-040-048:1786:1840 [7] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-040-048:1785:1838 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1786:1840 [7] NCCL INFO + PCI[24.0] - GPU/0-b000 (18) +n116-040-048:1786:1840 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1785:1838 [6] NCCL INFO + PCI[24.0] - GPU/0-8c000 (23) +n116-040-048:1786:1840 [7] NCCL INFO + PCI[24.0] - GPU/0-c000 (19) +n116-040-048:1785:1838 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1781:1842 [2] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-040-048:1782:1835 [3] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-040-048:1784:1836 [5] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-040-048:1785:1838 [6] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-040-048:1786:1840 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1779:1841 [0] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-048:1781:1842 [2] NCCL INFO CPU/0-0 (1/1/3) +n116-040-048:1785:1838 [6] NCCL INFO ========================================== +n116-040-048:1786:1840 [7] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-040-048:1786:1840 [7] NCCL INFO CPU/0-1 (1/1/3) +n116-040-048:1781:1842 [2] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-040-048:1786:1840 [7] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-040-048:1782:1835 [3] NCCL INFO Setting affinity for GPU 3 to 0fffffff,fffffe00,00000000,000fffff,fffffffe +n116-040-048:1786:1840 [7] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-040-048:1781:1842 [2] NCCL INFO + PCI[24.0] - GPU/0-4000 (16) +n116-040-048:1779:1841 [0] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-048:1785:1838 [6] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-048:1781:1842 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1784:1836 [5] NCCL INFO Setting affinity for GPU 5 to 7fff,ffffffff,e0000000,0000007f,ffffffff,ffe00000,00000000 +n116-040-048:1786:1840 [7] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-040-048:1780:1839 [1] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-040-048:1781:1842 [2] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-040-048:1782:1835 [3] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-040-048:1786:1840 [7] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-040-048:1785:1838 [6] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-048:1781:1842 [2] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-040-048:1780:1839 [1] NCCL INFO CPU/0-0 (1/1/3) +n116-040-048:1779:1841 [0] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-048:1781:1842 [2] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-040-048:1786:1840 [7] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-040-048:1780:1839 [1] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-040-048:1781:1842 [2] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-040-048:1786:1840 [7] NCCL INFO + PCI[24.0] - GPU/0-84000 (20) +n116-040-048:1781:1842 [2] NCCL INFO + PCI[24.0] - GPU/0-5000 (17) +n116-040-048:1781:1842 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1780:1839 [1] NCCL INFO + PCI[24.0] - GPU/0-4000 (16) +n116-040-048:1786:1840 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1785:1838 [6] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-048:1786:1840 [7] NCCL INFO + PCI[24.0] - GPU/0-85000 (21) +n116-040-048:1779:1841 [0] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-048:1781:1842 [2] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-040-048:1780:1839 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1786:1840 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1781:1842 [2] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-040-048:1786:1840 [7] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-040-048:1780:1839 [1] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-040-048:1785:1838 [6] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-048:1781:1842 [2] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-040-048:1779:1841 [0] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-040-048:1781:1842 [2] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-040-048:1786:1840 [7] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-040-048:1782:1835 [3] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-040-048:1780:1839 [1] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-040-048:1781:1842 [2] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-040-048:1786:1840 [7] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-040-048:1781:1842 [2] NCCL INFO + PCI[24.0] - GPU/0-b000 (18) +n116-040-048:1780:1839 [1] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-040-048:1786:1840 [7] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-040-048:1780:1839 [1] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-040-048:1785:1838 [6] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-048:1781:1842 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1779:1841 [0] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-040-048:1786:1840 [7] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-040-048:1786:1840 [7] NCCL INFO + PCI[24.0] - GPU/0-8b000 (22) +n116-040-048:1780:1839 [1] NCCL INFO + PCI[24.0] - GPU/0-5000 (17) +n116-040-048:1781:1842 [2] NCCL INFO + PCI[24.0] - GPU/0-c000 (19) +n116-040-048:1780:1839 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1786:1840 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1784:1836 [5] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-040-048:1785:1838 [6] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-048:1779:1841 [0] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-040-048:1781:1842 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1780:1839 [1] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-040-048:1786:1840 [7] NCCL INFO + PCI[24.0] - GPU/0-8c000 (23) +n116-040-048:1781:1842 [2] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-040-048:1786:1840 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1781:1842 [2] NCCL INFO CPU/0-1 (1/1/3) +n116-040-048:1780:1839 [1] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-040-048:1786:1840 [7] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-040-048:1781:1842 [2] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-040-048:1780:1839 [1] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-040-048:1786:1840 [7] NCCL INFO ========================================== +n116-040-048:1779:1841 [0] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-040-048:1780:1839 [1] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-040-048:1785:1838 [6] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-040-048:1781:1842 [2] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-040-048:1780:1839 [1] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-040-048:1781:1842 [2] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-040-048:1780:1839 [1] NCCL INFO + PCI[24.0] - GPU/0-b000 (18) +n116-040-048:1780:1839 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1786:1840 [7] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-048:1780:1839 [1] NCCL INFO + PCI[24.0] - GPU/0-c000 (19) +n116-040-048:1781:1842 [2] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-040-048:1779:1841 [0] NCCL INFO Setting affinity for GPU 0 to 0fffffff,fffffe00,00000000,000fffff,fffffffe +n116-040-048:1785:1838 [6] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-040-048:1780:1839 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1781:1842 [2] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-040-048:1780:1839 [1] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-040-048:1781:1842 [2] NCCL INFO + PCI[24.0] - GPU/0-84000 (20) +n116-040-048:1786:1840 [7] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-048:1780:1839 [1] NCCL INFO CPU/0-1 (1/1/3) +n116-040-048:1781:1842 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1780:1839 [1] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-040-048:1781:1842 [2] NCCL INFO + PCI[24.0] - GPU/0-85000 (21) +n116-040-048:1785:1838 [6] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-048:1781:1842 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1780:1839 [1] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-040-048:1781:1842 [2] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-040-048:1780:1839 [1] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-040-048:1786:1840 [7] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-048:1781:1842 [2] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-040-048:1780:1839 [1] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-040-048:1785:1838 [6] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-048:1781:1842 [2] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-040-048:1780:1839 [1] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-040-048:1781:1842 [2] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-040-048:1784:1836 [5] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-040-048:1780:1839 [1] NCCL INFO + PCI[24.0] - GPU/0-84000 (20) +n116-040-048:1781:1842 [2] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-040-048:1786:1840 [7] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-048:1780:1839 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1781:1842 [2] NCCL INFO + PCI[24.0] - GPU/0-8b000 (22) +n116-040-048:1785:1838 [6] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-048:1780:1839 [1] NCCL INFO + PCI[24.0] - GPU/0-85000 (21) +n116-040-048:1781:1842 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1781:1842 [2] NCCL INFO + PCI[24.0] - GPU/0-8c000 (23) +n116-040-048:1780:1839 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1786:1840 [7] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-048:1781:1842 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1780:1839 [1] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-040-048:1781:1842 [2] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-040-048:1780:1839 [1] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-040-048:1785:1838 [6] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-048:1781:1842 [2] NCCL INFO ========================================== +n116-040-048:1780:1839 [1] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-040-048:1780:1839 [1] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-040-048:1786:1840 [7] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-048:1780:1839 [1] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-040-048:1780:1839 [1] NCCL INFO + PCI[24.0] - GPU/0-8b000 (22) +n116-040-048:1780:1839 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1781:1842 [2] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-048:1780:1839 [1] NCCL INFO + PCI[24.0] - GPU/0-8c000 (23) +n116-040-048:1786:1840 [7] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-040-048:1785:1838 [6] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-040-048:1780:1839 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-048:1780:1839 [1] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-040-048:1780:1839 [1] NCCL INFO ========================================== +n116-040-048:1781:1842 [2] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-048:1779:1841 [0] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-040-048:1786:1840 [7] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-040-048:1785:1838 [6] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-040-048:1780:1839 [1] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-048:1785:1838 [6] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-040-048:1781:1842 [2] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-048:1780:1839 [1] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-048:1786:1840 [7] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-048:1785:1838 [6] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-040-048:1781:1842 [2] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-048:1780:1839 [1] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-048:1786:1840 [7] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-048:1785:1838 [6] NCCL INFO Setting affinity for GPU 6 to 7fff,ffffffff,e0000000,0000007f,ffffffff,ffe00000,00000000 +n116-040-048:1780:1839 [1] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-048:1786:1840 [7] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-048:1785:1838 [6] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-040-048:1779:1841 [0] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-040-048:1781:1842 [2] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-048:1786:1840 [7] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-048:1780:1839 [1] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-048:1781:1842 [2] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-048:1786:1840 [7] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-040-048:1780:1839 [1] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-048:1781:1842 [2] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-040-048:1786:1840 [7] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-040-048:1780:1839 [1] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-040-048:1781:1842 [2] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-040-048:1785:1838 [6] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-040-048:1780:1839 [1] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-040-048:1781:1842 [2] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-048:1786:1840 [7] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-040-048:1780:1839 [1] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-048:1781:1842 [2] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-048:1786:1840 [7] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-040-048:1781:1842 [2] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-048:1780:1839 [1] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-048:1786:1840 [7] NCCL INFO Setting affinity for GPU 7 to 7fff,ffffffff,e0000000,0000007f,ffffffff,ffe00000,00000000 +n116-040-048:1781:1842 [2] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-048:1780:1839 [1] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-048:1786:1840 [7] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-040-048:1781:1842 [2] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-040-048:1780:1839 [1] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-048:1781:1842 [2] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-040-048:1780:1839 [1] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-040-048:1780:1839 [1] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-040-048:1781:1842 [2] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-040-048:1780:1839 [1] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-040-048:1781:1842 [2] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-040-048:1786:1840 [7] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-040-048:1780:1839 [1] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-040-048:1781:1842 [2] NCCL INFO Setting affinity for GPU 2 to 0fffffff,fffffe00,00000000,000fffff,fffffffe +n116-040-048:1781:1842 [2] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-040-048:1780:1839 [1] NCCL INFO Setting affinity for GPU 1 to 0fffffff,fffffe00,00000000,000fffff,fffffffe +n116-040-048:1780:1839 [1] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-040-048:1781:1842 [2] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-040-048:1780:1839 [1] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-040-048:1785:1838 [6] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-040-048:1785:1838 [6] NCCL INFO 0 : NET/0-0 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 NET/0-0 +n116-040-048:1785:1838 [6] NCCL INFO 1 : NET/0-1 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 NET/0-1 +n116-040-048:1785:1838 [6] NCCL INFO 2 : NET/0-2 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 NET/0-2 +n116-040-048:1785:1838 [6] NCCL INFO 3 : NET/0-3 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 NET/0-3 +n116-040-048:1785:1838 [6] NCCL INFO 4 : NET/0-4 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 NET/0-4 +n116-040-048:1785:1838 [6] NCCL INFO 5 : NET/0-5 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 NET/0-5 +n116-040-048:1785:1838 [6] NCCL INFO 6 : NET/0-6 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 NET/0-6 +n116-040-048:1785:1838 [6] NCCL INFO 7 : NET/0-7 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 NET/0-7 +n116-040-048:1785:1838 [6] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-040-048:1785:1838 [6] NCCL INFO 0 : NET/0-0 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 NET/0-0 +n116-040-048:1785:1838 [6] NCCL INFO 1 : NET/0-1 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 NET/0-1 +n116-040-048:1785:1838 [6] NCCL INFO 2 : NET/0-2 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 NET/0-2 +n116-040-048:1785:1838 [6] NCCL INFO 3 : NET/0-3 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 NET/0-3 +n116-040-048:1785:1838 [6] NCCL INFO 4 : NET/0-4 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 NET/0-4 +n116-040-048:1785:1838 [6] NCCL INFO 5 : NET/0-5 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 NET/0-5 +n116-040-048:1785:1838 [6] NCCL INFO 6 : NET/0-6 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 NET/0-6 +n116-040-048:1785:1838 [6] NCCL INFO 7 : NET/0-7 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 NET/0-7 +n116-040-048:1786:1840 [7] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-040-048:1786:1840 [7] NCCL INFO 0 : NET/0-0 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 NET/0-0 +n116-040-048:1786:1840 [7] NCCL INFO 1 : NET/0-1 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 NET/0-1 +n116-040-048:1786:1840 [7] NCCL INFO 2 : NET/0-2 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 NET/0-2 +n116-040-048:1786:1840 [7] NCCL INFO 3 : NET/0-3 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 NET/0-3 +n116-040-048:1786:1840 [7] NCCL INFO 4 : NET/0-4 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 NET/0-4 +n116-040-048:1786:1840 [7] NCCL INFO 5 : NET/0-5 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 NET/0-5 +n116-040-048:1786:1840 [7] NCCL INFO 6 : NET/0-6 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 NET/0-6 +n116-040-048:1786:1840 [7] NCCL INFO 7 : NET/0-7 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 NET/0-7 +n116-040-048:1786:1840 [7] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-040-048:1786:1840 [7] NCCL INFO 0 : NET/0-0 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 NET/0-0 +n116-040-048:1786:1840 [7] NCCL INFO 1 : NET/0-1 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 NET/0-1 +n116-040-048:1786:1840 [7] NCCL INFO 2 : NET/0-2 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 NET/0-2 +n116-040-048:1786:1840 [7] NCCL INFO 3 : NET/0-3 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 NET/0-3 +n116-040-048:1786:1840 [7] NCCL INFO 4 : NET/0-4 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 NET/0-4 +n116-040-048:1786:1840 [7] NCCL INFO 5 : NET/0-5 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 NET/0-5 +n116-040-048:1786:1840 [7] NCCL INFO 6 : NET/0-6 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 NET/0-6 +n116-040-048:1786:1840 [7] NCCL INFO 7 : NET/0-7 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 NET/0-7 +n116-040-048:1782:1835 [3] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-040-048:1782:1835 [3] NCCL INFO 0 : NET/0-0 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 NET/0-0 +n116-040-048:1782:1835 [3] NCCL INFO 1 : NET/0-1 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 NET/0-1 +n116-040-048:1782:1835 [3] NCCL INFO 2 : NET/0-2 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 NET/0-2 +n116-040-048:1782:1835 [3] NCCL INFO 3 : NET/0-3 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 NET/0-3 +n116-040-048:1782:1835 [3] NCCL INFO 4 : NET/0-4 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 NET/0-4 +n116-040-048:1782:1835 [3] NCCL INFO 5 : NET/0-5 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 NET/0-5 +n116-040-048:1782:1835 [3] NCCL INFO 6 : NET/0-6 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 NET/0-6 +n116-040-048:1782:1835 [3] NCCL INFO 7 : NET/0-7 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 NET/0-7 +n116-040-048:1784:1836 [5] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-040-048:1784:1836 [5] NCCL INFO 0 : NET/0-0 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 NET/0-0 +n116-040-048:1782:1835 [3] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-040-048:1784:1836 [5] NCCL INFO 1 : NET/0-1 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 NET/0-1 +n116-040-048:1782:1835 [3] NCCL INFO 0 : NET/0-0 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 NET/0-0 +n116-040-048:1784:1836 [5] NCCL INFO 2 : NET/0-2 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 NET/0-2 +n116-040-048:1782:1835 [3] NCCL INFO 1 : NET/0-1 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 NET/0-1 +n116-040-048:1784:1836 [5] NCCL INFO 3 : NET/0-3 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 NET/0-3 +n116-040-048:1782:1835 [3] NCCL INFO 2 : NET/0-2 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 NET/0-2 +n116-040-048:1784:1836 [5] NCCL INFO 4 : NET/0-4 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 NET/0-4 +n116-040-048:1782:1835 [3] NCCL INFO 3 : NET/0-3 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 NET/0-3 +n116-040-048:1784:1836 [5] NCCL INFO 5 : NET/0-5 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 NET/0-5 +n116-040-048:1782:1835 [3] NCCL INFO 4 : NET/0-4 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 NET/0-4 +n116-040-048:1784:1836 [5] NCCL INFO 6 : NET/0-6 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 NET/0-6 +n116-040-048:1782:1835 [3] NCCL INFO 5 : NET/0-5 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 NET/0-5 +n116-040-048:1784:1836 [5] NCCL INFO 7 : NET/0-7 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 NET/0-7 +n116-040-048:1782:1835 [3] NCCL INFO 6 : NET/0-6 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 NET/0-6 +n116-040-048:1782:1835 [3] NCCL INFO 7 : NET/0-7 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 NET/0-7 +n116-040-048:1781:1842 [2] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-040-048:1781:1842 [2] NCCL INFO 0 : NET/0-0 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 NET/0-0 +n116-040-048:1781:1842 [2] NCCL INFO 1 : NET/0-1 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 NET/0-1 +n116-040-048:1781:1842 [2] NCCL INFO 2 : NET/0-2 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 NET/0-2 +n116-040-048:1781:1842 [2] NCCL INFO 3 : NET/0-3 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 NET/0-3 +n116-040-048:1781:1842 [2] NCCL INFO 4 : NET/0-4 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 NET/0-4 +n116-040-048:1781:1842 [2] NCCL INFO 5 : NET/0-5 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 NET/0-5 +n116-040-048:1781:1842 [2] NCCL INFO 6 : NET/0-6 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 NET/0-6 +n116-040-048:1781:1842 [2] NCCL INFO 7 : NET/0-7 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 NET/0-7 +n116-040-048:1783:1837 [4] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-040-048:1783:1837 [4] NCCL INFO 0 : NET/0-0 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 NET/0-0 +n116-040-048:1783:1837 [4] NCCL INFO 1 : NET/0-1 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 NET/0-1 +n116-040-048:1783:1837 [4] NCCL INFO 2 : NET/0-2 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 NET/0-2 +n116-040-048:1783:1837 [4] NCCL INFO 3 : NET/0-3 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 NET/0-3 +n116-040-048:1783:1837 [4] NCCL INFO 4 : NET/0-4 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 NET/0-4 +n116-040-048:1783:1837 [4] NCCL INFO 5 : NET/0-5 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 NET/0-5 +n116-040-048:1783:1837 [4] NCCL INFO 6 : NET/0-6 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 NET/0-6 +n116-040-048:1784:1836 [5] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-040-048:1783:1837 [4] NCCL INFO 7 : NET/0-7 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 NET/0-7 +n116-040-048:1784:1836 [5] NCCL INFO 0 : NET/0-0 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 NET/0-0 +n116-040-048:1784:1836 [5] NCCL INFO 1 : NET/0-1 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 NET/0-1 +n116-040-048:1784:1836 [5] NCCL INFO 2 : NET/0-2 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 NET/0-2 +n116-040-048:1784:1836 [5] NCCL INFO 3 : NET/0-3 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 NET/0-3 +n116-040-048:1784:1836 [5] NCCL INFO 4 : NET/0-4 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 NET/0-4 +n116-040-048:1784:1836 [5] NCCL INFO 5 : NET/0-5 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 NET/0-5 +n116-040-048:1784:1836 [5] NCCL INFO 6 : NET/0-6 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 NET/0-6 +n116-040-048:1784:1836 [5] NCCL INFO 7 : NET/0-7 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 NET/0-7 +n116-040-048:1781:1842 [2] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-040-048:1781:1842 [2] NCCL INFO 0 : NET/0-0 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 NET/0-0 +n116-040-048:1781:1842 [2] NCCL INFO 1 : NET/0-1 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 NET/0-1 +n116-040-048:1781:1842 [2] NCCL INFO 2 : NET/0-2 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 NET/0-2 +n116-040-048:1781:1842 [2] NCCL INFO 3 : NET/0-3 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 NET/0-3 +n116-040-048:1781:1842 [2] NCCL INFO 4 : NET/0-4 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 NET/0-4 +n116-040-048:1781:1842 [2] NCCL INFO 5 : NET/0-5 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 NET/0-5 +n116-040-048:1781:1842 [2] NCCL INFO 6 : NET/0-6 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 NET/0-6 +n116-040-048:1781:1842 [2] NCCL INFO 7 : NET/0-7 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 NET/0-7 +n116-040-048:1783:1837 [4] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-040-048:1783:1837 [4] NCCL INFO 0 : NET/0-0 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 NET/0-0 +n116-040-048:1783:1837 [4] NCCL INFO 1 : NET/0-1 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 NET/0-1 +n116-040-048:1783:1837 [4] NCCL INFO 2 : NET/0-2 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 NET/0-2 +n116-040-048:1783:1837 [4] NCCL INFO 3 : NET/0-3 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 NET/0-3 +n116-040-048:1783:1837 [4] NCCL INFO 4 : NET/0-4 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 NET/0-4 +n116-040-048:1783:1837 [4] NCCL INFO 5 : NET/0-5 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 NET/0-5 +n116-040-048:1783:1837 [4] NCCL INFO 6 : NET/0-6 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 NET/0-6 +n116-040-048:1783:1837 [4] NCCL INFO 7 : NET/0-7 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 NET/0-7 +n116-040-048:1780:1839 [1] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-040-048:1780:1839 [1] NCCL INFO 0 : NET/0-0 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 NET/0-0 +n116-040-048:1780:1839 [1] NCCL INFO 1 : NET/0-1 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 NET/0-1 +n116-040-048:1780:1839 [1] NCCL INFO 2 : NET/0-2 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 NET/0-2 +n116-040-048:1780:1839 [1] NCCL INFO 3 : NET/0-3 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 NET/0-3 +n116-040-048:1780:1839 [1] NCCL INFO 4 : NET/0-4 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 NET/0-4 +n116-040-048:1780:1839 [1] NCCL INFO 5 : NET/0-5 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 NET/0-5 +n116-040-048:1780:1839 [1] NCCL INFO 6 : NET/0-6 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 NET/0-6 +n116-040-048:1780:1839 [1] NCCL INFO 7 : NET/0-7 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 NET/0-7 +n116-040-048:1779:1841 [0] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-040-048:1779:1841 [0] NCCL INFO 0 : NET/0-0 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 NET/0-0 +n116-040-048:1779:1841 [0] NCCL INFO 1 : NET/0-1 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 NET/0-1 +n116-040-048:1779:1841 [0] NCCL INFO 2 : NET/0-2 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 NET/0-2 +n116-040-048:1779:1841 [0] NCCL INFO 3 : NET/0-3 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 NET/0-3 +n116-040-048:1779:1841 [0] NCCL INFO 4 : NET/0-4 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 NET/0-4 +n116-040-048:1779:1841 [0] NCCL INFO 5 : NET/0-5 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 NET/0-5 +n116-040-048:1779:1841 [0] NCCL INFO 6 : NET/0-6 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 NET/0-6 +n116-040-048:1779:1841 [0] NCCL INFO 7 : NET/0-7 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 NET/0-7 +n116-040-048:1780:1839 [1] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-040-048:1780:1839 [1] NCCL INFO 0 : NET/0-0 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 NET/0-0 +n116-040-048:1780:1839 [1] NCCL INFO 1 : NET/0-1 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 NET/0-1 +n116-040-048:1780:1839 [1] NCCL INFO 2 : NET/0-2 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 NET/0-2 +n116-040-048:1780:1839 [1] NCCL INFO 3 : NET/0-3 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 NET/0-3 +n116-040-048:1780:1839 [1] NCCL INFO 4 : NET/0-4 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 NET/0-4 +n116-040-048:1780:1839 [1] NCCL INFO 5 : NET/0-5 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 NET/0-5 +n116-040-048:1780:1839 [1] NCCL INFO 6 : NET/0-6 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 NET/0-6 +n116-040-048:1780:1839 [1] NCCL INFO 7 : NET/0-7 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 NET/0-7 +n116-040-048:1779:1841 [0] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-040-048:1779:1841 [0] NCCL INFO 0 : NET/0-0 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 NET/0-0 +n116-040-048:1779:1841 [0] NCCL INFO 1 : NET/0-1 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 NET/0-1 +n116-040-048:1779:1841 [0] NCCL INFO 2 : NET/0-2 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 NET/0-2 +n116-040-048:1779:1841 [0] NCCL INFO 3 : NET/0-3 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 NET/0-3 +n116-040-048:1779:1841 [0] NCCL INFO 4 : NET/0-4 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 NET/0-4 +n116-040-048:1779:1841 [0] NCCL INFO 5 : NET/0-5 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 NET/0-5 +n116-040-048:1779:1841 [0] NCCL INFO 6 : NET/0-6 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 NET/0-6 +n116-040-048:1779:1841 [0] NCCL INFO 7 : NET/0-7 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 NET/0-7 +n116-040-048:1782:1835 [3] NCCL INFO comm 0xc150970 rank 19 nRanks 32 nNodes 4 localRanks 8 localRank 3 MNNVL 0 +n116-040-048:1784:1836 [5] NCCL INFO comm 0xc314c50 rank 21 nRanks 32 nNodes 4 localRanks 8 localRank 5 MNNVL 0 +n116-040-048:1781:1842 [2] NCCL INFO comm 0xcd2af00 rank 18 nRanks 32 nNodes 4 localRanks 8 localRank 2 MNNVL 0 +n116-040-048:1783:1837 [4] NCCL INFO comm 0xb926bd0 rank 20 nRanks 32 nNodes 4 localRanks 8 localRank 4 MNNVL 0 +n116-040-048:1780:1839 [1] NCCL INFO comm 0xbfe3fc0 rank 17 nRanks 32 nNodes 4 localRanks 8 localRank 1 MNNVL 0 +n116-040-048:1784:1836 [5] NCCL INFO Tree 4 : 20 -> 21 -> 22/12/-1 +n116-040-048:1784:1836 [5] NCCL INFO Tree 12 : 20 -> 21 -> 22/-1/-1 +n116-040-048:1784:1836 [5] NCCL INFO Tree 5 : 5 -> 21 -> 20/29/-1 +n116-040-048:1781:1842 [2] NCCL INFO Tree 2 : 2 -> 18 -> 19/26/-1 +n116-040-048:1784:1836 [5] NCCL INFO Tree 13 : 12 -> 21 -> 20/-1/-1 +n116-040-048:1781:1842 [2] NCCL INFO Tree 10 : 11 -> 18 -> 19/-1/-1 +n116-040-048:1781:1842 [2] NCCL INFO Tree 3 : 19 -> 18 -> 17/11/-1 +n116-040-048:1781:1842 [2] NCCL INFO Tree 11 : 19 -> 18 -> 17/-1/-1 +n116-040-048:1783:1837 [4] NCCL INFO Tree 4 : 4 -> 20 -> 21/28/-1 +n116-040-048:1784:1836 [5] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-040-048:1782:1835 [3] NCCL INFO Tree 2 : 18 -> 19 -> 20/10/-1 +n116-040-048:1783:1837 [4] NCCL INFO Tree 12 : 13 -> 20 -> 21/-1/-1 +n116-040-048:1782:1835 [3] NCCL INFO Tree 10 : 18 -> 19 -> 20/-1/-1 +n116-040-048:1783:1837 [4] NCCL INFO Tree 5 : 21 -> 20 -> 19/13/-1 +n116-040-048:1783:1837 [4] NCCL INFO Tree 13 : 21 -> 20 -> 19/-1/-1 +n116-040-048:1782:1835 [3] NCCL INFO Tree 3 : 3 -> 19 -> 18/27/-1 +n116-040-048:1781:1842 [2] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-040-048:1782:1835 [3] NCCL INFO Tree 11 : 10 -> 19 -> 18/-1/-1 +n116-040-048:1779:1841 [0] NCCL INFO comm 0xb7cca60 rank 16 nRanks 32 nNodes 4 localRanks 8 localRank 0 MNNVL 0 +n116-040-048:1783:1837 [4] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-040-048:1782:1835 [3] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-040-048:1784:1836 [5] NCCL INFO Ring 00 : 22 -> 21 -> 20 +n116-040-048:1781:1842 [2] NCCL INFO Ring 00 : 19 -> 18 -> 17 +n116-040-048:1784:1836 [5] NCCL INFO Ring 01 : 20 -> 21 -> 22 +n116-040-048:1781:1842 [2] NCCL INFO Ring 01 : 17 -> 18 -> 19 +n116-040-048:1784:1836 [5] NCCL INFO Ring 02 : 22 -> 21 -> 20 +n116-040-048:1781:1842 [2] NCCL INFO Ring 02 : 11 -> 18 -> 17 +n116-040-048:1783:1837 [4] NCCL INFO Ring 00 : 21 -> 20 -> 19 +n116-040-048:1782:1835 [3] NCCL INFO Ring 00 : 20 -> 19 -> 18 +n116-040-048:1780:1839 [1] NCCL INFO Tree 0 : 16 -> 17 -> 18/8/-1 +n116-040-048:1784:1836 [5] NCCL INFO Ring 03 : 20 -> 21 -> 22 +n116-040-048:1781:1842 [2] NCCL INFO Ring 03 : 17 -> 18 -> 27 +n116-040-048:1783:1837 [4] NCCL INFO Ring 01 : 19 -> 20 -> 21 +n116-040-048:1782:1835 [3] NCCL INFO Ring 01 : 18 -> 19 -> 20 +n116-040-048:1780:1839 [1] NCCL INFO Tree 8 : 16 -> 17 -> 18/-1/-1 +n116-040-048:1783:1837 [4] NCCL INFO Ring 02 : 21 -> 20 -> 19 +n116-040-048:1784:1836 [5] NCCL INFO Ring 04 : 22 -> 21 -> 28 +n116-040-048:1781:1842 [2] NCCL INFO Ring 04 : 19 -> 18 -> 17 +n116-040-048:1779:1841 [0] NCCL INFO Tree 0 : 0 -> 16 -> 17/24/-1 +n116-040-048:1784:1836 [5] NCCL INFO Ring 05 : 12 -> 21 -> 22 +n116-040-048:1781:1842 [2] NCCL INFO Ring 05 : 17 -> 18 -> 19 +n116-040-048:1782:1835 [3] NCCL INFO Ring 02 : 20 -> 19 -> 26 +n116-040-048:1780:1839 [1] NCCL INFO Tree 1 : 1 -> 17 -> 16/25/-1 +n116-040-048:1783:1837 [4] NCCL INFO Ring 03 : 19 -> 20 -> 21 +n116-040-048:1779:1841 [0] NCCL INFO Tree 8 : 9 -> 16 -> 17/-1/-1 +n116-040-048:1781:1842 [2] NCCL INFO Ring 06 : 19 -> 18 -> 17 +n116-040-048:1784:1836 [5] NCCL INFO Ring 06 : 22 -> 21 -> 20 +n116-040-048:1782:1835 [3] NCCL INFO Ring 03 : 10 -> 19 -> 20 +n116-040-048:1780:1839 [1] NCCL INFO Tree 9 : 8 -> 17 -> 16/-1/-1 +n116-040-048:1783:1837 [4] NCCL INFO Ring 04 : 13 -> 20 -> 19 +n116-040-048:1779:1841 [0] NCCL INFO Tree 1 : 17 -> 16 -> 23/9/-1 +n116-040-048:1782:1835 [3] NCCL INFO Ring 04 : 20 -> 19 -> 18 +n116-040-048:1781:1842 [2] NCCL INFO Ring 07 : 17 -> 18 -> 19 +n116-040-048:1785:1838 [6] NCCL INFO comm 0xc067bd0 rank 22 nRanks 32 nNodes 4 localRanks 8 localRank 6 MNNVL 0 +n116-040-048:1784:1836 [5] NCCL INFO Ring 07 : 20 -> 21 -> 22 +n116-040-048:1782:1835 [3] NCCL INFO Ring 05 : 18 -> 19 -> 20 +n116-040-048:1779:1841 [0] NCCL INFO Tree 9 : 17 -> 16 -> 23/-1/-1 +n116-040-048:1783:1837 [4] NCCL INFO Ring 05 : 19 -> 20 -> 29 +n116-040-048:1780:1839 [1] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-040-048:1781:1842 [2] NCCL INFO Ring 08 : 19 -> 18 -> 17 +n116-040-048:1784:1836 [5] NCCL INFO Ring 08 : 22 -> 21 -> 20 +n116-040-048:1782:1835 [3] NCCL INFO Ring 06 : 20 -> 19 -> 18 +n116-040-048:1786:1840 [7] NCCL INFO comm 0xcc263e0 rank 23 nRanks 32 nNodes 4 localRanks 8 localRank 7 MNNVL 0 +n116-040-048:1784:1836 [5] NCCL INFO Ring 09 : 20 -> 21 -> 22 +n116-040-048:1783:1837 [4] NCCL INFO Ring 06 : 21 -> 20 -> 19 +n116-040-048:1781:1842 [2] NCCL INFO Ring 09 : 17 -> 18 -> 19 +n116-040-048:1779:1841 [0] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-040-048:1782:1835 [3] NCCL INFO Ring 07 : 18 -> 19 -> 20 +n116-040-048:1783:1837 [4] NCCL INFO Ring 07 : 19 -> 20 -> 21 +n116-040-048:1784:1836 [5] NCCL INFO Ring 10 : 22 -> 21 -> 20 +n116-040-048:1781:1842 [2] NCCL INFO Ring 10 : 11 -> 18 -> 17 +n116-040-048:1782:1835 [3] NCCL INFO Ring 08 : 20 -> 19 -> 18 +n116-040-048:1783:1837 [4] NCCL INFO Ring 08 : 21 -> 20 -> 19 +n116-040-048:1784:1836 [5] NCCL INFO Ring 11 : 20 -> 21 -> 22 +n116-040-048:1781:1842 [2] NCCL INFO Ring 11 : 17 -> 18 -> 27 +n116-040-048:1780:1839 [1] NCCL INFO Ring 00 : 18 -> 17 -> 24 +n116-040-048:1784:1836 [5] NCCL INFO Ring 12 : 22 -> 21 -> 28 +n116-040-048:1785:1838 [6] NCCL INFO Tree 6 : 6 -> 22 -> 23/30/-1 +n116-040-048:1783:1837 [4] NCCL INFO Ring 09 : 19 -> 20 -> 21 +n116-040-048:1782:1835 [3] NCCL INFO Ring 09 : 18 -> 19 -> 20 +n116-040-048:1781:1842 [2] NCCL INFO Ring 12 : 19 -> 18 -> 17 +n116-040-048:1780:1839 [1] NCCL INFO Ring 01 : 8 -> 17 -> 18 +n116-040-048:1779:1841 [0] NCCL INFO Ring 00 : 9 -> 16 -> 23 +n116-040-048:1781:1842 [2] NCCL INFO Ring 13 : 17 -> 18 -> 19 +n116-040-048:1784:1836 [5] NCCL INFO Ring 13 : 12 -> 21 -> 22 +n116-040-048:1785:1838 [6] NCCL INFO Tree 14 : 15 -> 22 -> 23/-1/-1 +n116-040-048:1780:1839 [1] NCCL INFO Ring 02 : 18 -> 17 -> 16 +n116-040-048:1781:1842 [2] NCCL INFO Ring 14 : 19 -> 18 -> 17 +n116-040-048:1780:1839 [1] NCCL INFO Ring 03 : 16 -> 17 -> 18 +n116-040-048:1783:1837 [4] NCCL INFO Ring 10 : 21 -> 20 -> 19 +n116-040-048:1782:1835 [3] NCCL INFO Ring 10 : 20 -> 19 -> 26 +n116-040-048:1786:1840 [7] NCCL INFO Tree 6 : 22 -> 23 -> 16/14/-1 +n116-040-048:1786:1840 [7] NCCL INFO Tree 14 : 22 -> 23 -> 16/-1/-1 +n116-040-048:1779:1841 [0] NCCL INFO Ring 01 : 23 -> 16 -> 25 +n116-040-048:1784:1836 [5] NCCL INFO Ring 14 : 22 -> 21 -> 20 +n116-040-048:1785:1838 [6] NCCL INFO Tree 7 : 23 -> 22 -> 21/15/-1 +n116-040-048:1781:1842 [2] NCCL INFO Ring 15 : 17 -> 18 -> 19 +n116-040-048:1780:1839 [1] NCCL INFO Ring 04 : 18 -> 17 -> 16 +n116-040-048:1785:1838 [6] NCCL INFO Tree 15 : 23 -> 22 -> 21/-1/-1 +n116-040-048:1786:1840 [7] NCCL INFO Tree 7 : 7 -> 23 -> 22/31/-1 +n116-040-048:1783:1837 [4] NCCL INFO Ring 11 : 19 -> 20 -> 21 +n116-040-048:1782:1835 [3] NCCL INFO Ring 11 : 10 -> 19 -> 20 +n116-040-048:1779:1841 [0] NCCL INFO Ring 02 : 17 -> 16 -> 23 +n116-040-048:1784:1836 [5] NCCL INFO Ring 15 : 20 -> 21 -> 22 +n116-040-048:1780:1839 [1] NCCL INFO Ring 05 : 16 -> 17 -> 18 +n116-040-048:1781:1842 [2] NCCL INFO Trees [0] 19/-1/-1->18->17 [1] -1/-1/-1->18->19 [2] 19/26/-1->18->2 [3] 17/11/-1->18->19 [4] 19/-1/-1->18->17 [5] 17/-1/-1->18->19 [6] 19/-1/-1->18->17 [7] 17/-1/-1->18->19 [8] 19/-1/-1->18->17 [9] -1/-1/-1->18->19 [10] 19/-1/-1->18->11 [11] 17/-1/-1->18->19 [12] 19/-1/-1->18->17 [13] 17/-1/-1->18->19 [14] 19/-1/-1->18->17 [15] 17/-1/-1->18->19 +n116-040-048:1786:1840 [7] NCCL INFO Tree 15 : 14 -> 23 -> 22/-1/-1 +n116-040-048:1780:1839 [1] NCCL INFO Ring 06 : 18 -> 17 -> 16 +n116-040-048:1783:1837 [4] NCCL INFO Ring 12 : 13 -> 20 -> 19 +n116-040-048:1782:1835 [3] NCCL INFO Ring 12 : 20 -> 19 -> 18 +n116-040-048:1781:1842 [2] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-040-048:1785:1838 [6] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-040-048:1782:1835 [3] NCCL INFO Ring 13 : 18 -> 19 -> 20 +n116-040-048:1781:1842 [2] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-040-048:1779:1841 [0] NCCL INFO Ring 03 : 23 -> 16 -> 17 +n116-040-048:1784:1836 [5] NCCL INFO Trees [0] 22/-1/-1->21->20 [1] 20/-1/-1->21->22 [2] 22/-1/-1->21->20 [3] 20/-1/-1->21->22 [4] 22/12/-1->21->20 [5] 20/29/-1->21->5 [6] -1/-1/-1->21->20 [7] 20/-1/-1->21->22 [8] 22/-1/-1->21->20 [9] 20/-1/-1->21->22 [10] 22/-1/-1->21->20 [11] 20/-1/-1->21->22 [12] 22/-1/-1->21->20 [13] 20/-1/-1->21->12 [14] -1/-1/-1->21->20 [15] 20/-1/-1->21->22 +n116-040-048:1780:1839 [1] NCCL INFO Ring 07 : 16 -> 17 -> 18 +n116-040-048:1783:1837 [4] NCCL INFO Ring 13 : 19 -> 20 -> 29 +n116-040-048:1786:1840 [7] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-040-048:1784:1836 [5] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-040-048:1782:1835 [3] NCCL INFO Ring 14 : 20 -> 19 -> 18 +n116-040-048:1781:1842 [2] NCCL INFO P2P Chunksize set to 524288 +n116-040-048:1779:1841 [0] NCCL INFO Ring 04 : 17 -> 16 -> 23 +n116-040-048:1780:1839 [1] NCCL INFO Ring 08 : 18 -> 17 -> 24 +n116-040-048:1783:1837 [4] NCCL INFO Ring 14 : 21 -> 20 -> 19 +n116-040-048:1784:1836 [5] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-040-048:1782:1835 [3] NCCL INFO Ring 15 : 18 -> 19 -> 20 +n116-040-048:1783:1837 [4] NCCL INFO Ring 15 : 19 -> 20 -> 21 +n116-040-048:1779:1841 [0] NCCL INFO Ring 05 : 23 -> 16 -> 17 +n116-040-048:1780:1839 [1] NCCL INFO Ring 09 : 8 -> 17 -> 18 +n116-040-048:1784:1836 [5] NCCL INFO P2P Chunksize set to 524288 +n116-040-048:1782:1835 [3] NCCL INFO Trees [0] 20/-1/-1->19->18 [1] 18/-1/-1->19->20 [2] 20/10/-1->19->18 [3] 18/27/-1->19->3 [4] -1/-1/-1->19->18 [5] 18/-1/-1->19->20 [6] 20/-1/-1->19->18 [7] 18/-1/-1->19->20 [8] 20/-1/-1->19->18 [9] 18/-1/-1->19->20 [10] 20/-1/-1->19->18 [11] 18/-1/-1->19->10 [12] -1/-1/-1->19->18 [13] 18/-1/-1->19->20 [14] 20/-1/-1->19->18 [15] 18/-1/-1->19->20 +n116-040-048:1780:1839 [1] NCCL INFO Ring 10 : 18 -> 17 -> 16 +n116-040-048:1783:1837 [4] NCCL INFO Trees [0] 21/-1/-1->20->19 [1] 19/-1/-1->20->21 [2] 21/-1/-1->20->19 [3] -1/-1/-1->20->21 [4] 21/28/-1->20->4 [5] 19/13/-1->20->21 [6] 21/-1/-1->20->19 [7] 19/-1/-1->20->21 [8] 21/-1/-1->20->19 [9] 19/-1/-1->20->21 [10] 21/-1/-1->20->19 [11] -1/-1/-1->20->21 [12] 21/-1/-1->20->13 [13] 19/-1/-1->20->21 [14] 21/-1/-1->20->19 [15] 19/-1/-1->20->21 +n116-040-048:1785:1838 [6] NCCL INFO Ring 00 : 23 -> 22 -> 21 +n116-040-048:1779:1841 [0] NCCL INFO Ring 06 : 17 -> 16 -> 23 +n116-040-048:1780:1839 [1] NCCL INFO Ring 11 : 16 -> 17 -> 18 +n116-040-048:1782:1835 [3] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-040-048:1785:1838 [6] NCCL INFO Ring 01 : 21 -> 22 -> 23 +n116-040-048:1780:1839 [1] NCCL INFO Ring 12 : 18 -> 17 -> 16 +n116-040-048:1786:1840 [7] NCCL INFO Ring 00 : 16 -> 23 -> 22 +n116-040-048:1783:1837 [4] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-040-048:1783:1837 [4] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-040-048:1779:1841 [0] NCCL INFO Ring 07 : 23 -> 16 -> 17 +n116-040-048:1782:1835 [3] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-040-048:1785:1838 [6] NCCL INFO Ring 02 : 23 -> 22 -> 21 +n116-040-048:1780:1839 [1] NCCL INFO Ring 13 : 16 -> 17 -> 18 +n116-040-048:1786:1840 [7] NCCL INFO Ring 01 : 22 -> 23 -> 16 +n116-040-048:1785:1838 [6] NCCL INFO Ring 03 : 21 -> 22 -> 23 +n116-040-048:1783:1837 [4] NCCL INFO P2P Chunksize set to 524288 +n116-040-048:1779:1841 [0] NCCL INFO Ring 08 : 9 -> 16 -> 23 +n116-040-048:1782:1835 [3] NCCL INFO P2P Chunksize set to 524288 +n116-040-048:1780:1839 [1] NCCL INFO Ring 14 : 18 -> 17 -> 16 +n116-040-048:1779:1841 [0] NCCL INFO Ring 09 : 23 -> 16 -> 25 +n116-040-048:1786:1840 [7] NCCL INFO Ring 02 : 16 -> 23 -> 22 +n116-040-048:1785:1838 [6] NCCL INFO Ring 04 : 23 -> 22 -> 21 +n116-040-048:1780:1839 [1] NCCL INFO Ring 15 : 16 -> 17 -> 18 +n116-040-048:1779:1841 [0] NCCL INFO Ring 10 : 17 -> 16 -> 23 +n116-040-048:1786:1840 [7] NCCL INFO Ring 03 : 22 -> 23 -> 16 +n116-040-048:1779:1841 [0] NCCL INFO Ring 11 : 23 -> 16 -> 17 +n116-040-048:1785:1838 [6] NCCL INFO Ring 05 : 21 -> 22 -> 23 +n116-040-048:1786:1840 [7] NCCL INFO Ring 04 : 16 -> 23 -> 22 +n116-040-048:1780:1839 [1] NCCL INFO Trees [0] 18/8/-1->17->16 [1] 16/25/-1->17->1 [2] -1/-1/-1->17->16 [3] 16/-1/-1->17->18 [4] 18/-1/-1->17->16 [5] 16/-1/-1->17->18 [6] 18/-1/-1->17->16 [7] 16/-1/-1->17->18 [8] 18/-1/-1->17->16 [9] 16/-1/-1->17->8 [10] -1/-1/-1->17->16 [11] 16/-1/-1->17->18 [12] 18/-1/-1->17->16 [13] 16/-1/-1->17->18 [14] 18/-1/-1->17->16 [15] 16/-1/-1->17->18 +n116-040-048:1779:1841 [0] NCCL INFO Ring 12 : 17 -> 16 -> 23 +n116-040-048:1785:1838 [6] NCCL INFO Ring 06 : 15 -> 22 -> 21 +n116-040-048:1779:1841 [0] NCCL INFO Ring 13 : 23 -> 16 -> 17 +n116-040-048:1785:1838 [6] NCCL INFO Ring 07 : 21 -> 22 -> 31 +n116-040-048:1786:1840 [7] NCCL INFO Ring 05 : 22 -> 23 -> 16 +n116-040-048:1780:1839 [1] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-040-048:1780:1839 [1] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-040-048:1785:1838 [6] NCCL INFO Ring 08 : 23 -> 22 -> 21 +n116-040-048:1779:1841 [0] NCCL INFO Ring 14 : 17 -> 16 -> 23 +n116-040-048:1786:1840 [7] NCCL INFO Ring 06 : 16 -> 23 -> 30 +n116-040-048:1780:1839 [1] NCCL INFO P2P Chunksize set to 524288 +n116-040-048:1785:1838 [6] NCCL INFO Ring 09 : 21 -> 22 -> 23 +n116-040-048:1779:1841 [0] NCCL INFO Ring 15 : 23 -> 16 -> 17 +n116-040-048:1786:1840 [7] NCCL INFO Ring 07 : 14 -> 23 -> 16 +n116-040-048:1785:1838 [6] NCCL INFO Ring 10 : 23 -> 22 -> 21 +n116-040-048:1779:1841 [0] NCCL INFO Trees [0] 17/24/-1->16->0 [1] 23/9/-1->16->17 [2] 17/-1/-1->16->23 [3] 23/-1/-1->16->17 [4] 17/-1/-1->16->23 [5] 23/-1/-1->16->17 [6] 17/-1/-1->16->23 [7] -1/-1/-1->16->17 [8] 17/-1/-1->16->9 [9] 23/-1/-1->16->17 [10] 17/-1/-1->16->23 [11] 23/-1/-1->16->17 [12] 17/-1/-1->16->23 [13] 23/-1/-1->16->17 [14] 17/-1/-1->16->23 [15] -1/-1/-1->16->17 +n116-040-048:1786:1840 [7] NCCL INFO Ring 08 : 16 -> 23 -> 22 +n116-040-048:1785:1838 [6] NCCL INFO Ring 11 : 21 -> 22 -> 23 +n116-040-048:1779:1841 [0] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-040-048:1785:1838 [6] NCCL INFO Ring 12 : 23 -> 22 -> 21 +n116-040-048:1786:1840 [7] NCCL INFO Ring 09 : 22 -> 23 -> 16 +n116-040-048:1779:1841 [0] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-040-048:1785:1838 [6] NCCL INFO Ring 13 : 21 -> 22 -> 23 +n116-040-048:1786:1840 [7] NCCL INFO Ring 10 : 16 -> 23 -> 22 +n116-040-048:1779:1841 [0] NCCL INFO P2P Chunksize set to 524288 +n116-040-048:1785:1838 [6] NCCL INFO Ring 14 : 15 -> 22 -> 21 +n116-040-048:1786:1840 [7] NCCL INFO Ring 11 : 22 -> 23 -> 16 +n116-040-048:1785:1838 [6] NCCL INFO Ring 15 : 21 -> 22 -> 31 +n116-040-048:1786:1840 [7] NCCL INFO Ring 12 : 16 -> 23 -> 22 +n116-040-048:1785:1838 [6] NCCL INFO Trees [0] 23/-1/-1->22->21 [1] 21/-1/-1->22->23 [2] 23/-1/-1->22->21 [3] 21/-1/-1->22->23 [4] 23/-1/-1->22->21 [5] -1/-1/-1->22->23 [6] 23/30/-1->22->6 [7] 21/15/-1->22->23 [8] 23/-1/-1->22->21 [9] 21/-1/-1->22->23 [10] 23/-1/-1->22->21 [11] 21/-1/-1->22->23 [12] 23/-1/-1->22->21 [13] -1/-1/-1->22->23 [14] 23/-1/-1->22->15 [15] 21/-1/-1->22->23 +n116-040-048:1786:1840 [7] NCCL INFO Ring 13 : 22 -> 23 -> 16 +n116-040-048:1786:1840 [7] NCCL INFO Ring 14 : 16 -> 23 -> 30 +n116-040-048:1785:1838 [6] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-040-048:1785:1838 [6] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-040-048:1786:1840 [7] NCCL INFO Ring 15 : 14 -> 23 -> 16 +n116-040-048:1785:1838 [6] NCCL INFO P2P Chunksize set to 524288 +n116-040-048:1786:1840 [7] NCCL INFO Trees [0] -1/-1/-1->23->22 [1] 22/-1/-1->23->16 [2] 16/-1/-1->23->22 [3] 22/-1/-1->23->16 [4] 16/-1/-1->23->22 [5] 22/-1/-1->23->16 [6] 16/14/-1->23->22 [7] 22/31/-1->23->7 [8] -1/-1/-1->23->22 [9] 22/-1/-1->23->16 [10] 16/-1/-1->23->22 [11] 22/-1/-1->23->16 [12] 16/-1/-1->23->22 [13] 22/-1/-1->23->16 [14] 16/-1/-1->23->22 [15] 22/-1/-1->23->14 +n116-040-048:1786:1840 [7] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-040-048:1786:1840 [7] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-040-048:1786:1840 [7] NCCL INFO P2P Chunksize set to 524288 +n116-040-048:1781:1876 [2] NCCL INFO [Proxy Service] Device 2 CPU core 110 +n116-040-048:1782:1877 [3] NCCL INFO [Proxy Service] Device 3 CPU core 46 +n116-040-048:1780:1884 [1] NCCL INFO [Proxy Service UDS] Device 1 CPU core 144 +n116-040-048:1784:1880 [5] NCCL INFO [Proxy Service UDS] Device 5 CPU core 61 +n116-040-048:1779:1885 [0] NCCL INFO [Proxy Service] Device 0 CPU core 4 +n116-040-048:1785:1883 [6] NCCL INFO [Proxy Service] Device 6 CPU core 64 +n116-040-048:1783:1886 [4] NCCL INFO [Proxy Service UDS] Device 4 CPU core 65 +n116-040-048:1779:1890 [0] NCCL INFO [Proxy Service UDS] Device 0 CPU core 5 +n116-040-048:1783:1879 [4] NCCL INFO [Proxy Service] Device 4 CPU core 161 +n116-040-048:1786:1887 [7] NCCL INFO [Proxy Service] Device 7 CPU core 89 +n116-040-048:1780:1878 [1] NCCL INFO [Proxy Service] Device 1 CPU core 19 +n116-040-048:1784:1875 [5] NCCL INFO [Proxy Service] Device 5 CPU core 206 +n116-040-048:1781:1881 [2] NCCL INFO [Proxy Service UDS] Device 2 CPU core 21 +n116-040-048:1782:1882 [3] NCCL INFO [Proxy Service UDS] Device 3 CPU core 134 +n116-040-048:1785:1888 [6] NCCL INFO [Proxy Service UDS] Device 6 CPU core 66 +n116-040-048:1786:1889 [7] NCCL INFO [Proxy Service UDS] Device 7 CPU core 95 +n116-040-048:1779:1841 [0] NCCL INFO NCCL_PROTO set by environment to Simple +n116-040-048:1779:1841 [0] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-040-048:1779:1841 [0] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512 +n116-040-048:1779:1841 [0] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 2 p2p channels per peer +n116-040-048:1781:1842 [2] NCCL INFO NCCL_PROTO set by environment to Simple +n116-040-048:1781:1842 [2] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-040-048:1781:1842 [2] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512 +n116-040-048:1781:1842 [2] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 2 p2p channels per peer +n116-040-048:1786:1840 [7] NCCL INFO NCCL_PROTO set by environment to Simple +n116-040-048:1786:1840 [7] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-040-048:1786:1840 [7] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512 +n116-040-048:1786:1840 [7] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 2 p2p channels per peer +n116-040-048:1780:1839 [1] NCCL INFO NCCL_PROTO set by environment to Simple +n116-040-048:1780:1839 [1] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-040-048:1780:1839 [1] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512 +n116-040-048:1780:1839 [1] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 2 p2p channels per peer +n116-040-048:1782:1835 [3] NCCL INFO NCCL_PROTO set by environment to Simple +n116-040-048:1782:1835 [3] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-040-048:1782:1835 [3] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512 +n116-040-048:1782:1835 [3] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 2 p2p channels per peer +n116-040-048:1783:1837 [4] NCCL INFO NCCL_PROTO set by environment to Simple +n116-040-048:1783:1837 [4] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-040-048:1783:1837 [4] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512 +n116-040-048:1783:1837 [4] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 2 p2p channels per peer +n116-040-048:1784:1836 [5] NCCL INFO NCCL_PROTO set by environment to Simple +n116-040-048:1784:1836 [5] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-040-048:1784:1836 [5] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512 +n116-040-048:1784:1836 [5] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 2 p2p channels per peer +n116-040-048:1785:1838 [6] NCCL INFO NCCL_PROTO set by environment to Simple +n116-040-048:1785:1838 [6] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-040-048:1785:1838 [6] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512 +n116-040-048:1785:1838 [6] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 2 p2p channels per peer +n116-040-048:1782:1835 [3] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-040-048:1782:1835 [3] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-040-048:1782:1835 [3] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-040-048:1782:1835 [3] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-040-048:1782:1835 [3] NCCL INFO ncclCommInitRankConfig comm 0xc150970 rank 19 nranks 32 cudaDev 3 nvmlDev 3 busId c000 commId 0x2a4d35ea441cbcba - Init COMPLETE +n116-040-048:1782:1835 [3] NCCL INFO Init timings - ncclCommInitRankConfig: rank 19 nranks 32 total 6.50 (kernels 0.69, alloc 3.40, bootstrap 0.85, allgathers 0.14, topo 0.73, graphs 0.08, connections 0.04, rest 0.57) +n116-040-048:1784:1836 [5] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-040-048:1784:1836 [5] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-040-048:1784:1836 [5] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-040-048:1784:1836 [5] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-040-048:1784:1836 [5] NCCL INFO ncclCommInitRankConfig comm 0xc314c50 rank 21 nranks 32 cudaDev 5 nvmlDev 5 busId 85000 commId 0x2a4d35ea441cbcba - Init COMPLETE +n116-040-048:1784:1836 [5] NCCL INFO Init timings - ncclCommInitRankConfig: rank 21 nranks 32 total 6.50 (kernels 0.68, alloc 3.40, bootstrap 0.85, allgathers 0.14, topo 0.73, graphs 0.08, connections 0.04, rest 0.57) +n116-040-048:1783:1837 [4] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-040-048:1783:1837 [4] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-040-048:1783:1837 [4] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-040-048:1783:1837 [4] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-040-048:1783:1837 [4] NCCL INFO ncclCommInitRankConfig comm 0xb926bd0 rank 20 nranks 32 cudaDev 4 nvmlDev 4 busId 84000 commId 0x2a4d35ea441cbcba - Init COMPLETE +n116-040-048:1783:1837 [4] NCCL INFO Init timings - ncclCommInitRankConfig: rank 20 nranks 32 total 6.48 (kernels 0.67, alloc 3.40, bootstrap 0.85, allgathers 0.14, topo 0.73, graphs 0.08, connections 0.04, rest 0.57) +n116-040-048:1782:1891 [3] NCCL INFO Channel 00/0 : 19[3] -> 20[4] via P2P/CUMEM +n116-040-048:1780:1839 [1] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-040-048:1780:1839 [1] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-040-048:1780:1839 [1] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-040-048:1780:1839 [1] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-040-048:1780:1839 [1] NCCL INFO ncclCommInitRankConfig comm 0xbfe3fc0 rank 17 nranks 32 cudaDev 1 nvmlDev 1 busId 5000 commId 0x2a4d35ea441cbcba - Init COMPLETE +n116-040-048:1780:1839 [1] NCCL INFO Init timings - ncclCommInitRankConfig: rank 17 nranks 32 total 6.48 (kernels 0.66, alloc 3.40, bootstrap 0.86, allgathers 0.14, topo 0.74, graphs 0.08, connections 0.04, rest 0.58) +n116-040-048:1782:1891 [3] NCCL INFO Channel 01/0 : 19[3] -> 20[4] via P2P/CUMEM +n116-040-048:1782:1891 [3] NCCL INFO Channel 02/0 : 19[3] -> 20[4] via P2P/CUMEM +n116-040-048:1782:1891 [3] NCCL INFO Channel 05/0 : 19[3] -> 20[4] via P2P/CUMEM +n116-040-048:1782:1891 [3] NCCL INFO Channel 06/0 : 19[3] -> 20[4] via P2P/CUMEM +n116-040-048:1782:1891 [3] NCCL INFO Channel 07/0 : 19[3] -> 20[4] via P2P/CUMEM +n116-040-048:1782:1891 [3] NCCL INFO Channel 08/0 : 19[3] -> 20[4] via P2P/CUMEM +n116-040-048:1782:1891 [3] NCCL INFO Channel 09/0 : 19[3] -> 20[4] via P2P/CUMEM +n116-040-048:1781:1842 [2] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-040-048:1781:1842 [2] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-040-048:1781:1842 [2] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-040-048:1781:1842 [2] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-040-048:1781:1842 [2] NCCL INFO ncclCommInitRankConfig comm 0xcd2af00 rank 18 nranks 32 cudaDev 2 nvmlDev 2 busId b000 commId 0x2a4d35ea441cbcba - Init COMPLETE +n116-040-048:1781:1842 [2] NCCL INFO Init timings - ncclCommInitRankConfig: rank 18 nranks 32 total 6.47 (kernels 0.65, alloc 3.36, bootstrap 0.89, allgathers 0.14, topo 0.74, graphs 0.08, connections 0.03, rest 0.59) +n116-040-048:1782:1891 [3] NCCL INFO Channel 10/0 : 19[3] -> 20[4] via P2P/CUMEM +n116-040-048:1782:1891 [3] NCCL INFO Channel 13/0 : 19[3] -> 20[4] via P2P/CUMEM +n116-040-048:1782:1891 [3] NCCL INFO Channel 14/0 : 19[3] -> 20[4] via P2P/CUMEM +n116-040-048:1782:1891 [3] NCCL INFO Channel 15/0 : 19[3] -> 20[4] via P2P/CUMEM +n116-040-048:1785:1838 [6] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-040-048:1785:1838 [6] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-040-048:1785:1838 [6] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-040-048:1785:1838 [6] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-040-048:1785:1838 [6] NCCL INFO ncclCommInitRankConfig comm 0xc067bd0 rank 22 nranks 32 cudaDev 6 nvmlDev 6 busId 8b000 commId 0x2a4d35ea441cbcba - Init COMPLETE +n116-040-048:1785:1838 [6] NCCL INFO Init timings - ncclCommInitRankConfig: rank 22 nranks 32 total 6.49 (kernels 0.65, alloc 3.41, bootstrap 0.85, allgathers 0.14, topo 0.74, graphs 0.08, connections 0.04, rest 0.58) +n116-040-048:1786:1840 [7] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-040-048:1786:1840 [7] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-040-048:1786:1840 [7] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-040-048:1786:1840 [7] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-040-048:1786:1840 [7] NCCL INFO ncclCommInitRankConfig comm 0xcc263e0 rank 23 nranks 32 cudaDev 7 nvmlDev 7 busId 8c000 commId 0x2a4d35ea441cbcba - Init COMPLETE +n116-040-048:1786:1840 [7] NCCL INFO Init timings - ncclCommInitRankConfig: rank 23 nranks 32 total 6.48 (kernels 0.65, alloc 3.40, bootstrap 0.86, allgathers 0.14, topo 0.74, graphs 0.08, connections 0.03, rest 0.59) +n116-040-048:1779:1841 [0] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-040-048:1779:1841 [0] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-040-048:1779:1841 [0] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-040-048:1779:1841 [0] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-040-048:1779:1841 [0] NCCL INFO ncclCommInitRankConfig comm 0xb7cca60 rank 16 nranks 32 cudaDev 0 nvmlDev 0 busId 4000 commId 0x2a4d35ea441cbcba - Init COMPLETE +n116-040-048:1779:1841 [0] NCCL INFO Init timings - ncclCommInitRankConfig: rank 16 nranks 32 total 6.48 (kernels 0.65, alloc 3.40, bootstrap 0.86, allgathers 0.14, topo 0.73, graphs 0.08, connections 0.03, rest 0.59) +n116-040-048:1783:1893 [4] NCCL INFO Channel 00/0 : 20[4] -> 21[5] via P2P/CUMEM +n116-040-048:1783:1893 [4] NCCL INFO Channel 01/0 : 20[4] -> 21[5] via P2P/CUMEM +n116-040-048:1783:1893 [4] NCCL INFO Channel 02/0 : 20[4] -> 21[5] via P2P/CUMEM +n116-040-048:1783:1893 [4] NCCL INFO Channel 03/0 : 20[4] -> 21[5] via P2P/CUMEM +n116-040-048:1784:1892 [5] NCCL INFO Channel 00/0 : 21[5] -> 22[6] via P2P/CUMEM +n116-040-048:1783:1893 [4] NCCL INFO Channel 04/0 : 20[4] -> 21[5] via P2P/CUMEM +n116-040-048:1784:1892 [5] NCCL INFO Channel 01/0 : 21[5] -> 22[6] via P2P/CUMEM +n116-040-048:1783:1893 [4] NCCL INFO Channel 05/0 : 20[4] -> 21[5] via P2P/CUMEM +n116-040-048:1784:1892 [5] NCCL INFO Channel 02/0 : 21[5] -> 22[6] via P2P/CUMEM +n116-040-048:1783:1893 [4] NCCL INFO Channel 06/0 : 20[4] -> 21[5] via P2P/CUMEM +n116-040-048:1784:1892 [5] NCCL INFO Channel 03/0 : 21[5] -> 22[6] via P2P/CUMEM +n116-040-048:1783:1893 [4] NCCL INFO Channel 07/0 : 20[4] -> 21[5] via P2P/CUMEM +n116-040-048:1784:1892 [5] NCCL INFO Channel 04/0 : 21[5] -> 22[6] via P2P/CUMEM +n116-040-048:1783:1893 [4] NCCL INFO Channel 08/0 : 20[4] -> 21[5] via P2P/CUMEM +n116-040-048:1784:1892 [5] NCCL INFO Channel 07/0 : 21[5] -> 22[6] via P2P/CUMEM +n116-040-048:1784:1892 [5] NCCL INFO Channel 08/0 : 21[5] -> 22[6] via P2P/CUMEM +n116-040-048:1784:1892 [5] NCCL INFO Channel 09/0 : 21[5] -> 22[6] via P2P/CUMEM +n116-040-048:1784:1892 [5] NCCL INFO Channel 10/0 : 21[5] -> 22[6] via P2P/CUMEM +n116-040-048:1784:1892 [5] NCCL INFO Channel 11/0 : 21[5] -> 22[6] via P2P/CUMEM +n116-040-048:1784:1892 [5] NCCL INFO Channel 12/0 : 21[5] -> 22[6] via P2P/CUMEM +n116-040-048:1784:1892 [5] NCCL INFO Channel 15/0 : 21[5] -> 22[6] via P2P/CUMEM +n116-040-048:1783:1893 [4] NCCL INFO Channel 09/0 : 20[4] -> 21[5] via P2P/CUMEM +n116-040-048:1783:1893 [4] NCCL INFO Channel 10/0 : 20[4] -> 21[5] via P2P/CUMEM +n116-040-048:1783:1893 [4] NCCL INFO Channel 11/0 : 20[4] -> 21[5] via P2P/CUMEM +n116-040-048:1780:1894 [1] NCCL INFO Channel 00/0 : 17[1] -> 18[2] via P2P/CUMEM +n116-040-048:1783:1893 [4] NCCL INFO Channel 12/0 : 20[4] -> 21[5] via P2P/CUMEM +n116-040-048:1780:1894 [1] NCCL INFO Channel 03/0 : 17[1] -> 18[2] via P2P/CUMEM +n116-040-048:1783:1893 [4] NCCL INFO Channel 13/0 : 20[4] -> 21[5] via P2P/CUMEM +n116-040-048:1780:1894 [1] NCCL INFO Channel 04/0 : 17[1] -> 18[2] via P2P/CUMEM +n116-040-048:1783:1893 [4] NCCL INFO Channel 14/0 : 20[4] -> 21[5] via P2P/CUMEM +n116-040-048:1780:1894 [1] NCCL INFO Channel 05/0 : 17[1] -> 18[2] via P2P/CUMEM +n116-040-048:1783:1893 [4] NCCL INFO Channel 15/0 : 20[4] -> 21[5] via P2P/CUMEM +n116-040-048:1780:1894 [1] NCCL INFO Channel 06/0 : 17[1] -> 18[2] via P2P/CUMEM +n116-040-048:1780:1894 [1] NCCL INFO Channel 07/0 : 17[1] -> 18[2] via P2P/CUMEM +n116-040-048:1779:1898 [0] NCCL INFO Channel 00/0 : 16[0] -> 17[1] via P2P/CUMEM +n116-040-048:1780:1894 [1] NCCL INFO Channel 08/0 : 17[1] -> 18[2] via P2P/CUMEM +n116-040-048:1780:1894 [1] NCCL INFO Channel 11/0 : 17[1] -> 18[2] via P2P/CUMEM +n116-040-048:1779:1898 [0] NCCL INFO Channel 01/0 : 16[0] -> 17[1] via P2P/CUMEM +n116-040-048:1780:1894 [1] NCCL INFO Channel 12/0 : 17[1] -> 18[2] via P2P/CUMEM +n116-040-048:1779:1898 [0] NCCL INFO Channel 02/0 : 16[0] -> 17[1] via P2P/CUMEM +n116-040-048:1780:1894 [1] NCCL INFO Channel 13/0 : 17[1] -> 18[2] via P2P/CUMEM +n116-040-048:1781:1895 [2] NCCL INFO Channel 00/0 : 18[2] -> 19[3] via P2P/CUMEM +n116-040-048:1779:1898 [0] NCCL INFO Channel 03/0 : 16[0] -> 17[1] via P2P/CUMEM +n116-040-048:1785:1896 [6] NCCL INFO Channel 00/0 : 22[6] -> 23[7] via P2P/CUMEM +n116-040-048:1780:1894 [1] NCCL INFO Channel 14/0 : 17[1] -> 18[2] via P2P/CUMEM +n116-040-048:1779:1898 [0] NCCL INFO Channel 04/0 : 16[0] -> 17[1] via P2P/CUMEM +n116-040-048:1781:1895 [2] NCCL INFO Channel 01/0 : 18[2] -> 19[3] via P2P/CUMEM +n116-040-048:1785:1896 [6] NCCL INFO Channel 01/0 : 22[6] -> 23[7] via P2P/CUMEM +n116-040-048:1780:1894 [1] NCCL INFO Channel 15/0 : 17[1] -> 18[2] via P2P/CUMEM +n116-040-048:1779:1898 [0] NCCL INFO Channel 05/0 : 16[0] -> 17[1] via P2P/CUMEM +n116-040-048:1781:1895 [2] NCCL INFO Channel 02/0 : 18[2] -> 19[3] via P2P/CUMEM +n116-040-048:1785:1896 [6] NCCL INFO Channel 02/0 : 22[6] -> 23[7] via P2P/CUMEM +n116-040-048:1779:1898 [0] NCCL INFO Channel 06/0 : 16[0] -> 17[1] via P2P/CUMEM +n116-040-048:1781:1895 [2] NCCL INFO Channel 03/0 : 18[2] -> 19[3] via P2P/CUMEM +n116-040-048:1785:1896 [6] NCCL INFO Channel 03/0 : 22[6] -> 23[7] via P2P/CUMEM +n116-040-048:1779:1898 [0] NCCL INFO Channel 07/0 : 16[0] -> 17[1] via P2P/CUMEM +n116-040-048:1781:1895 [2] NCCL INFO Channel 04/0 : 18[2] -> 19[3] via P2P/CUMEM +n116-040-048:1785:1896 [6] NCCL INFO Channel 04/0 : 22[6] -> 23[7] via P2P/CUMEM +n116-040-048:1779:1898 [0] NCCL INFO Channel 08/0 : 16[0] -> 17[1] via P2P/CUMEM +n116-040-048:1781:1895 [2] NCCL INFO Channel 05/0 : 18[2] -> 19[3] via P2P/CUMEM +n116-040-048:1785:1896 [6] NCCL INFO Channel 05/0 : 22[6] -> 23[7] via P2P/CUMEM +n116-040-048:1779:1898 [0] NCCL INFO Channel 09/0 : 16[0] -> 17[1] via P2P/CUMEM +n116-040-048:1781:1895 [2] NCCL INFO Channel 06/0 : 18[2] -> 19[3] via P2P/CUMEM +n116-040-048:1785:1896 [6] NCCL INFO Channel 06/0 : 22[6] -> 23[7] via P2P/CUMEM +n116-040-048:1781:1895 [2] NCCL INFO Channel 07/0 : 18[2] -> 19[3] via P2P/CUMEM +n116-040-048:1779:1898 [0] NCCL INFO Channel 10/0 : 16[0] -> 17[1] via P2P/CUMEM +n116-040-048:1785:1896 [6] NCCL INFO Channel 07/0 : 22[6] -> 23[7] via P2P/CUMEM +n116-040-048:1781:1895 [2] NCCL INFO Channel 08/0 : 18[2] -> 19[3] via P2P/CUMEM +n116-040-048:1779:1898 [0] NCCL INFO Channel 11/0 : 16[0] -> 17[1] via P2P/CUMEM +n116-040-048:1785:1896 [6] NCCL INFO Channel 08/0 : 22[6] -> 23[7] via P2P/CUMEM +n116-040-048:1781:1895 [2] NCCL INFO Channel 09/0 : 18[2] -> 19[3] via P2P/CUMEM +n116-040-048:1779:1898 [0] NCCL INFO Channel 12/0 : 16[0] -> 17[1] via P2P/CUMEM +n116-040-048:1785:1896 [6] NCCL INFO Channel 09/0 : 22[6] -> 23[7] via P2P/CUMEM +n116-040-048:1781:1895 [2] NCCL INFO Channel 10/0 : 18[2] -> 19[3] via P2P/CUMEM +n116-040-048:1779:1898 [0] NCCL INFO Channel 13/0 : 16[0] -> 17[1] via P2P/CUMEM +n116-040-048:1785:1896 [6] NCCL INFO Channel 10/0 : 22[6] -> 23[7] via P2P/CUMEM +n116-040-048:1781:1895 [2] NCCL INFO Channel 11/0 : 18[2] -> 19[3] via P2P/CUMEM +n116-040-048:1779:1898 [0] NCCL INFO Channel 14/0 : 16[0] -> 17[1] via P2P/CUMEM +n116-040-048:1785:1896 [6] NCCL INFO Channel 11/0 : 22[6] -> 23[7] via P2P/CUMEM +n116-040-048:1781:1895 [2] NCCL INFO Channel 12/0 : 18[2] -> 19[3] via P2P/CUMEM +n116-040-048:1779:1898 [0] NCCL INFO Channel 15/0 : 16[0] -> 17[1] via P2P/CUMEM +n116-040-048:1785:1896 [6] NCCL INFO Channel 12/0 : 22[6] -> 23[7] via P2P/CUMEM +n116-040-048:1781:1895 [2] NCCL INFO Channel 13/0 : 18[2] -> 19[3] via P2P/CUMEM +n116-040-048:1785:1896 [6] NCCL INFO Channel 13/0 : 22[6] -> 23[7] via P2P/CUMEM +n116-040-048:1781:1895 [2] NCCL INFO Channel 14/0 : 18[2] -> 19[3] via P2P/CUMEM +n116-040-048:1785:1896 [6] NCCL INFO Channel 14/0 : 22[6] -> 23[7] via P2P/CUMEM +n116-040-048:1781:1895 [2] NCCL INFO Channel 15/0 : 18[2] -> 19[3] via P2P/CUMEM +n116-040-048:1785:1896 [6] NCCL INFO Channel 15/0 : 22[6] -> 23[7] via P2P/CUMEM +n116-040-048:1783:1899 [4] NCCL INFO [Proxy Progress] Device 4 CPU core 186 +n116-040-048:1783:1893 [4] NCCL INFO Channel 05/0 : 13[5] -> 20[4] [receive] via NET/FasTrak/5/GDRDMA +n116-040-048:1783:1893 [4] NCCL INFO Channel 12/0 : 13[5] -> 20[4] [receive] via NET/FasTrak/4/GDRDMA +n116-040-048:1779:1900 [0] NCCL INFO [Proxy Progress] Device 0 CPU core 5 +n116-040-048:1780:1901 [1] NCCL INFO [Proxy Progress] Device 1 CPU core 21 +n116-040-048:1780:1894 [1] NCCL INFO Channel 01/0 : 17[1] -> 25[1] [send] via NET/FasTrak/1/GDRDMA +n116-040-048:1781:1902 [2] NCCL INFO [Proxy Progress] Device 2 CPU core 45 +n116-040-048:1779:1898 [0] NCCL INFO Channel 01/0 : 9[1] -> 16[0] [receive] via NET/FasTrak/1/GDRDMA +n116-040-048:1784:1903 [5] NCCL INFO [Proxy Progress] Device 5 CPU core 167 +n116-040-048:1782:1904 [3] NCCL INFO [Proxy Progress] Device 3 CPU core 153 +n116-040-048:1781:1895 [2] NCCL INFO Channel 03/0 : 11[3] -> 18[2] [receive] via NET/FasTrak/3/GDRDMA +n116-040-048:1782:1891 [3] NCCL INFO Channel 03/0 : 19[3] -> 27[3] [send] via NET/FasTrak/3/GDRDMA +n116-040-048:1784:1892 [5] NCCL INFO Channel 05/0 : 21[5] -> 29[5] [send] via NET/FasTrak/5/GDRDMA +n116-040-048:1785:1905 [6] NCCL INFO [Proxy Progress] Device 6 CPU core 182 +n116-040-048:1779:1898 [0] NCCL INFO Channel 08/0 : 9[1] -> 16[0] [receive] via NET/FasTrak/0/GDRDMA +n116-040-048:1779:1898 [0] NCCL INFO Channel 01/0 : 16[0] -> 23[7] via P2P/CUMEM +n116-040-048:1785:1896 [6] NCCL INFO Channel 07/0 : 15[7] -> 22[6] [receive] via NET/FasTrak/7/GDRDMA +n116-040-048:1779:1898 [0] NCCL INFO Channel 02/0 : 16[0] -> 23[7] via P2P/CUMEM +n116-040-048:1781:1895 [2] NCCL INFO Channel 10/0 : 11[3] -> 18[2] [receive] via NET/FasTrak/2/GDRDMA +n116-040-048:1779:1898 [0] NCCL INFO Channel 03/0 : 16[0] -> 23[7] via P2P/CUMEM +n116-040-048:1779:1898 [0] NCCL INFO Channel 04/0 : 16[0] -> 23[7] via P2P/CUMEM +n116-040-048:1779:1898 [0] NCCL INFO Channel 05/0 : 16[0] -> 23[7] via P2P/CUMEM +n116-040-048:1779:1898 [0] NCCL INFO Channel 06/0 : 16[0] -> 23[7] via P2P/CUMEM +n116-040-048:1785:1896 [6] NCCL INFO Channel 14/0 : 15[7] -> 22[6] [receive] via NET/FasTrak/6/GDRDMA +n116-040-048:1779:1898 [0] NCCL INFO Channel 09/0 : 16[0] -> 23[7] via P2P/CUMEM +n116-040-048:1779:1898 [0] NCCL INFO Channel 10/0 : 16[0] -> 23[7] via P2P/CUMEM +n116-040-048:1779:1898 [0] NCCL INFO Channel 11/0 : 16[0] -> 23[7] via P2P/CUMEM +n116-040-048:1779:1898 [0] NCCL INFO Channel 12/0 : 16[0] -> 23[7] via P2P/CUMEM +n116-040-048:1779:1898 [0] NCCL INFO Channel 13/0 : 16[0] -> 23[7] via P2P/CUMEM +n116-040-048:1779:1898 [0] NCCL INFO Channel 14/0 : 16[0] -> 23[7] via P2P/CUMEM +n116-040-048:1786:1906 [7] NCCL INFO [Proxy Progress] Device 7 CPU core 79 +n116-040-048:1786:1897 [7] NCCL INFO Channel 07/0 : 23[7] -> 31[7] [send] via NET/FasTrak/7/GDRDMA +n116-040-048:1783:1893 [4] NCCL INFO Channel 04/0 : 20[4] -> 28[4] [send] via NET/FasTrak/4/GDRDMA +n116-040-048:1785:1896 [6] NCCL INFO Channel 06/0 : 22[6] -> 30[6] [send] via NET/FasTrak/6/GDRDMA +n116-040-048:1781:1895 [2] NCCL INFO Channel 02/0 : 18[2] -> 26[2] [send] via NET/FasTrak/2/GDRDMA +n116-040-048:1779:1898 [0] NCCL INFO Channel 00/0 : 16[0] -> 24[0] [send] via NET/FasTrak/0/GDRDMA +n116-040-048:1785:1896 [6] NCCL INFO Channel 06/0 : 6[6] -> 22[6] [receive] via NET/FasTrak/6/GDRDMA +n116-040-048:1785:1896 [6] NCCL INFO Channel 06/0 : 22[6] -> 6[6] [send] via NET/FasTrak/6/GDRDMA +n116-040-048:1780:1894 [1] NCCL INFO Channel 00/0 : 8[0] -> 17[1] [receive] via NET/FasTrak/0/GDRDMA +n116-040-048:1785:1896 [6] NCCL INFO Channel 06/0 : 30[6] -> 22[6] [receive] via NET/FasTrak/6/GDRDMA +n116-040-048:1780:1894 [1] NCCL INFO Channel 09/0 : 8[0] -> 17[1] [receive] via NET/FasTrak/1/GDRDMA +n116-040-048:1780:1894 [1] NCCL INFO Channel 01/0 : 1[1] -> 17[1] [receive] via NET/FasTrak/1/GDRDMA +n116-040-048:1780:1894 [1] NCCL INFO Channel 01/0 : 17[1] -> 1[1] [send] via NET/FasTrak/1/GDRDMA +n116-040-048:1780:1894 [1] NCCL INFO Channel 00/0 : 17[1] -> 8[0] [send] via NET/FasTrak/0/GDRDMA +n116-040-048:1780:1894 [1] NCCL INFO Channel 09/0 : 17[1] -> 8[0] [send] via NET/FasTrak/1/GDRDMA +n116-040-048:1784:1892 [5] NCCL INFO Channel 04/0 : 12[4] -> 21[5] [receive] via NET/FasTrak/4/GDRDMA +n116-040-048:1784:1892 [5] NCCL INFO Channel 13/0 : 12[4] -> 21[5] [receive] via NET/FasTrak/5/GDRDMA +n116-040-048:1784:1892 [5] NCCL INFO Channel 05/0 : 5[5] -> 21[5] [receive] via NET/FasTrak/5/GDRDMA +n116-040-048:1784:1892 [5] NCCL INFO Channel 05/0 : 21[5] -> 5[5] [send] via NET/FasTrak/5/GDRDMA +n116-040-048:1784:1892 [5] NCCL INFO Channel 04/0 : 21[5] -> 12[4] [send] via NET/FasTrak/4/GDRDMA +n116-040-048:1784:1892 [5] NCCL INFO Channel 13/0 : 21[5] -> 12[4] [send] via NET/FasTrak/5/GDRDMA +n116-040-048:1781:1895 [2] NCCL INFO Channel 02/0 : 2[2] -> 18[2] [receive] via NET/FasTrak/2/GDRDMA +n116-040-048:1781:1895 [2] NCCL INFO Channel 02/0 : 18[2] -> 2[2] [send] via NET/FasTrak/2/GDRDMA +n116-040-048:1783:1893 [4] NCCL INFO Channel 04/0 : 4[4] -> 20[4] [receive] via NET/FasTrak/4/GDRDMA +n116-040-048:1783:1893 [4] NCCL INFO Channel 04/0 : 20[4] -> 4[4] [send] via NET/FasTrak/4/GDRDMA +n116-040-048:1782:1891 [3] NCCL INFO Channel 02/0 : 10[2] -> 19[3] [receive] via NET/FasTrak/2/GDRDMA +n116-040-048:1781:1895 [2] NCCL INFO Channel 02/0 : 26[2] -> 18[2] [receive] via NET/FasTrak/2/GDRDMA +n116-040-048:1782:1891 [3] NCCL INFO Channel 11/0 : 10[2] -> 19[3] [receive] via NET/FasTrak/3/GDRDMA +n116-040-048:1782:1891 [3] NCCL INFO Channel 03/0 : 3[3] -> 19[3] [receive] via NET/FasTrak/3/GDRDMA +n116-040-048:1782:1891 [3] NCCL INFO Channel 03/0 : 19[3] -> 3[3] [send] via NET/FasTrak/3/GDRDMA +n116-040-048:1783:1893 [4] NCCL INFO Channel 04/0 : 28[4] -> 20[4] [receive] via NET/FasTrak/4/GDRDMA +n116-040-048:1783:1893 [4] NCCL INFO Channel 05/0 : 20[4] -> 13[5] [send] via NET/FasTrak/5/GDRDMA +n116-040-048:1783:1893 [4] NCCL INFO Channel 12/0 : 20[4] -> 13[5] [send] via NET/FasTrak/4/GDRDMA +n116-040-048:1784:1892 [5] NCCL INFO Channel 05/0 : 29[5] -> 21[5] [receive] via NET/FasTrak/5/GDRDMA +n116-040-048:1782:1891 [3] NCCL INFO Channel 02/0 : 19[3] -> 10[2] [send] via NET/FasTrak/2/GDRDMA +n116-040-048:1782:1891 [3] NCCL INFO Channel 11/0 : 19[3] -> 10[2] [send] via NET/FasTrak/3/GDRDMA +n116-040-048:1781:1895 [2] NCCL INFO Channel 03/0 : 18[2] -> 11[3] [send] via NET/FasTrak/3/GDRDMA +n116-040-048:1781:1895 [2] NCCL INFO Channel 10/0 : 18[2] -> 11[3] [send] via NET/FasTrak/2/GDRDMA +n116-040-048:1782:1891 [3] NCCL INFO Channel 03/0 : 27[3] -> 19[3] [receive] via NET/FasTrak/3/GDRDMA +n116-040-048:1783:1893 [4] NCCL INFO Channel 00/0 : 20[4] -> 19[3] via P2P/CUMEM +n116-040-048:1784:1892 [5] NCCL INFO Channel 00/0 : 21[5] -> 20[4] via P2P/CUMEM +n116-040-048:1783:1893 [4] NCCL INFO Channel 01/0 : 20[4] -> 19[3] via P2P/CUMEM +n116-040-048:1784:1892 [5] NCCL INFO Channel 01/0 : 21[5] -> 20[4] via P2P/CUMEM +n116-040-048:1783:1893 [4] NCCL INFO Channel 02/0 : 20[4] -> 19[3] via P2P/CUMEM +n116-040-048:1784:1892 [5] NCCL INFO Channel 02/0 : 21[5] -> 20[4] via P2P/CUMEM +n116-040-048:1783:1893 [4] NCCL INFO Channel 05/0 : 20[4] -> 19[3] via P2P/CUMEM +n116-040-048:1784:1892 [5] NCCL INFO Channel 03/0 : 21[5] -> 20[4] via P2P/CUMEM +n116-040-048:1783:1893 [4] NCCL INFO Channel 06/0 : 20[4] -> 19[3] via P2P/CUMEM +n116-040-048:1784:1892 [5] NCCL INFO Channel 04/0 : 21[5] -> 20[4] via P2P/CUMEM +n116-040-048:1783:1893 [4] NCCL INFO Channel 07/0 : 20[4] -> 19[3] via P2P/CUMEM +n116-040-048:1784:1892 [5] NCCL INFO Channel 05/0 : 21[5] -> 20[4] via P2P/CUMEM +n116-040-048:1783:1893 [4] NCCL INFO Channel 08/0 : 20[4] -> 19[3] via P2P/CUMEM +n116-040-048:1784:1892 [5] NCCL INFO Channel 06/0 : 21[5] -> 20[4] via P2P/CUMEM +n116-040-048:1783:1893 [4] NCCL INFO Channel 09/0 : 20[4] -> 19[3] via P2P/CUMEM +n116-040-048:1784:1892 [5] NCCL INFO Channel 07/0 : 21[5] -> 20[4] via P2P/CUMEM +n116-040-048:1783:1893 [4] NCCL INFO Channel 10/0 : 20[4] -> 19[3] via P2P/CUMEM +n116-040-048:1784:1892 [5] NCCL INFO Channel 08/0 : 21[5] -> 20[4] via P2P/CUMEM +n116-040-048:1783:1893 [4] NCCL INFO Channel 13/0 : 20[4] -> 19[3] via P2P/CUMEM +n116-040-048:1784:1892 [5] NCCL INFO Channel 09/0 : 21[5] -> 20[4] via P2P/CUMEM +n116-040-048:1782:1891 [3] NCCL INFO Channel 00/0 : 19[3] -> 18[2] via P2P/CUMEM +n116-040-048:1783:1893 [4] NCCL INFO Channel 14/0 : 20[4] -> 19[3] via P2P/CUMEM +n116-040-048:1784:1892 [5] NCCL INFO Channel 10/0 : 21[5] -> 20[4] via P2P/CUMEM +n116-040-048:1782:1891 [3] NCCL INFO Channel 01/0 : 19[3] -> 18[2] via P2P/CUMEM +n116-040-048:1783:1893 [4] NCCL INFO Channel 15/0 : 20[4] -> 19[3] via P2P/CUMEM +n116-040-048:1784:1892 [5] NCCL INFO Channel 11/0 : 21[5] -> 20[4] via P2P/CUMEM +n116-040-048:1782:1891 [3] NCCL INFO Channel 02/0 : 19[3] -> 18[2] via P2P/CUMEM +n116-040-048:1784:1892 [5] NCCL INFO Channel 12/0 : 21[5] -> 20[4] via P2P/CUMEM +n116-040-048:1782:1891 [3] NCCL INFO Channel 03/0 : 19[3] -> 18[2] via P2P/CUMEM +n116-040-048:1784:1892 [5] NCCL INFO Channel 13/0 : 21[5] -> 20[4] via P2P/CUMEM +n116-040-048:1782:1891 [3] NCCL INFO Channel 04/0 : 19[3] -> 18[2] via P2P/CUMEM +n116-040-048:1784:1892 [5] NCCL INFO Channel 14/0 : 21[5] -> 20[4] via P2P/CUMEM +n116-040-048:1782:1891 [3] NCCL INFO Channel 05/0 : 19[3] -> 18[2] via P2P/CUMEM +n116-040-048:1784:1892 [5] NCCL INFO Channel 15/0 : 21[5] -> 20[4] via P2P/CUMEM +n116-040-048:1782:1891 [3] NCCL INFO Channel 06/0 : 19[3] -> 18[2] via P2P/CUMEM +n116-040-048:1782:1891 [3] NCCL INFO Channel 07/0 : 19[3] -> 18[2] via P2P/CUMEM +n116-040-048:1782:1891 [3] NCCL INFO Channel 08/0 : 19[3] -> 18[2] via P2P/CUMEM +n116-040-048:1782:1891 [3] NCCL INFO Channel 09/0 : 19[3] -> 18[2] via P2P/CUMEM +n116-040-048:1782:1891 [3] NCCL INFO Channel 10/0 : 19[3] -> 18[2] via P2P/CUMEM +n116-040-048:1782:1891 [3] NCCL INFO Channel 11/0 : 19[3] -> 18[2] via P2P/CUMEM +n116-040-048:1781:1895 [2] NCCL INFO Channel 00/0 : 18[2] -> 17[1] via P2P/CUMEM +n116-040-048:1782:1891 [3] NCCL INFO Channel 12/0 : 19[3] -> 18[2] via P2P/CUMEM +n116-040-048:1781:1895 [2] NCCL INFO Channel 03/0 : 18[2] -> 17[1] via P2P/CUMEM +n116-040-048:1782:1891 [3] NCCL INFO Channel 13/0 : 19[3] -> 18[2] via P2P/CUMEM +n116-040-048:1781:1895 [2] NCCL INFO Channel 04/0 : 18[2] -> 17[1] via P2P/CUMEM +n116-040-048:1782:1891 [3] NCCL INFO Channel 14/0 : 19[3] -> 18[2] via P2P/CUMEM +n116-040-048:1781:1895 [2] NCCL INFO Channel 05/0 : 18[2] -> 17[1] via P2P/CUMEM +n116-040-048:1782:1891 [3] NCCL INFO Channel 15/0 : 19[3] -> 18[2] via P2P/CUMEM +n116-040-048:1781:1895 [2] NCCL INFO Channel 06/0 : 18[2] -> 17[1] via P2P/CUMEM +n116-040-048:1781:1895 [2] NCCL INFO Channel 07/0 : 18[2] -> 17[1] via P2P/CUMEM +n116-040-048:1781:1895 [2] NCCL INFO Channel 08/0 : 18[2] -> 17[1] via P2P/CUMEM +n116-040-048:1781:1895 [2] NCCL INFO Channel 11/0 : 18[2] -> 17[1] via P2P/CUMEM +n116-040-048:1781:1895 [2] NCCL INFO Channel 12/0 : 18[2] -> 17[1] via P2P/CUMEM +n116-040-048:1781:1895 [2] NCCL INFO Channel 13/0 : 18[2] -> 17[1] via P2P/CUMEM +n116-040-048:1781:1895 [2] NCCL INFO Channel 14/0 : 18[2] -> 17[1] via P2P/CUMEM +n116-040-048:1781:1895 [2] NCCL INFO Channel 15/0 : 18[2] -> 17[1] via P2P/CUMEM +n116-040-048:1786:1897 [7] NCCL INFO Channel 06/0 : 14[6] -> 23[7] [receive] via NET/FasTrak/6/GDRDMA +n116-040-048:1779:1898 [0] NCCL INFO Channel 00/0 : 0[0] -> 16[0] [receive] via NET/FasTrak/0/GDRDMA +n116-040-048:1779:1898 [0] NCCL INFO Channel 00/0 : 16[0] -> 0[0] [send] via NET/FasTrak/0/GDRDMA +n116-040-048:1786:1897 [7] NCCL INFO Channel 15/0 : 14[6] -> 23[7] [receive] via NET/FasTrak/7/GDRDMA +n116-040-048:1779:1898 [0] NCCL INFO Channel 00/0 : 24[0] -> 16[0] [receive] via NET/FasTrak/0/GDRDMA +n116-040-048:1786:1897 [7] NCCL INFO Channel 07/0 : 7[7] -> 23[7] [receive] via NET/FasTrak/7/GDRDMA +n116-040-048:1786:1897 [7] NCCL INFO Channel 07/0 : 23[7] -> 7[7] [send] via NET/FasTrak/7/GDRDMA +n116-040-048:1786:1897 [7] NCCL INFO Channel 06/0 : 23[7] -> 14[6] [send] via NET/FasTrak/6/GDRDMA +n116-040-048:1786:1897 [7] NCCL INFO Channel 15/0 : 23[7] -> 14[6] [send] via NET/FasTrak/7/GDRDMA +n116-040-048:1780:1894 [1] NCCL INFO Channel 01/0 : 25[1] -> 17[1] [receive] via NET/FasTrak/1/GDRDMA +n116-040-048:1785:1896 [6] NCCL INFO Channel 07/0 : 22[6] -> 15[7] [send] via NET/FasTrak/7/GDRDMA +n116-040-048:1785:1896 [6] NCCL INFO Channel 14/0 : 22[6] -> 15[7] [send] via NET/FasTrak/6/GDRDMA +n116-040-048:1786:1897 [7] NCCL INFO Channel 07/0 : 31[7] -> 23[7] [receive] via NET/FasTrak/7/GDRDMA +n116-040-048:1786:1897 [7] NCCL INFO Channel 01/0 : 23[7] -> 16[0] via P2P/CUMEM +n116-040-048:1786:1897 [7] NCCL INFO Channel 02/0 : 23[7] -> 16[0] via P2P/CUMEM +n116-040-048:1786:1897 [7] NCCL INFO Channel 03/0 : 23[7] -> 16[0] via P2P/CUMEM +n116-040-048:1786:1897 [7] NCCL INFO Channel 04/0 : 23[7] -> 16[0] via P2P/CUMEM +n116-040-048:1786:1897 [7] NCCL INFO Channel 05/0 : 23[7] -> 16[0] via P2P/CUMEM +n116-040-048:1786:1897 [7] NCCL INFO Channel 06/0 : 23[7] -> 16[0] via P2P/CUMEM +n116-040-048:1786:1897 [7] NCCL INFO Channel 09/0 : 23[7] -> 16[0] via P2P/CUMEM +n116-040-048:1786:1897 [7] NCCL INFO Channel 10/0 : 23[7] -> 16[0] via P2P/CUMEM +n116-040-048:1786:1897 [7] NCCL INFO Channel 11/0 : 23[7] -> 16[0] via P2P/CUMEM +n116-040-048:1780:1894 [1] NCCL INFO Channel 00/0 : 17[1] -> 16[0] via P2P/CUMEM +n116-040-048:1786:1897 [7] NCCL INFO Channel 12/0 : 23[7] -> 16[0] via P2P/CUMEM +n116-040-048:1780:1894 [1] NCCL INFO Channel 01/0 : 17[1] -> 16[0] via P2P/CUMEM +n116-040-048:1786:1897 [7] NCCL INFO Channel 13/0 : 23[7] -> 16[0] via P2P/CUMEM +n116-040-048:1786:1897 [7] NCCL INFO Channel 14/0 : 23[7] -> 16[0] via P2P/CUMEM +n116-040-048:1779:1898 [0] NCCL INFO Channel 01/0 : 16[0] -> 9[1] [send] via NET/FasTrak/1/GDRDMA +n116-040-048:1779:1898 [0] NCCL INFO Channel 08/0 : 16[0] -> 9[1] [send] via NET/FasTrak/0/GDRDMA +n116-040-048:1780:1894 [1] NCCL INFO Channel 02/0 : 17[1] -> 16[0] via P2P/CUMEM +n116-040-048:1786:1897 [7] NCCL INFO Channel 00/0 : 23[7] -> 22[6] via P2P/CUMEM +n116-040-048:1780:1894 [1] NCCL INFO Channel 03/0 : 17[1] -> 16[0] via P2P/CUMEM +n116-040-048:1786:1897 [7] NCCL INFO Channel 01/0 : 23[7] -> 22[6] via P2P/CUMEM +n116-040-048:1780:1894 [1] NCCL INFO Channel 04/0 : 17[1] -> 16[0] via P2P/CUMEM +n116-040-048:1786:1897 [7] NCCL INFO Channel 02/0 : 23[7] -> 22[6] via P2P/CUMEM +n116-040-048:1780:1894 [1] NCCL INFO Channel 05/0 : 17[1] -> 16[0] via P2P/CUMEM +n116-040-048:1786:1897 [7] NCCL INFO Channel 03/0 : 23[7] -> 22[6] via P2P/CUMEM +n116-040-048:1780:1894 [1] NCCL INFO Channel 06/0 : 17[1] -> 16[0] via P2P/CUMEM +n116-040-048:1786:1897 [7] NCCL INFO Channel 04/0 : 23[7] -> 22[6] via P2P/CUMEM +n116-040-048:1780:1894 [1] NCCL INFO Channel 07/0 : 17[1] -> 16[0] via P2P/CUMEM +n116-040-048:1786:1897 [7] NCCL INFO Channel 05/0 : 23[7] -> 22[6] via P2P/CUMEM +n116-040-048:1786:1897 [7] NCCL INFO Channel 06/0 : 23[7] -> 22[6] via P2P/CUMEM +n116-040-048:1780:1894 [1] NCCL INFO Channel 08/0 : 17[1] -> 16[0] via P2P/CUMEM +n116-040-048:1786:1897 [7] NCCL INFO Channel 07/0 : 23[7] -> 22[6] via P2P/CUMEM +n116-040-048:1780:1894 [1] NCCL INFO Channel 09/0 : 17[1] -> 16[0] via P2P/CUMEM +n116-040-048:1786:1897 [7] NCCL INFO Channel 08/0 : 23[7] -> 22[6] via P2P/CUMEM +n116-040-048:1780:1894 [1] NCCL INFO Channel 10/0 : 17[1] -> 16[0] via P2P/CUMEM +n116-040-048:1780:1894 [1] NCCL INFO Channel 11/0 : 17[1] -> 16[0] via P2P/CUMEM +n116-040-048:1786:1897 [7] NCCL INFO Channel 09/0 : 23[7] -> 22[6] via P2P/CUMEM +n116-040-048:1780:1894 [1] NCCL INFO Channel 12/0 : 17[1] -> 16[0] via P2P/CUMEM +n116-040-048:1786:1897 [7] NCCL INFO Channel 10/0 : 23[7] -> 22[6] via P2P/CUMEM +n116-040-048:1780:1894 [1] NCCL INFO Channel 13/0 : 17[1] -> 16[0] via P2P/CUMEM +n116-040-048:1786:1897 [7] NCCL INFO Channel 11/0 : 23[7] -> 22[6] via P2P/CUMEM +n116-040-048:1780:1894 [1] NCCL INFO Channel 14/0 : 17[1] -> 16[0] via P2P/CUMEM +n116-040-048:1786:1897 [7] NCCL INFO Channel 12/0 : 23[7] -> 22[6] via P2P/CUMEM +n116-040-048:1780:1894 [1] NCCL INFO Channel 15/0 : 17[1] -> 16[0] via P2P/CUMEM +n116-040-048:1786:1897 [7] NCCL INFO Channel 13/0 : 23[7] -> 22[6] via P2P/CUMEM +n116-040-048:1786:1897 [7] NCCL INFO Channel 14/0 : 23[7] -> 22[6] via P2P/CUMEM +n116-040-048:1786:1897 [7] NCCL INFO Channel 15/0 : 23[7] -> 22[6] via P2P/CUMEM +n116-040-048:1785:1896 [6] NCCL INFO Channel 00/0 : 22[6] -> 21[5] via P2P/CUMEM +n116-040-048:1785:1896 [6] NCCL INFO Channel 01/0 : 22[6] -> 21[5] via P2P/CUMEM +n116-040-048:1785:1896 [6] NCCL INFO Channel 02/0 : 22[6] -> 21[5] via P2P/CUMEM +n116-040-048:1785:1896 [6] NCCL INFO Channel 03/0 : 22[6] -> 21[5] via P2P/CUMEM +n116-040-048:1785:1896 [6] NCCL INFO Channel 04/0 : 22[6] -> 21[5] via P2P/CUMEM +n116-040-048:1785:1896 [6] NCCL INFO Channel 07/0 : 22[6] -> 21[5] via P2P/CUMEM +n116-040-048:1785:1896 [6] NCCL INFO Channel 08/0 : 22[6] -> 21[5] via P2P/CUMEM +n116-040-048:1785:1896 [6] NCCL INFO Channel 09/0 : 22[6] -> 21[5] via P2P/CUMEM +n116-040-048:1785:1896 [6] NCCL INFO Channel 10/0 : 22[6] -> 21[5] via P2P/CUMEM +n116-040-048:1785:1896 [6] NCCL INFO Channel 11/0 : 22[6] -> 21[5] via P2P/CUMEM +n116-040-048:1785:1896 [6] NCCL INFO Channel 12/0 : 22[6] -> 21[5] via P2P/CUMEM +n116-040-048:1785:1896 [6] NCCL INFO Channel 15/0 : 22[6] -> 21[5] via P2P/CUMEM +n116-040-048:1783:1893 [4] NCCL INFO Connected all trees +n116-040-048:1781:1895 [2] NCCL INFO Connected all trees +n116-040-048:1782:1891 [3] NCCL INFO Connected all trees +n116-040-048:1779:1898 [0] NCCL INFO Connected all trees +n116-040-048:1780:1894 [1] NCCL INFO Connected all trees +n116-040-048:1784:1892 [5] NCCL INFO Connected all trees +n116-040-048:1786:1897 [7] NCCL INFO Connected all trees +n116-040-048:1785:1896 [6] NCCL INFO Connected all trees +n116-040-048:1779:1936 [0] NCCL INFO Channel 00/0 : 9[1] -> 16[0] [receive] via NET/FasTrak/0/GDRDMA +n116-040-048:1779:1936 [0] NCCL INFO Channel 00/0 : 16[0] -> 23[7] via P2P/CUMEM +n116-040-048:1779:1936 [0] NCCL INFO Channel 08/0 : 16[0] -> 23[7] via P2P/CUMEM +n116-040-048:1786:1939 [7] NCCL INFO Channel 06/0 : 23[7] -> 30[6] [send] via NET/FasTrak/6/GDRDMA +n116-040-048:1784:1940 [5] NCCL INFO Channel 05/0 : 21[5] -> 22[6] via P2P/CUMEM +n116-040-048:1786:1939 [7] NCCL INFO Channel 14/0 : 23[7] -> 30[6] [send] via NET/FasTrak/6/GDRDMA +n116-040-048:1784:1940 [5] NCCL INFO Channel 13/0 : 21[5] -> 22[6] via P2P/CUMEM +n116-040-048:1786:1939 [7] NCCL INFO Channel 07/0 : 14[6] -> 23[7] [receive] via NET/FasTrak/7/GDRDMA +n116-040-048:1779:1936 [0] NCCL INFO Channel 01/0 : 16[0] -> 25[1] [send] via NET/FasTrak/1/GDRDMA +n116-040-048:1779:1936 [0] NCCL INFO Channel 09/0 : 16[0] -> 25[1] [send] via NET/FasTrak/1/GDRDMA +n116-040-048:1780:1941 [1] NCCL INFO Channel 01/0 : 17[1] -> 18[2] via P2P/CUMEM +n116-040-048:1780:1941 [1] NCCL INFO Channel 09/0 : 17[1] -> 18[2] via P2P/CUMEM +n116-040-048:1780:1941 [1] NCCL INFO Channel 00/0 : 17[1] -> 24[0] [send] via NET/FasTrak/0/GDRDMA +n116-040-048:1780:1941 [1] NCCL INFO Channel 08/0 : 17[1] -> 24[0] [send] via NET/FasTrak/0/GDRDMA +n116-040-048:1781:1938 [2] NCCL INFO Channel 02/0 : 11[3] -> 18[2] [receive] via NET/FasTrak/2/GDRDMA +n116-040-048:1780:1941 [1] NCCL INFO Channel 01/0 : 8[0] -> 17[1] [receive] via NET/FasTrak/1/GDRDMA +n116-040-048:1782:1942 [3] NCCL INFO Channel 03/0 : 19[3] -> 20[4] via P2P/CUMEM +n116-040-048:1782:1942 [3] NCCL INFO Channel 11/0 : 19[3] -> 20[4] via P2P/CUMEM +n116-040-048:1782:1942 [3] NCCL INFO Channel 02/0 : 19[3] -> 26[2] [send] via NET/FasTrak/2/GDRDMA +n116-040-048:1782:1942 [3] NCCL INFO Channel 10/0 : 19[3] -> 26[2] [send] via NET/FasTrak/2/GDRDMA +n116-040-048:1783:1937 [4] NCCL INFO Channel 04/0 : 13[5] -> 20[4] [receive] via NET/FasTrak/4/GDRDMA +n116-040-048:1782:1942 [3] NCCL INFO Channel 03/0 : 10[2] -> 19[3] [receive] via NET/FasTrak/3/GDRDMA +n116-040-048:1784:1940 [5] NCCL INFO Channel 04/0 : 21[5] -> 28[4] [send] via NET/FasTrak/4/GDRDMA +n116-040-048:1784:1940 [5] NCCL INFO Channel 12/0 : 21[5] -> 28[4] [send] via NET/FasTrak/4/GDRDMA +n116-040-048:1785:1943 [6] NCCL INFO Channel 06/0 : 15[7] -> 22[6] [receive] via NET/FasTrak/6/GDRDMA +n116-040-048:1784:1940 [5] NCCL INFO Channel 05/0 : 12[4] -> 21[5] [receive] via NET/FasTrak/5/GDRDMA +n116-040-048:1781:1938 [2] NCCL INFO Channel 03/0 : 18[2] -> 27[3] [send] via NET/FasTrak/3/GDRDMA +n116-040-048:1781:1938 [2] NCCL INFO Channel 11/0 : 18[2] -> 27[3] [send] via NET/FasTrak/3/GDRDMA +n116-040-048:1781:1938 [2] NCCL INFO Channel 02/0 : 18[2] -> 17[1] via P2P/CUMEM +n116-040-048:1781:1938 [2] NCCL INFO Channel 10/0 : 18[2] -> 17[1] via P2P/CUMEM +n116-040-048:1785:1943 [6] NCCL INFO Channel 07/0 : 22[6] -> 31[7] [send] via NET/FasTrak/7/GDRDMA +n116-040-048:1785:1943 [6] NCCL INFO Channel 15/0 : 22[6] -> 31[7] [send] via NET/FasTrak/7/GDRDMA +n116-040-048:1785:1943 [6] NCCL INFO Channel 06/0 : 22[6] -> 21[5] via P2P/CUMEM +n116-040-048:1785:1943 [6] NCCL INFO Channel 14/0 : 22[6] -> 21[5] via P2P/CUMEM +n116-040-048:1786:1939 [7] NCCL INFO Channel 07/0 : 23[7] -> 16[0] via P2P/CUMEM +n116-040-048:1786:1939 [7] NCCL INFO Channel 15/0 : 23[7] -> 16[0] via P2P/CUMEM +n116-040-048:1783:1937 [4] NCCL INFO Channel 05/0 : 20[4] -> 29[5] [send] via NET/FasTrak/5/GDRDMA +n116-040-048:1783:1937 [4] NCCL INFO Channel 13/0 : 20[4] -> 29[5] [send] via NET/FasTrak/5/GDRDMA +n116-040-048:1783:1937 [4] NCCL INFO Channel 04/0 : 20[4] -> 19[3] via P2P/CUMEM +n116-040-048:1783:1937 [4] NCCL INFO Channel 12/0 : 20[4] -> 19[3] via P2P/CUMEM +n116-040-048:1779:1936 [0] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-040-048:1780:1941 [1] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-040-048:1781:1938 [2] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-040-048:1782:1942 [3] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-040-048:1786:1939 [7] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-040-048:1783:1937 [4] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-040-048:1785:1943 [6] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-040-048:1784:1940 [5] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +[rank22]:[W727 05:39:59.401310306 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n116-040-048.byted.org]:38656, remote=[n116-042-185.byted.org]:25003): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f6da93785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f6d926cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7f6d926cff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7f6d926d084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7f6d926ca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f6d53dc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f6dac86b4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f6daf4421f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f6daf4c289c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank22]:[W727 05:39:59.415594507 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 22] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank21]:[W727 05:39:59.476067259 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n116-040-048.byted.org]:38700, remote=[n116-042-185.byted.org]:25003): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f87ed5785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f87d6ccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7f87d6ccff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7f87d6cd084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7f87d6cca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f87983c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f87f0cdc4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f87f38b31f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f87f393389c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank21]:[W727 05:39:59.482388782 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 21] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank17]:[W727 05:39:59.524302143 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n116-040-048.byted.org]:38730, remote=[n116-042-185.byted.org]:25003): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f1c339785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f1c1cccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7f1c1cccff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7f1c1ccd084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7f1c1ccca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f1bde3c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f1c36e9d4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f1c39a741f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f1c39af489c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank17]:[W727 05:39:59.531652563 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 17] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank19]:[W727 05:39:59.660244957 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n116-040-048.byted.org]:38690, remote=[n116-042-185.byted.org]:25003): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f3cf2f785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f3cdc6cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7f3cdc6cff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7f3cdc6d084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7f3cdc6ca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f3c9ddc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f3cf67a24a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f3cf93791f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f3cf93f989c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank19]:[W727 05:39:59.668030308 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 19] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank23]:[W727 05:39:59.661040934 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n116-040-048.byted.org]:38676, remote=[n116-042-185.byted.org]:25003): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f54b59785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f549eccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7f549eccff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7f549ecd084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7f549ecca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f54603c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f54b8dff4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f54bb9d61f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f54bba5689c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank23]:[W727 05:39:59.668138345 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 23] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank20]:[W727 05:40:00.976340618 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n116-040-048.byted.org]:38716, remote=[n116-042-185.byted.org]:25003): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7ff901f785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7ff8eb2cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7ff8eb2cff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7ff8eb2d084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7ff8eb2ca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7ff8ac9c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7ff90542b4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7ff9080021f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7ff90808289c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank20]:[W727 05:40:00.982980758 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 20] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank18]:[W727 05:40:00.976633633 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n116-040-048.byted.org]:38670, remote=[n116-042-185.byted.org]:25003): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f9b2e7785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f9b17ecdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7f9b17ecff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7f9b17ed084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7f9b17eca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f9ad95c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f9b31f5c4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f9b34b331f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f9b34bb389c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank18]:[W727 05:40:00.983248302 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 18] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank16]:[W727 05:40:00.277296862 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=71, addr=[n116-040-048.byted.org]:38388, remote=[n116-042-185.byted.org]:25003): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7fc21e3785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7fc2076cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7fc2076cff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7fc2076d084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7fc2076ca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fc1c8dc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7fc2217da4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7fc2243b11f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7fc22443189c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank16]:[W727 05:40:00.284731518 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 16] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank22]:[W727 05:40:00.415765027 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-040-048.byted.org]:38656, remote=[n116-042-185.byted.org]:25003): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f6da93785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f6d926cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f6d926cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f6d926d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f6d926ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f6d53dc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f6dac86b4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f6daf4421f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f6daf4c289c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank22]:[W727 05:40:00.423458532 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 22] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank21]:[W727 05:40:00.482553814 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-040-048.byted.org]:38700, remote=[n116-042-185.byted.org]:25003): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f87ed5785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f87d6ccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f87d6ccf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f87d6cd0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f87d6cca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f87983c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f87f0cdc4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f87f38b31f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f87f393389c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank21]:[W727 05:40:00.489557973 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 21] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank17]:[W727 05:40:00.531794760 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-040-048.byted.org]:38730, remote=[n116-042-185.byted.org]:25003): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f1c339785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f1c1cccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f1c1cccf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f1c1ccd0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f1c1ccca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f1bde3c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f1c36e9d4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f1c39a741f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f1c39af489c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank17]:[W727 05:40:00.539492196 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 17] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank23]:[W727 05:40:00.668277760 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-040-048.byted.org]:38676, remote=[n116-042-185.byted.org]:25003): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f54b59785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f549eccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f549eccf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f549ecd0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f549ecca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f54603c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f54b8dff4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f54bb9d61f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f54bba5689c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank23]:[W727 05:40:00.674444154 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 23] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank19]:[W727 05:40:00.668176327 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-040-048.byted.org]:38690, remote=[n116-042-185.byted.org]:25003): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f3cf2f785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f3cdc6cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f3cdc6cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f3cdc6d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f3cdc6ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f3c9ddc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f3cf67a24a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f3cf93791f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f3cf93f989c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank19]:[W727 05:40:00.674479296 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 19] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank20]:[W727 05:40:01.983147588 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-040-048.byted.org]:38716, remote=[n116-042-185.byted.org]:25003): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7ff901f785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7ff8eb2cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7ff8eb2cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7ff8eb2d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7ff8eb2ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7ff8ac9c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7ff90542b4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7ff9080021f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7ff90808289c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank20]:[W727 05:40:01.989671251 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 20] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank18]:[W727 05:40:01.983404462 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-040-048.byted.org]:38670, remote=[n116-042-185.byted.org]:25003): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f9b2e7785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f9b17ecdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f9b17ecf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f9b17ed0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f9b17eca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f9ad95c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f9b31f5c4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f9b34b331f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f9b34bb389c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank18]:[W727 05:40:01.990309146 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 18] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +E0727 05:40:01.201000 1714 torch/distributed/elastic/multiprocessing/api.py:874] failed (exitcode: -15) local_rank: 0 (pid: 1779) of binary: /usr/bin/python3 +Traceback (most recent call last): + File "/usr/local/bin/torchrun", line 7, in + sys.exit(main()) + ^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 355, in wrapper + return f(*args, **kwargs) + ^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 892, in main + run(args) + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 883, in run + elastic_launch( + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 139, in __call__ + return launch_agent(self._config, self._entrypoint, list(args)) + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 270, in launch_agent + raise ChildFailedError( +torch.distributed.elastic.multiprocessing.errors.ChildFailedError: +============================================================ +main_cross_2direct_balance_realscale_perturb_v4_accum.py FAILED +------------------------------------------------------------ +Failures: +[1]: + time : 2026-07-27_05:40:01 + host : n116-040-048.byted.org + rank : 17 (local_rank: 1) + exitcode : -15 (pid: 1780) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 1780 +[2]: + time : 2026-07-27_05:40:01 + host : n116-040-048.byted.org + rank : 18 (local_rank: 2) + exitcode : -15 (pid: 1781) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 1781 +[3]: + time : 2026-07-27_05:40:01 + host : n116-040-048.byted.org + rank : 19 (local_rank: 3) + exitcode : -15 (pid: 1782) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 1782 +[4]: + time : 2026-07-27_05:40:01 + host : n116-040-048.byted.org + rank : 20 (local_rank: 4) + exitcode : -15 (pid: 1783) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 1783 +[5]: + time : 2026-07-27_05:40:01 + host : n116-040-048.byted.org + rank : 21 (local_rank: 5) + exitcode : -15 (pid: 1784) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 1784 +[6]: + time : 2026-07-27_05:40:01 + host : n116-040-048.byted.org + rank : 22 (local_rank: 6) + exitcode : -15 (pid: 1785) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 1785 +[7]: + time : 2026-07-27_05:40:01 + host : n116-040-048.byted.org + rank : 23 (local_rank: 7) + exitcode : -15 (pid: 1786) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 1786 +------------------------------------------------------------ +Root Cause (first observed failure): +[0]: + time : 2026-07-27_05:40:01 + host : n116-040-048.byted.org + rank : 16 (local_rank: 0) + exitcode : -15 (pid: 1779) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 1779 +============================================================ diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1603_vith32_launch_stub_master_n116-042-185_port25003/launcher_rank3_n111-046-125.log b/ScaleDF_rebuttal/_host_logs/20260726_1603_vith32_launch_stub_master_n116-042-185_port25003/launcher_rank3_n111-046-125.log new file mode 100644 index 0000000000000000000000000000000000000000..8506ca90ccb1ca84974ebeab543895b7d9d1cfc5 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1603_vith32_launch_stub_master_n116-042-185_port25003/launcher_rank3_n111-046-125.log @@ -0,0 +1,1948 @@ +started_utc=2026-07-26T16:03:57Z +hostname=n111-046-125 +node_rank=3 +nnodes=4 +nproc_per_node=8 +world_size=32 +master_addr=10.116.42.185 +master_port=25003 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +main=main_cross_2direct_balance_realscale_perturb_v4_accum.py +model_name=deit_huge_patch14_LS +pretrain=/tmp/ScaleDF_repro_20260724/pretrained/deit_3_huge_224_21k_v1.pth +resume=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_24card_effbs3072_from_epoch7_n116-041-246_n116-042-186_n116-042-185_20260726_0736_vith24_from_epoch7/checkpoint.pth +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_32card_effbs2048_from_latest24_master_n116-042-185_20260726_1603_vith32_stub_port25003 +batch_per_gpu=64 +accum_iter=1 +effective_batch=2048 +eval_batch_per_gpu=64 +epochs=40 +W0726 16:04:12.799000 1593 torch/distributed/run.py:766] +W0726 16:04:12.799000 1593 torch/distributed/run.py:766] ***************************************** +W0726 16:04:12.799000 1593 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 16:04:12.799000 1593 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 24): env:// +| distributed init (rank 27): env:// +[rank27]:[W726 16:05:01.417457204 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 27] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank24]:[W726 16:05:01.787010489 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 24] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 31): env:// +[rank31]:[W726 16:05:01.860818316 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 31] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 29): env:// +| distributed init (rank 26): env:// +[rank26]:[W726 16:05:01.927154489 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 26] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank29]:[W726 16:05:01.927148035 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 29] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 30): env:// +| distributed init (rank 25): env:// +[rank30]:[W726 16:05:01.954376930 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 30] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank25]:[W726 16:05:01.958376517 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 25] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 28): env:// +[rank28]:[W726 16:05:01.982022425 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 28] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +n111-046-125:1664:1664 [6] NCCL INFO cudaDriverVersion 12060 +n111-046-125:1664:1664 [6] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n111-046-125:1664:1664 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n111-046-125:1664:1664 [6] NCCL INFO NCCL version 2.26.2+cuda12.2 +n111-046-125:1664:1664 [6] NCCL INFO Comm config Blocking set to 1 +n111-046-125:1659:1659 [1] NCCL INFO cudaDriverVersion 12060 +n111-046-125:1659:1659 [1] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n111-046-125:1659:1659 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n111-046-125:1659:1659 [1] NCCL INFO NCCL version 2.26.2+cuda12.2 +n111-046-125:1659:1659 [1] NCCL INFO Comm config Blocking set to 1 +n111-046-125:1661:1661 [3] NCCL INFO cudaDriverVersion 12060 +n111-046-125:1661:1661 [3] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n111-046-125:1661:1661 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n111-046-125:1661:1661 [3] NCCL INFO NCCL version 2.26.2+cuda12.2 +n111-046-125:1661:1661 [3] NCCL INFO Comm config Blocking set to 1 +n111-046-125:1660:1660 [2] NCCL INFO cudaDriverVersion 12060 +n111-046-125:1660:1660 [2] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n111-046-125:1660:1660 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n111-046-125:1660:1660 [2] NCCL INFO NCCL version 2.26.2+cuda12.2 +n111-046-125:1660:1660 [2] NCCL INFO Comm config Blocking set to 1 +n111-046-125:1665:1665 [7] NCCL INFO cudaDriverVersion 12060 +n111-046-125:1665:1665 [7] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n111-046-125:1665:1665 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n111-046-125:1665:1665 [7] NCCL INFO NCCL version 2.26.2+cuda12.2 +n111-046-125:1662:1662 [4] NCCL INFO cudaDriverVersion 12060 +n111-046-125:1662:1662 [4] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n111-046-125:1662:1662 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n111-046-125:1662:1662 [4] NCCL INFO NCCL version 2.26.2+cuda12.2 +n111-046-125:1663:1663 [5] NCCL INFO cudaDriverVersion 12060 +n111-046-125:1663:1663 [5] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n111-046-125:1663:1663 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n111-046-125:1663:1663 [5] NCCL INFO NCCL version 2.26.2+cuda12.2 +n111-046-125:1658:1658 [0] NCCL INFO cudaDriverVersion 12060 +n111-046-125:1658:1658 [0] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n111-046-125:1658:1658 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n111-046-125:1665:1665 [7] NCCL INFO Comm config Blocking set to 1 +n111-046-125:1662:1662 [4] NCCL INFO Comm config Blocking set to 1 +n111-046-125:1663:1663 [5] NCCL INFO Comm config Blocking set to 1 +n111-046-125:1658:1658 [0] NCCL INFO NCCL version 2.26.2+cuda12.2 +n111-046-125:1658:1658 [0] NCCL INFO Comm config Blocking set to 1 +n111-046-125:1658:1721 [0] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n111-046-125:1659:1715 [1] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n111-046-125:1662:1719 [4] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n111-046-125:1661:1716 [3] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n111-046-125:1665:1718 [7] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n111-046-125:1664:1714 [6] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n111-046-125:1660:1717 [2] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n111-046-125:1658:1721 [0] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n111-046-125:1663:1720 [5] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n111-046-125:1661:1716 [3] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n111-046-125:1662:1719 [4] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n111-046-125:1664:1714 [6] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n111-046-125:1659:1715 [1] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n111-046-125:1665:1718 [7] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n111-046-125:1658:1721 [0] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n111-046-125:1664:1714 [6] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n111-046-125:1660:1717 [2] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n111-046-125:1661:1716 [3] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n111-046-125:1659:1715 [1] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n111-046-125:1665:1718 [7] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n111-046-125:1661:1716 [3] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n111-046-125:1659:1715 [1] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n111-046-125:1658:1721 [0] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n111-046-125:1664:1714 [6] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n111-046-125:1660:1717 [2] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n111-046-125:1663:1720 [5] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n111-046-125:1665:1718 [7] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n111-046-125:1662:1719 [4] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n111-046-125:1660:1717 [2] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n111-046-125:1661:1716 [3] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n111-046-125:1659:1715 [1] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n111-046-125:1662:1719 [4] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n111-046-125:1663:1720 [5] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n111-046-125:1664:1714 [6] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n111-046-125:1661:1716 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n111-046-125:1659:1715 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n111-046-125:1665:1718 [7] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n111-046-125:1664:1714 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n111-046-125:1663:1720 [5] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n111-046-125:1662:1719 [4] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n111-046-125:1658:1721 [0] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n111-046-125:1661:1716 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n111-046-125:1660:1717 [2] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n111-046-125:1659:1715 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n111-046-125:1665:1718 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n111-046-125:1662:1719 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n111-046-125:1658:1721 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n111-046-125:1660:1717 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n111-046-125:1665:1718 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n111-046-125:1664:1714 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n111-046-125:1661:1716 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n111-046-125:1659:1715 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n111-046-125:1662:1719 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n111-046-125:1663:1720 [5] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n111-046-125:1664:1714 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n111-046-125:1658:1721 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n111-046-125:1665:1718 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n111-046-125:1659:1715 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n111-046-125:1658:1721 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n111-046-125:1665:1718 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n111-046-125:1660:1717 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n111-046-125:1661:1716 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n111-046-125:1662:1719 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n111-046-125:1663:1720 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n111-046-125:1664:1714 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n111-046-125:1659:1715 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n111-046-125:1658:1721 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n111-046-125:1665:1718 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n111-046-125:1658:1721 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n111-046-125:1660:1717 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n111-046-125:1662:1719 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n111-046-125:1661:1716 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n111-046-125:1663:1720 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n111-046-125:1664:1714 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n111-046-125:1660:1717 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n111-046-125:1662:1719 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n111-046-125:1663:1720 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n111-046-125:1660:1717 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n111-046-125:1663:1720 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n111-046-125:1663:1720 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n111-046-125:1661:1716 [3] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n111-046-125:1661:1716 [3] NCCL INFO Using network FasTrak +n111-046-125:1663:1720 [5] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n111-046-125:1663:1720 [5] NCCL INFO Using network FasTrak +n111-046-125:1658:1721 [0] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n111-046-125:1658:1721 [0] NCCL INFO Using network FasTrak +n111-046-125:1662:1719 [4] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n111-046-125:1662:1719 [4] NCCL INFO Using network FasTrak +n111-046-125:1665:1718 [7] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n111-046-125:1665:1718 [7] NCCL INFO Using network FasTrak +n111-046-125:1664:1714 [6] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n111-046-125:1664:1714 [6] NCCL INFO Using network FasTrak +n111-046-125:1659:1715 [1] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n111-046-125:1659:1715 [1] NCCL INFO Using network FasTrak +n111-046-125:1660:1717 [2] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n111-046-125:1660:1717 [2] NCCL INFO Using network FasTrak +n111-046-125:1662:1719 [4] NCCL INFO DMA-BUF is available on GPU device 4 +n111-046-125:1661:1716 [3] NCCL INFO DMA-BUF is available on GPU device 3 +n111-046-125:1662:1719 [4] NCCL INFO ncclCommInitRankConfig comm 0xbcd41a0 rank 28 nranks 32 cudaDev 4 nvmlDev 4 busId 84000 commId 0x2a4d35ea441cbcba - Init START +n111-046-125:1661:1716 [3] NCCL INFO ncclCommInitRankConfig comm 0xbffbfe0 rank 27 nranks 32 cudaDev 3 nvmlDev 3 busId c000 commId 0x2a4d35ea441cbcba - Init START +n111-046-125:1663:1720 [5] NCCL INFO DMA-BUF is available on GPU device 5 +n111-046-125:1658:1721 [0] NCCL INFO DMA-BUF is available on GPU device 0 +n111-046-125:1663:1720 [5] NCCL INFO ncclCommInitRankConfig comm 0xceea760 rank 29 nranks 32 cudaDev 5 nvmlDev 5 busId 85000 commId 0x2a4d35ea441cbcba - Init START +n111-046-125:1658:1721 [0] NCCL INFO ncclCommInitRankConfig comm 0xc9c9a40 rank 24 nranks 32 cudaDev 0 nvmlDev 0 busId 4000 commId 0x2a4d35ea441cbcba - Init START +n111-046-125:1662:1719 [4] NCCL INFO RAS client listening socket at ::1<28028> +n111-046-125:1664:1714 [6] NCCL INFO DMA-BUF is available on GPU device 6 +n111-046-125:1659:1715 [1] NCCL INFO DMA-BUF is available on GPU device 1 +n111-046-125:1665:1718 [7] NCCL INFO DMA-BUF is available on GPU device 7 +n111-046-125:1664:1714 [6] NCCL INFO ncclCommInitRankConfig comm 0xc0ec280 rank 30 nranks 32 cudaDev 6 nvmlDev 6 busId 8b000 commId 0x2a4d35ea441cbcba - Init START +n111-046-125:1663:1720 [5] NCCL INFO RAS client listening socket at ::1<28028> +n111-046-125:1659:1715 [1] NCCL INFO ncclCommInitRankConfig comm 0xbe02940 rank 25 nranks 32 cudaDev 1 nvmlDev 1 busId 5000 commId 0x2a4d35ea441cbcba - Init START +n111-046-125:1658:1721 [0] NCCL INFO RAS client listening socket at ::1<28028> +n111-046-125:1665:1718 [7] NCCL INFO ncclCommInitRankConfig comm 0xbfcbe50 rank 31 nranks 32 cudaDev 7 nvmlDev 7 busId 8c000 commId 0x2a4d35ea441cbcba - Init START +n111-046-125:1665:1718 [7] NCCL INFO RAS client listening socket at ::1<28028> +n111-046-125:1664:1714 [6] NCCL INFO RAS client listening socket at ::1<28028> +n111-046-125:1660:1717 [2] NCCL INFO DMA-BUF is available on GPU device 2 +n111-046-125:1660:1717 [2] NCCL INFO ncclCommInitRankConfig comm 0xb131d80 rank 26 nranks 32 cudaDev 2 nvmlDev 2 busId b000 commId 0x2a4d35ea441cbcba - Init START +n111-046-125:1659:1715 [1] NCCL INFO RAS client listening socket at ::1<28028> +n111-046-125:1660:1717 [2] NCCL INFO RAS client listening socket at ::1<28028> +n111-046-125:1661:1716 [3] NCCL INFO RAS client listening socket at ::1<28028> +n111-046-125:1659:1715 [1] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n111-046-125:1659:1715 [1] NCCL INFO TOPO/NET : Importing network plugins to topology +n111-046-125:1659:1715 [1] NCCL INFO Retrieving state for FasTrak +n111-046-125:1659:1715 [1] NCCL INFO Initialized state 0 for FasTrak +n111-046-125:1659:1715 [1] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n111-046-125:1659:1715 [1] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n111-046-125:1659:1715 [1] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n111-046-125:1659:1715 [1] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n111-046-125:1659:1715 [1] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n111-046-125:1659:1715 [1] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n111-046-125:1659:1715 [1] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n111-046-125:1659:1715 [1] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n111-046-125:1665:1718 [7] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n111-046-125:1665:1718 [7] NCCL INFO TOPO/NET : Importing network plugins to topology +n111-046-125:1665:1718 [7] NCCL INFO Retrieving state for FasTrak +n111-046-125:1665:1718 [7] NCCL INFO Initialized state 0 for FasTrak +n111-046-125:1665:1718 [7] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n111-046-125:1665:1718 [7] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n111-046-125:1665:1718 [7] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n111-046-125:1665:1718 [7] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n111-046-125:1665:1718 [7] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n111-046-125:1665:1718 [7] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n111-046-125:1665:1718 [7] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n111-046-125:1665:1718 [7] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n111-046-125:1664:1714 [6] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n111-046-125:1664:1714 [6] NCCL INFO TOPO/NET : Importing network plugins to topology +n111-046-125:1664:1714 [6] NCCL INFO Retrieving state for FasTrak +n111-046-125:1664:1714 [6] NCCL INFO Initialized state 0 for FasTrak +n111-046-125:1664:1714 [6] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n111-046-125:1664:1714 [6] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n111-046-125:1664:1714 [6] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n111-046-125:1664:1714 [6] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n111-046-125:1664:1714 [6] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n111-046-125:1662:1719 [4] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n111-046-125:1662:1719 [4] NCCL INFO TOPO/NET : Importing network plugins to topology +n111-046-125:1662:1719 [4] NCCL INFO Retrieving state for FasTrak +n111-046-125:1662:1719 [4] NCCL INFO Initialized state 0 for FasTrak +n111-046-125:1664:1714 [6] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n111-046-125:1664:1714 [6] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n111-046-125:1662:1719 [4] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n111-046-125:1664:1714 [6] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n111-046-125:1662:1719 [4] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n111-046-125:1662:1719 [4] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n111-046-125:1662:1719 [4] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n111-046-125:1660:1717 [2] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n111-046-125:1660:1717 [2] NCCL INFO TOPO/NET : Importing network plugins to topology +n111-046-125:1660:1717 [2] NCCL INFO Retrieving state for FasTrak +n111-046-125:1660:1717 [2] NCCL INFO Initialized state 0 for FasTrak +n111-046-125:1662:1719 [4] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n111-046-125:1662:1719 [4] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n111-046-125:1660:1717 [2] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n111-046-125:1662:1719 [4] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n111-046-125:1660:1717 [2] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n111-046-125:1662:1719 [4] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n111-046-125:1660:1717 [2] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n111-046-125:1660:1717 [2] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n111-046-125:1660:1717 [2] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n111-046-125:1660:1717 [2] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n111-046-125:1660:1717 [2] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n111-046-125:1660:1717 [2] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n111-046-125:1661:1716 [3] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n111-046-125:1661:1716 [3] NCCL INFO TOPO/NET : Importing network plugins to topology +n111-046-125:1661:1716 [3] NCCL INFO Retrieving state for FasTrak +n111-046-125:1661:1716 [3] NCCL INFO Initialized state 0 for FasTrak +n111-046-125:1661:1716 [3] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n111-046-125:1661:1716 [3] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n111-046-125:1661:1716 [3] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n111-046-125:1661:1716 [3] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n111-046-125:1661:1716 [3] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n111-046-125:1661:1716 [3] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n111-046-125:1661:1716 [3] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n111-046-125:1661:1716 [3] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n111-046-125:1663:1720 [5] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n111-046-125:1663:1720 [5] NCCL INFO TOPO/NET : Importing network plugins to topology +n111-046-125:1663:1720 [5] NCCL INFO Retrieving state for FasTrak +n111-046-125:1663:1720 [5] NCCL INFO Initialized state 0 for FasTrak +n111-046-125:1663:1720 [5] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n111-046-125:1663:1720 [5] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n111-046-125:1663:1720 [5] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n111-046-125:1663:1720 [5] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n111-046-125:1663:1720 [5] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n111-046-125:1663:1720 [5] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n111-046-125:1663:1720 [5] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n111-046-125:1663:1720 [5] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n111-046-125:1658:1721 [0] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n111-046-125:1658:1721 [0] NCCL INFO TOPO/NET : Importing network plugins to topology +n111-046-125:1658:1721 [0] NCCL INFO Retrieving state for FasTrak +n111-046-125:1658:1721 [0] NCCL INFO Initialized state 0 for FasTrak +n111-046-125:1658:1721 [0] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n111-046-125:1658:1721 [0] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n111-046-125:1658:1721 [0] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n111-046-125:1658:1721 [0] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n111-046-125:1658:1721 [0] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n111-046-125:1658:1721 [0] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n111-046-125:1658:1721 [0] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n111-046-125:1658:1721 [0] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n111-046-125:1660:1717 [2] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n111-046-125:1659:1715 [1] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n111-046-125:1658:1721 [0] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n111-046-125:1662:1719 [4] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n111-046-125:1664:1714 [6] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n111-046-125:1663:1720 [5] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n111-046-125:1665:1718 [7] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n111-046-125:1661:1716 [3] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n111-046-125:1660:1717 [2] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n111-046-125:1660:1717 [2] NCCL INFO CPU/0-0 (1/1/3) +n111-046-125:1660:1717 [2] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n111-046-125:1660:1717 [2] NCCL INFO + PCI[24.0] - GPU/0-4000 (24) +n111-046-125:1660:1717 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1660:1717 [2] NCCL INFO + PCI[24.0] - NIC/0-6000 +n111-046-125:1660:1717 [2] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n111-046-125:1660:1717 [2] NCCL INFO + PCI[24.0] - NIC/0-7000 +n111-046-125:1660:1717 [2] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n111-046-125:1660:1717 [2] NCCL INFO + PCI[24.0] - GPU/0-5000 (25) +n111-046-125:1660:1717 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1660:1717 [2] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n111-046-125:1660:1717 [2] NCCL INFO + PCI[24.0] - NIC/0-d000 +n111-046-125:1660:1717 [2] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n111-046-125:1660:1717 [2] NCCL INFO + PCI[24.0] - NIC/0-e000 +n111-046-125:1660:1717 [2] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n111-046-125:1660:1717 [2] NCCL INFO + PCI[24.0] - GPU/0-b000 (26) +n111-046-125:1660:1717 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1660:1717 [2] NCCL INFO + PCI[24.0] - GPU/0-c000 (27) +n111-046-125:1660:1717 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1660:1717 [2] NCCL INFO + SYS[22.0] - CPU/0-1 +n111-046-125:1660:1717 [2] NCCL INFO CPU/0-1 (1/1/3) +n111-046-125:1660:1717 [2] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n111-046-125:1660:1717 [2] NCCL INFO + PCI[24.0] - NIC/0-86000 +n111-046-125:1660:1717 [2] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n111-046-125:1660:1717 [2] NCCL INFO + PCI[24.0] - NIC/0-87000 +n111-046-125:1660:1717 [2] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n111-046-125:1660:1717 [2] NCCL INFO + PCI[24.0] - GPU/0-84000 (28) +n111-046-125:1660:1717 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1660:1717 [2] NCCL INFO + PCI[24.0] - GPU/0-85000 (29) +n111-046-125:1660:1717 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1660:1717 [2] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n111-046-125:1660:1717 [2] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n111-046-125:1660:1717 [2] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n111-046-125:1660:1717 [2] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n111-046-125:1660:1717 [2] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n111-046-125:1660:1717 [2] NCCL INFO + PCI[24.0] - GPU/0-8b000 (30) +n111-046-125:1660:1717 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1660:1717 [2] NCCL INFO + PCI[24.0] - GPU/0-8c000 (31) +n111-046-125:1660:1717 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1660:1717 [2] NCCL INFO + SYS[22.0] - CPU/0-0 +n111-046-125:1660:1717 [2] NCCL INFO ========================================== +n111-046-125:1659:1715 [1] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n111-046-125:1659:1715 [1] NCCL INFO CPU/0-0 (1/1/3) +n111-046-125:1660:1717 [2] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-125:1659:1715 [1] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n111-046-125:1659:1715 [1] NCCL INFO + PCI[24.0] - GPU/0-4000 (24) +n111-046-125:1659:1715 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1659:1715 [1] NCCL INFO + PCI[24.0] - NIC/0-6000 +n111-046-125:1659:1715 [1] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n111-046-125:1660:1717 [2] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-125:1659:1715 [1] NCCL INFO + PCI[24.0] - NIC/0-7000 +n111-046-125:1659:1715 [1] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n111-046-125:1660:1717 [2] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-125:1659:1715 [1] NCCL INFO + PCI[24.0] - GPU/0-5000 (25) +n111-046-125:1659:1715 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1659:1715 [1] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n111-046-125:1659:1715 [1] NCCL INFO + PCI[24.0] - NIC/0-d000 +n111-046-125:1659:1715 [1] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n111-046-125:1658:1721 [0] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n111-046-125:1659:1715 [1] NCCL INFO + PCI[24.0] - NIC/0-e000 +n111-046-125:1658:1721 [0] NCCL INFO CPU/0-0 (1/1/3) +n111-046-125:1660:1717 [2] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-125:1659:1715 [1] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n111-046-125:1658:1721 [0] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n111-046-125:1659:1715 [1] NCCL INFO + PCI[24.0] - GPU/0-b000 (26) +n111-046-125:1659:1715 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1658:1721 [0] NCCL INFO + PCI[24.0] - GPU/0-4000 (24) +n111-046-125:1659:1715 [1] NCCL INFO + PCI[24.0] - GPU/0-c000 (27) +n111-046-125:1658:1721 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1659:1715 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1658:1721 [0] NCCL INFO + PCI[24.0] - NIC/0-6000 +n111-046-125:1659:1715 [1] NCCL INFO + SYS[22.0] - CPU/0-1 +n111-046-125:1659:1715 [1] NCCL INFO CPU/0-1 (1/1/3) +n111-046-125:1658:1721 [0] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n111-046-125:1660:1717 [2] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-125:1659:1715 [1] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n111-046-125:1658:1721 [0] NCCL INFO + PCI[24.0] - NIC/0-7000 +n111-046-125:1659:1715 [1] NCCL INFO + PCI[24.0] - NIC/0-86000 +n111-046-125:1658:1721 [0] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n111-046-125:1659:1715 [1] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n111-046-125:1658:1721 [0] NCCL INFO + PCI[24.0] - GPU/0-5000 (25) +n111-046-125:1659:1715 [1] NCCL INFO + PCI[24.0] - NIC/0-87000 +n111-046-125:1658:1721 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1659:1715 [1] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n111-046-125:1658:1721 [0] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n111-046-125:1659:1715 [1] NCCL INFO + PCI[24.0] - GPU/0-84000 (28) +n111-046-125:1664:1714 [6] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n111-046-125:1658:1721 [0] NCCL INFO + PCI[24.0] - NIC/0-d000 +n111-046-125:1660:1717 [2] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-125:1658:1721 [0] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n111-046-125:1659:1715 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1662:1719 [4] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n111-046-125:1664:1714 [6] NCCL INFO CPU/0-0 (1/1/3) +n111-046-125:1658:1721 [0] NCCL INFO + PCI[24.0] - NIC/0-e000 +n111-046-125:1659:1715 [1] NCCL INFO + PCI[24.0] - GPU/0-85000 (29) +n111-046-125:1658:1721 [0] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n111-046-125:1659:1715 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1662:1719 [4] NCCL INFO CPU/0-0 (1/1/3) +n111-046-125:1664:1714 [6] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n111-046-125:1658:1721 [0] NCCL INFO + PCI[24.0] - GPU/0-b000 (26) +n111-046-125:1659:1715 [1] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n111-046-125:1660:1717 [2] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n111-046-125:1659:1715 [1] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n111-046-125:1658:1721 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1664:1714 [6] NCCL INFO + PCI[24.0] - GPU/0-4000 (24) +n111-046-125:1658:1721 [0] NCCL INFO + PCI[24.0] - GPU/0-c000 (27) +n111-046-125:1662:1719 [4] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n111-046-125:1659:1715 [1] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n111-046-125:1664:1714 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1658:1721 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1659:1715 [1] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n111-046-125:1662:1719 [4] NCCL INFO + PCI[24.0] - GPU/0-4000 (24) +n111-046-125:1664:1714 [6] NCCL INFO + PCI[24.0] - NIC/0-6000 +n111-046-125:1658:1721 [0] NCCL INFO + SYS[22.0] - CPU/0-1 +n111-046-125:1659:1715 [1] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n111-046-125:1662:1719 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1658:1721 [0] NCCL INFO CPU/0-1 (1/1/3) +n111-046-125:1659:1715 [1] NCCL INFO + PCI[24.0] - GPU/0-8b000 (30) +n111-046-125:1664:1714 [6] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n111-046-125:1662:1719 [4] NCCL INFO + PCI[24.0] - NIC/0-6000 +n111-046-125:1659:1715 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1660:1717 [2] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n111-046-125:1658:1721 [0] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n111-046-125:1664:1714 [6] NCCL INFO + PCI[24.0] - NIC/0-7000 +n111-046-125:1659:1715 [1] NCCL INFO + PCI[24.0] - GPU/0-8c000 (31) +n111-046-125:1662:1719 [4] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n111-046-125:1664:1714 [6] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n111-046-125:1658:1721 [0] NCCL INFO + PCI[24.0] - NIC/0-86000 +n111-046-125:1659:1715 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1664:1714 [6] NCCL INFO + PCI[24.0] - GPU/0-5000 (25) +n111-046-125:1662:1719 [4] NCCL INFO + PCI[24.0] - NIC/0-7000 +n111-046-125:1659:1715 [1] NCCL INFO + SYS[22.0] - CPU/0-0 +n111-046-125:1658:1721 [0] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n111-046-125:1664:1714 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1662:1719 [4] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n111-046-125:1659:1715 [1] NCCL INFO ========================================== +n111-046-125:1664:1714 [6] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n111-046-125:1658:1721 [0] NCCL INFO + PCI[24.0] - NIC/0-87000 +n111-046-125:1660:1717 [2] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-125:1662:1719 [4] NCCL INFO + PCI[24.0] - GPU/0-5000 (25) +n111-046-125:1664:1714 [6] NCCL INFO + PCI[24.0] - NIC/0-d000 +n111-046-125:1658:1721 [0] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n111-046-125:1662:1719 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1664:1714 [6] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n111-046-125:1658:1721 [0] NCCL INFO + PCI[24.0] - GPU/0-84000 (28) +n111-046-125:1662:1719 [4] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n111-046-125:1664:1714 [6] NCCL INFO + PCI[24.0] - NIC/0-e000 +n111-046-125:1658:1721 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1662:1719 [4] NCCL INFO + PCI[24.0] - NIC/0-d000 +n111-046-125:1658:1721 [0] NCCL INFO + PCI[24.0] - GPU/0-85000 (29) +n111-046-125:1664:1714 [6] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n111-046-125:1659:1715 [1] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-125:1660:1717 [2] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-125:1662:1719 [4] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n111-046-125:1664:1714 [6] NCCL INFO + PCI[24.0] - GPU/0-b000 (26) +n111-046-125:1658:1721 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1662:1719 [4] NCCL INFO + PCI[24.0] - NIC/0-e000 +n111-046-125:1664:1714 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1658:1721 [0] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n111-046-125:1662:1719 [4] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n111-046-125:1664:1714 [6] NCCL INFO + PCI[24.0] - GPU/0-c000 (27) +n111-046-125:1658:1721 [0] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n111-046-125:1662:1719 [4] NCCL INFO + PCI[24.0] - GPU/0-b000 (26) +n111-046-125:1664:1714 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1658:1721 [0] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n111-046-125:1658:1721 [0] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n111-046-125:1660:1717 [2] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-125:1659:1715 [1] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-125:1658:1721 [0] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n111-046-125:1664:1714 [6] NCCL INFO + SYS[22.0] - CPU/0-1 +n111-046-125:1662:1719 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1664:1714 [6] NCCL INFO CPU/0-1 (1/1/3) +n111-046-125:1658:1721 [0] NCCL INFO + PCI[24.0] - GPU/0-8b000 (30) +n111-046-125:1662:1719 [4] NCCL INFO + PCI[24.0] - GPU/0-c000 (27) +n111-046-125:1664:1714 [6] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n111-046-125:1662:1719 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1658:1721 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1664:1714 [6] NCCL INFO + PCI[24.0] - NIC/0-86000 +n111-046-125:1658:1721 [0] NCCL INFO + PCI[24.0] - GPU/0-8c000 (31) +n111-046-125:1662:1719 [4] NCCL INFO + SYS[22.0] - CPU/0-1 +n111-046-125:1658:1721 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1660:1717 [2] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-125:1664:1714 [6] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n111-046-125:1662:1719 [4] NCCL INFO CPU/0-1 (1/1/3) +n111-046-125:1658:1721 [0] NCCL INFO + SYS[22.0] - CPU/0-0 +n111-046-125:1664:1714 [6] NCCL INFO + PCI[24.0] - NIC/0-87000 +n111-046-125:1659:1715 [1] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-125:1658:1721 [0] NCCL INFO ========================================== +n111-046-125:1662:1719 [4] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n111-046-125:1664:1714 [6] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n111-046-125:1662:1719 [4] NCCL INFO + PCI[24.0] - NIC/0-86000 +n111-046-125:1664:1714 [6] NCCL INFO + PCI[24.0] - GPU/0-84000 (28) +n111-046-125:1662:1719 [4] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n111-046-125:1664:1714 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1660:1717 [2] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n111-046-125:1662:1719 [4] NCCL INFO + PCI[24.0] - NIC/0-87000 +n111-046-125:1664:1714 [6] NCCL INFO + PCI[24.0] - GPU/0-85000 (29) +n111-046-125:1664:1714 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1662:1719 [4] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n111-046-125:1659:1715 [1] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-125:1664:1714 [6] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n111-046-125:1662:1719 [4] NCCL INFO + PCI[24.0] - GPU/0-84000 (28) +n111-046-125:1664:1714 [6] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n111-046-125:1658:1721 [0] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-125:1662:1719 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1664:1714 [6] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n111-046-125:1662:1719 [4] NCCL INFO + PCI[24.0] - GPU/0-85000 (29) +n111-046-125:1660:1717 [2] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n111-046-125:1662:1719 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1664:1714 [6] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n111-046-125:1662:1719 [4] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n111-046-125:1664:1714 [6] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n111-046-125:1662:1719 [4] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n111-046-125:1664:1714 [6] NCCL INFO + PCI[24.0] - GPU/0-8b000 (30) +n111-046-125:1659:1715 [1] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-125:1662:1719 [4] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n111-046-125:1658:1721 [0] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-125:1664:1714 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1662:1719 [4] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n111-046-125:1664:1714 [6] NCCL INFO + PCI[24.0] - GPU/0-8c000 (31) +n111-046-125:1660:1717 [2] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n111-046-125:1662:1719 [4] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n111-046-125:1664:1714 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1662:1719 [4] NCCL INFO + PCI[24.0] - GPU/0-8b000 (30) +n111-046-125:1664:1714 [6] NCCL INFO + SYS[22.0] - CPU/0-0 +n111-046-125:1659:1715 [1] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-125:1662:1719 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1664:1714 [6] NCCL INFO ========================================== +n111-046-125:1658:1721 [0] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-125:1662:1719 [4] NCCL INFO + PCI[24.0] - GPU/0-8c000 (31) +n111-046-125:1662:1719 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1660:1717 [2] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n111-046-125:1662:1719 [4] NCCL INFO + SYS[22.0] - CPU/0-0 +n111-046-125:1662:1719 [4] NCCL INFO ========================================== +n111-046-125:1659:1715 [1] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n111-046-125:1658:1721 [0] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-125:1664:1714 [6] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-125:1659:1715 [1] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n111-046-125:1662:1719 [4] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-125:1660:1717 [2] NCCL INFO Setting affinity for GPU 2 to 0fffffff,fffffe00,00000000,000fffff,fffffffe +n111-046-125:1658:1721 [0] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-125:1664:1714 [6] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-125:1659:1715 [1] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-125:1662:1719 [4] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-125:1658:1721 [0] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-125:1664:1714 [6] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-125:1659:1715 [1] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-125:1662:1719 [4] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-125:1664:1714 [6] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-125:1659:1715 [1] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-125:1658:1721 [0] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n111-046-125:1662:1719 [4] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-125:1659:1715 [1] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-125:1658:1721 [0] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n111-046-125:1664:1714 [6] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-125:1659:1715 [1] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n111-046-125:1658:1721 [0] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-125:1664:1714 [6] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-125:1662:1719 [4] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-125:1658:1721 [0] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-125:1659:1715 [1] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n111-046-125:1664:1714 [6] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n111-046-125:1662:1719 [4] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-125:1658:1721 [0] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-125:1660:1717 [2] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n111-046-125:1659:1715 [1] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n111-046-125:1664:1714 [6] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n111-046-125:1658:1721 [0] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-125:1659:1715 [1] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n111-046-125:1664:1714 [6] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-125:1658:1721 [0] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n111-046-125:1664:1714 [6] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-125:1659:1715 [1] NCCL INFO Setting affinity for GPU 1 to 0fffffff,fffffe00,00000000,000fffff,fffffffe +n111-046-125:1662:1719 [4] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n111-046-125:1658:1721 [0] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n111-046-125:1664:1714 [6] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-125:1659:1715 [1] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n111-046-125:1658:1721 [0] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n111-046-125:1664:1714 [6] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-125:1662:1719 [4] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n111-046-125:1658:1721 [0] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n111-046-125:1664:1714 [6] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n111-046-125:1662:1719 [4] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-125:1658:1721 [0] NCCL INFO Setting affinity for GPU 0 to 0fffffff,fffffe00,00000000,000fffff,fffffffe +n111-046-125:1664:1714 [6] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n111-046-125:1660:1717 [2] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n111-046-125:1659:1715 [1] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n111-046-125:1658:1721 [0] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n111-046-125:1664:1714 [6] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n111-046-125:1662:1719 [4] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-125:1664:1714 [6] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n111-046-125:1662:1719 [4] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-125:1664:1714 [6] NCCL INFO Setting affinity for GPU 6 to 7fff,ffffffff,e0000000,0000007f,ffffffff,ffe00000,00000000 +n111-046-125:1662:1719 [4] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-125:1658:1721 [0] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n111-046-125:1662:1719 [4] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n111-046-125:1662:1719 [4] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n111-046-125:1662:1719 [4] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n111-046-125:1662:1719 [4] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n111-046-125:1662:1719 [4] NCCL INFO Setting affinity for GPU 4 to 7fff,ffffffff,e0000000,0000007f,ffffffff,ffe00000,00000000 +n111-046-125:1664:1714 [6] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n111-046-125:1664:1714 [6] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n111-046-125:1663:1720 [5] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n111-046-125:1663:1720 [5] NCCL INFO CPU/0-0 (1/1/3) +n111-046-125:1663:1720 [5] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n111-046-125:1662:1719 [4] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n111-046-125:1663:1720 [5] NCCL INFO + PCI[24.0] - GPU/0-4000 (24) +n111-046-125:1663:1720 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1663:1720 [5] NCCL INFO + PCI[24.0] - NIC/0-6000 +n111-046-125:1663:1720 [5] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n111-046-125:1663:1720 [5] NCCL INFO + PCI[24.0] - NIC/0-7000 +n111-046-125:1663:1720 [5] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n111-046-125:1663:1720 [5] NCCL INFO + PCI[24.0] - GPU/0-5000 (25) +n111-046-125:1663:1720 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1663:1720 [5] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n111-046-125:1663:1720 [5] NCCL INFO + PCI[24.0] - NIC/0-d000 +n111-046-125:1663:1720 [5] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n111-046-125:1663:1720 [5] NCCL INFO + PCI[24.0] - NIC/0-e000 +n111-046-125:1663:1720 [5] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n111-046-125:1663:1720 [5] NCCL INFO + PCI[24.0] - GPU/0-b000 (26) +n111-046-125:1663:1720 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1663:1720 [5] NCCL INFO + PCI[24.0] - GPU/0-c000 (27) +n111-046-125:1663:1720 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1663:1720 [5] NCCL INFO + SYS[22.0] - CPU/0-1 +n111-046-125:1663:1720 [5] NCCL INFO CPU/0-1 (1/1/3) +n111-046-125:1663:1720 [5] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n111-046-125:1663:1720 [5] NCCL INFO + PCI[24.0] - NIC/0-86000 +n111-046-125:1663:1720 [5] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n111-046-125:1663:1720 [5] NCCL INFO + PCI[24.0] - NIC/0-87000 +n111-046-125:1662:1719 [4] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n111-046-125:1663:1720 [5] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n111-046-125:1663:1720 [5] NCCL INFO + PCI[24.0] - GPU/0-84000 (28) +n111-046-125:1663:1720 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1663:1720 [5] NCCL INFO + PCI[24.0] - GPU/0-85000 (29) +n111-046-125:1663:1720 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1663:1720 [5] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n111-046-125:1663:1720 [5] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n111-046-125:1663:1720 [5] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n111-046-125:1663:1720 [5] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n111-046-125:1663:1720 [5] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n111-046-125:1663:1720 [5] NCCL INFO + PCI[24.0] - GPU/0-8b000 (30) +n111-046-125:1663:1720 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1663:1720 [5] NCCL INFO + PCI[24.0] - GPU/0-8c000 (31) +n111-046-125:1663:1720 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1663:1720 [5] NCCL INFO + SYS[22.0] - CPU/0-0 +n111-046-125:1663:1720 [5] NCCL INFO ========================================== +n111-046-125:1663:1720 [5] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-125:1663:1720 [5] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-125:1663:1720 [5] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-125:1663:1720 [5] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-125:1663:1720 [5] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-125:1663:1720 [5] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-125:1663:1720 [5] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n111-046-125:1661:1716 [3] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n111-046-125:1661:1716 [3] NCCL INFO CPU/0-0 (1/1/3) +n111-046-125:1663:1720 [5] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n111-046-125:1661:1716 [3] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n111-046-125:1661:1716 [3] NCCL INFO + PCI[24.0] - GPU/0-4000 (24) +n111-046-125:1661:1716 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1661:1716 [3] NCCL INFO + PCI[24.0] - NIC/0-6000 +n111-046-125:1661:1716 [3] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n111-046-125:1663:1720 [5] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-125:1661:1716 [3] NCCL INFO + PCI[24.0] - NIC/0-7000 +n111-046-125:1661:1716 [3] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n111-046-125:1661:1716 [3] NCCL INFO + PCI[24.0] - GPU/0-5000 (25) +n111-046-125:1661:1716 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1661:1716 [3] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n111-046-125:1661:1716 [3] NCCL INFO + PCI[24.0] - NIC/0-d000 +n111-046-125:1665:1718 [7] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n111-046-125:1661:1716 [3] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n111-046-125:1663:1720 [5] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-125:1661:1716 [3] NCCL INFO + PCI[24.0] - NIC/0-e000 +n111-046-125:1665:1718 [7] NCCL INFO CPU/0-0 (1/1/3) +n111-046-125:1661:1716 [3] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n111-046-125:1665:1718 [7] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n111-046-125:1661:1716 [3] NCCL INFO + PCI[24.0] - GPU/0-b000 (26) +n111-046-125:1665:1718 [7] NCCL INFO + PCI[24.0] - GPU/0-4000 (24) +n111-046-125:1661:1716 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1665:1718 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1661:1716 [3] NCCL INFO + PCI[24.0] - GPU/0-c000 (27) +n111-046-125:1665:1718 [7] NCCL INFO + PCI[24.0] - NIC/0-6000 +n111-046-125:1661:1716 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1665:1718 [7] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n111-046-125:1663:1720 [5] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-125:1665:1718 [7] NCCL INFO + PCI[24.0] - NIC/0-7000 +n111-046-125:1661:1716 [3] NCCL INFO + SYS[22.0] - CPU/0-1 +n111-046-125:1665:1718 [7] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n111-046-125:1661:1716 [3] NCCL INFO CPU/0-1 (1/1/3) +n111-046-125:1665:1718 [7] NCCL INFO + PCI[24.0] - GPU/0-5000 (25) +n111-046-125:1661:1716 [3] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n111-046-125:1665:1718 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1661:1716 [3] NCCL INFO + PCI[24.0] - NIC/0-86000 +n111-046-125:1665:1718 [7] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n111-046-125:1661:1716 [3] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n111-046-125:1665:1718 [7] NCCL INFO + PCI[24.0] - NIC/0-d000 +n111-046-125:1661:1716 [3] NCCL INFO + PCI[24.0] - NIC/0-87000 +n111-046-125:1663:1720 [5] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-125:1665:1718 [7] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n111-046-125:1661:1716 [3] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n111-046-125:1665:1718 [7] NCCL INFO + PCI[24.0] - NIC/0-e000 +n111-046-125:1661:1716 [3] NCCL INFO + PCI[24.0] - GPU/0-84000 (28) +n111-046-125:1661:1716 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1665:1718 [7] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n111-046-125:1661:1716 [3] NCCL INFO + PCI[24.0] - GPU/0-85000 (29) +n111-046-125:1665:1718 [7] NCCL INFO + PCI[24.0] - GPU/0-b000 (26) +n111-046-125:1661:1716 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1665:1718 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1661:1716 [3] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n111-046-125:1665:1718 [7] NCCL INFO + PCI[24.0] - GPU/0-c000 (27) +n111-046-125:1661:1716 [3] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n111-046-125:1665:1718 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1663:1720 [5] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n111-046-125:1665:1718 [7] NCCL INFO + SYS[22.0] - CPU/0-1 +n111-046-125:1661:1716 [3] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n111-046-125:1665:1718 [7] NCCL INFO CPU/0-1 (1/1/3) +n111-046-125:1661:1716 [3] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n111-046-125:1665:1718 [7] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n111-046-125:1661:1716 [3] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n111-046-125:1665:1718 [7] NCCL INFO + PCI[24.0] - NIC/0-86000 +n111-046-125:1661:1716 [3] NCCL INFO + PCI[24.0] - GPU/0-8b000 (30) +n111-046-125:1665:1718 [7] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n111-046-125:1661:1716 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1665:1718 [7] NCCL INFO + PCI[24.0] - NIC/0-87000 +n111-046-125:1661:1716 [3] NCCL INFO + PCI[24.0] - GPU/0-8c000 (31) +n111-046-125:1665:1718 [7] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n111-046-125:1661:1716 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1663:1720 [5] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n111-046-125:1661:1716 [3] NCCL INFO + SYS[22.0] - CPU/0-0 +n111-046-125:1661:1716 [3] NCCL INFO ========================================== +n111-046-125:1665:1718 [7] NCCL INFO + PCI[24.0] - GPU/0-84000 (28) +n111-046-125:1665:1718 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1665:1718 [7] NCCL INFO + PCI[24.0] - GPU/0-85000 (29) +n111-046-125:1665:1718 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1665:1718 [7] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n111-046-125:1665:1718 [7] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n111-046-125:1663:1720 [5] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n111-046-125:1665:1718 [7] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n111-046-125:1661:1716 [3] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-125:1665:1718 [7] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n111-046-125:1665:1718 [7] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n111-046-125:1665:1718 [7] NCCL INFO + PCI[24.0] - GPU/0-8b000 (30) +n111-046-125:1665:1718 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1665:1718 [7] NCCL INFO + PCI[24.0] - GPU/0-8c000 (31) +n111-046-125:1663:1720 [5] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n111-046-125:1665:1718 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-125:1665:1718 [7] NCCL INFO + SYS[22.0] - CPU/0-0 +n111-046-125:1661:1716 [3] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-125:1665:1718 [7] NCCL INFO ========================================== +n111-046-125:1663:1720 [5] NCCL INFO Setting affinity for GPU 5 to 7fff,ffffffff,e0000000,0000007f,ffffffff,ffe00000,00000000 +n111-046-125:1661:1716 [3] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-125:1665:1718 [7] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-125:1663:1720 [5] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n111-046-125:1661:1716 [3] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-125:1665:1718 [7] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-125:1665:1718 [7] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-125:1661:1716 [3] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-125:1665:1718 [7] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-125:1663:1720 [5] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n111-046-125:1661:1716 [3] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-125:1661:1716 [3] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n111-046-125:1665:1718 [7] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-125:1665:1718 [7] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-125:1661:1716 [3] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n111-046-125:1661:1716 [3] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-125:1665:1718 [7] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n111-046-125:1665:1718 [7] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n111-046-125:1661:1716 [3] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-125:1661:1716 [3] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-125:1665:1718 [7] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-125:1665:1718 [7] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-125:1661:1716 [3] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-125:1661:1716 [3] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n111-046-125:1665:1718 [7] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-125:1661:1716 [3] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n111-046-125:1665:1718 [7] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-125:1661:1716 [3] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n111-046-125:1665:1718 [7] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n111-046-125:1661:1716 [3] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n111-046-125:1665:1718 [7] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n111-046-125:1661:1716 [3] NCCL INFO Setting affinity for GPU 3 to 0fffffff,fffffe00,00000000,000fffff,fffffffe +n111-046-125:1665:1718 [7] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n111-046-125:1661:1716 [3] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n111-046-125:1665:1718 [7] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n111-046-125:1665:1718 [7] NCCL INFO Setting affinity for GPU 7 to 7fff,ffffffff,e0000000,0000007f,ffffffff,ffe00000,00000000 +n111-046-125:1661:1716 [3] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n111-046-125:1665:1718 [7] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n111-046-125:1665:1718 [7] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n111-046-125:1662:1719 [4] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n111-046-125:1662:1719 [4] NCCL INFO 0 : NET/0-0 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 NET/0-0 +n111-046-125:1662:1719 [4] NCCL INFO 1 : NET/0-1 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 NET/0-1 +n111-046-125:1662:1719 [4] NCCL INFO 2 : NET/0-2 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 NET/0-2 +n111-046-125:1662:1719 [4] NCCL INFO 3 : NET/0-3 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 NET/0-3 +n111-046-125:1662:1719 [4] NCCL INFO 4 : NET/0-4 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 NET/0-4 +n111-046-125:1662:1719 [4] NCCL INFO 5 : NET/0-5 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 NET/0-5 +n111-046-125:1662:1719 [4] NCCL INFO 6 : NET/0-6 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 NET/0-6 +n111-046-125:1662:1719 [4] NCCL INFO 7 : NET/0-7 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 NET/0-7 +n111-046-125:1659:1715 [1] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n111-046-125:1659:1715 [1] NCCL INFO 0 : NET/0-0 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 NET/0-0 +n111-046-125:1659:1715 [1] NCCL INFO 1 : NET/0-1 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 NET/0-1 +n111-046-125:1659:1715 [1] NCCL INFO 2 : NET/0-2 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 NET/0-2 +n111-046-125:1659:1715 [1] NCCL INFO 3 : NET/0-3 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 NET/0-3 +n111-046-125:1659:1715 [1] NCCL INFO 4 : NET/0-4 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 NET/0-4 +n111-046-125:1659:1715 [1] NCCL INFO 5 : NET/0-5 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 NET/0-5 +n111-046-125:1659:1715 [1] NCCL INFO 6 : NET/0-6 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 NET/0-6 +n111-046-125:1659:1715 [1] NCCL INFO 7 : NET/0-7 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 NET/0-7 +n111-046-125:1662:1719 [4] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n111-046-125:1662:1719 [4] NCCL INFO 0 : NET/0-0 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 NET/0-0 +n111-046-125:1662:1719 [4] NCCL INFO 1 : NET/0-1 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 NET/0-1 +n111-046-125:1662:1719 [4] NCCL INFO 2 : NET/0-2 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 NET/0-2 +n111-046-125:1662:1719 [4] NCCL INFO 3 : NET/0-3 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 NET/0-3 +n111-046-125:1662:1719 [4] NCCL INFO 4 : NET/0-4 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 NET/0-4 +n111-046-125:1662:1719 [4] NCCL INFO 5 : NET/0-5 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 NET/0-5 +n111-046-125:1662:1719 [4] NCCL INFO 6 : NET/0-6 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 NET/0-6 +n111-046-125:1662:1719 [4] NCCL INFO 7 : NET/0-7 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 NET/0-7 +n111-046-125:1659:1715 [1] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n111-046-125:1659:1715 [1] NCCL INFO 0 : NET/0-0 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 NET/0-0 +n111-046-125:1659:1715 [1] NCCL INFO 1 : NET/0-1 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 NET/0-1 +n111-046-125:1659:1715 [1] NCCL INFO 2 : NET/0-2 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 NET/0-2 +n111-046-125:1659:1715 [1] NCCL INFO 3 : NET/0-3 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 NET/0-3 +n111-046-125:1659:1715 [1] NCCL INFO 4 : NET/0-4 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 NET/0-4 +n111-046-125:1659:1715 [1] NCCL INFO 5 : NET/0-5 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 NET/0-5 +n111-046-125:1659:1715 [1] NCCL INFO 6 : NET/0-6 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 NET/0-6 +n111-046-125:1659:1715 [1] NCCL INFO 7 : NET/0-7 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 NET/0-7 +n111-046-125:1660:1717 [2] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n111-046-125:1660:1717 [2] NCCL INFO 0 : NET/0-0 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 NET/0-0 +n111-046-125:1660:1717 [2] NCCL INFO 1 : NET/0-1 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 NET/0-1 +n111-046-125:1660:1717 [2] NCCL INFO 2 : NET/0-2 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 NET/0-2 +n111-046-125:1660:1717 [2] NCCL INFO 3 : NET/0-3 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 NET/0-3 +n111-046-125:1660:1717 [2] NCCL INFO 4 : NET/0-4 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 NET/0-4 +n111-046-125:1660:1717 [2] NCCL INFO 5 : NET/0-5 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 NET/0-5 +n111-046-125:1660:1717 [2] NCCL INFO 6 : NET/0-6 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 NET/0-6 +n111-046-125:1660:1717 [2] NCCL INFO 7 : NET/0-7 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 NET/0-7 +n111-046-125:1661:1716 [3] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n111-046-125:1661:1716 [3] NCCL INFO 0 : NET/0-0 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 NET/0-0 +n111-046-125:1661:1716 [3] NCCL INFO 1 : NET/0-1 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 NET/0-1 +n111-046-125:1660:1717 [2] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n111-046-125:1661:1716 [3] NCCL INFO 2 : NET/0-2 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 NET/0-2 +n111-046-125:1661:1716 [3] NCCL INFO 3 : NET/0-3 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 NET/0-3 +n111-046-125:1660:1717 [2] NCCL INFO 0 : NET/0-0 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 NET/0-0 +n111-046-125:1661:1716 [3] NCCL INFO 4 : NET/0-4 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 NET/0-4 +n111-046-125:1660:1717 [2] NCCL INFO 1 : NET/0-1 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 NET/0-1 +n111-046-125:1661:1716 [3] NCCL INFO 5 : NET/0-5 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 NET/0-5 +n111-046-125:1660:1717 [2] NCCL INFO 2 : NET/0-2 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 NET/0-2 +n111-046-125:1660:1717 [2] NCCL INFO 3 : NET/0-3 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 NET/0-3 +n111-046-125:1660:1717 [2] NCCL INFO 4 : NET/0-4 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 NET/0-4 +n111-046-125:1661:1716 [3] NCCL INFO 6 : NET/0-6 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 NET/0-6 +n111-046-125:1660:1717 [2] NCCL INFO 5 : NET/0-5 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 NET/0-5 +n111-046-125:1661:1716 [3] NCCL INFO 7 : NET/0-7 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 NET/0-7 +n111-046-125:1660:1717 [2] NCCL INFO 6 : NET/0-6 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 NET/0-6 +n111-046-125:1660:1717 [2] NCCL INFO 7 : NET/0-7 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 NET/0-7 +n111-046-125:1661:1716 [3] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n111-046-125:1661:1716 [3] NCCL INFO 0 : NET/0-0 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 NET/0-0 +n111-046-125:1661:1716 [3] NCCL INFO 1 : NET/0-1 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 NET/0-1 +n111-046-125:1661:1716 [3] NCCL INFO 2 : NET/0-2 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 NET/0-2 +n111-046-125:1661:1716 [3] NCCL INFO 3 : NET/0-3 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 NET/0-3 +n111-046-125:1661:1716 [3] NCCL INFO 4 : NET/0-4 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 NET/0-4 +n111-046-125:1661:1716 [3] NCCL INFO 5 : NET/0-5 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 NET/0-5 +n111-046-125:1661:1716 [3] NCCL INFO 6 : NET/0-6 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 NET/0-6 +n111-046-125:1661:1716 [3] NCCL INFO 7 : NET/0-7 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 NET/0-7 +n111-046-125:1658:1721 [0] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n111-046-125:1658:1721 [0] NCCL INFO 0 : NET/0-0 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 NET/0-0 +n111-046-125:1658:1721 [0] NCCL INFO 1 : NET/0-1 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 NET/0-1 +n111-046-125:1658:1721 [0] NCCL INFO 2 : NET/0-2 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 NET/0-2 +n111-046-125:1658:1721 [0] NCCL INFO 3 : NET/0-3 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 NET/0-3 +n111-046-125:1658:1721 [0] NCCL INFO 4 : NET/0-4 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 NET/0-4 +n111-046-125:1658:1721 [0] NCCL INFO 5 : NET/0-5 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 NET/0-5 +n111-046-125:1658:1721 [0] NCCL INFO 6 : NET/0-6 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 NET/0-6 +n111-046-125:1658:1721 [0] NCCL INFO 7 : NET/0-7 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 NET/0-7 +n111-046-125:1658:1721 [0] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n111-046-125:1658:1721 [0] NCCL INFO 0 : NET/0-0 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 NET/0-0 +n111-046-125:1658:1721 [0] NCCL INFO 1 : NET/0-1 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 NET/0-1 +n111-046-125:1658:1721 [0] NCCL INFO 2 : NET/0-2 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 NET/0-2 +n111-046-125:1658:1721 [0] NCCL INFO 3 : NET/0-3 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 NET/0-3 +n111-046-125:1658:1721 [0] NCCL INFO 4 : NET/0-4 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 NET/0-4 +n111-046-125:1658:1721 [0] NCCL INFO 5 : NET/0-5 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 NET/0-5 +n111-046-125:1658:1721 [0] NCCL INFO 6 : NET/0-6 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 NET/0-6 +n111-046-125:1658:1721 [0] NCCL INFO 7 : NET/0-7 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 NET/0-7 +n111-046-125:1663:1720 [5] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n111-046-125:1663:1720 [5] NCCL INFO 0 : NET/0-0 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 NET/0-0 +n111-046-125:1663:1720 [5] NCCL INFO 1 : NET/0-1 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 NET/0-1 +n111-046-125:1663:1720 [5] NCCL INFO 2 : NET/0-2 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 NET/0-2 +n111-046-125:1663:1720 [5] NCCL INFO 3 : NET/0-3 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 NET/0-3 +n111-046-125:1663:1720 [5] NCCL INFO 4 : NET/0-4 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 NET/0-4 +n111-046-125:1663:1720 [5] NCCL INFO 5 : NET/0-5 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 NET/0-5 +n111-046-125:1663:1720 [5] NCCL INFO 6 : NET/0-6 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 NET/0-6 +n111-046-125:1663:1720 [5] NCCL INFO 7 : NET/0-7 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 NET/0-7 +n111-046-125:1663:1720 [5] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n111-046-125:1663:1720 [5] NCCL INFO 0 : NET/0-0 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 NET/0-0 +n111-046-125:1663:1720 [5] NCCL INFO 1 : NET/0-1 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 NET/0-1 +n111-046-125:1663:1720 [5] NCCL INFO 2 : NET/0-2 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 NET/0-2 +n111-046-125:1663:1720 [5] NCCL INFO 3 : NET/0-3 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 NET/0-3 +n111-046-125:1663:1720 [5] NCCL INFO 4 : NET/0-4 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 NET/0-4 +n111-046-125:1663:1720 [5] NCCL INFO 5 : NET/0-5 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 NET/0-5 +n111-046-125:1663:1720 [5] NCCL INFO 6 : NET/0-6 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 NET/0-6 +n111-046-125:1663:1720 [5] NCCL INFO 7 : NET/0-7 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 NET/0-7 +n111-046-125:1665:1718 [7] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n111-046-125:1665:1718 [7] NCCL INFO 0 : NET/0-0 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 NET/0-0 +n111-046-125:1665:1718 [7] NCCL INFO 1 : NET/0-1 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 NET/0-1 +n111-046-125:1665:1718 [7] NCCL INFO 2 : NET/0-2 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 NET/0-2 +n111-046-125:1665:1718 [7] NCCL INFO 3 : NET/0-3 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 NET/0-3 +n111-046-125:1665:1718 [7] NCCL INFO 4 : NET/0-4 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 NET/0-4 +n111-046-125:1665:1718 [7] NCCL INFO 5 : NET/0-5 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 NET/0-5 +n111-046-125:1665:1718 [7] NCCL INFO 6 : NET/0-6 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 NET/0-6 +n111-046-125:1665:1718 [7] NCCL INFO 7 : NET/0-7 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 NET/0-7 +n111-046-125:1665:1718 [7] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n111-046-125:1665:1718 [7] NCCL INFO 0 : NET/0-0 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 NET/0-0 +n111-046-125:1665:1718 [7] NCCL INFO 1 : NET/0-1 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 NET/0-1 +n111-046-125:1665:1718 [7] NCCL INFO 2 : NET/0-2 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 NET/0-2 +n111-046-125:1665:1718 [7] NCCL INFO 3 : NET/0-3 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 NET/0-3 +n111-046-125:1665:1718 [7] NCCL INFO 4 : NET/0-4 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 NET/0-4 +n111-046-125:1665:1718 [7] NCCL INFO 5 : NET/0-5 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 NET/0-5 +n111-046-125:1665:1718 [7] NCCL INFO 6 : NET/0-6 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 NET/0-6 +n111-046-125:1665:1718 [7] NCCL INFO 7 : NET/0-7 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 NET/0-7 +n111-046-125:1664:1714 [6] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n111-046-125:1664:1714 [6] NCCL INFO 0 : NET/0-0 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 NET/0-0 +n111-046-125:1664:1714 [6] NCCL INFO 1 : NET/0-1 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 NET/0-1 +n111-046-125:1664:1714 [6] NCCL INFO 2 : NET/0-2 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 NET/0-2 +n111-046-125:1664:1714 [6] NCCL INFO 3 : NET/0-3 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 NET/0-3 +n111-046-125:1664:1714 [6] NCCL INFO 4 : NET/0-4 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 NET/0-4 +n111-046-125:1664:1714 [6] NCCL INFO 5 : NET/0-5 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 NET/0-5 +n111-046-125:1664:1714 [6] NCCL INFO 6 : NET/0-6 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 NET/0-6 +n111-046-125:1664:1714 [6] NCCL INFO 7 : NET/0-7 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 NET/0-7 +n111-046-125:1664:1714 [6] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n111-046-125:1664:1714 [6] NCCL INFO 0 : NET/0-0 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 NET/0-0 +n111-046-125:1664:1714 [6] NCCL INFO 1 : NET/0-1 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 NET/0-1 +n111-046-125:1664:1714 [6] NCCL INFO 2 : NET/0-2 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 NET/0-2 +n111-046-125:1664:1714 [6] NCCL INFO 3 : NET/0-3 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 NET/0-3 +n111-046-125:1664:1714 [6] NCCL INFO 4 : NET/0-4 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 NET/0-4 +n111-046-125:1664:1714 [6] NCCL INFO 5 : NET/0-5 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 NET/0-5 +n111-046-125:1664:1714 [6] NCCL INFO 6 : NET/0-6 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 NET/0-6 +n111-046-125:1664:1714 [6] NCCL INFO 7 : NET/0-7 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 NET/0-7 +n111-046-125:1658:1721 [0] NCCL INFO comm 0xc9c9a40 rank 24 nRanks 32 nNodes 4 localRanks 8 localRank 0 MNNVL 0 +n111-046-125:1659:1715 [1] NCCL INFO comm 0xbe02940 rank 25 nRanks 32 nNodes 4 localRanks 8 localRank 1 MNNVL 0 +n111-046-125:1660:1717 [2] NCCL INFO comm 0xb131d80 rank 26 nRanks 32 nNodes 4 localRanks 8 localRank 2 MNNVL 0 +n111-046-125:1661:1716 [3] NCCL INFO comm 0xbffbfe0 rank 27 nRanks 32 nNodes 4 localRanks 8 localRank 3 MNNVL 0 +n111-046-125:1658:1721 [0] NCCL INFO Tree 0 : 16 -> 24 -> 25/-1/-1 +n111-046-125:1658:1721 [0] NCCL INFO Tree 8 : -1 -> 24 -> 25/8/-1 +n111-046-125:1662:1719 [4] NCCL INFO comm 0xbcd41a0 rank 28 nRanks 32 nNodes 4 localRanks 8 localRank 4 MNNVL 0 +n111-046-125:1658:1721 [0] NCCL INFO Tree 1 : 25 -> 24 -> 31/-1/-1 +n111-046-125:1658:1721 [0] NCCL INFO Tree 9 : 25 -> 24 -> 31/-1/-1 +n111-046-125:1664:1714 [6] NCCL INFO comm 0xc0ec280 rank 30 nRanks 32 nNodes 4 localRanks 8 localRank 6 MNNVL 0 +n111-046-125:1665:1718 [7] NCCL INFO comm 0xbfcbe50 rank 31 nRanks 32 nNodes 4 localRanks 8 localRank 7 MNNVL 0 +n111-046-125:1659:1715 [1] NCCL INFO Tree 0 : 24 -> 25 -> 26/-1/-1 +n111-046-125:1663:1720 [5] NCCL INFO comm 0xceea760 rank 29 nRanks 32 nNodes 4 localRanks 8 localRank 5 MNNVL 0 +n111-046-125:1659:1715 [1] NCCL INFO Tree 8 : 24 -> 25 -> 26/-1/-1 +n111-046-125:1658:1721 [0] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n111-046-125:1659:1715 [1] NCCL INFO Tree 1 : 17 -> 25 -> 24/-1/-1 +n111-046-125:1659:1715 [1] NCCL INFO Tree 9 : -1 -> 25 -> 24/9/-1 +n111-046-125:1659:1715 [1] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n111-046-125:1661:1716 [3] NCCL INFO Tree 2 : 26 -> 27 -> 28/-1/-1 +n111-046-125:1661:1716 [3] NCCL INFO Tree 10 : 26 -> 27 -> 28/-1/-1 +n111-046-125:1662:1719 [4] NCCL INFO Tree 4 : 20 -> 28 -> 29/-1/-1 +n111-046-125:1661:1716 [3] NCCL INFO Tree 3 : 19 -> 27 -> 26/-1/-1 +n111-046-125:1662:1719 [4] NCCL INFO Tree 12 : -1 -> 28 -> 29/12/-1 +n111-046-125:1658:1721 [0] NCCL INFO Ring 00 : 17 -> 24 -> 31 +n111-046-125:1661:1716 [3] NCCL INFO Tree 11 : -1 -> 27 -> 26/11/-1 +n111-046-125:1662:1719 [4] NCCL INFO Tree 5 : 29 -> 28 -> 27/-1/-1 +n111-046-125:1660:1717 [2] NCCL INFO Tree 2 : 18 -> 26 -> 27/-1/-1 +n111-046-125:1658:1721 [0] NCCL INFO Ring 01 : 31 -> 24 -> 1 +n111-046-125:1662:1719 [4] NCCL INFO Tree 13 : 29 -> 28 -> 27/-1/-1 +n111-046-125:1660:1717 [2] NCCL INFO Tree 10 : -1 -> 26 -> 27/10/-1 +n111-046-125:1658:1721 [0] NCCL INFO Ring 02 : 25 -> 24 -> 31 +n111-046-125:1660:1717 [2] NCCL INFO Tree 3 : 27 -> 26 -> 25/-1/-1 +n111-046-125:1661:1716 [3] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n111-046-125:1660:1717 [2] NCCL INFO Tree 11 : 27 -> 26 -> 25/-1/-1 +n111-046-125:1658:1721 [0] NCCL INFO Ring 03 : 31 -> 24 -> 25 +n111-046-125:1663:1720 [5] NCCL INFO Tree 4 : 28 -> 29 -> 30/-1/-1 +n111-046-125:1665:1718 [7] NCCL INFO Tree 6 : 30 -> 31 -> 24/-1/-1 +n111-046-125:1664:1714 [6] NCCL INFO Tree 6 : 22 -> 30 -> 31/-1/-1 +n111-046-125:1662:1719 [4] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n111-046-125:1658:1721 [0] NCCL INFO Ring 04 : 25 -> 24 -> 31 +n111-046-125:1663:1720 [5] NCCL INFO Tree 12 : 28 -> 29 -> 30/-1/-1 +n111-046-125:1665:1718 [7] NCCL INFO Tree 14 : 30 -> 31 -> 24/-1/-1 +n111-046-125:1664:1714 [6] NCCL INFO Tree 14 : -1 -> 30 -> 31/14/-1 +n111-046-125:1658:1721 [0] NCCL INFO Ring 05 : 31 -> 24 -> 25 +n111-046-125:1659:1715 [1] NCCL INFO Ring 00 : 26 -> 25 -> 0 +n111-046-125:1660:1717 [2] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n111-046-125:1658:1721 [0] NCCL INFO Ring 06 : 25 -> 24 -> 31 +n111-046-125:1658:1721 [0] NCCL INFO Ring 07 : 31 -> 24 -> 25 +n111-046-125:1663:1720 [5] NCCL INFO Tree 5 : 21 -> 29 -> 28/-1/-1 +n111-046-125:1664:1714 [6] NCCL INFO Tree 7 : 31 -> 30 -> 29/-1/-1 +n111-046-125:1658:1721 [0] NCCL INFO Ring 08 : 17 -> 24 -> 31 +n111-046-125:1663:1720 [5] NCCL INFO Tree 13 : -1 -> 29 -> 28/13/-1 +n111-046-125:1661:1716 [3] NCCL INFO Ring 00 : 28 -> 27 -> 26 +n111-046-125:1658:1721 [0] NCCL INFO Ring 09 : 31 -> 24 -> 1 +n111-046-125:1665:1718 [7] NCCL INFO Tree 7 : 23 -> 31 -> 30/-1/-1 +n111-046-125:1659:1715 [1] NCCL INFO Ring 01 : 16 -> 25 -> 26 +n111-046-125:1661:1716 [3] NCCL INFO Ring 01 : 26 -> 27 -> 28 +n111-046-125:1665:1718 [7] NCCL INFO Tree 15 : -1 -> 31 -> 30/15/-1 +n111-046-125:1664:1714 [6] NCCL INFO Tree 15 : 31 -> 30 -> 29/-1/-1 +n111-046-125:1658:1721 [0] NCCL INFO Ring 10 : 25 -> 24 -> 31 +n111-046-125:1659:1715 [1] NCCL INFO Ring 02 : 26 -> 25 -> 24 +n111-046-125:1662:1719 [4] NCCL INFO Ring 00 : 29 -> 28 -> 27 +n111-046-125:1661:1716 [3] NCCL INFO Ring 02 : 28 -> 27 -> 2 +n111-046-125:1663:1720 [5] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n111-046-125:1658:1721 [0] NCCL INFO Ring 11 : 31 -> 24 -> 25 +n111-046-125:1659:1715 [1] NCCL INFO Ring 03 : 24 -> 25 -> 26 +n111-046-125:1665:1718 [7] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n111-046-125:1662:1719 [4] NCCL INFO Ring 01 : 27 -> 28 -> 29 +n111-046-125:1664:1714 [6] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n111-046-125:1661:1716 [3] NCCL INFO Ring 03 : 18 -> 27 -> 28 +n111-046-125:1660:1717 [2] NCCL INFO Ring 00 : 27 -> 26 -> 25 +n111-046-125:1659:1715 [1] NCCL INFO Ring 04 : 26 -> 25 -> 24 +n111-046-125:1658:1721 [0] NCCL INFO Ring 12 : 25 -> 24 -> 31 +n111-046-125:1662:1719 [4] NCCL INFO Ring 02 : 29 -> 28 -> 27 +n111-046-125:1661:1716 [3] NCCL INFO Ring 04 : 28 -> 27 -> 26 +n111-046-125:1658:1721 [0] NCCL INFO Ring 13 : 31 -> 24 -> 25 +n111-046-125:1662:1719 [4] NCCL INFO Ring 03 : 27 -> 28 -> 29 +n111-046-125:1660:1717 [2] NCCL INFO Ring 01 : 25 -> 26 -> 27 +n111-046-125:1659:1715 [1] NCCL INFO Ring 05 : 24 -> 25 -> 26 +n111-046-125:1658:1721 [0] NCCL INFO Ring 14 : 25 -> 24 -> 31 +n111-046-125:1661:1716 [3] NCCL INFO Ring 05 : 26 -> 27 -> 28 +n111-046-125:1659:1715 [1] NCCL INFO Ring 06 : 26 -> 25 -> 24 +n111-046-125:1659:1715 [1] NCCL INFO Ring 07 : 24 -> 25 -> 26 +n111-046-125:1662:1719 [4] NCCL INFO Ring 04 : 21 -> 28 -> 27 +n111-046-125:1660:1717 [2] NCCL INFO Ring 02 : 19 -> 26 -> 25 +n111-046-125:1658:1721 [0] NCCL INFO Ring 15 : 31 -> 24 -> 25 +n111-046-125:1661:1716 [3] NCCL INFO Ring 06 : 28 -> 27 -> 26 +n111-046-125:1663:1720 [5] NCCL INFO Ring 00 : 30 -> 29 -> 28 +n111-046-125:1659:1715 [1] NCCL INFO Ring 08 : 26 -> 25 -> 0 +n111-046-125:1662:1719 [4] NCCL INFO Ring 05 : 27 -> 28 -> 5 +n111-046-125:1660:1717 [2] NCCL INFO Ring 03 : 25 -> 26 -> 3 +n111-046-125:1658:1721 [0] NCCL INFO Trees [0] 25/-1/-1->24->16 [1] 31/-1/-1->24->25 [2] 25/-1/-1->24->31 [3] 31/-1/-1->24->25 [4] 25/-1/-1->24->31 [5] 31/-1/-1->24->25 [6] 25/-1/-1->24->31 [7] -1/-1/-1->24->25 [8] 25/8/-1->24->-1 [9] 31/-1/-1->24->25 [10] 25/-1/-1->24->31 [11] 31/-1/-1->24->25 [12] 25/-1/-1->24->31 [13] 31/-1/-1->24->25 [14] 25/-1/-1->24->31 [15] -1/-1/-1->24->25 +n111-046-125:1661:1716 [3] NCCL INFO Ring 07 : 26 -> 27 -> 28 +n111-046-125:1660:1717 [2] NCCL INFO Ring 04 : 27 -> 26 -> 25 +n111-046-125:1664:1714 [6] NCCL INFO Ring 00 : 31 -> 30 -> 29 +n111-046-125:1663:1720 [5] NCCL INFO Ring 01 : 28 -> 29 -> 30 +n111-046-125:1659:1715 [1] NCCL INFO Ring 09 : 16 -> 25 -> 26 +n111-046-125:1662:1719 [4] NCCL INFO Ring 06 : 29 -> 28 -> 27 +n111-046-125:1660:1717 [2] NCCL INFO Ring 05 : 25 -> 26 -> 27 +n111-046-125:1658:1721 [0] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n111-046-125:1664:1714 [6] NCCL INFO Ring 01 : 29 -> 30 -> 31 +n111-046-125:1661:1716 [3] NCCL INFO Ring 08 : 28 -> 27 -> 26 +n111-046-125:1659:1715 [1] NCCL INFO Ring 10 : 26 -> 25 -> 24 +n111-046-125:1663:1720 [5] NCCL INFO Ring 02 : 30 -> 29 -> 28 +n111-046-125:1662:1719 [4] NCCL INFO Ring 07 : 27 -> 28 -> 29 +n111-046-125:1664:1714 [6] NCCL INFO Ring 02 : 31 -> 30 -> 29 +n111-046-125:1660:1717 [2] NCCL INFO Ring 06 : 27 -> 26 -> 25 +n111-046-125:1663:1720 [5] NCCL INFO Ring 03 : 28 -> 29 -> 30 +n111-046-125:1658:1721 [0] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n111-046-125:1661:1716 [3] NCCL INFO Ring 09 : 26 -> 27 -> 28 +n111-046-125:1659:1715 [1] NCCL INFO Ring 11 : 24 -> 25 -> 26 +n111-046-125:1665:1718 [7] NCCL INFO Ring 00 : 24 -> 31 -> 30 +n111-046-125:1662:1719 [4] NCCL INFO Ring 08 : 29 -> 28 -> 27 +n111-046-125:1664:1714 [6] NCCL INFO Ring 03 : 29 -> 30 -> 31 +n111-046-125:1665:1718 [7] NCCL INFO Ring 01 : 30 -> 31 -> 24 +n111-046-125:1662:1719 [4] NCCL INFO Ring 09 : 27 -> 28 -> 29 +n111-046-125:1660:1717 [2] NCCL INFO Ring 07 : 25 -> 26 -> 27 +n111-046-125:1658:1721 [0] NCCL INFO P2P Chunksize set to 524288 +n111-046-125:1664:1714 [6] NCCL INFO Ring 04 : 31 -> 30 -> 29 +n111-046-125:1663:1720 [5] NCCL INFO Ring 04 : 30 -> 29 -> 4 +n111-046-125:1664:1714 [6] NCCL INFO Ring 05 : 29 -> 30 -> 31 +n111-046-125:1661:1716 [3] NCCL INFO Ring 10 : 28 -> 27 -> 2 +n111-046-125:1659:1715 [1] NCCL INFO Ring 12 : 26 -> 25 -> 24 +n111-046-125:1664:1714 [6] NCCL INFO Ring 06 : 23 -> 30 -> 29 +n111-046-125:1662:1719 [4] NCCL INFO Ring 10 : 29 -> 28 -> 27 +n111-046-125:1665:1718 [7] NCCL INFO Ring 02 : 24 -> 31 -> 30 +n111-046-125:1660:1717 [2] NCCL INFO Ring 08 : 27 -> 26 -> 25 +n111-046-125:1663:1720 [5] NCCL INFO Ring 05 : 20 -> 29 -> 30 +n111-046-125:1661:1716 [3] NCCL INFO Ring 11 : 18 -> 27 -> 28 +n111-046-125:1659:1715 [1] NCCL INFO Ring 13 : 24 -> 25 -> 26 +n111-046-125:1664:1714 [6] NCCL INFO Ring 07 : 29 -> 30 -> 7 +n111-046-125:1662:1719 [4] NCCL INFO Ring 11 : 27 -> 28 -> 29 +n111-046-125:1665:1718 [7] NCCL INFO Ring 03 : 30 -> 31 -> 24 +n111-046-125:1660:1717 [2] NCCL INFO Ring 09 : 25 -> 26 -> 27 +n111-046-125:1660:1717 [2] NCCL INFO Ring 10 : 19 -> 26 -> 25 +n111-046-125:1661:1716 [3] NCCL INFO Ring 12 : 28 -> 27 -> 26 +n111-046-125:1663:1720 [5] NCCL INFO Ring 06 : 30 -> 29 -> 28 +n111-046-125:1659:1715 [1] NCCL INFO Ring 14 : 26 -> 25 -> 24 +n111-046-125:1662:1719 [4] NCCL INFO Ring 12 : 21 -> 28 -> 27 +n111-046-125:1664:1714 [6] NCCL INFO Ring 08 : 31 -> 30 -> 29 +n111-046-125:1665:1718 [7] NCCL INFO Ring 04 : 24 -> 31 -> 30 +n111-046-125:1660:1717 [2] NCCL INFO Ring 11 : 25 -> 26 -> 3 +n111-046-125:1661:1716 [3] NCCL INFO Ring 13 : 26 -> 27 -> 28 +n111-046-125:1659:1715 [1] NCCL INFO Ring 15 : 24 -> 25 -> 26 +n111-046-125:1661:1716 [3] NCCL INFO Ring 14 : 28 -> 27 -> 26 +n111-046-125:1663:1720 [5] NCCL INFO Ring 07 : 28 -> 29 -> 30 +n111-046-125:1662:1719 [4] NCCL INFO Ring 13 : 27 -> 28 -> 5 +n111-046-125:1664:1714 [6] NCCL INFO Ring 09 : 29 -> 30 -> 31 +n111-046-125:1665:1718 [7] NCCL INFO Ring 05 : 30 -> 31 -> 24 +n111-046-125:1660:1717 [2] NCCL INFO Ring 12 : 27 -> 26 -> 25 +n111-046-125:1659:1715 [1] NCCL INFO Trees [0] 26/-1/-1->25->24 [1] 24/-1/-1->25->17 [2] -1/-1/-1->25->24 [3] 24/-1/-1->25->26 [4] 26/-1/-1->25->24 [5] 24/-1/-1->25->26 [6] 26/-1/-1->25->24 [7] 24/-1/-1->25->26 [8] 26/-1/-1->25->24 [9] 24/9/-1->25->-1 [10] -1/-1/-1->25->24 [11] 24/-1/-1->25->26 [12] 26/-1/-1->25->24 [13] 24/-1/-1->25->26 [14] 26/-1/-1->25->24 [15] 24/-1/-1->25->26 +n111-046-125:1661:1716 [3] NCCL INFO Ring 15 : 26 -> 27 -> 28 +n111-046-125:1661:1716 [3] NCCL INFO Trees [0] 28/-1/-1->27->26 [1] 26/-1/-1->27->28 [2] 28/-1/-1->27->26 [3] 26/-1/-1->27->19 [4] -1/-1/-1->27->26 [5] 26/-1/-1->27->28 [6] 28/-1/-1->27->26 [7] 26/-1/-1->27->28 [8] 28/-1/-1->27->26 [9] 26/-1/-1->27->28 [10] 28/-1/-1->27->26 [11] 26/11/-1->27->-1 [12] -1/-1/-1->27->26 [13] 26/-1/-1->27->28 [14] 28/-1/-1->27->26 [15] 26/-1/-1->27->28 +n111-046-125:1659:1715 [1] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n111-046-125:1663:1720 [5] NCCL INFO Ring 08 : 30 -> 29 -> 28 +n111-046-125:1662:1719 [4] NCCL INFO Ring 14 : 29 -> 28 -> 27 +n111-046-125:1659:1715 [1] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n111-046-125:1663:1720 [5] NCCL INFO Ring 09 : 28 -> 29 -> 30 +n111-046-125:1665:1718 [7] NCCL INFO Ring 06 : 24 -> 31 -> 6 +n111-046-125:1664:1714 [6] NCCL INFO Ring 10 : 31 -> 30 -> 29 +n111-046-125:1660:1717 [2] NCCL INFO Ring 13 : 25 -> 26 -> 27 +n111-046-125:1661:1716 [3] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n111-046-125:1665:1718 [7] NCCL INFO Ring 07 : 22 -> 31 -> 24 +n111-046-125:1660:1717 [2] NCCL INFO Ring 14 : 27 -> 26 -> 25 +n111-046-125:1662:1719 [4] NCCL INFO Ring 15 : 27 -> 28 -> 29 +n111-046-125:1659:1715 [1] NCCL INFO P2P Chunksize set to 524288 +n111-046-125:1663:1720 [5] NCCL INFO Ring 10 : 30 -> 29 -> 28 +n111-046-125:1664:1714 [6] NCCL INFO Ring 11 : 29 -> 30 -> 31 +n111-046-125:1661:1716 [3] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n111-046-125:1664:1714 [6] NCCL INFO Ring 12 : 31 -> 30 -> 29 +n111-046-125:1660:1717 [2] NCCL INFO Ring 15 : 25 -> 26 -> 27 +n111-046-125:1661:1716 [3] NCCL INFO P2P Chunksize set to 524288 +n111-046-125:1662:1719 [4] NCCL INFO Trees [0] 29/-1/-1->28->27 [1] 27/-1/-1->28->29 [2] 29/-1/-1->28->27 [3] -1/-1/-1->28->29 [4] 29/-1/-1->28->20 [5] 27/-1/-1->28->29 [6] 29/-1/-1->28->27 [7] 27/-1/-1->28->29 [8] 29/-1/-1->28->27 [9] 27/-1/-1->28->29 [10] 29/-1/-1->28->27 [11] -1/-1/-1->28->29 [12] 29/12/-1->28->-1 [13] 27/-1/-1->28->29 [14] 29/-1/-1->28->27 [15] 27/-1/-1->28->29 +n111-046-125:1664:1714 [6] NCCL INFO Ring 13 : 29 -> 30 -> 31 +n111-046-125:1663:1720 [5] NCCL INFO Ring 11 : 28 -> 29 -> 30 +n111-046-125:1665:1718 [7] NCCL INFO Ring 08 : 24 -> 31 -> 30 +n111-046-125:1660:1717 [2] NCCL INFO Trees [0] 27/-1/-1->26->25 [1] -1/-1/-1->26->27 [2] 27/-1/-1->26->18 [3] 25/-1/-1->26->27 [4] 27/-1/-1->26->25 [5] 25/-1/-1->26->27 [6] 27/-1/-1->26->25 [7] 25/-1/-1->26->27 [8] 27/-1/-1->26->25 [9] -1/-1/-1->26->27 [10] 27/10/-1->26->-1 [11] 25/-1/-1->26->27 [12] 27/-1/-1->26->25 [13] 25/-1/-1->26->27 [14] 27/-1/-1->26->25 [15] 25/-1/-1->26->27 +n111-046-125:1664:1714 [6] NCCL INFO Ring 14 : 23 -> 30 -> 29 +n111-046-125:1662:1719 [4] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n111-046-125:1663:1720 [5] NCCL INFO Ring 12 : 30 -> 29 -> 4 +n111-046-125:1665:1718 [7] NCCL INFO Ring 09 : 30 -> 31 -> 24 +n111-046-125:1665:1718 [7] NCCL INFO Ring 10 : 24 -> 31 -> 30 +n111-046-125:1664:1714 [6] NCCL INFO Ring 15 : 29 -> 30 -> 7 +n111-046-125:1660:1717 [2] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n111-046-125:1662:1719 [4] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n111-046-125:1664:1714 [6] NCCL INFO Trees [0] 31/-1/-1->30->29 [1] 29/-1/-1->30->31 [2] 31/-1/-1->30->29 [3] 29/-1/-1->30->31 [4] 31/-1/-1->30->29 [5] -1/-1/-1->30->31 [6] 31/-1/-1->30->22 [7] 29/-1/-1->30->31 [8] 31/-1/-1->30->29 [9] 29/-1/-1->30->31 [10] 31/-1/-1->30->29 [11] 29/-1/-1->30->31 [12] 31/-1/-1->30->29 [13] -1/-1/-1->30->31 [14] 31/14/-1->30->-1 [15] 29/-1/-1->30->31 +n111-046-125:1660:1717 [2] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n111-046-125:1663:1720 [5] NCCL INFO Ring 13 : 20 -> 29 -> 30 +n111-046-125:1665:1718 [7] NCCL INFO Ring 11 : 30 -> 31 -> 24 +n111-046-125:1662:1719 [4] NCCL INFO P2P Chunksize set to 524288 +n111-046-125:1660:1717 [2] NCCL INFO P2P Chunksize set to 524288 +n111-046-125:1664:1714 [6] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n111-046-125:1663:1720 [5] NCCL INFO Ring 14 : 30 -> 29 -> 28 +n111-046-125:1664:1714 [6] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n111-046-125:1665:1718 [7] NCCL INFO Ring 12 : 24 -> 31 -> 30 +n111-046-125:1663:1720 [5] NCCL INFO Ring 15 : 28 -> 29 -> 30 +n111-046-125:1664:1714 [6] NCCL INFO P2P Chunksize set to 524288 +n111-046-125:1665:1718 [7] NCCL INFO Ring 13 : 30 -> 31 -> 24 +n111-046-125:1663:1720 [5] NCCL INFO Trees [0] 30/-1/-1->29->28 [1] 28/-1/-1->29->30 [2] 30/-1/-1->29->28 [3] 28/-1/-1->29->30 [4] 30/-1/-1->29->28 [5] 28/-1/-1->29->21 [6] -1/-1/-1->29->28 [7] 28/-1/-1->29->30 [8] 30/-1/-1->29->28 [9] 28/-1/-1->29->30 [10] 30/-1/-1->29->28 [11] 28/-1/-1->29->30 [12] 30/-1/-1->29->28 [13] 28/13/-1->29->-1 [14] -1/-1/-1->29->28 [15] 28/-1/-1->29->30 +n111-046-125:1665:1718 [7] NCCL INFO Ring 14 : 24 -> 31 -> 6 +n111-046-125:1663:1720 [5] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n111-046-125:1663:1720 [5] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n111-046-125:1665:1718 [7] NCCL INFO Ring 15 : 22 -> 31 -> 24 +n111-046-125:1663:1720 [5] NCCL INFO P2P Chunksize set to 524288 +n111-046-125:1665:1718 [7] NCCL INFO Trees [0] -1/-1/-1->31->30 [1] 30/-1/-1->31->24 [2] 24/-1/-1->31->30 [3] 30/-1/-1->31->24 [4] 24/-1/-1->31->30 [5] 30/-1/-1->31->24 [6] 24/-1/-1->31->30 [7] 30/-1/-1->31->23 [8] -1/-1/-1->31->30 [9] 30/-1/-1->31->24 [10] 24/-1/-1->31->30 [11] 30/-1/-1->31->24 [12] 24/-1/-1->31->30 [13] 30/-1/-1->31->24 [14] 24/-1/-1->31->30 [15] 30/15/-1->31->-1 +n111-046-125:1665:1718 [7] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n111-046-125:1665:1718 [7] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n111-046-125:1665:1718 [7] NCCL INFO P2P Chunksize set to 524288 +n111-046-125:1658:1754 [0] NCCL INFO [Proxy Service] Device 0 CPU core 15 +n111-046-125:1658:1756 [0] NCCL INFO [Proxy Service UDS] Device 0 CPU core 18 +n111-046-125:1661:1757 [3] NCCL INFO [Proxy Service] Device 3 CPU core 25 +n111-046-125:1659:1755 [1] NCCL INFO [Proxy Service] Device 1 CPU core 122 +n111-046-125:1662:1758 [4] NCCL INFO [Proxy Service] Device 4 CPU core 58 +n111-046-125:1662:1762 [4] NCCL INFO [Proxy Service UDS] Device 4 CPU core 64 +n111-046-125:1664:1765 [6] NCCL INFO [Proxy Service] Device 6 CPU core 75 +n111-046-125:1660:1761 [2] NCCL INFO [Proxy Service] Device 2 CPU core 5 +n111-046-125:1660:1764 [2] NCCL INFO [Proxy Service UDS] Device 2 CPU core 7 +n111-046-125:1661:1759 [3] NCCL INFO [Proxy Service UDS] Device 3 CPU core 31 +n111-046-125:1664:1768 [6] NCCL INFO [Proxy Service UDS] Device 6 CPU core 185 +n111-046-125:1659:1760 [1] NCCL INFO [Proxy Service UDS] Device 1 CPU core 19 +n111-046-125:1663:1767 [5] NCCL INFO [Proxy Service UDS] Device 5 CPU core 61 +n111-046-125:1663:1763 [5] NCCL INFO [Proxy Service] Device 5 CPU core 175 +n111-046-125:1665:1766 [7] NCCL INFO [Proxy Service] Device 7 CPU core 181 +n111-046-125:1665:1769 [7] NCCL INFO [Proxy Service UDS] Device 7 CPU core 82 +n111-046-125:1660:1717 [2] NCCL INFO NCCL_PROTO set by environment to Simple +n111-046-125:1660:1717 [2] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n111-046-125:1660:1717 [2] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512 +n111-046-125:1660:1717 [2] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 2 p2p channels per peer +n111-046-125:1658:1721 [0] NCCL INFO NCCL_PROTO set by environment to Simple +n111-046-125:1658:1721 [0] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n111-046-125:1658:1721 [0] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512 +n111-046-125:1658:1721 [0] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 2 p2p channels per peer +n111-046-125:1661:1716 [3] NCCL INFO NCCL_PROTO set by environment to Simple +n111-046-125:1661:1716 [3] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n111-046-125:1661:1716 [3] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512 +n111-046-125:1661:1716 [3] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 2 p2p channels per peer +n111-046-125:1665:1718 [7] NCCL INFO NCCL_PROTO set by environment to Simple +n111-046-125:1665:1718 [7] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n111-046-125:1665:1718 [7] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512 +n111-046-125:1665:1718 [7] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 2 p2p channels per peer +n111-046-125:1662:1719 [4] NCCL INFO NCCL_PROTO set by environment to Simple +n111-046-125:1662:1719 [4] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n111-046-125:1662:1719 [4] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512 +n111-046-125:1662:1719 [4] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 2 p2p channels per peer +n111-046-125:1664:1714 [6] NCCL INFO NCCL_PROTO set by environment to Simple +n111-046-125:1664:1714 [6] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n111-046-125:1664:1714 [6] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512 +n111-046-125:1664:1714 [6] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 2 p2p channels per peer +n111-046-125:1659:1715 [1] NCCL INFO NCCL_PROTO set by environment to Simple +n111-046-125:1659:1715 [1] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n111-046-125:1659:1715 [1] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512 +n111-046-125:1659:1715 [1] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 2 p2p channels per peer +n111-046-125:1663:1720 [5] NCCL INFO NCCL_PROTO set by environment to Simple +n111-046-125:1663:1720 [5] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n111-046-125:1663:1720 [5] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512 +n111-046-125:1663:1720 [5] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 2 p2p channels per peer +n111-046-125:1658:1721 [0] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n111-046-125:1658:1721 [0] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n111-046-125:1658:1721 [0] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n111-046-125:1658:1721 [0] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n111-046-125:1658:1721 [0] NCCL INFO ncclCommInitRankConfig comm 0xc9c9a40 rank 24 nranks 32 cudaDev 0 nvmlDev 0 busId 4000 commId 0x2a4d35ea441cbcba - Init COMPLETE +n111-046-125:1658:1721 [0] NCCL INFO Init timings - ncclCommInitRankConfig: rank 24 nranks 32 total 6.67 (kernels 0.67, alloc 4.18, bootstrap 0.07, allgathers 0.06, topo 0.81, graphs 0.08, connections 0.06, rest 0.74) +n111-046-125:1662:1719 [4] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n111-046-125:1662:1719 [4] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n111-046-125:1662:1719 [4] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n111-046-125:1662:1719 [4] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n111-046-125:1662:1719 [4] NCCL INFO ncclCommInitRankConfig comm 0xbcd41a0 rank 28 nranks 32 cudaDev 4 nvmlDev 4 busId 84000 commId 0x2a4d35ea441cbcba - Init COMPLETE +n111-046-125:1662:1719 [4] NCCL INFO Init timings - ncclCommInitRankConfig: rank 28 nranks 32 total 6.67 (kernels 0.68, alloc 4.12, bootstrap 0.11, allgathers 0.06, topo 0.81, graphs 0.08, connections 0.09, rest 0.72) +n111-046-125:1659:1715 [1] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n111-046-125:1659:1715 [1] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n111-046-125:1659:1715 [1] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n111-046-125:1659:1715 [1] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n111-046-125:1659:1715 [1] NCCL INFO ncclCommInitRankConfig comm 0xbe02940 rank 25 nranks 32 cudaDev 1 nvmlDev 1 busId 5000 commId 0x2a4d35ea441cbcba - Init COMPLETE +n111-046-125:1659:1715 [1] NCCL INFO Init timings - ncclCommInitRankConfig: rank 25 nranks 32 total 6.72 (kernels 0.72, alloc 4.20, bootstrap 0.03, allgathers 0.06, topo 0.81, graphs 0.08, connections 0.16, rest 0.65) +n111-046-125:1660:1717 [2] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n111-046-125:1660:1717 [2] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n111-046-125:1660:1717 [2] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n111-046-125:1660:1717 [2] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n111-046-125:1660:1717 [2] NCCL INFO ncclCommInitRankConfig comm 0xb131d80 rank 26 nranks 32 cudaDev 2 nvmlDev 2 busId b000 commId 0x2a4d35ea441cbcba - Init COMPLETE +n111-046-125:1660:1717 [2] NCCL INFO Init timings - ncclCommInitRankConfig: rank 26 nranks 32 total 6.70 (kernels 0.69, alloc 4.22, bootstrap 0.01, allgathers 0.06, topo 0.81, graphs 0.08, connections 0.06, rest 0.76) +n111-046-125:1664:1714 [6] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n111-046-125:1664:1714 [6] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n111-046-125:1664:1714 [6] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n111-046-125:1664:1714 [6] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n111-046-125:1664:1714 [6] NCCL INFO ncclCommInitRankConfig comm 0xc0ec280 rank 30 nranks 32 cudaDev 6 nvmlDev 6 busId 8b000 commId 0x2a4d35ea441cbcba - Init COMPLETE +n111-046-125:1664:1714 [6] NCCL INFO Init timings - ncclCommInitRankConfig: rank 30 nranks 32 total 6.75 (kernels 0.73, alloc 4.20, bootstrap 0.03, allgathers 0.00, topo 0.81, graphs 0.14, connections 0.16, rest 0.67) +n111-046-125:1661:1716 [3] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n111-046-125:1661:1716 [3] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n111-046-125:1661:1716 [3] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n111-046-125:1661:1716 [3] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n111-046-125:1661:1716 [3] NCCL INFO ncclCommInitRankConfig comm 0xbffbfe0 rank 27 nranks 32 cudaDev 3 nvmlDev 3 busId c000 commId 0x2a4d35ea441cbcba - Init COMPLETE +n111-046-125:1661:1716 [3] NCCL INFO Init timings - ncclCommInitRankConfig: rank 27 nranks 32 total 6.74 (kernels 0.71, alloc 4.12, bootstrap 0.11, allgathers 0.06, topo 0.81, graphs 0.08, connections 0.09, rest 0.75) +n111-046-125:1665:1718 [7] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n111-046-125:1665:1718 [7] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n111-046-125:1665:1718 [7] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n111-046-125:1665:1718 [7] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n111-046-125:1665:1718 [7] NCCL INFO ncclCommInitRankConfig comm 0xbfcbe50 rank 31 nranks 32 cudaDev 7 nvmlDev 7 busId 8c000 commId 0x2a4d35ea441cbcba - Init COMPLETE +n111-046-125:1665:1718 [7] NCCL INFO Init timings - ncclCommInitRankConfig: rank 31 nranks 32 total 6.71 (kernels 0.68, alloc 4.21, bootstrap 0.02, allgathers 0.01, topo 0.81, graphs 0.13, connections 0.09, rest 0.75) +n111-046-125:1663:1720 [5] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n111-046-125:1663:1720 [5] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n111-046-125:1663:1720 [5] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n111-046-125:1663:1720 [5] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n111-046-125:1663:1720 [5] NCCL INFO ncclCommInitRankConfig comm 0xceea760 rank 29 nranks 32 cudaDev 5 nvmlDev 5 busId 85000 commId 0x2a4d35ea441cbcba - Init COMPLETE +n111-046-125:1663:1720 [5] NCCL INFO Init timings - ncclCommInitRankConfig: rank 29 nranks 32 total 6.71 (kernels 0.68, alloc 4.16, bootstrap 0.07, allgathers 0.02, topo 0.81, graphs 0.12, connections 0.16, rest 0.68) +n111-046-125:1658:1771 [0] NCCL INFO Channel 00/0 : 24[0] -> 25[1] via P2P/CUMEM +n111-046-125:1658:1771 [0] NCCL INFO Channel 01/0 : 24[0] -> 25[1] via P2P/CUMEM +n111-046-125:1658:1771 [0] NCCL INFO Channel 02/0 : 24[0] -> 25[1] via P2P/CUMEM +n111-046-125:1658:1771 [0] NCCL INFO Channel 03/0 : 24[0] -> 25[1] via P2P/CUMEM +n111-046-125:1658:1771 [0] NCCL INFO Channel 04/0 : 24[0] -> 25[1] via P2P/CUMEM +n111-046-125:1658:1771 [0] NCCL INFO Channel 05/0 : 24[0] -> 25[1] via P2P/CUMEM +n111-046-125:1658:1771 [0] NCCL INFO Channel 06/0 : 24[0] -> 25[1] via P2P/CUMEM +n111-046-125:1658:1771 [0] NCCL INFO Channel 07/0 : 24[0] -> 25[1] via P2P/CUMEM +n111-046-125:1658:1771 [0] NCCL INFO Channel 08/0 : 24[0] -> 25[1] via P2P/CUMEM +n111-046-125:1658:1771 [0] NCCL INFO Channel 09/0 : 24[0] -> 25[1] via P2P/CUMEM +n111-046-125:1658:1771 [0] NCCL INFO Channel 10/0 : 24[0] -> 25[1] via P2P/CUMEM +n111-046-125:1658:1771 [0] NCCL INFO Channel 11/0 : 24[0] -> 25[1] via P2P/CUMEM +n111-046-125:1658:1771 [0] NCCL INFO Channel 12/0 : 24[0] -> 25[1] via P2P/CUMEM +n111-046-125:1658:1771 [0] NCCL INFO Channel 13/0 : 24[0] -> 25[1] via P2P/CUMEM +n111-046-125:1658:1771 [0] NCCL INFO Channel 14/0 : 24[0] -> 25[1] via P2P/CUMEM +n111-046-125:1662:1770 [4] NCCL INFO Channel 00/0 : 28[4] -> 29[5] via P2P/CUMEM +n111-046-125:1658:1771 [0] NCCL INFO Channel 15/0 : 24[0] -> 25[1] via P2P/CUMEM +n111-046-125:1662:1770 [4] NCCL INFO Channel 01/0 : 28[4] -> 29[5] via P2P/CUMEM +n111-046-125:1662:1770 [4] NCCL INFO Channel 02/0 : 28[4] -> 29[5] via P2P/CUMEM +n111-046-125:1664:1775 [6] NCCL INFO Channel 00/0 : 30[6] -> 31[7] via P2P/CUMEM +n111-046-125:1659:1774 [1] NCCL INFO Channel 00/0 : 25[1] -> 26[2] via P2P/CUMEM +n111-046-125:1660:1773 [2] NCCL INFO Channel 00/0 : 26[2] -> 27[3] via P2P/CUMEM +n111-046-125:1662:1770 [4] NCCL INFO Channel 03/0 : 28[4] -> 29[5] via P2P/CUMEM +n111-046-125:1664:1775 [6] NCCL INFO Channel 01/0 : 30[6] -> 31[7] via P2P/CUMEM +n111-046-125:1659:1774 [1] NCCL INFO Channel 03/0 : 25[1] -> 26[2] via P2P/CUMEM +n111-046-125:1660:1773 [2] NCCL INFO Channel 01/0 : 26[2] -> 27[3] via P2P/CUMEM +n111-046-125:1662:1770 [4] NCCL INFO Channel 04/0 : 28[4] -> 29[5] via P2P/CUMEM +n111-046-125:1664:1775 [6] NCCL INFO Channel 02/0 : 30[6] -> 31[7] via P2P/CUMEM +n111-046-125:1659:1774 [1] NCCL INFO Channel 04/0 : 25[1] -> 26[2] via P2P/CUMEM +n111-046-125:1660:1773 [2] NCCL INFO Channel 02/0 : 26[2] -> 27[3] via P2P/CUMEM +n111-046-125:1662:1770 [4] NCCL INFO Channel 05/0 : 28[4] -> 29[5] via P2P/CUMEM +n111-046-125:1664:1775 [6] NCCL INFO Channel 03/0 : 30[6] -> 31[7] via P2P/CUMEM +n111-046-125:1659:1774 [1] NCCL INFO Channel 05/0 : 25[1] -> 26[2] via P2P/CUMEM +n111-046-125:1660:1773 [2] NCCL INFO Channel 03/0 : 26[2] -> 27[3] via P2P/CUMEM +n111-046-125:1662:1770 [4] NCCL INFO Channel 06/0 : 28[4] -> 29[5] via P2P/CUMEM +n111-046-125:1664:1775 [6] NCCL INFO Channel 04/0 : 30[6] -> 31[7] via P2P/CUMEM +n111-046-125:1659:1774 [1] NCCL INFO Channel 06/0 : 25[1] -> 26[2] via P2P/CUMEM +n111-046-125:1660:1773 [2] NCCL INFO Channel 04/0 : 26[2] -> 27[3] via P2P/CUMEM +n111-046-125:1662:1770 [4] NCCL INFO Channel 07/0 : 28[4] -> 29[5] via P2P/CUMEM +n111-046-125:1664:1775 [6] NCCL INFO Channel 05/0 : 30[6] -> 31[7] via P2P/CUMEM +n111-046-125:1659:1774 [1] NCCL INFO Channel 07/0 : 25[1] -> 26[2] via P2P/CUMEM +n111-046-125:1660:1773 [2] NCCL INFO Channel 05/0 : 26[2] -> 27[3] via P2P/CUMEM +n111-046-125:1662:1770 [4] NCCL INFO Channel 08/0 : 28[4] -> 29[5] via P2P/CUMEM +n111-046-125:1664:1775 [6] NCCL INFO Channel 06/0 : 30[6] -> 31[7] via P2P/CUMEM +n111-046-125:1659:1774 [1] NCCL INFO Channel 08/0 : 25[1] -> 26[2] via P2P/CUMEM +n111-046-125:1664:1775 [6] NCCL INFO Channel 07/0 : 30[6] -> 31[7] via P2P/CUMEM +n111-046-125:1660:1773 [2] NCCL INFO Channel 06/0 : 26[2] -> 27[3] via P2P/CUMEM +n111-046-125:1659:1774 [1] NCCL INFO Channel 11/0 : 25[1] -> 26[2] via P2P/CUMEM +n111-046-125:1664:1775 [6] NCCL INFO Channel 08/0 : 30[6] -> 31[7] via P2P/CUMEM +n111-046-125:1663:1776 [5] NCCL INFO Channel 00/0 : 29[5] -> 30[6] via P2P/CUMEM +n111-046-125:1660:1773 [2] NCCL INFO Channel 07/0 : 26[2] -> 27[3] via P2P/CUMEM +n111-046-125:1659:1774 [1] NCCL INFO Channel 12/0 : 25[1] -> 26[2] via P2P/CUMEM +n111-046-125:1664:1775 [6] NCCL INFO Channel 09/0 : 30[6] -> 31[7] via P2P/CUMEM +n111-046-125:1663:1776 [5] NCCL INFO Channel 01/0 : 29[5] -> 30[6] via P2P/CUMEM +n111-046-125:1660:1773 [2] NCCL INFO Channel 08/0 : 26[2] -> 27[3] via P2P/CUMEM +n111-046-125:1659:1774 [1] NCCL INFO Channel 13/0 : 25[1] -> 26[2] via P2P/CUMEM +n111-046-125:1664:1775 [6] NCCL INFO Channel 10/0 : 30[6] -> 31[7] via P2P/CUMEM +n111-046-125:1663:1776 [5] NCCL INFO Channel 02/0 : 29[5] -> 30[6] via P2P/CUMEM +n111-046-125:1660:1773 [2] NCCL INFO Channel 09/0 : 26[2] -> 27[3] via P2P/CUMEM +n111-046-125:1659:1774 [1] NCCL INFO Channel 14/0 : 25[1] -> 26[2] via P2P/CUMEM +n111-046-125:1664:1775 [6] NCCL INFO Channel 11/0 : 30[6] -> 31[7] via P2P/CUMEM +n111-046-125:1663:1776 [5] NCCL INFO Channel 03/0 : 29[5] -> 30[6] via P2P/CUMEM +n111-046-125:1660:1773 [2] NCCL INFO Channel 10/0 : 26[2] -> 27[3] via P2P/CUMEM +n111-046-125:1659:1774 [1] NCCL INFO Channel 15/0 : 25[1] -> 26[2] via P2P/CUMEM +n111-046-125:1664:1775 [6] NCCL INFO Channel 12/0 : 30[6] -> 31[7] via P2P/CUMEM +n111-046-125:1663:1776 [5] NCCL INFO Channel 04/0 : 29[5] -> 30[6] via P2P/CUMEM +n111-046-125:1660:1773 [2] NCCL INFO Channel 11/0 : 26[2] -> 27[3] via P2P/CUMEM +n111-046-125:1658:1771 [0] NCCL INFO Channel 01/0 : 24[0] -> 31[7] via P2P/CUMEM +n111-046-125:1664:1775 [6] NCCL INFO Channel 13/0 : 30[6] -> 31[7] via P2P/CUMEM +n111-046-125:1663:1776 [5] NCCL INFO Channel 07/0 : 29[5] -> 30[6] via P2P/CUMEM +n111-046-125:1662:1770 [4] NCCL INFO Channel 09/0 : 28[4] -> 29[5] via P2P/CUMEM +n111-046-125:1664:1775 [6] NCCL INFO Channel 14/0 : 30[6] -> 31[7] via P2P/CUMEM +n111-046-125:1660:1773 [2] NCCL INFO Channel 12/0 : 26[2] -> 27[3] via P2P/CUMEM +n111-046-125:1663:1776 [5] NCCL INFO Channel 08/0 : 29[5] -> 30[6] via P2P/CUMEM +n111-046-125:1662:1770 [4] NCCL INFO Channel 10/0 : 28[4] -> 29[5] via P2P/CUMEM +n111-046-125:1663:1776 [5] NCCL INFO Channel 09/0 : 29[5] -> 30[6] via P2P/CUMEM +n111-046-125:1660:1773 [2] NCCL INFO Channel 13/0 : 26[2] -> 27[3] via P2P/CUMEM +n111-046-125:1661:1777 [3] NCCL INFO Channel 00/0 : 27[3] -> 28[4] via P2P/CUMEM +n111-046-125:1658:1771 [0] NCCL INFO Channel 02/0 : 24[0] -> 31[7] via P2P/CUMEM +n111-046-125:1663:1776 [5] NCCL INFO Channel 10/0 : 29[5] -> 30[6] via P2P/CUMEM +n111-046-125:1662:1770 [4] NCCL INFO Channel 11/0 : 28[4] -> 29[5] via P2P/CUMEM +n111-046-125:1660:1773 [2] NCCL INFO Channel 14/0 : 26[2] -> 27[3] via P2P/CUMEM +n111-046-125:1661:1777 [3] NCCL INFO Channel 01/0 : 27[3] -> 28[4] via P2P/CUMEM +n111-046-125:1658:1771 [0] NCCL INFO Channel 03/0 : 24[0] -> 31[7] via P2P/CUMEM +n111-046-125:1664:1775 [6] NCCL INFO Channel 15/0 : 30[6] -> 31[7] via P2P/CUMEM +n111-046-125:1660:1773 [2] NCCL INFO Channel 15/0 : 26[2] -> 27[3] via P2P/CUMEM +n111-046-125:1661:1777 [3] NCCL INFO Channel 02/0 : 27[3] -> 28[4] via P2P/CUMEM +n111-046-125:1663:1776 [5] NCCL INFO Channel 11/0 : 29[5] -> 30[6] via P2P/CUMEM +n111-046-125:1661:1777 [3] NCCL INFO Channel 05/0 : 27[3] -> 28[4] via P2P/CUMEM +n111-046-125:1663:1776 [5] NCCL INFO Channel 12/0 : 29[5] -> 30[6] via P2P/CUMEM +n111-046-125:1662:1770 [4] NCCL INFO Channel 12/0 : 28[4] -> 29[5] via P2P/CUMEM +n111-046-125:1661:1777 [3] NCCL INFO Channel 06/0 : 27[3] -> 28[4] via P2P/CUMEM +n111-046-125:1658:1771 [0] NCCL INFO Channel 04/0 : 24[0] -> 31[7] via P2P/CUMEM +n111-046-125:1663:1776 [5] NCCL INFO Channel 15/0 : 29[5] -> 30[6] via P2P/CUMEM +n111-046-125:1662:1770 [4] NCCL INFO Channel 13/0 : 28[4] -> 29[5] via P2P/CUMEM +n111-046-125:1661:1777 [3] NCCL INFO Channel 07/0 : 27[3] -> 28[4] via P2P/CUMEM +n111-046-125:1662:1770 [4] NCCL INFO Channel 14/0 : 28[4] -> 29[5] via P2P/CUMEM +n111-046-125:1658:1771 [0] NCCL INFO Channel 05/0 : 24[0] -> 31[7] via P2P/CUMEM +n111-046-125:1661:1777 [3] NCCL INFO Channel 08/0 : 27[3] -> 28[4] via P2P/CUMEM +n111-046-125:1662:1770 [4] NCCL INFO Channel 15/0 : 28[4] -> 29[5] via P2P/CUMEM +n111-046-125:1658:1771 [0] NCCL INFO Channel 06/0 : 24[0] -> 31[7] via P2P/CUMEM +n111-046-125:1661:1777 [3] NCCL INFO Channel 09/0 : 27[3] -> 28[4] via P2P/CUMEM +n111-046-125:1658:1771 [0] NCCL INFO Channel 09/0 : 24[0] -> 31[7] via P2P/CUMEM +n111-046-125:1661:1777 [3] NCCL INFO Channel 10/0 : 27[3] -> 28[4] via P2P/CUMEM +n111-046-125:1659:1778 [1] NCCL INFO [Proxy Progress] Device 1 CPU core 24 +n111-046-125:1658:1771 [0] NCCL INFO Channel 10/0 : 24[0] -> 31[7] via P2P/CUMEM +n111-046-125:1661:1777 [3] NCCL INFO Channel 13/0 : 27[3] -> 28[4] via P2P/CUMEM +n111-046-125:1658:1771 [0] NCCL INFO Channel 11/0 : 24[0] -> 31[7] via P2P/CUMEM +n111-046-125:1664:1779 [6] NCCL INFO [Proxy Progress] Device 6 CPU core 88 +n111-046-125:1664:1775 [6] NCCL INFO Channel 06/0 : 22[6] -> 30[6] [receive] via NET/FasTrak/6/GDRDMA +n111-046-125:1659:1774 [1] NCCL INFO Channel 01/0 : 17[1] -> 25[1] [receive] via NET/FasTrak/1/GDRDMA +n111-046-125:1664:1775 [6] NCCL INFO Channel 14/0 : 14[6] -> 30[6] [receive] via NET/FasTrak/6/GDRDMA +n111-046-125:1664:1775 [6] NCCL INFO Channel 14/0 : 30[6] -> 14[6] [send] via NET/FasTrak/6/GDRDMA +n111-046-125:1659:1774 [1] NCCL INFO Channel 09/0 : 9[1] -> 25[1] [receive] via NET/FasTrak/1/GDRDMA +n111-046-125:1659:1774 [1] NCCL INFO Channel 09/0 : 25[1] -> 9[1] [send] via NET/FasTrak/1/GDRDMA +n111-046-125:1659:1774 [1] NCCL INFO Channel 01/0 : 25[1] -> 17[1] [send] via NET/FasTrak/1/GDRDMA +n111-046-125:1661:1777 [3] NCCL INFO Channel 14/0 : 27[3] -> 28[4] via P2P/CUMEM +n111-046-125:1658:1771 [0] NCCL INFO Channel 12/0 : 24[0] -> 31[7] via P2P/CUMEM +n111-046-125:1661:1777 [3] NCCL INFO Channel 15/0 : 27[3] -> 28[4] via P2P/CUMEM +n111-046-125:1658:1771 [0] NCCL INFO Channel 13/0 : 24[0] -> 31[7] via P2P/CUMEM +n111-046-125:1663:1780 [5] NCCL INFO [Proxy Progress] Device 5 CPU core 174 +n111-046-125:1663:1776 [5] NCCL INFO Channel 05/0 : 21[5] -> 29[5] [receive] via NET/FasTrak/5/GDRDMA +n111-046-125:1663:1776 [5] NCCL INFO Channel 13/0 : 13[5] -> 29[5] [receive] via NET/FasTrak/5/GDRDMA +n111-046-125:1663:1776 [5] NCCL INFO Channel 13/0 : 29[5] -> 13[5] [send] via NET/FasTrak/5/GDRDMA +n111-046-125:1663:1776 [5] NCCL INFO Channel 05/0 : 29[5] -> 21[5] [send] via NET/FasTrak/5/GDRDMA +n111-046-125:1658:1771 [0] NCCL INFO Channel 14/0 : 24[0] -> 31[7] via P2P/CUMEM +n111-046-125:1661:1781 [3] NCCL INFO [Proxy Progress] Device 3 CPU core 135 +n111-046-125:1661:1777 [3] NCCL INFO Channel 03/0 : 19[3] -> 27[3] [receive] via NET/FasTrak/3/GDRDMA +n111-046-125:1661:1777 [3] NCCL INFO Channel 11/0 : 11[3] -> 27[3] [receive] via NET/FasTrak/3/GDRDMA +n111-046-125:1661:1777 [3] NCCL INFO Channel 11/0 : 27[3] -> 11[3] [send] via NET/FasTrak/3/GDRDMA +n111-046-125:1661:1777 [3] NCCL INFO Channel 03/0 : 27[3] -> 19[3] [send] via NET/FasTrak/3/GDRDMA +n111-046-125:1660:1782 [2] NCCL INFO [Proxy Progress] Device 2 CPU core 7 +n111-046-125:1662:1783 [4] NCCL INFO [Proxy Progress] Device 4 CPU core 55 +n111-046-125:1660:1773 [2] NCCL INFO Channel 02/0 : 18[2] -> 26[2] [receive] via NET/FasTrak/2/GDRDMA +n111-046-125:1662:1770 [4] NCCL INFO Channel 04/0 : 20[4] -> 28[4] [receive] via NET/FasTrak/4/GDRDMA +n111-046-125:1660:1773 [2] NCCL INFO Channel 10/0 : 10[2] -> 26[2] [receive] via NET/FasTrak/2/GDRDMA +n111-046-125:1660:1773 [2] NCCL INFO Channel 10/0 : 26[2] -> 10[2] [send] via NET/FasTrak/2/GDRDMA +n111-046-125:1662:1770 [4] NCCL INFO Channel 12/0 : 12[4] -> 28[4] [receive] via NET/FasTrak/4/GDRDMA +n111-046-125:1662:1770 [4] NCCL INFO Channel 12/0 : 28[4] -> 12[4] [send] via NET/FasTrak/4/GDRDMA +n111-046-125:1662:1770 [4] NCCL INFO Channel 04/0 : 28[4] -> 20[4] [send] via NET/FasTrak/4/GDRDMA +n111-046-125:1660:1773 [2] NCCL INFO Channel 02/0 : 26[2] -> 18[2] [send] via NET/FasTrak/2/GDRDMA +n111-046-125:1663:1776 [5] NCCL INFO Channel 00/0 : 29[5] -> 28[4] via P2P/CUMEM +n111-046-125:1661:1777 [3] NCCL INFO Channel 00/0 : 27[3] -> 26[2] via P2P/CUMEM +n111-046-125:1663:1776 [5] NCCL INFO Channel 01/0 : 29[5] -> 28[4] via P2P/CUMEM +n111-046-125:1661:1777 [3] NCCL INFO Channel 01/0 : 27[3] -> 26[2] via P2P/CUMEM +n111-046-125:1662:1770 [4] NCCL INFO Channel 00/0 : 28[4] -> 27[3] via P2P/CUMEM +n111-046-125:1660:1773 [2] NCCL INFO Channel 00/0 : 26[2] -> 25[1] via P2P/CUMEM +n111-046-125:1663:1776 [5] NCCL INFO Channel 02/0 : 29[5] -> 28[4] via P2P/CUMEM +n111-046-125:1661:1777 [3] NCCL INFO Channel 02/0 : 27[3] -> 26[2] via P2P/CUMEM +n111-046-125:1662:1770 [4] NCCL INFO Channel 01/0 : 28[4] -> 27[3] via P2P/CUMEM +n111-046-125:1660:1773 [2] NCCL INFO Channel 03/0 : 26[2] -> 25[1] via P2P/CUMEM +n111-046-125:1663:1776 [5] NCCL INFO Channel 03/0 : 29[5] -> 28[4] via P2P/CUMEM +n111-046-125:1661:1777 [3] NCCL INFO Channel 03/0 : 27[3] -> 26[2] via P2P/CUMEM +n111-046-125:1662:1770 [4] NCCL INFO Channel 02/0 : 28[4] -> 27[3] via P2P/CUMEM +n111-046-125:1660:1773 [2] NCCL INFO Channel 04/0 : 26[2] -> 25[1] via P2P/CUMEM +n111-046-125:1663:1776 [5] NCCL INFO Channel 04/0 : 29[5] -> 28[4] via P2P/CUMEM +n111-046-125:1661:1777 [3] NCCL INFO Channel 04/0 : 27[3] -> 26[2] via P2P/CUMEM +n111-046-125:1662:1770 [4] NCCL INFO Channel 05/0 : 28[4] -> 27[3] via P2P/CUMEM +n111-046-125:1660:1773 [2] NCCL INFO Channel 05/0 : 26[2] -> 25[1] via P2P/CUMEM +n111-046-125:1663:1776 [5] NCCL INFO Channel 05/0 : 29[5] -> 28[4] via P2P/CUMEM +n111-046-125:1661:1777 [3] NCCL INFO Channel 05/0 : 27[3] -> 26[2] via P2P/CUMEM +n111-046-125:1662:1770 [4] NCCL INFO Channel 06/0 : 28[4] -> 27[3] via P2P/CUMEM +n111-046-125:1660:1773 [2] NCCL INFO Channel 06/0 : 26[2] -> 25[1] via P2P/CUMEM +n111-046-125:1663:1776 [5] NCCL INFO Channel 06/0 : 29[5] -> 28[4] via P2P/CUMEM +n111-046-125:1661:1777 [3] NCCL INFO Channel 06/0 : 27[3] -> 26[2] via P2P/CUMEM +n111-046-125:1662:1770 [4] NCCL INFO Channel 07/0 : 28[4] -> 27[3] via P2P/CUMEM +n111-046-125:1660:1773 [2] NCCL INFO Channel 07/0 : 26[2] -> 25[1] via P2P/CUMEM +n111-046-125:1663:1776 [5] NCCL INFO Channel 07/0 : 29[5] -> 28[4] via P2P/CUMEM +n111-046-125:1661:1777 [3] NCCL INFO Channel 07/0 : 27[3] -> 26[2] via P2P/CUMEM +n111-046-125:1662:1770 [4] NCCL INFO Channel 08/0 : 28[4] -> 27[3] via P2P/CUMEM +n111-046-125:1660:1773 [2] NCCL INFO Channel 08/0 : 26[2] -> 25[1] via P2P/CUMEM +n111-046-125:1661:1777 [3] NCCL INFO Channel 08/0 : 27[3] -> 26[2] via P2P/CUMEM +n111-046-125:1663:1776 [5] NCCL INFO Channel 08/0 : 29[5] -> 28[4] via P2P/CUMEM +n111-046-125:1662:1770 [4] NCCL INFO Channel 09/0 : 28[4] -> 27[3] via P2P/CUMEM +n111-046-125:1660:1773 [2] NCCL INFO Channel 11/0 : 26[2] -> 25[1] via P2P/CUMEM +n111-046-125:1661:1777 [3] NCCL INFO Channel 09/0 : 27[3] -> 26[2] via P2P/CUMEM +n111-046-125:1663:1776 [5] NCCL INFO Channel 09/0 : 29[5] -> 28[4] via P2P/CUMEM +n111-046-125:1662:1770 [4] NCCL INFO Channel 10/0 : 28[4] -> 27[3] via P2P/CUMEM +n111-046-125:1660:1773 [2] NCCL INFO Channel 12/0 : 26[2] -> 25[1] via P2P/CUMEM +n111-046-125:1661:1777 [3] NCCL INFO Channel 10/0 : 27[3] -> 26[2] via P2P/CUMEM +n111-046-125:1663:1776 [5] NCCL INFO Channel 10/0 : 29[5] -> 28[4] via P2P/CUMEM +n111-046-125:1662:1770 [4] NCCL INFO Channel 13/0 : 28[4] -> 27[3] via P2P/CUMEM +n111-046-125:1661:1777 [3] NCCL INFO Channel 11/0 : 27[3] -> 26[2] via P2P/CUMEM +n111-046-125:1660:1773 [2] NCCL INFO Channel 13/0 : 26[2] -> 25[1] via P2P/CUMEM +n111-046-125:1663:1776 [5] NCCL INFO Channel 11/0 : 29[5] -> 28[4] via P2P/CUMEM +n111-046-125:1662:1770 [4] NCCL INFO Channel 14/0 : 28[4] -> 27[3] via P2P/CUMEM +n111-046-125:1661:1777 [3] NCCL INFO Channel 12/0 : 27[3] -> 26[2] via P2P/CUMEM +n111-046-125:1660:1773 [2] NCCL INFO Channel 14/0 : 26[2] -> 25[1] via P2P/CUMEM +n111-046-125:1662:1770 [4] NCCL INFO Channel 15/0 : 28[4] -> 27[3] via P2P/CUMEM +n111-046-125:1663:1776 [5] NCCL INFO Channel 12/0 : 29[5] -> 28[4] via P2P/CUMEM +n111-046-125:1661:1777 [3] NCCL INFO Channel 13/0 : 27[3] -> 26[2] via P2P/CUMEM +n111-046-125:1660:1773 [2] NCCL INFO Channel 15/0 : 26[2] -> 25[1] via P2P/CUMEM +n111-046-125:1663:1776 [5] NCCL INFO Channel 13/0 : 29[5] -> 28[4] via P2P/CUMEM +n111-046-125:1661:1777 [3] NCCL INFO Channel 14/0 : 27[3] -> 26[2] via P2P/CUMEM +n111-046-125:1663:1776 [5] NCCL INFO Channel 14/0 : 29[5] -> 28[4] via P2P/CUMEM +n111-046-125:1665:1784 [7] NCCL INFO [Proxy Progress] Device 7 CPU core 86 +n111-046-125:1658:1785 [0] NCCL INFO [Proxy Progress] Device 0 CPU core 139 +n111-046-125:1661:1777 [3] NCCL INFO Channel 15/0 : 27[3] -> 26[2] via P2P/CUMEM +n111-046-125:1665:1772 [7] NCCL INFO Channel 07/0 : 23[7] -> 31[7] [receive] via NET/FasTrak/7/GDRDMA +n111-046-125:1663:1776 [5] NCCL INFO Channel 15/0 : 29[5] -> 28[4] via P2P/CUMEM +n111-046-125:1658:1771 [0] NCCL INFO Channel 00/0 : 16[0] -> 24[0] [receive] via NET/FasTrak/0/GDRDMA +n111-046-125:1665:1772 [7] NCCL INFO Channel 15/0 : 15[7] -> 31[7] [receive] via NET/FasTrak/7/GDRDMA +n111-046-125:1665:1772 [7] NCCL INFO Channel 15/0 : 31[7] -> 15[7] [send] via NET/FasTrak/7/GDRDMA +n111-046-125:1658:1771 [0] NCCL INFO Channel 08/0 : 8[0] -> 24[0] [receive] via NET/FasTrak/0/GDRDMA +n111-046-125:1658:1771 [0] NCCL INFO Channel 08/0 : 24[0] -> 8[0] [send] via NET/FasTrak/0/GDRDMA +n111-046-125:1665:1772 [7] NCCL INFO Channel 07/0 : 31[7] -> 23[7] [send] via NET/FasTrak/7/GDRDMA +n111-046-125:1658:1771 [0] NCCL INFO Channel 00/0 : 24[0] -> 16[0] [send] via NET/FasTrak/0/GDRDMA +n111-046-125:1664:1775 [6] NCCL INFO Channel 06/0 : 30[6] -> 22[6] [send] via NET/FasTrak/6/GDRDMA +n111-046-125:1665:1772 [7] NCCL INFO Channel 01/0 : 31[7] -> 24[0] via P2P/CUMEM +n111-046-125:1665:1772 [7] NCCL INFO Channel 02/0 : 31[7] -> 24[0] via P2P/CUMEM +n111-046-125:1665:1772 [7] NCCL INFO Channel 03/0 : 31[7] -> 24[0] via P2P/CUMEM +n111-046-125:1665:1772 [7] NCCL INFO Channel 04/0 : 31[7] -> 24[0] via P2P/CUMEM +n111-046-125:1665:1772 [7] NCCL INFO Channel 05/0 : 31[7] -> 24[0] via P2P/CUMEM +n111-046-125:1665:1772 [7] NCCL INFO Channel 06/0 : 31[7] -> 24[0] via P2P/CUMEM +n111-046-125:1665:1772 [7] NCCL INFO Channel 09/0 : 31[7] -> 24[0] via P2P/CUMEM +n111-046-125:1665:1772 [7] NCCL INFO Channel 10/0 : 31[7] -> 24[0] via P2P/CUMEM +n111-046-125:1665:1772 [7] NCCL INFO Channel 11/0 : 31[7] -> 24[0] via P2P/CUMEM +n111-046-125:1665:1772 [7] NCCL INFO Channel 12/0 : 31[7] -> 24[0] via P2P/CUMEM +n111-046-125:1659:1774 [1] NCCL INFO Channel 00/0 : 25[1] -> 24[0] via P2P/CUMEM +n111-046-125:1665:1772 [7] NCCL INFO Channel 13/0 : 31[7] -> 24[0] via P2P/CUMEM +n111-046-125:1659:1774 [1] NCCL INFO Channel 01/0 : 25[1] -> 24[0] via P2P/CUMEM +n111-046-125:1665:1772 [7] NCCL INFO Channel 14/0 : 31[7] -> 24[0] via P2P/CUMEM +n111-046-125:1659:1774 [1] NCCL INFO Channel 02/0 : 25[1] -> 24[0] via P2P/CUMEM +n111-046-125:1665:1772 [7] NCCL INFO Channel 00/0 : 31[7] -> 30[6] via P2P/CUMEM +n111-046-125:1659:1774 [1] NCCL INFO Channel 03/0 : 25[1] -> 24[0] via P2P/CUMEM +n111-046-125:1665:1772 [7] NCCL INFO Channel 01/0 : 31[7] -> 30[6] via P2P/CUMEM +n111-046-125:1659:1774 [1] NCCL INFO Channel 04/0 : 25[1] -> 24[0] via P2P/CUMEM +n111-046-125:1665:1772 [7] NCCL INFO Channel 02/0 : 31[7] -> 30[6] via P2P/CUMEM +n111-046-125:1659:1774 [1] NCCL INFO Channel 05/0 : 25[1] -> 24[0] via P2P/CUMEM +n111-046-125:1665:1772 [7] NCCL INFO Channel 03/0 : 31[7] -> 30[6] via P2P/CUMEM +n111-046-125:1665:1772 [7] NCCL INFO Channel 04/0 : 31[7] -> 30[6] via P2P/CUMEM +n111-046-125:1664:1775 [6] NCCL INFO Channel 00/0 : 30[6] -> 29[5] via P2P/CUMEM +n111-046-125:1665:1772 [7] NCCL INFO Channel 05/0 : 31[7] -> 30[6] via P2P/CUMEM +n111-046-125:1664:1775 [6] NCCL INFO Channel 01/0 : 30[6] -> 29[5] via P2P/CUMEM +n111-046-125:1664:1775 [6] NCCL INFO Channel 02/0 : 30[6] -> 29[5] via P2P/CUMEM +n111-046-125:1659:1774 [1] NCCL INFO Channel 06/0 : 25[1] -> 24[0] via P2P/CUMEM +n111-046-125:1664:1775 [6] NCCL INFO Channel 03/0 : 30[6] -> 29[5] via P2P/CUMEM +n111-046-125:1659:1774 [1] NCCL INFO Channel 07/0 : 25[1] -> 24[0] via P2P/CUMEM +n111-046-125:1664:1775 [6] NCCL INFO Channel 04/0 : 30[6] -> 29[5] via P2P/CUMEM +n111-046-125:1665:1772 [7] NCCL INFO Channel 06/0 : 31[7] -> 30[6] via P2P/CUMEM +n111-046-125:1664:1775 [6] NCCL INFO Channel 07/0 : 30[6] -> 29[5] via P2P/CUMEM +n111-046-125:1659:1774 [1] NCCL INFO Channel 08/0 : 25[1] -> 24[0] via P2P/CUMEM +n111-046-125:1664:1775 [6] NCCL INFO Channel 08/0 : 30[6] -> 29[5] via P2P/CUMEM +n111-046-125:1659:1774 [1] NCCL INFO Channel 09/0 : 25[1] -> 24[0] via P2P/CUMEM +n111-046-125:1665:1772 [7] NCCL INFO Channel 07/0 : 31[7] -> 30[6] via P2P/CUMEM +n111-046-125:1664:1775 [6] NCCL INFO Channel 09/0 : 30[6] -> 29[5] via P2P/CUMEM +n111-046-125:1659:1774 [1] NCCL INFO Channel 10/0 : 25[1] -> 24[0] via P2P/CUMEM +n111-046-125:1664:1775 [6] NCCL INFO Channel 10/0 : 30[6] -> 29[5] via P2P/CUMEM +n111-046-125:1665:1772 [7] NCCL INFO Channel 08/0 : 31[7] -> 30[6] via P2P/CUMEM +n111-046-125:1659:1774 [1] NCCL INFO Channel 11/0 : 25[1] -> 24[0] via P2P/CUMEM +n111-046-125:1664:1775 [6] NCCL INFO Channel 11/0 : 30[6] -> 29[5] via P2P/CUMEM +n111-046-125:1665:1772 [7] NCCL INFO Channel 09/0 : 31[7] -> 30[6] via P2P/CUMEM +n111-046-125:1664:1775 [6] NCCL INFO Channel 12/0 : 30[6] -> 29[5] via P2P/CUMEM +n111-046-125:1665:1772 [7] NCCL INFO Channel 10/0 : 31[7] -> 30[6] via P2P/CUMEM +n111-046-125:1664:1775 [6] NCCL INFO Channel 15/0 : 30[6] -> 29[5] via P2P/CUMEM +n111-046-125:1659:1774 [1] NCCL INFO Channel 12/0 : 25[1] -> 24[0] via P2P/CUMEM +n111-046-125:1665:1772 [7] NCCL INFO Channel 11/0 : 31[7] -> 30[6] via P2P/CUMEM +n111-046-125:1659:1774 [1] NCCL INFO Channel 13/0 : 25[1] -> 24[0] via P2P/CUMEM +n111-046-125:1665:1772 [7] NCCL INFO Channel 12/0 : 31[7] -> 30[6] via P2P/CUMEM +n111-046-125:1659:1774 [1] NCCL INFO Channel 14/0 : 25[1] -> 24[0] via P2P/CUMEM +n111-046-125:1665:1772 [7] NCCL INFO Channel 13/0 : 31[7] -> 30[6] via P2P/CUMEM +n111-046-125:1665:1772 [7] NCCL INFO Channel 14/0 : 31[7] -> 30[6] via P2P/CUMEM +n111-046-125:1659:1774 [1] NCCL INFO Channel 15/0 : 25[1] -> 24[0] via P2P/CUMEM +n111-046-125:1665:1772 [7] NCCL INFO Channel 15/0 : 31[7] -> 30[6] via P2P/CUMEM +n111-046-125:1661:1777 [3] NCCL INFO Connected all trees +n111-046-125:1662:1770 [4] NCCL INFO Connected all trees +n111-046-125:1660:1773 [2] NCCL INFO Connected all trees +n111-046-125:1663:1776 [5] NCCL INFO Connected all trees +n111-046-125:1658:1771 [0] NCCL INFO Connected all trees +n111-046-125:1659:1774 [1] NCCL INFO Connected all trees +n111-046-125:1665:1772 [7] NCCL INFO Connected all trees +n111-046-125:1664:1775 [6] NCCL INFO Connected all trees +n111-046-125:1661:1807 [3] NCCL INFO Channel 03/0 : 27[3] -> 28[4] via P2P/CUMEM +n111-046-125:1661:1807 [3] NCCL INFO Channel 11/0 : 27[3] -> 28[4] via P2P/CUMEM +n111-046-125:1659:1808 [1] NCCL INFO Channel 01/0 : 25[1] -> 26[2] via P2P/CUMEM +n111-046-125:1659:1808 [1] NCCL INFO Channel 09/0 : 25[1] -> 26[2] via P2P/CUMEM +n111-046-125:1665:1809 [7] NCCL INFO Channel 06/0 : 31[7] -> 6[6] [send] via NET/FasTrak/6/GDRDMA +n111-046-125:1665:1809 [7] NCCL INFO Channel 14/0 : 31[7] -> 6[6] [send] via NET/FasTrak/6/GDRDMA +n111-046-125:1663:1810 [5] NCCL INFO Channel 05/0 : 29[5] -> 30[6] via P2P/CUMEM +n111-046-125:1663:1810 [5] NCCL INFO Channel 13/0 : 29[5] -> 30[6] via P2P/CUMEM +n111-046-125:1658:1811 [0] NCCL INFO Channel 00/0 : 17[1] -> 24[0] [receive] via NET/FasTrak/0/GDRDMA +n111-046-125:1658:1811 [0] NCCL INFO Channel 08/0 : 17[1] -> 24[0] [receive] via NET/FasTrak/0/GDRDMA +n111-046-125:1658:1811 [0] NCCL INFO Channel 00/0 : 24[0] -> 31[7] via P2P/CUMEM +n111-046-125:1658:1811 [0] NCCL INFO Channel 08/0 : 24[0] -> 31[7] via P2P/CUMEM +n111-046-125:1662:1812 [4] NCCL INFO Channel 04/0 : 21[5] -> 28[4] [receive] via NET/FasTrak/4/GDRDMA +n111-046-125:1662:1812 [4] NCCL INFO Channel 12/0 : 21[5] -> 28[4] [receive] via NET/FasTrak/4/GDRDMA +n111-046-125:1661:1807 [3] NCCL INFO Channel 02/0 : 27[3] -> 2[2] [send] via NET/FasTrak/2/GDRDMA +n111-046-125:1661:1807 [3] NCCL INFO Channel 10/0 : 27[3] -> 2[2] [send] via NET/FasTrak/2/GDRDMA +n111-046-125:1661:1807 [3] NCCL INFO Channel 03/0 : 18[2] -> 27[3] [receive] via NET/FasTrak/3/GDRDMA +n111-046-125:1661:1807 [3] NCCL INFO Channel 11/0 : 18[2] -> 27[3] [receive] via NET/FasTrak/3/GDRDMA +n111-046-125:1663:1810 [5] NCCL INFO Channel 04/0 : 29[5] -> 4[4] [send] via NET/FasTrak/4/GDRDMA +n111-046-125:1663:1810 [5] NCCL INFO Channel 12/0 : 29[5] -> 4[4] [send] via NET/FasTrak/4/GDRDMA +n111-046-125:1664:1813 [6] NCCL INFO Channel 06/0 : 23[7] -> 30[6] [receive] via NET/FasTrak/6/GDRDMA +n111-046-125:1664:1813 [6] NCCL INFO Channel 14/0 : 23[7] -> 30[6] [receive] via NET/FasTrak/6/GDRDMA +n111-046-125:1664:1813 [6] NCCL INFO Channel 07/0 : 30[6] -> 7[7] [send] via NET/FasTrak/7/GDRDMA +n111-046-125:1664:1813 [6] NCCL INFO Channel 15/0 : 30[6] -> 7[7] [send] via NET/FasTrak/7/GDRDMA +n111-046-125:1658:1811 [0] NCCL INFO Channel 01/0 : 24[0] -> 1[1] [send] via NET/FasTrak/1/GDRDMA +n111-046-125:1658:1811 [0] NCCL INFO Channel 09/0 : 24[0] -> 1[1] [send] via NET/FasTrak/1/GDRDMA +n111-046-125:1659:1808 [1] NCCL INFO Channel 00/0 : 25[1] -> 0[0] [send] via NET/FasTrak/0/GDRDMA +n111-046-125:1659:1808 [1] NCCL INFO Channel 08/0 : 25[1] -> 0[0] [send] via NET/FasTrak/0/GDRDMA +n111-046-125:1660:1814 [2] NCCL INFO Channel 02/0 : 19[3] -> 26[2] [receive] via NET/FasTrak/2/GDRDMA +n111-046-125:1660:1814 [2] NCCL INFO Channel 10/0 : 19[3] -> 26[2] [receive] via NET/FasTrak/2/GDRDMA +n111-046-125:1660:1814 [2] NCCL INFO Channel 03/0 : 26[2] -> 3[3] [send] via NET/FasTrak/3/GDRDMA +n111-046-125:1660:1814 [2] NCCL INFO Channel 11/0 : 26[2] -> 3[3] [send] via NET/FasTrak/3/GDRDMA +n111-046-125:1659:1808 [1] NCCL INFO Channel 01/0 : 16[0] -> 25[1] [receive] via NET/FasTrak/1/GDRDMA +n111-046-125:1659:1808 [1] NCCL INFO Channel 09/0 : 16[0] -> 25[1] [receive] via NET/FasTrak/1/GDRDMA +n111-046-125:1662:1812 [4] NCCL INFO Channel 05/0 : 28[4] -> 5[5] [send] via NET/FasTrak/5/GDRDMA +n111-046-125:1662:1812 [4] NCCL INFO Channel 13/0 : 28[4] -> 5[5] [send] via NET/FasTrak/5/GDRDMA +n111-046-125:1665:1809 [7] NCCL INFO Channel 07/0 : 22[6] -> 31[7] [receive] via NET/FasTrak/7/GDRDMA +n111-046-125:1662:1812 [4] NCCL INFO Channel 04/0 : 28[4] -> 27[3] via P2P/CUMEM +n111-046-125:1662:1812 [4] NCCL INFO Channel 12/0 : 28[4] -> 27[3] via P2P/CUMEM +n111-046-125:1665:1809 [7] NCCL INFO Channel 15/0 : 22[6] -> 31[7] [receive] via NET/FasTrak/7/GDRDMA +n111-046-125:1660:1814 [2] NCCL INFO Channel 02/0 : 26[2] -> 25[1] via P2P/CUMEM +n111-046-125:1663:1810 [5] NCCL INFO Channel 05/0 : 20[4] -> 29[5] [receive] via NET/FasTrak/5/GDRDMA +n111-046-125:1663:1810 [5] NCCL INFO Channel 13/0 : 20[4] -> 29[5] [receive] via NET/FasTrak/5/GDRDMA +n111-046-125:1660:1814 [2] NCCL INFO Channel 10/0 : 26[2] -> 25[1] via P2P/CUMEM +n111-046-125:1665:1809 [7] NCCL INFO Channel 07/0 : 31[7] -> 24[0] via P2P/CUMEM +n111-046-125:1665:1809 [7] NCCL INFO Channel 15/0 : 31[7] -> 24[0] via P2P/CUMEM +n111-046-125:1664:1813 [6] NCCL INFO Channel 06/0 : 30[6] -> 29[5] via P2P/CUMEM +n111-046-125:1664:1813 [6] NCCL INFO Channel 14/0 : 30[6] -> 29[5] via P2P/CUMEM +n111-046-125:1658:1811 [0] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n111-046-125:1660:1814 [2] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n111-046-125:1659:1808 [1] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n111-046-125:1663:1810 [5] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n111-046-125:1664:1813 [6] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n111-046-125:1665:1809 [7] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n111-046-125:1661:1807 [3] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n111-046-125:1662:1812 [4] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +[rank28]:[W727 05:39:59.035634776 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n111-046-125.byted.org]:45394, remote=[n116-042-185.byted.org]:25003): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f2336f785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f23202cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7f23202cff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7f23202d084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7f23202ca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f22e19c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f233a3e44a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f233cfbb1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f233d03b89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank24]:[W727 05:39:59.035623361 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=71, addr=[n111-046-125.byted.org]:45334, remote=[n116-042-185.byted.org]:25003): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7fce0c1785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7fcdf58cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7fcdf58cff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7fcdf58d084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7fcdf58ca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fcdb6fc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7fce0f8d74a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7fce124ae1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7fce1252e89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank28]:[W727 05:39:59.048147344 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 28] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank24]:[W727 05:39:59.048176893 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 24] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank27]:[W727 05:39:59.094060379 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n111-046-125.byted.org]:45340, remote=[n116-042-185.byted.org]:25003): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7fd6577785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7fd640acdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7fd640acff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7fd640ad084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7fd640aca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fd6021c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7fd65ac044a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7fd65d7db1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7fd65d85b89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank27]:[W727 05:39:59.100379201 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 27] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank30]:[W727 05:39:59.192064058 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n111-046-125.byted.org]:45368, remote=[n116-042-185.byted.org]:25003): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7fb7151785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7fb6fe4cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7fb6fe4cff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7fb6fe4d084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7fb6fe4ca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fb6bfbc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7fb7185b84a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7fb71b18f1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7fb71b20f89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank30]:[W727 05:40:00.199211690 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 30] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank29]:[W727 05:39:59.197644186 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n111-046-125.byted.org]:45350, remote=[n116-042-185.byted.org]:25003): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f86959785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f867eccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7f867eccff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7f867ecd084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7f867ecca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f86403c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f8698e264a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f869b9fd1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f869ba7d89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank29]:[W727 05:40:00.212120333 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 29] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank26]:[W727 05:40:00.226347284 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n111-046-125.byted.org]:45356, remote=[n116-042-185.byted.org]:25003): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f4804b785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f47ee2cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7f47ee2cff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7f47ee2d084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7f47ee2ca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f47af9c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f480831f4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f480aef61f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f480af7689c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank26]:[W727 05:40:00.232610631 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 26] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank25]:[W727 05:40:00.225892598 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n111-046-125.byted.org]:45384, remote=[n116-042-185.byted.org]:25003): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f275d5785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f2746ccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7f2746ccff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7f2746cd084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7f2746cca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f27083c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f2760d6c4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f27639431f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f27639c389c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank25]:[W727 05:40:00.233835716 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 25] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank31]:[W727 05:40:00.459811618 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n111-046-125.byted.org]:45346, remote=[n116-042-185.byted.org]:25003): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7fdd55f785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7fdd3f2cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7fdd3f2cff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7fdd3f2d084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7fdd3f2ca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fdd009c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7fdd593e54a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7fdd5bfbc1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7fdd5c03c89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank31]:[W727 05:40:00.466395901 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 31] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank28]:[W727 05:40:00.048331069 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n111-046-125.byted.org]:45394, remote=[n116-042-185.byted.org]:25003): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f2336f785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f23202cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f23202cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f23202d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f23202ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f22e19c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f233a3e44a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f233cfbb1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f233d03b89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank28]:[W727 05:40:00.055788394 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 28] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank24]:[W727 05:40:00.048355580 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=71, addr=[n111-046-125.byted.org]:45334, remote=[n116-042-185.byted.org]:25003): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7fce0c1785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7fcdf58cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7fcdf58cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7fcdf58d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7fcdf58ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fcdb6fc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7fce0f8d74a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7fce124ae1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7fce1252e89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank24]:[W727 05:40:00.056060919 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 24] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank27]:[W727 05:40:00.100544733 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n111-046-125.byted.org]:45340, remote=[n116-042-185.byted.org]:25003): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7fd6577785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7fd640acdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7fd640acf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7fd640ad0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7fd640aca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fd6021c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7fd65ac044a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7fd65d7db1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7fd65d85b89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank27]:[W727 05:40:00.107688727 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 27] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank30]:[W727 05:40:01.199385808 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n111-046-125.byted.org]:45368, remote=[n116-042-185.byted.org]:25003): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7fb7151785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7fb6fe4cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7fb6fe4cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7fb6fe4d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7fb6fe4ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fb6bfbc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7fb7185b84a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7fb71b18f1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7fb71b20f89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank30]:[W727 05:40:01.206278558 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 30] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank29]:[W727 05:40:01.212291012 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n111-046-125.byted.org]:45350, remote=[n116-042-185.byted.org]:25003): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f86959785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f867eccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f867eccf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f867ecd0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f867ecca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f86403c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f8698e264a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f869b9fd1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f869ba7d89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank29]:[W727 05:40:01.218337851 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 29] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank26]:[W727 05:40:01.232774893 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n111-046-125.byted.org]:45356, remote=[n116-042-185.byted.org]:25003): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f4804b785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f47ee2cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f47ee2cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f47ee2d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f47ee2ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f47af9c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f480831f4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f480aef61f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f480af7689c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank26]:[W727 05:40:01.239890254 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 26] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank25]:[W727 05:40:01.234021271 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n111-046-125.byted.org]:45384, remote=[n116-042-185.byted.org]:25003): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f275d5785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f2746ccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f2746ccf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f2746cd0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f2746cca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f27083c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f2760d6c4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f27639431f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f27639c389c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank25]:[W727 05:40:01.240491921 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 25] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank31]:[W727 05:40:01.466572089 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n111-046-125.byted.org]:45346, remote=[n116-042-185.byted.org]:25003): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7fdd55f785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7fdd3f2cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7fdd3f2cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7fdd3f2d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7fdd3f2ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fdd009c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7fdd593e54a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7fdd5bfbc1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7fdd5c03c89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank31]:[W727 05:40:01.473262926 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 31] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank28]:[W727 05:40:01.055965538 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n111-046-125.byted.org]:45394, remote=[n116-042-185.byted.org]:25003): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f2336f785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f23202cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f23202cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f23202d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f23202ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f22e19c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f233a3e44a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f233cfbb1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f233d03b89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank28]:[W727 05:40:01.061903912 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 28] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank24]:[W727 05:40:01.056223248 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=71, addr=[n111-046-125.byted.org]:45334, remote=[n116-042-185.byted.org]:25003): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7fce0c1785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7fcdf58cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7fcdf58cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7fcdf58d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7fcdf58ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fcdb6fc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7fce0f8d74a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7fce124ae1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7fce1252e89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank24]:[W727 05:40:01.063070488 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 24] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank27]:[W727 05:40:01.107851624 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n111-046-125.byted.org]:45340, remote=[n116-042-185.byted.org]:25003): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7fd6577785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7fd640acdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7fd640acf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7fd640ad0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7fd640aca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fd6021c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7fd65ac044a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7fd65d7db1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7fd65d85b89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank27]:[W727 05:40:01.114052866 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 27] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank30]:[W727 05:40:02.206455526 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n111-046-125.byted.org]:45368, remote=[n116-042-185.byted.org]:25003): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7fb7151785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7fb6fe4cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7fb6fe4cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7fb6fe4d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7fb6fe4ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fb6bfbc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7fb7185b84a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7fb71b18f1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7fb71b20f89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank30]:[W727 05:40:02.213571885 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 30] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank29]:[W727 05:40:02.218497689 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n111-046-125.byted.org]:45350, remote=[n116-042-185.byted.org]:25003): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f86959785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f867eccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f867eccf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f867ecd0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f867ecca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f86403c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f8698e264a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f869b9fd1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f869ba7d89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank29]:[W727 05:40:02.224710919 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 29] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank26]:[W727 05:40:02.240061393 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n111-046-125.byted.org]:45356, remote=[n116-042-185.byted.org]:25003): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f4804b785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f47ee2cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f47ee2cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f47ee2d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f47ee2ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f47af9c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f480831f4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f480aef61f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f480af7689c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank26]:[W727 05:40:02.246167780 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 26] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank25]:[W727 05:40:02.240669479 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n111-046-125.byted.org]:45384, remote=[n116-042-185.byted.org]:25003): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f275d5785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f2746ccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f2746ccf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f2746cd0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f2746cca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f27083c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f2760d6c4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f27639431f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f27639c389c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank25]:[W727 05:40:02.247118311 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 25] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +E0727 05:40:02.144000 1593 torch/distributed/elastic/multiprocessing/api.py:874] failed (exitcode: -15) local_rank: 0 (pid: 1658) of binary: /usr/bin/python3 +Traceback (most recent call last): + File "/usr/local/bin/torchrun", line 7, in + sys.exit(main()) + ^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 355, in wrapper + return f(*args, **kwargs) + ^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 892, in main + run(args) + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 883, in run + elastic_launch( + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 139, in __call__ + return launch_agent(self._config, self._entrypoint, list(args)) + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 270, in launch_agent + raise ChildFailedError( +torch.distributed.elastic.multiprocessing.errors.ChildFailedError: +============================================================ +main_cross_2direct_balance_realscale_perturb_v4_accum.py FAILED +------------------------------------------------------------ +Failures: +[1]: + time : 2026-07-27_05:40:02 + host : n111-046-125.byted.org + rank : 25 (local_rank: 1) + exitcode : -15 (pid: 1659) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 1659 +[2]: + time : 2026-07-27_05:40:02 + host : n111-046-125.byted.org + rank : 26 (local_rank: 2) + exitcode : -15 (pid: 1660) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 1660 +[3]: + time : 2026-07-27_05:40:02 + host : n111-046-125.byted.org + rank : 27 (local_rank: 3) + exitcode : -15 (pid: 1661) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 1661 +[4]: + time : 2026-07-27_05:40:02 + host : n111-046-125.byted.org + rank : 28 (local_rank: 4) + exitcode : -15 (pid: 1662) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 1662 +[5]: + time : 2026-07-27_05:40:02 + host : n111-046-125.byted.org + rank : 29 (local_rank: 5) + exitcode : -15 (pid: 1663) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 1663 +[6]: + time : 2026-07-27_05:40:02 + host : n111-046-125.byted.org + rank : 30 (local_rank: 6) + exitcode : -15 (pid: 1664) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 1664 +[7]: + time : 2026-07-27_05:40:02 + host : n111-046-125.byted.org + rank : 31 (local_rank: 7) + exitcode : -15 (pid: 1665) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 1665 +------------------------------------------------------------ +Root Cause (first observed failure): +[0]: + time : 2026-07-27_05:40:02 + host : n111-046-125.byted.org + rank : 24 (local_rank: 0) + exitcode : -15 (pid: 1658) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 1658 +============================================================ diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1603_vith32_launch_stub_master_n116-042-185_port25003/output_root.txt b/ScaleDF_rebuttal/_host_logs/20260726_1603_vith32_launch_stub_master_n116-042-185_port25003/output_root.txt new file mode 100644 index 0000000000000000000000000000000000000000..eea600474314d176b8d14a4a698b20e8f5a6aa28 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1603_vith32_launch_stub_master_n116-042-185_port25003/output_root.txt @@ -0,0 +1 @@ +/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_32card_effbs2048_from_latest24_master_n116-042-185_20260726_1603_vith32_stub_port25003 diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1603_vith32_launch_stub_master_n116-042-185_port25003/pid_rank0_n116-042-185.txt b/ScaleDF_rebuttal/_host_logs/20260726_1603_vith32_launch_stub_master_n116-042-185_port25003/pid_rank0_n116-042-185.txt new file mode 100644 index 0000000000000000000000000000000000000000..c9ac2c6be71413802bdbaaf66a7465204a4099d2 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1603_vith32_launch_stub_master_n116-042-185_port25003/pid_rank0_n116-042-185.txt @@ -0,0 +1 @@ +1907 diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1603_vith32_launch_stub_master_n116-042-185_port25003/pid_rank1_n111-046-078.txt b/ScaleDF_rebuttal/_host_logs/20260726_1603_vith32_launch_stub_master_n116-042-185_port25003/pid_rank1_n111-046-078.txt new file mode 100644 index 0000000000000000000000000000000000000000..c989d24ff3203a1ac94aea670a6359c946acadb9 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1603_vith32_launch_stub_master_n116-042-185_port25003/pid_rank1_n111-046-078.txt @@ -0,0 +1 @@ +3240 diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1603_vith32_launch_stub_master_n116-042-185_port25003/pid_rank2_n116-040-048.txt b/ScaleDF_rebuttal/_host_logs/20260726_1603_vith32_launch_stub_master_n116-042-185_port25003/pid_rank2_n116-040-048.txt new file mode 100644 index 0000000000000000000000000000000000000000..859a8b222aac96a4330a50775302c04f48c4c9d4 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1603_vith32_launch_stub_master_n116-042-185_port25003/pid_rank2_n116-040-048.txt @@ -0,0 +1 @@ +1706 diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1603_vith32_launch_stub_master_n116-042-185_port25003/pid_rank3_n111-046-125.txt b/ScaleDF_rebuttal/_host_logs/20260726_1603_vith32_launch_stub_master_n116-042-185_port25003/pid_rank3_n111-046-125.txt new file mode 100644 index 0000000000000000000000000000000000000000..5882485b11949f20cd1c239d14ef72512657eeb6 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1603_vith32_launch_stub_master_n116-042-185_port25003/pid_rank3_n111-046-125.txt @@ -0,0 +1 @@ +1585 diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1603_vith32_launch_stub_master_n116-042-185_port25003/resume.txt b/ScaleDF_rebuttal/_host_logs/20260726_1603_vith32_launch_stub_master_n116-042-185_port25003/resume.txt new file mode 100644 index 0000000000000000000000000000000000000000..2158d35929432dee54a9499c09d7990967633387 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1603_vith32_launch_stub_master_n116-042-185_port25003/resume.txt @@ -0,0 +1 @@ +/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_24card_effbs3072_from_epoch7_n116-041-246_n116-042-186_n116-042-185_20260726_0736_vith24_from_epoch7/checkpoint.pth diff --git a/ScaleDF_rebuttal/_host_logs/20260726_1603_vith32_launch_stub_master_n116-042-185_port25003/run_name.txt b/ScaleDF_rebuttal/_host_logs/20260726_1603_vith32_launch_stub_master_n116-042-185_port25003/run_name.txt new file mode 100644 index 0000000000000000000000000000000000000000..5b36ca8a799e431533ae29c649d346576ff33935 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_1603_vith32_launch_stub_master_n116-042-185_port25003/run_name.txt @@ -0,0 +1 @@ +modelsize_vith_40epoch_32card_effbs2048_from_latest24_master_n116-042-185_20260726_1603_vith32_stub_port25003 diff --git a/ScaleDF_rebuttal/_host_logs/20260726_2338_vith64_free8_port25164_launch/launch_meta.txt b/ScaleDF_rebuttal/_host_logs/20260726_2338_vith64_free8_port25164_launch/launch_meta.txt new file mode 100644 index 0000000000000000000000000000000000000000..88db7950b34995f52aef673a1873c92897f6d93b --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_2338_vith64_free8_port25164_launch/launch_meta.txt @@ -0,0 +1,15 @@ +tag=20260726_2338_vith64_free8_port25164 +run_name=modelsize_vith_40epoch_64card_effbs4096_from_epoch22_free8_master_n116-040-168_20260726_2338_vith64_free8_port25164 +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_64card_effbs4096_from_epoch22_free8_master_n116-040-168_20260726_2338_vith64_free8_port25164 +resume=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_32card_effbs2048_from_latest24_master_n116-042-185_20260726_1603_vith32_stub_port25003/checkpoint.pth +master_addr=10.116.40.168 +master_port=25164 +effective_batch=4096 +rank0=n116-040-168 port=10338 host=fdbd:dccd:cdc1:1368:0:e4:: +rank1=n116-041-153 port=11208 host=fdbd:dccd:cdc1:1368:0:1d6:: +rank2=n116-041-188 port=9869 host=fdbd:dccd:cdc1:1368:0:1f8:: +rank3=n111-046-081 port=11986 host=fdbd:dccd:cdc1:1302:0:8d:: +rank4=n116-041-023 port=11693 host=fdbd:dccd:cdc1:1368:0:153:: +rank5=n116-040-213 port=9999 host=fdbd:dccd:cdc1:1368:0:111:: +rank6=n116-041-181 port=11062 host=fdbd:dccd:cdc1:1368:0:1f1:: +rank7=n116-042-235 port=11369 host=fdbd:dccd:cdc1:1368:0:327:: diff --git a/ScaleDF_rebuttal/_host_logs/20260726_2338_vith64_free8_port25164_launch/launcher_rank0_n116-040-168.log b/ScaleDF_rebuttal/_host_logs/20260726_2338_vith64_free8_port25164_launch/launcher_rank0_n116-040-168.log new file mode 100644 index 0000000000000000000000000000000000000000..b2c7719f384189e00fe66ff22bad34f02de40b07 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_2338_vith64_free8_port25164_launch/launcher_rank0_n116-040-168.log @@ -0,0 +1,113 @@ +started_utc=2026-07-26T23:38:09Z +hostname=n116-040-168 +node_rank=0 +nnodes=8 +nproc_per_node=8 +world_size=64 +master_addr=10.116.40.168 +master_port=25164 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +main=main_cross_2direct_balance_realscale_perturb_v4_accum.py +model_name=deit_huge_patch14_LS +pretrain=/tmp/ScaleDF_repro_20260724/pretrained/deit_3_huge_224_21k_v1.pth +resume=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_32card_effbs2048_from_latest24_master_n116-042-185_20260726_1603_vith32_stub_port25003/checkpoint.pth +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_64card_effbs4096_from_epoch22_free8_master_n116-040-168_20260726_2338_vith64_free8_port25164 +batch_per_gpu=64 +accum_iter=1 +effective_batch=4096 +eval_batch_per_gpu=64 +epochs=40 +W0726 23:38:24.454000 3350238 torch/distributed/run.py:766] +W0726 23:38:24.454000 3350238 torch/distributed/run.py:766] ***************************************** +W0726 23:38:24.454000 3350238 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 23:38:24.454000 3350238 torch/distributed/run.py:766] ***************************************** +/usr/bin/python3: can't open file '/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py': [Errno 2] No such file or directory +/usr/bin/python3: can't open file '/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py': [Errno 2] No such file or directory +/usr/bin/python3: can't open file '/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py': [Errno 2] No such file or directory +/usr/bin/python3: can't open file '/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py': [Errno 2] No such file or directory +/usr/bin/python3: can't open file '/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py': [Errno 2] No such file or directory +/usr/bin/python3: can't open file '/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py': [Errno 2] No such file or directory +/usr/bin/python3: can't open file '/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py': [Errno 2] No such file or directory +/usr/bin/python3: can't open file '/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py': [Errno 2] No such file or directory +E0726 23:38:28.162000 3350238 torch/distributed/elastic/multiprocessing/api.py:874] failed (exitcode: 2) local_rank: 0 (pid: 3350304) of binary: /usr/bin/python3 +Traceback (most recent call last): + File "/usr/local/bin/torchrun", line 7, in + sys.exit(main()) + ^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 355, in wrapper + return f(*args, **kwargs) + ^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 892, in main + run(args) + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 883, in run + elastic_launch( + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 139, in __call__ + return launch_agent(self._config, self._entrypoint, list(args)) + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 270, in launch_agent + raise ChildFailedError( +torch.distributed.elastic.multiprocessing.errors.ChildFailedError: +============================================================ +main_cross_2direct_balance_realscale_perturb_v4_accum.py FAILED +------------------------------------------------------------ +Failures: +[1]: + time : 2026-07-26_23:38:28 + host : n116-040-168.byted.org + rank : 1 (local_rank: 1) + exitcode : 2 (pid: 3350305) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +[2]: + time : 2026-07-26_23:38:28 + host : n116-040-168.byted.org + rank : 2 (local_rank: 2) + exitcode : 2 (pid: 3350306) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +[3]: + time : 2026-07-26_23:38:28 + host : n116-040-168.byted.org + rank : 3 (local_rank: 3) + exitcode : 2 (pid: 3350307) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +[4]: + time : 2026-07-26_23:38:28 + host : n116-040-168.byted.org + rank : 4 (local_rank: 4) + exitcode : 2 (pid: 3350308) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +[5]: + time : 2026-07-26_23:38:28 + host : n116-040-168.byted.org + rank : 5 (local_rank: 5) + exitcode : 2 (pid: 3350309) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +[6]: + time : 2026-07-26_23:38:28 + host : n116-040-168.byted.org + rank : 6 (local_rank: 6) + exitcode : 2 (pid: 3350310) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +[7]: + time : 2026-07-26_23:38:28 + host : n116-040-168.byted.org + rank : 7 (local_rank: 7) + exitcode : 2 (pid: 3350311) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +------------------------------------------------------------ +Root Cause (first observed failure): +[0]: + time : 2026-07-26_23:38:28 + host : n116-040-168.byted.org + rank : 0 (local_rank: 0) + exitcode : 2 (pid: 3350304) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +============================================================ diff --git a/ScaleDF_rebuttal/_host_logs/20260726_2338_vith64_free8_port25164_launch/launcher_rank0_n116-040-168.pid b/ScaleDF_rebuttal/_host_logs/20260726_2338_vith64_free8_port25164_launch/launcher_rank0_n116-040-168.pid new file mode 100644 index 0000000000000000000000000000000000000000..ff3864df556e8d3a2e9a15d1c9f2d4ec10d5bfc1 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_2338_vith64_free8_port25164_launch/launcher_rank0_n116-040-168.pid @@ -0,0 +1 @@ +3350232 diff --git a/ScaleDF_rebuttal/_host_logs/20260726_2338_vith64_free8_port25164_launch/launcher_rank1_n116-041-153.log b/ScaleDF_rebuttal/_host_logs/20260726_2338_vith64_free8_port25164_launch/launcher_rank1_n116-041-153.log new file mode 100644 index 0000000000000000000000000000000000000000..b94e9cf6e085b533593f036d4528ae9cefe5550f --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_2338_vith64_free8_port25164_launch/launcher_rank1_n116-041-153.log @@ -0,0 +1,113 @@ +started_utc=2026-07-26T23:38:10Z +hostname=n116-041-153 +node_rank=1 +nnodes=8 +nproc_per_node=8 +world_size=64 +master_addr=10.116.40.168 +master_port=25164 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +main=main_cross_2direct_balance_realscale_perturb_v4_accum.py +model_name=deit_huge_patch14_LS +pretrain=/tmp/ScaleDF_repro_20260724/pretrained/deit_3_huge_224_21k_v1.pth +resume=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_32card_effbs2048_from_latest24_master_n116-042-185_20260726_1603_vith32_stub_port25003/checkpoint.pth +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_64card_effbs4096_from_epoch22_free8_master_n116-040-168_20260726_2338_vith64_free8_port25164 +batch_per_gpu=64 +accum_iter=1 +effective_batch=4096 +eval_batch_per_gpu=64 +epochs=40 +W0726 23:38:22.614000 1379410 torch/distributed/run.py:766] +W0726 23:38:22.614000 1379410 torch/distributed/run.py:766] ***************************************** +W0726 23:38:22.614000 1379410 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 23:38:22.614000 1379410 torch/distributed/run.py:766] ***************************************** +/usr/bin/python3: can't open file '/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py': [Errno 2] No such file or directory +/usr/bin/python3: can't open file '/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py': [Errno 2] No such file or directory +/usr/bin/python3: can't open file '/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py': [Errno 2] No such file or directory +/usr/bin/python3: can't open file '/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py': [Errno 2] No such file or directory +/usr/bin/python3: can't open file '/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py': [Errno 2] No such file or directory +/usr/bin/python3: can't open file '/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py': [Errno 2] No such file or directory +/usr/bin/python3: can't open file '/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py': [Errno 2] No such file or directory +/usr/bin/python3: can't open file '/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py': [Errno 2] No such file or directory +E0726 23:38:28.060000 1379410 torch/distributed/elastic/multiprocessing/api.py:874] failed (exitcode: 2) local_rank: 0 (pid: 1379475) of binary: /usr/bin/python3 +Traceback (most recent call last): + File "/usr/local/bin/torchrun", line 7, in + sys.exit(main()) + ^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 355, in wrapper + return f(*args, **kwargs) + ^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 892, in main + run(args) + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 883, in run + elastic_launch( + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 139, in __call__ + return launch_agent(self._config, self._entrypoint, list(args)) + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 270, in launch_agent + raise ChildFailedError( +torch.distributed.elastic.multiprocessing.errors.ChildFailedError: +============================================================ +main_cross_2direct_balance_realscale_perturb_v4_accum.py FAILED +------------------------------------------------------------ +Failures: +[1]: + time : 2026-07-26_23:38:28 + host : n116-041-153.byted.org + rank : 9 (local_rank: 1) + exitcode : 2 (pid: 1379476) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +[2]: + time : 2026-07-26_23:38:28 + host : n116-041-153.byted.org + rank : 10 (local_rank: 2) + exitcode : 2 (pid: 1379477) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +[3]: + time : 2026-07-26_23:38:28 + host : n116-041-153.byted.org + rank : 11 (local_rank: 3) + exitcode : 2 (pid: 1379478) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +[4]: + time : 2026-07-26_23:38:28 + host : n116-041-153.byted.org + rank : 12 (local_rank: 4) + exitcode : 2 (pid: 1379479) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +[5]: + time : 2026-07-26_23:38:28 + host : n116-041-153.byted.org + rank : 13 (local_rank: 5) + exitcode : 2 (pid: 1379480) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +[6]: + time : 2026-07-26_23:38:28 + host : n116-041-153.byted.org + rank : 14 (local_rank: 6) + exitcode : 2 (pid: 1379481) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +[7]: + time : 2026-07-26_23:38:28 + host : n116-041-153.byted.org + rank : 15 (local_rank: 7) + exitcode : 2 (pid: 1379482) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +------------------------------------------------------------ +Root Cause (first observed failure): +[0]: + time : 2026-07-26_23:38:28 + host : n116-041-153.byted.org + rank : 8 (local_rank: 0) + exitcode : 2 (pid: 1379475) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +============================================================ diff --git a/ScaleDF_rebuttal/_host_logs/20260726_2338_vith64_free8_port25164_launch/launcher_rank1_n116-041-153.pid b/ScaleDF_rebuttal/_host_logs/20260726_2338_vith64_free8_port25164_launch/launcher_rank1_n116-041-153.pid new file mode 100644 index 0000000000000000000000000000000000000000..20d4c7c59093c7c65e715fd1a21fae27f859f5f1 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_2338_vith64_free8_port25164_launch/launcher_rank1_n116-041-153.pid @@ -0,0 +1 @@ +1379404 diff --git a/ScaleDF_rebuttal/_host_logs/20260726_2338_vith64_free8_port25164_launch/launcher_rank2_n116-041-188.log b/ScaleDF_rebuttal/_host_logs/20260726_2338_vith64_free8_port25164_launch/launcher_rank2_n116-041-188.log new file mode 100644 index 0000000000000000000000000000000000000000..bfaa3ec044c8beeadd3b58cafa9de0bedef6c002 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_2338_vith64_free8_port25164_launch/launcher_rank2_n116-041-188.log @@ -0,0 +1,113 @@ +started_utc=2026-07-26T23:38:10Z +hostname=n116-041-188 +node_rank=2 +nnodes=8 +nproc_per_node=8 +world_size=64 +master_addr=10.116.40.168 +master_port=25164 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +main=main_cross_2direct_balance_realscale_perturb_v4_accum.py +model_name=deit_huge_patch14_LS +pretrain=/tmp/ScaleDF_repro_20260724/pretrained/deit_3_huge_224_21k_v1.pth +resume=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_32card_effbs2048_from_latest24_master_n116-042-185_20260726_1603_vith32_stub_port25003/checkpoint.pth +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_64card_effbs4096_from_epoch22_free8_master_n116-040-168_20260726_2338_vith64_free8_port25164 +batch_per_gpu=64 +accum_iter=1 +effective_batch=4096 +eval_batch_per_gpu=64 +epochs=40 +W0726 23:38:23.481000 1376746 torch/distributed/run.py:766] +W0726 23:38:23.481000 1376746 torch/distributed/run.py:766] ***************************************** +W0726 23:38:23.481000 1376746 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 23:38:23.481000 1376746 torch/distributed/run.py:766] ***************************************** +/usr/bin/python3: can't open file '/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py': [Errno 2] No such file or directory +/usr/bin/python3: can't open file '/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py': [Errno 2] No such file or directory +/usr/bin/python3: can't open file '/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py': [Errno 2] No such file or directory +/usr/bin/python3: can't open file '/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py': [Errno 2] No such file or directory +/usr/bin/python3: can't open file '/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py': [Errno 2] No such file or directory +/usr/bin/python3: can't open file '/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py': [Errno 2] No such file or directory +/usr/bin/python3: can't open file '/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py': [Errno 2] No such file or directory +/usr/bin/python3: can't open file '/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py': [Errno 2] No such file or directory +E0726 23:38:28.056000 1376746 torch/distributed/elastic/multiprocessing/api.py:874] failed (exitcode: 2) local_rank: 0 (pid: 1376811) of binary: /usr/bin/python3 +Traceback (most recent call last): + File "/usr/local/bin/torchrun", line 7, in + sys.exit(main()) + ^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 355, in wrapper + return f(*args, **kwargs) + ^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 892, in main + run(args) + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 883, in run + elastic_launch( + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 139, in __call__ + return launch_agent(self._config, self._entrypoint, list(args)) + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 270, in launch_agent + raise ChildFailedError( +torch.distributed.elastic.multiprocessing.errors.ChildFailedError: +============================================================ +main_cross_2direct_balance_realscale_perturb_v4_accum.py FAILED +------------------------------------------------------------ +Failures: +[1]: + time : 2026-07-26_23:38:28 + host : n116-041-188.byted.org + rank : 17 (local_rank: 1) + exitcode : 2 (pid: 1376812) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +[2]: + time : 2026-07-26_23:38:28 + host : n116-041-188.byted.org + rank : 18 (local_rank: 2) + exitcode : 2 (pid: 1376813) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +[3]: + time : 2026-07-26_23:38:28 + host : n116-041-188.byted.org + rank : 19 (local_rank: 3) + exitcode : 2 (pid: 1376814) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +[4]: + time : 2026-07-26_23:38:28 + host : n116-041-188.byted.org + rank : 20 (local_rank: 4) + exitcode : 2 (pid: 1376815) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +[5]: + time : 2026-07-26_23:38:28 + host : n116-041-188.byted.org + rank : 21 (local_rank: 5) + exitcode : 2 (pid: 1376816) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +[6]: + time : 2026-07-26_23:38:28 + host : n116-041-188.byted.org + rank : 22 (local_rank: 6) + exitcode : 2 (pid: 1376817) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +[7]: + time : 2026-07-26_23:38:28 + host : n116-041-188.byted.org + rank : 23 (local_rank: 7) + exitcode : 2 (pid: 1376818) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +------------------------------------------------------------ +Root Cause (first observed failure): +[0]: + time : 2026-07-26_23:38:28 + host : n116-041-188.byted.org + rank : 16 (local_rank: 0) + exitcode : 2 (pid: 1376811) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +============================================================ diff --git a/ScaleDF_rebuttal/_host_logs/20260726_2338_vith64_free8_port25164_launch/launcher_rank2_n116-041-188.pid b/ScaleDF_rebuttal/_host_logs/20260726_2338_vith64_free8_port25164_launch/launcher_rank2_n116-041-188.pid new file mode 100644 index 0000000000000000000000000000000000000000..c3793dac0182a19213f41fe94f935bb18d7c2ed4 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_2338_vith64_free8_port25164_launch/launcher_rank2_n116-041-188.pid @@ -0,0 +1 @@ +1376740 diff --git a/ScaleDF_rebuttal/_host_logs/20260726_2338_vith64_free8_port25164_launch/launcher_rank3_n111-046-081.log b/ScaleDF_rebuttal/_host_logs/20260726_2338_vith64_free8_port25164_launch/launcher_rank3_n111-046-081.log new file mode 100644 index 0000000000000000000000000000000000000000..1fe915ca40998338bd5bb31eb9b3883d581ca3ce --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_2338_vith64_free8_port25164_launch/launcher_rank3_n111-046-081.log @@ -0,0 +1,113 @@ +started_utc=2026-07-26T23:38:11Z +hostname=n111-046-081 +node_rank=3 +nnodes=8 +nproc_per_node=8 +world_size=64 +master_addr=10.116.40.168 +master_port=25164 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +main=main_cross_2direct_balance_realscale_perturb_v4_accum.py +model_name=deit_huge_patch14_LS +pretrain=/tmp/ScaleDF_repro_20260724/pretrained/deit_3_huge_224_21k_v1.pth +resume=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_32card_effbs2048_from_latest24_master_n116-042-185_20260726_1603_vith32_stub_port25003/checkpoint.pth +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_64card_effbs4096_from_epoch22_free8_master_n116-040-168_20260726_2338_vith64_free8_port25164 +batch_per_gpu=64 +accum_iter=1 +effective_batch=4096 +eval_batch_per_gpu=64 +epochs=40 +W0726 23:38:23.731000 3352031 torch/distributed/run.py:766] +W0726 23:38:23.731000 3352031 torch/distributed/run.py:766] ***************************************** +W0726 23:38:23.731000 3352031 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 23:38:23.731000 3352031 torch/distributed/run.py:766] ***************************************** +/usr/bin/python3: can't open file '/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py': [Errno 2] No such file or directory +/usr/bin/python3: can't open file '/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py': [Errno 2] No such file or directory +/usr/bin/python3: can't open file '/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py': [Errno 2] No such file or directory +/usr/bin/python3: can't open file '/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py': [Errno 2] No such file or directory +/usr/bin/python3: can't open file '/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py': [Errno 2] No such file or directory +/usr/bin/python3: can't open file '/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py': [Errno 2] No such file or directory +/usr/bin/python3: can't open file '/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py': [Errno 2] No such file or directory +/usr/bin/python3: can't open file '/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py': [Errno 2] No such file or directory +E0726 23:38:28.048000 3352031 torch/distributed/elastic/multiprocessing/api.py:874] failed (exitcode: 2) local_rank: 0 (pid: 3352096) of binary: /usr/bin/python3 +Traceback (most recent call last): + File "/usr/local/bin/torchrun", line 7, in + sys.exit(main()) + ^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 355, in wrapper + return f(*args, **kwargs) + ^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 892, in main + run(args) + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 883, in run + elastic_launch( + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 139, in __call__ + return launch_agent(self._config, self._entrypoint, list(args)) + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 270, in launch_agent + raise ChildFailedError( +torch.distributed.elastic.multiprocessing.errors.ChildFailedError: +============================================================ +main_cross_2direct_balance_realscale_perturb_v4_accum.py FAILED +------------------------------------------------------------ +Failures: +[1]: + time : 2026-07-26_23:38:28 + host : n111-046-081.byted.org + rank : 25 (local_rank: 1) + exitcode : 2 (pid: 3352097) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +[2]: + time : 2026-07-26_23:38:28 + host : n111-046-081.byted.org + rank : 26 (local_rank: 2) + exitcode : 2 (pid: 3352098) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +[3]: + time : 2026-07-26_23:38:28 + host : n111-046-081.byted.org + rank : 27 (local_rank: 3) + exitcode : 2 (pid: 3352099) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +[4]: + time : 2026-07-26_23:38:28 + host : n111-046-081.byted.org + rank : 28 (local_rank: 4) + exitcode : 2 (pid: 3352100) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +[5]: + time : 2026-07-26_23:38:28 + host : n111-046-081.byted.org + rank : 29 (local_rank: 5) + exitcode : 2 (pid: 3352101) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +[6]: + time : 2026-07-26_23:38:28 + host : n111-046-081.byted.org + rank : 30 (local_rank: 6) + exitcode : 2 (pid: 3352102) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +[7]: + time : 2026-07-26_23:38:28 + host : n111-046-081.byted.org + rank : 31 (local_rank: 7) + exitcode : 2 (pid: 3352103) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +------------------------------------------------------------ +Root Cause (first observed failure): +[0]: + time : 2026-07-26_23:38:28 + host : n111-046-081.byted.org + rank : 24 (local_rank: 0) + exitcode : 2 (pid: 3352096) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +============================================================ diff --git a/ScaleDF_rebuttal/_host_logs/20260726_2338_vith64_free8_port25164_launch/launcher_rank3_n111-046-081.pid b/ScaleDF_rebuttal/_host_logs/20260726_2338_vith64_free8_port25164_launch/launcher_rank3_n111-046-081.pid new file mode 100644 index 0000000000000000000000000000000000000000..729c89a875b455e5503f4fc5ad67e1cc5e4f0a06 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_2338_vith64_free8_port25164_launch/launcher_rank3_n111-046-081.pid @@ -0,0 +1 @@ +3352025 diff --git a/ScaleDF_rebuttal/_host_logs/20260726_2338_vith64_free8_port25164_launch/launcher_rank4_n116-041-023.log b/ScaleDF_rebuttal/_host_logs/20260726_2338_vith64_free8_port25164_launch/launcher_rank4_n116-041-023.log new file mode 100644 index 0000000000000000000000000000000000000000..847824334dd113ab1ddd4a87f790bb86eca5846a --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_2338_vith64_free8_port25164_launch/launcher_rank4_n116-041-023.log @@ -0,0 +1,113 @@ +started_utc=2026-07-26T23:38:12Z +hostname=n116-041-023 +node_rank=4 +nnodes=8 +nproc_per_node=8 +world_size=64 +master_addr=10.116.40.168 +master_port=25164 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +main=main_cross_2direct_balance_realscale_perturb_v4_accum.py +model_name=deit_huge_patch14_LS +pretrain=/tmp/ScaleDF_repro_20260724/pretrained/deit_3_huge_224_21k_v1.pth +resume=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_32card_effbs2048_from_latest24_master_n116-042-185_20260726_1603_vith32_stub_port25003/checkpoint.pth +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_64card_effbs4096_from_epoch22_free8_master_n116-040-168_20260726_2338_vith64_free8_port25164 +batch_per_gpu=64 +accum_iter=1 +effective_batch=4096 +eval_batch_per_gpu=64 +epochs=40 +W0726 23:38:25.046000 1382637 torch/distributed/run.py:766] +W0726 23:38:25.046000 1382637 torch/distributed/run.py:766] ***************************************** +W0726 23:38:25.046000 1382637 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 23:38:25.046000 1382637 torch/distributed/run.py:766] ***************************************** +/usr/bin/python3: can't open file '/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py': [Errno 2] No such file or directory +/usr/bin/python3: can't open file '/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py': [Errno 2] No such file or directory +/usr/bin/python3: can't open file '/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py': [Errno 2] No such file or directory +/usr/bin/python3: can't open file '/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py': [Errno 2] No such file or directory +/usr/bin/python3: can't open file '/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py': [Errno 2] No such file or directory +/usr/bin/python3: can't open file '/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py': [Errno 2] No such file or directory +/usr/bin/python3: can't open file '/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py': [Errno 2] No such file or directory +/usr/bin/python3: can't open file '/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py': [Errno 2] No such file or directory +E0726 23:38:28.052000 1382637 torch/distributed/elastic/multiprocessing/api.py:874] failed (exitcode: 2) local_rank: 0 (pid: 1382702) of binary: /usr/bin/python3 +Traceback (most recent call last): + File "/usr/local/bin/torchrun", line 7, in + sys.exit(main()) + ^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 355, in wrapper + return f(*args, **kwargs) + ^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 892, in main + run(args) + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 883, in run + elastic_launch( + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 139, in __call__ + return launch_agent(self._config, self._entrypoint, list(args)) + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 270, in launch_agent + raise ChildFailedError( +torch.distributed.elastic.multiprocessing.errors.ChildFailedError: +============================================================ +main_cross_2direct_balance_realscale_perturb_v4_accum.py FAILED +------------------------------------------------------------ +Failures: +[1]: + time : 2026-07-26_23:38:28 + host : n116-041-023.byted.org + rank : 33 (local_rank: 1) + exitcode : 2 (pid: 1382703) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +[2]: + time : 2026-07-26_23:38:28 + host : n116-041-023.byted.org + rank : 34 (local_rank: 2) + exitcode : 2 (pid: 1382704) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +[3]: + time : 2026-07-26_23:38:28 + host : n116-041-023.byted.org + rank : 35 (local_rank: 3) + exitcode : 2 (pid: 1382705) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +[4]: + time : 2026-07-26_23:38:28 + host : n116-041-023.byted.org + rank : 36 (local_rank: 4) + exitcode : 2 (pid: 1382706) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +[5]: + time : 2026-07-26_23:38:28 + host : n116-041-023.byted.org + rank : 37 (local_rank: 5) + exitcode : 2 (pid: 1382707) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +[6]: + time : 2026-07-26_23:38:28 + host : n116-041-023.byted.org + rank : 38 (local_rank: 6) + exitcode : 2 (pid: 1382708) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +[7]: + time : 2026-07-26_23:38:28 + host : n116-041-023.byted.org + rank : 39 (local_rank: 7) + exitcode : 2 (pid: 1382709) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +------------------------------------------------------------ +Root Cause (first observed failure): +[0]: + time : 2026-07-26_23:38:28 + host : n116-041-023.byted.org + rank : 32 (local_rank: 0) + exitcode : 2 (pid: 1382702) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +============================================================ diff --git a/ScaleDF_rebuttal/_host_logs/20260726_2338_vith64_free8_port25164_launch/launcher_rank4_n116-041-023.pid b/ScaleDF_rebuttal/_host_logs/20260726_2338_vith64_free8_port25164_launch/launcher_rank4_n116-041-023.pid new file mode 100644 index 0000000000000000000000000000000000000000..fa0114fdd6e7269589b255064dbd482dc4c6bf6b --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_2338_vith64_free8_port25164_launch/launcher_rank4_n116-041-023.pid @@ -0,0 +1 @@ +1382631 diff --git a/ScaleDF_rebuttal/_host_logs/20260726_2338_vith64_free8_port25164_launch/launcher_rank5_n116-040-213.log b/ScaleDF_rebuttal/_host_logs/20260726_2338_vith64_free8_port25164_launch/launcher_rank5_n116-040-213.log new file mode 100644 index 0000000000000000000000000000000000000000..efbd0299e2baebc49ed1f29a6e10deb70e77fb06 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_2338_vith64_free8_port25164_launch/launcher_rank5_n116-040-213.log @@ -0,0 +1,113 @@ +started_utc=2026-07-26T23:38:12Z +hostname=n116-040-213 +node_rank=5 +nnodes=8 +nproc_per_node=8 +world_size=64 +master_addr=10.116.40.168 +master_port=25164 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +main=main_cross_2direct_balance_realscale_perturb_v4_accum.py +model_name=deit_huge_patch14_LS +pretrain=/tmp/ScaleDF_repro_20260724/pretrained/deit_3_huge_224_21k_v1.pth +resume=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_32card_effbs2048_from_latest24_master_n116-042-185_20260726_1603_vith32_stub_port25003/checkpoint.pth +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_64card_effbs4096_from_epoch22_free8_master_n116-040-168_20260726_2338_vith64_free8_port25164 +batch_per_gpu=64 +accum_iter=1 +effective_batch=4096 +eval_batch_per_gpu=64 +epochs=40 +W0726 23:38:26.271000 1384774 torch/distributed/run.py:766] +W0726 23:38:26.271000 1384774 torch/distributed/run.py:766] ***************************************** +W0726 23:38:26.271000 1384774 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 23:38:26.271000 1384774 torch/distributed/run.py:766] ***************************************** +/usr/bin/python3: can't open file '/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py': [Errno 2] No such file or directory +/usr/bin/python3: can't open file '/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py': [Errno 2] No such file or directory +/usr/bin/python3: can't open file '/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py': [Errno 2] No such file or directory +/usr/bin/python3: can't open file '/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py': [Errno 2] No such file or directory +/usr/bin/python3: can't open file '/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py': [Errno 2] No such file or directory +/usr/bin/python3: can't open file '/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py': [Errno 2] No such file or directory +/usr/bin/python3: can't open file '/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py': [Errno 2] No such file or directory +/usr/bin/python3: can't open file '/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py': [Errno 2] No such file or directory +E0726 23:38:27.564000 1384774 torch/distributed/elastic/multiprocessing/api.py:874] failed (exitcode: 2) local_rank: 0 (pid: 1384839) of binary: /usr/bin/python3 +Traceback (most recent call last): + File "/usr/local/bin/torchrun", line 7, in + sys.exit(main()) + ^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 355, in wrapper + return f(*args, **kwargs) + ^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 892, in main + run(args) + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 883, in run + elastic_launch( + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 139, in __call__ + return launch_agent(self._config, self._entrypoint, list(args)) + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 270, in launch_agent + raise ChildFailedError( +torch.distributed.elastic.multiprocessing.errors.ChildFailedError: +============================================================ +main_cross_2direct_balance_realscale_perturb_v4_accum.py FAILED +------------------------------------------------------------ +Failures: +[1]: + time : 2026-07-26_23:38:27 + host : n116-040-213.byted.org + rank : 41 (local_rank: 1) + exitcode : 2 (pid: 1384840) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +[2]: + time : 2026-07-26_23:38:27 + host : n116-040-213.byted.org + rank : 42 (local_rank: 2) + exitcode : 2 (pid: 1384841) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +[3]: + time : 2026-07-26_23:38:27 + host : n116-040-213.byted.org + rank : 43 (local_rank: 3) + exitcode : 2 (pid: 1384842) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +[4]: + time : 2026-07-26_23:38:27 + host : n116-040-213.byted.org + rank : 44 (local_rank: 4) + exitcode : 2 (pid: 1384843) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +[5]: + time : 2026-07-26_23:38:27 + host : n116-040-213.byted.org + rank : 45 (local_rank: 5) + exitcode : 2 (pid: 1384844) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +[6]: + time : 2026-07-26_23:38:27 + host : n116-040-213.byted.org + rank : 46 (local_rank: 6) + exitcode : 2 (pid: 1384845) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +[7]: + time : 2026-07-26_23:38:27 + host : n116-040-213.byted.org + rank : 47 (local_rank: 7) + exitcode : 2 (pid: 1384846) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +------------------------------------------------------------ +Root Cause (first observed failure): +[0]: + time : 2026-07-26_23:38:27 + host : n116-040-213.byted.org + rank : 40 (local_rank: 0) + exitcode : 2 (pid: 1384839) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +============================================================ diff --git a/ScaleDF_rebuttal/_host_logs/20260726_2338_vith64_free8_port25164_launch/launcher_rank5_n116-040-213.pid b/ScaleDF_rebuttal/_host_logs/20260726_2338_vith64_free8_port25164_launch/launcher_rank5_n116-040-213.pid new file mode 100644 index 0000000000000000000000000000000000000000..6d3d7bb745c970fb665d7a676304757f5a83b006 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_2338_vith64_free8_port25164_launch/launcher_rank5_n116-040-213.pid @@ -0,0 +1 @@ +1384768 diff --git a/ScaleDF_rebuttal/_host_logs/20260726_2338_vith64_free8_port25164_launch/launcher_rank6_n116-041-181.log b/ScaleDF_rebuttal/_host_logs/20260726_2338_vith64_free8_port25164_launch/launcher_rank6_n116-041-181.log new file mode 100644 index 0000000000000000000000000000000000000000..c0134a231bb6e1bac22a2aa5d354b12067a7a1b9 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_2338_vith64_free8_port25164_launch/launcher_rank6_n116-041-181.log @@ -0,0 +1,419 @@ +started_utc=2026-07-26T23:38:13Z +hostname=n116-041-181 +node_rank=6 +nnodes=8 +nproc_per_node=8 +world_size=64 +master_addr=10.116.40.168 +master_port=25164 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +main=main_cross_2direct_balance_realscale_perturb_v4_accum.py +model_name=deit_huge_patch14_LS +pretrain=/tmp/ScaleDF_repro_20260724/pretrained/deit_3_huge_224_21k_v1.pth +resume=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_32card_effbs2048_from_latest24_master_n116-042-185_20260726_1603_vith32_stub_port25003/checkpoint.pth +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_64card_effbs4096_from_epoch22_free8_master_n116-040-168_20260726_2338_vith64_free8_port25164 +batch_per_gpu=64 +accum_iter=1 +effective_batch=4096 +eval_batch_per_gpu=64 +epochs=40 +W0726 23:38:25.912000 72410 torch/distributed/run.py:766] +W0726 23:38:25.912000 72410 torch/distributed/run.py:766] ***************************************** +W0726 23:38:25.912000 72410 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 23:38:25.912000 72410 torch/distributed/run.py:766] ***************************************** +Traceback (most recent call last): + File "/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py", line 13, in + from timm.data import Mixup + File "/home/tiger/.local/lib/python3.11/site-packages/timm/__init__.py", line 2, in +Traceback (most recent call last): + File "/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py", line 13, in +Traceback (most recent call last): + File "/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py", line 13, in +Traceback (most recent call last): + from timm.data import Mixup + from timm.data import Mixup + File "/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py", line 13, in +Traceback (most recent call last): + File "/home/tiger/.local/lib/python3.11/site-packages/timm/__init__.py", line 2, in + File "/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py", line 13, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/__init__.py", line 2, in + Traceback (most recent call last): +from timm.data import Mixup + from timm.data import Mixup File "/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py", line 13, in + +Traceback (most recent call last): + File "/home/tiger/.local/lib/python3.11/site-packages/timm/__init__.py", line 2, in +Traceback (most recent call last): + File "/home/tiger/.local/lib/python3.11/site-packages/timm/__init__.py", line 2, in + File "/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py", line 13, in + File "/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_accum.py", line 13, in + from timm.data import Mixup + File "/home/tiger/.local/lib/python3.11/site-packages/timm/__init__.py", line 2, in + from timm.data import Mixup + File "/home/tiger/.local/lib/python3.11/site-packages/timm/__init__.py", line 2, in +from timm.data import Mixup + File "/home/tiger/.local/lib/python3.11/site-packages/timm/__init__.py", line 2, in + from .models import create_model, list_models, is_model, list_modules, model_entrypoint, \ + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/__init__.py", line 1, in + from .models import create_model, list_models, is_model, list_modules, model_entrypoint, \ + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/__init__.py", line 1, in + from .models import create_model, list_models, is_model, list_modules, model_entrypoint, \ + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/__init__.py", line 1, in + from .models import create_model, list_models, is_model, list_modules, model_entrypoint, \ + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/__init__.py", line 1, in + from .models import create_model, list_models, is_model, list_modules, model_entrypoint, \ + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/__init__.py", line 1, in + from .models import create_model, list_models, is_model, list_modules, model_entrypoint, \ + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/__init__.py", line 1, in + from .models import create_model, list_models, is_model, list_modules, model_entrypoint, \ + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/__init__.py", line 1, in + from .models import create_model, list_models, is_model, list_modules, model_entrypoint, \ + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/__init__.py", line 1, in + from .byoanet import * + from .byoanet import * + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/byoanet.py", line 15, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/byoanet.py", line 15, in + from .byoanet import * + from .byoanet import * + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/byoanet.py", line 15, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/byoanet.py", line 15, in + from .byoanet import * + from .byoanet import * + from .byoanet import * + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/byoanet.py", line 15, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/byoanet.py", line 15, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/byoanet.py", line 15, in + from .byoanet import * + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/byoanet.py", line 15, in + from timm.data import IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STD + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/__init__.py", line 5, in + from timm.data import IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STD +from timm.data import IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STD + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/__init__.py", line 5, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/__init__.py", line 5, in + from timm.data import IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STD + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/__init__.py", line 5, in + from timm.data import IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STD + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/__init__.py", line 5, in + from timm.data import IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STD + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/__init__.py", line 5, in + from timm.data import IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STD + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/__init__.py", line 5, in + from timm.data import IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STD + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/__init__.py", line 5, in + from .dataset import ImageDataset, IterableImageDataset, AugMixDataset + from .dataset import ImageDataset, IterableImageDataset, AugMixDataset + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/dataset.py", line 12, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/dataset.py", line 12, in + from .dataset import ImageDataset, IterableImageDataset, AugMixDataset + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/dataset.py", line 12, in +from .dataset import ImageDataset, IterableImageDataset, AugMixDataset + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/dataset.py", line 12, in + from .dataset import ImageDataset, IterableImageDataset, AugMixDataset + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/dataset.py", line 12, in + from .dataset import ImageDataset, IterableImageDataset, AugMixDataset +from .dataset import ImageDataset, IterableImageDataset, AugMixDataset + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/dataset.py", line 12, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/dataset.py", line 12, in + from .dataset import ImageDataset, IterableImageDataset, AugMixDataset + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/dataset.py", line 12, in + from .parsers import create_parser + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/__init__.py", line 1, in + from .parsers import create_parser + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/__init__.py", line 1, in + from .parsers import create_parser + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/__init__.py", line 1, in + from .parsers import create_parser + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/__init__.py", line 1, in + from .parsers import create_parser + from .parsers import create_parser + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/__init__.py", line 1, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/__init__.py", line 1, in + from .parsers import create_parser + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/__init__.py", line 1, in + from .parsers import create_parser + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/__init__.py", line 1, in + from .parser_factory import create_parser + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_factory.py", line 3, in + from .parser_factory import create_parser + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_factory.py", line 3, in + from .parser_factory import create_parser + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_factory.py", line 3, in + from .parser_factory import create_parser + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_factory.py", line 3, in + from .parser_factory import create_parser + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_factory.py", line 3, in + from .parser_factory import create_parser + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_factory.py", line 3, in + from .parser_factory import create_parser + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_factory.py", line 3, in + from .parser_factory import create_parser + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_factory.py", line 3, in + from .parser_image_folder import ParserImageFolder + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_image_folder.py", line 10, in +from .parser_image_folder import ParserImageFolder + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_image_folder.py", line 10, in + from .parser_image_folder import ParserImageFolder + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_image_folder.py", line 10, in + from .parser_image_folder import ParserImageFolder + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_image_folder.py", line 10, in + from .parser_image_folder import ParserImageFolder + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_image_folder.py", line 10, in + from .parser_image_folder import ParserImageFolder +from .parser_image_folder import ParserImageFolder + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_image_folder.py", line 10, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_image_folder.py", line 10, in + from timm.utils.misc import natural_key + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/__init__.py", line 13, in + from timm.utils.misc import natural_key + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/__init__.py", line 13, in + from .parser_image_folder import ParserImageFolder + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_image_folder.py", line 10, in + from timm.utils.misc import natural_key + from timm.utils.misc import natural_key + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/__init__.py", line 13, in + from timm.utils.misc import natural_key + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/__init__.py", line 13, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/__init__.py", line 13, in + from timm.utils.misc import natural_key + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/__init__.py", line 13, in + from timm.utils.misc import natural_key + from .summary import update_summary, get_outdir + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/__init__.py", line 13, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/summary.py", line 9, in + from timm.utils.misc import natural_key + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/__init__.py", line 13, in + from .summary import update_summary, get_outdir + from .summary import update_summary, get_outdir File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/summary.py", line 9, in + + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/summary.py", line 9, in + from .summary import update_summary, get_outdir + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/summary.py", line 9, in + from .summary import update_summary, get_outdir + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/summary.py", line 9, in + from .summary import update_summary, get_outdir + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/summary.py", line 9, in + import wandb + from .summary import update_summary, get_outdir + File "/usr/local/lib/python3.11/dist-packages/wandb/__init__.py", line 35, in + import wandb + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/summary.py", line 9, in + File "/usr/local/lib/python3.11/dist-packages/wandb/__init__.py", line 35, in + from .summary import update_summary, get_outdir + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/summary.py", line 9, in + import wandb + File "/usr/local/lib/python3.11/dist-packages/wandb/__init__.py", line 35, in + import wandb +import wandb + File "/usr/local/lib/python3.11/dist-packages/wandb/__init__.py", line 35, in + File "/usr/local/lib/python3.11/dist-packages/wandb/__init__.py", line 35, in + import wandb + File "/usr/local/lib/python3.11/dist-packages/wandb/__init__.py", line 35, in + from wandb import sdk as wandb_sdk + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/__init__.py", line 9, in + import wandb + import wandb + File "/usr/local/lib/python3.11/dist-packages/wandb/__init__.py", line 35, in + File "/usr/local/lib/python3.11/dist-packages/wandb/__init__.py", line 35, in + from wandb import sdk as wandb_sdk +from wandb import sdk as wandb_sdk + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/__init__.py", line 9, in +from wandb import sdk as wandb_sdk + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/__init__.py", line 9, in +from wandb import sdk as wandb_sdk + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/__init__.py", line 9, in + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/__init__.py", line 9, in + from wandb import sdk as wandb_sdk + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/__init__.py", line 9, in +from .wandb_init import _attach, init # noqa: F401 + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/wandb_init.py", line 31, in + from wandb import sdk as wandb_sdk + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/__init__.py", line 9, in + from .wandb_init import _attach, init # noqa: F401 + from .wandb_init import _attach, init # noqa: F401 +^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ ^^ ^ ^^^^from wandb import sdk as wandb_sdk^^ + +^^^^^^^^^^^ File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/wandb_init.py", line 31, in +^^^^^^^^^^^^^^^^^^^^^^^ + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/wandb_init.py", line 31, in + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/__init__.py", line 9, in + from .wandb_init import _attach, init # noqa: F401 + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/wandb_init.py", line 31, in + from .wandb_init import _attach, init # noqa: F401 + ^^^^^^^^^^^ ^from .wandb_init import _attach, init # noqa: F401^^ +^^^^^^^^^^^^^^^^^^^^ ^ ^ ^^ +^^^^^^^ File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/wandb_init.py", line 31, in +^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/wandb_init.py", line 31, in + from .backend.backend import Backend + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/backend/backend.py", line 17, in + from .backend.backend import Backend + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/backend/backend.py", line 17, in + from .backend.backend import Backend + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/backend/backend.py", line 17, in + from .wandb_init import _attach, init # noqa: F401 + ^^^^^^^^^^^^^ ^^from .backend.backend import Backend^ +^^^^^^^^^^^^^^^^^^^^^ + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/backend/backend.py", line 17, in + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/wandb_init.py", line 31, in +from .wandb_init import _attach, init # noqa: F401 + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/wandb_init.py", line 31, in + from .backend.backend import Backend + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/backend/backend.py", line 17, in + from .backend.backend import Backend + from ..interface.interface import InterfaceBase + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/backend/backend.py", line 17, in + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/interface/interface.py", line 34, in + from .backend.backend import Backend + from .backend.backend import Backend + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/backend/backend.py", line 17, in + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/backend/backend.py", line 17, in + from ..interface.interface import InterfaceBase + from ..interface.interface import InterfaceBase + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/interface/interface.py", line 34, in + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/interface/interface.py", line 34, in + from ..interface.interface import InterfaceBase + from ..interface.interface import InterfaceBase + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/interface/interface.py", line 34, in + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/interface/interface.py", line 34, in + from ..interface.interface import InterfaceBase + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/interface/interface.py", line 34, in + from ..interface.interface import InterfaceBase + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/interface/interface.py", line 34, in + from ..interface.interface import InterfaceBase + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/interface/interface.py", line 34, in + from ..lib.mailbox import MailboxHandle + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 92, in + from ..lib.mailbox import MailboxHandle + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 92, in + from ..lib.mailbox import MailboxHandle + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 92, in + from ..lib.mailbox import MailboxHandle + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 92, in + from ..lib.mailbox import MailboxHandle + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 92, in + from ..lib.mailbox import MailboxHandle + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 92, in + from ..lib.mailbox import MailboxHandle + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 92, in + from ..lib.mailbox import MailboxHandle + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 92, in + class _MailboxSlot: + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 93, in _MailboxSlot + class _MailboxSlot: + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 93, in _MailboxSlot + class _MailboxSlot: + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 93, in _MailboxSlot + class _MailboxSlot: + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 93, in _MailboxSlot + class _MailboxSlot: + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 93, in _MailboxSlot + class _MailboxSlot: + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 93, in _MailboxSlot + class _MailboxSlot: + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 93, in _MailboxSlot + class _MailboxSlot: + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 93, in _MailboxSlot + _result: Optional[pb.Result] + ^^^^^^^^^ +AttributeError _result: Optional[pb.Result] +: module 'wandb.proto.wandb_internal_pb2' has no attribute 'Result' + ^^^^^^^^^ +AttributeError: module 'wandb.proto.wandb_internal_pb2' has no attribute 'Result' + _result: Optional[pb.Result] + ^^^^^^^^^ +AttributeError : _result: Optional[pb.Result]module 'wandb.proto.wandb_internal_pb2' has no attribute 'Result' + + ^^^^^^^ ^^_result: Optional[pb.Result] + +AttributeError: module 'wandb.proto.wandb_internal_pb2' has no attribute 'Result' + ^^^^^^^^^ +AttributeError: module 'wandb.proto.wandb_internal_pb2' has no attribute 'Result' + _result: Optional[pb.Result]_result: Optional[pb.Result] + + _result: Optional[pb.Result] + ^ ^ ^ ^^ ^ ^^ ^^^^ + ^AttributeError ^: ^module 'wandb.proto.wandb_internal_pb2' has no attribute 'Result' ^ +^ ^ + AttributeError : module 'wandb.proto.wandb_internal_pb2' has no attribute 'Result' + ^^^^^^^^^ +AttributeError: module 'wandb.proto.wandb_internal_pb2' has no attribute 'Result' +W0726 23:38:40.169000 72410 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 72478 closing signal SIGTERM +E0726 23:38:40.283000 72410 torch/distributed/elastic/multiprocessing/api.py:874] failed (exitcode: 1) local_rank: 0 (pid: 72475) of binary: /usr/bin/python3 +Traceback (most recent call last): + File "/usr/local/bin/torchrun", line 7, in + sys.exit(main()) + ^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 355, in wrapper + return f(*args, **kwargs) + ^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 892, in main + run(args) + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 883, in run + elastic_launch( + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 139, in __call__ + return launch_agent(self._config, self._entrypoint, list(args)) + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 270, in launch_agent + raise ChildFailedError( +torch.distributed.elastic.multiprocessing.errors.ChildFailedError: +============================================================ +main_cross_2direct_balance_realscale_perturb_v4_accum.py FAILED +------------------------------------------------------------ +Failures: +[1]: + time : 2026-07-26_23:38:40 + host : n116-041-181.byted.org + rank : 49 (local_rank: 1) + exitcode : 1 (pid: 72476) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +[2]: + time : 2026-07-26_23:38:40 + host : n116-041-181.byted.org + rank : 50 (local_rank: 2) + exitcode : 1 (pid: 72477) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +[3]: + time : 2026-07-26_23:38:40 + host : n116-041-181.byted.org + rank : 52 (local_rank: 4) + exitcode : 1 (pid: 72479) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +[4]: + time : 2026-07-26_23:38:40 + host : n116-041-181.byted.org + rank : 53 (local_rank: 5) + exitcode : 1 (pid: 72480) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +[5]: + time : 2026-07-26_23:38:40 + host : n116-041-181.byted.org + rank : 54 (local_rank: 6) + exitcode : 1 (pid: 72481) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +[6]: + time : 2026-07-26_23:38:40 + host : n116-041-181.byted.org + rank : 55 (local_rank: 7) + exitcode : 1 (pid: 72482) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +------------------------------------------------------------ +Root Cause (first observed failure): +[0]: + time : 2026-07-26_23:38:40 + host : n116-041-181.byted.org + rank : 48 (local_rank: 0) + exitcode : 1 (pid: 72475) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +============================================================ diff --git a/ScaleDF_rebuttal/_host_logs/20260726_2338_vith64_free8_port25164_launch/launcher_rank6_n116-041-181.pid b/ScaleDF_rebuttal/_host_logs/20260726_2338_vith64_free8_port25164_launch/launcher_rank6_n116-041-181.pid new file mode 100644 index 0000000000000000000000000000000000000000..a263ce26cd2dfeb940e77274cb51bb0f2117ff6d --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_2338_vith64_free8_port25164_launch/launcher_rank6_n116-041-181.pid @@ -0,0 +1 @@ +72404 diff --git a/ScaleDF_rebuttal/_host_logs/20260726_2338_vith64_free8_port25164_launch/launcher_rank7_n116-042-235.log b/ScaleDF_rebuttal/_host_logs/20260726_2338_vith64_free8_port25164_launch/launcher_rank7_n116-042-235.log new file mode 100644 index 0000000000000000000000000000000000000000..e322375d35e49cf986c5aa124e2ef60e6bb6dd0a --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_2338_vith64_free8_port25164_launch/launcher_rank7_n116-042-235.log @@ -0,0 +1,33 @@ +started_utc=2026-07-26T23:38:14Z +hostname=n116-042-235 +node_rank=7 +nnodes=8 +nproc_per_node=8 +world_size=64 +master_addr=10.116.40.168 +master_port=25164 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +main=main_cross_2direct_balance_realscale_perturb_v4_accum.py +model_name=deit_huge_patch14_LS +pretrain=/tmp/ScaleDF_repro_20260724/pretrained/deit_3_huge_224_21k_v1.pth +resume=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_32card_effbs2048_from_latest24_master_n116-042-185_20260726_1603_vith32_stub_port25003/checkpoint.pth +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_64card_effbs4096_from_epoch22_free8_master_n116-040-168_20260726_2338_vith64_free8_port25164 +batch_per_gpu=64 +accum_iter=1 +effective_batch=4096 +eval_batch_per_gpu=64 +epochs=40 +W0726 23:38:27.181000 5398 torch/distributed/run.py:766] +W0726 23:38:27.181000 5398 torch/distributed/run.py:766] ***************************************** +W0726 23:38:27.181000 5398 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 23:38:27.181000 5398 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 58): env:// +| distributed init (rank 56): env:// +| distributed init (rank 60): env:// +| distributed init (rank 63): env:// +| distributed init (rank 59): env:// +| distributed init (rank 62): env:// +| distributed init (rank 61): env:// +| distributed init (rank 57): env:// +Terminated diff --git a/ScaleDF_rebuttal/_host_logs/20260726_2338_vith64_free8_port25164_launch/launcher_rank7_n116-042-235.pid b/ScaleDF_rebuttal/_host_logs/20260726_2338_vith64_free8_port25164_launch/launcher_rank7_n116-042-235.pid new file mode 100644 index 0000000000000000000000000000000000000000..63bee27e64bda97e8dcaf3416b2dcca632a12870 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_2338_vith64_free8_port25164_launch/launcher_rank7_n116-042-235.pid @@ -0,0 +1 @@ +5392 diff --git a/ScaleDF_rebuttal/_host_logs/20260726_2338_vith64_free8_port25164_launch/output_root.txt b/ScaleDF_rebuttal/_host_logs/20260726_2338_vith64_free8_port25164_launch/output_root.txt new file mode 100644 index 0000000000000000000000000000000000000000..9730838c90c854149c8991c3deb435244bd83c2c --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_2338_vith64_free8_port25164_launch/output_root.txt @@ -0,0 +1 @@ +/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_64card_effbs4096_from_epoch22_free8_master_n116-040-168_20260726_2338_vith64_free8_port25164 diff --git a/ScaleDF_rebuttal/_host_logs/20260726_2350_vith64_free8_port25264_launch/launch_meta.txt b/ScaleDF_rebuttal/_host_logs/20260726_2350_vith64_free8_port25264_launch/launch_meta.txt new file mode 100644 index 0000000000000000000000000000000000000000..c99fb9fb692f0f4fd207d4d437ee7c8c4ad59256 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_2350_vith64_free8_port25264_launch/launch_meta.txt @@ -0,0 +1,13 @@ +started_utc=2026-07-26T23:50:24Z +run_name=modelsize_vith_40epoch_64card_effbs4096_from_epoch22_snapshot_free8_master_n116-040-168_20260726_2350_vith64_port25264 +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_64card_effbs4096_from_epoch22_snapshot_free8_master_n116-040-168_20260726_2350_vith64_port25264 +log_dir=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/_host_logs/20260726_2350_vith64_free8_port25264_launch +resume=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/checkpoint_snapshots/vith32_latest_for_64card_effbs4096_20260726_234604.pth +nnodes=8 +nproc_per_node=8 +batch_per_gpu=64 +accum_iter=1 +effective_batch=4096 +master_addr=10.116.40.168 +master_port=25264 +old_32card_left_running=true diff --git a/ScaleDF_rebuttal/_host_logs/20260726_2350_vith64_free8_port25264_launch/launcher_rank0_n116-040-168.log b/ScaleDF_rebuttal/_host_logs/20260726_2350_vith64_free8_port25264_launch/launcher_rank0_n116-040-168.log new file mode 100644 index 0000000000000000000000000000000000000000..79ff2a65a1d687d953254a9ddad4bfd3205e1902 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_2350_vith64_free8_port25264_launch/launcher_rank0_n116-040-168.log @@ -0,0 +1,6940 @@ +started_utc=2026-07-26T23:50:25Z +hostname=n116-040-168 +node_rank=0 +nnodes=8 +nproc_per_node=8 +world_size=64 +master_addr=10.116.40.168 +master_port=25264 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +main=main_cross_2direct_balance_realscale_perturb_v4_accum.py +model_name=deit_huge_patch14_LS +pretrain=/tmp/ScaleDF_repro_20260724/pretrained/deit_3_huge_224_21k_v1.pth +resume=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/checkpoint_snapshots/vith32_latest_for_64card_effbs4096_20260726_234604.pth +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_64card_effbs4096_from_epoch22_snapshot_free8_master_n116-040-168_20260726_2350_vith64_port25264 +batch_per_gpu=64 +accum_iter=1 +effective_batch=4096 +eval_batch_per_gpu=64 +epochs=40 +W0726 23:50:40.095000 3350668 torch/distributed/run.py:766] +W0726 23:50:40.095000 3350668 torch/distributed/run.py:766] ***************************************** +W0726 23:50:40.095000 3350668 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 23:50:40.095000 3350668 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +| distributed init (rank 5): env:// +[rank5]:[W726 23:51:14.511944565 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +[rank7]:[W726 23:51:14.547611710 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +[rank0]:[W726 23:51:14.105011636 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank6]:[W726 23:51:14.106613380 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +n116-040-168:3350746:3350746 [0] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-040-168:3350746:3350746 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-040-168:3350746:3350746 [0] NCCL INFO cudaDriverVersion 12060 +n116-040-168:3350746:3350746 [0] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-040-168:3350746:3350746 [0] NCCL INFO Comm config Blocking set to 1 +| distributed init (rank 4): env:// +[rank4]:[W726 23:51:15.351182554 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +[rank2]:[W726 23:51:15.373646435 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +n116-040-168:3350752:3350752 [6] NCCL INFO cudaDriverVersion 12060 +n116-040-168:3350752:3350752 [6] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-040-168:3350752:3350752 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-040-168:3350752:3350752 [6] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-040-168:3350752:3350752 [6] NCCL INFO Comm config Blocking set to 1 +n116-040-168:3350753:3350753 [7] NCCL INFO cudaDriverVersion 12060 +n116-040-168:3350753:3350753 [7] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-040-168:3350753:3350753 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-040-168:3350753:3350753 [7] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-040-168:3350753:3350753 [7] NCCL INFO Comm config Blocking set to 1 +n116-040-168:3350751:3350751 [5] NCCL INFO cudaDriverVersion 12060 +n116-040-168:3350751:3350751 [5] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-040-168:3350751:3350751 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-040-168:3350751:3350751 [5] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-040-168:3350748:3350748 [2] NCCL INFO cudaDriverVersion 12060 +n116-040-168:3350748:3350748 [2] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-040-168:3350748:3350748 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-040-168:3350748:3350748 [2] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-040-168:3350751:3350751 [5] NCCL INFO Comm config Blocking set to 1 +n116-040-168:3350748:3350748 [2] NCCL INFO Comm config Blocking set to 1 +n116-040-168:3350750:3350750 [4] NCCL INFO cudaDriverVersion 12060 +n116-040-168:3350750:3350750 [4] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-040-168:3350750:3350750 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-040-168:3350750:3350750 [4] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-040-168:3350750:3350750 [4] NCCL INFO Comm config Blocking set to 1 +| distributed init (rank 3): env:// +[rank3]:[W726 23:51:15.508374744 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +[rank1]:[W726 23:51:15.528687308 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +n116-040-168:3350749:3350749 [3] NCCL INFO cudaDriverVersion 12060 +n116-040-168:3350749:3350749 [3] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-040-168:3350749:3350749 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-040-168:3350749:3350749 [3] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-040-168:3350749:3350749 [3] NCCL INFO Comm config Blocking set to 1 +n116-040-168:3350747:3350747 [1] NCCL INFO cudaDriverVersion 12060 +n116-040-168:3350747:3350747 [1] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-040-168:3350747:3350747 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-040-168:3350747:3350747 [1] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-040-168:3350747:3350747 [1] NCCL INFO Comm config Blocking set to 1 +n116-040-168:3350746:3350795 [0] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-040-168:3350746:3350795 [0] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-040-168:3350746:3350795 [0] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-040-168:3350746:3350795 [0] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-040-168:3350746:3350795 [0] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-040-168:3350746:3350795 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-040-168:3350746:3350795 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-040-168:3350746:3350795 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-040-168:3350746:3350795 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-040-168:3350746:3350795 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-040-168:3350752:3350800 [6] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-040-168:3350752:3350800 [6] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-040-168:3350752:3350800 [6] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-040-168:3350752:3350800 [6] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-040-168:3350752:3350800 [6] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-040-168:3350752:3350800 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-040-168:3350752:3350800 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-040-168:3350752:3350800 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-040-168:3350752:3350800 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-040-168:3350752:3350800 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-040-168:3350753:3350801 [7] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-040-168:3350753:3350801 [7] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-040-168:3350753:3350801 [7] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-040-168:3350753:3350801 [7] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-040-168:3350753:3350801 [7] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-040-168:3350753:3350801 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-040-168:3350753:3350801 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-040-168:3350753:3350801 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-040-168:3350753:3350801 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-040-168:3350753:3350801 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-040-168:3350748:3350803 [2] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-040-168:3350748:3350803 [2] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-040-168:3350748:3350803 [2] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-040-168:3350748:3350803 [2] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-040-168:3350748:3350803 [2] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-040-168:3350748:3350803 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-040-168:3350748:3350803 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-040-168:3350748:3350803 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-040-168:3350748:3350803 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-040-168:3350748:3350803 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-040-168:3350746:3350795 [0] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-040-168:3350746:3350795 [0] NCCL INFO Using network FasTrak +n116-040-168:3350751:3350802 [5] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-040-168:3350751:3350802 [5] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-040-168:3350751:3350802 [5] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-040-168:3350751:3350802 [5] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-040-168:3350751:3350802 [5] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-040-168:3350751:3350802 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-040-168:3350751:3350802 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-040-168:3350751:3350802 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-040-168:3350751:3350802 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-040-168:3350751:3350802 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-040-168:3350750:3350804 [4] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-040-168:3350750:3350804 [4] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-040-168:3350750:3350804 [4] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-040-168:3350750:3350804 [4] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-040-168:3350750:3350804 [4] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-040-168:3350750:3350804 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-040-168:3350750:3350804 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-040-168:3350750:3350804 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-040-168:3350750:3350804 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-040-168:3350750:3350804 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-040-168:3350752:3350800 [6] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-040-168:3350752:3350800 [6] NCCL INFO Using network FasTrak +n116-040-168:3350746:3350795 [0] NCCL INFO DMA-BUF is available on GPU device 0 +n116-040-168:3350746:3350795 [0] NCCL INFO ncclCommInitRankConfig comm 0xb583d70 rank 0 nranks 64 cudaDev 0 nvmlDev 0 busId 4000 commId 0xfb2d11625d05007a - Init START +n116-040-168:3350753:3350801 [7] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-040-168:3350753:3350801 [7] NCCL INFO Using network FasTrak +n116-040-168:3350748:3350803 [2] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-040-168:3350748:3350803 [2] NCCL INFO Using network FasTrak +n116-040-168:3350751:3350802 [5] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-040-168:3350751:3350802 [5] NCCL INFO Using network FasTrak +n116-040-168:3350750:3350804 [4] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-040-168:3350750:3350804 [4] NCCL INFO Using network FasTrak +n116-040-168:3350749:3350809 [3] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-040-168:3350749:3350809 [3] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-040-168:3350749:3350809 [3] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-040-168:3350749:3350809 [3] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-040-168:3350749:3350809 [3] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-040-168:3350749:3350809 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-040-168:3350749:3350809 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-040-168:3350749:3350809 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-040-168:3350749:3350809 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-040-168:3350749:3350809 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-040-168:3350747:3350810 [1] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-040-168:3350747:3350810 [1] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-040-168:3350747:3350810 [1] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-040-168:3350747:3350810 [1] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-040-168:3350747:3350810 [1] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-040-168:3350747:3350810 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-040-168:3350747:3350810 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-040-168:3350747:3350810 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-040-168:3350747:3350810 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-040-168:3350747:3350810 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-040-168:3350749:3350809 [3] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-040-168:3350749:3350809 [3] NCCL INFO Using network FasTrak +n116-040-168:3350747:3350810 [1] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-040-168:3350747:3350810 [1] NCCL INFO Using network FasTrak +n116-040-168:3350752:3350800 [6] NCCL INFO DMA-BUF is available on GPU device 6 +n116-040-168:3350752:3350800 [6] NCCL INFO ncclCommInitRankConfig comm 0xb33fce0 rank 6 nranks 64 cudaDev 6 nvmlDev 6 busId 8b000 commId 0xfb2d11625d05007a - Init START +n116-040-168:3350753:3350801 [7] NCCL INFO DMA-BUF is available on GPU device 7 +n116-040-168:3350753:3350801 [7] NCCL INFO ncclCommInitRankConfig comm 0xccbc9b0 rank 7 nranks 64 cudaDev 7 nvmlDev 7 busId 8c000 commId 0xfb2d11625d05007a - Init START +n116-040-168:3350753:3350801 [7] NCCL INFO RAS client listening socket at ::1<28028> +n116-040-168:3350748:3350803 [2] NCCL INFO DMA-BUF is available on GPU device 2 +n116-040-168:3350751:3350802 [5] NCCL INFO DMA-BUF is available on GPU device 5 +n116-040-168:3350748:3350803 [2] NCCL INFO ncclCommInitRankConfig comm 0xbb733d0 rank 2 nranks 64 cudaDev 2 nvmlDev 2 busId b000 commId 0xfb2d11625d05007a - Init START +n116-040-168:3350751:3350802 [5] NCCL INFO ncclCommInitRankConfig comm 0xccadb00 rank 5 nranks 64 cudaDev 5 nvmlDev 5 busId 85000 commId 0xfb2d11625d05007a - Init START +n116-040-168:3350752:3350800 [6] NCCL INFO RAS client listening socket at ::1<28028> +n116-040-168:3350750:3350804 [4] NCCL INFO DMA-BUF is available on GPU device 4 +n116-040-168:3350750:3350804 [4] NCCL INFO ncclCommInitRankConfig comm 0xca06ab0 rank 4 nranks 64 cudaDev 4 nvmlDev 4 busId 84000 commId 0xfb2d11625d05007a - Init START +n116-040-168:3350751:3350802 [5] NCCL INFO RAS client listening socket at ::1<28028> +n116-040-168:3350747:3350810 [1] NCCL INFO DMA-BUF is available on GPU device 1 +n116-040-168:3350749:3350809 [3] NCCL INFO DMA-BUF is available on GPU device 3 +n116-040-168:3350749:3350809 [3] NCCL INFO ncclCommInitRankConfig comm 0xbe60d40 rank 3 nranks 64 cudaDev 3 nvmlDev 3 busId c000 commId 0xfb2d11625d05007a - Init START +n116-040-168:3350747:3350810 [1] NCCL INFO ncclCommInitRankConfig comm 0xc967cd0 rank 1 nranks 64 cudaDev 1 nvmlDev 1 busId 5000 commId 0xfb2d11625d05007a - Init START +n116-040-168:3350749:3350809 [3] NCCL INFO RAS client listening socket at ::1<28028> +n116-040-168:3350747:3350810 [1] NCCL INFO RAS client listening socket at ::1<28028> +n116-040-168:3350750:3350804 [4] NCCL INFO RAS client listening socket at ::1<28028> +n116-040-168:3350746:3350795 [0] NCCL INFO RAS client listening socket at ::1<28028> +n116-040-168:3350748:3350803 [2] NCCL INFO RAS client listening socket at ::1<28028> +n116-040-168:3350750:3350804 [4] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-040-168:3350750:3350804 [4] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-040-168:3350750:3350804 [4] NCCL INFO Retrieving state for FasTrak +n116-040-168:3350750:3350804 [4] NCCL INFO Initialized state 0 for FasTrak +n116-040-168:3350750:3350804 [4] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-040-168:3350750:3350804 [4] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-040-168:3350750:3350804 [4] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-040-168:3350750:3350804 [4] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-040-168:3350750:3350804 [4] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-040-168:3350750:3350804 [4] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-040-168:3350750:3350804 [4] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-040-168:3350750:3350804 [4] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-040-168:3350749:3350809 [3] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-040-168:3350753:3350801 [7] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-040-168:3350747:3350810 [1] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-040-168:3350753:3350801 [7] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-040-168:3350749:3350809 [3] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-040-168:3350753:3350801 [7] NCCL INFO Retrieving state for FasTrak +n116-040-168:3350749:3350809 [3] NCCL INFO Retrieving state for FasTrak +n116-040-168:3350751:3350802 [5] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-040-168:3350753:3350801 [7] NCCL INFO Initialized state 0 for FasTrak +n116-040-168:3350749:3350809 [3] NCCL INFO Initialized state 0 for FasTrak +n116-040-168:3350747:3350810 [1] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-040-168:3350747:3350810 [1] NCCL INFO Retrieving state for FasTrak +n116-040-168:3350747:3350810 [1] NCCL INFO Initialized state 0 for FasTrak +n116-040-168:3350751:3350802 [5] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-040-168:3350751:3350802 [5] NCCL INFO Retrieving state for FasTrak +n116-040-168:3350751:3350802 [5] NCCL INFO Initialized state 0 for FasTrak +n116-040-168:3350747:3350810 [1] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-040-168:3350749:3350809 [3] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-040-168:3350753:3350801 [7] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-040-168:3350751:3350802 [5] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-040-168:3350747:3350810 [1] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-040-168:3350749:3350809 [3] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-040-168:3350753:3350801 [7] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-040-168:3350751:3350802 [5] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-040-168:3350749:3350809 [3] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-040-168:3350747:3350810 [1] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-040-168:3350753:3350801 [7] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-040-168:3350749:3350809 [3] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-040-168:3350747:3350810 [1] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-040-168:3350753:3350801 [7] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-040-168:3350751:3350802 [5] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-040-168:3350749:3350809 [3] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-040-168:3350747:3350810 [1] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-040-168:3350753:3350801 [7] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-040-168:3350751:3350802 [5] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-040-168:3350749:3350809 [3] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-040-168:3350753:3350801 [7] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-040-168:3350747:3350810 [1] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-040-168:3350751:3350802 [5] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-040-168:3350749:3350809 [3] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-040-168:3350753:3350801 [7] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-040-168:3350751:3350802 [5] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-040-168:3350747:3350810 [1] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-040-168:3350753:3350801 [7] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-040-168:3350749:3350809 [3] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-040-168:3350747:3350810 [1] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-040-168:3350751:3350802 [5] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-040-168:3350751:3350802 [5] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-040-168:3350752:3350800 [6] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-040-168:3350748:3350803 [2] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-040-168:3350752:3350800 [6] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-040-168:3350752:3350800 [6] NCCL INFO Retrieving state for FasTrak +n116-040-168:3350748:3350803 [2] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-040-168:3350752:3350800 [6] NCCL INFO Initialized state 0 for FasTrak +n116-040-168:3350748:3350803 [2] NCCL INFO Retrieving state for FasTrak +n116-040-168:3350748:3350803 [2] NCCL INFO Initialized state 0 for FasTrak +n116-040-168:3350748:3350803 [2] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-040-168:3350752:3350800 [6] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-040-168:3350748:3350803 [2] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-040-168:3350752:3350800 [6] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-040-168:3350748:3350803 [2] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-040-168:3350752:3350800 [6] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-040-168:3350748:3350803 [2] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-040-168:3350752:3350800 [6] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-040-168:3350748:3350803 [2] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-040-168:3350752:3350800 [6] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-040-168:3350748:3350803 [2] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-040-168:3350752:3350800 [6] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-040-168:3350752:3350800 [6] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-040-168:3350748:3350803 [2] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-040-168:3350752:3350800 [6] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-040-168:3350748:3350803 [2] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-040-168:3350746:3350795 [0] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-040-168:3350746:3350795 [0] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-040-168:3350746:3350795 [0] NCCL INFO Retrieving state for FasTrak +n116-040-168:3350746:3350795 [0] NCCL INFO Initialized state 0 for FasTrak +n116-040-168:3350746:3350795 [0] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-040-168:3350746:3350795 [0] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-040-168:3350746:3350795 [0] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-040-168:3350746:3350795 [0] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-040-168:3350746:3350795 [0] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-040-168:3350746:3350795 [0] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-040-168:3350746:3350795 [0] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-040-168:3350746:3350795 [0] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-040-168:3350747:3350810 [1] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-040-168:3350746:3350795 [0] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-040-168:3350752:3350800 [6] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-040-168:3350751:3350802 [5] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-040-168:3350753:3350801 [7] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-040-168:3350750:3350804 [4] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-040-168:3350748:3350803 [2] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-040-168:3350749:3350809 [3] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-040-168:3350746:3350795 [0] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-040-168:3350746:3350795 [0] NCCL INFO CPU/0-0 (1/1/3) +n116-040-168:3350746:3350795 [0] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-040-168:3350747:3350810 [1] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-040-168:3350746:3350795 [0] NCCL INFO + PCI[24.0] - GPU/0-4000 (0) +n116-040-168:3350746:3350795 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350747:3350810 [1] NCCL INFO CPU/0-0 (1/1/3) +n116-040-168:3350746:3350795 [0] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-040-168:3350746:3350795 [0] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-040-168:3350747:3350810 [1] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-040-168:3350746:3350795 [0] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-040-168:3350747:3350810 [1] NCCL INFO + PCI[24.0] - GPU/0-4000 (0) +n116-040-168:3350746:3350795 [0] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-040-168:3350746:3350795 [0] NCCL INFO + PCI[24.0] - GPU/0-5000 (1) +n116-040-168:3350747:3350810 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350746:3350795 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350747:3350810 [1] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-040-168:3350746:3350795 [0] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-040-168:3350746:3350795 [0] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-040-168:3350746:3350795 [0] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-040-168:3350747:3350810 [1] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-040-168:3350746:3350795 [0] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-040-168:3350747:3350810 [1] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-040-168:3350746:3350795 [0] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-040-168:3350746:3350795 [0] NCCL INFO + PCI[24.0] - GPU/0-b000 (2) +n116-040-168:3350746:3350795 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350747:3350810 [1] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-040-168:3350746:3350795 [0] NCCL INFO + PCI[24.0] - GPU/0-c000 (3) +n116-040-168:3350747:3350810 [1] NCCL INFO + PCI[24.0] - GPU/0-5000 (1) +n116-040-168:3350746:3350795 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350746:3350795 [0] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-040-168:3350747:3350810 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350746:3350795 [0] NCCL INFO CPU/0-1 (1/1/3) +n116-040-168:3350746:3350795 [0] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-040-168:3350747:3350810 [1] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-040-168:3350746:3350795 [0] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-040-168:3350747:3350810 [1] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-040-168:3350746:3350795 [0] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-040-168:3350746:3350795 [0] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-040-168:3350747:3350810 [1] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-040-168:3350746:3350795 [0] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-040-168:3350747:3350810 [1] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-040-168:3350746:3350795 [0] NCCL INFO + PCI[24.0] - GPU/0-84000 (4) +n116-040-168:3350746:3350795 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350747:3350810 [1] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-040-168:3350746:3350795 [0] NCCL INFO + PCI[24.0] - GPU/0-85000 (5) +n116-040-168:3350746:3350795 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350747:3350810 [1] NCCL INFO + PCI[24.0] - GPU/0-b000 (2) +n116-040-168:3350746:3350795 [0] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-040-168:3350746:3350795 [0] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-040-168:3350747:3350810 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350746:3350795 [0] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-040-168:3350747:3350810 [1] NCCL INFO + PCI[24.0] - GPU/0-c000 (3) +n116-040-168:3350746:3350795 [0] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-040-168:3350747:3350810 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350746:3350795 [0] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-040-168:3350746:3350795 [0] NCCL INFO + PCI[24.0] - GPU/0-8b000 (6) +n116-040-168:3350747:3350810 [1] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-040-168:3350746:3350795 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350746:3350795 [0] NCCL INFO + PCI[24.0] - GPU/0-8c000 (7) +n116-040-168:3350747:3350810 [1] NCCL INFO CPU/0-1 (1/1/3) +n116-040-168:3350746:3350795 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350746:3350795 [0] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-040-168:3350747:3350810 [1] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-040-168:3350746:3350795 [0] NCCL INFO ========================================== +n116-040-168:3350747:3350810 [1] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-040-168:3350747:3350810 [1] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-040-168:3350747:3350810 [1] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-040-168:3350747:3350810 [1] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-040-168:3350746:3350795 [0] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-168:3350747:3350810 [1] NCCL INFO + PCI[24.0] - GPU/0-84000 (4) +n116-040-168:3350747:3350810 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350747:3350810 [1] NCCL INFO + PCI[24.0] - GPU/0-85000 (5) +n116-040-168:3350747:3350810 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350747:3350810 [1] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-040-168:3350746:3350795 [0] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-168:3350747:3350810 [1] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-040-168:3350747:3350810 [1] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-040-168:3350746:3350795 [0] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-168:3350747:3350810 [1] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-040-168:3350747:3350810 [1] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-040-168:3350747:3350810 [1] NCCL INFO + PCI[24.0] - GPU/0-8b000 (6) +n116-040-168:3350747:3350810 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350746:3350795 [0] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-168:3350747:3350810 [1] NCCL INFO + PCI[24.0] - GPU/0-8c000 (7) +n116-040-168:3350747:3350810 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350747:3350810 [1] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-040-168:3350747:3350810 [1] NCCL INFO ========================================== +n116-040-168:3350746:3350795 [0] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-168:3350746:3350795 [0] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-168:3350747:3350810 [1] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-168:3350746:3350795 [0] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-040-168:3350746:3350795 [0] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-040-168:3350752:3350800 [6] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-040-168:3350747:3350810 [1] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-168:3350752:3350800 [6] NCCL INFO CPU/0-0 (1/1/3) +n116-040-168:3350752:3350800 [6] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-040-168:3350746:3350795 [0] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-168:3350752:3350800 [6] NCCL INFO + PCI[24.0] - GPU/0-4000 (0) +n116-040-168:3350752:3350800 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350747:3350810 [1] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-168:3350752:3350800 [6] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-040-168:3350746:3350795 [0] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-168:3350752:3350800 [6] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-040-168:3350752:3350800 [6] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-040-168:3350752:3350800 [6] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-040-168:3350752:3350800 [6] NCCL INFO + PCI[24.0] - GPU/0-5000 (1) +n116-040-168:3350752:3350800 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350752:3350800 [6] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-040-168:3350746:3350795 [0] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-168:3350752:3350800 [6] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-040-168:3350752:3350800 [6] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-040-168:3350747:3350810 [1] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-168:3350752:3350800 [6] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-040-168:3350752:3350800 [6] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-040-168:3350746:3350795 [0] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-168:3350752:3350800 [6] NCCL INFO + PCI[24.0] - GPU/0-b000 (2) +n116-040-168:3350752:3350800 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350752:3350800 [6] NCCL INFO + PCI[24.0] - GPU/0-c000 (3) +n116-040-168:3350752:3350800 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350746:3350795 [0] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-040-168:3350752:3350800 [6] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-040-168:3350747:3350810 [1] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-168:3350752:3350800 [6] NCCL INFO CPU/0-1 (1/1/3) +n116-040-168:3350752:3350800 [6] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-040-168:3350752:3350800 [6] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-040-168:3350746:3350795 [0] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-040-168:3350752:3350800 [6] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-040-168:3350752:3350800 [6] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-040-168:3350752:3350800 [6] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-040-168:3350752:3350800 [6] NCCL INFO + PCI[24.0] - GPU/0-84000 (4) +n116-040-168:3350752:3350800 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350752:3350800 [6] NCCL INFO + PCI[24.0] - GPU/0-85000 (5) +n116-040-168:3350746:3350795 [0] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-040-168:3350747:3350810 [1] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-168:3350752:3350800 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350752:3350800 [6] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-040-168:3350752:3350800 [6] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-040-168:3350752:3350800 [6] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-040-168:3350746:3350795 [0] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-040-168:3350752:3350800 [6] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-040-168:3350752:3350800 [6] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-040-168:3350752:3350800 [6] NCCL INFO + PCI[24.0] - GPU/0-8b000 (6) +n116-040-168:3350747:3350810 [1] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-040-168:3350752:3350800 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350746:3350795 [0] NCCL INFO Setting affinity for GPU 0 to 0fffffff,fffffe00,00000000,000fffff,fffffffe +n116-040-168:3350752:3350800 [6] NCCL INFO + PCI[24.0] - GPU/0-8c000 (7) +n116-040-168:3350752:3350800 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350752:3350800 [6] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-040-168:3350752:3350800 [6] NCCL INFO ========================================== +n116-040-168:3350747:3350810 [1] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-040-168:3350752:3350800 [6] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-168:3350752:3350800 [6] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-168:3350747:3350810 [1] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-168:3350752:3350800 [6] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-168:3350747:3350810 [1] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-168:3350752:3350800 [6] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-168:3350747:3350810 [1] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-168:3350752:3350800 [6] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-168:3350746:3350795 [0] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-040-168:3350752:3350800 [6] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-168:3350747:3350810 [1] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-168:3350752:3350800 [6] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-040-168:3350752:3350800 [6] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-040-168:3350747:3350810 [1] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-040-168:3350752:3350800 [6] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-168:3350752:3350800 [6] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-168:3350746:3350795 [0] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-040-168:3350747:3350810 [1] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-040-168:3350752:3350800 [6] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-168:3350752:3350800 [6] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-168:3350747:3350810 [1] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-040-168:3350752:3350800 [6] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-040-168:3350747:3350810 [1] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-040-168:3350747:3350810 [1] NCCL INFO Setting affinity for GPU 1 to 0fffffff,fffffe00,00000000,000fffff,fffffffe +n116-040-168:3350752:3350800 [6] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-040-168:3350752:3350800 [6] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-040-168:3350752:3350800 [6] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-040-168:3350752:3350800 [6] NCCL INFO Setting affinity for GPU 6 to 7fff,ffffffff,e0000000,0000007f,ffffffff,ffe00000,00000000 +n116-040-168:3350752:3350800 [6] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-040-168:3350747:3350810 [1] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-040-168:3350747:3350810 [1] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-040-168:3350752:3350800 [6] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-040-168:3350751:3350802 [5] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-040-168:3350750:3350804 [4] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-040-168:3350750:3350804 [4] NCCL INFO CPU/0-0 (1/1/3) +n116-040-168:3350750:3350804 [4] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-040-168:3350751:3350802 [5] NCCL INFO CPU/0-0 (1/1/3) +n116-040-168:3350750:3350804 [4] NCCL INFO + PCI[24.0] - GPU/0-4000 (0) +n116-040-168:3350750:3350804 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350751:3350802 [5] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-040-168:3350750:3350804 [4] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-040-168:3350751:3350802 [5] NCCL INFO + PCI[24.0] - GPU/0-4000 (0) +n116-040-168:3350750:3350804 [4] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-040-168:3350751:3350802 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350750:3350804 [4] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-040-168:3350751:3350802 [5] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-040-168:3350750:3350804 [4] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-040-168:3350751:3350802 [5] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-040-168:3350750:3350804 [4] NCCL INFO + PCI[24.0] - GPU/0-5000 (1) +n116-040-168:3350751:3350802 [5] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-040-168:3350750:3350804 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350750:3350804 [4] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-040-168:3350751:3350802 [5] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-040-168:3350750:3350804 [4] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-040-168:3350751:3350802 [5] NCCL INFO + PCI[24.0] - GPU/0-5000 (1) +n116-040-168:3350750:3350804 [4] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-040-168:3350751:3350802 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350750:3350804 [4] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-040-168:3350751:3350802 [5] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-040-168:3350750:3350804 [4] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-040-168:3350751:3350802 [5] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-040-168:3350750:3350804 [4] NCCL INFO + PCI[24.0] - GPU/0-b000 (2) +n116-040-168:3350750:3350804 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350751:3350802 [5] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-040-168:3350750:3350804 [4] NCCL INFO + PCI[24.0] - GPU/0-c000 (3) +n116-040-168:3350751:3350802 [5] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-040-168:3350750:3350804 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350751:3350802 [5] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-040-168:3350750:3350804 [4] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-040-168:3350751:3350802 [5] NCCL INFO + PCI[24.0] - GPU/0-b000 (2) +n116-040-168:3350750:3350804 [4] NCCL INFO CPU/0-1 (1/1/3) +n116-040-168:3350751:3350802 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350750:3350804 [4] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-040-168:3350751:3350802 [5] NCCL INFO + PCI[24.0] - GPU/0-c000 (3) +n116-040-168:3350750:3350804 [4] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-040-168:3350751:3350802 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350750:3350804 [4] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-040-168:3350751:3350802 [5] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-040-168:3350750:3350804 [4] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-040-168:3350751:3350802 [5] NCCL INFO CPU/0-1 (1/1/3) +n116-040-168:3350750:3350804 [4] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-040-168:3350751:3350802 [5] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-040-168:3350750:3350804 [4] NCCL INFO + PCI[24.0] - GPU/0-84000 (4) +n116-040-168:3350751:3350802 [5] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-040-168:3350750:3350804 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350751:3350802 [5] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-040-168:3350750:3350804 [4] NCCL INFO + PCI[24.0] - GPU/0-85000 (5) +n116-040-168:3350751:3350802 [5] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-040-168:3350750:3350804 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350751:3350802 [5] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-040-168:3350750:3350804 [4] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-040-168:3350750:3350804 [4] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-040-168:3350751:3350802 [5] NCCL INFO + PCI[24.0] - GPU/0-84000 (4) +n116-040-168:3350750:3350804 [4] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-040-168:3350751:3350802 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350750:3350804 [4] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-040-168:3350750:3350804 [4] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-040-168:3350751:3350802 [5] NCCL INFO + PCI[24.0] - GPU/0-85000 (5) +n116-040-168:3350753:3350801 [7] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-040-168:3350751:3350802 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350750:3350804 [4] NCCL INFO + PCI[24.0] - GPU/0-8b000 (6) +n116-040-168:3350751:3350802 [5] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-040-168:3350753:3350801 [7] NCCL INFO CPU/0-0 (1/1/3) +n116-040-168:3350750:3350804 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350753:3350801 [7] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-040-168:3350751:3350802 [5] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-040-168:3350750:3350804 [4] NCCL INFO + PCI[24.0] - GPU/0-8c000 (7) +n116-040-168:3350753:3350801 [7] NCCL INFO + PCI[24.0] - GPU/0-4000 (0) +n116-040-168:3350751:3350802 [5] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-040-168:3350750:3350804 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350753:3350801 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350750:3350804 [4] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-040-168:3350751:3350802 [5] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-040-168:3350751:3350802 [5] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-040-168:3350750:3350804 [4] NCCL INFO ========================================== +n116-040-168:3350753:3350801 [7] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-040-168:3350751:3350802 [5] NCCL INFO + PCI[24.0] - GPU/0-8b000 (6) +n116-040-168:3350751:3350802 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350753:3350801 [7] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-040-168:3350751:3350802 [5] NCCL INFO + PCI[24.0] - GPU/0-8c000 (7) +n116-040-168:3350753:3350801 [7] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-040-168:3350751:3350802 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350753:3350801 [7] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-040-168:3350751:3350802 [5] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-040-168:3350751:3350802 [5] NCCL INFO ========================================== +n116-040-168:3350753:3350801 [7] NCCL INFO + PCI[24.0] - GPU/0-5000 (1) +n116-040-168:3350753:3350801 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350750:3350804 [4] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-168:3350753:3350801 [7] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-040-168:3350753:3350801 [7] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-040-168:3350753:3350801 [7] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-040-168:3350751:3350802 [5] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-168:3350753:3350801 [7] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-040-168:3350753:3350801 [7] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-040-168:3350750:3350804 [4] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-168:3350753:3350801 [7] NCCL INFO + PCI[24.0] - GPU/0-b000 (2) +n116-040-168:3350753:3350801 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350751:3350802 [5] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-168:3350753:3350801 [7] NCCL INFO + PCI[24.0] - GPU/0-c000 (3) +n116-040-168:3350753:3350801 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350753:3350801 [7] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-040-168:3350750:3350804 [4] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-168:3350753:3350801 [7] NCCL INFO CPU/0-1 (1/1/3) +n116-040-168:3350753:3350801 [7] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-040-168:3350753:3350801 [7] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-040-168:3350751:3350802 [5] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-168:3350753:3350801 [7] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-040-168:3350753:3350801 [7] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-040-168:3350750:3350804 [4] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-168:3350753:3350801 [7] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-040-168:3350753:3350801 [7] NCCL INFO + PCI[24.0] - GPU/0-84000 (4) +n116-040-168:3350751:3350802 [5] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-168:3350753:3350801 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350753:3350801 [7] NCCL INFO + PCI[24.0] - GPU/0-85000 (5) +n116-040-168:3350753:3350801 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350750:3350804 [4] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-168:3350753:3350801 [7] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-040-168:3350753:3350801 [7] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-040-168:3350751:3350802 [5] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-168:3350753:3350801 [7] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-040-168:3350753:3350801 [7] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-040-168:3350750:3350804 [4] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-168:3350753:3350801 [7] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-040-168:3350753:3350801 [7] NCCL INFO + PCI[24.0] - GPU/0-8b000 (6) +n116-040-168:3350753:3350801 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350753:3350801 [7] NCCL INFO + PCI[24.0] - GPU/0-8c000 (7) +n116-040-168:3350751:3350802 [5] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-168:3350753:3350801 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350753:3350801 [7] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-040-168:3350753:3350801 [7] NCCL INFO ========================================== +n116-040-168:3350750:3350804 [4] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-040-168:3350751:3350802 [5] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-040-168:3350750:3350804 [4] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-040-168:3350753:3350801 [7] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-168:3350751:3350802 [5] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-040-168:3350750:3350804 [4] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-168:3350753:3350801 [7] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-168:3350750:3350804 [4] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-168:3350751:3350802 [5] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-168:3350753:3350801 [7] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-168:3350750:3350804 [4] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-168:3350751:3350802 [5] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-168:3350750:3350804 [4] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-168:3350753:3350801 [7] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-168:3350751:3350802 [5] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-168:3350750:3350804 [4] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-040-168:3350753:3350801 [7] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-168:3350751:3350802 [5] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-168:3350750:3350804 [4] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-040-168:3350753:3350801 [7] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-168:3350751:3350802 [5] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-040-168:3350750:3350804 [4] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-040-168:3350751:3350802 [5] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-040-168:3350753:3350801 [7] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-040-168:3350750:3350804 [4] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-040-168:3350749:3350809 [3] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-040-168:3350751:3350802 [5] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-040-168:3350753:3350801 [7] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-040-168:3350750:3350804 [4] NCCL INFO Setting affinity for GPU 4 to 7fff,ffffffff,e0000000,0000007f,ffffffff,ffe00000,00000000 +n116-040-168:3350749:3350809 [3] NCCL INFO CPU/0-0 (1/1/3) +n116-040-168:3350749:3350809 [3] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-040-168:3350749:3350809 [3] NCCL INFO + PCI[24.0] - GPU/0-4000 (0) +n116-040-168:3350749:3350809 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350751:3350802 [5] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-040-168:3350750:3350804 [4] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-040-168:3350749:3350809 [3] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-040-168:3350753:3350801 [7] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-168:3350749:3350809 [3] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-040-168:3350749:3350809 [3] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-040-168:3350749:3350809 [3] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-040-168:3350749:3350809 [3] NCCL INFO + PCI[24.0] - GPU/0-5000 (1) +n116-040-168:3350751:3350802 [5] NCCL INFO Setting affinity for GPU 5 to 7fff,ffffffff,e0000000,0000007f,ffffffff,ffe00000,00000000 +n116-040-168:3350749:3350809 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350753:3350801 [7] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-168:3350749:3350809 [3] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-040-168:3350749:3350809 [3] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-040-168:3350749:3350809 [3] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-040-168:3350749:3350809 [3] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-040-168:3350749:3350809 [3] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-040-168:3350749:3350809 [3] NCCL INFO + PCI[24.0] - GPU/0-b000 (2) +n116-040-168:3350749:3350809 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350753:3350801 [7] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-168:3350749:3350809 [3] NCCL INFO + PCI[24.0] - GPU/0-c000 (3) +n116-040-168:3350749:3350809 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350749:3350809 [3] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-040-168:3350749:3350809 [3] NCCL INFO CPU/0-1 (1/1/3) +n116-040-168:3350749:3350809 [3] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-040-168:3350749:3350809 [3] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-040-168:3350749:3350809 [3] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-040-168:3350753:3350801 [7] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-168:3350749:3350809 [3] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-040-168:3350750:3350804 [4] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-040-168:3350749:3350809 [3] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-040-168:3350749:3350809 [3] NCCL INFO + PCI[24.0] - GPU/0-84000 (4) +n116-040-168:3350749:3350809 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350749:3350809 [3] NCCL INFO + PCI[24.0] - GPU/0-85000 (5) +n116-040-168:3350749:3350809 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350749:3350809 [3] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-040-168:3350753:3350801 [7] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-040-168:3350749:3350809 [3] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-040-168:3350749:3350809 [3] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-040-168:3350749:3350809 [3] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-040-168:3350749:3350809 [3] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-040-168:3350749:3350809 [3] NCCL INFO + PCI[24.0] - GPU/0-8b000 (6) +n116-040-168:3350749:3350809 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350749:3350809 [3] NCCL INFO + PCI[24.0] - GPU/0-8c000 (7) +n116-040-168:3350753:3350801 [7] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-040-168:3350749:3350809 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350749:3350809 [3] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-040-168:3350749:3350809 [3] NCCL INFO ========================================== +n116-040-168:3350753:3350801 [7] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-040-168:3350749:3350809 [3] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-168:3350753:3350801 [7] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-040-168:3350749:3350809 [3] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-168:3350753:3350801 [7] NCCL INFO Setting affinity for GPU 7 to 7fff,ffffffff,e0000000,0000007f,ffffffff,ffe00000,00000000 +n116-040-168:3350751:3350802 [5] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-040-168:3350753:3350801 [7] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-040-168:3350749:3350809 [3] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-168:3350748:3350803 [2] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-040-168:3350748:3350803 [2] NCCL INFO CPU/0-0 (1/1/3) +n116-040-168:3350748:3350803 [2] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-040-168:3350749:3350809 [3] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-168:3350748:3350803 [2] NCCL INFO + PCI[24.0] - GPU/0-4000 (0) +n116-040-168:3350748:3350803 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350748:3350803 [2] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-040-168:3350748:3350803 [2] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-040-168:3350748:3350803 [2] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-040-168:3350748:3350803 [2] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-040-168:3350749:3350809 [3] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-168:3350748:3350803 [2] NCCL INFO + PCI[24.0] - GPU/0-5000 (1) +n116-040-168:3350748:3350803 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350748:3350803 [2] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-040-168:3350748:3350803 [2] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-040-168:3350748:3350803 [2] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-040-168:3350748:3350803 [2] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-040-168:3350748:3350803 [2] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-040-168:3350749:3350809 [3] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-168:3350748:3350803 [2] NCCL INFO + PCI[24.0] - GPU/0-b000 (2) +n116-040-168:3350748:3350803 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350748:3350803 [2] NCCL INFO + PCI[24.0] - GPU/0-c000 (3) +n116-040-168:3350748:3350803 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350748:3350803 [2] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-040-168:3350748:3350803 [2] NCCL INFO CPU/0-1 (1/1/3) +n116-040-168:3350748:3350803 [2] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-040-168:3350748:3350803 [2] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-040-168:3350749:3350809 [3] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-040-168:3350748:3350803 [2] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-040-168:3350748:3350803 [2] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-040-168:3350748:3350803 [2] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-040-168:3350748:3350803 [2] NCCL INFO + PCI[24.0] - GPU/0-84000 (4) +n116-040-168:3350753:3350801 [7] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-040-168:3350749:3350809 [3] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-040-168:3350748:3350803 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350748:3350803 [2] NCCL INFO + PCI[24.0] - GPU/0-85000 (5) +n116-040-168:3350748:3350803 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350751:3350802 [5] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-040-168:3350748:3350803 [2] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-040-168:3350748:3350803 [2] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-040-168:3350749:3350809 [3] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-168:3350748:3350803 [2] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-040-168:3350748:3350803 [2] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-040-168:3350748:3350803 [2] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-040-168:3350748:3350803 [2] NCCL INFO + PCI[24.0] - GPU/0-8b000 (6) +n116-040-168:3350748:3350803 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350748:3350803 [2] NCCL INFO + PCI[24.0] - GPU/0-8c000 (7) +n116-040-168:3350748:3350803 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-168:3350748:3350803 [2] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-040-168:3350749:3350809 [3] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-168:3350748:3350803 [2] NCCL INFO ========================================== +n116-040-168:3350749:3350809 [3] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-168:3350748:3350803 [2] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-168:3350749:3350809 [3] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-168:3350748:3350803 [2] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-168:3350749:3350809 [3] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-040-168:3350748:3350803 [2] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-168:3350749:3350809 [3] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-040-168:3350748:3350803 [2] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-168:3350749:3350809 [3] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-040-168:3350748:3350803 [2] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-168:3350749:3350809 [3] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-040-168:3350748:3350803 [2] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-168:3350749:3350809 [3] NCCL INFO Setting affinity for GPU 3 to 0fffffff,fffffe00,00000000,000fffff,fffffffe +n116-040-168:3350748:3350803 [2] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-040-168:3350748:3350803 [2] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-040-168:3350748:3350803 [2] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-168:3350748:3350803 [2] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-168:3350748:3350803 [2] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-168:3350748:3350803 [2] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-168:3350748:3350803 [2] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-040-168:3350748:3350803 [2] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-040-168:3350748:3350803 [2] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-040-168:3350748:3350803 [2] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-040-168:3350748:3350803 [2] NCCL INFO Setting affinity for GPU 2 to 0fffffff,fffffe00,00000000,000fffff,fffffffe +n116-040-168:3350748:3350803 [2] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-040-168:3350749:3350809 [3] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-040-168:3350748:3350803 [2] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-040-168:3350749:3350809 [3] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-040-168:3350750:3350804 [4] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-040-168:3350750:3350804 [4] NCCL INFO 0 : NET/0-0 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 NET/0-0 +n116-040-168:3350750:3350804 [4] NCCL INFO 1 : NET/0-1 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 NET/0-1 +n116-040-168:3350750:3350804 [4] NCCL INFO 2 : NET/0-2 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 NET/0-2 +n116-040-168:3350750:3350804 [4] NCCL INFO 3 : NET/0-3 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 NET/0-3 +n116-040-168:3350750:3350804 [4] NCCL INFO 4 : NET/0-4 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 NET/0-4 +n116-040-168:3350750:3350804 [4] NCCL INFO 5 : NET/0-5 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 NET/0-5 +n116-040-168:3350750:3350804 [4] NCCL INFO 6 : NET/0-6 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 NET/0-6 +n116-040-168:3350750:3350804 [4] NCCL INFO 7 : NET/0-7 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 NET/0-7 +n116-040-168:3350750:3350804 [4] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-040-168:3350750:3350804 [4] NCCL INFO 0 : NET/0-0 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 NET/0-0 +n116-040-168:3350750:3350804 [4] NCCL INFO 1 : NET/0-1 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 NET/0-1 +n116-040-168:3350750:3350804 [4] NCCL INFO 2 : NET/0-2 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 NET/0-2 +n116-040-168:3350750:3350804 [4] NCCL INFO 3 : NET/0-3 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 NET/0-3 +n116-040-168:3350750:3350804 [4] NCCL INFO 4 : NET/0-4 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 NET/0-4 +n116-040-168:3350750:3350804 [4] NCCL INFO 5 : NET/0-5 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 NET/0-5 +n116-040-168:3350750:3350804 [4] NCCL INFO 6 : NET/0-6 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 NET/0-6 +n116-040-168:3350750:3350804 [4] NCCL INFO 7 : NET/0-7 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 NET/0-7 +n116-040-168:3350747:3350810 [1] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-040-168:3350747:3350810 [1] NCCL INFO 0 : NET/0-0 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 NET/0-0 +n116-040-168:3350747:3350810 [1] NCCL INFO 1 : NET/0-1 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 NET/0-1 +n116-040-168:3350747:3350810 [1] NCCL INFO 2 : NET/0-2 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 NET/0-2 +n116-040-168:3350747:3350810 [1] NCCL INFO 3 : NET/0-3 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 NET/0-3 +n116-040-168:3350747:3350810 [1] NCCL INFO 4 : NET/0-4 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 NET/0-4 +n116-040-168:3350747:3350810 [1] NCCL INFO 5 : NET/0-5 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 NET/0-5 +n116-040-168:3350747:3350810 [1] NCCL INFO 6 : NET/0-6 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 NET/0-6 +n116-040-168:3350747:3350810 [1] NCCL INFO 7 : NET/0-7 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 NET/0-7 +n116-040-168:3350747:3350810 [1] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-040-168:3350747:3350810 [1] NCCL INFO 0 : NET/0-0 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 NET/0-0 +n116-040-168:3350747:3350810 [1] NCCL INFO 1 : NET/0-1 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 NET/0-1 +n116-040-168:3350747:3350810 [1] NCCL INFO 2 : NET/0-2 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 NET/0-2 +n116-040-168:3350747:3350810 [1] NCCL INFO 3 : NET/0-3 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 NET/0-3 +n116-040-168:3350747:3350810 [1] NCCL INFO 4 : NET/0-4 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 NET/0-4 +n116-040-168:3350747:3350810 [1] NCCL INFO 5 : NET/0-5 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 NET/0-5 +n116-040-168:3350747:3350810 [1] NCCL INFO 6 : NET/0-6 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 NET/0-6 +n116-040-168:3350747:3350810 [1] NCCL INFO 7 : NET/0-7 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 NET/0-7 +n116-040-168:3350749:3350809 [3] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-040-168:3350749:3350809 [3] NCCL INFO 0 : NET/0-0 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 NET/0-0 +n116-040-168:3350749:3350809 [3] NCCL INFO 1 : NET/0-1 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 NET/0-1 +n116-040-168:3350749:3350809 [3] NCCL INFO 2 : NET/0-2 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 NET/0-2 +n116-040-168:3350749:3350809 [3] NCCL INFO 3 : NET/0-3 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 NET/0-3 +n116-040-168:3350749:3350809 [3] NCCL INFO 4 : NET/0-4 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 NET/0-4 +n116-040-168:3350749:3350809 [3] NCCL INFO 5 : NET/0-5 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 NET/0-5 +n116-040-168:3350749:3350809 [3] NCCL INFO 6 : NET/0-6 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 NET/0-6 +n116-040-168:3350749:3350809 [3] NCCL INFO 7 : NET/0-7 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 NET/0-7 +n116-040-168:3350749:3350809 [3] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-040-168:3350749:3350809 [3] NCCL INFO 0 : NET/0-0 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 NET/0-0 +n116-040-168:3350749:3350809 [3] NCCL INFO 1 : NET/0-1 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 NET/0-1 +n116-040-168:3350749:3350809 [3] NCCL INFO 2 : NET/0-2 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 NET/0-2 +n116-040-168:3350749:3350809 [3] NCCL INFO 3 : NET/0-3 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 NET/0-3 +n116-040-168:3350749:3350809 [3] NCCL INFO 4 : NET/0-4 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 NET/0-4 +n116-040-168:3350749:3350809 [3] NCCL INFO 5 : NET/0-5 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 NET/0-5 +n116-040-168:3350749:3350809 [3] NCCL INFO 6 : NET/0-6 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 NET/0-6 +n116-040-168:3350749:3350809 [3] NCCL INFO 7 : NET/0-7 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 NET/0-7 +n116-040-168:3350746:3350795 [0] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-040-168:3350753:3350801 [7] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-040-168:3350746:3350795 [0] NCCL INFO 0 : NET/0-0 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 NET/0-0 +n116-040-168:3350753:3350801 [7] NCCL INFO 0 : NET/0-0 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 NET/0-0 +n116-040-168:3350746:3350795 [0] NCCL INFO 1 : NET/0-1 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 NET/0-1 +n116-040-168:3350753:3350801 [7] NCCL INFO 1 : NET/0-1 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 NET/0-1 +n116-040-168:3350746:3350795 [0] NCCL INFO 2 : NET/0-2 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 NET/0-2 +n116-040-168:3350753:3350801 [7] NCCL INFO 2 : NET/0-2 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 NET/0-2 +n116-040-168:3350746:3350795 [0] NCCL INFO 3 : NET/0-3 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 NET/0-3 +n116-040-168:3350753:3350801 [7] NCCL INFO 3 : NET/0-3 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 NET/0-3 +n116-040-168:3350746:3350795 [0] NCCL INFO 4 : NET/0-4 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 NET/0-4 +n116-040-168:3350753:3350801 [7] NCCL INFO 4 : NET/0-4 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 NET/0-4 +n116-040-168:3350746:3350795 [0] NCCL INFO 5 : NET/0-5 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 NET/0-5 +n116-040-168:3350753:3350801 [7] NCCL INFO 5 : NET/0-5 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 NET/0-5 +n116-040-168:3350746:3350795 [0] NCCL INFO 6 : NET/0-6 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 NET/0-6 +n116-040-168:3350753:3350801 [7] NCCL INFO 6 : NET/0-6 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 NET/0-6 +n116-040-168:3350746:3350795 [0] NCCL INFO 7 : NET/0-7 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 NET/0-7 +n116-040-168:3350753:3350801 [7] NCCL INFO 7 : NET/0-7 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 NET/0-7 +n116-040-168:3350753:3350801 [7] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-040-168:3350746:3350795 [0] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-040-168:3350746:3350795 [0] NCCL INFO 0 : NET/0-0 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 NET/0-0 +n116-040-168:3350753:3350801 [7] NCCL INFO 0 : NET/0-0 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 NET/0-0 +n116-040-168:3350746:3350795 [0] NCCL INFO 1 : NET/0-1 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 NET/0-1 +n116-040-168:3350753:3350801 [7] NCCL INFO 1 : NET/0-1 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 NET/0-1 +n116-040-168:3350746:3350795 [0] NCCL INFO 2 : NET/0-2 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 NET/0-2 +n116-040-168:3350753:3350801 [7] NCCL INFO 2 : NET/0-2 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 NET/0-2 +n116-040-168:3350746:3350795 [0] NCCL INFO 3 : NET/0-3 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 NET/0-3 +n116-040-168:3350753:3350801 [7] NCCL INFO 3 : NET/0-3 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 NET/0-3 +n116-040-168:3350746:3350795 [0] NCCL INFO 4 : NET/0-4 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 NET/0-4 +n116-040-168:3350753:3350801 [7] NCCL INFO 4 : NET/0-4 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 NET/0-4 +n116-040-168:3350746:3350795 [0] NCCL INFO 5 : NET/0-5 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 NET/0-5 +n116-040-168:3350753:3350801 [7] NCCL INFO 5 : NET/0-5 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 NET/0-5 +n116-040-168:3350746:3350795 [0] NCCL INFO 6 : NET/0-6 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 NET/0-6 +n116-040-168:3350753:3350801 [7] NCCL INFO 6 : NET/0-6 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 NET/0-6 +n116-040-168:3350746:3350795 [0] NCCL INFO 7 : NET/0-7 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 NET/0-7 +n116-040-168:3350753:3350801 [7] NCCL INFO 7 : NET/0-7 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 NET/0-7 +n116-040-168:3350748:3350803 [2] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-040-168:3350748:3350803 [2] NCCL INFO 0 : NET/0-0 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 NET/0-0 +n116-040-168:3350748:3350803 [2] NCCL INFO 1 : NET/0-1 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 NET/0-1 +n116-040-168:3350748:3350803 [2] NCCL INFO 2 : NET/0-2 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 NET/0-2 +n116-040-168:3350748:3350803 [2] NCCL INFO 3 : NET/0-3 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 NET/0-3 +n116-040-168:3350748:3350803 [2] NCCL INFO 4 : NET/0-4 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 NET/0-4 +n116-040-168:3350748:3350803 [2] NCCL INFO 5 : NET/0-5 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 NET/0-5 +n116-040-168:3350748:3350803 [2] NCCL INFO 6 : NET/0-6 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 NET/0-6 +n116-040-168:3350748:3350803 [2] NCCL INFO 7 : NET/0-7 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 NET/0-7 +n116-040-168:3350748:3350803 [2] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-040-168:3350748:3350803 [2] NCCL INFO 0 : NET/0-0 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 NET/0-0 +n116-040-168:3350748:3350803 [2] NCCL INFO 1 : NET/0-1 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 NET/0-1 +n116-040-168:3350748:3350803 [2] NCCL INFO 2 : NET/0-2 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 NET/0-2 +n116-040-168:3350748:3350803 [2] NCCL INFO 3 : NET/0-3 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 NET/0-3 +n116-040-168:3350748:3350803 [2] NCCL INFO 4 : NET/0-4 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 NET/0-4 +n116-040-168:3350748:3350803 [2] NCCL INFO 5 : NET/0-5 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 NET/0-5 +n116-040-168:3350748:3350803 [2] NCCL INFO 6 : NET/0-6 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 NET/0-6 +n116-040-168:3350748:3350803 [2] NCCL INFO 7 : NET/0-7 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 NET/0-7 +n116-040-168:3350752:3350800 [6] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-040-168:3350752:3350800 [6] NCCL INFO 0 : NET/0-0 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 NET/0-0 +n116-040-168:3350752:3350800 [6] NCCL INFO 1 : NET/0-1 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 NET/0-1 +n116-040-168:3350752:3350800 [6] NCCL INFO 2 : NET/0-2 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 NET/0-2 +n116-040-168:3350752:3350800 [6] NCCL INFO 3 : NET/0-3 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 NET/0-3 +n116-040-168:3350752:3350800 [6] NCCL INFO 4 : NET/0-4 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 NET/0-4 +n116-040-168:3350752:3350800 [6] NCCL INFO 5 : NET/0-5 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 NET/0-5 +n116-040-168:3350752:3350800 [6] NCCL INFO 6 : NET/0-6 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 NET/0-6 +n116-040-168:3350752:3350800 [6] NCCL INFO 7 : NET/0-7 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 NET/0-7 +n116-040-168:3350752:3350800 [6] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-040-168:3350752:3350800 [6] NCCL INFO 0 : NET/0-0 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 NET/0-0 +n116-040-168:3350752:3350800 [6] NCCL INFO 1 : NET/0-1 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 NET/0-1 +n116-040-168:3350752:3350800 [6] NCCL INFO 2 : NET/0-2 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 NET/0-2 +n116-040-168:3350752:3350800 [6] NCCL INFO 3 : NET/0-3 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 NET/0-3 +n116-040-168:3350752:3350800 [6] NCCL INFO 4 : NET/0-4 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 NET/0-4 +n116-040-168:3350752:3350800 [6] NCCL INFO 5 : NET/0-5 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 NET/0-5 +n116-040-168:3350752:3350800 [6] NCCL INFO 6 : NET/0-6 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 NET/0-6 +n116-040-168:3350752:3350800 [6] NCCL INFO 7 : NET/0-7 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 NET/0-7 +n116-040-168:3350751:3350802 [5] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-040-168:3350751:3350802 [5] NCCL INFO 0 : NET/0-0 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 NET/0-0 +n116-040-168:3350751:3350802 [5] NCCL INFO 1 : NET/0-1 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 NET/0-1 +n116-040-168:3350751:3350802 [5] NCCL INFO 2 : NET/0-2 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 NET/0-2 +n116-040-168:3350751:3350802 [5] NCCL INFO 3 : NET/0-3 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 NET/0-3 +n116-040-168:3350751:3350802 [5] NCCL INFO 4 : NET/0-4 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 NET/0-4 +n116-040-168:3350751:3350802 [5] NCCL INFO 5 : NET/0-5 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 NET/0-5 +n116-040-168:3350751:3350802 [5] NCCL INFO 6 : NET/0-6 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 NET/0-6 +n116-040-168:3350751:3350802 [5] NCCL INFO 7 : NET/0-7 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 NET/0-7 +n116-040-168:3350751:3350802 [5] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-040-168:3350751:3350802 [5] NCCL INFO 0 : NET/0-0 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 NET/0-0 +n116-040-168:3350751:3350802 [5] NCCL INFO 1 : NET/0-1 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 NET/0-1 +n116-040-168:3350751:3350802 [5] NCCL INFO 2 : NET/0-2 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 NET/0-2 +n116-040-168:3350751:3350802 [5] NCCL INFO 3 : NET/0-3 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 GPU/5 GPU/4 NET/0-3 +n116-040-168:3350751:3350802 [5] NCCL INFO 4 : NET/0-4 GPU/4 GPU/5 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 NET/0-4 +n116-040-168:3350751:3350802 [5] NCCL INFO 5 : NET/0-5 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 GPU/7 GPU/6 NET/0-5 +n116-040-168:3350751:3350802 [5] NCCL INFO 6 : NET/0-6 GPU/6 GPU/7 GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 NET/0-6 +n116-040-168:3350751:3350802 [5] NCCL INFO 7 : NET/0-7 GPU/7 GPU/6 GPU/5 GPU/4 GPU/3 GPU/2 GPU/1 GPU/0 NET/0-7 +n116-040-168:3350753:3350801 [7] NCCL INFO comm 0xccbc9b0 rank 7 nRanks 64 nNodes 8 localRanks 8 localRank 7 MNNVL 0 +n116-040-168:3350751:3350802 [5] NCCL INFO comm 0xccadb00 rank 5 nRanks 64 nNodes 8 localRanks 8 localRank 5 MNNVL 0 +n116-040-168:3350752:3350800 [6] NCCL INFO comm 0xb33fce0 rank 6 nRanks 64 nNodes 8 localRanks 8 localRank 6 MNNVL 0 +n116-040-168:3350750:3350804 [4] NCCL INFO comm 0xca06ab0 rank 4 nRanks 64 nNodes 8 localRanks 8 localRank 4 MNNVL 0 +n116-040-168:3350749:3350809 [3] NCCL INFO comm 0xbe60d40 rank 3 nRanks 64 nNodes 8 localRanks 8 localRank 3 MNNVL 0 +n116-040-168:3350748:3350803 [2] NCCL INFO comm 0xbb733d0 rank 2 nRanks 64 nNodes 8 localRanks 8 localRank 2 MNNVL 0 +n116-040-168:3350747:3350810 [1] NCCL INFO comm 0xc967cd0 rank 1 nRanks 64 nNodes 8 localRanks 8 localRank 1 MNNVL 0 +n116-040-168:3350753:3350801 [7] NCCL INFO Tree 6 : 6 -> 7 -> 0/-1/-1 +n116-040-168:3350753:3350801 [7] NCCL INFO Tree 14 : 6 -> 7 -> 0/-1/-1 +n116-040-168:3350750:3350804 [4] NCCL INFO Tree 4 : -1 -> 4 -> 5/36/-1 +n116-040-168:3350753:3350801 [7] NCCL INFO Tree 7 : -1 -> 7 -> 6/39/-1 +n116-040-168:3350750:3350804 [4] NCCL INFO Tree 12 : 12 -> 4 -> 5/-1/-1 +n116-040-168:3350753:3350801 [7] NCCL INFO Tree 15 : 15 -> 7 -> 6/-1/-1 +n116-040-168:3350750:3350804 [4] NCCL INFO Tree 5 : 5 -> 4 -> 3/-1/-1 +n116-040-168:3350750:3350804 [4] NCCL INFO Tree 13 : 5 -> 4 -> 3/-1/-1 +n116-040-168:3350753:3350801 [7] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-040-168:3350750:3350804 [4] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-040-168:3350749:3350809 [3] NCCL INFO Tree 2 : 2 -> 3 -> 4/-1/-1 +n116-040-168:3350747:3350810 [1] NCCL INFO Tree 0 : 0 -> 1 -> 2/-1/-1 +n116-040-168:3350751:3350802 [5] NCCL INFO Tree 4 : 4 -> 5 -> 6/-1/-1 +n116-040-168:3350747:3350810 [1] NCCL INFO Tree 8 : 0 -> 1 -> 2/-1/-1 +n116-040-168:3350748:3350803 [2] NCCL INFO Tree 2 : -1 -> 2 -> 3/34/-1 +n116-040-168:3350752:3350800 [6] NCCL INFO Tree 6 : -1 -> 6 -> 7/38/-1 +n116-040-168:3350749:3350809 [3] NCCL INFO Tree 10 : 2 -> 3 -> 4/-1/-1 +n116-040-168:3350752:3350800 [6] NCCL INFO Tree 14 : 14 -> 6 -> 7/-1/-1 +n116-040-168:3350751:3350802 [5] NCCL INFO Tree 12 : 4 -> 5 -> 6/-1/-1 +n116-040-168:3350747:3350810 [1] NCCL INFO Tree 1 : -1 -> 1 -> 0/33/-1 +n116-040-168:3350748:3350803 [2] NCCL INFO Tree 10 : 10 -> 2 -> 3/-1/-1 +n116-040-168:3350746:3350795 [0] NCCL INFO comm 0xb583d70 rank 0 nRanks 64 nNodes 8 localRanks 8 localRank 0 MNNVL 0 +n116-040-168:3350747:3350810 [1] NCCL INFO Tree 9 : 9 -> 1 -> 0/-1/-1 +n116-040-168:3350749:3350809 [3] NCCL INFO Tree 3 : -1 -> 3 -> 2/35/-1 +n116-040-168:3350751:3350802 [5] NCCL INFO Tree 5 : -1 -> 5 -> 4/37/-1 +n116-040-168:3350752:3350800 [6] NCCL INFO Tree 7 : 7 -> 6 -> 5/-1/-1 +n116-040-168:3350748:3350803 [2] NCCL INFO Tree 3 : 3 -> 2 -> 1/-1/-1 +n116-040-168:3350749:3350809 [3] NCCL INFO Tree 11 : 11 -> 3 -> 2/-1/-1 +n116-040-168:3350752:3350800 [6] NCCL INFO Tree 15 : 7 -> 6 -> 5/-1/-1 +n116-040-168:3350748:3350803 [2] NCCL INFO Tree 11 : 3 -> 2 -> 1/-1/-1 +n116-040-168:3350751:3350802 [5] NCCL INFO Tree 13 : 13 -> 5 -> 4/-1/-1 +n116-040-168:3350747:3350810 [1] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-040-168:3350750:3350804 [4] NCCL INFO Ring 00 : 5 -> 4 -> 3 +n116-040-168:3350750:3350804 [4] NCCL INFO Ring 01 : 3 -> 4 -> 5 +n116-040-168:3350748:3350803 [2] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-040-168:3350750:3350804 [4] NCCL INFO Ring 02 : 5 -> 4 -> 3 +n116-040-168:3350749:3350809 [3] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-040-168:3350752:3350800 [6] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-040-168:3350751:3350802 [5] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-040-168:3350753:3350801 [7] NCCL INFO Ring 00 : 0 -> 7 -> 6 +n116-040-168:3350750:3350804 [4] NCCL INFO Ring 03 : 3 -> 4 -> 5 +n116-040-168:3350750:3350804 [4] NCCL INFO Ring 04 : 61 -> 4 -> 3 +n116-040-168:3350753:3350801 [7] NCCL INFO Ring 01 : 6 -> 7 -> 0 +n116-040-168:3350750:3350804 [4] NCCL INFO Ring 05 : 3 -> 4 -> 13 +n116-040-168:3350753:3350801 [7] NCCL INFO Ring 02 : 0 -> 7 -> 6 +n116-040-168:3350750:3350804 [4] NCCL INFO Ring 06 : 5 -> 4 -> 3 +n116-040-168:3350753:3350801 [7] NCCL INFO Ring 03 : 6 -> 7 -> 0 +n116-040-168:3350750:3350804 [4] NCCL INFO Ring 07 : 3 -> 4 -> 5 +n116-040-168:3350746:3350795 [0] NCCL INFO Tree 0 : -1 -> 0 -> 1/32/-1 +n116-040-168:3350746:3350795 [0] NCCL INFO Tree 8 : 8 -> 0 -> 1/-1/-1 +n116-040-168:3350753:3350801 [7] NCCL INFO Ring 04 : 0 -> 7 -> 6 +n116-040-168:3350746:3350795 [0] NCCL INFO Tree 1 : 1 -> 0 -> 7/-1/-1 +n116-040-168:3350750:3350804 [4] NCCL INFO Ring 08 : 5 -> 4 -> 3 +n116-040-168:3350753:3350801 [7] NCCL INFO Ring 05 : 6 -> 7 -> 0 +n116-040-168:3350746:3350795 [0] NCCL INFO Tree 9 : 1 -> 0 -> 7/-1/-1 +n116-040-168:3350750:3350804 [4] NCCL INFO Ring 09 : 3 -> 4 -> 5 +n116-040-168:3350753:3350801 [7] NCCL INFO Ring 06 : 0 -> 7 -> 14 +n116-040-168:3350750:3350804 [4] NCCL INFO Ring 10 : 5 -> 4 -> 3 +n116-040-168:3350753:3350801 [7] NCCL INFO Ring 07 : 62 -> 7 -> 0 +n116-040-168:3350750:3350804 [4] NCCL INFO Ring 11 : 3 -> 4 -> 5 +n116-040-168:3350746:3350795 [0] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-040-168:3350750:3350804 [4] NCCL INFO Ring 12 : 61 -> 4 -> 3 +n116-040-168:3350753:3350801 [7] NCCL INFO Ring 08 : 0 -> 7 -> 6 +n116-040-168:3350750:3350804 [4] NCCL INFO Ring 13 : 3 -> 4 -> 13 +n116-040-168:3350747:3350810 [1] NCCL INFO Ring 00 : 2 -> 1 -> 8 +n116-040-168:3350753:3350801 [7] NCCL INFO Ring 09 : 6 -> 7 -> 0 +n116-040-168:3350750:3350804 [4] NCCL INFO Ring 14 : 5 -> 4 -> 3 +n116-040-168:3350747:3350810 [1] NCCL INFO Ring 01 : 56 -> 1 -> 2 +n116-040-168:3350753:3350801 [7] NCCL INFO Ring 10 : 0 -> 7 -> 6 +n116-040-168:3350748:3350803 [2] NCCL INFO Ring 00 : 3 -> 2 -> 1 +n116-040-168:3350746:3350795 [0] NCCL INFO Channel 00/16 : 0 7 6 5 4 3 2 1 8 15 14 13 12 11 10 9 16 23 22 21 20 19 18 17 24 31 30 29 28 27 26 25 32 39 38 37 ... +n116-040-168:3350750:3350804 [4] NCCL INFO Ring 15 : 3 -> 4 -> 5 +n116-040-168:3350749:3350809 [3] NCCL INFO Ring 00 : 4 -> 3 -> 2 +n116-040-168:3350747:3350810 [1] NCCL INFO Ring 02 : 2 -> 1 -> 0 +n116-040-168:3350753:3350801 [7] NCCL INFO Ring 11 : 6 -> 7 -> 0 +n116-040-168:3350748:3350803 [2] NCCL INFO Ring 01 : 1 -> 2 -> 3 +n116-040-168:3350750:3350804 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 3/-1/-1->4->5 [2] 5/-1/-1->4->3 [3] -1/-1/-1->4->5 [4] 5/36/-1->4->-1 [5] 3/-1/-1->4->5 [6] 5/-1/-1->4->3 [7] 3/-1/-1->4->5 [8] 5/-1/-1->4->3 [9] 3/-1/-1->4->5 [10] 5/-1/-1->4->3 [11] -1/-1/-1->4->5 [12] 5/-1/-1->4->12 [13] 3/-1/-1->4->5 [14] 5/-1/-1->4->3 [15] 3/-1/-1->4->5 +n116-040-168:3350749:3350809 [3] NCCL INFO Ring 01 : 2 -> 3 -> 4 +n116-040-168:3350747:3350810 [1] NCCL INFO Ring 03 : 0 -> 1 -> 2 +n116-040-168:3350753:3350801 [7] NCCL INFO Ring 12 : 0 -> 7 -> 6 +n116-040-168:3350746:3350795 [0] NCCL INFO Channel 01/16 : 0 9 10 11 12 13 14 15 8 17 18 19 20 21 22 23 16 25 26 27 28 29 30 31 24 33 34 35 36 37 38 39 32 41 42 43 ... +n116-040-168:3350748:3350803 [2] NCCL INFO Ring 02 : 59 -> 2 -> 1 +n116-040-168:3350750:3350804 [4] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-040-168:3350750:3350804 [4] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-040-168:3350749:3350809 [3] NCCL INFO Ring 02 : 4 -> 3 -> 10 +n116-040-168:3350747:3350810 [1] NCCL INFO Ring 04 : 2 -> 1 -> 0 +n116-040-168:3350753:3350801 [7] NCCL INFO Ring 13 : 6 -> 7 -> 0 +n116-040-168:3350746:3350795 [0] NCCL INFO Channel 02/16 : 0 7 6 5 4 3 10 9 8 15 14 13 12 11 18 17 16 23 22 21 20 19 26 25 24 31 30 29 28 27 34 33 32 39 38 37 ... +n116-040-168:3350748:3350803 [2] NCCL INFO Ring 03 : 1 -> 2 -> 11 +n116-040-168:3350753:3350801 [7] NCCL INFO Ring 14 : 0 -> 7 -> 14 +n116-040-168:3350750:3350804 [4] NCCL INFO P2P Chunksize set to 524288 +n116-040-168:3350749:3350809 [3] NCCL INFO Ring 03 : 58 -> 3 -> 4 +n116-040-168:3350747:3350810 [1] NCCL INFO Ring 05 : 0 -> 1 -> 2 +n116-040-168:3350753:3350801 [7] NCCL INFO Ring 15 : 62 -> 7 -> 0 +n116-040-168:3350748:3350803 [2] NCCL INFO Ring 04 : 3 -> 2 -> 1 +n116-040-168:3350746:3350795 [0] NCCL INFO Channel 03/16 : 0 1 2 11 12 13 14 15 8 9 10 19 20 21 22 23 16 17 18 27 28 29 30 31 24 25 26 35 36 37 38 39 32 33 34 43 ... +n116-040-168:3350749:3350809 [3] NCCL INFO Ring 04 : 4 -> 3 -> 2 +n116-040-168:3350747:3350810 [1] NCCL INFO Ring 06 : 2 -> 1 -> 0 +n116-040-168:3350753:3350801 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] 6/-1/-1->7->0 [2] 0/-1/-1->7->6 [3] 6/-1/-1->7->0 [4] 0/-1/-1->7->6 [5] 6/-1/-1->7->0 [6] 0/-1/-1->7->6 [7] 6/39/-1->7->-1 [8] -1/-1/-1->7->6 [9] 6/-1/-1->7->0 [10] 0/-1/-1->7->6 [11] 6/-1/-1->7->0 [12] 0/-1/-1->7->6 [13] 6/-1/-1->7->0 [14] 0/-1/-1->7->6 [15] 6/-1/-1->7->15 +n116-040-168:3350748:3350803 [2] NCCL INFO Ring 05 : 1 -> 2 -> 3 +n116-040-168:3350751:3350802 [5] NCCL INFO Ring 00 : 6 -> 5 -> 4 +n116-040-168:3350752:3350800 [6] NCCL INFO Ring 00 : 7 -> 6 -> 5 +n116-040-168:3350749:3350809 [3] NCCL INFO Ring 05 : 2 -> 3 -> 4 +n116-040-168:3350747:3350810 [1] NCCL INFO Ring 07 : 0 -> 1 -> 2 +n116-040-168:3350748:3350803 [2] NCCL INFO Ring 06 : 3 -> 2 -> 1 +n116-040-168:3350753:3350801 [7] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-040-168:3350746:3350795 [0] NCCL INFO Channel 04/16 : 0 7 6 5 12 11 10 9 8 15 14 13 20 19 18 17 16 23 22 21 28 27 26 25 24 31 30 29 36 35 34 33 32 39 38 37 ... +n116-040-168:3350751:3350802 [5] NCCL INFO Ring 01 : 4 -> 5 -> 6 +n116-040-168:3350753:3350801 [7] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-040-168:3350752:3350800 [6] NCCL INFO Ring 01 : 5 -> 6 -> 7 +n116-040-168:3350749:3350809 [3] NCCL INFO Ring 06 : 4 -> 3 -> 2 +n116-040-168:3350747:3350810 [1] NCCL INFO Ring 08 : 2 -> 1 -> 8 +n116-040-168:3350748:3350803 [2] NCCL INFO Ring 07 : 1 -> 2 -> 3 +n116-040-168:3350746:3350795 [0] NCCL INFO Channel 05/16 : 0 1 2 3 4 13 14 15 8 9 10 11 12 21 22 23 16 17 18 19 20 29 30 31 24 25 26 27 28 37 38 39 32 33 34 35 ... +n116-040-168:3350749:3350809 [3] NCCL INFO Ring 07 : 2 -> 3 -> 4 +n116-040-168:3350747:3350810 [1] NCCL INFO Ring 09 : 56 -> 1 -> 2 +n116-040-168:3350751:3350802 [5] NCCL INFO Ring 02 : 6 -> 5 -> 4 +n116-040-168:3350752:3350800 [6] NCCL INFO Ring 02 : 7 -> 6 -> 5 +n116-040-168:3350753:3350801 [7] NCCL INFO P2P Chunksize set to 524288 +n116-040-168:3350748:3350803 [2] NCCL INFO Ring 08 : 3 -> 2 -> 1 +n116-040-168:3350748:3350803 [2] NCCL INFO Ring 09 : 1 -> 2 -> 3 +n116-040-168:3350749:3350809 [3] NCCL INFO Ring 08 : 4 -> 3 -> 2 +n116-040-168:3350747:3350810 [1] NCCL INFO Ring 10 : 2 -> 1 -> 0 +n116-040-168:3350748:3350803 [2] NCCL INFO Ring 10 : 59 -> 2 -> 1 +n116-040-168:3350751:3350802 [5] NCCL INFO Ring 03 : 4 -> 5 -> 6 +n116-040-168:3350752:3350800 [6] NCCL INFO Ring 03 : 5 -> 6 -> 7 +n116-040-168:3350748:3350803 [2] NCCL INFO Ring 11 : 1 -> 2 -> 11 +n116-040-168:3350746:3350795 [0] NCCL INFO Channel 06/16 : 0 7 14 13 12 11 10 9 8 15 22 21 20 19 18 17 16 23 30 29 28 27 26 25 24 31 38 37 36 35 34 33 32 39 46 45 ... +n116-040-168:3350749:3350809 [3] NCCL INFO Ring 09 : 2 -> 3 -> 4 +n116-040-168:3350747:3350810 [1] NCCL INFO Ring 11 : 0 -> 1 -> 2 +n116-040-168:3350751:3350802 [5] NCCL INFO Ring 04 : 6 -> 5 -> 12 +n116-040-168:3350752:3350800 [6] NCCL INFO Ring 04 : 7 -> 6 -> 5 +n116-040-168:3350748:3350803 [2] NCCL INFO Ring 12 : 3 -> 2 -> 1 +n116-040-168:3350747:3350810 [1] NCCL INFO Ring 12 : 2 -> 1 -> 0 +n116-040-168:3350749:3350809 [3] NCCL INFO Ring 10 : 4 -> 3 -> 10 +n116-040-168:3350747:3350810 [1] NCCL INFO Ring 13 : 0 -> 1 -> 2 +n116-040-168:3350752:3350800 [6] NCCL INFO Ring 05 : 5 -> 6 -> 7 +n116-040-168:3350746:3350795 [0] NCCL INFO Channel 07/16 : 0 1 2 3 4 5 6 15 8 9 10 11 12 13 14 23 16 17 18 19 20 21 22 31 24 25 26 27 28 29 30 39 32 33 34 35 ... +n116-040-168:3350751:3350802 [5] NCCL INFO Ring 05 : 60 -> 5 -> 6 +n116-040-168:3350748:3350803 [2] NCCL INFO Ring 13 : 1 -> 2 -> 3 +n116-040-168:3350749:3350809 [3] NCCL INFO Ring 11 : 58 -> 3 -> 4 +n116-040-168:3350747:3350810 [1] NCCL INFO Ring 14 : 2 -> 1 -> 0 +n116-040-168:3350751:3350802 [5] NCCL INFO Ring 06 : 6 -> 5 -> 4 +n116-040-168:3350749:3350809 [3] NCCL INFO Ring 12 : 4 -> 3 -> 2 +n116-040-168:3350752:3350800 [6] NCCL INFO Ring 06 : 63 -> 6 -> 5 +n116-040-168:3350746:3350795 [0] NCCL INFO Channel 08/16 : 0 7 6 5 4 3 2 1 8 15 14 13 12 11 10 9 16 23 22 21 20 19 18 17 24 31 30 29 28 27 26 25 32 39 38 37 ... +n116-040-168:3350748:3350803 [2] NCCL INFO Ring 14 : 3 -> 2 -> 1 +n116-040-168:3350747:3350810 [1] NCCL INFO Ring 15 : 0 -> 1 -> 2 +n116-040-168:3350751:3350802 [5] NCCL INFO Ring 07 : 4 -> 5 -> 6 +n116-040-168:3350749:3350809 [3] NCCL INFO Ring 13 : 2 -> 3 -> 4 +n116-040-168:3350751:3350802 [5] NCCL INFO Ring 08 : 6 -> 5 -> 4 +n116-040-168:3350752:3350800 [6] NCCL INFO Ring 07 : 5 -> 6 -> 15 +n116-040-168:3350748:3350803 [2] NCCL INFO Ring 15 : 1 -> 2 -> 3 +n116-040-168:3350747:3350810 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 0/33/-1->1->-1 [2] -1/-1/-1->1->0 [3] 0/-1/-1->1->2 [4] 2/-1/-1->1->0 [5] 0/-1/-1->1->2 [6] 2/-1/-1->1->0 [7] 0/-1/-1->1->2 [8] 2/-1/-1->1->0 [9] 0/-1/-1->1->9 [10] -1/-1/-1->1->0 [11] 0/-1/-1->1->2 [12] 2/-1/-1->1->0 [13] 0/-1/-1->1->2 [14] 2/-1/-1->1->0 [15] 0/-1/-1->1->2 +n116-040-168:3350749:3350809 [3] NCCL INFO Ring 14 : 4 -> 3 -> 2 +n116-040-168:3350748:3350803 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] -1/-1/-1->2->3 [2] 3/34/-1->2->-1 [3] 1/-1/-1->2->3 [4] 3/-1/-1->2->1 [5] 1/-1/-1->2->3 [6] 3/-1/-1->2->1 [7] 1/-1/-1->2->3 [8] 3/-1/-1->2->1 [9] -1/-1/-1->2->3 [10] 3/-1/-1->2->10 [11] 1/-1/-1->2->3 [12] 3/-1/-1->2->1 [13] 1/-1/-1->2->3 [14] 3/-1/-1->2->1 [15] 1/-1/-1->2->3 +n116-040-168:3350746:3350795 [0] NCCL INFO Channel 09/16 : 0 9 10 11 12 13 14 15 8 17 18 19 20 21 22 23 16 25 26 27 28 29 30 31 24 33 34 35 36 37 38 39 32 41 42 43 ... +n116-040-168:3350752:3350800 [6] NCCL INFO Ring 08 : 7 -> 6 -> 5 +n116-040-168:3350751:3350802 [5] NCCL INFO Ring 09 : 4 -> 5 -> 6 +n116-040-168:3350748:3350803 [2] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-040-168:3350749:3350809 [3] NCCL INFO Ring 15 : 2 -> 3 -> 4 +n116-040-168:3350747:3350810 [1] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-040-168:3350751:3350802 [5] NCCL INFO Ring 10 : 6 -> 5 -> 4 +n116-040-168:3350749:3350809 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 2/-1/-1->3->4 [2] 4/-1/-1->3->2 [3] 2/35/-1->3->-1 [4] -1/-1/-1->3->2 [5] 2/-1/-1->3->4 [6] 4/-1/-1->3->2 [7] 2/-1/-1->3->4 [8] 4/-1/-1->3->2 [9] 2/-1/-1->3->4 [10] 4/-1/-1->3->2 [11] 2/-1/-1->3->11 [12] -1/-1/-1->3->2 [13] 2/-1/-1->3->4 [14] 4/-1/-1->3->2 [15] 2/-1/-1->3->4 +n116-040-168:3350746:3350795 [0] NCCL INFO Channel 10/16 : 0 7 6 5 4 3 10 9 8 15 14 13 12 11 18 17 16 23 22 21 20 19 26 25 24 31 30 29 28 27 34 33 32 39 38 37 ... +n116-040-168:3350752:3350800 [6] NCCL INFO Ring 09 : 5 -> 6 -> 7 +n116-040-168:3350748:3350803 [2] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-040-168:3350747:3350810 [1] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-040-168:3350751:3350802 [5] NCCL INFO Ring 11 : 4 -> 5 -> 6 +n116-040-168:3350749:3350809 [3] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-040-168:3350752:3350800 [6] NCCL INFO Ring 10 : 7 -> 6 -> 5 +n116-040-168:3350746:3350795 [0] NCCL INFO Channel 11/16 : 0 1 2 11 12 13 14 15 8 9 10 19 20 21 22 23 16 17 18 27 28 29 30 31 24 25 26 35 36 37 38 39 32 33 34 43 ... +n116-040-168:3350748:3350803 [2] NCCL INFO P2P Chunksize set to 524288 +n116-040-168:3350747:3350810 [1] NCCL INFO P2P Chunksize set to 524288 +n116-040-168:3350751:3350802 [5] NCCL INFO Ring 12 : 6 -> 5 -> 12 +n116-040-168:3350749:3350809 [3] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-040-168:3350752:3350800 [6] NCCL INFO Ring 11 : 5 -> 6 -> 7 +n116-040-168:3350751:3350802 [5] NCCL INFO Ring 13 : 60 -> 5 -> 6 +n116-040-168:3350746:3350795 [0] NCCL INFO Channel 12/16 : 0 7 6 5 12 11 10 9 8 15 14 13 20 19 18 17 16 23 22 21 28 27 26 25 24 31 30 29 36 35 34 33 32 39 38 37 ... +n116-040-168:3350749:3350809 [3] NCCL INFO P2P Chunksize set to 524288 +n116-040-168:3350752:3350800 [6] NCCL INFO Ring 12 : 7 -> 6 -> 5 +n116-040-168:3350751:3350802 [5] NCCL INFO Ring 14 : 6 -> 5 -> 4 +n116-040-168:3350752:3350800 [6] NCCL INFO Ring 13 : 5 -> 6 -> 7 +n116-040-168:3350751:3350802 [5] NCCL INFO Ring 15 : 4 -> 5 -> 6 +n116-040-168:3350752:3350800 [6] NCCL INFO Ring 14 : 63 -> 6 -> 5 +n116-040-168:3350746:3350795 [0] NCCL INFO Channel 13/16 : 0 1 2 3 4 13 14 15 8 9 10 11 12 21 22 23 16 17 18 19 20 29 30 31 24 25 26 27 28 37 38 39 32 33 34 35 ... +n116-040-168:3350751:3350802 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 4/-1/-1->5->6 [2] 6/-1/-1->5->4 [3] 4/-1/-1->5->6 [4] 6/-1/-1->5->4 [5] 4/37/-1->5->-1 [6] -1/-1/-1->5->4 [7] 4/-1/-1->5->6 [8] 6/-1/-1->5->4 [9] 4/-1/-1->5->6 [10] 6/-1/-1->5->4 [11] 4/-1/-1->5->6 [12] 6/-1/-1->5->4 [13] 4/-1/-1->5->13 [14] -1/-1/-1->5->4 [15] 4/-1/-1->5->6 +n116-040-168:3350752:3350800 [6] NCCL INFO Ring 15 : 5 -> 6 -> 15 +n116-040-168:3350752:3350800 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 5/-1/-1->6->7 [2] 7/-1/-1->6->5 [3] 5/-1/-1->6->7 [4] 7/-1/-1->6->5 [5] -1/-1/-1->6->7 [6] 7/38/-1->6->-1 [7] 5/-1/-1->6->7 [8] 7/-1/-1->6->5 [9] 5/-1/-1->6->7 [10] 7/-1/-1->6->5 [11] 5/-1/-1->6->7 [12] 7/-1/-1->6->5 [13] -1/-1/-1->6->7 [14] 7/-1/-1->6->14 [15] 5/-1/-1->6->7 +n116-040-168:3350751:3350802 [5] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-040-168:3350746:3350795 [0] NCCL INFO Channel 14/16 : 0 7 14 13 12 11 10 9 8 15 22 21 20 19 18 17 16 23 30 29 28 27 26 25 24 31 38 37 36 35 34 33 32 39 46 45 ... +n116-040-168:3350751:3350802 [5] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-040-168:3350752:3350800 [6] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-040-168:3350751:3350802 [5] NCCL INFO P2P Chunksize set to 524288 +n116-040-168:3350752:3350800 [6] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-040-168:3350746:3350795 [0] NCCL INFO Channel 15/16 : 0 1 2 3 4 5 6 15 8 9 10 11 12 13 14 23 16 17 18 19 20 21 22 31 24 25 26 27 28 29 30 39 32 33 34 35 ... +n116-040-168:3350752:3350800 [6] NCCL INFO P2P Chunksize set to 524288 +n116-040-168:3350746:3350795 [0] NCCL INFO Ring 00 : 57 -> 0 -> 7 +n116-040-168:3350746:3350795 [0] NCCL INFO Ring 01 : 7 -> 0 -> 9 +n116-040-168:3350746:3350795 [0] NCCL INFO Ring 02 : 1 -> 0 -> 7 +n116-040-168:3350746:3350795 [0] NCCL INFO Ring 03 : 7 -> 0 -> 1 +n116-040-168:3350746:3350795 [0] NCCL INFO Ring 04 : 1 -> 0 -> 7 +n116-040-168:3350746:3350795 [0] NCCL INFO Ring 05 : 7 -> 0 -> 1 +n116-040-168:3350746:3350795 [0] NCCL INFO Ring 06 : 1 -> 0 -> 7 +n116-040-168:3350746:3350795 [0] NCCL INFO Ring 07 : 7 -> 0 -> 1 +n116-040-168:3350746:3350795 [0] NCCL INFO Ring 08 : 57 -> 0 -> 7 +n116-040-168:3350746:3350795 [0] NCCL INFO Ring 09 : 7 -> 0 -> 9 +n116-040-168:3350746:3350795 [0] NCCL INFO Ring 10 : 1 -> 0 -> 7 +n116-040-168:3350746:3350795 [0] NCCL INFO Ring 11 : 7 -> 0 -> 1 +n116-040-168:3350746:3350795 [0] NCCL INFO Ring 12 : 1 -> 0 -> 7 +n116-040-168:3350746:3350795 [0] NCCL INFO Ring 13 : 7 -> 0 -> 1 +n116-040-168:3350746:3350795 [0] NCCL INFO Ring 14 : 1 -> 0 -> 7 +n116-040-168:3350746:3350795 [0] NCCL INFO Ring 15 : 7 -> 0 -> 1 +n116-040-168:3350746:3350795 [0] NCCL INFO Trees [0] 1/32/-1->0->-1 [1] 7/-1/-1->0->1 [2] 1/-1/-1->0->7 [3] 7/-1/-1->0->1 [4] 1/-1/-1->0->7 [5] 7/-1/-1->0->1 [6] 1/-1/-1->0->7 [7] -1/-1/-1->0->1 [8] 1/-1/-1->0->8 [9] 7/-1/-1->0->1 [10] 1/-1/-1->0->7 [11] 7/-1/-1->0->1 [12] 1/-1/-1->0->7 [13] 7/-1/-1->0->1 [14] 1/-1/-1->0->7 [15] -1/-1/-1->0->1 +n116-040-168:3350746:3350795 [0] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-040-168:3350746:3350795 [0] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-040-168:3350746:3350795 [0] NCCL INFO P2P Chunksize set to 524288 +n116-040-168:3350746:3350795 [0] NCCL INFO Check P2P Type intraNodeP2pSupport 1 directMode 0 +n116-040-168:3350750:3350843 [4] NCCL INFO [Proxy Service] Device 4 CPU core 85 +n116-040-168:3350753:3350844 [7] NCCL INFO [Proxy Service] Device 7 CPU core 101 +n116-040-168:3350750:3350845 [4] NCCL INFO [Proxy Service UDS] Device 4 CPU core 101 +n116-040-168:3350748:3350847 [2] NCCL INFO [Proxy Service] Device 2 CPU core 31 +n116-040-168:3350746:3350853 [0] NCCL INFO [Proxy Service] Device 0 CPU core 117 +n116-040-168:3350748:3350850 [2] NCCL INFO [Proxy Service UDS] Device 2 CPU core 11 +n116-040-168:3350753:3350846 [7] NCCL INFO [Proxy Service UDS] Device 7 CPU core 61 +n116-040-168:3350747:3350848 [1] NCCL INFO [Proxy Service] Device 1 CPU core 135 +n116-040-168:3350751:3350855 [5] NCCL INFO [Proxy Service] Device 5 CPU core 80 +n116-040-168:3350747:3350851 [1] NCCL INFO [Proxy Service UDS] Device 1 CPU core 115 +n116-040-168:3350746:3350856 [0] NCCL INFO [Proxy Service UDS] Device 0 CPU core 148 +n116-040-168:3350749:3350849 [3] NCCL INFO [Proxy Service] Device 3 CPU core 147 +n116-040-168:3350749:3350852 [3] NCCL INFO [Proxy Service UDS] Device 3 CPU core 23 +n116-040-168:3350752:3350854 [6] NCCL INFO [Proxy Service] Device 6 CPU core 80 +n116-040-168:3350751:3350858 [5] NCCL INFO [Proxy Service UDS] Device 5 CPU core 97 +n116-040-168:3350752:3350857 [6] NCCL INFO [Proxy Service UDS] Device 6 CPU core 97 +n116-040-168:3350750:3350804 [4] NCCL INFO NCCL_PROTO set by environment to Simple +n116-040-168:3350750:3350804 [4] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-040-168:3350750:3350804 [4] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n116-040-168:3350750:3350804 [4] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n116-040-168:3350753:3350801 [7] NCCL INFO NCCL_PROTO set by environment to Simple +n116-040-168:3350753:3350801 [7] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-040-168:3350753:3350801 [7] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n116-040-168:3350753:3350801 [7] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n116-040-168:3350751:3350802 [5] NCCL INFO NCCL_PROTO set by environment to Simple +n116-040-168:3350751:3350802 [5] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-040-168:3350751:3350802 [5] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n116-040-168:3350751:3350802 [5] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n116-040-168:3350747:3350810 [1] NCCL INFO NCCL_PROTO set by environment to Simple +n116-040-168:3350747:3350810 [1] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-040-168:3350747:3350810 [1] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n116-040-168:3350747:3350810 [1] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n116-040-168:3350746:3350795 [0] NCCL INFO NCCL_PROTO set by environment to Simple +n116-040-168:3350746:3350795 [0] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-040-168:3350746:3350795 [0] NCCL INFO Enabled NCCL Func/Proto/Algo Matrix: + Function | LL LL128 Simple | Tree Ring CollNetDirect CollNetChain NVLS NVLSTree PAT + Broadcast | 0 0 1 | 1 1 0 0 0 0 0 + Reduce | 0 0 1 | 1 1 0 0 0 0 0 + AllGather | 0 0 1 | 1 1 0 0 0 0 0 +ReduceScatter | 0 0 1 | 1 1 0 0 0 0 0 + AllReduce | 0 0 1 | 1 1 0 0 0 0 0 + +n116-040-168:3350746:3350795 [0] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n116-040-168:3350746:3350795 [0] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n116-040-168:3350746:3350795 [0] NCCL INFO CC Off, workFifoBytes 1048576 +n116-040-168:3350749:3350809 [3] NCCL INFO NCCL_PROTO set by environment to Simple +n116-040-168:3350749:3350809 [3] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-040-168:3350749:3350809 [3] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n116-040-168:3350749:3350809 [3] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n116-040-168:3350752:3350800 [6] NCCL INFO NCCL_PROTO set by environment to Simple +n116-040-168:3350752:3350800 [6] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-040-168:3350752:3350800 [6] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n116-040-168:3350752:3350800 [6] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n116-040-168:3350748:3350803 [2] NCCL INFO NCCL_PROTO set by environment to Simple +n116-040-168:3350748:3350803 [2] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-040-168:3350748:3350803 [2] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n116-040-168:3350748:3350803 [2] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n116-040-168:3350750:3350804 [4] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-040-168:3350750:3350804 [4] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-040-168:3350750:3350804 [4] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-040-168:3350750:3350804 [4] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-040-168:3350750:3350804 [4] NCCL INFO ncclCommInitRankConfig comm 0xca06ab0 rank 4 nranks 64 cudaDev 4 nvmlDev 4 busId 84000 commId 0xfb2d11625d05007a - Init COMPLETE +n116-040-168:3350750:3350804 [4] NCCL INFO Init timings - ncclCommInitRankConfig: rank 4 nranks 64 total 4.93 (kernels 0.40, alloc 1.58, bootstrap 2.06, allgathers 0.11, topo 0.66, graphs 0.08, connections 0.02, rest 0.02) +n116-040-168:3350752:3350800 [6] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-040-168:3350752:3350800 [6] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-040-168:3350752:3350800 [6] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-040-168:3350752:3350800 [6] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-040-168:3350752:3350800 [6] NCCL INFO ncclCommInitRankConfig comm 0xb33fce0 rank 6 nranks 64 cudaDev 6 nvmlDev 6 busId 8b000 commId 0xfb2d11625d05007a - Init COMPLETE +n116-040-168:3350752:3350800 [6] NCCL INFO Init timings - ncclCommInitRankConfig: rank 6 nranks 64 total 4.99 (kernels 0.40, alloc 1.35, bootstrap 2.36, allgathers 0.05, topo 0.66, graphs 0.13, connections 0.04, rest 0.00) +n116-040-168:3350747:3350810 [1] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-040-168:3350747:3350810 [1] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-040-168:3350747:3350810 [1] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-040-168:3350747:3350810 [1] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-040-168:3350747:3350810 [1] NCCL INFO ncclCommInitRankConfig comm 0xc967cd0 rank 1 nranks 64 cudaDev 1 nvmlDev 1 busId 5000 commId 0xfb2d11625d05007a - Init COMPLETE +n116-040-168:3350747:3350810 [1] NCCL INFO Init timings - ncclCommInitRankConfig: rank 1 nranks 64 total 4.76 (kernels 0.92, alloc 1.10, bootstrap 1.85, allgathers 0.10, topo 0.66, graphs 0.08, connections 0.03, rest 0.01) +n116-040-168:3350751:3350802 [5] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-040-168:3350753:3350801 [7] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-040-168:3350753:3350801 [7] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-040-168:3350753:3350801 [7] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-040-168:3350751:3350802 [5] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-040-168:3350753:3350801 [7] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-040-168:3350751:3350802 [5] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-040-168:3350751:3350802 [5] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-040-168:3350753:3350801 [7] NCCL INFO ncclCommInitRankConfig comm 0xccbc9b0 rank 7 nranks 64 cudaDev 7 nvmlDev 7 busId 8c000 commId 0xfb2d11625d05007a - Init COMPLETE +n116-040-168:3350751:3350802 [5] NCCL INFO ncclCommInitRankConfig comm 0xccadb00 rank 5 nranks 64 cudaDev 5 nvmlDev 5 busId 85000 commId 0xfb2d11625d05007a - Init COMPLETE +n116-040-168:3350753:3350801 [7] NCCL INFO Init timings - ncclCommInitRankConfig: rank 7 nranks 64 total 4.98 (kernels 0.38, alloc 1.50, bootstrap 2.21, allgathers 0.10, topo 0.66, graphs 0.08, connections 0.03, rest 0.02) +n116-040-168:3350751:3350802 [5] NCCL INFO Init timings - ncclCommInitRankConfig: rank 5 nranks 64 total 4.95 (kernels 0.40, alloc 1.56, bootstrap 2.10, allgathers 0.05, topo 0.66, graphs 0.14, connections 0.03, rest 0.01) +n116-040-168:3350746:3350795 [0] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-040-168:3350746:3350795 [0] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-040-168:3350746:3350795 [0] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-040-168:3350746:3350795 [0] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-040-168:3350746:3350795 [0] NCCL INFO ncclCommInitRankConfig comm 0xb583d70 rank 0 nranks 64 cudaDev 0 nvmlDev 0 busId 4000 commId 0xfb2d11625d05007a - Init COMPLETE +n116-040-168:3350746:3350795 [0] NCCL INFO Init timings - ncclCommInitRankConfig: rank 0 nranks 64 total 5.09 (kernels 0.41, alloc 0.39, bootstrap 3.40, allgathers 0.10, topo 0.66, graphs 0.08, connections 0.04, rest 0.01) +n116-040-168:3350749:3350809 [3] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-040-168:3350749:3350809 [3] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-040-168:3350749:3350809 [3] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-040-168:3350749:3350809 [3] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-040-168:3350749:3350809 [3] NCCL INFO ncclCommInitRankConfig comm 0xbe60d40 rank 3 nranks 64 cudaDev 3 nvmlDev 3 busId c000 commId 0xfb2d11625d05007a - Init COMPLETE +n116-040-168:3350748:3350803 [2] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-040-168:3350748:3350803 [2] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-040-168:3350749:3350809 [3] NCCL INFO Init timings - ncclCommInitRankConfig: rank 3 nranks 64 total 4.79 (kernels 0.79, alloc 1.27, bootstrap 1.85, allgathers 0.10, topo 0.66, graphs 0.08, connections 0.04, rest 0.00) +n116-040-168:3350748:3350803 [2] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-040-168:3350748:3350803 [2] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-040-168:3350748:3350803 [2] NCCL INFO ncclCommInitRankConfig comm 0xbb733d0 rank 2 nranks 64 cudaDev 2 nvmlDev 2 busId b000 commId 0xfb2d11625d05007a - Init COMPLETE +n116-040-168:3350748:3350803 [2] NCCL INFO Init timings - ncclCommInitRankConfig: rank 2 nranks 64 total 4.95 (kernels 0.37, alloc 1.58, bootstrap 2.11, allgathers 0.10, topo 0.66, graphs 0.08, connections 0.04, rest 0.00) +n116-040-168:3350746:3350866 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/CUMEM +n116-040-168:3350751:3350861 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/CUMEM +n116-040-168:3350746:3350866 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/CUMEM +n116-040-168:3350751:3350861 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/CUMEM +n116-040-168:3350746:3350866 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/CUMEM +n116-040-168:3350751:3350861 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/CUMEM +n116-040-168:3350749:3350863 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/CUMEM +n116-040-168:3350746:3350866 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/CUMEM +n116-040-168:3350751:3350861 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/CUMEM +n116-040-168:3350749:3350863 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/CUMEM +n116-040-168:3350746:3350866 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/CUMEM +n116-040-168:3350751:3350861 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/CUMEM +n116-040-168:3350749:3350863 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/CUMEM +n116-040-168:3350746:3350866 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/CUMEM +n116-040-168:3350751:3350861 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/CUMEM +n116-040-168:3350749:3350863 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/CUMEM +n116-040-168:3350746:3350866 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/CUMEM +n116-040-168:3350751:3350861 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/CUMEM +n116-040-168:3350749:3350863 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/CUMEM +n116-040-168:3350746:3350866 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/CUMEM +n116-040-168:3350751:3350861 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/CUMEM +n116-040-168:3350749:3350863 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/CUMEM +n116-040-168:3350746:3350866 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/CUMEM +n116-040-168:3350751:3350861 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/CUMEM +n116-040-168:3350746:3350866 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/CUMEM +n116-040-168:3350751:3350861 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/CUMEM +n116-040-168:3350748:3350865 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/CUMEM +n116-040-168:3350746:3350866 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/CUMEM +n116-040-168:3350750:3350859 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/CUMEM +n116-040-168:3350751:3350861 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/CUMEM +n116-040-168:3350748:3350865 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/CUMEM +n116-040-168:3350746:3350866 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/CUMEM +n116-040-168:3350750:3350859 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/CUMEM +n116-040-168:3350751:3350861 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/CUMEM +n116-040-168:3350748:3350865 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/CUMEM +n116-040-168:3350746:3350866 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/CUMEM +n116-040-168:3350750:3350859 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/CUMEM +n116-040-168:3350749:3350863 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/CUMEM +n116-040-168:3350748:3350865 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/CUMEM +n116-040-168:3350746:3350866 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/CUMEM +n116-040-168:3350750:3350859 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/CUMEM +n116-040-168:3350749:3350863 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/CUMEM +n116-040-168:3350746:3350866 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/CUMEM +n116-040-168:3350748:3350865 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/CUMEM +n116-040-168:3350750:3350859 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/CUMEM +n116-040-168:3350749:3350863 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/CUMEM +n116-040-168:3350746:3350866 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/CUMEM +n116-040-168:3350748:3350865 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/CUMEM +n116-040-168:3350750:3350859 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/CUMEM +n116-040-168:3350749:3350863 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/CUMEM +n116-040-168:3350748:3350865 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/CUMEM +n116-040-168:3350750:3350859 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/CUMEM +n116-040-168:3350749:3350863 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/CUMEM +n116-040-168:3350748:3350865 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/CUMEM +n116-040-168:3350750:3350859 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/CUMEM +n116-040-168:3350747:3350862 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/CUMEM +n116-040-168:3350748:3350865 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/CUMEM +n116-040-168:3350749:3350863 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/CUMEM +n116-040-168:3350750:3350859 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/CUMEM +n116-040-168:3350747:3350862 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/CUMEM +n116-040-168:3350748:3350865 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/CUMEM +n116-040-168:3350747:3350862 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/CUMEM +n116-040-168:3350750:3350859 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/CUMEM +n116-040-168:3350748:3350865 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/CUMEM +n116-040-168:3350747:3350862 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/CUMEM +n116-040-168:3350750:3350859 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/CUMEM +n116-040-168:3350748:3350865 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/CUMEM +n116-040-168:3350747:3350862 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/CUMEM +n116-040-168:3350750:3350859 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/CUMEM +n116-040-168:3350748:3350865 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/CUMEM +n116-040-168:3350747:3350862 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/CUMEM +n116-040-168:3350750:3350859 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/CUMEM +n116-040-168:3350748:3350865 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/CUMEM +n116-040-168:3350747:3350862 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/CUMEM +n116-040-168:3350750:3350859 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/CUMEM +n116-040-168:3350748:3350865 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/CUMEM +n116-040-168:3350747:3350862 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/CUMEM +n116-040-168:3350752:3350864 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/CUMEM +n116-040-168:3350748:3350865 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/CUMEM +n116-040-168:3350750:3350859 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/CUMEM +n116-040-168:3350747:3350862 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/CUMEM +n116-040-168:3350752:3350864 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/CUMEM +n116-040-168:3350750:3350859 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/CUMEM +n116-040-168:3350747:3350862 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/CUMEM +n116-040-168:3350752:3350864 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/CUMEM +n116-040-168:3350747:3350862 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/CUMEM +n116-040-168:3350752:3350864 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/CUMEM +n116-040-168:3350747:3350862 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/CUMEM +n116-040-168:3350752:3350864 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/CUMEM +n116-040-168:3350752:3350864 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/CUMEM +n116-040-168:3350746:3350866 [0] NCCL INFO Channel 01/0 : 0[0] -> 7[7] via P2P/CUMEM +n116-040-168:3350752:3350864 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/CUMEM +n116-040-168:3350746:3350866 [0] NCCL INFO Channel 02/0 : 0[0] -> 7[7] via P2P/CUMEM +n116-040-168:3350752:3350864 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/CUMEM +n116-040-168:3350746:3350866 [0] NCCL INFO Channel 03/0 : 0[0] -> 7[7] via P2P/CUMEM +n116-040-168:3350752:3350864 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/CUMEM +n116-040-168:3350746:3350866 [0] NCCL INFO Channel 04/0 : 0[0] -> 7[7] via P2P/CUMEM +n116-040-168:3350752:3350864 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/CUMEM +n116-040-168:3350746:3350866 [0] NCCL INFO Channel 05/0 : 0[0] -> 7[7] via P2P/CUMEM +n116-040-168:3350752:3350864 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/CUMEM +n116-040-168:3350746:3350866 [0] NCCL INFO Channel 06/0 : 0[0] -> 7[7] via P2P/CUMEM +n116-040-168:3350752:3350864 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/CUMEM +n116-040-168:3350746:3350866 [0] NCCL INFO Channel 09/0 : 0[0] -> 7[7] via P2P/CUMEM +n116-040-168:3350752:3350864 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/CUMEM +n116-040-168:3350746:3350866 [0] NCCL INFO Channel 10/0 : 0[0] -> 7[7] via P2P/CUMEM +n116-040-168:3350752:3350864 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/CUMEM +n116-040-168:3350746:3350866 [0] NCCL INFO Channel 11/0 : 0[0] -> 7[7] via P2P/CUMEM +n116-040-168:3350752:3350864 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/CUMEM +n116-040-168:3350746:3350866 [0] NCCL INFO Channel 12/0 : 0[0] -> 7[7] via P2P/CUMEM +n116-040-168:3350752:3350864 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/CUMEM +n116-040-168:3350746:3350866 [0] NCCL INFO Channel 13/0 : 0[0] -> 7[7] via P2P/CUMEM +n116-040-168:3350746:3350866 [0] NCCL INFO Channel 14/0 : 0[0] -> 7[7] via P2P/CUMEM +n116-040-168:3350750:3350867 [4] NCCL INFO [Proxy Progress] Device 4 CPU core 193 +n116-040-168:3350750:3350859 [4] NCCL INFO Channel 12/0 : 4[4] -> 12[4] [send] via NET/FasTrak/4/GDRDMA +n116-040-168:3350749:3350868 [3] NCCL INFO [Proxy Progress] Device 3 CPU core 36 +n116-040-168:3350749:3350863 [3] NCCL INFO Channel 11/0 : 3[3] -> 11[3] [send] via NET/FasTrak/3/GDRDMA +n116-040-168:3350752:3350869 [6] NCCL INFO [Proxy Progress] Device 6 CPU core 180 +n116-040-168:3350752:3350864 [6] NCCL INFO Channel 14/0 : 6[6] -> 14[6] [send] via NET/FasTrak/6/GDRDMA +n116-040-168:3350751:3350870 [5] NCCL INFO [Proxy Progress] Device 5 CPU core 96 +n116-040-168:3350751:3350861 [5] NCCL INFO Channel 13/0 : 5[5] -> 13[5] [send] via NET/FasTrak/5/GDRDMA +n116-040-168:3350746:3350871 [0] NCCL INFO [Proxy Progress] Device 0 CPU core 155 +n116-040-168:3350746:3350866 [0] NCCL INFO Channel 08/0 : 0[0] -> 8[0] [send] via NET/FasTrak/0/GDRDMA +n116-040-168:3350753:3350872 [7] NCCL INFO [Proxy Progress] Device 7 CPU core 172 +n116-040-168:3350753:3350860 [7] NCCL INFO Channel 15/0 : 7[7] -> 15[7] [send] via NET/FasTrak/7/GDRDMA +n116-040-168:3350748:3350873 [2] NCCL INFO [Proxy Progress] Device 2 CPU core 109 +n116-040-168:3350748:3350865 [2] NCCL INFO Channel 10/0 : 2[2] -> 10[2] [send] via NET/FasTrak/2/GDRDMA +n116-040-168:3350747:3350862 [1] NCCL INFO Channel 09/0 : 1[1] -> 9[1] [send] via NET/FasTrak/1/GDRDMA +n116-040-168:3350747:3350874 [1] NCCL INFO [Proxy Progress] Device 1 CPU core 129 +n116-040-168:3350750:3350859 [4] NCCL INFO Channel 04/0 : 36[4] -> 4[4] [receive] via NET/FasTrak/4/GDRDMA +n116-040-168:3350750:3350859 [4] NCCL INFO Channel 04/0 : 4[4] -> 36[4] [send] via NET/FasTrak/4/GDRDMA +n116-040-168:3350749:3350863 [3] NCCL INFO Channel 03/0 : 35[3] -> 3[3] [receive] via NET/FasTrak/3/GDRDMA +n116-040-168:3350747:3350862 [1] NCCL INFO Channel 01/0 : 33[1] -> 1[1] [receive] via NET/FasTrak/1/GDRDMA +n116-040-168:3350749:3350863 [3] NCCL INFO Channel 03/0 : 3[3] -> 35[3] [send] via NET/FasTrak/3/GDRDMA +n116-040-168:3350747:3350862 [1] NCCL INFO Channel 01/0 : 1[1] -> 33[1] [send] via NET/FasTrak/1/GDRDMA +n116-040-168:3350751:3350861 [5] NCCL INFO Channel 05/0 : 37[5] -> 5[5] [receive] via NET/FasTrak/5/GDRDMA +n116-040-168:3350751:3350861 [5] NCCL INFO Channel 05/0 : 5[5] -> 37[5] [send] via NET/FasTrak/5/GDRDMA +n116-040-168:3350748:3350865 [2] NCCL INFO Channel 02/0 : 34[2] -> 2[2] [receive] via NET/FasTrak/2/GDRDMA +n116-040-168:3350748:3350865 [2] NCCL INFO Channel 02/0 : 2[2] -> 34[2] [send] via NET/FasTrak/2/GDRDMA +n116-040-168:3350752:3350864 [6] NCCL INFO Channel 06/0 : 38[6] -> 6[6] [receive] via NET/FasTrak/6/GDRDMA +n116-040-168:3350752:3350864 [6] NCCL INFO Channel 06/0 : 6[6] -> 38[6] [send] via NET/FasTrak/6/GDRDMA +n116-040-168:3350753:3350860 [7] NCCL INFO Channel 07/0 : 39[7] -> 7[7] [receive] via NET/FasTrak/7/GDRDMA +n116-040-168:3350753:3350860 [7] NCCL INFO Channel 07/0 : 7[7] -> 39[7] [send] via NET/FasTrak/7/GDRDMA +n116-040-168:3350746:3350866 [0] NCCL INFO Channel 00/0 : 32[0] -> 0[0] [receive] via NET/FasTrak/0/GDRDMA +n116-040-168:3350746:3350866 [0] NCCL INFO Channel 00/0 : 0[0] -> 32[0] [send] via NET/FasTrak/0/GDRDMA +n116-040-168:3350746:3350866 [0] NCCL INFO Channel 08/0 : 8[0] -> 0[0] [receive] via NET/FasTrak/0/GDRDMA +n116-040-168:3350752:3350864 [6] NCCL INFO Channel 14/0 : 14[6] -> 6[6] [receive] via NET/FasTrak/6/GDRDMA +n116-040-168:3350750:3350859 [4] NCCL INFO Channel 12/0 : 12[4] -> 4[4] [receive] via NET/FasTrak/4/GDRDMA +n116-040-168:3350749:3350863 [3] NCCL INFO Channel 11/0 : 11[3] -> 3[3] [receive] via NET/FasTrak/3/GDRDMA +n116-040-168:3350748:3350865 [2] NCCL INFO Channel 10/0 : 10[2] -> 2[2] [receive] via NET/FasTrak/2/GDRDMA +n116-040-168:3350751:3350861 [5] NCCL INFO Channel 13/0 : 13[5] -> 5[5] [receive] via NET/FasTrak/5/GDRDMA +n116-040-168:3350747:3350862 [1] NCCL INFO Channel 09/0 : 9[1] -> 1[1] [receive] via NET/FasTrak/1/GDRDMA +n116-040-168:3350749:3350863 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/CUMEM +n116-040-168:3350749:3350863 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/CUMEM +n116-040-168:3350749:3350863 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/CUMEM +n116-040-168:3350749:3350863 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/CUMEM +n116-040-168:3350751:3350861 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/CUMEM +n116-040-168:3350749:3350863 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/CUMEM +n116-040-168:3350751:3350861 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/CUMEM +n116-040-168:3350747:3350862 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/CUMEM +n116-040-168:3350749:3350863 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/CUMEM +n116-040-168:3350751:3350861 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/CUMEM +n116-040-168:3350752:3350864 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/CUMEM +n116-040-168:3350747:3350862 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/CUMEM +n116-040-168:3350749:3350863 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/CUMEM +n116-040-168:3350751:3350861 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/CUMEM +n116-040-168:3350752:3350864 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/CUMEM +n116-040-168:3350747:3350862 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/CUMEM +n116-040-168:3350750:3350859 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/CUMEM +n116-040-168:3350749:3350863 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/CUMEM +n116-040-168:3350751:3350861 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/CUMEM +n116-040-168:3350752:3350864 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/CUMEM +n116-040-168:3350747:3350862 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/CUMEM +n116-040-168:3350750:3350859 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/CUMEM +n116-040-168:3350749:3350863 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/CUMEM +n116-040-168:3350751:3350861 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/CUMEM +n116-040-168:3350752:3350864 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/CUMEM +n116-040-168:3350747:3350862 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/CUMEM +n116-040-168:3350750:3350859 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/CUMEM +n116-040-168:3350749:3350863 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/CUMEM +n116-040-168:3350751:3350861 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/CUMEM +n116-040-168:3350752:3350864 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/CUMEM +n116-040-168:3350747:3350862 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/CUMEM +n116-040-168:3350750:3350859 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/CUMEM +n116-040-168:3350749:3350863 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/CUMEM +n116-040-168:3350751:3350861 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/CUMEM +n116-040-168:3350752:3350864 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/CUMEM +n116-040-168:3350747:3350862 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/CUMEM +n116-040-168:3350750:3350859 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/CUMEM +n116-040-168:3350749:3350863 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/CUMEM +n116-040-168:3350751:3350861 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/CUMEM +n116-040-168:3350752:3350864 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/CUMEM +n116-040-168:3350747:3350862 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/CUMEM +n116-040-168:3350750:3350859 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/CUMEM +n116-040-168:3350749:3350863 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/CUMEM +n116-040-168:3350751:3350861 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/CUMEM +n116-040-168:3350752:3350864 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/CUMEM +n116-040-168:3350747:3350862 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/CUMEM +n116-040-168:3350750:3350859 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/CUMEM +n116-040-168:3350749:3350863 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/CUMEM +n116-040-168:3350751:3350861 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/CUMEM +n116-040-168:3350748:3350865 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/CUMEM +n116-040-168:3350752:3350864 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/CUMEM +n116-040-168:3350747:3350862 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/CUMEM +n116-040-168:3350750:3350859 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/CUMEM +n116-040-168:3350749:3350863 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/CUMEM +n116-040-168:3350751:3350861 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/CUMEM +n116-040-168:3350748:3350865 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/CUMEM +n116-040-168:3350752:3350864 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/CUMEM +n116-040-168:3350747:3350862 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/CUMEM +n116-040-168:3350750:3350859 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/CUMEM +n116-040-168:3350749:3350863 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/CUMEM +n116-040-168:3350751:3350861 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/CUMEM +n116-040-168:3350748:3350865 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/CUMEM +n116-040-168:3350752:3350864 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/CUMEM +n116-040-168:3350747:3350862 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/CUMEM +n116-040-168:3350750:3350859 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/CUMEM +n116-040-168:3350751:3350861 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/CUMEM +n116-040-168:3350753:3350860 [7] NCCL INFO Channel 15/0 : 15[7] -> 7[7] [receive] via NET/FasTrak/7/GDRDMA +n116-040-168:3350748:3350865 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/CUMEM +n116-040-168:3350752:3350864 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/CUMEM +n116-040-168:3350747:3350862 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/CUMEM +n116-040-168:3350750:3350859 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/CUMEM +n116-040-168:3350751:3350861 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/CUMEM +n116-040-168:3350748:3350865 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/CUMEM +n116-040-168:3350747:3350862 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/CUMEM +n116-040-168:3350750:3350859 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/CUMEM +n116-040-168:3350751:3350861 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/CUMEM +n116-040-168:3350748:3350865 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/CUMEM +n116-040-168:3350747:3350862 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/CUMEM +n116-040-168:3350747:3350862 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/CUMEM +n116-040-168:3350748:3350865 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/CUMEM +n116-040-168:3350748:3350865 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/CUMEM +n116-040-168:3350748:3350865 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/CUMEM +n116-040-168:3350748:3350865 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/CUMEM +n116-040-168:3350753:3350860 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/CUMEM +n116-040-168:3350753:3350860 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/CUMEM +n116-040-168:3350748:3350865 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/CUMEM +n116-040-168:3350748:3350865 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/CUMEM +n116-040-168:3350753:3350860 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/CUMEM +n116-040-168:3350753:3350860 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/CUMEM +n116-040-168:3350753:3350860 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/CUMEM +n116-040-168:3350753:3350860 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/CUMEM +n116-040-168:3350753:3350860 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/CUMEM +n116-040-168:3350753:3350860 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/CUMEM +n116-040-168:3350753:3350860 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/CUMEM +n116-040-168:3350753:3350860 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/CUMEM +n116-040-168:3350753:3350860 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/CUMEM +n116-040-168:3350753:3350860 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/CUMEM +n116-040-168:3350753:3350860 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/CUMEM +n116-040-168:3350753:3350860 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/CUMEM +n116-040-168:3350753:3350860 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/CUMEM +n116-040-168:3350753:3350860 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/CUMEM +n116-040-168:3350753:3350860 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/CUMEM +n116-040-168:3350753:3350860 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/CUMEM +n116-040-168:3350753:3350860 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/CUMEM +n116-040-168:3350753:3350860 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/CUMEM +n116-040-168:3350753:3350860 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/CUMEM +n116-040-168:3350753:3350860 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/CUMEM +n116-040-168:3350753:3350860 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/CUMEM +n116-040-168:3350753:3350860 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/CUMEM +n116-040-168:3350753:3350860 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/CUMEM +n116-040-168:3350753:3350860 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/CUMEM +n116-040-168:3350753:3350860 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/CUMEM +n116-040-168:3350753:3350860 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/CUMEM +n116-040-168:3350748:3350865 [2] NCCL INFO Connected all trees +n116-040-168:3350746:3350866 [0] NCCL INFO Connected all trees +n116-040-168:3350747:3350862 [1] NCCL INFO Connected all trees +n116-040-168:3350753:3350860 [7] NCCL INFO Connected all trees +n116-040-168:3350749:3350863 [3] NCCL INFO Connected all trees +n116-040-168:3350750:3350859 [4] NCCL INFO Connected all trees +n116-040-168:3350751:3350861 [5] NCCL INFO Connected all trees +n116-040-168:3350752:3350864 [6] NCCL INFO Connected all trees +Namespace(batch_size=64, accum_iter=1, eval_batch_size=64, epochs=40, bce_loss=False, unscale_lr=True, select_models='scaledf_v4_train_fix.pkl', model='deit_huge_patch14_LS', input_size=224, drop=0.0, drop_path=0.2, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.1, sched='cosine', lr=1e-05, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-06, decay_epochs=15, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=False, train_mode=True, ThreeAugment=False, src=False, reprob=0.0, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/ScaleDF', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_64card_effbs4096_from_epoch22_snapshot_free8_master_n116-040-168_20260726_2350_vith64_port25264', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/checkpoint_snapshots/vith32_latest_for_64card_effbs4096_20260726_234604.pth', start_epoch=0, eval=False, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=64, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_huge_patch14_LS +n116-040-168:3350747:3350888 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/CUMEM +n116-040-168:3350747:3350888 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/CUMEM +n116-040-168:3350746:3350889 [0] NCCL INFO Channel 00/0 : 57[1] -> 0[0] [receive] via NET/FasTrak/0/GDRDMA +n116-040-168:3350746:3350889 [0] NCCL INFO Channel 08/0 : 57[1] -> 0[0] [receive] via NET/FasTrak/0/GDRDMA +n116-040-168:3350746:3350889 [0] NCCL INFO Channel 00/0 : 0[0] -> 7[7] via P2P/CUMEM +n116-040-168:3350746:3350889 [0] NCCL INFO Channel 08/0 : 0[0] -> 7[7] via P2P/CUMEM +n116-040-168:3350751:3350890 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/CUMEM +n116-040-168:3350751:3350890 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/CUMEM +n116-040-168:3350753:3350891 [7] NCCL INFO Channel 06/0 : 7[7] -> 14[6] [send] via NET/FasTrak/6/GDRDMA +n116-040-168:3350753:3350891 [7] NCCL INFO Channel 14/0 : 7[7] -> 14[6] [send] via NET/FasTrak/6/GDRDMA +n116-040-168:3350746:3350889 [0] NCCL INFO Channel 01/0 : 0[0] -> 9[1] [send] via NET/FasTrak/1/GDRDMA +n116-040-168:3350746:3350889 [0] NCCL INFO Channel 09/0 : 0[0] -> 9[1] [send] via NET/FasTrak/1/GDRDMA +n116-040-168:3350753:3350891 [7] NCCL INFO Channel 07/0 : 62[6] -> 7[7] [receive] via NET/FasTrak/7/GDRDMA +n116-040-168:3350753:3350891 [7] NCCL INFO Channel 15/0 : 62[6] -> 7[7] [receive] via NET/FasTrak/7/GDRDMA +n116-040-168:3350753:3350891 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/CUMEM +n116-040-168:3350753:3350891 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/CUMEM +n116-040-168:3350751:3350890 [5] NCCL INFO Channel 04/0 : 5[5] -> 12[4] [send] via NET/FasTrak/4/GDRDMA +n116-040-168:3350751:3350890 [5] NCCL INFO Channel 12/0 : 5[5] -> 12[4] [send] via NET/FasTrak/4/GDRDMA +n116-040-168:3350752:3350894 [6] NCCL INFO Channel 06/0 : 63[7] -> 6[6] [receive] via NET/FasTrak/6/GDRDMA +n116-040-168:3350751:3350890 [5] NCCL INFO Channel 05/0 : 60[4] -> 5[5] [receive] via NET/FasTrak/5/GDRDMA +n116-040-168:3350752:3350894 [6] NCCL INFO Channel 14/0 : 63[7] -> 6[6] [receive] via NET/FasTrak/6/GDRDMA +n116-040-168:3350751:3350890 [5] NCCL INFO Channel 13/0 : 60[4] -> 5[5] [receive] via NET/FasTrak/5/GDRDMA +n116-040-168:3350752:3350894 [6] NCCL INFO Channel 07/0 : 6[6] -> 15[7] [send] via NET/FasTrak/7/GDRDMA +n116-040-168:3350752:3350894 [6] NCCL INFO Channel 15/0 : 6[6] -> 15[7] [send] via NET/FasTrak/7/GDRDMA +n116-040-168:3350752:3350894 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/CUMEM +n116-040-168:3350752:3350894 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/CUMEM +n116-040-168:3350749:3350897 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/CUMEM +n116-040-168:3350749:3350897 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/CUMEM +n116-040-168:3350749:3350897 [3] NCCL INFO Channel 02/0 : 3[3] -> 10[2] [send] via NET/FasTrak/2/GDRDMA +n116-040-168:3350749:3350897 [3] NCCL INFO Channel 10/0 : 3[3] -> 10[2] [send] via NET/FasTrak/2/GDRDMA +n116-040-168:3350749:3350897 [3] NCCL INFO Channel 03/0 : 58[2] -> 3[3] [receive] via NET/FasTrak/3/GDRDMA +n116-040-168:3350750:3350887 [4] NCCL INFO Channel 04/0 : 61[5] -> 4[4] [receive] via NET/FasTrak/4/GDRDMA +n116-040-168:3350749:3350897 [3] NCCL INFO Channel 11/0 : 58[2] -> 3[3] [receive] via NET/FasTrak/3/GDRDMA +n116-040-168:3350750:3350887 [4] NCCL INFO Channel 12/0 : 61[5] -> 4[4] [receive] via NET/FasTrak/4/GDRDMA +n116-040-168:3350750:3350887 [4] NCCL INFO Channel 05/0 : 4[4] -> 13[5] [send] via NET/FasTrak/5/GDRDMA +n116-040-168:3350750:3350887 [4] NCCL INFO Channel 13/0 : 4[4] -> 13[5] [send] via NET/FasTrak/5/GDRDMA +n116-040-168:3350750:3350887 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/CUMEM +n116-040-168:3350750:3350887 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/CUMEM +n116-040-168:3350748:3350900 [2] NCCL INFO Channel 02/0 : 59[3] -> 2[2] [receive] via NET/FasTrak/2/GDRDMA +n116-040-168:3350748:3350900 [2] NCCL INFO Channel 10/0 : 59[3] -> 2[2] [receive] via NET/FasTrak/2/GDRDMA +n116-040-168:3350747:3350888 [1] NCCL INFO Channel 00/0 : 1[1] -> 8[0] [send] via NET/FasTrak/0/GDRDMA +n116-040-168:3350747:3350888 [1] NCCL INFO Channel 08/0 : 1[1] -> 8[0] [send] via NET/FasTrak/0/GDRDMA +n116-040-168:3350747:3350888 [1] NCCL INFO Channel 01/0 : 56[0] -> 1[1] [receive] via NET/FasTrak/1/GDRDMA +n116-040-168:3350747:3350888 [1] NCCL INFO Channel 09/0 : 56[0] -> 1[1] [receive] via NET/FasTrak/1/GDRDMA +n116-040-168:3350748:3350900 [2] NCCL INFO Channel 03/0 : 2[2] -> 11[3] [send] via NET/FasTrak/3/GDRDMA +n116-040-168:3350748:3350900 [2] NCCL INFO Channel 11/0 : 2[2] -> 11[3] [send] via NET/FasTrak/3/GDRDMA +n116-040-168:3350748:3350900 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/CUMEM +n116-040-168:3350748:3350900 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/CUMEM +n116-040-168:3350753:3350891 [7] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-040-168:3350752:3350894 [6] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-040-168:3350751:3350890 [5] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-040-168:3350750:3350887 [4] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-040-168:3350749:3350897 [3] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-040-168:3350748:3350900 [2] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-040-168:3350746:3350889 [0] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-040-168:3350747:3350888 [1] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +number of params: 630848002 +Start training for 40 epochs +Epoch: [23] [ 0/3634] eta: 16:32:34 lr: 0.000004 loss: 0.6041 (0.6041) time: 16.3882 data: 3.9490 max mem: 52124 +Epoch: [23] [ 10/3634] eta: 1:55:10 lr: 0.000004 loss: 0.5513 (0.5592) time: 1.9070 data: 0.3592 max mem: 52124 +Epoch: [23] [ 20/3634] eta: 1:11:09 lr: 0.000004 loss: 0.5438 (0.5550) time: 0.4209 data: 0.0002 max mem: 52124 +Epoch: [23] [ 30/3634] eta: 0:55:28 lr: 0.000004 loss: 0.5649 (0.5557) time: 0.3827 data: 0.0002 max mem: 52124 +Epoch: [23] [ 40/3634] eta: 0:47:24 lr: 0.000004 loss: 0.5407 (0.5446) time: 0.3823 data: 0.0002 max mem: 52124 +Epoch: [23] [ 50/3634] eta: 0:42:30 lr: 0.000004 loss: 0.5377 (0.5449) time: 0.3831 data: 0.0002 max mem: 52124 +Epoch: [23] [ 60/3634] eta: 0:39:10 lr: 0.000004 loss: 0.5433 (0.5417) time: 0.3834 data: 0.0002 max mem: 52124 +Epoch: [23] [ 70/3634] eta: 0:36:49 lr: 0.000004 loss: 0.5250 (0.5353) time: 0.3866 data: 0.0002 max mem: 52124 +Epoch: [23] [ 80/3634] eta: 0:34:59 lr: 0.000004 loss: 0.5250 (0.5349) time: 0.3867 data: 0.0002 max mem: 52124 +Epoch: [23] [ 90/3634] eta: 0:33:32 lr: 0.000004 loss: 0.5391 (0.5337) time: 0.3824 data: 0.0002 max mem: 52124 +Epoch: [23] [ 100/3634] eta: 0:32:21 lr: 0.000004 loss: 0.5476 (0.5351) time: 0.3815 data: 0.0002 max mem: 52124 +Epoch: [23] [ 110/3634] eta: 0:31:23 lr: 0.000004 loss: 0.5476 (0.5354) time: 0.3820 data: 0.0002 max mem: 52124 +Epoch: [23] [ 120/3634] eta: 0:30:33 lr: 0.000004 loss: 0.5321 (0.5356) time: 0.3828 data: 0.0002 max mem: 52124 +Epoch: [23] [ 130/3634] eta: 0:29:51 lr: 0.000004 loss: 0.5341 (0.5353) time: 0.3823 data: 0.0002 max mem: 52124 +Epoch: [23] [ 140/3634] eta: 0:29:14 lr: 0.000004 loss: 0.5035 (0.5325) time: 0.3833 data: 0.0002 max mem: 52124 +Epoch: [23] [ 150/3634] eta: 0:28:41 lr: 0.000004 loss: 0.4806 (0.5300) time: 0.3835 data: 0.0002 max mem: 52124 +Epoch: [23] [ 160/3634] eta: 0:28:12 lr: 0.000004 loss: 0.4969 (0.5306) time: 0.3821 data: 0.0002 max mem: 52124 +Epoch: [23] [ 170/3634] eta: 0:27:46 lr: 0.000004 loss: 0.5009 (0.5299) time: 0.3815 data: 0.0002 max mem: 52124 +Epoch: [23] [ 180/3634] eta: 0:27:22 lr: 0.000004 loss: 0.5144 (0.5289) time: 0.3817 data: 0.0002 max mem: 52124 +Epoch: [23] [ 190/3634] eta: 0:27:01 lr: 0.000004 loss: 0.5231 (0.5278) time: 0.3827 data: 0.0002 max mem: 52124 +Epoch: [23] [ 200/3634] eta: 0:26:41 lr: 0.000004 loss: 0.5137 (0.5264) time: 0.3820 data: 0.0002 max mem: 52124 +Epoch: [23] [ 210/3634] eta: 0:26:22 lr: 0.000004 loss: 0.5221 (0.5259) time: 0.3814 data: 0.0002 max mem: 52124 +Epoch: [23] [ 220/3634] eta: 0:26:05 lr: 0.000004 loss: 0.5123 (0.5243) time: 0.3819 data: 0.0002 max mem: 52124 +Epoch: [23] [ 230/3634] eta: 0:25:49 lr: 0.000004 loss: 0.5046 (0.5228) time: 0.3818 data: 0.0002 max mem: 52124 +Epoch: [23] [ 240/3634] eta: 0:25:35 lr: 0.000004 loss: 0.5044 (0.5219) time: 0.3819 data: 0.0002 max mem: 52124 +Epoch: [23] [ 250/3634] eta: 0:25:33 lr: 0.000004 loss: 0.5040 (0.5207) time: 0.4278 data: 0.0002 max mem: 52124 +Epoch: [23] [ 260/3634] eta: 0:25:19 lr: 0.000004 loss: 0.4722 (0.5191) time: 0.4275 data: 0.0002 max mem: 52124 +Epoch: [23] [ 270/3634] eta: 0:25:06 lr: 0.000004 loss: 0.4722 (0.5182) time: 0.3821 data: 0.0002 max mem: 52124 +Epoch: [23] [ 280/3634] eta: 0:25:24 lr: 0.000004 loss: 0.5318 (0.5190) time: 0.5082 data: 0.0002 max mem: 52124 +Epoch: [23] [ 290/3634] eta: 0:25:11 lr: 0.000004 loss: 0.5318 (0.5185) time: 0.5073 data: 0.0002 max mem: 52124 +Epoch: [23] [ 300/3634] eta: 0:24:59 lr: 0.000004 loss: 0.4673 (0.5170) time: 0.3810 data: 0.0002 max mem: 52124 +Epoch: [23] [ 310/3634] eta: 0:25:14 lr: 0.000004 loss: 0.4793 (0.5168) time: 0.5066 data: 0.0002 max mem: 52124 +Epoch: [23] [ 320/3634] eta: 0:25:01 lr: 0.000004 loss: 0.5432 (0.5178) time: 0.5073 data: 0.0002 max mem: 52124 +Epoch: [23] [ 330/3634] eta: 0:24:50 lr: 0.000004 loss: 0.5531 (0.5173) time: 0.3823 data: 0.0002 max mem: 52124 +Epoch: [23] [ 340/3634] eta: 0:24:51 lr: 0.000004 loss: 0.5273 (0.5179) time: 0.4443 data: 0.0002 max mem: 52124 +Epoch: [23] [ 350/3634] eta: 0:24:40 lr: 0.000004 loss: 0.5199 (0.5182) time: 0.4445 data: 0.0002 max mem: 52124 +Epoch: [23] [ 360/3634] eta: 0:24:37 lr: 0.000004 loss: 0.5161 (0.5181) time: 0.4275 data: 0.0002 max mem: 52124 +Epoch: [23] [ 370/3634] eta: 0:24:38 lr: 0.000004 loss: 0.5032 (0.5175) time: 0.4926 data: 0.0002 max mem: 52124 +Epoch: [23] [ 380/3634] eta: 0:24:27 lr: 0.000004 loss: 0.5322 (0.5180) time: 0.4473 data: 0.0002 max mem: 52124 +Epoch: [23] [ 390/3634] eta: 0:24:17 lr: 0.000004 loss: 0.5288 (0.5176) time: 0.3821 data: 0.0002 max mem: 52124 +Epoch: [23] [ 400/3634] eta: 0:24:17 lr: 0.000004 loss: 0.5524 (0.5184) time: 0.4452 data: 0.0002 max mem: 52124 +Epoch: [23] [ 410/3634] eta: 0:24:08 lr: 0.000004 loss: 0.5552 (0.5190) time: 0.4462 data: 0.0002 max mem: 52124 +Epoch: [23] [ 420/3634] eta: 0:23:58 lr: 0.000004 loss: 0.5398 (0.5188) time: 0.3830 data: 0.0001 max mem: 52124 +Epoch: [23] [ 430/3634] eta: 0:23:58 lr: 0.000004 loss: 0.5515 (0.5200) time: 0.4467 data: 0.0002 max mem: 52124 +Epoch: [23] [ 440/3634] eta: 0:23:49 lr: 0.000004 loss: 0.5533 (0.5200) time: 0.4466 data: 0.0002 max mem: 52124 +Epoch: [23] [ 450/3634] eta: 0:23:47 lr: 0.000004 loss: 0.5265 (0.5201) time: 0.4304 data: 0.0002 max mem: 52124 +Epoch: [23] [ 460/3634] eta: 0:23:38 lr: 0.000004 loss: 0.5154 (0.5197) time: 0.4306 data: 0.0002 max mem: 52124 +Epoch: [23] [ 470/3634] eta: 0:23:38 lr: 0.000004 loss: 0.4752 (0.5186) time: 0.4473 data: 0.0001 max mem: 52124 +Epoch: [23] [ 480/3634] eta: 0:23:29 lr: 0.000004 loss: 0.4756 (0.5181) time: 0.4470 data: 0.0001 max mem: 52124 +Epoch: [23] [ 490/3634] eta: 0:23:20 lr: 0.000004 loss: 0.5009 (0.5182) time: 0.3825 data: 0.0002 max mem: 52124 +Epoch: [23] [ 500/3634] eta: 0:23:12 lr: 0.000004 loss: 0.5322 (0.5184) time: 0.3827 data: 0.0002 max mem: 52124 +Epoch: [23] [ 510/3634] eta: 0:23:04 lr: 0.000004 loss: 0.5452 (0.5190) time: 0.3829 data: 0.0002 max mem: 52124 +Epoch: [23] [ 520/3634] eta: 0:22:56 lr: 0.000004 loss: 0.5452 (0.5193) time: 0.3834 data: 0.0002 max mem: 52124 +Epoch: [23] [ 530/3634] eta: 0:22:48 lr: 0.000004 loss: 0.5368 (0.5185) time: 0.3829 data: 0.0002 max mem: 52124 +Epoch: [23] [ 540/3634] eta: 0:22:40 lr: 0.000004 loss: 0.5184 (0.5185) time: 0.3825 data: 0.0002 max mem: 52124 +Epoch: [23] [ 550/3634] eta: 0:22:32 lr: 0.000004 loss: 0.5197 (0.5183) time: 0.3828 data: 0.0002 max mem: 52124 +Epoch: [23] [ 560/3634] eta: 0:22:25 lr: 0.000004 loss: 0.5475 (0.5187) time: 0.3833 data: 0.0002 max mem: 52124 +Epoch: [23] [ 570/3634] eta: 0:22:18 lr: 0.000004 loss: 0.5329 (0.5188) time: 0.3830 data: 0.0002 max mem: 52124 +Epoch: [23] [ 580/3634] eta: 0:22:10 lr: 0.000004 loss: 0.5368 (0.5197) time: 0.3825 data: 0.0001 max mem: 52124 +Epoch: [23] [ 590/3634] eta: 0:22:03 lr: 0.000004 loss: 0.5740 (0.5200) time: 0.3830 data: 0.0001 max mem: 52124 +Epoch: [23] [ 600/3634] eta: 0:21:56 lr: 0.000004 loss: 0.5426 (0.5197) time: 0.3829 data: 0.0001 max mem: 52124 +Epoch: [23] [ 610/3634] eta: 0:21:50 lr: 0.000004 loss: 0.4929 (0.5193) time: 0.3825 data: 0.0001 max mem: 52124 +Epoch: [23] [ 620/3634] eta: 0:21:43 lr: 0.000004 loss: 0.5341 (0.5200) time: 0.3830 data: 0.0001 max mem: 52124 +Epoch: [23] [ 630/3634] eta: 0:21:36 lr: 0.000004 loss: 0.5370 (0.5203) time: 0.3844 data: 0.0002 max mem: 52124 +Epoch: [23] [ 640/3634] eta: 0:21:30 lr: 0.000004 loss: 0.5224 (0.5202) time: 0.3836 data: 0.0002 max mem: 52124 +Epoch: [23] [ 650/3634] eta: 0:21:23 lr: 0.000004 loss: 0.5066 (0.5199) time: 0.3816 data: 0.0002 max mem: 52124 +Epoch: [23] [ 660/3634] eta: 0:21:17 lr: 0.000004 loss: 0.4878 (0.5193) time: 0.3824 data: 0.0002 max mem: 52124 +Epoch: [23] [ 670/3634] eta: 0:21:10 lr: 0.000004 loss: 0.5267 (0.5199) time: 0.3824 data: 0.0002 max mem: 52124 +Epoch: [23] [ 680/3634] eta: 0:21:04 lr: 0.000004 loss: 0.5598 (0.5201) time: 0.3818 data: 0.0002 max mem: 52124 +Epoch: [23] [ 690/3634] eta: 0:20:58 lr: 0.000004 loss: 0.5079 (0.5199) time: 0.3820 data: 0.0002 max mem: 52124 +Epoch: [23] [ 700/3634] eta: 0:20:51 lr: 0.000004 loss: 0.4940 (0.5195) time: 0.3817 data: 0.0002 max mem: 52124 +Epoch: [23] [ 710/3634] eta: 0:20:45 lr: 0.000004 loss: 0.5127 (0.5195) time: 0.3818 data: 0.0002 max mem: 52124 +Epoch: [23] [ 720/3634] eta: 0:20:39 lr: 0.000004 loss: 0.5039 (0.5190) time: 0.3829 data: 0.0002 max mem: 52124 +Epoch: [23] [ 730/3634] eta: 0:20:33 lr: 0.000004 loss: 0.5166 (0.5192) time: 0.3829 data: 0.0002 max mem: 52124 +Epoch: [23] [ 740/3634] eta: 0:20:27 lr: 0.000004 loss: 0.5406 (0.5195) time: 0.3818 data: 0.0002 max mem: 52124 +Epoch: [23] [ 750/3634] eta: 0:20:22 lr: 0.000004 loss: 0.5402 (0.5191) time: 0.3825 data: 0.0001 max mem: 52124 +Epoch: [23] [ 760/3634] eta: 0:20:16 lr: 0.000004 loss: 0.5216 (0.5193) time: 0.3832 data: 0.0002 max mem: 52124 +Epoch: [23] [ 770/3634] eta: 0:20:10 lr: 0.000004 loss: 0.5184 (0.5194) time: 0.3829 data: 0.0002 max mem: 52124 +Epoch: [23] [ 780/3634] eta: 0:20:04 lr: 0.000004 loss: 0.5325 (0.5194) time: 0.3824 data: 0.0001 max mem: 52124 +Epoch: [23] [ 790/3634] eta: 0:19:59 lr: 0.000004 loss: 0.5325 (0.5195) time: 0.3827 data: 0.0002 max mem: 52124 +Epoch: [23] [ 800/3634] eta: 0:19:57 lr: 0.000004 loss: 0.5337 (0.5196) time: 0.4422 data: 0.0002 max mem: 52124 +Epoch: [23] [ 810/3634] eta: 0:19:55 lr: 0.000004 loss: 0.4999 (0.5186) time: 0.4862 data: 0.0002 max mem: 52124 +Epoch: [23] [ 820/3634] eta: 0:19:49 lr: 0.000004 loss: 0.5162 (0.5191) time: 0.4273 data: 0.0002 max mem: 52124 +Epoch: [23] [ 830/3634] eta: 0:19:47 lr: 0.000004 loss: 0.4985 (0.5183) time: 0.4272 data: 0.0002 max mem: 52124 +Epoch: [23] [ 840/3634] eta: 0:19:47 lr: 0.000004 loss: 0.4592 (0.5182) time: 0.5188 data: 0.0002 max mem: 52124 +Epoch: [23] [ 850/3634] eta: 0:19:45 lr: 0.000004 loss: 0.5068 (0.5180) time: 0.5337 data: 0.0001 max mem: 52124 +Epoch: [23] [ 860/3634] eta: 0:19:43 lr: 0.000004 loss: 0.5477 (0.5181) time: 0.4927 data: 0.0002 max mem: 52124 +Epoch: [23] [ 870/3634] eta: 0:19:40 lr: 0.000004 loss: 0.5174 (0.5179) time: 0.4781 data: 0.0002 max mem: 52124 +Epoch: [23] [ 880/3634] eta: 0:19:41 lr: 0.000004 loss: 0.4986 (0.5177) time: 0.5359 data: 0.0001 max mem: 52124 +Epoch: [23] [ 890/3634] eta: 0:19:43 lr: 0.000004 loss: 0.5108 (0.5175) time: 0.6091 data: 0.0001 max mem: 52124 +Epoch: [23] [ 900/3634] eta: 0:19:47 lr: 0.000004 loss: 0.5205 (0.5175) time: 0.6646 data: 0.0001 max mem: 52124 +Epoch: [23] [ 910/3634] eta: 0:19:48 lr: 0.000004 loss: 0.5225 (0.5174) time: 0.6569 data: 0.0001 max mem: 52124 +Epoch: [23] [ 920/3634] eta: 0:19:44 lr: 0.000004 loss: 0.5393 (0.5176) time: 0.5382 data: 0.0001 max mem: 52124 +Epoch: [23] [ 930/3634] eta: 0:19:44 lr: 0.000004 loss: 0.5588 (0.5179) time: 0.5198 data: 0.0001 max mem: 52124 +Epoch: [23] [ 940/3634] eta: 0:19:44 lr: 0.000004 loss: 0.5508 (0.5178) time: 0.5784 data: 0.0001 max mem: 52124 +Epoch: [23] [ 950/3634] eta: 0:19:41 lr: 0.000004 loss: 0.4597 (0.5172) time: 0.5448 data: 0.0001 max mem: 52124 +Epoch: [23] [ 960/3634] eta: 0:19:38 lr: 0.000004 loss: 0.4597 (0.5171) time: 0.4864 data: 0.0001 max mem: 52124 +Epoch: [23] [ 970/3634] eta: 0:19:37 lr: 0.000004 loss: 0.5198 (0.5174) time: 0.5320 data: 0.0001 max mem: 52124 +Epoch: [23] [ 980/3634] eta: 0:19:36 lr: 0.000004 loss: 0.5460 (0.5175) time: 0.5776 data: 0.0002 max mem: 52124 +Epoch: [23] [ 990/3634] eta: 0:19:33 lr: 0.000004 loss: 0.5516 (0.5176) time: 0.5189 data: 0.0002 max mem: 52124 +Epoch: [23] [1000/3634] eta: 0:19:30 lr: 0.000004 loss: 0.5516 (0.5178) time: 0.4870 data: 0.0002 max mem: 52124 +Epoch: [23] [1010/3634] eta: 0:19:27 lr: 0.000004 loss: 0.5536 (0.5180) time: 0.5019 data: 0.0002 max mem: 52124 +Epoch: [23] [1020/3634] eta: 0:19:29 lr: 0.000004 loss: 0.5439 (0.5181) time: 0.6074 data: 0.0002 max mem: 52124 +Epoch: [23] [1030/3634] eta: 0:19:23 lr: 0.000004 loss: 0.4958 (0.5177) time: 0.5465 data: 0.0002 max mem: 52124 +Epoch: [23] [1040/3634] eta: 0:19:19 lr: 0.000004 loss: 0.5030 (0.5179) time: 0.4305 data: 0.0002 max mem: 52124 +Epoch: [23] [1050/3634] eta: 0:19:21 lr: 0.000004 loss: 0.5464 (0.5180) time: 0.5935 data: 0.0002 max mem: 52124 +Epoch: [23] [1060/3634] eta: 0:19:19 lr: 0.000004 loss: 0.5408 (0.5181) time: 0.6339 data: 0.0002 max mem: 52124 +Epoch: [23] [1070/3634] eta: 0:19:13 lr: 0.000004 loss: 0.5382 (0.5181) time: 0.4715 data: 0.0002 max mem: 52124 +Epoch: [23] [1080/3634] eta: 0:19:12 lr: 0.000004 loss: 0.5408 (0.5182) time: 0.4751 data: 0.0001 max mem: 52124 +Epoch: [23] [1090/3634] eta: 0:19:05 lr: 0.000004 loss: 0.5383 (0.5182) time: 0.4750 data: 0.0002 max mem: 52124 +Epoch: [23] [1100/3634] eta: 0:19:01 lr: 0.000004 loss: 0.5114 (0.5182) time: 0.4277 data: 0.0002 max mem: 52124 +Epoch: [23] [1110/3634] eta: 0:18:55 lr: 0.000004 loss: 0.4896 (0.5182) time: 0.4282 data: 0.0002 max mem: 52124 +Epoch: [23] [1120/3634] eta: 0:18:52 lr: 0.000004 loss: 0.5470 (0.5184) time: 0.4474 data: 0.0002 max mem: 52124 +Epoch: [23] [1130/3634] eta: 0:18:49 lr: 0.000004 loss: 0.5605 (0.5187) time: 0.5068 data: 0.0002 max mem: 52124 +Epoch: [23] [1140/3634] eta: 0:18:46 lr: 0.000004 loss: 0.5288 (0.5186) time: 0.5034 data: 0.0002 max mem: 52124 +Epoch: [23] [1150/3634] eta: 0:18:40 lr: 0.000004 loss: 0.5048 (0.5184) time: 0.4436 data: 0.0001 max mem: 52124 +Epoch: [23] [1160/3634] eta: 0:18:34 lr: 0.000004 loss: 0.5082 (0.5184) time: 0.3830 data: 0.0002 max mem: 52124 +Epoch: [23] [1170/3634] eta: 0:18:28 lr: 0.000004 loss: 0.5175 (0.5183) time: 0.3830 data: 0.0002 max mem: 52124 +Epoch: [23] [1180/3634] eta: 0:18:22 lr: 0.000004 loss: 0.5630 (0.5188) time: 0.3826 data: 0.0002 max mem: 52124 +Epoch: [23] [1190/3634] eta: 0:18:16 lr: 0.000004 loss: 0.5575 (0.5190) time: 0.3823 data: 0.0001 max mem: 52124 +Epoch: [23] [1200/3634] eta: 0:18:10 lr: 0.000004 loss: 0.5320 (0.5190) time: 0.3823 data: 0.0001 max mem: 52124 +Epoch: [23] [1210/3634] eta: 0:18:04 lr: 0.000004 loss: 0.5498 (0.5191) time: 0.3826 data: 0.0001 max mem: 52124 +Epoch: [23] [1220/3634] eta: 0:17:59 lr: 0.000004 loss: 0.5444 (0.5190) time: 0.3823 data: 0.0001 max mem: 52124 +Epoch: [23] [1230/3634] eta: 0:17:53 lr: 0.000004 loss: 0.5248 (0.5192) time: 0.3826 data: 0.0001 max mem: 52124 +Epoch: [23] [1240/3634] eta: 0:17:47 lr: 0.000004 loss: 0.5248 (0.5191) time: 0.3829 data: 0.0001 max mem: 52124 +Epoch: [23] [1250/3634] eta: 0:17:42 lr: 0.000004 loss: 0.5157 (0.5189) time: 0.3823 data: 0.0002 max mem: 52124 +Epoch: [23] [1260/3634] eta: 0:17:36 lr: 0.000004 loss: 0.4967 (0.5190) time: 0.3825 data: 0.0002 max mem: 52124 +Epoch: [23] [1270/3634] eta: 0:17:30 lr: 0.000004 loss: 0.5261 (0.5192) time: 0.3824 data: 0.0002 max mem: 52124 +Epoch: [23] [1280/3634] eta: 0:17:25 lr: 0.000004 loss: 0.5432 (0.5192) time: 0.3821 data: 0.0002 max mem: 52124 +Epoch: [23] [1290/3634] eta: 0:17:19 lr: 0.000004 loss: 0.5624 (0.5195) time: 0.3832 data: 0.0001 max mem: 52124 +Epoch: [23] [1300/3634] eta: 0:17:14 lr: 0.000004 loss: 0.5624 (0.5197) time: 0.3832 data: 0.0001 max mem: 52124 +Epoch: [23] [1310/3634] eta: 0:17:08 lr: 0.000004 loss: 0.5216 (0.5194) time: 0.3822 data: 0.0001 max mem: 52124 +Epoch: [23] [1320/3634] eta: 0:17:03 lr: 0.000004 loss: 0.5030 (0.5192) time: 0.3822 data: 0.0002 max mem: 52124 +Epoch: [23] [1330/3634] eta: 0:16:59 lr: 0.000004 loss: 0.5030 (0.5193) time: 0.4260 data: 0.0002 max mem: 52124 +Epoch: [23] [1340/3634] eta: 0:16:53 lr: 0.000004 loss: 0.5524 (0.5195) time: 0.4258 data: 0.0002 max mem: 52124 +Epoch: [23] [1350/3634] eta: 0:16:48 lr: 0.000004 loss: 0.5571 (0.5195) time: 0.3822 data: 0.0002 max mem: 52124 +Epoch: [23] [1360/3634] eta: 0:16:47 lr: 0.000004 loss: 0.5556 (0.5197) time: 0.5342 data: 0.0001 max mem: 52124 +Epoch: [23] [1370/3634] eta: 0:16:44 lr: 0.000004 loss: 0.5432 (0.5197) time: 0.5959 data: 0.0002 max mem: 52124 +Epoch: [23] [1380/3634] eta: 0:16:39 lr: 0.000004 loss: 0.5087 (0.5199) time: 0.4439 data: 0.0002 max mem: 52124 +Epoch: [23] [1390/3634] eta: 0:16:35 lr: 0.000004 loss: 0.5486 (0.5200) time: 0.4412 data: 0.0001 max mem: 52124 +Epoch: [23] [1400/3634] eta: 0:16:32 lr: 0.000004 loss: 0.5392 (0.5200) time: 0.5019 data: 0.0001 max mem: 52124 +Epoch: [23] [1410/3634] eta: 0:16:28 lr: 0.000004 loss: 0.5425 (0.5202) time: 0.4921 data: 0.0001 max mem: 52124 +Epoch: [23] [1420/3634] eta: 0:16:25 lr: 0.000004 loss: 0.5425 (0.5202) time: 0.5242 data: 0.0002 max mem: 52124 +Epoch: [23] [1430/3634] eta: 0:16:20 lr: 0.000004 loss: 0.5109 (0.5202) time: 0.4748 data: 0.0002 max mem: 52124 +Epoch: [23] [1440/3634] eta: 0:16:21 lr: 0.000004 loss: 0.5227 (0.5201) time: 0.6000 data: 0.0002 max mem: 52124 +Epoch: [23] [1450/3634] eta: 0:16:24 lr: 0.000004 loss: 0.5283 (0.5201) time: 0.8862 data: 0.0002 max mem: 52124 +Epoch: [23] [1460/3634] eta: 0:16:22 lr: 0.000004 loss: 0.5360 (0.5202) time: 0.7741 data: 0.0001 max mem: 52124 +Epoch: [23] [1470/3634] eta: 0:16:20 lr: 0.000004 loss: 0.5420 (0.5202) time: 0.6232 data: 0.0001 max mem: 52124 +Epoch: [23] [1480/3634] eta: 0:16:20 lr: 0.000004 loss: 0.5030 (0.5199) time: 0.7101 data: 0.0001 max mem: 52124 +Epoch: [23] [1490/3634] eta: 0:16:15 lr: 0.000004 loss: 0.4717 (0.5195) time: 0.5746 data: 0.0002 max mem: 52124 +Epoch: [23] [1500/3634] eta: 0:16:10 lr: 0.000004 loss: 0.4801 (0.5194) time: 0.4290 data: 0.0002 max mem: 52124 +Epoch: [23] [1510/3634] eta: 0:16:05 lr: 0.000004 loss: 0.5366 (0.5194) time: 0.4288 data: 0.0002 max mem: 52124 +Epoch: [23] [1520/3634] eta: 0:16:03 lr: 0.000004 loss: 0.5475 (0.5196) time: 0.4998 data: 0.0002 max mem: 52124 +Epoch: [23] [1530/3634] eta: 0:15:59 lr: 0.000004 loss: 0.5218 (0.5194) time: 0.5602 data: 0.0001 max mem: 52124 +Epoch: [23] [1540/3634] eta: 0:15:57 lr: 0.000004 loss: 0.5194 (0.5196) time: 0.5798 data: 0.0002 max mem: 52124 +Epoch: [23] [1550/3634] eta: 0:15:54 lr: 0.000004 loss: 0.5194 (0.5194) time: 0.6264 data: 0.0002 max mem: 52124 +Epoch: [23] [1560/3634] eta: 0:15:49 lr: 0.000004 loss: 0.4968 (0.5194) time: 0.4887 data: 0.0002 max mem: 52124 +Epoch: [23] [1570/3634] eta: 0:15:45 lr: 0.000004 loss: 0.4818 (0.5191) time: 0.4418 data: 0.0001 max mem: 52124 +Epoch: [23] [1580/3634] eta: 0:15:40 lr: 0.000004 loss: 0.4730 (0.5190) time: 0.4934 data: 0.0001 max mem: 52124 +Epoch: [23] [1590/3634] eta: 0:15:37 lr: 0.000004 loss: 0.5387 (0.5192) time: 0.5382 data: 0.0002 max mem: 52124 +Epoch: [23] [1600/3634] eta: 0:15:33 lr: 0.000004 loss: 0.5402 (0.5193) time: 0.5325 data: 0.0002 max mem: 52124 +Epoch: [23] [1610/3634] eta: 0:15:27 lr: 0.000004 loss: 0.5356 (0.5194) time: 0.4280 data: 0.0002 max mem: 52124 +Epoch: [23] [1620/3634] eta: 0:15:23 lr: 0.000004 loss: 0.4872 (0.5192) time: 0.4303 data: 0.0002 max mem: 52124 +Epoch: [23] [1630/3634] eta: 0:15:19 lr: 0.000004 loss: 0.5099 (0.5192) time: 0.4912 data: 0.0002 max mem: 52124 +Epoch: [23] [1640/3634] eta: 0:15:17 lr: 0.000004 loss: 0.5192 (0.5191) time: 0.5739 data: 0.0002 max mem: 52124 +Epoch: [23] [1650/3634] eta: 0:15:13 lr: 0.000004 loss: 0.4831 (0.5188) time: 0.5722 data: 0.0002 max mem: 52124 +Epoch: [23] [1660/3634] eta: 0:15:08 lr: 0.000004 loss: 0.4730 (0.5185) time: 0.4852 data: 0.0001 max mem: 52124 +Epoch: [23] [1670/3634] eta: 0:15:03 lr: 0.000004 loss: 0.4920 (0.5186) time: 0.4257 data: 0.0001 max mem: 52124 +Epoch: [23] [1680/3634] eta: 0:15:00 lr: 0.000004 loss: 0.5388 (0.5185) time: 0.5167 data: 0.0001 max mem: 52124 +Epoch: [23] [1690/3634] eta: 0:14:56 lr: 0.000004 loss: 0.5388 (0.5185) time: 0.5608 data: 0.0001 max mem: 52124 +Epoch: [23] [1700/3634] eta: 0:14:50 lr: 0.000004 loss: 0.5518 (0.5189) time: 0.4255 data: 0.0001 max mem: 52124 +Epoch: [23] [1710/3634] eta: 0:14:46 lr: 0.000004 loss: 0.5478 (0.5188) time: 0.4407 data: 0.0002 max mem: 52124 +Epoch: [23] [1720/3634] eta: 0:14:41 lr: 0.000004 loss: 0.5184 (0.5187) time: 0.4405 data: 0.0002 max mem: 52124 +Epoch: [23] [1730/3634] eta: 0:14:35 lr: 0.000004 loss: 0.5339 (0.5188) time: 0.3815 data: 0.0001 max mem: 52124 +Epoch: [23] [1740/3634] eta: 0:14:30 lr: 0.000004 loss: 0.5334 (0.5187) time: 0.3821 data: 0.0001 max mem: 52124 +Epoch: [23] [1750/3634] eta: 0:14:26 lr: 0.000004 loss: 0.4948 (0.5186) time: 0.4429 data: 0.0002 max mem: 52124 +Epoch: [23] [1760/3634] eta: 0:14:22 lr: 0.000004 loss: 0.4931 (0.5186) time: 0.5074 data: 0.0001 max mem: 52124 +Epoch: [23] [1770/3634] eta: 0:14:17 lr: 0.000004 loss: 0.5549 (0.5187) time: 0.5060 data: 0.0001 max mem: 52124 +Epoch: [23] [1780/3634] eta: 0:14:12 lr: 0.000004 loss: 0.5324 (0.5186) time: 0.4413 data: 0.0002 max mem: 52124 +Epoch: [23] [1790/3634] eta: 0:14:07 lr: 0.000004 loss: 0.5089 (0.5186) time: 0.3819 data: 0.0002 max mem: 52124 +Epoch: [23] [1800/3634] eta: 0:14:01 lr: 0.000004 loss: 0.5327 (0.5188) time: 0.3815 data: 0.0002 max mem: 52124 +Epoch: [23] [1810/3634] eta: 0:13:56 lr: 0.000004 loss: 0.5430 (0.5187) time: 0.3821 data: 0.0002 max mem: 52124 +Epoch: [23] [1820/3634] eta: 0:13:50 lr: 0.000004 loss: 0.5353 (0.5187) time: 0.3825 data: 0.0002 max mem: 52124 +Epoch: [23] [1830/3634] eta: 0:13:45 lr: 0.000004 loss: 0.5361 (0.5189) time: 0.3820 data: 0.0002 max mem: 52124 +Epoch: [23] [1840/3634] eta: 0:13:40 lr: 0.000004 loss: 0.5469 (0.5189) time: 0.3819 data: 0.0002 max mem: 52124 +Epoch: [23] [1850/3634] eta: 0:13:34 lr: 0.000004 loss: 0.4889 (0.5188) time: 0.3815 data: 0.0002 max mem: 52124 +Epoch: [23] [1860/3634] eta: 0:13:29 lr: 0.000004 loss: 0.4831 (0.5187) time: 0.3818 data: 0.0002 max mem: 52124 +Epoch: [23] [1870/3634] eta: 0:13:25 lr: 0.000004 loss: 0.5014 (0.5186) time: 0.4420 data: 0.0002 max mem: 52124 +Epoch: [23] [1880/3634] eta: 0:13:20 lr: 0.000004 loss: 0.5431 (0.5189) time: 0.4417 data: 0.0001 max mem: 52124 +Epoch: [23] [1890/3634] eta: 0:13:15 lr: 0.000004 loss: 0.5379 (0.5187) time: 0.3811 data: 0.0001 max mem: 52124 +Epoch: [23] [1900/3634] eta: 0:13:10 lr: 0.000004 loss: 0.5271 (0.5187) time: 0.4294 data: 0.0002 max mem: 52124 +Epoch: [23] [1910/3634] eta: 0:13:06 lr: 0.000004 loss: 0.5391 (0.5188) time: 0.4765 data: 0.0002 max mem: 52124 +Epoch: [23] [1920/3634] eta: 0:13:01 lr: 0.000004 loss: 0.4948 (0.5184) time: 0.4280 data: 0.0001 max mem: 52124 +Epoch: [23] [1930/3634] eta: 0:12:57 lr: 0.000004 loss: 0.4875 (0.5184) time: 0.4851 data: 0.0001 max mem: 52124 +Epoch: [23] [1940/3634] eta: 0:12:54 lr: 0.000004 loss: 0.5206 (0.5184) time: 0.5912 data: 0.0001 max mem: 52124 +Epoch: [23] [1950/3634] eta: 0:12:50 lr: 0.000004 loss: 0.5164 (0.5183) time: 0.5945 data: 0.0001 max mem: 52124 +Epoch: [23] [1960/3634] eta: 0:12:46 lr: 0.000004 loss: 0.5143 (0.5182) time: 0.5338 data: 0.0002 max mem: 52124 +Epoch: [23] [1970/3634] eta: 0:12:44 lr: 0.000004 loss: 0.5143 (0.5181) time: 0.6107 data: 0.0002 max mem: 52124 +Epoch: [23] [1980/3634] eta: 0:12:40 lr: 0.000004 loss: 0.5178 (0.5181) time: 0.6757 data: 0.0001 max mem: 52124 +Epoch: [23] [1990/3634] eta: 0:12:36 lr: 0.000004 loss: 0.5380 (0.5181) time: 0.5371 data: 0.0001 max mem: 52124 +Epoch: [23] [2000/3634] eta: 0:12:32 lr: 0.000004 loss: 0.5002 (0.5180) time: 0.5175 data: 0.0001 max mem: 52124 +Epoch: [23] [2010/3634] eta: 0:12:28 lr: 0.000004 loss: 0.5093 (0.5180) time: 0.5169 data: 0.0001 max mem: 52124 +Epoch: [23] [2020/3634] eta: 0:12:24 lr: 0.000004 loss: 0.5051 (0.5179) time: 0.5174 data: 0.0002 max mem: 52124 +Epoch: [23] [2030/3634] eta: 0:12:22 lr: 0.000004 loss: 0.5008 (0.5179) time: 0.6812 data: 0.0002 max mem: 52124 +Epoch: [23] [2040/3634] eta: 0:12:17 lr: 0.000004 loss: 0.5503 (0.5180) time: 0.6343 data: 0.0001 max mem: 52124 +Epoch: [23] [2050/3634] eta: 0:12:14 lr: 0.000004 loss: 0.5295 (0.5180) time: 0.5483 data: 0.0001 max mem: 52124 +Epoch: [23] [2060/3634] eta: 0:12:10 lr: 0.000004 loss: 0.5010 (0.5179) time: 0.6091 data: 0.0001 max mem: 52124 +Epoch: [23] [2070/3634] eta: 0:12:07 lr: 0.000004 loss: 0.5010 (0.5179) time: 0.6080 data: 0.0002 max mem: 52124 +Epoch: [23] [2080/3634] eta: 0:12:02 lr: 0.000004 loss: 0.5453 (0.5179) time: 0.5469 data: 0.0002 max mem: 52124 +Epoch: [23] [2090/3634] eta: 0:11:58 lr: 0.000004 loss: 0.5453 (0.5181) time: 0.5167 data: 0.0001 max mem: 52124 +Epoch: [23] [2100/3634] eta: 0:11:54 lr: 0.000004 loss: 0.5205 (0.5180) time: 0.5625 data: 0.0001 max mem: 52124 +Epoch: [23] [2110/3634] eta: 0:11:49 lr: 0.000004 loss: 0.5265 (0.5181) time: 0.4715 data: 0.0002 max mem: 52124 +Epoch: [23] [2120/3634] eta: 0:11:45 lr: 0.000004 loss: 0.5390 (0.5181) time: 0.4414 data: 0.0002 max mem: 52124 +Epoch: [23] [2130/3634] eta: 0:11:40 lr: 0.000004 loss: 0.5378 (0.5181) time: 0.5004 data: 0.0002 max mem: 52124 +Epoch: [23] [2140/3634] eta: 0:11:36 lr: 0.000004 loss: 0.5488 (0.5183) time: 0.4895 data: 0.0002 max mem: 52124 +Epoch: [23] [2150/3634] eta: 0:11:32 lr: 0.000004 loss: 0.5488 (0.5184) time: 0.5272 data: 0.0002 max mem: 52124 +Epoch: [23] [2160/3634] eta: 0:11:27 lr: 0.000004 loss: 0.5479 (0.5185) time: 0.5244 data: 0.0002 max mem: 52124 +Epoch: [23] [2170/3634] eta: 0:11:23 lr: 0.000004 loss: 0.5479 (0.5186) time: 0.4767 data: 0.0002 max mem: 52124 +Epoch: [23] [2180/3634] eta: 0:11:19 lr: 0.000004 loss: 0.5331 (0.5186) time: 0.5357 data: 0.0002 max mem: 52124 +Epoch: [23] [2190/3634] eta: 0:11:15 lr: 0.000004 loss: 0.5389 (0.5187) time: 0.5904 data: 0.0002 max mem: 52124 +Epoch: [23] [2200/3634] eta: 0:11:10 lr: 0.000004 loss: 0.5220 (0.5186) time: 0.5309 data: 0.0002 max mem: 52124 +Epoch: [23] [2210/3634] eta: 0:11:05 lr: 0.000004 loss: 0.4844 (0.5186) time: 0.4274 data: 0.0001 max mem: 52124 +Epoch: [23] [2220/3634] eta: 0:11:00 lr: 0.000004 loss: 0.4862 (0.5184) time: 0.3821 data: 0.0002 max mem: 52124 +Epoch: [23] [2230/3634] eta: 0:10:55 lr: 0.000004 loss: 0.4918 (0.5182) time: 0.4249 data: 0.0002 max mem: 52124 +Epoch: [23] [2240/3634] eta: 0:10:52 lr: 0.000004 loss: 0.5016 (0.5183) time: 0.5746 data: 0.0001 max mem: 52124 +Epoch: [23] [2250/3634] eta: 0:10:47 lr: 0.000004 loss: 0.5010 (0.5182) time: 0.5319 data: 0.0001 max mem: 52124 +Epoch: [23] [2260/3634] eta: 0:10:42 lr: 0.000004 loss: 0.4983 (0.5182) time: 0.4255 data: 0.0002 max mem: 52124 +Epoch: [23] [2270/3634] eta: 0:10:38 lr: 0.000004 loss: 0.5094 (0.5182) time: 0.4844 data: 0.0002 max mem: 52124 +Epoch: [23] [2280/3634] eta: 0:10:33 lr: 0.000004 loss: 0.5185 (0.5183) time: 0.4854 data: 0.0001 max mem: 52124 +Epoch: [23] [2290/3634] eta: 0:10:28 lr: 0.000004 loss: 0.5185 (0.5182) time: 0.4261 data: 0.0001 max mem: 52124 +Epoch: [23] [2300/3634] eta: 0:10:23 lr: 0.000004 loss: 0.5225 (0.5181) time: 0.3819 data: 0.0002 max mem: 52124 +Epoch: [23] [2310/3634] eta: 0:10:18 lr: 0.000004 loss: 0.5225 (0.5180) time: 0.4445 data: 0.0002 max mem: 52124 +Epoch: [23] [2320/3634] eta: 0:10:13 lr: 0.000004 loss: 0.5216 (0.5182) time: 0.4444 data: 0.0002 max mem: 52124 +Epoch: [23] [2330/3634] eta: 0:10:08 lr: 0.000004 loss: 0.5510 (0.5184) time: 0.3818 data: 0.0002 max mem: 52124 +Epoch: [23] [2340/3634] eta: 0:10:03 lr: 0.000004 loss: 0.5415 (0.5184) time: 0.3819 data: 0.0001 max mem: 52124 +Epoch: [23] [2350/3634] eta: 0:09:59 lr: 0.000004 loss: 0.5011 (0.5183) time: 0.4858 data: 0.0001 max mem: 52124 +Epoch: [23] [2360/3634] eta: 0:09:54 lr: 0.000004 loss: 0.5079 (0.5183) time: 0.5334 data: 0.0001 max mem: 52124 +Epoch: [23] [2370/3634] eta: 0:09:49 lr: 0.000004 loss: 0.5079 (0.5182) time: 0.4296 data: 0.0001 max mem: 52124 +Epoch: [23] [2380/3634] eta: 0:09:44 lr: 0.000004 loss: 0.5206 (0.5183) time: 0.3828 data: 0.0002 max mem: 52124 +Epoch: [23] [2390/3634] eta: 0:09:39 lr: 0.000004 loss: 0.5052 (0.5182) time: 0.4258 data: 0.0001 max mem: 52124 +Epoch: [23] [2400/3634] eta: 0:09:35 lr: 0.000004 loss: 0.4806 (0.5181) time: 0.4745 data: 0.0001 max mem: 52124 +Epoch: [23] [2410/3634] eta: 0:09:31 lr: 0.000004 loss: 0.5285 (0.5182) time: 0.5367 data: 0.0001 max mem: 52124 +Epoch: [23] [2420/3634] eta: 0:09:26 lr: 0.000004 loss: 0.5250 (0.5181) time: 0.5372 data: 0.0001 max mem: 52124 +Epoch: [23] [2430/3634] eta: 0:09:21 lr: 0.000004 loss: 0.5099 (0.5181) time: 0.4321 data: 0.0002 max mem: 52124 +Epoch: [23] [2440/3634] eta: 0:09:16 lr: 0.000004 loss: 0.5546 (0.5182) time: 0.4280 data: 0.0002 max mem: 52124 +Epoch: [23] [2450/3634] eta: 0:09:12 lr: 0.000004 loss: 0.5514 (0.5183) time: 0.5346 data: 0.0002 max mem: 52124 +Epoch: [23] [2460/3634] eta: 0:09:07 lr: 0.000004 loss: 0.5385 (0.5183) time: 0.4881 data: 0.0001 max mem: 52124 +Epoch: [23] [2470/3634] eta: 0:09:03 lr: 0.000004 loss: 0.5231 (0.5182) time: 0.4285 data: 0.0001 max mem: 52124 +Epoch: [23] [2480/3634] eta: 0:08:57 lr: 0.000004 loss: 0.4782 (0.5181) time: 0.4292 data: 0.0002 max mem: 52124 +Epoch: [23] [2490/3634] eta: 0:08:52 lr: 0.000004 loss: 0.4845 (0.5182) time: 0.3820 data: 0.0001 max mem: 52124 +Epoch: [23] [2500/3634] eta: 0:08:49 lr: 0.000004 loss: 0.5174 (0.5182) time: 0.5450 data: 0.0001 max mem: 52124 +Epoch: [23] [2510/3634] eta: 0:08:44 lr: 0.000004 loss: 0.4976 (0.5181) time: 0.5456 data: 0.0001 max mem: 52124 +Epoch: [23] [2520/3634] eta: 0:08:39 lr: 0.000004 loss: 0.4900 (0.5181) time: 0.4293 data: 0.0001 max mem: 52124 +Epoch: [23] [2530/3634] eta: 0:08:35 lr: 0.000004 loss: 0.5331 (0.5181) time: 0.4870 data: 0.0001 max mem: 52124 +Epoch: [23] [2540/3634] eta: 0:08:31 lr: 0.000004 loss: 0.5331 (0.5182) time: 0.6475 data: 0.0001 max mem: 52124 +Epoch: [23] [2550/3634] eta: 0:08:28 lr: 0.000004 loss: 0.5436 (0.5182) time: 0.7214 data: 0.0001 max mem: 52124 +Epoch: [23] [2560/3634] eta: 0:08:22 lr: 0.000004 loss: 0.5487 (0.5184) time: 0.5142 data: 0.0002 max mem: 52124 +Epoch: [23] [2570/3634] eta: 0:08:18 lr: 0.000004 loss: 0.5524 (0.5184) time: 0.4427 data: 0.0002 max mem: 52124 +Epoch: [23] [2580/3634] eta: 0:08:14 lr: 0.000004 loss: 0.5194 (0.5184) time: 0.5765 data: 0.0002 max mem: 52124 +Epoch: [23] [2590/3634] eta: 0:08:09 lr: 0.000004 loss: 0.5001 (0.5184) time: 0.5594 data: 0.0002 max mem: 52124 +Epoch: [23] [2600/3634] eta: 0:08:05 lr: 0.000004 loss: 0.5060 (0.5184) time: 0.4872 data: 0.0002 max mem: 52124 +Epoch: [23] [2610/3634] eta: 0:08:00 lr: 0.000004 loss: 0.5043 (0.5182) time: 0.4435 data: 0.0002 max mem: 52124 +Epoch: [23] [2620/3634] eta: 0:07:55 lr: 0.000004 loss: 0.5118 (0.5183) time: 0.4418 data: 0.0002 max mem: 52124 +Epoch: [23] [2630/3634] eta: 0:07:50 lr: 0.000004 loss: 0.5221 (0.5182) time: 0.4896 data: 0.0001 max mem: 52124 +Epoch: [23] [2640/3634] eta: 0:07:46 lr: 0.000004 loss: 0.5330 (0.5184) time: 0.4884 data: 0.0001 max mem: 52124 +Epoch: [23] [2650/3634] eta: 0:07:42 lr: 0.000004 loss: 0.5578 (0.5184) time: 0.5904 data: 0.0001 max mem: 52124 +Epoch: [23] [2660/3634] eta: 0:07:38 lr: 0.000004 loss: 0.5500 (0.5185) time: 0.6336 data: 0.0001 max mem: 52124 +Epoch: [23] [2670/3634] eta: 0:07:33 lr: 0.000004 loss: 0.5229 (0.5185) time: 0.5750 data: 0.0001 max mem: 52124 +Epoch: [23] [2680/3634] eta: 0:07:29 lr: 0.000004 loss: 0.5524 (0.5186) time: 0.5202 data: 0.0001 max mem: 52124 +Epoch: [23] [2690/3634] eta: 0:07:24 lr: 0.000004 loss: 0.5683 (0.5187) time: 0.4291 data: 0.0001 max mem: 52124 +Epoch: [23] [2700/3634] eta: 0:07:19 lr: 0.000004 loss: 0.5327 (0.5187) time: 0.3820 data: 0.0002 max mem: 52124 +Epoch: [23] [2710/3634] eta: 0:07:14 lr: 0.000004 loss: 0.5291 (0.5188) time: 0.3816 data: 0.0001 max mem: 52124 +Epoch: [23] [2720/3634] eta: 0:07:09 lr: 0.000004 loss: 0.5420 (0.5190) time: 0.4729 data: 0.0002 max mem: 52124 +Epoch: [23] [2730/3634] eta: 0:07:05 lr: 0.000004 loss: 0.5410 (0.5191) time: 0.5175 data: 0.0002 max mem: 52124 +Epoch: [23] [2740/3634] eta: 0:07:00 lr: 0.000004 loss: 0.5215 (0.5189) time: 0.5334 data: 0.0001 max mem: 52124 +Epoch: [23] [2750/3634] eta: 0:06:56 lr: 0.000004 loss: 0.4788 (0.5188) time: 0.5922 data: 0.0001 max mem: 52124 +Epoch: [23] [2760/3634] eta: 0:06:51 lr: 0.000004 loss: 0.5266 (0.5187) time: 0.5299 data: 0.0001 max mem: 52124 +Epoch: [23] [2770/3634] eta: 0:06:46 lr: 0.000004 loss: 0.5407 (0.5189) time: 0.4266 data: 0.0001 max mem: 52124 +Epoch: [23] [2780/3634] eta: 0:06:42 lr: 0.000004 loss: 0.5341 (0.5188) time: 0.4424 data: 0.0001 max mem: 52124 +Epoch: [23] [2790/3634] eta: 0:06:37 lr: 0.000004 loss: 0.4885 (0.5187) time: 0.5016 data: 0.0001 max mem: 52124 +Epoch: [23] [2800/3634] eta: 0:06:32 lr: 0.000004 loss: 0.4919 (0.5187) time: 0.4413 data: 0.0001 max mem: 52124 +Epoch: [23] [2810/3634] eta: 0:06:27 lr: 0.000004 loss: 0.5163 (0.5186) time: 0.4460 data: 0.0001 max mem: 52124 +Epoch: [23] [2820/3634] eta: 0:06:22 lr: 0.000004 loss: 0.5163 (0.5186) time: 0.4460 data: 0.0001 max mem: 52124 +Epoch: [23] [2830/3634] eta: 0:06:18 lr: 0.000004 loss: 0.5017 (0.5186) time: 0.3825 data: 0.0001 max mem: 52124 +Epoch: [23] [2840/3634] eta: 0:06:13 lr: 0.000004 loss: 0.5133 (0.5187) time: 0.3825 data: 0.0002 max mem: 52124 +Epoch: [23] [2850/3634] eta: 0:06:08 lr: 0.000004 loss: 0.5024 (0.5185) time: 0.4294 data: 0.0002 max mem: 52124 +Epoch: [23] [2860/3634] eta: 0:06:03 lr: 0.000004 loss: 0.4577 (0.5182) time: 0.4741 data: 0.0002 max mem: 52124 +Epoch: [23] [2870/3634] eta: 0:05:58 lr: 0.000004 loss: 0.4536 (0.5182) time: 0.4265 data: 0.0001 max mem: 52124 +Epoch: [23] [2880/3634] eta: 0:05:54 lr: 0.000004 loss: 0.5168 (0.5181) time: 0.4293 data: 0.0001 max mem: 52124 +Epoch: [23] [2890/3634] eta: 0:05:49 lr: 0.000004 loss: 0.5101 (0.5181) time: 0.4723 data: 0.0001 max mem: 52124 +Epoch: [23] [2900/3634] eta: 0:05:44 lr: 0.000004 loss: 0.5199 (0.5182) time: 0.4691 data: 0.0001 max mem: 52124 +Epoch: [23] [2910/3634] eta: 0:05:40 lr: 0.000004 loss: 0.5242 (0.5181) time: 0.5187 data: 0.0001 max mem: 52124 +Epoch: [23] [2920/3634] eta: 0:05:35 lr: 0.000004 loss: 0.5242 (0.5181) time: 0.5783 data: 0.0001 max mem: 52124 +Epoch: [23] [2930/3634] eta: 0:05:31 lr: 0.000004 loss: 0.5550 (0.5182) time: 0.6040 data: 0.0001 max mem: 52124 +Epoch: [23] [2940/3634] eta: 0:05:26 lr: 0.000004 loss: 0.5503 (0.5183) time: 0.4999 data: 0.0001 max mem: 52124 +Epoch: [23] [2950/3634] eta: 0:05:21 lr: 0.000004 loss: 0.5353 (0.5182) time: 0.3823 data: 0.0001 max mem: 52124 +Epoch: [23] [2960/3634] eta: 0:05:16 lr: 0.000004 loss: 0.5020 (0.5182) time: 0.4267 data: 0.0001 max mem: 52124 +Epoch: [23] [2970/3634] eta: 0:05:12 lr: 0.000004 loss: 0.5211 (0.5182) time: 0.5319 data: 0.0001 max mem: 52124 +Epoch: [23] [2980/3634] eta: 0:05:07 lr: 0.000004 loss: 0.5211 (0.5182) time: 0.5349 data: 0.0001 max mem: 52124 +Epoch: [23] [2990/3634] eta: 0:05:03 lr: 0.000004 loss: 0.5191 (0.5181) time: 0.4742 data: 0.0001 max mem: 52124 +Epoch: [23] [3000/3634] eta: 0:04:58 lr: 0.000004 loss: 0.5221 (0.5181) time: 0.4272 data: 0.0001 max mem: 52124 +Epoch: [23] [3010/3634] eta: 0:04:53 lr: 0.000004 loss: 0.5237 (0.5182) time: 0.4269 data: 0.0001 max mem: 52124 +Epoch: [23] [3020/3634] eta: 0:04:48 lr: 0.000004 loss: 0.5179 (0.5182) time: 0.4891 data: 0.0001 max mem: 52124 +Epoch: [23] [3030/3634] eta: 0:04:44 lr: 0.000004 loss: 0.4962 (0.5181) time: 0.4446 data: 0.0001 max mem: 52124 +Epoch: [23] [3040/3634] eta: 0:04:39 lr: 0.000004 loss: 0.5173 (0.5182) time: 0.4267 data: 0.0001 max mem: 52124 +Epoch: [23] [3050/3634] eta: 0:04:34 lr: 0.000004 loss: 0.5336 (0.5183) time: 0.4718 data: 0.0001 max mem: 52124 +Epoch: [23] [3060/3634] eta: 0:04:29 lr: 0.000004 loss: 0.5249 (0.5182) time: 0.4715 data: 0.0001 max mem: 52124 +Epoch: [23] [3070/3634] eta: 0:04:25 lr: 0.000004 loss: 0.4921 (0.5182) time: 0.6365 data: 0.0001 max mem: 52124 +Epoch: [23] [3080/3634] eta: 0:04:21 lr: 0.000004 loss: 0.4921 (0.5181) time: 0.6531 data: 0.0001 max mem: 52124 +Epoch: [23] [3090/3634] eta: 0:04:16 lr: 0.000004 loss: 0.4902 (0.5181) time: 0.6050 data: 0.0002 max mem: 52124 +Epoch: [23] [3100/3634] eta: 0:04:12 lr: 0.000004 loss: 0.5045 (0.5181) time: 0.6041 data: 0.0002 max mem: 52124 +Epoch: [23] [3110/3634] eta: 0:04:07 lr: 0.000004 loss: 0.5014 (0.5180) time: 0.4925 data: 0.0002 max mem: 52124 +Epoch: [23] [3120/3634] eta: 0:04:03 lr: 0.000004 loss: 0.5282 (0.5181) time: 0.5828 data: 0.0001 max mem: 52124 +Epoch: [23] [3130/3634] eta: 0:03:58 lr: 0.000004 loss: 0.5532 (0.5181) time: 0.6529 data: 0.0001 max mem: 52124 +Epoch: [23] [3140/3634] eta: 0:03:53 lr: 0.000004 loss: 0.5262 (0.5180) time: 0.5030 data: 0.0001 max mem: 52124 +Epoch: [23] [3150/3634] eta: 0:03:49 lr: 0.000004 loss: 0.5073 (0.5180) time: 0.4297 data: 0.0001 max mem: 52124 +Epoch: [23] [3160/3634] eta: 0:03:44 lr: 0.000004 loss: 0.5093 (0.5180) time: 0.4733 data: 0.0001 max mem: 52124 +Epoch: [23] [3170/3634] eta: 0:03:39 lr: 0.000004 loss: 0.5290 (0.5180) time: 0.5278 data: 0.0002 max mem: 52124 +Epoch: [23] [3180/3634] eta: 0:03:34 lr: 0.000004 loss: 0.5290 (0.5180) time: 0.4844 data: 0.0002 max mem: 52124 +Epoch: [23] [3190/3634] eta: 0:03:30 lr: 0.000004 loss: 0.5088 (0.5180) time: 0.4271 data: 0.0002 max mem: 52124 +Epoch: [23] [3200/3634] eta: 0:03:25 lr: 0.000004 loss: 0.5240 (0.5180) time: 0.5186 data: 0.0001 max mem: 52124 +Epoch: [23] [3210/3634] eta: 0:03:20 lr: 0.000004 loss: 0.5254 (0.5180) time: 0.5185 data: 0.0001 max mem: 52124 +Epoch: [23] [3220/3634] eta: 0:03:16 lr: 0.000004 loss: 0.5328 (0.5181) time: 0.4872 data: 0.0001 max mem: 52124 +Epoch: [23] [3230/3634] eta: 0:03:11 lr: 0.000004 loss: 0.5328 (0.5181) time: 0.4942 data: 0.0001 max mem: 52124 +Epoch: [23] [3240/3634] eta: 0:03:06 lr: 0.000004 loss: 0.5398 (0.5183) time: 0.4794 data: 0.0001 max mem: 52124 +Epoch: [23] [3250/3634] eta: 0:03:01 lr: 0.000004 loss: 0.5399 (0.5184) time: 0.4715 data: 0.0002 max mem: 52124 +Epoch: [23] [3260/3634] eta: 0:02:57 lr: 0.000004 loss: 0.4913 (0.5183) time: 0.4704 data: 0.0002 max mem: 52124 +Epoch: [23] [3270/3634] eta: 0:02:52 lr: 0.000004 loss: 0.4947 (0.5183) time: 0.4707 data: 0.0002 max mem: 52124 +Epoch: [23] [3280/3634] eta: 0:02:47 lr: 0.000004 loss: 0.4947 (0.5182) time: 0.5479 data: 0.0001 max mem: 52124 +Epoch: [23] [3290/3634] eta: 0:02:43 lr: 0.000004 loss: 0.5059 (0.5182) time: 0.5657 data: 0.0001 max mem: 52124 +Epoch: [23] [3300/3634] eta: 0:02:38 lr: 0.000004 loss: 0.5072 (0.5182) time: 0.4882 data: 0.0002 max mem: 52124 +Epoch: [23] [3310/3634] eta: 0:02:33 lr: 0.000004 loss: 0.5124 (0.5182) time: 0.4266 data: 0.0002 max mem: 52124 +Epoch: [23] [3320/3634] eta: 0:02:28 lr: 0.000004 loss: 0.5179 (0.5183) time: 0.4292 data: 0.0002 max mem: 52124 +Epoch: [23] [3330/3634] eta: 0:02:24 lr: 0.000004 loss: 0.4904 (0.5181) time: 0.4905 data: 0.0002 max mem: 52124 +Epoch: [23] [3340/3634] eta: 0:02:19 lr: 0.000004 loss: 0.4904 (0.5181) time: 0.4439 data: 0.0001 max mem: 52124 +Epoch: [23] [3350/3634] eta: 0:02:14 lr: 0.000004 loss: 0.5041 (0.5180) time: 0.3831 data: 0.0001 max mem: 52124 +Epoch: [23] [3360/3634] eta: 0:02:09 lr: 0.000004 loss: 0.5024 (0.5180) time: 0.4456 data: 0.0001 max mem: 52124 +Epoch: [23] [3370/3634] eta: 0:02:05 lr: 0.000004 loss: 0.5082 (0.5180) time: 0.4882 data: 0.0001 max mem: 52124 +Epoch: [23] [3380/3634] eta: 0:02:00 lr: 0.000004 loss: 0.5345 (0.5180) time: 0.4258 data: 0.0001 max mem: 52124 +Epoch: [23] [3390/3634] eta: 0:01:55 lr: 0.000004 loss: 0.5431 (0.5181) time: 0.4860 data: 0.0001 max mem: 52124 +Epoch: [23] [3400/3634] eta: 0:01:50 lr: 0.000004 loss: 0.5306 (0.5180) time: 0.5931 data: 0.0001 max mem: 52124 +Epoch: [23] [3410/3634] eta: 0:01:46 lr: 0.000004 loss: 0.4881 (0.5180) time: 0.4904 data: 0.0001 max mem: 52124 +Epoch: [23] [3420/3634] eta: 0:01:41 lr: 0.000004 loss: 0.5295 (0.5181) time: 0.3832 data: 0.0001 max mem: 52124 +Epoch: [23] [3430/3634] eta: 0:01:36 lr: 0.000004 loss: 0.5289 (0.5181) time: 0.3827 data: 0.0001 max mem: 52124 +Epoch: [23] [3440/3634] eta: 0:01:31 lr: 0.000004 loss: 0.4893 (0.5180) time: 0.4272 data: 0.0002 max mem: 52124 +Epoch: [23] [3450/3634] eta: 0:01:27 lr: 0.000004 loss: 0.4881 (0.5180) time: 0.4270 data: 0.0002 max mem: 52124 +Epoch: [23] [3460/3634] eta: 0:01:22 lr: 0.000004 loss: 0.5458 (0.5181) time: 0.3822 data: 0.0002 max mem: 52124 +Epoch: [23] [3470/3634] eta: 0:01:17 lr: 0.000004 loss: 0.5470 (0.5180) time: 0.5330 data: 0.0002 max mem: 52124 +Epoch: [23] [3480/3634] eta: 0:01:12 lr: 0.000004 loss: 0.5180 (0.5180) time: 0.5795 data: 0.0001 max mem: 52124 +Epoch: [23] [3490/3634] eta: 0:01:08 lr: 0.000004 loss: 0.5070 (0.5180) time: 0.4293 data: 0.0002 max mem: 52124 +Epoch: [23] [3500/3634] eta: 0:01:03 lr: 0.000004 loss: 0.5164 (0.5180) time: 0.4278 data: 0.0002 max mem: 52124 +Epoch: [23] [3510/3634] eta: 0:00:58 lr: 0.000004 loss: 0.5303 (0.5180) time: 0.4738 data: 0.0001 max mem: 52124 +Epoch: [23] [3520/3634] eta: 0:00:53 lr: 0.000004 loss: 0.4862 (0.5179) time: 0.4745 data: 0.0002 max mem: 52124 +Epoch: [23] [3530/3634] eta: 0:00:49 lr: 0.000004 loss: 0.4931 (0.5180) time: 0.4759 data: 0.0002 max mem: 52124 +Epoch: [23] [3540/3634] eta: 0:00:44 lr: 0.000004 loss: 0.5182 (0.5180) time: 0.4309 data: 0.0001 max mem: 52124 +Epoch: [23] [3550/3634] eta: 0:00:39 lr: 0.000004 loss: 0.5015 (0.5178) time: 0.4285 data: 0.0001 max mem: 52124 +Epoch: [23] [3560/3634] eta: 0:00:35 lr: 0.000004 loss: 0.5095 (0.5180) time: 0.5493 data: 0.0001 max mem: 52124 +Epoch: [23] [3570/3634] eta: 0:00:30 lr: 0.000004 loss: 0.5624 (0.5180) time: 0.7184 data: 0.0001 max mem: 52124 +Epoch: [23] [3580/3634] eta: 0:00:25 lr: 0.000004 loss: 0.5422 (0.5181) time: 0.5973 data: 0.0002 max mem: 52124 +Epoch: [23] [3590/3634] eta: 0:00:20 lr: 0.000004 loss: 0.5408 (0.5182) time: 0.5953 data: 0.0002 max mem: 52124 +Epoch: [23] [3600/3634] eta: 0:00:16 lr: 0.000004 loss: 0.5177 (0.5182) time: 0.6959 data: 0.0001 max mem: 52124 +Epoch: [23] [3610/3634] eta: 0:00:11 lr: 0.000004 loss: 0.5272 (0.5182) time: 0.5265 data: 0.0002 max mem: 52124 +Epoch: [23] [3620/3634] eta: 0:00:06 lr: 0.000004 loss: 0.5549 (0.5183) time: 0.4843 data: 0.0026 max mem: 52124 +Epoch: [23] [3630/3634] eta: 0:00:01 lr: 0.000004 loss: 0.5362 (0.5183) time: 0.4979 data: 0.0026 max mem: 52124 +Epoch: [23] [3633/3634] eta: 0:00:00 lr: 0.000004 loss: 0.5362 (0.5183) time: 0.4972 data: 0.0025 max mem: 52124 +Epoch: [23] Total time: 0:28:48 (0.4757 s / it) +Averaged stats: lr: 0.000004 loss: 0.5362 (0.5184) +Test: [ 0/1403] eta: 1:11:58 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0921 (0.0921) acc1: 100.0000 (100.0000) acc5: 100.0000 (100.0000) time: 3.0778 data: 2.8606 max mem: 52124 +Test: [ 10/1403] eta: 0:08:31 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1772 (0.1849) acc1: 95.3125 (93.6080) acc5: 100.0000 (100.0000) time: 0.3671 data: 0.2603 max mem: 52124 +Test: [ 20/1403] eta: 0:05:29 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1367 (0.1656) acc1: 95.3125 (94.4196) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52124 +Test: [ 30/1403] eta: 0:04:23 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0852 (0.1563) acc1: 100.0000 (95.1613) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52124 +Test: [ 40/1403] eta: 0:03:50 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0986 (0.1556) acc1: 100.0000 (95.5793) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52124 +Test: [ 50/1403] eta: 0:03:29 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1095 (0.1679) acc1: 98.4375 (94.9755) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52124 +Test: [ 60/1403] eta: 0:03:14 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.1132 (0.1647) acc1: 100.0000 (95.4406) acc5: 100.0000 (100.0000) time: 0.0959 data: 0.0003 max mem: 52124 +Test: [ 70/1403] eta: 0:03:03 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.1304 (0.1711) acc1: 98.4375 (95.0484) acc5: 100.0000 (100.0000) time: 0.0959 data: 0.0003 max mem: 52124 +Test: [ 80/1403] eta: 0:02:55 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1546 (0.1807) acc1: 95.3125 (94.5795) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52124 +Test: [ 90/1403] eta: 0:02:49 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1961 (0.2004) acc1: 93.7500 (93.5955) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52124 +Test: [ 100/1403] eta: 0:02:43 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1961 (0.1998) acc1: 95.3125 (93.6881) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52124 +Test: [ 110/1403] eta: 0:02:38 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1565 (0.1973) acc1: 96.8750 (93.8485) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52124 +Test: [ 120/1403] eta: 0:02:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0817 (0.1877) acc1: 100.0000 (94.3311) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52124 +Test: [ 130/1403] eta: 0:02:31 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0818 (0.1811) acc1: 100.0000 (94.6565) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52124 +Test: [ 140/1403] eta: 0:02:27 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0780 (0.1736) acc1: 100.0000 (95.0355) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52124 +Test: [ 150/1403] eta: 0:02:25 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0749 (0.1675) acc1: 100.0000 (95.3332) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52124 +Test: [ 160/1403] eta: 0:02:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0792 (0.1625) acc1: 100.0000 (95.5648) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52124 +Test: [ 170/1403] eta: 0:02:19 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0792 (0.1582) acc1: 100.0000 (95.7968) acc5: 100.0000 (100.0000) time: 0.0962 data: 0.0003 max mem: 52124 +Test: [ 180/1403] eta: 0:02:17 acc1_pos: 93.750 acc5_pos: 100.000 loss: 0.0836 (0.1604) acc1: 100.0000 (95.7700) acc5: 100.0000 (100.0000) time: 0.0962 data: 0.0003 max mem: 52124 +Test: [ 190/1403] eta: 0:02:15 acc1_pos: 90.625 acc5_pos: 100.000 loss: 0.2347 (0.1659) acc1: 93.7500 (95.5007) acc5: 100.0000 (100.0000) time: 0.0962 data: 0.0003 max mem: 52124 +Test: [ 200/1403] eta: 0:02:13 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.2347 (0.1739) acc1: 90.6250 (95.1493) acc5: 100.0000 (100.0000) time: 0.0963 data: 0.0003 max mem: 52124 +Test: [ 210/1403] eta: 0:02:11 acc1_pos: 76.562 acc5_pos: 100.000 loss: 0.2557 (0.1819) acc1: 87.5000 (94.7127) acc5: 100.0000 (100.0000) time: 0.0962 data: 0.0003 max mem: 52124 +Test: [ 220/1403] eta: 0:02:09 acc1_pos: 92.188 acc5_pos: 100.000 loss: 0.3027 (0.1882) acc1: 92.1875 (94.4499) acc5: 100.0000 (100.0000) time: 0.0962 data: 0.0003 max mem: 52124 +Test: [ 230/1403] eta: 0:02:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2147 (0.1845) acc1: 93.7500 (94.6293) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52124 +Test: [ 240/1403] eta: 0:02:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0812 (0.1804) acc1: 100.0000 (94.8392) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52124 +Test: [ 250/1403] eta: 0:02:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0800 (0.1768) acc1: 100.0000 (95.0261) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52124 +Test: [ 260/1403] eta: 0:02:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0811 (0.1733) acc1: 100.0000 (95.1988) acc5: 100.0000 (100.0000) time: 0.0963 data: 0.0003 max mem: 52124 +Test: [ 270/1403] eta: 0:02:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0826 (0.1702) acc1: 100.0000 (95.3529) acc5: 100.0000 (100.0000) time: 0.0963 data: 0.0003 max mem: 52124 +Test: [ 280/1403] eta: 0:01:59 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0819 (0.1670) acc1: 100.0000 (95.5127) acc5: 100.0000 (100.0000) time: 0.0962 data: 0.0003 max mem: 52124 +Test: [ 290/1403] eta: 0:01:58 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0783 (0.1640) acc1: 100.0000 (95.6615) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52124 +Test: [ 300/1403] eta: 0:01:56 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0767 (0.1613) acc1: 100.0000 (95.8005) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52124 +Test: [ 310/1403] eta: 0:01:55 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0814 (0.1589) acc1: 100.0000 (95.9305) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52124 +Test: [ 320/1403] eta: 0:01:54 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1037 (0.1573) acc1: 100.0000 (96.0572) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52124 +Test: [ 330/1403] eta: 0:01:52 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1113 (0.1560) acc1: 100.0000 (96.1764) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52124 +Test: [ 340/1403] eta: 0:01:51 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1113 (0.1547) acc1: 100.0000 (96.2885) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52124 +Test: [ 350/1403] eta: 0:01:50 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1084 (0.1534) acc1: 100.0000 (96.3942) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52124 +Test: [ 360/1403] eta: 0:01:48 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1119 (0.1524) acc1: 100.0000 (96.4855) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52124 +Test: [ 370/1403] eta: 0:01:47 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1111 (0.1512) acc1: 100.0000 (96.5802) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52124 +Test: [ 380/1403] eta: 0:01:46 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1090 (0.1501) acc1: 100.0000 (96.6699) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0004 max mem: 52124 +Test: [ 390/1403] eta: 0:01:45 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1097 (0.1509) acc1: 100.0000 (96.6272) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52124 +Test: [ 400/1403] eta: 0:01:43 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2543 (0.1544) acc1: 90.6250 (96.4269) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52124 +Test: [ 410/1403] eta: 0:01:42 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2773 (0.1573) acc1: 89.0625 (96.2515) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52124 +Test: [ 420/1403] eta: 0:01:41 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2409 (0.1587) acc1: 90.6250 (96.1550) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52124 +Test: [ 430/1403] eta: 0:01:40 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2287 (0.1614) acc1: 90.6250 (96.0013) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0961 data: 0.0003 max mem: 52124 +Test: [ 440/1403] eta: 0:01:39 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2758 (0.1648) acc1: 89.0625 (95.8262) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52124 +Test: [ 450/1403] eta: 0:01:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2638 (0.1671) acc1: 89.0625 (95.7109) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52124 +Test: [ 460/1403] eta: 0:01:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2638 (0.1694) acc1: 89.0625 (95.5531) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52124 +Test: [ 470/1403] eta: 0:01:35 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2213 (0.1686) acc1: 90.6250 (95.5812) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52124 +Test: [ 480/1403] eta: 0:01:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0626 (0.1664) acc1: 100.0000 (95.6731) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52124 +Test: [ 490/1403] eta: 0:01:33 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0630 (0.1643) acc1: 100.0000 (95.7612) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52124 +Test: [ 500/1403] eta: 0:01:32 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0634 (0.1623) acc1: 100.0000 (95.8458) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52124 +Test: [ 510/1403] eta: 0:01:31 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0636 (0.1604) acc1: 100.0000 (95.9271) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52124 +Test: [ 520/1403] eta: 0:01:29 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0677 (0.1587) acc1: 100.0000 (96.0053) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52124 +Test: [ 530/1403] eta: 0:01:28 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0593 (0.1568) acc1: 100.0000 (96.0805) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52124 +Test: [ 540/1403] eta: 0:01:27 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0589 (0.1550) acc1: 100.0000 (96.1530) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52124 +Test: [ 550/1403] eta: 0:01:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0599 (0.1538) acc1: 100.0000 (96.2001) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52124 +Test: [ 560/1403] eta: 0:01:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0858 (0.1528) acc1: 98.4375 (96.2372) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52124 +Test: [ 570/1403] eta: 0:01:24 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0887 (0.1517) acc1: 98.4375 (96.2839) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52124 +Test: [ 580/1403] eta: 0:01:23 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0958 (0.1512) acc1: 98.4375 (96.3022) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52124 +Test: [ 590/1403] eta: 0:01:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1182 (0.1511) acc1: 96.8750 (96.2881) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52124 +Test: [ 600/1403] eta: 0:01:21 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1286 (0.1514) acc1: 95.3125 (96.2354) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52124 +Test: [ 610/1403] eta: 0:01:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1259 (0.1508) acc1: 95.3125 (96.2408) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52124 +Test: [ 620/1403] eta: 0:01:19 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1338 (0.1516) acc1: 95.3125 (96.1629) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52124 +Test: [ 630/1403] eta: 0:01:18 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1594 (0.1516) acc1: 93.7500 (96.1544) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52124 +Test: [ 640/1403] eta: 0:01:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1352 (0.1513) acc1: 95.3125 (96.1583) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52124 +Test: [ 650/1403] eta: 0:01:15 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1274 (0.1510) acc1: 96.8750 (96.1742) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52124 +Test: [ 660/1403] eta: 0:01:14 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0996 (0.1500) acc1: 98.4375 (96.2179) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52124 +Test: [ 670/1403] eta: 0:01:13 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0734 (0.1489) acc1: 100.0000 (96.2672) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52124 +Test: [ 680/1403] eta: 0:01:12 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0903 (0.1505) acc1: 98.4375 (96.1546) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52124 +Test: [ 690/1403] eta: 0:01:11 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3009 (0.1552) acc1: 87.5000 (95.8959) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52124 +Test: [ 700/1403] eta: 0:01:10 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1823 (0.1554) acc1: 89.0625 (95.8742) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52124 +Test: [ 710/1403] eta: 0:01:09 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3038 (0.1598) acc1: 85.9375 (95.6421) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52124 +Test: [ 720/1403] eta: 0:01:08 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3038 (0.1598) acc1: 84.3750 (95.6462) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0969 data: 0.0003 max mem: 52124 +Test: [ 730/1403] eta: 0:01:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0676 (0.1587) acc1: 100.0000 (95.6908) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52124 +Test: [ 740/1403] eta: 0:01:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0770 (0.1578) acc1: 98.4375 (95.7279) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52124 +Test: [ 750/1403] eta: 0:01:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1024 (0.1594) acc1: 96.8750 (95.6350) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52124 +Test: [ 760/1403] eta: 0:01:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3279 (0.1633) acc1: 81.2500 (95.3638) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52124 +Test: [ 770/1403] eta: 0:01:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4828 (0.1678) acc1: 73.4375 (95.0875) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52124 +Test: [ 780/1403] eta: 0:01:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4024 (0.1701) acc1: 81.2500 (94.9444) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52124 +Test: [ 790/1403] eta: 0:01:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3185 (0.1725) acc1: 84.3750 (94.8266) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52124 +Test: [ 800/1403] eta: 0:01:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2108 (0.1729) acc1: 90.6250 (94.8053) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52124 +Test: [ 810/1403] eta: 0:00:59 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2349 (0.1762) acc1: 89.0625 (94.6478) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52124 +Test: [ 820/1403] eta: 0:00:58 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6450 (0.1854) acc1: 70.3125 (94.2543) acc5: 100.0000 (100.0000) acc1_neg: 43.750 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52124 +Test: [ 830/1403] eta: 0:00:57 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6984 (0.1880) acc1: 70.3125 (94.1204) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52124 +Test: [ 840/1403] eta: 0:00:56 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5008 (0.1962) acc1: 71.8750 (93.7667) acc5: 100.0000 (100.0000) acc1_neg: 46.875 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52124 +Test: [ 850/1403] eta: 0:00:55 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.0707 (0.2074) acc1: 54.6875 (93.2965) acc5: 100.0000 (100.0000) acc1_neg: 54.688 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52124 +Test: [ 860/1403] eta: 0:00:54 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.0259 (0.2110) acc1: 56.2500 (93.1348) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52124 +Test: [ 870/1403] eta: 0:00:53 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1452 (0.2100) acc1: 95.3125 (93.1706) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52124 +Test: [ 880/1403] eta: 0:00:52 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0988 (0.2089) acc1: 98.4375 (93.2215) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52124 +Test: [ 890/1403] eta: 0:00:51 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0937 (0.2075) acc1: 98.4375 (93.2870) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52124 +Test: [ 900/1403] eta: 0:00:50 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0752 (0.2060) acc1: 100.0000 (93.3598) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52124 +Test: [ 910/1403] eta: 0:00:49 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0773 (0.2052) acc1: 100.0000 (93.3915) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52124 +Test: [ 920/1403] eta: 0:00:48 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1064 (0.2043) acc1: 96.8750 (93.4226) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52124 +Test: [ 930/1403] eta: 0:00:47 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1031 (0.2031) acc1: 98.4375 (93.4764) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52124 +Test: [ 940/1403] eta: 0:00:46 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1131 (0.2034) acc1: 96.8750 (93.4445) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52124 +Test: [ 950/1403] eta: 0:00:45 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2618 (0.2045) acc1: 87.5000 (93.3721) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52124 +Test: [ 960/1403] eta: 0:00:44 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2854 (0.2053) acc1: 87.5000 (93.3159) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52124 +Test: [ 970/1403] eta: 0:00:43 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2580 (0.2064) acc1: 89.0625 (93.2576) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52124 +Test: [ 980/1403] eta: 0:00:42 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2699 (0.2073) acc1: 89.0625 (93.1925) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52124 +Test: [ 990/1403] eta: 0:00:40 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2595 (0.2079) acc1: 89.0625 (93.1493) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52124 +Test: [1000/1403] eta: 0:00:39 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2595 (0.2088) acc1: 89.0625 (93.0944) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52124 +Test: [1010/1403] eta: 0:00:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2985 (0.2098) acc1: 87.5000 (93.0422) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52124 +Test: [1020/1403] eta: 0:00:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4498 (0.2163) acc1: 81.2500 (92.6956) acc5: 100.0000 (100.0000) acc1_neg: 34.375 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52124 +Test: [1030/1403] eta: 0:00:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5888 (0.2183) acc1: 62.5000 (92.5573) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52124 +Test: [1040/1403] eta: 0:00:35 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2423 (0.2182) acc1: 89.0625 (92.5627) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52124 +Test: [1050/1403] eta: 0:00:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1853 (0.2179) acc1: 93.7500 (92.5815) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52124 +Test: [1060/1403] eta: 0:00:33 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1498 (0.2172) acc1: 96.8750 (92.6205) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52124 +Test: [1070/1403] eta: 0:00:32 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1498 (0.2178) acc1: 96.8750 (92.5945) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52124 +Test: [1080/1403] eta: 0:00:31 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4134 (0.2199) acc1: 82.8125 (92.4853) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52124 +Test: [1090/1403] eta: 0:00:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4646 (0.2221) acc1: 81.2500 (92.3880) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52124 +Test: [1100/1403] eta: 0:00:29 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3543 (0.2226) acc1: 85.9375 (92.3663) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52124 +Test: [1110/1403] eta: 0:00:28 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2841 (0.2231) acc1: 89.0625 (92.3436) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52124 +Test: [1120/1403] eta: 0:00:27 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2887 (0.2237) acc1: 89.0625 (92.3116) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52124 +Test: [1130/1403] eta: 0:00:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2981 (0.2245) acc1: 89.0625 (92.2732) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52124 +Test: [1140/1403] eta: 0:00:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2520 (0.2247) acc1: 90.6250 (92.2628) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52124 +Test: [1150/1403] eta: 0:00:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2505 (0.2251) acc1: 90.6250 (92.2377) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52124 +Test: [1160/1403] eta: 0:00:24 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2819 (0.2257) acc1: 89.0625 (92.2090) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52124 +Test: [1170/1403] eta: 0:00:23 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2730 (0.2256) acc1: 89.0625 (92.2115) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52124 +Test: [1180/1403] eta: 0:00:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0711 (0.2243) acc1: 100.0000 (92.2775) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0003 max mem: 52124 +Test: [1190/1403] eta: 0:00:21 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0700 (0.2230) acc1: 100.0000 (92.3423) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0960 data: 0.0002 max mem: 52124 +Test: [1200/1403] eta: 0:00:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0705 (0.2217) acc1: 100.0000 (92.4061) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0002 max mem: 52124 +Test: [1210/1403] eta: 0:00:19 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0693 (0.2205) acc1: 100.0000 (92.4688) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0003 max mem: 52124 +Test: [1220/1403] eta: 0:00:18 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0690 (0.2192) acc1: 100.0000 (92.5305) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0003 max mem: 52124 +Test: [1230/1403] eta: 0:00:17 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0702 (0.2180) acc1: 100.0000 (92.5911) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0002 max mem: 52124 +Test: [1240/1403] eta: 0:00:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0701 (0.2168) acc1: 100.0000 (92.6508) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0002 max mem: 52124 +Test: [1250/1403] eta: 0:00:15 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0700 (0.2157) acc1: 100.0000 (92.7096) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0002 max mem: 52124 +Test: [1260/1403] eta: 0:00:14 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0750 (0.2146) acc1: 100.0000 (92.7674) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0002 max mem: 52124 +Test: [1270/1403] eta: 0:00:13 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0764 (0.2135) acc1: 100.0000 (92.8243) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0002 max mem: 52124 +Test: [1280/1403] eta: 0:00:12 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0762 (0.2124) acc1: 100.0000 (92.8803) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0002 max mem: 52124 +Test: [1290/1403] eta: 0:00:11 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0759 (0.2114) acc1: 100.0000 (92.9355) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52124 +Test: [1300/1403] eta: 0:00:10 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0774 (0.2104) acc1: 100.0000 (92.9898) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0002 max mem: 52124 +Test: [1310/1403] eta: 0:00:09 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0771 (0.2093) acc1: 100.0000 (93.0432) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0002 max mem: 52124 +Test: [1320/1403] eta: 0:00:08 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0751 (0.2083) acc1: 100.0000 (93.0959) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0003 max mem: 52124 +Test: [1330/1403] eta: 0:00:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0749 (0.2073) acc1: 100.0000 (93.1466) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0003 max mem: 52124 +Test: [1340/1403] eta: 0:00:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0677 (0.2063) acc1: 100.0000 (93.1977) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0002 max mem: 52124 +Test: [1350/1403] eta: 0:00:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0665 (0.2052) acc1: 100.0000 (93.2457) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52124 +Test: [1360/1403] eta: 0:00:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0665 (0.2042) acc1: 100.0000 (93.2954) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52124 +Test: [1370/1403] eta: 0:00:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0672 (0.2033) acc1: 100.0000 (93.3409) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52124 +Test: [1380/1403] eta: 0:00:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0660 (0.2023) acc1: 100.0000 (93.3891) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0002 max mem: 52124 +Test: [1390/1403] eta: 0:00:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0657 (0.2013) acc1: 100.0000 (93.4366) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52124 +Test: [1400/1403] eta: 0:00:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0657 (0.2004) acc1: 100.0000 (93.4812) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0001 max mem: 52124 +Test: [1402/1403] eta: 0:00:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0662 (0.2002) acc1: 100.0000 (93.4859) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0983 data: 0.0001 max mem: 52124 +Test: Total time: 0:02:18 (0.0988 s / it) +* Overall Acc@1 93.486 Acc@5 100.000 + target == real Acc@1 95.894 Acc@5 100.000 + target == fake Acc@1 92.807 Acc@5 100.000 + loss 0.200 +Accuracy of the network on the 89729 test images: 93.5% +Max accuracy: 93.49% +Epoch: [24] [ 0/3634] eta: 5:23:58 lr: 0.000004 loss: 0.5037 (0.5037) time: 5.3492 data: 4.8804 max mem: 52125 +Epoch: [24] [ 10/3634] eta: 1:19:07 lr: 0.000004 loss: 0.5257 (0.5092) time: 1.3101 data: 0.4438 max mem: 52126 +Epoch: [24] [ 20/3634] eta: 0:55:50 lr: 0.000004 loss: 0.5257 (0.5185) time: 0.7060 data: 0.0001 max mem: 52126 +Epoch: [24] [ 30/3634] eta: 0:45:07 lr: 0.000004 loss: 0.5192 (0.5184) time: 0.4438 data: 0.0001 max mem: 52126 +Epoch: [24] [ 40/3634] eta: 0:39:36 lr: 0.000004 loss: 0.5390 (0.5223) time: 0.3823 data: 0.0001 max mem: 52126 +Epoch: [24] [ 50/3634] eta: 0:38:33 lr: 0.000004 loss: 0.5393 (0.5233) time: 0.4817 data: 0.0001 max mem: 52126 +Epoch: [24] [ 60/3634] eta: 0:37:04 lr: 0.000004 loss: 0.5195 (0.5216) time: 0.5423 data: 0.0001 max mem: 52126 +Epoch: [24] [ 70/3634] eta: 0:37:17 lr: 0.000004 loss: 0.5162 (0.5184) time: 0.5823 data: 0.0001 max mem: 52126 +Epoch: [24] [ 80/3634] eta: 0:35:23 lr: 0.000004 loss: 0.5241 (0.5194) time: 0.5216 data: 0.0001 max mem: 52126 +Epoch: [24] [ 90/3634] eta: 0:35:28 lr: 0.000004 loss: 0.5597 (0.5218) time: 0.5043 data: 0.0001 max mem: 52126 +Epoch: [24] [ 100/3634] eta: 0:36:05 lr: 0.000004 loss: 0.5574 (0.5232) time: 0.6743 data: 0.0001 max mem: 52126 +Epoch: [24] [ 110/3634] eta: 0:34:46 lr: 0.000004 loss: 0.5574 (0.5252) time: 0.5528 data: 0.0002 max mem: 52126 +Epoch: [24] [ 120/3634] eta: 0:34:14 lr: 0.000004 loss: 0.5429 (0.5230) time: 0.4437 data: 0.0001 max mem: 52126 +Epoch: [24] [ 130/3634] eta: 0:33:39 lr: 0.000004 loss: 0.5258 (0.5242) time: 0.4886 data: 0.0001 max mem: 52126 +Epoch: [24] [ 140/3634] eta: 0:33:08 lr: 0.000004 loss: 0.5229 (0.5227) time: 0.4751 data: 0.0002 max mem: 52126 +Epoch: [24] [ 150/3634] eta: 0:32:49 lr: 0.000004 loss: 0.5276 (0.5229) time: 0.4928 data: 0.0001 max mem: 52126 +Epoch: [24] [ 160/3634] eta: 0:32:25 lr: 0.000004 loss: 0.5276 (0.5227) time: 0.4945 data: 0.0002 max mem: 52126 +Epoch: [24] [ 170/3634] eta: 0:31:43 lr: 0.000004 loss: 0.5335 (0.5228) time: 0.4322 data: 0.0002 max mem: 52126 +Epoch: [24] [ 180/3634] eta: 0:31:29 lr: 0.000004 loss: 0.4875 (0.5198) time: 0.4447 data: 0.0001 max mem: 52126 +Epoch: [24] [ 190/3634] eta: 0:30:54 lr: 0.000004 loss: 0.5469 (0.5220) time: 0.4437 data: 0.0002 max mem: 52126 +Epoch: [24] [ 200/3634] eta: 0:30:23 lr: 0.000004 loss: 0.5663 (0.5232) time: 0.3836 data: 0.0002 max mem: 52126 +Epoch: [24] [ 210/3634] eta: 0:30:33 lr: 0.000004 loss: 0.5144 (0.5217) time: 0.5079 data: 0.0001 max mem: 52126 +Epoch: [24] [ 220/3634] eta: 0:30:24 lr: 0.000004 loss: 0.5106 (0.5215) time: 0.5685 data: 0.0001 max mem: 52126 +Epoch: [24] [ 230/3634] eta: 0:30:14 lr: 0.000004 loss: 0.5026 (0.5200) time: 0.5049 data: 0.0001 max mem: 52126 +Epoch: [24] [ 240/3634] eta: 0:30:00 lr: 0.000004 loss: 0.5029 (0.5202) time: 0.4890 data: 0.0001 max mem: 52126 +Epoch: [24] [ 250/3634] eta: 0:29:35 lr: 0.000004 loss: 0.4930 (0.5188) time: 0.4287 data: 0.0001 max mem: 52126 +Epoch: [24] [ 260/3634] eta: 0:29:11 lr: 0.000004 loss: 0.4930 (0.5177) time: 0.3821 data: 0.0002 max mem: 52126 +Epoch: [24] [ 270/3634] eta: 0:29:05 lr: 0.000004 loss: 0.4979 (0.5166) time: 0.4453 data: 0.0002 max mem: 52126 +Epoch: [24] [ 280/3634] eta: 0:29:10 lr: 0.000004 loss: 0.5071 (0.5170) time: 0.5566 data: 0.0002 max mem: 52126 +Epoch: [24] [ 290/3634] eta: 0:29:13 lr: 0.000004 loss: 0.5193 (0.5168) time: 0.6011 data: 0.0002 max mem: 52126 +Epoch: [24] [ 300/3634] eta: 0:29:06 lr: 0.000004 loss: 0.5188 (0.5157) time: 0.5542 data: 0.0002 max mem: 52126 +Epoch: [24] [ 310/3634] eta: 0:28:46 lr: 0.000004 loss: 0.4802 (0.5150) time: 0.4465 data: 0.0002 max mem: 52126 +Epoch: [24] [ 320/3634] eta: 0:28:39 lr: 0.000004 loss: 0.4972 (0.5144) time: 0.4436 data: 0.0002 max mem: 52126 +Epoch: [24] [ 330/3634] eta: 0:28:21 lr: 0.000004 loss: 0.5333 (0.5153) time: 0.4434 data: 0.0002 max mem: 52126 +Epoch: [24] [ 340/3634] eta: 0:28:12 lr: 0.000004 loss: 0.5361 (0.5163) time: 0.4297 data: 0.0001 max mem: 52126 +Epoch: [24] [ 350/3634] eta: 0:28:06 lr: 0.000004 loss: 0.5367 (0.5159) time: 0.4906 data: 0.0001 max mem: 52126 +Epoch: [24] [ 360/3634] eta: 0:27:49 lr: 0.000004 loss: 0.5367 (0.5165) time: 0.4432 data: 0.0001 max mem: 52126 +Epoch: [24] [ 370/3634] eta: 0:27:32 lr: 0.000004 loss: 0.5442 (0.5165) time: 0.3829 data: 0.0001 max mem: 52126 +Epoch: [24] [ 380/3634] eta: 0:27:35 lr: 0.000004 loss: 0.4999 (0.5156) time: 0.4902 data: 0.0002 max mem: 52126 +Epoch: [24] [ 390/3634] eta: 0:27:30 lr: 0.000004 loss: 0.5323 (0.5168) time: 0.5522 data: 0.0002 max mem: 52126 +Epoch: [24] [ 400/3634] eta: 0:27:25 lr: 0.000004 loss: 0.5250 (0.5163) time: 0.5064 data: 0.0002 max mem: 52126 +Epoch: [24] [ 410/3634] eta: 0:27:10 lr: 0.000004 loss: 0.5250 (0.5178) time: 0.4441 data: 0.0002 max mem: 52126 +Epoch: [24] [ 420/3634] eta: 0:27:14 lr: 0.000004 loss: 0.5639 (0.5187) time: 0.5066 data: 0.0002 max mem: 52126 +Epoch: [24] [ 430/3634] eta: 0:27:09 lr: 0.000004 loss: 0.5348 (0.5176) time: 0.5673 data: 0.0002 max mem: 52126 +Epoch: [24] [ 440/3634] eta: 0:27:04 lr: 0.000004 loss: 0.4751 (0.5168) time: 0.5058 data: 0.0002 max mem: 52126 +Epoch: [24] [ 450/3634] eta: 0:27:07 lr: 0.000004 loss: 0.4464 (0.5153) time: 0.5685 data: 0.0002 max mem: 52126 +Epoch: [24] [ 460/3634] eta: 0:26:53 lr: 0.000004 loss: 0.4907 (0.5157) time: 0.5068 data: 0.0002 max mem: 52126 +Epoch: [24] [ 470/3634] eta: 0:26:46 lr: 0.000004 loss: 0.5073 (0.5145) time: 0.4327 data: 0.0002 max mem: 52126 +Epoch: [24] [ 480/3634] eta: 0:26:53 lr: 0.000004 loss: 0.5180 (0.5151) time: 0.5864 data: 0.0002 max mem: 52126 +Epoch: [24] [ 490/3634] eta: 0:26:40 lr: 0.000004 loss: 0.5218 (0.5149) time: 0.5374 data: 0.0002 max mem: 52126 +Epoch: [24] [ 500/3634] eta: 0:26:40 lr: 0.000004 loss: 0.4841 (0.5146) time: 0.4894 data: 0.0002 max mem: 52126 +Epoch: [24] [ 510/3634] eta: 0:26:33 lr: 0.000004 loss: 0.5049 (0.5147) time: 0.5344 data: 0.0002 max mem: 52126 +Epoch: [24] [ 520/3634] eta: 0:26:25 lr: 0.000004 loss: 0.5321 (0.5150) time: 0.4724 data: 0.0002 max mem: 52126 +Epoch: [24] [ 530/3634] eta: 0:26:13 lr: 0.000004 loss: 0.5284 (0.5150) time: 0.4273 data: 0.0002 max mem: 52126 +Epoch: [24] [ 540/3634] eta: 0:26:19 lr: 0.000004 loss: 0.5288 (0.5155) time: 0.5468 data: 0.0002 max mem: 52126 +Epoch: [24] [ 550/3634] eta: 0:26:17 lr: 0.000004 loss: 0.5480 (0.5160) time: 0.6372 data: 0.0001 max mem: 52126 +Epoch: [24] [ 560/3634] eta: 0:26:16 lr: 0.000004 loss: 0.5133 (0.5154) time: 0.5765 data: 0.0001 max mem: 52126 +Epoch: [24] [ 570/3634] eta: 0:26:04 lr: 0.000004 loss: 0.5133 (0.5159) time: 0.4857 data: 0.0002 max mem: 52126 +Epoch: [24] [ 580/3634] eta: 0:25:59 lr: 0.000004 loss: 0.5475 (0.5161) time: 0.4433 data: 0.0001 max mem: 52126 +Epoch: [24] [ 590/3634] eta: 0:25:47 lr: 0.000004 loss: 0.5320 (0.5163) time: 0.4439 data: 0.0001 max mem: 52126 +Epoch: [24] [ 600/3634] eta: 0:25:53 lr: 0.000004 loss: 0.5237 (0.5159) time: 0.5502 data: 0.0001 max mem: 52126 +Epoch: [24] [ 610/3634] eta: 0:25:41 lr: 0.000004 loss: 0.5194 (0.5160) time: 0.5493 data: 0.0001 max mem: 52126 +Epoch: [24] [ 620/3634] eta: 0:25:36 lr: 0.000004 loss: 0.5194 (0.5159) time: 0.4411 data: 0.0001 max mem: 52126 +Epoch: [24] [ 630/3634] eta: 0:25:29 lr: 0.000004 loss: 0.5126 (0.5160) time: 0.4854 data: 0.0001 max mem: 52126 +Epoch: [24] [ 640/3634] eta: 0:25:27 lr: 0.000004 loss: 0.5126 (0.5156) time: 0.5315 data: 0.0001 max mem: 52126 +Epoch: [24] [ 650/3634] eta: 0:25:21 lr: 0.000004 loss: 0.4990 (0.5154) time: 0.5322 data: 0.0001 max mem: 52126 +Epoch: [24] [ 660/3634] eta: 0:25:10 lr: 0.000004 loss: 0.5458 (0.5159) time: 0.4276 data: 0.0001 max mem: 52126 +Epoch: [24] [ 670/3634] eta: 0:25:13 lr: 0.000004 loss: 0.5048 (0.5154) time: 0.5362 data: 0.0001 max mem: 52126 +Epoch: [24] [ 680/3634] eta: 0:25:06 lr: 0.000004 loss: 0.4939 (0.5157) time: 0.5843 data: 0.0001 max mem: 52126 +Epoch: [24] [ 690/3634] eta: 0:25:01 lr: 0.000004 loss: 0.5420 (0.5163) time: 0.4905 data: 0.0001 max mem: 52126 +Epoch: [24] [ 700/3634] eta: 0:24:50 lr: 0.000004 loss: 0.5252 (0.5160) time: 0.4419 data: 0.0001 max mem: 52126 +Epoch: [24] [ 710/3634] eta: 0:24:45 lr: 0.000004 loss: 0.4752 (0.5154) time: 0.4437 data: 0.0001 max mem: 52126 +Epoch: [24] [ 720/3634] eta: 0:24:39 lr: 0.000004 loss: 0.5102 (0.5155) time: 0.4938 data: 0.0001 max mem: 52126 +Epoch: [24] [ 730/3634] eta: 0:24:39 lr: 0.000004 loss: 0.5144 (0.5154) time: 0.5540 data: 0.0001 max mem: 52126 +Epoch: [24] [ 740/3634] eta: 0:24:29 lr: 0.000004 loss: 0.5387 (0.5158) time: 0.5045 data: 0.0001 max mem: 52126 +Epoch: [24] [ 750/3634] eta: 0:24:27 lr: 0.000004 loss: 0.5387 (0.5154) time: 0.4921 data: 0.0001 max mem: 52126 +Epoch: [24] [ 760/3634] eta: 0:24:17 lr: 0.000004 loss: 0.4919 (0.5153) time: 0.4916 data: 0.0001 max mem: 52126 +Epoch: [24] [ 770/3634] eta: 0:24:12 lr: 0.000004 loss: 0.5066 (0.5155) time: 0.4425 data: 0.0002 max mem: 52126 +Epoch: [24] [ 780/3634] eta: 0:24:07 lr: 0.000004 loss: 0.5197 (0.5153) time: 0.5030 data: 0.0002 max mem: 52126 +Epoch: [24] [ 790/3634] eta: 0:24:00 lr: 0.000004 loss: 0.5197 (0.5152) time: 0.4857 data: 0.0002 max mem: 52126 +Epoch: [24] [ 800/3634] eta: 0:23:54 lr: 0.000004 loss: 0.5310 (0.5150) time: 0.4717 data: 0.0001 max mem: 52126 +Epoch: [24] [ 810/3634] eta: 0:23:54 lr: 0.000004 loss: 0.5331 (0.5154) time: 0.5557 data: 0.0001 max mem: 52126 +Epoch: [24] [ 820/3634] eta: 0:23:48 lr: 0.000004 loss: 0.5081 (0.5149) time: 0.5683 data: 0.0002 max mem: 52126 +Epoch: [24] [ 830/3634] eta: 0:23:42 lr: 0.000004 loss: 0.4956 (0.5146) time: 0.4873 data: 0.0002 max mem: 52126 +Epoch: [24] [ 840/3634] eta: 0:23:37 lr: 0.000004 loss: 0.5453 (0.5151) time: 0.4882 data: 0.0002 max mem: 52126 +Epoch: [24] [ 850/3634] eta: 0:23:28 lr: 0.000004 loss: 0.5453 (0.5153) time: 0.4434 data: 0.0002 max mem: 52126 +Epoch: [24] [ 860/3634] eta: 0:23:28 lr: 0.000004 loss: 0.5296 (0.5148) time: 0.5332 data: 0.0001 max mem: 52126 +Epoch: [24] [ 870/3634] eta: 0:23:23 lr: 0.000004 loss: 0.5248 (0.5148) time: 0.5908 data: 0.0001 max mem: 52126 +Epoch: [24] [ 880/3634] eta: 0:23:25 lr: 0.000004 loss: 0.5248 (0.5147) time: 0.6214 data: 0.0001 max mem: 52126 +Epoch: [24] [ 890/3634] eta: 0:23:16 lr: 0.000004 loss: 0.5154 (0.5148) time: 0.5632 data: 0.0001 max mem: 52126 +Epoch: [24] [ 900/3634] eta: 0:23:07 lr: 0.000004 loss: 0.5285 (0.5150) time: 0.3823 data: 0.0001 max mem: 52126 +Epoch: [24] [ 910/3634] eta: 0:22:59 lr: 0.000004 loss: 0.5285 (0.5148) time: 0.3825 data: 0.0001 max mem: 52126 +Epoch: [24] [ 920/3634] eta: 0:22:57 lr: 0.000004 loss: 0.5247 (0.5150) time: 0.5010 data: 0.0001 max mem: 52126 +Epoch: [24] [ 930/3634] eta: 0:22:48 lr: 0.000004 loss: 0.5259 (0.5151) time: 0.5007 data: 0.0001 max mem: 52126 +Epoch: [24] [ 940/3634] eta: 0:22:39 lr: 0.000004 loss: 0.5259 (0.5151) time: 0.3828 data: 0.0001 max mem: 52126 +Epoch: [24] [ 950/3634] eta: 0:22:34 lr: 0.000004 loss: 0.5460 (0.5155) time: 0.4277 data: 0.0002 max mem: 52126 +Epoch: [24] [ 960/3634] eta: 0:22:25 lr: 0.000004 loss: 0.5429 (0.5155) time: 0.4279 data: 0.0001 max mem: 52126 +Epoch: [24] [ 970/3634] eta: 0:22:19 lr: 0.000004 loss: 0.5237 (0.5154) time: 0.4271 data: 0.0001 max mem: 52126 +Epoch: [24] [ 980/3634] eta: 0:22:14 lr: 0.000004 loss: 0.5190 (0.5153) time: 0.4782 data: 0.0002 max mem: 52126 +Epoch: [24] [ 990/3634] eta: 0:22:08 lr: 0.000004 loss: 0.5190 (0.5152) time: 0.4847 data: 0.0002 max mem: 52126 +Epoch: [24] [1000/3634] eta: 0:22:06 lr: 0.000004 loss: 0.5169 (0.5153) time: 0.5401 data: 0.0001 max mem: 52126 +Epoch: [24] [1010/3634] eta: 0:21:57 lr: 0.000004 loss: 0.5497 (0.5157) time: 0.4901 data: 0.0001 max mem: 52126 +Epoch: [24] [1020/3634] eta: 0:21:49 lr: 0.000004 loss: 0.5575 (0.5159) time: 0.3839 data: 0.0001 max mem: 52126 +Epoch: [24] [1030/3634] eta: 0:21:47 lr: 0.000004 loss: 0.5489 (0.5162) time: 0.4880 data: 0.0001 max mem: 52126 +Epoch: [24] [1040/3634] eta: 0:21:42 lr: 0.000004 loss: 0.5489 (0.5165) time: 0.5460 data: 0.0001 max mem: 52126 +Epoch: [24] [1050/3634] eta: 0:21:39 lr: 0.000004 loss: 0.5301 (0.5163) time: 0.5447 data: 0.0001 max mem: 52126 +Epoch: [24] [1060/3634] eta: 0:21:33 lr: 0.000004 loss: 0.5407 (0.5166) time: 0.5306 data: 0.0001 max mem: 52126 +Epoch: [24] [1070/3634] eta: 0:21:30 lr: 0.000004 loss: 0.5595 (0.5170) time: 0.5362 data: 0.0001 max mem: 52126 +Epoch: [24] [1080/3634] eta: 0:21:25 lr: 0.000004 loss: 0.5324 (0.5170) time: 0.5503 data: 0.0001 max mem: 52126 +Epoch: [24] [1090/3634] eta: 0:21:20 lr: 0.000004 loss: 0.5157 (0.5170) time: 0.5000 data: 0.0001 max mem: 52126 +Epoch: [24] [1100/3634] eta: 0:21:15 lr: 0.000004 loss: 0.5188 (0.5171) time: 0.5004 data: 0.0001 max mem: 52126 +Epoch: [24] [1110/3634] eta: 0:21:12 lr: 0.000004 loss: 0.5112 (0.5169) time: 0.5494 data: 0.0001 max mem: 52126 +Epoch: [24] [1120/3634] eta: 0:21:06 lr: 0.000004 loss: 0.5292 (0.5171) time: 0.5345 data: 0.0001 max mem: 52126 +Epoch: [24] [1130/3634] eta: 0:21:01 lr: 0.000004 loss: 0.5292 (0.5171) time: 0.4715 data: 0.0001 max mem: 52126 +Epoch: [24] [1140/3634] eta: 0:21:02 lr: 0.000004 loss: 0.4993 (0.5169) time: 0.6338 data: 0.0001 max mem: 52126 +Epoch: [24] [1150/3634] eta: 0:20:56 lr: 0.000004 loss: 0.4993 (0.5167) time: 0.6403 data: 0.0001 max mem: 52126 +Epoch: [24] [1160/3634] eta: 0:20:49 lr: 0.000004 loss: 0.5187 (0.5168) time: 0.4341 data: 0.0002 max mem: 52126 +Epoch: [24] [1170/3634] eta: 0:20:41 lr: 0.000004 loss: 0.5247 (0.5168) time: 0.3830 data: 0.0002 max mem: 52126 +Epoch: [24] [1180/3634] eta: 0:20:34 lr: 0.000004 loss: 0.5112 (0.5167) time: 0.3832 data: 0.0002 max mem: 52126 +Epoch: [24] [1190/3634] eta: 0:20:28 lr: 0.000004 loss: 0.4787 (0.5165) time: 0.4292 data: 0.0002 max mem: 52126 +Epoch: [24] [1200/3634] eta: 0:20:27 lr: 0.000004 loss: 0.5243 (0.5166) time: 0.5837 data: 0.0001 max mem: 52126 +Epoch: [24] [1210/3634] eta: 0:20:19 lr: 0.000004 loss: 0.5396 (0.5169) time: 0.5378 data: 0.0002 max mem: 52126 +Epoch: [24] [1220/3634] eta: 0:20:14 lr: 0.000004 loss: 0.5509 (0.5167) time: 0.4291 data: 0.0002 max mem: 52126 +Epoch: [24] [1230/3634] eta: 0:20:06 lr: 0.000004 loss: 0.4571 (0.5163) time: 0.4291 data: 0.0002 max mem: 52126 +Epoch: [24] [1240/3634] eta: 0:20:01 lr: 0.000004 loss: 0.4619 (0.5160) time: 0.4418 data: 0.0001 max mem: 52126 +Epoch: [24] [1250/3634] eta: 0:19:58 lr: 0.000004 loss: 0.4909 (0.5159) time: 0.5458 data: 0.0001 max mem: 52126 +Epoch: [24] [1260/3634] eta: 0:19:51 lr: 0.000004 loss: 0.5204 (0.5159) time: 0.4886 data: 0.0001 max mem: 52126 +Epoch: [24] [1270/3634] eta: 0:19:44 lr: 0.000004 loss: 0.5234 (0.5159) time: 0.3844 data: 0.0002 max mem: 52126 +Epoch: [24] [1280/3634] eta: 0:19:43 lr: 0.000004 loss: 0.5160 (0.5158) time: 0.5610 data: 0.0002 max mem: 52126 +Epoch: [24] [1290/3634] eta: 0:19:38 lr: 0.000004 loss: 0.4984 (0.5157) time: 0.6206 data: 0.0002 max mem: 52126 +Epoch: [24] [1300/3634] eta: 0:19:32 lr: 0.000004 loss: 0.5314 (0.5159) time: 0.4872 data: 0.0002 max mem: 52126 +Epoch: [24] [1310/3634] eta: 0:19:25 lr: 0.000004 loss: 0.5395 (0.5160) time: 0.4280 data: 0.0001 max mem: 52126 +Epoch: [24] [1320/3634] eta: 0:19:22 lr: 0.000004 loss: 0.5354 (0.5161) time: 0.4868 data: 0.0001 max mem: 52126 +Epoch: [24] [1330/3634] eta: 0:19:16 lr: 0.000004 loss: 0.5274 (0.5161) time: 0.5372 data: 0.0001 max mem: 52126 +Epoch: [24] [1340/3634] eta: 0:19:11 lr: 0.000004 loss: 0.5217 (0.5162) time: 0.4826 data: 0.0001 max mem: 52126 +Epoch: [24] [1350/3634] eta: 0:19:08 lr: 0.000004 loss: 0.5310 (0.5164) time: 0.5391 data: 0.0001 max mem: 52126 +Epoch: [24] [1360/3634] eta: 0:19:04 lr: 0.000004 loss: 0.5308 (0.5164) time: 0.5940 data: 0.0001 max mem: 52126 +Epoch: [24] [1370/3634] eta: 0:18:57 lr: 0.000004 loss: 0.5493 (0.5166) time: 0.4876 data: 0.0001 max mem: 52126 +Epoch: [24] [1380/3634] eta: 0:18:52 lr: 0.000004 loss: 0.5577 (0.5166) time: 0.4290 data: 0.0001 max mem: 52126 +Epoch: [24] [1390/3634] eta: 0:18:45 lr: 0.000004 loss: 0.4866 (0.5165) time: 0.4292 data: 0.0001 max mem: 52126 +Epoch: [24] [1400/3634] eta: 0:18:39 lr: 0.000004 loss: 0.4837 (0.5163) time: 0.4261 data: 0.0001 max mem: 52126 +Epoch: [24] [1410/3634] eta: 0:18:37 lr: 0.000004 loss: 0.5002 (0.5164) time: 0.5917 data: 0.0001 max mem: 52126 +Epoch: [24] [1420/3634] eta: 0:18:31 lr: 0.000004 loss: 0.5434 (0.5164) time: 0.5506 data: 0.0002 max mem: 52126 +Epoch: [24] [1430/3634] eta: 0:18:24 lr: 0.000004 loss: 0.5434 (0.5166) time: 0.3837 data: 0.0002 max mem: 52126 +Epoch: [24] [1440/3634] eta: 0:18:17 lr: 0.000004 loss: 0.5427 (0.5165) time: 0.3823 data: 0.0002 max mem: 52126 +Epoch: [24] [1450/3634] eta: 0:18:12 lr: 0.000004 loss: 0.5053 (0.5164) time: 0.4412 data: 0.0002 max mem: 52126 +Epoch: [24] [1460/3634] eta: 0:18:08 lr: 0.000004 loss: 0.5008 (0.5161) time: 0.5311 data: 0.0002 max mem: 52126 +Epoch: [24] [1470/3634] eta: 0:18:02 lr: 0.000004 loss: 0.5185 (0.5163) time: 0.5178 data: 0.0002 max mem: 52126 +Epoch: [24] [1480/3634] eta: 0:17:57 lr: 0.000004 loss: 0.5304 (0.5162) time: 0.4877 data: 0.0002 max mem: 52126 +Epoch: [24] [1490/3634] eta: 0:17:52 lr: 0.000004 loss: 0.5042 (0.5161) time: 0.5025 data: 0.0001 max mem: 52126 +Epoch: [24] [1500/3634] eta: 0:17:47 lr: 0.000004 loss: 0.5096 (0.5161) time: 0.5030 data: 0.0001 max mem: 52126 +Epoch: [24] [1510/3634] eta: 0:17:41 lr: 0.000004 loss: 0.5263 (0.5163) time: 0.4430 data: 0.0001 max mem: 52126 +Epoch: [24] [1520/3634] eta: 0:17:36 lr: 0.000004 loss: 0.5263 (0.5162) time: 0.4427 data: 0.0001 max mem: 52126 +Epoch: [24] [1530/3634] eta: 0:17:31 lr: 0.000004 loss: 0.5319 (0.5165) time: 0.4900 data: 0.0001 max mem: 52126 +Epoch: [24] [1540/3634] eta: 0:17:26 lr: 0.000004 loss: 0.5319 (0.5164) time: 0.4892 data: 0.0001 max mem: 52126 +Epoch: [24] [1550/3634] eta: 0:17:21 lr: 0.000004 loss: 0.5109 (0.5165) time: 0.5008 data: 0.0002 max mem: 52126 +Epoch: [24] [1560/3634] eta: 0:17:17 lr: 0.000004 loss: 0.4861 (0.5163) time: 0.5348 data: 0.0002 max mem: 52126 +Epoch: [24] [1570/3634] eta: 0:17:10 lr: 0.000004 loss: 0.5061 (0.5164) time: 0.4760 data: 0.0001 max mem: 52126 +Epoch: [24] [1580/3634] eta: 0:17:03 lr: 0.000004 loss: 0.5536 (0.5166) time: 0.3835 data: 0.0001 max mem: 52126 +Epoch: [24] [1590/3634] eta: 0:16:59 lr: 0.000004 loss: 0.5621 (0.5168) time: 0.4424 data: 0.0001 max mem: 52126 +Epoch: [24] [1600/3634] eta: 0:16:54 lr: 0.000004 loss: 0.5435 (0.5166) time: 0.5034 data: 0.0001 max mem: 52126 +Epoch: [24] [1610/3634] eta: 0:16:52 lr: 0.000004 loss: 0.5388 (0.5168) time: 0.6209 data: 0.0001 max mem: 52126 +Epoch: [24] [1620/3634] eta: 0:16:47 lr: 0.000004 loss: 0.5388 (0.5169) time: 0.6181 data: 0.0001 max mem: 52126 +Epoch: [24] [1630/3634] eta: 0:16:42 lr: 0.000004 loss: 0.5181 (0.5169) time: 0.5015 data: 0.0001 max mem: 52126 +Epoch: [24] [1640/3634] eta: 0:16:38 lr: 0.000004 loss: 0.5207 (0.5168) time: 0.5484 data: 0.0001 max mem: 52126 +Epoch: [24] [1650/3634] eta: 0:16:34 lr: 0.000004 loss: 0.5332 (0.5170) time: 0.5917 data: 0.0001 max mem: 52126 +Epoch: [24] [1660/3634] eta: 0:16:30 lr: 0.000004 loss: 0.5395 (0.5171) time: 0.5745 data: 0.0001 max mem: 52126 +Epoch: [24] [1670/3634] eta: 0:16:27 lr: 0.000004 loss: 0.5355 (0.5173) time: 0.6191 data: 0.0001 max mem: 52126 +Epoch: [24] [1680/3634] eta: 0:16:20 lr: 0.000004 loss: 0.5355 (0.5172) time: 0.5311 data: 0.0001 max mem: 52126 +Epoch: [24] [1690/3634] eta: 0:16:14 lr: 0.000004 loss: 0.5006 (0.5170) time: 0.3834 data: 0.0001 max mem: 52126 +Epoch: [24] [1700/3634] eta: 0:16:07 lr: 0.000004 loss: 0.4830 (0.5169) time: 0.3834 data: 0.0002 max mem: 52126 +Epoch: [24] [1710/3634] eta: 0:16:04 lr: 0.000004 loss: 0.5201 (0.5169) time: 0.5220 data: 0.0001 max mem: 52126 +Epoch: [24] [1720/3634] eta: 0:16:00 lr: 0.000004 loss: 0.5364 (0.5170) time: 0.6098 data: 0.0001 max mem: 52126 +Epoch: [24] [1730/3634] eta: 0:15:54 lr: 0.000004 loss: 0.5235 (0.5170) time: 0.4711 data: 0.0001 max mem: 52126 +Epoch: [24] [1740/3634] eta: 0:15:48 lr: 0.000004 loss: 0.5028 (0.5170) time: 0.4316 data: 0.0001 max mem: 52126 +Epoch: [24] [1750/3634] eta: 0:15:44 lr: 0.000004 loss: 0.5367 (0.5171) time: 0.5380 data: 0.0001 max mem: 52126 +Epoch: [24] [1760/3634] eta: 0:15:40 lr: 0.000004 loss: 0.5405 (0.5171) time: 0.5980 data: 0.0001 max mem: 52126 +Epoch: [24] [1770/3634] eta: 0:15:35 lr: 0.000004 loss: 0.5361 (0.5172) time: 0.5506 data: 0.0001 max mem: 52126 +Epoch: [24] [1780/3634] eta: 0:15:30 lr: 0.000004 loss: 0.5182 (0.5171) time: 0.5018 data: 0.0001 max mem: 52126 +Epoch: [24] [1790/3634] eta: 0:15:24 lr: 0.000004 loss: 0.4979 (0.5169) time: 0.4425 data: 0.0001 max mem: 52126 +Epoch: [24] [1800/3634] eta: 0:15:18 lr: 0.000004 loss: 0.5115 (0.5169) time: 0.3827 data: 0.0001 max mem: 52126 +Epoch: [24] [1810/3634] eta: 0:15:13 lr: 0.000004 loss: 0.5392 (0.5171) time: 0.4309 data: 0.0001 max mem: 52126 +Epoch: [24] [1820/3634] eta: 0:15:07 lr: 0.000004 loss: 0.5346 (0.5171) time: 0.4750 data: 0.0001 max mem: 52126 +Epoch: [24] [1830/3634] eta: 0:15:03 lr: 0.000004 loss: 0.5293 (0.5171) time: 0.5308 data: 0.0001 max mem: 52126 +Epoch: [24] [1840/3634] eta: 0:14:57 lr: 0.000004 loss: 0.5262 (0.5172) time: 0.4868 data: 0.0001 max mem: 52126 +Epoch: [24] [1850/3634] eta: 0:14:51 lr: 0.000004 loss: 0.5113 (0.5171) time: 0.3834 data: 0.0001 max mem: 52126 +Epoch: [24] [1860/3634] eta: 0:14:45 lr: 0.000004 loss: 0.5249 (0.5173) time: 0.3832 data: 0.0001 max mem: 52126 +Epoch: [24] [1870/3634] eta: 0:14:41 lr: 0.000004 loss: 0.5457 (0.5174) time: 0.4875 data: 0.0001 max mem: 52126 +Epoch: [24] [1880/3634] eta: 0:14:35 lr: 0.000004 loss: 0.5302 (0.5173) time: 0.4871 data: 0.0001 max mem: 52126 +Epoch: [24] [1890/3634] eta: 0:14:31 lr: 0.000004 loss: 0.5086 (0.5173) time: 0.5354 data: 0.0001 max mem: 52126 +Epoch: [24] [1900/3634] eta: 0:14:26 lr: 0.000004 loss: 0.5133 (0.5172) time: 0.5791 data: 0.0001 max mem: 52126 +Epoch: [24] [1910/3634] eta: 0:14:22 lr: 0.000004 loss: 0.4978 (0.5171) time: 0.5317 data: 0.0001 max mem: 52126 +Epoch: [24] [1920/3634] eta: 0:14:16 lr: 0.000004 loss: 0.4978 (0.5170) time: 0.4884 data: 0.0002 max mem: 52126 +Epoch: [24] [1930/3634] eta: 0:14:12 lr: 0.000004 loss: 0.5229 (0.5171) time: 0.5052 data: 0.0002 max mem: 52126 +Epoch: [24] [1940/3634] eta: 0:14:06 lr: 0.000004 loss: 0.5003 (0.5170) time: 0.5052 data: 0.0002 max mem: 52126 +Epoch: [24] [1950/3634] eta: 0:14:01 lr: 0.000004 loss: 0.5467 (0.5170) time: 0.4458 data: 0.0001 max mem: 52126 +Epoch: [24] [1960/3634] eta: 0:13:58 lr: 0.000004 loss: 0.5444 (0.5170) time: 0.6099 data: 0.0001 max mem: 52126 +Epoch: [24] [1970/3634] eta: 0:13:55 lr: 0.000004 loss: 0.5186 (0.5169) time: 0.7306 data: 0.0001 max mem: 52126 +Epoch: [24] [1980/3634] eta: 0:13:51 lr: 0.000004 loss: 0.5295 (0.5170) time: 0.6854 data: 0.0001 max mem: 52126 +Epoch: [24] [1990/3634] eta: 0:13:47 lr: 0.000004 loss: 0.5371 (0.5170) time: 0.6062 data: 0.0002 max mem: 52126 +Epoch: [24] [2000/3634] eta: 0:13:41 lr: 0.000004 loss: 0.5623 (0.5173) time: 0.4876 data: 0.0002 max mem: 52126 +Epoch: [24] [2010/3634] eta: 0:13:35 lr: 0.000004 loss: 0.5624 (0.5173) time: 0.3827 data: 0.0002 max mem: 52126 +Epoch: [24] [2020/3634] eta: 0:13:29 lr: 0.000004 loss: 0.5389 (0.5173) time: 0.3824 data: 0.0002 max mem: 52126 +Epoch: [24] [2030/3634] eta: 0:13:24 lr: 0.000004 loss: 0.5080 (0.5173) time: 0.4417 data: 0.0001 max mem: 52126 +Epoch: [24] [2040/3634] eta: 0:13:18 lr: 0.000004 loss: 0.5080 (0.5173) time: 0.4410 data: 0.0002 max mem: 52126 +Epoch: [24] [2050/3634] eta: 0:13:13 lr: 0.000004 loss: 0.5138 (0.5173) time: 0.4883 data: 0.0002 max mem: 52126 +Epoch: [24] [2060/3634] eta: 0:13:08 lr: 0.000004 loss: 0.5248 (0.5174) time: 0.4883 data: 0.0001 max mem: 52126 +Epoch: [24] [2070/3634] eta: 0:13:02 lr: 0.000004 loss: 0.5563 (0.5174) time: 0.3825 data: 0.0001 max mem: 52126 +Epoch: [24] [2080/3634] eta: 0:12:56 lr: 0.000004 loss: 0.5463 (0.5173) time: 0.4310 data: 0.0001 max mem: 52126 +Epoch: [24] [2090/3634] eta: 0:12:51 lr: 0.000004 loss: 0.4553 (0.5171) time: 0.4313 data: 0.0001 max mem: 52126 +Epoch: [24] [2100/3634] eta: 0:12:47 lr: 0.000004 loss: 0.5204 (0.5171) time: 0.5051 data: 0.0002 max mem: 52126 +Epoch: [24] [2110/3634] eta: 0:12:42 lr: 0.000004 loss: 0.5204 (0.5170) time: 0.6142 data: 0.0002 max mem: 52126 +Epoch: [24] [2120/3634] eta: 0:12:37 lr: 0.000004 loss: 0.5310 (0.5172) time: 0.5529 data: 0.0001 max mem: 52126 +Epoch: [24] [2130/3634] eta: 0:12:31 lr: 0.000004 loss: 0.5413 (0.5172) time: 0.4432 data: 0.0001 max mem: 52126 +Epoch: [24] [2140/3634] eta: 0:12:26 lr: 0.000004 loss: 0.5322 (0.5171) time: 0.3843 data: 0.0001 max mem: 52126 +Epoch: [24] [2150/3634] eta: 0:12:20 lr: 0.000004 loss: 0.5361 (0.5172) time: 0.4303 data: 0.0002 max mem: 52126 +Epoch: [24] [2160/3634] eta: 0:12:15 lr: 0.000004 loss: 0.5300 (0.5171) time: 0.4284 data: 0.0002 max mem: 52126 +Epoch: [24] [2170/3634] eta: 0:12:10 lr: 0.000004 loss: 0.5147 (0.5172) time: 0.4709 data: 0.0001 max mem: 52126 +Epoch: [24] [2180/3634] eta: 0:12:04 lr: 0.000004 loss: 0.5192 (0.5171) time: 0.4716 data: 0.0001 max mem: 52126 +Epoch: [24] [2190/3634] eta: 0:12:00 lr: 0.000004 loss: 0.5229 (0.5172) time: 0.4710 data: 0.0002 max mem: 52126 +Epoch: [24] [2200/3634] eta: 0:11:54 lr: 0.000004 loss: 0.5345 (0.5172) time: 0.4706 data: 0.0002 max mem: 52126 +Epoch: [24] [2210/3634] eta: 0:11:50 lr: 0.000004 loss: 0.5006 (0.5172) time: 0.5331 data: 0.0002 max mem: 52126 +Epoch: [24] [2220/3634] eta: 0:11:45 lr: 0.000004 loss: 0.5201 (0.5171) time: 0.5923 data: 0.0001 max mem: 52126 +Epoch: [24] [2230/3634] eta: 0:11:40 lr: 0.000004 loss: 0.5494 (0.5173) time: 0.4419 data: 0.0001 max mem: 52126 +Epoch: [24] [2240/3634] eta: 0:11:36 lr: 0.000004 loss: 0.5586 (0.5176) time: 0.5506 data: 0.0001 max mem: 52126 +Epoch: [24] [2250/3634] eta: 0:11:33 lr: 0.000004 loss: 0.5575 (0.5177) time: 0.7570 data: 0.0001 max mem: 52126 +Epoch: [24] [2260/3634] eta: 0:11:28 lr: 0.000004 loss: 0.5205 (0.5175) time: 0.6346 data: 0.0001 max mem: 52126 +Epoch: [24] [2270/3634] eta: 0:11:23 lr: 0.000004 loss: 0.4799 (0.5174) time: 0.4888 data: 0.0001 max mem: 52126 +Epoch: [24] [2280/3634] eta: 0:11:17 lr: 0.000004 loss: 0.5358 (0.5175) time: 0.4432 data: 0.0001 max mem: 52126 +Epoch: [24] [2290/3634] eta: 0:11:11 lr: 0.000004 loss: 0.5402 (0.5176) time: 0.3827 data: 0.0002 max mem: 52126 +Epoch: [24] [2300/3634] eta: 0:11:05 lr: 0.000004 loss: 0.5444 (0.5177) time: 0.3823 data: 0.0002 max mem: 52126 +Epoch: [24] [2310/3634] eta: 0:11:00 lr: 0.000004 loss: 0.5168 (0.5177) time: 0.3817 data: 0.0001 max mem: 52126 +Epoch: [24] [2320/3634] eta: 0:10:54 lr: 0.000004 loss: 0.5066 (0.5176) time: 0.3821 data: 0.0001 max mem: 52126 +Epoch: [24] [2330/3634] eta: 0:10:49 lr: 0.000004 loss: 0.4943 (0.5176) time: 0.4466 data: 0.0001 max mem: 52126 +Epoch: [24] [2340/3634] eta: 0:10:44 lr: 0.000004 loss: 0.5579 (0.5176) time: 0.4473 data: 0.0001 max mem: 52126 +Epoch: [24] [2350/3634] eta: 0:10:39 lr: 0.000004 loss: 0.5613 (0.5177) time: 0.4279 data: 0.0001 max mem: 52126 +Epoch: [24] [2360/3634] eta: 0:10:33 lr: 0.000004 loss: 0.5139 (0.5178) time: 0.4714 data: 0.0001 max mem: 52126 +Epoch: [24] [2370/3634] eta: 0:10:28 lr: 0.000004 loss: 0.5495 (0.5179) time: 0.4724 data: 0.0001 max mem: 52126 +Epoch: [24] [2380/3634] eta: 0:10:23 lr: 0.000004 loss: 0.5495 (0.5180) time: 0.4712 data: 0.0001 max mem: 52126 +Epoch: [24] [2390/3634] eta: 0:10:18 lr: 0.000004 loss: 0.5423 (0.5180) time: 0.4694 data: 0.0001 max mem: 52126 +Epoch: [24] [2400/3634] eta: 0:10:14 lr: 0.000004 loss: 0.5027 (0.5180) time: 0.5323 data: 0.0002 max mem: 52126 +Epoch: [24] [2410/3634] eta: 0:10:10 lr: 0.000004 loss: 0.5206 (0.5180) time: 0.6698 data: 0.0001 max mem: 52126 +Epoch: [24] [2420/3634] eta: 0:10:05 lr: 0.000004 loss: 0.5194 (0.5180) time: 0.6712 data: 0.0001 max mem: 52126 +Epoch: [24] [2430/3634] eta: 0:10:00 lr: 0.000004 loss: 0.5235 (0.5181) time: 0.4899 data: 0.0001 max mem: 52126 +Epoch: [24] [2440/3634] eta: 0:09:56 lr: 0.000004 loss: 0.5294 (0.5181) time: 0.5672 data: 0.0001 max mem: 52126 +Epoch: [24] [2450/3634] eta: 0:09:52 lr: 0.000004 loss: 0.5524 (0.5182) time: 0.6719 data: 0.0001 max mem: 52126 +Epoch: [24] [2460/3634] eta: 0:09:46 lr: 0.000004 loss: 0.5711 (0.5183) time: 0.5351 data: 0.0001 max mem: 52126 +Epoch: [24] [2470/3634] eta: 0:09:42 lr: 0.000004 loss: 0.5535 (0.5183) time: 0.5377 data: 0.0001 max mem: 52126 +Epoch: [24] [2480/3634] eta: 0:09:36 lr: 0.000004 loss: 0.4926 (0.5181) time: 0.4898 data: 0.0001 max mem: 52126 +Epoch: [24] [2490/3634] eta: 0:09:32 lr: 0.000004 loss: 0.5024 (0.5181) time: 0.5604 data: 0.0001 max mem: 52126 +Epoch: [24] [2500/3634] eta: 0:09:27 lr: 0.000004 loss: 0.5075 (0.5181) time: 0.6076 data: 0.0002 max mem: 52126 +Epoch: [24] [2510/3634] eta: 0:09:22 lr: 0.000004 loss: 0.4689 (0.5179) time: 0.4299 data: 0.0002 max mem: 52126 +Epoch: [24] [2520/3634] eta: 0:09:17 lr: 0.000004 loss: 0.4868 (0.5179) time: 0.4862 data: 0.0002 max mem: 52126 +Epoch: [24] [2530/3634] eta: 0:09:12 lr: 0.000004 loss: 0.5115 (0.5178) time: 0.5317 data: 0.0001 max mem: 52126 +Epoch: [24] [2540/3634] eta: 0:09:07 lr: 0.000004 loss: 0.5115 (0.5180) time: 0.4751 data: 0.0001 max mem: 52126 +Epoch: [24] [2550/3634] eta: 0:09:02 lr: 0.000004 loss: 0.5166 (0.5180) time: 0.5355 data: 0.0001 max mem: 52126 +Epoch: [24] [2560/3634] eta: 0:08:57 lr: 0.000004 loss: 0.5400 (0.5181) time: 0.4882 data: 0.0002 max mem: 52126 +Epoch: [24] [2570/3634] eta: 0:08:53 lr: 0.000004 loss: 0.5424 (0.5182) time: 0.5336 data: 0.0002 max mem: 52126 +Epoch: [24] [2580/3634] eta: 0:08:47 lr: 0.000004 loss: 0.5096 (0.5180) time: 0.5336 data: 0.0002 max mem: 52126 +Epoch: [24] [2590/3634] eta: 0:08:42 lr: 0.000004 loss: 0.4983 (0.5181) time: 0.3821 data: 0.0002 max mem: 52126 +Epoch: [24] [2600/3634] eta: 0:08:37 lr: 0.000004 loss: 0.5012 (0.5179) time: 0.5026 data: 0.0001 max mem: 52126 +Epoch: [24] [2610/3634] eta: 0:08:32 lr: 0.000004 loss: 0.5057 (0.5179) time: 0.5647 data: 0.0001 max mem: 52126 +Epoch: [24] [2620/3634] eta: 0:08:27 lr: 0.000004 loss: 0.5357 (0.5181) time: 0.4452 data: 0.0001 max mem: 52126 +Epoch: [24] [2630/3634] eta: 0:08:21 lr: 0.000004 loss: 0.5606 (0.5182) time: 0.3833 data: 0.0001 max mem: 52126 +Epoch: [24] [2640/3634] eta: 0:08:16 lr: 0.000004 loss: 0.5445 (0.5183) time: 0.4405 data: 0.0001 max mem: 52126 +Epoch: [24] [2650/3634] eta: 0:08:11 lr: 0.000004 loss: 0.5534 (0.5184) time: 0.4870 data: 0.0001 max mem: 52126 +Epoch: [24] [2660/3634] eta: 0:08:06 lr: 0.000004 loss: 0.5536 (0.5185) time: 0.4749 data: 0.0001 max mem: 52126 +Epoch: [24] [2670/3634] eta: 0:08:01 lr: 0.000004 loss: 0.5519 (0.5185) time: 0.4284 data: 0.0001 max mem: 52126 +Epoch: [24] [2680/3634] eta: 0:07:56 lr: 0.000004 loss: 0.5289 (0.5185) time: 0.4428 data: 0.0001 max mem: 52126 +Epoch: [24] [2690/3634] eta: 0:07:51 lr: 0.000004 loss: 0.5251 (0.5186) time: 0.5016 data: 0.0001 max mem: 52126 +Epoch: [24] [2700/3634] eta: 0:07:46 lr: 0.000004 loss: 0.5251 (0.5186) time: 0.5006 data: 0.0001 max mem: 52126 +Epoch: [24] [2710/3634] eta: 0:07:41 lr: 0.000004 loss: 0.5524 (0.5186) time: 0.4899 data: 0.0001 max mem: 52126 +Epoch: [24] [2720/3634] eta: 0:07:36 lr: 0.000004 loss: 0.5524 (0.5186) time: 0.4904 data: 0.0001 max mem: 52126 +Epoch: [24] [2730/3634] eta: 0:07:31 lr: 0.000004 loss: 0.5383 (0.5187) time: 0.6071 data: 0.0001 max mem: 52126 +Epoch: [24] [2740/3634] eta: 0:07:26 lr: 0.000004 loss: 0.5578 (0.5188) time: 0.5476 data: 0.0001 max mem: 52126 +Epoch: [24] [2750/3634] eta: 0:07:21 lr: 0.000004 loss: 0.5196 (0.5188) time: 0.4862 data: 0.0001 max mem: 52126 +Epoch: [24] [2760/3634] eta: 0:07:16 lr: 0.000004 loss: 0.5202 (0.5189) time: 0.5374 data: 0.0001 max mem: 52126 +Epoch: [24] [2770/3634] eta: 0:07:12 lr: 0.000004 loss: 0.5353 (0.5190) time: 0.6274 data: 0.0001 max mem: 52126 +Epoch: [24] [2780/3634] eta: 0:07:07 lr: 0.000004 loss: 0.5209 (0.5190) time: 0.6211 data: 0.0001 max mem: 52126 +Epoch: [24] [2790/3634] eta: 0:07:02 lr: 0.000004 loss: 0.5169 (0.5191) time: 0.4281 data: 0.0001 max mem: 52126 +Epoch: [24] [2800/3634] eta: 0:06:57 lr: 0.000004 loss: 0.4852 (0.5189) time: 0.4308 data: 0.0001 max mem: 52126 +Epoch: [24] [2810/3634] eta: 0:06:52 lr: 0.000004 loss: 0.5087 (0.5190) time: 0.4895 data: 0.0002 max mem: 52126 +Epoch: [24] [2820/3634] eta: 0:06:47 lr: 0.000004 loss: 0.5420 (0.5189) time: 0.5042 data: 0.0002 max mem: 52126 +Epoch: [24] [2830/3634] eta: 0:06:42 lr: 0.000004 loss: 0.5137 (0.5189) time: 0.5058 data: 0.0002 max mem: 52126 +Epoch: [24] [2840/3634] eta: 0:06:36 lr: 0.000004 loss: 0.5137 (0.5188) time: 0.4868 data: 0.0002 max mem: 52126 +Epoch: [24] [2850/3634] eta: 0:06:31 lr: 0.000004 loss: 0.5260 (0.5188) time: 0.4857 data: 0.0002 max mem: 52126 +Epoch: [24] [2860/3634] eta: 0:06:26 lr: 0.000004 loss: 0.5445 (0.5188) time: 0.4889 data: 0.0002 max mem: 52126 +Epoch: [24] [2870/3634] eta: 0:06:22 lr: 0.000004 loss: 0.5418 (0.5188) time: 0.5337 data: 0.0001 max mem: 52126 +Epoch: [24] [2880/3634] eta: 0:06:17 lr: 0.000004 loss: 0.5299 (0.5189) time: 0.5322 data: 0.0001 max mem: 52126 +Epoch: [24] [2890/3634] eta: 0:06:11 lr: 0.000004 loss: 0.5299 (0.5188) time: 0.4274 data: 0.0001 max mem: 52126 +Epoch: [24] [2900/3634] eta: 0:06:06 lr: 0.000004 loss: 0.5200 (0.5188) time: 0.4723 data: 0.0001 max mem: 52126 +Epoch: [24] [2910/3634] eta: 0:06:02 lr: 0.000004 loss: 0.5200 (0.5189) time: 0.5931 data: 0.0001 max mem: 52126 +Epoch: [24] [2920/3634] eta: 0:05:57 lr: 0.000004 loss: 0.5133 (0.5189) time: 0.5619 data: 0.0001 max mem: 52126 +Epoch: [24] [2930/3634] eta: 0:05:52 lr: 0.000004 loss: 0.5042 (0.5188) time: 0.5448 data: 0.0002 max mem: 52126 +Epoch: [24] [2940/3634] eta: 0:05:47 lr: 0.000004 loss: 0.5251 (0.5189) time: 0.5486 data: 0.0002 max mem: 52126 +Epoch: [24] [2950/3634] eta: 0:05:42 lr: 0.000004 loss: 0.5301 (0.5188) time: 0.4924 data: 0.0001 max mem: 52126 +Epoch: [24] [2960/3634] eta: 0:05:37 lr: 0.000004 loss: 0.5255 (0.5189) time: 0.5387 data: 0.0001 max mem: 52126 +Epoch: [24] [2970/3634] eta: 0:05:32 lr: 0.000004 loss: 0.5255 (0.5189) time: 0.4913 data: 0.0001 max mem: 52126 +Epoch: [24] [2980/3634] eta: 0:05:27 lr: 0.000004 loss: 0.5057 (0.5188) time: 0.3847 data: 0.0001 max mem: 52126 +Epoch: [24] [2990/3634] eta: 0:05:22 lr: 0.000004 loss: 0.5435 (0.5189) time: 0.4453 data: 0.0001 max mem: 52126 +Epoch: [24] [3000/3634] eta: 0:05:17 lr: 0.000004 loss: 0.5426 (0.5189) time: 0.5029 data: 0.0001 max mem: 52126 +Epoch: [24] [3010/3634] eta: 0:05:12 lr: 0.000004 loss: 0.5174 (0.5190) time: 0.5031 data: 0.0001 max mem: 52126 +Epoch: [24] [3020/3634] eta: 0:05:07 lr: 0.000004 loss: 0.5061 (0.5189) time: 0.5331 data: 0.0001 max mem: 52126 +Epoch: [24] [3030/3634] eta: 0:05:02 lr: 0.000004 loss: 0.5006 (0.5189) time: 0.5313 data: 0.0001 max mem: 52126 +Epoch: [24] [3040/3634] eta: 0:04:57 lr: 0.000004 loss: 0.5006 (0.5188) time: 0.4854 data: 0.0001 max mem: 52126 +Epoch: [24] [3050/3634] eta: 0:04:52 lr: 0.000004 loss: 0.4855 (0.5187) time: 0.5169 data: 0.0001 max mem: 52126 +Epoch: [24] [3060/3634] eta: 0:04:47 lr: 0.000004 loss: 0.4999 (0.5187) time: 0.5777 data: 0.0001 max mem: 52126 +Epoch: [24] [3070/3634] eta: 0:04:42 lr: 0.000004 loss: 0.5130 (0.5187) time: 0.5473 data: 0.0001 max mem: 52126 +Epoch: [24] [3080/3634] eta: 0:04:37 lr: 0.000004 loss: 0.5383 (0.5188) time: 0.5033 data: 0.0001 max mem: 52126 +Epoch: [24] [3090/3634] eta: 0:04:32 lr: 0.000004 loss: 0.5423 (0.5190) time: 0.4902 data: 0.0001 max mem: 52126 +Epoch: [24] [3100/3634] eta: 0:04:27 lr: 0.000004 loss: 0.5090 (0.5189) time: 0.4298 data: 0.0001 max mem: 52126 +Epoch: [24] [3110/3634] eta: 0:04:21 lr: 0.000004 loss: 0.4875 (0.5188) time: 0.3833 data: 0.0001 max mem: 52126 +Epoch: [24] [3120/3634] eta: 0:04:17 lr: 0.000004 loss: 0.4780 (0.5187) time: 0.4895 data: 0.0001 max mem: 52126 +Epoch: [24] [3130/3634] eta: 0:04:12 lr: 0.000004 loss: 0.4780 (0.5186) time: 0.5474 data: 0.0001 max mem: 52126 +Epoch: [24] [3140/3634] eta: 0:04:06 lr: 0.000004 loss: 0.5189 (0.5186) time: 0.4411 data: 0.0001 max mem: 52126 +Epoch: [24] [3150/3634] eta: 0:04:02 lr: 0.000004 loss: 0.5231 (0.5186) time: 0.4886 data: 0.0002 max mem: 52126 +Epoch: [24] [3160/3634] eta: 0:03:57 lr: 0.000004 loss: 0.5410 (0.5187) time: 0.5791 data: 0.0002 max mem: 52126 +Epoch: [24] [3170/3634] eta: 0:03:52 lr: 0.000004 loss: 0.5457 (0.5187) time: 0.5341 data: 0.0002 max mem: 52126 +Epoch: [24] [3180/3634] eta: 0:03:47 lr: 0.000004 loss: 0.5173 (0.5188) time: 0.4920 data: 0.0001 max mem: 52126 +Epoch: [24] [3190/3634] eta: 0:03:41 lr: 0.000004 loss: 0.5192 (0.5187) time: 0.4306 data: 0.0002 max mem: 52126 +Epoch: [24] [3200/3634] eta: 0:03:36 lr: 0.000004 loss: 0.5192 (0.5187) time: 0.3820 data: 0.0002 max mem: 52126 +Epoch: [24] [3210/3634] eta: 0:03:31 lr: 0.000004 loss: 0.5199 (0.5187) time: 0.3819 data: 0.0002 max mem: 52126 +Epoch: [24] [3220/3634] eta: 0:03:27 lr: 0.000004 loss: 0.4708 (0.5186) time: 0.5647 data: 0.0002 max mem: 52126 +Epoch: [24] [3230/3634] eta: 0:03:21 lr: 0.000004 loss: 0.4642 (0.5185) time: 0.5645 data: 0.0001 max mem: 52126 +Epoch: [24] [3240/3634] eta: 0:03:17 lr: 0.000004 loss: 0.5077 (0.5186) time: 0.5297 data: 0.0001 max mem: 52126 +Epoch: [24] [3250/3634] eta: 0:03:12 lr: 0.000004 loss: 0.5268 (0.5186) time: 0.6400 data: 0.0001 max mem: 52126 +Epoch: [24] [3260/3634] eta: 0:03:07 lr: 0.000004 loss: 0.5244 (0.5186) time: 0.5957 data: 0.0001 max mem: 52126 +Epoch: [24] [3270/3634] eta: 0:03:02 lr: 0.000004 loss: 0.5244 (0.5185) time: 0.5455 data: 0.0001 max mem: 52126 +Epoch: [24] [3280/3634] eta: 0:02:57 lr: 0.000004 loss: 0.5191 (0.5185) time: 0.5542 data: 0.0001 max mem: 52126 +Epoch: [24] [3290/3634] eta: 0:02:52 lr: 0.000004 loss: 0.4676 (0.5184) time: 0.4945 data: 0.0001 max mem: 52126 +Epoch: [24] [3300/3634] eta: 0:02:47 lr: 0.000004 loss: 0.4967 (0.5184) time: 0.3825 data: 0.0001 max mem: 52126 +Epoch: [24] [3310/3634] eta: 0:02:42 lr: 0.000004 loss: 0.5133 (0.5183) time: 0.4437 data: 0.0001 max mem: 52126 +Epoch: [24] [3320/3634] eta: 0:02:37 lr: 0.000004 loss: 0.4849 (0.5182) time: 0.4431 data: 0.0001 max mem: 52126 +Epoch: [24] [3330/3634] eta: 0:02:32 lr: 0.000004 loss: 0.5465 (0.5183) time: 0.4717 data: 0.0001 max mem: 52126 +Epoch: [24] [3340/3634] eta: 0:02:27 lr: 0.000004 loss: 0.5037 (0.5182) time: 0.5160 data: 0.0001 max mem: 52126 +Epoch: [24] [3350/3634] eta: 0:02:22 lr: 0.000004 loss: 0.5164 (0.5182) time: 0.4698 data: 0.0001 max mem: 52126 +Epoch: [24] [3360/3634] eta: 0:02:17 lr: 0.000004 loss: 0.5202 (0.5182) time: 0.5478 data: 0.0001 max mem: 52126 +Epoch: [24] [3370/3634] eta: 0:02:12 lr: 0.000004 loss: 0.4749 (0.5181) time: 0.6076 data: 0.0001 max mem: 52126 +Epoch: [24] [3380/3634] eta: 0:02:07 lr: 0.000004 loss: 0.4969 (0.5180) time: 0.4854 data: 0.0002 max mem: 52126 +Epoch: [24] [3390/3634] eta: 0:02:02 lr: 0.000004 loss: 0.5228 (0.5180) time: 0.4412 data: 0.0002 max mem: 52126 +Epoch: [24] [3400/3634] eta: 0:01:57 lr: 0.000004 loss: 0.5033 (0.5180) time: 0.5037 data: 0.0001 max mem: 52126 +Epoch: [24] [3410/3634] eta: 0:01:52 lr: 0.000004 loss: 0.4993 (0.5179) time: 0.5036 data: 0.0001 max mem: 52126 +Epoch: [24] [3420/3634] eta: 0:01:47 lr: 0.000004 loss: 0.5412 (0.5178) time: 0.4416 data: 0.0001 max mem: 52126 +Epoch: [24] [3430/3634] eta: 0:01:41 lr: 0.000004 loss: 0.5426 (0.5179) time: 0.3824 data: 0.0001 max mem: 52126 +Epoch: [24] [3440/3634] eta: 0:01:36 lr: 0.000004 loss: 0.5322 (0.5179) time: 0.4420 data: 0.0001 max mem: 52126 +Epoch: [24] [3450/3634] eta: 0:01:31 lr: 0.000004 loss: 0.5034 (0.5179) time: 0.4880 data: 0.0001 max mem: 52126 +Epoch: [24] [3460/3634] eta: 0:01:26 lr: 0.000004 loss: 0.5172 (0.5179) time: 0.5477 data: 0.0001 max mem: 52126 +Epoch: [24] [3470/3634] eta: 0:01:21 lr: 0.000004 loss: 0.4902 (0.5178) time: 0.5610 data: 0.0001 max mem: 52126 +Epoch: [24] [3480/3634] eta: 0:01:16 lr: 0.000004 loss: 0.4851 (0.5177) time: 0.5023 data: 0.0001 max mem: 52126 +Epoch: [24] [3490/3634] eta: 0:01:11 lr: 0.000004 loss: 0.4994 (0.5177) time: 0.4433 data: 0.0001 max mem: 52126 +Epoch: [24] [3500/3634] eta: 0:01:06 lr: 0.000004 loss: 0.5290 (0.5178) time: 0.4905 data: 0.0001 max mem: 52126 +Epoch: [24] [3510/3634] eta: 0:01:01 lr: 0.000004 loss: 0.5290 (0.5178) time: 0.5378 data: 0.0001 max mem: 52126 +Epoch: [24] [3520/3634] eta: 0:00:57 lr: 0.000004 loss: 0.5467 (0.5179) time: 0.5803 data: 0.0001 max mem: 52126 +Epoch: [24] [3530/3634] eta: 0:00:52 lr: 0.000004 loss: 0.5542 (0.5179) time: 0.5331 data: 0.0001 max mem: 52126 +Epoch: [24] [3540/3634] eta: 0:00:47 lr: 0.000004 loss: 0.5283 (0.5179) time: 0.4897 data: 0.0001 max mem: 52126 +Epoch: [24] [3550/3634] eta: 0:00:42 lr: 0.000004 loss: 0.5228 (0.5178) time: 0.5340 data: 0.0001 max mem: 52126 +Epoch: [24] [3560/3634] eta: 0:00:36 lr: 0.000004 loss: 0.4913 (0.5178) time: 0.4269 data: 0.0001 max mem: 52126 +Epoch: [24] [3570/3634] eta: 0:00:31 lr: 0.000004 loss: 0.5367 (0.5178) time: 0.3818 data: 0.0001 max mem: 52126 +Epoch: [24] [3580/3634] eta: 0:00:27 lr: 0.000004 loss: 0.5072 (0.5177) time: 0.5480 data: 0.0001 max mem: 52126 +Epoch: [24] [3590/3634] eta: 0:00:21 lr: 0.000004 loss: 0.4866 (0.5177) time: 0.5486 data: 0.0002 max mem: 52126 +Epoch: [24] [3600/3634] eta: 0:00:17 lr: 0.000004 loss: 0.5233 (0.5177) time: 0.5005 data: 0.0002 max mem: 52126 +Epoch: [24] [3610/3634] eta: 0:00:12 lr: 0.000004 loss: 0.5428 (0.5178) time: 0.5463 data: 0.0001 max mem: 52126 +Epoch: [24] [3620/3634] eta: 0:00:06 lr: 0.000004 loss: 0.5184 (0.5177) time: 0.4297 data: 0.0013 max mem: 52126 +Epoch: [24] [3630/3634] eta: 0:00:01 lr: 0.000004 loss: 0.5166 (0.5178) time: 0.4250 data: 0.0013 max mem: 52126 +Epoch: [24] [3633/3634] eta: 0:00:00 lr: 0.000004 loss: 0.5327 (0.5178) time: 0.5297 data: 0.0013 max mem: 52126 +Epoch: [24] Total time: 0:30:19 (0.5006 s / it) +Averaged stats: lr: 0.000004 loss: 0.5327 (0.5180) +Test: [ 0/1403] eta: 1:06:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0868 (0.0868) acc1: 100.0000 (100.0000) acc5: 100.0000 (100.0000) time: 2.8385 data: 2.7424 max mem: 52126 +Test: [ 10/1403] eta: 0:08:00 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1337 (0.1590) acc1: 98.4375 (96.1648) acc5: 100.0000 (100.0000) time: 0.3451 data: 0.2495 max mem: 52126 +Test: [ 20/1403] eta: 0:05:13 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1261 (0.1484) acc1: 98.4375 (96.4286) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 30/1403] eta: 0:04:13 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0822 (0.1436) acc1: 100.0000 (96.4214) acc5: 100.0000 (100.0000) time: 0.0959 data: 0.0002 max mem: 52126 +Test: [ 40/1403] eta: 0:03:41 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0975 (0.1441) acc1: 100.0000 (96.6082) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 50/1403] eta: 0:03:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1140 (0.1608) acc1: 100.0000 (95.7414) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 60/1403] eta: 0:03:09 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.1194 (0.1571) acc1: 100.0000 (96.2346) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 70/1403] eta: 0:02:59 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.1290 (0.1601) acc1: 100.0000 (96.0387) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 80/1403] eta: 0:02:51 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1406 (0.1671) acc1: 96.8750 (95.6983) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 90/1403] eta: 0:02:45 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1724 (0.1869) acc1: 95.3125 (94.7974) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 100/1403] eta: 0:02:40 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1618 (0.1856) acc1: 96.8750 (94.8793) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 110/1403] eta: 0:02:35 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1368 (0.1835) acc1: 98.4375 (94.9887) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 120/1403] eta: 0:02:31 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0788 (0.1748) acc1: 100.0000 (95.3900) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 130/1403] eta: 0:02:28 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0788 (0.1687) acc1: 100.0000 (95.6345) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 140/1403] eta: 0:02:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0756 (0.1620) acc1: 100.0000 (95.9441) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 150/1403] eta: 0:02:22 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0728 (0.1563) acc1: 100.0000 (96.1921) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 160/1403] eta: 0:02:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0758 (0.1518) acc1: 100.0000 (96.3800) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 170/1403] eta: 0:02:17 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0768 (0.1480) acc1: 100.0000 (96.5643) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 180/1403] eta: 0:02:15 acc1_pos: 92.188 acc5_pos: 100.000 loss: 0.0840 (0.1510) acc1: 100.0000 (96.4606) acc5: 100.0000 (100.0000) time: 0.0959 data: 0.0003 max mem: 52126 +Test: [ 190/1403] eta: 0:02:13 acc1_pos: 90.625 acc5_pos: 100.000 loss: 0.2386 (0.1580) acc1: 92.1875 (96.0978) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 200/1403] eta: 0:02:11 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.2386 (0.1680) acc1: 90.6250 (95.6390) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 210/1403] eta: 0:02:09 acc1_pos: 78.125 acc5_pos: 100.000 loss: 0.3004 (0.1768) acc1: 87.5000 (95.1274) acc5: 100.0000 (100.0000) time: 0.0959 data: 0.0003 max mem: 52126 +Test: [ 220/1403] eta: 0:02:08 acc1_pos: 90.625 acc5_pos: 100.000 loss: 0.3147 (0.1842) acc1: 89.0625 (94.7893) acc5: 100.0000 (100.0000) time: 0.0959 data: 0.0003 max mem: 52126 +Test: [ 230/1403] eta: 0:02:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2209 (0.1806) acc1: 93.7500 (94.9608) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 240/1403] eta: 0:02:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0762 (0.1764) acc1: 100.0000 (95.1569) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 250/1403] eta: 0:02:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0796 (0.1729) acc1: 100.0000 (95.3312) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 260/1403] eta: 0:02:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0796 (0.1695) acc1: 100.0000 (95.4921) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 270/1403] eta: 0:02:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0814 (0.1665) acc1: 100.0000 (95.6296) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 280/1403] eta: 0:01:58 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0812 (0.1634) acc1: 100.0000 (95.7796) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 290/1403] eta: 0:01:57 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0756 (0.1604) acc1: 100.0000 (95.9192) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 300/1403] eta: 0:01:55 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0748 (0.1577) acc1: 100.0000 (96.0496) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 310/1403] eta: 0:01:54 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0801 (0.1555) acc1: 100.0000 (96.1666) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 320/1403] eta: 0:01:53 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1012 (0.1539) acc1: 100.0000 (96.2860) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 330/1403] eta: 0:01:51 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1089 (0.1526) acc1: 100.0000 (96.3935) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 340/1403] eta: 0:01:50 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1089 (0.1513) acc1: 100.0000 (96.4947) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 350/1403] eta: 0:01:49 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1053 (0.1500) acc1: 100.0000 (96.5946) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 360/1403] eta: 0:01:47 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1091 (0.1490) acc1: 100.0000 (96.6802) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0959 data: 0.0003 max mem: 52126 +Test: [ 370/1403] eta: 0:01:46 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1082 (0.1478) acc1: 100.0000 (96.7697) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [ 380/1403] eta: 0:01:45 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1065 (0.1468) acc1: 100.0000 (96.8545) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 390/1403] eta: 0:01:44 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1080 (0.1479) acc1: 100.0000 (96.7911) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 400/1403] eta: 0:01:42 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2648 (0.1522) acc1: 89.0625 (96.5516) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 410/1403] eta: 0:01:41 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2932 (0.1558) acc1: 89.0625 (96.3428) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 420/1403] eta: 0:01:40 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2668 (0.1579) acc1: 89.0625 (96.2255) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 430/1403] eta: 0:01:39 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2569 (0.1612) acc1: 89.0625 (96.0339) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [ 440/1403] eta: 0:01:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3096 (0.1651) acc1: 87.5000 (95.8369) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 450/1403] eta: 0:01:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2985 (0.1677) acc1: 87.5000 (95.7040) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 460/1403] eta: 0:01:35 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2758 (0.1706) acc1: 87.5000 (95.5193) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0957 data: 0.0002 max mem: 52126 +Test: [ 470/1403] eta: 0:01:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2386 (0.1699) acc1: 90.6250 (95.5381) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 480/1403] eta: 0:01:33 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0628 (0.1677) acc1: 100.0000 (95.6308) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 490/1403] eta: 0:01:32 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0629 (0.1655) acc1: 100.0000 (95.7198) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 500/1403] eta: 0:01:31 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0637 (0.1635) acc1: 100.0000 (95.8053) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 510/1403] eta: 0:01:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0639 (0.1616) acc1: 100.0000 (95.8874) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 520/1403] eta: 0:01:29 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0673 (0.1598) acc1: 100.0000 (95.9663) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 530/1403] eta: 0:01:28 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0587 (0.1579) acc1: 100.0000 (96.0423) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 540/1403] eta: 0:01:27 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0580 (0.1561) acc1: 100.0000 (96.1154) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 550/1403] eta: 0:01:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0591 (0.1548) acc1: 100.0000 (96.1632) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 560/1403] eta: 0:01:24 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0846 (0.1538) acc1: 98.4375 (96.2010) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 570/1403] eta: 0:01:23 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0877 (0.1527) acc1: 98.4375 (96.2456) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 580/1403] eta: 0:01:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0903 (0.1522) acc1: 98.4375 (96.2591) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 590/1403] eta: 0:01:21 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1146 (0.1519) acc1: 96.8750 (96.2537) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 600/1403] eta: 0:01:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1330 (0.1522) acc1: 93.7500 (96.1990) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 610/1403] eta: 0:01:19 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1305 (0.1518) acc1: 95.3125 (96.2050) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 620/1403] eta: 0:01:18 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1370 (0.1527) acc1: 95.3125 (96.1051) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 630/1403] eta: 0:01:17 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1497 (0.1525) acc1: 93.7500 (96.0999) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 640/1403] eta: 0:01:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1177 (0.1521) acc1: 96.8750 (96.1218) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 650/1403] eta: 0:01:15 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1177 (0.1517) acc1: 96.8750 (96.1406) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 660/1403] eta: 0:01:14 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0963 (0.1506) acc1: 100.0000 (96.1871) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 670/1403] eta: 0:01:13 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0720 (0.1495) acc1: 100.0000 (96.2393) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 680/1403] eta: 0:01:12 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0825 (0.1508) acc1: 98.4375 (96.1408) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 690/1403] eta: 0:01:11 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2526 (0.1549) acc1: 87.5000 (95.9095) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 700/1403] eta: 0:01:10 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1793 (0.1551) acc1: 93.7500 (95.8943) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [ 710/1403] eta: 0:01:09 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3248 (0.1596) acc1: 85.9375 (95.6553) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 720/1403] eta: 0:01:08 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3248 (0.1596) acc1: 84.3750 (95.6527) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 730/1403] eta: 0:01:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0674 (0.1585) acc1: 100.0000 (95.7015) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 740/1403] eta: 0:01:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0752 (0.1576) acc1: 100.0000 (95.7384) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 750/1403] eta: 0:01:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1044 (0.1592) acc1: 96.8750 (95.6433) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 760/1403] eta: 0:01:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3305 (0.1635) acc1: 81.2500 (95.3700) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 770/1403] eta: 0:01:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4540 (0.1681) acc1: 75.0000 (95.0875) acc5: 100.0000 (100.0000) acc1_neg: 70.312 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 780/1403] eta: 0:01:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4203 (0.1706) acc1: 79.6875 (94.9424) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 790/1403] eta: 0:01:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3133 (0.1730) acc1: 84.3750 (94.8167) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [ 800/1403] eta: 0:00:59 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2214 (0.1734) acc1: 92.1875 (94.7975) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [ 810/1403] eta: 0:00:58 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2261 (0.1766) acc1: 89.0625 (94.6536) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.0959 data: 0.0003 max mem: 52126 +Test: [ 820/1403] eta: 0:00:57 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6422 (0.1856) acc1: 71.8750 (94.2639) acc5: 100.0000 (100.0000) acc1_neg: 42.188 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 830/1403] eta: 0:00:56 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.7072 (0.1884) acc1: 70.3125 (94.1242) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 840/1403] eta: 0:00:55 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4995 (0.1966) acc1: 70.3125 (93.7667) acc5: 100.0000 (100.0000) acc1_neg: 50.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 850/1403] eta: 0:00:54 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.0648 (0.2079) acc1: 54.6875 (93.2891) acc5: 100.0000 (100.0000) acc1_neg: 54.688 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 860/1403] eta: 0:00:53 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.0287 (0.2114) acc1: 56.2500 (93.1294) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 870/1403] eta: 0:00:52 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1303 (0.2103) acc1: 96.8750 (93.1777) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 880/1403] eta: 0:00:51 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0968 (0.2092) acc1: 98.4375 (93.2303) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 890/1403] eta: 0:00:50 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0870 (0.2078) acc1: 98.4375 (93.2958) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 900/1403] eta: 0:00:49 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0751 (0.2063) acc1: 100.0000 (93.3667) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 910/1403] eta: 0:00:48 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0789 (0.2054) acc1: 98.4375 (93.3984) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 920/1403] eta: 0:00:47 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1110 (0.2046) acc1: 96.8750 (93.4260) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 930/1403] eta: 0:00:46 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1055 (0.2035) acc1: 96.8750 (93.4781) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 940/1403] eta: 0:00:45 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1185 (0.2038) acc1: 96.8750 (93.4378) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 950/1403] eta: 0:00:44 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2696 (0.2049) acc1: 87.5000 (93.3590) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 960/1403] eta: 0:00:43 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2919 (0.2058) acc1: 87.5000 (93.2996) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 970/1403] eta: 0:00:42 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2715 (0.2069) acc1: 87.5000 (93.2367) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 980/1403] eta: 0:00:41 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2806 (0.2080) acc1: 89.0625 (93.1686) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 990/1403] eta: 0:00:40 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2754 (0.2086) acc1: 89.0625 (93.1225) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1000/1403] eta: 0:00:39 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2450 (0.2095) acc1: 89.0625 (93.0694) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [1010/1403] eta: 0:00:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2957 (0.2105) acc1: 89.0625 (93.0113) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1020/1403] eta: 0:00:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4411 (0.2170) acc1: 81.2500 (92.6543) acc5: 100.0000 (100.0000) acc1_neg: 35.938 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1030/1403] eta: 0:00:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5952 (0.2190) acc1: 57.8125 (92.5255) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1040/1403] eta: 0:00:35 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2172 (0.2188) acc1: 92.1875 (92.5312) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1050/1403] eta: 0:00:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1745 (0.2184) acc1: 93.7500 (92.5532) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [1060/1403] eta: 0:00:33 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1345 (0.2176) acc1: 96.8750 (92.5969) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [1070/1403] eta: 0:00:32 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1477 (0.2179) acc1: 96.8750 (92.5814) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1080/1403] eta: 0:00:31 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3607 (0.2196) acc1: 84.3750 (92.4867) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1090/1403] eta: 0:00:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4006 (0.2213) acc1: 81.2500 (92.3980) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [1100/1403] eta: 0:00:29 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3536 (0.2219) acc1: 85.9375 (92.3677) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [1110/1403] eta: 0:00:28 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2830 (0.2224) acc1: 89.0625 (92.3436) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [1120/1403] eta: 0:00:27 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2851 (0.2230) acc1: 89.0625 (92.3116) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [1130/1403] eta: 0:00:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2978 (0.2238) acc1: 87.5000 (92.2690) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1140/1403] eta: 0:00:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2437 (0.2239) acc1: 90.6250 (92.2560) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1150/1403] eta: 0:00:24 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2437 (0.2244) acc1: 90.6250 (92.2255) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.0959 data: 0.0003 max mem: 52126 +Test: [1160/1403] eta: 0:00:23 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2779 (0.2249) acc1: 89.0625 (92.1929) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0959 data: 0.0003 max mem: 52126 +Test: [1170/1403] eta: 0:00:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2654 (0.2248) acc1: 89.0625 (92.1942) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0960 data: 0.0003 max mem: 52126 +Test: [1180/1403] eta: 0:00:21 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0695 (0.2235) acc1: 100.0000 (92.2603) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0959 data: 0.0003 max mem: 52126 +Test: [1190/1403] eta: 0:00:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0688 (0.2222) acc1: 100.0000 (92.3253) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1200/1403] eta: 0:00:19 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0690 (0.2210) acc1: 100.0000 (92.3892) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [1210/1403] eta: 0:00:18 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0686 (0.2197) acc1: 100.0000 (92.4520) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0959 data: 0.0003 max mem: 52126 +Test: [1220/1403] eta: 0:00:17 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0682 (0.2185) acc1: 100.0000 (92.5138) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0959 data: 0.0003 max mem: 52126 +Test: [1230/1403] eta: 0:00:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0690 (0.2173) acc1: 100.0000 (92.5746) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [1240/1403] eta: 0:00:15 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0689 (0.2161) acc1: 100.0000 (92.6345) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1250/1403] eta: 0:00:14 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0694 (0.2149) acc1: 100.0000 (92.6933) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [1260/1403] eta: 0:00:14 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0750 (0.2138) acc1: 100.0000 (92.7513) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1270/1403] eta: 0:00:13 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0764 (0.2127) acc1: 100.0000 (92.8083) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1280/1403] eta: 0:00:12 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0761 (0.2117) acc1: 100.0000 (92.8645) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1290/1403] eta: 0:00:11 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0757 (0.2106) acc1: 100.0000 (92.9197) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [1300/1403] eta: 0:00:10 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0764 (0.2096) acc1: 100.0000 (92.9742) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [1310/1403] eta: 0:00:09 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0769 (0.2086) acc1: 100.0000 (93.0277) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [1320/1403] eta: 0:00:08 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0749 (0.2076) acc1: 100.0000 (93.0805) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1330/1403] eta: 0:00:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0742 (0.2066) acc1: 100.0000 (93.1313) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1340/1403] eta: 0:00:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0674 (0.2055) acc1: 100.0000 (93.1814) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1350/1403] eta: 0:00:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0660 (0.2045) acc1: 100.0000 (93.2284) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0959 data: 0.0003 max mem: 52126 +Test: [1360/1403] eta: 0:00:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0660 (0.2035) acc1: 100.0000 (93.2759) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [1370/1403] eta: 0:00:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0671 (0.2026) acc1: 100.0000 (93.3215) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1380/1403] eta: 0:00:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0654 (0.2016) acc1: 100.0000 (93.3687) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1390/1403] eta: 0:00:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0654 (0.2006) acc1: 100.0000 (93.4164) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0957 data: 0.0002 max mem: 52126 +Test: [1400/1403] eta: 0:00:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0657 (0.1997) acc1: 100.0000 (93.4611) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0956 data: 0.0001 max mem: 52126 +Test: [1402/1403] eta: 0:00:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0657 (0.1995) acc1: 100.0000 (93.4659) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0916 data: 0.0001 max mem: 52126 +Test: Total time: 0:02:17 (0.0981 s / it) +* Overall Acc@1 93.466 Acc@5 100.000 + target == real Acc@1 96.133 Acc@5 100.000 + target == fake Acc@1 92.714 Acc@5 100.000 + loss 0.200 +Accuracy of the network on the 89729 test images: 93.5% +Max accuracy: 93.49% +Epoch: [25] [ 0/3634] eta: 6:29:56 lr: 0.000004 loss: 0.4684 (0.4684) time: 6.4382 data: 4.1541 max mem: 52126 +Epoch: [25] [ 10/3634] eta: 1:16:38 lr: 0.000004 loss: 0.4684 (0.4836) time: 1.2690 data: 0.3778 max mem: 52126 +Epoch: [25] [ 20/3634] eta: 0:51:10 lr: 0.000004 loss: 0.5257 (0.4961) time: 0.5701 data: 0.0001 max mem: 52126 +Epoch: [25] [ 30/3634] eta: 0:44:18 lr: 0.000004 loss: 0.5190 (0.4951) time: 0.4454 data: 0.0001 max mem: 52126 +Epoch: [25] [ 40/3634] eta: 0:38:59 lr: 0.000004 loss: 0.5233 (0.5116) time: 0.4424 data: 0.0002 max mem: 52126 +Epoch: [25] [ 50/3634] eta: 0:36:50 lr: 0.000004 loss: 0.5590 (0.5167) time: 0.4292 data: 0.0002 max mem: 52126 +Epoch: [25] [ 60/3634] eta: 0:34:26 lr: 0.000004 loss: 0.4961 (0.5159) time: 0.4295 data: 0.0002 max mem: 52126 +Epoch: [25] [ 70/3634] eta: 0:35:30 lr: 0.000004 loss: 0.4961 (0.5161) time: 0.5499 data: 0.0002 max mem: 52126 +Epoch: [25] [ 80/3634] eta: 0:33:50 lr: 0.000004 loss: 0.5439 (0.5225) time: 0.5496 data: 0.0002 max mem: 52126 +Epoch: [25] [ 90/3634] eta: 0:34:52 lr: 0.000004 loss: 0.5605 (0.5240) time: 0.5636 data: 0.0001 max mem: 52126 +Epoch: [25] [ 100/3634] eta: 0:34:09 lr: 0.000004 loss: 0.5242 (0.5216) time: 0.6157 data: 0.0001 max mem: 52126 +Epoch: [25] [ 110/3634] eta: 0:33:32 lr: 0.000004 loss: 0.5242 (0.5230) time: 0.4840 data: 0.0001 max mem: 52126 +Epoch: [25] [ 120/3634] eta: 0:33:01 lr: 0.000004 loss: 0.5535 (0.5250) time: 0.4819 data: 0.0001 max mem: 52126 +Epoch: [25] [ 130/3634] eta: 0:32:40 lr: 0.000004 loss: 0.5248 (0.5214) time: 0.4954 data: 0.0001 max mem: 52126 +Epoch: [25] [ 140/3634] eta: 0:31:51 lr: 0.000004 loss: 0.5171 (0.5207) time: 0.4460 data: 0.0002 max mem: 52126 +Epoch: [25] [ 150/3634] eta: 0:32:06 lr: 0.000004 loss: 0.5251 (0.5208) time: 0.5085 data: 0.0002 max mem: 52126 +Epoch: [25] [ 160/3634] eta: 0:32:30 lr: 0.000004 loss: 0.5244 (0.5216) time: 0.6617 data: 0.0001 max mem: 52126 +Epoch: [25] [ 170/3634] eta: 0:32:08 lr: 0.000004 loss: 0.5244 (0.5219) time: 0.5859 data: 0.0002 max mem: 52126 +Epoch: [25] [ 180/3634] eta: 0:32:11 lr: 0.000004 loss: 0.5101 (0.5198) time: 0.5419 data: 0.0002 max mem: 52126 +Epoch: [25] [ 190/3634] eta: 0:31:56 lr: 0.000004 loss: 0.5050 (0.5179) time: 0.5532 data: 0.0002 max mem: 52126 +Epoch: [25] [ 200/3634] eta: 0:31:20 lr: 0.000004 loss: 0.5403 (0.5198) time: 0.4431 data: 0.0002 max mem: 52126 +Epoch: [25] [ 210/3634] eta: 0:31:27 lr: 0.000004 loss: 0.5368 (0.5186) time: 0.5038 data: 0.0001 max mem: 52126 +Epoch: [25] [ 220/3634] eta: 0:31:15 lr: 0.000004 loss: 0.5194 (0.5180) time: 0.5652 data: 0.0001 max mem: 52126 +Epoch: [25] [ 230/3634] eta: 0:31:03 lr: 0.000004 loss: 0.5238 (0.5178) time: 0.5050 data: 0.0001 max mem: 52126 +Epoch: [25] [ 240/3634] eta: 0:30:47 lr: 0.000004 loss: 0.5132 (0.5163) time: 0.4920 data: 0.0001 max mem: 52126 +Epoch: [25] [ 250/3634] eta: 0:30:21 lr: 0.000004 loss: 0.5176 (0.5166) time: 0.4324 data: 0.0002 max mem: 52126 +Epoch: [25] [ 260/3634] eta: 0:30:08 lr: 0.000004 loss: 0.5334 (0.5168) time: 0.4331 data: 0.0002 max mem: 52126 +Epoch: [25] [ 270/3634] eta: 0:29:55 lr: 0.000004 loss: 0.5116 (0.5159) time: 0.4767 data: 0.0002 max mem: 52126 +Epoch: [25] [ 280/3634] eta: 0:29:42 lr: 0.000004 loss: 0.5276 (0.5174) time: 0.4747 data: 0.0002 max mem: 52126 +Epoch: [25] [ 290/3634] eta: 0:29:20 lr: 0.000004 loss: 0.5316 (0.5160) time: 0.4289 data: 0.0002 max mem: 52126 +Epoch: [25] [ 300/3634] eta: 0:28:59 lr: 0.000004 loss: 0.4867 (0.5138) time: 0.3815 data: 0.0002 max mem: 52126 +Epoch: [25] [ 310/3634] eta: 0:28:49 lr: 0.000004 loss: 0.4939 (0.5143) time: 0.4312 data: 0.0002 max mem: 52126 +Epoch: [25] [ 320/3634] eta: 0:28:52 lr: 0.000004 loss: 0.5256 (0.5150) time: 0.5389 data: 0.0002 max mem: 52126 +Epoch: [25] [ 330/3634] eta: 0:28:42 lr: 0.000004 loss: 0.5193 (0.5146) time: 0.5392 data: 0.0002 max mem: 52126 +Epoch: [25] [ 340/3634] eta: 0:28:45 lr: 0.000004 loss: 0.5261 (0.5154) time: 0.5403 data: 0.0001 max mem: 52126 +Epoch: [25] [ 350/3634] eta: 0:28:38 lr: 0.000004 loss: 0.5150 (0.5146) time: 0.5550 data: 0.0002 max mem: 52126 +Epoch: [25] [ 360/3634] eta: 0:28:40 lr: 0.000004 loss: 0.5058 (0.5148) time: 0.5555 data: 0.0002 max mem: 52126 +Epoch: [25] [ 370/3634] eta: 0:28:22 lr: 0.000004 loss: 0.5487 (0.5153) time: 0.4910 data: 0.0001 max mem: 52126 +Epoch: [25] [ 380/3634] eta: 0:28:34 lr: 0.000004 loss: 0.5145 (0.5154) time: 0.5511 data: 0.0001 max mem: 52126 +Epoch: [25] [ 390/3634] eta: 0:28:27 lr: 0.000004 loss: 0.5336 (0.5158) time: 0.6126 data: 0.0001 max mem: 52126 +Epoch: [25] [ 400/3634] eta: 0:28:10 lr: 0.000004 loss: 0.5497 (0.5167) time: 0.4440 data: 0.0001 max mem: 52126 +Epoch: [25] [ 410/3634] eta: 0:28:20 lr: 0.000004 loss: 0.5229 (0.5151) time: 0.5525 data: 0.0002 max mem: 52126 +Epoch: [25] [ 420/3634] eta: 0:28:04 lr: 0.000004 loss: 0.4866 (0.5153) time: 0.5521 data: 0.0002 max mem: 52126 +Epoch: [25] [ 430/3634] eta: 0:27:55 lr: 0.000004 loss: 0.5403 (0.5160) time: 0.4275 data: 0.0001 max mem: 52126 +Epoch: [25] [ 440/3634] eta: 0:27:40 lr: 0.000004 loss: 0.5437 (0.5165) time: 0.4280 data: 0.0001 max mem: 52126 +Epoch: [25] [ 450/3634] eta: 0:27:40 lr: 0.000004 loss: 0.5006 (0.5154) time: 0.4896 data: 0.0002 max mem: 52126 +Epoch: [25] [ 460/3634] eta: 0:27:33 lr: 0.000004 loss: 0.4692 (0.5150) time: 0.5494 data: 0.0001 max mem: 52126 +Epoch: [25] [ 470/3634] eta: 0:27:25 lr: 0.000004 loss: 0.5121 (0.5149) time: 0.4925 data: 0.0001 max mem: 52126 +Epoch: [25] [ 480/3634] eta: 0:27:23 lr: 0.000004 loss: 0.5098 (0.5149) time: 0.5263 data: 0.0002 max mem: 52126 +Epoch: [25] [ 490/3634] eta: 0:27:09 lr: 0.000004 loss: 0.4849 (0.5147) time: 0.4757 data: 0.0002 max mem: 52126 +Epoch: [25] [ 500/3634] eta: 0:27:03 lr: 0.000004 loss: 0.4817 (0.5143) time: 0.4410 data: 0.0002 max mem: 52126 +Epoch: [25] [ 510/3634] eta: 0:27:03 lr: 0.000004 loss: 0.4878 (0.5143) time: 0.5517 data: 0.0001 max mem: 52126 +Epoch: [25] [ 520/3634] eta: 0:26:50 lr: 0.000004 loss: 0.5179 (0.5147) time: 0.4928 data: 0.0001 max mem: 52126 +Epoch: [25] [ 530/3634] eta: 0:26:56 lr: 0.000004 loss: 0.5243 (0.5149) time: 0.5525 data: 0.0002 max mem: 52126 +Epoch: [25] [ 540/3634] eta: 0:26:43 lr: 0.000004 loss: 0.5147 (0.5148) time: 0.5527 data: 0.0002 max mem: 52126 +Epoch: [25] [ 550/3634] eta: 0:26:31 lr: 0.000004 loss: 0.5150 (0.5150) time: 0.3826 data: 0.0001 max mem: 52126 +Epoch: [25] [ 560/3634] eta: 0:26:30 lr: 0.000004 loss: 0.5150 (0.5146) time: 0.4890 data: 0.0001 max mem: 52126 +Epoch: [25] [ 570/3634] eta: 0:26:17 lr: 0.000004 loss: 0.4896 (0.5147) time: 0.4891 data: 0.0001 max mem: 52126 +Epoch: [25] [ 580/3634] eta: 0:26:12 lr: 0.000004 loss: 0.5733 (0.5153) time: 0.4418 data: 0.0001 max mem: 52126 +Epoch: [25] [ 590/3634] eta: 0:26:00 lr: 0.000004 loss: 0.5237 (0.5147) time: 0.4414 data: 0.0001 max mem: 52126 +Epoch: [25] [ 600/3634] eta: 0:25:54 lr: 0.000004 loss: 0.5237 (0.5153) time: 0.4432 data: 0.0002 max mem: 52126 +Epoch: [25] [ 610/3634] eta: 0:25:42 lr: 0.000004 loss: 0.5323 (0.5153) time: 0.4437 data: 0.0001 max mem: 52126 +Epoch: [25] [ 620/3634] eta: 0:25:43 lr: 0.000004 loss: 0.5323 (0.5153) time: 0.5024 data: 0.0001 max mem: 52126 +Epoch: [25] [ 630/3634] eta: 0:25:41 lr: 0.000004 loss: 0.5395 (0.5158) time: 0.6055 data: 0.0001 max mem: 52126 +Epoch: [25] [ 640/3634] eta: 0:25:36 lr: 0.000004 loss: 0.5430 (0.5162) time: 0.5449 data: 0.0001 max mem: 52126 +Epoch: [25] [ 650/3634] eta: 0:25:29 lr: 0.000004 loss: 0.5380 (0.5161) time: 0.4878 data: 0.0001 max mem: 52126 +Epoch: [25] [ 660/3634] eta: 0:25:18 lr: 0.000004 loss: 0.5145 (0.5161) time: 0.4288 data: 0.0002 max mem: 52126 +Epoch: [25] [ 670/3634] eta: 0:25:18 lr: 0.000004 loss: 0.5253 (0.5159) time: 0.5042 data: 0.0002 max mem: 52126 +Epoch: [25] [ 680/3634] eta: 0:25:20 lr: 0.000004 loss: 0.5269 (0.5163) time: 0.6516 data: 0.0001 max mem: 52126 +Epoch: [25] [ 690/3634] eta: 0:25:09 lr: 0.000004 loss: 0.5384 (0.5162) time: 0.5298 data: 0.0001 max mem: 52126 +Epoch: [25] [ 700/3634] eta: 0:25:08 lr: 0.000004 loss: 0.5046 (0.5159) time: 0.5013 data: 0.0001 max mem: 52126 +Epoch: [25] [ 710/3634] eta: 0:25:05 lr: 0.000004 loss: 0.5046 (0.5157) time: 0.5886 data: 0.0001 max mem: 52126 +Epoch: [25] [ 720/3634] eta: 0:25:00 lr: 0.000004 loss: 0.5160 (0.5160) time: 0.5319 data: 0.0001 max mem: 52126 +Epoch: [25] [ 730/3634] eta: 0:24:54 lr: 0.000004 loss: 0.5160 (0.5155) time: 0.5035 data: 0.0002 max mem: 52126 +Epoch: [25] [ 740/3634] eta: 0:24:43 lr: 0.000004 loss: 0.4937 (0.5155) time: 0.4415 data: 0.0002 max mem: 52126 +Epoch: [25] [ 750/3634] eta: 0:24:38 lr: 0.000004 loss: 0.5328 (0.5160) time: 0.4406 data: 0.0002 max mem: 52126 +Epoch: [25] [ 760/3634] eta: 0:24:28 lr: 0.000004 loss: 0.5294 (0.5155) time: 0.4407 data: 0.0001 max mem: 52126 +Epoch: [25] [ 770/3634] eta: 0:24:21 lr: 0.000004 loss: 0.4944 (0.5157) time: 0.4272 data: 0.0001 max mem: 52126 +Epoch: [25] [ 780/3634] eta: 0:24:15 lr: 0.000004 loss: 0.4944 (0.5155) time: 0.4715 data: 0.0001 max mem: 52126 +Epoch: [25] [ 790/3634] eta: 0:24:14 lr: 0.000004 loss: 0.4993 (0.5152) time: 0.5488 data: 0.0002 max mem: 52126 +Epoch: [25] [ 800/3634] eta: 0:24:15 lr: 0.000004 loss: 0.5154 (0.5152) time: 0.6548 data: 0.0002 max mem: 52126 +Epoch: [25] [ 810/3634] eta: 0:24:12 lr: 0.000004 loss: 0.5203 (0.5153) time: 0.6375 data: 0.0001 max mem: 52126 +Epoch: [25] [ 820/3634] eta: 0:24:17 lr: 0.000004 loss: 0.5150 (0.5153) time: 0.6935 data: 0.0002 max mem: 52126 +Epoch: [25] [ 830/3634] eta: 0:24:07 lr: 0.000004 loss: 0.5150 (0.5157) time: 0.5888 data: 0.0001 max mem: 52126 +Epoch: [25] [ 840/3634] eta: 0:23:57 lr: 0.000004 loss: 0.5117 (0.5152) time: 0.3824 data: 0.0002 max mem: 52126 +Epoch: [25] [ 850/3634] eta: 0:23:52 lr: 0.000004 loss: 0.5023 (0.5152) time: 0.4406 data: 0.0001 max mem: 52126 +Epoch: [25] [ 860/3634] eta: 0:23:42 lr: 0.000004 loss: 0.5293 (0.5154) time: 0.4417 data: 0.0002 max mem: 52126 +Epoch: [25] [ 870/3634] eta: 0:23:44 lr: 0.000004 loss: 0.5188 (0.5152) time: 0.5484 data: 0.0002 max mem: 52126 +Epoch: [25] [ 880/3634] eta: 0:23:38 lr: 0.000004 loss: 0.4972 (0.5151) time: 0.6065 data: 0.0002 max mem: 52126 +Epoch: [25] [ 890/3634] eta: 0:23:32 lr: 0.000004 loss: 0.4972 (0.5151) time: 0.5000 data: 0.0002 max mem: 52126 +Epoch: [25] [ 900/3634] eta: 0:23:26 lr: 0.000004 loss: 0.5173 (0.5150) time: 0.4871 data: 0.0002 max mem: 52126 +Epoch: [25] [ 910/3634] eta: 0:23:17 lr: 0.000004 loss: 0.5275 (0.5153) time: 0.4288 data: 0.0002 max mem: 52126 +Epoch: [25] [ 920/3634] eta: 0:23:08 lr: 0.000004 loss: 0.5501 (0.5155) time: 0.3827 data: 0.0002 max mem: 52126 +Epoch: [25] [ 930/3634] eta: 0:23:02 lr: 0.000004 loss: 0.5103 (0.5149) time: 0.4272 data: 0.0002 max mem: 52126 +Epoch: [25] [ 940/3634] eta: 0:22:56 lr: 0.000004 loss: 0.5103 (0.5154) time: 0.4779 data: 0.0001 max mem: 52126 +Epoch: [25] [ 950/3634] eta: 0:22:47 lr: 0.000004 loss: 0.5484 (0.5153) time: 0.4334 data: 0.0002 max mem: 52126 +Epoch: [25] [ 960/3634] eta: 0:22:38 lr: 0.000004 loss: 0.5332 (0.5151) time: 0.3823 data: 0.0002 max mem: 52126 +Epoch: [25] [ 970/3634] eta: 0:22:30 lr: 0.000004 loss: 0.5131 (0.5149) time: 0.3835 data: 0.0002 max mem: 52126 +Epoch: [25] [ 980/3634] eta: 0:22:25 lr: 0.000004 loss: 0.5052 (0.5148) time: 0.4421 data: 0.0001 max mem: 52126 +Epoch: [25] [ 990/3634] eta: 0:22:20 lr: 0.000004 loss: 0.5144 (0.5150) time: 0.5037 data: 0.0002 max mem: 52126 +Epoch: [25] [1000/3634] eta: 0:22:20 lr: 0.000004 loss: 0.5549 (0.5154) time: 0.6092 data: 0.0002 max mem: 52126 +Epoch: [25] [1010/3634] eta: 0:22:12 lr: 0.000004 loss: 0.5591 (0.5156) time: 0.5460 data: 0.0001 max mem: 52126 +Epoch: [25] [1020/3634] eta: 0:22:09 lr: 0.000004 loss: 0.5244 (0.5157) time: 0.4866 data: 0.0001 max mem: 52126 +Epoch: [25] [1030/3634] eta: 0:22:03 lr: 0.000004 loss: 0.5134 (0.5155) time: 0.5339 data: 0.0001 max mem: 52126 +Epoch: [25] [1040/3634] eta: 0:21:58 lr: 0.000004 loss: 0.5129 (0.5154) time: 0.4903 data: 0.0002 max mem: 52126 +Epoch: [25] [1050/3634] eta: 0:21:49 lr: 0.000004 loss: 0.4810 (0.5150) time: 0.4434 data: 0.0002 max mem: 52126 +Epoch: [25] [1060/3634] eta: 0:21:44 lr: 0.000004 loss: 0.5072 (0.5151) time: 0.4412 data: 0.0001 max mem: 52126 +Epoch: [25] [1070/3634] eta: 0:21:36 lr: 0.000004 loss: 0.5194 (0.5151) time: 0.4404 data: 0.0001 max mem: 52126 +Epoch: [25] [1080/3634] eta: 0:21:28 lr: 0.000004 loss: 0.5022 (0.5148) time: 0.3817 data: 0.0002 max mem: 52126 +Epoch: [25] [1090/3634] eta: 0:21:22 lr: 0.000004 loss: 0.5206 (0.5151) time: 0.4294 data: 0.0002 max mem: 52126 +Epoch: [25] [1100/3634] eta: 0:21:17 lr: 0.000004 loss: 0.5339 (0.5149) time: 0.4792 data: 0.0001 max mem: 52126 +Epoch: [25] [1110/3634] eta: 0:21:09 lr: 0.000004 loss: 0.5463 (0.5152) time: 0.4324 data: 0.0001 max mem: 52126 +Epoch: [25] [1120/3634] eta: 0:21:09 lr: 0.000004 loss: 0.5463 (0.5153) time: 0.5632 data: 0.0001 max mem: 52126 +Epoch: [25] [1130/3634] eta: 0:21:04 lr: 0.000004 loss: 0.5335 (0.5154) time: 0.6226 data: 0.0001 max mem: 52126 +Epoch: [25] [1140/3634] eta: 0:20:59 lr: 0.000004 loss: 0.5406 (0.5155) time: 0.4863 data: 0.0001 max mem: 52126 +Epoch: [25] [1150/3634] eta: 0:20:53 lr: 0.000004 loss: 0.4829 (0.5150) time: 0.4882 data: 0.0001 max mem: 52126 +Epoch: [25] [1160/3634] eta: 0:20:46 lr: 0.000004 loss: 0.4829 (0.5150) time: 0.4441 data: 0.0001 max mem: 52126 +Epoch: [25] [1170/3634] eta: 0:20:43 lr: 0.000004 loss: 0.5065 (0.5149) time: 0.4999 data: 0.0001 max mem: 52126 +Epoch: [25] [1180/3634] eta: 0:20:37 lr: 0.000004 loss: 0.5510 (0.5152) time: 0.5443 data: 0.0001 max mem: 52126 +Epoch: [25] [1190/3634] eta: 0:20:35 lr: 0.000004 loss: 0.5647 (0.5155) time: 0.5488 data: 0.0002 max mem: 52126 +Epoch: [25] [1200/3634] eta: 0:20:30 lr: 0.000004 loss: 0.5392 (0.5154) time: 0.5626 data: 0.0002 max mem: 52126 +Epoch: [25] [1210/3634] eta: 0:20:26 lr: 0.000004 loss: 0.4935 (0.5150) time: 0.5432 data: 0.0002 max mem: 52126 +Epoch: [25] [1220/3634] eta: 0:20:22 lr: 0.000004 loss: 0.4972 (0.5150) time: 0.5722 data: 0.0002 max mem: 52126 +Epoch: [25] [1230/3634] eta: 0:20:21 lr: 0.000004 loss: 0.5311 (0.5150) time: 0.6191 data: 0.0002 max mem: 52126 +Epoch: [25] [1240/3634] eta: 0:20:15 lr: 0.000004 loss: 0.5279 (0.5153) time: 0.5763 data: 0.0002 max mem: 52126 +Epoch: [25] [1250/3634] eta: 0:20:10 lr: 0.000004 loss: 0.5330 (0.5153) time: 0.4868 data: 0.0002 max mem: 52126 +Epoch: [25] [1260/3634] eta: 0:20:04 lr: 0.000004 loss: 0.5424 (0.5154) time: 0.4887 data: 0.0001 max mem: 52126 +Epoch: [25] [1270/3634] eta: 0:19:57 lr: 0.000004 loss: 0.5489 (0.5158) time: 0.4298 data: 0.0001 max mem: 52126 +Epoch: [25] [1280/3634] eta: 0:19:51 lr: 0.000004 loss: 0.5417 (0.5162) time: 0.4272 data: 0.0002 max mem: 52126 +Epoch: [25] [1290/3634] eta: 0:19:45 lr: 0.000004 loss: 0.5202 (0.5162) time: 0.4701 data: 0.0002 max mem: 52126 +Epoch: [25] [1300/3634] eta: 0:19:42 lr: 0.000004 loss: 0.5202 (0.5164) time: 0.5448 data: 0.0001 max mem: 52126 +Epoch: [25] [1310/3634] eta: 0:19:36 lr: 0.000004 loss: 0.5300 (0.5163) time: 0.5451 data: 0.0001 max mem: 52126 +Epoch: [25] [1320/3634] eta: 0:19:31 lr: 0.000004 loss: 0.5244 (0.5162) time: 0.4867 data: 0.0002 max mem: 52126 +Epoch: [25] [1330/3634] eta: 0:19:28 lr: 0.000004 loss: 0.5506 (0.5165) time: 0.5668 data: 0.0002 max mem: 52126 +Epoch: [25] [1340/3634] eta: 0:19:25 lr: 0.000004 loss: 0.5571 (0.5166) time: 0.6114 data: 0.0001 max mem: 52126 +Epoch: [25] [1350/3634] eta: 0:19:23 lr: 0.000004 loss: 0.5571 (0.5169) time: 0.6377 data: 0.0001 max mem: 52126 +Epoch: [25] [1360/3634] eta: 0:19:17 lr: 0.000004 loss: 0.5675 (0.5172) time: 0.5923 data: 0.0002 max mem: 52126 +Epoch: [25] [1370/3634] eta: 0:19:10 lr: 0.000004 loss: 0.5146 (0.5169) time: 0.4420 data: 0.0002 max mem: 52126 +Epoch: [25] [1380/3634] eta: 0:19:05 lr: 0.000004 loss: 0.4896 (0.5168) time: 0.4422 data: 0.0002 max mem: 52126 +Epoch: [25] [1390/3634] eta: 0:19:00 lr: 0.000004 loss: 0.5120 (0.5166) time: 0.5024 data: 0.0002 max mem: 52126 +Epoch: [25] [1400/3634] eta: 0:18:53 lr: 0.000004 loss: 0.5120 (0.5167) time: 0.4428 data: 0.0001 max mem: 52126 +Epoch: [25] [1410/3634] eta: 0:18:46 lr: 0.000004 loss: 0.5205 (0.5167) time: 0.3825 data: 0.0002 max mem: 52126 +Epoch: [25] [1420/3634] eta: 0:18:42 lr: 0.000004 loss: 0.4954 (0.5164) time: 0.4860 data: 0.0002 max mem: 52126 +Epoch: [25] [1430/3634] eta: 0:18:36 lr: 0.000004 loss: 0.4949 (0.5162) time: 0.5354 data: 0.0002 max mem: 52126 +Epoch: [25] [1440/3634] eta: 0:18:31 lr: 0.000004 loss: 0.4949 (0.5163) time: 0.4916 data: 0.0002 max mem: 52126 +Epoch: [25] [1450/3634] eta: 0:18:24 lr: 0.000004 loss: 0.5258 (0.5164) time: 0.4423 data: 0.0002 max mem: 52126 +Epoch: [25] [1460/3634] eta: 0:18:21 lr: 0.000004 loss: 0.5258 (0.5163) time: 0.4909 data: 0.0002 max mem: 52126 +Epoch: [25] [1470/3634] eta: 0:18:14 lr: 0.000004 loss: 0.5204 (0.5165) time: 0.4911 data: 0.0001 max mem: 52126 +Epoch: [25] [1480/3634] eta: 0:18:08 lr: 0.000004 loss: 0.5414 (0.5165) time: 0.4287 data: 0.0001 max mem: 52126 +Epoch: [25] [1490/3634] eta: 0:18:03 lr: 0.000004 loss: 0.5384 (0.5166) time: 0.4874 data: 0.0002 max mem: 52126 +Epoch: [25] [1500/3634] eta: 0:17:56 lr: 0.000004 loss: 0.5236 (0.5166) time: 0.4412 data: 0.0002 max mem: 52126 +Epoch: [25] [1510/3634] eta: 0:17:51 lr: 0.000004 loss: 0.5146 (0.5167) time: 0.4421 data: 0.0001 max mem: 52126 +Epoch: [25] [1520/3634] eta: 0:17:46 lr: 0.000004 loss: 0.5303 (0.5168) time: 0.4870 data: 0.0002 max mem: 52126 +Epoch: [25] [1530/3634] eta: 0:17:39 lr: 0.000004 loss: 0.5412 (0.5168) time: 0.4279 data: 0.0001 max mem: 52126 +Epoch: [25] [1540/3634] eta: 0:17:35 lr: 0.000004 loss: 0.5137 (0.5168) time: 0.4895 data: 0.0001 max mem: 52126 +Epoch: [25] [1550/3634] eta: 0:17:30 lr: 0.000004 loss: 0.5219 (0.5169) time: 0.5384 data: 0.0001 max mem: 52126 +Epoch: [25] [1560/3634] eta: 0:17:25 lr: 0.000004 loss: 0.5219 (0.5169) time: 0.4781 data: 0.0001 max mem: 52126 +Epoch: [25] [1570/3634] eta: 0:17:22 lr: 0.000004 loss: 0.5313 (0.5170) time: 0.5897 data: 0.0001 max mem: 52126 +Epoch: [25] [1580/3634] eta: 0:17:15 lr: 0.000004 loss: 0.5331 (0.5171) time: 0.5431 data: 0.0002 max mem: 52126 +Epoch: [25] [1590/3634] eta: 0:17:12 lr: 0.000004 loss: 0.5162 (0.5171) time: 0.4864 data: 0.0002 max mem: 52126 +Epoch: [25] [1600/3634] eta: 0:17:06 lr: 0.000004 loss: 0.5088 (0.5172) time: 0.5389 data: 0.0001 max mem: 52126 +Epoch: [25] [1610/3634] eta: 0:17:01 lr: 0.000004 loss: 0.5106 (0.5172) time: 0.4805 data: 0.0001 max mem: 52126 +Epoch: [25] [1620/3634] eta: 0:16:54 lr: 0.000004 loss: 0.5182 (0.5172) time: 0.4285 data: 0.0001 max mem: 52126 +Epoch: [25] [1630/3634] eta: 0:16:48 lr: 0.000004 loss: 0.5182 (0.5172) time: 0.3822 data: 0.0002 max mem: 52126 +Epoch: [25] [1640/3634] eta: 0:16:43 lr: 0.000004 loss: 0.5376 (0.5172) time: 0.4417 data: 0.0002 max mem: 52126 +Epoch: [25] [1650/3634] eta: 0:16:39 lr: 0.000004 loss: 0.4931 (0.5170) time: 0.5463 data: 0.0002 max mem: 52126 +Epoch: [25] [1660/3634] eta: 0:16:32 lr: 0.000004 loss: 0.5130 (0.5171) time: 0.4875 data: 0.0002 max mem: 52126 +Epoch: [25] [1670/3634] eta: 0:16:27 lr: 0.000004 loss: 0.5373 (0.5172) time: 0.4411 data: 0.0002 max mem: 52126 +Epoch: [25] [1680/3634] eta: 0:16:21 lr: 0.000004 loss: 0.5219 (0.5172) time: 0.4410 data: 0.0002 max mem: 52126 +Epoch: [25] [1690/3634] eta: 0:16:16 lr: 0.000004 loss: 0.5009 (0.5172) time: 0.4444 data: 0.0001 max mem: 52126 +Epoch: [25] [1700/3634] eta: 0:16:11 lr: 0.000004 loss: 0.5324 (0.5175) time: 0.4959 data: 0.0001 max mem: 52126 +Epoch: [25] [1710/3634] eta: 0:16:04 lr: 0.000004 loss: 0.5440 (0.5173) time: 0.4346 data: 0.0001 max mem: 52126 +Epoch: [25] [1720/3634] eta: 0:15:58 lr: 0.000004 loss: 0.5230 (0.5175) time: 0.3829 data: 0.0002 max mem: 52126 +Epoch: [25] [1730/3634] eta: 0:15:53 lr: 0.000004 loss: 0.5116 (0.5172) time: 0.4287 data: 0.0001 max mem: 52126 +Epoch: [25] [1740/3634] eta: 0:15:50 lr: 0.000004 loss: 0.4757 (0.5170) time: 0.5791 data: 0.0002 max mem: 52126 +Epoch: [25] [1750/3634] eta: 0:15:45 lr: 0.000004 loss: 0.5255 (0.5171) time: 0.6222 data: 0.0002 max mem: 52126 +Epoch: [25] [1760/3634] eta: 0:15:42 lr: 0.000004 loss: 0.5283 (0.5171) time: 0.5947 data: 0.0001 max mem: 52126 +Epoch: [25] [1770/3634] eta: 0:15:37 lr: 0.000004 loss: 0.5227 (0.5171) time: 0.5660 data: 0.0002 max mem: 52126 +Epoch: [25] [1780/3634] eta: 0:15:33 lr: 0.000004 loss: 0.5021 (0.5168) time: 0.5895 data: 0.0002 max mem: 52126 +Epoch: [25] [1790/3634] eta: 0:15:29 lr: 0.000004 loss: 0.4810 (0.5168) time: 0.6476 data: 0.0001 max mem: 52126 +Epoch: [25] [1800/3634] eta: 0:15:23 lr: 0.000004 loss: 0.5423 (0.5168) time: 0.5008 data: 0.0002 max mem: 52126 +Epoch: [25] [1810/3634] eta: 0:15:17 lr: 0.000004 loss: 0.5225 (0.5168) time: 0.3827 data: 0.0002 max mem: 52126 +Epoch: [25] [1820/3634] eta: 0:15:13 lr: 0.000004 loss: 0.5248 (0.5170) time: 0.5076 data: 0.0002 max mem: 52126 +Epoch: [25] [1830/3634] eta: 0:15:10 lr: 0.000004 loss: 0.5421 (0.5171) time: 0.6836 data: 0.0002 max mem: 52126 +Epoch: [25] [1840/3634] eta: 0:15:05 lr: 0.000004 loss: 0.5421 (0.5173) time: 0.6034 data: 0.0002 max mem: 52126 +Epoch: [25] [1850/3634] eta: 0:14:59 lr: 0.000004 loss: 0.4977 (0.5169) time: 0.4276 data: 0.0001 max mem: 52126 +Epoch: [25] [1860/3634] eta: 0:14:53 lr: 0.000004 loss: 0.4824 (0.5168) time: 0.4273 data: 0.0002 max mem: 52126 +Epoch: [25] [1870/3634] eta: 0:14:47 lr: 0.000004 loss: 0.4941 (0.5167) time: 0.4290 data: 0.0002 max mem: 52126 +Epoch: [25] [1880/3634] eta: 0:14:45 lr: 0.000004 loss: 0.5171 (0.5168) time: 0.5666 data: 0.0002 max mem: 52126 +Epoch: [25] [1890/3634] eta: 0:14:38 lr: 0.000004 loss: 0.5111 (0.5167) time: 0.5645 data: 0.0001 max mem: 52126 +Epoch: [25] [1900/3634] eta: 0:14:32 lr: 0.000004 loss: 0.4993 (0.5166) time: 0.3831 data: 0.0002 max mem: 52126 +Epoch: [25] [1910/3634] eta: 0:14:27 lr: 0.000004 loss: 0.4907 (0.5164) time: 0.4284 data: 0.0002 max mem: 52126 +Epoch: [25] [1920/3634] eta: 0:14:23 lr: 0.000004 loss: 0.4876 (0.5164) time: 0.5329 data: 0.0002 max mem: 52126 +Epoch: [25] [1930/3634] eta: 0:14:18 lr: 0.000004 loss: 0.5012 (0.5164) time: 0.5472 data: 0.0001 max mem: 52126 +Epoch: [25] [1940/3634] eta: 0:14:12 lr: 0.000004 loss: 0.5012 (0.5163) time: 0.4420 data: 0.0002 max mem: 52126 +Epoch: [25] [1950/3634] eta: 0:14:08 lr: 0.000004 loss: 0.5346 (0.5164) time: 0.5013 data: 0.0002 max mem: 52126 +Epoch: [25] [1960/3634] eta: 0:14:01 lr: 0.000004 loss: 0.5242 (0.5164) time: 0.5017 data: 0.0002 max mem: 52126 +Epoch: [25] [1970/3634] eta: 0:13:55 lr: 0.000004 loss: 0.4959 (0.5164) time: 0.3824 data: 0.0001 max mem: 52126 +Epoch: [25] [1980/3634] eta: 0:13:51 lr: 0.000004 loss: 0.4902 (0.5162) time: 0.5021 data: 0.0002 max mem: 52126 +Epoch: [25] [1990/3634] eta: 0:13:46 lr: 0.000004 loss: 0.4972 (0.5163) time: 0.5505 data: 0.0002 max mem: 52126 +Epoch: [25] [2000/3634] eta: 0:13:41 lr: 0.000004 loss: 0.5570 (0.5164) time: 0.4918 data: 0.0001 max mem: 52126 +Epoch: [25] [2010/3634] eta: 0:13:35 lr: 0.000004 loss: 0.5368 (0.5165) time: 0.4434 data: 0.0002 max mem: 52126 +Epoch: [25] [2020/3634] eta: 0:13:31 lr: 0.000004 loss: 0.5238 (0.5164) time: 0.4884 data: 0.0002 max mem: 52126 +Epoch: [25] [2030/3634] eta: 0:13:27 lr: 0.000004 loss: 0.5116 (0.5163) time: 0.5954 data: 0.0001 max mem: 52126 +Epoch: [25] [2040/3634] eta: 0:13:21 lr: 0.000004 loss: 0.5116 (0.5163) time: 0.5363 data: 0.0001 max mem: 52126 +Epoch: [25] [2050/3634] eta: 0:13:15 lr: 0.000004 loss: 0.5161 (0.5161) time: 0.4296 data: 0.0002 max mem: 52126 +Epoch: [25] [2060/3634] eta: 0:13:12 lr: 0.000004 loss: 0.4720 (0.5159) time: 0.5456 data: 0.0002 max mem: 52126 +Epoch: [25] [2070/3634] eta: 0:13:06 lr: 0.000004 loss: 0.4872 (0.5158) time: 0.5455 data: 0.0002 max mem: 52126 +Epoch: [25] [2080/3634] eta: 0:13:00 lr: 0.000004 loss: 0.4946 (0.5158) time: 0.3819 data: 0.0001 max mem: 52126 +Epoch: [25] [2090/3634] eta: 0:12:55 lr: 0.000004 loss: 0.5075 (0.5159) time: 0.4407 data: 0.0002 max mem: 52126 +Epoch: [25] [2100/3634] eta: 0:12:51 lr: 0.000004 loss: 0.5577 (0.5161) time: 0.5431 data: 0.0002 max mem: 52126 +Epoch: [25] [2110/3634] eta: 0:12:46 lr: 0.000004 loss: 0.5458 (0.5160) time: 0.5891 data: 0.0002 max mem: 52126 +Epoch: [25] [2120/3634] eta: 0:12:40 lr: 0.000004 loss: 0.5241 (0.5160) time: 0.4868 data: 0.0002 max mem: 52126 +Epoch: [25] [2130/3634] eta: 0:12:34 lr: 0.000004 loss: 0.5317 (0.5162) time: 0.3820 data: 0.0002 max mem: 52126 +Epoch: [25] [2140/3634] eta: 0:12:29 lr: 0.000004 loss: 0.5257 (0.5162) time: 0.3815 data: 0.0002 max mem: 52126 +Epoch: [25] [2150/3634] eta: 0:12:24 lr: 0.000004 loss: 0.5313 (0.5162) time: 0.4400 data: 0.0002 max mem: 52126 +Epoch: [25] [2160/3634] eta: 0:12:18 lr: 0.000004 loss: 0.5426 (0.5164) time: 0.4402 data: 0.0002 max mem: 52126 +Epoch: [25] [2170/3634] eta: 0:12:12 lr: 0.000004 loss: 0.5465 (0.5164) time: 0.3817 data: 0.0002 max mem: 52126 +Epoch: [25] [2180/3634] eta: 0:12:07 lr: 0.000004 loss: 0.5255 (0.5165) time: 0.4266 data: 0.0002 max mem: 52126 +Epoch: [25] [2190/3634] eta: 0:12:02 lr: 0.000004 loss: 0.5082 (0.5164) time: 0.4743 data: 0.0002 max mem: 52126 +Epoch: [25] [2200/3634] eta: 0:11:58 lr: 0.000004 loss: 0.5085 (0.5165) time: 0.5995 data: 0.0001 max mem: 52126 +Epoch: [25] [2210/3634] eta: 0:11:53 lr: 0.000004 loss: 0.5171 (0.5165) time: 0.5968 data: 0.0001 max mem: 52126 +Epoch: [25] [2220/3634] eta: 0:11:47 lr: 0.000004 loss: 0.5283 (0.5167) time: 0.4268 data: 0.0001 max mem: 52126 +Epoch: [25] [2230/3634] eta: 0:11:42 lr: 0.000004 loss: 0.5044 (0.5166) time: 0.4265 data: 0.0001 max mem: 52126 +Epoch: [25] [2240/3634] eta: 0:11:36 lr: 0.000004 loss: 0.4821 (0.5164) time: 0.4263 data: 0.0002 max mem: 52126 +Epoch: [25] [2250/3634] eta: 0:11:32 lr: 0.000004 loss: 0.4685 (0.5162) time: 0.5023 data: 0.0002 max mem: 52126 +Epoch: [25] [2260/3634] eta: 0:11:26 lr: 0.000004 loss: 0.4958 (0.5163) time: 0.5032 data: 0.0001 max mem: 52126 +Epoch: [25] [2270/3634] eta: 0:11:22 lr: 0.000004 loss: 0.5433 (0.5164) time: 0.4894 data: 0.0001 max mem: 52126 +Epoch: [25] [2280/3634] eta: 0:11:17 lr: 0.000004 loss: 0.5020 (0.5164) time: 0.5798 data: 0.0001 max mem: 52126 +Epoch: [25] [2290/3634] eta: 0:11:11 lr: 0.000004 loss: 0.5020 (0.5164) time: 0.4730 data: 0.0001 max mem: 52126 +Epoch: [25] [2300/3634] eta: 0:11:07 lr: 0.000004 loss: 0.5198 (0.5164) time: 0.5336 data: 0.0001 max mem: 52126 +Epoch: [25] [2310/3634] eta: 0:11:03 lr: 0.000004 loss: 0.5198 (0.5163) time: 0.5970 data: 0.0001 max mem: 52126 +Epoch: [25] [2320/3634] eta: 0:10:59 lr: 0.000004 loss: 0.4977 (0.5162) time: 0.6116 data: 0.0001 max mem: 52126 +Epoch: [25] [2330/3634] eta: 0:10:54 lr: 0.000004 loss: 0.4904 (0.5160) time: 0.6507 data: 0.0001 max mem: 52126 +Epoch: [25] [2340/3634] eta: 0:10:50 lr: 0.000004 loss: 0.4904 (0.5160) time: 0.5883 data: 0.0001 max mem: 52126 +Epoch: [25] [2350/3634] eta: 0:10:46 lr: 0.000004 loss: 0.4916 (0.5159) time: 0.6509 data: 0.0001 max mem: 52126 +Epoch: [25] [2360/3634] eta: 0:10:41 lr: 0.000004 loss: 0.5410 (0.5160) time: 0.6054 data: 0.0001 max mem: 52126 +Epoch: [25] [2370/3634] eta: 0:10:35 lr: 0.000004 loss: 0.5157 (0.5160) time: 0.4403 data: 0.0002 max mem: 52126 +Epoch: [25] [2380/3634] eta: 0:10:31 lr: 0.000004 loss: 0.5100 (0.5159) time: 0.5511 data: 0.0002 max mem: 52126 +Epoch: [25] [2390/3634] eta: 0:10:25 lr: 0.000004 loss: 0.4950 (0.5159) time: 0.5517 data: 0.0001 max mem: 52126 +Epoch: [25] [2400/3634] eta: 0:10:21 lr: 0.000004 loss: 0.5493 (0.5161) time: 0.4877 data: 0.0002 max mem: 52126 +Epoch: [25] [2410/3634] eta: 0:10:16 lr: 0.000004 loss: 0.5600 (0.5163) time: 0.5459 data: 0.0002 max mem: 52126 +Epoch: [25] [2420/3634] eta: 0:10:11 lr: 0.000004 loss: 0.5478 (0.5164) time: 0.5593 data: 0.0002 max mem: 52126 +Epoch: [25] [2430/3634] eta: 0:10:06 lr: 0.000004 loss: 0.5318 (0.5164) time: 0.5483 data: 0.0002 max mem: 52126 +Epoch: [25] [2440/3634] eta: 0:10:01 lr: 0.000004 loss: 0.5357 (0.5165) time: 0.4300 data: 0.0002 max mem: 52126 +Epoch: [25] [2450/3634] eta: 0:09:55 lr: 0.000004 loss: 0.5364 (0.5165) time: 0.3829 data: 0.0001 max mem: 52126 +Epoch: [25] [2460/3634] eta: 0:09:49 lr: 0.000004 loss: 0.5214 (0.5165) time: 0.3832 data: 0.0001 max mem: 52126 +Epoch: [25] [2470/3634] eta: 0:09:44 lr: 0.000004 loss: 0.5121 (0.5165) time: 0.4428 data: 0.0002 max mem: 52126 +Epoch: [25] [2480/3634] eta: 0:09:39 lr: 0.000004 loss: 0.5272 (0.5165) time: 0.4432 data: 0.0001 max mem: 52126 +Epoch: [25] [2490/3634] eta: 0:09:33 lr: 0.000004 loss: 0.4866 (0.5164) time: 0.3831 data: 0.0002 max mem: 52126 +Epoch: [25] [2500/3634] eta: 0:09:28 lr: 0.000004 loss: 0.4963 (0.5163) time: 0.4418 data: 0.0001 max mem: 52126 +Epoch: [25] [2510/3634] eta: 0:09:23 lr: 0.000004 loss: 0.5118 (0.5162) time: 0.5028 data: 0.0001 max mem: 52126 +Epoch: [25] [2520/3634] eta: 0:09:19 lr: 0.000004 loss: 0.5118 (0.5163) time: 0.6078 data: 0.0002 max mem: 52126 +Epoch: [25] [2530/3634] eta: 0:09:14 lr: 0.000004 loss: 0.5216 (0.5163) time: 0.6494 data: 0.0001 max mem: 52126 +Epoch: [25] [2540/3634] eta: 0:09:10 lr: 0.000004 loss: 0.5343 (0.5163) time: 0.6354 data: 0.0001 max mem: 52126 +Epoch: [25] [2550/3634] eta: 0:09:05 lr: 0.000004 loss: 0.5374 (0.5163) time: 0.5922 data: 0.0001 max mem: 52126 +Epoch: [25] [2560/3634] eta: 0:09:00 lr: 0.000004 loss: 0.5066 (0.5163) time: 0.5030 data: 0.0002 max mem: 52126 +Epoch: [25] [2570/3634] eta: 0:08:55 lr: 0.000004 loss: 0.5006 (0.5162) time: 0.4437 data: 0.0002 max mem: 52126 +Epoch: [25] [2580/3634] eta: 0:08:49 lr: 0.000004 loss: 0.5128 (0.5162) time: 0.4331 data: 0.0002 max mem: 52126 +Epoch: [25] [2590/3634] eta: 0:08:45 lr: 0.000004 loss: 0.5076 (0.5161) time: 0.5353 data: 0.0002 max mem: 52126 +Epoch: [25] [2600/3634] eta: 0:08:40 lr: 0.000004 loss: 0.5222 (0.5162) time: 0.5287 data: 0.0001 max mem: 52126 +Epoch: [25] [2610/3634] eta: 0:08:35 lr: 0.000004 loss: 0.5364 (0.5162) time: 0.4748 data: 0.0002 max mem: 52126 +Epoch: [25] [2620/3634] eta: 0:08:29 lr: 0.000004 loss: 0.5305 (0.5162) time: 0.4310 data: 0.0002 max mem: 52126 +Epoch: [25] [2630/3634] eta: 0:08:24 lr: 0.000004 loss: 0.5232 (0.5163) time: 0.3827 data: 0.0002 max mem: 52126 +Epoch: [25] [2640/3634] eta: 0:08:18 lr: 0.000004 loss: 0.5232 (0.5163) time: 0.3829 data: 0.0001 max mem: 52126 +Epoch: [25] [2650/3634] eta: 0:08:13 lr: 0.000004 loss: 0.5396 (0.5164) time: 0.4272 data: 0.0001 max mem: 52126 +Epoch: [25] [2660/3634] eta: 0:08:08 lr: 0.000004 loss: 0.5173 (0.5163) time: 0.5347 data: 0.0001 max mem: 52126 +Epoch: [25] [2670/3634] eta: 0:08:03 lr: 0.000004 loss: 0.5068 (0.5163) time: 0.5482 data: 0.0002 max mem: 52126 +Epoch: [25] [2680/3634] eta: 0:07:58 lr: 0.000004 loss: 0.5700 (0.5165) time: 0.4411 data: 0.0002 max mem: 52126 +Epoch: [25] [2690/3634] eta: 0:07:52 lr: 0.000004 loss: 0.5216 (0.5164) time: 0.3832 data: 0.0001 max mem: 52126 +Epoch: [25] [2700/3634] eta: 0:07:47 lr: 0.000004 loss: 0.4639 (0.5164) time: 0.3824 data: 0.0002 max mem: 52126 +Epoch: [25] [2710/3634] eta: 0:07:42 lr: 0.000004 loss: 0.5306 (0.5164) time: 0.3819 data: 0.0002 max mem: 52126 +Epoch: [25] [2720/3634] eta: 0:07:36 lr: 0.000004 loss: 0.5345 (0.5164) time: 0.4313 data: 0.0001 max mem: 52126 +Epoch: [25] [2730/3634] eta: 0:07:31 lr: 0.000004 loss: 0.4913 (0.5164) time: 0.4909 data: 0.0002 max mem: 52126 +Epoch: [25] [2740/3634] eta: 0:07:26 lr: 0.000004 loss: 0.5185 (0.5165) time: 0.5004 data: 0.0002 max mem: 52126 +Epoch: [25] [2750/3634] eta: 0:07:21 lr: 0.000004 loss: 0.5593 (0.5165) time: 0.4426 data: 0.0002 max mem: 52126 +Epoch: [25] [2760/3634] eta: 0:07:16 lr: 0.000004 loss: 0.5585 (0.5166) time: 0.4421 data: 0.0001 max mem: 52126 +Epoch: [25] [2770/3634] eta: 0:07:11 lr: 0.000004 loss: 0.5373 (0.5167) time: 0.4405 data: 0.0001 max mem: 52126 +Epoch: [25] [2780/3634] eta: 0:07:06 lr: 0.000004 loss: 0.5345 (0.5166) time: 0.4999 data: 0.0001 max mem: 52126 +Epoch: [25] [2790/3634] eta: 0:07:01 lr: 0.000004 loss: 0.5408 (0.5167) time: 0.5591 data: 0.0002 max mem: 52126 +Epoch: [25] [2800/3634] eta: 0:06:56 lr: 0.000004 loss: 0.5443 (0.5168) time: 0.5030 data: 0.0002 max mem: 52126 +Epoch: [25] [2810/3634] eta: 0:06:51 lr: 0.000004 loss: 0.5438 (0.5169) time: 0.5508 data: 0.0002 max mem: 52126 +Epoch: [25] [2820/3634] eta: 0:06:47 lr: 0.000004 loss: 0.5473 (0.5170) time: 0.5819 data: 0.0001 max mem: 52126 +Epoch: [25] [2830/3634] eta: 0:06:41 lr: 0.000004 loss: 0.5415 (0.5171) time: 0.4747 data: 0.0001 max mem: 52126 +Epoch: [25] [2840/3634] eta: 0:06:36 lr: 0.000004 loss: 0.4917 (0.5170) time: 0.3822 data: 0.0001 max mem: 52126 +Epoch: [25] [2850/3634] eta: 0:06:31 lr: 0.000004 loss: 0.5312 (0.5170) time: 0.4439 data: 0.0001 max mem: 52126 +Epoch: [25] [2860/3634] eta: 0:06:26 lr: 0.000004 loss: 0.5452 (0.5170) time: 0.4882 data: 0.0001 max mem: 52126 +Epoch: [25] [2870/3634] eta: 0:06:21 lr: 0.000004 loss: 0.5410 (0.5171) time: 0.4874 data: 0.0001 max mem: 52126 +Epoch: [25] [2880/3634] eta: 0:06:17 lr: 0.000004 loss: 0.5091 (0.5169) time: 0.6542 data: 0.0001 max mem: 52126 +Epoch: [25] [2890/3634] eta: 0:06:12 lr: 0.000004 loss: 0.5066 (0.5170) time: 0.7147 data: 0.0001 max mem: 52126 +Epoch: [25] [2900/3634] eta: 0:06:08 lr: 0.000004 loss: 0.5524 (0.5170) time: 0.6742 data: 0.0001 max mem: 52126 +Epoch: [25] [2910/3634] eta: 0:06:03 lr: 0.000004 loss: 0.5332 (0.5171) time: 0.5962 data: 0.0001 max mem: 52126 +Epoch: [25] [2920/3634] eta: 0:05:57 lr: 0.000004 loss: 0.5339 (0.5170) time: 0.4255 data: 0.0001 max mem: 52126 +Epoch: [25] [2930/3634] eta: 0:05:52 lr: 0.000004 loss: 0.5506 (0.5172) time: 0.4303 data: 0.0002 max mem: 52126 +Epoch: [25] [2940/3634] eta: 0:05:47 lr: 0.000004 loss: 0.5184 (0.5172) time: 0.4328 data: 0.0002 max mem: 52126 +Epoch: [25] [2950/3634] eta: 0:05:42 lr: 0.000004 loss: 0.5184 (0.5172) time: 0.4453 data: 0.0002 max mem: 52126 +Epoch: [25] [2960/3634] eta: 0:05:37 lr: 0.000004 loss: 0.4925 (0.5171) time: 0.5020 data: 0.0002 max mem: 52126 +Epoch: [25] [2970/3634] eta: 0:05:32 lr: 0.000004 loss: 0.5322 (0.5171) time: 0.4410 data: 0.0002 max mem: 52126 +Epoch: [25] [2980/3634] eta: 0:05:27 lr: 0.000004 loss: 0.5499 (0.5172) time: 0.4731 data: 0.0001 max mem: 52126 +Epoch: [25] [2990/3634] eta: 0:05:22 lr: 0.000004 loss: 0.5719 (0.5173) time: 0.5197 data: 0.0001 max mem: 52126 +Epoch: [25] [3000/3634] eta: 0:05:17 lr: 0.000004 loss: 0.5643 (0.5174) time: 0.4870 data: 0.0001 max mem: 52126 +Epoch: [25] [3010/3634] eta: 0:05:12 lr: 0.000004 loss: 0.5132 (0.5173) time: 0.5030 data: 0.0001 max mem: 52126 +Epoch: [25] [3020/3634] eta: 0:05:06 lr: 0.000004 loss: 0.5388 (0.5174) time: 0.4446 data: 0.0002 max mem: 52126 +Epoch: [25] [3030/3634] eta: 0:05:01 lr: 0.000004 loss: 0.5287 (0.5175) time: 0.3825 data: 0.0002 max mem: 52126 +Epoch: [25] [3040/3634] eta: 0:04:56 lr: 0.000004 loss: 0.5287 (0.5176) time: 0.5025 data: 0.0002 max mem: 52126 +Epoch: [25] [3050/3634] eta: 0:04:52 lr: 0.000004 loss: 0.4896 (0.5174) time: 0.6501 data: 0.0002 max mem: 52126 +Epoch: [25] [3060/3634] eta: 0:04:47 lr: 0.000004 loss: 0.4719 (0.5174) time: 0.6321 data: 0.0001 max mem: 52126 +Epoch: [25] [3070/3634] eta: 0:04:42 lr: 0.000004 loss: 0.5268 (0.5174) time: 0.4846 data: 0.0001 max mem: 52126 +Epoch: [25] [3080/3634] eta: 0:04:37 lr: 0.000004 loss: 0.5398 (0.5175) time: 0.4415 data: 0.0002 max mem: 52126 +Epoch: [25] [3090/3634] eta: 0:04:32 lr: 0.000004 loss: 0.5440 (0.5175) time: 0.5624 data: 0.0002 max mem: 52126 +Epoch: [25] [3100/3634] eta: 0:04:27 lr: 0.000004 loss: 0.5540 (0.5176) time: 0.5030 data: 0.0001 max mem: 52126 +Epoch: [25] [3110/3634] eta: 0:04:22 lr: 0.000004 loss: 0.5225 (0.5175) time: 0.6079 data: 0.0001 max mem: 52126 +Epoch: [25] [3120/3634] eta: 0:04:17 lr: 0.000004 loss: 0.4409 (0.5174) time: 0.6535 data: 0.0001 max mem: 52126 +Epoch: [25] [3130/3634] eta: 0:04:12 lr: 0.000004 loss: 0.4951 (0.5174) time: 0.4869 data: 0.0002 max mem: 52126 +Epoch: [25] [3140/3634] eta: 0:04:07 lr: 0.000004 loss: 0.4927 (0.5173) time: 0.5496 data: 0.0002 max mem: 52126 +Epoch: [25] [3150/3634] eta: 0:04:02 lr: 0.000004 loss: 0.5022 (0.5173) time: 0.4905 data: 0.0001 max mem: 52126 +Epoch: [25] [3160/3634] eta: 0:03:57 lr: 0.000004 loss: 0.5022 (0.5172) time: 0.4767 data: 0.0002 max mem: 52126 +Epoch: [25] [3170/3634] eta: 0:03:52 lr: 0.000004 loss: 0.4628 (0.5171) time: 0.6237 data: 0.0002 max mem: 52126 +Epoch: [25] [3180/3634] eta: 0:03:48 lr: 0.000004 loss: 0.5091 (0.5172) time: 0.6801 data: 0.0001 max mem: 52126 +Epoch: [25] [3190/3634] eta: 0:03:43 lr: 0.000004 loss: 0.5157 (0.5172) time: 0.5783 data: 0.0001 max mem: 52126 +Epoch: [25] [3200/3634] eta: 0:03:37 lr: 0.000004 loss: 0.5135 (0.5171) time: 0.4284 data: 0.0001 max mem: 52126 +Epoch: [25] [3210/3634] eta: 0:03:32 lr: 0.000004 loss: 0.5477 (0.5172) time: 0.3828 data: 0.0002 max mem: 52126 +Epoch: [25] [3220/3634] eta: 0:03:27 lr: 0.000004 loss: 0.5090 (0.5171) time: 0.4412 data: 0.0002 max mem: 52126 +Epoch: [25] [3230/3634] eta: 0:03:22 lr: 0.000004 loss: 0.5090 (0.5172) time: 0.4414 data: 0.0002 max mem: 52126 +Epoch: [25] [3240/3634] eta: 0:03:17 lr: 0.000004 loss: 0.5089 (0.5172) time: 0.3824 data: 0.0002 max mem: 52126 +Epoch: [25] [3250/3634] eta: 0:03:12 lr: 0.000004 loss: 0.5199 (0.5173) time: 0.3825 data: 0.0002 max mem: 52126 +Epoch: [25] [3260/3634] eta: 0:03:07 lr: 0.000004 loss: 0.5355 (0.5173) time: 0.3828 data: 0.0002 max mem: 52126 +Epoch: [25] [3270/3634] eta: 0:03:02 lr: 0.000004 loss: 0.4834 (0.5172) time: 0.5586 data: 0.0001 max mem: 52126 +Epoch: [25] [3280/3634] eta: 0:02:57 lr: 0.000004 loss: 0.4859 (0.5172) time: 0.6166 data: 0.0001 max mem: 52126 +Epoch: [25] [3290/3634] eta: 0:02:52 lr: 0.000004 loss: 0.5365 (0.5172) time: 0.4403 data: 0.0002 max mem: 52126 +Epoch: [25] [3300/3634] eta: 0:02:47 lr: 0.000004 loss: 0.4611 (0.5170) time: 0.3815 data: 0.0001 max mem: 52126 +Epoch: [25] [3310/3634] eta: 0:02:42 lr: 0.000004 loss: 0.4719 (0.5170) time: 0.4879 data: 0.0001 max mem: 52126 +Epoch: [25] [3320/3634] eta: 0:02:37 lr: 0.000004 loss: 0.4963 (0.5170) time: 0.5358 data: 0.0001 max mem: 52126 +Epoch: [25] [3330/3634] eta: 0:02:32 lr: 0.000004 loss: 0.4941 (0.5169) time: 0.4884 data: 0.0001 max mem: 52126 +Epoch: [25] [3340/3634] eta: 0:02:27 lr: 0.000004 loss: 0.4981 (0.5169) time: 0.5473 data: 0.0001 max mem: 52126 +Epoch: [25] [3350/3634] eta: 0:02:22 lr: 0.000004 loss: 0.5194 (0.5169) time: 0.5328 data: 0.0002 max mem: 52126 +Epoch: [25] [3360/3634] eta: 0:02:17 lr: 0.000004 loss: 0.5219 (0.5169) time: 0.4741 data: 0.0002 max mem: 52126 +Epoch: [25] [3370/3634] eta: 0:02:12 lr: 0.000004 loss: 0.5451 (0.5170) time: 0.5390 data: 0.0001 max mem: 52126 +Epoch: [25] [3380/3634] eta: 0:02:07 lr: 0.000004 loss: 0.5363 (0.5169) time: 0.4916 data: 0.0001 max mem: 52126 +Epoch: [25] [3390/3634] eta: 0:02:02 lr: 0.000004 loss: 0.4987 (0.5169) time: 0.4914 data: 0.0002 max mem: 52126 +Epoch: [25] [3400/3634] eta: 0:01:57 lr: 0.000004 loss: 0.5006 (0.5169) time: 0.5345 data: 0.0002 max mem: 52126 +Epoch: [25] [3410/3634] eta: 0:01:52 lr: 0.000004 loss: 0.5087 (0.5170) time: 0.4261 data: 0.0002 max mem: 52126 +Epoch: [25] [3420/3634] eta: 0:01:47 lr: 0.000004 loss: 0.5421 (0.5170) time: 0.3829 data: 0.0001 max mem: 52126 +Epoch: [25] [3430/3634] eta: 0:01:41 lr: 0.000004 loss: 0.5117 (0.5170) time: 0.3826 data: 0.0002 max mem: 52126 +Epoch: [25] [3440/3634] eta: 0:01:36 lr: 0.000004 loss: 0.5117 (0.5170) time: 0.3823 data: 0.0002 max mem: 52126 +Epoch: [25] [3450/3634] eta: 0:01:31 lr: 0.000004 loss: 0.5131 (0.5170) time: 0.3828 data: 0.0001 max mem: 52126 +Epoch: [25] [3460/3634] eta: 0:01:26 lr: 0.000004 loss: 0.5180 (0.5169) time: 0.4420 data: 0.0001 max mem: 52126 +Epoch: [25] [3470/3634] eta: 0:01:21 lr: 0.000004 loss: 0.5044 (0.5169) time: 0.5018 data: 0.0001 max mem: 52126 +Epoch: [25] [3480/3634] eta: 0:01:16 lr: 0.000004 loss: 0.5044 (0.5169) time: 0.5520 data: 0.0002 max mem: 52126 +Epoch: [25] [3490/3634] eta: 0:01:11 lr: 0.000004 loss: 0.5200 (0.5169) time: 0.4924 data: 0.0002 max mem: 52126 +Epoch: [25] [3500/3634] eta: 0:01:06 lr: 0.000004 loss: 0.5063 (0.5169) time: 0.5649 data: 0.0001 max mem: 52126 +Epoch: [25] [3510/3634] eta: 0:01:01 lr: 0.000004 loss: 0.5063 (0.5169) time: 0.6091 data: 0.0001 max mem: 52126 +Epoch: [25] [3520/3634] eta: 0:00:57 lr: 0.000004 loss: 0.5445 (0.5170) time: 0.6534 data: 0.0001 max mem: 52126 +Epoch: [25] [3530/3634] eta: 0:00:52 lr: 0.000004 loss: 0.5305 (0.5169) time: 0.6084 data: 0.0001 max mem: 52126 +Epoch: [25] [3540/3634] eta: 0:00:47 lr: 0.000004 loss: 0.5085 (0.5169) time: 0.5618 data: 0.0001 max mem: 52126 +Epoch: [25] [3550/3634] eta: 0:00:42 lr: 0.000004 loss: 0.5360 (0.5169) time: 0.6211 data: 0.0001 max mem: 52126 +Epoch: [25] [3560/3634] eta: 0:00:37 lr: 0.000004 loss: 0.5124 (0.5169) time: 0.4873 data: 0.0001 max mem: 52126 +Epoch: [25] [3570/3634] eta: 0:00:32 lr: 0.000004 loss: 0.5124 (0.5169) time: 0.4891 data: 0.0001 max mem: 52126 +Epoch: [25] [3580/3634] eta: 0:00:27 lr: 0.000004 loss: 0.5407 (0.5170) time: 0.4433 data: 0.0001 max mem: 52126 +Epoch: [25] [3590/3634] eta: 0:00:22 lr: 0.000004 loss: 0.5229 (0.5170) time: 0.4874 data: 0.0001 max mem: 52126 +Epoch: [25] [3600/3634] eta: 0:00:17 lr: 0.000004 loss: 0.5124 (0.5170) time: 0.5931 data: 0.0001 max mem: 52126 +Epoch: [25] [3610/3634] eta: 0:00:12 lr: 0.000004 loss: 0.5345 (0.5170) time: 0.5471 data: 0.0001 max mem: 52126 +Epoch: [25] [3620/3634] eta: 0:00:07 lr: 0.000004 loss: 0.5345 (0.5170) time: 0.4457 data: 0.0024 max mem: 52126 +Epoch: [25] [3630/3634] eta: 0:00:02 lr: 0.000004 loss: 0.5094 (0.5170) time: 0.5201 data: 0.0024 max mem: 52126 +Epoch: [25] [3633/3634] eta: 0:00:00 lr: 0.000004 loss: 0.5094 (0.5169) time: 0.5623 data: 0.0024 max mem: 52126 +Epoch: [25] Total time: 0:30:24 (0.5019 s / it) +Averaged stats: lr: 0.000004 loss: 0.5094 (0.5175) +Test: [ 0/1403] eta: 1:07:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0873 (0.0873) acc1: 100.0000 (100.0000) acc5: 100.0000 (100.0000) time: 2.8843 data: 2.7881 max mem: 52126 +Test: [ 10/1403] eta: 0:08:07 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1572 (0.1793) acc1: 96.8750 (95.0284) acc5: 100.0000 (100.0000) time: 0.3498 data: 0.2537 max mem: 52126 +Test: [ 20/1403] eta: 0:05:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1298 (0.1623) acc1: 96.8750 (95.5357) acc5: 100.0000 (100.0000) time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 30/1403] eta: 0:04:15 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0924 (0.1566) acc1: 98.4375 (95.6653) acc5: 100.0000 (100.0000) time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 40/1403] eta: 0:03:43 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1058 (0.1553) acc1: 100.0000 (96.0747) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 50/1403] eta: 0:03:24 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1137 (0.1706) acc1: 100.0000 (95.2512) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 60/1403] eta: 0:03:10 acc1_pos: 92.188 acc5_pos: 100.000 loss: 0.1194 (0.1680) acc1: 100.0000 (95.6199) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 70/1403] eta: 0:03:00 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.1426 (0.1707) acc1: 98.4375 (95.4005) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 80/1403] eta: 0:02:52 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1426 (0.1782) acc1: 96.8750 (95.0039) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 90/1403] eta: 0:02:46 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.1877 (0.1992) acc1: 93.7500 (93.9904) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 100/1403] eta: 0:02:41 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1833 (0.1966) acc1: 95.3125 (94.1832) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 110/1403] eta: 0:02:36 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1538 (0.1944) acc1: 98.4375 (94.3131) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 120/1403] eta: 0:02:32 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0770 (0.1846) acc1: 100.0000 (94.7701) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 130/1403] eta: 0:02:29 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0770 (0.1776) acc1: 100.0000 (95.0620) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 140/1403] eta: 0:02:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0725 (0.1700) acc1: 100.0000 (95.4122) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 150/1403] eta: 0:02:23 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0704 (0.1638) acc1: 100.0000 (95.6954) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 160/1403] eta: 0:02:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0741 (0.1587) acc1: 100.0000 (95.9239) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 170/1403] eta: 0:02:18 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0753 (0.1545) acc1: 100.0000 (96.1349) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 180/1403] eta: 0:02:16 acc1_pos: 89.062 acc5_pos: 100.000 loss: 0.0855 (0.1578) acc1: 100.0000 (96.0290) acc5: 100.0000 (100.0000) time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 190/1403] eta: 0:02:14 acc1_pos: 90.625 acc5_pos: 100.000 loss: 0.2685 (0.1654) acc1: 92.1875 (95.6479) acc5: 100.0000 (100.0000) time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 200/1403] eta: 0:02:12 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.2685 (0.1771) acc1: 90.6250 (95.0637) acc5: 100.0000 (100.0000) time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 210/1403] eta: 0:02:10 acc1_pos: 76.562 acc5_pos: 100.000 loss: 0.3193 (0.1876) acc1: 85.9375 (94.4831) acc5: 100.0000 (100.0000) time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 220/1403] eta: 0:02:08 acc1_pos: 87.500 acc5_pos: 100.000 loss: 0.3730 (0.1963) acc1: 85.9375 (94.0682) acc5: 100.0000 (100.0000) time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 230/1403] eta: 0:02:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2651 (0.1923) acc1: 90.6250 (94.2641) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 240/1403] eta: 0:02:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0811 (0.1878) acc1: 100.0000 (94.4891) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 250/1403] eta: 0:02:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0786 (0.1837) acc1: 100.0000 (94.6900) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 260/1403] eta: 0:02:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0787 (0.1800) acc1: 100.0000 (94.8755) acc5: 100.0000 (100.0000) time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 270/1403] eta: 0:02:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0789 (0.1765) acc1: 100.0000 (95.0357) acc5: 100.0000 (100.0000) time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 280/1403] eta: 0:01:59 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0797 (0.1730) acc1: 100.0000 (95.2069) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 290/1403] eta: 0:01:57 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0761 (0.1698) acc1: 100.0000 (95.3662) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 300/1403] eta: 0:01:56 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0751 (0.1668) acc1: 100.0000 (95.5150) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 310/1403] eta: 0:01:54 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0784 (0.1642) acc1: 100.0000 (95.6491) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 320/1403] eta: 0:01:53 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1027 (0.1624) acc1: 100.0000 (95.7847) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 330/1403] eta: 0:01:52 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1101 (0.1610) acc1: 100.0000 (95.9073) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 340/1403] eta: 0:01:50 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1101 (0.1594) acc1: 100.0000 (96.0273) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 350/1403] eta: 0:01:49 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1076 (0.1580) acc1: 100.0000 (96.1405) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 360/1403] eta: 0:01:48 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1109 (0.1568) acc1: 100.0000 (96.2387) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 370/1403] eta: 0:01:47 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1099 (0.1554) acc1: 100.0000 (96.3401) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 380/1403] eta: 0:01:45 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1078 (0.1542) acc1: 100.0000 (96.4362) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 390/1403] eta: 0:01:44 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1095 (0.1549) acc1: 100.0000 (96.3995) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [ 400/1403] eta: 0:01:43 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2435 (0.1584) acc1: 90.6250 (96.2087) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 410/1403] eta: 0:01:42 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2696 (0.1613) acc1: 89.0625 (96.0462) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 420/1403] eta: 0:01:41 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2464 (0.1629) acc1: 90.6250 (95.9583) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 430/1403] eta: 0:01:39 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2349 (0.1656) acc1: 90.6250 (95.8019) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 440/1403] eta: 0:01:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2878 (0.1690) acc1: 89.0625 (95.6349) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 450/1403] eta: 0:01:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2653 (0.1711) acc1: 89.0625 (95.5100) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 460/1403] eta: 0:01:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2556 (0.1735) acc1: 89.0625 (95.3667) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 470/1403] eta: 0:01:35 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2158 (0.1726) acc1: 92.1875 (95.3954) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 480/1403] eta: 0:01:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0645 (0.1704) acc1: 100.0000 (95.4912) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 490/1403] eta: 0:01:32 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0644 (0.1682) acc1: 100.0000 (95.5830) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 500/1403] eta: 0:01:31 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0649 (0.1662) acc1: 100.0000 (95.6712) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 510/1403] eta: 0:01:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0649 (0.1642) acc1: 100.0000 (95.7559) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 520/1403] eta: 0:01:29 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0700 (0.1625) acc1: 100.0000 (95.8373) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 530/1403] eta: 0:01:28 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0615 (0.1605) acc1: 100.0000 (95.9157) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 540/1403] eta: 0:01:27 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0606 (0.1587) acc1: 100.0000 (95.9912) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 550/1403] eta: 0:01:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0617 (0.1574) acc1: 100.0000 (96.0470) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 560/1403] eta: 0:01:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0850 (0.1563) acc1: 98.4375 (96.0951) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 570/1403] eta: 0:01:24 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0862 (0.1551) acc1: 98.4375 (96.1471) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 580/1403] eta: 0:01:23 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0903 (0.1544) acc1: 98.4375 (96.1731) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [ 590/1403] eta: 0:01:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1093 (0.1541) acc1: 96.8750 (96.1797) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52126 +Test: [ 600/1403] eta: 0:01:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1236 (0.1541) acc1: 95.3125 (96.1392) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [ 610/1403] eta: 0:01:19 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1179 (0.1535) acc1: 95.3125 (96.1538) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 620/1403] eta: 0:01:18 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1269 (0.1541) acc1: 95.3125 (96.0749) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 630/1403] eta: 0:01:17 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1453 (0.1539) acc1: 95.3125 (96.0727) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 640/1403] eta: 0:01:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1271 (0.1535) acc1: 96.8750 (96.0877) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 650/1403] eta: 0:01:15 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1271 (0.1531) acc1: 96.8750 (96.1070) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 660/1403] eta: 0:01:14 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1015 (0.1521) acc1: 100.0000 (96.1517) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 670/1403] eta: 0:01:13 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0722 (0.1509) acc1: 100.0000 (96.2067) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 680/1403] eta: 0:01:12 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0829 (0.1519) acc1: 98.4375 (96.1385) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 690/1403] eta: 0:01:11 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2371 (0.1557) acc1: 90.6250 (95.9230) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 700/1403] eta: 0:01:10 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1674 (0.1558) acc1: 92.1875 (95.9009) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 710/1403] eta: 0:01:09 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3264 (0.1604) acc1: 84.3750 (95.6619) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 720/1403] eta: 0:01:08 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3359 (0.1604) acc1: 84.3750 (95.6592) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 730/1403] eta: 0:01:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0701 (0.1593) acc1: 100.0000 (95.7101) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 740/1403] eta: 0:01:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0774 (0.1584) acc1: 100.0000 (95.7469) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 750/1403] eta: 0:01:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1051 (0.1595) acc1: 96.8750 (95.6849) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 760/1403] eta: 0:01:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3017 (0.1625) acc1: 87.5000 (95.4932) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 770/1403] eta: 0:01:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4010 (0.1664) acc1: 79.6875 (95.2578) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 780/1403] eta: 0:01:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3773 (0.1688) acc1: 81.2500 (95.1204) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 790/1403] eta: 0:01:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3051 (0.1713) acc1: 87.5000 (94.9945) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 800/1403] eta: 0:01:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2198 (0.1717) acc1: 90.6250 (94.9750) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [ 810/1403] eta: 0:00:59 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2198 (0.1749) acc1: 90.6250 (94.8289) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 820/1403] eta: 0:00:58 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6376 (0.1839) acc1: 71.8750 (94.4389) acc5: 100.0000 (100.0000) acc1_neg: 40.625 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 830/1403] eta: 0:00:57 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.7224 (0.1869) acc1: 70.3125 (94.2878) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [ 840/1403] eta: 0:00:56 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5381 (0.1953) acc1: 70.3125 (93.9302) acc5: 100.0000 (100.0000) acc1_neg: 48.438 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [ 850/1403] eta: 0:00:55 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.0766 (0.2066) acc1: 54.6875 (93.4507) acc5: 100.0000 (100.0000) acc1_neg: 50.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 860/1403] eta: 0:00:53 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.0375 (0.2102) acc1: 56.2500 (93.2945) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 870/1403] eta: 0:00:52 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1324 (0.2091) acc1: 96.8750 (93.3428) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [ 880/1403] eta: 0:00:51 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1009 (0.2079) acc1: 98.4375 (93.3988) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [ 890/1403] eta: 0:00:50 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0840 (0.2066) acc1: 100.0000 (93.4642) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 900/1403] eta: 0:00:49 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0752 (0.2051) acc1: 100.0000 (93.5332) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 910/1403] eta: 0:00:48 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0765 (0.2041) acc1: 100.0000 (93.5751) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [ 920/1403] eta: 0:00:47 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1008 (0.2030) acc1: 98.4375 (93.6177) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 930/1403] eta: 0:00:46 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0942 (0.2018) acc1: 98.4375 (93.6711) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 940/1403] eta: 0:00:45 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1083 (0.2017) acc1: 96.8750 (93.6620) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 950/1403] eta: 0:00:44 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2019 (0.2023) acc1: 89.0625 (93.6169) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 960/1403] eta: 0:00:43 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2352 (0.2026) acc1: 89.0625 (93.5858) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 970/1403] eta: 0:00:42 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2133 (0.2031) acc1: 92.1875 (93.5505) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [ 980/1403] eta: 0:00:41 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2219 (0.2036) acc1: 92.1875 (93.5175) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 990/1403] eta: 0:00:40 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2194 (0.2038) acc1: 92.1875 (93.5025) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [1000/1403] eta: 0:00:39 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2124 (0.2042) acc1: 92.1875 (93.4815) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [1010/1403] eta: 0:00:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2247 (0.2048) acc1: 90.6250 (93.4440) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [1020/1403] eta: 0:00:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3946 (0.2104) acc1: 81.2500 (93.1287) acc5: 100.0000 (100.0000) acc1_neg: 37.500 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [1030/1403] eta: 0:00:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4910 (0.2121) acc1: 64.0625 (93.0180) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [1040/1403] eta: 0:00:35 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2167 (0.2119) acc1: 90.6250 (93.0115) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [1050/1403] eta: 0:00:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1735 (0.2117) acc1: 93.7500 (93.0275) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [1060/1403] eta: 0:00:33 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1506 (0.2110) acc1: 95.3125 (93.0637) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52126 +Test: [1070/1403] eta: 0:00:32 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1506 (0.2115) acc1: 96.8750 (93.0453) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52126 +Test: [1080/1403] eta: 0:00:31 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3560 (0.2133) acc1: 85.9375 (92.9536) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [1090/1403] eta: 0:00:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4163 (0.2152) acc1: 82.8125 (92.8592) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [1100/1403] eta: 0:00:29 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3287 (0.2156) acc1: 85.9375 (92.8318) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [1110/1403] eta: 0:00:28 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2687 (0.2161) acc1: 90.6250 (92.8091) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [1120/1403] eta: 0:00:27 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2708 (0.2166) acc1: 89.0625 (92.7855) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [1130/1403] eta: 0:00:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2753 (0.2172) acc1: 89.0625 (92.7567) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [1140/1403] eta: 0:00:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2267 (0.2172) acc1: 92.1875 (92.7490) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [1150/1403] eta: 0:00:24 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2224 (0.2176) acc1: 92.1875 (92.7291) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [1160/1403] eta: 0:00:23 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2741 (0.2180) acc1: 89.0625 (92.7110) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [1170/1403] eta: 0:00:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2507 (0.2179) acc1: 90.6250 (92.7132) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [1180/1403] eta: 0:00:21 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0713 (0.2166) acc1: 100.0000 (92.7749) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [1190/1403] eta: 0:00:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0700 (0.2154) acc1: 100.0000 (92.8356) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [1200/1403] eta: 0:00:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0703 (0.2142) acc1: 100.0000 (92.8952) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0002 max mem: 52126 +Test: [1210/1403] eta: 0:00:19 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0695 (0.2130) acc1: 100.0000 (92.9539) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0002 max mem: 52126 +Test: [1220/1403] eta: 0:00:18 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0693 (0.2119) acc1: 100.0000 (93.0116) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0002 max mem: 52126 +Test: [1230/1403] eta: 0:00:17 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0703 (0.2107) acc1: 100.0000 (93.0684) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0002 max mem: 52126 +Test: [1240/1403] eta: 0:00:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0700 (0.2096) acc1: 100.0000 (93.1242) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0002 max mem: 52126 +Test: [1250/1403] eta: 0:00:15 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0703 (0.2085) acc1: 100.0000 (93.1792) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [1260/1403] eta: 0:00:14 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0763 (0.2075) acc1: 100.0000 (93.2333) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [1270/1403] eta: 0:00:13 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0775 (0.2064) acc1: 100.0000 (93.2865) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [1280/1403] eta: 0:00:12 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0771 (0.2054) acc1: 100.0000 (93.3389) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0003 max mem: 52126 +Test: [1290/1403] eta: 0:00:11 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0767 (0.2044) acc1: 100.0000 (93.3905) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0003 max mem: 52126 +Test: [1300/1403] eta: 0:00:10 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0777 (0.2035) acc1: 100.0000 (93.4413) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0002 max mem: 52126 +Test: [1310/1403] eta: 0:00:09 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0782 (0.2025) acc1: 100.0000 (93.4914) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0002 max mem: 52126 +Test: [1320/1403] eta: 0:00:08 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0760 (0.2016) acc1: 100.0000 (93.5406) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0002 max mem: 52126 +Test: [1330/1403] eta: 0:00:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0749 (0.2006) acc1: 100.0000 (93.5880) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0002 max mem: 52126 +Test: [1340/1403] eta: 0:00:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0673 (0.1996) acc1: 100.0000 (93.6358) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0002 max mem: 52126 +Test: [1350/1403] eta: 0:00:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0661 (0.1986) acc1: 100.0000 (93.6806) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0002 max mem: 52126 +Test: [1360/1403] eta: 0:00:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0660 (0.1976) acc1: 100.0000 (93.7270) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0002 max mem: 52126 +Test: [1370/1403] eta: 0:00:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0660 (0.1967) acc1: 100.0000 (93.7705) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0002 max mem: 52126 +Test: [1380/1403] eta: 0:00:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0654 (0.1958) acc1: 100.0000 (93.8156) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0002 max mem: 52126 +Test: [1390/1403] eta: 0:00:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0652 (0.1949) acc1: 100.0000 (93.8601) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0002 max mem: 52126 +Test: [1400/1403] eta: 0:00:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0660 (0.1940) acc1: 100.0000 (93.9017) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0960 data: 0.0001 max mem: 52126 +Test: [1402/1403] eta: 0:00:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0661 (0.1938) acc1: 100.0000 (93.9061) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0919 data: 0.0001 max mem: 52126 +Test: Total time: 0:02:18 (0.0986 s / it) +* Overall Acc@1 93.906 Acc@5 100.000 + target == real Acc@1 95.611 Acc@5 100.000 + target == fake Acc@1 93.426 Acc@5 100.000 + loss 0.194 +Accuracy of the network on the 89729 test images: 93.9% +Max accuracy: 93.91% +Epoch: [26] [ 0/3634] eta: 4:43:47 lr: 0.000004 loss: 0.5068 (0.5068) time: 4.6856 data: 4.3117 max mem: 52126 +Epoch: [26] [ 10/3634] eta: 0:46:44 lr: 0.000004 loss: 0.5355 (0.5095) time: 0.7740 data: 0.3921 max mem: 52126 +Epoch: [26] [ 20/3634] eta: 0:40:43 lr: 0.000004 loss: 0.5210 (0.5017) time: 0.4758 data: 0.0001 max mem: 52126 +Epoch: [26] [ 30/3634] eta: 0:38:42 lr: 0.000004 loss: 0.4904 (0.4988) time: 0.5730 data: 0.0001 max mem: 52126 +Epoch: [26] [ 40/3634] eta: 0:34:45 lr: 0.000004 loss: 0.4924 (0.4987) time: 0.4794 data: 0.0002 max mem: 52126 +Epoch: [26] [ 50/3634] eta: 0:34:49 lr: 0.000004 loss: 0.4983 (0.5053) time: 0.4880 data: 0.0002 max mem: 52126 +Epoch: [26] [ 60/3634] eta: 0:33:58 lr: 0.000004 loss: 0.5049 (0.5051) time: 0.5501 data: 0.0002 max mem: 52126 +Epoch: [26] [ 70/3634] eta: 0:32:20 lr: 0.000004 loss: 0.5208 (0.5094) time: 0.4459 data: 0.0002 max mem: 52126 +Epoch: [26] [ 80/3634] eta: 0:31:03 lr: 0.000004 loss: 0.5263 (0.5114) time: 0.3837 data: 0.0002 max mem: 52126 +Epoch: [26] [ 90/3634] eta: 0:30:50 lr: 0.000004 loss: 0.5327 (0.5096) time: 0.4431 data: 0.0002 max mem: 52126 +Epoch: [26] [ 100/3634] eta: 0:30:29 lr: 0.000004 loss: 0.5363 (0.5116) time: 0.4915 data: 0.0002 max mem: 52126 +Epoch: [26] [ 110/3634] eta: 0:30:44 lr: 0.000004 loss: 0.5386 (0.5140) time: 0.5287 data: 0.0001 max mem: 52126 +Epoch: [26] [ 120/3634] eta: 0:29:57 lr: 0.000004 loss: 0.5386 (0.5144) time: 0.4805 data: 0.0001 max mem: 52126 +Epoch: [26] [ 130/3634] eta: 0:29:17 lr: 0.000004 loss: 0.5276 (0.5156) time: 0.3818 data: 0.0002 max mem: 52126 +Epoch: [26] [ 140/3634] eta: 0:29:29 lr: 0.000004 loss: 0.5185 (0.5145) time: 0.4754 data: 0.0002 max mem: 52126 +Epoch: [26] [ 150/3634] eta: 0:29:18 lr: 0.000004 loss: 0.4965 (0.5147) time: 0.5249 data: 0.0002 max mem: 52126 +Epoch: [26] [ 160/3634] eta: 0:28:46 lr: 0.000004 loss: 0.5433 (0.5147) time: 0.4308 data: 0.0002 max mem: 52126 +Epoch: [26] [ 170/3634] eta: 0:29:08 lr: 0.000004 loss: 0.5238 (0.5153) time: 0.5043 data: 0.0002 max mem: 52126 +Epoch: [26] [ 180/3634] eta: 0:28:39 lr: 0.000004 loss: 0.5221 (0.5142) time: 0.5047 data: 0.0002 max mem: 52126 +Epoch: [26] [ 190/3634] eta: 0:28:31 lr: 0.000004 loss: 0.5491 (0.5164) time: 0.4308 data: 0.0002 max mem: 52126 +Epoch: [26] [ 200/3634] eta: 0:28:28 lr: 0.000004 loss: 0.5372 (0.5162) time: 0.4930 data: 0.0002 max mem: 52126 +Epoch: [26] [ 210/3634] eta: 0:28:24 lr: 0.000004 loss: 0.5320 (0.5156) time: 0.5058 data: 0.0002 max mem: 52126 +Epoch: [26] [ 220/3634] eta: 0:28:02 lr: 0.000004 loss: 0.5115 (0.5150) time: 0.4462 data: 0.0002 max mem: 52126 +Epoch: [26] [ 230/3634] eta: 0:28:12 lr: 0.000004 loss: 0.5211 (0.5158) time: 0.4922 data: 0.0001 max mem: 52126 +Epoch: [26] [ 240/3634] eta: 0:27:51 lr: 0.000004 loss: 0.5386 (0.5162) time: 0.4902 data: 0.0001 max mem: 52126 +Epoch: [26] [ 250/3634] eta: 0:27:31 lr: 0.000004 loss: 0.5349 (0.5172) time: 0.3826 data: 0.0002 max mem: 52126 +Epoch: [26] [ 260/3634] eta: 0:27:13 lr: 0.000004 loss: 0.5370 (0.5180) time: 0.3814 data: 0.0002 max mem: 52126 +Epoch: [26] [ 270/3634] eta: 0:27:07 lr: 0.000004 loss: 0.5017 (0.5173) time: 0.4292 data: 0.0002 max mem: 52126 +Epoch: [26] [ 280/3634] eta: 0:27:05 lr: 0.000004 loss: 0.5000 (0.5172) time: 0.4913 data: 0.0002 max mem: 52126 +Epoch: [26] [ 290/3634] eta: 0:26:48 lr: 0.000004 loss: 0.5088 (0.5166) time: 0.4430 data: 0.0002 max mem: 52126 +Epoch: [26] [ 300/3634] eta: 0:26:43 lr: 0.000004 loss: 0.5201 (0.5173) time: 0.4284 data: 0.0002 max mem: 52126 +Epoch: [26] [ 310/3634] eta: 0:26:54 lr: 0.000004 loss: 0.5436 (0.5169) time: 0.5519 data: 0.0002 max mem: 52126 +Epoch: [26] [ 320/3634] eta: 0:26:38 lr: 0.000004 loss: 0.5017 (0.5166) time: 0.5050 data: 0.0002 max mem: 52126 +Epoch: [26] [ 330/3634] eta: 0:26:35 lr: 0.000004 loss: 0.5311 (0.5179) time: 0.4431 data: 0.0001 max mem: 52126 +Epoch: [26] [ 340/3634] eta: 0:26:41 lr: 0.000004 loss: 0.5361 (0.5169) time: 0.5503 data: 0.0001 max mem: 52126 +Epoch: [26] [ 350/3634] eta: 0:26:48 lr: 0.000004 loss: 0.5116 (0.5167) time: 0.6013 data: 0.0001 max mem: 52126 +Epoch: [26] [ 360/3634] eta: 0:26:55 lr: 0.000004 loss: 0.5081 (0.5167) time: 0.6162 data: 0.0001 max mem: 52126 +Epoch: [26] [ 370/3634] eta: 0:27:20 lr: 0.000004 loss: 0.5331 (0.5167) time: 0.7260 data: 0.0001 max mem: 52126 +Epoch: [26] [ 380/3634] eta: 0:27:22 lr: 0.000004 loss: 0.5331 (0.5158) time: 0.7103 data: 0.0001 max mem: 52126 +Epoch: [26] [ 390/3634] eta: 0:27:07 lr: 0.000004 loss: 0.5358 (0.5164) time: 0.4880 data: 0.0001 max mem: 52126 +Epoch: [26] [ 400/3634] eta: 0:27:12 lr: 0.000004 loss: 0.5479 (0.5168) time: 0.5051 data: 0.0002 max mem: 52126 +Epoch: [26] [ 410/3634] eta: 0:27:05 lr: 0.000004 loss: 0.5327 (0.5168) time: 0.5550 data: 0.0002 max mem: 52126 +Epoch: [26] [ 420/3634] eta: 0:27:00 lr: 0.000004 loss: 0.5250 (0.5169) time: 0.4918 data: 0.0002 max mem: 52126 +Epoch: [26] [ 430/3634] eta: 0:26:53 lr: 0.000004 loss: 0.5192 (0.5170) time: 0.4894 data: 0.0001 max mem: 52126 +Epoch: [26] [ 440/3634] eta: 0:26:48 lr: 0.000004 loss: 0.5732 (0.5183) time: 0.4902 data: 0.0001 max mem: 52126 +Epoch: [26] [ 450/3634] eta: 0:27:01 lr: 0.000004 loss: 0.5700 (0.5186) time: 0.6282 data: 0.0001 max mem: 52126 +Epoch: [26] [ 460/3634] eta: 0:26:47 lr: 0.000004 loss: 0.5078 (0.5180) time: 0.5670 data: 0.0001 max mem: 52126 +Epoch: [26] [ 470/3634] eta: 0:26:40 lr: 0.000004 loss: 0.5114 (0.5185) time: 0.4281 data: 0.0001 max mem: 52126 +Epoch: [26] [ 480/3634] eta: 0:26:33 lr: 0.000004 loss: 0.5403 (0.5187) time: 0.4750 data: 0.0002 max mem: 52126 +Epoch: [26] [ 490/3634] eta: 0:26:32 lr: 0.000004 loss: 0.5205 (0.5184) time: 0.5243 data: 0.0002 max mem: 52126 +Epoch: [26] [ 500/3634] eta: 0:26:19 lr: 0.000004 loss: 0.5143 (0.5179) time: 0.4782 data: 0.0001 max mem: 52126 +Epoch: [26] [ 510/3634] eta: 0:26:07 lr: 0.000004 loss: 0.5070 (0.5179) time: 0.3835 data: 0.0002 max mem: 52126 +Epoch: [26] [ 520/3634] eta: 0:26:00 lr: 0.000004 loss: 0.5460 (0.5184) time: 0.4294 data: 0.0002 max mem: 52126 +Epoch: [26] [ 530/3634] eta: 0:25:59 lr: 0.000004 loss: 0.5409 (0.5186) time: 0.5208 data: 0.0002 max mem: 52126 +Epoch: [26] [ 540/3634] eta: 0:26:05 lr: 0.000004 loss: 0.5007 (0.5183) time: 0.6281 data: 0.0002 max mem: 52126 +Epoch: [26] [ 550/3634] eta: 0:26:03 lr: 0.000004 loss: 0.5130 (0.5185) time: 0.6293 data: 0.0002 max mem: 52126 +Epoch: [26] [ 560/3634] eta: 0:25:58 lr: 0.000004 loss: 0.5394 (0.5189) time: 0.5355 data: 0.0002 max mem: 52126 +Epoch: [26] [ 570/3634] eta: 0:25:52 lr: 0.000004 loss: 0.5394 (0.5191) time: 0.5016 data: 0.0001 max mem: 52126 +Epoch: [26] [ 580/3634] eta: 0:25:46 lr: 0.000004 loss: 0.5337 (0.5193) time: 0.4872 data: 0.0001 max mem: 52126 +Epoch: [26] [ 590/3634] eta: 0:25:48 lr: 0.000004 loss: 0.5517 (0.5200) time: 0.5624 data: 0.0001 max mem: 52126 +Epoch: [26] [ 600/3634] eta: 0:25:36 lr: 0.000004 loss: 0.5594 (0.5206) time: 0.5168 data: 0.0001 max mem: 52126 +Epoch: [26] [ 610/3634] eta: 0:25:31 lr: 0.000004 loss: 0.5403 (0.5210) time: 0.4407 data: 0.0001 max mem: 52126 +Epoch: [26] [ 620/3634] eta: 0:25:25 lr: 0.000004 loss: 0.5227 (0.5203) time: 0.4895 data: 0.0001 max mem: 52126 +Epoch: [26] [ 630/3634] eta: 0:25:14 lr: 0.000004 loss: 0.4892 (0.5200) time: 0.4317 data: 0.0001 max mem: 52126 +Epoch: [26] [ 640/3634] eta: 0:25:07 lr: 0.000004 loss: 0.5440 (0.5203) time: 0.4297 data: 0.0001 max mem: 52126 +Epoch: [26] [ 650/3634] eta: 0:25:08 lr: 0.000004 loss: 0.5479 (0.5206) time: 0.5481 data: 0.0001 max mem: 52126 +Epoch: [26] [ 660/3634] eta: 0:25:06 lr: 0.000004 loss: 0.4777 (0.5197) time: 0.5949 data: 0.0002 max mem: 52126 +Epoch: [26] [ 670/3634] eta: 0:24:55 lr: 0.000004 loss: 0.4860 (0.5198) time: 0.4762 data: 0.0002 max mem: 52126 +Epoch: [26] [ 680/3634] eta: 0:24:45 lr: 0.000004 loss: 0.5055 (0.5188) time: 0.3831 data: 0.0002 max mem: 52126 +Epoch: [26] [ 690/3634] eta: 0:24:40 lr: 0.000004 loss: 0.4890 (0.5189) time: 0.4445 data: 0.0002 max mem: 52126 +Epoch: [26] [ 700/3634] eta: 0:24:30 lr: 0.000004 loss: 0.5208 (0.5189) time: 0.4441 data: 0.0002 max mem: 52126 +Epoch: [26] [ 710/3634] eta: 0:24:20 lr: 0.000004 loss: 0.5202 (0.5187) time: 0.3819 data: 0.0002 max mem: 52126 +Epoch: [26] [ 720/3634] eta: 0:24:10 lr: 0.000004 loss: 0.5340 (0.5187) time: 0.3815 data: 0.0001 max mem: 52126 +Epoch: [26] [ 730/3634] eta: 0:24:00 lr: 0.000004 loss: 0.5163 (0.5185) time: 0.3813 data: 0.0001 max mem: 52126 +Epoch: [26] [ 740/3634] eta: 0:23:51 lr: 0.000004 loss: 0.5144 (0.5188) time: 0.3812 data: 0.0001 max mem: 52126 +Epoch: [26] [ 750/3634] eta: 0:23:46 lr: 0.000004 loss: 0.5550 (0.5189) time: 0.4421 data: 0.0001 max mem: 52126 +Epoch: [26] [ 760/3634] eta: 0:23:42 lr: 0.000004 loss: 0.5383 (0.5190) time: 0.5011 data: 0.0002 max mem: 52126 +Epoch: [26] [ 770/3634] eta: 0:23:36 lr: 0.000004 loss: 0.4911 (0.5187) time: 0.4861 data: 0.0002 max mem: 52126 +Epoch: [26] [ 780/3634] eta: 0:23:39 lr: 0.000004 loss: 0.5114 (0.5187) time: 0.5899 data: 0.0002 max mem: 52126 +Epoch: [26] [ 790/3634] eta: 0:23:30 lr: 0.000004 loss: 0.5040 (0.5185) time: 0.5451 data: 0.0001 max mem: 52126 +Epoch: [26] [ 800/3634] eta: 0:23:24 lr: 0.000004 loss: 0.5182 (0.5188) time: 0.4275 data: 0.0002 max mem: 52126 +Epoch: [26] [ 810/3634] eta: 0:23:23 lr: 0.000004 loss: 0.5366 (0.5188) time: 0.5449 data: 0.0002 max mem: 52126 +Epoch: [26] [ 820/3634] eta: 0:23:20 lr: 0.000004 loss: 0.5185 (0.5182) time: 0.5900 data: 0.0002 max mem: 52126 +Epoch: [26] [ 830/3634] eta: 0:23:16 lr: 0.000004 loss: 0.4842 (0.5179) time: 0.5325 data: 0.0001 max mem: 52126 +Epoch: [26] [ 840/3634] eta: 0:23:07 lr: 0.000004 loss: 0.4982 (0.5180) time: 0.4423 data: 0.0002 max mem: 52126 +Epoch: [26] [ 850/3634] eta: 0:23:06 lr: 0.000004 loss: 0.5175 (0.5177) time: 0.5000 data: 0.0002 max mem: 52126 +Epoch: [26] [ 860/3634] eta: 0:23:01 lr: 0.000004 loss: 0.5140 (0.5178) time: 0.5580 data: 0.0002 max mem: 52126 +Epoch: [26] [ 870/3634] eta: 0:22:55 lr: 0.000004 loss: 0.5197 (0.5177) time: 0.4851 data: 0.0002 max mem: 52126 +Epoch: [26] [ 880/3634] eta: 0:22:49 lr: 0.000004 loss: 0.5298 (0.5179) time: 0.4740 data: 0.0002 max mem: 52126 +Epoch: [26] [ 890/3634] eta: 0:22:44 lr: 0.000004 loss: 0.5298 (0.5178) time: 0.4748 data: 0.0002 max mem: 52126 +Epoch: [26] [ 900/3634] eta: 0:22:39 lr: 0.000004 loss: 0.5363 (0.5179) time: 0.4888 data: 0.0002 max mem: 52126 +Epoch: [26] [ 910/3634] eta: 0:22:30 lr: 0.000004 loss: 0.5393 (0.5181) time: 0.4430 data: 0.0002 max mem: 52126 +Epoch: [26] [ 920/3634] eta: 0:22:31 lr: 0.000004 loss: 0.5354 (0.5181) time: 0.5298 data: 0.0002 max mem: 52126 +Epoch: [26] [ 930/3634] eta: 0:22:22 lr: 0.000004 loss: 0.4840 (0.5177) time: 0.5288 data: 0.0002 max mem: 52126 +Epoch: [26] [ 940/3634] eta: 0:22:21 lr: 0.000004 loss: 0.5166 (0.5177) time: 0.4990 data: 0.0002 max mem: 52126 +Epoch: [26] [ 950/3634] eta: 0:22:13 lr: 0.000004 loss: 0.5341 (0.5180) time: 0.5003 data: 0.0002 max mem: 52126 +Epoch: [26] [ 960/3634] eta: 0:22:07 lr: 0.000004 loss: 0.5380 (0.5176) time: 0.4332 data: 0.0002 max mem: 52126 +Epoch: [26] [ 970/3634] eta: 0:22:09 lr: 0.000004 loss: 0.4618 (0.5174) time: 0.6147 data: 0.0002 max mem: 52126 +Epoch: [26] [ 980/3634] eta: 0:22:01 lr: 0.000004 loss: 0.5193 (0.5175) time: 0.5639 data: 0.0002 max mem: 52126 +Epoch: [26] [ 990/3634] eta: 0:22:02 lr: 0.000004 loss: 0.5060 (0.5173) time: 0.5447 data: 0.0001 max mem: 52126 +Epoch: [26] [1000/3634] eta: 0:21:54 lr: 0.000004 loss: 0.5060 (0.5177) time: 0.5450 data: 0.0001 max mem: 52126 +Epoch: [26] [1010/3634] eta: 0:21:48 lr: 0.000004 loss: 0.5447 (0.5179) time: 0.4269 data: 0.0001 max mem: 52126 +Epoch: [26] [1020/3634] eta: 0:21:40 lr: 0.000004 loss: 0.5277 (0.5179) time: 0.4270 data: 0.0001 max mem: 52126 +Epoch: [26] [1030/3634] eta: 0:21:38 lr: 0.000004 loss: 0.5253 (0.5179) time: 0.4997 data: 0.0001 max mem: 52126 +Epoch: [26] [1040/3634] eta: 0:21:32 lr: 0.000004 loss: 0.5214 (0.5178) time: 0.5440 data: 0.0001 max mem: 52126 +Epoch: [26] [1050/3634] eta: 0:21:30 lr: 0.000004 loss: 0.5036 (0.5178) time: 0.5315 data: 0.0001 max mem: 52126 +Epoch: [26] [1060/3634] eta: 0:21:24 lr: 0.000004 loss: 0.5036 (0.5177) time: 0.5317 data: 0.0001 max mem: 52126 +Epoch: [26] [1070/3634] eta: 0:21:20 lr: 0.000004 loss: 0.5209 (0.5178) time: 0.5150 data: 0.0002 max mem: 52126 +Epoch: [26] [1080/3634] eta: 0:21:13 lr: 0.000004 loss: 0.5123 (0.5178) time: 0.4703 data: 0.0002 max mem: 52126 +Epoch: [26] [1090/3634] eta: 0:21:08 lr: 0.000004 loss: 0.4991 (0.5174) time: 0.4416 data: 0.0001 max mem: 52126 +Epoch: [26] [1100/3634] eta: 0:21:02 lr: 0.000004 loss: 0.5283 (0.5177) time: 0.4865 data: 0.0002 max mem: 52126 +Epoch: [26] [1110/3634] eta: 0:20:59 lr: 0.000004 loss: 0.5516 (0.5178) time: 0.5300 data: 0.0002 max mem: 52126 +Epoch: [26] [1120/3634] eta: 0:20:53 lr: 0.000004 loss: 0.4889 (0.5175) time: 0.5286 data: 0.0002 max mem: 52126 +Epoch: [26] [1130/3634] eta: 0:20:46 lr: 0.000004 loss: 0.4924 (0.5175) time: 0.4260 data: 0.0002 max mem: 52126 +Epoch: [26] [1140/3634] eta: 0:20:41 lr: 0.000004 loss: 0.5189 (0.5176) time: 0.4467 data: 0.0002 max mem: 52126 +Epoch: [26] [1150/3634] eta: 0:20:38 lr: 0.000004 loss: 0.5189 (0.5176) time: 0.5378 data: 0.0002 max mem: 52126 +Epoch: [26] [1160/3634] eta: 0:20:37 lr: 0.000004 loss: 0.5102 (0.5176) time: 0.6215 data: 0.0002 max mem: 52126 +Epoch: [26] [1170/3634] eta: 0:20:33 lr: 0.000004 loss: 0.5448 (0.5180) time: 0.6312 data: 0.0001 max mem: 52126 +Epoch: [26] [1180/3634] eta: 0:20:30 lr: 0.000004 loss: 0.5448 (0.5179) time: 0.5886 data: 0.0001 max mem: 52126 +Epoch: [26] [1190/3634] eta: 0:20:25 lr: 0.000004 loss: 0.5018 (0.5179) time: 0.5327 data: 0.0001 max mem: 52126 +Epoch: [26] [1200/3634] eta: 0:20:17 lr: 0.000004 loss: 0.5007 (0.5176) time: 0.4273 data: 0.0001 max mem: 52126 +Epoch: [26] [1210/3634] eta: 0:20:10 lr: 0.000004 loss: 0.4846 (0.5175) time: 0.3817 data: 0.0001 max mem: 52126 +Epoch: [26] [1220/3634] eta: 0:20:03 lr: 0.000004 loss: 0.4900 (0.5174) time: 0.3825 data: 0.0001 max mem: 52126 +Epoch: [26] [1230/3634] eta: 0:19:55 lr: 0.000004 loss: 0.4888 (0.5173) time: 0.3830 data: 0.0001 max mem: 52126 +Epoch: [26] [1240/3634] eta: 0:19:50 lr: 0.000004 loss: 0.4720 (0.5169) time: 0.4287 data: 0.0001 max mem: 52126 +Epoch: [26] [1250/3634] eta: 0:19:43 lr: 0.000004 loss: 0.4764 (0.5168) time: 0.4286 data: 0.0001 max mem: 52126 +Epoch: [26] [1260/3634] eta: 0:19:36 lr: 0.000004 loss: 0.5132 (0.5167) time: 0.3828 data: 0.0001 max mem: 52126 +Epoch: [26] [1270/3634] eta: 0:19:29 lr: 0.000004 loss: 0.5151 (0.5168) time: 0.3825 data: 0.0001 max mem: 52126 +Epoch: [26] [1280/3634] eta: 0:19:24 lr: 0.000004 loss: 0.4982 (0.5165) time: 0.4407 data: 0.0001 max mem: 52126 +Epoch: [26] [1290/3634] eta: 0:19:17 lr: 0.000004 loss: 0.4977 (0.5163) time: 0.4411 data: 0.0002 max mem: 52126 +Epoch: [26] [1300/3634] eta: 0:19:10 lr: 0.000004 loss: 0.4982 (0.5162) time: 0.3827 data: 0.0002 max mem: 52126 +Epoch: [26] [1310/3634] eta: 0:19:06 lr: 0.000004 loss: 0.4905 (0.5161) time: 0.4750 data: 0.0001 max mem: 52126 +Epoch: [26] [1320/3634] eta: 0:19:03 lr: 0.000004 loss: 0.4855 (0.5160) time: 0.5793 data: 0.0002 max mem: 52126 +Epoch: [26] [1330/3634] eta: 0:19:02 lr: 0.000004 loss: 0.4991 (0.5160) time: 0.6518 data: 0.0002 max mem: 52126 +Epoch: [26] [1340/3634] eta: 0:18:55 lr: 0.000004 loss: 0.5161 (0.5161) time: 0.5472 data: 0.0002 max mem: 52126 +Epoch: [26] [1350/3634] eta: 0:18:51 lr: 0.000004 loss: 0.5379 (0.5160) time: 0.4704 data: 0.0002 max mem: 52126 +Epoch: [26] [1360/3634] eta: 0:18:46 lr: 0.000004 loss: 0.4993 (0.5160) time: 0.5305 data: 0.0001 max mem: 52126 +Epoch: [26] [1370/3634] eta: 0:18:41 lr: 0.000004 loss: 0.4874 (0.5159) time: 0.4871 data: 0.0001 max mem: 52126 +Epoch: [26] [1380/3634] eta: 0:18:36 lr: 0.000004 loss: 0.5144 (0.5160) time: 0.4867 data: 0.0002 max mem: 52126 +Epoch: [26] [1390/3634] eta: 0:18:31 lr: 0.000004 loss: 0.5272 (0.5161) time: 0.4921 data: 0.0002 max mem: 52126 +Epoch: [26] [1400/3634] eta: 0:18:26 lr: 0.000004 loss: 0.5391 (0.5161) time: 0.4912 data: 0.0002 max mem: 52126 +Epoch: [26] [1410/3634] eta: 0:18:19 lr: 0.000004 loss: 0.5185 (0.5159) time: 0.4404 data: 0.0001 max mem: 52126 +Epoch: [26] [1420/3634] eta: 0:18:14 lr: 0.000004 loss: 0.5016 (0.5158) time: 0.4281 data: 0.0001 max mem: 52126 +Epoch: [26] [1430/3634] eta: 0:18:09 lr: 0.000004 loss: 0.5083 (0.5156) time: 0.4892 data: 0.0001 max mem: 52126 +Epoch: [26] [1440/3634] eta: 0:18:04 lr: 0.000004 loss: 0.5569 (0.5159) time: 0.4999 data: 0.0001 max mem: 52126 +Epoch: [26] [1450/3634] eta: 0:18:01 lr: 0.000004 loss: 0.5407 (0.5157) time: 0.5566 data: 0.0002 max mem: 52126 +Epoch: [26] [1460/3634] eta: 0:17:54 lr: 0.000004 loss: 0.5038 (0.5157) time: 0.4996 data: 0.0002 max mem: 52126 +Epoch: [26] [1470/3634] eta: 0:17:48 lr: 0.000004 loss: 0.4728 (0.5154) time: 0.3827 data: 0.0001 max mem: 52126 +Epoch: [26] [1480/3634] eta: 0:17:44 lr: 0.000004 loss: 0.4983 (0.5154) time: 0.4863 data: 0.0001 max mem: 52126 +Epoch: [26] [1490/3634] eta: 0:17:40 lr: 0.000004 loss: 0.5234 (0.5153) time: 0.5451 data: 0.0001 max mem: 52126 +Epoch: [26] [1500/3634] eta: 0:17:37 lr: 0.000004 loss: 0.4868 (0.5152) time: 0.5952 data: 0.0002 max mem: 52126 +Epoch: [26] [1510/3634] eta: 0:17:32 lr: 0.000004 loss: 0.4937 (0.5152) time: 0.5948 data: 0.0001 max mem: 52126 +Epoch: [26] [1520/3634] eta: 0:17:29 lr: 0.000004 loss: 0.5028 (0.5151) time: 0.5583 data: 0.0001 max mem: 52126 +Epoch: [26] [1530/3634] eta: 0:17:25 lr: 0.000004 loss: 0.4968 (0.5149) time: 0.5879 data: 0.0001 max mem: 52126 +Epoch: [26] [1540/3634] eta: 0:17:20 lr: 0.000004 loss: 0.4994 (0.5148) time: 0.5155 data: 0.0001 max mem: 52126 +Epoch: [26] [1550/3634] eta: 0:17:16 lr: 0.000004 loss: 0.5242 (0.5148) time: 0.5477 data: 0.0001 max mem: 52126 +Epoch: [26] [1560/3634] eta: 0:17:11 lr: 0.000004 loss: 0.5152 (0.5148) time: 0.5485 data: 0.0002 max mem: 52126 +Epoch: [26] [1570/3634] eta: 0:17:06 lr: 0.000004 loss: 0.5019 (0.5149) time: 0.4769 data: 0.0002 max mem: 52126 +Epoch: [26] [1580/3634] eta: 0:17:02 lr: 0.000004 loss: 0.5019 (0.5149) time: 0.5340 data: 0.0001 max mem: 52126 +Epoch: [26] [1590/3634] eta: 0:16:57 lr: 0.000004 loss: 0.4936 (0.5148) time: 0.5438 data: 0.0002 max mem: 52126 +Epoch: [26] [1600/3634] eta: 0:16:51 lr: 0.000004 loss: 0.4936 (0.5147) time: 0.4413 data: 0.0002 max mem: 52126 +Epoch: [26] [1610/3634] eta: 0:16:46 lr: 0.000004 loss: 0.5250 (0.5148) time: 0.4441 data: 0.0002 max mem: 52126 +Epoch: [26] [1620/3634] eta: 0:16:41 lr: 0.000004 loss: 0.5257 (0.5150) time: 0.5084 data: 0.0002 max mem: 52126 +Epoch: [26] [1630/3634] eta: 0:16:35 lr: 0.000004 loss: 0.5257 (0.5151) time: 0.4471 data: 0.0002 max mem: 52126 +Epoch: [26] [1640/3634] eta: 0:16:28 lr: 0.000004 loss: 0.5144 (0.5151) time: 0.3819 data: 0.0002 max mem: 52126 +Epoch: [26] [1650/3634] eta: 0:16:22 lr: 0.000004 loss: 0.5177 (0.5151) time: 0.3815 data: 0.0002 max mem: 52126 +Epoch: [26] [1660/3634] eta: 0:16:16 lr: 0.000004 loss: 0.4902 (0.5149) time: 0.3817 data: 0.0002 max mem: 52126 +Epoch: [26] [1670/3634] eta: 0:16:09 lr: 0.000004 loss: 0.4762 (0.5149) time: 0.3821 data: 0.0002 max mem: 52126 +Epoch: [26] [1680/3634] eta: 0:16:05 lr: 0.000004 loss: 0.5401 (0.5150) time: 0.4420 data: 0.0002 max mem: 52126 +Epoch: [26] [1690/3634] eta: 0:16:00 lr: 0.000004 loss: 0.5128 (0.5150) time: 0.5296 data: 0.0002 max mem: 52126 +Epoch: [26] [1700/3634] eta: 0:15:55 lr: 0.000004 loss: 0.5068 (0.5148) time: 0.5129 data: 0.0002 max mem: 52126 +Epoch: [26] [1710/3634] eta: 0:15:50 lr: 0.000004 loss: 0.5095 (0.5148) time: 0.4881 data: 0.0002 max mem: 52126 +Epoch: [26] [1720/3634] eta: 0:15:45 lr: 0.000004 loss: 0.4905 (0.5147) time: 0.4910 data: 0.0002 max mem: 52126 +Epoch: [26] [1730/3634] eta: 0:15:41 lr: 0.000004 loss: 0.5059 (0.5149) time: 0.5317 data: 0.0002 max mem: 52126 +Epoch: [26] [1740/3634] eta: 0:15:35 lr: 0.000004 loss: 0.5450 (0.5149) time: 0.4859 data: 0.0002 max mem: 52126 +Epoch: [26] [1750/3634] eta: 0:15:30 lr: 0.000004 loss: 0.5307 (0.5150) time: 0.4410 data: 0.0002 max mem: 52126 +Epoch: [26] [1760/3634] eta: 0:15:25 lr: 0.000004 loss: 0.5278 (0.5149) time: 0.4852 data: 0.0002 max mem: 52126 +Epoch: [26] [1770/3634] eta: 0:15:20 lr: 0.000004 loss: 0.5422 (0.5151) time: 0.4722 data: 0.0002 max mem: 52126 +Epoch: [26] [1780/3634] eta: 0:15:17 lr: 0.000004 loss: 0.5354 (0.5151) time: 0.5812 data: 0.0002 max mem: 52126 +Epoch: [26] [1790/3634] eta: 0:15:12 lr: 0.000004 loss: 0.5295 (0.5153) time: 0.5801 data: 0.0002 max mem: 52126 +Epoch: [26] [1800/3634] eta: 0:15:06 lr: 0.000004 loss: 0.5344 (0.5153) time: 0.4267 data: 0.0002 max mem: 52126 +Epoch: [26] [1810/3634] eta: 0:15:03 lr: 0.000004 loss: 0.5136 (0.5152) time: 0.5269 data: 0.0002 max mem: 52126 +Epoch: [26] [1820/3634] eta: 0:14:57 lr: 0.000004 loss: 0.5349 (0.5154) time: 0.5271 data: 0.0002 max mem: 52126 +Epoch: [26] [1830/3634] eta: 0:14:50 lr: 0.000004 loss: 0.5472 (0.5155) time: 0.3814 data: 0.0002 max mem: 52126 +Epoch: [26] [1840/3634] eta: 0:14:44 lr: 0.000004 loss: 0.5027 (0.5153) time: 0.3812 data: 0.0002 max mem: 52126 +Epoch: [26] [1850/3634] eta: 0:14:40 lr: 0.000004 loss: 0.4503 (0.5151) time: 0.4858 data: 0.0002 max mem: 52126 +Epoch: [26] [1860/3634] eta: 0:14:36 lr: 0.000004 loss: 0.4967 (0.5151) time: 0.5896 data: 0.0002 max mem: 52126 +Epoch: [26] [1870/3634] eta: 0:14:30 lr: 0.000004 loss: 0.5329 (0.5152) time: 0.4855 data: 0.0002 max mem: 52126 +Epoch: [26] [1880/3634] eta: 0:14:27 lr: 0.000004 loss: 0.5153 (0.5150) time: 0.5037 data: 0.0002 max mem: 52126 +Epoch: [26] [1890/3634] eta: 0:14:21 lr: 0.000004 loss: 0.4920 (0.5150) time: 0.5033 data: 0.0002 max mem: 52126 +Epoch: [26] [1900/3634] eta: 0:14:16 lr: 0.000004 loss: 0.4933 (0.5148) time: 0.4256 data: 0.0002 max mem: 52126 +Epoch: [26] [1910/3634] eta: 0:14:11 lr: 0.000004 loss: 0.4933 (0.5149) time: 0.5297 data: 0.0002 max mem: 52126 +Epoch: [26] [1920/3634] eta: 0:14:06 lr: 0.000004 loss: 0.5290 (0.5149) time: 0.4857 data: 0.0002 max mem: 52126 +Epoch: [26] [1930/3634] eta: 0:14:00 lr: 0.000004 loss: 0.5495 (0.5151) time: 0.3819 data: 0.0002 max mem: 52126 +Epoch: [26] [1940/3634] eta: 0:13:55 lr: 0.000004 loss: 0.5359 (0.5149) time: 0.4742 data: 0.0002 max mem: 52126 +Epoch: [26] [1950/3634] eta: 0:13:49 lr: 0.000004 loss: 0.5315 (0.5151) time: 0.4744 data: 0.0002 max mem: 52126 +Epoch: [26] [1960/3634] eta: 0:13:46 lr: 0.000004 loss: 0.5111 (0.5149) time: 0.5015 data: 0.0002 max mem: 52126 +Epoch: [26] [1970/3634] eta: 0:13:40 lr: 0.000004 loss: 0.5111 (0.5150) time: 0.5467 data: 0.0002 max mem: 52126 +Epoch: [26] [1980/3634] eta: 0:13:35 lr: 0.000004 loss: 0.5288 (0.5150) time: 0.4707 data: 0.0002 max mem: 52126 +Epoch: [26] [1990/3634] eta: 0:13:29 lr: 0.000004 loss: 0.5016 (0.5150) time: 0.4253 data: 0.0002 max mem: 52126 +Epoch: [26] [2000/3634] eta: 0:13:25 lr: 0.000004 loss: 0.5227 (0.5151) time: 0.4402 data: 0.0002 max mem: 52126 +Epoch: [26] [2010/3634] eta: 0:13:23 lr: 0.000004 loss: 0.5508 (0.5152) time: 0.6708 data: 0.0002 max mem: 52126 +Epoch: [26] [2020/3634] eta: 0:13:17 lr: 0.000004 loss: 0.5425 (0.5153) time: 0.6124 data: 0.0002 max mem: 52126 +Epoch: [26] [2030/3634] eta: 0:13:12 lr: 0.000004 loss: 0.5072 (0.5151) time: 0.4407 data: 0.0002 max mem: 52126 +Epoch: [26] [2040/3634] eta: 0:13:06 lr: 0.000004 loss: 0.5056 (0.5152) time: 0.4411 data: 0.0002 max mem: 52126 +Epoch: [26] [2050/3634] eta: 0:13:02 lr: 0.000004 loss: 0.5132 (0.5152) time: 0.4771 data: 0.0002 max mem: 52126 +Epoch: [26] [2060/3634] eta: 0:12:57 lr: 0.000004 loss: 0.5132 (0.5152) time: 0.5702 data: 0.0002 max mem: 52126 +Epoch: [26] [2070/3634] eta: 0:12:52 lr: 0.000004 loss: 0.5247 (0.5152) time: 0.5350 data: 0.0002 max mem: 52126 +Epoch: [26] [2080/3634] eta: 0:12:47 lr: 0.000004 loss: 0.5123 (0.5151) time: 0.4858 data: 0.0002 max mem: 52126 +Epoch: [26] [2090/3634] eta: 0:12:41 lr: 0.000004 loss: 0.5123 (0.5150) time: 0.4261 data: 0.0002 max mem: 52126 +Epoch: [26] [2100/3634] eta: 0:12:38 lr: 0.000004 loss: 0.4944 (0.5147) time: 0.5466 data: 0.0002 max mem: 52126 +Epoch: [26] [2110/3634] eta: 0:12:34 lr: 0.000004 loss: 0.5190 (0.5148) time: 0.6725 data: 0.0002 max mem: 52126 +Epoch: [26] [2120/3634] eta: 0:12:29 lr: 0.000004 loss: 0.5629 (0.5151) time: 0.5668 data: 0.0002 max mem: 52126 +Epoch: [26] [2130/3634] eta: 0:12:24 lr: 0.000004 loss: 0.5585 (0.5150) time: 0.4843 data: 0.0002 max mem: 52126 +Epoch: [26] [2140/3634] eta: 0:12:19 lr: 0.000004 loss: 0.5448 (0.5151) time: 0.4762 data: 0.0002 max mem: 52126 +Epoch: [26] [2150/3634] eta: 0:12:14 lr: 0.000004 loss: 0.5448 (0.5152) time: 0.4779 data: 0.0002 max mem: 52126 +Epoch: [26] [2160/3634] eta: 0:12:08 lr: 0.000004 loss: 0.5121 (0.5152) time: 0.4269 data: 0.0002 max mem: 52126 +Epoch: [26] [2170/3634] eta: 0:12:03 lr: 0.000004 loss: 0.4721 (0.5150) time: 0.3822 data: 0.0002 max mem: 52126 +Epoch: [26] [2180/3634] eta: 0:11:58 lr: 0.000004 loss: 0.4809 (0.5149) time: 0.4856 data: 0.0002 max mem: 52126 +Epoch: [26] [2190/3634] eta: 0:11:54 lr: 0.000004 loss: 0.5436 (0.5150) time: 0.5956 data: 0.0002 max mem: 52126 +Epoch: [26] [2200/3634] eta: 0:11:50 lr: 0.000004 loss: 0.5456 (0.5151) time: 0.5834 data: 0.0002 max mem: 52126 +Epoch: [26] [2210/3634] eta: 0:11:44 lr: 0.000004 loss: 0.5353 (0.5150) time: 0.4733 data: 0.0002 max mem: 52126 +Epoch: [26] [2220/3634] eta: 0:11:39 lr: 0.000004 loss: 0.4866 (0.5150) time: 0.4424 data: 0.0002 max mem: 52126 +Epoch: [26] [2230/3634] eta: 0:11:33 lr: 0.000004 loss: 0.4941 (0.5150) time: 0.4424 data: 0.0002 max mem: 52126 +Epoch: [26] [2240/3634] eta: 0:11:28 lr: 0.000004 loss: 0.5174 (0.5150) time: 0.3823 data: 0.0002 max mem: 52126 +Epoch: [26] [2250/3634] eta: 0:11:22 lr: 0.000004 loss: 0.5636 (0.5152) time: 0.3817 data: 0.0002 max mem: 52126 +Epoch: [26] [2260/3634] eta: 0:11:17 lr: 0.000004 loss: 0.5400 (0.5151) time: 0.4271 data: 0.0002 max mem: 52126 +Epoch: [26] [2270/3634] eta: 0:11:11 lr: 0.000004 loss: 0.4986 (0.5151) time: 0.4280 data: 0.0002 max mem: 52126 +Epoch: [26] [2280/3634] eta: 0:11:07 lr: 0.000004 loss: 0.5274 (0.5153) time: 0.4871 data: 0.0002 max mem: 52126 +Epoch: [26] [2290/3634] eta: 0:11:02 lr: 0.000004 loss: 0.5257 (0.5151) time: 0.5502 data: 0.0002 max mem: 52126 +Epoch: [26] [2300/3634] eta: 0:10:57 lr: 0.000004 loss: 0.4889 (0.5151) time: 0.5043 data: 0.0002 max mem: 52126 +Epoch: [26] [2310/3634] eta: 0:10:52 lr: 0.000004 loss: 0.5360 (0.5152) time: 0.4407 data: 0.0002 max mem: 52126 +Epoch: [26] [2320/3634] eta: 0:10:47 lr: 0.000004 loss: 0.5312 (0.5153) time: 0.4268 data: 0.0002 max mem: 52126 +Epoch: [26] [2330/3634] eta: 0:10:43 lr: 0.000004 loss: 0.5067 (0.5153) time: 0.5911 data: 0.0002 max mem: 52126 +Epoch: [26] [2340/3634] eta: 0:10:38 lr: 0.000004 loss: 0.5040 (0.5152) time: 0.5902 data: 0.0002 max mem: 52126 +Epoch: [26] [2350/3634] eta: 0:10:32 lr: 0.000004 loss: 0.4974 (0.5152) time: 0.4260 data: 0.0002 max mem: 52126 +Epoch: [26] [2360/3634] eta: 0:10:28 lr: 0.000004 loss: 0.5046 (0.5152) time: 0.4863 data: 0.0002 max mem: 52126 +Epoch: [26] [2370/3634] eta: 0:10:23 lr: 0.000004 loss: 0.5111 (0.5152) time: 0.5305 data: 0.0002 max mem: 52126 +Epoch: [26] [2380/3634] eta: 0:10:19 lr: 0.000004 loss: 0.5112 (0.5152) time: 0.5907 data: 0.0002 max mem: 52126 +Epoch: [26] [2390/3634] eta: 0:10:14 lr: 0.000004 loss: 0.5341 (0.5153) time: 0.5913 data: 0.0002 max mem: 52126 +Epoch: [26] [2400/3634] eta: 0:10:09 lr: 0.000004 loss: 0.5227 (0.5152) time: 0.4697 data: 0.0002 max mem: 52126 +Epoch: [26] [2410/3634] eta: 0:10:04 lr: 0.000004 loss: 0.4630 (0.5149) time: 0.4243 data: 0.0002 max mem: 52126 +Epoch: [26] [2420/3634] eta: 0:09:59 lr: 0.000004 loss: 0.4609 (0.5149) time: 0.4291 data: 0.0002 max mem: 52126 +Epoch: [26] [2430/3634] eta: 0:09:56 lr: 0.000004 loss: 0.4938 (0.5149) time: 0.6985 data: 0.0002 max mem: 52126 +Epoch: [26] [2440/3634] eta: 0:09:50 lr: 0.000004 loss: 0.5294 (0.5150) time: 0.6503 data: 0.0002 max mem: 52126 +Epoch: [26] [2450/3634] eta: 0:09:45 lr: 0.000004 loss: 0.5494 (0.5149) time: 0.3816 data: 0.0002 max mem: 52126 +Epoch: [26] [2460/3634] eta: 0:09:39 lr: 0.000004 loss: 0.5335 (0.5151) time: 0.3842 data: 0.0002 max mem: 52126 +Epoch: [26] [2470/3634] eta: 0:09:34 lr: 0.000004 loss: 0.5350 (0.5151) time: 0.3847 data: 0.0002 max mem: 52126 +Epoch: [26] [2480/3634] eta: 0:09:29 lr: 0.000004 loss: 0.5487 (0.5152) time: 0.4419 data: 0.0002 max mem: 52126 +Epoch: [26] [2490/3634] eta: 0:09:24 lr: 0.000004 loss: 0.5310 (0.5151) time: 0.5002 data: 0.0002 max mem: 52126 +Epoch: [26] [2500/3634] eta: 0:09:19 lr: 0.000004 loss: 0.4843 (0.5149) time: 0.5011 data: 0.0002 max mem: 52126 +Epoch: [26] [2510/3634] eta: 0:09:14 lr: 0.000004 loss: 0.4543 (0.5147) time: 0.5052 data: 0.0002 max mem: 52126 +Epoch: [26] [2520/3634] eta: 0:09:09 lr: 0.000004 loss: 0.4970 (0.5147) time: 0.5038 data: 0.0002 max mem: 52126 +Epoch: [26] [2530/3634] eta: 0:09:04 lr: 0.000004 loss: 0.5480 (0.5149) time: 0.4891 data: 0.0002 max mem: 52126 +Epoch: [26] [2540/3634] eta: 0:08:59 lr: 0.000004 loss: 0.5704 (0.5151) time: 0.4308 data: 0.0002 max mem: 52126 +Epoch: [26] [2550/3634] eta: 0:08:53 lr: 0.000004 loss: 0.5461 (0.5150) time: 0.3821 data: 0.0002 max mem: 52126 +Epoch: [26] [2560/3634] eta: 0:08:49 lr: 0.000004 loss: 0.5153 (0.5151) time: 0.4902 data: 0.0002 max mem: 52126 +Epoch: [26] [2570/3634] eta: 0:08:44 lr: 0.000004 loss: 0.5153 (0.5151) time: 0.4911 data: 0.0002 max mem: 52126 +Epoch: [26] [2580/3634] eta: 0:08:39 lr: 0.000004 loss: 0.5321 (0.5152) time: 0.4419 data: 0.0002 max mem: 52126 +Epoch: [26] [2590/3634] eta: 0:08:35 lr: 0.000004 loss: 0.5329 (0.5152) time: 0.6066 data: 0.0002 max mem: 52126 +Epoch: [26] [2600/3634] eta: 0:08:30 lr: 0.000004 loss: 0.5120 (0.5152) time: 0.6538 data: 0.0002 max mem: 52126 +Epoch: [26] [2610/3634] eta: 0:08:25 lr: 0.000004 loss: 0.4902 (0.5153) time: 0.4885 data: 0.0002 max mem: 52126 +Epoch: [26] [2620/3634] eta: 0:08:19 lr: 0.000004 loss: 0.4871 (0.5151) time: 0.3819 data: 0.0002 max mem: 52126 +Epoch: [26] [2630/3634] eta: 0:08:15 lr: 0.000004 loss: 0.4871 (0.5150) time: 0.4414 data: 0.0002 max mem: 52126 +Epoch: [26] [2640/3634] eta: 0:08:10 lr: 0.000004 loss: 0.4996 (0.5150) time: 0.5003 data: 0.0002 max mem: 52126 +Epoch: [26] [2650/3634] eta: 0:08:05 lr: 0.000004 loss: 0.5114 (0.5149) time: 0.5640 data: 0.0002 max mem: 52126 +Epoch: [26] [2660/3634] eta: 0:08:01 lr: 0.000004 loss: 0.5311 (0.5150) time: 0.6769 data: 0.0002 max mem: 52126 +Epoch: [26] [2670/3634] eta: 0:07:57 lr: 0.000004 loss: 0.5013 (0.5149) time: 0.7380 data: 0.0002 max mem: 52126 +Epoch: [26] [2680/3634] eta: 0:07:52 lr: 0.000004 loss: 0.5013 (0.5149) time: 0.6106 data: 0.0002 max mem: 52126 +Epoch: [26] [2690/3634] eta: 0:07:47 lr: 0.000004 loss: 0.5132 (0.5148) time: 0.4701 data: 0.0002 max mem: 52126 +Epoch: [26] [2700/3634] eta: 0:07:42 lr: 0.000004 loss: 0.5047 (0.5148) time: 0.4712 data: 0.0002 max mem: 52126 +Epoch: [26] [2710/3634] eta: 0:07:37 lr: 0.000004 loss: 0.5055 (0.5147) time: 0.4271 data: 0.0002 max mem: 52126 +Epoch: [26] [2720/3634] eta: 0:07:31 lr: 0.000004 loss: 0.5455 (0.5147) time: 0.3814 data: 0.0002 max mem: 52126 +Epoch: [26] [2730/3634] eta: 0:07:26 lr: 0.000004 loss: 0.5525 (0.5149) time: 0.3816 data: 0.0002 max mem: 52126 +Epoch: [26] [2740/3634] eta: 0:07:21 lr: 0.000004 loss: 0.5011 (0.5147) time: 0.4413 data: 0.0002 max mem: 52126 +Epoch: [26] [2750/3634] eta: 0:07:16 lr: 0.000004 loss: 0.4726 (0.5146) time: 0.4415 data: 0.0002 max mem: 52126 +Epoch: [26] [2760/3634] eta: 0:07:10 lr: 0.000004 loss: 0.5027 (0.5145) time: 0.3819 data: 0.0002 max mem: 52126 +Epoch: [26] [2770/3634] eta: 0:07:05 lr: 0.000004 loss: 0.5054 (0.5145) time: 0.3814 data: 0.0002 max mem: 52126 +Epoch: [26] [2780/3634] eta: 0:07:00 lr: 0.000004 loss: 0.5521 (0.5146) time: 0.4426 data: 0.0002 max mem: 52126 +Epoch: [26] [2790/3634] eta: 0:06:55 lr: 0.000004 loss: 0.5345 (0.5144) time: 0.4431 data: 0.0002 max mem: 52126 +Epoch: [26] [2800/3634] eta: 0:06:50 lr: 0.000004 loss: 0.4621 (0.5143) time: 0.4887 data: 0.0002 max mem: 52126 +Epoch: [26] [2810/3634] eta: 0:06:45 lr: 0.000004 loss: 0.5047 (0.5144) time: 0.5326 data: 0.0002 max mem: 52126 +Epoch: [26] [2820/3634] eta: 0:06:41 lr: 0.000004 loss: 0.5290 (0.5144) time: 0.5147 data: 0.0002 max mem: 52126 +Epoch: [26] [2830/3634] eta: 0:06:36 lr: 0.000004 loss: 0.5282 (0.5143) time: 0.5164 data: 0.0002 max mem: 52126 +Epoch: [26] [2840/3634] eta: 0:06:30 lr: 0.000004 loss: 0.5031 (0.5144) time: 0.4278 data: 0.0002 max mem: 52126 +Epoch: [26] [2850/3634] eta: 0:06:26 lr: 0.000004 loss: 0.5273 (0.5145) time: 0.4719 data: 0.0002 max mem: 52126 +Epoch: [26] [2860/3634] eta: 0:06:21 lr: 0.000004 loss: 0.5242 (0.5145) time: 0.5790 data: 0.0002 max mem: 52126 +Epoch: [26] [2870/3634] eta: 0:06:16 lr: 0.000004 loss: 0.5167 (0.5145) time: 0.4888 data: 0.0002 max mem: 52126 +Epoch: [26] [2880/3634] eta: 0:06:11 lr: 0.000004 loss: 0.5343 (0.5145) time: 0.3809 data: 0.0002 max mem: 52126 +Epoch: [26] [2890/3634] eta: 0:06:06 lr: 0.000004 loss: 0.5215 (0.5145) time: 0.4897 data: 0.0002 max mem: 52126 +Epoch: [26] [2900/3634] eta: 0:06:01 lr: 0.000004 loss: 0.5070 (0.5145) time: 0.5336 data: 0.0002 max mem: 52126 +Epoch: [26] [2910/3634] eta: 0:05:56 lr: 0.000004 loss: 0.4941 (0.5145) time: 0.5288 data: 0.0002 max mem: 52126 +Epoch: [26] [2920/3634] eta: 0:05:51 lr: 0.000004 loss: 0.4927 (0.5145) time: 0.5439 data: 0.0002 max mem: 52126 +Epoch: [26] [2930/3634] eta: 0:05:46 lr: 0.000004 loss: 0.4927 (0.5144) time: 0.4405 data: 0.0002 max mem: 52126 +Epoch: [26] [2940/3634] eta: 0:05:41 lr: 0.000004 loss: 0.5069 (0.5144) time: 0.4265 data: 0.0002 max mem: 52126 +Epoch: [26] [2950/3634] eta: 0:05:36 lr: 0.000004 loss: 0.5328 (0.5144) time: 0.4834 data: 0.0002 max mem: 52126 +Epoch: [26] [2960/3634] eta: 0:05:32 lr: 0.000004 loss: 0.5140 (0.5145) time: 0.5271 data: 0.0002 max mem: 52126 +Epoch: [26] [2970/3634] eta: 0:05:27 lr: 0.000004 loss: 0.5334 (0.5146) time: 0.5752 data: 0.0002 max mem: 52126 +Epoch: [26] [2980/3634] eta: 0:05:22 lr: 0.000004 loss: 0.5165 (0.5145) time: 0.5475 data: 0.0002 max mem: 52126 +Epoch: [26] [2990/3634] eta: 0:05:17 lr: 0.000004 loss: 0.4907 (0.5146) time: 0.4883 data: 0.0002 max mem: 52126 +Epoch: [26] [3000/3634] eta: 0:05:12 lr: 0.000004 loss: 0.4907 (0.5145) time: 0.4771 data: 0.0002 max mem: 52126 +Epoch: [26] [3010/3634] eta: 0:05:07 lr: 0.000004 loss: 0.5293 (0.5146) time: 0.4747 data: 0.0002 max mem: 52126 +Epoch: [26] [3020/3634] eta: 0:05:02 lr: 0.000004 loss: 0.5294 (0.5146) time: 0.4858 data: 0.0002 max mem: 52126 +Epoch: [26] [3030/3634] eta: 0:04:57 lr: 0.000004 loss: 0.5196 (0.5146) time: 0.5063 data: 0.0002 max mem: 52126 +Epoch: [26] [3040/3634] eta: 0:04:52 lr: 0.000004 loss: 0.5196 (0.5146) time: 0.5078 data: 0.0002 max mem: 52126 +Epoch: [26] [3050/3634] eta: 0:04:48 lr: 0.000004 loss: 0.4888 (0.5145) time: 0.5482 data: 0.0002 max mem: 52126 +Epoch: [26] [3060/3634] eta: 0:04:43 lr: 0.000004 loss: 0.4958 (0.5144) time: 0.5495 data: 0.0002 max mem: 52126 +Epoch: [26] [3070/3634] eta: 0:04:38 lr: 0.000004 loss: 0.5191 (0.5144) time: 0.5485 data: 0.0002 max mem: 52126 +Epoch: [26] [3080/3634] eta: 0:04:33 lr: 0.000004 loss: 0.5512 (0.5144) time: 0.4853 data: 0.0002 max mem: 52126 +Epoch: [26] [3090/3634] eta: 0:04:28 lr: 0.000004 loss: 0.5313 (0.5145) time: 0.4912 data: 0.0002 max mem: 52126 +Epoch: [26] [3100/3634] eta: 0:04:23 lr: 0.000004 loss: 0.5170 (0.5145) time: 0.4910 data: 0.0002 max mem: 52126 +Epoch: [26] [3110/3634] eta: 0:04:18 lr: 0.000004 loss: 0.5170 (0.5145) time: 0.4880 data: 0.0002 max mem: 52126 +Epoch: [26] [3120/3634] eta: 0:04:13 lr: 0.000004 loss: 0.5072 (0.5145) time: 0.5474 data: 0.0002 max mem: 52126 +Epoch: [26] [3130/3634] eta: 0:04:08 lr: 0.000004 loss: 0.5072 (0.5146) time: 0.4403 data: 0.0002 max mem: 52126 +Epoch: [26] [3140/3634] eta: 0:04:03 lr: 0.000004 loss: 0.5107 (0.5146) time: 0.3809 data: 0.0002 max mem: 52126 +Epoch: [26] [3150/3634] eta: 0:03:58 lr: 0.000004 loss: 0.5166 (0.5146) time: 0.4417 data: 0.0002 max mem: 52126 +Epoch: [26] [3160/3634] eta: 0:03:53 lr: 0.000004 loss: 0.5166 (0.5146) time: 0.4902 data: 0.0002 max mem: 52126 +Epoch: [26] [3170/3634] eta: 0:03:48 lr: 0.000004 loss: 0.5245 (0.5146) time: 0.5339 data: 0.0002 max mem: 52126 +Epoch: [26] [3180/3634] eta: 0:03:44 lr: 0.000004 loss: 0.5366 (0.5148) time: 0.6630 data: 0.0002 max mem: 52126 +Epoch: [26] [3190/3634] eta: 0:03:39 lr: 0.000004 loss: 0.5360 (0.5147) time: 0.6210 data: 0.0002 max mem: 52126 +Epoch: [26] [3200/3634] eta: 0:03:34 lr: 0.000004 loss: 0.5167 (0.5148) time: 0.4431 data: 0.0002 max mem: 52126 +Epoch: [26] [3210/3634] eta: 0:03:29 lr: 0.000004 loss: 0.4924 (0.5147) time: 0.3819 data: 0.0002 max mem: 52126 +Epoch: [26] [3220/3634] eta: 0:03:24 lr: 0.000004 loss: 0.4924 (0.5147) time: 0.5116 data: 0.0002 max mem: 52126 +Epoch: [26] [3230/3634] eta: 0:03:19 lr: 0.000004 loss: 0.5171 (0.5148) time: 0.5112 data: 0.0002 max mem: 52126 +Epoch: [26] [3240/3634] eta: 0:03:14 lr: 0.000004 loss: 0.5420 (0.5149) time: 0.6012 data: 0.0002 max mem: 52126 +Epoch: [26] [3250/3634] eta: 0:03:10 lr: 0.000004 loss: 0.5353 (0.5149) time: 0.7376 data: 0.0002 max mem: 52126 +Epoch: [26] [3260/3634] eta: 0:03:04 lr: 0.000004 loss: 0.5376 (0.5150) time: 0.5183 data: 0.0002 max mem: 52126 +Epoch: [26] [3270/3634] eta: 0:03:00 lr: 0.000004 loss: 0.5385 (0.5148) time: 0.4712 data: 0.0002 max mem: 52126 +Epoch: [26] [3280/3634] eta: 0:02:55 lr: 0.000004 loss: 0.4809 (0.5148) time: 0.5323 data: 0.0002 max mem: 52126 +Epoch: [26] [3290/3634] eta: 0:02:50 lr: 0.000004 loss: 0.5391 (0.5150) time: 0.5015 data: 0.0002 max mem: 52126 +Epoch: [26] [3300/3634] eta: 0:02:45 lr: 0.000004 loss: 0.5576 (0.5150) time: 0.4408 data: 0.0002 max mem: 52126 +Epoch: [26] [3310/3634] eta: 0:02:40 lr: 0.000004 loss: 0.4865 (0.5149) time: 0.4435 data: 0.0002 max mem: 52126 +Epoch: [26] [3320/3634] eta: 0:02:35 lr: 0.000004 loss: 0.4855 (0.5149) time: 0.4435 data: 0.0002 max mem: 52126 +Epoch: [26] [3330/3634] eta: 0:02:30 lr: 0.000004 loss: 0.4949 (0.5148) time: 0.4261 data: 0.0002 max mem: 52126 +Epoch: [26] [3340/3634] eta: 0:02:25 lr: 0.000004 loss: 0.4949 (0.5148) time: 0.4255 data: 0.0002 max mem: 52126 +Epoch: [26] [3350/3634] eta: 0:02:20 lr: 0.000004 loss: 0.5075 (0.5148) time: 0.3813 data: 0.0002 max mem: 52126 +Epoch: [26] [3360/3634] eta: 0:02:15 lr: 0.000004 loss: 0.5073 (0.5148) time: 0.4877 data: 0.0002 max mem: 52126 +Epoch: [26] [3370/3634] eta: 0:02:10 lr: 0.000004 loss: 0.4948 (0.5148) time: 0.5323 data: 0.0002 max mem: 52126 +Epoch: [26] [3380/3634] eta: 0:02:05 lr: 0.000004 loss: 0.4922 (0.5147) time: 0.4712 data: 0.0002 max mem: 52126 +Epoch: [26] [3390/3634] eta: 0:02:00 lr: 0.000004 loss: 0.5160 (0.5148) time: 0.4727 data: 0.0002 max mem: 52126 +Epoch: [26] [3400/3634] eta: 0:01:55 lr: 0.000004 loss: 0.5150 (0.5148) time: 0.4280 data: 0.0002 max mem: 52126 +Epoch: [26] [3410/3634] eta: 0:01:50 lr: 0.000004 loss: 0.4922 (0.5148) time: 0.4848 data: 0.0002 max mem: 52126 +Epoch: [26] [3420/3634] eta: 0:01:45 lr: 0.000004 loss: 0.5239 (0.5148) time: 0.4846 data: 0.0002 max mem: 52126 +Epoch: [26] [3430/3634] eta: 0:01:40 lr: 0.000004 loss: 0.5115 (0.5149) time: 0.4754 data: 0.0002 max mem: 52126 +Epoch: [26] [3440/3634] eta: 0:01:35 lr: 0.000004 loss: 0.5184 (0.5149) time: 0.5361 data: 0.0002 max mem: 52126 +Epoch: [26] [3450/3634] eta: 0:01:30 lr: 0.000004 loss: 0.5449 (0.5150) time: 0.5380 data: 0.0002 max mem: 52126 +Epoch: [26] [3460/3634] eta: 0:01:25 lr: 0.000004 loss: 0.5349 (0.5149) time: 0.4774 data: 0.0002 max mem: 52126 +Epoch: [26] [3470/3634] eta: 0:01:20 lr: 0.000004 loss: 0.5186 (0.5150) time: 0.3820 data: 0.0002 max mem: 52126 +Epoch: [26] [3480/3634] eta: 0:01:15 lr: 0.000004 loss: 0.5277 (0.5150) time: 0.4433 data: 0.0002 max mem: 52126 +Epoch: [26] [3490/3634] eta: 0:01:10 lr: 0.000004 loss: 0.5277 (0.5150) time: 0.4882 data: 0.0002 max mem: 52126 +Epoch: [26] [3500/3634] eta: 0:01:06 lr: 0.000004 loss: 0.5427 (0.5151) time: 0.4713 data: 0.0002 max mem: 52126 +Epoch: [26] [3510/3634] eta: 0:01:01 lr: 0.000004 loss: 0.5427 (0.5152) time: 0.5381 data: 0.0002 max mem: 52126 +Epoch: [26] [3520/3634] eta: 0:00:56 lr: 0.000004 loss: 0.5366 (0.5152) time: 0.4930 data: 0.0002 max mem: 52126 +Epoch: [26] [3530/3634] eta: 0:00:51 lr: 0.000004 loss: 0.5385 (0.5153) time: 0.5970 data: 0.0002 max mem: 52126 +Epoch: [26] [3540/3634] eta: 0:00:46 lr: 0.000004 loss: 0.5385 (0.5153) time: 0.6848 data: 0.0002 max mem: 52126 +Epoch: [26] [3550/3634] eta: 0:00:41 lr: 0.000004 loss: 0.5354 (0.5154) time: 0.5792 data: 0.0002 max mem: 52126 +Epoch: [26] [3560/3634] eta: 0:00:36 lr: 0.000004 loss: 0.5717 (0.5155) time: 0.5404 data: 0.0002 max mem: 52126 +Epoch: [26] [3570/3634] eta: 0:00:31 lr: 0.000004 loss: 0.5414 (0.5156) time: 0.4303 data: 0.0002 max mem: 52126 +Epoch: [26] [3580/3634] eta: 0:00:26 lr: 0.000004 loss: 0.5283 (0.5155) time: 0.4423 data: 0.0002 max mem: 52126 +Epoch: [26] [3590/3634] eta: 0:00:21 lr: 0.000004 loss: 0.5298 (0.5156) time: 0.6524 data: 0.0002 max mem: 52126 +Epoch: [26] [3600/3634] eta: 0:00:16 lr: 0.000004 loss: 0.5527 (0.5157) time: 0.5923 data: 0.0002 max mem: 52126 +Epoch: [26] [3610/3634] eta: 0:00:11 lr: 0.000004 loss: 0.5371 (0.5157) time: 0.4413 data: 0.0002 max mem: 52126 +Epoch: [26] [3620/3634] eta: 0:00:06 lr: 0.000004 loss: 0.5659 (0.5158) time: 0.4435 data: 0.0023 max mem: 52126 +Epoch: [26] [3630/3634] eta: 0:00:01 lr: 0.000004 loss: 0.5435 (0.5158) time: 0.3847 data: 0.0023 max mem: 52126 +Epoch: [26] [3633/3634] eta: 0:00:00 lr: 0.000004 loss: 0.5435 (0.5158) time: 0.3849 data: 0.0022 max mem: 52126 +Epoch: [26] Total time: 0:29:56 (0.4944 s / it) +Averaged stats: lr: 0.000004 loss: 0.5435 (0.5169) +Test: [ 0/1403] eta: 1:06:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0899 (0.0899) acc1: 100.0000 (100.0000) acc5: 100.0000 (100.0000) time: 2.8502 data: 2.7536 max mem: 52126 +Test: [ 10/1403] eta: 0:08:02 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1539 (0.1745) acc1: 98.4375 (95.4545) acc5: 100.0000 (100.0000) time: 0.3463 data: 0.2506 max mem: 52126 +Test: [ 20/1403] eta: 0:05:14 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1331 (0.1567) acc1: 98.4375 (95.9077) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 30/1403] eta: 0:04:13 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0874 (0.1481) acc1: 100.0000 (96.2702) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 40/1403] eta: 0:03:42 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0964 (0.1483) acc1: 100.0000 (96.6082) acc5: 100.0000 (100.0000) time: 0.0959 data: 0.0003 max mem: 52126 +Test: [ 50/1403] eta: 0:03:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1213 (0.1631) acc1: 100.0000 (95.7721) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 60/1403] eta: 0:03:09 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.1263 (0.1608) acc1: 100.0000 (96.1066) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 70/1403] eta: 0:02:59 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.1369 (0.1646) acc1: 98.4375 (95.8407) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 80/1403] eta: 0:02:51 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1590 (0.1698) acc1: 96.8750 (95.5633) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 90/1403] eta: 0:02:45 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.1743 (0.1893) acc1: 93.7500 (94.6257) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 100/1403] eta: 0:02:40 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1743 (0.1883) acc1: 95.3125 (94.7246) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 110/1403] eta: 0:02:36 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1440 (0.1864) acc1: 98.4375 (94.8480) acc5: 100.0000 (100.0000) time: 0.0959 data: 0.0003 max mem: 52126 +Test: [ 120/1403] eta: 0:02:32 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0802 (0.1774) acc1: 100.0000 (95.2608) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 130/1403] eta: 0:02:28 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.0802 (0.1713) acc1: 100.0000 (95.4914) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 140/1403] eta: 0:02:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0753 (0.1644) acc1: 100.0000 (95.8112) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 150/1403] eta: 0:02:22 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0729 (0.1586) acc1: 100.0000 (96.0679) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 160/1403] eta: 0:02:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0746 (0.1540) acc1: 100.0000 (96.2636) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 170/1403] eta: 0:02:18 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0764 (0.1501) acc1: 100.0000 (96.4547) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 180/1403] eta: 0:02:15 acc1_pos: 90.625 acc5_pos: 100.000 loss: 0.0816 (0.1529) acc1: 100.0000 (96.3570) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 190/1403] eta: 0:02:13 acc1_pos: 90.625 acc5_pos: 100.000 loss: 0.2388 (0.1591) acc1: 93.7500 (96.0406) acc5: 100.0000 (100.0000) time: 0.0959 data: 0.0003 max mem: 52126 +Test: [ 200/1403] eta: 0:02:11 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.2388 (0.1688) acc1: 90.6250 (95.5613) acc5: 100.0000 (100.0000) time: 0.0959 data: 0.0003 max mem: 52126 +Test: [ 210/1403] eta: 0:02:09 acc1_pos: 79.688 acc5_pos: 100.000 loss: 0.2824 (0.1774) acc1: 89.0625 (95.1052) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 220/1403] eta: 0:02:08 acc1_pos: 89.062 acc5_pos: 100.000 loss: 0.2995 (0.1849) acc1: 89.0625 (94.7257) acc5: 100.0000 (100.0000) time: 0.0959 data: 0.0003 max mem: 52126 +Test: [ 230/1403] eta: 0:02:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2367 (0.1811) acc1: 92.1875 (94.8999) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 240/1403] eta: 0:02:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0765 (0.1769) acc1: 100.0000 (95.0985) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 250/1403] eta: 0:02:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0770 (0.1732) acc1: 100.0000 (95.2751) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 260/1403] eta: 0:02:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0770 (0.1697) acc1: 100.0000 (95.4382) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 270/1403] eta: 0:02:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0768 (0.1666) acc1: 100.0000 (95.5893) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 280/1403] eta: 0:01:58 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0771 (0.1633) acc1: 100.0000 (95.7407) acc5: 100.0000 (100.0000) time: 0.0959 data: 0.0003 max mem: 52126 +Test: [ 290/1403] eta: 0:01:57 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0739 (0.1603) acc1: 100.0000 (95.8817) acc5: 100.0000 (100.0000) time: 0.0959 data: 0.0003 max mem: 52126 +Test: [ 300/1403] eta: 0:01:55 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0730 (0.1576) acc1: 100.0000 (96.0133) acc5: 100.0000 (100.0000) time: 0.0959 data: 0.0003 max mem: 52126 +Test: [ 310/1403] eta: 0:01:54 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0762 (0.1552) acc1: 100.0000 (96.1365) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0959 data: 0.0003 max mem: 52126 +Test: [ 320/1403] eta: 0:01:53 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0994 (0.1536) acc1: 100.0000 (96.2568) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 330/1403] eta: 0:01:51 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1080 (0.1523) acc1: 100.0000 (96.3699) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 340/1403] eta: 0:01:50 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1069 (0.1510) acc1: 100.0000 (96.4764) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 350/1403] eta: 0:01:49 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1045 (0.1497) acc1: 100.0000 (96.5767) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 360/1403] eta: 0:01:47 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1089 (0.1487) acc1: 100.0000 (96.6629) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 370/1403] eta: 0:01:46 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1068 (0.1475) acc1: 100.0000 (96.7529) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 380/1403] eta: 0:01:45 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1056 (0.1464) acc1: 100.0000 (96.8381) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 390/1403] eta: 0:01:44 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1065 (0.1474) acc1: 100.0000 (96.7791) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 400/1403] eta: 0:01:43 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2574 (0.1514) acc1: 89.0625 (96.5594) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 410/1403] eta: 0:01:41 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2851 (0.1547) acc1: 87.5000 (96.3694) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 420/1403] eta: 0:01:40 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2562 (0.1565) acc1: 89.0625 (96.2626) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 430/1403] eta: 0:01:39 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2441 (0.1595) acc1: 90.6250 (96.0992) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 440/1403] eta: 0:01:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2976 (0.1631) acc1: 89.0625 (95.9219) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 450/1403] eta: 0:01:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2841 (0.1656) acc1: 89.0625 (95.7906) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 460/1403] eta: 0:01:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2549 (0.1682) acc1: 89.0625 (95.6243) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 470/1403] eta: 0:01:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2215 (0.1674) acc1: 90.6250 (95.6442) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 480/1403] eta: 0:01:33 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0625 (0.1652) acc1: 100.0000 (95.7348) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 490/1403] eta: 0:01:32 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0625 (0.1631) acc1: 100.0000 (95.8217) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 500/1403] eta: 0:01:31 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0628 (0.1611) acc1: 100.0000 (95.9051) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 510/1403] eta: 0:01:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0631 (0.1592) acc1: 100.0000 (95.9852) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 520/1403] eta: 0:01:29 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0671 (0.1575) acc1: 100.0000 (96.0623) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 530/1403] eta: 0:01:28 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0595 (0.1557) acc1: 100.0000 (96.1364) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [ 540/1403] eta: 0:01:27 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0588 (0.1539) acc1: 100.0000 (96.2078) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 550/1403] eta: 0:01:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0597 (0.1526) acc1: 100.0000 (96.2625) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 560/1403] eta: 0:01:24 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0842 (0.1516) acc1: 98.4375 (96.3012) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 570/1403] eta: 0:01:23 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0848 (0.1505) acc1: 98.4375 (96.3469) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 580/1403] eta: 0:01:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0924 (0.1499) acc1: 98.4375 (96.3694) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 590/1403] eta: 0:01:21 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1063 (0.1495) acc1: 96.8750 (96.3700) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 600/1403] eta: 0:01:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1253 (0.1498) acc1: 95.3125 (96.3238) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 610/1403] eta: 0:01:19 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1253 (0.1494) acc1: 95.3125 (96.3329) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 620/1403] eta: 0:01:18 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1341 (0.1503) acc1: 95.3125 (96.2384) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 630/1403] eta: 0:01:17 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1505 (0.1501) acc1: 93.7500 (96.2262) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 640/1403] eta: 0:01:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1254 (0.1498) acc1: 96.8750 (96.2363) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 650/1403] eta: 0:01:15 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1166 (0.1493) acc1: 96.8750 (96.2558) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 660/1403] eta: 0:01:14 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0876 (0.1482) acc1: 100.0000 (96.3030) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 670/1403] eta: 0:01:13 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0680 (0.1470) acc1: 100.0000 (96.3557) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 680/1403] eta: 0:01:12 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0788 (0.1481) acc1: 98.4375 (96.2785) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 690/1403] eta: 0:01:11 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2506 (0.1518) acc1: 89.0625 (96.0497) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 700/1403] eta: 0:01:10 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1639 (0.1519) acc1: 93.7500 (96.0347) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 710/1403] eta: 0:01:09 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2728 (0.1560) acc1: 85.9375 (95.8048) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 720/1403] eta: 0:01:08 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2728 (0.1559) acc1: 84.3750 (95.8023) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 730/1403] eta: 0:01:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0660 (0.1548) acc1: 100.0000 (95.8511) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0959 data: 0.0003 max mem: 52126 +Test: [ 740/1403] eta: 0:01:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0716 (0.1540) acc1: 100.0000 (95.8860) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0959 data: 0.0003 max mem: 52126 +Test: [ 750/1403] eta: 0:01:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0988 (0.1554) acc1: 96.8750 (95.7994) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 760/1403] eta: 0:01:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3187 (0.1590) acc1: 81.2500 (95.5609) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [ 770/1403] eta: 0:01:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4473 (0.1635) acc1: 75.0000 (95.2821) acc5: 100.0000 (100.0000) acc1_neg: 68.750 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 780/1403] eta: 0:01:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4014 (0.1659) acc1: 78.1250 (95.1324) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 790/1403] eta: 0:01:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3021 (0.1683) acc1: 84.3750 (95.0103) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 800/1403] eta: 0:00:59 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2308 (0.1686) acc1: 90.6250 (94.9965) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 810/1403] eta: 0:00:58 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2131 (0.1715) acc1: 90.6250 (94.8597) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 820/1403] eta: 0:00:57 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6266 (0.1804) acc1: 71.8750 (94.4808) acc5: 100.0000 (100.0000) acc1_neg: 40.625 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 830/1403] eta: 0:00:56 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6907 (0.1834) acc1: 68.7500 (94.3272) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 840/1403] eta: 0:00:55 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5195 (0.1917) acc1: 71.8750 (93.9674) acc5: 100.0000 (100.0000) acc1_neg: 50.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 850/1403] eta: 0:00:54 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.0487 (0.2028) acc1: 54.6875 (93.4948) acc5: 100.0000 (100.0000) acc1_neg: 54.688 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 860/1403] eta: 0:00:53 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.0063 (0.2064) acc1: 57.8125 (93.3417) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 870/1403] eta: 0:00:52 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1307 (0.2053) acc1: 95.3125 (93.3858) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 880/1403] eta: 0:00:51 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1016 (0.2042) acc1: 98.4375 (93.4379) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 890/1403] eta: 0:00:50 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0797 (0.2028) acc1: 98.4375 (93.5010) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 900/1403] eta: 0:00:49 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0709 (0.2013) acc1: 100.0000 (93.5696) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 910/1403] eta: 0:00:48 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0749 (0.2004) acc1: 100.0000 (93.6111) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 920/1403] eta: 0:00:47 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0993 (0.1994) acc1: 98.4375 (93.6567) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 930/1403] eta: 0:00:46 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0955 (0.1983) acc1: 98.4375 (93.7114) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 940/1403] eta: 0:00:45 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1081 (0.1984) acc1: 96.8750 (93.6902) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 950/1403] eta: 0:00:44 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2271 (0.1991) acc1: 89.0625 (93.6317) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 960/1403] eta: 0:00:43 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2537 (0.1997) acc1: 89.0625 (93.5923) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 970/1403] eta: 0:00:42 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2281 (0.2005) acc1: 90.6250 (93.5408) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 980/1403] eta: 0:00:41 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2520 (0.2011) acc1: 90.6250 (93.4920) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0959 data: 0.0003 max mem: 52126 +Test: [ 990/1403] eta: 0:00:40 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2376 (0.2015) acc1: 90.6250 (93.4662) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1000/1403] eta: 0:00:39 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2246 (0.2021) acc1: 90.6250 (93.4300) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1010/1403] eta: 0:00:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2479 (0.2028) acc1: 90.6250 (93.3945) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [1020/1403] eta: 0:00:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3987 (0.2089) acc1: 82.8125 (93.0552) acc5: 100.0000 (100.0000) acc1_neg: 37.500 acc5_neg: 100.000 time: 0.0959 data: 0.0003 max mem: 52126 +Test: [1030/1403] eta: 0:00:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5551 (0.2107) acc1: 60.9375 (92.9362) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0959 data: 0.0003 max mem: 52126 +Test: [1040/1403] eta: 0:00:35 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2161 (0.2105) acc1: 92.1875 (92.9380) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0959 data: 0.0003 max mem: 52126 +Test: [1050/1403] eta: 0:00:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1678 (0.2102) acc1: 93.7500 (92.9561) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0959 data: 0.0003 max mem: 52126 +Test: [1060/1403] eta: 0:00:33 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1362 (0.2094) acc1: 96.8750 (92.9945) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [1070/1403] eta: 0:00:32 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1362 (0.2099) acc1: 96.8750 (92.9782) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1080/1403] eta: 0:00:31 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3453 (0.2116) acc1: 85.9375 (92.8871) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1090/1403] eta: 0:00:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3979 (0.2134) acc1: 82.8125 (92.8033) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0959 data: 0.0003 max mem: 52126 +Test: [1100/1403] eta: 0:00:29 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3277 (0.2138) acc1: 85.9375 (92.7835) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0959 data: 0.0003 max mem: 52126 +Test: [1110/1403] eta: 0:00:28 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2510 (0.2141) acc1: 90.6250 (92.7712) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0960 data: 0.0002 max mem: 52126 +Test: [1120/1403] eta: 0:00:27 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2548 (0.2145) acc1: 90.6250 (92.7478) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0960 data: 0.0003 max mem: 52126 +Test: [1130/1403] eta: 0:00:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2594 (0.2151) acc1: 90.6250 (92.7221) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0959 data: 0.0003 max mem: 52126 +Test: [1140/1403] eta: 0:00:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2246 (0.2152) acc1: 90.6250 (92.7175) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1150/1403] eta: 0:00:24 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2145 (0.2154) acc1: 90.6250 (92.6979) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1160/1403] eta: 0:00:23 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2494 (0.2158) acc1: 90.6250 (92.6774) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [1170/1403] eta: 0:00:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2356 (0.2156) acc1: 90.6250 (92.6852) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1180/1403] eta: 0:00:21 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0689 (0.2144) acc1: 100.0000 (92.7471) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [1190/1403] eta: 0:00:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0682 (0.2132) acc1: 100.0000 (92.8080) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1200/1403] eta: 0:00:19 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0687 (0.2120) acc1: 100.0000 (92.8679) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0957 data: 0.0002 max mem: 52126 +Test: [1210/1403] eta: 0:00:18 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0677 (0.2108) acc1: 100.0000 (92.9268) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0957 data: 0.0002 max mem: 52126 +Test: [1220/1403] eta: 0:00:17 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0676 (0.2096) acc1: 100.0000 (92.9847) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [1230/1403] eta: 0:00:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0681 (0.2085) acc1: 100.0000 (93.0417) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [1240/1403] eta: 0:00:15 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0681 (0.2073) acc1: 100.0000 (93.0978) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1250/1403] eta: 0:00:14 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0682 (0.2063) acc1: 100.0000 (93.1530) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [1260/1403] eta: 0:00:14 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0746 (0.2052) acc1: 100.0000 (93.2073) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [1270/1403] eta: 0:00:13 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0758 (0.2042) acc1: 100.0000 (93.2607) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0957 data: 0.0002 max mem: 52126 +Test: [1280/1403] eta: 0:00:12 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0757 (0.2032) acc1: 100.0000 (93.3133) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1290/1403] eta: 0:00:11 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0753 (0.2022) acc1: 100.0000 (93.3651) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [1300/1403] eta: 0:00:10 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0759 (0.2013) acc1: 100.0000 (93.4161) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [1310/1403] eta: 0:00:09 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0763 (0.2003) acc1: 100.0000 (93.4663) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1320/1403] eta: 0:00:08 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0746 (0.1994) acc1: 100.0000 (93.5158) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1330/1403] eta: 0:00:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0741 (0.1984) acc1: 100.0000 (93.5622) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [1340/1403] eta: 0:00:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0677 (0.1974) acc1: 100.0000 (93.6078) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [1350/1403] eta: 0:00:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0655 (0.1965) acc1: 100.0000 (93.6517) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1360/1403] eta: 0:00:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0655 (0.1956) acc1: 100.0000 (93.6960) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1370/1403] eta: 0:00:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0671 (0.1947) acc1: 100.0000 (93.7397) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [1380/1403] eta: 0:00:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0650 (0.1937) acc1: 100.0000 (93.7851) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0960 data: 0.0002 max mem: 52126 +Test: [1390/1403] eta: 0:00:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0646 (0.1928) acc1: 100.0000 (93.8298) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1400/1403] eta: 0:00:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0661 (0.1920) acc1: 100.0000 (93.8716) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0957 data: 0.0001 max mem: 52126 +Test: [1402/1403] eta: 0:00:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0661 (0.1918) acc1: 100.0000 (93.8760) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0916 data: 0.0001 max mem: 52126 +Test: Total time: 0:02:17 (0.0981 s / it) +* Overall Acc@1 93.876 Acc@5 100.000 + target == real Acc@1 96.102 Acc@5 100.000 + target == fake Acc@1 93.249 Acc@5 100.000 + loss 0.192 +Accuracy of the network on the 89729 test images: 93.9% +Max accuracy: 93.91% +Epoch: [27] [ 0/3634] eta: 6:17:08 lr: 0.000003 loss: 0.5502 (0.5502) time: 6.2268 data: 4.4761 max mem: 52126 +Epoch: [27] [ 10/3634] eta: 1:08:31 lr: 0.000003 loss: 0.5060 (0.4946) time: 1.1345 data: 0.4071 max mem: 52126 +Epoch: [27] [ 20/3634] eta: 0:52:18 lr: 0.000003 loss: 0.5288 (0.5249) time: 0.6006 data: 0.0001 max mem: 52126 +Epoch: [27] [ 30/3634] eta: 0:42:44 lr: 0.000003 loss: 0.5191 (0.5131) time: 0.4791 data: 0.0002 max mem: 52126 +Epoch: [27] [ 40/3634] eta: 0:41:06 lr: 0.000003 loss: 0.5138 (0.5121) time: 0.4947 data: 0.0002 max mem: 52126 +Epoch: [27] [ 50/3634] eta: 0:42:52 lr: 0.000003 loss: 0.5251 (0.5116) time: 0.7270 data: 0.0002 max mem: 52126 +Epoch: [27] [ 60/3634] eta: 0:43:56 lr: 0.000003 loss: 0.5251 (0.5137) time: 0.8430 data: 0.0002 max mem: 52126 +Epoch: [27] [ 70/3634] eta: 0:42:38 lr: 0.000003 loss: 0.5701 (0.5219) time: 0.7186 data: 0.0002 max mem: 52126 +Epoch: [27] [ 80/3634] eta: 0:40:04 lr: 0.000003 loss: 0.5681 (0.5243) time: 0.4899 data: 0.0002 max mem: 52126 +Epoch: [27] [ 90/3634] eta: 0:38:41 lr: 0.000003 loss: 0.5164 (0.5243) time: 0.4316 data: 0.0002 max mem: 52126 +Epoch: [27] [ 100/3634] eta: 0:37:47 lr: 0.000003 loss: 0.5037 (0.5227) time: 0.5002 data: 0.0002 max mem: 52126 +Epoch: [27] [ 110/3634] eta: 0:36:47 lr: 0.000003 loss: 0.4755 (0.5180) time: 0.4965 data: 0.0002 max mem: 52126 +Epoch: [27] [ 120/3634] eta: 0:35:30 lr: 0.000003 loss: 0.5124 (0.5202) time: 0.4282 data: 0.0002 max mem: 52126 +Epoch: [27] [ 130/3634] eta: 0:34:24 lr: 0.000003 loss: 0.5401 (0.5183) time: 0.3815 data: 0.0002 max mem: 52126 +Epoch: [27] [ 140/3634] eta: 0:33:26 lr: 0.000003 loss: 0.5286 (0.5171) time: 0.3814 data: 0.0002 max mem: 52126 +Epoch: [27] [ 150/3634] eta: 0:32:36 lr: 0.000003 loss: 0.5030 (0.5165) time: 0.3824 data: 0.0002 max mem: 52126 +Epoch: [27] [ 160/3634] eta: 0:32:38 lr: 0.000003 loss: 0.4796 (0.5112) time: 0.4899 data: 0.0002 max mem: 52126 +Epoch: [27] [ 170/3634] eta: 0:31:56 lr: 0.000003 loss: 0.4455 (0.5105) time: 0.4894 data: 0.0002 max mem: 52126 +Epoch: [27] [ 180/3634] eta: 0:31:18 lr: 0.000003 loss: 0.5152 (0.5113) time: 0.3820 data: 0.0002 max mem: 52126 +Epoch: [27] [ 190/3634] eta: 0:30:43 lr: 0.000003 loss: 0.5178 (0.5108) time: 0.3822 data: 0.0002 max mem: 52126 +Epoch: [27] [ 200/3634] eta: 0:30:28 lr: 0.000003 loss: 0.5054 (0.5108) time: 0.4290 data: 0.0002 max mem: 52126 +Epoch: [27] [ 210/3634] eta: 0:30:18 lr: 0.000003 loss: 0.5183 (0.5123) time: 0.4918 data: 0.0002 max mem: 52126 +Epoch: [27] [ 220/3634] eta: 0:30:39 lr: 0.000003 loss: 0.5089 (0.5127) time: 0.6022 data: 0.0002 max mem: 52126 +Epoch: [27] [ 230/3634] eta: 0:30:28 lr: 0.000003 loss: 0.4981 (0.5110) time: 0.6006 data: 0.0002 max mem: 52126 +Epoch: [27] [ 240/3634] eta: 0:30:18 lr: 0.000003 loss: 0.4610 (0.5101) time: 0.5051 data: 0.0002 max mem: 52126 +Epoch: [27] [ 250/3634] eta: 0:30:09 lr: 0.000003 loss: 0.5169 (0.5106) time: 0.5044 data: 0.0002 max mem: 52126 +Epoch: [27] [ 260/3634] eta: 0:29:43 lr: 0.000003 loss: 0.5335 (0.5120) time: 0.4426 data: 0.0002 max mem: 52126 +Epoch: [27] [ 270/3634] eta: 0:29:32 lr: 0.000003 loss: 0.5335 (0.5120) time: 0.4286 data: 0.0002 max mem: 52126 +Epoch: [27] [ 280/3634] eta: 0:29:09 lr: 0.000003 loss: 0.4980 (0.5121) time: 0.4300 data: 0.0002 max mem: 52126 +Epoch: [27] [ 290/3634] eta: 0:28:48 lr: 0.000003 loss: 0.4827 (0.5117) time: 0.3834 data: 0.0002 max mem: 52126 +Epoch: [27] [ 300/3634] eta: 0:28:28 lr: 0.000003 loss: 0.5104 (0.5123) time: 0.3818 data: 0.0002 max mem: 52126 +Epoch: [27] [ 310/3634] eta: 0:28:22 lr: 0.000003 loss: 0.5166 (0.5129) time: 0.4441 data: 0.0002 max mem: 52126 +Epoch: [27] [ 320/3634] eta: 0:28:14 lr: 0.000003 loss: 0.5134 (0.5127) time: 0.4925 data: 0.0002 max mem: 52126 +Epoch: [27] [ 330/3634] eta: 0:28:05 lr: 0.000003 loss: 0.5314 (0.5134) time: 0.4757 data: 0.0002 max mem: 52126 +Epoch: [27] [ 340/3634] eta: 0:27:56 lr: 0.000003 loss: 0.5366 (0.5142) time: 0.4747 data: 0.0002 max mem: 52126 +Epoch: [27] [ 350/3634] eta: 0:27:51 lr: 0.000003 loss: 0.5267 (0.5135) time: 0.4899 data: 0.0002 max mem: 52126 +Epoch: [27] [ 360/3634] eta: 0:27:34 lr: 0.000003 loss: 0.4770 (0.5129) time: 0.4428 data: 0.0002 max mem: 52126 +Epoch: [27] [ 370/3634] eta: 0:27:48 lr: 0.000003 loss: 0.5118 (0.5130) time: 0.5540 data: 0.0002 max mem: 52126 +Epoch: [27] [ 380/3634] eta: 0:27:43 lr: 0.000003 loss: 0.5191 (0.5133) time: 0.6165 data: 0.0002 max mem: 52126 +Epoch: [27] [ 390/3634] eta: 0:27:27 lr: 0.000003 loss: 0.5231 (0.5132) time: 0.4445 data: 0.0001 max mem: 52126 +Epoch: [27] [ 400/3634] eta: 0:27:29 lr: 0.000003 loss: 0.4983 (0.5126) time: 0.4879 data: 0.0002 max mem: 52126 +Epoch: [27] [ 410/3634] eta: 0:27:24 lr: 0.000003 loss: 0.4983 (0.5135) time: 0.5500 data: 0.0002 max mem: 52126 +Epoch: [27] [ 420/3634] eta: 0:27:25 lr: 0.000003 loss: 0.5245 (0.5135) time: 0.5524 data: 0.0002 max mem: 52126 +Epoch: [27] [ 430/3634] eta: 0:27:10 lr: 0.000003 loss: 0.4984 (0.5135) time: 0.4894 data: 0.0002 max mem: 52126 +Epoch: [27] [ 440/3634] eta: 0:27:05 lr: 0.000003 loss: 0.5010 (0.5134) time: 0.4421 data: 0.0002 max mem: 52126 +Epoch: [27] [ 450/3634] eta: 0:26:51 lr: 0.000003 loss: 0.5374 (0.5139) time: 0.4428 data: 0.0002 max mem: 52126 +Epoch: [27] [ 460/3634] eta: 0:26:44 lr: 0.000003 loss: 0.5456 (0.5146) time: 0.4299 data: 0.0002 max mem: 52126 +Epoch: [27] [ 470/3634] eta: 0:26:39 lr: 0.000003 loss: 0.5525 (0.5152) time: 0.4921 data: 0.0002 max mem: 52126 +Epoch: [27] [ 480/3634] eta: 0:26:34 lr: 0.000003 loss: 0.5572 (0.5155) time: 0.5053 data: 0.0002 max mem: 52126 +Epoch: [27] [ 490/3634] eta: 0:26:34 lr: 0.000003 loss: 0.5393 (0.5156) time: 0.5459 data: 0.0002 max mem: 52126 +Epoch: [27] [ 500/3634] eta: 0:26:36 lr: 0.000003 loss: 0.5057 (0.5155) time: 0.6060 data: 0.0002 max mem: 52126 +Epoch: [27] [ 510/3634] eta: 0:26:29 lr: 0.000003 loss: 0.5407 (0.5153) time: 0.5481 data: 0.0002 max mem: 52126 +Epoch: [27] [ 520/3634] eta: 0:26:33 lr: 0.000003 loss: 0.5073 (0.5150) time: 0.5706 data: 0.0002 max mem: 52126 +Epoch: [27] [ 530/3634] eta: 0:26:28 lr: 0.000003 loss: 0.4824 (0.5143) time: 0.5879 data: 0.0002 max mem: 52126 +Epoch: [27] [ 540/3634] eta: 0:26:34 lr: 0.000003 loss: 0.5186 (0.5146) time: 0.6086 data: 0.0002 max mem: 52126 +Epoch: [27] [ 550/3634] eta: 0:26:33 lr: 0.000003 loss: 0.5186 (0.5144) time: 0.6528 data: 0.0002 max mem: 52126 +Epoch: [27] [ 560/3634] eta: 0:26:34 lr: 0.000003 loss: 0.5410 (0.5150) time: 0.6064 data: 0.0001 max mem: 52126 +Epoch: [27] [ 570/3634] eta: 0:26:21 lr: 0.000003 loss: 0.5296 (0.5144) time: 0.4999 data: 0.0001 max mem: 52126 +Epoch: [27] [ 580/3634] eta: 0:26:27 lr: 0.000003 loss: 0.5115 (0.5146) time: 0.5493 data: 0.0002 max mem: 52126 +Epoch: [27] [ 590/3634] eta: 0:26:31 lr: 0.000003 loss: 0.5308 (0.5145) time: 0.7116 data: 0.0002 max mem: 52126 +Epoch: [27] [ 600/3634] eta: 0:26:23 lr: 0.000003 loss: 0.4895 (0.5141) time: 0.5888 data: 0.0002 max mem: 52126 +Epoch: [27] [ 610/3634] eta: 0:26:17 lr: 0.000003 loss: 0.4961 (0.5142) time: 0.4857 data: 0.0002 max mem: 52126 +Epoch: [27] [ 620/3634] eta: 0:26:05 lr: 0.000003 loss: 0.5289 (0.5146) time: 0.4405 data: 0.0002 max mem: 52126 +Epoch: [27] [ 630/3634] eta: 0:25:57 lr: 0.000003 loss: 0.5420 (0.5147) time: 0.4266 data: 0.0002 max mem: 52126 +Epoch: [27] [ 640/3634] eta: 0:25:50 lr: 0.000003 loss: 0.5282 (0.5148) time: 0.4759 data: 0.0002 max mem: 52126 +Epoch: [27] [ 650/3634] eta: 0:25:45 lr: 0.000003 loss: 0.5238 (0.5152) time: 0.4922 data: 0.0002 max mem: 52126 +Epoch: [27] [ 660/3634] eta: 0:25:33 lr: 0.000003 loss: 0.5238 (0.5150) time: 0.4422 data: 0.0002 max mem: 52126 +Epoch: [27] [ 670/3634] eta: 0:25:22 lr: 0.000003 loss: 0.5121 (0.5147) time: 0.3805 data: 0.0002 max mem: 52126 +Epoch: [27] [ 680/3634] eta: 0:25:16 lr: 0.000003 loss: 0.5168 (0.5146) time: 0.4398 data: 0.0002 max mem: 52126 +Epoch: [27] [ 690/3634] eta: 0:25:15 lr: 0.000003 loss: 0.5179 (0.5147) time: 0.5479 data: 0.0002 max mem: 52126 +Epoch: [27] [ 700/3634] eta: 0:25:14 lr: 0.000003 loss: 0.5265 (0.5148) time: 0.6082 data: 0.0001 max mem: 52126 +Epoch: [27] [ 710/3634] eta: 0:25:03 lr: 0.000003 loss: 0.5369 (0.5152) time: 0.5005 data: 0.0002 max mem: 52126 +Epoch: [27] [ 720/3634] eta: 0:25:06 lr: 0.000003 loss: 0.5471 (0.5153) time: 0.5458 data: 0.0002 max mem: 52126 +Epoch: [27] [ 730/3634] eta: 0:24:59 lr: 0.000003 loss: 0.5092 (0.5151) time: 0.5900 data: 0.0002 max mem: 52126 +Epoch: [27] [ 740/3634] eta: 0:24:53 lr: 0.000003 loss: 0.5581 (0.5157) time: 0.4857 data: 0.0002 max mem: 52126 +Epoch: [27] [ 750/3634] eta: 0:24:46 lr: 0.000003 loss: 0.5604 (0.5158) time: 0.4867 data: 0.0002 max mem: 52126 +Epoch: [27] [ 760/3634] eta: 0:24:40 lr: 0.000003 loss: 0.5613 (0.5165) time: 0.4728 data: 0.0002 max mem: 52126 +Epoch: [27] [ 770/3634] eta: 0:24:30 lr: 0.000003 loss: 0.5388 (0.5162) time: 0.4272 data: 0.0002 max mem: 52126 +Epoch: [27] [ 780/3634] eta: 0:24:20 lr: 0.000003 loss: 0.5115 (0.5162) time: 0.3816 data: 0.0002 max mem: 52126 +Epoch: [27] [ 790/3634] eta: 0:24:14 lr: 0.000003 loss: 0.5054 (0.5157) time: 0.4431 data: 0.0002 max mem: 52126 +Epoch: [27] [ 800/3634] eta: 0:24:08 lr: 0.000003 loss: 0.5288 (0.5161) time: 0.4926 data: 0.0002 max mem: 52126 +Epoch: [27] [ 810/3634] eta: 0:23:58 lr: 0.000003 loss: 0.5147 (0.5158) time: 0.4320 data: 0.0002 max mem: 52126 +Epoch: [27] [ 820/3634] eta: 0:23:52 lr: 0.000003 loss: 0.5053 (0.5162) time: 0.4281 data: 0.0002 max mem: 52126 +Epoch: [27] [ 830/3634] eta: 0:23:49 lr: 0.000003 loss: 0.5282 (0.5161) time: 0.5167 data: 0.0002 max mem: 52126 +Epoch: [27] [ 840/3634] eta: 0:23:50 lr: 0.000003 loss: 0.5243 (0.5166) time: 0.6234 data: 0.0001 max mem: 52126 +Epoch: [27] [ 850/3634] eta: 0:23:40 lr: 0.000003 loss: 0.5634 (0.5169) time: 0.5349 data: 0.0002 max mem: 52126 +Epoch: [27] [ 860/3634] eta: 0:23:35 lr: 0.000003 loss: 0.4961 (0.5162) time: 0.4410 data: 0.0002 max mem: 52126 +Epoch: [27] [ 870/3634] eta: 0:23:26 lr: 0.000003 loss: 0.4730 (0.5161) time: 0.4410 data: 0.0002 max mem: 52126 +Epoch: [27] [ 880/3634] eta: 0:23:17 lr: 0.000003 loss: 0.5199 (0.5163) time: 0.3831 data: 0.0002 max mem: 52126 +Epoch: [27] [ 890/3634] eta: 0:23:11 lr: 0.000003 loss: 0.5037 (0.5159) time: 0.4403 data: 0.0002 max mem: 52126 +Epoch: [27] [ 900/3634] eta: 0:23:02 lr: 0.000003 loss: 0.4976 (0.5158) time: 0.4397 data: 0.0002 max mem: 52126 +Epoch: [27] [ 910/3634] eta: 0:22:57 lr: 0.000003 loss: 0.5311 (0.5159) time: 0.4406 data: 0.0002 max mem: 52126 +Epoch: [27] [ 920/3634] eta: 0:22:52 lr: 0.000003 loss: 0.5307 (0.5159) time: 0.5000 data: 0.0001 max mem: 52126 +Epoch: [27] [ 930/3634] eta: 0:22:46 lr: 0.000003 loss: 0.5106 (0.5156) time: 0.4864 data: 0.0001 max mem: 52126 +Epoch: [27] [ 940/3634] eta: 0:22:41 lr: 0.000003 loss: 0.4821 (0.5156) time: 0.4860 data: 0.0002 max mem: 52126 +Epoch: [27] [ 950/3634] eta: 0:22:32 lr: 0.000003 loss: 0.5162 (0.5158) time: 0.4415 data: 0.0002 max mem: 52126 +Epoch: [27] [ 960/3634] eta: 0:22:26 lr: 0.000003 loss: 0.5215 (0.5161) time: 0.4273 data: 0.0002 max mem: 52126 +Epoch: [27] [ 970/3634] eta: 0:22:18 lr: 0.000003 loss: 0.5259 (0.5162) time: 0.4268 data: 0.0002 max mem: 52126 +Epoch: [27] [ 980/3634] eta: 0:22:10 lr: 0.000003 loss: 0.5145 (0.5160) time: 0.3818 data: 0.0002 max mem: 52126 +Epoch: [27] [ 990/3634] eta: 0:22:17 lr: 0.000003 loss: 0.5165 (0.5164) time: 0.6769 data: 0.0002 max mem: 52126 +Epoch: [27] [1000/3634] eta: 0:22:11 lr: 0.000003 loss: 0.5491 (0.5164) time: 0.7224 data: 0.0002 max mem: 52126 +Epoch: [27] [1010/3634] eta: 0:22:08 lr: 0.000003 loss: 0.5339 (0.5166) time: 0.5344 data: 0.0002 max mem: 52126 +Epoch: [27] [1020/3634] eta: 0:22:03 lr: 0.000003 loss: 0.5339 (0.5167) time: 0.5490 data: 0.0002 max mem: 52126 +Epoch: [27] [1030/3634] eta: 0:22:00 lr: 0.000003 loss: 0.5279 (0.5168) time: 0.5453 data: 0.0001 max mem: 52126 +Epoch: [27] [1040/3634] eta: 0:21:52 lr: 0.000003 loss: 0.5279 (0.5170) time: 0.4854 data: 0.0001 max mem: 52126 +Epoch: [27] [1050/3634] eta: 0:21:54 lr: 0.000003 loss: 0.5166 (0.5168) time: 0.5885 data: 0.0001 max mem: 52126 +Epoch: [27] [1060/3634] eta: 0:21:49 lr: 0.000003 loss: 0.4765 (0.5164) time: 0.6494 data: 0.0002 max mem: 52126 +Epoch: [27] [1070/3634] eta: 0:21:47 lr: 0.000003 loss: 0.4999 (0.5165) time: 0.5761 data: 0.0002 max mem: 52126 +Epoch: [27] [1080/3634] eta: 0:21:39 lr: 0.000003 loss: 0.5377 (0.5166) time: 0.5153 data: 0.0001 max mem: 52126 +Epoch: [27] [1090/3634] eta: 0:21:36 lr: 0.000003 loss: 0.5493 (0.5169) time: 0.4888 data: 0.0002 max mem: 52126 +Epoch: [27] [1100/3634] eta: 0:21:28 lr: 0.000003 loss: 0.5046 (0.5165) time: 0.4885 data: 0.0002 max mem: 52126 +Epoch: [27] [1110/3634] eta: 0:21:20 lr: 0.000003 loss: 0.4780 (0.5163) time: 0.3825 data: 0.0002 max mem: 52126 +Epoch: [27] [1120/3634] eta: 0:21:14 lr: 0.000003 loss: 0.5031 (0.5164) time: 0.4275 data: 0.0002 max mem: 52126 +Epoch: [27] [1130/3634] eta: 0:21:09 lr: 0.000003 loss: 0.5013 (0.5162) time: 0.4769 data: 0.0002 max mem: 52126 +Epoch: [27] [1140/3634] eta: 0:21:08 lr: 0.000003 loss: 0.5208 (0.5165) time: 0.5997 data: 0.0002 max mem: 52126 +Epoch: [27] [1150/3634] eta: 0:21:00 lr: 0.000003 loss: 0.5208 (0.5161) time: 0.5501 data: 0.0002 max mem: 52126 +Epoch: [27] [1160/3634] eta: 0:21:00 lr: 0.000003 loss: 0.5066 (0.5161) time: 0.5470 data: 0.0002 max mem: 52126 +Epoch: [27] [1170/3634] eta: 0:20:56 lr: 0.000003 loss: 0.5131 (0.5163) time: 0.6519 data: 0.0002 max mem: 52126 +Epoch: [27] [1180/3634] eta: 0:20:48 lr: 0.000003 loss: 0.5192 (0.5161) time: 0.4874 data: 0.0002 max mem: 52126 +Epoch: [27] [1190/3634] eta: 0:20:50 lr: 0.000003 loss: 0.4923 (0.5159) time: 0.6066 data: 0.0001 max mem: 52126 +Epoch: [27] [1200/3634] eta: 0:20:42 lr: 0.000003 loss: 0.4923 (0.5160) time: 0.6069 data: 0.0002 max mem: 52126 +Epoch: [27] [1210/3634] eta: 0:20:35 lr: 0.000003 loss: 0.5467 (0.5160) time: 0.3825 data: 0.0002 max mem: 52126 +Epoch: [27] [1220/3634] eta: 0:20:33 lr: 0.000003 loss: 0.5196 (0.5159) time: 0.5351 data: 0.0002 max mem: 52126 +Epoch: [27] [1230/3634] eta: 0:20:30 lr: 0.000003 loss: 0.5138 (0.5159) time: 0.6551 data: 0.0002 max mem: 52126 +Epoch: [27] [1240/3634] eta: 0:20:25 lr: 0.000003 loss: 0.5305 (0.5161) time: 0.5637 data: 0.0002 max mem: 52126 +Epoch: [27] [1250/3634] eta: 0:20:19 lr: 0.000003 loss: 0.5184 (0.5160) time: 0.4888 data: 0.0002 max mem: 52126 +Epoch: [27] [1260/3634] eta: 0:20:14 lr: 0.000003 loss: 0.5355 (0.5161) time: 0.4860 data: 0.0002 max mem: 52126 +Epoch: [27] [1270/3634] eta: 0:20:06 lr: 0.000003 loss: 0.5150 (0.5163) time: 0.4413 data: 0.0002 max mem: 52126 +Epoch: [27] [1280/3634] eta: 0:20:00 lr: 0.000003 loss: 0.5098 (0.5162) time: 0.4277 data: 0.0002 max mem: 52126 +Epoch: [27] [1290/3634] eta: 0:19:55 lr: 0.000003 loss: 0.5176 (0.5163) time: 0.4883 data: 0.0002 max mem: 52126 +Epoch: [27] [1300/3634] eta: 0:19:48 lr: 0.000003 loss: 0.5195 (0.5163) time: 0.4428 data: 0.0002 max mem: 52126 +Epoch: [27] [1310/3634] eta: 0:19:44 lr: 0.000003 loss: 0.5232 (0.5165) time: 0.4784 data: 0.0001 max mem: 52126 +Epoch: [27] [1320/3634] eta: 0:19:40 lr: 0.000003 loss: 0.5232 (0.5164) time: 0.5806 data: 0.0001 max mem: 52126 +Epoch: [27] [1330/3634] eta: 0:19:38 lr: 0.000003 loss: 0.5159 (0.5164) time: 0.6319 data: 0.0001 max mem: 52126 +Epoch: [27] [1340/3634] eta: 0:19:30 lr: 0.000003 loss: 0.5263 (0.5165) time: 0.5301 data: 0.0001 max mem: 52126 +Epoch: [27] [1350/3634] eta: 0:19:25 lr: 0.000003 loss: 0.5339 (0.5168) time: 0.4321 data: 0.0002 max mem: 52126 +Epoch: [27] [1360/3634] eta: 0:19:20 lr: 0.000003 loss: 0.5463 (0.5167) time: 0.4916 data: 0.0002 max mem: 52126 +Epoch: [27] [1370/3634] eta: 0:19:14 lr: 0.000003 loss: 0.5463 (0.5170) time: 0.5006 data: 0.0002 max mem: 52126 +Epoch: [27] [1380/3634] eta: 0:19:07 lr: 0.000003 loss: 0.5304 (0.5167) time: 0.4410 data: 0.0001 max mem: 52126 +Epoch: [27] [1390/3634] eta: 0:19:03 lr: 0.000003 loss: 0.4474 (0.5162) time: 0.4711 data: 0.0001 max mem: 52126 +Epoch: [27] [1400/3634] eta: 0:18:56 lr: 0.000003 loss: 0.4970 (0.5160) time: 0.4707 data: 0.0001 max mem: 52126 +Epoch: [27] [1410/3634] eta: 0:18:49 lr: 0.000003 loss: 0.4974 (0.5159) time: 0.3811 data: 0.0001 max mem: 52126 +Epoch: [27] [1420/3634] eta: 0:18:42 lr: 0.000003 loss: 0.4957 (0.5157) time: 0.3813 data: 0.0001 max mem: 52126 +Epoch: [27] [1430/3634] eta: 0:18:36 lr: 0.000003 loss: 0.4909 (0.5157) time: 0.4268 data: 0.0002 max mem: 52126 +Epoch: [27] [1440/3634] eta: 0:18:31 lr: 0.000003 loss: 0.5258 (0.5157) time: 0.4766 data: 0.0002 max mem: 52126 +Epoch: [27] [1450/3634] eta: 0:18:25 lr: 0.000003 loss: 0.5034 (0.5153) time: 0.4905 data: 0.0001 max mem: 52126 +Epoch: [27] [1460/3634] eta: 0:18:18 lr: 0.000003 loss: 0.5001 (0.5156) time: 0.4405 data: 0.0001 max mem: 52126 +Epoch: [27] [1470/3634] eta: 0:18:15 lr: 0.000003 loss: 0.5330 (0.5156) time: 0.4992 data: 0.0001 max mem: 52126 +Epoch: [27] [1480/3634] eta: 0:18:08 lr: 0.000003 loss: 0.5009 (0.5155) time: 0.4993 data: 0.0001 max mem: 52126 +Epoch: [27] [1490/3634] eta: 0:18:03 lr: 0.000003 loss: 0.4959 (0.5155) time: 0.4415 data: 0.0002 max mem: 52126 +Epoch: [27] [1500/3634] eta: 0:17:56 lr: 0.000003 loss: 0.5432 (0.5155) time: 0.4413 data: 0.0002 max mem: 52126 +Epoch: [27] [1510/3634] eta: 0:17:50 lr: 0.000003 loss: 0.5256 (0.5154) time: 0.3830 data: 0.0002 max mem: 52126 +Epoch: [27] [1520/3634] eta: 0:17:44 lr: 0.000003 loss: 0.4901 (0.5153) time: 0.4284 data: 0.0002 max mem: 52126 +Epoch: [27] [1530/3634] eta: 0:17:42 lr: 0.000003 loss: 0.5112 (0.5153) time: 0.5931 data: 0.0001 max mem: 52126 +Epoch: [27] [1540/3634] eta: 0:17:38 lr: 0.000003 loss: 0.5219 (0.5152) time: 0.6675 data: 0.0001 max mem: 52126 +Epoch: [27] [1550/3634] eta: 0:17:33 lr: 0.000003 loss: 0.5324 (0.5156) time: 0.5618 data: 0.0002 max mem: 52126 +Epoch: [27] [1560/3634] eta: 0:17:28 lr: 0.000003 loss: 0.5197 (0.5152) time: 0.4906 data: 0.0002 max mem: 52126 +Epoch: [27] [1570/3634] eta: 0:17:24 lr: 0.000003 loss: 0.4851 (0.5151) time: 0.5357 data: 0.0001 max mem: 52126 +Epoch: [27] [1580/3634] eta: 0:17:19 lr: 0.000003 loss: 0.5071 (0.5151) time: 0.5462 data: 0.0001 max mem: 52126 +Epoch: [27] [1590/3634] eta: 0:17:13 lr: 0.000003 loss: 0.5013 (0.5150) time: 0.4856 data: 0.0002 max mem: 52126 +Epoch: [27] [1600/3634] eta: 0:17:07 lr: 0.000003 loss: 0.5432 (0.5153) time: 0.4290 data: 0.0002 max mem: 52126 +Epoch: [27] [1610/3634] eta: 0:17:00 lr: 0.000003 loss: 0.5610 (0.5154) time: 0.3851 data: 0.0002 max mem: 52126 +Epoch: [27] [1620/3634] eta: 0:16:55 lr: 0.000003 loss: 0.5126 (0.5154) time: 0.4271 data: 0.0002 max mem: 52126 +Epoch: [27] [1630/3634] eta: 0:16:50 lr: 0.000003 loss: 0.5338 (0.5155) time: 0.5166 data: 0.0001 max mem: 52126 +Epoch: [27] [1640/3634] eta: 0:16:44 lr: 0.000003 loss: 0.5213 (0.5153) time: 0.4725 data: 0.0001 max mem: 52126 +Epoch: [27] [1650/3634] eta: 0:16:37 lr: 0.000003 loss: 0.4795 (0.5151) time: 0.3824 data: 0.0002 max mem: 52126 +Epoch: [27] [1660/3634] eta: 0:16:32 lr: 0.000003 loss: 0.5264 (0.5154) time: 0.4270 data: 0.0002 max mem: 52126 +Epoch: [27] [1670/3634] eta: 0:16:28 lr: 0.000003 loss: 0.5504 (0.5155) time: 0.5317 data: 0.0002 max mem: 52126 +Epoch: [27] [1680/3634] eta: 0:16:26 lr: 0.000003 loss: 0.5209 (0.5155) time: 0.6644 data: 0.0002 max mem: 52126 +Epoch: [27] [1690/3634] eta: 0:16:22 lr: 0.000003 loss: 0.5184 (0.5155) time: 0.6659 data: 0.0002 max mem: 52126 +Epoch: [27] [1700/3634] eta: 0:16:17 lr: 0.000003 loss: 0.5184 (0.5153) time: 0.5478 data: 0.0001 max mem: 52126 +Epoch: [27] [1710/3634] eta: 0:16:13 lr: 0.000003 loss: 0.4801 (0.5151) time: 0.5462 data: 0.0001 max mem: 52126 +Epoch: [27] [1720/3634] eta: 0:16:06 lr: 0.000003 loss: 0.4947 (0.5151) time: 0.4869 data: 0.0002 max mem: 52126 +Epoch: [27] [1730/3634] eta: 0:16:01 lr: 0.000003 loss: 0.5190 (0.5152) time: 0.4430 data: 0.0002 max mem: 52126 +Epoch: [27] [1740/3634] eta: 0:15:57 lr: 0.000003 loss: 0.5195 (0.5149) time: 0.5473 data: 0.0002 max mem: 52126 +Epoch: [27] [1750/3634] eta: 0:15:51 lr: 0.000003 loss: 0.5177 (0.5150) time: 0.4863 data: 0.0002 max mem: 52126 +Epoch: [27] [1760/3634] eta: 0:15:46 lr: 0.000003 loss: 0.5122 (0.5149) time: 0.4887 data: 0.0002 max mem: 52126 +Epoch: [27] [1770/3634] eta: 0:15:40 lr: 0.000003 loss: 0.4947 (0.5148) time: 0.4889 data: 0.0002 max mem: 52126 +Epoch: [27] [1780/3634] eta: 0:15:36 lr: 0.000003 loss: 0.5072 (0.5148) time: 0.5038 data: 0.0002 max mem: 52126 +Epoch: [27] [1790/3634] eta: 0:15:30 lr: 0.000003 loss: 0.5172 (0.5148) time: 0.5033 data: 0.0002 max mem: 52126 +Epoch: [27] [1800/3634] eta: 0:15:26 lr: 0.000003 loss: 0.5128 (0.5148) time: 0.4850 data: 0.0002 max mem: 52126 +Epoch: [27] [1810/3634] eta: 0:15:22 lr: 0.000003 loss: 0.5078 (0.5147) time: 0.5906 data: 0.0001 max mem: 52126 +Epoch: [27] [1820/3634] eta: 0:15:17 lr: 0.000003 loss: 0.5010 (0.5147) time: 0.5949 data: 0.0001 max mem: 52126 +Epoch: [27] [1830/3634] eta: 0:15:11 lr: 0.000003 loss: 0.4960 (0.5146) time: 0.4899 data: 0.0001 max mem: 52126 +Epoch: [27] [1840/3634] eta: 0:15:06 lr: 0.000003 loss: 0.5170 (0.5147) time: 0.4300 data: 0.0001 max mem: 52126 +Epoch: [27] [1850/3634] eta: 0:15:02 lr: 0.000003 loss: 0.5327 (0.5146) time: 0.5319 data: 0.0002 max mem: 52126 +Epoch: [27] [1860/3634] eta: 0:14:57 lr: 0.000003 loss: 0.5327 (0.5148) time: 0.5881 data: 0.0002 max mem: 52126 +Epoch: [27] [1870/3634] eta: 0:14:53 lr: 0.000003 loss: 0.5317 (0.5147) time: 0.5913 data: 0.0002 max mem: 52126 +Epoch: [27] [1880/3634] eta: 0:14:49 lr: 0.000003 loss: 0.5161 (0.5147) time: 0.5899 data: 0.0002 max mem: 52126 +Epoch: [27] [1890/3634] eta: 0:14:43 lr: 0.000003 loss: 0.5161 (0.5146) time: 0.5292 data: 0.0002 max mem: 52126 +Epoch: [27] [1900/3634] eta: 0:14:39 lr: 0.000003 loss: 0.4696 (0.5145) time: 0.5300 data: 0.0002 max mem: 52126 +Epoch: [27] [1910/3634] eta: 0:14:34 lr: 0.000003 loss: 0.5313 (0.5147) time: 0.5494 data: 0.0002 max mem: 52126 +Epoch: [27] [1920/3634] eta: 0:14:28 lr: 0.000003 loss: 0.5137 (0.5146) time: 0.4462 data: 0.0002 max mem: 52126 +Epoch: [27] [1930/3634] eta: 0:14:22 lr: 0.000003 loss: 0.4979 (0.5144) time: 0.3825 data: 0.0002 max mem: 52126 +Epoch: [27] [1940/3634] eta: 0:14:17 lr: 0.000003 loss: 0.5045 (0.5145) time: 0.4424 data: 0.0002 max mem: 52126 +Epoch: [27] [1950/3634] eta: 0:14:10 lr: 0.000003 loss: 0.4942 (0.5145) time: 0.4426 data: 0.0002 max mem: 52126 +Epoch: [27] [1960/3634] eta: 0:14:05 lr: 0.000003 loss: 0.4925 (0.5144) time: 0.4416 data: 0.0001 max mem: 52126 +Epoch: [27] [1970/3634] eta: 0:13:59 lr: 0.000003 loss: 0.5442 (0.5146) time: 0.4412 data: 0.0001 max mem: 52126 +Epoch: [27] [1980/3634] eta: 0:13:53 lr: 0.000003 loss: 0.5588 (0.5148) time: 0.3821 data: 0.0002 max mem: 52126 +Epoch: [27] [1990/3634] eta: 0:13:49 lr: 0.000003 loss: 0.5316 (0.5148) time: 0.4721 data: 0.0002 max mem: 52126 +Epoch: [27] [2000/3634] eta: 0:13:43 lr: 0.000003 loss: 0.4698 (0.5146) time: 0.5187 data: 0.0002 max mem: 52126 +Epoch: [27] [2010/3634] eta: 0:13:38 lr: 0.000003 loss: 0.5098 (0.5146) time: 0.4869 data: 0.0002 max mem: 52126 +Epoch: [27] [2020/3634] eta: 0:13:33 lr: 0.000003 loss: 0.4880 (0.5145) time: 0.4990 data: 0.0001 max mem: 52126 +Epoch: [27] [2030/3634] eta: 0:13:28 lr: 0.000003 loss: 0.4880 (0.5145) time: 0.4901 data: 0.0001 max mem: 52126 +Epoch: [27] [2040/3634] eta: 0:13:24 lr: 0.000003 loss: 0.5280 (0.5146) time: 0.5366 data: 0.0001 max mem: 52126 +Epoch: [27] [2050/3634] eta: 0:13:19 lr: 0.000003 loss: 0.5744 (0.5148) time: 0.5904 data: 0.0001 max mem: 52126 +Epoch: [27] [2060/3634] eta: 0:13:15 lr: 0.000003 loss: 0.5344 (0.5145) time: 0.5915 data: 0.0001 max mem: 52126 +Epoch: [27] [2070/3634] eta: 0:13:11 lr: 0.000003 loss: 0.4955 (0.5145) time: 0.6371 data: 0.0001 max mem: 52126 +Epoch: [27] [2080/3634] eta: 0:13:06 lr: 0.000003 loss: 0.5333 (0.5146) time: 0.5897 data: 0.0001 max mem: 52126 +Epoch: [27] [2090/3634] eta: 0:13:00 lr: 0.000003 loss: 0.5296 (0.5145) time: 0.4415 data: 0.0002 max mem: 52126 +Epoch: [27] [2100/3634] eta: 0:12:54 lr: 0.000003 loss: 0.5002 (0.5145) time: 0.3824 data: 0.0002 max mem: 52126 +Epoch: [27] [2110/3634] eta: 0:12:50 lr: 0.000003 loss: 0.5201 (0.5146) time: 0.5011 data: 0.0001 max mem: 52126 +Epoch: [27] [2120/3634] eta: 0:12:45 lr: 0.000003 loss: 0.5181 (0.5145) time: 0.6064 data: 0.0001 max mem: 52126 +Epoch: [27] [2130/3634] eta: 0:12:40 lr: 0.000003 loss: 0.5129 (0.5144) time: 0.5312 data: 0.0001 max mem: 52126 +Epoch: [27] [2140/3634] eta: 0:12:35 lr: 0.000003 loss: 0.5392 (0.5143) time: 0.5172 data: 0.0002 max mem: 52126 +Epoch: [27] [2150/3634] eta: 0:12:30 lr: 0.000003 loss: 0.5222 (0.5143) time: 0.5200 data: 0.0002 max mem: 52126 +Epoch: [27] [2160/3634] eta: 0:12:24 lr: 0.000003 loss: 0.5222 (0.5143) time: 0.4292 data: 0.0002 max mem: 52126 +Epoch: [27] [2170/3634] eta: 0:12:20 lr: 0.000003 loss: 0.5163 (0.5144) time: 0.4881 data: 0.0001 max mem: 52126 +Epoch: [27] [2180/3634] eta: 0:12:15 lr: 0.000003 loss: 0.5033 (0.5144) time: 0.5462 data: 0.0001 max mem: 52126 +Epoch: [27] [2190/3634] eta: 0:12:09 lr: 0.000003 loss: 0.5161 (0.5142) time: 0.4404 data: 0.0001 max mem: 52126 +Epoch: [27] [2200/3634] eta: 0:12:05 lr: 0.000003 loss: 0.5104 (0.5142) time: 0.5033 data: 0.0002 max mem: 52126 +Epoch: [27] [2210/3634] eta: 0:12:00 lr: 0.000003 loss: 0.5237 (0.5143) time: 0.5632 data: 0.0001 max mem: 52126 +Epoch: [27] [2220/3634] eta: 0:11:54 lr: 0.000003 loss: 0.5566 (0.5145) time: 0.5008 data: 0.0001 max mem: 52126 +Epoch: [27] [2230/3634] eta: 0:11:49 lr: 0.000003 loss: 0.5543 (0.5146) time: 0.4407 data: 0.0001 max mem: 52126 +Epoch: [27] [2240/3634] eta: 0:11:43 lr: 0.000003 loss: 0.5531 (0.5149) time: 0.4283 data: 0.0002 max mem: 52126 +Epoch: [27] [2250/3634] eta: 0:11:39 lr: 0.000003 loss: 0.5420 (0.5148) time: 0.5477 data: 0.0001 max mem: 52126 +Epoch: [27] [2260/3634] eta: 0:11:35 lr: 0.000003 loss: 0.4512 (0.5146) time: 0.6701 data: 0.0001 max mem: 52126 +Epoch: [27] [2270/3634] eta: 0:11:29 lr: 0.000003 loss: 0.4894 (0.5146) time: 0.5515 data: 0.0001 max mem: 52126 +Epoch: [27] [2280/3634] eta: 0:11:24 lr: 0.000003 loss: 0.5347 (0.5145) time: 0.4441 data: 0.0001 max mem: 52126 +Epoch: [27] [2290/3634] eta: 0:11:20 lr: 0.000003 loss: 0.5090 (0.5145) time: 0.5631 data: 0.0001 max mem: 52126 +Epoch: [27] [2300/3634] eta: 0:11:16 lr: 0.000003 loss: 0.5090 (0.5145) time: 0.6204 data: 0.0001 max mem: 52126 +Epoch: [27] [2310/3634] eta: 0:11:10 lr: 0.000003 loss: 0.5247 (0.5146) time: 0.5462 data: 0.0001 max mem: 52126 +Epoch: [27] [2320/3634] eta: 0:11:06 lr: 0.000003 loss: 0.5456 (0.5147) time: 0.5464 data: 0.0001 max mem: 52126 +Epoch: [27] [2330/3634] eta: 0:11:01 lr: 0.000003 loss: 0.4933 (0.5145) time: 0.5448 data: 0.0001 max mem: 52126 +Epoch: [27] [2340/3634] eta: 0:10:55 lr: 0.000003 loss: 0.5144 (0.5146) time: 0.4264 data: 0.0002 max mem: 52126 +Epoch: [27] [2350/3634] eta: 0:10:49 lr: 0.000003 loss: 0.5444 (0.5146) time: 0.3829 data: 0.0002 max mem: 52126 +Epoch: [27] [2360/3634] eta: 0:10:44 lr: 0.000003 loss: 0.4805 (0.5145) time: 0.4271 data: 0.0001 max mem: 52126 +Epoch: [27] [2370/3634] eta: 0:10:39 lr: 0.000003 loss: 0.4546 (0.5144) time: 0.5342 data: 0.0002 max mem: 52126 +Epoch: [27] [2380/3634] eta: 0:10:34 lr: 0.000003 loss: 0.5134 (0.5144) time: 0.4904 data: 0.0002 max mem: 52126 +Epoch: [27] [2390/3634] eta: 0:10:28 lr: 0.000003 loss: 0.5112 (0.5144) time: 0.4286 data: 0.0001 max mem: 52126 +Epoch: [27] [2400/3634] eta: 0:10:23 lr: 0.000003 loss: 0.5098 (0.5143) time: 0.4726 data: 0.0001 max mem: 52126 +Epoch: [27] [2410/3634] eta: 0:10:17 lr: 0.000003 loss: 0.5137 (0.5144) time: 0.4270 data: 0.0001 max mem: 52126 +Epoch: [27] [2420/3634] eta: 0:10:12 lr: 0.000003 loss: 0.5137 (0.5144) time: 0.3824 data: 0.0001 max mem: 52126 +Epoch: [27] [2430/3634] eta: 0:10:06 lr: 0.000003 loss: 0.5020 (0.5144) time: 0.3825 data: 0.0001 max mem: 52126 +Epoch: [27] [2440/3634] eta: 0:10:01 lr: 0.000003 loss: 0.5374 (0.5145) time: 0.4433 data: 0.0002 max mem: 52126 +Epoch: [27] [2450/3634] eta: 0:09:55 lr: 0.000003 loss: 0.5383 (0.5143) time: 0.4428 data: 0.0002 max mem: 52126 +Epoch: [27] [2460/3634] eta: 0:09:51 lr: 0.000003 loss: 0.5289 (0.5144) time: 0.5316 data: 0.0001 max mem: 52126 +Epoch: [27] [2470/3634] eta: 0:09:47 lr: 0.000003 loss: 0.5288 (0.5144) time: 0.6529 data: 0.0001 max mem: 52126 +Epoch: [27] [2480/3634] eta: 0:09:42 lr: 0.000003 loss: 0.5258 (0.5143) time: 0.5541 data: 0.0002 max mem: 52126 +Epoch: [27] [2490/3634] eta: 0:09:36 lr: 0.000003 loss: 0.5106 (0.5143) time: 0.4774 data: 0.0002 max mem: 52126 +Epoch: [27] [2500/3634] eta: 0:09:31 lr: 0.000003 loss: 0.4995 (0.5142) time: 0.4271 data: 0.0002 max mem: 52126 +Epoch: [27] [2510/3634] eta: 0:09:26 lr: 0.000003 loss: 0.5243 (0.5143) time: 0.4716 data: 0.0001 max mem: 52126 +Epoch: [27] [2520/3634] eta: 0:09:21 lr: 0.000003 loss: 0.5221 (0.5142) time: 0.5291 data: 0.0001 max mem: 52126 +Epoch: [27] [2530/3634] eta: 0:09:16 lr: 0.000003 loss: 0.5026 (0.5141) time: 0.4890 data: 0.0001 max mem: 52126 +Epoch: [27] [2540/3634] eta: 0:09:10 lr: 0.000003 loss: 0.4973 (0.5141) time: 0.4314 data: 0.0002 max mem: 52126 +Epoch: [27] [2550/3634] eta: 0:09:06 lr: 0.000003 loss: 0.5389 (0.5142) time: 0.5634 data: 0.0002 max mem: 52126 +Epoch: [27] [2560/3634] eta: 0:09:01 lr: 0.000003 loss: 0.5260 (0.5141) time: 0.5630 data: 0.0002 max mem: 52126 +Epoch: [27] [2570/3634] eta: 0:08:56 lr: 0.000003 loss: 0.5088 (0.5142) time: 0.5008 data: 0.0002 max mem: 52126 +Epoch: [27] [2580/3634] eta: 0:08:51 lr: 0.000003 loss: 0.5429 (0.5143) time: 0.5500 data: 0.0002 max mem: 52126 +Epoch: [27] [2590/3634] eta: 0:08:46 lr: 0.000003 loss: 0.5325 (0.5143) time: 0.4313 data: 0.0002 max mem: 52126 +Epoch: [27] [2600/3634] eta: 0:08:41 lr: 0.000003 loss: 0.4854 (0.5143) time: 0.5020 data: 0.0002 max mem: 52126 +Epoch: [27] [2610/3634] eta: 0:08:36 lr: 0.000003 loss: 0.5110 (0.5143) time: 0.5485 data: 0.0002 max mem: 52126 +Epoch: [27] [2620/3634] eta: 0:08:30 lr: 0.000003 loss: 0.5297 (0.5143) time: 0.4282 data: 0.0002 max mem: 52126 +Epoch: [27] [2630/3634] eta: 0:08:25 lr: 0.000003 loss: 0.5399 (0.5144) time: 0.4675 data: 0.0002 max mem: 52126 +Epoch: [27] [2640/3634] eta: 0:08:21 lr: 0.000003 loss: 0.5279 (0.5144) time: 0.5757 data: 0.0002 max mem: 52126 +Epoch: [27] [2650/3634] eta: 0:08:16 lr: 0.000003 loss: 0.4842 (0.5142) time: 0.5970 data: 0.0002 max mem: 52126 +Epoch: [27] [2660/3634] eta: 0:08:11 lr: 0.000003 loss: 0.4864 (0.5141) time: 0.4892 data: 0.0002 max mem: 52126 +Epoch: [27] [2670/3634] eta: 0:08:05 lr: 0.000003 loss: 0.5246 (0.5144) time: 0.4279 data: 0.0002 max mem: 52126 +Epoch: [27] [2680/3634] eta: 0:08:00 lr: 0.000003 loss: 0.5390 (0.5144) time: 0.4739 data: 0.0002 max mem: 52126 +Epoch: [27] [2690/3634] eta: 0:07:55 lr: 0.000003 loss: 0.5060 (0.5144) time: 0.4278 data: 0.0002 max mem: 52126 +Epoch: [27] [2700/3634] eta: 0:07:50 lr: 0.000003 loss: 0.4830 (0.5144) time: 0.4441 data: 0.0002 max mem: 52126 +Epoch: [27] [2710/3634] eta: 0:07:45 lr: 0.000003 loss: 0.5482 (0.5145) time: 0.5040 data: 0.0002 max mem: 52126 +Epoch: [27] [2720/3634] eta: 0:07:40 lr: 0.000003 loss: 0.5508 (0.5146) time: 0.5504 data: 0.0002 max mem: 52126 +Epoch: [27] [2730/3634] eta: 0:07:36 lr: 0.000003 loss: 0.5310 (0.5146) time: 0.6693 data: 0.0002 max mem: 52126 +Epoch: [27] [2740/3634] eta: 0:07:30 lr: 0.000003 loss: 0.5310 (0.5145) time: 0.5607 data: 0.0002 max mem: 52126 +Epoch: [27] [2750/3634] eta: 0:07:26 lr: 0.000003 loss: 0.5237 (0.5145) time: 0.5325 data: 0.0002 max mem: 52126 +Epoch: [27] [2760/3634] eta: 0:07:20 lr: 0.000003 loss: 0.5244 (0.5146) time: 0.5324 data: 0.0002 max mem: 52126 +Epoch: [27] [2770/3634] eta: 0:07:15 lr: 0.000003 loss: 0.5261 (0.5145) time: 0.3832 data: 0.0002 max mem: 52126 +Epoch: [27] [2780/3634] eta: 0:07:10 lr: 0.000003 loss: 0.5095 (0.5145) time: 0.4414 data: 0.0002 max mem: 52126 +Epoch: [27] [2790/3634] eta: 0:07:06 lr: 0.000003 loss: 0.5136 (0.5145) time: 0.6747 data: 0.0002 max mem: 52126 +Epoch: [27] [2800/3634] eta: 0:07:01 lr: 0.000003 loss: 0.5226 (0.5146) time: 0.7654 data: 0.0002 max mem: 52126 +Epoch: [27] [2810/3634] eta: 0:06:56 lr: 0.000003 loss: 0.5130 (0.5145) time: 0.5315 data: 0.0002 max mem: 52126 +Epoch: [27] [2820/3634] eta: 0:06:51 lr: 0.000003 loss: 0.5056 (0.5145) time: 0.5012 data: 0.0002 max mem: 52126 +Epoch: [27] [2830/3634] eta: 0:06:46 lr: 0.000003 loss: 0.4868 (0.5143) time: 0.5009 data: 0.0002 max mem: 52126 +Epoch: [27] [2840/3634] eta: 0:06:41 lr: 0.000003 loss: 0.5232 (0.5144) time: 0.4421 data: 0.0002 max mem: 52126 +Epoch: [27] [2850/3634] eta: 0:06:35 lr: 0.000003 loss: 0.5130 (0.5142) time: 0.4421 data: 0.0002 max mem: 52126 +Epoch: [27] [2860/3634] eta: 0:06:30 lr: 0.000003 loss: 0.5053 (0.5142) time: 0.3823 data: 0.0002 max mem: 52126 +Epoch: [27] [2870/3634] eta: 0:06:25 lr: 0.000003 loss: 0.5402 (0.5143) time: 0.3823 data: 0.0002 max mem: 52126 +Epoch: [27] [2880/3634] eta: 0:06:20 lr: 0.000003 loss: 0.5394 (0.5143) time: 0.4703 data: 0.0002 max mem: 52126 +Epoch: [27] [2890/3634] eta: 0:06:15 lr: 0.000003 loss: 0.5148 (0.5143) time: 0.5872 data: 0.0002 max mem: 52126 +Epoch: [27] [2900/3634] eta: 0:06:11 lr: 0.000003 loss: 0.5320 (0.5145) time: 0.6782 data: 0.0002 max mem: 52126 +Epoch: [27] [2910/3634] eta: 0:06:05 lr: 0.000003 loss: 0.5345 (0.5144) time: 0.5611 data: 0.0002 max mem: 52126 +Epoch: [27] [2920/3634] eta: 0:06:00 lr: 0.000003 loss: 0.4924 (0.5143) time: 0.4423 data: 0.0002 max mem: 52126 +Epoch: [27] [2930/3634] eta: 0:05:55 lr: 0.000003 loss: 0.4924 (0.5143) time: 0.4426 data: 0.0002 max mem: 52126 +Epoch: [27] [2940/3634] eta: 0:05:49 lr: 0.000003 loss: 0.5022 (0.5142) time: 0.3830 data: 0.0002 max mem: 52126 +Epoch: [27] [2950/3634] eta: 0:05:44 lr: 0.000003 loss: 0.5163 (0.5143) time: 0.4450 data: 0.0002 max mem: 52126 +Epoch: [27] [2960/3634] eta: 0:05:39 lr: 0.000003 loss: 0.5222 (0.5142) time: 0.4448 data: 0.0002 max mem: 52126 +Epoch: [27] [2970/3634] eta: 0:05:34 lr: 0.000003 loss: 0.4802 (0.5142) time: 0.4286 data: 0.0002 max mem: 52126 +Epoch: [27] [2980/3634] eta: 0:05:29 lr: 0.000003 loss: 0.5240 (0.5143) time: 0.4734 data: 0.0002 max mem: 52126 +Epoch: [27] [2990/3634] eta: 0:05:24 lr: 0.000003 loss: 0.5448 (0.5143) time: 0.4750 data: 0.0002 max mem: 52126 +Epoch: [27] [3000/3634] eta: 0:05:19 lr: 0.000003 loss: 0.5448 (0.5143) time: 0.4802 data: 0.0002 max mem: 52126 +Epoch: [27] [3010/3634] eta: 0:05:14 lr: 0.000003 loss: 0.5405 (0.5144) time: 0.5407 data: 0.0002 max mem: 52126 +Epoch: [27] [3020/3634] eta: 0:05:09 lr: 0.000003 loss: 0.4997 (0.5143) time: 0.5805 data: 0.0002 max mem: 52126 +Epoch: [27] [3030/3634] eta: 0:05:04 lr: 0.000003 loss: 0.4912 (0.5143) time: 0.5904 data: 0.0002 max mem: 52126 +Epoch: [27] [3040/3634] eta: 0:04:59 lr: 0.000003 loss: 0.5326 (0.5144) time: 0.4994 data: 0.0002 max mem: 52126 +Epoch: [27] [3050/3634] eta: 0:04:54 lr: 0.000003 loss: 0.5600 (0.5145) time: 0.4861 data: 0.0002 max mem: 52126 +Epoch: [27] [3060/3634] eta: 0:04:49 lr: 0.000003 loss: 0.5381 (0.5146) time: 0.4868 data: 0.0002 max mem: 52126 +Epoch: [27] [3070/3634] eta: 0:04:44 lr: 0.000003 loss: 0.5240 (0.5146) time: 0.4451 data: 0.0002 max mem: 52126 +Epoch: [27] [3080/3634] eta: 0:04:39 lr: 0.000003 loss: 0.5310 (0.5146) time: 0.5038 data: 0.0002 max mem: 52126 +Epoch: [27] [3090/3634] eta: 0:04:34 lr: 0.000003 loss: 0.5416 (0.5146) time: 0.5001 data: 0.0002 max mem: 52126 +Epoch: [27] [3100/3634] eta: 0:04:29 lr: 0.000003 loss: 0.4965 (0.5146) time: 0.4992 data: 0.0002 max mem: 52126 +Epoch: [27] [3110/3634] eta: 0:04:23 lr: 0.000003 loss: 0.4965 (0.5146) time: 0.4872 data: 0.0002 max mem: 52126 +Epoch: [27] [3120/3634] eta: 0:04:18 lr: 0.000003 loss: 0.5215 (0.5146) time: 0.4293 data: 0.0002 max mem: 52126 +Epoch: [27] [3130/3634] eta: 0:04:13 lr: 0.000003 loss: 0.5429 (0.5146) time: 0.4294 data: 0.0002 max mem: 52126 +Epoch: [27] [3140/3634] eta: 0:04:08 lr: 0.000003 loss: 0.5441 (0.5148) time: 0.4878 data: 0.0002 max mem: 52126 +Epoch: [27] [3150/3634] eta: 0:04:03 lr: 0.000003 loss: 0.5222 (0.5146) time: 0.6033 data: 0.0002 max mem: 52126 +Epoch: [27] [3160/3634] eta: 0:03:58 lr: 0.000003 loss: 0.4490 (0.5144) time: 0.6040 data: 0.0002 max mem: 52126 +Epoch: [27] [3170/3634] eta: 0:03:53 lr: 0.000003 loss: 0.5159 (0.5145) time: 0.5444 data: 0.0002 max mem: 52126 +Epoch: [27] [3180/3634] eta: 0:03:48 lr: 0.000003 loss: 0.5355 (0.5146) time: 0.4872 data: 0.0002 max mem: 52126 +Epoch: [27] [3190/3634] eta: 0:03:43 lr: 0.000003 loss: 0.5237 (0.5145) time: 0.5476 data: 0.0002 max mem: 52126 +Epoch: [27] [3200/3634] eta: 0:03:38 lr: 0.000003 loss: 0.5094 (0.5146) time: 0.6052 data: 0.0002 max mem: 52126 +Epoch: [27] [3210/3634] eta: 0:03:33 lr: 0.000003 loss: 0.4957 (0.5145) time: 0.5042 data: 0.0002 max mem: 52126 +Epoch: [27] [3220/3634] eta: 0:03:28 lr: 0.000003 loss: 0.5241 (0.5146) time: 0.5369 data: 0.0002 max mem: 52126 +Epoch: [27] [3230/3634] eta: 0:03:23 lr: 0.000003 loss: 0.5461 (0.5146) time: 0.4737 data: 0.0002 max mem: 52126 +Epoch: [27] [3240/3634] eta: 0:03:18 lr: 0.000003 loss: 0.5009 (0.5146) time: 0.3823 data: 0.0002 max mem: 52126 +Epoch: [27] [3250/3634] eta: 0:03:13 lr: 0.000003 loss: 0.4976 (0.5145) time: 0.3831 data: 0.0002 max mem: 52126 +Epoch: [27] [3260/3634] eta: 0:03:08 lr: 0.000003 loss: 0.4737 (0.5144) time: 0.4423 data: 0.0002 max mem: 52126 +Epoch: [27] [3270/3634] eta: 0:03:03 lr: 0.000003 loss: 0.5085 (0.5145) time: 0.5477 data: 0.0002 max mem: 52126 +Epoch: [27] [3280/3634] eta: 0:02:58 lr: 0.000003 loss: 0.5659 (0.5145) time: 0.5978 data: 0.0002 max mem: 52126 +Epoch: [27] [3290/3634] eta: 0:02:53 lr: 0.000003 loss: 0.5233 (0.5145) time: 0.5992 data: 0.0002 max mem: 52126 +Epoch: [27] [3300/3634] eta: 0:02:48 lr: 0.000003 loss: 0.5224 (0.5145) time: 0.6391 data: 0.0002 max mem: 52126 +Epoch: [27] [3310/3634] eta: 0:02:43 lr: 0.000003 loss: 0.5170 (0.5144) time: 0.6637 data: 0.0002 max mem: 52126 +Epoch: [27] [3320/3634] eta: 0:02:38 lr: 0.000003 loss: 0.5020 (0.5144) time: 0.5137 data: 0.0002 max mem: 52126 +Epoch: [27] [3330/3634] eta: 0:02:33 lr: 0.000003 loss: 0.5125 (0.5144) time: 0.3814 data: 0.0002 max mem: 52126 +Epoch: [27] [3340/3634] eta: 0:02:28 lr: 0.000003 loss: 0.5258 (0.5144) time: 0.3815 data: 0.0002 max mem: 52126 +Epoch: [27] [3350/3634] eta: 0:02:23 lr: 0.000003 loss: 0.4937 (0.5143) time: 0.4301 data: 0.0002 max mem: 52126 +Epoch: [27] [3360/3634] eta: 0:02:18 lr: 0.000003 loss: 0.5127 (0.5143) time: 0.5388 data: 0.0002 max mem: 52126 +Epoch: [27] [3370/3634] eta: 0:02:13 lr: 0.000003 loss: 0.5268 (0.5143) time: 0.4912 data: 0.0002 max mem: 52126 +Epoch: [27] [3380/3634] eta: 0:02:07 lr: 0.000003 loss: 0.5277 (0.5143) time: 0.3831 data: 0.0002 max mem: 52126 +Epoch: [27] [3390/3634] eta: 0:02:02 lr: 0.000003 loss: 0.5277 (0.5143) time: 0.3824 data: 0.0002 max mem: 52126 +Epoch: [27] [3400/3634] eta: 0:01:57 lr: 0.000003 loss: 0.4898 (0.5142) time: 0.4423 data: 0.0002 max mem: 52126 +Epoch: [27] [3410/3634] eta: 0:01:52 lr: 0.000003 loss: 0.5015 (0.5142) time: 0.4865 data: 0.0002 max mem: 52126 +Epoch: [27] [3420/3634] eta: 0:01:47 lr: 0.000003 loss: 0.5015 (0.5142) time: 0.4855 data: 0.0002 max mem: 52126 +Epoch: [27] [3430/3634] eta: 0:01:42 lr: 0.000003 loss: 0.5022 (0.5142) time: 0.4877 data: 0.0002 max mem: 52126 +Epoch: [27] [3440/3634] eta: 0:01:37 lr: 0.000003 loss: 0.5284 (0.5142) time: 0.4292 data: 0.0002 max mem: 52126 +Epoch: [27] [3450/3634] eta: 0:01:32 lr: 0.000003 loss: 0.4890 (0.5141) time: 0.5049 data: 0.0002 max mem: 52126 +Epoch: [27] [3460/3634] eta: 0:01:27 lr: 0.000003 loss: 0.4834 (0.5141) time: 0.6261 data: 0.0002 max mem: 52126 +Epoch: [27] [3470/3634] eta: 0:01:22 lr: 0.000003 loss: 0.4935 (0.5140) time: 0.6099 data: 0.0001 max mem: 52126 +Epoch: [27] [3480/3634] eta: 0:01:17 lr: 0.000003 loss: 0.5368 (0.5141) time: 0.4880 data: 0.0002 max mem: 52126 +Epoch: [27] [3490/3634] eta: 0:01:12 lr: 0.000003 loss: 0.5373 (0.5142) time: 0.4869 data: 0.0002 max mem: 52126 +Epoch: [27] [3500/3634] eta: 0:01:07 lr: 0.000003 loss: 0.5116 (0.5141) time: 0.4873 data: 0.0002 max mem: 52126 +Epoch: [27] [3510/3634] eta: 0:01:02 lr: 0.000003 loss: 0.5065 (0.5141) time: 0.4886 data: 0.0002 max mem: 52126 +Epoch: [27] [3520/3634] eta: 0:00:57 lr: 0.000003 loss: 0.5150 (0.5141) time: 0.5977 data: 0.0002 max mem: 52126 +Epoch: [27] [3530/3634] eta: 0:00:52 lr: 0.000003 loss: 0.5150 (0.5141) time: 0.5508 data: 0.0002 max mem: 52126 +Epoch: [27] [3540/3634] eta: 0:00:47 lr: 0.000003 loss: 0.5153 (0.5141) time: 0.4407 data: 0.0002 max mem: 52126 +Epoch: [27] [3550/3634] eta: 0:00:42 lr: 0.000003 loss: 0.5340 (0.5141) time: 0.3817 data: 0.0002 max mem: 52126 +Epoch: [27] [3560/3634] eta: 0:00:37 lr: 0.000003 loss: 0.5340 (0.5142) time: 0.5341 data: 0.0002 max mem: 52126 +Epoch: [27] [3570/3634] eta: 0:00:32 lr: 0.000003 loss: 0.5247 (0.5142) time: 0.6388 data: 0.0002 max mem: 52126 +Epoch: [27] [3580/3634] eta: 0:00:27 lr: 0.000003 loss: 0.5234 (0.5142) time: 0.4875 data: 0.0002 max mem: 52126 +Epoch: [27] [3590/3634] eta: 0:00:22 lr: 0.000003 loss: 0.5250 (0.5142) time: 0.4870 data: 0.0002 max mem: 52126 +Epoch: [27] [3600/3634] eta: 0:00:17 lr: 0.000003 loss: 0.5083 (0.5142) time: 0.5465 data: 0.0002 max mem: 52126 +Epoch: [27] [3610/3634] eta: 0:00:12 lr: 0.000003 loss: 0.5161 (0.5142) time: 0.5563 data: 0.0002 max mem: 52126 +Epoch: [27] [3620/3634] eta: 0:00:07 lr: 0.000003 loss: 0.5361 (0.5142) time: 0.4999 data: 0.0030 max mem: 52126 +Epoch: [27] [3630/3634] eta: 0:00:02 lr: 0.000003 loss: 0.5470 (0.5143) time: 0.5000 data: 0.0029 max mem: 52126 +Epoch: [27] [3633/3634] eta: 0:00:00 lr: 0.000003 loss: 0.5361 (0.5143) time: 0.4996 data: 0.0029 max mem: 52126 +Epoch: [27] Total time: 0:30:33 (0.5045 s / it) +Averaged stats: lr: 0.000003 loss: 0.5361 (0.5164) +Test: [ 0/1403] eta: 1:07:21 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0775 (0.0775) acc1: 100.0000 (100.0000) acc5: 100.0000 (100.0000) time: 2.8808 data: 2.7843 max mem: 52126 +Test: [ 10/1403] eta: 0:08:06 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1491 (0.1770) acc1: 98.4375 (95.1705) acc5: 100.0000 (100.0000) time: 0.3492 data: 0.2534 max mem: 52126 +Test: [ 20/1403] eta: 0:05:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1318 (0.1547) acc1: 98.4375 (96.2798) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 30/1403] eta: 0:04:15 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0930 (0.1522) acc1: 98.4375 (96.0685) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 40/1403] eta: 0:03:43 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1061 (0.1530) acc1: 98.4375 (96.2652) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 50/1403] eta: 0:03:23 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1198 (0.1756) acc1: 98.4375 (94.7917) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 60/1403] eta: 0:03:10 acc1_pos: 90.625 acc5_pos: 100.000 loss: 0.1210 (0.1721) acc1: 100.0000 (95.2613) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 70/1403] eta: 0:03:00 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.1414 (0.1731) acc1: 96.8750 (95.1805) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 80/1403] eta: 0:02:52 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.1457 (0.1772) acc1: 96.8750 (94.9460) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 90/1403] eta: 0:02:46 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.1756 (0.1968) acc1: 93.7500 (94.0762) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 100/1403] eta: 0:02:41 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1756 (0.1943) acc1: 95.3125 (94.2760) acc5: 100.0000 (100.0000) time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 110/1403] eta: 0:02:36 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1537 (0.1921) acc1: 98.4375 (94.3834) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 120/1403] eta: 0:02:32 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0801 (0.1827) acc1: 100.0000 (94.8347) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 130/1403] eta: 0:02:29 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0772 (0.1759) acc1: 100.0000 (95.1574) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 140/1403] eta: 0:02:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0741 (0.1686) acc1: 100.0000 (95.5009) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 150/1403] eta: 0:02:23 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0728 (0.1625) acc1: 100.0000 (95.7885) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 160/1403] eta: 0:02:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0744 (0.1576) acc1: 100.0000 (96.0016) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 170/1403] eta: 0:02:18 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0767 (0.1537) acc1: 100.0000 (96.2080) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 180/1403] eta: 0:02:16 acc1_pos: 87.500 acc5_pos: 100.000 loss: 0.0914 (0.1585) acc1: 100.0000 (96.0290) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 190/1403] eta: 0:02:14 acc1_pos: 89.062 acc5_pos: 100.000 loss: 0.2815 (0.1667) acc1: 90.6250 (95.5988) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 200/1403] eta: 0:02:12 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.2815 (0.1784) acc1: 89.0625 (95.0715) acc5: 100.0000 (100.0000) time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 210/1403] eta: 0:02:10 acc1_pos: 76.562 acc5_pos: 100.000 loss: 0.3176 (0.1891) acc1: 87.5000 (94.4313) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 220/1403] eta: 0:02:08 acc1_pos: 89.062 acc5_pos: 100.000 loss: 0.3335 (0.1981) acc1: 85.9375 (93.9975) acc5: 100.0000 (100.0000) time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 230/1403] eta: 0:02:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2762 (0.1942) acc1: 90.6250 (94.1964) acc5: 100.0000 (100.0000) time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 240/1403] eta: 0:02:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0833 (0.1897) acc1: 100.0000 (94.4243) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 250/1403] eta: 0:02:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0827 (0.1857) acc1: 100.0000 (94.6277) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 260/1403] eta: 0:02:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0833 (0.1821) acc1: 100.0000 (94.8156) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 270/1403] eta: 0:02:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0839 (0.1787) acc1: 100.0000 (94.9781) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 280/1403] eta: 0:01:59 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0843 (0.1753) acc1: 100.0000 (95.1512) acc5: 100.0000 (100.0000) time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 290/1403] eta: 0:01:57 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0807 (0.1721) acc1: 100.0000 (95.3125) acc5: 100.0000 (100.0000) time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 300/1403] eta: 0:01:56 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0780 (0.1691) acc1: 100.0000 (95.4630) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 310/1403] eta: 0:01:54 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0841 (0.1666) acc1: 100.0000 (95.5939) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 320/1403] eta: 0:01:53 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1011 (0.1647) acc1: 100.0000 (95.7311) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 330/1403] eta: 0:01:52 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1076 (0.1630) acc1: 100.0000 (95.8601) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 340/1403] eta: 0:01:50 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1076 (0.1613) acc1: 100.0000 (95.9769) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 350/1403] eta: 0:01:49 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1054 (0.1598) acc1: 100.0000 (96.0915) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 360/1403] eta: 0:01:48 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1084 (0.1585) acc1: 100.0000 (96.1955) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 370/1403] eta: 0:01:47 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1067 (0.1570) acc1: 100.0000 (96.2980) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 380/1403] eta: 0:01:45 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1056 (0.1557) acc1: 100.0000 (96.3952) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [ 390/1403] eta: 0:01:44 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1070 (0.1564) acc1: 100.0000 (96.3595) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [ 400/1403] eta: 0:01:43 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2447 (0.1597) acc1: 90.6250 (96.1736) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 410/1403] eta: 0:01:42 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2636 (0.1623) acc1: 90.6250 (96.0310) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 420/1403] eta: 0:01:41 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2363 (0.1636) acc1: 92.1875 (95.9397) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 430/1403] eta: 0:01:39 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2270 (0.1661) acc1: 90.6250 (95.7874) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 440/1403] eta: 0:01:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2852 (0.1692) acc1: 89.0625 (95.6314) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 450/1403] eta: 0:01:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2563 (0.1712) acc1: 89.0625 (95.5273) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 460/1403] eta: 0:01:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2425 (0.1733) acc1: 90.6250 (95.3871) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [ 470/1403] eta: 0:01:35 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2211 (0.1724) acc1: 92.1875 (95.4187) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 480/1403] eta: 0:01:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0629 (0.1701) acc1: 100.0000 (95.5139) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 490/1403] eta: 0:01:32 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0627 (0.1679) acc1: 100.0000 (95.6053) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 500/1403] eta: 0:01:31 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0627 (0.1658) acc1: 100.0000 (95.6930) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 510/1403] eta: 0:01:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0625 (0.1638) acc1: 100.0000 (95.7773) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 520/1403] eta: 0:01:29 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0675 (0.1620) acc1: 100.0000 (95.8583) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 530/1403] eta: 0:01:28 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0594 (0.1601) acc1: 100.0000 (95.9363) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 540/1403] eta: 0:01:27 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0583 (0.1582) acc1: 100.0000 (96.0114) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 550/1403] eta: 0:01:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0593 (0.1568) acc1: 100.0000 (96.0725) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 560/1403] eta: 0:01:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0788 (0.1556) acc1: 100.0000 (96.1258) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 570/1403] eta: 0:01:24 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0795 (0.1544) acc1: 100.0000 (96.1827) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52126 +Test: [ 580/1403] eta: 0:01:23 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0846 (0.1537) acc1: 100.0000 (96.2054) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52126 +Test: [ 590/1403] eta: 0:01:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1077 (0.1532) acc1: 98.4375 (96.2088) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [ 600/1403] eta: 0:01:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1193 (0.1533) acc1: 95.3125 (96.1600) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 610/1403] eta: 0:01:19 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1193 (0.1526) acc1: 95.3125 (96.1820) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [ 620/1403] eta: 0:01:18 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1242 (0.1533) acc1: 95.3125 (96.0975) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 630/1403] eta: 0:01:17 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1381 (0.1529) acc1: 95.3125 (96.1024) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 640/1403] eta: 0:01:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1166 (0.1524) acc1: 96.8750 (96.1315) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 650/1403] eta: 0:01:15 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1094 (0.1517) acc1: 98.4375 (96.1646) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [ 660/1403] eta: 0:01:14 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0866 (0.1506) acc1: 100.0000 (96.2155) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [ 670/1403] eta: 0:01:13 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0679 (0.1494) acc1: 100.0000 (96.2696) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [ 680/1403] eta: 0:01:12 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0767 (0.1501) acc1: 98.4375 (96.2257) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 690/1403] eta: 0:01:11 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2196 (0.1532) acc1: 90.6250 (96.0293) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 700/1403] eta: 0:01:10 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1511 (0.1532) acc1: 92.1875 (96.0191) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [ 710/1403] eta: 0:01:09 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2631 (0.1572) acc1: 85.9375 (95.7916) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [ 720/1403] eta: 0:01:08 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2631 (0.1570) acc1: 85.9375 (95.7979) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [ 730/1403] eta: 0:01:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0640 (0.1558) acc1: 100.0000 (95.8469) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 740/1403] eta: 0:01:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0687 (0.1549) acc1: 100.0000 (95.8860) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 750/1403] eta: 0:01:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0964 (0.1559) acc1: 98.4375 (95.8285) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 760/1403] eta: 0:01:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2810 (0.1584) acc1: 89.0625 (95.6780) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 770/1403] eta: 0:01:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3944 (0.1623) acc1: 79.6875 (95.4280) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 780/1403] eta: 0:01:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3623 (0.1645) acc1: 81.2500 (95.2905) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [ 790/1403] eta: 0:01:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2929 (0.1670) acc1: 85.9375 (95.1683) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [ 800/1403] eta: 0:01:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2222 (0.1675) acc1: 92.1875 (95.1447) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [ 810/1403] eta: 0:00:59 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2207 (0.1706) acc1: 89.0625 (94.9985) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [ 820/1403] eta: 0:00:58 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6436 (0.1797) acc1: 71.8750 (94.6102) acc5: 100.0000 (100.0000) acc1_neg: 43.750 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52126 +Test: [ 830/1403] eta: 0:00:57 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.7125 (0.1825) acc1: 70.3125 (94.4645) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52126 +Test: [ 840/1403] eta: 0:00:56 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5251 (0.1906) acc1: 70.3125 (94.1216) acc5: 100.0000 (100.0000) acc1_neg: 51.562 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [ 850/1403] eta: 0:00:55 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.0363 (0.2015) acc1: 54.6875 (93.6564) acc5: 100.0000 (100.0000) acc1_neg: 53.125 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [ 860/1403] eta: 0:00:54 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.9815 (0.2049) acc1: 59.3750 (93.5086) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52126 +Test: [ 870/1403] eta: 0:00:53 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1185 (0.2038) acc1: 96.8750 (93.5598) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [ 880/1403] eta: 0:00:51 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0905 (0.2026) acc1: 98.4375 (93.6134) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 890/1403] eta: 0:00:50 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0770 (0.2012) acc1: 100.0000 (93.6781) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 900/1403] eta: 0:00:49 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0711 (0.1997) acc1: 100.0000 (93.7465) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 910/1403] eta: 0:00:48 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0719 (0.1987) acc1: 100.0000 (93.7894) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 920/1403] eta: 0:00:47 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0867 (0.1976) acc1: 98.4375 (93.8399) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [ 930/1403] eta: 0:00:46 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0803 (0.1964) acc1: 98.4375 (93.8943) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [ 940/1403] eta: 0:00:45 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1004 (0.1963) acc1: 96.8750 (93.8812) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 950/1403] eta: 0:00:44 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1891 (0.1969) acc1: 89.0625 (93.8338) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 960/1403] eta: 0:00:43 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2377 (0.1972) acc1: 89.0625 (93.8037) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 970/1403] eta: 0:00:42 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2125 (0.1978) acc1: 92.1875 (93.7629) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 980/1403] eta: 0:00:41 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2320 (0.1983) acc1: 90.6250 (93.7150) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 990/1403] eta: 0:00:40 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2283 (0.1984) acc1: 90.6250 (93.7059) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [1000/1403] eta: 0:00:39 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2013 (0.1989) acc1: 92.1875 (93.6813) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [1010/1403] eta: 0:00:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2176 (0.1994) acc1: 90.6250 (93.6495) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [1020/1403] eta: 0:00:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3721 (0.2046) acc1: 82.8125 (93.3598) acc5: 100.0000 (100.0000) acc1_neg: 40.625 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [1030/1403] eta: 0:00:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4278 (0.2062) acc1: 68.7500 (93.2620) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52126 +Test: [1040/1403] eta: 0:00:35 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2093 (0.2059) acc1: 92.1875 (93.2667) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [1050/1403] eta: 0:00:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1561 (0.2057) acc1: 95.3125 (93.2817) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [1060/1403] eta: 0:00:33 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1337 (0.2049) acc1: 96.8750 (93.3215) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [1070/1403] eta: 0:00:32 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1337 (0.2053) acc1: 96.8750 (93.3065) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [1080/1403] eta: 0:00:31 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3403 (0.2070) acc1: 85.9375 (93.2152) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [1090/1403] eta: 0:00:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3860 (0.2088) acc1: 84.3750 (93.1284) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [1100/1403] eta: 0:00:29 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3282 (0.2092) acc1: 85.9375 (93.1000) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [1110/1403] eta: 0:00:28 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2490 (0.2096) acc1: 90.6250 (93.0848) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [1120/1403] eta: 0:00:27 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2502 (0.2100) acc1: 90.6250 (93.0656) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [1130/1403] eta: 0:00:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2588 (0.2106) acc1: 90.6250 (93.0344) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [1140/1403] eta: 0:00:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2183 (0.2107) acc1: 90.6250 (93.0228) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [1150/1403] eta: 0:00:24 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2179 (0.2109) acc1: 92.1875 (93.0034) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [1160/1403] eta: 0:00:23 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2370 (0.2113) acc1: 90.6250 (92.9856) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [1170/1403] eta: 0:00:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2272 (0.2112) acc1: 90.6250 (92.9881) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [1180/1403] eta: 0:00:21 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0679 (0.2099) acc1: 100.0000 (93.0475) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [1190/1403] eta: 0:00:21 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0676 (0.2088) acc1: 100.0000 (93.1058) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [1200/1403] eta: 0:00:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0679 (0.2076) acc1: 100.0000 (93.1632) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [1210/1403] eta: 0:00:19 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0667 (0.2064) acc1: 100.0000 (93.2197) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0002 max mem: 52126 +Test: [1220/1403] eta: 0:00:18 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0664 (0.2053) acc1: 100.0000 (93.2752) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0002 max mem: 52126 +Test: [1230/1403] eta: 0:00:17 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0673 (0.2042) acc1: 100.0000 (93.3299) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0002 max mem: 52126 +Test: [1240/1403] eta: 0:00:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0673 (0.2031) acc1: 100.0000 (93.3836) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0002 max mem: 52126 +Test: [1250/1403] eta: 0:00:15 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0675 (0.2020) acc1: 100.0000 (93.4365) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0002 max mem: 52126 +Test: [1260/1403] eta: 0:00:14 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0743 (0.2010) acc1: 100.0000 (93.4886) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0960 data: 0.0002 max mem: 52126 +Test: [1270/1403] eta: 0:00:13 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0761 (0.2000) acc1: 100.0000 (93.5398) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0002 max mem: 52126 +Test: [1280/1403] eta: 0:00:12 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0760 (0.1991) acc1: 100.0000 (93.5902) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0002 max mem: 52126 +Test: [1290/1403] eta: 0:00:11 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0752 (0.1981) acc1: 100.0000 (93.6399) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0002 max mem: 52126 +Test: [1300/1403] eta: 0:00:10 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0763 (0.1972) acc1: 100.0000 (93.6887) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0960 data: 0.0003 max mem: 52126 +Test: [1310/1403] eta: 0:00:09 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0764 (0.1963) acc1: 100.0000 (93.7369) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0002 max mem: 52126 +Test: [1320/1403] eta: 0:00:08 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0748 (0.1953) acc1: 100.0000 (93.7843) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0002 max mem: 52126 +Test: [1330/1403] eta: 0:00:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0740 (0.1944) acc1: 100.0000 (93.8298) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0002 max mem: 52126 +Test: [1340/1403] eta: 0:00:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0667 (0.1935) acc1: 100.0000 (93.8735) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0002 max mem: 52126 +Test: [1350/1403] eta: 0:00:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0648 (0.1925) acc1: 100.0000 (93.9154) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0002 max mem: 52126 +Test: [1360/1403] eta: 0:00:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0648 (0.1916) acc1: 100.0000 (93.9589) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0960 data: 0.0002 max mem: 52126 +Test: [1370/1403] eta: 0:00:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0652 (0.1908) acc1: 100.0000 (93.9996) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0002 max mem: 52126 +Test: [1380/1403] eta: 0:00:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0639 (0.1898) acc1: 100.0000 (94.0430) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0002 max mem: 52126 +Test: [1390/1403] eta: 0:00:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0638 (0.1889) acc1: 100.0000 (94.0859) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0960 data: 0.0002 max mem: 52126 +Test: [1400/1403] eta: 0:00:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0648 (0.1881) acc1: 100.0000 (94.1258) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0959 data: 0.0001 max mem: 52126 +Test: [1402/1403] eta: 0:00:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0651 (0.1880) acc1: 100.0000 (94.1301) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0918 data: 0.0001 max mem: 52126 +Test: Total time: 0:02:18 (0.0986 s / it) +* Overall Acc@1 94.130 Acc@5 100.000 + target == real Acc@1 95.555 Acc@5 100.000 + target == fake Acc@1 93.729 Acc@5 100.000 + loss 0.188 +Accuracy of the network on the 89729 test images: 94.1% +Max accuracy: 94.13% +Epoch: [28] [ 0/3634] eta: 4:25:54 lr: 0.000003 loss: 0.5063 (0.5063) time: 4.3903 data: 4.0163 max mem: 52126 +Epoch: [28] [ 10/3634] eta: 0:50:15 lr: 0.000003 loss: 0.5111 (0.5114) time: 0.8322 data: 0.3653 max mem: 52126 +Epoch: [28] [ 20/3634] eta: 0:46:23 lr: 0.000003 loss: 0.5217 (0.5191) time: 0.5892 data: 0.0001 max mem: 52126 +Epoch: [28] [ 30/3634] eta: 0:40:32 lr: 0.000003 loss: 0.5354 (0.5166) time: 0.5883 data: 0.0002 max mem: 52126 +Epoch: [28] [ 40/3634] eta: 0:41:00 lr: 0.000003 loss: 0.5486 (0.5207) time: 0.5945 data: 0.0002 max mem: 52126 +Epoch: [28] [ 50/3634] eta: 0:38:25 lr: 0.000003 loss: 0.5481 (0.5242) time: 0.5947 data: 0.0002 max mem: 52126 +Epoch: [28] [ 60/3634] eta: 0:36:40 lr: 0.000003 loss: 0.5441 (0.5218) time: 0.4743 data: 0.0002 max mem: 52126 +Epoch: [28] [ 70/3634] eta: 0:34:36 lr: 0.000003 loss: 0.5441 (0.5260) time: 0.4281 data: 0.0002 max mem: 52126 +Epoch: [28] [ 80/3634] eta: 0:33:02 lr: 0.000003 loss: 0.5465 (0.5255) time: 0.3819 data: 0.0002 max mem: 52126 +Epoch: [28] [ 90/3634] eta: 0:31:47 lr: 0.000003 loss: 0.5488 (0.5288) time: 0.3808 data: 0.0002 max mem: 52126 +Epoch: [28] [ 100/3634] eta: 0:30:49 lr: 0.000003 loss: 0.5467 (0.5292) time: 0.3832 data: 0.0002 max mem: 52126 +Epoch: [28] [ 110/3634] eta: 0:31:07 lr: 0.000003 loss: 0.5467 (0.5297) time: 0.4910 data: 0.0001 max mem: 52126 +Epoch: [28] [ 120/3634] eta: 0:31:30 lr: 0.000003 loss: 0.5232 (0.5283) time: 0.6126 data: 0.0001 max mem: 52126 +Epoch: [28] [ 130/3634] eta: 0:31:07 lr: 0.000003 loss: 0.5145 (0.5291) time: 0.5512 data: 0.0002 max mem: 52126 +Epoch: [28] [ 140/3634] eta: 0:30:47 lr: 0.000003 loss: 0.5094 (0.5268) time: 0.4735 data: 0.0002 max mem: 52126 +Epoch: [28] [ 150/3634] eta: 0:30:08 lr: 0.000003 loss: 0.4989 (0.5245) time: 0.4272 data: 0.0001 max mem: 52126 +Epoch: [28] [ 160/3634] eta: 0:29:58 lr: 0.000003 loss: 0.5118 (0.5256) time: 0.4394 data: 0.0002 max mem: 52126 +Epoch: [28] [ 170/3634] eta: 0:29:25 lr: 0.000003 loss: 0.5144 (0.5243) time: 0.4396 data: 0.0002 max mem: 52126 +Epoch: [28] [ 180/3634] eta: 0:29:20 lr: 0.000003 loss: 0.4858 (0.5216) time: 0.4439 data: 0.0002 max mem: 52126 +Epoch: [28] [ 190/3634] eta: 0:29:08 lr: 0.000003 loss: 0.4921 (0.5199) time: 0.4898 data: 0.0002 max mem: 52126 +Epoch: [28] [ 200/3634] eta: 0:29:14 lr: 0.000003 loss: 0.4931 (0.5190) time: 0.5229 data: 0.0002 max mem: 52126 +Epoch: [28] [ 210/3634] eta: 0:28:48 lr: 0.000003 loss: 0.5396 (0.5201) time: 0.4774 data: 0.0002 max mem: 52126 +Epoch: [28] [ 220/3634] eta: 0:28:24 lr: 0.000003 loss: 0.5485 (0.5212) time: 0.3820 data: 0.0002 max mem: 52126 +Epoch: [28] [ 230/3634] eta: 0:28:15 lr: 0.000003 loss: 0.5358 (0.5205) time: 0.4284 data: 0.0002 max mem: 52126 +Epoch: [28] [ 240/3634] eta: 0:27:54 lr: 0.000003 loss: 0.4933 (0.5206) time: 0.4280 data: 0.0002 max mem: 52126 +Epoch: [28] [ 250/3634] eta: 0:27:34 lr: 0.000003 loss: 0.5145 (0.5202) time: 0.3818 data: 0.0002 max mem: 52126 +Epoch: [28] [ 260/3634] eta: 0:27:15 lr: 0.000003 loss: 0.4718 (0.5178) time: 0.3816 data: 0.0002 max mem: 52126 +Epoch: [28] [ 270/3634] eta: 0:27:09 lr: 0.000003 loss: 0.4650 (0.5169) time: 0.4284 data: 0.0002 max mem: 52126 +Epoch: [28] [ 280/3634] eta: 0:27:44 lr: 0.000003 loss: 0.4882 (0.5160) time: 0.6466 data: 0.0002 max mem: 52126 +Epoch: [28] [ 290/3634] eta: 0:27:40 lr: 0.000003 loss: 0.5289 (0.5157) time: 0.6615 data: 0.0002 max mem: 52126 +Epoch: [28] [ 300/3634] eta: 0:27:47 lr: 0.000003 loss: 0.5105 (0.5153) time: 0.5534 data: 0.0002 max mem: 52126 +Epoch: [28] [ 310/3634] eta: 0:27:39 lr: 0.000003 loss: 0.5167 (0.5155) time: 0.5379 data: 0.0002 max mem: 52126 +Epoch: [28] [ 320/3634] eta: 0:27:22 lr: 0.000003 loss: 0.5236 (0.5153) time: 0.4277 data: 0.0002 max mem: 52126 +Epoch: [28] [ 330/3634] eta: 0:27:29 lr: 0.000003 loss: 0.5133 (0.5155) time: 0.4968 data: 0.0002 max mem: 52126 +Epoch: [28] [ 340/3634] eta: 0:27:58 lr: 0.000003 loss: 0.5238 (0.5154) time: 0.7341 data: 0.0002 max mem: 52126 +Epoch: [28] [ 350/3634] eta: 0:28:01 lr: 0.000003 loss: 0.5238 (0.5159) time: 0.7257 data: 0.0002 max mem: 52126 +Epoch: [28] [ 360/3634] eta: 0:27:44 lr: 0.000003 loss: 0.5236 (0.5162) time: 0.4891 data: 0.0001 max mem: 52126 +Epoch: [28] [ 370/3634] eta: 0:28:01 lr: 0.000003 loss: 0.4829 (0.5151) time: 0.5677 data: 0.0002 max mem: 52126 +Epoch: [28] [ 380/3634] eta: 0:28:05 lr: 0.000003 loss: 0.4816 (0.5147) time: 0.6884 data: 0.0002 max mem: 52126 +Epoch: [28] [ 390/3634] eta: 0:27:58 lr: 0.000003 loss: 0.5084 (0.5150) time: 0.5632 data: 0.0002 max mem: 52126 +Epoch: [28] [ 400/3634] eta: 0:27:59 lr: 0.000003 loss: 0.5084 (0.5142) time: 0.5496 data: 0.0002 max mem: 52126 +Epoch: [28] [ 410/3634] eta: 0:27:43 lr: 0.000003 loss: 0.4998 (0.5143) time: 0.4889 data: 0.0002 max mem: 52126 +Epoch: [28] [ 420/3634] eta: 0:27:44 lr: 0.000003 loss: 0.5264 (0.5140) time: 0.4876 data: 0.0001 max mem: 52126 +Epoch: [28] [ 430/3634] eta: 0:27:38 lr: 0.000003 loss: 0.5158 (0.5140) time: 0.5488 data: 0.0002 max mem: 52126 +Epoch: [28] [ 440/3634] eta: 0:27:32 lr: 0.000003 loss: 0.5293 (0.5142) time: 0.5045 data: 0.0002 max mem: 52126 +Epoch: [28] [ 450/3634] eta: 0:27:24 lr: 0.000003 loss: 0.5144 (0.5137) time: 0.4923 data: 0.0002 max mem: 52126 +Epoch: [28] [ 460/3634] eta: 0:27:16 lr: 0.000003 loss: 0.4599 (0.5124) time: 0.4762 data: 0.0002 max mem: 52126 +Epoch: [28] [ 470/3634] eta: 0:27:10 lr: 0.000003 loss: 0.4603 (0.5130) time: 0.4887 data: 0.0002 max mem: 52126 +Epoch: [28] [ 480/3634] eta: 0:27:03 lr: 0.000003 loss: 0.5392 (0.5133) time: 0.4925 data: 0.0002 max mem: 52126 +Epoch: [28] [ 490/3634] eta: 0:27:05 lr: 0.000003 loss: 0.5210 (0.5135) time: 0.5543 data: 0.0002 max mem: 52126 +Epoch: [28] [ 500/3634] eta: 0:26:51 lr: 0.000003 loss: 0.5107 (0.5132) time: 0.5048 data: 0.0002 max mem: 52126 +Epoch: [28] [ 510/3634] eta: 0:26:44 lr: 0.000003 loss: 0.5029 (0.5129) time: 0.4285 data: 0.0002 max mem: 52126 +Epoch: [28] [ 520/3634] eta: 0:26:36 lr: 0.000003 loss: 0.5243 (0.5133) time: 0.4774 data: 0.0002 max mem: 52126 +Epoch: [28] [ 530/3634] eta: 0:26:29 lr: 0.000003 loss: 0.5276 (0.5138) time: 0.4752 data: 0.0002 max mem: 52126 +Epoch: [28] [ 540/3634] eta: 0:26:26 lr: 0.000003 loss: 0.5165 (0.5132) time: 0.5161 data: 0.0002 max mem: 52126 +Epoch: [28] [ 550/3634] eta: 0:26:24 lr: 0.000003 loss: 0.4797 (0.5124) time: 0.5598 data: 0.0002 max mem: 52126 +Epoch: [28] [ 560/3634] eta: 0:26:22 lr: 0.000003 loss: 0.4906 (0.5127) time: 0.5624 data: 0.0002 max mem: 52126 +Epoch: [28] [ 570/3634] eta: 0:26:09 lr: 0.000003 loss: 0.5258 (0.5121) time: 0.4733 data: 0.0002 max mem: 52126 +Epoch: [28] [ 580/3634] eta: 0:25:57 lr: 0.000003 loss: 0.5163 (0.5122) time: 0.3806 data: 0.0002 max mem: 52126 +Epoch: [28] [ 590/3634] eta: 0:25:46 lr: 0.000003 loss: 0.5163 (0.5119) time: 0.3812 data: 0.0002 max mem: 52126 +Epoch: [28] [ 600/3634] eta: 0:25:39 lr: 0.000003 loss: 0.5216 (0.5120) time: 0.4274 data: 0.0002 max mem: 52126 +Epoch: [28] [ 610/3634] eta: 0:25:27 lr: 0.000003 loss: 0.5291 (0.5123) time: 0.4269 data: 0.0002 max mem: 52126 +Epoch: [28] [ 620/3634] eta: 0:25:16 lr: 0.000003 loss: 0.5085 (0.5124) time: 0.3821 data: 0.0002 max mem: 52126 +Epoch: [28] [ 630/3634] eta: 0:25:15 lr: 0.000003 loss: 0.5286 (0.5127) time: 0.4820 data: 0.0002 max mem: 52126 +Epoch: [28] [ 640/3634] eta: 0:25:15 lr: 0.000003 loss: 0.5286 (0.5127) time: 0.5916 data: 0.0002 max mem: 52126 +Epoch: [28] [ 650/3634] eta: 0:25:08 lr: 0.000003 loss: 0.4922 (0.5125) time: 0.5385 data: 0.0002 max mem: 52126 +Epoch: [28] [ 660/3634] eta: 0:25:08 lr: 0.000003 loss: 0.4773 (0.5123) time: 0.5450 data: 0.0001 max mem: 52126 +Epoch: [28] [ 670/3634] eta: 0:24:57 lr: 0.000003 loss: 0.5247 (0.5123) time: 0.4986 data: 0.0001 max mem: 52126 +Epoch: [28] [ 680/3634] eta: 0:24:47 lr: 0.000003 loss: 0.4973 (0.5117) time: 0.3816 data: 0.0002 max mem: 52126 +Epoch: [28] [ 690/3634] eta: 0:24:44 lr: 0.000003 loss: 0.4973 (0.5116) time: 0.4699 data: 0.0002 max mem: 52126 +Epoch: [28] [ 700/3634] eta: 0:24:34 lr: 0.000003 loss: 0.5120 (0.5118) time: 0.4693 data: 0.0002 max mem: 52126 +Epoch: [28] [ 710/3634] eta: 0:24:34 lr: 0.000003 loss: 0.5120 (0.5118) time: 0.5002 data: 0.0002 max mem: 52126 +Epoch: [28] [ 720/3634] eta: 0:24:24 lr: 0.000003 loss: 0.5252 (0.5124) time: 0.5004 data: 0.0002 max mem: 52126 +Epoch: [28] [ 730/3634] eta: 0:24:14 lr: 0.000003 loss: 0.5278 (0.5125) time: 0.3810 data: 0.0002 max mem: 52126 +Epoch: [28] [ 740/3634] eta: 0:24:09 lr: 0.000003 loss: 0.4944 (0.5121) time: 0.4403 data: 0.0002 max mem: 52126 +Epoch: [28] [ 750/3634] eta: 0:23:59 lr: 0.000003 loss: 0.4820 (0.5119) time: 0.4406 data: 0.0001 max mem: 52126 +Epoch: [28] [ 760/3634] eta: 0:23:53 lr: 0.000003 loss: 0.5136 (0.5123) time: 0.4255 data: 0.0002 max mem: 52126 +Epoch: [28] [ 770/3634] eta: 0:23:47 lr: 0.000003 loss: 0.5030 (0.5119) time: 0.4723 data: 0.0002 max mem: 52126 +Epoch: [28] [ 780/3634] eta: 0:23:38 lr: 0.000003 loss: 0.4886 (0.5119) time: 0.4298 data: 0.0001 max mem: 52126 +Epoch: [28] [ 790/3634] eta: 0:23:32 lr: 0.000003 loss: 0.4862 (0.5114) time: 0.4278 data: 0.0002 max mem: 52126 +Epoch: [28] [ 800/3634] eta: 0:23:32 lr: 0.000003 loss: 0.4862 (0.5116) time: 0.5488 data: 0.0002 max mem: 52126 +Epoch: [28] [ 810/3634] eta: 0:23:26 lr: 0.000003 loss: 0.5216 (0.5117) time: 0.5487 data: 0.0001 max mem: 52126 +Epoch: [28] [ 820/3634] eta: 0:23:21 lr: 0.000003 loss: 0.5187 (0.5116) time: 0.4862 data: 0.0002 max mem: 52126 +Epoch: [28] [ 830/3634] eta: 0:23:15 lr: 0.000003 loss: 0.5125 (0.5113) time: 0.4852 data: 0.0002 max mem: 52126 +Epoch: [28] [ 840/3634] eta: 0:23:21 lr: 0.000003 loss: 0.5317 (0.5117) time: 0.6466 data: 0.0002 max mem: 52126 +Epoch: [28] [ 850/3634] eta: 0:23:12 lr: 0.000003 loss: 0.5446 (0.5117) time: 0.6029 data: 0.0001 max mem: 52126 +Epoch: [28] [ 860/3634] eta: 0:23:03 lr: 0.000003 loss: 0.5124 (0.5115) time: 0.3815 data: 0.0002 max mem: 52126 +Epoch: [28] [ 870/3634] eta: 0:23:01 lr: 0.000003 loss: 0.5093 (0.5114) time: 0.4894 data: 0.0002 max mem: 52126 +Epoch: [28] [ 880/3634] eta: 0:22:53 lr: 0.000003 loss: 0.5093 (0.5112) time: 0.4892 data: 0.0002 max mem: 52126 +Epoch: [28] [ 890/3634] eta: 0:22:51 lr: 0.000003 loss: 0.5207 (0.5114) time: 0.4989 data: 0.0001 max mem: 52126 +Epoch: [28] [ 900/3634] eta: 0:22:46 lr: 0.000003 loss: 0.5354 (0.5117) time: 0.5586 data: 0.0002 max mem: 52126 +Epoch: [28] [ 910/3634] eta: 0:22:44 lr: 0.000003 loss: 0.5326 (0.5118) time: 0.5462 data: 0.0002 max mem: 52126 +Epoch: [28] [ 920/3634] eta: 0:22:51 lr: 0.000003 loss: 0.5034 (0.5114) time: 0.7580 data: 0.0002 max mem: 52126 +Epoch: [28] [ 930/3634] eta: 0:22:46 lr: 0.000003 loss: 0.5086 (0.5118) time: 0.7122 data: 0.0001 max mem: 52126 +Epoch: [28] [ 940/3634] eta: 0:22:38 lr: 0.000003 loss: 0.5668 (0.5123) time: 0.4411 data: 0.0002 max mem: 52126 +Epoch: [28] [ 950/3634] eta: 0:22:33 lr: 0.000003 loss: 0.5300 (0.5120) time: 0.4425 data: 0.0002 max mem: 52126 +Epoch: [28] [ 960/3634] eta: 0:22:24 lr: 0.000003 loss: 0.5219 (0.5118) time: 0.4423 data: 0.0002 max mem: 52126 +Epoch: [28] [ 970/3634] eta: 0:22:16 lr: 0.000003 loss: 0.5165 (0.5114) time: 0.3810 data: 0.0002 max mem: 52126 +Epoch: [28] [ 980/3634] eta: 0:22:10 lr: 0.000003 loss: 0.5165 (0.5117) time: 0.4248 data: 0.0002 max mem: 52126 +Epoch: [28] [ 990/3634] eta: 0:22:02 lr: 0.000003 loss: 0.5260 (0.5115) time: 0.4250 data: 0.0002 max mem: 52126 +Epoch: [28] [1000/3634] eta: 0:21:59 lr: 0.000003 loss: 0.5698 (0.5120) time: 0.4845 data: 0.0002 max mem: 52126 +Epoch: [28] [1010/3634] eta: 0:21:56 lr: 0.000003 loss: 0.5528 (0.5122) time: 0.5871 data: 0.0002 max mem: 52126 +Epoch: [28] [1020/3634] eta: 0:21:55 lr: 0.000003 loss: 0.5467 (0.5121) time: 0.6208 data: 0.0002 max mem: 52126 +Epoch: [28] [1030/3634] eta: 0:21:47 lr: 0.000003 loss: 0.5089 (0.5122) time: 0.5184 data: 0.0001 max mem: 52126 +Epoch: [28] [1040/3634] eta: 0:21:41 lr: 0.000003 loss: 0.5197 (0.5125) time: 0.4242 data: 0.0001 max mem: 52126 +Epoch: [28] [1050/3634] eta: 0:21:38 lr: 0.000003 loss: 0.5363 (0.5128) time: 0.5274 data: 0.0002 max mem: 52126 +Epoch: [28] [1060/3634] eta: 0:21:30 lr: 0.000003 loss: 0.5512 (0.5131) time: 0.4843 data: 0.0002 max mem: 52126 +Epoch: [28] [1070/3634] eta: 0:21:27 lr: 0.000003 loss: 0.5423 (0.5134) time: 0.4697 data: 0.0002 max mem: 52126 +Epoch: [28] [1080/3634] eta: 0:21:23 lr: 0.000003 loss: 0.5290 (0.5131) time: 0.5627 data: 0.0001 max mem: 52126 +Epoch: [28] [1090/3634] eta: 0:21:15 lr: 0.000003 loss: 0.4798 (0.5128) time: 0.4745 data: 0.0001 max mem: 52126 +Epoch: [28] [1100/3634] eta: 0:21:16 lr: 0.000003 loss: 0.4971 (0.5129) time: 0.5780 data: 0.0002 max mem: 52126 +Epoch: [28] [1110/3634] eta: 0:21:11 lr: 0.000003 loss: 0.4786 (0.5124) time: 0.6261 data: 0.0002 max mem: 52126 +Epoch: [28] [1120/3634] eta: 0:21:08 lr: 0.000003 loss: 0.4786 (0.5125) time: 0.5327 data: 0.0002 max mem: 52126 +Epoch: [28] [1130/3634] eta: 0:21:02 lr: 0.000003 loss: 0.5091 (0.5124) time: 0.5340 data: 0.0002 max mem: 52126 +Epoch: [28] [1140/3634] eta: 0:20:56 lr: 0.000003 loss: 0.5208 (0.5125) time: 0.4751 data: 0.0002 max mem: 52126 +Epoch: [28] [1150/3634] eta: 0:20:49 lr: 0.000003 loss: 0.5220 (0.5124) time: 0.4254 data: 0.0001 max mem: 52126 +Epoch: [28] [1160/3634] eta: 0:20:43 lr: 0.000003 loss: 0.4976 (0.5122) time: 0.4246 data: 0.0001 max mem: 52126 +Epoch: [28] [1170/3634] eta: 0:20:40 lr: 0.000003 loss: 0.4936 (0.5121) time: 0.5295 data: 0.0001 max mem: 52126 +Epoch: [28] [1180/3634] eta: 0:20:32 lr: 0.000003 loss: 0.5209 (0.5121) time: 0.4857 data: 0.0002 max mem: 52126 +Epoch: [28] [1190/3634] eta: 0:20:25 lr: 0.000003 loss: 0.5298 (0.5123) time: 0.3816 data: 0.0002 max mem: 52126 +Epoch: [28] [1200/3634] eta: 0:20:20 lr: 0.000003 loss: 0.5359 (0.5122) time: 0.4428 data: 0.0002 max mem: 52126 +Epoch: [28] [1210/3634] eta: 0:20:15 lr: 0.000003 loss: 0.5046 (0.5120) time: 0.5007 data: 0.0002 max mem: 52126 +Epoch: [28] [1220/3634] eta: 0:20:07 lr: 0.000003 loss: 0.4999 (0.5118) time: 0.4396 data: 0.0002 max mem: 52126 +Epoch: [28] [1230/3634] eta: 0:20:00 lr: 0.000003 loss: 0.5135 (0.5118) time: 0.3803 data: 0.0002 max mem: 52126 +Epoch: [28] [1240/3634] eta: 0:19:53 lr: 0.000003 loss: 0.5506 (0.5121) time: 0.3809 data: 0.0002 max mem: 52126 +Epoch: [28] [1250/3634] eta: 0:19:45 lr: 0.000003 loss: 0.5490 (0.5123) time: 0.3810 data: 0.0002 max mem: 52126 +Epoch: [28] [1260/3634] eta: 0:19:42 lr: 0.000003 loss: 0.5490 (0.5126) time: 0.4867 data: 0.0002 max mem: 52126 +Epoch: [28] [1270/3634] eta: 0:19:35 lr: 0.000003 loss: 0.5615 (0.5128) time: 0.4865 data: 0.0002 max mem: 52126 +Epoch: [28] [1280/3634] eta: 0:19:28 lr: 0.000003 loss: 0.5475 (0.5128) time: 0.3810 data: 0.0002 max mem: 52126 +Epoch: [28] [1290/3634] eta: 0:19:21 lr: 0.000003 loss: 0.5201 (0.5130) time: 0.3836 data: 0.0001 max mem: 52126 +Epoch: [28] [1300/3634] eta: 0:19:14 lr: 0.000003 loss: 0.5201 (0.5129) time: 0.3833 data: 0.0001 max mem: 52126 +Epoch: [28] [1310/3634] eta: 0:19:09 lr: 0.000003 loss: 0.5366 (0.5133) time: 0.4400 data: 0.0001 max mem: 52126 +Epoch: [28] [1320/3634] eta: 0:19:02 lr: 0.000003 loss: 0.5580 (0.5135) time: 0.4398 data: 0.0002 max mem: 52126 +Epoch: [28] [1330/3634] eta: 0:18:57 lr: 0.000003 loss: 0.4962 (0.5135) time: 0.4266 data: 0.0002 max mem: 52126 +Epoch: [28] [1340/3634] eta: 0:18:52 lr: 0.000003 loss: 0.4962 (0.5134) time: 0.4858 data: 0.0002 max mem: 52126 +Epoch: [28] [1350/3634] eta: 0:18:50 lr: 0.000003 loss: 0.5284 (0.5135) time: 0.5858 data: 0.0002 max mem: 52126 +Epoch: [28] [1360/3634] eta: 0:18:43 lr: 0.000003 loss: 0.5186 (0.5135) time: 0.5272 data: 0.0002 max mem: 52126 +Epoch: [28] [1370/3634] eta: 0:18:38 lr: 0.000003 loss: 0.5224 (0.5138) time: 0.4410 data: 0.0001 max mem: 52126 +Epoch: [28] [1380/3634] eta: 0:18:36 lr: 0.000003 loss: 0.5344 (0.5137) time: 0.5622 data: 0.0001 max mem: 52126 +Epoch: [28] [1390/3634] eta: 0:18:29 lr: 0.000003 loss: 0.5133 (0.5135) time: 0.5032 data: 0.0001 max mem: 52126 +Epoch: [28] [1400/3634] eta: 0:18:22 lr: 0.000003 loss: 0.5178 (0.5139) time: 0.3820 data: 0.0002 max mem: 52126 +Epoch: [28] [1410/3634] eta: 0:18:19 lr: 0.000003 loss: 0.5399 (0.5140) time: 0.4844 data: 0.0002 max mem: 52126 +Epoch: [28] [1420/3634] eta: 0:18:13 lr: 0.000003 loss: 0.5048 (0.5139) time: 0.5320 data: 0.0002 max mem: 52126 +Epoch: [28] [1430/3634] eta: 0:18:08 lr: 0.000003 loss: 0.4955 (0.5139) time: 0.4878 data: 0.0002 max mem: 52126 +Epoch: [28] [1440/3634] eta: 0:18:02 lr: 0.000003 loss: 0.5257 (0.5140) time: 0.4387 data: 0.0002 max mem: 52126 +Epoch: [28] [1450/3634] eta: 0:17:58 lr: 0.000003 loss: 0.5315 (0.5141) time: 0.4853 data: 0.0002 max mem: 52126 +Epoch: [28] [1460/3634] eta: 0:17:55 lr: 0.000003 loss: 0.5438 (0.5141) time: 0.6052 data: 0.0002 max mem: 52126 +Epoch: [28] [1470/3634] eta: 0:17:49 lr: 0.000003 loss: 0.5120 (0.5140) time: 0.5011 data: 0.0002 max mem: 52126 +Epoch: [28] [1480/3634] eta: 0:17:43 lr: 0.000003 loss: 0.4975 (0.5138) time: 0.4278 data: 0.0002 max mem: 52126 +Epoch: [28] [1490/3634] eta: 0:17:38 lr: 0.000003 loss: 0.4932 (0.5137) time: 0.4726 data: 0.0002 max mem: 52126 +Epoch: [28] [1500/3634] eta: 0:17:33 lr: 0.000003 loss: 0.4709 (0.5135) time: 0.4730 data: 0.0001 max mem: 52126 +Epoch: [28] [1510/3634] eta: 0:17:28 lr: 0.000003 loss: 0.4795 (0.5134) time: 0.4718 data: 0.0002 max mem: 52126 +Epoch: [28] [1520/3634] eta: 0:17:22 lr: 0.000003 loss: 0.5444 (0.5136) time: 0.4701 data: 0.0002 max mem: 52126 +Epoch: [28] [1530/3634] eta: 0:17:20 lr: 0.000003 loss: 0.5536 (0.5136) time: 0.5904 data: 0.0002 max mem: 52126 +Epoch: [28] [1540/3634] eta: 0:17:17 lr: 0.000003 loss: 0.5318 (0.5136) time: 0.6498 data: 0.0002 max mem: 52126 +Epoch: [28] [1550/3634] eta: 0:17:12 lr: 0.000003 loss: 0.5304 (0.5136) time: 0.5447 data: 0.0001 max mem: 52126 +Epoch: [28] [1560/3634] eta: 0:17:07 lr: 0.000003 loss: 0.5249 (0.5138) time: 0.5014 data: 0.0001 max mem: 52126 +Epoch: [28] [1570/3634] eta: 0:17:03 lr: 0.000003 loss: 0.5112 (0.5138) time: 0.5455 data: 0.0002 max mem: 52126 +Epoch: [28] [1580/3634] eta: 0:17:00 lr: 0.000003 loss: 0.5061 (0.5138) time: 0.5946 data: 0.0002 max mem: 52126 +Epoch: [28] [1590/3634] eta: 0:16:56 lr: 0.000003 loss: 0.5135 (0.5138) time: 0.6102 data: 0.0002 max mem: 52126 +Epoch: [28] [1600/3634] eta: 0:16:52 lr: 0.000003 loss: 0.5536 (0.5140) time: 0.5902 data: 0.0001 max mem: 52126 +Epoch: [28] [1610/3634] eta: 0:16:46 lr: 0.000003 loss: 0.5069 (0.5139) time: 0.4709 data: 0.0001 max mem: 52126 +Epoch: [28] [1620/3634] eta: 0:16:40 lr: 0.000003 loss: 0.4887 (0.5139) time: 0.4257 data: 0.0002 max mem: 52126 +Epoch: [28] [1630/3634] eta: 0:16:37 lr: 0.000003 loss: 0.5309 (0.5140) time: 0.5381 data: 0.0001 max mem: 52126 +Epoch: [28] [1640/3634] eta: 0:16:32 lr: 0.000003 loss: 0.4700 (0.5136) time: 0.5521 data: 0.0001 max mem: 52126 +Epoch: [28] [1650/3634] eta: 0:16:27 lr: 0.000003 loss: 0.4769 (0.5137) time: 0.5005 data: 0.0002 max mem: 52126 +Epoch: [28] [1660/3634] eta: 0:16:22 lr: 0.000003 loss: 0.5059 (0.5136) time: 0.4866 data: 0.0002 max mem: 52126 +Epoch: [28] [1670/3634] eta: 0:16:18 lr: 0.000003 loss: 0.5349 (0.5135) time: 0.5565 data: 0.0001 max mem: 52126 +Epoch: [28] [1680/3634] eta: 0:16:12 lr: 0.000003 loss: 0.4868 (0.5133) time: 0.5123 data: 0.0002 max mem: 52126 +Epoch: [28] [1690/3634] eta: 0:16:08 lr: 0.000003 loss: 0.5042 (0.5134) time: 0.5000 data: 0.0002 max mem: 52126 +Epoch: [28] [1700/3634] eta: 0:16:02 lr: 0.000003 loss: 0.5123 (0.5133) time: 0.4998 data: 0.0001 max mem: 52126 +Epoch: [28] [1710/3634] eta: 0:15:56 lr: 0.000003 loss: 0.5157 (0.5136) time: 0.3814 data: 0.0002 max mem: 52126 +Epoch: [28] [1720/3634] eta: 0:15:51 lr: 0.000003 loss: 0.5442 (0.5137) time: 0.4426 data: 0.0002 max mem: 52126 +Epoch: [28] [1730/3634] eta: 0:15:45 lr: 0.000003 loss: 0.5381 (0.5138) time: 0.4432 data: 0.0002 max mem: 52126 +Epoch: [28] [1740/3634] eta: 0:15:41 lr: 0.000003 loss: 0.5185 (0.5137) time: 0.4924 data: 0.0002 max mem: 52126 +Epoch: [28] [1750/3634] eta: 0:15:35 lr: 0.000003 loss: 0.5076 (0.5140) time: 0.4926 data: 0.0001 max mem: 52126 +Epoch: [28] [1760/3634] eta: 0:15:28 lr: 0.000003 loss: 0.5097 (0.5140) time: 0.3819 data: 0.0001 max mem: 52126 +Epoch: [28] [1770/3634] eta: 0:15:22 lr: 0.000003 loss: 0.5200 (0.5142) time: 0.3814 data: 0.0001 max mem: 52126 +Epoch: [28] [1780/3634] eta: 0:15:17 lr: 0.000003 loss: 0.5028 (0.5140) time: 0.4402 data: 0.0001 max mem: 52126 +Epoch: [28] [1790/3634] eta: 0:15:11 lr: 0.000003 loss: 0.5002 (0.5138) time: 0.4404 data: 0.0002 max mem: 52126 +Epoch: [28] [1800/3634] eta: 0:15:06 lr: 0.000003 loss: 0.5124 (0.5138) time: 0.4426 data: 0.0002 max mem: 52126 +Epoch: [28] [1810/3634] eta: 0:15:00 lr: 0.000003 loss: 0.5175 (0.5137) time: 0.4420 data: 0.0002 max mem: 52126 +Epoch: [28] [1820/3634] eta: 0:14:54 lr: 0.000003 loss: 0.5036 (0.5136) time: 0.3812 data: 0.0001 max mem: 52126 +Epoch: [28] [1830/3634] eta: 0:14:48 lr: 0.000003 loss: 0.5021 (0.5136) time: 0.3814 data: 0.0001 max mem: 52126 +Epoch: [28] [1840/3634] eta: 0:14:43 lr: 0.000003 loss: 0.4999 (0.5135) time: 0.4298 data: 0.0002 max mem: 52126 +Epoch: [28] [1850/3634] eta: 0:14:40 lr: 0.000003 loss: 0.5012 (0.5136) time: 0.5985 data: 0.0002 max mem: 52126 +Epoch: [28] [1860/3634] eta: 0:14:36 lr: 0.000003 loss: 0.5210 (0.5135) time: 0.6572 data: 0.0001 max mem: 52126 +Epoch: [28] [1870/3634] eta: 0:14:32 lr: 0.000003 loss: 0.5280 (0.5137) time: 0.5954 data: 0.0001 max mem: 52126 +Epoch: [28] [1880/3634] eta: 0:14:26 lr: 0.000003 loss: 0.5194 (0.5137) time: 0.4877 data: 0.0002 max mem: 52126 +Epoch: [28] [1890/3634] eta: 0:14:22 lr: 0.000003 loss: 0.5085 (0.5136) time: 0.4837 data: 0.0002 max mem: 52126 +Epoch: [28] [1900/3634] eta: 0:14:16 lr: 0.000003 loss: 0.5239 (0.5138) time: 0.4843 data: 0.0002 max mem: 52126 +Epoch: [28] [1910/3634] eta: 0:14:11 lr: 0.000003 loss: 0.5500 (0.5140) time: 0.4257 data: 0.0002 max mem: 52126 +Epoch: [28] [1920/3634] eta: 0:14:05 lr: 0.000003 loss: 0.5365 (0.5140) time: 0.4253 data: 0.0002 max mem: 52126 +Epoch: [28] [1930/3634] eta: 0:14:00 lr: 0.000003 loss: 0.5322 (0.5142) time: 0.4414 data: 0.0002 max mem: 52126 +Epoch: [28] [1940/3634] eta: 0:13:55 lr: 0.000003 loss: 0.5322 (0.5141) time: 0.4867 data: 0.0002 max mem: 52126 +Epoch: [28] [1950/3634] eta: 0:13:50 lr: 0.000003 loss: 0.4573 (0.5139) time: 0.4854 data: 0.0001 max mem: 52126 +Epoch: [28] [1960/3634] eta: 0:13:46 lr: 0.000003 loss: 0.4789 (0.5139) time: 0.4990 data: 0.0002 max mem: 52126 +Epoch: [28] [1970/3634] eta: 0:13:41 lr: 0.000003 loss: 0.5060 (0.5139) time: 0.5455 data: 0.0002 max mem: 52126 +Epoch: [28] [1980/3634] eta: 0:13:37 lr: 0.000003 loss: 0.4996 (0.5138) time: 0.5464 data: 0.0001 max mem: 52126 +Epoch: [28] [1990/3634] eta: 0:13:31 lr: 0.000003 loss: 0.4947 (0.5138) time: 0.4419 data: 0.0001 max mem: 52126 +Epoch: [28] [2000/3634] eta: 0:13:28 lr: 0.000003 loss: 0.5085 (0.5139) time: 0.5595 data: 0.0002 max mem: 52126 +Epoch: [28] [2010/3634] eta: 0:13:22 lr: 0.000003 loss: 0.4854 (0.5137) time: 0.5590 data: 0.0002 max mem: 52126 +Epoch: [28] [2020/3634] eta: 0:13:18 lr: 0.000003 loss: 0.4763 (0.5136) time: 0.4855 data: 0.0001 max mem: 52126 +Epoch: [28] [2030/3634] eta: 0:13:12 lr: 0.000003 loss: 0.5023 (0.5135) time: 0.4855 data: 0.0001 max mem: 52126 +Epoch: [28] [2040/3634] eta: 0:13:07 lr: 0.000003 loss: 0.4833 (0.5134) time: 0.4399 data: 0.0001 max mem: 52126 +Epoch: [28] [2050/3634] eta: 0:13:03 lr: 0.000003 loss: 0.5195 (0.5136) time: 0.5298 data: 0.0001 max mem: 52126 +Epoch: [28] [2060/3634] eta: 0:12:57 lr: 0.000003 loss: 0.5322 (0.5136) time: 0.5192 data: 0.0001 max mem: 52126 +Epoch: [28] [2070/3634] eta: 0:12:54 lr: 0.000003 loss: 0.5308 (0.5136) time: 0.5527 data: 0.0001 max mem: 52126 +Epoch: [28] [2080/3634] eta: 0:12:48 lr: 0.000003 loss: 0.5547 (0.5137) time: 0.5052 data: 0.0001 max mem: 52126 +Epoch: [28] [2090/3634] eta: 0:12:43 lr: 0.000003 loss: 0.5182 (0.5135) time: 0.4849 data: 0.0001 max mem: 52126 +Epoch: [28] [2100/3634] eta: 0:12:41 lr: 0.000003 loss: 0.4879 (0.5134) time: 0.7097 data: 0.0001 max mem: 52126 +Epoch: [28] [2110/3634] eta: 0:12:35 lr: 0.000003 loss: 0.5233 (0.5135) time: 0.6062 data: 0.0001 max mem: 52126 +Epoch: [28] [2120/3634] eta: 0:12:31 lr: 0.000003 loss: 0.5233 (0.5135) time: 0.4991 data: 0.0002 max mem: 52126 +Epoch: [28] [2130/3634] eta: 0:12:26 lr: 0.000003 loss: 0.5264 (0.5136) time: 0.5438 data: 0.0002 max mem: 52126 +Epoch: [28] [2140/3634] eta: 0:12:20 lr: 0.000003 loss: 0.5269 (0.5136) time: 0.4264 data: 0.0001 max mem: 52126 +Epoch: [28] [2150/3634] eta: 0:12:15 lr: 0.000003 loss: 0.5157 (0.5136) time: 0.4263 data: 0.0001 max mem: 52126 +Epoch: [28] [2160/3634] eta: 0:12:11 lr: 0.000003 loss: 0.5320 (0.5136) time: 0.5343 data: 0.0002 max mem: 52126 +Epoch: [28] [2170/3634] eta: 0:12:06 lr: 0.000003 loss: 0.5314 (0.5137) time: 0.5499 data: 0.0002 max mem: 52126 +Epoch: [28] [2180/3634] eta: 0:12:02 lr: 0.000003 loss: 0.5094 (0.5136) time: 0.5450 data: 0.0001 max mem: 52126 +Epoch: [28] [2190/3634] eta: 0:11:56 lr: 0.000003 loss: 0.4783 (0.5135) time: 0.5306 data: 0.0001 max mem: 52126 +Epoch: [28] [2200/3634] eta: 0:11:52 lr: 0.000003 loss: 0.5405 (0.5137) time: 0.5318 data: 0.0002 max mem: 52126 +Epoch: [28] [2210/3634] eta: 0:11:47 lr: 0.000003 loss: 0.5213 (0.5136) time: 0.5474 data: 0.0002 max mem: 52126 +Epoch: [28] [2220/3634] eta: 0:11:42 lr: 0.000003 loss: 0.5166 (0.5136) time: 0.4868 data: 0.0001 max mem: 52126 +Epoch: [28] [2230/3634] eta: 0:11:36 lr: 0.000003 loss: 0.5182 (0.5136) time: 0.4246 data: 0.0001 max mem: 52126 +Epoch: [28] [2240/3634] eta: 0:11:31 lr: 0.000003 loss: 0.5150 (0.5134) time: 0.3808 data: 0.0002 max mem: 52126 +Epoch: [28] [2250/3634] eta: 0:11:26 lr: 0.000003 loss: 0.5177 (0.5135) time: 0.4415 data: 0.0002 max mem: 52126 +Epoch: [28] [2260/3634] eta: 0:11:20 lr: 0.000003 loss: 0.5177 (0.5136) time: 0.4416 data: 0.0002 max mem: 52126 +Epoch: [28] [2270/3634] eta: 0:11:14 lr: 0.000003 loss: 0.5082 (0.5135) time: 0.3816 data: 0.0002 max mem: 52126 +Epoch: [28] [2280/3634] eta: 0:11:09 lr: 0.000003 loss: 0.4898 (0.5134) time: 0.3818 data: 0.0002 max mem: 52126 +Epoch: [28] [2290/3634] eta: 0:11:04 lr: 0.000003 loss: 0.5334 (0.5135) time: 0.4415 data: 0.0001 max mem: 52126 +Epoch: [28] [2300/3634] eta: 0:11:00 lr: 0.000003 loss: 0.5165 (0.5135) time: 0.5488 data: 0.0001 max mem: 52126 +Epoch: [28] [2310/3634] eta: 0:10:55 lr: 0.000003 loss: 0.5092 (0.5136) time: 0.5521 data: 0.0001 max mem: 52126 +Epoch: [28] [2320/3634] eta: 0:10:49 lr: 0.000003 loss: 0.5410 (0.5137) time: 0.4447 data: 0.0002 max mem: 52126 +Epoch: [28] [2330/3634] eta: 0:10:43 lr: 0.000003 loss: 0.5268 (0.5137) time: 0.3819 data: 0.0002 max mem: 52126 +Epoch: [28] [2340/3634] eta: 0:10:39 lr: 0.000003 loss: 0.5214 (0.5138) time: 0.4413 data: 0.0002 max mem: 52126 +Epoch: [28] [2350/3634] eta: 0:10:36 lr: 0.000003 loss: 0.5418 (0.5139) time: 0.6821 data: 0.0002 max mem: 52126 +Epoch: [28] [2360/3634] eta: 0:10:31 lr: 0.000003 loss: 0.5367 (0.5139) time: 0.6843 data: 0.0001 max mem: 52126 +Epoch: [28] [2370/3634] eta: 0:10:26 lr: 0.000003 loss: 0.5258 (0.5140) time: 0.5061 data: 0.0001 max mem: 52126 +Epoch: [28] [2380/3634] eta: 0:10:21 lr: 0.000003 loss: 0.5258 (0.5140) time: 0.4887 data: 0.0001 max mem: 52126 +Epoch: [28] [2390/3634] eta: 0:10:16 lr: 0.000003 loss: 0.5085 (0.5139) time: 0.4712 data: 0.0001 max mem: 52126 +Epoch: [28] [2400/3634] eta: 0:10:11 lr: 0.000003 loss: 0.5419 (0.5140) time: 0.4863 data: 0.0002 max mem: 52126 +Epoch: [28] [2410/3634] eta: 0:10:05 lr: 0.000003 loss: 0.5382 (0.5141) time: 0.4408 data: 0.0002 max mem: 52126 +Epoch: [28] [2420/3634] eta: 0:10:01 lr: 0.000003 loss: 0.5382 (0.5143) time: 0.5454 data: 0.0001 max mem: 52126 +Epoch: [28] [2430/3634] eta: 0:09:56 lr: 0.000003 loss: 0.5565 (0.5144) time: 0.5948 data: 0.0001 max mem: 52126 +Epoch: [28] [2440/3634] eta: 0:09:52 lr: 0.000003 loss: 0.5517 (0.5144) time: 0.5335 data: 0.0001 max mem: 52126 +Epoch: [28] [2450/3634] eta: 0:09:47 lr: 0.000003 loss: 0.5478 (0.5145) time: 0.5443 data: 0.0001 max mem: 52126 +Epoch: [28] [2460/3634] eta: 0:09:41 lr: 0.000003 loss: 0.5425 (0.5145) time: 0.4417 data: 0.0001 max mem: 52126 +Epoch: [28] [2470/3634] eta: 0:09:36 lr: 0.000003 loss: 0.5160 (0.5145) time: 0.3821 data: 0.0002 max mem: 52126 +Epoch: [28] [2480/3634] eta: 0:09:30 lr: 0.000003 loss: 0.5185 (0.5146) time: 0.3817 data: 0.0001 max mem: 52126 +Epoch: [28] [2490/3634] eta: 0:09:26 lr: 0.000003 loss: 0.5229 (0.5146) time: 0.4843 data: 0.0002 max mem: 52126 +Epoch: [28] [2500/3634] eta: 0:09:21 lr: 0.000003 loss: 0.5365 (0.5146) time: 0.5287 data: 0.0002 max mem: 52126 +Epoch: [28] [2510/3634] eta: 0:09:16 lr: 0.000003 loss: 0.5335 (0.5146) time: 0.4857 data: 0.0001 max mem: 52126 +Epoch: [28] [2520/3634] eta: 0:09:11 lr: 0.000003 loss: 0.5107 (0.5145) time: 0.5009 data: 0.0001 max mem: 52126 +Epoch: [28] [2530/3634] eta: 0:09:06 lr: 0.000003 loss: 0.5107 (0.5145) time: 0.4407 data: 0.0001 max mem: 52126 +Epoch: [28] [2540/3634] eta: 0:09:00 lr: 0.000003 loss: 0.5256 (0.5146) time: 0.4273 data: 0.0001 max mem: 52126 +Epoch: [28] [2550/3634] eta: 0:08:55 lr: 0.000003 loss: 0.5165 (0.5145) time: 0.4273 data: 0.0001 max mem: 52126 +Epoch: [28] [2560/3634] eta: 0:08:50 lr: 0.000003 loss: 0.5005 (0.5145) time: 0.3820 data: 0.0001 max mem: 52126 +Epoch: [28] [2570/3634] eta: 0:08:45 lr: 0.000003 loss: 0.5040 (0.5144) time: 0.4301 data: 0.0001 max mem: 52126 +Epoch: [28] [2580/3634] eta: 0:08:39 lr: 0.000003 loss: 0.5068 (0.5144) time: 0.4300 data: 0.0002 max mem: 52126 +Epoch: [28] [2590/3634] eta: 0:08:35 lr: 0.000003 loss: 0.5289 (0.5145) time: 0.4846 data: 0.0002 max mem: 52126 +Epoch: [28] [2600/3634] eta: 0:08:29 lr: 0.000003 loss: 0.5225 (0.5144) time: 0.4842 data: 0.0001 max mem: 52126 +Epoch: [28] [2610/3634] eta: 0:08:25 lr: 0.000003 loss: 0.5168 (0.5145) time: 0.5056 data: 0.0002 max mem: 52126 +Epoch: [28] [2620/3634] eta: 0:08:20 lr: 0.000003 loss: 0.5459 (0.5145) time: 0.5062 data: 0.0002 max mem: 52126 +Epoch: [28] [2630/3634] eta: 0:08:15 lr: 0.000003 loss: 0.5438 (0.5144) time: 0.5476 data: 0.0002 max mem: 52126 +Epoch: [28] [2640/3634] eta: 0:08:11 lr: 0.000003 loss: 0.5355 (0.5144) time: 0.6395 data: 0.0001 max mem: 52126 +Epoch: [28] [2650/3634] eta: 0:08:06 lr: 0.000003 loss: 0.5497 (0.5145) time: 0.5782 data: 0.0002 max mem: 52126 +Epoch: [28] [2660/3634] eta: 0:08:01 lr: 0.000003 loss: 0.5497 (0.5145) time: 0.5323 data: 0.0002 max mem: 52126 +Epoch: [28] [2670/3634] eta: 0:07:56 lr: 0.000003 loss: 0.5359 (0.5147) time: 0.4295 data: 0.0002 max mem: 52126 +Epoch: [28] [2680/3634] eta: 0:07:51 lr: 0.000003 loss: 0.5365 (0.5147) time: 0.4419 data: 0.0002 max mem: 52126 +Epoch: [28] [2690/3634] eta: 0:07:46 lr: 0.000003 loss: 0.5478 (0.5148) time: 0.4411 data: 0.0001 max mem: 52126 +Epoch: [28] [2700/3634] eta: 0:07:41 lr: 0.000003 loss: 0.5243 (0.5147) time: 0.4260 data: 0.0001 max mem: 52126 +Epoch: [28] [2710/3634] eta: 0:07:36 lr: 0.000003 loss: 0.5050 (0.5147) time: 0.5289 data: 0.0001 max mem: 52126 +Epoch: [28] [2720/3634] eta: 0:07:32 lr: 0.000003 loss: 0.5152 (0.5147) time: 0.6932 data: 0.0002 max mem: 52126 +Epoch: [28] [2730/3634] eta: 0:07:28 lr: 0.000003 loss: 0.5152 (0.5147) time: 0.7086 data: 0.0002 max mem: 52126 +Epoch: [28] [2740/3634] eta: 0:07:23 lr: 0.000003 loss: 0.4802 (0.5146) time: 0.6824 data: 0.0002 max mem: 52126 +Epoch: [28] [2750/3634] eta: 0:07:19 lr: 0.000003 loss: 0.5290 (0.5147) time: 0.6852 data: 0.0002 max mem: 52126 +Epoch: [28] [2760/3634] eta: 0:07:13 lr: 0.000003 loss: 0.5670 (0.5148) time: 0.5029 data: 0.0001 max mem: 52126 +Epoch: [28] [2770/3634] eta: 0:07:08 lr: 0.000003 loss: 0.5204 (0.5148) time: 0.3809 data: 0.0001 max mem: 52126 +Epoch: [28] [2780/3634] eta: 0:07:03 lr: 0.000003 loss: 0.4976 (0.5147) time: 0.4701 data: 0.0001 max mem: 52126 +Epoch: [28] [2790/3634] eta: 0:06:58 lr: 0.000003 loss: 0.4872 (0.5147) time: 0.5298 data: 0.0001 max mem: 52126 +Epoch: [28] [2800/3634] eta: 0:06:53 lr: 0.000003 loss: 0.5270 (0.5147) time: 0.4993 data: 0.0001 max mem: 52126 +Epoch: [28] [2810/3634] eta: 0:06:49 lr: 0.000003 loss: 0.5169 (0.5146) time: 0.5594 data: 0.0002 max mem: 52126 +Epoch: [28] [2820/3634] eta: 0:06:44 lr: 0.000003 loss: 0.4941 (0.5146) time: 0.6215 data: 0.0002 max mem: 52126 +Epoch: [28] [2830/3634] eta: 0:06:39 lr: 0.000003 loss: 0.5206 (0.5147) time: 0.5674 data: 0.0002 max mem: 52126 +Epoch: [28] [2840/3634] eta: 0:06:35 lr: 0.000003 loss: 0.5328 (0.5148) time: 0.5387 data: 0.0002 max mem: 52126 +Epoch: [28] [2850/3634] eta: 0:06:30 lr: 0.000003 loss: 0.5247 (0.5146) time: 0.5181 data: 0.0001 max mem: 52126 +Epoch: [28] [2860/3634] eta: 0:06:24 lr: 0.000003 loss: 0.5091 (0.5148) time: 0.4705 data: 0.0002 max mem: 52126 +Epoch: [28] [2870/3634] eta: 0:06:20 lr: 0.000003 loss: 0.5575 (0.5148) time: 0.5127 data: 0.0002 max mem: 52126 +Epoch: [28] [2880/3634] eta: 0:06:15 lr: 0.000003 loss: 0.5241 (0.5148) time: 0.5276 data: 0.0002 max mem: 52126 +Epoch: [28] [2890/3634] eta: 0:06:10 lr: 0.000003 loss: 0.5235 (0.5148) time: 0.4906 data: 0.0001 max mem: 52126 +Epoch: [28] [2900/3634] eta: 0:06:04 lr: 0.000003 loss: 0.4969 (0.5147) time: 0.4308 data: 0.0001 max mem: 52126 +Epoch: [28] [2910/3634] eta: 0:06:00 lr: 0.000003 loss: 0.4751 (0.5147) time: 0.4982 data: 0.0001 max mem: 52126 +Epoch: [28] [2920/3634] eta: 0:05:55 lr: 0.000003 loss: 0.5036 (0.5147) time: 0.6021 data: 0.0001 max mem: 52126 +Epoch: [28] [2930/3634] eta: 0:05:50 lr: 0.000003 loss: 0.5024 (0.5147) time: 0.4845 data: 0.0001 max mem: 52126 +Epoch: [28] [2940/3634] eta: 0:05:45 lr: 0.000003 loss: 0.5018 (0.5147) time: 0.4258 data: 0.0001 max mem: 52126 +Epoch: [28] [2950/3634] eta: 0:05:39 lr: 0.000003 loss: 0.4889 (0.5146) time: 0.4258 data: 0.0002 max mem: 52126 +Epoch: [28] [2960/3634] eta: 0:05:34 lr: 0.000003 loss: 0.5018 (0.5147) time: 0.4436 data: 0.0002 max mem: 52126 +Epoch: [28] [2970/3634] eta: 0:05:29 lr: 0.000003 loss: 0.4864 (0.5146) time: 0.4438 data: 0.0002 max mem: 52126 +Epoch: [28] [2980/3634] eta: 0:05:25 lr: 0.000003 loss: 0.5229 (0.5147) time: 0.5010 data: 0.0001 max mem: 52126 +Epoch: [28] [2990/3634] eta: 0:05:20 lr: 0.000003 loss: 0.5476 (0.5147) time: 0.5446 data: 0.0001 max mem: 52126 +Epoch: [28] [3000/3634] eta: 0:05:14 lr: 0.000003 loss: 0.5279 (0.5147) time: 0.4251 data: 0.0001 max mem: 52126 +Epoch: [28] [3010/3634] eta: 0:05:10 lr: 0.000003 loss: 0.5164 (0.5147) time: 0.4820 data: 0.0001 max mem: 52126 +Epoch: [28] [3020/3634] eta: 0:05:04 lr: 0.000003 loss: 0.5204 (0.5147) time: 0.4824 data: 0.0002 max mem: 52126 +Epoch: [28] [3030/3634] eta: 0:04:59 lr: 0.000003 loss: 0.5503 (0.5148) time: 0.3820 data: 0.0002 max mem: 52126 +Epoch: [28] [3040/3634] eta: 0:04:54 lr: 0.000003 loss: 0.5544 (0.5149) time: 0.3822 data: 0.0001 max mem: 52126 +Epoch: [28] [3050/3634] eta: 0:04:49 lr: 0.000003 loss: 0.5140 (0.5148) time: 0.3825 data: 0.0001 max mem: 52126 +Epoch: [28] [3060/3634] eta: 0:04:44 lr: 0.000003 loss: 0.5112 (0.5149) time: 0.4294 data: 0.0001 max mem: 52126 +Epoch: [28] [3070/3634] eta: 0:04:39 lr: 0.000003 loss: 0.5101 (0.5148) time: 0.4300 data: 0.0001 max mem: 52126 +Epoch: [28] [3080/3634] eta: 0:04:34 lr: 0.000003 loss: 0.5101 (0.5148) time: 0.5034 data: 0.0001 max mem: 52126 +Epoch: [28] [3090/3634] eta: 0:04:29 lr: 0.000003 loss: 0.5262 (0.5149) time: 0.5018 data: 0.0001 max mem: 52126 +Epoch: [28] [3100/3634] eta: 0:04:24 lr: 0.000003 loss: 0.5237 (0.5148) time: 0.4292 data: 0.0001 max mem: 52126 +Epoch: [28] [3110/3634] eta: 0:04:19 lr: 0.000003 loss: 0.5347 (0.5150) time: 0.5397 data: 0.0001 max mem: 52126 +Epoch: [28] [3120/3634] eta: 0:04:14 lr: 0.000003 loss: 0.5269 (0.5149) time: 0.5992 data: 0.0001 max mem: 52126 +Epoch: [28] [3130/3634] eta: 0:04:09 lr: 0.000003 loss: 0.4912 (0.5149) time: 0.4889 data: 0.0001 max mem: 52126 +Epoch: [28] [3140/3634] eta: 0:04:04 lr: 0.000003 loss: 0.4795 (0.5147) time: 0.4404 data: 0.0002 max mem: 52126 +Epoch: [28] [3150/3634] eta: 0:03:59 lr: 0.000003 loss: 0.4454 (0.5146) time: 0.4919 data: 0.0002 max mem: 52126 +Epoch: [28] [3160/3634] eta: 0:03:54 lr: 0.000003 loss: 0.4917 (0.5146) time: 0.4325 data: 0.0001 max mem: 52126 +Epoch: [28] [3170/3634] eta: 0:03:49 lr: 0.000003 loss: 0.5375 (0.5147) time: 0.4255 data: 0.0001 max mem: 52126 +Epoch: [28] [3180/3634] eta: 0:03:44 lr: 0.000003 loss: 0.5358 (0.5147) time: 0.5307 data: 0.0001 max mem: 52126 +Epoch: [28] [3190/3634] eta: 0:03:39 lr: 0.000003 loss: 0.5462 (0.5148) time: 0.4871 data: 0.0001 max mem: 52126 +Epoch: [28] [3200/3634] eta: 0:03:34 lr: 0.000003 loss: 0.5462 (0.5148) time: 0.4743 data: 0.0001 max mem: 52126 +Epoch: [28] [3210/3634] eta: 0:03:29 lr: 0.000003 loss: 0.5120 (0.5148) time: 0.5783 data: 0.0001 max mem: 52126 +Epoch: [28] [3220/3634] eta: 0:03:25 lr: 0.000003 loss: 0.4837 (0.5147) time: 0.5456 data: 0.0001 max mem: 52126 +Epoch: [28] [3230/3634] eta: 0:03:19 lr: 0.000003 loss: 0.5284 (0.5148) time: 0.4415 data: 0.0001 max mem: 52126 +Epoch: [28] [3240/3634] eta: 0:03:14 lr: 0.000003 loss: 0.5232 (0.5147) time: 0.4460 data: 0.0001 max mem: 52126 +Epoch: [28] [3250/3634] eta: 0:03:10 lr: 0.000003 loss: 0.4896 (0.5147) time: 0.5052 data: 0.0002 max mem: 52126 +Epoch: [28] [3260/3634] eta: 0:03:04 lr: 0.000003 loss: 0.5328 (0.5146) time: 0.4416 data: 0.0002 max mem: 52126 +Epoch: [28] [3270/3634] eta: 0:03:00 lr: 0.000003 loss: 0.5328 (0.5146) time: 0.4391 data: 0.0002 max mem: 52126 +Epoch: [28] [3280/3634] eta: 0:02:55 lr: 0.000003 loss: 0.5261 (0.5146) time: 0.5447 data: 0.0001 max mem: 52126 +Epoch: [28] [3290/3634] eta: 0:02:50 lr: 0.000003 loss: 0.4783 (0.5145) time: 0.5914 data: 0.0001 max mem: 52126 +Epoch: [28] [3300/3634] eta: 0:02:45 lr: 0.000003 loss: 0.5127 (0.5145) time: 0.5297 data: 0.0002 max mem: 52126 +Epoch: [28] [3310/3634] eta: 0:02:40 lr: 0.000003 loss: 0.5233 (0.5145) time: 0.4259 data: 0.0002 max mem: 52126 +Epoch: [28] [3320/3634] eta: 0:02:35 lr: 0.000003 loss: 0.4916 (0.5145) time: 0.5118 data: 0.0002 max mem: 52126 +Epoch: [28] [3330/3634] eta: 0:02:30 lr: 0.000003 loss: 0.4916 (0.5144) time: 0.5754 data: 0.0002 max mem: 52126 +Epoch: [28] [3340/3634] eta: 0:02:25 lr: 0.000003 loss: 0.5011 (0.5143) time: 0.6243 data: 0.0001 max mem: 52126 +Epoch: [28] [3350/3634] eta: 0:02:20 lr: 0.000003 loss: 0.5210 (0.5143) time: 0.6213 data: 0.0002 max mem: 52126 +Epoch: [28] [3360/3634] eta: 0:02:15 lr: 0.000003 loss: 0.5164 (0.5143) time: 0.4422 data: 0.0002 max mem: 52126 +Epoch: [28] [3370/3634] eta: 0:02:10 lr: 0.000003 loss: 0.4921 (0.5142) time: 0.3817 data: 0.0002 max mem: 52126 +Epoch: [28] [3380/3634] eta: 0:02:05 lr: 0.000003 loss: 0.4642 (0.5141) time: 0.3811 data: 0.0002 max mem: 52126 +Epoch: [28] [3390/3634] eta: 0:02:00 lr: 0.000003 loss: 0.4918 (0.5140) time: 0.4882 data: 0.0002 max mem: 52126 +Epoch: [28] [3400/3634] eta: 0:01:55 lr: 0.000003 loss: 0.5245 (0.5141) time: 0.5320 data: 0.0002 max mem: 52126 +Epoch: [28] [3410/3634] eta: 0:01:51 lr: 0.000003 loss: 0.5477 (0.5141) time: 0.5447 data: 0.0002 max mem: 52126 +Epoch: [28] [3420/3634] eta: 0:01:46 lr: 0.000003 loss: 0.5298 (0.5141) time: 0.7145 data: 0.0001 max mem: 52126 +Epoch: [28] [3430/3634] eta: 0:01:41 lr: 0.000003 loss: 0.4623 (0.5139) time: 0.7156 data: 0.0001 max mem: 52126 +Epoch: [28] [3440/3634] eta: 0:01:36 lr: 0.000003 loss: 0.4444 (0.5138) time: 0.6493 data: 0.0001 max mem: 52126 +Epoch: [28] [3450/3634] eta: 0:01:31 lr: 0.000003 loss: 0.4887 (0.5138) time: 0.5732 data: 0.0001 max mem: 52126 +Epoch: [28] [3460/3634] eta: 0:01:26 lr: 0.000003 loss: 0.5217 (0.5138) time: 0.4257 data: 0.0001 max mem: 52126 +Epoch: [28] [3470/3634] eta: 0:01:21 lr: 0.000003 loss: 0.5291 (0.5139) time: 0.4263 data: 0.0001 max mem: 52126 +Epoch: [28] [3480/3634] eta: 0:01:16 lr: 0.000003 loss: 0.5362 (0.5138) time: 0.4271 data: 0.0001 max mem: 52126 +Epoch: [28] [3490/3634] eta: 0:01:11 lr: 0.000003 loss: 0.5116 (0.5138) time: 0.3818 data: 0.0001 max mem: 52126 +Epoch: [28] [3500/3634] eta: 0:01:06 lr: 0.000003 loss: 0.5116 (0.5138) time: 0.4420 data: 0.0002 max mem: 52126 +Epoch: [28] [3510/3634] eta: 0:01:01 lr: 0.000003 loss: 0.5269 (0.5139) time: 0.5025 data: 0.0001 max mem: 52126 +Epoch: [28] [3520/3634] eta: 0:00:56 lr: 0.000003 loss: 0.5164 (0.5138) time: 0.4423 data: 0.0001 max mem: 52126 +Epoch: [28] [3530/3634] eta: 0:00:51 lr: 0.000003 loss: 0.5223 (0.5139) time: 0.3817 data: 0.0001 max mem: 52126 +Epoch: [28] [3540/3634] eta: 0:00:46 lr: 0.000003 loss: 0.5438 (0.5139) time: 0.3816 data: 0.0001 max mem: 52126 +Epoch: [28] [3550/3634] eta: 0:00:41 lr: 0.000003 loss: 0.5434 (0.5140) time: 0.4404 data: 0.0001 max mem: 52126 +Epoch: [28] [3560/3634] eta: 0:00:36 lr: 0.000003 loss: 0.5272 (0.5140) time: 0.5776 data: 0.0001 max mem: 52126 +Epoch: [28] [3570/3634] eta: 0:00:31 lr: 0.000003 loss: 0.5428 (0.5140) time: 0.5189 data: 0.0001 max mem: 52126 +Epoch: [28] [3580/3634] eta: 0:00:26 lr: 0.000003 loss: 0.5269 (0.5140) time: 0.3812 data: 0.0001 max mem: 52126 +Epoch: [28] [3590/3634] eta: 0:00:21 lr: 0.000003 loss: 0.5470 (0.5141) time: 0.4421 data: 0.0002 max mem: 52126 +Epoch: [28] [3600/3634] eta: 0:00:16 lr: 0.000003 loss: 0.5309 (0.5141) time: 0.5029 data: 0.0002 max mem: 52126 +Epoch: [28] [3610/3634] eta: 0:00:11 lr: 0.000003 loss: 0.4835 (0.5140) time: 0.4424 data: 0.0002 max mem: 52126 +Epoch: [28] [3620/3634] eta: 0:00:06 lr: 0.000003 loss: 0.5274 (0.5141) time: 0.5485 data: 0.0020 max mem: 52126 +Epoch: [28] [3630/3634] eta: 0:00:01 lr: 0.000003 loss: 0.5061 (0.5140) time: 0.5472 data: 0.0020 max mem: 52126 +Epoch: [28] [3633/3634] eta: 0:00:00 lr: 0.000003 loss: 0.5260 (0.5141) time: 0.4888 data: 0.0020 max mem: 52126 +Epoch: [28] Total time: 0:30:00 (0.4956 s / it) +Averaged stats: lr: 0.000003 loss: 0.5260 (0.5159) +Test: [ 0/1403] eta: 1:07:11 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0914 (0.0914) acc1: 100.0000 (100.0000) acc5: 100.0000 (100.0000) time: 2.8735 data: 2.7772 max mem: 52126 +Test: [ 10/1403] eta: 0:08:05 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.1639 (0.1858) acc1: 96.8750 (95.0284) acc5: 100.0000 (100.0000) time: 0.3485 data: 0.2528 max mem: 52126 +Test: [ 20/1403] eta: 0:05:15 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1492 (0.1687) acc1: 96.8750 (95.3125) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 30/1403] eta: 0:04:14 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0937 (0.1574) acc1: 98.4375 (95.7661) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 40/1403] eta: 0:03:43 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1022 (0.1597) acc1: 98.4375 (96.0747) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 50/1403] eta: 0:03:23 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1305 (0.1796) acc1: 98.4375 (94.8836) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 60/1403] eta: 0:03:10 acc1_pos: 92.188 acc5_pos: 100.000 loss: 0.1337 (0.1762) acc1: 100.0000 (95.2357) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 70/1403] eta: 0:03:00 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.1389 (0.1787) acc1: 98.4375 (95.0704) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 80/1403] eta: 0:02:52 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1659 (0.1858) acc1: 96.8750 (94.6181) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 90/1403] eta: 0:02:46 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1821 (0.2038) acc1: 93.7500 (93.7672) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 100/1403] eta: 0:02:40 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1821 (0.2017) acc1: 95.3125 (93.8892) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 110/1403] eta: 0:02:36 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1521 (0.2000) acc1: 98.4375 (93.9893) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 120/1403] eta: 0:02:32 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0839 (0.1902) acc1: 100.0000 (94.4731) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 130/1403] eta: 0:02:29 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.0839 (0.1835) acc1: 100.0000 (94.7758) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 140/1403] eta: 0:02:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0795 (0.1760) acc1: 100.0000 (95.1463) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 150/1403] eta: 0:02:23 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0756 (0.1696) acc1: 100.0000 (95.4367) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 160/1403] eta: 0:02:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0776 (0.1646) acc1: 100.0000 (95.6522) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 170/1403] eta: 0:02:18 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0816 (0.1604) acc1: 100.0000 (95.8790) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 180/1403] eta: 0:02:16 acc1_pos: 89.062 acc5_pos: 100.000 loss: 0.0902 (0.1637) acc1: 100.0000 (95.7787) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 190/1403] eta: 0:02:14 acc1_pos: 90.625 acc5_pos: 100.000 loss: 0.2608 (0.1707) acc1: 92.1875 (95.4352) acc5: 100.0000 (100.0000) time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 200/1403] eta: 0:02:12 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.2608 (0.1810) acc1: 90.6250 (94.9471) acc5: 100.0000 (100.0000) time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 210/1403] eta: 0:02:10 acc1_pos: 78.125 acc5_pos: 100.000 loss: 0.2935 (0.1901) acc1: 85.9375 (94.4091) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 220/1403] eta: 0:02:08 acc1_pos: 90.625 acc5_pos: 100.000 loss: 0.3397 (0.1983) acc1: 85.9375 (94.0257) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0004 max mem: 52126 +Test: [ 230/1403] eta: 0:02:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2555 (0.1942) acc1: 90.6250 (94.2302) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 240/1403] eta: 0:02:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0827 (0.1897) acc1: 100.0000 (94.4567) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 250/1403] eta: 0:02:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0815 (0.1858) acc1: 100.0000 (94.6589) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 260/1403] eta: 0:02:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0843 (0.1821) acc1: 100.0000 (94.8455) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 270/1403] eta: 0:02:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0843 (0.1788) acc1: 100.0000 (95.0127) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 280/1403] eta: 0:01:58 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0851 (0.1754) acc1: 100.0000 (95.1846) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 290/1403] eta: 0:01:57 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0818 (0.1722) acc1: 100.0000 (95.3447) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 300/1403] eta: 0:01:56 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0787 (0.1693) acc1: 100.0000 (95.4942) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 310/1403] eta: 0:01:54 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0837 (0.1667) acc1: 100.0000 (95.6340) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 320/1403] eta: 0:01:53 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1003 (0.1647) acc1: 100.0000 (95.7701) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 330/1403] eta: 0:01:52 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1064 (0.1631) acc1: 100.0000 (95.8978) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 340/1403] eta: 0:01:50 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1064 (0.1613) acc1: 100.0000 (96.0181) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 350/1403] eta: 0:01:49 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1041 (0.1597) acc1: 100.0000 (96.1316) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 360/1403] eta: 0:01:48 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1066 (0.1584) acc1: 100.0000 (96.2344) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 370/1403] eta: 0:01:46 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1058 (0.1569) acc1: 100.0000 (96.3359) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 380/1403] eta: 0:01:45 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1047 (0.1556) acc1: 100.0000 (96.4321) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 390/1403] eta: 0:01:44 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1059 (0.1559) acc1: 100.0000 (96.4114) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 400/1403] eta: 0:01:43 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2319 (0.1589) acc1: 90.6250 (96.2282) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [ 410/1403] eta: 0:01:42 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2598 (0.1612) acc1: 89.0625 (96.0880) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 420/1403] eta: 0:01:40 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2280 (0.1623) acc1: 92.1875 (96.0028) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 430/1403] eta: 0:01:39 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2190 (0.1646) acc1: 90.6250 (95.8599) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 440/1403] eta: 0:01:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2792 (0.1674) acc1: 89.0625 (95.7129) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 450/1403] eta: 0:01:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2558 (0.1691) acc1: 89.0625 (95.6070) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 460/1403] eta: 0:01:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2264 (0.1709) acc1: 90.6250 (95.4955) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [ 470/1403] eta: 0:01:35 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2048 (0.1699) acc1: 93.7500 (95.5248) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0960 data: 0.0002 max mem: 52126 +Test: [ 480/1403] eta: 0:01:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0639 (0.1677) acc1: 100.0000 (95.6179) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0960 data: 0.0002 max mem: 52126 +Test: [ 490/1403] eta: 0:01:32 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0639 (0.1656) acc1: 100.0000 (95.7071) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0960 data: 0.0002 max mem: 52126 +Test: [ 500/1403] eta: 0:01:31 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0642 (0.1636) acc1: 100.0000 (95.7928) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 510/1403] eta: 0:01:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0643 (0.1617) acc1: 100.0000 (95.8751) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 520/1403] eta: 0:01:29 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0686 (0.1600) acc1: 100.0000 (95.9543) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0960 data: 0.0002 max mem: 52126 +Test: [ 530/1403] eta: 0:01:28 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0598 (0.1581) acc1: 100.0000 (96.0305) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [ 540/1403] eta: 0:01:27 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0590 (0.1562) acc1: 100.0000 (96.1039) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [ 550/1403] eta: 0:01:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0595 (0.1549) acc1: 100.0000 (96.1632) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0960 data: 0.0002 max mem: 52126 +Test: [ 560/1403] eta: 0:01:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0814 (0.1538) acc1: 98.4375 (96.2066) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0960 data: 0.0002 max mem: 52126 +Test: [ 570/1403] eta: 0:01:24 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0822 (0.1527) acc1: 98.4375 (96.2566) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [ 580/1403] eta: 0:01:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0897 (0.1520) acc1: 98.4375 (96.2780) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0960 data: 0.0002 max mem: 52126 +Test: [ 590/1403] eta: 0:01:21 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1060 (0.1516) acc1: 98.4375 (96.2828) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0960 data: 0.0002 max mem: 52126 +Test: [ 600/1403] eta: 0:01:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1209 (0.1518) acc1: 96.8750 (96.2432) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [ 610/1403] eta: 0:01:19 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1209 (0.1512) acc1: 96.8750 (96.2587) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0960 data: 0.0002 max mem: 52126 +Test: [ 620/1403] eta: 0:01:18 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1282 (0.1519) acc1: 95.3125 (96.1831) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0961 data: 0.0002 max mem: 52126 +Test: [ 630/1403] eta: 0:01:17 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1377 (0.1515) acc1: 95.3125 (96.1866) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0960 data: 0.0002 max mem: 52126 +Test: [ 640/1403] eta: 0:01:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1160 (0.1510) acc1: 96.8750 (96.2095) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0960 data: 0.0002 max mem: 52126 +Test: [ 650/1403] eta: 0:01:15 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1160 (0.1505) acc1: 98.4375 (96.2318) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0960 data: 0.0002 max mem: 52126 +Test: [ 660/1403] eta: 0:01:14 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0946 (0.1495) acc1: 100.0000 (96.2769) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0960 data: 0.0002 max mem: 52126 +Test: [ 670/1403] eta: 0:01:13 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0682 (0.1483) acc1: 100.0000 (96.3301) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0960 data: 0.0002 max mem: 52126 +Test: [ 680/1403] eta: 0:01:12 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0769 (0.1495) acc1: 98.4375 (96.2417) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [ 690/1403] eta: 0:01:11 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2478 (0.1533) acc1: 89.0625 (96.0225) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [ 700/1403] eta: 0:01:10 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1757 (0.1534) acc1: 90.6250 (96.0146) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [ 710/1403] eta: 0:01:09 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2449 (0.1571) acc1: 89.0625 (95.7960) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [ 720/1403] eta: 0:01:08 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2449 (0.1570) acc1: 85.9375 (95.8023) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0960 data: 0.0002 max mem: 52126 +Test: [ 730/1403] eta: 0:01:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0670 (0.1559) acc1: 100.0000 (95.8533) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0960 data: 0.0002 max mem: 52126 +Test: [ 740/1403] eta: 0:01:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0738 (0.1550) acc1: 100.0000 (95.8924) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [ 750/1403] eta: 0:01:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0990 (0.1561) acc1: 98.4375 (95.8326) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [ 760/1403] eta: 0:01:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2951 (0.1592) acc1: 85.9375 (95.6328) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.0960 data: 0.0002 max mem: 52126 +Test: [ 770/1403] eta: 0:01:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4143 (0.1634) acc1: 76.5625 (95.3733) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [ 780/1403] eta: 0:01:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4120 (0.1657) acc1: 79.6875 (95.2405) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 790/1403] eta: 0:01:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2927 (0.1680) acc1: 87.5000 (95.1169) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 800/1403] eta: 0:00:59 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2272 (0.1683) acc1: 89.0625 (95.1018) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 810/1403] eta: 0:00:58 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2203 (0.1710) acc1: 89.0625 (94.9619) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.0960 data: 0.0002 max mem: 52126 +Test: [ 820/1403] eta: 0:00:57 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6035 (0.1794) acc1: 71.8750 (94.5912) acc5: 100.0000 (100.0000) acc1_neg: 40.625 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [ 830/1403] eta: 0:00:56 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6593 (0.1824) acc1: 70.3125 (94.4344) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [ 840/1403] eta: 0:00:55 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5159 (0.1904) acc1: 71.8750 (94.0807) acc5: 100.0000 (100.0000) acc1_neg: 50.000 acc5_neg: 100.000 time: 0.0960 data: 0.0002 max mem: 52126 +Test: [ 850/1403] eta: 0:00:54 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.0055 (0.2010) acc1: 56.2500 (93.6178) acc5: 100.0000 (100.0000) acc1_neg: 56.250 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [ 860/1403] eta: 0:00:53 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.9725 (0.2044) acc1: 57.8125 (93.4669) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 870/1403] eta: 0:00:52 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1216 (0.2032) acc1: 96.8750 (93.5204) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [ 880/1403] eta: 0:00:51 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0907 (0.2020) acc1: 98.4375 (93.5744) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [ 890/1403] eta: 0:00:50 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0786 (0.2007) acc1: 98.4375 (93.6395) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0960 data: 0.0002 max mem: 52126 +Test: [ 900/1403] eta: 0:00:49 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0729 (0.1992) acc1: 100.0000 (93.7084) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [ 910/1403] eta: 0:00:48 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0744 (0.1983) acc1: 100.0000 (93.7551) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [ 920/1403] eta: 0:00:47 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0978 (0.1973) acc1: 98.4375 (93.8009) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0960 data: 0.0002 max mem: 52126 +Test: [ 930/1403] eta: 0:00:46 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0881 (0.1961) acc1: 98.4375 (93.8557) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0960 data: 0.0002 max mem: 52126 +Test: [ 940/1403] eta: 0:00:45 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1066 (0.1962) acc1: 98.4375 (93.8347) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [ 950/1403] eta: 0:00:44 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2376 (0.1969) acc1: 89.0625 (93.7796) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [ 960/1403] eta: 0:00:43 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2580 (0.1974) acc1: 89.0625 (93.7321) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [ 970/1403] eta: 0:00:42 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2410 (0.1982) acc1: 89.0625 (93.6840) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [ 980/1403] eta: 0:00:41 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2515 (0.1989) acc1: 90.6250 (93.6258) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [ 990/1403] eta: 0:00:40 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2437 (0.1992) acc1: 90.6250 (93.6049) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [1000/1403] eta: 0:00:39 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2191 (0.1998) acc1: 92.1875 (93.5767) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0960 data: 0.0002 max mem: 52126 +Test: [1010/1403] eta: 0:00:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2475 (0.2005) acc1: 90.6250 (93.5352) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [1020/1403] eta: 0:00:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4012 (0.2062) acc1: 81.2500 (93.2128) acc5: 100.0000 (100.0000) acc1_neg: 35.938 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [1030/1403] eta: 0:00:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5173 (0.2078) acc1: 67.1875 (93.1135) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [1040/1403] eta: 0:00:35 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2068 (0.2074) acc1: 90.6250 (93.1256) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0959 data: 0.0003 max mem: 52126 +Test: [1050/1403] eta: 0:00:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1498 (0.2070) acc1: 95.3125 (93.1464) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [1060/1403] eta: 0:00:33 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1165 (0.2063) acc1: 96.8750 (93.1845) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0959 data: 0.0003 max mem: 52126 +Test: [1070/1403] eta: 0:00:32 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1421 (0.2066) acc1: 96.8750 (93.1737) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0959 data: 0.0003 max mem: 52126 +Test: [1080/1403] eta: 0:00:31 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3142 (0.2080) acc1: 87.5000 (93.0909) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [1090/1403] eta: 0:00:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3482 (0.2093) acc1: 84.3750 (93.0225) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0959 data: 0.0003 max mem: 52126 +Test: [1100/1403] eta: 0:00:29 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3358 (0.2099) acc1: 87.5000 (92.9922) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0960 data: 0.0003 max mem: 52126 +Test: [1110/1403] eta: 0:00:28 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2770 (0.2105) acc1: 90.6250 (92.9652) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0960 data: 0.0002 max mem: 52126 +Test: [1120/1403] eta: 0:00:27 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2770 (0.2110) acc1: 90.6250 (92.9332) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0960 data: 0.0003 max mem: 52126 +Test: [1130/1403] eta: 0:00:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2828 (0.2118) acc1: 87.5000 (92.8893) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0960 data: 0.0003 max mem: 52126 +Test: [1140/1403] eta: 0:00:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2383 (0.2120) acc1: 89.0625 (92.8722) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0960 data: 0.0003 max mem: 52126 +Test: [1150/1403] eta: 0:00:24 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2383 (0.2125) acc1: 90.6250 (92.8473) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [1160/1403] eta: 0:00:23 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2599 (0.2130) acc1: 90.6250 (92.8187) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1170/1403] eta: 0:00:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2422 (0.2129) acc1: 90.6250 (92.8186) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [1180/1403] eta: 0:00:21 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0696 (0.2117) acc1: 100.0000 (92.8794) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [1190/1403] eta: 0:00:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0685 (0.2105) acc1: 100.0000 (92.9392) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1200/1403] eta: 0:00:19 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0690 (0.2093) acc1: 100.0000 (92.9980) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1210/1403] eta: 0:00:18 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0678 (0.2082) acc1: 100.0000 (93.0558) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [1220/1403] eta: 0:00:17 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0677 (0.2070) acc1: 100.0000 (93.1127) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [1230/1403] eta: 0:00:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0685 (0.2059) acc1: 100.0000 (93.1687) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [1240/1403] eta: 0:00:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0685 (0.2048) acc1: 100.0000 (93.2237) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [1250/1403] eta: 0:00:15 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0687 (0.2037) acc1: 100.0000 (93.2779) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0959 data: 0.0003 max mem: 52126 +Test: [1260/1403] eta: 0:00:14 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0749 (0.2027) acc1: 100.0000 (93.3312) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0959 data: 0.0003 max mem: 52126 +Test: [1270/1403] eta: 0:00:13 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0763 (0.2018) acc1: 100.0000 (93.3837) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [1280/1403] eta: 0:00:12 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0763 (0.2008) acc1: 100.0000 (93.4353) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0959 data: 0.0003 max mem: 52126 +Test: [1290/1403] eta: 0:00:11 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0756 (0.1998) acc1: 100.0000 (93.4862) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [1300/1403] eta: 0:00:10 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0763 (0.1989) acc1: 100.0000 (93.5362) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [1310/1403] eta: 0:00:09 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0765 (0.1979) acc1: 100.0000 (93.5855) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1320/1403] eta: 0:00:08 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0747 (0.1970) acc1: 100.0000 (93.6341) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1330/1403] eta: 0:00:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0743 (0.1961) acc1: 100.0000 (93.6796) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0959 data: 0.0003 max mem: 52126 +Test: [1340/1403] eta: 0:00:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0671 (0.1951) acc1: 100.0000 (93.7244) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0960 data: 0.0002 max mem: 52126 +Test: [1350/1403] eta: 0:00:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0651 (0.1942) acc1: 100.0000 (93.7673) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0960 data: 0.0002 max mem: 52126 +Test: [1360/1403] eta: 0:00:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0654 (0.1933) acc1: 100.0000 (93.8120) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [1370/1403] eta: 0:00:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0656 (0.1924) acc1: 100.0000 (93.8537) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1380/1403] eta: 0:00:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0649 (0.1915) acc1: 100.0000 (93.8982) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1390/1403] eta: 0:00:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0643 (0.1906) acc1: 100.0000 (93.9421) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1400/1403] eta: 0:00:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0651 (0.1897) acc1: 100.0000 (93.9820) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0957 data: 0.0001 max mem: 52126 +Test: [1402/1403] eta: 0:00:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0653 (0.1896) acc1: 100.0000 (93.9863) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0916 data: 0.0001 max mem: 52126 +Test: Total time: 0:02:17 (0.0983 s / it) +* Overall Acc@1 93.986 Acc@5 100.000 + target == real Acc@1 95.595 Acc@5 100.000 + target == fake Acc@1 93.533 Acc@5 100.000 + loss 0.190 +Accuracy of the network on the 89729 test images: 94.0% +Max accuracy: 94.13% +Epoch: [29] [ 0/3634] eta: 6:32:59 lr: 0.000003 loss: 0.4660 (0.4660) time: 6.4887 data: 4.2693 max mem: 52126 +Epoch: [29] [ 10/3634] eta: 1:07:01 lr: 0.000003 loss: 0.5332 (0.5107) time: 1.1098 data: 0.3883 max mem: 52126 +Epoch: [29] [ 20/3634] eta: 0:45:58 lr: 0.000003 loss: 0.5368 (0.5319) time: 0.4769 data: 0.0002 max mem: 52126 +Epoch: [29] [ 30/3634] eta: 0:40:48 lr: 0.000003 loss: 0.5636 (0.5324) time: 0.4425 data: 0.0002 max mem: 52126 +Epoch: [29] [ 40/3634] eta: 0:36:23 lr: 0.000003 loss: 0.5150 (0.5224) time: 0.4439 data: 0.0002 max mem: 52126 +Epoch: [29] [ 50/3634] eta: 0:33:37 lr: 0.000003 loss: 0.4771 (0.5104) time: 0.3827 data: 0.0002 max mem: 52126 +Epoch: [29] [ 60/3634] eta: 0:32:57 lr: 0.000003 loss: 0.4800 (0.5089) time: 0.4423 data: 0.0002 max mem: 52126 +Epoch: [29] [ 70/3634] eta: 0:32:13 lr: 0.000003 loss: 0.5462 (0.5159) time: 0.4905 data: 0.0002 max mem: 52126 +Epoch: [29] [ 80/3634] eta: 0:31:38 lr: 0.000003 loss: 0.5125 (0.5108) time: 0.4763 data: 0.0002 max mem: 52126 +Epoch: [29] [ 90/3634] eta: 0:31:21 lr: 0.000003 loss: 0.5129 (0.5140) time: 0.4888 data: 0.0002 max mem: 52126 +Epoch: [29] [ 100/3634] eta: 0:30:56 lr: 0.000003 loss: 0.5240 (0.5159) time: 0.4893 data: 0.0002 max mem: 52126 +Epoch: [29] [ 110/3634] eta: 0:30:06 lr: 0.000003 loss: 0.5059 (0.5123) time: 0.4307 data: 0.0002 max mem: 52126 +Epoch: [29] [ 120/3634] eta: 0:30:25 lr: 0.000003 loss: 0.5273 (0.5150) time: 0.4901 data: 0.0002 max mem: 52126 +Epoch: [29] [ 130/3634] eta: 0:30:16 lr: 0.000003 loss: 0.5356 (0.5164) time: 0.5500 data: 0.0002 max mem: 52126 +Epoch: [29] [ 140/3634] eta: 0:29:37 lr: 0.000003 loss: 0.5145 (0.5140) time: 0.4433 data: 0.0002 max mem: 52126 +Epoch: [29] [ 150/3634] eta: 0:29:46 lr: 0.000003 loss: 0.5027 (0.5145) time: 0.4762 data: 0.0002 max mem: 52126 +Epoch: [29] [ 160/3634] eta: 0:29:58 lr: 0.000003 loss: 0.4861 (0.5122) time: 0.5811 data: 0.0001 max mem: 52126 +Epoch: [29] [ 170/3634] eta: 0:30:59 lr: 0.000003 loss: 0.5035 (0.5125) time: 0.7171 data: 0.0002 max mem: 52126 +Epoch: [29] [ 180/3634] eta: 0:30:59 lr: 0.000003 loss: 0.5296 (0.5140) time: 0.7048 data: 0.0002 max mem: 52126 +Epoch: [29] [ 190/3634] eta: 0:30:26 lr: 0.000003 loss: 0.5216 (0.5140) time: 0.4743 data: 0.0002 max mem: 52126 +Epoch: [29] [ 200/3634] eta: 0:30:33 lr: 0.000003 loss: 0.5235 (0.5151) time: 0.4931 data: 0.0002 max mem: 52126 +Epoch: [29] [ 210/3634] eta: 0:30:43 lr: 0.000003 loss: 0.5263 (0.5151) time: 0.6173 data: 0.0002 max mem: 52126 +Epoch: [29] [ 220/3634] eta: 0:30:47 lr: 0.000003 loss: 0.5013 (0.5138) time: 0.6127 data: 0.0002 max mem: 52126 +Epoch: [29] [ 230/3634] eta: 0:30:18 lr: 0.000003 loss: 0.4928 (0.5123) time: 0.4882 data: 0.0002 max mem: 52126 +Epoch: [29] [ 240/3634] eta: 0:30:08 lr: 0.000003 loss: 0.4956 (0.5121) time: 0.4421 data: 0.0002 max mem: 52126 +Epoch: [29] [ 250/3634] eta: 0:30:13 lr: 0.000003 loss: 0.5108 (0.5127) time: 0.5550 data: 0.0002 max mem: 52126 +Epoch: [29] [ 260/3634] eta: 0:29:47 lr: 0.000003 loss: 0.5248 (0.5127) time: 0.4942 data: 0.0002 max mem: 52126 +Epoch: [29] [ 270/3634] eta: 0:29:47 lr: 0.000003 loss: 0.5268 (0.5133) time: 0.4751 data: 0.0002 max mem: 52126 +Epoch: [29] [ 280/3634] eta: 0:29:53 lr: 0.000003 loss: 0.5611 (0.5132) time: 0.5974 data: 0.0002 max mem: 52126 +Epoch: [29] [ 290/3634] eta: 0:29:30 lr: 0.000003 loss: 0.4893 (0.5116) time: 0.5040 data: 0.0002 max mem: 52126 +Epoch: [29] [ 300/3634] eta: 0:30:00 lr: 0.000003 loss: 0.5017 (0.5126) time: 0.6152 data: 0.0002 max mem: 52126 +Epoch: [29] [ 310/3634] eta: 0:30:11 lr: 0.000003 loss: 0.5149 (0.5111) time: 0.7696 data: 0.0002 max mem: 52126 +Epoch: [29] [ 320/3634] eta: 0:29:58 lr: 0.000003 loss: 0.4884 (0.5105) time: 0.5829 data: 0.0002 max mem: 52126 +Epoch: [29] [ 330/3634] eta: 0:29:36 lr: 0.000003 loss: 0.5056 (0.5117) time: 0.4284 data: 0.0002 max mem: 52126 +Epoch: [29] [ 340/3634] eta: 0:29:39 lr: 0.000003 loss: 0.5324 (0.5109) time: 0.5025 data: 0.0002 max mem: 52126 +Epoch: [29] [ 350/3634] eta: 0:29:30 lr: 0.000003 loss: 0.5125 (0.5110) time: 0.5632 data: 0.0002 max mem: 52126 +Epoch: [29] [ 360/3634] eta: 0:29:22 lr: 0.000003 loss: 0.5295 (0.5109) time: 0.5034 data: 0.0002 max mem: 52126 +Epoch: [29] [ 370/3634] eta: 0:29:24 lr: 0.000003 loss: 0.5443 (0.5115) time: 0.5663 data: 0.0002 max mem: 52126 +Epoch: [29] [ 380/3634] eta: 0:29:14 lr: 0.000003 loss: 0.5522 (0.5125) time: 0.5531 data: 0.0002 max mem: 52126 +Epoch: [29] [ 390/3634] eta: 0:29:14 lr: 0.000003 loss: 0.5451 (0.5121) time: 0.5418 data: 0.0002 max mem: 52126 +Epoch: [29] [ 400/3634] eta: 0:28:56 lr: 0.000003 loss: 0.5341 (0.5120) time: 0.4940 data: 0.0002 max mem: 52126 +Epoch: [29] [ 410/3634] eta: 0:28:38 lr: 0.000003 loss: 0.5252 (0.5116) time: 0.3816 data: 0.0002 max mem: 52126 +Epoch: [29] [ 420/3634] eta: 0:28:28 lr: 0.000003 loss: 0.5313 (0.5122) time: 0.4276 data: 0.0002 max mem: 52126 +Epoch: [29] [ 430/3634] eta: 0:28:19 lr: 0.000003 loss: 0.5314 (0.5125) time: 0.4736 data: 0.0002 max mem: 52126 +Epoch: [29] [ 440/3634] eta: 0:28:03 lr: 0.000003 loss: 0.5283 (0.5124) time: 0.4280 data: 0.0002 max mem: 52126 +Epoch: [29] [ 450/3634] eta: 0:27:54 lr: 0.000003 loss: 0.5353 (0.5130) time: 0.4301 data: 0.0002 max mem: 52126 +Epoch: [29] [ 460/3634] eta: 0:27:39 lr: 0.000003 loss: 0.5471 (0.5139) time: 0.4297 data: 0.0002 max mem: 52126 +Epoch: [29] [ 470/3634] eta: 0:27:31 lr: 0.000003 loss: 0.5558 (0.5152) time: 0.4313 data: 0.0002 max mem: 52126 +Epoch: [29] [ 480/3634] eta: 0:27:23 lr: 0.000003 loss: 0.5371 (0.5153) time: 0.4809 data: 0.0002 max mem: 52126 +Epoch: [29] [ 490/3634] eta: 0:27:16 lr: 0.000003 loss: 0.5093 (0.5150) time: 0.4921 data: 0.0002 max mem: 52126 +Epoch: [29] [ 500/3634] eta: 0:27:02 lr: 0.000003 loss: 0.4785 (0.5146) time: 0.4431 data: 0.0002 max mem: 52126 +Epoch: [29] [ 510/3634] eta: 0:26:57 lr: 0.000003 loss: 0.5076 (0.5154) time: 0.4439 data: 0.0002 max mem: 52126 +Epoch: [29] [ 520/3634] eta: 0:26:49 lr: 0.000003 loss: 0.5586 (0.5155) time: 0.4910 data: 0.0002 max mem: 52126 +Epoch: [29] [ 530/3634] eta: 0:26:42 lr: 0.000003 loss: 0.5250 (0.5161) time: 0.4783 data: 0.0002 max mem: 52126 +Epoch: [29] [ 540/3634] eta: 0:26:29 lr: 0.000003 loss: 0.5546 (0.5167) time: 0.4309 data: 0.0002 max mem: 52126 +Epoch: [29] [ 550/3634] eta: 0:26:33 lr: 0.000003 loss: 0.5031 (0.5167) time: 0.5308 data: 0.0002 max mem: 52126 +Epoch: [29] [ 560/3634] eta: 0:26:25 lr: 0.000003 loss: 0.5119 (0.5168) time: 0.5753 data: 0.0002 max mem: 52126 +Epoch: [29] [ 570/3634] eta: 0:26:13 lr: 0.000003 loss: 0.5229 (0.5165) time: 0.4263 data: 0.0002 max mem: 52126 +Epoch: [29] [ 580/3634] eta: 0:26:01 lr: 0.000003 loss: 0.5112 (0.5166) time: 0.3814 data: 0.0002 max mem: 52126 +Epoch: [29] [ 590/3634] eta: 0:26:00 lr: 0.000003 loss: 0.5167 (0.5169) time: 0.4857 data: 0.0002 max mem: 52126 +Epoch: [29] [ 600/3634] eta: 0:25:54 lr: 0.000003 loss: 0.5180 (0.5170) time: 0.5458 data: 0.0002 max mem: 52126 +Epoch: [29] [ 610/3634] eta: 0:25:47 lr: 0.000003 loss: 0.5125 (0.5165) time: 0.4858 data: 0.0002 max mem: 52126 +Epoch: [29] [ 620/3634] eta: 0:25:36 lr: 0.000003 loss: 0.5048 (0.5163) time: 0.4276 data: 0.0002 max mem: 52126 +Epoch: [29] [ 630/3634] eta: 0:25:24 lr: 0.000003 loss: 0.5023 (0.5161) time: 0.3836 data: 0.0002 max mem: 52126 +Epoch: [29] [ 640/3634] eta: 0:25:14 lr: 0.000003 loss: 0.5272 (0.5166) time: 0.3826 data: 0.0002 max mem: 52126 +Epoch: [29] [ 650/3634] eta: 0:25:07 lr: 0.000003 loss: 0.5499 (0.5171) time: 0.4284 data: 0.0002 max mem: 52126 +Epoch: [29] [ 660/3634] eta: 0:25:06 lr: 0.000003 loss: 0.5451 (0.5173) time: 0.5320 data: 0.0002 max mem: 52126 +Epoch: [29] [ 670/3634] eta: 0:25:01 lr: 0.000003 loss: 0.5305 (0.5171) time: 0.5466 data: 0.0002 max mem: 52126 +Epoch: [29] [ 680/3634] eta: 0:24:50 lr: 0.000003 loss: 0.5209 (0.5170) time: 0.4424 data: 0.0002 max mem: 52126 +Epoch: [29] [ 690/3634] eta: 0:24:43 lr: 0.000003 loss: 0.4907 (0.5164) time: 0.4255 data: 0.0002 max mem: 52126 +Epoch: [29] [ 700/3634] eta: 0:24:42 lr: 0.000003 loss: 0.4851 (0.5162) time: 0.5307 data: 0.0002 max mem: 52126 +Epoch: [29] [ 710/3634] eta: 0:24:41 lr: 0.000003 loss: 0.5340 (0.5168) time: 0.5924 data: 0.0002 max mem: 52126 +Epoch: [29] [ 720/3634] eta: 0:24:40 lr: 0.000003 loss: 0.5425 (0.5171) time: 0.6089 data: 0.0002 max mem: 52126 +Epoch: [29] [ 730/3634] eta: 0:24:34 lr: 0.000003 loss: 0.5367 (0.5170) time: 0.5478 data: 0.0002 max mem: 52126 +Epoch: [29] [ 740/3634] eta: 0:24:31 lr: 0.000003 loss: 0.5227 (0.5166) time: 0.5177 data: 0.0002 max mem: 52126 +Epoch: [29] [ 750/3634] eta: 0:24:26 lr: 0.000003 loss: 0.5032 (0.5164) time: 0.5338 data: 0.0002 max mem: 52126 +Epoch: [29] [ 760/3634] eta: 0:24:30 lr: 0.000003 loss: 0.5151 (0.5162) time: 0.6262 data: 0.0002 max mem: 52126 +Epoch: [29] [ 770/3634] eta: 0:24:24 lr: 0.000003 loss: 0.5310 (0.5163) time: 0.6256 data: 0.0002 max mem: 52126 +Epoch: [29] [ 780/3634] eta: 0:24:19 lr: 0.000003 loss: 0.5379 (0.5163) time: 0.5027 data: 0.0002 max mem: 52126 +Epoch: [29] [ 790/3634] eta: 0:24:25 lr: 0.000003 loss: 0.4473 (0.5152) time: 0.6582 data: 0.0002 max mem: 52126 +Epoch: [29] [ 800/3634] eta: 0:24:23 lr: 0.000003 loss: 0.5245 (0.5156) time: 0.7172 data: 0.0002 max mem: 52126 +Epoch: [29] [ 810/3634] eta: 0:24:13 lr: 0.000003 loss: 0.5245 (0.5155) time: 0.5022 data: 0.0002 max mem: 52126 +Epoch: [29] [ 820/3634] eta: 0:24:04 lr: 0.000003 loss: 0.5111 (0.5157) time: 0.3821 data: 0.0002 max mem: 52126 +Epoch: [29] [ 830/3634] eta: 0:23:57 lr: 0.000003 loss: 0.5425 (0.5161) time: 0.4246 data: 0.0002 max mem: 52126 +Epoch: [29] [ 840/3634] eta: 0:23:51 lr: 0.000003 loss: 0.5553 (0.5165) time: 0.4834 data: 0.0002 max mem: 52126 +Epoch: [29] [ 850/3634] eta: 0:23:46 lr: 0.000003 loss: 0.5436 (0.5167) time: 0.4993 data: 0.0002 max mem: 52126 +Epoch: [29] [ 860/3634] eta: 0:23:37 lr: 0.000003 loss: 0.5346 (0.5166) time: 0.4426 data: 0.0002 max mem: 52126 +Epoch: [29] [ 870/3634] eta: 0:23:34 lr: 0.000003 loss: 0.4777 (0.5161) time: 0.4894 data: 0.0002 max mem: 52126 +Epoch: [29] [ 880/3634] eta: 0:23:31 lr: 0.000003 loss: 0.4777 (0.5157) time: 0.5911 data: 0.0002 max mem: 52126 +Epoch: [29] [ 890/3634] eta: 0:23:35 lr: 0.000003 loss: 0.4771 (0.5151) time: 0.6923 data: 0.0002 max mem: 52126 +Epoch: [29] [ 900/3634] eta: 0:23:29 lr: 0.000003 loss: 0.4771 (0.5150) time: 0.6488 data: 0.0002 max mem: 52126 +Epoch: [29] [ 910/3634] eta: 0:23:20 lr: 0.000003 loss: 0.4874 (0.5149) time: 0.4423 data: 0.0002 max mem: 52126 +Epoch: [29] [ 920/3634] eta: 0:23:14 lr: 0.000003 loss: 0.4998 (0.5147) time: 0.4259 data: 0.0002 max mem: 52126 +Epoch: [29] [ 930/3634] eta: 0:23:07 lr: 0.000003 loss: 0.5137 (0.5147) time: 0.4696 data: 0.0002 max mem: 52126 +Epoch: [29] [ 940/3634] eta: 0:23:01 lr: 0.000003 loss: 0.5174 (0.5146) time: 0.4701 data: 0.0002 max mem: 52126 +Epoch: [29] [ 950/3634] eta: 0:22:56 lr: 0.000003 loss: 0.5360 (0.5149) time: 0.4850 data: 0.0002 max mem: 52126 +Epoch: [29] [ 960/3634] eta: 0:22:47 lr: 0.000003 loss: 0.5422 (0.5150) time: 0.4426 data: 0.0002 max mem: 52126 +Epoch: [29] [ 970/3634] eta: 0:22:38 lr: 0.000003 loss: 0.5182 (0.5147) time: 0.3840 data: 0.0002 max mem: 52126 +Epoch: [29] [ 980/3634] eta: 0:22:32 lr: 0.000003 loss: 0.5173 (0.5150) time: 0.4257 data: 0.0002 max mem: 52126 +Epoch: [29] [ 990/3634] eta: 0:22:24 lr: 0.000003 loss: 0.5210 (0.5147) time: 0.4254 data: 0.0002 max mem: 52126 +Epoch: [29] [1000/3634] eta: 0:22:15 lr: 0.000003 loss: 0.5598 (0.5154) time: 0.3817 data: 0.0002 max mem: 52126 +Epoch: [29] [1010/3634] eta: 0:22:15 lr: 0.000003 loss: 0.5659 (0.5153) time: 0.5328 data: 0.0002 max mem: 52126 +Epoch: [29] [1020/3634] eta: 0:22:11 lr: 0.000003 loss: 0.4972 (0.5148) time: 0.6202 data: 0.0002 max mem: 52126 +Epoch: [29] [1030/3634] eta: 0:22:02 lr: 0.000003 loss: 0.4844 (0.5149) time: 0.4691 data: 0.0002 max mem: 52126 +Epoch: [29] [1040/3634] eta: 0:21:54 lr: 0.000003 loss: 0.5332 (0.5148) time: 0.3814 data: 0.0002 max mem: 52126 +Epoch: [29] [1050/3634] eta: 0:21:46 lr: 0.000003 loss: 0.5161 (0.5147) time: 0.3817 data: 0.0002 max mem: 52126 +Epoch: [29] [1060/3634] eta: 0:21:46 lr: 0.000003 loss: 0.5085 (0.5143) time: 0.5465 data: 0.0002 max mem: 52126 +Epoch: [29] [1070/3634] eta: 0:21:44 lr: 0.000003 loss: 0.4917 (0.5141) time: 0.6651 data: 0.0002 max mem: 52126 +Epoch: [29] [1080/3634] eta: 0:21:36 lr: 0.000003 loss: 0.4998 (0.5141) time: 0.5010 data: 0.0002 max mem: 52126 +Epoch: [29] [1090/3634] eta: 0:21:28 lr: 0.000003 loss: 0.5206 (0.5143) time: 0.3828 data: 0.0002 max mem: 52126 +Epoch: [29] [1100/3634] eta: 0:21:25 lr: 0.000003 loss: 0.5422 (0.5144) time: 0.4892 data: 0.0002 max mem: 52126 +Epoch: [29] [1110/3634] eta: 0:21:17 lr: 0.000003 loss: 0.5661 (0.5146) time: 0.4889 data: 0.0002 max mem: 52126 +Epoch: [29] [1120/3634] eta: 0:21:09 lr: 0.000003 loss: 0.5250 (0.5143) time: 0.3820 data: 0.0002 max mem: 52126 +Epoch: [29] [1130/3634] eta: 0:21:03 lr: 0.000003 loss: 0.5001 (0.5144) time: 0.4266 data: 0.0002 max mem: 52126 +Epoch: [29] [1140/3634] eta: 0:20:55 lr: 0.000003 loss: 0.5387 (0.5147) time: 0.4269 data: 0.0002 max mem: 52126 +Epoch: [29] [1150/3634] eta: 0:20:48 lr: 0.000003 loss: 0.5454 (0.5147) time: 0.3824 data: 0.0002 max mem: 52126 +Epoch: [29] [1160/3634] eta: 0:20:45 lr: 0.000003 loss: 0.5221 (0.5146) time: 0.4883 data: 0.0002 max mem: 52126 +Epoch: [29] [1170/3634] eta: 0:20:40 lr: 0.000003 loss: 0.5310 (0.5146) time: 0.5483 data: 0.0002 max mem: 52126 +Epoch: [29] [1180/3634] eta: 0:20:36 lr: 0.000003 loss: 0.5290 (0.5146) time: 0.5448 data: 0.0002 max mem: 52126 +Epoch: [29] [1190/3634] eta: 0:20:30 lr: 0.000003 loss: 0.5307 (0.5147) time: 0.5287 data: 0.0002 max mem: 52126 +Epoch: [29] [1200/3634] eta: 0:20:29 lr: 0.000003 loss: 0.5307 (0.5147) time: 0.5766 data: 0.0002 max mem: 52126 +Epoch: [29] [1210/3634] eta: 0:20:24 lr: 0.000003 loss: 0.5360 (0.5149) time: 0.5913 data: 0.0002 max mem: 52126 +Epoch: [29] [1220/3634] eta: 0:20:23 lr: 0.000003 loss: 0.5222 (0.5149) time: 0.6085 data: 0.0002 max mem: 52126 +Epoch: [29] [1230/3634] eta: 0:20:21 lr: 0.000003 loss: 0.5172 (0.5149) time: 0.6965 data: 0.0002 max mem: 52126 +Epoch: [29] [1240/3634] eta: 0:20:16 lr: 0.000003 loss: 0.5172 (0.5150) time: 0.5743 data: 0.0002 max mem: 52126 +Epoch: [29] [1250/3634] eta: 0:20:14 lr: 0.000003 loss: 0.5061 (0.5147) time: 0.5956 data: 0.0002 max mem: 52126 +Epoch: [29] [1260/3634] eta: 0:20:11 lr: 0.000003 loss: 0.4906 (0.5145) time: 0.6566 data: 0.0002 max mem: 52126 +Epoch: [29] [1270/3634] eta: 0:20:06 lr: 0.000003 loss: 0.5340 (0.5148) time: 0.5477 data: 0.0002 max mem: 52126 +Epoch: [29] [1280/3634] eta: 0:20:01 lr: 0.000003 loss: 0.5503 (0.5148) time: 0.5048 data: 0.0002 max mem: 52126 +Epoch: [29] [1290/3634] eta: 0:19:55 lr: 0.000003 loss: 0.5330 (0.5149) time: 0.4922 data: 0.0002 max mem: 52126 +Epoch: [29] [1300/3634] eta: 0:19:50 lr: 0.000003 loss: 0.5319 (0.5149) time: 0.4883 data: 0.0002 max mem: 52126 +Epoch: [29] [1310/3634] eta: 0:19:44 lr: 0.000003 loss: 0.4943 (0.5146) time: 0.4927 data: 0.0002 max mem: 52126 +Epoch: [29] [1320/3634] eta: 0:19:40 lr: 0.000003 loss: 0.4974 (0.5145) time: 0.5358 data: 0.0002 max mem: 52126 +Epoch: [29] [1330/3634] eta: 0:19:33 lr: 0.000003 loss: 0.5153 (0.5146) time: 0.4852 data: 0.0002 max mem: 52126 +Epoch: [29] [1340/3634] eta: 0:19:28 lr: 0.000003 loss: 0.5332 (0.5145) time: 0.4406 data: 0.0002 max mem: 52126 +Epoch: [29] [1350/3634] eta: 0:19:25 lr: 0.000003 loss: 0.5399 (0.5146) time: 0.5875 data: 0.0002 max mem: 52126 +Epoch: [29] [1360/3634] eta: 0:19:20 lr: 0.000003 loss: 0.5203 (0.5147) time: 0.5764 data: 0.0002 max mem: 52126 +Epoch: [29] [1370/3634] eta: 0:19:13 lr: 0.000003 loss: 0.4794 (0.5144) time: 0.4286 data: 0.0002 max mem: 52126 +Epoch: [29] [1380/3634] eta: 0:19:05 lr: 0.000003 loss: 0.4760 (0.5142) time: 0.3829 data: 0.0002 max mem: 52126 +Epoch: [29] [1390/3634] eta: 0:18:58 lr: 0.000003 loss: 0.5301 (0.5144) time: 0.3831 data: 0.0002 max mem: 52126 +Epoch: [29] [1400/3634] eta: 0:18:51 lr: 0.000003 loss: 0.5377 (0.5143) time: 0.3819 data: 0.0002 max mem: 52126 +Epoch: [29] [1410/3634] eta: 0:18:49 lr: 0.000003 loss: 0.5010 (0.5143) time: 0.5344 data: 0.0002 max mem: 52126 +Epoch: [29] [1420/3634] eta: 0:18:44 lr: 0.000003 loss: 0.5236 (0.5144) time: 0.5953 data: 0.0002 max mem: 52126 +Epoch: [29] [1430/3634] eta: 0:18:37 lr: 0.000003 loss: 0.5286 (0.5145) time: 0.4426 data: 0.0002 max mem: 52126 +Epoch: [29] [1440/3634] eta: 0:18:30 lr: 0.000003 loss: 0.5152 (0.5144) time: 0.3816 data: 0.0001 max mem: 52126 +Epoch: [29] [1450/3634] eta: 0:18:23 lr: 0.000003 loss: 0.5017 (0.5145) time: 0.3824 data: 0.0001 max mem: 52126 +Epoch: [29] [1460/3634] eta: 0:18:16 lr: 0.000003 loss: 0.5170 (0.5146) time: 0.3818 data: 0.0002 max mem: 52126 +Epoch: [29] [1470/3634] eta: 0:18:10 lr: 0.000003 loss: 0.4990 (0.5145) time: 0.4255 data: 0.0002 max mem: 52126 +Epoch: [29] [1480/3634] eta: 0:18:05 lr: 0.000003 loss: 0.4865 (0.5144) time: 0.4706 data: 0.0001 max mem: 52126 +Epoch: [29] [1490/3634] eta: 0:17:59 lr: 0.000003 loss: 0.5002 (0.5143) time: 0.4691 data: 0.0001 max mem: 52126 +Epoch: [29] [1500/3634] eta: 0:17:53 lr: 0.000003 loss: 0.5318 (0.5143) time: 0.4241 data: 0.0001 max mem: 52126 +Epoch: [29] [1510/3634] eta: 0:17:50 lr: 0.000003 loss: 0.5600 (0.5146) time: 0.5146 data: 0.0002 max mem: 52126 +Epoch: [29] [1520/3634] eta: 0:17:44 lr: 0.000003 loss: 0.5489 (0.5146) time: 0.5599 data: 0.0002 max mem: 52126 +Epoch: [29] [1530/3634] eta: 0:17:39 lr: 0.000003 loss: 0.5180 (0.5145) time: 0.4734 data: 0.0002 max mem: 52126 +Epoch: [29] [1540/3634] eta: 0:17:32 lr: 0.000003 loss: 0.5108 (0.5144) time: 0.4290 data: 0.0002 max mem: 52126 +Epoch: [29] [1550/3634] eta: 0:17:27 lr: 0.000003 loss: 0.5254 (0.5145) time: 0.4431 data: 0.0002 max mem: 52126 +Epoch: [29] [1560/3634] eta: 0:17:22 lr: 0.000003 loss: 0.5501 (0.5146) time: 0.5053 data: 0.0002 max mem: 52126 +Epoch: [29] [1570/3634] eta: 0:17:17 lr: 0.000003 loss: 0.5343 (0.5145) time: 0.4925 data: 0.0002 max mem: 52126 +Epoch: [29] [1580/3634] eta: 0:17:12 lr: 0.000003 loss: 0.4637 (0.5142) time: 0.4894 data: 0.0001 max mem: 52126 +Epoch: [29] [1590/3634] eta: 0:17:07 lr: 0.000003 loss: 0.5099 (0.5142) time: 0.5026 data: 0.0002 max mem: 52126 +Epoch: [29] [1600/3634] eta: 0:17:02 lr: 0.000003 loss: 0.4942 (0.5139) time: 0.5022 data: 0.0002 max mem: 52126 +Epoch: [29] [1610/3634] eta: 0:16:58 lr: 0.000003 loss: 0.4981 (0.5139) time: 0.5438 data: 0.0002 max mem: 52126 +Epoch: [29] [1620/3634] eta: 0:16:53 lr: 0.000003 loss: 0.5169 (0.5139) time: 0.5440 data: 0.0002 max mem: 52126 +Epoch: [29] [1630/3634] eta: 0:16:46 lr: 0.000003 loss: 0.5157 (0.5139) time: 0.4410 data: 0.0002 max mem: 52126 +Epoch: [29] [1640/3634] eta: 0:16:41 lr: 0.000003 loss: 0.5157 (0.5139) time: 0.4413 data: 0.0002 max mem: 52126 +Epoch: [29] [1650/3634] eta: 0:16:37 lr: 0.000003 loss: 0.5307 (0.5141) time: 0.5459 data: 0.0002 max mem: 52126 +Epoch: [29] [1660/3634] eta: 0:16:33 lr: 0.000003 loss: 0.5385 (0.5141) time: 0.5756 data: 0.0002 max mem: 52126 +Epoch: [29] [1670/3634] eta: 0:16:26 lr: 0.000003 loss: 0.5385 (0.5143) time: 0.4714 data: 0.0002 max mem: 52126 +Epoch: [29] [1680/3634] eta: 0:16:22 lr: 0.000003 loss: 0.5237 (0.5143) time: 0.4696 data: 0.0002 max mem: 52126 +Epoch: [29] [1690/3634] eta: 0:16:17 lr: 0.000003 loss: 0.4696 (0.5139) time: 0.5324 data: 0.0002 max mem: 52126 +Epoch: [29] [1700/3634] eta: 0:16:10 lr: 0.000003 loss: 0.4665 (0.5138) time: 0.4453 data: 0.0002 max mem: 52126 +Epoch: [29] [1710/3634] eta: 0:16:05 lr: 0.000003 loss: 0.5353 (0.5140) time: 0.4415 data: 0.0002 max mem: 52126 +Epoch: [29] [1720/3634] eta: 0:16:03 lr: 0.000003 loss: 0.5281 (0.5140) time: 0.6119 data: 0.0002 max mem: 52126 +Epoch: [29] [1730/3634] eta: 0:15:59 lr: 0.000003 loss: 0.5281 (0.5141) time: 0.6602 data: 0.0002 max mem: 52126 +Epoch: [29] [1740/3634] eta: 0:15:53 lr: 0.000003 loss: 0.5158 (0.5140) time: 0.4897 data: 0.0002 max mem: 52126 +Epoch: [29] [1750/3634] eta: 0:15:49 lr: 0.000003 loss: 0.4936 (0.5140) time: 0.5055 data: 0.0002 max mem: 52126 +Epoch: [29] [1760/3634] eta: 0:15:47 lr: 0.000003 loss: 0.5141 (0.5141) time: 0.7160 data: 0.0002 max mem: 52126 +Epoch: [29] [1770/3634] eta: 0:15:43 lr: 0.000003 loss: 0.5129 (0.5140) time: 0.7026 data: 0.0001 max mem: 52126 +Epoch: [29] [1780/3634] eta: 0:15:37 lr: 0.000003 loss: 0.4938 (0.5140) time: 0.4919 data: 0.0002 max mem: 52126 +Epoch: [29] [1790/3634] eta: 0:15:31 lr: 0.000003 loss: 0.4981 (0.5141) time: 0.4262 data: 0.0002 max mem: 52126 +Epoch: [29] [1800/3634] eta: 0:15:25 lr: 0.000003 loss: 0.5291 (0.5142) time: 0.4260 data: 0.0002 max mem: 52126 +Epoch: [29] [1810/3634] eta: 0:15:20 lr: 0.000003 loss: 0.5466 (0.5145) time: 0.4406 data: 0.0002 max mem: 52126 +Epoch: [29] [1820/3634] eta: 0:15:14 lr: 0.000003 loss: 0.5419 (0.5145) time: 0.4411 data: 0.0002 max mem: 52126 +Epoch: [29] [1830/3634] eta: 0:15:07 lr: 0.000003 loss: 0.5118 (0.5145) time: 0.3837 data: 0.0002 max mem: 52126 +Epoch: [29] [1840/3634] eta: 0:15:02 lr: 0.000003 loss: 0.4970 (0.5144) time: 0.4271 data: 0.0002 max mem: 52126 +Epoch: [29] [1850/3634] eta: 0:14:57 lr: 0.000003 loss: 0.4840 (0.5142) time: 0.4850 data: 0.0002 max mem: 52126 +Epoch: [29] [1860/3634] eta: 0:14:52 lr: 0.000003 loss: 0.4996 (0.5143) time: 0.4872 data: 0.0001 max mem: 52126 +Epoch: [29] [1870/3634] eta: 0:14:45 lr: 0.000003 loss: 0.5230 (0.5144) time: 0.4279 data: 0.0002 max mem: 52126 +Epoch: [29] [1880/3634] eta: 0:14:40 lr: 0.000003 loss: 0.5162 (0.5142) time: 0.4417 data: 0.0002 max mem: 52126 +Epoch: [29] [1890/3634] eta: 0:14:35 lr: 0.000003 loss: 0.5077 (0.5142) time: 0.5019 data: 0.0001 max mem: 52126 +Epoch: [29] [1900/3634] eta: 0:14:31 lr: 0.000003 loss: 0.4967 (0.5141) time: 0.5593 data: 0.0001 max mem: 52126 +Epoch: [29] [1910/3634] eta: 0:14:27 lr: 0.000003 loss: 0.5408 (0.5142) time: 0.6171 data: 0.0002 max mem: 52126 +Epoch: [29] [1920/3634] eta: 0:14:24 lr: 0.000003 loss: 0.5532 (0.5142) time: 0.6362 data: 0.0002 max mem: 52126 +Epoch: [29] [1930/3634] eta: 0:14:19 lr: 0.000003 loss: 0.5076 (0.5141) time: 0.5778 data: 0.0001 max mem: 52126 +Epoch: [29] [1940/3634] eta: 0:14:13 lr: 0.000003 loss: 0.5076 (0.5141) time: 0.4422 data: 0.0001 max mem: 52126 +Epoch: [29] [1950/3634] eta: 0:14:08 lr: 0.000003 loss: 0.5084 (0.5141) time: 0.4897 data: 0.0001 max mem: 52126 +Epoch: [29] [1960/3634] eta: 0:14:03 lr: 0.000003 loss: 0.5382 (0.5141) time: 0.5388 data: 0.0002 max mem: 52126 +Epoch: [29] [1970/3634] eta: 0:13:57 lr: 0.000003 loss: 0.5049 (0.5139) time: 0.4320 data: 0.0002 max mem: 52126 +Epoch: [29] [1980/3634] eta: 0:13:51 lr: 0.000003 loss: 0.4765 (0.5139) time: 0.3825 data: 0.0002 max mem: 52126 +Epoch: [29] [1990/3634] eta: 0:13:47 lr: 0.000003 loss: 0.5133 (0.5138) time: 0.4882 data: 0.0001 max mem: 52126 +Epoch: [29] [2000/3634] eta: 0:13:42 lr: 0.000003 loss: 0.4935 (0.5137) time: 0.5754 data: 0.0002 max mem: 52126 +Epoch: [29] [2010/3634] eta: 0:13:36 lr: 0.000003 loss: 0.4935 (0.5138) time: 0.4696 data: 0.0002 max mem: 52126 +Epoch: [29] [2020/3634] eta: 0:13:31 lr: 0.000003 loss: 0.4888 (0.5137) time: 0.4258 data: 0.0001 max mem: 52126 +Epoch: [29] [2030/3634] eta: 0:13:25 lr: 0.000003 loss: 0.4978 (0.5137) time: 0.4276 data: 0.0002 max mem: 52126 +Epoch: [29] [2040/3634] eta: 0:13:20 lr: 0.000003 loss: 0.5462 (0.5137) time: 0.4734 data: 0.0002 max mem: 52126 +Epoch: [29] [2050/3634] eta: 0:13:14 lr: 0.000003 loss: 0.5378 (0.5136) time: 0.4721 data: 0.0002 max mem: 52126 +Epoch: [29] [2060/3634] eta: 0:13:12 lr: 0.000003 loss: 0.5348 (0.5137) time: 0.6372 data: 0.0001 max mem: 52126 +Epoch: [29] [2070/3634] eta: 0:13:07 lr: 0.000003 loss: 0.5464 (0.5138) time: 0.6996 data: 0.0001 max mem: 52126 +Epoch: [29] [2080/3634] eta: 0:13:01 lr: 0.000003 loss: 0.5181 (0.5138) time: 0.4452 data: 0.0002 max mem: 52126 +Epoch: [29] [2090/3634] eta: 0:12:56 lr: 0.000003 loss: 0.5161 (0.5137) time: 0.4273 data: 0.0002 max mem: 52126 +Epoch: [29] [2100/3634] eta: 0:12:51 lr: 0.000003 loss: 0.5304 (0.5137) time: 0.4712 data: 0.0002 max mem: 52126 +Epoch: [29] [2110/3634] eta: 0:12:46 lr: 0.000003 loss: 0.5352 (0.5138) time: 0.4843 data: 0.0002 max mem: 52126 +Epoch: [29] [2120/3634] eta: 0:12:41 lr: 0.000003 loss: 0.5218 (0.5138) time: 0.4850 data: 0.0002 max mem: 52126 +Epoch: [29] [2130/3634] eta: 0:12:35 lr: 0.000003 loss: 0.5273 (0.5140) time: 0.4268 data: 0.0002 max mem: 52126 +Epoch: [29] [2140/3634] eta: 0:12:30 lr: 0.000003 loss: 0.5275 (0.5139) time: 0.4418 data: 0.0002 max mem: 52126 +Epoch: [29] [2150/3634] eta: 0:12:24 lr: 0.000003 loss: 0.4906 (0.5139) time: 0.4420 data: 0.0002 max mem: 52126 +Epoch: [29] [2160/3634] eta: 0:12:18 lr: 0.000003 loss: 0.5304 (0.5140) time: 0.3828 data: 0.0001 max mem: 52126 +Epoch: [29] [2170/3634] eta: 0:12:14 lr: 0.000003 loss: 0.5429 (0.5141) time: 0.4863 data: 0.0001 max mem: 52126 +Epoch: [29] [2180/3634] eta: 0:12:10 lr: 0.000003 loss: 0.5243 (0.5140) time: 0.6669 data: 0.0001 max mem: 52126 +Epoch: [29] [2190/3634] eta: 0:12:04 lr: 0.000003 loss: 0.5060 (0.5141) time: 0.5631 data: 0.0002 max mem: 52126 +Epoch: [29] [2200/3634] eta: 0:12:00 lr: 0.000003 loss: 0.5341 (0.5141) time: 0.5033 data: 0.0001 max mem: 52126 +Epoch: [29] [2210/3634] eta: 0:11:54 lr: 0.000003 loss: 0.5541 (0.5142) time: 0.5034 data: 0.0002 max mem: 52126 +Epoch: [29] [2220/3634] eta: 0:11:49 lr: 0.000003 loss: 0.5353 (0.5142) time: 0.3829 data: 0.0002 max mem: 52126 +Epoch: [29] [2230/3634] eta: 0:11:44 lr: 0.000003 loss: 0.4503 (0.5140) time: 0.5022 data: 0.0002 max mem: 52126 +Epoch: [29] [2240/3634] eta: 0:11:38 lr: 0.000003 loss: 0.4979 (0.5140) time: 0.5018 data: 0.0002 max mem: 52126 +Epoch: [29] [2250/3634] eta: 0:11:34 lr: 0.000003 loss: 0.5283 (0.5141) time: 0.4892 data: 0.0002 max mem: 52126 +Epoch: [29] [2260/3634] eta: 0:11:30 lr: 0.000003 loss: 0.5097 (0.5140) time: 0.5946 data: 0.0002 max mem: 52126 +Epoch: [29] [2270/3634] eta: 0:11:25 lr: 0.000003 loss: 0.4859 (0.5140) time: 0.5466 data: 0.0002 max mem: 52126 +Epoch: [29] [2280/3634] eta: 0:11:20 lr: 0.000003 loss: 0.5081 (0.5140) time: 0.5008 data: 0.0002 max mem: 52126 +Epoch: [29] [2290/3634] eta: 0:11:16 lr: 0.000003 loss: 0.5517 (0.5141) time: 0.6192 data: 0.0002 max mem: 52126 +Epoch: [29] [2300/3634] eta: 0:11:11 lr: 0.000003 loss: 0.5442 (0.5141) time: 0.6237 data: 0.0001 max mem: 52126 +Epoch: [29] [2310/3634] eta: 0:11:06 lr: 0.000003 loss: 0.5344 (0.5141) time: 0.5060 data: 0.0002 max mem: 52126 +Epoch: [29] [2320/3634] eta: 0:11:00 lr: 0.000003 loss: 0.5344 (0.5140) time: 0.4419 data: 0.0002 max mem: 52126 +Epoch: [29] [2330/3634] eta: 0:10:55 lr: 0.000003 loss: 0.5314 (0.5140) time: 0.4272 data: 0.0002 max mem: 52126 +Epoch: [29] [2340/3634] eta: 0:10:50 lr: 0.000003 loss: 0.5389 (0.5142) time: 0.4774 data: 0.0002 max mem: 52126 +Epoch: [29] [2350/3634] eta: 0:10:45 lr: 0.000003 loss: 0.5389 (0.5142) time: 0.5504 data: 0.0002 max mem: 52126 +Epoch: [29] [2360/3634] eta: 0:10:40 lr: 0.000003 loss: 0.5545 (0.5144) time: 0.5005 data: 0.0002 max mem: 52126 +Epoch: [29] [2370/3634] eta: 0:10:34 lr: 0.000003 loss: 0.5369 (0.5142) time: 0.3831 data: 0.0002 max mem: 52126 +Epoch: [29] [2380/3634] eta: 0:10:29 lr: 0.000003 loss: 0.5145 (0.5144) time: 0.4419 data: 0.0002 max mem: 52126 +Epoch: [29] [2390/3634] eta: 0:10:24 lr: 0.000003 loss: 0.5431 (0.5144) time: 0.4860 data: 0.0002 max mem: 52126 +Epoch: [29] [2400/3634] eta: 0:10:18 lr: 0.000003 loss: 0.5502 (0.5145) time: 0.4270 data: 0.0002 max mem: 52126 +Epoch: [29] [2410/3634] eta: 0:10:14 lr: 0.000003 loss: 0.5443 (0.5145) time: 0.5007 data: 0.0002 max mem: 52126 +Epoch: [29] [2420/3634] eta: 0:10:08 lr: 0.000003 loss: 0.5314 (0.5144) time: 0.5007 data: 0.0002 max mem: 52126 +Epoch: [29] [2430/3634] eta: 0:10:04 lr: 0.000003 loss: 0.5314 (0.5143) time: 0.5002 data: 0.0002 max mem: 52126 +Epoch: [29] [2440/3634] eta: 0:09:59 lr: 0.000003 loss: 0.4964 (0.5143) time: 0.5442 data: 0.0002 max mem: 52126 +Epoch: [29] [2450/3634] eta: 0:09:55 lr: 0.000003 loss: 0.4964 (0.5143) time: 0.5974 data: 0.0002 max mem: 52126 +Epoch: [29] [2460/3634] eta: 0:09:50 lr: 0.000003 loss: 0.5246 (0.5143) time: 0.6144 data: 0.0002 max mem: 52126 +Epoch: [29] [2470/3634] eta: 0:09:44 lr: 0.000003 loss: 0.5277 (0.5144) time: 0.4436 data: 0.0002 max mem: 52126 +Epoch: [29] [2480/3634] eta: 0:09:39 lr: 0.000003 loss: 0.5210 (0.5143) time: 0.4421 data: 0.0002 max mem: 52126 +Epoch: [29] [2490/3634] eta: 0:09:34 lr: 0.000003 loss: 0.5112 (0.5142) time: 0.4862 data: 0.0002 max mem: 52126 +Epoch: [29] [2500/3634] eta: 0:09:29 lr: 0.000003 loss: 0.5305 (0.5144) time: 0.4763 data: 0.0002 max mem: 52126 +Epoch: [29] [2510/3634] eta: 0:09:23 lr: 0.000003 loss: 0.5186 (0.5144) time: 0.4319 data: 0.0001 max mem: 52126 +Epoch: [29] [2520/3634] eta: 0:09:19 lr: 0.000003 loss: 0.5147 (0.5144) time: 0.5030 data: 0.0002 max mem: 52126 +Epoch: [29] [2530/3634] eta: 0:09:14 lr: 0.000003 loss: 0.4966 (0.5143) time: 0.6350 data: 0.0002 max mem: 52126 +Epoch: [29] [2540/3634] eta: 0:09:09 lr: 0.000003 loss: 0.4973 (0.5142) time: 0.5726 data: 0.0002 max mem: 52126 +Epoch: [29] [2550/3634] eta: 0:09:04 lr: 0.000003 loss: 0.5083 (0.5142) time: 0.4846 data: 0.0002 max mem: 52126 +Epoch: [29] [2560/3634] eta: 0:09:00 lr: 0.000003 loss: 0.5359 (0.5143) time: 0.5611 data: 0.0002 max mem: 52126 +Epoch: [29] [2570/3634] eta: 0:08:54 lr: 0.000003 loss: 0.5075 (0.5143) time: 0.5176 data: 0.0002 max mem: 52126 +Epoch: [29] [2580/3634] eta: 0:08:50 lr: 0.000003 loss: 0.4725 (0.5141) time: 0.4863 data: 0.0001 max mem: 52126 +Epoch: [29] [2590/3634] eta: 0:08:44 lr: 0.000003 loss: 0.4669 (0.5141) time: 0.5319 data: 0.0001 max mem: 52126 +Epoch: [29] [2600/3634] eta: 0:08:39 lr: 0.000003 loss: 0.5391 (0.5142) time: 0.4723 data: 0.0001 max mem: 52126 +Epoch: [29] [2610/3634] eta: 0:08:34 lr: 0.000003 loss: 0.5391 (0.5142) time: 0.4867 data: 0.0002 max mem: 52126 +Epoch: [29] [2620/3634] eta: 0:08:29 lr: 0.000003 loss: 0.5143 (0.5141) time: 0.4428 data: 0.0002 max mem: 52126 +Epoch: [29] [2630/3634] eta: 0:08:23 lr: 0.000003 loss: 0.5022 (0.5142) time: 0.3823 data: 0.0002 max mem: 52126 +Epoch: [29] [2640/3634] eta: 0:08:18 lr: 0.000003 loss: 0.5022 (0.5140) time: 0.3819 data: 0.0002 max mem: 52126 +Epoch: [29] [2650/3634] eta: 0:08:13 lr: 0.000003 loss: 0.4479 (0.5139) time: 0.4714 data: 0.0002 max mem: 52126 +Epoch: [29] [2660/3634] eta: 0:08:08 lr: 0.000003 loss: 0.4973 (0.5140) time: 0.5155 data: 0.0002 max mem: 52126 +Epoch: [29] [2670/3634] eta: 0:08:03 lr: 0.000003 loss: 0.5291 (0.5141) time: 0.4848 data: 0.0001 max mem: 52126 +Epoch: [29] [2680/3634] eta: 0:07:58 lr: 0.000003 loss: 0.5302 (0.5141) time: 0.4857 data: 0.0001 max mem: 52126 +Epoch: [29] [2690/3634] eta: 0:07:53 lr: 0.000003 loss: 0.5357 (0.5142) time: 0.5979 data: 0.0001 max mem: 52126 +Epoch: [29] [2700/3634] eta: 0:07:48 lr: 0.000003 loss: 0.4989 (0.5141) time: 0.5532 data: 0.0002 max mem: 52126 +Epoch: [29] [2710/3634] eta: 0:07:43 lr: 0.000003 loss: 0.5139 (0.5141) time: 0.4254 data: 0.0002 max mem: 52126 +Epoch: [29] [2720/3634] eta: 0:07:39 lr: 0.000003 loss: 0.5370 (0.5141) time: 0.5916 data: 0.0002 max mem: 52126 +Epoch: [29] [2730/3634] eta: 0:07:34 lr: 0.000003 loss: 0.5229 (0.5142) time: 0.6708 data: 0.0002 max mem: 52126 +Epoch: [29] [2740/3634] eta: 0:07:29 lr: 0.000003 loss: 0.5249 (0.5143) time: 0.5048 data: 0.0002 max mem: 52126 +Epoch: [29] [2750/3634] eta: 0:07:24 lr: 0.000003 loss: 0.5196 (0.5143) time: 0.4415 data: 0.0002 max mem: 52126 +Epoch: [29] [2760/3634] eta: 0:07:18 lr: 0.000003 loss: 0.5120 (0.5143) time: 0.4413 data: 0.0002 max mem: 52126 +Epoch: [29] [2770/3634] eta: 0:07:13 lr: 0.000003 loss: 0.5326 (0.5144) time: 0.3830 data: 0.0001 max mem: 52126 +Epoch: [29] [2780/3634] eta: 0:07:08 lr: 0.000003 loss: 0.5317 (0.5143) time: 0.5618 data: 0.0002 max mem: 52126 +Epoch: [29] [2790/3634] eta: 0:07:04 lr: 0.000003 loss: 0.5335 (0.5144) time: 0.6666 data: 0.0002 max mem: 52126 +Epoch: [29] [2800/3634] eta: 0:06:59 lr: 0.000003 loss: 0.5170 (0.5143) time: 0.5887 data: 0.0002 max mem: 52126 +Epoch: [29] [2810/3634] eta: 0:06:54 lr: 0.000003 loss: 0.5098 (0.5143) time: 0.5429 data: 0.0002 max mem: 52126 +Epoch: [29] [2820/3634] eta: 0:06:49 lr: 0.000003 loss: 0.5480 (0.5144) time: 0.4871 data: 0.0002 max mem: 52126 +Epoch: [29] [2830/3634] eta: 0:06:43 lr: 0.000003 loss: 0.5305 (0.5144) time: 0.4275 data: 0.0001 max mem: 52126 +Epoch: [29] [2840/3634] eta: 0:06:38 lr: 0.000003 loss: 0.5185 (0.5144) time: 0.4416 data: 0.0002 max mem: 52126 +Epoch: [29] [2850/3634] eta: 0:06:33 lr: 0.000003 loss: 0.5239 (0.5145) time: 0.4416 data: 0.0002 max mem: 52126 +Epoch: [29] [2860/3634] eta: 0:06:28 lr: 0.000003 loss: 0.5294 (0.5146) time: 0.4739 data: 0.0002 max mem: 52126 +Epoch: [29] [2870/3634] eta: 0:06:23 lr: 0.000003 loss: 0.5321 (0.5145) time: 0.4744 data: 0.0002 max mem: 52126 +Epoch: [29] [2880/3634] eta: 0:06:18 lr: 0.000003 loss: 0.5159 (0.5146) time: 0.3853 data: 0.0002 max mem: 52126 +Epoch: [29] [2890/3634] eta: 0:06:13 lr: 0.000003 loss: 0.5449 (0.5146) time: 0.4449 data: 0.0002 max mem: 52126 +Epoch: [29] [2900/3634] eta: 0:06:08 lr: 0.000003 loss: 0.5449 (0.5147) time: 0.5557 data: 0.0002 max mem: 52126 +Epoch: [29] [2910/3634] eta: 0:06:02 lr: 0.000003 loss: 0.5534 (0.5147) time: 0.4957 data: 0.0002 max mem: 52126 +Epoch: [29] [2920/3634] eta: 0:05:57 lr: 0.000003 loss: 0.5055 (0.5147) time: 0.3823 data: 0.0002 max mem: 52126 +Epoch: [29] [2930/3634] eta: 0:05:53 lr: 0.000003 loss: 0.5230 (0.5147) time: 0.5995 data: 0.0002 max mem: 52126 +Epoch: [29] [2940/3634] eta: 0:05:48 lr: 0.000003 loss: 0.5602 (0.5149) time: 0.6586 data: 0.0001 max mem: 52126 +Epoch: [29] [2950/3634] eta: 0:05:43 lr: 0.000003 loss: 0.5602 (0.5150) time: 0.4408 data: 0.0002 max mem: 52126 +Epoch: [29] [2960/3634] eta: 0:05:37 lr: 0.000003 loss: 0.5406 (0.5150) time: 0.3820 data: 0.0001 max mem: 52126 +Epoch: [29] [2970/3634] eta: 0:05:32 lr: 0.000003 loss: 0.5194 (0.5151) time: 0.3821 data: 0.0002 max mem: 52126 +Epoch: [29] [2980/3634] eta: 0:05:27 lr: 0.000003 loss: 0.5182 (0.5150) time: 0.5457 data: 0.0002 max mem: 52126 +Epoch: [29] [2990/3634] eta: 0:05:23 lr: 0.000003 loss: 0.5025 (0.5149) time: 0.6521 data: 0.0001 max mem: 52126 +Epoch: [29] [3000/3634] eta: 0:05:17 lr: 0.000003 loss: 0.4623 (0.5149) time: 0.4886 data: 0.0002 max mem: 52126 +Epoch: [29] [3010/3634] eta: 0:05:12 lr: 0.000003 loss: 0.4934 (0.5149) time: 0.3820 data: 0.0002 max mem: 52126 +Epoch: [29] [3020/3634] eta: 0:05:07 lr: 0.000003 loss: 0.5619 (0.5150) time: 0.5183 data: 0.0002 max mem: 52126 +Epoch: [29] [3030/3634] eta: 0:05:02 lr: 0.000003 loss: 0.5238 (0.5149) time: 0.5622 data: 0.0002 max mem: 52126 +Epoch: [29] [3040/3634] eta: 0:04:57 lr: 0.000003 loss: 0.4977 (0.5149) time: 0.4262 data: 0.0002 max mem: 52126 +Epoch: [29] [3050/3634] eta: 0:04:52 lr: 0.000003 loss: 0.5390 (0.5150) time: 0.4419 data: 0.0002 max mem: 52126 +Epoch: [29] [3060/3634] eta: 0:04:47 lr: 0.000003 loss: 0.5418 (0.5152) time: 0.4418 data: 0.0002 max mem: 52126 +Epoch: [29] [3070/3634] eta: 0:04:42 lr: 0.000003 loss: 0.5356 (0.5151) time: 0.3830 data: 0.0002 max mem: 52126 +Epoch: [29] [3080/3634] eta: 0:04:37 lr: 0.000003 loss: 0.5108 (0.5151) time: 0.5734 data: 0.0002 max mem: 52126 +Epoch: [29] [3090/3634] eta: 0:04:32 lr: 0.000003 loss: 0.5108 (0.5151) time: 0.7048 data: 0.0002 max mem: 52126 +Epoch: [29] [3100/3634] eta: 0:04:27 lr: 0.000003 loss: 0.5105 (0.5150) time: 0.5729 data: 0.0002 max mem: 52126 +Epoch: [29] [3110/3634] eta: 0:04:22 lr: 0.000003 loss: 0.4885 (0.5150) time: 0.4857 data: 0.0002 max mem: 52126 +Epoch: [29] [3120/3634] eta: 0:04:17 lr: 0.000003 loss: 0.5038 (0.5151) time: 0.4883 data: 0.0002 max mem: 52126 +Epoch: [29] [3130/3634] eta: 0:04:12 lr: 0.000003 loss: 0.5252 (0.5150) time: 0.5074 data: 0.0002 max mem: 52126 +Epoch: [29] [3140/3634] eta: 0:04:07 lr: 0.000003 loss: 0.5390 (0.5151) time: 0.4901 data: 0.0002 max mem: 52126 +Epoch: [29] [3150/3634] eta: 0:04:02 lr: 0.000003 loss: 0.5688 (0.5152) time: 0.4857 data: 0.0001 max mem: 52126 +Epoch: [29] [3160/3634] eta: 0:03:57 lr: 0.000003 loss: 0.5688 (0.5152) time: 0.4859 data: 0.0001 max mem: 52126 +Epoch: [29] [3170/3634] eta: 0:03:52 lr: 0.000003 loss: 0.5356 (0.5153) time: 0.4887 data: 0.0002 max mem: 52126 +Epoch: [29] [3180/3634] eta: 0:03:47 lr: 0.000003 loss: 0.5336 (0.5153) time: 0.6272 data: 0.0002 max mem: 52126 +Epoch: [29] [3190/3634] eta: 0:03:42 lr: 0.000003 loss: 0.5126 (0.5152) time: 0.6258 data: 0.0002 max mem: 52126 +Epoch: [29] [3200/3634] eta: 0:03:37 lr: 0.000003 loss: 0.4956 (0.5153) time: 0.5002 data: 0.0002 max mem: 52126 +Epoch: [29] [3210/3634] eta: 0:03:32 lr: 0.000003 loss: 0.5497 (0.5153) time: 0.4397 data: 0.0001 max mem: 52126 +Epoch: [29] [3220/3634] eta: 0:03:27 lr: 0.000003 loss: 0.5497 (0.5153) time: 0.4420 data: 0.0001 max mem: 52126 +Epoch: [29] [3230/3634] eta: 0:03:22 lr: 0.000003 loss: 0.5504 (0.5154) time: 0.4860 data: 0.0001 max mem: 52126 +Epoch: [29] [3240/3634] eta: 0:03:17 lr: 0.000003 loss: 0.5404 (0.5153) time: 0.5357 data: 0.0001 max mem: 52126 +Epoch: [29] [3250/3634] eta: 0:03:12 lr: 0.000003 loss: 0.5397 (0.5154) time: 0.5505 data: 0.0002 max mem: 52126 +Epoch: [29] [3260/3634] eta: 0:03:07 lr: 0.000003 loss: 0.5134 (0.5153) time: 0.4868 data: 0.0002 max mem: 52126 +Epoch: [29] [3270/3634] eta: 0:03:02 lr: 0.000003 loss: 0.4815 (0.5152) time: 0.4283 data: 0.0002 max mem: 52126 +Epoch: [29] [3280/3634] eta: 0:02:57 lr: 0.000003 loss: 0.4993 (0.5152) time: 0.4405 data: 0.0001 max mem: 52126 +Epoch: [29] [3290/3634] eta: 0:02:52 lr: 0.000003 loss: 0.5199 (0.5152) time: 0.4979 data: 0.0002 max mem: 52126 +Epoch: [29] [3300/3634] eta: 0:02:47 lr: 0.000003 loss: 0.5241 (0.5151) time: 0.4904 data: 0.0002 max mem: 52126 +Epoch: [29] [3310/3634] eta: 0:02:42 lr: 0.000003 loss: 0.4619 (0.5149) time: 0.4325 data: 0.0002 max mem: 52126 +Epoch: [29] [3320/3634] eta: 0:02:37 lr: 0.000003 loss: 0.4432 (0.5148) time: 0.4308 data: 0.0002 max mem: 52126 +Epoch: [29] [3330/3634] eta: 0:02:32 lr: 0.000003 loss: 0.4915 (0.5149) time: 0.4758 data: 0.0001 max mem: 52126 +Epoch: [29] [3340/3634] eta: 0:02:27 lr: 0.000003 loss: 0.5431 (0.5149) time: 0.4273 data: 0.0002 max mem: 52126 +Epoch: [29] [3350/3634] eta: 0:02:22 lr: 0.000003 loss: 0.5145 (0.5149) time: 0.4874 data: 0.0002 max mem: 52126 +Epoch: [29] [3360/3634] eta: 0:02:17 lr: 0.000003 loss: 0.5145 (0.5149) time: 0.5496 data: 0.0002 max mem: 52126 +Epoch: [29] [3370/3634] eta: 0:02:12 lr: 0.000003 loss: 0.5268 (0.5149) time: 0.5029 data: 0.0002 max mem: 52126 +Epoch: [29] [3380/3634] eta: 0:02:07 lr: 0.000003 loss: 0.5121 (0.5149) time: 0.4847 data: 0.0001 max mem: 52126 +Epoch: [29] [3390/3634] eta: 0:02:02 lr: 0.000003 loss: 0.5121 (0.5149) time: 0.5304 data: 0.0002 max mem: 52126 +Epoch: [29] [3400/3634] eta: 0:01:57 lr: 0.000003 loss: 0.5295 (0.5150) time: 0.5315 data: 0.0001 max mem: 52126 +Epoch: [29] [3410/3634] eta: 0:01:52 lr: 0.000003 loss: 0.5745 (0.5152) time: 0.5195 data: 0.0001 max mem: 52126 +Epoch: [29] [3420/3634] eta: 0:01:47 lr: 0.000003 loss: 0.5725 (0.5153) time: 0.5190 data: 0.0002 max mem: 52126 +Epoch: [29] [3430/3634] eta: 0:01:42 lr: 0.000003 loss: 0.5306 (0.5153) time: 0.4271 data: 0.0002 max mem: 52126 +Epoch: [29] [3440/3634] eta: 0:01:37 lr: 0.000003 loss: 0.5205 (0.5153) time: 0.5002 data: 0.0002 max mem: 52126 +Epoch: [29] [3450/3634] eta: 0:01:32 lr: 0.000003 loss: 0.5041 (0.5153) time: 0.5002 data: 0.0002 max mem: 52126 +Epoch: [29] [3460/3634] eta: 0:01:27 lr: 0.000003 loss: 0.5085 (0.5153) time: 0.5047 data: 0.0001 max mem: 52126 +Epoch: [29] [3470/3634] eta: 0:01:22 lr: 0.000003 loss: 0.5085 (0.5153) time: 0.5640 data: 0.0001 max mem: 52126 +Epoch: [29] [3480/3634] eta: 0:01:17 lr: 0.000003 loss: 0.5110 (0.5153) time: 0.4415 data: 0.0001 max mem: 52126 +Epoch: [29] [3490/3634] eta: 0:01:12 lr: 0.000003 loss: 0.5453 (0.5153) time: 0.3818 data: 0.0001 max mem: 52126 +Epoch: [29] [3500/3634] eta: 0:01:07 lr: 0.000003 loss: 0.5505 (0.5154) time: 0.5283 data: 0.0002 max mem: 52126 +Epoch: [29] [3510/3634] eta: 0:01:02 lr: 0.000003 loss: 0.5068 (0.5154) time: 0.5288 data: 0.0002 max mem: 52126 +Epoch: [29] [3520/3634] eta: 0:00:57 lr: 0.000003 loss: 0.4851 (0.5153) time: 0.4431 data: 0.0002 max mem: 52126 +Epoch: [29] [3530/3634] eta: 0:00:52 lr: 0.000003 loss: 0.5162 (0.5153) time: 0.4904 data: 0.0001 max mem: 52126 +Epoch: [29] [3540/3634] eta: 0:00:47 lr: 0.000003 loss: 0.5254 (0.5153) time: 0.5786 data: 0.0001 max mem: 52126 +Epoch: [29] [3550/3634] eta: 0:00:42 lr: 0.000003 loss: 0.5356 (0.5154) time: 0.5304 data: 0.0001 max mem: 52126 +Epoch: [29] [3560/3634] eta: 0:00:37 lr: 0.000003 loss: 0.5679 (0.5155) time: 0.3815 data: 0.0002 max mem: 52126 +Epoch: [29] [3570/3634] eta: 0:00:32 lr: 0.000003 loss: 0.5379 (0.5155) time: 0.5313 data: 0.0002 max mem: 52126 +Epoch: [29] [3580/3634] eta: 0:00:27 lr: 0.000003 loss: 0.5009 (0.5155) time: 0.6376 data: 0.0002 max mem: 52126 +Epoch: [29] [3590/3634] eta: 0:00:22 lr: 0.000003 loss: 0.5009 (0.5154) time: 0.5512 data: 0.0002 max mem: 52126 +Epoch: [29] [3600/3634] eta: 0:00:17 lr: 0.000003 loss: 0.4868 (0.5154) time: 0.4448 data: 0.0002 max mem: 52126 +Epoch: [29] [3610/3634] eta: 0:00:12 lr: 0.000003 loss: 0.4942 (0.5153) time: 0.3818 data: 0.0002 max mem: 52126 +Epoch: [29] [3620/3634] eta: 0:00:07 lr: 0.000003 loss: 0.4763 (0.5152) time: 0.4452 data: 0.0029 max mem: 52126 +Epoch: [29] [3630/3634] eta: 0:00:02 lr: 0.000003 loss: 0.4963 (0.5152) time: 0.5015 data: 0.0029 max mem: 52126 +Epoch: [29] [3633/3634] eta: 0:00:00 lr: 0.000003 loss: 0.4963 (0.5152) time: 0.5617 data: 0.0029 max mem: 52126 +Epoch: [29] Total time: 0:30:21 (0.5012 s / it) +Averaged stats: lr: 0.000003 loss: 0.4963 (0.5154) +Test: [ 0/1403] eta: 1:09:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0812 (0.0812) acc1: 100.0000 (100.0000) acc5: 100.0000 (100.0000) time: 2.9687 data: 2.8724 max mem: 52126 +Test: [ 10/1403] eta: 0:08:17 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1482 (0.1722) acc1: 98.4375 (95.5966) acc5: 100.0000 (100.0000) time: 0.3569 data: 0.2613 max mem: 52126 +Test: [ 20/1403] eta: 0:05:21 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1384 (0.1582) acc1: 98.4375 (95.7589) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 30/1403] eta: 0:04:18 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0881 (0.1482) acc1: 100.0000 (96.1694) acc5: 100.0000 (100.0000) time: 0.0959 data: 0.0003 max mem: 52126 +Test: [ 40/1403] eta: 0:03:46 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0984 (0.1467) acc1: 100.0000 (96.5320) acc5: 100.0000 (100.0000) time: 0.0959 data: 0.0003 max mem: 52126 +Test: [ 50/1403] eta: 0:03:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1094 (0.1558) acc1: 98.4375 (96.2010) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 60/1403] eta: 0:03:11 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.1179 (0.1542) acc1: 100.0000 (96.4139) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 70/1403] eta: 0:03:01 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.1291 (0.1580) acc1: 98.4375 (96.1268) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 80/1403] eta: 0:02:53 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1488 (0.1638) acc1: 96.8750 (95.7755) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 90/1403] eta: 0:02:47 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1681 (0.1806) acc1: 95.3125 (94.9176) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 100/1403] eta: 0:02:41 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1649 (0.1791) acc1: 95.3125 (95.0340) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 110/1403] eta: 0:02:37 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1321 (0.1775) acc1: 98.4375 (95.1436) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 120/1403] eta: 0:02:33 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0797 (0.1692) acc1: 100.0000 (95.5320) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 130/1403] eta: 0:02:29 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.0797 (0.1636) acc1: 100.0000 (95.7657) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 140/1403] eta: 0:02:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0738 (0.1572) acc1: 100.0000 (96.0660) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 150/1403] eta: 0:02:23 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0726 (0.1519) acc1: 100.0000 (96.2955) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 160/1403] eta: 0:02:21 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0753 (0.1476) acc1: 100.0000 (96.4771) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 170/1403] eta: 0:02:18 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0774 (0.1442) acc1: 100.0000 (96.6557) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 180/1403] eta: 0:02:16 acc1_pos: 82.812 acc5_pos: 100.000 loss: 0.0866 (0.1488) acc1: 100.0000 (96.4434) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 190/1403] eta: 0:02:14 acc1_pos: 89.062 acc5_pos: 100.000 loss: 0.2476 (0.1578) acc1: 92.1875 (95.9588) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 200/1403] eta: 0:02:12 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.2669 (0.1698) acc1: 89.0625 (95.3669) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 210/1403] eta: 0:02:10 acc1_pos: 76.562 acc5_pos: 100.000 loss: 0.3024 (0.1806) acc1: 84.3750 (94.7793) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 220/1403] eta: 0:02:08 acc1_pos: 89.062 acc5_pos: 100.000 loss: 0.3386 (0.1894) acc1: 85.9375 (94.3722) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 230/1403] eta: 0:02:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2443 (0.1856) acc1: 90.6250 (94.5617) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 240/1403] eta: 0:02:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0789 (0.1814) acc1: 100.0000 (94.7744) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 250/1403] eta: 0:02:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0781 (0.1777) acc1: 100.0000 (94.9639) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 260/1403] eta: 0:02:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0816 (0.1742) acc1: 100.0000 (95.1389) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 270/1403] eta: 0:02:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0828 (0.1710) acc1: 100.0000 (95.3010) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 280/1403] eta: 0:01:59 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0821 (0.1678) acc1: 100.0000 (95.4626) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 290/1403] eta: 0:01:57 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0763 (0.1647) acc1: 100.0000 (95.6132) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 300/1403] eta: 0:01:56 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0750 (0.1619) acc1: 100.0000 (95.7537) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 310/1403] eta: 0:01:54 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0807 (0.1595) acc1: 100.0000 (95.8903) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 320/1403] eta: 0:01:53 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1019 (0.1579) acc1: 100.0000 (96.0183) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 330/1403] eta: 0:01:52 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1089 (0.1565) acc1: 100.0000 (96.1386) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0957 data: 0.0002 max mem: 52126 +Test: [ 340/1403] eta: 0:01:50 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1089 (0.1550) acc1: 100.0000 (96.2473) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 350/1403] eta: 0:01:49 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1066 (0.1537) acc1: 100.0000 (96.3542) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 360/1403] eta: 0:01:48 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1098 (0.1526) acc1: 100.0000 (96.4508) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 370/1403] eta: 0:01:46 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1082 (0.1514) acc1: 100.0000 (96.5465) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 380/1403] eta: 0:01:45 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1073 (0.1503) acc1: 100.0000 (96.6371) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [ 390/1403] eta: 0:01:44 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1090 (0.1511) acc1: 100.0000 (96.5873) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0959 data: 0.0003 max mem: 52126 +Test: [ 400/1403] eta: 0:01:43 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2557 (0.1547) acc1: 90.6250 (96.3879) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 410/1403] eta: 0:01:42 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2784 (0.1577) acc1: 89.0625 (96.2135) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 420/1403] eta: 0:01:40 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2432 (0.1592) acc1: 90.6250 (96.1216) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 430/1403] eta: 0:01:39 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2371 (0.1618) acc1: 90.6250 (95.9723) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 440/1403] eta: 0:01:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2939 (0.1652) acc1: 89.0625 (95.8085) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 450/1403] eta: 0:01:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2660 (0.1673) acc1: 89.0625 (95.6901) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 460/1403] eta: 0:01:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2594 (0.1697) acc1: 89.0625 (95.5362) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 470/1403] eta: 0:01:35 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2137 (0.1688) acc1: 92.1875 (95.5646) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 480/1403] eta: 0:01:33 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0632 (0.1666) acc1: 100.0000 (95.6568) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 490/1403] eta: 0:01:32 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0632 (0.1645) acc1: 100.0000 (95.7453) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 500/1403] eta: 0:01:31 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0635 (0.1625) acc1: 100.0000 (95.8302) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 510/1403] eta: 0:01:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0635 (0.1606) acc1: 100.0000 (95.9118) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 520/1403] eta: 0:01:29 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0682 (0.1589) acc1: 100.0000 (95.9903) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 530/1403] eta: 0:01:28 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0590 (0.1570) acc1: 100.0000 (96.0658) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 540/1403] eta: 0:01:27 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0585 (0.1552) acc1: 100.0000 (96.1385) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 550/1403] eta: 0:01:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0593 (0.1539) acc1: 100.0000 (96.1973) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 560/1403] eta: 0:01:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0819 (0.1529) acc1: 98.4375 (96.2372) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 570/1403] eta: 0:01:23 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0827 (0.1518) acc1: 98.4375 (96.2839) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 580/1403] eta: 0:01:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0971 (0.1513) acc1: 98.4375 (96.2995) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 590/1403] eta: 0:01:21 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1152 (0.1510) acc1: 96.8750 (96.3013) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 600/1403] eta: 0:01:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1271 (0.1513) acc1: 95.3125 (96.2458) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 610/1403] eta: 0:01:19 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1271 (0.1509) acc1: 95.3125 (96.2485) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 620/1403] eta: 0:01:18 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1367 (0.1518) acc1: 95.3125 (96.1629) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 630/1403] eta: 0:01:17 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1476 (0.1516) acc1: 95.3125 (96.1594) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 640/1403] eta: 0:01:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1213 (0.1511) acc1: 96.8750 (96.1827) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 650/1403] eta: 0:01:15 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1213 (0.1507) acc1: 98.4375 (96.2078) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 660/1403] eta: 0:01:14 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0939 (0.1497) acc1: 98.4375 (96.2557) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 670/1403] eta: 0:01:13 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0693 (0.1485) acc1: 100.0000 (96.3115) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 680/1403] eta: 0:01:12 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0803 (0.1497) acc1: 100.0000 (96.2326) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 690/1403] eta: 0:01:11 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2483 (0.1537) acc1: 89.0625 (95.9999) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 700/1403] eta: 0:01:10 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1605 (0.1538) acc1: 90.6250 (95.9856) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 710/1403] eta: 0:01:09 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2761 (0.1580) acc1: 87.5000 (95.7608) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 720/1403] eta: 0:01:08 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2761 (0.1579) acc1: 85.9375 (95.7589) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 730/1403] eta: 0:01:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0661 (0.1568) acc1: 100.0000 (95.8063) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 740/1403] eta: 0:01:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0760 (0.1560) acc1: 100.0000 (95.8481) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 750/1403] eta: 0:01:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1009 (0.1574) acc1: 98.4375 (95.7681) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 760/1403] eta: 0:01:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3229 (0.1611) acc1: 82.8125 (95.5301) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 770/1403] eta: 0:01:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4565 (0.1659) acc1: 75.0000 (95.2395) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [ 780/1403] eta: 0:01:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4235 (0.1685) acc1: 78.1250 (95.0924) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [ 790/1403] eta: 0:01:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3135 (0.1711) acc1: 85.9375 (94.9550) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.0959 data: 0.0003 max mem: 52126 +Test: [ 800/1403] eta: 0:00:59 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2328 (0.1715) acc1: 89.0625 (94.9321) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 810/1403] eta: 0:00:58 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2205 (0.1745) acc1: 89.0625 (94.7962) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 820/1403] eta: 0:00:57 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6402 (0.1833) acc1: 71.8750 (94.4180) acc5: 100.0000 (100.0000) acc1_neg: 42.188 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 830/1403] eta: 0:00:56 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.7211 (0.1864) acc1: 68.7500 (94.2690) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 840/1403] eta: 0:00:55 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5265 (0.1946) acc1: 70.3125 (93.9154) acc5: 100.0000 (100.0000) acc1_neg: 50.000 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 850/1403] eta: 0:00:54 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.0587 (0.2057) acc1: 53.1250 (93.4415) acc5: 100.0000 (100.0000) acc1_neg: 51.562 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 860/1403] eta: 0:00:53 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.0016 (0.2092) acc1: 56.2500 (93.2891) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 870/1403] eta: 0:00:52 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1344 (0.2081) acc1: 95.3125 (93.3320) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 880/1403] eta: 0:00:51 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1029 (0.2070) acc1: 96.8750 (93.3864) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 890/1403] eta: 0:00:50 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0825 (0.2056) acc1: 98.4375 (93.4519) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 900/1403] eta: 0:00:49 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0741 (0.2041) acc1: 100.0000 (93.5228) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 910/1403] eta: 0:00:48 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0759 (0.2031) acc1: 100.0000 (93.5665) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 920/1403] eta: 0:00:47 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0988 (0.2021) acc1: 98.4375 (93.6058) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 930/1403] eta: 0:00:46 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0911 (0.2009) acc1: 98.4375 (93.6610) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 940/1403] eta: 0:00:45 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1083 (0.2009) acc1: 96.8750 (93.6454) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 950/1403] eta: 0:00:44 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2257 (0.2016) acc1: 89.0625 (93.5956) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [ 960/1403] eta: 0:00:43 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2435 (0.2019) acc1: 89.0625 (93.5646) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0959 data: 0.0003 max mem: 52126 +Test: [ 970/1403] eta: 0:00:42 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2191 (0.2026) acc1: 92.1875 (93.5231) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [ 980/1403] eta: 0:00:41 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2299 (0.2032) acc1: 89.0625 (93.4776) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [ 990/1403] eta: 0:00:40 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2286 (0.2034) acc1: 92.1875 (93.4552) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0959 data: 0.0003 max mem: 52126 +Test: [1000/1403] eta: 0:00:39 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2138 (0.2038) acc1: 92.1875 (93.4269) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [1010/1403] eta: 0:00:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2446 (0.2045) acc1: 90.6250 (93.3884) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1020/1403] eta: 0:00:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3861 (0.2105) acc1: 81.2500 (93.0629) acc5: 100.0000 (100.0000) acc1_neg: 35.938 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1030/1403] eta: 0:00:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5279 (0.2122) acc1: 65.6250 (92.9544) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [1040/1403] eta: 0:00:35 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2217 (0.2119) acc1: 90.6250 (92.9590) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1050/1403] eta: 0:00:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1689 (0.2115) acc1: 95.3125 (92.9784) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1060/1403] eta: 0:00:33 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1314 (0.2107) acc1: 96.8750 (93.0181) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1070/1403] eta: 0:00:32 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1452 (0.2111) acc1: 96.8750 (93.0030) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1080/1403] eta: 0:00:31 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3443 (0.2127) acc1: 85.9375 (92.9174) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1090/1403] eta: 0:00:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3861 (0.2143) acc1: 84.3750 (92.8363) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1100/1403] eta: 0:00:29 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3096 (0.2146) acc1: 85.9375 (92.8162) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [1110/1403] eta: 0:00:28 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2417 (0.2148) acc1: 90.6250 (92.8077) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1120/1403] eta: 0:00:27 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2457 (0.2151) acc1: 90.6250 (92.7924) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0957 data: 0.0002 max mem: 52126 +Test: [1130/1403] eta: 0:00:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2521 (0.2156) acc1: 90.6250 (92.7719) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0957 data: 0.0002 max mem: 52126 +Test: [1140/1403] eta: 0:00:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2192 (0.2156) acc1: 92.1875 (92.7668) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1150/1403] eta: 0:00:24 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2111 (0.2157) acc1: 92.1875 (92.7522) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1160/1403] eta: 0:00:23 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2265 (0.2159) acc1: 90.6250 (92.7433) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1170/1403] eta: 0:00:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2202 (0.2157) acc1: 92.1875 (92.7546) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [1180/1403] eta: 0:00:21 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0692 (0.2144) acc1: 100.0000 (92.8159) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1190/1403] eta: 0:00:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0690 (0.2132) acc1: 100.0000 (92.8763) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0957 data: 0.0002 max mem: 52126 +Test: [1200/1403] eta: 0:00:19 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0692 (0.2120) acc1: 100.0000 (92.9356) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [1210/1403] eta: 0:00:18 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0681 (0.2108) acc1: 100.0000 (92.9939) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [1220/1403] eta: 0:00:17 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0679 (0.2097) acc1: 100.0000 (93.0513) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0959 data: 0.0003 max mem: 52126 +Test: [1230/1403] eta: 0:00:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0689 (0.2085) acc1: 100.0000 (93.1077) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0959 data: 0.0003 max mem: 52126 +Test: [1240/1403] eta: 0:00:15 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0686 (0.2074) acc1: 100.0000 (93.1633) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0959 data: 0.0003 max mem: 52126 +Test: [1250/1403] eta: 0:00:15 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0689 (0.2063) acc1: 100.0000 (93.2179) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1260/1403] eta: 0:00:14 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0758 (0.2053) acc1: 100.0000 (93.2717) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1270/1403] eta: 0:00:13 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0773 (0.2043) acc1: 100.0000 (93.3246) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1280/1403] eta: 0:00:12 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0769 (0.2033) acc1: 100.0000 (93.3768) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0957 data: 0.0002 max mem: 52126 +Test: [1290/1403] eta: 0:00:11 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0763 (0.2023) acc1: 100.0000 (93.4281) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [1300/1403] eta: 0:00:10 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0774 (0.2014) acc1: 100.0000 (93.4786) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0959 data: 0.0003 max mem: 52126 +Test: [1310/1403] eta: 0:00:09 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0774 (0.2004) acc1: 100.0000 (93.5283) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1320/1403] eta: 0:00:08 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0754 (0.1995) acc1: 100.0000 (93.5773) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1330/1403] eta: 0:00:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0747 (0.1985) acc1: 100.0000 (93.6244) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1340/1403] eta: 0:00:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0669 (0.1975) acc1: 100.0000 (93.6696) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1350/1403] eta: 0:00:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0652 (0.1966) acc1: 100.0000 (93.7130) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0959 data: 0.0002 max mem: 52126 +Test: [1360/1403] eta: 0:00:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0652 (0.1956) acc1: 100.0000 (93.7580) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0003 max mem: 52126 +Test: [1370/1403] eta: 0:00:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0650 (0.1947) acc1: 100.0000 (93.8013) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1380/1403] eta: 0:00:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0646 (0.1938) acc1: 100.0000 (93.8462) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0958 data: 0.0002 max mem: 52126 +Test: [1390/1403] eta: 0:00:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0643 (0.1929) acc1: 100.0000 (93.8904) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0957 data: 0.0002 max mem: 52126 +Test: [1400/1403] eta: 0:00:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0646 (0.1920) acc1: 100.0000 (93.9318) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0956 data: 0.0001 max mem: 52126 +Test: [1402/1403] eta: 0:00:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0649 (0.1918) acc1: 100.0000 (93.9362) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0916 data: 0.0001 max mem: 52126 +Test: Total time: 0:02:17 (0.0982 s / it) +* Overall Acc@1 93.936 Acc@5 100.000 + target == real Acc@1 95.854 Acc@5 100.000 + target == fake Acc@1 93.396 Acc@5 100.000 + loss 0.192 +Accuracy of the network on the 89729 test images: 93.9% +Max accuracy: 94.13% +Epoch: [30] [ 0/3634] eta: 6:28:12 lr: 0.000003 loss: 0.5293 (0.5293) time: 6.4095 data: 4.2494 max mem: 52126 +Epoch: [30] [ 10/3634] eta: 1:08:26 lr: 0.000003 loss: 0.5420 (0.5568) time: 1.1330 data: 0.3864 max mem: 52126 +Epoch: [30] [ 20/3634] eta: 0:49:21 lr: 0.000003 loss: 0.5307 (0.5118) time: 0.5401 data: 0.0002 max mem: 52126 +Epoch: [30] [ 30/3634] eta: 0:45:02 lr: 0.000003 loss: 0.4493 (0.4971) time: 0.5393 data: 0.0002 max mem: 52126 +Epoch: [30] [ 40/3634] eta: 0:40:52 lr: 0.000003 loss: 0.5064 (0.4992) time: 0.5383 data: 0.0002 max mem: 52126 +Epoch: [30] [ 50/3634] eta: 0:41:16 lr: 0.000003 loss: 0.5229 (0.4994) time: 0.5994 data: 0.0002 max mem: 52126 +Epoch: [30] [ 60/3634] eta: 0:40:18 lr: 0.000003 loss: 0.4971 (0.4983) time: 0.6653 data: 0.0002 max mem: 52126 +Epoch: [30] [ 70/3634] eta: 0:37:43 lr: 0.000003 loss: 0.4810 (0.4968) time: 0.4930 data: 0.0002 max mem: 52126 +Epoch: [30] [ 80/3634] eta: 0:35:45 lr: 0.000003 loss: 0.4859 (0.4947) time: 0.3813 data: 0.0002 max mem: 52126 +Epoch: [30] [ 90/3634] eta: 0:34:13 lr: 0.000003 loss: 0.4637 (0.4910) time: 0.3811 data: 0.0002 max mem: 52126 +Epoch: [30] [ 100/3634] eta: 0:32:58 lr: 0.000003 loss: 0.4565 (0.4889) time: 0.3818 data: 0.0002 max mem: 52126 +Epoch: [30] [ 110/3634] eta: 0:32:25 lr: 0.000003 loss: 0.4830 (0.4909) time: 0.4288 data: 0.0002 max mem: 52126 +Epoch: [30] [ 120/3634] eta: 0:32:33 lr: 0.000003 loss: 0.5061 (0.4894) time: 0.5365 data: 0.0002 max mem: 52126 +Epoch: [30] [ 130/3634] eta: 0:32:46 lr: 0.000003 loss: 0.5214 (0.4932) time: 0.6104 data: 0.0002 max mem: 52126 +Epoch: [30] [ 140/3634] eta: 0:32:26 lr: 0.000003 loss: 0.5468 (0.4942) time: 0.5632 data: 0.0002 max mem: 52126 +Epoch: [30] [ 150/3634] eta: 0:32:37 lr: 0.000003 loss: 0.5418 (0.4974) time: 0.5664 data: 0.0002 max mem: 52126 +Epoch: [30] [ 160/3634] eta: 0:31:52 lr: 0.000003 loss: 0.5362 (0.4981) time: 0.5054 data: 0.0001 max mem: 52126 +Epoch: [30] [ 170/3634] eta: 0:31:32 lr: 0.000003 loss: 0.5362 (0.4998) time: 0.4304 data: 0.0002 max mem: 52126 +Epoch: [30] [ 180/3634] eta: 0:30:55 lr: 0.000003 loss: 0.5615 (0.5040) time: 0.4305 data: 0.0002 max mem: 52126 +Epoch: [30] [ 190/3634] eta: 0:31:19 lr: 0.000003 loss: 0.5449 (0.5032) time: 0.5397 data: 0.0002 max mem: 52126 +Epoch: [30] [ 200/3634] eta: 0:30:45 lr: 0.000003 loss: 0.4849 (0.5020) time: 0.5395 data: 0.0002 max mem: 52126 +Epoch: [30] [ 210/3634] eta: 0:30:35 lr: 0.000003 loss: 0.5005 (0.5030) time: 0.4438 data: 0.0002 max mem: 52126 +Epoch: [30] [ 220/3634] eta: 0:30:40 lr: 0.000003 loss: 0.5337 (0.5044) time: 0.5550 data: 0.0002 max mem: 52126 +Epoch: [30] [ 230/3634] eta: 0:30:25 lr: 0.000003 loss: 0.5333 (0.5034) time: 0.5388 data: 0.0001 max mem: 52126 +Epoch: [30] [ 240/3634] eta: 0:30:15 lr: 0.000003 loss: 0.5144 (0.5044) time: 0.4889 data: 0.0002 max mem: 52126 +Epoch: [30] [ 250/3634] eta: 0:30:02 lr: 0.000003 loss: 0.4819 (0.5039) time: 0.4897 data: 0.0002 max mem: 52126 +Epoch: [30] [ 260/3634] eta: 0:29:37 lr: 0.000003 loss: 0.4693 (0.5029) time: 0.4290 data: 0.0002 max mem: 52126 +Epoch: [30] [ 270/3634] eta: 0:29:26 lr: 0.000003 loss: 0.5064 (0.5039) time: 0.4299 data: 0.0002 max mem: 52126 +Epoch: [30] [ 280/3634] eta: 0:29:03 lr: 0.000003 loss: 0.5250 (0.5039) time: 0.4296 data: 0.0002 max mem: 52126 +Epoch: [30] [ 290/3634] eta: 0:29:09 lr: 0.000003 loss: 0.5209 (0.5048) time: 0.4984 data: 0.0002 max mem: 52126 +Epoch: [30] [ 300/3634] eta: 0:29:12 lr: 0.000003 loss: 0.5461 (0.5053) time: 0.6079 data: 0.0002 max mem: 52126 +Epoch: [30] [ 310/3634] eta: 0:28:52 lr: 0.000003 loss: 0.4991 (0.5049) time: 0.4918 data: 0.0002 max mem: 52126 +Epoch: [30] [ 320/3634] eta: 0:28:42 lr: 0.000003 loss: 0.5277 (0.5059) time: 0.4283 data: 0.0002 max mem: 52126 +Epoch: [30] [ 330/3634] eta: 0:28:23 lr: 0.000003 loss: 0.5176 (0.5044) time: 0.4281 data: 0.0002 max mem: 52126 +Epoch: [30] [ 340/3634] eta: 0:28:17 lr: 0.000003 loss: 0.4837 (0.5038) time: 0.4440 data: 0.0002 max mem: 52126 +Epoch: [30] [ 350/3634] eta: 0:27:59 lr: 0.000003 loss: 0.5067 (0.5051) time: 0.4440 data: 0.0002 max mem: 52126 +Epoch: [30] [ 360/3634] eta: 0:27:53 lr: 0.000003 loss: 0.5383 (0.5061) time: 0.4422 data: 0.0002 max mem: 52126 +Epoch: [30] [ 370/3634] eta: 0:27:37 lr: 0.000003 loss: 0.5291 (0.5064) time: 0.4420 data: 0.0002 max mem: 52126 +Epoch: [30] [ 380/3634] eta: 0:27:40 lr: 0.000003 loss: 0.5142 (0.5065) time: 0.4915 data: 0.0002 max mem: 52126 +Epoch: [30] [ 390/3634] eta: 0:27:32 lr: 0.000003 loss: 0.4984 (0.5066) time: 0.5412 data: 0.0002 max mem: 52126 +Epoch: [30] [ 400/3634] eta: 0:27:17 lr: 0.000003 loss: 0.5135 (0.5069) time: 0.4316 data: 0.0002 max mem: 52126 +Epoch: [30] [ 410/3634] eta: 0:27:16 lr: 0.000003 loss: 0.5296 (0.5082) time: 0.4740 data: 0.0002 max mem: 52126 +Epoch: [30] [ 420/3634] eta: 0:27:11 lr: 0.000003 loss: 0.5610 (0.5088) time: 0.5354 data: 0.0002 max mem: 52126 +Epoch: [30] [ 430/3634] eta: 0:27:04 lr: 0.000003 loss: 0.5447 (0.5094) time: 0.4909 data: 0.0002 max mem: 52126 +Epoch: [30] [ 440/3634] eta: 0:26:56 lr: 0.000003 loss: 0.5431 (0.5092) time: 0.4759 data: 0.0002 max mem: 52126 +Epoch: [30] [ 450/3634] eta: 0:27:13 lr: 0.000003 loss: 0.4640 (0.5080) time: 0.6435 data: 0.0002 max mem: 52126 +Epoch: [30] [ 460/3634] eta: 0:27:07 lr: 0.000003 loss: 0.5011 (0.5084) time: 0.6588 data: 0.0001 max mem: 52126 +Epoch: [30] [ 470/3634] eta: 0:27:02 lr: 0.000003 loss: 0.4911 (0.5076) time: 0.5063 data: 0.0002 max mem: 52126 +Epoch: [30] [ 480/3634] eta: 0:26:54 lr: 0.000003 loss: 0.4879 (0.5075) time: 0.4933 data: 0.0002 max mem: 52126 +Epoch: [30] [ 490/3634] eta: 0:26:55 lr: 0.000003 loss: 0.5148 (0.5081) time: 0.5419 data: 0.0002 max mem: 52126 +Epoch: [30] [ 500/3634] eta: 0:26:42 lr: 0.000003 loss: 0.5149 (0.5080) time: 0.4944 data: 0.0002 max mem: 52126 +Epoch: [30] [ 510/3634] eta: 0:26:29 lr: 0.000003 loss: 0.5016 (0.5077) time: 0.3830 data: 0.0002 max mem: 52126 +Epoch: [30] [ 520/3634] eta: 0:26:21 lr: 0.000003 loss: 0.4863 (0.5074) time: 0.4244 data: 0.0002 max mem: 52126 +Epoch: [30] [ 530/3634] eta: 0:26:09 lr: 0.000003 loss: 0.5116 (0.5084) time: 0.4253 data: 0.0002 max mem: 52126 +Epoch: [30] [ 540/3634] eta: 0:26:14 lr: 0.000003 loss: 0.5392 (0.5085) time: 0.5326 data: 0.0002 max mem: 52126 +Epoch: [30] [ 550/3634] eta: 0:26:32 lr: 0.000003 loss: 0.4955 (0.5084) time: 0.8036 data: 0.0002 max mem: 52126 +Epoch: [30] [ 560/3634] eta: 0:26:26 lr: 0.000003 loss: 0.4955 (0.5081) time: 0.7126 data: 0.0002 max mem: 52126 +Epoch: [30] [ 570/3634] eta: 0:26:14 lr: 0.000003 loss: 0.5243 (0.5084) time: 0.4413 data: 0.0002 max mem: 52126 +Epoch: [30] [ 580/3634] eta: 0:26:07 lr: 0.000003 loss: 0.5154 (0.5079) time: 0.4323 data: 0.0002 max mem: 52126 +Epoch: [30] [ 590/3634] eta: 0:25:55 lr: 0.000003 loss: 0.4979 (0.5084) time: 0.4322 data: 0.0002 max mem: 52126 +Epoch: [30] [ 600/3634] eta: 0:25:43 lr: 0.000003 loss: 0.5043 (0.5083) time: 0.3817 data: 0.0002 max mem: 52126 +Epoch: [30] [ 610/3634] eta: 0:25:32 lr: 0.000003 loss: 0.5043 (0.5082) time: 0.3813 data: 0.0002 max mem: 52126 +Epoch: [30] [ 620/3634] eta: 0:25:21 lr: 0.000003 loss: 0.4809 (0.5082) time: 0.3819 data: 0.0002 max mem: 52126 +Epoch: [30] [ 630/3634] eta: 0:25:10 lr: 0.000003 loss: 0.4842 (0.5078) time: 0.3823 data: 0.0002 max mem: 52126 +Epoch: [30] [ 640/3634] eta: 0:25:09 lr: 0.000003 loss: 0.5007 (0.5077) time: 0.4864 data: 0.0001 max mem: 52126 +Epoch: [30] [ 650/3634] eta: 0:25:19 lr: 0.000003 loss: 0.5249 (0.5085) time: 0.7080 data: 0.0001 max mem: 52126 +Epoch: [30] [ 660/3634] eta: 0:25:27 lr: 0.000003 loss: 0.5283 (0.5085) time: 0.8153 data: 0.0001 max mem: 52126 +Epoch: [30] [ 670/3634] eta: 0:25:16 lr: 0.000003 loss: 0.5111 (0.5088) time: 0.5938 data: 0.0001 max mem: 52126 +Epoch: [30] [ 680/3634] eta: 0:25:09 lr: 0.000003 loss: 0.5004 (0.5085) time: 0.4257 data: 0.0002 max mem: 52126 +Epoch: [30] [ 690/3634] eta: 0:24:58 lr: 0.000003 loss: 0.4840 (0.5084) time: 0.4247 data: 0.0002 max mem: 52126 +Epoch: [30] [ 700/3634] eta: 0:24:52 lr: 0.000003 loss: 0.5122 (0.5084) time: 0.4249 data: 0.0002 max mem: 52126 +Epoch: [30] [ 710/3634] eta: 0:24:50 lr: 0.000003 loss: 0.4984 (0.5082) time: 0.5285 data: 0.0002 max mem: 52126 +Epoch: [30] [ 720/3634] eta: 0:24:40 lr: 0.000003 loss: 0.4984 (0.5083) time: 0.4845 data: 0.0002 max mem: 52126 +Epoch: [30] [ 730/3634] eta: 0:24:33 lr: 0.000003 loss: 0.5028 (0.5079) time: 0.4277 data: 0.0002 max mem: 52126 +Epoch: [30] [ 740/3634] eta: 0:24:32 lr: 0.000003 loss: 0.5125 (0.5083) time: 0.5460 data: 0.0002 max mem: 52126 +Epoch: [30] [ 750/3634] eta: 0:24:22 lr: 0.000003 loss: 0.5125 (0.5082) time: 0.4997 data: 0.0002 max mem: 52126 +Epoch: [30] [ 760/3634] eta: 0:24:13 lr: 0.000003 loss: 0.5112 (0.5084) time: 0.3815 data: 0.0001 max mem: 52126 +Epoch: [30] [ 770/3634] eta: 0:24:03 lr: 0.000003 loss: 0.5112 (0.5084) time: 0.3813 data: 0.0002 max mem: 52126 +Epoch: [30] [ 780/3634] eta: 0:23:57 lr: 0.000003 loss: 0.5144 (0.5087) time: 0.4266 data: 0.0002 max mem: 52126 +Epoch: [30] [ 790/3634] eta: 0:23:47 lr: 0.000003 loss: 0.5246 (0.5088) time: 0.4272 data: 0.0002 max mem: 52126 +Epoch: [30] [ 800/3634] eta: 0:23:45 lr: 0.000003 loss: 0.5267 (0.5088) time: 0.4875 data: 0.0001 max mem: 52126 +Epoch: [30] [ 810/3634] eta: 0:23:39 lr: 0.000003 loss: 0.5239 (0.5087) time: 0.5319 data: 0.0001 max mem: 52126 +Epoch: [30] [ 820/3634] eta: 0:23:30 lr: 0.000003 loss: 0.5325 (0.5091) time: 0.4263 data: 0.0001 max mem: 52126 +Epoch: [30] [ 830/3634] eta: 0:23:24 lr: 0.000003 loss: 0.5372 (0.5092) time: 0.4268 data: 0.0001 max mem: 52126 +Epoch: [30] [ 840/3634] eta: 0:23:15 lr: 0.000003 loss: 0.5310 (0.5095) time: 0.4268 data: 0.0001 max mem: 52126 +Epoch: [30] [ 850/3634] eta: 0:23:10 lr: 0.000003 loss: 0.5384 (0.5098) time: 0.4405 data: 0.0002 max mem: 52126 +Epoch: [30] [ 860/3634] eta: 0:23:05 lr: 0.000003 loss: 0.5124 (0.5096) time: 0.5002 data: 0.0002 max mem: 52126 +Epoch: [30] [ 870/3634] eta: 0:22:56 lr: 0.000003 loss: 0.5140 (0.5100) time: 0.4412 data: 0.0002 max mem: 52126 +Epoch: [30] [ 880/3634] eta: 0:22:51 lr: 0.000003 loss: 0.5198 (0.5095) time: 0.4264 data: 0.0002 max mem: 52126 +Epoch: [30] [ 890/3634] eta: 0:22:46 lr: 0.000003 loss: 0.5116 (0.5099) time: 0.4859 data: 0.0002 max mem: 52126 +Epoch: [30] [ 900/3634] eta: 0:22:37 lr: 0.000003 loss: 0.5200 (0.5098) time: 0.4412 data: 0.0002 max mem: 52126 +Epoch: [30] [ 910/3634] eta: 0:22:32 lr: 0.000003 loss: 0.5076 (0.5100) time: 0.4410 data: 0.0001 max mem: 52126 +Epoch: [30] [ 920/3634] eta: 0:22:37 lr: 0.000003 loss: 0.5131 (0.5099) time: 0.6599 data: 0.0001 max mem: 52126 +Epoch: [30] [ 930/3634] eta: 0:22:35 lr: 0.000003 loss: 0.5086 (0.5101) time: 0.7063 data: 0.0001 max mem: 52126 +Epoch: [30] [ 940/3634] eta: 0:22:29 lr: 0.000003 loss: 0.5438 (0.5102) time: 0.5313 data: 0.0002 max mem: 52126 +Epoch: [30] [ 950/3634] eta: 0:22:24 lr: 0.000003 loss: 0.5098 (0.5103) time: 0.4843 data: 0.0002 max mem: 52126 +Epoch: [30] [ 960/3634] eta: 0:22:18 lr: 0.000003 loss: 0.4946 (0.5103) time: 0.4843 data: 0.0001 max mem: 52126 +Epoch: [30] [ 970/3634] eta: 0:22:12 lr: 0.000003 loss: 0.4811 (0.5103) time: 0.4701 data: 0.0001 max mem: 52126 +Epoch: [30] [ 980/3634] eta: 0:22:04 lr: 0.000003 loss: 0.5586 (0.5108) time: 0.4257 data: 0.0002 max mem: 52126 +Epoch: [30] [ 990/3634] eta: 0:21:58 lr: 0.000003 loss: 0.5673 (0.5108) time: 0.4277 data: 0.0002 max mem: 52126 +Epoch: [30] [1000/3634] eta: 0:21:50 lr: 0.000003 loss: 0.5593 (0.5113) time: 0.4283 data: 0.0002 max mem: 52126 +Epoch: [30] [1010/3634] eta: 0:21:42 lr: 0.000003 loss: 0.5373 (0.5112) time: 0.3816 data: 0.0002 max mem: 52126 +Epoch: [30] [1020/3634] eta: 0:21:39 lr: 0.000003 loss: 0.5449 (0.5115) time: 0.4773 data: 0.0002 max mem: 52126 +Epoch: [30] [1030/3634] eta: 0:21:31 lr: 0.000003 loss: 0.5674 (0.5118) time: 0.4775 data: 0.0001 max mem: 52126 +Epoch: [30] [1040/3634] eta: 0:21:23 lr: 0.000003 loss: 0.5009 (0.5114) time: 0.3818 data: 0.0002 max mem: 52126 +Epoch: [30] [1050/3634] eta: 0:21:18 lr: 0.000003 loss: 0.5009 (0.5115) time: 0.4274 data: 0.0002 max mem: 52126 +Epoch: [30] [1060/3634] eta: 0:21:18 lr: 0.000003 loss: 0.5172 (0.5114) time: 0.5838 data: 0.0001 max mem: 52126 +Epoch: [30] [1070/3634] eta: 0:21:13 lr: 0.000003 loss: 0.5343 (0.5118) time: 0.5989 data: 0.0002 max mem: 52126 +Epoch: [30] [1080/3634] eta: 0:21:05 lr: 0.000003 loss: 0.5177 (0.5117) time: 0.4448 data: 0.0002 max mem: 52126 +Epoch: [30] [1090/3634] eta: 0:21:01 lr: 0.000003 loss: 0.5198 (0.5119) time: 0.4459 data: 0.0001 max mem: 52126 +Epoch: [30] [1100/3634] eta: 0:20:58 lr: 0.000003 loss: 0.5424 (0.5121) time: 0.5483 data: 0.0001 max mem: 52126 +Epoch: [30] [1110/3634] eta: 0:20:55 lr: 0.000003 loss: 0.5381 (0.5124) time: 0.5913 data: 0.0001 max mem: 52126 +Epoch: [30] [1120/3634] eta: 0:20:50 lr: 0.000003 loss: 0.5164 (0.5123) time: 0.5453 data: 0.0001 max mem: 52126 +Epoch: [30] [1130/3634] eta: 0:20:45 lr: 0.000003 loss: 0.5111 (0.5125) time: 0.4857 data: 0.0001 max mem: 52126 +Epoch: [30] [1140/3634] eta: 0:20:40 lr: 0.000003 loss: 0.5069 (0.5124) time: 0.4849 data: 0.0002 max mem: 52126 +Epoch: [30] [1150/3634] eta: 0:20:32 lr: 0.000003 loss: 0.5035 (0.5125) time: 0.4395 data: 0.0002 max mem: 52126 +Epoch: [30] [1160/3634] eta: 0:20:25 lr: 0.000003 loss: 0.5197 (0.5124) time: 0.3835 data: 0.0002 max mem: 52126 +Epoch: [30] [1170/3634] eta: 0:20:24 lr: 0.000003 loss: 0.5197 (0.5126) time: 0.5312 data: 0.0001 max mem: 52126 +Epoch: [30] [1180/3634] eta: 0:20:16 lr: 0.000003 loss: 0.4862 (0.5122) time: 0.5297 data: 0.0001 max mem: 52126 +Epoch: [30] [1190/3634] eta: 0:20:13 lr: 0.000003 loss: 0.4862 (0.5122) time: 0.4749 data: 0.0002 max mem: 52126 +Epoch: [30] [1200/3634] eta: 0:20:13 lr: 0.000003 loss: 0.5093 (0.5119) time: 0.6526 data: 0.0002 max mem: 52126 +Epoch: [30] [1210/3634] eta: 0:20:06 lr: 0.000003 loss: 0.4978 (0.5119) time: 0.5601 data: 0.0002 max mem: 52126 +Epoch: [30] [1220/3634] eta: 0:20:03 lr: 0.000003 loss: 0.5543 (0.5123) time: 0.4999 data: 0.0002 max mem: 52126 +Epoch: [30] [1230/3634] eta: 0:20:00 lr: 0.000003 loss: 0.5642 (0.5126) time: 0.6191 data: 0.0002 max mem: 52126 +Epoch: [30] [1240/3634] eta: 0:19:57 lr: 0.000003 loss: 0.5312 (0.5127) time: 0.5946 data: 0.0002 max mem: 52126 +Epoch: [30] [1250/3634] eta: 0:19:53 lr: 0.000003 loss: 0.5312 (0.5129) time: 0.5623 data: 0.0002 max mem: 52126 +Epoch: [30] [1260/3634] eta: 0:19:48 lr: 0.000003 loss: 0.5286 (0.5129) time: 0.5286 data: 0.0002 max mem: 52126 +Epoch: [30] [1270/3634] eta: 0:19:44 lr: 0.000003 loss: 0.5391 (0.5132) time: 0.5471 data: 0.0002 max mem: 52126 +Epoch: [30] [1280/3634] eta: 0:19:39 lr: 0.000003 loss: 0.5391 (0.5133) time: 0.5311 data: 0.0002 max mem: 52126 +Epoch: [30] [1290/3634] eta: 0:19:32 lr: 0.000003 loss: 0.5369 (0.5136) time: 0.4258 data: 0.0002 max mem: 52126 +Epoch: [30] [1300/3634] eta: 0:19:26 lr: 0.000003 loss: 0.5402 (0.5136) time: 0.4258 data: 0.0001 max mem: 52126 +Epoch: [30] [1310/3634] eta: 0:19:21 lr: 0.000003 loss: 0.5236 (0.5137) time: 0.4735 data: 0.0001 max mem: 52126 +Epoch: [30] [1320/3634] eta: 0:19:16 lr: 0.000003 loss: 0.5011 (0.5134) time: 0.4896 data: 0.0002 max mem: 52126 +Epoch: [30] [1330/3634] eta: 0:19:10 lr: 0.000003 loss: 0.5080 (0.5136) time: 0.4867 data: 0.0002 max mem: 52126 +Epoch: [30] [1340/3634] eta: 0:19:05 lr: 0.000003 loss: 0.5182 (0.5136) time: 0.4870 data: 0.0002 max mem: 52126 +Epoch: [30] [1350/3634] eta: 0:19:01 lr: 0.000003 loss: 0.5330 (0.5138) time: 0.5319 data: 0.0002 max mem: 52126 +Epoch: [30] [1360/3634] eta: 0:18:54 lr: 0.000003 loss: 0.5374 (0.5139) time: 0.4719 data: 0.0002 max mem: 52126 +Epoch: [30] [1370/3634] eta: 0:18:52 lr: 0.000003 loss: 0.5374 (0.5140) time: 0.5053 data: 0.0002 max mem: 52126 +Epoch: [30] [1380/3634] eta: 0:18:45 lr: 0.000003 loss: 0.5314 (0.5139) time: 0.5044 data: 0.0001 max mem: 52126 +Epoch: [30] [1390/3634] eta: 0:18:38 lr: 0.000003 loss: 0.5239 (0.5140) time: 0.3811 data: 0.0001 max mem: 52126 +Epoch: [30] [1400/3634] eta: 0:18:31 lr: 0.000003 loss: 0.5043 (0.5137) time: 0.3820 data: 0.0001 max mem: 52126 +Epoch: [30] [1410/3634] eta: 0:18:25 lr: 0.000003 loss: 0.5250 (0.5139) time: 0.4243 data: 0.0002 max mem: 52126 +Epoch: [30] [1420/3634] eta: 0:18:20 lr: 0.000003 loss: 0.5413 (0.5140) time: 0.4695 data: 0.0001 max mem: 52126 +Epoch: [30] [1430/3634] eta: 0:18:13 lr: 0.000003 loss: 0.5518 (0.5143) time: 0.4277 data: 0.0001 max mem: 52126 +Epoch: [30] [1440/3634] eta: 0:18:10 lr: 0.000003 loss: 0.5477 (0.5144) time: 0.4851 data: 0.0001 max mem: 52126 +Epoch: [30] [1450/3634] eta: 0:18:04 lr: 0.000003 loss: 0.5320 (0.5144) time: 0.5292 data: 0.0002 max mem: 52126 +Epoch: [30] [1460/3634] eta: 0:18:00 lr: 0.000003 loss: 0.5197 (0.5144) time: 0.5128 data: 0.0002 max mem: 52126 +Epoch: [30] [1470/3634] eta: 0:17:57 lr: 0.000003 loss: 0.5482 (0.5145) time: 0.5773 data: 0.0001 max mem: 52126 +Epoch: [30] [1480/3634] eta: 0:17:50 lr: 0.000003 loss: 0.5411 (0.5145) time: 0.4911 data: 0.0001 max mem: 52126 +Epoch: [30] [1490/3634] eta: 0:17:45 lr: 0.000003 loss: 0.5112 (0.5145) time: 0.4413 data: 0.0001 max mem: 52126 +Epoch: [30] [1500/3634] eta: 0:17:40 lr: 0.000003 loss: 0.5110 (0.5146) time: 0.5000 data: 0.0001 max mem: 52126 +Epoch: [30] [1510/3634] eta: 0:17:36 lr: 0.000003 loss: 0.5110 (0.5146) time: 0.5299 data: 0.0001 max mem: 52126 +Epoch: [30] [1520/3634] eta: 0:17:31 lr: 0.000003 loss: 0.5356 (0.5148) time: 0.5304 data: 0.0002 max mem: 52126 +Epoch: [30] [1530/3634] eta: 0:17:28 lr: 0.000003 loss: 0.5359 (0.5149) time: 0.5450 data: 0.0002 max mem: 52126 +Epoch: [30] [1540/3634] eta: 0:17:21 lr: 0.000003 loss: 0.5049 (0.5147) time: 0.4859 data: 0.0001 max mem: 52126 +Epoch: [30] [1550/3634] eta: 0:17:16 lr: 0.000003 loss: 0.5064 (0.5147) time: 0.4297 data: 0.0002 max mem: 52126 +Epoch: [30] [1560/3634] eta: 0:17:12 lr: 0.000003 loss: 0.5124 (0.5148) time: 0.5336 data: 0.0002 max mem: 52126 +Epoch: [30] [1570/3634] eta: 0:17:06 lr: 0.000003 loss: 0.4921 (0.5146) time: 0.4858 data: 0.0002 max mem: 52126 +Epoch: [30] [1580/3634] eta: 0:17:01 lr: 0.000003 loss: 0.4849 (0.5145) time: 0.4439 data: 0.0002 max mem: 52126 +Epoch: [30] [1590/3634] eta: 0:16:55 lr: 0.000003 loss: 0.5198 (0.5145) time: 0.4896 data: 0.0001 max mem: 52126 +Epoch: [30] [1600/3634] eta: 0:16:53 lr: 0.000003 loss: 0.5027 (0.5142) time: 0.5754 data: 0.0001 max mem: 52126 +Epoch: [30] [1610/3634] eta: 0:16:48 lr: 0.000003 loss: 0.5275 (0.5145) time: 0.5898 data: 0.0001 max mem: 52126 +Epoch: [30] [1620/3634] eta: 0:16:41 lr: 0.000003 loss: 0.5275 (0.5143) time: 0.4415 data: 0.0001 max mem: 52126 +Epoch: [30] [1630/3634] eta: 0:16:35 lr: 0.000003 loss: 0.4898 (0.5143) time: 0.3822 data: 0.0001 max mem: 52126 +Epoch: [30] [1640/3634] eta: 0:16:32 lr: 0.000003 loss: 0.5328 (0.5142) time: 0.5013 data: 0.0002 max mem: 52126 +Epoch: [30] [1650/3634] eta: 0:16:28 lr: 0.000003 loss: 0.5328 (0.5144) time: 0.6053 data: 0.0002 max mem: 52126 +Epoch: [30] [1660/3634] eta: 0:16:21 lr: 0.000003 loss: 0.5330 (0.5144) time: 0.4868 data: 0.0002 max mem: 52126 +Epoch: [30] [1670/3634] eta: 0:16:17 lr: 0.000003 loss: 0.5318 (0.5143) time: 0.4856 data: 0.0001 max mem: 52126 +Epoch: [30] [1680/3634] eta: 0:16:12 lr: 0.000003 loss: 0.5328 (0.5144) time: 0.5313 data: 0.0001 max mem: 52126 +Epoch: [30] [1690/3634] eta: 0:16:07 lr: 0.000003 loss: 0.5388 (0.5145) time: 0.4872 data: 0.0001 max mem: 52126 +Epoch: [30] [1700/3634] eta: 0:16:02 lr: 0.000003 loss: 0.5190 (0.5145) time: 0.5017 data: 0.0001 max mem: 52126 +Epoch: [30] [1710/3634] eta: 0:15:57 lr: 0.000003 loss: 0.5138 (0.5145) time: 0.4859 data: 0.0001 max mem: 52126 +Epoch: [30] [1720/3634] eta: 0:15:52 lr: 0.000003 loss: 0.5467 (0.5148) time: 0.4861 data: 0.0001 max mem: 52126 +Epoch: [30] [1730/3634] eta: 0:15:47 lr: 0.000003 loss: 0.5467 (0.5148) time: 0.4891 data: 0.0001 max mem: 52126 +Epoch: [30] [1740/3634] eta: 0:15:41 lr: 0.000003 loss: 0.4975 (0.5147) time: 0.4287 data: 0.0001 max mem: 52126 +Epoch: [30] [1750/3634] eta: 0:15:36 lr: 0.000003 loss: 0.5036 (0.5147) time: 0.4418 data: 0.0002 max mem: 52126 +Epoch: [30] [1760/3634] eta: 0:15:31 lr: 0.000003 loss: 0.5158 (0.5147) time: 0.5012 data: 0.0002 max mem: 52126 +Epoch: [30] [1770/3634] eta: 0:15:26 lr: 0.000003 loss: 0.5246 (0.5147) time: 0.5309 data: 0.0001 max mem: 52126 +Epoch: [30] [1780/3634] eta: 0:15:20 lr: 0.000003 loss: 0.5246 (0.5147) time: 0.4712 data: 0.0001 max mem: 52126 +Epoch: [30] [1790/3634] eta: 0:15:17 lr: 0.000003 loss: 0.4656 (0.5143) time: 0.5444 data: 0.0001 max mem: 52126 +Epoch: [30] [1800/3634] eta: 0:15:13 lr: 0.000003 loss: 0.4884 (0.5144) time: 0.6489 data: 0.0002 max mem: 52126 +Epoch: [30] [1810/3634] eta: 0:15:10 lr: 0.000003 loss: 0.5126 (0.5141) time: 0.6270 data: 0.0002 max mem: 52126 +Epoch: [30] [1820/3634] eta: 0:15:05 lr: 0.000003 loss: 0.5161 (0.5143) time: 0.5715 data: 0.0001 max mem: 52126 +Epoch: [30] [1830/3634] eta: 0:15:01 lr: 0.000003 loss: 0.5203 (0.5143) time: 0.5214 data: 0.0001 max mem: 52126 +Epoch: [30] [1840/3634] eta: 0:14:55 lr: 0.000003 loss: 0.5203 (0.5144) time: 0.5192 data: 0.0002 max mem: 52126 +Epoch: [30] [1850/3634] eta: 0:14:50 lr: 0.000003 loss: 0.5499 (0.5144) time: 0.4904 data: 0.0002 max mem: 52126 +Epoch: [30] [1860/3634] eta: 0:14:45 lr: 0.000003 loss: 0.5398 (0.5144) time: 0.4867 data: 0.0002 max mem: 52126 +Epoch: [30] [1870/3634] eta: 0:14:39 lr: 0.000003 loss: 0.5043 (0.5142) time: 0.4248 data: 0.0002 max mem: 52126 +Epoch: [30] [1880/3634] eta: 0:14:33 lr: 0.000003 loss: 0.5140 (0.5143) time: 0.3814 data: 0.0001 max mem: 52126 +Epoch: [30] [1890/3634] eta: 0:14:27 lr: 0.000003 loss: 0.5453 (0.5143) time: 0.3815 data: 0.0001 max mem: 52126 +Epoch: [30] [1900/3634] eta: 0:14:21 lr: 0.000003 loss: 0.5288 (0.5143) time: 0.3809 data: 0.0001 max mem: 52126 +Epoch: [30] [1910/3634] eta: 0:14:17 lr: 0.000003 loss: 0.5235 (0.5142) time: 0.4855 data: 0.0001 max mem: 52126 +Epoch: [30] [1920/3634] eta: 0:14:12 lr: 0.000003 loss: 0.5311 (0.5142) time: 0.5447 data: 0.0001 max mem: 52126 +Epoch: [30] [1930/3634] eta: 0:14:07 lr: 0.000003 loss: 0.4829 (0.5142) time: 0.4994 data: 0.0001 max mem: 52126 +Epoch: [30] [1940/3634] eta: 0:14:01 lr: 0.000003 loss: 0.4810 (0.5141) time: 0.4410 data: 0.0001 max mem: 52126 +Epoch: [30] [1950/3634] eta: 0:13:55 lr: 0.000003 loss: 0.5165 (0.5142) time: 0.3817 data: 0.0002 max mem: 52126 +Epoch: [30] [1960/3634] eta: 0:13:51 lr: 0.000003 loss: 0.5280 (0.5142) time: 0.4880 data: 0.0002 max mem: 52126 +Epoch: [30] [1970/3634] eta: 0:13:46 lr: 0.000003 loss: 0.5280 (0.5141) time: 0.5758 data: 0.0001 max mem: 52126 +Epoch: [30] [1980/3634] eta: 0:13:42 lr: 0.000003 loss: 0.5302 (0.5141) time: 0.5901 data: 0.0001 max mem: 52126 +Epoch: [30] [1990/3634] eta: 0:13:36 lr: 0.000003 loss: 0.5115 (0.5141) time: 0.5028 data: 0.0001 max mem: 52126 +Epoch: [30] [2000/3634] eta: 0:13:33 lr: 0.000003 loss: 0.5210 (0.5141) time: 0.5019 data: 0.0001 max mem: 52126 +Epoch: [30] [2010/3634] eta: 0:13:28 lr: 0.000003 loss: 0.5165 (0.5139) time: 0.6175 data: 0.0002 max mem: 52126 +Epoch: [30] [2020/3634] eta: 0:13:23 lr: 0.000003 loss: 0.5137 (0.5139) time: 0.5477 data: 0.0002 max mem: 52126 +Epoch: [30] [2030/3634] eta: 0:13:18 lr: 0.000003 loss: 0.5137 (0.5137) time: 0.4773 data: 0.0002 max mem: 52126 +Epoch: [30] [2040/3634] eta: 0:13:13 lr: 0.000003 loss: 0.4978 (0.5137) time: 0.4722 data: 0.0002 max mem: 52126 +Epoch: [30] [2050/3634] eta: 0:13:08 lr: 0.000003 loss: 0.5288 (0.5138) time: 0.4717 data: 0.0001 max mem: 52126 +Epoch: [30] [2060/3634] eta: 0:13:03 lr: 0.000003 loss: 0.5288 (0.5137) time: 0.4865 data: 0.0001 max mem: 52126 +Epoch: [30] [2070/3634] eta: 0:12:59 lr: 0.000003 loss: 0.4966 (0.5136) time: 0.5611 data: 0.0002 max mem: 52126 +Epoch: [30] [2080/3634] eta: 0:12:53 lr: 0.000003 loss: 0.4950 (0.5136) time: 0.5011 data: 0.0002 max mem: 52126 +Epoch: [30] [2090/3634] eta: 0:12:48 lr: 0.000003 loss: 0.5279 (0.5136) time: 0.4403 data: 0.0002 max mem: 52126 +Epoch: [30] [2100/3634] eta: 0:12:43 lr: 0.000003 loss: 0.5483 (0.5138) time: 0.4854 data: 0.0001 max mem: 52126 +Epoch: [30] [2110/3634] eta: 0:12:40 lr: 0.000003 loss: 0.5567 (0.5139) time: 0.6489 data: 0.0001 max mem: 52126 +Epoch: [30] [2120/3634] eta: 0:12:35 lr: 0.000003 loss: 0.5392 (0.5139) time: 0.6497 data: 0.0001 max mem: 52126 +Epoch: [30] [2130/3634] eta: 0:12:29 lr: 0.000003 loss: 0.5090 (0.5138) time: 0.4279 data: 0.0001 max mem: 52126 +Epoch: [30] [2140/3634] eta: 0:12:25 lr: 0.000003 loss: 0.5131 (0.5138) time: 0.4920 data: 0.0002 max mem: 52126 +Epoch: [30] [2150/3634] eta: 0:12:19 lr: 0.000003 loss: 0.5073 (0.5137) time: 0.4916 data: 0.0002 max mem: 52126 +Epoch: [30] [2160/3634] eta: 0:12:14 lr: 0.000003 loss: 0.5063 (0.5138) time: 0.4265 data: 0.0001 max mem: 52126 +Epoch: [30] [2170/3634] eta: 0:12:08 lr: 0.000003 loss: 0.5172 (0.5138) time: 0.4269 data: 0.0002 max mem: 52126 +Epoch: [30] [2180/3634] eta: 0:12:03 lr: 0.000003 loss: 0.5070 (0.5138) time: 0.4272 data: 0.0002 max mem: 52126 +Epoch: [30] [2190/3634] eta: 0:11:57 lr: 0.000003 loss: 0.5056 (0.5138) time: 0.4267 data: 0.0002 max mem: 52126 +Epoch: [30] [2200/3634] eta: 0:11:52 lr: 0.000003 loss: 0.4901 (0.5138) time: 0.4405 data: 0.0002 max mem: 52126 +Epoch: [30] [2210/3634] eta: 0:11:47 lr: 0.000003 loss: 0.5057 (0.5137) time: 0.4402 data: 0.0002 max mem: 52126 +Epoch: [30] [2220/3634] eta: 0:11:42 lr: 0.000003 loss: 0.5057 (0.5138) time: 0.4989 data: 0.0002 max mem: 52126 +Epoch: [30] [2230/3634] eta: 0:11:37 lr: 0.000003 loss: 0.5376 (0.5138) time: 0.4995 data: 0.0002 max mem: 52126 +Epoch: [30] [2240/3634] eta: 0:11:31 lr: 0.000003 loss: 0.5236 (0.5136) time: 0.3821 data: 0.0002 max mem: 52126 +Epoch: [30] [2250/3634] eta: 0:11:25 lr: 0.000003 loss: 0.5212 (0.5137) time: 0.3816 data: 0.0002 max mem: 52126 +Epoch: [30] [2260/3634] eta: 0:11:21 lr: 0.000003 loss: 0.5546 (0.5137) time: 0.4831 data: 0.0002 max mem: 52126 +Epoch: [30] [2270/3634] eta: 0:11:16 lr: 0.000003 loss: 0.5546 (0.5139) time: 0.5279 data: 0.0002 max mem: 52126 +Epoch: [30] [2280/3634] eta: 0:11:11 lr: 0.000003 loss: 0.5112 (0.5138) time: 0.5293 data: 0.0002 max mem: 52126 +Epoch: [30] [2290/3634] eta: 0:11:09 lr: 0.000003 loss: 0.5408 (0.5140) time: 0.7380 data: 0.0002 max mem: 52126 +Epoch: [30] [2300/3634] eta: 0:11:04 lr: 0.000003 loss: 0.5381 (0.5139) time: 0.6954 data: 0.0002 max mem: 52126 +Epoch: [30] [2310/3634] eta: 0:10:59 lr: 0.000003 loss: 0.5036 (0.5140) time: 0.4881 data: 0.0002 max mem: 52126 +Epoch: [30] [2320/3634] eta: 0:10:53 lr: 0.000003 loss: 0.5316 (0.5139) time: 0.4283 data: 0.0002 max mem: 52126 +Epoch: [30] [2330/3634] eta: 0:10:48 lr: 0.000003 loss: 0.5087 (0.5139) time: 0.4266 data: 0.0002 max mem: 52126 +Epoch: [30] [2340/3634] eta: 0:10:42 lr: 0.000003 loss: 0.5222 (0.5139) time: 0.4263 data: 0.0002 max mem: 52126 +Epoch: [30] [2350/3634] eta: 0:10:37 lr: 0.000003 loss: 0.5078 (0.5138) time: 0.4426 data: 0.0002 max mem: 52126 +Epoch: [30] [2360/3634] eta: 0:10:32 lr: 0.000003 loss: 0.5063 (0.5137) time: 0.4427 data: 0.0002 max mem: 52126 +Epoch: [30] [2370/3634] eta: 0:10:26 lr: 0.000003 loss: 0.5321 (0.5139) time: 0.3823 data: 0.0002 max mem: 52126 +Epoch: [30] [2380/3634] eta: 0:10:22 lr: 0.000003 loss: 0.5538 (0.5140) time: 0.4768 data: 0.0002 max mem: 52126 +Epoch: [30] [2390/3634] eta: 0:10:17 lr: 0.000003 loss: 0.5418 (0.5140) time: 0.5224 data: 0.0002 max mem: 52126 +Epoch: [30] [2400/3634] eta: 0:10:12 lr: 0.000003 loss: 0.5360 (0.5140) time: 0.4884 data: 0.0001 max mem: 52126 +Epoch: [30] [2410/3634] eta: 0:10:07 lr: 0.000003 loss: 0.5360 (0.5141) time: 0.5487 data: 0.0001 max mem: 52126 +Epoch: [30] [2420/3634] eta: 0:10:03 lr: 0.000003 loss: 0.5576 (0.5143) time: 0.5936 data: 0.0001 max mem: 52126 +Epoch: [30] [2430/3634] eta: 0:09:58 lr: 0.000003 loss: 0.5643 (0.5144) time: 0.5899 data: 0.0002 max mem: 52126 +Epoch: [30] [2440/3634] eta: 0:09:53 lr: 0.000003 loss: 0.5643 (0.5145) time: 0.4842 data: 0.0002 max mem: 52126 +Epoch: [30] [2450/3634] eta: 0:09:48 lr: 0.000003 loss: 0.5287 (0.5145) time: 0.4699 data: 0.0002 max mem: 52126 +Epoch: [30] [2460/3634] eta: 0:09:43 lr: 0.000003 loss: 0.4960 (0.5144) time: 0.4702 data: 0.0002 max mem: 52126 +Epoch: [30] [2470/3634] eta: 0:09:37 lr: 0.000003 loss: 0.4960 (0.5143) time: 0.3826 data: 0.0001 max mem: 52126 +Epoch: [30] [2480/3634] eta: 0:09:32 lr: 0.000003 loss: 0.5258 (0.5143) time: 0.4311 data: 0.0001 max mem: 52126 +Epoch: [30] [2490/3634] eta: 0:09:27 lr: 0.000003 loss: 0.4756 (0.5142) time: 0.5349 data: 0.0001 max mem: 52126 +Epoch: [30] [2500/3634] eta: 0:09:22 lr: 0.000003 loss: 0.4755 (0.5141) time: 0.4862 data: 0.0001 max mem: 52126 +Epoch: [30] [2510/3634] eta: 0:09:18 lr: 0.000003 loss: 0.5126 (0.5141) time: 0.5038 data: 0.0001 max mem: 52126 +Epoch: [30] [2520/3634] eta: 0:09:13 lr: 0.000003 loss: 0.5032 (0.5140) time: 0.6079 data: 0.0001 max mem: 52126 +Epoch: [30] [2530/3634] eta: 0:09:09 lr: 0.000003 loss: 0.5032 (0.5141) time: 0.6680 data: 0.0001 max mem: 52126 +Epoch: [30] [2540/3634] eta: 0:09:04 lr: 0.000003 loss: 0.5400 (0.5142) time: 0.6065 data: 0.0001 max mem: 52126 +Epoch: [30] [2550/3634] eta: 0:08:59 lr: 0.000003 loss: 0.5287 (0.5141) time: 0.4249 data: 0.0002 max mem: 52126 +Epoch: [30] [2560/3634] eta: 0:08:54 lr: 0.000003 loss: 0.4967 (0.5140) time: 0.4428 data: 0.0002 max mem: 52126 +Epoch: [30] [2570/3634] eta: 0:08:48 lr: 0.000003 loss: 0.5049 (0.5140) time: 0.4428 data: 0.0001 max mem: 52126 +Epoch: [30] [2580/3634] eta: 0:08:43 lr: 0.000003 loss: 0.5083 (0.5141) time: 0.4268 data: 0.0001 max mem: 52126 +Epoch: [30] [2590/3634] eta: 0:08:39 lr: 0.000003 loss: 0.4973 (0.5139) time: 0.5449 data: 0.0001 max mem: 52126 +Epoch: [30] [2600/3634] eta: 0:08:34 lr: 0.000003 loss: 0.5250 (0.5140) time: 0.5442 data: 0.0001 max mem: 52126 +Epoch: [30] [2610/3634] eta: 0:08:28 lr: 0.000003 loss: 0.5449 (0.5141) time: 0.4710 data: 0.0001 max mem: 52126 +Epoch: [30] [2620/3634] eta: 0:08:23 lr: 0.000003 loss: 0.5318 (0.5141) time: 0.4270 data: 0.0001 max mem: 52126 +Epoch: [30] [2630/3634] eta: 0:08:18 lr: 0.000003 loss: 0.5027 (0.5140) time: 0.4868 data: 0.0001 max mem: 52126 +Epoch: [30] [2640/3634] eta: 0:08:14 lr: 0.000003 loss: 0.5081 (0.5141) time: 0.6546 data: 0.0001 max mem: 52126 +Epoch: [30] [2650/3634] eta: 0:08:09 lr: 0.000003 loss: 0.5420 (0.5142) time: 0.5953 data: 0.0002 max mem: 52126 +Epoch: [30] [2660/3634] eta: 0:08:04 lr: 0.000003 loss: 0.5101 (0.5140) time: 0.4276 data: 0.0002 max mem: 52126 +Epoch: [30] [2670/3634] eta: 0:07:59 lr: 0.000003 loss: 0.4861 (0.5139) time: 0.4435 data: 0.0002 max mem: 52126 +Epoch: [30] [2680/3634] eta: 0:07:54 lr: 0.000003 loss: 0.5060 (0.5140) time: 0.4956 data: 0.0002 max mem: 52126 +Epoch: [30] [2690/3634] eta: 0:07:49 lr: 0.000003 loss: 0.4891 (0.5138) time: 0.5395 data: 0.0002 max mem: 52126 +Epoch: [30] [2700/3634] eta: 0:07:44 lr: 0.000003 loss: 0.4437 (0.5136) time: 0.5495 data: 0.0002 max mem: 52126 +Epoch: [30] [2710/3634] eta: 0:07:39 lr: 0.000003 loss: 0.4892 (0.5137) time: 0.5033 data: 0.0002 max mem: 52126 +Epoch: [30] [2720/3634] eta: 0:07:34 lr: 0.000003 loss: 0.5229 (0.5137) time: 0.5000 data: 0.0002 max mem: 52126 +Epoch: [30] [2730/3634] eta: 0:07:30 lr: 0.000003 loss: 0.5129 (0.5137) time: 0.5452 data: 0.0002 max mem: 52126 +Epoch: [30] [2740/3634] eta: 0:07:25 lr: 0.000003 loss: 0.5186 (0.5138) time: 0.5463 data: 0.0002 max mem: 52126 +Epoch: [30] [2750/3634] eta: 0:07:19 lr: 0.000003 loss: 0.5134 (0.5137) time: 0.4423 data: 0.0002 max mem: 52126 +Epoch: [30] [2760/3634] eta: 0:07:14 lr: 0.000003 loss: 0.5013 (0.5137) time: 0.3822 data: 0.0002 max mem: 52126 +Epoch: [30] [2770/3634] eta: 0:07:09 lr: 0.000003 loss: 0.5014 (0.5136) time: 0.3826 data: 0.0002 max mem: 52126 +Epoch: [30] [2780/3634] eta: 0:07:03 lr: 0.000003 loss: 0.5109 (0.5136) time: 0.3825 data: 0.0002 max mem: 52126 +Epoch: [30] [2790/3634] eta: 0:06:59 lr: 0.000003 loss: 0.5109 (0.5137) time: 0.4853 data: 0.0002 max mem: 52126 +Epoch: [30] [2800/3634] eta: 0:06:54 lr: 0.000003 loss: 0.5362 (0.5136) time: 0.5305 data: 0.0002 max mem: 52126 +Epoch: [30] [2810/3634] eta: 0:06:48 lr: 0.000003 loss: 0.5083 (0.5136) time: 0.4276 data: 0.0002 max mem: 52126 +Epoch: [30] [2820/3634] eta: 0:06:43 lr: 0.000003 loss: 0.5018 (0.5135) time: 0.4285 data: 0.0002 max mem: 52126 +Epoch: [30] [2830/3634] eta: 0:06:38 lr: 0.000003 loss: 0.5172 (0.5137) time: 0.4281 data: 0.0002 max mem: 52126 +Epoch: [30] [2840/3634] eta: 0:06:33 lr: 0.000003 loss: 0.5159 (0.5136) time: 0.4854 data: 0.0002 max mem: 52126 +Epoch: [30] [2850/3634] eta: 0:06:28 lr: 0.000003 loss: 0.5033 (0.5135) time: 0.5459 data: 0.0002 max mem: 52126 +Epoch: [30] [2860/3634] eta: 0:06:23 lr: 0.000003 loss: 0.5033 (0.5135) time: 0.4924 data: 0.0002 max mem: 52126 +Epoch: [30] [2870/3634] eta: 0:06:19 lr: 0.000003 loss: 0.5222 (0.5135) time: 0.5213 data: 0.0002 max mem: 52126 +Epoch: [30] [2880/3634] eta: 0:06:14 lr: 0.000003 loss: 0.5138 (0.5135) time: 0.5842 data: 0.0002 max mem: 52126 +Epoch: [30] [2890/3634] eta: 0:06:09 lr: 0.000003 loss: 0.5163 (0.5136) time: 0.4952 data: 0.0002 max mem: 52126 +Epoch: [30] [2900/3634] eta: 0:06:03 lr: 0.000003 loss: 0.5288 (0.5136) time: 0.3850 data: 0.0002 max mem: 52126 +Epoch: [30] [2910/3634] eta: 0:05:59 lr: 0.000003 loss: 0.5129 (0.5134) time: 0.4458 data: 0.0002 max mem: 52126 +Epoch: [30] [2920/3634] eta: 0:05:54 lr: 0.000003 loss: 0.5189 (0.5135) time: 0.4881 data: 0.0002 max mem: 52126 +Epoch: [30] [2930/3634] eta: 0:05:49 lr: 0.000003 loss: 0.5335 (0.5135) time: 0.4866 data: 0.0002 max mem: 52126 +Epoch: [30] [2940/3634] eta: 0:05:44 lr: 0.000003 loss: 0.5335 (0.5135) time: 0.5910 data: 0.0002 max mem: 52126 +Epoch: [30] [2950/3634] eta: 0:05:39 lr: 0.000003 loss: 0.5048 (0.5134) time: 0.5751 data: 0.0002 max mem: 52126 +Epoch: [30] [2960/3634] eta: 0:05:34 lr: 0.000003 loss: 0.4882 (0.5134) time: 0.4708 data: 0.0002 max mem: 52126 +Epoch: [30] [2970/3634] eta: 0:05:29 lr: 0.000003 loss: 0.5010 (0.5134) time: 0.4869 data: 0.0002 max mem: 52126 +Epoch: [30] [2980/3634] eta: 0:05:24 lr: 0.000003 loss: 0.5342 (0.5135) time: 0.4868 data: 0.0002 max mem: 52126 +Epoch: [30] [2990/3634] eta: 0:05:19 lr: 0.000003 loss: 0.5331 (0.5135) time: 0.4267 data: 0.0002 max mem: 52126 +Epoch: [30] [3000/3634] eta: 0:05:14 lr: 0.000003 loss: 0.5001 (0.5134) time: 0.4853 data: 0.0002 max mem: 52126 +Epoch: [30] [3010/3634] eta: 0:05:09 lr: 0.000003 loss: 0.5169 (0.5135) time: 0.5338 data: 0.0002 max mem: 52126 +Epoch: [30] [3020/3634] eta: 0:05:04 lr: 0.000003 loss: 0.5218 (0.5134) time: 0.5369 data: 0.0002 max mem: 52126 +Epoch: [30] [3030/3634] eta: 0:04:59 lr: 0.000003 loss: 0.5007 (0.5134) time: 0.4886 data: 0.0002 max mem: 52126 +Epoch: [30] [3040/3634] eta: 0:04:54 lr: 0.000003 loss: 0.4972 (0.5133) time: 0.4268 data: 0.0002 max mem: 52126 +Epoch: [30] [3050/3634] eta: 0:04:49 lr: 0.000003 loss: 0.4835 (0.5133) time: 0.4274 data: 0.0002 max mem: 52126 +Epoch: [30] [3060/3634] eta: 0:04:44 lr: 0.000003 loss: 0.5000 (0.5133) time: 0.4944 data: 0.0002 max mem: 52126 +Epoch: [30] [3070/3634] eta: 0:04:39 lr: 0.000003 loss: 0.5235 (0.5133) time: 0.4933 data: 0.0002 max mem: 52126 +Epoch: [30] [3080/3634] eta: 0:04:34 lr: 0.000003 loss: 0.5386 (0.5133) time: 0.4284 data: 0.0002 max mem: 52126 +Epoch: [30] [3090/3634] eta: 0:04:29 lr: 0.000003 loss: 0.5091 (0.5132) time: 0.5496 data: 0.0002 max mem: 52126 +Epoch: [30] [3100/3634] eta: 0:04:24 lr: 0.000003 loss: 0.5091 (0.5133) time: 0.5647 data: 0.0002 max mem: 52126 +Epoch: [30] [3110/3634] eta: 0:04:19 lr: 0.000003 loss: 0.5272 (0.5133) time: 0.5030 data: 0.0002 max mem: 52126 +Epoch: [30] [3120/3634] eta: 0:04:15 lr: 0.000003 loss: 0.5125 (0.5134) time: 0.5652 data: 0.0002 max mem: 52126 +Epoch: [30] [3130/3634] eta: 0:04:10 lr: 0.000003 loss: 0.5237 (0.5134) time: 0.6077 data: 0.0002 max mem: 52126 +Epoch: [30] [3140/3634] eta: 0:04:05 lr: 0.000003 loss: 0.5237 (0.5134) time: 0.4848 data: 0.0002 max mem: 52126 +Epoch: [30] [3150/3634] eta: 0:04:00 lr: 0.000003 loss: 0.5308 (0.5135) time: 0.5026 data: 0.0002 max mem: 52126 +Epoch: [30] [3160/3634] eta: 0:03:55 lr: 0.000003 loss: 0.5308 (0.5136) time: 0.5460 data: 0.0002 max mem: 52126 +Epoch: [30] [3170/3634] eta: 0:03:50 lr: 0.000003 loss: 0.5259 (0.5136) time: 0.5290 data: 0.0002 max mem: 52126 +Epoch: [30] [3180/3634] eta: 0:03:45 lr: 0.000003 loss: 0.5316 (0.5137) time: 0.4848 data: 0.0002 max mem: 52126 +Epoch: [30] [3190/3634] eta: 0:03:40 lr: 0.000003 loss: 0.5349 (0.5137) time: 0.3826 data: 0.0002 max mem: 52126 +Epoch: [30] [3200/3634] eta: 0:03:35 lr: 0.000003 loss: 0.5410 (0.5138) time: 0.5035 data: 0.0002 max mem: 52126 +Epoch: [30] [3210/3634] eta: 0:03:30 lr: 0.000003 loss: 0.5156 (0.5137) time: 0.6123 data: 0.0002 max mem: 52126 +Epoch: [30] [3220/3634] eta: 0:03:25 lr: 0.000003 loss: 0.5156 (0.5138) time: 0.5785 data: 0.0002 max mem: 52126 +Epoch: [30] [3230/3634] eta: 0:03:21 lr: 0.000003 loss: 0.5012 (0.5137) time: 0.6319 data: 0.0002 max mem: 52126 +Epoch: [30] [3240/3634] eta: 0:03:16 lr: 0.000003 loss: 0.4973 (0.5137) time: 0.6676 data: 0.0002 max mem: 52126 +Epoch: [30] [3250/3634] eta: 0:03:11 lr: 0.000003 loss: 0.4909 (0.5136) time: 0.6261 data: 0.0002 max mem: 52126 +Epoch: [30] [3260/3634] eta: 0:03:06 lr: 0.000003 loss: 0.5208 (0.5137) time: 0.6097 data: 0.0002 max mem: 52126 +Epoch: [30] [3270/3634] eta: 0:03:01 lr: 0.000003 loss: 0.5272 (0.5137) time: 0.5477 data: 0.0002 max mem: 52126 +Epoch: [30] [3280/3634] eta: 0:02:56 lr: 0.000003 loss: 0.5369 (0.5137) time: 0.5007 data: 0.0002 max mem: 52126 +Epoch: [30] [3290/3634] eta: 0:02:51 lr: 0.000003 loss: 0.5369 (0.5138) time: 0.4418 data: 0.0002 max mem: 52126 +Epoch: [30] [3300/3634] eta: 0:02:46 lr: 0.000003 loss: 0.5200 (0.5138) time: 0.3826 data: 0.0002 max mem: 52126 +Epoch: [30] [3310/3634] eta: 0:02:41 lr: 0.000003 loss: 0.5211 (0.5138) time: 0.3823 data: 0.0002 max mem: 52126 +Epoch: [30] [3320/3634] eta: 0:02:36 lr: 0.000003 loss: 0.5211 (0.5138) time: 0.4416 data: 0.0002 max mem: 52126 +Epoch: [30] [3330/3634] eta: 0:02:31 lr: 0.000003 loss: 0.5003 (0.5137) time: 0.4882 data: 0.0002 max mem: 52126 +Epoch: [30] [3340/3634] eta: 0:02:26 lr: 0.000003 loss: 0.4884 (0.5137) time: 0.4288 data: 0.0002 max mem: 52126 +Epoch: [30] [3350/3634] eta: 0:02:21 lr: 0.000003 loss: 0.5305 (0.5137) time: 0.3821 data: 0.0002 max mem: 52126 +Epoch: [30] [3360/3634] eta: 0:02:16 lr: 0.000003 loss: 0.5285 (0.5137) time: 0.3817 data: 0.0002 max mem: 52126 +Epoch: [30] [3370/3634] eta: 0:02:11 lr: 0.000003 loss: 0.5285 (0.5137) time: 0.4307 data: 0.0002 max mem: 52126 +Epoch: [30] [3380/3634] eta: 0:02:06 lr: 0.000003 loss: 0.5492 (0.5137) time: 0.4920 data: 0.0002 max mem: 52126 +Epoch: [30] [3390/3634] eta: 0:02:01 lr: 0.000003 loss: 0.5329 (0.5137) time: 0.5030 data: 0.0002 max mem: 52126 +Epoch: [30] [3400/3634] eta: 0:01:56 lr: 0.000003 loss: 0.5070 (0.5137) time: 0.5640 data: 0.0002 max mem: 52126 +Epoch: [30] [3410/3634] eta: 0:01:51 lr: 0.000003 loss: 0.5070 (0.5136) time: 0.5478 data: 0.0002 max mem: 52126 +Epoch: [30] [3420/3634] eta: 0:01:46 lr: 0.000003 loss: 0.5236 (0.5137) time: 0.4257 data: 0.0002 max mem: 52126 +Epoch: [30] [3430/3634] eta: 0:01:41 lr: 0.000003 loss: 0.5493 (0.5138) time: 0.4260 data: 0.0002 max mem: 52126 +Epoch: [30] [3440/3634] eta: 0:01:36 lr: 0.000003 loss: 0.5624 (0.5139) time: 0.4699 data: 0.0002 max mem: 52126 +Epoch: [30] [3450/3634] eta: 0:01:31 lr: 0.000003 loss: 0.5238 (0.5138) time: 0.4258 data: 0.0002 max mem: 52126 +Epoch: [30] [3460/3634] eta: 0:01:26 lr: 0.000003 loss: 0.4518 (0.5137) time: 0.4693 data: 0.0002 max mem: 52126 +Epoch: [30] [3470/3634] eta: 0:01:21 lr: 0.000003 loss: 0.4944 (0.5137) time: 0.5768 data: 0.0002 max mem: 52126 +Epoch: [30] [3480/3634] eta: 0:01:16 lr: 0.000003 loss: 0.5217 (0.5137) time: 0.5786 data: 0.0002 max mem: 52126 +Epoch: [30] [3490/3634] eta: 0:01:11 lr: 0.000003 loss: 0.4974 (0.5136) time: 0.5780 data: 0.0002 max mem: 52126 +Epoch: [30] [3500/3634] eta: 0:01:06 lr: 0.000003 loss: 0.4824 (0.5135) time: 0.6553 data: 0.0002 max mem: 52126 +Epoch: [30] [3510/3634] eta: 0:01:01 lr: 0.000003 loss: 0.4681 (0.5135) time: 0.5940 data: 0.0002 max mem: 52126 +Epoch: [30] [3520/3634] eta: 0:00:56 lr: 0.000003 loss: 0.4890 (0.5135) time: 0.4902 data: 0.0002 max mem: 52126 +Epoch: [30] [3530/3634] eta: 0:00:51 lr: 0.000003 loss: 0.5117 (0.5135) time: 0.5632 data: 0.0002 max mem: 52126 +Epoch: [30] [3540/3634] eta: 0:00:46 lr: 0.000003 loss: 0.5311 (0.5136) time: 0.5012 data: 0.0002 max mem: 52126 +Epoch: [30] [3550/3634] eta: 0:00:41 lr: 0.000003 loss: 0.5484 (0.5136) time: 0.4424 data: 0.0002 max mem: 52126 +Epoch: [30] [3560/3634] eta: 0:00:36 lr: 0.000003 loss: 0.5261 (0.5135) time: 0.4416 data: 0.0002 max mem: 52126 +Epoch: [30] [3570/3634] eta: 0:00:31 lr: 0.000003 loss: 0.5173 (0.5135) time: 0.3821 data: 0.0002 max mem: 52126 +Epoch: [30] [3580/3634] eta: 0:00:26 lr: 0.000003 loss: 0.5173 (0.5135) time: 0.4873 data: 0.0002 max mem: 52126 +Epoch: [30] [3590/3634] eta: 0:00:21 lr: 0.000003 loss: 0.5197 (0.5135) time: 0.4873 data: 0.0002 max mem: 52126 +Epoch: [30] [3600/3634] eta: 0:00:16 lr: 0.000003 loss: 0.5174 (0.5136) time: 0.4410 data: 0.0002 max mem: 52126 +Epoch: [30] [3610/3634] eta: 0:00:11 lr: 0.000003 loss: 0.4994 (0.5136) time: 0.5043 data: 0.0002 max mem: 52126 +Epoch: [30] [3620/3634] eta: 0:00:06 lr: 0.000003 loss: 0.5212 (0.5136) time: 0.4488 data: 0.0033 max mem: 52126 +Epoch: [30] [3630/3634] eta: 0:00:01 lr: 0.000003 loss: 0.5212 (0.5136) time: 0.4893 data: 0.0033 max mem: 52126 +Epoch: [30] [3633/3634] eta: 0:00:00 lr: 0.000003 loss: 0.5112 (0.5136) time: 0.4888 data: 0.0033 max mem: 52126 +Epoch: [30] Total time: 0:30:07 (0.4975 s / it) +Averaged stats: lr: 0.000003 loss: 0.5112 (0.5147) +Test: [ 0/1403] eta: 1:08:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0882 (0.0882) acc1: 100.0000 (100.0000) acc5: 100.0000 (100.0000) time: 2.9347 data: 2.8382 max mem: 52126 +Test: [ 10/1403] eta: 0:08:13 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1662 (0.1806) acc1: 98.4375 (95.1705) acc5: 100.0000 (100.0000) time: 0.3541 data: 0.2583 max mem: 52126 +Test: [ 20/1403] eta: 0:05:19 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1295 (0.1622) acc1: 98.4375 (95.6845) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 30/1403] eta: 0:04:17 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0885 (0.1474) acc1: 100.0000 (96.2702) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 40/1403] eta: 0:03:45 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0977 (0.1480) acc1: 100.0000 (96.5320) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0004 max mem: 52126 +Test: [ 50/1403] eta: 0:03:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1122 (0.1606) acc1: 98.4375 (95.9252) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 60/1403] eta: 0:03:11 acc1_pos: 89.062 acc5_pos: 100.000 loss: 0.1198 (0.1607) acc1: 100.0000 (95.9529) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 70/1403] eta: 0:03:01 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.1282 (0.1666) acc1: 98.4375 (95.5546) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 80/1403] eta: 0:02:53 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1579 (0.1741) acc1: 95.3125 (95.1968) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 90/1403] eta: 0:02:47 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.1760 (0.1920) acc1: 95.3125 (94.3338) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 100/1403] eta: 0:02:41 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1754 (0.1903) acc1: 95.3125 (94.4926) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 110/1403] eta: 0:02:37 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1447 (0.1886) acc1: 98.4375 (94.5946) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 120/1403] eta: 0:02:33 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0817 (0.1794) acc1: 100.0000 (95.0284) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 130/1403] eta: 0:02:29 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0807 (0.1731) acc1: 100.0000 (95.3125) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 140/1403] eta: 0:02:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0750 (0.1660) acc1: 100.0000 (95.6449) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 150/1403] eta: 0:02:23 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0730 (0.1602) acc1: 100.0000 (95.9127) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 160/1403] eta: 0:02:21 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0774 (0.1554) acc1: 100.0000 (96.1180) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 170/1403] eta: 0:02:18 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0774 (0.1515) acc1: 100.0000 (96.3176) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 180/1403] eta: 0:02:16 acc1_pos: 92.188 acc5_pos: 100.000 loss: 0.0838 (0.1545) acc1: 100.0000 (96.2535) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 190/1403] eta: 0:02:14 acc1_pos: 90.625 acc5_pos: 100.000 loss: 0.2223 (0.1610) acc1: 93.7500 (95.9179) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 200/1403] eta: 0:02:12 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.2421 (0.1710) acc1: 90.6250 (95.4213) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 210/1403] eta: 0:02:10 acc1_pos: 76.562 acc5_pos: 100.000 loss: 0.2596 (0.1800) acc1: 87.5000 (94.9422) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52126 +Test: [ 220/1403] eta: 0:02:08 acc1_pos: 92.188 acc5_pos: 100.000 loss: 0.3192 (0.1873) acc1: 87.5000 (94.5984) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52126 +Test: [ 230/1403] eta: 0:02:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2286 (0.1834) acc1: 92.1875 (94.7849) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 240/1403] eta: 0:02:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0765 (0.1792) acc1: 100.0000 (94.9883) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 250/1403] eta: 0:02:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0765 (0.1755) acc1: 100.0000 (95.1693) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 260/1403] eta: 0:02:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0804 (0.1721) acc1: 100.0000 (95.3364) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 270/1403] eta: 0:02:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0814 (0.1690) acc1: 100.0000 (95.4912) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 280/1403] eta: 0:01:59 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0788 (0.1658) acc1: 100.0000 (95.6461) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 290/1403] eta: 0:01:57 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0755 (0.1628) acc1: 100.0000 (95.7904) acc5: 100.0000 (100.0000) time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 300/1403] eta: 0:01:56 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0741 (0.1600) acc1: 100.0000 (95.9250) acc5: 100.0000 (100.0000) time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 310/1403] eta: 0:01:54 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0800 (0.1575) acc1: 100.0000 (96.0561) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 320/1403] eta: 0:01:53 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0944 (0.1557) acc1: 100.0000 (96.1789) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 330/1403] eta: 0:01:52 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1012 (0.1542) acc1: 100.0000 (96.2944) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 340/1403] eta: 0:01:50 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1012 (0.1525) acc1: 100.0000 (96.4030) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 350/1403] eta: 0:01:49 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0992 (0.1511) acc1: 100.0000 (96.5055) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 360/1403] eta: 0:01:48 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1018 (0.1498) acc1: 100.0000 (96.5980) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 370/1403] eta: 0:01:47 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1007 (0.1484) acc1: 100.0000 (96.6897) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 380/1403] eta: 0:01:45 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0996 (0.1472) acc1: 100.0000 (96.7766) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 390/1403] eta: 0:01:44 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1008 (0.1478) acc1: 100.0000 (96.7311) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 400/1403] eta: 0:01:43 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2313 (0.1510) acc1: 90.6250 (96.5555) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 410/1403] eta: 0:01:42 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2533 (0.1535) acc1: 90.6250 (96.4112) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 420/1403] eta: 0:01:41 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2234 (0.1547) acc1: 92.1875 (96.3257) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 430/1403] eta: 0:01:39 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2168 (0.1570) acc1: 92.1875 (96.2043) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 440/1403] eta: 0:01:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2766 (0.1599) acc1: 89.0625 (96.0424) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 450/1403] eta: 0:01:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2511 (0.1618) acc1: 89.0625 (95.9396) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 460/1403] eta: 0:01:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2166 (0.1638) acc1: 90.6250 (95.8073) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 470/1403] eta: 0:01:35 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1932 (0.1629) acc1: 92.1875 (95.8300) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 480/1403] eta: 0:01:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0631 (0.1608) acc1: 100.0000 (95.9167) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 490/1403] eta: 0:01:33 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0630 (0.1588) acc1: 100.0000 (95.9999) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 500/1403] eta: 0:01:31 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0629 (0.1569) acc1: 100.0000 (96.0797) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 510/1403] eta: 0:01:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0630 (0.1551) acc1: 100.0000 (96.1564) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 520/1403] eta: 0:01:29 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0666 (0.1535) acc1: 100.0000 (96.2302) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 530/1403] eta: 0:01:28 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0599 (0.1517) acc1: 100.0000 (96.3012) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 540/1403] eta: 0:01:27 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0593 (0.1500) acc1: 100.0000 (96.3696) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 550/1403] eta: 0:01:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0601 (0.1488) acc1: 100.0000 (96.4213) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 560/1403] eta: 0:01:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0819 (0.1478) acc1: 98.4375 (96.4600) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 570/1403] eta: 0:01:24 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0847 (0.1469) acc1: 98.4375 (96.5001) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 580/1403] eta: 0:01:23 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0948 (0.1465) acc1: 98.4375 (96.5093) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 590/1403] eta: 0:01:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1144 (0.1463) acc1: 96.8750 (96.5022) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 600/1403] eta: 0:01:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1285 (0.1466) acc1: 95.3125 (96.4486) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 610/1403] eta: 0:01:19 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1285 (0.1462) acc1: 95.3125 (96.4505) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 620/1403] eta: 0:01:18 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1362 (0.1471) acc1: 95.3125 (96.3617) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 630/1403] eta: 0:01:17 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1459 (0.1470) acc1: 95.3125 (96.3575) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 640/1403] eta: 0:01:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1237 (0.1466) acc1: 96.8750 (96.3753) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 650/1403] eta: 0:01:15 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1188 (0.1464) acc1: 96.8750 (96.3830) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 660/1403] eta: 0:01:14 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1025 (0.1454) acc1: 98.4375 (96.4259) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 670/1403] eta: 0:01:13 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0684 (0.1443) acc1: 100.0000 (96.4791) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 680/1403] eta: 0:01:12 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0790 (0.1459) acc1: 100.0000 (96.3771) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 690/1403] eta: 0:01:11 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2850 (0.1505) acc1: 89.0625 (96.1243) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 700/1403] eta: 0:01:10 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1887 (0.1507) acc1: 90.6250 (96.1060) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 710/1403] eta: 0:01:09 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2650 (0.1548) acc1: 89.0625 (95.8817) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 720/1403] eta: 0:01:08 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2650 (0.1548) acc1: 84.3750 (95.8781) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 730/1403] eta: 0:01:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0666 (0.1537) acc1: 100.0000 (95.9238) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 740/1403] eta: 0:01:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0790 (0.1529) acc1: 100.0000 (95.9577) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 750/1403] eta: 0:01:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1016 (0.1544) acc1: 98.4375 (95.8805) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 760/1403] eta: 0:01:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3219 (0.1582) acc1: 81.2500 (95.6328) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 770/1403] eta: 0:01:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4717 (0.1629) acc1: 75.0000 (95.3409) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 780/1403] eta: 0:01:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4484 (0.1656) acc1: 78.1250 (95.1905) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 790/1403] eta: 0:01:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3124 (0.1682) acc1: 84.3750 (95.0577) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [ 800/1403] eta: 0:01:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2318 (0.1686) acc1: 89.0625 (95.0336) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 810/1403] eta: 0:00:59 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2203 (0.1717) acc1: 89.0625 (94.8848) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 820/1403] eta: 0:00:58 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6467 (0.1807) acc1: 71.8750 (94.4979) acc5: 100.0000 (100.0000) acc1_neg: 43.750 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 830/1403] eta: 0:00:57 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.7260 (0.1837) acc1: 68.7500 (94.3442) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 840/1403] eta: 0:00:56 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5365 (0.1919) acc1: 70.3125 (93.9878) acc5: 100.0000 (100.0000) acc1_neg: 48.438 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 850/1403] eta: 0:00:55 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.0477 (0.2028) acc1: 56.2500 (93.5278) acc5: 100.0000 (100.0000) acc1_neg: 54.688 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 860/1403] eta: 0:00:54 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.9783 (0.2063) acc1: 57.8125 (93.3798) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 870/1403] eta: 0:00:53 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1313 (0.2052) acc1: 95.3125 (93.4235) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52126 +Test: [ 880/1403] eta: 0:00:51 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1002 (0.2040) acc1: 98.4375 (93.4769) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52126 +Test: [ 890/1403] eta: 0:00:50 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0781 (0.2026) acc1: 98.4375 (93.5413) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 900/1403] eta: 0:00:49 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0723 (0.2012) acc1: 100.0000 (93.6095) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 910/1403] eta: 0:00:48 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0734 (0.2002) acc1: 100.0000 (93.6522) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 920/1403] eta: 0:00:47 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0950 (0.1992) acc1: 98.4375 (93.6991) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 930/1403] eta: 0:00:46 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0927 (0.1981) acc1: 98.4375 (93.7517) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 940/1403] eta: 0:00:45 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1096 (0.1982) acc1: 96.8750 (93.7317) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 950/1403] eta: 0:00:44 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2377 (0.1991) acc1: 89.0625 (93.6744) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 960/1403] eta: 0:00:43 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2840 (0.1997) acc1: 89.0625 (93.6281) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 970/1403] eta: 0:00:42 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2314 (0.2005) acc1: 90.6250 (93.5762) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 980/1403] eta: 0:00:41 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2527 (0.2013) acc1: 90.6250 (93.5206) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [ 990/1403] eta: 0:00:40 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2489 (0.2017) acc1: 90.6250 (93.4946) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [1000/1403] eta: 0:00:39 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2341 (0.2024) acc1: 90.6250 (93.4581) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [1010/1403] eta: 0:00:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2671 (0.2032) acc1: 90.6250 (93.4100) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [1020/1403] eta: 0:00:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4101 (0.2095) acc1: 81.2500 (93.0720) acc5: 100.0000 (100.0000) acc1_neg: 32.812 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [1030/1403] eta: 0:00:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6052 (0.2114) acc1: 64.0625 (92.9513) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [1040/1403] eta: 0:00:35 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2109 (0.2111) acc1: 92.1875 (92.9545) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [1050/1403] eta: 0:00:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1661 (0.2108) acc1: 93.7500 (92.9740) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [1060/1403] eta: 0:00:33 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1266 (0.2100) acc1: 96.8750 (93.0092) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [1070/1403] eta: 0:00:32 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1482 (0.2104) acc1: 96.8750 (92.9928) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [1080/1403] eta: 0:00:31 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3229 (0.2119) acc1: 87.5000 (92.9073) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [1090/1403] eta: 0:00:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3672 (0.2134) acc1: 84.3750 (92.8320) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [1100/1403] eta: 0:00:29 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3414 (0.2139) acc1: 85.9375 (92.7992) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [1110/1403] eta: 0:00:28 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2774 (0.2144) acc1: 90.6250 (92.7726) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [1120/1403] eta: 0:00:27 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2774 (0.2150) acc1: 90.6250 (92.7436) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [1130/1403] eta: 0:00:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2866 (0.2157) acc1: 89.0625 (92.7014) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52126 +Test: [1140/1403] eta: 0:00:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2478 (0.2159) acc1: 90.6250 (92.6887) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52126 +Test: [1150/1403] eta: 0:00:24 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2460 (0.2164) acc1: 90.6250 (92.6586) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52126 +Test: [1160/1403] eta: 0:00:23 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2587 (0.2169) acc1: 89.0625 (92.6343) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [1170/1403] eta: 0:00:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2531 (0.2168) acc1: 90.6250 (92.6358) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [1180/1403] eta: 0:00:21 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0687 (0.2155) acc1: 100.0000 (92.6982) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [1190/1403] eta: 0:00:21 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0685 (0.2143) acc1: 100.0000 (92.7595) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [1200/1403] eta: 0:00:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0686 (0.2131) acc1: 100.0000 (92.8198) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0002 max mem: 52126 +Test: [1210/1403] eta: 0:00:19 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0676 (0.2119) acc1: 100.0000 (92.8791) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0002 max mem: 52126 +Test: [1220/1403] eta: 0:00:18 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0675 (0.2107) acc1: 100.0000 (92.9374) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0002 max mem: 52126 +Test: [1230/1403] eta: 0:00:17 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0683 (0.2096) acc1: 100.0000 (92.9948) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0002 max mem: 52126 +Test: [1240/1403] eta: 0:00:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0682 (0.2084) acc1: 100.0000 (93.0512) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0002 max mem: 52126 +Test: [1250/1403] eta: 0:00:15 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0682 (0.2073) acc1: 100.0000 (93.1068) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0002 max mem: 52126 +Test: [1260/1403] eta: 0:00:14 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0727 (0.2063) acc1: 100.0000 (93.1614) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0960 data: 0.0002 max mem: 52126 +Test: [1270/1403] eta: 0:00:13 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0741 (0.2053) acc1: 100.0000 (93.2152) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0002 max mem: 52126 +Test: [1280/1403] eta: 0:00:12 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0737 (0.2042) acc1: 100.0000 (93.2682) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0002 max mem: 52126 +Test: [1290/1403] eta: 0:00:11 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0730 (0.2032) acc1: 100.0000 (93.3203) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0003 max mem: 52126 +Test: [1300/1403] eta: 0:00:10 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0740 (0.2022) acc1: 100.0000 (93.3717) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0003 max mem: 52126 +Test: [1310/1403] eta: 0:00:09 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0743 (0.2012) acc1: 100.0000 (93.4222) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0002 max mem: 52126 +Test: [1320/1403] eta: 0:00:08 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0725 (0.2003) acc1: 100.0000 (93.4720) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0002 max mem: 52126 +Test: [1330/1403] eta: 0:00:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0721 (0.1993) acc1: 100.0000 (93.5187) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0002 max mem: 52126 +Test: [1340/1403] eta: 0:00:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0676 (0.1983) acc1: 100.0000 (93.5647) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0002 max mem: 52126 +Test: [1350/1403] eta: 0:00:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0662 (0.1974) acc1: 100.0000 (93.6089) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0003 max mem: 52126 +Test: [1360/1403] eta: 0:00:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0664 (0.1964) acc1: 100.0000 (93.6547) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0002 max mem: 52126 +Test: [1370/1403] eta: 0:00:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0664 (0.1955) acc1: 100.0000 (93.6987) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0002 max mem: 52126 +Test: [1380/1403] eta: 0:00:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0651 (0.1946) acc1: 100.0000 (93.7443) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0960 data: 0.0002 max mem: 52126 +Test: [1390/1403] eta: 0:00:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0648 (0.1937) acc1: 100.0000 (93.7893) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0960 data: 0.0002 max mem: 52126 +Test: [1400/1403] eta: 0:00:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0663 (0.1928) acc1: 100.0000 (93.8314) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0959 data: 0.0001 max mem: 52126 +Test: [1402/1403] eta: 0:00:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0666 (0.1926) acc1: 100.0000 (93.8359) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0919 data: 0.0001 max mem: 52126 +Test: Total time: 0:02:18 (0.0986 s / it) +* Overall Acc@1 93.836 Acc@5 100.000 + target == real Acc@1 96.021 Acc@5 100.000 + target == fake Acc@1 93.220 Acc@5 100.000 + loss 0.193 +Accuracy of the network on the 89729 test images: 93.8% +Max accuracy: 94.13% +Epoch: [31] [ 0/3634] eta: 6:29:49 lr: 0.000002 loss: 0.4393 (0.4393) time: 6.4362 data: 4.5717 max mem: 52126 +Epoch: [31] [ 10/3634] eta: 1:03:08 lr: 0.000002 loss: 0.5301 (0.5352) time: 1.0454 data: 0.4157 max mem: 52126 +Epoch: [31] [ 20/3634] eta: 0:53:52 lr: 0.000002 loss: 0.5294 (0.5156) time: 0.6174 data: 0.0001 max mem: 52126 +Epoch: [31] [ 30/3634] eta: 0:45:41 lr: 0.000002 loss: 0.5260 (0.5174) time: 0.6041 data: 0.0002 max mem: 52126 +Epoch: [31] [ 40/3634] eta: 0:43:09 lr: 0.000002 loss: 0.5341 (0.5173) time: 0.5379 data: 0.0002 max mem: 52126 +Epoch: [31] [ 50/3634] eta: 0:39:03 lr: 0.000002 loss: 0.5300 (0.5154) time: 0.4881 data: 0.0002 max mem: 52126 +Epoch: [31] [ 60/3634] eta: 0:38:23 lr: 0.000002 loss: 0.5030 (0.5048) time: 0.4886 data: 0.0002 max mem: 52126 +Epoch: [31] [ 70/3634] eta: 0:37:07 lr: 0.000002 loss: 0.5168 (0.5113) time: 0.5517 data: 0.0002 max mem: 52126 +Epoch: [31] [ 80/3634] eta: 0:37:30 lr: 0.000002 loss: 0.5697 (0.5169) time: 0.5985 data: 0.0002 max mem: 52126 +Epoch: [31] [ 90/3634] eta: 0:36:24 lr: 0.000002 loss: 0.5419 (0.5166) time: 0.5852 data: 0.0002 max mem: 52126 +Epoch: [31] [ 100/3634] eta: 0:36:21 lr: 0.000002 loss: 0.4960 (0.5139) time: 0.5525 data: 0.0002 max mem: 52126 +Epoch: [31] [ 110/3634] eta: 0:36:03 lr: 0.000002 loss: 0.4790 (0.5112) time: 0.6026 data: 0.0002 max mem: 52126 +Epoch: [31] [ 120/3634] eta: 0:35:54 lr: 0.000002 loss: 0.4818 (0.5111) time: 0.5920 data: 0.0002 max mem: 52126 +Epoch: [31] [ 130/3634] eta: 0:36:16 lr: 0.000002 loss: 0.5072 (0.5114) time: 0.6619 data: 0.0002 max mem: 52126 +Epoch: [31] [ 140/3634] eta: 0:35:10 lr: 0.000002 loss: 0.5283 (0.5105) time: 0.5507 data: 0.0002 max mem: 52126 +Epoch: [31] [ 150/3634] eta: 0:34:56 lr: 0.000002 loss: 0.5283 (0.5122) time: 0.4749 data: 0.0002 max mem: 52126 +Epoch: [31] [ 160/3634] eta: 0:34:29 lr: 0.000002 loss: 0.5126 (0.5117) time: 0.5365 data: 0.0001 max mem: 52126 +Epoch: [31] [ 170/3634] eta: 0:33:40 lr: 0.000002 loss: 0.5126 (0.5112) time: 0.4439 data: 0.0002 max mem: 52126 +Epoch: [31] [ 180/3634] eta: 0:32:55 lr: 0.000002 loss: 0.5159 (0.5118) time: 0.3812 data: 0.0002 max mem: 52126 +Epoch: [31] [ 190/3634] eta: 0:32:15 lr: 0.000002 loss: 0.5246 (0.5118) time: 0.3804 data: 0.0002 max mem: 52126 +Epoch: [31] [ 200/3634] eta: 0:31:39 lr: 0.000002 loss: 0.5325 (0.5131) time: 0.3810 data: 0.0002 max mem: 52126 +Epoch: [31] [ 210/3634] eta: 0:31:45 lr: 0.000002 loss: 0.5423 (0.5159) time: 0.5035 data: 0.0002 max mem: 52126 +Epoch: [31] [ 220/3634] eta: 0:31:12 lr: 0.000002 loss: 0.5335 (0.5165) time: 0.5039 data: 0.0002 max mem: 52126 +Epoch: [31] [ 230/3634] eta: 0:30:57 lr: 0.000002 loss: 0.5169 (0.5160) time: 0.4312 data: 0.0002 max mem: 52126 +Epoch: [31] [ 240/3634] eta: 0:30:46 lr: 0.000002 loss: 0.5028 (0.5163) time: 0.4921 data: 0.0002 max mem: 52126 +Epoch: [31] [ 250/3634] eta: 0:30:18 lr: 0.000002 loss: 0.5016 (0.5157) time: 0.4417 data: 0.0002 max mem: 52126 +Epoch: [31] [ 260/3634] eta: 0:30:09 lr: 0.000002 loss: 0.5177 (0.5161) time: 0.4431 data: 0.0002 max mem: 52126 +Epoch: [31] [ 270/3634] eta: 0:29:56 lr: 0.000002 loss: 0.5484 (0.5171) time: 0.4926 data: 0.0002 max mem: 52126 +Epoch: [31] [ 280/3634] eta: 0:29:32 lr: 0.000002 loss: 0.5632 (0.5183) time: 0.4293 data: 0.0002 max mem: 52126 +Epoch: [31] [ 290/3634] eta: 0:29:24 lr: 0.000002 loss: 0.5321 (0.5165) time: 0.4408 data: 0.0002 max mem: 52126 +Epoch: [31] [ 300/3634] eta: 0:29:14 lr: 0.000002 loss: 0.4815 (0.5160) time: 0.4914 data: 0.0002 max mem: 52126 +Epoch: [31] [ 310/3634] eta: 0:29:16 lr: 0.000002 loss: 0.5181 (0.5165) time: 0.5394 data: 0.0002 max mem: 52126 +Epoch: [31] [ 320/3634] eta: 0:29:08 lr: 0.000002 loss: 0.5197 (0.5166) time: 0.5508 data: 0.0002 max mem: 52126 +Epoch: [31] [ 330/3634] eta: 0:28:57 lr: 0.000002 loss: 0.5377 (0.5168) time: 0.4876 data: 0.0002 max mem: 52126 +Epoch: [31] [ 340/3634] eta: 0:28:50 lr: 0.000002 loss: 0.5323 (0.5167) time: 0.4890 data: 0.0002 max mem: 52126 +Epoch: [31] [ 350/3634] eta: 0:28:49 lr: 0.000002 loss: 0.5071 (0.5157) time: 0.5395 data: 0.0002 max mem: 52126 +Epoch: [31] [ 360/3634] eta: 0:28:42 lr: 0.000002 loss: 0.4953 (0.5148) time: 0.5392 data: 0.0002 max mem: 52126 +Epoch: [31] [ 370/3634] eta: 0:28:32 lr: 0.000002 loss: 0.5067 (0.5150) time: 0.4894 data: 0.0002 max mem: 52126 +Epoch: [31] [ 380/3634] eta: 0:28:30 lr: 0.000002 loss: 0.4731 (0.5139) time: 0.5197 data: 0.0002 max mem: 52126 +Epoch: [31] [ 390/3634] eta: 0:28:32 lr: 0.000002 loss: 0.4731 (0.5138) time: 0.5841 data: 0.0002 max mem: 52126 +Epoch: [31] [ 400/3634] eta: 0:28:22 lr: 0.000002 loss: 0.4964 (0.5132) time: 0.5374 data: 0.0001 max mem: 52126 +Epoch: [31] [ 410/3634] eta: 0:28:05 lr: 0.000002 loss: 0.5317 (0.5139) time: 0.4266 data: 0.0002 max mem: 52126 +Epoch: [31] [ 420/3634] eta: 0:28:08 lr: 0.000002 loss: 0.5389 (0.5138) time: 0.5029 data: 0.0002 max mem: 52126 +Epoch: [31] [ 430/3634] eta: 0:27:52 lr: 0.000002 loss: 0.5268 (0.5143) time: 0.5049 data: 0.0002 max mem: 52126 +Epoch: [31] [ 440/3634] eta: 0:27:43 lr: 0.000002 loss: 0.5477 (0.5151) time: 0.4284 data: 0.0002 max mem: 52126 +Epoch: [31] [ 450/3634] eta: 0:27:44 lr: 0.000002 loss: 0.5134 (0.5141) time: 0.5372 data: 0.0002 max mem: 52126 +Epoch: [31] [ 460/3634] eta: 0:27:37 lr: 0.000002 loss: 0.4595 (0.5133) time: 0.5513 data: 0.0002 max mem: 52126 +Epoch: [31] [ 470/3634] eta: 0:27:29 lr: 0.000002 loss: 0.5212 (0.5139) time: 0.4910 data: 0.0002 max mem: 52126 +Epoch: [31] [ 480/3634] eta: 0:27:27 lr: 0.000002 loss: 0.5246 (0.5138) time: 0.5285 data: 0.0001 max mem: 52126 +Epoch: [31] [ 490/3634] eta: 0:27:13 lr: 0.000002 loss: 0.5269 (0.5139) time: 0.4783 data: 0.0001 max mem: 52126 +Epoch: [31] [ 500/3634] eta: 0:26:59 lr: 0.000002 loss: 0.5269 (0.5141) time: 0.3805 data: 0.0002 max mem: 52126 +Epoch: [31] [ 510/3634] eta: 0:26:51 lr: 0.000002 loss: 0.5426 (0.5144) time: 0.4276 data: 0.0002 max mem: 52126 +Epoch: [31] [ 520/3634] eta: 0:26:53 lr: 0.000002 loss: 0.4966 (0.5139) time: 0.5496 data: 0.0002 max mem: 52126 +Epoch: [31] [ 530/3634] eta: 0:26:40 lr: 0.000002 loss: 0.4905 (0.5133) time: 0.5030 data: 0.0002 max mem: 52126 +Epoch: [31] [ 540/3634] eta: 0:26:33 lr: 0.000002 loss: 0.5071 (0.5132) time: 0.4401 data: 0.0002 max mem: 52126 +Epoch: [31] [ 550/3634] eta: 0:26:34 lr: 0.000002 loss: 0.4697 (0.5121) time: 0.5613 data: 0.0002 max mem: 52126 +Epoch: [31] [ 560/3634] eta: 0:26:22 lr: 0.000002 loss: 0.4561 (0.5116) time: 0.5027 data: 0.0002 max mem: 52126 +Epoch: [31] [ 570/3634] eta: 0:26:21 lr: 0.000002 loss: 0.5096 (0.5120) time: 0.4918 data: 0.0002 max mem: 52126 +Epoch: [31] [ 580/3634] eta: 0:26:09 lr: 0.000002 loss: 0.5431 (0.5124) time: 0.4943 data: 0.0002 max mem: 52126 +Epoch: [31] [ 590/3634] eta: 0:26:09 lr: 0.000002 loss: 0.5210 (0.5122) time: 0.5016 data: 0.0002 max mem: 52126 +Epoch: [31] [ 600/3634] eta: 0:26:02 lr: 0.000002 loss: 0.5144 (0.5122) time: 0.5448 data: 0.0002 max mem: 52126 +Epoch: [31] [ 610/3634] eta: 0:26:01 lr: 0.000002 loss: 0.5113 (0.5123) time: 0.5349 data: 0.0002 max mem: 52126 +Epoch: [31] [ 620/3634] eta: 0:25:49 lr: 0.000002 loss: 0.5445 (0.5127) time: 0.4893 data: 0.0002 max mem: 52126 +Epoch: [31] [ 630/3634] eta: 0:25:51 lr: 0.000002 loss: 0.5377 (0.5126) time: 0.5179 data: 0.0002 max mem: 52126 +Epoch: [31] [ 640/3634] eta: 0:25:43 lr: 0.000002 loss: 0.5309 (0.5128) time: 0.5618 data: 0.0002 max mem: 52126 +Epoch: [31] [ 650/3634] eta: 0:25:48 lr: 0.000002 loss: 0.5490 (0.5134) time: 0.5944 data: 0.0002 max mem: 52126 +Epoch: [31] [ 660/3634] eta: 0:25:36 lr: 0.000002 loss: 0.5280 (0.5133) time: 0.5511 data: 0.0002 max mem: 52126 +Epoch: [31] [ 670/3634] eta: 0:25:29 lr: 0.000002 loss: 0.5077 (0.5129) time: 0.4261 data: 0.0002 max mem: 52126 +Epoch: [31] [ 680/3634] eta: 0:25:18 lr: 0.000002 loss: 0.5059 (0.5131) time: 0.4251 data: 0.0002 max mem: 52126 +Epoch: [31] [ 690/3634] eta: 0:25:17 lr: 0.000002 loss: 0.5059 (0.5126) time: 0.5001 data: 0.0002 max mem: 52126 +Epoch: [31] [ 700/3634] eta: 0:25:14 lr: 0.000002 loss: 0.4941 (0.5126) time: 0.5915 data: 0.0002 max mem: 52126 +Epoch: [31] [ 710/3634] eta: 0:25:11 lr: 0.000002 loss: 0.4927 (0.5121) time: 0.5606 data: 0.0002 max mem: 52126 +Epoch: [31] [ 720/3634] eta: 0:25:00 lr: 0.000002 loss: 0.4980 (0.5123) time: 0.4691 data: 0.0002 max mem: 52126 +Epoch: [31] [ 730/3634] eta: 0:24:53 lr: 0.000002 loss: 0.5203 (0.5125) time: 0.4257 data: 0.0002 max mem: 52126 +Epoch: [31] [ 740/3634] eta: 0:24:51 lr: 0.000002 loss: 0.5203 (0.5127) time: 0.5300 data: 0.0002 max mem: 52126 +Epoch: [31] [ 750/3634] eta: 0:24:44 lr: 0.000002 loss: 0.5417 (0.5129) time: 0.5302 data: 0.0002 max mem: 52126 +Epoch: [31] [ 760/3634] eta: 0:24:38 lr: 0.000002 loss: 0.5445 (0.5132) time: 0.4859 data: 0.0002 max mem: 52126 +Epoch: [31] [ 770/3634] eta: 0:24:28 lr: 0.000002 loss: 0.5547 (0.5135) time: 0.4406 data: 0.0002 max mem: 52126 +Epoch: [31] [ 780/3634] eta: 0:24:27 lr: 0.000002 loss: 0.5488 (0.5137) time: 0.5008 data: 0.0002 max mem: 52126 +Epoch: [31] [ 790/3634] eta: 0:24:20 lr: 0.000002 loss: 0.5219 (0.5136) time: 0.5444 data: 0.0002 max mem: 52126 +Epoch: [31] [ 800/3634] eta: 0:24:19 lr: 0.000002 loss: 0.5204 (0.5137) time: 0.5415 data: 0.0002 max mem: 52126 +Epoch: [31] [ 810/3634] eta: 0:24:12 lr: 0.000002 loss: 0.4998 (0.5134) time: 0.5428 data: 0.0002 max mem: 52126 +Epoch: [31] [ 820/3634] eta: 0:24:10 lr: 0.000002 loss: 0.4428 (0.5128) time: 0.5315 data: 0.0002 max mem: 52126 +Epoch: [31] [ 830/3634] eta: 0:24:00 lr: 0.000002 loss: 0.4923 (0.5125) time: 0.4862 data: 0.0002 max mem: 52126 +Epoch: [31] [ 840/3634] eta: 0:23:53 lr: 0.000002 loss: 0.5121 (0.5127) time: 0.4264 data: 0.0002 max mem: 52126 +Epoch: [31] [ 850/3634] eta: 0:23:57 lr: 0.000002 loss: 0.5120 (0.5126) time: 0.6235 data: 0.0002 max mem: 52126 +Epoch: [31] [ 860/3634] eta: 0:23:47 lr: 0.000002 loss: 0.4969 (0.5124) time: 0.5781 data: 0.0002 max mem: 52126 +Epoch: [31] [ 870/3634] eta: 0:23:45 lr: 0.000002 loss: 0.5004 (0.5126) time: 0.4885 data: 0.0002 max mem: 52126 +Epoch: [31] [ 880/3634] eta: 0:23:38 lr: 0.000002 loss: 0.5319 (0.5131) time: 0.5326 data: 0.0002 max mem: 52126 +Epoch: [31] [ 890/3634] eta: 0:23:32 lr: 0.000002 loss: 0.5585 (0.5134) time: 0.4726 data: 0.0002 max mem: 52126 +Epoch: [31] [ 900/3634] eta: 0:23:23 lr: 0.000002 loss: 0.5751 (0.5138) time: 0.4278 data: 0.0002 max mem: 52126 +Epoch: [31] [ 910/3634] eta: 0:23:16 lr: 0.000002 loss: 0.5641 (0.5135) time: 0.4266 data: 0.0001 max mem: 52126 +Epoch: [31] [ 920/3634] eta: 0:23:07 lr: 0.000002 loss: 0.5285 (0.5138) time: 0.4273 data: 0.0002 max mem: 52126 +Epoch: [31] [ 930/3634] eta: 0:23:09 lr: 0.000002 loss: 0.5292 (0.5138) time: 0.5636 data: 0.0002 max mem: 52126 +Epoch: [31] [ 940/3634] eta: 0:23:00 lr: 0.000002 loss: 0.5203 (0.5140) time: 0.5634 data: 0.0002 max mem: 52126 +Epoch: [31] [ 950/3634] eta: 0:22:57 lr: 0.000002 loss: 0.5209 (0.5140) time: 0.4872 data: 0.0002 max mem: 52126 +Epoch: [31] [ 960/3634] eta: 0:22:48 lr: 0.000002 loss: 0.5160 (0.5141) time: 0.4872 data: 0.0002 max mem: 52126 +Epoch: [31] [ 970/3634] eta: 0:22:43 lr: 0.000002 loss: 0.4870 (0.5138) time: 0.4424 data: 0.0002 max mem: 52126 +Epoch: [31] [ 980/3634] eta: 0:22:41 lr: 0.000002 loss: 0.5119 (0.5139) time: 0.5616 data: 0.0002 max mem: 52126 +Epoch: [31] [ 990/3634] eta: 0:22:35 lr: 0.000002 loss: 0.5303 (0.5141) time: 0.5458 data: 0.0002 max mem: 52126 +Epoch: [31] [1000/3634] eta: 0:22:26 lr: 0.000002 loss: 0.5303 (0.5140) time: 0.4266 data: 0.0002 max mem: 52126 +Epoch: [31] [1010/3634] eta: 0:22:18 lr: 0.000002 loss: 0.4930 (0.5138) time: 0.3814 data: 0.0002 max mem: 52126 +Epoch: [31] [1020/3634] eta: 0:22:12 lr: 0.000002 loss: 0.5105 (0.5139) time: 0.4408 data: 0.0002 max mem: 52126 +Epoch: [31] [1030/3634] eta: 0:22:04 lr: 0.000002 loss: 0.5472 (0.5142) time: 0.4410 data: 0.0002 max mem: 52126 +Epoch: [31] [1040/3634] eta: 0:21:56 lr: 0.000002 loss: 0.5481 (0.5144) time: 0.3809 data: 0.0002 max mem: 52126 +Epoch: [31] [1050/3634] eta: 0:21:47 lr: 0.000002 loss: 0.5481 (0.5145) time: 0.3809 data: 0.0002 max mem: 52126 +Epoch: [31] [1060/3634] eta: 0:21:42 lr: 0.000002 loss: 0.5383 (0.5147) time: 0.4416 data: 0.0002 max mem: 52126 +Epoch: [31] [1070/3634] eta: 0:21:41 lr: 0.000002 loss: 0.5383 (0.5148) time: 0.5890 data: 0.0002 max mem: 52126 +Epoch: [31] [1080/3634] eta: 0:21:33 lr: 0.000002 loss: 0.5232 (0.5148) time: 0.5292 data: 0.0002 max mem: 52126 +Epoch: [31] [1090/3634] eta: 0:21:31 lr: 0.000002 loss: 0.5129 (0.5148) time: 0.5159 data: 0.0002 max mem: 52126 +Epoch: [31] [1100/3634] eta: 0:21:23 lr: 0.000002 loss: 0.5113 (0.5149) time: 0.5158 data: 0.0002 max mem: 52126 +Epoch: [31] [1110/3634] eta: 0:21:16 lr: 0.000002 loss: 0.5287 (0.5150) time: 0.3820 data: 0.0002 max mem: 52126 +Epoch: [31] [1120/3634] eta: 0:21:08 lr: 0.000002 loss: 0.5191 (0.5151) time: 0.3827 data: 0.0002 max mem: 52126 +Epoch: [31] [1130/3634] eta: 0:21:00 lr: 0.000002 loss: 0.5057 (0.5149) time: 0.3825 data: 0.0002 max mem: 52126 +Epoch: [31] [1140/3634] eta: 0:20:55 lr: 0.000002 loss: 0.4886 (0.5152) time: 0.4415 data: 0.0002 max mem: 52126 +Epoch: [31] [1150/3634] eta: 0:20:47 lr: 0.000002 loss: 0.4720 (0.5147) time: 0.4414 data: 0.0002 max mem: 52126 +Epoch: [31] [1160/3634] eta: 0:20:44 lr: 0.000002 loss: 0.4720 (0.5147) time: 0.4856 data: 0.0002 max mem: 52126 +Epoch: [31] [1170/3634] eta: 0:20:37 lr: 0.000002 loss: 0.5216 (0.5147) time: 0.4853 data: 0.0002 max mem: 52126 +Epoch: [31] [1180/3634] eta: 0:20:33 lr: 0.000002 loss: 0.5442 (0.5150) time: 0.4688 data: 0.0002 max mem: 52126 +Epoch: [31] [1190/3634] eta: 0:20:28 lr: 0.000002 loss: 0.5419 (0.5153) time: 0.5287 data: 0.0002 max mem: 52126 +Epoch: [31] [1200/3634] eta: 0:20:26 lr: 0.000002 loss: 0.5413 (0.5154) time: 0.5881 data: 0.0002 max mem: 52126 +Epoch: [31] [1210/3634] eta: 0:20:24 lr: 0.000002 loss: 0.5311 (0.5154) time: 0.6645 data: 0.0002 max mem: 52126 +Epoch: [31] [1220/3634] eta: 0:20:17 lr: 0.000002 loss: 0.5057 (0.5151) time: 0.5175 data: 0.0002 max mem: 52126 +Epoch: [31] [1230/3634] eta: 0:20:16 lr: 0.000002 loss: 0.5113 (0.5152) time: 0.5472 data: 0.0002 max mem: 52126 +Epoch: [31] [1240/3634] eta: 0:20:10 lr: 0.000002 loss: 0.5380 (0.5153) time: 0.5927 data: 0.0002 max mem: 52126 +Epoch: [31] [1250/3634] eta: 0:20:02 lr: 0.000002 loss: 0.5315 (0.5154) time: 0.4269 data: 0.0002 max mem: 52126 +Epoch: [31] [1260/3634] eta: 0:19:57 lr: 0.000002 loss: 0.5246 (0.5154) time: 0.4314 data: 0.0002 max mem: 52126 +Epoch: [31] [1270/3634] eta: 0:19:51 lr: 0.000002 loss: 0.5119 (0.5153) time: 0.4755 data: 0.0001 max mem: 52126 +Epoch: [31] [1280/3634] eta: 0:19:46 lr: 0.000002 loss: 0.5119 (0.5156) time: 0.4698 data: 0.0002 max mem: 52126 +Epoch: [31] [1290/3634] eta: 0:19:40 lr: 0.000002 loss: 0.5528 (0.5158) time: 0.4712 data: 0.0002 max mem: 52126 +Epoch: [31] [1300/3634] eta: 0:19:35 lr: 0.000002 loss: 0.5122 (0.5158) time: 0.4857 data: 0.0002 max mem: 52126 +Epoch: [31] [1310/3634] eta: 0:19:30 lr: 0.000002 loss: 0.5102 (0.5159) time: 0.4999 data: 0.0002 max mem: 52126 +Epoch: [31] [1320/3634] eta: 0:19:23 lr: 0.000002 loss: 0.5117 (0.5156) time: 0.4416 data: 0.0002 max mem: 52126 +Epoch: [31] [1330/3634] eta: 0:19:21 lr: 0.000002 loss: 0.4946 (0.5155) time: 0.5439 data: 0.0002 max mem: 52126 +Epoch: [31] [1340/3634] eta: 0:19:16 lr: 0.000002 loss: 0.4815 (0.5153) time: 0.6041 data: 0.0002 max mem: 52126 +Epoch: [31] [1350/3634] eta: 0:19:10 lr: 0.000002 loss: 0.4869 (0.5153) time: 0.4879 data: 0.0002 max mem: 52126 +Epoch: [31] [1360/3634] eta: 0:19:07 lr: 0.000002 loss: 0.5054 (0.5154) time: 0.5324 data: 0.0002 max mem: 52126 +Epoch: [31] [1370/3634] eta: 0:19:00 lr: 0.000002 loss: 0.5110 (0.5154) time: 0.4866 data: 0.0002 max mem: 52126 +Epoch: [31] [1380/3634] eta: 0:18:53 lr: 0.000002 loss: 0.5016 (0.5152) time: 0.3819 data: 0.0001 max mem: 52126 +Epoch: [31] [1390/3634] eta: 0:18:50 lr: 0.000002 loss: 0.5021 (0.5153) time: 0.4996 data: 0.0002 max mem: 52126 +Epoch: [31] [1400/3634] eta: 0:18:43 lr: 0.000002 loss: 0.5252 (0.5153) time: 0.4990 data: 0.0002 max mem: 52126 +Epoch: [31] [1410/3634] eta: 0:18:37 lr: 0.000002 loss: 0.5384 (0.5154) time: 0.4251 data: 0.0002 max mem: 52126 +Epoch: [31] [1420/3634] eta: 0:18:34 lr: 0.000002 loss: 0.5384 (0.5156) time: 0.5358 data: 0.0002 max mem: 52126 +Epoch: [31] [1430/3634] eta: 0:18:31 lr: 0.000002 loss: 0.5232 (0.5155) time: 0.6404 data: 0.0002 max mem: 52126 +Epoch: [31] [1440/3634] eta: 0:18:24 lr: 0.000002 loss: 0.4800 (0.5152) time: 0.5301 data: 0.0002 max mem: 52126 +Epoch: [31] [1450/3634] eta: 0:18:19 lr: 0.000002 loss: 0.4673 (0.5151) time: 0.4401 data: 0.0002 max mem: 52126 +Epoch: [31] [1460/3634] eta: 0:18:17 lr: 0.000002 loss: 0.4644 (0.5147) time: 0.5822 data: 0.0002 max mem: 52126 +Epoch: [31] [1470/3634] eta: 0:18:11 lr: 0.000002 loss: 0.4845 (0.5148) time: 0.5827 data: 0.0002 max mem: 52126 +Epoch: [31] [1480/3634] eta: 0:18:06 lr: 0.000002 loss: 0.4976 (0.5146) time: 0.4854 data: 0.0002 max mem: 52126 +Epoch: [31] [1490/3634] eta: 0:17:59 lr: 0.000002 loss: 0.4976 (0.5144) time: 0.4267 data: 0.0001 max mem: 52126 +Epoch: [31] [1500/3634] eta: 0:17:57 lr: 0.000002 loss: 0.4959 (0.5142) time: 0.5315 data: 0.0002 max mem: 52126 +Epoch: [31] [1510/3634] eta: 0:17:54 lr: 0.000002 loss: 0.4867 (0.5142) time: 0.6790 data: 0.0002 max mem: 52126 +Epoch: [31] [1520/3634] eta: 0:17:47 lr: 0.000002 loss: 0.5376 (0.5144) time: 0.5290 data: 0.0001 max mem: 52126 +Epoch: [31] [1530/3634] eta: 0:17:42 lr: 0.000002 loss: 0.5349 (0.5143) time: 0.4259 data: 0.0002 max mem: 52126 +Epoch: [31] [1540/3634] eta: 0:17:35 lr: 0.000002 loss: 0.5185 (0.5144) time: 0.4257 data: 0.0002 max mem: 52126 +Epoch: [31] [1550/3634] eta: 0:17:28 lr: 0.000002 loss: 0.5004 (0.5144) time: 0.3813 data: 0.0001 max mem: 52126 +Epoch: [31] [1560/3634] eta: 0:17:23 lr: 0.000002 loss: 0.5242 (0.5145) time: 0.4271 data: 0.0002 max mem: 52126 +Epoch: [31] [1570/3634] eta: 0:17:18 lr: 0.000002 loss: 0.5252 (0.5144) time: 0.4857 data: 0.0002 max mem: 52126 +Epoch: [31] [1580/3634] eta: 0:17:13 lr: 0.000002 loss: 0.5124 (0.5144) time: 0.5008 data: 0.0002 max mem: 52126 +Epoch: [31] [1590/3634] eta: 0:17:06 lr: 0.000002 loss: 0.5257 (0.5146) time: 0.4422 data: 0.0002 max mem: 52126 +Epoch: [31] [1600/3634] eta: 0:17:00 lr: 0.000002 loss: 0.5355 (0.5145) time: 0.3811 data: 0.0002 max mem: 52126 +Epoch: [31] [1610/3634] eta: 0:16:54 lr: 0.000002 loss: 0.4751 (0.5143) time: 0.4259 data: 0.0002 max mem: 52126 +Epoch: [31] [1620/3634] eta: 0:16:50 lr: 0.000002 loss: 0.4564 (0.5142) time: 0.5154 data: 0.0001 max mem: 52126 +Epoch: [31] [1630/3634] eta: 0:16:43 lr: 0.000002 loss: 0.5237 (0.5142) time: 0.4707 data: 0.0002 max mem: 52126 +Epoch: [31] [1640/3634] eta: 0:16:38 lr: 0.000002 loss: 0.5257 (0.5142) time: 0.4398 data: 0.0002 max mem: 52126 +Epoch: [31] [1650/3634] eta: 0:16:33 lr: 0.000002 loss: 0.5198 (0.5142) time: 0.4868 data: 0.0002 max mem: 52126 +Epoch: [31] [1660/3634] eta: 0:16:27 lr: 0.000002 loss: 0.5287 (0.5143) time: 0.4286 data: 0.0002 max mem: 52126 +Epoch: [31] [1670/3634] eta: 0:16:22 lr: 0.000002 loss: 0.5028 (0.5141) time: 0.4398 data: 0.0002 max mem: 52126 +Epoch: [31] [1680/3634] eta: 0:16:16 lr: 0.000002 loss: 0.4963 (0.5143) time: 0.4847 data: 0.0002 max mem: 52126 +Epoch: [31] [1690/3634] eta: 0:16:11 lr: 0.000002 loss: 0.5597 (0.5144) time: 0.4711 data: 0.0002 max mem: 52126 +Epoch: [31] [1700/3634] eta: 0:16:05 lr: 0.000002 loss: 0.5744 (0.5146) time: 0.4268 data: 0.0002 max mem: 52126 +Epoch: [31] [1710/3634] eta: 0:15:58 lr: 0.000002 loss: 0.5610 (0.5147) time: 0.3819 data: 0.0002 max mem: 52126 +Epoch: [31] [1720/3634] eta: 0:15:54 lr: 0.000002 loss: 0.5233 (0.5149) time: 0.4852 data: 0.0002 max mem: 52126 +Epoch: [31] [1730/3634] eta: 0:15:50 lr: 0.000002 loss: 0.5469 (0.5150) time: 0.5730 data: 0.0002 max mem: 52126 +Epoch: [31] [1740/3634] eta: 0:15:46 lr: 0.000002 loss: 0.5432 (0.5152) time: 0.5641 data: 0.0002 max mem: 52126 +Epoch: [31] [1750/3634] eta: 0:15:39 lr: 0.000002 loss: 0.5523 (0.5154) time: 0.4765 data: 0.0002 max mem: 52126 +Epoch: [31] [1760/3634] eta: 0:15:36 lr: 0.000002 loss: 0.5560 (0.5155) time: 0.5157 data: 0.0002 max mem: 52126 +Epoch: [31] [1770/3634] eta: 0:15:31 lr: 0.000002 loss: 0.5205 (0.5155) time: 0.5744 data: 0.0001 max mem: 52126 +Epoch: [31] [1780/3634] eta: 0:15:26 lr: 0.000002 loss: 0.5022 (0.5155) time: 0.4989 data: 0.0002 max mem: 52126 +Epoch: [31] [1790/3634] eta: 0:15:20 lr: 0.000002 loss: 0.5199 (0.5157) time: 0.4402 data: 0.0002 max mem: 52126 +Epoch: [31] [1800/3634] eta: 0:15:14 lr: 0.000002 loss: 0.5308 (0.5158) time: 0.3817 data: 0.0002 max mem: 52126 +Epoch: [31] [1810/3634] eta: 0:15:09 lr: 0.000002 loss: 0.5308 (0.5158) time: 0.4421 data: 0.0002 max mem: 52126 +Epoch: [31] [1820/3634] eta: 0:15:04 lr: 0.000002 loss: 0.5122 (0.5158) time: 0.4901 data: 0.0002 max mem: 52126 +Epoch: [31] [1830/3634] eta: 0:14:57 lr: 0.000002 loss: 0.5010 (0.5157) time: 0.4298 data: 0.0002 max mem: 52126 +Epoch: [31] [1840/3634] eta: 0:14:54 lr: 0.000002 loss: 0.4987 (0.5157) time: 0.5315 data: 0.0002 max mem: 52126 +Epoch: [31] [1850/3634] eta: 0:14:50 lr: 0.000002 loss: 0.4917 (0.5156) time: 0.6370 data: 0.0002 max mem: 52126 +Epoch: [31] [1860/3634] eta: 0:14:45 lr: 0.000002 loss: 0.5259 (0.5158) time: 0.5470 data: 0.0002 max mem: 52126 +Epoch: [31] [1870/3634] eta: 0:14:42 lr: 0.000002 loss: 0.5328 (0.5155) time: 0.5901 data: 0.0001 max mem: 52126 +Epoch: [31] [1880/3634] eta: 0:14:38 lr: 0.000002 loss: 0.4805 (0.5154) time: 0.6480 data: 0.0002 max mem: 52126 +Epoch: [31] [1890/3634] eta: 0:14:32 lr: 0.000002 loss: 0.4926 (0.5152) time: 0.4998 data: 0.0002 max mem: 52126 +Epoch: [31] [1900/3634] eta: 0:14:27 lr: 0.000002 loss: 0.5176 (0.5152) time: 0.4412 data: 0.0002 max mem: 52126 +Epoch: [31] [1910/3634] eta: 0:14:23 lr: 0.000002 loss: 0.5421 (0.5153) time: 0.5449 data: 0.0002 max mem: 52126 +Epoch: [31] [1920/3634] eta: 0:14:17 lr: 0.000002 loss: 0.5240 (0.5151) time: 0.4861 data: 0.0002 max mem: 52126 +Epoch: [31] [1930/3634] eta: 0:14:12 lr: 0.000002 loss: 0.5240 (0.5151) time: 0.4713 data: 0.0002 max mem: 52126 +Epoch: [31] [1940/3634] eta: 0:14:10 lr: 0.000002 loss: 0.5188 (0.5151) time: 0.6704 data: 0.0002 max mem: 52126 +Epoch: [31] [1950/3634] eta: 0:14:04 lr: 0.000002 loss: 0.5188 (0.5151) time: 0.5813 data: 0.0002 max mem: 52126 +Epoch: [31] [1960/3634] eta: 0:13:57 lr: 0.000002 loss: 0.5149 (0.5152) time: 0.3814 data: 0.0002 max mem: 52126 +Epoch: [31] [1970/3634] eta: 0:13:51 lr: 0.000002 loss: 0.5334 (0.5154) time: 0.3818 data: 0.0002 max mem: 52126 +Epoch: [31] [1980/3634] eta: 0:13:47 lr: 0.000002 loss: 0.5447 (0.5155) time: 0.4431 data: 0.0002 max mem: 52126 +Epoch: [31] [1990/3634] eta: 0:13:42 lr: 0.000002 loss: 0.5447 (0.5157) time: 0.5329 data: 0.0002 max mem: 52126 +Epoch: [31] [2000/3634] eta: 0:13:38 lr: 0.000002 loss: 0.5362 (0.5157) time: 0.6084 data: 0.0002 max mem: 52126 +Epoch: [31] [2010/3634] eta: 0:13:34 lr: 0.000002 loss: 0.4957 (0.5157) time: 0.6230 data: 0.0002 max mem: 52126 +Epoch: [31] [2020/3634] eta: 0:13:29 lr: 0.000002 loss: 0.5424 (0.5160) time: 0.5483 data: 0.0002 max mem: 52126 +Epoch: [31] [2030/3634] eta: 0:13:24 lr: 0.000002 loss: 0.5568 (0.5162) time: 0.5024 data: 0.0002 max mem: 52126 +Epoch: [31] [2040/3634] eta: 0:13:18 lr: 0.000002 loss: 0.5507 (0.5163) time: 0.4408 data: 0.0001 max mem: 52126 +Epoch: [31] [2050/3634] eta: 0:13:12 lr: 0.000002 loss: 0.5241 (0.5163) time: 0.3824 data: 0.0002 max mem: 52126 +Epoch: [31] [2060/3634] eta: 0:13:08 lr: 0.000002 loss: 0.5352 (0.5165) time: 0.4862 data: 0.0002 max mem: 52126 +Epoch: [31] [2070/3634] eta: 0:13:03 lr: 0.000002 loss: 0.5484 (0.5166) time: 0.5447 data: 0.0002 max mem: 52126 +Epoch: [31] [2080/3634] eta: 0:12:58 lr: 0.000002 loss: 0.5226 (0.5165) time: 0.4905 data: 0.0002 max mem: 52126 +Epoch: [31] [2090/3634] eta: 0:12:53 lr: 0.000002 loss: 0.5084 (0.5163) time: 0.4908 data: 0.0002 max mem: 52126 +Epoch: [31] [2100/3634] eta: 0:12:47 lr: 0.000002 loss: 0.4837 (0.5163) time: 0.4412 data: 0.0001 max mem: 52126 +Epoch: [31] [2110/3634] eta: 0:12:43 lr: 0.000002 loss: 0.4877 (0.5163) time: 0.5442 data: 0.0001 max mem: 52126 +Epoch: [31] [2120/3634] eta: 0:12:38 lr: 0.000002 loss: 0.4765 (0.5161) time: 0.5891 data: 0.0002 max mem: 52126 +Epoch: [31] [2130/3634] eta: 0:12:32 lr: 0.000002 loss: 0.4765 (0.5160) time: 0.4264 data: 0.0002 max mem: 52126 +Epoch: [31] [2140/3634] eta: 0:12:28 lr: 0.000002 loss: 0.5037 (0.5159) time: 0.5163 data: 0.0002 max mem: 52126 +Epoch: [31] [2150/3634] eta: 0:12:23 lr: 0.000002 loss: 0.5397 (0.5159) time: 0.5609 data: 0.0002 max mem: 52126 +Epoch: [31] [2160/3634] eta: 0:12:17 lr: 0.000002 loss: 0.5311 (0.5159) time: 0.4254 data: 0.0002 max mem: 52126 +Epoch: [31] [2170/3634] eta: 0:12:11 lr: 0.000002 loss: 0.5253 (0.5159) time: 0.3818 data: 0.0002 max mem: 52126 +Epoch: [31] [2180/3634] eta: 0:12:06 lr: 0.000002 loss: 0.5183 (0.5160) time: 0.3821 data: 0.0002 max mem: 52126 +Epoch: [31] [2190/3634] eta: 0:12:01 lr: 0.000002 loss: 0.5613 (0.5162) time: 0.4399 data: 0.0002 max mem: 52126 +Epoch: [31] [2200/3634] eta: 0:11:56 lr: 0.000002 loss: 0.5426 (0.5160) time: 0.4991 data: 0.0002 max mem: 52126 +Epoch: [31] [2210/3634] eta: 0:11:51 lr: 0.000002 loss: 0.5361 (0.5162) time: 0.5307 data: 0.0002 max mem: 52126 +Epoch: [31] [2220/3634] eta: 0:11:45 lr: 0.000002 loss: 0.5304 (0.5161) time: 0.4732 data: 0.0002 max mem: 52126 +Epoch: [31] [2230/3634] eta: 0:11:40 lr: 0.000002 loss: 0.4769 (0.5159) time: 0.3832 data: 0.0002 max mem: 52126 +Epoch: [31] [2240/3634] eta: 0:11:34 lr: 0.000002 loss: 0.5180 (0.5159) time: 0.3816 data: 0.0002 max mem: 52126 +Epoch: [31] [2250/3634] eta: 0:11:29 lr: 0.000002 loss: 0.4968 (0.5157) time: 0.4427 data: 0.0002 max mem: 52126 +Epoch: [31] [2260/3634] eta: 0:11:24 lr: 0.000002 loss: 0.4868 (0.5158) time: 0.5314 data: 0.0002 max mem: 52126 +Epoch: [31] [2270/3634] eta: 0:11:19 lr: 0.000002 loss: 0.5352 (0.5159) time: 0.5174 data: 0.0002 max mem: 52126 +Epoch: [31] [2280/3634] eta: 0:11:14 lr: 0.000002 loss: 0.5227 (0.5159) time: 0.4884 data: 0.0002 max mem: 52126 +Epoch: [31] [2290/3634] eta: 0:11:09 lr: 0.000002 loss: 0.5015 (0.5160) time: 0.4860 data: 0.0002 max mem: 52126 +Epoch: [31] [2300/3634] eta: 0:11:04 lr: 0.000002 loss: 0.5100 (0.5159) time: 0.4847 data: 0.0002 max mem: 52126 +Epoch: [31] [2310/3634] eta: 0:10:59 lr: 0.000002 loss: 0.5100 (0.5159) time: 0.5290 data: 0.0002 max mem: 52126 +Epoch: [31] [2320/3634] eta: 0:10:54 lr: 0.000002 loss: 0.5432 (0.5161) time: 0.5146 data: 0.0002 max mem: 52126 +Epoch: [31] [2330/3634] eta: 0:10:49 lr: 0.000002 loss: 0.5346 (0.5158) time: 0.4850 data: 0.0002 max mem: 52126 +Epoch: [31] [2340/3634] eta: 0:10:44 lr: 0.000002 loss: 0.4947 (0.5158) time: 0.4899 data: 0.0002 max mem: 52126 +Epoch: [31] [2350/3634] eta: 0:10:39 lr: 0.000002 loss: 0.5216 (0.5158) time: 0.4309 data: 0.0002 max mem: 52126 +Epoch: [31] [2360/3634] eta: 0:10:34 lr: 0.000002 loss: 0.5440 (0.5159) time: 0.4859 data: 0.0002 max mem: 52126 +Epoch: [31] [2370/3634] eta: 0:10:30 lr: 0.000002 loss: 0.5408 (0.5159) time: 0.5931 data: 0.0001 max mem: 52126 +Epoch: [31] [2380/3634] eta: 0:10:25 lr: 0.000002 loss: 0.5304 (0.5159) time: 0.5364 data: 0.0002 max mem: 52126 +Epoch: [31] [2390/3634] eta: 0:10:19 lr: 0.000002 loss: 0.5198 (0.5159) time: 0.4294 data: 0.0002 max mem: 52126 +Epoch: [31] [2400/3634] eta: 0:10:14 lr: 0.000002 loss: 0.5146 (0.5159) time: 0.4707 data: 0.0002 max mem: 52126 +Epoch: [31] [2410/3634] eta: 0:10:09 lr: 0.000002 loss: 0.5185 (0.5159) time: 0.5283 data: 0.0002 max mem: 52126 +Epoch: [31] [2420/3634] eta: 0:10:04 lr: 0.000002 loss: 0.5285 (0.5159) time: 0.4839 data: 0.0002 max mem: 52126 +Epoch: [31] [2430/3634] eta: 0:09:59 lr: 0.000002 loss: 0.5421 (0.5159) time: 0.4848 data: 0.0002 max mem: 52126 +Epoch: [31] [2440/3634] eta: 0:09:54 lr: 0.000002 loss: 0.5019 (0.5157) time: 0.4408 data: 0.0002 max mem: 52126 +Epoch: [31] [2450/3634] eta: 0:09:50 lr: 0.000002 loss: 0.5107 (0.5158) time: 0.5424 data: 0.0002 max mem: 52126 +Epoch: [31] [2460/3634] eta: 0:09:44 lr: 0.000002 loss: 0.5203 (0.5158) time: 0.5427 data: 0.0002 max mem: 52126 +Epoch: [31] [2470/3634] eta: 0:09:39 lr: 0.000002 loss: 0.5340 (0.5158) time: 0.4417 data: 0.0002 max mem: 52126 +Epoch: [31] [2480/3634] eta: 0:09:34 lr: 0.000002 loss: 0.4919 (0.5157) time: 0.5005 data: 0.0002 max mem: 52126 +Epoch: [31] [2490/3634] eta: 0:09:29 lr: 0.000002 loss: 0.5123 (0.5157) time: 0.4412 data: 0.0002 max mem: 52126 +Epoch: [31] [2500/3634] eta: 0:09:24 lr: 0.000002 loss: 0.5287 (0.5158) time: 0.4419 data: 0.0002 max mem: 52126 +Epoch: [31] [2510/3634] eta: 0:09:18 lr: 0.000002 loss: 0.5415 (0.5159) time: 0.4408 data: 0.0002 max mem: 52126 +Epoch: [31] [2520/3634] eta: 0:09:14 lr: 0.000002 loss: 0.5436 (0.5159) time: 0.5355 data: 0.0002 max mem: 52126 +Epoch: [31] [2530/3634] eta: 0:09:10 lr: 0.000002 loss: 0.5414 (0.5159) time: 0.6877 data: 0.0002 max mem: 52126 +Epoch: [31] [2540/3634] eta: 0:09:05 lr: 0.000002 loss: 0.4947 (0.5157) time: 0.6389 data: 0.0001 max mem: 52126 +Epoch: [31] [2550/3634] eta: 0:09:00 lr: 0.000002 loss: 0.4786 (0.5157) time: 0.4873 data: 0.0002 max mem: 52126 +Epoch: [31] [2560/3634] eta: 0:08:55 lr: 0.000002 loss: 0.4849 (0.5156) time: 0.4322 data: 0.0002 max mem: 52126 +Epoch: [31] [2570/3634] eta: 0:08:49 lr: 0.000002 loss: 0.5187 (0.5157) time: 0.4319 data: 0.0002 max mem: 52126 +Epoch: [31] [2580/3634] eta: 0:08:44 lr: 0.000002 loss: 0.5542 (0.5158) time: 0.4416 data: 0.0002 max mem: 52126 +Epoch: [31] [2590/3634] eta: 0:08:39 lr: 0.000002 loss: 0.5405 (0.5157) time: 0.4416 data: 0.0002 max mem: 52126 +Epoch: [31] [2600/3634] eta: 0:08:34 lr: 0.000002 loss: 0.5216 (0.5158) time: 0.4984 data: 0.0002 max mem: 52126 +Epoch: [31] [2610/3634] eta: 0:08:29 lr: 0.000002 loss: 0.5467 (0.5159) time: 0.4990 data: 0.0002 max mem: 52126 +Epoch: [31] [2620/3634] eta: 0:08:24 lr: 0.000002 loss: 0.5295 (0.5158) time: 0.3818 data: 0.0002 max mem: 52126 +Epoch: [31] [2630/3634] eta: 0:08:19 lr: 0.000002 loss: 0.5181 (0.5159) time: 0.4259 data: 0.0002 max mem: 52126 +Epoch: [31] [2640/3634] eta: 0:08:13 lr: 0.000002 loss: 0.5213 (0.5159) time: 0.4263 data: 0.0002 max mem: 52126 +Epoch: [31] [2650/3634] eta: 0:08:08 lr: 0.000002 loss: 0.4963 (0.5158) time: 0.4269 data: 0.0002 max mem: 52126 +Epoch: [31] [2660/3634] eta: 0:08:03 lr: 0.000002 loss: 0.4879 (0.5157) time: 0.4855 data: 0.0002 max mem: 52126 +Epoch: [31] [2670/3634] eta: 0:07:58 lr: 0.000002 loss: 0.5193 (0.5157) time: 0.5454 data: 0.0002 max mem: 52126 +Epoch: [31] [2680/3634] eta: 0:07:53 lr: 0.000002 loss: 0.5304 (0.5157) time: 0.4862 data: 0.0002 max mem: 52126 +Epoch: [31] [2690/3634] eta: 0:07:49 lr: 0.000002 loss: 0.5124 (0.5157) time: 0.5289 data: 0.0002 max mem: 52126 +Epoch: [31] [2700/3634] eta: 0:07:45 lr: 0.000002 loss: 0.5567 (0.5159) time: 0.6960 data: 0.0002 max mem: 52126 +Epoch: [31] [2710/3634] eta: 0:07:39 lr: 0.000002 loss: 0.5567 (0.5159) time: 0.5964 data: 0.0002 max mem: 52126 +Epoch: [31] [2720/3634] eta: 0:07:34 lr: 0.000002 loss: 0.5285 (0.5160) time: 0.4752 data: 0.0002 max mem: 52126 +Epoch: [31] [2730/3634] eta: 0:07:29 lr: 0.000002 loss: 0.5407 (0.5160) time: 0.4273 data: 0.0002 max mem: 52126 +Epoch: [31] [2740/3634] eta: 0:07:24 lr: 0.000002 loss: 0.5378 (0.5161) time: 0.4420 data: 0.0002 max mem: 52126 +Epoch: [31] [2750/3634] eta: 0:07:19 lr: 0.000002 loss: 0.5282 (0.5161) time: 0.4865 data: 0.0002 max mem: 52126 +Epoch: [31] [2760/3634] eta: 0:07:14 lr: 0.000002 loss: 0.5282 (0.5161) time: 0.4277 data: 0.0002 max mem: 52126 +Epoch: [31] [2770/3634] eta: 0:07:09 lr: 0.000002 loss: 0.4966 (0.5160) time: 0.4276 data: 0.0002 max mem: 52126 +Epoch: [31] [2780/3634] eta: 0:07:03 lr: 0.000002 loss: 0.5035 (0.5160) time: 0.4261 data: 0.0002 max mem: 52126 +Epoch: [31] [2790/3634] eta: 0:06:58 lr: 0.000002 loss: 0.5331 (0.5160) time: 0.3815 data: 0.0002 max mem: 52126 +Epoch: [31] [2800/3634] eta: 0:06:53 lr: 0.000002 loss: 0.5216 (0.5160) time: 0.4275 data: 0.0002 max mem: 52126 +Epoch: [31] [2810/3634] eta: 0:06:48 lr: 0.000002 loss: 0.5191 (0.5160) time: 0.4754 data: 0.0002 max mem: 52126 +Epoch: [31] [2820/3634] eta: 0:06:43 lr: 0.000002 loss: 0.4910 (0.5160) time: 0.4880 data: 0.0001 max mem: 52126 +Epoch: [31] [2830/3634] eta: 0:06:39 lr: 0.000002 loss: 0.5037 (0.5160) time: 0.5901 data: 0.0001 max mem: 52126 +Epoch: [31] [2840/3634] eta: 0:06:34 lr: 0.000002 loss: 0.5084 (0.5160) time: 0.5797 data: 0.0001 max mem: 52126 +Epoch: [31] [2850/3634] eta: 0:06:29 lr: 0.000002 loss: 0.5146 (0.5160) time: 0.4940 data: 0.0001 max mem: 52126 +Epoch: [31] [2860/3634] eta: 0:06:23 lr: 0.000002 loss: 0.5246 (0.5161) time: 0.4462 data: 0.0002 max mem: 52126 +Epoch: [31] [2870/3634] eta: 0:06:18 lr: 0.000002 loss: 0.5246 (0.5161) time: 0.4254 data: 0.0002 max mem: 52126 +Epoch: [31] [2880/3634] eta: 0:06:13 lr: 0.000002 loss: 0.5254 (0.5161) time: 0.4855 data: 0.0002 max mem: 52126 +Epoch: [31] [2890/3634] eta: 0:06:08 lr: 0.000002 loss: 0.5254 (0.5160) time: 0.5039 data: 0.0002 max mem: 52126 +Epoch: [31] [2900/3634] eta: 0:06:03 lr: 0.000002 loss: 0.5205 (0.5160) time: 0.4886 data: 0.0002 max mem: 52126 +Epoch: [31] [2910/3634] eta: 0:05:58 lr: 0.000002 loss: 0.4922 (0.5158) time: 0.4749 data: 0.0001 max mem: 52126 +Epoch: [31] [2920/3634] eta: 0:05:53 lr: 0.000002 loss: 0.4756 (0.5158) time: 0.4746 data: 0.0001 max mem: 52126 +Epoch: [31] [2930/3634] eta: 0:05:48 lr: 0.000002 loss: 0.5310 (0.5158) time: 0.4251 data: 0.0002 max mem: 52126 +Epoch: [31] [2940/3634] eta: 0:05:43 lr: 0.000002 loss: 0.5084 (0.5157) time: 0.4400 data: 0.0002 max mem: 52126 +Epoch: [31] [2950/3634] eta: 0:05:39 lr: 0.000002 loss: 0.5028 (0.5157) time: 0.5467 data: 0.0002 max mem: 52126 +Epoch: [31] [2960/3634] eta: 0:05:34 lr: 0.000002 loss: 0.4874 (0.5157) time: 0.5451 data: 0.0002 max mem: 52126 +Epoch: [31] [2970/3634] eta: 0:05:28 lr: 0.000002 loss: 0.4795 (0.5156) time: 0.4394 data: 0.0002 max mem: 52126 +Epoch: [31] [2980/3634] eta: 0:05:23 lr: 0.000002 loss: 0.4809 (0.5156) time: 0.4417 data: 0.0002 max mem: 52126 +Epoch: [31] [2990/3634] eta: 0:05:19 lr: 0.000002 loss: 0.5048 (0.5157) time: 0.5317 data: 0.0001 max mem: 52126 +Epoch: [31] [3000/3634] eta: 0:05:14 lr: 0.000002 loss: 0.5191 (0.5157) time: 0.6245 data: 0.0001 max mem: 52126 +Epoch: [31] [3010/3634] eta: 0:05:09 lr: 0.000002 loss: 0.5195 (0.5158) time: 0.5831 data: 0.0002 max mem: 52126 +Epoch: [31] [3020/3634] eta: 0:05:04 lr: 0.000002 loss: 0.5402 (0.5157) time: 0.4735 data: 0.0001 max mem: 52126 +Epoch: [31] [3030/3634] eta: 0:04:59 lr: 0.000002 loss: 0.5075 (0.5157) time: 0.4273 data: 0.0001 max mem: 52126 +Epoch: [31] [3040/3634] eta: 0:04:54 lr: 0.000002 loss: 0.5117 (0.5158) time: 0.4271 data: 0.0002 max mem: 52126 +Epoch: [31] [3050/3634] eta: 0:04:49 lr: 0.000002 loss: 0.5555 (0.5159) time: 0.4740 data: 0.0002 max mem: 52126 +Epoch: [31] [3060/3634] eta: 0:04:44 lr: 0.000002 loss: 0.5455 (0.5159) time: 0.4309 data: 0.0002 max mem: 52126 +Epoch: [31] [3070/3634] eta: 0:04:39 lr: 0.000002 loss: 0.5136 (0.5160) time: 0.5453 data: 0.0002 max mem: 52126 +Epoch: [31] [3080/3634] eta: 0:04:34 lr: 0.000002 loss: 0.5363 (0.5160) time: 0.6486 data: 0.0001 max mem: 52126 +Epoch: [31] [3090/3634] eta: 0:04:30 lr: 0.000002 loss: 0.5470 (0.5161) time: 0.6629 data: 0.0001 max mem: 52126 +Epoch: [31] [3100/3634] eta: 0:04:25 lr: 0.000002 loss: 0.5434 (0.5163) time: 0.5592 data: 0.0002 max mem: 52126 +Epoch: [31] [3110/3634] eta: 0:04:20 lr: 0.000002 loss: 0.5374 (0.5162) time: 0.4894 data: 0.0002 max mem: 52126 +Epoch: [31] [3120/3634] eta: 0:04:15 lr: 0.000002 loss: 0.5222 (0.5163) time: 0.5344 data: 0.0002 max mem: 52126 +Epoch: [31] [3130/3634] eta: 0:04:10 lr: 0.000002 loss: 0.5240 (0.5163) time: 0.4267 data: 0.0002 max mem: 52126 +Epoch: [31] [3140/3634] eta: 0:04:05 lr: 0.000002 loss: 0.5359 (0.5164) time: 0.4260 data: 0.0002 max mem: 52126 +Epoch: [31] [3150/3634] eta: 0:04:00 lr: 0.000002 loss: 0.5359 (0.5164) time: 0.5284 data: 0.0002 max mem: 52126 +Epoch: [31] [3160/3634] eta: 0:03:55 lr: 0.000002 loss: 0.5139 (0.5163) time: 0.5916 data: 0.0002 max mem: 52126 +Epoch: [31] [3170/3634] eta: 0:03:50 lr: 0.000002 loss: 0.4907 (0.5164) time: 0.5350 data: 0.0002 max mem: 52126 +Epoch: [31] [3180/3634] eta: 0:03:45 lr: 0.000002 loss: 0.4843 (0.5162) time: 0.4274 data: 0.0002 max mem: 52126 +Epoch: [31] [3190/3634] eta: 0:03:40 lr: 0.000002 loss: 0.5099 (0.5163) time: 0.3821 data: 0.0002 max mem: 52126 +Epoch: [31] [3200/3634] eta: 0:03:35 lr: 0.000002 loss: 0.5524 (0.5164) time: 0.4723 data: 0.0002 max mem: 52126 +Epoch: [31] [3210/3634] eta: 0:03:30 lr: 0.000002 loss: 0.5455 (0.5164) time: 0.5218 data: 0.0002 max mem: 52126 +Epoch: [31] [3220/3634] eta: 0:03:25 lr: 0.000002 loss: 0.5068 (0.5163) time: 0.4770 data: 0.0002 max mem: 52126 +Epoch: [31] [3230/3634] eta: 0:03:20 lr: 0.000002 loss: 0.5186 (0.5164) time: 0.4759 data: 0.0002 max mem: 52126 +Epoch: [31] [3240/3634] eta: 0:03:15 lr: 0.000002 loss: 0.5198 (0.5164) time: 0.4305 data: 0.0002 max mem: 52126 +Epoch: [31] [3250/3634] eta: 0:03:10 lr: 0.000002 loss: 0.5358 (0.5165) time: 0.5016 data: 0.0002 max mem: 52126 +Epoch: [31] [3260/3634] eta: 0:03:05 lr: 0.000002 loss: 0.5414 (0.5165) time: 0.5018 data: 0.0002 max mem: 52126 +Epoch: [31] [3270/3634] eta: 0:03:00 lr: 0.000002 loss: 0.5338 (0.5166) time: 0.3829 data: 0.0002 max mem: 52126 +Epoch: [31] [3280/3634] eta: 0:02:55 lr: 0.000002 loss: 0.5385 (0.5167) time: 0.4455 data: 0.0002 max mem: 52126 +Epoch: [31] [3290/3634] eta: 0:02:50 lr: 0.000002 loss: 0.5275 (0.5167) time: 0.4924 data: 0.0002 max mem: 52126 +Epoch: [31] [3300/3634] eta: 0:02:45 lr: 0.000002 loss: 0.4839 (0.5166) time: 0.4925 data: 0.0001 max mem: 52126 +Epoch: [31] [3310/3634] eta: 0:02:40 lr: 0.000002 loss: 0.4818 (0.5166) time: 0.4453 data: 0.0002 max mem: 52126 +Epoch: [31] [3320/3634] eta: 0:02:35 lr: 0.000002 loss: 0.5477 (0.5167) time: 0.4286 data: 0.0002 max mem: 52126 +Epoch: [31] [3330/3634] eta: 0:02:30 lr: 0.000002 loss: 0.5351 (0.5166) time: 0.4281 data: 0.0002 max mem: 52126 +Epoch: [31] [3340/3634] eta: 0:02:25 lr: 0.000002 loss: 0.5328 (0.5167) time: 0.4423 data: 0.0002 max mem: 52126 +Epoch: [31] [3350/3634] eta: 0:02:20 lr: 0.000002 loss: 0.5326 (0.5167) time: 0.6285 data: 0.0002 max mem: 52126 +Epoch: [31] [3360/3634] eta: 0:02:16 lr: 0.000002 loss: 0.5175 (0.5165) time: 0.7827 data: 0.0002 max mem: 52126 +Epoch: [31] [3370/3634] eta: 0:02:11 lr: 0.000002 loss: 0.5397 (0.5167) time: 0.5966 data: 0.0001 max mem: 52126 +Epoch: [31] [3380/3634] eta: 0:02:06 lr: 0.000002 loss: 0.5326 (0.5165) time: 0.4409 data: 0.0001 max mem: 52126 +Epoch: [31] [3390/3634] eta: 0:02:01 lr: 0.000002 loss: 0.4817 (0.5165) time: 0.5033 data: 0.0001 max mem: 52126 +Epoch: [31] [3400/3634] eta: 0:01:56 lr: 0.000002 loss: 0.4866 (0.5165) time: 0.5055 data: 0.0002 max mem: 52126 +Epoch: [31] [3410/3634] eta: 0:01:51 lr: 0.000002 loss: 0.5140 (0.5165) time: 0.5487 data: 0.0002 max mem: 52126 +Epoch: [31] [3420/3634] eta: 0:01:46 lr: 0.000002 loss: 0.5001 (0.5164) time: 0.5469 data: 0.0002 max mem: 52126 +Epoch: [31] [3430/3634] eta: 0:01:41 lr: 0.000002 loss: 0.4645 (0.5164) time: 0.5317 data: 0.0001 max mem: 52126 +Epoch: [31] [3440/3634] eta: 0:01:36 lr: 0.000002 loss: 0.5081 (0.5163) time: 0.5196 data: 0.0002 max mem: 52126 +Epoch: [31] [3450/3634] eta: 0:01:31 lr: 0.000002 loss: 0.5202 (0.5163) time: 0.4739 data: 0.0002 max mem: 52126 +Epoch: [31] [3460/3634] eta: 0:01:26 lr: 0.000002 loss: 0.5202 (0.5163) time: 0.4732 data: 0.0002 max mem: 52126 +Epoch: [31] [3470/3634] eta: 0:01:21 lr: 0.000002 loss: 0.4940 (0.5162) time: 0.5336 data: 0.0002 max mem: 52126 +Epoch: [31] [3480/3634] eta: 0:01:16 lr: 0.000002 loss: 0.4722 (0.5162) time: 0.5979 data: 0.0002 max mem: 52126 +Epoch: [31] [3490/3634] eta: 0:01:11 lr: 0.000002 loss: 0.5210 (0.5162) time: 0.4935 data: 0.0002 max mem: 52126 +Epoch: [31] [3500/3634] eta: 0:01:06 lr: 0.000002 loss: 0.5365 (0.5162) time: 0.3830 data: 0.0002 max mem: 52126 +Epoch: [31] [3510/3634] eta: 0:01:01 lr: 0.000002 loss: 0.5249 (0.5162) time: 0.3819 data: 0.0002 max mem: 52126 +Epoch: [31] [3520/3634] eta: 0:00:56 lr: 0.000002 loss: 0.5249 (0.5162) time: 0.4270 data: 0.0002 max mem: 52126 +Epoch: [31] [3530/3634] eta: 0:00:51 lr: 0.000002 loss: 0.4862 (0.5162) time: 0.4276 data: 0.0002 max mem: 52126 +Epoch: [31] [3540/3634] eta: 0:00:46 lr: 0.000002 loss: 0.5178 (0.5162) time: 0.6041 data: 0.0002 max mem: 52126 +Epoch: [31] [3550/3634] eta: 0:00:41 lr: 0.000002 loss: 0.4941 (0.5161) time: 0.6639 data: 0.0002 max mem: 52126 +Epoch: [31] [3560/3634] eta: 0:00:36 lr: 0.000002 loss: 0.4802 (0.5160) time: 0.5370 data: 0.0002 max mem: 52126 +Epoch: [31] [3570/3634] eta: 0:00:31 lr: 0.000002 loss: 0.5319 (0.5161) time: 0.5684 data: 0.0002 max mem: 52126 +Epoch: [31] [3580/3634] eta: 0:00:26 lr: 0.000002 loss: 0.5406 (0.5161) time: 0.4732 data: 0.0002 max mem: 52126 +Epoch: [31] [3590/3634] eta: 0:00:21 lr: 0.000002 loss: 0.5039 (0.5160) time: 0.3821 data: 0.0002 max mem: 52126 +Epoch: [31] [3600/3634] eta: 0:00:16 lr: 0.000002 loss: 0.4910 (0.5160) time: 0.4319 data: 0.0002 max mem: 52126 +Epoch: [31] [3610/3634] eta: 0:00:11 lr: 0.000002 loss: 0.5625 (0.5161) time: 0.4764 data: 0.0002 max mem: 52126 +Epoch: [31] [3620/3634] eta: 0:00:06 lr: 0.000002 loss: 0.5367 (0.5160) time: 0.4877 data: 0.0033 max mem: 52126 +Epoch: [31] [3630/3634] eta: 0:00:01 lr: 0.000002 loss: 0.4665 (0.5159) time: 0.4855 data: 0.0033 max mem: 52126 +Epoch: [31] [3633/3634] eta: 0:00:00 lr: 0.000002 loss: 0.4486 (0.5159) time: 0.4856 data: 0.0032 max mem: 52126 +Epoch: [31] Total time: 0:30:04 (0.4967 s / it) +Averaged stats: lr: 0.000002 loss: 0.4486 (0.5145) +Test: [ 0/1403] eta: 1:10:32 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0796 (0.0796) acc1: 100.0000 (100.0000) acc5: 100.0000 (100.0000) time: 3.0171 data: 2.9188 max mem: 52126 +Test: [ 10/1403] eta: 0:08:23 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1336 (0.1629) acc1: 98.4375 (95.8807) acc5: 100.0000 (100.0000) time: 0.3616 data: 0.2656 max mem: 52126 +Test: [ 20/1403] eta: 0:05:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1237 (0.1450) acc1: 98.4375 (96.5774) acc5: 100.0000 (100.0000) time: 0.0959 data: 0.0003 max mem: 52126 +Test: [ 30/1403] eta: 0:04:21 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0857 (0.1348) acc1: 100.0000 (97.0262) acc5: 100.0000 (100.0000) time: 0.0959 data: 0.0003 max mem: 52126 +Test: [ 40/1403] eta: 0:03:47 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0945 (0.1385) acc1: 100.0000 (97.1799) acc5: 100.0000 (100.0000) time: 0.0959 data: 0.0003 max mem: 52126 +Test: [ 50/1403] eta: 0:03:27 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1159 (0.1565) acc1: 100.0000 (96.1703) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 60/1403] eta: 0:03:13 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1207 (0.1541) acc1: 100.0000 (96.4652) acc5: 100.0000 (100.0000) time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 70/1403] eta: 0:03:02 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.1303 (0.1559) acc1: 98.4375 (96.2808) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 80/1403] eta: 0:02:54 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1340 (0.1597) acc1: 96.8750 (96.0648) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 90/1403] eta: 0:02:48 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1489 (0.1745) acc1: 95.3125 (95.2782) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 100/1403] eta: 0:02:42 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1488 (0.1725) acc1: 96.8750 (95.4672) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 110/1403] eta: 0:02:38 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1300 (0.1708) acc1: 98.4375 (95.5377) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 120/1403] eta: 0:02:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0749 (0.1629) acc1: 100.0000 (95.8936) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 130/1403] eta: 0:02:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0749 (0.1573) acc1: 100.0000 (96.1355) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 140/1403] eta: 0:02:27 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0723 (0.1512) acc1: 100.0000 (96.4096) acc5: 100.0000 (100.0000) time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 150/1403] eta: 0:02:24 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0697 (0.1462) acc1: 100.0000 (96.6267) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 160/1403] eta: 0:02:21 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0727 (0.1420) acc1: 100.0000 (96.7974) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 170/1403] eta: 0:02:19 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0738 (0.1387) acc1: 100.0000 (96.9664) acc5: 100.0000 (100.0000) time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 180/1403] eta: 0:02:17 acc1_pos: 89.062 acc5_pos: 100.000 loss: 0.0798 (0.1427) acc1: 100.0000 (96.7973) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 190/1403] eta: 0:02:15 acc1_pos: 90.625 acc5_pos: 100.000 loss: 0.2544 (0.1506) acc1: 93.7500 (96.3351) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 200/1403] eta: 0:02:13 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.2544 (0.1621) acc1: 90.6250 (95.7945) acc5: 100.0000 (100.0000) time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 210/1403] eta: 0:02:11 acc1_pos: 81.250 acc5_pos: 100.000 loss: 0.3095 (0.1724) acc1: 87.5000 (95.2533) acc5: 100.0000 (100.0000) time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 220/1403] eta: 0:02:09 acc1_pos: 87.500 acc5_pos: 100.000 loss: 0.3239 (0.1814) acc1: 87.5000 (94.8388) acc5: 100.0000 (100.0000) time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 230/1403] eta: 0:02:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2550 (0.1777) acc1: 90.6250 (95.0149) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 240/1403] eta: 0:02:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0741 (0.1736) acc1: 100.0000 (95.2088) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 250/1403] eta: 0:02:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0741 (0.1699) acc1: 100.0000 (95.3810) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 260/1403] eta: 0:02:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0759 (0.1665) acc1: 100.0000 (95.5400) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 270/1403] eta: 0:02:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0763 (0.1634) acc1: 100.0000 (95.6873) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 280/1403] eta: 0:01:59 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0754 (0.1603) acc1: 100.0000 (95.8352) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 290/1403] eta: 0:01:58 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0722 (0.1573) acc1: 100.0000 (95.9729) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 300/1403] eta: 0:01:56 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0704 (0.1545) acc1: 100.0000 (96.1015) acc5: 100.0000 (100.0000) time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 310/1403] eta: 0:01:55 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0746 (0.1522) acc1: 100.0000 (96.2269) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [ 320/1403] eta: 0:01:53 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0971 (0.1506) acc1: 100.0000 (96.3444) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 330/1403] eta: 0:01:52 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1039 (0.1493) acc1: 100.0000 (96.4549) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 340/1403] eta: 0:01:51 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1034 (0.1479) acc1: 100.0000 (96.5543) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 350/1403] eta: 0:01:49 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1016 (0.1466) acc1: 100.0000 (96.6524) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 360/1403] eta: 0:01:48 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1045 (0.1456) acc1: 100.0000 (96.7408) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 370/1403] eta: 0:01:47 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1035 (0.1444) acc1: 100.0000 (96.8287) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 380/1403] eta: 0:01:46 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1025 (0.1433) acc1: 100.0000 (96.9119) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 390/1403] eta: 0:01:44 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1030 (0.1443) acc1: 100.0000 (96.8430) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 400/1403] eta: 0:01:43 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2458 (0.1481) acc1: 90.6250 (96.6451) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 410/1403] eta: 0:01:42 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2747 (0.1512) acc1: 89.0625 (96.4720) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [ 420/1403] eta: 0:01:41 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2472 (0.1529) acc1: 90.6250 (96.3777) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 430/1403] eta: 0:01:40 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2304 (0.1558) acc1: 90.6250 (96.2079) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 440/1403] eta: 0:01:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2845 (0.1593) acc1: 89.0625 (96.0353) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 450/1403] eta: 0:01:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2718 (0.1617) acc1: 89.0625 (95.9084) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 460/1403] eta: 0:01:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2401 (0.1642) acc1: 89.0625 (95.7463) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 470/1403] eta: 0:01:35 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2160 (0.1635) acc1: 92.1875 (95.7703) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 480/1403] eta: 0:01:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0632 (0.1614) acc1: 100.0000 (95.8582) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 490/1403] eta: 0:01:33 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0632 (0.1594) acc1: 100.0000 (95.9426) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 500/1403] eta: 0:01:32 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0634 (0.1575) acc1: 100.0000 (96.0236) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 510/1403] eta: 0:01:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0634 (0.1557) acc1: 100.0000 (96.1014) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 520/1403] eta: 0:01:29 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0672 (0.1541) acc1: 100.0000 (96.1762) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 530/1403] eta: 0:01:28 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0597 (0.1523) acc1: 100.0000 (96.2482) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [ 540/1403] eta: 0:01:27 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0592 (0.1506) acc1: 100.0000 (96.3176) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [ 550/1403] eta: 0:01:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0600 (0.1494) acc1: 100.0000 (96.3674) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 560/1403] eta: 0:01:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0825 (0.1484) acc1: 98.4375 (96.4099) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 570/1403] eta: 0:01:24 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0850 (0.1474) acc1: 98.4375 (96.4536) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 580/1403] eta: 0:01:23 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0970 (0.1470) acc1: 98.4375 (96.4635) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 590/1403] eta: 0:01:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1140 (0.1468) acc1: 96.8750 (96.4626) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 600/1403] eta: 0:01:21 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1366 (0.1472) acc1: 95.3125 (96.4096) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 610/1403] eta: 0:01:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1371 (0.1469) acc1: 95.3125 (96.4121) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 620/1403] eta: 0:01:18 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1508 (0.1481) acc1: 95.3125 (96.3064) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 630/1403] eta: 0:01:17 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1511 (0.1480) acc1: 95.3125 (96.3005) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 640/1403] eta: 0:01:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1267 (0.1476) acc1: 96.8750 (96.3192) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 650/1403] eta: 0:01:15 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1267 (0.1473) acc1: 96.8750 (96.3398) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 660/1403] eta: 0:01:14 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0952 (0.1463) acc1: 100.0000 (96.3833) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 670/1403] eta: 0:01:13 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0690 (0.1452) acc1: 100.0000 (96.4349) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 680/1403] eta: 0:01:12 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0793 (0.1469) acc1: 98.4375 (96.3289) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 690/1403] eta: 0:01:11 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2898 (0.1514) acc1: 87.5000 (96.0700) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 700/1403] eta: 0:01:10 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1905 (0.1517) acc1: 89.0625 (96.0525) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 710/1403] eta: 0:01:09 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3227 (0.1565) acc1: 85.9375 (95.7982) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 720/1403] eta: 0:01:08 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3227 (0.1565) acc1: 82.8125 (95.7936) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 730/1403] eta: 0:01:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0679 (0.1555) acc1: 100.0000 (95.8405) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 740/1403] eta: 0:01:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0818 (0.1546) acc1: 100.0000 (95.8776) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [ 750/1403] eta: 0:01:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1034 (0.1564) acc1: 96.8750 (95.7744) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52126 +Test: [ 760/1403] eta: 0:01:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3476 (0.1604) acc1: 82.8125 (95.5260) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52126 +Test: [ 770/1403] eta: 0:01:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5302 (0.1658) acc1: 67.1875 (95.1929) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [ 780/1403] eta: 0:01:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5181 (0.1690) acc1: 73.4375 (95.0024) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 790/1403] eta: 0:01:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3230 (0.1718) acc1: 81.2500 (94.8700) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 800/1403] eta: 0:01:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2268 (0.1724) acc1: 90.6250 (94.8424) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 810/1403] eta: 0:00:59 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2256 (0.1755) acc1: 89.0625 (94.6998) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 820/1403] eta: 0:00:58 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6503 (0.1846) acc1: 71.8750 (94.3076) acc5: 100.0000 (100.0000) acc1_neg: 40.625 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [ 830/1403] eta: 0:00:57 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.7335 (0.1876) acc1: 70.3125 (94.1655) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 840/1403] eta: 0:00:56 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5344 (0.1959) acc1: 70.3125 (93.8095) acc5: 100.0000 (100.0000) acc1_neg: 50.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 850/1403] eta: 0:00:55 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.0776 (0.2072) acc1: 53.1250 (93.3222) acc5: 100.0000 (100.0000) acc1_neg: 48.438 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 860/1403] eta: 0:00:54 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.0132 (0.2108) acc1: 56.2500 (93.1747) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 870/1403] eta: 0:00:53 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1339 (0.2097) acc1: 96.8750 (93.2208) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 880/1403] eta: 0:00:52 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1031 (0.2085) acc1: 98.4375 (93.2729) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 890/1403] eta: 0:00:51 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0837 (0.2071) acc1: 98.4375 (93.3379) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 900/1403] eta: 0:00:50 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0734 (0.2056) acc1: 100.0000 (93.4084) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 910/1403] eta: 0:00:49 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0755 (0.2047) acc1: 100.0000 (93.4464) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [ 920/1403] eta: 0:00:48 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1044 (0.2037) acc1: 98.4375 (93.4921) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [ 930/1403] eta: 0:00:47 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1010 (0.2027) acc1: 98.4375 (93.5436) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 940/1403] eta: 0:00:45 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1165 (0.2028) acc1: 96.8750 (93.5175) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [ 950/1403] eta: 0:00:44 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2557 (0.2037) acc1: 89.0625 (93.4575) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [ 960/1403] eta: 0:00:43 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2766 (0.2043) acc1: 89.0625 (93.4167) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 970/1403] eta: 0:00:42 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2325 (0.2052) acc1: 90.6250 (93.3654) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [ 980/1403] eta: 0:00:41 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2566 (0.2060) acc1: 89.0625 (93.3056) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52126 +Test: [ 990/1403] eta: 0:00:40 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2531 (0.2064) acc1: 89.0625 (93.2817) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52126 +Test: [1000/1403] eta: 0:00:39 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2378 (0.2071) acc1: 90.6250 (93.2427) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [1010/1403] eta: 0:00:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2575 (0.2079) acc1: 89.0625 (93.1967) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [1020/1403] eta: 0:00:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4072 (0.2142) acc1: 81.2500 (92.8593) acc5: 100.0000 (100.0000) acc1_neg: 34.375 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [1030/1403] eta: 0:00:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5877 (0.2162) acc1: 62.5000 (92.7346) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [1040/1403] eta: 0:00:35 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2216 (0.2159) acc1: 90.6250 (92.7384) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [1050/1403] eta: 0:00:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1573 (0.2156) acc1: 93.7500 (92.7554) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52126 +Test: [1060/1403] eta: 0:00:33 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1349 (0.2148) acc1: 96.8750 (92.7928) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52126 +Test: [1070/1403] eta: 0:00:32 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1415 (0.2152) acc1: 96.8750 (92.7754) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [1080/1403] eta: 0:00:31 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3626 (0.2169) acc1: 85.9375 (92.6876) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [1090/1403] eta: 0:00:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3924 (0.2186) acc1: 82.8125 (92.5957) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [1100/1403] eta: 0:00:29 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3439 (0.2190) acc1: 85.9375 (92.5636) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52126 +Test: [1110/1403] eta: 0:00:28 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2662 (0.2194) acc1: 89.0625 (92.5433) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52126 +Test: [1120/1403] eta: 0:00:27 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2662 (0.2199) acc1: 90.6250 (92.5206) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52126 +Test: [1130/1403] eta: 0:00:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2758 (0.2206) acc1: 89.0625 (92.4818) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [1140/1403] eta: 0:00:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2404 (0.2207) acc1: 90.6250 (92.4682) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [1150/1403] eta: 0:00:24 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2265 (0.2210) acc1: 90.6250 (92.4481) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [1160/1403] eta: 0:00:24 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2513 (0.2215) acc1: 90.6250 (92.4257) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [1170/1403] eta: 0:00:23 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2405 (0.2213) acc1: 90.6250 (92.4357) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [1180/1403] eta: 0:00:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0682 (0.2200) acc1: 100.0000 (92.4997) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [1190/1403] eta: 0:00:21 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0678 (0.2187) acc1: 100.0000 (92.5627) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [1200/1403] eta: 0:00:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0679 (0.2174) acc1: 100.0000 (92.6246) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [1210/1403] eta: 0:00:19 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0671 (0.2162) acc1: 100.0000 (92.6855) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [1220/1403] eta: 0:00:18 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0670 (0.2150) acc1: 100.0000 (92.7454) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [1230/1403] eta: 0:00:17 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0679 (0.2138) acc1: 100.0000 (92.8044) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0003 max mem: 52126 +Test: [1240/1403] eta: 0:00:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0677 (0.2126) acc1: 100.0000 (92.8624) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0002 max mem: 52126 +Test: [1250/1403] eta: 0:00:15 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0680 (0.2115) acc1: 100.0000 (92.9194) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0002 max mem: 52126 +Test: [1260/1403] eta: 0:00:14 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0732 (0.2104) acc1: 100.0000 (92.9756) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0002 max mem: 52126 +Test: [1270/1403] eta: 0:00:13 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0747 (0.2093) acc1: 100.0000 (93.0308) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0003 max mem: 52126 +Test: [1280/1403] eta: 0:00:12 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0743 (0.2083) acc1: 100.0000 (93.0852) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [1290/1403] eta: 0:00:11 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0737 (0.2072) acc1: 100.0000 (93.1388) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [1300/1403] eta: 0:00:10 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0744 (0.2062) acc1: 100.0000 (93.1915) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [1310/1403] eta: 0:00:09 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0752 (0.2052) acc1: 100.0000 (93.2435) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [1320/1403] eta: 0:00:08 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0732 (0.2042) acc1: 100.0000 (93.2946) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [1330/1403] eta: 0:00:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0731 (0.2032) acc1: 100.0000 (93.3438) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [1340/1403] eta: 0:00:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0672 (0.2022) acc1: 100.0000 (93.3911) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [1350/1403] eta: 0:00:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0658 (0.2013) acc1: 100.0000 (93.4366) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [1360/1403] eta: 0:00:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0660 (0.2003) acc1: 100.0000 (93.4837) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0002 max mem: 52126 +Test: [1370/1403] eta: 0:00:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0660 (0.1993) acc1: 100.0000 (93.5289) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0002 max mem: 52126 +Test: [1380/1403] eta: 0:00:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0653 (0.1984) acc1: 100.0000 (93.5758) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0002 max mem: 52126 +Test: [1390/1403] eta: 0:00:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0651 (0.1974) acc1: 100.0000 (93.6219) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0960 data: 0.0002 max mem: 52126 +Test: [1400/1403] eta: 0:00:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0662 (0.1966) acc1: 100.0000 (93.6652) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0959 data: 0.0001 max mem: 52126 +Test: [1402/1403] eta: 0:00:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0662 (0.1964) acc1: 100.0000 (93.6698) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0918 data: 0.0001 max mem: 52126 +Test: Total time: 0:02:18 (0.0987 s / it) +* Overall Acc@1 93.670 Acc@5 100.000 + target == real Acc@1 96.193 Acc@5 100.000 + target == fake Acc@1 92.959 Acc@5 100.000 + loss 0.196 +Accuracy of the network on the 89729 test images: 93.7% +Max accuracy: 94.13% +Epoch: [32] [ 0/3634] eta: 6:24:47 lr: 0.000002 loss: 0.6189 (0.6189) time: 6.3532 data: 4.4162 max mem: 52126 +Epoch: [32] [ 10/3634] eta: 1:07:35 lr: 0.000002 loss: 0.5370 (0.5510) time: 1.1189 data: 0.4016 max mem: 52126 +Epoch: [32] [ 20/3634] eta: 0:46:17 lr: 0.000002 loss: 0.5358 (0.5394) time: 0.4892 data: 0.0001 max mem: 52126 +Epoch: [32] [ 30/3634] eta: 0:41:02 lr: 0.000002 loss: 0.5470 (0.5451) time: 0.4435 data: 0.0001 max mem: 52126 +Epoch: [32] [ 40/3634] eta: 0:39:38 lr: 0.000002 loss: 0.5312 (0.5322) time: 0.5498 data: 0.0002 max mem: 52126 +Epoch: [32] [ 50/3634] eta: 0:37:41 lr: 0.000002 loss: 0.4855 (0.5245) time: 0.5502 data: 0.0002 max mem: 52126 +Epoch: [32] [ 60/3634] eta: 0:36:03 lr: 0.000002 loss: 0.4951 (0.5235) time: 0.4895 data: 0.0002 max mem: 52126 +Epoch: [32] [ 70/3634] eta: 0:34:04 lr: 0.000002 loss: 0.5681 (0.5257) time: 0.4276 data: 0.0002 max mem: 52126 +Epoch: [32] [ 80/3634] eta: 0:32:35 lr: 0.000002 loss: 0.5538 (0.5283) time: 0.3818 data: 0.0002 max mem: 52126 +Epoch: [32] [ 90/3634] eta: 0:32:01 lr: 0.000002 loss: 0.5382 (0.5297) time: 0.4306 data: 0.0002 max mem: 52126 +Epoch: [32] [ 100/3634] eta: 0:31:02 lr: 0.000002 loss: 0.5066 (0.5241) time: 0.4329 data: 0.0002 max mem: 52126 +Epoch: [32] [ 110/3634] eta: 0:31:58 lr: 0.000002 loss: 0.5131 (0.5261) time: 0.5538 data: 0.0002 max mem: 52126 +Epoch: [32] [ 120/3634] eta: 0:31:34 lr: 0.000002 loss: 0.5489 (0.5284) time: 0.6003 data: 0.0002 max mem: 52126 +Epoch: [32] [ 130/3634] eta: 0:30:47 lr: 0.000002 loss: 0.5451 (0.5282) time: 0.4316 data: 0.0002 max mem: 52126 +Epoch: [32] [ 140/3634] eta: 0:30:30 lr: 0.000002 loss: 0.5393 (0.5273) time: 0.4321 data: 0.0002 max mem: 52126 +Epoch: [32] [ 150/3634] eta: 0:29:52 lr: 0.000002 loss: 0.5653 (0.5300) time: 0.4318 data: 0.0002 max mem: 52126 +Epoch: [32] [ 160/3634] eta: 0:29:18 lr: 0.000002 loss: 0.5245 (0.5280) time: 0.3810 data: 0.0002 max mem: 52126 +Epoch: [32] [ 170/3634] eta: 0:29:31 lr: 0.000002 loss: 0.5069 (0.5259) time: 0.4889 data: 0.0002 max mem: 52126 +Epoch: [32] [ 180/3634] eta: 0:29:25 lr: 0.000002 loss: 0.4711 (0.5247) time: 0.5520 data: 0.0002 max mem: 52126 +Epoch: [32] [ 190/3634] eta: 0:29:13 lr: 0.000002 loss: 0.5148 (0.5234) time: 0.4892 data: 0.0002 max mem: 52126 +Epoch: [32] [ 200/3634] eta: 0:29:44 lr: 0.000002 loss: 0.5181 (0.5233) time: 0.5961 data: 0.0002 max mem: 52126 +Epoch: [32] [ 210/3634] eta: 0:29:32 lr: 0.000002 loss: 0.5358 (0.5238) time: 0.5984 data: 0.0002 max mem: 52126 +Epoch: [32] [ 220/3634] eta: 0:29:25 lr: 0.000002 loss: 0.5322 (0.5227) time: 0.4919 data: 0.0002 max mem: 52126 +Epoch: [32] [ 230/3634] eta: 0:29:00 lr: 0.000002 loss: 0.5215 (0.5228) time: 0.4440 data: 0.0002 max mem: 52126 +Epoch: [32] [ 240/3634] eta: 0:28:37 lr: 0.000002 loss: 0.5281 (0.5229) time: 0.3836 data: 0.0002 max mem: 52126 +Epoch: [32] [ 250/3634] eta: 0:28:41 lr: 0.000002 loss: 0.5008 (0.5217) time: 0.4779 data: 0.0002 max mem: 52126 +Epoch: [32] [ 260/3634] eta: 0:28:55 lr: 0.000002 loss: 0.4945 (0.5208) time: 0.6150 data: 0.0002 max mem: 52126 +Epoch: [32] [ 270/3634] eta: 0:28:33 lr: 0.000002 loss: 0.4903 (0.5200) time: 0.5206 data: 0.0002 max mem: 52126 +Epoch: [32] [ 280/3634] eta: 0:28:39 lr: 0.000002 loss: 0.4627 (0.5189) time: 0.4886 data: 0.0002 max mem: 52126 +Epoch: [32] [ 290/3634] eta: 0:28:18 lr: 0.000002 loss: 0.4768 (0.5179) time: 0.4888 data: 0.0002 max mem: 52126 +Epoch: [32] [ 300/3634] eta: 0:28:13 lr: 0.000002 loss: 0.4768 (0.5175) time: 0.4432 data: 0.0002 max mem: 52126 +Epoch: [32] [ 310/3634] eta: 0:28:15 lr: 0.000002 loss: 0.4935 (0.5172) time: 0.5386 data: 0.0002 max mem: 52126 +Epoch: [32] [ 320/3634] eta: 0:27:56 lr: 0.000002 loss: 0.5472 (0.5187) time: 0.4770 data: 0.0002 max mem: 52126 +Epoch: [32] [ 330/3634] eta: 0:27:49 lr: 0.000002 loss: 0.5499 (0.5188) time: 0.4311 data: 0.0002 max mem: 52126 +Epoch: [32] [ 340/3634] eta: 0:27:41 lr: 0.000002 loss: 0.5464 (0.5197) time: 0.4787 data: 0.0001 max mem: 52126 +Epoch: [32] [ 350/3634] eta: 0:27:47 lr: 0.000002 loss: 0.5306 (0.5197) time: 0.5494 data: 0.0002 max mem: 52126 +Epoch: [32] [ 360/3634] eta: 0:27:42 lr: 0.000002 loss: 0.5138 (0.5195) time: 0.5628 data: 0.0002 max mem: 52126 +Epoch: [32] [ 370/3634] eta: 0:27:45 lr: 0.000002 loss: 0.5138 (0.5188) time: 0.5524 data: 0.0001 max mem: 52126 +Epoch: [32] [ 380/3634] eta: 0:28:06 lr: 0.000002 loss: 0.5125 (0.5189) time: 0.7095 data: 0.0001 max mem: 52126 +Epoch: [32] [ 390/3634] eta: 0:27:49 lr: 0.000002 loss: 0.5079 (0.5187) time: 0.5995 data: 0.0001 max mem: 52126 +Epoch: [32] [ 400/3634] eta: 0:27:43 lr: 0.000002 loss: 0.5236 (0.5188) time: 0.4430 data: 0.0001 max mem: 52126 +Epoch: [32] [ 410/3634] eta: 0:27:28 lr: 0.000002 loss: 0.5623 (0.5198) time: 0.4428 data: 0.0002 max mem: 52126 +Epoch: [32] [ 420/3634] eta: 0:27:32 lr: 0.000002 loss: 0.5619 (0.5201) time: 0.5050 data: 0.0002 max mem: 52126 +Epoch: [32] [ 430/3634] eta: 0:27:17 lr: 0.000002 loss: 0.5295 (0.5202) time: 0.5050 data: 0.0002 max mem: 52126 +Epoch: [32] [ 440/3634] eta: 0:27:02 lr: 0.000002 loss: 0.5067 (0.5188) time: 0.3810 data: 0.0002 max mem: 52126 +Epoch: [32] [ 450/3634] eta: 0:26:57 lr: 0.000002 loss: 0.4875 (0.5189) time: 0.4443 data: 0.0001 max mem: 52126 +Epoch: [32] [ 460/3634] eta: 0:27:00 lr: 0.000002 loss: 0.5250 (0.5191) time: 0.5689 data: 0.0002 max mem: 52126 +Epoch: [32] [ 470/3634] eta: 0:26:53 lr: 0.000002 loss: 0.5088 (0.5183) time: 0.5525 data: 0.0002 max mem: 52126 +Epoch: [32] [ 480/3634] eta: 0:27:01 lr: 0.000002 loss: 0.5087 (0.5181) time: 0.5932 data: 0.0001 max mem: 52126 +Epoch: [32] [ 490/3634] eta: 0:26:47 lr: 0.000002 loss: 0.5173 (0.5181) time: 0.5470 data: 0.0002 max mem: 52126 +Epoch: [32] [ 500/3634] eta: 0:26:49 lr: 0.000002 loss: 0.5173 (0.5183) time: 0.5005 data: 0.0002 max mem: 52126 +Epoch: [32] [ 510/3634] eta: 0:26:42 lr: 0.000002 loss: 0.5269 (0.5179) time: 0.5480 data: 0.0002 max mem: 52126 +Epoch: [32] [ 520/3634] eta: 0:26:48 lr: 0.000002 loss: 0.5037 (0.5177) time: 0.5953 data: 0.0001 max mem: 52126 +Epoch: [32] [ 530/3634] eta: 0:26:48 lr: 0.000002 loss: 0.5287 (0.5181) time: 0.6533 data: 0.0001 max mem: 52126 +Epoch: [32] [ 540/3634] eta: 0:26:42 lr: 0.000002 loss: 0.5675 (0.5192) time: 0.5490 data: 0.0001 max mem: 52126 +Epoch: [32] [ 550/3634] eta: 0:26:29 lr: 0.000002 loss: 0.5289 (0.5186) time: 0.4425 data: 0.0002 max mem: 52126 +Epoch: [32] [ 560/3634] eta: 0:26:21 lr: 0.000002 loss: 0.4962 (0.5193) time: 0.4249 data: 0.0002 max mem: 52126 +Epoch: [32] [ 570/3634] eta: 0:26:09 lr: 0.000002 loss: 0.5287 (0.5191) time: 0.4250 data: 0.0002 max mem: 52126 +Epoch: [32] [ 580/3634] eta: 0:26:03 lr: 0.000002 loss: 0.5206 (0.5197) time: 0.4401 data: 0.0001 max mem: 52126 +Epoch: [32] [ 590/3634] eta: 0:25:58 lr: 0.000002 loss: 0.5365 (0.5196) time: 0.5020 data: 0.0002 max mem: 52126 +Epoch: [32] [ 600/3634] eta: 0:25:51 lr: 0.000002 loss: 0.5365 (0.5200) time: 0.4888 data: 0.0002 max mem: 52126 +Epoch: [32] [ 610/3634] eta: 0:25:39 lr: 0.000002 loss: 0.5416 (0.5200) time: 0.4265 data: 0.0002 max mem: 52126 +Epoch: [32] [ 620/3634] eta: 0:25:32 lr: 0.000002 loss: 0.5300 (0.5202) time: 0.4256 data: 0.0001 max mem: 52126 +Epoch: [32] [ 630/3634] eta: 0:25:30 lr: 0.000002 loss: 0.5267 (0.5202) time: 0.5182 data: 0.0001 max mem: 52126 +Epoch: [32] [ 640/3634] eta: 0:25:19 lr: 0.000002 loss: 0.4925 (0.5201) time: 0.4734 data: 0.0002 max mem: 52126 +Epoch: [32] [ 650/3634] eta: 0:25:08 lr: 0.000002 loss: 0.4925 (0.5203) time: 0.3816 data: 0.0002 max mem: 52126 +Epoch: [32] [ 660/3634] eta: 0:25:08 lr: 0.000002 loss: 0.5149 (0.5199) time: 0.4995 data: 0.0001 max mem: 52126 +Epoch: [32] [ 670/3634] eta: 0:25:01 lr: 0.000002 loss: 0.5149 (0.5198) time: 0.5449 data: 0.0002 max mem: 52126 +Epoch: [32] [ 680/3634] eta: 0:24:51 lr: 0.000002 loss: 0.5178 (0.5199) time: 0.4271 data: 0.0002 max mem: 52126 +Epoch: [32] [ 690/3634] eta: 0:24:44 lr: 0.000002 loss: 0.5140 (0.5198) time: 0.4269 data: 0.0001 max mem: 52126 +Epoch: [32] [ 700/3634] eta: 0:24:38 lr: 0.000002 loss: 0.5141 (0.5196) time: 0.4709 data: 0.0002 max mem: 52126 +Epoch: [32] [ 710/3634] eta: 0:24:28 lr: 0.000002 loss: 0.4586 (0.5188) time: 0.4278 data: 0.0002 max mem: 52126 +Epoch: [32] [ 720/3634] eta: 0:24:23 lr: 0.000002 loss: 0.4620 (0.5188) time: 0.4434 data: 0.0002 max mem: 52126 +Epoch: [32] [ 730/3634] eta: 0:24:13 lr: 0.000002 loss: 0.4837 (0.5181) time: 0.4414 data: 0.0002 max mem: 52126 +Epoch: [32] [ 740/3634] eta: 0:24:03 lr: 0.000002 loss: 0.5135 (0.5182) time: 0.3819 data: 0.0002 max mem: 52126 +Epoch: [32] [ 750/3634] eta: 0:24:05 lr: 0.000002 loss: 0.5166 (0.5182) time: 0.5303 data: 0.0002 max mem: 52126 +Epoch: [32] [ 760/3634] eta: 0:24:05 lr: 0.000002 loss: 0.5113 (0.5182) time: 0.6500 data: 0.0002 max mem: 52126 +Epoch: [32] [ 770/3634] eta: 0:23:58 lr: 0.000002 loss: 0.5433 (0.5188) time: 0.5449 data: 0.0001 max mem: 52126 +Epoch: [32] [ 780/3634] eta: 0:23:52 lr: 0.000002 loss: 0.5481 (0.5190) time: 0.4708 data: 0.0001 max mem: 52126 +Epoch: [32] [ 790/3634] eta: 0:23:47 lr: 0.000002 loss: 0.5251 (0.5188) time: 0.4854 data: 0.0002 max mem: 52126 +Epoch: [32] [ 800/3634] eta: 0:23:41 lr: 0.000002 loss: 0.5469 (0.5191) time: 0.4839 data: 0.0001 max mem: 52126 +Epoch: [32] [ 810/3634] eta: 0:23:36 lr: 0.000002 loss: 0.5469 (0.5193) time: 0.4847 data: 0.0001 max mem: 52126 +Epoch: [32] [ 820/3634] eta: 0:23:27 lr: 0.000002 loss: 0.5320 (0.5193) time: 0.4411 data: 0.0001 max mem: 52126 +Epoch: [32] [ 830/3634] eta: 0:23:25 lr: 0.000002 loss: 0.5545 (0.5196) time: 0.4893 data: 0.0002 max mem: 52126 +Epoch: [32] [ 840/3634] eta: 0:23:20 lr: 0.000002 loss: 0.5577 (0.5198) time: 0.5485 data: 0.0002 max mem: 52126 +Epoch: [32] [ 850/3634] eta: 0:23:11 lr: 0.000002 loss: 0.5460 (0.5196) time: 0.4413 data: 0.0002 max mem: 52126 +Epoch: [32] [ 860/3634] eta: 0:23:06 lr: 0.000002 loss: 0.5187 (0.5197) time: 0.4422 data: 0.0002 max mem: 52126 +Epoch: [32] [ 870/3634] eta: 0:23:07 lr: 0.000002 loss: 0.5212 (0.5196) time: 0.5879 data: 0.0001 max mem: 52126 +Epoch: [32] [ 880/3634] eta: 0:23:01 lr: 0.000002 loss: 0.5212 (0.5195) time: 0.5715 data: 0.0002 max mem: 52126 +Epoch: [32] [ 890/3634] eta: 0:22:56 lr: 0.000002 loss: 0.5371 (0.5197) time: 0.4848 data: 0.0002 max mem: 52126 +Epoch: [32] [ 900/3634] eta: 0:22:51 lr: 0.000002 loss: 0.5458 (0.5199) time: 0.4998 data: 0.0002 max mem: 52126 +Epoch: [32] [ 910/3634] eta: 0:22:45 lr: 0.000002 loss: 0.5441 (0.5201) time: 0.4864 data: 0.0002 max mem: 52126 +Epoch: [32] [ 920/3634] eta: 0:22:42 lr: 0.000002 loss: 0.5365 (0.5200) time: 0.5309 data: 0.0002 max mem: 52126 +Epoch: [32] [ 930/3634] eta: 0:22:37 lr: 0.000002 loss: 0.5352 (0.5199) time: 0.5344 data: 0.0001 max mem: 52126 +Epoch: [32] [ 940/3634] eta: 0:22:31 lr: 0.000002 loss: 0.5190 (0.5195) time: 0.4751 data: 0.0002 max mem: 52126 +Epoch: [32] [ 950/3634] eta: 0:22:29 lr: 0.000002 loss: 0.4773 (0.5193) time: 0.5463 data: 0.0002 max mem: 52126 +Epoch: [32] [ 960/3634] eta: 0:22:21 lr: 0.000002 loss: 0.5060 (0.5192) time: 0.5026 data: 0.0002 max mem: 52126 +Epoch: [32] [ 970/3634] eta: 0:22:18 lr: 0.000002 loss: 0.5228 (0.5191) time: 0.4872 data: 0.0002 max mem: 52126 +Epoch: [32] [ 980/3634] eta: 0:22:16 lr: 0.000002 loss: 0.5171 (0.5189) time: 0.6041 data: 0.0002 max mem: 52126 +Epoch: [32] [ 990/3634] eta: 0:22:10 lr: 0.000002 loss: 0.5171 (0.5189) time: 0.5441 data: 0.0002 max mem: 52126 +Epoch: [32] [1000/3634] eta: 0:22:11 lr: 0.000002 loss: 0.5176 (0.5187) time: 0.5886 data: 0.0002 max mem: 52126 +Epoch: [32] [1010/3634] eta: 0:22:05 lr: 0.000002 loss: 0.5012 (0.5186) time: 0.6027 data: 0.0002 max mem: 52126 +Epoch: [32] [1020/3634] eta: 0:21:57 lr: 0.000002 loss: 0.4780 (0.5181) time: 0.4404 data: 0.0002 max mem: 52126 +Epoch: [32] [1030/3634] eta: 0:21:51 lr: 0.000002 loss: 0.5051 (0.5182) time: 0.4262 data: 0.0002 max mem: 52126 +Epoch: [32] [1040/3634] eta: 0:21:46 lr: 0.000002 loss: 0.5487 (0.5186) time: 0.4850 data: 0.0002 max mem: 52126 +Epoch: [32] [1050/3634] eta: 0:21:41 lr: 0.000002 loss: 0.5327 (0.5182) time: 0.4908 data: 0.0001 max mem: 52126 +Epoch: [32] [1060/3634] eta: 0:21:33 lr: 0.000002 loss: 0.5164 (0.5182) time: 0.4323 data: 0.0001 max mem: 52126 +Epoch: [32] [1070/3634] eta: 0:21:25 lr: 0.000002 loss: 0.5330 (0.5182) time: 0.3821 data: 0.0001 max mem: 52126 +Epoch: [32] [1080/3634] eta: 0:21:17 lr: 0.000002 loss: 0.4773 (0.5176) time: 0.3818 data: 0.0001 max mem: 52126 +Epoch: [32] [1090/3634] eta: 0:21:14 lr: 0.000002 loss: 0.4773 (0.5173) time: 0.4861 data: 0.0002 max mem: 52126 +Epoch: [32] [1100/3634] eta: 0:21:06 lr: 0.000002 loss: 0.5222 (0.5176) time: 0.4859 data: 0.0002 max mem: 52126 +Epoch: [32] [1110/3634] eta: 0:21:04 lr: 0.000002 loss: 0.5322 (0.5175) time: 0.4913 data: 0.0001 max mem: 52126 +Epoch: [32] [1120/3634] eta: 0:20:58 lr: 0.000002 loss: 0.5328 (0.5177) time: 0.5501 data: 0.0002 max mem: 52126 +Epoch: [32] [1130/3634] eta: 0:20:51 lr: 0.000002 loss: 0.5375 (0.5175) time: 0.4402 data: 0.0002 max mem: 52126 +Epoch: [32] [1140/3634] eta: 0:20:49 lr: 0.000002 loss: 0.5262 (0.5177) time: 0.5027 data: 0.0002 max mem: 52126 +Epoch: [32] [1150/3634] eta: 0:20:47 lr: 0.000002 loss: 0.5588 (0.5181) time: 0.6495 data: 0.0002 max mem: 52126 +Epoch: [32] [1160/3634] eta: 0:20:42 lr: 0.000002 loss: 0.5487 (0.5183) time: 0.5724 data: 0.0002 max mem: 52126 +Epoch: [32] [1170/3634] eta: 0:20:34 lr: 0.000002 loss: 0.5487 (0.5188) time: 0.4262 data: 0.0002 max mem: 52126 +Epoch: [32] [1180/3634] eta: 0:20:27 lr: 0.000002 loss: 0.4995 (0.5183) time: 0.3827 data: 0.0002 max mem: 52126 +Epoch: [32] [1190/3634] eta: 0:20:24 lr: 0.000002 loss: 0.4995 (0.5185) time: 0.5027 data: 0.0002 max mem: 52126 +Epoch: [32] [1200/3634] eta: 0:20:19 lr: 0.000002 loss: 0.5406 (0.5182) time: 0.5714 data: 0.0001 max mem: 52126 +Epoch: [32] [1210/3634] eta: 0:20:12 lr: 0.000002 loss: 0.4694 (0.5179) time: 0.4510 data: 0.0001 max mem: 52126 +Epoch: [32] [1220/3634] eta: 0:20:07 lr: 0.000002 loss: 0.4854 (0.5178) time: 0.4414 data: 0.0001 max mem: 52126 +Epoch: [32] [1230/3634] eta: 0:20:00 lr: 0.000002 loss: 0.5203 (0.5179) time: 0.4416 data: 0.0001 max mem: 52126 +Epoch: [32] [1240/3634] eta: 0:19:54 lr: 0.000002 loss: 0.5198 (0.5176) time: 0.4266 data: 0.0002 max mem: 52126 +Epoch: [32] [1250/3634] eta: 0:19:49 lr: 0.000002 loss: 0.4563 (0.5172) time: 0.4860 data: 0.0002 max mem: 52126 +Epoch: [32] [1260/3634] eta: 0:19:42 lr: 0.000002 loss: 0.4910 (0.5173) time: 0.4414 data: 0.0002 max mem: 52126 +Epoch: [32] [1270/3634] eta: 0:19:39 lr: 0.000002 loss: 0.5058 (0.5173) time: 0.4857 data: 0.0002 max mem: 52126 +Epoch: [32] [1280/3634] eta: 0:19:32 lr: 0.000002 loss: 0.4950 (0.5172) time: 0.4859 data: 0.0001 max mem: 52126 +Epoch: [32] [1290/3634] eta: 0:19:25 lr: 0.000002 loss: 0.4950 (0.5173) time: 0.3817 data: 0.0001 max mem: 52126 +Epoch: [32] [1300/3634] eta: 0:19:22 lr: 0.000002 loss: 0.5288 (0.5173) time: 0.5002 data: 0.0002 max mem: 52126 +Epoch: [32] [1310/3634] eta: 0:19:15 lr: 0.000002 loss: 0.5247 (0.5173) time: 0.5007 data: 0.0002 max mem: 52126 +Epoch: [32] [1320/3634] eta: 0:19:11 lr: 0.000002 loss: 0.5097 (0.5171) time: 0.4861 data: 0.0002 max mem: 52126 +Epoch: [32] [1330/3634] eta: 0:19:08 lr: 0.000002 loss: 0.5040 (0.5172) time: 0.5908 data: 0.0001 max mem: 52126 +Epoch: [32] [1340/3634] eta: 0:19:03 lr: 0.000002 loss: 0.5374 (0.5171) time: 0.5328 data: 0.0001 max mem: 52126 +Epoch: [32] [1350/3634] eta: 0:18:57 lr: 0.000002 loss: 0.5165 (0.5169) time: 0.4725 data: 0.0001 max mem: 52126 +Epoch: [32] [1360/3634] eta: 0:18:55 lr: 0.000002 loss: 0.5448 (0.5172) time: 0.5749 data: 0.0002 max mem: 52126 +Epoch: [32] [1370/3634] eta: 0:18:52 lr: 0.000002 loss: 0.5513 (0.5174) time: 0.6339 data: 0.0002 max mem: 52126 +Epoch: [32] [1380/3634] eta: 0:18:47 lr: 0.000002 loss: 0.5065 (0.5173) time: 0.5454 data: 0.0002 max mem: 52126 +Epoch: [32] [1390/3634] eta: 0:18:43 lr: 0.000002 loss: 0.5241 (0.5175) time: 0.5442 data: 0.0002 max mem: 52126 +Epoch: [32] [1400/3634] eta: 0:18:38 lr: 0.000002 loss: 0.5305 (0.5175) time: 0.5280 data: 0.0002 max mem: 52126 +Epoch: [32] [1410/3634] eta: 0:18:32 lr: 0.000002 loss: 0.5199 (0.5172) time: 0.4702 data: 0.0002 max mem: 52126 +Epoch: [32] [1420/3634] eta: 0:18:27 lr: 0.000002 loss: 0.5199 (0.5173) time: 0.4909 data: 0.0002 max mem: 52126 +Epoch: [32] [1430/3634] eta: 0:18:22 lr: 0.000002 loss: 0.5280 (0.5174) time: 0.4915 data: 0.0002 max mem: 52126 +Epoch: [32] [1440/3634] eta: 0:18:15 lr: 0.000002 loss: 0.5049 (0.5173) time: 0.4267 data: 0.0002 max mem: 52126 +Epoch: [32] [1450/3634] eta: 0:18:13 lr: 0.000002 loss: 0.5128 (0.5170) time: 0.5312 data: 0.0001 max mem: 52126 +Epoch: [32] [1460/3634] eta: 0:18:09 lr: 0.000002 loss: 0.5414 (0.5171) time: 0.6398 data: 0.0002 max mem: 52126 +Epoch: [32] [1470/3634] eta: 0:18:03 lr: 0.000002 loss: 0.4653 (0.5167) time: 0.4903 data: 0.0002 max mem: 52126 +Epoch: [32] [1480/3634] eta: 0:17:56 lr: 0.000002 loss: 0.4927 (0.5168) time: 0.3820 data: 0.0001 max mem: 52126 +Epoch: [32] [1490/3634] eta: 0:17:50 lr: 0.000002 loss: 0.5117 (0.5165) time: 0.4263 data: 0.0001 max mem: 52126 +Epoch: [32] [1500/3634] eta: 0:17:44 lr: 0.000002 loss: 0.4867 (0.5162) time: 0.4261 data: 0.0001 max mem: 52126 +Epoch: [32] [1510/3634] eta: 0:17:40 lr: 0.000002 loss: 0.4957 (0.5162) time: 0.4870 data: 0.0002 max mem: 52126 +Epoch: [32] [1520/3634] eta: 0:17:39 lr: 0.000002 loss: 0.4949 (0.5159) time: 0.6794 data: 0.0002 max mem: 52126 +Epoch: [32] [1530/3634] eta: 0:17:32 lr: 0.000002 loss: 0.5209 (0.5160) time: 0.5747 data: 0.0001 max mem: 52126 +Epoch: [32] [1540/3634] eta: 0:17:25 lr: 0.000002 loss: 0.5349 (0.5160) time: 0.3821 data: 0.0002 max mem: 52126 +Epoch: [32] [1550/3634] eta: 0:17:23 lr: 0.000002 loss: 0.5349 (0.5160) time: 0.5323 data: 0.0002 max mem: 52126 +Epoch: [32] [1560/3634] eta: 0:17:16 lr: 0.000002 loss: 0.4566 (0.5158) time: 0.5332 data: 0.0002 max mem: 52126 +Epoch: [32] [1570/3634] eta: 0:17:10 lr: 0.000002 loss: 0.5410 (0.5161) time: 0.3827 data: 0.0002 max mem: 52126 +Epoch: [32] [1580/3634] eta: 0:17:06 lr: 0.000002 loss: 0.5632 (0.5161) time: 0.5031 data: 0.0002 max mem: 52126 +Epoch: [32] [1590/3634] eta: 0:17:00 lr: 0.000002 loss: 0.4929 (0.5159) time: 0.5035 data: 0.0002 max mem: 52126 +Epoch: [32] [1600/3634] eta: 0:16:53 lr: 0.000002 loss: 0.4985 (0.5161) time: 0.3825 data: 0.0002 max mem: 52126 +Epoch: [32] [1610/3634] eta: 0:16:49 lr: 0.000002 loss: 0.5292 (0.5161) time: 0.4421 data: 0.0001 max mem: 52126 +Epoch: [32] [1620/3634] eta: 0:16:45 lr: 0.000002 loss: 0.5324 (0.5163) time: 0.5461 data: 0.0001 max mem: 52126 +Epoch: [32] [1630/3634] eta: 0:16:39 lr: 0.000002 loss: 0.5650 (0.5164) time: 0.5319 data: 0.0001 max mem: 52126 +Epoch: [32] [1640/3634] eta: 0:16:33 lr: 0.000002 loss: 0.5150 (0.5161) time: 0.4280 data: 0.0002 max mem: 52126 +Epoch: [32] [1650/3634] eta: 0:16:28 lr: 0.000002 loss: 0.4825 (0.5162) time: 0.4254 data: 0.0002 max mem: 52126 +Epoch: [32] [1660/3634] eta: 0:16:22 lr: 0.000002 loss: 0.5464 (0.5164) time: 0.4717 data: 0.0001 max mem: 52126 +Epoch: [32] [1670/3634] eta: 0:16:19 lr: 0.000002 loss: 0.5584 (0.5168) time: 0.5409 data: 0.0001 max mem: 52126 +Epoch: [32] [1680/3634] eta: 0:16:13 lr: 0.000002 loss: 0.5312 (0.5167) time: 0.5423 data: 0.0001 max mem: 52126 +Epoch: [32] [1690/3634] eta: 0:16:07 lr: 0.000002 loss: 0.4942 (0.5166) time: 0.4296 data: 0.0002 max mem: 52126 +Epoch: [32] [1700/3634] eta: 0:16:01 lr: 0.000002 loss: 0.5300 (0.5167) time: 0.3824 data: 0.0002 max mem: 52126 +Epoch: [32] [1710/3634] eta: 0:15:56 lr: 0.000002 loss: 0.5389 (0.5167) time: 0.4447 data: 0.0002 max mem: 52126 +Epoch: [32] [1720/3634] eta: 0:15:51 lr: 0.000002 loss: 0.4983 (0.5164) time: 0.5047 data: 0.0001 max mem: 52126 +Epoch: [32] [1730/3634] eta: 0:15:45 lr: 0.000002 loss: 0.4865 (0.5165) time: 0.4425 data: 0.0002 max mem: 52126 +Epoch: [32] [1740/3634] eta: 0:15:39 lr: 0.000002 loss: 0.5199 (0.5165) time: 0.3819 data: 0.0002 max mem: 52126 +Epoch: [32] [1750/3634] eta: 0:15:34 lr: 0.000002 loss: 0.5327 (0.5165) time: 0.4421 data: 0.0002 max mem: 52126 +Epoch: [32] [1760/3634] eta: 0:15:29 lr: 0.000002 loss: 0.5049 (0.5164) time: 0.4896 data: 0.0001 max mem: 52126 +Epoch: [32] [1770/3634] eta: 0:15:26 lr: 0.000002 loss: 0.4886 (0.5164) time: 0.6093 data: 0.0001 max mem: 52126 +Epoch: [32] [1780/3634] eta: 0:15:20 lr: 0.000002 loss: 0.5054 (0.5164) time: 0.5624 data: 0.0001 max mem: 52126 +Epoch: [32] [1790/3634] eta: 0:15:14 lr: 0.000002 loss: 0.5190 (0.5164) time: 0.3824 data: 0.0001 max mem: 52126 +Epoch: [32] [1800/3634] eta: 0:15:10 lr: 0.000002 loss: 0.5392 (0.5165) time: 0.4847 data: 0.0001 max mem: 52126 +Epoch: [32] [1810/3634] eta: 0:15:04 lr: 0.000002 loss: 0.5350 (0.5164) time: 0.4851 data: 0.0001 max mem: 52126 +Epoch: [32] [1820/3634] eta: 0:14:58 lr: 0.000002 loss: 0.5320 (0.5164) time: 0.3827 data: 0.0001 max mem: 52126 +Epoch: [32] [1830/3634] eta: 0:14:54 lr: 0.000002 loss: 0.5005 (0.5164) time: 0.5009 data: 0.0001 max mem: 52126 +Epoch: [32] [1840/3634] eta: 0:14:49 lr: 0.000002 loss: 0.4819 (0.5161) time: 0.5488 data: 0.0001 max mem: 52126 +Epoch: [32] [1850/3634] eta: 0:14:44 lr: 0.000002 loss: 0.5084 (0.5162) time: 0.4753 data: 0.0001 max mem: 52126 +Epoch: [32] [1860/3634] eta: 0:14:40 lr: 0.000002 loss: 0.5145 (0.5162) time: 0.5331 data: 0.0001 max mem: 52126 +Epoch: [32] [1870/3634] eta: 0:14:36 lr: 0.000002 loss: 0.5149 (0.5162) time: 0.5980 data: 0.0002 max mem: 52126 +Epoch: [32] [1880/3634] eta: 0:14:31 lr: 0.000002 loss: 0.5192 (0.5163) time: 0.5519 data: 0.0002 max mem: 52126 +Epoch: [32] [1890/3634] eta: 0:14:26 lr: 0.000002 loss: 0.4864 (0.5161) time: 0.5301 data: 0.0002 max mem: 52126 +Epoch: [32] [1900/3634] eta: 0:14:21 lr: 0.000002 loss: 0.5053 (0.5161) time: 0.5299 data: 0.0001 max mem: 52126 +Epoch: [32] [1910/3634] eta: 0:14:17 lr: 0.000002 loss: 0.5031 (0.5158) time: 0.5509 data: 0.0001 max mem: 52126 +Epoch: [32] [1920/3634] eta: 0:14:11 lr: 0.000002 loss: 0.4757 (0.5156) time: 0.4910 data: 0.0001 max mem: 52126 +Epoch: [32] [1930/3634] eta: 0:14:06 lr: 0.000002 loss: 0.4988 (0.5155) time: 0.4275 data: 0.0002 max mem: 52126 +Epoch: [32] [1940/3634] eta: 0:14:01 lr: 0.000002 loss: 0.4988 (0.5154) time: 0.4872 data: 0.0002 max mem: 52126 +Epoch: [32] [1950/3634] eta: 0:13:57 lr: 0.000002 loss: 0.4874 (0.5153) time: 0.5356 data: 0.0002 max mem: 52126 +Epoch: [32] [1960/3634] eta: 0:13:52 lr: 0.000002 loss: 0.4942 (0.5153) time: 0.5371 data: 0.0002 max mem: 52126 +Epoch: [32] [1970/3634] eta: 0:13:48 lr: 0.000002 loss: 0.5057 (0.5152) time: 0.5478 data: 0.0002 max mem: 52126 +Epoch: [32] [1980/3634] eta: 0:13:44 lr: 0.000002 loss: 0.5157 (0.5152) time: 0.5897 data: 0.0002 max mem: 52126 +Epoch: [32] [1990/3634] eta: 0:13:38 lr: 0.000002 loss: 0.5326 (0.5154) time: 0.5294 data: 0.0002 max mem: 52126 +Epoch: [32] [2000/3634] eta: 0:13:33 lr: 0.000002 loss: 0.5523 (0.5155) time: 0.4730 data: 0.0002 max mem: 52126 +Epoch: [32] [2010/3634] eta: 0:13:29 lr: 0.000002 loss: 0.5513 (0.5157) time: 0.5178 data: 0.0002 max mem: 52126 +Epoch: [32] [2020/3634] eta: 0:13:24 lr: 0.000002 loss: 0.5485 (0.5157) time: 0.5336 data: 0.0002 max mem: 52126 +Epoch: [32] [2030/3634] eta: 0:13:18 lr: 0.000002 loss: 0.5210 (0.5155) time: 0.4445 data: 0.0002 max mem: 52126 +Epoch: [32] [2040/3634] eta: 0:13:12 lr: 0.000002 loss: 0.4910 (0.5154) time: 0.3828 data: 0.0002 max mem: 52126 +Epoch: [32] [2050/3634] eta: 0:13:06 lr: 0.000002 loss: 0.4917 (0.5154) time: 0.3830 data: 0.0002 max mem: 52126 +Epoch: [32] [2060/3634] eta: 0:13:01 lr: 0.000002 loss: 0.5285 (0.5154) time: 0.4298 data: 0.0002 max mem: 52126 +Epoch: [32] [2070/3634] eta: 0:12:57 lr: 0.000002 loss: 0.5328 (0.5155) time: 0.5329 data: 0.0002 max mem: 52126 +Epoch: [32] [2080/3634] eta: 0:12:52 lr: 0.000002 loss: 0.5236 (0.5155) time: 0.5761 data: 0.0002 max mem: 52126 +Epoch: [32] [2090/3634] eta: 0:12:48 lr: 0.000002 loss: 0.5150 (0.5156) time: 0.5611 data: 0.0002 max mem: 52126 +Epoch: [32] [2100/3634] eta: 0:12:44 lr: 0.000002 loss: 0.5315 (0.5155) time: 0.6346 data: 0.0002 max mem: 52126 +Epoch: [32] [2110/3634] eta: 0:12:39 lr: 0.000002 loss: 0.5314 (0.5156) time: 0.5460 data: 0.0002 max mem: 52126 +Epoch: [32] [2120/3634] eta: 0:12:34 lr: 0.000002 loss: 0.5288 (0.5154) time: 0.4422 data: 0.0002 max mem: 52126 +Epoch: [32] [2130/3634] eta: 0:12:30 lr: 0.000002 loss: 0.5126 (0.5154) time: 0.5612 data: 0.0002 max mem: 52126 +Epoch: [32] [2140/3634] eta: 0:12:24 lr: 0.000002 loss: 0.5126 (0.5155) time: 0.5011 data: 0.0001 max mem: 52126 +Epoch: [32] [2150/3634] eta: 0:12:19 lr: 0.000002 loss: 0.5302 (0.5155) time: 0.4414 data: 0.0001 max mem: 52126 +Epoch: [32] [2160/3634] eta: 0:12:14 lr: 0.000002 loss: 0.5318 (0.5157) time: 0.4881 data: 0.0001 max mem: 52126 +Epoch: [32] [2170/3634] eta: 0:12:08 lr: 0.000002 loss: 0.5553 (0.5159) time: 0.4744 data: 0.0002 max mem: 52126 +Epoch: [32] [2180/3634] eta: 0:12:03 lr: 0.000002 loss: 0.5058 (0.5158) time: 0.4717 data: 0.0002 max mem: 52126 +Epoch: [32] [2190/3634] eta: 0:11:58 lr: 0.000002 loss: 0.5038 (0.5157) time: 0.4267 data: 0.0002 max mem: 52126 +Epoch: [32] [2200/3634] eta: 0:11:52 lr: 0.000002 loss: 0.5465 (0.5158) time: 0.3826 data: 0.0001 max mem: 52126 +Epoch: [32] [2210/3634] eta: 0:11:47 lr: 0.000002 loss: 0.5310 (0.5159) time: 0.4296 data: 0.0002 max mem: 52126 +Epoch: [32] [2220/3634] eta: 0:11:42 lr: 0.000002 loss: 0.5259 (0.5159) time: 0.4892 data: 0.0002 max mem: 52126 +Epoch: [32] [2230/3634] eta: 0:11:38 lr: 0.000002 loss: 0.5340 (0.5159) time: 0.5981 data: 0.0002 max mem: 52126 +Epoch: [32] [2240/3634] eta: 0:11:32 lr: 0.000002 loss: 0.5494 (0.5161) time: 0.5391 data: 0.0002 max mem: 52126 +Epoch: [32] [2250/3634] eta: 0:11:27 lr: 0.000002 loss: 0.5573 (0.5163) time: 0.4299 data: 0.0002 max mem: 52126 +Epoch: [32] [2260/3634] eta: 0:11:22 lr: 0.000002 loss: 0.5539 (0.5164) time: 0.4300 data: 0.0002 max mem: 52126 +Epoch: [32] [2270/3634] eta: 0:11:16 lr: 0.000002 loss: 0.5301 (0.5164) time: 0.3830 data: 0.0002 max mem: 52126 +Epoch: [32] [2280/3634] eta: 0:11:11 lr: 0.000002 loss: 0.5006 (0.5161) time: 0.4426 data: 0.0002 max mem: 52126 +Epoch: [32] [2290/3634] eta: 0:11:05 lr: 0.000002 loss: 0.5006 (0.5161) time: 0.4426 data: 0.0002 max mem: 52126 +Epoch: [32] [2300/3634] eta: 0:11:00 lr: 0.000002 loss: 0.5431 (0.5161) time: 0.3821 data: 0.0001 max mem: 52126 +Epoch: [32] [2310/3634] eta: 0:10:55 lr: 0.000002 loss: 0.5262 (0.5160) time: 0.4287 data: 0.0001 max mem: 52126 +Epoch: [32] [2320/3634] eta: 0:10:49 lr: 0.000002 loss: 0.5252 (0.5160) time: 0.4294 data: 0.0002 max mem: 52126 +Epoch: [32] [2330/3634] eta: 0:10:44 lr: 0.000002 loss: 0.5091 (0.5159) time: 0.4271 data: 0.0002 max mem: 52126 +Epoch: [32] [2340/3634] eta: 0:10:39 lr: 0.000002 loss: 0.4832 (0.5157) time: 0.4886 data: 0.0002 max mem: 52126 +Epoch: [32] [2350/3634] eta: 0:10:34 lr: 0.000002 loss: 0.4832 (0.5157) time: 0.4441 data: 0.0002 max mem: 52126 +Epoch: [32] [2360/3634] eta: 0:10:29 lr: 0.000002 loss: 0.5214 (0.5157) time: 0.4259 data: 0.0002 max mem: 52126 +Epoch: [32] [2370/3634] eta: 0:10:25 lr: 0.000002 loss: 0.5406 (0.5157) time: 0.6367 data: 0.0002 max mem: 52126 +Epoch: [32] [2380/3634] eta: 0:10:20 lr: 0.000002 loss: 0.5456 (0.5156) time: 0.5928 data: 0.0002 max mem: 52126 +Epoch: [32] [2390/3634] eta: 0:10:15 lr: 0.000002 loss: 0.5233 (0.5156) time: 0.4883 data: 0.0001 max mem: 52126 +Epoch: [32] [2400/3634] eta: 0:10:10 lr: 0.000002 loss: 0.5043 (0.5155) time: 0.4885 data: 0.0002 max mem: 52126 +Epoch: [32] [2410/3634] eta: 0:10:05 lr: 0.000002 loss: 0.5153 (0.5156) time: 0.4420 data: 0.0002 max mem: 52126 +Epoch: [32] [2420/3634] eta: 0:10:00 lr: 0.000002 loss: 0.5270 (0.5156) time: 0.5443 data: 0.0002 max mem: 52126 +Epoch: [32] [2430/3634] eta: 0:09:55 lr: 0.000002 loss: 0.5451 (0.5158) time: 0.4844 data: 0.0002 max mem: 52126 +Epoch: [32] [2440/3634] eta: 0:09:51 lr: 0.000002 loss: 0.5464 (0.5159) time: 0.5617 data: 0.0002 max mem: 52126 +Epoch: [32] [2450/3634] eta: 0:09:46 lr: 0.000002 loss: 0.5484 (0.5159) time: 0.6208 data: 0.0001 max mem: 52126 +Epoch: [32] [2460/3634] eta: 0:09:41 lr: 0.000002 loss: 0.5222 (0.5159) time: 0.4875 data: 0.0001 max mem: 52126 +Epoch: [32] [2470/3634] eta: 0:09:38 lr: 0.000002 loss: 0.5095 (0.5159) time: 0.6372 data: 0.0001 max mem: 52126 +Epoch: [32] [2480/3634] eta: 0:09:33 lr: 0.000002 loss: 0.5384 (0.5160) time: 0.6355 data: 0.0001 max mem: 52126 +Epoch: [32] [2490/3634] eta: 0:09:27 lr: 0.000002 loss: 0.5434 (0.5161) time: 0.4267 data: 0.0002 max mem: 52126 +Epoch: [32] [2500/3634] eta: 0:09:22 lr: 0.000002 loss: 0.5411 (0.5162) time: 0.4271 data: 0.0002 max mem: 52126 +Epoch: [32] [2510/3634] eta: 0:09:18 lr: 0.000002 loss: 0.5419 (0.5162) time: 0.5936 data: 0.0001 max mem: 52126 +Epoch: [32] [2520/3634] eta: 0:09:13 lr: 0.000002 loss: 0.5382 (0.5162) time: 0.6072 data: 0.0001 max mem: 52126 +Epoch: [32] [2530/3634] eta: 0:09:09 lr: 0.000002 loss: 0.5264 (0.5163) time: 0.5478 data: 0.0001 max mem: 52126 +Epoch: [32] [2540/3634] eta: 0:09:03 lr: 0.000002 loss: 0.5270 (0.5163) time: 0.4886 data: 0.0001 max mem: 52126 +Epoch: [32] [2550/3634] eta: 0:08:58 lr: 0.000002 loss: 0.5574 (0.5165) time: 0.4697 data: 0.0001 max mem: 52126 +Epoch: [32] [2560/3634] eta: 0:08:53 lr: 0.000002 loss: 0.5617 (0.5165) time: 0.5294 data: 0.0001 max mem: 52126 +Epoch: [32] [2570/3634] eta: 0:08:49 lr: 0.000002 loss: 0.5023 (0.5164) time: 0.5472 data: 0.0001 max mem: 52126 +Epoch: [32] [2580/3634] eta: 0:08:44 lr: 0.000002 loss: 0.4972 (0.5163) time: 0.5963 data: 0.0001 max mem: 52126 +Epoch: [32] [2590/3634] eta: 0:08:39 lr: 0.000002 loss: 0.4708 (0.5162) time: 0.4904 data: 0.0001 max mem: 52126 +Epoch: [32] [2600/3634] eta: 0:08:34 lr: 0.000002 loss: 0.5135 (0.5162) time: 0.4711 data: 0.0001 max mem: 52126 +Epoch: [32] [2610/3634] eta: 0:08:29 lr: 0.000002 loss: 0.5246 (0.5162) time: 0.5341 data: 0.0001 max mem: 52126 +Epoch: [32] [2620/3634] eta: 0:08:24 lr: 0.000002 loss: 0.4963 (0.5161) time: 0.4466 data: 0.0002 max mem: 52126 +Epoch: [32] [2630/3634] eta: 0:08:19 lr: 0.000002 loss: 0.4986 (0.5160) time: 0.4439 data: 0.0002 max mem: 52126 +Epoch: [32] [2640/3634] eta: 0:08:15 lr: 0.000002 loss: 0.4914 (0.5159) time: 0.6546 data: 0.0002 max mem: 52126 +Epoch: [32] [2650/3634] eta: 0:08:10 lr: 0.000002 loss: 0.4914 (0.5158) time: 0.6407 data: 0.0002 max mem: 52126 +Epoch: [32] [2660/3634] eta: 0:08:05 lr: 0.000002 loss: 0.5115 (0.5158) time: 0.5511 data: 0.0002 max mem: 52126 +Epoch: [32] [2670/3634] eta: 0:08:00 lr: 0.000002 loss: 0.5155 (0.5158) time: 0.5531 data: 0.0002 max mem: 52126 +Epoch: [32] [2680/3634] eta: 0:07:56 lr: 0.000002 loss: 0.5206 (0.5158) time: 0.5932 data: 0.0002 max mem: 52126 +Epoch: [32] [2690/3634] eta: 0:07:51 lr: 0.000002 loss: 0.4402 (0.5155) time: 0.5888 data: 0.0001 max mem: 52126 +Epoch: [32] [2700/3634] eta: 0:07:46 lr: 0.000002 loss: 0.4673 (0.5155) time: 0.5313 data: 0.0001 max mem: 52126 +Epoch: [32] [2710/3634] eta: 0:07:41 lr: 0.000002 loss: 0.5159 (0.5155) time: 0.4871 data: 0.0002 max mem: 52126 +Epoch: [32] [2720/3634] eta: 0:07:36 lr: 0.000002 loss: 0.4984 (0.5155) time: 0.4293 data: 0.0002 max mem: 52126 +Epoch: [32] [2730/3634] eta: 0:07:30 lr: 0.000002 loss: 0.5329 (0.5155) time: 0.4294 data: 0.0001 max mem: 52126 +Epoch: [32] [2740/3634] eta: 0:07:25 lr: 0.000002 loss: 0.5302 (0.5154) time: 0.3821 data: 0.0002 max mem: 52126 +Epoch: [32] [2750/3634] eta: 0:07:20 lr: 0.000002 loss: 0.4922 (0.5154) time: 0.3823 data: 0.0002 max mem: 52126 +Epoch: [32] [2760/3634] eta: 0:07:14 lr: 0.000002 loss: 0.4922 (0.5154) time: 0.3820 data: 0.0002 max mem: 52126 +Epoch: [32] [2770/3634] eta: 0:07:09 lr: 0.000002 loss: 0.5074 (0.5155) time: 0.3816 data: 0.0002 max mem: 52126 +Epoch: [32] [2780/3634] eta: 0:07:04 lr: 0.000002 loss: 0.5269 (0.5155) time: 0.3817 data: 0.0002 max mem: 52126 +Epoch: [32] [2790/3634] eta: 0:06:59 lr: 0.000002 loss: 0.5139 (0.5154) time: 0.4268 data: 0.0002 max mem: 52126 +Epoch: [32] [2800/3634] eta: 0:06:53 lr: 0.000002 loss: 0.4796 (0.5153) time: 0.4265 data: 0.0002 max mem: 52126 +Epoch: [32] [2810/3634] eta: 0:06:48 lr: 0.000002 loss: 0.5217 (0.5154) time: 0.3829 data: 0.0002 max mem: 52126 +Epoch: [32] [2820/3634] eta: 0:06:43 lr: 0.000002 loss: 0.5374 (0.5154) time: 0.3848 data: 0.0002 max mem: 52126 +Epoch: [32] [2830/3634] eta: 0:06:38 lr: 0.000002 loss: 0.4811 (0.5152) time: 0.3837 data: 0.0002 max mem: 52126 +Epoch: [32] [2840/3634] eta: 0:06:33 lr: 0.000002 loss: 0.5063 (0.5153) time: 0.4427 data: 0.0002 max mem: 52126 +Epoch: [32] [2850/3634] eta: 0:06:28 lr: 0.000002 loss: 0.5424 (0.5153) time: 0.4866 data: 0.0001 max mem: 52126 +Epoch: [32] [2860/3634] eta: 0:06:23 lr: 0.000002 loss: 0.5324 (0.5153) time: 0.4756 data: 0.0002 max mem: 52126 +Epoch: [32] [2870/3634] eta: 0:06:17 lr: 0.000002 loss: 0.5541 (0.5155) time: 0.4317 data: 0.0002 max mem: 52126 +Epoch: [32] [2880/3634] eta: 0:06:12 lr: 0.000002 loss: 0.5614 (0.5155) time: 0.3823 data: 0.0002 max mem: 52126 +Epoch: [32] [2890/3634] eta: 0:06:07 lr: 0.000002 loss: 0.5251 (0.5154) time: 0.4701 data: 0.0002 max mem: 52126 +Epoch: [32] [2900/3634] eta: 0:06:02 lr: 0.000002 loss: 0.5408 (0.5156) time: 0.4701 data: 0.0002 max mem: 52126 +Epoch: [32] [2910/3634] eta: 0:05:57 lr: 0.000002 loss: 0.5538 (0.5156) time: 0.5006 data: 0.0001 max mem: 52126 +Epoch: [32] [2920/3634] eta: 0:05:53 lr: 0.000002 loss: 0.5654 (0.5158) time: 0.5605 data: 0.0002 max mem: 52126 +Epoch: [32] [2930/3634] eta: 0:05:48 lr: 0.000002 loss: 0.5350 (0.5157) time: 0.5025 data: 0.0002 max mem: 52126 +Epoch: [32] [2940/3634] eta: 0:05:43 lr: 0.000002 loss: 0.4567 (0.5155) time: 0.6214 data: 0.0002 max mem: 52126 +Epoch: [32] [2950/3634] eta: 0:05:38 lr: 0.000002 loss: 0.5050 (0.5156) time: 0.6093 data: 0.0002 max mem: 52126 +Epoch: [32] [2960/3634] eta: 0:05:33 lr: 0.000002 loss: 0.5391 (0.5156) time: 0.4934 data: 0.0002 max mem: 52126 +Epoch: [32] [2970/3634] eta: 0:05:28 lr: 0.000002 loss: 0.5141 (0.5155) time: 0.4447 data: 0.0002 max mem: 52126 +Epoch: [32] [2980/3634] eta: 0:05:23 lr: 0.000002 loss: 0.5556 (0.5157) time: 0.4406 data: 0.0002 max mem: 52126 +Epoch: [32] [2990/3634] eta: 0:05:18 lr: 0.000002 loss: 0.5300 (0.5156) time: 0.5494 data: 0.0001 max mem: 52126 +Epoch: [32] [3000/3634] eta: 0:05:14 lr: 0.000002 loss: 0.4997 (0.5156) time: 0.5963 data: 0.0001 max mem: 52126 +Epoch: [32] [3010/3634] eta: 0:05:09 lr: 0.000002 loss: 0.5006 (0.5155) time: 0.4888 data: 0.0002 max mem: 52126 +Epoch: [32] [3020/3634] eta: 0:05:04 lr: 0.000002 loss: 0.5051 (0.5154) time: 0.4893 data: 0.0002 max mem: 52126 +Epoch: [32] [3030/3634] eta: 0:04:59 lr: 0.000002 loss: 0.5269 (0.5154) time: 0.5934 data: 0.0002 max mem: 52126 +Epoch: [32] [3040/3634] eta: 0:04:54 lr: 0.000002 loss: 0.5341 (0.5154) time: 0.4872 data: 0.0002 max mem: 52126 +Epoch: [32] [3050/3634] eta: 0:04:49 lr: 0.000002 loss: 0.5469 (0.5155) time: 0.4886 data: 0.0002 max mem: 52126 +Epoch: [32] [3060/3634] eta: 0:04:44 lr: 0.000002 loss: 0.5159 (0.5153) time: 0.5479 data: 0.0002 max mem: 52126 +Epoch: [32] [3070/3634] eta: 0:04:39 lr: 0.000002 loss: 0.4951 (0.5153) time: 0.4409 data: 0.0002 max mem: 52126 +Epoch: [32] [3080/3634] eta: 0:04:34 lr: 0.000002 loss: 0.4971 (0.5152) time: 0.5308 data: 0.0002 max mem: 52126 +Epoch: [32] [3090/3634] eta: 0:04:30 lr: 0.000002 loss: 0.5239 (0.5153) time: 0.7469 data: 0.0002 max mem: 52126 +Epoch: [32] [3100/3634] eta: 0:04:25 lr: 0.000002 loss: 0.5306 (0.5154) time: 0.6896 data: 0.0002 max mem: 52126 +Epoch: [32] [3110/3634] eta: 0:04:20 lr: 0.000002 loss: 0.4898 (0.5153) time: 0.4744 data: 0.0002 max mem: 52126 +Epoch: [32] [3120/3634] eta: 0:04:15 lr: 0.000002 loss: 0.4983 (0.5153) time: 0.5555 data: 0.0002 max mem: 52126 +Epoch: [32] [3130/3634] eta: 0:04:11 lr: 0.000002 loss: 0.5177 (0.5153) time: 0.6584 data: 0.0002 max mem: 52126 +Epoch: [32] [3140/3634] eta: 0:04:06 lr: 0.000002 loss: 0.5387 (0.5155) time: 0.5904 data: 0.0002 max mem: 52126 +Epoch: [32] [3150/3634] eta: 0:04:01 lr: 0.000002 loss: 0.5329 (0.5154) time: 0.5486 data: 0.0002 max mem: 52126 +Epoch: [32] [3160/3634] eta: 0:03:56 lr: 0.000002 loss: 0.5195 (0.5153) time: 0.4874 data: 0.0002 max mem: 52126 +Epoch: [32] [3170/3634] eta: 0:03:51 lr: 0.000002 loss: 0.5213 (0.5154) time: 0.4892 data: 0.0002 max mem: 52126 +Epoch: [32] [3180/3634] eta: 0:03:46 lr: 0.000002 loss: 0.5157 (0.5153) time: 0.5507 data: 0.0002 max mem: 52126 +Epoch: [32] [3190/3634] eta: 0:03:41 lr: 0.000002 loss: 0.4912 (0.5152) time: 0.5758 data: 0.0002 max mem: 52126 +Epoch: [32] [3200/3634] eta: 0:03:36 lr: 0.000002 loss: 0.5099 (0.5153) time: 0.4709 data: 0.0002 max mem: 52126 +Epoch: [32] [3210/3634] eta: 0:03:31 lr: 0.000002 loss: 0.5075 (0.5152) time: 0.5318 data: 0.0002 max mem: 52126 +Epoch: [32] [3220/3634] eta: 0:03:26 lr: 0.000002 loss: 0.5205 (0.5152) time: 0.6951 data: 0.0002 max mem: 52126 +Epoch: [32] [3230/3634] eta: 0:03:21 lr: 0.000002 loss: 0.5161 (0.5152) time: 0.6050 data: 0.0002 max mem: 52126 +Epoch: [32] [3240/3634] eta: 0:03:16 lr: 0.000002 loss: 0.5026 (0.5153) time: 0.4998 data: 0.0002 max mem: 52126 +Epoch: [32] [3250/3634] eta: 0:03:11 lr: 0.000002 loss: 0.5487 (0.5154) time: 0.4998 data: 0.0002 max mem: 52126 +Epoch: [32] [3260/3634] eta: 0:03:06 lr: 0.000002 loss: 0.5114 (0.5153) time: 0.4411 data: 0.0002 max mem: 52126 +Epoch: [32] [3270/3634] eta: 0:03:01 lr: 0.000002 loss: 0.4844 (0.5151) time: 0.3823 data: 0.0002 max mem: 52126 +Epoch: [32] [3280/3634] eta: 0:02:56 lr: 0.000002 loss: 0.4428 (0.5150) time: 0.4264 data: 0.0002 max mem: 52126 +Epoch: [32] [3290/3634] eta: 0:02:51 lr: 0.000002 loss: 0.4641 (0.5150) time: 0.4262 data: 0.0002 max mem: 52126 +Epoch: [32] [3300/3634] eta: 0:02:46 lr: 0.000002 loss: 0.5434 (0.5150) time: 0.3847 data: 0.0002 max mem: 52126 +Epoch: [32] [3310/3634] eta: 0:02:41 lr: 0.000002 loss: 0.5434 (0.5151) time: 0.3847 data: 0.0002 max mem: 52126 +Epoch: [32] [3320/3634] eta: 0:02:36 lr: 0.000002 loss: 0.5237 (0.5150) time: 0.3825 data: 0.0002 max mem: 52126 +Epoch: [32] [3330/3634] eta: 0:02:31 lr: 0.000002 loss: 0.5030 (0.5150) time: 0.3827 data: 0.0002 max mem: 52126 +Epoch: [32] [3340/3634] eta: 0:02:26 lr: 0.000002 loss: 0.5309 (0.5151) time: 0.3827 data: 0.0002 max mem: 52126 +Epoch: [32] [3350/3634] eta: 0:02:21 lr: 0.000002 loss: 0.5209 (0.5150) time: 0.4434 data: 0.0002 max mem: 52126 +Epoch: [32] [3360/3634] eta: 0:02:16 lr: 0.000002 loss: 0.5209 (0.5151) time: 0.5368 data: 0.0002 max mem: 52126 +Epoch: [32] [3370/3634] eta: 0:02:11 lr: 0.000002 loss: 0.5068 (0.5150) time: 0.4765 data: 0.0002 max mem: 52126 +Epoch: [32] [3380/3634] eta: 0:02:06 lr: 0.000002 loss: 0.4644 (0.5149) time: 0.4258 data: 0.0002 max mem: 52126 +Epoch: [32] [3390/3634] eta: 0:02:01 lr: 0.000002 loss: 0.5006 (0.5149) time: 0.4250 data: 0.0002 max mem: 52126 +Epoch: [32] [3400/3634] eta: 0:01:56 lr: 0.000002 loss: 0.4999 (0.5148) time: 0.3826 data: 0.0002 max mem: 52126 +Epoch: [32] [3410/3634] eta: 0:01:51 lr: 0.000002 loss: 0.4815 (0.5147) time: 0.3829 data: 0.0002 max mem: 52126 +Epoch: [32] [3420/3634] eta: 0:01:45 lr: 0.000002 loss: 0.5113 (0.5147) time: 0.3823 data: 0.0002 max mem: 52126 +Epoch: [32] [3430/3634] eta: 0:01:41 lr: 0.000002 loss: 0.4972 (0.5146) time: 0.4265 data: 0.0002 max mem: 52126 +Epoch: [32] [3440/3634] eta: 0:01:35 lr: 0.000002 loss: 0.4972 (0.5146) time: 0.4265 data: 0.0002 max mem: 52126 +Epoch: [32] [3450/3634] eta: 0:01:30 lr: 0.000002 loss: 0.5247 (0.5146) time: 0.3824 data: 0.0002 max mem: 52126 +Epoch: [32] [3460/3634] eta: 0:01:26 lr: 0.000002 loss: 0.5376 (0.5147) time: 0.4313 data: 0.0002 max mem: 52126 +Epoch: [32] [3470/3634] eta: 0:01:21 lr: 0.000002 loss: 0.5081 (0.5147) time: 0.5980 data: 0.0002 max mem: 52126 +Epoch: [32] [3480/3634] eta: 0:01:16 lr: 0.000002 loss: 0.4974 (0.5147) time: 0.5947 data: 0.0002 max mem: 52126 +Epoch: [32] [3490/3634] eta: 0:01:11 lr: 0.000002 loss: 0.5043 (0.5146) time: 0.4281 data: 0.0002 max mem: 52126 +Epoch: [32] [3500/3634] eta: 0:01:06 lr: 0.000002 loss: 0.5098 (0.5146) time: 0.5014 data: 0.0002 max mem: 52126 +Epoch: [32] [3510/3634] eta: 0:01:01 lr: 0.000002 loss: 0.5086 (0.5146) time: 0.5011 data: 0.0002 max mem: 52126 +Epoch: [32] [3520/3634] eta: 0:00:56 lr: 0.000002 loss: 0.5223 (0.5146) time: 0.6093 data: 0.0002 max mem: 52126 +Epoch: [32] [3530/3634] eta: 0:00:51 lr: 0.000002 loss: 0.5189 (0.5146) time: 0.6529 data: 0.0002 max mem: 52126 +Epoch: [32] [3540/3634] eta: 0:00:46 lr: 0.000002 loss: 0.5189 (0.5146) time: 0.4259 data: 0.0002 max mem: 52126 +Epoch: [32] [3550/3634] eta: 0:00:41 lr: 0.000002 loss: 0.5255 (0.5147) time: 0.5646 data: 0.0002 max mem: 52126 +Epoch: [32] [3560/3634] eta: 0:00:36 lr: 0.000002 loss: 0.5523 (0.5147) time: 0.6093 data: 0.0002 max mem: 52126 +Epoch: [32] [3570/3634] eta: 0:00:31 lr: 0.000002 loss: 0.5590 (0.5149) time: 0.4891 data: 0.0002 max mem: 52126 +Epoch: [32] [3580/3634] eta: 0:00:26 lr: 0.000002 loss: 0.5549 (0.5149) time: 0.4442 data: 0.0002 max mem: 52126 +Epoch: [32] [3590/3634] eta: 0:00:21 lr: 0.000002 loss: 0.5091 (0.5149) time: 0.4990 data: 0.0002 max mem: 52126 +Epoch: [32] [3600/3634] eta: 0:00:16 lr: 0.000002 loss: 0.4983 (0.5148) time: 0.4991 data: 0.0002 max mem: 52126 +Epoch: [32] [3610/3634] eta: 0:00:11 lr: 0.000002 loss: 0.4996 (0.5147) time: 0.4255 data: 0.0002 max mem: 52126 +Epoch: [32] [3620/3634] eta: 0:00:06 lr: 0.000002 loss: 0.5004 (0.5148) time: 0.5898 data: 0.0028 max mem: 52126 +Epoch: [32] [3630/3634] eta: 0:00:01 lr: 0.000002 loss: 0.4926 (0.5146) time: 0.5455 data: 0.0028 max mem: 52126 +Epoch: [32] [3633/3634] eta: 0:00:00 lr: 0.000002 loss: 0.4926 (0.5146) time: 0.4848 data: 0.0028 max mem: 52126 +Epoch: [32] Total time: 0:30:03 (0.4963 s / it) +Averaged stats: lr: 0.000002 loss: 0.4926 (0.5144) +Test: [ 0/1403] eta: 1:07:42 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0813 (0.0813) acc1: 100.0000 (100.0000) acc5: 100.0000 (100.0000) time: 2.8954 data: 2.7989 max mem: 52126 +Test: [ 10/1403] eta: 0:08:08 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1309 (0.1598) acc1: 98.4375 (96.1648) acc5: 100.0000 (100.0000) time: 0.3504 data: 0.2547 max mem: 52126 +Test: [ 20/1403] eta: 0:05:17 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1258 (0.1434) acc1: 98.4375 (96.6518) acc5: 100.0000 (100.0000) time: 0.0959 data: 0.0003 max mem: 52126 +Test: [ 30/1403] eta: 0:04:15 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0853 (0.1345) acc1: 100.0000 (97.0262) acc5: 100.0000 (100.0000) time: 0.0959 data: 0.0003 max mem: 52126 +Test: [ 40/1403] eta: 0:03:43 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0960 (0.1369) acc1: 100.0000 (97.1418) acc5: 100.0000 (100.0000) time: 0.0959 data: 0.0003 max mem: 52126 +Test: [ 50/1403] eta: 0:03:24 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1192 (0.1518) acc1: 98.4375 (96.3542) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 60/1403] eta: 0:03:10 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1220 (0.1483) acc1: 100.0000 (96.7725) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 70/1403] eta: 0:03:00 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.1243 (0.1501) acc1: 98.4375 (96.5889) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 80/1403] eta: 0:02:52 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1258 (0.1543) acc1: 96.8750 (96.3542) acc5: 100.0000 (100.0000) time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 90/1403] eta: 0:02:46 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1536 (0.1706) acc1: 95.3125 (95.5529) acc5: 100.0000 (100.0000) time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 100/1403] eta: 0:02:41 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1536 (0.1684) acc1: 96.8750 (95.7302) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 110/1403] eta: 0:02:36 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1248 (0.1669) acc1: 98.4375 (95.8193) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 120/1403] eta: 0:02:32 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0763 (0.1594) acc1: 100.0000 (96.1519) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 130/1403] eta: 0:02:29 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.0760 (0.1544) acc1: 100.0000 (96.3263) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 140/1403] eta: 0:02:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0732 (0.1485) acc1: 100.0000 (96.5869) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 150/1403] eta: 0:02:23 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0716 (0.1438) acc1: 100.0000 (96.7922) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 160/1403] eta: 0:02:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0754 (0.1399) acc1: 100.0000 (96.9526) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0003 max mem: 52126 +Test: [ 170/1403] eta: 0:02:18 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0758 (0.1367) acc1: 100.0000 (97.1034) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 180/1403] eta: 0:02:16 acc1_pos: 90.625 acc5_pos: 100.000 loss: 0.0801 (0.1406) acc1: 100.0000 (96.9613) acc5: 100.0000 (100.0000) time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 190/1403] eta: 0:02:14 acc1_pos: 89.062 acc5_pos: 100.000 loss: 0.2502 (0.1475) acc1: 90.6250 (96.6296) acc5: 100.0000 (100.0000) time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 200/1403] eta: 0:02:12 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.2621 (0.1580) acc1: 89.0625 (96.1210) acc5: 100.0000 (100.0000) time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 210/1403] eta: 0:02:10 acc1_pos: 82.812 acc5_pos: 100.000 loss: 0.2812 (0.1673) acc1: 87.5000 (95.6013) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 220/1403] eta: 0:02:08 acc1_pos: 89.062 acc5_pos: 100.000 loss: 0.3088 (0.1752) acc1: 89.0625 (95.2206) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0004 max mem: 52126 +Test: [ 230/1403] eta: 0:02:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2245 (0.1717) acc1: 92.1875 (95.3801) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 240/1403] eta: 0:02:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0744 (0.1679) acc1: 100.0000 (95.5589) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 250/1403] eta: 0:02:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0728 (0.1644) acc1: 100.0000 (95.7171) acc5: 100.0000 (100.0000) time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 260/1403] eta: 0:02:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0766 (0.1612) acc1: 100.0000 (95.8633) acc5: 100.0000 (100.0000) time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 270/1403] eta: 0:02:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0766 (0.1584) acc1: 100.0000 (95.9986) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0003 max mem: 52126 +Test: [ 280/1403] eta: 0:01:59 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0762 (0.1554) acc1: 100.0000 (96.1355) acc5: 100.0000 (100.0000) time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 290/1403] eta: 0:01:57 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.0721 (0.1526) acc1: 100.0000 (96.2629) acc5: 100.0000 (100.0000) time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 300/1403] eta: 0:01:56 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0702 (0.1501) acc1: 100.0000 (96.3819) acc5: 100.0000 (100.0000) time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 310/1403] eta: 0:01:54 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0753 (0.1479) acc1: 100.0000 (96.4982) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 320/1403] eta: 0:01:53 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1007 (0.1466) acc1: 100.0000 (96.6073) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 330/1403] eta: 0:01:52 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1070 (0.1455) acc1: 100.0000 (96.7098) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 340/1403] eta: 0:01:50 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1070 (0.1443) acc1: 100.0000 (96.8063) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 350/1403] eta: 0:01:49 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1050 (0.1433) acc1: 100.0000 (96.8928) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0004 max mem: 52126 +Test: [ 360/1403] eta: 0:01:48 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1087 (0.1424) acc1: 100.0000 (96.9702) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0004 max mem: 52126 +Test: [ 370/1403] eta: 0:01:47 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1069 (0.1414) acc1: 100.0000 (97.0519) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 380/1403] eta: 0:01:45 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1057 (0.1405) acc1: 100.0000 (97.1293) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 390/1403] eta: 0:01:44 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1071 (0.1417) acc1: 100.0000 (97.0628) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 400/1403] eta: 0:01:43 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2489 (0.1456) acc1: 90.6250 (96.8399) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 410/1403] eta: 0:01:42 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2825 (0.1488) acc1: 89.0625 (96.6545) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 420/1403] eta: 0:01:41 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2371 (0.1506) acc1: 90.6250 (96.5373) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 430/1403] eta: 0:01:39 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2311 (0.1536) acc1: 90.6250 (96.3783) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 440/1403] eta: 0:01:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2945 (0.1571) acc1: 89.0625 (96.2054) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0961 data: 0.0003 max mem: 52126 +Test: [ 450/1403] eta: 0:01:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2638 (0.1595) acc1: 89.0625 (96.0851) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 460/1403] eta: 0:01:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2427 (0.1621) acc1: 90.6250 (95.9361) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52126 +Test: [ 470/1403] eta: 0:01:35 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2175 (0.1614) acc1: 92.1875 (95.9561) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [ 480/1403] eta: 0:01:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0639 (0.1593) acc1: 100.0000 (96.0402) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 490/1403] eta: 0:01:33 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0639 (0.1574) acc1: 100.0000 (96.1208) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 500/1403] eta: 0:01:31 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0640 (0.1556) acc1: 100.0000 (96.1982) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 510/1403] eta: 0:01:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0641 (0.1538) acc1: 100.0000 (96.2726) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 520/1403] eta: 0:01:29 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0679 (0.1522) acc1: 100.0000 (96.3442) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [ 530/1403] eta: 0:01:28 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0601 (0.1505) acc1: 100.0000 (96.4130) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 540/1403] eta: 0:01:27 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0595 (0.1488) acc1: 100.0000 (96.4793) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 550/1403] eta: 0:01:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0601 (0.1476) acc1: 100.0000 (96.5290) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 560/1403] eta: 0:01:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0779 (0.1465) acc1: 100.0000 (96.5770) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 570/1403] eta: 0:01:24 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0819 (0.1455) acc1: 98.4375 (96.6205) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 580/1403] eta: 0:01:23 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0892 (0.1450) acc1: 98.4375 (96.6410) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 590/1403] eta: 0:01:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1083 (0.1447) acc1: 96.8750 (96.6397) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [ 600/1403] eta: 0:01:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1282 (0.1450) acc1: 95.3125 (96.5890) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 610/1403] eta: 0:01:19 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1282 (0.1447) acc1: 95.3125 (96.5886) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 620/1403] eta: 0:01:18 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1384 (0.1457) acc1: 95.3125 (96.4875) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 630/1403] eta: 0:01:17 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1378 (0.1455) acc1: 93.7500 (96.4813) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 640/1403] eta: 0:01:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1200 (0.1451) acc1: 96.8750 (96.5020) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 650/1403] eta: 0:01:15 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1139 (0.1446) acc1: 96.8750 (96.5222) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 660/1403] eta: 0:01:14 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0885 (0.1437) acc1: 100.0000 (96.5630) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 670/1403] eta: 0:01:13 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0686 (0.1426) acc1: 100.0000 (96.6142) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 680/1403] eta: 0:01:12 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0759 (0.1436) acc1: 100.0000 (96.5469) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 690/1403] eta: 0:01:11 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2147 (0.1473) acc1: 90.6250 (96.3210) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 700/1403] eta: 0:01:10 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1684 (0.1474) acc1: 92.1875 (96.3111) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 710/1403] eta: 0:01:09 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2589 (0.1515) acc1: 87.5000 (96.0861) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 720/1403] eta: 0:01:08 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2589 (0.1515) acc1: 82.8125 (96.0840) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [ 730/1403] eta: 0:01:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0665 (0.1505) acc1: 100.0000 (96.1290) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [ 740/1403] eta: 0:01:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0765 (0.1497) acc1: 100.0000 (96.1665) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52126 +Test: [ 750/1403] eta: 0:01:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0978 (0.1511) acc1: 98.4375 (96.0865) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0965 data: 0.0003 max mem: 52126 +Test: [ 760/1403] eta: 0:01:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3058 (0.1547) acc1: 82.8125 (95.8545) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 770/1403] eta: 0:01:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4870 (0.1597) acc1: 73.4375 (95.5496) acc5: 100.0000 (100.0000) acc1_neg: 70.312 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 780/1403] eta: 0:01:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4422 (0.1627) acc1: 76.5625 (95.3845) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 790/1403] eta: 0:01:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3157 (0.1653) acc1: 85.9375 (95.2532) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 800/1403] eta: 0:01:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2279 (0.1659) acc1: 89.0625 (95.2267) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 810/1403] eta: 0:00:59 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2166 (0.1691) acc1: 89.0625 (95.0832) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 820/1403] eta: 0:00:58 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6466 (0.1783) acc1: 71.8750 (94.6902) acc5: 100.0000 (100.0000) acc1_neg: 40.625 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 830/1403] eta: 0:00:57 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.7056 (0.1811) acc1: 70.3125 (94.5472) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 840/1403] eta: 0:00:56 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5186 (0.1894) acc1: 71.8750 (94.1866) acc5: 100.0000 (100.0000) acc1_neg: 48.438 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 850/1403] eta: 0:00:55 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.0654 (0.2007) acc1: 54.6875 (93.7133) acc5: 100.0000 (100.0000) acc1_neg: 53.125 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 860/1403] eta: 0:00:54 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.0107 (0.2043) acc1: 57.8125 (93.5595) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 870/1403] eta: 0:00:52 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1215 (0.2032) acc1: 96.8750 (93.6119) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 880/1403] eta: 0:00:51 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0949 (0.2020) acc1: 98.4375 (93.6631) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 890/1403] eta: 0:00:50 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0781 (0.2006) acc1: 98.4375 (93.7272) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 900/1403] eta: 0:00:49 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0707 (0.1992) acc1: 100.0000 (93.7934) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [ 910/1403] eta: 0:00:48 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0743 (0.1982) acc1: 100.0000 (93.8358) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [ 920/1403] eta: 0:00:47 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0957 (0.1972) acc1: 98.4375 (93.8789) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [ 930/1403] eta: 0:00:46 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0920 (0.1961) acc1: 98.4375 (93.9296) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0964 data: 0.0004 max mem: 52126 +Test: [ 940/1403] eta: 0:00:45 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1076 (0.1962) acc1: 96.8750 (93.9061) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [ 950/1403] eta: 0:00:44 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2309 (0.1971) acc1: 89.0625 (93.8469) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [ 960/1403] eta: 0:00:43 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2711 (0.1977) acc1: 89.0625 (93.8053) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52126 +Test: [ 970/1403] eta: 0:00:42 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2280 (0.1986) acc1: 90.6250 (93.7484) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.0968 data: 0.0003 max mem: 52126 +Test: [ 980/1403] eta: 0:00:41 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2555 (0.1994) acc1: 89.0625 (93.6895) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.0966 data: 0.0003 max mem: 52126 +Test: [ 990/1403] eta: 0:00:40 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2497 (0.1997) acc1: 89.0625 (93.6633) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [1000/1403] eta: 0:00:39 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2231 (0.2004) acc1: 92.1875 (93.6282) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [1010/1403] eta: 0:00:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2549 (0.2012) acc1: 89.0625 (93.5846) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [1020/1403] eta: 0:00:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4088 (0.2073) acc1: 81.2500 (93.2526) acc5: 100.0000 (100.0000) acc1_neg: 34.375 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [1030/1403] eta: 0:00:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5503 (0.2092) acc1: 64.0625 (93.1332) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [1040/1403] eta: 0:00:35 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2199 (0.2090) acc1: 90.6250 (93.1331) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [1050/1403] eta: 0:00:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1695 (0.2087) acc1: 93.7500 (93.1494) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [1060/1403] eta: 0:00:33 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1228 (0.2079) acc1: 95.3125 (93.1860) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [1070/1403] eta: 0:00:32 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1336 (0.2083) acc1: 95.3125 (93.1723) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [1080/1403] eta: 0:00:31 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3372 (0.2098) acc1: 85.9375 (93.0808) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [1090/1403] eta: 0:00:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3852 (0.2115) acc1: 82.8125 (93.0010) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [1100/1403] eta: 0:00:29 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3590 (0.2123) acc1: 85.9375 (92.9553) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [1110/1403] eta: 0:00:28 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2913 (0.2130) acc1: 89.0625 (92.9202) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [1120/1403] eta: 0:00:27 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2969 (0.2138) acc1: 87.5000 (92.8761) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [1130/1403] eta: 0:00:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3151 (0.2148) acc1: 87.5000 (92.8230) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [1140/1403] eta: 0:00:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2636 (0.2152) acc1: 89.0625 (92.8024) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [1150/1403] eta: 0:00:24 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2636 (0.2159) acc1: 89.0625 (92.7577) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.0964 data: 0.0003 max mem: 52126 +Test: [1160/1403] eta: 0:00:23 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2863 (0.2167) acc1: 87.5000 (92.7258) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [1170/1403] eta: 0:00:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2623 (0.2167) acc1: 89.0625 (92.7252) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [1180/1403] eta: 0:00:21 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0696 (0.2154) acc1: 100.0000 (92.7868) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [1190/1403] eta: 0:00:21 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0691 (0.2142) acc1: 100.0000 (92.8474) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0003 max mem: 52126 +Test: [1200/1403] eta: 0:00:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0692 (0.2130) acc1: 100.0000 (92.9070) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0003 max mem: 52126 +Test: [1210/1403] eta: 0:00:19 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0685 (0.2118) acc1: 100.0000 (92.9655) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0002 max mem: 52126 +Test: [1220/1403] eta: 0:00:18 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0682 (0.2107) acc1: 100.0000 (93.0231) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0002 max mem: 52126 +Test: [1230/1403] eta: 0:00:17 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0692 (0.2095) acc1: 100.0000 (93.0798) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0002 max mem: 52126 +Test: [1240/1403] eta: 0:00:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0692 (0.2084) acc1: 100.0000 (93.1356) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0002 max mem: 52126 +Test: [1250/1403] eta: 0:00:15 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0694 (0.2073) acc1: 100.0000 (93.1904) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0002 max mem: 52126 +Test: [1260/1403] eta: 0:00:14 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0747 (0.2063) acc1: 100.0000 (93.2444) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0002 max mem: 52126 +Test: [1270/1403] eta: 0:00:13 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0758 (0.2052) acc1: 100.0000 (93.2976) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0002 max mem: 52126 +Test: [1280/1403] eta: 0:00:12 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0758 (0.2042) acc1: 100.0000 (93.3499) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0002 max mem: 52126 +Test: [1290/1403] eta: 0:00:11 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0751 (0.2032) acc1: 100.0000 (93.4014) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0002 max mem: 52126 +Test: [1300/1403] eta: 0:00:10 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0760 (0.2023) acc1: 100.0000 (93.4522) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0002 max mem: 52126 +Test: [1310/1403] eta: 0:00:09 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0763 (0.2013) acc1: 100.0000 (93.5021) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0963 data: 0.0002 max mem: 52126 +Test: [1320/1403] eta: 0:00:08 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0745 (0.2003) acc1: 100.0000 (93.5513) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0962 data: 0.0002 max mem: 52126 +Test: [1330/1403] eta: 0:00:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0743 (0.1994) acc1: 100.0000 (93.5986) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0002 max mem: 52126 +Test: [1340/1403] eta: 0:00:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0673 (0.1984) acc1: 100.0000 (93.6440) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0002 max mem: 52126 +Test: [1350/1403] eta: 0:00:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0656 (0.1975) acc1: 100.0000 (93.6875) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0002 max mem: 52126 +Test: [1360/1403] eta: 0:00:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0657 (0.1965) acc1: 100.0000 (93.7328) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0961 data: 0.0002 max mem: 52126 +Test: [1370/1403] eta: 0:00:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0663 (0.1956) acc1: 100.0000 (93.7762) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0960 data: 0.0002 max mem: 52126 +Test: [1380/1403] eta: 0:00:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0651 (0.1947) acc1: 100.0000 (93.8213) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0960 data: 0.0002 max mem: 52126 +Test: [1390/1403] eta: 0:00:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0651 (0.1937) acc1: 100.0000 (93.8657) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0960 data: 0.0002 max mem: 52126 +Test: [1400/1403] eta: 0:00:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0656 (0.1929) acc1: 100.0000 (93.9073) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0959 data: 0.0001 max mem: 52126 +Test: [1402/1403] eta: 0:00:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0659 (0.1927) acc1: 100.0000 (93.9117) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0918 data: 0.0001 max mem: 52126 +Test: Total time: 0:02:18 (0.0986 s / it) +* Overall Acc@1 93.912 Acc@5 100.000 + target == real Acc@1 96.467 Acc@5 100.000 + target == fake Acc@1 93.191 Acc@5 100.000 + loss 0.193 +Accuracy of the network on the 89729 test images: 93.9% +Max accuracy: 94.13% +Epoch: [33] [ 0/3634] eta: 6:59:13 lr: 0.000002 loss: 0.3714 (0.3714) time: 6.9218 data: 3.9275 max mem: 52126 +Epoch: [33] [ 10/3634] eta: 1:17:37 lr: 0.000002 loss: 0.4826 (0.4800) time: 1.2851 data: 0.3572 max mem: 52126 +Epoch: [33] [ 20/3634] eta: 0:55:00 lr: 0.000002 loss: 0.5174 (0.5040) time: 0.6130 data: 0.0001 max mem: 52126 +Epoch: [33] [ 30/3634] eta: 0:51:17 lr: 0.000002 loss: 0.5281 (0.5137) time: 0.6169 data: 0.0002 max mem: 52126 +Epoch: [33] [ 40/3634] eta: 0:49:08 lr: 0.000002 loss: 0.5257 (0.5092) time: 0.7225 data: 0.0002 max mem: 52126 +Epoch: [33] [ 50/3634] eta: 0:47:32 lr: 0.000002 loss: 0.5075 (0.5085) time: 0.7056 data: 0.0001 max mem: 52126 +Epoch: [33] [ 60/3634] eta: 0:44:17 lr: 0.000002 loss: 0.5217 (0.5146) time: 0.5859 data: 0.0002 max mem: 52126 +Epoch: [33] [ 70/3634] eta: 0:42:09 lr: 0.000002 loss: 0.5475 (0.5148) time: 0.4903 data: 0.0002 max mem: 52126 +Epoch: [33] [ 80/3634] eta: 0:41:43 lr: 0.000002 loss: 0.5121 (0.5112) time: 0.5855 data: 0.0002 max mem: 52126 +Epoch: [33] [ 90/3634] eta: 0:40:48 lr: 0.000002 loss: 0.5049 (0.5093) time: 0.6236 data: 0.0001 max mem: 52126 +Epoch: [33] [ 100/3634] eta: 0:39:25 lr: 0.000002 loss: 0.5409 (0.5128) time: 0.5273 data: 0.0002 max mem: 52126 +Epoch: [33] [ 110/3634] eta: 0:37:47 lr: 0.000002 loss: 0.5416 (0.5137) time: 0.4277 data: 0.0002 max mem: 52126 +Epoch: [33] [ 120/3634] eta: 0:36:25 lr: 0.000002 loss: 0.5153 (0.5131) time: 0.3819 data: 0.0001 max mem: 52126 +Epoch: [33] [ 130/3634] eta: 0:35:48 lr: 0.000002 loss: 0.5044 (0.5103) time: 0.4447 data: 0.0001 max mem: 52126 +Epoch: [33] [ 140/3634] eta: 0:36:38 lr: 0.000002 loss: 0.4991 (0.5096) time: 0.6733 data: 0.0001 max mem: 52126 +Epoch: [33] [ 150/3634] eta: 0:35:34 lr: 0.000002 loss: 0.5305 (0.5104) time: 0.6109 data: 0.0001 max mem: 52126 +Epoch: [33] [ 160/3634] eta: 0:35:25 lr: 0.000002 loss: 0.5305 (0.5105) time: 0.4906 data: 0.0001 max mem: 52126 +Epoch: [33] [ 170/3634] eta: 0:34:32 lr: 0.000002 loss: 0.5213 (0.5111) time: 0.4896 data: 0.0002 max mem: 52126 +Epoch: [33] [ 180/3634] eta: 0:34:08 lr: 0.000002 loss: 0.5233 (0.5121) time: 0.4416 data: 0.0002 max mem: 52126 +Epoch: [33] [ 190/3634] eta: 0:33:58 lr: 0.000002 loss: 0.5435 (0.5127) time: 0.5356 data: 0.0002 max mem: 52126 +Epoch: [33] [ 200/3634] eta: 0:33:32 lr: 0.000002 loss: 0.4808 (0.5104) time: 0.5213 data: 0.0002 max mem: 52126 +Epoch: [33] [ 210/3634] eta: 0:32:53 lr: 0.000002 loss: 0.4808 (0.5113) time: 0.4279 data: 0.0001 max mem: 52126 +Epoch: [33] [ 220/3634] eta: 0:32:17 lr: 0.000002 loss: 0.5052 (0.5105) time: 0.3815 data: 0.0001 max mem: 52126 +Epoch: [33] [ 230/3634] eta: 0:32:02 lr: 0.000002 loss: 0.5052 (0.5111) time: 0.4437 data: 0.0001 max mem: 52126 +Epoch: [33] [ 240/3634] eta: 0:31:31 lr: 0.000002 loss: 0.5210 (0.5113) time: 0.4438 data: 0.0001 max mem: 52126 +Epoch: [33] [ 250/3634] eta: 0:31:01 lr: 0.000002 loss: 0.5183 (0.5115) time: 0.3817 data: 0.0001 max mem: 52126 +Epoch: [33] [ 260/3634] eta: 0:30:34 lr: 0.000002 loss: 0.5112 (0.5116) time: 0.3812 data: 0.0002 max mem: 52126 +Epoch: [33] [ 270/3634] eta: 0:30:08 lr: 0.000002 loss: 0.5116 (0.5122) time: 0.3810 data: 0.0001 max mem: 52126 +Epoch: [33] [ 280/3634] eta: 0:29:59 lr: 0.000002 loss: 0.5178 (0.5119) time: 0.4414 data: 0.0001 max mem: 52126 +Epoch: [33] [ 290/3634] eta: 0:29:47 lr: 0.000002 loss: 0.4890 (0.5108) time: 0.4915 data: 0.0001 max mem: 52126 +Epoch: [33] [ 300/3634] eta: 0:29:25 lr: 0.000002 loss: 0.5078 (0.5101) time: 0.4313 data: 0.0001 max mem: 52126 +Epoch: [33] [ 310/3634] eta: 0:29:04 lr: 0.000002 loss: 0.5340 (0.5101) time: 0.3816 data: 0.0002 max mem: 52126 +Epoch: [33] [ 320/3634] eta: 0:28:44 lr: 0.000002 loss: 0.5779 (0.5120) time: 0.3819 data: 0.0002 max mem: 52126 +Epoch: [33] [ 330/3634] eta: 0:28:34 lr: 0.000002 loss: 0.5411 (0.5122) time: 0.4268 data: 0.0001 max mem: 52126 +Epoch: [33] [ 340/3634] eta: 0:28:24 lr: 0.000002 loss: 0.5292 (0.5119) time: 0.4717 data: 0.0001 max mem: 52126 +Epoch: [33] [ 350/3634] eta: 0:28:27 lr: 0.000002 loss: 0.4974 (0.5116) time: 0.5372 data: 0.0001 max mem: 52126 +Epoch: [33] [ 360/3634] eta: 0:28:20 lr: 0.000002 loss: 0.4982 (0.5115) time: 0.5538 data: 0.0002 max mem: 52126 +Epoch: [33] [ 370/3634] eta: 0:28:22 lr: 0.000002 loss: 0.5188 (0.5120) time: 0.5530 data: 0.0002 max mem: 52126 +Epoch: [33] [ 380/3634] eta: 0:28:33 lr: 0.000002 loss: 0.5145 (0.5114) time: 0.6536 data: 0.0002 max mem: 52126 +Epoch: [33] [ 390/3634] eta: 0:28:15 lr: 0.000002 loss: 0.4817 (0.5108) time: 0.5438 data: 0.0002 max mem: 52126 +Epoch: [33] [ 400/3634] eta: 0:27:59 lr: 0.000002 loss: 0.4877 (0.5112) time: 0.3809 data: 0.0001 max mem: 52126 +Epoch: [33] [ 410/3634] eta: 0:27:52 lr: 0.000002 loss: 0.5260 (0.5113) time: 0.4405 data: 0.0002 max mem: 52126 +Epoch: [33] [ 420/3634] eta: 0:27:36 lr: 0.000002 loss: 0.5461 (0.5125) time: 0.4411 data: 0.0002 max mem: 52126 +Epoch: [33] [ 430/3634] eta: 0:27:30 lr: 0.000002 loss: 0.5446 (0.5126) time: 0.4425 data: 0.0002 max mem: 52126 +Epoch: [33] [ 440/3634] eta: 0:27:22 lr: 0.000002 loss: 0.5198 (0.5129) time: 0.4908 data: 0.0001 max mem: 52126 +Epoch: [33] [ 450/3634] eta: 0:27:23 lr: 0.000002 loss: 0.5198 (0.5129) time: 0.5358 data: 0.0001 max mem: 52126 +Epoch: [33] [ 460/3634] eta: 0:27:08 lr: 0.000002 loss: 0.4764 (0.5122) time: 0.4876 data: 0.0001 max mem: 52126 +Epoch: [33] [ 470/3634] eta: 0:27:07 lr: 0.000002 loss: 0.4829 (0.5118) time: 0.4758 data: 0.0002 max mem: 52126 +Epoch: [33] [ 480/3634] eta: 0:27:02 lr: 0.000002 loss: 0.5003 (0.5115) time: 0.5390 data: 0.0002 max mem: 52126 +Epoch: [33] [ 490/3634] eta: 0:26:56 lr: 0.000002 loss: 0.5366 (0.5122) time: 0.5046 data: 0.0002 max mem: 52126 +Epoch: [33] [ 500/3634] eta: 0:26:42 lr: 0.000002 loss: 0.5276 (0.5120) time: 0.4418 data: 0.0001 max mem: 52126 +Epoch: [33] [ 510/3634] eta: 0:26:37 lr: 0.000002 loss: 0.5266 (0.5122) time: 0.4445 data: 0.0001 max mem: 52126 +Epoch: [33] [ 520/3634] eta: 0:26:24 lr: 0.000002 loss: 0.5064 (0.5119) time: 0.4435 data: 0.0001 max mem: 52126 +Epoch: [33] [ 530/3634] eta: 0:26:11 lr: 0.000002 loss: 0.4954 (0.5123) time: 0.3814 data: 0.0002 max mem: 52126 +Epoch: [33] [ 540/3634] eta: 0:25:59 lr: 0.000002 loss: 0.5212 (0.5120) time: 0.3822 data: 0.0002 max mem: 52126 +Epoch: [33] [ 550/3634] eta: 0:26:11 lr: 0.000002 loss: 0.5272 (0.5127) time: 0.5894 data: 0.0001 max mem: 52126 +Epoch: [33] [ 560/3634] eta: 0:26:03 lr: 0.000002 loss: 0.5337 (0.5129) time: 0.6338 data: 0.0001 max mem: 52126 +Epoch: [33] [ 570/3634] eta: 0:25:51 lr: 0.000002 loss: 0.5283 (0.5128) time: 0.4264 data: 0.0002 max mem: 52126 +Epoch: [33] [ 580/3634] eta: 0:25:53 lr: 0.000002 loss: 0.5253 (0.5125) time: 0.5035 data: 0.0002 max mem: 52126 +Epoch: [33] [ 590/3634] eta: 0:25:41 lr: 0.000002 loss: 0.5282 (0.5128) time: 0.5039 data: 0.0001 max mem: 52126 +Epoch: [33] [ 600/3634] eta: 0:25:40 lr: 0.000002 loss: 0.5211 (0.5129) time: 0.4853 data: 0.0001 max mem: 52126 +Epoch: [33] [ 610/3634] eta: 0:25:39 lr: 0.000002 loss: 0.5065 (0.5124) time: 0.5922 data: 0.0001 max mem: 52126 +Epoch: [33] [ 620/3634] eta: 0:25:34 lr: 0.000002 loss: 0.5065 (0.5127) time: 0.5478 data: 0.0001 max mem: 52126 +Epoch: [33] [ 630/3634] eta: 0:25:34 lr: 0.000002 loss: 0.5284 (0.5136) time: 0.5644 data: 0.0002 max mem: 52126 +Epoch: [33] [ 640/3634] eta: 0:25:23 lr: 0.000002 loss: 0.5284 (0.5136) time: 0.5059 data: 0.0002 max mem: 52126 +Epoch: [33] [ 650/3634] eta: 0:25:29 lr: 0.000002 loss: 0.5253 (0.5140) time: 0.5624 data: 0.0002 max mem: 52126 +Epoch: [33] [ 660/3634] eta: 0:25:18 lr: 0.000002 loss: 0.5452 (0.5146) time: 0.5614 data: 0.0002 max mem: 52126 +Epoch: [33] [ 670/3634] eta: 0:25:07 lr: 0.000002 loss: 0.5468 (0.5147) time: 0.3813 data: 0.0002 max mem: 52126 +Epoch: [33] [ 680/3634] eta: 0:24:57 lr: 0.000002 loss: 0.5228 (0.5147) time: 0.3824 data: 0.0002 max mem: 52126 +Epoch: [33] [ 690/3634] eta: 0:24:50 lr: 0.000002 loss: 0.5038 (0.5143) time: 0.4280 data: 0.0001 max mem: 52126 +Epoch: [33] [ 700/3634] eta: 0:24:40 lr: 0.000002 loss: 0.5089 (0.5143) time: 0.4278 data: 0.0001 max mem: 52126 +Epoch: [33] [ 710/3634] eta: 0:24:40 lr: 0.000002 loss: 0.4668 (0.5137) time: 0.5026 data: 0.0002 max mem: 52126 +Epoch: [33] [ 720/3634] eta: 0:24:30 lr: 0.000002 loss: 0.4970 (0.5139) time: 0.5020 data: 0.0001 max mem: 52126 +Epoch: [33] [ 730/3634] eta: 0:24:20 lr: 0.000002 loss: 0.5188 (0.5138) time: 0.3819 data: 0.0002 max mem: 52126 +Epoch: [33] [ 740/3634] eta: 0:24:17 lr: 0.000002 loss: 0.5232 (0.5138) time: 0.4729 data: 0.0002 max mem: 52126 +Epoch: [33] [ 750/3634] eta: 0:24:12 lr: 0.000002 loss: 0.5054 (0.5132) time: 0.5316 data: 0.0001 max mem: 52126 +Epoch: [33] [ 760/3634] eta: 0:24:05 lr: 0.000002 loss: 0.4779 (0.5132) time: 0.4852 data: 0.0002 max mem: 52126 +Epoch: [33] [ 770/3634] eta: 0:24:10 lr: 0.000002 loss: 0.5095 (0.5128) time: 0.6173 data: 0.0002 max mem: 52126 +Epoch: [33] [ 780/3634] eta: 0:24:14 lr: 0.000002 loss: 0.4634 (0.5127) time: 0.7580 data: 0.0001 max mem: 52126 +Epoch: [33] [ 790/3634] eta: 0:24:04 lr: 0.000002 loss: 0.5234 (0.5128) time: 0.5664 data: 0.0001 max mem: 52126 +Epoch: [33] [ 800/3634] eta: 0:23:55 lr: 0.000002 loss: 0.5293 (0.5132) time: 0.3813 data: 0.0002 max mem: 52126 +Epoch: [33] [ 810/3634] eta: 0:23:49 lr: 0.000002 loss: 0.5498 (0.5133) time: 0.4273 data: 0.0002 max mem: 52126 +Epoch: [33] [ 820/3634] eta: 0:23:43 lr: 0.000002 loss: 0.5141 (0.5130) time: 0.4866 data: 0.0001 max mem: 52126 +Epoch: [33] [ 830/3634] eta: 0:23:40 lr: 0.000002 loss: 0.5278 (0.5135) time: 0.5305 data: 0.0001 max mem: 52126 +Epoch: [33] [ 840/3634] eta: 0:23:31 lr: 0.000002 loss: 0.5519 (0.5138) time: 0.4711 data: 0.0002 max mem: 52126 +Epoch: [33] [ 850/3634] eta: 0:23:25 lr: 0.000002 loss: 0.5086 (0.5137) time: 0.4264 data: 0.0001 max mem: 52126 +Epoch: [33] [ 860/3634] eta: 0:23:16 lr: 0.000002 loss: 0.4954 (0.5136) time: 0.4261 data: 0.0002 max mem: 52126 +Epoch: [33] [ 870/3634] eta: 0:23:07 lr: 0.000002 loss: 0.4847 (0.5131) time: 0.3800 data: 0.0002 max mem: 52126 +Epoch: [33] [ 880/3634] eta: 0:23:04 lr: 0.000002 loss: 0.4788 (0.5129) time: 0.4711 data: 0.0002 max mem: 52126 +Epoch: [33] [ 890/3634] eta: 0:22:55 lr: 0.000002 loss: 0.4998 (0.5126) time: 0.4720 data: 0.0002 max mem: 52126 +Epoch: [33] [ 900/3634] eta: 0:22:50 lr: 0.000002 loss: 0.5008 (0.5127) time: 0.4412 data: 0.0002 max mem: 52126 +Epoch: [33] [ 910/3634] eta: 0:22:47 lr: 0.000002 loss: 0.4912 (0.5125) time: 0.5318 data: 0.0001 max mem: 52126 +Epoch: [33] [ 920/3634] eta: 0:22:38 lr: 0.000002 loss: 0.4977 (0.5127) time: 0.4721 data: 0.0001 max mem: 52126 +Epoch: [33] [ 930/3634] eta: 0:22:33 lr: 0.000002 loss: 0.5169 (0.5127) time: 0.4413 data: 0.0001 max mem: 52126 +Epoch: [33] [ 940/3634] eta: 0:22:25 lr: 0.000002 loss: 0.5140 (0.5128) time: 0.4421 data: 0.0001 max mem: 52126 +Epoch: [33] [ 950/3634] eta: 0:22:16 lr: 0.000002 loss: 0.5386 (0.5132) time: 0.3824 data: 0.0002 max mem: 52126 +Epoch: [33] [ 960/3634] eta: 0:22:16 lr: 0.000002 loss: 0.5293 (0.5130) time: 0.5161 data: 0.0001 max mem: 52126 +Epoch: [33] [ 970/3634] eta: 0:22:14 lr: 0.000002 loss: 0.5151 (0.5131) time: 0.6358 data: 0.0001 max mem: 52126 +Epoch: [33] [ 980/3634] eta: 0:22:15 lr: 0.000002 loss: 0.5432 (0.5134) time: 0.6657 data: 0.0001 max mem: 52126 +Epoch: [33] [ 990/3634] eta: 0:22:06 lr: 0.000002 loss: 0.5501 (0.5137) time: 0.5460 data: 0.0001 max mem: 52126 +Epoch: [33] [1000/3634] eta: 0:22:01 lr: 0.000002 loss: 0.5127 (0.5134) time: 0.4284 data: 0.0002 max mem: 52126 +E0727 05:39:50.188000 3350668 torch/distributed/elastic/multiprocessing/api.py:874] failed (exitcode: -15) local_rank: 0 (pid: 3350746) of binary: /usr/bin/python3 +Traceback (most recent call last): + File "/usr/local/bin/torchrun", line 7, in + sys.exit(main()) + ^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 355, in wrapper + return f(*args, **kwargs) + ^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 892, in main + run(args) + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 883, in run + elastic_launch( + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 139, in __call__ + return launch_agent(self._config, self._entrypoint, list(args)) + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 270, in launch_agent + raise ChildFailedError( +torch.distributed.elastic.multiprocessing.errors.ChildFailedError: +============================================================ +main_cross_2direct_balance_realscale_perturb_v4_accum.py FAILED +------------------------------------------------------------ +Failures: +[1]: + time : 2026-07-27_05:39:50 + host : n116-040-168.byted.org + rank : 1 (local_rank: 1) + exitcode : -15 (pid: 3350747) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 3350747 +[2]: + time : 2026-07-27_05:39:50 + host : n116-040-168.byted.org + rank : 2 (local_rank: 2) + exitcode : -15 (pid: 3350748) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 3350748 +[3]: + time : 2026-07-27_05:39:50 + host : n116-040-168.byted.org + rank : 3 (local_rank: 3) + exitcode : -15 (pid: 3350749) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 3350749 +[4]: + time : 2026-07-27_05:39:50 + host : n116-040-168.byted.org + rank : 4 (local_rank: 4) + exitcode : -15 (pid: 3350750) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 3350750 +[5]: + time : 2026-07-27_05:39:50 + host : n116-040-168.byted.org + rank : 5 (local_rank: 5) + exitcode : -15 (pid: 3350751) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 3350751 +[6]: + time : 2026-07-27_05:39:50 + host : n116-040-168.byted.org + rank : 6 (local_rank: 6) + exitcode : -15 (pid: 3350752) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 3350752 +[7]: + time : 2026-07-27_05:39:50 + host : n116-040-168.byted.org + rank : 7 (local_rank: 7) + exitcode : -15 (pid: 3350753) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 3350753 +------------------------------------------------------------ +Root Cause (first observed failure): +[0]: + time : 2026-07-27_05:39:50 + host : n116-040-168.byted.org + rank : 0 (local_rank: 0) + exitcode : -15 (pid: 3350746) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 3350746 +============================================================ diff --git a/ScaleDF_rebuttal/_host_logs/20260726_2350_vith64_free8_port25264_launch/launcher_rank1_n116-041-153.log b/ScaleDF_rebuttal/_host_logs/20260726_2350_vith64_free8_port25264_launch/launcher_rank1_n116-041-153.log new file mode 100644 index 0000000000000000000000000000000000000000..86a758b64b0e462c26da518eb92fb6fc67d41e27 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_2350_vith64_free8_port25264_launch/launcher_rank1_n116-041-153.log @@ -0,0 +1,1777 @@ +started_utc=2026-07-26T23:50:25Z +hostname=n116-041-153 +node_rank=1 +nnodes=8 +nproc_per_node=8 +world_size=64 +master_addr=10.116.40.168 +master_port=25264 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +main=main_cross_2direct_balance_realscale_perturb_v4_accum.py +model_name=deit_huge_patch14_LS +pretrain=/tmp/ScaleDF_repro_20260724/pretrained/deit_3_huge_224_21k_v1.pth +resume=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/checkpoint_snapshots/vith32_latest_for_64card_effbs4096_20260726_234604.pth +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_64card_effbs4096_from_epoch22_snapshot_free8_master_n116-040-168_20260726_2350_vith64_port25264 +batch_per_gpu=64 +accum_iter=1 +effective_batch=4096 +eval_batch_per_gpu=64 +epochs=40 +W0726 23:50:38.282000 1379839 torch/distributed/run.py:766] +W0726 23:50:38.282000 1379839 torch/distributed/run.py:766] ***************************************** +W0726 23:50:38.282000 1379839 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 23:50:38.282000 1379839 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 8): env:// +| distributed init (rank 11): env:// +[rank11]:[W726 23:51:06.310556538 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 11] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 13): env:// +[rank13]:[W726 23:51:07.492230064 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 13] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 14): env:// +[rank14]:[W726 23:51:07.514957076 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 14] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank8]:[W726 23:51:07.775621934 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 8] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 9): env:// +[rank9]:[W726 23:51:07.823497268 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 9] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 15): env:// +[rank15]:[W726 23:51:07.838071305 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 15] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 10): env:// +[rank10]:[W726 23:51:07.880701782 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 10] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 12): env:// +[rank12]:[W726 23:51:07.901509122 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 12] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +n116-041-153:1379916:1379916 [0] NCCL INFO cudaDriverVersion 12060 +n116-041-153:1379916:1379916 [0] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-041-153:1379916:1379916 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-041-153:1379916:1379916 [0] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-041-153:1379916:1379916 [0] NCCL INFO Comm config Blocking set to 1 +n116-041-153:1379923:1379923 [7] NCCL INFO cudaDriverVersion 12060 +n116-041-153:1379923:1379923 [7] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-041-153:1379923:1379923 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-041-153:1379923:1379923 [7] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-041-153:1379923:1379923 [7] NCCL INFO Comm config Blocking set to 1 +n116-041-153:1379917:1379917 [1] NCCL INFO cudaDriverVersion 12060 +n116-041-153:1379917:1379917 [1] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-041-153:1379917:1379917 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-041-153:1379917:1379917 [1] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-041-153:1379917:1379917 [1] NCCL INFO Comm config Blocking set to 1 +n116-041-153:1379918:1379918 [2] NCCL INFO cudaDriverVersion 12060 +n116-041-153:1379918:1379918 [2] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-041-153:1379918:1379918 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-041-153:1379918:1379918 [2] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-041-153:1379919:1379919 [3] NCCL INFO cudaDriverVersion 12060 +n116-041-153:1379919:1379919 [3] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-041-153:1379919:1379919 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-041-153:1379919:1379919 [3] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-041-153:1379918:1379918 [2] NCCL INFO Comm config Blocking set to 1 +n116-041-153:1379922:1379922 [6] NCCL INFO cudaDriverVersion 12060 +n116-041-153:1379922:1379922 [6] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-041-153:1379922:1379922 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-041-153:1379922:1379922 [6] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-041-153:1379919:1379919 [3] NCCL INFO Comm config Blocking set to 1 +n116-041-153:1379922:1379922 [6] NCCL INFO Comm config Blocking set to 1 +n116-041-153:1379920:1379920 [4] NCCL INFO cudaDriverVersion 12060 +n116-041-153:1379920:1379920 [4] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-041-153:1379920:1379920 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-041-153:1379920:1379920 [4] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-041-153:1379920:1379920 [4] NCCL INFO Comm config Blocking set to 1 +n116-041-153:1379921:1379921 [5] NCCL INFO cudaDriverVersion 12060 +n116-041-153:1379921:1379921 [5] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-041-153:1379921:1379921 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-041-153:1379921:1379921 [5] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-041-153:1379921:1379921 [5] NCCL INFO Comm config Blocking set to 1 +n116-041-153:1379916:1379972 [0] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-041-153:1379916:1379972 [0] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-041-153:1379916:1379972 [0] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-041-153:1379916:1379972 [0] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-041-153:1379916:1379972 [0] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-041-153:1379916:1379972 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-041-153:1379916:1379972 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-041-153:1379916:1379972 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-041-153:1379916:1379972 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-041-153:1379916:1379972 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-041-153:1379923:1379973 [7] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-041-153:1379923:1379973 [7] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-041-153:1379923:1379973 [7] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-041-153:1379923:1379973 [7] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-041-153:1379923:1379973 [7] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-041-153:1379923:1379973 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-041-153:1379923:1379973 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-041-153:1379923:1379973 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-041-153:1379923:1379973 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-041-153:1379923:1379973 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-041-153:1379917:1379974 [1] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-041-153:1379917:1379974 [1] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-041-153:1379917:1379974 [1] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-041-153:1379917:1379974 [1] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-041-153:1379917:1379974 [1] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-041-153:1379917:1379974 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-041-153:1379917:1379974 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-041-153:1379917:1379974 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-041-153:1379917:1379974 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-041-153:1379917:1379974 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-041-153:1379919:1379976 [3] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-041-153:1379919:1379976 [3] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-041-153:1379919:1379976 [3] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-041-153:1379919:1379976 [3] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-041-153:1379919:1379976 [3] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-041-153:1379919:1379976 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-041-153:1379919:1379976 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-041-153:1379919:1379976 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-041-153:1379919:1379976 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-041-153:1379919:1379976 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-041-153:1379918:1379975 [2] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-041-153:1379918:1379975 [2] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-041-153:1379918:1379975 [2] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-041-153:1379918:1379975 [2] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-041-153:1379918:1379975 [2] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-041-153:1379918:1379975 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-041-153:1379918:1379975 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-041-153:1379918:1379975 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-041-153:1379918:1379975 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-041-153:1379918:1379975 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-041-153:1379922:1379977 [6] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-041-153:1379922:1379977 [6] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-041-153:1379922:1379977 [6] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-041-153:1379922:1379977 [6] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-041-153:1379922:1379977 [6] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-041-153:1379922:1379977 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-041-153:1379922:1379977 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-041-153:1379922:1379977 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-041-153:1379922:1379977 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-041-153:1379922:1379977 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-041-153:1379920:1379978 [4] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-041-153:1379920:1379978 [4] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-041-153:1379920:1379978 [4] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-041-153:1379920:1379978 [4] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-041-153:1379920:1379978 [4] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-041-153:1379920:1379978 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-041-153:1379920:1379978 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-041-153:1379920:1379978 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-041-153:1379920:1379978 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-041-153:1379920:1379978 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-041-153:1379921:1379979 [5] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-041-153:1379921:1379979 [5] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-041-153:1379921:1379979 [5] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-041-153:1379921:1379979 [5] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-041-153:1379921:1379979 [5] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-041-153:1379921:1379979 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-041-153:1379921:1379979 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-041-153:1379921:1379979 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-041-153:1379921:1379979 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-041-153:1379921:1379979 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-041-153:1379923:1379973 [7] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-041-153:1379923:1379973 [7] NCCL INFO Using network FasTrak +n116-041-153:1379917:1379974 [1] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-041-153:1379917:1379974 [1] NCCL INFO Using network FasTrak +n116-041-153:1379916:1379972 [0] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-041-153:1379916:1379972 [0] NCCL INFO Using network FasTrak +n116-041-153:1379918:1379975 [2] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-041-153:1379918:1379975 [2] NCCL INFO Using network FasTrak +n116-041-153:1379919:1379976 [3] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-041-153:1379919:1379976 [3] NCCL INFO Using network FasTrak +n116-041-153:1379922:1379977 [6] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-041-153:1379922:1379977 [6] NCCL INFO Using network FasTrak +n116-041-153:1379920:1379978 [4] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-041-153:1379920:1379978 [4] NCCL INFO Using network FasTrak +n116-041-153:1379921:1379979 [5] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-041-153:1379921:1379979 [5] NCCL INFO Using network FasTrak +n116-041-153:1379923:1379973 [7] NCCL INFO DMA-BUF is available on GPU device 7 +n116-041-153:1379923:1379973 [7] NCCL INFO ncclCommInitRankConfig comm 0xc3c7290 rank 15 nranks 64 cudaDev 7 nvmlDev 7 busId 8c000 commId 0xfb2d11625d05007a - Init START +n116-041-153:1379917:1379974 [1] NCCL INFO DMA-BUF is available on GPU device 1 +n116-041-153:1379918:1379975 [2] NCCL INFO DMA-BUF is available on GPU device 2 +n116-041-153:1379916:1379972 [0] NCCL INFO DMA-BUF is available on GPU device 0 +n116-041-153:1379920:1379978 [4] NCCL INFO DMA-BUF is available on GPU device 4 +n116-041-153:1379922:1379977 [6] NCCL INFO DMA-BUF is available on GPU device 6 +n116-041-153:1379917:1379974 [1] NCCL INFO ncclCommInitRankConfig comm 0xc46e860 rank 9 nranks 64 cudaDev 1 nvmlDev 1 busId 5000 commId 0xfb2d11625d05007a - Init START +n116-041-153:1379916:1379972 [0] NCCL INFO ncclCommInitRankConfig comm 0xc333170 rank 8 nranks 64 cudaDev 0 nvmlDev 0 busId 4000 commId 0xfb2d11625d05007a - Init START +n116-041-153:1379918:1379975 [2] NCCL INFO ncclCommInitRankConfig comm 0xc9e6060 rank 10 nranks 64 cudaDev 2 nvmlDev 2 busId b000 commId 0xfb2d11625d05007a - Init START +n116-041-153:1379922:1379977 [6] NCCL INFO ncclCommInitRankConfig comm 0xc818680 rank 14 nranks 64 cudaDev 6 nvmlDev 6 busId 8b000 commId 0xfb2d11625d05007a - Init START +n116-041-153:1379917:1379974 [1] NCCL INFO RAS client listening socket at ::1<28028> +n116-041-153:1379919:1379976 [3] NCCL INFO DMA-BUF is available on GPU device 3 +n116-041-153:1379920:1379978 [4] NCCL INFO ncclCommInitRankConfig comm 0xb74e300 rank 12 nranks 64 cudaDev 4 nvmlDev 4 busId 84000 commId 0xfb2d11625d05007a - Init START +n116-041-153:1379921:1379979 [5] NCCL INFO DMA-BUF is available on GPU device 5 +n116-041-153:1379919:1379976 [3] NCCL INFO ncclCommInitRankConfig comm 0xcaa0ca0 rank 11 nranks 64 cudaDev 3 nvmlDev 3 busId c000 commId 0xfb2d11625d05007a - Init START +n116-041-153:1379921:1379979 [5] NCCL INFO ncclCommInitRankConfig comm 0xcaa7660 rank 13 nranks 64 cudaDev 5 nvmlDev 5 busId 85000 commId 0xfb2d11625d05007a - Init START +n116-041-153:1379918:1379975 [2] NCCL INFO RAS client listening socket at ::1<28028> +n116-041-153:1379919:1379976 [3] NCCL INFO RAS client listening socket at ::1<28028> +n116-041-153:1379920:1379978 [4] NCCL INFO RAS client listening socket at ::1<28028> +n116-041-153:1379921:1379979 [5] NCCL INFO RAS client listening socket at ::1<28028> +n116-041-153:1379922:1379977 [6] NCCL INFO RAS client listening socket at ::1<28028> +n116-041-153:1379923:1379973 [7] NCCL INFO RAS client listening socket at ::1<28028> +n116-041-153:1379916:1379972 [0] NCCL INFO RAS client listening socket at ::1<28028> +n116-041-153:1379920:1379978 [4] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-041-153:1379920:1379978 [4] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-041-153:1379920:1379978 [4] NCCL INFO Retrieving state for FasTrak +n116-041-153:1379920:1379978 [4] NCCL INFO Initialized state 0 for FasTrak +n116-041-153:1379920:1379978 [4] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-041-153:1379920:1379978 [4] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-041-153:1379920:1379978 [4] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-041-153:1379919:1379976 [3] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-041-153:1379919:1379976 [3] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-041-153:1379919:1379976 [3] NCCL INFO Retrieving state for FasTrak +n116-041-153:1379919:1379976 [3] NCCL INFO Initialized state 0 for FasTrak +n116-041-153:1379920:1379978 [4] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-041-153:1379920:1379978 [4] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-041-153:1379919:1379976 [3] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-041-153:1379920:1379978 [4] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-041-153:1379919:1379976 [3] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-041-153:1379919:1379976 [3] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-041-153:1379920:1379978 [4] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-041-153:1379919:1379976 [3] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-041-153:1379920:1379978 [4] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-041-153:1379919:1379976 [3] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-041-153:1379919:1379976 [3] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-041-153:1379919:1379976 [3] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-041-153:1379919:1379976 [3] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-041-153:1379922:1379977 [6] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-041-153:1379922:1379977 [6] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-041-153:1379922:1379977 [6] NCCL INFO Retrieving state for FasTrak +n116-041-153:1379922:1379977 [6] NCCL INFO Initialized state 0 for FasTrak +n116-041-153:1379922:1379977 [6] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-041-153:1379922:1379977 [6] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-041-153:1379922:1379977 [6] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-041-153:1379922:1379977 [6] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-041-153:1379917:1379974 [1] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-041-153:1379917:1379974 [1] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-041-153:1379917:1379974 [1] NCCL INFO Retrieving state for FasTrak +n116-041-153:1379917:1379974 [1] NCCL INFO Initialized state 0 for FasTrak +n116-041-153:1379922:1379977 [6] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-041-153:1379917:1379974 [1] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-041-153:1379922:1379977 [6] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-041-153:1379917:1379974 [1] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-041-153:1379922:1379977 [6] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-041-153:1379922:1379977 [6] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-041-153:1379917:1379974 [1] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-041-153:1379917:1379974 [1] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-041-153:1379923:1379973 [7] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-041-153:1379923:1379973 [7] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-041-153:1379923:1379973 [7] NCCL INFO Retrieving state for FasTrak +n116-041-153:1379923:1379973 [7] NCCL INFO Initialized state 0 for FasTrak +n116-041-153:1379917:1379974 [1] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-041-153:1379917:1379974 [1] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-041-153:1379923:1379973 [7] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-041-153:1379917:1379974 [1] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-041-153:1379923:1379973 [7] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-041-153:1379917:1379974 [1] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-041-153:1379923:1379973 [7] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-041-153:1379921:1379979 [5] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-041-153:1379921:1379979 [5] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-041-153:1379921:1379979 [5] NCCL INFO Retrieving state for FasTrak +n116-041-153:1379921:1379979 [5] NCCL INFO Initialized state 0 for FasTrak +n116-041-153:1379923:1379973 [7] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-041-153:1379923:1379973 [7] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-041-153:1379921:1379979 [5] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-041-153:1379923:1379973 [7] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-041-153:1379921:1379979 [5] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-041-153:1379923:1379973 [7] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-041-153:1379923:1379973 [7] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-041-153:1379921:1379979 [5] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-041-153:1379918:1379975 [2] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-041-153:1379918:1379975 [2] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-041-153:1379918:1379975 [2] NCCL INFO Retrieving state for FasTrak +n116-041-153:1379921:1379979 [5] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-041-153:1379918:1379975 [2] NCCL INFO Initialized state 0 for FasTrak +n116-041-153:1379921:1379979 [5] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-041-153:1379921:1379979 [5] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-041-153:1379918:1379975 [2] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-041-153:1379918:1379975 [2] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-041-153:1379921:1379979 [5] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-041-153:1379918:1379975 [2] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-041-153:1379921:1379979 [5] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-041-153:1379918:1379975 [2] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-041-153:1379918:1379975 [2] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-041-153:1379918:1379975 [2] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-041-153:1379918:1379975 [2] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-041-153:1379918:1379975 [2] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-041-153:1379916:1379972 [0] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-041-153:1379916:1379972 [0] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-041-153:1379916:1379972 [0] NCCL INFO Retrieving state for FasTrak +n116-041-153:1379916:1379972 [0] NCCL INFO Initialized state 0 for FasTrak +n116-041-153:1379916:1379972 [0] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-041-153:1379916:1379972 [0] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-041-153:1379916:1379972 [0] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-041-153:1379916:1379972 [0] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-041-153:1379916:1379972 [0] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-041-153:1379916:1379972 [0] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-041-153:1379916:1379972 [0] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-041-153:1379916:1379972 [0] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-041-153:1379917:1379974 [1] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-041-153:1379918:1379975 [2] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-041-153:1379916:1379972 [0] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-041-153:1379923:1379973 [7] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-041-153:1379919:1379976 [3] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-041-153:1379922:1379977 [6] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-041-153:1379921:1379979 [5] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-041-153:1379920:1379978 [4] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-041-153:1379918:1379975 [2] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-041-153:1379918:1379975 [2] NCCL INFO CPU/0-0 (1/1/3) +n116-041-153:1379918:1379975 [2] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-041-153:1379918:1379975 [2] NCCL INFO + PCI[24.0] - GPU/0-4000 (8) +n116-041-153:1379918:1379975 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379918:1379975 [2] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-041-153:1379918:1379975 [2] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-041-153:1379918:1379975 [2] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-041-153:1379918:1379975 [2] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-041-153:1379918:1379975 [2] NCCL INFO + PCI[24.0] - GPU/0-5000 (9) +n116-041-153:1379918:1379975 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379918:1379975 [2] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-041-153:1379918:1379975 [2] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-041-153:1379918:1379975 [2] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-041-153:1379918:1379975 [2] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-041-153:1379918:1379975 [2] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-041-153:1379918:1379975 [2] NCCL INFO + PCI[24.0] - GPU/0-b000 (10) +n116-041-153:1379918:1379975 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379918:1379975 [2] NCCL INFO + PCI[24.0] - GPU/0-c000 (11) +n116-041-153:1379918:1379975 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379918:1379975 [2] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-041-153:1379918:1379975 [2] NCCL INFO CPU/0-1 (1/1/3) +n116-041-153:1379918:1379975 [2] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-041-153:1379918:1379975 [2] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-041-153:1379918:1379975 [2] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-041-153:1379918:1379975 [2] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-041-153:1379918:1379975 [2] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-041-153:1379918:1379975 [2] NCCL INFO + PCI[24.0] - GPU/0-84000 (12) +n116-041-153:1379918:1379975 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379918:1379975 [2] NCCL INFO + PCI[24.0] - GPU/0-85000 (13) +n116-041-153:1379918:1379975 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379918:1379975 [2] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-041-153:1379918:1379975 [2] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-041-153:1379918:1379975 [2] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-041-153:1379918:1379975 [2] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-041-153:1379918:1379975 [2] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-041-153:1379918:1379975 [2] NCCL INFO + PCI[24.0] - GPU/0-8b000 (14) +n116-041-153:1379918:1379975 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379918:1379975 [2] NCCL INFO + PCI[24.0] - GPU/0-8c000 (15) +n116-041-153:1379918:1379975 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379918:1379975 [2] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-041-153:1379918:1379975 [2] NCCL INFO ========================================== +n116-041-153:1379918:1379975 [2] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-153:1379917:1379974 [1] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-041-153:1379918:1379975 [2] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-153:1379917:1379974 [1] NCCL INFO CPU/0-0 (1/1/3) +n116-041-153:1379917:1379974 [1] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-041-153:1379917:1379974 [1] NCCL INFO + PCI[24.0] - GPU/0-4000 (8) +n116-041-153:1379918:1379975 [2] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-153:1379917:1379974 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379917:1379974 [1] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-041-153:1379917:1379974 [1] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-041-153:1379918:1379975 [2] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-153:1379917:1379974 [1] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-041-153:1379917:1379974 [1] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-041-153:1379917:1379974 [1] NCCL INFO + PCI[24.0] - GPU/0-5000 (9) +n116-041-153:1379917:1379974 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379917:1379974 [1] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-041-153:1379917:1379974 [1] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-041-153:1379918:1379975 [2] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-153:1379917:1379974 [1] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-041-153:1379917:1379974 [1] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-041-153:1379917:1379974 [1] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-041-153:1379917:1379974 [1] NCCL INFO + PCI[24.0] - GPU/0-b000 (10) +n116-041-153:1379917:1379974 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379917:1379974 [1] NCCL INFO + PCI[24.0] - GPU/0-c000 (11) +n116-041-153:1379918:1379975 [2] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-153:1379917:1379974 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379917:1379974 [1] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-041-153:1379917:1379974 [1] NCCL INFO CPU/0-1 (1/1/3) +n116-041-153:1379917:1379974 [1] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-041-153:1379917:1379974 [1] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-041-153:1379917:1379974 [1] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-041-153:1379917:1379974 [1] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-041-153:1379918:1379975 [2] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-153:1379917:1379974 [1] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-041-153:1379917:1379974 [1] NCCL INFO + PCI[24.0] - GPU/0-84000 (12) +n116-041-153:1379917:1379974 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379917:1379974 [1] NCCL INFO + PCI[24.0] - GPU/0-85000 (13) +n116-041-153:1379917:1379974 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379917:1379974 [1] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-041-153:1379918:1379975 [2] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-153:1379917:1379974 [1] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-041-153:1379917:1379974 [1] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-041-153:1379917:1379974 [1] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-041-153:1379917:1379974 [1] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-041-153:1379917:1379974 [1] NCCL INFO + PCI[24.0] - GPU/0-8b000 (14) +n116-041-153:1379917:1379974 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379917:1379974 [1] NCCL INFO + PCI[24.0] - GPU/0-8c000 (15) +n116-041-153:1379918:1379975 [2] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-153:1379917:1379974 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379917:1379974 [1] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-041-153:1379917:1379974 [1] NCCL INFO ========================================== +n116-041-153:1379918:1379975 [2] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-153:1379917:1379974 [1] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-153:1379918:1379975 [2] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-153:1379917:1379974 [1] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-153:1379918:1379975 [2] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-153:1379917:1379974 [1] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-153:1379917:1379974 [1] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-153:1379918:1379975 [2] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-153:1379918:1379975 [2] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-153:1379917:1379974 [1] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-153:1379918:1379975 [2] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-041-153:1379917:1379974 [1] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-153:1379917:1379974 [1] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-153:1379918:1379975 [2] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-041-153:1379918:1379975 [2] NCCL INFO Setting affinity for GPU 2 to 0fffffff,fffffe00,00000000,000fffff,fffffffe +n116-041-153:1379917:1379974 [1] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-153:1379917:1379974 [1] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-153:1379917:1379974 [1] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-153:1379917:1379974 [1] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-153:1379917:1379974 [1] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-153:1379917:1379974 [1] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-153:1379917:1379974 [1] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-153:1379918:1379975 [2] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-041-153:1379917:1379974 [1] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-041-153:1379917:1379974 [1] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-041-153:1379917:1379974 [1] NCCL INFO Setting affinity for GPU 1 to 0fffffff,fffffe00,00000000,000fffff,fffffffe +n116-041-153:1379918:1379975 [2] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-041-153:1379917:1379974 [1] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-041-153:1379917:1379974 [1] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-041-153:1379919:1379976 [3] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-041-153:1379919:1379976 [3] NCCL INFO CPU/0-0 (1/1/3) +n116-041-153:1379919:1379976 [3] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-041-153:1379919:1379976 [3] NCCL INFO + PCI[24.0] - GPU/0-4000 (8) +n116-041-153:1379919:1379976 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379919:1379976 [3] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-041-153:1379919:1379976 [3] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-041-153:1379919:1379976 [3] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-041-153:1379919:1379976 [3] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-041-153:1379919:1379976 [3] NCCL INFO + PCI[24.0] - GPU/0-5000 (9) +n116-041-153:1379919:1379976 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379919:1379976 [3] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-041-153:1379919:1379976 [3] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-041-153:1379919:1379976 [3] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-041-153:1379919:1379976 [3] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-041-153:1379919:1379976 [3] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-041-153:1379919:1379976 [3] NCCL INFO + PCI[24.0] - GPU/0-b000 (10) +n116-041-153:1379919:1379976 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379919:1379976 [3] NCCL INFO + PCI[24.0] - GPU/0-c000 (11) +n116-041-153:1379919:1379976 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379919:1379976 [3] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-041-153:1379919:1379976 [3] NCCL INFO CPU/0-1 (1/1/3) +n116-041-153:1379919:1379976 [3] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-041-153:1379919:1379976 [3] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-041-153:1379919:1379976 [3] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-041-153:1379919:1379976 [3] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-041-153:1379919:1379976 [3] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-041-153:1379919:1379976 [3] NCCL INFO + PCI[24.0] - GPU/0-84000 (12) +n116-041-153:1379919:1379976 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379919:1379976 [3] NCCL INFO + PCI[24.0] - GPU/0-85000 (13) +n116-041-153:1379919:1379976 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379919:1379976 [3] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-041-153:1379919:1379976 [3] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-041-153:1379919:1379976 [3] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-041-153:1379919:1379976 [3] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-041-153:1379919:1379976 [3] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-041-153:1379919:1379976 [3] NCCL INFO + PCI[24.0] - GPU/0-8b000 (14) +n116-041-153:1379919:1379976 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379919:1379976 [3] NCCL INFO + PCI[24.0] - GPU/0-8c000 (15) +n116-041-153:1379919:1379976 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379919:1379976 [3] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-041-153:1379919:1379976 [3] NCCL INFO ========================================== +n116-041-153:1379919:1379976 [3] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-153:1379919:1379976 [3] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-153:1379919:1379976 [3] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-153:1379919:1379976 [3] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-153:1379916:1379972 [0] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-041-153:1379916:1379972 [0] NCCL INFO CPU/0-0 (1/1/3) +n116-041-153:1379919:1379976 [3] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-153:1379916:1379972 [0] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-041-153:1379919:1379976 [3] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-153:1379916:1379972 [0] NCCL INFO + PCI[24.0] - GPU/0-4000 (8) +n116-041-153:1379916:1379972 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379916:1379972 [0] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-041-153:1379916:1379972 [0] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-041-153:1379916:1379972 [0] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-041-153:1379916:1379972 [0] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-041-153:1379919:1379976 [3] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-153:1379916:1379972 [0] NCCL INFO + PCI[24.0] - GPU/0-5000 (9) +n116-041-153:1379916:1379972 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379916:1379972 [0] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-041-153:1379916:1379972 [0] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-041-153:1379916:1379972 [0] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-041-153:1379916:1379972 [0] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-041-153:1379916:1379972 [0] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-041-153:1379919:1379976 [3] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-153:1379916:1379972 [0] NCCL INFO + PCI[24.0] - GPU/0-b000 (10) +n116-041-153:1379916:1379972 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379916:1379972 [0] NCCL INFO + PCI[24.0] - GPU/0-c000 (11) +n116-041-153:1379916:1379972 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379916:1379972 [0] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-041-153:1379916:1379972 [0] NCCL INFO CPU/0-1 (1/1/3) +n116-041-153:1379916:1379972 [0] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-041-153:1379916:1379972 [0] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-041-153:1379919:1379976 [3] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-153:1379916:1379972 [0] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-041-153:1379916:1379972 [0] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-041-153:1379916:1379972 [0] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-041-153:1379916:1379972 [0] NCCL INFO + PCI[24.0] - GPU/0-84000 (12) +n116-041-153:1379916:1379972 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379916:1379972 [0] NCCL INFO + PCI[24.0] - GPU/0-85000 (13) +n116-041-153:1379919:1379976 [3] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-153:1379916:1379972 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379916:1379972 [0] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-041-153:1379916:1379972 [0] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-041-153:1379916:1379972 [0] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-041-153:1379916:1379972 [0] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-041-153:1379916:1379972 [0] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-041-153:1379919:1379976 [3] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-153:1379916:1379972 [0] NCCL INFO + PCI[24.0] - GPU/0-8b000 (14) +n116-041-153:1379916:1379972 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379916:1379972 [0] NCCL INFO + PCI[24.0] - GPU/0-8c000 (15) +n116-041-153:1379923:1379973 [7] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-041-153:1379916:1379972 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379916:1379972 [0] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-041-153:1379916:1379972 [0] NCCL INFO ========================================== +n116-041-153:1379923:1379973 [7] NCCL INFO CPU/0-0 (1/1/3) +n116-041-153:1379919:1379976 [3] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-153:1379923:1379973 [7] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-041-153:1379923:1379973 [7] NCCL INFO + PCI[24.0] - GPU/0-4000 (8) +n116-041-153:1379923:1379973 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379923:1379973 [7] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-041-153:1379919:1379976 [3] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-153:1379923:1379973 [7] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-041-153:1379916:1379972 [0] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-153:1379923:1379973 [7] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-041-153:1379923:1379973 [7] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-041-153:1379923:1379973 [7] NCCL INFO + PCI[24.0] - GPU/0-5000 (9) +n116-041-153:1379923:1379973 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379923:1379973 [7] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-041-153:1379919:1379976 [3] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-153:1379923:1379973 [7] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-041-153:1379916:1379972 [0] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-153:1379923:1379973 [7] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-041-153:1379923:1379973 [7] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-041-153:1379923:1379973 [7] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-041-153:1379923:1379973 [7] NCCL INFO + PCI[24.0] - GPU/0-b000 (10) +n116-041-153:1379923:1379973 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379919:1379976 [3] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-041-153:1379916:1379972 [0] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-153:1379923:1379973 [7] NCCL INFO + PCI[24.0] - GPU/0-c000 (11) +n116-041-153:1379923:1379973 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379923:1379973 [7] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-041-153:1379923:1379973 [7] NCCL INFO CPU/0-1 (1/1/3) +n116-041-153:1379923:1379973 [7] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-041-153:1379923:1379973 [7] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-041-153:1379919:1379976 [3] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-041-153:1379923:1379973 [7] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-041-153:1379916:1379972 [0] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-153:1379923:1379973 [7] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-041-153:1379923:1379973 [7] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-041-153:1379923:1379973 [7] NCCL INFO + PCI[24.0] - GPU/0-84000 (12) +n116-041-153:1379923:1379973 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379919:1379976 [3] NCCL INFO Setting affinity for GPU 3 to 0fffffff,fffffe00,00000000,000fffff,fffffffe +n116-041-153:1379923:1379973 [7] NCCL INFO + PCI[24.0] - GPU/0-85000 (13) +n116-041-153:1379916:1379972 [0] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-153:1379923:1379973 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379923:1379973 [7] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-041-153:1379923:1379973 [7] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-041-153:1379923:1379973 [7] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-041-153:1379923:1379973 [7] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-041-153:1379919:1379976 [3] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-041-153:1379923:1379973 [7] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-041-153:1379916:1379972 [0] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-153:1379923:1379973 [7] NCCL INFO + PCI[24.0] - GPU/0-8b000 (14) +n116-041-153:1379923:1379973 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379923:1379973 [7] NCCL INFO + PCI[24.0] - GPU/0-8c000 (15) +n116-041-153:1379923:1379973 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379923:1379973 [7] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-041-153:1379923:1379973 [7] NCCL INFO ========================================== +n116-041-153:1379916:1379972 [0] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-153:1379923:1379973 [7] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-153:1379916:1379972 [0] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-153:1379923:1379973 [7] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-153:1379916:1379972 [0] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-153:1379919:1379976 [3] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-041-153:1379923:1379973 [7] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-153:1379916:1379972 [0] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-153:1379923:1379973 [7] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-153:1379923:1379973 [7] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-153:1379916:1379972 [0] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-153:1379916:1379972 [0] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-153:1379923:1379973 [7] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-153:1379916:1379972 [0] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-153:1379923:1379973 [7] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-153:1379916:1379972 [0] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-153:1379923:1379973 [7] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-153:1379916:1379972 [0] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-041-153:1379923:1379973 [7] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-153:1379923:1379973 [7] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-153:1379916:1379972 [0] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-041-153:1379916:1379972 [0] NCCL INFO Setting affinity for GPU 0 to 0fffffff,fffffe00,00000000,000fffff,fffffffe +n116-041-153:1379923:1379973 [7] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-153:1379923:1379973 [7] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-153:1379923:1379973 [7] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-153:1379923:1379973 [7] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-153:1379923:1379973 [7] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-041-153:1379923:1379973 [7] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-041-153:1379916:1379972 [0] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-041-153:1379923:1379973 [7] NCCL INFO Setting affinity for GPU 7 to 7fff,ffffffff,e0000000,0000007f,ffffffff,ffe00000,00000000 +n116-041-153:1379923:1379973 [7] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-041-153:1379916:1379972 [0] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-041-153:1379922:1379977 [6] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-041-153:1379922:1379977 [6] NCCL INFO CPU/0-0 (1/1/3) +n116-041-153:1379922:1379977 [6] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-041-153:1379923:1379973 [7] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-041-153:1379922:1379977 [6] NCCL INFO + PCI[24.0] - GPU/0-4000 (8) +n116-041-153:1379922:1379977 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379922:1379977 [6] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-041-153:1379922:1379977 [6] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-041-153:1379922:1379977 [6] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-041-153:1379922:1379977 [6] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-041-153:1379922:1379977 [6] NCCL INFO + PCI[24.0] - GPU/0-5000 (9) +n116-041-153:1379922:1379977 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379922:1379977 [6] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-041-153:1379922:1379977 [6] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-041-153:1379922:1379977 [6] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-041-153:1379922:1379977 [6] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-041-153:1379922:1379977 [6] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-041-153:1379922:1379977 [6] NCCL INFO + PCI[24.0] - GPU/0-b000 (10) +n116-041-153:1379922:1379977 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379922:1379977 [6] NCCL INFO + PCI[24.0] - GPU/0-c000 (11) +n116-041-153:1379922:1379977 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379922:1379977 [6] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-041-153:1379922:1379977 [6] NCCL INFO CPU/0-1 (1/1/3) +n116-041-153:1379922:1379977 [6] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-041-153:1379922:1379977 [6] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-041-153:1379922:1379977 [6] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-041-153:1379922:1379977 [6] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-041-153:1379922:1379977 [6] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-041-153:1379922:1379977 [6] NCCL INFO + PCI[24.0] - GPU/0-84000 (12) +n116-041-153:1379922:1379977 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379922:1379977 [6] NCCL INFO + PCI[24.0] - GPU/0-85000 (13) +n116-041-153:1379922:1379977 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379922:1379977 [6] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-041-153:1379922:1379977 [6] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-041-153:1379922:1379977 [6] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-041-153:1379922:1379977 [6] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-041-153:1379922:1379977 [6] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-041-153:1379922:1379977 [6] NCCL INFO + PCI[24.0] - GPU/0-8b000 (14) +n116-041-153:1379922:1379977 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379922:1379977 [6] NCCL INFO + PCI[24.0] - GPU/0-8c000 (15) +n116-041-153:1379922:1379977 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379922:1379977 [6] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-041-153:1379922:1379977 [6] NCCL INFO ========================================== +n116-041-153:1379922:1379977 [6] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-153:1379922:1379977 [6] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-153:1379922:1379977 [6] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-153:1379922:1379977 [6] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-153:1379922:1379977 [6] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-153:1379922:1379977 [6] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-153:1379922:1379977 [6] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-153:1379922:1379977 [6] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-153:1379922:1379977 [6] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-153:1379922:1379977 [6] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-153:1379922:1379977 [6] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-153:1379921:1379979 [5] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-041-153:1379921:1379979 [5] NCCL INFO CPU/0-0 (1/1/3) +n116-041-153:1379921:1379979 [5] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-041-153:1379921:1379979 [5] NCCL INFO + PCI[24.0] - GPU/0-4000 (8) +n116-041-153:1379921:1379979 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379921:1379979 [5] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-041-153:1379922:1379977 [6] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-153:1379921:1379979 [5] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-041-153:1379921:1379979 [5] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-041-153:1379921:1379979 [5] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-041-153:1379922:1379977 [6] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-153:1379921:1379979 [5] NCCL INFO + PCI[24.0] - GPU/0-5000 (9) +n116-041-153:1379921:1379979 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379921:1379979 [5] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-041-153:1379921:1379979 [5] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-041-153:1379921:1379979 [5] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-041-153:1379922:1379977 [6] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-153:1379921:1379979 [5] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-041-153:1379921:1379979 [5] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-041-153:1379921:1379979 [5] NCCL INFO + PCI[24.0] - GPU/0-b000 (10) +n116-041-153:1379921:1379979 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379921:1379979 [5] NCCL INFO + PCI[24.0] - GPU/0-c000 (11) +n116-041-153:1379921:1379979 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379921:1379979 [5] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-041-153:1379922:1379977 [6] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-041-153:1379921:1379979 [5] NCCL INFO CPU/0-1 (1/1/3) +n116-041-153:1379921:1379979 [5] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-041-153:1379921:1379979 [5] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-041-153:1379921:1379979 [5] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-041-153:1379921:1379979 [5] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-041-153:1379921:1379979 [5] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-041-153:1379922:1379977 [6] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-041-153:1379921:1379979 [5] NCCL INFO + PCI[24.0] - GPU/0-84000 (12) +n116-041-153:1379921:1379979 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379921:1379979 [5] NCCL INFO + PCI[24.0] - GPU/0-85000 (13) +n116-041-153:1379921:1379979 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379921:1379979 [5] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-041-153:1379921:1379979 [5] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-041-153:1379921:1379979 [5] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-041-153:1379922:1379977 [6] NCCL INFO Setting affinity for GPU 6 to 7fff,ffffffff,e0000000,0000007f,ffffffff,ffe00000,00000000 +n116-041-153:1379921:1379979 [5] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-041-153:1379921:1379979 [5] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-041-153:1379921:1379979 [5] NCCL INFO + PCI[24.0] - GPU/0-8b000 (14) +n116-041-153:1379921:1379979 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379921:1379979 [5] NCCL INFO + PCI[24.0] - GPU/0-8c000 (15) +n116-041-153:1379922:1379977 [6] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-041-153:1379921:1379979 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379921:1379979 [5] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-041-153:1379921:1379979 [5] NCCL INFO ========================================== +n116-041-153:1379921:1379979 [5] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-153:1379921:1379979 [5] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-153:1379922:1379977 [6] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-041-153:1379921:1379979 [5] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-153:1379921:1379979 [5] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-153:1379921:1379979 [5] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-153:1379921:1379979 [5] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-153:1379921:1379979 [5] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-153:1379921:1379979 [5] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-153:1379921:1379979 [5] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-153:1379921:1379979 [5] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-153:1379921:1379979 [5] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-153:1379921:1379979 [5] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-153:1379921:1379979 [5] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-153:1379921:1379979 [5] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-153:1379920:1379978 [4] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-041-153:1379921:1379979 [5] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-041-153:1379920:1379978 [4] NCCL INFO CPU/0-0 (1/1/3) +n116-041-153:1379920:1379978 [4] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-041-153:1379921:1379979 [5] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-041-153:1379920:1379978 [4] NCCL INFO + PCI[24.0] - GPU/0-4000 (8) +n116-041-153:1379920:1379978 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379920:1379978 [4] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-041-153:1379921:1379979 [5] NCCL INFO Setting affinity for GPU 5 to 7fff,ffffffff,e0000000,0000007f,ffffffff,ffe00000,00000000 +n116-041-153:1379920:1379978 [4] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-041-153:1379920:1379978 [4] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-041-153:1379920:1379978 [4] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-041-153:1379920:1379978 [4] NCCL INFO + PCI[24.0] - GPU/0-5000 (9) +n116-041-153:1379920:1379978 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379921:1379979 [5] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-041-153:1379920:1379978 [4] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-041-153:1379920:1379978 [4] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-041-153:1379920:1379978 [4] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-041-153:1379920:1379978 [4] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-041-153:1379920:1379978 [4] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-041-153:1379920:1379978 [4] NCCL INFO + PCI[24.0] - GPU/0-b000 (10) +n116-041-153:1379920:1379978 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379920:1379978 [4] NCCL INFO + PCI[24.0] - GPU/0-c000 (11) +n116-041-153:1379920:1379978 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379920:1379978 [4] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-041-153:1379920:1379978 [4] NCCL INFO CPU/0-1 (1/1/3) +n116-041-153:1379920:1379978 [4] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-041-153:1379920:1379978 [4] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-041-153:1379920:1379978 [4] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-041-153:1379920:1379978 [4] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-041-153:1379920:1379978 [4] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-041-153:1379920:1379978 [4] NCCL INFO + PCI[24.0] - GPU/0-84000 (12) +n116-041-153:1379920:1379978 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379920:1379978 [4] NCCL INFO + PCI[24.0] - GPU/0-85000 (13) +n116-041-153:1379920:1379978 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379920:1379978 [4] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-041-153:1379920:1379978 [4] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-041-153:1379920:1379978 [4] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-041-153:1379920:1379978 [4] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-041-153:1379920:1379978 [4] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-041-153:1379920:1379978 [4] NCCL INFO + PCI[24.0] - GPU/0-8b000 (14) +n116-041-153:1379920:1379978 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379920:1379978 [4] NCCL INFO + PCI[24.0] - GPU/0-8c000 (15) +n116-041-153:1379921:1379979 [5] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-041-153:1379920:1379978 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-153:1379920:1379978 [4] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-041-153:1379920:1379978 [4] NCCL INFO ========================================== +n116-041-153:1379920:1379978 [4] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-153:1379920:1379978 [4] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-153:1379920:1379978 [4] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-153:1379920:1379978 [4] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-153:1379920:1379978 [4] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-153:1379920:1379978 [4] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-153:1379920:1379978 [4] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-153:1379920:1379978 [4] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-153:1379920:1379978 [4] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-153:1379920:1379978 [4] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-153:1379920:1379978 [4] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-153:1379920:1379978 [4] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-153:1379920:1379978 [4] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-153:1379920:1379978 [4] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-153:1379920:1379978 [4] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-041-153:1379920:1379978 [4] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-041-153:1379920:1379978 [4] NCCL INFO Setting affinity for GPU 4 to 7fff,ffffffff,e0000000,0000007f,ffffffff,ffe00000,00000000 +n116-041-153:1379920:1379978 [4] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-041-153:1379920:1379978 [4] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-041-153:1379917:1379974 [1] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-153:1379917:1379974 [1] NCCL INFO 0 : NET/0-0 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 NET/0-0 +n116-041-153:1379917:1379974 [1] NCCL INFO 1 : NET/0-1 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 NET/0-1 +n116-041-153:1379917:1379974 [1] NCCL INFO 2 : NET/0-2 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 NET/0-2 +n116-041-153:1379917:1379974 [1] NCCL INFO 3 : NET/0-3 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 NET/0-3 +n116-041-153:1379917:1379974 [1] NCCL INFO 4 : NET/0-4 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 NET/0-4 +n116-041-153:1379917:1379974 [1] NCCL INFO 5 : NET/0-5 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 NET/0-5 +n116-041-153:1379917:1379974 [1] NCCL INFO 6 : NET/0-6 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 NET/0-6 +n116-041-153:1379917:1379974 [1] NCCL INFO 7 : NET/0-7 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 NET/0-7 +n116-041-153:1379917:1379974 [1] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-153:1379917:1379974 [1] NCCL INFO 0 : NET/0-0 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 NET/0-0 +n116-041-153:1379917:1379974 [1] NCCL INFO 1 : NET/0-1 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 NET/0-1 +n116-041-153:1379917:1379974 [1] NCCL INFO 2 : NET/0-2 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 NET/0-2 +n116-041-153:1379917:1379974 [1] NCCL INFO 3 : NET/0-3 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 NET/0-3 +n116-041-153:1379917:1379974 [1] NCCL INFO 4 : NET/0-4 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 NET/0-4 +n116-041-153:1379917:1379974 [1] NCCL INFO 5 : NET/0-5 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 NET/0-5 +n116-041-153:1379917:1379974 [1] NCCL INFO 6 : NET/0-6 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 NET/0-6 +n116-041-153:1379917:1379974 [1] NCCL INFO 7 : NET/0-7 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 NET/0-7 +n116-041-153:1379922:1379977 [6] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-153:1379922:1379977 [6] NCCL INFO 0 : NET/0-0 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 NET/0-0 +n116-041-153:1379922:1379977 [6] NCCL INFO 1 : NET/0-1 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 NET/0-1 +n116-041-153:1379922:1379977 [6] NCCL INFO 2 : NET/0-2 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 NET/0-2 +n116-041-153:1379922:1379977 [6] NCCL INFO 3 : NET/0-3 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 NET/0-3 +n116-041-153:1379922:1379977 [6] NCCL INFO 4 : NET/0-4 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 NET/0-4 +n116-041-153:1379922:1379977 [6] NCCL INFO 5 : NET/0-5 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 NET/0-5 +n116-041-153:1379922:1379977 [6] NCCL INFO 6 : NET/0-6 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 NET/0-6 +n116-041-153:1379922:1379977 [6] NCCL INFO 7 : NET/0-7 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 NET/0-7 +n116-041-153:1379923:1379973 [7] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-153:1379923:1379973 [7] NCCL INFO 0 : NET/0-0 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 NET/0-0 +n116-041-153:1379923:1379973 [7] NCCL INFO 1 : NET/0-1 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 NET/0-1 +n116-041-153:1379923:1379973 [7] NCCL INFO 2 : NET/0-2 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 NET/0-2 +n116-041-153:1379923:1379973 [7] NCCL INFO 3 : NET/0-3 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 NET/0-3 +n116-041-153:1379923:1379973 [7] NCCL INFO 4 : NET/0-4 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 NET/0-4 +n116-041-153:1379923:1379973 [7] NCCL INFO 5 : NET/0-5 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 NET/0-5 +n116-041-153:1379923:1379973 [7] NCCL INFO 6 : NET/0-6 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 NET/0-6 +n116-041-153:1379923:1379973 [7] NCCL INFO 7 : NET/0-7 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 NET/0-7 +n116-041-153:1379919:1379976 [3] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-153:1379919:1379976 [3] NCCL INFO 0 : NET/0-0 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 NET/0-0 +n116-041-153:1379922:1379977 [6] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-153:1379919:1379976 [3] NCCL INFO 1 : NET/0-1 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 NET/0-1 +n116-041-153:1379922:1379977 [6] NCCL INFO 0 : NET/0-0 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 NET/0-0 +n116-041-153:1379919:1379976 [3] NCCL INFO 2 : NET/0-2 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 NET/0-2 +n116-041-153:1379922:1379977 [6] NCCL INFO 1 : NET/0-1 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 NET/0-1 +n116-041-153:1379919:1379976 [3] NCCL INFO 3 : NET/0-3 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 NET/0-3 +n116-041-153:1379922:1379977 [6] NCCL INFO 2 : NET/0-2 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 NET/0-2 +n116-041-153:1379919:1379976 [3] NCCL INFO 4 : NET/0-4 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 NET/0-4 +n116-041-153:1379922:1379977 [6] NCCL INFO 3 : NET/0-3 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 NET/0-3 +n116-041-153:1379919:1379976 [3] NCCL INFO 5 : NET/0-5 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 NET/0-5 +n116-041-153:1379922:1379977 [6] NCCL INFO 4 : NET/0-4 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 NET/0-4 +n116-041-153:1379919:1379976 [3] NCCL INFO 6 : NET/0-6 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 NET/0-6 +n116-041-153:1379922:1379977 [6] NCCL INFO 5 : NET/0-5 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 NET/0-5 +n116-041-153:1379919:1379976 [3] NCCL INFO 7 : NET/0-7 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 NET/0-7 +n116-041-153:1379922:1379977 [6] NCCL INFO 6 : NET/0-6 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 NET/0-6 +n116-041-153:1379922:1379977 [6] NCCL INFO 7 : NET/0-7 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 NET/0-7 +n116-041-153:1379923:1379973 [7] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-153:1379923:1379973 [7] NCCL INFO 0 : NET/0-0 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 NET/0-0 +n116-041-153:1379923:1379973 [7] NCCL INFO 1 : NET/0-1 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 NET/0-1 +n116-041-153:1379923:1379973 [7] NCCL INFO 2 : NET/0-2 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 NET/0-2 +n116-041-153:1379923:1379973 [7] NCCL INFO 3 : NET/0-3 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 NET/0-3 +n116-041-153:1379923:1379973 [7] NCCL INFO 4 : NET/0-4 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 NET/0-4 +n116-041-153:1379923:1379973 [7] NCCL INFO 5 : NET/0-5 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 NET/0-5 +n116-041-153:1379923:1379973 [7] NCCL INFO 6 : NET/0-6 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 NET/0-6 +n116-041-153:1379923:1379973 [7] NCCL INFO 7 : NET/0-7 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 NET/0-7 +n116-041-153:1379919:1379976 [3] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-153:1379919:1379976 [3] NCCL INFO 0 : NET/0-0 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 NET/0-0 +n116-041-153:1379919:1379976 [3] NCCL INFO 1 : NET/0-1 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 NET/0-1 +n116-041-153:1379919:1379976 [3] NCCL INFO 2 : NET/0-2 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 NET/0-2 +n116-041-153:1379919:1379976 [3] NCCL INFO 3 : NET/0-3 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 NET/0-3 +n116-041-153:1379919:1379976 [3] NCCL INFO 4 : NET/0-4 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 NET/0-4 +n116-041-153:1379919:1379976 [3] NCCL INFO 5 : NET/0-5 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 NET/0-5 +n116-041-153:1379919:1379976 [3] NCCL INFO 6 : NET/0-6 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 NET/0-6 +n116-041-153:1379919:1379976 [3] NCCL INFO 7 : NET/0-7 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 NET/0-7 +n116-041-153:1379920:1379978 [4] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-153:1379920:1379978 [4] NCCL INFO 0 : NET/0-0 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 NET/0-0 +n116-041-153:1379920:1379978 [4] NCCL INFO 1 : NET/0-1 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 NET/0-1 +n116-041-153:1379920:1379978 [4] NCCL INFO 2 : NET/0-2 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 NET/0-2 +n116-041-153:1379920:1379978 [4] NCCL INFO 3 : NET/0-3 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 NET/0-3 +n116-041-153:1379920:1379978 [4] NCCL INFO 4 : NET/0-4 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 NET/0-4 +n116-041-153:1379920:1379978 [4] NCCL INFO 5 : NET/0-5 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 NET/0-5 +n116-041-153:1379920:1379978 [4] NCCL INFO 6 : NET/0-6 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 NET/0-6 +n116-041-153:1379920:1379978 [4] NCCL INFO 7 : NET/0-7 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 NET/0-7 +n116-041-153:1379921:1379979 [5] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-153:1379921:1379979 [5] NCCL INFO 0 : NET/0-0 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 NET/0-0 +n116-041-153:1379921:1379979 [5] NCCL INFO 1 : NET/0-1 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 NET/0-1 +n116-041-153:1379921:1379979 [5] NCCL INFO 2 : NET/0-2 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 NET/0-2 +n116-041-153:1379921:1379979 [5] NCCL INFO 3 : NET/0-3 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 NET/0-3 +n116-041-153:1379921:1379979 [5] NCCL INFO 4 : NET/0-4 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 NET/0-4 +n116-041-153:1379921:1379979 [5] NCCL INFO 5 : NET/0-5 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 NET/0-5 +n116-041-153:1379921:1379979 [5] NCCL INFO 6 : NET/0-6 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 NET/0-6 +n116-041-153:1379921:1379979 [5] NCCL INFO 7 : NET/0-7 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 NET/0-7 +n116-041-153:1379920:1379978 [4] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-153:1379920:1379978 [4] NCCL INFO 0 : NET/0-0 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 NET/0-0 +n116-041-153:1379920:1379978 [4] NCCL INFO 1 : NET/0-1 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 NET/0-1 +n116-041-153:1379920:1379978 [4] NCCL INFO 2 : NET/0-2 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 NET/0-2 +n116-041-153:1379920:1379978 [4] NCCL INFO 3 : NET/0-3 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 NET/0-3 +n116-041-153:1379920:1379978 [4] NCCL INFO 4 : NET/0-4 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 NET/0-4 +n116-041-153:1379920:1379978 [4] NCCL INFO 5 : NET/0-5 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 NET/0-5 +n116-041-153:1379920:1379978 [4] NCCL INFO 6 : NET/0-6 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 NET/0-6 +n116-041-153:1379920:1379978 [4] NCCL INFO 7 : NET/0-7 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 NET/0-7 +n116-041-153:1379918:1379975 [2] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-153:1379918:1379975 [2] NCCL INFO 0 : NET/0-0 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 NET/0-0 +n116-041-153:1379918:1379975 [2] NCCL INFO 1 : NET/0-1 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 NET/0-1 +n116-041-153:1379918:1379975 [2] NCCL INFO 2 : NET/0-2 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 NET/0-2 +n116-041-153:1379918:1379975 [2] NCCL INFO 3 : NET/0-3 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 NET/0-3 +n116-041-153:1379918:1379975 [2] NCCL INFO 4 : NET/0-4 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 NET/0-4 +n116-041-153:1379918:1379975 [2] NCCL INFO 5 : NET/0-5 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 NET/0-5 +n116-041-153:1379918:1379975 [2] NCCL INFO 6 : NET/0-6 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 NET/0-6 +n116-041-153:1379918:1379975 [2] NCCL INFO 7 : NET/0-7 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 NET/0-7 +n116-041-153:1379921:1379979 [5] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-153:1379921:1379979 [5] NCCL INFO 0 : NET/0-0 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 NET/0-0 +n116-041-153:1379921:1379979 [5] NCCL INFO 1 : NET/0-1 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 NET/0-1 +n116-041-153:1379921:1379979 [5] NCCL INFO 2 : NET/0-2 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 NET/0-2 +n116-041-153:1379921:1379979 [5] NCCL INFO 3 : NET/0-3 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 NET/0-3 +n116-041-153:1379921:1379979 [5] NCCL INFO 4 : NET/0-4 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 NET/0-4 +n116-041-153:1379921:1379979 [5] NCCL INFO 5 : NET/0-5 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 NET/0-5 +n116-041-153:1379921:1379979 [5] NCCL INFO 6 : NET/0-6 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 NET/0-6 +n116-041-153:1379921:1379979 [5] NCCL INFO 7 : NET/0-7 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 NET/0-7 +n116-041-153:1379918:1379975 [2] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-153:1379918:1379975 [2] NCCL INFO 0 : NET/0-0 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 NET/0-0 +n116-041-153:1379918:1379975 [2] NCCL INFO 1 : NET/0-1 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 NET/0-1 +n116-041-153:1379918:1379975 [2] NCCL INFO 2 : NET/0-2 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 NET/0-2 +n116-041-153:1379918:1379975 [2] NCCL INFO 3 : NET/0-3 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 NET/0-3 +n116-041-153:1379918:1379975 [2] NCCL INFO 4 : NET/0-4 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 NET/0-4 +n116-041-153:1379918:1379975 [2] NCCL INFO 5 : NET/0-5 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 NET/0-5 +n116-041-153:1379918:1379975 [2] NCCL INFO 6 : NET/0-6 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 NET/0-6 +n116-041-153:1379918:1379975 [2] NCCL INFO 7 : NET/0-7 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 NET/0-7 +n116-041-153:1379916:1379972 [0] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-153:1379916:1379972 [0] NCCL INFO 0 : NET/0-0 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 NET/0-0 +n116-041-153:1379916:1379972 [0] NCCL INFO 1 : NET/0-1 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 NET/0-1 +n116-041-153:1379916:1379972 [0] NCCL INFO 2 : NET/0-2 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 NET/0-2 +n116-041-153:1379916:1379972 [0] NCCL INFO 3 : NET/0-3 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 NET/0-3 +n116-041-153:1379916:1379972 [0] NCCL INFO 4 : NET/0-4 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 NET/0-4 +n116-041-153:1379916:1379972 [0] NCCL INFO 5 : NET/0-5 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 NET/0-5 +n116-041-153:1379916:1379972 [0] NCCL INFO 6 : NET/0-6 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 NET/0-6 +n116-041-153:1379916:1379972 [0] NCCL INFO 7 : NET/0-7 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 NET/0-7 +n116-041-153:1379916:1379972 [0] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-153:1379916:1379972 [0] NCCL INFO 0 : NET/0-0 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 NET/0-0 +n116-041-153:1379916:1379972 [0] NCCL INFO 1 : NET/0-1 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 NET/0-1 +n116-041-153:1379916:1379972 [0] NCCL INFO 2 : NET/0-2 GPU/10 GPU/11 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 NET/0-2 +n116-041-153:1379916:1379972 [0] NCCL INFO 3 : NET/0-3 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 GPU/13 GPU/12 NET/0-3 +n116-041-153:1379916:1379972 [0] NCCL INFO 4 : NET/0-4 GPU/12 GPU/13 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 NET/0-4 +n116-041-153:1379916:1379972 [0] NCCL INFO 5 : NET/0-5 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 GPU/15 GPU/14 NET/0-5 +n116-041-153:1379916:1379972 [0] NCCL INFO 6 : NET/0-6 GPU/14 GPU/15 GPU/8 GPU/9 GPU/10 GPU/11 GPU/12 GPU/13 NET/0-6 +n116-041-153:1379916:1379972 [0] NCCL INFO 7 : NET/0-7 GPU/15 GPU/14 GPU/13 GPU/12 GPU/11 GPU/10 GPU/9 GPU/8 NET/0-7 +n116-041-153:1379923:1379973 [7] NCCL INFO comm 0xc3c7290 rank 15 nRanks 64 nNodes 8 localRanks 8 localRank 7 MNNVL 0 +n116-041-153:1379922:1379977 [6] NCCL INFO comm 0xc818680 rank 14 nRanks 64 nNodes 8 localRanks 8 localRank 6 MNNVL 0 +n116-041-153:1379921:1379979 [5] NCCL INFO comm 0xcaa7660 rank 13 nRanks 64 nNodes 8 localRanks 8 localRank 5 MNNVL 0 +n116-041-153:1379923:1379973 [7] NCCL INFO Tree 6 : 14 -> 15 -> 8/-1/-1 +n116-041-153:1379923:1379973 [7] NCCL INFO Tree 14 : 14 -> 15 -> 8/22/-1 +n116-041-153:1379921:1379979 [5] NCCL INFO Tree 4 : 12 -> 13 -> 14/-1/-1 +n116-041-153:1379919:1379976 [3] NCCL INFO comm 0xcaa0ca0 rank 11 nRanks 64 nNodes 8 localRanks 8 localRank 3 MNNVL 0 +n116-041-153:1379921:1379979 [5] NCCL INFO Tree 12 : 12 -> 13 -> 14/20/-1 +n116-041-153:1379922:1379977 [6] NCCL INFO Tree 6 : 23 -> 14 -> 15/-1/-1 +n116-041-153:1379923:1379973 [7] NCCL INFO Tree 7 : 22 -> 15 -> 14/-1/-1 +n116-041-153:1379921:1379979 [5] NCCL INFO Tree 5 : 20 -> 13 -> 12/-1/-1 +n116-041-153:1379921:1379979 [5] NCCL INFO Tree 13 : 29 -> 13 -> 12/5/-1 +n116-041-153:1379922:1379977 [6] NCCL INFO Tree 14 : 30 -> 14 -> 15/6/-1 +n116-041-153:1379923:1379973 [7] NCCL INFO Tree 15 : 31 -> 15 -> 14/7/-1 +n116-041-153:1379922:1379977 [6] NCCL INFO Tree 7 : 15 -> 14 -> 13/-1/-1 +n116-041-153:1379920:1379978 [4] NCCL INFO comm 0xb74e300 rank 12 nRanks 64 nNodes 8 localRanks 8 localRank 4 MNNVL 0 +n116-041-153:1379922:1379977 [6] NCCL INFO Tree 15 : 15 -> 14 -> 13/23/-1 +n116-041-153:1379918:1379975 [2] NCCL INFO comm 0xc9e6060 rank 10 nRanks 64 nNodes 8 localRanks 8 localRank 2 MNNVL 0 +n116-041-153:1379917:1379974 [1] NCCL INFO comm 0xc46e860 rank 9 nRanks 64 nNodes 8 localRanks 8 localRank 1 MNNVL 0 +n116-041-153:1379921:1379979 [5] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-041-153:1379923:1379973 [7] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-041-153:1379922:1379977 [6] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-041-153:1379916:1379972 [0] NCCL INFO comm 0xc333170 rank 8 nRanks 64 nNodes 8 localRanks 8 localRank 0 MNNVL 0 +n116-041-153:1379919:1379976 [3] NCCL INFO Tree 2 : 10 -> 11 -> 12/-1/-1 +n116-041-153:1379919:1379976 [3] NCCL INFO Tree 10 : 10 -> 11 -> 12/18/-1 +n116-041-153:1379919:1379976 [3] NCCL INFO Tree 3 : 18 -> 11 -> 10/-1/-1 +n116-041-153:1379919:1379976 [3] NCCL INFO Tree 11 : 27 -> 11 -> 10/3/-1 +n116-041-153:1379917:1379974 [1] NCCL INFO Tree 0 : 8 -> 9 -> 10/-1/-1 +n116-041-153:1379917:1379974 [1] NCCL INFO Tree 8 : 8 -> 9 -> 10/16/-1 +n116-041-153:1379918:1379975 [2] NCCL INFO Tree 2 : 19 -> 10 -> 11/-1/-1 +n116-041-153:1379919:1379976 [3] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-041-153:1379920:1379978 [4] NCCL INFO Tree 4 : 21 -> 12 -> 13/-1/-1 +n116-041-153:1379918:1379975 [2] NCCL INFO Tree 10 : 26 -> 10 -> 11/2/-1 +n116-041-153:1379917:1379974 [1] NCCL INFO Tree 1 : 16 -> 9 -> 8/-1/-1 +n116-041-153:1379916:1379972 [0] NCCL INFO Tree 0 : 17 -> 8 -> 9/-1/-1 +n116-041-153:1379920:1379978 [4] NCCL INFO Tree 12 : 28 -> 12 -> 13/4/-1 +n116-041-153:1379916:1379972 [0] NCCL INFO Tree 8 : 24 -> 8 -> 9/0/-1 +n116-041-153:1379918:1379975 [2] NCCL INFO Tree 3 : 11 -> 10 -> 9/-1/-1 +n116-041-153:1379917:1379974 [1] NCCL INFO Tree 9 : 25 -> 9 -> 8/1/-1 +n116-041-153:1379920:1379978 [4] NCCL INFO Tree 5 : 13 -> 12 -> 11/-1/-1 +n116-041-153:1379916:1379972 [0] NCCL INFO Tree 1 : 9 -> 8 -> 15/-1/-1 +n116-041-153:1379918:1379975 [2] NCCL INFO Tree 11 : 11 -> 10 -> 9/19/-1 +n116-041-153:1379920:1379978 [4] NCCL INFO Tree 13 : 13 -> 12 -> 11/21/-1 +n116-041-153:1379921:1379979 [5] NCCL INFO Ring 00 : 14 -> 13 -> 12 +n116-041-153:1379922:1379977 [6] NCCL INFO Ring 00 : 15 -> 14 -> 13 +n116-041-153:1379923:1379973 [7] NCCL INFO Ring 00 : 8 -> 15 -> 14 +n116-041-153:1379917:1379974 [1] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-041-153:1379918:1379975 [2] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-041-153:1379916:1379972 [0] NCCL INFO Tree 9 : 9 -> 8 -> 15/17/-1 +n116-041-153:1379921:1379979 [5] NCCL INFO Ring 01 : 12 -> 13 -> 14 +n116-041-153:1379922:1379977 [6] NCCL INFO Ring 01 : 13 -> 14 -> 15 +n116-041-153:1379921:1379979 [5] NCCL INFO Ring 02 : 14 -> 13 -> 12 +n116-041-153:1379923:1379973 [7] NCCL INFO Ring 01 : 14 -> 15 -> 8 +n116-041-153:1379920:1379978 [4] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-041-153:1379921:1379979 [5] NCCL INFO Ring 03 : 12 -> 13 -> 14 +n116-041-153:1379923:1379973 [7] NCCL INFO Ring 02 : 8 -> 15 -> 14 +n116-041-153:1379922:1379977 [6] NCCL INFO Ring 02 : 15 -> 14 -> 13 +n116-041-153:1379921:1379979 [5] NCCL INFO Ring 04 : 14 -> 13 -> 20 +n116-041-153:1379923:1379973 [7] NCCL INFO Ring 03 : 14 -> 15 -> 8 +n116-041-153:1379922:1379977 [6] NCCL INFO Ring 03 : 13 -> 14 -> 15 +n116-041-153:1379919:1379976 [3] NCCL INFO Ring 00 : 12 -> 11 -> 10 +n116-041-153:1379916:1379972 [0] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-041-153:1379921:1379979 [5] NCCL INFO Ring 05 : 4 -> 13 -> 14 +n116-041-153:1379922:1379977 [6] NCCL INFO Ring 04 : 15 -> 14 -> 13 +n116-041-153:1379921:1379979 [5] NCCL INFO Ring 06 : 14 -> 13 -> 12 +n116-041-153:1379923:1379973 [7] NCCL INFO Ring 04 : 8 -> 15 -> 14 +n116-041-153:1379922:1379977 [6] NCCL INFO Ring 05 : 13 -> 14 -> 15 +n116-041-153:1379921:1379979 [5] NCCL INFO Ring 07 : 12 -> 13 -> 14 +n116-041-153:1379921:1379979 [5] NCCL INFO Ring 08 : 14 -> 13 -> 12 +n116-041-153:1379919:1379976 [3] NCCL INFO Ring 01 : 10 -> 11 -> 12 +n116-041-153:1379923:1379973 [7] NCCL INFO Ring 05 : 14 -> 15 -> 8 +n116-041-153:1379922:1379977 [6] NCCL INFO Ring 06 : 7 -> 14 -> 13 +n116-041-153:1379919:1379976 [3] NCCL INFO Ring 02 : 12 -> 11 -> 18 +n116-041-153:1379923:1379973 [7] NCCL INFO Ring 06 : 8 -> 15 -> 22 +n116-041-153:1379921:1379979 [5] NCCL INFO Ring 09 : 12 -> 13 -> 14 +n116-041-153:1379922:1379977 [6] NCCL INFO Ring 07 : 13 -> 14 -> 23 +n116-041-153:1379921:1379979 [5] NCCL INFO Ring 10 : 14 -> 13 -> 12 +n116-041-153:1379919:1379976 [3] NCCL INFO Ring 03 : 2 -> 11 -> 12 +n116-041-153:1379923:1379973 [7] NCCL INFO Ring 07 : 6 -> 15 -> 8 +n116-041-153:1379917:1379974 [1] NCCL INFO Ring 00 : 10 -> 9 -> 16 +n116-041-153:1379922:1379977 [6] NCCL INFO Ring 08 : 15 -> 14 -> 13 +n116-041-153:1379919:1379976 [3] NCCL INFO Ring 04 : 12 -> 11 -> 10 +n116-041-153:1379918:1379975 [2] NCCL INFO Ring 00 : 11 -> 10 -> 9 +n116-041-153:1379917:1379974 [1] NCCL INFO Ring 01 : 0 -> 9 -> 10 +n116-041-153:1379919:1379976 [3] NCCL INFO Ring 05 : 10 -> 11 -> 12 +n116-041-153:1379921:1379979 [5] NCCL INFO Ring 11 : 12 -> 13 -> 14 +n116-041-153:1379923:1379973 [7] NCCL INFO Ring 08 : 8 -> 15 -> 14 +n116-041-153:1379922:1379977 [6] NCCL INFO Ring 09 : 13 -> 14 -> 15 +n116-041-153:1379920:1379978 [4] NCCL INFO Ring 00 : 13 -> 12 -> 11 +n116-041-153:1379918:1379975 [2] NCCL INFO Ring 01 : 9 -> 10 -> 11 +n116-041-153:1379919:1379976 [3] NCCL INFO Ring 06 : 12 -> 11 -> 10 +n116-041-153:1379917:1379974 [1] NCCL INFO Ring 02 : 10 -> 9 -> 8 +n116-041-153:1379921:1379979 [5] NCCL INFO Ring 12 : 14 -> 13 -> 20 +n116-041-153:1379923:1379973 [7] NCCL INFO Ring 09 : 14 -> 15 -> 8 +n116-041-153:1379917:1379974 [1] NCCL INFO Ring 03 : 8 -> 9 -> 10 +n116-041-153:1379922:1379977 [6] NCCL INFO Ring 10 : 15 -> 14 -> 13 +n116-041-153:1379916:1379972 [0] NCCL INFO Ring 00 : 1 -> 8 -> 15 +n116-041-153:1379917:1379974 [1] NCCL INFO Ring 04 : 10 -> 9 -> 8 +n116-041-153:1379920:1379978 [4] NCCL INFO Ring 01 : 11 -> 12 -> 13 +n116-041-153:1379918:1379975 [2] NCCL INFO Ring 02 : 3 -> 10 -> 9 +n116-041-153:1379919:1379976 [3] NCCL INFO Ring 07 : 10 -> 11 -> 12 +n116-041-153:1379919:1379976 [3] NCCL INFO Ring 08 : 12 -> 11 -> 10 +n116-041-153:1379923:1379973 [7] NCCL INFO Ring 10 : 8 -> 15 -> 14 +n116-041-153:1379921:1379979 [5] NCCL INFO Ring 13 : 4 -> 13 -> 14 +n116-041-153:1379922:1379977 [6] NCCL INFO Ring 11 : 13 -> 14 -> 15 +n116-041-153:1379916:1379972 [0] NCCL INFO Ring 01 : 15 -> 8 -> 17 +n116-041-153:1379917:1379974 [1] NCCL INFO Ring 05 : 8 -> 9 -> 10 +n116-041-153:1379920:1379978 [4] NCCL INFO Ring 02 : 13 -> 12 -> 11 +n116-041-153:1379918:1379975 [2] NCCL INFO Ring 03 : 9 -> 10 -> 19 +n116-041-153:1379919:1379976 [3] NCCL INFO Ring 09 : 10 -> 11 -> 12 +n116-041-153:1379921:1379979 [5] NCCL INFO Ring 14 : 14 -> 13 -> 12 +n116-041-153:1379923:1379973 [7] NCCL INFO Ring 11 : 14 -> 15 -> 8 +n116-041-153:1379922:1379977 [6] NCCL INFO Ring 12 : 15 -> 14 -> 13 +n116-041-153:1379921:1379979 [5] NCCL INFO Ring 15 : 12 -> 13 -> 14 +n116-041-153:1379916:1379972 [0] NCCL INFO Ring 02 : 9 -> 8 -> 15 +n116-041-153:1379917:1379974 [1] NCCL INFO Ring 06 : 10 -> 9 -> 8 +n116-041-153:1379920:1379978 [4] NCCL INFO Ring 03 : 11 -> 12 -> 13 +n116-041-153:1379918:1379975 [2] NCCL INFO Ring 04 : 11 -> 10 -> 9 +n116-041-153:1379919:1379976 [3] NCCL INFO Ring 10 : 12 -> 11 -> 18 +n116-041-153:1379923:1379973 [7] NCCL INFO Ring 12 : 8 -> 15 -> 14 +n116-041-153:1379921:1379979 [5] NCCL INFO Trees [0] 14/-1/-1->13->12 [1] 12/-1/-1->13->14 [2] 14/-1/-1->13->12 [3] 12/-1/-1->13->14 [4] 14/-1/-1->13->12 [5] 12/-1/-1->13->20 [6] -1/-1/-1->13->12 [7] 12/-1/-1->13->14 [8] 14/-1/-1->13->12 [9] 12/-1/-1->13->14 [10] 14/-1/-1->13->12 [11] 12/-1/-1->13->14 [12] 14/20/-1->13->12 [13] 12/5/-1->13->29 [14] -1/-1/-1->13->12 [15] 12/-1/-1->13->14 +n116-041-153:1379922:1379977 [6] NCCL INFO Ring 13 : 13 -> 14 -> 15 +n116-041-153:1379923:1379973 [7] NCCL INFO Ring 13 : 14 -> 15 -> 8 +n116-041-153:1379916:1379972 [0] NCCL INFO Ring 03 : 15 -> 8 -> 9 +n116-041-153:1379917:1379974 [1] NCCL INFO Ring 07 : 8 -> 9 -> 10 +n116-041-153:1379920:1379978 [4] NCCL INFO Ring 04 : 5 -> 12 -> 11 +n116-041-153:1379918:1379975 [2] NCCL INFO Ring 05 : 9 -> 10 -> 11 +n116-041-153:1379919:1379976 [3] NCCL INFO Ring 11 : 2 -> 11 -> 12 +n116-041-153:1379922:1379977 [6] NCCL INFO Ring 14 : 7 -> 14 -> 13 +n116-041-153:1379923:1379973 [7] NCCL INFO Ring 14 : 8 -> 15 -> 22 +n116-041-153:1379921:1379979 [5] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-041-153:1379922:1379977 [6] NCCL INFO Ring 15 : 13 -> 14 -> 23 +n116-041-153:1379923:1379973 [7] NCCL INFO Ring 15 : 6 -> 15 -> 8 +n116-041-153:1379916:1379972 [0] NCCL INFO Ring 04 : 9 -> 8 -> 15 +n116-041-153:1379917:1379974 [1] NCCL INFO Ring 08 : 10 -> 9 -> 16 +n116-041-153:1379920:1379978 [4] NCCL INFO Ring 05 : 11 -> 12 -> 21 +n116-041-153:1379918:1379975 [2] NCCL INFO Ring 06 : 11 -> 10 -> 9 +n116-041-153:1379917:1379974 [1] NCCL INFO Ring 09 : 0 -> 9 -> 10 +n116-041-153:1379919:1379976 [3] NCCL INFO Ring 12 : 12 -> 11 -> 10 +n116-041-153:1379920:1379978 [4] NCCL INFO Ring 06 : 13 -> 12 -> 11 +n116-041-153:1379921:1379979 [5] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-041-153:1379922:1379977 [6] NCCL INFO Trees [0] 15/-1/-1->14->13 [1] 13/-1/-1->14->15 [2] 15/-1/-1->14->13 [3] 13/-1/-1->14->15 [4] 15/-1/-1->14->13 [5] -1/-1/-1->14->15 [6] 15/-1/-1->14->23 [7] 13/-1/-1->14->15 [8] 15/-1/-1->14->13 [9] 13/-1/-1->14->15 [10] 15/-1/-1->14->13 [11] 13/-1/-1->14->15 [12] 15/-1/-1->14->13 [13] -1/-1/-1->14->15 [14] 15/6/-1->14->30 [15] 13/23/-1->14->15 +n116-041-153:1379923:1379973 [7] NCCL INFO Trees [0] -1/-1/-1->15->14 [1] 14/-1/-1->15->8 [2] 8/-1/-1->15->14 [3] 14/-1/-1->15->8 [4] 8/-1/-1->15->14 [5] 14/-1/-1->15->8 [6] 8/-1/-1->15->14 [7] 14/-1/-1->15->22 [8] -1/-1/-1->15->14 [9] 14/-1/-1->15->8 [10] 8/-1/-1->15->14 [11] 14/-1/-1->15->8 [12] 8/-1/-1->15->14 [13] 14/-1/-1->15->8 [14] 8/22/-1->15->14 [15] 14/7/-1->15->31 +n116-041-153:1379921:1379979 [5] NCCL INFO P2P Chunksize set to 524288 +n116-041-153:1379916:1379972 [0] NCCL INFO Ring 05 : 15 -> 8 -> 9 +n116-041-153:1379918:1379975 [2] NCCL INFO Ring 07 : 9 -> 10 -> 11 +n116-041-153:1379917:1379974 [1] NCCL INFO Ring 10 : 10 -> 9 -> 8 +n116-041-153:1379922:1379977 [6] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-041-153:1379919:1379976 [3] NCCL INFO Ring 13 : 10 -> 11 -> 12 +n116-041-153:1379916:1379972 [0] NCCL INFO Ring 06 : 9 -> 8 -> 15 +n116-041-153:1379920:1379978 [4] NCCL INFO Ring 07 : 11 -> 12 -> 13 +n116-041-153:1379923:1379973 [7] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-041-153:1379918:1379975 [2] NCCL INFO Ring 08 : 11 -> 10 -> 9 +n116-041-153:1379916:1379972 [0] NCCL INFO Ring 07 : 15 -> 8 -> 9 +n116-041-153:1379917:1379974 [1] NCCL INFO Ring 11 : 8 -> 9 -> 10 +n116-041-153:1379919:1379976 [3] NCCL INFO Ring 14 : 12 -> 11 -> 10 +n116-041-153:1379922:1379977 [6] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-041-153:1379920:1379978 [4] NCCL INFO Ring 08 : 13 -> 12 -> 11 +n116-041-153:1379923:1379973 [7] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-041-153:1379916:1379972 [0] NCCL INFO Ring 08 : 1 -> 8 -> 15 +n116-041-153:1379918:1379975 [2] NCCL INFO Ring 09 : 9 -> 10 -> 11 +n116-041-153:1379917:1379974 [1] NCCL INFO Ring 12 : 10 -> 9 -> 8 +n116-041-153:1379919:1379976 [3] NCCL INFO Ring 15 : 10 -> 11 -> 12 +n116-041-153:1379918:1379975 [2] NCCL INFO Ring 10 : 3 -> 10 -> 9 +n116-041-153:1379922:1379977 [6] NCCL INFO P2P Chunksize set to 524288 +n116-041-153:1379920:1379978 [4] NCCL INFO Ring 09 : 11 -> 12 -> 13 +n116-041-153:1379923:1379973 [7] NCCL INFO P2P Chunksize set to 524288 +n116-041-153:1379916:1379972 [0] NCCL INFO Ring 09 : 15 -> 8 -> 17 +n116-041-153:1379917:1379974 [1] NCCL INFO Ring 13 : 8 -> 9 -> 10 +n116-041-153:1379918:1379975 [2] NCCL INFO Ring 11 : 9 -> 10 -> 19 +n116-041-153:1379919:1379976 [3] NCCL INFO Trees [0] 12/-1/-1->11->10 [1] 10/-1/-1->11->12 [2] 12/-1/-1->11->10 [3] 10/-1/-1->11->18 [4] -1/-1/-1->11->10 [5] 10/-1/-1->11->12 [6] 12/-1/-1->11->10 [7] 10/-1/-1->11->12 [8] 12/-1/-1->11->10 [9] 10/-1/-1->11->12 [10] 12/18/-1->11->10 [11] 10/3/-1->11->27 [12] -1/-1/-1->11->10 [13] 10/-1/-1->11->12 [14] 12/-1/-1->11->10 [15] 10/-1/-1->11->12 +n116-041-153:1379920:1379978 [4] NCCL INFO Ring 10 : 13 -> 12 -> 11 +n116-041-153:1379916:1379972 [0] NCCL INFO Ring 10 : 9 -> 8 -> 15 +n116-041-153:1379917:1379974 [1] NCCL INFO Ring 14 : 10 -> 9 -> 8 +n116-041-153:1379918:1379975 [2] NCCL INFO Ring 12 : 11 -> 10 -> 9 +n116-041-153:1379919:1379976 [3] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-041-153:1379916:1379972 [0] NCCL INFO Ring 11 : 15 -> 8 -> 9 +n116-041-153:1379920:1379978 [4] NCCL INFO Ring 11 : 11 -> 12 -> 13 +n116-041-153:1379917:1379974 [1] NCCL INFO Ring 15 : 8 -> 9 -> 10 +n116-041-153:1379918:1379975 [2] NCCL INFO Ring 13 : 9 -> 10 -> 11 +n116-041-153:1379919:1379976 [3] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-041-153:1379916:1379972 [0] NCCL INFO Ring 12 : 9 -> 8 -> 15 +n116-041-153:1379918:1379975 [2] NCCL INFO Ring 14 : 11 -> 10 -> 9 +n116-041-153:1379920:1379978 [4] NCCL INFO Ring 12 : 5 -> 12 -> 11 +n116-041-153:1379917:1379974 [1] NCCL INFO Trees [0] 10/-1/-1->9->8 [1] 8/-1/-1->9->16 [2] -1/-1/-1->9->8 [3] 8/-1/-1->9->10 [4] 10/-1/-1->9->8 [5] 8/-1/-1->9->10 [6] 10/-1/-1->9->8 [7] 8/-1/-1->9->10 [8] 10/16/-1->9->8 [9] 8/1/-1->9->25 [10] -1/-1/-1->9->8 [11] 8/-1/-1->9->10 [12] 10/-1/-1->9->8 [13] 8/-1/-1->9->10 [14] 10/-1/-1->9->8 [15] 8/-1/-1->9->10 +n116-041-153:1379919:1379976 [3] NCCL INFO P2P Chunksize set to 524288 +n116-041-153:1379916:1379972 [0] NCCL INFO Ring 13 : 15 -> 8 -> 9 +n116-041-153:1379918:1379975 [2] NCCL INFO Ring 15 : 9 -> 10 -> 11 +n116-041-153:1379920:1379978 [4] NCCL INFO Ring 13 : 11 -> 12 -> 21 +n116-041-153:1379918:1379975 [2] NCCL INFO Trees [0] 11/-1/-1->10->9 [1] -1/-1/-1->10->11 [2] 11/-1/-1->10->19 [3] 9/-1/-1->10->11 [4] 11/-1/-1->10->9 [5] 9/-1/-1->10->11 [6] 11/-1/-1->10->9 [7] 9/-1/-1->10->11 [8] 11/-1/-1->10->9 [9] -1/-1/-1->10->11 [10] 11/2/-1->10->26 [11] 9/19/-1->10->11 [12] 11/-1/-1->10->9 [13] 9/-1/-1->10->11 [14] 11/-1/-1->10->9 [15] 9/-1/-1->10->11 +n116-041-153:1379917:1379974 [1] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-041-153:1379916:1379972 [0] NCCL INFO Ring 14 : 9 -> 8 -> 15 +n116-041-153:1379920:1379978 [4] NCCL INFO Ring 14 : 13 -> 12 -> 11 +n116-041-153:1379917:1379974 [1] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-041-153:1379916:1379972 [0] NCCL INFO Ring 15 : 15 -> 8 -> 9 +n116-041-153:1379920:1379978 [4] NCCL INFO Ring 15 : 11 -> 12 -> 13 +n116-041-153:1379916:1379972 [0] NCCL INFO Trees [0] 9/-1/-1->8->17 [1] 15/-1/-1->8->9 [2] 9/-1/-1->8->15 [3] 15/-1/-1->8->9 [4] 9/-1/-1->8->15 [5] 15/-1/-1->8->9 [6] 9/-1/-1->8->15 [7] -1/-1/-1->8->9 [8] 9/0/-1->8->24 [9] 15/17/-1->8->9 [10] 9/-1/-1->8->15 [11] 15/-1/-1->8->9 [12] 9/-1/-1->8->15 [13] 15/-1/-1->8->9 [14] 9/-1/-1->8->15 [15] -1/-1/-1->8->9 +n116-041-153:1379920:1379978 [4] NCCL INFO Trees [0] 13/-1/-1->12->11 [1] 11/-1/-1->12->13 [2] 13/-1/-1->12->11 [3] -1/-1/-1->12->13 [4] 13/-1/-1->12->21 [5] 11/-1/-1->12->13 [6] 13/-1/-1->12->11 [7] 11/-1/-1->12->13 [8] 13/-1/-1->12->11 [9] 11/-1/-1->12->13 [10] 13/-1/-1->12->11 [11] -1/-1/-1->12->13 [12] 13/4/-1->12->28 [13] 11/21/-1->12->13 [14] 13/-1/-1->12->11 [15] 11/-1/-1->12->13 +n116-041-153:1379918:1379975 [2] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-041-153:1379917:1379974 [1] NCCL INFO P2P Chunksize set to 524288 +n116-041-153:1379916:1379972 [0] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-041-153:1379918:1379975 [2] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-041-153:1379920:1379978 [4] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-041-153:1379916:1379972 [0] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-041-153:1379918:1379975 [2] NCCL INFO P2P Chunksize set to 524288 +n116-041-153:1379920:1379978 [4] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-041-153:1379916:1379972 [0] NCCL INFO P2P Chunksize set to 524288 +n116-041-153:1379920:1379978 [4] NCCL INFO P2P Chunksize set to 524288 +n116-041-153:1379923:1380014 [7] NCCL INFO [Proxy Service] Device 7 CPU core 171 +n116-041-153:1379919:1380015 [3] NCCL INFO [Proxy Service] Device 3 CPU core 120 +n116-041-153:1379922:1380013 [6] NCCL INFO [Proxy Service] Device 6 CPU core 66 +n116-041-153:1379921:1380016 [5] NCCL INFO [Proxy Service UDS] Device 5 CPU core 56 +n116-041-153:1379923:1380018 [7] NCCL INFO [Proxy Service UDS] Device 7 CPU core 199 +n116-041-153:1379921:1380012 [5] NCCL INFO [Proxy Service] Device 5 CPU core 65 +n116-041-153:1379918:1380022 [2] NCCL INFO [Proxy Service] Device 2 CPU core 136 +n116-041-153:1379922:1380017 [6] NCCL INFO [Proxy Service UDS] Device 6 CPU core 69 +n116-041-153:1379917:1380024 [1] NCCL INFO [Proxy Service UDS] Device 1 CPU core 125 +n116-041-153:1379916:1380019 [0] NCCL INFO [Proxy Service] Device 0 CPU core 140 +n116-041-153:1379917:1380021 [1] NCCL INFO [Proxy Service] Device 1 CPU core 139 +n116-041-153:1379919:1380023 [3] NCCL INFO [Proxy Service UDS] Device 3 CPU core 35 +n116-041-153:1379918:1380026 [2] NCCL INFO [Proxy Service UDS] Device 2 CPU core 13 +n116-041-153:1379920:1380020 [4] NCCL INFO [Proxy Service] Device 4 CPU core 74 +n116-041-153:1379920:1380025 [4] NCCL INFO [Proxy Service UDS] Device 4 CPU core 68 +n116-041-153:1379916:1380027 [0] NCCL INFO [Proxy Service UDS] Device 0 CPU core 14 +n116-041-153:1379918:1379975 [2] NCCL INFO NCCL_PROTO set by environment to Simple +n116-041-153:1379918:1379975 [2] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-041-153:1379918:1379975 [2] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n116-041-153:1379918:1379975 [2] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n116-041-153:1379922:1379977 [6] NCCL INFO NCCL_PROTO set by environment to Simple +n116-041-153:1379922:1379977 [6] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-041-153:1379922:1379977 [6] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n116-041-153:1379922:1379977 [6] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n116-041-153:1379921:1379979 [5] NCCL INFO NCCL_PROTO set by environment to Simple +n116-041-153:1379921:1379979 [5] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-041-153:1379921:1379979 [5] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n116-041-153:1379921:1379979 [5] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n116-041-153:1379920:1379978 [4] NCCL INFO NCCL_PROTO set by environment to Simple +n116-041-153:1379920:1379978 [4] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-041-153:1379920:1379978 [4] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n116-041-153:1379920:1379978 [4] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n116-041-153:1379917:1379974 [1] NCCL INFO NCCL_PROTO set by environment to Simple +n116-041-153:1379917:1379974 [1] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-041-153:1379917:1379974 [1] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n116-041-153:1379917:1379974 [1] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n116-041-153:1379923:1379973 [7] NCCL INFO NCCL_PROTO set by environment to Simple +n116-041-153:1379923:1379973 [7] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-041-153:1379923:1379973 [7] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n116-041-153:1379923:1379973 [7] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n116-041-153:1379916:1379972 [0] NCCL INFO NCCL_PROTO set by environment to Simple +n116-041-153:1379916:1379972 [0] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-041-153:1379916:1379972 [0] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n116-041-153:1379916:1379972 [0] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n116-041-153:1379919:1379976 [3] NCCL INFO NCCL_PROTO set by environment to Simple +n116-041-153:1379919:1379976 [3] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-041-153:1379919:1379976 [3] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n116-041-153:1379919:1379976 [3] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n116-041-153:1379918:1379975 [2] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-041-153:1379919:1379976 [3] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-041-153:1379918:1379975 [2] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-041-153:1379918:1379975 [2] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-041-153:1379919:1379976 [3] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-041-153:1379918:1379975 [2] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-041-153:1379919:1379976 [3] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-041-153:1379916:1379972 [0] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-041-153:1379919:1379976 [3] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-041-153:1379918:1379975 [2] NCCL INFO ncclCommInitRankConfig comm 0xc9e6060 rank 10 nranks 64 cudaDev 2 nvmlDev 2 busId b000 commId 0xfb2d11625d05007a - Init COMPLETE +n116-041-153:1379916:1379972 [0] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-041-153:1379919:1379976 [3] NCCL INFO ncclCommInitRankConfig comm 0xcaa0ca0 rank 11 nranks 64 cudaDev 3 nvmlDev 3 busId c000 commId 0xfb2d11625d05007a - Init COMPLETE +n116-041-153:1379916:1379972 [0] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-041-153:1379918:1379975 [2] NCCL INFO Init timings - ncclCommInitRankConfig: rank 10 nranks 64 total 5.01 (kernels 0.31, alloc 1.55, bootstrap 2.25, allgathers 0.12, topo 0.64, graphs 0.08, connections 0.04, rest 0.01) +n116-041-153:1379916:1379972 [0] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-041-153:1379919:1379976 [3] NCCL INFO Init timings - ncclCommInitRankConfig: rank 11 nranks 64 total 5.00 (kernels 0.30, alloc 1.56, bootstrap 2.24, allgathers 0.12, topo 0.64, graphs 0.08, connections 0.05, rest 0.00) +n116-041-153:1379916:1379972 [0] NCCL INFO ncclCommInitRankConfig comm 0xc333170 rank 8 nranks 64 cudaDev 0 nvmlDev 0 busId 4000 commId 0xfb2d11625d05007a - Init COMPLETE +n116-041-153:1379917:1379974 [1] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-041-153:1379917:1379974 [1] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-041-153:1379916:1379972 [0] NCCL INFO Init timings - ncclCommInitRankConfig: rank 8 nranks 64 total 5.06 (kernels 0.30, alloc 1.62, bootstrap 2.25, allgathers 0.12, topo 0.64, graphs 0.08, connections 0.05, rest 0.00) +n116-041-153:1379917:1379974 [1] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-041-153:1379917:1379974 [1] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-041-153:1379917:1379974 [1] NCCL INFO ncclCommInitRankConfig comm 0xc46e860 rank 9 nranks 64 cudaDev 1 nvmlDev 1 busId 5000 commId 0xfb2d11625d05007a - Init COMPLETE +n116-041-153:1379922:1379977 [6] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-041-153:1379917:1379974 [1] NCCL INFO Init timings - ncclCommInitRankConfig: rank 9 nranks 64 total 5.01 (kernels 0.29, alloc 1.58, bootstrap 2.25, allgathers 0.12, topo 0.64, graphs 0.08, connections 0.05, rest 0.00) +n116-041-153:1379922:1379977 [6] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-041-153:1379922:1379977 [6] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-041-153:1379922:1379977 [6] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-041-153:1379922:1379977 [6] NCCL INFO ncclCommInitRankConfig comm 0xc818680 rank 14 nranks 64 cudaDev 6 nvmlDev 6 busId 8b000 commId 0xfb2d11625d05007a - Init COMPLETE +n116-041-153:1379923:1379973 [7] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-041-153:1379923:1379973 [7] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-041-153:1379922:1379977 [6] NCCL INFO Init timings - ncclCommInitRankConfig: rank 14 nranks 64 total 5.00 (kernels 0.33, alloc 1.52, bootstrap 2.25, allgathers 0.12, topo 0.64, graphs 0.08, connections 0.04, rest 0.01) +n116-041-153:1379923:1379973 [7] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-041-153:1379923:1379973 [7] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-041-153:1379923:1379973 [7] NCCL INFO ncclCommInitRankConfig comm 0xc3c7290 rank 15 nranks 64 cudaDev 7 nvmlDev 7 busId 8c000 commId 0xfb2d11625d05007a - Init COMPLETE +n116-041-153:1379923:1379973 [7] NCCL INFO Init timings - ncclCommInitRankConfig: rank 15 nranks 64 total 5.03 (kernels 0.29, alloc 1.41, bootstrap 2.44, allgathers 0.12, topo 0.64, graphs 0.08, connections 0.05, rest 0.00) +n116-041-153:1379920:1379978 [4] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-041-153:1379920:1379978 [4] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-041-153:1379920:1379978 [4] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-041-153:1379920:1379978 [4] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-041-153:1379920:1379978 [4] NCCL INFO ncclCommInitRankConfig comm 0xb74e300 rank 12 nranks 64 cudaDev 4 nvmlDev 4 busId 84000 commId 0xfb2d11625d05007a - Init COMPLETE +n116-041-153:1379920:1379978 [4] NCCL INFO Init timings - ncclCommInitRankConfig: rank 12 nranks 64 total 4.99 (kernels 0.33, alloc 1.52, bootstrap 2.24, allgathers 0.12, topo 0.65, graphs 0.08, connections 0.05, rest 0.00) +n116-041-153:1379921:1379979 [5] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-041-153:1379921:1379979 [5] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-041-153:1379921:1379979 [5] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-041-153:1379921:1379979 [5] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-041-153:1379921:1379979 [5] NCCL INFO ncclCommInitRankConfig comm 0xcaa7660 rank 13 nranks 64 cudaDev 5 nvmlDev 5 busId 85000 commId 0xfb2d11625d05007a - Init COMPLETE +n116-041-153:1379921:1379979 [5] NCCL INFO Init timings - ncclCommInitRankConfig: rank 13 nranks 64 total 4.99 (kernels 0.34, alloc 1.51, bootstrap 2.24, allgathers 0.12, topo 0.64, graphs 0.08, connections 0.04, rest 0.01) +n116-041-153:1379918:1380028 [2] NCCL INFO Channel 00/0 : 10[2] -> 11[3] via P2P/CUMEM +n116-041-153:1379918:1380028 [2] NCCL INFO Channel 01/0 : 10[2] -> 11[3] via P2P/CUMEM +n116-041-153:1379918:1380028 [2] NCCL INFO Channel 02/0 : 10[2] -> 11[3] via P2P/CUMEM +n116-041-153:1379918:1380028 [2] NCCL INFO Channel 03/0 : 10[2] -> 11[3] via P2P/CUMEM +n116-041-153:1379918:1380028 [2] NCCL INFO Channel 04/0 : 10[2] -> 11[3] via P2P/CUMEM +n116-041-153:1379916:1380032 [0] NCCL INFO Channel 00/0 : 8[0] -> 9[1] via P2P/CUMEM +n116-041-153:1379918:1380028 [2] NCCL INFO Channel 05/0 : 10[2] -> 11[3] via P2P/CUMEM +n116-041-153:1379919:1380029 [3] NCCL INFO Channel 00/0 : 11[3] -> 12[4] via P2P/CUMEM +n116-041-153:1379918:1380028 [2] NCCL INFO Channel 06/0 : 10[2] -> 11[3] via P2P/CUMEM +n116-041-153:1379916:1380032 [0] NCCL INFO Channel 01/0 : 8[0] -> 9[1] via P2P/CUMEM +n116-041-153:1379919:1380029 [3] NCCL INFO Channel 01/0 : 11[3] -> 12[4] via P2P/CUMEM +n116-041-153:1379918:1380028 [2] NCCL INFO Channel 07/0 : 10[2] -> 11[3] via P2P/CUMEM +n116-041-153:1379916:1380032 [0] NCCL INFO Channel 02/0 : 8[0] -> 9[1] via P2P/CUMEM +n116-041-153:1379919:1380029 [3] NCCL INFO Channel 02/0 : 11[3] -> 12[4] via P2P/CUMEM +n116-041-153:1379918:1380028 [2] NCCL INFO Channel 08/0 : 10[2] -> 11[3] via P2P/CUMEM +n116-041-153:1379916:1380032 [0] NCCL INFO Channel 03/0 : 8[0] -> 9[1] via P2P/CUMEM +n116-041-153:1379919:1380029 [3] NCCL INFO Channel 05/0 : 11[3] -> 12[4] via P2P/CUMEM +n116-041-153:1379920:1380031 [4] NCCL INFO Channel 00/0 : 12[4] -> 13[5] via P2P/CUMEM +n116-041-153:1379918:1380028 [2] NCCL INFO Channel 09/0 : 10[2] -> 11[3] via P2P/CUMEM +n116-041-153:1379916:1380032 [0] NCCL INFO Channel 04/0 : 8[0] -> 9[1] via P2P/CUMEM +n116-041-153:1379919:1380029 [3] NCCL INFO Channel 06/0 : 11[3] -> 12[4] via P2P/CUMEM +n116-041-153:1379920:1380031 [4] NCCL INFO Channel 01/0 : 12[4] -> 13[5] via P2P/CUMEM +n116-041-153:1379918:1380028 [2] NCCL INFO Channel 10/0 : 10[2] -> 11[3] via P2P/CUMEM +n116-041-153:1379916:1380032 [0] NCCL INFO Channel 05/0 : 8[0] -> 9[1] via P2P/CUMEM +n116-041-153:1379920:1380031 [4] NCCL INFO Channel 02/0 : 12[4] -> 13[5] via P2P/CUMEM +n116-041-153:1379919:1380029 [3] NCCL INFO Channel 07/0 : 11[3] -> 12[4] via P2P/CUMEM +n116-041-153:1379918:1380028 [2] NCCL INFO Channel 11/0 : 10[2] -> 11[3] via P2P/CUMEM +n116-041-153:1379916:1380032 [0] NCCL INFO Channel 06/0 : 8[0] -> 9[1] via P2P/CUMEM +n116-041-153:1379919:1380029 [3] NCCL INFO Channel 08/0 : 11[3] -> 12[4] via P2P/CUMEM +n116-041-153:1379920:1380031 [4] NCCL INFO Channel 03/0 : 12[4] -> 13[5] via P2P/CUMEM +n116-041-153:1379918:1380028 [2] NCCL INFO Channel 12/0 : 10[2] -> 11[3] via P2P/CUMEM +n116-041-153:1379916:1380032 [0] NCCL INFO Channel 07/0 : 8[0] -> 9[1] via P2P/CUMEM +n116-041-153:1379918:1380028 [2] NCCL INFO Channel 13/0 : 10[2] -> 11[3] via P2P/CUMEM +n116-041-153:1379916:1380032 [0] NCCL INFO Channel 08/0 : 8[0] -> 9[1] via P2P/CUMEM +n116-041-153:1379919:1380029 [3] NCCL INFO Channel 09/0 : 11[3] -> 12[4] via P2P/CUMEM +n116-041-153:1379920:1380031 [4] NCCL INFO Channel 04/0 : 12[4] -> 13[5] via P2P/CUMEM +n116-041-153:1379916:1380032 [0] NCCL INFO Channel 09/0 : 8[0] -> 9[1] via P2P/CUMEM +n116-041-153:1379918:1380028 [2] NCCL INFO Channel 14/0 : 10[2] -> 11[3] via P2P/CUMEM +n116-041-153:1379919:1380029 [3] NCCL INFO Channel 10/0 : 11[3] -> 12[4] via P2P/CUMEM +n116-041-153:1379920:1380031 [4] NCCL INFO Channel 05/0 : 12[4] -> 13[5] via P2P/CUMEM +n116-041-153:1379916:1380032 [0] NCCL INFO Channel 10/0 : 8[0] -> 9[1] via P2P/CUMEM +n116-041-153:1379918:1380028 [2] NCCL INFO Channel 15/0 : 10[2] -> 11[3] via P2P/CUMEM +n116-041-153:1379919:1380029 [3] NCCL INFO Channel 13/0 : 11[3] -> 12[4] via P2P/CUMEM +n116-041-153:1379920:1380031 [4] NCCL INFO Channel 06/0 : 12[4] -> 13[5] via P2P/CUMEM +n116-041-153:1379916:1380032 [0] NCCL INFO Channel 11/0 : 8[0] -> 9[1] via P2P/CUMEM +n116-041-153:1379920:1380031 [4] NCCL INFO Channel 07/0 : 12[4] -> 13[5] via P2P/CUMEM +n116-041-153:1379919:1380029 [3] NCCL INFO Channel 14/0 : 11[3] -> 12[4] via P2P/CUMEM +n116-041-153:1379916:1380032 [0] NCCL INFO Channel 12/0 : 8[0] -> 9[1] via P2P/CUMEM +n116-041-153:1379920:1380031 [4] NCCL INFO Channel 08/0 : 12[4] -> 13[5] via P2P/CUMEM +n116-041-153:1379919:1380029 [3] NCCL INFO Channel 15/0 : 11[3] -> 12[4] via P2P/CUMEM +n116-041-153:1379916:1380032 [0] NCCL INFO Channel 13/0 : 8[0] -> 9[1] via P2P/CUMEM +n116-041-153:1379920:1380031 [4] NCCL INFO Channel 09/0 : 12[4] -> 13[5] via P2P/CUMEM +n116-041-153:1379916:1380032 [0] NCCL INFO Channel 14/0 : 8[0] -> 9[1] via P2P/CUMEM +n116-041-153:1379920:1380031 [4] NCCL INFO Channel 10/0 : 12[4] -> 13[5] via P2P/CUMEM +n116-041-153:1379916:1380032 [0] NCCL INFO Channel 15/0 : 8[0] -> 9[1] via P2P/CUMEM +n116-041-153:1379920:1380031 [4] NCCL INFO Channel 11/0 : 12[4] -> 13[5] via P2P/CUMEM +n116-041-153:1379920:1380031 [4] NCCL INFO Channel 12/0 : 12[4] -> 13[5] via P2P/CUMEM +n116-041-153:1379917:1380030 [1] NCCL INFO Channel 00/0 : 9[1] -> 10[2] via P2P/CUMEM +n116-041-153:1379920:1380031 [4] NCCL INFO Channel 13/0 : 12[4] -> 13[5] via P2P/CUMEM +n116-041-153:1379917:1380030 [1] NCCL INFO Channel 03/0 : 9[1] -> 10[2] via P2P/CUMEM +n116-041-153:1379920:1380031 [4] NCCL INFO Channel 14/0 : 12[4] -> 13[5] via P2P/CUMEM +n116-041-153:1379917:1380030 [1] NCCL INFO Channel 04/0 : 9[1] -> 10[2] via P2P/CUMEM +n116-041-153:1379920:1380031 [4] NCCL INFO Channel 15/0 : 12[4] -> 13[5] via P2P/CUMEM +n116-041-153:1379921:1380034 [5] NCCL INFO Channel 00/0 : 13[5] -> 14[6] via P2P/CUMEM +n116-041-153:1379917:1380030 [1] NCCL INFO Channel 05/0 : 9[1] -> 10[2] via P2P/CUMEM +n116-041-153:1379921:1380034 [5] NCCL INFO Channel 01/0 : 13[5] -> 14[6] via P2P/CUMEM +n116-041-153:1379917:1380030 [1] NCCL INFO Channel 06/0 : 9[1] -> 10[2] via P2P/CUMEM +n116-041-153:1379921:1380034 [5] NCCL INFO Channel 02/0 : 13[5] -> 14[6] via P2P/CUMEM +n116-041-153:1379917:1380030 [1] NCCL INFO Channel 07/0 : 9[1] -> 10[2] via P2P/CUMEM +n116-041-153:1379921:1380034 [5] NCCL INFO Channel 03/0 : 13[5] -> 14[6] via P2P/CUMEM +n116-041-153:1379922:1380033 [6] NCCL INFO Channel 00/0 : 14[6] -> 15[7] via P2P/CUMEM +n116-041-153:1379917:1380030 [1] NCCL INFO Channel 08/0 : 9[1] -> 10[2] via P2P/CUMEM +n116-041-153:1379921:1380034 [5] NCCL INFO Channel 04/0 : 13[5] -> 14[6] via P2P/CUMEM +n116-041-153:1379922:1380033 [6] NCCL INFO Channel 01/0 : 14[6] -> 15[7] via P2P/CUMEM +n116-041-153:1379917:1380030 [1] NCCL INFO Channel 11/0 : 9[1] -> 10[2] via P2P/CUMEM +n116-041-153:1379921:1380034 [5] NCCL INFO Channel 07/0 : 13[5] -> 14[6] via P2P/CUMEM +n116-041-153:1379922:1380033 [6] NCCL INFO Channel 02/0 : 14[6] -> 15[7] via P2P/CUMEM +n116-041-153:1379917:1380030 [1] NCCL INFO Channel 12/0 : 9[1] -> 10[2] via P2P/CUMEM +n116-041-153:1379921:1380034 [5] NCCL INFO Channel 08/0 : 13[5] -> 14[6] via P2P/CUMEM +n116-041-153:1379922:1380033 [6] NCCL INFO Channel 03/0 : 14[6] -> 15[7] via P2P/CUMEM +n116-041-153:1379917:1380030 [1] NCCL INFO Channel 13/0 : 9[1] -> 10[2] via P2P/CUMEM +n116-041-153:1379921:1380034 [5] NCCL INFO Channel 09/0 : 13[5] -> 14[6] via P2P/CUMEM +n116-041-153:1379922:1380033 [6] NCCL INFO Channel 04/0 : 14[6] -> 15[7] via P2P/CUMEM +n116-041-153:1379917:1380030 [1] NCCL INFO Channel 14/0 : 9[1] -> 10[2] via P2P/CUMEM +n116-041-153:1379921:1380034 [5] NCCL INFO Channel 10/0 : 13[5] -> 14[6] via P2P/CUMEM +n116-041-153:1379922:1380033 [6] NCCL INFO Channel 05/0 : 14[6] -> 15[7] via P2P/CUMEM +n116-041-153:1379917:1380030 [1] NCCL INFO Channel 15/0 : 9[1] -> 10[2] via P2P/CUMEM +n116-041-153:1379921:1380034 [5] NCCL INFO Channel 11/0 : 13[5] -> 14[6] via P2P/CUMEM +n116-041-153:1379922:1380033 [6] NCCL INFO Channel 06/0 : 14[6] -> 15[7] via P2P/CUMEM +n116-041-153:1379921:1380034 [5] NCCL INFO Channel 12/0 : 13[5] -> 14[6] via P2P/CUMEM +n116-041-153:1379916:1380032 [0] NCCL INFO Channel 01/0 : 8[0] -> 15[7] via P2P/CUMEM +n116-041-153:1379922:1380033 [6] NCCL INFO Channel 07/0 : 14[6] -> 15[7] via P2P/CUMEM +n116-041-153:1379921:1380034 [5] NCCL INFO Channel 15/0 : 13[5] -> 14[6] via P2P/CUMEM +n116-041-153:1379916:1380032 [0] NCCL INFO Channel 02/0 : 8[0] -> 15[7] via P2P/CUMEM +n116-041-153:1379922:1380033 [6] NCCL INFO Channel 08/0 : 14[6] -> 15[7] via P2P/CUMEM +n116-041-153:1379916:1380032 [0] NCCL INFO Channel 03/0 : 8[0] -> 15[7] via P2P/CUMEM +n116-041-153:1379922:1380033 [6] NCCL INFO Channel 09/0 : 14[6] -> 15[7] via P2P/CUMEM +n116-041-153:1379916:1380032 [0] NCCL INFO Channel 04/0 : 8[0] -> 15[7] via P2P/CUMEM +n116-041-153:1379922:1380033 [6] NCCL INFO Channel 10/0 : 14[6] -> 15[7] via P2P/CUMEM +n116-041-153:1379916:1380032 [0] NCCL INFO Channel 05/0 : 8[0] -> 15[7] via P2P/CUMEM +n116-041-153:1379922:1380033 [6] NCCL INFO Channel 11/0 : 14[6] -> 15[7] via P2P/CUMEM +n116-041-153:1379916:1380032 [0] NCCL INFO Channel 06/0 : 8[0] -> 15[7] via P2P/CUMEM +n116-041-153:1379922:1380033 [6] NCCL INFO Channel 12/0 : 14[6] -> 15[7] via P2P/CUMEM +n116-041-153:1379916:1380032 [0] NCCL INFO Channel 09/0 : 8[0] -> 15[7] via P2P/CUMEM +n116-041-153:1379922:1380033 [6] NCCL INFO Channel 13/0 : 14[6] -> 15[7] via P2P/CUMEM +n116-041-153:1379916:1380032 [0] NCCL INFO Channel 10/0 : 8[0] -> 15[7] via P2P/CUMEM +n116-041-153:1379922:1380033 [6] NCCL INFO Channel 14/0 : 14[6] -> 15[7] via P2P/CUMEM +n116-041-153:1379916:1380032 [0] NCCL INFO Channel 11/0 : 8[0] -> 15[7] via P2P/CUMEM +n116-041-153:1379922:1380033 [6] NCCL INFO Channel 15/0 : 14[6] -> 15[7] via P2P/CUMEM +n116-041-153:1379916:1380032 [0] NCCL INFO Channel 12/0 : 8[0] -> 15[7] via P2P/CUMEM +n116-041-153:1379916:1380032 [0] NCCL INFO Channel 13/0 : 8[0] -> 15[7] via P2P/CUMEM +n116-041-153:1379916:1380032 [0] NCCL INFO Channel 14/0 : 8[0] -> 15[7] via P2P/CUMEM +n116-041-153:1379920:1380036 [4] NCCL INFO [Proxy Progress] Device 4 CPU core 203 +n116-041-153:1379917:1380037 [1] NCCL INFO [Proxy Progress] Device 1 CPU core 135 +n116-041-153:1379917:1380030 [1] NCCL INFO Channel 01/0 : 9[1] -> 16[0] [send] via NET/FasTrak/1/GDRDMA +n116-041-153:1379917:1380030 [1] NCCL INFO Channel 08/0 : 9[1] -> 16[0] [send] via NET/FasTrak/0/GDRDMA +n116-041-153:1379920:1380031 [4] NCCL INFO Channel 12/0 : 4[4] -> 12[4] [receive] via NET/FasTrak/4/GDRDMA +n116-041-153:1379919:1380038 [3] NCCL INFO [Proxy Progress] Device 3 CPU core 137 +n116-041-153:1379919:1380029 [3] NCCL INFO Channel 03/0 : 11[3] -> 18[2] [send] via NET/FasTrak/3/GDRDMA +n116-041-153:1379919:1380029 [3] NCCL INFO Channel 10/0 : 11[3] -> 18[2] [send] via NET/FasTrak/2/GDRDMA +n116-041-153:1379920:1380031 [4] NCCL INFO Channel 04/0 : 12[4] -> 21[5] [send] via NET/FasTrak/4/GDRDMA +n116-041-153:1379920:1380031 [4] NCCL INFO Channel 13/0 : 12[4] -> 21[5] [send] via NET/FasTrak/5/GDRDMA +n116-041-153:1379920:1380031 [4] NCCL INFO Channel 12/0 : 12[4] -> 28[4] [send] via NET/FasTrak/4/GDRDMA +n116-041-153:1379917:1380030 [1] NCCL INFO Channel 09/0 : 1[1] -> 9[1] [receive] via NET/FasTrak/1/GDRDMA +n116-041-153:1379919:1380029 [3] NCCL INFO Channel 11/0 : 3[3] -> 11[3] [receive] via NET/FasTrak/3/GDRDMA +n116-041-153:1379917:1380030 [1] NCCL INFO Channel 09/0 : 9[1] -> 25[1] [send] via NET/FasTrak/1/GDRDMA +n116-041-153:1379921:1380039 [5] NCCL INFO [Proxy Progress] Device 5 CPU core 161 +n116-041-153:1379919:1380029 [3] NCCL INFO Channel 11/0 : 11[3] -> 27[3] [send] via NET/FasTrak/3/GDRDMA +n116-041-153:1379921:1380034 [5] NCCL INFO Channel 05/0 : 13[5] -> 20[4] [send] via NET/FasTrak/5/GDRDMA +n116-041-153:1379921:1380034 [5] NCCL INFO Channel 12/0 : 13[5] -> 20[4] [send] via NET/FasTrak/4/GDRDMA +n116-041-153:1379918:1380040 [2] NCCL INFO [Proxy Progress] Device 2 CPU core 34 +n116-041-153:1379921:1380034 [5] NCCL INFO Channel 13/0 : 5[5] -> 13[5] [receive] via NET/FasTrak/5/GDRDMA +n116-041-153:1379921:1380034 [5] NCCL INFO Channel 13/0 : 13[5] -> 29[5] [send] via NET/FasTrak/5/GDRDMA +n116-041-153:1379918:1380028 [2] NCCL INFO Channel 10/0 : 2[2] -> 10[2] [receive] via NET/FasTrak/2/GDRDMA +n116-041-153:1379918:1380028 [2] NCCL INFO Channel 02/0 : 10[2] -> 19[3] [send] via NET/FasTrak/2/GDRDMA +n116-041-153:1379918:1380028 [2] NCCL INFO Channel 11/0 : 10[2] -> 19[3] [send] via NET/FasTrak/3/GDRDMA +n116-041-153:1379918:1380028 [2] NCCL INFO Channel 10/0 : 10[2] -> 26[2] [send] via NET/FasTrak/2/GDRDMA +n116-041-153:1379920:1380031 [4] NCCL INFO Channel 12/0 : 28[4] -> 12[4] [receive] via NET/FasTrak/4/GDRDMA +n116-041-153:1379922:1380041 [6] NCCL INFO [Proxy Progress] Device 6 CPU core 179 +n116-041-153:1379918:1380028 [2] NCCL INFO Channel 10/0 : 26[2] -> 10[2] [receive] via NET/FasTrak/2/GDRDMA +n116-041-153:1379922:1380033 [6] NCCL INFO Channel 14/0 : 6[6] -> 14[6] [receive] via NET/FasTrak/6/GDRDMA +n116-041-153:1379922:1380033 [6] NCCL INFO Channel 06/0 : 14[6] -> 23[7] [send] via NET/FasTrak/6/GDRDMA +n116-041-153:1379922:1380033 [6] NCCL INFO Channel 15/0 : 14[6] -> 23[7] [send] via NET/FasTrak/7/GDRDMA +n116-041-153:1379922:1380033 [6] NCCL INFO Channel 14/0 : 14[6] -> 30[6] [send] via NET/FasTrak/6/GDRDMA +n116-041-153:1379923:1380042 [7] NCCL INFO [Proxy Progress] Device 7 CPU core 100 +n116-041-153:1379923:1380035 [7] NCCL INFO Channel 07/0 : 15[7] -> 22[6] [send] via NET/FasTrak/7/GDRDMA +n116-041-153:1379923:1380035 [7] NCCL INFO Channel 14/0 : 15[7] -> 22[6] [send] via NET/FasTrak/6/GDRDMA +n116-041-153:1379923:1380035 [7] NCCL INFO Channel 15/0 : 7[7] -> 15[7] [receive] via NET/FasTrak/7/GDRDMA +n116-041-153:1379923:1380035 [7] NCCL INFO Channel 15/0 : 15[7] -> 31[7] [send] via NET/FasTrak/7/GDRDMA +n116-041-153:1379922:1380033 [6] NCCL INFO Channel 14/0 : 30[6] -> 14[6] [receive] via NET/FasTrak/6/GDRDMA +n116-041-153:1379916:1380043 [0] NCCL INFO [Proxy Progress] Device 0 CPU core 44 +n116-041-153:1379916:1380032 [0] NCCL INFO Channel 08/0 : 0[0] -> 8[0] [receive] via NET/FasTrak/0/GDRDMA +n116-041-153:1379916:1380032 [0] NCCL INFO Channel 00/0 : 8[0] -> 17[1] [send] via NET/FasTrak/0/GDRDMA +n116-041-153:1379916:1380032 [0] NCCL INFO Channel 09/0 : 8[0] -> 17[1] [send] via NET/FasTrak/1/GDRDMA +n116-041-153:1379916:1380032 [0] NCCL INFO Channel 08/0 : 8[0] -> 24[0] [send] via NET/FasTrak/0/GDRDMA +n116-041-153:1379916:1380032 [0] NCCL INFO Channel 08/0 : 24[0] -> 8[0] [receive] via NET/FasTrak/0/GDRDMA +n116-041-153:1379921:1380034 [5] NCCL INFO Channel 13/0 : 29[5] -> 13[5] [receive] via NET/FasTrak/5/GDRDMA +n116-041-153:1379921:1380034 [5] NCCL INFO Channel 13/0 : 13[5] -> 5[5] [send] via NET/FasTrak/5/GDRDMA +n116-041-153:1379916:1380032 [0] NCCL INFO Channel 00/0 : 17[1] -> 8[0] [receive] via NET/FasTrak/0/GDRDMA +n116-041-153:1379919:1380029 [3] NCCL INFO Channel 11/0 : 27[3] -> 11[3] [receive] via NET/FasTrak/3/GDRDMA +n116-041-153:1379916:1380032 [0] NCCL INFO Channel 09/0 : 17[1] -> 8[0] [receive] via NET/FasTrak/1/GDRDMA +n116-041-153:1379919:1380029 [3] NCCL INFO Channel 11/0 : 11[3] -> 3[3] [send] via NET/FasTrak/3/GDRDMA +n116-041-153:1379922:1380033 [6] NCCL INFO Channel 06/0 : 23[7] -> 14[6] [receive] via NET/FasTrak/6/GDRDMA +n116-041-153:1379916:1380032 [0] NCCL INFO Channel 08/0 : 8[0] -> 0[0] [send] via NET/FasTrak/0/GDRDMA +n116-041-153:1379920:1380031 [4] NCCL INFO Channel 04/0 : 21[5] -> 12[4] [receive] via NET/FasTrak/4/GDRDMA +n116-041-153:1379922:1380033 [6] NCCL INFO Channel 15/0 : 23[7] -> 14[6] [receive] via NET/FasTrak/7/GDRDMA +n116-041-153:1379917:1380030 [1] NCCL INFO Channel 09/0 : 25[1] -> 9[1] [receive] via NET/FasTrak/1/GDRDMA +n116-041-153:1379920:1380031 [4] NCCL INFO Channel 13/0 : 21[5] -> 12[4] [receive] via NET/FasTrak/5/GDRDMA +n116-041-153:1379922:1380033 [6] NCCL INFO Channel 14/0 : 14[6] -> 6[6] [send] via NET/FasTrak/6/GDRDMA +n116-041-153:1379920:1380031 [4] NCCL INFO Channel 12/0 : 12[4] -> 4[4] [send] via NET/FasTrak/4/GDRDMA +n116-041-153:1379917:1380030 [1] NCCL INFO Channel 09/0 : 9[1] -> 1[1] [send] via NET/FasTrak/1/GDRDMA +n116-041-153:1379919:1380029 [3] NCCL INFO Channel 03/0 : 18[2] -> 11[3] [receive] via NET/FasTrak/3/GDRDMA +n116-041-153:1379921:1380034 [5] NCCL INFO Channel 05/0 : 20[4] -> 13[5] [receive] via NET/FasTrak/5/GDRDMA +n116-041-153:1379919:1380029 [3] NCCL INFO Channel 10/0 : 18[2] -> 11[3] [receive] via NET/FasTrak/2/GDRDMA +n116-041-153:1379918:1380028 [2] NCCL INFO Channel 02/0 : 19[3] -> 10[2] [receive] via NET/FasTrak/2/GDRDMA +n116-041-153:1379921:1380034 [5] NCCL INFO Channel 12/0 : 20[4] -> 13[5] [receive] via NET/FasTrak/4/GDRDMA +n116-041-153:1379917:1380030 [1] NCCL INFO Channel 01/0 : 16[0] -> 9[1] [receive] via NET/FasTrak/1/GDRDMA +n116-041-153:1379918:1380028 [2] NCCL INFO Channel 11/0 : 19[3] -> 10[2] [receive] via NET/FasTrak/3/GDRDMA +n116-041-153:1379918:1380028 [2] NCCL INFO Channel 10/0 : 10[2] -> 2[2] [send] via NET/FasTrak/2/GDRDMA +n116-041-153:1379917:1380030 [1] NCCL INFO Channel 08/0 : 16[0] -> 9[1] [receive] via NET/FasTrak/0/GDRDMA +n116-041-153:1379922:1380033 [6] NCCL INFO Channel 00/0 : 14[6] -> 13[5] via P2P/CUMEM +n116-041-153:1379922:1380033 [6] NCCL INFO Channel 01/0 : 14[6] -> 13[5] via P2P/CUMEM +n116-041-153:1379920:1380031 [4] NCCL INFO Channel 00/0 : 12[4] -> 11[3] via P2P/CUMEM +n116-041-153:1379922:1380033 [6] NCCL INFO Channel 02/0 : 14[6] -> 13[5] via P2P/CUMEM +n116-041-153:1379920:1380031 [4] NCCL INFO Channel 01/0 : 12[4] -> 11[3] via P2P/CUMEM +n116-041-153:1379922:1380033 [6] NCCL INFO Channel 03/0 : 14[6] -> 13[5] via P2P/CUMEM +n116-041-153:1379919:1380029 [3] NCCL INFO Channel 00/0 : 11[3] -> 10[2] via P2P/CUMEM +n116-041-153:1379920:1380031 [4] NCCL INFO Channel 02/0 : 12[4] -> 11[3] via P2P/CUMEM +n116-041-153:1379922:1380033 [6] NCCL INFO Channel 04/0 : 14[6] -> 13[5] via P2P/CUMEM +n116-041-153:1379920:1380031 [4] NCCL INFO Channel 05/0 : 12[4] -> 11[3] via P2P/CUMEM +n116-041-153:1379919:1380029 [3] NCCL INFO Channel 01/0 : 11[3] -> 10[2] via P2P/CUMEM +n116-041-153:1379922:1380033 [6] NCCL INFO Channel 07/0 : 14[6] -> 13[5] via P2P/CUMEM +n116-041-153:1379920:1380031 [4] NCCL INFO Channel 06/0 : 12[4] -> 11[3] via P2P/CUMEM +n116-041-153:1379919:1380029 [3] NCCL INFO Channel 02/0 : 11[3] -> 10[2] via P2P/CUMEM +n116-041-153:1379921:1380034 [5] NCCL INFO Channel 00/0 : 13[5] -> 12[4] via P2P/CUMEM +n116-041-153:1379922:1380033 [6] NCCL INFO Channel 08/0 : 14[6] -> 13[5] via P2P/CUMEM +n116-041-153:1379920:1380031 [4] NCCL INFO Channel 07/0 : 12[4] -> 11[3] via P2P/CUMEM +n116-041-153:1379919:1380029 [3] NCCL INFO Channel 03/0 : 11[3] -> 10[2] via P2P/CUMEM +n116-041-153:1379921:1380034 [5] NCCL INFO Channel 01/0 : 13[5] -> 12[4] via P2P/CUMEM +n116-041-153:1379922:1380033 [6] NCCL INFO Channel 09/0 : 14[6] -> 13[5] via P2P/CUMEM +n116-041-153:1379920:1380031 [4] NCCL INFO Channel 08/0 : 12[4] -> 11[3] via P2P/CUMEM +n116-041-153:1379919:1380029 [3] NCCL INFO Channel 04/0 : 11[3] -> 10[2] via P2P/CUMEM +n116-041-153:1379921:1380034 [5] NCCL INFO Channel 02/0 : 13[5] -> 12[4] via P2P/CUMEM +n116-041-153:1379922:1380033 [6] NCCL INFO Channel 10/0 : 14[6] -> 13[5] via P2P/CUMEM +n116-041-153:1379920:1380031 [4] NCCL INFO Channel 09/0 : 12[4] -> 11[3] via P2P/CUMEM +n116-041-153:1379919:1380029 [3] NCCL INFO Channel 05/0 : 11[3] -> 10[2] via P2P/CUMEM +n116-041-153:1379921:1380034 [5] NCCL INFO Channel 03/0 : 13[5] -> 12[4] via P2P/CUMEM +n116-041-153:1379922:1380033 [6] NCCL INFO Channel 11/0 : 14[6] -> 13[5] via P2P/CUMEM +n116-041-153:1379920:1380031 [4] NCCL INFO Channel 10/0 : 12[4] -> 11[3] via P2P/CUMEM +n116-041-153:1379919:1380029 [3] NCCL INFO Channel 06/0 : 11[3] -> 10[2] via P2P/CUMEM +n116-041-153:1379921:1380034 [5] NCCL INFO Channel 04/0 : 13[5] -> 12[4] via P2P/CUMEM +n116-041-153:1379922:1380033 [6] NCCL INFO Channel 12/0 : 14[6] -> 13[5] via P2P/CUMEM +n116-041-153:1379920:1380031 [4] NCCL INFO Channel 13/0 : 12[4] -> 11[3] via P2P/CUMEM +n116-041-153:1379919:1380029 [3] NCCL INFO Channel 07/0 : 11[3] -> 10[2] via P2P/CUMEM +n116-041-153:1379921:1380034 [5] NCCL INFO Channel 05/0 : 13[5] -> 12[4] via P2P/CUMEM +n116-041-153:1379922:1380033 [6] NCCL INFO Channel 15/0 : 14[6] -> 13[5] via P2P/CUMEM +n116-041-153:1379920:1380031 [4] NCCL INFO Channel 14/0 : 12[4] -> 11[3] via P2P/CUMEM +n116-041-153:1379918:1380028 [2] NCCL INFO Channel 00/0 : 10[2] -> 9[1] via P2P/CUMEM +n116-041-153:1379919:1380029 [3] NCCL INFO Channel 08/0 : 11[3] -> 10[2] via P2P/CUMEM +n116-041-153:1379921:1380034 [5] NCCL INFO Channel 06/0 : 13[5] -> 12[4] via P2P/CUMEM +n116-041-153:1379920:1380031 [4] NCCL INFO Channel 15/0 : 12[4] -> 11[3] via P2P/CUMEM +n116-041-153:1379918:1380028 [2] NCCL INFO Channel 03/0 : 10[2] -> 9[1] via P2P/CUMEM +n116-041-153:1379919:1380029 [3] NCCL INFO Channel 09/0 : 11[3] -> 10[2] via P2P/CUMEM +n116-041-153:1379921:1380034 [5] NCCL INFO Channel 07/0 : 13[5] -> 12[4] via P2P/CUMEM +n116-041-153:1379918:1380028 [2] NCCL INFO Channel 04/0 : 10[2] -> 9[1] via P2P/CUMEM +n116-041-153:1379921:1380034 [5] NCCL INFO Channel 08/0 : 13[5] -> 12[4] via P2P/CUMEM +n116-041-153:1379919:1380029 [3] NCCL INFO Channel 10/0 : 11[3] -> 10[2] via P2P/CUMEM +n116-041-153:1379918:1380028 [2] NCCL INFO Channel 05/0 : 10[2] -> 9[1] via P2P/CUMEM +n116-041-153:1379921:1380034 [5] NCCL INFO Channel 09/0 : 13[5] -> 12[4] via P2P/CUMEM +n116-041-153:1379919:1380029 [3] NCCL INFO Channel 11/0 : 11[3] -> 10[2] via P2P/CUMEM +n116-041-153:1379918:1380028 [2] NCCL INFO Channel 06/0 : 10[2] -> 9[1] via P2P/CUMEM +n116-041-153:1379921:1380034 [5] NCCL INFO Channel 10/0 : 13[5] -> 12[4] via P2P/CUMEM +n116-041-153:1379919:1380029 [3] NCCL INFO Channel 12/0 : 11[3] -> 10[2] via P2P/CUMEM +n116-041-153:1379918:1380028 [2] NCCL INFO Channel 07/0 : 10[2] -> 9[1] via P2P/CUMEM +n116-041-153:1379921:1380034 [5] NCCL INFO Channel 11/0 : 13[5] -> 12[4] via P2P/CUMEM +n116-041-153:1379919:1380029 [3] NCCL INFO Channel 13/0 : 11[3] -> 10[2] via P2P/CUMEM +n116-041-153:1379918:1380028 [2] NCCL INFO Channel 08/0 : 10[2] -> 9[1] via P2P/CUMEM +n116-041-153:1379921:1380034 [5] NCCL INFO Channel 12/0 : 13[5] -> 12[4] via P2P/CUMEM +n116-041-153:1379919:1380029 [3] NCCL INFO Channel 14/0 : 11[3] -> 10[2] via P2P/CUMEM +n116-041-153:1379918:1380028 [2] NCCL INFO Channel 11/0 : 10[2] -> 9[1] via P2P/CUMEM +n116-041-153:1379921:1380034 [5] NCCL INFO Channel 13/0 : 13[5] -> 12[4] via P2P/CUMEM +n116-041-153:1379919:1380029 [3] NCCL INFO Channel 15/0 : 11[3] -> 10[2] via P2P/CUMEM +n116-041-153:1379918:1380028 [2] NCCL INFO Channel 12/0 : 10[2] -> 9[1] via P2P/CUMEM +n116-041-153:1379921:1380034 [5] NCCL INFO Channel 14/0 : 13[5] -> 12[4] via P2P/CUMEM +n116-041-153:1379918:1380028 [2] NCCL INFO Channel 13/0 : 10[2] -> 9[1] via P2P/CUMEM +n116-041-153:1379918:1380028 [2] NCCL INFO Channel 14/0 : 10[2] -> 9[1] via P2P/CUMEM +n116-041-153:1379921:1380034 [5] NCCL INFO Channel 15/0 : 13[5] -> 12[4] via P2P/CUMEM +n116-041-153:1379918:1380028 [2] NCCL INFO Channel 15/0 : 10[2] -> 9[1] via P2P/CUMEM +n116-041-153:1379917:1380030 [1] NCCL INFO Channel 00/0 : 9[1] -> 8[0] via P2P/CUMEM +n116-041-153:1379923:1380035 [7] NCCL INFO Channel 15/0 : 31[7] -> 15[7] [receive] via NET/FasTrak/7/GDRDMA +n116-041-153:1379923:1380035 [7] NCCL INFO Channel 15/0 : 15[7] -> 7[7] [send] via NET/FasTrak/7/GDRDMA +n116-041-153:1379917:1380030 [1] NCCL INFO Channel 01/0 : 9[1] -> 8[0] via P2P/CUMEM +n116-041-153:1379923:1380035 [7] NCCL INFO Channel 07/0 : 22[6] -> 15[7] [receive] via NET/FasTrak/7/GDRDMA +n116-041-153:1379923:1380035 [7] NCCL INFO Channel 14/0 : 22[6] -> 15[7] [receive] via NET/FasTrak/6/GDRDMA +n116-041-153:1379923:1380035 [7] NCCL INFO Channel 01/0 : 15[7] -> 8[0] via P2P/CUMEM +n116-041-153:1379917:1380030 [1] NCCL INFO Channel 02/0 : 9[1] -> 8[0] via P2P/CUMEM +n116-041-153:1379923:1380035 [7] NCCL INFO Channel 02/0 : 15[7] -> 8[0] via P2P/CUMEM +n116-041-153:1379917:1380030 [1] NCCL INFO Channel 03/0 : 9[1] -> 8[0] via P2P/CUMEM +n116-041-153:1379923:1380035 [7] NCCL INFO Channel 03/0 : 15[7] -> 8[0] via P2P/CUMEM +n116-041-153:1379917:1380030 [1] NCCL INFO Channel 04/0 : 9[1] -> 8[0] via P2P/CUMEM +n116-041-153:1379923:1380035 [7] NCCL INFO Channel 04/0 : 15[7] -> 8[0] via P2P/CUMEM +n116-041-153:1379923:1380035 [7] NCCL INFO Channel 05/0 : 15[7] -> 8[0] via P2P/CUMEM +n116-041-153:1379917:1380030 [1] NCCL INFO Channel 05/0 : 9[1] -> 8[0] via P2P/CUMEM +n116-041-153:1379923:1380035 [7] NCCL INFO Channel 06/0 : 15[7] -> 8[0] via P2P/CUMEM +n116-041-153:1379917:1380030 [1] NCCL INFO Channel 06/0 : 9[1] -> 8[0] via P2P/CUMEM +n116-041-153:1379923:1380035 [7] NCCL INFO Channel 09/0 : 15[7] -> 8[0] via P2P/CUMEM +n116-041-153:1379917:1380030 [1] NCCL INFO Channel 07/0 : 9[1] -> 8[0] via P2P/CUMEM +n116-041-153:1379923:1380035 [7] NCCL INFO Channel 10/0 : 15[7] -> 8[0] via P2P/CUMEM +n116-041-153:1379917:1380030 [1] NCCL INFO Channel 08/0 : 9[1] -> 8[0] via P2P/CUMEM +n116-041-153:1379923:1380035 [7] NCCL INFO Channel 11/0 : 15[7] -> 8[0] via P2P/CUMEM +n116-041-153:1379923:1380035 [7] NCCL INFO Channel 12/0 : 15[7] -> 8[0] via P2P/CUMEM +n116-041-153:1379917:1380030 [1] NCCL INFO Channel 09/0 : 9[1] -> 8[0] via P2P/CUMEM +n116-041-153:1379923:1380035 [7] NCCL INFO Channel 13/0 : 15[7] -> 8[0] via P2P/CUMEM +n116-041-153:1379917:1380030 [1] NCCL INFO Channel 10/0 : 9[1] -> 8[0] via P2P/CUMEM +n116-041-153:1379923:1380035 [7] NCCL INFO Channel 14/0 : 15[7] -> 8[0] via P2P/CUMEM +n116-041-153:1379917:1380030 [1] NCCL INFO Channel 11/0 : 9[1] -> 8[0] via P2P/CUMEM +n116-041-153:1379923:1380035 [7] NCCL INFO Channel 00/0 : 15[7] -> 14[6] via P2P/CUMEM +n116-041-153:1379923:1380035 [7] NCCL INFO Channel 01/0 : 15[7] -> 14[6] via P2P/CUMEM +n116-041-153:1379923:1380035 [7] NCCL INFO Channel 02/0 : 15[7] -> 14[6] via P2P/CUMEM +n116-041-153:1379917:1380030 [1] NCCL INFO Channel 12/0 : 9[1] -> 8[0] via P2P/CUMEM +n116-041-153:1379923:1380035 [7] NCCL INFO Channel 03/0 : 15[7] -> 14[6] via P2P/CUMEM +n116-041-153:1379917:1380030 [1] NCCL INFO Channel 13/0 : 9[1] -> 8[0] via P2P/CUMEM +n116-041-153:1379917:1380030 [1] NCCL INFO Channel 14/0 : 9[1] -> 8[0] via P2P/CUMEM +n116-041-153:1379923:1380035 [7] NCCL INFO Channel 04/0 : 15[7] -> 14[6] via P2P/CUMEM +n116-041-153:1379917:1380030 [1] NCCL INFO Channel 15/0 : 9[1] -> 8[0] via P2P/CUMEM +n116-041-153:1379923:1380035 [7] NCCL INFO Channel 05/0 : 15[7] -> 14[6] via P2P/CUMEM +n116-041-153:1379923:1380035 [7] NCCL INFO Channel 06/0 : 15[7] -> 14[6] via P2P/CUMEM +n116-041-153:1379923:1380035 [7] NCCL INFO Channel 07/0 : 15[7] -> 14[6] via P2P/CUMEM +n116-041-153:1379923:1380035 [7] NCCL INFO Channel 08/0 : 15[7] -> 14[6] via P2P/CUMEM +n116-041-153:1379923:1380035 [7] NCCL INFO Channel 09/0 : 15[7] -> 14[6] via P2P/CUMEM +n116-041-153:1379923:1380035 [7] NCCL INFO Channel 10/0 : 15[7] -> 14[6] via P2P/CUMEM +n116-041-153:1379923:1380035 [7] NCCL INFO Channel 11/0 : 15[7] -> 14[6] via P2P/CUMEM +n116-041-153:1379923:1380035 [7] NCCL INFO Channel 12/0 : 15[7] -> 14[6] via P2P/CUMEM +n116-041-153:1379923:1380035 [7] NCCL INFO Channel 13/0 : 15[7] -> 14[6] via P2P/CUMEM +n116-041-153:1379923:1380035 [7] NCCL INFO Channel 14/0 : 15[7] -> 14[6] via P2P/CUMEM +n116-041-153:1379923:1380035 [7] NCCL INFO Channel 15/0 : 15[7] -> 14[6] via P2P/CUMEM +n116-041-153:1379918:1380028 [2] NCCL INFO Connected all trees +n116-041-153:1379916:1380032 [0] NCCL INFO Connected all trees +n116-041-153:1379919:1380029 [3] NCCL INFO Connected all trees +n116-041-153:1379917:1380030 [1] NCCL INFO Connected all trees +n116-041-153:1379923:1380035 [7] NCCL INFO Connected all trees +n116-041-153:1379922:1380033 [6] NCCL INFO Connected all trees +n116-041-153:1379921:1380034 [5] NCCL INFO Connected all trees +n116-041-153:1379920:1380031 [4] NCCL INFO Connected all trees +n116-041-153:1379921:1380064 [5] NCCL INFO Channel 05/0 : 13[5] -> 14[6] via P2P/CUMEM +n116-041-153:1379921:1380064 [5] NCCL INFO Channel 13/0 : 13[5] -> 14[6] via P2P/CUMEM +n116-041-153:1379917:1380066 [1] NCCL INFO Channel 01/0 : 9[1] -> 10[2] via P2P/CUMEM +n116-041-153:1379917:1380066 [1] NCCL INFO Channel 09/0 : 9[1] -> 10[2] via P2P/CUMEM +n116-041-153:1379921:1380064 [5] NCCL INFO Channel 04/0 : 13[5] -> 20[4] [send] via NET/FasTrak/4/GDRDMA +n116-041-153:1379922:1380067 [6] NCCL INFO Channel 06/0 : 7[7] -> 14[6] [receive] via NET/FasTrak/6/GDRDMA +n116-041-153:1379922:1380067 [6] NCCL INFO Channel 14/0 : 7[7] -> 14[6] [receive] via NET/FasTrak/6/GDRDMA +n116-041-153:1379916:1380068 [0] NCCL INFO Channel 00/0 : 1[1] -> 8[0] [receive] via NET/FasTrak/0/GDRDMA +n116-041-153:1379916:1380068 [0] NCCL INFO Channel 08/0 : 1[1] -> 8[0] [receive] via NET/FasTrak/0/GDRDMA +n116-041-153:1379916:1380068 [0] NCCL INFO Channel 00/0 : 8[0] -> 15[7] via P2P/CUMEM +n116-041-153:1379916:1380068 [0] NCCL INFO Channel 08/0 : 8[0] -> 15[7] via P2P/CUMEM +n116-041-153:1379923:1380069 [7] NCCL INFO Channel 06/0 : 15[7] -> 22[6] [send] via NET/FasTrak/6/GDRDMA +n116-041-153:1379917:1380066 [1] NCCL INFO Channel 00/0 : 9[1] -> 16[0] [send] via NET/FasTrak/0/GDRDMA +n116-041-153:1379918:1380070 [2] NCCL INFO Channel 02/0 : 3[3] -> 10[2] [receive] via NET/FasTrak/2/GDRDMA +n116-041-153:1379917:1380066 [1] NCCL INFO Channel 01/0 : 0[0] -> 9[1] [receive] via NET/FasTrak/1/GDRDMA +n116-041-153:1379918:1380070 [2] NCCL INFO Channel 10/0 : 3[3] -> 10[2] [receive] via NET/FasTrak/2/GDRDMA +n116-041-153:1379917:1380066 [1] NCCL INFO Channel 09/0 : 0[0] -> 9[1] [receive] via NET/FasTrak/1/GDRDMA +n116-041-153:1379919:1380071 [3] NCCL INFO Channel 03/0 : 11[3] -> 12[4] via P2P/CUMEM +n116-041-153:1379919:1380071 [3] NCCL INFO Channel 11/0 : 11[3] -> 12[4] via P2P/CUMEM +n116-041-153:1379919:1380071 [3] NCCL INFO Channel 02/0 : 11[3] -> 18[2] [send] via NET/FasTrak/2/GDRDMA +n116-041-153:1379919:1380071 [3] NCCL INFO Channel 03/0 : 2[2] -> 11[3] [receive] via NET/FasTrak/3/GDRDMA +n116-041-153:1379920:1380065 [4] NCCL INFO Channel 04/0 : 5[5] -> 12[4] [receive] via NET/FasTrak/4/GDRDMA +n116-041-153:1379919:1380071 [3] NCCL INFO Channel 11/0 : 2[2] -> 11[3] [receive] via NET/FasTrak/3/GDRDMA +n116-041-153:1379920:1380065 [4] NCCL INFO Channel 12/0 : 5[5] -> 12[4] [receive] via NET/FasTrak/4/GDRDMA +n116-041-153:1379921:1380064 [5] NCCL INFO Channel 05/0 : 4[4] -> 13[5] [receive] via NET/FasTrak/5/GDRDMA +n116-041-153:1379921:1380064 [5] NCCL INFO Channel 13/0 : 4[4] -> 13[5] [receive] via NET/FasTrak/5/GDRDMA +n116-041-153:1379923:1380069 [7] NCCL INFO Channel 07/0 : 6[6] -> 15[7] [receive] via NET/FasTrak/7/GDRDMA +n116-041-153:1379923:1380069 [7] NCCL INFO Channel 15/0 : 6[6] -> 15[7] [receive] via NET/FasTrak/7/GDRDMA +n116-041-153:1379922:1380067 [6] NCCL INFO Channel 07/0 : 14[6] -> 23[7] [send] via NET/FasTrak/7/GDRDMA +n116-041-153:1379920:1380065 [4] NCCL INFO Channel 05/0 : 12[4] -> 21[5] [send] via NET/FasTrak/5/GDRDMA +n116-041-153:1379923:1380069 [7] NCCL INFO Channel 07/0 : 15[7] -> 8[0] via P2P/CUMEM +n116-041-153:1379923:1380069 [7] NCCL INFO Channel 15/0 : 15[7] -> 8[0] via P2P/CUMEM +n116-041-153:1379918:1380070 [2] NCCL INFO Channel 03/0 : 10[2] -> 19[3] [send] via NET/FasTrak/3/GDRDMA +n116-041-153:1379918:1380070 [2] NCCL INFO Channel 02/0 : 10[2] -> 9[1] via P2P/CUMEM +n116-041-153:1379918:1380070 [2] NCCL INFO Channel 10/0 : 10[2] -> 9[1] via P2P/CUMEM +n116-041-153:1379916:1380068 [0] NCCL INFO Channel 01/0 : 8[0] -> 17[1] [send] via NET/FasTrak/1/GDRDMA +n116-041-153:1379920:1380065 [4] NCCL INFO Channel 04/0 : 12[4] -> 11[3] via P2P/CUMEM +n116-041-153:1379920:1380065 [4] NCCL INFO Channel 12/0 : 12[4] -> 11[3] via P2P/CUMEM +n116-041-153:1379922:1380067 [6] NCCL INFO Channel 06/0 : 14[6] -> 13[5] via P2P/CUMEM +n116-041-153:1379922:1380067 [6] NCCL INFO Channel 14/0 : 14[6] -> 13[5] via P2P/CUMEM +n116-041-153:1379922:1380067 [6] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-041-153:1379923:1380069 [7] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-041-153:1379921:1380064 [5] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-041-153:1379918:1380070 [2] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-041-153:1379917:1380066 [1] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-041-153:1379920:1380065 [4] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-041-153:1379919:1380071 [3] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-041-153:1379916:1380068 [0] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +[rank12]:[W727 05:39:50.890084899 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n116-041-153.byted.org]:41112, remote=[n116-040-168.byted.org]:25264): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f8f4ff785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f8f392cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7f8f392cff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7f8f392d084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7f8f392ca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f8efa9c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f8f5349b4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f8f560721f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f8f560f289c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank12]:[W727 05:39:50.902488346 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 12] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank15]:[W727 05:39:50.080263391 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n116-041-153.byted.org]:41096, remote=[n116-040-168.byted.org]:25264): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f57c5b785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f57af2cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7f57af2cff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7f57af2d084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7f57af2ca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f57709c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f57c933b4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f57cbf121f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f57cbf9289c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank15]:[W727 05:39:50.091963770 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 15] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank11]:[W727 05:39:50.099476181 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n116-041-153.byted.org]:41056, remote=[n116-040-168.byted.org]:25264): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7fd10a9785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7fd0f3ccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7fd0f3ccff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7fd0f3cd084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7fd0f3cca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fd0b53c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7fd10de884a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7fd110a5f1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7fd110adf89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank11]:[W727 05:39:50.106509632 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 11] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank9]:[W727 05:39:50.166995684 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n116-041-153.byted.org]:41080, remote=[n116-040-168.byted.org]:25264): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7fbdf9f785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7fbde32cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7fbde32cff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7fbde32d084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7fbde32ca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fbda49c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7fbdfd45c4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7fbe000331f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7fbe000b389c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank9]:[W727 05:39:50.174060951 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 9] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank8]:[W727 05:39:50.316030959 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=71, addr=[n116-041-153.byted.org]:41050, remote=[n116-040-168.byted.org]:25264): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7fa6de7785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7fa6c7ecdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7fa6c7ecff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7fa6c7ed084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7fa6c7eca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fa6895c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7fa6e1f604a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7fa6e4b371f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7fa6e4bb789c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank8]:[W727 05:39:50.324092018 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 8] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank10]:[W727 05:39:50.319769404 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n116-041-153.byted.org]:41110, remote=[n116-040-168.byted.org]:25264): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f725d1785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f72464cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7f72464cff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7f72464d084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7f72464ca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f7207bc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f72605cf4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f72631a61f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f726322689c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank10]:[W727 05:39:50.327834042 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 10] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank14]:[W727 05:39:51.594858624 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n116-041-153.byted.org]:41072, remote=[n116-040-168.byted.org]:25264): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f4c8f1785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f4c784cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7f4c784cff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7f4c784d084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7f4c784ca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f4c39bc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f4c9267e4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f4c952551f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f4c952d589c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank14]:[W727 05:39:51.602898435 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 14] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank13]:[W727 05:39:51.632319323 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n116-041-153.byted.org]:41060, remote=[n116-040-168.byted.org]:25264): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f7225b785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f720f2cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7f720f2cff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7f720f2d084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7f720f2ca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f71d09c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f72293994a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f722bf701f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f722bff089c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank13]:[W727 05:39:51.638906754 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 13] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +E0727 05:39:51.521000 1379839 torch/distributed/elastic/multiprocessing/api.py:874] failed (exitcode: -15) local_rank: 0 (pid: 1379916) of binary: /usr/bin/python3 +Traceback (most recent call last): + File "/usr/local/bin/torchrun", line 7, in + sys.exit(main()) + ^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 355, in wrapper + return f(*args, **kwargs) + ^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 892, in main + run(args) + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 883, in run + elastic_launch( + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 139, in __call__ + return launch_agent(self._config, self._entrypoint, list(args)) + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 270, in launch_agent + raise ChildFailedError( +torch.distributed.elastic.multiprocessing.errors.ChildFailedError: +============================================================ +main_cross_2direct_balance_realscale_perturb_v4_accum.py FAILED +------------------------------------------------------------ +Failures: +[1]: + time : 2026-07-27_05:39:51 + host : n116-041-153.byted.org + rank : 9 (local_rank: 1) + exitcode : -15 (pid: 1379917) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 1379917 +[2]: + time : 2026-07-27_05:39:51 + host : n116-041-153.byted.org + rank : 10 (local_rank: 2) + exitcode : -15 (pid: 1379918) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 1379918 +[3]: + time : 2026-07-27_05:39:51 + host : n116-041-153.byted.org + rank : 11 (local_rank: 3) + exitcode : -15 (pid: 1379919) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 1379919 +[4]: + time : 2026-07-27_05:39:51 + host : n116-041-153.byted.org + rank : 12 (local_rank: 4) + exitcode : -15 (pid: 1379920) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 1379920 +[5]: + time : 2026-07-27_05:39:51 + host : n116-041-153.byted.org + rank : 13 (local_rank: 5) + exitcode : -15 (pid: 1379921) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 1379921 +[6]: + time : 2026-07-27_05:39:51 + host : n116-041-153.byted.org + rank : 14 (local_rank: 6) + exitcode : -15 (pid: 1379922) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 1379922 +[7]: + time : 2026-07-27_05:39:51 + host : n116-041-153.byted.org + rank : 15 (local_rank: 7) + exitcode : -15 (pid: 1379923) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 1379923 +------------------------------------------------------------ +Root Cause (first observed failure): +[0]: + time : 2026-07-27_05:39:51 + host : n116-041-153.byted.org + rank : 8 (local_rank: 0) + exitcode : -15 (pid: 1379916) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 1379916 +============================================================ diff --git a/ScaleDF_rebuttal/_host_logs/20260726_2350_vith64_free8_port25264_launch/launcher_rank2_n116-041-188.log b/ScaleDF_rebuttal/_host_logs/20260726_2350_vith64_free8_port25264_launch/launcher_rank2_n116-041-188.log new file mode 100644 index 0000000000000000000000000000000000000000..71d62c7e5089915487da8772c2e84a8f59fd0993 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_2350_vith64_free8_port25264_launch/launcher_rank2_n116-041-188.log @@ -0,0 +1,1881 @@ +started_utc=2026-07-26T23:50:26Z +hostname=n116-041-188 +node_rank=2 +nnodes=8 +nproc_per_node=8 +world_size=64 +master_addr=10.116.40.168 +master_port=25264 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +main=main_cross_2direct_balance_realscale_perturb_v4_accum.py +model_name=deit_huge_patch14_LS +pretrain=/tmp/ScaleDF_repro_20260724/pretrained/deit_3_huge_224_21k_v1.pth +resume=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/checkpoint_snapshots/vith32_latest_for_64card_effbs4096_20260726_234604.pth +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_64card_effbs4096_from_epoch22_snapshot_free8_master_n116-040-168_20260726_2350_vith64_port25264 +batch_per_gpu=64 +accum_iter=1 +effective_batch=4096 +eval_batch_per_gpu=64 +epochs=40 +W0726 23:50:38.757000 1377175 torch/distributed/run.py:766] +W0726 23:50:38.757000 1377175 torch/distributed/run.py:766] ***************************************** +W0726 23:50:38.757000 1377175 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 23:50:38.757000 1377175 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 16): env:// +| distributed init (rank 20): env:// +[rank20]:[W726 23:51:06.186014957 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 20] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 19): env:// +[rank19]:[W726 23:51:06.584208541 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 19] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 18): env:// +[rank18]:[W726 23:51:06.665885486 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 18] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 17): env:// +[rank17]:[W726 23:51:06.875515635 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 17] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 22): env:// +[rank22]:[W726 23:51:07.907228961 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 22] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank16]:[W726 23:51:07.907480366 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 16] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 21): env:// +| distributed init (rank 23): env:// +[rank21]:[W726 23:51:07.924047125 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 21] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank23]:[W726 23:51:07.924167265 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 23] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +n116-041-188:1377257:1377257 [5] NCCL INFO cudaDriverVersion 12060 +n116-041-188:1377257:1377257 [5] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-041-188:1377257:1377257 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-041-188:1377255:1377255 [3] NCCL INFO cudaDriverVersion 12060 +n116-041-188:1377255:1377255 [3] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-041-188:1377255:1377255 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-041-188:1377257:1377257 [5] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-041-188:1377255:1377255 [3] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-041-188:1377257:1377257 [5] NCCL INFO Comm config Blocking set to 1 +n116-041-188:1377255:1377255 [3] NCCL INFO Comm config Blocking set to 1 +n116-041-188:1377253:1377253 [1] NCCL INFO cudaDriverVersion 12060 +n116-041-188:1377253:1377253 [1] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-041-188:1377253:1377253 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-041-188:1377253:1377253 [1] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-041-188:1377253:1377253 [1] NCCL INFO Comm config Blocking set to 1 +n116-041-188:1377256:1377256 [4] NCCL INFO cudaDriverVersion 12060 +n116-041-188:1377256:1377256 [4] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-041-188:1377256:1377256 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-041-188:1377256:1377256 [4] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-041-188:1377256:1377256 [4] NCCL INFO Comm config Blocking set to 1 +n116-041-188:1377252:1377252 [0] NCCL INFO cudaDriverVersion 12060 +n116-041-188:1377252:1377252 [0] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-041-188:1377252:1377252 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-041-188:1377252:1377252 [0] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-041-188:1377259:1377259 [7] NCCL INFO cudaDriverVersion 12060 +n116-041-188:1377259:1377259 [7] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-041-188:1377259:1377259 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-041-188:1377258:1377258 [6] NCCL INFO cudaDriverVersion 12060 +n116-041-188:1377258:1377258 [6] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-041-188:1377258:1377258 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-041-188:1377259:1377259 [7] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-041-188:1377258:1377258 [6] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-041-188:1377252:1377252 [0] NCCL INFO Comm config Blocking set to 1 +n116-041-188:1377254:1377254 [2] NCCL INFO cudaDriverVersion 12060 +n116-041-188:1377254:1377254 [2] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-041-188:1377254:1377254 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-041-188:1377259:1377259 [7] NCCL INFO Comm config Blocking set to 1 +n116-041-188:1377258:1377258 [6] NCCL INFO Comm config Blocking set to 1 +n116-041-188:1377254:1377254 [2] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-041-188:1377254:1377254 [2] NCCL INFO Comm config Blocking set to 1 +n116-041-188:1377257:1377308 [5] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-041-188:1377255:1377309 [3] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-041-188:1377257:1377308 [5] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-041-188:1377255:1377309 [3] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-041-188:1377257:1377308 [5] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-041-188:1377257:1377308 [5] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-041-188:1377255:1377309 [3] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-041-188:1377255:1377309 [3] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-041-188:1377257:1377308 [5] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-041-188:1377255:1377309 [3] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-041-188:1377257:1377308 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-041-188:1377255:1377309 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-041-188:1377257:1377308 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-041-188:1377255:1377309 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-041-188:1377257:1377308 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-041-188:1377255:1377309 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-041-188:1377257:1377308 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-041-188:1377255:1377309 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-041-188:1377257:1377308 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-041-188:1377255:1377309 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-041-188:1377253:1377310 [1] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-041-188:1377253:1377310 [1] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-041-188:1377253:1377310 [1] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-041-188:1377253:1377310 [1] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-041-188:1377253:1377310 [1] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-041-188:1377253:1377310 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-041-188:1377253:1377310 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-041-188:1377253:1377310 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-041-188:1377253:1377310 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-041-188:1377253:1377310 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-041-188:1377256:1377311 [4] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-041-188:1377256:1377311 [4] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-041-188:1377256:1377311 [4] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-041-188:1377256:1377311 [4] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-041-188:1377256:1377311 [4] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-041-188:1377256:1377311 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-041-188:1377256:1377311 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-041-188:1377256:1377311 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-041-188:1377256:1377311 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-041-188:1377256:1377311 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-041-188:1377252:1377312 [0] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-041-188:1377252:1377312 [0] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-041-188:1377252:1377312 [0] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-041-188:1377252:1377312 [0] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-041-188:1377252:1377312 [0] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-041-188:1377252:1377312 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-041-188:1377252:1377312 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-041-188:1377252:1377312 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-041-188:1377252:1377312 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-041-188:1377252:1377312 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-041-188:1377254:1377315 [2] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-041-188:1377254:1377315 [2] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-041-188:1377254:1377315 [2] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-041-188:1377254:1377315 [2] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-041-188:1377254:1377315 [2] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-041-188:1377254:1377315 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-041-188:1377254:1377315 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-041-188:1377254:1377315 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-041-188:1377254:1377315 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-041-188:1377254:1377315 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-041-188:1377259:1377313 [7] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-041-188:1377259:1377313 [7] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-041-188:1377259:1377313 [7] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-041-188:1377259:1377313 [7] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-041-188:1377259:1377313 [7] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-041-188:1377259:1377313 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-041-188:1377259:1377313 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-041-188:1377259:1377313 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-041-188:1377259:1377313 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-041-188:1377259:1377313 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-041-188:1377258:1377314 [6] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-041-188:1377258:1377314 [6] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-041-188:1377258:1377314 [6] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-041-188:1377258:1377314 [6] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-041-188:1377258:1377314 [6] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-041-188:1377258:1377314 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-041-188:1377258:1377314 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-041-188:1377258:1377314 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-041-188:1377258:1377314 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-041-188:1377258:1377314 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-041-188:1377253:1377310 [1] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-041-188:1377253:1377310 [1] NCCL INFO Using network FasTrak +n116-041-188:1377255:1377309 [3] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-041-188:1377255:1377309 [3] NCCL INFO Using network FasTrak +n116-041-188:1377256:1377311 [4] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-041-188:1377256:1377311 [4] NCCL INFO Using network FasTrak +n116-041-188:1377257:1377308 [5] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-041-188:1377257:1377308 [5] NCCL INFO Using network FasTrak +n116-041-188:1377252:1377312 [0] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-041-188:1377252:1377312 [0] NCCL INFO Using network FasTrak +n116-041-188:1377254:1377315 [2] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-041-188:1377254:1377315 [2] NCCL INFO Using network FasTrak +n116-041-188:1377259:1377313 [7] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-041-188:1377259:1377313 [7] NCCL INFO Using network FasTrak +n116-041-188:1377258:1377314 [6] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-041-188:1377258:1377314 [6] NCCL INFO Using network FasTrak +n116-041-188:1377253:1377310 [1] NCCL INFO DMA-BUF is available on GPU device 1 +n116-041-188:1377255:1377309 [3] NCCL INFO DMA-BUF is available on GPU device 3 +n116-041-188:1377256:1377311 [4] NCCL INFO DMA-BUF is available on GPU device 4 +n116-041-188:1377253:1377310 [1] NCCL INFO ncclCommInitRankConfig comm 0xcbee300 rank 17 nranks 64 cudaDev 1 nvmlDev 1 busId 5000 commId 0xfb2d11625d05007a - Init START +n116-041-188:1377255:1377309 [3] NCCL INFO ncclCommInitRankConfig comm 0xcab1d40 rank 19 nranks 64 cudaDev 3 nvmlDev 3 busId c000 commId 0xfb2d11625d05007a - Init START +n116-041-188:1377257:1377308 [5] NCCL INFO DMA-BUF is available on GPU device 5 +n116-041-188:1377254:1377315 [2] NCCL INFO DMA-BUF is available on GPU device 2 +n116-041-188:1377258:1377314 [6] NCCL INFO DMA-BUF is available on GPU device 6 +n116-041-188:1377256:1377311 [4] NCCL INFO ncclCommInitRankConfig comm 0xb9355b0 rank 20 nranks 64 cudaDev 4 nvmlDev 4 busId 84000 commId 0xfb2d11625d05007a - Init START +n116-041-188:1377252:1377312 [0] NCCL INFO DMA-BUF is available on GPU device 0 +n116-041-188:1377259:1377313 [7] NCCL INFO DMA-BUF is available on GPU device 7 +n116-041-188:1377254:1377315 [2] NCCL INFO ncclCommInitRankConfig comm 0xb3d7aa0 rank 18 nranks 64 cudaDev 2 nvmlDev 2 busId b000 commId 0xfb2d11625d05007a - Init START +n116-041-188:1377258:1377314 [6] NCCL INFO ncclCommInitRankConfig comm 0xc3c2c30 rank 22 nranks 64 cudaDev 6 nvmlDev 6 busId 8b000 commId 0xfb2d11625d05007a - Init START +n116-041-188:1377259:1377313 [7] NCCL INFO ncclCommInitRankConfig comm 0xc17e960 rank 23 nranks 64 cudaDev 7 nvmlDev 7 busId 8c000 commId 0xfb2d11625d05007a - Init START +n116-041-188:1377257:1377308 [5] NCCL INFO ncclCommInitRankConfig comm 0xb18a360 rank 21 nranks 64 cudaDev 5 nvmlDev 5 busId 85000 commId 0xfb2d11625d05007a - Init START +n116-041-188:1377252:1377312 [0] NCCL INFO ncclCommInitRankConfig comm 0xcaf2670 rank 16 nranks 64 cudaDev 0 nvmlDev 0 busId 4000 commId 0xfb2d11625d05007a - Init START +n116-041-188:1377256:1377311 [4] NCCL INFO RAS client listening socket at ::1<28028> +n116-041-188:1377252:1377312 [0] NCCL INFO RAS client listening socket at ::1<28028> +n116-041-188:1377254:1377315 [2] NCCL INFO RAS client listening socket at ::1<28028> +n116-041-188:1377257:1377308 [5] NCCL INFO RAS client listening socket at ::1<28028> +n116-041-188:1377255:1377309 [3] NCCL INFO RAS client listening socket at ::1<28028> +n116-041-188:1377258:1377314 [6] NCCL INFO RAS client listening socket at ::1<28028> +n116-041-188:1377253:1377310 [1] NCCL INFO RAS client listening socket at ::1<28028> +n116-041-188:1377259:1377313 [7] NCCL INFO RAS client listening socket at ::1<28028> +n116-041-188:1377259:1377313 [7] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-041-188:1377259:1377313 [7] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-041-188:1377259:1377313 [7] NCCL INFO Retrieving state for FasTrak +n116-041-188:1377259:1377313 [7] NCCL INFO Initialized state 0 for FasTrak +n116-041-188:1377259:1377313 [7] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-041-188:1377259:1377313 [7] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-041-188:1377259:1377313 [7] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-041-188:1377259:1377313 [7] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-041-188:1377259:1377313 [7] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-041-188:1377259:1377313 [7] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-041-188:1377259:1377313 [7] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-041-188:1377259:1377313 [7] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-041-188:1377257:1377308 [5] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-041-188:1377257:1377308 [5] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-041-188:1377257:1377308 [5] NCCL INFO Retrieving state for FasTrak +n116-041-188:1377257:1377308 [5] NCCL INFO Initialized state 0 for FasTrak +n116-041-188:1377257:1377308 [5] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-041-188:1377257:1377308 [5] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-041-188:1377257:1377308 [5] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-041-188:1377258:1377314 [6] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-041-188:1377257:1377308 [5] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-041-188:1377258:1377314 [6] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-041-188:1377258:1377314 [6] NCCL INFO Retrieving state for FasTrak +n116-041-188:1377258:1377314 [6] NCCL INFO Initialized state 0 for FasTrak +n116-041-188:1377257:1377308 [5] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-041-188:1377257:1377308 [5] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-041-188:1377258:1377314 [6] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-041-188:1377258:1377314 [6] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-041-188:1377257:1377308 [5] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-041-188:1377257:1377308 [5] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-041-188:1377258:1377314 [6] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-041-188:1377256:1377311 [4] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-041-188:1377256:1377311 [4] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-041-188:1377256:1377311 [4] NCCL INFO Retrieving state for FasTrak +n116-041-188:1377256:1377311 [4] NCCL INFO Initialized state 0 for FasTrak +n116-041-188:1377258:1377314 [6] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-041-188:1377258:1377314 [6] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-041-188:1377256:1377311 [4] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-041-188:1377258:1377314 [6] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-041-188:1377256:1377311 [4] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-041-188:1377258:1377314 [6] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-041-188:1377258:1377314 [6] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-041-188:1377256:1377311 [4] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-041-188:1377256:1377311 [4] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-041-188:1377254:1377315 [2] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-041-188:1377254:1377315 [2] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-041-188:1377254:1377315 [2] NCCL INFO Retrieving state for FasTrak +n116-041-188:1377254:1377315 [2] NCCL INFO Initialized state 0 for FasTrak +n116-041-188:1377256:1377311 [4] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-041-188:1377256:1377311 [4] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-041-188:1377254:1377315 [2] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-041-188:1377254:1377315 [2] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-041-188:1377256:1377311 [4] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-041-188:1377254:1377315 [2] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-041-188:1377256:1377311 [4] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-041-188:1377254:1377315 [2] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-041-188:1377254:1377315 [2] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-041-188:1377254:1377315 [2] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-041-188:1377254:1377315 [2] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-041-188:1377254:1377315 [2] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-041-188:1377253:1377310 [1] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-041-188:1377253:1377310 [1] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-041-188:1377253:1377310 [1] NCCL INFO Retrieving state for FasTrak +n116-041-188:1377253:1377310 [1] NCCL INFO Initialized state 0 for FasTrak +n116-041-188:1377253:1377310 [1] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-041-188:1377253:1377310 [1] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-041-188:1377253:1377310 [1] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-041-188:1377253:1377310 [1] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-041-188:1377255:1377309 [3] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-041-188:1377255:1377309 [3] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-041-188:1377255:1377309 [3] NCCL INFO Retrieving state for FasTrak +n116-041-188:1377255:1377309 [3] NCCL INFO Initialized state 0 for FasTrak +n116-041-188:1377253:1377310 [1] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-041-188:1377253:1377310 [1] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-041-188:1377255:1377309 [3] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-041-188:1377253:1377310 [1] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-041-188:1377255:1377309 [3] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-041-188:1377253:1377310 [1] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-041-188:1377255:1377309 [3] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-041-188:1377255:1377309 [3] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-041-188:1377255:1377309 [3] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-041-188:1377255:1377309 [3] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-041-188:1377255:1377309 [3] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-041-188:1377255:1377309 [3] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-041-188:1377252:1377312 [0] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-041-188:1377252:1377312 [0] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-041-188:1377252:1377312 [0] NCCL INFO Retrieving state for FasTrak +n116-041-188:1377252:1377312 [0] NCCL INFO Initialized state 0 for FasTrak +n116-041-188:1377252:1377312 [0] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-041-188:1377252:1377312 [0] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-041-188:1377252:1377312 [0] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-041-188:1377252:1377312 [0] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-041-188:1377252:1377312 [0] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-041-188:1377252:1377312 [0] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-041-188:1377252:1377312 [0] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-041-188:1377252:1377312 [0] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-041-188:1377258:1377314 [6] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-041-188:1377257:1377308 [5] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-041-188:1377259:1377313 [7] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-041-188:1377256:1377311 [4] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-041-188:1377255:1377309 [3] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-041-188:1377254:1377315 [2] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-041-188:1377258:1377314 [6] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-041-188:1377258:1377314 [6] NCCL INFO CPU/0-0 (1/1/3) +n116-041-188:1377258:1377314 [6] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-041-188:1377258:1377314 [6] NCCL INFO + PCI[24.0] - GPU/0-4000 (16) +n116-041-188:1377258:1377314 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377258:1377314 [6] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-041-188:1377258:1377314 [6] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-041-188:1377258:1377314 [6] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-041-188:1377258:1377314 [6] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-041-188:1377258:1377314 [6] NCCL INFO + PCI[24.0] - GPU/0-5000 (17) +n116-041-188:1377258:1377314 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377258:1377314 [6] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-041-188:1377258:1377314 [6] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-041-188:1377258:1377314 [6] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-041-188:1377258:1377314 [6] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-041-188:1377258:1377314 [6] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-041-188:1377258:1377314 [6] NCCL INFO + PCI[24.0] - GPU/0-b000 (18) +n116-041-188:1377258:1377314 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377258:1377314 [6] NCCL INFO + PCI[24.0] - GPU/0-c000 (19) +n116-041-188:1377258:1377314 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377258:1377314 [6] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-041-188:1377258:1377314 [6] NCCL INFO CPU/0-1 (1/1/3) +n116-041-188:1377258:1377314 [6] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-041-188:1377258:1377314 [6] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-041-188:1377258:1377314 [6] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-041-188:1377258:1377314 [6] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-041-188:1377258:1377314 [6] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-041-188:1377258:1377314 [6] NCCL INFO + PCI[24.0] - GPU/0-84000 (20) +n116-041-188:1377258:1377314 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377258:1377314 [6] NCCL INFO + PCI[24.0] - GPU/0-85000 (21) +n116-041-188:1377258:1377314 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377258:1377314 [6] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-041-188:1377258:1377314 [6] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-041-188:1377258:1377314 [6] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-041-188:1377258:1377314 [6] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-041-188:1377258:1377314 [6] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-041-188:1377258:1377314 [6] NCCL INFO + PCI[24.0] - GPU/0-8b000 (22) +n116-041-188:1377258:1377314 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377258:1377314 [6] NCCL INFO + PCI[24.0] - GPU/0-8c000 (23) +n116-041-188:1377258:1377314 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377258:1377314 [6] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-041-188:1377258:1377314 [6] NCCL INFO ========================================== +n116-041-188:1377258:1377314 [6] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-188:1377258:1377314 [6] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-188:1377258:1377314 [6] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-188:1377258:1377314 [6] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-188:1377258:1377314 [6] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-188:1377258:1377314 [6] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-188:1377253:1377310 [1] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-041-188:1377258:1377314 [6] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-188:1377257:1377308 [5] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-041-188:1377258:1377314 [6] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-188:1377257:1377308 [5] NCCL INFO CPU/0-0 (1/1/3) +n116-041-188:1377257:1377308 [5] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-041-188:1377258:1377314 [6] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-188:1377257:1377308 [5] NCCL INFO + PCI[24.0] - GPU/0-4000 (16) +n116-041-188:1377257:1377308 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377257:1377308 [5] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-041-188:1377257:1377308 [5] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-041-188:1377258:1377314 [6] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-188:1377252:1377312 [0] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-041-188:1377257:1377308 [5] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-041-188:1377257:1377308 [5] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-041-188:1377257:1377308 [5] NCCL INFO + PCI[24.0] - GPU/0-5000 (17) +n116-041-188:1377257:1377308 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377257:1377308 [5] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-041-188:1377257:1377308 [5] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-041-188:1377258:1377314 [6] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-188:1377257:1377308 [5] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-041-188:1377257:1377308 [5] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-041-188:1377257:1377308 [5] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-041-188:1377257:1377308 [5] NCCL INFO + PCI[24.0] - GPU/0-b000 (18) +n116-041-188:1377257:1377308 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377257:1377308 [5] NCCL INFO + PCI[24.0] - GPU/0-c000 (19) +n116-041-188:1377257:1377308 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377258:1377314 [6] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-188:1377257:1377308 [5] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-041-188:1377257:1377308 [5] NCCL INFO CPU/0-1 (1/1/3) +n116-041-188:1377257:1377308 [5] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-041-188:1377257:1377308 [5] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-041-188:1377257:1377308 [5] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-041-188:1377258:1377314 [6] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-188:1377257:1377308 [5] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-041-188:1377257:1377308 [5] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-041-188:1377257:1377308 [5] NCCL INFO + PCI[24.0] - GPU/0-84000 (20) +n116-041-188:1377257:1377308 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377257:1377308 [5] NCCL INFO + PCI[24.0] - GPU/0-85000 (21) +n116-041-188:1377257:1377308 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377258:1377314 [6] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-188:1377257:1377308 [5] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-041-188:1377257:1377308 [5] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-041-188:1377257:1377308 [5] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-041-188:1377257:1377308 [5] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-041-188:1377257:1377308 [5] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-041-188:1377257:1377308 [5] NCCL INFO + PCI[24.0] - GPU/0-8b000 (22) +n116-041-188:1377258:1377314 [6] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-041-188:1377257:1377308 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377257:1377308 [5] NCCL INFO + PCI[24.0] - GPU/0-8c000 (23) +n116-041-188:1377257:1377308 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377257:1377308 [5] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-041-188:1377257:1377308 [5] NCCL INFO ========================================== +n116-041-188:1377258:1377314 [6] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-041-188:1377257:1377308 [5] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-188:1377258:1377314 [6] NCCL INFO Setting affinity for GPU 6 to 7fff,ffffffff,e0000000,0000007f,ffffffff,ffe00000,00000000 +n116-041-188:1377257:1377308 [5] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-188:1377258:1377314 [6] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-041-188:1377257:1377308 [5] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-188:1377259:1377313 [7] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-041-188:1377259:1377313 [7] NCCL INFO CPU/0-0 (1/1/3) +n116-041-188:1377257:1377308 [5] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-188:1377259:1377313 [7] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-041-188:1377259:1377313 [7] NCCL INFO + PCI[24.0] - GPU/0-4000 (16) +n116-041-188:1377257:1377308 [5] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-188:1377259:1377313 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377259:1377313 [7] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-041-188:1377259:1377313 [7] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-041-188:1377259:1377313 [7] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-041-188:1377257:1377308 [5] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-188:1377259:1377313 [7] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-041-188:1377259:1377313 [7] NCCL INFO + PCI[24.0] - GPU/0-5000 (17) +n116-041-188:1377259:1377313 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377259:1377313 [7] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-041-188:1377259:1377313 [7] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-041-188:1377258:1377314 [6] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-041-188:1377259:1377313 [7] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-041-188:1377259:1377313 [7] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-041-188:1377257:1377308 [5] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-188:1377259:1377313 [7] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-041-188:1377259:1377313 [7] NCCL INFO + PCI[24.0] - GPU/0-b000 (18) +n116-041-188:1377259:1377313 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377259:1377313 [7] NCCL INFO + PCI[24.0] - GPU/0-c000 (19) +n116-041-188:1377259:1377313 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377259:1377313 [7] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-041-188:1377257:1377308 [5] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-188:1377259:1377313 [7] NCCL INFO CPU/0-1 (1/1/3) +n116-041-188:1377259:1377313 [7] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-041-188:1377259:1377313 [7] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-041-188:1377257:1377308 [5] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-188:1377259:1377313 [7] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-041-188:1377259:1377313 [7] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-041-188:1377259:1377313 [7] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-041-188:1377259:1377313 [7] NCCL INFO + PCI[24.0] - GPU/0-84000 (20) +n116-041-188:1377257:1377308 [5] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-188:1377259:1377313 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377259:1377313 [7] NCCL INFO + PCI[24.0] - GPU/0-85000 (21) +n116-041-188:1377259:1377313 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377259:1377313 [7] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-041-188:1377257:1377308 [5] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-188:1377259:1377313 [7] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-041-188:1377259:1377313 [7] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-041-188:1377259:1377313 [7] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-041-188:1377257:1377308 [5] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-188:1377259:1377313 [7] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-041-188:1377259:1377313 [7] NCCL INFO + PCI[24.0] - GPU/0-8b000 (22) +n116-041-188:1377259:1377313 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377257:1377308 [5] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-188:1377259:1377313 [7] NCCL INFO + PCI[24.0] - GPU/0-8c000 (23) +n116-041-188:1377259:1377313 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377259:1377313 [7] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-041-188:1377259:1377313 [7] NCCL INFO ========================================== +n116-041-188:1377257:1377308 [5] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-188:1377257:1377308 [5] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-041-188:1377259:1377313 [7] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-188:1377257:1377308 [5] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-041-188:1377259:1377313 [7] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-188:1377257:1377308 [5] NCCL INFO Setting affinity for GPU 5 to 7fff,ffffffff,e0000000,0000007f,ffffffff,ffe00000,00000000 +n116-041-188:1377259:1377313 [7] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-188:1377257:1377308 [5] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-041-188:1377259:1377313 [7] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-188:1377259:1377313 [7] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-188:1377257:1377308 [5] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-041-188:1377259:1377313 [7] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-188:1377259:1377313 [7] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-188:1377259:1377313 [7] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-188:1377259:1377313 [7] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-188:1377259:1377313 [7] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-188:1377259:1377313 [7] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-188:1377259:1377313 [7] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-188:1377259:1377313 [7] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-188:1377259:1377313 [7] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-188:1377256:1377311 [4] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-041-188:1377259:1377313 [7] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-041-188:1377256:1377311 [4] NCCL INFO CPU/0-0 (1/1/3) +n116-041-188:1377256:1377311 [4] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-041-188:1377256:1377311 [4] NCCL INFO + PCI[24.0] - GPU/0-4000 (16) +n116-041-188:1377256:1377311 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377259:1377313 [7] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-041-188:1377256:1377311 [4] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-041-188:1377256:1377311 [4] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-041-188:1377256:1377311 [4] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-041-188:1377256:1377311 [4] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-041-188:1377256:1377311 [4] NCCL INFO + PCI[24.0] - GPU/0-5000 (17) +n116-041-188:1377256:1377311 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377259:1377313 [7] NCCL INFO Setting affinity for GPU 7 to 7fff,ffffffff,e0000000,0000007f,ffffffff,ffe00000,00000000 +n116-041-188:1377256:1377311 [4] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-041-188:1377256:1377311 [4] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-041-188:1377256:1377311 [4] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-041-188:1377256:1377311 [4] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-041-188:1377256:1377311 [4] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-041-188:1377256:1377311 [4] NCCL INFO + PCI[24.0] - GPU/0-b000 (18) +n116-041-188:1377256:1377311 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377256:1377311 [4] NCCL INFO + PCI[24.0] - GPU/0-c000 (19) +n116-041-188:1377256:1377311 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377256:1377311 [4] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-041-188:1377256:1377311 [4] NCCL INFO CPU/0-1 (1/1/3) +n116-041-188:1377256:1377311 [4] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-041-188:1377256:1377311 [4] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-041-188:1377256:1377311 [4] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-041-188:1377256:1377311 [4] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-041-188:1377256:1377311 [4] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-041-188:1377256:1377311 [4] NCCL INFO + PCI[24.0] - GPU/0-84000 (20) +n116-041-188:1377256:1377311 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377256:1377311 [4] NCCL INFO + PCI[24.0] - GPU/0-85000 (21) +n116-041-188:1377256:1377311 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377256:1377311 [4] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-041-188:1377256:1377311 [4] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-041-188:1377256:1377311 [4] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-041-188:1377256:1377311 [4] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-041-188:1377256:1377311 [4] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-041-188:1377256:1377311 [4] NCCL INFO + PCI[24.0] - GPU/0-8b000 (22) +n116-041-188:1377256:1377311 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377256:1377311 [4] NCCL INFO + PCI[24.0] - GPU/0-8c000 (23) +n116-041-188:1377256:1377311 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377256:1377311 [4] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-041-188:1377256:1377311 [4] NCCL INFO ========================================== +n116-041-188:1377256:1377311 [4] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-188:1377259:1377313 [7] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-041-188:1377256:1377311 [4] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-188:1377256:1377311 [4] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-188:1377256:1377311 [4] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-188:1377256:1377311 [4] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-188:1377259:1377313 [7] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-041-188:1377256:1377311 [4] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-188:1377256:1377311 [4] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-188:1377256:1377311 [4] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-188:1377256:1377311 [4] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-188:1377256:1377311 [4] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-188:1377256:1377311 [4] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-188:1377256:1377311 [4] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-188:1377256:1377311 [4] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-188:1377256:1377311 [4] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-188:1377256:1377311 [4] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-041-188:1377256:1377311 [4] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-041-188:1377256:1377311 [4] NCCL INFO Setting affinity for GPU 4 to 7fff,ffffffff,e0000000,0000007f,ffffffff,ffe00000,00000000 +n116-041-188:1377256:1377311 [4] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-041-188:1377256:1377311 [4] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-041-188:1377255:1377309 [3] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-041-188:1377255:1377309 [3] NCCL INFO CPU/0-0 (1/1/3) +n116-041-188:1377254:1377315 [2] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-041-188:1377255:1377309 [3] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-041-188:1377254:1377315 [2] NCCL INFO CPU/0-0 (1/1/3) +n116-041-188:1377255:1377309 [3] NCCL INFO + PCI[24.0] - GPU/0-4000 (16) +n116-041-188:1377255:1377309 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377254:1377315 [2] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-041-188:1377255:1377309 [3] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-041-188:1377254:1377315 [2] NCCL INFO + PCI[24.0] - GPU/0-4000 (16) +n116-041-188:1377255:1377309 [3] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-041-188:1377254:1377315 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377255:1377309 [3] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-041-188:1377254:1377315 [2] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-041-188:1377255:1377309 [3] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-041-188:1377254:1377315 [2] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-041-188:1377255:1377309 [3] NCCL INFO + PCI[24.0] - GPU/0-5000 (17) +n116-041-188:1377254:1377315 [2] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-041-188:1377255:1377309 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377255:1377309 [3] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-041-188:1377254:1377315 [2] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-041-188:1377255:1377309 [3] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-041-188:1377254:1377315 [2] NCCL INFO + PCI[24.0] - GPU/0-5000 (17) +n116-041-188:1377255:1377309 [3] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-041-188:1377254:1377315 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377255:1377309 [3] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-041-188:1377254:1377315 [2] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-041-188:1377255:1377309 [3] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-041-188:1377254:1377315 [2] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-041-188:1377255:1377309 [3] NCCL INFO + PCI[24.0] - GPU/0-b000 (18) +n116-041-188:1377255:1377309 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377254:1377315 [2] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-041-188:1377255:1377309 [3] NCCL INFO + PCI[24.0] - GPU/0-c000 (19) +n116-041-188:1377254:1377315 [2] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-041-188:1377255:1377309 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377254:1377315 [2] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-041-188:1377255:1377309 [3] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-041-188:1377254:1377315 [2] NCCL INFO + PCI[24.0] - GPU/0-b000 (18) +n116-041-188:1377255:1377309 [3] NCCL INFO CPU/0-1 (1/1/3) +n116-041-188:1377254:1377315 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377255:1377309 [3] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-041-188:1377254:1377315 [2] NCCL INFO + PCI[24.0] - GPU/0-c000 (19) +n116-041-188:1377255:1377309 [3] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-041-188:1377254:1377315 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377255:1377309 [3] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-041-188:1377254:1377315 [2] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-041-188:1377255:1377309 [3] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-041-188:1377254:1377315 [2] NCCL INFO CPU/0-1 (1/1/3) +n116-041-188:1377255:1377309 [3] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-041-188:1377254:1377315 [2] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-041-188:1377254:1377315 [2] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-041-188:1377255:1377309 [3] NCCL INFO + PCI[24.0] - GPU/0-84000 (20) +n116-041-188:1377255:1377309 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377254:1377315 [2] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-041-188:1377255:1377309 [3] NCCL INFO + PCI[24.0] - GPU/0-85000 (21) +n116-041-188:1377254:1377315 [2] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-041-188:1377255:1377309 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377254:1377315 [2] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-041-188:1377255:1377309 [3] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-041-188:1377254:1377315 [2] NCCL INFO + PCI[24.0] - GPU/0-84000 (20) +n116-041-188:1377255:1377309 [3] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-041-188:1377254:1377315 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377254:1377315 [2] NCCL INFO + PCI[24.0] - GPU/0-85000 (21) +n116-041-188:1377255:1377309 [3] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-041-188:1377254:1377315 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377255:1377309 [3] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-041-188:1377254:1377315 [2] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-041-188:1377255:1377309 [3] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-041-188:1377254:1377315 [2] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-041-188:1377255:1377309 [3] NCCL INFO + PCI[24.0] - GPU/0-8b000 (22) +n116-041-188:1377254:1377315 [2] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-041-188:1377255:1377309 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377254:1377315 [2] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-041-188:1377255:1377309 [3] NCCL INFO + PCI[24.0] - GPU/0-8c000 (23) +n116-041-188:1377255:1377309 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377254:1377315 [2] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-041-188:1377255:1377309 [3] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-041-188:1377254:1377315 [2] NCCL INFO + PCI[24.0] - GPU/0-8b000 (22) +n116-041-188:1377255:1377309 [3] NCCL INFO ========================================== +n116-041-188:1377254:1377315 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377254:1377315 [2] NCCL INFO + PCI[24.0] - GPU/0-8c000 (23) +n116-041-188:1377254:1377315 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377254:1377315 [2] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-041-188:1377254:1377315 [2] NCCL INFO ========================================== +n116-041-188:1377255:1377309 [3] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-188:1377254:1377315 [2] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-188:1377255:1377309 [3] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-188:1377254:1377315 [2] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-188:1377255:1377309 [3] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-188:1377254:1377315 [2] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-188:1377254:1377315 [2] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-188:1377255:1377309 [3] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-188:1377255:1377309 [3] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-188:1377254:1377315 [2] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-188:1377254:1377315 [2] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-188:1377255:1377309 [3] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-188:1377255:1377309 [3] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-188:1377254:1377315 [2] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-188:1377254:1377315 [2] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-188:1377255:1377309 [3] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-188:1377255:1377309 [3] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-188:1377254:1377315 [2] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-188:1377254:1377315 [2] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-188:1377255:1377309 [3] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-188:1377255:1377309 [3] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-188:1377254:1377315 [2] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-188:1377254:1377315 [2] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-188:1377255:1377309 [3] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-188:1377254:1377315 [2] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-188:1377255:1377309 [3] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-188:1377254:1377315 [2] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-188:1377255:1377309 [3] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-188:1377255:1377309 [3] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-041-188:1377254:1377315 [2] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-041-188:1377255:1377309 [3] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-041-188:1377254:1377315 [2] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-041-188:1377254:1377315 [2] NCCL INFO Setting affinity for GPU 2 to 0fffffff,fffffe00,00000000,000fffff,fffffffe +n116-041-188:1377255:1377309 [3] NCCL INFO Setting affinity for GPU 3 to 0fffffff,fffffe00,00000000,000fffff,fffffffe +n116-041-188:1377254:1377315 [2] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-041-188:1377255:1377309 [3] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-041-188:1377255:1377309 [3] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-041-188:1377254:1377315 [2] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-041-188:1377252:1377312 [0] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-041-188:1377252:1377312 [0] NCCL INFO CPU/0-0 (1/1/3) +n116-041-188:1377252:1377312 [0] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-041-188:1377252:1377312 [0] NCCL INFO + PCI[24.0] - GPU/0-4000 (16) +n116-041-188:1377252:1377312 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377252:1377312 [0] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-041-188:1377252:1377312 [0] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-041-188:1377252:1377312 [0] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-041-188:1377252:1377312 [0] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-041-188:1377252:1377312 [0] NCCL INFO + PCI[24.0] - GPU/0-5000 (17) +n116-041-188:1377252:1377312 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377252:1377312 [0] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-041-188:1377252:1377312 [0] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-041-188:1377252:1377312 [0] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-041-188:1377252:1377312 [0] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-041-188:1377252:1377312 [0] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-041-188:1377252:1377312 [0] NCCL INFO + PCI[24.0] - GPU/0-b000 (18) +n116-041-188:1377252:1377312 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377252:1377312 [0] NCCL INFO + PCI[24.0] - GPU/0-c000 (19) +n116-041-188:1377252:1377312 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377252:1377312 [0] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-041-188:1377252:1377312 [0] NCCL INFO CPU/0-1 (1/1/3) +n116-041-188:1377252:1377312 [0] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-041-188:1377252:1377312 [0] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-041-188:1377252:1377312 [0] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-041-188:1377252:1377312 [0] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-041-188:1377252:1377312 [0] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-041-188:1377252:1377312 [0] NCCL INFO + PCI[24.0] - GPU/0-84000 (20) +n116-041-188:1377252:1377312 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377252:1377312 [0] NCCL INFO + PCI[24.0] - GPU/0-85000 (21) +n116-041-188:1377252:1377312 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377252:1377312 [0] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-041-188:1377252:1377312 [0] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-041-188:1377252:1377312 [0] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-041-188:1377252:1377312 [0] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-041-188:1377252:1377312 [0] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-041-188:1377252:1377312 [0] NCCL INFO + PCI[24.0] - GPU/0-8b000 (22) +n116-041-188:1377252:1377312 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377252:1377312 [0] NCCL INFO + PCI[24.0] - GPU/0-8c000 (23) +n116-041-188:1377252:1377312 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377252:1377312 [0] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-041-188:1377252:1377312 [0] NCCL INFO ========================================== +n116-041-188:1377252:1377312 [0] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-188:1377252:1377312 [0] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-188:1377252:1377312 [0] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-188:1377252:1377312 [0] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-188:1377252:1377312 [0] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-188:1377252:1377312 [0] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-188:1377252:1377312 [0] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-188:1377252:1377312 [0] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-188:1377252:1377312 [0] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-188:1377252:1377312 [0] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-188:1377252:1377312 [0] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-188:1377252:1377312 [0] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-188:1377252:1377312 [0] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-188:1377252:1377312 [0] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-188:1377252:1377312 [0] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-041-188:1377252:1377312 [0] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-041-188:1377252:1377312 [0] NCCL INFO Setting affinity for GPU 0 to 0fffffff,fffffe00,00000000,000fffff,fffffffe +n116-041-188:1377252:1377312 [0] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-041-188:1377252:1377312 [0] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-041-188:1377253:1377310 [1] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-041-188:1377253:1377310 [1] NCCL INFO CPU/0-0 (1/1/3) +n116-041-188:1377253:1377310 [1] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-041-188:1377253:1377310 [1] NCCL INFO + PCI[24.0] - GPU/0-4000 (16) +n116-041-188:1377253:1377310 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377253:1377310 [1] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-041-188:1377253:1377310 [1] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-041-188:1377253:1377310 [1] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-041-188:1377253:1377310 [1] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-041-188:1377253:1377310 [1] NCCL INFO + PCI[24.0] - GPU/0-5000 (17) +n116-041-188:1377253:1377310 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377253:1377310 [1] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-041-188:1377253:1377310 [1] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-041-188:1377253:1377310 [1] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-041-188:1377253:1377310 [1] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-041-188:1377253:1377310 [1] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-041-188:1377253:1377310 [1] NCCL INFO + PCI[24.0] - GPU/0-b000 (18) +n116-041-188:1377253:1377310 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377253:1377310 [1] NCCL INFO + PCI[24.0] - GPU/0-c000 (19) +n116-041-188:1377253:1377310 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377253:1377310 [1] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-041-188:1377253:1377310 [1] NCCL INFO CPU/0-1 (1/1/3) +n116-041-188:1377253:1377310 [1] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-041-188:1377253:1377310 [1] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-041-188:1377253:1377310 [1] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-041-188:1377253:1377310 [1] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-041-188:1377253:1377310 [1] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-041-188:1377253:1377310 [1] NCCL INFO + PCI[24.0] - GPU/0-84000 (20) +n116-041-188:1377253:1377310 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377253:1377310 [1] NCCL INFO + PCI[24.0] - GPU/0-85000 (21) +n116-041-188:1377253:1377310 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377253:1377310 [1] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-041-188:1377253:1377310 [1] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-041-188:1377253:1377310 [1] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-041-188:1377253:1377310 [1] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-041-188:1377253:1377310 [1] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-041-188:1377253:1377310 [1] NCCL INFO + PCI[24.0] - GPU/0-8b000 (22) +n116-041-188:1377253:1377310 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377253:1377310 [1] NCCL INFO + PCI[24.0] - GPU/0-8c000 (23) +n116-041-188:1377253:1377310 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-188:1377253:1377310 [1] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-041-188:1377253:1377310 [1] NCCL INFO ========================================== +n116-041-188:1377253:1377310 [1] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-188:1377253:1377310 [1] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-188:1377253:1377310 [1] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-188:1377253:1377310 [1] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-188:1377253:1377310 [1] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-188:1377253:1377310 [1] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-188:1377253:1377310 [1] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-188:1377253:1377310 [1] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-188:1377253:1377310 [1] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-188:1377253:1377310 [1] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-188:1377253:1377310 [1] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-188:1377253:1377310 [1] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-188:1377253:1377310 [1] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-188:1377253:1377310 [1] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-188:1377253:1377310 [1] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-041-188:1377253:1377310 [1] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-041-188:1377253:1377310 [1] NCCL INFO Setting affinity for GPU 1 to 0fffffff,fffffe00,00000000,000fffff,fffffffe +n116-041-188:1377253:1377310 [1] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-041-188:1377253:1377310 [1] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-041-188:1377258:1377314 [6] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-188:1377258:1377314 [6] NCCL INFO 0 : NET/0-0 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 NET/0-0 +n116-041-188:1377258:1377314 [6] NCCL INFO 1 : NET/0-1 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 NET/0-1 +n116-041-188:1377258:1377314 [6] NCCL INFO 2 : NET/0-2 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 NET/0-2 +n116-041-188:1377258:1377314 [6] NCCL INFO 3 : NET/0-3 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 NET/0-3 +n116-041-188:1377258:1377314 [6] NCCL INFO 4 : NET/0-4 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 NET/0-4 +n116-041-188:1377258:1377314 [6] NCCL INFO 5 : NET/0-5 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 NET/0-5 +n116-041-188:1377258:1377314 [6] NCCL INFO 6 : NET/0-6 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 NET/0-6 +n116-041-188:1377258:1377314 [6] NCCL INFO 7 : NET/0-7 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 NET/0-7 +n116-041-188:1377258:1377314 [6] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-188:1377258:1377314 [6] NCCL INFO 0 : NET/0-0 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 NET/0-0 +n116-041-188:1377258:1377314 [6] NCCL INFO 1 : NET/0-1 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 NET/0-1 +n116-041-188:1377258:1377314 [6] NCCL INFO 2 : NET/0-2 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 NET/0-2 +n116-041-188:1377258:1377314 [6] NCCL INFO 3 : NET/0-3 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 NET/0-3 +n116-041-188:1377258:1377314 [6] NCCL INFO 4 : NET/0-4 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 NET/0-4 +n116-041-188:1377258:1377314 [6] NCCL INFO 5 : NET/0-5 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 NET/0-5 +n116-041-188:1377258:1377314 [6] NCCL INFO 6 : NET/0-6 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 NET/0-6 +n116-041-188:1377258:1377314 [6] NCCL INFO 7 : NET/0-7 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 NET/0-7 +n116-041-188:1377257:1377308 [5] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-188:1377257:1377308 [5] NCCL INFO 0 : NET/0-0 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 NET/0-0 +n116-041-188:1377257:1377308 [5] NCCL INFO 1 : NET/0-1 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 NET/0-1 +n116-041-188:1377257:1377308 [5] NCCL INFO 2 : NET/0-2 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 NET/0-2 +n116-041-188:1377257:1377308 [5] NCCL INFO 3 : NET/0-3 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 NET/0-3 +n116-041-188:1377257:1377308 [5] NCCL INFO 4 : NET/0-4 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 NET/0-4 +n116-041-188:1377257:1377308 [5] NCCL INFO 5 : NET/0-5 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 NET/0-5 +n116-041-188:1377257:1377308 [5] NCCL INFO 6 : NET/0-6 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 NET/0-6 +n116-041-188:1377257:1377308 [5] NCCL INFO 7 : NET/0-7 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 NET/0-7 +n116-041-188:1377257:1377308 [5] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-188:1377257:1377308 [5] NCCL INFO 0 : NET/0-0 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 NET/0-0 +n116-041-188:1377257:1377308 [5] NCCL INFO 1 : NET/0-1 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 NET/0-1 +n116-041-188:1377257:1377308 [5] NCCL INFO 2 : NET/0-2 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 NET/0-2 +n116-041-188:1377257:1377308 [5] NCCL INFO 3 : NET/0-3 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 NET/0-3 +n116-041-188:1377257:1377308 [5] NCCL INFO 4 : NET/0-4 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 NET/0-4 +n116-041-188:1377257:1377308 [5] NCCL INFO 5 : NET/0-5 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 NET/0-5 +n116-041-188:1377257:1377308 [5] NCCL INFO 6 : NET/0-6 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 NET/0-6 +n116-041-188:1377257:1377308 [5] NCCL INFO 7 : NET/0-7 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 NET/0-7 +n116-041-188:1377259:1377313 [7] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-188:1377256:1377311 [4] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-188:1377256:1377311 [4] NCCL INFO 0 : NET/0-0 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 NET/0-0 +n116-041-188:1377256:1377311 [4] NCCL INFO 1 : NET/0-1 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 NET/0-1 +n116-041-188:1377256:1377311 [4] NCCL INFO 2 : NET/0-2 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 NET/0-2 +n116-041-188:1377259:1377313 [7] NCCL INFO 0 : NET/0-0 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 NET/0-0 +n116-041-188:1377256:1377311 [4] NCCL INFO 3 : NET/0-3 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 NET/0-3 +n116-041-188:1377259:1377313 [7] NCCL INFO 1 : NET/0-1 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 NET/0-1 +n116-041-188:1377256:1377311 [4] NCCL INFO 4 : NET/0-4 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 NET/0-4 +n116-041-188:1377259:1377313 [7] NCCL INFO 2 : NET/0-2 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 NET/0-2 +n116-041-188:1377256:1377311 [4] NCCL INFO 5 : NET/0-5 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 NET/0-5 +n116-041-188:1377259:1377313 [7] NCCL INFO 3 : NET/0-3 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 NET/0-3 +n116-041-188:1377256:1377311 [4] NCCL INFO 6 : NET/0-6 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 NET/0-6 +n116-041-188:1377259:1377313 [7] NCCL INFO 4 : NET/0-4 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 NET/0-4 +n116-041-188:1377256:1377311 [4] NCCL INFO 7 : NET/0-7 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 NET/0-7 +n116-041-188:1377259:1377313 [7] NCCL INFO 5 : NET/0-5 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 NET/0-5 +n116-041-188:1377259:1377313 [7] NCCL INFO 6 : NET/0-6 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 NET/0-6 +n116-041-188:1377259:1377313 [7] NCCL INFO 7 : NET/0-7 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 NET/0-7 +n116-041-188:1377252:1377312 [0] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-188:1377252:1377312 [0] NCCL INFO 0 : NET/0-0 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 NET/0-0 +n116-041-188:1377252:1377312 [0] NCCL INFO 1 : NET/0-1 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 NET/0-1 +n116-041-188:1377252:1377312 [0] NCCL INFO 2 : NET/0-2 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 NET/0-2 +n116-041-188:1377252:1377312 [0] NCCL INFO 3 : NET/0-3 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 NET/0-3 +n116-041-188:1377252:1377312 [0] NCCL INFO 4 : NET/0-4 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 NET/0-4 +n116-041-188:1377252:1377312 [0] NCCL INFO 5 : NET/0-5 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 NET/0-5 +n116-041-188:1377252:1377312 [0] NCCL INFO 6 : NET/0-6 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 NET/0-6 +n116-041-188:1377252:1377312 [0] NCCL INFO 7 : NET/0-7 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 NET/0-7 +n116-041-188:1377256:1377311 [4] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-188:1377256:1377311 [4] NCCL INFO 0 : NET/0-0 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 NET/0-0 +n116-041-188:1377256:1377311 [4] NCCL INFO 1 : NET/0-1 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 NET/0-1 +n116-041-188:1377256:1377311 [4] NCCL INFO 2 : NET/0-2 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 NET/0-2 +n116-041-188:1377256:1377311 [4] NCCL INFO 3 : NET/0-3 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 NET/0-3 +n116-041-188:1377256:1377311 [4] NCCL INFO 4 : NET/0-4 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 NET/0-4 +n116-041-188:1377259:1377313 [7] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-188:1377256:1377311 [4] NCCL INFO 5 : NET/0-5 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 NET/0-5 +n116-041-188:1377259:1377313 [7] NCCL INFO 0 : NET/0-0 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 NET/0-0 +n116-041-188:1377256:1377311 [4] NCCL INFO 6 : NET/0-6 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 NET/0-6 +n116-041-188:1377259:1377313 [7] NCCL INFO 1 : NET/0-1 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 NET/0-1 +n116-041-188:1377256:1377311 [4] NCCL INFO 7 : NET/0-7 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 NET/0-7 +n116-041-188:1377259:1377313 [7] NCCL INFO 2 : NET/0-2 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 NET/0-2 +n116-041-188:1377259:1377313 [7] NCCL INFO 3 : NET/0-3 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 NET/0-3 +n116-041-188:1377259:1377313 [7] NCCL INFO 4 : NET/0-4 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 NET/0-4 +n116-041-188:1377259:1377313 [7] NCCL INFO 5 : NET/0-5 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 NET/0-5 +n116-041-188:1377259:1377313 [7] NCCL INFO 6 : NET/0-6 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 NET/0-6 +n116-041-188:1377259:1377313 [7] NCCL INFO 7 : NET/0-7 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 NET/0-7 +n116-041-188:1377252:1377312 [0] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-188:1377252:1377312 [0] NCCL INFO 0 : NET/0-0 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 NET/0-0 +n116-041-188:1377252:1377312 [0] NCCL INFO 1 : NET/0-1 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 NET/0-1 +n116-041-188:1377252:1377312 [0] NCCL INFO 2 : NET/0-2 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 NET/0-2 +n116-041-188:1377252:1377312 [0] NCCL INFO 3 : NET/0-3 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 NET/0-3 +n116-041-188:1377252:1377312 [0] NCCL INFO 4 : NET/0-4 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 NET/0-4 +n116-041-188:1377252:1377312 [0] NCCL INFO 5 : NET/0-5 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 NET/0-5 +n116-041-188:1377252:1377312 [0] NCCL INFO 6 : NET/0-6 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 NET/0-6 +n116-041-188:1377252:1377312 [0] NCCL INFO 7 : NET/0-7 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 NET/0-7 +n116-041-188:1377253:1377310 [1] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-188:1377253:1377310 [1] NCCL INFO 0 : NET/0-0 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 NET/0-0 +n116-041-188:1377253:1377310 [1] NCCL INFO 1 : NET/0-1 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 NET/0-1 +n116-041-188:1377253:1377310 [1] NCCL INFO 2 : NET/0-2 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 NET/0-2 +n116-041-188:1377253:1377310 [1] NCCL INFO 3 : NET/0-3 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 NET/0-3 +n116-041-188:1377253:1377310 [1] NCCL INFO 4 : NET/0-4 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 NET/0-4 +n116-041-188:1377253:1377310 [1] NCCL INFO 5 : NET/0-5 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 NET/0-5 +n116-041-188:1377253:1377310 [1] NCCL INFO 6 : NET/0-6 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 NET/0-6 +n116-041-188:1377253:1377310 [1] NCCL INFO 7 : NET/0-7 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 NET/0-7 +n116-041-188:1377253:1377310 [1] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-188:1377253:1377310 [1] NCCL INFO 0 : NET/0-0 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 NET/0-0 +n116-041-188:1377253:1377310 [1] NCCL INFO 1 : NET/0-1 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 NET/0-1 +n116-041-188:1377253:1377310 [1] NCCL INFO 2 : NET/0-2 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 NET/0-2 +n116-041-188:1377253:1377310 [1] NCCL INFO 3 : NET/0-3 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 NET/0-3 +n116-041-188:1377253:1377310 [1] NCCL INFO 4 : NET/0-4 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 NET/0-4 +n116-041-188:1377253:1377310 [1] NCCL INFO 5 : NET/0-5 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 NET/0-5 +n116-041-188:1377253:1377310 [1] NCCL INFO 6 : NET/0-6 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 NET/0-6 +n116-041-188:1377253:1377310 [1] NCCL INFO 7 : NET/0-7 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 NET/0-7 +n116-041-188:1377255:1377309 [3] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-188:1377255:1377309 [3] NCCL INFO 0 : NET/0-0 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 NET/0-0 +n116-041-188:1377255:1377309 [3] NCCL INFO 1 : NET/0-1 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 NET/0-1 +n116-041-188:1377255:1377309 [3] NCCL INFO 2 : NET/0-2 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 NET/0-2 +n116-041-188:1377255:1377309 [3] NCCL INFO 3 : NET/0-3 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 NET/0-3 +n116-041-188:1377255:1377309 [3] NCCL INFO 4 : NET/0-4 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 NET/0-4 +n116-041-188:1377255:1377309 [3] NCCL INFO 5 : NET/0-5 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 NET/0-5 +n116-041-188:1377255:1377309 [3] NCCL INFO 6 : NET/0-6 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 NET/0-6 +n116-041-188:1377255:1377309 [3] NCCL INFO 7 : NET/0-7 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 NET/0-7 +n116-041-188:1377255:1377309 [3] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-188:1377255:1377309 [3] NCCL INFO 0 : NET/0-0 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 NET/0-0 +n116-041-188:1377255:1377309 [3] NCCL INFO 1 : NET/0-1 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 NET/0-1 +n116-041-188:1377255:1377309 [3] NCCL INFO 2 : NET/0-2 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 NET/0-2 +n116-041-188:1377255:1377309 [3] NCCL INFO 3 : NET/0-3 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 NET/0-3 +n116-041-188:1377255:1377309 [3] NCCL INFO 4 : NET/0-4 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 NET/0-4 +n116-041-188:1377255:1377309 [3] NCCL INFO 5 : NET/0-5 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 NET/0-5 +n116-041-188:1377255:1377309 [3] NCCL INFO 6 : NET/0-6 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 NET/0-6 +n116-041-188:1377255:1377309 [3] NCCL INFO 7 : NET/0-7 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 NET/0-7 +n116-041-188:1377254:1377315 [2] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-188:1377254:1377315 [2] NCCL INFO 0 : NET/0-0 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 NET/0-0 +n116-041-188:1377254:1377315 [2] NCCL INFO 1 : NET/0-1 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 NET/0-1 +n116-041-188:1377254:1377315 [2] NCCL INFO 2 : NET/0-2 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 NET/0-2 +n116-041-188:1377254:1377315 [2] NCCL INFO 3 : NET/0-3 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 NET/0-3 +n116-041-188:1377254:1377315 [2] NCCL INFO 4 : NET/0-4 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 NET/0-4 +n116-041-188:1377254:1377315 [2] NCCL INFO 5 : NET/0-5 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 NET/0-5 +n116-041-188:1377254:1377315 [2] NCCL INFO 6 : NET/0-6 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 NET/0-6 +n116-041-188:1377254:1377315 [2] NCCL INFO 7 : NET/0-7 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 NET/0-7 +n116-041-188:1377254:1377315 [2] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-188:1377254:1377315 [2] NCCL INFO 0 : NET/0-0 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 NET/0-0 +n116-041-188:1377254:1377315 [2] NCCL INFO 1 : NET/0-1 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 NET/0-1 +n116-041-188:1377254:1377315 [2] NCCL INFO 2 : NET/0-2 GPU/18 GPU/19 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 NET/0-2 +n116-041-188:1377254:1377315 [2] NCCL INFO 3 : NET/0-3 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 GPU/21 GPU/20 NET/0-3 +n116-041-188:1377254:1377315 [2] NCCL INFO 4 : NET/0-4 GPU/20 GPU/21 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 NET/0-4 +n116-041-188:1377254:1377315 [2] NCCL INFO 5 : NET/0-5 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 GPU/23 GPU/22 NET/0-5 +n116-041-188:1377254:1377315 [2] NCCL INFO 6 : NET/0-6 GPU/22 GPU/23 GPU/16 GPU/17 GPU/18 GPU/19 GPU/20 GPU/21 NET/0-6 +n116-041-188:1377254:1377315 [2] NCCL INFO 7 : NET/0-7 GPU/23 GPU/22 GPU/21 GPU/20 GPU/19 GPU/18 GPU/17 GPU/16 NET/0-7 +n116-041-188:1377259:1377313 [7] NCCL INFO comm 0xc17e960 rank 23 nRanks 64 nNodes 8 localRanks 8 localRank 7 MNNVL 0 +n116-041-188:1377258:1377314 [6] NCCL INFO comm 0xc3c2c30 rank 22 nRanks 64 nNodes 8 localRanks 8 localRank 6 MNNVL 0 +n116-041-188:1377257:1377308 [5] NCCL INFO comm 0xb18a360 rank 21 nRanks 64 nNodes 8 localRanks 8 localRank 5 MNNVL 0 +n116-041-188:1377256:1377311 [4] NCCL INFO comm 0xb9355b0 rank 20 nRanks 64 nNodes 8 localRanks 8 localRank 4 MNNVL 0 +n116-041-188:1377255:1377309 [3] NCCL INFO comm 0xcab1d40 rank 19 nRanks 64 nNodes 8 localRanks 8 localRank 3 MNNVL 0 +n116-041-188:1377258:1377314 [6] NCCL INFO Tree 6 : 39 -> 22 -> 23/30/-1 +n116-041-188:1377258:1377314 [6] NCCL INFO Tree 14 : 15 -> 22 -> 23/-1/-1 +n116-041-188:1377259:1377313 [7] NCCL INFO Tree 6 : 22 -> 23 -> 16/14/-1 +n116-041-188:1377258:1377314 [6] NCCL INFO Tree 7 : 23 -> 22 -> 21/15/-1 +n116-041-188:1377253:1377310 [1] NCCL INFO comm 0xcbee300 rank 17 nRanks 64 nNodes 8 localRanks 8 localRank 1 MNNVL 0 +n116-041-188:1377258:1377314 [6] NCCL INFO Tree 15 : 23 -> 22 -> 21/-1/-1 +n116-041-188:1377259:1377313 [7] NCCL INFO Tree 14 : 22 -> 23 -> 16/-1/-1 +n116-041-188:1377259:1377313 [7] NCCL INFO Tree 7 : 38 -> 23 -> 22/31/-1 +n116-041-188:1377254:1377315 [2] NCCL INFO comm 0xb3d7aa0 rank 18 nRanks 64 nNodes 8 localRanks 8 localRank 2 MNNVL 0 +n116-041-188:1377257:1377308 [5] NCCL INFO Tree 4 : 20 -> 21 -> 22/12/-1 +n116-041-188:1377259:1377313 [7] NCCL INFO Tree 15 : 14 -> 23 -> 22/-1/-1 +n116-041-188:1377252:1377312 [0] NCCL INFO comm 0xcaf2670 rank 16 nRanks 64 nNodes 8 localRanks 8 localRank 0 MNNVL 0 +n116-041-188:1377257:1377308 [5] NCCL INFO Tree 12 : 20 -> 21 -> 22/-1/-1 +n116-041-188:1377258:1377314 [6] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-041-188:1377257:1377308 [5] NCCL INFO Tree 5 : 36 -> 21 -> 20/29/-1 +n116-041-188:1377257:1377308 [5] NCCL INFO Tree 13 : 12 -> 21 -> 20/-1/-1 +n116-041-188:1377259:1377313 [7] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-041-188:1377256:1377311 [4] NCCL INFO Tree 4 : 37 -> 20 -> 21/28/-1 +n116-041-188:1377256:1377311 [4] NCCL INFO Tree 12 : 13 -> 20 -> 21/-1/-1 +n116-041-188:1377257:1377308 [5] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-041-188:1377256:1377311 [4] NCCL INFO Tree 5 : 21 -> 20 -> 19/13/-1 +n116-041-188:1377256:1377311 [4] NCCL INFO Tree 13 : 21 -> 20 -> 19/-1/-1 +n116-041-188:1377256:1377311 [4] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-041-188:1377255:1377309 [3] NCCL INFO Tree 2 : 18 -> 19 -> 20/10/-1 +n116-041-188:1377253:1377310 [1] NCCL INFO Tree 0 : 16 -> 17 -> 18/8/-1 +n116-041-188:1377255:1377309 [3] NCCL INFO Tree 10 : 18 -> 19 -> 20/-1/-1 +n116-041-188:1377253:1377310 [1] NCCL INFO Tree 8 : 16 -> 17 -> 18/-1/-1 +n116-041-188:1377253:1377310 [1] NCCL INFO Tree 1 : 32 -> 17 -> 16/25/-1 +n116-041-188:1377255:1377309 [3] NCCL INFO Tree 3 : 34 -> 19 -> 18/27/-1 +n116-041-188:1377253:1377310 [1] NCCL INFO Tree 9 : 8 -> 17 -> 16/-1/-1 +n116-041-188:1377252:1377312 [0] NCCL INFO Tree 0 : 33 -> 16 -> 17/24/-1 +n116-041-188:1377255:1377309 [3] NCCL INFO Tree 11 : 10 -> 19 -> 18/-1/-1 +n116-041-188:1377252:1377312 [0] NCCL INFO Tree 8 : 9 -> 16 -> 17/-1/-1 +n116-041-188:1377254:1377315 [2] NCCL INFO Tree 2 : 35 -> 18 -> 19/26/-1 +n116-041-188:1377253:1377310 [1] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-041-188:1377252:1377312 [0] NCCL INFO Tree 1 : 17 -> 16 -> 23/9/-1 +n116-041-188:1377254:1377315 [2] NCCL INFO Tree 10 : 11 -> 18 -> 19/-1/-1 +n116-041-188:1377258:1377314 [6] NCCL INFO Ring 00 : 23 -> 22 -> 21 +n116-041-188:1377255:1377309 [3] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-041-188:1377252:1377312 [0] NCCL INFO Tree 9 : 17 -> 16 -> 23/-1/-1 +n116-041-188:1377259:1377313 [7] NCCL INFO Ring 00 : 16 -> 23 -> 22 +n116-041-188:1377257:1377308 [5] NCCL INFO Ring 00 : 22 -> 21 -> 20 +n116-041-188:1377254:1377315 [2] NCCL INFO Tree 3 : 19 -> 18 -> 17/11/-1 +n116-041-188:1377258:1377314 [6] NCCL INFO Ring 01 : 21 -> 22 -> 23 +n116-041-188:1377259:1377313 [7] NCCL INFO Ring 01 : 22 -> 23 -> 16 +n116-041-188:1377259:1377313 [7] NCCL INFO Ring 02 : 16 -> 23 -> 22 +n116-041-188:1377257:1377308 [5] NCCL INFO Ring 01 : 20 -> 21 -> 22 +n116-041-188:1377254:1377315 [2] NCCL INFO Tree 11 : 19 -> 18 -> 17/-1/-1 +n116-041-188:1377258:1377314 [6] NCCL INFO Ring 02 : 23 -> 22 -> 21 +n116-041-188:1377252:1377312 [0] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-041-188:1377258:1377314 [6] NCCL INFO Ring 03 : 21 -> 22 -> 23 +n116-041-188:1377256:1377311 [4] NCCL INFO Ring 00 : 21 -> 20 -> 19 +n116-041-188:1377259:1377313 [7] NCCL INFO Ring 03 : 22 -> 23 -> 16 +n116-041-188:1377257:1377308 [5] NCCL INFO Ring 02 : 22 -> 21 -> 20 +n116-041-188:1377258:1377314 [6] NCCL INFO Ring 04 : 23 -> 22 -> 21 +n116-041-188:1377259:1377313 [7] NCCL INFO Ring 04 : 16 -> 23 -> 22 +n116-041-188:1377256:1377311 [4] NCCL INFO Ring 01 : 19 -> 20 -> 21 +n116-041-188:1377257:1377308 [5] NCCL INFO Ring 03 : 20 -> 21 -> 22 +n116-041-188:1377258:1377314 [6] NCCL INFO Ring 05 : 21 -> 22 -> 23 +n116-041-188:1377254:1377315 [2] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-041-188:1377259:1377313 [7] NCCL INFO Ring 05 : 22 -> 23 -> 16 +n116-041-188:1377256:1377311 [4] NCCL INFO Ring 02 : 21 -> 20 -> 19 +n116-041-188:1377259:1377313 [7] NCCL INFO Ring 06 : 16 -> 23 -> 30 +n116-041-188:1377257:1377308 [5] NCCL INFO Ring 04 : 22 -> 21 -> 28 +n116-041-188:1377258:1377314 [6] NCCL INFO Ring 06 : 15 -> 22 -> 21 +n116-041-188:1377253:1377310 [1] NCCL INFO Ring 00 : 18 -> 17 -> 24 +n116-041-188:1377259:1377313 [7] NCCL INFO Ring 07 : 14 -> 23 -> 16 +n116-041-188:1377256:1377311 [4] NCCL INFO Ring 03 : 19 -> 20 -> 21 +n116-041-188:1377257:1377308 [5] NCCL INFO Ring 05 : 12 -> 21 -> 22 +n116-041-188:1377259:1377313 [7] NCCL INFO Ring 08 : 16 -> 23 -> 22 +n116-041-188:1377258:1377314 [6] NCCL INFO Ring 07 : 21 -> 22 -> 31 +n116-041-188:1377253:1377310 [1] NCCL INFO Ring 01 : 8 -> 17 -> 18 +n116-041-188:1377257:1377308 [5] NCCL INFO Ring 06 : 22 -> 21 -> 20 +n116-041-188:1377259:1377313 [7] NCCL INFO Ring 09 : 22 -> 23 -> 16 +n116-041-188:1377256:1377311 [4] NCCL INFO Ring 04 : 13 -> 20 -> 19 +n116-041-188:1377258:1377314 [6] NCCL INFO Ring 08 : 23 -> 22 -> 21 +n116-041-188:1377253:1377310 [1] NCCL INFO Ring 02 : 18 -> 17 -> 16 +n116-041-188:1377259:1377313 [7] NCCL INFO Ring 10 : 16 -> 23 -> 22 +n116-041-188:1377257:1377308 [5] NCCL INFO Ring 07 : 20 -> 21 -> 22 +n116-041-188:1377256:1377311 [4] NCCL INFO Ring 05 : 19 -> 20 -> 29 +n116-041-188:1377258:1377314 [6] NCCL INFO Ring 09 : 21 -> 22 -> 23 +n116-041-188:1377257:1377308 [5] NCCL INFO Ring 08 : 22 -> 21 -> 20 +n116-041-188:1377256:1377311 [4] NCCL INFO Ring 06 : 21 -> 20 -> 19 +n116-041-188:1377258:1377314 [6] NCCL INFO Ring 10 : 23 -> 22 -> 21 +n116-041-188:1377256:1377311 [4] NCCL INFO Ring 07 : 19 -> 20 -> 21 +n116-041-188:1377253:1377310 [1] NCCL INFO Ring 03 : 16 -> 17 -> 18 +n116-041-188:1377259:1377313 [7] NCCL INFO Ring 11 : 22 -> 23 -> 16 +n116-041-188:1377255:1377309 [3] NCCL INFO Ring 00 : 20 -> 19 -> 18 +n116-041-188:1377257:1377308 [5] NCCL INFO Ring 09 : 20 -> 21 -> 22 +n116-041-188:1377258:1377314 [6] NCCL INFO Ring 11 : 21 -> 22 -> 23 +n116-041-188:1377256:1377311 [4] NCCL INFO Ring 08 : 21 -> 20 -> 19 +n116-041-188:1377257:1377308 [5] NCCL INFO Ring 10 : 22 -> 21 -> 20 +n116-041-188:1377258:1377314 [6] NCCL INFO Ring 12 : 23 -> 22 -> 21 +n116-041-188:1377253:1377310 [1] NCCL INFO Ring 04 : 18 -> 17 -> 16 +n116-041-188:1377252:1377312 [0] NCCL INFO Ring 00 : 9 -> 16 -> 23 +n116-041-188:1377258:1377314 [6] NCCL INFO Ring 13 : 21 -> 22 -> 23 +n116-041-188:1377253:1377310 [1] NCCL INFO Ring 05 : 16 -> 17 -> 18 +n116-041-188:1377259:1377313 [7] NCCL INFO Ring 12 : 16 -> 23 -> 22 +n116-041-188:1377253:1377310 [1] NCCL INFO Ring 06 : 18 -> 17 -> 16 +n116-041-188:1377255:1377309 [3] NCCL INFO Ring 01 : 18 -> 19 -> 20 +n116-041-188:1377253:1377310 [1] NCCL INFO Ring 07 : 16 -> 17 -> 18 +n116-041-188:1377256:1377311 [4] NCCL INFO Ring 09 : 19 -> 20 -> 21 +n116-041-188:1377257:1377308 [5] NCCL INFO Ring 11 : 20 -> 21 -> 22 +n116-041-188:1377252:1377312 [0] NCCL INFO Ring 01 : 23 -> 16 -> 25 +n116-041-188:1377258:1377314 [6] NCCL INFO Ring 14 : 15 -> 22 -> 21 +n116-041-188:1377259:1377313 [7] NCCL INFO Ring 13 : 22 -> 23 -> 16 +n116-041-188:1377255:1377309 [3] NCCL INFO Ring 02 : 20 -> 19 -> 26 +n116-041-188:1377259:1377313 [7] NCCL INFO Ring 14 : 16 -> 23 -> 30 +n116-041-188:1377253:1377310 [1] NCCL INFO Ring 08 : 18 -> 17 -> 24 +n116-041-188:1377256:1377311 [4] NCCL INFO Ring 10 : 21 -> 20 -> 19 +n116-041-188:1377257:1377308 [5] NCCL INFO Ring 12 : 22 -> 21 -> 28 +n116-041-188:1377252:1377312 [0] NCCL INFO Ring 02 : 17 -> 16 -> 23 +n116-041-188:1377258:1377314 [6] NCCL INFO Ring 15 : 21 -> 22 -> 31 +n116-041-188:1377257:1377308 [5] NCCL INFO Ring 13 : 12 -> 21 -> 22 +n116-041-188:1377259:1377313 [7] NCCL INFO Ring 15 : 14 -> 23 -> 16 +n116-041-188:1377252:1377312 [0] NCCL INFO Ring 03 : 23 -> 16 -> 17 +n116-041-188:1377257:1377308 [5] NCCL INFO Ring 14 : 22 -> 21 -> 20 +n116-041-188:1377255:1377309 [3] NCCL INFO Ring 03 : 10 -> 19 -> 20 +n116-041-188:1377256:1377311 [4] NCCL INFO Ring 11 : 19 -> 20 -> 21 +n116-041-188:1377253:1377310 [1] NCCL INFO Ring 09 : 8 -> 17 -> 18 +n116-041-188:1377254:1377315 [2] NCCL INFO Ring 00 : 19 -> 18 -> 17 +n116-041-188:1377258:1377314 [6] NCCL INFO Trees [0] 23/-1/-1->22->21 [1] 21/-1/-1->22->23 [2] 23/-1/-1->22->21 [3] 21/-1/-1->22->23 [4] 23/-1/-1->22->21 [5] -1/-1/-1->22->23 [6] 23/30/-1->22->39 [7] 21/15/-1->22->23 [8] 23/-1/-1->22->21 [9] 21/-1/-1->22->23 [10] 23/-1/-1->22->21 [11] 21/-1/-1->22->23 [12] 23/-1/-1->22->21 [13] -1/-1/-1->22->23 [14] 23/-1/-1->22->15 [15] 21/-1/-1->22->23 +n116-041-188:1377259:1377313 [7] NCCL INFO Trees [0] -1/-1/-1->23->22 [1] 22/-1/-1->23->16 [2] 16/-1/-1->23->22 [3] 22/-1/-1->23->16 [4] 16/-1/-1->23->22 [5] 22/-1/-1->23->16 [6] 16/14/-1->23->22 [7] 22/31/-1->23->38 [8] -1/-1/-1->23->22 [9] 22/-1/-1->23->16 [10] 16/-1/-1->23->22 [11] 22/-1/-1->23->16 [12] 16/-1/-1->23->22 [13] 22/-1/-1->23->16 [14] 16/-1/-1->23->22 [15] 22/-1/-1->23->14 +n116-041-188:1377252:1377312 [0] NCCL INFO Ring 04 : 17 -> 16 -> 23 +n116-041-188:1377257:1377308 [5] NCCL INFO Ring 15 : 20 -> 21 -> 22 +n116-041-188:1377255:1377309 [3] NCCL INFO Ring 04 : 20 -> 19 -> 18 +n116-041-188:1377257:1377308 [5] NCCL INFO Trees [0] 22/-1/-1->21->20 [1] 20/-1/-1->21->22 [2] 22/-1/-1->21->20 [3] 20/-1/-1->21->22 [4] 22/12/-1->21->20 [5] 20/29/-1->21->36 [6] -1/-1/-1->21->20 [7] 20/-1/-1->21->22 [8] 22/-1/-1->21->20 [9] 20/-1/-1->21->22 [10] 22/-1/-1->21->20 [11] 20/-1/-1->21->22 [12] 22/-1/-1->21->20 [13] 20/-1/-1->21->12 [14] -1/-1/-1->21->20 [15] 20/-1/-1->21->22 +n116-041-188:1377256:1377311 [4] NCCL INFO Ring 12 : 13 -> 20 -> 19 +n116-041-188:1377253:1377310 [1] NCCL INFO Ring 10 : 18 -> 17 -> 16 +n116-041-188:1377254:1377315 [2] NCCL INFO Ring 01 : 17 -> 18 -> 19 +n116-041-188:1377252:1377312 [0] NCCL INFO Ring 05 : 23 -> 16 -> 17 +n116-041-188:1377258:1377314 [6] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-041-188:1377259:1377313 [7] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-041-188:1377255:1377309 [3] NCCL INFO Ring 05 : 18 -> 19 -> 20 +n116-041-188:1377257:1377308 [5] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-041-188:1377259:1377313 [7] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-041-188:1377256:1377311 [4] NCCL INFO Ring 13 : 19 -> 20 -> 29 +n116-041-188:1377253:1377310 [1] NCCL INFO Ring 11 : 16 -> 17 -> 18 +n116-041-188:1377254:1377315 [2] NCCL INFO Ring 02 : 11 -> 18 -> 17 +n116-041-188:1377252:1377312 [0] NCCL INFO Ring 06 : 17 -> 16 -> 23 +n116-041-188:1377258:1377314 [6] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-041-188:1377257:1377308 [5] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-041-188:1377255:1377309 [3] NCCL INFO Ring 06 : 20 -> 19 -> 18 +n116-041-188:1377259:1377313 [7] NCCL INFO P2P Chunksize set to 524288 +n116-041-188:1377256:1377311 [4] NCCL INFO Ring 14 : 21 -> 20 -> 19 +n116-041-188:1377253:1377310 [1] NCCL INFO Ring 12 : 18 -> 17 -> 16 +n116-041-188:1377252:1377312 [0] NCCL INFO Ring 07 : 23 -> 16 -> 17 +n116-041-188:1377254:1377315 [2] NCCL INFO Ring 03 : 17 -> 18 -> 27 +n116-041-188:1377258:1377314 [6] NCCL INFO P2P Chunksize set to 524288 +n116-041-188:1377254:1377315 [2] NCCL INFO Ring 04 : 19 -> 18 -> 17 +n116-041-188:1377257:1377308 [5] NCCL INFO P2P Chunksize set to 524288 +n116-041-188:1377255:1377309 [3] NCCL INFO Ring 07 : 18 -> 19 -> 20 +n116-041-188:1377256:1377311 [4] NCCL INFO Ring 15 : 19 -> 20 -> 21 +n116-041-188:1377253:1377310 [1] NCCL INFO Ring 13 : 16 -> 17 -> 18 +n116-041-188:1377252:1377312 [0] NCCL INFO Ring 08 : 9 -> 16 -> 23 +n116-041-188:1377256:1377311 [4] NCCL INFO Trees [0] 21/-1/-1->20->19 [1] 19/-1/-1->20->21 [2] 21/-1/-1->20->19 [3] -1/-1/-1->20->21 [4] 21/28/-1->20->37 [5] 19/13/-1->20->21 [6] 21/-1/-1->20->19 [7] 19/-1/-1->20->21 [8] 21/-1/-1->20->19 [9] 19/-1/-1->20->21 [10] 21/-1/-1->20->19 [11] -1/-1/-1->20->21 [12] 21/-1/-1->20->13 [13] 19/-1/-1->20->21 [14] 21/-1/-1->20->19 [15] 19/-1/-1->20->21 +n116-041-188:1377254:1377315 [2] NCCL INFO Ring 05 : 17 -> 18 -> 19 +n116-041-188:1377256:1377311 [4] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-041-188:1377255:1377309 [3] NCCL INFO Ring 08 : 20 -> 19 -> 18 +n116-041-188:1377253:1377310 [1] NCCL INFO Ring 14 : 18 -> 17 -> 16 +n116-041-188:1377256:1377311 [4] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-041-188:1377252:1377312 [0] NCCL INFO Ring 09 : 23 -> 16 -> 25 +n116-041-188:1377253:1377310 [1] NCCL INFO Ring 15 : 16 -> 17 -> 18 +n116-041-188:1377252:1377312 [0] NCCL INFO Ring 10 : 17 -> 16 -> 23 +n116-041-188:1377254:1377315 [2] NCCL INFO Ring 06 : 19 -> 18 -> 17 +n116-041-188:1377255:1377309 [3] NCCL INFO Ring 09 : 18 -> 19 -> 20 +n116-041-188:1377252:1377312 [0] NCCL INFO Ring 11 : 23 -> 16 -> 17 +n116-041-188:1377256:1377311 [4] NCCL INFO P2P Chunksize set to 524288 +n116-041-188:1377253:1377310 [1] NCCL INFO Trees [0] 18/8/-1->17->16 [1] 16/25/-1->17->32 [2] -1/-1/-1->17->16 [3] 16/-1/-1->17->18 [4] 18/-1/-1->17->16 [5] 16/-1/-1->17->18 [6] 18/-1/-1->17->16 [7] 16/-1/-1->17->18 [8] 18/-1/-1->17->16 [9] 16/-1/-1->17->8 [10] -1/-1/-1->17->16 [11] 16/-1/-1->17->18 [12] 18/-1/-1->17->16 [13] 16/-1/-1->17->18 [14] 18/-1/-1->17->16 [15] 16/-1/-1->17->18 +n116-041-188:1377254:1377315 [2] NCCL INFO Ring 07 : 17 -> 18 -> 19 +n116-041-188:1377252:1377312 [0] NCCL INFO Ring 12 : 17 -> 16 -> 23 +n116-041-188:1377255:1377309 [3] NCCL INFO Ring 10 : 20 -> 19 -> 26 +n116-041-188:1377255:1377309 [3] NCCL INFO Ring 11 : 10 -> 19 -> 20 +n116-041-188:1377254:1377315 [2] NCCL INFO Ring 08 : 19 -> 18 -> 17 +n116-041-188:1377252:1377312 [0] NCCL INFO Ring 13 : 23 -> 16 -> 17 +n116-041-188:1377253:1377310 [1] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-041-188:1377254:1377315 [2] NCCL INFO Ring 09 : 17 -> 18 -> 19 +n116-041-188:1377255:1377309 [3] NCCL INFO Ring 12 : 20 -> 19 -> 18 +n116-041-188:1377252:1377312 [0] NCCL INFO Ring 14 : 17 -> 16 -> 23 +n116-041-188:1377253:1377310 [1] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-041-188:1377254:1377315 [2] NCCL INFO Ring 10 : 11 -> 18 -> 17 +n116-041-188:1377253:1377310 [1] NCCL INFO P2P Chunksize set to 524288 +n116-041-188:1377255:1377309 [3] NCCL INFO Ring 13 : 18 -> 19 -> 20 +n116-041-188:1377252:1377312 [0] NCCL INFO Ring 15 : 23 -> 16 -> 17 +n116-041-188:1377254:1377315 [2] NCCL INFO Ring 11 : 17 -> 18 -> 27 +n116-041-188:1377254:1377315 [2] NCCL INFO Ring 12 : 19 -> 18 -> 17 +n116-041-188:1377255:1377309 [3] NCCL INFO Ring 14 : 20 -> 19 -> 18 +n116-041-188:1377252:1377312 [0] NCCL INFO Trees [0] 17/24/-1->16->33 [1] 23/9/-1->16->17 [2] 17/-1/-1->16->23 [3] 23/-1/-1->16->17 [4] 17/-1/-1->16->23 [5] 23/-1/-1->16->17 [6] 17/-1/-1->16->23 [7] -1/-1/-1->16->17 [8] 17/-1/-1->16->9 [9] 23/-1/-1->16->17 [10] 17/-1/-1->16->23 [11] 23/-1/-1->16->17 [12] 17/-1/-1->16->23 [13] 23/-1/-1->16->17 [14] 17/-1/-1->16->23 [15] -1/-1/-1->16->17 +n116-041-188:1377255:1377309 [3] NCCL INFO Ring 15 : 18 -> 19 -> 20 +n116-041-188:1377254:1377315 [2] NCCL INFO Ring 13 : 17 -> 18 -> 19 +n116-041-188:1377252:1377312 [0] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-041-188:1377255:1377309 [3] NCCL INFO Trees [0] 20/-1/-1->19->18 [1] 18/-1/-1->19->20 [2] 20/10/-1->19->18 [3] 18/27/-1->19->34 [4] -1/-1/-1->19->18 [5] 18/-1/-1->19->20 [6] 20/-1/-1->19->18 [7] 18/-1/-1->19->20 [8] 20/-1/-1->19->18 [9] 18/-1/-1->19->20 [10] 20/-1/-1->19->18 [11] 18/-1/-1->19->10 [12] -1/-1/-1->19->18 [13] 18/-1/-1->19->20 [14] 20/-1/-1->19->18 [15] 18/-1/-1->19->20 +n116-041-188:1377254:1377315 [2] NCCL INFO Ring 14 : 19 -> 18 -> 17 +n116-041-188:1377252:1377312 [0] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-041-188:1377254:1377315 [2] NCCL INFO Ring 15 : 17 -> 18 -> 19 +n116-041-188:1377255:1377309 [3] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-041-188:1377255:1377309 [3] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-041-188:1377252:1377312 [0] NCCL INFO P2P Chunksize set to 524288 +n116-041-188:1377254:1377315 [2] NCCL INFO Trees [0] 19/-1/-1->18->17 [1] -1/-1/-1->18->19 [2] 19/26/-1->18->35 [3] 17/11/-1->18->19 [4] 19/-1/-1->18->17 [5] 17/-1/-1->18->19 [6] 19/-1/-1->18->17 [7] 17/-1/-1->18->19 [8] 19/-1/-1->18->17 [9] -1/-1/-1->18->19 [10] 19/-1/-1->18->11 [11] 17/-1/-1->18->19 [12] 19/-1/-1->18->17 [13] 17/-1/-1->18->19 [14] 19/-1/-1->18->17 [15] 17/-1/-1->18->19 +n116-041-188:1377255:1377309 [3] NCCL INFO P2P Chunksize set to 524288 +n116-041-188:1377254:1377315 [2] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-041-188:1377254:1377315 [2] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-041-188:1377254:1377315 [2] NCCL INFO P2P Chunksize set to 524288 +n116-041-188:1377256:1377350 [4] NCCL INFO [Proxy Service] Device 4 CPU core 160 +n116-041-188:1377259:1377352 [7] NCCL INFO [Proxy Service UDS] Device 7 CPU core 78 +n116-041-188:1377259:1377348 [7] NCCL INFO [Proxy Service] Device 7 CPU core 159 +n116-041-188:1377258:1377349 [6] NCCL INFO [Proxy Service] Device 6 CPU core 167 +n116-041-188:1377253:1377353 [1] NCCL INFO [Proxy Service] Device 1 CPU core 108 +n116-041-188:1377258:1377355 [6] NCCL INFO [Proxy Service UDS] Device 6 CPU core 75 +n116-041-188:1377257:1377351 [5] NCCL INFO [Proxy Service] Device 5 CPU core 181 +n116-041-188:1377256:1377354 [4] NCCL INFO [Proxy Service UDS] Device 4 CPU core 183 +n116-041-188:1377254:1377360 [2] NCCL INFO [Proxy Service] Device 2 CPU core 111 +n116-041-188:1377255:1377361 [3] NCCL INFO [Proxy Service] Device 3 CPU core 7 +n116-041-188:1377257:1377356 [5] NCCL INFO [Proxy Service UDS] Device 5 CPU core 86 +n116-041-188:1377252:1377357 [0] NCCL INFO [Proxy Service] Device 0 CPU core 35 +n116-041-188:1377253:1377358 [1] NCCL INFO [Proxy Service UDS] Device 1 CPU core 110 +n116-041-188:1377252:1377359 [0] NCCL INFO [Proxy Service UDS] Device 0 CPU core 38 +n116-041-188:1377254:1377362 [2] NCCL INFO [Proxy Service UDS] Device 2 CPU core 126 +n116-041-188:1377255:1377363 [3] NCCL INFO [Proxy Service UDS] Device 3 CPU core 22 +n116-041-188:1377256:1377311 [4] NCCL INFO NCCL_PROTO set by environment to Simple +n116-041-188:1377256:1377311 [4] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-041-188:1377256:1377311 [4] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n116-041-188:1377256:1377311 [4] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n116-041-188:1377259:1377313 [7] NCCL INFO NCCL_PROTO set by environment to Simple +n116-041-188:1377259:1377313 [7] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-041-188:1377259:1377313 [7] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n116-041-188:1377259:1377313 [7] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n116-041-188:1377258:1377314 [6] NCCL INFO NCCL_PROTO set by environment to Simple +n116-041-188:1377258:1377314 [6] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-041-188:1377258:1377314 [6] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n116-041-188:1377258:1377314 [6] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n116-041-188:1377252:1377312 [0] NCCL INFO NCCL_PROTO set by environment to Simple +n116-041-188:1377252:1377312 [0] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-041-188:1377252:1377312 [0] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n116-041-188:1377252:1377312 [0] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n116-041-188:1377257:1377308 [5] NCCL INFO NCCL_PROTO set by environment to Simple +n116-041-188:1377257:1377308 [5] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-041-188:1377257:1377308 [5] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n116-041-188:1377257:1377308 [5] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n116-041-188:1377254:1377315 [2] NCCL INFO NCCL_PROTO set by environment to Simple +n116-041-188:1377254:1377315 [2] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-041-188:1377254:1377315 [2] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n116-041-188:1377254:1377315 [2] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n116-041-188:1377253:1377310 [1] NCCL INFO NCCL_PROTO set by environment to Simple +n116-041-188:1377253:1377310 [1] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-041-188:1377253:1377310 [1] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n116-041-188:1377253:1377310 [1] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n116-041-188:1377255:1377309 [3] NCCL INFO NCCL_PROTO set by environment to Simple +n116-041-188:1377255:1377309 [3] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-041-188:1377255:1377309 [3] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n116-041-188:1377255:1377309 [3] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n116-041-188:1377252:1377312 [0] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-041-188:1377252:1377312 [0] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-041-188:1377252:1377312 [0] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-041-188:1377252:1377312 [0] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-041-188:1377252:1377312 [0] NCCL INFO ncclCommInitRankConfig comm 0xcaf2670 rank 16 nranks 64 cudaDev 0 nvmlDev 0 busId 4000 commId 0xfb2d11625d05007a - Init COMPLETE +n116-041-188:1377252:1377312 [0] NCCL INFO Init timings - ncclCommInitRankConfig: rank 16 nranks 64 total 5.01 (kernels 0.30, alloc 1.59, bootstrap 2.24, allgathers 0.07, topo 0.70, graphs 0.08, connections 0.03, rest 0.01) +n116-041-188:1377256:1377311 [4] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-041-188:1377258:1377314 [6] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-041-188:1377256:1377311 [4] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-041-188:1377258:1377314 [6] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-041-188:1377256:1377311 [4] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-041-188:1377258:1377314 [6] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-041-188:1377256:1377311 [4] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-041-188:1377258:1377314 [6] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-041-188:1377257:1377308 [5] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-041-188:1377256:1377311 [4] NCCL INFO ncclCommInitRankConfig comm 0xb9355b0 rank 20 nranks 64 cudaDev 4 nvmlDev 4 busId 84000 commId 0xfb2d11625d05007a - Init COMPLETE +n116-041-188:1377258:1377314 [6] NCCL INFO ncclCommInitRankConfig comm 0xc3c2c30 rank 22 nranks 64 cudaDev 6 nvmlDev 6 busId 8b000 commId 0xfb2d11625d05007a - Init COMPLETE +n116-041-188:1377257:1377308 [5] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-041-188:1377257:1377308 [5] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-041-188:1377257:1377308 [5] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-041-188:1377258:1377314 [6] NCCL INFO Init timings - ncclCommInitRankConfig: rank 22 nranks 64 total 5.01 (kernels 0.31, alloc 1.57, bootstrap 2.24, allgathers 0.07, topo 0.69, graphs 0.08, connections 0.03, rest 0.01) +n116-041-188:1377256:1377311 [4] NCCL INFO Init timings - ncclCommInitRankConfig: rank 20 nranks 64 total 5.02 (kernels 0.30, alloc 1.60, bootstrap 2.24, allgathers 0.07, topo 0.69, graphs 0.08, connections 0.03, rest 0.01) +n116-041-188:1377257:1377308 [5] NCCL INFO ncclCommInitRankConfig comm 0xb18a360 rank 21 nranks 64 cudaDev 5 nvmlDev 5 busId 85000 commId 0xfb2d11625d05007a - Init COMPLETE +n116-041-188:1377257:1377308 [5] NCCL INFO Init timings - ncclCommInitRankConfig: rank 21 nranks 64 total 5.05 (kernels 0.30, alloc 1.62, bootstrap 2.24, allgathers 0.07, topo 0.69, graphs 0.08, connections 0.03, rest 0.00) +n116-041-188:1377259:1377313 [7] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-041-188:1377259:1377313 [7] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-041-188:1377259:1377313 [7] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-041-188:1377259:1377313 [7] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-041-188:1377259:1377313 [7] NCCL INFO ncclCommInitRankConfig comm 0xc17e960 rank 23 nranks 64 cudaDev 7 nvmlDev 7 busId 8c000 commId 0xfb2d11625d05007a - Init COMPLETE +n116-041-188:1377259:1377313 [7] NCCL INFO Init timings - ncclCommInitRankConfig: rank 23 nranks 64 total 5.01 (kernels 0.31, alloc 1.57, bootstrap 2.24, allgathers 0.07, topo 0.69, graphs 0.08, connections 0.03, rest 0.01) +n116-041-188:1377253:1377310 [1] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-041-188:1377253:1377310 [1] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-041-188:1377253:1377310 [1] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-041-188:1377253:1377310 [1] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-041-188:1377253:1377310 [1] NCCL INFO ncclCommInitRankConfig comm 0xcbee300 rank 17 nranks 64 cudaDev 1 nvmlDev 1 busId 5000 commId 0xfb2d11625d05007a - Init COMPLETE +n116-041-188:1377253:1377310 [1] NCCL INFO Init timings - ncclCommInitRankConfig: rank 17 nranks 64 total 5.03 (kernels 0.30, alloc 1.60, bootstrap 2.25, allgathers 0.07, topo 0.70, graphs 0.08, connections 0.04, rest 0.00) +n116-041-188:1377254:1377315 [2] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-041-188:1377254:1377315 [2] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-041-188:1377254:1377315 [2] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-041-188:1377254:1377315 [2] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-041-188:1377255:1377309 [3] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-041-188:1377254:1377315 [2] NCCL INFO ncclCommInitRankConfig comm 0xb3d7aa0 rank 18 nranks 64 cudaDev 2 nvmlDev 2 busId b000 commId 0xfb2d11625d05007a - Init COMPLETE +n116-041-188:1377255:1377309 [3] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-041-188:1377255:1377309 [3] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-041-188:1377254:1377315 [2] NCCL INFO Init timings - ncclCommInitRankConfig: rank 18 nranks 64 total 5.00 (kernels 0.31, alloc 1.58, bootstrap 2.24, allgathers 0.01, topo 0.69, graphs 0.14, connections 0.03, rest 0.00) +n116-041-188:1377255:1377309 [3] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-041-188:1377255:1377309 [3] NCCL INFO ncclCommInitRankConfig comm 0xcab1d40 rank 19 nranks 64 cudaDev 3 nvmlDev 3 busId c000 commId 0xfb2d11625d05007a - Init COMPLETE +n116-041-188:1377255:1377309 [3] NCCL INFO Init timings - ncclCommInitRankConfig: rank 19 nranks 64 total 5.05 (kernels 0.30, alloc 1.61, bootstrap 2.25, allgathers 0.01, topo 0.69, graphs 0.13, connections 0.04, rest 0.00) +n116-041-188:1377252:1377364 [0] NCCL INFO Channel 00/0 : 16[0] -> 17[1] via P2P/CUMEM +n116-041-188:1377252:1377364 [0] NCCL INFO Channel 01/0 : 16[0] -> 17[1] via P2P/CUMEM +n116-041-188:1377252:1377364 [0] NCCL INFO Channel 02/0 : 16[0] -> 17[1] via P2P/CUMEM +n116-041-188:1377252:1377364 [0] NCCL INFO Channel 03/0 : 16[0] -> 17[1] via P2P/CUMEM +n116-041-188:1377252:1377364 [0] NCCL INFO Channel 04/0 : 16[0] -> 17[1] via P2P/CUMEM +n116-041-188:1377252:1377364 [0] NCCL INFO Channel 05/0 : 16[0] -> 17[1] via P2P/CUMEM +n116-041-188:1377252:1377364 [0] NCCL INFO Channel 06/0 : 16[0] -> 17[1] via P2P/CUMEM +n116-041-188:1377252:1377364 [0] NCCL INFO Channel 07/0 : 16[0] -> 17[1] via P2P/CUMEM +n116-041-188:1377252:1377364 [0] NCCL INFO Channel 08/0 : 16[0] -> 17[1] via P2P/CUMEM +n116-041-188:1377252:1377364 [0] NCCL INFO Channel 09/0 : 16[0] -> 17[1] via P2P/CUMEM +n116-041-188:1377252:1377364 [0] NCCL INFO Channel 10/0 : 16[0] -> 17[1] via P2P/CUMEM +n116-041-188:1377252:1377364 [0] NCCL INFO Channel 11/0 : 16[0] -> 17[1] via P2P/CUMEM +n116-041-188:1377252:1377364 [0] NCCL INFO Channel 12/0 : 16[0] -> 17[1] via P2P/CUMEM +n116-041-188:1377252:1377364 [0] NCCL INFO Channel 13/0 : 16[0] -> 17[1] via P2P/CUMEM +n116-041-188:1377252:1377364 [0] NCCL INFO Channel 14/0 : 16[0] -> 17[1] via P2P/CUMEM +n116-041-188:1377256:1377366 [4] NCCL INFO Channel 00/0 : 20[4] -> 21[5] via P2P/CUMEM +n116-041-188:1377254:1377368 [2] NCCL INFO Channel 00/0 : 18[2] -> 19[3] via P2P/CUMEM +n116-041-188:1377252:1377364 [0] NCCL INFO Channel 15/0 : 16[0] -> 17[1] via P2P/CUMEM +n116-041-188:1377256:1377366 [4] NCCL INFO Channel 01/0 : 20[4] -> 21[5] via P2P/CUMEM +n116-041-188:1377254:1377368 [2] NCCL INFO Channel 01/0 : 18[2] -> 19[3] via P2P/CUMEM +n116-041-188:1377258:1377367 [6] NCCL INFO Channel 00/0 : 22[6] -> 23[7] via P2P/CUMEM +n116-041-188:1377256:1377366 [4] NCCL INFO Channel 02/0 : 20[4] -> 21[5] via P2P/CUMEM +n116-041-188:1377254:1377368 [2] NCCL INFO Channel 02/0 : 18[2] -> 19[3] via P2P/CUMEM +n116-041-188:1377258:1377367 [6] NCCL INFO Channel 01/0 : 22[6] -> 23[7] via P2P/CUMEM +n116-041-188:1377256:1377366 [4] NCCL INFO Channel 03/0 : 20[4] -> 21[5] via P2P/CUMEM +n116-041-188:1377257:1377365 [5] NCCL INFO Channel 00/0 : 21[5] -> 22[6] via P2P/CUMEM +n116-041-188:1377254:1377368 [2] NCCL INFO Channel 03/0 : 18[2] -> 19[3] via P2P/CUMEM +n116-041-188:1377258:1377367 [6] NCCL INFO Channel 02/0 : 22[6] -> 23[7] via P2P/CUMEM +n116-041-188:1377256:1377366 [4] NCCL INFO Channel 04/0 : 20[4] -> 21[5] via P2P/CUMEM +n116-041-188:1377253:1377370 [1] NCCL INFO Channel 00/0 : 17[1] -> 18[2] via P2P/CUMEM +n116-041-188:1377257:1377365 [5] NCCL INFO Channel 01/0 : 21[5] -> 22[6] via P2P/CUMEM +n116-041-188:1377254:1377368 [2] NCCL INFO Channel 04/0 : 18[2] -> 19[3] via P2P/CUMEM +n116-041-188:1377258:1377367 [6] NCCL INFO Channel 03/0 : 22[6] -> 23[7] via P2P/CUMEM +n116-041-188:1377256:1377366 [4] NCCL INFO Channel 05/0 : 20[4] -> 21[5] via P2P/CUMEM +n116-041-188:1377253:1377370 [1] NCCL INFO Channel 03/0 : 17[1] -> 18[2] via P2P/CUMEM +n116-041-188:1377255:1377371 [3] NCCL INFO Channel 00/0 : 19[3] -> 20[4] via P2P/CUMEM +n116-041-188:1377254:1377368 [2] NCCL INFO Channel 05/0 : 18[2] -> 19[3] via P2P/CUMEM +n116-041-188:1377257:1377365 [5] NCCL INFO Channel 02/0 : 21[5] -> 22[6] via P2P/CUMEM +n116-041-188:1377258:1377367 [6] NCCL INFO Channel 04/0 : 22[6] -> 23[7] via P2P/CUMEM +n116-041-188:1377256:1377366 [4] NCCL INFO Channel 06/0 : 20[4] -> 21[5] via P2P/CUMEM +n116-041-188:1377253:1377370 [1] NCCL INFO Channel 04/0 : 17[1] -> 18[2] via P2P/CUMEM +n116-041-188:1377255:1377371 [3] NCCL INFO Channel 01/0 : 19[3] -> 20[4] via P2P/CUMEM +n116-041-188:1377254:1377368 [2] NCCL INFO Channel 06/0 : 18[2] -> 19[3] via P2P/CUMEM +n116-041-188:1377257:1377365 [5] NCCL INFO Channel 03/0 : 21[5] -> 22[6] via P2P/CUMEM +n116-041-188:1377258:1377367 [6] NCCL INFO Channel 05/0 : 22[6] -> 23[7] via P2P/CUMEM +n116-041-188:1377256:1377366 [4] NCCL INFO Channel 07/0 : 20[4] -> 21[5] via P2P/CUMEM +n116-041-188:1377253:1377370 [1] NCCL INFO Channel 05/0 : 17[1] -> 18[2] via P2P/CUMEM +n116-041-188:1377255:1377371 [3] NCCL INFO Channel 02/0 : 19[3] -> 20[4] via P2P/CUMEM +n116-041-188:1377254:1377368 [2] NCCL INFO Channel 07/0 : 18[2] -> 19[3] via P2P/CUMEM +n116-041-188:1377257:1377365 [5] NCCL INFO Channel 04/0 : 21[5] -> 22[6] via P2P/CUMEM +n116-041-188:1377258:1377367 [6] NCCL INFO Channel 06/0 : 22[6] -> 23[7] via P2P/CUMEM +n116-041-188:1377256:1377366 [4] NCCL INFO Channel 08/0 : 20[4] -> 21[5] via P2P/CUMEM +n116-041-188:1377253:1377370 [1] NCCL INFO Channel 06/0 : 17[1] -> 18[2] via P2P/CUMEM +n116-041-188:1377255:1377371 [3] NCCL INFO Channel 05/0 : 19[3] -> 20[4] via P2P/CUMEM +n116-041-188:1377254:1377368 [2] NCCL INFO Channel 08/0 : 18[2] -> 19[3] via P2P/CUMEM +n116-041-188:1377257:1377365 [5] NCCL INFO Channel 07/0 : 21[5] -> 22[6] via P2P/CUMEM +n116-041-188:1377258:1377367 [6] NCCL INFO Channel 07/0 : 22[6] -> 23[7] via P2P/CUMEM +n116-041-188:1377256:1377366 [4] NCCL INFO Channel 09/0 : 20[4] -> 21[5] via P2P/CUMEM +n116-041-188:1377253:1377370 [1] NCCL INFO Channel 07/0 : 17[1] -> 18[2] via P2P/CUMEM +n116-041-188:1377255:1377371 [3] NCCL INFO Channel 06/0 : 19[3] -> 20[4] via P2P/CUMEM +n116-041-188:1377254:1377368 [2] NCCL INFO Channel 09/0 : 18[2] -> 19[3] via P2P/CUMEM +n116-041-188:1377257:1377365 [5] NCCL INFO Channel 08/0 : 21[5] -> 22[6] via P2P/CUMEM +n116-041-188:1377258:1377367 [6] NCCL INFO Channel 08/0 : 22[6] -> 23[7] via P2P/CUMEM +n116-041-188:1377256:1377366 [4] NCCL INFO Channel 10/0 : 20[4] -> 21[5] via P2P/CUMEM +n116-041-188:1377253:1377370 [1] NCCL INFO Channel 08/0 : 17[1] -> 18[2] via P2P/CUMEM +n116-041-188:1377255:1377371 [3] NCCL INFO Channel 07/0 : 19[3] -> 20[4] via P2P/CUMEM +n116-041-188:1377254:1377368 [2] NCCL INFO Channel 10/0 : 18[2] -> 19[3] via P2P/CUMEM +n116-041-188:1377257:1377365 [5] NCCL INFO Channel 09/0 : 21[5] -> 22[6] via P2P/CUMEM +n116-041-188:1377258:1377367 [6] NCCL INFO Channel 09/0 : 22[6] -> 23[7] via P2P/CUMEM +n116-041-188:1377256:1377366 [4] NCCL INFO Channel 11/0 : 20[4] -> 21[5] via P2P/CUMEM +n116-041-188:1377253:1377370 [1] NCCL INFO Channel 11/0 : 17[1] -> 18[2] via P2P/CUMEM +n116-041-188:1377255:1377371 [3] NCCL INFO Channel 08/0 : 19[3] -> 20[4] via P2P/CUMEM +n116-041-188:1377254:1377368 [2] NCCL INFO Channel 11/0 : 18[2] -> 19[3] via P2P/CUMEM +n116-041-188:1377257:1377365 [5] NCCL INFO Channel 10/0 : 21[5] -> 22[6] via P2P/CUMEM +n116-041-188:1377258:1377367 [6] NCCL INFO Channel 10/0 : 22[6] -> 23[7] via P2P/CUMEM +n116-041-188:1377256:1377366 [4] NCCL INFO Channel 12/0 : 20[4] -> 21[5] via P2P/CUMEM +n116-041-188:1377253:1377370 [1] NCCL INFO Channel 12/0 : 17[1] -> 18[2] via P2P/CUMEM +n116-041-188:1377255:1377371 [3] NCCL INFO Channel 09/0 : 19[3] -> 20[4] via P2P/CUMEM +n116-041-188:1377254:1377368 [2] NCCL INFO Channel 12/0 : 18[2] -> 19[3] via P2P/CUMEM +n116-041-188:1377258:1377367 [6] NCCL INFO Channel 11/0 : 22[6] -> 23[7] via P2P/CUMEM +n116-041-188:1377257:1377365 [5] NCCL INFO Channel 11/0 : 21[5] -> 22[6] via P2P/CUMEM +n116-041-188:1377256:1377366 [4] NCCL INFO Channel 13/0 : 20[4] -> 21[5] via P2P/CUMEM +n116-041-188:1377253:1377370 [1] NCCL INFO Channel 13/0 : 17[1] -> 18[2] via P2P/CUMEM +n116-041-188:1377255:1377371 [3] NCCL INFO Channel 10/0 : 19[3] -> 20[4] via P2P/CUMEM +n116-041-188:1377254:1377368 [2] NCCL INFO Channel 13/0 : 18[2] -> 19[3] via P2P/CUMEM +n116-041-188:1377258:1377367 [6] NCCL INFO Channel 12/0 : 22[6] -> 23[7] via P2P/CUMEM +n116-041-188:1377257:1377365 [5] NCCL INFO Channel 12/0 : 21[5] -> 22[6] via P2P/CUMEM +n116-041-188:1377256:1377366 [4] NCCL INFO Channel 14/0 : 20[4] -> 21[5] via P2P/CUMEM +n116-041-188:1377253:1377370 [1] NCCL INFO Channel 14/0 : 17[1] -> 18[2] via P2P/CUMEM +n116-041-188:1377255:1377371 [3] NCCL INFO Channel 13/0 : 19[3] -> 20[4] via P2P/CUMEM +n116-041-188:1377254:1377368 [2] NCCL INFO Channel 14/0 : 18[2] -> 19[3] via P2P/CUMEM +n116-041-188:1377258:1377367 [6] NCCL INFO Channel 13/0 : 22[6] -> 23[7] via P2P/CUMEM +n116-041-188:1377257:1377365 [5] NCCL INFO Channel 15/0 : 21[5] -> 22[6] via P2P/CUMEM +n116-041-188:1377256:1377366 [4] NCCL INFO Channel 15/0 : 20[4] -> 21[5] via P2P/CUMEM +n116-041-188:1377253:1377370 [1] NCCL INFO Channel 15/0 : 17[1] -> 18[2] via P2P/CUMEM +n116-041-188:1377255:1377371 [3] NCCL INFO Channel 14/0 : 19[3] -> 20[4] via P2P/CUMEM +n116-041-188:1377254:1377368 [2] NCCL INFO Channel 15/0 : 18[2] -> 19[3] via P2P/CUMEM +n116-041-188:1377258:1377367 [6] NCCL INFO Channel 14/0 : 22[6] -> 23[7] via P2P/CUMEM +n116-041-188:1377255:1377371 [3] NCCL INFO Channel 15/0 : 19[3] -> 20[4] via P2P/CUMEM +n116-041-188:1377258:1377367 [6] NCCL INFO Channel 15/0 : 22[6] -> 23[7] via P2P/CUMEM +n116-041-188:1377252:1377372 [0] NCCL INFO [Proxy Progress] Device 0 CPU core 39 +n116-041-188:1377256:1377373 [4] NCCL INFO [Proxy Progress] Device 4 CPU core 196 +n116-041-188:1377258:1377374 [6] NCCL INFO [Proxy Progress] Device 6 CPU core 167 +n116-041-188:1377257:1377375 [5] NCCL INFO [Proxy Progress] Device 5 CPU core 84 +n116-041-188:1377252:1377364 [0] NCCL INFO Channel 01/0 : 9[1] -> 16[0] [receive] via NET/FasTrak/1/GDRDMA +n116-041-188:1377253:1377376 [1] NCCL INFO [Proxy Progress] Device 1 CPU core 14 +n116-041-188:1377257:1377365 [5] NCCL INFO Channel 05/0 : 21[5] -> 29[5] [send] via NET/FasTrak/5/GDRDMA +n116-041-188:1377253:1377370 [1] NCCL INFO Channel 01/0 : 17[1] -> 25[1] [send] via NET/FasTrak/1/GDRDMA +n116-041-188:1377256:1377366 [4] NCCL INFO Channel 05/0 : 13[5] -> 20[4] [receive] via NET/FasTrak/5/GDRDMA +n116-041-188:1377258:1377367 [6] NCCL INFO Channel 07/0 : 15[7] -> 22[6] [receive] via NET/FasTrak/7/GDRDMA +n116-041-188:1377253:1377370 [1] NCCL INFO Channel 00/0 : 8[0] -> 17[1] [receive] via NET/FasTrak/0/GDRDMA +n116-041-188:1377252:1377364 [0] NCCL INFO Channel 08/0 : 9[1] -> 16[0] [receive] via NET/FasTrak/0/GDRDMA +n116-041-188:1377252:1377364 [0] NCCL INFO Channel 01/0 : 16[0] -> 23[7] via P2P/CUMEM +n116-041-188:1377252:1377364 [0] NCCL INFO Channel 02/0 : 16[0] -> 23[7] via P2P/CUMEM +n116-041-188:1377252:1377364 [0] NCCL INFO Channel 03/0 : 16[0] -> 23[7] via P2P/CUMEM +n116-041-188:1377256:1377366 [4] NCCL INFO Channel 12/0 : 13[5] -> 20[4] [receive] via NET/FasTrak/4/GDRDMA +n116-041-188:1377252:1377364 [0] NCCL INFO Channel 04/0 : 16[0] -> 23[7] via P2P/CUMEM +n116-041-188:1377258:1377367 [6] NCCL INFO Channel 14/0 : 15[7] -> 22[6] [receive] via NET/FasTrak/6/GDRDMA +n116-041-188:1377252:1377364 [0] NCCL INFO Channel 05/0 : 16[0] -> 23[7] via P2P/CUMEM +n116-041-188:1377253:1377370 [1] NCCL INFO Channel 09/0 : 8[0] -> 17[1] [receive] via NET/FasTrak/1/GDRDMA +n116-041-188:1377252:1377364 [0] NCCL INFO Channel 06/0 : 16[0] -> 23[7] via P2P/CUMEM +n116-041-188:1377252:1377364 [0] NCCL INFO Channel 09/0 : 16[0] -> 23[7] via P2P/CUMEM +n116-041-188:1377252:1377364 [0] NCCL INFO Channel 10/0 : 16[0] -> 23[7] via P2P/CUMEM +n116-041-188:1377252:1377364 [0] NCCL INFO Channel 11/0 : 16[0] -> 23[7] via P2P/CUMEM +n116-041-188:1377252:1377364 [0] NCCL INFO Channel 12/0 : 16[0] -> 23[7] via P2P/CUMEM +n116-041-188:1377252:1377364 [0] NCCL INFO Channel 13/0 : 16[0] -> 23[7] via P2P/CUMEM +n116-041-188:1377252:1377364 [0] NCCL INFO Channel 14/0 : 16[0] -> 23[7] via P2P/CUMEM +n116-041-188:1377257:1377365 [5] NCCL INFO Channel 04/0 : 12[4] -> 21[5] [receive] via NET/FasTrak/4/GDRDMA +n116-041-188:1377254:1377377 [2] NCCL INFO [Proxy Progress] Device 2 CPU core 112 +n116-041-188:1377255:1377378 [3] NCCL INFO [Proxy Progress] Device 3 CPU core 43 +n116-041-188:1377255:1377371 [3] NCCL INFO Channel 03/0 : 19[3] -> 27[3] [send] via NET/FasTrak/3/GDRDMA +n116-041-188:1377257:1377365 [5] NCCL INFO Channel 13/0 : 12[4] -> 21[5] [receive] via NET/FasTrak/5/GDRDMA +n116-041-188:1377255:1377371 [3] NCCL INFO Channel 02/0 : 10[2] -> 19[3] [receive] via NET/FasTrak/2/GDRDMA +n116-041-188:1377254:1377368 [2] NCCL INFO Channel 03/0 : 11[3] -> 18[2] [receive] via NET/FasTrak/3/GDRDMA +n116-041-188:1377255:1377371 [3] NCCL INFO Channel 11/0 : 10[2] -> 19[3] [receive] via NET/FasTrak/3/GDRDMA +n116-041-188:1377254:1377368 [2] NCCL INFO Channel 10/0 : 11[3] -> 18[2] [receive] via NET/FasTrak/2/GDRDMA +n116-041-188:1377252:1377364 [0] NCCL INFO Channel 00/0 : 16[0] -> 24[0] [send] via NET/FasTrak/0/GDRDMA +n116-041-188:1377254:1377368 [2] NCCL INFO Channel 02/0 : 18[2] -> 26[2] [send] via NET/FasTrak/2/GDRDMA +n116-041-188:1377252:1377364 [0] NCCL INFO Channel 00/0 : 16[0] -> 33[1] [send] via NET/FasTrak/0/GDRDMA +n116-041-188:1377257:1377365 [5] NCCL INFO Channel 05/0 : 21[5] -> 36[4] [send] via NET/FasTrak/5/GDRDMA +n116-041-188:1377254:1377368 [2] NCCL INFO Channel 02/0 : 18[2] -> 35[3] [send] via NET/FasTrak/2/GDRDMA +n116-041-188:1377256:1377366 [4] NCCL INFO Channel 04/0 : 20[4] -> 28[4] [send] via NET/FasTrak/4/GDRDMA +n116-041-188:1377257:1377365 [5] NCCL INFO Channel 05/0 : 36[4] -> 21[5] [receive] via NET/FasTrak/5/GDRDMA +n116-041-188:1377256:1377366 [4] NCCL INFO Channel 04/0 : 20[4] -> 37[5] [send] via NET/FasTrak/4/GDRDMA +n116-041-188:1377255:1377371 [3] NCCL INFO Channel 03/0 : 19[3] -> 34[2] [send] via NET/FasTrak/3/GDRDMA +n116-041-188:1377259:1377379 [7] NCCL INFO [Proxy Progress] Device 7 CPU core 165 +n116-041-188:1377259:1377369 [7] NCCL INFO Channel 07/0 : 23[7] -> 31[7] [send] via NET/FasTrak/7/GDRDMA +n116-041-188:1377259:1377369 [7] NCCL INFO Channel 06/0 : 14[6] -> 23[7] [receive] via NET/FasTrak/6/GDRDMA +n116-041-188:1377259:1377369 [7] NCCL INFO Channel 15/0 : 14[6] -> 23[7] [receive] via NET/FasTrak/7/GDRDMA +n116-041-188:1377259:1377369 [7] NCCL INFO Channel 07/0 : 23[7] -> 38[6] [send] via NET/FasTrak/7/GDRDMA +n116-041-188:1377259:1377369 [7] NCCL INFO Channel 07/0 : 38[6] -> 23[7] [receive] via NET/FasTrak/7/GDRDMA +n116-041-188:1377258:1377367 [6] NCCL INFO Channel 06/0 : 22[6] -> 30[6] [send] via NET/FasTrak/6/GDRDMA +n116-041-188:1377255:1377371 [3] NCCL INFO Channel 03/0 : 34[2] -> 19[3] [receive] via NET/FasTrak/3/GDRDMA +n116-041-188:1377258:1377367 [6] NCCL INFO Channel 06/0 : 22[6] -> 39[7] [send] via NET/FasTrak/6/GDRDMA +n116-041-188:1377253:1377370 [1] NCCL INFO Channel 01/0 : 17[1] -> 32[0] [send] via NET/FasTrak/1/GDRDMA +n116-041-188:1377253:1377370 [1] NCCL INFO Channel 01/0 : 32[0] -> 17[1] [receive] via NET/FasTrak/1/GDRDMA +n116-041-188:1377257:1377365 [5] NCCL INFO Channel 04/0 : 21[5] -> 12[4] [send] via NET/FasTrak/4/GDRDMA +n116-041-188:1377257:1377365 [5] NCCL INFO Channel 13/0 : 21[5] -> 12[4] [send] via NET/FasTrak/5/GDRDMA +n116-041-188:1377253:1377370 [1] NCCL INFO Channel 00/0 : 17[1] -> 8[0] [send] via NET/FasTrak/0/GDRDMA +n116-041-188:1377253:1377370 [1] NCCL INFO Channel 09/0 : 17[1] -> 8[0] [send] via NET/FasTrak/1/GDRDMA +n116-041-188:1377254:1377368 [2] NCCL INFO Channel 02/0 : 35[3] -> 18[2] [receive] via NET/FasTrak/2/GDRDMA +n116-041-188:1377259:1377369 [7] NCCL INFO Channel 06/0 : 23[7] -> 14[6] [send] via NET/FasTrak/6/GDRDMA +n116-041-188:1377256:1377366 [4] NCCL INFO Channel 04/0 : 37[5] -> 20[4] [receive] via NET/FasTrak/4/GDRDMA +n116-041-188:1377259:1377369 [7] NCCL INFO Channel 15/0 : 23[7] -> 14[6] [send] via NET/FasTrak/7/GDRDMA +n116-041-188:1377254:1377368 [2] NCCL INFO Channel 02/0 : 26[2] -> 18[2] [receive] via NET/FasTrak/2/GDRDMA +n116-041-188:1377253:1377370 [1] NCCL INFO Channel 01/0 : 25[1] -> 17[1] [receive] via NET/FasTrak/1/GDRDMA +n116-041-188:1377252:1377364 [0] NCCL INFO Channel 00/0 : 33[1] -> 16[0] [receive] via NET/FasTrak/0/GDRDMA +n116-041-188:1377256:1377366 [4] NCCL INFO Channel 04/0 : 28[4] -> 20[4] [receive] via NET/FasTrak/4/GDRDMA +n116-041-188:1377259:1377369 [7] NCCL INFO Channel 07/0 : 31[7] -> 23[7] [receive] via NET/FasTrak/7/GDRDMA +n116-041-188:1377257:1377365 [5] NCCL INFO Channel 05/0 : 29[5] -> 21[5] [receive] via NET/FasTrak/5/GDRDMA +n116-041-188:1377255:1377371 [3] NCCL INFO Channel 02/0 : 19[3] -> 10[2] [send] via NET/FasTrak/2/GDRDMA +n116-041-188:1377256:1377366 [4] NCCL INFO Channel 05/0 : 20[4] -> 13[5] [send] via NET/FasTrak/5/GDRDMA +n116-041-188:1377252:1377364 [0] NCCL INFO Channel 00/0 : 24[0] -> 16[0] [receive] via NET/FasTrak/0/GDRDMA +n116-041-188:1377255:1377371 [3] NCCL INFO Channel 11/0 : 19[3] -> 10[2] [send] via NET/FasTrak/3/GDRDMA +n116-041-188:1377256:1377366 [4] NCCL INFO Channel 12/0 : 20[4] -> 13[5] [send] via NET/FasTrak/4/GDRDMA +n116-041-188:1377254:1377368 [2] NCCL INFO Channel 03/0 : 18[2] -> 11[3] [send] via NET/FasTrak/3/GDRDMA +n116-041-188:1377254:1377368 [2] NCCL INFO Channel 10/0 : 18[2] -> 11[3] [send] via NET/FasTrak/2/GDRDMA +n116-041-188:1377255:1377371 [3] NCCL INFO Channel 03/0 : 27[3] -> 19[3] [receive] via NET/FasTrak/3/GDRDMA +n116-041-188:1377257:1377365 [5] NCCL INFO Channel 00/0 : 21[5] -> 20[4] via P2P/CUMEM +n116-041-188:1377257:1377365 [5] NCCL INFO Channel 01/0 : 21[5] -> 20[4] via P2P/CUMEM +n116-041-188:1377257:1377365 [5] NCCL INFO Channel 02/0 : 21[5] -> 20[4] via P2P/CUMEM +n116-041-188:1377257:1377365 [5] NCCL INFO Channel 03/0 : 21[5] -> 20[4] via P2P/CUMEM +n116-041-188:1377257:1377365 [5] NCCL INFO Channel 04/0 : 21[5] -> 20[4] via P2P/CUMEM +n116-041-188:1377257:1377365 [5] NCCL INFO Channel 05/0 : 21[5] -> 20[4] via P2P/CUMEM +n116-041-188:1377257:1377365 [5] NCCL INFO Channel 06/0 : 21[5] -> 20[4] via P2P/CUMEM +n116-041-188:1377257:1377365 [5] NCCL INFO Channel 07/0 : 21[5] -> 20[4] via P2P/CUMEM +n116-041-188:1377257:1377365 [5] NCCL INFO Channel 08/0 : 21[5] -> 20[4] via P2P/CUMEM +n116-041-188:1377257:1377365 [5] NCCL INFO Channel 09/0 : 21[5] -> 20[4] via P2P/CUMEM +n116-041-188:1377257:1377365 [5] NCCL INFO Channel 10/0 : 21[5] -> 20[4] via P2P/CUMEM +n116-041-188:1377252:1377364 [0] NCCL INFO Channel 01/0 : 16[0] -> 9[1] [send] via NET/FasTrak/1/GDRDMA +n116-041-188:1377252:1377364 [0] NCCL INFO Channel 08/0 : 16[0] -> 9[1] [send] via NET/FasTrak/0/GDRDMA +n116-041-188:1377257:1377365 [5] NCCL INFO Channel 11/0 : 21[5] -> 20[4] via P2P/CUMEM +n116-041-188:1377257:1377365 [5] NCCL INFO Channel 12/0 : 21[5] -> 20[4] via P2P/CUMEM +n116-041-188:1377257:1377365 [5] NCCL INFO Channel 13/0 : 21[5] -> 20[4] via P2P/CUMEM +n116-041-188:1377257:1377365 [5] NCCL INFO Channel 14/0 : 21[5] -> 20[4] via P2P/CUMEM +n116-041-188:1377253:1377370 [1] NCCL INFO Channel 00/0 : 17[1] -> 16[0] via P2P/CUMEM +n116-041-188:1377257:1377365 [5] NCCL INFO Channel 15/0 : 21[5] -> 20[4] via P2P/CUMEM +n116-041-188:1377254:1377368 [2] NCCL INFO Channel 00/0 : 18[2] -> 17[1] via P2P/CUMEM +n116-041-188:1377254:1377368 [2] NCCL INFO Channel 03/0 : 18[2] -> 17[1] via P2P/CUMEM +n116-041-188:1377254:1377368 [2] NCCL INFO Channel 04/0 : 18[2] -> 17[1] via P2P/CUMEM +n116-041-188:1377254:1377368 [2] NCCL INFO Channel 05/0 : 18[2] -> 17[1] via P2P/CUMEM +n116-041-188:1377254:1377368 [2] NCCL INFO Channel 06/0 : 18[2] -> 17[1] via P2P/CUMEM +n116-041-188:1377254:1377368 [2] NCCL INFO Channel 07/0 : 18[2] -> 17[1] via P2P/CUMEM +n116-041-188:1377258:1377367 [6] NCCL INFO Channel 06/0 : 39[7] -> 22[6] [receive] via NET/FasTrak/6/GDRDMA +n116-041-188:1377254:1377368 [2] NCCL INFO Channel 08/0 : 18[2] -> 17[1] via P2P/CUMEM +n116-041-188:1377253:1377370 [1] NCCL INFO Channel 01/0 : 17[1] -> 16[0] via P2P/CUMEM +n116-041-188:1377255:1377371 [3] NCCL INFO Channel 00/0 : 19[3] -> 18[2] via P2P/CUMEM +n116-041-188:1377254:1377368 [2] NCCL INFO Channel 11/0 : 18[2] -> 17[1] via P2P/CUMEM +n116-041-188:1377253:1377370 [1] NCCL INFO Channel 02/0 : 17[1] -> 16[0] via P2P/CUMEM +n116-041-188:1377255:1377371 [3] NCCL INFO Channel 01/0 : 19[3] -> 18[2] via P2P/CUMEM +n116-041-188:1377254:1377368 [2] NCCL INFO Channel 12/0 : 18[2] -> 17[1] via P2P/CUMEM +n116-041-188:1377253:1377370 [1] NCCL INFO Channel 03/0 : 17[1] -> 16[0] via P2P/CUMEM +n116-041-188:1377258:1377367 [6] NCCL INFO Channel 06/0 : 30[6] -> 22[6] [receive] via NET/FasTrak/6/GDRDMA +n116-041-188:1377254:1377368 [2] NCCL INFO Channel 13/0 : 18[2] -> 17[1] via P2P/CUMEM +n116-041-188:1377255:1377371 [3] NCCL INFO Channel 02/0 : 19[3] -> 18[2] via P2P/CUMEM +n116-041-188:1377253:1377370 [1] NCCL INFO Channel 04/0 : 17[1] -> 16[0] via P2P/CUMEM +n116-041-188:1377254:1377368 [2] NCCL INFO Channel 14/0 : 18[2] -> 17[1] via P2P/CUMEM +n116-041-188:1377258:1377367 [6] NCCL INFO Channel 07/0 : 22[6] -> 15[7] [send] via NET/FasTrak/7/GDRDMA +n116-041-188:1377253:1377370 [1] NCCL INFO Channel 05/0 : 17[1] -> 16[0] via P2P/CUMEM +n116-041-188:1377258:1377367 [6] NCCL INFO Channel 14/0 : 22[6] -> 15[7] [send] via NET/FasTrak/6/GDRDMA +n116-041-188:1377255:1377371 [3] NCCL INFO Channel 03/0 : 19[3] -> 18[2] via P2P/CUMEM +n116-041-188:1377253:1377370 [1] NCCL INFO Channel 06/0 : 17[1] -> 16[0] via P2P/CUMEM +n116-041-188:1377254:1377368 [2] NCCL INFO Channel 15/0 : 18[2] -> 17[1] via P2P/CUMEM +n116-041-188:1377255:1377371 [3] NCCL INFO Channel 04/0 : 19[3] -> 18[2] via P2P/CUMEM +n116-041-188:1377253:1377370 [1] NCCL INFO Channel 07/0 : 17[1] -> 16[0] via P2P/CUMEM +n116-041-188:1377255:1377371 [3] NCCL INFO Channel 05/0 : 19[3] -> 18[2] via P2P/CUMEM +n116-041-188:1377253:1377370 [1] NCCL INFO Channel 08/0 : 17[1] -> 16[0] via P2P/CUMEM +n116-041-188:1377255:1377371 [3] NCCL INFO Channel 06/0 : 19[3] -> 18[2] via P2P/CUMEM +n116-041-188:1377253:1377370 [1] NCCL INFO Channel 09/0 : 17[1] -> 16[0] via P2P/CUMEM +n116-041-188:1377255:1377371 [3] NCCL INFO Channel 07/0 : 19[3] -> 18[2] via P2P/CUMEM +n116-041-188:1377253:1377370 [1] NCCL INFO Channel 10/0 : 17[1] -> 16[0] via P2P/CUMEM +n116-041-188:1377256:1377366 [4] NCCL INFO Channel 00/0 : 20[4] -> 19[3] via P2P/CUMEM +n116-041-188:1377255:1377371 [3] NCCL INFO Channel 08/0 : 19[3] -> 18[2] via P2P/CUMEM +n116-041-188:1377259:1377369 [7] NCCL INFO Channel 01/0 : 23[7] -> 16[0] via P2P/CUMEM +n116-041-188:1377253:1377370 [1] NCCL INFO Channel 11/0 : 17[1] -> 16[0] via P2P/CUMEM +n116-041-188:1377256:1377366 [4] NCCL INFO Channel 01/0 : 20[4] -> 19[3] via P2P/CUMEM +n116-041-188:1377255:1377371 [3] NCCL INFO Channel 09/0 : 19[3] -> 18[2] via P2P/CUMEM +n116-041-188:1377253:1377370 [1] NCCL INFO Channel 12/0 : 17[1] -> 16[0] via P2P/CUMEM +n116-041-188:1377256:1377366 [4] NCCL INFO Channel 02/0 : 20[4] -> 19[3] via P2P/CUMEM +n116-041-188:1377259:1377369 [7] NCCL INFO Channel 02/0 : 23[7] -> 16[0] via P2P/CUMEM +n116-041-188:1377255:1377371 [3] NCCL INFO Channel 10/0 : 19[3] -> 18[2] via P2P/CUMEM +n116-041-188:1377253:1377370 [1] NCCL INFO Channel 13/0 : 17[1] -> 16[0] via P2P/CUMEM +n116-041-188:1377256:1377366 [4] NCCL INFO Channel 05/0 : 20[4] -> 19[3] via P2P/CUMEM +n116-041-188:1377259:1377369 [7] NCCL INFO Channel 03/0 : 23[7] -> 16[0] via P2P/CUMEM +n116-041-188:1377255:1377371 [3] NCCL INFO Channel 11/0 : 19[3] -> 18[2] via P2P/CUMEM +n116-041-188:1377253:1377370 [1] NCCL INFO Channel 14/0 : 17[1] -> 16[0] via P2P/CUMEM +n116-041-188:1377256:1377366 [4] NCCL INFO Channel 06/0 : 20[4] -> 19[3] via P2P/CUMEM +n116-041-188:1377259:1377369 [7] NCCL INFO Channel 04/0 : 23[7] -> 16[0] via P2P/CUMEM +n116-041-188:1377255:1377371 [3] NCCL INFO Channel 12/0 : 19[3] -> 18[2] via P2P/CUMEM +n116-041-188:1377253:1377370 [1] NCCL INFO Channel 15/0 : 17[1] -> 16[0] via P2P/CUMEM +n116-041-188:1377256:1377366 [4] NCCL INFO Channel 07/0 : 20[4] -> 19[3] via P2P/CUMEM +n116-041-188:1377259:1377369 [7] NCCL INFO Channel 05/0 : 23[7] -> 16[0] via P2P/CUMEM +n116-041-188:1377255:1377371 [3] NCCL INFO Channel 13/0 : 19[3] -> 18[2] via P2P/CUMEM +n116-041-188:1377256:1377366 [4] NCCL INFO Channel 08/0 : 20[4] -> 19[3] via P2P/CUMEM +n116-041-188:1377259:1377369 [7] NCCL INFO Channel 06/0 : 23[7] -> 16[0] via P2P/CUMEM +n116-041-188:1377255:1377371 [3] NCCL INFO Channel 14/0 : 19[3] -> 18[2] via P2P/CUMEM +n116-041-188:1377256:1377366 [4] NCCL INFO Channel 09/0 : 20[4] -> 19[3] via P2P/CUMEM +n116-041-188:1377259:1377369 [7] NCCL INFO Channel 09/0 : 23[7] -> 16[0] via P2P/CUMEM +n116-041-188:1377255:1377371 [3] NCCL INFO Channel 15/0 : 19[3] -> 18[2] via P2P/CUMEM +n116-041-188:1377256:1377366 [4] NCCL INFO Channel 10/0 : 20[4] -> 19[3] via P2P/CUMEM +n116-041-188:1377259:1377369 [7] NCCL INFO Channel 10/0 : 23[7] -> 16[0] via P2P/CUMEM +n116-041-188:1377256:1377366 [4] NCCL INFO Channel 13/0 : 20[4] -> 19[3] via P2P/CUMEM +n116-041-188:1377259:1377369 [7] NCCL INFO Channel 11/0 : 23[7] -> 16[0] via P2P/CUMEM +n116-041-188:1377256:1377366 [4] NCCL INFO Channel 14/0 : 20[4] -> 19[3] via P2P/CUMEM +n116-041-188:1377259:1377369 [7] NCCL INFO Channel 12/0 : 23[7] -> 16[0] via P2P/CUMEM +n116-041-188:1377256:1377366 [4] NCCL INFO Channel 15/0 : 20[4] -> 19[3] via P2P/CUMEM +n116-041-188:1377259:1377369 [7] NCCL INFO Channel 13/0 : 23[7] -> 16[0] via P2P/CUMEM +n116-041-188:1377259:1377369 [7] NCCL INFO Channel 14/0 : 23[7] -> 16[0] via P2P/CUMEM +n116-041-188:1377259:1377369 [7] NCCL INFO Channel 00/0 : 23[7] -> 22[6] via P2P/CUMEM +n116-041-188:1377259:1377369 [7] NCCL INFO Channel 01/0 : 23[7] -> 22[6] via P2P/CUMEM +n116-041-188:1377259:1377369 [7] NCCL INFO Channel 02/0 : 23[7] -> 22[6] via P2P/CUMEM +n116-041-188:1377259:1377369 [7] NCCL INFO Channel 03/0 : 23[7] -> 22[6] via P2P/CUMEM +n116-041-188:1377259:1377369 [7] NCCL INFO Channel 04/0 : 23[7] -> 22[6] via P2P/CUMEM +n116-041-188:1377259:1377369 [7] NCCL INFO Channel 05/0 : 23[7] -> 22[6] via P2P/CUMEM +n116-041-188:1377259:1377369 [7] NCCL INFO Channel 06/0 : 23[7] -> 22[6] via P2P/CUMEM +n116-041-188:1377259:1377369 [7] NCCL INFO Channel 07/0 : 23[7] -> 22[6] via P2P/CUMEM +n116-041-188:1377259:1377369 [7] NCCL INFO Channel 08/0 : 23[7] -> 22[6] via P2P/CUMEM +n116-041-188:1377259:1377369 [7] NCCL INFO Channel 09/0 : 23[7] -> 22[6] via P2P/CUMEM +n116-041-188:1377259:1377369 [7] NCCL INFO Channel 10/0 : 23[7] -> 22[6] via P2P/CUMEM +n116-041-188:1377259:1377369 [7] NCCL INFO Channel 11/0 : 23[7] -> 22[6] via P2P/CUMEM +n116-041-188:1377259:1377369 [7] NCCL INFO Channel 12/0 : 23[7] -> 22[6] via P2P/CUMEM +n116-041-188:1377259:1377369 [7] NCCL INFO Channel 13/0 : 23[7] -> 22[6] via P2P/CUMEM +n116-041-188:1377259:1377369 [7] NCCL INFO Channel 14/0 : 23[7] -> 22[6] via P2P/CUMEM +n116-041-188:1377259:1377369 [7] NCCL INFO Channel 15/0 : 23[7] -> 22[6] via P2P/CUMEM +n116-041-188:1377258:1377367 [6] NCCL INFO Channel 00/0 : 22[6] -> 21[5] via P2P/CUMEM +n116-041-188:1377258:1377367 [6] NCCL INFO Channel 01/0 : 22[6] -> 21[5] via P2P/CUMEM +n116-041-188:1377258:1377367 [6] NCCL INFO Channel 02/0 : 22[6] -> 21[5] via P2P/CUMEM +n116-041-188:1377258:1377367 [6] NCCL INFO Channel 03/0 : 22[6] -> 21[5] via P2P/CUMEM +n116-041-188:1377258:1377367 [6] NCCL INFO Channel 04/0 : 22[6] -> 21[5] via P2P/CUMEM +n116-041-188:1377258:1377367 [6] NCCL INFO Channel 07/0 : 22[6] -> 21[5] via P2P/CUMEM +n116-041-188:1377258:1377367 [6] NCCL INFO Channel 08/0 : 22[6] -> 21[5] via P2P/CUMEM +n116-041-188:1377258:1377367 [6] NCCL INFO Channel 09/0 : 22[6] -> 21[5] via P2P/CUMEM +n116-041-188:1377258:1377367 [6] NCCL INFO Channel 10/0 : 22[6] -> 21[5] via P2P/CUMEM +n116-041-188:1377258:1377367 [6] NCCL INFO Channel 11/0 : 22[6] -> 21[5] via P2P/CUMEM +n116-041-188:1377258:1377367 [6] NCCL INFO Channel 12/0 : 22[6] -> 21[5] via P2P/CUMEM +n116-041-188:1377258:1377367 [6] NCCL INFO Channel 15/0 : 22[6] -> 21[5] via P2P/CUMEM +n116-041-188:1377254:1377368 [2] NCCL INFO Connected all trees +n116-041-188:1377252:1377364 [0] NCCL INFO Connected all trees +n116-041-188:1377253:1377370 [1] NCCL INFO Connected all trees +n116-041-188:1377259:1377369 [7] NCCL INFO Connected all trees +n116-041-188:1377258:1377367 [6] NCCL INFO Connected all trees +n116-041-188:1377255:1377371 [3] NCCL INFO Connected all trees +n116-041-188:1377257:1377365 [5] NCCL INFO Connected all trees +n116-041-188:1377256:1377366 [4] NCCL INFO Connected all trees +n116-041-188:1377252:1377400 [0] NCCL INFO Channel 00/0 : 9[1] -> 16[0] [receive] via NET/FasTrak/0/GDRDMA +n116-041-188:1377252:1377400 [0] NCCL INFO Channel 00/0 : 16[0] -> 23[7] via P2P/CUMEM +n116-041-188:1377252:1377400 [0] NCCL INFO Channel 08/0 : 16[0] -> 23[7] via P2P/CUMEM +n116-041-188:1377259:1377401 [7] NCCL INFO Channel 06/0 : 23[7] -> 30[6] [send] via NET/FasTrak/6/GDRDMA +n116-041-188:1377259:1377401 [7] NCCL INFO Channel 14/0 : 23[7] -> 30[6] [send] via NET/FasTrak/6/GDRDMA +n116-041-188:1377257:1377402 [5] NCCL INFO Channel 05/0 : 21[5] -> 22[6] via P2P/CUMEM +n116-041-188:1377257:1377402 [5] NCCL INFO Channel 13/0 : 21[5] -> 22[6] via P2P/CUMEM +n116-041-188:1377253:1377404 [1] NCCL INFO Channel 01/0 : 17[1] -> 18[2] via P2P/CUMEM +n116-041-188:1377253:1377404 [1] NCCL INFO Channel 09/0 : 17[1] -> 18[2] via P2P/CUMEM +n116-041-188:1377253:1377404 [1] NCCL INFO Channel 00/0 : 17[1] -> 24[0] [send] via NET/FasTrak/0/GDRDMA +n116-041-188:1377253:1377404 [1] NCCL INFO Channel 08/0 : 17[1] -> 24[0] [send] via NET/FasTrak/0/GDRDMA +n116-041-188:1377254:1377403 [2] NCCL INFO Channel 02/0 : 11[3] -> 18[2] [receive] via NET/FasTrak/2/GDRDMA +n116-041-188:1377255:1377405 [3] NCCL INFO Channel 03/0 : 19[3] -> 20[4] via P2P/CUMEM +n116-041-188:1377255:1377405 [3] NCCL INFO Channel 11/0 : 19[3] -> 20[4] via P2P/CUMEM +n116-041-188:1377253:1377404 [1] NCCL INFO Channel 01/0 : 8[0] -> 17[1] [receive] via NET/FasTrak/1/GDRDMA +n116-041-188:1377252:1377400 [0] NCCL INFO Channel 01/0 : 16[0] -> 25[1] [send] via NET/FasTrak/1/GDRDMA +n116-041-188:1377252:1377400 [0] NCCL INFO Channel 09/0 : 16[0] -> 25[1] [send] via NET/FasTrak/1/GDRDMA +n116-041-188:1377254:1377403 [2] NCCL INFO Channel 03/0 : 18[2] -> 27[3] [send] via NET/FasTrak/3/GDRDMA +n116-041-188:1377254:1377403 [2] NCCL INFO Channel 11/0 : 18[2] -> 27[3] [send] via NET/FasTrak/3/GDRDMA +n116-041-188:1377255:1377405 [3] NCCL INFO Channel 02/0 : 19[3] -> 26[2] [send] via NET/FasTrak/2/GDRDMA +n116-041-188:1377255:1377405 [3] NCCL INFO Channel 10/0 : 19[3] -> 26[2] [send] via NET/FasTrak/2/GDRDMA +n116-041-188:1377256:1377406 [4] NCCL INFO Channel 04/0 : 13[5] -> 20[4] [receive] via NET/FasTrak/4/GDRDMA +n116-041-188:1377256:1377406 [4] NCCL INFO Channel 05/0 : 20[4] -> 29[5] [send] via NET/FasTrak/5/GDRDMA +n116-041-188:1377256:1377406 [4] NCCL INFO Channel 13/0 : 20[4] -> 29[5] [send] via NET/FasTrak/5/GDRDMA +n116-041-188:1377255:1377405 [3] NCCL INFO Channel 03/0 : 10[2] -> 19[3] [receive] via NET/FasTrak/3/GDRDMA +n116-041-188:1377257:1377402 [5] NCCL INFO Channel 04/0 : 21[5] -> 28[4] [send] via NET/FasTrak/4/GDRDMA +n116-041-188:1377257:1377402 [5] NCCL INFO Channel 12/0 : 21[5] -> 28[4] [send] via NET/FasTrak/4/GDRDMA +n116-041-188:1377258:1377407 [6] NCCL INFO Channel 06/0 : 15[7] -> 22[6] [receive] via NET/FasTrak/6/GDRDMA +n116-041-188:1377258:1377407 [6] NCCL INFO Channel 07/0 : 22[6] -> 31[7] [send] via NET/FasTrak/7/GDRDMA +n116-041-188:1377258:1377407 [6] NCCL INFO Channel 15/0 : 22[6] -> 31[7] [send] via NET/FasTrak/7/GDRDMA +n116-041-188:1377258:1377407 [6] NCCL INFO Channel 06/0 : 22[6] -> 21[5] via P2P/CUMEM +n116-041-188:1377258:1377407 [6] NCCL INFO Channel 14/0 : 22[6] -> 21[5] via P2P/CUMEM +n116-041-188:1377257:1377402 [5] NCCL INFO Channel 05/0 : 12[4] -> 21[5] [receive] via NET/FasTrak/5/GDRDMA +n116-041-188:1377254:1377403 [2] NCCL INFO Channel 02/0 : 18[2] -> 17[1] via P2P/CUMEM +n116-041-188:1377254:1377403 [2] NCCL INFO Channel 10/0 : 18[2] -> 17[1] via P2P/CUMEM +n116-041-188:1377259:1377401 [7] NCCL INFO Channel 07/0 : 14[6] -> 23[7] [receive] via NET/FasTrak/7/GDRDMA +n116-041-188:1377256:1377406 [4] NCCL INFO Channel 04/0 : 20[4] -> 19[3] via P2P/CUMEM +n116-041-188:1377259:1377401 [7] NCCL INFO Channel 07/0 : 23[7] -> 16[0] via P2P/CUMEM +n116-041-188:1377256:1377406 [4] NCCL INFO Channel 12/0 : 20[4] -> 19[3] via P2P/CUMEM +n116-041-188:1377259:1377401 [7] NCCL INFO Channel 15/0 : 23[7] -> 16[0] via P2P/CUMEM +n116-041-188:1377259:1377401 [7] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-041-188:1377258:1377407 [6] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-041-188:1377257:1377402 [5] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-041-188:1377255:1377405 [3] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-041-188:1377256:1377406 [4] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-041-188:1377254:1377403 [2] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-041-188:1377253:1377404 [1] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-041-188:1377252:1377400 [0] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +[rank18]:[W727 05:39:50.511823752 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n116-041-188.byted.org]:49260, remote=[n116-040-168.byted.org]:25264): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7fbb9cf785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7fbb862cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7fbb862cff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7fbb862d084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7fbb862ca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fbb479c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7fbba03cc4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7fbba2fa31f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7fbba302389c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank18]:[W727 05:39:50.525036727 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 18] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank20]:[W727 05:39:50.553146848 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n116-041-188.byted.org]:42766, remote=[n116-040-168.byted.org]:25264): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7efd1c9785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7efd060cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7efd060cff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7efd060d084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7efd060ca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7efcc77c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7efd201014a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7efd22cd81f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7efd22d5889c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank20]:[W727 05:39:50.560638741 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 20] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank21]:[W727 05:39:50.719512805 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n116-041-188.byted.org]:49290, remote=[n116-040-168.byted.org]:25264): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7efd131785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7efcfc4cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7efcfc4cff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7efcfc4d084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7efcfc4ca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7efcbdbc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7efd1660f4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7efd191e61f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7efd1926689c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank21]:[W727 05:39:50.725522523 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 21] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank17]:[W727 05:39:50.732752915 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n116-041-188.byted.org]:49272, remote=[n116-040-168.byted.org]:25264): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f08207785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f0809ecdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7f0809ecff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7f0809ed084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7f0809eca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f07cb5c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f0823f054a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f0826adc1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f0826b5c89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank17]:[W727 05:39:50.739143931 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 17] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank23]:[W727 05:39:50.881838021 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n116-041-188.byted.org]:49298, remote=[n116-040-168.byted.org]:25264): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7ffa04f785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7ff9ee6cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7ff9ee6cff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7ff9ee6d084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7ff9ee6ca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7ff9afdc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7ffa087174a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7ffa0b2ee1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7ffa0b36e89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank23]:[W727 05:39:50.889313514 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 23] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank16]:[W727 05:39:51.916691135 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=71, addr=[n116-041-188.byted.org]:42756, remote=[n116-040-168.byted.org]:25264): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7fe91f9785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7fe9090cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7fe9090cff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7fe9090d084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7fe9090ca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fe8ca7c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7fe9231974a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7fe925d6e1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7fe925dee89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank16]:[W727 05:39:51.922997569 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 16] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank19]:[W727 05:39:51.017356607 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n116-041-188.byted.org]:49250, remote=[n116-040-168.byted.org]:25264): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7fd01c9785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7fd0060cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7fd0060cff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7fd0060d084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7fd0060ca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fcfc77c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7fd0200d14a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7fd022ca81f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7fd022d2889c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank19]:[W727 05:39:51.023630363 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 19] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank22]:[W727 05:39:51.112716939 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n116-041-188.byted.org]:49286, remote=[n116-040-168.byted.org]:25264): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f37c1d785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f37ab4cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7f37ab4cff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7f37ab4d084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7f37ab4ca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f376cbc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f37c55434a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f37c811a1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f37c819a89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank22]:[W727 05:39:51.119829252 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 22] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank18]:[W727 05:39:51.525223220 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-188.byted.org]:49260, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7fbb9cf785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7fbb862cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7fbb862cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7fbb862d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7fbb862ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fbb479c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7fbba03cc4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7fbba2fa31f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7fbba302389c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank18]:[W727 05:39:51.532246025 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 18] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank20]:[W727 05:39:51.560831941 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-188.byted.org]:42766, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7efd1c9785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7efd060cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7efd060cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7efd060d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7efd060ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7efcc77c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7efd201014a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7efd22cd81f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7efd22d5889c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank20]:[W727 05:39:51.567917673 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 20] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank21]:[W727 05:39:51.725693470 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-188.byted.org]:49290, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7efd131785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7efcfc4cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7efcfc4cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7efcfc4d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7efcfc4ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7efcbdbc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7efd1660f4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7efd191e61f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7efd1926689c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank21]:[W727 05:39:51.732080641 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 21] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank17]:[W727 05:39:51.739311080 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-188.byted.org]:49272, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f08207785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f0809ecdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f0809ecf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f0809ed0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f0809eca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f07cb5c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f0823f054a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f0826adc1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f0826b5c89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank17]:[W727 05:39:51.746708377 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 17] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank23]:[W727 05:39:51.889478933 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-188.byted.org]:49298, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7ffa04f785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7ff9ee6cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7ff9ee6cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7ff9ee6d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7ff9ee6ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7ff9afdc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7ffa087174a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7ffa0b2ee1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7ffa0b36e89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank23]:[W727 05:39:51.895698733 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 23] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank16]:[W727 05:39:52.923174823 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=71, addr=[n116-041-188.byted.org]:42756, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7fe91f9785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7fe9090cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7fe9090cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7fe9090d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7fe9090ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fe8ca7c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7fe9231974a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7fe925d6e1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7fe925dee89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank16]:[W727 05:39:52.929421546 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 16] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank19]:[W727 05:39:52.023792087 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-188.byted.org]:49250, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7fd01c9785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7fd0060cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7fd0060cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7fd0060d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7fd0060ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fcfc77c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7fd0200d14a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7fd022ca81f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7fd022d2889c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank19]:[W727 05:39:52.029952841 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 19] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank22]:[W727 05:39:52.119994832 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-188.byted.org]:49286, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f37c1d785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f37ab4cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f37ab4cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f37ab4d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f37ab4ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f376cbc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f37c55434a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f37c811a1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f37c819a89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank22]:[W727 05:39:52.126411851 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 22] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +E0727 05:39:52.643000 1377175 torch/distributed/elastic/multiprocessing/api.py:874] failed (exitcode: -15) local_rank: 0 (pid: 1377252) of binary: /usr/bin/python3 +Traceback (most recent call last): + File "/usr/local/bin/torchrun", line 7, in + sys.exit(main()) + ^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 355, in wrapper + return f(*args, **kwargs) + ^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 892, in main + run(args) + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 883, in run + elastic_launch( + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 139, in __call__ + return launch_agent(self._config, self._entrypoint, list(args)) + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 270, in launch_agent + raise ChildFailedError( +torch.distributed.elastic.multiprocessing.errors.ChildFailedError: +============================================================ +main_cross_2direct_balance_realscale_perturb_v4_accum.py FAILED +------------------------------------------------------------ +Failures: +[1]: + time : 2026-07-27_05:39:52 + host : n116-041-188.byted.org + rank : 17 (local_rank: 1) + exitcode : -15 (pid: 1377253) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 1377253 +[2]: + time : 2026-07-27_05:39:52 + host : n116-041-188.byted.org + rank : 18 (local_rank: 2) + exitcode : -15 (pid: 1377254) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 1377254 +[3]: + time : 2026-07-27_05:39:52 + host : n116-041-188.byted.org + rank : 19 (local_rank: 3) + exitcode : -15 (pid: 1377255) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 1377255 +[4]: + time : 2026-07-27_05:39:52 + host : n116-041-188.byted.org + rank : 20 (local_rank: 4) + exitcode : -15 (pid: 1377256) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 1377256 +[5]: + time : 2026-07-27_05:39:52 + host : n116-041-188.byted.org + rank : 21 (local_rank: 5) + exitcode : -15 (pid: 1377257) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 1377257 +[6]: + time : 2026-07-27_05:39:52 + host : n116-041-188.byted.org + rank : 22 (local_rank: 6) + exitcode : -15 (pid: 1377258) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 1377258 +[7]: + time : 2026-07-27_05:39:52 + host : n116-041-188.byted.org + rank : 23 (local_rank: 7) + exitcode : -15 (pid: 1377259) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 1377259 +------------------------------------------------------------ +Root Cause (first observed failure): +[0]: + time : 2026-07-27_05:39:52 + host : n116-041-188.byted.org + rank : 16 (local_rank: 0) + exitcode : -15 (pid: 1377252) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 1377252 +============================================================ diff --git a/ScaleDF_rebuttal/_host_logs/20260726_2350_vith64_free8_port25264_launch/launcher_rank3_n111-046-081.log b/ScaleDF_rebuttal/_host_logs/20260726_2350_vith64_free8_port25264_launch/launcher_rank3_n111-046-081.log new file mode 100644 index 0000000000000000000000000000000000000000..9519a0627ee71014e15febe8610c55d4a2ad3b4b --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_2350_vith64_free8_port25264_launch/launcher_rank3_n111-046-081.log @@ -0,0 +1,2045 @@ +started_utc=2026-07-26T23:50:26Z +hostname=n111-046-081 +node_rank=3 +nnodes=8 +nproc_per_node=8 +world_size=64 +master_addr=10.116.40.168 +master_port=25264 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +main=main_cross_2direct_balance_realscale_perturb_v4_accum.py +model_name=deit_huge_patch14_LS +pretrain=/tmp/ScaleDF_repro_20260724/pretrained/deit_3_huge_224_21k_v1.pth +resume=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/checkpoint_snapshots/vith32_latest_for_64card_effbs4096_20260726_234604.pth +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_64card_effbs4096_from_epoch22_snapshot_free8_master_n116-040-168_20260726_2350_vith64_port25264 +batch_per_gpu=64 +accum_iter=1 +effective_batch=4096 +eval_batch_per_gpu=64 +epochs=40 +W0726 23:50:38.774000 3352460 torch/distributed/run.py:766] +W0726 23:50:38.774000 3352460 torch/distributed/run.py:766] ***************************************** +W0726 23:50:38.774000 3352460 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 23:50:38.774000 3352460 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 24): env:// +| distributed init (rank 27): env:// +[rank27]:[W726 23:51:08.882291338 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 27] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 30): env:// +[rank30]:[W726 23:51:08.114065573 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 30] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 25): env:// +[rank25]:[W726 23:51:08.156681823 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 25] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 28): env:// +[rank28]:[W726 23:51:08.394584331 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 28] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 31): env:// +[rank31]:[W726 23:51:08.486083809 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 31] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 26): env:// +[rank26]:[W726 23:51:08.527265513 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 26] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank24]:[W726 23:51:08.555684977 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 24] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 29): env:// +[rank29]:[W726 23:51:09.572554756 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 29] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +n111-046-081:3352527:3352527 [2] NCCL INFO cudaDriverVersion 12060 +n111-046-081:3352527:3352527 [2] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n111-046-081:3352527:3352527 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n111-046-081:3352527:3352527 [2] NCCL INFO NCCL version 2.26.2+cuda12.2 +n111-046-081:3352527:3352527 [2] NCCL INFO Comm config Blocking set to 1 +n111-046-081:3352528:3352528 [3] NCCL INFO cudaDriverVersion 12060 +n111-046-081:3352528:3352528 [3] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n111-046-081:3352528:3352528 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n111-046-081:3352528:3352528 [3] NCCL INFO NCCL version 2.26.2+cuda12.2 +n111-046-081:3352530:3352530 [5] NCCL INFO cudaDriverVersion 12060 +n111-046-081:3352530:3352530 [5] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n111-046-081:3352530:3352530 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n111-046-081:3352530:3352530 [5] NCCL INFO NCCL version 2.26.2+cuda12.2 +n111-046-081:3352528:3352528 [3] NCCL INFO Comm config Blocking set to 1 +n111-046-081:3352531:3352531 [6] NCCL INFO cudaDriverVersion 12060 +n111-046-081:3352531:3352531 [6] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n111-046-081:3352531:3352531 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n111-046-081:3352531:3352531 [6] NCCL INFO NCCL version 2.26.2+cuda12.2 +n111-046-081:3352526:3352526 [1] NCCL INFO cudaDriverVersion 12060 +n111-046-081:3352526:3352526 [1] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n111-046-081:3352526:3352526 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n111-046-081:3352526:3352526 [1] NCCL INFO NCCL version 2.26.2+cuda12.2 +n111-046-081:3352530:3352530 [5] NCCL INFO Comm config Blocking set to 1 +n111-046-081:3352531:3352531 [6] NCCL INFO Comm config Blocking set to 1 +n111-046-081:3352525:3352525 [0] NCCL INFO cudaDriverVersion 12060 +n111-046-081:3352525:3352525 [0] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n111-046-081:3352525:3352525 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n111-046-081:3352525:3352525 [0] NCCL INFO NCCL version 2.26.2+cuda12.2 +n111-046-081:3352526:3352526 [1] NCCL INFO Comm config Blocking set to 1 +n111-046-081:3352525:3352525 [0] NCCL INFO Comm config Blocking set to 1 +n111-046-081:3352532:3352532 [7] NCCL INFO cudaDriverVersion 12060 +n111-046-081:3352532:3352532 [7] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n111-046-081:3352532:3352532 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n111-046-081:3352532:3352532 [7] NCCL INFO NCCL version 2.26.2+cuda12.2 +n111-046-081:3352532:3352532 [7] NCCL INFO Comm config Blocking set to 1 +n111-046-081:3352529:3352529 [4] NCCL INFO cudaDriverVersion 12060 +n111-046-081:3352529:3352529 [4] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n111-046-081:3352529:3352529 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n111-046-081:3352529:3352529 [4] NCCL INFO NCCL version 2.26.2+cuda12.2 +n111-046-081:3352529:3352529 [4] NCCL INFO Comm config Blocking set to 1 +n111-046-081:3352527:3352593 [2] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n111-046-081:3352527:3352593 [2] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n111-046-081:3352527:3352593 [2] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n111-046-081:3352527:3352593 [2] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n111-046-081:3352527:3352593 [2] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n111-046-081:3352527:3352593 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n111-046-081:3352527:3352593 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n111-046-081:3352527:3352593 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n111-046-081:3352527:3352593 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n111-046-081:3352527:3352593 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n111-046-081:3352530:3352595 [5] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n111-046-081:3352530:3352595 [5] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n111-046-081:3352530:3352595 [5] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n111-046-081:3352530:3352595 [5] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n111-046-081:3352530:3352595 [5] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n111-046-081:3352530:3352595 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n111-046-081:3352530:3352595 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n111-046-081:3352530:3352595 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n111-046-081:3352530:3352595 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n111-046-081:3352530:3352595 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n111-046-081:3352531:3352596 [6] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n111-046-081:3352531:3352596 [6] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n111-046-081:3352531:3352596 [6] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n111-046-081:3352531:3352596 [6] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n111-046-081:3352531:3352596 [6] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n111-046-081:3352531:3352596 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n111-046-081:3352531:3352596 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n111-046-081:3352531:3352596 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n111-046-081:3352531:3352596 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n111-046-081:3352531:3352596 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n111-046-081:3352525:3352598 [0] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n111-046-081:3352525:3352598 [0] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n111-046-081:3352525:3352598 [0] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n111-046-081:3352525:3352598 [0] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n111-046-081:3352525:3352598 [0] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n111-046-081:3352525:3352598 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n111-046-081:3352525:3352598 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n111-046-081:3352525:3352598 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n111-046-081:3352525:3352598 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n111-046-081:3352525:3352598 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n111-046-081:3352529:3352600 [4] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n111-046-081:3352529:3352600 [4] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n111-046-081:3352529:3352600 [4] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n111-046-081:3352529:3352600 [4] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n111-046-081:3352529:3352600 [4] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n111-046-081:3352529:3352600 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n111-046-081:3352526:3352597 [1] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n111-046-081:3352529:3352600 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n111-046-081:3352529:3352600 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n111-046-081:3352529:3352600 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n111-046-081:3352529:3352600 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n111-046-081:3352526:3352597 [1] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n111-046-081:3352526:3352597 [1] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n111-046-081:3352526:3352597 [1] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n111-046-081:3352526:3352597 [1] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n111-046-081:3352526:3352597 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n111-046-081:3352526:3352597 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n111-046-081:3352526:3352597 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n111-046-081:3352526:3352597 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n111-046-081:3352526:3352597 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n111-046-081:3352532:3352599 [7] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n111-046-081:3352532:3352599 [7] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n111-046-081:3352532:3352599 [7] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n111-046-081:3352532:3352599 [7] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n111-046-081:3352532:3352599 [7] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n111-046-081:3352532:3352599 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n111-046-081:3352532:3352599 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n111-046-081:3352532:3352599 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n111-046-081:3352532:3352599 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n111-046-081:3352532:3352599 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n111-046-081:3352527:3352593 [2] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n111-046-081:3352527:3352593 [2] NCCL INFO Using network FasTrak +n111-046-081:3352532:3352599 [7] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n111-046-081:3352532:3352599 [7] NCCL INFO Using network FasTrak +n111-046-081:3352525:3352598 [0] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n111-046-081:3352525:3352598 [0] NCCL INFO Using network FasTrak +n111-046-081:3352530:3352595 [5] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n111-046-081:3352530:3352595 [5] NCCL INFO Using network FasTrak +n111-046-081:3352528:3352594 [3] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n111-046-081:3352528:3352594 [3] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n111-046-081:3352528:3352594 [3] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n111-046-081:3352528:3352594 [3] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n111-046-081:3352528:3352594 [3] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n111-046-081:3352528:3352594 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n111-046-081:3352528:3352594 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n111-046-081:3352528:3352594 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n111-046-081:3352528:3352594 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n111-046-081:3352528:3352594 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n111-046-081:3352531:3352596 [6] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n111-046-081:3352531:3352596 [6] NCCL INFO Using network FasTrak +n111-046-081:3352526:3352597 [1] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n111-046-081:3352526:3352597 [1] NCCL INFO Using network FasTrak +n111-046-081:3352529:3352600 [4] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n111-046-081:3352529:3352600 [4] NCCL INFO Using network FasTrak +n111-046-081:3352528:3352594 [3] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n111-046-081:3352528:3352594 [3] NCCL INFO Using network FasTrak +n111-046-081:3352527:3352593 [2] NCCL INFO DMA-BUF is available on GPU device 2 +n111-046-081:3352527:3352593 [2] NCCL INFO ncclCommInitRankConfig comm 0xbfe2f00 rank 26 nranks 64 cudaDev 2 nvmlDev 2 busId b000 commId 0xfb2d11625d05007a - Init START +n111-046-081:3352532:3352599 [7] NCCL INFO DMA-BUF is available on GPU device 7 +n111-046-081:3352525:3352598 [0] NCCL INFO DMA-BUF is available on GPU device 0 +n111-046-081:3352532:3352599 [7] NCCL INFO ncclCommInitRankConfig comm 0xcc43090 rank 31 nranks 64 cudaDev 7 nvmlDev 7 busId 8c000 commId 0xfb2d11625d05007a - Init START +n111-046-081:3352530:3352595 [5] NCCL INFO DMA-BUF is available on GPU device 5 +n111-046-081:3352529:3352600 [4] NCCL INFO DMA-BUF is available on GPU device 4 +n111-046-081:3352525:3352598 [0] NCCL INFO ncclCommInitRankConfig comm 0xc925780 rank 24 nranks 64 cudaDev 0 nvmlDev 0 busId 4000 commId 0xfb2d11625d05007a - Init START +n111-046-081:3352531:3352596 [6] NCCL INFO DMA-BUF is available on GPU device 6 +n111-046-081:3352530:3352595 [5] NCCL INFO ncclCommInitRankConfig comm 0xc14dc20 rank 29 nranks 64 cudaDev 5 nvmlDev 5 busId 85000 commId 0xfb2d11625d05007a - Init START +n111-046-081:3352529:3352600 [4] NCCL INFO ncclCommInitRankConfig comm 0xc569f60 rank 28 nranks 64 cudaDev 4 nvmlDev 4 busId 84000 commId 0xfb2d11625d05007a - Init START +n111-046-081:3352531:3352596 [6] NCCL INFO ncclCommInitRankConfig comm 0xb2b9f40 rank 30 nranks 64 cudaDev 6 nvmlDev 6 busId 8b000 commId 0xfb2d11625d05007a - Init START +n111-046-081:3352530:3352595 [5] NCCL INFO RAS client listening socket at ::1<28028> +n111-046-081:3352531:3352596 [6] NCCL INFO RAS client listening socket at ::1<28028> +n111-046-081:3352526:3352597 [1] NCCL INFO DMA-BUF is available on GPU device 1 +n111-046-081:3352532:3352599 [7] NCCL INFO RAS client listening socket at ::1<28028> +n111-046-081:3352526:3352597 [1] NCCL INFO ncclCommInitRankConfig comm 0xbdbfa40 rank 25 nranks 64 cudaDev 1 nvmlDev 1 busId 5000 commId 0xfb2d11625d05007a - Init START +n111-046-081:3352525:3352598 [0] NCCL INFO RAS client listening socket at ::1<28028> +n111-046-081:3352526:3352597 [1] NCCL INFO RAS client listening socket at ::1<28028> +n111-046-081:3352528:3352594 [3] NCCL INFO DMA-BUF is available on GPU device 3 +n111-046-081:3352528:3352594 [3] NCCL INFO ncclCommInitRankConfig comm 0xcd5e030 rank 27 nranks 64 cudaDev 3 nvmlDev 3 busId c000 commId 0xfb2d11625d05007a - Init START +n111-046-081:3352527:3352593 [2] NCCL INFO RAS client listening socket at ::1<28028> +n111-046-081:3352528:3352594 [3] NCCL INFO RAS client listening socket at ::1<28028> +n111-046-081:3352529:3352600 [4] NCCL INFO RAS client listening socket at ::1<28028> +n111-046-081:3352530:3352595 [5] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n111-046-081:3352530:3352595 [5] NCCL INFO TOPO/NET : Importing network plugins to topology +n111-046-081:3352530:3352595 [5] NCCL INFO Retrieving state for FasTrak +n111-046-081:3352530:3352595 [5] NCCL INFO Initialized state 0 for FasTrak +n111-046-081:3352530:3352595 [5] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n111-046-081:3352530:3352595 [5] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n111-046-081:3352530:3352595 [5] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n111-046-081:3352530:3352595 [5] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n111-046-081:3352530:3352595 [5] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n111-046-081:3352530:3352595 [5] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n111-046-081:3352530:3352595 [5] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n111-046-081:3352530:3352595 [5] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n111-046-081:3352529:3352600 [4] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n111-046-081:3352529:3352600 [4] NCCL INFO TOPO/NET : Importing network plugins to topology +n111-046-081:3352529:3352600 [4] NCCL INFO Retrieving state for FasTrak +n111-046-081:3352529:3352600 [4] NCCL INFO Initialized state 0 for FasTrak +n111-046-081:3352529:3352600 [4] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n111-046-081:3352529:3352600 [4] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n111-046-081:3352529:3352600 [4] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n111-046-081:3352531:3352596 [6] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n111-046-081:3352529:3352600 [4] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n111-046-081:3352531:3352596 [6] NCCL INFO TOPO/NET : Importing network plugins to topology +n111-046-081:3352531:3352596 [6] NCCL INFO Retrieving state for FasTrak +n111-046-081:3352531:3352596 [6] NCCL INFO Initialized state 0 for FasTrak +n111-046-081:3352529:3352600 [4] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n111-046-081:3352529:3352600 [4] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n111-046-081:3352531:3352596 [6] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n111-046-081:3352531:3352596 [6] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n111-046-081:3352529:3352600 [4] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n111-046-081:3352529:3352600 [4] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n111-046-081:3352531:3352596 [6] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n111-046-081:3352531:3352596 [6] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n111-046-081:3352531:3352596 [6] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n111-046-081:3352531:3352596 [6] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n111-046-081:3352531:3352596 [6] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n111-046-081:3352531:3352596 [6] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n111-046-081:3352528:3352594 [3] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n111-046-081:3352528:3352594 [3] NCCL INFO TOPO/NET : Importing network plugins to topology +n111-046-081:3352528:3352594 [3] NCCL INFO Retrieving state for FasTrak +n111-046-081:3352528:3352594 [3] NCCL INFO Initialized state 0 for FasTrak +n111-046-081:3352528:3352594 [3] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n111-046-081:3352528:3352594 [3] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n111-046-081:3352528:3352594 [3] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n111-046-081:3352528:3352594 [3] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n111-046-081:3352532:3352599 [7] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n111-046-081:3352532:3352599 [7] NCCL INFO TOPO/NET : Importing network plugins to topology +n111-046-081:3352532:3352599 [7] NCCL INFO Retrieving state for FasTrak +n111-046-081:3352532:3352599 [7] NCCL INFO Initialized state 0 for FasTrak +n111-046-081:3352528:3352594 [3] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n111-046-081:3352528:3352594 [3] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n111-046-081:3352532:3352599 [7] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n111-046-081:3352528:3352594 [3] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n111-046-081:3352532:3352599 [7] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n111-046-081:3352528:3352594 [3] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n111-046-081:3352532:3352599 [7] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n111-046-081:3352532:3352599 [7] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n111-046-081:3352532:3352599 [7] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n111-046-081:3352532:3352599 [7] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n111-046-081:3352532:3352599 [7] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n111-046-081:3352532:3352599 [7] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n111-046-081:3352527:3352593 [2] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n111-046-081:3352527:3352593 [2] NCCL INFO TOPO/NET : Importing network plugins to topology +n111-046-081:3352527:3352593 [2] NCCL INFO Retrieving state for FasTrak +n111-046-081:3352527:3352593 [2] NCCL INFO Initialized state 0 for FasTrak +n111-046-081:3352527:3352593 [2] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n111-046-081:3352527:3352593 [2] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n111-046-081:3352527:3352593 [2] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n111-046-081:3352527:3352593 [2] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n111-046-081:3352526:3352597 [1] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n111-046-081:3352526:3352597 [1] NCCL INFO TOPO/NET : Importing network plugins to topology +n111-046-081:3352526:3352597 [1] NCCL INFO Retrieving state for FasTrak +n111-046-081:3352526:3352597 [1] NCCL INFO Initialized state 0 for FasTrak +n111-046-081:3352527:3352593 [2] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n111-046-081:3352527:3352593 [2] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n111-046-081:3352526:3352597 [1] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n111-046-081:3352526:3352597 [1] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n111-046-081:3352527:3352593 [2] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n111-046-081:3352527:3352593 [2] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n111-046-081:3352526:3352597 [1] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n111-046-081:3352526:3352597 [1] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n111-046-081:3352526:3352597 [1] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n111-046-081:3352526:3352597 [1] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n111-046-081:3352526:3352597 [1] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n111-046-081:3352526:3352597 [1] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n111-046-081:3352525:3352598 [0] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n111-046-081:3352525:3352598 [0] NCCL INFO TOPO/NET : Importing network plugins to topology +n111-046-081:3352525:3352598 [0] NCCL INFO Retrieving state for FasTrak +n111-046-081:3352525:3352598 [0] NCCL INFO Initialized state 0 for FasTrak +n111-046-081:3352525:3352598 [0] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n111-046-081:3352525:3352598 [0] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n111-046-081:3352525:3352598 [0] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n111-046-081:3352525:3352598 [0] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n111-046-081:3352525:3352598 [0] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n111-046-081:3352525:3352598 [0] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n111-046-081:3352525:3352598 [0] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n111-046-081:3352525:3352598 [0] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n111-046-081:3352532:3352599 [7] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n111-046-081:3352530:3352595 [5] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n111-046-081:3352527:3352593 [2] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n111-046-081:3352531:3352596 [6] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n111-046-081:3352525:3352598 [0] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n111-046-081:3352528:3352594 [3] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n111-046-081:3352529:3352600 [4] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n111-046-081:3352526:3352597 [1] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n111-046-081:3352527:3352593 [2] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n111-046-081:3352527:3352593 [2] NCCL INFO CPU/0-0 (1/1/3) +n111-046-081:3352527:3352593 [2] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n111-046-081:3352527:3352593 [2] NCCL INFO + PCI[24.0] - GPU/0-4000 (24) +n111-046-081:3352527:3352593 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352527:3352593 [2] NCCL INFO + PCI[24.0] - NIC/0-6000 +n111-046-081:3352527:3352593 [2] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n111-046-081:3352527:3352593 [2] NCCL INFO + PCI[24.0] - NIC/0-7000 +n111-046-081:3352527:3352593 [2] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n111-046-081:3352527:3352593 [2] NCCL INFO + PCI[24.0] - GPU/0-5000 (25) +n111-046-081:3352527:3352593 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352527:3352593 [2] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n111-046-081:3352527:3352593 [2] NCCL INFO + PCI[24.0] - NIC/0-d000 +n111-046-081:3352527:3352593 [2] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n111-046-081:3352527:3352593 [2] NCCL INFO + PCI[24.0] - NIC/0-e000 +n111-046-081:3352527:3352593 [2] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n111-046-081:3352527:3352593 [2] NCCL INFO + PCI[24.0] - GPU/0-b000 (26) +n111-046-081:3352527:3352593 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352527:3352593 [2] NCCL INFO + PCI[24.0] - GPU/0-c000 (27) +n111-046-081:3352527:3352593 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352527:3352593 [2] NCCL INFO + SYS[22.0] - CPU/0-1 +n111-046-081:3352527:3352593 [2] NCCL INFO CPU/0-1 (1/1/3) +n111-046-081:3352527:3352593 [2] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n111-046-081:3352527:3352593 [2] NCCL INFO + PCI[24.0] - NIC/0-86000 +n111-046-081:3352527:3352593 [2] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n111-046-081:3352527:3352593 [2] NCCL INFO + PCI[24.0] - NIC/0-87000 +n111-046-081:3352527:3352593 [2] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n111-046-081:3352527:3352593 [2] NCCL INFO + PCI[24.0] - GPU/0-84000 (28) +n111-046-081:3352527:3352593 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352527:3352593 [2] NCCL INFO + PCI[24.0] - GPU/0-85000 (29) +n111-046-081:3352530:3352595 [5] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n111-046-081:3352527:3352593 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352527:3352593 [2] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n111-046-081:3352530:3352595 [5] NCCL INFO CPU/0-0 (1/1/3) +n111-046-081:3352532:3352599 [7] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n111-046-081:3352527:3352593 [2] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n111-046-081:3352527:3352593 [2] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n111-046-081:3352532:3352599 [7] NCCL INFO CPU/0-0 (1/1/3) +n111-046-081:3352527:3352593 [2] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n111-046-081:3352532:3352599 [7] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n111-046-081:3352527:3352593 [2] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n111-046-081:3352530:3352595 [5] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n111-046-081:3352527:3352593 [2] NCCL INFO + PCI[24.0] - GPU/0-8b000 (30) +n111-046-081:3352532:3352599 [7] NCCL INFO + PCI[24.0] - GPU/0-4000 (24) +n111-046-081:3352527:3352593 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352530:3352595 [5] NCCL INFO + PCI[24.0] - GPU/0-4000 (24) +n111-046-081:3352532:3352599 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352527:3352593 [2] NCCL INFO + PCI[24.0] - GPU/0-8c000 (31) +n111-046-081:3352530:3352595 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352532:3352599 [7] NCCL INFO + PCI[24.0] - NIC/0-6000 +n111-046-081:3352527:3352593 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352530:3352595 [5] NCCL INFO + PCI[24.0] - NIC/0-6000 +n111-046-081:3352527:3352593 [2] NCCL INFO + SYS[22.0] - CPU/0-0 +n111-046-081:3352532:3352599 [7] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n111-046-081:3352530:3352595 [5] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n111-046-081:3352527:3352593 [2] NCCL INFO ========================================== +n111-046-081:3352532:3352599 [7] NCCL INFO + PCI[24.0] - NIC/0-7000 +n111-046-081:3352530:3352595 [5] NCCL INFO + PCI[24.0] - NIC/0-7000 +n111-046-081:3352530:3352595 [5] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n111-046-081:3352532:3352599 [7] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n111-046-081:3352530:3352595 [5] NCCL INFO + PCI[24.0] - GPU/0-5000 (25) +n111-046-081:3352532:3352599 [7] NCCL INFO + PCI[24.0] - GPU/0-5000 (25) +n111-046-081:3352530:3352595 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352532:3352599 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352530:3352595 [5] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n111-046-081:3352532:3352599 [7] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n111-046-081:3352527:3352593 [2] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-081:3352530:3352595 [5] NCCL INFO + PCI[24.0] - NIC/0-d000 +n111-046-081:3352532:3352599 [7] NCCL INFO + PCI[24.0] - NIC/0-d000 +n111-046-081:3352530:3352595 [5] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n111-046-081:3352532:3352599 [7] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n111-046-081:3352530:3352595 [5] NCCL INFO + PCI[24.0] - NIC/0-e000 +n111-046-081:3352532:3352599 [7] NCCL INFO + PCI[24.0] - NIC/0-e000 +n111-046-081:3352530:3352595 [5] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n111-046-081:3352532:3352599 [7] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n111-046-081:3352527:3352593 [2] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-081:3352530:3352595 [5] NCCL INFO + PCI[24.0] - GPU/0-b000 (26) +n111-046-081:3352532:3352599 [7] NCCL INFO + PCI[24.0] - GPU/0-b000 (26) +n111-046-081:3352530:3352595 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352532:3352599 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352530:3352595 [5] NCCL INFO + PCI[24.0] - GPU/0-c000 (27) +n111-046-081:3352532:3352599 [7] NCCL INFO + PCI[24.0] - GPU/0-c000 (27) +n111-046-081:3352532:3352599 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352530:3352595 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352532:3352599 [7] NCCL INFO + SYS[22.0] - CPU/0-1 +n111-046-081:3352527:3352593 [2] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-081:3352530:3352595 [5] NCCL INFO + SYS[22.0] - CPU/0-1 +n111-046-081:3352532:3352599 [7] NCCL INFO CPU/0-1 (1/1/3) +n111-046-081:3352530:3352595 [5] NCCL INFO CPU/0-1 (1/1/3) +n111-046-081:3352532:3352599 [7] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n111-046-081:3352530:3352595 [5] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n111-046-081:3352532:3352599 [7] NCCL INFO + PCI[24.0] - NIC/0-86000 +n111-046-081:3352530:3352595 [5] NCCL INFO + PCI[24.0] - NIC/0-86000 +n111-046-081:3352527:3352593 [2] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-081:3352532:3352599 [7] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n111-046-081:3352530:3352595 [5] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n111-046-081:3352532:3352599 [7] NCCL INFO + PCI[24.0] - NIC/0-87000 +n111-046-081:3352530:3352595 [5] NCCL INFO + PCI[24.0] - NIC/0-87000 +n111-046-081:3352532:3352599 [7] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n111-046-081:3352530:3352595 [5] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n111-046-081:3352532:3352599 [7] NCCL INFO + PCI[24.0] - GPU/0-84000 (28) +n111-046-081:3352530:3352595 [5] NCCL INFO + PCI[24.0] - GPU/0-84000 (28) +n111-046-081:3352532:3352599 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352530:3352595 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352532:3352599 [7] NCCL INFO + PCI[24.0] - GPU/0-85000 (29) +n111-046-081:3352530:3352595 [5] NCCL INFO + PCI[24.0] - GPU/0-85000 (29) +n111-046-081:3352527:3352593 [2] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-081:3352532:3352599 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352530:3352595 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352532:3352599 [7] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n111-046-081:3352530:3352595 [5] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n111-046-081:3352532:3352599 [7] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n111-046-081:3352530:3352595 [5] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n111-046-081:3352532:3352599 [7] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n111-046-081:3352530:3352595 [5] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n111-046-081:3352527:3352593 [2] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-081:3352530:3352595 [5] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n111-046-081:3352532:3352599 [7] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n111-046-081:3352530:3352595 [5] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n111-046-081:3352532:3352599 [7] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n111-046-081:3352530:3352595 [5] NCCL INFO + PCI[24.0] - GPU/0-8b000 (30) +n111-046-081:3352530:3352595 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352532:3352599 [7] NCCL INFO + PCI[24.0] - GPU/0-8b000 (30) +n111-046-081:3352527:3352593 [2] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n111-046-081:3352530:3352595 [5] NCCL INFO + PCI[24.0] - GPU/0-8c000 (31) +n111-046-081:3352532:3352599 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352530:3352595 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352532:3352599 [7] NCCL INFO + PCI[24.0] - GPU/0-8c000 (31) +n111-046-081:3352530:3352595 [5] NCCL INFO + SYS[22.0] - CPU/0-0 +n111-046-081:3352532:3352599 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352527:3352593 [2] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n111-046-081:3352530:3352595 [5] NCCL INFO ========================================== +n111-046-081:3352532:3352599 [7] NCCL INFO + SYS[22.0] - CPU/0-0 +n111-046-081:3352532:3352599 [7] NCCL INFO ========================================== +n111-046-081:3352527:3352593 [2] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-081:3352530:3352595 [5] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-081:3352532:3352599 [7] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-081:3352527:3352593 [2] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-081:3352530:3352595 [5] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-081:3352532:3352599 [7] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-081:3352530:3352595 [5] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-081:3352527:3352593 [2] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-081:3352532:3352599 [7] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-081:3352527:3352593 [2] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-081:3352532:3352599 [7] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-081:3352530:3352595 [5] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-081:3352527:3352593 [2] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n111-046-081:3352532:3352599 [7] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-081:3352530:3352595 [5] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-081:3352527:3352593 [2] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n111-046-081:3352532:3352599 [7] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-081:3352530:3352595 [5] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-081:3352527:3352593 [2] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n111-046-081:3352532:3352599 [7] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n111-046-081:3352530:3352595 [5] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n111-046-081:3352527:3352593 [2] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n111-046-081:3352532:3352599 [7] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n111-046-081:3352527:3352593 [2] NCCL INFO Setting affinity for GPU 2 to 07ffffff,ffffff00,00000000,0007ffff,ffffffff +n111-046-081:3352530:3352595 [5] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n111-046-081:3352527:3352593 [2] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n111-046-081:3352530:3352595 [5] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-081:3352532:3352599 [7] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-081:3352530:3352595 [5] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-081:3352532:3352599 [7] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-081:3352530:3352595 [5] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-081:3352532:3352599 [7] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-081:3352532:3352599 [7] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-081:3352530:3352595 [5] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-081:3352527:3352593 [2] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n111-046-081:3352532:3352599 [7] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n111-046-081:3352530:3352595 [5] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n111-046-081:3352525:3352598 [0] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n111-046-081:3352532:3352599 [7] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n111-046-081:3352525:3352598 [0] NCCL INFO CPU/0-0 (1/1/3) +n111-046-081:3352530:3352595 [5] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n111-046-081:3352525:3352598 [0] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n111-046-081:3352525:3352598 [0] NCCL INFO + PCI[24.0] - GPU/0-4000 (24) +n111-046-081:3352525:3352598 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352525:3352598 [0] NCCL INFO + PCI[24.0] - NIC/0-6000 +n111-046-081:3352532:3352599 [7] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n111-046-081:3352525:3352598 [0] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n111-046-081:3352530:3352595 [5] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n111-046-081:3352525:3352598 [0] NCCL INFO + PCI[24.0] - NIC/0-7000 +n111-046-081:3352525:3352598 [0] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n111-046-081:3352531:3352596 [6] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n111-046-081:3352532:3352599 [7] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n111-046-081:3352525:3352598 [0] NCCL INFO + PCI[24.0] - GPU/0-5000 (25) +n111-046-081:3352530:3352595 [5] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n111-046-081:3352531:3352596 [6] NCCL INFO CPU/0-0 (1/1/3) +n111-046-081:3352525:3352598 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352525:3352598 [0] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n111-046-081:3352532:3352599 [7] NCCL INFO Setting affinity for GPU 7 to 7fff,ffffffff,f0000000,0000007f,ffffffff,fff00000,00000000 +n111-046-081:3352531:3352596 [6] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n111-046-081:3352525:3352598 [0] NCCL INFO + PCI[24.0] - NIC/0-d000 +n111-046-081:3352531:3352596 [6] NCCL INFO + PCI[24.0] - GPU/0-4000 (24) +n111-046-081:3352530:3352595 [5] NCCL INFO Setting affinity for GPU 5 to 7fff,ffffffff,f0000000,0000007f,ffffffff,fff00000,00000000 +n111-046-081:3352525:3352598 [0] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n111-046-081:3352531:3352596 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352525:3352598 [0] NCCL INFO + PCI[24.0] - NIC/0-e000 +n111-046-081:3352531:3352596 [6] NCCL INFO + PCI[24.0] - NIC/0-6000 +n111-046-081:3352525:3352598 [0] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n111-046-081:3352530:3352595 [5] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n111-046-081:3352531:3352596 [6] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n111-046-081:3352525:3352598 [0] NCCL INFO + PCI[24.0] - GPU/0-b000 (26) +n111-046-081:3352531:3352596 [6] NCCL INFO + PCI[24.0] - NIC/0-7000 +n111-046-081:3352531:3352596 [6] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n111-046-081:3352525:3352598 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352531:3352596 [6] NCCL INFO + PCI[24.0] - GPU/0-5000 (25) +n111-046-081:3352525:3352598 [0] NCCL INFO + PCI[24.0] - GPU/0-c000 (27) +n111-046-081:3352531:3352596 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352525:3352598 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352531:3352596 [6] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n111-046-081:3352525:3352598 [0] NCCL INFO + SYS[22.0] - CPU/0-1 +n111-046-081:3352531:3352596 [6] NCCL INFO + PCI[24.0] - NIC/0-d000 +n111-046-081:3352525:3352598 [0] NCCL INFO CPU/0-1 (1/1/3) +n111-046-081:3352531:3352596 [6] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n111-046-081:3352525:3352598 [0] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n111-046-081:3352531:3352596 [6] NCCL INFO + PCI[24.0] - NIC/0-e000 +n111-046-081:3352525:3352598 [0] NCCL INFO + PCI[24.0] - NIC/0-86000 +n111-046-081:3352531:3352596 [6] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n111-046-081:3352528:3352594 [3] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n111-046-081:3352525:3352598 [0] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n111-046-081:3352531:3352596 [6] NCCL INFO + PCI[24.0] - GPU/0-b000 (26) +n111-046-081:3352525:3352598 [0] NCCL INFO + PCI[24.0] - NIC/0-87000 +n111-046-081:3352525:3352598 [0] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n111-046-081:3352528:3352594 [3] NCCL INFO CPU/0-0 (1/1/3) +n111-046-081:3352531:3352596 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352525:3352598 [0] NCCL INFO + PCI[24.0] - GPU/0-84000 (28) +n111-046-081:3352531:3352596 [6] NCCL INFO + PCI[24.0] - GPU/0-c000 (27) +n111-046-081:3352525:3352598 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352530:3352595 [5] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n111-046-081:3352525:3352598 [0] NCCL INFO + PCI[24.0] - GPU/0-85000 (29) +n111-046-081:3352528:3352594 [3] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n111-046-081:3352531:3352596 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352525:3352598 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352528:3352594 [3] NCCL INFO + PCI[24.0] - GPU/0-4000 (24) +n111-046-081:3352531:3352596 [6] NCCL INFO + SYS[22.0] - CPU/0-1 +n111-046-081:3352525:3352598 [0] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n111-046-081:3352528:3352594 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352531:3352596 [6] NCCL INFO CPU/0-1 (1/1/3) +n111-046-081:3352525:3352598 [0] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n111-046-081:3352528:3352594 [3] NCCL INFO + PCI[24.0] - NIC/0-6000 +n111-046-081:3352532:3352599 [7] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n111-046-081:3352531:3352596 [6] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n111-046-081:3352525:3352598 [0] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n111-046-081:3352528:3352594 [3] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n111-046-081:3352531:3352596 [6] NCCL INFO + PCI[24.0] - NIC/0-86000 +n111-046-081:3352525:3352598 [0] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n111-046-081:3352528:3352594 [3] NCCL INFO + PCI[24.0] - NIC/0-7000 +n111-046-081:3352531:3352596 [6] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n111-046-081:3352525:3352598 [0] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n111-046-081:3352528:3352594 [3] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n111-046-081:3352528:3352594 [3] NCCL INFO + PCI[24.0] - GPU/0-5000 (25) +n111-046-081:3352531:3352596 [6] NCCL INFO + PCI[24.0] - NIC/0-87000 +n111-046-081:3352525:3352598 [0] NCCL INFO + PCI[24.0] - GPU/0-8b000 (30) +n111-046-081:3352528:3352594 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352525:3352598 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352531:3352596 [6] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n111-046-081:3352528:3352594 [3] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n111-046-081:3352528:3352594 [3] NCCL INFO + PCI[24.0] - NIC/0-d000 +n111-046-081:3352525:3352598 [0] NCCL INFO + PCI[24.0] - GPU/0-8c000 (31) +n111-046-081:3352525:3352598 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352531:3352596 [6] NCCL INFO + PCI[24.0] - GPU/0-84000 (28) +n111-046-081:3352528:3352594 [3] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n111-046-081:3352525:3352598 [0] NCCL INFO + SYS[22.0] - CPU/0-0 +n111-046-081:3352531:3352596 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352525:3352598 [0] NCCL INFO ========================================== +n111-046-081:3352528:3352594 [3] NCCL INFO + PCI[24.0] - NIC/0-e000 +n111-046-081:3352531:3352596 [6] NCCL INFO + PCI[24.0] - GPU/0-85000 (29) +n111-046-081:3352528:3352594 [3] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n111-046-081:3352531:3352596 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352528:3352594 [3] NCCL INFO + PCI[24.0] - GPU/0-b000 (26) +n111-046-081:3352529:3352600 [4] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n111-046-081:3352531:3352596 [6] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n111-046-081:3352528:3352594 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352528:3352594 [3] NCCL INFO + PCI[24.0] - GPU/0-c000 (27) +n111-046-081:3352531:3352596 [6] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n111-046-081:3352529:3352600 [4] NCCL INFO CPU/0-0 (1/1/3) +n111-046-081:3352525:3352598 [0] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-081:3352528:3352594 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352531:3352596 [6] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n111-046-081:3352532:3352599 [7] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n111-046-081:3352529:3352600 [4] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n111-046-081:3352531:3352596 [6] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n111-046-081:3352528:3352594 [3] NCCL INFO + SYS[22.0] - CPU/0-1 +n111-046-081:3352529:3352600 [4] NCCL INFO + PCI[24.0] - GPU/0-4000 (24) +n111-046-081:3352531:3352596 [6] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n111-046-081:3352528:3352594 [3] NCCL INFO CPU/0-1 (1/1/3) +n111-046-081:3352525:3352598 [0] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-081:3352529:3352600 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352531:3352596 [6] NCCL INFO + PCI[24.0] - GPU/0-8b000 (30) +n111-046-081:3352528:3352594 [3] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n111-046-081:3352529:3352600 [4] NCCL INFO + PCI[24.0] - NIC/0-6000 +n111-046-081:3352531:3352596 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352528:3352594 [3] NCCL INFO + PCI[24.0] - NIC/0-86000 +n111-046-081:3352529:3352600 [4] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n111-046-081:3352525:3352598 [0] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-081:3352528:3352594 [3] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n111-046-081:3352531:3352596 [6] NCCL INFO + PCI[24.0] - GPU/0-8c000 (31) +n111-046-081:3352529:3352600 [4] NCCL INFO + PCI[24.0] - NIC/0-7000 +n111-046-081:3352528:3352594 [3] NCCL INFO + PCI[24.0] - NIC/0-87000 +n111-046-081:3352529:3352600 [4] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n111-046-081:3352531:3352596 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352529:3352600 [4] NCCL INFO + PCI[24.0] - GPU/0-5000 (25) +n111-046-081:3352528:3352594 [3] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n111-046-081:3352531:3352596 [6] NCCL INFO + SYS[22.0] - CPU/0-0 +n111-046-081:3352525:3352598 [0] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-081:3352529:3352600 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352528:3352594 [3] NCCL INFO + PCI[24.0] - GPU/0-84000 (28) +n111-046-081:3352531:3352596 [6] NCCL INFO ========================================== +n111-046-081:3352529:3352600 [4] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n111-046-081:3352528:3352594 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352529:3352600 [4] NCCL INFO + PCI[24.0] - NIC/0-d000 +n111-046-081:3352528:3352594 [3] NCCL INFO + PCI[24.0] - GPU/0-85000 (29) +n111-046-081:3352528:3352594 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352529:3352600 [4] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n111-046-081:3352525:3352598 [0] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-081:3352528:3352594 [3] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n111-046-081:3352529:3352600 [4] NCCL INFO + PCI[24.0] - NIC/0-e000 +n111-046-081:3352531:3352596 [6] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-081:3352528:3352594 [3] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n111-046-081:3352529:3352600 [4] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n111-046-081:3352529:3352600 [4] NCCL INFO + PCI[24.0] - GPU/0-b000 (26) +n111-046-081:3352528:3352594 [3] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n111-046-081:3352529:3352600 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352525:3352598 [0] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-081:3352528:3352594 [3] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n111-046-081:3352528:3352594 [3] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n111-046-081:3352531:3352596 [6] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-081:3352529:3352600 [4] NCCL INFO + PCI[24.0] - GPU/0-c000 (27) +n111-046-081:3352528:3352594 [3] NCCL INFO + PCI[24.0] - GPU/0-8b000 (30) +n111-046-081:3352529:3352600 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352525:3352598 [0] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n111-046-081:3352528:3352594 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352529:3352600 [4] NCCL INFO + SYS[22.0] - CPU/0-1 +n111-046-081:3352528:3352594 [3] NCCL INFO + PCI[24.0] - GPU/0-8c000 (31) +n111-046-081:3352529:3352600 [4] NCCL INFO CPU/0-1 (1/1/3) +n111-046-081:3352528:3352594 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352531:3352596 [6] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-081:3352529:3352600 [4] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n111-046-081:3352528:3352594 [3] NCCL INFO + SYS[22.0] - CPU/0-0 +n111-046-081:3352529:3352600 [4] NCCL INFO + PCI[24.0] - NIC/0-86000 +n111-046-081:3352525:3352598 [0] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n111-046-081:3352528:3352594 [3] NCCL INFO ========================================== +n111-046-081:3352529:3352600 [4] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n111-046-081:3352529:3352600 [4] NCCL INFO + PCI[24.0] - NIC/0-87000 +n111-046-081:3352531:3352596 [6] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-081:3352529:3352600 [4] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n111-046-081:3352529:3352600 [4] NCCL INFO + PCI[24.0] - GPU/0-84000 (28) +n111-046-081:3352525:3352598 [0] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-081:3352529:3352600 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352528:3352594 [3] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-081:3352531:3352596 [6] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-081:3352529:3352600 [4] NCCL INFO + PCI[24.0] - GPU/0-85000 (29) +n111-046-081:3352529:3352600 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352525:3352598 [0] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-081:3352529:3352600 [4] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n111-046-081:3352529:3352600 [4] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n111-046-081:3352528:3352594 [3] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-081:3352531:3352596 [6] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-081:3352529:3352600 [4] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n111-046-081:3352525:3352598 [0] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-081:3352529:3352600 [4] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n111-046-081:3352529:3352600 [4] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n111-046-081:3352529:3352600 [4] NCCL INFO + PCI[24.0] - GPU/0-8b000 (30) +n111-046-081:3352529:3352600 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352528:3352594 [3] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-081:3352531:3352596 [6] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n111-046-081:3352525:3352598 [0] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-081:3352529:3352600 [4] NCCL INFO + PCI[24.0] - GPU/0-8c000 (31) +n111-046-081:3352529:3352600 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352529:3352600 [4] NCCL INFO + SYS[22.0] - CPU/0-0 +n111-046-081:3352529:3352600 [4] NCCL INFO ========================================== +n111-046-081:3352528:3352594 [3] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-081:3352525:3352598 [0] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n111-046-081:3352531:3352596 [6] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n111-046-081:3352529:3352600 [4] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-081:3352528:3352594 [3] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-081:3352531:3352596 [6] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-081:3352525:3352598 [0] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n111-046-081:3352529:3352600 [4] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-081:3352528:3352594 [3] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-081:3352531:3352596 [6] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-081:3352525:3352598 [0] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n111-046-081:3352529:3352600 [4] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-081:3352528:3352594 [3] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n111-046-081:3352525:3352598 [0] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n111-046-081:3352531:3352596 [6] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-081:3352528:3352594 [3] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n111-046-081:3352525:3352598 [0] NCCL INFO Setting affinity for GPU 0 to 07ffffff,ffffff00,00000000,0007ffff,ffffffff +n111-046-081:3352529:3352600 [4] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-081:3352531:3352596 [6] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-081:3352528:3352594 [3] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-081:3352529:3352600 [4] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-081:3352528:3352594 [3] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-081:3352531:3352596 [6] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n111-046-081:3352529:3352600 [4] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-081:3352528:3352594 [3] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-081:3352529:3352600 [4] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n111-046-081:3352528:3352594 [3] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-081:3352531:3352596 [6] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n111-046-081:3352529:3352600 [4] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n111-046-081:3352528:3352594 [3] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n111-046-081:3352529:3352600 [4] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-081:3352528:3352594 [3] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n111-046-081:3352531:3352596 [6] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n111-046-081:3352529:3352600 [4] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-081:3352528:3352594 [3] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n111-046-081:3352525:3352598 [0] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n111-046-081:3352531:3352596 [6] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n111-046-081:3352529:3352600 [4] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-081:3352528:3352594 [3] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n111-046-081:3352529:3352600 [4] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-081:3352531:3352596 [6] NCCL INFO Setting affinity for GPU 6 to 7fff,ffffffff,f0000000,0000007f,ffffffff,fff00000,00000000 +n111-046-081:3352528:3352594 [3] NCCL INFO Setting affinity for GPU 3 to 07ffffff,ffffff00,00000000,0007ffff,ffffffff +n111-046-081:3352529:3352600 [4] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n111-046-081:3352529:3352600 [4] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n111-046-081:3352525:3352598 [0] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n111-046-081:3352529:3352600 [4] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n111-046-081:3352529:3352600 [4] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n111-046-081:3352529:3352600 [4] NCCL INFO Setting affinity for GPU 4 to 7fff,ffffffff,f0000000,0000007f,ffffffff,fff00000,00000000 +n111-046-081:3352531:3352596 [6] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n111-046-081:3352528:3352594 [3] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n111-046-081:3352531:3352596 [6] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n111-046-081:3352529:3352600 [4] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n111-046-081:3352528:3352594 [3] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n111-046-081:3352529:3352600 [4] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n111-046-081:3352526:3352597 [1] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n111-046-081:3352526:3352597 [1] NCCL INFO CPU/0-0 (1/1/3) +n111-046-081:3352526:3352597 [1] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n111-046-081:3352526:3352597 [1] NCCL INFO + PCI[24.0] - GPU/0-4000 (24) +n111-046-081:3352526:3352597 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352526:3352597 [1] NCCL INFO + PCI[24.0] - NIC/0-6000 +n111-046-081:3352526:3352597 [1] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n111-046-081:3352526:3352597 [1] NCCL INFO + PCI[24.0] - NIC/0-7000 +n111-046-081:3352526:3352597 [1] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n111-046-081:3352526:3352597 [1] NCCL INFO + PCI[24.0] - GPU/0-5000 (25) +n111-046-081:3352526:3352597 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352526:3352597 [1] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n111-046-081:3352526:3352597 [1] NCCL INFO + PCI[24.0] - NIC/0-d000 +n111-046-081:3352526:3352597 [1] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n111-046-081:3352526:3352597 [1] NCCL INFO + PCI[24.0] - NIC/0-e000 +n111-046-081:3352526:3352597 [1] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n111-046-081:3352526:3352597 [1] NCCL INFO + PCI[24.0] - GPU/0-b000 (26) +n111-046-081:3352526:3352597 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352526:3352597 [1] NCCL INFO + PCI[24.0] - GPU/0-c000 (27) +n111-046-081:3352526:3352597 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352526:3352597 [1] NCCL INFO + SYS[22.0] - CPU/0-1 +n111-046-081:3352526:3352597 [1] NCCL INFO CPU/0-1 (1/1/3) +n111-046-081:3352526:3352597 [1] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n111-046-081:3352526:3352597 [1] NCCL INFO + PCI[24.0] - NIC/0-86000 +n111-046-081:3352526:3352597 [1] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n111-046-081:3352526:3352597 [1] NCCL INFO + PCI[24.0] - NIC/0-87000 +n111-046-081:3352526:3352597 [1] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n111-046-081:3352526:3352597 [1] NCCL INFO + PCI[24.0] - GPU/0-84000 (28) +n111-046-081:3352526:3352597 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352526:3352597 [1] NCCL INFO + PCI[24.0] - GPU/0-85000 (29) +n111-046-081:3352526:3352597 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352526:3352597 [1] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n111-046-081:3352526:3352597 [1] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n111-046-081:3352526:3352597 [1] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n111-046-081:3352526:3352597 [1] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n111-046-081:3352526:3352597 [1] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n111-046-081:3352526:3352597 [1] NCCL INFO + PCI[24.0] - GPU/0-8b000 (30) +n111-046-081:3352526:3352597 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352526:3352597 [1] NCCL INFO + PCI[24.0] - GPU/0-8c000 (31) +n111-046-081:3352526:3352597 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n111-046-081:3352526:3352597 [1] NCCL INFO + SYS[22.0] - CPU/0-0 +n111-046-081:3352526:3352597 [1] NCCL INFO ========================================== +n111-046-081:3352526:3352597 [1] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-081:3352526:3352597 [1] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-081:3352526:3352597 [1] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-081:3352526:3352597 [1] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-081:3352526:3352597 [1] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-081:3352526:3352597 [1] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n111-046-081:3352526:3352597 [1] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n111-046-081:3352526:3352597 [1] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n111-046-081:3352526:3352597 [1] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-081:3352526:3352597 [1] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-081:3352526:3352597 [1] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-081:3352526:3352597 [1] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n111-046-081:3352526:3352597 [1] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n111-046-081:3352526:3352597 [1] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n111-046-081:3352526:3352597 [1] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n111-046-081:3352526:3352597 [1] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n111-046-081:3352526:3352597 [1] NCCL INFO Setting affinity for GPU 1 to 07ffffff,ffffff00,00000000,0007ffff,ffffffff +n111-046-081:3352526:3352597 [1] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n111-046-081:3352526:3352597 [1] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n111-046-081:3352530:3352595 [5] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n111-046-081:3352530:3352595 [5] NCCL INFO 0 : NET/0-0 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 NET/0-0 +n111-046-081:3352530:3352595 [5] NCCL INFO 1 : NET/0-1 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 NET/0-1 +n111-046-081:3352530:3352595 [5] NCCL INFO 2 : NET/0-2 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 NET/0-2 +n111-046-081:3352530:3352595 [5] NCCL INFO 3 : NET/0-3 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 NET/0-3 +n111-046-081:3352530:3352595 [5] NCCL INFO 4 : NET/0-4 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 NET/0-4 +n111-046-081:3352530:3352595 [5] NCCL INFO 5 : NET/0-5 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 NET/0-5 +n111-046-081:3352530:3352595 [5] NCCL INFO 6 : NET/0-6 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 NET/0-6 +n111-046-081:3352530:3352595 [5] NCCL INFO 7 : NET/0-7 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 NET/0-7 +n111-046-081:3352530:3352595 [5] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n111-046-081:3352530:3352595 [5] NCCL INFO 0 : NET/0-0 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 NET/0-0 +n111-046-081:3352530:3352595 [5] NCCL INFO 1 : NET/0-1 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 NET/0-1 +n111-046-081:3352530:3352595 [5] NCCL INFO 2 : NET/0-2 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 NET/0-2 +n111-046-081:3352530:3352595 [5] NCCL INFO 3 : NET/0-3 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 NET/0-3 +n111-046-081:3352530:3352595 [5] NCCL INFO 4 : NET/0-4 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 NET/0-4 +n111-046-081:3352530:3352595 [5] NCCL INFO 5 : NET/0-5 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 NET/0-5 +n111-046-081:3352530:3352595 [5] NCCL INFO 6 : NET/0-6 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 NET/0-6 +n111-046-081:3352530:3352595 [5] NCCL INFO 7 : NET/0-7 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 NET/0-7 +n111-046-081:3352531:3352596 [6] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n111-046-081:3352531:3352596 [6] NCCL INFO 0 : NET/0-0 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 NET/0-0 +n111-046-081:3352531:3352596 [6] NCCL INFO 1 : NET/0-1 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 NET/0-1 +n111-046-081:3352531:3352596 [6] NCCL INFO 2 : NET/0-2 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 NET/0-2 +n111-046-081:3352531:3352596 [6] NCCL INFO 3 : NET/0-3 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 NET/0-3 +n111-046-081:3352531:3352596 [6] NCCL INFO 4 : NET/0-4 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 NET/0-4 +n111-046-081:3352531:3352596 [6] NCCL INFO 5 : NET/0-5 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 NET/0-5 +n111-046-081:3352531:3352596 [6] NCCL INFO 6 : NET/0-6 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 NET/0-6 +n111-046-081:3352531:3352596 [6] NCCL INFO 7 : NET/0-7 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 NET/0-7 +n111-046-081:3352531:3352596 [6] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n111-046-081:3352531:3352596 [6] NCCL INFO 0 : NET/0-0 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 NET/0-0 +n111-046-081:3352531:3352596 [6] NCCL INFO 1 : NET/0-1 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 NET/0-1 +n111-046-081:3352531:3352596 [6] NCCL INFO 2 : NET/0-2 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 NET/0-2 +n111-046-081:3352531:3352596 [6] NCCL INFO 3 : NET/0-3 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 NET/0-3 +n111-046-081:3352531:3352596 [6] NCCL INFO 4 : NET/0-4 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 NET/0-4 +n111-046-081:3352531:3352596 [6] NCCL INFO 5 : NET/0-5 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 NET/0-5 +n111-046-081:3352531:3352596 [6] NCCL INFO 6 : NET/0-6 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 NET/0-6 +n111-046-081:3352531:3352596 [6] NCCL INFO 7 : NET/0-7 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 NET/0-7 +n111-046-081:3352529:3352600 [4] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n111-046-081:3352529:3352600 [4] NCCL INFO 0 : NET/0-0 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 NET/0-0 +n111-046-081:3352529:3352600 [4] NCCL INFO 1 : NET/0-1 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 NET/0-1 +n111-046-081:3352529:3352600 [4] NCCL INFO 2 : NET/0-2 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 NET/0-2 +n111-046-081:3352529:3352600 [4] NCCL INFO 3 : NET/0-3 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 NET/0-3 +n111-046-081:3352529:3352600 [4] NCCL INFO 4 : NET/0-4 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 NET/0-4 +n111-046-081:3352529:3352600 [4] NCCL INFO 5 : NET/0-5 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 NET/0-5 +n111-046-081:3352529:3352600 [4] NCCL INFO 6 : NET/0-6 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 NET/0-6 +n111-046-081:3352529:3352600 [4] NCCL INFO 7 : NET/0-7 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 NET/0-7 +n111-046-081:3352529:3352600 [4] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n111-046-081:3352529:3352600 [4] NCCL INFO 0 : NET/0-0 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 NET/0-0 +n111-046-081:3352529:3352600 [4] NCCL INFO 1 : NET/0-1 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 NET/0-1 +n111-046-081:3352529:3352600 [4] NCCL INFO 2 : NET/0-2 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 NET/0-2 +n111-046-081:3352529:3352600 [4] NCCL INFO 3 : NET/0-3 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 NET/0-3 +n111-046-081:3352529:3352600 [4] NCCL INFO 4 : NET/0-4 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 NET/0-4 +n111-046-081:3352529:3352600 [4] NCCL INFO 5 : NET/0-5 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 NET/0-5 +n111-046-081:3352529:3352600 [4] NCCL INFO 6 : NET/0-6 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 NET/0-6 +n111-046-081:3352529:3352600 [4] NCCL INFO 7 : NET/0-7 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 NET/0-7 +n111-046-081:3352527:3352593 [2] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n111-046-081:3352527:3352593 [2] NCCL INFO 0 : NET/0-0 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 NET/0-0 +n111-046-081:3352527:3352593 [2] NCCL INFO 1 : NET/0-1 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 NET/0-1 +n111-046-081:3352527:3352593 [2] NCCL INFO 2 : NET/0-2 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 NET/0-2 +n111-046-081:3352527:3352593 [2] NCCL INFO 3 : NET/0-3 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 NET/0-3 +n111-046-081:3352527:3352593 [2] NCCL INFO 4 : NET/0-4 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 NET/0-4 +n111-046-081:3352527:3352593 [2] NCCL INFO 5 : NET/0-5 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 NET/0-5 +n111-046-081:3352527:3352593 [2] NCCL INFO 6 : NET/0-6 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 NET/0-6 +n111-046-081:3352527:3352593 [2] NCCL INFO 7 : NET/0-7 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 NET/0-7 +n111-046-081:3352527:3352593 [2] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n111-046-081:3352527:3352593 [2] NCCL INFO 0 : NET/0-0 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 NET/0-0 +n111-046-081:3352527:3352593 [2] NCCL INFO 1 : NET/0-1 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 NET/0-1 +n111-046-081:3352527:3352593 [2] NCCL INFO 2 : NET/0-2 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 NET/0-2 +n111-046-081:3352527:3352593 [2] NCCL INFO 3 : NET/0-3 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 NET/0-3 +n111-046-081:3352527:3352593 [2] NCCL INFO 4 : NET/0-4 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 NET/0-4 +n111-046-081:3352527:3352593 [2] NCCL INFO 5 : NET/0-5 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 NET/0-5 +n111-046-081:3352527:3352593 [2] NCCL INFO 6 : NET/0-6 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 NET/0-6 +n111-046-081:3352527:3352593 [2] NCCL INFO 7 : NET/0-7 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 NET/0-7 +n111-046-081:3352532:3352599 [7] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n111-046-081:3352532:3352599 [7] NCCL INFO 0 : NET/0-0 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 NET/0-0 +n111-046-081:3352532:3352599 [7] NCCL INFO 1 : NET/0-1 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 NET/0-1 +n111-046-081:3352532:3352599 [7] NCCL INFO 2 : NET/0-2 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 NET/0-2 +n111-046-081:3352532:3352599 [7] NCCL INFO 3 : NET/0-3 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 NET/0-3 +n111-046-081:3352532:3352599 [7] NCCL INFO 4 : NET/0-4 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 NET/0-4 +n111-046-081:3352532:3352599 [7] NCCL INFO 5 : NET/0-5 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 NET/0-5 +n111-046-081:3352532:3352599 [7] NCCL INFO 6 : NET/0-6 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 NET/0-6 +n111-046-081:3352532:3352599 [7] NCCL INFO 7 : NET/0-7 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 NET/0-7 +n111-046-081:3352532:3352599 [7] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n111-046-081:3352532:3352599 [7] NCCL INFO 0 : NET/0-0 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 NET/0-0 +n111-046-081:3352532:3352599 [7] NCCL INFO 1 : NET/0-1 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 NET/0-1 +n111-046-081:3352532:3352599 [7] NCCL INFO 2 : NET/0-2 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 NET/0-2 +n111-046-081:3352532:3352599 [7] NCCL INFO 3 : NET/0-3 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 NET/0-3 +n111-046-081:3352532:3352599 [7] NCCL INFO 4 : NET/0-4 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 NET/0-4 +n111-046-081:3352532:3352599 [7] NCCL INFO 5 : NET/0-5 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 NET/0-5 +n111-046-081:3352532:3352599 [7] NCCL INFO 6 : NET/0-6 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 NET/0-6 +n111-046-081:3352532:3352599 [7] NCCL INFO 7 : NET/0-7 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 NET/0-7 +n111-046-081:3352526:3352597 [1] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n111-046-081:3352526:3352597 [1] NCCL INFO 0 : NET/0-0 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 NET/0-0 +n111-046-081:3352526:3352597 [1] NCCL INFO 1 : NET/0-1 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 NET/0-1 +n111-046-081:3352526:3352597 [1] NCCL INFO 2 : NET/0-2 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 NET/0-2 +n111-046-081:3352526:3352597 [1] NCCL INFO 3 : NET/0-3 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 NET/0-3 +n111-046-081:3352526:3352597 [1] NCCL INFO 4 : NET/0-4 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 NET/0-4 +n111-046-081:3352526:3352597 [1] NCCL INFO 5 : NET/0-5 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 NET/0-5 +n111-046-081:3352526:3352597 [1] NCCL INFO 6 : NET/0-6 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 NET/0-6 +n111-046-081:3352526:3352597 [1] NCCL INFO 7 : NET/0-7 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 NET/0-7 +n111-046-081:3352526:3352597 [1] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n111-046-081:3352526:3352597 [1] NCCL INFO 0 : NET/0-0 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 NET/0-0 +n111-046-081:3352526:3352597 [1] NCCL INFO 1 : NET/0-1 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 NET/0-1 +n111-046-081:3352526:3352597 [1] NCCL INFO 2 : NET/0-2 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 NET/0-2 +n111-046-081:3352526:3352597 [1] NCCL INFO 3 : NET/0-3 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 NET/0-3 +n111-046-081:3352526:3352597 [1] NCCL INFO 4 : NET/0-4 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 NET/0-4 +n111-046-081:3352526:3352597 [1] NCCL INFO 5 : NET/0-5 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 NET/0-5 +n111-046-081:3352526:3352597 [1] NCCL INFO 6 : NET/0-6 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 NET/0-6 +n111-046-081:3352526:3352597 [1] NCCL INFO 7 : NET/0-7 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 NET/0-7 +n111-046-081:3352528:3352594 [3] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n111-046-081:3352528:3352594 [3] NCCL INFO 0 : NET/0-0 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 NET/0-0 +n111-046-081:3352528:3352594 [3] NCCL INFO 1 : NET/0-1 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 NET/0-1 +n111-046-081:3352528:3352594 [3] NCCL INFO 2 : NET/0-2 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 NET/0-2 +n111-046-081:3352528:3352594 [3] NCCL INFO 3 : NET/0-3 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 NET/0-3 +n111-046-081:3352528:3352594 [3] NCCL INFO 4 : NET/0-4 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 NET/0-4 +n111-046-081:3352528:3352594 [3] NCCL INFO 5 : NET/0-5 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 NET/0-5 +n111-046-081:3352528:3352594 [3] NCCL INFO 6 : NET/0-6 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 NET/0-6 +n111-046-081:3352528:3352594 [3] NCCL INFO 7 : NET/0-7 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 NET/0-7 +n111-046-081:3352525:3352598 [0] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n111-046-081:3352525:3352598 [0] NCCL INFO 0 : NET/0-0 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 NET/0-0 +n111-046-081:3352525:3352598 [0] NCCL INFO 1 : NET/0-1 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 NET/0-1 +n111-046-081:3352525:3352598 [0] NCCL INFO 2 : NET/0-2 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 NET/0-2 +n111-046-081:3352525:3352598 [0] NCCL INFO 3 : NET/0-3 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 NET/0-3 +n111-046-081:3352525:3352598 [0] NCCL INFO 4 : NET/0-4 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 NET/0-4 +n111-046-081:3352525:3352598 [0] NCCL INFO 5 : NET/0-5 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 NET/0-5 +n111-046-081:3352525:3352598 [0] NCCL INFO 6 : NET/0-6 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 NET/0-6 +n111-046-081:3352525:3352598 [0] NCCL INFO 7 : NET/0-7 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 NET/0-7 +n111-046-081:3352528:3352594 [3] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n111-046-081:3352528:3352594 [3] NCCL INFO 0 : NET/0-0 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 NET/0-0 +n111-046-081:3352528:3352594 [3] NCCL INFO 1 : NET/0-1 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 NET/0-1 +n111-046-081:3352528:3352594 [3] NCCL INFO 2 : NET/0-2 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 NET/0-2 +n111-046-081:3352528:3352594 [3] NCCL INFO 3 : NET/0-3 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 NET/0-3 +n111-046-081:3352528:3352594 [3] NCCL INFO 4 : NET/0-4 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 NET/0-4 +n111-046-081:3352528:3352594 [3] NCCL INFO 5 : NET/0-5 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 NET/0-5 +n111-046-081:3352528:3352594 [3] NCCL INFO 6 : NET/0-6 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 NET/0-6 +n111-046-081:3352528:3352594 [3] NCCL INFO 7 : NET/0-7 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 NET/0-7 +n111-046-081:3352525:3352598 [0] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n111-046-081:3352525:3352598 [0] NCCL INFO 0 : NET/0-0 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 NET/0-0 +n111-046-081:3352525:3352598 [0] NCCL INFO 1 : NET/0-1 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 NET/0-1 +n111-046-081:3352525:3352598 [0] NCCL INFO 2 : NET/0-2 GPU/26 GPU/27 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 NET/0-2 +n111-046-081:3352525:3352598 [0] NCCL INFO 3 : NET/0-3 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 GPU/29 GPU/28 NET/0-3 +n111-046-081:3352525:3352598 [0] NCCL INFO 4 : NET/0-4 GPU/28 GPU/29 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 NET/0-4 +n111-046-081:3352525:3352598 [0] NCCL INFO 5 : NET/0-5 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 GPU/31 GPU/30 NET/0-5 +n111-046-081:3352525:3352598 [0] NCCL INFO 6 : NET/0-6 GPU/30 GPU/31 GPU/24 GPU/25 GPU/26 GPU/27 GPU/28 GPU/29 NET/0-6 +n111-046-081:3352525:3352598 [0] NCCL INFO 7 : NET/0-7 GPU/31 GPU/30 GPU/29 GPU/28 GPU/27 GPU/26 GPU/25 GPU/24 NET/0-7 +n111-046-081:3352532:3352599 [7] NCCL INFO comm 0xcc43090 rank 31 nRanks 64 nNodes 8 localRanks 8 localRank 7 MNNVL 0 +n111-046-081:3352531:3352596 [6] NCCL INFO comm 0xb2b9f40 rank 30 nRanks 64 nNodes 8 localRanks 8 localRank 6 MNNVL 0 +n111-046-081:3352530:3352595 [5] NCCL INFO comm 0xc14dc20 rank 29 nRanks 64 nNodes 8 localRanks 8 localRank 5 MNNVL 0 +n111-046-081:3352529:3352600 [4] NCCL INFO comm 0xc569f60 rank 28 nRanks 64 nNodes 8 localRanks 8 localRank 4 MNNVL 0 +n111-046-081:3352531:3352596 [6] NCCL INFO Tree 6 : 22 -> 30 -> 31/-1/-1 +n111-046-081:3352531:3352596 [6] NCCL INFO Tree 14 : 62 -> 30 -> 31/14/-1 +n111-046-081:3352528:3352594 [3] NCCL INFO comm 0xcd5e030 rank 27 nRanks 64 nNodes 8 localRanks 8 localRank 3 MNNVL 0 +n111-046-081:3352531:3352596 [6] NCCL INFO Tree 7 : 31 -> 30 -> 29/-1/-1 +n111-046-081:3352532:3352599 [7] NCCL INFO Tree 6 : 30 -> 31 -> 24/-1/-1 +n111-046-081:3352532:3352599 [7] NCCL INFO Tree 14 : 30 -> 31 -> 24/46/-1 +n111-046-081:3352531:3352596 [6] NCCL INFO Tree 15 : 31 -> 30 -> 29/47/-1 +n111-046-081:3352532:3352599 [7] NCCL INFO Tree 7 : 23 -> 31 -> 30/-1/-1 +n111-046-081:3352530:3352595 [5] NCCL INFO Tree 4 : 28 -> 29 -> 30/-1/-1 +n111-046-081:3352532:3352599 [7] NCCL INFO Tree 15 : 63 -> 31 -> 30/15/-1 +n111-046-081:3352530:3352595 [5] NCCL INFO Tree 12 : 28 -> 29 -> 30/44/-1 +n111-046-081:3352531:3352596 [6] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n111-046-081:3352530:3352595 [5] NCCL INFO Tree 5 : 21 -> 29 -> 28/-1/-1 +n111-046-081:3352527:3352593 [2] NCCL INFO comm 0xbfe2f00 rank 26 nRanks 64 nNodes 8 localRanks 8 localRank 2 MNNVL 0 +n111-046-081:3352530:3352595 [5] NCCL INFO Tree 13 : 61 -> 29 -> 28/13/-1 +n111-046-081:3352532:3352599 [7] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n111-046-081:3352530:3352595 [5] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n111-046-081:3352529:3352600 [4] NCCL INFO Tree 4 : 20 -> 28 -> 29/-1/-1 +n111-046-081:3352529:3352600 [4] NCCL INFO Tree 12 : 60 -> 28 -> 29/12/-1 +n111-046-081:3352529:3352600 [4] NCCL INFO Tree 5 : 29 -> 28 -> 27/-1/-1 +n111-046-081:3352529:3352600 [4] NCCL INFO Tree 13 : 29 -> 28 -> 27/45/-1 +n111-046-081:3352529:3352600 [4] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n111-046-081:3352526:3352597 [1] NCCL INFO comm 0xbdbfa40 rank 25 nRanks 64 nNodes 8 localRanks 8 localRank 1 MNNVL 0 +n111-046-081:3352527:3352593 [2] NCCL INFO Tree 2 : 18 -> 26 -> 27/-1/-1 +n111-046-081:3352528:3352594 [3] NCCL INFO Tree 2 : 26 -> 27 -> 28/-1/-1 +n111-046-081:3352527:3352593 [2] NCCL INFO Tree 10 : 58 -> 26 -> 27/10/-1 +n111-046-081:3352528:3352594 [3] NCCL INFO Tree 10 : 26 -> 27 -> 28/42/-1 +n111-046-081:3352527:3352593 [2] NCCL INFO Tree 3 : 27 -> 26 -> 25/-1/-1 +n111-046-081:3352527:3352593 [2] NCCL INFO Tree 11 : 27 -> 26 -> 25/43/-1 +n111-046-081:3352525:3352598 [0] NCCL INFO comm 0xc925780 rank 24 nRanks 64 nNodes 8 localRanks 8 localRank 0 MNNVL 0 +n111-046-081:3352528:3352594 [3] NCCL INFO Tree 3 : 19 -> 27 -> 26/-1/-1 +n111-046-081:3352528:3352594 [3] NCCL INFO Tree 11 : 59 -> 27 -> 26/11/-1 +n111-046-081:3352531:3352596 [6] NCCL INFO Ring 00 : 31 -> 30 -> 29 +n111-046-081:3352532:3352599 [7] NCCL INFO Ring 00 : 24 -> 31 -> 30 +n111-046-081:3352531:3352596 [6] NCCL INFO Ring 01 : 29 -> 30 -> 31 +n111-046-081:3352527:3352593 [2] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n111-046-081:3352532:3352599 [7] NCCL INFO Ring 01 : 30 -> 31 -> 24 +n111-046-081:3352528:3352594 [3] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n111-046-081:3352531:3352596 [6] NCCL INFO Ring 02 : 31 -> 30 -> 29 +n111-046-081:3352532:3352599 [7] NCCL INFO Ring 02 : 24 -> 31 -> 30 +n111-046-081:3352532:3352599 [7] NCCL INFO Ring 03 : 30 -> 31 -> 24 +n111-046-081:3352531:3352596 [6] NCCL INFO Ring 03 : 29 -> 30 -> 31 +n111-046-081:3352532:3352599 [7] NCCL INFO Ring 04 : 24 -> 31 -> 30 +n111-046-081:3352531:3352596 [6] NCCL INFO Ring 04 : 31 -> 30 -> 29 +n111-046-081:3352532:3352599 [7] NCCL INFO Ring 05 : 30 -> 31 -> 24 +n111-046-081:3352530:3352595 [5] NCCL INFO Ring 00 : 30 -> 29 -> 28 +n111-046-081:3352531:3352596 [6] NCCL INFO Ring 05 : 29 -> 30 -> 31 +n111-046-081:3352532:3352599 [7] NCCL INFO Ring 06 : 24 -> 31 -> 38 +n111-046-081:3352531:3352596 [6] NCCL INFO Ring 06 : 23 -> 30 -> 29 +n111-046-081:3352532:3352599 [7] NCCL INFO Ring 07 : 22 -> 31 -> 24 +n111-046-081:3352530:3352595 [5] NCCL INFO Ring 01 : 28 -> 29 -> 30 +n111-046-081:3352531:3352596 [6] NCCL INFO Ring 07 : 29 -> 30 -> 39 +n111-046-081:3352526:3352597 [1] NCCL INFO Tree 0 : 24 -> 25 -> 26/-1/-1 +n111-046-081:3352532:3352599 [7] NCCL INFO Ring 08 : 24 -> 31 -> 30 +n111-046-081:3352530:3352595 [5] NCCL INFO Ring 02 : 30 -> 29 -> 28 +n111-046-081:3352531:3352596 [6] NCCL INFO Ring 08 : 31 -> 30 -> 29 +n111-046-081:3352531:3352596 [6] NCCL INFO Ring 09 : 29 -> 30 -> 31 +n111-046-081:3352526:3352597 [1] NCCL INFO Tree 8 : 24 -> 25 -> 26/40/-1 +n111-046-081:3352531:3352596 [6] NCCL INFO Ring 10 : 31 -> 30 -> 29 +n111-046-081:3352532:3352599 [7] NCCL INFO Ring 09 : 30 -> 31 -> 24 +n111-046-081:3352530:3352595 [5] NCCL INFO Ring 03 : 28 -> 29 -> 30 +n111-046-081:3352525:3352598 [0] NCCL INFO Tree 0 : 16 -> 24 -> 25/-1/-1 +n111-046-081:3352526:3352597 [1] NCCL INFO Tree 1 : 17 -> 25 -> 24/-1/-1 +n111-046-081:3352525:3352598 [0] NCCL INFO Tree 8 : 56 -> 24 -> 25/8/-1 +n111-046-081:3352531:3352596 [6] NCCL INFO Ring 11 : 29 -> 30 -> 31 +n111-046-081:3352532:3352599 [7] NCCL INFO Ring 10 : 24 -> 31 -> 30 +n111-046-081:3352530:3352595 [5] NCCL INFO Ring 04 : 30 -> 29 -> 36 +n111-046-081:3352532:3352599 [7] NCCL INFO Ring 11 : 30 -> 31 -> 24 +n111-046-081:3352526:3352597 [1] NCCL INFO Tree 9 : 57 -> 25 -> 24/9/-1 +n111-046-081:3352531:3352596 [6] NCCL INFO Ring 12 : 31 -> 30 -> 29 +n111-046-081:3352529:3352600 [4] NCCL INFO Ring 00 : 29 -> 28 -> 27 +n111-046-081:3352525:3352598 [0] NCCL INFO Tree 1 : 25 -> 24 -> 31/-1/-1 +n111-046-081:3352527:3352593 [2] NCCL INFO Ring 00 : 27 -> 26 -> 25 +n111-046-081:3352530:3352595 [5] NCCL INFO Ring 05 : 20 -> 29 -> 30 +n111-046-081:3352532:3352599 [7] NCCL INFO Ring 12 : 24 -> 31 -> 30 +n111-046-081:3352531:3352596 [6] NCCL INFO Ring 13 : 29 -> 30 -> 31 +n111-046-081:3352526:3352597 [1] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n111-046-081:3352532:3352599 [7] NCCL INFO Ring 13 : 30 -> 31 -> 24 +n111-046-081:3352531:3352596 [6] NCCL INFO Ring 14 : 23 -> 30 -> 29 +n111-046-081:3352529:3352600 [4] NCCL INFO Ring 01 : 27 -> 28 -> 29 +n111-046-081:3352525:3352598 [0] NCCL INFO Tree 9 : 25 -> 24 -> 31/41/-1 +n111-046-081:3352529:3352600 [4] NCCL INFO Ring 02 : 29 -> 28 -> 27 +n111-046-081:3352530:3352595 [5] NCCL INFO Ring 06 : 30 -> 29 -> 28 +n111-046-081:3352527:3352593 [2] NCCL INFO Ring 01 : 25 -> 26 -> 27 +n111-046-081:3352528:3352594 [3] NCCL INFO Ring 00 : 28 -> 27 -> 26 +n111-046-081:3352531:3352596 [6] NCCL INFO Ring 15 : 29 -> 30 -> 39 +n111-046-081:3352532:3352599 [7] NCCL INFO Ring 14 : 24 -> 31 -> 38 +n111-046-081:3352529:3352600 [4] NCCL INFO Ring 03 : 27 -> 28 -> 29 +n111-046-081:3352530:3352595 [5] NCCL INFO Ring 07 : 28 -> 29 -> 30 +n111-046-081:3352529:3352600 [4] NCCL INFO Ring 04 : 21 -> 28 -> 27 +n111-046-081:3352525:3352598 [0] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n111-046-081:3352527:3352593 [2] NCCL INFO Ring 02 : 19 -> 26 -> 25 +n111-046-081:3352531:3352596 [6] NCCL INFO Trees [0] 31/-1/-1->30->29 [1] 29/-1/-1->30->31 [2] 31/-1/-1->30->29 [3] 29/-1/-1->30->31 [4] 31/-1/-1->30->29 [5] -1/-1/-1->30->31 [6] 31/-1/-1->30->22 [7] 29/-1/-1->30->31 [8] 31/-1/-1->30->29 [9] 29/-1/-1->30->31 [10] 31/-1/-1->30->29 [11] 29/-1/-1->30->31 [12] 31/-1/-1->30->29 [13] -1/-1/-1->30->31 [14] 31/14/-1->30->62 [15] 29/47/-1->30->31 +n111-046-081:3352532:3352599 [7] NCCL INFO Ring 15 : 22 -> 31 -> 24 +n111-046-081:3352528:3352594 [3] NCCL INFO Ring 01 : 26 -> 27 -> 28 +n111-046-081:3352530:3352595 [5] NCCL INFO Ring 08 : 30 -> 29 -> 28 +n111-046-081:3352530:3352595 [5] NCCL INFO Ring 09 : 28 -> 29 -> 30 +n111-046-081:3352529:3352600 [4] NCCL INFO Ring 05 : 27 -> 28 -> 37 +n111-046-081:3352529:3352600 [4] NCCL INFO Ring 06 : 29 -> 28 -> 27 +n111-046-081:3352527:3352593 [2] NCCL INFO Ring 03 : 25 -> 26 -> 35 +n111-046-081:3352532:3352599 [7] NCCL INFO Trees [0] -1/-1/-1->31->30 [1] 30/-1/-1->31->24 [2] 24/-1/-1->31->30 [3] 30/-1/-1->31->24 [4] 24/-1/-1->31->30 [5] 30/-1/-1->31->24 [6] 24/-1/-1->31->30 [7] 30/-1/-1->31->23 [8] -1/-1/-1->31->30 [9] 30/-1/-1->31->24 [10] 24/-1/-1->31->30 [11] 30/-1/-1->31->24 [12] 24/-1/-1->31->30 [13] 30/-1/-1->31->24 [14] 24/46/-1->31->30 [15] 30/15/-1->31->63 +n111-046-081:3352529:3352600 [4] NCCL INFO Ring 07 : 27 -> 28 -> 29 +n111-046-081:3352531:3352596 [6] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n111-046-081:3352528:3352594 [3] NCCL INFO Ring 02 : 28 -> 27 -> 34 +n111-046-081:3352532:3352599 [7] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n111-046-081:3352532:3352599 [7] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n111-046-081:3352530:3352595 [5] NCCL INFO Ring 10 : 30 -> 29 -> 28 +n111-046-081:3352529:3352600 [4] NCCL INFO Ring 08 : 29 -> 28 -> 27 +n111-046-081:3352527:3352593 [2] NCCL INFO Ring 04 : 27 -> 26 -> 25 +n111-046-081:3352531:3352596 [6] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n111-046-081:3352529:3352600 [4] NCCL INFO Ring 09 : 27 -> 28 -> 29 +n111-046-081:3352528:3352594 [3] NCCL INFO Ring 03 : 18 -> 27 -> 28 +n111-046-081:3352532:3352599 [7] NCCL INFO P2P Chunksize set to 524288 +n111-046-081:3352530:3352595 [5] NCCL INFO Ring 11 : 28 -> 29 -> 30 +n111-046-081:3352531:3352596 [6] NCCL INFO P2P Chunksize set to 524288 +n111-046-081:3352528:3352594 [3] NCCL INFO Ring 04 : 28 -> 27 -> 26 +n111-046-081:3352527:3352593 [2] NCCL INFO Ring 05 : 25 -> 26 -> 27 +n111-046-081:3352529:3352600 [4] NCCL INFO Ring 10 : 29 -> 28 -> 27 +n111-046-081:3352530:3352595 [5] NCCL INFO Ring 12 : 30 -> 29 -> 36 +n111-046-081:3352526:3352597 [1] NCCL INFO Ring 00 : 26 -> 25 -> 32 +n111-046-081:3352528:3352594 [3] NCCL INFO Ring 05 : 26 -> 27 -> 28 +n111-046-081:3352527:3352593 [2] NCCL INFO Ring 06 : 27 -> 26 -> 25 +n111-046-081:3352529:3352600 [4] NCCL INFO Ring 11 : 27 -> 28 -> 29 +n111-046-081:3352530:3352595 [5] NCCL INFO Ring 13 : 20 -> 29 -> 30 +n111-046-081:3352526:3352597 [1] NCCL INFO Ring 01 : 16 -> 25 -> 26 +n111-046-081:3352527:3352593 [2] NCCL INFO Ring 07 : 25 -> 26 -> 27 +n111-046-081:3352528:3352594 [3] NCCL INFO Ring 06 : 28 -> 27 -> 26 +n111-046-081:3352529:3352600 [4] NCCL INFO Ring 12 : 21 -> 28 -> 27 +n111-046-081:3352525:3352598 [0] NCCL INFO Ring 00 : 17 -> 24 -> 31 +n111-046-081:3352528:3352594 [3] NCCL INFO Ring 07 : 26 -> 27 -> 28 +n111-046-081:3352530:3352595 [5] NCCL INFO Ring 14 : 30 -> 29 -> 28 +n111-046-081:3352526:3352597 [1] NCCL INFO Ring 02 : 26 -> 25 -> 24 +n111-046-081:3352528:3352594 [3] NCCL INFO Ring 08 : 28 -> 27 -> 26 +n111-046-081:3352527:3352593 [2] NCCL INFO Ring 08 : 27 -> 26 -> 25 +n111-046-081:3352529:3352600 [4] NCCL INFO Ring 13 : 27 -> 28 -> 37 +n111-046-081:3352530:3352595 [5] NCCL INFO Ring 15 : 28 -> 29 -> 30 +n111-046-081:3352526:3352597 [1] NCCL INFO Ring 03 : 24 -> 25 -> 26 +n111-046-081:3352525:3352598 [0] NCCL INFO Ring 01 : 31 -> 24 -> 33 +n111-046-081:3352528:3352594 [3] NCCL INFO Ring 09 : 26 -> 27 -> 28 +n111-046-081:3352527:3352593 [2] NCCL INFO Ring 09 : 25 -> 26 -> 27 +n111-046-081:3352527:3352593 [2] NCCL INFO Ring 10 : 19 -> 26 -> 25 +n111-046-081:3352529:3352600 [4] NCCL INFO Ring 14 : 29 -> 28 -> 27 +n111-046-081:3352527:3352593 [2] NCCL INFO Ring 11 : 25 -> 26 -> 35 +n111-046-081:3352530:3352595 [5] NCCL INFO Trees [0] 30/-1/-1->29->28 [1] 28/-1/-1->29->30 [2] 30/-1/-1->29->28 [3] 28/-1/-1->29->30 [4] 30/-1/-1->29->28 [5] 28/-1/-1->29->21 [6] -1/-1/-1->29->28 [7] 28/-1/-1->29->30 [8] 30/-1/-1->29->28 [9] 28/-1/-1->29->30 [10] 30/-1/-1->29->28 [11] 28/-1/-1->29->30 [12] 30/44/-1->29->28 [13] 28/13/-1->29->61 [14] -1/-1/-1->29->28 [15] 28/-1/-1->29->30 +n111-046-081:3352526:3352597 [1] NCCL INFO Ring 04 : 26 -> 25 -> 24 +n111-046-081:3352527:3352593 [2] NCCL INFO Ring 12 : 27 -> 26 -> 25 +n111-046-081:3352525:3352598 [0] NCCL INFO Ring 02 : 25 -> 24 -> 31 +n111-046-081:3352528:3352594 [3] NCCL INFO Ring 10 : 28 -> 27 -> 34 +n111-046-081:3352528:3352594 [3] NCCL INFO Ring 11 : 18 -> 27 -> 28 +n111-046-081:3352529:3352600 [4] NCCL INFO Ring 15 : 27 -> 28 -> 29 +n111-046-081:3352526:3352597 [1] NCCL INFO Ring 05 : 24 -> 25 -> 26 +n111-046-081:3352527:3352593 [2] NCCL INFO Ring 13 : 25 -> 26 -> 27 +n111-046-081:3352529:3352600 [4] NCCL INFO Trees [0] 29/-1/-1->28->27 [1] 27/-1/-1->28->29 [2] 29/-1/-1->28->27 [3] -1/-1/-1->28->29 [4] 29/-1/-1->28->20 [5] 27/-1/-1->28->29 [6] 29/-1/-1->28->27 [7] 27/-1/-1->28->29 [8] 29/-1/-1->28->27 [9] 27/-1/-1->28->29 [10] 29/-1/-1->28->27 [11] -1/-1/-1->28->29 [12] 29/12/-1->28->60 [13] 27/45/-1->28->29 [14] 29/-1/-1->28->27 [15] 27/-1/-1->28->29 +n111-046-081:3352530:3352595 [5] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n111-046-081:3352525:3352598 [0] NCCL INFO Ring 03 : 31 -> 24 -> 25 +n111-046-081:3352528:3352594 [3] NCCL INFO Ring 12 : 28 -> 27 -> 26 +n111-046-081:3352526:3352597 [1] NCCL INFO Ring 06 : 26 -> 25 -> 24 +n111-046-081:3352526:3352597 [1] NCCL INFO Ring 07 : 24 -> 25 -> 26 +n111-046-081:3352527:3352593 [2] NCCL INFO Ring 14 : 27 -> 26 -> 25 +n111-046-081:3352530:3352595 [5] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n111-046-081:3352529:3352600 [4] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n111-046-081:3352525:3352598 [0] NCCL INFO Ring 04 : 25 -> 24 -> 31 +n111-046-081:3352528:3352594 [3] NCCL INFO Ring 13 : 26 -> 27 -> 28 +n111-046-081:3352526:3352597 [1] NCCL INFO Ring 08 : 26 -> 25 -> 32 +n111-046-081:3352527:3352593 [2] NCCL INFO Ring 15 : 25 -> 26 -> 27 +n111-046-081:3352530:3352595 [5] NCCL INFO P2P Chunksize set to 524288 +n111-046-081:3352529:3352600 [4] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n111-046-081:3352525:3352598 [0] NCCL INFO Ring 05 : 31 -> 24 -> 25 +n111-046-081:3352526:3352597 [1] NCCL INFO Ring 09 : 16 -> 25 -> 26 +n111-046-081:3352528:3352594 [3] NCCL INFO Ring 14 : 28 -> 27 -> 26 +n111-046-081:3352527:3352593 [2] NCCL INFO Trees [0] 27/-1/-1->26->25 [1] -1/-1/-1->26->27 [2] 27/-1/-1->26->18 [3] 25/-1/-1->26->27 [4] 27/-1/-1->26->25 [5] 25/-1/-1->26->27 [6] 27/-1/-1->26->25 [7] 25/-1/-1->26->27 [8] 27/-1/-1->26->25 [9] -1/-1/-1->26->27 [10] 27/10/-1->26->58 [11] 25/43/-1->26->27 [12] 27/-1/-1->26->25 [13] 25/-1/-1->26->27 [14] 27/-1/-1->26->25 [15] 25/-1/-1->26->27 +n111-046-081:3352526:3352597 [1] NCCL INFO Ring 10 : 26 -> 25 -> 24 +n111-046-081:3352529:3352600 [4] NCCL INFO P2P Chunksize set to 524288 +n111-046-081:3352525:3352598 [0] NCCL INFO Ring 06 : 25 -> 24 -> 31 +n111-046-081:3352528:3352594 [3] NCCL INFO Ring 15 : 26 -> 27 -> 28 +n111-046-081:3352526:3352597 [1] NCCL INFO Ring 11 : 24 -> 25 -> 26 +n111-046-081:3352527:3352593 [2] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n111-046-081:3352528:3352594 [3] NCCL INFO Trees [0] 28/-1/-1->27->26 [1] 26/-1/-1->27->28 [2] 28/-1/-1->27->26 [3] 26/-1/-1->27->19 [4] -1/-1/-1->27->26 [5] 26/-1/-1->27->28 [6] 28/-1/-1->27->26 [7] 26/-1/-1->27->28 [8] 28/-1/-1->27->26 [9] 26/-1/-1->27->28 [10] 28/42/-1->27->26 [11] 26/11/-1->27->59 [12] -1/-1/-1->27->26 [13] 26/-1/-1->27->28 [14] 28/-1/-1->27->26 [15] 26/-1/-1->27->28 +n111-046-081:3352526:3352597 [1] NCCL INFO Ring 12 : 26 -> 25 -> 24 +n111-046-081:3352525:3352598 [0] NCCL INFO Ring 07 : 31 -> 24 -> 25 +n111-046-081:3352527:3352593 [2] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n111-046-081:3352526:3352597 [1] NCCL INFO Ring 13 : 24 -> 25 -> 26 +n111-046-081:3352528:3352594 [3] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n111-046-081:3352527:3352593 [2] NCCL INFO P2P Chunksize set to 524288 +n111-046-081:3352525:3352598 [0] NCCL INFO Ring 08 : 17 -> 24 -> 31 +n111-046-081:3352528:3352594 [3] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n111-046-081:3352526:3352597 [1] NCCL INFO Ring 14 : 26 -> 25 -> 24 +n111-046-081:3352528:3352594 [3] NCCL INFO P2P Chunksize set to 524288 +n111-046-081:3352525:3352598 [0] NCCL INFO Ring 09 : 31 -> 24 -> 33 +n111-046-081:3352526:3352597 [1] NCCL INFO Ring 15 : 24 -> 25 -> 26 +n111-046-081:3352525:3352598 [0] NCCL INFO Ring 10 : 25 -> 24 -> 31 +n111-046-081:3352526:3352597 [1] NCCL INFO Trees [0] 26/-1/-1->25->24 [1] 24/-1/-1->25->17 [2] -1/-1/-1->25->24 [3] 24/-1/-1->25->26 [4] 26/-1/-1->25->24 [5] 24/-1/-1->25->26 [6] 26/-1/-1->25->24 [7] 24/-1/-1->25->26 [8] 26/40/-1->25->24 [9] 24/9/-1->25->57 [10] -1/-1/-1->25->24 [11] 24/-1/-1->25->26 [12] 26/-1/-1->25->24 [13] 24/-1/-1->25->26 [14] 26/-1/-1->25->24 [15] 24/-1/-1->25->26 +n111-046-081:3352525:3352598 [0] NCCL INFO Ring 11 : 31 -> 24 -> 25 +n111-046-081:3352526:3352597 [1] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n111-046-081:3352525:3352598 [0] NCCL INFO Ring 12 : 25 -> 24 -> 31 +n111-046-081:3352525:3352598 [0] NCCL INFO Ring 13 : 31 -> 24 -> 25 +n111-046-081:3352525:3352598 [0] NCCL INFO Ring 14 : 25 -> 24 -> 31 +n111-046-081:3352525:3352598 [0] NCCL INFO Ring 15 : 31 -> 24 -> 25 +n111-046-081:3352526:3352597 [1] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n111-046-081:3352525:3352598 [0] NCCL INFO Trees [0] 25/-1/-1->24->16 [1] 31/-1/-1->24->25 [2] 25/-1/-1->24->31 [3] 31/-1/-1->24->25 [4] 25/-1/-1->24->31 [5] 31/-1/-1->24->25 [6] 25/-1/-1->24->31 [7] -1/-1/-1->24->25 [8] 25/8/-1->24->56 [9] 31/41/-1->24->25 [10] 25/-1/-1->24->31 [11] 31/-1/-1->24->25 [12] 25/-1/-1->24->31 [13] 31/-1/-1->24->25 [14] 25/-1/-1->24->31 [15] -1/-1/-1->24->25 +n111-046-081:3352526:3352597 [1] NCCL INFO P2P Chunksize set to 524288 +n111-046-081:3352525:3352598 [0] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n111-046-081:3352525:3352598 [0] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n111-046-081:3352525:3352598 [0] NCCL INFO P2P Chunksize set to 524288 +n111-046-081:3352532:3352633 [7] NCCL INFO [Proxy Service] Device 7 CPU core 73 +n111-046-081:3352531:3352637 [6] NCCL INFO [Proxy Service UDS] Device 6 CPU core 199 +n111-046-081:3352531:3352634 [6] NCCL INFO [Proxy Service] Device 6 CPU core 89 +n111-046-081:3352529:3352642 [4] NCCL INFO [Proxy Service UDS] Device 4 CPU core 200 +n111-046-081:3352526:3352640 [1] NCCL INFO [Proxy Service] Device 1 CPU core 120 +n111-046-081:3352530:3352643 [5] NCCL INFO [Proxy Service UDS] Device 5 CPU core 96 +n111-046-081:3352529:3352638 [4] NCCL INFO [Proxy Service] Device 4 CPU core 199 +n111-046-081:3352528:3352645 [3] NCCL INFO [Proxy Service] Device 3 CPU core 35 +n111-046-081:3352527:3352636 [2] NCCL INFO [Proxy Service] Device 2 CPU core 110 +n111-046-081:3352526:3352644 [1] NCCL INFO [Proxy Service UDS] Device 1 CPU core 14 +n111-046-081:3352532:3352635 [7] NCCL INFO [Proxy Service UDS] Device 7 CPU core 198 +n111-046-081:3352527:3352641 [2] NCCL INFO [Proxy Service UDS] Device 2 CPU core 121 +n111-046-081:3352528:3352648 [3] NCCL INFO [Proxy Service UDS] Device 3 CPU core 26 +n111-046-081:3352525:3352647 [0] NCCL INFO [Proxy Service UDS] Device 0 CPU core 130 +n111-046-081:3352525:3352646 [0] NCCL INFO [Proxy Service] Device 0 CPU core 25 +n111-046-081:3352530:3352639 [5] NCCL INFO [Proxy Service] Device 5 CPU core 97 +n111-046-081:3352526:3352597 [1] NCCL INFO NCCL_PROTO set by environment to Simple +n111-046-081:3352526:3352597 [1] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n111-046-081:3352526:3352597 [1] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n111-046-081:3352526:3352597 [1] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n111-046-081:3352531:3352596 [6] NCCL INFO NCCL_PROTO set by environment to Simple +n111-046-081:3352531:3352596 [6] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n111-046-081:3352531:3352596 [6] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n111-046-081:3352531:3352596 [6] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n111-046-081:3352525:3352598 [0] NCCL INFO NCCL_PROTO set by environment to Simple +n111-046-081:3352525:3352598 [0] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n111-046-081:3352525:3352598 [0] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n111-046-081:3352525:3352598 [0] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n111-046-081:3352527:3352593 [2] NCCL INFO NCCL_PROTO set by environment to Simple +n111-046-081:3352527:3352593 [2] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n111-046-081:3352527:3352593 [2] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n111-046-081:3352527:3352593 [2] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n111-046-081:3352532:3352599 [7] NCCL INFO NCCL_PROTO set by environment to Simple +n111-046-081:3352532:3352599 [7] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n111-046-081:3352532:3352599 [7] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n111-046-081:3352532:3352599 [7] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n111-046-081:3352530:3352595 [5] NCCL INFO NCCL_PROTO set by environment to Simple +n111-046-081:3352530:3352595 [5] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n111-046-081:3352530:3352595 [5] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n111-046-081:3352530:3352595 [5] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n111-046-081:3352528:3352594 [3] NCCL INFO NCCL_PROTO set by environment to Simple +n111-046-081:3352528:3352594 [3] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n111-046-081:3352528:3352594 [3] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n111-046-081:3352528:3352594 [3] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n111-046-081:3352529:3352600 [4] NCCL INFO NCCL_PROTO set by environment to Simple +n111-046-081:3352529:3352600 [4] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n111-046-081:3352529:3352600 [4] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n111-046-081:3352529:3352600 [4] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n111-046-081:3352526:3352597 [1] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n111-046-081:3352526:3352597 [1] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n111-046-081:3352526:3352597 [1] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n111-046-081:3352526:3352597 [1] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n111-046-081:3352526:3352597 [1] NCCL INFO ncclCommInitRankConfig comm 0xbdbfa40 rank 25 nranks 64 cudaDev 1 nvmlDev 1 busId 5000 commId 0xfb2d11625d05007a - Init COMPLETE +n111-046-081:3352526:3352597 [1] NCCL INFO Init timings - ncclCommInitRankConfig: rank 25 nranks 64 total 4.92 (kernels 0.49, alloc 1.75, bootstrap 1.79, allgathers 0.08, topo 0.68, graphs 0.09, connections 0.03, rest 0.01) +n111-046-081:3352527:3352593 [2] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n111-046-081:3352527:3352593 [2] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n111-046-081:3352527:3352593 [2] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n111-046-081:3352527:3352593 [2] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n111-046-081:3352527:3352593 [2] NCCL INFO ncclCommInitRankConfig comm 0xbfe2f00 rank 26 nranks 64 cudaDev 2 nvmlDev 2 busId b000 commId 0xfb2d11625d05007a - Init COMPLETE +n111-046-081:3352527:3352593 [2] NCCL INFO Init timings - ncclCommInitRankConfig: rank 26 nranks 64 total 4.96 (kernels 0.47, alloc 1.67, bootstrap 1.93, allgathers 0.08, topo 0.68, graphs 0.08, connections 0.04, rest 0.00) +n111-046-081:3352532:3352599 [7] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n111-046-081:3352531:3352596 [6] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n111-046-081:3352532:3352599 [7] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n111-046-081:3352531:3352596 [6] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n111-046-081:3352532:3352599 [7] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n111-046-081:3352532:3352599 [7] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n111-046-081:3352531:3352596 [6] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n111-046-081:3352531:3352596 [6] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n111-046-081:3352532:3352599 [7] NCCL INFO ncclCommInitRankConfig comm 0xcc43090 rank 31 nranks 64 cudaDev 7 nvmlDev 7 busId 8c000 commId 0xfb2d11625d05007a - Init COMPLETE +n111-046-081:3352531:3352596 [6] NCCL INFO ncclCommInitRankConfig comm 0xb2b9f40 rank 30 nranks 64 cudaDev 6 nvmlDev 6 busId 8b000 commId 0xfb2d11625d05007a - Init COMPLETE +n111-046-081:3352529:3352600 [4] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n111-046-081:3352532:3352599 [7] NCCL INFO Init timings - ncclCommInitRankConfig: rank 31 nranks 64 total 4.90 (kernels 0.49, alloc 1.69, bootstrap 1.84, allgathers 0.08, topo 0.68, graphs 0.08, connections 0.04, rest 0.00) +n111-046-081:3352531:3352596 [6] NCCL INFO Init timings - ncclCommInitRankConfig: rank 30 nranks 64 total 4.93 (kernels 0.49, alloc 1.74, bootstrap 1.81, allgathers 0.09, topo 0.68, graphs 0.08, connections 0.03, rest 0.00) +n111-046-081:3352529:3352600 [4] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n111-046-081:3352529:3352600 [4] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n111-046-081:3352529:3352600 [4] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n111-046-081:3352529:3352600 [4] NCCL INFO ncclCommInitRankConfig comm 0xc569f60 rank 28 nranks 64 cudaDev 4 nvmlDev 4 busId 84000 commId 0xfb2d11625d05007a - Init COMPLETE +n111-046-081:3352529:3352600 [4] NCCL INFO Init timings - ncclCommInitRankConfig: rank 28 nranks 64 total 4.88 (kernels 0.46, alloc 1.73, bootstrap 1.81, allgathers 0.08, topo 0.68, graphs 0.08, connections 0.04, rest 0.00) +n111-046-081:3352530:3352595 [5] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n111-046-081:3352530:3352595 [5] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n111-046-081:3352530:3352595 [5] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n111-046-081:3352530:3352595 [5] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n111-046-081:3352530:3352595 [5] NCCL INFO ncclCommInitRankConfig comm 0xc14dc20 rank 29 nranks 64 cudaDev 5 nvmlDev 5 busId 85000 commId 0xfb2d11625d05007a - Init COMPLETE +n111-046-081:3352530:3352595 [5] NCCL INFO Init timings - ncclCommInitRankConfig: rank 29 nranks 64 total 4.93 (kernels 0.46, alloc 1.77, bootstrap 1.82, allgathers 0.09, topo 0.68, graphs 0.08, connections 0.04, rest 0.00) +n111-046-081:3352528:3352594 [3] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n111-046-081:3352528:3352594 [3] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n111-046-081:3352528:3352594 [3] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n111-046-081:3352528:3352594 [3] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n111-046-081:3352528:3352594 [3] NCCL INFO ncclCommInitRankConfig comm 0xcd5e030 rank 27 nranks 64 cudaDev 3 nvmlDev 3 busId c000 commId 0xfb2d11625d05007a - Init COMPLETE +n111-046-081:3352525:3352598 [0] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n111-046-081:3352525:3352598 [0] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n111-046-081:3352528:3352594 [3] NCCL INFO Init timings - ncclCommInitRankConfig: rank 27 nranks 64 total 4.94 (kernels 0.66, alloc 1.65, bootstrap 1.75, allgathers 0.02, topo 0.68, graphs 0.14, connections 0.04, rest 0.00) +n111-046-081:3352525:3352598 [0] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n111-046-081:3352525:3352598 [0] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n111-046-081:3352525:3352598 [0] NCCL INFO ncclCommInitRankConfig comm 0xc925780 rank 24 nranks 64 cudaDev 0 nvmlDev 0 busId 4000 commId 0xfb2d11625d05007a - Init COMPLETE +n111-046-081:3352525:3352598 [0] NCCL INFO Init timings - ncclCommInitRankConfig: rank 24 nranks 64 total 4.92 (kernels 0.49, alloc 1.72, bootstrap 1.82, allgathers 0.02, topo 0.68, graphs 0.14, connections 0.03, rest 0.00) +n111-046-081:3352525:3352656 [0] NCCL INFO Channel 00/0 : 24[0] -> 25[1] via P2P/CUMEM +n111-046-081:3352525:3352656 [0] NCCL INFO Channel 01/0 : 24[0] -> 25[1] via P2P/CUMEM +n111-046-081:3352525:3352656 [0] NCCL INFO Channel 02/0 : 24[0] -> 25[1] via P2P/CUMEM +n111-046-081:3352525:3352656 [0] NCCL INFO Channel 03/0 : 24[0] -> 25[1] via P2P/CUMEM +n111-046-081:3352525:3352656 [0] NCCL INFO Channel 04/0 : 24[0] -> 25[1] via P2P/CUMEM +n111-046-081:3352525:3352656 [0] NCCL INFO Channel 05/0 : 24[0] -> 25[1] via P2P/CUMEM +n111-046-081:3352525:3352656 [0] NCCL INFO Channel 06/0 : 24[0] -> 25[1] via P2P/CUMEM +n111-046-081:3352525:3352656 [0] NCCL INFO Channel 07/0 : 24[0] -> 25[1] via P2P/CUMEM +n111-046-081:3352525:3352656 [0] NCCL INFO Channel 08/0 : 24[0] -> 25[1] via P2P/CUMEM +n111-046-081:3352525:3352656 [0] NCCL INFO Channel 09/0 : 24[0] -> 25[1] via P2P/CUMEM +n111-046-081:3352527:3352650 [2] NCCL INFO Channel 00/0 : 26[2] -> 27[3] via P2P/CUMEM +n111-046-081:3352525:3352656 [0] NCCL INFO Channel 10/0 : 24[0] -> 25[1] via P2P/CUMEM +n111-046-081:3352527:3352650 [2] NCCL INFO Channel 01/0 : 26[2] -> 27[3] via P2P/CUMEM +n111-046-081:3352531:3352651 [6] NCCL INFO Channel 00/0 : 30[6] -> 31[7] via P2P/CUMEM +n111-046-081:3352529:3352654 [4] NCCL INFO Channel 00/0 : 28[4] -> 29[5] via P2P/CUMEM +n111-046-081:3352525:3352656 [0] NCCL INFO Channel 11/0 : 24[0] -> 25[1] via P2P/CUMEM +n111-046-081:3352527:3352650 [2] NCCL INFO Channel 02/0 : 26[2] -> 27[3] via P2P/CUMEM +n111-046-081:3352531:3352651 [6] NCCL INFO Channel 01/0 : 30[6] -> 31[7] via P2P/CUMEM +n111-046-081:3352526:3352649 [1] NCCL INFO Channel 00/0 : 25[1] -> 26[2] via P2P/CUMEM +n111-046-081:3352529:3352654 [4] NCCL INFO Channel 01/0 : 28[4] -> 29[5] via P2P/CUMEM +n111-046-081:3352525:3352656 [0] NCCL INFO Channel 12/0 : 24[0] -> 25[1] via P2P/CUMEM +n111-046-081:3352527:3352650 [2] NCCL INFO Channel 03/0 : 26[2] -> 27[3] via P2P/CUMEM +n111-046-081:3352531:3352651 [6] NCCL INFO Channel 02/0 : 30[6] -> 31[7] via P2P/CUMEM +n111-046-081:3352526:3352649 [1] NCCL INFO Channel 03/0 : 25[1] -> 26[2] via P2P/CUMEM +n111-046-081:3352529:3352654 [4] NCCL INFO Channel 02/0 : 28[4] -> 29[5] via P2P/CUMEM +n111-046-081:3352525:3352656 [0] NCCL INFO Channel 13/0 : 24[0] -> 25[1] via P2P/CUMEM +n111-046-081:3352527:3352650 [2] NCCL INFO Channel 04/0 : 26[2] -> 27[3] via P2P/CUMEM +n111-046-081:3352531:3352651 [6] NCCL INFO Channel 03/0 : 30[6] -> 31[7] via P2P/CUMEM +n111-046-081:3352529:3352654 [4] NCCL INFO Channel 03/0 : 28[4] -> 29[5] via P2P/CUMEM +n111-046-081:3352526:3352649 [1] NCCL INFO Channel 04/0 : 25[1] -> 26[2] via P2P/CUMEM +n111-046-081:3352525:3352656 [0] NCCL INFO Channel 14/0 : 24[0] -> 25[1] via P2P/CUMEM +n111-046-081:3352527:3352650 [2] NCCL INFO Channel 05/0 : 26[2] -> 27[3] via P2P/CUMEM +n111-046-081:3352531:3352651 [6] NCCL INFO Channel 04/0 : 30[6] -> 31[7] via P2P/CUMEM +n111-046-081:3352529:3352654 [4] NCCL INFO Channel 04/0 : 28[4] -> 29[5] via P2P/CUMEM +n111-046-081:3352528:3352653 [3] NCCL INFO Channel 00/0 : 27[3] -> 28[4] via P2P/CUMEM +n111-046-081:3352526:3352649 [1] NCCL INFO Channel 05/0 : 25[1] -> 26[2] via P2P/CUMEM +n111-046-081:3352525:3352656 [0] NCCL INFO Channel 15/0 : 24[0] -> 25[1] via P2P/CUMEM +n111-046-081:3352527:3352650 [2] NCCL INFO Channel 06/0 : 26[2] -> 27[3] via P2P/CUMEM +n111-046-081:3352530:3352655 [5] NCCL INFO Channel 00/0 : 29[5] -> 30[6] via P2P/CUMEM +n111-046-081:3352531:3352651 [6] NCCL INFO Channel 05/0 : 30[6] -> 31[7] via P2P/CUMEM +n111-046-081:3352529:3352654 [4] NCCL INFO Channel 05/0 : 28[4] -> 29[5] via P2P/CUMEM +n111-046-081:3352528:3352653 [3] NCCL INFO Channel 01/0 : 27[3] -> 28[4] via P2P/CUMEM +n111-046-081:3352526:3352649 [1] NCCL INFO Channel 06/0 : 25[1] -> 26[2] via P2P/CUMEM +n111-046-081:3352527:3352650 [2] NCCL INFO Channel 07/0 : 26[2] -> 27[3] via P2P/CUMEM +n111-046-081:3352530:3352655 [5] NCCL INFO Channel 01/0 : 29[5] -> 30[6] via P2P/CUMEM +n111-046-081:3352531:3352651 [6] NCCL INFO Channel 06/0 : 30[6] -> 31[7] via P2P/CUMEM +n111-046-081:3352529:3352654 [4] NCCL INFO Channel 06/0 : 28[4] -> 29[5] via P2P/CUMEM +n111-046-081:3352528:3352653 [3] NCCL INFO Channel 02/0 : 27[3] -> 28[4] via P2P/CUMEM +n111-046-081:3352526:3352649 [1] NCCL INFO Channel 07/0 : 25[1] -> 26[2] via P2P/CUMEM +n111-046-081:3352527:3352650 [2] NCCL INFO Channel 08/0 : 26[2] -> 27[3] via P2P/CUMEM +n111-046-081:3352530:3352655 [5] NCCL INFO Channel 02/0 : 29[5] -> 30[6] via P2P/CUMEM +n111-046-081:3352531:3352651 [6] NCCL INFO Channel 07/0 : 30[6] -> 31[7] via P2P/CUMEM +n111-046-081:3352529:3352654 [4] NCCL INFO Channel 07/0 : 28[4] -> 29[5] via P2P/CUMEM +n111-046-081:3352528:3352653 [3] NCCL INFO Channel 05/0 : 27[3] -> 28[4] via P2P/CUMEM +n111-046-081:3352526:3352649 [1] NCCL INFO Channel 08/0 : 25[1] -> 26[2] via P2P/CUMEM +n111-046-081:3352527:3352650 [2] NCCL INFO Channel 09/0 : 26[2] -> 27[3] via P2P/CUMEM +n111-046-081:3352530:3352655 [5] NCCL INFO Channel 03/0 : 29[5] -> 30[6] via P2P/CUMEM +n111-046-081:3352531:3352651 [6] NCCL INFO Channel 08/0 : 30[6] -> 31[7] via P2P/CUMEM +n111-046-081:3352529:3352654 [4] NCCL INFO Channel 08/0 : 28[4] -> 29[5] via P2P/CUMEM +n111-046-081:3352528:3352653 [3] NCCL INFO Channel 06/0 : 27[3] -> 28[4] via P2P/CUMEM +n111-046-081:3352526:3352649 [1] NCCL INFO Channel 11/0 : 25[1] -> 26[2] via P2P/CUMEM +n111-046-081:3352527:3352650 [2] NCCL INFO Channel 10/0 : 26[2] -> 27[3] via P2P/CUMEM +n111-046-081:3352530:3352655 [5] NCCL INFO Channel 04/0 : 29[5] -> 30[6] via P2P/CUMEM +n111-046-081:3352531:3352651 [6] NCCL INFO Channel 09/0 : 30[6] -> 31[7] via P2P/CUMEM +n111-046-081:3352529:3352654 [4] NCCL INFO Channel 09/0 : 28[4] -> 29[5] via P2P/CUMEM +n111-046-081:3352528:3352653 [3] NCCL INFO Channel 07/0 : 27[3] -> 28[4] via P2P/CUMEM +n111-046-081:3352526:3352649 [1] NCCL INFO Channel 12/0 : 25[1] -> 26[2] via P2P/CUMEM +n111-046-081:3352527:3352650 [2] NCCL INFO Channel 11/0 : 26[2] -> 27[3] via P2P/CUMEM +n111-046-081:3352531:3352651 [6] NCCL INFO Channel 10/0 : 30[6] -> 31[7] via P2P/CUMEM +n111-046-081:3352530:3352655 [5] NCCL INFO Channel 07/0 : 29[5] -> 30[6] via P2P/CUMEM +n111-046-081:3352529:3352654 [4] NCCL INFO Channel 10/0 : 28[4] -> 29[5] via P2P/CUMEM +n111-046-081:3352528:3352653 [3] NCCL INFO Channel 08/0 : 27[3] -> 28[4] via P2P/CUMEM +n111-046-081:3352530:3352655 [5] NCCL INFO Channel 08/0 : 29[5] -> 30[6] via P2P/CUMEM +n111-046-081:3352526:3352649 [1] NCCL INFO Channel 13/0 : 25[1] -> 26[2] via P2P/CUMEM +n111-046-081:3352529:3352654 [4] NCCL INFO Channel 11/0 : 28[4] -> 29[5] via P2P/CUMEM +n111-046-081:3352527:3352650 [2] NCCL INFO Channel 12/0 : 26[2] -> 27[3] via P2P/CUMEM +n111-046-081:3352528:3352653 [3] NCCL INFO Channel 09/0 : 27[3] -> 28[4] via P2P/CUMEM +n111-046-081:3352530:3352655 [5] NCCL INFO Channel 09/0 : 29[5] -> 30[6] via P2P/CUMEM +n111-046-081:3352529:3352654 [4] NCCL INFO Channel 12/0 : 28[4] -> 29[5] via P2P/CUMEM +n111-046-081:3352528:3352653 [3] NCCL INFO Channel 10/0 : 27[3] -> 28[4] via P2P/CUMEM +n111-046-081:3352526:3352649 [1] NCCL INFO Channel 14/0 : 25[1] -> 26[2] via P2P/CUMEM +n111-046-081:3352527:3352650 [2] NCCL INFO Channel 13/0 : 26[2] -> 27[3] via P2P/CUMEM +n111-046-081:3352529:3352654 [4] NCCL INFO Channel 13/0 : 28[4] -> 29[5] via P2P/CUMEM +n111-046-081:3352528:3352653 [3] NCCL INFO Channel 13/0 : 27[3] -> 28[4] via P2P/CUMEM +n111-046-081:3352526:3352649 [1] NCCL INFO Channel 15/0 : 25[1] -> 26[2] via P2P/CUMEM +n111-046-081:3352527:3352650 [2] NCCL INFO Channel 14/0 : 26[2] -> 27[3] via P2P/CUMEM +n111-046-081:3352529:3352654 [4] NCCL INFO Channel 14/0 : 28[4] -> 29[5] via P2P/CUMEM +n111-046-081:3352528:3352653 [3] NCCL INFO Channel 14/0 : 27[3] -> 28[4] via P2P/CUMEM +n111-046-081:3352527:3352650 [2] NCCL INFO Channel 15/0 : 26[2] -> 27[3] via P2P/CUMEM +n111-046-081:3352529:3352654 [4] NCCL INFO Channel 15/0 : 28[4] -> 29[5] via P2P/CUMEM +n111-046-081:3352528:3352653 [3] NCCL INFO Channel 15/0 : 27[3] -> 28[4] via P2P/CUMEM +n111-046-081:3352525:3352656 [0] NCCL INFO Channel 01/0 : 24[0] -> 31[7] via P2P/CUMEM +n111-046-081:3352531:3352651 [6] NCCL INFO Channel 11/0 : 30[6] -> 31[7] via P2P/CUMEM +n111-046-081:3352530:3352655 [5] NCCL INFO Channel 10/0 : 29[5] -> 30[6] via P2P/CUMEM +n111-046-081:3352525:3352656 [0] NCCL INFO Channel 02/0 : 24[0] -> 31[7] via P2P/CUMEM +n111-046-081:3352531:3352651 [6] NCCL INFO Channel 12/0 : 30[6] -> 31[7] via P2P/CUMEM +n111-046-081:3352530:3352655 [5] NCCL INFO Channel 11/0 : 29[5] -> 30[6] via P2P/CUMEM +n111-046-081:3352525:3352656 [0] NCCL INFO Channel 03/0 : 24[0] -> 31[7] via P2P/CUMEM +n111-046-081:3352531:3352651 [6] NCCL INFO Channel 13/0 : 30[6] -> 31[7] via P2P/CUMEM +n111-046-081:3352530:3352655 [5] NCCL INFO Channel 12/0 : 29[5] -> 30[6] via P2P/CUMEM +n111-046-081:3352525:3352656 [0] NCCL INFO Channel 04/0 : 24[0] -> 31[7] via P2P/CUMEM +n111-046-081:3352531:3352651 [6] NCCL INFO Channel 14/0 : 30[6] -> 31[7] via P2P/CUMEM +n111-046-081:3352530:3352655 [5] NCCL INFO Channel 15/0 : 29[5] -> 30[6] via P2P/CUMEM +n111-046-081:3352525:3352656 [0] NCCL INFO Channel 05/0 : 24[0] -> 31[7] via P2P/CUMEM +n111-046-081:3352531:3352651 [6] NCCL INFO Channel 15/0 : 30[6] -> 31[7] via P2P/CUMEM +n111-046-081:3352525:3352656 [0] NCCL INFO Channel 06/0 : 24[0] -> 31[7] via P2P/CUMEM +n111-046-081:3352525:3352656 [0] NCCL INFO Channel 09/0 : 24[0] -> 31[7] via P2P/CUMEM +n111-046-081:3352525:3352656 [0] NCCL INFO Channel 10/0 : 24[0] -> 31[7] via P2P/CUMEM +n111-046-081:3352525:3352656 [0] NCCL INFO Channel 11/0 : 24[0] -> 31[7] via P2P/CUMEM +n111-046-081:3352525:3352656 [0] NCCL INFO Channel 12/0 : 24[0] -> 31[7] via P2P/CUMEM +n111-046-081:3352525:3352656 [0] NCCL INFO Channel 13/0 : 24[0] -> 31[7] via P2P/CUMEM +n111-046-081:3352525:3352656 [0] NCCL INFO Channel 14/0 : 24[0] -> 31[7] via P2P/CUMEM +n111-046-081:3352526:3352657 [1] NCCL INFO [Proxy Progress] Device 1 CPU core 150 +n111-046-081:3352526:3352649 [1] NCCL INFO Channel 01/0 : 17[1] -> 25[1] [receive] via NET/FasTrak/1/GDRDMA +n111-046-081:3352527:3352658 [2] NCCL INFO [Proxy Progress] Device 2 CPU core 114 +n111-046-081:3352528:3352659 [3] NCCL INFO [Proxy Progress] Device 3 CPU core 39 +n111-046-081:3352527:3352650 [2] NCCL INFO Channel 02/0 : 18[2] -> 26[2] [receive] via NET/FasTrak/2/GDRDMA +n111-046-081:3352528:3352653 [3] NCCL INFO Channel 03/0 : 19[3] -> 27[3] [receive] via NET/FasTrak/3/GDRDMA +n111-046-081:3352526:3352649 [1] NCCL INFO Channel 08/0 : 25[1] -> 40[0] [send] via NET/FasTrak/0/GDRDMA +n111-046-081:3352530:3352662 [5] NCCL INFO [Proxy Progress] Device 5 CPU core 101 +n111-046-081:3352531:3352661 [6] NCCL INFO [Proxy Progress] Device 6 CPU core 93 +n111-046-081:3352529:3352660 [4] NCCL INFO [Proxy Progress] Device 4 CPU core 203 +n111-046-081:3352530:3352655 [5] NCCL INFO Channel 05/0 : 21[5] -> 29[5] [receive] via NET/FasTrak/5/GDRDMA +n111-046-081:3352529:3352654 [4] NCCL INFO Channel 04/0 : 20[4] -> 28[4] [receive] via NET/FasTrak/4/GDRDMA +n111-046-081:3352531:3352651 [6] NCCL INFO Channel 06/0 : 22[6] -> 30[6] [receive] via NET/FasTrak/6/GDRDMA +n111-046-081:3352530:3352655 [5] NCCL INFO Channel 12/0 : 29[5] -> 44[4] [send] via NET/FasTrak/4/GDRDMA +n111-046-081:3352525:3352663 [0] NCCL INFO [Proxy Progress] Device 0 CPU core 27 +n111-046-081:3352532:3352664 [7] NCCL INFO [Proxy Progress] Device 7 CPU core 74 +n111-046-081:3352532:3352652 [7] NCCL INFO Channel 07/0 : 23[7] -> 31[7] [receive] via NET/FasTrak/7/GDRDMA +n111-046-081:3352525:3352656 [0] NCCL INFO Channel 00/0 : 16[0] -> 24[0] [receive] via NET/FasTrak/0/GDRDMA +n111-046-081:3352528:3352653 [3] NCCL INFO Channel 10/0 : 27[3] -> 42[2] [send] via NET/FasTrak/2/GDRDMA +n111-046-081:3352527:3352650 [2] NCCL INFO Channel 10/0 : 10[2] -> 26[2] [receive] via NET/FasTrak/2/GDRDMA +n111-046-081:3352525:3352656 [0] NCCL INFO Channel 08/0 : 8[0] -> 24[0] [receive] via NET/FasTrak/0/GDRDMA +n111-046-081:3352529:3352654 [4] NCCL INFO Channel 12/0 : 12[4] -> 28[4] [receive] via NET/FasTrak/4/GDRDMA +n111-046-081:3352529:3352654 [4] NCCL INFO Channel 13/0 : 28[4] -> 45[5] [send] via NET/FasTrak/5/GDRDMA +n111-046-081:3352527:3352650 [2] NCCL INFO Channel 11/0 : 26[2] -> 43[3] [send] via NET/FasTrak/3/GDRDMA +n111-046-081:3352532:3352652 [7] NCCL INFO Channel 14/0 : 31[7] -> 46[6] [send] via NET/FasTrak/6/GDRDMA +n111-046-081:3352531:3352651 [6] NCCL INFO Channel 14/0 : 14[6] -> 30[6] [receive] via NET/FasTrak/6/GDRDMA +n111-046-081:3352531:3352651 [6] NCCL INFO Channel 15/0 : 30[6] -> 47[7] [send] via NET/FasTrak/7/GDRDMA +n111-046-081:3352525:3352656 [0] NCCL INFO Channel 09/0 : 24[0] -> 41[1] [send] via NET/FasTrak/1/GDRDMA +n111-046-081:3352531:3352651 [6] NCCL INFO Channel 14/0 : 62[6] -> 30[6] [receive] via NET/FasTrak/6/GDRDMA +n111-046-081:3352531:3352651 [6] NCCL INFO Channel 14/0 : 30[6] -> 62[6] [send] via NET/FasTrak/6/GDRDMA +n111-046-081:3352525:3352656 [0] NCCL INFO Channel 08/0 : 56[0] -> 24[0] [receive] via NET/FasTrak/0/GDRDMA +n111-046-081:3352525:3352656 [0] NCCL INFO Channel 08/0 : 24[0] -> 56[0] [send] via NET/FasTrak/0/GDRDMA +n111-046-081:3352530:3352655 [5] NCCL INFO Channel 13/0 : 13[5] -> 29[5] [receive] via NET/FasTrak/5/GDRDMA +n111-046-081:3352529:3352654 [4] NCCL INFO Channel 12/0 : 60[4] -> 28[4] [receive] via NET/FasTrak/4/GDRDMA +n111-046-081:3352529:3352654 [4] NCCL INFO Channel 12/0 : 28[4] -> 60[4] [send] via NET/FasTrak/4/GDRDMA +n111-046-081:3352525:3352656 [0] NCCL INFO Channel 09/0 : 41[1] -> 24[0] [receive] via NET/FasTrak/1/GDRDMA +n111-046-081:3352530:3352655 [5] NCCL INFO Channel 13/0 : 61[5] -> 29[5] [receive] via NET/FasTrak/5/GDRDMA +n111-046-081:3352530:3352655 [5] NCCL INFO Channel 13/0 : 29[5] -> 61[5] [send] via NET/FasTrak/5/GDRDMA +n111-046-081:3352525:3352656 [0] NCCL INFO Channel 08/0 : 24[0] -> 8[0] [send] via NET/FasTrak/0/GDRDMA +n111-046-081:3352530:3352655 [5] NCCL INFO Channel 13/0 : 29[5] -> 13[5] [send] via NET/FasTrak/5/GDRDMA +n111-046-081:3352531:3352651 [6] NCCL INFO Channel 15/0 : 47[7] -> 30[6] [receive] via NET/FasTrak/7/GDRDMA +n111-046-081:3352525:3352656 [0] NCCL INFO Channel 00/0 : 24[0] -> 16[0] [send] via NET/FasTrak/0/GDRDMA +n111-046-081:3352528:3352653 [3] NCCL INFO Channel 11/0 : 11[3] -> 27[3] [receive] via NET/FasTrak/3/GDRDMA +n111-046-081:3352531:3352651 [6] NCCL INFO Channel 14/0 : 30[6] -> 14[6] [send] via NET/FasTrak/6/GDRDMA +n111-046-081:3352527:3352650 [2] NCCL INFO Channel 10/0 : 58[2] -> 26[2] [receive] via NET/FasTrak/2/GDRDMA +n111-046-081:3352529:3352654 [4] NCCL INFO Channel 13/0 : 45[5] -> 28[4] [receive] via NET/FasTrak/5/GDRDMA +n111-046-081:3352527:3352650 [2] NCCL INFO Channel 10/0 : 26[2] -> 58[2] [send] via NET/FasTrak/2/GDRDMA +n111-046-081:3352531:3352651 [6] NCCL INFO Channel 06/0 : 30[6] -> 22[6] [send] via NET/FasTrak/6/GDRDMA +n111-046-081:3352528:3352653 [3] NCCL INFO Channel 11/0 : 59[3] -> 27[3] [receive] via NET/FasTrak/3/GDRDMA +n111-046-081:3352529:3352654 [4] NCCL INFO Channel 12/0 : 28[4] -> 12[4] [send] via NET/FasTrak/4/GDRDMA +n111-046-081:3352528:3352653 [3] NCCL INFO Channel 11/0 : 27[3] -> 59[3] [send] via NET/FasTrak/3/GDRDMA +n111-046-081:3352530:3352655 [5] NCCL INFO Channel 12/0 : 44[4] -> 29[5] [receive] via NET/FasTrak/4/GDRDMA +n111-046-081:3352528:3352653 [3] NCCL INFO Channel 11/0 : 27[3] -> 11[3] [send] via NET/FasTrak/3/GDRDMA +n111-046-081:3352530:3352655 [5] NCCL INFO Channel 05/0 : 29[5] -> 21[5] [send] via NET/FasTrak/5/GDRDMA +n111-046-081:3352529:3352654 [4] NCCL INFO Channel 04/0 : 28[4] -> 20[4] [send] via NET/FasTrak/4/GDRDMA +n111-046-081:3352528:3352653 [3] NCCL INFO Channel 10/0 : 42[2] -> 27[3] [receive] via NET/FasTrak/2/GDRDMA +n111-046-081:3352526:3352649 [1] NCCL INFO Channel 09/0 : 9[1] -> 25[1] [receive] via NET/FasTrak/1/GDRDMA +n111-046-081:3352528:3352653 [3] NCCL INFO Channel 03/0 : 27[3] -> 19[3] [send] via NET/FasTrak/3/GDRDMA +n111-046-081:3352526:3352649 [1] NCCL INFO Channel 09/0 : 57[1] -> 25[1] [receive] via NET/FasTrak/1/GDRDMA +n111-046-081:3352526:3352649 [1] NCCL INFO Channel 09/0 : 25[1] -> 57[1] [send] via NET/FasTrak/1/GDRDMA +n111-046-081:3352526:3352649 [1] NCCL INFO Channel 09/0 : 25[1] -> 9[1] [send] via NET/FasTrak/1/GDRDMA +n111-046-081:3352527:3352650 [2] NCCL INFO Channel 11/0 : 43[3] -> 26[2] [receive] via NET/FasTrak/3/GDRDMA +n111-046-081:3352527:3352650 [2] NCCL INFO Channel 10/0 : 26[2] -> 10[2] [send] via NET/FasTrak/2/GDRDMA +n111-046-081:3352526:3352649 [1] NCCL INFO Channel 08/0 : 40[0] -> 25[1] [receive] via NET/FasTrak/0/GDRDMA +n111-046-081:3352527:3352650 [2] NCCL INFO Channel 02/0 : 26[2] -> 18[2] [send] via NET/FasTrak/2/GDRDMA +n111-046-081:3352526:3352649 [1] NCCL INFO Channel 01/0 : 25[1] -> 17[1] [send] via NET/FasTrak/1/GDRDMA +n111-046-081:3352526:3352649 [1] NCCL INFO Channel 00/0 : 25[1] -> 24[0] via P2P/CUMEM +n111-046-081:3352530:3352655 [5] NCCL INFO Channel 00/0 : 29[5] -> 28[4] via P2P/CUMEM +n111-046-081:3352526:3352649 [1] NCCL INFO Channel 01/0 : 25[1] -> 24[0] via P2P/CUMEM +n111-046-081:3352530:3352655 [5] NCCL INFO Channel 01/0 : 29[5] -> 28[4] via P2P/CUMEM +n111-046-081:3352526:3352649 [1] NCCL INFO Channel 02/0 : 25[1] -> 24[0] via P2P/CUMEM +n111-046-081:3352530:3352655 [5] NCCL INFO Channel 02/0 : 29[5] -> 28[4] via P2P/CUMEM +n111-046-081:3352526:3352649 [1] NCCL INFO Channel 03/0 : 25[1] -> 24[0] via P2P/CUMEM +n111-046-081:3352530:3352655 [5] NCCL INFO Channel 03/0 : 29[5] -> 28[4] via P2P/CUMEM +n111-046-081:3352527:3352650 [2] NCCL INFO Channel 00/0 : 26[2] -> 25[1] via P2P/CUMEM +n111-046-081:3352529:3352654 [4] NCCL INFO Channel 00/0 : 28[4] -> 27[3] via P2P/CUMEM +n111-046-081:3352526:3352649 [1] NCCL INFO Channel 04/0 : 25[1] -> 24[0] via P2P/CUMEM +n111-046-081:3352530:3352655 [5] NCCL INFO Channel 04/0 : 29[5] -> 28[4] via P2P/CUMEM +n111-046-081:3352527:3352650 [2] NCCL INFO Channel 03/0 : 26[2] -> 25[1] via P2P/CUMEM +n111-046-081:3352529:3352654 [4] NCCL INFO Channel 01/0 : 28[4] -> 27[3] via P2P/CUMEM +n111-046-081:3352526:3352649 [1] NCCL INFO Channel 05/0 : 25[1] -> 24[0] via P2P/CUMEM +n111-046-081:3352530:3352655 [5] NCCL INFO Channel 05/0 : 29[5] -> 28[4] via P2P/CUMEM +n111-046-081:3352527:3352650 [2] NCCL INFO Channel 04/0 : 26[2] -> 25[1] via P2P/CUMEM +n111-046-081:3352529:3352654 [4] NCCL INFO Channel 02/0 : 28[4] -> 27[3] via P2P/CUMEM +n111-046-081:3352526:3352649 [1] NCCL INFO Channel 06/0 : 25[1] -> 24[0] via P2P/CUMEM +n111-046-081:3352530:3352655 [5] NCCL INFO Channel 06/0 : 29[5] -> 28[4] via P2P/CUMEM +n111-046-081:3352527:3352650 [2] NCCL INFO Channel 05/0 : 26[2] -> 25[1] via P2P/CUMEM +n111-046-081:3352529:3352654 [4] NCCL INFO Channel 05/0 : 28[4] -> 27[3] via P2P/CUMEM +n111-046-081:3352526:3352649 [1] NCCL INFO Channel 07/0 : 25[1] -> 24[0] via P2P/CUMEM +n111-046-081:3352530:3352655 [5] NCCL INFO Channel 07/0 : 29[5] -> 28[4] via P2P/CUMEM +n111-046-081:3352527:3352650 [2] NCCL INFO Channel 06/0 : 26[2] -> 25[1] via P2P/CUMEM +n111-046-081:3352528:3352653 [3] NCCL INFO Channel 00/0 : 27[3] -> 26[2] via P2P/CUMEM +n111-046-081:3352529:3352654 [4] NCCL INFO Channel 06/0 : 28[4] -> 27[3] via P2P/CUMEM +n111-046-081:3352526:3352649 [1] NCCL INFO Channel 08/0 : 25[1] -> 24[0] via P2P/CUMEM +n111-046-081:3352530:3352655 [5] NCCL INFO Channel 08/0 : 29[5] -> 28[4] via P2P/CUMEM +n111-046-081:3352527:3352650 [2] NCCL INFO Channel 07/0 : 26[2] -> 25[1] via P2P/CUMEM +n111-046-081:3352528:3352653 [3] NCCL INFO Channel 01/0 : 27[3] -> 26[2] via P2P/CUMEM +n111-046-081:3352529:3352654 [4] NCCL INFO Channel 07/0 : 28[4] -> 27[3] via P2P/CUMEM +n111-046-081:3352526:3352649 [1] NCCL INFO Channel 09/0 : 25[1] -> 24[0] via P2P/CUMEM +n111-046-081:3352530:3352655 [5] NCCL INFO Channel 09/0 : 29[5] -> 28[4] via P2P/CUMEM +n111-046-081:3352527:3352650 [2] NCCL INFO Channel 08/0 : 26[2] -> 25[1] via P2P/CUMEM +n111-046-081:3352528:3352653 [3] NCCL INFO Channel 02/0 : 27[3] -> 26[2] via P2P/CUMEM +n111-046-081:3352529:3352654 [4] NCCL INFO Channel 08/0 : 28[4] -> 27[3] via P2P/CUMEM +n111-046-081:3352526:3352649 [1] NCCL INFO Channel 10/0 : 25[1] -> 24[0] via P2P/CUMEM +n111-046-081:3352530:3352655 [5] NCCL INFO Channel 10/0 : 29[5] -> 28[4] via P2P/CUMEM +n111-046-081:3352527:3352650 [2] NCCL INFO Channel 11/0 : 26[2] -> 25[1] via P2P/CUMEM +n111-046-081:3352528:3352653 [3] NCCL INFO Channel 03/0 : 27[3] -> 26[2] via P2P/CUMEM +n111-046-081:3352529:3352654 [4] NCCL INFO Channel 09/0 : 28[4] -> 27[3] via P2P/CUMEM +n111-046-081:3352526:3352649 [1] NCCL INFO Channel 11/0 : 25[1] -> 24[0] via P2P/CUMEM +n111-046-081:3352530:3352655 [5] NCCL INFO Channel 11/0 : 29[5] -> 28[4] via P2P/CUMEM +n111-046-081:3352527:3352650 [2] NCCL INFO Channel 12/0 : 26[2] -> 25[1] via P2P/CUMEM +n111-046-081:3352528:3352653 [3] NCCL INFO Channel 04/0 : 27[3] -> 26[2] via P2P/CUMEM +n111-046-081:3352529:3352654 [4] NCCL INFO Channel 10/0 : 28[4] -> 27[3] via P2P/CUMEM +n111-046-081:3352526:3352649 [1] NCCL INFO Channel 12/0 : 25[1] -> 24[0] via P2P/CUMEM +n111-046-081:3352530:3352655 [5] NCCL INFO Channel 12/0 : 29[5] -> 28[4] via P2P/CUMEM +n111-046-081:3352527:3352650 [2] NCCL INFO Channel 13/0 : 26[2] -> 25[1] via P2P/CUMEM +n111-046-081:3352528:3352653 [3] NCCL INFO Channel 05/0 : 27[3] -> 26[2] via P2P/CUMEM +n111-046-081:3352529:3352654 [4] NCCL INFO Channel 13/0 : 28[4] -> 27[3] via P2P/CUMEM +n111-046-081:3352526:3352649 [1] NCCL INFO Channel 13/0 : 25[1] -> 24[0] via P2P/CUMEM +n111-046-081:3352530:3352655 [5] NCCL INFO Channel 13/0 : 29[5] -> 28[4] via P2P/CUMEM +n111-046-081:3352527:3352650 [2] NCCL INFO Channel 14/0 : 26[2] -> 25[1] via P2P/CUMEM +n111-046-081:3352528:3352653 [3] NCCL INFO Channel 06/0 : 27[3] -> 26[2] via P2P/CUMEM +n111-046-081:3352529:3352654 [4] NCCL INFO Channel 14/0 : 28[4] -> 27[3] via P2P/CUMEM +n111-046-081:3352526:3352649 [1] NCCL INFO Channel 14/0 : 25[1] -> 24[0] via P2P/CUMEM +n111-046-081:3352530:3352655 [5] NCCL INFO Channel 14/0 : 29[5] -> 28[4] via P2P/CUMEM +n111-046-081:3352527:3352650 [2] NCCL INFO Channel 15/0 : 26[2] -> 25[1] via P2P/CUMEM +n111-046-081:3352528:3352653 [3] NCCL INFO Channel 07/0 : 27[3] -> 26[2] via P2P/CUMEM +n111-046-081:3352529:3352654 [4] NCCL INFO Channel 15/0 : 28[4] -> 27[3] via P2P/CUMEM +n111-046-081:3352526:3352649 [1] NCCL INFO Channel 15/0 : 25[1] -> 24[0] via P2P/CUMEM +n111-046-081:3352530:3352655 [5] NCCL INFO Channel 15/0 : 29[5] -> 28[4] via P2P/CUMEM +n111-046-081:3352528:3352653 [3] NCCL INFO Channel 08/0 : 27[3] -> 26[2] via P2P/CUMEM +n111-046-081:3352528:3352653 [3] NCCL INFO Channel 09/0 : 27[3] -> 26[2] via P2P/CUMEM +n111-046-081:3352532:3352652 [7] NCCL INFO Channel 15/0 : 15[7] -> 31[7] [receive] via NET/FasTrak/7/GDRDMA +n111-046-081:3352528:3352653 [3] NCCL INFO Channel 10/0 : 27[3] -> 26[2] via P2P/CUMEM +n111-046-081:3352528:3352653 [3] NCCL INFO Channel 11/0 : 27[3] -> 26[2] via P2P/CUMEM +n111-046-081:3352528:3352653 [3] NCCL INFO Channel 12/0 : 27[3] -> 26[2] via P2P/CUMEM +n111-046-081:3352528:3352653 [3] NCCL INFO Channel 13/0 : 27[3] -> 26[2] via P2P/CUMEM +n111-046-081:3352532:3352652 [7] NCCL INFO Channel 15/0 : 63[7] -> 31[7] [receive] via NET/FasTrak/7/GDRDMA +n111-046-081:3352528:3352653 [3] NCCL INFO Channel 14/0 : 27[3] -> 26[2] via P2P/CUMEM +n111-046-081:3352532:3352652 [7] NCCL INFO Channel 15/0 : 31[7] -> 63[7] [send] via NET/FasTrak/7/GDRDMA +n111-046-081:3352528:3352653 [3] NCCL INFO Channel 15/0 : 27[3] -> 26[2] via P2P/CUMEM +n111-046-081:3352532:3352652 [7] NCCL INFO Channel 15/0 : 31[7] -> 15[7] [send] via NET/FasTrak/7/GDRDMA +n111-046-081:3352532:3352652 [7] NCCL INFO Channel 14/0 : 46[6] -> 31[7] [receive] via NET/FasTrak/6/GDRDMA +n111-046-081:3352532:3352652 [7] NCCL INFO Channel 07/0 : 31[7] -> 23[7] [send] via NET/FasTrak/7/GDRDMA +n111-046-081:3352532:3352652 [7] NCCL INFO Channel 01/0 : 31[7] -> 24[0] via P2P/CUMEM +n111-046-081:3352532:3352652 [7] NCCL INFO Channel 02/0 : 31[7] -> 24[0] via P2P/CUMEM +n111-046-081:3352532:3352652 [7] NCCL INFO Channel 03/0 : 31[7] -> 24[0] via P2P/CUMEM +n111-046-081:3352532:3352652 [7] NCCL INFO Channel 04/0 : 31[7] -> 24[0] via P2P/CUMEM +n111-046-081:3352532:3352652 [7] NCCL INFO Channel 05/0 : 31[7] -> 24[0] via P2P/CUMEM +n111-046-081:3352532:3352652 [7] NCCL INFO Channel 06/0 : 31[7] -> 24[0] via P2P/CUMEM +n111-046-081:3352532:3352652 [7] NCCL INFO Channel 09/0 : 31[7] -> 24[0] via P2P/CUMEM +n111-046-081:3352531:3352651 [6] NCCL INFO Channel 00/0 : 30[6] -> 29[5] via P2P/CUMEM +n111-046-081:3352532:3352652 [7] NCCL INFO Channel 10/0 : 31[7] -> 24[0] via P2P/CUMEM +n111-046-081:3352531:3352651 [6] NCCL INFO Channel 01/0 : 30[6] -> 29[5] via P2P/CUMEM +n111-046-081:3352532:3352652 [7] NCCL INFO Channel 11/0 : 31[7] -> 24[0] via P2P/CUMEM +n111-046-081:3352531:3352651 [6] NCCL INFO Channel 02/0 : 30[6] -> 29[5] via P2P/CUMEM +n111-046-081:3352532:3352652 [7] NCCL INFO Channel 12/0 : 31[7] -> 24[0] via P2P/CUMEM +n111-046-081:3352532:3352652 [7] NCCL INFO Channel 13/0 : 31[7] -> 24[0] via P2P/CUMEM +n111-046-081:3352531:3352651 [6] NCCL INFO Channel 03/0 : 30[6] -> 29[5] via P2P/CUMEM +n111-046-081:3352532:3352652 [7] NCCL INFO Channel 14/0 : 31[7] -> 24[0] via P2P/CUMEM +n111-046-081:3352531:3352651 [6] NCCL INFO Channel 04/0 : 30[6] -> 29[5] via P2P/CUMEM +n111-046-081:3352532:3352652 [7] NCCL INFO Channel 00/0 : 31[7] -> 30[6] via P2P/CUMEM +n111-046-081:3352531:3352651 [6] NCCL INFO Channel 07/0 : 30[6] -> 29[5] via P2P/CUMEM +n111-046-081:3352532:3352652 [7] NCCL INFO Channel 01/0 : 31[7] -> 30[6] via P2P/CUMEM +n111-046-081:3352531:3352651 [6] NCCL INFO Channel 08/0 : 30[6] -> 29[5] via P2P/CUMEM +n111-046-081:3352532:3352652 [7] NCCL INFO Channel 02/0 : 31[7] -> 30[6] via P2P/CUMEM +n111-046-081:3352531:3352651 [6] NCCL INFO Channel 09/0 : 30[6] -> 29[5] via P2P/CUMEM +n111-046-081:3352532:3352652 [7] NCCL INFO Channel 03/0 : 31[7] -> 30[6] via P2P/CUMEM +n111-046-081:3352531:3352651 [6] NCCL INFO Channel 10/0 : 30[6] -> 29[5] via P2P/CUMEM +n111-046-081:3352532:3352652 [7] NCCL INFO Channel 04/0 : 31[7] -> 30[6] via P2P/CUMEM +n111-046-081:3352532:3352652 [7] NCCL INFO Channel 05/0 : 31[7] -> 30[6] via P2P/CUMEM +n111-046-081:3352532:3352652 [7] NCCL INFO Channel 06/0 : 31[7] -> 30[6] via P2P/CUMEM +n111-046-081:3352531:3352651 [6] NCCL INFO Channel 11/0 : 30[6] -> 29[5] via P2P/CUMEM +n111-046-081:3352532:3352652 [7] NCCL INFO Channel 07/0 : 31[7] -> 30[6] via P2P/CUMEM +n111-046-081:3352532:3352652 [7] NCCL INFO Channel 08/0 : 31[7] -> 30[6] via P2P/CUMEM +n111-046-081:3352532:3352652 [7] NCCL INFO Channel 09/0 : 31[7] -> 30[6] via P2P/CUMEM +n111-046-081:3352531:3352651 [6] NCCL INFO Channel 12/0 : 30[6] -> 29[5] via P2P/CUMEM +n111-046-081:3352532:3352652 [7] NCCL INFO Channel 10/0 : 31[7] -> 30[6] via P2P/CUMEM +n111-046-081:3352531:3352651 [6] NCCL INFO Channel 15/0 : 30[6] -> 29[5] via P2P/CUMEM +n111-046-081:3352532:3352652 [7] NCCL INFO Channel 11/0 : 31[7] -> 30[6] via P2P/CUMEM +n111-046-081:3352532:3352652 [7] NCCL INFO Channel 12/0 : 31[7] -> 30[6] via P2P/CUMEM +n111-046-081:3352532:3352652 [7] NCCL INFO Channel 13/0 : 31[7] -> 30[6] via P2P/CUMEM +n111-046-081:3352532:3352652 [7] NCCL INFO Channel 14/0 : 31[7] -> 30[6] via P2P/CUMEM +n111-046-081:3352532:3352652 [7] NCCL INFO Channel 15/0 : 31[7] -> 30[6] via P2P/CUMEM +n111-046-081:3352527:3352650 [2] NCCL INFO Connected all trees +n111-046-081:3352528:3352653 [3] NCCL INFO Connected all trees +n111-046-081:3352532:3352652 [7] NCCL INFO Connected all trees +n111-046-081:3352525:3352656 [0] NCCL INFO Connected all trees +n111-046-081:3352526:3352649 [1] NCCL INFO Connected all trees +n111-046-081:3352529:3352654 [4] NCCL INFO Connected all trees +n111-046-081:3352531:3352651 [6] NCCL INFO Connected all trees +n111-046-081:3352530:3352655 [5] NCCL INFO Connected all trees +n111-046-081:3352532:3352685 [7] NCCL INFO Channel 06/0 : 31[7] -> 38[6] [send] via NET/FasTrak/6/GDRDMA +n111-046-081:3352532:3352685 [7] NCCL INFO Channel 14/0 : 31[7] -> 38[6] [send] via NET/FasTrak/6/GDRDMA +n111-046-081:3352525:3352686 [0] NCCL INFO Channel 00/0 : 17[1] -> 24[0] [receive] via NET/FasTrak/0/GDRDMA +n111-046-081:3352525:3352686 [0] NCCL INFO Channel 08/0 : 17[1] -> 24[0] [receive] via NET/FasTrak/0/GDRDMA +n111-046-081:3352525:3352686 [0] NCCL INFO Channel 00/0 : 24[0] -> 31[7] via P2P/CUMEM +n111-046-081:3352525:3352686 [0] NCCL INFO Channel 08/0 : 24[0] -> 31[7] via P2P/CUMEM +n111-046-081:3352525:3352686 [0] NCCL INFO Channel 01/0 : 24[0] -> 33[1] [send] via NET/FasTrak/1/GDRDMA +n111-046-081:3352525:3352686 [0] NCCL INFO Channel 09/0 : 24[0] -> 33[1] [send] via NET/FasTrak/1/GDRDMA +n111-046-081:3352526:3352688 [1] NCCL INFO Channel 01/0 : 25[1] -> 26[2] via P2P/CUMEM +n111-046-081:3352526:3352688 [1] NCCL INFO Channel 09/0 : 25[1] -> 26[2] via P2P/CUMEM +n111-046-081:3352526:3352688 [1] NCCL INFO Channel 00/0 : 25[1] -> 32[0] [send] via NET/FasTrak/0/GDRDMA +n111-046-081:3352526:3352688 [1] NCCL INFO Channel 08/0 : 25[1] -> 32[0] [send] via NET/FasTrak/0/GDRDMA +n111-046-081:3352527:3352689 [2] NCCL INFO Channel 02/0 : 19[3] -> 26[2] [receive] via NET/FasTrak/2/GDRDMA +n111-046-081:3352527:3352689 [2] NCCL INFO Channel 10/0 : 19[3] -> 26[2] [receive] via NET/FasTrak/2/GDRDMA +n111-046-081:3352526:3352688 [1] NCCL INFO Channel 01/0 : 16[0] -> 25[1] [receive] via NET/FasTrak/1/GDRDMA +n111-046-081:3352526:3352688 [1] NCCL INFO Channel 09/0 : 16[0] -> 25[1] [receive] via NET/FasTrak/1/GDRDMA +n111-046-081:3352527:3352689 [2] NCCL INFO Channel 03/0 : 26[2] -> 35[3] [send] via NET/FasTrak/3/GDRDMA +n111-046-081:3352527:3352689 [2] NCCL INFO Channel 11/0 : 26[2] -> 35[3] [send] via NET/FasTrak/3/GDRDMA +n111-046-081:3352527:3352689 [2] NCCL INFO Channel 02/0 : 26[2] -> 25[1] via P2P/CUMEM +n111-046-081:3352527:3352689 [2] NCCL INFO Channel 10/0 : 26[2] -> 25[1] via P2P/CUMEM +n111-046-081:3352532:3352685 [7] NCCL INFO Channel 07/0 : 22[6] -> 31[7] [receive] via NET/FasTrak/7/GDRDMA +n111-046-081:3352532:3352685 [7] NCCL INFO Channel 15/0 : 22[6] -> 31[7] [receive] via NET/FasTrak/7/GDRDMA +n111-046-081:3352532:3352685 [7] NCCL INFO Channel 07/0 : 31[7] -> 24[0] via P2P/CUMEM +n111-046-081:3352532:3352685 [7] NCCL INFO Channel 15/0 : 31[7] -> 24[0] via P2P/CUMEM +n111-046-081:3352528:3352690 [3] NCCL INFO Channel 03/0 : 27[3] -> 28[4] via P2P/CUMEM +n111-046-081:3352528:3352690 [3] NCCL INFO Channel 11/0 : 27[3] -> 28[4] via P2P/CUMEM +n111-046-081:3352528:3352690 [3] NCCL INFO Channel 02/0 : 27[3] -> 34[2] [send] via NET/FasTrak/2/GDRDMA +n111-046-081:3352528:3352690 [3] NCCL INFO Channel 10/0 : 27[3] -> 34[2] [send] via NET/FasTrak/2/GDRDMA +n111-046-081:3352529:3352691 [4] NCCL INFO Channel 04/0 : 21[5] -> 28[4] [receive] via NET/FasTrak/4/GDRDMA +n111-046-081:3352529:3352691 [4] NCCL INFO Channel 12/0 : 21[5] -> 28[4] [receive] via NET/FasTrak/4/GDRDMA +n111-046-081:3352529:3352691 [4] NCCL INFO Channel 05/0 : 28[4] -> 37[5] [send] via NET/FasTrak/5/GDRDMA +n111-046-081:3352529:3352691 [4] NCCL INFO Channel 13/0 : 28[4] -> 37[5] [send] via NET/FasTrak/5/GDRDMA +n111-046-081:3352529:3352691 [4] NCCL INFO Channel 04/0 : 28[4] -> 27[3] via P2P/CUMEM +n111-046-081:3352529:3352691 [4] NCCL INFO Channel 12/0 : 28[4] -> 27[3] via P2P/CUMEM +n111-046-081:3352528:3352690 [3] NCCL INFO Channel 03/0 : 18[2] -> 27[3] [receive] via NET/FasTrak/3/GDRDMA +n111-046-081:3352528:3352690 [3] NCCL INFO Channel 11/0 : 18[2] -> 27[3] [receive] via NET/FasTrak/3/GDRDMA +n111-046-081:3352530:3352692 [5] NCCL INFO Channel 05/0 : 29[5] -> 30[6] via P2P/CUMEM +n111-046-081:3352530:3352692 [5] NCCL INFO Channel 13/0 : 29[5] -> 30[6] via P2P/CUMEM +n111-046-081:3352530:3352692 [5] NCCL INFO Channel 04/0 : 29[5] -> 36[4] [send] via NET/FasTrak/4/GDRDMA +n111-046-081:3352530:3352692 [5] NCCL INFO Channel 12/0 : 29[5] -> 36[4] [send] via NET/FasTrak/4/GDRDMA +n111-046-081:3352531:3352687 [6] NCCL INFO Channel 06/0 : 23[7] -> 30[6] [receive] via NET/FasTrak/6/GDRDMA +n111-046-081:3352530:3352692 [5] NCCL INFO Channel 05/0 : 20[4] -> 29[5] [receive] via NET/FasTrak/5/GDRDMA +n111-046-081:3352531:3352687 [6] NCCL INFO Channel 14/0 : 23[7] -> 30[6] [receive] via NET/FasTrak/6/GDRDMA +n111-046-081:3352531:3352687 [6] NCCL INFO Channel 07/0 : 30[6] -> 39[7] [send] via NET/FasTrak/7/GDRDMA +n111-046-081:3352531:3352687 [6] NCCL INFO Channel 15/0 : 30[6] -> 39[7] [send] via NET/FasTrak/7/GDRDMA +n111-046-081:3352530:3352692 [5] NCCL INFO Channel 13/0 : 20[4] -> 29[5] [receive] via NET/FasTrak/5/GDRDMA +n111-046-081:3352531:3352687 [6] NCCL INFO Channel 06/0 : 30[6] -> 29[5] via P2P/CUMEM +n111-046-081:3352531:3352687 [6] NCCL INFO Channel 14/0 : 30[6] -> 29[5] via P2P/CUMEM +n111-046-081:3352532:3352685 [7] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n111-046-081:3352531:3352687 [6] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n111-046-081:3352530:3352692 [5] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n111-046-081:3352528:3352690 [3] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n111-046-081:3352529:3352691 [4] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n111-046-081:3352527:3352689 [2] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n111-046-081:3352526:3352688 [1] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n111-046-081:3352525:3352686 [0] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +[rank31]:[W727 05:39:50.928667012 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n111-046-081.byted.org]:63306, remote=[n116-040-168.byted.org]:25264): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7feabcf785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7feaa62cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7feaa62cff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7feaa62d084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7feaa62ca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fea679c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7feac04124a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7feac2fe91f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7feac306989c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank29]:[W727 05:39:50.929275494 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n111-046-081.byted.org]:63326, remote=[n116-040-168.byted.org]:25264): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f044bb785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f04352cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7f04352cff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7f04352d084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7f04352ca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f03f69c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f044f2ea4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f0451ec11f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f0451f4189c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank29]:[W727 05:39:50.944051521 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 29] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank31]:[W727 05:39:50.944060848 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 31] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank26]:[W727 05:39:50.979546625 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n111-046-081.byted.org]:63316, remote=[n116-040-168.byted.org]:25264): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f6d229785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f6d0c0cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7f6d0c0cff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7f6d0c0d084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7f6d0c0ca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f6ccd7c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f6d260bb4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f6d28c921f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f6d28d1289c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank26]:[W727 05:39:50.986011602 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 26] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank25]:[W727 05:39:50.128338490 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n111-046-081.byted.org]:63288, remote=[n116-040-168.byted.org]:25264): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f9ea63785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f9e8facdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7f9e8facff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7f9e8fad084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7f9e8faca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f9e511c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f9ea9b1f4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f9eac6f61f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f9eac77689c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank25]:[W727 05:39:50.134566732 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 25] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank28]:[W727 05:39:50.526510919 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n111-046-081.byted.org]:63292, remote=[n116-040-168.byted.org]:25264): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7fb93f9785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7fb9290cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7fb9290cff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7fb9290d084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7fb9290ca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fb8ea7c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7fb9431274a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7fb945cfe1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7fb945d7e89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank28]:[W727 05:39:50.533677602 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 28] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank27]:[W727 05:39:51.694155872 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n111-046-081.byted.org]:63272, remote=[n116-040-168.byted.org]:25264): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f5c4e1785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f5c374cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7f5c374cff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7f5c374d084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7f5c374ca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f5bf8bc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f5c515f34a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f5c541ca1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f5c5424a89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank27]:[W727 05:39:51.701499152 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 27] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank30]:[W727 05:39:51.853804550 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n111-046-081.byted.org]:63278, remote=[n116-040-168.byted.org]:25264): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f0dc25785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f0dabccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7f0dabccff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7f0dabcd084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7f0dabcca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f0d6d3c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f0dc5ce94a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f0dc88c01f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f0dc894089c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank30]:[W727 05:39:51.860867290 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 30] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank24]:[W727 05:39:51.875492555 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=71, addr=[n111-046-081.byted.org]:63264, remote=[n116-040-168.byted.org]:25264): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f85767785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f855fecdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7f855fecff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7f855fed084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7f855feca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f85215c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f8579f264a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f857cafd1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f857cb7d89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank24]:[W727 05:39:51.881841714 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 24] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank29]:[W727 05:39:51.944197122 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n111-046-081.byted.org]:63326, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f044bb785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f04352cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f04352cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f04352d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f04352ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f03f69c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f044f2ea4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f0451ec11f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f0451f4189c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank29]:[W727 05:39:51.950085158 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 29] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank31]:[W727 05:39:51.944199292 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n111-046-081.byted.org]:63306, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7feabcf785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7feaa62cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7feaa62cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7feaa62d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7feaa62ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fea679c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7feac04124a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7feac2fe91f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7feac306989c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank31]:[W727 05:39:51.951508889 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 31] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank26]:[W727 05:39:51.986144217 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n111-046-081.byted.org]:63316, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f6d229785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f6d0c0cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f6d0c0cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f6d0c0d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f6d0c0ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f6ccd7c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f6d260bb4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f6d28c921f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f6d28d1289c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank26]:[W727 05:39:51.995463981 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 26] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank25]:[W727 05:39:51.134712492 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n111-046-081.byted.org]:63288, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f9ea63785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f9e8facdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f9e8facf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f9e8fad0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f9e8faca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f9e511c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f9ea9b1f4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f9eac6f61f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f9eac77689c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank25]:[W727 05:39:51.140806724 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 25] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank28]:[W727 05:39:51.533832734 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n111-046-081.byted.org]:63292, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7fb93f9785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7fb9290cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7fb9290cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7fb9290d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7fb9290ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fb8ea7c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7fb9431274a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7fb945cfe1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7fb945d7e89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank28]:[W727 05:39:51.540261449 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 28] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank27]:[W727 05:39:52.701638502 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n111-046-081.byted.org]:63272, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f5c4e1785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f5c374cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f5c374cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f5c374d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f5c374ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f5bf8bc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f5c515f34a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f5c541ca1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f5c5424a89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank27]:[W727 05:39:52.708841351 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 27] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank30]:[W727 05:39:52.861037913 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n111-046-081.byted.org]:63278, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f0dc25785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f0dabccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f0dabccf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f0dabcd0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f0dabcca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f0d6d3c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f0dc5ce94a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f0dc88c01f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f0dc894089c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank30]:[W727 05:39:52.868946921 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 30] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank24]:[W727 05:39:52.881986023 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=71, addr=[n111-046-081.byted.org]:63264, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f85767785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f855fecdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f855fecf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f855fed0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f855feca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f85215c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f8579f264a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f857cafd1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f857cb7d89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank24]:[W727 05:39:52.888212650 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 24] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank29]:[W727 05:39:52.950225781 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n111-046-081.byted.org]:63326, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f044bb785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f04352cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f04352cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f04352d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f04352ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f03f69c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f044f2ea4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f0451ec11f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f0451f4189c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank29]:[W727 05:39:52.957114950 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 29] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank31]:[W727 05:39:52.951633166 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n111-046-081.byted.org]:63306, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7feabcf785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7feaa62cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7feaa62cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7feaa62d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7feaa62ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fea679c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7feac04124a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7feac2fe91f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7feac306989c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank31]:[W727 05:39:52.957822450 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 31] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank26]:[W727 05:39:52.995599030 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n111-046-081.byted.org]:63316, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f6d229785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f6d0c0cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f6d0c0cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f6d0c0d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f6d0c0ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f6ccd7c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f6d260bb4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f6d28c921f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f6d28d1289c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank26]:[W727 05:39:52.001419036 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 26] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank25]:[W727 05:39:52.140956005 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n111-046-081.byted.org]:63288, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f9ea63785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f9e8facdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f9e8facf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f9e8fad0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f9e8faca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f9e511c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f9ea9b1f4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f9eac6f61f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f9eac77689c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank25]:[W727 05:39:52.147526010 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 25] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank28]:[W727 05:39:52.540419824 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n111-046-081.byted.org]:63292, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7fb93f9785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7fb9290cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7fb9290cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7fb9290d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7fb9290ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fb8ea7c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7fb9431274a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7fb945cfe1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7fb945d7e89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank28]:[W727 05:39:52.547761462 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 28] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank27]:[W727 05:39:53.708968784 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n111-046-081.byted.org]:63272, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f5c4e1785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f5c374cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f5c374cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f5c374d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f5c374ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f5bf8bc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f5c515f34a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f5c541ca1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f5c5424a89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank27]:[W727 05:39:53.715537955 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 27] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank30]:[W727 05:39:53.869088486 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n111-046-081.byted.org]:63278, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f0dc25785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f0dabccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f0dabccf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f0dabcd0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f0dabcca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f0d6d3c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f0dc5ce94a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f0dc88c01f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f0dc894089c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank30]:[W727 05:39:53.876202748 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 30] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank24]:[W727 05:39:53.888369023 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=71, addr=[n111-046-081.byted.org]:63264, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f85767785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f855fecdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f855fecf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f855fed0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f855feca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f85215c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f8579f264a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f857cafd1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f857cb7d89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank24]:[W727 05:39:53.894614298 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 24] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank29]:[W727 05:39:53.957299506 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n111-046-081.byted.org]:63326, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f044bb785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f04352cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f04352cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f04352d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f04352ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f03f69c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f044f2ea4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f0451ec11f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f0451f4189c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank29]:[W727 05:39:53.963647064 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 29] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank31]:[W727 05:39:53.957971418 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n111-046-081.byted.org]:63306, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7feabcf785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7feaa62cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7feaa62cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7feaa62d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7feaa62ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fea679c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7feac04124a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7feac2fe91f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7feac306989c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank31]:[W727 05:39:53.971604944 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 31] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank26]:[W727 05:39:53.001559457 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n111-046-081.byted.org]:63316, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f6d229785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f6d0c0cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f6d0c0cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f6d0c0d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f6d0c0ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f6ccd7c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f6d260bb4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f6d28c921f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f6d28d1289c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank26]:[W727 05:39:53.007804567 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 26] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank25]:[W727 05:39:53.147686565 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n111-046-081.byted.org]:63288, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f9ea63785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f9e8facdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f9e8facf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f9e8fad0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f9e8faca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f9e511c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f9ea9b1f4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f9eac6f61f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f9eac77689c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank25]:[W727 05:39:53.153515541 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 25] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +E0727 05:39:53.826000 3352460 torch/distributed/elastic/multiprocessing/api.py:874] failed (exitcode: -15) local_rank: 0 (pid: 3352525) of binary: /usr/bin/python3 +Traceback (most recent call last): + File "/usr/local/bin/torchrun", line 7, in + sys.exit(main()) + ^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 355, in wrapper + return f(*args, **kwargs) + ^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 892, in main + run(args) + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 883, in run + elastic_launch( + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 139, in __call__ + return launch_agent(self._config, self._entrypoint, list(args)) + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 270, in launch_agent + raise ChildFailedError( +torch.distributed.elastic.multiprocessing.errors.ChildFailedError: +============================================================ +main_cross_2direct_balance_realscale_perturb_v4_accum.py FAILED +------------------------------------------------------------ +Failures: +[1]: + time : 2026-07-27_05:39:53 + host : n111-046-081.byted.org + rank : 25 (local_rank: 1) + exitcode : -15 (pid: 3352526) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 3352526 +[2]: + time : 2026-07-27_05:39:53 + host : n111-046-081.byted.org + rank : 26 (local_rank: 2) + exitcode : -15 (pid: 3352527) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 3352527 +[3]: + time : 2026-07-27_05:39:53 + host : n111-046-081.byted.org + rank : 27 (local_rank: 3) + exitcode : -15 (pid: 3352528) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 3352528 +[4]: + time : 2026-07-27_05:39:53 + host : n111-046-081.byted.org + rank : 28 (local_rank: 4) + exitcode : -15 (pid: 3352529) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 3352529 +[5]: + time : 2026-07-27_05:39:53 + host : n111-046-081.byted.org + rank : 29 (local_rank: 5) + exitcode : -15 (pid: 3352530) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 3352530 +[6]: + time : 2026-07-27_05:39:53 + host : n111-046-081.byted.org + rank : 30 (local_rank: 6) + exitcode : -15 (pid: 3352531) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 3352531 +[7]: + time : 2026-07-27_05:39:53 + host : n111-046-081.byted.org + rank : 31 (local_rank: 7) + exitcode : -15 (pid: 3352532) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 3352532 +------------------------------------------------------------ +Root Cause (first observed failure): +[0]: + time : 2026-07-27_05:39:53 + host : n111-046-081.byted.org + rank : 24 (local_rank: 0) + exitcode : -15 (pid: 3352525) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 3352525 +============================================================ diff --git a/ScaleDF_rebuttal/_host_logs/20260726_2350_vith64_free8_port25264_launch/launcher_rank4_n116-041-023.log b/ScaleDF_rebuttal/_host_logs/20260726_2350_vith64_free8_port25264_launch/launcher_rank4_n116-041-023.log new file mode 100644 index 0000000000000000000000000000000000000000..4d82e886ac9c2e0964a1e9da36c5216026112b8b --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_2350_vith64_free8_port25264_launch/launcher_rank4_n116-041-023.log @@ -0,0 +1,2149 @@ +started_utc=2026-07-26T23:50:27Z +hostname=n116-041-023 +node_rank=4 +nnodes=8 +nproc_per_node=8 +world_size=64 +master_addr=10.116.40.168 +master_port=25264 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +main=main_cross_2direct_balance_realscale_perturb_v4_accum.py +model_name=deit_huge_patch14_LS +pretrain=/tmp/ScaleDF_repro_20260724/pretrained/deit_3_huge_224_21k_v1.pth +resume=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/checkpoint_snapshots/vith32_latest_for_64card_effbs4096_20260726_234604.pth +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_64card_effbs4096_from_epoch22_snapshot_free8_master_n116-040-168_20260726_2350_vith64_port25264 +batch_per_gpu=64 +accum_iter=1 +effective_batch=4096 +eval_batch_per_gpu=64 +epochs=40 +W0726 23:50:40.442000 1383066 torch/distributed/run.py:766] +W0726 23:50:40.442000 1383066 torch/distributed/run.py:766] ***************************************** +W0726 23:50:40.442000 1383066 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 23:50:40.442000 1383066 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 39): env:// +[rank39]:[W726 23:51:08.632099740 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 39] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 37): env:// +[rank37]:[W726 23:51:08.785023062 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 37] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 32): env:// +| distributed init (rank 34): env:// +[rank34]:[W726 23:51:08.155786140 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 34] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 33): env:// +[rank33]:[W726 23:51:09.590319967 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 33] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 38): env:// +[rank38]:[W726 23:51:09.772340029 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 38] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 36): env:// +[rank36]:[W726 23:51:09.782852110 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 36] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 35): env:// +[rank35]:[W726 23:51:09.795660160 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 35] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank32]:[W726 23:51:09.795972846 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 32] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +n116-041-023:1383131:1383131 [0] NCCL INFO cudaDriverVersion 12060 +n116-041-023:1383131:1383131 [0] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-041-023:1383131:1383131 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-041-023:1383131:1383131 [0] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-041-023:1383131:1383131 [0] NCCL INFO Comm config Blocking set to 1 +n116-041-023:1383138:1383138 [7] NCCL INFO cudaDriverVersion 12060 +n116-041-023:1383138:1383138 [7] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-041-023:1383138:1383138 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-041-023:1383138:1383138 [7] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-041-023:1383135:1383135 [4] NCCL INFO cudaDriverVersion 12060 +n116-041-023:1383135:1383135 [4] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-041-023:1383135:1383135 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-041-023:1383135:1383135 [4] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-041-023:1383138:1383138 [7] NCCL INFO Comm config Blocking set to 1 +n116-041-023:1383135:1383135 [4] NCCL INFO Comm config Blocking set to 1 +n116-041-023:1383134:1383134 [3] NCCL INFO cudaDriverVersion 12060 +n116-041-023:1383134:1383134 [3] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-041-023:1383134:1383134 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-041-023:1383134:1383134 [3] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-041-023:1383134:1383134 [3] NCCL INFO Comm config Blocking set to 1 +n116-041-023:1383133:1383133 [2] NCCL INFO cudaDriverVersion 12060 +n116-041-023:1383133:1383133 [2] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-041-023:1383133:1383133 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-041-023:1383133:1383133 [2] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-041-023:1383136:1383136 [5] NCCL INFO cudaDriverVersion 12060 +n116-041-023:1383136:1383136 [5] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-041-023:1383136:1383136 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-041-023:1383136:1383136 [5] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-041-023:1383133:1383133 [2] NCCL INFO Comm config Blocking set to 1 +n116-041-023:1383132:1383132 [1] NCCL INFO cudaDriverVersion 12060 +n116-041-023:1383132:1383132 [1] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-041-023:1383132:1383132 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-041-023:1383132:1383132 [1] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-041-023:1383136:1383136 [5] NCCL INFO Comm config Blocking set to 1 +n116-041-023:1383137:1383137 [6] NCCL INFO cudaDriverVersion 12060 +n116-041-023:1383137:1383137 [6] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-041-023:1383137:1383137 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-041-023:1383132:1383132 [1] NCCL INFO Comm config Blocking set to 1 +n116-041-023:1383137:1383137 [6] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-041-023:1383137:1383137 [6] NCCL INFO Comm config Blocking set to 1 +n116-041-023:1383131:1383199 [0] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-041-023:1383131:1383199 [0] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-041-023:1383131:1383199 [0] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-041-023:1383131:1383199 [0] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-041-023:1383131:1383199 [0] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-041-023:1383131:1383199 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-041-023:1383131:1383199 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-041-023:1383131:1383199 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-041-023:1383131:1383199 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-041-023:1383131:1383199 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-041-023:1383135:1383201 [4] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-041-023:1383138:1383200 [7] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-041-023:1383138:1383200 [7] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-041-023:1383135:1383201 [4] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-041-023:1383138:1383200 [7] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-041-023:1383135:1383201 [4] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-041-023:1383138:1383200 [7] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-041-023:1383135:1383201 [4] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-041-023:1383135:1383201 [4] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-041-023:1383138:1383200 [7] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-041-023:1383135:1383201 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-041-023:1383138:1383200 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-041-023:1383135:1383201 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-041-023:1383138:1383200 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-041-023:1383135:1383201 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-041-023:1383138:1383200 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-041-023:1383135:1383201 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-041-023:1383138:1383200 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-041-023:1383135:1383201 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-041-023:1383138:1383200 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-041-023:1383134:1383202 [3] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-041-023:1383134:1383202 [3] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-041-023:1383134:1383202 [3] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-041-023:1383134:1383202 [3] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-041-023:1383134:1383202 [3] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-041-023:1383134:1383202 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-041-023:1383134:1383202 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-041-023:1383134:1383202 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-041-023:1383134:1383202 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-041-023:1383134:1383202 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-041-023:1383133:1383203 [2] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-041-023:1383133:1383203 [2] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-041-023:1383133:1383203 [2] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-041-023:1383133:1383203 [2] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-041-023:1383133:1383203 [2] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-041-023:1383133:1383203 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-041-023:1383133:1383203 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-041-023:1383133:1383203 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-041-023:1383133:1383203 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-041-023:1383133:1383203 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-041-023:1383137:1383206 [6] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-041-023:1383137:1383206 [6] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-041-023:1383137:1383206 [6] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-041-023:1383137:1383206 [6] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-041-023:1383137:1383206 [6] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-041-023:1383137:1383206 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-041-023:1383137:1383206 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-041-023:1383137:1383206 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-041-023:1383137:1383206 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-041-023:1383137:1383206 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-041-023:1383132:1383205 [1] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-041-023:1383132:1383205 [1] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-041-023:1383132:1383205 [1] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-041-023:1383132:1383205 [1] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-041-023:1383132:1383205 [1] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-041-023:1383132:1383205 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-041-023:1383132:1383205 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-041-023:1383132:1383205 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-041-023:1383132:1383205 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-041-023:1383132:1383205 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-041-023:1383136:1383204 [5] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-041-023:1383136:1383204 [5] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-041-023:1383136:1383204 [5] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-041-023:1383136:1383204 [5] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-041-023:1383136:1383204 [5] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-041-023:1383136:1383204 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-041-023:1383136:1383204 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-041-023:1383136:1383204 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-041-023:1383136:1383204 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-041-023:1383136:1383204 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-041-023:1383131:1383199 [0] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-041-023:1383131:1383199 [0] NCCL INFO Using network FasTrak +n116-041-023:1383138:1383200 [7] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-041-023:1383138:1383200 [7] NCCL INFO Using network FasTrak +n116-041-023:1383135:1383201 [4] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-041-023:1383135:1383201 [4] NCCL INFO Using network FasTrak +n116-041-023:1383134:1383202 [3] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-041-023:1383134:1383202 [3] NCCL INFO Using network FasTrak +n116-041-023:1383133:1383203 [2] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-041-023:1383133:1383203 [2] NCCL INFO Using network FasTrak +n116-041-023:1383137:1383206 [6] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-041-023:1383137:1383206 [6] NCCL INFO Using network FasTrak +n116-041-023:1383132:1383205 [1] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-041-023:1383132:1383205 [1] NCCL INFO Using network FasTrak +n116-041-023:1383136:1383204 [5] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-041-023:1383136:1383204 [5] NCCL INFO Using network FasTrak +n116-041-023:1383131:1383199 [0] NCCL INFO DMA-BUF is available on GPU device 0 +n116-041-023:1383138:1383200 [7] NCCL INFO DMA-BUF is available on GPU device 7 +n116-041-023:1383131:1383199 [0] NCCL INFO ncclCommInitRankConfig comm 0xccd6b80 rank 32 nranks 64 cudaDev 0 nvmlDev 0 busId 4000 commId 0xfb2d11625d05007a - Init START +n116-041-023:1383135:1383201 [4] NCCL INFO DMA-BUF is available on GPU device 4 +n116-041-023:1383138:1383200 [7] NCCL INFO ncclCommInitRankConfig comm 0xc365770 rank 39 nranks 64 cudaDev 7 nvmlDev 7 busId 8c000 commId 0xfb2d11625d05007a - Init START +n116-041-023:1383134:1383202 [3] NCCL INFO DMA-BUF is available on GPU device 3 +n116-041-023:1383135:1383201 [4] NCCL INFO ncclCommInitRankConfig comm 0xcdb65e0 rank 36 nranks 64 cudaDev 4 nvmlDev 4 busId 84000 commId 0xfb2d11625d05007a - Init START +n116-041-023:1383133:1383203 [2] NCCL INFO DMA-BUF is available on GPU device 2 +n116-041-023:1383134:1383202 [3] NCCL INFO ncclCommInitRankConfig comm 0xc850b30 rank 35 nranks 64 cudaDev 3 nvmlDev 3 busId c000 commId 0xfb2d11625d05007a - Init START +n116-041-023:1383133:1383203 [2] NCCL INFO ncclCommInitRankConfig comm 0xce0fdb0 rank 34 nranks 64 cudaDev 2 nvmlDev 2 busId b000 commId 0xfb2d11625d05007a - Init START +n116-041-023:1383137:1383206 [6] NCCL INFO DMA-BUF is available on GPU device 6 +n116-041-023:1383134:1383202 [3] NCCL INFO RAS client listening socket at ::1<28028> +n116-041-023:1383137:1383206 [6] NCCL INFO ncclCommInitRankConfig comm 0xb3a0c60 rank 38 nranks 64 cudaDev 6 nvmlDev 6 busId 8b000 commId 0xfb2d11625d05007a - Init START +n116-041-023:1383132:1383205 [1] NCCL INFO DMA-BUF is available on GPU device 1 +n116-041-023:1383132:1383205 [1] NCCL INFO ncclCommInitRankConfig comm 0xb86b310 rank 33 nranks 64 cudaDev 1 nvmlDev 1 busId 5000 commId 0xfb2d11625d05007a - Init START +n116-041-023:1383132:1383205 [1] NCCL INFO RAS client listening socket at ::1<28028> +n116-041-023:1383133:1383203 [2] NCCL INFO RAS client listening socket at ::1<28028> +n116-041-023:1383136:1383204 [5] NCCL INFO DMA-BUF is available on GPU device 5 +n116-041-023:1383136:1383204 [5] NCCL INFO ncclCommInitRankConfig comm 0xbecc3f0 rank 37 nranks 64 cudaDev 5 nvmlDev 5 busId 85000 commId 0xfb2d11625d05007a - Init START +n116-041-023:1383135:1383201 [4] NCCL INFO RAS client listening socket at ::1<28028> +n116-041-023:1383136:1383204 [5] NCCL INFO RAS client listening socket at ::1<28028> +n116-041-023:1383137:1383206 [6] NCCL INFO RAS client listening socket at ::1<28028> +n116-041-023:1383138:1383200 [7] NCCL INFO RAS client listening socket at ::1<28028> +n116-041-023:1383131:1383199 [0] NCCL INFO RAS client listening socket at ::1<28028> +n116-041-023:1383137:1383206 [6] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-041-023:1383137:1383206 [6] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-041-023:1383137:1383206 [6] NCCL INFO Retrieving state for FasTrak +n116-041-023:1383137:1383206 [6] NCCL INFO Initialized state 0 for FasTrak +n116-041-023:1383137:1383206 [6] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-041-023:1383137:1383206 [6] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-041-023:1383137:1383206 [6] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-041-023:1383137:1383206 [6] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-041-023:1383137:1383206 [6] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-041-023:1383137:1383206 [6] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-041-023:1383137:1383206 [6] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-041-023:1383137:1383206 [6] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-041-023:1383135:1383201 [4] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-041-023:1383135:1383201 [4] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-041-023:1383135:1383201 [4] NCCL INFO Retrieving state for FasTrak +n116-041-023:1383135:1383201 [4] NCCL INFO Initialized state 0 for FasTrak +n116-041-023:1383135:1383201 [4] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-041-023:1383135:1383201 [4] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-041-023:1383135:1383201 [4] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-041-023:1383135:1383201 [4] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-041-023:1383135:1383201 [4] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-041-023:1383135:1383201 [4] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-041-023:1383136:1383204 [5] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-041-023:1383136:1383204 [5] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-041-023:1383136:1383204 [5] NCCL INFO Retrieving state for FasTrak +n116-041-023:1383136:1383204 [5] NCCL INFO Initialized state 0 for FasTrak +n116-041-023:1383135:1383201 [4] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-041-023:1383136:1383204 [5] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-041-023:1383135:1383201 [4] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-041-023:1383136:1383204 [5] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-041-023:1383136:1383204 [5] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-041-023:1383136:1383204 [5] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-041-023:1383134:1383202 [3] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-041-023:1383134:1383202 [3] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-041-023:1383134:1383202 [3] NCCL INFO Retrieving state for FasTrak +n116-041-023:1383134:1383202 [3] NCCL INFO Initialized state 0 for FasTrak +n116-041-023:1383136:1383204 [5] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-041-023:1383136:1383204 [5] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-041-023:1383134:1383202 [3] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-041-023:1383136:1383204 [5] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-041-023:1383134:1383202 [3] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-041-023:1383136:1383204 [5] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-041-023:1383134:1383202 [3] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-041-023:1383134:1383202 [3] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-041-023:1383133:1383203 [2] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-041-023:1383133:1383203 [2] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-041-023:1383133:1383203 [2] NCCL INFO Retrieving state for FasTrak +n116-041-023:1383133:1383203 [2] NCCL INFO Initialized state 0 for FasTrak +n116-041-023:1383134:1383202 [3] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-041-023:1383134:1383202 [3] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-041-023:1383133:1383203 [2] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-041-023:1383133:1383203 [2] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-041-023:1383134:1383202 [3] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-041-023:1383133:1383203 [2] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-041-023:1383134:1383202 [3] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-041-023:1383133:1383203 [2] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-041-023:1383138:1383200 [7] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-041-023:1383138:1383200 [7] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-041-023:1383138:1383200 [7] NCCL INFO Retrieving state for FasTrak +n116-041-023:1383138:1383200 [7] NCCL INFO Initialized state 0 for FasTrak +n116-041-023:1383133:1383203 [2] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-041-023:1383133:1383203 [2] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-041-023:1383138:1383200 [7] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-041-023:1383133:1383203 [2] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-041-023:1383138:1383200 [7] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-041-023:1383133:1383203 [2] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-041-023:1383138:1383200 [7] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-041-023:1383138:1383200 [7] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-041-023:1383138:1383200 [7] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-041-023:1383138:1383200 [7] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-041-023:1383138:1383200 [7] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-041-023:1383138:1383200 [7] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-041-023:1383132:1383205 [1] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-041-023:1383132:1383205 [1] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-041-023:1383132:1383205 [1] NCCL INFO Retrieving state for FasTrak +n116-041-023:1383132:1383205 [1] NCCL INFO Initialized state 0 for FasTrak +n116-041-023:1383132:1383205 [1] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-041-023:1383132:1383205 [1] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-041-023:1383132:1383205 [1] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-041-023:1383132:1383205 [1] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-041-023:1383132:1383205 [1] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-041-023:1383132:1383205 [1] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-041-023:1383132:1383205 [1] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-041-023:1383132:1383205 [1] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-041-023:1383131:1383199 [0] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-041-023:1383131:1383199 [0] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-041-023:1383131:1383199 [0] NCCL INFO Retrieving state for FasTrak +n116-041-023:1383131:1383199 [0] NCCL INFO Initialized state 0 for FasTrak +n116-041-023:1383131:1383199 [0] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-041-023:1383131:1383199 [0] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-041-023:1383131:1383199 [0] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-041-023:1383131:1383199 [0] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-041-023:1383131:1383199 [0] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-041-023:1383131:1383199 [0] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-041-023:1383131:1383199 [0] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-041-023:1383131:1383199 [0] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-041-023:1383131:1383199 [0] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-041-023:1383138:1383200 [7] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-041-023:1383136:1383204 [5] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-041-023:1383134:1383202 [3] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-041-023:1383137:1383206 [6] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-041-023:1383132:1383205 [1] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-041-023:1383131:1383199 [0] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-041-023:1383131:1383199 [0] NCCL INFO CPU/0-0 (1/1/3) +n116-041-023:1383131:1383199 [0] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-041-023:1383131:1383199 [0] NCCL INFO + PCI[24.0] - GPU/0-4000 (32) +n116-041-023:1383131:1383199 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383131:1383199 [0] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-041-023:1383131:1383199 [0] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-041-023:1383131:1383199 [0] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-041-023:1383131:1383199 [0] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-041-023:1383131:1383199 [0] NCCL INFO + PCI[24.0] - GPU/0-5000 (33) +n116-041-023:1383131:1383199 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383131:1383199 [0] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-041-023:1383131:1383199 [0] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-041-023:1383131:1383199 [0] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-041-023:1383131:1383199 [0] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-041-023:1383131:1383199 [0] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-041-023:1383131:1383199 [0] NCCL INFO + PCI[24.0] - GPU/0-b000 (34) +n116-041-023:1383131:1383199 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383131:1383199 [0] NCCL INFO + PCI[24.0] - GPU/0-c000 (35) +n116-041-023:1383131:1383199 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383131:1383199 [0] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-041-023:1383131:1383199 [0] NCCL INFO CPU/0-1 (1/1/3) +n116-041-023:1383131:1383199 [0] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-041-023:1383131:1383199 [0] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-041-023:1383131:1383199 [0] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-041-023:1383131:1383199 [0] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-041-023:1383131:1383199 [0] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-041-023:1383131:1383199 [0] NCCL INFO + PCI[24.0] - GPU/0-84000 (36) +n116-041-023:1383131:1383199 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383131:1383199 [0] NCCL INFO + PCI[24.0] - GPU/0-85000 (37) +n116-041-023:1383131:1383199 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383131:1383199 [0] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-041-023:1383131:1383199 [0] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-041-023:1383131:1383199 [0] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-041-023:1383131:1383199 [0] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-041-023:1383131:1383199 [0] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-041-023:1383131:1383199 [0] NCCL INFO + PCI[24.0] - GPU/0-8b000 (38) +n116-041-023:1383131:1383199 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383131:1383199 [0] NCCL INFO + PCI[24.0] - GPU/0-8c000 (39) +n116-041-023:1383131:1383199 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383131:1383199 [0] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-041-023:1383131:1383199 [0] NCCL INFO ========================================== +n116-041-023:1383131:1383199 [0] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-023:1383131:1383199 [0] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-023:1383131:1383199 [0] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-023:1383133:1383203 [2] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-041-023:1383131:1383199 [0] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-023:1383131:1383199 [0] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-023:1383131:1383199 [0] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-023:1383131:1383199 [0] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-023:1383131:1383199 [0] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-023:1383131:1383199 [0] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-023:1383131:1383199 [0] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-023:1383131:1383199 [0] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-023:1383131:1383199 [0] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-023:1383131:1383199 [0] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-023:1383131:1383199 [0] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-023:1383131:1383199 [0] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-041-023:1383131:1383199 [0] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-041-023:1383131:1383199 [0] NCCL INFO Setting affinity for GPU 0 to 07ffffff,ffffff00,00000000,0007ffff,ffffffff +n116-041-023:1383131:1383199 [0] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-041-023:1383131:1383199 [0] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-041-023:1383135:1383201 [4] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-041-023:1383138:1383200 [7] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-041-023:1383138:1383200 [7] NCCL INFO CPU/0-0 (1/1/3) +n116-041-023:1383138:1383200 [7] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-041-023:1383138:1383200 [7] NCCL INFO + PCI[24.0] - GPU/0-4000 (32) +n116-041-023:1383138:1383200 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383138:1383200 [7] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-041-023:1383138:1383200 [7] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-041-023:1383138:1383200 [7] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-041-023:1383138:1383200 [7] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-041-023:1383138:1383200 [7] NCCL INFO + PCI[24.0] - GPU/0-5000 (33) +n116-041-023:1383138:1383200 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383138:1383200 [7] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-041-023:1383138:1383200 [7] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-041-023:1383138:1383200 [7] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-041-023:1383138:1383200 [7] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-041-023:1383138:1383200 [7] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-041-023:1383138:1383200 [7] NCCL INFO + PCI[24.0] - GPU/0-b000 (34) +n116-041-023:1383138:1383200 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383138:1383200 [7] NCCL INFO + PCI[24.0] - GPU/0-c000 (35) +n116-041-023:1383138:1383200 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383138:1383200 [7] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-041-023:1383138:1383200 [7] NCCL INFO CPU/0-1 (1/1/3) +n116-041-023:1383138:1383200 [7] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-041-023:1383138:1383200 [7] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-041-023:1383138:1383200 [7] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-041-023:1383138:1383200 [7] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-041-023:1383138:1383200 [7] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-041-023:1383138:1383200 [7] NCCL INFO + PCI[24.0] - GPU/0-84000 (36) +n116-041-023:1383138:1383200 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383138:1383200 [7] NCCL INFO + PCI[24.0] - GPU/0-85000 (37) +n116-041-023:1383138:1383200 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383138:1383200 [7] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-041-023:1383138:1383200 [7] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-041-023:1383138:1383200 [7] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-041-023:1383138:1383200 [7] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-041-023:1383138:1383200 [7] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-041-023:1383138:1383200 [7] NCCL INFO + PCI[24.0] - GPU/0-8b000 (38) +n116-041-023:1383138:1383200 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383138:1383200 [7] NCCL INFO + PCI[24.0] - GPU/0-8c000 (39) +n116-041-023:1383138:1383200 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383138:1383200 [7] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-041-023:1383138:1383200 [7] NCCL INFO ========================================== +n116-041-023:1383138:1383200 [7] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-023:1383138:1383200 [7] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-023:1383138:1383200 [7] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-023:1383138:1383200 [7] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-023:1383138:1383200 [7] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-023:1383138:1383200 [7] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-023:1383138:1383200 [7] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-023:1383138:1383200 [7] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-023:1383138:1383200 [7] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-023:1383138:1383200 [7] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-023:1383138:1383200 [7] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-023:1383138:1383200 [7] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-023:1383138:1383200 [7] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-023:1383138:1383200 [7] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-023:1383138:1383200 [7] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-041-023:1383138:1383200 [7] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-041-023:1383138:1383200 [7] NCCL INFO Setting affinity for GPU 7 to 7fff,ffffffff,f0000000,0000007f,ffffffff,fff00000,00000000 +n116-041-023:1383138:1383200 [7] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-041-023:1383138:1383200 [7] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-041-023:1383136:1383204 [5] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-041-023:1383136:1383204 [5] NCCL INFO CPU/0-0 (1/1/3) +n116-041-023:1383136:1383204 [5] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-041-023:1383136:1383204 [5] NCCL INFO + PCI[24.0] - GPU/0-4000 (32) +n116-041-023:1383136:1383204 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383136:1383204 [5] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-041-023:1383136:1383204 [5] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-041-023:1383136:1383204 [5] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-041-023:1383136:1383204 [5] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-041-023:1383136:1383204 [5] NCCL INFO + PCI[24.0] - GPU/0-5000 (33) +n116-041-023:1383136:1383204 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383136:1383204 [5] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-041-023:1383136:1383204 [5] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-041-023:1383136:1383204 [5] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-041-023:1383136:1383204 [5] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-041-023:1383136:1383204 [5] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-041-023:1383136:1383204 [5] NCCL INFO + PCI[24.0] - GPU/0-b000 (34) +n116-041-023:1383136:1383204 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383136:1383204 [5] NCCL INFO + PCI[24.0] - GPU/0-c000 (35) +n116-041-023:1383136:1383204 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383136:1383204 [5] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-041-023:1383136:1383204 [5] NCCL INFO CPU/0-1 (1/1/3) +n116-041-023:1383136:1383204 [5] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-041-023:1383136:1383204 [5] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-041-023:1383136:1383204 [5] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-041-023:1383136:1383204 [5] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-041-023:1383136:1383204 [5] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-041-023:1383136:1383204 [5] NCCL INFO + PCI[24.0] - GPU/0-84000 (36) +n116-041-023:1383136:1383204 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383136:1383204 [5] NCCL INFO + PCI[24.0] - GPU/0-85000 (37) +n116-041-023:1383136:1383204 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383136:1383204 [5] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-041-023:1383136:1383204 [5] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-041-023:1383136:1383204 [5] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-041-023:1383136:1383204 [5] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-041-023:1383136:1383204 [5] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-041-023:1383136:1383204 [5] NCCL INFO + PCI[24.0] - GPU/0-8b000 (38) +n116-041-023:1383136:1383204 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383136:1383204 [5] NCCL INFO + PCI[24.0] - GPU/0-8c000 (39) +n116-041-023:1383136:1383204 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383136:1383204 [5] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-041-023:1383136:1383204 [5] NCCL INFO ========================================== +n116-041-023:1383136:1383204 [5] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-023:1383136:1383204 [5] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-023:1383136:1383204 [5] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-023:1383136:1383204 [5] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-023:1383134:1383202 [3] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-041-023:1383134:1383202 [3] NCCL INFO CPU/0-0 (1/1/3) +n116-041-023:1383134:1383202 [3] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-041-023:1383134:1383202 [3] NCCL INFO + PCI[24.0] - GPU/0-4000 (32) +n116-041-023:1383136:1383204 [5] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-023:1383134:1383202 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383134:1383202 [3] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-041-023:1383134:1383202 [3] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-041-023:1383136:1383204 [5] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-023:1383134:1383202 [3] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-041-023:1383134:1383202 [3] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-041-023:1383134:1383202 [3] NCCL INFO + PCI[24.0] - GPU/0-5000 (33) +n116-041-023:1383134:1383202 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383134:1383202 [3] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-041-023:1383134:1383202 [3] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-041-023:1383134:1383202 [3] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-041-023:1383134:1383202 [3] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-041-023:1383136:1383204 [5] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-023:1383134:1383202 [3] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-041-023:1383134:1383202 [3] NCCL INFO + PCI[24.0] - GPU/0-b000 (34) +n116-041-023:1383134:1383202 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383134:1383202 [3] NCCL INFO + PCI[24.0] - GPU/0-c000 (35) +n116-041-023:1383134:1383202 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383134:1383202 [3] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-041-023:1383134:1383202 [3] NCCL INFO CPU/0-1 (1/1/3) +n116-041-023:1383136:1383204 [5] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-023:1383134:1383202 [3] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-041-023:1383134:1383202 [3] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-041-023:1383134:1383202 [3] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-041-023:1383134:1383202 [3] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-041-023:1383134:1383202 [3] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-041-023:1383134:1383202 [3] NCCL INFO + PCI[24.0] - GPU/0-84000 (36) +n116-041-023:1383136:1383204 [5] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-023:1383134:1383202 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383134:1383202 [3] NCCL INFO + PCI[24.0] - GPU/0-85000 (37) +n116-041-023:1383134:1383202 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383134:1383202 [3] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-041-023:1383134:1383202 [3] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-041-023:1383134:1383202 [3] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-041-023:1383134:1383202 [3] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-041-023:1383136:1383204 [5] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-023:1383134:1383202 [3] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-041-023:1383134:1383202 [3] NCCL INFO + PCI[24.0] - GPU/0-8b000 (38) +n116-041-023:1383134:1383202 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383134:1383202 [3] NCCL INFO + PCI[24.0] - GPU/0-8c000 (39) +n116-041-023:1383134:1383202 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383134:1383202 [3] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-041-023:1383134:1383202 [3] NCCL INFO ========================================== +n116-041-023:1383136:1383204 [5] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-023:1383136:1383204 [5] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-023:1383134:1383202 [3] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-023:1383136:1383204 [5] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-023:1383134:1383202 [3] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-023:1383134:1383202 [3] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-023:1383136:1383204 [5] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-023:1383136:1383204 [5] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-041-023:1383134:1383202 [3] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-023:1383136:1383204 [5] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-041-023:1383134:1383202 [3] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-023:1383136:1383204 [5] NCCL INFO Setting affinity for GPU 5 to 7fff,ffffffff,f0000000,0000007f,ffffffff,fff00000,00000000 +n116-041-023:1383134:1383202 [3] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-023:1383136:1383204 [5] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-041-023:1383134:1383202 [3] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-023:1383134:1383202 [3] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-023:1383134:1383202 [3] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-023:1383136:1383204 [5] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-041-023:1383134:1383202 [3] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-023:1383137:1383206 [6] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-041-023:1383134:1383202 [3] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-023:1383137:1383206 [6] NCCL INFO CPU/0-0 (1/1/3) +n116-041-023:1383134:1383202 [3] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-023:1383137:1383206 [6] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-041-023:1383137:1383206 [6] NCCL INFO + PCI[24.0] - GPU/0-4000 (32) +n116-041-023:1383137:1383206 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383137:1383206 [6] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-041-023:1383137:1383206 [6] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-041-023:1383134:1383202 [3] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-023:1383137:1383206 [6] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-041-023:1383137:1383206 [6] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-041-023:1383137:1383206 [6] NCCL INFO + PCI[24.0] - GPU/0-5000 (33) +n116-041-023:1383137:1383206 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383137:1383206 [6] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-041-023:1383137:1383206 [6] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-041-023:1383134:1383202 [3] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-023:1383137:1383206 [6] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-041-023:1383137:1383206 [6] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-041-023:1383137:1383206 [6] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-041-023:1383137:1383206 [6] NCCL INFO + PCI[24.0] - GPU/0-b000 (34) +n116-041-023:1383137:1383206 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383137:1383206 [6] NCCL INFO + PCI[24.0] - GPU/0-c000 (35) +n116-041-023:1383134:1383202 [3] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-041-023:1383137:1383206 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383137:1383206 [6] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-041-023:1383137:1383206 [6] NCCL INFO CPU/0-1 (1/1/3) +n116-041-023:1383137:1383206 [6] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-041-023:1383137:1383206 [6] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-041-023:1383137:1383206 [6] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-041-023:1383134:1383202 [3] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-041-023:1383137:1383206 [6] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-041-023:1383137:1383206 [6] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-041-023:1383137:1383206 [6] NCCL INFO + PCI[24.0] - GPU/0-84000 (36) +n116-041-023:1383137:1383206 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383137:1383206 [6] NCCL INFO + PCI[24.0] - GPU/0-85000 (37) +n116-041-023:1383134:1383202 [3] NCCL INFO Setting affinity for GPU 3 to 07ffffff,ffffff00,00000000,0007ffff,ffffffff +n116-041-023:1383137:1383206 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383137:1383206 [6] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-041-023:1383137:1383206 [6] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-041-023:1383137:1383206 [6] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-041-023:1383137:1383206 [6] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-041-023:1383137:1383206 [6] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-041-023:1383137:1383206 [6] NCCL INFO + PCI[24.0] - GPU/0-8b000 (38) +n116-041-023:1383137:1383206 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383137:1383206 [6] NCCL INFO + PCI[24.0] - GPU/0-8c000 (39) +n116-041-023:1383137:1383206 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383137:1383206 [6] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-041-023:1383137:1383206 [6] NCCL INFO ========================================== +n116-041-023:1383137:1383206 [6] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-023:1383137:1383206 [6] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-023:1383134:1383202 [3] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-041-023:1383137:1383206 [6] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-023:1383137:1383206 [6] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-023:1383132:1383205 [1] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-041-023:1383137:1383206 [6] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-023:1383132:1383205 [1] NCCL INFO CPU/0-0 (1/1/3) +n116-041-023:1383132:1383205 [1] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-041-023:1383134:1383202 [3] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-041-023:1383132:1383205 [1] NCCL INFO + PCI[24.0] - GPU/0-4000 (32) +n116-041-023:1383137:1383206 [6] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-023:1383132:1383205 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383132:1383205 [1] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-041-023:1383132:1383205 [1] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-041-023:1383132:1383205 [1] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-041-023:1383137:1383206 [6] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-023:1383132:1383205 [1] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-041-023:1383132:1383205 [1] NCCL INFO + PCI[24.0] - GPU/0-5000 (33) +n116-041-023:1383132:1383205 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383132:1383205 [1] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-041-023:1383132:1383205 [1] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-041-023:1383132:1383205 [1] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-041-023:1383137:1383206 [6] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-023:1383132:1383205 [1] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-041-023:1383132:1383205 [1] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-041-023:1383132:1383205 [1] NCCL INFO + PCI[24.0] - GPU/0-b000 (34) +n116-041-023:1383132:1383205 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383132:1383205 [1] NCCL INFO + PCI[24.0] - GPU/0-c000 (35) +n116-041-023:1383132:1383205 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383137:1383206 [6] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-023:1383132:1383205 [1] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-041-023:1383132:1383205 [1] NCCL INFO CPU/0-1 (1/1/3) +n116-041-023:1383132:1383205 [1] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-041-023:1383132:1383205 [1] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-041-023:1383132:1383205 [1] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-041-023:1383132:1383205 [1] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-041-023:1383137:1383206 [6] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-023:1383132:1383205 [1] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-041-023:1383132:1383205 [1] NCCL INFO + PCI[24.0] - GPU/0-84000 (36) +n116-041-023:1383132:1383205 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383132:1383205 [1] NCCL INFO + PCI[24.0] - GPU/0-85000 (37) +n116-041-023:1383132:1383205 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383132:1383205 [1] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-041-023:1383132:1383205 [1] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-041-023:1383137:1383206 [6] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-023:1383132:1383205 [1] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-041-023:1383132:1383205 [1] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-041-023:1383132:1383205 [1] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-041-023:1383132:1383205 [1] NCCL INFO + PCI[24.0] - GPU/0-8b000 (38) +n116-041-023:1383132:1383205 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383132:1383205 [1] NCCL INFO + PCI[24.0] - GPU/0-8c000 (39) +n116-041-023:1383132:1383205 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383137:1383206 [6] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-023:1383132:1383205 [1] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-041-023:1383132:1383205 [1] NCCL INFO ========================================== +n116-041-023:1383137:1383206 [6] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-023:1383132:1383205 [1] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-023:1383137:1383206 [6] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-023:1383132:1383205 [1] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-023:1383137:1383206 [6] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-041-023:1383132:1383205 [1] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-023:1383132:1383205 [1] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-023:1383137:1383206 [6] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-041-023:1383137:1383206 [6] NCCL INFO Setting affinity for GPU 6 to 7fff,ffffffff,f0000000,0000007f,ffffffff,fff00000,00000000 +n116-041-023:1383132:1383205 [1] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-023:1383132:1383205 [1] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-023:1383132:1383205 [1] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-023:1383132:1383205 [1] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-023:1383132:1383205 [1] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-023:1383137:1383206 [6] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-041-023:1383132:1383205 [1] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-023:1383132:1383205 [1] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-023:1383132:1383205 [1] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-023:1383132:1383205 [1] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-023:1383132:1383205 [1] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-023:1383132:1383205 [1] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-041-023:1383137:1383206 [6] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-041-023:1383132:1383205 [1] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-041-023:1383132:1383205 [1] NCCL INFO Setting affinity for GPU 1 to 07ffffff,ffffff00,00000000,0007ffff,ffffffff +n116-041-023:1383132:1383205 [1] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-041-023:1383132:1383205 [1] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-041-023:1383133:1383203 [2] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-041-023:1383133:1383203 [2] NCCL INFO CPU/0-0 (1/1/3) +n116-041-023:1383133:1383203 [2] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-041-023:1383133:1383203 [2] NCCL INFO + PCI[24.0] - GPU/0-4000 (32) +n116-041-023:1383133:1383203 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383133:1383203 [2] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-041-023:1383133:1383203 [2] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-041-023:1383133:1383203 [2] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-041-023:1383133:1383203 [2] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-041-023:1383133:1383203 [2] NCCL INFO + PCI[24.0] - GPU/0-5000 (33) +n116-041-023:1383133:1383203 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383133:1383203 [2] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-041-023:1383133:1383203 [2] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-041-023:1383133:1383203 [2] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-041-023:1383133:1383203 [2] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-041-023:1383133:1383203 [2] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-041-023:1383133:1383203 [2] NCCL INFO + PCI[24.0] - GPU/0-b000 (34) +n116-041-023:1383133:1383203 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383133:1383203 [2] NCCL INFO + PCI[24.0] - GPU/0-c000 (35) +n116-041-023:1383133:1383203 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383133:1383203 [2] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-041-023:1383133:1383203 [2] NCCL INFO CPU/0-1 (1/1/3) +n116-041-023:1383133:1383203 [2] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-041-023:1383133:1383203 [2] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-041-023:1383133:1383203 [2] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-041-023:1383133:1383203 [2] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-041-023:1383133:1383203 [2] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-041-023:1383133:1383203 [2] NCCL INFO + PCI[24.0] - GPU/0-84000 (36) +n116-041-023:1383133:1383203 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383133:1383203 [2] NCCL INFO + PCI[24.0] - GPU/0-85000 (37) +n116-041-023:1383133:1383203 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383133:1383203 [2] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-041-023:1383133:1383203 [2] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-041-023:1383133:1383203 [2] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-041-023:1383133:1383203 [2] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-041-023:1383133:1383203 [2] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-041-023:1383133:1383203 [2] NCCL INFO + PCI[24.0] - GPU/0-8b000 (38) +n116-041-023:1383133:1383203 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383133:1383203 [2] NCCL INFO + PCI[24.0] - GPU/0-8c000 (39) +n116-041-023:1383133:1383203 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383133:1383203 [2] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-041-023:1383133:1383203 [2] NCCL INFO ========================================== +n116-041-023:1383133:1383203 [2] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-023:1383133:1383203 [2] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-023:1383133:1383203 [2] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-023:1383133:1383203 [2] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-023:1383133:1383203 [2] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-023:1383133:1383203 [2] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-023:1383133:1383203 [2] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-023:1383133:1383203 [2] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-023:1383133:1383203 [2] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-023:1383133:1383203 [2] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-023:1383133:1383203 [2] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-023:1383133:1383203 [2] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-023:1383133:1383203 [2] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-023:1383133:1383203 [2] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-023:1383133:1383203 [2] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-041-023:1383133:1383203 [2] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-041-023:1383133:1383203 [2] NCCL INFO Setting affinity for GPU 2 to 07ffffff,ffffff00,00000000,0007ffff,ffffffff +n116-041-023:1383133:1383203 [2] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-041-023:1383133:1383203 [2] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-041-023:1383135:1383201 [4] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-041-023:1383135:1383201 [4] NCCL INFO CPU/0-0 (1/1/3) +n116-041-023:1383135:1383201 [4] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-041-023:1383135:1383201 [4] NCCL INFO + PCI[24.0] - GPU/0-4000 (32) +n116-041-023:1383135:1383201 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383135:1383201 [4] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-041-023:1383135:1383201 [4] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-041-023:1383135:1383201 [4] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-041-023:1383135:1383201 [4] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-041-023:1383135:1383201 [4] NCCL INFO + PCI[24.0] - GPU/0-5000 (33) +n116-041-023:1383135:1383201 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383135:1383201 [4] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-041-023:1383135:1383201 [4] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-041-023:1383135:1383201 [4] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-041-023:1383135:1383201 [4] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-041-023:1383135:1383201 [4] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-041-023:1383135:1383201 [4] NCCL INFO + PCI[24.0] - GPU/0-b000 (34) +n116-041-023:1383135:1383201 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383135:1383201 [4] NCCL INFO + PCI[24.0] - GPU/0-c000 (35) +n116-041-023:1383135:1383201 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383135:1383201 [4] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-041-023:1383135:1383201 [4] NCCL INFO CPU/0-1 (1/1/3) +n116-041-023:1383135:1383201 [4] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-041-023:1383135:1383201 [4] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-041-023:1383135:1383201 [4] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-041-023:1383135:1383201 [4] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-041-023:1383135:1383201 [4] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-041-023:1383135:1383201 [4] NCCL INFO + PCI[24.0] - GPU/0-84000 (36) +n116-041-023:1383135:1383201 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383135:1383201 [4] NCCL INFO + PCI[24.0] - GPU/0-85000 (37) +n116-041-023:1383135:1383201 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383135:1383201 [4] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-041-023:1383135:1383201 [4] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-041-023:1383135:1383201 [4] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-041-023:1383135:1383201 [4] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-041-023:1383135:1383201 [4] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-041-023:1383135:1383201 [4] NCCL INFO + PCI[24.0] - GPU/0-8b000 (38) +n116-041-023:1383135:1383201 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383135:1383201 [4] NCCL INFO + PCI[24.0] - GPU/0-8c000 (39) +n116-041-023:1383135:1383201 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-023:1383135:1383201 [4] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-041-023:1383135:1383201 [4] NCCL INFO ========================================== +n116-041-023:1383135:1383201 [4] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-023:1383135:1383201 [4] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-023:1383135:1383201 [4] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-023:1383135:1383201 [4] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-023:1383135:1383201 [4] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-023:1383135:1383201 [4] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-023:1383135:1383201 [4] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-023:1383135:1383201 [4] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-023:1383135:1383201 [4] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-023:1383135:1383201 [4] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-023:1383135:1383201 [4] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-023:1383135:1383201 [4] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-023:1383135:1383201 [4] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-023:1383135:1383201 [4] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-023:1383135:1383201 [4] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-041-023:1383135:1383201 [4] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-041-023:1383135:1383201 [4] NCCL INFO Setting affinity for GPU 4 to 7fff,ffffffff,f0000000,0000007f,ffffffff,fff00000,00000000 +n116-041-023:1383135:1383201 [4] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-041-023:1383135:1383201 [4] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-041-023:1383132:1383205 [1] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-023:1383132:1383205 [1] NCCL INFO 0 : NET/0-0 GPU/32 GPU/39 GPU/38 GPU/37 GPU/36 GPU/35 GPU/34 GPU/33 NET/0-0 +n116-041-023:1383132:1383205 [1] NCCL INFO 1 : NET/0-1 GPU/33 GPU/34 GPU/35 GPU/36 GPU/37 GPU/38 GPU/39 GPU/32 NET/0-1 +n116-041-023:1383132:1383205 [1] NCCL INFO 2 : NET/0-2 GPU/34 GPU/33 GPU/32 GPU/39 GPU/38 GPU/37 GPU/36 GPU/35 NET/0-2 +n116-041-023:1383132:1383205 [1] NCCL INFO 3 : NET/0-3 GPU/35 GPU/36 GPU/37 GPU/38 GPU/39 GPU/32 GPU/33 GPU/34 NET/0-3 +n116-041-023:1383132:1383205 [1] NCCL INFO 4 : NET/0-4 GPU/36 GPU/35 GPU/34 GPU/33 GPU/32 GPU/39 GPU/38 GPU/37 NET/0-4 +n116-041-023:1383132:1383205 [1] NCCL INFO 5 : NET/0-5 GPU/37 GPU/38 GPU/39 GPU/32 GPU/33 GPU/34 GPU/35 GPU/36 NET/0-5 +n116-041-023:1383132:1383205 [1] NCCL INFO 6 : NET/0-6 GPU/38 GPU/37 GPU/36 GPU/35 GPU/34 GPU/33 GPU/32 GPU/39 NET/0-6 +n116-041-023:1383132:1383205 [1] NCCL INFO 7 : NET/0-7 GPU/39 GPU/32 GPU/33 GPU/34 GPU/35 GPU/36 GPU/37 GPU/38 NET/0-7 +n116-041-023:1383131:1383199 [0] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-023:1383131:1383199 [0] NCCL INFO 0 : NET/0-0 GPU/32 GPU/39 GPU/38 GPU/37 GPU/36 GPU/35 GPU/34 GPU/33 NET/0-0 +n116-041-023:1383131:1383199 [0] NCCL INFO 1 : NET/0-1 GPU/33 GPU/34 GPU/35 GPU/36 GPU/37 GPU/38 GPU/39 GPU/32 NET/0-1 +n116-041-023:1383131:1383199 [0] NCCL INFO 2 : NET/0-2 GPU/34 GPU/33 GPU/32 GPU/39 GPU/38 GPU/37 GPU/36 GPU/35 NET/0-2 +n116-041-023:1383131:1383199 [0] NCCL INFO 3 : NET/0-3 GPU/35 GPU/36 GPU/37 GPU/38 GPU/39 GPU/32 GPU/33 GPU/34 NET/0-3 +n116-041-023:1383131:1383199 [0] NCCL INFO 4 : NET/0-4 GPU/36 GPU/35 GPU/34 GPU/33 GPU/32 GPU/39 GPU/38 GPU/37 NET/0-4 +n116-041-023:1383131:1383199 [0] NCCL INFO 5 : NET/0-5 GPU/37 GPU/38 GPU/39 GPU/32 GPU/33 GPU/34 GPU/35 GPU/36 NET/0-5 +n116-041-023:1383131:1383199 [0] NCCL INFO 6 : NET/0-6 GPU/38 GPU/37 GPU/36 GPU/35 GPU/34 GPU/33 GPU/32 GPU/39 NET/0-6 +n116-041-023:1383131:1383199 [0] NCCL INFO 7 : NET/0-7 GPU/39 GPU/32 GPU/33 GPU/34 GPU/35 GPU/36 GPU/37 GPU/38 NET/0-7 +n116-041-023:1383132:1383205 [1] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-023:1383132:1383205 [1] NCCL INFO 0 : NET/0-0 GPU/32 GPU/33 GPU/34 GPU/35 GPU/36 GPU/37 GPU/38 GPU/39 NET/0-0 +n116-041-023:1383132:1383205 [1] NCCL INFO 1 : NET/0-1 GPU/33 GPU/32 GPU/39 GPU/38 GPU/37 GPU/36 GPU/35 GPU/34 NET/0-1 +n116-041-023:1383132:1383205 [1] NCCL INFO 2 : NET/0-2 GPU/34 GPU/35 GPU/36 GPU/37 GPU/38 GPU/39 GPU/32 GPU/33 NET/0-2 +n116-041-023:1383132:1383205 [1] NCCL INFO 3 : NET/0-3 GPU/35 GPU/34 GPU/33 GPU/32 GPU/39 GPU/38 GPU/37 GPU/36 NET/0-3 +n116-041-023:1383132:1383205 [1] NCCL INFO 4 : NET/0-4 GPU/36 GPU/37 GPU/38 GPU/39 GPU/32 GPU/33 GPU/34 GPU/35 NET/0-4 +n116-041-023:1383132:1383205 [1] NCCL INFO 5 : NET/0-5 GPU/37 GPU/36 GPU/35 GPU/34 GPU/33 GPU/32 GPU/39 GPU/38 NET/0-5 +n116-041-023:1383132:1383205 [1] NCCL INFO 6 : NET/0-6 GPU/38 GPU/39 GPU/32 GPU/33 GPU/34 GPU/35 GPU/36 GPU/37 NET/0-6 +n116-041-023:1383132:1383205 [1] NCCL INFO 7 : NET/0-7 GPU/39 GPU/38 GPU/37 GPU/36 GPU/35 GPU/34 GPU/33 GPU/32 NET/0-7 +n116-041-023:1383131:1383199 [0] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-023:1383131:1383199 [0] NCCL INFO 0 : NET/0-0 GPU/32 GPU/33 GPU/34 GPU/35 GPU/36 GPU/37 GPU/38 GPU/39 NET/0-0 +n116-041-023:1383131:1383199 [0] NCCL INFO 1 : NET/0-1 GPU/33 GPU/32 GPU/39 GPU/38 GPU/37 GPU/36 GPU/35 GPU/34 NET/0-1 +n116-041-023:1383131:1383199 [0] NCCL INFO 2 : NET/0-2 GPU/34 GPU/35 GPU/36 GPU/37 GPU/38 GPU/39 GPU/32 GPU/33 NET/0-2 +n116-041-023:1383131:1383199 [0] NCCL INFO 3 : NET/0-3 GPU/35 GPU/34 GPU/33 GPU/32 GPU/39 GPU/38 GPU/37 GPU/36 NET/0-3 +n116-041-023:1383131:1383199 [0] NCCL INFO 4 : NET/0-4 GPU/36 GPU/37 GPU/38 GPU/39 GPU/32 GPU/33 GPU/34 GPU/35 NET/0-4 +n116-041-023:1383131:1383199 [0] NCCL INFO 5 : NET/0-5 GPU/37 GPU/36 GPU/35 GPU/34 GPU/33 GPU/32 GPU/39 GPU/38 NET/0-5 +n116-041-023:1383131:1383199 [0] NCCL INFO 6 : NET/0-6 GPU/38 GPU/39 GPU/32 GPU/33 GPU/34 GPU/35 GPU/36 GPU/37 NET/0-6 +n116-041-023:1383131:1383199 [0] NCCL INFO 7 : NET/0-7 GPU/39 GPU/38 GPU/37 GPU/36 GPU/35 GPU/34 GPU/33 GPU/32 NET/0-7 +n116-041-023:1383136:1383204 [5] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-023:1383136:1383204 [5] NCCL INFO 0 : NET/0-0 GPU/32 GPU/39 GPU/38 GPU/37 GPU/36 GPU/35 GPU/34 GPU/33 NET/0-0 +n116-041-023:1383136:1383204 [5] NCCL INFO 1 : NET/0-1 GPU/33 GPU/34 GPU/35 GPU/36 GPU/37 GPU/38 GPU/39 GPU/32 NET/0-1 +n116-041-023:1383136:1383204 [5] NCCL INFO 2 : NET/0-2 GPU/34 GPU/33 GPU/32 GPU/39 GPU/38 GPU/37 GPU/36 GPU/35 NET/0-2 +n116-041-023:1383136:1383204 [5] NCCL INFO 3 : NET/0-3 GPU/35 GPU/36 GPU/37 GPU/38 GPU/39 GPU/32 GPU/33 GPU/34 NET/0-3 +n116-041-023:1383136:1383204 [5] NCCL INFO 4 : NET/0-4 GPU/36 GPU/35 GPU/34 GPU/33 GPU/32 GPU/39 GPU/38 GPU/37 NET/0-4 +n116-041-023:1383136:1383204 [5] NCCL INFO 5 : NET/0-5 GPU/37 GPU/38 GPU/39 GPU/32 GPU/33 GPU/34 GPU/35 GPU/36 NET/0-5 +n116-041-023:1383136:1383204 [5] NCCL INFO 6 : NET/0-6 GPU/38 GPU/37 GPU/36 GPU/35 GPU/34 GPU/33 GPU/32 GPU/39 NET/0-6 +n116-041-023:1383136:1383204 [5] NCCL INFO 7 : NET/0-7 GPU/39 GPU/32 GPU/33 GPU/34 GPU/35 GPU/36 GPU/37 GPU/38 NET/0-7 +n116-041-023:1383136:1383204 [5] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-023:1383136:1383204 [5] NCCL INFO 0 : NET/0-0 GPU/32 GPU/33 GPU/34 GPU/35 GPU/36 GPU/37 GPU/38 GPU/39 NET/0-0 +n116-041-023:1383136:1383204 [5] NCCL INFO 1 : NET/0-1 GPU/33 GPU/32 GPU/39 GPU/38 GPU/37 GPU/36 GPU/35 GPU/34 NET/0-1 +n116-041-023:1383136:1383204 [5] NCCL INFO 2 : NET/0-2 GPU/34 GPU/35 GPU/36 GPU/37 GPU/38 GPU/39 GPU/32 GPU/33 NET/0-2 +n116-041-023:1383136:1383204 [5] NCCL INFO 3 : NET/0-3 GPU/35 GPU/34 GPU/33 GPU/32 GPU/39 GPU/38 GPU/37 GPU/36 NET/0-3 +n116-041-023:1383136:1383204 [5] NCCL INFO 4 : NET/0-4 GPU/36 GPU/37 GPU/38 GPU/39 GPU/32 GPU/33 GPU/34 GPU/35 NET/0-4 +n116-041-023:1383136:1383204 [5] NCCL INFO 5 : NET/0-5 GPU/37 GPU/36 GPU/35 GPU/34 GPU/33 GPU/32 GPU/39 GPU/38 NET/0-5 +n116-041-023:1383136:1383204 [5] NCCL INFO 6 : NET/0-6 GPU/38 GPU/39 GPU/32 GPU/33 GPU/34 GPU/35 GPU/36 GPU/37 NET/0-6 +n116-041-023:1383136:1383204 [5] NCCL INFO 7 : NET/0-7 GPU/39 GPU/38 GPU/37 GPU/36 GPU/35 GPU/34 GPU/33 GPU/32 NET/0-7 +n116-041-023:1383134:1383202 [3] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-023:1383134:1383202 [3] NCCL INFO 0 : NET/0-0 GPU/32 GPU/39 GPU/38 GPU/37 GPU/36 GPU/35 GPU/34 GPU/33 NET/0-0 +n116-041-023:1383134:1383202 [3] NCCL INFO 1 : NET/0-1 GPU/33 GPU/34 GPU/35 GPU/36 GPU/37 GPU/38 GPU/39 GPU/32 NET/0-1 +n116-041-023:1383134:1383202 [3] NCCL INFO 2 : NET/0-2 GPU/34 GPU/33 GPU/32 GPU/39 GPU/38 GPU/37 GPU/36 GPU/35 NET/0-2 +n116-041-023:1383134:1383202 [3] NCCL INFO 3 : NET/0-3 GPU/35 GPU/36 GPU/37 GPU/38 GPU/39 GPU/32 GPU/33 GPU/34 NET/0-3 +n116-041-023:1383134:1383202 [3] NCCL INFO 4 : NET/0-4 GPU/36 GPU/35 GPU/34 GPU/33 GPU/32 GPU/39 GPU/38 GPU/37 NET/0-4 +n116-041-023:1383134:1383202 [3] NCCL INFO 5 : NET/0-5 GPU/37 GPU/38 GPU/39 GPU/32 GPU/33 GPU/34 GPU/35 GPU/36 NET/0-5 +n116-041-023:1383134:1383202 [3] NCCL INFO 6 : NET/0-6 GPU/38 GPU/37 GPU/36 GPU/35 GPU/34 GPU/33 GPU/32 GPU/39 NET/0-6 +n116-041-023:1383134:1383202 [3] NCCL INFO 7 : NET/0-7 GPU/39 GPU/32 GPU/33 GPU/34 GPU/35 GPU/36 GPU/37 GPU/38 NET/0-7 +n116-041-023:1383134:1383202 [3] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-023:1383134:1383202 [3] NCCL INFO 0 : NET/0-0 GPU/32 GPU/33 GPU/34 GPU/35 GPU/36 GPU/37 GPU/38 GPU/39 NET/0-0 +n116-041-023:1383134:1383202 [3] NCCL INFO 1 : NET/0-1 GPU/33 GPU/32 GPU/39 GPU/38 GPU/37 GPU/36 GPU/35 GPU/34 NET/0-1 +n116-041-023:1383134:1383202 [3] NCCL INFO 2 : NET/0-2 GPU/34 GPU/35 GPU/36 GPU/37 GPU/38 GPU/39 GPU/32 GPU/33 NET/0-2 +n116-041-023:1383134:1383202 [3] NCCL INFO 3 : NET/0-3 GPU/35 GPU/34 GPU/33 GPU/32 GPU/39 GPU/38 GPU/37 GPU/36 NET/0-3 +n116-041-023:1383134:1383202 [3] NCCL INFO 4 : NET/0-4 GPU/36 GPU/37 GPU/38 GPU/39 GPU/32 GPU/33 GPU/34 GPU/35 NET/0-4 +n116-041-023:1383134:1383202 [3] NCCL INFO 5 : NET/0-5 GPU/37 GPU/36 GPU/35 GPU/34 GPU/33 GPU/32 GPU/39 GPU/38 NET/0-5 +n116-041-023:1383134:1383202 [3] NCCL INFO 6 : NET/0-6 GPU/38 GPU/39 GPU/32 GPU/33 GPU/34 GPU/35 GPU/36 GPU/37 NET/0-6 +n116-041-023:1383134:1383202 [3] NCCL INFO 7 : NET/0-7 GPU/39 GPU/38 GPU/37 GPU/36 GPU/35 GPU/34 GPU/33 GPU/32 NET/0-7 +n116-041-023:1383133:1383203 [2] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-023:1383133:1383203 [2] NCCL INFO 0 : NET/0-0 GPU/32 GPU/39 GPU/38 GPU/37 GPU/36 GPU/35 GPU/34 GPU/33 NET/0-0 +n116-041-023:1383133:1383203 [2] NCCL INFO 1 : NET/0-1 GPU/33 GPU/34 GPU/35 GPU/36 GPU/37 GPU/38 GPU/39 GPU/32 NET/0-1 +n116-041-023:1383133:1383203 [2] NCCL INFO 2 : NET/0-2 GPU/34 GPU/33 GPU/32 GPU/39 GPU/38 GPU/37 GPU/36 GPU/35 NET/0-2 +n116-041-023:1383133:1383203 [2] NCCL INFO 3 : NET/0-3 GPU/35 GPU/36 GPU/37 GPU/38 GPU/39 GPU/32 GPU/33 GPU/34 NET/0-3 +n116-041-023:1383133:1383203 [2] NCCL INFO 4 : NET/0-4 GPU/36 GPU/35 GPU/34 GPU/33 GPU/32 GPU/39 GPU/38 GPU/37 NET/0-4 +n116-041-023:1383133:1383203 [2] NCCL INFO 5 : NET/0-5 GPU/37 GPU/38 GPU/39 GPU/32 GPU/33 GPU/34 GPU/35 GPU/36 NET/0-5 +n116-041-023:1383133:1383203 [2] NCCL INFO 6 : NET/0-6 GPU/38 GPU/37 GPU/36 GPU/35 GPU/34 GPU/33 GPU/32 GPU/39 NET/0-6 +n116-041-023:1383133:1383203 [2] NCCL INFO 7 : NET/0-7 GPU/39 GPU/32 GPU/33 GPU/34 GPU/35 GPU/36 GPU/37 GPU/38 NET/0-7 +n116-041-023:1383133:1383203 [2] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-023:1383133:1383203 [2] NCCL INFO 0 : NET/0-0 GPU/32 GPU/33 GPU/34 GPU/35 GPU/36 GPU/37 GPU/38 GPU/39 NET/0-0 +n116-041-023:1383133:1383203 [2] NCCL INFO 1 : NET/0-1 GPU/33 GPU/32 GPU/39 GPU/38 GPU/37 GPU/36 GPU/35 GPU/34 NET/0-1 +n116-041-023:1383133:1383203 [2] NCCL INFO 2 : NET/0-2 GPU/34 GPU/35 GPU/36 GPU/37 GPU/38 GPU/39 GPU/32 GPU/33 NET/0-2 +n116-041-023:1383133:1383203 [2] NCCL INFO 3 : NET/0-3 GPU/35 GPU/34 GPU/33 GPU/32 GPU/39 GPU/38 GPU/37 GPU/36 NET/0-3 +n116-041-023:1383133:1383203 [2] NCCL INFO 4 : NET/0-4 GPU/36 GPU/37 GPU/38 GPU/39 GPU/32 GPU/33 GPU/34 GPU/35 NET/0-4 +n116-041-023:1383133:1383203 [2] NCCL INFO 5 : NET/0-5 GPU/37 GPU/36 GPU/35 GPU/34 GPU/33 GPU/32 GPU/39 GPU/38 NET/0-5 +n116-041-023:1383133:1383203 [2] NCCL INFO 6 : NET/0-6 GPU/38 GPU/39 GPU/32 GPU/33 GPU/34 GPU/35 GPU/36 GPU/37 NET/0-6 +n116-041-023:1383133:1383203 [2] NCCL INFO 7 : NET/0-7 GPU/39 GPU/38 GPU/37 GPU/36 GPU/35 GPU/34 GPU/33 GPU/32 NET/0-7 +n116-041-023:1383135:1383201 [4] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-023:1383135:1383201 [4] NCCL INFO 0 : NET/0-0 GPU/32 GPU/39 GPU/38 GPU/37 GPU/36 GPU/35 GPU/34 GPU/33 NET/0-0 +n116-041-023:1383135:1383201 [4] NCCL INFO 1 : NET/0-1 GPU/33 GPU/34 GPU/35 GPU/36 GPU/37 GPU/38 GPU/39 GPU/32 NET/0-1 +n116-041-023:1383135:1383201 [4] NCCL INFO 2 : NET/0-2 GPU/34 GPU/33 GPU/32 GPU/39 GPU/38 GPU/37 GPU/36 GPU/35 NET/0-2 +n116-041-023:1383135:1383201 [4] NCCL INFO 3 : NET/0-3 GPU/35 GPU/36 GPU/37 GPU/38 GPU/39 GPU/32 GPU/33 GPU/34 NET/0-3 +n116-041-023:1383135:1383201 [4] NCCL INFO 4 : NET/0-4 GPU/36 GPU/35 GPU/34 GPU/33 GPU/32 GPU/39 GPU/38 GPU/37 NET/0-4 +n116-041-023:1383135:1383201 [4] NCCL INFO 5 : NET/0-5 GPU/37 GPU/38 GPU/39 GPU/32 GPU/33 GPU/34 GPU/35 GPU/36 NET/0-5 +n116-041-023:1383135:1383201 [4] NCCL INFO 6 : NET/0-6 GPU/38 GPU/37 GPU/36 GPU/35 GPU/34 GPU/33 GPU/32 GPU/39 NET/0-6 +n116-041-023:1383135:1383201 [4] NCCL INFO 7 : NET/0-7 GPU/39 GPU/32 GPU/33 GPU/34 GPU/35 GPU/36 GPU/37 GPU/38 NET/0-7 +n116-041-023:1383135:1383201 [4] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-023:1383135:1383201 [4] NCCL INFO 0 : NET/0-0 GPU/32 GPU/33 GPU/34 GPU/35 GPU/36 GPU/37 GPU/38 GPU/39 NET/0-0 +n116-041-023:1383135:1383201 [4] NCCL INFO 1 : NET/0-1 GPU/33 GPU/32 GPU/39 GPU/38 GPU/37 GPU/36 GPU/35 GPU/34 NET/0-1 +n116-041-023:1383135:1383201 [4] NCCL INFO 2 : NET/0-2 GPU/34 GPU/35 GPU/36 GPU/37 GPU/38 GPU/39 GPU/32 GPU/33 NET/0-2 +n116-041-023:1383135:1383201 [4] NCCL INFO 3 : NET/0-3 GPU/35 GPU/34 GPU/33 GPU/32 GPU/39 GPU/38 GPU/37 GPU/36 NET/0-3 +n116-041-023:1383135:1383201 [4] NCCL INFO 4 : NET/0-4 GPU/36 GPU/37 GPU/38 GPU/39 GPU/32 GPU/33 GPU/34 GPU/35 NET/0-4 +n116-041-023:1383135:1383201 [4] NCCL INFO 5 : NET/0-5 GPU/37 GPU/36 GPU/35 GPU/34 GPU/33 GPU/32 GPU/39 GPU/38 NET/0-5 +n116-041-023:1383135:1383201 [4] NCCL INFO 6 : NET/0-6 GPU/38 GPU/39 GPU/32 GPU/33 GPU/34 GPU/35 GPU/36 GPU/37 NET/0-6 +n116-041-023:1383135:1383201 [4] NCCL INFO 7 : NET/0-7 GPU/39 GPU/38 GPU/37 GPU/36 GPU/35 GPU/34 GPU/33 GPU/32 NET/0-7 +n116-041-023:1383138:1383200 [7] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-023:1383138:1383200 [7] NCCL INFO 0 : NET/0-0 GPU/32 GPU/39 GPU/38 GPU/37 GPU/36 GPU/35 GPU/34 GPU/33 NET/0-0 +n116-041-023:1383138:1383200 [7] NCCL INFO 1 : NET/0-1 GPU/33 GPU/34 GPU/35 GPU/36 GPU/37 GPU/38 GPU/39 GPU/32 NET/0-1 +n116-041-023:1383138:1383200 [7] NCCL INFO 2 : NET/0-2 GPU/34 GPU/33 GPU/32 GPU/39 GPU/38 GPU/37 GPU/36 GPU/35 NET/0-2 +n116-041-023:1383138:1383200 [7] NCCL INFO 3 : NET/0-3 GPU/35 GPU/36 GPU/37 GPU/38 GPU/39 GPU/32 GPU/33 GPU/34 NET/0-3 +n116-041-023:1383138:1383200 [7] NCCL INFO 4 : NET/0-4 GPU/36 GPU/35 GPU/34 GPU/33 GPU/32 GPU/39 GPU/38 GPU/37 NET/0-4 +n116-041-023:1383138:1383200 [7] NCCL INFO 5 : NET/0-5 GPU/37 GPU/38 GPU/39 GPU/32 GPU/33 GPU/34 GPU/35 GPU/36 NET/0-5 +n116-041-023:1383138:1383200 [7] NCCL INFO 6 : NET/0-6 GPU/38 GPU/37 GPU/36 GPU/35 GPU/34 GPU/33 GPU/32 GPU/39 NET/0-6 +n116-041-023:1383138:1383200 [7] NCCL INFO 7 : NET/0-7 GPU/39 GPU/32 GPU/33 GPU/34 GPU/35 GPU/36 GPU/37 GPU/38 NET/0-7 +n116-041-023:1383138:1383200 [7] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-023:1383138:1383200 [7] NCCL INFO 0 : NET/0-0 GPU/32 GPU/33 GPU/34 GPU/35 GPU/36 GPU/37 GPU/38 GPU/39 NET/0-0 +n116-041-023:1383138:1383200 [7] NCCL INFO 1 : NET/0-1 GPU/33 GPU/32 GPU/39 GPU/38 GPU/37 GPU/36 GPU/35 GPU/34 NET/0-1 +n116-041-023:1383138:1383200 [7] NCCL INFO 2 : NET/0-2 GPU/34 GPU/35 GPU/36 GPU/37 GPU/38 GPU/39 GPU/32 GPU/33 NET/0-2 +n116-041-023:1383138:1383200 [7] NCCL INFO 3 : NET/0-3 GPU/35 GPU/34 GPU/33 GPU/32 GPU/39 GPU/38 GPU/37 GPU/36 NET/0-3 +n116-041-023:1383138:1383200 [7] NCCL INFO 4 : NET/0-4 GPU/36 GPU/37 GPU/38 GPU/39 GPU/32 GPU/33 GPU/34 GPU/35 NET/0-4 +n116-041-023:1383138:1383200 [7] NCCL INFO 5 : NET/0-5 GPU/37 GPU/36 GPU/35 GPU/34 GPU/33 GPU/32 GPU/39 GPU/38 NET/0-5 +n116-041-023:1383138:1383200 [7] NCCL INFO 6 : NET/0-6 GPU/38 GPU/39 GPU/32 GPU/33 GPU/34 GPU/35 GPU/36 GPU/37 NET/0-6 +n116-041-023:1383138:1383200 [7] NCCL INFO 7 : NET/0-7 GPU/39 GPU/38 GPU/37 GPU/36 GPU/35 GPU/34 GPU/33 GPU/32 NET/0-7 +n116-041-023:1383137:1383206 [6] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-023:1383137:1383206 [6] NCCL INFO 0 : NET/0-0 GPU/32 GPU/39 GPU/38 GPU/37 GPU/36 GPU/35 GPU/34 GPU/33 NET/0-0 +n116-041-023:1383137:1383206 [6] NCCL INFO 1 : NET/0-1 GPU/33 GPU/34 GPU/35 GPU/36 GPU/37 GPU/38 GPU/39 GPU/32 NET/0-1 +n116-041-023:1383137:1383206 [6] NCCL INFO 2 : NET/0-2 GPU/34 GPU/33 GPU/32 GPU/39 GPU/38 GPU/37 GPU/36 GPU/35 NET/0-2 +n116-041-023:1383137:1383206 [6] NCCL INFO 3 : NET/0-3 GPU/35 GPU/36 GPU/37 GPU/38 GPU/39 GPU/32 GPU/33 GPU/34 NET/0-3 +n116-041-023:1383137:1383206 [6] NCCL INFO 4 : NET/0-4 GPU/36 GPU/35 GPU/34 GPU/33 GPU/32 GPU/39 GPU/38 GPU/37 NET/0-4 +n116-041-023:1383137:1383206 [6] NCCL INFO 5 : NET/0-5 GPU/37 GPU/38 GPU/39 GPU/32 GPU/33 GPU/34 GPU/35 GPU/36 NET/0-5 +n116-041-023:1383137:1383206 [6] NCCL INFO 6 : NET/0-6 GPU/38 GPU/37 GPU/36 GPU/35 GPU/34 GPU/33 GPU/32 GPU/39 NET/0-6 +n116-041-023:1383137:1383206 [6] NCCL INFO 7 : NET/0-7 GPU/39 GPU/32 GPU/33 GPU/34 GPU/35 GPU/36 GPU/37 GPU/38 NET/0-7 +n116-041-023:1383137:1383206 [6] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-023:1383137:1383206 [6] NCCL INFO 0 : NET/0-0 GPU/32 GPU/33 GPU/34 GPU/35 GPU/36 GPU/37 GPU/38 GPU/39 NET/0-0 +n116-041-023:1383137:1383206 [6] NCCL INFO 1 : NET/0-1 GPU/33 GPU/32 GPU/39 GPU/38 GPU/37 GPU/36 GPU/35 GPU/34 NET/0-1 +n116-041-023:1383137:1383206 [6] NCCL INFO 2 : NET/0-2 GPU/34 GPU/35 GPU/36 GPU/37 GPU/38 GPU/39 GPU/32 GPU/33 NET/0-2 +n116-041-023:1383137:1383206 [6] NCCL INFO 3 : NET/0-3 GPU/35 GPU/34 GPU/33 GPU/32 GPU/39 GPU/38 GPU/37 GPU/36 NET/0-3 +n116-041-023:1383137:1383206 [6] NCCL INFO 4 : NET/0-4 GPU/36 GPU/37 GPU/38 GPU/39 GPU/32 GPU/33 GPU/34 GPU/35 NET/0-4 +n116-041-023:1383137:1383206 [6] NCCL INFO 5 : NET/0-5 GPU/37 GPU/36 GPU/35 GPU/34 GPU/33 GPU/32 GPU/39 GPU/38 NET/0-5 +n116-041-023:1383137:1383206 [6] NCCL INFO 6 : NET/0-6 GPU/38 GPU/39 GPU/32 GPU/33 GPU/34 GPU/35 GPU/36 GPU/37 NET/0-6 +n116-041-023:1383137:1383206 [6] NCCL INFO 7 : NET/0-7 GPU/39 GPU/38 GPU/37 GPU/36 GPU/35 GPU/34 GPU/33 GPU/32 NET/0-7 +n116-041-023:1383136:1383204 [5] NCCL INFO comm 0xbecc3f0 rank 37 nRanks 64 nNodes 8 localRanks 8 localRank 5 MNNVL 0 +n116-041-023:1383134:1383202 [3] NCCL INFO comm 0xc850b30 rank 35 nRanks 64 nNodes 8 localRanks 8 localRank 3 MNNVL 0 +n116-041-023:1383135:1383201 [4] NCCL INFO comm 0xcdb65e0 rank 36 nRanks 64 nNodes 8 localRanks 8 localRank 4 MNNVL 0 +n116-041-023:1383138:1383200 [7] NCCL INFO comm 0xc365770 rank 39 nRanks 64 nNodes 8 localRanks 8 localRank 7 MNNVL 0 +n116-041-023:1383137:1383206 [6] NCCL INFO comm 0xb3a0c60 rank 38 nRanks 64 nNodes 8 localRanks 8 localRank 6 MNNVL 0 +n116-041-023:1383132:1383205 [1] NCCL INFO comm 0xb86b310 rank 33 nRanks 64 nNodes 8 localRanks 8 localRank 1 MNNVL 0 +n116-041-023:1383133:1383203 [2] NCCL INFO comm 0xce0fdb0 rank 34 nRanks 64 nNodes 8 localRanks 8 localRank 2 MNNVL 0 +n116-041-023:1383134:1383202 [3] NCCL INFO Tree 2 : 34 -> 35 -> 36/18/-1 +n116-041-023:1383136:1383204 [5] NCCL INFO Tree 4 : 36 -> 37 -> 38/20/-1 +n116-041-023:1383137:1383206 [6] NCCL INFO Tree 6 : 6 -> 38 -> 39/54/-1 +n116-041-023:1383134:1383202 [3] NCCL INFO Tree 10 : 34 -> 35 -> 36/-1/-1 +n116-041-023:1383136:1383204 [5] NCCL INFO Tree 12 : 36 -> 37 -> 38/-1/-1 +n116-041-023:1383137:1383206 [6] NCCL INFO Tree 14 : 46 -> 38 -> 39/-1/-1 +n116-041-023:1383135:1383201 [4] NCCL INFO Tree 4 : 4 -> 36 -> 37/52/-1 +n116-041-023:1383136:1383204 [5] NCCL INFO Tree 5 : 5 -> 37 -> 36/53/-1 +n116-041-023:1383137:1383206 [6] NCCL INFO Tree 7 : 39 -> 38 -> 37/23/-1 +n116-041-023:1383134:1383202 [3] NCCL INFO Tree 3 : 3 -> 35 -> 34/51/-1 +n116-041-023:1383138:1383200 [7] NCCL INFO Tree 6 : 38 -> 39 -> 32/22/-1 +n116-041-023:1383131:1383199 [0] NCCL INFO comm 0xccd6b80 rank 32 nRanks 64 nNodes 8 localRanks 8 localRank 0 MNNVL 0 +n116-041-023:1383136:1383204 [5] NCCL INFO Tree 13 : 45 -> 37 -> 36/-1/-1 +n116-041-023:1383138:1383200 [7] NCCL INFO Tree 14 : 38 -> 39 -> 32/-1/-1 +n116-041-023:1383137:1383206 [6] NCCL INFO Tree 15 : 39 -> 38 -> 37/-1/-1 +n116-041-023:1383135:1383201 [4] NCCL INFO Tree 12 : 44 -> 36 -> 37/-1/-1 +n116-041-023:1383134:1383202 [3] NCCL INFO Tree 11 : 43 -> 35 -> 34/-1/-1 +n116-041-023:1383138:1383200 [7] NCCL INFO Tree 7 : 7 -> 39 -> 38/55/-1 +n116-041-023:1383135:1383201 [4] NCCL INFO Tree 5 : 37 -> 36 -> 35/21/-1 +n116-041-023:1383138:1383200 [7] NCCL INFO Tree 15 : 47 -> 39 -> 38/-1/-1 +n116-041-023:1383136:1383204 [5] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-041-023:1383135:1383201 [4] NCCL INFO Tree 13 : 37 -> 36 -> 35/-1/-1 +n116-041-023:1383134:1383202 [3] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-041-023:1383137:1383206 [6] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-041-023:1383132:1383205 [1] NCCL INFO Tree 0 : 32 -> 33 -> 34/16/-1 +n116-041-023:1383132:1383205 [1] NCCL INFO Tree 8 : 32 -> 33 -> 34/-1/-1 +n116-041-023:1383132:1383205 [1] NCCL INFO Tree 1 : 1 -> 33 -> 32/49/-1 +n116-041-023:1383138:1383200 [7] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-041-023:1383132:1383205 [1] NCCL INFO Tree 9 : 41 -> 33 -> 32/-1/-1 +n116-041-023:1383135:1383201 [4] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-041-023:1383132:1383205 [1] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-041-023:1383133:1383203 [2] NCCL INFO Tree 2 : 2 -> 34 -> 35/50/-1 +n116-041-023:1383133:1383203 [2] NCCL INFO Tree 10 : 42 -> 34 -> 35/-1/-1 +n116-041-023:1383131:1383199 [0] NCCL INFO Tree 0 : 0 -> 32 -> 33/48/-1 +n116-041-023:1383131:1383199 [0] NCCL INFO Tree 8 : 40 -> 32 -> 33/-1/-1 +n116-041-023:1383131:1383199 [0] NCCL INFO Tree 1 : 33 -> 32 -> 39/17/-1 +n116-041-023:1383131:1383199 [0] NCCL INFO Tree 9 : 33 -> 32 -> 39/-1/-1 +n116-041-023:1383133:1383203 [2] NCCL INFO Tree 3 : 35 -> 34 -> 33/19/-1 +n116-041-023:1383133:1383203 [2] NCCL INFO Tree 11 : 35 -> 34 -> 33/-1/-1 +n116-041-023:1383131:1383199 [0] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-041-023:1383133:1383203 [2] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-041-023:1383134:1383202 [3] NCCL INFO Ring 00 : 36 -> 35 -> 34 +n116-041-023:1383132:1383205 [1] NCCL INFO Ring 00 : 34 -> 33 -> 40 +n116-041-023:1383132:1383205 [1] NCCL INFO Ring 01 : 24 -> 33 -> 34 +n116-041-023:1383132:1383205 [1] NCCL INFO Ring 02 : 34 -> 33 -> 32 +n116-041-023:1383137:1383206 [6] NCCL INFO Ring 00 : 39 -> 38 -> 37 +n116-041-023:1383132:1383205 [1] NCCL INFO Ring 03 : 32 -> 33 -> 34 +n116-041-023:1383134:1383202 [3] NCCL INFO Ring 01 : 34 -> 35 -> 36 +n116-041-023:1383132:1383205 [1] NCCL INFO Ring 04 : 34 -> 33 -> 32 +n116-041-023:1383134:1383202 [3] NCCL INFO Ring 02 : 36 -> 35 -> 42 +n116-041-023:1383137:1383206 [6] NCCL INFO Ring 01 : 37 -> 38 -> 39 +n116-041-023:1383132:1383205 [1] NCCL INFO Ring 05 : 32 -> 33 -> 34 +n116-041-023:1383134:1383202 [3] NCCL INFO Ring 03 : 26 -> 35 -> 36 +n116-041-023:1383136:1383204 [5] NCCL INFO Ring 00 : 38 -> 37 -> 36 +n116-041-023:1383132:1383205 [1] NCCL INFO Ring 06 : 34 -> 33 -> 32 +n116-041-023:1383137:1383206 [6] NCCL INFO Ring 02 : 39 -> 38 -> 37 +n116-041-023:1383134:1383202 [3] NCCL INFO Ring 04 : 36 -> 35 -> 34 +n116-041-023:1383132:1383205 [1] NCCL INFO Ring 07 : 32 -> 33 -> 34 +n116-041-023:1383136:1383204 [5] NCCL INFO Ring 01 : 36 -> 37 -> 38 +n116-041-023:1383137:1383206 [6] NCCL INFO Ring 03 : 37 -> 38 -> 39 +n116-041-023:1383131:1383199 [0] NCCL INFO Ring 00 : 25 -> 32 -> 39 +n116-041-023:1383132:1383205 [1] NCCL INFO Ring 08 : 34 -> 33 -> 40 +n116-041-023:1383134:1383202 [3] NCCL INFO Ring 05 : 34 -> 35 -> 36 +n116-041-023:1383136:1383204 [5] NCCL INFO Ring 02 : 38 -> 37 -> 36 +n116-041-023:1383131:1383199 [0] NCCL INFO Ring 01 : 39 -> 32 -> 41 +n116-041-023:1383137:1383206 [6] NCCL INFO Ring 04 : 39 -> 38 -> 37 +n116-041-023:1383138:1383200 [7] NCCL INFO Ring 00 : 32 -> 39 -> 38 +n116-041-023:1383132:1383205 [1] NCCL INFO Ring 09 : 24 -> 33 -> 34 +n116-041-023:1383134:1383202 [3] NCCL INFO Ring 06 : 36 -> 35 -> 34 +n116-041-023:1383135:1383201 [4] NCCL INFO Ring 00 : 37 -> 36 -> 35 +n116-041-023:1383136:1383204 [5] NCCL INFO Ring 03 : 36 -> 37 -> 38 +n116-041-023:1383131:1383199 [0] NCCL INFO Ring 02 : 33 -> 32 -> 39 +n116-041-023:1383137:1383206 [6] NCCL INFO Ring 05 : 37 -> 38 -> 39 +n116-041-023:1383138:1383200 [7] NCCL INFO Ring 01 : 38 -> 39 -> 32 +n116-041-023:1383136:1383204 [5] NCCL INFO Ring 04 : 38 -> 37 -> 44 +n116-041-023:1383135:1383201 [4] NCCL INFO Ring 01 : 35 -> 36 -> 37 +n116-041-023:1383132:1383205 [1] NCCL INFO Ring 10 : 34 -> 33 -> 32 +n116-041-023:1383134:1383202 [3] NCCL INFO Ring 07 : 34 -> 35 -> 36 +n116-041-023:1383138:1383200 [7] NCCL INFO Ring 02 : 32 -> 39 -> 38 +n116-041-023:1383137:1383206 [6] NCCL INFO Ring 06 : 31 -> 38 -> 37 +n116-041-023:1383135:1383201 [4] NCCL INFO Ring 02 : 37 -> 36 -> 35 +n116-041-023:1383131:1383199 [0] NCCL INFO Ring 03 : 39 -> 32 -> 33 +n116-041-023:1383137:1383206 [6] NCCL INFO Ring 07 : 37 -> 38 -> 47 +n116-041-023:1383136:1383204 [5] NCCL INFO Ring 05 : 28 -> 37 -> 38 +n116-041-023:1383135:1383201 [4] NCCL INFO Ring 03 : 35 -> 36 -> 37 +n116-041-023:1383132:1383205 [1] NCCL INFO Ring 11 : 32 -> 33 -> 34 +n116-041-023:1383134:1383202 [3] NCCL INFO Ring 08 : 36 -> 35 -> 34 +n116-041-023:1383138:1383200 [7] NCCL INFO Ring 03 : 38 -> 39 -> 32 +n116-041-023:1383131:1383199 [0] NCCL INFO Ring 04 : 33 -> 32 -> 39 +n116-041-023:1383137:1383206 [6] NCCL INFO Ring 08 : 39 -> 38 -> 37 +n116-041-023:1383131:1383199 [0] NCCL INFO Ring 05 : 39 -> 32 -> 33 +n116-041-023:1383136:1383204 [5] NCCL INFO Ring 06 : 38 -> 37 -> 36 +n116-041-023:1383135:1383201 [4] NCCL INFO Ring 04 : 29 -> 36 -> 35 +n116-041-023:1383132:1383205 [1] NCCL INFO Ring 12 : 34 -> 33 -> 32 +n116-041-023:1383134:1383202 [3] NCCL INFO Ring 09 : 34 -> 35 -> 36 +n116-041-023:1383133:1383203 [2] NCCL INFO Ring 00 : 35 -> 34 -> 33 +n116-041-023:1383138:1383200 [7] NCCL INFO Ring 04 : 32 -> 39 -> 38 +n116-041-023:1383137:1383206 [6] NCCL INFO Ring 09 : 37 -> 38 -> 39 +n116-041-023:1383131:1383199 [0] NCCL INFO Ring 06 : 33 -> 32 -> 39 +n116-041-023:1383133:1383203 [2] NCCL INFO Ring 01 : 33 -> 34 -> 35 +n116-041-023:1383136:1383204 [5] NCCL INFO Ring 07 : 36 -> 37 -> 38 +n116-041-023:1383131:1383199 [0] NCCL INFO Ring 07 : 39 -> 32 -> 33 +n116-041-023:1383133:1383203 [2] NCCL INFO Ring 02 : 27 -> 34 -> 33 +n116-041-023:1383132:1383205 [1] NCCL INFO Ring 13 : 32 -> 33 -> 34 +n116-041-023:1383134:1383202 [3] NCCL INFO Ring 10 : 36 -> 35 -> 42 +n116-041-023:1383135:1383201 [4] NCCL INFO Ring 05 : 35 -> 36 -> 45 +n116-041-023:1383138:1383200 [7] NCCL INFO Ring 05 : 38 -> 39 -> 32 +n116-041-023:1383137:1383206 [6] NCCL INFO Ring 10 : 39 -> 38 -> 37 +n116-041-023:1383135:1383201 [4] NCCL INFO Ring 06 : 37 -> 36 -> 35 +n116-041-023:1383136:1383204 [5] NCCL INFO Ring 08 : 38 -> 37 -> 36 +n116-041-023:1383131:1383199 [0] NCCL INFO Ring 08 : 25 -> 32 -> 39 +n116-041-023:1383132:1383205 [1] NCCL INFO Ring 14 : 34 -> 33 -> 32 +n116-041-023:1383133:1383203 [2] NCCL INFO Ring 03 : 33 -> 34 -> 43 +n116-041-023:1383134:1383202 [3] NCCL INFO Ring 11 : 26 -> 35 -> 36 +n116-041-023:1383138:1383200 [7] NCCL INFO Ring 06 : 32 -> 39 -> 46 +n116-041-023:1383137:1383206 [6] NCCL INFO Ring 11 : 37 -> 38 -> 39 +n116-041-023:1383135:1383201 [4] NCCL INFO Ring 07 : 35 -> 36 -> 37 +n116-041-023:1383136:1383204 [5] NCCL INFO Ring 09 : 36 -> 37 -> 38 +n116-041-023:1383131:1383199 [0] NCCL INFO Ring 09 : 39 -> 32 -> 41 +n116-041-023:1383132:1383205 [1] NCCL INFO Ring 15 : 32 -> 33 -> 34 +n116-041-023:1383134:1383202 [3] NCCL INFO Ring 12 : 36 -> 35 -> 34 +n116-041-023:1383133:1383203 [2] NCCL INFO Ring 04 : 35 -> 34 -> 33 +n116-041-023:1383132:1383205 [1] NCCL INFO Trees [0] 34/16/-1->33->32 [1] 32/49/-1->33->1 [2] -1/-1/-1->33->32 [3] 32/-1/-1->33->34 [4] 34/-1/-1->33->32 [5] 32/-1/-1->33->34 [6] 34/-1/-1->33->32 [7] 32/-1/-1->33->34 [8] 34/-1/-1->33->32 [9] 32/-1/-1->33->41 [10] -1/-1/-1->33->32 [11] 32/-1/-1->33->34 [12] 34/-1/-1->33->32 [13] 32/-1/-1->33->34 [14] 34/-1/-1->33->32 [15] 32/-1/-1->33->34 +n116-041-023:1383138:1383200 [7] NCCL INFO Ring 07 : 30 -> 39 -> 32 +n116-041-023:1383137:1383206 [6] NCCL INFO Ring 12 : 39 -> 38 -> 37 +n116-041-023:1383135:1383201 [4] NCCL INFO Ring 08 : 37 -> 36 -> 35 +n116-041-023:1383136:1383204 [5] NCCL INFO Ring 10 : 38 -> 37 -> 36 +n116-041-023:1383131:1383199 [0] NCCL INFO Ring 10 : 33 -> 32 -> 39 +n116-041-023:1383138:1383200 [7] NCCL INFO Ring 08 : 32 -> 39 -> 38 +n116-041-023:1383137:1383206 [6] NCCL INFO Ring 13 : 37 -> 38 -> 39 +n116-041-023:1383135:1383201 [4] NCCL INFO Ring 09 : 35 -> 36 -> 37 +n116-041-023:1383131:1383199 [0] NCCL INFO Ring 11 : 39 -> 32 -> 33 +n116-041-023:1383134:1383202 [3] NCCL INFO Ring 13 : 34 -> 35 -> 36 +n116-041-023:1383133:1383203 [2] NCCL INFO Ring 05 : 33 -> 34 -> 35 +n116-041-023:1383132:1383205 [1] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-041-023:1383134:1383202 [3] NCCL INFO Ring 14 : 36 -> 35 -> 34 +n116-041-023:1383136:1383204 [5] NCCL INFO Ring 11 : 36 -> 37 -> 38 +n116-041-023:1383138:1383200 [7] NCCL INFO Ring 09 : 38 -> 39 -> 32 +n116-041-023:1383138:1383200 [7] NCCL INFO Ring 10 : 32 -> 39 -> 38 +n116-041-023:1383137:1383206 [6] NCCL INFO Ring 14 : 31 -> 38 -> 37 +n116-041-023:1383131:1383199 [0] NCCL INFO Ring 12 : 33 -> 32 -> 39 +n116-041-023:1383135:1383201 [4] NCCL INFO Ring 10 : 37 -> 36 -> 35 +n116-041-023:1383137:1383206 [6] NCCL INFO Ring 15 : 37 -> 38 -> 47 +n116-041-023:1383131:1383199 [0] NCCL INFO Ring 13 : 39 -> 32 -> 33 +n116-041-023:1383132:1383205 [1] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-041-023:1383134:1383202 [3] NCCL INFO Ring 15 : 34 -> 35 -> 36 +n116-041-023:1383132:1383205 [1] NCCL INFO P2P Chunksize set to 524288 +n116-041-023:1383136:1383204 [5] NCCL INFO Ring 12 : 38 -> 37 -> 44 +n116-041-023:1383134:1383202 [3] NCCL INFO Trees [0] 36/-1/-1->35->34 [1] 34/-1/-1->35->36 [2] 36/18/-1->35->34 [3] 34/51/-1->35->3 [4] -1/-1/-1->35->34 [5] 34/-1/-1->35->36 [6] 36/-1/-1->35->34 [7] 34/-1/-1->35->36 [8] 36/-1/-1->35->34 [9] 34/-1/-1->35->36 [10] 36/-1/-1->35->34 [11] 34/-1/-1->35->43 [12] -1/-1/-1->35->34 [13] 34/-1/-1->35->36 [14] 36/-1/-1->35->34 [15] 34/-1/-1->35->36 +n116-041-023:1383133:1383203 [2] NCCL INFO Ring 06 : 35 -> 34 -> 33 +n116-041-023:1383138:1383200 [7] NCCL INFO Ring 11 : 38 -> 39 -> 32 +n116-041-023:1383137:1383206 [6] NCCL INFO Trees [0] 39/-1/-1->38->37 [1] 37/-1/-1->38->39 [2] 39/-1/-1->38->37 [3] 37/-1/-1->38->39 [4] 39/-1/-1->38->37 [5] -1/-1/-1->38->39 [6] 39/54/-1->38->6 [7] 37/23/-1->38->39 [8] 39/-1/-1->38->37 [9] 37/-1/-1->38->39 [10] 39/-1/-1->38->37 [11] 37/-1/-1->38->39 [12] 39/-1/-1->38->37 [13] -1/-1/-1->38->39 [14] 39/-1/-1->38->46 [15] 37/-1/-1->38->39 +n116-041-023:1383134:1383202 [3] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-041-023:1383135:1383201 [4] NCCL INFO Ring 11 : 35 -> 36 -> 37 +n116-041-023:1383131:1383199 [0] NCCL INFO Ring 14 : 33 -> 32 -> 39 +n116-041-023:1383136:1383204 [5] NCCL INFO Ring 13 : 28 -> 37 -> 38 +n116-041-023:1383134:1383202 [3] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-041-023:1383133:1383203 [2] NCCL INFO Ring 07 : 33 -> 34 -> 35 +n116-041-023:1383136:1383204 [5] NCCL INFO Ring 14 : 38 -> 37 -> 36 +n116-041-023:1383138:1383200 [7] NCCL INFO Ring 12 : 32 -> 39 -> 38 +n116-041-023:1383137:1383206 [6] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-041-023:1383131:1383199 [0] NCCL INFO Ring 15 : 39 -> 32 -> 33 +n116-041-023:1383135:1383201 [4] NCCL INFO Ring 12 : 29 -> 36 -> 35 +n116-041-023:1383134:1383202 [3] NCCL INFO P2P Chunksize set to 524288 +n116-041-023:1383133:1383203 [2] NCCL INFO Ring 08 : 35 -> 34 -> 33 +n116-041-023:1383135:1383201 [4] NCCL INFO Ring 13 : 35 -> 36 -> 45 +n116-041-023:1383136:1383204 [5] NCCL INFO Ring 15 : 36 -> 37 -> 38 +n116-041-023:1383138:1383200 [7] NCCL INFO Ring 13 : 38 -> 39 -> 32 +n116-041-023:1383131:1383199 [0] NCCL INFO Trees [0] 33/48/-1->32->0 [1] 39/17/-1->32->33 [2] 33/-1/-1->32->39 [3] 39/-1/-1->32->33 [4] 33/-1/-1->32->39 [5] 39/-1/-1->32->33 [6] 33/-1/-1->32->39 [7] -1/-1/-1->32->33 [8] 33/-1/-1->32->40 [9] 39/-1/-1->32->33 [10] 33/-1/-1->32->39 [11] 39/-1/-1->32->33 [12] 33/-1/-1->32->39 [13] 39/-1/-1->32->33 [14] 33/-1/-1->32->39 [15] -1/-1/-1->32->33 +n116-041-023:1383137:1383206 [6] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-041-023:1383133:1383203 [2] NCCL INFO Ring 09 : 33 -> 34 -> 35 +n116-041-023:1383136:1383204 [5] NCCL INFO Trees [0] 38/-1/-1->37->36 [1] 36/-1/-1->37->38 [2] 38/-1/-1->37->36 [3] 36/-1/-1->37->38 [4] 38/20/-1->37->36 [5] 36/53/-1->37->5 [6] -1/-1/-1->37->36 [7] 36/-1/-1->37->38 [8] 38/-1/-1->37->36 [9] 36/-1/-1->37->38 [10] 38/-1/-1->37->36 [11] 36/-1/-1->37->38 [12] 38/-1/-1->37->36 [13] 36/-1/-1->37->45 [14] -1/-1/-1->37->36 [15] 36/-1/-1->37->38 +n116-041-023:1383135:1383201 [4] NCCL INFO Ring 14 : 37 -> 36 -> 35 +n116-041-023:1383138:1383200 [7] NCCL INFO Ring 14 : 32 -> 39 -> 46 +n116-041-023:1383137:1383206 [6] NCCL INFO P2P Chunksize set to 524288 +n116-041-023:1383135:1383201 [4] NCCL INFO Ring 15 : 35 -> 36 -> 37 +n116-041-023:1383133:1383203 [2] NCCL INFO Ring 10 : 27 -> 34 -> 33 +n116-041-023:1383131:1383199 [0] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-041-023:1383136:1383204 [5] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-041-023:1383138:1383200 [7] NCCL INFO Ring 15 : 30 -> 39 -> 32 +n116-041-023:1383138:1383200 [7] NCCL INFO Trees [0] -1/-1/-1->39->38 [1] 38/-1/-1->39->32 [2] 32/-1/-1->39->38 [3] 38/-1/-1->39->32 [4] 32/-1/-1->39->38 [5] 38/-1/-1->39->32 [6] 32/22/-1->39->38 [7] 38/55/-1->39->7 [8] -1/-1/-1->39->38 [9] 38/-1/-1->39->32 [10] 32/-1/-1->39->38 [11] 38/-1/-1->39->32 [12] 32/-1/-1->39->38 [13] 38/-1/-1->39->32 [14] 32/-1/-1->39->38 [15] 38/-1/-1->39->47 +n116-041-023:1383133:1383203 [2] NCCL INFO Ring 11 : 33 -> 34 -> 43 +n116-041-023:1383131:1383199 [0] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-041-023:1383135:1383201 [4] NCCL INFO Trees [0] 37/-1/-1->36->35 [1] 35/-1/-1->36->37 [2] 37/-1/-1->36->35 [3] -1/-1/-1->36->37 [4] 37/52/-1->36->4 [5] 35/21/-1->36->37 [6] 37/-1/-1->36->35 [7] 35/-1/-1->36->37 [8] 37/-1/-1->36->35 [9] 35/-1/-1->36->37 [10] 37/-1/-1->36->35 [11] -1/-1/-1->36->37 [12] 37/-1/-1->36->44 [13] 35/-1/-1->36->37 [14] 37/-1/-1->36->35 [15] 35/-1/-1->36->37 +n116-041-023:1383133:1383203 [2] NCCL INFO Ring 12 : 35 -> 34 -> 33 +n116-041-023:1383131:1383199 [0] NCCL INFO P2P Chunksize set to 524288 +n116-041-023:1383136:1383204 [5] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-041-023:1383138:1383200 [7] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-041-023:1383133:1383203 [2] NCCL INFO Ring 13 : 33 -> 34 -> 35 +n116-041-023:1383136:1383204 [5] NCCL INFO P2P Chunksize set to 524288 +n116-041-023:1383138:1383200 [7] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-041-023:1383133:1383203 [2] NCCL INFO Ring 14 : 35 -> 34 -> 33 +n116-041-023:1383135:1383201 [4] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-041-023:1383138:1383200 [7] NCCL INFO P2P Chunksize set to 524288 +n116-041-023:1383133:1383203 [2] NCCL INFO Ring 15 : 33 -> 34 -> 35 +n116-041-023:1383135:1383201 [4] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-041-023:1383133:1383203 [2] NCCL INFO Trees [0] 35/-1/-1->34->33 [1] -1/-1/-1->34->35 [2] 35/50/-1->34->2 [3] 33/19/-1->34->35 [4] 35/-1/-1->34->33 [5] 33/-1/-1->34->35 [6] 35/-1/-1->34->33 [7] 33/-1/-1->34->35 [8] 35/-1/-1->34->33 [9] -1/-1/-1->34->35 [10] 35/-1/-1->34->42 [11] 33/-1/-1->34->35 [12] 35/-1/-1->34->33 [13] 33/-1/-1->34->35 [14] 35/-1/-1->34->33 [15] 33/-1/-1->34->35 +n116-041-023:1383135:1383201 [4] NCCL INFO P2P Chunksize set to 524288 +n116-041-023:1383133:1383203 [2] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-041-023:1383133:1383203 [2] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-041-023:1383133:1383203 [2] NCCL INFO P2P Chunksize set to 524288 +n116-041-023:1383132:1383239 [1] NCCL INFO [Proxy Service] Device 1 CPU core 18 +n116-041-023:1383136:1383242 [5] NCCL INFO [Proxy Service] Device 5 CPU core 59 +n116-041-023:1383132:1383243 [1] NCCL INFO [Proxy Service UDS] Device 1 CPU core 12 +n116-041-023:1383131:1383244 [0] NCCL INFO [Proxy Service UDS] Device 0 CPU core 144 +n116-041-023:1383133:1383245 [2] NCCL INFO [Proxy Service] Device 2 CPU core 2 +n116-041-023:1383134:1383241 [3] NCCL INFO [Proxy Service] Device 3 CPU core 28 +n116-041-023:1383135:1383250 [4] NCCL INFO [Proxy Service] Device 4 CPU core 52 +n116-041-023:1383133:1383248 [2] NCCL INFO [Proxy Service UDS] Device 2 CPU core 5 +n116-041-023:1383136:1383247 [5] NCCL INFO [Proxy Service UDS] Device 5 CPU core 65 +n116-041-023:1383135:1383252 [4] NCCL INFO [Proxy Service UDS] Device 4 CPU core 183 +n116-041-023:1383138:1383251 [7] NCCL INFO [Proxy Service] Device 7 CPU core 178 +n116-041-023:1383131:1383240 [0] NCCL INFO [Proxy Service] Device 0 CPU core 140 +n116-041-023:1383134:1383246 [3] NCCL INFO [Proxy Service UDS] Device 3 CPU core 145 +n116-041-023:1383138:1383253 [7] NCCL INFO [Proxy Service UDS] Device 7 CPU core 185 +n116-041-023:1383137:1383254 [6] NCCL INFO [Proxy Service UDS] Device 6 CPU core 184 +n116-041-023:1383137:1383249 [6] NCCL INFO [Proxy Service] Device 6 CPU core 179 +n116-041-023:1383134:1383202 [3] NCCL INFO NCCL_PROTO set by environment to Simple +n116-041-023:1383134:1383202 [3] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-041-023:1383134:1383202 [3] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n116-041-023:1383134:1383202 [3] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n116-041-023:1383131:1383199 [0] NCCL INFO NCCL_PROTO set by environment to Simple +n116-041-023:1383131:1383199 [0] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-041-023:1383131:1383199 [0] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n116-041-023:1383131:1383199 [0] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n116-041-023:1383133:1383203 [2] NCCL INFO NCCL_PROTO set by environment to Simple +n116-041-023:1383133:1383203 [2] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-041-023:1383133:1383203 [2] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n116-041-023:1383133:1383203 [2] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n116-041-023:1383136:1383204 [5] NCCL INFO NCCL_PROTO set by environment to Simple +n116-041-023:1383136:1383204 [5] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-041-023:1383136:1383204 [5] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n116-041-023:1383136:1383204 [5] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n116-041-023:1383137:1383206 [6] NCCL INFO NCCL_PROTO set by environment to Simple +n116-041-023:1383137:1383206 [6] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-041-023:1383137:1383206 [6] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n116-041-023:1383137:1383206 [6] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n116-041-023:1383138:1383200 [7] NCCL INFO NCCL_PROTO set by environment to Simple +n116-041-023:1383138:1383200 [7] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-041-023:1383138:1383200 [7] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n116-041-023:1383138:1383200 [7] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n116-041-023:1383132:1383205 [1] NCCL INFO NCCL_PROTO set by environment to Simple +n116-041-023:1383132:1383205 [1] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-041-023:1383132:1383205 [1] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n116-041-023:1383132:1383205 [1] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n116-041-023:1383135:1383201 [4] NCCL INFO NCCL_PROTO set by environment to Simple +n116-041-023:1383135:1383201 [4] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-041-023:1383135:1383201 [4] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n116-041-023:1383135:1383201 [4] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n116-041-023:1383131:1383199 [0] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-041-023:1383131:1383199 [0] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-041-023:1383131:1383199 [0] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-041-023:1383131:1383199 [0] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-041-023:1383131:1383199 [0] NCCL INFO ncclCommInitRankConfig comm 0xccd6b80 rank 32 nranks 64 cudaDev 0 nvmlDev 0 busId 4000 commId 0xfb2d11625d05007a - Init COMPLETE +n116-041-023:1383131:1383199 [0] NCCL INFO Init timings - ncclCommInitRankConfig: rank 32 nranks 64 total 5.05 (kernels 0.29, alloc 1.50, bootstrap 2.38, allgathers 0.12, topo 0.65, graphs 0.08, connections 0.03, rest 0.01) +n116-041-023:1383138:1383200 [7] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-041-023:1383138:1383200 [7] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-041-023:1383138:1383200 [7] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-041-023:1383138:1383200 [7] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-041-023:1383138:1383200 [7] NCCL INFO ncclCommInitRankConfig comm 0xc365770 rank 39 nranks 64 cudaDev 7 nvmlDev 7 busId 8c000 commId 0xfb2d11625d05007a - Init COMPLETE +n116-041-023:1383132:1383205 [1] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-041-023:1383138:1383200 [7] NCCL INFO Init timings - ncclCommInitRankConfig: rank 39 nranks 64 total 5.02 (kernels 0.29, alloc 1.50, bootstrap 2.35, allgathers 0.06, topo 0.65, graphs 0.13, connections 0.03, rest 0.00) +n116-041-023:1383132:1383205 [1] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-041-023:1383132:1383205 [1] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-041-023:1383132:1383205 [1] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-041-023:1383132:1383205 [1] NCCL INFO ncclCommInitRankConfig comm 0xb86b310 rank 33 nranks 64 cudaDev 1 nvmlDev 1 busId 5000 commId 0xfb2d11625d05007a - Init COMPLETE +n116-041-023:1383132:1383205 [1] NCCL INFO Init timings - ncclCommInitRankConfig: rank 33 nranks 64 total 4.98 (kernels 0.33, alloc 1.49, bootstrap 2.29, allgathers 0.12, topo 0.65, graphs 0.08, connections 0.03, rest 0.00) +n116-041-023:1383135:1383201 [4] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-041-023:1383135:1383201 [4] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-041-023:1383135:1383201 [4] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-041-023:1383135:1383201 [4] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-041-023:1383135:1383201 [4] NCCL INFO ncclCommInitRankConfig comm 0xcdb65e0 rank 36 nranks 64 cudaDev 4 nvmlDev 4 busId 84000 commId 0xfb2d11625d05007a - Init COMPLETE +n116-041-023:1383135:1383201 [4] NCCL INFO Init timings - ncclCommInitRankConfig: rank 36 nranks 64 total 5.02 (kernels 0.29, alloc 1.52, bootstrap 2.33, allgathers 0.07, topo 0.65, graphs 0.12, connections 0.03, rest 0.00) +n116-041-023:1383134:1383202 [3] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-041-023:1383136:1383204 [5] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-041-023:1383134:1383202 [3] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-041-023:1383134:1383202 [3] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-041-023:1383136:1383204 [5] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-041-023:1383134:1383202 [3] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-041-023:1383136:1383204 [5] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-041-023:1383136:1383204 [5] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-041-023:1383134:1383202 [3] NCCL INFO ncclCommInitRankConfig comm 0xc850b30 rank 35 nranks 64 cudaDev 3 nvmlDev 3 busId c000 commId 0xfb2d11625d05007a - Init COMPLETE +n116-041-023:1383136:1383204 [5] NCCL INFO ncclCommInitRankConfig comm 0xbecc3f0 rank 37 nranks 64 cudaDev 5 nvmlDev 5 busId 85000 commId 0xfb2d11625d05007a - Init COMPLETE +n116-041-023:1383136:1383204 [5] NCCL INFO Init timings - ncclCommInitRankConfig: rank 37 nranks 64 total 4.98 (kernels 0.36, alloc 1.47, bootstrap 2.28, allgathers 0.12, topo 0.65, graphs 0.08, connections 0.03, rest 0.00) +n116-041-023:1383134:1383202 [3] NCCL INFO Init timings - ncclCommInitRankConfig: rank 35 nranks 64 total 5.00 (kernels 0.31, alloc 1.49, bootstrap 2.32, allgathers 0.11, topo 0.65, graphs 0.08, connections 0.03, rest 0.01) +n116-041-023:1383133:1383203 [2] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-041-023:1383133:1383203 [2] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-041-023:1383133:1383203 [2] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-041-023:1383133:1383203 [2] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-041-023:1383133:1383203 [2] NCCL INFO ncclCommInitRankConfig comm 0xce0fdb0 rank 34 nranks 64 cudaDev 2 nvmlDev 2 busId b000 commId 0xfb2d11625d05007a - Init COMPLETE +n116-041-023:1383133:1383203 [2] NCCL INFO Init timings - ncclCommInitRankConfig: rank 34 nranks 64 total 4.99 (kernels 0.32, alloc 1.47, bootstrap 2.31, allgathers 0.10, topo 0.65, graphs 0.09, connections 0.03, rest 0.01) +n116-041-023:1383137:1383206 [6] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-041-023:1383137:1383206 [6] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-041-023:1383137:1383206 [6] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-041-023:1383137:1383206 [6] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-041-023:1383137:1383206 [6] NCCL INFO ncclCommInitRankConfig comm 0xb3a0c60 rank 38 nranks 64 cudaDev 6 nvmlDev 6 busId 8b000 commId 0xfb2d11625d05007a - Init COMPLETE +n116-041-023:1383137:1383206 [6] NCCL INFO Init timings - ncclCommInitRankConfig: rank 38 nranks 64 total 4.98 (kernels 0.33, alloc 1.48, bootstrap 2.29, allgathers 0.06, topo 0.65, graphs 0.13, connections 0.03, rest 0.00) +n116-041-023:1383131:1383260 [0] NCCL INFO Channel 00/0 : 32[0] -> 33[1] via P2P/CUMEM +n116-041-023:1383131:1383260 [0] NCCL INFO Channel 01/0 : 32[0] -> 33[1] via P2P/CUMEM +n116-041-023:1383131:1383260 [0] NCCL INFO Channel 02/0 : 32[0] -> 33[1] via P2P/CUMEM +n116-041-023:1383131:1383260 [0] NCCL INFO Channel 03/0 : 32[0] -> 33[1] via P2P/CUMEM +n116-041-023:1383135:1383257 [4] NCCL INFO Channel 00/0 : 36[4] -> 37[5] via P2P/CUMEM +n116-041-023:1383131:1383260 [0] NCCL INFO Channel 04/0 : 32[0] -> 33[1] via P2P/CUMEM +n116-041-023:1383135:1383257 [4] NCCL INFO Channel 01/0 : 36[4] -> 37[5] via P2P/CUMEM +n116-041-023:1383131:1383260 [0] NCCL INFO Channel 05/0 : 32[0] -> 33[1] via P2P/CUMEM +n116-041-023:1383135:1383257 [4] NCCL INFO Channel 02/0 : 36[4] -> 37[5] via P2P/CUMEM +n116-041-023:1383131:1383260 [0] NCCL INFO Channel 06/0 : 32[0] -> 33[1] via P2P/CUMEM +n116-041-023:1383135:1383257 [4] NCCL INFO Channel 03/0 : 36[4] -> 37[5] via P2P/CUMEM +n116-041-023:1383131:1383260 [0] NCCL INFO Channel 07/0 : 32[0] -> 33[1] via P2P/CUMEM +n116-041-023:1383135:1383257 [4] NCCL INFO Channel 04/0 : 36[4] -> 37[5] via P2P/CUMEM +n116-041-023:1383131:1383260 [0] NCCL INFO Channel 08/0 : 32[0] -> 33[1] via P2P/CUMEM +n116-041-023:1383135:1383257 [4] NCCL INFO Channel 05/0 : 36[4] -> 37[5] via P2P/CUMEM +n116-041-023:1383133:1383256 [2] NCCL INFO Channel 00/0 : 34[2] -> 35[3] via P2P/CUMEM +n116-041-023:1383131:1383260 [0] NCCL INFO Channel 09/0 : 32[0] -> 33[1] via P2P/CUMEM +n116-041-023:1383137:1383259 [6] NCCL INFO Channel 00/0 : 38[6] -> 39[7] via P2P/CUMEM +n116-041-023:1383135:1383257 [4] NCCL INFO Channel 06/0 : 36[4] -> 37[5] via P2P/CUMEM +n116-041-023:1383133:1383256 [2] NCCL INFO Channel 01/0 : 34[2] -> 35[3] via P2P/CUMEM +n116-041-023:1383131:1383260 [0] NCCL INFO Channel 10/0 : 32[0] -> 33[1] via P2P/CUMEM +n116-041-023:1383137:1383259 [6] NCCL INFO Channel 01/0 : 38[6] -> 39[7] via P2P/CUMEM +n116-041-023:1383135:1383257 [4] NCCL INFO Channel 07/0 : 36[4] -> 37[5] via P2P/CUMEM +n116-041-023:1383133:1383256 [2] NCCL INFO Channel 02/0 : 34[2] -> 35[3] via P2P/CUMEM +n116-041-023:1383131:1383260 [0] NCCL INFO Channel 11/0 : 32[0] -> 33[1] via P2P/CUMEM +n116-041-023:1383137:1383259 [6] NCCL INFO Channel 02/0 : 38[6] -> 39[7] via P2P/CUMEM +n116-041-023:1383135:1383257 [4] NCCL INFO Channel 08/0 : 36[4] -> 37[5] via P2P/CUMEM +n116-041-023:1383133:1383256 [2] NCCL INFO Channel 03/0 : 34[2] -> 35[3] via P2P/CUMEM +n116-041-023:1383131:1383260 [0] NCCL INFO Channel 12/0 : 32[0] -> 33[1] via P2P/CUMEM +n116-041-023:1383137:1383259 [6] NCCL INFO Channel 03/0 : 38[6] -> 39[7] via P2P/CUMEM +n116-041-023:1383135:1383257 [4] NCCL INFO Channel 09/0 : 36[4] -> 37[5] via P2P/CUMEM +n116-041-023:1383133:1383256 [2] NCCL INFO Channel 04/0 : 34[2] -> 35[3] via P2P/CUMEM +n116-041-023:1383131:1383260 [0] NCCL INFO Channel 13/0 : 32[0] -> 33[1] via P2P/CUMEM +n116-041-023:1383137:1383259 [6] NCCL INFO Channel 04/0 : 38[6] -> 39[7] via P2P/CUMEM +n116-041-023:1383135:1383257 [4] NCCL INFO Channel 10/0 : 36[4] -> 37[5] via P2P/CUMEM +n116-041-023:1383133:1383256 [2] NCCL INFO Channel 05/0 : 34[2] -> 35[3] via P2P/CUMEM +n116-041-023:1383131:1383260 [0] NCCL INFO Channel 14/0 : 32[0] -> 33[1] via P2P/CUMEM +n116-041-023:1383137:1383259 [6] NCCL INFO Channel 05/0 : 38[6] -> 39[7] via P2P/CUMEM +n116-041-023:1383133:1383256 [2] NCCL INFO Channel 06/0 : 34[2] -> 35[3] via P2P/CUMEM +n116-041-023:1383131:1383260 [0] NCCL INFO Channel 15/0 : 32[0] -> 33[1] via P2P/CUMEM +n116-041-023:1383137:1383259 [6] NCCL INFO Channel 06/0 : 38[6] -> 39[7] via P2P/CUMEM +n116-041-023:1383133:1383256 [2] NCCL INFO Channel 07/0 : 34[2] -> 35[3] via P2P/CUMEM +n116-041-023:1383137:1383259 [6] NCCL INFO Channel 07/0 : 38[6] -> 39[7] via P2P/CUMEM +n116-041-023:1383133:1383256 [2] NCCL INFO Channel 08/0 : 34[2] -> 35[3] via P2P/CUMEM +n116-041-023:1383137:1383259 [6] NCCL INFO Channel 08/0 : 38[6] -> 39[7] via P2P/CUMEM +n116-041-023:1383133:1383256 [2] NCCL INFO Channel 09/0 : 34[2] -> 35[3] via P2P/CUMEM +n116-041-023:1383136:1383258 [5] NCCL INFO Channel 00/0 : 37[5] -> 38[6] via P2P/CUMEM +n116-041-023:1383137:1383259 [6] NCCL INFO Channel 09/0 : 38[6] -> 39[7] via P2P/CUMEM +n116-041-023:1383133:1383256 [2] NCCL INFO Channel 10/0 : 34[2] -> 35[3] via P2P/CUMEM +n116-041-023:1383136:1383258 [5] NCCL INFO Channel 01/0 : 37[5] -> 38[6] via P2P/CUMEM +n116-041-023:1383137:1383259 [6] NCCL INFO Channel 10/0 : 38[6] -> 39[7] via P2P/CUMEM +n116-041-023:1383133:1383256 [2] NCCL INFO Channel 11/0 : 34[2] -> 35[3] via P2P/CUMEM +n116-041-023:1383136:1383258 [5] NCCL INFO Channel 02/0 : 37[5] -> 38[6] via P2P/CUMEM +n116-041-023:1383137:1383259 [6] NCCL INFO Channel 11/0 : 38[6] -> 39[7] via P2P/CUMEM +n116-041-023:1383133:1383256 [2] NCCL INFO Channel 12/0 : 34[2] -> 35[3] via P2P/CUMEM +n116-041-023:1383136:1383258 [5] NCCL INFO Channel 03/0 : 37[5] -> 38[6] via P2P/CUMEM +n116-041-023:1383137:1383259 [6] NCCL INFO Channel 12/0 : 38[6] -> 39[7] via P2P/CUMEM +n116-041-023:1383133:1383256 [2] NCCL INFO Channel 13/0 : 34[2] -> 35[3] via P2P/CUMEM +n116-041-023:1383136:1383258 [5] NCCL INFO Channel 04/0 : 37[5] -> 38[6] via P2P/CUMEM +n116-041-023:1383137:1383259 [6] NCCL INFO Channel 13/0 : 38[6] -> 39[7] via P2P/CUMEM +n116-041-023:1383133:1383256 [2] NCCL INFO Channel 14/0 : 34[2] -> 35[3] via P2P/CUMEM +n116-041-023:1383136:1383258 [5] NCCL INFO Channel 07/0 : 37[5] -> 38[6] via P2P/CUMEM +n116-041-023:1383137:1383259 [6] NCCL INFO Channel 14/0 : 38[6] -> 39[7] via P2P/CUMEM +n116-041-023:1383133:1383256 [2] NCCL INFO Channel 15/0 : 34[2] -> 35[3] via P2P/CUMEM +n116-041-023:1383136:1383258 [5] NCCL INFO Channel 08/0 : 37[5] -> 38[6] via P2P/CUMEM +n116-041-023:1383137:1383259 [6] NCCL INFO Channel 15/0 : 38[6] -> 39[7] via P2P/CUMEM +n116-041-023:1383136:1383258 [5] NCCL INFO Channel 09/0 : 37[5] -> 38[6] via P2P/CUMEM +n116-041-023:1383135:1383257 [4] NCCL INFO Channel 11/0 : 36[4] -> 37[5] via P2P/CUMEM +n116-041-023:1383136:1383258 [5] NCCL INFO Channel 10/0 : 37[5] -> 38[6] via P2P/CUMEM +n116-041-023:1383135:1383257 [4] NCCL INFO Channel 12/0 : 36[4] -> 37[5] via P2P/CUMEM +n116-041-023:1383136:1383258 [5] NCCL INFO Channel 11/0 : 37[5] -> 38[6] via P2P/CUMEM +n116-041-023:1383135:1383257 [4] NCCL INFO Channel 13/0 : 36[4] -> 37[5] via P2P/CUMEM +n116-041-023:1383136:1383258 [5] NCCL INFO Channel 12/0 : 37[5] -> 38[6] via P2P/CUMEM +n116-041-023:1383135:1383257 [4] NCCL INFO Channel 14/0 : 36[4] -> 37[5] via P2P/CUMEM +n116-041-023:1383134:1383262 [3] NCCL INFO Channel 00/0 : 35[3] -> 36[4] via P2P/CUMEM +n116-041-023:1383136:1383258 [5] NCCL INFO Channel 15/0 : 37[5] -> 38[6] via P2P/CUMEM +n116-041-023:1383135:1383257 [4] NCCL INFO Channel 15/0 : 36[4] -> 37[5] via P2P/CUMEM +n116-041-023:1383134:1383262 [3] NCCL INFO Channel 01/0 : 35[3] -> 36[4] via P2P/CUMEM +n116-041-023:1383134:1383262 [3] NCCL INFO Channel 02/0 : 35[3] -> 36[4] via P2P/CUMEM +n116-041-023:1383134:1383262 [3] NCCL INFO Channel 05/0 : 35[3] -> 36[4] via P2P/CUMEM +n116-041-023:1383134:1383262 [3] NCCL INFO Channel 06/0 : 35[3] -> 36[4] via P2P/CUMEM +n116-041-023:1383134:1383262 [3] NCCL INFO Channel 07/0 : 35[3] -> 36[4] via P2P/CUMEM +n116-041-023:1383134:1383262 [3] NCCL INFO Channel 08/0 : 35[3] -> 36[4] via P2P/CUMEM +n116-041-023:1383134:1383262 [3] NCCL INFO Channel 09/0 : 35[3] -> 36[4] via P2P/CUMEM +n116-041-023:1383134:1383262 [3] NCCL INFO Channel 10/0 : 35[3] -> 36[4] via P2P/CUMEM +n116-041-023:1383134:1383262 [3] NCCL INFO Channel 13/0 : 35[3] -> 36[4] via P2P/CUMEM +n116-041-023:1383132:1383261 [1] NCCL INFO Channel 00/0 : 33[1] -> 34[2] via P2P/CUMEM +n116-041-023:1383134:1383262 [3] NCCL INFO Channel 14/0 : 35[3] -> 36[4] via P2P/CUMEM +n116-041-023:1383134:1383262 [3] NCCL INFO Channel 15/0 : 35[3] -> 36[4] via P2P/CUMEM +n116-041-023:1383132:1383261 [1] NCCL INFO Channel 03/0 : 33[1] -> 34[2] via P2P/CUMEM +n116-041-023:1383132:1383261 [1] NCCL INFO Channel 04/0 : 33[1] -> 34[2] via P2P/CUMEM +n116-041-023:1383132:1383261 [1] NCCL INFO Channel 05/0 : 33[1] -> 34[2] via P2P/CUMEM +n116-041-023:1383132:1383261 [1] NCCL INFO Channel 06/0 : 33[1] -> 34[2] via P2P/CUMEM +n116-041-023:1383132:1383261 [1] NCCL INFO Channel 07/0 : 33[1] -> 34[2] via P2P/CUMEM +n116-041-023:1383132:1383261 [1] NCCL INFO Channel 08/0 : 33[1] -> 34[2] via P2P/CUMEM +n116-041-023:1383132:1383261 [1] NCCL INFO Channel 11/0 : 33[1] -> 34[2] via P2P/CUMEM +n116-041-023:1383132:1383261 [1] NCCL INFO Channel 12/0 : 33[1] -> 34[2] via P2P/CUMEM +n116-041-023:1383132:1383261 [1] NCCL INFO Channel 13/0 : 33[1] -> 34[2] via P2P/CUMEM +n116-041-023:1383132:1383261 [1] NCCL INFO Channel 14/0 : 33[1] -> 34[2] via P2P/CUMEM +n116-041-023:1383132:1383261 [1] NCCL INFO Channel 15/0 : 33[1] -> 34[2] via P2P/CUMEM +n116-041-023:1383131:1383260 [0] NCCL INFO Channel 01/0 : 32[0] -> 39[7] via P2P/CUMEM +n116-041-023:1383131:1383260 [0] NCCL INFO Channel 02/0 : 32[0] -> 39[7] via P2P/CUMEM +n116-041-023:1383131:1383260 [0] NCCL INFO Channel 03/0 : 32[0] -> 39[7] via P2P/CUMEM +n116-041-023:1383131:1383260 [0] NCCL INFO Channel 04/0 : 32[0] -> 39[7] via P2P/CUMEM +n116-041-023:1383131:1383260 [0] NCCL INFO Channel 05/0 : 32[0] -> 39[7] via P2P/CUMEM +n116-041-023:1383131:1383260 [0] NCCL INFO Channel 06/0 : 32[0] -> 39[7] via P2P/CUMEM +n116-041-023:1383131:1383260 [0] NCCL INFO Channel 09/0 : 32[0] -> 39[7] via P2P/CUMEM +n116-041-023:1383131:1383260 [0] NCCL INFO Channel 10/0 : 32[0] -> 39[7] via P2P/CUMEM +n116-041-023:1383131:1383260 [0] NCCL INFO Channel 11/0 : 32[0] -> 39[7] via P2P/CUMEM +n116-041-023:1383131:1383260 [0] NCCL INFO Channel 12/0 : 32[0] -> 39[7] via P2P/CUMEM +n116-041-023:1383131:1383260 [0] NCCL INFO Channel 13/0 : 32[0] -> 39[7] via P2P/CUMEM +n116-041-023:1383131:1383260 [0] NCCL INFO Channel 14/0 : 32[0] -> 39[7] via P2P/CUMEM +n116-041-023:1383136:1383258 [5] NCCL INFO Channel 13/0 : 37[5] -> 45[5] [send] via NET/FasTrak/5/GDRDMA +n116-041-023:1383136:1383263 [5] NCCL INFO [Proxy Progress] Device 5 CPU core 64 +n116-041-023:1383137:1383259 [6] NCCL INFO Channel 14/0 : 38[6] -> 46[6] [send] via NET/FasTrak/6/GDRDMA +n116-041-023:1383137:1383264 [6] NCCL INFO [Proxy Progress] Device 6 CPU core 92 +n116-041-023:1383135:1383265 [4] NCCL INFO [Proxy Progress] Device 4 CPU core 93 +n116-041-023:1383135:1383257 [4] NCCL INFO Channel 12/0 : 36[4] -> 44[4] [send] via NET/FasTrak/4/GDRDMA +n116-041-023:1383134:1383266 [3] NCCL INFO [Proxy Progress] Device 3 CPU core 47 +n116-041-023:1383134:1383262 [3] NCCL INFO Channel 11/0 : 35[3] -> 43[3] [send] via NET/FasTrak/3/GDRDMA +n116-041-023:1383132:1383267 [1] NCCL INFO [Proxy Progress] Device 1 CPU core 49 +n116-041-023:1383132:1383261 [1] NCCL INFO Channel 09/0 : 33[1] -> 41[1] [send] via NET/FasTrak/1/GDRDMA +n116-041-023:1383133:1383268 [2] NCCL INFO [Proxy Progress] Device 2 CPU core 111 +n116-041-023:1383133:1383256 [2] NCCL INFO Channel 10/0 : 34[2] -> 42[2] [send] via NET/FasTrak/2/GDRDMA +n116-041-023:1383138:1383269 [7] NCCL INFO [Proxy Progress] Device 7 CPU core 94 +n116-041-023:1383138:1383255 [7] NCCL INFO Channel 15/0 : 39[7] -> 47[7] [send] via NET/FasTrak/7/GDRDMA +n116-041-023:1383131:1383270 [0] NCCL INFO [Proxy Progress] Device 0 CPU core 14 +n116-041-023:1383131:1383260 [0] NCCL INFO Channel 08/0 : 32[0] -> 40[0] [send] via NET/FasTrak/0/GDRDMA +n116-041-023:1383137:1383259 [6] NCCL INFO Channel 07/0 : 23[7] -> 38[6] [receive] via NET/FasTrak/7/GDRDMA +n116-041-023:1383135:1383257 [4] NCCL INFO Channel 05/0 : 21[5] -> 36[4] [receive] via NET/FasTrak/5/GDRDMA +n116-041-023:1383135:1383257 [4] NCCL INFO Channel 04/0 : 36[4] -> 52[4] [send] via NET/FasTrak/4/GDRDMA +n116-041-023:1383133:1383256 [2] NCCL INFO Channel 03/0 : 19[3] -> 34[2] [receive] via NET/FasTrak/3/GDRDMA +n116-041-023:1383137:1383259 [6] NCCL INFO Channel 06/0 : 38[6] -> 54[6] [send] via NET/FasTrak/6/GDRDMA +n116-041-023:1383133:1383256 [2] NCCL INFO Channel 02/0 : 34[2] -> 50[2] [send] via NET/FasTrak/2/GDRDMA +n116-041-023:1383131:1383260 [0] NCCL INFO Channel 01/0 : 17[1] -> 32[0] [receive] via NET/FasTrak/1/GDRDMA +n116-041-023:1383131:1383260 [0] NCCL INFO Channel 00/0 : 32[0] -> 48[0] [send] via NET/FasTrak/0/GDRDMA +n116-041-023:1383131:1383260 [0] NCCL INFO Channel 00/0 : 0[0] -> 32[0] [receive] via NET/FasTrak/0/GDRDMA +n116-041-023:1383131:1383260 [0] NCCL INFO Channel 00/0 : 32[0] -> 0[0] [send] via NET/FasTrak/0/GDRDMA +n116-041-023:1383137:1383259 [6] NCCL INFO Channel 06/0 : 6[6] -> 38[6] [receive] via NET/FasTrak/6/GDRDMA +n116-041-023:1383137:1383259 [6] NCCL INFO Channel 06/0 : 38[6] -> 6[6] [send] via NET/FasTrak/6/GDRDMA +n116-041-023:1383136:1383258 [5] NCCL INFO Channel 05/0 : 37[5] -> 53[5] [send] via NET/FasTrak/5/GDRDMA +n116-041-023:1383132:1383261 [1] NCCL INFO Channel 01/0 : 33[1] -> 49[1] [send] via NET/FasTrak/1/GDRDMA +n116-041-023:1383131:1383260 [0] NCCL INFO Channel 00/0 : 48[0] -> 32[0] [receive] via NET/FasTrak/0/GDRDMA +n116-041-023:1383135:1383257 [4] NCCL INFO Channel 04/0 : 4[4] -> 36[4] [receive] via NET/FasTrak/4/GDRDMA +n116-041-023:1383131:1383260 [0] NCCL INFO Channel 01/0 : 32[0] -> 17[1] [send] via NET/FasTrak/1/GDRDMA +n116-041-023:1383135:1383257 [4] NCCL INFO Channel 04/0 : 36[4] -> 4[4] [send] via NET/FasTrak/4/GDRDMA +n116-041-023:1383138:1383255 [7] NCCL INFO Channel 07/0 : 39[7] -> 55[7] [send] via NET/FasTrak/7/GDRDMA +n116-041-023:1383135:1383257 [4] NCCL INFO Channel 04/0 : 52[4] -> 36[4] [receive] via NET/FasTrak/4/GDRDMA +n116-041-023:1383135:1383257 [4] NCCL INFO Channel 05/0 : 36[4] -> 21[5] [send] via NET/FasTrak/5/GDRDMA +n116-041-023:1383134:1383262 [3] NCCL INFO Channel 03/0 : 35[3] -> 51[3] [send] via NET/FasTrak/3/GDRDMA +n116-041-023:1383135:1383257 [4] NCCL INFO Channel 12/0 : 44[4] -> 36[4] [receive] via NET/FasTrak/4/GDRDMA +n116-041-023:1383134:1383262 [3] NCCL INFO Channel 02/0 : 18[2] -> 35[3] [receive] via NET/FasTrak/2/GDRDMA +n116-041-023:1383136:1383258 [5] NCCL INFO Channel 04/0 : 20[4] -> 37[5] [receive] via NET/FasTrak/4/GDRDMA +n116-041-023:1383137:1383259 [6] NCCL INFO Channel 06/0 : 54[6] -> 38[6] [receive] via NET/FasTrak/6/GDRDMA +n116-041-023:1383131:1383260 [0] NCCL INFO Channel 08/0 : 40[0] -> 32[0] [receive] via NET/FasTrak/0/GDRDMA +n116-041-023:1383137:1383259 [6] NCCL INFO Channel 07/0 : 38[6] -> 23[7] [send] via NET/FasTrak/7/GDRDMA +n116-041-023:1383134:1383262 [3] NCCL INFO Channel 03/0 : 3[3] -> 35[3] [receive] via NET/FasTrak/3/GDRDMA +n116-041-023:1383134:1383262 [3] NCCL INFO Channel 03/0 : 35[3] -> 3[3] [send] via NET/FasTrak/3/GDRDMA +n116-041-023:1383134:1383262 [3] NCCL INFO Channel 02/0 : 35[3] -> 18[2] [send] via NET/FasTrak/2/GDRDMA +n116-041-023:1383136:1383258 [5] NCCL INFO Channel 05/0 : 5[5] -> 37[5] [receive] via NET/FasTrak/5/GDRDMA +n116-041-023:1383136:1383258 [5] NCCL INFO Channel 05/0 : 37[5] -> 5[5] [send] via NET/FasTrak/5/GDRDMA +n116-041-023:1383132:1383261 [1] NCCL INFO Channel 00/0 : 16[0] -> 33[1] [receive] via NET/FasTrak/0/GDRDMA +n116-041-023:1383133:1383256 [2] NCCL INFO Channel 02/0 : 2[2] -> 34[2] [receive] via NET/FasTrak/2/GDRDMA +n116-041-023:1383136:1383258 [5] NCCL INFO Channel 04/0 : 37[5] -> 20[4] [send] via NET/FasTrak/4/GDRDMA +n116-041-023:1383137:1383259 [6] NCCL INFO Channel 14/0 : 46[6] -> 38[6] [receive] via NET/FasTrak/6/GDRDMA +n116-041-023:1383133:1383256 [2] NCCL INFO Channel 02/0 : 34[2] -> 2[2] [send] via NET/FasTrak/2/GDRDMA +n116-041-023:1383132:1383261 [1] NCCL INFO Channel 01/0 : 1[1] -> 33[1] [receive] via NET/FasTrak/1/GDRDMA +n116-041-023:1383134:1383262 [3] NCCL INFO Channel 03/0 : 51[3] -> 35[3] [receive] via NET/FasTrak/3/GDRDMA +n116-041-023:1383132:1383261 [1] NCCL INFO Channel 01/0 : 33[1] -> 1[1] [send] via NET/FasTrak/1/GDRDMA +n116-041-023:1383133:1383256 [2] NCCL INFO Channel 02/0 : 50[2] -> 34[2] [receive] via NET/FasTrak/2/GDRDMA +n116-041-023:1383132:1383261 [1] NCCL INFO Channel 00/0 : 33[1] -> 16[0] [send] via NET/FasTrak/0/GDRDMA +n116-041-023:1383133:1383256 [2] NCCL INFO Channel 03/0 : 34[2] -> 19[3] [send] via NET/FasTrak/3/GDRDMA +n116-041-023:1383136:1383258 [5] NCCL INFO Channel 05/0 : 53[5] -> 37[5] [receive] via NET/FasTrak/5/GDRDMA +n116-041-023:1383132:1383261 [1] NCCL INFO Channel 01/0 : 49[1] -> 33[1] [receive] via NET/FasTrak/1/GDRDMA +n116-041-023:1383134:1383262 [3] NCCL INFO Channel 11/0 : 43[3] -> 35[3] [receive] via NET/FasTrak/3/GDRDMA +n116-041-023:1383133:1383256 [2] NCCL INFO Channel 10/0 : 42[2] -> 34[2] [receive] via NET/FasTrak/2/GDRDMA +n116-041-023:1383136:1383258 [5] NCCL INFO Channel 13/0 : 45[5] -> 37[5] [receive] via NET/FasTrak/5/GDRDMA +n116-041-023:1383132:1383261 [1] NCCL INFO Channel 09/0 : 41[1] -> 33[1] [receive] via NET/FasTrak/1/GDRDMA +n116-041-023:1383138:1383255 [7] NCCL INFO Channel 06/0 : 22[6] -> 39[7] [receive] via NET/FasTrak/6/GDRDMA +n116-041-023:1383138:1383255 [7] NCCL INFO Channel 07/0 : 7[7] -> 39[7] [receive] via NET/FasTrak/7/GDRDMA +n116-041-023:1383138:1383255 [7] NCCL INFO Channel 07/0 : 39[7] -> 7[7] [send] via NET/FasTrak/7/GDRDMA +n116-041-023:1383138:1383255 [7] NCCL INFO Channel 06/0 : 39[7] -> 22[6] [send] via NET/FasTrak/6/GDRDMA +n116-041-023:1383138:1383255 [7] NCCL INFO Channel 07/0 : 55[7] -> 39[7] [receive] via NET/FasTrak/7/GDRDMA +n116-041-023:1383138:1383255 [7] NCCL INFO Channel 15/0 : 47[7] -> 39[7] [receive] via NET/FasTrak/7/GDRDMA +n116-041-023:1383138:1383255 [7] NCCL INFO Channel 01/0 : 39[7] -> 32[0] via P2P/CUMEM +n116-041-023:1383138:1383255 [7] NCCL INFO Channel 02/0 : 39[7] -> 32[0] via P2P/CUMEM +n116-041-023:1383138:1383255 [7] NCCL INFO Channel 03/0 : 39[7] -> 32[0] via P2P/CUMEM +n116-041-023:1383138:1383255 [7] NCCL INFO Channel 04/0 : 39[7] -> 32[0] via P2P/CUMEM +n116-041-023:1383134:1383262 [3] NCCL INFO Channel 00/0 : 35[3] -> 34[2] via P2P/CUMEM +n116-041-023:1383138:1383255 [7] NCCL INFO Channel 05/0 : 39[7] -> 32[0] via P2P/CUMEM +n116-041-023:1383134:1383262 [3] NCCL INFO Channel 01/0 : 35[3] -> 34[2] via P2P/CUMEM +n116-041-023:1383138:1383255 [7] NCCL INFO Channel 06/0 : 39[7] -> 32[0] via P2P/CUMEM +n116-041-023:1383134:1383262 [3] NCCL INFO Channel 02/0 : 35[3] -> 34[2] via P2P/CUMEM +n116-041-023:1383138:1383255 [7] NCCL INFO Channel 09/0 : 39[7] -> 32[0] via P2P/CUMEM +n116-041-023:1383134:1383262 [3] NCCL INFO Channel 03/0 : 35[3] -> 34[2] via P2P/CUMEM +n116-041-023:1383136:1383258 [5] NCCL INFO Channel 00/0 : 37[5] -> 36[4] via P2P/CUMEM +n116-041-023:1383138:1383255 [7] NCCL INFO Channel 10/0 : 39[7] -> 32[0] via P2P/CUMEM +n116-041-023:1383134:1383262 [3] NCCL INFO Channel 04/0 : 35[3] -> 34[2] via P2P/CUMEM +n116-041-023:1383136:1383258 [5] NCCL INFO Channel 01/0 : 37[5] -> 36[4] via P2P/CUMEM +n116-041-023:1383138:1383255 [7] NCCL INFO Channel 11/0 : 39[7] -> 32[0] via P2P/CUMEM +n116-041-023:1383134:1383262 [3] NCCL INFO Channel 05/0 : 35[3] -> 34[2] via P2P/CUMEM +n116-041-023:1383136:1383258 [5] NCCL INFO Channel 02/0 : 37[5] -> 36[4] via P2P/CUMEM +n116-041-023:1383135:1383257 [4] NCCL INFO Channel 00/0 : 36[4] -> 35[3] via P2P/CUMEM +n116-041-023:1383134:1383262 [3] NCCL INFO Channel 06/0 : 35[3] -> 34[2] via P2P/CUMEM +n116-041-023:1383138:1383255 [7] NCCL INFO Channel 12/0 : 39[7] -> 32[0] via P2P/CUMEM +n116-041-023:1383136:1383258 [5] NCCL INFO Channel 03/0 : 37[5] -> 36[4] via P2P/CUMEM +n116-041-023:1383134:1383262 [3] NCCL INFO Channel 07/0 : 35[3] -> 34[2] via P2P/CUMEM +n116-041-023:1383138:1383255 [7] NCCL INFO Channel 13/0 : 39[7] -> 32[0] via P2P/CUMEM +n116-041-023:1383136:1383258 [5] NCCL INFO Channel 04/0 : 37[5] -> 36[4] via P2P/CUMEM +n116-041-023:1383134:1383262 [3] NCCL INFO Channel 08/0 : 35[3] -> 34[2] via P2P/CUMEM +n116-041-023:1383138:1383255 [7] NCCL INFO Channel 14/0 : 39[7] -> 32[0] via P2P/CUMEM +n116-041-023:1383136:1383258 [5] NCCL INFO Channel 05/0 : 37[5] -> 36[4] via P2P/CUMEM +n116-041-023:1383134:1383262 [3] NCCL INFO Channel 09/0 : 35[3] -> 34[2] via P2P/CUMEM +n116-041-023:1383136:1383258 [5] NCCL INFO Channel 06/0 : 37[5] -> 36[4] via P2P/CUMEM +n116-041-023:1383138:1383255 [7] NCCL INFO Channel 00/0 : 39[7] -> 38[6] via P2P/CUMEM +n116-041-023:1383134:1383262 [3] NCCL INFO Channel 10/0 : 35[3] -> 34[2] via P2P/CUMEM +n116-041-023:1383136:1383258 [5] NCCL INFO Channel 07/0 : 37[5] -> 36[4] via P2P/CUMEM +n116-041-023:1383132:1383261 [1] NCCL INFO Channel 00/0 : 33[1] -> 32[0] via P2P/CUMEM +n116-041-023:1383138:1383255 [7] NCCL INFO Channel 01/0 : 39[7] -> 38[6] via P2P/CUMEM +n116-041-023:1383134:1383262 [3] NCCL INFO Channel 11/0 : 35[3] -> 34[2] via P2P/CUMEM +n116-041-023:1383136:1383258 [5] NCCL INFO Channel 08/0 : 37[5] -> 36[4] via P2P/CUMEM +n116-041-023:1383132:1383261 [1] NCCL INFO Channel 01/0 : 33[1] -> 32[0] via P2P/CUMEM +n116-041-023:1383138:1383255 [7] NCCL INFO Channel 02/0 : 39[7] -> 38[6] via P2P/CUMEM +n116-041-023:1383134:1383262 [3] NCCL INFO Channel 12/0 : 35[3] -> 34[2] via P2P/CUMEM +n116-041-023:1383135:1383257 [4] NCCL INFO Channel 01/0 : 36[4] -> 35[3] via P2P/CUMEM +n116-041-023:1383136:1383258 [5] NCCL INFO Channel 09/0 : 37[5] -> 36[4] via P2P/CUMEM +n116-041-023:1383132:1383261 [1] NCCL INFO Channel 02/0 : 33[1] -> 32[0] via P2P/CUMEM +n116-041-023:1383138:1383255 [7] NCCL INFO Channel 03/0 : 39[7] -> 38[6] via P2P/CUMEM +n116-041-023:1383134:1383262 [3] NCCL INFO Channel 13/0 : 35[3] -> 34[2] via P2P/CUMEM +n116-041-023:1383135:1383257 [4] NCCL INFO Channel 02/0 : 36[4] -> 35[3] via P2P/CUMEM +n116-041-023:1383136:1383258 [5] NCCL INFO Channel 10/0 : 37[5] -> 36[4] via P2P/CUMEM +n116-041-023:1383132:1383261 [1] NCCL INFO Channel 03/0 : 33[1] -> 32[0] via P2P/CUMEM +n116-041-023:1383138:1383255 [7] NCCL INFO Channel 04/0 : 39[7] -> 38[6] via P2P/CUMEM +n116-041-023:1383134:1383262 [3] NCCL INFO Channel 14/0 : 35[3] -> 34[2] via P2P/CUMEM +n116-041-023:1383135:1383257 [4] NCCL INFO Channel 05/0 : 36[4] -> 35[3] via P2P/CUMEM +n116-041-023:1383132:1383261 [1] NCCL INFO Channel 04/0 : 33[1] -> 32[0] via P2P/CUMEM +n116-041-023:1383136:1383258 [5] NCCL INFO Channel 11/0 : 37[5] -> 36[4] via P2P/CUMEM +n116-041-023:1383138:1383255 [7] NCCL INFO Channel 05/0 : 39[7] -> 38[6] via P2P/CUMEM +n116-041-023:1383134:1383262 [3] NCCL INFO Channel 15/0 : 35[3] -> 34[2] via P2P/CUMEM +n116-041-023:1383135:1383257 [4] NCCL INFO Channel 06/0 : 36[4] -> 35[3] via P2P/CUMEM +n116-041-023:1383136:1383258 [5] NCCL INFO Channel 12/0 : 37[5] -> 36[4] via P2P/CUMEM +n116-041-023:1383132:1383261 [1] NCCL INFO Channel 05/0 : 33[1] -> 32[0] via P2P/CUMEM +n116-041-023:1383133:1383256 [2] NCCL INFO Channel 00/0 : 34[2] -> 33[1] via P2P/CUMEM +n116-041-023:1383136:1383258 [5] NCCL INFO Channel 13/0 : 37[5] -> 36[4] via P2P/CUMEM +n116-041-023:1383132:1383261 [1] NCCL INFO Channel 06/0 : 33[1] -> 32[0] via P2P/CUMEM +n116-041-023:1383135:1383257 [4] NCCL INFO Channel 07/0 : 36[4] -> 35[3] via P2P/CUMEM +n116-041-023:1383133:1383256 [2] NCCL INFO Channel 03/0 : 34[2] -> 33[1] via P2P/CUMEM +n116-041-023:1383136:1383258 [5] NCCL INFO Channel 14/0 : 37[5] -> 36[4] via P2P/CUMEM +n116-041-023:1383132:1383261 [1] NCCL INFO Channel 07/0 : 33[1] -> 32[0] via P2P/CUMEM +n116-041-023:1383135:1383257 [4] NCCL INFO Channel 08/0 : 36[4] -> 35[3] via P2P/CUMEM +n116-041-023:1383133:1383256 [2] NCCL INFO Channel 04/0 : 34[2] -> 33[1] via P2P/CUMEM +n116-041-023:1383136:1383258 [5] NCCL INFO Channel 15/0 : 37[5] -> 36[4] via P2P/CUMEM +n116-041-023:1383132:1383261 [1] NCCL INFO Channel 08/0 : 33[1] -> 32[0] via P2P/CUMEM +n116-041-023:1383135:1383257 [4] NCCL INFO Channel 09/0 : 36[4] -> 35[3] via P2P/CUMEM +n116-041-023:1383138:1383255 [7] NCCL INFO Channel 06/0 : 39[7] -> 38[6] via P2P/CUMEM +n116-041-023:1383133:1383256 [2] NCCL INFO Channel 05/0 : 34[2] -> 33[1] via P2P/CUMEM +n116-041-023:1383132:1383261 [1] NCCL INFO Channel 09/0 : 33[1] -> 32[0] via P2P/CUMEM +n116-041-023:1383135:1383257 [4] NCCL INFO Channel 10/0 : 36[4] -> 35[3] via P2P/CUMEM +n116-041-023:1383133:1383256 [2] NCCL INFO Channel 06/0 : 34[2] -> 33[1] via P2P/CUMEM +n116-041-023:1383132:1383261 [1] NCCL INFO Channel 10/0 : 33[1] -> 32[0] via P2P/CUMEM +n116-041-023:1383138:1383255 [7] NCCL INFO Channel 07/0 : 39[7] -> 38[6] via P2P/CUMEM +n116-041-023:1383135:1383257 [4] NCCL INFO Channel 13/0 : 36[4] -> 35[3] via P2P/CUMEM +n116-041-023:1383133:1383256 [2] NCCL INFO Channel 07/0 : 34[2] -> 33[1] via P2P/CUMEM +n116-041-023:1383132:1383261 [1] NCCL INFO Channel 11/0 : 33[1] -> 32[0] via P2P/CUMEM +n116-041-023:1383138:1383255 [7] NCCL INFO Channel 08/0 : 39[7] -> 38[6] via P2P/CUMEM +n116-041-023:1383135:1383257 [4] NCCL INFO Channel 14/0 : 36[4] -> 35[3] via P2P/CUMEM +n116-041-023:1383133:1383256 [2] NCCL INFO Channel 08/0 : 34[2] -> 33[1] via P2P/CUMEM +n116-041-023:1383138:1383255 [7] NCCL INFO Channel 09/0 : 39[7] -> 38[6] via P2P/CUMEM +n116-041-023:1383135:1383257 [4] NCCL INFO Channel 15/0 : 36[4] -> 35[3] via P2P/CUMEM +n116-041-023:1383138:1383255 [7] NCCL INFO Channel 10/0 : 39[7] -> 38[6] via P2P/CUMEM +n116-041-023:1383138:1383255 [7] NCCL INFO Channel 11/0 : 39[7] -> 38[6] via P2P/CUMEM +n116-041-023:1383133:1383256 [2] NCCL INFO Channel 11/0 : 34[2] -> 33[1] via P2P/CUMEM +n116-041-023:1383132:1383261 [1] NCCL INFO Channel 12/0 : 33[1] -> 32[0] via P2P/CUMEM +n116-041-023:1383138:1383255 [7] NCCL INFO Channel 12/0 : 39[7] -> 38[6] via P2P/CUMEM +n116-041-023:1383133:1383256 [2] NCCL INFO Channel 12/0 : 34[2] -> 33[1] via P2P/CUMEM +n116-041-023:1383132:1383261 [1] NCCL INFO Channel 13/0 : 33[1] -> 32[0] via P2P/CUMEM +n116-041-023:1383138:1383255 [7] NCCL INFO Channel 13/0 : 39[7] -> 38[6] via P2P/CUMEM +n116-041-023:1383132:1383261 [1] NCCL INFO Channel 14/0 : 33[1] -> 32[0] via P2P/CUMEM +n116-041-023:1383132:1383261 [1] NCCL INFO Channel 15/0 : 33[1] -> 32[0] via P2P/CUMEM +n116-041-023:1383133:1383256 [2] NCCL INFO Channel 13/0 : 34[2] -> 33[1] via P2P/CUMEM +n116-041-023:1383138:1383255 [7] NCCL INFO Channel 14/0 : 39[7] -> 38[6] via P2P/CUMEM +n116-041-023:1383133:1383256 [2] NCCL INFO Channel 14/0 : 34[2] -> 33[1] via P2P/CUMEM +n116-041-023:1383138:1383255 [7] NCCL INFO Channel 15/0 : 39[7] -> 38[6] via P2P/CUMEM +n116-041-023:1383133:1383256 [2] NCCL INFO Channel 15/0 : 34[2] -> 33[1] via P2P/CUMEM +n116-041-023:1383137:1383259 [6] NCCL INFO Channel 00/0 : 38[6] -> 37[5] via P2P/CUMEM +n116-041-023:1383137:1383259 [6] NCCL INFO Channel 01/0 : 38[6] -> 37[5] via P2P/CUMEM +n116-041-023:1383137:1383259 [6] NCCL INFO Channel 02/0 : 38[6] -> 37[5] via P2P/CUMEM +n116-041-023:1383137:1383259 [6] NCCL INFO Channel 03/0 : 38[6] -> 37[5] via P2P/CUMEM +n116-041-023:1383137:1383259 [6] NCCL INFO Channel 04/0 : 38[6] -> 37[5] via P2P/CUMEM +n116-041-023:1383137:1383259 [6] NCCL INFO Channel 07/0 : 38[6] -> 37[5] via P2P/CUMEM +n116-041-023:1383137:1383259 [6] NCCL INFO Channel 08/0 : 38[6] -> 37[5] via P2P/CUMEM +n116-041-023:1383137:1383259 [6] NCCL INFO Channel 09/0 : 38[6] -> 37[5] via P2P/CUMEM +n116-041-023:1383137:1383259 [6] NCCL INFO Channel 10/0 : 38[6] -> 37[5] via P2P/CUMEM +n116-041-023:1383137:1383259 [6] NCCL INFO Channel 11/0 : 38[6] -> 37[5] via P2P/CUMEM +n116-041-023:1383137:1383259 [6] NCCL INFO Channel 12/0 : 38[6] -> 37[5] via P2P/CUMEM +n116-041-023:1383137:1383259 [6] NCCL INFO Channel 15/0 : 38[6] -> 37[5] via P2P/CUMEM +n116-041-023:1383133:1383256 [2] NCCL INFO Connected all trees +n116-041-023:1383131:1383260 [0] NCCL INFO Connected all trees +n116-041-023:1383132:1383261 [1] NCCL INFO Connected all trees +n116-041-023:1383138:1383255 [7] NCCL INFO Connected all trees +n116-041-023:1383137:1383259 [6] NCCL INFO Connected all trees +n116-041-023:1383134:1383262 [3] NCCL INFO Connected all trees +n116-041-023:1383136:1383258 [5] NCCL INFO Connected all trees +n116-041-023:1383135:1383257 [4] NCCL INFO Connected all trees +n116-041-023:1383131:1383291 [0] NCCL INFO Channel 00/0 : 25[1] -> 32[0] [receive] via NET/FasTrak/0/GDRDMA +n116-041-023:1383131:1383291 [0] NCCL INFO Channel 08/0 : 25[1] -> 32[0] [receive] via NET/FasTrak/0/GDRDMA +n116-041-023:1383131:1383291 [0] NCCL INFO Channel 00/0 : 32[0] -> 39[7] via P2P/CUMEM +n116-041-023:1383131:1383291 [0] NCCL INFO Channel 08/0 : 32[0] -> 39[7] via P2P/CUMEM +n116-041-023:1383132:1383292 [1] NCCL INFO Channel 01/0 : 33[1] -> 34[2] via P2P/CUMEM +n116-041-023:1383132:1383292 [1] NCCL INFO Channel 09/0 : 33[1] -> 34[2] via P2P/CUMEM +n116-041-023:1383134:1383293 [3] NCCL INFO Channel 03/0 : 35[3] -> 36[4] via P2P/CUMEM +n116-041-023:1383134:1383293 [3] NCCL INFO Channel 11/0 : 35[3] -> 36[4] via P2P/CUMEM +n116-041-023:1383136:1383294 [5] NCCL INFO Channel 05/0 : 37[5] -> 38[6] via P2P/CUMEM +n116-041-023:1383136:1383294 [5] NCCL INFO Channel 13/0 : 37[5] -> 38[6] via P2P/CUMEM +n116-041-023:1383134:1383293 [3] NCCL INFO Channel 02/0 : 35[3] -> 42[2] [send] via NET/FasTrak/2/GDRDMA +n116-041-023:1383134:1383293 [3] NCCL INFO Channel 10/0 : 35[3] -> 42[2] [send] via NET/FasTrak/2/GDRDMA +n116-041-023:1383135:1383295 [4] NCCL INFO Channel 04/0 : 29[5] -> 36[4] [receive] via NET/FasTrak/4/GDRDMA +n116-041-023:1383135:1383295 [4] NCCL INFO Channel 12/0 : 29[5] -> 36[4] [receive] via NET/FasTrak/4/GDRDMA +n116-041-023:1383134:1383293 [3] NCCL INFO Channel 03/0 : 26[2] -> 35[3] [receive] via NET/FasTrak/3/GDRDMA +n116-041-023:1383134:1383293 [3] NCCL INFO Channel 11/0 : 26[2] -> 35[3] [receive] via NET/FasTrak/3/GDRDMA +n116-041-023:1383138:1383296 [7] NCCL INFO Channel 06/0 : 39[7] -> 46[6] [send] via NET/FasTrak/6/GDRDMA +n116-041-023:1383138:1383296 [7] NCCL INFO Channel 14/0 : 39[7] -> 46[6] [send] via NET/FasTrak/6/GDRDMA +n116-041-023:1383131:1383291 [0] NCCL INFO Channel 01/0 : 32[0] -> 41[1] [send] via NET/FasTrak/1/GDRDMA +n116-041-023:1383131:1383291 [0] NCCL INFO Channel 09/0 : 32[0] -> 41[1] [send] via NET/FasTrak/1/GDRDMA +n116-041-023:1383138:1383296 [7] NCCL INFO Channel 07/0 : 30[6] -> 39[7] [receive] via NET/FasTrak/7/GDRDMA +n116-041-023:1383138:1383296 [7] NCCL INFO Channel 15/0 : 30[6] -> 39[7] [receive] via NET/FasTrak/7/GDRDMA +n116-041-023:1383136:1383294 [5] NCCL INFO Channel 04/0 : 37[5] -> 44[4] [send] via NET/FasTrak/4/GDRDMA +n116-041-023:1383136:1383294 [5] NCCL INFO Channel 12/0 : 37[5] -> 44[4] [send] via NET/FasTrak/4/GDRDMA +n116-041-023:1383137:1383297 [6] NCCL INFO Channel 06/0 : 31[7] -> 38[6] [receive] via NET/FasTrak/6/GDRDMA +n116-041-023:1383136:1383294 [5] NCCL INFO Channel 05/0 : 28[4] -> 37[5] [receive] via NET/FasTrak/5/GDRDMA +n116-041-023:1383137:1383297 [6] NCCL INFO Channel 14/0 : 31[7] -> 38[6] [receive] via NET/FasTrak/6/GDRDMA +n116-041-023:1383137:1383297 [6] NCCL INFO Channel 07/0 : 38[6] -> 47[7] [send] via NET/FasTrak/7/GDRDMA +n116-041-023:1383137:1383297 [6] NCCL INFO Channel 15/0 : 38[6] -> 47[7] [send] via NET/FasTrak/7/GDRDMA +n116-041-023:1383136:1383294 [5] NCCL INFO Channel 13/0 : 28[4] -> 37[5] [receive] via NET/FasTrak/5/GDRDMA +n116-041-023:1383137:1383297 [6] NCCL INFO Channel 06/0 : 38[6] -> 37[5] via P2P/CUMEM +n116-041-023:1383137:1383297 [6] NCCL INFO Channel 14/0 : 38[6] -> 37[5] via P2P/CUMEM +n116-041-023:1383132:1383292 [1] NCCL INFO Channel 00/0 : 33[1] -> 40[0] [send] via NET/FasTrak/0/GDRDMA +n116-041-023:1383132:1383292 [1] NCCL INFO Channel 08/0 : 33[1] -> 40[0] [send] via NET/FasTrak/0/GDRDMA +n116-041-023:1383133:1383298 [2] NCCL INFO Channel 02/0 : 27[3] -> 34[2] [receive] via NET/FasTrak/2/GDRDMA +n116-041-023:1383132:1383292 [1] NCCL INFO Channel 01/0 : 24[0] -> 33[1] [receive] via NET/FasTrak/1/GDRDMA +n116-041-023:1383133:1383298 [2] NCCL INFO Channel 10/0 : 27[3] -> 34[2] [receive] via NET/FasTrak/2/GDRDMA +n116-041-023:1383132:1383292 [1] NCCL INFO Channel 09/0 : 24[0] -> 33[1] [receive] via NET/FasTrak/1/GDRDMA +n116-041-023:1383133:1383298 [2] NCCL INFO Channel 03/0 : 34[2] -> 43[3] [send] via NET/FasTrak/3/GDRDMA +n116-041-023:1383133:1383298 [2] NCCL INFO Channel 11/0 : 34[2] -> 43[3] [send] via NET/FasTrak/3/GDRDMA +n116-041-023:1383133:1383298 [2] NCCL INFO Channel 02/0 : 34[2] -> 33[1] via P2P/CUMEM +n116-041-023:1383133:1383298 [2] NCCL INFO Channel 10/0 : 34[2] -> 33[1] via P2P/CUMEM +n116-041-023:1383135:1383295 [4] NCCL INFO Channel 05/0 : 36[4] -> 45[5] [send] via NET/FasTrak/5/GDRDMA +n116-041-023:1383135:1383295 [4] NCCL INFO Channel 13/0 : 36[4] -> 45[5] [send] via NET/FasTrak/5/GDRDMA +n116-041-023:1383135:1383295 [4] NCCL INFO Channel 04/0 : 36[4] -> 35[3] via P2P/CUMEM +n116-041-023:1383138:1383296 [7] NCCL INFO Channel 07/0 : 39[7] -> 32[0] via P2P/CUMEM +n116-041-023:1383135:1383295 [4] NCCL INFO Channel 12/0 : 36[4] -> 35[3] via P2P/CUMEM +n116-041-023:1383138:1383296 [7] NCCL INFO Channel 15/0 : 39[7] -> 32[0] via P2P/CUMEM +n116-041-023:1383132:1383292 [1] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-041-023:1383133:1383298 [2] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-041-023:1383131:1383291 [0] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-041-023:1383138:1383296 [7] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-041-023:1383137:1383297 [6] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-041-023:1383136:1383294 [5] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-041-023:1383135:1383295 [4] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-041-023:1383134:1383293 [3] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +[rank34]:[W727 05:39:50.986981717 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n116-041-023.byted.org]:38014, remote=[n116-040-168.byted.org]:25264): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f7ccd9785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f7cb70cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7f7cb70cff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7f7cb70d084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7f7cb70ca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f7c787c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f7cd11004a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f7cd3cd71f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f7cd3d5789c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank34]:[W727 05:39:50.001794841 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 34] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank35]:[W727 05:39:50.084537287 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n116-041-023.byted.org]:38058, remote=[n116-040-168.byted.org]:25264): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f10779785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f1060ccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7f1060ccff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7f1060cd084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7f1060cca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f10223c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f107adf94a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f107d9d01f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f107da5089c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank35]:[W727 05:39:50.092952237 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 35] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank37]:[W727 05:39:50.096927152 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n116-041-023.byted.org]:37990, remote=[n116-040-168.byted.org]:25264): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f27601785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f27494cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7f27494cff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7f27494d084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7f27494ca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f270abc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f276365b4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f27662321f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f27662b289c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank37]:[W727 05:39:50.103875918 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 37] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank39]:[W727 05:39:50.336256363 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n116-041-023.byted.org]:37986, remote=[n116-040-168.byted.org]:25264): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7eff41f785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7eff2b2cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7eff2b2cff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7eff2b2d084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7eff2b2ca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7efeec9c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7eff453e94a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7eff47fc01f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7eff4804089c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank39]:[W727 05:39:50.344033496 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 39] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank36]:[W727 05:39:50.342185433 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n116-041-023.byted.org]:38048, remote=[n116-040-168.byted.org]:25264): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7fee60b785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7fee49ecdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7fee49ecff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7fee49ed084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7fee49eca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fee0b5c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7fee6400b4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7fee66be21f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7fee66c6289c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank36]:[W727 05:39:50.349246037 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 36] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank32]:[W727 05:39:51.585114592 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=71, addr=[n116-041-023.byted.org]:37998, remote=[n116-040-168.byted.org]:25264): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f78687785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f7851acdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7f7851acff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7f7851ad084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7f7851aca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f78131c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f786bbca4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f786e7a11f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f786e82189c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank32]:[W727 05:39:51.591428788 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 32] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank33]:[W727 05:39:51.585128201 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n116-041-023.byted.org]:38022, remote=[n116-040-168.byted.org]:25264): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f1e677785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f1e50ecdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7f1e50ecff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7f1e50ed084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7f1e50eca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f1e125c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f1e6af1a4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f1e6daf11f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f1e6db7189c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank33]:[W727 05:39:51.593064613 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 33] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank38]:[W727 05:39:51.618957001 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n116-041-023.byted.org]:38032, remote=[n116-040-168.byted.org]:25264): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7ff889f785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7ff8736cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7ff8736cff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7ff8736d084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7ff8736ca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7ff834dc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7ff88d74c4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7ff8903231f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7ff8903a389c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank38]:[W727 05:39:51.625962334 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 38] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank34]:[W727 05:39:51.001967982 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-023.byted.org]:38014, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f7ccd9785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f7cb70cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f7cb70cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f7cb70d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f7cb70ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f7c787c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f7cd11004a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f7cd3cd71f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f7cd3d5789c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank34]:[W727 05:39:51.009062180 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 34] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank35]:[W727 05:39:51.093095678 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-023.byted.org]:38058, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f10779785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f1060ccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f1060ccf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f1060cd0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f1060cca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f10223c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f107adf94a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f107d9d01f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f107da5089c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank35]:[W727 05:39:51.100551759 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 35] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank37]:[W727 05:39:51.104058118 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-023.byted.org]:37990, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f27601785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f27494cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f27494cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f27494d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f27494ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f270abc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f276365b4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f27662321f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f27662b289c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank37]:[W727 05:39:51.111217736 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 37] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank39]:[W727 05:39:51.344198160 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-023.byted.org]:37986, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7eff41f785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7eff2b2cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7eff2b2cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7eff2b2d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7eff2b2ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7efeec9c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7eff453e94a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7eff47fc01f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7eff4804089c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank39]:[W727 05:39:51.351146866 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 39] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank36]:[W727 05:39:51.349409189 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-023.byted.org]:38048, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7fee60b785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7fee49ecdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7fee49ecf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7fee49ed0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7fee49eca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fee0b5c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7fee6400b4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7fee66be21f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7fee66c6289c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank36]:[W727 05:39:51.355628578 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 36] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank32]:[W727 05:39:52.591601551 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=71, addr=[n116-041-023.byted.org]:37998, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f78687785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f7851acdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f7851acf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f7851ad0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f7851aca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f78131c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f786bbca4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f786e7a11f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f786e82189c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank32]:[W727 05:39:52.598082077 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 32] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank33]:[W727 05:39:52.593232518 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-023.byted.org]:38022, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f1e677785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f1e50ecdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f1e50ecf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f1e50ed0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f1e50eca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f1e125c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f1e6af1a4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f1e6daf11f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f1e6db7189c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank33]:[W727 05:39:52.600265515 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 33] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank38]:[W727 05:39:52.626128467 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-023.byted.org]:38032, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7ff889f785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7ff8736cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7ff8736cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7ff8736d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7ff8736ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7ff834dc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7ff88d74c4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7ff8903231f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7ff8903a389c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank38]:[W727 05:39:52.633042491 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 38] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank34]:[W727 05:39:52.009230029 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-023.byted.org]:38014, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f7ccd9785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f7cb70cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f7cb70cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f7cb70d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f7cb70ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f7c787c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f7cd11004a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f7cd3cd71f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f7cd3d5789c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank34]:[W727 05:39:52.016618335 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 34] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank35]:[W727 05:39:52.100833450 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-023.byted.org]:38058, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f10779785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f1060ccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f1060ccf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f1060cd0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f1060cca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f10223c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f107adf94a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f107d9d01f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f107da5089c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank35]:[W727 05:39:52.107864368 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 35] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank37]:[W727 05:39:52.111374661 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-023.byted.org]:37990, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f27601785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f27494cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f27494cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f27494d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f27494ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f270abc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f276365b4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f27662321f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f27662b289c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank37]:[W727 05:39:52.118268515 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 37] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank39]:[W727 05:39:52.351323227 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-023.byted.org]:37986, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7eff41f785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7eff2b2cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7eff2b2cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7eff2b2d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7eff2b2ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7efeec9c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7eff453e94a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7eff47fc01f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7eff4804089c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank39]:[W727 05:39:52.357371348 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 39] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank36]:[W727 05:39:52.355789528 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-023.byted.org]:38048, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7fee60b785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7fee49ecdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7fee49ecf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7fee49ed0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7fee49eca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fee0b5c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7fee6400b4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7fee66be21f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7fee66c6289c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank36]:[W727 05:39:52.362188713 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 36] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank32]:[W727 05:39:53.598248401 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=71, addr=[n116-041-023.byted.org]:37998, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f78687785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f7851acdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f7851acf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f7851ad0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f7851aca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f78131c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f786bbca4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f786e7a11f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f786e82189c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank32]:[W727 05:39:53.604325830 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 32] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank33]:[W727 05:39:53.600428122 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-023.byted.org]:38022, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f1e677785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f1e50ecdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f1e50ecf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f1e50ed0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f1e50eca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f1e125c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f1e6af1a4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f1e6daf11f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f1e6db7189c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank33]:[W727 05:39:53.607458993 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 33] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank38]:[W727 05:39:53.633223132 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-023.byted.org]:38032, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7ff889f785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7ff8736cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7ff8736cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7ff8736d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7ff8736ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7ff834dc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7ff88d74c4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7ff8903231f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7ff8903a389c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank38]:[W727 05:39:53.640231169 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 38] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank34]:[W727 05:39:53.016776890 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-023.byted.org]:38014, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f7ccd9785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f7cb70cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f7cb70cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f7cb70d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f7cb70ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f7c787c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f7cd11004a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f7cd3cd71f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f7cd3d5789c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank34]:[W727 05:39:53.023451843 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 34] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank35]:[W727 05:39:53.108023318 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-023.byted.org]:38058, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f10779785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f1060ccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f1060ccf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f1060cd0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f1060cca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f10223c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f107adf94a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f107d9d01f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f107da5089c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank35]:[W727 05:39:53.120328200 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 35] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank37]:[W727 05:39:53.118426287 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-023.byted.org]:37990, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f27601785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f27494cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f27494cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f27494d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f27494ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f270abc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f276365b4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f27662321f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f27662b289c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank37]:[W727 05:39:53.124464459 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 37] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank39]:[W727 05:39:53.357550640 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-023.byted.org]:37986, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7eff41f785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7eff2b2cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7eff2b2cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7eff2b2d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7eff2b2ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7efeec9c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7eff453e94a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7eff47fc01f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7eff4804089c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank39]:[W727 05:39:53.366191471 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 39] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank36]:[W727 05:39:53.362374691 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-023.byted.org]:38048, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7fee60b785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7fee49ecdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7fee49ecf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7fee49ed0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7fee49eca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fee0b5c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7fee6400b4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7fee66be21f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7fee66c6289c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank36]:[W727 05:39:53.368702845 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 36] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank32]:[W727 05:39:54.604539647 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=71, addr=[n116-041-023.byted.org]:37998, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f78687785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f7851acdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f7851acf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f7851ad0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f7851aca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f78131c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f786bbca4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f786e7a11f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f786e82189c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank32]:[W727 05:39:54.611951135 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 32] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank33]:[W727 05:39:54.607636091 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-023.byted.org]:38022, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f1e677785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f1e50ecdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f1e50ecf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f1e50ed0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f1e50eca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f1e125c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f1e6af1a4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f1e6daf11f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f1e6db7189c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank33]:[W727 05:39:54.614362618 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 33] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank38]:[W727 05:39:54.640377920 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-023.byted.org]:38032, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7ff889f785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7ff8736cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7ff8736cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7ff8736d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7ff8736ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7ff834dc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7ff88d74c4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7ff8903231f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7ff8903a389c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank38]:[W727 05:39:54.646564613 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 38] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank34]:[W727 05:39:54.023616902 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-023.byted.org]:38014, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f7ccd9785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f7cb70cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f7cb70cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f7cb70d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f7cb70ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f7c787c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f7cd11004a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f7cd3cd71f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f7cd3d5789c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank34]:[W727 05:39:54.029984858 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 34] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank35]:[W727 05:39:54.120489669 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-023.byted.org]:38058, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f10779785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f1060ccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f1060ccf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f1060cd0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f1060cca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f10223c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f107adf94a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f107d9d01f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f107da5089c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank35]:[W727 05:39:54.127863166 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 35] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank37]:[W727 05:39:54.124624186 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-023.byted.org]:37990, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f27601785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f27494cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f27494cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f27494d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f27494ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f270abc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f276365b4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f27662321f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f27662b289c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank37]:[W727 05:39:54.130677493 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 37] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank39]:[W727 05:39:54.366354670 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-023.byted.org]:37986, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7eff41f785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7eff2b2cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7eff2b2cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7eff2b2d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7eff2b2ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7efeec9c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7eff453e94a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7eff47fc01f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7eff4804089c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank39]:[W727 05:39:54.372473630 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 39] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +E0727 05:39:55.015000 1383066 torch/distributed/elastic/multiprocessing/api.py:874] failed (exitcode: -15) local_rank: 0 (pid: 1383131) of binary: /usr/bin/python3 +Traceback (most recent call last): + File "/usr/local/bin/torchrun", line 7, in + sys.exit(main()) + ^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 355, in wrapper + return f(*args, **kwargs) + ^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 892, in main + run(args) + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 883, in run + elastic_launch( + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 139, in __call__ + return launch_agent(self._config, self._entrypoint, list(args)) + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 270, in launch_agent + raise ChildFailedError( +torch.distributed.elastic.multiprocessing.errors.ChildFailedError: +============================================================ +main_cross_2direct_balance_realscale_perturb_v4_accum.py FAILED +------------------------------------------------------------ +Failures: +[1]: + time : 2026-07-27_05:39:55 + host : n116-041-023.byted.org + rank : 33 (local_rank: 1) + exitcode : -15 (pid: 1383132) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 1383132 +[2]: + time : 2026-07-27_05:39:55 + host : n116-041-023.byted.org + rank : 34 (local_rank: 2) + exitcode : -15 (pid: 1383133) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 1383133 +[3]: + time : 2026-07-27_05:39:55 + host : n116-041-023.byted.org + rank : 35 (local_rank: 3) + exitcode : -15 (pid: 1383134) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 1383134 +[4]: + time : 2026-07-27_05:39:55 + host : n116-041-023.byted.org + rank : 36 (local_rank: 4) + exitcode : -15 (pid: 1383135) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 1383135 +[5]: + time : 2026-07-27_05:39:55 + host : n116-041-023.byted.org + rank : 37 (local_rank: 5) + exitcode : -15 (pid: 1383136) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 1383136 +[6]: + time : 2026-07-27_05:39:55 + host : n116-041-023.byted.org + rank : 38 (local_rank: 6) + exitcode : -15 (pid: 1383137) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 1383137 +[7]: + time : 2026-07-27_05:39:55 + host : n116-041-023.byted.org + rank : 39 (local_rank: 7) + exitcode : -15 (pid: 1383138) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 1383138 +------------------------------------------------------------ +Root Cause (first observed failure): +[0]: + time : 2026-07-27_05:39:55 + host : n116-041-023.byted.org + rank : 32 (local_rank: 0) + exitcode : -15 (pid: 1383131) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 1383131 +============================================================ diff --git a/ScaleDF_rebuttal/_host_logs/20260726_2350_vith64_free8_port25264_launch/launcher_rank5_n116-040-213.log b/ScaleDF_rebuttal/_host_logs/20260726_2350_vith64_free8_port25264_launch/launcher_rank5_n116-040-213.log new file mode 100644 index 0000000000000000000000000000000000000000..46d82d602d5bed4ecdbc4a10c8aebfba856c4bc4 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_2350_vith64_free8_port25264_launch/launcher_rank5_n116-040-213.log @@ -0,0 +1,2271 @@ +started_utc=2026-07-26T23:50:27Z +hostname=n116-040-213 +node_rank=5 +nnodes=8 +nproc_per_node=8 +world_size=64 +master_addr=10.116.40.168 +master_port=25264 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +main=main_cross_2direct_balance_realscale_perturb_v4_accum.py +model_name=deit_huge_patch14_LS +pretrain=/tmp/ScaleDF_repro_20260724/pretrained/deit_3_huge_224_21k_v1.pth +resume=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/checkpoint_snapshots/vith32_latest_for_64card_effbs4096_20260726_234604.pth +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_64card_effbs4096_from_epoch22_snapshot_free8_master_n116-040-168_20260726_2350_vith64_port25264 +batch_per_gpu=64 +accum_iter=1 +effective_batch=4096 +eval_batch_per_gpu=64 +epochs=40 +W0726 23:50:41.147000 1385203 torch/distributed/run.py:766] +W0726 23:50:41.147000 1385203 torch/distributed/run.py:766] ***************************************** +W0726 23:50:41.147000 1385203 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 23:50:41.147000 1385203 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 40): env:// +| distributed init (rank 44): env:// +[rank44]:[W726 23:51:08.286707805 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 44] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 46): env:// +[rank46]:[W726 23:51:08.338091180 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 46] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 45): env:// +[rank45]:[W726 23:51:08.537803121 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 45] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 43): env:// +[rank43]:[W726 23:51:08.587829262 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 43] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank40]:[W726 23:51:08.615703668 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 40] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 47): env:// +| distributed init (rank 41): env:// +| distributed init (rank 42): env:// +[rank47]:[W726 23:51:08.813629803 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 47] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank42]:[W726 23:51:08.813812833 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 42] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank41]:[W726 23:51:08.813883239 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 41] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +n116-040-213:1385274:1385274 [6] NCCL INFO cudaDriverVersion 12060 +n116-040-213:1385274:1385274 [6] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-040-213:1385274:1385274 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-040-213:1385274:1385274 [6] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-040-213:1385274:1385274 [6] NCCL INFO Comm config Blocking set to 1 +n116-040-213:1385271:1385271 [3] NCCL INFO cudaDriverVersion 12060 +n116-040-213:1385271:1385271 [3] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-040-213:1385271:1385271 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-040-213:1385271:1385271 [3] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-040-213:1385271:1385271 [3] NCCL INFO Comm config Blocking set to 1 +n116-040-213:1385273:1385273 [5] NCCL INFO cudaDriverVersion 12060 +n116-040-213:1385273:1385273 [5] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-040-213:1385273:1385273 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-040-213:1385273:1385273 [5] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-040-213:1385273:1385273 [5] NCCL INFO Comm config Blocking set to 1 +n116-040-213:1385272:1385272 [4] NCCL INFO cudaDriverVersion 12060 +n116-040-213:1385272:1385272 [4] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-040-213:1385272:1385272 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-040-213:1385272:1385272 [4] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-040-213:1385272:1385272 [4] NCCL INFO Comm config Blocking set to 1 +n116-040-213:1385268:1385268 [0] NCCL INFO cudaDriverVersion 12060 +n116-040-213:1385268:1385268 [0] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-040-213:1385268:1385268 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-040-213:1385268:1385268 [0] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-040-213:1385268:1385268 [0] NCCL INFO Comm config Blocking set to 1 +n116-040-213:1385275:1385275 [7] NCCL INFO cudaDriverVersion 12060 +n116-040-213:1385275:1385275 [7] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-040-213:1385275:1385275 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-040-213:1385275:1385275 [7] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-040-213:1385269:1385269 [1] NCCL INFO cudaDriverVersion 12060 +n116-040-213:1385269:1385269 [1] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-040-213:1385269:1385269 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-040-213:1385269:1385269 [1] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-040-213:1385275:1385275 [7] NCCL INFO Comm config Blocking set to 1 +n116-040-213:1385269:1385269 [1] NCCL INFO Comm config Blocking set to 1 +n116-040-213:1385270:1385270 [2] NCCL INFO cudaDriverVersion 12060 +n116-040-213:1385270:1385270 [2] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-040-213:1385270:1385270 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-040-213:1385270:1385270 [2] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-040-213:1385270:1385270 [2] NCCL INFO Comm config Blocking set to 1 +n116-040-213:1385274:1385336 [6] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-040-213:1385274:1385336 [6] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-040-213:1385274:1385336 [6] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-040-213:1385274:1385336 [6] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-040-213:1385274:1385336 [6] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-040-213:1385274:1385336 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-040-213:1385274:1385336 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-040-213:1385274:1385336 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-040-213:1385274:1385336 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-040-213:1385274:1385336 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-040-213:1385271:1385337 [3] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-040-213:1385271:1385337 [3] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-040-213:1385271:1385337 [3] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-040-213:1385271:1385337 [3] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-040-213:1385271:1385337 [3] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-040-213:1385271:1385337 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-040-213:1385271:1385337 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-040-213:1385271:1385337 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-040-213:1385271:1385337 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-040-213:1385271:1385337 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-040-213:1385273:1385338 [5] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-040-213:1385273:1385338 [5] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-040-213:1385273:1385338 [5] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-040-213:1385273:1385338 [5] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-040-213:1385273:1385338 [5] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-040-213:1385273:1385338 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-040-213:1385273:1385338 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-040-213:1385273:1385338 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-040-213:1385273:1385338 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-040-213:1385273:1385338 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-040-213:1385268:1385340 [0] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-040-213:1385268:1385340 [0] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-040-213:1385268:1385340 [0] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-040-213:1385268:1385340 [0] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-040-213:1385268:1385340 [0] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-040-213:1385268:1385340 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-040-213:1385268:1385340 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-040-213:1385268:1385340 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-040-213:1385268:1385340 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-040-213:1385268:1385340 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-040-213:1385272:1385339 [4] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-040-213:1385272:1385339 [4] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-040-213:1385272:1385339 [4] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-040-213:1385272:1385339 [4] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-040-213:1385272:1385339 [4] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-040-213:1385272:1385339 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-040-213:1385272:1385339 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-040-213:1385272:1385339 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-040-213:1385272:1385339 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-040-213:1385272:1385339 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-040-213:1385275:1385341 [7] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-040-213:1385275:1385341 [7] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-040-213:1385275:1385341 [7] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-040-213:1385275:1385341 [7] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-040-213:1385275:1385341 [7] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-040-213:1385275:1385341 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-040-213:1385275:1385341 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-040-213:1385275:1385341 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-040-213:1385275:1385341 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-040-213:1385275:1385341 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-040-213:1385269:1385342 [1] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-040-213:1385269:1385342 [1] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-040-213:1385269:1385342 [1] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-040-213:1385269:1385342 [1] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-040-213:1385269:1385342 [1] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-040-213:1385269:1385342 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-040-213:1385269:1385342 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-040-213:1385269:1385342 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-040-213:1385269:1385342 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-040-213:1385269:1385342 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-040-213:1385270:1385343 [2] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-040-213:1385270:1385343 [2] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-040-213:1385270:1385343 [2] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-040-213:1385270:1385343 [2] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-040-213:1385270:1385343 [2] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-040-213:1385270:1385343 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-040-213:1385270:1385343 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-040-213:1385270:1385343 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-040-213:1385270:1385343 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-040-213:1385270:1385343 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-040-213:1385271:1385337 [3] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-040-213:1385271:1385337 [3] NCCL INFO Using network FasTrak +n116-040-213:1385274:1385336 [6] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-040-213:1385274:1385336 [6] NCCL INFO Using network FasTrak +n116-040-213:1385273:1385338 [5] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-040-213:1385273:1385338 [5] NCCL INFO Using network FasTrak +n116-040-213:1385268:1385340 [0] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-040-213:1385268:1385340 [0] NCCL INFO Using network FasTrak +n116-040-213:1385275:1385341 [7] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-040-213:1385275:1385341 [7] NCCL INFO Using network FasTrak +n116-040-213:1385272:1385339 [4] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-040-213:1385272:1385339 [4] NCCL INFO Using network FasTrak +n116-040-213:1385269:1385342 [1] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-040-213:1385269:1385342 [1] NCCL INFO Using network FasTrak +n116-040-213:1385270:1385343 [2] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-040-213:1385270:1385343 [2] NCCL INFO Using network FasTrak +n116-040-213:1385274:1385336 [6] NCCL INFO DMA-BUF is available on GPU device 6 +n116-040-213:1385271:1385337 [3] NCCL INFO DMA-BUF is available on GPU device 3 +n116-040-213:1385274:1385336 [6] NCCL INFO ncclCommInitRankConfig comm 0xc6f23a0 rank 46 nranks 64 cudaDev 6 nvmlDev 6 busId 8b000 commId 0xfb2d11625d05007a - Init START +n116-040-213:1385271:1385337 [3] NCCL INFO ncclCommInitRankConfig comm 0xc1512d0 rank 43 nranks 64 cudaDev 3 nvmlDev 3 busId c000 commId 0xfb2d11625d05007a - Init START +n116-040-213:1385273:1385338 [5] NCCL INFO DMA-BUF is available on GPU device 5 +n116-040-213:1385268:1385340 [0] NCCL INFO DMA-BUF is available on GPU device 0 +n116-040-213:1385273:1385338 [5] NCCL INFO ncclCommInitRankConfig comm 0xbc479c0 rank 45 nranks 64 cudaDev 5 nvmlDev 5 busId 85000 commId 0xfb2d11625d05007a - Init START +n116-040-213:1385268:1385340 [0] NCCL INFO ncclCommInitRankConfig comm 0xcdf2640 rank 40 nranks 64 cudaDev 0 nvmlDev 0 busId 4000 commId 0xfb2d11625d05007a - Init START +n116-040-213:1385275:1385341 [7] NCCL INFO DMA-BUF is available on GPU device 7 +n116-040-213:1385269:1385342 [1] NCCL INFO DMA-BUF is available on GPU device 1 +n116-040-213:1385272:1385339 [4] NCCL INFO DMA-BUF is available on GPU device 4 +n116-040-213:1385275:1385341 [7] NCCL INFO ncclCommInitRankConfig comm 0xb2afe50 rank 47 nranks 64 cudaDev 7 nvmlDev 7 busId 8c000 commId 0xfb2d11625d05007a - Init START +n116-040-213:1385272:1385339 [4] NCCL INFO ncclCommInitRankConfig comm 0xd0e41d0 rank 44 nranks 64 cudaDev 4 nvmlDev 4 busId 84000 commId 0xfb2d11625d05007a - Init START +n116-040-213:1385269:1385342 [1] NCCL INFO ncclCommInitRankConfig comm 0xcb0efc0 rank 41 nranks 64 cudaDev 1 nvmlDev 1 busId 5000 commId 0xfb2d11625d05007a - Init START +n116-040-213:1385270:1385343 [2] NCCL INFO DMA-BUF is available on GPU device 2 +n116-040-213:1385272:1385339 [4] NCCL INFO RAS client listening socket at ::1<28028> +n116-040-213:1385268:1385340 [0] NCCL INFO RAS client listening socket at ::1<28028> +n116-040-213:1385273:1385338 [5] NCCL INFO RAS client listening socket at ::1<28028> +n116-040-213:1385274:1385336 [6] NCCL INFO RAS client listening socket at ::1<28028> +n116-040-213:1385270:1385343 [2] NCCL INFO ncclCommInitRankConfig comm 0xc5bfe00 rank 42 nranks 64 cudaDev 2 nvmlDev 2 busId b000 commId 0xfb2d11625d05007a - Init START +n116-040-213:1385269:1385342 [1] NCCL INFO RAS client listening socket at ::1<28028> +n116-040-213:1385270:1385343 [2] NCCL INFO RAS client listening socket at ::1<28028> +n116-040-213:1385271:1385337 [3] NCCL INFO RAS client listening socket at ::1<28028> +n116-040-213:1385275:1385341 [7] NCCL INFO RAS client listening socket at ::1<28028> +n116-040-213:1385274:1385336 [6] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-040-213:1385274:1385336 [6] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-040-213:1385274:1385336 [6] NCCL INFO Retrieving state for FasTrak +n116-040-213:1385274:1385336 [6] NCCL INFO Initialized state 0 for FasTrak +n116-040-213:1385274:1385336 [6] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-040-213:1385274:1385336 [6] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-040-213:1385270:1385343 [2] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-040-213:1385274:1385336 [6] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-040-213:1385270:1385343 [2] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-040-213:1385270:1385343 [2] NCCL INFO Retrieving state for FasTrak +n116-040-213:1385270:1385343 [2] NCCL INFO Initialized state 0 for FasTrak +n116-040-213:1385274:1385336 [6] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-040-213:1385270:1385343 [2] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-040-213:1385274:1385336 [6] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-040-213:1385274:1385336 [6] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-040-213:1385270:1385343 [2] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-040-213:1385275:1385341 [7] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-040-213:1385275:1385341 [7] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-040-213:1385275:1385341 [7] NCCL INFO Retrieving state for FasTrak +n116-040-213:1385275:1385341 [7] NCCL INFO Initialized state 0 for FasTrak +n116-040-213:1385270:1385343 [2] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-040-213:1385274:1385336 [6] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-040-213:1385270:1385343 [2] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-040-213:1385274:1385336 [6] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-040-213:1385275:1385341 [7] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-040-213:1385270:1385343 [2] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-040-213:1385275:1385341 [7] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-040-213:1385270:1385343 [2] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-040-213:1385273:1385338 [5] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-040-213:1385273:1385338 [5] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-040-213:1385273:1385338 [5] NCCL INFO Retrieving state for FasTrak +n116-040-213:1385273:1385338 [5] NCCL INFO Initialized state 0 for FasTrak +n116-040-213:1385275:1385341 [7] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-040-213:1385270:1385343 [2] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-040-213:1385275:1385341 [7] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-040-213:1385273:1385338 [5] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-040-213:1385270:1385343 [2] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-040-213:1385273:1385338 [5] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-040-213:1385275:1385341 [7] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-040-213:1385275:1385341 [7] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-040-213:1385273:1385338 [5] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-040-213:1385271:1385337 [3] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-040-213:1385271:1385337 [3] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-040-213:1385271:1385337 [3] NCCL INFO Retrieving state for FasTrak +n116-040-213:1385271:1385337 [3] NCCL INFO Initialized state 0 for FasTrak +n116-040-213:1385275:1385341 [7] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-040-213:1385273:1385338 [5] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-040-213:1385275:1385341 [7] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-040-213:1385273:1385338 [5] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-040-213:1385271:1385337 [3] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-040-213:1385273:1385338 [5] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-040-213:1385271:1385337 [3] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-040-213:1385271:1385337 [3] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-040-213:1385273:1385338 [5] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-040-213:1385271:1385337 [3] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-040-213:1385273:1385338 [5] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-040-213:1385271:1385337 [3] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-040-213:1385271:1385337 [3] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-040-213:1385271:1385337 [3] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-040-213:1385271:1385337 [3] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-040-213:1385272:1385339 [4] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-040-213:1385272:1385339 [4] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-040-213:1385272:1385339 [4] NCCL INFO Retrieving state for FasTrak +n116-040-213:1385272:1385339 [4] NCCL INFO Initialized state 0 for FasTrak +n116-040-213:1385272:1385339 [4] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-040-213:1385272:1385339 [4] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-040-213:1385272:1385339 [4] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-040-213:1385269:1385342 [1] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-040-213:1385269:1385342 [1] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-040-213:1385272:1385339 [4] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-040-213:1385269:1385342 [1] NCCL INFO Retrieving state for FasTrak +n116-040-213:1385269:1385342 [1] NCCL INFO Initialized state 0 for FasTrak +n116-040-213:1385272:1385339 [4] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-040-213:1385269:1385342 [1] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-040-213:1385272:1385339 [4] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-040-213:1385269:1385342 [1] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-040-213:1385272:1385339 [4] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-040-213:1385269:1385342 [1] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-040-213:1385272:1385339 [4] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-040-213:1385269:1385342 [1] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-040-213:1385269:1385342 [1] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-040-213:1385269:1385342 [1] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-040-213:1385269:1385342 [1] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-040-213:1385269:1385342 [1] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-040-213:1385268:1385340 [0] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-040-213:1385268:1385340 [0] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-040-213:1385268:1385340 [0] NCCL INFO Retrieving state for FasTrak +n116-040-213:1385268:1385340 [0] NCCL INFO Initialized state 0 for FasTrak +n116-040-213:1385268:1385340 [0] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-040-213:1385268:1385340 [0] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-040-213:1385268:1385340 [0] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-040-213:1385268:1385340 [0] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-040-213:1385268:1385340 [0] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-040-213:1385268:1385340 [0] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-040-213:1385268:1385340 [0] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-040-213:1385268:1385340 [0] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-040-213:1385270:1385343 [2] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-040-213:1385269:1385342 [1] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-040-213:1385268:1385340 [0] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-040-213:1385273:1385338 [5] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-040-213:1385275:1385341 [7] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-040-213:1385272:1385339 [4] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-040-213:1385274:1385336 [6] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-040-213:1385271:1385337 [3] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-040-213:1385270:1385343 [2] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-040-213:1385270:1385343 [2] NCCL INFO CPU/0-0 (1/1/3) +n116-040-213:1385270:1385343 [2] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-040-213:1385270:1385343 [2] NCCL INFO + PCI[24.0] - GPU/0-4000 (40) +n116-040-213:1385270:1385343 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385270:1385343 [2] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-040-213:1385270:1385343 [2] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-040-213:1385270:1385343 [2] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-040-213:1385270:1385343 [2] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-040-213:1385270:1385343 [2] NCCL INFO + PCI[24.0] - GPU/0-5000 (41) +n116-040-213:1385270:1385343 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385270:1385343 [2] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-040-213:1385270:1385343 [2] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-040-213:1385270:1385343 [2] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-040-213:1385270:1385343 [2] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-040-213:1385270:1385343 [2] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-040-213:1385270:1385343 [2] NCCL INFO + PCI[24.0] - GPU/0-b000 (42) +n116-040-213:1385270:1385343 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385270:1385343 [2] NCCL INFO + PCI[24.0] - GPU/0-c000 (43) +n116-040-213:1385270:1385343 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385270:1385343 [2] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-040-213:1385270:1385343 [2] NCCL INFO CPU/0-1 (1/1/3) +n116-040-213:1385270:1385343 [2] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-040-213:1385270:1385343 [2] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-040-213:1385270:1385343 [2] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-040-213:1385270:1385343 [2] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-040-213:1385270:1385343 [2] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-040-213:1385270:1385343 [2] NCCL INFO + PCI[24.0] - GPU/0-84000 (44) +n116-040-213:1385270:1385343 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385270:1385343 [2] NCCL INFO + PCI[24.0] - GPU/0-85000 (45) +n116-040-213:1385270:1385343 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385270:1385343 [2] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-040-213:1385270:1385343 [2] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-040-213:1385270:1385343 [2] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-040-213:1385270:1385343 [2] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-040-213:1385270:1385343 [2] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-040-213:1385270:1385343 [2] NCCL INFO + PCI[24.0] - GPU/0-8b000 (46) +n116-040-213:1385270:1385343 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385270:1385343 [2] NCCL INFO + PCI[24.0] - GPU/0-8c000 (47) +n116-040-213:1385270:1385343 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385270:1385343 [2] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-040-213:1385270:1385343 [2] NCCL INFO ========================================== +n116-040-213:1385270:1385343 [2] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-213:1385270:1385343 [2] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-213:1385270:1385343 [2] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-213:1385270:1385343 [2] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-213:1385270:1385343 [2] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-213:1385270:1385343 [2] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-213:1385270:1385343 [2] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-040-213:1385270:1385343 [2] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-040-213:1385270:1385343 [2] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-213:1385270:1385343 [2] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-213:1385270:1385343 [2] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-213:1385270:1385343 [2] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-213:1385270:1385343 [2] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-040-213:1385270:1385343 [2] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-040-213:1385270:1385343 [2] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-040-213:1385270:1385343 [2] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-040-213:1385270:1385343 [2] NCCL INFO Setting affinity for GPU 2 to 0fffffff,fffffe00,00000000,000fffff,fffffffe +n116-040-213:1385270:1385343 [2] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-040-213:1385270:1385343 [2] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-040-213:1385269:1385342 [1] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-040-213:1385269:1385342 [1] NCCL INFO CPU/0-0 (1/1/3) +n116-040-213:1385269:1385342 [1] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-040-213:1385269:1385342 [1] NCCL INFO + PCI[24.0] - GPU/0-4000 (40) +n116-040-213:1385269:1385342 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385269:1385342 [1] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-040-213:1385269:1385342 [1] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-040-213:1385269:1385342 [1] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-040-213:1385269:1385342 [1] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-040-213:1385269:1385342 [1] NCCL INFO + PCI[24.0] - GPU/0-5000 (41) +n116-040-213:1385269:1385342 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385269:1385342 [1] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-040-213:1385269:1385342 [1] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-040-213:1385269:1385342 [1] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-040-213:1385269:1385342 [1] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-040-213:1385269:1385342 [1] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-040-213:1385269:1385342 [1] NCCL INFO + PCI[24.0] - GPU/0-b000 (42) +n116-040-213:1385269:1385342 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385269:1385342 [1] NCCL INFO + PCI[24.0] - GPU/0-c000 (43) +n116-040-213:1385269:1385342 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385269:1385342 [1] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-040-213:1385269:1385342 [1] NCCL INFO CPU/0-1 (1/1/3) +n116-040-213:1385269:1385342 [1] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-040-213:1385269:1385342 [1] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-040-213:1385269:1385342 [1] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-040-213:1385269:1385342 [1] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-040-213:1385269:1385342 [1] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-040-213:1385269:1385342 [1] NCCL INFO + PCI[24.0] - GPU/0-84000 (44) +n116-040-213:1385269:1385342 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385269:1385342 [1] NCCL INFO + PCI[24.0] - GPU/0-85000 (45) +n116-040-213:1385269:1385342 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385269:1385342 [1] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-040-213:1385269:1385342 [1] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-040-213:1385269:1385342 [1] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-040-213:1385269:1385342 [1] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-040-213:1385269:1385342 [1] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-040-213:1385269:1385342 [1] NCCL INFO + PCI[24.0] - GPU/0-8b000 (46) +n116-040-213:1385269:1385342 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385269:1385342 [1] NCCL INFO + PCI[24.0] - GPU/0-8c000 (47) +n116-040-213:1385269:1385342 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385269:1385342 [1] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-040-213:1385269:1385342 [1] NCCL INFO ========================================== +n116-040-213:1385269:1385342 [1] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-213:1385269:1385342 [1] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-213:1385269:1385342 [1] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-213:1385269:1385342 [1] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-213:1385269:1385342 [1] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-213:1385269:1385342 [1] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-213:1385269:1385342 [1] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-040-213:1385269:1385342 [1] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-040-213:1385269:1385342 [1] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-213:1385269:1385342 [1] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-213:1385269:1385342 [1] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-213:1385269:1385342 [1] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-213:1385269:1385342 [1] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-040-213:1385269:1385342 [1] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-040-213:1385269:1385342 [1] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-040-213:1385269:1385342 [1] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-040-213:1385269:1385342 [1] NCCL INFO Setting affinity for GPU 1 to 0fffffff,fffffe00,00000000,000fffff,fffffffe +n116-040-213:1385269:1385342 [1] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-040-213:1385268:1385340 [0] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-040-213:1385268:1385340 [0] NCCL INFO CPU/0-0 (1/1/3) +n116-040-213:1385268:1385340 [0] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-040-213:1385268:1385340 [0] NCCL INFO + PCI[24.0] - GPU/0-4000 (40) +n116-040-213:1385268:1385340 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385268:1385340 [0] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-040-213:1385268:1385340 [0] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-040-213:1385268:1385340 [0] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-040-213:1385268:1385340 [0] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-040-213:1385268:1385340 [0] NCCL INFO + PCI[24.0] - GPU/0-5000 (41) +n116-040-213:1385268:1385340 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385268:1385340 [0] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-040-213:1385268:1385340 [0] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-040-213:1385268:1385340 [0] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-040-213:1385273:1385338 [5] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-040-213:1385268:1385340 [0] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-040-213:1385268:1385340 [0] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-040-213:1385273:1385338 [5] NCCL INFO CPU/0-0 (1/1/3) +n116-040-213:1385268:1385340 [0] NCCL INFO + PCI[24.0] - GPU/0-b000 (42) +n116-040-213:1385268:1385340 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385273:1385338 [5] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-040-213:1385268:1385340 [0] NCCL INFO + PCI[24.0] - GPU/0-c000 (43) +n116-040-213:1385273:1385338 [5] NCCL INFO + PCI[24.0] - GPU/0-4000 (40) +n116-040-213:1385268:1385340 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385273:1385338 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385268:1385340 [0] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-040-213:1385273:1385338 [5] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-040-213:1385268:1385340 [0] NCCL INFO CPU/0-1 (1/1/3) +n116-040-213:1385269:1385342 [1] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-040-213:1385268:1385340 [0] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-040-213:1385273:1385338 [5] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-040-213:1385268:1385340 [0] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-040-213:1385273:1385338 [5] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-040-213:1385268:1385340 [0] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-040-213:1385273:1385338 [5] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-040-213:1385268:1385340 [0] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-040-213:1385273:1385338 [5] NCCL INFO + PCI[24.0] - GPU/0-5000 (41) +n116-040-213:1385268:1385340 [0] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-040-213:1385273:1385338 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385268:1385340 [0] NCCL INFO + PCI[24.0] - GPU/0-84000 (44) +n116-040-213:1385273:1385338 [5] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-040-213:1385268:1385340 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385273:1385338 [5] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-040-213:1385268:1385340 [0] NCCL INFO + PCI[24.0] - GPU/0-85000 (45) +n116-040-213:1385273:1385338 [5] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-040-213:1385268:1385340 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385273:1385338 [5] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-040-213:1385268:1385340 [0] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-040-213:1385273:1385338 [5] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-040-213:1385268:1385340 [0] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-040-213:1385273:1385338 [5] NCCL INFO + PCI[24.0] - GPU/0-b000 (42) +n116-040-213:1385273:1385338 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385268:1385340 [0] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-040-213:1385273:1385338 [5] NCCL INFO + PCI[24.0] - GPU/0-c000 (43) +n116-040-213:1385268:1385340 [0] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-040-213:1385273:1385338 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385268:1385340 [0] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-040-213:1385273:1385338 [5] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-040-213:1385268:1385340 [0] NCCL INFO + PCI[24.0] - GPU/0-8b000 (46) +n116-040-213:1385273:1385338 [5] NCCL INFO CPU/0-1 (1/1/3) +n116-040-213:1385268:1385340 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385273:1385338 [5] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-040-213:1385268:1385340 [0] NCCL INFO + PCI[24.0] - GPU/0-8c000 (47) +n116-040-213:1385273:1385338 [5] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-040-213:1385268:1385340 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385273:1385338 [5] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-040-213:1385268:1385340 [0] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-040-213:1385268:1385340 [0] NCCL INFO ========================================== +n116-040-213:1385273:1385338 [5] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-040-213:1385273:1385338 [5] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-040-213:1385273:1385338 [5] NCCL INFO + PCI[24.0] - GPU/0-84000 (44) +n116-040-213:1385273:1385338 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385273:1385338 [5] NCCL INFO + PCI[24.0] - GPU/0-85000 (45) +n116-040-213:1385273:1385338 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385273:1385338 [5] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-040-213:1385273:1385338 [5] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-040-213:1385273:1385338 [5] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-040-213:1385268:1385340 [0] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-213:1385273:1385338 [5] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-040-213:1385273:1385338 [5] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-040-213:1385273:1385338 [5] NCCL INFO + PCI[24.0] - GPU/0-8b000 (46) +n116-040-213:1385273:1385338 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385273:1385338 [5] NCCL INFO + PCI[24.0] - GPU/0-8c000 (47) +n116-040-213:1385273:1385338 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385273:1385338 [5] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-040-213:1385268:1385340 [0] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-213:1385273:1385338 [5] NCCL INFO ========================================== +n116-040-213:1385268:1385340 [0] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-213:1385273:1385338 [5] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-213:1385273:1385338 [5] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-213:1385268:1385340 [0] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-213:1385268:1385340 [0] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-213:1385273:1385338 [5] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-213:1385268:1385340 [0] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-213:1385273:1385338 [5] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-213:1385273:1385338 [5] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-213:1385268:1385340 [0] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-040-213:1385268:1385340 [0] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-040-213:1385273:1385338 [5] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-213:1385273:1385338 [5] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-040-213:1385268:1385340 [0] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-213:1385268:1385340 [0] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-213:1385273:1385338 [5] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-040-213:1385273:1385338 [5] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-213:1385268:1385340 [0] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-213:1385268:1385340 [0] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-213:1385273:1385338 [5] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-213:1385273:1385338 [5] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-213:1385268:1385340 [0] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-040-213:1385273:1385338 [5] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-213:1385268:1385340 [0] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-040-213:1385273:1385338 [5] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-040-213:1385268:1385340 [0] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-040-213:1385268:1385340 [0] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-040-213:1385273:1385338 [5] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-040-213:1385273:1385338 [5] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-040-213:1385275:1385341 [7] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-040-213:1385268:1385340 [0] NCCL INFO Setting affinity for GPU 0 to 0fffffff,fffffe00,00000000,000fffff,fffffffe +n116-040-213:1385275:1385341 [7] NCCL INFO CPU/0-0 (1/1/3) +n116-040-213:1385275:1385341 [7] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-040-213:1385273:1385338 [5] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-040-213:1385275:1385341 [7] NCCL INFO + PCI[24.0] - GPU/0-4000 (40) +n116-040-213:1385275:1385341 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385275:1385341 [7] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-040-213:1385273:1385338 [5] NCCL INFO Setting affinity for GPU 5 to 7fff,ffffffff,e0000000,0000007f,ffffffff,ffe00000,00000000 +n116-040-213:1385275:1385341 [7] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-040-213:1385275:1385341 [7] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-040-213:1385275:1385341 [7] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-040-213:1385275:1385341 [7] NCCL INFO + PCI[24.0] - GPU/0-5000 (41) +n116-040-213:1385275:1385341 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385273:1385338 [5] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-040-213:1385275:1385341 [7] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-040-213:1385275:1385341 [7] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-040-213:1385275:1385341 [7] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-040-213:1385275:1385341 [7] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-040-213:1385275:1385341 [7] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-040-213:1385275:1385341 [7] NCCL INFO + PCI[24.0] - GPU/0-b000 (42) +n116-040-213:1385275:1385341 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385275:1385341 [7] NCCL INFO + PCI[24.0] - GPU/0-c000 (43) +n116-040-213:1385275:1385341 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385275:1385341 [7] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-040-213:1385275:1385341 [7] NCCL INFO CPU/0-1 (1/1/3) +n116-040-213:1385275:1385341 [7] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-040-213:1385275:1385341 [7] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-040-213:1385275:1385341 [7] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-040-213:1385273:1385338 [5] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-040-213:1385275:1385341 [7] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-040-213:1385275:1385341 [7] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-040-213:1385275:1385341 [7] NCCL INFO + PCI[24.0] - GPU/0-84000 (44) +n116-040-213:1385275:1385341 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385275:1385341 [7] NCCL INFO + PCI[24.0] - GPU/0-85000 (45) +n116-040-213:1385275:1385341 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385275:1385341 [7] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-040-213:1385275:1385341 [7] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-040-213:1385275:1385341 [7] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-040-213:1385275:1385341 [7] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-040-213:1385275:1385341 [7] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-040-213:1385275:1385341 [7] NCCL INFO + PCI[24.0] - GPU/0-8b000 (46) +n116-040-213:1385275:1385341 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385275:1385341 [7] NCCL INFO + PCI[24.0] - GPU/0-8c000 (47) +n116-040-213:1385275:1385341 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385275:1385341 [7] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-040-213:1385275:1385341 [7] NCCL INFO ========================================== +n116-040-213:1385275:1385341 [7] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-213:1385268:1385340 [0] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-040-213:1385275:1385341 [7] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-213:1385275:1385341 [7] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-213:1385275:1385341 [7] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-213:1385275:1385341 [7] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-213:1385268:1385340 [0] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-040-213:1385275:1385341 [7] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-213:1385275:1385341 [7] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-040-213:1385275:1385341 [7] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-040-213:1385275:1385341 [7] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-213:1385275:1385341 [7] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-213:1385275:1385341 [7] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-213:1385275:1385341 [7] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-213:1385275:1385341 [7] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-040-213:1385275:1385341 [7] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-040-213:1385275:1385341 [7] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-040-213:1385275:1385341 [7] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-040-213:1385275:1385341 [7] NCCL INFO Setting affinity for GPU 7 to 7fff,ffffffff,e0000000,0000007f,ffffffff,ffe00000,00000000 +n116-040-213:1385272:1385339 [4] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-040-213:1385272:1385339 [4] NCCL INFO CPU/0-0 (1/1/3) +n116-040-213:1385272:1385339 [4] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-040-213:1385272:1385339 [4] NCCL INFO + PCI[24.0] - GPU/0-4000 (40) +n116-040-213:1385272:1385339 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385272:1385339 [4] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-040-213:1385272:1385339 [4] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-040-213:1385272:1385339 [4] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-040-213:1385272:1385339 [4] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-040-213:1385272:1385339 [4] NCCL INFO + PCI[24.0] - GPU/0-5000 (41) +n116-040-213:1385272:1385339 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385272:1385339 [4] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-040-213:1385272:1385339 [4] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-040-213:1385272:1385339 [4] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-040-213:1385272:1385339 [4] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-040-213:1385272:1385339 [4] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-040-213:1385272:1385339 [4] NCCL INFO + PCI[24.0] - GPU/0-b000 (42) +n116-040-213:1385272:1385339 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385272:1385339 [4] NCCL INFO + PCI[24.0] - GPU/0-c000 (43) +n116-040-213:1385272:1385339 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385275:1385341 [7] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-040-213:1385272:1385339 [4] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-040-213:1385272:1385339 [4] NCCL INFO CPU/0-1 (1/1/3) +n116-040-213:1385272:1385339 [4] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-040-213:1385272:1385339 [4] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-040-213:1385272:1385339 [4] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-040-213:1385272:1385339 [4] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-040-213:1385272:1385339 [4] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-040-213:1385272:1385339 [4] NCCL INFO + PCI[24.0] - GPU/0-84000 (44) +n116-040-213:1385272:1385339 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385272:1385339 [4] NCCL INFO + PCI[24.0] - GPU/0-85000 (45) +n116-040-213:1385272:1385339 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385272:1385339 [4] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-040-213:1385272:1385339 [4] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-040-213:1385272:1385339 [4] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-040-213:1385272:1385339 [4] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-040-213:1385272:1385339 [4] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-040-213:1385272:1385339 [4] NCCL INFO + PCI[24.0] - GPU/0-8b000 (46) +n116-040-213:1385272:1385339 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385272:1385339 [4] NCCL INFO + PCI[24.0] - GPU/0-8c000 (47) +n116-040-213:1385272:1385339 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385272:1385339 [4] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-040-213:1385272:1385339 [4] NCCL INFO ========================================== +n116-040-213:1385272:1385339 [4] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-213:1385274:1385336 [6] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-040-213:1385274:1385336 [6] NCCL INFO CPU/0-0 (1/1/3) +n116-040-213:1385274:1385336 [6] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-040-213:1385274:1385336 [6] NCCL INFO + PCI[24.0] - GPU/0-4000 (40) +n116-040-213:1385272:1385339 [4] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-213:1385275:1385341 [7] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-040-213:1385274:1385336 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385274:1385336 [6] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-040-213:1385274:1385336 [6] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-040-213:1385274:1385336 [6] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-040-213:1385274:1385336 [6] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-040-213:1385272:1385339 [4] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-213:1385274:1385336 [6] NCCL INFO + PCI[24.0] - GPU/0-5000 (41) +n116-040-213:1385274:1385336 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385274:1385336 [6] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-040-213:1385274:1385336 [6] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-040-213:1385272:1385339 [4] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-213:1385274:1385336 [6] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-040-213:1385271:1385337 [3] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-040-213:1385274:1385336 [6] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-040-213:1385274:1385336 [6] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-040-213:1385271:1385337 [3] NCCL INFO CPU/0-0 (1/1/3) +n116-040-213:1385274:1385336 [6] NCCL INFO + PCI[24.0] - GPU/0-b000 (42) +n116-040-213:1385272:1385339 [4] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-213:1385271:1385337 [3] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-040-213:1385274:1385336 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385271:1385337 [3] NCCL INFO + PCI[24.0] - GPU/0-4000 (40) +n116-040-213:1385274:1385336 [6] NCCL INFO + PCI[24.0] - GPU/0-c000 (43) +n116-040-213:1385271:1385337 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385274:1385336 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385271:1385337 [3] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-040-213:1385274:1385336 [6] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-040-213:1385272:1385339 [4] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-213:1385271:1385337 [3] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-040-213:1385274:1385336 [6] NCCL INFO CPU/0-1 (1/1/3) +n116-040-213:1385271:1385337 [3] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-040-213:1385274:1385336 [6] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-040-213:1385274:1385336 [6] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-040-213:1385271:1385337 [3] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-040-213:1385274:1385336 [6] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-040-213:1385271:1385337 [3] NCCL INFO + PCI[24.0] - GPU/0-5000 (41) +n116-040-213:1385271:1385337 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385274:1385336 [6] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-040-213:1385272:1385339 [4] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-040-213:1385271:1385337 [3] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-040-213:1385274:1385336 [6] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-040-213:1385271:1385337 [3] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-040-213:1385274:1385336 [6] NCCL INFO + PCI[24.0] - GPU/0-84000 (44) +n116-040-213:1385274:1385336 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385271:1385337 [3] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-040-213:1385274:1385336 [6] NCCL INFO + PCI[24.0] - GPU/0-85000 (45) +n116-040-213:1385272:1385339 [4] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-040-213:1385271:1385337 [3] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-040-213:1385274:1385336 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385274:1385336 [6] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-040-213:1385271:1385337 [3] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-040-213:1385274:1385336 [6] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-040-213:1385271:1385337 [3] NCCL INFO + PCI[24.0] - GPU/0-b000 (42) +n116-040-213:1385271:1385337 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385274:1385336 [6] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-040-213:1385271:1385337 [3] NCCL INFO + PCI[24.0] - GPU/0-c000 (43) +n116-040-213:1385272:1385339 [4] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-213:1385274:1385336 [6] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-040-213:1385271:1385337 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385274:1385336 [6] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-040-213:1385271:1385337 [3] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-040-213:1385274:1385336 [6] NCCL INFO + PCI[24.0] - GPU/0-8b000 (46) +n116-040-213:1385271:1385337 [3] NCCL INFO CPU/0-1 (1/1/3) +n116-040-213:1385274:1385336 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385271:1385337 [3] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-040-213:1385274:1385336 [6] NCCL INFO + PCI[24.0] - GPU/0-8c000 (47) +n116-040-213:1385271:1385337 [3] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-040-213:1385272:1385339 [4] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-213:1385274:1385336 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385271:1385337 [3] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-040-213:1385274:1385336 [6] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-040-213:1385271:1385337 [3] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-040-213:1385271:1385337 [3] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-040-213:1385274:1385336 [6] NCCL INFO ========================================== +n116-040-213:1385271:1385337 [3] NCCL INFO + PCI[24.0] - GPU/0-84000 (44) +n116-040-213:1385272:1385339 [4] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-213:1385271:1385337 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385271:1385337 [3] NCCL INFO + PCI[24.0] - GPU/0-85000 (45) +n116-040-213:1385271:1385337 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385271:1385337 [3] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-040-213:1385274:1385336 [6] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-213:1385272:1385339 [4] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-213:1385271:1385337 [3] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-040-213:1385271:1385337 [3] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-040-213:1385271:1385337 [3] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-040-213:1385274:1385336 [6] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-213:1385271:1385337 [3] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-040-213:1385272:1385339 [4] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-040-213:1385271:1385337 [3] NCCL INFO + PCI[24.0] - GPU/0-8b000 (46) +n116-040-213:1385271:1385337 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385271:1385337 [3] NCCL INFO + PCI[24.0] - GPU/0-8c000 (47) +n116-040-213:1385271:1385337 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-040-213:1385271:1385337 [3] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-040-213:1385271:1385337 [3] NCCL INFO ========================================== +n116-040-213:1385272:1385339 [4] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-040-213:1385274:1385336 [6] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-213:1385271:1385337 [3] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-213:1385272:1385339 [4] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-040-213:1385274:1385336 [6] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-213:1385272:1385339 [4] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-040-213:1385271:1385337 [3] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-213:1385274:1385336 [6] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-213:1385272:1385339 [4] NCCL INFO Setting affinity for GPU 4 to 7fff,ffffffff,e0000000,0000007f,ffffffff,ffe00000,00000000 +n116-040-213:1385271:1385337 [3] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-213:1385272:1385339 [4] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-040-213:1385271:1385337 [3] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-213:1385274:1385336 [6] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-213:1385271:1385337 [3] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-213:1385274:1385336 [6] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-040-213:1385274:1385336 [6] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-040-213:1385271:1385337 [3] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-040-213:1385274:1385336 [6] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-213:1385271:1385337 [3] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-040-213:1385272:1385339 [4] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-040-213:1385274:1385336 [6] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-213:1385271:1385337 [3] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-040-213:1385274:1385336 [6] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-213:1385271:1385337 [3] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-213:1385274:1385336 [6] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-213:1385271:1385337 [3] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-213:1385274:1385336 [6] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-040-213:1385271:1385337 [3] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-213:1385271:1385337 [3] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-040-213:1385274:1385336 [6] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-040-213:1385271:1385337 [3] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-040-213:1385274:1385336 [6] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-040-213:1385271:1385337 [3] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-040-213:1385274:1385336 [6] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-040-213:1385274:1385336 [6] NCCL INFO Setting affinity for GPU 6 to 7fff,ffffffff,e0000000,0000007f,ffffffff,ffe00000,00000000 +n116-040-213:1385271:1385337 [3] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-040-213:1385271:1385337 [3] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-040-213:1385271:1385337 [3] NCCL INFO Setting affinity for GPU 3 to 0fffffff,fffffe00,00000000,000fffff,fffffffe +n116-040-213:1385271:1385337 [3] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-040-213:1385274:1385336 [6] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-040-213:1385271:1385337 [3] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-040-213:1385274:1385336 [6] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-040-213:1385273:1385338 [5] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-040-213:1385273:1385338 [5] NCCL INFO 0 : NET/0-0 GPU/40 GPU/47 GPU/46 GPU/45 GPU/44 GPU/43 GPU/42 GPU/41 NET/0-0 +n116-040-213:1385273:1385338 [5] NCCL INFO 1 : NET/0-1 GPU/41 GPU/42 GPU/43 GPU/44 GPU/45 GPU/46 GPU/47 GPU/40 NET/0-1 +n116-040-213:1385273:1385338 [5] NCCL INFO 2 : NET/0-2 GPU/42 GPU/41 GPU/40 GPU/47 GPU/46 GPU/45 GPU/44 GPU/43 NET/0-2 +n116-040-213:1385273:1385338 [5] NCCL INFO 3 : NET/0-3 GPU/43 GPU/44 GPU/45 GPU/46 GPU/47 GPU/40 GPU/41 GPU/42 NET/0-3 +n116-040-213:1385273:1385338 [5] NCCL INFO 4 : NET/0-4 GPU/44 GPU/43 GPU/42 GPU/41 GPU/40 GPU/47 GPU/46 GPU/45 NET/0-4 +n116-040-213:1385273:1385338 [5] NCCL INFO 5 : NET/0-5 GPU/45 GPU/46 GPU/47 GPU/40 GPU/41 GPU/42 GPU/43 GPU/44 NET/0-5 +n116-040-213:1385273:1385338 [5] NCCL INFO 6 : NET/0-6 GPU/46 GPU/45 GPU/44 GPU/43 GPU/42 GPU/41 GPU/40 GPU/47 NET/0-6 +n116-040-213:1385273:1385338 [5] NCCL INFO 7 : NET/0-7 GPU/47 GPU/40 GPU/41 GPU/42 GPU/43 GPU/44 GPU/45 GPU/46 NET/0-7 +n116-040-213:1385273:1385338 [5] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-040-213:1385273:1385338 [5] NCCL INFO 0 : NET/0-0 GPU/40 GPU/41 GPU/42 GPU/43 GPU/44 GPU/45 GPU/46 GPU/47 NET/0-0 +n116-040-213:1385273:1385338 [5] NCCL INFO 1 : NET/0-1 GPU/41 GPU/40 GPU/47 GPU/46 GPU/45 GPU/44 GPU/43 GPU/42 NET/0-1 +n116-040-213:1385273:1385338 [5] NCCL INFO 2 : NET/0-2 GPU/42 GPU/43 GPU/44 GPU/45 GPU/46 GPU/47 GPU/40 GPU/41 NET/0-2 +n116-040-213:1385273:1385338 [5] NCCL INFO 3 : NET/0-3 GPU/43 GPU/42 GPU/41 GPU/40 GPU/47 GPU/46 GPU/45 GPU/44 NET/0-3 +n116-040-213:1385273:1385338 [5] NCCL INFO 4 : NET/0-4 GPU/44 GPU/45 GPU/46 GPU/47 GPU/40 GPU/41 GPU/42 GPU/43 NET/0-4 +n116-040-213:1385273:1385338 [5] NCCL INFO 5 : NET/0-5 GPU/45 GPU/44 GPU/43 GPU/42 GPU/41 GPU/40 GPU/47 GPU/46 NET/0-5 +n116-040-213:1385273:1385338 [5] NCCL INFO 6 : NET/0-6 GPU/46 GPU/47 GPU/40 GPU/41 GPU/42 GPU/43 GPU/44 GPU/45 NET/0-6 +n116-040-213:1385273:1385338 [5] NCCL INFO 7 : NET/0-7 GPU/47 GPU/46 GPU/45 GPU/44 GPU/43 GPU/42 GPU/41 GPU/40 NET/0-7 +n116-040-213:1385268:1385340 [0] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-040-213:1385268:1385340 [0] NCCL INFO 0 : NET/0-0 GPU/40 GPU/47 GPU/46 GPU/45 GPU/44 GPU/43 GPU/42 GPU/41 NET/0-0 +n116-040-213:1385268:1385340 [0] NCCL INFO 1 : NET/0-1 GPU/41 GPU/42 GPU/43 GPU/44 GPU/45 GPU/46 GPU/47 GPU/40 NET/0-1 +n116-040-213:1385268:1385340 [0] NCCL INFO 2 : NET/0-2 GPU/42 GPU/41 GPU/40 GPU/47 GPU/46 GPU/45 GPU/44 GPU/43 NET/0-2 +n116-040-213:1385268:1385340 [0] NCCL INFO 3 : NET/0-3 GPU/43 GPU/44 GPU/45 GPU/46 GPU/47 GPU/40 GPU/41 GPU/42 NET/0-3 +n116-040-213:1385268:1385340 [0] NCCL INFO 4 : NET/0-4 GPU/44 GPU/43 GPU/42 GPU/41 GPU/40 GPU/47 GPU/46 GPU/45 NET/0-4 +n116-040-213:1385268:1385340 [0] NCCL INFO 5 : NET/0-5 GPU/45 GPU/46 GPU/47 GPU/40 GPU/41 GPU/42 GPU/43 GPU/44 NET/0-5 +n116-040-213:1385268:1385340 [0] NCCL INFO 6 : NET/0-6 GPU/46 GPU/45 GPU/44 GPU/43 GPU/42 GPU/41 GPU/40 GPU/47 NET/0-6 +n116-040-213:1385268:1385340 [0] NCCL INFO 7 : NET/0-7 GPU/47 GPU/40 GPU/41 GPU/42 GPU/43 GPU/44 GPU/45 GPU/46 NET/0-7 +n116-040-213:1385268:1385340 [0] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-040-213:1385268:1385340 [0] NCCL INFO 0 : NET/0-0 GPU/40 GPU/41 GPU/42 GPU/43 GPU/44 GPU/45 GPU/46 GPU/47 NET/0-0 +n116-040-213:1385268:1385340 [0] NCCL INFO 1 : NET/0-1 GPU/41 GPU/40 GPU/47 GPU/46 GPU/45 GPU/44 GPU/43 GPU/42 NET/0-1 +n116-040-213:1385268:1385340 [0] NCCL INFO 2 : NET/0-2 GPU/42 GPU/43 GPU/44 GPU/45 GPU/46 GPU/47 GPU/40 GPU/41 NET/0-2 +n116-040-213:1385268:1385340 [0] NCCL INFO 3 : NET/0-3 GPU/43 GPU/42 GPU/41 GPU/40 GPU/47 GPU/46 GPU/45 GPU/44 NET/0-3 +n116-040-213:1385268:1385340 [0] NCCL INFO 4 : NET/0-4 GPU/44 GPU/45 GPU/46 GPU/47 GPU/40 GPU/41 GPU/42 GPU/43 NET/0-4 +n116-040-213:1385268:1385340 [0] NCCL INFO 5 : NET/0-5 GPU/45 GPU/44 GPU/43 GPU/42 GPU/41 GPU/40 GPU/47 GPU/46 NET/0-5 +n116-040-213:1385268:1385340 [0] NCCL INFO 6 : NET/0-6 GPU/46 GPU/47 GPU/40 GPU/41 GPU/42 GPU/43 GPU/44 GPU/45 NET/0-6 +n116-040-213:1385268:1385340 [0] NCCL INFO 7 : NET/0-7 GPU/47 GPU/46 GPU/45 GPU/44 GPU/43 GPU/42 GPU/41 GPU/40 NET/0-7 +n116-040-213:1385272:1385339 [4] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-040-213:1385272:1385339 [4] NCCL INFO 0 : NET/0-0 GPU/40 GPU/47 GPU/46 GPU/45 GPU/44 GPU/43 GPU/42 GPU/41 NET/0-0 +n116-040-213:1385272:1385339 [4] NCCL INFO 1 : NET/0-1 GPU/41 GPU/42 GPU/43 GPU/44 GPU/45 GPU/46 GPU/47 GPU/40 NET/0-1 +n116-040-213:1385272:1385339 [4] NCCL INFO 2 : NET/0-2 GPU/42 GPU/41 GPU/40 GPU/47 GPU/46 GPU/45 GPU/44 GPU/43 NET/0-2 +n116-040-213:1385272:1385339 [4] NCCL INFO 3 : NET/0-3 GPU/43 GPU/44 GPU/45 GPU/46 GPU/47 GPU/40 GPU/41 GPU/42 NET/0-3 +n116-040-213:1385272:1385339 [4] NCCL INFO 4 : NET/0-4 GPU/44 GPU/43 GPU/42 GPU/41 GPU/40 GPU/47 GPU/46 GPU/45 NET/0-4 +n116-040-213:1385272:1385339 [4] NCCL INFO 5 : NET/0-5 GPU/45 GPU/46 GPU/47 GPU/40 GPU/41 GPU/42 GPU/43 GPU/44 NET/0-5 +n116-040-213:1385272:1385339 [4] NCCL INFO 6 : NET/0-6 GPU/46 GPU/45 GPU/44 GPU/43 GPU/42 GPU/41 GPU/40 GPU/47 NET/0-6 +n116-040-213:1385272:1385339 [4] NCCL INFO 7 : NET/0-7 GPU/47 GPU/40 GPU/41 GPU/42 GPU/43 GPU/44 GPU/45 GPU/46 NET/0-7 +n116-040-213:1385272:1385339 [4] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-040-213:1385272:1385339 [4] NCCL INFO 0 : NET/0-0 GPU/40 GPU/41 GPU/42 GPU/43 GPU/44 GPU/45 GPU/46 GPU/47 NET/0-0 +n116-040-213:1385272:1385339 [4] NCCL INFO 1 : NET/0-1 GPU/41 GPU/40 GPU/47 GPU/46 GPU/45 GPU/44 GPU/43 GPU/42 NET/0-1 +n116-040-213:1385272:1385339 [4] NCCL INFO 2 : NET/0-2 GPU/42 GPU/43 GPU/44 GPU/45 GPU/46 GPU/47 GPU/40 GPU/41 NET/0-2 +n116-040-213:1385272:1385339 [4] NCCL INFO 3 : NET/0-3 GPU/43 GPU/42 GPU/41 GPU/40 GPU/47 GPU/46 GPU/45 GPU/44 NET/0-3 +n116-040-213:1385272:1385339 [4] NCCL INFO 4 : NET/0-4 GPU/44 GPU/45 GPU/46 GPU/47 GPU/40 GPU/41 GPU/42 GPU/43 NET/0-4 +n116-040-213:1385272:1385339 [4] NCCL INFO 5 : NET/0-5 GPU/45 GPU/44 GPU/43 GPU/42 GPU/41 GPU/40 GPU/47 GPU/46 NET/0-5 +n116-040-213:1385272:1385339 [4] NCCL INFO 6 : NET/0-6 GPU/46 GPU/47 GPU/40 GPU/41 GPU/42 GPU/43 GPU/44 GPU/45 NET/0-6 +n116-040-213:1385272:1385339 [4] NCCL INFO 7 : NET/0-7 GPU/47 GPU/46 GPU/45 GPU/44 GPU/43 GPU/42 GPU/41 GPU/40 NET/0-7 +n116-040-213:1385271:1385337 [3] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-040-213:1385271:1385337 [3] NCCL INFO 0 : NET/0-0 GPU/40 GPU/47 GPU/46 GPU/45 GPU/44 GPU/43 GPU/42 GPU/41 NET/0-0 +n116-040-213:1385271:1385337 [3] NCCL INFO 1 : NET/0-1 GPU/41 GPU/42 GPU/43 GPU/44 GPU/45 GPU/46 GPU/47 GPU/40 NET/0-1 +n116-040-213:1385271:1385337 [3] NCCL INFO 2 : NET/0-2 GPU/42 GPU/41 GPU/40 GPU/47 GPU/46 GPU/45 GPU/44 GPU/43 NET/0-2 +n116-040-213:1385271:1385337 [3] NCCL INFO 3 : NET/0-3 GPU/43 GPU/44 GPU/45 GPU/46 GPU/47 GPU/40 GPU/41 GPU/42 NET/0-3 +n116-040-213:1385271:1385337 [3] NCCL INFO 4 : NET/0-4 GPU/44 GPU/43 GPU/42 GPU/41 GPU/40 GPU/47 GPU/46 GPU/45 NET/0-4 +n116-040-213:1385271:1385337 [3] NCCL INFO 5 : NET/0-5 GPU/45 GPU/46 GPU/47 GPU/40 GPU/41 GPU/42 GPU/43 GPU/44 NET/0-5 +n116-040-213:1385271:1385337 [3] NCCL INFO 6 : NET/0-6 GPU/46 GPU/45 GPU/44 GPU/43 GPU/42 GPU/41 GPU/40 GPU/47 NET/0-6 +n116-040-213:1385271:1385337 [3] NCCL INFO 7 : NET/0-7 GPU/47 GPU/40 GPU/41 GPU/42 GPU/43 GPU/44 GPU/45 GPU/46 NET/0-7 +n116-040-213:1385271:1385337 [3] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-040-213:1385271:1385337 [3] NCCL INFO 0 : NET/0-0 GPU/40 GPU/41 GPU/42 GPU/43 GPU/44 GPU/45 GPU/46 GPU/47 NET/0-0 +n116-040-213:1385271:1385337 [3] NCCL INFO 1 : NET/0-1 GPU/41 GPU/40 GPU/47 GPU/46 GPU/45 GPU/44 GPU/43 GPU/42 NET/0-1 +n116-040-213:1385271:1385337 [3] NCCL INFO 2 : NET/0-2 GPU/42 GPU/43 GPU/44 GPU/45 GPU/46 GPU/47 GPU/40 GPU/41 NET/0-2 +n116-040-213:1385271:1385337 [3] NCCL INFO 3 : NET/0-3 GPU/43 GPU/42 GPU/41 GPU/40 GPU/47 GPU/46 GPU/45 GPU/44 NET/0-3 +n116-040-213:1385271:1385337 [3] NCCL INFO 4 : NET/0-4 GPU/44 GPU/45 GPU/46 GPU/47 GPU/40 GPU/41 GPU/42 GPU/43 NET/0-4 +n116-040-213:1385271:1385337 [3] NCCL INFO 5 : NET/0-5 GPU/45 GPU/44 GPU/43 GPU/42 GPU/41 GPU/40 GPU/47 GPU/46 NET/0-5 +n116-040-213:1385271:1385337 [3] NCCL INFO 6 : NET/0-6 GPU/46 GPU/47 GPU/40 GPU/41 GPU/42 GPU/43 GPU/44 GPU/45 NET/0-6 +n116-040-213:1385271:1385337 [3] NCCL INFO 7 : NET/0-7 GPU/47 GPU/46 GPU/45 GPU/44 GPU/43 GPU/42 GPU/41 GPU/40 NET/0-7 +n116-040-213:1385270:1385343 [2] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-040-213:1385270:1385343 [2] NCCL INFO 0 : NET/0-0 GPU/40 GPU/47 GPU/46 GPU/45 GPU/44 GPU/43 GPU/42 GPU/41 NET/0-0 +n116-040-213:1385270:1385343 [2] NCCL INFO 1 : NET/0-1 GPU/41 GPU/42 GPU/43 GPU/44 GPU/45 GPU/46 GPU/47 GPU/40 NET/0-1 +n116-040-213:1385270:1385343 [2] NCCL INFO 2 : NET/0-2 GPU/42 GPU/41 GPU/40 GPU/47 GPU/46 GPU/45 GPU/44 GPU/43 NET/0-2 +n116-040-213:1385270:1385343 [2] NCCL INFO 3 : NET/0-3 GPU/43 GPU/44 GPU/45 GPU/46 GPU/47 GPU/40 GPU/41 GPU/42 NET/0-3 +n116-040-213:1385270:1385343 [2] NCCL INFO 4 : NET/0-4 GPU/44 GPU/43 GPU/42 GPU/41 GPU/40 GPU/47 GPU/46 GPU/45 NET/0-4 +n116-040-213:1385270:1385343 [2] NCCL INFO 5 : NET/0-5 GPU/45 GPU/46 GPU/47 GPU/40 GPU/41 GPU/42 GPU/43 GPU/44 NET/0-5 +n116-040-213:1385270:1385343 [2] NCCL INFO 6 : NET/0-6 GPU/46 GPU/45 GPU/44 GPU/43 GPU/42 GPU/41 GPU/40 GPU/47 NET/0-6 +n116-040-213:1385270:1385343 [2] NCCL INFO 7 : NET/0-7 GPU/47 GPU/40 GPU/41 GPU/42 GPU/43 GPU/44 GPU/45 GPU/46 NET/0-7 +n116-040-213:1385270:1385343 [2] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-040-213:1385270:1385343 [2] NCCL INFO 0 : NET/0-0 GPU/40 GPU/41 GPU/42 GPU/43 GPU/44 GPU/45 GPU/46 GPU/47 NET/0-0 +n116-040-213:1385270:1385343 [2] NCCL INFO 1 : NET/0-1 GPU/41 GPU/40 GPU/47 GPU/46 GPU/45 GPU/44 GPU/43 GPU/42 NET/0-1 +n116-040-213:1385270:1385343 [2] NCCL INFO 2 : NET/0-2 GPU/42 GPU/43 GPU/44 GPU/45 GPU/46 GPU/47 GPU/40 GPU/41 NET/0-2 +n116-040-213:1385270:1385343 [2] NCCL INFO 3 : NET/0-3 GPU/43 GPU/42 GPU/41 GPU/40 GPU/47 GPU/46 GPU/45 GPU/44 NET/0-3 +n116-040-213:1385270:1385343 [2] NCCL INFO 4 : NET/0-4 GPU/44 GPU/45 GPU/46 GPU/47 GPU/40 GPU/41 GPU/42 GPU/43 NET/0-4 +n116-040-213:1385270:1385343 [2] NCCL INFO 5 : NET/0-5 GPU/45 GPU/44 GPU/43 GPU/42 GPU/41 GPU/40 GPU/47 GPU/46 NET/0-5 +n116-040-213:1385270:1385343 [2] NCCL INFO 6 : NET/0-6 GPU/46 GPU/47 GPU/40 GPU/41 GPU/42 GPU/43 GPU/44 GPU/45 NET/0-6 +n116-040-213:1385270:1385343 [2] NCCL INFO 7 : NET/0-7 GPU/47 GPU/46 GPU/45 GPU/44 GPU/43 GPU/42 GPU/41 GPU/40 NET/0-7 +n116-040-213:1385269:1385342 [1] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-040-213:1385269:1385342 [1] NCCL INFO 0 : NET/0-0 GPU/40 GPU/47 GPU/46 GPU/45 GPU/44 GPU/43 GPU/42 GPU/41 NET/0-0 +n116-040-213:1385269:1385342 [1] NCCL INFO 1 : NET/0-1 GPU/41 GPU/42 GPU/43 GPU/44 GPU/45 GPU/46 GPU/47 GPU/40 NET/0-1 +n116-040-213:1385269:1385342 [1] NCCL INFO 2 : NET/0-2 GPU/42 GPU/41 GPU/40 GPU/47 GPU/46 GPU/45 GPU/44 GPU/43 NET/0-2 +n116-040-213:1385269:1385342 [1] NCCL INFO 3 : NET/0-3 GPU/43 GPU/44 GPU/45 GPU/46 GPU/47 GPU/40 GPU/41 GPU/42 NET/0-3 +n116-040-213:1385269:1385342 [1] NCCL INFO 4 : NET/0-4 GPU/44 GPU/43 GPU/42 GPU/41 GPU/40 GPU/47 GPU/46 GPU/45 NET/0-4 +n116-040-213:1385269:1385342 [1] NCCL INFO 5 : NET/0-5 GPU/45 GPU/46 GPU/47 GPU/40 GPU/41 GPU/42 GPU/43 GPU/44 NET/0-5 +n116-040-213:1385269:1385342 [1] NCCL INFO 6 : NET/0-6 GPU/46 GPU/45 GPU/44 GPU/43 GPU/42 GPU/41 GPU/40 GPU/47 NET/0-6 +n116-040-213:1385269:1385342 [1] NCCL INFO 7 : NET/0-7 GPU/47 GPU/40 GPU/41 GPU/42 GPU/43 GPU/44 GPU/45 GPU/46 NET/0-7 +n116-040-213:1385269:1385342 [1] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-040-213:1385269:1385342 [1] NCCL INFO 0 : NET/0-0 GPU/40 GPU/41 GPU/42 GPU/43 GPU/44 GPU/45 GPU/46 GPU/47 NET/0-0 +n116-040-213:1385269:1385342 [1] NCCL INFO 1 : NET/0-1 GPU/41 GPU/40 GPU/47 GPU/46 GPU/45 GPU/44 GPU/43 GPU/42 NET/0-1 +n116-040-213:1385269:1385342 [1] NCCL INFO 2 : NET/0-2 GPU/42 GPU/43 GPU/44 GPU/45 GPU/46 GPU/47 GPU/40 GPU/41 NET/0-2 +n116-040-213:1385269:1385342 [1] NCCL INFO 3 : NET/0-3 GPU/43 GPU/42 GPU/41 GPU/40 GPU/47 GPU/46 GPU/45 GPU/44 NET/0-3 +n116-040-213:1385269:1385342 [1] NCCL INFO 4 : NET/0-4 GPU/44 GPU/45 GPU/46 GPU/47 GPU/40 GPU/41 GPU/42 GPU/43 NET/0-4 +n116-040-213:1385269:1385342 [1] NCCL INFO 5 : NET/0-5 GPU/45 GPU/44 GPU/43 GPU/42 GPU/41 GPU/40 GPU/47 GPU/46 NET/0-5 +n116-040-213:1385269:1385342 [1] NCCL INFO 6 : NET/0-6 GPU/46 GPU/47 GPU/40 GPU/41 GPU/42 GPU/43 GPU/44 GPU/45 NET/0-6 +n116-040-213:1385269:1385342 [1] NCCL INFO 7 : NET/0-7 GPU/47 GPU/46 GPU/45 GPU/44 GPU/43 GPU/42 GPU/41 GPU/40 NET/0-7 +n116-040-213:1385275:1385341 [7] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-040-213:1385275:1385341 [7] NCCL INFO 0 : NET/0-0 GPU/40 GPU/47 GPU/46 GPU/45 GPU/44 GPU/43 GPU/42 GPU/41 NET/0-0 +n116-040-213:1385275:1385341 [7] NCCL INFO 1 : NET/0-1 GPU/41 GPU/42 GPU/43 GPU/44 GPU/45 GPU/46 GPU/47 GPU/40 NET/0-1 +n116-040-213:1385275:1385341 [7] NCCL INFO 2 : NET/0-2 GPU/42 GPU/41 GPU/40 GPU/47 GPU/46 GPU/45 GPU/44 GPU/43 NET/0-2 +n116-040-213:1385275:1385341 [7] NCCL INFO 3 : NET/0-3 GPU/43 GPU/44 GPU/45 GPU/46 GPU/47 GPU/40 GPU/41 GPU/42 NET/0-3 +n116-040-213:1385275:1385341 [7] NCCL INFO 4 : NET/0-4 GPU/44 GPU/43 GPU/42 GPU/41 GPU/40 GPU/47 GPU/46 GPU/45 NET/0-4 +n116-040-213:1385275:1385341 [7] NCCL INFO 5 : NET/0-5 GPU/45 GPU/46 GPU/47 GPU/40 GPU/41 GPU/42 GPU/43 GPU/44 NET/0-5 +n116-040-213:1385275:1385341 [7] NCCL INFO 6 : NET/0-6 GPU/46 GPU/45 GPU/44 GPU/43 GPU/42 GPU/41 GPU/40 GPU/47 NET/0-6 +n116-040-213:1385275:1385341 [7] NCCL INFO 7 : NET/0-7 GPU/47 GPU/40 GPU/41 GPU/42 GPU/43 GPU/44 GPU/45 GPU/46 NET/0-7 +n116-040-213:1385275:1385341 [7] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-040-213:1385275:1385341 [7] NCCL INFO 0 : NET/0-0 GPU/40 GPU/41 GPU/42 GPU/43 GPU/44 GPU/45 GPU/46 GPU/47 NET/0-0 +n116-040-213:1385275:1385341 [7] NCCL INFO 1 : NET/0-1 GPU/41 GPU/40 GPU/47 GPU/46 GPU/45 GPU/44 GPU/43 GPU/42 NET/0-1 +n116-040-213:1385275:1385341 [7] NCCL INFO 2 : NET/0-2 GPU/42 GPU/43 GPU/44 GPU/45 GPU/46 GPU/47 GPU/40 GPU/41 NET/0-2 +n116-040-213:1385275:1385341 [7] NCCL INFO 3 : NET/0-3 GPU/43 GPU/42 GPU/41 GPU/40 GPU/47 GPU/46 GPU/45 GPU/44 NET/0-3 +n116-040-213:1385275:1385341 [7] NCCL INFO 4 : NET/0-4 GPU/44 GPU/45 GPU/46 GPU/47 GPU/40 GPU/41 GPU/42 GPU/43 NET/0-4 +n116-040-213:1385275:1385341 [7] NCCL INFO 5 : NET/0-5 GPU/45 GPU/44 GPU/43 GPU/42 GPU/41 GPU/40 GPU/47 GPU/46 NET/0-5 +n116-040-213:1385275:1385341 [7] NCCL INFO 6 : NET/0-6 GPU/46 GPU/47 GPU/40 GPU/41 GPU/42 GPU/43 GPU/44 GPU/45 NET/0-6 +n116-040-213:1385275:1385341 [7] NCCL INFO 7 : NET/0-7 GPU/47 GPU/46 GPU/45 GPU/44 GPU/43 GPU/42 GPU/41 GPU/40 NET/0-7 +n116-040-213:1385274:1385336 [6] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-040-213:1385274:1385336 [6] NCCL INFO 0 : NET/0-0 GPU/40 GPU/47 GPU/46 GPU/45 GPU/44 GPU/43 GPU/42 GPU/41 NET/0-0 +n116-040-213:1385274:1385336 [6] NCCL INFO 1 : NET/0-1 GPU/41 GPU/42 GPU/43 GPU/44 GPU/45 GPU/46 GPU/47 GPU/40 NET/0-1 +n116-040-213:1385274:1385336 [6] NCCL INFO 2 : NET/0-2 GPU/42 GPU/41 GPU/40 GPU/47 GPU/46 GPU/45 GPU/44 GPU/43 NET/0-2 +n116-040-213:1385274:1385336 [6] NCCL INFO 3 : NET/0-3 GPU/43 GPU/44 GPU/45 GPU/46 GPU/47 GPU/40 GPU/41 GPU/42 NET/0-3 +n116-040-213:1385274:1385336 [6] NCCL INFO 4 : NET/0-4 GPU/44 GPU/43 GPU/42 GPU/41 GPU/40 GPU/47 GPU/46 GPU/45 NET/0-4 +n116-040-213:1385274:1385336 [6] NCCL INFO 5 : NET/0-5 GPU/45 GPU/46 GPU/47 GPU/40 GPU/41 GPU/42 GPU/43 GPU/44 NET/0-5 +n116-040-213:1385274:1385336 [6] NCCL INFO 6 : NET/0-6 GPU/46 GPU/45 GPU/44 GPU/43 GPU/42 GPU/41 GPU/40 GPU/47 NET/0-6 +n116-040-213:1385274:1385336 [6] NCCL INFO 7 : NET/0-7 GPU/47 GPU/40 GPU/41 GPU/42 GPU/43 GPU/44 GPU/45 GPU/46 NET/0-7 +n116-040-213:1385274:1385336 [6] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-040-213:1385274:1385336 [6] NCCL INFO 0 : NET/0-0 GPU/40 GPU/41 GPU/42 GPU/43 GPU/44 GPU/45 GPU/46 GPU/47 NET/0-0 +n116-040-213:1385274:1385336 [6] NCCL INFO 1 : NET/0-1 GPU/41 GPU/40 GPU/47 GPU/46 GPU/45 GPU/44 GPU/43 GPU/42 NET/0-1 +n116-040-213:1385274:1385336 [6] NCCL INFO 2 : NET/0-2 GPU/42 GPU/43 GPU/44 GPU/45 GPU/46 GPU/47 GPU/40 GPU/41 NET/0-2 +n116-040-213:1385274:1385336 [6] NCCL INFO 3 : NET/0-3 GPU/43 GPU/42 GPU/41 GPU/40 GPU/47 GPU/46 GPU/45 GPU/44 NET/0-3 +n116-040-213:1385274:1385336 [6] NCCL INFO 4 : NET/0-4 GPU/44 GPU/45 GPU/46 GPU/47 GPU/40 GPU/41 GPU/42 GPU/43 NET/0-4 +n116-040-213:1385274:1385336 [6] NCCL INFO 5 : NET/0-5 GPU/45 GPU/44 GPU/43 GPU/42 GPU/41 GPU/40 GPU/47 GPU/46 NET/0-5 +n116-040-213:1385274:1385336 [6] NCCL INFO 6 : NET/0-6 GPU/46 GPU/47 GPU/40 GPU/41 GPU/42 GPU/43 GPU/44 GPU/45 NET/0-6 +n116-040-213:1385274:1385336 [6] NCCL INFO 7 : NET/0-7 GPU/47 GPU/46 GPU/45 GPU/44 GPU/43 GPU/42 GPU/41 GPU/40 NET/0-7 +n116-040-213:1385274:1385336 [6] NCCL INFO comm 0xc6f23a0 rank 46 nRanks 64 nNodes 8 localRanks 8 localRank 6 MNNVL 0 +n116-040-213:1385275:1385341 [7] NCCL INFO comm 0xb2afe50 rank 47 nRanks 64 nNodes 8 localRanks 8 localRank 7 MNNVL 0 +n116-040-213:1385273:1385338 [5] NCCL INFO comm 0xbc479c0 rank 45 nRanks 64 nNodes 8 localRanks 8 localRank 5 MNNVL 0 +n116-040-213:1385272:1385339 [4] NCCL INFO comm 0xd0e41d0 rank 44 nRanks 64 nNodes 8 localRanks 8 localRank 4 MNNVL 0 +n116-040-213:1385273:1385338 [5] NCCL INFO Tree 4 : 44 -> 45 -> 46/-1/-1 +n116-040-213:1385275:1385341 [7] NCCL INFO Tree 6 : 46 -> 47 -> 40/-1/-1 +n116-040-213:1385273:1385338 [5] NCCL INFO Tree 12 : 44 -> 45 -> 46/52/-1 +n116-040-213:1385273:1385338 [5] NCCL INFO Tree 5 : 52 -> 45 -> 44/-1/-1 +n116-040-213:1385275:1385341 [7] NCCL INFO Tree 14 : 46 -> 47 -> 40/54/-1 +n116-040-213:1385273:1385338 [5] NCCL INFO Tree 13 : 28 -> 45 -> 44/37/-1 +n116-040-213:1385274:1385336 [6] NCCL INFO Tree 6 : 55 -> 46 -> 47/-1/-1 +n116-040-213:1385275:1385341 [7] NCCL INFO Tree 7 : 54 -> 47 -> 46/-1/-1 +n116-040-213:1385275:1385341 [7] NCCL INFO Tree 15 : 30 -> 47 -> 46/39/-1 +n116-040-213:1385274:1385336 [6] NCCL INFO Tree 14 : 31 -> 46 -> 47/38/-1 +n116-040-213:1385274:1385336 [6] NCCL INFO Tree 7 : 47 -> 46 -> 45/-1/-1 +n116-040-213:1385273:1385338 [5] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-040-213:1385274:1385336 [6] NCCL INFO Tree 15 : 47 -> 46 -> 45/55/-1 +n116-040-213:1385275:1385341 [7] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-040-213:1385274:1385336 [6] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-040-213:1385271:1385337 [3] NCCL INFO comm 0xc1512d0 rank 43 nRanks 64 nNodes 8 localRanks 8 localRank 3 MNNVL 0 +n116-040-213:1385272:1385339 [4] NCCL INFO Tree 4 : 53 -> 44 -> 45/-1/-1 +n116-040-213:1385272:1385339 [4] NCCL INFO Tree 12 : 29 -> 44 -> 45/36/-1 +n116-040-213:1385272:1385339 [4] NCCL INFO Tree 5 : 45 -> 44 -> 43/-1/-1 +n116-040-213:1385272:1385339 [4] NCCL INFO Tree 13 : 45 -> 44 -> 43/53/-1 +n116-040-213:1385272:1385339 [4] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-040-213:1385273:1385338 [5] NCCL INFO Ring 00 : 46 -> 45 -> 44 +n116-040-213:1385273:1385338 [5] NCCL INFO Ring 01 : 44 -> 45 -> 46 +n116-040-213:1385273:1385338 [5] NCCL INFO Ring 02 : 46 -> 45 -> 44 +n116-040-213:1385273:1385338 [5] NCCL INFO Ring 03 : 44 -> 45 -> 46 +n116-040-213:1385273:1385338 [5] NCCL INFO Ring 04 : 46 -> 45 -> 52 +n116-040-213:1385273:1385338 [5] NCCL INFO Ring 05 : 36 -> 45 -> 46 +n116-040-213:1385273:1385338 [5] NCCL INFO Ring 06 : 46 -> 45 -> 44 +n116-040-213:1385275:1385341 [7] NCCL INFO Ring 00 : 40 -> 47 -> 46 +n116-040-213:1385273:1385338 [5] NCCL INFO Ring 07 : 44 -> 45 -> 46 +n116-040-213:1385274:1385336 [6] NCCL INFO Ring 00 : 47 -> 46 -> 45 +n116-040-213:1385272:1385339 [4] NCCL INFO Ring 00 : 45 -> 44 -> 43 +n116-040-213:1385271:1385337 [3] NCCL INFO Tree 2 : 42 -> 43 -> 44/-1/-1 +n116-040-213:1385273:1385338 [5] NCCL INFO Ring 08 : 46 -> 45 -> 44 +n116-040-213:1385272:1385339 [4] NCCL INFO Ring 01 : 43 -> 44 -> 45 +n116-040-213:1385272:1385339 [4] NCCL INFO Ring 02 : 45 -> 44 -> 43 +n116-040-213:1385275:1385341 [7] NCCL INFO Ring 01 : 46 -> 47 -> 40 +n116-040-213:1385274:1385336 [6] NCCL INFO Ring 01 : 45 -> 46 -> 47 +n116-040-213:1385271:1385337 [3] NCCL INFO Tree 10 : 42 -> 43 -> 44/50/-1 +n116-040-213:1385270:1385343 [2] NCCL INFO comm 0xc5bfe00 rank 42 nRanks 64 nNodes 8 localRanks 8 localRank 2 MNNVL 0 +n116-040-213:1385273:1385338 [5] NCCL INFO Ring 09 : 44 -> 45 -> 46 +n116-040-213:1385272:1385339 [4] NCCL INFO Ring 03 : 43 -> 44 -> 45 +n116-040-213:1385275:1385341 [7] NCCL INFO Ring 02 : 40 -> 47 -> 46 +n116-040-213:1385274:1385336 [6] NCCL INFO Ring 02 : 47 -> 46 -> 45 +n116-040-213:1385271:1385337 [3] NCCL INFO Tree 3 : 50 -> 43 -> 42/-1/-1 +n116-040-213:1385273:1385338 [5] NCCL INFO Ring 10 : 46 -> 45 -> 44 +n116-040-213:1385272:1385339 [4] NCCL INFO Ring 04 : 37 -> 44 -> 43 +n116-040-213:1385275:1385341 [7] NCCL INFO Ring 03 : 46 -> 47 -> 40 +n116-040-213:1385268:1385340 [0] NCCL INFO comm 0xcdf2640 rank 40 nRanks 64 nNodes 8 localRanks 8 localRank 0 MNNVL 0 +n116-040-213:1385272:1385339 [4] NCCL INFO Ring 05 : 43 -> 44 -> 53 +n116-040-213:1385274:1385336 [6] NCCL INFO Ring 03 : 45 -> 46 -> 47 +n116-040-213:1385274:1385336 [6] NCCL INFO Ring 04 : 47 -> 46 -> 45 +n116-040-213:1385273:1385338 [5] NCCL INFO Ring 11 : 44 -> 45 -> 46 +n116-040-213:1385275:1385341 [7] NCCL INFO Ring 04 : 40 -> 47 -> 46 +n116-040-213:1385272:1385339 [4] NCCL INFO Ring 06 : 45 -> 44 -> 43 +n116-040-213:1385274:1385336 [6] NCCL INFO Ring 05 : 45 -> 46 -> 47 +n116-040-213:1385271:1385337 [3] NCCL INFO Tree 11 : 26 -> 43 -> 42/35/-1 +n116-040-213:1385273:1385338 [5] NCCL INFO Ring 12 : 46 -> 45 -> 52 +n116-040-213:1385275:1385341 [7] NCCL INFO Ring 05 : 46 -> 47 -> 40 +n116-040-213:1385272:1385339 [4] NCCL INFO Ring 07 : 43 -> 44 -> 45 +n116-040-213:1385274:1385336 [6] NCCL INFO Ring 06 : 39 -> 46 -> 45 +n116-040-213:1385269:1385342 [1] NCCL INFO comm 0xcb0efc0 rank 41 nRanks 64 nNodes 8 localRanks 8 localRank 1 MNNVL 0 +n116-040-213:1385273:1385338 [5] NCCL INFO Ring 13 : 36 -> 45 -> 46 +n116-040-213:1385275:1385341 [7] NCCL INFO Ring 06 : 40 -> 47 -> 54 +n116-040-213:1385272:1385339 [4] NCCL INFO Ring 08 : 45 -> 44 -> 43 +n116-040-213:1385274:1385336 [6] NCCL INFO Ring 07 : 45 -> 46 -> 55 +n116-040-213:1385271:1385337 [3] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-040-213:1385273:1385338 [5] NCCL INFO Ring 14 : 46 -> 45 -> 44 +n116-040-213:1385274:1385336 [6] NCCL INFO Ring 08 : 47 -> 46 -> 45 +n116-040-213:1385275:1385341 [7] NCCL INFO Ring 07 : 38 -> 47 -> 40 +n116-040-213:1385272:1385339 [4] NCCL INFO Ring 09 : 43 -> 44 -> 45 +n116-040-213:1385273:1385338 [5] NCCL INFO Ring 15 : 44 -> 45 -> 46 +n116-040-213:1385274:1385336 [6] NCCL INFO Ring 09 : 45 -> 46 -> 47 +n116-040-213:1385272:1385339 [4] NCCL INFO Ring 10 : 45 -> 44 -> 43 +n116-040-213:1385275:1385341 [7] NCCL INFO Ring 08 : 40 -> 47 -> 46 +n116-040-213:1385273:1385338 [5] NCCL INFO Trees [0] 46/-1/-1->45->44 [1] 44/-1/-1->45->46 [2] 46/-1/-1->45->44 [3] 44/-1/-1->45->46 [4] 46/-1/-1->45->44 [5] 44/-1/-1->45->52 [6] -1/-1/-1->45->44 [7] 44/-1/-1->45->46 [8] 46/-1/-1->45->44 [9] 44/-1/-1->45->46 [10] 46/-1/-1->45->44 [11] 44/-1/-1->45->46 [12] 46/52/-1->45->44 [13] 44/37/-1->45->28 [14] -1/-1/-1->45->44 [15] 44/-1/-1->45->46 +n116-040-213:1385272:1385339 [4] NCCL INFO Ring 11 : 43 -> 44 -> 45 +n116-040-213:1385274:1385336 [6] NCCL INFO Ring 10 : 47 -> 46 -> 45 +n116-040-213:1385275:1385341 [7] NCCL INFO Ring 09 : 46 -> 47 -> 40 +n116-040-213:1385272:1385339 [4] NCCL INFO Ring 12 : 37 -> 44 -> 43 +n116-040-213:1385274:1385336 [6] NCCL INFO Ring 11 : 45 -> 46 -> 47 +n116-040-213:1385272:1385339 [4] NCCL INFO Ring 13 : 43 -> 44 -> 53 +n116-040-213:1385275:1385341 [7] NCCL INFO Ring 10 : 40 -> 47 -> 46 +n116-040-213:1385268:1385340 [0] NCCL INFO Tree 0 : 49 -> 40 -> 41/-1/-1 +n116-040-213:1385272:1385339 [4] NCCL INFO Ring 14 : 45 -> 44 -> 43 +n116-040-213:1385273:1385338 [5] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-040-213:1385274:1385336 [6] NCCL INFO Ring 12 : 47 -> 46 -> 45 +n116-040-213:1385273:1385338 [5] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-040-213:1385275:1385341 [7] NCCL INFO Ring 11 : 46 -> 47 -> 40 +n116-040-213:1385270:1385343 [2] NCCL INFO Tree 2 : 51 -> 42 -> 43/-1/-1 +n116-040-213:1385272:1385339 [4] NCCL INFO Ring 15 : 43 -> 44 -> 45 +n116-040-213:1385275:1385341 [7] NCCL INFO Ring 12 : 40 -> 47 -> 46 +n116-040-213:1385268:1385340 [0] NCCL INFO Tree 8 : 25 -> 40 -> 41/32/-1 +n116-040-213:1385270:1385343 [2] NCCL INFO Tree 10 : 27 -> 42 -> 43/34/-1 +n116-040-213:1385274:1385336 [6] NCCL INFO Ring 13 : 45 -> 46 -> 47 +n116-040-213:1385275:1385341 [7] NCCL INFO Ring 13 : 46 -> 47 -> 40 +n116-040-213:1385273:1385338 [5] NCCL INFO P2P Chunksize set to 524288 +n116-040-213:1385270:1385343 [2] NCCL INFO Tree 3 : 43 -> 42 -> 41/-1/-1 +n116-040-213:1385272:1385339 [4] NCCL INFO Trees [0] 45/-1/-1->44->43 [1] 43/-1/-1->44->45 [2] 45/-1/-1->44->43 [3] -1/-1/-1->44->45 [4] 45/-1/-1->44->53 [5] 43/-1/-1->44->45 [6] 45/-1/-1->44->43 [7] 43/-1/-1->44->45 [8] 45/-1/-1->44->43 [9] 43/-1/-1->44->45 [10] 45/-1/-1->44->43 [11] -1/-1/-1->44->45 [12] 45/36/-1->44->29 [13] 43/53/-1->44->45 [14] 45/-1/-1->44->43 [15] 43/-1/-1->44->45 +n116-040-213:1385268:1385340 [0] NCCL INFO Tree 1 : 41 -> 40 -> 47/-1/-1 +n116-040-213:1385274:1385336 [6] NCCL INFO Ring 14 : 39 -> 46 -> 45 +n116-040-213:1385275:1385341 [7] NCCL INFO Ring 14 : 40 -> 47 -> 54 +n116-040-213:1385270:1385343 [2] NCCL INFO Tree 11 : 43 -> 42 -> 41/51/-1 +n116-040-213:1385268:1385340 [0] NCCL INFO Tree 9 : 41 -> 40 -> 47/49/-1 +n116-040-213:1385272:1385339 [4] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-040-213:1385274:1385336 [6] NCCL INFO Ring 15 : 45 -> 46 -> 55 +n116-040-213:1385275:1385341 [7] NCCL INFO Ring 15 : 38 -> 47 -> 40 +n116-040-213:1385269:1385342 [1] NCCL INFO Tree 0 : 40 -> 41 -> 42/-1/-1 +n116-040-213:1385272:1385339 [4] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-040-213:1385275:1385341 [7] NCCL INFO Trees [0] -1/-1/-1->47->46 [1] 46/-1/-1->47->40 [2] 40/-1/-1->47->46 [3] 46/-1/-1->47->40 [4] 40/-1/-1->47->46 [5] 46/-1/-1->47->40 [6] 40/-1/-1->47->46 [7] 46/-1/-1->47->54 [8] -1/-1/-1->47->46 [9] 46/-1/-1->47->40 [10] 40/-1/-1->47->46 [11] 46/-1/-1->47->40 [12] 40/-1/-1->47->46 [13] 46/-1/-1->47->40 [14] 40/54/-1->47->46 [15] 46/39/-1->47->30 +n116-040-213:1385274:1385336 [6] NCCL INFO Trees [0] 47/-1/-1->46->45 [1] 45/-1/-1->46->47 [2] 47/-1/-1->46->45 [3] 45/-1/-1->46->47 [4] 47/-1/-1->46->45 [5] -1/-1/-1->46->47 [6] 47/-1/-1->46->55 [7] 45/-1/-1->46->47 [8] 47/-1/-1->46->45 [9] 45/-1/-1->46->47 [10] 47/-1/-1->46->45 [11] 45/-1/-1->46->47 [12] 47/-1/-1->46->45 [13] -1/-1/-1->46->47 [14] 47/38/-1->46->31 [15] 45/55/-1->46->47 +n116-040-213:1385272:1385339 [4] NCCL INFO P2P Chunksize set to 524288 +n116-040-213:1385268:1385340 [0] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-040-213:1385269:1385342 [1] NCCL INFO Tree 8 : 40 -> 41 -> 42/48/-1 +n116-040-213:1385274:1385336 [6] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-040-213:1385275:1385341 [7] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-040-213:1385269:1385342 [1] NCCL INFO Tree 1 : 48 -> 41 -> 40/-1/-1 +n116-040-213:1385274:1385336 [6] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-040-213:1385275:1385341 [7] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-040-213:1385271:1385337 [3] NCCL INFO Ring 00 : 44 -> 43 -> 42 +n116-040-213:1385274:1385336 [6] NCCL INFO P2P Chunksize set to 524288 +n116-040-213:1385269:1385342 [1] NCCL INFO Tree 9 : 24 -> 41 -> 40/33/-1 +n116-040-213:1385271:1385337 [3] NCCL INFO Ring 01 : 42 -> 43 -> 44 +n116-040-213:1385270:1385343 [2] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-040-213:1385275:1385341 [7] NCCL INFO P2P Chunksize set to 524288 +n116-040-213:1385271:1385337 [3] NCCL INFO Ring 02 : 44 -> 43 -> 50 +n116-040-213:1385271:1385337 [3] NCCL INFO Ring 03 : 34 -> 43 -> 44 +n116-040-213:1385269:1385342 [1] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-040-213:1385271:1385337 [3] NCCL INFO Ring 04 : 44 -> 43 -> 42 +n116-040-213:1385271:1385337 [3] NCCL INFO Ring 05 : 42 -> 43 -> 44 +n116-040-213:1385268:1385340 [0] NCCL INFO Ring 00 : 33 -> 40 -> 47 +n116-040-213:1385271:1385337 [3] NCCL INFO Ring 06 : 44 -> 43 -> 42 +n116-040-213:1385268:1385340 [0] NCCL INFO Ring 01 : 47 -> 40 -> 49 +n116-040-213:1385271:1385337 [3] NCCL INFO Ring 07 : 42 -> 43 -> 44 +n116-040-213:1385268:1385340 [0] NCCL INFO Ring 02 : 41 -> 40 -> 47 +n116-040-213:1385271:1385337 [3] NCCL INFO Ring 08 : 44 -> 43 -> 42 +n116-040-213:1385271:1385337 [3] NCCL INFO Ring 09 : 42 -> 43 -> 44 +n116-040-213:1385268:1385340 [0] NCCL INFO Ring 03 : 47 -> 40 -> 41 +n116-040-213:1385268:1385340 [0] NCCL INFO Ring 04 : 41 -> 40 -> 47 +n116-040-213:1385271:1385337 [3] NCCL INFO Ring 10 : 44 -> 43 -> 50 +n116-040-213:1385268:1385340 [0] NCCL INFO Ring 05 : 47 -> 40 -> 41 +n116-040-213:1385271:1385337 [3] NCCL INFO Ring 11 : 34 -> 43 -> 44 +n116-040-213:1385271:1385337 [3] NCCL INFO Ring 12 : 44 -> 43 -> 42 +n116-040-213:1385268:1385340 [0] NCCL INFO Ring 06 : 41 -> 40 -> 47 +n116-040-213:1385268:1385340 [0] NCCL INFO Ring 07 : 47 -> 40 -> 41 +n116-040-213:1385271:1385337 [3] NCCL INFO Ring 13 : 42 -> 43 -> 44 +n116-040-213:1385268:1385340 [0] NCCL INFO Ring 08 : 33 -> 40 -> 47 +n116-040-213:1385271:1385337 [3] NCCL INFO Ring 14 : 44 -> 43 -> 42 +n116-040-213:1385268:1385340 [0] NCCL INFO Ring 09 : 47 -> 40 -> 49 +n116-040-213:1385268:1385340 [0] NCCL INFO Ring 10 : 41 -> 40 -> 47 +n116-040-213:1385271:1385337 [3] NCCL INFO Ring 15 : 42 -> 43 -> 44 +n116-040-213:1385270:1385343 [2] NCCL INFO Ring 00 : 43 -> 42 -> 41 +n116-040-213:1385268:1385340 [0] NCCL INFO Ring 11 : 47 -> 40 -> 41 +n116-040-213:1385271:1385337 [3] NCCL INFO Trees [0] 44/-1/-1->43->42 [1] 42/-1/-1->43->44 [2] 44/-1/-1->43->42 [3] 42/-1/-1->43->50 [4] -1/-1/-1->43->42 [5] 42/-1/-1->43->44 [6] 44/-1/-1->43->42 [7] 42/-1/-1->43->44 [8] 44/-1/-1->43->42 [9] 42/-1/-1->43->44 [10] 44/50/-1->43->42 [11] 42/35/-1->43->26 [12] -1/-1/-1->43->42 [13] 42/-1/-1->43->44 [14] 44/-1/-1->43->42 [15] 42/-1/-1->43->44 +n116-040-213:1385270:1385343 [2] NCCL INFO Ring 01 : 41 -> 42 -> 43 +n116-040-213:1385268:1385340 [0] NCCL INFO Ring 12 : 41 -> 40 -> 47 +n116-040-213:1385269:1385342 [1] NCCL INFO Ring 00 : 42 -> 41 -> 48 +n116-040-213:1385271:1385337 [3] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-040-213:1385268:1385340 [0] NCCL INFO Ring 13 : 47 -> 40 -> 41 +n116-040-213:1385269:1385342 [1] NCCL INFO Ring 01 : 32 -> 41 -> 42 +n116-040-213:1385270:1385343 [2] NCCL INFO Ring 02 : 35 -> 42 -> 41 +n116-040-213:1385268:1385340 [0] NCCL INFO Ring 14 : 41 -> 40 -> 47 +n116-040-213:1385271:1385337 [3] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-040-213:1385269:1385342 [1] NCCL INFO Ring 02 : 42 -> 41 -> 40 +n116-040-213:1385270:1385343 [2] NCCL INFO Ring 03 : 41 -> 42 -> 51 +n116-040-213:1385268:1385340 [0] NCCL INFO Ring 15 : 47 -> 40 -> 41 +n116-040-213:1385268:1385340 [0] NCCL INFO Trees [0] 41/-1/-1->40->49 [1] 47/-1/-1->40->41 [2] 41/-1/-1->40->47 [3] 47/-1/-1->40->41 [4] 41/-1/-1->40->47 [5] 47/-1/-1->40->41 [6] 41/-1/-1->40->47 [7] -1/-1/-1->40->41 [8] 41/32/-1->40->25 [9] 47/49/-1->40->41 [10] 41/-1/-1->40->47 [11] 47/-1/-1->40->41 [12] 41/-1/-1->40->47 [13] 47/-1/-1->40->41 [14] 41/-1/-1->40->47 [15] -1/-1/-1->40->41 +n116-040-213:1385271:1385337 [3] NCCL INFO P2P Chunksize set to 524288 +n116-040-213:1385270:1385343 [2] NCCL INFO Ring 04 : 43 -> 42 -> 41 +n116-040-213:1385269:1385342 [1] NCCL INFO Ring 03 : 40 -> 41 -> 42 +n116-040-213:1385270:1385343 [2] NCCL INFO Ring 05 : 41 -> 42 -> 43 +n116-040-213:1385269:1385342 [1] NCCL INFO Ring 04 : 42 -> 41 -> 40 +n116-040-213:1385268:1385340 [0] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-040-213:1385270:1385343 [2] NCCL INFO Ring 06 : 43 -> 42 -> 41 +n116-040-213:1385269:1385342 [1] NCCL INFO Ring 05 : 40 -> 41 -> 42 +n116-040-213:1385268:1385340 [0] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-040-213:1385268:1385340 [0] NCCL INFO P2P Chunksize set to 524288 +n116-040-213:1385270:1385343 [2] NCCL INFO Ring 07 : 41 -> 42 -> 43 +n116-040-213:1385269:1385342 [1] NCCL INFO Ring 06 : 42 -> 41 -> 40 +n116-040-213:1385270:1385343 [2] NCCL INFO Ring 08 : 43 -> 42 -> 41 +n116-040-213:1385269:1385342 [1] NCCL INFO Ring 07 : 40 -> 41 -> 42 +n116-040-213:1385270:1385343 [2] NCCL INFO Ring 09 : 41 -> 42 -> 43 +n116-040-213:1385269:1385342 [1] NCCL INFO Ring 08 : 42 -> 41 -> 48 +n116-040-213:1385270:1385343 [2] NCCL INFO Ring 10 : 35 -> 42 -> 41 +n116-040-213:1385269:1385342 [1] NCCL INFO Ring 09 : 32 -> 41 -> 42 +n116-040-213:1385270:1385343 [2] NCCL INFO Ring 11 : 41 -> 42 -> 51 +n116-040-213:1385269:1385342 [1] NCCL INFO Ring 10 : 42 -> 41 -> 40 +n116-040-213:1385270:1385343 [2] NCCL INFO Ring 12 : 43 -> 42 -> 41 +n116-040-213:1385269:1385342 [1] NCCL INFO Ring 11 : 40 -> 41 -> 42 +n116-040-213:1385270:1385343 [2] NCCL INFO Ring 13 : 41 -> 42 -> 43 +n116-040-213:1385269:1385342 [1] NCCL INFO Ring 12 : 42 -> 41 -> 40 +n116-040-213:1385270:1385343 [2] NCCL INFO Ring 14 : 43 -> 42 -> 41 +n116-040-213:1385269:1385342 [1] NCCL INFO Ring 13 : 40 -> 41 -> 42 +n116-040-213:1385270:1385343 [2] NCCL INFO Ring 15 : 41 -> 42 -> 43 +n116-040-213:1385269:1385342 [1] NCCL INFO Ring 14 : 42 -> 41 -> 40 +n116-040-213:1385270:1385343 [2] NCCL INFO Trees [0] 43/-1/-1->42->41 [1] -1/-1/-1->42->43 [2] 43/-1/-1->42->51 [3] 41/-1/-1->42->43 [4] 43/-1/-1->42->41 [5] 41/-1/-1->42->43 [6] 43/-1/-1->42->41 [7] 41/-1/-1->42->43 [8] 43/-1/-1->42->41 [9] -1/-1/-1->42->43 [10] 43/34/-1->42->27 [11] 41/51/-1->42->43 [12] 43/-1/-1->42->41 [13] 41/-1/-1->42->43 [14] 43/-1/-1->42->41 [15] 41/-1/-1->42->43 +n116-040-213:1385269:1385342 [1] NCCL INFO Ring 15 : 40 -> 41 -> 42 +n116-040-213:1385269:1385342 [1] NCCL INFO Trees [0] 42/-1/-1->41->40 [1] 40/-1/-1->41->48 [2] -1/-1/-1->41->40 [3] 40/-1/-1->41->42 [4] 42/-1/-1->41->40 [5] 40/-1/-1->41->42 [6] 42/-1/-1->41->40 [7] 40/-1/-1->41->42 [8] 42/48/-1->41->40 [9] 40/33/-1->41->24 [10] -1/-1/-1->41->40 [11] 40/-1/-1->41->42 [12] 42/-1/-1->41->40 [13] 40/-1/-1->41->42 [14] 42/-1/-1->41->40 [15] 40/-1/-1->41->42 +n116-040-213:1385270:1385343 [2] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-040-213:1385269:1385342 [1] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-040-213:1385270:1385343 [2] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-040-213:1385269:1385342 [1] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-040-213:1385270:1385343 [2] NCCL INFO P2P Chunksize set to 524288 +n116-040-213:1385269:1385342 [1] NCCL INFO P2P Chunksize set to 524288 +n116-040-213:1385273:1385378 [5] NCCL INFO [Proxy Service UDS] Device 5 CPU core 196 +n116-040-213:1385272:1385377 [4] NCCL INFO [Proxy Service] Device 4 CPU core 78 +n116-040-213:1385272:1385379 [4] NCCL INFO [Proxy Service UDS] Device 4 CPU core 197 +n116-040-213:1385273:1385376 [5] NCCL INFO [Proxy Service] Device 5 CPU core 162 +n116-040-213:1385268:1385382 [0] NCCL INFO [Proxy Service] Device 0 CPU core 137 +n116-040-213:1385274:1385380 [6] NCCL INFO [Proxy Service] Device 6 CPU core 86 +n116-040-213:1385275:1385383 [7] NCCL INFO [Proxy Service UDS] Device 7 CPU core 90 +n116-040-213:1385271:1385386 [3] NCCL INFO [Proxy Service] Device 3 CPU core 130 +n116-040-213:1385275:1385381 [7] NCCL INFO [Proxy Service] Device 7 CPU core 85 +n116-040-213:1385269:1385391 [1] NCCL INFO [Proxy Service UDS] Device 1 CPU core 118 +n116-040-213:1385274:1385384 [6] NCCL INFO [Proxy Service UDS] Device 6 CPU core 91 +n116-040-213:1385268:1385385 [0] NCCL INFO [Proxy Service UDS] Device 0 CPU core 35 +n116-040-213:1385270:1385390 [2] NCCL INFO [Proxy Service UDS] Device 2 CPU core 118 +n116-040-213:1385271:1385387 [3] NCCL INFO [Proxy Service UDS] Device 3 CPU core 38 +n116-040-213:1385269:1385389 [1] NCCL INFO [Proxy Service] Device 1 CPU core 145 +n116-040-213:1385270:1385388 [2] NCCL INFO [Proxy Service] Device 2 CPU core 145 +n116-040-213:1385275:1385341 [7] NCCL INFO NCCL_PROTO set by environment to Simple +n116-040-213:1385275:1385341 [7] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-040-213:1385275:1385341 [7] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n116-040-213:1385275:1385341 [7] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n116-040-213:1385273:1385338 [5] NCCL INFO NCCL_PROTO set by environment to Simple +n116-040-213:1385273:1385338 [5] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-040-213:1385273:1385338 [5] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n116-040-213:1385273:1385338 [5] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n116-040-213:1385274:1385336 [6] NCCL INFO NCCL_PROTO set by environment to Simple +n116-040-213:1385274:1385336 [6] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-040-213:1385274:1385336 [6] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n116-040-213:1385274:1385336 [6] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n116-040-213:1385272:1385339 [4] NCCL INFO NCCL_PROTO set by environment to Simple +n116-040-213:1385272:1385339 [4] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-040-213:1385272:1385339 [4] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n116-040-213:1385272:1385339 [4] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n116-040-213:1385270:1385343 [2] NCCL INFO NCCL_PROTO set by environment to Simple +n116-040-213:1385270:1385343 [2] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-040-213:1385270:1385343 [2] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n116-040-213:1385270:1385343 [2] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n116-040-213:1385271:1385337 [3] NCCL INFO NCCL_PROTO set by environment to Simple +n116-040-213:1385271:1385337 [3] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-040-213:1385271:1385337 [3] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n116-040-213:1385271:1385337 [3] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n116-040-213:1385268:1385340 [0] NCCL INFO NCCL_PROTO set by environment to Simple +n116-040-213:1385268:1385340 [0] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-040-213:1385268:1385340 [0] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n116-040-213:1385268:1385340 [0] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n116-040-213:1385269:1385342 [1] NCCL INFO NCCL_PROTO set by environment to Simple +n116-040-213:1385269:1385342 [1] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-040-213:1385269:1385342 [1] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n116-040-213:1385269:1385342 [1] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n116-040-213:1385271:1385337 [3] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-040-213:1385271:1385337 [3] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-040-213:1385271:1385337 [3] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-040-213:1385271:1385337 [3] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-040-213:1385271:1385337 [3] NCCL INFO ncclCommInitRankConfig comm 0xc1512d0 rank 43 nranks 64 cudaDev 3 nvmlDev 3 busId c000 commId 0xfb2d11625d05007a - Init COMPLETE +n116-040-213:1385271:1385337 [3] NCCL INFO Init timings - ncclCommInitRankConfig: rank 43 nranks 64 total 5.03 (kernels 0.43, alloc 1.76, bootstrap 1.94, allgathers 0.11, topo 0.66, graphs 0.08, connections 0.05, rest 0.00) +n116-040-213:1385272:1385339 [4] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-040-213:1385272:1385339 [4] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-040-213:1385272:1385339 [4] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-040-213:1385272:1385339 [4] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-040-213:1385272:1385339 [4] NCCL INFO ncclCommInitRankConfig comm 0xd0e41d0 rank 44 nranks 64 cudaDev 4 nvmlDev 4 busId 84000 commId 0xfb2d11625d05007a - Init COMPLETE +n116-040-213:1385272:1385339 [4] NCCL INFO Init timings - ncclCommInitRankConfig: rank 44 nranks 64 total 4.99 (kernels 0.46, alloc 1.79, bootstrap 1.85, allgathers 0.11, topo 0.65, graphs 0.08, connections 0.05, rest 0.01) +n116-040-213:1385268:1385340 [0] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-040-213:1385268:1385340 [0] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-040-213:1385268:1385340 [0] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-040-213:1385268:1385340 [0] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-040-213:1385268:1385340 [0] NCCL INFO ncclCommInitRankConfig comm 0xcdf2640 rank 40 nranks 64 cudaDev 0 nvmlDev 0 busId 4000 commId 0xfb2d11625d05007a - Init COMPLETE +n116-040-213:1385268:1385340 [0] NCCL INFO Init timings - ncclCommInitRankConfig: rank 40 nranks 64 total 4.97 (kernels 0.42, alloc 1.79, bootstrap 1.87, allgathers 0.11, topo 0.65, graphs 0.08, connections 0.05, rest 0.00) +n116-040-213:1385273:1385338 [5] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-040-213:1385273:1385338 [5] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-040-213:1385273:1385338 [5] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-040-213:1385273:1385338 [5] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-040-213:1385273:1385338 [5] NCCL INFO ncclCommInitRankConfig comm 0xbc479c0 rank 45 nranks 64 cudaDev 5 nvmlDev 5 busId 85000 commId 0xfb2d11625d05007a - Init COMPLETE +n116-040-213:1385273:1385338 [5] NCCL INFO Init timings - ncclCommInitRankConfig: rank 45 nranks 64 total 5.01 (kernels 0.44, alloc 1.78, bootstrap 1.89, allgathers 0.11, topo 0.65, graphs 0.08, connections 0.05, rest 0.01) +n116-040-213:1385275:1385341 [7] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-040-213:1385275:1385341 [7] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-040-213:1385275:1385341 [7] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-040-213:1385275:1385341 [7] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-040-213:1385275:1385341 [7] NCCL INFO ncclCommInitRankConfig comm 0xb2afe50 rank 47 nranks 64 cudaDev 7 nvmlDev 7 busId 8c000 commId 0xfb2d11625d05007a - Init COMPLETE +n116-040-213:1385275:1385341 [7] NCCL INFO Init timings - ncclCommInitRankConfig: rank 47 nranks 64 total 4.97 (kernels 0.45, alloc 1.77, bootstrap 1.85, allgathers 0.05, topo 0.65, graphs 0.13, connections 0.04, rest 0.01) +n116-040-213:1385270:1385343 [2] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-040-213:1385270:1385343 [2] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-040-213:1385270:1385343 [2] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-040-213:1385270:1385343 [2] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-040-213:1385269:1385342 [1] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-040-213:1385270:1385343 [2] NCCL INFO ncclCommInitRankConfig comm 0xc5bfe00 rank 42 nranks 64 cudaDev 2 nvmlDev 2 busId b000 commId 0xfb2d11625d05007a - Init COMPLETE +n116-040-213:1385269:1385342 [1] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-040-213:1385269:1385342 [1] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-040-213:1385269:1385342 [1] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-040-213:1385270:1385343 [2] NCCL INFO Init timings - ncclCommInitRankConfig: rank 42 nranks 64 total 4.95 (kernels 0.48, alloc 1.73, bootstrap 1.84, allgathers 0.06, topo 0.65, graphs 0.13, connections 0.05, rest 0.00) +n116-040-213:1385269:1385342 [1] NCCL INFO ncclCommInitRankConfig comm 0xcb0efc0 rank 41 nranks 64 cudaDev 1 nvmlDev 1 busId 5000 commId 0xfb2d11625d05007a - Init COMPLETE +n116-040-213:1385269:1385342 [1] NCCL INFO Init timings - ncclCommInitRankConfig: rank 41 nranks 64 total 4.96 (kernels 0.45, alloc 1.76, bootstrap 1.85, allgathers 0.06, topo 0.65, graphs 0.13, connections 0.05, rest 0.00) +n116-040-213:1385274:1385336 [6] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-040-213:1385274:1385336 [6] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-040-213:1385274:1385336 [6] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-040-213:1385274:1385336 [6] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-040-213:1385274:1385336 [6] NCCL INFO ncclCommInitRankConfig comm 0xc6f23a0 rank 46 nranks 64 cudaDev 6 nvmlDev 6 busId 8b000 commId 0xfb2d11625d05007a - Init COMPLETE +n116-040-213:1385274:1385336 [6] NCCL INFO Init timings - ncclCommInitRankConfig: rank 46 nranks 64 total 5.05 (kernels 0.43, alloc 1.78, bootstrap 1.94, allgathers 0.05, topo 0.65, graphs 0.14, connections 0.05, rest 0.01) +n116-040-213:1385274:1385394 [6] NCCL INFO Channel 00/0 : 46[6] -> 47[7] via P2P/CUMEM +n116-040-213:1385274:1385394 [6] NCCL INFO Channel 01/0 : 46[6] -> 47[7] via P2P/CUMEM +n116-040-213:1385271:1385392 [3] NCCL INFO Channel 00/0 : 43[3] -> 44[4] via P2P/CUMEM +n116-040-213:1385273:1385393 [5] NCCL INFO Channel 00/0 : 45[5] -> 46[6] via P2P/CUMEM +n116-040-213:1385274:1385394 [6] NCCL INFO Channel 02/0 : 46[6] -> 47[7] via P2P/CUMEM +n116-040-213:1385271:1385392 [3] NCCL INFO Channel 01/0 : 43[3] -> 44[4] via P2P/CUMEM +n116-040-213:1385273:1385393 [5] NCCL INFO Channel 01/0 : 45[5] -> 46[6] via P2P/CUMEM +n116-040-213:1385274:1385394 [6] NCCL INFO Channel 03/0 : 46[6] -> 47[7] via P2P/CUMEM +n116-040-213:1385268:1385399 [0] NCCL INFO Channel 00/0 : 40[0] -> 41[1] via P2P/CUMEM +n116-040-213:1385271:1385392 [3] NCCL INFO Channel 02/0 : 43[3] -> 44[4] via P2P/CUMEM +n116-040-213:1385273:1385393 [5] NCCL INFO Channel 02/0 : 45[5] -> 46[6] via P2P/CUMEM +n116-040-213:1385272:1385396 [4] NCCL INFO Channel 00/0 : 44[4] -> 45[5] via P2P/CUMEM +n116-040-213:1385274:1385394 [6] NCCL INFO Channel 04/0 : 46[6] -> 47[7] via P2P/CUMEM +n116-040-213:1385271:1385392 [3] NCCL INFO Channel 05/0 : 43[3] -> 44[4] via P2P/CUMEM +n116-040-213:1385273:1385393 [5] NCCL INFO Channel 03/0 : 45[5] -> 46[6] via P2P/CUMEM +n116-040-213:1385268:1385399 [0] NCCL INFO Channel 01/0 : 40[0] -> 41[1] via P2P/CUMEM +n116-040-213:1385272:1385396 [4] NCCL INFO Channel 01/0 : 44[4] -> 45[5] via P2P/CUMEM +n116-040-213:1385274:1385394 [6] NCCL INFO Channel 05/0 : 46[6] -> 47[7] via P2P/CUMEM +n116-040-213:1385273:1385393 [5] NCCL INFO Channel 04/0 : 45[5] -> 46[6] via P2P/CUMEM +n116-040-213:1385269:1385395 [1] NCCL INFO Channel 00/0 : 41[1] -> 42[2] via P2P/CUMEM +n116-040-213:1385268:1385399 [0] NCCL INFO Channel 02/0 : 40[0] -> 41[1] via P2P/CUMEM +n116-040-213:1385272:1385396 [4] NCCL INFO Channel 02/0 : 44[4] -> 45[5] via P2P/CUMEM +n116-040-213:1385273:1385393 [5] NCCL INFO Channel 07/0 : 45[5] -> 46[6] via P2P/CUMEM +n116-040-213:1385269:1385395 [1] NCCL INFO Channel 03/0 : 41[1] -> 42[2] via P2P/CUMEM +n116-040-213:1385268:1385399 [0] NCCL INFO Channel 03/0 : 40[0] -> 41[1] via P2P/CUMEM +n116-040-213:1385272:1385396 [4] NCCL INFO Channel 03/0 : 44[4] -> 45[5] via P2P/CUMEM +n116-040-213:1385273:1385393 [5] NCCL INFO Channel 08/0 : 45[5] -> 46[6] via P2P/CUMEM +n116-040-213:1385269:1385395 [1] NCCL INFO Channel 04/0 : 41[1] -> 42[2] via P2P/CUMEM +n116-040-213:1385268:1385399 [0] NCCL INFO Channel 04/0 : 40[0] -> 41[1] via P2P/CUMEM +n116-040-213:1385272:1385396 [4] NCCL INFO Channel 04/0 : 44[4] -> 45[5] via P2P/CUMEM +n116-040-213:1385273:1385393 [5] NCCL INFO Channel 09/0 : 45[5] -> 46[6] via P2P/CUMEM +n116-040-213:1385269:1385395 [1] NCCL INFO Channel 05/0 : 41[1] -> 42[2] via P2P/CUMEM +n116-040-213:1385268:1385399 [0] NCCL INFO Channel 05/0 : 40[0] -> 41[1] via P2P/CUMEM +n116-040-213:1385272:1385396 [4] NCCL INFO Channel 05/0 : 44[4] -> 45[5] via P2P/CUMEM +n116-040-213:1385273:1385393 [5] NCCL INFO Channel 10/0 : 45[5] -> 46[6] via P2P/CUMEM +n116-040-213:1385269:1385395 [1] NCCL INFO Channel 06/0 : 41[1] -> 42[2] via P2P/CUMEM +n116-040-213:1385268:1385399 [0] NCCL INFO Channel 06/0 : 40[0] -> 41[1] via P2P/CUMEM +n116-040-213:1385272:1385396 [4] NCCL INFO Channel 06/0 : 44[4] -> 45[5] via P2P/CUMEM +n116-040-213:1385273:1385393 [5] NCCL INFO Channel 11/0 : 45[5] -> 46[6] via P2P/CUMEM +n116-040-213:1385269:1385395 [1] NCCL INFO Channel 07/0 : 41[1] -> 42[2] via P2P/CUMEM +n116-040-213:1385268:1385399 [0] NCCL INFO Channel 07/0 : 40[0] -> 41[1] via P2P/CUMEM +n116-040-213:1385272:1385396 [4] NCCL INFO Channel 07/0 : 44[4] -> 45[5] via P2P/CUMEM +n116-040-213:1385273:1385393 [5] NCCL INFO Channel 12/0 : 45[5] -> 46[6] via P2P/CUMEM +n116-040-213:1385269:1385395 [1] NCCL INFO Channel 08/0 : 41[1] -> 42[2] via P2P/CUMEM +n116-040-213:1385268:1385399 [0] NCCL INFO Channel 08/0 : 40[0] -> 41[1] via P2P/CUMEM +n116-040-213:1385272:1385396 [4] NCCL INFO Channel 08/0 : 44[4] -> 45[5] via P2P/CUMEM +n116-040-213:1385273:1385393 [5] NCCL INFO Channel 15/0 : 45[5] -> 46[6] via P2P/CUMEM +n116-040-213:1385269:1385395 [1] NCCL INFO Channel 11/0 : 41[1] -> 42[2] via P2P/CUMEM +n116-040-213:1385268:1385399 [0] NCCL INFO Channel 09/0 : 40[0] -> 41[1] via P2P/CUMEM +n116-040-213:1385272:1385396 [4] NCCL INFO Channel 09/0 : 44[4] -> 45[5] via P2P/CUMEM +n116-040-213:1385269:1385395 [1] NCCL INFO Channel 12/0 : 41[1] -> 42[2] via P2P/CUMEM +n116-040-213:1385268:1385399 [0] NCCL INFO Channel 10/0 : 40[0] -> 41[1] via P2P/CUMEM +n116-040-213:1385271:1385392 [3] NCCL INFO Channel 06/0 : 43[3] -> 44[4] via P2P/CUMEM +n116-040-213:1385272:1385396 [4] NCCL INFO Channel 10/0 : 44[4] -> 45[5] via P2P/CUMEM +n116-040-213:1385269:1385395 [1] NCCL INFO Channel 13/0 : 41[1] -> 42[2] via P2P/CUMEM +n116-040-213:1385274:1385394 [6] NCCL INFO Channel 06/0 : 46[6] -> 47[7] via P2P/CUMEM +n116-040-213:1385268:1385399 [0] NCCL INFO Channel 11/0 : 40[0] -> 41[1] via P2P/CUMEM +n116-040-213:1385271:1385392 [3] NCCL INFO Channel 07/0 : 43[3] -> 44[4] via P2P/CUMEM +n116-040-213:1385272:1385396 [4] NCCL INFO Channel 11/0 : 44[4] -> 45[5] via P2P/CUMEM +n116-040-213:1385269:1385395 [1] NCCL INFO Channel 14/0 : 41[1] -> 42[2] via P2P/CUMEM +n116-040-213:1385274:1385394 [6] NCCL INFO Channel 07/0 : 46[6] -> 47[7] via P2P/CUMEM +n116-040-213:1385268:1385399 [0] NCCL INFO Channel 12/0 : 40[0] -> 41[1] via P2P/CUMEM +n116-040-213:1385271:1385392 [3] NCCL INFO Channel 08/0 : 43[3] -> 44[4] via P2P/CUMEM +n116-040-213:1385272:1385396 [4] NCCL INFO Channel 12/0 : 44[4] -> 45[5] via P2P/CUMEM +n116-040-213:1385269:1385395 [1] NCCL INFO Channel 15/0 : 41[1] -> 42[2] via P2P/CUMEM +n116-040-213:1385274:1385394 [6] NCCL INFO Channel 08/0 : 46[6] -> 47[7] via P2P/CUMEM +n116-040-213:1385268:1385399 [0] NCCL INFO Channel 13/0 : 40[0] -> 41[1] via P2P/CUMEM +n116-040-213:1385271:1385392 [3] NCCL INFO Channel 09/0 : 43[3] -> 44[4] via P2P/CUMEM +n116-040-213:1385272:1385396 [4] NCCL INFO Channel 13/0 : 44[4] -> 45[5] via P2P/CUMEM +n116-040-213:1385274:1385394 [6] NCCL INFO Channel 09/0 : 46[6] -> 47[7] via P2P/CUMEM +n116-040-213:1385271:1385392 [3] NCCL INFO Channel 10/0 : 43[3] -> 44[4] via P2P/CUMEM +n116-040-213:1385268:1385399 [0] NCCL INFO Channel 14/0 : 40[0] -> 41[1] via P2P/CUMEM +n116-040-213:1385274:1385394 [6] NCCL INFO Channel 10/0 : 46[6] -> 47[7] via P2P/CUMEM +n116-040-213:1385271:1385392 [3] NCCL INFO Channel 13/0 : 43[3] -> 44[4] via P2P/CUMEM +n116-040-213:1385268:1385399 [0] NCCL INFO Channel 15/0 : 40[0] -> 41[1] via P2P/CUMEM +n116-040-213:1385274:1385394 [6] NCCL INFO Channel 11/0 : 46[6] -> 47[7] via P2P/CUMEM +n116-040-213:1385271:1385392 [3] NCCL INFO Channel 14/0 : 43[3] -> 44[4] via P2P/CUMEM +n116-040-213:1385268:1385399 [0] NCCL INFO Channel 01/0 : 40[0] -> 47[7] via P2P/CUMEM +n116-040-213:1385274:1385394 [6] NCCL INFO Channel 12/0 : 46[6] -> 47[7] via P2P/CUMEM +n116-040-213:1385271:1385392 [3] NCCL INFO Channel 15/0 : 43[3] -> 44[4] via P2P/CUMEM +n116-040-213:1385272:1385396 [4] NCCL INFO Channel 14/0 : 44[4] -> 45[5] via P2P/CUMEM +n116-040-213:1385274:1385394 [6] NCCL INFO Channel 13/0 : 46[6] -> 47[7] via P2P/CUMEM +n116-040-213:1385268:1385399 [0] NCCL INFO Channel 02/0 : 40[0] -> 47[7] via P2P/CUMEM +n116-040-213:1385272:1385396 [4] NCCL INFO Channel 15/0 : 44[4] -> 45[5] via P2P/CUMEM +n116-040-213:1385274:1385394 [6] NCCL INFO Channel 14/0 : 46[6] -> 47[7] via P2P/CUMEM +n116-040-213:1385268:1385399 [0] NCCL INFO Channel 03/0 : 40[0] -> 47[7] via P2P/CUMEM +n116-040-213:1385270:1385398 [2] NCCL INFO Channel 00/0 : 42[2] -> 43[3] via P2P/CUMEM +n116-040-213:1385274:1385394 [6] NCCL INFO Channel 15/0 : 46[6] -> 47[7] via P2P/CUMEM +n116-040-213:1385268:1385399 [0] NCCL INFO Channel 04/0 : 40[0] -> 47[7] via P2P/CUMEM +n116-040-213:1385270:1385398 [2] NCCL INFO Channel 01/0 : 42[2] -> 43[3] via P2P/CUMEM +n116-040-213:1385268:1385399 [0] NCCL INFO Channel 05/0 : 40[0] -> 47[7] via P2P/CUMEM +n116-040-213:1385270:1385398 [2] NCCL INFO Channel 02/0 : 42[2] -> 43[3] via P2P/CUMEM +n116-040-213:1385268:1385399 [0] NCCL INFO Channel 06/0 : 40[0] -> 47[7] via P2P/CUMEM +n116-040-213:1385270:1385398 [2] NCCL INFO Channel 03/0 : 42[2] -> 43[3] via P2P/CUMEM +n116-040-213:1385268:1385399 [0] NCCL INFO Channel 09/0 : 40[0] -> 47[7] via P2P/CUMEM +n116-040-213:1385270:1385398 [2] NCCL INFO Channel 04/0 : 42[2] -> 43[3] via P2P/CUMEM +n116-040-213:1385268:1385399 [0] NCCL INFO Channel 10/0 : 40[0] -> 47[7] via P2P/CUMEM +n116-040-213:1385270:1385398 [2] NCCL INFO Channel 05/0 : 42[2] -> 43[3] via P2P/CUMEM +n116-040-213:1385268:1385399 [0] NCCL INFO Channel 11/0 : 40[0] -> 47[7] via P2P/CUMEM +n116-040-213:1385270:1385398 [2] NCCL INFO Channel 06/0 : 42[2] -> 43[3] via P2P/CUMEM +n116-040-213:1385268:1385399 [0] NCCL INFO Channel 12/0 : 40[0] -> 47[7] via P2P/CUMEM +n116-040-213:1385270:1385398 [2] NCCL INFO Channel 07/0 : 42[2] -> 43[3] via P2P/CUMEM +n116-040-213:1385268:1385399 [0] NCCL INFO Channel 13/0 : 40[0] -> 47[7] via P2P/CUMEM +n116-040-213:1385270:1385398 [2] NCCL INFO Channel 08/0 : 42[2] -> 43[3] via P2P/CUMEM +n116-040-213:1385268:1385399 [0] NCCL INFO Channel 14/0 : 40[0] -> 47[7] via P2P/CUMEM +n116-040-213:1385270:1385398 [2] NCCL INFO Channel 09/0 : 42[2] -> 43[3] via P2P/CUMEM +n116-040-213:1385270:1385398 [2] NCCL INFO Channel 10/0 : 42[2] -> 43[3] via P2P/CUMEM +n116-040-213:1385270:1385398 [2] NCCL INFO Channel 11/0 : 42[2] -> 43[3] via P2P/CUMEM +n116-040-213:1385270:1385398 [2] NCCL INFO Channel 12/0 : 42[2] -> 43[3] via P2P/CUMEM +n116-040-213:1385270:1385398 [2] NCCL INFO Channel 13/0 : 42[2] -> 43[3] via P2P/CUMEM +n116-040-213:1385270:1385398 [2] NCCL INFO Channel 14/0 : 42[2] -> 43[3] via P2P/CUMEM +n116-040-213:1385270:1385398 [2] NCCL INFO Channel 15/0 : 42[2] -> 43[3] via P2P/CUMEM +n116-040-213:1385273:1385400 [5] NCCL INFO [Proxy Progress] Device 5 CPU core 167 +n116-040-213:1385273:1385393 [5] NCCL INFO Channel 05/0 : 45[5] -> 52[4] [send] via NET/FasTrak/5/GDRDMA +n116-040-213:1385272:1385401 [4] NCCL INFO [Proxy Progress] Device 4 CPU core 96 +n116-040-213:1385273:1385393 [5] NCCL INFO Channel 12/0 : 45[5] -> 52[4] [send] via NET/FasTrak/4/GDRDMA +n116-040-213:1385274:1385402 [6] NCCL INFO [Proxy Progress] Device 6 CPU core 100 +n116-040-213:1385272:1385396 [4] NCCL INFO Channel 12/0 : 36[4] -> 44[4] [receive] via NET/FasTrak/4/GDRDMA +n116-040-213:1385274:1385394 [6] NCCL INFO Channel 14/0 : 38[6] -> 46[6] [receive] via NET/FasTrak/6/GDRDMA +n116-040-213:1385272:1385396 [4] NCCL INFO Channel 04/0 : 44[4] -> 53[5] [send] via NET/FasTrak/4/GDRDMA +n116-040-213:1385272:1385396 [4] NCCL INFO Channel 13/0 : 44[4] -> 53[5] [send] via NET/FasTrak/5/GDRDMA +n116-040-213:1385274:1385394 [6] NCCL INFO Channel 06/0 : 46[6] -> 55[7] [send] via NET/FasTrak/6/GDRDMA +n116-040-213:1385274:1385394 [6] NCCL INFO Channel 15/0 : 46[6] -> 55[7] [send] via NET/FasTrak/7/GDRDMA +n116-040-213:1385275:1385403 [7] NCCL INFO [Proxy Progress] Device 7 CPU core 71 +n116-040-213:1385275:1385397 [7] NCCL INFO Channel 07/0 : 47[7] -> 54[6] [send] via NET/FasTrak/7/GDRDMA +n116-040-213:1385275:1385397 [7] NCCL INFO Channel 14/0 : 47[7] -> 54[6] [send] via NET/FasTrak/6/GDRDMA +n116-040-213:1385270:1385405 [2] NCCL INFO [Proxy Progress] Device 2 CPU core 124 +n116-040-213:1385271:1385404 [3] NCCL INFO [Proxy Progress] Device 3 CPU core 122 +n116-040-213:1385271:1385392 [3] NCCL INFO Channel 03/0 : 43[3] -> 50[2] [send] via NET/FasTrak/3/GDRDMA +n116-040-213:1385269:1385406 [1] NCCL INFO [Proxy Progress] Device 1 CPU core 42 +n116-040-213:1385271:1385392 [3] NCCL INFO Channel 10/0 : 43[3] -> 50[2] [send] via NET/FasTrak/2/GDRDMA +n116-040-213:1385270:1385398 [2] NCCL INFO Channel 10/0 : 34[2] -> 42[2] [receive] via NET/FasTrak/2/GDRDMA +n116-040-213:1385269:1385395 [1] NCCL INFO Channel 01/0 : 41[1] -> 48[0] [send] via NET/FasTrak/1/GDRDMA +n116-040-213:1385269:1385395 [1] NCCL INFO Channel 08/0 : 41[1] -> 48[0] [send] via NET/FasTrak/0/GDRDMA +n116-040-213:1385270:1385398 [2] NCCL INFO Channel 02/0 : 42[2] -> 51[3] [send] via NET/FasTrak/2/GDRDMA +n116-040-213:1385270:1385398 [2] NCCL INFO Channel 11/0 : 42[2] -> 51[3] [send] via NET/FasTrak/3/GDRDMA +n116-040-213:1385268:1385407 [0] NCCL INFO [Proxy Progress] Device 0 CPU core 13 +n116-040-213:1385268:1385399 [0] NCCL INFO Channel 08/0 : 32[0] -> 40[0] [receive] via NET/FasTrak/0/GDRDMA +n116-040-213:1385268:1385399 [0] NCCL INFO Channel 00/0 : 40[0] -> 49[1] [send] via NET/FasTrak/0/GDRDMA +n116-040-213:1385268:1385399 [0] NCCL INFO Channel 09/0 : 40[0] -> 49[1] [send] via NET/FasTrak/1/GDRDMA +n116-040-213:1385275:1385397 [7] NCCL INFO Channel 15/0 : 39[7] -> 47[7] [receive] via NET/FasTrak/7/GDRDMA +n116-040-213:1385275:1385397 [7] NCCL INFO Channel 15/0 : 30[6] -> 47[7] [receive] via NET/FasTrak/7/GDRDMA +n116-040-213:1385269:1385395 [1] NCCL INFO Channel 09/0 : 33[1] -> 41[1] [receive] via NET/FasTrak/1/GDRDMA +n116-040-213:1385275:1385397 [7] NCCL INFO Channel 15/0 : 47[7] -> 30[6] [send] via NET/FasTrak/7/GDRDMA +n116-040-213:1385273:1385393 [5] NCCL INFO Channel 13/0 : 37[5] -> 45[5] [receive] via NET/FasTrak/5/GDRDMA +n116-040-213:1385272:1385396 [4] NCCL INFO Channel 12/0 : 29[5] -> 44[4] [receive] via NET/FasTrak/4/GDRDMA +n116-040-213:1385272:1385396 [4] NCCL INFO Channel 12/0 : 44[4] -> 29[5] [send] via NET/FasTrak/4/GDRDMA +n116-040-213:1385269:1385395 [1] NCCL INFO Channel 09/0 : 24[0] -> 41[1] [receive] via NET/FasTrak/1/GDRDMA +n116-040-213:1385273:1385393 [5] NCCL INFO Channel 13/0 : 28[4] -> 45[5] [receive] via NET/FasTrak/5/GDRDMA +n116-040-213:1385269:1385395 [1] NCCL INFO Channel 09/0 : 41[1] -> 24[0] [send] via NET/FasTrak/1/GDRDMA +n116-040-213:1385273:1385393 [5] NCCL INFO Channel 13/0 : 45[5] -> 28[4] [send] via NET/FasTrak/5/GDRDMA +n116-040-213:1385271:1385392 [3] NCCL INFO Channel 11/0 : 35[3] -> 43[3] [receive] via NET/FasTrak/3/GDRDMA +n116-040-213:1385270:1385398 [2] NCCL INFO Channel 10/0 : 27[3] -> 42[2] [receive] via NET/FasTrak/2/GDRDMA +n116-040-213:1385268:1385399 [0] NCCL INFO Channel 08/0 : 25[1] -> 40[0] [receive] via NET/FasTrak/0/GDRDMA +n116-040-213:1385271:1385392 [3] NCCL INFO Channel 11/0 : 26[2] -> 43[3] [receive] via NET/FasTrak/3/GDRDMA +n116-040-213:1385270:1385398 [2] NCCL INFO Channel 10/0 : 42[2] -> 27[3] [send] via NET/FasTrak/2/GDRDMA +n116-040-213:1385269:1385395 [1] NCCL INFO Channel 09/0 : 41[1] -> 33[1] [send] via NET/FasTrak/1/GDRDMA +n116-040-213:1385271:1385392 [3] NCCL INFO Channel 11/0 : 43[3] -> 26[2] [send] via NET/FasTrak/3/GDRDMA +n116-040-213:1385275:1385397 [7] NCCL INFO Channel 15/0 : 47[7] -> 39[7] [send] via NET/FasTrak/7/GDRDMA +n116-040-213:1385273:1385393 [5] NCCL INFO Channel 13/0 : 45[5] -> 37[5] [send] via NET/FasTrak/5/GDRDMA +n116-040-213:1385268:1385399 [0] NCCL INFO Channel 08/0 : 40[0] -> 25[1] [send] via NET/FasTrak/0/GDRDMA +n116-040-213:1385272:1385396 [4] NCCL INFO Channel 04/0 : 53[5] -> 44[4] [receive] via NET/FasTrak/4/GDRDMA +n116-040-213:1385270:1385398 [2] NCCL INFO Channel 02/0 : 51[3] -> 42[2] [receive] via NET/FasTrak/2/GDRDMA +n116-040-213:1385272:1385396 [4] NCCL INFO Channel 13/0 : 53[5] -> 44[4] [receive] via NET/FasTrak/5/GDRDMA +n116-040-213:1385270:1385398 [2] NCCL INFO Channel 11/0 : 51[3] -> 42[2] [receive] via NET/FasTrak/3/GDRDMA +n116-040-213:1385271:1385392 [3] NCCL INFO Channel 11/0 : 43[3] -> 35[3] [send] via NET/FasTrak/3/GDRDMA +n116-040-213:1385270:1385398 [2] NCCL INFO Channel 10/0 : 42[2] -> 34[2] [send] via NET/FasTrak/2/GDRDMA +n116-040-213:1385272:1385396 [4] NCCL INFO Channel 12/0 : 44[4] -> 36[4] [send] via NET/FasTrak/4/GDRDMA +n116-040-213:1385268:1385399 [0] NCCL INFO Channel 00/0 : 49[1] -> 40[0] [receive] via NET/FasTrak/0/GDRDMA +n116-040-213:1385271:1385392 [3] NCCL INFO Channel 03/0 : 50[2] -> 43[3] [receive] via NET/FasTrak/3/GDRDMA +n116-040-213:1385269:1385395 [1] NCCL INFO Channel 01/0 : 48[0] -> 41[1] [receive] via NET/FasTrak/1/GDRDMA +n116-040-213:1385268:1385399 [0] NCCL INFO Channel 09/0 : 49[1] -> 40[0] [receive] via NET/FasTrak/1/GDRDMA +n116-040-213:1385271:1385392 [3] NCCL INFO Channel 10/0 : 50[2] -> 43[3] [receive] via NET/FasTrak/2/GDRDMA +n116-040-213:1385273:1385393 [5] NCCL INFO Channel 05/0 : 52[4] -> 45[5] [receive] via NET/FasTrak/5/GDRDMA +n116-040-213:1385268:1385399 [0] NCCL INFO Channel 08/0 : 40[0] -> 32[0] [send] via NET/FasTrak/0/GDRDMA +n116-040-213:1385269:1385395 [1] NCCL INFO Channel 08/0 : 48[0] -> 41[1] [receive] via NET/FasTrak/0/GDRDMA +n116-040-213:1385273:1385393 [5] NCCL INFO Channel 12/0 : 52[4] -> 45[5] [receive] via NET/FasTrak/4/GDRDMA +n116-040-213:1385270:1385398 [2] NCCL INFO Channel 00/0 : 42[2] -> 41[1] via P2P/CUMEM +n116-040-213:1385273:1385393 [5] NCCL INFO Channel 00/0 : 45[5] -> 44[4] via P2P/CUMEM +n116-040-213:1385269:1385395 [1] NCCL INFO Channel 00/0 : 41[1] -> 40[0] via P2P/CUMEM +n116-040-213:1385270:1385398 [2] NCCL INFO Channel 03/0 : 42[2] -> 41[1] via P2P/CUMEM +n116-040-213:1385273:1385393 [5] NCCL INFO Channel 01/0 : 45[5] -> 44[4] via P2P/CUMEM +n116-040-213:1385271:1385392 [3] NCCL INFO Channel 00/0 : 43[3] -> 42[2] via P2P/CUMEM +n116-040-213:1385269:1385395 [1] NCCL INFO Channel 01/0 : 41[1] -> 40[0] via P2P/CUMEM +n116-040-213:1385273:1385393 [5] NCCL INFO Channel 02/0 : 45[5] -> 44[4] via P2P/CUMEM +n116-040-213:1385270:1385398 [2] NCCL INFO Channel 04/0 : 42[2] -> 41[1] via P2P/CUMEM +n116-040-213:1385272:1385396 [4] NCCL INFO Channel 00/0 : 44[4] -> 43[3] via P2P/CUMEM +n116-040-213:1385271:1385392 [3] NCCL INFO Channel 01/0 : 43[3] -> 42[2] via P2P/CUMEM +n116-040-213:1385269:1385395 [1] NCCL INFO Channel 02/0 : 41[1] -> 40[0] via P2P/CUMEM +n116-040-213:1385273:1385393 [5] NCCL INFO Channel 03/0 : 45[5] -> 44[4] via P2P/CUMEM +n116-040-213:1385270:1385398 [2] NCCL INFO Channel 05/0 : 42[2] -> 41[1] via P2P/CUMEM +n116-040-213:1385272:1385396 [4] NCCL INFO Channel 01/0 : 44[4] -> 43[3] via P2P/CUMEM +n116-040-213:1385271:1385392 [3] NCCL INFO Channel 02/0 : 43[3] -> 42[2] via P2P/CUMEM +n116-040-213:1385269:1385395 [1] NCCL INFO Channel 03/0 : 41[1] -> 40[0] via P2P/CUMEM +n116-040-213:1385273:1385393 [5] NCCL INFO Channel 04/0 : 45[5] -> 44[4] via P2P/CUMEM +n116-040-213:1385270:1385398 [2] NCCL INFO Channel 06/0 : 42[2] -> 41[1] via P2P/CUMEM +n116-040-213:1385274:1385394 [6] NCCL INFO Channel 14/0 : 31[7] -> 46[6] [receive] via NET/FasTrak/6/GDRDMA +n116-040-213:1385272:1385396 [4] NCCL INFO Channel 02/0 : 44[4] -> 43[3] via P2P/CUMEM +n116-040-213:1385271:1385392 [3] NCCL INFO Channel 03/0 : 43[3] -> 42[2] via P2P/CUMEM +n116-040-213:1385269:1385395 [1] NCCL INFO Channel 04/0 : 41[1] -> 40[0] via P2P/CUMEM +n116-040-213:1385273:1385393 [5] NCCL INFO Channel 05/0 : 45[5] -> 44[4] via P2P/CUMEM +n116-040-213:1385270:1385398 [2] NCCL INFO Channel 07/0 : 42[2] -> 41[1] via P2P/CUMEM +n116-040-213:1385272:1385396 [4] NCCL INFO Channel 05/0 : 44[4] -> 43[3] via P2P/CUMEM +n116-040-213:1385271:1385392 [3] NCCL INFO Channel 04/0 : 43[3] -> 42[2] via P2P/CUMEM +n116-040-213:1385269:1385395 [1] NCCL INFO Channel 05/0 : 41[1] -> 40[0] via P2P/CUMEM +n116-040-213:1385274:1385394 [6] NCCL INFO Channel 14/0 : 46[6] -> 31[7] [send] via NET/FasTrak/6/GDRDMA +n116-040-213:1385273:1385393 [5] NCCL INFO Channel 06/0 : 45[5] -> 44[4] via P2P/CUMEM +n116-040-213:1385270:1385398 [2] NCCL INFO Channel 08/0 : 42[2] -> 41[1] via P2P/CUMEM +n116-040-213:1385272:1385396 [4] NCCL INFO Channel 06/0 : 44[4] -> 43[3] via P2P/CUMEM +n116-040-213:1385271:1385392 [3] NCCL INFO Channel 05/0 : 43[3] -> 42[2] via P2P/CUMEM +n116-040-213:1385269:1385395 [1] NCCL INFO Channel 06/0 : 41[1] -> 40[0] via P2P/CUMEM +n116-040-213:1385273:1385393 [5] NCCL INFO Channel 07/0 : 45[5] -> 44[4] via P2P/CUMEM +n116-040-213:1385270:1385398 [2] NCCL INFO Channel 11/0 : 42[2] -> 41[1] via P2P/CUMEM +n116-040-213:1385272:1385396 [4] NCCL INFO Channel 07/0 : 44[4] -> 43[3] via P2P/CUMEM +n116-040-213:1385271:1385392 [3] NCCL INFO Channel 06/0 : 43[3] -> 42[2] via P2P/CUMEM +n116-040-213:1385269:1385395 [1] NCCL INFO Channel 07/0 : 41[1] -> 40[0] via P2P/CUMEM +n116-040-213:1385273:1385393 [5] NCCL INFO Channel 08/0 : 45[5] -> 44[4] via P2P/CUMEM +n116-040-213:1385270:1385398 [2] NCCL INFO Channel 12/0 : 42[2] -> 41[1] via P2P/CUMEM +n116-040-213:1385272:1385396 [4] NCCL INFO Channel 08/0 : 44[4] -> 43[3] via P2P/CUMEM +n116-040-213:1385269:1385395 [1] NCCL INFO Channel 08/0 : 41[1] -> 40[0] via P2P/CUMEM +n116-040-213:1385271:1385392 [3] NCCL INFO Channel 07/0 : 43[3] -> 42[2] via P2P/CUMEM +n116-040-213:1385273:1385393 [5] NCCL INFO Channel 09/0 : 45[5] -> 44[4] via P2P/CUMEM +n116-040-213:1385270:1385398 [2] NCCL INFO Channel 13/0 : 42[2] -> 41[1] via P2P/CUMEM +n116-040-213:1385272:1385396 [4] NCCL INFO Channel 09/0 : 44[4] -> 43[3] via P2P/CUMEM +n116-040-213:1385269:1385395 [1] NCCL INFO Channel 09/0 : 41[1] -> 40[0] via P2P/CUMEM +n116-040-213:1385271:1385392 [3] NCCL INFO Channel 08/0 : 43[3] -> 42[2] via P2P/CUMEM +n116-040-213:1385273:1385393 [5] NCCL INFO Channel 10/0 : 45[5] -> 44[4] via P2P/CUMEM +n116-040-213:1385270:1385398 [2] NCCL INFO Channel 14/0 : 42[2] -> 41[1] via P2P/CUMEM +n116-040-213:1385272:1385396 [4] NCCL INFO Channel 10/0 : 44[4] -> 43[3] via P2P/CUMEM +n116-040-213:1385269:1385395 [1] NCCL INFO Channel 10/0 : 41[1] -> 40[0] via P2P/CUMEM +n116-040-213:1385271:1385392 [3] NCCL INFO Channel 09/0 : 43[3] -> 42[2] via P2P/CUMEM +n116-040-213:1385273:1385393 [5] NCCL INFO Channel 11/0 : 45[5] -> 44[4] via P2P/CUMEM +n116-040-213:1385270:1385398 [2] NCCL INFO Channel 15/0 : 42[2] -> 41[1] via P2P/CUMEM +n116-040-213:1385272:1385396 [4] NCCL INFO Channel 13/0 : 44[4] -> 43[3] via P2P/CUMEM +n116-040-213:1385269:1385395 [1] NCCL INFO Channel 11/0 : 41[1] -> 40[0] via P2P/CUMEM +n116-040-213:1385271:1385392 [3] NCCL INFO Channel 10/0 : 43[3] -> 42[2] via P2P/CUMEM +n116-040-213:1385273:1385393 [5] NCCL INFO Channel 12/0 : 45[5] -> 44[4] via P2P/CUMEM +n116-040-213:1385272:1385396 [4] NCCL INFO Channel 14/0 : 44[4] -> 43[3] via P2P/CUMEM +n116-040-213:1385271:1385392 [3] NCCL INFO Channel 11/0 : 43[3] -> 42[2] via P2P/CUMEM +n116-040-213:1385269:1385395 [1] NCCL INFO Channel 12/0 : 41[1] -> 40[0] via P2P/CUMEM +n116-040-213:1385273:1385393 [5] NCCL INFO Channel 13/0 : 45[5] -> 44[4] via P2P/CUMEM +n116-040-213:1385272:1385396 [4] NCCL INFO Channel 15/0 : 44[4] -> 43[3] via P2P/CUMEM +n116-040-213:1385269:1385395 [1] NCCL INFO Channel 13/0 : 41[1] -> 40[0] via P2P/CUMEM +n116-040-213:1385271:1385392 [3] NCCL INFO Channel 12/0 : 43[3] -> 42[2] via P2P/CUMEM +n116-040-213:1385273:1385393 [5] NCCL INFO Channel 14/0 : 45[5] -> 44[4] via P2P/CUMEM +n116-040-213:1385269:1385395 [1] NCCL INFO Channel 14/0 : 41[1] -> 40[0] via P2P/CUMEM +n116-040-213:1385271:1385392 [3] NCCL INFO Channel 13/0 : 43[3] -> 42[2] via P2P/CUMEM +n116-040-213:1385273:1385393 [5] NCCL INFO Channel 15/0 : 45[5] -> 44[4] via P2P/CUMEM +n116-040-213:1385269:1385395 [1] NCCL INFO Channel 15/0 : 41[1] -> 40[0] via P2P/CUMEM +n116-040-213:1385271:1385392 [3] NCCL INFO Channel 14/0 : 43[3] -> 42[2] via P2P/CUMEM +n116-040-213:1385271:1385392 [3] NCCL INFO Channel 15/0 : 43[3] -> 42[2] via P2P/CUMEM +n116-040-213:1385275:1385397 [7] NCCL INFO Channel 07/0 : 54[6] -> 47[7] [receive] via NET/FasTrak/7/GDRDMA +n116-040-213:1385275:1385397 [7] NCCL INFO Channel 14/0 : 54[6] -> 47[7] [receive] via NET/FasTrak/6/GDRDMA +n116-040-213:1385275:1385397 [7] NCCL INFO Channel 01/0 : 47[7] -> 40[0] via P2P/CUMEM +n116-040-213:1385274:1385394 [6] NCCL INFO Channel 06/0 : 55[7] -> 46[6] [receive] via NET/FasTrak/6/GDRDMA +n116-040-213:1385275:1385397 [7] NCCL INFO Channel 02/0 : 47[7] -> 40[0] via P2P/CUMEM +n116-040-213:1385274:1385394 [6] NCCL INFO Channel 15/0 : 55[7] -> 46[6] [receive] via NET/FasTrak/7/GDRDMA +n116-040-213:1385274:1385394 [6] NCCL INFO Channel 14/0 : 46[6] -> 38[6] [send] via NET/FasTrak/6/GDRDMA +n116-040-213:1385275:1385397 [7] NCCL INFO Channel 03/0 : 47[7] -> 40[0] via P2P/CUMEM +n116-040-213:1385275:1385397 [7] NCCL INFO Channel 04/0 : 47[7] -> 40[0] via P2P/CUMEM +n116-040-213:1385275:1385397 [7] NCCL INFO Channel 05/0 : 47[7] -> 40[0] via P2P/CUMEM +n116-040-213:1385275:1385397 [7] NCCL INFO Channel 06/0 : 47[7] -> 40[0] via P2P/CUMEM +n116-040-213:1385275:1385397 [7] NCCL INFO Channel 09/0 : 47[7] -> 40[0] via P2P/CUMEM +n116-040-213:1385275:1385397 [7] NCCL INFO Channel 10/0 : 47[7] -> 40[0] via P2P/CUMEM +n116-040-213:1385275:1385397 [7] NCCL INFO Channel 11/0 : 47[7] -> 40[0] via P2P/CUMEM +n116-040-213:1385275:1385397 [7] NCCL INFO Channel 12/0 : 47[7] -> 40[0] via P2P/CUMEM +n116-040-213:1385275:1385397 [7] NCCL INFO Channel 13/0 : 47[7] -> 40[0] via P2P/CUMEM +n116-040-213:1385275:1385397 [7] NCCL INFO Channel 14/0 : 47[7] -> 40[0] via P2P/CUMEM +n116-040-213:1385275:1385397 [7] NCCL INFO Channel 00/0 : 47[7] -> 46[6] via P2P/CUMEM +n116-040-213:1385275:1385397 [7] NCCL INFO Channel 01/0 : 47[7] -> 46[6] via P2P/CUMEM +n116-040-213:1385275:1385397 [7] NCCL INFO Channel 02/0 : 47[7] -> 46[6] via P2P/CUMEM +n116-040-213:1385274:1385394 [6] NCCL INFO Channel 00/0 : 46[6] -> 45[5] via P2P/CUMEM +n116-040-213:1385275:1385397 [7] NCCL INFO Channel 03/0 : 47[7] -> 46[6] via P2P/CUMEM +n116-040-213:1385274:1385394 [6] NCCL INFO Channel 01/0 : 46[6] -> 45[5] via P2P/CUMEM +n116-040-213:1385275:1385397 [7] NCCL INFO Channel 04/0 : 47[7] -> 46[6] via P2P/CUMEM +n116-040-213:1385275:1385397 [7] NCCL INFO Channel 05/0 : 47[7] -> 46[6] via P2P/CUMEM +n116-040-213:1385274:1385394 [6] NCCL INFO Channel 02/0 : 46[6] -> 45[5] via P2P/CUMEM +n116-040-213:1385275:1385397 [7] NCCL INFO Channel 06/0 : 47[7] -> 46[6] via P2P/CUMEM +n116-040-213:1385275:1385397 [7] NCCL INFO Channel 07/0 : 47[7] -> 46[6] via P2P/CUMEM +n116-040-213:1385274:1385394 [6] NCCL INFO Channel 03/0 : 46[6] -> 45[5] via P2P/CUMEM +n116-040-213:1385274:1385394 [6] NCCL INFO Channel 04/0 : 46[6] -> 45[5] via P2P/CUMEM +n116-040-213:1385275:1385397 [7] NCCL INFO Channel 08/0 : 47[7] -> 46[6] via P2P/CUMEM +n116-040-213:1385274:1385394 [6] NCCL INFO Channel 07/0 : 46[6] -> 45[5] via P2P/CUMEM +n116-040-213:1385275:1385397 [7] NCCL INFO Channel 09/0 : 47[7] -> 46[6] via P2P/CUMEM +n116-040-213:1385274:1385394 [6] NCCL INFO Channel 08/0 : 46[6] -> 45[5] via P2P/CUMEM +n116-040-213:1385275:1385397 [7] NCCL INFO Channel 10/0 : 47[7] -> 46[6] via P2P/CUMEM +n116-040-213:1385274:1385394 [6] NCCL INFO Channel 09/0 : 46[6] -> 45[5] via P2P/CUMEM +n116-040-213:1385275:1385397 [7] NCCL INFO Channel 11/0 : 47[7] -> 46[6] via P2P/CUMEM +n116-040-213:1385274:1385394 [6] NCCL INFO Channel 10/0 : 46[6] -> 45[5] via P2P/CUMEM +n116-040-213:1385275:1385397 [7] NCCL INFO Channel 12/0 : 47[7] -> 46[6] via P2P/CUMEM +n116-040-213:1385274:1385394 [6] NCCL INFO Channel 11/0 : 46[6] -> 45[5] via P2P/CUMEM +n116-040-213:1385275:1385397 [7] NCCL INFO Channel 13/0 : 47[7] -> 46[6] via P2P/CUMEM +n116-040-213:1385274:1385394 [6] NCCL INFO Channel 12/0 : 46[6] -> 45[5] via P2P/CUMEM +n116-040-213:1385275:1385397 [7] NCCL INFO Channel 14/0 : 47[7] -> 46[6] via P2P/CUMEM +n116-040-213:1385274:1385394 [6] NCCL INFO Channel 15/0 : 46[6] -> 45[5] via P2P/CUMEM +n116-040-213:1385275:1385397 [7] NCCL INFO Channel 15/0 : 47[7] -> 46[6] via P2P/CUMEM +n116-040-213:1385270:1385398 [2] NCCL INFO Connected all trees +n116-040-213:1385268:1385399 [0] NCCL INFO Connected all trees +n116-040-213:1385269:1385395 [1] NCCL INFO Connected all trees +n116-040-213:1385275:1385397 [7] NCCL INFO Connected all trees +n116-040-213:1385274:1385394 [6] NCCL INFO Connected all trees +n116-040-213:1385271:1385392 [3] NCCL INFO Connected all trees +n116-040-213:1385273:1385393 [5] NCCL INFO Connected all trees +n116-040-213:1385272:1385396 [4] NCCL INFO Connected all trees +n116-040-213:1385268:1385428 [0] NCCL INFO Channel 00/0 : 33[1] -> 40[0] [receive] via NET/FasTrak/0/GDRDMA +n116-040-213:1385268:1385428 [0] NCCL INFO Channel 08/0 : 33[1] -> 40[0] [receive] via NET/FasTrak/0/GDRDMA +n116-040-213:1385268:1385428 [0] NCCL INFO Channel 00/0 : 40[0] -> 47[7] via P2P/CUMEM +n116-040-213:1385268:1385428 [0] NCCL INFO Channel 08/0 : 40[0] -> 47[7] via P2P/CUMEM +n116-040-213:1385273:1385429 [5] NCCL INFO Channel 05/0 : 45[5] -> 46[6] via P2P/CUMEM +n116-040-213:1385273:1385429 [5] NCCL INFO Channel 13/0 : 45[5] -> 46[6] via P2P/CUMEM +n116-040-213:1385273:1385429 [5] NCCL INFO Channel 04/0 : 45[5] -> 52[4] [send] via NET/FasTrak/4/GDRDMA +n116-040-213:1385274:1385430 [6] NCCL INFO Channel 06/0 : 39[7] -> 46[6] [receive] via NET/FasTrak/6/GDRDMA +n116-040-213:1385274:1385430 [6] NCCL INFO Channel 14/0 : 39[7] -> 46[6] [receive] via NET/FasTrak/6/GDRDMA +n116-040-213:1385271:1385431 [3] NCCL INFO Channel 03/0 : 43[3] -> 44[4] via P2P/CUMEM +n116-040-213:1385271:1385431 [3] NCCL INFO Channel 11/0 : 43[3] -> 44[4] via P2P/CUMEM +n116-040-213:1385271:1385431 [3] NCCL INFO Channel 02/0 : 43[3] -> 50[2] [send] via NET/FasTrak/2/GDRDMA +n116-040-213:1385272:1385432 [4] NCCL INFO Channel 04/0 : 37[5] -> 44[4] [receive] via NET/FasTrak/4/GDRDMA +n116-040-213:1385272:1385432 [4] NCCL INFO Channel 12/0 : 37[5] -> 44[4] [receive] via NET/FasTrak/4/GDRDMA +n116-040-213:1385269:1385433 [1] NCCL INFO Channel 01/0 : 41[1] -> 42[2] via P2P/CUMEM +n116-040-213:1385269:1385433 [1] NCCL INFO Channel 09/0 : 41[1] -> 42[2] via P2P/CUMEM +n116-040-213:1385269:1385433 [1] NCCL INFO Channel 00/0 : 41[1] -> 48[0] [send] via NET/FasTrak/0/GDRDMA +n116-040-213:1385270:1385434 [2] NCCL INFO Channel 02/0 : 35[3] -> 42[2] [receive] via NET/FasTrak/2/GDRDMA +n116-040-213:1385270:1385434 [2] NCCL INFO Channel 10/0 : 35[3] -> 42[2] [receive] via NET/FasTrak/2/GDRDMA +n116-040-213:1385270:1385434 [2] NCCL INFO Channel 03/0 : 42[2] -> 51[3] [send] via NET/FasTrak/3/GDRDMA +n116-040-213:1385273:1385429 [5] NCCL INFO Channel 05/0 : 36[4] -> 45[5] [receive] via NET/FasTrak/5/GDRDMA +n116-040-213:1385270:1385434 [2] NCCL INFO Channel 02/0 : 42[2] -> 41[1] via P2P/CUMEM +n116-040-213:1385273:1385429 [5] NCCL INFO Channel 13/0 : 36[4] -> 45[5] [receive] via NET/FasTrak/5/GDRDMA +n116-040-213:1385270:1385434 [2] NCCL INFO Channel 10/0 : 42[2] -> 41[1] via P2P/CUMEM +n116-040-213:1385275:1385435 [7] NCCL INFO Channel 06/0 : 47[7] -> 54[6] [send] via NET/FasTrak/6/GDRDMA +n116-040-213:1385275:1385435 [7] NCCL INFO Channel 07/0 : 38[6] -> 47[7] [receive] via NET/FasTrak/7/GDRDMA +n116-040-213:1385275:1385435 [7] NCCL INFO Channel 15/0 : 38[6] -> 47[7] [receive] via NET/FasTrak/7/GDRDMA +n116-040-213:1385269:1385433 [1] NCCL INFO Channel 01/0 : 32[0] -> 41[1] [receive] via NET/FasTrak/1/GDRDMA +n116-040-213:1385269:1385433 [1] NCCL INFO Channel 09/0 : 32[0] -> 41[1] [receive] via NET/FasTrak/1/GDRDMA +n116-040-213:1385271:1385431 [3] NCCL INFO Channel 03/0 : 34[2] -> 43[3] [receive] via NET/FasTrak/3/GDRDMA +n116-040-213:1385271:1385431 [3] NCCL INFO Channel 11/0 : 34[2] -> 43[3] [receive] via NET/FasTrak/3/GDRDMA +n116-040-213:1385274:1385430 [6] NCCL INFO Channel 07/0 : 46[6] -> 55[7] [send] via NET/FasTrak/7/GDRDMA +n116-040-213:1385274:1385430 [6] NCCL INFO Channel 06/0 : 46[6] -> 45[5] via P2P/CUMEM +n116-040-213:1385274:1385430 [6] NCCL INFO Channel 14/0 : 46[6] -> 45[5] via P2P/CUMEM +n116-040-213:1385272:1385432 [4] NCCL INFO Channel 05/0 : 44[4] -> 53[5] [send] via NET/FasTrak/5/GDRDMA +n116-040-213:1385272:1385432 [4] NCCL INFO Channel 04/0 : 44[4] -> 43[3] via P2P/CUMEM +n116-040-213:1385275:1385435 [7] NCCL INFO Channel 07/0 : 47[7] -> 40[0] via P2P/CUMEM +n116-040-213:1385272:1385432 [4] NCCL INFO Channel 12/0 : 44[4] -> 43[3] via P2P/CUMEM +n116-040-213:1385275:1385435 [7] NCCL INFO Channel 15/0 : 47[7] -> 40[0] via P2P/CUMEM +n116-040-213:1385268:1385428 [0] NCCL INFO Channel 01/0 : 40[0] -> 49[1] [send] via NET/FasTrak/1/GDRDMA +n116-040-213:1385274:1385430 [6] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-040-213:1385275:1385435 [7] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-040-213:1385273:1385429 [5] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-040-213:1385269:1385433 [1] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-040-213:1385270:1385434 [2] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-040-213:1385271:1385431 [3] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-040-213:1385268:1385428 [0] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-040-213:1385272:1385432 [4] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +[rank45]:[W727 05:39:50.595340843 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n116-040-213.byted.org]:43390, remote=[n116-040-168.byted.org]:25264): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7fbfde3785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7fbfc76cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7fbfc76cff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7fbfc76d084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7fbfc76ca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fbf88dc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7fbfe189b4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7fbfe44721f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7fbfe44f289c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank45]:[W727 05:39:50.607325362 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 45] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank47]:[W727 05:39:50.791872240 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n116-040-213.byted.org]:43418, remote=[n116-040-168.byted.org]:25264): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f440cd785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f43f60cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7f43f60cff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7f43f60d084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7f43f60ca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f43b77c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f44102314a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f4412e081f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f4412e8889c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank47]:[W727 05:39:50.800375902 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 47] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank42]:[W727 05:39:50.818402338 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n116-040-213.byted.org]:43422, remote=[n116-040-168.byted.org]:25264): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f1d363785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f1d1facdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7f1d1facff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7f1d1fad084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7f1d1faca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f1ce11c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f1d39ae24a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f1d3c6b91f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f1d3c73989c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank42]:[W727 05:39:50.825108873 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 42] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank41]:[W727 05:39:50.988243520 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n116-040-213.byted.org]:43412, remote=[n116-040-168.byted.org]:25264): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f82527785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f823bacdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7f823bacff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7f823bad084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7f823baca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f81fd1c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f8255bd74a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f82587ae1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f825882e89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank41]:[W727 05:39:50.000378534 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 41] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank46]:[W727 05:39:51.173698000 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n116-040-213.byted.org]:43374, remote=[n116-040-168.byted.org]:25264): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f63047785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f62edacdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7f62edacff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7f62edad084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7f62edaca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f62af1c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f6307bac4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f630a7831f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f630a80389c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank46]:[W727 05:39:51.180941808 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 46] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank44]:[W727 05:39:51.212885712 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n116-040-213.byted.org]:43372, remote=[n116-040-168.byted.org]:25264): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f94d09785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f94b9ccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7f94b9ccff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7f94b9cd084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7f94b9cca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f947b3c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f94d3e3e4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f94d6a151f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f94d6a9589c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank44]:[W727 05:39:51.219646123 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 44] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank43]:[W727 05:39:51.338993172 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n116-040-213.byted.org]:43398, remote=[n116-040-168.byted.org]:25264): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f9ba99785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f9b930cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7f9b930cff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7f9b930d084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7f9b930ca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f9b547c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f9bad1844a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f9bafd5b1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f9bafddb89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank43]:[W727 05:39:51.345345912 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 43] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank40]:[W727 05:39:51.432566190 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=71, addr=[n116-040-213.byted.org]:43362, remote=[n116-040-168.byted.org]:25264): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f561db785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f56072cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7f56072cff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7f56072d084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7f56072ca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f55c89c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f562134b4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f5623f221f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f5623fa289c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank40]:[W727 05:39:51.441513512 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 40] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank45]:[W727 05:39:51.607510457 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-040-213.byted.org]:43390, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7fbfde3785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7fbfc76cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7fbfc76cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7fbfc76d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7fbfc76ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fbf88dc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7fbfe189b4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7fbfe44721f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7fbfe44f289c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank45]:[W727 05:39:51.616036780 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 45] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank47]:[W727 05:39:51.800575779 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-040-213.byted.org]:43418, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f440cd785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f43f60cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f43f60cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f43f60d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f43f60ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f43b77c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f44102314a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f4412e081f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f4412e8889c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank47]:[W727 05:39:51.808414052 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 47] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank42]:[W727 05:39:51.825286982 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-040-213.byted.org]:43422, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f1d363785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f1d1facdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f1d1facf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f1d1fad0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f1d1faca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f1ce11c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f1d39ae24a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f1d3c6b91f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f1d3c73989c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank42]:[W727 05:39:51.831268719 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 42] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank41]:[W727 05:39:51.000551479 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-040-213.byted.org]:43412, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f82527785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f823bacdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f823bacf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f823bad0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f823baca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f81fd1c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f8255bd74a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f82587ae1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f825882e89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank41]:[W727 05:39:51.006911814 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 41] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank46]:[W727 05:39:52.181125097 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-040-213.byted.org]:43374, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f63047785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f62edacdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f62edacf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f62edad0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f62edaca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f62af1c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f6307bac4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f630a7831f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f630a80389c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank46]:[W727 05:39:52.187368565 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 46] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank44]:[W727 05:39:52.219810233 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-040-213.byted.org]:43372, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f94d09785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f94b9ccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f94b9ccf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f94b9cd0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f94b9cca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f947b3c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f94d3e3e4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f94d6a151f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f94d6a9589c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank44]:[W727 05:39:52.225947743 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 44] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank43]:[W727 05:39:52.345522587 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-040-213.byted.org]:43398, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f9ba99785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f9b930cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f9b930cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f9b930d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f9b930ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f9b547c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f9bad1844a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f9bafd5b1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f9bafddb89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank43]:[W727 05:39:52.351594283 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 43] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank40]:[W727 05:39:52.441687932 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=71, addr=[n116-040-213.byted.org]:43362, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f561db785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f56072cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f56072cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f56072d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f56072ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f55c89c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f562134b4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f5623f221f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f5623fa289c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank40]:[W727 05:39:52.448568263 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 40] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank45]:[W727 05:39:52.616220487 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-040-213.byted.org]:43390, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7fbfde3785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7fbfc76cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7fbfc76cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7fbfc76d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7fbfc76ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fbf88dc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7fbfe189b4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7fbfe44721f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7fbfe44f289c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank45]:[W727 05:39:52.622400015 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 45] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank47]:[W727 05:39:52.808598867 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-040-213.byted.org]:43418, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f440cd785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f43f60cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f43f60cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f43f60d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f43f60ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f43b77c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f44102314a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f4412e081f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f4412e8889c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank47]:[W727 05:39:52.815840299 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 47] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank42]:[W727 05:39:52.831431233 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-040-213.byted.org]:43422, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f1d363785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f1d1facdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f1d1facf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f1d1fad0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f1d1faca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f1ce11c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f1d39ae24a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f1d3c6b91f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f1d3c73989c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank42]:[W727 05:39:52.838387050 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 42] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank41]:[W727 05:39:52.007095741 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-040-213.byted.org]:43412, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f82527785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f823bacdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f823bacf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f823bad0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f823baca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f81fd1c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f8255bd74a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f82587ae1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f825882e89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank41]:[W727 05:39:52.013715315 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 41] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank46]:[W727 05:39:53.187589424 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-040-213.byted.org]:43374, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f63047785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f62edacdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f62edacf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f62edad0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f62edaca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f62af1c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f6307bac4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f630a7831f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f630a80389c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank46]:[W727 05:39:53.212569113 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 46] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank44]:[W727 05:39:53.226113955 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-040-213.byted.org]:43372, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f94d09785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f94b9ccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f94b9ccf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f94b9cd0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f94b9cca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f947b3c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f94d3e3e4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f94d6a151f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f94d6a9589c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank44]:[W727 05:39:53.232353147 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 44] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank43]:[W727 05:39:53.351771506 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-040-213.byted.org]:43398, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f9ba99785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f9b930cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f9b930cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f9b930d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f9b930ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f9b547c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f9bad1844a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f9bafd5b1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f9bafddb89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank43]:[W727 05:39:53.357913533 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 43] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank40]:[W727 05:39:53.448751143 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=71, addr=[n116-040-213.byted.org]:43362, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f561db785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f56072cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f56072cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f56072d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f56072ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f55c89c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f562134b4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f5623f221f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f5623fa289c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank40]:[W727 05:39:53.455070948 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 40] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank45]:[W727 05:39:53.622580790 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-040-213.byted.org]:43390, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7fbfde3785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7fbfc76cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7fbfc76cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7fbfc76d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7fbfc76ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fbf88dc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7fbfe189b4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7fbfe44721f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7fbfe44f289c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank45]:[W727 05:39:53.629097026 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 45] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank47]:[W727 05:39:53.816052142 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-040-213.byted.org]:43418, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f440cd785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f43f60cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f43f60cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f43f60d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f43f60ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f43b77c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f44102314a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f4412e081f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f4412e8889c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank47]:[W727 05:39:53.823575097 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 47] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank42]:[W727 05:39:53.838588367 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-040-213.byted.org]:43422, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f1d363785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f1d1facdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f1d1facf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f1d1fad0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f1d1faca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f1ce11c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f1d39ae24a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f1d3c6b91f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f1d3c73989c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank42]:[W727 05:39:53.844633117 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 42] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank41]:[W727 05:39:53.013879229 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-040-213.byted.org]:43412, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f82527785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f823bacdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f823bacf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f823bad0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f823baca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f81fd1c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f8255bd74a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f82587ae1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f825882e89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank41]:[W727 05:39:53.020349079 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 41] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank46]:[W727 05:39:54.212748083 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-040-213.byted.org]:43374, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f63047785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f62edacdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f62edacf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f62edad0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f62edaca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f62af1c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f6307bac4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f630a7831f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f630a80389c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank46]:[W727 05:39:54.219049091 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 46] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank44]:[W727 05:39:54.232507634 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-040-213.byted.org]:43372, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f94d09785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f94b9ccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f94b9ccf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f94b9cd0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f94b9cca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f947b3c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f94d3e3e4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f94d6a151f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f94d6a9589c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank44]:[W727 05:39:54.238955101 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 44] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank43]:[W727 05:39:54.358096721 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-040-213.byted.org]:43398, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f9ba99785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f9b930cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f9b930cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f9b930d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f9b930ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f9b547c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f9bad1844a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f9bafd5b1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f9bafddb89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank43]:[W727 05:39:54.364872135 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 43] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank40]:[W727 05:39:54.455227770 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=71, addr=[n116-040-213.byted.org]:43362, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f561db785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f56072cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f56072cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f56072d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f56072ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f55c89c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f562134b4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f5623f221f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f5623fa289c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank40]:[W727 05:39:54.462030984 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 40] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank45]:[W727 05:39:54.629268837 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-040-213.byted.org]:43390, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7fbfde3785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7fbfc76cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7fbfc76cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7fbfc76d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7fbfc76ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fbf88dc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7fbfe189b4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7fbfe44721f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7fbfe44f289c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank45]:[W727 05:39:54.635609124 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 45] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank47]:[W727 05:39:54.823792181 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-040-213.byted.org]:43418, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f440cd785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f43f60cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f43f60cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f43f60d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f43f60ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f43b77c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f44102314a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f4412e081f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f4412e8889c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank47]:[W727 05:39:54.831081170 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 47] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank42]:[W727 05:39:54.844798020 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-040-213.byted.org]:43422, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f1d363785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f1d1facdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f1d1facf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f1d1fad0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f1d1faca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f1ce11c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f1d39ae24a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f1d3c6b91f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f1d3c73989c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank42]:[W727 05:39:54.851571694 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 42] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank41]:[W727 05:39:54.020497162 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-040-213.byted.org]:43412, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f82527785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f823bacdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f823bacf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f823bad0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f823baca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f81fd1c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f8255bd74a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f82587ae1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f825882e89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank41]:[W727 05:39:54.034822815 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 41] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank46]:[W727 05:39:55.219208517 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-040-213.byted.org]:43374, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f63047785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f62edacdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f62edacf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f62edad0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f62edaca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f62af1c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f6307bac4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f630a7831f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f630a80389c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank46]:[W727 05:39:55.226068368 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 46] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank44]:[W727 05:39:55.239113593 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-040-213.byted.org]:43372, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f94d09785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f94b9ccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f94b9ccf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f94b9cd0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f94b9cca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f947b3c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f94d3e3e4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f94d6a151f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f94d6a9589c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank44]:[W727 05:39:55.246285267 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 44] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank43]:[W727 05:39:55.365052117 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-040-213.byted.org]:43398, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f9ba99785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f9b930cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f9b930cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f9b930d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f9b930ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f9b547c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f9bad1844a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f9bafd5b1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f9bafddb89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank43]:[W727 05:39:55.372228730 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 43] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank40]:[W727 05:39:55.462186998 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=71, addr=[n116-040-213.byted.org]:43362, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f561db785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f56072cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f56072cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f56072d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f56072ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f55c89c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f562134b4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f5623f221f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f5623fa289c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank40]:[W727 05:39:55.467878621 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 40] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank45]:[W727 05:39:55.635790499 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-040-213.byted.org]:43390, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7fbfde3785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7fbfc76cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7fbfc76cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7fbfc76d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7fbfc76ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fbf88dc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7fbfe189b4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7fbfe44721f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7fbfe44f289c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank45]:[W727 05:39:55.642989193 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 45] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank47]:[W727 05:39:55.831261821 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-040-213.byted.org]:43418, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f440cd785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f43f60cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f43f60cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f43f60d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f43f60ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f43b77c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f44102314a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f4412e081f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f4412e8889c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank47]:[W727 05:39:55.838346311 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 47] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank42]:[W727 05:39:55.851729651 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-040-213.byted.org]:43422, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f1d363785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f1d1facdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f1d1facf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f1d1fad0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f1d1faca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f1ce11c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f1d39ae24a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f1d3c6b91f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f1d3c73989c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank42]:[W727 05:39:55.858758998 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 42] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank41]:[W727 05:39:55.034972500 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-040-213.byted.org]:43412, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f82527785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f823bacdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f823bacf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f823bad0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f823baca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f81fd1c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f8255bd74a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f82587ae1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f825882e89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank41]:[W727 05:39:55.041720259 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 41] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank46]:[W727 05:39:56.226255171 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-040-213.byted.org]:43374, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f63047785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f62edacdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f62edacf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f62edad0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f62edaca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f62af1c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f6307bac4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f630a7831f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f630a80389c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank46]:[W727 05:39:56.232221131 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 46] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank44]:[W727 05:39:56.246443609 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-040-213.byted.org]:43372, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f94d09785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f94b9ccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f94b9ccf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f94b9cd0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f94b9cca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f947b3c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f94d3e3e4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f94d6a151f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f94d6a9589c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank44]:[W727 05:39:56.252465566 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 44] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +E0727 05:39:56.156000 1385203 torch/distributed/elastic/multiprocessing/api.py:874] failed (exitcode: -15) local_rank: 0 (pid: 1385268) of binary: /usr/bin/python3 +Traceback (most recent call last): + File "/usr/local/bin/torchrun", line 7, in + sys.exit(main()) + ^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 355, in wrapper + return f(*args, **kwargs) + ^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 892, in main + run(args) + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 883, in run + elastic_launch( + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 139, in __call__ + return launch_agent(self._config, self._entrypoint, list(args)) + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 270, in launch_agent + raise ChildFailedError( +torch.distributed.elastic.multiprocessing.errors.ChildFailedError: +============================================================ +main_cross_2direct_balance_realscale_perturb_v4_accum.py FAILED +------------------------------------------------------------ +Failures: +[1]: + time : 2026-07-27_05:39:56 + host : n116-040-213.byted.org + rank : 41 (local_rank: 1) + exitcode : -15 (pid: 1385269) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 1385269 +[2]: + time : 2026-07-27_05:39:56 + host : n116-040-213.byted.org + rank : 42 (local_rank: 2) + exitcode : -15 (pid: 1385270) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 1385270 +[3]: + time : 2026-07-27_05:39:56 + host : n116-040-213.byted.org + rank : 43 (local_rank: 3) + exitcode : -15 (pid: 1385271) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 1385271 +[4]: + time : 2026-07-27_05:39:56 + host : n116-040-213.byted.org + rank : 44 (local_rank: 4) + exitcode : -15 (pid: 1385272) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 1385272 +[5]: + time : 2026-07-27_05:39:56 + host : n116-040-213.byted.org + rank : 45 (local_rank: 5) + exitcode : -15 (pid: 1385273) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 1385273 +[6]: + time : 2026-07-27_05:39:56 + host : n116-040-213.byted.org + rank : 46 (local_rank: 6) + exitcode : -15 (pid: 1385274) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 1385274 +[7]: + time : 2026-07-27_05:39:56 + host : n116-040-213.byted.org + rank : 47 (local_rank: 7) + exitcode : -15 (pid: 1385275) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 1385275 +------------------------------------------------------------ +Root Cause (first observed failure): +[0]: + time : 2026-07-27_05:39:56 + host : n116-040-213.byted.org + rank : 40 (local_rank: 0) + exitcode : -15 (pid: 1385268) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 1385268 +============================================================ diff --git a/ScaleDF_rebuttal/_host_logs/20260726_2350_vith64_free8_port25264_launch/launcher_rank6_n116-041-181.log b/ScaleDF_rebuttal/_host_logs/20260726_2350_vith64_free8_port25264_launch/launcher_rank6_n116-041-181.log new file mode 100644 index 0000000000000000000000000000000000000000..7525cdaf943c4dbd5e360885cf273827233a0ca8 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_2350_vith64_free8_port25264_launch/launcher_rank6_n116-041-181.log @@ -0,0 +1,2336 @@ +started_utc=2026-07-26T23:50:28Z +hostname=n116-041-181 +node_rank=6 +nnodes=8 +nproc_per_node=8 +world_size=64 +master_addr=10.116.40.168 +master_port=25264 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +main=main_cross_2direct_balance_realscale_perturb_v4_accum.py +model_name=deit_huge_patch14_LS +pretrain=/tmp/ScaleDF_repro_20260724/pretrained/deit_3_huge_224_21k_v1.pth +resume=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/checkpoint_snapshots/vith32_latest_for_64card_effbs4096_20260726_234604.pth +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_64card_effbs4096_from_epoch22_snapshot_free8_master_n116-040-168_20260726_2350_vith64_port25264 +batch_per_gpu=64 +accum_iter=1 +effective_batch=4096 +eval_batch_per_gpu=64 +epochs=40 +W0726 23:50:40.661000 72845 torch/distributed/run.py:766] +W0726 23:50:40.661000 72845 torch/distributed/run.py:766] ***************************************** +W0726 23:50:40.661000 72845 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 23:50:40.661000 72845 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 48): env:// +| distributed init (rank 49): env:// +| distributed init (rank 54): env:// +[rank49]:[W726 23:51:08.570804313 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 49] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank54]:[W726 23:51:08.571307615 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 54] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 50): env:// +[rank50]:[W726 23:51:09.312774428 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 50] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank48]:[W726 23:51:09.312798952 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 48] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 51): env:// +[rank51]:[W726 23:51:09.533814649 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 51] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 55): env:// +[rank55]:[W726 23:51:09.596944816 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 55] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 52): env:// +[rank52]:[W726 23:51:09.626237207 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 52] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 53): env:// +[rank53]:[W726 23:51:09.636834654 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 53] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +n116-041-181:72915:72915 [5] NCCL INFO cudaDriverVersion 12060 +n116-041-181:72915:72915 [5] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-041-181:72915:72915 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-041-181:72915:72915 [5] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-041-181:72915:72915 [5] NCCL INFO Comm config Blocking set to 1 +n116-041-181:72917:72917 [7] NCCL INFO cudaDriverVersion 12060 +n116-041-181:72917:72917 [7] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-041-181:72917:72917 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-041-181:72917:72917 [7] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-041-181:72917:72917 [7] NCCL INFO Comm config Blocking set to 1 +n116-041-181:72912:72912 [2] NCCL INFO cudaDriverVersion 12060 +n116-041-181:72912:72912 [2] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-041-181:72912:72912 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-041-181:72912:72912 [2] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-041-181:72912:72912 [2] NCCL INFO Comm config Blocking set to 1 +n116-041-181:72916:72916 [6] NCCL INFO cudaDriverVersion 12060 +n116-041-181:72916:72916 [6] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-041-181:72916:72916 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-041-181:72916:72916 [6] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-041-181:72911:72911 [1] NCCL INFO cudaDriverVersion 12060 +n116-041-181:72911:72911 [1] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-041-181:72911:72911 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-041-181:72911:72911 [1] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-041-181:72916:72916 [6] NCCL INFO Comm config Blocking set to 1 +n116-041-181:72911:72911 [1] NCCL INFO Comm config Blocking set to 1 +n116-041-181:72910:72910 [0] NCCL INFO cudaDriverVersion 12060 +n116-041-181:72910:72910 [0] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-041-181:72910:72910 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-041-181:72910:72910 [0] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-041-181:72913:72913 [3] NCCL INFO cudaDriverVersion 12060 +n116-041-181:72913:72913 [3] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-041-181:72913:72913 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-041-181:72913:72913 [3] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-041-181:72914:72914 [4] NCCL INFO cudaDriverVersion 12060 +n116-041-181:72914:72914 [4] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-041-181:72914:72914 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-041-181:72910:72910 [0] NCCL INFO Comm config Blocking set to 1 +n116-041-181:72913:72913 [3] NCCL INFO Comm config Blocking set to 1 +n116-041-181:72914:72914 [4] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-041-181:72914:72914 [4] NCCL INFO Comm config Blocking set to 1 +n116-041-181:72917:72967 [7] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-041-181:72917:72967 [7] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-041-181:72917:72967 [7] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-041-181:72917:72967 [7] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-041-181:72913:72972 [3] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-041-181:72912:72968 [2] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-041-181:72910:72971 [0] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-041-181:72917:72967 [7] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-041-181:72917:72967 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-041-181:72917:72967 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-041-181:72915:72966 [5] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-041-181:72917:72967 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-041-181:72913:72972 [3] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-041-181:72917:72967 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-041-181:72917:72967 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-041-181:72913:72972 [3] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-041-181:72910:72971 [0] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-041-181:72912:72968 [2] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-041-181:72913:72972 [3] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-041-181:72912:72968 [2] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-041-181:72910:72971 [0] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-041-181:72915:72966 [5] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-041-181:72912:72968 [2] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-041-181:72910:72971 [0] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-041-181:72913:72972 [3] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-041-181:72915:72966 [5] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-041-181:72915:72966 [5] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-041-181:72913:72972 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-041-181:72912:72968 [2] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-041-181:72913:72972 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-041-181:72912:72968 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-041-181:72910:72971 [0] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-041-181:72912:72968 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-041-181:72913:72972 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-041-181:72913:72972 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-041-181:72915:72966 [5] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-041-181:72912:72968 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-041-181:72910:72971 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-041-181:72913:72972 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-041-181:72915:72966 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-041-181:72910:72971 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-041-181:72912:72968 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-041-181:72915:72966 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-041-181:72912:72968 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-041-181:72910:72971 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-041-181:72915:72966 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-041-181:72910:72971 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-041-181:72915:72966 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-041-181:72910:72971 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-041-181:72915:72966 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-041-181:72911:72970 [1] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-041-181:72914:72973 [4] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-041-181:72911:72970 [1] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-041-181:72911:72970 [1] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-041-181:72911:72970 [1] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-041-181:72914:72973 [4] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-041-181:72916:72969 [6] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-041-181:72911:72970 [1] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-041-181:72914:72973 [4] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-041-181:72914:72973 [4] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-041-181:72911:72970 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-041-181:72911:72970 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-041-181:72916:72969 [6] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-041-181:72914:72973 [4] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-041-181:72914:72973 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-041-181:72916:72969 [6] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-041-181:72911:72970 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-041-181:72916:72969 [6] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-041-181:72911:72970 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-041-181:72914:72973 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-041-181:72911:72970 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-041-181:72914:72973 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-041-181:72914:72973 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-041-181:72916:72969 [6] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-041-181:72914:72973 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-041-181:72916:72969 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-041-181:72916:72969 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-041-181:72916:72969 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-041-181:72916:72969 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-041-181:72916:72969 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-041-181:72912:72968 [2] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-041-181:72912:72968 [2] NCCL INFO Using network FasTrak +n116-041-181:72914:72973 [4] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-041-181:72914:72973 [4] NCCL INFO Using network FasTrak +n116-041-181:72917:72967 [7] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-041-181:72917:72967 [7] NCCL INFO Using network FasTrak +n116-041-181:72910:72971 [0] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-041-181:72910:72971 [0] NCCL INFO Using network FasTrak +n116-041-181:72916:72969 [6] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-041-181:72916:72969 [6] NCCL INFO Using network FasTrak +n116-041-181:72915:72966 [5] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-041-181:72915:72966 [5] NCCL INFO Using network FasTrak +n116-041-181:72911:72970 [1] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-041-181:72911:72970 [1] NCCL INFO Using network FasTrak +n116-041-181:72913:72972 [3] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-041-181:72913:72972 [3] NCCL INFO Using network FasTrak +n116-041-181:72912:72968 [2] NCCL INFO DMA-BUF is available on GPU device 2 +n116-041-181:72912:72968 [2] NCCL INFO ncclCommInitRankConfig comm 0xbba68e0 rank 50 nranks 64 cudaDev 2 nvmlDev 2 busId b000 commId 0xfb2d11625d05007a - Init START +n116-041-181:72917:72967 [7] NCCL INFO DMA-BUF is available on GPU device 7 +n116-041-181:72915:72966 [5] NCCL INFO DMA-BUF is available on GPU device 5 +n116-041-181:72914:72973 [4] NCCL INFO DMA-BUF is available on GPU device 4 +n116-041-181:72910:72971 [0] NCCL INFO DMA-BUF is available on GPU device 0 +n116-041-181:72913:72972 [3] NCCL INFO DMA-BUF is available on GPU device 3 +n116-041-181:72915:72966 [5] NCCL INFO ncclCommInitRankConfig comm 0xc1893a0 rank 53 nranks 64 cudaDev 5 nvmlDev 5 busId 85000 commId 0xfb2d11625d05007a - Init START +n116-041-181:72914:72973 [4] NCCL INFO ncclCommInitRankConfig comm 0xbc34310 rank 52 nranks 64 cudaDev 4 nvmlDev 4 busId 84000 commId 0xfb2d11625d05007a - Init START +n116-041-181:72913:72972 [3] NCCL INFO ncclCommInitRankConfig comm 0xc145f50 rank 51 nranks 64 cudaDev 3 nvmlDev 3 busId c000 commId 0xfb2d11625d05007a - Init START +n116-041-181:72914:72973 [4] NCCL INFO RAS client listening socket at ::1<28028> +n116-041-181:72916:72969 [6] NCCL INFO DMA-BUF is available on GPU device 6 +n116-041-181:72917:72967 [7] NCCL INFO ncclCommInitRankConfig comm 0xbede0d0 rank 55 nranks 64 cudaDev 7 nvmlDev 7 busId 8c000 commId 0xfb2d11625d05007a - Init START +n116-041-181:72911:72970 [1] NCCL INFO DMA-BUF is available on GPU device 1 +n116-041-181:72913:72972 [3] NCCL INFO RAS client listening socket at ::1<28028> +n116-041-181:72910:72971 [0] NCCL INFO ncclCommInitRankConfig comm 0xc56ed00 rank 48 nranks 64 cudaDev 0 nvmlDev 0 busId 4000 commId 0xfb2d11625d05007a - Init START +n116-041-181:72916:72969 [6] NCCL INFO ncclCommInitRankConfig comm 0xbc00590 rank 54 nranks 64 cudaDev 6 nvmlDev 6 busId 8b000 commId 0xfb2d11625d05007a - Init START +n116-041-181:72915:72966 [5] NCCL INFO RAS client listening socket at ::1<28028> +n116-041-181:72916:72969 [6] NCCL INFO RAS client listening socket at ::1<28028> +n116-041-181:72917:72967 [7] NCCL INFO RAS client listening socket at ::1<28028> +n116-041-181:72911:72970 [1] NCCL INFO ncclCommInitRankConfig comm 0xc348340 rank 49 nranks 64 cudaDev 1 nvmlDev 1 busId 5000 commId 0xfb2d11625d05007a - Init START +n116-041-181:72910:72971 [0] NCCL INFO RAS client listening socket at ::1<28028> +n116-041-181:72911:72970 [1] NCCL INFO RAS client listening socket at ::1<28028> +n116-041-181:72912:72968 [2] NCCL INFO RAS client listening socket at ::1<28028> +n116-041-181:72917:72967 [7] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-041-181:72917:72967 [7] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-041-181:72917:72967 [7] NCCL INFO Retrieving state for FasTrak +n116-041-181:72917:72967 [7] NCCL INFO Initialized state 0 for FasTrak +n116-041-181:72917:72967 [7] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-041-181:72917:72967 [7] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-041-181:72917:72967 [7] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-041-181:72917:72967 [7] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-041-181:72917:72967 [7] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-041-181:72917:72967 [7] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-041-181:72917:72967 [7] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-041-181:72917:72967 [7] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-041-181:72911:72970 [1] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-041-181:72911:72970 [1] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-041-181:72911:72970 [1] NCCL INFO Retrieving state for FasTrak +n116-041-181:72911:72970 [1] NCCL INFO Initialized state 0 for FasTrak +n116-041-181:72911:72970 [1] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-041-181:72911:72970 [1] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-041-181:72911:72970 [1] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-041-181:72911:72970 [1] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-041-181:72911:72970 [1] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-041-181:72911:72970 [1] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-041-181:72911:72970 [1] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-041-181:72911:72970 [1] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-041-181:72915:72966 [5] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-041-181:72915:72966 [5] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-041-181:72915:72966 [5] NCCL INFO Retrieving state for FasTrak +n116-041-181:72915:72966 [5] NCCL INFO Initialized state 0 for FasTrak +n116-041-181:72915:72966 [5] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-041-181:72915:72966 [5] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-041-181:72915:72966 [5] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-041-181:72915:72966 [5] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-041-181:72915:72966 [5] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-041-181:72915:72966 [5] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-041-181:72915:72966 [5] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-041-181:72915:72966 [5] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-041-181:72916:72969 [6] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-041-181:72916:72969 [6] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-041-181:72916:72969 [6] NCCL INFO Retrieving state for FasTrak +n116-041-181:72916:72969 [6] NCCL INFO Initialized state 0 for FasTrak +n116-041-181:72916:72969 [6] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-041-181:72916:72969 [6] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-041-181:72916:72969 [6] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-041-181:72916:72969 [6] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-041-181:72914:72973 [4] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-041-181:72914:72973 [4] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-041-181:72914:72973 [4] NCCL INFO Retrieving state for FasTrak +n116-041-181:72914:72973 [4] NCCL INFO Initialized state 0 for FasTrak +n116-041-181:72916:72969 [6] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-041-181:72916:72969 [6] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-041-181:72914:72973 [4] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-041-181:72916:72969 [6] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-041-181:72914:72973 [4] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-041-181:72916:72969 [6] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-041-181:72914:72973 [4] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-041-181:72912:72968 [2] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-041-181:72912:72968 [2] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-041-181:72912:72968 [2] NCCL INFO Retrieving state for FasTrak +n116-041-181:72912:72968 [2] NCCL INFO Initialized state 0 for FasTrak +n116-041-181:72914:72973 [4] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-041-181:72914:72973 [4] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-041-181:72914:72973 [4] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-041-181:72912:72968 [2] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-041-181:72912:72968 [2] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-041-181:72914:72973 [4] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-041-181:72912:72968 [2] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-041-181:72914:72973 [4] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-041-181:72912:72968 [2] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-041-181:72913:72972 [3] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-041-181:72913:72972 [3] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-041-181:72913:72972 [3] NCCL INFO Retrieving state for FasTrak +n116-041-181:72913:72972 [3] NCCL INFO Initialized state 0 for FasTrak +n116-041-181:72912:72968 [2] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-041-181:72912:72968 [2] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-041-181:72913:72972 [3] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-041-181:72913:72972 [3] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-041-181:72912:72968 [2] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-041-181:72913:72972 [3] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-041-181:72912:72968 [2] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-041-181:72913:72972 [3] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-041-181:72913:72972 [3] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-041-181:72913:72972 [3] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-041-181:72913:72972 [3] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-041-181:72913:72972 [3] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-041-181:72910:72971 [0] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-041-181:72910:72971 [0] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-041-181:72910:72971 [0] NCCL INFO Retrieving state for FasTrak +n116-041-181:72910:72971 [0] NCCL INFO Initialized state 0 for FasTrak +n116-041-181:72910:72971 [0] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-041-181:72910:72971 [0] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-041-181:72910:72971 [0] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-041-181:72910:72971 [0] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-041-181:72910:72971 [0] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-041-181:72910:72971 [0] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-041-181:72910:72971 [0] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-041-181:72910:72971 [0] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-041-181:72910:72971 [0] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-041-181:72915:72966 [5] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-041-181:72911:72970 [1] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-041-181:72913:72972 [3] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-041-181:72912:72968 [2] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-041-181:72914:72973 [4] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-041-181:72916:72969 [6] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-041-181:72917:72967 [7] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-041-181:72911:72970 [1] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-041-181:72911:72970 [1] NCCL INFO CPU/0-0 (1/1/3) +n116-041-181:72911:72970 [1] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-041-181:72911:72970 [1] NCCL INFO + PCI[24.0] - GPU/0-4000 (48) +n116-041-181:72911:72970 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72911:72970 [1] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-041-181:72911:72970 [1] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-041-181:72911:72970 [1] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-041-181:72911:72970 [1] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-041-181:72911:72970 [1] NCCL INFO + PCI[24.0] - GPU/0-5000 (49) +n116-041-181:72911:72970 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72911:72970 [1] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-041-181:72911:72970 [1] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-041-181:72911:72970 [1] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-041-181:72911:72970 [1] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-041-181:72911:72970 [1] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-041-181:72911:72970 [1] NCCL INFO + PCI[24.0] - GPU/0-b000 (50) +n116-041-181:72911:72970 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72911:72970 [1] NCCL INFO + PCI[24.0] - GPU/0-c000 (51) +n116-041-181:72911:72970 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72911:72970 [1] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-041-181:72911:72970 [1] NCCL INFO CPU/0-1 (1/1/3) +n116-041-181:72911:72970 [1] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-041-181:72911:72970 [1] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-041-181:72911:72970 [1] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-041-181:72911:72970 [1] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-041-181:72911:72970 [1] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-041-181:72911:72970 [1] NCCL INFO + PCI[24.0] - GPU/0-84000 (52) +n116-041-181:72911:72970 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72911:72970 [1] NCCL INFO + PCI[24.0] - GPU/0-85000 (53) +n116-041-181:72911:72970 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72911:72970 [1] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-041-181:72911:72970 [1] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-041-181:72911:72970 [1] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-041-181:72911:72970 [1] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-041-181:72911:72970 [1] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-041-181:72911:72970 [1] NCCL INFO + PCI[24.0] - GPU/0-8b000 (54) +n116-041-181:72911:72970 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72911:72970 [1] NCCL INFO + PCI[24.0] - GPU/0-8c000 (55) +n116-041-181:72911:72970 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72911:72970 [1] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-041-181:72911:72970 [1] NCCL INFO ========================================== +n116-041-181:72911:72970 [1] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-181:72911:72970 [1] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-181:72911:72970 [1] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-181:72911:72970 [1] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-181:72911:72970 [1] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-181:72915:72966 [5] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-041-181:72911:72970 [1] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-181:72915:72966 [5] NCCL INFO CPU/0-0 (1/1/3) +n116-041-181:72915:72966 [5] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-041-181:72915:72966 [5] NCCL INFO + PCI[24.0] - GPU/0-4000 (48) +n116-041-181:72911:72970 [1] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-181:72915:72966 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72915:72966 [5] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-041-181:72915:72966 [5] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-041-181:72915:72966 [5] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-041-181:72915:72966 [5] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-041-181:72915:72966 [5] NCCL INFO + PCI[24.0] - GPU/0-5000 (49) +n116-041-181:72911:72970 [1] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-181:72915:72966 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72915:72966 [5] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-041-181:72915:72966 [5] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-041-181:72915:72966 [5] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-041-181:72915:72966 [5] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-041-181:72915:72966 [5] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-041-181:72915:72966 [5] NCCL INFO + PCI[24.0] - GPU/0-b000 (50) +n116-041-181:72911:72970 [1] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-181:72915:72966 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72915:72966 [5] NCCL INFO + PCI[24.0] - GPU/0-c000 (51) +n116-041-181:72913:72972 [3] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-041-181:72915:72966 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72912:72968 [2] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-041-181:72915:72966 [5] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-041-181:72913:72972 [3] NCCL INFO CPU/0-0 (1/1/3) +n116-041-181:72915:72966 [5] NCCL INFO CPU/0-1 (1/1/3) +n116-041-181:72911:72970 [1] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-181:72912:72968 [2] NCCL INFO CPU/0-0 (1/1/3) +n116-041-181:72915:72966 [5] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-041-181:72913:72972 [3] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-041-181:72915:72966 [5] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-041-181:72912:72968 [2] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-041-181:72912:72968 [2] NCCL INFO + PCI[24.0] - GPU/0-4000 (48) +n116-041-181:72913:72972 [3] NCCL INFO + PCI[24.0] - GPU/0-4000 (48) +n116-041-181:72915:72966 [5] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-041-181:72912:72968 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72911:72970 [1] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-181:72913:72972 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72913:72972 [3] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-041-181:72915:72966 [5] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-041-181:72912:72968 [2] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-041-181:72915:72966 [5] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-041-181:72913:72972 [3] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-041-181:72912:72968 [2] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-041-181:72915:72966 [5] NCCL INFO + PCI[24.0] - GPU/0-84000 (52) +n116-041-181:72912:72968 [2] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-041-181:72913:72972 [3] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-041-181:72911:72970 [1] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-181:72915:72966 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72912:72968 [2] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-041-181:72913:72972 [3] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-041-181:72913:72972 [3] NCCL INFO + PCI[24.0] - GPU/0-5000 (49) +n116-041-181:72912:72968 [2] NCCL INFO + PCI[24.0] - GPU/0-5000 (49) +n116-041-181:72915:72966 [5] NCCL INFO + PCI[24.0] - GPU/0-85000 (53) +n116-041-181:72911:72970 [1] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-181:72913:72972 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72915:72966 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72910:72971 [0] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-041-181:72913:72972 [3] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-041-181:72912:72968 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72915:72966 [5] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-041-181:72913:72972 [3] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-041-181:72911:72970 [1] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-181:72910:72971 [0] NCCL INFO CPU/0-0 (1/1/3) +n116-041-181:72912:72968 [2] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-041-181:72910:72971 [0] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-041-181:72915:72966 [5] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-041-181:72913:72972 [3] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-041-181:72912:72968 [2] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-041-181:72910:72971 [0] NCCL INFO + PCI[24.0] - GPU/0-4000 (48) +n116-041-181:72915:72966 [5] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-041-181:72913:72972 [3] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-041-181:72915:72966 [5] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-041-181:72911:72970 [1] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-041-181:72912:72968 [2] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-041-181:72910:72971 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72913:72972 [3] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-041-181:72915:72966 [5] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-041-181:72913:72972 [3] NCCL INFO + PCI[24.0] - GPU/0-b000 (50) +n116-041-181:72912:72968 [2] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-041-181:72910:72971 [0] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-041-181:72915:72966 [5] NCCL INFO + PCI[24.0] - GPU/0-8b000 (54) +n116-041-181:72912:72968 [2] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-041-181:72913:72972 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72911:72970 [1] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-041-181:72910:72971 [0] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-041-181:72915:72966 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72912:72968 [2] NCCL INFO + PCI[24.0] - GPU/0-b000 (50) +n116-041-181:72913:72972 [3] NCCL INFO + PCI[24.0] - GPU/0-c000 (51) +n116-041-181:72910:72971 [0] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-041-181:72915:72966 [5] NCCL INFO + PCI[24.0] - GPU/0-8c000 (55) +n116-041-181:72912:72968 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72913:72972 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72915:72966 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72910:72971 [0] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-041-181:72912:72968 [2] NCCL INFO + PCI[24.0] - GPU/0-c000 (51) +n116-041-181:72913:72972 [3] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-041-181:72915:72966 [5] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-041-181:72910:72971 [0] NCCL INFO + PCI[24.0] - GPU/0-5000 (49) +n116-041-181:72912:72968 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72911:72970 [1] NCCL INFO Setting affinity for GPU 1 to 0fffffff,fffffe00,00000000,000fffff,fffffffe +n116-041-181:72913:72972 [3] NCCL INFO CPU/0-1 (1/1/3) +n116-041-181:72912:72968 [2] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-041-181:72915:72966 [5] NCCL INFO ========================================== +n116-041-181:72910:72971 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72913:72972 [3] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-041-181:72912:72968 [2] NCCL INFO CPU/0-1 (1/1/3) +n116-041-181:72910:72971 [0] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-041-181:72913:72972 [3] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-041-181:72912:72968 [2] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-041-181:72910:72971 [0] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-041-181:72913:72972 [3] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-041-181:72912:72968 [2] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-041-181:72910:72971 [0] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-041-181:72912:72968 [2] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-041-181:72913:72972 [3] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-041-181:72915:72966 [5] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-181:72910:72971 [0] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-041-181:72912:72968 [2] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-041-181:72913:72972 [3] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-041-181:72910:72971 [0] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-041-181:72912:72968 [2] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-041-181:72913:72972 [3] NCCL INFO + PCI[24.0] - GPU/0-84000 (52) +n116-041-181:72910:72971 [0] NCCL INFO + PCI[24.0] - GPU/0-b000 (50) +n116-041-181:72910:72971 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72912:72968 [2] NCCL INFO + PCI[24.0] - GPU/0-84000 (52) +n116-041-181:72913:72972 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72912:72968 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72915:72966 [5] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-181:72912:72968 [2] NCCL INFO + PCI[24.0] - GPU/0-85000 (53) +n116-041-181:72910:72971 [0] NCCL INFO + PCI[24.0] - GPU/0-c000 (51) +n116-041-181:72913:72972 [3] NCCL INFO + PCI[24.0] - GPU/0-85000 (53) +n116-041-181:72912:72968 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72910:72971 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72913:72972 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72912:72968 [2] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-041-181:72910:72971 [0] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-041-181:72915:72966 [5] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-181:72912:72968 [2] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-041-181:72913:72972 [3] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-041-181:72910:72971 [0] NCCL INFO CPU/0-1 (1/1/3) +n116-041-181:72911:72970 [1] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-041-181:72912:72968 [2] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-041-181:72913:72972 [3] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-041-181:72910:72971 [0] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-041-181:72912:72968 [2] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-041-181:72910:72971 [0] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-041-181:72913:72972 [3] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-041-181:72912:72968 [2] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-041-181:72915:72966 [5] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-181:72912:72968 [2] NCCL INFO + PCI[24.0] - GPU/0-8b000 (54) +n116-041-181:72910:72971 [0] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-041-181:72913:72972 [3] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-041-181:72913:72972 [3] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-041-181:72912:72968 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72910:72971 [0] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-041-181:72912:72968 [2] NCCL INFO + PCI[24.0] - GPU/0-8c000 (55) +n116-041-181:72915:72966 [5] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-181:72913:72972 [3] NCCL INFO + PCI[24.0] - GPU/0-8b000 (54) +n116-041-181:72910:72971 [0] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-041-181:72912:72968 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72913:72972 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72910:72971 [0] NCCL INFO + PCI[24.0] - GPU/0-84000 (52) +n116-041-181:72912:72968 [2] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-041-181:72913:72972 [3] NCCL INFO + PCI[24.0] - GPU/0-8c000 (55) +n116-041-181:72912:72968 [2] NCCL INFO ========================================== +n116-041-181:72910:72971 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72915:72966 [5] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-181:72913:72972 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72910:72971 [0] NCCL INFO + PCI[24.0] - GPU/0-85000 (53) +n116-041-181:72910:72971 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72913:72972 [3] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-041-181:72913:72972 [3] NCCL INFO ========================================== +n116-041-181:72910:72971 [0] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-041-181:72912:72968 [2] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-181:72910:72971 [0] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-041-181:72915:72966 [5] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-181:72910:72971 [0] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-041-181:72910:72971 [0] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-041-181:72913:72972 [3] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-181:72910:72971 [0] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-041-181:72912:72968 [2] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-181:72910:72971 [0] NCCL INFO + PCI[24.0] - GPU/0-8b000 (54) +n116-041-181:72915:72966 [5] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-181:72910:72971 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72910:72971 [0] NCCL INFO + PCI[24.0] - GPU/0-8c000 (55) +n116-041-181:72911:72970 [1] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-041-181:72913:72972 [3] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-181:72912:72968 [2] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-181:72910:72971 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72915:72966 [5] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-181:72910:72971 [0] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-041-181:72910:72971 [0] NCCL INFO ========================================== +n116-041-181:72913:72972 [3] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-181:72912:72968 [2] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-181:72915:72966 [5] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-181:72910:72971 [0] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-181:72913:72972 [3] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-181:72912:72968 [2] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-181:72915:72966 [5] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-181:72910:72971 [0] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-181:72913:72972 [3] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-181:72915:72966 [5] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-181:72916:72969 [6] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-041-181:72910:72971 [0] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-181:72914:72973 [4] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-041-181:72913:72972 [3] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-181:72916:72969 [6] NCCL INFO CPU/0-0 (1/1/3) +n116-041-181:72912:72968 [2] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-181:72914:72973 [4] NCCL INFO CPU/0-0 (1/1/3) +n116-041-181:72916:72969 [6] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-041-181:72915:72966 [5] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-181:72916:72969 [6] NCCL INFO + PCI[24.0] - GPU/0-4000 (48) +n116-041-181:72914:72973 [4] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-041-181:72910:72971 [0] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-181:72913:72972 [3] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-181:72912:72968 [2] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-181:72916:72969 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72914:72973 [4] NCCL INFO + PCI[24.0] - GPU/0-4000 (48) +n116-041-181:72916:72969 [6] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-041-181:72914:72973 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72916:72969 [6] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-041-181:72916:72969 [6] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-041-181:72914:72973 [4] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-041-181:72915:72966 [5] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-181:72916:72969 [6] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-041-181:72910:72971 [0] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-181:72916:72969 [6] NCCL INFO + PCI[24.0] - GPU/0-5000 (49) +n116-041-181:72914:72973 [4] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-041-181:72913:72972 [3] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-181:72912:72968 [2] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-181:72916:72969 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72914:72973 [4] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-041-181:72916:72969 [6] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-041-181:72915:72966 [5] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-041-181:72916:72969 [6] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-041-181:72914:72973 [4] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-041-181:72910:72971 [0] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-181:72916:72969 [6] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-041-181:72914:72973 [4] NCCL INFO + PCI[24.0] - GPU/0-5000 (49) +n116-041-181:72916:72969 [6] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-041-181:72913:72972 [3] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-181:72912:72968 [2] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-181:72914:72973 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72916:72969 [6] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-041-181:72916:72969 [6] NCCL INFO + PCI[24.0] - GPU/0-b000 (50) +n116-041-181:72915:72966 [5] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-041-181:72914:72973 [4] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-041-181:72916:72969 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72910:72971 [0] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-181:72916:72969 [6] NCCL INFO + PCI[24.0] - GPU/0-c000 (51) +n116-041-181:72914:72973 [4] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-041-181:72913:72972 [3] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-181:72916:72969 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72912:72968 [2] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-181:72916:72969 [6] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-041-181:72914:72973 [4] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-041-181:72916:72969 [6] NCCL INFO CPU/0-1 (1/1/3) +n116-041-181:72914:72973 [4] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-041-181:72916:72969 [6] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-041-181:72916:72969 [6] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-041-181:72914:72973 [4] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-041-181:72910:72971 [0] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-181:72916:72969 [6] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-041-181:72915:72966 [5] NCCL INFO Setting affinity for GPU 5 to 7fff,ffffffff,e0000000,0000007f,ffffffff,ffe00000,00000000 +n116-041-181:72914:72973 [4] NCCL INFO + PCI[24.0] - GPU/0-b000 (50) +n116-041-181:72910:72971 [0] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-181:72913:72972 [3] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-181:72912:72968 [2] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-181:72916:72969 [6] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-041-181:72914:72973 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72916:72969 [6] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-041-181:72916:72969 [6] NCCL INFO + PCI[24.0] - GPU/0-84000 (52) +n116-041-181:72914:72973 [4] NCCL INFO + PCI[24.0] - GPU/0-c000 (51) +n116-041-181:72916:72969 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72914:72973 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72916:72969 [6] NCCL INFO + PCI[24.0] - GPU/0-85000 (53) +n116-041-181:72915:72966 [5] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-041-181:72910:72971 [0] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-181:72913:72972 [3] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-181:72912:72968 [2] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-181:72914:72973 [4] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-041-181:72916:72969 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72916:72969 [6] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-041-181:72914:72973 [4] NCCL INFO CPU/0-1 (1/1/3) +n116-041-181:72916:72969 [6] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-041-181:72916:72969 [6] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-041-181:72914:72973 [4] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-041-181:72910:72971 [0] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-181:72916:72969 [6] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-041-181:72913:72972 [3] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-181:72912:72968 [2] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-181:72914:72973 [4] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-041-181:72916:72969 [6] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-041-181:72916:72969 [6] NCCL INFO + PCI[24.0] - GPU/0-8b000 (54) +n116-041-181:72916:72969 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72914:72973 [4] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-041-181:72916:72969 [6] NCCL INFO + PCI[24.0] - GPU/0-8c000 (55) +n116-041-181:72914:72973 [4] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-041-181:72910:72971 [0] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-181:72916:72969 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72913:72972 [3] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-181:72912:72968 [2] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-181:72914:72973 [4] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-041-181:72916:72969 [6] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-041-181:72916:72969 [6] NCCL INFO ========================================== +n116-041-181:72914:72973 [4] NCCL INFO + PCI[24.0] - GPU/0-84000 (52) +n116-041-181:72914:72973 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72910:72971 [0] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-181:72914:72973 [4] NCCL INFO + PCI[24.0] - GPU/0-85000 (53) +n116-041-181:72913:72972 [3] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-041-181:72912:72968 [2] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-041-181:72914:72973 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72914:72973 [4] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-041-181:72916:72969 [6] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-181:72914:72973 [4] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-041-181:72910:72971 [0] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-181:72913:72972 [3] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-041-181:72914:72973 [4] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-041-181:72912:72968 [2] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-041-181:72914:72973 [4] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-041-181:72915:72966 [5] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-041-181:72916:72969 [6] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-181:72914:72973 [4] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-041-181:72910:72971 [0] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-041-181:72914:72973 [4] NCCL INFO + PCI[24.0] - GPU/0-8b000 (54) +n116-041-181:72914:72973 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72914:72973 [4] NCCL INFO + PCI[24.0] - GPU/0-8c000 (55) +n116-041-181:72913:72972 [3] NCCL INFO Setting affinity for GPU 3 to 0fffffff,fffffe00,00000000,000fffff,fffffffe +n116-041-181:72912:72968 [2] NCCL INFO Setting affinity for GPU 2 to 0fffffff,fffffe00,00000000,000fffff,fffffffe +n116-041-181:72916:72969 [6] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-181:72910:72971 [0] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-041-181:72914:72973 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72914:72973 [4] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-041-181:72914:72973 [4] NCCL INFO ========================================== +n116-041-181:72912:72968 [2] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-041-181:72913:72972 [3] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-041-181:72916:72969 [6] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-181:72910:72971 [0] NCCL INFO Setting affinity for GPU 0 to 0fffffff,fffffe00,00000000,000fffff,fffffffe +n116-041-181:72916:72969 [6] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-181:72910:72971 [0] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-041-181:72914:72973 [4] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-181:72916:72969 [6] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-181:72914:72973 [4] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-181:72916:72969 [6] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-181:72912:72968 [2] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-041-181:72913:72972 [3] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-041-181:72910:72971 [0] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-041-181:72916:72969 [6] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-181:72914:72973 [4] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-181:72916:72969 [6] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-181:72916:72969 [6] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-181:72914:72973 [4] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-181:72916:72969 [6] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-181:72916:72969 [6] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-181:72914:72973 [4] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-181:72916:72969 [6] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-181:72916:72969 [6] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-181:72914:72973 [4] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-181:72916:72969 [6] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-041-181:72916:72969 [6] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-041-181:72914:72973 [4] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-181:72916:72969 [6] NCCL INFO Setting affinity for GPU 6 to 7fff,ffffffff,e0000000,0000007f,ffffffff,ffe00000,00000000 +n116-041-181:72917:72967 [7] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-041-181:72914:72973 [4] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-181:72917:72967 [7] NCCL INFO CPU/0-0 (1/1/3) +n116-041-181:72917:72967 [7] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-041-181:72917:72967 [7] NCCL INFO + PCI[24.0] - GPU/0-4000 (48) +n116-041-181:72917:72967 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72917:72967 [7] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-041-181:72914:72973 [4] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-181:72917:72967 [7] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-041-181:72917:72967 [7] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-041-181:72917:72967 [7] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-041-181:72917:72967 [7] NCCL INFO + PCI[24.0] - GPU/0-5000 (49) +n116-041-181:72917:72967 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72917:72967 [7] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-041-181:72917:72967 [7] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-041-181:72917:72967 [7] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-041-181:72917:72967 [7] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-041-181:72917:72967 [7] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-041-181:72917:72967 [7] NCCL INFO + PCI[24.0] - GPU/0-b000 (50) +n116-041-181:72917:72967 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72917:72967 [7] NCCL INFO + PCI[24.0] - GPU/0-c000 (51) +n116-041-181:72917:72967 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72917:72967 [7] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-041-181:72914:72973 [4] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-181:72917:72967 [7] NCCL INFO CPU/0-1 (1/1/3) +n116-041-181:72917:72967 [7] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-041-181:72917:72967 [7] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-041-181:72917:72967 [7] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-041-181:72917:72967 [7] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-041-181:72917:72967 [7] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-041-181:72916:72969 [6] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-041-181:72917:72967 [7] NCCL INFO + PCI[24.0] - GPU/0-84000 (52) +n116-041-181:72914:72973 [4] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-181:72917:72967 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72917:72967 [7] NCCL INFO + PCI[24.0] - GPU/0-85000 (53) +n116-041-181:72917:72967 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72917:72967 [7] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-041-181:72917:72967 [7] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-041-181:72917:72967 [7] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-041-181:72914:72973 [4] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-181:72917:72967 [7] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-041-181:72917:72967 [7] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-041-181:72917:72967 [7] NCCL INFO + PCI[24.0] - GPU/0-8b000 (54) +n116-041-181:72917:72967 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72917:72967 [7] NCCL INFO + PCI[24.0] - GPU/0-8c000 (55) +n116-041-181:72917:72967 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-041-181:72917:72967 [7] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-041-181:72917:72967 [7] NCCL INFO ========================================== +n116-041-181:72914:72973 [4] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-181:72917:72967 [7] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-181:72914:72973 [4] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-181:72917:72967 [7] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-181:72917:72967 [7] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-181:72914:72973 [4] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-041-181:72917:72967 [7] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-181:72914:72973 [4] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-041-181:72917:72967 [7] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-181:72914:72973 [4] NCCL INFO Setting affinity for GPU 4 to 7fff,ffffffff,e0000000,0000007f,ffffffff,ffe00000,00000000 +n116-041-181:72916:72969 [6] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-041-181:72917:72967 [7] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-041-181:72917:72967 [7] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-181:72917:72967 [7] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-041-181:72917:72967 [7] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-181:72917:72967 [7] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-181:72917:72967 [7] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-181:72917:72967 [7] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-041-181:72917:72967 [7] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-181:72917:72967 [7] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-041-181:72917:72967 [7] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-041-181:72917:72967 [7] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-041-181:72917:72967 [7] NCCL INFO Setting affinity for GPU 7 to 7fff,ffffffff,e0000000,0000007f,ffffffff,ffe00000,00000000 +n116-041-181:72914:72973 [4] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-041-181:72917:72967 [7] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-041-181:72914:72973 [4] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-041-181:72917:72967 [7] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-041-181:72912:72968 [2] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-181:72912:72968 [2] NCCL INFO 0 : NET/0-0 GPU/48 GPU/55 GPU/54 GPU/53 GPU/52 GPU/51 GPU/50 GPU/49 NET/0-0 +n116-041-181:72912:72968 [2] NCCL INFO 1 : NET/0-1 GPU/49 GPU/50 GPU/51 GPU/52 GPU/53 GPU/54 GPU/55 GPU/48 NET/0-1 +n116-041-181:72912:72968 [2] NCCL INFO 2 : NET/0-2 GPU/50 GPU/49 GPU/48 GPU/55 GPU/54 GPU/53 GPU/52 GPU/51 NET/0-2 +n116-041-181:72912:72968 [2] NCCL INFO 3 : NET/0-3 GPU/51 GPU/52 GPU/53 GPU/54 GPU/55 GPU/48 GPU/49 GPU/50 NET/0-3 +n116-041-181:72912:72968 [2] NCCL INFO 4 : NET/0-4 GPU/52 GPU/51 GPU/50 GPU/49 GPU/48 GPU/55 GPU/54 GPU/53 NET/0-4 +n116-041-181:72912:72968 [2] NCCL INFO 5 : NET/0-5 GPU/53 GPU/54 GPU/55 GPU/48 GPU/49 GPU/50 GPU/51 GPU/52 NET/0-5 +n116-041-181:72912:72968 [2] NCCL INFO 6 : NET/0-6 GPU/54 GPU/53 GPU/52 GPU/51 GPU/50 GPU/49 GPU/48 GPU/55 NET/0-6 +n116-041-181:72912:72968 [2] NCCL INFO 7 : NET/0-7 GPU/55 GPU/48 GPU/49 GPU/50 GPU/51 GPU/52 GPU/53 GPU/54 NET/0-7 +n116-041-181:72910:72971 [0] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-181:72910:72971 [0] NCCL INFO 0 : NET/0-0 GPU/48 GPU/55 GPU/54 GPU/53 GPU/52 GPU/51 GPU/50 GPU/49 NET/0-0 +n116-041-181:72910:72971 [0] NCCL INFO 1 : NET/0-1 GPU/49 GPU/50 GPU/51 GPU/52 GPU/53 GPU/54 GPU/55 GPU/48 NET/0-1 +n116-041-181:72910:72971 [0] NCCL INFO 2 : NET/0-2 GPU/50 GPU/49 GPU/48 GPU/55 GPU/54 GPU/53 GPU/52 GPU/51 NET/0-2 +n116-041-181:72910:72971 [0] NCCL INFO 3 : NET/0-3 GPU/51 GPU/52 GPU/53 GPU/54 GPU/55 GPU/48 GPU/49 GPU/50 NET/0-3 +n116-041-181:72910:72971 [0] NCCL INFO 4 : NET/0-4 GPU/52 GPU/51 GPU/50 GPU/49 GPU/48 GPU/55 GPU/54 GPU/53 NET/0-4 +n116-041-181:72910:72971 [0] NCCL INFO 5 : NET/0-5 GPU/53 GPU/54 GPU/55 GPU/48 GPU/49 GPU/50 GPU/51 GPU/52 NET/0-5 +n116-041-181:72910:72971 [0] NCCL INFO 6 : NET/0-6 GPU/54 GPU/53 GPU/52 GPU/51 GPU/50 GPU/49 GPU/48 GPU/55 NET/0-6 +n116-041-181:72910:72971 [0] NCCL INFO 7 : NET/0-7 GPU/55 GPU/48 GPU/49 GPU/50 GPU/51 GPU/52 GPU/53 GPU/54 NET/0-7 +n116-041-181:72912:72968 [2] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-181:72912:72968 [2] NCCL INFO 0 : NET/0-0 GPU/48 GPU/49 GPU/50 GPU/51 GPU/52 GPU/53 GPU/54 GPU/55 NET/0-0 +n116-041-181:72912:72968 [2] NCCL INFO 1 : NET/0-1 GPU/49 GPU/48 GPU/55 GPU/54 GPU/53 GPU/52 GPU/51 GPU/50 NET/0-1 +n116-041-181:72912:72968 [2] NCCL INFO 2 : NET/0-2 GPU/50 GPU/51 GPU/52 GPU/53 GPU/54 GPU/55 GPU/48 GPU/49 NET/0-2 +n116-041-181:72912:72968 [2] NCCL INFO 3 : NET/0-3 GPU/51 GPU/50 GPU/49 GPU/48 GPU/55 GPU/54 GPU/53 GPU/52 NET/0-3 +n116-041-181:72912:72968 [2] NCCL INFO 4 : NET/0-4 GPU/52 GPU/53 GPU/54 GPU/55 GPU/48 GPU/49 GPU/50 GPU/51 NET/0-4 +n116-041-181:72912:72968 [2] NCCL INFO 5 : NET/0-5 GPU/53 GPU/52 GPU/51 GPU/50 GPU/49 GPU/48 GPU/55 GPU/54 NET/0-5 +n116-041-181:72912:72968 [2] NCCL INFO 6 : NET/0-6 GPU/54 GPU/55 GPU/48 GPU/49 GPU/50 GPU/51 GPU/52 GPU/53 NET/0-6 +n116-041-181:72912:72968 [2] NCCL INFO 7 : NET/0-7 GPU/55 GPU/54 GPU/53 GPU/52 GPU/51 GPU/50 GPU/49 GPU/48 NET/0-7 +n116-041-181:72910:72971 [0] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-181:72910:72971 [0] NCCL INFO 0 : NET/0-0 GPU/48 GPU/49 GPU/50 GPU/51 GPU/52 GPU/53 GPU/54 GPU/55 NET/0-0 +n116-041-181:72910:72971 [0] NCCL INFO 1 : NET/0-1 GPU/49 GPU/48 GPU/55 GPU/54 GPU/53 GPU/52 GPU/51 GPU/50 NET/0-1 +n116-041-181:72910:72971 [0] NCCL INFO 2 : NET/0-2 GPU/50 GPU/51 GPU/52 GPU/53 GPU/54 GPU/55 GPU/48 GPU/49 NET/0-2 +n116-041-181:72910:72971 [0] NCCL INFO 3 : NET/0-3 GPU/51 GPU/50 GPU/49 GPU/48 GPU/55 GPU/54 GPU/53 GPU/52 NET/0-3 +n116-041-181:72910:72971 [0] NCCL INFO 4 : NET/0-4 GPU/52 GPU/53 GPU/54 GPU/55 GPU/48 GPU/49 GPU/50 GPU/51 NET/0-4 +n116-041-181:72910:72971 [0] NCCL INFO 5 : NET/0-5 GPU/53 GPU/52 GPU/51 GPU/50 GPU/49 GPU/48 GPU/55 GPU/54 NET/0-5 +n116-041-181:72910:72971 [0] NCCL INFO 6 : NET/0-6 GPU/54 GPU/55 GPU/48 GPU/49 GPU/50 GPU/51 GPU/52 GPU/53 NET/0-6 +n116-041-181:72910:72971 [0] NCCL INFO 7 : NET/0-7 GPU/55 GPU/54 GPU/53 GPU/52 GPU/51 GPU/50 GPU/49 GPU/48 NET/0-7 +n116-041-181:72913:72972 [3] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-181:72913:72972 [3] NCCL INFO 0 : NET/0-0 GPU/48 GPU/55 GPU/54 GPU/53 GPU/52 GPU/51 GPU/50 GPU/49 NET/0-0 +n116-041-181:72913:72972 [3] NCCL INFO 1 : NET/0-1 GPU/49 GPU/50 GPU/51 GPU/52 GPU/53 GPU/54 GPU/55 GPU/48 NET/0-1 +n116-041-181:72913:72972 [3] NCCL INFO 2 : NET/0-2 GPU/50 GPU/49 GPU/48 GPU/55 GPU/54 GPU/53 GPU/52 GPU/51 NET/0-2 +n116-041-181:72913:72972 [3] NCCL INFO 3 : NET/0-3 GPU/51 GPU/52 GPU/53 GPU/54 GPU/55 GPU/48 GPU/49 GPU/50 NET/0-3 +n116-041-181:72913:72972 [3] NCCL INFO 4 : NET/0-4 GPU/52 GPU/51 GPU/50 GPU/49 GPU/48 GPU/55 GPU/54 GPU/53 NET/0-4 +n116-041-181:72913:72972 [3] NCCL INFO 5 : NET/0-5 GPU/53 GPU/54 GPU/55 GPU/48 GPU/49 GPU/50 GPU/51 GPU/52 NET/0-5 +n116-041-181:72913:72972 [3] NCCL INFO 6 : NET/0-6 GPU/54 GPU/53 GPU/52 GPU/51 GPU/50 GPU/49 GPU/48 GPU/55 NET/0-6 +n116-041-181:72913:72972 [3] NCCL INFO 7 : NET/0-7 GPU/55 GPU/48 GPU/49 GPU/50 GPU/51 GPU/52 GPU/53 GPU/54 NET/0-7 +n116-041-181:72913:72972 [3] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-181:72913:72972 [3] NCCL INFO 0 : NET/0-0 GPU/48 GPU/49 GPU/50 GPU/51 GPU/52 GPU/53 GPU/54 GPU/55 NET/0-0 +n116-041-181:72913:72972 [3] NCCL INFO 1 : NET/0-1 GPU/49 GPU/48 GPU/55 GPU/54 GPU/53 GPU/52 GPU/51 GPU/50 NET/0-1 +n116-041-181:72913:72972 [3] NCCL INFO 2 : NET/0-2 GPU/50 GPU/51 GPU/52 GPU/53 GPU/54 GPU/55 GPU/48 GPU/49 NET/0-2 +n116-041-181:72913:72972 [3] NCCL INFO 3 : NET/0-3 GPU/51 GPU/50 GPU/49 GPU/48 GPU/55 GPU/54 GPU/53 GPU/52 NET/0-3 +n116-041-181:72913:72972 [3] NCCL INFO 4 : NET/0-4 GPU/52 GPU/53 GPU/54 GPU/55 GPU/48 GPU/49 GPU/50 GPU/51 NET/0-4 +n116-041-181:72913:72972 [3] NCCL INFO 5 : NET/0-5 GPU/53 GPU/52 GPU/51 GPU/50 GPU/49 GPU/48 GPU/55 GPU/54 NET/0-5 +n116-041-181:72913:72972 [3] NCCL INFO 6 : NET/0-6 GPU/54 GPU/55 GPU/48 GPU/49 GPU/50 GPU/51 GPU/52 GPU/53 NET/0-6 +n116-041-181:72913:72972 [3] NCCL INFO 7 : NET/0-7 GPU/55 GPU/54 GPU/53 GPU/52 GPU/51 GPU/50 GPU/49 GPU/48 NET/0-7 +n116-041-181:72917:72967 [7] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-181:72917:72967 [7] NCCL INFO 0 : NET/0-0 GPU/48 GPU/55 GPU/54 GPU/53 GPU/52 GPU/51 GPU/50 GPU/49 NET/0-0 +n116-041-181:72917:72967 [7] NCCL INFO 1 : NET/0-1 GPU/49 GPU/50 GPU/51 GPU/52 GPU/53 GPU/54 GPU/55 GPU/48 NET/0-1 +n116-041-181:72917:72967 [7] NCCL INFO 2 : NET/0-2 GPU/50 GPU/49 GPU/48 GPU/55 GPU/54 GPU/53 GPU/52 GPU/51 NET/0-2 +n116-041-181:72917:72967 [7] NCCL INFO 3 : NET/0-3 GPU/51 GPU/52 GPU/53 GPU/54 GPU/55 GPU/48 GPU/49 GPU/50 NET/0-3 +n116-041-181:72917:72967 [7] NCCL INFO 4 : NET/0-4 GPU/52 GPU/51 GPU/50 GPU/49 GPU/48 GPU/55 GPU/54 GPU/53 NET/0-4 +n116-041-181:72917:72967 [7] NCCL INFO 5 : NET/0-5 GPU/53 GPU/54 GPU/55 GPU/48 GPU/49 GPU/50 GPU/51 GPU/52 NET/0-5 +n116-041-181:72917:72967 [7] NCCL INFO 6 : NET/0-6 GPU/54 GPU/53 GPU/52 GPU/51 GPU/50 GPU/49 GPU/48 GPU/55 NET/0-6 +n116-041-181:72917:72967 [7] NCCL INFO 7 : NET/0-7 GPU/55 GPU/48 GPU/49 GPU/50 GPU/51 GPU/52 GPU/53 GPU/54 NET/0-7 +n116-041-181:72917:72967 [7] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-181:72917:72967 [7] NCCL INFO 0 : NET/0-0 GPU/48 GPU/49 GPU/50 GPU/51 GPU/52 GPU/53 GPU/54 GPU/55 NET/0-0 +n116-041-181:72917:72967 [7] NCCL INFO 1 : NET/0-1 GPU/49 GPU/48 GPU/55 GPU/54 GPU/53 GPU/52 GPU/51 GPU/50 NET/0-1 +n116-041-181:72917:72967 [7] NCCL INFO 2 : NET/0-2 GPU/50 GPU/51 GPU/52 GPU/53 GPU/54 GPU/55 GPU/48 GPU/49 NET/0-2 +n116-041-181:72917:72967 [7] NCCL INFO 3 : NET/0-3 GPU/51 GPU/50 GPU/49 GPU/48 GPU/55 GPU/54 GPU/53 GPU/52 NET/0-3 +n116-041-181:72917:72967 [7] NCCL INFO 4 : NET/0-4 GPU/52 GPU/53 GPU/54 GPU/55 GPU/48 GPU/49 GPU/50 GPU/51 NET/0-4 +n116-041-181:72917:72967 [7] NCCL INFO 5 : NET/0-5 GPU/53 GPU/52 GPU/51 GPU/50 GPU/49 GPU/48 GPU/55 GPU/54 NET/0-5 +n116-041-181:72917:72967 [7] NCCL INFO 6 : NET/0-6 GPU/54 GPU/55 GPU/48 GPU/49 GPU/50 GPU/51 GPU/52 GPU/53 NET/0-6 +n116-041-181:72917:72967 [7] NCCL INFO 7 : NET/0-7 GPU/55 GPU/54 GPU/53 GPU/52 GPU/51 GPU/50 GPU/49 GPU/48 NET/0-7 +n116-041-181:72915:72966 [5] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-181:72915:72966 [5] NCCL INFO 0 : NET/0-0 GPU/48 GPU/55 GPU/54 GPU/53 GPU/52 GPU/51 GPU/50 GPU/49 NET/0-0 +n116-041-181:72915:72966 [5] NCCL INFO 1 : NET/0-1 GPU/49 GPU/50 GPU/51 GPU/52 GPU/53 GPU/54 GPU/55 GPU/48 NET/0-1 +n116-041-181:72915:72966 [5] NCCL INFO 2 : NET/0-2 GPU/50 GPU/49 GPU/48 GPU/55 GPU/54 GPU/53 GPU/52 GPU/51 NET/0-2 +n116-041-181:72915:72966 [5] NCCL INFO 3 : NET/0-3 GPU/51 GPU/52 GPU/53 GPU/54 GPU/55 GPU/48 GPU/49 GPU/50 NET/0-3 +n116-041-181:72915:72966 [5] NCCL INFO 4 : NET/0-4 GPU/52 GPU/51 GPU/50 GPU/49 GPU/48 GPU/55 GPU/54 GPU/53 NET/0-4 +n116-041-181:72915:72966 [5] NCCL INFO 5 : NET/0-5 GPU/53 GPU/54 GPU/55 GPU/48 GPU/49 GPU/50 GPU/51 GPU/52 NET/0-5 +n116-041-181:72915:72966 [5] NCCL INFO 6 : NET/0-6 GPU/54 GPU/53 GPU/52 GPU/51 GPU/50 GPU/49 GPU/48 GPU/55 NET/0-6 +n116-041-181:72915:72966 [5] NCCL INFO 7 : NET/0-7 GPU/55 GPU/48 GPU/49 GPU/50 GPU/51 GPU/52 GPU/53 GPU/54 NET/0-7 +n116-041-181:72915:72966 [5] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-181:72915:72966 [5] NCCL INFO 0 : NET/0-0 GPU/48 GPU/49 GPU/50 GPU/51 GPU/52 GPU/53 GPU/54 GPU/55 NET/0-0 +n116-041-181:72915:72966 [5] NCCL INFO 1 : NET/0-1 GPU/49 GPU/48 GPU/55 GPU/54 GPU/53 GPU/52 GPU/51 GPU/50 NET/0-1 +n116-041-181:72915:72966 [5] NCCL INFO 2 : NET/0-2 GPU/50 GPU/51 GPU/52 GPU/53 GPU/54 GPU/55 GPU/48 GPU/49 NET/0-2 +n116-041-181:72915:72966 [5] NCCL INFO 3 : NET/0-3 GPU/51 GPU/50 GPU/49 GPU/48 GPU/55 GPU/54 GPU/53 GPU/52 NET/0-3 +n116-041-181:72915:72966 [5] NCCL INFO 4 : NET/0-4 GPU/52 GPU/53 GPU/54 GPU/55 GPU/48 GPU/49 GPU/50 GPU/51 NET/0-4 +n116-041-181:72915:72966 [5] NCCL INFO 5 : NET/0-5 GPU/53 GPU/52 GPU/51 GPU/50 GPU/49 GPU/48 GPU/55 GPU/54 NET/0-5 +n116-041-181:72915:72966 [5] NCCL INFO 6 : NET/0-6 GPU/54 GPU/55 GPU/48 GPU/49 GPU/50 GPU/51 GPU/52 GPU/53 NET/0-6 +n116-041-181:72915:72966 [5] NCCL INFO 7 : NET/0-7 GPU/55 GPU/54 GPU/53 GPU/52 GPU/51 GPU/50 GPU/49 GPU/48 NET/0-7 +n116-041-181:72911:72970 [1] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-181:72911:72970 [1] NCCL INFO 0 : NET/0-0 GPU/48 GPU/55 GPU/54 GPU/53 GPU/52 GPU/51 GPU/50 GPU/49 NET/0-0 +n116-041-181:72911:72970 [1] NCCL INFO 1 : NET/0-1 GPU/49 GPU/50 GPU/51 GPU/52 GPU/53 GPU/54 GPU/55 GPU/48 NET/0-1 +n116-041-181:72911:72970 [1] NCCL INFO 2 : NET/0-2 GPU/50 GPU/49 GPU/48 GPU/55 GPU/54 GPU/53 GPU/52 GPU/51 NET/0-2 +n116-041-181:72911:72970 [1] NCCL INFO 3 : NET/0-3 GPU/51 GPU/52 GPU/53 GPU/54 GPU/55 GPU/48 GPU/49 GPU/50 NET/0-3 +n116-041-181:72911:72970 [1] NCCL INFO 4 : NET/0-4 GPU/52 GPU/51 GPU/50 GPU/49 GPU/48 GPU/55 GPU/54 GPU/53 NET/0-4 +n116-041-181:72911:72970 [1] NCCL INFO 5 : NET/0-5 GPU/53 GPU/54 GPU/55 GPU/48 GPU/49 GPU/50 GPU/51 GPU/52 NET/0-5 +n116-041-181:72911:72970 [1] NCCL INFO 6 : NET/0-6 GPU/54 GPU/53 GPU/52 GPU/51 GPU/50 GPU/49 GPU/48 GPU/55 NET/0-6 +n116-041-181:72911:72970 [1] NCCL INFO 7 : NET/0-7 GPU/55 GPU/48 GPU/49 GPU/50 GPU/51 GPU/52 GPU/53 GPU/54 NET/0-7 +n116-041-181:72911:72970 [1] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-181:72911:72970 [1] NCCL INFO 0 : NET/0-0 GPU/48 GPU/49 GPU/50 GPU/51 GPU/52 GPU/53 GPU/54 GPU/55 NET/0-0 +n116-041-181:72911:72970 [1] NCCL INFO 1 : NET/0-1 GPU/49 GPU/48 GPU/55 GPU/54 GPU/53 GPU/52 GPU/51 GPU/50 NET/0-1 +n116-041-181:72911:72970 [1] NCCL INFO 2 : NET/0-2 GPU/50 GPU/51 GPU/52 GPU/53 GPU/54 GPU/55 GPU/48 GPU/49 NET/0-2 +n116-041-181:72911:72970 [1] NCCL INFO 3 : NET/0-3 GPU/51 GPU/50 GPU/49 GPU/48 GPU/55 GPU/54 GPU/53 GPU/52 NET/0-3 +n116-041-181:72911:72970 [1] NCCL INFO 4 : NET/0-4 GPU/52 GPU/53 GPU/54 GPU/55 GPU/48 GPU/49 GPU/50 GPU/51 NET/0-4 +n116-041-181:72911:72970 [1] NCCL INFO 5 : NET/0-5 GPU/53 GPU/52 GPU/51 GPU/50 GPU/49 GPU/48 GPU/55 GPU/54 NET/0-5 +n116-041-181:72911:72970 [1] NCCL INFO 6 : NET/0-6 GPU/54 GPU/55 GPU/48 GPU/49 GPU/50 GPU/51 GPU/52 GPU/53 NET/0-6 +n116-041-181:72911:72970 [1] NCCL INFO 7 : NET/0-7 GPU/55 GPU/54 GPU/53 GPU/52 GPU/51 GPU/50 GPU/49 GPU/48 NET/0-7 +n116-041-181:72914:72973 [4] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-181:72916:72969 [6] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-181:72916:72969 [6] NCCL INFO 0 : NET/0-0 GPU/48 GPU/55 GPU/54 GPU/53 GPU/52 GPU/51 GPU/50 GPU/49 NET/0-0 +n116-041-181:72916:72969 [6] NCCL INFO 1 : NET/0-1 GPU/49 GPU/50 GPU/51 GPU/52 GPU/53 GPU/54 GPU/55 GPU/48 NET/0-1 +n116-041-181:72914:72973 [4] NCCL INFO 0 : NET/0-0 GPU/48 GPU/55 GPU/54 GPU/53 GPU/52 GPU/51 GPU/50 GPU/49 NET/0-0 +n116-041-181:72916:72969 [6] NCCL INFO 2 : NET/0-2 GPU/50 GPU/49 GPU/48 GPU/55 GPU/54 GPU/53 GPU/52 GPU/51 NET/0-2 +n116-041-181:72914:72973 [4] NCCL INFO 1 : NET/0-1 GPU/49 GPU/50 GPU/51 GPU/52 GPU/53 GPU/54 GPU/55 GPU/48 NET/0-1 +n116-041-181:72916:72969 [6] NCCL INFO 3 : NET/0-3 GPU/51 GPU/52 GPU/53 GPU/54 GPU/55 GPU/48 GPU/49 GPU/50 NET/0-3 +n116-041-181:72914:72973 [4] NCCL INFO 2 : NET/0-2 GPU/50 GPU/49 GPU/48 GPU/55 GPU/54 GPU/53 GPU/52 GPU/51 NET/0-2 +n116-041-181:72916:72969 [6] NCCL INFO 4 : NET/0-4 GPU/52 GPU/51 GPU/50 GPU/49 GPU/48 GPU/55 GPU/54 GPU/53 NET/0-4 +n116-041-181:72914:72973 [4] NCCL INFO 3 : NET/0-3 GPU/51 GPU/52 GPU/53 GPU/54 GPU/55 GPU/48 GPU/49 GPU/50 NET/0-3 +n116-041-181:72916:72969 [6] NCCL INFO 5 : NET/0-5 GPU/53 GPU/54 GPU/55 GPU/48 GPU/49 GPU/50 GPU/51 GPU/52 NET/0-5 +n116-041-181:72914:72973 [4] NCCL INFO 4 : NET/0-4 GPU/52 GPU/51 GPU/50 GPU/49 GPU/48 GPU/55 GPU/54 GPU/53 NET/0-4 +n116-041-181:72916:72969 [6] NCCL INFO 6 : NET/0-6 GPU/54 GPU/53 GPU/52 GPU/51 GPU/50 GPU/49 GPU/48 GPU/55 NET/0-6 +n116-041-181:72914:72973 [4] NCCL INFO 5 : NET/0-5 GPU/53 GPU/54 GPU/55 GPU/48 GPU/49 GPU/50 GPU/51 GPU/52 NET/0-5 +n116-041-181:72916:72969 [6] NCCL INFO 7 : NET/0-7 GPU/55 GPU/48 GPU/49 GPU/50 GPU/51 GPU/52 GPU/53 GPU/54 NET/0-7 +n116-041-181:72914:72973 [4] NCCL INFO 6 : NET/0-6 GPU/54 GPU/53 GPU/52 GPU/51 GPU/50 GPU/49 GPU/48 GPU/55 NET/0-6 +n116-041-181:72914:72973 [4] NCCL INFO 7 : NET/0-7 GPU/55 GPU/48 GPU/49 GPU/50 GPU/51 GPU/52 GPU/53 GPU/54 NET/0-7 +n116-041-181:72916:72969 [6] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-181:72914:72973 [4] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-041-181:72914:72973 [4] NCCL INFO 0 : NET/0-0 GPU/48 GPU/49 GPU/50 GPU/51 GPU/52 GPU/53 GPU/54 GPU/55 NET/0-0 +n116-041-181:72916:72969 [6] NCCL INFO 0 : NET/0-0 GPU/48 GPU/49 GPU/50 GPU/51 GPU/52 GPU/53 GPU/54 GPU/55 NET/0-0 +n116-041-181:72914:72973 [4] NCCL INFO 1 : NET/0-1 GPU/49 GPU/48 GPU/55 GPU/54 GPU/53 GPU/52 GPU/51 GPU/50 NET/0-1 +n116-041-181:72916:72969 [6] NCCL INFO 1 : NET/0-1 GPU/49 GPU/48 GPU/55 GPU/54 GPU/53 GPU/52 GPU/51 GPU/50 NET/0-1 +n116-041-181:72914:72973 [4] NCCL INFO 2 : NET/0-2 GPU/50 GPU/51 GPU/52 GPU/53 GPU/54 GPU/55 GPU/48 GPU/49 NET/0-2 +n116-041-181:72916:72969 [6] NCCL INFO 2 : NET/0-2 GPU/50 GPU/51 GPU/52 GPU/53 GPU/54 GPU/55 GPU/48 GPU/49 NET/0-2 +n116-041-181:72914:72973 [4] NCCL INFO 3 : NET/0-3 GPU/51 GPU/50 GPU/49 GPU/48 GPU/55 GPU/54 GPU/53 GPU/52 NET/0-3 +n116-041-181:72916:72969 [6] NCCL INFO 3 : NET/0-3 GPU/51 GPU/50 GPU/49 GPU/48 GPU/55 GPU/54 GPU/53 GPU/52 NET/0-3 +n116-041-181:72914:72973 [4] NCCL INFO 4 : NET/0-4 GPU/52 GPU/53 GPU/54 GPU/55 GPU/48 GPU/49 GPU/50 GPU/51 NET/0-4 +n116-041-181:72916:72969 [6] NCCL INFO 4 : NET/0-4 GPU/52 GPU/53 GPU/54 GPU/55 GPU/48 GPU/49 GPU/50 GPU/51 NET/0-4 +n116-041-181:72914:72973 [4] NCCL INFO 5 : NET/0-5 GPU/53 GPU/52 GPU/51 GPU/50 GPU/49 GPU/48 GPU/55 GPU/54 NET/0-5 +n116-041-181:72916:72969 [6] NCCL INFO 5 : NET/0-5 GPU/53 GPU/52 GPU/51 GPU/50 GPU/49 GPU/48 GPU/55 GPU/54 NET/0-5 +n116-041-181:72914:72973 [4] NCCL INFO 6 : NET/0-6 GPU/54 GPU/55 GPU/48 GPU/49 GPU/50 GPU/51 GPU/52 GPU/53 NET/0-6 +n116-041-181:72916:72969 [6] NCCL INFO 6 : NET/0-6 GPU/54 GPU/55 GPU/48 GPU/49 GPU/50 GPU/51 GPU/52 GPU/53 NET/0-6 +n116-041-181:72914:72973 [4] NCCL INFO 7 : NET/0-7 GPU/55 GPU/54 GPU/53 GPU/52 GPU/51 GPU/50 GPU/49 GPU/48 NET/0-7 +n116-041-181:72916:72969 [6] NCCL INFO 7 : NET/0-7 GPU/55 GPU/54 GPU/53 GPU/52 GPU/51 GPU/50 GPU/49 GPU/48 NET/0-7 +n116-041-181:72917:72967 [7] NCCL INFO comm 0xbede0d0 rank 55 nRanks 64 nNodes 8 localRanks 8 localRank 7 MNNVL 0 +n116-041-181:72917:72967 [7] NCCL INFO Tree 6 : 54 -> 55 -> 48/46/-1 +n116-041-181:72917:72967 [7] NCCL INFO Tree 14 : 54 -> 55 -> 48/-1/-1 +n116-041-181:72917:72967 [7] NCCL INFO Tree 7 : 39 -> 55 -> 54/63/-1 +n116-041-181:72917:72967 [7] NCCL INFO Tree 15 : 46 -> 55 -> 54/-1/-1 +n116-041-181:72917:72967 [7] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-041-181:72915:72966 [5] NCCL INFO comm 0xc1893a0 rank 53 nRanks 64 nNodes 8 localRanks 8 localRank 5 MNNVL 0 +n116-041-181:72916:72969 [6] NCCL INFO comm 0xbc00590 rank 54 nRanks 64 nNodes 8 localRanks 8 localRank 6 MNNVL 0 +n116-041-181:72915:72966 [5] NCCL INFO Tree 4 : 52 -> 53 -> 54/44/-1 +n116-041-181:72917:72967 [7] NCCL INFO Ring 00 : 48 -> 55 -> 54 +n116-041-181:72916:72969 [6] NCCL INFO Tree 6 : 38 -> 54 -> 55/62/-1 +n116-041-181:72917:72967 [7] NCCL INFO Ring 01 : 54 -> 55 -> 48 +n116-041-181:72913:72972 [3] NCCL INFO comm 0xc145f50 rank 51 nRanks 64 nNodes 8 localRanks 8 localRank 3 MNNVL 0 +n116-041-181:72914:72973 [4] NCCL INFO comm 0xbc34310 rank 52 nRanks 64 nNodes 8 localRanks 8 localRank 4 MNNVL 0 +n116-041-181:72916:72969 [6] NCCL INFO Tree 14 : 47 -> 54 -> 55/-1/-1 +n116-041-181:72917:72967 [7] NCCL INFO Ring 02 : 48 -> 55 -> 54 +n116-041-181:72915:72966 [5] NCCL INFO Tree 12 : 52 -> 53 -> 54/-1/-1 +n116-041-181:72917:72967 [7] NCCL INFO Ring 03 : 54 -> 55 -> 48 +n116-041-181:72916:72969 [6] NCCL INFO Tree 7 : 55 -> 54 -> 53/47/-1 +n116-041-181:72915:72966 [5] NCCL INFO Tree 5 : 37 -> 53 -> 52/61/-1 +n116-041-181:72917:72967 [7] NCCL INFO Ring 04 : 48 -> 55 -> 54 +n116-041-181:72915:72966 [5] NCCL INFO Tree 13 : 44 -> 53 -> 52/-1/-1 +n116-041-181:72916:72969 [6] NCCL INFO Tree 15 : 55 -> 54 -> 53/-1/-1 +n116-041-181:72917:72967 [7] NCCL INFO Ring 05 : 54 -> 55 -> 48 +n116-041-181:72917:72967 [7] NCCL INFO Ring 06 : 48 -> 55 -> 62 +n116-041-181:72912:72968 [2] NCCL INFO comm 0xbba68e0 rank 50 nRanks 64 nNodes 8 localRanks 8 localRank 2 MNNVL 0 +n116-041-181:72917:72967 [7] NCCL INFO Ring 07 : 46 -> 55 -> 48 +n116-041-181:72915:72966 [5] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-041-181:72916:72969 [6] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-041-181:72917:72967 [7] NCCL INFO Ring 08 : 48 -> 55 -> 54 +n116-041-181:72917:72967 [7] NCCL INFO Ring 09 : 54 -> 55 -> 48 +n116-041-181:72917:72967 [7] NCCL INFO Ring 10 : 48 -> 55 -> 54 +n116-041-181:72917:72967 [7] NCCL INFO Ring 11 : 54 -> 55 -> 48 +n116-041-181:72911:72970 [1] NCCL INFO comm 0xc348340 rank 49 nRanks 64 nNodes 8 localRanks 8 localRank 1 MNNVL 0 +n116-041-181:72917:72967 [7] NCCL INFO Ring 12 : 48 -> 55 -> 54 +n116-041-181:72917:72967 [7] NCCL INFO Ring 13 : 54 -> 55 -> 48 +n116-041-181:72917:72967 [7] NCCL INFO Ring 14 : 48 -> 55 -> 62 +n116-041-181:72917:72967 [7] NCCL INFO Ring 15 : 46 -> 55 -> 48 +n116-041-181:72917:72967 [7] NCCL INFO Trees [0] -1/-1/-1->55->54 [1] 54/-1/-1->55->48 [2] 48/-1/-1->55->54 [3] 54/-1/-1->55->48 [4] 48/-1/-1->55->54 [5] 54/-1/-1->55->48 [6] 48/46/-1->55->54 [7] 54/63/-1->55->39 [8] -1/-1/-1->55->54 [9] 54/-1/-1->55->48 [10] 48/-1/-1->55->54 [11] 54/-1/-1->55->48 [12] 48/-1/-1->55->54 [13] 54/-1/-1->55->48 [14] 48/-1/-1->55->54 [15] 54/-1/-1->55->46 +n116-041-181:72913:72972 [3] NCCL INFO Tree 2 : 50 -> 51 -> 52/42/-1 +n116-041-181:72917:72967 [7] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-041-181:72910:72971 [0] NCCL INFO comm 0xc56ed00 rank 48 nRanks 64 nNodes 8 localRanks 8 localRank 0 MNNVL 0 +n116-041-181:72913:72972 [3] NCCL INFO Tree 10 : 50 -> 51 -> 52/-1/-1 +n116-041-181:72917:72967 [7] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-041-181:72914:72973 [4] NCCL INFO Tree 4 : 36 -> 52 -> 53/60/-1 +n116-041-181:72917:72967 [7] NCCL INFO P2P Chunksize set to 524288 +n116-041-181:72913:72972 [3] NCCL INFO Tree 3 : 35 -> 51 -> 50/59/-1 +n116-041-181:72912:72968 [2] NCCL INFO Tree 2 : 34 -> 50 -> 51/58/-1 +n116-041-181:72914:72973 [4] NCCL INFO Tree 12 : 45 -> 52 -> 53/-1/-1 +n116-041-181:72913:72972 [3] NCCL INFO Tree 11 : 42 -> 51 -> 50/-1/-1 +n116-041-181:72912:72968 [2] NCCL INFO Tree 10 : 43 -> 50 -> 51/-1/-1 +n116-041-181:72914:72973 [4] NCCL INFO Tree 5 : 53 -> 52 -> 51/45/-1 +n116-041-181:72912:72968 [2] NCCL INFO Tree 3 : 51 -> 50 -> 49/43/-1 +n116-041-181:72915:72966 [5] NCCL INFO Ring 00 : 54 -> 53 -> 52 +n116-041-181:72913:72972 [3] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-041-181:72911:72970 [1] NCCL INFO Tree 0 : 48 -> 49 -> 50/40/-1 +n116-041-181:72914:72973 [4] NCCL INFO Tree 13 : 53 -> 52 -> 51/-1/-1 +n116-041-181:72910:72971 [0] NCCL INFO Tree 0 : 32 -> 48 -> 49/56/-1 +n116-041-181:72912:72968 [2] NCCL INFO Tree 11 : 51 -> 50 -> 49/-1/-1 +n116-041-181:72915:72966 [5] NCCL INFO Ring 01 : 52 -> 53 -> 54 +n116-041-181:72916:72969 [6] NCCL INFO Ring 00 : 55 -> 54 -> 53 +n116-041-181:72916:72969 [6] NCCL INFO Ring 01 : 53 -> 54 -> 55 +n116-041-181:72911:72970 [1] NCCL INFO Tree 8 : 48 -> 49 -> 50/-1/-1 +n116-041-181:72910:72971 [0] NCCL INFO Tree 8 : 41 -> 48 -> 49/-1/-1 +n116-041-181:72915:72966 [5] NCCL INFO Ring 02 : 54 -> 53 -> 52 +n116-041-181:72914:72973 [4] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-041-181:72912:72968 [2] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-041-181:72916:72969 [6] NCCL INFO Ring 02 : 55 -> 54 -> 53 +n116-041-181:72910:72971 [0] NCCL INFO Tree 1 : 49 -> 48 -> 55/41/-1 +n116-041-181:72911:72970 [1] NCCL INFO Tree 1 : 33 -> 49 -> 48/57/-1 +n116-041-181:72915:72966 [5] NCCL INFO Ring 03 : 52 -> 53 -> 54 +n116-041-181:72916:72969 [6] NCCL INFO Ring 03 : 53 -> 54 -> 55 +n116-041-181:72910:72971 [0] NCCL INFO Tree 9 : 49 -> 48 -> 55/-1/-1 +n116-041-181:72911:72970 [1] NCCL INFO Tree 9 : 40 -> 49 -> 48/-1/-1 +n116-041-181:72915:72966 [5] NCCL INFO Ring 04 : 54 -> 53 -> 60 +n116-041-181:72916:72969 [6] NCCL INFO Ring 04 : 55 -> 54 -> 53 +n116-041-181:72915:72966 [5] NCCL INFO Ring 05 : 44 -> 53 -> 54 +n116-041-181:72915:72966 [5] NCCL INFO Ring 06 : 54 -> 53 -> 52 +n116-041-181:72916:72969 [6] NCCL INFO Ring 05 : 53 -> 54 -> 55 +n116-041-181:72910:72971 [0] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-041-181:72911:72970 [1] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-041-181:72915:72966 [5] NCCL INFO Ring 07 : 52 -> 53 -> 54 +n116-041-181:72916:72969 [6] NCCL INFO Ring 06 : 47 -> 54 -> 53 +n116-041-181:72913:72972 [3] NCCL INFO Ring 00 : 52 -> 51 -> 50 +n116-041-181:72916:72969 [6] NCCL INFO Ring 07 : 53 -> 54 -> 63 +n116-041-181:72915:72966 [5] NCCL INFO Ring 08 : 54 -> 53 -> 52 +n116-041-181:72913:72972 [3] NCCL INFO Ring 01 : 50 -> 51 -> 52 +n116-041-181:72915:72966 [5] NCCL INFO Ring 09 : 52 -> 53 -> 54 +n116-041-181:72916:72969 [6] NCCL INFO Ring 08 : 55 -> 54 -> 53 +n116-041-181:72913:72972 [3] NCCL INFO Ring 02 : 52 -> 51 -> 58 +n116-041-181:72915:72966 [5] NCCL INFO Ring 10 : 54 -> 53 -> 52 +n116-041-181:72912:72968 [2] NCCL INFO Ring 00 : 51 -> 50 -> 49 +n116-041-181:72913:72972 [3] NCCL INFO Ring 03 : 42 -> 51 -> 52 +n116-041-181:72916:72969 [6] NCCL INFO Ring 09 : 53 -> 54 -> 55 +n116-041-181:72912:72968 [2] NCCL INFO Ring 01 : 49 -> 50 -> 51 +n116-041-181:72915:72966 [5] NCCL INFO Ring 11 : 52 -> 53 -> 54 +n116-041-181:72914:72973 [4] NCCL INFO Ring 00 : 53 -> 52 -> 51 +n116-041-181:72913:72972 [3] NCCL INFO Ring 04 : 52 -> 51 -> 50 +n116-041-181:72916:72969 [6] NCCL INFO Ring 10 : 55 -> 54 -> 53 +n116-041-181:72912:72968 [2] NCCL INFO Ring 02 : 43 -> 50 -> 49 +n116-041-181:72913:72972 [3] NCCL INFO Ring 05 : 50 -> 51 -> 52 +n116-041-181:72915:72966 [5] NCCL INFO Ring 12 : 54 -> 53 -> 60 +n116-041-181:72914:72973 [4] NCCL INFO Ring 01 : 51 -> 52 -> 53 +n116-041-181:72916:72969 [6] NCCL INFO Ring 11 : 53 -> 54 -> 55 +n116-041-181:72915:72966 [5] NCCL INFO Ring 13 : 44 -> 53 -> 54 +n116-041-181:72912:72968 [2] NCCL INFO Ring 03 : 49 -> 50 -> 59 +n116-041-181:72914:72973 [4] NCCL INFO Ring 02 : 53 -> 52 -> 51 +n116-041-181:72916:72969 [6] NCCL INFO Ring 12 : 55 -> 54 -> 53 +n116-041-181:72915:72966 [5] NCCL INFO Ring 14 : 54 -> 53 -> 52 +n116-041-181:72913:72972 [3] NCCL INFO Ring 06 : 52 -> 51 -> 50 +n116-041-181:72910:72971 [0] NCCL INFO Ring 00 : 41 -> 48 -> 55 +n116-041-181:72912:72968 [2] NCCL INFO Ring 04 : 51 -> 50 -> 49 +n116-041-181:72915:72966 [5] NCCL INFO Ring 15 : 52 -> 53 -> 54 +n116-041-181:72914:72973 [4] NCCL INFO Ring 03 : 51 -> 52 -> 53 +n116-041-181:72915:72966 [5] NCCL INFO Trees [0] 54/-1/-1->53->52 [1] 52/-1/-1->53->54 [2] 54/-1/-1->53->52 [3] 52/-1/-1->53->54 [4] 54/44/-1->53->52 [5] 52/61/-1->53->37 [6] -1/-1/-1->53->52 [7] 52/-1/-1->53->54 [8] 54/-1/-1->53->52 [9] 52/-1/-1->53->54 [10] 54/-1/-1->53->52 [11] 52/-1/-1->53->54 [12] 54/-1/-1->53->52 [13] 52/-1/-1->53->44 [14] -1/-1/-1->53->52 [15] 52/-1/-1->53->54 +n116-041-181:72916:72969 [6] NCCL INFO Ring 13 : 53 -> 54 -> 55 +n116-041-181:72913:72972 [3] NCCL INFO Ring 07 : 50 -> 51 -> 52 +n116-041-181:72910:72971 [0] NCCL INFO Ring 01 : 55 -> 48 -> 57 +n116-041-181:72912:72968 [2] NCCL INFO Ring 05 : 49 -> 50 -> 51 +n116-041-181:72911:72970 [1] NCCL INFO Ring 00 : 50 -> 49 -> 56 +n116-041-181:72914:72973 [4] NCCL INFO Ring 04 : 45 -> 52 -> 51 +n116-041-181:72916:72969 [6] NCCL INFO Ring 14 : 47 -> 54 -> 53 +n116-041-181:72913:72972 [3] NCCL INFO Ring 08 : 52 -> 51 -> 50 +n116-041-181:72911:72970 [1] NCCL INFO Ring 01 : 40 -> 49 -> 50 +n116-041-181:72916:72969 [6] NCCL INFO Ring 15 : 53 -> 54 -> 63 +n116-041-181:72915:72966 [5] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-041-181:72912:72968 [2] NCCL INFO Ring 06 : 51 -> 50 -> 49 +n116-041-181:72910:72971 [0] NCCL INFO Ring 02 : 49 -> 48 -> 55 +n116-041-181:72914:72973 [4] NCCL INFO Ring 05 : 51 -> 52 -> 61 +n116-041-181:72913:72972 [3] NCCL INFO Ring 09 : 50 -> 51 -> 52 +n116-041-181:72910:72971 [0] NCCL INFO Ring 03 : 55 -> 48 -> 49 +n116-041-181:72916:72969 [6] NCCL INFO Trees [0] 55/-1/-1->54->53 [1] 53/-1/-1->54->55 [2] 55/-1/-1->54->53 [3] 53/-1/-1->54->55 [4] 55/-1/-1->54->53 [5] -1/-1/-1->54->55 [6] 55/62/-1->54->38 [7] 53/47/-1->54->55 [8] 55/-1/-1->54->53 [9] 53/-1/-1->54->55 [10] 55/-1/-1->54->53 [11] 53/-1/-1->54->55 [12] 55/-1/-1->54->53 [13] -1/-1/-1->54->55 [14] 55/-1/-1->54->47 [15] 53/-1/-1->54->55 +n116-041-181:72911:72970 [1] NCCL INFO Ring 02 : 50 -> 49 -> 48 +n116-041-181:72915:72966 [5] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-041-181:72912:72968 [2] NCCL INFO Ring 07 : 49 -> 50 -> 51 +n116-041-181:72915:72966 [5] NCCL INFO P2P Chunksize set to 524288 +n116-041-181:72914:72973 [4] NCCL INFO Ring 06 : 53 -> 52 -> 51 +n116-041-181:72913:72972 [3] NCCL INFO Ring 10 : 52 -> 51 -> 58 +n116-041-181:72910:72971 [0] NCCL INFO Ring 04 : 49 -> 48 -> 55 +n116-041-181:72916:72969 [6] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-041-181:72911:72970 [1] NCCL INFO Ring 03 : 48 -> 49 -> 50 +n116-041-181:72912:72968 [2] NCCL INFO Ring 08 : 51 -> 50 -> 49 +n116-041-181:72913:72972 [3] NCCL INFO Ring 11 : 42 -> 51 -> 52 +n116-041-181:72914:72973 [4] NCCL INFO Ring 07 : 51 -> 52 -> 53 +n116-041-181:72910:72971 [0] NCCL INFO Ring 05 : 55 -> 48 -> 49 +n116-041-181:72913:72972 [3] NCCL INFO Ring 12 : 52 -> 51 -> 50 +n116-041-181:72916:72969 [6] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-041-181:72914:72973 [4] NCCL INFO Ring 08 : 53 -> 52 -> 51 +n116-041-181:72911:72970 [1] NCCL INFO Ring 04 : 50 -> 49 -> 48 +n116-041-181:72911:72970 [1] NCCL INFO Ring 05 : 48 -> 49 -> 50 +n116-041-181:72912:72968 [2] NCCL INFO Ring 09 : 49 -> 50 -> 51 +n116-041-181:72911:72970 [1] NCCL INFO Ring 06 : 50 -> 49 -> 48 +n116-041-181:72910:72971 [0] NCCL INFO Ring 06 : 49 -> 48 -> 55 +n116-041-181:72913:72972 [3] NCCL INFO Ring 13 : 50 -> 51 -> 52 +n116-041-181:72916:72969 [6] NCCL INFO P2P Chunksize set to 524288 +n116-041-181:72914:72973 [4] NCCL INFO Ring 09 : 51 -> 52 -> 53 +n116-041-181:72912:72968 [2] NCCL INFO Ring 10 : 43 -> 50 -> 49 +n116-041-181:72910:72971 [0] NCCL INFO Ring 07 : 55 -> 48 -> 49 +n116-041-181:72911:72970 [1] NCCL INFO Ring 07 : 48 -> 49 -> 50 +n116-041-181:72913:72972 [3] NCCL INFO Ring 14 : 52 -> 51 -> 50 +n116-041-181:72912:72968 [2] NCCL INFO Ring 11 : 49 -> 50 -> 59 +n116-041-181:72910:72971 [0] NCCL INFO Ring 08 : 41 -> 48 -> 55 +n116-041-181:72914:72973 [4] NCCL INFO Ring 10 : 53 -> 52 -> 51 +n116-041-181:72911:72970 [1] NCCL INFO Ring 08 : 50 -> 49 -> 56 +n116-041-181:72913:72972 [3] NCCL INFO Ring 15 : 50 -> 51 -> 52 +n116-041-181:72912:72968 [2] NCCL INFO Ring 12 : 51 -> 50 -> 49 +n116-041-181:72910:72971 [0] NCCL INFO Ring 09 : 55 -> 48 -> 57 +n116-041-181:72913:72972 [3] NCCL INFO Trees [0] 52/-1/-1->51->50 [1] 50/-1/-1->51->52 [2] 52/42/-1->51->50 [3] 50/59/-1->51->35 [4] -1/-1/-1->51->50 [5] 50/-1/-1->51->52 [6] 52/-1/-1->51->50 [7] 50/-1/-1->51->52 [8] 52/-1/-1->51->50 [9] 50/-1/-1->51->52 [10] 52/-1/-1->51->50 [11] 50/-1/-1->51->42 [12] -1/-1/-1->51->50 [13] 50/-1/-1->51->52 [14] 52/-1/-1->51->50 [15] 50/-1/-1->51->52 +n116-041-181:72914:72973 [4] NCCL INFO Ring 11 : 51 -> 52 -> 53 +n116-041-181:72911:72970 [1] NCCL INFO Ring 09 : 40 -> 49 -> 50 +n116-041-181:72912:72968 [2] NCCL INFO Ring 13 : 49 -> 50 -> 51 +n116-041-181:72910:72971 [0] NCCL INFO Ring 10 : 49 -> 48 -> 55 +n116-041-181:72913:72972 [3] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-041-181:72914:72973 [4] NCCL INFO Ring 12 : 45 -> 52 -> 51 +n116-041-181:72912:72968 [2] NCCL INFO Ring 14 : 51 -> 50 -> 49 +n116-041-181:72913:72972 [3] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-041-181:72911:72970 [1] NCCL INFO Ring 10 : 50 -> 49 -> 48 +n116-041-181:72910:72971 [0] NCCL INFO Ring 11 : 55 -> 48 -> 49 +n116-041-181:72914:72973 [4] NCCL INFO Ring 13 : 51 -> 52 -> 61 +n116-041-181:72913:72972 [3] NCCL INFO P2P Chunksize set to 524288 +n116-041-181:72910:72971 [0] NCCL INFO Ring 12 : 49 -> 48 -> 55 +n116-041-181:72910:72971 [0] NCCL INFO Ring 13 : 55 -> 48 -> 49 +n116-041-181:72912:72968 [2] NCCL INFO Ring 15 : 49 -> 50 -> 51 +n116-041-181:72911:72970 [1] NCCL INFO Ring 11 : 48 -> 49 -> 50 +n116-041-181:72910:72971 [0] NCCL INFO Ring 14 : 49 -> 48 -> 55 +n116-041-181:72912:72968 [2] NCCL INFO Trees [0] 51/-1/-1->50->49 [1] -1/-1/-1->50->51 [2] 51/58/-1->50->34 [3] 49/43/-1->50->51 [4] 51/-1/-1->50->49 [5] 49/-1/-1->50->51 [6] 51/-1/-1->50->49 [7] 49/-1/-1->50->51 [8] 51/-1/-1->50->49 [9] -1/-1/-1->50->51 [10] 51/-1/-1->50->43 [11] 49/-1/-1->50->51 [12] 51/-1/-1->50->49 [13] 49/-1/-1->50->51 [14] 51/-1/-1->50->49 [15] 49/-1/-1->50->51 +n116-041-181:72914:72973 [4] NCCL INFO Ring 14 : 53 -> 52 -> 51 +n116-041-181:72910:72971 [0] NCCL INFO Ring 15 : 55 -> 48 -> 49 +n116-041-181:72911:72970 [1] NCCL INFO Ring 12 : 50 -> 49 -> 48 +n116-041-181:72914:72973 [4] NCCL INFO Ring 15 : 51 -> 52 -> 53 +n116-041-181:72912:72968 [2] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-041-181:72910:72971 [0] NCCL INFO Trees [0] 49/56/-1->48->32 [1] 55/41/-1->48->49 [2] 49/-1/-1->48->55 [3] 55/-1/-1->48->49 [4] 49/-1/-1->48->55 [5] 55/-1/-1->48->49 [6] 49/-1/-1->48->55 [7] -1/-1/-1->48->49 [8] 49/-1/-1->48->41 [9] 55/-1/-1->48->49 [10] 49/-1/-1->48->55 [11] 55/-1/-1->48->49 [12] 49/-1/-1->48->55 [13] 55/-1/-1->48->49 [14] 49/-1/-1->48->55 [15] -1/-1/-1->48->49 +n116-041-181:72914:72973 [4] NCCL INFO Trees [0] 53/-1/-1->52->51 [1] 51/-1/-1->52->53 [2] 53/-1/-1->52->51 [3] -1/-1/-1->52->53 [4] 53/60/-1->52->36 [5] 51/45/-1->52->53 [6] 53/-1/-1->52->51 [7] 51/-1/-1->52->53 [8] 53/-1/-1->52->51 [9] 51/-1/-1->52->53 [10] 53/-1/-1->52->51 [11] -1/-1/-1->52->53 [12] 53/-1/-1->52->45 [13] 51/-1/-1->52->53 [14] 53/-1/-1->52->51 [15] 51/-1/-1->52->53 +n116-041-181:72912:72968 [2] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-041-181:72912:72968 [2] NCCL INFO P2P Chunksize set to 524288 +n116-041-181:72910:72971 [0] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-041-181:72910:72971 [0] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-041-181:72914:72973 [4] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-041-181:72910:72971 [0] NCCL INFO P2P Chunksize set to 524288 +n116-041-181:72911:72970 [1] NCCL INFO Ring 13 : 48 -> 49 -> 50 +n116-041-181:72914:72973 [4] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-041-181:72911:72970 [1] NCCL INFO Ring 14 : 50 -> 49 -> 48 +n116-041-181:72914:72973 [4] NCCL INFO P2P Chunksize set to 524288 +n116-041-181:72911:72970 [1] NCCL INFO Ring 15 : 48 -> 49 -> 50 +n116-041-181:72911:72970 [1] NCCL INFO Trees [0] 50/40/-1->49->48 [1] 48/57/-1->49->33 [2] -1/-1/-1->49->48 [3] 48/-1/-1->49->50 [4] 50/-1/-1->49->48 [5] 48/-1/-1->49->50 [6] 50/-1/-1->49->48 [7] 48/-1/-1->49->50 [8] 50/-1/-1->49->48 [9] 48/-1/-1->49->40 [10] -1/-1/-1->49->48 [11] 48/-1/-1->49->50 [12] 50/-1/-1->49->48 [13] 48/-1/-1->49->50 [14] 50/-1/-1->49->48 [15] 48/-1/-1->49->50 +n116-041-181:72911:72970 [1] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-041-181:72911:72970 [1] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-041-181:72911:72970 [1] NCCL INFO P2P Chunksize set to 524288 +n116-041-181:72917:73006 [7] NCCL INFO [Proxy Service] Device 7 CPU core 186 +n116-041-181:72917:73007 [7] NCCL INFO [Proxy Service UDS] Device 7 CPU core 87 +n116-041-181:72915:73009 [5] NCCL INFO [Proxy Service UDS] Device 5 CPU core 169 +n116-041-181:72910:73012 [0] NCCL INFO [Proxy Service] Device 0 CPU core 130 +n116-041-181:72915:73008 [5] NCCL INFO [Proxy Service] Device 5 CPU core 59 +n116-041-181:72912:73011 [2] NCCL INFO [Proxy Service] Device 2 CPU core 139 +n116-041-181:72912:73015 [2] NCCL INFO [Proxy Service UDS] Device 2 CPU core 114 +n116-041-181:72913:73013 [3] NCCL INFO [Proxy Service UDS] Device 3 CPU core 112 +n116-041-181:72913:73010 [3] NCCL INFO [Proxy Service] Device 3 CPU core 144 +n116-041-181:72911:73019 [1] NCCL INFO [Proxy Service] Device 1 CPU core 111 +n116-041-181:72916:73014 [6] NCCL INFO [Proxy Service] Device 6 CPU core 77 +n116-041-181:72914:73020 [4] NCCL INFO [Proxy Service UDS] Device 4 CPU core 60 +n116-041-181:72910:73016 [0] NCCL INFO [Proxy Service UDS] Device 0 CPU core 141 +n116-041-181:72916:73018 [6] NCCL INFO [Proxy Service UDS] Device 6 CPU core 58 +n116-041-181:72911:73021 [1] NCCL INFO [Proxy Service UDS] Device 1 CPU core 125 +n116-041-181:72914:73017 [4] NCCL INFO [Proxy Service] Device 4 CPU core 90 +n116-041-181:72914:72973 [4] NCCL INFO NCCL_PROTO set by environment to Simple +n116-041-181:72914:72973 [4] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-041-181:72914:72973 [4] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n116-041-181:72914:72973 [4] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n116-041-181:72910:72971 [0] NCCL INFO NCCL_PROTO set by environment to Simple +n116-041-181:72910:72971 [0] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-041-181:72910:72971 [0] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n116-041-181:72910:72971 [0] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n116-041-181:72915:72966 [5] NCCL INFO NCCL_PROTO set by environment to Simple +n116-041-181:72915:72966 [5] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-041-181:72915:72966 [5] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n116-041-181:72915:72966 [5] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n116-041-181:72916:72969 [6] NCCL INFO NCCL_PROTO set by environment to Simple +n116-041-181:72916:72969 [6] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-041-181:72916:72969 [6] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n116-041-181:72916:72969 [6] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n116-041-181:72913:72972 [3] NCCL INFO NCCL_PROTO set by environment to Simple +n116-041-181:72913:72972 [3] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-041-181:72913:72972 [3] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n116-041-181:72913:72972 [3] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n116-041-181:72917:72967 [7] NCCL INFO NCCL_PROTO set by environment to Simple +n116-041-181:72917:72967 [7] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-041-181:72917:72967 [7] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n116-041-181:72917:72967 [7] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n116-041-181:72912:72968 [2] NCCL INFO NCCL_PROTO set by environment to Simple +n116-041-181:72912:72968 [2] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-041-181:72912:72968 [2] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n116-041-181:72912:72968 [2] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n116-041-181:72911:72970 [1] NCCL INFO NCCL_PROTO set by environment to Simple +n116-041-181:72911:72970 [1] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-041-181:72911:72970 [1] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n116-041-181:72911:72970 [1] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n116-041-181:72910:72971 [0] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-041-181:72910:72971 [0] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-041-181:72910:72971 [0] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-041-181:72910:72971 [0] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-041-181:72910:72971 [0] NCCL INFO ncclCommInitRankConfig comm 0xc56ed00 rank 48 nranks 64 cudaDev 0 nvmlDev 0 busId 4000 commId 0xfb2d11625d05007a - Init COMPLETE +n116-041-181:72910:72971 [0] NCCL INFO Init timings - ncclCommInitRankConfig: rank 48 nranks 64 total 5.44 (kernels 0.61, alloc 3.46, bootstrap 0.01, allgathers 0.06, topo 0.69, graphs 0.08, connections 0.03, rest 0.49) +n116-041-181:72911:72970 [1] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-041-181:72911:72970 [1] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-041-181:72911:72970 [1] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-041-181:72911:72970 [1] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-041-181:72911:72970 [1] NCCL INFO ncclCommInitRankConfig comm 0xc348340 rank 49 nranks 64 cudaDev 1 nvmlDev 1 busId 5000 commId 0xfb2d11625d05007a - Init COMPLETE +n116-041-181:72911:72970 [1] NCCL INFO Init timings - ncclCommInitRankConfig: rank 49 nranks 64 total 5.45 (kernels 0.60, alloc 3.48, bootstrap 0.00, allgathers 0.02, topo 0.69, graphs 0.12, connections 0.04, rest 0.48) +n116-041-181:72910:73022 [0] NCCL INFO Channel 00/0 : 48[0] -> 49[1] via P2P/CUMEM +n116-041-181:72910:73022 [0] NCCL INFO Channel 01/0 : 48[0] -> 49[1] via P2P/CUMEM +n116-041-181:72910:73022 [0] NCCL INFO Channel 02/0 : 48[0] -> 49[1] via P2P/CUMEM +n116-041-181:72910:73022 [0] NCCL INFO Channel 03/0 : 48[0] -> 49[1] via P2P/CUMEM +n116-041-181:72910:73022 [0] NCCL INFO Channel 04/0 : 48[0] -> 49[1] via P2P/CUMEM +n116-041-181:72910:73022 [0] NCCL INFO Channel 05/0 : 48[0] -> 49[1] via P2P/CUMEM +n116-041-181:72910:73022 [0] NCCL INFO Channel 06/0 : 48[0] -> 49[1] via P2P/CUMEM +n116-041-181:72910:73022 [0] NCCL INFO Channel 07/0 : 48[0] -> 49[1] via P2P/CUMEM +n116-041-181:72910:73022 [0] NCCL INFO Channel 08/0 : 48[0] -> 49[1] via P2P/CUMEM +n116-041-181:72910:73022 [0] NCCL INFO Channel 09/0 : 48[0] -> 49[1] via P2P/CUMEM +n116-041-181:72910:73022 [0] NCCL INFO Channel 10/0 : 48[0] -> 49[1] via P2P/CUMEM +n116-041-181:72910:73022 [0] NCCL INFO Channel 11/0 : 48[0] -> 49[1] via P2P/CUMEM +n116-041-181:72910:73022 [0] NCCL INFO Channel 12/0 : 48[0] -> 49[1] via P2P/CUMEM +n116-041-181:72910:73022 [0] NCCL INFO Channel 13/0 : 48[0] -> 49[1] via P2P/CUMEM +n116-041-181:72910:73022 [0] NCCL INFO Channel 14/0 : 48[0] -> 49[1] via P2P/CUMEM +n116-041-181:72910:73022 [0] NCCL INFO Channel 15/0 : 48[0] -> 49[1] via P2P/CUMEM +n116-041-181:72915:72966 [5] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-041-181:72915:72966 [5] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-041-181:72915:72966 [5] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-041-181:72915:72966 [5] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-041-181:72915:72966 [5] NCCL INFO ncclCommInitRankConfig comm 0xc1893a0 rank 53 nranks 64 cudaDev 5 nvmlDev 5 busId 85000 commId 0xfb2d11625d05007a - Init COMPLETE +n116-041-181:72915:72966 [5] NCCL INFO Init timings - ncclCommInitRankConfig: rank 53 nranks 64 total 5.50 (kernels 0.70, alloc 3.41, bootstrap 0.02, allgathers 0.06, topo 0.70, graphs 0.08, connections 0.04, rest 0.50) +n116-041-181:72914:72973 [4] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-041-181:72914:72973 [4] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-041-181:72914:72973 [4] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-041-181:72914:72973 [4] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-041-181:72914:72973 [4] NCCL INFO ncclCommInitRankConfig comm 0xbc34310 rank 52 nranks 64 cudaDev 4 nvmlDev 4 busId 84000 commId 0xfb2d11625d05007a - Init COMPLETE +n116-041-181:72914:72973 [4] NCCL INFO Init timings - ncclCommInitRankConfig: rank 52 nranks 64 total 5.45 (kernels 0.59, alloc 3.47, bootstrap 0.02, allgathers 0.01, topo 0.70, graphs 0.13, connections 0.03, rest 0.51) +n116-041-181:72916:72969 [6] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-041-181:72916:72969 [6] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-041-181:72916:72969 [6] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-041-181:72916:72969 [6] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-041-181:72916:72969 [6] NCCL INFO ncclCommInitRankConfig comm 0xbc00590 rank 54 nranks 64 cudaDev 6 nvmlDev 6 busId 8b000 commId 0xfb2d11625d05007a - Init COMPLETE +n116-041-181:72916:72969 [6] NCCL INFO Init timings - ncclCommInitRankConfig: rank 54 nranks 64 total 5.47 (kernels 0.60, alloc 3.48, bootstrap 0.01, allgathers 0.01, topo 0.70, graphs 0.13, connections 0.04, rest 0.51) +n116-041-181:72913:72972 [3] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-041-181:72913:72972 [3] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-041-181:72913:72972 [3] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-041-181:72913:72972 [3] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-041-181:72913:72972 [3] NCCL INFO ncclCommInitRankConfig comm 0xc145f50 rank 51 nranks 64 cudaDev 3 nvmlDev 3 busId c000 commId 0xfb2d11625d05007a - Init COMPLETE +n116-041-181:72913:72972 [3] NCCL INFO Init timings - ncclCommInitRankConfig: rank 51 nranks 64 total 5.47 (kernels 0.60, alloc 3.46, bootstrap 0.02, allgathers 0.06, topo 0.70, graphs 0.08, connections 0.04, rest 0.52) +n116-041-181:72912:72968 [2] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-041-181:72912:72968 [2] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-041-181:72912:72968 [2] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-041-181:72912:72968 [2] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-041-181:72912:72968 [2] NCCL INFO ncclCommInitRankConfig comm 0xbba68e0 rank 50 nranks 64 cudaDev 2 nvmlDev 2 busId b000 commId 0xfb2d11625d05007a - Init COMPLETE +n116-041-181:72912:72968 [2] NCCL INFO Init timings - ncclCommInitRankConfig: rank 50 nranks 64 total 5.49 (kernels 0.62, alloc 3.40, bootstrap 0.07, allgathers 0.06, topo 0.69, graphs 0.08, connections 0.04, rest 0.52) +n116-041-181:72917:72967 [7] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-041-181:72917:72967 [7] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-041-181:72917:72967 [7] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-041-181:72917:72967 [7] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-041-181:72917:72967 [7] NCCL INFO ncclCommInitRankConfig comm 0xbede0d0 rank 55 nranks 64 cudaDev 7 nvmlDev 7 busId 8c000 commId 0xfb2d11625d05007a - Init COMPLETE +n116-041-181:72917:72967 [7] NCCL INFO Init timings - ncclCommInitRankConfig: rank 55 nranks 64 total 5.50 (kernels 0.62, alloc 3.47, bootstrap 0.02, allgathers 0.06, topo 0.70, graphs 0.08, connections 0.04, rest 0.52) +n116-041-181:72911:73023 [1] NCCL INFO Channel 00/0 : 49[1] -> 50[2] via P2P/CUMEM +n116-041-181:72911:73023 [1] NCCL INFO Channel 03/0 : 49[1] -> 50[2] via P2P/CUMEM +n116-041-181:72911:73023 [1] NCCL INFO Channel 04/0 : 49[1] -> 50[2] via P2P/CUMEM +n116-041-181:72911:73023 [1] NCCL INFO Channel 05/0 : 49[1] -> 50[2] via P2P/CUMEM +n116-041-181:72911:73023 [1] NCCL INFO Channel 06/0 : 49[1] -> 50[2] via P2P/CUMEM +n116-041-181:72911:73023 [1] NCCL INFO Channel 07/0 : 49[1] -> 50[2] via P2P/CUMEM +n116-041-181:72911:73023 [1] NCCL INFO Channel 08/0 : 49[1] -> 50[2] via P2P/CUMEM +n116-041-181:72911:73023 [1] NCCL INFO Channel 11/0 : 49[1] -> 50[2] via P2P/CUMEM +n116-041-181:72911:73023 [1] NCCL INFO Channel 12/0 : 49[1] -> 50[2] via P2P/CUMEM +n116-041-181:72911:73023 [1] NCCL INFO Channel 13/0 : 49[1] -> 50[2] via P2P/CUMEM +n116-041-181:72911:73023 [1] NCCL INFO Channel 14/0 : 49[1] -> 50[2] via P2P/CUMEM +n116-041-181:72911:73023 [1] NCCL INFO Channel 15/0 : 49[1] -> 50[2] via P2P/CUMEM +n116-041-181:72916:73024 [6] NCCL INFO Channel 00/0 : 54[6] -> 55[7] via P2P/CUMEM +n116-041-181:72916:73024 [6] NCCL INFO Channel 01/0 : 54[6] -> 55[7] via P2P/CUMEM +n116-041-181:72916:73024 [6] NCCL INFO Channel 02/0 : 54[6] -> 55[7] via P2P/CUMEM +n116-041-181:72916:73024 [6] NCCL INFO Channel 03/0 : 54[6] -> 55[7] via P2P/CUMEM +n116-041-181:72916:73024 [6] NCCL INFO Channel 04/0 : 54[6] -> 55[7] via P2P/CUMEM +n116-041-181:72916:73024 [6] NCCL INFO Channel 05/0 : 54[6] -> 55[7] via P2P/CUMEM +n116-041-181:72914:73027 [4] NCCL INFO Channel 00/0 : 52[4] -> 53[5] via P2P/CUMEM +n116-041-181:72916:73024 [6] NCCL INFO Channel 06/0 : 54[6] -> 55[7] via P2P/CUMEM +n116-041-181:72915:73026 [5] NCCL INFO Channel 00/0 : 53[5] -> 54[6] via P2P/CUMEM +n116-041-181:72914:73027 [4] NCCL INFO Channel 01/0 : 52[4] -> 53[5] via P2P/CUMEM +n116-041-181:72916:73024 [6] NCCL INFO Channel 07/0 : 54[6] -> 55[7] via P2P/CUMEM +n116-041-181:72915:73026 [5] NCCL INFO Channel 01/0 : 53[5] -> 54[6] via P2P/CUMEM +n116-041-181:72914:73027 [4] NCCL INFO Channel 02/0 : 52[4] -> 53[5] via P2P/CUMEM +n116-041-181:72913:73025 [3] NCCL INFO Channel 00/0 : 51[3] -> 52[4] via P2P/CUMEM +n116-041-181:72916:73024 [6] NCCL INFO Channel 08/0 : 54[6] -> 55[7] via P2P/CUMEM +n116-041-181:72915:73026 [5] NCCL INFO Channel 02/0 : 53[5] -> 54[6] via P2P/CUMEM +n116-041-181:72914:73027 [4] NCCL INFO Channel 03/0 : 52[4] -> 53[5] via P2P/CUMEM +n116-041-181:72916:73024 [6] NCCL INFO Channel 09/0 : 54[6] -> 55[7] via P2P/CUMEM +n116-041-181:72915:73026 [5] NCCL INFO Channel 03/0 : 53[5] -> 54[6] via P2P/CUMEM +n116-041-181:72913:73025 [3] NCCL INFO Channel 01/0 : 51[3] -> 52[4] via P2P/CUMEM +n116-041-181:72914:73027 [4] NCCL INFO Channel 04/0 : 52[4] -> 53[5] via P2P/CUMEM +n116-041-181:72916:73024 [6] NCCL INFO Channel 10/0 : 54[6] -> 55[7] via P2P/CUMEM +n116-041-181:72915:73026 [5] NCCL INFO Channel 04/0 : 53[5] -> 54[6] via P2P/CUMEM +n116-041-181:72913:73025 [3] NCCL INFO Channel 02/0 : 51[3] -> 52[4] via P2P/CUMEM +n116-041-181:72914:73027 [4] NCCL INFO Channel 05/0 : 52[4] -> 53[5] via P2P/CUMEM +n116-041-181:72916:73024 [6] NCCL INFO Channel 11/0 : 54[6] -> 55[7] via P2P/CUMEM +n116-041-181:72915:73026 [5] NCCL INFO Channel 07/0 : 53[5] -> 54[6] via P2P/CUMEM +n116-041-181:72913:73025 [3] NCCL INFO Channel 05/0 : 51[3] -> 52[4] via P2P/CUMEM +n116-041-181:72914:73027 [4] NCCL INFO Channel 06/0 : 52[4] -> 53[5] via P2P/CUMEM +n116-041-181:72916:73024 [6] NCCL INFO Channel 12/0 : 54[6] -> 55[7] via P2P/CUMEM +n116-041-181:72915:73026 [5] NCCL INFO Channel 08/0 : 53[5] -> 54[6] via P2P/CUMEM +n116-041-181:72913:73025 [3] NCCL INFO Channel 06/0 : 51[3] -> 52[4] via P2P/CUMEM +n116-041-181:72914:73027 [4] NCCL INFO Channel 07/0 : 52[4] -> 53[5] via P2P/CUMEM +n116-041-181:72916:73024 [6] NCCL INFO Channel 13/0 : 54[6] -> 55[7] via P2P/CUMEM +n116-041-181:72915:73026 [5] NCCL INFO Channel 09/0 : 53[5] -> 54[6] via P2P/CUMEM +n116-041-181:72913:73025 [3] NCCL INFO Channel 07/0 : 51[3] -> 52[4] via P2P/CUMEM +n116-041-181:72914:73027 [4] NCCL INFO Channel 08/0 : 52[4] -> 53[5] via P2P/CUMEM +n116-041-181:72916:73024 [6] NCCL INFO Channel 14/0 : 54[6] -> 55[7] via P2P/CUMEM +n116-041-181:72915:73026 [5] NCCL INFO Channel 10/0 : 53[5] -> 54[6] via P2P/CUMEM +n116-041-181:72912:73029 [2] NCCL INFO Channel 00/0 : 50[2] -> 51[3] via P2P/CUMEM +n116-041-181:72913:73025 [3] NCCL INFO Channel 08/0 : 51[3] -> 52[4] via P2P/CUMEM +n116-041-181:72914:73027 [4] NCCL INFO Channel 09/0 : 52[4] -> 53[5] via P2P/CUMEM +n116-041-181:72916:73024 [6] NCCL INFO Channel 15/0 : 54[6] -> 55[7] via P2P/CUMEM +n116-041-181:72915:73026 [5] NCCL INFO Channel 11/0 : 53[5] -> 54[6] via P2P/CUMEM +n116-041-181:72912:73029 [2] NCCL INFO Channel 01/0 : 50[2] -> 51[3] via P2P/CUMEM +n116-041-181:72913:73025 [3] NCCL INFO Channel 09/0 : 51[3] -> 52[4] via P2P/CUMEM +n116-041-181:72914:73027 [4] NCCL INFO Channel 10/0 : 52[4] -> 53[5] via P2P/CUMEM +n116-041-181:72915:73026 [5] NCCL INFO Channel 12/0 : 53[5] -> 54[6] via P2P/CUMEM +n116-041-181:72912:73029 [2] NCCL INFO Channel 02/0 : 50[2] -> 51[3] via P2P/CUMEM +n116-041-181:72914:73027 [4] NCCL INFO Channel 11/0 : 52[4] -> 53[5] via P2P/CUMEM +n116-041-181:72913:73025 [3] NCCL INFO Channel 10/0 : 51[3] -> 52[4] via P2P/CUMEM +n116-041-181:72915:73026 [5] NCCL INFO Channel 15/0 : 53[5] -> 54[6] via P2P/CUMEM +n116-041-181:72912:73029 [2] NCCL INFO Channel 03/0 : 50[2] -> 51[3] via P2P/CUMEM +n116-041-181:72914:73027 [4] NCCL INFO Channel 12/0 : 52[4] -> 53[5] via P2P/CUMEM +n116-041-181:72913:73025 [3] NCCL INFO Channel 13/0 : 51[3] -> 52[4] via P2P/CUMEM +n116-041-181:72912:73029 [2] NCCL INFO Channel 04/0 : 50[2] -> 51[3] via P2P/CUMEM +n116-041-181:72914:73027 [4] NCCL INFO Channel 13/0 : 52[4] -> 53[5] via P2P/CUMEM +n116-041-181:72913:73025 [3] NCCL INFO Channel 14/0 : 51[3] -> 52[4] via P2P/CUMEM +n116-041-181:72912:73029 [2] NCCL INFO Channel 05/0 : 50[2] -> 51[3] via P2P/CUMEM +n116-041-181:72914:73027 [4] NCCL INFO Channel 14/0 : 52[4] -> 53[5] via P2P/CUMEM +n116-041-181:72913:73025 [3] NCCL INFO Channel 15/0 : 51[3] -> 52[4] via P2P/CUMEM +n116-041-181:72912:73029 [2] NCCL INFO Channel 06/0 : 50[2] -> 51[3] via P2P/CUMEM +n116-041-181:72914:73027 [4] NCCL INFO Channel 15/0 : 52[4] -> 53[5] via P2P/CUMEM +n116-041-181:72912:73029 [2] NCCL INFO Channel 07/0 : 50[2] -> 51[3] via P2P/CUMEM +n116-041-181:72912:73029 [2] NCCL INFO Channel 08/0 : 50[2] -> 51[3] via P2P/CUMEM +n116-041-181:72912:73029 [2] NCCL INFO Channel 09/0 : 50[2] -> 51[3] via P2P/CUMEM +n116-041-181:72912:73029 [2] NCCL INFO Channel 10/0 : 50[2] -> 51[3] via P2P/CUMEM +n116-041-181:72912:73029 [2] NCCL INFO Channel 11/0 : 50[2] -> 51[3] via P2P/CUMEM +n116-041-181:72912:73029 [2] NCCL INFO Channel 12/0 : 50[2] -> 51[3] via P2P/CUMEM +n116-041-181:72912:73029 [2] NCCL INFO Channel 13/0 : 50[2] -> 51[3] via P2P/CUMEM +n116-041-181:72912:73029 [2] NCCL INFO Channel 14/0 : 50[2] -> 51[3] via P2P/CUMEM +n116-041-181:72912:73029 [2] NCCL INFO Channel 15/0 : 50[2] -> 51[3] via P2P/CUMEM +n116-041-181:72910:73030 [0] NCCL INFO [Proxy Progress] Device 0 CPU core 118 +n116-041-181:72910:73022 [0] NCCL INFO Channel 01/0 : 41[1] -> 48[0] [receive] via NET/FasTrak/1/GDRDMA +n116-041-181:72910:73022 [0] NCCL INFO Channel 08/0 : 41[1] -> 48[0] [receive] via NET/FasTrak/0/GDRDMA +n116-041-181:72910:73022 [0] NCCL INFO Channel 01/0 : 48[0] -> 55[7] via P2P/CUMEM +n116-041-181:72910:73022 [0] NCCL INFO Channel 02/0 : 48[0] -> 55[7] via P2P/CUMEM +n116-041-181:72910:73022 [0] NCCL INFO Channel 03/0 : 48[0] -> 55[7] via P2P/CUMEM +n116-041-181:72910:73022 [0] NCCL INFO Channel 04/0 : 48[0] -> 55[7] via P2P/CUMEM +n116-041-181:72910:73022 [0] NCCL INFO Channel 05/0 : 48[0] -> 55[7] via P2P/CUMEM +n116-041-181:72910:73022 [0] NCCL INFO Channel 06/0 : 48[0] -> 55[7] via P2P/CUMEM +n116-041-181:72910:73022 [0] NCCL INFO Channel 09/0 : 48[0] -> 55[7] via P2P/CUMEM +n116-041-181:72910:73022 [0] NCCL INFO Channel 10/0 : 48[0] -> 55[7] via P2P/CUMEM +n116-041-181:72910:73022 [0] NCCL INFO Channel 11/0 : 48[0] -> 55[7] via P2P/CUMEM +n116-041-181:72910:73022 [0] NCCL INFO Channel 12/0 : 48[0] -> 55[7] via P2P/CUMEM +n116-041-181:72910:73022 [0] NCCL INFO Channel 13/0 : 48[0] -> 55[7] via P2P/CUMEM +n116-041-181:72910:73022 [0] NCCL INFO Channel 14/0 : 48[0] -> 55[7] via P2P/CUMEM +n116-041-181:72916:73031 [6] NCCL INFO [Proxy Progress] Device 6 CPU core 168 +n116-041-181:72910:73022 [0] NCCL INFO Channel 00/0 : 48[0] -> 56[0] [send] via NET/FasTrak/0/GDRDMA +n116-041-181:72916:73024 [6] NCCL INFO Channel 07/0 : 47[7] -> 54[6] [receive] via NET/FasTrak/7/GDRDMA +n116-041-181:72915:73032 [5] NCCL INFO [Proxy Progress] Device 5 CPU core 184 +n116-041-181:72915:73026 [5] NCCL INFO Channel 05/0 : 53[5] -> 61[5] [send] via NET/FasTrak/5/GDRDMA +n116-041-181:72914:73033 [4] NCCL INFO [Proxy Progress] Device 4 CPU core 195 +n116-041-181:72910:73022 [0] NCCL INFO Channel 00/0 : 32[0] -> 48[0] [receive] via NET/FasTrak/0/GDRDMA +n116-041-181:72916:73024 [6] NCCL INFO Channel 14/0 : 47[7] -> 54[6] [receive] via NET/FasTrak/6/GDRDMA +n116-041-181:72910:73022 [0] NCCL INFO Channel 00/0 : 48[0] -> 32[0] [send] via NET/FasTrak/0/GDRDMA +n116-041-181:72914:73027 [4] NCCL INFO Channel 05/0 : 45[5] -> 52[4] [receive] via NET/FasTrak/5/GDRDMA +n116-041-181:72916:73024 [6] NCCL INFO Channel 06/0 : 54[6] -> 62[6] [send] via NET/FasTrak/6/GDRDMA +n116-041-181:72915:73026 [5] NCCL INFO Channel 04/0 : 44[4] -> 53[5] [receive] via NET/FasTrak/4/GDRDMA +n116-041-181:72914:73027 [4] NCCL INFO Channel 12/0 : 45[5] -> 52[4] [receive] via NET/FasTrak/4/GDRDMA +n116-041-181:72915:73026 [5] NCCL INFO Channel 13/0 : 44[4] -> 53[5] [receive] via NET/FasTrak/5/GDRDMA +n116-041-181:72914:73027 [4] NCCL INFO Channel 04/0 : 52[4] -> 60[4] [send] via NET/FasTrak/4/GDRDMA +n116-041-181:72916:73024 [6] NCCL INFO Channel 06/0 : 38[6] -> 54[6] [receive] via NET/FasTrak/6/GDRDMA +n116-041-181:72912:73034 [2] NCCL INFO [Proxy Progress] Device 2 CPU core 126 +n116-041-181:72911:73035 [1] NCCL INFO [Proxy Progress] Device 1 CPU core 138 +n116-041-181:72911:73023 [1] NCCL INFO Channel 01/0 : 49[1] -> 57[1] [send] via NET/FasTrak/1/GDRDMA +n116-041-181:72913:73036 [3] NCCL INFO [Proxy Progress] Device 3 CPU core 24 +n116-041-181:72914:73027 [4] NCCL INFO Channel 04/0 : 36[4] -> 52[4] [receive] via NET/FasTrak/4/GDRDMA +n116-041-181:72913:73025 [3] NCCL INFO Channel 03/0 : 51[3] -> 59[3] [send] via NET/FasTrak/3/GDRDMA +n116-041-181:72916:73024 [6] NCCL INFO Channel 06/0 : 54[6] -> 38[6] [send] via NET/FasTrak/6/GDRDMA +n116-041-181:72915:73026 [5] NCCL INFO Channel 05/0 : 37[5] -> 53[5] [receive] via NET/FasTrak/5/GDRDMA +n116-041-181:72914:73027 [4] NCCL INFO Channel 04/0 : 52[4] -> 36[4] [send] via NET/FasTrak/4/GDRDMA +n116-041-181:72912:73029 [2] NCCL INFO Channel 03/0 : 43[3] -> 50[2] [receive] via NET/FasTrak/3/GDRDMA +n116-041-181:72913:73025 [3] NCCL INFO Channel 02/0 : 42[2] -> 51[3] [receive] via NET/FasTrak/2/GDRDMA +n116-041-181:72911:73023 [1] NCCL INFO Channel 00/0 : 40[0] -> 49[1] [receive] via NET/FasTrak/0/GDRDMA +n116-041-181:72912:73029 [2] NCCL INFO Channel 10/0 : 43[3] -> 50[2] [receive] via NET/FasTrak/2/GDRDMA +n116-041-181:72913:73025 [3] NCCL INFO Channel 11/0 : 42[2] -> 51[3] [receive] via NET/FasTrak/3/GDRDMA +n116-041-181:72912:73029 [2] NCCL INFO Channel 02/0 : 50[2] -> 58[2] [send] via NET/FasTrak/2/GDRDMA +n116-041-181:72911:73023 [1] NCCL INFO Channel 09/0 : 40[0] -> 49[1] [receive] via NET/FasTrak/1/GDRDMA +n116-041-181:72910:73022 [0] NCCL INFO Channel 00/0 : 56[0] -> 48[0] [receive] via NET/FasTrak/0/GDRDMA +n116-041-181:72912:73029 [2] NCCL INFO Channel 02/0 : 34[2] -> 50[2] [receive] via NET/FasTrak/2/GDRDMA +n116-041-181:72913:73025 [3] NCCL INFO Channel 03/0 : 35[3] -> 51[3] [receive] via NET/FasTrak/3/GDRDMA +n116-041-181:72915:73026 [5] NCCL INFO Channel 05/0 : 53[5] -> 37[5] [send] via NET/FasTrak/5/GDRDMA +n116-041-181:72912:73029 [2] NCCL INFO Channel 02/0 : 50[2] -> 34[2] [send] via NET/FasTrak/2/GDRDMA +n116-041-181:72911:73023 [1] NCCL INFO Channel 01/0 : 33[1] -> 49[1] [receive] via NET/FasTrak/1/GDRDMA +n116-041-181:72911:73023 [1] NCCL INFO Channel 01/0 : 49[1] -> 33[1] [send] via NET/FasTrak/1/GDRDMA +n116-041-181:72913:73025 [3] NCCL INFO Channel 03/0 : 51[3] -> 35[3] [send] via NET/FasTrak/3/GDRDMA +n116-041-181:72914:73027 [4] NCCL INFO Channel 04/0 : 60[4] -> 52[4] [receive] via NET/FasTrak/4/GDRDMA +n116-041-181:72914:73027 [4] NCCL INFO Channel 05/0 : 52[4] -> 45[5] [send] via NET/FasTrak/5/GDRDMA +n116-041-181:72914:73027 [4] NCCL INFO Channel 12/0 : 52[4] -> 45[5] [send] via NET/FasTrak/4/GDRDMA +n116-041-181:72910:73022 [0] NCCL INFO Channel 01/0 : 48[0] -> 41[1] [send] via NET/FasTrak/1/GDRDMA +n116-041-181:72910:73022 [0] NCCL INFO Channel 08/0 : 48[0] -> 41[1] [send] via NET/FasTrak/0/GDRDMA +n116-041-181:72916:73024 [6] NCCL INFO Channel 06/0 : 62[6] -> 54[6] [receive] via NET/FasTrak/6/GDRDMA +n116-041-181:72916:73024 [6] NCCL INFO Channel 07/0 : 54[6] -> 47[7] [send] via NET/FasTrak/7/GDRDMA +n116-041-181:72916:73024 [6] NCCL INFO Channel 14/0 : 54[6] -> 47[7] [send] via NET/FasTrak/6/GDRDMA +n116-041-181:72913:73025 [3] NCCL INFO Channel 02/0 : 51[3] -> 42[2] [send] via NET/FasTrak/2/GDRDMA +n116-041-181:72913:73025 [3] NCCL INFO Channel 11/0 : 51[3] -> 42[2] [send] via NET/FasTrak/3/GDRDMA +n116-041-181:72912:73029 [2] NCCL INFO Channel 02/0 : 58[2] -> 50[2] [receive] via NET/FasTrak/2/GDRDMA +n116-041-181:72915:73026 [5] NCCL INFO Channel 04/0 : 53[5] -> 44[4] [send] via NET/FasTrak/4/GDRDMA +n116-041-181:72913:73025 [3] NCCL INFO Channel 03/0 : 59[3] -> 51[3] [receive] via NET/FasTrak/3/GDRDMA +n116-041-181:72915:73026 [5] NCCL INFO Channel 13/0 : 53[5] -> 44[4] [send] via NET/FasTrak/5/GDRDMA +n116-041-181:72912:73029 [2] NCCL INFO Channel 03/0 : 50[2] -> 43[3] [send] via NET/FasTrak/3/GDRDMA +n116-041-181:72912:73029 [2] NCCL INFO Channel 10/0 : 50[2] -> 43[3] [send] via NET/FasTrak/2/GDRDMA +n116-041-181:72911:73023 [1] NCCL INFO Channel 00/0 : 49[1] -> 40[0] [send] via NET/FasTrak/0/GDRDMA +n116-041-181:72911:73023 [1] NCCL INFO Channel 09/0 : 49[1] -> 40[0] [send] via NET/FasTrak/1/GDRDMA +n116-041-181:72915:73026 [5] NCCL INFO Channel 05/0 : 61[5] -> 53[5] [receive] via NET/FasTrak/5/GDRDMA +n116-041-181:72911:73023 [1] NCCL INFO Channel 01/0 : 57[1] -> 49[1] [receive] via NET/FasTrak/1/GDRDMA +n116-041-181:72917:73037 [7] NCCL INFO [Proxy Progress] Device 7 CPU core 187 +n116-041-181:72917:73028 [7] NCCL INFO Channel 07/0 : 55[7] -> 63[7] [send] via NET/FasTrak/7/GDRDMA +n116-041-181:72913:73025 [3] NCCL INFO Channel 00/0 : 51[3] -> 50[2] via P2P/CUMEM +n116-041-181:72913:73025 [3] NCCL INFO Channel 01/0 : 51[3] -> 50[2] via P2P/CUMEM +n116-041-181:72913:73025 [3] NCCL INFO Channel 02/0 : 51[3] -> 50[2] via P2P/CUMEM +n116-041-181:72917:73028 [7] NCCL INFO Channel 06/0 : 46[6] -> 55[7] [receive] via NET/FasTrak/6/GDRDMA +n116-041-181:72913:73025 [3] NCCL INFO Channel 03/0 : 51[3] -> 50[2] via P2P/CUMEM +n116-041-181:72913:73025 [3] NCCL INFO Channel 04/0 : 51[3] -> 50[2] via P2P/CUMEM +n116-041-181:72913:73025 [3] NCCL INFO Channel 05/0 : 51[3] -> 50[2] via P2P/CUMEM +n116-041-181:72917:73028 [7] NCCL INFO Channel 15/0 : 46[6] -> 55[7] [receive] via NET/FasTrak/7/GDRDMA +n116-041-181:72915:73026 [5] NCCL INFO Channel 00/0 : 53[5] -> 52[4] via P2P/CUMEM +n116-041-181:72913:73025 [3] NCCL INFO Channel 06/0 : 51[3] -> 50[2] via P2P/CUMEM +n116-041-181:72915:73026 [5] NCCL INFO Channel 01/0 : 53[5] -> 52[4] via P2P/CUMEM +n116-041-181:72913:73025 [3] NCCL INFO Channel 07/0 : 51[3] -> 50[2] via P2P/CUMEM +n116-041-181:72915:73026 [5] NCCL INFO Channel 02/0 : 53[5] -> 52[4] via P2P/CUMEM +n116-041-181:72913:73025 [3] NCCL INFO Channel 08/0 : 51[3] -> 50[2] via P2P/CUMEM +n116-041-181:72915:73026 [5] NCCL INFO Channel 03/0 : 53[5] -> 52[4] via P2P/CUMEM +n116-041-181:72917:73028 [7] NCCL INFO Channel 07/0 : 39[7] -> 55[7] [receive] via NET/FasTrak/7/GDRDMA +n116-041-181:72913:73025 [3] NCCL INFO Channel 09/0 : 51[3] -> 50[2] via P2P/CUMEM +n116-041-181:72915:73026 [5] NCCL INFO Channel 04/0 : 53[5] -> 52[4] via P2P/CUMEM +n116-041-181:72917:73028 [7] NCCL INFO Channel 07/0 : 55[7] -> 39[7] [send] via NET/FasTrak/7/GDRDMA +n116-041-181:72913:73025 [3] NCCL INFO Channel 10/0 : 51[3] -> 50[2] via P2P/CUMEM +n116-041-181:72912:73029 [2] NCCL INFO Channel 00/0 : 50[2] -> 49[1] via P2P/CUMEM +n116-041-181:72915:73026 [5] NCCL INFO Channel 05/0 : 53[5] -> 52[4] via P2P/CUMEM +n116-041-181:72913:73025 [3] NCCL INFO Channel 11/0 : 51[3] -> 50[2] via P2P/CUMEM +n116-041-181:72912:73029 [2] NCCL INFO Channel 03/0 : 50[2] -> 49[1] via P2P/CUMEM +n116-041-181:72915:73026 [5] NCCL INFO Channel 06/0 : 53[5] -> 52[4] via P2P/CUMEM +n116-041-181:72913:73025 [3] NCCL INFO Channel 12/0 : 51[3] -> 50[2] via P2P/CUMEM +n116-041-181:72912:73029 [2] NCCL INFO Channel 04/0 : 50[2] -> 49[1] via P2P/CUMEM +n116-041-181:72915:73026 [5] NCCL INFO Channel 07/0 : 53[5] -> 52[4] via P2P/CUMEM +n116-041-181:72915:73026 [5] NCCL INFO Channel 08/0 : 53[5] -> 52[4] via P2P/CUMEM +n116-041-181:72915:73026 [5] NCCL INFO Channel 09/0 : 53[5] -> 52[4] via P2P/CUMEM +n116-041-181:72915:73026 [5] NCCL INFO Channel 10/0 : 53[5] -> 52[4] via P2P/CUMEM +n116-041-181:72915:73026 [5] NCCL INFO Channel 11/0 : 53[5] -> 52[4] via P2P/CUMEM +n116-041-181:72913:73025 [3] NCCL INFO Channel 13/0 : 51[3] -> 50[2] via P2P/CUMEM +n116-041-181:72915:73026 [5] NCCL INFO Channel 12/0 : 53[5] -> 52[4] via P2P/CUMEM +n116-041-181:72913:73025 [3] NCCL INFO Channel 14/0 : 51[3] -> 50[2] via P2P/CUMEM +n116-041-181:72915:73026 [5] NCCL INFO Channel 13/0 : 53[5] -> 52[4] via P2P/CUMEM +n116-041-181:72911:73023 [1] NCCL INFO Channel 00/0 : 49[1] -> 48[0] via P2P/CUMEM +n116-041-181:72913:73025 [3] NCCL INFO Channel 15/0 : 51[3] -> 50[2] via P2P/CUMEM +n116-041-181:72912:73029 [2] NCCL INFO Channel 05/0 : 50[2] -> 49[1] via P2P/CUMEM +n116-041-181:72915:73026 [5] NCCL INFO Channel 14/0 : 53[5] -> 52[4] via P2P/CUMEM +n116-041-181:72911:73023 [1] NCCL INFO Channel 01/0 : 49[1] -> 48[0] via P2P/CUMEM +n116-041-181:72912:73029 [2] NCCL INFO Channel 06/0 : 50[2] -> 49[1] via P2P/CUMEM +n116-041-181:72915:73026 [5] NCCL INFO Channel 15/0 : 53[5] -> 52[4] via P2P/CUMEM +n116-041-181:72911:73023 [1] NCCL INFO Channel 02/0 : 49[1] -> 48[0] via P2P/CUMEM +n116-041-181:72914:73027 [4] NCCL INFO Channel 00/0 : 52[4] -> 51[3] via P2P/CUMEM +n116-041-181:72912:73029 [2] NCCL INFO Channel 07/0 : 50[2] -> 49[1] via P2P/CUMEM +n116-041-181:72911:73023 [1] NCCL INFO Channel 03/0 : 49[1] -> 48[0] via P2P/CUMEM +n116-041-181:72914:73027 [4] NCCL INFO Channel 01/0 : 52[4] -> 51[3] via P2P/CUMEM +n116-041-181:72912:73029 [2] NCCL INFO Channel 08/0 : 50[2] -> 49[1] via P2P/CUMEM +n116-041-181:72911:73023 [1] NCCL INFO Channel 04/0 : 49[1] -> 48[0] via P2P/CUMEM +n116-041-181:72914:73027 [4] NCCL INFO Channel 02/0 : 52[4] -> 51[3] via P2P/CUMEM +n116-041-181:72912:73029 [2] NCCL INFO Channel 11/0 : 50[2] -> 49[1] via P2P/CUMEM +n116-041-181:72911:73023 [1] NCCL INFO Channel 05/0 : 49[1] -> 48[0] via P2P/CUMEM +n116-041-181:72914:73027 [4] NCCL INFO Channel 05/0 : 52[4] -> 51[3] via P2P/CUMEM +n116-041-181:72917:73028 [7] NCCL INFO Channel 06/0 : 55[7] -> 46[6] [send] via NET/FasTrak/6/GDRDMA +n116-041-181:72912:73029 [2] NCCL INFO Channel 12/0 : 50[2] -> 49[1] via P2P/CUMEM +n116-041-181:72917:73028 [7] NCCL INFO Channel 15/0 : 55[7] -> 46[6] [send] via NET/FasTrak/7/GDRDMA +n116-041-181:72911:73023 [1] NCCL INFO Channel 06/0 : 49[1] -> 48[0] via P2P/CUMEM +n116-041-181:72914:73027 [4] NCCL INFO Channel 06/0 : 52[4] -> 51[3] via P2P/CUMEM +n116-041-181:72912:73029 [2] NCCL INFO Channel 13/0 : 50[2] -> 49[1] via P2P/CUMEM +n116-041-181:72911:73023 [1] NCCL INFO Channel 07/0 : 49[1] -> 48[0] via P2P/CUMEM +n116-041-181:72914:73027 [4] NCCL INFO Channel 07/0 : 52[4] -> 51[3] via P2P/CUMEM +n116-041-181:72912:73029 [2] NCCL INFO Channel 14/0 : 50[2] -> 49[1] via P2P/CUMEM +n116-041-181:72911:73023 [1] NCCL INFO Channel 08/0 : 49[1] -> 48[0] via P2P/CUMEM +n116-041-181:72914:73027 [4] NCCL INFO Channel 08/0 : 52[4] -> 51[3] via P2P/CUMEM +n116-041-181:72912:73029 [2] NCCL INFO Channel 15/0 : 50[2] -> 49[1] via P2P/CUMEM +n116-041-181:72911:73023 [1] NCCL INFO Channel 09/0 : 49[1] -> 48[0] via P2P/CUMEM +n116-041-181:72914:73027 [4] NCCL INFO Channel 09/0 : 52[4] -> 51[3] via P2P/CUMEM +n116-041-181:72911:73023 [1] NCCL INFO Channel 10/0 : 49[1] -> 48[0] via P2P/CUMEM +n116-041-181:72914:73027 [4] NCCL INFO Channel 10/0 : 52[4] -> 51[3] via P2P/CUMEM +n116-041-181:72911:73023 [1] NCCL INFO Channel 11/0 : 49[1] -> 48[0] via P2P/CUMEM +n116-041-181:72914:73027 [4] NCCL INFO Channel 13/0 : 52[4] -> 51[3] via P2P/CUMEM +n116-041-181:72911:73023 [1] NCCL INFO Channel 12/0 : 49[1] -> 48[0] via P2P/CUMEM +n116-041-181:72914:73027 [4] NCCL INFO Channel 14/0 : 52[4] -> 51[3] via P2P/CUMEM +n116-041-181:72911:73023 [1] NCCL INFO Channel 13/0 : 49[1] -> 48[0] via P2P/CUMEM +n116-041-181:72914:73027 [4] NCCL INFO Channel 15/0 : 52[4] -> 51[3] via P2P/CUMEM +n116-041-181:72911:73023 [1] NCCL INFO Channel 14/0 : 49[1] -> 48[0] via P2P/CUMEM +n116-041-181:72911:73023 [1] NCCL INFO Channel 15/0 : 49[1] -> 48[0] via P2P/CUMEM +n116-041-181:72917:73028 [7] NCCL INFO Channel 07/0 : 63[7] -> 55[7] [receive] via NET/FasTrak/7/GDRDMA +n116-041-181:72917:73028 [7] NCCL INFO Channel 01/0 : 55[7] -> 48[0] via P2P/CUMEM +n116-041-181:72917:73028 [7] NCCL INFO Channel 02/0 : 55[7] -> 48[0] via P2P/CUMEM +n116-041-181:72917:73028 [7] NCCL INFO Channel 03/0 : 55[7] -> 48[0] via P2P/CUMEM +n116-041-181:72917:73028 [7] NCCL INFO Channel 04/0 : 55[7] -> 48[0] via P2P/CUMEM +n116-041-181:72917:73028 [7] NCCL INFO Channel 05/0 : 55[7] -> 48[0] via P2P/CUMEM +n116-041-181:72917:73028 [7] NCCL INFO Channel 06/0 : 55[7] -> 48[0] via P2P/CUMEM +n116-041-181:72917:73028 [7] NCCL INFO Channel 09/0 : 55[7] -> 48[0] via P2P/CUMEM +n116-041-181:72917:73028 [7] NCCL INFO Channel 10/0 : 55[7] -> 48[0] via P2P/CUMEM +n116-041-181:72917:73028 [7] NCCL INFO Channel 11/0 : 55[7] -> 48[0] via P2P/CUMEM +n116-041-181:72917:73028 [7] NCCL INFO Channel 12/0 : 55[7] -> 48[0] via P2P/CUMEM +n116-041-181:72917:73028 [7] NCCL INFO Channel 13/0 : 55[7] -> 48[0] via P2P/CUMEM +n116-041-181:72917:73028 [7] NCCL INFO Channel 14/0 : 55[7] -> 48[0] via P2P/CUMEM +n116-041-181:72917:73028 [7] NCCL INFO Channel 00/0 : 55[7] -> 54[6] via P2P/CUMEM +n116-041-181:72917:73028 [7] NCCL INFO Channel 01/0 : 55[7] -> 54[6] via P2P/CUMEM +n116-041-181:72917:73028 [7] NCCL INFO Channel 02/0 : 55[7] -> 54[6] via P2P/CUMEM +n116-041-181:72917:73028 [7] NCCL INFO Channel 03/0 : 55[7] -> 54[6] via P2P/CUMEM +n116-041-181:72917:73028 [7] NCCL INFO Channel 04/0 : 55[7] -> 54[6] via P2P/CUMEM +n116-041-181:72917:73028 [7] NCCL INFO Channel 05/0 : 55[7] -> 54[6] via P2P/CUMEM +n116-041-181:72917:73028 [7] NCCL INFO Channel 06/0 : 55[7] -> 54[6] via P2P/CUMEM +n116-041-181:72917:73028 [7] NCCL INFO Channel 07/0 : 55[7] -> 54[6] via P2P/CUMEM +n116-041-181:72917:73028 [7] NCCL INFO Channel 08/0 : 55[7] -> 54[6] via P2P/CUMEM +n116-041-181:72916:73024 [6] NCCL INFO Channel 00/0 : 54[6] -> 53[5] via P2P/CUMEM +n116-041-181:72917:73028 [7] NCCL INFO Channel 09/0 : 55[7] -> 54[6] via P2P/CUMEM +n116-041-181:72916:73024 [6] NCCL INFO Channel 01/0 : 54[6] -> 53[5] via P2P/CUMEM +n116-041-181:72917:73028 [7] NCCL INFO Channel 10/0 : 55[7] -> 54[6] via P2P/CUMEM +n116-041-181:72916:73024 [6] NCCL INFO Channel 02/0 : 54[6] -> 53[5] via P2P/CUMEM +n116-041-181:72917:73028 [7] NCCL INFO Channel 11/0 : 55[7] -> 54[6] via P2P/CUMEM +n116-041-181:72916:73024 [6] NCCL INFO Channel 03/0 : 54[6] -> 53[5] via P2P/CUMEM +n116-041-181:72917:73028 [7] NCCL INFO Channel 12/0 : 55[7] -> 54[6] via P2P/CUMEM +n116-041-181:72917:73028 [7] NCCL INFO Channel 13/0 : 55[7] -> 54[6] via P2P/CUMEM +n116-041-181:72916:73024 [6] NCCL INFO Channel 04/0 : 54[6] -> 53[5] via P2P/CUMEM +n116-041-181:72917:73028 [7] NCCL INFO Channel 14/0 : 55[7] -> 54[6] via P2P/CUMEM +n116-041-181:72917:73028 [7] NCCL INFO Channel 15/0 : 55[7] -> 54[6] via P2P/CUMEM +n116-041-181:72916:73024 [6] NCCL INFO Channel 07/0 : 54[6] -> 53[5] via P2P/CUMEM +n116-041-181:72916:73024 [6] NCCL INFO Channel 08/0 : 54[6] -> 53[5] via P2P/CUMEM +n116-041-181:72916:73024 [6] NCCL INFO Channel 09/0 : 54[6] -> 53[5] via P2P/CUMEM +n116-041-181:72916:73024 [6] NCCL INFO Channel 10/0 : 54[6] -> 53[5] via P2P/CUMEM +n116-041-181:72916:73024 [6] NCCL INFO Channel 11/0 : 54[6] -> 53[5] via P2P/CUMEM +n116-041-181:72916:73024 [6] NCCL INFO Channel 12/0 : 54[6] -> 53[5] via P2P/CUMEM +n116-041-181:72916:73024 [6] NCCL INFO Channel 15/0 : 54[6] -> 53[5] via P2P/CUMEM +n116-041-181:72912:73029 [2] NCCL INFO Connected all trees +n116-041-181:72913:73025 [3] NCCL INFO Connected all trees +n116-041-181:72917:73028 [7] NCCL INFO Connected all trees +n116-041-181:72910:73022 [0] NCCL INFO Connected all trees +n116-041-181:72911:73023 [1] NCCL INFO Connected all trees +n116-041-181:72914:73027 [4] NCCL INFO Connected all trees +n116-041-181:72915:73026 [5] NCCL INFO Connected all trees +n116-041-181:72916:73024 [6] NCCL INFO Connected all trees +n116-041-181:72913:73046 [3] NCCL INFO Channel 03/0 : 51[3] -> 52[4] via P2P/CUMEM +n116-041-181:72913:73046 [3] NCCL INFO Channel 11/0 : 51[3] -> 52[4] via P2P/CUMEM +n116-041-181:72913:73046 [3] NCCL INFO Channel 02/0 : 51[3] -> 58[2] [send] via NET/FasTrak/2/GDRDMA +n116-041-181:72913:73046 [3] NCCL INFO Channel 10/0 : 51[3] -> 58[2] [send] via NET/FasTrak/2/GDRDMA +n116-041-181:72914:73047 [4] NCCL INFO Channel 04/0 : 45[5] -> 52[4] [receive] via NET/FasTrak/4/GDRDMA +n116-041-181:72914:73047 [4] NCCL INFO Channel 05/0 : 52[4] -> 61[5] [send] via NET/FasTrak/5/GDRDMA +n116-041-181:72914:73047 [4] NCCL INFO Channel 13/0 : 52[4] -> 61[5] [send] via NET/FasTrak/5/GDRDMA +n116-041-181:72913:73046 [3] NCCL INFO Channel 03/0 : 42[2] -> 51[3] [receive] via NET/FasTrak/3/GDRDMA +n116-041-181:72917:73048 [7] NCCL INFO Channel 06/0 : 55[7] -> 62[6] [send] via NET/FasTrak/6/GDRDMA +n116-041-181:72917:73048 [7] NCCL INFO Channel 14/0 : 55[7] -> 62[6] [send] via NET/FasTrak/6/GDRDMA +n116-041-181:72915:73050 [5] NCCL INFO Channel 05/0 : 53[5] -> 54[6] via P2P/CUMEM +n116-041-181:72915:73050 [5] NCCL INFO Channel 13/0 : 53[5] -> 54[6] via P2P/CUMEM +n116-041-181:72915:73050 [5] NCCL INFO Channel 04/0 : 53[5] -> 60[4] [send] via NET/FasTrak/4/GDRDMA +n116-041-181:72915:73050 [5] NCCL INFO Channel 12/0 : 53[5] -> 60[4] [send] via NET/FasTrak/4/GDRDMA +n116-041-181:72915:73050 [5] NCCL INFO Channel 05/0 : 44[4] -> 53[5] [receive] via NET/FasTrak/5/GDRDMA +n116-041-181:72916:73049 [6] NCCL INFO Channel 06/0 : 47[7] -> 54[6] [receive] via NET/FasTrak/6/GDRDMA +n116-041-181:72916:73049 [6] NCCL INFO Channel 07/0 : 54[6] -> 63[7] [send] via NET/FasTrak/7/GDRDMA +n116-041-181:72916:73049 [6] NCCL INFO Channel 15/0 : 54[6] -> 63[7] [send] via NET/FasTrak/7/GDRDMA +n116-041-181:72911:73051 [1] NCCL INFO Channel 01/0 : 49[1] -> 50[2] via P2P/CUMEM +n116-041-181:72911:73051 [1] NCCL INFO Channel 09/0 : 49[1] -> 50[2] via P2P/CUMEM +n116-041-181:72910:73052 [0] NCCL INFO Channel 00/0 : 41[1] -> 48[0] [receive] via NET/FasTrak/0/GDRDMA +n116-041-181:72910:73052 [0] NCCL INFO Channel 00/0 : 48[0] -> 55[7] via P2P/CUMEM +n116-041-181:72910:73052 [0] NCCL INFO Channel 08/0 : 48[0] -> 55[7] via P2P/CUMEM +n116-041-181:72910:73052 [0] NCCL INFO Channel 01/0 : 48[0] -> 57[1] [send] via NET/FasTrak/1/GDRDMA +n116-041-181:72910:73052 [0] NCCL INFO Channel 09/0 : 48[0] -> 57[1] [send] via NET/FasTrak/1/GDRDMA +n116-041-181:72917:73048 [7] NCCL INFO Channel 07/0 : 46[6] -> 55[7] [receive] via NET/FasTrak/7/GDRDMA +n116-041-181:72911:73051 [1] NCCL INFO Channel 00/0 : 49[1] -> 56[0] [send] via NET/FasTrak/0/GDRDMA +n116-041-181:72911:73051 [1] NCCL INFO Channel 08/0 : 49[1] -> 56[0] [send] via NET/FasTrak/0/GDRDMA +n116-041-181:72912:73053 [2] NCCL INFO Channel 02/0 : 43[3] -> 50[2] [receive] via NET/FasTrak/2/GDRDMA +n116-041-181:72912:73053 [2] NCCL INFO Channel 03/0 : 50[2] -> 59[3] [send] via NET/FasTrak/3/GDRDMA +n116-041-181:72912:73053 [2] NCCL INFO Channel 11/0 : 50[2] -> 59[3] [send] via NET/FasTrak/3/GDRDMA +n116-041-181:72911:73051 [1] NCCL INFO Channel 01/0 : 40[0] -> 49[1] [receive] via NET/FasTrak/1/GDRDMA +n116-041-181:72917:73048 [7] NCCL INFO Channel 07/0 : 55[7] -> 48[0] via P2P/CUMEM +n116-041-181:72917:73048 [7] NCCL INFO Channel 15/0 : 55[7] -> 48[0] via P2P/CUMEM +n116-041-181:72916:73049 [6] NCCL INFO Channel 06/0 : 54[6] -> 53[5] via P2P/CUMEM +n116-041-181:72916:73049 [6] NCCL INFO Channel 14/0 : 54[6] -> 53[5] via P2P/CUMEM +n116-041-181:72914:73047 [4] NCCL INFO Channel 04/0 : 52[4] -> 51[3] via P2P/CUMEM +n116-041-181:72914:73047 [4] NCCL INFO Channel 12/0 : 52[4] -> 51[3] via P2P/CUMEM +n116-041-181:72912:73053 [2] NCCL INFO Channel 02/0 : 50[2] -> 49[1] via P2P/CUMEM +n116-041-181:72912:73053 [2] NCCL INFO Channel 10/0 : 50[2] -> 49[1] via P2P/CUMEM +n116-041-181:72917:73048 [7] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-041-181:72916:73049 [6] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-041-181:72915:73050 [5] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-041-181:72914:73047 [4] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-041-181:72913:73046 [3] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-041-181:72912:73053 [2] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-041-181:72911:73051 [1] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-041-181:72910:73052 [0] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +[rank50]:[W727 05:39:50.505443271 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n116-041-181.byted.org]:63166, remote=[n116-040-168.byted.org]:25264): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7fe279f785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7fe2636cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7fe2636cff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7fe2636d084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7fe2636ca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fe224dc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7fe27d7404a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7fe2803171f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7fe28039789c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank49]:[W727 05:39:50.505451408 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n116-041-181.byted.org]:63150, remote=[n116-040-168.byted.org]:25264): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f7abed785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f7aa84cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7f7aa84cff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7f7aa84d084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7f7aa84ca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f7a69bc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f7ac25454a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f7ac511c1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f7ac519c89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank49]:[W727 05:39:50.517661788 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 49] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank50]:[W727 05:39:50.517691496 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 50] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank51]:[W727 05:39:50.522538723 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n116-041-181.byted.org]:63178, remote=[n116-040-168.byted.org]:25264): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f14cb5785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f14b48cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7f14b48cff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7f14b48d084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7f14b48ca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f1475fc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f14cea2e4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f14d16051f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f14d168589c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank51]:[W727 05:39:50.530536129 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 51] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank54]:[W727 05:39:51.290637286 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n116-041-181.byted.org]:63164, remote=[n116-040-168.byted.org]:25264): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f3d2ed785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f3d184cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7f3d184cff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7f3d184d084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7f3d184ca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f3cd9bc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f3d324d14a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f3d350a81f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f3d3512889c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank54]:[W727 05:39:51.297385910 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 54] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank52]:[W727 05:39:51.291255784 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n116-041-181.byted.org]:63188, remote=[n116-040-168.byted.org]:25264): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f248bf785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f24752cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7f24752cff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7f24752d084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7f24752ca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f24369c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f248f3f14a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f2491fc81f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f249204889c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank52]:[W727 05:39:51.297867326 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 52] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank53]:[W727 05:39:51.291326808 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n116-041-181.byted.org]:63198, remote=[n116-040-168.byted.org]:25264): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7fcb237785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7fcb0cecdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7fcb0cecff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7fcb0ced084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7fcb0ceca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fcace5c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7fcb26efc4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7fcb29ad31f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7fcb29b5389c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank53]:[W727 05:39:51.298686780 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 53] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank48]:[W727 05:39:51.353119551 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=71, addr=[n116-041-181.byted.org]:63140, remote=[n116-040-168.byted.org]:25264): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f3d4f1785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f3d384cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7f3d384cff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7f3d384d084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7f3d384ca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f3cf9bc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f3d526094a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f3d551e01f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f3d5526089c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank48]:[W727 05:39:51.359400272 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 48] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank55]:[W727 05:39:51.357071752 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n116-041-181.byted.org]:63184, remote=[n116-040-168.byted.org]:25264): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f5addb785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f5ac72cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7f5ac72cff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7f5ac72d084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7f5ac72ca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f5a889c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f5ae130e4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f5ae3ee51f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f5ae3f6589c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank55]:[W727 05:39:51.363983478 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 55] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank50]:[W727 05:39:51.517859564 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-181.byted.org]:63166, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7fe279f785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7fe2636cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7fe2636cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7fe2636d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7fe2636ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fe224dc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7fe27d7404a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7fe2803171f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7fe28039789c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank50]:[W727 05:39:51.524828735 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 50] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank49]:[W727 05:39:51.517832723 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-181.byted.org]:63150, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f7abed785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f7aa84cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f7aa84cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f7aa84d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f7aa84ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f7a69bc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f7ac25454a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f7ac511c1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f7ac519c89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank49]:[W727 05:39:51.526098521 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 49] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank51]:[W727 05:39:51.530703019 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-181.byted.org]:63178, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f14cb5785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f14b48cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f14b48cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f14b48d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f14b48ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f1475fc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f14cea2e4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f14d16051f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f14d168589c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank51]:[W727 05:39:51.537094633 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 51] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank54]:[W727 05:39:52.297565091 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-181.byted.org]:63164, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f3d2ed785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f3d184cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f3d184cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f3d184d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f3d184ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f3cd9bc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f3d324d14a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f3d350a81f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f3d3512889c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank54]:[W727 05:39:52.304084546 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 54] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank52]:[W727 05:39:52.298042739 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-181.byted.org]:63188, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f248bf785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f24752cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f24752cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f24752d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f24752ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f24369c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f248f3f14a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f2491fc81f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f249204889c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank52]:[W727 05:39:52.304454900 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 52] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank53]:[W727 05:39:52.298881595 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-181.byted.org]:63198, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7fcb237785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7fcb0cecdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7fcb0cecf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7fcb0ced0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7fcb0ceca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fcace5c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7fcb26efc4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7fcb29ad31f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7fcb29b5389c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank53]:[W727 05:39:52.305756999 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 53] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank48]:[W727 05:39:52.359559907 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=71, addr=[n116-041-181.byted.org]:63140, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f3d4f1785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f3d384cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f3d384cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f3d384d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f3d384ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f3cf9bc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f3d526094a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f3d551e01f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f3d5526089c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank48]:[W727 05:39:52.365617146 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 48] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank55]:[W727 05:39:52.364090656 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-181.byted.org]:63184, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f5addb785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f5ac72cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f5ac72cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f5ac72d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f5ac72ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f5a889c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f5ae130e4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f5ae3ee51f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f5ae3f6589c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank55]:[W727 05:39:52.370566592 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 55] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank50]:[W727 05:39:52.524986427 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-181.byted.org]:63166, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7fe279f785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7fe2636cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7fe2636cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7fe2636d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7fe2636ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fe224dc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7fe27d7404a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7fe2803171f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7fe28039789c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank50]:[W727 05:39:52.532759571 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 50] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank49]:[W727 05:39:52.526261344 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-181.byted.org]:63150, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f7abed785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f7aa84cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f7aa84cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f7aa84d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f7aa84ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f7a69bc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f7ac25454a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f7ac511c1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f7ac519c89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank49]:[W727 05:39:52.534303553 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 49] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank51]:[W727 05:39:52.537257103 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-181.byted.org]:63178, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f14cb5785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f14b48cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f14b48cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f14b48d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f14b48ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f1475fc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f14cea2e4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f14d16051f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f14d168589c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank51]:[W727 05:39:52.543548716 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 51] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank54]:[W727 05:39:53.304263919 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-181.byted.org]:63164, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f3d2ed785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f3d184cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f3d184cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f3d184d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f3d184ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f3cd9bc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f3d324d14a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f3d350a81f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f3d3512889c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank54]:[W727 05:39:53.310625305 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 54] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank52]:[W727 05:39:53.304624053 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-181.byted.org]:63188, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f248bf785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f24752cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f24752cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f24752d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f24752ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f24369c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f248f3f14a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f2491fc81f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f249204889c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank52]:[W727 05:39:53.310807669 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 52] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank53]:[W727 05:39:53.305939458 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-181.byted.org]:63198, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7fcb237785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7fcb0cecdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7fcb0cecf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7fcb0ced0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7fcb0ceca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fcace5c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7fcb26efc4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7fcb29ad31f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7fcb29b5389c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank53]:[W727 05:39:53.313526640 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 53] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank48]:[W727 05:39:53.365740053 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=71, addr=[n116-041-181.byted.org]:63140, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f3d4f1785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f3d384cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f3d384cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f3d384d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f3d384ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f3cf9bc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f3d526094a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f3d551e01f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f3d5526089c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank48]:[W727 05:39:53.371860051 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 48] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank55]:[W727 05:39:53.370718457 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-181.byted.org]:63184, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f5addb785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f5ac72cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f5ac72cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f5ac72d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f5ac72ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f5a889c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f5ae130e4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f5ae3ee51f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f5ae3f6589c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank55]:[W727 05:39:53.377399038 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 55] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank50]:[W727 05:39:53.532968116 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-181.byted.org]:63166, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7fe279f785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7fe2636cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7fe2636cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7fe2636d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7fe2636ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fe224dc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7fe27d7404a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7fe2803171f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7fe28039789c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank50]:[W727 05:39:53.539211612 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 50] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank49]:[W727 05:39:53.534479422 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-181.byted.org]:63150, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f7abed785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f7aa84cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f7aa84cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f7aa84d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f7aa84ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f7a69bc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f7ac25454a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f7ac511c1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f7ac519c89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank49]:[W727 05:39:53.541768988 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 49] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank51]:[W727 05:39:53.543706774 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-181.byted.org]:63178, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f14cb5785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f14b48cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f14b48cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f14b48d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f14b48ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f1475fc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f14cea2e4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f14d16051f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f14d168589c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank51]:[W727 05:39:53.549858417 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 51] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank52]:[W727 05:39:54.310980693 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-181.byted.org]:63188, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f248bf785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f24752cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f24752cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f24752d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f24752ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f24369c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f248f3f14a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f2491fc81f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f249204889c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank52]:[W727 05:39:54.317442227 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 52] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank54]:[W727 05:39:54.310800508 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-181.byted.org]:63164, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f3d2ed785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f3d184cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f3d184cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f3d184d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f3d184ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f3cd9bc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f3d324d14a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f3d350a81f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f3d3512889c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank54]:[W727 05:39:54.317973701 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 54] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank53]:[W727 05:39:54.313686372 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-181.byted.org]:63198, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7fcb237785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7fcb0cecdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7fcb0cecf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7fcb0ced0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7fcb0ceca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fcace5c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7fcb26efc4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7fcb29ad31f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7fcb29b5389c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank53]:[W727 05:39:54.320461685 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 53] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank48]:[W727 05:39:54.372025868 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=71, addr=[n116-041-181.byted.org]:63140, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f3d4f1785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f3d384cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f3d384cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f3d384d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f3d384ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f3cf9bc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f3d526094a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f3d551e01f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f3d5526089c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank48]:[W727 05:39:54.378228066 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 48] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank55]:[W727 05:39:54.377545093 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-181.byted.org]:63184, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f5addb785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f5ac72cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f5ac72cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f5ac72d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f5ac72ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f5a889c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f5ae130e4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f5ae3ee51f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f5ae3f6589c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank55]:[W727 05:39:54.383647567 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 55] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank50]:[W727 05:39:54.539376058 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-181.byted.org]:63166, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7fe279f785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7fe2636cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7fe2636cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7fe2636d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7fe2636ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fe224dc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7fe27d7404a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7fe2803171f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7fe28039789c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank50]:[W727 05:39:54.545547568 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 50] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank49]:[W727 05:39:54.541937356 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-181.byted.org]:63150, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f7abed785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f7aa84cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f7aa84cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f7aa84d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f7aa84ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f7a69bc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f7ac25454a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f7ac511c1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f7ac519c89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank49]:[W727 05:39:54.550072219 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 49] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank51]:[W727 05:39:54.550042504 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-181.byted.org]:63178, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f14cb5785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f14b48cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f14b48cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f14b48d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f14b48ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f1475fc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f14cea2e4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f14d16051f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f14d168589c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank51]:[W727 05:39:54.556299246 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 51] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank52]:[W727 05:39:55.317622950 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-181.byted.org]:63188, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f248bf785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f24752cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f24752cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f24752d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f24752ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f24369c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f248f3f14a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f2491fc81f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f249204889c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank52]:[W727 05:39:55.324076955 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 52] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank54]:[W727 05:39:55.318173273 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-181.byted.org]:63164, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f3d2ed785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f3d184cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f3d184cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f3d184d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f3d184ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f3cd9bc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f3d324d14a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f3d350a81f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f3d3512889c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank54]:[W727 05:39:55.326078793 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 54] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank53]:[W727 05:39:55.320641221 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-181.byted.org]:63198, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7fcb237785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7fcb0cecdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7fcb0cecf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7fcb0ced0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7fcb0ceca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fcace5c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7fcb26efc4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7fcb29ad31f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7fcb29b5389c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank53]:[W727 05:39:55.328553386 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 53] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank48]:[W727 05:39:55.378386675 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=71, addr=[n116-041-181.byted.org]:63140, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f3d4f1785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f3d384cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f3d384cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f3d384d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f3d384ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f3cf9bc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f3d526094a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f3d551e01f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f3d5526089c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank48]:[W727 05:39:55.384597021 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 48] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank55]:[W727 05:39:55.383811487 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-181.byted.org]:63184, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f5addb785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f5ac72cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f5ac72cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f5ac72d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f5ac72ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f5a889c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f5ae130e4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f5ae3ee51f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f5ae3f6589c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank55]:[W727 05:39:55.390092527 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 55] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank50]:[W727 05:39:55.545720827 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-181.byted.org]:63166, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7fe279f785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7fe2636cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7fe2636cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7fe2636d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7fe2636ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fe224dc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7fe27d7404a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7fe2803171f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7fe28039789c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank50]:[W727 05:39:55.551766242 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 50] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank49]:[W727 05:39:55.550246949 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-181.byted.org]:63150, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f7abed785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f7aa84cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f7aa84cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f7aa84d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f7aa84ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f7a69bc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f7ac25454a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f7ac511c1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f7ac519c89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank49]:[W727 05:39:55.557253710 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 49] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank51]:[W727 05:39:55.556462507 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-181.byted.org]:63178, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f14cb5785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f14b48cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f14b48cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f14b48d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f14b48ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f1475fc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f14cea2e4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f14d16051f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f14d168589c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank51]:[W727 05:39:55.562405496 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 51] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank52]:[W727 05:39:56.324261975 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-181.byted.org]:63188, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f248bf785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f24752cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f24752cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f24752d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f24752ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f24369c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f248f3f14a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f2491fc81f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f249204889c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank52]:[W727 05:39:56.330317502 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 52] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank54]:[W727 05:39:56.326242962 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-181.byted.org]:63164, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f3d2ed785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f3d184cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f3d184cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f3d184d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f3d184ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f3cd9bc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f3d324d14a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f3d350a81f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f3d3512889c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank54]:[W727 05:39:56.333375405 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 54] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank53]:[W727 05:39:56.328727788 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-181.byted.org]:63198, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7fcb237785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7fcb0cecdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7fcb0cecf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7fcb0ced0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7fcb0ceca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fcace5c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7fcb26efc4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7fcb29ad31f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7fcb29b5389c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank53]:[W727 05:39:56.335836406 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 53] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank48]:[W727 05:39:56.384765531 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=71, addr=[n116-041-181.byted.org]:63140, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f3d4f1785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f3d384cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f3d384cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f3d384d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f3d384ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f3cf9bc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f3d526094a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f3d551e01f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f3d5526089c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank48]:[W727 05:39:56.391061099 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 48] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank55]:[W727 05:39:56.390247929 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-181.byted.org]:63184, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f5addb785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f5ac72cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f5ac72cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f5ac72d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f5ac72ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f5a889c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f5ae130e4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f5ae3ee51f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f5ae3f6589c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank55]:[W727 05:39:56.396543698 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 55] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank50]:[W727 05:39:56.551941969 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-181.byted.org]:63166, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7fe279f785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7fe2636cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7fe2636cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7fe2636d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7fe2636ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7fe224dc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7fe27d7404a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7fe2803171f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7fe28039789c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank50]:[W727 05:39:56.558650154 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 50] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank49]:[W727 05:39:56.557422771 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-181.byted.org]:63150, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f7abed785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f7aa84cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f7aa84cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f7aa84d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f7aa84ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f7a69bc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f7ac25454a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f7ac511c1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f7ac519c89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank49]:[W727 05:39:56.565203520 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 49] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank51]:[W727 05:39:56.562574382 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-041-181.byted.org]:63178, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f14cb5785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f14b48cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f14b48cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f14b48d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f14b48ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f1475fc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f14cea2e4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f14d16051f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f14d168589c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank51]:[W727 05:39:56.568640226 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 51] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +E0727 05:39:57.142000 72845 torch/distributed/elastic/multiprocessing/api.py:874] failed (exitcode: -15) local_rank: 0 (pid: 72910) of binary: /usr/bin/python3 +Traceback (most recent call last): + File "/usr/local/bin/torchrun", line 7, in + sys.exit(main()) + ^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 355, in wrapper + return f(*args, **kwargs) + ^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 892, in main + run(args) + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 883, in run + elastic_launch( + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 139, in __call__ + return launch_agent(self._config, self._entrypoint, list(args)) + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 270, in launch_agent + raise ChildFailedError( +torch.distributed.elastic.multiprocessing.errors.ChildFailedError: +============================================================ +main_cross_2direct_balance_realscale_perturb_v4_accum.py FAILED +------------------------------------------------------------ +Failures: +[1]: + time : 2026-07-27_05:39:57 + host : n116-041-181.byted.org + rank : 49 (local_rank: 1) + exitcode : -15 (pid: 72911) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 72911 +[2]: + time : 2026-07-27_05:39:57 + host : n116-041-181.byted.org + rank : 50 (local_rank: 2) + exitcode : -15 (pid: 72912) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 72912 +[3]: + time : 2026-07-27_05:39:57 + host : n116-041-181.byted.org + rank : 51 (local_rank: 3) + exitcode : -15 (pid: 72913) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 72913 +[4]: + time : 2026-07-27_05:39:57 + host : n116-041-181.byted.org + rank : 52 (local_rank: 4) + exitcode : -15 (pid: 72914) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 72914 +[5]: + time : 2026-07-27_05:39:57 + host : n116-041-181.byted.org + rank : 53 (local_rank: 5) + exitcode : -15 (pid: 72915) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 72915 +[6]: + time : 2026-07-27_05:39:57 + host : n116-041-181.byted.org + rank : 54 (local_rank: 6) + exitcode : -15 (pid: 72916) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 72916 +[7]: + time : 2026-07-27_05:39:57 + host : n116-041-181.byted.org + rank : 55 (local_rank: 7) + exitcode : -15 (pid: 72917) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 72917 +------------------------------------------------------------ +Root Cause (first observed failure): +[0]: + time : 2026-07-27_05:39:57 + host : n116-041-181.byted.org + rank : 48 (local_rank: 0) + exitcode : -15 (pid: 72910) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 72910 +============================================================ diff --git a/ScaleDF_rebuttal/_host_logs/20260726_2350_vith64_free8_port25264_launch/launcher_rank7_n116-042-235.log b/ScaleDF_rebuttal/_host_logs/20260726_2350_vith64_free8_port25264_launch/launcher_rank7_n116-042-235.log new file mode 100644 index 0000000000000000000000000000000000000000..94e4ccb494bda9daa7ae0400caa910ffe2b56624 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/20260726_2350_vith64_free8_port25264_launch/launcher_rank7_n116-042-235.log @@ -0,0 +1,2455 @@ +started_utc=2026-07-26T23:50:29Z +hostname=n116-042-235 +node_rank=7 +nnodes=8 +nproc_per_node=8 +world_size=64 +master_addr=10.116.40.168 +master_port=25264 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +main=main_cross_2direct_balance_realscale_perturb_v4_accum.py +model_name=deit_huge_patch14_LS +pretrain=/tmp/ScaleDF_repro_20260724/pretrained/deit_3_huge_224_21k_v1.pth +resume=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/checkpoint_snapshots/vith32_latest_for_64card_effbs4096_20260726_234604.pth +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/modelsize_vith_40epoch_64card_effbs4096_from_epoch22_snapshot_free8_master_n116-040-168_20260726_2350_vith64_port25264 +batch_per_gpu=64 +accum_iter=1 +effective_batch=4096 +eval_batch_per_gpu=64 +epochs=40 +W0726 23:50:37.414000 5854 torch/distributed/run.py:766] +W0726 23:50:37.414000 5854 torch/distributed/run.py:766] ***************************************** +W0726 23:50:37.414000 5854 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 23:50:37.414000 5854 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 56): env:// +| distributed init (rank 63): env:// +[rank63]:[W726 23:51:09.339191900 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 63] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank56]:[W726 23:51:09.389717703 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 56] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 60): env:// +[rank60]:[W726 23:51:09.571555637 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 60] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 57): env:// +[rank57]:[W726 23:51:09.699868209 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 57] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 58): env:// +[rank58]:[W726 23:51:10.942987181 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 58] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 62): env:// +[rank62]:[W726 23:51:10.962828272 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 62] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 61): env:// +| distributed init (rank 59): env:// +[rank61]:[W726 23:51:10.978188193 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 61] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank59]:[W726 23:51:10.978212143 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 59] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +n116-042-235:5926:5926 [7] NCCL INFO cudaDriverVersion 12060 +n116-042-235:5926:5926 [7] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-042-235:5926:5926 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-042-235:5926:5926 [7] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-042-235:5921:5921 [2] NCCL INFO cudaDriverVersion 12060 +n116-042-235:5921:5921 [2] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-042-235:5921:5921 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-042-235:5921:5921 [2] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-042-235:5926:5926 [7] NCCL INFO Comm config Blocking set to 1 +n116-042-235:5921:5921 [2] NCCL INFO Comm config Blocking set to 1 +n116-042-235:5924:5924 [5] NCCL INFO cudaDriverVersion 12060 +n116-042-235:5924:5924 [5] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-042-235:5924:5924 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-042-235:5924:5924 [5] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-042-235:5924:5924 [5] NCCL INFO Comm config Blocking set to 1 +n116-042-235:5925:5925 [6] NCCL INFO cudaDriverVersion 12060 +n116-042-235:5925:5925 [6] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-042-235:5925:5925 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-042-235:5925:5925 [6] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-042-235:5925:5925 [6] NCCL INFO Comm config Blocking set to 1 +n116-042-235:5919:5919 [0] NCCL INFO cudaDriverVersion 12060 +n116-042-235:5919:5919 [0] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-042-235:5919:5919 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-042-235:5919:5919 [0] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-042-235:5919:5919 [0] NCCL INFO Comm config Blocking set to 1 +n116-042-235:5923:5923 [4] NCCL INFO cudaDriverVersion 12060 +n116-042-235:5923:5923 [4] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-042-235:5923:5923 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-042-235:5923:5923 [4] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-042-235:5923:5923 [4] NCCL INFO Comm config Blocking set to 1 +n116-042-235:5922:5922 [3] NCCL INFO cudaDriverVersion 12060 +n116-042-235:5922:5922 [3] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-042-235:5922:5922 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-042-235:5920:5920 [1] NCCL INFO cudaDriverVersion 12060 +n116-042-235:5920:5920 [1] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET +n116-042-235:5920:5920 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to eth0 +n116-042-235:5922:5922 [3] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-042-235:5920:5920 [1] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-042-235:5922:5922 [3] NCCL INFO Comm config Blocking set to 1 +n116-042-235:5920:5920 [1] NCCL INFO Comm config Blocking set to 1 +n116-042-235:5926:5975 [7] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-042-235:5926:5975 [7] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-042-235:5926:5975 [7] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-042-235:5926:5975 [7] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-042-235:5926:5975 [7] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-042-235:5926:5975 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-042-235:5926:5975 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-042-235:5926:5975 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-042-235:5926:5975 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-042-235:5926:5975 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-042-235:5921:5976 [2] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-042-235:5921:5976 [2] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-042-235:5921:5976 [2] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-042-235:5921:5976 [2] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-042-235:5921:5976 [2] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-042-235:5921:5976 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-042-235:5921:5976 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-042-235:5921:5976 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-042-235:5921:5976 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-042-235:5921:5976 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-042-235:5924:5977 [5] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-042-235:5924:5977 [5] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-042-235:5924:5977 [5] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-042-235:5924:5977 [5] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-042-235:5924:5977 [5] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-042-235:5924:5977 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-042-235:5924:5977 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-042-235:5924:5977 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-042-235:5924:5977 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-042-235:5924:5977 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-042-235:5925:5978 [6] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-042-235:5925:5978 [6] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-042-235:5925:5978 [6] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-042-235:5925:5978 [6] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-042-235:5925:5978 [6] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-042-235:5925:5978 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-042-235:5925:5978 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-042-235:5925:5978 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-042-235:5925:5978 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-042-235:5925:5978 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-042-235:5919:5979 [0] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-042-235:5919:5979 [0] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-042-235:5919:5979 [0] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-042-235:5919:5979 [0] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-042-235:5919:5979 [0] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-042-235:5919:5979 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-042-235:5919:5979 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-042-235:5919:5979 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-042-235:5919:5979 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-042-235:5919:5979 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-042-235:5923:5980 [4] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-042-235:5923:5980 [4] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-042-235:5923:5980 [4] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-042-235:5923:5980 [4] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-042-235:5923:5980 [4] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-042-235:5923:5980 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-042-235:5923:5980 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-042-235:5923:5980 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-042-235:5923:5980 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-042-235:5923:5980 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-042-235:5922:5981 [3] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-042-235:5922:5981 [3] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-042-235:5922:5981 [3] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-042-235:5922:5981 [3] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-042-235:5922:5981 [3] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-042-235:5922:5981 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-042-235:5922:5981 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-042-235:5922:5981 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-042-235:5922:5981 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-042-235:5922:5981 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-042-235:5926:5975 [7] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-042-235:5926:5975 [7] NCCL INFO Using network FasTrak +n116-042-235:5921:5976 [2] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-042-235:5921:5976 [2] NCCL INFO Using network FasTrak +n116-042-235:5924:5977 [5] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-042-235:5924:5977 [5] NCCL INFO Using network FasTrak +n116-042-235:5925:5978 [6] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-042-235:5925:5978 [6] NCCL INFO Using network FasTrak +n116-042-235:5919:5979 [0] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-042-235:5919:5979 [0] NCCL INFO Using network FasTrak +n116-042-235:5920:5982 [1] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-042-235:5920:5982 [1] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-042-235:5920:5982 [1] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-042-235:5920:5982 [1] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-042-235:5920:5982 [1] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-042-235:5920:5982 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-042-235:5920:5982 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-042-235:5920:5982 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-042-235:5920:5982 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-042-235:5920:5982 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-042-235:5923:5980 [4] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-042-235:5923:5980 [4] NCCL INFO Using network FasTrak +n116-042-235:5922:5981 [3] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-042-235:5922:5981 [3] NCCL INFO Using network FasTrak +n116-042-235:5920:5982 [1] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-042-235:5920:5982 [1] NCCL INFO Using network FasTrak +n116-042-235:5921:5976 [2] NCCL INFO DMA-BUF is available on GPU device 2 +n116-042-235:5926:5975 [7] NCCL INFO DMA-BUF is available on GPU device 7 +n116-042-235:5921:5976 [2] NCCL INFO ncclCommInitRankConfig comm 0xc5cd980 rank 58 nranks 64 cudaDev 2 nvmlDev 2 busId b000 commId 0xfb2d11625d05007a - Init START +n116-042-235:5926:5975 [7] NCCL INFO ncclCommInitRankConfig comm 0xc302a60 rank 63 nranks 64 cudaDev 7 nvmlDev 7 busId 8c000 commId 0xfb2d11625d05007a - Init START +n116-042-235:5924:5977 [5] NCCL INFO DMA-BUF is available on GPU device 5 +n116-042-235:5925:5978 [6] NCCL INFO DMA-BUF is available on GPU device 6 +n116-042-235:5924:5977 [5] NCCL INFO ncclCommInitRankConfig comm 0xd08cdf0 rank 61 nranks 64 cudaDev 5 nvmlDev 5 busId 85000 commId 0xfb2d11625d05007a - Init START +n116-042-235:5919:5979 [0] NCCL INFO DMA-BUF is available on GPU device 0 +n116-042-235:5925:5978 [6] NCCL INFO ncclCommInitRankConfig comm 0xb61dba0 rank 62 nranks 64 cudaDev 6 nvmlDev 6 busId 8b000 commId 0xfb2d11625d05007a - Init START +n116-042-235:5925:5978 [6] NCCL INFO RAS client listening socket at ::1<28028> +n116-042-235:5926:5975 [7] NCCL INFO RAS client listening socket at ::1<28028> +n116-042-235:5919:5979 [0] NCCL INFO ncclCommInitRankConfig comm 0xcb319f0 rank 56 nranks 64 cudaDev 0 nvmlDev 0 busId 4000 commId 0xfb2d11625d05007a - Init START +n116-042-235:5923:5980 [4] NCCL INFO DMA-BUF is available on GPU device 4 +n116-042-235:5923:5980 [4] NCCL INFO ncclCommInitRankConfig comm 0xb3ebb20 rank 60 nranks 64 cudaDev 4 nvmlDev 4 busId 84000 commId 0xfb2d11625d05007a - Init START +n116-042-235:5924:5977 [5] NCCL INFO RAS client listening socket at ::1<28028> +n116-042-235:5920:5982 [1] NCCL INFO DMA-BUF is available on GPU device 1 +n116-042-235:5922:5981 [3] NCCL INFO DMA-BUF is available on GPU device 3 +n116-042-235:5920:5982 [1] NCCL INFO ncclCommInitRankConfig comm 0xb1bcf90 rank 57 nranks 64 cudaDev 1 nvmlDev 1 busId 5000 commId 0xfb2d11625d05007a - Init START +n116-042-235:5922:5981 [3] NCCL INFO ncclCommInitRankConfig comm 0xc5c7cf0 rank 59 nranks 64 cudaDev 3 nvmlDev 3 busId c000 commId 0xfb2d11625d05007a - Init START +n116-042-235:5921:5976 [2] NCCL INFO RAS client listening socket at ::1<28028> +n116-042-235:5920:5982 [1] NCCL INFO RAS client listening socket at ::1<28028> +n116-042-235:5922:5981 [3] NCCL INFO RAS client listening socket at ::1<28028> +n116-042-235:5923:5980 [4] NCCL INFO RAS client listening socket at ::1<28028> +n116-042-235:5919:5979 [0] NCCL INFO RAS client listening socket at ::1<28028> +n116-042-235:5921:5976 [2] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-042-235:5921:5976 [2] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-042-235:5921:5976 [2] NCCL INFO Retrieving state for FasTrak +n116-042-235:5921:5976 [2] NCCL INFO Initialized state 0 for FasTrak +n116-042-235:5921:5976 [2] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-042-235:5921:5976 [2] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-042-235:5921:5976 [2] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-042-235:5921:5976 [2] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-042-235:5925:5978 [6] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-042-235:5925:5978 [6] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-042-235:5925:5978 [6] NCCL INFO Retrieving state for FasTrak +n116-042-235:5925:5978 [6] NCCL INFO Initialized state 0 for FasTrak +n116-042-235:5921:5976 [2] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-042-235:5921:5976 [2] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-042-235:5925:5978 [6] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-042-235:5925:5978 [6] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-042-235:5921:5976 [2] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-042-235:5921:5976 [2] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-042-235:5925:5978 [6] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-042-235:5922:5981 [3] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-042-235:5922:5981 [3] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-042-235:5922:5981 [3] NCCL INFO Retrieving state for FasTrak +n116-042-235:5922:5981 [3] NCCL INFO Initialized state 0 for FasTrak +n116-042-235:5925:5978 [6] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-042-235:5925:5978 [6] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-042-235:5922:5981 [3] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-042-235:5925:5978 [6] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-042-235:5922:5981 [3] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-042-235:5925:5978 [6] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-042-235:5922:5981 [3] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-042-235:5925:5978 [6] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-042-235:5922:5981 [3] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-042-235:5926:5975 [7] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-042-235:5926:5975 [7] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-042-235:5926:5975 [7] NCCL INFO Retrieving state for FasTrak +n116-042-235:5926:5975 [7] NCCL INFO Initialized state 0 for FasTrak +n116-042-235:5922:5981 [3] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-042-235:5922:5981 [3] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-042-235:5926:5975 [7] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-042-235:5922:5981 [3] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-042-235:5926:5975 [7] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-042-235:5922:5981 [3] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-042-235:5926:5975 [7] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-042-235:5920:5982 [1] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-042-235:5920:5982 [1] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-042-235:5920:5982 [1] NCCL INFO Retrieving state for FasTrak +n116-042-235:5920:5982 [1] NCCL INFO Initialized state 0 for FasTrak +n116-042-235:5926:5975 [7] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-042-235:5920:5982 [1] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-042-235:5926:5975 [7] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-042-235:5920:5982 [1] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-042-235:5926:5975 [7] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-042-235:5926:5975 [7] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-042-235:5923:5980 [4] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-042-235:5923:5980 [4] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-042-235:5923:5980 [4] NCCL INFO Retrieving state for FasTrak +n116-042-235:5923:5980 [4] NCCL INFO Initialized state 0 for FasTrak +n116-042-235:5920:5982 [1] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-042-235:5926:5975 [7] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-042-235:5920:5982 [1] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-042-235:5923:5980 [4] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-042-235:5923:5980 [4] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-042-235:5920:5982 [1] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-042-235:5924:5977 [5] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-042-235:5924:5977 [5] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-042-235:5924:5977 [5] NCCL INFO Retrieving state for FasTrak +n116-042-235:5924:5977 [5] NCCL INFO Initialized state 0 for FasTrak +n116-042-235:5923:5980 [4] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-042-235:5920:5982 [1] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-042-235:5923:5980 [4] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-042-235:5924:5977 [5] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-042-235:5923:5980 [4] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-042-235:5920:5982 [1] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-042-235:5924:5977 [5] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-042-235:5923:5980 [4] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-042-235:5920:5982 [1] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-042-235:5924:5977 [5] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-042-235:5923:5980 [4] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-042-235:5924:5977 [5] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-042-235:5923:5980 [4] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-042-235:5924:5977 [5] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-042-235:5924:5977 [5] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-042-235:5924:5977 [5] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-042-235:5924:5977 [5] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-042-235:5919:5979 [0] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-042-235:5919:5979 [0] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-042-235:5919:5979 [0] NCCL INFO Retrieving state for FasTrak +n116-042-235:5919:5979 [0] NCCL INFO Initialized state 0 for FasTrak +n116-042-235:5919:5979 [0] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-042-235:5919:5979 [0] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-042-235:5919:5979 [0] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-042-235:5919:5979 [0] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-042-235:5919:5979 [0] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-042-235:5919:5979 [0] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-042-235:5919:5979 [0] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-042-235:5919:5979 [0] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-042-235:5919:5979 [0] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-042-235:5925:5978 [6] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-042-235:5922:5981 [3] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-042-235:5926:5975 [7] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-042-235:5924:5977 [5] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-042-235:5921:5976 [2] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-042-235:5923:5980 [4] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-042-235:5920:5982 [1] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-042-235:5919:5979 [0] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-042-235:5919:5979 [0] NCCL INFO CPU/0-0 (1/1/3) +n116-042-235:5919:5979 [0] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-042-235:5919:5979 [0] NCCL INFO + PCI[24.0] - GPU/0-4000 (56) +n116-042-235:5919:5979 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5919:5979 [0] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-042-235:5919:5979 [0] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-042-235:5919:5979 [0] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-042-235:5919:5979 [0] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-042-235:5919:5979 [0] NCCL INFO + PCI[24.0] - GPU/0-5000 (57) +n116-042-235:5919:5979 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5919:5979 [0] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-042-235:5919:5979 [0] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-042-235:5919:5979 [0] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-042-235:5919:5979 [0] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-042-235:5919:5979 [0] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-042-235:5919:5979 [0] NCCL INFO + PCI[24.0] - GPU/0-b000 (58) +n116-042-235:5919:5979 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5919:5979 [0] NCCL INFO + PCI[24.0] - GPU/0-c000 (59) +n116-042-235:5919:5979 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5919:5979 [0] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-042-235:5919:5979 [0] NCCL INFO CPU/0-1 (1/1/3) +n116-042-235:5919:5979 [0] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-042-235:5919:5979 [0] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-042-235:5919:5979 [0] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-042-235:5919:5979 [0] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-042-235:5919:5979 [0] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-042-235:5919:5979 [0] NCCL INFO + PCI[24.0] - GPU/0-84000 (60) +n116-042-235:5919:5979 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5919:5979 [0] NCCL INFO + PCI[24.0] - GPU/0-85000 (61) +n116-042-235:5919:5979 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5919:5979 [0] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-042-235:5919:5979 [0] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-042-235:5919:5979 [0] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-042-235:5919:5979 [0] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-042-235:5919:5979 [0] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-042-235:5919:5979 [0] NCCL INFO + PCI[24.0] - GPU/0-8b000 (62) +n116-042-235:5919:5979 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5919:5979 [0] NCCL INFO + PCI[24.0] - GPU/0-8c000 (63) +n116-042-235:5919:5979 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5919:5979 [0] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-042-235:5919:5979 [0] NCCL INFO ========================================== +n116-042-235:5919:5979 [0] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-235:5919:5979 [0] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-235:5919:5979 [0] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-235:5919:5979 [0] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-235:5919:5979 [0] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-235:5919:5979 [0] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-235:5919:5979 [0] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-042-235:5919:5979 [0] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-042-235:5919:5979 [0] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-235:5919:5979 [0] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-235:5926:5975 [7] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-042-235:5926:5975 [7] NCCL INFO CPU/0-0 (1/1/3) +n116-042-235:5926:5975 [7] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-042-235:5919:5979 [0] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-235:5926:5975 [7] NCCL INFO + PCI[24.0] - GPU/0-4000 (56) +n116-042-235:5926:5975 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5926:5975 [7] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-042-235:5926:5975 [7] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-042-235:5919:5979 [0] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-235:5926:5975 [7] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-042-235:5926:5975 [7] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-042-235:5926:5975 [7] NCCL INFO + PCI[24.0] - GPU/0-5000 (57) +n116-042-235:5926:5975 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5926:5975 [7] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-042-235:5926:5975 [7] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-042-235:5926:5975 [7] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-042-235:5919:5979 [0] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-042-235:5926:5975 [7] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-042-235:5926:5975 [7] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-042-235:5926:5975 [7] NCCL INFO + PCI[24.0] - GPU/0-b000 (58) +n116-042-235:5926:5975 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5926:5975 [7] NCCL INFO + PCI[24.0] - GPU/0-c000 (59) +n116-042-235:5926:5975 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5926:5975 [7] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-042-235:5926:5975 [7] NCCL INFO CPU/0-1 (1/1/3) +n116-042-235:5926:5975 [7] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-042-235:5919:5979 [0] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-042-235:5926:5975 [7] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-042-235:5926:5975 [7] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-042-235:5926:5975 [7] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-042-235:5926:5975 [7] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-042-235:5926:5975 [7] NCCL INFO + PCI[24.0] - GPU/0-84000 (60) +n116-042-235:5926:5975 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5926:5975 [7] NCCL INFO + PCI[24.0] - GPU/0-85000 (61) +n116-042-235:5926:5975 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5926:5975 [7] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-042-235:5926:5975 [7] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-042-235:5919:5979 [0] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-042-235:5926:5975 [7] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-042-235:5926:5975 [7] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-042-235:5926:5975 [7] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-042-235:5926:5975 [7] NCCL INFO + PCI[24.0] - GPU/0-8b000 (62) +n116-042-235:5926:5975 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5926:5975 [7] NCCL INFO + PCI[24.0] - GPU/0-8c000 (63) +n116-042-235:5926:5975 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5919:5979 [0] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-042-235:5926:5975 [7] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-042-235:5926:5975 [7] NCCL INFO ========================================== +n116-042-235:5919:5979 [0] NCCL INFO Setting affinity for GPU 0 to 0fffffff,fffffe00,00000000,000fffff,fffffffe +n116-042-235:5926:5975 [7] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-235:5919:5979 [0] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-042-235:5926:5975 [7] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-235:5926:5975 [7] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-235:5926:5975 [7] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-235:5919:5979 [0] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-042-235:5925:5978 [6] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-042-235:5926:5975 [7] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-235:5925:5978 [6] NCCL INFO CPU/0-0 (1/1/3) +n116-042-235:5925:5978 [6] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-042-235:5926:5975 [7] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-235:5925:5978 [6] NCCL INFO + PCI[24.0] - GPU/0-4000 (56) +n116-042-235:5925:5978 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5925:5978 [6] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-042-235:5925:5978 [6] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-042-235:5926:5975 [7] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-042-235:5925:5978 [6] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-042-235:5925:5978 [6] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-042-235:5925:5978 [6] NCCL INFO + PCI[24.0] - GPU/0-5000 (57) +n116-042-235:5925:5978 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5925:5978 [6] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-042-235:5925:5978 [6] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-042-235:5926:5975 [7] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-042-235:5925:5978 [6] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-042-235:5925:5978 [6] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-042-235:5925:5978 [6] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-042-235:5925:5978 [6] NCCL INFO + PCI[24.0] - GPU/0-b000 (58) +n116-042-235:5925:5978 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5925:5978 [6] NCCL INFO + PCI[24.0] - GPU/0-c000 (59) +n116-042-235:5925:5978 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5926:5975 [7] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-235:5925:5978 [6] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-042-235:5925:5978 [6] NCCL INFO CPU/0-1 (1/1/3) +n116-042-235:5925:5978 [6] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-042-235:5925:5978 [6] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-042-235:5925:5978 [6] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-042-235:5925:5978 [6] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-042-235:5926:5975 [7] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-235:5925:5978 [6] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-042-235:5925:5978 [6] NCCL INFO + PCI[24.0] - GPU/0-84000 (60) +n116-042-235:5925:5978 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5925:5978 [6] NCCL INFO + PCI[24.0] - GPU/0-85000 (61) +n116-042-235:5925:5978 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5925:5978 [6] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-042-235:5926:5975 [7] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-235:5925:5978 [6] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-042-235:5925:5978 [6] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-042-235:5925:5978 [6] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-042-235:5925:5978 [6] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-042-235:5925:5978 [6] NCCL INFO + PCI[24.0] - GPU/0-8b000 (62) +n116-042-235:5926:5975 [7] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-235:5925:5978 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5925:5978 [6] NCCL INFO + PCI[24.0] - GPU/0-8c000 (63) +n116-042-235:5925:5978 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5925:5978 [6] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-042-235:5925:5978 [6] NCCL INFO ========================================== +n116-042-235:5926:5975 [7] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-042-235:5926:5975 [7] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-042-235:5925:5978 [6] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-235:5925:5978 [6] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-235:5926:5975 [7] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-042-235:5925:5978 [6] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-235:5926:5975 [7] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-042-235:5926:5975 [7] NCCL INFO Setting affinity for GPU 7 to 7fff,ffffffff,e0000000,0000007f,ffffffff,ffe00000,00000000 +n116-042-235:5925:5978 [6] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-235:5925:5978 [6] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-235:5925:5978 [6] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-235:5925:5978 [6] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-042-235:5925:5978 [6] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-042-235:5925:5978 [6] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-235:5925:5978 [6] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-235:5926:5975 [7] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-042-235:5925:5978 [6] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-235:5925:5978 [6] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-235:5925:5978 [6] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-042-235:5922:5981 [3] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-042-235:5925:5978 [6] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-042-235:5922:5981 [3] NCCL INFO CPU/0-0 (1/1/3) +n116-042-235:5922:5981 [3] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-042-235:5922:5981 [3] NCCL INFO + PCI[24.0] - GPU/0-4000 (56) +n116-042-235:5925:5978 [6] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-042-235:5922:5981 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5922:5981 [3] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-042-235:5922:5981 [3] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-042-235:5922:5981 [3] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-042-235:5922:5981 [3] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-042-235:5925:5978 [6] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-042-235:5922:5981 [3] NCCL INFO + PCI[24.0] - GPU/0-5000 (57) +n116-042-235:5922:5981 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5922:5981 [3] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-042-235:5922:5981 [3] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-042-235:5922:5981 [3] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-042-235:5925:5978 [6] NCCL INFO Setting affinity for GPU 6 to 7fff,ffffffff,e0000000,0000007f,ffffffff,ffe00000,00000000 +n116-042-235:5926:5975 [7] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-042-235:5922:5981 [3] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-042-235:5922:5981 [3] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-042-235:5922:5981 [3] NCCL INFO + PCI[24.0] - GPU/0-b000 (58) +n116-042-235:5925:5978 [6] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-042-235:5922:5981 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5922:5981 [3] NCCL INFO + PCI[24.0] - GPU/0-c000 (59) +n116-042-235:5922:5981 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5922:5981 [3] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-042-235:5922:5981 [3] NCCL INFO CPU/0-1 (1/1/3) +n116-042-235:5922:5981 [3] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-042-235:5922:5981 [3] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-042-235:5922:5981 [3] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-042-235:5922:5981 [3] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-042-235:5922:5981 [3] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-042-235:5922:5981 [3] NCCL INFO + PCI[24.0] - GPU/0-84000 (60) +n116-042-235:5922:5981 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5922:5981 [3] NCCL INFO + PCI[24.0] - GPU/0-85000 (61) +n116-042-235:5922:5981 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5922:5981 [3] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-042-235:5922:5981 [3] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-042-235:5922:5981 [3] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-042-235:5922:5981 [3] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-042-235:5922:5981 [3] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-042-235:5922:5981 [3] NCCL INFO + PCI[24.0] - GPU/0-8b000 (62) +n116-042-235:5922:5981 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5922:5981 [3] NCCL INFO + PCI[24.0] - GPU/0-8c000 (63) +n116-042-235:5922:5981 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5922:5981 [3] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-042-235:5922:5981 [3] NCCL INFO ========================================== +n116-042-235:5925:5978 [6] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-042-235:5921:5976 [2] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-042-235:5923:5980 [4] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-042-235:5922:5981 [3] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-235:5921:5976 [2] NCCL INFO CPU/0-0 (1/1/3) +n116-042-235:5923:5980 [4] NCCL INFO CPU/0-0 (1/1/3) +n116-042-235:5921:5976 [2] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-042-235:5923:5980 [4] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-042-235:5921:5976 [2] NCCL INFO + PCI[24.0] - GPU/0-4000 (56) +n116-042-235:5923:5980 [4] NCCL INFO + PCI[24.0] - GPU/0-4000 (56) +n116-042-235:5921:5976 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5922:5981 [3] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-235:5921:5976 [2] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-042-235:5923:5980 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5921:5976 [2] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-042-235:5923:5980 [4] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-042-235:5921:5976 [2] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-042-235:5923:5980 [4] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-042-235:5921:5976 [2] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-042-235:5923:5980 [4] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-042-235:5921:5976 [2] NCCL INFO + PCI[24.0] - GPU/0-5000 (57) +n116-042-235:5922:5981 [3] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-235:5921:5976 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5923:5980 [4] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-042-235:5921:5976 [2] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-042-235:5923:5980 [4] NCCL INFO + PCI[24.0] - GPU/0-5000 (57) +n116-042-235:5921:5976 [2] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-042-235:5923:5980 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5923:5980 [4] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-042-235:5921:5976 [2] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-042-235:5922:5981 [3] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-235:5923:5980 [4] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-042-235:5921:5976 [2] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-042-235:5923:5980 [4] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-042-235:5921:5976 [2] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-042-235:5923:5980 [4] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-042-235:5921:5976 [2] NCCL INFO + PCI[24.0] - GPU/0-b000 (58) +n116-042-235:5923:5980 [4] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-042-235:5923:5980 [4] NCCL INFO + PCI[24.0] - GPU/0-b000 (58) +n116-042-235:5921:5976 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5922:5981 [3] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-235:5923:5980 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5921:5976 [2] NCCL INFO + PCI[24.0] - GPU/0-c000 (59) +n116-042-235:5921:5976 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5923:5980 [4] NCCL INFO + PCI[24.0] - GPU/0-c000 (59) +n116-042-235:5921:5976 [2] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-042-235:5923:5980 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5923:5980 [4] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-042-235:5921:5976 [2] NCCL INFO CPU/0-1 (1/1/3) +n116-042-235:5922:5981 [3] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-235:5923:5980 [4] NCCL INFO CPU/0-1 (1/1/3) +n116-042-235:5921:5976 [2] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-042-235:5923:5980 [4] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-042-235:5921:5976 [2] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-042-235:5923:5980 [4] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-042-235:5921:5976 [2] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-042-235:5923:5980 [4] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-042-235:5922:5981 [3] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-042-235:5921:5976 [2] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-042-235:5923:5980 [4] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-042-235:5921:5976 [2] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-042-235:5923:5980 [4] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-042-235:5921:5976 [2] NCCL INFO + PCI[24.0] - GPU/0-84000 (60) +n116-042-235:5923:5980 [4] NCCL INFO + PCI[24.0] - GPU/0-84000 (60) +n116-042-235:5921:5976 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5923:5980 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5921:5976 [2] NCCL INFO + PCI[24.0] - GPU/0-85000 (61) +n116-042-235:5922:5981 [3] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-042-235:5923:5980 [4] NCCL INFO + PCI[24.0] - GPU/0-85000 (61) +n116-042-235:5921:5976 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5923:5980 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5921:5976 [2] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-042-235:5923:5980 [4] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-042-235:5921:5976 [2] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-042-235:5923:5980 [4] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-042-235:5922:5981 [3] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-235:5923:5980 [4] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-042-235:5921:5976 [2] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-042-235:5921:5976 [2] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-042-235:5921:5976 [2] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-042-235:5923:5980 [4] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-042-235:5922:5981 [3] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-235:5923:5980 [4] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-042-235:5923:5980 [4] NCCL INFO + PCI[24.0] - GPU/0-8b000 (62) +n116-042-235:5921:5976 [2] NCCL INFO + PCI[24.0] - GPU/0-8b000 (62) +n116-042-235:5923:5980 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5921:5976 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5922:5981 [3] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-235:5921:5976 [2] NCCL INFO + PCI[24.0] - GPU/0-8c000 (63) +n116-042-235:5923:5980 [4] NCCL INFO + PCI[24.0] - GPU/0-8c000 (63) +n116-042-235:5921:5976 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5921:5976 [2] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-042-235:5923:5980 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5921:5976 [2] NCCL INFO ========================================== +n116-042-235:5923:5980 [4] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-042-235:5922:5981 [3] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-235:5923:5980 [4] NCCL INFO ========================================== +n116-042-235:5921:5976 [2] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-235:5922:5981 [3] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-042-235:5923:5980 [4] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-235:5922:5981 [3] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-042-235:5921:5976 [2] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-235:5923:5980 [4] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-235:5922:5981 [3] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-042-235:5921:5976 [2] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-235:5923:5980 [4] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-235:5921:5976 [2] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-235:5923:5980 [4] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-235:5922:5981 [3] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-042-235:5923:5980 [4] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-235:5922:5981 [3] NCCL INFO Setting affinity for GPU 3 to 0fffffff,fffffe00,00000000,000fffff,fffffffe +n116-042-235:5921:5976 [2] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-235:5921:5976 [2] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-235:5923:5980 [4] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-235:5924:5977 [5] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-042-235:5924:5977 [5] NCCL INFO CPU/0-0 (1/1/3) +n116-042-235:5921:5976 [2] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-042-235:5924:5977 [5] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-042-235:5923:5980 [4] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-042-235:5924:5977 [5] NCCL INFO + PCI[24.0] - GPU/0-4000 (56) +n116-042-235:5924:5977 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5924:5977 [5] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-042-235:5921:5976 [2] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-042-235:5923:5980 [4] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-042-235:5924:5977 [5] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-042-235:5924:5977 [5] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-042-235:5924:5977 [5] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-042-235:5923:5980 [4] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-235:5921:5976 [2] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-235:5924:5977 [5] NCCL INFO + PCI[24.0] - GPU/0-5000 (57) +n116-042-235:5924:5977 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5924:5977 [5] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-042-235:5923:5980 [4] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-235:5924:5977 [5] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-042-235:5924:5977 [5] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-042-235:5924:5977 [5] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-042-235:5924:5977 [5] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-042-235:5924:5977 [5] NCCL INFO + PCI[24.0] - GPU/0-b000 (58) +n116-042-235:5923:5980 [4] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-235:5924:5977 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5924:5977 [5] NCCL INFO + PCI[24.0] - GPU/0-c000 (59) +n116-042-235:5921:5976 [2] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-235:5924:5977 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5924:5977 [5] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-042-235:5923:5980 [4] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-235:5924:5977 [5] NCCL INFO CPU/0-1 (1/1/3) +n116-042-235:5922:5981 [3] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-042-235:5924:5977 [5] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-042-235:5924:5977 [5] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-042-235:5921:5976 [2] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-235:5924:5977 [5] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-042-235:5924:5977 [5] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-042-235:5923:5980 [4] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-042-235:5924:5977 [5] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-042-235:5924:5977 [5] NCCL INFO + PCI[24.0] - GPU/0-84000 (60) +n116-042-235:5924:5977 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5921:5976 [2] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-235:5923:5980 [4] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-042-235:5924:5977 [5] NCCL INFO + PCI[24.0] - GPU/0-85000 (61) +n116-042-235:5924:5977 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5924:5977 [5] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-042-235:5924:5977 [5] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-042-235:5924:5977 [5] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-042-235:5923:5980 [4] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-042-235:5921:5976 [2] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-042-235:5924:5977 [5] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-042-235:5924:5977 [5] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-042-235:5924:5977 [5] NCCL INFO + PCI[24.0] - GPU/0-8b000 (62) +n116-042-235:5924:5977 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5924:5977 [5] NCCL INFO + PCI[24.0] - GPU/0-8c000 (63) +n116-042-235:5921:5976 [2] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-042-235:5923:5980 [4] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-042-235:5924:5977 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5924:5977 [5] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-042-235:5924:5977 [5] NCCL INFO ========================================== +n116-042-235:5921:5976 [2] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-042-235:5923:5980 [4] NCCL INFO Setting affinity for GPU 4 to 7fff,ffffffff,e0000000,0000007f,ffffffff,ffe00000,00000000 +n116-042-235:5922:5981 [3] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-042-235:5924:5977 [5] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-235:5924:5977 [5] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-235:5921:5976 [2] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-042-235:5921:5976 [2] NCCL INFO Setting affinity for GPU 2 to 0fffffff,fffffe00,00000000,000fffff,fffffffe +n116-042-235:5924:5977 [5] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-235:5924:5977 [5] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-235:5924:5977 [5] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-235:5924:5977 [5] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-235:5924:5977 [5] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-042-235:5923:5980 [4] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-042-235:5924:5977 [5] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-042-235:5924:5977 [5] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-235:5924:5977 [5] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-235:5924:5977 [5] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-235:5923:5980 [4] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-042-235:5924:5977 [5] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-235:5924:5977 [5] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-042-235:5924:5977 [5] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-042-235:5920:5982 [1] NCCL INFO === System : maxBw 24.0 totalBw 370.8 === +n116-042-235:5924:5977 [5] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-042-235:5920:5982 [1] NCCL INFO CPU/0-0 (1/1/3) +n116-042-235:5920:5982 [1] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-042-235:5920:5982 [1] NCCL INFO + PCI[24.0] - GPU/0-4000 (56) +n116-042-235:5924:5977 [5] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-042-235:5920:5982 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5920:5982 [1] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-042-235:5920:5982 [1] NCCL INFO + NET[25.0] - NET/0-0 (0/0/0/25.000000) +n116-042-235:5924:5977 [5] NCCL INFO Setting affinity for GPU 5 to 7fff,ffffffff,e0000000,0000007f,ffffffff,ffe00000,00000000 +n116-042-235:5920:5982 [1] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-042-235:5920:5982 [1] NCCL INFO + NET[25.0] - NET/0-1 (0/1/0/25.000000) +n116-042-235:5920:5982 [1] NCCL INFO + PCI[24.0] - GPU/0-5000 (57) +n116-042-235:5920:5982 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5924:5977 [5] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-042-235:5920:5982 [1] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-042-235:5920:5982 [1] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-042-235:5920:5982 [1] NCCL INFO + NET[25.0] - NET/0-2 (0/2/0/25.000000) +n116-042-235:5920:5982 [1] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-042-235:5920:5982 [1] NCCL INFO + NET[25.0] - NET/0-3 (0/3/0/25.000000) +n116-042-235:5920:5982 [1] NCCL INFO + PCI[24.0] - GPU/0-b000 (58) +n116-042-235:5920:5982 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5920:5982 [1] NCCL INFO + PCI[24.0] - GPU/0-c000 (59) +n116-042-235:5920:5982 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5920:5982 [1] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-042-235:5920:5982 [1] NCCL INFO CPU/0-1 (1/1/3) +n116-042-235:5920:5982 [1] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-042-235:5920:5982 [1] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-042-235:5920:5982 [1] NCCL INFO + NET[25.0] - NET/0-4 (0/4/0/25.000000) +n116-042-235:5920:5982 [1] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-042-235:5920:5982 [1] NCCL INFO + NET[25.0] - NET/0-5 (0/5/0/25.000000) +n116-042-235:5920:5982 [1] NCCL INFO + PCI[24.0] - GPU/0-84000 (60) +n116-042-235:5920:5982 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5920:5982 [1] NCCL INFO + PCI[24.0] - GPU/0-85000 (61) +n116-042-235:5920:5982 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5920:5982 [1] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-042-235:5920:5982 [1] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-042-235:5924:5977 [5] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-042-235:5920:5982 [1] NCCL INFO + NET[25.0] - NET/0-6 (0/6/0/25.000000) +n116-042-235:5920:5982 [1] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-042-235:5920:5982 [1] NCCL INFO + NET[25.0] - NET/0-7 (0/7/0/25.000000) +n116-042-235:5920:5982 [1] NCCL INFO + PCI[24.0] - GPU/0-8b000 (62) +n116-042-235:5920:5982 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5920:5982 [1] NCCL INFO + PCI[24.0] - GPU/0-8c000 (63) +n116-042-235:5920:5982 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-235:5920:5982 [1] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-042-235:5920:5982 [1] NCCL INFO ========================================== +n116-042-235:5920:5982 [1] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-235:5920:5982 [1] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (3/24.0/PIX) NET/0-1 (3/24.0/PIX) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-235:5920:5982 [1] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-235:5921:5976 [2] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-042-235:5920:5982 [1] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (3/24.0/PIX) NET/0-3 (3/24.0/PIX) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-235:5920:5982 [1] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-235:5920:5982 [1] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (3/24.0/PIX) NET/0-5 (3/24.0/PIX) NET/0-6 (5/24.0/PXN) NET/0-7 (5/24.0/PXN) +n116-042-235:5920:5982 [1] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-042-235:5920:5982 [1] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) NET/0-0 (5/24.0/PXN) NET/0-1 (5/24.0/PXN) NET/0-2 (5/24.0/PXN) NET/0-3 (5/24.0/PXN) NET/0-4 (5/24.0/PXN) NET/0-5 (5/24.0/PXN) NET/0-6 (3/24.0/PIX) NET/0-7 (3/24.0/PIX) +n116-042-235:5921:5976 [2] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-042-235:5920:5982 [1] NCCL INFO NET/0-0 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (0/5000.0/LOC) NET/0-1 (4/24.0/PIX) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-235:5920:5982 [1] NCCL INFO NET/0-1 :GPU/0-4000 (3/24.0/PIX) GPU/0-5000 (3/24.0/PIX) GPU/0-b000 (5/24.0/PHB) GPU/0-c000 (5/24.0/PHB) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (4/24.0/PIX) NET/0-1 (0/5000.0/LOC) NET/0-2 (6/24.0/PHB) NET/0-3 (6/24.0/PHB) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-235:5920:5982 [1] NCCL INFO NET/0-2 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (0/5000.0/LOC) NET/0-3 (4/24.0/PIX) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-235:5920:5982 [1] NCCL INFO NET/0-3 :GPU/0-4000 (5/24.0/PHB) GPU/0-5000 (5/24.0/PHB) GPU/0-b000 (3/24.0/PIX) GPU/0-c000 (3/24.0/PIX) GPU/0-84000 (6/22.0/SYS) GPU/0-85000 (6/22.0/SYS) GPU/0-8b000 (6/22.0/SYS) GPU/0-8c000 (6/22.0/SYS) CPU/0-0 (3/24.0/PHB) CPU/0-1 (4/22.0/SYS) NET/0-0 (6/24.0/PHB) NET/0-1 (6/24.0/PHB) NET/0-2 (4/24.0/PIX) NET/0-3 (0/5000.0/LOC) NET/0-4 (7/22.0/SYS) NET/0-5 (7/22.0/SYS) NET/0-6 (7/22.0/SYS) NET/0-7 (7/22.0/SYS) +n116-042-235:5920:5982 [1] NCCL INFO NET/0-4 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (0/5000.0/LOC) NET/0-5 (4/24.0/PIX) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-042-235:5920:5982 [1] NCCL INFO NET/0-5 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (3/24.0/PIX) GPU/0-85000 (3/24.0/PIX) GPU/0-8b000 (5/24.0/PHB) GPU/0-8c000 (5/24.0/PHB) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (4/24.0/PIX) NET/0-5 (0/5000.0/LOC) NET/0-6 (6/24.0/PHB) NET/0-7 (6/24.0/PHB) +n116-042-235:5920:5982 [1] NCCL INFO NET/0-6 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (0/5000.0/LOC) NET/0-7 (4/24.0/PIX) +n116-042-235:5920:5982 [1] NCCL INFO NET/0-7 :GPU/0-4000 (6/22.0/SYS) GPU/0-5000 (6/22.0/SYS) GPU/0-b000 (6/22.0/SYS) GPU/0-c000 (6/22.0/SYS) GPU/0-84000 (5/24.0/PHB) GPU/0-85000 (5/24.0/PHB) GPU/0-8b000 (3/24.0/PIX) GPU/0-8c000 (3/24.0/PIX) CPU/0-0 (4/22.0/SYS) CPU/0-1 (3/24.0/PHB) NET/0-0 (7/22.0/SYS) NET/0-1 (7/22.0/SYS) NET/0-2 (7/22.0/SYS) NET/0-3 (7/22.0/SYS) NET/0-4 (6/24.0/PHB) NET/0-5 (6/24.0/PHB) NET/0-6 (4/24.0/PIX) NET/0-7 (0/5000.0/LOC) +n116-042-235:5920:5982 [1] NCCL INFO Setting affinity for GPU 1 to 0fffffff,fffffe00,00000000,000fffff,fffffffe +n116-042-235:5920:5982 [1] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-042-235:5920:5982 [1] NCCL INFO NCCL_CROSS_NIC set by environment to 0. +n116-042-235:5925:5978 [6] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-042-235:5925:5978 [6] NCCL INFO 0 : NET/0-0 GPU/56 GPU/63 GPU/62 GPU/61 GPU/60 GPU/59 GPU/58 GPU/57 NET/0-0 +n116-042-235:5925:5978 [6] NCCL INFO 1 : NET/0-1 GPU/57 GPU/58 GPU/59 GPU/60 GPU/61 GPU/62 GPU/63 GPU/56 NET/0-1 +n116-042-235:5925:5978 [6] NCCL INFO 2 : NET/0-2 GPU/58 GPU/57 GPU/56 GPU/63 GPU/62 GPU/61 GPU/60 GPU/59 NET/0-2 +n116-042-235:5925:5978 [6] NCCL INFO 3 : NET/0-3 GPU/59 GPU/60 GPU/61 GPU/62 GPU/63 GPU/56 GPU/57 GPU/58 NET/0-3 +n116-042-235:5925:5978 [6] NCCL INFO 4 : NET/0-4 GPU/60 GPU/59 GPU/58 GPU/57 GPU/56 GPU/63 GPU/62 GPU/61 NET/0-4 +n116-042-235:5925:5978 [6] NCCL INFO 5 : NET/0-5 GPU/61 GPU/62 GPU/63 GPU/56 GPU/57 GPU/58 GPU/59 GPU/60 NET/0-5 +n116-042-235:5925:5978 [6] NCCL INFO 6 : NET/0-6 GPU/62 GPU/61 GPU/60 GPU/59 GPU/58 GPU/57 GPU/56 GPU/63 NET/0-6 +n116-042-235:5925:5978 [6] NCCL INFO 7 : NET/0-7 GPU/63 GPU/56 GPU/57 GPU/58 GPU/59 GPU/60 GPU/61 GPU/62 NET/0-7 +n116-042-235:5919:5979 [0] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-042-235:5919:5979 [0] NCCL INFO 0 : NET/0-0 GPU/56 GPU/63 GPU/62 GPU/61 GPU/60 GPU/59 GPU/58 GPU/57 NET/0-0 +n116-042-235:5919:5979 [0] NCCL INFO 1 : NET/0-1 GPU/57 GPU/58 GPU/59 GPU/60 GPU/61 GPU/62 GPU/63 GPU/56 NET/0-1 +n116-042-235:5919:5979 [0] NCCL INFO 2 : NET/0-2 GPU/58 GPU/57 GPU/56 GPU/63 GPU/62 GPU/61 GPU/60 GPU/59 NET/0-2 +n116-042-235:5919:5979 [0] NCCL INFO 3 : NET/0-3 GPU/59 GPU/60 GPU/61 GPU/62 GPU/63 GPU/56 GPU/57 GPU/58 NET/0-3 +n116-042-235:5919:5979 [0] NCCL INFO 4 : NET/0-4 GPU/60 GPU/59 GPU/58 GPU/57 GPU/56 GPU/63 GPU/62 GPU/61 NET/0-4 +n116-042-235:5919:5979 [0] NCCL INFO 5 : NET/0-5 GPU/61 GPU/62 GPU/63 GPU/56 GPU/57 GPU/58 GPU/59 GPU/60 NET/0-5 +n116-042-235:5919:5979 [0] NCCL INFO 6 : NET/0-6 GPU/62 GPU/61 GPU/60 GPU/59 GPU/58 GPU/57 GPU/56 GPU/63 NET/0-6 +n116-042-235:5919:5979 [0] NCCL INFO 7 : NET/0-7 GPU/63 GPU/56 GPU/57 GPU/58 GPU/59 GPU/60 GPU/61 GPU/62 NET/0-7 +n116-042-235:5925:5978 [6] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-042-235:5925:5978 [6] NCCL INFO 0 : NET/0-0 GPU/56 GPU/57 GPU/58 GPU/59 GPU/60 GPU/61 GPU/62 GPU/63 NET/0-0 +n116-042-235:5925:5978 [6] NCCL INFO 1 : NET/0-1 GPU/57 GPU/56 GPU/63 GPU/62 GPU/61 GPU/60 GPU/59 GPU/58 NET/0-1 +n116-042-235:5925:5978 [6] NCCL INFO 2 : NET/0-2 GPU/58 GPU/59 GPU/60 GPU/61 GPU/62 GPU/63 GPU/56 GPU/57 NET/0-2 +n116-042-235:5925:5978 [6] NCCL INFO 3 : NET/0-3 GPU/59 GPU/58 GPU/57 GPU/56 GPU/63 GPU/62 GPU/61 GPU/60 NET/0-3 +n116-042-235:5925:5978 [6] NCCL INFO 4 : NET/0-4 GPU/60 GPU/61 GPU/62 GPU/63 GPU/56 GPU/57 GPU/58 GPU/59 NET/0-4 +n116-042-235:5925:5978 [6] NCCL INFO 5 : NET/0-5 GPU/61 GPU/60 GPU/59 GPU/58 GPU/57 GPU/56 GPU/63 GPU/62 NET/0-5 +n116-042-235:5925:5978 [6] NCCL INFO 6 : NET/0-6 GPU/62 GPU/63 GPU/56 GPU/57 GPU/58 GPU/59 GPU/60 GPU/61 NET/0-6 +n116-042-235:5925:5978 [6] NCCL INFO 7 : NET/0-7 GPU/63 GPU/62 GPU/61 GPU/60 GPU/59 GPU/58 GPU/57 GPU/56 NET/0-7 +n116-042-235:5919:5979 [0] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-042-235:5919:5979 [0] NCCL INFO 0 : NET/0-0 GPU/56 GPU/57 GPU/58 GPU/59 GPU/60 GPU/61 GPU/62 GPU/63 NET/0-0 +n116-042-235:5919:5979 [0] NCCL INFO 1 : NET/0-1 GPU/57 GPU/56 GPU/63 GPU/62 GPU/61 GPU/60 GPU/59 GPU/58 NET/0-1 +n116-042-235:5919:5979 [0] NCCL INFO 2 : NET/0-2 GPU/58 GPU/59 GPU/60 GPU/61 GPU/62 GPU/63 GPU/56 GPU/57 NET/0-2 +n116-042-235:5919:5979 [0] NCCL INFO 3 : NET/0-3 GPU/59 GPU/58 GPU/57 GPU/56 GPU/63 GPU/62 GPU/61 GPU/60 NET/0-3 +n116-042-235:5919:5979 [0] NCCL INFO 4 : NET/0-4 GPU/60 GPU/61 GPU/62 GPU/63 GPU/56 GPU/57 GPU/58 GPU/59 NET/0-4 +n116-042-235:5919:5979 [0] NCCL INFO 5 : NET/0-5 GPU/61 GPU/60 GPU/59 GPU/58 GPU/57 GPU/56 GPU/63 GPU/62 NET/0-5 +n116-042-235:5919:5979 [0] NCCL INFO 6 : NET/0-6 GPU/62 GPU/63 GPU/56 GPU/57 GPU/58 GPU/59 GPU/60 GPU/61 NET/0-6 +n116-042-235:5919:5979 [0] NCCL INFO 7 : NET/0-7 GPU/63 GPU/62 GPU/61 GPU/60 GPU/59 GPU/58 GPU/57 GPU/56 NET/0-7 +n116-042-235:5921:5976 [2] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-042-235:5921:5976 [2] NCCL INFO 0 : NET/0-0 GPU/56 GPU/63 GPU/62 GPU/61 GPU/60 GPU/59 GPU/58 GPU/57 NET/0-0 +n116-042-235:5921:5976 [2] NCCL INFO 1 : NET/0-1 GPU/57 GPU/58 GPU/59 GPU/60 GPU/61 GPU/62 GPU/63 GPU/56 NET/0-1 +n116-042-235:5921:5976 [2] NCCL INFO 2 : NET/0-2 GPU/58 GPU/57 GPU/56 GPU/63 GPU/62 GPU/61 GPU/60 GPU/59 NET/0-2 +n116-042-235:5921:5976 [2] NCCL INFO 3 : NET/0-3 GPU/59 GPU/60 GPU/61 GPU/62 GPU/63 GPU/56 GPU/57 GPU/58 NET/0-3 +n116-042-235:5921:5976 [2] NCCL INFO 4 : NET/0-4 GPU/60 GPU/59 GPU/58 GPU/57 GPU/56 GPU/63 GPU/62 GPU/61 NET/0-4 +n116-042-235:5921:5976 [2] NCCL INFO 5 : NET/0-5 GPU/61 GPU/62 GPU/63 GPU/56 GPU/57 GPU/58 GPU/59 GPU/60 NET/0-5 +n116-042-235:5921:5976 [2] NCCL INFO 6 : NET/0-6 GPU/62 GPU/61 GPU/60 GPU/59 GPU/58 GPU/57 GPU/56 GPU/63 NET/0-6 +n116-042-235:5921:5976 [2] NCCL INFO 7 : NET/0-7 GPU/63 GPU/56 GPU/57 GPU/58 GPU/59 GPU/60 GPU/61 GPU/62 NET/0-7 +n116-042-235:5921:5976 [2] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-042-235:5921:5976 [2] NCCL INFO 0 : NET/0-0 GPU/56 GPU/57 GPU/58 GPU/59 GPU/60 GPU/61 GPU/62 GPU/63 NET/0-0 +n116-042-235:5921:5976 [2] NCCL INFO 1 : NET/0-1 GPU/57 GPU/56 GPU/63 GPU/62 GPU/61 GPU/60 GPU/59 GPU/58 NET/0-1 +n116-042-235:5921:5976 [2] NCCL INFO 2 : NET/0-2 GPU/58 GPU/59 GPU/60 GPU/61 GPU/62 GPU/63 GPU/56 GPU/57 NET/0-2 +n116-042-235:5921:5976 [2] NCCL INFO 3 : NET/0-3 GPU/59 GPU/58 GPU/57 GPU/56 GPU/63 GPU/62 GPU/61 GPU/60 NET/0-3 +n116-042-235:5921:5976 [2] NCCL INFO 4 : NET/0-4 GPU/60 GPU/61 GPU/62 GPU/63 GPU/56 GPU/57 GPU/58 GPU/59 NET/0-4 +n116-042-235:5921:5976 [2] NCCL INFO 5 : NET/0-5 GPU/61 GPU/60 GPU/59 GPU/58 GPU/57 GPU/56 GPU/63 GPU/62 NET/0-5 +n116-042-235:5921:5976 [2] NCCL INFO 6 : NET/0-6 GPU/62 GPU/63 GPU/56 GPU/57 GPU/58 GPU/59 GPU/60 GPU/61 NET/0-6 +n116-042-235:5921:5976 [2] NCCL INFO 7 : NET/0-7 GPU/63 GPU/62 GPU/61 GPU/60 GPU/59 GPU/58 GPU/57 GPU/56 NET/0-7 +n116-042-235:5924:5977 [5] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-042-235:5924:5977 [5] NCCL INFO 0 : NET/0-0 GPU/56 GPU/63 GPU/62 GPU/61 GPU/60 GPU/59 GPU/58 GPU/57 NET/0-0 +n116-042-235:5924:5977 [5] NCCL INFO 1 : NET/0-1 GPU/57 GPU/58 GPU/59 GPU/60 GPU/61 GPU/62 GPU/63 GPU/56 NET/0-1 +n116-042-235:5924:5977 [5] NCCL INFO 2 : NET/0-2 GPU/58 GPU/57 GPU/56 GPU/63 GPU/62 GPU/61 GPU/60 GPU/59 NET/0-2 +n116-042-235:5924:5977 [5] NCCL INFO 3 : NET/0-3 GPU/59 GPU/60 GPU/61 GPU/62 GPU/63 GPU/56 GPU/57 GPU/58 NET/0-3 +n116-042-235:5924:5977 [5] NCCL INFO 4 : NET/0-4 GPU/60 GPU/59 GPU/58 GPU/57 GPU/56 GPU/63 GPU/62 GPU/61 NET/0-4 +n116-042-235:5924:5977 [5] NCCL INFO 5 : NET/0-5 GPU/61 GPU/62 GPU/63 GPU/56 GPU/57 GPU/58 GPU/59 GPU/60 NET/0-5 +n116-042-235:5924:5977 [5] NCCL INFO 6 : NET/0-6 GPU/62 GPU/61 GPU/60 GPU/59 GPU/58 GPU/57 GPU/56 GPU/63 NET/0-6 +n116-042-235:5924:5977 [5] NCCL INFO 7 : NET/0-7 GPU/63 GPU/56 GPU/57 GPU/58 GPU/59 GPU/60 GPU/61 GPU/62 NET/0-7 +n116-042-235:5924:5977 [5] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-042-235:5924:5977 [5] NCCL INFO 0 : NET/0-0 GPU/56 GPU/57 GPU/58 GPU/59 GPU/60 GPU/61 GPU/62 GPU/63 NET/0-0 +n116-042-235:5924:5977 [5] NCCL INFO 1 : NET/0-1 GPU/57 GPU/56 GPU/63 GPU/62 GPU/61 GPU/60 GPU/59 GPU/58 NET/0-1 +n116-042-235:5924:5977 [5] NCCL INFO 2 : NET/0-2 GPU/58 GPU/59 GPU/60 GPU/61 GPU/62 GPU/63 GPU/56 GPU/57 NET/0-2 +n116-042-235:5924:5977 [5] NCCL INFO 3 : NET/0-3 GPU/59 GPU/58 GPU/57 GPU/56 GPU/63 GPU/62 GPU/61 GPU/60 NET/0-3 +n116-042-235:5924:5977 [5] NCCL INFO 4 : NET/0-4 GPU/60 GPU/61 GPU/62 GPU/63 GPU/56 GPU/57 GPU/58 GPU/59 NET/0-4 +n116-042-235:5924:5977 [5] NCCL INFO 5 : NET/0-5 GPU/61 GPU/60 GPU/59 GPU/58 GPU/57 GPU/56 GPU/63 GPU/62 NET/0-5 +n116-042-235:5924:5977 [5] NCCL INFO 6 : NET/0-6 GPU/62 GPU/63 GPU/56 GPU/57 GPU/58 GPU/59 GPU/60 GPU/61 NET/0-6 +n116-042-235:5924:5977 [5] NCCL INFO 7 : NET/0-7 GPU/63 GPU/62 GPU/61 GPU/60 GPU/59 GPU/58 GPU/57 GPU/56 NET/0-7 +n116-042-235:5923:5980 [4] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-042-235:5923:5980 [4] NCCL INFO 0 : NET/0-0 GPU/56 GPU/63 GPU/62 GPU/61 GPU/60 GPU/59 GPU/58 GPU/57 NET/0-0 +n116-042-235:5923:5980 [4] NCCL INFO 1 : NET/0-1 GPU/57 GPU/58 GPU/59 GPU/60 GPU/61 GPU/62 GPU/63 GPU/56 NET/0-1 +n116-042-235:5923:5980 [4] NCCL INFO 2 : NET/0-2 GPU/58 GPU/57 GPU/56 GPU/63 GPU/62 GPU/61 GPU/60 GPU/59 NET/0-2 +n116-042-235:5923:5980 [4] NCCL INFO 3 : NET/0-3 GPU/59 GPU/60 GPU/61 GPU/62 GPU/63 GPU/56 GPU/57 GPU/58 NET/0-3 +n116-042-235:5923:5980 [4] NCCL INFO 4 : NET/0-4 GPU/60 GPU/59 GPU/58 GPU/57 GPU/56 GPU/63 GPU/62 GPU/61 NET/0-4 +n116-042-235:5923:5980 [4] NCCL INFO 5 : NET/0-5 GPU/61 GPU/62 GPU/63 GPU/56 GPU/57 GPU/58 GPU/59 GPU/60 NET/0-5 +n116-042-235:5923:5980 [4] NCCL INFO 6 : NET/0-6 GPU/62 GPU/61 GPU/60 GPU/59 GPU/58 GPU/57 GPU/56 GPU/63 NET/0-6 +n116-042-235:5923:5980 [4] NCCL INFO 7 : NET/0-7 GPU/63 GPU/56 GPU/57 GPU/58 GPU/59 GPU/60 GPU/61 GPU/62 NET/0-7 +n116-042-235:5923:5980 [4] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-042-235:5923:5980 [4] NCCL INFO 0 : NET/0-0 GPU/56 GPU/57 GPU/58 GPU/59 GPU/60 GPU/61 GPU/62 GPU/63 NET/0-0 +n116-042-235:5923:5980 [4] NCCL INFO 1 : NET/0-1 GPU/57 GPU/56 GPU/63 GPU/62 GPU/61 GPU/60 GPU/59 GPU/58 NET/0-1 +n116-042-235:5923:5980 [4] NCCL INFO 2 : NET/0-2 GPU/58 GPU/59 GPU/60 GPU/61 GPU/62 GPU/63 GPU/56 GPU/57 NET/0-2 +n116-042-235:5923:5980 [4] NCCL INFO 3 : NET/0-3 GPU/59 GPU/58 GPU/57 GPU/56 GPU/63 GPU/62 GPU/61 GPU/60 NET/0-3 +n116-042-235:5923:5980 [4] NCCL INFO 4 : NET/0-4 GPU/60 GPU/61 GPU/62 GPU/63 GPU/56 GPU/57 GPU/58 GPU/59 NET/0-4 +n116-042-235:5923:5980 [4] NCCL INFO 5 : NET/0-5 GPU/61 GPU/60 GPU/59 GPU/58 GPU/57 GPU/56 GPU/63 GPU/62 NET/0-5 +n116-042-235:5923:5980 [4] NCCL INFO 6 : NET/0-6 GPU/62 GPU/63 GPU/56 GPU/57 GPU/58 GPU/59 GPU/60 GPU/61 NET/0-6 +n116-042-235:5923:5980 [4] NCCL INFO 7 : NET/0-7 GPU/63 GPU/62 GPU/61 GPU/60 GPU/59 GPU/58 GPU/57 GPU/56 NET/0-7 +n116-042-235:5926:5975 [7] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-042-235:5926:5975 [7] NCCL INFO 0 : NET/0-0 GPU/56 GPU/63 GPU/62 GPU/61 GPU/60 GPU/59 GPU/58 GPU/57 NET/0-0 +n116-042-235:5926:5975 [7] NCCL INFO 1 : NET/0-1 GPU/57 GPU/58 GPU/59 GPU/60 GPU/61 GPU/62 GPU/63 GPU/56 NET/0-1 +n116-042-235:5926:5975 [7] NCCL INFO 2 : NET/0-2 GPU/58 GPU/57 GPU/56 GPU/63 GPU/62 GPU/61 GPU/60 GPU/59 NET/0-2 +n116-042-235:5926:5975 [7] NCCL INFO 3 : NET/0-3 GPU/59 GPU/60 GPU/61 GPU/62 GPU/63 GPU/56 GPU/57 GPU/58 NET/0-3 +n116-042-235:5926:5975 [7] NCCL INFO 4 : NET/0-4 GPU/60 GPU/59 GPU/58 GPU/57 GPU/56 GPU/63 GPU/62 GPU/61 NET/0-4 +n116-042-235:5926:5975 [7] NCCL INFO 5 : NET/0-5 GPU/61 GPU/62 GPU/63 GPU/56 GPU/57 GPU/58 GPU/59 GPU/60 NET/0-5 +n116-042-235:5926:5975 [7] NCCL INFO 6 : NET/0-6 GPU/62 GPU/61 GPU/60 GPU/59 GPU/58 GPU/57 GPU/56 GPU/63 NET/0-6 +n116-042-235:5926:5975 [7] NCCL INFO 7 : NET/0-7 GPU/63 GPU/56 GPU/57 GPU/58 GPU/59 GPU/60 GPU/61 GPU/62 NET/0-7 +n116-042-235:5926:5975 [7] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-042-235:5926:5975 [7] NCCL INFO 0 : NET/0-0 GPU/56 GPU/57 GPU/58 GPU/59 GPU/60 GPU/61 GPU/62 GPU/63 NET/0-0 +n116-042-235:5926:5975 [7] NCCL INFO 1 : NET/0-1 GPU/57 GPU/56 GPU/63 GPU/62 GPU/61 GPU/60 GPU/59 GPU/58 NET/0-1 +n116-042-235:5926:5975 [7] NCCL INFO 2 : NET/0-2 GPU/58 GPU/59 GPU/60 GPU/61 GPU/62 GPU/63 GPU/56 GPU/57 NET/0-2 +n116-042-235:5926:5975 [7] NCCL INFO 3 : NET/0-3 GPU/59 GPU/58 GPU/57 GPU/56 GPU/63 GPU/62 GPU/61 GPU/60 NET/0-3 +n116-042-235:5926:5975 [7] NCCL INFO 4 : NET/0-4 GPU/60 GPU/61 GPU/62 GPU/63 GPU/56 GPU/57 GPU/58 GPU/59 NET/0-4 +n116-042-235:5926:5975 [7] NCCL INFO 5 : NET/0-5 GPU/61 GPU/60 GPU/59 GPU/58 GPU/57 GPU/56 GPU/63 GPU/62 NET/0-5 +n116-042-235:5926:5975 [7] NCCL INFO 6 : NET/0-6 GPU/62 GPU/63 GPU/56 GPU/57 GPU/58 GPU/59 GPU/60 GPU/61 NET/0-6 +n116-042-235:5926:5975 [7] NCCL INFO 7 : NET/0-7 GPU/63 GPU/62 GPU/61 GPU/60 GPU/59 GPU/58 GPU/57 GPU/56 NET/0-7 +n116-042-235:5920:5982 [1] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-042-235:5920:5982 [1] NCCL INFO 0 : NET/0-0 GPU/56 GPU/63 GPU/62 GPU/61 GPU/60 GPU/59 GPU/58 GPU/57 NET/0-0 +n116-042-235:5920:5982 [1] NCCL INFO 1 : NET/0-1 GPU/57 GPU/58 GPU/59 GPU/60 GPU/61 GPU/62 GPU/63 GPU/56 NET/0-1 +n116-042-235:5920:5982 [1] NCCL INFO 2 : NET/0-2 GPU/58 GPU/57 GPU/56 GPU/63 GPU/62 GPU/61 GPU/60 GPU/59 NET/0-2 +n116-042-235:5920:5982 [1] NCCL INFO 3 : NET/0-3 GPU/59 GPU/60 GPU/61 GPU/62 GPU/63 GPU/56 GPU/57 GPU/58 NET/0-3 +n116-042-235:5920:5982 [1] NCCL INFO 4 : NET/0-4 GPU/60 GPU/59 GPU/58 GPU/57 GPU/56 GPU/63 GPU/62 GPU/61 NET/0-4 +n116-042-235:5920:5982 [1] NCCL INFO 5 : NET/0-5 GPU/61 GPU/62 GPU/63 GPU/56 GPU/57 GPU/58 GPU/59 GPU/60 NET/0-5 +n116-042-235:5920:5982 [1] NCCL INFO 6 : NET/0-6 GPU/62 GPU/61 GPU/60 GPU/59 GPU/58 GPU/57 GPU/56 GPU/63 NET/0-6 +n116-042-235:5920:5982 [1] NCCL INFO 7 : NET/0-7 GPU/63 GPU/56 GPU/57 GPU/58 GPU/59 GPU/60 GPU/61 GPU/62 NET/0-7 +n116-042-235:5920:5982 [1] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-042-235:5920:5982 [1] NCCL INFO 0 : NET/0-0 GPU/56 GPU/57 GPU/58 GPU/59 GPU/60 GPU/61 GPU/62 GPU/63 NET/0-0 +n116-042-235:5920:5982 [1] NCCL INFO 1 : NET/0-1 GPU/57 GPU/56 GPU/63 GPU/62 GPU/61 GPU/60 GPU/59 GPU/58 NET/0-1 +n116-042-235:5920:5982 [1] NCCL INFO 2 : NET/0-2 GPU/58 GPU/59 GPU/60 GPU/61 GPU/62 GPU/63 GPU/56 GPU/57 NET/0-2 +n116-042-235:5920:5982 [1] NCCL INFO 3 : NET/0-3 GPU/59 GPU/58 GPU/57 GPU/56 GPU/63 GPU/62 GPU/61 GPU/60 NET/0-3 +n116-042-235:5920:5982 [1] NCCL INFO 4 : NET/0-4 GPU/60 GPU/61 GPU/62 GPU/63 GPU/56 GPU/57 GPU/58 GPU/59 NET/0-4 +n116-042-235:5920:5982 [1] NCCL INFO 5 : NET/0-5 GPU/61 GPU/60 GPU/59 GPU/58 GPU/57 GPU/56 GPU/63 GPU/62 NET/0-5 +n116-042-235:5920:5982 [1] NCCL INFO 6 : NET/0-6 GPU/62 GPU/63 GPU/56 GPU/57 GPU/58 GPU/59 GPU/60 GPU/61 NET/0-6 +n116-042-235:5920:5982 [1] NCCL INFO 7 : NET/0-7 GPU/63 GPU/62 GPU/61 GPU/60 GPU/59 GPU/58 GPU/57 GPU/56 NET/0-7 +n116-042-235:5922:5981 [3] NCCL INFO Pattern 4, crossNic 0, nChannels 8, bw 24.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-042-235:5922:5981 [3] NCCL INFO 0 : NET/0-0 GPU/56 GPU/63 GPU/62 GPU/61 GPU/60 GPU/59 GPU/58 GPU/57 NET/0-0 +n116-042-235:5922:5981 [3] NCCL INFO 1 : NET/0-1 GPU/57 GPU/58 GPU/59 GPU/60 GPU/61 GPU/62 GPU/63 GPU/56 NET/0-1 +n116-042-235:5922:5981 [3] NCCL INFO 2 : NET/0-2 GPU/58 GPU/57 GPU/56 GPU/63 GPU/62 GPU/61 GPU/60 GPU/59 NET/0-2 +n116-042-235:5922:5981 [3] NCCL INFO 3 : NET/0-3 GPU/59 GPU/60 GPU/61 GPU/62 GPU/63 GPU/56 GPU/57 GPU/58 NET/0-3 +n116-042-235:5922:5981 [3] NCCL INFO 4 : NET/0-4 GPU/60 GPU/59 GPU/58 GPU/57 GPU/56 GPU/63 GPU/62 GPU/61 NET/0-4 +n116-042-235:5922:5981 [3] NCCL INFO 5 : NET/0-5 GPU/61 GPU/62 GPU/63 GPU/56 GPU/57 GPU/58 GPU/59 GPU/60 NET/0-5 +n116-042-235:5922:5981 [3] NCCL INFO 6 : NET/0-6 GPU/62 GPU/61 GPU/60 GPU/59 GPU/58 GPU/57 GPU/56 GPU/63 NET/0-6 +n116-042-235:5922:5981 [3] NCCL INFO 7 : NET/0-7 GPU/63 GPU/56 GPU/57 GPU/58 GPU/59 GPU/60 GPU/61 GPU/62 NET/0-7 +n116-042-235:5922:5981 [3] NCCL INFO Pattern 1, crossNic 0, nChannels 8, bw 48.000000/24.000000, type NVL/PIX, sameChannels 0 +n116-042-235:5922:5981 [3] NCCL INFO 0 : NET/0-0 GPU/56 GPU/57 GPU/58 GPU/59 GPU/60 GPU/61 GPU/62 GPU/63 NET/0-0 +n116-042-235:5922:5981 [3] NCCL INFO 1 : NET/0-1 GPU/57 GPU/56 GPU/63 GPU/62 GPU/61 GPU/60 GPU/59 GPU/58 NET/0-1 +n116-042-235:5922:5981 [3] NCCL INFO 2 : NET/0-2 GPU/58 GPU/59 GPU/60 GPU/61 GPU/62 GPU/63 GPU/56 GPU/57 NET/0-2 +n116-042-235:5922:5981 [3] NCCL INFO 3 : NET/0-3 GPU/59 GPU/58 GPU/57 GPU/56 GPU/63 GPU/62 GPU/61 GPU/60 NET/0-3 +n116-042-235:5922:5981 [3] NCCL INFO 4 : NET/0-4 GPU/60 GPU/61 GPU/62 GPU/63 GPU/56 GPU/57 GPU/58 GPU/59 NET/0-4 +n116-042-235:5922:5981 [3] NCCL INFO 5 : NET/0-5 GPU/61 GPU/60 GPU/59 GPU/58 GPU/57 GPU/56 GPU/63 GPU/62 NET/0-5 +n116-042-235:5922:5981 [3] NCCL INFO 6 : NET/0-6 GPU/62 GPU/63 GPU/56 GPU/57 GPU/58 GPU/59 GPU/60 GPU/61 NET/0-6 +n116-042-235:5922:5981 [3] NCCL INFO 7 : NET/0-7 GPU/63 GPU/62 GPU/61 GPU/60 GPU/59 GPU/58 GPU/57 GPU/56 NET/0-7 +n116-042-235:5926:5975 [7] NCCL INFO comm 0xc302a60 rank 63 nRanks 64 nNodes 8 localRanks 8 localRank 7 MNNVL 0 +n116-042-235:5925:5978 [6] NCCL INFO comm 0xb61dba0 rank 62 nRanks 64 nNodes 8 localRanks 8 localRank 6 MNNVL 0 +n116-042-235:5924:5977 [5] NCCL INFO comm 0xd08cdf0 rank 61 nRanks 64 nNodes 8 localRanks 8 localRank 5 MNNVL 0 +n116-042-235:5926:5975 [7] NCCL INFO Tree 6 : 62 -> 63 -> 56/-1/-1 +n116-042-235:5926:5975 [7] NCCL INFO Tree 14 : 62 -> 63 -> 56/-1/-1 +n116-042-235:5926:5975 [7] NCCL INFO Tree 7 : 55 -> 63 -> 62/-1/-1 +n116-042-235:5926:5975 [7] NCCL INFO Tree 15 : -1 -> 63 -> 62/31/-1 +n116-042-235:5923:5980 [4] NCCL INFO comm 0xb3ebb20 rank 60 nRanks 64 nNodes 8 localRanks 8 localRank 4 MNNVL 0 +n116-042-235:5920:5982 [1] NCCL INFO comm 0xb1bcf90 rank 57 nRanks 64 nNodes 8 localRanks 8 localRank 1 MNNVL 0 +n116-042-235:5922:5981 [3] NCCL INFO comm 0xc5c7cf0 rank 59 nRanks 64 nNodes 8 localRanks 8 localRank 3 MNNVL 0 +n116-042-235:5926:5975 [7] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-042-235:5925:5978 [6] NCCL INFO Tree 6 : 54 -> 62 -> 63/-1/-1 +n116-042-235:5925:5978 [6] NCCL INFO Tree 14 : -1 -> 62 -> 63/30/-1 +n116-042-235:5925:5978 [6] NCCL INFO Tree 7 : 63 -> 62 -> 61/-1/-1 +n116-042-235:5925:5978 [6] NCCL INFO Tree 15 : 63 -> 62 -> 61/-1/-1 +n116-042-235:5924:5977 [5] NCCL INFO Tree 4 : 60 -> 61 -> 62/-1/-1 +n116-042-235:5921:5976 [2] NCCL INFO comm 0xc5cd980 rank 58 nRanks 64 nNodes 8 localRanks 8 localRank 2 MNNVL 0 +n116-042-235:5919:5979 [0] NCCL INFO comm 0xcb319f0 rank 56 nRanks 64 nNodes 8 localRanks 8 localRank 0 MNNVL 0 +n116-042-235:5925:5978 [6] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-042-235:5924:5977 [5] NCCL INFO Tree 12 : 60 -> 61 -> 62/-1/-1 +n116-042-235:5924:5977 [5] NCCL INFO Tree 5 : 53 -> 61 -> 60/-1/-1 +n116-042-235:5924:5977 [5] NCCL INFO Tree 13 : -1 -> 61 -> 60/29/-1 +n116-042-235:5924:5977 [5] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-042-235:5923:5980 [4] NCCL INFO Tree 4 : 52 -> 60 -> 61/-1/-1 +n116-042-235:5923:5980 [4] NCCL INFO Tree 12 : -1 -> 60 -> 61/28/-1 +n116-042-235:5923:5980 [4] NCCL INFO Tree 5 : 61 -> 60 -> 59/-1/-1 +n116-042-235:5923:5980 [4] NCCL INFO Tree 13 : 61 -> 60 -> 59/-1/-1 +n116-042-235:5926:5975 [7] NCCL INFO Ring 00 : 56 -> 63 -> 62 +n116-042-235:5923:5980 [4] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-042-235:5920:5982 [1] NCCL INFO Tree 0 : 56 -> 57 -> 58/-1/-1 +n116-042-235:5926:5975 [7] NCCL INFO Ring 01 : 62 -> 63 -> 56 +n116-042-235:5919:5979 [0] NCCL INFO Tree 0 : 48 -> 56 -> 57/-1/-1 +n116-042-235:5926:5975 [7] NCCL INFO Ring 02 : 56 -> 63 -> 62 +n116-042-235:5920:5982 [1] NCCL INFO Tree 8 : 56 -> 57 -> 58/-1/-1 +n116-042-235:5922:5981 [3] NCCL INFO Tree 2 : 58 -> 59 -> 60/-1/-1 +n116-042-235:5919:5979 [0] NCCL INFO Tree 8 : -1 -> 56 -> 57/24/-1 +n116-042-235:5926:5975 [7] NCCL INFO Ring 03 : 62 -> 63 -> 56 +n116-042-235:5920:5982 [1] NCCL INFO Tree 1 : 49 -> 57 -> 56/-1/-1 +n116-042-235:5922:5981 [3] NCCL INFO Tree 10 : 58 -> 59 -> 60/-1/-1 +n116-042-235:5919:5979 [0] NCCL INFO Tree 1 : 57 -> 56 -> 63/-1/-1 +n116-042-235:5925:5978 [6] NCCL INFO Ring 00 : 63 -> 62 -> 61 +n116-042-235:5926:5975 [7] NCCL INFO Ring 04 : 56 -> 63 -> 62 +n116-042-235:5920:5982 [1] NCCL INFO Tree 9 : -1 -> 57 -> 56/25/-1 +n116-042-235:5919:5979 [0] NCCL INFO Tree 9 : 57 -> 56 -> 63/-1/-1 +n116-042-235:5922:5981 [3] NCCL INFO Tree 3 : 51 -> 59 -> 58/-1/-1 +n116-042-235:5926:5975 [7] NCCL INFO Ring 05 : 62 -> 63 -> 56 +n116-042-235:5925:5978 [6] NCCL INFO Ring 01 : 61 -> 62 -> 63 +n116-042-235:5926:5975 [7] NCCL INFO Ring 06 : 56 -> 63 -> 6 +n116-042-235:5922:5981 [3] NCCL INFO Tree 11 : -1 -> 59 -> 58/27/-1 +n116-042-235:5921:5976 [2] NCCL INFO Tree 2 : 50 -> 58 -> 59/-1/-1 +n116-042-235:5926:5975 [7] NCCL INFO Ring 07 : 54 -> 63 -> 56 +n116-042-235:5926:5975 [7] NCCL INFO Ring 08 : 56 -> 63 -> 62 +n116-042-235:5919:5979 [0] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-042-235:5921:5976 [2] NCCL INFO Tree 10 : -1 -> 58 -> 59/26/-1 +n116-042-235:5925:5978 [6] NCCL INFO Ring 02 : 63 -> 62 -> 61 +n116-042-235:5920:5982 [1] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-042-235:5924:5977 [5] NCCL INFO Ring 00 : 62 -> 61 -> 60 +n116-042-235:5926:5975 [7] NCCL INFO Ring 09 : 62 -> 63 -> 56 +n116-042-235:5925:5978 [6] NCCL INFO Ring 03 : 61 -> 62 -> 63 +n116-042-235:5921:5976 [2] NCCL INFO Tree 3 : 59 -> 58 -> 57/-1/-1 +n116-042-235:5926:5975 [7] NCCL INFO Ring 10 : 56 -> 63 -> 62 +n116-042-235:5925:5978 [6] NCCL INFO Ring 04 : 63 -> 62 -> 61 +n116-042-235:5922:5981 [3] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-042-235:5924:5977 [5] NCCL INFO Ring 01 : 60 -> 61 -> 62 +n116-042-235:5921:5976 [2] NCCL INFO Tree 11 : 59 -> 58 -> 57/-1/-1 +n116-042-235:5924:5977 [5] NCCL INFO Ring 02 : 62 -> 61 -> 60 +n116-042-235:5926:5975 [7] NCCL INFO Ring 11 : 62 -> 63 -> 56 +n116-042-235:5925:5978 [6] NCCL INFO Ring 05 : 61 -> 62 -> 63 +n116-042-235:5923:5980 [4] NCCL INFO Ring 00 : 61 -> 60 -> 59 +n116-042-235:5924:5977 [5] NCCL INFO Ring 03 : 60 -> 61 -> 62 +n116-042-235:5926:5975 [7] NCCL INFO Ring 12 : 56 -> 63 -> 62 +n116-042-235:5925:5978 [6] NCCL INFO Ring 06 : 55 -> 62 -> 61 +n116-042-235:5923:5980 [4] NCCL INFO Ring 01 : 59 -> 60 -> 61 +n116-042-235:5924:5977 [5] NCCL INFO Ring 04 : 62 -> 61 -> 4 +n116-042-235:5925:5978 [6] NCCL INFO Ring 07 : 61 -> 62 -> 7 +n116-042-235:5923:5980 [4] NCCL INFO Ring 02 : 61 -> 60 -> 59 +n116-042-235:5924:5977 [5] NCCL INFO Ring 05 : 52 -> 61 -> 62 +n116-042-235:5926:5975 [7] NCCL INFO Ring 13 : 62 -> 63 -> 56 +n116-042-235:5925:5978 [6] NCCL INFO Ring 08 : 63 -> 62 -> 61 +n116-042-235:5921:5976 [2] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-042-235:5925:5978 [6] NCCL INFO Ring 09 : 61 -> 62 -> 63 +n116-042-235:5924:5977 [5] NCCL INFO Ring 06 : 62 -> 61 -> 60 +n116-042-235:5923:5980 [4] NCCL INFO Ring 03 : 59 -> 60 -> 61 +n116-042-235:5926:5975 [7] NCCL INFO Ring 14 : 56 -> 63 -> 6 +n116-042-235:5925:5978 [6] NCCL INFO Ring 10 : 63 -> 62 -> 61 +n116-042-235:5924:5977 [5] NCCL INFO Ring 07 : 60 -> 61 -> 62 +n116-042-235:5923:5980 [4] NCCL INFO Ring 04 : 53 -> 60 -> 59 +n116-042-235:5926:5975 [7] NCCL INFO Ring 15 : 54 -> 63 -> 56 +n116-042-235:5925:5978 [6] NCCL INFO Ring 11 : 61 -> 62 -> 63 +n116-042-235:5919:5979 [0] NCCL INFO Ring 00 : 49 -> 56 -> 63 +n116-042-235:5924:5977 [5] NCCL INFO Ring 08 : 62 -> 61 -> 60 +n116-042-235:5923:5980 [4] NCCL INFO Ring 05 : 59 -> 60 -> 5 +n116-042-235:5926:5975 [7] NCCL INFO Trees [0] -1/-1/-1->63->62 [1] 62/-1/-1->63->56 [2] 56/-1/-1->63->62 [3] 62/-1/-1->63->56 [4] 56/-1/-1->63->62 [5] 62/-1/-1->63->56 [6] 56/-1/-1->63->62 [7] 62/-1/-1->63->55 [8] -1/-1/-1->63->62 [9] 62/-1/-1->63->56 [10] 56/-1/-1->63->62 [11] 62/-1/-1->63->56 [12] 56/-1/-1->63->62 [13] 62/-1/-1->63->56 [14] 56/-1/-1->63->62 [15] 62/31/-1->63->-1 +n116-042-235:5925:5978 [6] NCCL INFO Ring 12 : 63 -> 62 -> 61 +n116-042-235:5919:5979 [0] NCCL INFO Ring 01 : 63 -> 56 -> 1 +n116-042-235:5924:5977 [5] NCCL INFO Ring 09 : 60 -> 61 -> 62 +n116-042-235:5919:5979 [0] NCCL INFO Ring 02 : 57 -> 56 -> 63 +n116-042-235:5926:5975 [7] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-042-235:5923:5980 [4] NCCL INFO Ring 06 : 61 -> 60 -> 59 +n116-042-235:5925:5978 [6] NCCL INFO Ring 13 : 61 -> 62 -> 63 +n116-042-235:5926:5975 [7] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-042-235:5920:5982 [1] NCCL INFO Ring 00 : 58 -> 57 -> 0 +n116-042-235:5924:5977 [5] NCCL INFO Ring 10 : 62 -> 61 -> 60 +n116-042-235:5919:5979 [0] NCCL INFO Ring 03 : 63 -> 56 -> 57 +n116-042-235:5923:5980 [4] NCCL INFO Ring 07 : 59 -> 60 -> 61 +n116-042-235:5926:5975 [7] NCCL INFO P2P Chunksize set to 524288 +n116-042-235:5925:5978 [6] NCCL INFO Ring 14 : 55 -> 62 -> 61 +n116-042-235:5924:5977 [5] NCCL INFO Ring 11 : 60 -> 61 -> 62 +n116-042-235:5920:5982 [1] NCCL INFO Ring 01 : 48 -> 57 -> 58 +n116-042-235:5922:5981 [3] NCCL INFO Ring 00 : 60 -> 59 -> 58 +n116-042-235:5919:5979 [0] NCCL INFO Ring 04 : 57 -> 56 -> 63 +n116-042-235:5923:5980 [4] NCCL INFO Ring 08 : 61 -> 60 -> 59 +n116-042-235:5920:5982 [1] NCCL INFO Ring 02 : 58 -> 57 -> 56 +n116-042-235:5923:5980 [4] NCCL INFO Ring 09 : 59 -> 60 -> 61 +n116-042-235:5923:5980 [4] NCCL INFO Ring 10 : 61 -> 60 -> 59 +n116-042-235:5920:5982 [1] NCCL INFO Ring 03 : 56 -> 57 -> 58 +n116-042-235:5925:5978 [6] NCCL INFO Ring 15 : 61 -> 62 -> 7 +n116-042-235:5924:5977 [5] NCCL INFO Ring 12 : 62 -> 61 -> 4 +n116-042-235:5922:5981 [3] NCCL INFO Ring 01 : 58 -> 59 -> 60 +n116-042-235:5919:5979 [0] NCCL INFO Ring 05 : 63 -> 56 -> 57 +n116-042-235:5919:5979 [0] NCCL INFO Ring 06 : 57 -> 56 -> 63 +n116-042-235:5923:5980 [4] NCCL INFO Ring 11 : 59 -> 60 -> 61 +n116-042-235:5920:5982 [1] NCCL INFO Ring 04 : 58 -> 57 -> 56 +n116-042-235:5925:5978 [6] NCCL INFO Trees [0] 63/-1/-1->62->61 [1] 61/-1/-1->62->63 [2] 63/-1/-1->62->61 [3] 61/-1/-1->62->63 [4] 63/-1/-1->62->61 [5] -1/-1/-1->62->63 [6] 63/-1/-1->62->54 [7] 61/-1/-1->62->63 [8] 63/-1/-1->62->61 [9] 61/-1/-1->62->63 [10] 63/-1/-1->62->61 [11] 61/-1/-1->62->63 [12] 63/-1/-1->62->61 [13] -1/-1/-1->62->63 [14] 63/30/-1->62->-1 [15] 61/-1/-1->62->63 +n116-042-235:5924:5977 [5] NCCL INFO Ring 13 : 52 -> 61 -> 62 +n116-042-235:5922:5981 [3] NCCL INFO Ring 02 : 60 -> 59 -> 2 +n116-042-235:5921:5976 [2] NCCL INFO Ring 00 : 59 -> 58 -> 57 +n116-042-235:5919:5979 [0] NCCL INFO Ring 07 : 63 -> 56 -> 57 +n116-042-235:5921:5976 [2] NCCL INFO Ring 01 : 57 -> 58 -> 59 +n116-042-235:5923:5980 [4] NCCL INFO Ring 12 : 53 -> 60 -> 59 +n116-042-235:5920:5982 [1] NCCL INFO Ring 05 : 56 -> 57 -> 58 +n116-042-235:5924:5977 [5] NCCL INFO Ring 14 : 62 -> 61 -> 60 +n116-042-235:5919:5979 [0] NCCL INFO Ring 08 : 49 -> 56 -> 63 +n116-042-235:5922:5981 [3] NCCL INFO Ring 03 : 50 -> 59 -> 60 +n116-042-235:5925:5978 [6] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-042-235:5922:5981 [3] NCCL INFO Ring 04 : 60 -> 59 -> 58 +n116-042-235:5923:5980 [4] NCCL INFO Ring 13 : 59 -> 60 -> 5 +n116-042-235:5925:5978 [6] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-042-235:5921:5976 [2] NCCL INFO Ring 02 : 51 -> 58 -> 57 +n116-042-235:5921:5976 [2] NCCL INFO Ring 03 : 57 -> 58 -> 3 +n116-042-235:5920:5982 [1] NCCL INFO Ring 06 : 58 -> 57 -> 56 +n116-042-235:5924:5977 [5] NCCL INFO Ring 15 : 60 -> 61 -> 62 +n116-042-235:5919:5979 [0] NCCL INFO Ring 09 : 63 -> 56 -> 1 +n116-042-235:5922:5981 [3] NCCL INFO Ring 05 : 58 -> 59 -> 60 +n116-042-235:5923:5980 [4] NCCL INFO Ring 14 : 61 -> 60 -> 59 +n116-042-235:5925:5978 [6] NCCL INFO P2P Chunksize set to 524288 +n116-042-235:5919:5979 [0] NCCL INFO Ring 10 : 57 -> 56 -> 63 +n116-042-235:5921:5976 [2] NCCL INFO Ring 04 : 59 -> 58 -> 57 +n116-042-235:5920:5982 [1] NCCL INFO Ring 07 : 56 -> 57 -> 58 +n116-042-235:5924:5977 [5] NCCL INFO Trees [0] 62/-1/-1->61->60 [1] 60/-1/-1->61->62 [2] 62/-1/-1->61->60 [3] 60/-1/-1->61->62 [4] 62/-1/-1->61->60 [5] 60/-1/-1->61->53 [6] -1/-1/-1->61->60 [7] 60/-1/-1->61->62 [8] 62/-1/-1->61->60 [9] 60/-1/-1->61->62 [10] 62/-1/-1->61->60 [11] 60/-1/-1->61->62 [12] 62/-1/-1->61->60 [13] 60/29/-1->61->-1 [14] -1/-1/-1->61->60 [15] 60/-1/-1->61->62 +n116-042-235:5922:5981 [3] NCCL INFO Ring 06 : 60 -> 59 -> 58 +n116-042-235:5923:5980 [4] NCCL INFO Ring 15 : 59 -> 60 -> 61 +n116-042-235:5919:5979 [0] NCCL INFO Ring 11 : 63 -> 56 -> 57 +n116-042-235:5921:5976 [2] NCCL INFO Ring 05 : 57 -> 58 -> 59 +n116-042-235:5923:5980 [4] NCCL INFO Trees [0] 61/-1/-1->60->59 [1] 59/-1/-1->60->61 [2] 61/-1/-1->60->59 [3] -1/-1/-1->60->61 [4] 61/-1/-1->60->52 [5] 59/-1/-1->60->61 [6] 61/-1/-1->60->59 [7] 59/-1/-1->60->61 [8] 61/-1/-1->60->59 [9] 59/-1/-1->60->61 [10] 61/-1/-1->60->59 [11] -1/-1/-1->60->61 [12] 61/28/-1->60->-1 [13] 59/-1/-1->60->61 [14] 61/-1/-1->60->59 [15] 59/-1/-1->60->61 +n116-042-235:5921:5976 [2] NCCL INFO Ring 06 : 59 -> 58 -> 57 +n116-042-235:5920:5982 [1] NCCL INFO Ring 08 : 58 -> 57 -> 0 +n116-042-235:5920:5982 [1] NCCL INFO Ring 09 : 48 -> 57 -> 58 +n116-042-235:5922:5981 [3] NCCL INFO Ring 07 : 58 -> 59 -> 60 +n116-042-235:5919:5979 [0] NCCL INFO Ring 12 : 57 -> 56 -> 63 +n116-042-235:5924:5977 [5] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-042-235:5921:5976 [2] NCCL INFO Ring 07 : 57 -> 58 -> 59 +n116-042-235:5923:5980 [4] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-042-235:5924:5977 [5] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-042-235:5922:5981 [3] NCCL INFO Ring 08 : 60 -> 59 -> 58 +n116-042-235:5924:5977 [5] NCCL INFO P2P Chunksize set to 524288 +n116-042-235:5920:5982 [1] NCCL INFO Ring 10 : 58 -> 57 -> 56 +n116-042-235:5922:5981 [3] NCCL INFO Ring 09 : 58 -> 59 -> 60 +n116-042-235:5919:5979 [0] NCCL INFO Ring 13 : 63 -> 56 -> 57 +n116-042-235:5921:5976 [2] NCCL INFO Ring 08 : 59 -> 58 -> 57 +n116-042-235:5923:5980 [4] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-042-235:5920:5982 [1] NCCL INFO Ring 11 : 56 -> 57 -> 58 +n116-042-235:5922:5981 [3] NCCL INFO Ring 10 : 60 -> 59 -> 2 +n116-042-235:5919:5979 [0] NCCL INFO Ring 14 : 57 -> 56 -> 63 +n116-042-235:5920:5982 [1] NCCL INFO Ring 12 : 58 -> 57 -> 56 +n116-042-235:5921:5976 [2] NCCL INFO Ring 09 : 57 -> 58 -> 59 +n116-042-235:5923:5980 [4] NCCL INFO P2P Chunksize set to 524288 +n116-042-235:5922:5981 [3] NCCL INFO Ring 11 : 50 -> 59 -> 60 +n116-042-235:5919:5979 [0] NCCL INFO Ring 15 : 63 -> 56 -> 57 +n116-042-235:5920:5982 [1] NCCL INFO Ring 13 : 56 -> 57 -> 58 +n116-042-235:5921:5976 [2] NCCL INFO Ring 10 : 51 -> 58 -> 57 +n116-042-235:5922:5981 [3] NCCL INFO Ring 12 : 60 -> 59 -> 58 +n116-042-235:5919:5979 [0] NCCL INFO Trees [0] 57/-1/-1->56->48 [1] 63/-1/-1->56->57 [2] 57/-1/-1->56->63 [3] 63/-1/-1->56->57 [4] 57/-1/-1->56->63 [5] 63/-1/-1->56->57 [6] 57/-1/-1->56->63 [7] -1/-1/-1->56->57 [8] 57/24/-1->56->-1 [9] 63/-1/-1->56->57 [10] 57/-1/-1->56->63 [11] 63/-1/-1->56->57 [12] 57/-1/-1->56->63 [13] 63/-1/-1->56->57 [14] 57/-1/-1->56->63 [15] -1/-1/-1->56->57 +n116-042-235:5920:5982 [1] NCCL INFO Ring 14 : 58 -> 57 -> 56 +n116-042-235:5922:5981 [3] NCCL INFO Ring 13 : 58 -> 59 -> 60 +n116-042-235:5921:5976 [2] NCCL INFO Ring 11 : 57 -> 58 -> 3 +n116-042-235:5920:5982 [1] NCCL INFO Ring 15 : 56 -> 57 -> 58 +n116-042-235:5922:5981 [3] NCCL INFO Ring 14 : 60 -> 59 -> 58 +n116-042-235:5919:5979 [0] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-042-235:5920:5982 [1] NCCL INFO Trees [0] 58/-1/-1->57->56 [1] 56/-1/-1->57->49 [2] -1/-1/-1->57->56 [3] 56/-1/-1->57->58 [4] 58/-1/-1->57->56 [5] 56/-1/-1->57->58 [6] 58/-1/-1->57->56 [7] 56/-1/-1->57->58 [8] 58/-1/-1->57->56 [9] 56/25/-1->57->-1 [10] -1/-1/-1->57->56 [11] 56/-1/-1->57->58 [12] 58/-1/-1->57->56 [13] 56/-1/-1->57->58 [14] 58/-1/-1->57->56 [15] 56/-1/-1->57->58 +n116-042-235:5922:5981 [3] NCCL INFO Ring 15 : 58 -> 59 -> 60 +n116-042-235:5919:5979 [0] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-042-235:5921:5976 [2] NCCL INFO Ring 12 : 59 -> 58 -> 57 +n116-042-235:5919:5979 [0] NCCL INFO P2P Chunksize set to 524288 +n116-042-235:5920:5982 [1] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-042-235:5922:5981 [3] NCCL INFO Trees [0] 60/-1/-1->59->58 [1] 58/-1/-1->59->60 [2] 60/-1/-1->59->58 [3] 58/-1/-1->59->51 [4] -1/-1/-1->59->58 [5] 58/-1/-1->59->60 [6] 60/-1/-1->59->58 [7] 58/-1/-1->59->60 [8] 60/-1/-1->59->58 [9] 58/-1/-1->59->60 [10] 60/-1/-1->59->58 [11] 58/27/-1->59->-1 [12] -1/-1/-1->59->58 [13] 58/-1/-1->59->60 [14] 60/-1/-1->59->58 [15] 58/-1/-1->59->60 +n116-042-235:5921:5976 [2] NCCL INFO Ring 13 : 57 -> 58 -> 59 +n116-042-235:5920:5982 [1] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-042-235:5921:5976 [2] NCCL INFO Ring 14 : 59 -> 58 -> 57 +n116-042-235:5920:5982 [1] NCCL INFO P2P Chunksize set to 524288 +n116-042-235:5922:5981 [3] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-042-235:5921:5976 [2] NCCL INFO Ring 15 : 57 -> 58 -> 59 +n116-042-235:5922:5981 [3] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-042-235:5921:5976 [2] NCCL INFO Trees [0] 59/-1/-1->58->57 [1] -1/-1/-1->58->59 [2] 59/-1/-1->58->50 [3] 57/-1/-1->58->59 [4] 59/-1/-1->58->57 [5] 57/-1/-1->58->59 [6] 59/-1/-1->58->57 [7] 57/-1/-1->58->59 [8] 59/-1/-1->58->57 [9] -1/-1/-1->58->59 [10] 59/26/-1->58->-1 [11] 57/-1/-1->58->59 [12] 59/-1/-1->58->57 [13] 57/-1/-1->58->59 [14] 59/-1/-1->58->57 [15] 57/-1/-1->58->59 +n116-042-235:5922:5981 [3] NCCL INFO P2P Chunksize set to 524288 +n116-042-235:5921:5976 [2] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-042-235:5921:5976 [2] NCCL INFO NCCL_P2P_NET_CHUNKSIZE set by environment to 524288. +n116-042-235:5921:5976 [2] NCCL INFO P2P Chunksize set to 524288 +n116-042-235:5925:6016 [6] NCCL INFO [Proxy Service] Device 6 CPU core 189 +n116-042-235:5925:6019 [6] NCCL INFO [Proxy Service UDS] Device 6 CPU core 59 +n116-042-235:5926:6015 [7] NCCL INFO [Proxy Service] Device 7 CPU core 179 +n116-042-235:5919:6021 [0] NCCL INFO [Proxy Service] Device 0 CPU core 29 +n116-042-235:5924:6018 [5] NCCL INFO [Proxy Service] Device 5 CPU core 194 +n116-042-235:5921:6025 [2] NCCL INFO [Proxy Service] Device 2 CPU core 50 +n116-042-235:5920:6024 [1] NCCL INFO [Proxy Service] Device 1 CPU core 153 +n116-042-235:5926:6017 [7] NCCL INFO [Proxy Service UDS] Device 7 CPU core 65 +n116-042-235:5924:6022 [5] NCCL INFO [Proxy Service UDS] Device 5 CPU core 193 +n116-042-235:5919:6026 [0] NCCL INFO [Proxy Service UDS] Device 0 CPU core 131 +n116-042-235:5923:6023 [4] NCCL INFO [Proxy Service UDS] Device 4 CPU core 162 +n116-042-235:5922:6027 [3] NCCL INFO [Proxy Service] Device 3 CPU core 134 +n116-042-235:5920:6028 [1] NCCL INFO [Proxy Service UDS] Device 1 CPU core 32 +n116-042-235:5923:6020 [4] NCCL INFO [Proxy Service] Device 4 CPU core 159 +n116-042-235:5922:6030 [3] NCCL INFO [Proxy Service UDS] Device 3 CPU core 106 +n116-042-235:5921:6029 [2] NCCL INFO [Proxy Service UDS] Device 2 CPU core 36 +n116-042-235:5922:5981 [3] NCCL INFO NCCL_PROTO set by environment to Simple +n116-042-235:5922:5981 [3] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-042-235:5922:5981 [3] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n116-042-235:5922:5981 [3] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n116-042-235:5919:5979 [0] NCCL INFO NCCL_PROTO set by environment to Simple +n116-042-235:5919:5979 [0] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-042-235:5919:5979 [0] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n116-042-235:5919:5979 [0] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n116-042-235:5921:5976 [2] NCCL INFO NCCL_PROTO set by environment to Simple +n116-042-235:5921:5976 [2] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-042-235:5921:5976 [2] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n116-042-235:5921:5976 [2] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n116-042-235:5925:5978 [6] NCCL INFO NCCL_PROTO set by environment to Simple +n116-042-235:5925:5978 [6] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-042-235:5925:5978 [6] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n116-042-235:5925:5978 [6] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n116-042-235:5920:5982 [1] NCCL INFO NCCL_PROTO set by environment to Simple +n116-042-235:5920:5982 [1] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-042-235:5920:5982 [1] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n116-042-235:5920:5982 [1] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n116-042-235:5924:5977 [5] NCCL INFO NCCL_PROTO set by environment to Simple +n116-042-235:5924:5977 [5] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-042-235:5924:5977 [5] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n116-042-235:5924:5977 [5] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n116-042-235:5926:5975 [7] NCCL INFO NCCL_PROTO set by environment to Simple +n116-042-235:5926:5975 [7] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-042-235:5926:5975 [7] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n116-042-235:5926:5975 [7] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n116-042-235:5923:5980 [4] NCCL INFO NCCL_PROTO set by environment to Simple +n116-042-235:5923:5980 [4] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-042-235:5923:5980 [4] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512 +n116-042-235:5923:5980 [4] NCCL INFO 16 coll channels, 16 collnet channels, 0 nvls channels, 16 p2p channels, 1 p2p channels per peer +n116-042-235:5926:5975 [7] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-042-235:5926:5975 [7] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-042-235:5926:5975 [7] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-042-235:5926:5975 [7] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-042-235:5919:5979 [0] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-042-235:5926:5975 [7] NCCL INFO ncclCommInitRankConfig comm 0xc302a60 rank 63 nranks 64 cudaDev 7 nvmlDev 7 busId 8c000 commId 0xfb2d11625d05007a - Init COMPLETE +n116-042-235:5919:5979 [0] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-042-235:5919:5979 [0] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-042-235:5926:5975 [7] NCCL INFO Init timings - ncclCommInitRankConfig: rank 63 nranks 64 total 5.05 (kernels 0.29, alloc 1.48, bootstrap 2.39, allgathers 0.13, topo 0.63, graphs 0.08, connections 0.03, rest 0.00) +n116-042-235:5919:5979 [0] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-042-235:5919:5979 [0] NCCL INFO ncclCommInitRankConfig comm 0xcb319f0 rank 56 nranks 64 cudaDev 0 nvmlDev 0 busId 4000 commId 0xfb2d11625d05007a - Init COMPLETE +n116-042-235:5921:5976 [2] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-042-235:5919:5979 [0] NCCL INFO Init timings - ncclCommInitRankConfig: rank 56 nranks 64 total 5.00 (kernels 0.30, alloc 1.52, bootstrap 2.29, allgathers 0.13, topo 0.63, graphs 0.08, connections 0.03, rest 0.01) +n116-042-235:5921:5976 [2] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-042-235:5921:5976 [2] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-042-235:5921:5976 [2] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-042-235:5921:5976 [2] NCCL INFO ncclCommInitRankConfig comm 0xc5cd980 rank 58 nranks 64 cudaDev 2 nvmlDev 2 busId b000 commId 0xfb2d11625d05007a - Init COMPLETE +n116-042-235:5921:5976 [2] NCCL INFO Init timings - ncclCommInitRankConfig: rank 58 nranks 64 total 5.04 (kernels 0.28, alloc 1.47, bootstrap 2.41, allgathers 0.13, topo 0.63, graphs 0.08, connections 0.03, rest 0.01) +n116-042-235:5920:5982 [1] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-042-235:5920:5982 [1] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-042-235:5920:5982 [1] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-042-235:5920:5982 [1] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-042-235:5920:5982 [1] NCCL INFO ncclCommInitRankConfig comm 0xb1bcf90 rank 57 nranks 64 cudaDev 1 nvmlDev 1 busId 5000 commId 0xfb2d11625d05007a - Init COMPLETE +n116-042-235:5922:5981 [3] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-042-235:5920:5982 [1] NCCL INFO Init timings - ncclCommInitRankConfig: rank 57 nranks 64 total 4.98 (kernels 0.39, alloc 1.46, bootstrap 2.25, allgathers 0.13, topo 0.63, graphs 0.08, connections 0.03, rest 0.01) +n116-042-235:5922:5981 [3] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-042-235:5922:5981 [3] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-042-235:5922:5981 [3] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-042-235:5922:5981 [3] NCCL INFO ncclCommInitRankConfig comm 0xc5c7cf0 rank 59 nranks 64 cudaDev 3 nvmlDev 3 busId c000 commId 0xfb2d11625d05007a - Init COMPLETE +n116-042-235:5922:5981 [3] NCCL INFO Init timings - ncclCommInitRankConfig: rank 59 nranks 64 total 4.98 (kernels 0.33, alloc 1.51, bootstrap 2.25, allgathers 0.13, topo 0.63, graphs 0.08, connections 0.02, rest 0.01) +n116-042-235:5924:5977 [5] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-042-235:5924:5977 [5] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-042-235:5924:5977 [5] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-042-235:5924:5977 [5] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-042-235:5923:5980 [4] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-042-235:5924:5977 [5] NCCL INFO ncclCommInitRankConfig comm 0xd08cdf0 rank 61 nranks 64 cudaDev 5 nvmlDev 5 busId 85000 commId 0xfb2d11625d05007a - Init COMPLETE +n116-042-235:5923:5980 [4] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-042-235:5924:5977 [5] NCCL INFO Init timings - ncclCommInitRankConfig: rank 61 nranks 64 total 5.02 (kernels 0.29, alloc 1.54, bootstrap 2.31, allgathers 0.13, topo 0.63, graphs 0.08, connections 0.03, rest 0.01) +n116-042-235:5923:5980 [4] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-042-235:5923:5980 [4] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-042-235:5923:5980 [4] NCCL INFO ncclCommInitRankConfig comm 0xb3ebb20 rank 60 nranks 64 cudaDev 4 nvmlDev 4 busId 84000 commId 0xfb2d11625d05007a - Init COMPLETE +n116-042-235:5923:5980 [4] NCCL INFO Init timings - ncclCommInitRankConfig: rank 60 nranks 64 total 4.98 (kernels 0.33, alloc 1.51, bootstrap 2.27, allgathers 0.13, topo 0.63, graphs 0.08, connections 0.04, rest 0.00) +n116-042-235:5925:5978 [6] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-042-235:5925:5978 [6] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-042-235:5925:5978 [6] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-042-235:5925:5978 [6] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-042-235:5925:5978 [6] NCCL INFO ncclCommInitRankConfig comm 0xb61dba0 rank 62 nranks 64 cudaDev 6 nvmlDev 6 busId 8b000 commId 0xfb2d11625d05007a - Init COMPLETE +n116-042-235:5925:5978 [6] NCCL INFO Init timings - ncclCommInitRankConfig: rank 62 nranks 64 total 5.01 (kernels 0.29, alloc 1.53, bootstrap 2.31, allgathers 0.13, topo 0.63, graphs 0.08, connections 0.03, rest 0.01) +n116-042-235:5919:6032 [0] NCCL INFO Channel 00/0 : 56[0] -> 57[1] via P2P/CUMEM +n116-042-235:5919:6032 [0] NCCL INFO Channel 01/0 : 56[0] -> 57[1] via P2P/CUMEM +n116-042-235:5919:6032 [0] NCCL INFO Channel 02/0 : 56[0] -> 57[1] via P2P/CUMEM +n116-042-235:5919:6032 [0] NCCL INFO Channel 03/0 : 56[0] -> 57[1] via P2P/CUMEM +n116-042-235:5919:6032 [0] NCCL INFO Channel 04/0 : 56[0] -> 57[1] via P2P/CUMEM +n116-042-235:5919:6032 [0] NCCL INFO Channel 05/0 : 56[0] -> 57[1] via P2P/CUMEM +n116-042-235:5919:6032 [0] NCCL INFO Channel 06/0 : 56[0] -> 57[1] via P2P/CUMEM +n116-042-235:5919:6032 [0] NCCL INFO Channel 07/0 : 56[0] -> 57[1] via P2P/CUMEM +n116-042-235:5919:6032 [0] NCCL INFO Channel 08/0 : 56[0] -> 57[1] via P2P/CUMEM +n116-042-235:5919:6032 [0] NCCL INFO Channel 09/0 : 56[0] -> 57[1] via P2P/CUMEM +n116-042-235:5919:6032 [0] NCCL INFO Channel 10/0 : 56[0] -> 57[1] via P2P/CUMEM +n116-042-235:5919:6032 [0] NCCL INFO Channel 11/0 : 56[0] -> 57[1] via P2P/CUMEM +n116-042-235:5925:6036 [6] NCCL INFO Channel 00/0 : 62[6] -> 63[7] via P2P/CUMEM +n116-042-235:5921:6033 [2] NCCL INFO Channel 00/0 : 58[2] -> 59[3] via P2P/CUMEM +n116-042-235:5923:6037 [4] NCCL INFO Channel 00/0 : 60[4] -> 61[5] via P2P/CUMEM +n116-042-235:5919:6032 [0] NCCL INFO Channel 12/0 : 56[0] -> 57[1] via P2P/CUMEM +n116-042-235:5925:6036 [6] NCCL INFO Channel 01/0 : 62[6] -> 63[7] via P2P/CUMEM +n116-042-235:5921:6033 [2] NCCL INFO Channel 01/0 : 58[2] -> 59[3] via P2P/CUMEM +n116-042-235:5923:6037 [4] NCCL INFO Channel 01/0 : 60[4] -> 61[5] via P2P/CUMEM +n116-042-235:5919:6032 [0] NCCL INFO Channel 13/0 : 56[0] -> 57[1] via P2P/CUMEM +n116-042-235:5925:6036 [6] NCCL INFO Channel 02/0 : 62[6] -> 63[7] via P2P/CUMEM +n116-042-235:5921:6033 [2] NCCL INFO Channel 02/0 : 58[2] -> 59[3] via P2P/CUMEM +n116-042-235:5923:6037 [4] NCCL INFO Channel 02/0 : 60[4] -> 61[5] via P2P/CUMEM +n116-042-235:5919:6032 [0] NCCL INFO Channel 14/0 : 56[0] -> 57[1] via P2P/CUMEM +n116-042-235:5925:6036 [6] NCCL INFO Channel 03/0 : 62[6] -> 63[7] via P2P/CUMEM +n116-042-235:5921:6033 [2] NCCL INFO Channel 03/0 : 58[2] -> 59[3] via P2P/CUMEM +n116-042-235:5923:6037 [4] NCCL INFO Channel 03/0 : 60[4] -> 61[5] via P2P/CUMEM +n116-042-235:5919:6032 [0] NCCL INFO Channel 15/0 : 56[0] -> 57[1] via P2P/CUMEM +n116-042-235:5924:6034 [5] NCCL INFO Channel 00/0 : 61[5] -> 62[6] via P2P/CUMEM +n116-042-235:5925:6036 [6] NCCL INFO Channel 04/0 : 62[6] -> 63[7] via P2P/CUMEM +n116-042-235:5921:6033 [2] NCCL INFO Channel 04/0 : 58[2] -> 59[3] via P2P/CUMEM +n116-042-235:5923:6037 [4] NCCL INFO Channel 04/0 : 60[4] -> 61[5] via P2P/CUMEM +n116-042-235:5920:6035 [1] NCCL INFO Channel 00/0 : 57[1] -> 58[2] via P2P/CUMEM +n116-042-235:5924:6034 [5] NCCL INFO Channel 01/0 : 61[5] -> 62[6] via P2P/CUMEM +n116-042-235:5925:6036 [6] NCCL INFO Channel 05/0 : 62[6] -> 63[7] via P2P/CUMEM +n116-042-235:5921:6033 [2] NCCL INFO Channel 05/0 : 58[2] -> 59[3] via P2P/CUMEM +n116-042-235:5923:6037 [4] NCCL INFO Channel 05/0 : 60[4] -> 61[5] via P2P/CUMEM +n116-042-235:5920:6035 [1] NCCL INFO Channel 03/0 : 57[1] -> 58[2] via P2P/CUMEM +n116-042-235:5924:6034 [5] NCCL INFO Channel 02/0 : 61[5] -> 62[6] via P2P/CUMEM +n116-042-235:5925:6036 [6] NCCL INFO Channel 06/0 : 62[6] -> 63[7] via P2P/CUMEM +n116-042-235:5923:6037 [4] NCCL INFO Channel 06/0 : 60[4] -> 61[5] via P2P/CUMEM +n116-042-235:5921:6033 [2] NCCL INFO Channel 06/0 : 58[2] -> 59[3] via P2P/CUMEM +n116-042-235:5920:6035 [1] NCCL INFO Channel 04/0 : 57[1] -> 58[2] via P2P/CUMEM +n116-042-235:5924:6034 [5] NCCL INFO Channel 03/0 : 61[5] -> 62[6] via P2P/CUMEM +n116-042-235:5925:6036 [6] NCCL INFO Channel 07/0 : 62[6] -> 63[7] via P2P/CUMEM +n116-042-235:5923:6037 [4] NCCL INFO Channel 07/0 : 60[4] -> 61[5] via P2P/CUMEM +n116-042-235:5921:6033 [2] NCCL INFO Channel 07/0 : 58[2] -> 59[3] via P2P/CUMEM +n116-042-235:5920:6035 [1] NCCL INFO Channel 05/0 : 57[1] -> 58[2] via P2P/CUMEM +n116-042-235:5924:6034 [5] NCCL INFO Channel 04/0 : 61[5] -> 62[6] via P2P/CUMEM +n116-042-235:5925:6036 [6] NCCL INFO Channel 08/0 : 62[6] -> 63[7] via P2P/CUMEM +n116-042-235:5923:6037 [4] NCCL INFO Channel 08/0 : 60[4] -> 61[5] via P2P/CUMEM +n116-042-235:5921:6033 [2] NCCL INFO Channel 08/0 : 58[2] -> 59[3] via P2P/CUMEM +n116-042-235:5920:6035 [1] NCCL INFO Channel 06/0 : 57[1] -> 58[2] via P2P/CUMEM +n116-042-235:5924:6034 [5] NCCL INFO Channel 07/0 : 61[5] -> 62[6] via P2P/CUMEM +n116-042-235:5925:6036 [6] NCCL INFO Channel 09/0 : 62[6] -> 63[7] via P2P/CUMEM +n116-042-235:5923:6037 [4] NCCL INFO Channel 09/0 : 60[4] -> 61[5] via P2P/CUMEM +n116-042-235:5921:6033 [2] NCCL INFO Channel 09/0 : 58[2] -> 59[3] via P2P/CUMEM +n116-042-235:5924:6034 [5] NCCL INFO Channel 08/0 : 61[5] -> 62[6] via P2P/CUMEM +n116-042-235:5920:6035 [1] NCCL INFO Channel 07/0 : 57[1] -> 58[2] via P2P/CUMEM +n116-042-235:5923:6037 [4] NCCL INFO Channel 10/0 : 60[4] -> 61[5] via P2P/CUMEM +n116-042-235:5925:6036 [6] NCCL INFO Channel 10/0 : 62[6] -> 63[7] via P2P/CUMEM +n116-042-235:5921:6033 [2] NCCL INFO Channel 10/0 : 58[2] -> 59[3] via P2P/CUMEM +n116-042-235:5924:6034 [5] NCCL INFO Channel 09/0 : 61[5] -> 62[6] via P2P/CUMEM +n116-042-235:5920:6035 [1] NCCL INFO Channel 08/0 : 57[1] -> 58[2] via P2P/CUMEM +n116-042-235:5923:6037 [4] NCCL INFO Channel 11/0 : 60[4] -> 61[5] via P2P/CUMEM +n116-042-235:5925:6036 [6] NCCL INFO Channel 11/0 : 62[6] -> 63[7] via P2P/CUMEM +n116-042-235:5921:6033 [2] NCCL INFO Channel 11/0 : 58[2] -> 59[3] via P2P/CUMEM +n116-042-235:5924:6034 [5] NCCL INFO Channel 10/0 : 61[5] -> 62[6] via P2P/CUMEM +n116-042-235:5920:6035 [1] NCCL INFO Channel 11/0 : 57[1] -> 58[2] via P2P/CUMEM +n116-042-235:5923:6037 [4] NCCL INFO Channel 12/0 : 60[4] -> 61[5] via P2P/CUMEM +n116-042-235:5925:6036 [6] NCCL INFO Channel 12/0 : 62[6] -> 63[7] via P2P/CUMEM +n116-042-235:5921:6033 [2] NCCL INFO Channel 12/0 : 58[2] -> 59[3] via P2P/CUMEM +n116-042-235:5922:6038 [3] NCCL INFO Channel 00/0 : 59[3] -> 60[4] via P2P/CUMEM +n116-042-235:5924:6034 [5] NCCL INFO Channel 11/0 : 61[5] -> 62[6] via P2P/CUMEM +n116-042-235:5923:6037 [4] NCCL INFO Channel 13/0 : 60[4] -> 61[5] via P2P/CUMEM +n116-042-235:5920:6035 [1] NCCL INFO Channel 12/0 : 57[1] -> 58[2] via P2P/CUMEM +n116-042-235:5925:6036 [6] NCCL INFO Channel 13/0 : 62[6] -> 63[7] via P2P/CUMEM +n116-042-235:5921:6033 [2] NCCL INFO Channel 13/0 : 58[2] -> 59[3] via P2P/CUMEM +n116-042-235:5922:6038 [3] NCCL INFO Channel 01/0 : 59[3] -> 60[4] via P2P/CUMEM +n116-042-235:5924:6034 [5] NCCL INFO Channel 12/0 : 61[5] -> 62[6] via P2P/CUMEM +n116-042-235:5923:6037 [4] NCCL INFO Channel 14/0 : 60[4] -> 61[5] via P2P/CUMEM +n116-042-235:5920:6035 [1] NCCL INFO Channel 13/0 : 57[1] -> 58[2] via P2P/CUMEM +n116-042-235:5925:6036 [6] NCCL INFO Channel 14/0 : 62[6] -> 63[7] via P2P/CUMEM +n116-042-235:5921:6033 [2] NCCL INFO Channel 14/0 : 58[2] -> 59[3] via P2P/CUMEM +n116-042-235:5922:6038 [3] NCCL INFO Channel 02/0 : 59[3] -> 60[4] via P2P/CUMEM +n116-042-235:5924:6034 [5] NCCL INFO Channel 15/0 : 61[5] -> 62[6] via P2P/CUMEM +n116-042-235:5923:6037 [4] NCCL INFO Channel 15/0 : 60[4] -> 61[5] via P2P/CUMEM +n116-042-235:5920:6035 [1] NCCL INFO Channel 14/0 : 57[1] -> 58[2] via P2P/CUMEM +n116-042-235:5925:6036 [6] NCCL INFO Channel 15/0 : 62[6] -> 63[7] via P2P/CUMEM +n116-042-235:5921:6033 [2] NCCL INFO Channel 15/0 : 58[2] -> 59[3] via P2P/CUMEM +n116-042-235:5922:6038 [3] NCCL INFO Channel 05/0 : 59[3] -> 60[4] via P2P/CUMEM +n116-042-235:5920:6035 [1] NCCL INFO Channel 15/0 : 57[1] -> 58[2] via P2P/CUMEM +n116-042-235:5922:6038 [3] NCCL INFO Channel 06/0 : 59[3] -> 60[4] via P2P/CUMEM +n116-042-235:5922:6038 [3] NCCL INFO Channel 07/0 : 59[3] -> 60[4] via P2P/CUMEM +n116-042-235:5919:6032 [0] NCCL INFO Channel 01/0 : 56[0] -> 63[7] via P2P/CUMEM +n116-042-235:5922:6038 [3] NCCL INFO Channel 08/0 : 59[3] -> 60[4] via P2P/CUMEM +n116-042-235:5919:6032 [0] NCCL INFO Channel 02/0 : 56[0] -> 63[7] via P2P/CUMEM +n116-042-235:5922:6038 [3] NCCL INFO Channel 09/0 : 59[3] -> 60[4] via P2P/CUMEM +n116-042-235:5919:6032 [0] NCCL INFO Channel 03/0 : 56[0] -> 63[7] via P2P/CUMEM +n116-042-235:5922:6038 [3] NCCL INFO Channel 10/0 : 59[3] -> 60[4] via P2P/CUMEM +n116-042-235:5919:6032 [0] NCCL INFO Channel 04/0 : 56[0] -> 63[7] via P2P/CUMEM +n116-042-235:5922:6038 [3] NCCL INFO Channel 13/0 : 59[3] -> 60[4] via P2P/CUMEM +n116-042-235:5919:6032 [0] NCCL INFO Channel 05/0 : 56[0] -> 63[7] via P2P/CUMEM +n116-042-235:5922:6038 [3] NCCL INFO Channel 14/0 : 59[3] -> 60[4] via P2P/CUMEM +n116-042-235:5919:6032 [0] NCCL INFO Channel 06/0 : 56[0] -> 63[7] via P2P/CUMEM +n116-042-235:5922:6038 [3] NCCL INFO Channel 15/0 : 59[3] -> 60[4] via P2P/CUMEM +n116-042-235:5919:6032 [0] NCCL INFO Channel 09/0 : 56[0] -> 63[7] via P2P/CUMEM +n116-042-235:5919:6032 [0] NCCL INFO Channel 10/0 : 56[0] -> 63[7] via P2P/CUMEM +n116-042-235:5919:6032 [0] NCCL INFO Channel 11/0 : 56[0] -> 63[7] via P2P/CUMEM +n116-042-235:5919:6032 [0] NCCL INFO Channel 12/0 : 56[0] -> 63[7] via P2P/CUMEM +n116-042-235:5919:6032 [0] NCCL INFO Channel 13/0 : 56[0] -> 63[7] via P2P/CUMEM +n116-042-235:5919:6032 [0] NCCL INFO Channel 14/0 : 56[0] -> 63[7] via P2P/CUMEM +n116-042-235:5924:6039 [5] NCCL INFO [Proxy Progress] Device 5 CPU core 181 +n116-042-235:5925:6040 [6] NCCL INFO [Proxy Progress] Device 6 CPU core 96 +n116-042-235:5925:6036 [6] NCCL INFO Channel 06/0 : 54[6] -> 62[6] [receive] via NET/FasTrak/6/GDRDMA +n116-042-235:5924:6034 [5] NCCL INFO Channel 05/0 : 53[5] -> 61[5] [receive] via NET/FasTrak/5/GDRDMA +n116-042-235:5920:6041 [1] NCCL INFO [Proxy Progress] Device 1 CPU core 21 +n116-042-235:5923:6042 [4] NCCL INFO [Proxy Progress] Device 4 CPU core 102 +n116-042-235:5920:6035 [1] NCCL INFO Channel 01/0 : 49[1] -> 57[1] [receive] via NET/FasTrak/1/GDRDMA +n116-042-235:5921:6043 [2] NCCL INFO [Proxy Progress] Device 2 CPU core 32 +n116-042-235:5922:6044 [3] NCCL INFO [Proxy Progress] Device 3 CPU core 140 +n116-042-235:5923:6037 [4] NCCL INFO Channel 04/0 : 52[4] -> 60[4] [receive] via NET/FasTrak/4/GDRDMA +n116-042-235:5921:6033 [2] NCCL INFO Channel 02/0 : 50[2] -> 58[2] [receive] via NET/FasTrak/2/GDRDMA +n116-042-235:5919:6045 [0] NCCL INFO [Proxy Progress] Device 0 CPU core 155 +n116-042-235:5926:6046 [7] NCCL INFO [Proxy Progress] Device 7 CPU core 170 +n116-042-235:5919:6032 [0] NCCL INFO Channel 00/0 : 48[0] -> 56[0] [receive] via NET/FasTrak/0/GDRDMA +n116-042-235:5922:6038 [3] NCCL INFO Channel 03/0 : 51[3] -> 59[3] [receive] via NET/FasTrak/3/GDRDMA +n116-042-235:5926:6031 [7] NCCL INFO Channel 07/0 : 55[7] -> 63[7] [receive] via NET/FasTrak/7/GDRDMA +n116-042-235:5924:6034 [5] NCCL INFO Channel 13/0 : 29[5] -> 61[5] [receive] via NET/FasTrak/5/GDRDMA +n116-042-235:5924:6034 [5] NCCL INFO Channel 13/0 : 61[5] -> 29[5] [send] via NET/FasTrak/5/GDRDMA +n116-042-235:5919:6032 [0] NCCL INFO Channel 08/0 : 24[0] -> 56[0] [receive] via NET/FasTrak/0/GDRDMA +n116-042-235:5919:6032 [0] NCCL INFO Channel 08/0 : 56[0] -> 24[0] [send] via NET/FasTrak/0/GDRDMA +n116-042-235:5925:6036 [6] NCCL INFO Channel 14/0 : 30[6] -> 62[6] [receive] via NET/FasTrak/6/GDRDMA +n116-042-235:5923:6037 [4] NCCL INFO Channel 12/0 : 28[4] -> 60[4] [receive] via NET/FasTrak/4/GDRDMA +n116-042-235:5923:6037 [4] NCCL INFO Channel 12/0 : 60[4] -> 28[4] [send] via NET/FasTrak/4/GDRDMA +n116-042-235:5925:6036 [6] NCCL INFO Channel 14/0 : 62[6] -> 30[6] [send] via NET/FasTrak/6/GDRDMA +n116-042-235:5922:6038 [3] NCCL INFO Channel 11/0 : 27[3] -> 59[3] [receive] via NET/FasTrak/3/GDRDMA +n116-042-235:5922:6038 [3] NCCL INFO Channel 11/0 : 59[3] -> 27[3] [send] via NET/FasTrak/3/GDRDMA +n116-042-235:5919:6032 [0] NCCL INFO Channel 00/0 : 56[0] -> 48[0] [send] via NET/FasTrak/0/GDRDMA +n116-042-235:5925:6036 [6] NCCL INFO Channel 06/0 : 62[6] -> 54[6] [send] via NET/FasTrak/6/GDRDMA +n116-042-235:5923:6037 [4] NCCL INFO Channel 04/0 : 60[4] -> 52[4] [send] via NET/FasTrak/4/GDRDMA +n116-042-235:5920:6035 [1] NCCL INFO Channel 09/0 : 25[1] -> 57[1] [receive] via NET/FasTrak/1/GDRDMA +n116-042-235:5920:6035 [1] NCCL INFO Channel 09/0 : 57[1] -> 25[1] [send] via NET/FasTrak/1/GDRDMA +n116-042-235:5921:6033 [2] NCCL INFO Channel 10/0 : 26[2] -> 58[2] [receive] via NET/FasTrak/2/GDRDMA +n116-042-235:5921:6033 [2] NCCL INFO Channel 10/0 : 58[2] -> 26[2] [send] via NET/FasTrak/2/GDRDMA +n116-042-235:5924:6034 [5] NCCL INFO Channel 05/0 : 61[5] -> 53[5] [send] via NET/FasTrak/5/GDRDMA +n116-042-235:5921:6033 [2] NCCL INFO Channel 02/0 : 58[2] -> 50[2] [send] via NET/FasTrak/2/GDRDMA +n116-042-235:5920:6035 [1] NCCL INFO Channel 01/0 : 57[1] -> 49[1] [send] via NET/FasTrak/1/GDRDMA +n116-042-235:5922:6038 [3] NCCL INFO Channel 03/0 : 59[3] -> 51[3] [send] via NET/FasTrak/3/GDRDMA +n116-042-235:5926:6031 [7] NCCL INFO Channel 15/0 : 31[7] -> 63[7] [receive] via NET/FasTrak/7/GDRDMA +n116-042-235:5926:6031 [7] NCCL INFO Channel 15/0 : 63[7] -> 31[7] [send] via NET/FasTrak/7/GDRDMA +n116-042-235:5926:6031 [7] NCCL INFO Channel 07/0 : 63[7] -> 55[7] [send] via NET/FasTrak/7/GDRDMA +n116-042-235:5926:6031 [7] NCCL INFO Channel 01/0 : 63[7] -> 56[0] via P2P/CUMEM +n116-042-235:5924:6034 [5] NCCL INFO Channel 00/0 : 61[5] -> 60[4] via P2P/CUMEM +n116-042-235:5924:6034 [5] NCCL INFO Channel 01/0 : 61[5] -> 60[4] via P2P/CUMEM +n116-042-235:5925:6036 [6] NCCL INFO Channel 00/0 : 62[6] -> 61[5] via P2P/CUMEM +n116-042-235:5921:6033 [2] NCCL INFO Channel 00/0 : 58[2] -> 57[1] via P2P/CUMEM +n116-042-235:5924:6034 [5] NCCL INFO Channel 02/0 : 61[5] -> 60[4] via P2P/CUMEM +n116-042-235:5925:6036 [6] NCCL INFO Channel 01/0 : 62[6] -> 61[5] via P2P/CUMEM +n116-042-235:5921:6033 [2] NCCL INFO Channel 03/0 : 58[2] -> 57[1] via P2P/CUMEM +n116-042-235:5925:6036 [6] NCCL INFO Channel 02/0 : 62[6] -> 61[5] via P2P/CUMEM +n116-042-235:5924:6034 [5] NCCL INFO Channel 03/0 : 61[5] -> 60[4] via P2P/CUMEM +n116-042-235:5921:6033 [2] NCCL INFO Channel 04/0 : 58[2] -> 57[1] via P2P/CUMEM +n116-042-235:5925:6036 [6] NCCL INFO Channel 03/0 : 62[6] -> 61[5] via P2P/CUMEM +n116-042-235:5921:6033 [2] NCCL INFO Channel 05/0 : 58[2] -> 57[1] via P2P/CUMEM +n116-042-235:5925:6036 [6] NCCL INFO Channel 04/0 : 62[6] -> 61[5] via P2P/CUMEM +n116-042-235:5923:6037 [4] NCCL INFO Channel 00/0 : 60[4] -> 59[3] via P2P/CUMEM +n116-042-235:5921:6033 [2] NCCL INFO Channel 06/0 : 58[2] -> 57[1] via P2P/CUMEM +n116-042-235:5925:6036 [6] NCCL INFO Channel 07/0 : 62[6] -> 61[5] via P2P/CUMEM +n116-042-235:5922:6038 [3] NCCL INFO Channel 00/0 : 59[3] -> 58[2] via P2P/CUMEM +n116-042-235:5923:6037 [4] NCCL INFO Channel 01/0 : 60[4] -> 59[3] via P2P/CUMEM +n116-042-235:5921:6033 [2] NCCL INFO Channel 07/0 : 58[2] -> 57[1] via P2P/CUMEM +n116-042-235:5925:6036 [6] NCCL INFO Channel 08/0 : 62[6] -> 61[5] via P2P/CUMEM +n116-042-235:5926:6031 [7] NCCL INFO Channel 02/0 : 63[7] -> 56[0] via P2P/CUMEM +n116-042-235:5922:6038 [3] NCCL INFO Channel 01/0 : 59[3] -> 58[2] via P2P/CUMEM +n116-042-235:5924:6034 [5] NCCL INFO Channel 04/0 : 61[5] -> 60[4] via P2P/CUMEM +n116-042-235:5923:6037 [4] NCCL INFO Channel 02/0 : 60[4] -> 59[3] via P2P/CUMEM +n116-042-235:5921:6033 [2] NCCL INFO Channel 08/0 : 58[2] -> 57[1] via P2P/CUMEM +n116-042-235:5925:6036 [6] NCCL INFO Channel 09/0 : 62[6] -> 61[5] via P2P/CUMEM +n116-042-235:5926:6031 [7] NCCL INFO Channel 03/0 : 63[7] -> 56[0] via P2P/CUMEM +n116-042-235:5922:6038 [3] NCCL INFO Channel 02/0 : 59[3] -> 58[2] via P2P/CUMEM +n116-042-235:5924:6034 [5] NCCL INFO Channel 05/0 : 61[5] -> 60[4] via P2P/CUMEM +n116-042-235:5923:6037 [4] NCCL INFO Channel 05/0 : 60[4] -> 59[3] via P2P/CUMEM +n116-042-235:5921:6033 [2] NCCL INFO Channel 11/0 : 58[2] -> 57[1] via P2P/CUMEM +n116-042-235:5925:6036 [6] NCCL INFO Channel 10/0 : 62[6] -> 61[5] via P2P/CUMEM +n116-042-235:5926:6031 [7] NCCL INFO Channel 04/0 : 63[7] -> 56[0] via P2P/CUMEM +n116-042-235:5922:6038 [3] NCCL INFO Channel 03/0 : 59[3] -> 58[2] via P2P/CUMEM +n116-042-235:5924:6034 [5] NCCL INFO Channel 06/0 : 61[5] -> 60[4] via P2P/CUMEM +n116-042-235:5921:6033 [2] NCCL INFO Channel 12/0 : 58[2] -> 57[1] via P2P/CUMEM +n116-042-235:5923:6037 [4] NCCL INFO Channel 06/0 : 60[4] -> 59[3] via P2P/CUMEM +n116-042-235:5926:6031 [7] NCCL INFO Channel 05/0 : 63[7] -> 56[0] via P2P/CUMEM +n116-042-235:5925:6036 [6] NCCL INFO Channel 11/0 : 62[6] -> 61[5] via P2P/CUMEM +n116-042-235:5921:6033 [2] NCCL INFO Channel 13/0 : 58[2] -> 57[1] via P2P/CUMEM +n116-042-235:5923:6037 [4] NCCL INFO Channel 07/0 : 60[4] -> 59[3] via P2P/CUMEM +n116-042-235:5926:6031 [7] NCCL INFO Channel 06/0 : 63[7] -> 56[0] via P2P/CUMEM +n116-042-235:5924:6034 [5] NCCL INFO Channel 07/0 : 61[5] -> 60[4] via P2P/CUMEM +n116-042-235:5925:6036 [6] NCCL INFO Channel 12/0 : 62[6] -> 61[5] via P2P/CUMEM +n116-042-235:5922:6038 [3] NCCL INFO Channel 04/0 : 59[3] -> 58[2] via P2P/CUMEM +n116-042-235:5923:6037 [4] NCCL INFO Channel 08/0 : 60[4] -> 59[3] via P2P/CUMEM +n116-042-235:5924:6034 [5] NCCL INFO Channel 08/0 : 61[5] -> 60[4] via P2P/CUMEM +n116-042-235:5925:6036 [6] NCCL INFO Channel 15/0 : 62[6] -> 61[5] via P2P/CUMEM +n116-042-235:5922:6038 [3] NCCL INFO Channel 05/0 : 59[3] -> 58[2] via P2P/CUMEM +n116-042-235:5923:6037 [4] NCCL INFO Channel 09/0 : 60[4] -> 59[3] via P2P/CUMEM +n116-042-235:5920:6035 [1] NCCL INFO Channel 00/0 : 57[1] -> 56[0] via P2P/CUMEM +n116-042-235:5924:6034 [5] NCCL INFO Channel 09/0 : 61[5] -> 60[4] via P2P/CUMEM +n116-042-235:5922:6038 [3] NCCL INFO Channel 06/0 : 59[3] -> 58[2] via P2P/CUMEM +n116-042-235:5923:6037 [4] NCCL INFO Channel 10/0 : 60[4] -> 59[3] via P2P/CUMEM +n116-042-235:5920:6035 [1] NCCL INFO Channel 01/0 : 57[1] -> 56[0] via P2P/CUMEM +n116-042-235:5924:6034 [5] NCCL INFO Channel 10/0 : 61[5] -> 60[4] via P2P/CUMEM +n116-042-235:5922:6038 [3] NCCL INFO Channel 07/0 : 59[3] -> 58[2] via P2P/CUMEM +n116-042-235:5921:6033 [2] NCCL INFO Channel 14/0 : 58[2] -> 57[1] via P2P/CUMEM +n116-042-235:5923:6037 [4] NCCL INFO Channel 13/0 : 60[4] -> 59[3] via P2P/CUMEM +n116-042-235:5920:6035 [1] NCCL INFO Channel 02/0 : 57[1] -> 56[0] via P2P/CUMEM +n116-042-235:5924:6034 [5] NCCL INFO Channel 11/0 : 61[5] -> 60[4] via P2P/CUMEM +n116-042-235:5922:6038 [3] NCCL INFO Channel 08/0 : 59[3] -> 58[2] via P2P/CUMEM +n116-042-235:5921:6033 [2] NCCL INFO Channel 15/0 : 58[2] -> 57[1] via P2P/CUMEM +n116-042-235:5923:6037 [4] NCCL INFO Channel 14/0 : 60[4] -> 59[3] via P2P/CUMEM +n116-042-235:5920:6035 [1] NCCL INFO Channel 03/0 : 57[1] -> 56[0] via P2P/CUMEM +n116-042-235:5924:6034 [5] NCCL INFO Channel 12/0 : 61[5] -> 60[4] via P2P/CUMEM +n116-042-235:5922:6038 [3] NCCL INFO Channel 09/0 : 59[3] -> 58[2] via P2P/CUMEM +n116-042-235:5923:6037 [4] NCCL INFO Channel 15/0 : 60[4] -> 59[3] via P2P/CUMEM +n116-042-235:5920:6035 [1] NCCL INFO Channel 04/0 : 57[1] -> 56[0] via P2P/CUMEM +n116-042-235:5924:6034 [5] NCCL INFO Channel 13/0 : 61[5] -> 60[4] via P2P/CUMEM +n116-042-235:5922:6038 [3] NCCL INFO Channel 10/0 : 59[3] -> 58[2] via P2P/CUMEM +n116-042-235:5920:6035 [1] NCCL INFO Channel 05/0 : 57[1] -> 56[0] via P2P/CUMEM +n116-042-235:5926:6031 [7] NCCL INFO Channel 09/0 : 63[7] -> 56[0] via P2P/CUMEM +n116-042-235:5924:6034 [5] NCCL INFO Channel 14/0 : 61[5] -> 60[4] via P2P/CUMEM +n116-042-235:5922:6038 [3] NCCL INFO Channel 11/0 : 59[3] -> 58[2] via P2P/CUMEM +n116-042-235:5920:6035 [1] NCCL INFO Channel 06/0 : 57[1] -> 56[0] via P2P/CUMEM +n116-042-235:5926:6031 [7] NCCL INFO Channel 10/0 : 63[7] -> 56[0] via P2P/CUMEM +n116-042-235:5924:6034 [5] NCCL INFO Channel 15/0 : 61[5] -> 60[4] via P2P/CUMEM +n116-042-235:5920:6035 [1] NCCL INFO Channel 07/0 : 57[1] -> 56[0] via P2P/CUMEM +n116-042-235:5922:6038 [3] NCCL INFO Channel 12/0 : 59[3] -> 58[2] via P2P/CUMEM +n116-042-235:5926:6031 [7] NCCL INFO Channel 11/0 : 63[7] -> 56[0] via P2P/CUMEM +n116-042-235:5922:6038 [3] NCCL INFO Channel 13/0 : 59[3] -> 58[2] via P2P/CUMEM +n116-042-235:5920:6035 [1] NCCL INFO Channel 08/0 : 57[1] -> 56[0] via P2P/CUMEM +n116-042-235:5922:6038 [3] NCCL INFO Channel 14/0 : 59[3] -> 58[2] via P2P/CUMEM +n116-042-235:5926:6031 [7] NCCL INFO Channel 12/0 : 63[7] -> 56[0] via P2P/CUMEM +n116-042-235:5922:6038 [3] NCCL INFO Channel 15/0 : 59[3] -> 58[2] via P2P/CUMEM +n116-042-235:5926:6031 [7] NCCL INFO Channel 13/0 : 63[7] -> 56[0] via P2P/CUMEM +n116-042-235:5920:6035 [1] NCCL INFO Channel 09/0 : 57[1] -> 56[0] via P2P/CUMEM +n116-042-235:5926:6031 [7] NCCL INFO Channel 14/0 : 63[7] -> 56[0] via P2P/CUMEM +n116-042-235:5920:6035 [1] NCCL INFO Channel 10/0 : 57[1] -> 56[0] via P2P/CUMEM +n116-042-235:5926:6031 [7] NCCL INFO Channel 00/0 : 63[7] -> 62[6] via P2P/CUMEM +n116-042-235:5920:6035 [1] NCCL INFO Channel 11/0 : 57[1] -> 56[0] via P2P/CUMEM +n116-042-235:5926:6031 [7] NCCL INFO Channel 01/0 : 63[7] -> 62[6] via P2P/CUMEM +n116-042-235:5920:6035 [1] NCCL INFO Channel 12/0 : 57[1] -> 56[0] via P2P/CUMEM +n116-042-235:5920:6035 [1] NCCL INFO Channel 13/0 : 57[1] -> 56[0] via P2P/CUMEM +n116-042-235:5926:6031 [7] NCCL INFO Channel 02/0 : 63[7] -> 62[6] via P2P/CUMEM +n116-042-235:5920:6035 [1] NCCL INFO Channel 14/0 : 57[1] -> 56[0] via P2P/CUMEM +n116-042-235:5920:6035 [1] NCCL INFO Channel 15/0 : 57[1] -> 56[0] via P2P/CUMEM +n116-042-235:5926:6031 [7] NCCL INFO Channel 03/0 : 63[7] -> 62[6] via P2P/CUMEM +n116-042-235:5926:6031 [7] NCCL INFO Channel 04/0 : 63[7] -> 62[6] via P2P/CUMEM +n116-042-235:5926:6031 [7] NCCL INFO Channel 05/0 : 63[7] -> 62[6] via P2P/CUMEM +n116-042-235:5926:6031 [7] NCCL INFO Channel 06/0 : 63[7] -> 62[6] via P2P/CUMEM +n116-042-235:5926:6031 [7] NCCL INFO Channel 07/0 : 63[7] -> 62[6] via P2P/CUMEM +n116-042-235:5926:6031 [7] NCCL INFO Channel 08/0 : 63[7] -> 62[6] via P2P/CUMEM +n116-042-235:5926:6031 [7] NCCL INFO Channel 09/0 : 63[7] -> 62[6] via P2P/CUMEM +n116-042-235:5926:6031 [7] NCCL INFO Channel 10/0 : 63[7] -> 62[6] via P2P/CUMEM +n116-042-235:5926:6031 [7] NCCL INFO Channel 11/0 : 63[7] -> 62[6] via P2P/CUMEM +n116-042-235:5926:6031 [7] NCCL INFO Channel 12/0 : 63[7] -> 62[6] via P2P/CUMEM +n116-042-235:5926:6031 [7] NCCL INFO Channel 13/0 : 63[7] -> 62[6] via P2P/CUMEM +n116-042-235:5926:6031 [7] NCCL INFO Channel 14/0 : 63[7] -> 62[6] via P2P/CUMEM +n116-042-235:5926:6031 [7] NCCL INFO Channel 15/0 : 63[7] -> 62[6] via P2P/CUMEM +n116-042-235:5921:6033 [2] NCCL INFO Connected all trees +n116-042-235:5922:6038 [3] NCCL INFO Connected all trees +n116-042-235:5926:6031 [7] NCCL INFO Connected all trees +n116-042-235:5919:6032 [0] NCCL INFO Connected all trees +n116-042-235:5920:6035 [1] NCCL INFO Connected all trees +n116-042-235:5923:6037 [4] NCCL INFO Connected all trees +n116-042-235:5924:6034 [5] NCCL INFO Connected all trees +n116-042-235:5925:6036 [6] NCCL INFO Connected all trees +n116-042-235:5919:6047 [0] NCCL INFO Channel 00/0 : 49[1] -> 56[0] [receive] via NET/FasTrak/0/GDRDMA +n116-042-235:5919:6047 [0] NCCL INFO Channel 08/0 : 49[1] -> 56[0] [receive] via NET/FasTrak/0/GDRDMA +n116-042-235:5919:6047 [0] NCCL INFO Channel 00/0 : 56[0] -> 63[7] via P2P/CUMEM +n116-042-235:5919:6047 [0] NCCL INFO Channel 08/0 : 56[0] -> 63[7] via P2P/CUMEM +n116-042-235:5926:6049 [7] NCCL INFO Channel 06/0 : 63[7] -> 6[6] [send] via NET/FasTrak/6/GDRDMA +n116-042-235:5926:6049 [7] NCCL INFO Channel 14/0 : 63[7] -> 6[6] [send] via NET/FasTrak/6/GDRDMA +n116-042-235:5924:6052 [5] NCCL INFO Channel 05/0 : 61[5] -> 62[6] via P2P/CUMEM +n116-042-235:5924:6052 [5] NCCL INFO Channel 13/0 : 61[5] -> 62[6] via P2P/CUMEM +n116-042-235:5924:6052 [5] NCCL INFO Channel 04/0 : 61[5] -> 4[4] [send] via NET/FasTrak/4/GDRDMA +n116-042-235:5924:6052 [5] NCCL INFO Channel 12/0 : 61[5] -> 4[4] [send] via NET/FasTrak/4/GDRDMA +n116-042-235:5925:6048 [6] NCCL INFO Channel 06/0 : 55[7] -> 62[6] [receive] via NET/FasTrak/6/GDRDMA +n116-042-235:5925:6048 [6] NCCL INFO Channel 14/0 : 55[7] -> 62[6] [receive] via NET/FasTrak/6/GDRDMA +n116-042-235:5922:6053 [3] NCCL INFO Channel 03/0 : 59[3] -> 60[4] via P2P/CUMEM +n116-042-235:5922:6053 [3] NCCL INFO Channel 11/0 : 59[3] -> 60[4] via P2P/CUMEM +n116-042-235:5922:6053 [3] NCCL INFO Channel 02/0 : 59[3] -> 2[2] [send] via NET/FasTrak/2/GDRDMA +n116-042-235:5922:6053 [3] NCCL INFO Channel 10/0 : 59[3] -> 2[2] [send] via NET/FasTrak/2/GDRDMA +n116-042-235:5923:6051 [4] NCCL INFO Channel 04/0 : 53[5] -> 60[4] [receive] via NET/FasTrak/4/GDRDMA +n116-042-235:5923:6051 [4] NCCL INFO Channel 12/0 : 53[5] -> 60[4] [receive] via NET/FasTrak/4/GDRDMA +n116-042-235:5920:6054 [1] NCCL INFO Channel 01/0 : 57[1] -> 58[2] via P2P/CUMEM +n116-042-235:5920:6054 [1] NCCL INFO Channel 09/0 : 57[1] -> 58[2] via P2P/CUMEM +n116-042-235:5920:6054 [1] NCCL INFO Channel 00/0 : 57[1] -> 0[0] [send] via NET/FasTrak/0/GDRDMA +n116-042-235:5920:6054 [1] NCCL INFO Channel 08/0 : 57[1] -> 0[0] [send] via NET/FasTrak/0/GDRDMA +n116-042-235:5921:6050 [2] NCCL INFO Channel 02/0 : 51[3] -> 58[2] [receive] via NET/FasTrak/2/GDRDMA +n116-042-235:5921:6050 [2] NCCL INFO Channel 10/0 : 51[3] -> 58[2] [receive] via NET/FasTrak/2/GDRDMA +n116-042-235:5921:6050 [2] NCCL INFO Channel 03/0 : 58[2] -> 3[3] [send] via NET/FasTrak/3/GDRDMA +n116-042-235:5921:6050 [2] NCCL INFO Channel 11/0 : 58[2] -> 3[3] [send] via NET/FasTrak/3/GDRDMA +n116-042-235:5925:6048 [6] NCCL INFO Channel 07/0 : 62[6] -> 7[7] [send] via NET/FasTrak/7/GDRDMA +n116-042-235:5925:6048 [6] NCCL INFO Channel 15/0 : 62[6] -> 7[7] [send] via NET/FasTrak/7/GDRDMA +n116-042-235:5923:6051 [4] NCCL INFO Channel 05/0 : 60[4] -> 5[5] [send] via NET/FasTrak/5/GDRDMA +n116-042-235:5923:6051 [4] NCCL INFO Channel 13/0 : 60[4] -> 5[5] [send] via NET/FasTrak/5/GDRDMA +n116-042-235:5919:6047 [0] NCCL INFO Channel 01/0 : 56[0] -> 1[1] [send] via NET/FasTrak/1/GDRDMA +n116-042-235:5919:6047 [0] NCCL INFO Channel 09/0 : 56[0] -> 1[1] [send] via NET/FasTrak/1/GDRDMA +n116-042-235:5920:6054 [1] NCCL INFO Channel 01/0 : 48[0] -> 57[1] [receive] via NET/FasTrak/1/GDRDMA +n116-042-235:5920:6054 [1] NCCL INFO Channel 09/0 : 48[0] -> 57[1] [receive] via NET/FasTrak/1/GDRDMA +n116-042-235:5925:6048 [6] NCCL INFO Channel 06/0 : 62[6] -> 61[5] via P2P/CUMEM +n116-042-235:5925:6048 [6] NCCL INFO Channel 14/0 : 62[6] -> 61[5] via P2P/CUMEM +n116-042-235:5923:6051 [4] NCCL INFO Channel 04/0 : 60[4] -> 59[3] via P2P/CUMEM +n116-042-235:5923:6051 [4] NCCL INFO Channel 12/0 : 60[4] -> 59[3] via P2P/CUMEM +n116-042-235:5926:6049 [7] NCCL INFO Channel 07/0 : 54[6] -> 63[7] [receive] via NET/FasTrak/7/GDRDMA +n116-042-235:5926:6049 [7] NCCL INFO Channel 15/0 : 54[6] -> 63[7] [receive] via NET/FasTrak/7/GDRDMA +n116-042-235:5926:6049 [7] NCCL INFO Channel 07/0 : 63[7] -> 56[0] via P2P/CUMEM +n116-042-235:5926:6049 [7] NCCL INFO Channel 15/0 : 63[7] -> 56[0] via P2P/CUMEM +n116-042-235:5921:6050 [2] NCCL INFO Channel 02/0 : 58[2] -> 57[1] via P2P/CUMEM +n116-042-235:5921:6050 [2] NCCL INFO Channel 10/0 : 58[2] -> 57[1] via P2P/CUMEM +n116-042-235:5924:6052 [5] NCCL INFO Channel 05/0 : 52[4] -> 61[5] [receive] via NET/FasTrak/5/GDRDMA +n116-042-235:5924:6052 [5] NCCL INFO Channel 13/0 : 52[4] -> 61[5] [receive] via NET/FasTrak/5/GDRDMA +n116-042-235:5922:6053 [3] NCCL INFO Channel 03/0 : 50[2] -> 59[3] [receive] via NET/FasTrak/3/GDRDMA +n116-042-235:5922:6053 [3] NCCL INFO Channel 11/0 : 50[2] -> 59[3] [receive] via NET/FasTrak/3/GDRDMA +n116-042-235:5922:6053 [3] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-042-235:5924:6052 [5] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-042-235:5923:6051 [4] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-042-235:5925:6048 [6] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-042-235:5926:6049 [7] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-042-235:5921:6050 [2] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-042-235:5920:6054 [1] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-042-235:5919:6047 [0] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +[rank57]:[W727 05:39:50.321440939 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n116-042-235.byted.org]:60750, remote=[n116-040-168.byted.org]:25264): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f221a5785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f2203ccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7f2203ccff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7f2203cd084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7f2203cca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f21c53c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f221dda74a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f222097e1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f22209fe89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank57]:[W727 05:39:50.334118270 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 57] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank59]:[W727 05:39:50.445739995 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n116-042-235.byted.org]:60780, remote=[n116-040-168.byted.org]:25264): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f6b6cf785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f6b566cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7f6b566cff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7f6b566d084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7f6b566ca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f6b17dc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f6b707514a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f6b733281f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f6b733a889c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank59]:[W727 05:39:50.452420633 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 59] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank56]:[W727 05:39:50.749150399 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=71, addr=[n116-042-235.byted.org]:53198, remote=[n116-040-168.byted.org]:25264): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f6adf5785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f6ac8ccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7f6ac8ccff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7f6ac8cd084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7f6ac8cca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f6a8a3c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f6ae2d284a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f6ae58ff1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f6ae597f89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank56]:[W727 05:39:50.755656396 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 56] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank62]:[W727 05:39:50.805449175 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n116-042-235.byted.org]:60764, remote=[n116-040-168.byted.org]:25264): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f31aed785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f31984cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7f31984cff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7f31984d084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7f31984ca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f3159bc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f31b24dd4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f31b50b41f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f31b513489c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank62]:[W727 05:39:50.812890648 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 62] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank58]:[W727 05:39:50.807546849 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n116-042-235.byted.org]:60762, remote=[n116-040-168.byted.org]:25264): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f326b9785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f32550cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7f32550cff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7f32550d084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7f32550ca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f32167c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f326f18b4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f3271d621f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f3271de289c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank58]:[W727 05:39:50.814337440 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 58] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank61]:[W727 05:39:51.877510125 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n116-042-235.byted.org]:60774, remote=[n116-040-168.byted.org]:25264): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f032b5785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f0314ccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7f0314ccff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7f0314cd084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7f0314cca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f02d63c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f032ecd24a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f03318a91f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f033192989c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank61]:[W727 05:39:51.883763338 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 61] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank63]:[W727 05:39:51.946808097 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n116-042-235.byted.org]:60734, remote=[n116-040-168.byted.org]:25264): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f243a3785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f2423acdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7f2423acff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7f2423ad084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7f2423aca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f23e51c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f243dadd4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f24406b41f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f244073489c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank63]:[W727 05:39:51.953844996 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 63] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank60]:[W727 05:39:51.113804000 TCPStore.cpp:125] [c10d] recvValue failed on SocketImpl(fd=91, addr=[n116-042-235.byted.org]:60736, remote=[n116-040-168.byted.org]:25264): failed to recv, got 0 bytes +Exception raised from recvBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:678 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f18071785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f17f04cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baaf40 (0x7f17f04cff40 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5bab84a (0x7f17f04d084a in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x2a9 (0x7f17f04ca2a9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f17b1bc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f180a5b34a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f180d18a1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f180d20a89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank60]:[W727 05:39:51.120123080 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 60] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: failed to recv, got 0 bytes +[rank57]:[W727 05:39:51.334283501 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-042-235.byted.org]:60750, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f221a5785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f2203ccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f2203ccf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f2203cd0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f2203cca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f21c53c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f221dda74a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f222097e1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f22209fe89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank57]:[W727 05:39:51.341316793 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 57] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank59]:[W727 05:39:51.452590853 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-042-235.byted.org]:60780, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f6b6cf785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f6b566cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f6b566cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f6b566d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f6b566ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f6b17dc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f6b707514a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f6b733281f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f6b733a889c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank59]:[W727 05:39:51.458787577 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 59] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank56]:[W727 05:39:51.755824105 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=71, addr=[n116-042-235.byted.org]:53198, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f6adf5785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f6ac8ccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f6ac8ccf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f6ac8cd0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f6ac8cca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f6a8a3c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f6ae2d284a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f6ae58ff1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f6ae597f89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank56]:[W727 05:39:51.763837751 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 56] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank62]:[W727 05:39:51.813095285 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-042-235.byted.org]:60764, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f31aed785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f31984cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f31984cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f31984d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f31984ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f3159bc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f31b24dd4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f31b50b41f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f31b513489c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank62]:[W727 05:39:51.820074995 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 62] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank58]:[W727 05:39:51.814498586 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-042-235.byted.org]:60762, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f326b9785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f32550cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f32550cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f32550d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f32550ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f32167c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f326f18b4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f3271d621f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f3271de289c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank58]:[W727 05:39:51.821185349 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 58] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank61]:[W727 05:39:52.883947975 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-042-235.byted.org]:60774, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f032b5785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f0314ccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f0314ccf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f0314cd0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f0314cca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f02d63c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f032ecd24a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f03318a91f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f033192989c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank61]:[W727 05:39:52.891820985 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 61] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank63]:[W727 05:39:52.954026903 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-042-235.byted.org]:60734, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f243a3785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f2423acdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f2423acf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f2423ad0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f2423aca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f23e51c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f243dadd4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f24406b41f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f244073489c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank63]:[W727 05:39:52.960391656 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 63] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank60]:[W727 05:39:52.120312503 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-042-235.byted.org]:60736, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f18071785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f17f04cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f17f04cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f17f04d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f17f04ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f17b1bc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f180a5b34a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f180d18a1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f180d20a89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank60]:[W727 05:39:52.126684018 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 60] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank57]:[W727 05:39:52.341488044 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-042-235.byted.org]:60750, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f221a5785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f2203ccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f2203ccf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f2203cd0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f2203cca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f21c53c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f221dda74a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f222097e1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f22209fe89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank57]:[W727 05:39:52.348732260 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 57] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank59]:[W727 05:39:52.458964665 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-042-235.byted.org]:60780, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f6b6cf785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f6b566cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f6b566cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f6b566d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f6b566ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f6b17dc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f6b707514a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f6b733281f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f6b733a889c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank59]:[W727 05:39:52.466013519 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 59] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank56]:[W727 05:39:52.764010970 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=71, addr=[n116-042-235.byted.org]:53198, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f6adf5785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f6ac8ccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f6ac8ccf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f6ac8cd0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f6ac8cca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f6a8a3c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f6ae2d284a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f6ae58ff1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f6ae597f89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank56]:[W727 05:39:52.770301203 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 56] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank62]:[W727 05:39:52.820267144 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-042-235.byted.org]:60764, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f31aed785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f31984cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f31984cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f31984d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f31984ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f3159bc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f31b24dd4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f31b50b41f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f31b513489c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank62]:[W727 05:39:52.828362961 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 62] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank58]:[W727 05:39:52.821331836 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-042-235.byted.org]:60762, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f326b9785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f32550cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f32550cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f32550d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f32550ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f32167c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f326f18b4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f3271d621f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f3271de289c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank58]:[W727 05:39:52.828592120 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 58] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank61]:[W727 05:39:53.892011066 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-042-235.byted.org]:60774, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f032b5785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f0314ccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f0314ccf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f0314cd0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f0314cca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f02d63c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f032ecd24a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f03318a91f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f033192989c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank61]:[W727 05:39:53.899354868 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 61] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank63]:[W727 05:39:53.960591631 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-042-235.byted.org]:60734, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f243a3785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f2423acdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f2423acf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f2423ad0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f2423aca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f23e51c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f243dadd4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f24406b41f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f244073489c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank63]:[W727 05:39:53.967477718 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 63] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank60]:[W727 05:39:53.126850634 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-042-235.byted.org]:60736, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f18071785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f17f04cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f17f04cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f17f04d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f17f04ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f17b1bc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f180a5b34a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f180d18a1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f180d20a89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank60]:[W727 05:39:53.133317859 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 60] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank57]:[W727 05:39:53.348898144 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-042-235.byted.org]:60750, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f221a5785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f2203ccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f2203ccf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f2203cd0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f2203cca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f21c53c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f221dda74a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f222097e1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f22209fe89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank57]:[W727 05:39:53.355519213 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 57] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank59]:[W727 05:39:53.466211759 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-042-235.byted.org]:60780, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f6b6cf785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f6b566cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f6b566cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f6b566d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f6b566ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f6b17dc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f6b707514a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f6b733281f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f6b733a889c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank59]:[W727 05:39:53.472738955 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 59] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank56]:[W727 05:39:53.770475414 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=71, addr=[n116-042-235.byted.org]:53198, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f6adf5785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f6ac8ccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f6ac8ccf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f6ac8cd0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f6ac8cca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f6a8a3c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f6ae2d284a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f6ae58ff1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f6ae597f89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank56]:[W727 05:39:53.778306199 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 56] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank58]:[W727 05:39:53.828753112 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-042-235.byted.org]:60762, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f326b9785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f32550cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f32550cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f32550d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f32550ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f32167c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f326f18b4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f3271d621f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f3271de289c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank58]:[W727 05:39:53.834860998 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 58] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank62]:[W727 05:39:53.828521346 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-042-235.byted.org]:60764, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f31aed785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f31984cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f31984cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f31984d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f31984ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f3159bc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f31b24dd4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f31b50b41f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f31b513489c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank62]:[W727 05:39:53.835407247 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 62] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank61]:[W727 05:39:54.899515038 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-042-235.byted.org]:60774, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f032b5785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f0314ccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f0314ccf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f0314cd0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f0314cca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f02d63c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f032ecd24a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f03318a91f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f033192989c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank61]:[W727 05:39:54.906381531 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 61] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank63]:[W727 05:39:54.967650403 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-042-235.byted.org]:60734, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f243a3785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f2423acdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f2423acf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f2423ad0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f2423aca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f23e51c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f243dadd4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f24406b41f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f244073489c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank63]:[W727 05:39:54.974647889 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 63] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank60]:[W727 05:39:54.133484326 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-042-235.byted.org]:60736, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f18071785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f17f04cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f17f04cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f17f04d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f17f04ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f17b1bc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f180a5b34a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f180d18a1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f180d20a89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank60]:[W727 05:39:54.140093240 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 60] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank57]:[W727 05:39:54.355693368 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-042-235.byted.org]:60750, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f221a5785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f2203ccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f2203ccf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f2203cd0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f2203cca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f21c53c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f221dda74a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f222097e1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f22209fe89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank57]:[W727 05:39:54.362118497 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 57] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank59]:[W727 05:39:54.473024824 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-042-235.byted.org]:60780, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f6b6cf785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f6b566cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f6b566cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f6b566d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f6b566ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f6b17dc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f6b707514a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f6b733281f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f6b733a889c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank59]:[W727 05:39:54.479658546 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 59] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank56]:[W727 05:39:54.778466945 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=71, addr=[n116-042-235.byted.org]:53198, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f6adf5785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f6ac8ccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f6ac8ccf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f6ac8cd0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f6ac8cca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f6a8a3c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f6ae2d284a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f6ae58ff1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f6ae597f89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank56]:[W727 05:39:54.784697762 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 56] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank58]:[W727 05:39:54.835034429 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-042-235.byted.org]:60762, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f326b9785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f32550cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f32550cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f32550d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f32550ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f32167c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f326f18b4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f3271d621f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f3271de289c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank58]:[W727 05:39:54.841078049 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 58] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank62]:[W727 05:39:54.835572369 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-042-235.byted.org]:60764, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f31aed785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f31984cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f31984cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f31984d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f31984ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f3159bc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f31b24dd4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f31b50b41f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f31b513489c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank62]:[W727 05:39:54.841581023 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 62] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank61]:[W727 05:39:55.906541991 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-042-235.byted.org]:60774, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f032b5785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f0314ccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f0314ccf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f0314cd0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f0314cca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f02d63c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f032ecd24a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f03318a91f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f033192989c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank61]:[W727 05:39:55.914258624 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 61] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank63]:[W727 05:39:55.974843495 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-042-235.byted.org]:60734, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f243a3785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f2423acdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f2423acf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f2423ad0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f2423aca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f23e51c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f243dadd4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f24406b41f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f244073489c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank63]:[W727 05:39:55.981469660 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 63] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank60]:[W727 05:39:55.140272653 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-042-235.byted.org]:60736, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f18071785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f17f04cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f17f04cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f17f04d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f17f04ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f17b1bc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f180a5b34a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f180d18a1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f180d20a89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank60]:[W727 05:39:55.147475999 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 60] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank57]:[W727 05:39:55.362297848 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-042-235.byted.org]:60750, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f221a5785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f2203ccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f2203ccf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f2203cd0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f2203cca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f21c53c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f221dda74a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f222097e1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f22209fe89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank57]:[W727 05:39:55.369108890 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 57] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank59]:[W727 05:39:55.479815356 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-042-235.byted.org]:60780, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f6b6cf785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f6b566cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f6b566cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f6b566d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f6b566ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f6b17dc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f6b707514a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f6b733281f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f6b733a889c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank59]:[W727 05:39:55.486759758 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 59] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank56]:[W727 05:39:55.784872191 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=71, addr=[n116-042-235.byted.org]:53198, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f6adf5785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f6ac8ccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f6ac8ccf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f6ac8cd0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f6ac8cca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f6a8a3c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f6ae2d284a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f6ae58ff1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f6ae597f89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank56]:[W727 05:39:55.792335301 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 56] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank58]:[W727 05:39:55.841233822 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-042-235.byted.org]:60762, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f326b9785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f32550cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f32550cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f32550d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f32550ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f32167c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f326f18b4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f3271d621f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f3271de289c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank58]:[W727 05:39:55.847464018 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 58] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank62]:[W727 05:39:55.841744354 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-042-235.byted.org]:60764, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f31aed785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f31984cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f31984cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f31984d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f31984ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f3159bc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f31b24dd4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f31b50b41f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f31b513489c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank62]:[W727 05:39:55.847940535 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 62] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank61]:[W727 05:39:56.914429183 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-042-235.byted.org]:60774, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f032b5785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f0314ccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f0314ccf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f0314cd0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f0314cca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f02d63c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f032ecd24a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f03318a91f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f033192989c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank61]:[W727 05:39:56.921264288 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 61] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank63]:[W727 05:39:56.981707483 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-042-235.byted.org]:60734, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f243a3785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f2423acdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f2423acf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f2423ad0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f2423aca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f23e51c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f243dadd4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f24406b41f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f244073489c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank63]:[W727 05:39:56.987901841 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 63] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank60]:[W727 05:39:56.147686975 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-042-235.byted.org]:60736, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f18071785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f17f04cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f17f04cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f17f04d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f17f04ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f17b1bc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f180a5b34a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f180d18a1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f180d20a89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank60]:[W727 05:39:56.154133274 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 60] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank57]:[W727 05:39:56.369286419 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-042-235.byted.org]:60750, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f221a5785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f2203ccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f2203ccf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f2203cd0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f2203cca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f21c53c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f221dda74a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f222097e1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f22209fe89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank57]:[W727 05:39:56.376619620 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 57] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank59]:[W727 05:39:56.486913800 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-042-235.byted.org]:60780, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f6b6cf785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f6b566cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f6b566cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f6b566d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f6b566ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f6b17dc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f6b707514a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f6b733281f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f6b733a889c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank59]:[W727 05:39:56.493754776 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 59] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank56]:[W727 05:39:56.792514312 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=71, addr=[n116-042-235.byted.org]:53198, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f6adf5785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f6ac8ccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f6ac8ccf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f6ac8cd0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f6ac8cca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f6a8a3c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f6ae2d284a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f6ae58ff1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f6ae597f89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank56]:[W727 05:39:56.799337271 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 56] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank58]:[W727 05:39:56.847622984 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-042-235.byted.org]:60762, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f326b9785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f32550cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f32550cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f32550d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f32550ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f32167c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f326f18b4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f3271d621f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f3271de289c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank58]:[W727 05:39:56.854572755 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 58] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank62]:[W727 05:39:56.848100075 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-042-235.byted.org]:60764, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f31aed785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f31984cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f31984cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f31984d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f31984ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f3159bc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f31b24dd4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f31b50b41f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f31b513489c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank62]:[W727 05:39:56.854922269 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 62] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank61]:[W727 05:39:57.921435764 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-042-235.byted.org]:60774, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f032b5785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f0314ccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f0314ccf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f0314cd0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f0314cca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f02d63c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f032ecd24a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f03318a91f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f033192989c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank61]:[W727 05:39:57.929609249 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 61] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank63]:[W727 05:39:57.988095058 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-042-235.byted.org]:60734, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f243a3785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f2423acdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f2423acf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f2423ad0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f2423aca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f23e51c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f243dadd4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f24406b41f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f244073489c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank63]:[W727 05:39:57.995036754 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 63] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank60]:[W727 05:39:57.154309902 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-042-235.byted.org]:60736, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f18071785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f17f04cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f17f04cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f17f04d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f17f04ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f17b1bc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f180a5b34a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f180d18a1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f180d20a89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank60]:[W727 05:39:57.161173720 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 60] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank57]:[W727 05:39:57.376776993 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-042-235.byted.org]:60750, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f221a5785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f2203ccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f2203ccf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f2203cd0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f2203cca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f21c53c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f221dda74a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f222097e1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f22209fe89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank57]:[W727 05:39:57.383819097 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 57] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank59]:[W727 05:39:57.493911268 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-042-235.byted.org]:60780, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f6b6cf785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f6b566cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f6b566cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f6b566d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f6b566ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f6b17dc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f6b707514a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f6b733281f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f6b733a889c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank59]:[W727 05:39:57.500943671 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 59] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank56]:[W727 05:39:57.799530600 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=71, addr=[n116-042-235.byted.org]:53198, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f6adf5785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f6ac8ccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f6ac8ccf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f6ac8cd0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f6ac8cca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f6a8a3c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f6ae2d284a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f6ae58ff1f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f6ae597f89c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank56]:[W727 05:39:57.805877730 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 56] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank62]:[W727 05:39:57.855086078 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-042-235.byted.org]:60764, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f31aed785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f31984cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f31984cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f31984d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f31984ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f3159bc99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f31b24dd4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f31b50b41f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f31b513489c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank62]:[W727 05:39:57.861366465 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 62] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank58]:[W727 05:39:57.854722893 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-042-235.byted.org]:60762, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f326b9785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f32550cdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f32550cf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f32550d0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f32550ca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f32167c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f326f18b4a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f3271d621f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f3271de289c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank58]:[W727 05:39:57.861550365 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 58] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +[rank61]:[W727 05:39:58.929792713 TCPStore.cpp:106] [c10d] sendBytes failed on SocketImpl(fd=91, addr=[n116-042-235.byted.org]:60774, remote=[n116-040-168.byted.org]:25264): Broken pipe +Exception raised from sendBytes at /pytorch/torch/csrc/distributed/c10d/Utils.hpp:653 (most recent call first): +frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string, std::allocator >) + 0x98 (0x7f032b5785e8 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libc10.so) +frame #1: + 0x5ba8bfe (0x7f0314ccdbfe in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #2: + 0x5baa458 (0x7f0314ccf458 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #3: + 0x5babc3e (0x7f0314cd0c3e in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #4: c10d::TCPStore::check(std::vector, std::allocator >, std::allocator, std::allocator > > > const&) + 0x298 (0x7f0314cca298 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cpu.so) +frame #5: c10d::ProcessGroupNCCL::heartbeatMonitor() + 0x379 (0x7f02d63c99f9 in /home/tiger/.local/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so) +frame #6: + 0xd44a3 (0x7f032ecd24a3 in /lib/x86_64-linux-gnu/libstdc++.so.6) +frame #7: + 0x891f5 (0x7f03318a91f5 in /lib/x86_64-linux-gnu/libc.so.6) +frame #8: + 0x10989c (0x7f033192989c in /lib/x86_64-linux-gnu/libc.so.6) + +[rank61]:[W727 05:39:58.936900446 ProcessGroupNCCL.cpp:1662] [PG ID 0 PG GUID 0(default_pg) Rank 61] Failed to check the "should dump" flag on TCPStore, (maybe TCPStore server has shut down too early), with error: Broken pipe +E0727 05:39:58.135000 5854 torch/distributed/elastic/multiprocessing/api.py:874] failed (exitcode: -15) local_rank: 0 (pid: 5919) of binary: /usr/bin/python3 +Traceback (most recent call last): + File "/usr/local/bin/torchrun", line 7, in + sys.exit(main()) + ^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 355, in wrapper + return f(*args, **kwargs) + ^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 892, in main + run(args) + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 883, in run + elastic_launch( + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 139, in __call__ + return launch_agent(self._config, self._entrypoint, list(args)) + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 270, in launch_agent + raise ChildFailedError( +torch.distributed.elastic.multiprocessing.errors.ChildFailedError: +============================================================ +main_cross_2direct_balance_realscale_perturb_v4_accum.py FAILED +------------------------------------------------------------ +Failures: +[1]: + time : 2026-07-27_05:39:58 + host : n116-042-235.byted.org + rank : 57 (local_rank: 1) + exitcode : -15 (pid: 5920) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 5920 +[2]: + time : 2026-07-27_05:39:58 + host : n116-042-235.byted.org + rank : 58 (local_rank: 2) + exitcode : -15 (pid: 5921) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 5921 +[3]: + time : 2026-07-27_05:39:58 + host : n116-042-235.byted.org + rank : 59 (local_rank: 3) + exitcode : -15 (pid: 5922) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 5922 +[4]: + time : 2026-07-27_05:39:58 + host : n116-042-235.byted.org + rank : 60 (local_rank: 4) + exitcode : -15 (pid: 5923) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 5923 +[5]: + time : 2026-07-27_05:39:58 + host : n116-042-235.byted.org + rank : 61 (local_rank: 5) + exitcode : -15 (pid: 5924) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 5924 +[6]: + time : 2026-07-27_05:39:58 + host : n116-042-235.byted.org + rank : 62 (local_rank: 6) + exitcode : -15 (pid: 5925) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 5925 +[7]: + time : 2026-07-27_05:39:58 + host : n116-042-235.byted.org + rank : 63 (local_rank: 7) + exitcode : -15 (pid: 5926) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 5926 +------------------------------------------------------------ +Root Cause (first observed failure): +[0]: + time : 2026-07-27_05:39:58 + host : n116-042-235.byted.org + rank : 56 (local_rank: 0) + exitcode : -15 (pid: 5919) + error_file: + traceback : Signal 15 (SIGTERM) received by PID 5919 +============================================================ diff --git a/ScaleDF_rebuttal/_host_logs/all_46real_91fake_11451_20260725_1215/eval_after_train.log b/ScaleDF_rebuttal/_host_logs/all_46real_91fake_11451_20260725_1215/eval_after_train.log new file mode 100644 index 0000000000000000000000000000000000000000..2b435477938837fac8fba20ac1a9545b0e9155c8 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/all_46real_91fake_11451_20260725_1215/eval_after_train.log @@ -0,0 +1,567 @@ +/home/tiger/.profile: line 29: /run/bin/env: No such file or directory +/home/tiger/.profile: line 31: /tmp/m9/bin/env: No such file or directory +[2026-07-25T12:51:56Z] waiting_for_checkpoint /mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_all_46real_91fake_imgseed20260725_n116-042-186_20260725_1215/best_checkpoint.pth +[2026-07-25T12:52:56Z] waiting_for_checkpoint /mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_all_46real_91fake_imgseed20260725_n116-042-186_20260725_1215/best_checkpoint.pth +[2026-07-25T12:53:56Z] waiting_for_checkpoint /mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_all_46real_91fake_imgseed20260725_n116-042-186_20260725_1215/best_checkpoint.pth +[2026-07-25T12:54:57Z] waiting_for_checkpoint /mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_all_46real_91fake_imgseed20260725_n116-042-186_20260725_1215/best_checkpoint.pth +[2026-07-25T12:55:57Z] waiting_for_checkpoint /mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_all_46real_91fake_imgseed20260725_n116-042-186_20260725_1215/best_checkpoint.pth +[2026-07-25T12:56:57Z] waiting_for_checkpoint /mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_all_46real_91fake_imgseed20260725_n116-042-186_20260725_1215/best_checkpoint.pth +[2026-07-25T12:57:57Z] waiting_for_checkpoint /mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_all_46real_91fake_imgseed20260725_n116-042-186_20260725_1215/best_checkpoint.pth +[2026-07-25T12:58:57Z] waiting_for_checkpoint /mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_all_46real_91fake_imgseed20260725_n116-042-186_20260725_1215/best_checkpoint.pth +[2026-07-25T12:59:57Z] waiting_for_checkpoint /mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_all_46real_91fake_imgseed20260725_n116-042-186_20260725_1215/best_checkpoint.pth +[2026-07-25T13:00:57Z] waiting_for_checkpoint /mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_all_46real_91fake_imgseed20260725_n116-042-186_20260725_1215/best_checkpoint.pth +[2026-07-25T13:01:57Z] waiting_for_checkpoint /mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_all_46real_91fake_imgseed20260725_n116-042-186_20260725_1215/best_checkpoint.pth +[2026-07-25T13:02:57Z] waiting_for_checkpoint /mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_all_46real_91fake_imgseed20260725_n116-042-186_20260725_1215/best_checkpoint.pth +[2026-07-25T13:03:57Z] waiting_for_checkpoint /mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_all_46real_91fake_imgseed20260725_n116-042-186_20260725_1215/best_checkpoint.pth +[2026-07-25T13:04:57Z] waiting_for_checkpoint /mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_all_46real_91fake_imgseed20260725_n116-042-186_20260725_1215/best_checkpoint.pth +[2026-07-25T13:05:57Z] waiting_for_checkpoint /mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_all_46real_91fake_imgseed20260725_n116-042-186_20260725_1215/best_checkpoint.pth +[2026-07-25T13:06:58Z] waiting_for_checkpoint /mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_all_46real_91fake_imgseed20260725_n116-042-186_20260725_1215/best_checkpoint.pth +[2026-07-25T13:07:58Z] waiting_for_checkpoint /mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_all_46real_91fake_imgseed20260725_n116-042-186_20260725_1215/best_checkpoint.pth +[2026-07-25T13:08:58Z] waiting_for_checkpoint /mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_all_46real_91fake_imgseed20260725_n116-042-186_20260725_1215/best_checkpoint.pth +[2026-07-25T13:09:58Z] starting_eval +started_utc=2026-07-25T13:09:58Z +hostname=n116-042-186 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_all_46real_91fake_imgseed20260725_n116-042-186_20260725_1215/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_all_46real_91fake_benchmark_eval_20260725_1215/fixed500k_all_46real_91fake_seed0 +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0725 13:10:15.897000 34871 torch/distributed/run.py:766] +W0725 13:10:15.897000 34871 torch/distributed/run.py:766] ***************************************** +W0725 13:10:15.897000 34871 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 13:10:15.897000 34871 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 13:11:04.130088003 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 13:11:04.723863657 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 13:11:05.349081176 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 13:11:06.070515536 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 13:11:06.310425959 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 13:11:06.435375124 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W725 13:11:06.435381716 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 13:11:06.554471010 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_all_46real_91fake_benchmark_eval_20260725_1215/fixed500k_all_46real_91fake_seed0/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_all_46real_91fake_imgseed20260725_n116-042-186_20260725_1215/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 1:05:06 acc1_pos: 39.583 acc5_pos: 100.000 loss: 0.8075 (0.8075) acc1: 39.5833 (39.5833) acc5: 100.0000 (100.0000) time: 16.6954 data: 10.9888 max mem: 3531 +Test: [ 10/234] eta: 0:06:21 acc1_pos: 50.781 acc5_pos: 100.000 loss: 0.8009 (0.8097) acc1: 46.6146 (41.9981) acc5: 100.0000 (100.0000) time: 1.7053 data: 0.9995 max mem: 3531 +Test: [ 20/234] eta: 0:03:32 acc1_pos: 96.354 acc5_pos: 100.000 loss: 0.7856 (0.7610) acc1: 46.6146 (45.9821) acc5: 100.0000 (100.0000) time: 0.2090 data: 0.0006 max mem: 3531 +Test: [ 30/234] eta: 0:03:04 acc1_pos: 80.990 acc5_pos: 100.000 loss: 0.2768 (0.6072) acc1: 94.0104 (61.3071) acc5: 100.0000 (100.0000) time: 0.4608 data: 0.2693 max mem: 3531 +Test: [ 40/234] eta: 0:02:53 acc1_pos: 98.958 acc5_pos: 100.000 loss: 0.2757 (0.5650) acc1: 94.2708 (65.6885) acc5: 100.0000 (100.0000) time: 0.7903 data: 0.5975 max mem: 3531 +Test: [ 50/234] eta: 0:02:35 acc1_pos: 99.219 acc5_pos: 100.000 loss: 0.1931 (0.4900) acc1: 97.3958 (72.0588) acc5: 100.0000 (100.0000) time: 0.7597 data: 0.5683 max mem: 3531 +Test: [ 60/234] eta: 0:02:22 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.1758 (0.4382) acc1: 99.2188 (76.6137) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.6591 data: 0.4695 max mem: 3531 +Test: [ 70/234] eta: 0:02:08 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.1788 (0.4579) acc1: 99.7396 (75.7629) acc5: 100.0000 (100.0000) acc1_neg: 53.385 acc5_neg: 100.000 time: 0.6239 data: 0.4518 max mem: 3531 +Test: [ 80/234] eta: 0:01:56 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.7585 (0.4733) acc1: 57.0312 (75.0675) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.5793 data: 0.4025 max mem: 3531 +Test: [ 90/234] eta: 0:01:45 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.3269 (0.4579) acc1: 91.1458 (76.7485) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.5549 data: 0.3614 max mem: 3531 +Test: [100/234] eta: 0:01:36 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.3395 (0.4684) acc1: 86.7188 (76.2041) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.5759 data: 0.4145 max mem: 3531 +Test: [110/234] eta: 0:01:27 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.4945 (0.4754) acc1: 71.0938 (76.0769) acc5: 100.0000 (100.0000) acc1_neg: 61.979 acc5_neg: 100.000 time: 0.5861 data: 0.4022 max mem: 3531 +Test: [120/234] eta: 0:01:18 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.4761 (0.4798) acc1: 80.2083 (76.0890) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.5437 data: 0.3443 max mem: 3531 +Test: [130/234] eta: 0:01:10 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.4134 (0.4760) acc1: 87.2396 (76.8050) acc5: 100.0000 (100.0000) acc1_neg: 74.740 acc5_neg: 100.000 time: 0.5203 data: 0.3642 max mem: 3531 +Test: [140/234] eta: 0:01:03 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.4976 (0.5044) acc1: 78.1250 (75.2770) acc5: 100.0000 (100.0000) acc1_neg: 26.823 acc5_neg: 100.000 time: 0.5856 data: 0.4225 max mem: 3531 +Test: [150/234] eta: 0:00:56 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.6734 (0.5179) acc1: 65.3646 (74.6327) acc5: 100.0000 (100.0000) acc1_neg: 72.135 acc5_neg: 100.000 time: 0.6239 data: 0.4396 max mem: 3531 +Test: [160/234] eta: 0:00:49 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.4472 (0.5094) acc1: 82.8125 (75.4933) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.5994 data: 0.4247 max mem: 3531 +Test: [170/234] eta: 0:00:42 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.4275 (0.5049) acc1: 84.6354 (76.0691) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.6246 data: 0.4501 max mem: 3531 +Test: [180/234] eta: 0:00:35 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.4403 (0.5153) acc1: 84.1146 (75.2532) acc5: 100.0000 (100.0000) acc1_neg: 27.083 acc5_neg: 100.000 time: 0.6395 data: 0.4755 max mem: 3531 +Test: [190/234] eta: 0:00:29 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.6446 (0.5236) acc1: 62.2396 (74.5351) acc5: 100.0000 (100.0000) acc1_neg: 63.542 acc5_neg: 100.000 time: 0.6103 data: 0.4680 max mem: 3531 +Test: [200/234] eta: 0:00:22 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.6319 (0.5117) acc1: 65.1042 (75.1257) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.6300 data: 0.4864 max mem: 3531 +Test: [210/234] eta: 0:00:15 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.0791 (0.4924) acc1: 99.7396 (76.2959) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.6438 data: 0.5048 max mem: 3532 +Test: [220/234] eta: 0:00:09 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.1532 (0.4783) acc1: 99.7396 (77.2530) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.5304 data: 0.3843 max mem: 3532 +Test: [230/234] eta: 0:00:02 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.3902 (0.4929) acc1: 45.5729 (75.7001) acc5: 100.0000 (100.0000) acc1_neg: 40.365 acc5_neg: 100.000 time: 0.3470 data: 0.1740 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.7904 (0.4970) acc1: 42.1875 (75.3045) acc5: 100.0000 (100.0000) acc1_neg: 47.471 acc5_neg: 100.000 time: 0.2981 data: 0.1068 max mem: 3532 +Test: Total time: 0:02:26 (0.6278 s / it) +* Overall Acc@1 75.305 Acc@5 100.000 + target == real Acc@1 72.254 Acc@5 100.000 + target == fake Acc@1 76.164 Acc@5 100.000 + loss 0.497 + PR-AUC 0.9501 ROC-AUC 0.8411 EER 25.45% (thr=0.5176) +Accuracy of the network on the 89729 test images: 75.3% +[rank0]:[W725 13:13:51.163485485 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 13:14:18.281000 35460 torch/distributed/run.py:766] +W0725 13:14:18.281000 35460 torch/distributed/run.py:766] ***************************************** +W0725 13:14:18.281000 35460 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 13:14:18.281000 35460 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 13:14:57.306185533 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 13:14:57.370686878 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 13:14:57.740443864 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 13:14:58.400903697 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 13:14:58.452073735 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 13:14:58.515584305 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 13:14:58.535174836 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 13:14:58.744551466 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_all_46real_91fake_benchmark_eval_20260725_1215/fixed500k_all_46real_91fake_seed0/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_all_46real_91fake_imgseed20260725_n116-042-186_20260725_1215/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 11:22:14 acc1_pos: 20.312 acc5_pos: 100.000 loss: 1.1162 (1.1162) acc1: 20.3125 (20.3125) acc5: 100.0000 (100.0000) time: 17.1488 data: 9.9694 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/2387] eta: 1:09:01 acc1_pos: 4.167 acc5_pos: 100.000 loss: 1.1568 (1.1391) acc1: 9.1146 (13.0445) acc5: 100.0000 (100.0000) time: 1.7424 data: 0.9068 max mem: 3531 +Test: [ 20/2387] eta: 0:40:07 acc1_pos: 5.208 acc5_pos: 100.000 loss: 1.1364 (1.1375) acc1: 9.1146 (12.8348) acc5: 100.0000 (100.0000) time: 0.2107 data: 0.0261 max mem: 3531 +Test: [ 30/2387] eta: 0:32:39 acc1_pos: 1.302 acc5_pos: 100.000 loss: 1.1355 (1.1434) acc1: 9.1146 (12.0212) acc5: 100.0000 (100.0000) time: 0.3305 data: 0.1639 max mem: 3531 +Test: [ 40/2387] eta: 0:31:44 acc1_pos: 3.385 acc5_pos: 100.000 loss: 1.1495 (1.1535) acc1: 3.9062 (10.3468) acc5: 100.0000 (100.0000) time: 0.5950 data: 0.4273 max mem: 3531 +Test: [ 50/2387] eta: 0:30:49 acc1_pos: 37.500 acc5_pos: 100.000 loss: 1.0876 (1.1428) acc1: 8.5938 (11.7902) acc5: 100.0000 (100.0000) time: 0.7293 data: 0.5651 max mem: 3531 +Test: [ 60/2387] eta: 0:30:06 acc1_pos: 28.646 acc5_pos: 100.000 loss: 1.0412 (1.1284) acc1: 17.9688 (12.9739) acc5: 100.0000 (100.0000) time: 0.7044 data: 0.5257 max mem: 3531 +Test: [ 70/2387] eta: 0:28:44 acc1_pos: 29.948 acc5_pos: 100.000 loss: 1.0938 (1.1236) acc1: 17.9688 (13.5343) acc5: 100.0000 (100.0000) time: 0.6245 data: 0.4626 max mem: 3531 +Test: [ 80/2387] eta: 0:28:22 acc1_pos: 36.719 acc5_pos: 100.000 loss: 1.0307 (1.1071) acc1: 18.4896 (14.4451) acc5: 100.0000 (100.0000) time: 0.6209 data: 0.4740 max mem: 3531 +Test: [ 90/2387] eta: 0:28:08 acc1_pos: 31.771 acc5_pos: 100.000 loss: 0.9534 (1.0864) acc1: 22.9167 (16.2031) acc5: 100.0000 (100.0000) time: 0.7023 data: 0.5284 max mem: 3531 +Test: [ 100/2387] eta: 0:27:44 acc1_pos: 19.271 acc5_pos: 100.000 loss: 0.9534 (1.0778) acc1: 22.9167 (16.7337) acc5: 100.0000 (100.0000) time: 0.6864 data: 0.4962 max mem: 3531 +Test: [ 110/2387] eta: 0:27:20 acc1_pos: 26.302 acc5_pos: 100.000 loss: 0.9271 (1.0642) acc1: 22.9167 (17.6004) acc5: 100.0000 (100.0000) time: 0.6541 data: 0.4449 max mem: 3531 +Test: [ 120/2387] eta: 0:27:14 acc1_pos: 22.396 acc5_pos: 100.000 loss: 0.9688 (1.0611) acc1: 26.3021 (18.2141) acc5: 100.0000 (100.0000) time: 0.6884 data: 0.4809 max mem: 3531 +Test: [ 130/2387] eta: 0:26:47 acc1_pos: 16.667 acc5_pos: 100.000 loss: 0.9934 (1.0551) acc1: 18.2292 (18.1477) acc5: 100.0000 (100.0000) time: 0.6648 data: 0.4700 max mem: 3531 +Test: [ 140/2387] eta: 0:26:38 acc1_pos: 11.979 acc5_pos: 100.000 loss: 0.9750 (1.0498) acc1: 16.6667 (18.3658) acc5: 100.0000 (100.0000) time: 0.6505 data: 0.4549 max mem: 3531 +Test: [ 150/2387] eta: 0:26:32 acc1_pos: 10.417 acc5_pos: 100.000 loss: 0.9546 (1.0436) acc1: 20.0521 (18.6310) acc5: 100.0000 (100.0000) time: 0.7106 data: 0.5100 max mem: 3531 +Test: [ 160/2387] eta: 0:26:31 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.9203 (1.0225) acc1: 22.9167 (20.6813) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.7400 data: 0.5386 max mem: 3531 +Test: [ 170/2387] eta: 0:26:09 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2440 (0.9763) acc1: 97.9167 (25.2482) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.6800 data: 0.4785 max mem: 3531 +Test: [ 180/2387] eta: 0:25:50 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2359 (0.9353) acc1: 98.6979 (29.3077) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.6055 data: 0.4080 max mem: 3531 +Test: [ 190/2387] eta: 0:25:32 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2302 (0.8984) acc1: 98.6979 (32.9475) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.6091 data: 0.4039 max mem: 3531 +Test: [ 200/2387] eta: 0:25:21 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2297 (0.8652) acc1: 98.6979 (36.2186) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.6344 data: 0.4199 max mem: 3531 +Test: [ 210/2387] eta: 0:25:01 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2324 (0.8355) acc1: 98.9583 (39.1909) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.6148 data: 0.4066 max mem: 3532 +Test: [ 220/2387] eta: 0:24:39 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2332 (0.8081) acc1: 98.9583 (41.8953) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.5506 data: 0.3471 max mem: 3532 +Test: [ 230/2387] eta: 0:24:26 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2311 (0.7832) acc1: 98.9583 (44.3655) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.5748 data: 0.3712 max mem: 3532 +Test: [ 240/2387] eta: 0:24:11 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2387 (0.7623) acc1: 98.4375 (46.5681) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.6040 data: 0.3973 max mem: 3532 +Test: [ 250/2387] eta: 0:23:54 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2306 (0.7368) acc1: 98.4375 (48.6875) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5693 data: 0.3554 max mem: 3532 +Test: [ 260/2387] eta: 0:23:44 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.0956 (0.7120) acc1: 100.0000 (50.6535) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5893 data: 0.4116 max mem: 3532 +Test: [ 270/2387] eta: 0:23:34 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.0856 (0.6888) acc1: 100.0000 (52.4744) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.6298 data: 0.4554 max mem: 3532 +Test: [ 280/2387] eta: 0:23:26 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.0879 (0.6676) acc1: 100.0000 (54.1657) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.6396 data: 0.4458 max mem: 3532 +Test: [ 290/2387] eta: 0:23:16 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.0934 (0.6479) acc1: 100.0000 (55.7408) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.6363 data: 0.4488 max mem: 3532 +Test: [ 300/2387] eta: 0:23:02 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.0958 (0.6294) acc1: 100.0000 (57.2112) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5904 data: 0.3786 max mem: 3532 +Test: [ 310/2387] eta: 0:22:52 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1007 (0.6159) acc1: 100.0000 (58.5711) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.5840 data: 0.3780 max mem: 3532 +Test: [ 320/2387] eta: 0:22:41 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.3243 (0.6069) acc1: 99.4792 (59.8390) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.6049 data: 0.3950 max mem: 3532 +Test: [ 330/2387] eta: 0:22:28 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.3283 (0.5985) acc1: 99.4792 (61.0406) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.5756 data: 0.3585 max mem: 3532 +Test: [ 340/2387] eta: 0:22:17 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.3239 (0.5904) acc1: 99.7396 (62.1724) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.5705 data: 0.3669 max mem: 3532 +Test: [ 350/2387] eta: 0:22:09 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.3196 (0.5785) acc1: 100.0000 (63.2479) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.6091 data: 0.4002 max mem: 3532 +Test: [ 360/2387] eta: 0:22:00 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1875 (0.5678) acc1: 99.7396 (64.2486) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.6185 data: 0.4068 max mem: 3532 +Test: [ 370/2387] eta: 0:21:52 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1985 (0.5579) acc1: 99.2188 (65.1933) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.6098 data: 0.4095 max mem: 3532 +Test: [ 380/2387] eta: 0:21:47 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1985 (0.5485) acc1: 99.4792 (66.0898) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.6508 data: 0.4490 max mem: 3532 +Test: [ 390/2387] eta: 0:21:35 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1974 (0.5396) acc1: 99.4792 (66.9384) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.6192 data: 0.4326 max mem: 3532 +Test: [ 400/2387] eta: 0:21:26 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1983 (0.5311) acc1: 99.2188 (67.7467) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.5645 data: 0.3775 max mem: 3532 +Test: [ 410/2387] eta: 0:21:14 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1986 (0.5230) acc1: 99.4792 (68.5175) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.5654 data: 0.3528 max mem: 3532 +Test: [ 420/2387] eta: 0:21:02 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1980 (0.5153) acc1: 99.2188 (69.2492) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.5302 data: 0.3229 max mem: 3532 +Test: [ 430/2387] eta: 0:20:55 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1980 (0.5085) acc1: 99.2188 (69.9373) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.5705 data: 0.3630 max mem: 3532 +Test: [ 440/2387] eta: 0:20:38 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2077 (0.5005) acc1: 99.2188 (70.6054) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5189 data: 0.3127 max mem: 3533 +Test: [ 450/2387] eta: 0:20:30 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.0991 (0.4916) acc1: 100.0000 (71.2572) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5036 data: 0.3030 max mem: 3533 +Test: [ 460/2387] eta: 0:20:19 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.0979 (0.4831) acc1: 100.0000 (71.8807) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5614 data: 0.3630 max mem: 3533 +Test: [ 470/2387] eta: 0:20:12 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.0979 (0.4749) acc1: 100.0000 (72.4777) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5680 data: 0.3611 max mem: 3533 +Test: [ 480/2387] eta: 0:20:05 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.0992 (0.4671) acc1: 100.0000 (73.0499) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.6114 data: 0.4070 max mem: 3533 +Test: [ 490/2387] eta: 0:19:57 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.0997 (0.4596) acc1: 100.0000 (73.5987) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5999 data: 0.4006 max mem: 3533 +Test: [ 500/2387] eta: 0:19:48 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.0990 (0.4524) acc1: 100.0000 (74.1257) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5791 data: 0.3633 max mem: 3533 +Test: [ 510/2387] eta: 0:19:39 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.0989 (0.4456) acc1: 100.0000 (74.6321) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5592 data: 0.3565 max mem: 3533 +Test: [ 520/2387] eta: 0:19:31 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1241 (0.4395) acc1: 100.0000 (75.1190) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5692 data: 0.3628 max mem: 3533 +Test: [ 530/2387] eta: 0:19:29 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1302 (0.4337) acc1: 100.0000 (75.5875) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.6647 data: 0.4568 max mem: 3533 +Test: [ 540/2387] eta: 0:19:19 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1327 (0.4282) acc1: 100.0000 (76.0388) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.6416 data: 0.4732 max mem: 3533 +Test: [ 550/2387] eta: 0:19:11 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1326 (0.4228) acc1: 100.0000 (76.4736) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5592 data: 0.4210 max mem: 3533 +Test: [ 560/2387] eta: 0:19:04 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1320 (0.4176) acc1: 100.0000 (76.8930) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5833 data: 0.4119 max mem: 3533 +Test: [ 570/2387] eta: 0:18:58 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1324 (0.4126) acc1: 100.0000 (77.2977) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.6099 data: 0.4117 max mem: 3533 +Test: [ 580/2387] eta: 0:18:48 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1310 (0.4078) acc1: 100.0000 (77.6884) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5754 data: 0.4038 max mem: 3533 +Test: [ 590/2387] eta: 0:18:40 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1311 (0.4031) acc1: 100.0000 (78.0660) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5349 data: 0.3969 max mem: 3533 +Test: [ 600/2387] eta: 0:18:33 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1301 (0.3986) acc1: 100.0000 (78.4309) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5692 data: 0.4279 max mem: 3533 +Test: [ 610/2387] eta: 0:18:25 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1564 (0.3958) acc1: 100.0000 (78.7750) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.5901 data: 0.4446 max mem: 3533 +Test: [ 620/2387] eta: 0:18:19 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2258 (0.3930) acc1: 99.4792 (79.1054) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.5961 data: 0.4790 max mem: 3533 +Test: [ 630/2387] eta: 0:18:09 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2259 (0.3904) acc1: 99.4792 (79.4271) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.5453 data: 0.4188 max mem: 3533 +Test: [ 640/2387] eta: 0:18:02 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2267 (0.3879) acc1: 99.2188 (79.7358) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.5487 data: 0.3847 max mem: 3533 +Test: [ 650/2387] eta: 0:17:56 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2265 (0.3854) acc1: 99.2188 (80.0355) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.6203 data: 0.4317 max mem: 3533 +Test: [ 660/2387] eta: 0:17:49 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2261 (0.3830) acc1: 99.4792 (80.3313) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.6055 data: 0.4033 max mem: 3533 +Test: [ 670/2387] eta: 0:17:42 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2267 (0.3807) acc1: 99.7396 (80.6197) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.5838 data: 0.4017 max mem: 3534 +Test: [ 680/2387] eta: 0:17:34 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2293 (0.3787) acc1: 99.7396 (80.9016) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5698 data: 0.4151 max mem: 3534 +Test: [ 690/2387] eta: 0:17:26 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2605 (0.3802) acc1: 99.2188 (80.8633) acc5: 100.0000 (100.0000) acc1_neg: 66.667 acc5_neg: 100.000 time: 0.5449 data: 0.3651 max mem: 3534 +Test: [ 700/2387] eta: 0:17:19 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.5611 (0.3828) acc1: 70.5729 (80.7165) acc5: 100.0000 (100.0000) acc1_neg: 72.396 acc5_neg: 100.000 time: 0.5605 data: 0.3939 max mem: 3534 +Test: [ 710/2387] eta: 0:17:10 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.5644 (0.3854) acc1: 69.7917 (80.5643) acc5: 100.0000 (100.0000) acc1_neg: 70.833 acc5_neg: 100.000 time: 0.5465 data: 0.3765 max mem: 3534 +Test: [ 720/2387] eta: 0:17:03 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.5645 (0.3878) acc1: 69.5312 (80.4142) acc5: 100.0000 (100.0000) acc1_neg: 69.792 acc5_neg: 100.000 time: 0.5351 data: 0.3309 max mem: 3534 +Test: [ 730/2387] eta: 0:16:56 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.5637 (0.3902) acc1: 69.5312 (80.2771) acc5: 100.0000 (100.0000) acc1_neg: 70.573 acc5_neg: 100.000 time: 0.5782 data: 0.3764 max mem: 3534 +Test: [ 740/2387] eta: 0:16:50 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.5637 (0.3926) acc1: 70.5729 (80.1451) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.5940 data: 0.3904 max mem: 3534 +Test: [ 750/2387] eta: 0:16:43 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.5564 (0.3948) acc1: 70.3125 (80.0114) acc5: 100.0000 (100.0000) acc1_neg: 72.135 acc5_neg: 100.000 time: 0.5903 data: 0.3782 max mem: 3534 +Test: [ 760/2387] eta: 0:16:38 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.5523 (0.3967) acc1: 70.5729 (79.9216) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.6200 data: 0.4166 max mem: 3534 +Test: [ 770/2387] eta: 0:16:31 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.4984 (0.3965) acc1: 76.8229 (80.0097) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.6260 data: 0.4298 max mem: 3534 +Test: [ 780/2387] eta: 0:16:25 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2964 (0.3950) acc1: 95.8333 (80.2300) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5998 data: 0.3935 max mem: 3534 +Test: [ 790/2387] eta: 0:16:19 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2536 (0.3929) acc1: 97.9167 (80.4269) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.6039 data: 0.3888 max mem: 3534 +Test: [ 800/2387] eta: 0:16:11 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2169 (0.3909) acc1: 100.0000 (80.6651) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5749 data: 0.3702 max mem: 3534 +Test: [ 810/2387] eta: 0:16:08 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2273 (0.3894) acc1: 99.7396 (80.8839) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.6448 data: 0.4408 max mem: 3534 +Test: [ 820/2387] eta: 0:16:00 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2307 (0.3879) acc1: 99.4792 (81.0860) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.6449 data: 0.4496 max mem: 3534 +Test: [ 830/2387] eta: 0:15:54 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.3142 (0.3879) acc1: 93.2292 (81.1873) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.5795 data: 0.3930 max mem: 3534 +Test: [ 840/2387] eta: 0:15:47 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2833 (0.3862) acc1: 97.1354 (81.3949) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5899 data: 0.3848 max mem: 3534 +Test: [ 850/2387] eta: 0:15:40 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2347 (0.3843) acc1: 99.4792 (81.6044) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5564 data: 0.3448 max mem: 3534 +Test: [ 860/2387] eta: 0:15:32 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1984 (0.3822) acc1: 99.7396 (81.8153) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5362 data: 0.3315 max mem: 3534 +Test: [ 870/2387] eta: 0:15:25 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2095 (0.3803) acc1: 100.0000 (82.0214) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5508 data: 0.3393 max mem: 3534 +Test: [ 880/2387] eta: 0:15:18 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2201 (0.3785) acc1: 99.7396 (82.2225) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5668 data: 0.3458 max mem: 3534 +Test: [ 890/2387] eta: 0:15:07 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2261 (0.3769) acc1: 99.7396 (82.4171) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.4349 data: 0.2304 max mem: 3534 +Test: [ 900/2387] eta: 0:14:56 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2665 (0.3784) acc1: 98.4375 (82.3365) acc5: 100.0000 (100.0000) acc1_neg: 70.312 acc5_neg: 100.000 time: 0.2896 data: 0.0932 max mem: 3535 +Test: [ 910/2387] eta: 0:14:45 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.5892 (0.3808) acc1: 68.4896 (82.1816) acc5: 100.0000 (100.0000) acc1_neg: 69.271 acc5_neg: 100.000 time: 0.2763 data: 0.0780 max mem: 3535 +Test: [ 920/2387] eta: 0:14:34 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.5896 (0.3830) acc1: 67.9688 (82.0267) acc5: 100.0000 (100.0000) acc1_neg: 65.104 acc5_neg: 100.000 time: 0.3130 data: 0.1269 max mem: 3535 +Test: [ 930/2387] eta: 0:14:22 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.5900 (0.3852) acc1: 67.9688 (81.8833) acc5: 100.0000 (100.0000) acc1_neg: 66.927 acc5_neg: 100.000 time: 0.2487 data: 0.1137 max mem: 3535 +Test: [ 940/2387] eta: 0:14:15 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.5919 (0.3874) acc1: 67.7083 (81.7241) acc5: 100.0000 (100.0000) acc1_neg: 64.062 acc5_neg: 100.000 time: 0.3399 data: 0.2384 max mem: 3535 +Test: [ 950/2387] eta: 0:14:07 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.5911 (0.3895) acc1: 67.7083 (81.5772) acc5: 100.0000 (100.0000) acc1_neg: 70.573 acc5_neg: 100.000 time: 0.5044 data: 0.3985 max mem: 3535 +Test: [ 960/2387] eta: 0:14:02 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.5825 (0.3915) acc1: 67.9688 (81.4451) acc5: 100.0000 (100.0000) acc1_neg: 72.396 acc5_neg: 100.000 time: 0.5553 data: 0.4525 max mem: 3535 +Test: [ 970/2387] eta: 0:13:55 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.5931 (0.3936) acc1: 68.2292 (81.3101) acc5: 100.0000 (100.0000) acc1_neg: 69.010 acc5_neg: 100.000 time: 0.5697 data: 0.4670 max mem: 3535 +Test: [ 980/2387] eta: 0:13:46 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.5608 (0.3939) acc1: 70.3125 (81.3517) acc5: 100.0000 (100.0000) acc1_neg: 70.312 acc5_neg: 100.000 time: 0.4607 data: 0.3413 max mem: 3535 +Test: [ 990/2387] eta: 0:13:37 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.3318 (0.3930) acc1: 93.7500 (81.4878) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.3945 data: 0.2401 max mem: 3535 +Test: [1000/2387] eta: 0:13:32 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2848 (0.3922) acc1: 98.4375 (81.6434) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.5086 data: 0.3647 max mem: 3535 +Test: [1010/2387] eta: 0:13:27 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.3042 (0.3916) acc1: 97.1354 (81.7495) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.6248 data: 0.5145 max mem: 3535 +Test: [1020/2387] eta: 0:13:20 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.3042 (0.3908) acc1: 94.5312 (81.8639) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.5798 data: 0.4658 max mem: 3535 +Test: [1030/2387] eta: 0:13:15 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2439 (0.3893) acc1: 97.9167 (82.0300) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5708 data: 0.4427 max mem: 3535 +Test: [1040/2387] eta: 0:13:08 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1968 (0.3873) acc1: 100.0000 (82.2014) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5699 data: 0.4425 max mem: 3535 +Test: [1050/2387] eta: 0:13:03 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1774 (0.3854) acc1: 100.0000 (82.3653) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5889 data: 0.4512 max mem: 3535 +Test: [1060/2387] eta: 0:12:58 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1781 (0.3839) acc1: 100.0000 (82.5270) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.6299 data: 0.4967 max mem: 3535 +Test: [1070/2387] eta: 0:12:51 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1964 (0.3825) acc1: 100.0000 (82.6681) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.5862 data: 0.4709 max mem: 3535 +Test: [1080/2387] eta: 0:12:45 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2587 (0.3818) acc1: 98.1771 (82.7855) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.5499 data: 0.4119 max mem: 3535 +Test: [1090/2387] eta: 0:12:40 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2700 (0.3807) acc1: 97.6562 (82.9252) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.5937 data: 0.4427 max mem: 3535 +Test: [1100/2387] eta: 0:12:34 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2328 (0.3791) acc1: 99.4792 (83.0755) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.6198 data: 0.4816 max mem: 3535 +Test: [1110/2387] eta: 0:12:28 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2184 (0.3781) acc1: 99.2188 (83.1693) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.6009 data: 0.4475 max mem: 3535 +Test: [1120/2387] eta: 0:12:23 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2543 (0.3770) acc1: 98.6979 (83.3085) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.6202 data: 0.4774 max mem: 3536 +Test: [1130/2387] eta: 0:12:18 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2296 (0.3758) acc1: 99.2188 (83.4475) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.6344 data: 0.5111 max mem: 3536 +Test: [1140/2387] eta: 0:12:12 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2699 (0.3751) acc1: 98.6979 (83.5556) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.6045 data: 0.4894 max mem: 3536 +Test: [1150/2387] eta: 0:12:07 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2699 (0.3738) acc1: 98.6979 (83.6788) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.6248 data: 0.4957 max mem: 3536 +Test: [1160/2387] eta: 0:12:01 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1612 (0.3718) acc1: 100.0000 (83.8192) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.6249 data: 0.4857 max mem: 3536 +Test: [1170/2387] eta: 0:11:55 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1309 (0.3697) acc1: 100.0000 (83.9574) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5692 data: 0.4332 max mem: 3536 +Test: [1180/2387] eta: 0:11:49 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1309 (0.3677) acc1: 100.0000 (84.0919) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.6105 data: 0.4532 max mem: 3536 +Test: [1190/2387] eta: 0:11:44 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1366 (0.3658) acc1: 100.0000 (84.2254) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.6248 data: 0.4429 max mem: 3536 +Test: [1200/2387] eta: 0:11:37 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1381 (0.3640) acc1: 100.0000 (84.3564) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5609 data: 0.3721 max mem: 3536 +Test: [1210/2387] eta: 0:11:31 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1502 (0.3623) acc1: 100.0000 (84.4855) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5556 data: 0.3752 max mem: 3536 +Test: [1220/2387] eta: 0:11:26 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1200 (0.3602) acc1: 100.0000 (84.6124) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5983 data: 0.4253 max mem: 3536 +Test: [1230/2387] eta: 0:11:20 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1077 (0.3581) acc1: 100.0000 (84.7374) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.6353 data: 0.4682 max mem: 3536 +Test: [1240/2387] eta: 0:11:15 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.0941 (0.3560) acc1: 100.0000 (84.8604) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.6297 data: 0.4490 max mem: 3536 +Test: [1250/2387] eta: 0:11:08 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.0956 (0.3540) acc1: 100.0000 (84.9814) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5838 data: 0.3945 max mem: 3536 +Test: [1260/2387] eta: 0:11:03 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1058 (0.3520) acc1: 100.0000 (85.1005) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5800 data: 0.3995 max mem: 3536 +Test: [1270/2387] eta: 0:10:57 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1071 (0.3502) acc1: 100.0000 (85.2177) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.6155 data: 0.4290 max mem: 3536 +Test: [1280/2387] eta: 0:10:50 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1315 (0.3487) acc1: 100.0000 (85.3319) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5458 data: 0.3395 max mem: 3536 +Test: [1290/2387] eta: 0:10:44 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1545 (0.3472) acc1: 99.7396 (85.4286) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.5099 data: 0.3124 max mem: 3536 +Test: [1300/2387] eta: 0:10:38 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1545 (0.3457) acc1: 99.7396 (85.5344) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.5848 data: 0.3868 max mem: 3536 +Test: [1310/2387] eta: 0:10:31 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1517 (0.3444) acc1: 99.7396 (85.6358) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.4902 data: 0.2903 max mem: 3536 +Test: [1320/2387] eta: 0:10:24 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1747 (0.3431) acc1: 99.7396 (85.7327) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.4351 data: 0.2365 max mem: 3536 +Test: [1330/2387] eta: 0:10:18 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1930 (0.3423) acc1: 99.4792 (85.8277) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.5238 data: 0.3180 max mem: 3536 +Test: [1340/2387] eta: 0:10:12 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2435 (0.3426) acc1: 96.6146 (85.8068) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5846 data: 0.3804 max mem: 3536 +Test: [1350/2387] eta: 0:10:07 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1208 (0.3407) acc1: 100.0000 (85.9119) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.6005 data: 0.4085 max mem: 3537 +Test: [1360/2387] eta: 0:10:01 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.0818 (0.3388) acc1: 100.0000 (86.0154) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.6047 data: 0.4032 max mem: 3537 +Test: [1370/2387] eta: 0:09:55 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.0726 (0.3368) acc1: 100.0000 (86.1174) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5998 data: 0.3872 max mem: 3537 +Test: [1380/2387] eta: 0:09:49 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.0777 (0.3350) acc1: 100.0000 (86.2179) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5746 data: 0.3651 max mem: 3537 +Test: [1390/2387] eta: 0:09:43 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.0865 (0.3332) acc1: 100.0000 (86.3170) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5509 data: 0.3444 max mem: 3537 +Test: [1400/2387] eta: 0:09:37 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.0882 (0.3315) acc1: 100.0000 (86.4147) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5801 data: 0.3708 max mem: 3537 +Test: [1410/2387] eta: 0:09:32 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1037 (0.3307) acc1: 100.0000 (86.5008) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.6239 data: 0.4139 max mem: 3537 +Test: [1420/2387] eta: 0:09:26 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2395 (0.3302) acc1: 99.2188 (86.5677) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.5840 data: 0.3808 max mem: 3537 +Test: [1430/2387] eta: 0:09:20 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2518 (0.3297) acc1: 99.4792 (86.6536) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5900 data: 0.3830 max mem: 3537 +Test: [1440/2387] eta: 0:09:15 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2457 (0.3291) acc1: 99.4792 (86.7316) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.6355 data: 0.4302 max mem: 3537 +Test: [1450/2387] eta: 0:09:08 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2629 (0.3290) acc1: 98.4375 (86.7916) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5900 data: 0.3889 max mem: 3537 +Test: [1460/2387] eta: 0:09:02 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2057 (0.3281) acc1: 97.3958 (86.8726) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5559 data: 0.3587 max mem: 3537 +Test: [1470/2387] eta: 0:08:56 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1759 (0.3269) acc1: 100.0000 (86.9613) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.5695 data: 0.3657 max mem: 3537 +Test: [1480/2387] eta: 0:08:50 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1604 (0.3258) acc1: 100.0000 (87.0483) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.5381 data: 0.3350 max mem: 3537 +Test: [1490/2387] eta: 0:08:44 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1643 (0.3247) acc1: 100.0000 (87.1336) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.5452 data: 0.3326 max mem: 3537 +Test: [1500/2387] eta: 0:08:38 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1615 (0.3237) acc1: 99.7396 (87.2177) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5611 data: 0.3436 max mem: 3537 +Test: [1510/2387] eta: 0:08:33 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1659 (0.3226) acc1: 100.0000 (87.3020) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5952 data: 0.3882 max mem: 3537 +Test: [1520/2387] eta: 0:08:27 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1908 (0.3219) acc1: 100.0000 (87.3778) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.6295 data: 0.4209 max mem: 3537 +Test: [1530/2387] eta: 0:08:21 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2000 (0.3211) acc1: 99.4792 (87.4583) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.6046 data: 0.3925 max mem: 3537 +Test: [1540/2387] eta: 0:08:15 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1870 (0.3202) acc1: 100.0000 (87.5385) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5654 data: 0.3558 max mem: 3537 +Test: [1550/2387] eta: 0:08:09 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1836 (0.3194) acc1: 100.0000 (87.6184) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5787 data: 0.3784 max mem: 3537 +Test: [1560/2387] eta: 0:08:04 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1852 (0.3185) acc1: 100.0000 (87.6965) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.6139 data: 0.4137 max mem: 3537 +Test: [1570/2387] eta: 0:07:58 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1902 (0.3178) acc1: 99.7396 (87.7735) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.5961 data: 0.3947 max mem: 3537 +Test: [1580/2387] eta: 0:07:52 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2178 (0.3174) acc1: 99.4792 (87.8410) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.5800 data: 0.3740 max mem: 3538 +Test: [1590/2387] eta: 0:07:46 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2321 (0.3169) acc1: 99.2188 (87.9107) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.5696 data: 0.3698 max mem: 3538 +Test: [1600/2387] eta: 0:07:40 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2604 (0.3167) acc1: 99.2188 (87.9481) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.5592 data: 0.3610 max mem: 3538 +Test: [1610/2387] eta: 0:07:34 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2293 (0.3160) acc1: 99.7396 (88.0202) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.5805 data: 0.3664 max mem: 3538 +Test: [1620/2387] eta: 0:07:28 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2251 (0.3156) acc1: 99.7396 (88.0843) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5926 data: 0.3745 max mem: 3538 +Test: [1630/2387] eta: 0:07:22 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2523 (0.3157) acc1: 97.9167 (88.1331) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.5350 data: 0.3226 max mem: 3538 +Test: [1640/2387] eta: 0:07:16 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.3739 (0.3164) acc1: 91.9271 (88.1178) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.5443 data: 0.3333 max mem: 3538 +Test: [1650/2387] eta: 0:07:10 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.3844 (0.3168) acc1: 89.3229 (88.1357) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.5604 data: 0.3502 max mem: 3538 +Test: [1660/2387] eta: 0:07:05 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.3743 (0.3172) acc1: 90.6250 (88.1522) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.5897 data: 0.3725 max mem: 3538 +Test: [1670/2387] eta: 0:06:58 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.3949 (0.3177) acc1: 88.8021 (88.1648) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.5561 data: 0.3416 max mem: 3538 +Test: [1680/2387] eta: 0:06:51 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.3949 (0.3181) acc1: 88.8021 (88.1792) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.3912 data: 0.1992 max mem: 3538 +Test: [1690/2387] eta: 0:06:45 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.3923 (0.3186) acc1: 92.1875 (88.1927) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.4144 data: 0.2239 max mem: 3538 +Test: [1700/2387] eta: 0:06:39 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.3914 (0.3185) acc1: 93.4896 (88.2341) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5596 data: 0.3583 max mem: 3538 +Test: [1710/2387] eta: 0:06:34 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1525 (0.3175) acc1: 100.0000 (88.3024) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5896 data: 0.3858 max mem: 3538 +Test: [1720/2387] eta: 0:06:28 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1283 (0.3163) acc1: 100.0000 (88.3702) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.5582 data: 0.3634 max mem: 3538 +Test: [1730/2387] eta: 0:06:22 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1203 (0.3152) acc1: 100.0000 (88.4373) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5498 data: 0.3571 max mem: 3538 +Test: [1740/2387] eta: 0:06:16 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1269 (0.3142) acc1: 100.0000 (88.5032) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5464 data: 0.3389 max mem: 3538 +Test: [1750/2387] eta: 0:06:10 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1321 (0.3131) acc1: 100.0000 (88.5687) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.6148 data: 0.4062 max mem: 3538 +Test: [1760/2387] eta: 0:06:04 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1412 (0.3123) acc1: 100.0000 (88.6334) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.6133 data: 0.4065 max mem: 3538 +Test: [1770/2387] eta: 0:05:58 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1813 (0.3116) acc1: 100.0000 (88.6933) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.5546 data: 0.3473 max mem: 3538 +Test: [1780/2387] eta: 0:05:53 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1902 (0.3109) acc1: 99.7396 (88.7560) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5761 data: 0.3677 max mem: 3538 +Test: [1790/2387] eta: 0:05:46 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1742 (0.3100) acc1: 100.0000 (88.8184) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5239 data: 0.3172 max mem: 3538 +Test: [1800/2387] eta: 0:05:40 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1590 (0.3092) acc1: 100.0000 (88.8797) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.4692 data: 0.2663 max mem: 3538 +Test: [1810/2387] eta: 0:05:35 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1542 (0.3084) acc1: 100.0000 (88.9404) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.5598 data: 0.3496 max mem: 3539 +Test: [1820/2387] eta: 0:05:29 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1605 (0.3076) acc1: 100.0000 (88.9991) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.6148 data: 0.4004 max mem: 3539 +Test: [1830/2387] eta: 0:05:23 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1722 (0.3068) acc1: 100.0000 (89.0585) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5767 data: 0.3651 max mem: 3539 +Test: [1840/2387] eta: 0:05:16 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1161 (0.3058) acc1: 100.0000 (89.1180) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.4527 data: 0.2532 max mem: 3539 +Test: [1850/2387] eta: 0:05:11 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1150 (0.3048) acc1: 100.0000 (89.1767) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.4374 data: 0.2410 max mem: 3539 +Test: [1860/2387] eta: 0:05:05 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1161 (0.3038) acc1: 100.0000 (89.2348) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5601 data: 0.3614 max mem: 3539 +Test: [1870/2387] eta: 0:04:59 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1180 (0.3028) acc1: 100.0000 (89.2923) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.6040 data: 0.3886 max mem: 3539 +Test: [1880/2387] eta: 0:04:53 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1176 (0.3018) acc1: 100.0000 (89.3491) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.6106 data: 0.4001 max mem: 3539 +Test: [1890/2387] eta: 0:04:48 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1167 (0.3008) acc1: 100.0000 (89.4054) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5755 data: 0.3777 max mem: 3539 +Test: [1900/2387] eta: 0:04:42 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1178 (0.2999) acc1: 100.0000 (89.4610) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5288 data: 0.3193 max mem: 3539 +Test: [1910/2387] eta: 0:04:36 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1263 (0.2990) acc1: 100.0000 (89.5157) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5448 data: 0.3411 max mem: 3539 +Test: [1920/2387] eta: 0:04:30 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1776 (0.2989) acc1: 100.0000 (89.5538) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.5898 data: 0.3796 max mem: 3539 +Test: [1930/2387] eta: 0:04:24 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2068 (0.2984) acc1: 100.0000 (89.6079) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5797 data: 0.3674 max mem: 3539 +Test: [1940/2387] eta: 0:04:18 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1963 (0.2982) acc1: 100.0000 (89.6269) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5842 data: 0.3800 max mem: 3539 +Test: [1950/2387] eta: 0:04:13 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2225 (0.2979) acc1: 100.0000 (89.6796) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5842 data: 0.3760 max mem: 3539 +Test: [1960/2387] eta: 0:04:07 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2152 (0.2976) acc1: 100.0000 (89.7257) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5700 data: 0.3608 max mem: 3539 +Test: [1970/2387] eta: 0:04:01 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2933 (0.2978) acc1: 100.0000 (89.7474) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.5548 data: 0.3486 max mem: 3539 +Test: [1980/2387] eta: 0:03:55 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.3045 (0.2975) acc1: 97.9167 (89.7799) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.4853 data: 0.2782 max mem: 3539 +Test: [1990/2387] eta: 0:03:48 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1398 (0.2967) acc1: 98.9583 (89.8269) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.3567 data: 0.1549 max mem: 3539 +Test: [2000/2387] eta: 0:03:42 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1384 (0.2959) acc1: 98.9583 (89.8732) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2451 data: 0.0499 max mem: 3539 +Test: [2010/2387] eta: 0:03:36 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1406 (0.2951) acc1: 98.9583 (89.9192) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2080 data: 0.0121 max mem: 3539 +Test: [2020/2387] eta: 0:03:29 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1400 (0.2944) acc1: 98.9583 (89.9653) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2712 data: 0.1156 max mem: 3539 +Test: [2030/2387] eta: 0:03:23 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1375 (0.2936) acc1: 98.9583 (90.0104) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.3605 data: 0.2584 max mem: 3540 +Test: [2040/2387] eta: 0:03:17 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1377 (0.2928) acc1: 99.2188 (90.0552) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.4030 data: 0.3114 max mem: 3540 +Test: [2050/2387] eta: 0:03:12 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1392 (0.2921) acc1: 99.2188 (90.0995) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.5186 data: 0.4211 max mem: 3540 +Test: [2060/2387] eta: 0:03:06 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1444 (0.2914) acc1: 98.6979 (90.1404) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.5649 data: 0.4671 max mem: 3540 +Test: [2070/2387] eta: 0:03:00 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1483 (0.2907) acc1: 98.9583 (90.1836) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.5110 data: 0.4078 max mem: 3540 +Test: [2080/2387] eta: 0:02:54 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1474 (0.2900) acc1: 98.9583 (90.2244) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.4245 data: 0.2870 max mem: 3540 +Test: [2090/2387] eta: 0:02:48 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1496 (0.2894) acc1: 98.9583 (90.2654) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.4688 data: 0.3264 max mem: 3540 +Test: [2100/2387] eta: 0:02:43 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1471 (0.2887) acc1: 98.9583 (90.3072) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.5949 data: 0.4551 max mem: 3540 +Test: [2110/2387] eta: 0:02:37 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1441 (0.2880) acc1: 98.9583 (90.3479) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.6153 data: 0.4734 max mem: 3540 +Test: [2120/2387] eta: 0:02:32 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1644 (0.2879) acc1: 98.4375 (90.3753) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.6199 data: 0.4929 max mem: 3540 +Test: [2130/2387] eta: 0:02:26 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2089 (0.2875) acc1: 98.4375 (90.4166) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.5948 data: 0.4742 max mem: 3540 +Test: [2140/2387] eta: 0:02:20 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2040 (0.2874) acc1: 99.2188 (90.4326) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5655 data: 0.4284 max mem: 3540 +Test: [2150/2387] eta: 0:02:14 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2225 (0.2872) acc1: 99.4792 (90.4705) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.5554 data: 0.4297 max mem: 3540 +Test: [2160/2387] eta: 0:02:09 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2037 (0.2869) acc1: 99.7396 (90.5084) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5591 data: 0.4309 max mem: 3540 +Test: [2170/2387] eta: 0:02:03 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2345 (0.2871) acc1: 98.6979 (90.5217) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.5555 data: 0.4204 max mem: 3540 +Test: [2180/2387] eta: 0:01:57 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.3669 (0.2885) acc1: 92.4479 (90.4248) acc5: 100.0000 (100.0000) acc1_neg: 33.333 acc5_neg: 100.000 time: 0.5668 data: 0.4266 max mem: 3540 +Test: [2190/2387] eta: 0:01:52 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.3669 (0.2888) acc1: 93.7500 (90.4125) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5906 data: 0.4438 max mem: 3540 +Test: [2200/2387] eta: 0:01:46 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2431 (0.2885) acc1: 98.1771 (90.4531) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.6347 data: 0.5040 max mem: 3540 +Test: [2210/2387] eta: 0:01:40 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1872 (0.2880) acc1: 100.0000 (90.4961) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5994 data: 0.4659 max mem: 3540 +Test: [2220/2387] eta: 0:01:35 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1700 (0.2875) acc1: 100.0000 (90.5385) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5514 data: 0.4149 max mem: 3540 +Test: [2230/2387] eta: 0:01:29 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1757 (0.2870) acc1: 100.0000 (90.5804) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5796 data: 0.4279 max mem: 3540 +Test: [2240/2387] eta: 0:01:23 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1830 (0.2866) acc1: 100.0000 (90.6187) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.6045 data: 0.4408 max mem: 3540 +Test: [2250/2387] eta: 0:01:18 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1964 (0.2862) acc1: 100.0000 (90.6580) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.6449 data: 0.4967 max mem: 3540 +Test: [2260/2387] eta: 0:01:12 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2005 (0.2858) acc1: 99.7396 (90.6974) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.6291 data: 0.4656 max mem: 3541 +Test: [2270/2387] eta: 0:01:06 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1711 (0.2853) acc1: 100.0000 (90.7384) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5599 data: 0.3829 max mem: 3541 +Test: [2280/2387] eta: 0:01:01 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1640 (0.2848) acc1: 100.0000 (90.7604) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.5544 data: 0.3956 max mem: 3541 +Test: [2290/2387] eta: 0:00:55 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1439 (0.2842) acc1: 100.0000 (90.7996) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.5849 data: 0.4227 max mem: 3541 +Test: [2300/2387] eta: 0:00:49 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1544 (0.2838) acc1: 100.0000 (90.8382) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.5812 data: 0.4264 max mem: 3541 +Test: [2310/2387] eta: 0:00:43 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1817 (0.2835) acc1: 100.0000 (90.8711) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.5748 data: 0.4165 max mem: 3541 +Test: [2320/2387] eta: 0:00:38 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2430 (0.2834) acc1: 98.9583 (90.8998) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.5846 data: 0.4113 max mem: 3541 +Test: [2330/2387] eta: 0:00:32 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.3003 (0.2836) acc1: 98.6979 (90.9188) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.5698 data: 0.3798 max mem: 3541 +Test: [2340/2387] eta: 0:00:26 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.3003 (0.2835) acc1: 98.9583 (90.9480) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.5442 data: 0.3397 max mem: 3541 +Test: [2350/2387] eta: 0:00:21 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2635 (0.2837) acc1: 99.2188 (90.9542) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.5806 data: 0.3776 max mem: 3541 +Test: [2360/2387] eta: 0:00:15 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2750 (0.2839) acc1: 98.6979 (90.9614) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.5599 data: 0.3597 max mem: 3541 +Test: [2370/2387] eta: 0:00:09 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2806 (0.2839) acc1: 98.4375 (90.9860) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.5296 data: 0.3338 max mem: 3541 +Test: [2380/2387] eta: 0:00:03 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.3421 (0.2843) acc1: 95.5729 (90.9811) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.5048 data: 0.3188 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.4034 (0.2847) acc1: 90.1042 (90.9738) acc5: 100.0000 (100.0000) acc1_neg: 86.047 acc5_neg: 100.000 time: 0.4114 data: 0.2269 max mem: 3541 +Test: Total time: 0:22:40 (0.5698 s / it) +* Overall Acc@1 90.974 Acc@5 100.000 + target == real Acc@1 18.581 Acc@5 100.000 + target == fake Acc@1 96.058 Acc@5 100.000 + loss 0.285 + PR-AUC 0.9880 ROC-AUC 0.8525 EER 22.11% (thr=0.7354) +Accuracy of the network on the 916267 test images: 91.0% +[rank0]:[W725 13:38:20.563786188 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 13:38:50.281000 52595 torch/distributed/run.py:766] +W0725 13:38:50.281000 52595 torch/distributed/run.py:766] ***************************************** +W0725 13:38:50.281000 52595 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 13:38:50.281000 52595 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W725 13:39:31.905631650 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +NCCL version 2.26.2+cuda12.2 +[rank6]:[W725 13:39:32.021488664 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 13:39:33.321129037 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 13:39:33.859600242 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank1]:[W725 13:39:33.859816066 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 13:39:34.725601264 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W725 13:39:34.727543509 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 13:39:34.856600426 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_all_46real_91fake_benchmark_eval_20260725_1215/fixed500k_all_46real_91fake_seed0/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_all_46real_91fake_imgseed20260725_n116-042-186_20260725_1215/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 2:43:48 acc1_pos: 20.312 acc5_pos: 100.000 loss: 1.1162 (1.1162) acc1: 20.3125 (20.3125) acc5: 100.0000 (100.0000) time: 19.0100 data: 14.1738 max mem: 3529 +Test: [ 10/517] eta: 0:16:06 acc1_pos: 4.167 acc5_pos: 100.000 loss: 1.1568 (1.1391) acc1: 9.1146 (13.0445) acc5: 100.0000 (100.0000) time: 1.9067 data: 1.2890 max mem: 3529 +Test: [ 20/517] eta: 0:09:10 acc1_pos: 5.208 acc5_pos: 100.000 loss: 1.1364 (1.1375) acc1: 9.1146 (12.8224) acc5: 100.0000 (100.0000) time: 0.2125 data: 0.0201 max mem: 3529 +Test: [ 30/517] eta: 0:07:39 acc1_pos: 1.302 acc5_pos: 100.000 loss: 1.1354 (1.1433) acc1: 9.1146 (12.0212) acc5: 100.0000 (100.0000) time: 0.4153 data: 0.2142 max mem: 3529 +Test: [ 40/517] eta: 0:07:02 acc1_pos: 3.385 acc5_pos: 100.000 loss: 1.1495 (1.1535) acc1: 3.9062 (10.3532) acc5: 100.0000 (100.0000) time: 0.6549 data: 0.4501 max mem: 3529 +Test: [ 50/517] eta: 0:06:47 acc1_pos: 15.138 acc5_pos: 100.000 loss: 1.0876 (1.1169) acc1: 8.5938 (14.5731) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.7638 data: 0.5683 max mem: 3529 +Test: [ 60/517] eta: 0:06:30 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.4277 (0.9974) acc1: 86.1979 (27.1602) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.7855 data: 0.5830 max mem: 3530 +Test: [ 70/517] eta: 0:06:13 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.3525 (0.9085) acc1: 96.3542 (37.0562) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.7399 data: 0.5322 max mem: 3530 +Test: [ 80/517] eta: 0:06:00 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.3857 (0.8447) acc1: 96.6146 (44.2612) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.7350 data: 0.5290 max mem: 3530 +Test: [ 90/517] eta: 0:05:48 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.3857 (0.7931) acc1: 95.8333 (49.9857) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.7406 data: 0.5398 max mem: 3530 +Test: [100/517] eta: 0:05:35 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.3634 (0.7496) acc1: 95.8333 (54.4348) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.7249 data: 0.5220 max mem: 3530 +Test: [110/517] eta: 0:05:23 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.3412 (0.7118) acc1: 95.5729 (58.2301) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.6999 data: 0.4949 max mem: 3530 +Test: [120/517] eta: 0:05:14 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.3634 (0.6843) acc1: 98.1771 (61.4626) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.7240 data: 0.5264 max mem: 3530 +Test: [130/517] eta: 0:05:02 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.3726 (0.6593) acc1: 98.4375 (64.2931) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.7112 data: 0.5115 max mem: 3530 +Test: [140/517] eta: 0:04:45 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.2667 (0.6309) acc1: 100.0000 (66.8237) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5498 data: 0.3508 max mem: 3530 +Test: [150/517] eta: 0:04:27 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.2577 (0.6069) acc1: 100.0000 (68.9794) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.3836 data: 0.1832 max mem: 3530 +Test: [160/517] eta: 0:04:10 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.2558 (0.5846) acc1: 99.7396 (70.8899) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.3149 data: 0.1624 max mem: 3530 +Test: [170/517] eta: 0:03:57 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.2467 (0.5638) acc1: 99.7396 (72.5816) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.3491 data: 0.2464 max mem: 3530 +Test: [180/517] eta: 0:03:48 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.2467 (0.5489) acc1: 99.7396 (73.9756) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.4899 data: 0.3878 max mem: 3530 +Test: [190/517] eta: 0:03:42 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.2318 (0.5324) acc1: 99.4792 (75.3190) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.6549 data: 0.5560 max mem: 3530 +Test: [200/517] eta: 0:03:36 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.2387 (0.5192) acc1: 99.7396 (76.5353) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.7099 data: 0.6116 max mem: 3530 +Test: [210/517] eta: 0:03:29 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.3106 (0.5151) acc1: 96.0938 (77.0019) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.6905 data: 0.5441 max mem: 3530 +Test: [220/517] eta: 0:03:19 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.4259 (0.5104) acc1: 86.7188 (77.5146) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.5599 data: 0.4135 max mem: 3530 +Test: [230/517] eta: 0:03:14 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.3922 (0.5054) acc1: 91.6667 (78.3719) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.6147 data: 0.4719 max mem: 3530 +Test: [240/517] eta: 0:03:06 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.3931 (0.5020) acc1: 98.1771 (79.1116) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.7098 data: 0.5668 max mem: 3530 +Test: [250/517] eta: 0:03:01 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.3909 (0.4940) acc1: 98.6979 (79.8763) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.7301 data: 0.5968 max mem: 3530 +Test: [260/517] eta: 0:02:54 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.2571 (0.4855) acc1: 100.0000 (80.6124) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.7349 data: 0.6017 max mem: 3530 +Test: [270/517] eta: 0:02:48 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.3325 (0.4812) acc1: 97.1354 (81.1433) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.6992 data: 0.6019 max mem: 3530 +Test: [280/517] eta: 0:02:42 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.3601 (0.4768) acc1: 95.0521 (81.6661) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.7549 data: 0.6480 max mem: 3530 +Test: [290/517] eta: 0:02:35 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.3280 (0.4709) acc1: 97.6562 (82.2550) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.7148 data: 0.6083 max mem: 3531 +Test: [300/517] eta: 0:02:28 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.3112 (0.4664) acc1: 99.4792 (82.8064) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.6948 data: 0.5982 max mem: 3531 +Test: [310/517] eta: 0:02:21 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.3291 (0.4618) acc1: 98.4375 (83.2797) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.6799 data: 0.5821 max mem: 3531 +Terminated diff --git a/ScaleDF_rebuttal/_host_logs/all_46real_91fake_11451_20260725_1215/eval_after_train.pid b/ScaleDF_rebuttal/_host_logs/all_46real_91fake_11451_20260725_1215/eval_after_train.pid new file mode 100644 index 0000000000000000000000000000000000000000..57ece9f44f6964ae324b026129656c3f40f31bb4 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/all_46real_91fake_11451_20260725_1215/eval_after_train.pid @@ -0,0 +1 @@ +6796 diff --git a/ScaleDF_rebuttal/_host_logs/all_46real_91fake_11451_20260725_1215/eval_after_train_complete.log b/ScaleDF_rebuttal/_host_logs/all_46real_91fake_11451_20260725_1215/eval_after_train_complete.log new file mode 100644 index 0000000000000000000000000000000000000000..52abaf6d69e5d7c8e76875d12148a249f9b792d7 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/all_46real_91fake_11451_20260725_1215/eval_after_train_complete.log @@ -0,0 +1,7 @@ +[2026-07-25T13:44:23Z] waiter_started run_dir=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_all_46real_91fake_imgseed20260725_n116-042-186_20260725_1215 +[2026-07-25T13:44:23Z] waiting_train_complete last=Epoch: [9] [170/488] eta: 0:04:21 lr: 0.000002 loss: 0.6368 (0.6324) time: 0.4285 data: 0.0005 max mem: 28828 +[2026-07-25T13:45:23Z] waiting_train_complete last=Epoch: [9] [310/488] eta: 0:01:56 lr: 0.000002 loss: 0.6387 (0.6334) time: 0.4396 data: 0.0696 max mem: 28828 +[2026-07-25T13:46:24Z] waiting_train_complete last=Epoch: [9] [440/488] eta: 0:00:28 lr: 0.000002 loss: 0.6296 (0.6323) time: 0.4054 data: 0.0853 max mem: 28828 +[2026-07-25T13:47:24Z] waiting_train_complete last=Averaged stats: lr: 0.000002 loss: 0.6272 (0.6320) +[2026-07-25T13:48:24Z] starting_eval checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_all_46real_91fake_imgseed20260725_n116-042-186_20260725_1215/best_checkpoint.pth output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_all_46real_91fake_benchmark_eval_20260725_1215/fixed500k_all_46real_91fake_seed0 +[2026-07-25T14:10:56Z] eval_done output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_all_46real_91fake_benchmark_eval_20260725_1215/fixed500k_all_46real_91fake_seed0 diff --git a/ScaleDF_rebuttal/_host_logs/all_46real_91fake_11451_20260725_1215/eval_after_train_complete.nohup.log b/ScaleDF_rebuttal/_host_logs/all_46real_91fake_11451_20260725_1215/eval_after_train_complete.nohup.log new file mode 100644 index 0000000000000000000000000000000000000000..91c3afec71f002caecb7f727aa3d1ab4ca7266cd --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/all_46real_91fake_11451_20260725_1215/eval_after_train_complete.nohup.log @@ -0,0 +1,1031 @@ +[2026-07-25T13:44:23Z] waiter_started run_dir=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_all_46real_91fake_imgseed20260725_n116-042-186_20260725_1215 +[2026-07-25T13:44:23Z] waiting_train_complete last=Epoch: [9] [170/488] eta: 0:04:21 lr: 0.000002 loss: 0.6368 (0.6324) time: 0.4285 data: 0.0005 max mem: 28828 +[2026-07-25T13:45:23Z] waiting_train_complete last=Epoch: [9] [310/488] eta: 0:01:56 lr: 0.000002 loss: 0.6387 (0.6334) time: 0.4396 data: 0.0696 max mem: 28828 +[2026-07-25T13:46:24Z] waiting_train_complete last=Epoch: [9] [440/488] eta: 0:00:28 lr: 0.000002 loss: 0.6296 (0.6323) time: 0.4054 data: 0.0853 max mem: 28828 +[2026-07-25T13:47:24Z] waiting_train_complete last=Averaged stats: lr: 0.000002 loss: 0.6272 (0.6320) +[2026-07-25T13:48:24Z] starting_eval checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_all_46real_91fake_imgseed20260725_n116-042-186_20260725_1215/best_checkpoint.pth output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_all_46real_91fake_benchmark_eval_20260725_1215/fixed500k_all_46real_91fake_seed0 +started_utc=2026-07-25T13:48:24Z +hostname=n116-042-186 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_all_46real_91fake_imgseed20260725_n116-042-186_20260725_1215/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_all_46real_91fake_benchmark_eval_20260725_1215/fixed500k_all_46real_91fake_seed0 +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0725 13:48:36.933000 59106 torch/distributed/run.py:766] +W0725 13:48:36.933000 59106 torch/distributed/run.py:766] ***************************************** +W0725 13:48:36.933000 59106 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 13:48:36.933000 59106 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 13:49:04.192451516 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 13:49:04.265115303 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 13:49:04.502762785 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank4]:[W725 13:49:04.502825110 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +[rank0]:[W725 13:49:04.830214120 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 13:49:04.834148580 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 13:49:04.987907189 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 13:49:04.014187037 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_all_46real_91fake_benchmark_eval_20260725_1215/fixed500k_all_46real_91fake_seed0/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_all_46real_91fake_imgseed20260725_n116-042-186_20260725_1215/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:25:38 acc1_pos: 67.969 acc5_pos: 100.000 loss: 0.6078 (0.6078) acc1: 67.9688 (67.9688) acc5: 100.0000 (100.0000) time: 6.5753 data: 4.9528 max mem: 3531 +Test: [ 10/234] eta: 0:02:32 acc1_pos: 65.365 acc5_pos: 100.000 loss: 0.6591 (0.6566) acc1: 65.3646 (61.6241) acc5: 100.0000 (100.0000) time: 0.6819 data: 0.4507 max mem: 3531 +Test: [ 20/234] eta: 0:01:26 acc1_pos: 96.354 acc5_pos: 100.000 loss: 0.6677 (0.6455) acc1: 55.7292 (60.6275) acc5: 100.0000 (100.0000) time: 0.0963 data: 0.0004 max mem: 3531 +Test: [ 30/234] eta: 0:01:11 acc1_pos: 84.635 acc5_pos: 100.000 loss: 0.2048 (0.5084) acc1: 95.5729 (71.7154) acc5: 100.0000 (100.0000) time: 0.1707 data: 0.0737 max mem: 3531 +Test: [ 40/234] eta: 0:01:02 acc1_pos: 99.219 acc5_pos: 100.000 loss: 0.2021 (0.4656) acc1: 95.8333 (75.6733) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1458 max mem: 3531 +Test: [ 50/234] eta: 0:00:56 acc1_pos: 99.219 acc5_pos: 100.000 loss: 0.1599 (0.4043) acc1: 97.9167 (80.1113) acc5: 100.0000 (100.0000) time: 0.2311 data: 0.1386 max mem: 3531 +Test: [ 60/234] eta: 0:00:50 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.1057 (0.3494) acc1: 99.2188 (83.3461) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2228 data: 0.1304 max mem: 3531 +Test: [ 70/234] eta: 0:00:46 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.0680 (0.3665) acc1: 100.0000 (82.1523) acc5: 100.0000 (100.0000) acc1_neg: 59.896 acc5_neg: 100.000 time: 0.2196 data: 0.1273 max mem: 3531 +Test: [ 80/234] eta: 0:00:40 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.6405 (0.3822) acc1: 63.2812 (81.3111) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1897 data: 0.0986 max mem: 3531 +Test: [ 90/234] eta: 0:00:37 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.2154 (0.3605) acc1: 92.7083 (82.8154) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.1937 data: 0.1026 max mem: 3531 +Test: [100/234] eta: 0:00:34 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.2233 (0.3662) acc1: 88.8021 (82.4928) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2275 data: 0.1369 max mem: 3531 +Test: [110/234] eta: 0:00:31 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.4363 (0.3692) acc1: 77.8646 (82.5052) acc5: 100.0000 (100.0000) acc1_neg: 72.135 acc5_neg: 100.000 time: 0.2009 data: 0.1092 max mem: 3531 +Test: [120/234] eta: 0:00:27 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.4326 (0.3790) acc1: 79.1667 (82.1259) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1596 data: 0.0672 max mem: 3531 +Test: [130/234] eta: 0:00:24 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.4317 (0.3824) acc1: 83.0729 (82.3851) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.1715 data: 0.0799 max mem: 3531 +Test: [140/234] eta: 0:00:22 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.4425 (0.4093) acc1: 81.7708 (81.0136) acc5: 100.0000 (100.0000) acc1_neg: 35.417 acc5_neg: 100.000 time: 0.1913 data: 0.0997 max mem: 3531 +Test: [150/234] eta: 0:00:19 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.4493 (0.4170) acc1: 76.0417 (80.5722) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.1796 data: 0.0871 max mem: 3531 +Test: [160/234] eta: 0:00:16 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.4108 (0.4161) acc1: 80.4688 (80.8521) acc5: 100.0000 (100.0000) acc1_neg: 77.344 acc5_neg: 100.000 time: 0.1785 data: 0.0857 max mem: 3531 +Test: [170/234] eta: 0:00:14 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.4775 (0.4199) acc1: 79.6875 (80.7429) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.1731 data: 0.0801 max mem: 3531 +Test: [180/234] eta: 0:00:11 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.4899 (0.4300) acc1: 76.8229 (80.0731) acc5: 100.0000 (100.0000) acc1_neg: 38.802 acc5_neg: 100.000 time: 0.1567 data: 0.0648 max mem: 3531 +Test: [190/234] eta: 0:00:09 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.7002 (0.4456) acc1: 60.9375 (78.9676) acc5: 100.0000 (100.0000) acc1_neg: 59.635 acc5_neg: 100.000 time: 0.1653 data: 0.0752 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.7002 (0.4396) acc1: 61.1979 (79.2638) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1735 data: 0.0834 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.1001 (0.4241) acc1: 100.0000 (80.2231) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1804 data: 0.0897 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.1370 (0.4123) acc1: 99.2188 (80.9837) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.1561 data: 0.0653 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.3535 (0.4282) acc1: 52.3438 (79.6041) acc5: 100.0000 (100.0000) acc1_neg: 51.302 acc5_neg: 100.000 time: 0.1194 data: 0.0294 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.7540 (0.4325) acc1: 51.3021 (79.2509) acc5: 100.0000 (100.0000) acc1_neg: 51.362 acc5_neg: 100.000 time: 0.1227 data: 0.0317 max mem: 3532 +Test: Total time: 0:00:48 (0.2054 s / it) +* Overall Acc@1 79.251 Acc@5 100.000 + target == real Acc@1 80.232 Acc@5 100.000 + target == fake Acc@1 78.974 Acc@5 100.000 + loss 0.433 + PR-AUC 0.9672 ROC-AUC 0.8892 EER 20.55% (thr=0.4927) +Accuracy of the network on the 89729 test images: 79.3% +[rank0]:[W725 13:50:01.378611758 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 13:50:20.054000 59704 torch/distributed/run.py:766] +W0725 13:50:20.054000 59704 torch/distributed/run.py:766] ***************************************** +W0725 13:50:20.054000 59704 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 13:50:20.054000 59704 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 13:50:49.017869383 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 13:50:50.046254439 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 13:50:50.481626565 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 13:50:50.649530355 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 13:50:50.661758424 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 13:50:50.670923850 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W725 13:50:50.780830075 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W725 13:50:50.785797137 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_all_46real_91fake_benchmark_eval_20260725_1215/fixed500k_all_46real_91fake_seed0/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_all_46real_91fake_imgseed20260725_n116-042-186_20260725_1215/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:23:17 acc1_pos: 39.844 acc5_pos: 100.000 loss: 0.8335 (0.8335) acc1: 39.8438 (39.8438) acc5: 100.0000 (100.0000) time: 6.6181 data: 5.1644 max mem: 3531 +Test: [ 10/2387] eta: 0:27:04 acc1_pos: 19.010 acc5_pos: 100.000 loss: 0.9475 (0.9517) acc1: 29.9479 (30.6581) acc5: 100.0000 (100.0000) time: 0.6834 data: 0.4698 max mem: 3531 +Test: [ 20/2387] eta: 0:15:54 acc1_pos: 17.969 acc5_pos: 100.000 loss: 0.9475 (0.9605) acc1: 28.6458 (30.2951) acc5: 100.0000 (100.0000) time: 0.0924 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:12:55 acc1_pos: 8.854 acc5_pos: 100.000 loss: 1.0121 (0.9747) acc1: 22.9167 (28.6290) acc5: 100.0000 (100.0000) time: 0.1338 data: 0.0409 max mem: 3531 +Test: [ 40/2387] eta: 0:12:17 acc1_pos: 23.698 acc5_pos: 100.000 loss: 1.0190 (0.9830) acc1: 22.9167 (26.9309) acc5: 100.0000 (100.0000) time: 0.2211 data: 0.1302 max mem: 3531 +Test: [ 50/2387] eta: 0:11:34 acc1_pos: 38.802 acc5_pos: 100.000 loss: 0.9618 (0.9773) acc1: 23.6979 (28.0586) acc5: 100.0000 (100.0000) time: 0.2484 data: 0.1566 max mem: 3531 +Test: [ 60/2387] eta: 0:10:58 acc1_pos: 41.927 acc5_pos: 100.000 loss: 0.9643 (0.9804) acc1: 28.1250 (28.2872) acc5: 100.0000 (100.0000) time: 0.2188 data: 0.1253 max mem: 3531 +Test: [ 70/2387] eta: 0:10:33 acc1_pos: 37.500 acc5_pos: 100.000 loss: 1.0655 (0.9948) acc1: 26.3021 (27.5161) acc5: 100.0000 (100.0000) time: 0.2124 data: 0.1197 max mem: 3531 +Test: [ 80/2387] eta: 0:10:21 acc1_pos: 59.896 acc5_pos: 100.000 loss: 0.9325 (0.9796) acc1: 26.5625 (28.6394) acc5: 100.0000 (100.0000) time: 0.2272 data: 0.1363 max mem: 3531 +Test: [ 90/2387] eta: 0:10:12 acc1_pos: 48.438 acc5_pos: 100.000 loss: 0.8280 (0.9574) acc1: 41.4062 (30.4659) acc5: 100.0000 (100.0000) time: 0.2435 data: 0.1526 max mem: 3531 +Test: [ 100/2387] eta: 0:10:04 acc1_pos: 42.188 acc5_pos: 100.000 loss: 0.7913 (0.9490) acc1: 41.4062 (31.1907) acc5: 100.0000 (100.0000) time: 0.2434 data: 0.1494 max mem: 3531 +Test: [ 110/2387] eta: 0:10:04 acc1_pos: 39.583 acc5_pos: 100.000 loss: 0.7913 (0.9351) acc1: 42.1875 (32.5380) acc5: 100.0000 (100.0000) time: 0.2588 data: 0.1650 max mem: 3531 +Test: [ 120/2387] eta: 0:09:53 acc1_pos: 33.073 acc5_pos: 100.000 loss: 0.8250 (0.9301) acc1: 43.7500 (33.0471) acc5: 100.0000 (100.0000) time: 0.2509 data: 0.1579 max mem: 3531 +Test: [ 130/2387] eta: 0:09:48 acc1_pos: 37.500 acc5_pos: 100.000 loss: 0.8455 (0.9225) acc1: 37.5000 (33.3453) acc5: 100.0000 (100.0000) time: 0.2349 data: 0.1387 max mem: 3531 +Test: [ 140/2387] eta: 0:09:41 acc1_pos: 31.771 acc5_pos: 100.000 loss: 0.8265 (0.9166) acc1: 35.4167 (33.5697) acc5: 100.0000 (100.0000) time: 0.2402 data: 0.1463 max mem: 3531 +Test: [ 150/2387] eta: 0:09:38 acc1_pos: 22.396 acc5_pos: 100.000 loss: 0.8026 (0.9080) acc1: 39.3229 (34.1198) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1510 max mem: 3531 +Test: [ 160/2387] eta: 0:09:33 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.7562 (0.8875) acc1: 45.3125 (36.1090) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2477 data: 0.1501 max mem: 3531 +Test: [ 170/2387] eta: 0:09:16 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2409 (0.8491) acc1: 96.8750 (39.7036) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1949 data: 0.0983 max mem: 3531 +Test: [ 180/2387] eta: 0:09:09 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2286 (0.8148) acc1: 97.9167 (42.9241) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1799 data: 0.0846 max mem: 3531 +Test: [ 190/2387] eta: 0:09:02 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2239 (0.7838) acc1: 97.9167 (45.8088) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2099 data: 0.1130 max mem: 3531 +Test: [ 200/2387] eta: 0:08:51 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2209 (0.7560) acc1: 97.9167 (48.4090) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1899 data: 0.0908 max mem: 3531 +Test: [ 210/2387] eta: 0:08:42 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2279 (0.7312) acc1: 97.9167 (50.7442) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1727 data: 0.0734 max mem: 3532 +Test: [ 220/2387] eta: 0:08:36 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2293 (0.7084) acc1: 98.1771 (52.8893) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1900 data: 0.0948 max mem: 3532 +Test: [ 230/2387] eta: 0:08:27 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2284 (0.6876) acc1: 98.1771 (54.8374) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1872 data: 0.0933 max mem: 3532 +Test: [ 240/2387] eta: 0:08:17 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2352 (0.6703) acc1: 97.1354 (56.5742) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1621 data: 0.0677 max mem: 3532 +Test: [ 250/2387] eta: 0:08:12 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2153 (0.6478) acc1: 97.6562 (58.2887) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1766 data: 0.0821 max mem: 3532 +Test: [ 260/2387] eta: 0:08:08 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.0805 (0.6260) acc1: 100.0000 (59.8869) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2026 data: 0.1046 max mem: 3532 +Test: [ 270/2387] eta: 0:07:59 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.0726 (0.6056) acc1: 100.0000 (61.3661) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1782 data: 0.0815 max mem: 3532 +Test: [ 280/2387] eta: 0:07:53 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.0762 (0.5869) acc1: 100.0000 (62.7410) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1599 data: 0.0649 max mem: 3532 +Test: [ 290/2387] eta: 0:07:49 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.0801 (0.5695) acc1: 100.0000 (64.0213) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0895 max mem: 3532 +Test: [ 300/2387] eta: 0:07:42 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.0760 (0.5532) acc1: 100.0000 (65.2166) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0801 max mem: 3532 +Test: [ 310/2387] eta: 0:07:39 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.0885 (0.5424) acc1: 100.0000 (66.2865) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1849 data: 0.0891 max mem: 3532 +Test: [ 320/2387] eta: 0:07:36 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.3725 (0.5372) acc1: 97.1354 (67.2435) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2149 data: 0.1167 max mem: 3532 +Test: [ 330/2387] eta: 0:07:32 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.3762 (0.5323) acc1: 96.8750 (68.1576) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1959 data: 0.0968 max mem: 3532 +Test: [ 340/2387] eta: 0:07:28 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.3729 (0.5276) acc1: 97.1354 (69.0058) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1849 data: 0.0876 max mem: 3532 +Test: [ 350/2387] eta: 0:07:24 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.3666 (0.5180) acc1: 98.4375 (69.8785) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1939 data: 0.0975 max mem: 3532 +Test: [ 360/2387] eta: 0:07:20 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1797 (0.5088) acc1: 99.4792 (70.6934) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1900 data: 0.0932 max mem: 3532 +Test: [ 370/2387] eta: 0:07:14 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1877 (0.5002) acc1: 99.2188 (71.4581) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1649 data: 0.0681 max mem: 3532 +Test: [ 380/2387] eta: 0:07:07 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1883 (0.4920) acc1: 98.9583 (72.1771) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1398 data: 0.0446 max mem: 3532 +Test: [ 390/2387] eta: 0:07:05 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1887 (0.4842) acc1: 98.9583 (72.8634) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1699 data: 0.0730 max mem: 3532 +Test: [ 400/2387] eta: 0:07:02 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1862 (0.4769) acc1: 98.9583 (73.5180) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2054 data: 0.1099 max mem: 3532 +Test: [ 410/2387] eta: 0:06:56 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1856 (0.4697) acc1: 99.2188 (74.1459) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1599 data: 0.0654 max mem: 3532 +Test: [ 420/2387] eta: 0:06:51 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1831 (0.4630) acc1: 99.2188 (74.7371) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1394 data: 0.0442 max mem: 3532 +Test: [ 430/2387] eta: 0:06:49 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1891 (0.4573) acc1: 98.6979 (75.2900) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1853 data: 0.0926 max mem: 3532 +Test: [ 440/2387] eta: 0:06:47 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1941 (0.4501) acc1: 98.6979 (75.8397) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2099 data: 0.1159 max mem: 3533 +Test: [ 450/2387] eta: 0:06:45 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.0861 (0.4420) acc1: 100.0000 (76.3754) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2095 data: 0.1155 max mem: 3533 +Test: [ 460/2387] eta: 0:06:42 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.0852 (0.4343) acc1: 100.0000 (76.8879) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1958 data: 0.1011 max mem: 3533 +Test: [ 470/2387] eta: 0:06:39 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.0855 (0.4269) acc1: 100.0000 (77.3786) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1816 data: 0.0877 max mem: 3533 +Test: [ 480/2387] eta: 0:06:35 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.0860 (0.4198) acc1: 100.0000 (77.8489) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1766 data: 0.0852 max mem: 3533 +Test: [ 490/2387] eta: 0:06:31 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.0864 (0.4130) acc1: 100.0000 (78.3000) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1656 data: 0.0729 max mem: 3533 +Test: [ 500/2387] eta: 0:06:28 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.0866 (0.4065) acc1: 100.0000 (78.7332) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1624 data: 0.0693 max mem: 3533 +Test: [ 510/2387] eta: 0:06:25 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.0869 (0.4006) acc1: 100.0000 (79.1493) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1730 data: 0.0814 max mem: 3533 +Test: [ 520/2387] eta: 0:06:22 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1329 (0.3957) acc1: 100.0000 (79.5495) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1807 data: 0.0896 max mem: 3533 +Test: [ 530/2387] eta: 0:06:18 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1319 (0.3908) acc1: 100.0000 (79.9347) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1748 data: 0.0835 max mem: 3533 +Test: [ 540/2387] eta: 0:06:16 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1310 (0.3859) acc1: 100.0000 (80.3056) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1752 data: 0.0817 max mem: 3533 +Test: [ 550/2387] eta: 0:06:13 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1300 (0.3813) acc1: 100.0000 (80.6630) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1796 data: 0.0853 max mem: 3533 +Test: [ 560/2387] eta: 0:06:10 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1300 (0.3768) acc1: 100.0000 (81.0077) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1742 data: 0.0802 max mem: 3533 +Test: [ 570/2387] eta: 0:06:07 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1298 (0.3725) acc1: 100.0000 (81.3403) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0857 max mem: 3533 +Test: [ 580/2387] eta: 0:06:03 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1292 (0.3683) acc1: 100.0000 (81.6615) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1635 data: 0.0707 max mem: 3533 +Test: [ 590/2387] eta: 0:05:59 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1305 (0.3643) acc1: 100.0000 (81.9718) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1401 data: 0.0482 max mem: 3533 +Test: [ 600/2387] eta: 0:05:56 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1355 (0.3608) acc1: 100.0000 (82.2717) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1552 data: 0.0619 max mem: 3533 +Test: [ 610/2387] eta: 0:05:53 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1630 (0.3580) acc1: 100.0000 (82.5563) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1644 data: 0.0696 max mem: 3533 +Test: [ 620/2387] eta: 0:05:50 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1823 (0.3552) acc1: 99.4792 (82.8297) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1545 data: 0.0601 max mem: 3533 +Test: [ 630/2387] eta: 0:05:47 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1848 (0.3525) acc1: 99.4792 (83.0944) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1573 data: 0.0632 max mem: 3533 +Test: [ 640/2387] eta: 0:05:44 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1856 (0.3499) acc1: 99.4792 (83.3480) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1715 data: 0.0790 max mem: 3533 +Test: [ 650/2387] eta: 0:05:41 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1879 (0.3474) acc1: 99.4792 (83.5958) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1575 data: 0.0647 max mem: 3533 +Test: [ 660/2387] eta: 0:05:38 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1867 (0.3450) acc1: 99.4792 (83.8384) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1502 data: 0.0570 max mem: 3533 +Test: [ 670/2387] eta: 0:05:35 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1855 (0.3426) acc1: 99.7396 (84.0731) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1651 data: 0.0726 max mem: 3534 +Test: [ 680/2387] eta: 0:05:32 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1886 (0.3405) acc1: 99.7396 (84.3046) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1651 data: 0.0703 max mem: 3534 +Test: [ 690/2387] eta: 0:05:30 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2069 (0.3410) acc1: 99.7396 (84.3057) acc5: 100.0000 (100.0000) acc1_neg: 76.042 acc5_neg: 100.000 time: 0.1591 data: 0.0618 max mem: 3534 +Test: [ 700/2387] eta: 0:05:28 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.4295 (0.3423) acc1: 79.1667 (84.2364) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.1878 data: 0.0897 max mem: 3534 +Test: [ 710/2387] eta: 0:05:27 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.4320 (0.3437) acc1: 79.4271 (84.1670) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.2199 data: 0.1212 max mem: 3534 +Test: [ 720/2387] eta: 0:05:23 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.4319 (0.3448) acc1: 79.4271 (84.1099) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.1599 data: 0.0608 max mem: 3534 +Test: [ 730/2387] eta: 0:05:19 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.4358 (0.3462) acc1: 78.9062 (84.0355) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.1072 data: 0.0088 max mem: 3534 +Test: [ 740/2387] eta: 0:05:17 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.4365 (0.3474) acc1: 78.9062 (83.9691) acc5: 100.0000 (100.0000) acc1_neg: 76.302 acc5_neg: 100.000 time: 0.1591 data: 0.0648 max mem: 3534 +Test: [ 750/2387] eta: 0:05:17 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.4268 (0.3485) acc1: 79.4271 (83.9093) acc5: 100.0000 (100.0000) acc1_neg: 79.427 acc5_neg: 100.000 time: 0.2273 data: 0.1336 max mem: 3534 +Test: [ 760/2387] eta: 0:05:14 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.4240 (0.3489) acc1: 79.4271 (83.9004) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.2110 data: 0.1154 max mem: 3534 +Test: [ 770/2387] eta: 0:05:12 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2999 (0.3477) acc1: 89.8438 (84.0214) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0809 max mem: 3534 +Test: [ 780/2387] eta: 0:05:09 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2095 (0.3457) acc1: 96.3542 (84.1946) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1691 data: 0.0752 max mem: 3534 +Test: [ 790/2387] eta: 0:05:06 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1730 (0.3436) acc1: 97.6562 (84.3444) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1541 data: 0.0583 max mem: 3534 +Test: [ 800/2387] eta: 0:05:04 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1587 (0.3414) acc1: 98.6979 (84.5262) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1594 data: 0.0611 max mem: 3534 +Test: [ 810/2387] eta: 0:05:02 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1685 (0.3394) acc1: 98.9583 (84.7003) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1799 data: 0.0839 max mem: 3534 +Test: [ 820/2387] eta: 0:05:01 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1688 (0.3377) acc1: 98.6979 (84.8549) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1999 data: 0.1054 max mem: 3534 +Test: [ 830/2387] eta: 0:04:58 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2379 (0.3370) acc1: 95.0521 (84.9526) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1812 data: 0.0862 max mem: 3534 +Test: [ 840/2387] eta: 0:04:56 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2671 (0.3360) acc1: 95.5729 (85.0996) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1817 data: 0.0892 max mem: 3534 +Test: [ 850/2387] eta: 0:04:54 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2489 (0.3349) acc1: 97.3958 (85.2346) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1891 data: 0.0980 max mem: 3534 +Test: [ 860/2387] eta: 0:04:52 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2210 (0.3336) acc1: 98.9583 (85.3922) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1762 data: 0.0853 max mem: 3534 +Test: [ 870/2387] eta: 0:04:50 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2165 (0.3324) acc1: 99.2188 (85.5381) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1882 data: 0.0963 max mem: 3534 +Test: [ 880/2387] eta: 0:04:48 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2231 (0.3313) acc1: 98.9583 (85.6777) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1898 data: 0.0987 max mem: 3534 +Test: [ 890/2387] eta: 0:04:46 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2231 (0.3301) acc1: 99.2188 (85.8288) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1882 data: 0.0960 max mem: 3534 +Test: [ 900/2387] eta: 0:04:44 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2431 (0.3330) acc1: 96.8750 (85.6386) acc5: 100.0000 (100.0000) acc1_neg: 65.104 acc5_neg: 100.000 time: 0.1968 data: 0.1043 max mem: 3535 +Test: [ 910/2387] eta: 0:04:42 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.6702 (0.3367) acc1: 60.9375 (85.3704) acc5: 100.0000 (100.0000) acc1_neg: 60.156 acc5_neg: 100.000 time: 0.1902 data: 0.0987 max mem: 3535 +Test: [ 920/2387] eta: 0:04:40 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.6776 (0.3404) acc1: 60.9375 (85.1056) acc5: 100.0000 (100.0000) acc1_neg: 58.854 acc5_neg: 100.000 time: 0.1862 data: 0.0929 max mem: 3535 +Test: [ 930/2387] eta: 0:04:38 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.6784 (0.3439) acc1: 61.4583 (84.8567) acc5: 100.0000 (100.0000) acc1_neg: 57.292 acc5_neg: 100.000 time: 0.1875 data: 0.0938 max mem: 3535 +Test: [ 940/2387] eta: 0:04:37 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.6799 (0.3475) acc1: 60.9375 (84.5986) acc5: 100.0000 (100.0000) acc1_neg: 59.375 acc5_neg: 100.000 time: 0.1997 data: 0.1076 max mem: 3535 +Test: [ 950/2387] eta: 0:04:35 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.6702 (0.3509) acc1: 60.4167 (84.3572) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.1949 data: 0.1006 max mem: 3535 +Test: [ 960/2387] eta: 0:04:33 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.6665 (0.3541) acc1: 62.5000 (84.1306) acc5: 100.0000 (100.0000) acc1_neg: 65.365 acc5_neg: 100.000 time: 0.1851 data: 0.0886 max mem: 3535 +Test: [ 970/2387] eta: 0:04:31 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.6692 (0.3575) acc1: 61.4583 (83.8896) acc5: 100.0000 (100.0000) acc1_neg: 62.240 acc5_neg: 100.000 time: 0.1972 data: 0.1035 max mem: 3535 +Test: [ 980/2387] eta: 0:04:29 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.6467 (0.3578) acc1: 63.0208 (83.8863) acc5: 100.0000 (100.0000) acc1_neg: 80.729 acc5_neg: 100.000 time: 0.1839 data: 0.0929 max mem: 3535 +Test: [ 990/2387] eta: 0:04:27 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2182 (0.3564) acc1: 95.5729 (84.0063) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1825 data: 0.0907 max mem: 3535 +Test: [1000/2387] eta: 0:04:25 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1847 (0.3547) acc1: 98.9583 (84.1393) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1867 data: 0.0924 max mem: 3535 +Test: [1010/2387] eta: 0:04:23 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1929 (0.3534) acc1: 97.9167 (84.2588) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1819 data: 0.0891 max mem: 3535 +Test: [1020/2387] eta: 0:04:20 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2095 (0.3522) acc1: 97.6562 (84.3714) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1683 data: 0.0775 max mem: 3535 +Test: [1030/2387] eta: 0:04:18 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1917 (0.3505) acc1: 99.2188 (84.5111) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1632 data: 0.0718 max mem: 3535 +Test: [1040/2387] eta: 0:04:17 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1482 (0.3485) acc1: 99.4792 (84.6539) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1908 data: 0.0974 max mem: 3535 +Test: [1050/2387] eta: 0:04:14 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1376 (0.3466) acc1: 99.7396 (84.7950) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1912 data: 0.0979 max mem: 3535 +Test: [1060/2387] eta: 0:04:13 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1456 (0.3449) acc1: 99.7396 (84.9307) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1963 data: 0.1029 max mem: 3535 +Test: [1070/2387] eta: 0:04:11 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1551 (0.3435) acc1: 98.9583 (85.0534) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1948 data: 0.1006 max mem: 3535 +Test: [1080/2387] eta: 0:04:09 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2207 (0.3425) acc1: 96.8750 (85.1536) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1771 data: 0.0847 max mem: 3535 +Test: [1090/2387] eta: 0:04:07 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2172 (0.3412) acc1: 96.6146 (85.2601) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1793 data: 0.0854 max mem: 3535 +Test: [1100/2387] eta: 0:04:04 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1843 (0.3398) acc1: 97.6562 (85.3762) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1663 data: 0.0710 max mem: 3535 +Test: [1110/2387] eta: 0:04:02 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1779 (0.3387) acc1: 97.3958 (85.4645) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1615 data: 0.0691 max mem: 3535 +Test: [1120/2387] eta: 0:04:00 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1983 (0.3375) acc1: 96.6146 (85.5642) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1635 data: 0.0705 max mem: 3536 +Test: [1130/2387] eta: 0:03:58 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1875 (0.3362) acc1: 97.3958 (85.6743) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1702 data: 0.0757 max mem: 3536 +Test: [1140/2387] eta: 0:03:56 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2131 (0.3353) acc1: 97.3958 (85.7711) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1669 data: 0.0739 max mem: 3536 +Test: [1150/2387] eta: 0:03:53 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2329 (0.3344) acc1: 98.4375 (85.8714) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1588 data: 0.0660 max mem: 3536 +Test: [1160/2387] eta: 0:03:51 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1861 (0.3330) acc1: 99.2188 (85.9848) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1629 data: 0.0704 max mem: 3536 +Test: [1170/2387] eta: 0:03:49 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1644 (0.3315) acc1: 100.0000 (86.1032) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1653 data: 0.0699 max mem: 3536 +Test: [1180/2387] eta: 0:03:47 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1640 (0.3301) acc1: 99.7396 (86.2145) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1527 data: 0.0567 max mem: 3536 +Test: [1190/2387] eta: 0:03:45 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1594 (0.3287) acc1: 99.4792 (86.3274) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1528 data: 0.0604 max mem: 3536 +Test: [1200/2387] eta: 0:03:42 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1557 (0.3273) acc1: 100.0000 (86.4410) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1630 data: 0.0712 max mem: 3536 +Test: [1210/2387] eta: 0:03:40 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1560 (0.3259) acc1: 100.0000 (86.5525) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1583 data: 0.0641 max mem: 3536 +Test: [1220/2387] eta: 0:03:38 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1101 (0.3240) acc1: 100.0000 (86.6618) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1615 data: 0.0671 max mem: 3536 +Test: [1230/2387] eta: 0:03:36 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.0863 (0.3220) acc1: 100.0000 (86.7702) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1665 data: 0.0736 max mem: 3536 +Test: [1240/2387] eta: 0:03:34 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.0795 (0.3201) acc1: 100.0000 (86.8768) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1775 data: 0.0815 max mem: 3536 +Test: [1250/2387] eta: 0:03:32 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.0809 (0.3182) acc1: 100.0000 (86.9812) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1703 data: 0.0719 max mem: 3536 +Test: [1260/2387] eta: 0:03:30 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.0850 (0.3164) acc1: 100.0000 (87.0845) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1364 data: 0.0383 max mem: 3536 +Test: [1270/2387] eta: 0:03:28 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.0831 (0.3146) acc1: 100.0000 (87.1859) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1610 data: 0.0650 max mem: 3536 +Test: [1280/2387] eta: 0:03:26 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.0947 (0.3131) acc1: 100.0000 (87.2831) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1990 data: 0.1040 max mem: 3536 +Test: [1290/2387] eta: 0:03:24 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1203 (0.3116) acc1: 99.7396 (87.3703) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1710 data: 0.0743 max mem: 3536 +Test: [1300/2387] eta: 0:03:22 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1203 (0.3102) acc1: 99.4792 (87.4596) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1353 data: 0.0414 max mem: 3536 +Test: [1310/2387] eta: 0:03:20 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1202 (0.3088) acc1: 99.4792 (87.5471) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1927 data: 0.1015 max mem: 3536 +Test: [1320/2387] eta: 0:03:18 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1330 (0.3075) acc1: 99.4792 (87.6321) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2117 data: 0.1186 max mem: 3536 +Test: [1330/2387] eta: 0:03:16 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1455 (0.3065) acc1: 99.7396 (87.7162) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1470 data: 0.0539 max mem: 3536 +Test: [1340/2387] eta: 0:03:14 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1891 (0.3065) acc1: 98.1771 (87.7245) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1380 data: 0.0444 max mem: 3536 +Test: [1350/2387] eta: 0:03:12 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.0817 (0.3047) acc1: 100.0000 (87.8154) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1744 data: 0.0813 max mem: 3537 +Test: [1360/2387] eta: 0:03:10 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.0608 (0.3029) acc1: 100.0000 (87.9049) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1941 data: 0.1030 max mem: 3537 +Test: [1370/2387] eta: 0:03:08 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.0561 (0.3011) acc1: 100.0000 (87.9931) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1730 data: 0.0798 max mem: 3537 +Test: [1380/2387] eta: 0:03:06 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.0591 (0.2994) acc1: 100.0000 (88.0800) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1551 data: 0.0604 max mem: 3537 +Test: [1390/2387] eta: 0:03:05 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.0610 (0.2977) acc1: 100.0000 (88.1657) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1915 data: 0.0985 max mem: 3537 +Test: [1400/2387] eta: 0:03:03 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.0622 (0.2960) acc1: 100.0000 (88.2502) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2095 data: 0.1155 max mem: 3537 +Test: [1410/2387] eta: 0:03:01 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.0699 (0.2951) acc1: 100.0000 (88.3156) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1957 data: 0.1022 max mem: 3537 +Test: [1420/2387] eta: 0:02:59 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1867 (0.2945) acc1: 97.6562 (88.3657) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1853 data: 0.0931 max mem: 3537 +Test: [1430/2387] eta: 0:02:57 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1973 (0.2939) acc1: 97.3958 (88.4270) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1892 data: 0.0964 max mem: 3537 +Test: [1440/2387] eta: 0:02:55 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1973 (0.2933) acc1: 97.6562 (88.4927) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1903 data: 0.0984 max mem: 3537 +Test: [1450/2387] eta: 0:02:54 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2056 (0.2928) acc1: 97.9167 (88.5445) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1882 data: 0.0960 max mem: 3537 +Test: [1460/2387] eta: 0:02:52 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1668 (0.2919) acc1: 98.1771 (88.6131) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2001 data: 0.1086 max mem: 3537 +Test: [1470/2387] eta: 0:02:50 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1357 (0.2908) acc1: 99.7396 (88.6895) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1847 data: 0.0919 max mem: 3537 +Test: [1480/2387] eta: 0:02:48 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1282 (0.2897) acc1: 99.7396 (88.7606) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1672 data: 0.0741 max mem: 3537 +Test: [1490/2387] eta: 0:02:46 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1290 (0.2887) acc1: 99.7396 (88.8333) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1634 data: 0.0727 max mem: 3537 +Test: [1500/2387] eta: 0:02:44 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1340 (0.2877) acc1: 99.7396 (88.9044) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1816 data: 0.0908 max mem: 3537 +Test: [1510/2387] eta: 0:02:42 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1462 (0.2869) acc1: 100.0000 (88.9777) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1924 data: 0.1002 max mem: 3537 +Test: [1520/2387] eta: 0:02:40 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1822 (0.2864) acc1: 100.0000 (89.0418) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1836 data: 0.0908 max mem: 3537 +Test: [1530/2387] eta: 0:02:38 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2143 (0.2859) acc1: 99.2188 (89.0986) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1807 data: 0.0874 max mem: 3537 +Test: [1540/2387] eta: 0:02:37 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1907 (0.2853) acc1: 99.2188 (89.1631) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2085 data: 0.1139 max mem: 3537 +Test: [1550/2387] eta: 0:02:35 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1931 (0.2847) acc1: 99.4792 (89.2272) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2180 data: 0.1249 max mem: 3537 +Test: [1560/2387] eta: 0:02:33 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1935 (0.2842) acc1: 99.4792 (89.2884) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1888 data: 0.0938 max mem: 3537 +Test: [1570/2387] eta: 0:02:31 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1849 (0.2836) acc1: 99.4792 (89.3533) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1486 data: 0.0524 max mem: 3537 +Test: [1580/2387] eta: 0:02:29 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1955 (0.2831) acc1: 99.2188 (89.4091) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1585 data: 0.0647 max mem: 3538 +Test: [1590/2387] eta: 0:02:27 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1816 (0.2824) acc1: 98.4375 (89.4689) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1874 data: 0.0914 max mem: 3538 +Test: [1600/2387] eta: 0:02:26 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1585 (0.2817) acc1: 98.4375 (89.5090) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1912 data: 0.0949 max mem: 3538 +Test: [1610/2387] eta: 0:02:24 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1585 (0.2810) acc1: 99.2188 (89.5701) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1994 data: 0.1061 max mem: 3538 +Test: [1620/2387] eta: 0:02:22 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1696 (0.2803) acc1: 98.6979 (89.6174) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2006 data: 0.1072 max mem: 3538 +Test: [1630/2387] eta: 0:02:20 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1950 (0.2800) acc1: 97.3958 (89.6558) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2019 data: 0.1090 max mem: 3538 +Test: [1640/2387] eta: 0:02:18 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2425 (0.2799) acc1: 95.0521 (89.6832) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1896 data: 0.0988 max mem: 3538 +Test: [1650/2387] eta: 0:02:16 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2419 (0.2797) acc1: 95.0521 (89.7149) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1855 data: 0.0952 max mem: 3538 +Test: [1660/2387] eta: 0:02:15 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2361 (0.2796) acc1: 95.0521 (89.7404) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1845 data: 0.0935 max mem: 3538 +Test: [1670/2387] eta: 0:02:13 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2501 (0.2793) acc1: 94.0104 (89.7707) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1767 data: 0.0851 max mem: 3538 +Test: [1680/2387] eta: 0:02:11 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2462 (0.2792) acc1: 95.8333 (89.8005) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1893 data: 0.0973 max mem: 3538 +Test: [1690/2387] eta: 0:02:09 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2462 (0.2791) acc1: 96.0938 (89.8276) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2197 data: 0.1260 max mem: 3538 +Test: [1700/2387] eta: 0:02:07 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2458 (0.2787) acc1: 96.0938 (89.8675) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1892 data: 0.0946 max mem: 3538 +Test: [1710/2387] eta: 0:02:06 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1126 (0.2777) acc1: 99.7396 (89.9262) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1878 data: 0.0932 max mem: 3538 +Test: [1720/2387] eta: 0:02:04 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1036 (0.2766) acc1: 100.0000 (89.9843) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2119 data: 0.1187 max mem: 3538 +Test: [1730/2387] eta: 0:02:02 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.0949 (0.2756) acc1: 100.0000 (90.0419) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1891 data: 0.0951 max mem: 3538 +Test: [1740/2387] eta: 0:02:00 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.0947 (0.2746) acc1: 100.0000 (90.0985) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1842 data: 0.0907 max mem: 3538 +Test: [1750/2387] eta: 0:01:58 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1003 (0.2736) acc1: 100.0000 (90.1547) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2032 data: 0.1108 max mem: 3538 +Test: [1760/2387] eta: 0:01:56 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1029 (0.2727) acc1: 100.0000 (90.2105) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2186 data: 0.1233 max mem: 3538 +Test: [1770/2387] eta: 0:01:54 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1264 (0.2720) acc1: 100.0000 (90.2611) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1767 data: 0.0826 max mem: 3538 +Test: [1780/2387] eta: 0:01:52 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1390 (0.2712) acc1: 99.7396 (90.3146) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1520 data: 0.0608 max mem: 3538 +Test: [1790/2387] eta: 0:01:51 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1333 (0.2704) acc1: 100.0000 (90.3681) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1681 data: 0.0777 max mem: 3538 +Test: [1800/2387] eta: 0:01:49 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1238 (0.2696) acc1: 100.0000 (90.4204) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1878 data: 0.0946 max mem: 3538 +Test: [1810/2387] eta: 0:01:47 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1226 (0.2687) acc1: 100.0000 (90.4723) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2000 data: 0.1051 max mem: 3539 +Test: [1820/2387] eta: 0:01:45 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1194 (0.2679) acc1: 100.0000 (90.5245) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2049 data: 0.1102 max mem: 3539 +Test: [1830/2387] eta: 0:01:43 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1194 (0.2671) acc1: 100.0000 (90.5761) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1923 data: 0.0973 max mem: 3539 +Test: [1840/2387] eta: 0:01:41 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1153 (0.2663) acc1: 100.0000 (90.6273) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1830 data: 0.0888 max mem: 3539 +Test: [1850/2387] eta: 0:01:40 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1156 (0.2655) acc1: 100.0000 (90.6778) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1969 data: 0.1032 max mem: 3539 +Test: [1860/2387] eta: 0:01:38 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1152 (0.2647) acc1: 100.0000 (90.7279) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2031 data: 0.1108 max mem: 3539 +Test: [1870/2387] eta: 0:01:36 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1149 (0.2639) acc1: 100.0000 (90.7774) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1987 data: 0.1069 max mem: 3539 +Test: [1880/2387] eta: 0:01:34 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1149 (0.2631) acc1: 100.0000 (90.8264) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1913 data: 0.1001 max mem: 3539 +Test: [1890/2387] eta: 0:01:32 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1144 (0.2623) acc1: 100.0000 (90.8750) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1978 data: 0.1074 max mem: 3539 +Test: [1900/2387] eta: 0:01:30 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1164 (0.2616) acc1: 100.0000 (90.9230) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2056 data: 0.1129 max mem: 3539 +Test: [1910/2387] eta: 0:01:29 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1333 (0.2609) acc1: 100.0000 (90.9690) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1834 data: 0.0902 max mem: 3539 +Test: [1920/2387] eta: 0:01:27 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1601 (0.2608) acc1: 99.7396 (90.9822) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1753 data: 0.0813 max mem: 3539 +Test: [1930/2387] eta: 0:01:25 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1662 (0.2602) acc1: 99.7396 (91.0262) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1802 data: 0.0857 max mem: 3539 +Test: [1940/2387] eta: 0:01:23 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1630 (0.2600) acc1: 99.4792 (91.0346) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1771 data: 0.0845 max mem: 3539 +Test: [1950/2387] eta: 0:01:21 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1838 (0.2596) acc1: 98.6979 (91.0672) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1865 data: 0.0942 max mem: 3539 +Test: [1960/2387] eta: 0:01:19 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1916 (0.2594) acc1: 98.4375 (91.1002) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1802 data: 0.0891 max mem: 3539 +Test: [1970/2387] eta: 0:01:17 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2349 (0.2595) acc1: 95.5729 (91.0884) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.1601 data: 0.0675 max mem: 3539 +Test: [1980/2387] eta: 0:01:15 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2529 (0.2593) acc1: 94.0104 (91.1117) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1746 data: 0.0816 max mem: 3539 +Test: [1990/2387] eta: 0:01:13 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1241 (0.2586) acc1: 98.9583 (91.1515) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1832 data: 0.0915 max mem: 3539 +Test: [2000/2387] eta: 0:01:12 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1253 (0.2580) acc1: 99.2188 (91.1911) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1703 data: 0.0788 max mem: 3539 +Test: [2010/2387] eta: 0:01:10 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1265 (0.2573) acc1: 99.2188 (91.2303) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1751 data: 0.0829 max mem: 3539 +Test: [2020/2387] eta: 0:01:08 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1259 (0.2567) acc1: 98.9583 (91.2691) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1734 data: 0.0793 max mem: 3539 +Test: [2030/2387] eta: 0:01:06 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1246 (0.2560) acc1: 98.9583 (91.3079) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1741 data: 0.0775 max mem: 3540 +Test: [2040/2387] eta: 0:01:04 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1242 (0.2554) acc1: 99.2188 (91.3469) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1863 data: 0.0896 max mem: 3540 +Test: [2050/2387] eta: 0:01:02 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1242 (0.2547) acc1: 99.2188 (91.3858) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1815 data: 0.0886 max mem: 3540 +Test: [2060/2387] eta: 0:01:00 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1265 (0.2541) acc1: 98.9583 (91.4220) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1788 data: 0.0879 max mem: 3540 +Test: [2070/2387] eta: 0:00:58 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1286 (0.2535) acc1: 98.9583 (91.4608) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1881 data: 0.0953 max mem: 3540 +Test: [2080/2387] eta: 0:00:57 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1273 (0.2529) acc1: 99.2188 (91.4977) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1745 data: 0.0799 max mem: 3540 +Test: [2090/2387] eta: 0:00:55 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1278 (0.2523) acc1: 99.2188 (91.5348) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1749 data: 0.0787 max mem: 3540 +Test: [2100/2387] eta: 0:00:53 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1278 (0.2517) acc1: 99.2188 (91.5718) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1799 data: 0.0857 max mem: 3540 +Test: [2110/2387] eta: 0:00:51 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1255 (0.2511) acc1: 99.2188 (91.6084) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1714 data: 0.0791 max mem: 3540 +Test: [2120/2387] eta: 0:00:49 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1392 (0.2509) acc1: 98.9583 (91.6338) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1827 data: 0.0898 max mem: 3540 +Test: [2130/2387] eta: 0:00:47 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1641 (0.2504) acc1: 98.4375 (91.6690) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1847 data: 0.0911 max mem: 3540 +Test: [2140/2387] eta: 0:00:45 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1609 (0.2502) acc1: 99.2188 (91.6850) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1835 data: 0.0901 max mem: 3540 +Test: [2150/2387] eta: 0:00:44 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1724 (0.2498) acc1: 98.9583 (91.7192) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1957 data: 0.1042 max mem: 3540 +Test: [2160/2387] eta: 0:00:42 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1607 (0.2495) acc1: 99.2188 (91.7499) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1834 data: 0.0912 max mem: 3540 +Test: [2170/2387] eta: 0:00:40 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1948 (0.2494) acc1: 96.8750 (91.7683) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1880 data: 0.0930 max mem: 3540 +Test: [2180/2387] eta: 0:00:38 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2682 (0.2504) acc1: 95.0521 (91.7079) acc5: 100.0000 (100.0000) acc1_neg: 53.385 acc5_neg: 100.000 time: 0.1927 data: 0.0981 max mem: 3540 +Test: [2190/2387] eta: 0:00:36 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2495 (0.2505) acc1: 95.8333 (91.7034) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1905 data: 0.0983 max mem: 3540 +Test: [2200/2387] eta: 0:00:34 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1683 (0.2501) acc1: 98.1771 (91.7372) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1921 data: 0.1008 max mem: 3540 +Test: [2210/2387] eta: 0:00:32 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1469 (0.2497) acc1: 99.7396 (91.7726) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1846 data: 0.0931 max mem: 3540 +Test: [2220/2387] eta: 0:00:31 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1535 (0.2493) acc1: 99.7396 (91.8067) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1953 data: 0.1022 max mem: 3540 +Test: [2230/2387] eta: 0:00:29 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1427 (0.2488) acc1: 99.7396 (91.8419) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1998 data: 0.1053 max mem: 3540 +Test: [2240/2387] eta: 0:00:27 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1406 (0.2483) acc1: 99.4792 (91.8741) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2063 data: 0.1125 max mem: 3540 +Test: [2250/2387] eta: 0:00:25 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1543 (0.2480) acc1: 99.7396 (91.9080) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1944 data: 0.0976 max mem: 3540 +Test: [2260/2387] eta: 0:00:23 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1703 (0.2477) acc1: 99.7396 (91.9416) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2135 data: 0.1174 max mem: 3541 +Test: [2270/2387] eta: 0:00:21 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1429 (0.2472) acc1: 100.0000 (91.9769) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2166 data: 0.1236 max mem: 3541 +Test: [2280/2387] eta: 0:00:19 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1239 (0.2468) acc1: 100.0000 (91.9958) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1796 data: 0.0863 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1104 (0.2462) acc1: 100.0000 (92.0300) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1601 data: 0.0675 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1399 (0.2458) acc1: 100.0000 (92.0618) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1564 data: 0.0623 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1465 (0.2455) acc1: 99.7396 (92.0885) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1831 data: 0.0893 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1970 (0.2454) acc1: 97.6562 (92.1058) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1987 data: 0.1059 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2467 (0.2454) acc1: 96.6146 (92.1196) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1762 data: 0.0839 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2158 (0.2453) acc1: 97.1354 (92.1394) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1622 data: 0.0675 max mem: 3541 +Test: [2350/2387] eta: 0:00:06 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1974 (0.2454) acc1: 97.1354 (92.1409) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1984 data: 0.1022 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2395 (0.2455) acc1: 95.5729 (92.1488) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1881 data: 0.0949 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2408 (0.2455) acc1: 93.2292 (92.1509) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.1699 data: 0.0760 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2603 (0.2457) acc1: 93.2292 (92.1506) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1895 data: 0.0941 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.3027 (0.2458) acc1: 91.9271 (92.1525) acc5: 100.0000 (100.0000) acc1_neg: 93.023 acc5_neg: 100.000 time: 0.1770 data: 0.0846 max mem: 3541 +Test: Total time: 0:07:24 (0.1861 s / it) +* Overall Acc@1 92.153 Acc@5 100.000 + target == real Acc@1 34.494 Acc@5 100.000 + target == fake Acc@1 96.202 Acc@5 100.000 + loss 0.246 + PR-AUC 0.9930 ROC-AUC 0.9096 EER 16.00% (thr=0.7310) +Accuracy of the network on the 916267 test images: 92.2% +[rank0]:[W725 13:58:44.695424695 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 13:59:06.464000 60455 torch/distributed/run.py:766] +W0725 13:59:06.464000 60455 torch/distributed/run.py:766] ***************************************** +W0725 13:59:06.464000 60455 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 13:59:06.464000 60455 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 13:59:40.123211103 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 13:59:40.865210881 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 13:59:41.218459283 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +[rank0]:[W725 13:59:41.434748891 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 13:59:41.443272743 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 13:59:41.480352251 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank1]:[W725 13:59:41.480421844 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 13:59:41.632187787 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_all_46real_91fake_benchmark_eval_20260725_1215/fixed500k_all_46real_91fake_seed0/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_all_46real_91fake_imgseed20260725_n116-042-186_20260725_1215/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:06:21 acc1_pos: 39.844 acc5_pos: 100.000 loss: 0.8335 (0.8335) acc1: 39.8438 (39.8438) acc5: 100.0000 (100.0000) time: 7.7011 data: 6.2422 max mem: 3531 +Test: [ 10/517] eta: 0:06:36 acc1_pos: 19.010 acc5_pos: 100.000 loss: 0.9475 (0.9517) acc1: 29.9479 (30.6581) acc5: 100.0000 (100.0000) time: 0.7824 data: 0.5679 max mem: 3531 +Test: [ 20/517] eta: 0:03:47 acc1_pos: 17.969 acc5_pos: 100.000 loss: 0.9475 (0.9605) acc1: 28.6458 (30.2951) acc5: 100.0000 (100.0000) time: 0.0952 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:03:14 acc1_pos: 8.854 acc5_pos: 100.000 loss: 1.0121 (0.9747) acc1: 22.9167 (28.6290) acc5: 100.0000 (100.0000) time: 0.1899 data: 0.0932 max mem: 3531 +Test: [ 40/517] eta: 0:02:54 acc1_pos: 23.698 acc5_pos: 100.000 loss: 1.0190 (0.9830) acc1: 22.9167 (26.9309) acc5: 100.0000 (100.0000) time: 0.2699 data: 0.1756 max mem: 3531 +Test: [ 50/517] eta: 0:02:39 acc1_pos: 39.908 acc5_pos: 100.000 loss: 0.9595 (0.9553) acc1: 23.6979 (30.4024) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.2499 data: 0.1555 max mem: 3531 +Test: [ 60/517] eta: 0:02:29 acc1_pos: 39.908 acc5_pos: 100.000 loss: 0.4642 (0.8714) acc1: 77.8646 (39.8608) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.2499 data: 0.1547 max mem: 3531 +Test: [ 70/517] eta: 0:02:22 acc1_pos: 39.908 acc5_pos: 100.000 loss: 0.4112 (0.8066) acc1: 91.9271 (47.2198) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.2599 data: 0.1644 max mem: 3531 +Test: [ 80/517] eta: 0:02:16 acc1_pos: 39.908 acc5_pos: 100.000 loss: 0.4186 (0.7616) acc1: 89.3229 (52.3438) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.2649 data: 0.1694 max mem: 3531 +Test: [ 90/517] eta: 0:02:10 acc1_pos: 39.908 acc5_pos: 100.000 loss: 0.4333 (0.7229) acc1: 88.2812 (56.6993) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.2603 data: 0.1667 max mem: 3531 +Test: [100/517] eta: 0:02:07 acc1_pos: 39.908 acc5_pos: 100.000 loss: 0.3845 (0.6886) acc1: 93.7500 (60.5275) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2749 data: 0.1797 max mem: 3531 +Test: [110/517] eta: 0:02:01 acc1_pos: 39.908 acc5_pos: 100.000 loss: 0.3367 (0.6547) acc1: 96.0938 (63.8490) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2696 data: 0.1739 max mem: 3531 +Test: [120/517] eta: 0:01:55 acc1_pos: 39.908 acc5_pos: 100.000 loss: 0.3577 (0.6332) acc1: 96.0938 (66.4106) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2249 data: 0.1326 max mem: 3531 +Test: [130/517] eta: 0:01:53 acc1_pos: 39.908 acc5_pos: 100.000 loss: 0.3892 (0.6133) acc1: 95.5729 (68.6248) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2549 data: 0.1584 max mem: 3531 +Test: [140/517] eta: 0:01:48 acc1_pos: 39.908 acc5_pos: 100.000 loss: 0.3165 (0.5912) acc1: 99.2188 (70.8241) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2649 data: 0.1659 max mem: 3531 +Test: [150/517] eta: 0:01:44 acc1_pos: 39.908 acc5_pos: 100.000 loss: 0.3068 (0.5711) acc1: 99.4792 (72.6976) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2299 data: 0.1319 max mem: 3531 +Test: [160/517] eta: 0:01:40 acc1_pos: 39.908 acc5_pos: 100.000 loss: 0.3085 (0.5549) acc1: 98.1771 (74.2026) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2349 data: 0.1383 max mem: 3531 +Test: [170/517] eta: 0:01:37 acc1_pos: 39.908 acc5_pos: 100.000 loss: 0.3065 (0.5392) acc1: 97.6562 (75.5543) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2499 data: 0.1543 max mem: 3531 +Test: [180/517] eta: 0:01:33 acc1_pos: 39.908 acc5_pos: 100.000 loss: 0.2891 (0.5263) acc1: 97.6562 (76.6531) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2499 data: 0.1556 max mem: 3531 +Test: [190/517] eta: 0:01:30 acc1_pos: 39.908 acc5_pos: 100.000 loss: 0.2387 (0.5113) acc1: 98.1771 (77.8087) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2499 data: 0.1541 max mem: 3531 +Test: [200/517] eta: 0:01:27 acc1_pos: 39.908 acc5_pos: 100.000 loss: 0.2495 (0.5000) acc1: 98.9583 (78.8635) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2600 data: 0.1616 max mem: 3531 +Test: [210/517] eta: 0:01:24 acc1_pos: 39.908 acc5_pos: 100.000 loss: 0.3298 (0.4978) acc1: 95.3125 (79.1482) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.2548 data: 0.1580 max mem: 3532 +Test: [220/517] eta: 0:01:21 acc1_pos: 39.908 acc5_pos: 100.000 loss: 0.4579 (0.4958) acc1: 84.8958 (79.4271) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.2548 data: 0.1579 max mem: 3532 +Test: [230/517] eta: 0:01:18 acc1_pos: 39.908 acc5_pos: 100.000 loss: 0.4244 (0.4921) acc1: 88.0208 (79.9908) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.2599 data: 0.1617 max mem: 3532 +Test: [240/517] eta: 0:01:15 acc1_pos: 39.908 acc5_pos: 100.000 loss: 0.4035 (0.4890) acc1: 94.5312 (80.5584) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.2599 data: 0.1618 max mem: 3532 +Test: [250/517] eta: 0:01:12 acc1_pos: 39.908 acc5_pos: 100.000 loss: 0.3897 (0.4833) acc1: 97.3958 (81.2355) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2599 data: 0.1621 max mem: 3532 +Test: [260/517] eta: 0:01:09 acc1_pos: 39.908 acc5_pos: 100.000 loss: 0.3242 (0.4774) acc1: 99.2188 (81.8876) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.2499 data: 0.1528 max mem: 3532 +Test: [270/517] eta: 0:01:06 acc1_pos: 39.908 acc5_pos: 100.000 loss: 0.3822 (0.4747) acc1: 95.5729 (82.3061) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2449 data: 0.1496 max mem: 3532 +Test: [280/517] eta: 0:01:03 acc1_pos: 39.908 acc5_pos: 100.000 loss: 0.3868 (0.4709) acc1: 93.4896 (82.6930) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2514 data: 0.1558 max mem: 3532 +Test: [290/517] eta: 0:01:01 acc1_pos: 39.908 acc5_pos: 100.000 loss: 0.3530 (0.4663) acc1: 95.8333 (83.1812) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2499 data: 0.1537 max mem: 3532 +Test: [300/517] eta: 0:00:58 acc1_pos: 39.908 acc5_pos: 100.000 loss: 0.3207 (0.4613) acc1: 98.6979 (83.6907) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2484 data: 0.1492 max mem: 3532 +Test: [310/517] eta: 0:00:55 acc1_pos: 39.908 acc5_pos: 100.000 loss: 0.3686 (0.4594) acc1: 97.9167 (84.0133) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.2499 data: 0.1490 max mem: 3532 +Test: [320/517] eta: 0:00:52 acc1_pos: 39.908 acc5_pos: 100.000 loss: 0.3758 (0.4550) acc1: 96.8750 (84.4561) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2499 data: 0.1488 max mem: 3532 +Test: [330/517] eta: 0:00:49 acc1_pos: 39.908 acc5_pos: 100.000 loss: 0.2879 (0.4494) acc1: 99.2188 (84.9069) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2499 data: 0.1469 max mem: 3532 +Test: [340/517] eta: 0:00:47 acc1_pos: 39.908 acc5_pos: 100.000 loss: 0.2655 (0.4462) acc1: 99.2188 (85.1585) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.2399 data: 0.1416 max mem: 3532 +Test: [350/517] eta: 0:00:44 acc1_pos: 39.908 acc5_pos: 100.000 loss: 0.4385 (0.4469) acc1: 90.3646 (85.1933) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.2399 data: 0.1467 max mem: 3532 +Test: [360/517] eta: 0:00:41 acc1_pos: 39.908 acc5_pos: 100.000 loss: 0.4380 (0.4458) acc1: 89.0625 (85.3784) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2499 data: 0.1558 max mem: 3532 +Test: [370/517] eta: 0:00:39 acc1_pos: 39.908 acc5_pos: 100.000 loss: 0.3393 (0.4419) acc1: 97.3958 (85.7634) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2799 data: 0.1858 max mem: 3532 +Test: [380/517] eta: 0:00:36 acc1_pos: 39.908 acc5_pos: 100.000 loss: 0.3053 (0.4388) acc1: 99.4792 (86.0756) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2599 data: 0.1653 max mem: 3532 +Test: [390/517] eta: 0:00:33 acc1_pos: 39.908 acc5_pos: 100.000 loss: 0.3371 (0.4366) acc1: 97.1354 (86.3724) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2299 data: 0.1331 max mem: 3532 +Test: [400/517] eta: 0:00:30 acc1_pos: 39.908 acc5_pos: 100.000 loss: 0.3280 (0.4332) acc1: 97.9167 (86.6824) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2499 data: 0.1540 max mem: 3532 +Test: [410/517] eta: 0:00:28 acc1_pos: 39.908 acc5_pos: 100.000 loss: 0.3124 (0.4323) acc1: 97.6562 (86.8562) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.2549 data: 0.1583 max mem: 3532 +Test: [420/517] eta: 0:00:25 acc1_pos: 39.908 acc5_pos: 100.000 loss: 0.3964 (0.4313) acc1: 93.7500 (87.0348) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2499 data: 0.1515 max mem: 3532 +Test: [430/517] eta: 0:00:22 acc1_pos: 39.908 acc5_pos: 100.000 loss: 0.3806 (0.4297) acc1: 95.8333 (87.2323) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.2349 data: 0.1396 max mem: 3532 +Test: [440/517] eta: 0:00:20 acc1_pos: 39.908 acc5_pos: 100.000 loss: 0.3767 (0.4287) acc1: 93.4896 (87.3795) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.2349 data: 0.1408 max mem: 3533 +Test: [450/517] eta: 0:00:17 acc1_pos: 39.908 acc5_pos: 100.000 loss: 0.3728 (0.4269) acc1: 96.3542 (87.6166) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2299 data: 0.1342 max mem: 3533 +Test: [460/517] eta: 0:00:14 acc1_pos: 39.908 acc5_pos: 100.000 loss: 0.3528 (0.4256) acc1: 97.6562 (87.7700) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.2299 data: 0.1348 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 39.908 acc5_pos: 100.000 loss: 0.3493 (0.4236) acc1: 97.6562 (87.9722) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2499 data: 0.1537 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 39.908 acc5_pos: 100.000 loss: 0.3286 (0.4212) acc1: 98.6979 (88.2055) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2399 data: 0.1430 max mem: 3533 +Test: [490/517] eta: 0:00:06 acc1_pos: 39.908 acc5_pos: 100.000 loss: 0.3086 (0.4188) acc1: 98.6979 (88.3804) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2249 data: 0.1294 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 39.908 acc5_pos: 100.000 loss: 0.3574 (0.4198) acc1: 92.9688 (88.3140) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.2363 data: 0.1435 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 39.908 acc5_pos: 100.000 loss: 0.4196 (0.4190) acc1: 90.6250 (88.3623) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2037 data: 0.1136 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 39.908 acc5_pos: 100.000 loss: 0.3226 (0.4178) acc1: 97.6562 (88.4814) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1644 data: 0.0731 max mem: 3533 +Test: Total time: 0:02:12 (0.2566 s / it) +* Overall Acc@1 88.481 Acc@5 100.000 + target == real Acc@1 27.555 Acc@5 100.000 + target == fake Acc@1 94.802 Acc@5 100.000 + loss 0.418 + PR-AUC 0.9532 ROC-AUC 0.7188 EER 34.62% (thr=0.6733) +Accuracy of the network on the 198427 test images: 88.5% +[rank0]:[W725 14:02:05.770805087 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0725 14:02:28.604000 61063 torch/distributed/run.py:766] +W0725 14:02:28.604000 61063 torch/distributed/run.py:766] ***************************************** +W0725 14:02:28.604000 61063 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 14:02:28.604000 61063 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 14:03:04.546764758 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 14:03:04.864632030 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 14:03:04.919299371 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 14:03:05.077165098 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 14:03:05.210364037 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 14:03:05.239397847 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 14:03:05.335922904 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W725 14:03:05.338674497 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_all_46real_91fake_benchmark_eval_20260725_1215/fixed500k_all_46real_91fake_seed0/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_all_46real_91fake_imgseed20260725_n116-042-186_20260725_1215/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:40:15 acc1_pos: 29.688 acc5_pos: 100.000 loss: 0.8683 (0.8683) acc1: 29.6875 (29.6875) acc5: 100.0000 (100.0000) time: 6.2090 data: 4.5399 max mem: 3531 +Test: [ 10/389] eta: 0:04:06 acc1_pos: 77.604 acc5_pos: 100.000 loss: 0.7887 (0.7762) acc1: 40.8854 (50.3788) acc5: 100.0000 (100.0000) time: 0.6515 data: 0.4131 max mem: 3531 +Test: [ 20/389] eta: 0:02:23 acc1_pos: 13.802 acc5_pos: 100.000 loss: 0.7213 (0.7924) acc1: 40.8854 (45.6473) acc5: 100.0000 (100.0000) time: 0.0978 data: 0.0004 max mem: 3531 +Test: [ 30/389] eta: 0:01:50 acc1_pos: 61.458 acc5_pos: 100.000 loss: 0.7582 (0.7878) acc1: 43.4896 (44.4640) acc5: 100.0000 (100.0000) time: 0.1172 data: 0.0216 max mem: 3531 +Test: [ 40/389] eta: 0:01:35 acc1_pos: 3.125 acc5_pos: 100.000 loss: 0.7400 (0.7898) acc1: 45.0521 (44.7282) acc5: 100.0000 (100.0000) time: 0.1500 data: 0.0571 max mem: 3531 +Test: [ 50/389] eta: 0:01:24 acc1_pos: 72.396 acc5_pos: 100.000 loss: 0.6660 (0.7740) acc1: 55.7292 (46.7933) acc5: 100.0000 (100.0000) time: 0.1576 data: 0.0646 max mem: 3531 +Test: [ 60/389] eta: 0:01:16 acc1_pos: 19.531 acc5_pos: 100.000 loss: 0.6288 (0.7535) acc1: 67.1875 (49.3212) acc5: 100.0000 (100.0000) time: 0.1463 data: 0.0538 max mem: 3531 +Test: [ 70/389] eta: 0:01:11 acc1_pos: 52.083 acc5_pos: 100.000 loss: 0.6474 (0.7663) acc1: 52.0833 (47.6086) acc5: 100.0000 (100.0000) time: 0.1576 data: 0.0659 max mem: 3531 +Test: [ 80/389] eta: 0:01:06 acc1_pos: 90.104 acc5_pos: 100.000 loss: 0.8493 (0.7717) acc1: 30.7292 (47.4601) acc5: 100.0000 (100.0000) time: 0.1659 data: 0.0727 max mem: 3531 +Test: [ 90/389] eta: 0:01:02 acc1_pos: 48.177 acc5_pos: 100.000 loss: 0.4745 (0.7410) acc1: 76.3021 (50.6639) acc5: 100.0000 (100.0000) time: 0.1572 data: 0.0610 max mem: 3531 +Test: [100/389] eta: 0:00:57 acc1_pos: 28.646 acc5_pos: 100.000 loss: 0.7214 (0.7564) acc1: 41.4062 (48.8552) acc5: 100.0000 (100.0000) time: 0.1326 data: 0.0385 max mem: 3531 +Test: [110/389] eta: 0:00:54 acc1_pos: 99.740 acc5_pos: 100.000 loss: 0.8595 (0.7642) acc1: 32.8125 (48.5970) acc5: 100.0000 (100.0000) time: 0.1265 data: 0.0341 max mem: 3531 +Test: [120/389] eta: 0:00:51 acc1_pos: 62.240 acc5_pos: 100.000 loss: 0.6411 (0.7579) acc1: 59.6354 (49.4813) acc5: 100.0000 (100.0000) time: 0.1454 data: 0.0538 max mem: 3531 +Test: [130/389] eta: 0:00:48 acc1_pos: 21.472 acc5_pos: 100.000 loss: 0.6075 (0.7458) acc1: 62.2396 (50.5208) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1537 data: 0.0626 max mem: 3531 +Test: [140/389] eta: 0:00:46 acc1_pos: 21.472 acc5_pos: 100.000 loss: 0.4321 (0.7268) acc1: 80.2083 (52.6965) acc5: 100.0000 (100.0000) acc1_neg: 51.562 acc5_neg: 100.000 time: 0.1707 data: 0.0775 max mem: 3531 +Test: [150/389] eta: 0:00:45 acc1_pos: 21.472 acc5_pos: 100.000 loss: 0.4771 (0.7156) acc1: 80.4688 (53.7614) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1945 data: 0.1011 max mem: 3531 +Test: [160/389] eta: 0:00:43 acc1_pos: 21.472 acc5_pos: 100.000 loss: 0.5386 (0.6951) acc1: 81.2500 (55.8246) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2100 data: 0.1149 max mem: 3531 +Test: [170/389] eta: 0:00:41 acc1_pos: 21.472 acc5_pos: 100.000 loss: 0.3364 (0.6778) acc1: 97.6562 (57.9328) acc5: 100.0000 (100.0000) acc1_neg: 77.083 acc5_neg: 100.000 time: 0.1800 data: 0.0818 max mem: 3531 +Test: [180/389] eta: 0:00:39 acc1_pos: 21.472 acc5_pos: 100.000 loss: 0.4288 (0.6674) acc1: 92.1875 (59.3477) acc5: 100.0000 (100.0000) acc1_neg: 68.490 acc5_neg: 100.000 time: 0.1653 data: 0.0677 max mem: 3531 +Test: [190/389] eta: 0:00:37 acc1_pos: 21.472 acc5_pos: 100.000 loss: 0.4619 (0.6608) acc1: 84.8958 (60.0335) acc5: 100.0000 (100.0000) acc1_neg: 42.448 acc5_neg: 100.000 time: 0.1984 data: 0.1041 max mem: 3531 +Test: [200/389] eta: 0:00:35 acc1_pos: 21.472 acc5_pos: 100.000 loss: 0.5196 (0.6550) acc1: 82.8125 (60.7302) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.1895 data: 0.0982 max mem: 3531 +Test: [210/389] eta: 0:00:33 acc1_pos: 21.472 acc5_pos: 100.000 loss: 0.3610 (0.6411) acc1: 85.1562 (62.1840) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1712 data: 0.0780 max mem: 3532 +Test: [220/389] eta: 0:00:31 acc1_pos: 21.472 acc5_pos: 100.000 loss: 0.2530 (0.6250) acc1: 99.4792 (63.7738) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1699 data: 0.0750 max mem: 3532 +Test: [230/389] eta: 0:00:29 acc1_pos: 21.472 acc5_pos: 100.000 loss: 0.3925 (0.6190) acc1: 90.8854 (64.4830) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1603 data: 0.0644 max mem: 3532 +Test: [240/389] eta: 0:00:27 acc1_pos: 21.472 acc5_pos: 100.000 loss: 0.4328 (0.6111) acc1: 85.4167 (65.5169) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1611 data: 0.0640 max mem: 3532 +Test: [250/389] eta: 0:00:25 acc1_pos: 21.472 acc5_pos: 100.000 loss: 0.4079 (0.6000) acc1: 94.5312 (66.7964) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1958 data: 0.1010 max mem: 3532 +Test: [260/389] eta: 0:00:24 acc1_pos: 21.472 acc5_pos: 100.000 loss: 0.3516 (0.5958) acc1: 94.7917 (67.3561) acc5: 100.0000 (100.0000) acc1_neg: 30.469 acc5_neg: 100.000 time: 0.2132 data: 0.1192 max mem: 3532 +Test: [270/389] eta: 0:00:21 acc1_pos: 21.472 acc5_pos: 100.000 loss: 0.5417 (0.5988) acc1: 86.9792 (66.9059) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1646 data: 0.0704 max mem: 3532 +Test: [280/389] eta: 0:00:20 acc1_pos: 21.472 acc5_pos: 100.000 loss: 0.4613 (0.5961) acc1: 79.1667 (67.2005) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1686 data: 0.0770 max mem: 3532 +Test: [290/389] eta: 0:00:18 acc1_pos: 21.472 acc5_pos: 100.000 loss: 0.4610 (0.5912) acc1: 87.5000 (67.7960) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1956 data: 0.1038 max mem: 3532 +Test: [300/389] eta: 0:00:16 acc1_pos: 21.472 acc5_pos: 100.000 loss: 0.4388 (0.5859) acc1: 94.2708 (68.5908) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.1898 data: 0.0969 max mem: 3532 +Test: [310/389] eta: 0:00:14 acc1_pos: 21.472 acc5_pos: 100.000 loss: 0.4553 (0.5868) acc1: 89.3229 (68.5063) acc5: 100.0000 (100.0000) acc1_neg: 32.292 acc5_neg: 100.000 time: 0.1940 data: 0.0989 max mem: 3532 +Test: [320/389] eta: 0:00:12 acc1_pos: 21.472 acc5_pos: 100.000 loss: 0.5218 (0.5820) acc1: 76.8229 (69.0339) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2099 data: 0.1142 max mem: 3532 +Test: [330/389] eta: 0:00:10 acc1_pos: 21.472 acc5_pos: 100.000 loss: 0.5611 (0.5817) acc1: 81.7708 (69.3582) acc5: 100.0000 (100.0000) acc1_neg: 78.646 acc5_neg: 100.000 time: 0.1915 data: 0.0978 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 21.472 acc5_pos: 100.000 loss: 0.3893 (0.5723) acc1: 84.1146 (70.1643) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1929 data: 0.0994 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 21.472 acc5_pos: 100.000 loss: 0.2708 (0.5750) acc1: 98.9583 (69.9638) acc5: 100.0000 (100.0000) acc1_neg: 55.729 acc5_neg: 100.000 time: 0.2083 data: 0.1163 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 21.472 acc5_pos: 100.000 loss: 0.4691 (0.5736) acc1: 81.5104 (70.1415) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.1769 data: 0.0858 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 21.472 acc5_pos: 100.000 loss: 0.4678 (0.5714) acc1: 81.7708 (70.3104) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1875 data: 0.0964 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 21.472 acc5_pos: 100.000 loss: 0.4489 (0.5681) acc1: 77.3438 (70.5722) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1747 data: 0.0830 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 21.472 acc5_pos: 100.000 loss: 0.4030 (0.5647) acc1: 89.5833 (71.0059) acc5: 100.0000 (100.0000) acc1_neg: 86.339 acc5_neg: 100.000 time: 0.1642 data: 0.0687 max mem: 3532 +Test: Total time: 0:01:12 (0.1860 s / it) +* Overall Acc@1 71.006 Acc@5 100.000 + target == real Acc@1 49.132 Acc@5 100.000 + target == fake Acc@1 81.174 Acc@5 100.000 + loss 0.565 + PR-AUC 0.8444 ROC-AUC 0.7361 EER 32.83% (thr=0.5850) +Accuracy of the network on the 149358 test images: 71.0% +[rank0]:[W725 14:04:31.715564116 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 14:04:52.520000 61664 torch/distributed/run.py:766] +W0725 14:04:52.520000 61664 torch/distributed/run.py:766] ***************************************** +W0725 14:04:52.520000 61664 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 14:04:52.520000 61664 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 14:05:27.977741568 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 14:05:28.229910459 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 14:05:28.355991605 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W725 14:05:28.356068298 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 14:05:28.661157188 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +[rank1]:[W725 14:05:28.887728238 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 14:05:28.896916731 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 14:05:28.919592500 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_all_46real_91fake_benchmark_eval_20260725_1215/fixed500k_all_46real_91fake_seed0/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_all_46real_91fake_imgseed20260725_n116-042-186_20260725_1215/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:29:47 acc1_pos: 87.240 acc5_pos: 100.000 loss: 0.3730 (0.3730) acc1: 87.2396 (87.2396) acc5: 100.0000 (100.0000) time: 5.3683 data: 3.8204 max mem: 3531 +Test: [ 10/333] eta: 0:03:04 acc1_pos: 86.458 acc5_pos: 100.000 loss: 0.2976 (0.2969) acc1: 88.5417 (89.3229) acc5: 100.0000 (100.0000) time: 0.5698 data: 0.3477 max mem: 3531 +Test: [ 20/333] eta: 0:01:47 acc1_pos: 95.833 acc5_pos: 100.000 loss: 0.2879 (0.2940) acc1: 89.3229 (88.9881) acc5: 100.0000 (100.0000) time: 0.0921 data: 0.0005 max mem: 3531 +Test: [ 30/333] eta: 0:01:38 acc1_pos: 60.156 acc5_pos: 100.000 loss: 0.3842 (0.4318) acc1: 77.8646 (78.3602) acc5: 100.0000 (100.0000) time: 0.1887 data: 0.0953 max mem: 3531 +Test: [ 40/333] eta: 0:01:28 acc1_pos: 57.292 acc5_pos: 100.000 loss: 0.7469 (0.5090) acc1: 53.1250 (71.9004) acc5: 100.0000 (100.0000) time: 0.2595 data: 0.1658 max mem: 3531 +Test: [ 50/333] eta: 0:01:18 acc1_pos: 53.385 acc5_pos: 100.000 loss: 0.7469 (0.5561) acc1: 53.1250 (68.1270) acc5: 100.0000 (100.0000) time: 0.2083 data: 0.1147 max mem: 3531 +Test: [ 60/333] eta: 0:01:11 acc1_pos: 49.479 acc5_pos: 100.000 loss: 0.7331 (0.5884) acc1: 53.9062 (65.7061) acc5: 100.0000 (100.0000) time: 0.1807 data: 0.0867 max mem: 3531 +Test: [ 70/333] eta: 0:01:04 acc1_pos: 54.948 acc5_pos: 100.000 loss: 0.7435 (0.6116) acc1: 53.9062 (64.0222) acc5: 100.0000 (100.0000) time: 0.1601 data: 0.0662 max mem: 3531 +Test: [ 80/333] eta: 0:01:00 acc1_pos: 57.552 acc5_pos: 100.000 loss: 0.7408 (0.6268) acc1: 52.0833 (62.5707) acc5: 100.0000 (100.0000) time: 0.1705 data: 0.0757 max mem: 3531 +Test: [ 90/333] eta: 0:00:56 acc1_pos: 67.708 acc5_pos: 100.000 loss: 0.6736 (0.6291) acc1: 57.5521 (62.5715) acc5: 100.0000 (100.0000) time: 0.1960 data: 0.0978 max mem: 3531 +Test: [100/333] eta: 0:00:54 acc1_pos: 68.229 acc5_pos: 100.000 loss: 0.6579 (0.6339) acc1: 60.9375 (62.3247) acc5: 100.0000 (100.0000) time: 0.2183 data: 0.1221 max mem: 3531 +Test: [110/333] eta: 0:00:52 acc1_pos: 66.146 acc5_pos: 100.000 loss: 0.6671 (0.6372) acc1: 56.7708 (62.0191) acc5: 100.0000 (100.0000) time: 0.2375 data: 0.1415 max mem: 3531 +Test: [120/333] eta: 0:00:49 acc1_pos: 65.365 acc5_pos: 100.000 loss: 0.6379 (0.6379) acc1: 57.5521 (61.9878) acc5: 100.0000 (100.0000) time: 0.2060 data: 0.1101 max mem: 3531 +Test: [130/333] eta: 0:00:46 acc1_pos: 82.031 acc5_pos: 100.000 loss: 0.6148 (0.6374) acc1: 63.2812 (62.2436) acc5: 100.0000 (100.0000) time: 0.1899 data: 0.0981 max mem: 3531 +Test: [140/333] eta: 0:00:43 acc1_pos: 82.552 acc5_pos: 100.000 loss: 0.6153 (0.6359) acc1: 65.6250 (62.4557) acc5: 100.0000 (100.0000) time: 0.2083 data: 0.1147 max mem: 3531 +Test: [150/333] eta: 0:00:41 acc1_pos: 68.229 acc5_pos: 100.000 loss: 0.6331 (0.6384) acc1: 64.8438 (62.3051) acc5: 100.0000 (100.0000) time: 0.2049 data: 0.1097 max mem: 3531 +Test: [160/333] eta: 0:00:38 acc1_pos: 51.645 acc5_pos: 100.000 loss: 0.6571 (0.6481) acc1: 58.5938 (61.6913) acc5: 100.0000 (100.0000) acc1_neg: 41.406 acc5_neg: 100.000 time: 0.1799 data: 0.0852 max mem: 3531 +Test: [170/333] eta: 0:00:35 acc1_pos: 51.645 acc5_pos: 100.000 loss: 0.8121 (0.6606) acc1: 49.4792 (60.8431) acc5: 100.0000 (100.0000) acc1_neg: 50.781 acc5_neg: 100.000 time: 0.1749 data: 0.0812 max mem: 3531 +Test: [180/333] eta: 0:00:33 acc1_pos: 51.645 acc5_pos: 100.000 loss: 0.8043 (0.6660) acc1: 50.5208 (60.4282) acc5: 100.0000 (100.0000) acc1_neg: 57.552 acc5_neg: 100.000 time: 0.1749 data: 0.0804 max mem: 3531 +Test: [190/333] eta: 0:00:30 acc1_pos: 51.645 acc5_pos: 100.000 loss: 0.7203 (0.6672) acc1: 55.2083 (60.3989) acc5: 100.0000 (100.0000) acc1_neg: 55.208 acc5_neg: 100.000 time: 0.1650 data: 0.0707 max mem: 3531 +Test: [200/333] eta: 0:00:28 acc1_pos: 51.645 acc5_pos: 100.000 loss: 0.6963 (0.6688) acc1: 58.5938 (60.4413) acc5: 100.0000 (100.0000) acc1_neg: 58.333 acc5_neg: 100.000 time: 0.1700 data: 0.0792 max mem: 3531 +Test: [210/333] eta: 0:00:25 acc1_pos: 51.645 acc5_pos: 100.000 loss: 0.7498 (0.6749) acc1: 56.5104 (60.1760) acc5: 100.0000 (100.0000) acc1_neg: 51.823 acc5_neg: 100.000 time: 0.1700 data: 0.0784 max mem: 3532 +Test: [220/333] eta: 0:00:23 acc1_pos: 51.645 acc5_pos: 100.000 loss: 0.7815 (0.6824) acc1: 54.6875 (59.7344) acc5: 100.0000 (100.0000) acc1_neg: 66.406 acc5_neg: 100.000 time: 0.1587 data: 0.0640 max mem: 3532 +Test: [230/333] eta: 0:00:21 acc1_pos: 51.645 acc5_pos: 100.000 loss: 0.5752 (0.6764) acc1: 69.5312 (60.4009) acc5: 100.0000 (100.0000) acc1_neg: 72.135 acc5_neg: 100.000 time: 0.1538 data: 0.0595 max mem: 3532 +Test: [240/333] eta: 0:00:19 acc1_pos: 51.645 acc5_pos: 100.000 loss: 0.5600 (0.6753) acc1: 71.3542 (60.4577) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.1799 data: 0.0847 max mem: 3532 +Test: [250/333] eta: 0:00:16 acc1_pos: 51.645 acc5_pos: 100.000 loss: 0.6693 (0.6755) acc1: 61.4583 (60.4934) acc5: 100.0000 (100.0000) acc1_neg: 53.125 acc5_neg: 100.000 time: 0.1802 data: 0.0808 max mem: 3532 +Test: [260/333] eta: 0:00:14 acc1_pos: 51.645 acc5_pos: 100.000 loss: 0.6833 (0.6768) acc1: 60.6771 (60.4875) acc5: 100.0000 (100.0000) acc1_neg: 65.625 acc5_neg: 100.000 time: 0.1330 data: 0.0357 max mem: 3532 +Test: [270/333] eta: 0:00:12 acc1_pos: 51.645 acc5_pos: 100.000 loss: 0.6825 (0.6761) acc1: 62.7604 (60.6165) acc5: 100.0000 (100.0000) acc1_neg: 62.760 acc5_neg: 100.000 time: 0.1501 data: 0.0573 max mem: 3532 +Test: [280/333] eta: 0:00:10 acc1_pos: 51.645 acc5_pos: 100.000 loss: 0.6686 (0.6794) acc1: 61.9792 (60.3240) acc5: 100.0000 (100.0000) acc1_neg: 58.854 acc5_neg: 100.000 time: 0.1745 data: 0.0822 max mem: 3532 +Test: [290/333] eta: 0:00:08 acc1_pos: 51.645 acc5_pos: 100.000 loss: 0.8614 (0.6861) acc1: 43.4896 (59.7455) acc5: 100.0000 (100.0000) acc1_neg: 45.573 acc5_neg: 100.000 time: 0.1599 data: 0.0674 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 51.645 acc5_pos: 100.000 loss: 0.8795 (0.6927) acc1: 43.4896 (59.2521) acc5: 100.0000 (100.0000) acc1_neg: 48.958 acc5_neg: 100.000 time: 0.1602 data: 0.0681 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 51.645 acc5_pos: 100.000 loss: 0.7943 (0.6907) acc1: 50.7812 (59.4797) acc5: 100.0000 (100.0000) acc1_neg: 67.969 acc5_neg: 100.000 time: 0.1822 data: 0.0874 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 51.645 acc5_pos: 100.000 loss: 0.6082 (0.6895) acc1: 67.4479 (59.6281) acc5: 100.0000 (100.0000) acc1_neg: 47.135 acc5_neg: 100.000 time: 0.1689 data: 0.0744 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 51.645 acc5_pos: 100.000 loss: 0.8528 (0.6969) acc1: 47.1354 (59.1571) acc5: 100.0000 (100.0000) acc1_neg: 46.354 acc5_neg: 100.000 time: 0.1277 data: 0.0368 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 51.645 acc5_pos: 100.000 loss: 0.9007 (0.6982) acc1: 46.3542 (59.0795) acc5: 100.0000 (100.0000) acc1_neg: 44.789 acc5_neg: 100.000 time: 0.1269 data: 0.0321 max mem: 3532 +Test: Total time: 0:01:03 (0.1914 s / it) +* Overall Acc@1 59.079 Acc@5 100.000 + target == real Acc@1 62.314 Acc@5 100.000 + target == fake Acc@1 56.200 Acc@5 100.000 + loss 0.698 + PR-AUC 0.6435 ROC-AUC 0.6321 EER 40.77% (thr=0.4805) +Accuracy of the network on the 127843 test images: 59.1% +[rank0]:[W725 14:06:45.199269352 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 14:07:05.631000 62241 torch/distributed/run.py:766] +W0725 14:07:05.631000 62241 torch/distributed/run.py:766] ***************************************** +W0725 14:07:05.631000 62241 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 14:07:05.631000 62241 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 14:07:40.496258520 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 14:07:41.440326801 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W725 14:07:42.143563233 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank4]:[W725 14:07:42.149840632 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 14:07:42.173655526 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 14:07:42.424575968 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 14:07:42.494300747 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 14:07:42.609710853 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_all_46real_91fake_benchmark_eval_20260725_1215/fixed500k_all_46real_91fake_seed0/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_all_46real_91fake_imgseed20260725_n116-042-186_20260725_1215/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:20:12 acc1_pos: 95.573 acc5_pos: 100.000 loss: 0.2025 (0.2025) acc1: 95.5729 (95.5729) acc5: 100.0000 (100.0000) time: 5.1591 data: 3.0472 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/235] eta: 0:02:05 acc1_pos: 94.531 acc5_pos: 100.000 loss: 0.2088 (0.2091) acc1: 95.5729 (95.4545) acc5: 100.0000 (100.0000) time: 0.5561 data: 0.2826 max mem: 3531 +Test: [ 20/235] eta: 0:01:13 acc1_pos: 96.094 acc5_pos: 100.000 loss: 0.2029 (0.2051) acc1: 95.8333 (95.8333) acc5: 100.0000 (100.0000) time: 0.1025 data: 0.0104 max mem: 3531 +Test: [ 30/235] eta: 0:00:59 acc1_pos: 96.615 acc5_pos: 100.000 loss: 0.1983 (0.2030) acc1: 96.3542 (95.9761) acc5: 100.0000 (100.0000) time: 0.1450 data: 0.0518 max mem: 3531 +Test: [ 40/235] eta: 0:00:55 acc1_pos: 95.833 acc5_pos: 100.000 loss: 0.1983 (0.2023) acc1: 96.3542 (96.0302) acc5: 100.0000 (100.0000) time: 0.2267 data: 0.1333 max mem: 3531 +Test: [ 50/235] eta: 0:00:51 acc1_pos: 93.750 acc5_pos: 100.000 loss: 0.1993 (0.2031) acc1: 96.0938 (95.9916) acc5: 100.0000 (100.0000) time: 0.2595 data: 0.1638 max mem: 3531 +Test: [ 60/235] eta: 0:00:45 acc1_pos: 95.052 acc5_pos: 100.000 loss: 0.2047 (0.2044) acc1: 95.5729 (95.9443) acc5: 100.0000 (100.0000) time: 0.1981 data: 0.1026 max mem: 3531 +Test: [ 70/235] eta: 0:00:40 acc1_pos: 98.462 acc5_pos: 100.000 loss: 0.2229 (0.3551) acc1: 95.0521 (86.4253) acc5: 100.0000 (100.0000) acc1_neg: 16.406 acc5_neg: 100.000 time: 0.1649 data: 0.0674 max mem: 3531 +Test: [ 80/235] eta: 0:00:36 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4119 (0.4880) acc1: 19.7917 (78.1507) acc5: 100.0000 (100.0000) acc1_neg: 15.625 acc5_neg: 100.000 time: 0.1801 data: 0.0833 max mem: 3531 +Test: [ 90/235] eta: 0:00:33 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4272 (0.5913) acc1: 19.2708 (71.6546) acc5: 100.0000 (100.0000) acc1_neg: 17.188 acc5_neg: 100.000 time: 0.1804 data: 0.0866 max mem: 3531 +Test: [100/235] eta: 0:00:30 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4272 (0.6721) acc1: 19.2708 (66.5558) acc5: 100.0000 (100.0000) acc1_neg: 16.927 acc5_neg: 100.000 time: 0.1746 data: 0.0800 max mem: 3531 +Test: [110/235] eta: 0:00:27 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4076 (0.7389) acc1: 20.0521 (62.4038) acc5: 100.0000 (100.0000) acc1_neg: 23.958 acc5_neg: 100.000 time: 0.1794 data: 0.0827 max mem: 3531 +Test: [120/235] eta: 0:00:25 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4727 (0.8130) acc1: 17.7083 (58.1482) acc5: 100.0000 (100.0000) acc1_neg: 11.198 acc5_neg: 100.000 time: 0.2099 data: 0.1148 max mem: 3531 +Test: [130/235] eta: 0:00:23 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.6426 (0.8767) acc1: 10.4167 (54.5165) acc5: 100.0000 (100.0000) acc1_neg: 9.635 acc5_neg: 100.000 time: 0.2099 data: 0.1160 max mem: 3531 +Test: [140/235] eta: 0:00:20 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.6695 (0.9343) acc1: 9.8958 (51.3575) acc5: 100.0000 (100.0000) acc1_neg: 6.771 acc5_neg: 100.000 time: 0.1649 data: 0.0696 max mem: 3531 +Test: [150/235] eta: 0:00:18 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.6705 (0.9820) acc1: 9.8958 (48.6100) acc5: 100.0000 (100.0000) acc1_neg: 12.240 acc5_neg: 100.000 time: 0.1704 data: 0.0776 max mem: 3531 +Test: [160/235] eta: 0:00:15 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.6519 (1.0240) acc1: 9.8958 (46.2361) acc5: 100.0000 (100.0000) acc1_neg: 10.677 acc5_neg: 100.000 time: 0.1971 data: 0.1041 max mem: 3531 +Test: [170/235] eta: 0:00:13 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.6200 (1.0339) acc1: 12.2396 (45.2257) acc5: 100.0000 (100.0000) acc1_neg: 26.823 acc5_neg: 100.000 time: 0.2002 data: 0.1031 max mem: 3531 +Test: [180/235] eta: 0:00:11 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.1439 (1.0393) acc1: 30.9896 (44.4694) acc5: 100.0000 (100.0000) acc1_neg: 32.031 acc5_neg: 100.000 time: 0.1987 data: 0.1012 max mem: 3531 +Test: [190/235] eta: 0:00:09 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.1182 (1.0441) acc1: 31.5104 (43.7923) acc5: 100.0000 (100.0000) acc1_neg: 34.115 acc5_neg: 100.000 time: 0.1790 data: 0.0822 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.1219 (1.0493) acc1: 29.6875 (43.1061) acc5: 100.0000 (100.0000) acc1_neg: 28.646 acc5_neg: 100.000 time: 0.1710 data: 0.0748 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.1404 (1.0531) acc1: 30.2083 (42.5676) acc5: 100.0000 (100.0000) acc1_neg: 30.208 acc5_neg: 100.000 time: 0.1817 data: 0.0880 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.1263 (1.0563) acc1: 31.2500 (42.0826) acc5: 100.0000 (100.0000) acc1_neg: 32.292 acc5_neg: 100.000 time: 0.1778 data: 0.0856 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.1188 (1.0589) acc1: 31.7708 (41.6531) acc5: 100.0000 (100.0000) acc1_neg: 31.771 acc5_neg: 100.000 time: 0.1519 data: 0.0608 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.1195 (1.0612) acc1: 31.7708 (41.5080) acc5: 100.0000 (100.0000) acc1_neg: 26.087 acc5_neg: 100.000 time: 0.1395 data: 0.0516 max mem: 3532 +Test: Total time: 0:00:47 (0.2013 s / it) +* Overall Acc@1 41.508 Acc@5 100.000 + target == real Acc@1 95.958 Acc@5 100.000 + target == fake Acc@1 21.815 Acc@5 100.000 + loss 1.061 + PR-AUC 0.8824 ROC-AUC 0.7434 EER 32.27% (thr=0.1753) +Accuracy of the network on the 89879 test images: 41.5% +[rank0]:[W725 14:08:42.204452837 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 14:09:03.995000 62836 torch/distributed/run.py:766] +W0725 14:09:03.995000 62836 torch/distributed/run.py:766] ***************************************** +W0725 14:09:03.995000 62836 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 14:09:03.995000 62836 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 14:09:40.687902198 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 14:09:42.115563622 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 14:09:42.134324433 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +[rank0]:[W725 14:09:42.299156778 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 14:09:42.305886254 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 14:09:42.317827096 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 14:09:42.553127249 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 14:09:42.802159420 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_all_46real_91fake_benchmark_eval_20260725_1215/fixed500k_all_46real_91fake_seed0/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_all_46real_91fake_imgseed20260725_n116-042-186_20260725_1215/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:26:06 acc1_pos: 59.115 acc5_pos: 100.000 loss: 0.6551 (0.6551) acc1: 59.1146 (59.1146) acc5: 100.0000 (100.0000) time: 5.8219 data: 4.3441 max mem: 3531 +Test: [ 10/269] eta: 0:02:38 acc1_pos: 81.510 acc5_pos: 100.000 loss: 0.5721 (0.5763) acc1: 71.3542 (71.5909) acc5: 100.0000 (100.0000) time: 0.6117 data: 0.3953 max mem: 3531 +Test: [ 20/269] eta: 0:01:37 acc1_pos: 83.854 acc5_pos: 100.000 loss: 0.5407 (0.5412) acc1: 75.2604 (76.2649) acc5: 100.0000 (100.0000) time: 0.1217 data: 0.0314 max mem: 3531 +Test: [ 30/269] eta: 0:01:27 acc1_pos: 72.135 acc5_pos: 100.000 loss: 0.5197 (0.5324) acc1: 79.6875 (77.3858) acc5: 100.0000 (100.0000) time: 0.2297 data: 0.1379 max mem: 3531 +Test: [ 40/269] eta: 0:01:17 acc1_pos: 75.862 acc5_pos: 100.000 loss: 0.5690 (0.5473) acc1: 74.4792 (76.2005) acc5: 100.0000 (100.0000) acc1_neg: 71.354 acc5_neg: 100.000 time: 0.2782 data: 0.1851 max mem: 3531 +Test: [ 50/269] eta: 0:01:07 acc1_pos: 75.862 acc5_pos: 100.000 loss: 0.6503 (0.5814) acc1: 63.0208 (72.1354) acc5: 100.0000 (100.0000) acc1_neg: 29.167 acc5_neg: 100.000 time: 0.2156 data: 0.1233 max mem: 3531 +Test: [ 60/269] eta: 0:00:59 acc1_pos: 75.862 acc5_pos: 100.000 loss: 0.7362 (0.6126) acc1: 48.1771 (67.9389) acc5: 100.0000 (100.0000) acc1_neg: 48.177 acc5_neg: 100.000 time: 0.1807 data: 0.0892 max mem: 3531 +Test: [ 70/269] eta: 0:00:55 acc1_pos: 75.862 acc5_pos: 100.000 loss: 0.7155 (0.6207) acc1: 48.6979 (66.6006) acc5: 100.0000 (100.0000) acc1_neg: 59.115 acc5_neg: 100.000 time: 0.2066 data: 0.1146 max mem: 3531 +Test: [ 80/269] eta: 0:00:51 acc1_pos: 75.862 acc5_pos: 100.000 loss: 0.6694 (0.6260) acc1: 55.7292 (65.8308) acc5: 100.0000 (100.0000) acc1_neg: 52.344 acc5_neg: 100.000 time: 0.2242 data: 0.1298 max mem: 3531 +Test: [ 90/269] eta: 0:00:47 acc1_pos: 75.862 acc5_pos: 100.000 loss: 0.7480 (0.6558) acc1: 50.0000 (62.1423) acc5: 100.0000 (100.0000) acc1_neg: 43.750 acc5_neg: 100.000 time: 0.2299 data: 0.1336 max mem: 3531 +Test: [100/269] eta: 0:00:44 acc1_pos: 75.862 acc5_pos: 100.000 loss: 0.8024 (0.6685) acc1: 39.0625 (60.3058) acc5: 100.0000 (100.0000) acc1_neg: 33.854 acc5_neg: 100.000 time: 0.2198 data: 0.1246 max mem: 3531 +Test: [110/269] eta: 0:00:40 acc1_pos: 75.862 acc5_pos: 100.000 loss: 0.7913 (0.6803) acc1: 39.0625 (58.4061) acc5: 100.0000 (100.0000) acc1_neg: 31.510 acc5_neg: 100.000 time: 0.1994 data: 0.1039 max mem: 3531 +Test: [120/269] eta: 0:00:37 acc1_pos: 75.862 acc5_pos: 100.000 loss: 0.7762 (0.6861) acc1: 43.7500 (57.6167) acc5: 100.0000 (100.0000) acc1_neg: 48.438 acc5_neg: 100.000 time: 0.1904 data: 0.0968 max mem: 3531 +Test: [130/269] eta: 0:00:34 acc1_pos: 75.862 acc5_pos: 100.000 loss: 0.7823 (0.6970) acc1: 43.7500 (56.3434) acc5: 100.0000 (100.0000) acc1_neg: 32.031 acc5_neg: 100.000 time: 0.1898 data: 0.0966 max mem: 3531 +Test: [140/269] eta: 0:00:31 acc1_pos: 75.862 acc5_pos: 100.000 loss: 0.8592 (0.7105) acc1: 36.1979 (54.5822) acc5: 100.0000 (100.0000) acc1_neg: 27.083 acc5_neg: 100.000 time: 0.2120 data: 0.1167 max mem: 3531 +Test: [150/269] eta: 0:00:28 acc1_pos: 75.862 acc5_pos: 100.000 loss: 0.8596 (0.7183) acc1: 33.0729 (53.5268) acc5: 100.0000 (100.0000) acc1_neg: 42.188 acc5_neg: 100.000 time: 0.1912 data: 0.0981 max mem: 3531 +Test: [160/269] eta: 0:00:25 acc1_pos: 75.862 acc5_pos: 100.000 loss: 0.7621 (0.7213) acc1: 44.0104 (53.1234) acc5: 100.0000 (100.0000) acc1_neg: 51.042 acc5_neg: 100.000 time: 0.1573 data: 0.0636 max mem: 3531 +Test: [170/269] eta: 0:00:22 acc1_pos: 75.862 acc5_pos: 100.000 loss: 0.8065 (0.7276) acc1: 38.5417 (52.1991) acc5: 100.0000 (100.0000) acc1_neg: 34.375 acc5_neg: 100.000 time: 0.1568 data: 0.0599 max mem: 3531 +Test: [180/269] eta: 0:00:20 acc1_pos: 75.862 acc5_pos: 100.000 loss: 0.8065 (0.7279) acc1: 38.5417 (51.9855) acc5: 100.0000 (100.0000) acc1_neg: 38.281 acc5_neg: 100.000 time: 0.1799 data: 0.0843 max mem: 3531 +Test: [190/269] eta: 0:00:17 acc1_pos: 75.862 acc5_pos: 100.000 loss: 0.7333 (0.7305) acc1: 44.0104 (51.4589) acc5: 100.0000 (100.0000) acc1_neg: 48.438 acc5_neg: 100.000 time: 0.1999 data: 0.1047 max mem: 3531 +Test: [200/269] eta: 0:00:15 acc1_pos: 75.862 acc5_pos: 100.000 loss: 0.8358 (0.7362) acc1: 37.5000 (50.6841) acc5: 100.0000 (100.0000) acc1_neg: 31.510 acc5_neg: 100.000 time: 0.2111 data: 0.1173 max mem: 3531 +Test: [210/269] eta: 0:00:13 acc1_pos: 75.862 acc5_pos: 100.000 loss: 0.7974 (0.7382) acc1: 39.3229 (50.3764) acc5: 100.0000 (100.0000) acc1_neg: 47.396 acc5_neg: 100.000 time: 0.1907 data: 0.0991 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 75.862 acc5_pos: 100.000 loss: 0.7337 (0.7390) acc1: 44.7917 (50.2321) acc5: 100.0000 (100.0000) acc1_neg: 69.531 acc5_neg: 100.000 time: 0.1588 data: 0.0657 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 75.862 acc5_pos: 100.000 loss: 0.6638 (0.7347) acc1: 54.1667 (50.6663) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.1805 data: 0.0884 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 75.862 acc5_pos: 100.000 loss: 0.7470 (0.7385) acc1: 46.8750 (50.2366) acc5: 100.0000 (100.0000) acc1_neg: 39.844 acc5_neg: 100.000 time: 0.1856 data: 0.0947 max mem: 3532 +Test: [250/269] eta: 0:00:04 acc1_pos: 75.862 acc5_pos: 100.000 loss: 0.8309 (0.7445) acc1: 36.1979 (49.5912) acc5: 100.0000 (100.0000) acc1_neg: 11.198 acc5_neg: 100.000 time: 0.1433 data: 0.0529 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 75.862 acc5_pos: 100.000 loss: 0.8740 (0.7510) acc1: 28.3854 (48.8625) acc5: 100.0000 (100.0000) acc1_neg: 51.302 acc5_neg: 100.000 time: 0.1130 data: 0.0234 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 75.862 acc5_pos: 100.000 loss: 0.8061 (0.7511) acc1: 33.0729 (48.7671) acc5: 100.0000 (100.0000) acc1_neg: 58.530 acc5_neg: 100.000 time: 0.1216 data: 0.0277 max mem: 3532 +Test: Total time: 0:00:55 (0.2075 s / it) +* Overall Acc@1 48.767 Acc@5 100.000 + target == real Acc@1 77.112 Acc@5 100.000 + target == fake Acc@1 44.899 Acc@5 100.000 + loss 0.751 + PR-AUC 0.9296 ROC-AUC 0.6644 EER 38.30% (thr=0.4297) +Accuracy of the network on the 103293 test images: 48.8% +[rank0]:[W725 14:10:50.150418647 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-25T14:10:56Z] eval_done output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_all_46real_91fake_benchmark_eval_20260725_1215/fixed500k_all_46real_91fake_seed0 diff --git a/ScaleDF_rebuttal/_host_logs/all_46real_91fake_11451_20260725_1215/eval_after_train_complete.pid b/ScaleDF_rebuttal/_host_logs/all_46real_91fake_11451_20260725_1215/eval_after_train_complete.pid new file mode 100644 index 0000000000000000000000000000000000000000..6d8c3053e2349deb30a4866d0bd1a875f5cff341 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/all_46real_91fake_11451_20260725_1215/eval_after_train_complete.pid @@ -0,0 +1 @@ +58606 diff --git a/ScaleDF_rebuttal/_host_logs/all_46real_91fake_11451_20260725_1215/train_all_46real_91fake.log b/ScaleDF_rebuttal/_host_logs/all_46real_91fake_11451_20260725_1215/train_all_46real_91fake.log new file mode 100644 index 0000000000000000000000000000000000000000..acafad655cdff6fe4bcfd67fa6cd32ec531004a6 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/all_46real_91fake_11451_20260725_1215/train_all_46real_91fake.log @@ -0,0 +1,375 @@ +started_utc=2026-07-25T12:46:19Z +hostname=n116-042-186 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +kind=nrealfake +selection=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/selections/fixed_budget_500k_all_46real_91fake_20260725/scaledf_v4_fixed500000_all_46real_91fake_imgseed20260725.pkl +model_name=deit_base_patch16_LS +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_all_46real_91fake_imgseed20260725_n116-042-186_20260725_1215 +main=main_cross_2direct_balance_realscale_perturb_v4_NRealFake.py +nproc_per_node=8 +batch_per_gpu=256 +W0725 12:46:42.548000 730 torch/distributed/run.py:766] +W0725 12:46:42.548000 730 torch/distributed/run.py:766] ***************************************** +W0725 12:46:42.548000 730 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 12:46:42.548000 730 torch/distributed/run.py:766] ***************************************** +Traceback (most recent call last): +Traceback (most recent call last): +Traceback (most recent call last): + File "/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_NRealFake.py", line 13, in + File "/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_NRealFake.py", line 13, in +Traceback (most recent call last): + File "/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_NRealFake.py", line 13, in + File "/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_NRealFake.py", line 13, in +Traceback (most recent call last): + File "/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_NRealFake.py", line 13, in +Traceback (most recent call last): +Traceback (most recent call last): + File "/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_NRealFake.py", line 13, in + File "/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_NRealFake.py", line 13, in + from timm.data import Mixupfrom timm.data import Mixupfrom timm.data import Mixup + + +from timm.data import Mixup + File "/home/tiger/.local/lib/python3.11/site-packages/timm/__init__.py", line 2, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/__init__.py", line 2, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/__init__.py", line 2, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/__init__.py", line 2, in + from timm.data import Mixup + from timm.data import Mixupfrom timm.data import Mixup + + File "/home/tiger/.local/lib/python3.11/site-packages/timm/__init__.py", line 2, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/__init__.py", line 2, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/__init__.py", line 2, in +Traceback (most recent call last): + File "/tmp/ScaleDF_repro_20260724/code/deit/main_cross_2direct_balance_realscale_perturb_v4_NRealFake.py", line 13, in + from timm.data import Mixup + File "/home/tiger/.local/lib/python3.11/site-packages/timm/__init__.py", line 2, in + from .models import create_model, list_models, is_model, list_modules, model_entrypoint, \ + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/__init__.py", line 1, in +from .models import create_model, list_models, is_model, list_modules, model_entrypoint, \ + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/__init__.py", line 1, in +from .models import create_model, list_models, is_model, list_modules, model_entrypoint, \ + from .models import create_model, list_models, is_model, list_modules, model_entrypoint, \ File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/__init__.py", line 1, in + + from .models import create_model, list_models, is_model, list_modules, model_entrypoint, \ + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/__init__.py", line 1, in +from .models import create_model, list_models, is_model, list_modules, model_entrypoint, \ File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/__init__.py", line 1, in + +from .models import create_model, list_models, is_model, list_modules, model_entrypoint, \ + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/__init__.py", line 1, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/__init__.py", line 1, in + from .models import create_model, list_models, is_model, list_modules, model_entrypoint, \ + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/__init__.py", line 1, in + from .byoanet import * + from .byoanet import * from .byoanet import * + from .byoanet import * from .byoanet import *from .byoanet import *from .byoanet import * + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/byoanet.py", line 15, in + + + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/byoanet.py", line 15, in + + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/byoanet.py", line 15, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/byoanet.py", line 15, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/byoanet.py", line 15, in + + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/byoanet.py", line 15, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/byoanet.py", line 15, in + from .byoanet import * + File "/home/tiger/.local/lib/python3.11/site-packages/timm/models/byoanet.py", line 15, in + from timm.data import IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STDfrom timm.data import IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STDfrom timm.data import IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STD + + + from timm.data import IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STDfrom timm.data import IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STD + + from timm.data import IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STD +from timm.data import IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STD + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/__init__.py", line 5, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/__init__.py", line 5, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/__init__.py", line 5, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/__init__.py", line 5, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/__init__.py", line 5, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/__init__.py", line 5, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/__init__.py", line 5, in + from timm.data import IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STD + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/__init__.py", line 5, in + from .dataset import ImageDataset, IterableImageDataset, AugMixDataset +from .dataset import ImageDataset, IterableImageDataset, AugMixDataset + from .dataset import ImageDataset, IterableImageDataset, AugMixDataset + from .dataset import ImageDataset, IterableImageDataset, AugMixDataset File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/dataset.py", line 12, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/dataset.py", line 12, in + + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/dataset.py", line 12, in +from .dataset import ImageDataset, IterableImageDataset, AugMixDataset from .dataset import ImageDataset, IterableImageDataset, AugMixDataset + + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/dataset.py", line 12, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/dataset.py", line 12, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/dataset.py", line 12, in +from .dataset import ImageDataset, IterableImageDataset, AugMixDataset + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/dataset.py", line 12, in + from .dataset import ImageDataset, IterableImageDataset, AugMixDataset + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/dataset.py", line 12, in + from .parsers import create_parser + from .parsers import create_parser +from .parsers import create_parser + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/__init__.py", line 1, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/__init__.py", line 1, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/__init__.py", line 1, in + from .parsers import create_parser + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/__init__.py", line 1, in + from .parsers import create_parserfrom .parsers import create_parser + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/__init__.py", line 1, in +from .parsers import create_parser + + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/__init__.py", line 1, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/__init__.py", line 1, in + from .parsers import create_parser + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/__init__.py", line 1, in +from .parser_factory import create_parser + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_factory.py", line 3, in + from .parser_factory import create_parser + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_factory.py", line 3, in + from .parser_factory import create_parser + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_factory.py", line 3, in + from .parser_factory import create_parser + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_factory.py", line 3, in + from .parser_image_folder import ParserImageFolder + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_image_folder.py", line 10, in + from .parser_factory import create_parser + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_factory.py", line 3, in + from .parser_factory import create_parser + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_factory.py", line 3, in + from .parser_factory import create_parser + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_factory.py", line 3, in + from .parser_factory import create_parser + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_factory.py", line 3, in + from .parser_image_folder import ParserImageFolder + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_image_folder.py", line 10, in + from .parser_image_folder import ParserImageFolder + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_image_folder.py", line 10, in + from .parser_image_folder import ParserImageFolder + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_image_folder.py", line 10, in + from .parser_image_folder import ParserImageFolder + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_image_folder.py", line 10, in + from timm.utils.misc import natural_key + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/__init__.py", line 13, in + from .parser_image_folder import ParserImageFolder + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_image_folder.py", line 10, in + from .parser_image_folder import ParserImageFolder + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_image_folder.py", line 10, in + from .parser_image_folder import ParserImageFolder + File "/home/tiger/.local/lib/python3.11/site-packages/timm/data/parsers/parser_image_folder.py", line 10, in + from timm.utils.misc import natural_key + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/__init__.py", line 13, in + from timm.utils.misc import natural_key + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/__init__.py", line 13, in + from timm.utils.misc import natural_key + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/__init__.py", line 13, in + from timm.utils.misc import natural_key + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/__init__.py", line 13, in + from timm.utils.misc import natural_key + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/__init__.py", line 13, in + from timm.utils.misc import natural_key + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/__init__.py", line 13, in + from timm.utils.misc import natural_key + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/__init__.py", line 13, in + from .summary import update_summary, get_outdir + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/summary.py", line 9, in + from .summary import update_summary, get_outdir + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/summary.py", line 9, in + from .summary import update_summary, get_outdir + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/summary.py", line 9, in + from .summary import update_summary, get_outdir + from .summary import update_summary, get_outdir + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/summary.py", line 9, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/summary.py", line 9, in + import wandb + File "/usr/local/lib/python3.11/dist-packages/wandb/__init__.py", line 35, in + from .summary import update_summary, get_outdir + from .summary import update_summary, get_outdir + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/summary.py", line 9, in + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/summary.py", line 9, in + from .summary import update_summary, get_outdir + File "/home/tiger/.local/lib/python3.11/site-packages/timm/utils/summary.py", line 9, in + import wandb + File "/usr/local/lib/python3.11/dist-packages/wandb/__init__.py", line 35, in + import wandb + File "/usr/local/lib/python3.11/dist-packages/wandb/__init__.py", line 35, in + import wandb + File "/usr/local/lib/python3.11/dist-packages/wandb/__init__.py", line 35, in + import wandb + File "/usr/local/lib/python3.11/dist-packages/wandb/__init__.py", line 35, in + import wandb + File "/usr/local/lib/python3.11/dist-packages/wandb/__init__.py", line 35, in + import wandb + File "/usr/local/lib/python3.11/dist-packages/wandb/__init__.py", line 35, in + import wandb + File "/usr/local/lib/python3.11/dist-packages/wandb/__init__.py", line 35, in + from wandb import sdk as wandb_sdk + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/__init__.py", line 9, in + from wandb import sdk as wandb_sdk + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/__init__.py", line 9, in +from wandb import sdk as wandb_sdk + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/__init__.py", line 9, in + from wandb import sdk as wandb_sdk + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/__init__.py", line 9, in + from wandb import sdk as wandb_sdk + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/__init__.py", line 9, in + from wandb import sdk as wandb_sdk + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/__init__.py", line 9, in +from wandb import sdk as wandb_sdk + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/__init__.py", line 9, in + from .wandb_init import _attach, init # noqa: F401 + from .wandb_init import _attach, init # noqa: F401 from .wandb_init import _attach, init # noqa: F401 +from .wandb_init import _attach, init # noqa: F401^ +^ +^^^ ^ ^ ^ ^^^ ^ ^ ^ ^ ^^^ ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ ^^^^^from wandb import sdk as wandb_sdk^^^^^^ +^^^^^^^^^^^^^^ File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/__init__.py", line 9, in +^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ ^^ +^^from .wandb_init import _attach, init # noqa: F401 File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/wandb_init.py", line 31, in +^ +^^^^^^^^^^ +^^^ File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/wandb_init.py", line 31, in +^^ ^^ + ^ ^ File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/wandb_init.py", line 31, in + +^^ File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/wandb_init.py", line 31, in +^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/wandb_init.py", line 31, in + from .wandb_init import _attach, init # noqa: F401 + ^^^^^^^^ ^from .wandb_init import _attach, init # noqa: F401^ +^^^^^^^^^^^^ ^ from .wandb_init import _attach, init # noqa: F401 ^ + ^^^^^^^^^ ^^^^ ^^ ^^ ^^^^^^^^^^^^^^ +^^^^ File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/wandb_init.py", line 31, in +^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/wandb_init.py", line 31, in +^^^^^^^^^^^^^ + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/wandb_init.py", line 31, in + from .backend.backend import Backend +from .backend.backend import Backend + from .backend.backend import Backendfrom .backend.backend import Backend File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/backend/backend.py", line 17, in + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/backend/backend.py", line 17, in +from .backend.backend import Backend + + +from .backend.backend import Backend +from .backend.backend import Backend + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/backend/backend.py", line 17, in + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/backend/backend.py", line 17, in + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/backend/backend.py", line 17, in + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/backend/backend.py", line 17, in + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/backend/backend.py", line 17, in + from .backend.backend import Backend + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/backend/backend.py", line 17, in + from ..interface.interface import InterfaceBase + from ..interface.interface import InterfaceBasefrom ..interface.interface import InterfaceBase + + from ..interface.interface import InterfaceBase + from ..interface.interface import InterfaceBasefrom ..interface.interface import InterfaceBase + + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/interface/interface.py", line 34, in + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/interface/interface.py", line 34, in + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/interface/interface.py", line 34, in + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/interface/interface.py", line 34, in + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/interface/interface.py", line 34, in + from ..interface.interface import InterfaceBase + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/interface/interface.py", line 34, in + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/interface/interface.py", line 34, in + from ..interface.interface import InterfaceBase + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/interface/interface.py", line 34, in + from ..lib.mailbox import MailboxHandle + from ..lib.mailbox import MailboxHandle + from ..lib.mailbox import MailboxHandlefrom ..lib.mailbox import MailboxHandle + +from ..lib.mailbox import MailboxHandle File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 92, in +from ..lib.mailbox import MailboxHandle + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 92, in + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 92, in + + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 92, in +from ..lib.mailbox import MailboxHandle + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 92, in + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 92, in + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 92, in + from ..lib.mailbox import MailboxHandle + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 92, in + class _MailboxSlot:class _MailboxSlot: + + class _MailboxSlot: + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 93, in _MailboxSlot + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 93, in _MailboxSlot + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 93, in _MailboxSlot + class _MailboxSlot: + class _MailboxSlot: File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 93, in _MailboxSlot + class _MailboxSlot: + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 93, in _MailboxSlot +class _MailboxSlot: + + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 93, in _MailboxSlot + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 93, in _MailboxSlot + class _MailboxSlot: + File "/usr/local/lib/python3.11/dist-packages/wandb/sdk/lib/mailbox.py", line 93, in _MailboxSlot + _result: Optional[pb.Result] + ^^^^^^^^^ +AttributeError: module 'wandb.proto.wandb_internal_pb2' has no attribute 'Result' + _result: Optional[pb.Result] + _result: Optional[pb.Result] +^ ^_result: Optional[pb.Result]^ +^^^^^^ +AttributeError: module 'wandb.proto.wandb_internal_pb2' has no attribute 'Result' + ^ ^ ^ ^^^^^^^^^^^ +^^^^AttributeError +: module 'wandb.proto.wandb_internal_pb2' has no attribute 'Result'AttributeError: module 'wandb.proto.wandb_internal_pb2' has no attribute 'Result' + + _result: Optional[pb.Result] + ^^^^ ^_result: Optional[pb.Result]^ +^^^ +AttributeError: module 'wandb.proto.wandb_internal_pb2' has no attribute 'Result' + _result: Optional[pb.Result] + ^^^^^^^^^ +AttributeError : module 'wandb.proto.wandb_internal_pb2' has no attribute 'Result' + _result: Optional[pb.Result] + ^^^^^^^^^ + AttributeError : module 'wandb.proto.wandb_internal_pb2' has no attribute 'Result' + ^^^^^^^^^ +AttributeError: module 'wandb.proto.wandb_internal_pb2' has no attribute 'Result' +W0725 12:47:03.861000 730 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 799 closing signal SIGTERM +W0725 12:47:03.864000 730 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 800 closing signal SIGTERM +W0725 12:47:03.864000 730 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 801 closing signal SIGTERM +W0725 12:47:03.864000 730 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 802 closing signal SIGTERM +W0725 12:47:03.865000 730 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 803 closing signal SIGTERM +W0725 12:47:03.865000 730 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 804 closing signal SIGTERM +W0725 12:47:03.865000 730 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 805 closing signal SIGTERM +E0725 12:47:03.979000 730 torch/distributed/elastic/multiprocessing/api.py:874] failed (exitcode: 1) local_rank: 0 (pid: 798) of binary: /usr/bin/python3 +Traceback (most recent call last): + File "/usr/local/bin/torchrun", line 7, in + sys.exit(main()) + ^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 355, in wrapper + return f(*args, **kwargs) + ^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 892, in main + run(args) + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 883, in run + elastic_launch( + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 139, in __call__ + return launch_agent(self._config, self._entrypoint, list(args)) + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 270, in launch_agent + raise ChildFailedError( +torch.distributed.elastic.multiprocessing.errors.ChildFailedError: +============================================================ +main_cross_2direct_balance_realscale_perturb_v4_NRealFake.py FAILED +------------------------------------------------------------ +Failures: + +------------------------------------------------------------ +Root Cause (first observed failure): +[0]: + time : 2026-07-25_12:47:03 + host : n116-042-186.byted.org + rank : 0 (local_rank: 0) + exitcode : 1 (pid: 798) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +============================================================ diff --git a/ScaleDF_rebuttal/_host_logs/all_46real_91fake_11451_20260725_1215/train_all_46real_91fake.pid b/ScaleDF_rebuttal/_host_logs/all_46real_91fake_11451_20260725_1215/train_all_46real_91fake.pid new file mode 100644 index 0000000000000000000000000000000000000000..55a98c62cb9ca312bf7e992cfe4d7e14387eae66 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/all_46real_91fake_11451_20260725_1215/train_all_46real_91fake.pid @@ -0,0 +1 @@ +722 diff --git a/ScaleDF_rebuttal/_host_logs/all_46real_91fake_11451_20260725_1215/train_all_46real_91fake_retry1.log b/ScaleDF_rebuttal/_host_logs/all_46real_91fake_11451_20260725_1215/train_all_46real_91fake_retry1.log new file mode 100644 index 0000000000000000000000000000000000000000..b297156eb4eb3a3eb0f63d7a6ef263341994856e --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/all_46real_91fake_11451_20260725_1215/train_all_46real_91fake_retry1.log @@ -0,0 +1,2157 @@ +started_utc=2026-07-25T12:48:33Z +hostname=n116-042-186 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +kind=nrealfake +selection=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/selections/fixed_budget_500k_all_46real_91fake_20260725/scaledf_v4_fixed500000_all_46real_91fake_imgseed20260725.pkl +model_name=deit_base_patch16_LS +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_all_46real_91fake_imgseed20260725_n116-042-186_20260725_1215 +main=main_cross_2direct_balance_realscale_perturb_v4_NRealFake.py +nproc_per_node=8 +batch_per_gpu=256 +W0725 12:48:45.798000 1138 torch/distributed/run.py:766] +W0725 12:48:45.798000 1138 torch/distributed/run.py:766] ***************************************** +W0725 12:48:45.798000 1138 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 12:48:45.798000 1138 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +| distributed init (rank 5): env:// +| distributed init (rank 2): env:// +[rank2]:[W725 12:49:22.550665334 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W725 12:49:22.550665645 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +[rank4]:[W725 12:49:22.894013439 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +[rank6]:[W725 12:49:22.934665964 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 12:49:22.985785874 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +[rank7]:[W725 12:49:22.004467679 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +| distributed init (rank 3): env:// +[rank1]:[W725 12:49:22.029738306 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +n116-042-186:1206:1206 [0] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET6 +n116-042-186:1206:1206 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to =eth0 +n116-042-186:1206:1206 [0] NCCL INFO cudaDriverVersion 12060 +n116-042-186:1206:1206 [0] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-042-186:1206:1206 [0] NCCL INFO Comm config Blocking set to 1 +[rank3]:[W725 12:49:23.042394080 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +n116-042-186:1213:1213 [7] NCCL INFO cudaDriverVersion 12060 +n116-042-186:1213:1213 [7] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET6 +n116-042-186:1213:1213 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to =eth0 +n116-042-186:1213:1213 [7] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-042-186:1213:1213 [7] NCCL INFO Comm config Blocking set to 1 +n116-042-186:1210:1210 [4] NCCL INFO cudaDriverVersion 12060 +n116-042-186:1210:1210 [4] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET6 +n116-042-186:1210:1210 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to =eth0 +n116-042-186:1210:1210 [4] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-042-186:1210:1210 [4] NCCL INFO Comm config Blocking set to 1 +n116-042-186:1212:1212 [6] NCCL INFO cudaDriverVersion 12060 +n116-042-186:1212:1212 [6] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET6 +n116-042-186:1212:1212 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to =eth0 +n116-042-186:1212:1212 [6] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-042-186:1208:1208 [2] NCCL INFO cudaDriverVersion 12060 +n116-042-186:1208:1208 [2] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET6 +n116-042-186:1208:1208 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to =eth0 +n116-042-186:1208:1208 [2] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-042-186:1212:1212 [6] NCCL INFO Comm config Blocking set to 1 +n116-042-186:1207:1207 [1] NCCL INFO cudaDriverVersion 12060 +n116-042-186:1207:1207 [1] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET6 +n116-042-186:1207:1207 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to =eth0 +n116-042-186:1207:1207 [1] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-042-186:1209:1209 [3] NCCL INFO cudaDriverVersion 12060 +n116-042-186:1209:1209 [3] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET6 +n116-042-186:1209:1209 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to =eth0 +n116-042-186:1209:1209 [3] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-042-186:1208:1208 [2] NCCL INFO Comm config Blocking set to 1 +n116-042-186:1207:1207 [1] NCCL INFO Comm config Blocking set to 1 +n116-042-186:1211:1211 [5] NCCL INFO cudaDriverVersion 12060 +n116-042-186:1211:1211 [5] NCCL INFO NCCL_SOCKET_FAMILY set by environment to AF_INET6 +n116-042-186:1211:1211 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to =eth0 +n116-042-186:1209:1209 [3] NCCL INFO Comm config Blocking set to 1 +n116-042-186:1211:1211 [5] NCCL INFO NCCL version 2.26.2+cuda12.2 +n116-042-186:1211:1211 [5] NCCL INFO Comm config Blocking set to 1 +n116-042-186:1208:1267 [2] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-042-186:1211:1270 [5] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-042-186:1209:1269 [3] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-042-186:1207:1268 [1] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-042-186:1211:1270 [5] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-042-186:1206:1263 [0] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-042-186:1208:1267 [2] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-042-186:1213:1264 [7] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-042-186:1210:1265 [4] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-042-186:1211:1270 [5] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-042-186:1208:1267 [2] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-042-186:1207:1268 [1] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-042-186:1211:1270 [5] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-042-186:1209:1269 [3] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-042-186:1208:1267 [2] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-042-186:1207:1268 [1] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-042-186:1212:1266 [6] NCCL INFO NET/Plugin: Plugin name set by env to libnccl-net-gcp-fastrak.so +n116-042-186:1209:1269 [3] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-042-186:1207:1268 [1] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-042-186:1206:1263 [0] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-042-186:1209:1269 [3] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-042-186:1210:1265 [4] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-042-186:1213:1264 [7] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-042-186:1206:1263 [0] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-042-186:1210:1265 [4] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-042-186:1206:1263 [0] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-042-186:1213:1264 [7] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-042-186:1210:1265 [4] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-042-186:1213:1264 [7] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-042-186:1211:1270 [5] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-042-186:1208:1267 [2] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-042-186:1209:1269 [3] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-042-186:1207:1268 [1] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-042-186:1208:1267 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-042-186:1209:1269 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-042-186:1206:1263 [0] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-042-186:1207:1268 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-042-186:1212:1266 [6] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v10 symbol. +n116-042-186:1211:1270 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-042-186:1210:1265 [4] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-042-186:1209:1269 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-042-186:1208:1267 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-042-186:1206:1263 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-042-186:1213:1264 [7] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-042-186:1211:1270 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-042-186:1212:1266 [6] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v9 symbol. +n116-042-186:1206:1263 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-042-186:1213:1264 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-042-186:1211:1270 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-042-186:1207:1268 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-042-186:1210:1265 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-042-186:1206:1263 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-042-186:1211:1270 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-042-186:1209:1269 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-042-186:1208:1267 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-042-186:1209:1269 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-042-186:1213:1264 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-042-186:1207:1268 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-042-186:1212:1266 [6] NCCL INFO NET/Plugin: Failed to find ncclNetPlugin_v8 symbol. +n116-042-186:1206:1263 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-042-186:1210:1265 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-042-186:1211:1270 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-042-186:1208:1267 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-042-186:1209:1269 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-042-186:1207:1268 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-042-186:1213:1264 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-042-186:1206:1263 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-042-186:1210:1265 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-042-186:1212:1266 [6] NCCL INFO NET/Plugin: Loaded net plugin FasTrak (v7) +n116-042-186:1208:1267 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-042-186:1207:1268 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-042-186:1213:1264 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-042-186:1210:1265 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-042-186:1212:1266 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v10 symbol. +n116-042-186:1213:1264 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-042-186:1210:1265 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-042-186:1212:1266 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v9 symbol. +n116-042-186:1212:1266 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v8 symbol. +n116-042-186:1212:1266 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v7 symbol. +n116-042-186:1212:1266 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol. +n116-042-186:1210:1265 [4] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-042-186:1210:1265 [4] NCCL INFO Using network FasTrak +n116-042-186:1207:1268 [1] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-042-186:1207:1268 [1] NCCL INFO Using network FasTrak +n116-042-186:1206:1263 [0] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-042-186:1206:1263 [0] NCCL INFO Using network FasTrak +n116-042-186:1211:1270 [5] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-042-186:1211:1270 [5] NCCL INFO Using network FasTrak +n116-042-186:1213:1264 [7] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-042-186:1213:1264 [7] NCCL INFO Using network FasTrak +n116-042-186:1208:1267 [2] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-042-186:1208:1267 [2] NCCL INFO Using network FasTrak +n116-042-186:1212:1266 [6] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-042-186:1209:1269 [3] NCCL INFO PROFILER/Plugin: Could not find: libnccl-profiler.so. +n116-042-186:1212:1266 [6] NCCL INFO Using network FasTrak +n116-042-186:1209:1269 [3] NCCL INFO Using network FasTrak +n116-042-186:1210:1265 [4] NCCL INFO DMA-BUF is available on GPU device 4 +n116-042-186:1210:1265 [4] NCCL INFO ncclCommInitRankConfig comm 0xb3b2a80 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 84000 commId 0x5e9d7d8d37acfcd1 - Init START +n116-042-186:1206:1263 [0] NCCL INFO DMA-BUF is available on GPU device 0 +n116-042-186:1207:1268 [1] NCCL INFO DMA-BUF is available on GPU device 1 +n116-042-186:1211:1270 [5] NCCL INFO DMA-BUF is available on GPU device 5 +n116-042-186:1212:1266 [6] NCCL INFO DMA-BUF is available on GPU device 6 +n116-042-186:1206:1263 [0] NCCL INFO ncclCommInitRankConfig comm 0xbec8fd0 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 4000 commId 0x5e9d7d8d37acfcd1 - Init START +n116-042-186:1207:1268 [1] NCCL INFO ncclCommInitRankConfig comm 0xb5957c0 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 5000 commId 0x5e9d7d8d37acfcd1 - Init START +n116-042-186:1211:1270 [5] NCCL INFO ncclCommInitRankConfig comm 0xbc38ba0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId 85000 commId 0x5e9d7d8d37acfcd1 - Init START +n116-042-186:1212:1266 [6] NCCL INFO ncclCommInitRankConfig comm 0xb67df70 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId 8b000 commId 0x5e9d7d8d37acfcd1 - Init START +n116-042-186:1208:1267 [2] NCCL INFO DMA-BUF is available on GPU device 2 +n116-042-186:1211:1270 [5] NCCL INFO RAS client listening socket at ::1<28028> +n116-042-186:1213:1264 [7] NCCL INFO DMA-BUF is available on GPU device 7 +n116-042-186:1209:1269 [3] NCCL INFO DMA-BUF is available on GPU device 3 +n116-042-186:1208:1267 [2] NCCL INFO ncclCommInitRankConfig comm 0xb4d5d00 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId b000 commId 0x5e9d7d8d37acfcd1 - Init START +n116-042-186:1213:1264 [7] NCCL INFO ncclCommInitRankConfig comm 0xbf6b490 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId 8c000 commId 0x5e9d7d8d37acfcd1 - Init START +n116-042-186:1209:1269 [3] NCCL INFO ncclCommInitRankConfig comm 0xd05c320 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId c000 commId 0x5e9d7d8d37acfcd1 - Init START +n116-042-186:1213:1264 [7] NCCL INFO RAS client listening socket at ::1<28028> +n116-042-186:1212:1266 [6] NCCL INFO RAS client listening socket at ::1<28028> +n116-042-186:1207:1268 [1] NCCL INFO RAS client listening socket at ::1<28028> +n116-042-186:1206:1263 [0] NCCL INFO RAS client listening socket at ::1<28028> +n116-042-186:1208:1267 [2] NCCL INFO RAS client listening socket at ::1<28028> +n116-042-186:1209:1269 [3] NCCL INFO RAS client listening socket at ::1<28028> +n116-042-186:1210:1265 [4] NCCL INFO RAS client listening socket at ::1<28028> +n116-042-186:1213:1264 [7] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-042-186:1213:1264 [7] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-042-186:1213:1264 [7] NCCL INFO Retrieving state for FasTrak +n116-042-186:1213:1264 [7] NCCL INFO Initialized state 0 for FasTrak +n116-042-186:1213:1264 [7] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-042-186:1213:1264 [7] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-042-186:1213:1264 [7] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-042-186:1213:1264 [7] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-042-186:1213:1264 [7] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-042-186:1213:1264 [7] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-042-186:1213:1264 [7] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-042-186:1213:1264 [7] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-042-186:1212:1266 [6] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-042-186:1212:1266 [6] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-042-186:1212:1266 [6] NCCL INFO Retrieving state for FasTrak +n116-042-186:1212:1266 [6] NCCL INFO Initialized state 0 for FasTrak +n116-042-186:1212:1266 [6] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-042-186:1212:1266 [6] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-042-186:1212:1266 [6] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-042-186:1212:1266 [6] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-042-186:1212:1266 [6] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-042-186:1212:1266 [6] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-042-186:1212:1266 [6] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-042-186:1212:1266 [6] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-042-186:1209:1269 [3] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-042-186:1209:1269 [3] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-042-186:1209:1269 [3] NCCL INFO Retrieving state for FasTrak +n116-042-186:1209:1269 [3] NCCL INFO Initialized state 0 for FasTrak +n116-042-186:1209:1269 [3] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-042-186:1209:1269 [3] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-042-186:1209:1269 [3] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-042-186:1209:1269 [3] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-042-186:1209:1269 [3] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-042-186:1209:1269 [3] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-042-186:1209:1269 [3] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-042-186:1208:1267 [2] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-042-186:1209:1269 [3] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-042-186:1208:1267 [2] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-042-186:1208:1267 [2] NCCL INFO Retrieving state for FasTrak +n116-042-186:1208:1267 [2] NCCL INFO Initialized state 0 for FasTrak +n116-042-186:1208:1267 [2] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-042-186:1208:1267 [2] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-042-186:1208:1267 [2] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-042-186:1208:1267 [2] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-042-186:1208:1267 [2] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-042-186:1208:1267 [2] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-042-186:1208:1267 [2] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-042-186:1208:1267 [2] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-042-186:1210:1265 [4] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-042-186:1211:1270 [5] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-042-186:1210:1265 [4] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-042-186:1210:1265 [4] NCCL INFO Retrieving state for FasTrak +n116-042-186:1210:1265 [4] NCCL INFO Initialized state 0 for FasTrak +n116-042-186:1211:1270 [5] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-042-186:1211:1270 [5] NCCL INFO Retrieving state for FasTrak +n116-042-186:1211:1270 [5] NCCL INFO Initialized state 0 for FasTrak +n116-042-186:1210:1265 [4] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-042-186:1211:1270 [5] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-042-186:1210:1265 [4] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-042-186:1211:1270 [5] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-042-186:1210:1265 [4] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-042-186:1211:1270 [5] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-042-186:1207:1268 [1] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-042-186:1207:1268 [1] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-042-186:1207:1268 [1] NCCL INFO Retrieving state for FasTrak +n116-042-186:1207:1268 [1] NCCL INFO Initialized state 0 for FasTrak +n116-042-186:1210:1265 [4] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-042-186:1211:1270 [5] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-042-186:1207:1268 [1] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-042-186:1210:1265 [4] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-042-186:1211:1270 [5] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-042-186:1207:1268 [1] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-042-186:1210:1265 [4] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-042-186:1211:1270 [5] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-042-186:1207:1268 [1] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-042-186:1210:1265 [4] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-042-186:1211:1270 [5] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-042-186:1207:1268 [1] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-042-186:1210:1265 [4] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-042-186:1211:1270 [5] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-042-186:1207:1268 [1] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-042-186:1207:1268 [1] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-042-186:1207:1268 [1] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-042-186:1207:1268 [1] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-042-186:1206:1263 [0] NCCL INFO Could not find real path of /sys/class/pci_bus/fffffff/../../fffffff:ff:f +n116-042-186:1206:1263 [0] NCCL INFO TOPO/NET : Importing network plugins to topology +n116-042-186:1206:1263 [0] NCCL INFO Retrieving state for FasTrak +n116-042-186:1206:1263 [0] NCCL INFO Initialized state 0 for FasTrak +n116-042-186:1206:1263 [0] NCCL INFO ncclTopoPopulateNics : Filled eth1 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:02.0/0000:06:00.0 keep=1 coll=(null) +n116-042-186:1206:1263 [0] NCCL INFO ncclTopoPopulateNics : Filled eth2 in topo with pciPath=/sys/devices/pci0000:01/0000:01:00.0/0000:02:00.0/0000:03:03.0/0000:07:00.0 keep=1 coll=(null) +n116-042-186:1206:1263 [0] NCCL INFO ncclTopoPopulateNics : Filled eth3 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:02.0/0000:0d:00.0 keep=1 coll=(null) +n116-042-186:1206:1263 [0] NCCL INFO ncclTopoPopulateNics : Filled eth4 in topo with pciPath=/sys/devices/pci0000:08/0000:08:00.0/0000:09:00.0/0000:0a:03.0/0000:0e:00.0 keep=1 coll=(null) +n116-042-186:1206:1263 [0] NCCL INFO ncclTopoPopulateNics : Filled eth5 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:02.0/0000:86:00.0 keep=1 coll=(null) +n116-042-186:1206:1263 [0] NCCL INFO ncclTopoPopulateNics : Filled eth6 in topo with pciPath=/sys/devices/pci0000:81/0000:81:00.0/0000:82:00.0/0000:83:03.0/0000:87:00.0 keep=1 coll=(null) +n116-042-186:1206:1263 [0] NCCL INFO ncclTopoPopulateNics : Filled eth7 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:02.0/0000:8d:00.0 keep=1 coll=(null) +n116-042-186:1206:1263 [0] NCCL INFO ncclTopoPopulateNics : Filled eth8 in topo with pciPath=/sys/devices/pci0000:88/0000:88:00.0/0000:89:00.0/0000:8a:03.0/0000:8e:00.0 keep=1 coll=(null) +n116-042-186:1210:1265 [4] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-042-186:1209:1269 [3] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-042-186:1208:1267 [2] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-042-186:1211:1270 [5] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-042-186:1212:1266 [6] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-042-186:1207:1268 [1] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-042-186:1206:1263 [0] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-042-186:1209:1269 [3] NCCL INFO === System : maxBw 370.8 totalBw 370.8 === +n116-042-186:1209:1269 [3] NCCL INFO CPU/0-0 (1/1/3) +n116-042-186:1209:1269 [3] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-042-186:1209:1269 [3] NCCL INFO + PCI[24.0] - GPU/0-4000 (0) +n116-042-186:1209:1269 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1209:1269 [3] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-042-186:1209:1269 [3] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-042-186:1209:1269 [3] NCCL INFO + PCI[24.0] - GPU/0-5000 (1) +n116-042-186:1209:1269 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1209:1269 [3] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-042-186:1209:1269 [3] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-042-186:1209:1269 [3] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-042-186:1209:1269 [3] NCCL INFO + PCI[24.0] - GPU/0-b000 (2) +n116-042-186:1209:1269 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1209:1269 [3] NCCL INFO + PCI[24.0] - GPU/0-c000 (3) +n116-042-186:1209:1269 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1209:1269 [3] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-042-186:1209:1269 [3] NCCL INFO CPU/0-1 (1/1/3) +n116-042-186:1209:1269 [3] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-042-186:1209:1269 [3] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-042-186:1209:1269 [3] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-042-186:1209:1269 [3] NCCL INFO + PCI[24.0] - GPU/0-84000 (4) +n116-042-186:1209:1269 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1209:1269 [3] NCCL INFO + PCI[24.0] - GPU/0-85000 (5) +n116-042-186:1209:1269 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1209:1269 [3] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-042-186:1209:1269 [3] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-042-186:1209:1269 [3] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-042-186:1209:1269 [3] NCCL INFO + PCI[24.0] - GPU/0-8b000 (6) +n116-042-186:1209:1269 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1209:1269 [3] NCCL INFO + PCI[24.0] - GPU/0-8c000 (7) +n116-042-186:1209:1269 [3] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1209:1269 [3] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-042-186:1209:1269 [3] NCCL INFO ========================================== +n116-042-186:1209:1269 [3] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) +n116-042-186:1209:1269 [3] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) +n116-042-186:1209:1269 [3] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) +n116-042-186:1209:1269 [3] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) +n116-042-186:1209:1269 [3] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) +n116-042-186:1209:1269 [3] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) +n116-042-186:1209:1269 [3] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) +n116-042-186:1209:1269 [3] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) +n116-042-186:1209:1269 [3] NCCL INFO Setting affinity for GPU 3 to 0fffffff,fffffe00,00000000,000fffff,fffffffe +n116-042-186:1210:1265 [4] NCCL INFO === System : maxBw 370.8 totalBw 370.8 === +n116-042-186:1210:1265 [4] NCCL INFO CPU/0-0 (1/1/3) +n116-042-186:1210:1265 [4] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-042-186:1210:1265 [4] NCCL INFO + PCI[24.0] - GPU/0-4000 (0) +n116-042-186:1210:1265 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1210:1265 [4] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-042-186:1210:1265 [4] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-042-186:1210:1265 [4] NCCL INFO + PCI[24.0] - GPU/0-5000 (1) +n116-042-186:1210:1265 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1210:1265 [4] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-042-186:1210:1265 [4] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-042-186:1210:1265 [4] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-042-186:1210:1265 [4] NCCL INFO + PCI[24.0] - GPU/0-b000 (2) +n116-042-186:1210:1265 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1210:1265 [4] NCCL INFO + PCI[24.0] - GPU/0-c000 (3) +n116-042-186:1210:1265 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1210:1265 [4] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-042-186:1210:1265 [4] NCCL INFO CPU/0-1 (1/1/3) +n116-042-186:1210:1265 [4] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-042-186:1210:1265 [4] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-042-186:1210:1265 [4] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-042-186:1210:1265 [4] NCCL INFO + PCI[24.0] - GPU/0-84000 (4) +n116-042-186:1210:1265 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1210:1265 [4] NCCL INFO + PCI[24.0] - GPU/0-85000 (5) +n116-042-186:1210:1265 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1209:1269 [3] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-042-186:1210:1265 [4] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-042-186:1210:1265 [4] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-042-186:1210:1265 [4] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-042-186:1210:1265 [4] NCCL INFO + PCI[24.0] - GPU/0-8b000 (6) +n116-042-186:1210:1265 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1210:1265 [4] NCCL INFO + PCI[24.0] - GPU/0-8c000 (7) +n116-042-186:1210:1265 [4] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1210:1265 [4] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-042-186:1210:1265 [4] NCCL INFO ========================================== +n116-042-186:1210:1265 [4] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) +n116-042-186:1210:1265 [4] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) +n116-042-186:1210:1265 [4] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) +n116-042-186:1210:1265 [4] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) +n116-042-186:1210:1265 [4] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) +n116-042-186:1210:1265 [4] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) +n116-042-186:1210:1265 [4] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) +n116-042-186:1210:1265 [4] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) +n116-042-186:1210:1265 [4] NCCL INFO Setting affinity for GPU 4 to 7fff,ffffffff,e0000000,0000007f,ffffffff,ffe00000,00000000 +n116-042-186:1210:1265 [4] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-042-186:1208:1267 [2] NCCL INFO === System : maxBw 370.8 totalBw 370.8 === +n116-042-186:1208:1267 [2] NCCL INFO CPU/0-0 (1/1/3) +n116-042-186:1208:1267 [2] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-042-186:1208:1267 [2] NCCL INFO + PCI[24.0] - GPU/0-4000 (0) +n116-042-186:1208:1267 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1208:1267 [2] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-042-186:1208:1267 [2] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-042-186:1208:1267 [2] NCCL INFO + PCI[24.0] - GPU/0-5000 (1) +n116-042-186:1208:1267 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1208:1267 [2] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-042-186:1208:1267 [2] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-042-186:1208:1267 [2] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-042-186:1208:1267 [2] NCCL INFO + PCI[24.0] - GPU/0-b000 (2) +n116-042-186:1208:1267 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1208:1267 [2] NCCL INFO + PCI[24.0] - GPU/0-c000 (3) +n116-042-186:1208:1267 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1208:1267 [2] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-042-186:1208:1267 [2] NCCL INFO CPU/0-1 (1/1/3) +n116-042-186:1208:1267 [2] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-042-186:1208:1267 [2] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-042-186:1208:1267 [2] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-042-186:1208:1267 [2] NCCL INFO + PCI[24.0] - GPU/0-84000 (4) +n116-042-186:1208:1267 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1208:1267 [2] NCCL INFO + PCI[24.0] - GPU/0-85000 (5) +n116-042-186:1208:1267 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1208:1267 [2] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-042-186:1208:1267 [2] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-042-186:1208:1267 [2] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-042-186:1208:1267 [2] NCCL INFO + PCI[24.0] - GPU/0-8b000 (6) +n116-042-186:1208:1267 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1208:1267 [2] NCCL INFO + PCI[24.0] - GPU/0-8c000 (7) +n116-042-186:1208:1267 [2] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1208:1267 [2] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-042-186:1208:1267 [2] NCCL INFO ========================================== +n116-042-186:1208:1267 [2] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) +n116-042-186:1208:1267 [2] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) +n116-042-186:1208:1267 [2] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) +n116-042-186:1213:1264 [7] NCCL INFO NCCL_NET_GDR_LEVEL set by environment to PIX +n116-042-186:1208:1267 [2] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) +n116-042-186:1208:1267 [2] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) +n116-042-186:1208:1267 [2] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) +n116-042-186:1208:1267 [2] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) +n116-042-186:1208:1267 [2] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) +n116-042-186:1208:1267 [2] NCCL INFO Setting affinity for GPU 2 to 0fffffff,fffffe00,00000000,000fffff,fffffffe +n116-042-186:1212:1266 [6] NCCL INFO === System : maxBw 370.8 totalBw 370.8 === +n116-042-186:1212:1266 [6] NCCL INFO CPU/0-0 (1/1/3) +n116-042-186:1212:1266 [6] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-042-186:1212:1266 [6] NCCL INFO + PCI[24.0] - GPU/0-4000 (0) +n116-042-186:1212:1266 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1212:1266 [6] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-042-186:1212:1266 [6] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-042-186:1212:1266 [6] NCCL INFO + PCI[24.0] - GPU/0-5000 (1) +n116-042-186:1212:1266 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1212:1266 [6] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-042-186:1212:1266 [6] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-042-186:1212:1266 [6] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-042-186:1212:1266 [6] NCCL INFO + PCI[24.0] - GPU/0-b000 (2) +n116-042-186:1212:1266 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1212:1266 [6] NCCL INFO + PCI[24.0] - GPU/0-c000 (3) +n116-042-186:1212:1266 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1212:1266 [6] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-042-186:1212:1266 [6] NCCL INFO CPU/0-1 (1/1/3) +n116-042-186:1212:1266 [6] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-042-186:1212:1266 [6] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-042-186:1212:1266 [6] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-042-186:1212:1266 [6] NCCL INFO + PCI[24.0] - GPU/0-84000 (4) +n116-042-186:1212:1266 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1212:1266 [6] NCCL INFO + PCI[24.0] - GPU/0-85000 (5) +n116-042-186:1212:1266 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1212:1266 [6] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-042-186:1212:1266 [6] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-042-186:1212:1266 [6] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-042-186:1212:1266 [6] NCCL INFO + PCI[24.0] - GPU/0-8b000 (6) +n116-042-186:1212:1266 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1212:1266 [6] NCCL INFO + PCI[24.0] - GPU/0-8c000 (7) +n116-042-186:1212:1266 [6] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1212:1266 [6] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-042-186:1212:1266 [6] NCCL INFO ========================================== +n116-042-186:1211:1270 [5] NCCL INFO === System : maxBw 370.8 totalBw 370.8 === +n116-042-186:1212:1266 [6] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) +n116-042-186:1211:1270 [5] NCCL INFO CPU/0-0 (1/1/3) +n116-042-186:1212:1266 [6] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) +n116-042-186:1211:1270 [5] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-042-186:1211:1270 [5] NCCL INFO + PCI[24.0] - GPU/0-4000 (0) +n116-042-186:1211:1270 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1212:1266 [6] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) +n116-042-186:1211:1270 [5] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-042-186:1211:1270 [5] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-042-186:1211:1270 [5] NCCL INFO + PCI[24.0] - GPU/0-5000 (1) +n116-042-186:1212:1266 [6] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) +n116-042-186:1211:1270 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1211:1270 [5] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-042-186:1211:1270 [5] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-042-186:1211:1270 [5] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-042-186:1212:1266 [6] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) +n116-042-186:1211:1270 [5] NCCL INFO + PCI[24.0] - GPU/0-b000 (2) +n116-042-186:1211:1270 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1212:1266 [6] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) +n116-042-186:1211:1270 [5] NCCL INFO + PCI[24.0] - GPU/0-c000 (3) +n116-042-186:1211:1270 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1211:1270 [5] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-042-186:1211:1270 [5] NCCL INFO CPU/0-1 (1/1/3) +n116-042-186:1212:1266 [6] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) +n116-042-186:1211:1270 [5] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-042-186:1211:1270 [5] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-042-186:1211:1270 [5] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-042-186:1212:1266 [6] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) +n116-042-186:1211:1270 [5] NCCL INFO + PCI[24.0] - GPU/0-84000 (4) +n116-042-186:1211:1270 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1211:1270 [5] NCCL INFO + PCI[24.0] - GPU/0-85000 (5) +n116-042-186:1211:1270 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1211:1270 [5] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-042-186:1212:1266 [6] NCCL INFO Setting affinity for GPU 6 to 7fff,ffffffff,e0000000,0000007f,ffffffff,ffe00000,00000000 +n116-042-186:1211:1270 [5] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-042-186:1211:1270 [5] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-042-186:1211:1270 [5] NCCL INFO + PCI[24.0] - GPU/0-8b000 (6) +n116-042-186:1211:1270 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1211:1270 [5] NCCL INFO + PCI[24.0] - GPU/0-8c000 (7) +n116-042-186:1211:1270 [5] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1211:1270 [5] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-042-186:1211:1270 [5] NCCL INFO ========================================== +n116-042-186:1211:1270 [5] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) +n116-042-186:1211:1270 [5] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) +n116-042-186:1211:1270 [5] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) +n116-042-186:1208:1267 [2] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-042-186:1211:1270 [5] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) +n116-042-186:1207:1268 [1] NCCL INFO === System : maxBw 370.8 totalBw 370.8 === +n116-042-186:1207:1268 [1] NCCL INFO CPU/0-0 (1/1/3) +n116-042-186:1211:1270 [5] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) +n116-042-186:1211:1270 [5] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) +n116-042-186:1207:1268 [1] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-042-186:1207:1268 [1] NCCL INFO + PCI[24.0] - GPU/0-4000 (0) +n116-042-186:1211:1270 [5] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) +n116-042-186:1207:1268 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1207:1268 [1] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-042-186:1207:1268 [1] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-042-186:1207:1268 [1] NCCL INFO + PCI[24.0] - GPU/0-5000 (1) +n116-042-186:1207:1268 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1211:1270 [5] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) +n116-042-186:1207:1268 [1] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-042-186:1207:1268 [1] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-042-186:1207:1268 [1] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-042-186:1207:1268 [1] NCCL INFO + PCI[24.0] - GPU/0-b000 (2) +n116-042-186:1207:1268 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1207:1268 [1] NCCL INFO + PCI[24.0] - GPU/0-c000 (3) +n116-042-186:1207:1268 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1207:1268 [1] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-042-186:1211:1270 [5] NCCL INFO Setting affinity for GPU 5 to 7fff,ffffffff,e0000000,0000007f,ffffffff,ffe00000,00000000 +n116-042-186:1207:1268 [1] NCCL INFO CPU/0-1 (1/1/3) +n116-042-186:1207:1268 [1] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-042-186:1207:1268 [1] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-042-186:1207:1268 [1] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-042-186:1207:1268 [1] NCCL INFO + PCI[24.0] - GPU/0-84000 (4) +n116-042-186:1207:1268 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1212:1266 [6] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-042-186:1207:1268 [1] NCCL INFO + PCI[24.0] - GPU/0-85000 (5) +n116-042-186:1211:1270 [5] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-042-186:1207:1268 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1207:1268 [1] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-042-186:1207:1268 [1] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-042-186:1207:1268 [1] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-042-186:1207:1268 [1] NCCL INFO + PCI[24.0] - GPU/0-8b000 (6) +n116-042-186:1207:1268 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1207:1268 [1] NCCL INFO + PCI[24.0] - GPU/0-8c000 (7) +n116-042-186:1207:1268 [1] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1207:1268 [1] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-042-186:1207:1268 [1] NCCL INFO ========================================== +n116-042-186:1207:1268 [1] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) +n116-042-186:1207:1268 [1] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) +n116-042-186:1206:1263 [0] NCCL INFO === System : maxBw 370.8 totalBw 370.8 === +n116-042-186:1206:1263 [0] NCCL INFO CPU/0-0 (1/1/3) +n116-042-186:1206:1263 [0] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-042-186:1206:1263 [0] NCCL INFO + PCI[24.0] - GPU/0-4000 (0) +n116-042-186:1206:1263 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1206:1263 [0] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-042-186:1206:1263 [0] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-042-186:1206:1263 [0] NCCL INFO + PCI[24.0] - GPU/0-5000 (1) +n116-042-186:1206:1263 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1206:1263 [0] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-042-186:1206:1263 [0] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-042-186:1206:1263 [0] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-042-186:1206:1263 [0] NCCL INFO + PCI[24.0] - GPU/0-b000 (2) +n116-042-186:1206:1263 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1206:1263 [0] NCCL INFO + PCI[24.0] - GPU/0-c000 (3) +n116-042-186:1206:1263 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1206:1263 [0] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-042-186:1206:1263 [0] NCCL INFO CPU/0-1 (1/1/3) +n116-042-186:1206:1263 [0] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-042-186:1206:1263 [0] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-042-186:1206:1263 [0] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-042-186:1206:1263 [0] NCCL INFO + PCI[24.0] - GPU/0-84000 (4) +n116-042-186:1206:1263 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1206:1263 [0] NCCL INFO + PCI[24.0] - GPU/0-85000 (5) +n116-042-186:1206:1263 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1206:1263 [0] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-042-186:1206:1263 [0] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-042-186:1206:1263 [0] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-042-186:1206:1263 [0] NCCL INFO + PCI[24.0] - GPU/0-8b000 (6) +n116-042-186:1206:1263 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1206:1263 [0] NCCL INFO + PCI[24.0] - GPU/0-8c000 (7) +n116-042-186:1206:1263 [0] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1206:1263 [0] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-042-186:1206:1263 [0] NCCL INFO ========================================== +n116-042-186:1206:1263 [0] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) +n116-042-186:1206:1263 [0] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) +n116-042-186:1206:1263 [0] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) +n116-042-186:1206:1263 [0] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) +n116-042-186:1206:1263 [0] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) +n116-042-186:1206:1263 [0] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) +n116-042-186:1206:1263 [0] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) +n116-042-186:1206:1263 [0] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) +n116-042-186:1206:1263 [0] NCCL INFO Setting affinity for GPU 0 to 0fffffff,fffffe00,00000000,000fffff,fffffffe +n116-042-186:1206:1263 [0] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-042-186:1207:1268 [1] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) +n116-042-186:1207:1268 [1] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) +n116-042-186:1207:1268 [1] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) +n116-042-186:1207:1268 [1] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) +n116-042-186:1207:1268 [1] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) +n116-042-186:1207:1268 [1] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) +n116-042-186:1207:1268 [1] NCCL INFO Setting affinity for GPU 1 to 0fffffff,fffffe00,00000000,000fffff,fffffffe +n116-042-186:1213:1264 [7] NCCL INFO === System : maxBw 370.8 totalBw 370.8 === +n116-042-186:1213:1264 [7] NCCL INFO CPU/0-0 (1/1/3) +n116-042-186:1213:1264 [7] NCCL INFO + PCI[24.0] - PCI/0-2000 (10b5879610b58796) +n116-042-186:1213:1264 [7] NCCL INFO + PCI[24.0] - GPU/0-4000 (0) +n116-042-186:1213:1264 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1213:1264 [7] NCCL INFO + PCI[24.0] - NIC/0-6000 +n116-042-186:1213:1264 [7] NCCL INFO + PCI[24.0] - NIC/0-7000 +n116-042-186:1213:1264 [7] NCCL INFO + PCI[24.0] - GPU/0-5000 (1) +n116-042-186:1213:1264 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1213:1264 [7] NCCL INFO + PCI[24.0] - PCI/0-9000 (10b5879610b58796) +n116-042-186:1213:1264 [7] NCCL INFO + PCI[24.0] - NIC/0-d000 +n116-042-186:1213:1264 [7] NCCL INFO + PCI[24.0] - NIC/0-e000 +n116-042-186:1213:1264 [7] NCCL INFO + PCI[24.0] - GPU/0-b000 (2) +n116-042-186:1213:1264 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1213:1264 [7] NCCL INFO + PCI[24.0] - GPU/0-c000 (3) +n116-042-186:1213:1264 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1213:1264 [7] NCCL INFO + SYS[22.0] - CPU/0-1 +n116-042-186:1213:1264 [7] NCCL INFO CPU/0-1 (1/1/3) +n116-042-186:1213:1264 [7] NCCL INFO + PCI[24.0] - PCI/0-82000 (10b5879610b58796) +n116-042-186:1213:1264 [7] NCCL INFO + PCI[24.0] - NIC/0-86000 +n116-042-186:1213:1264 [7] NCCL INFO + PCI[24.0] - NIC/0-87000 +n116-042-186:1213:1264 [7] NCCL INFO + PCI[24.0] - GPU/0-84000 (4) +n116-042-186:1213:1264 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1213:1264 [7] NCCL INFO + PCI[24.0] - GPU/0-85000 (5) +n116-042-186:1213:1264 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1213:1264 [7] NCCL INFO + PCI[24.0] - PCI/0-89000 (10b5879610b58796) +n116-042-186:1213:1264 [7] NCCL INFO + PCI[24.0] - NIC/0-8d000 +n116-042-186:1213:1264 [7] NCCL INFO + PCI[24.0] - NIC/0-8e000 +n116-042-186:1213:1264 [7] NCCL INFO + PCI[24.0] - GPU/0-8b000 (6) +n116-042-186:1213:1264 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1213:1264 [7] NCCL INFO + PCI[24.0] - GPU/0-8c000 (7) +n116-042-186:1213:1264 [7] NCCL INFO + NVL[370.8] - NVS/0-0 +n116-042-186:1213:1264 [7] NCCL INFO + SYS[22.0] - CPU/0-0 +n116-042-186:1213:1264 [7] NCCL INFO ========================================== +n116-042-186:1213:1264 [7] NCCL INFO GPU/0-4000 :GPU/0-4000 (0/5000.0/LOC) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) +n116-042-186:1213:1264 [7] NCCL INFO GPU/0-5000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (0/5000.0/LOC) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) +n116-042-186:1213:1264 [7] NCCL INFO GPU/0-b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (0/5000.0/LOC) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) +n116-042-186:1213:1264 [7] NCCL INFO GPU/0-c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (0/5000.0/LOC) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (2/24.0/PHB) CPU/0-1 (3/22.0/SYS) +n116-042-186:1213:1264 [7] NCCL INFO GPU/0-84000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (0/5000.0/LOC) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) +n116-042-186:1213:1264 [7] NCCL INFO GPU/0-85000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (0/5000.0/LOC) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) +n116-042-186:1213:1264 [7] NCCL INFO GPU/0-8b000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (0/5000.0/LOC) GPU/0-8c000 (2/370.8/NVL) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) +n116-042-186:1213:1264 [7] NCCL INFO GPU/0-8c000 :GPU/0-4000 (2/370.8/NVL) GPU/0-5000 (2/370.8/NVL) GPU/0-b000 (2/370.8/NVL) GPU/0-c000 (2/370.8/NVL) GPU/0-84000 (2/370.8/NVL) GPU/0-85000 (2/370.8/NVL) GPU/0-8b000 (2/370.8/NVL) GPU/0-8c000 (0/5000.0/LOC) NVS/0-0 (1/370.8/NVL) CPU/0-0 (3/22.0/SYS) CPU/0-1 (2/24.0/PHB) +n116-042-186:1213:1264 [7] NCCL INFO Setting affinity for GPU 7 to 7fff,ffffffff,e0000000,0000007f,ffffffff,ffe00000,00000000 +n116-042-186:1213:1264 [7] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-042-186:1207:1268 [1] NCCL INFO NCCL_NVLS_ENABLE set by environment to 0. +n116-042-186:1210:1265 [4] NCCL INFO Pattern 4, crossNic 0, nChannels 12, bw 30.000000/30.000000, type NVL/PIX, sameChannels 1 +n116-042-186:1210:1265 [4] NCCL INFO 0 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1210:1265 [4] NCCL INFO 1 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1210:1265 [4] NCCL INFO 2 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1210:1265 [4] NCCL INFO 3 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1210:1265 [4] NCCL INFO 4 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1210:1265 [4] NCCL INFO 5 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1210:1265 [4] NCCL INFO 6 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1210:1265 [4] NCCL INFO 7 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1210:1265 [4] NCCL INFO 8 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1210:1265 [4] NCCL INFO 9 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1210:1265 [4] NCCL INFO 10 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1210:1265 [4] NCCL INFO 11 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1210:1265 [4] NCCL INFO Pattern 1, crossNic 0, nChannels 12, bw 30.000000/30.000000, type NVL/PIX, sameChannels 1 +n116-042-186:1210:1265 [4] NCCL INFO 0 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1210:1265 [4] NCCL INFO 1 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1210:1265 [4] NCCL INFO 2 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1210:1265 [4] NCCL INFO 3 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1210:1265 [4] NCCL INFO 4 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1210:1265 [4] NCCL INFO 5 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1210:1265 [4] NCCL INFO 6 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1210:1265 [4] NCCL INFO 7 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1210:1265 [4] NCCL INFO 8 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1210:1265 [4] NCCL INFO 9 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1210:1265 [4] NCCL INFO 10 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1210:1265 [4] NCCL INFO 11 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1208:1267 [2] NCCL INFO Pattern 4, crossNic 0, nChannels 12, bw 30.000000/30.000000, type NVL/PIX, sameChannels 1 +n116-042-186:1208:1267 [2] NCCL INFO 0 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1208:1267 [2] NCCL INFO 1 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1208:1267 [2] NCCL INFO 2 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1208:1267 [2] NCCL INFO 3 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1208:1267 [2] NCCL INFO 4 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1208:1267 [2] NCCL INFO 5 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1208:1267 [2] NCCL INFO 6 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1208:1267 [2] NCCL INFO 7 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1208:1267 [2] NCCL INFO 8 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1208:1267 [2] NCCL INFO 9 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1208:1267 [2] NCCL INFO 10 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1208:1267 [2] NCCL INFO 11 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1208:1267 [2] NCCL INFO Pattern 1, crossNic 0, nChannels 12, bw 30.000000/30.000000, type NVL/PIX, sameChannels 1 +n116-042-186:1208:1267 [2] NCCL INFO 0 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1208:1267 [2] NCCL INFO 1 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1208:1267 [2] NCCL INFO 2 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1208:1267 [2] NCCL INFO 3 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1208:1267 [2] NCCL INFO 4 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1208:1267 [2] NCCL INFO 5 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1208:1267 [2] NCCL INFO 6 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1211:1270 [5] NCCL INFO Pattern 4, crossNic 0, nChannels 12, bw 30.000000/30.000000, type NVL/PIX, sameChannels 1 +n116-042-186:1208:1267 [2] NCCL INFO 7 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1208:1267 [2] NCCL INFO 8 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1211:1270 [5] NCCL INFO 0 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1208:1267 [2] NCCL INFO 9 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1211:1270 [5] NCCL INFO 1 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1208:1267 [2] NCCL INFO 10 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1211:1270 [5] NCCL INFO 2 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1208:1267 [2] NCCL INFO 11 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1211:1270 [5] NCCL INFO 3 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1211:1270 [5] NCCL INFO 4 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1211:1270 [5] NCCL INFO 5 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1211:1270 [5] NCCL INFO 6 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1211:1270 [5] NCCL INFO 7 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1211:1270 [5] NCCL INFO 8 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1211:1270 [5] NCCL INFO 9 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1211:1270 [5] NCCL INFO 10 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1211:1270 [5] NCCL INFO 11 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1212:1266 [6] NCCL INFO Pattern 4, crossNic 0, nChannels 12, bw 30.000000/30.000000, type NVL/PIX, sameChannels 1 +n116-042-186:1211:1270 [5] NCCL INFO Pattern 1, crossNic 0, nChannels 12, bw 30.000000/30.000000, type NVL/PIX, sameChannels 1 +n116-042-186:1211:1270 [5] NCCL INFO 0 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1211:1270 [5] NCCL INFO 1 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1211:1270 [5] NCCL INFO 2 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1212:1266 [6] NCCL INFO 0 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1211:1270 [5] NCCL INFO 3 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1211:1270 [5] NCCL INFO 4 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1212:1266 [6] NCCL INFO 1 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1211:1270 [5] NCCL INFO 5 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1212:1266 [6] NCCL INFO 2 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1211:1270 [5] NCCL INFO 6 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1212:1266 [6] NCCL INFO 3 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1211:1270 [5] NCCL INFO 7 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1212:1266 [6] NCCL INFO 4 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1211:1270 [5] NCCL INFO 8 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1212:1266 [6] NCCL INFO 5 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1211:1270 [5] NCCL INFO 9 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1212:1266 [6] NCCL INFO 6 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1211:1270 [5] NCCL INFO 10 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1212:1266 [6] NCCL INFO 7 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1211:1270 [5] NCCL INFO 11 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1212:1266 [6] NCCL INFO 8 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1212:1266 [6] NCCL INFO 9 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1212:1266 [6] NCCL INFO 10 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1212:1266 [6] NCCL INFO 11 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1212:1266 [6] NCCL INFO Pattern 1, crossNic 0, nChannels 12, bw 30.000000/30.000000, type NVL/PIX, sameChannels 1 +n116-042-186:1212:1266 [6] NCCL INFO 0 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1212:1266 [6] NCCL INFO 1 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1212:1266 [6] NCCL INFO 2 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1212:1266 [6] NCCL INFO 3 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1212:1266 [6] NCCL INFO 4 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1212:1266 [6] NCCL INFO 5 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1212:1266 [6] NCCL INFO 6 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1212:1266 [6] NCCL INFO 7 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1212:1266 [6] NCCL INFO 8 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1212:1266 [6] NCCL INFO 9 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1212:1266 [6] NCCL INFO 10 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1212:1266 [6] NCCL INFO 11 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1213:1264 [7] NCCL INFO Pattern 4, crossNic 0, nChannels 12, bw 30.000000/30.000000, type NVL/PIX, sameChannels 1 +n116-042-186:1213:1264 [7] NCCL INFO 0 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1213:1264 [7] NCCL INFO 1 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1213:1264 [7] NCCL INFO 2 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1213:1264 [7] NCCL INFO 3 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1213:1264 [7] NCCL INFO 4 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1213:1264 [7] NCCL INFO 5 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1213:1264 [7] NCCL INFO 6 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1213:1264 [7] NCCL INFO 7 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1213:1264 [7] NCCL INFO 8 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1213:1264 [7] NCCL INFO 9 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1213:1264 [7] NCCL INFO 10 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1213:1264 [7] NCCL INFO 11 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1213:1264 [7] NCCL INFO Pattern 1, crossNic 0, nChannels 12, bw 30.000000/30.000000, type NVL/PIX, sameChannels 1 +n116-042-186:1213:1264 [7] NCCL INFO 0 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1213:1264 [7] NCCL INFO 1 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1213:1264 [7] NCCL INFO 2 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1213:1264 [7] NCCL INFO 3 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1213:1264 [7] NCCL INFO 4 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1213:1264 [7] NCCL INFO 5 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1213:1264 [7] NCCL INFO 6 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1213:1264 [7] NCCL INFO 7 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1213:1264 [7] NCCL INFO 8 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1213:1264 [7] NCCL INFO 9 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1213:1264 [7] NCCL INFO 10 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1213:1264 [7] NCCL INFO 11 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1207:1268 [1] NCCL INFO Pattern 4, crossNic 0, nChannels 12, bw 30.000000/30.000000, type NVL/PIX, sameChannels 1 +n116-042-186:1207:1268 [1] NCCL INFO 0 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1207:1268 [1] NCCL INFO 1 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1207:1268 [1] NCCL INFO 2 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1207:1268 [1] NCCL INFO 3 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1207:1268 [1] NCCL INFO 4 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1207:1268 [1] NCCL INFO 5 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1207:1268 [1] NCCL INFO 6 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1207:1268 [1] NCCL INFO 7 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1207:1268 [1] NCCL INFO 8 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1207:1268 [1] NCCL INFO 9 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1207:1268 [1] NCCL INFO 10 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1207:1268 [1] NCCL INFO 11 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1207:1268 [1] NCCL INFO Pattern 1, crossNic 0, nChannels 12, bw 30.000000/30.000000, type NVL/PIX, sameChannels 1 +n116-042-186:1207:1268 [1] NCCL INFO 0 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1207:1268 [1] NCCL INFO 1 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1207:1268 [1] NCCL INFO 2 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1207:1268 [1] NCCL INFO 3 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1207:1268 [1] NCCL INFO 4 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1207:1268 [1] NCCL INFO 5 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1207:1268 [1] NCCL INFO 6 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1207:1268 [1] NCCL INFO 7 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1207:1268 [1] NCCL INFO 8 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1207:1268 [1] NCCL INFO 9 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1207:1268 [1] NCCL INFO 10 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1207:1268 [1] NCCL INFO 11 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1209:1269 [3] NCCL INFO Pattern 4, crossNic 0, nChannels 12, bw 30.000000/30.000000, type NVL/PIX, sameChannels 1 +n116-042-186:1209:1269 [3] NCCL INFO 0 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1209:1269 [3] NCCL INFO 1 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1209:1269 [3] NCCL INFO 2 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1209:1269 [3] NCCL INFO 3 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1209:1269 [3] NCCL INFO 4 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1209:1269 [3] NCCL INFO 5 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1209:1269 [3] NCCL INFO 6 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1209:1269 [3] NCCL INFO 7 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1209:1269 [3] NCCL INFO 8 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1209:1269 [3] NCCL INFO 9 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1209:1269 [3] NCCL INFO 10 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1209:1269 [3] NCCL INFO 11 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1209:1269 [3] NCCL INFO Pattern 1, crossNic 0, nChannels 12, bw 30.000000/30.000000, type NVL/PIX, sameChannels 1 +n116-042-186:1209:1269 [3] NCCL INFO 0 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1209:1269 [3] NCCL INFO 1 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1209:1269 [3] NCCL INFO 2 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1209:1269 [3] NCCL INFO 3 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1209:1269 [3] NCCL INFO 4 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1209:1269 [3] NCCL INFO 5 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1209:1269 [3] NCCL INFO 6 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1209:1269 [3] NCCL INFO 7 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1209:1269 [3] NCCL INFO 8 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1209:1269 [3] NCCL INFO 9 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1209:1269 [3] NCCL INFO 10 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1209:1269 [3] NCCL INFO 11 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1206:1263 [0] NCCL INFO Pattern 4, crossNic 0, nChannels 12, bw 30.000000/30.000000, type NVL/PIX, sameChannels 1 +n116-042-186:1206:1263 [0] NCCL INFO 0 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1206:1263 [0] NCCL INFO 1 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1206:1263 [0] NCCL INFO 2 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1206:1263 [0] NCCL INFO 3 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1206:1263 [0] NCCL INFO 4 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1206:1263 [0] NCCL INFO 5 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1206:1263 [0] NCCL INFO 6 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1206:1263 [0] NCCL INFO 7 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1206:1263 [0] NCCL INFO 8 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1206:1263 [0] NCCL INFO 9 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1206:1263 [0] NCCL INFO 10 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1206:1263 [0] NCCL INFO 11 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1206:1263 [0] NCCL INFO Pattern 1, crossNic 0, nChannels 12, bw 30.000000/30.000000, type NVL/PIX, sameChannels 1 +n116-042-186:1206:1263 [0] NCCL INFO 0 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1206:1263 [0] NCCL INFO 1 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1206:1263 [0] NCCL INFO 2 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1206:1263 [0] NCCL INFO 3 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1206:1263 [0] NCCL INFO 4 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1206:1263 [0] NCCL INFO 5 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1206:1263 [0] NCCL INFO 6 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1206:1263 [0] NCCL INFO 7 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1206:1263 [0] NCCL INFO 8 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1206:1263 [0] NCCL INFO 9 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1206:1263 [0] NCCL INFO 10 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1206:1263 [0] NCCL INFO 11 : GPU/0 GPU/1 GPU/2 GPU/3 GPU/4 GPU/5 GPU/6 GPU/7 +n116-042-186:1208:1267 [2] NCCL INFO comm 0xb4d5d00 rank 2 nRanks 8 nNodes 1 localRanks 8 localRank 2 MNNVL 0 +n116-042-186:1206:1263 [0] NCCL INFO comm 0xbec8fd0 rank 0 nRanks 8 nNodes 1 localRanks 8 localRank 0 MNNVL 0 +n116-042-186:1207:1268 [1] NCCL INFO comm 0xb5957c0 rank 1 nRanks 8 nNodes 1 localRanks 8 localRank 1 MNNVL 0 +n116-042-186:1213:1264 [7] NCCL INFO comm 0xbf6b490 rank 7 nRanks 8 nNodes 1 localRanks 8 localRank 7 MNNVL 0 +n116-042-186:1212:1266 [6] NCCL INFO comm 0xb67df70 rank 6 nRanks 8 nNodes 1 localRanks 8 localRank 6 MNNVL 0 +n116-042-186:1208:1267 [2] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-042-186:1206:1263 [0] NCCL INFO Tree 0 : -1 -> 0 -> 1/-1/-1 +n116-042-186:1207:1268 [1] NCCL INFO Tree 0 : 0 -> 1 -> 2/-1/-1 +n116-042-186:1207:1268 [1] NCCL INFO Tree 12 : 0 -> 1 -> 2/-1/-1 +n116-042-186:1211:1270 [5] NCCL INFO comm 0xbc38ba0 rank 5 nRanks 8 nNodes 1 localRanks 8 localRank 5 MNNVL 0 +n116-042-186:1213:1264 [7] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-042-186:1206:1263 [0] NCCL INFO Tree 12 : -1 -> 0 -> 1/-1/-1 +n116-042-186:1210:1265 [4] NCCL INFO comm 0xb3b2a80 rank 4 nRanks 8 nNodes 1 localRanks 8 localRank 4 MNNVL 0 +n116-042-186:1207:1268 [1] NCCL INFO Tree 1 : 0 -> 1 -> 2/-1/-1 +n116-042-186:1208:1267 [2] NCCL INFO Ring 00 : 1 -> 2 -> 3 +n116-042-186:1206:1263 [0] NCCL INFO Tree 1 : -1 -> 0 -> 1/-1/-1 +n116-042-186:1211:1270 [5] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-042-186:1207:1268 [1] NCCL INFO Tree 13 : 0 -> 1 -> 2/-1/-1 +n116-042-186:1212:1266 [6] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-042-186:1206:1263 [0] NCCL INFO Tree 13 : -1 -> 0 -> 1/-1/-1 +n116-042-186:1209:1269 [3] NCCL INFO comm 0xd05c320 rank 3 nRanks 8 nNodes 1 localRanks 8 localRank 3 MNNVL 0 +n116-042-186:1208:1267 [2] NCCL INFO Ring 01 : 1 -> 2 -> 3 +n116-042-186:1213:1264 [7] NCCL INFO Ring 00 : 6 -> 7 -> 0 +n116-042-186:1210:1265 [4] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-042-186:1207:1268 [1] NCCL INFO Tree 2 : 0 -> 1 -> 2/-1/-1 +n116-042-186:1206:1263 [0] NCCL INFO Tree 2 : -1 -> 0 -> 1/-1/-1 +n116-042-186:1212:1266 [6] NCCL INFO Ring 00 : 5 -> 6 -> 7 +n116-042-186:1208:1267 [2] NCCL INFO Ring 02 : 1 -> 2 -> 3 +n116-042-186:1210:1265 [4] NCCL INFO Ring 00 : 3 -> 4 -> 5 +n116-042-186:1213:1264 [7] NCCL INFO Ring 01 : 6 -> 7 -> 0 +n116-042-186:1211:1270 [5] NCCL INFO Ring 00 : 4 -> 5 -> 6 +n116-042-186:1207:1268 [1] NCCL INFO Tree 14 : 0 -> 1 -> 2/-1/-1 +n116-042-186:1206:1263 [0] NCCL INFO Tree 14 : -1 -> 0 -> 1/-1/-1 +n116-042-186:1212:1266 [6] NCCL INFO Ring 01 : 5 -> 6 -> 7 +n116-042-186:1209:1269 [3] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-042-186:1208:1267 [2] NCCL INFO Ring 03 : 1 -> 2 -> 3 +n116-042-186:1210:1265 [4] NCCL INFO Ring 01 : 3 -> 4 -> 5 +n116-042-186:1213:1264 [7] NCCL INFO Ring 02 : 6 -> 7 -> 0 +n116-042-186:1211:1270 [5] NCCL INFO Ring 01 : 4 -> 5 -> 6 +n116-042-186:1210:1265 [4] NCCL INFO Ring 02 : 3 -> 4 -> 5 +n116-042-186:1213:1264 [7] NCCL INFO Ring 03 : 6 -> 7 -> 0 +n116-042-186:1207:1268 [1] NCCL INFO Tree 3 : 0 -> 1 -> 2/-1/-1 +n116-042-186:1206:1263 [0] NCCL INFO Tree 3 : -1 -> 0 -> 1/-1/-1 +n116-042-186:1213:1264 [7] NCCL INFO Ring 04 : 6 -> 7 -> 0 +n116-042-186:1212:1266 [6] NCCL INFO Ring 02 : 5 -> 6 -> 7 +n116-042-186:1208:1267 [2] NCCL INFO Ring 04 : 1 -> 2 -> 3 +n116-042-186:1211:1270 [5] NCCL INFO Ring 02 : 4 -> 5 -> 6 +n116-042-186:1210:1265 [4] NCCL INFO Ring 03 : 3 -> 4 -> 5 +n116-042-186:1207:1268 [1] NCCL INFO Tree 15 : 0 -> 1 -> 2/-1/-1 +n116-042-186:1206:1263 [0] NCCL INFO Tree 15 : -1 -> 0 -> 1/-1/-1 +n116-042-186:1213:1264 [7] NCCL INFO Ring 05 : 6 -> 7 -> 0 +n116-042-186:1212:1266 [6] NCCL INFO Ring 03 : 5 -> 6 -> 7 +n116-042-186:1208:1267 [2] NCCL INFO Ring 05 : 1 -> 2 -> 3 +n116-042-186:1209:1269 [3] NCCL INFO Ring 00 : 2 -> 3 -> 4 +n116-042-186:1211:1270 [5] NCCL INFO Ring 03 : 4 -> 5 -> 6 +n116-042-186:1210:1265 [4] NCCL INFO Ring 04 : 3 -> 4 -> 5 +n116-042-186:1207:1268 [1] NCCL INFO Tree 4 : 0 -> 1 -> 2/-1/-1 +n116-042-186:1206:1263 [0] NCCL INFO Tree 4 : -1 -> 0 -> 1/-1/-1 +n116-042-186:1211:1270 [5] NCCL INFO Ring 04 : 4 -> 5 -> 6 +n116-042-186:1210:1265 [4] NCCL INFO Ring 05 : 3 -> 4 -> 5 +n116-042-186:1213:1264 [7] NCCL INFO Ring 06 : 6 -> 7 -> 0 +n116-042-186:1212:1266 [6] NCCL INFO Ring 04 : 5 -> 6 -> 7 +n116-042-186:1210:1265 [4] NCCL INFO Ring 06 : 3 -> 4 -> 5 +n116-042-186:1208:1267 [2] NCCL INFO Ring 06 : 1 -> 2 -> 3 +n116-042-186:1209:1269 [3] NCCL INFO Ring 01 : 2 -> 3 -> 4 +n116-042-186:1207:1268 [1] NCCL INFO Tree 16 : 0 -> 1 -> 2/-1/-1 +n116-042-186:1209:1269 [3] NCCL INFO Ring 02 : 2 -> 3 -> 4 +n116-042-186:1206:1263 [0] NCCL INFO Tree 16 : -1 -> 0 -> 1/-1/-1 +n116-042-186:1211:1270 [5] NCCL INFO Ring 05 : 4 -> 5 -> 6 +n116-042-186:1213:1264 [7] NCCL INFO Ring 07 : 6 -> 7 -> 0 +n116-042-186:1212:1266 [6] NCCL INFO Ring 05 : 5 -> 6 -> 7 +n116-042-186:1211:1270 [5] NCCL INFO Ring 06 : 4 -> 5 -> 6 +n116-042-186:1213:1264 [7] NCCL INFO Ring 08 : 6 -> 7 -> 0 +n116-042-186:1210:1265 [4] NCCL INFO Ring 07 : 3 -> 4 -> 5 +n116-042-186:1208:1267 [2] NCCL INFO Ring 07 : 1 -> 2 -> 3 +n116-042-186:1207:1268 [1] NCCL INFO Tree 5 : 0 -> 1 -> 2/-1/-1 +n116-042-186:1206:1263 [0] NCCL INFO Tree 5 : -1 -> 0 -> 1/-1/-1 +n116-042-186:1209:1269 [3] NCCL INFO Ring 03 : 2 -> 3 -> 4 +n116-042-186:1207:1268 [1] NCCL INFO Tree 17 : 0 -> 1 -> 2/-1/-1 +n116-042-186:1212:1266 [6] NCCL INFO Ring 06 : 5 -> 6 -> 7 +n116-042-186:1211:1270 [5] NCCL INFO Ring 07 : 4 -> 5 -> 6 +n116-042-186:1213:1264 [7] NCCL INFO Ring 09 : 6 -> 7 -> 0 +n116-042-186:1210:1265 [4] NCCL INFO Ring 08 : 3 -> 4 -> 5 +n116-042-186:1208:1267 [2] NCCL INFO Ring 08 : 1 -> 2 -> 3 +n116-042-186:1206:1263 [0] NCCL INFO Tree 17 : -1 -> 0 -> 1/-1/-1 +n116-042-186:1207:1268 [1] NCCL INFO Tree 6 : 0 -> 1 -> 2/-1/-1 +n116-042-186:1209:1269 [3] NCCL INFO Ring 04 : 2 -> 3 -> 4 +n116-042-186:1206:1263 [0] NCCL INFO Tree 6 : -1 -> 0 -> 1/-1/-1 +n116-042-186:1212:1266 [6] NCCL INFO Ring 07 : 5 -> 6 -> 7 +n116-042-186:1213:1264 [7] NCCL INFO Ring 10 : 6 -> 7 -> 0 +n116-042-186:1211:1270 [5] NCCL INFO Ring 08 : 4 -> 5 -> 6 +n116-042-186:1210:1265 [4] NCCL INFO Ring 09 : 3 -> 4 -> 5 +n116-042-186:1213:1264 [7] NCCL INFO Ring 11 : 6 -> 7 -> 0 +n116-042-186:1208:1267 [2] NCCL INFO Ring 09 : 1 -> 2 -> 3 +n116-042-186:1207:1268 [1] NCCL INFO Tree 18 : 0 -> 1 -> 2/-1/-1 +n116-042-186:1209:1269 [3] NCCL INFO Ring 05 : 2 -> 3 -> 4 +n116-042-186:1206:1263 [0] NCCL INFO Tree 18 : -1 -> 0 -> 1/-1/-1 +n116-042-186:1212:1266 [6] NCCL INFO Ring 08 : 5 -> 6 -> 7 +n116-042-186:1211:1270 [5] NCCL INFO Ring 09 : 4 -> 5 -> 6 +n116-042-186:1210:1265 [4] NCCL INFO Ring 10 : 3 -> 4 -> 5 +n116-042-186:1213:1264 [7] NCCL INFO Ring 12 : 6 -> 7 -> 0 +n116-042-186:1212:1266 [6] NCCL INFO Ring 09 : 5 -> 6 -> 7 +n116-042-186:1213:1264 [7] NCCL INFO Ring 13 : 6 -> 7 -> 0 +n116-042-186:1208:1267 [2] NCCL INFO Ring 10 : 1 -> 2 -> 3 +n116-042-186:1207:1268 [1] NCCL INFO Tree 7 : 0 -> 1 -> 2/-1/-1 +n116-042-186:1212:1266 [6] NCCL INFO Ring 10 : 5 -> 6 -> 7 +n116-042-186:1209:1269 [3] NCCL INFO Ring 06 : 2 -> 3 -> 4 +n116-042-186:1206:1263 [0] NCCL INFO Tree 7 : -1 -> 0 -> 1/-1/-1 +n116-042-186:1212:1266 [6] NCCL INFO Ring 11 : 5 -> 6 -> 7 +n116-042-186:1211:1270 [5] NCCL INFO Ring 10 : 4 -> 5 -> 6 +n116-042-186:1210:1265 [4] NCCL INFO Ring 11 : 3 -> 4 -> 5 +n116-042-186:1210:1265 [4] NCCL INFO Ring 12 : 3 -> 4 -> 5 +n116-042-186:1213:1264 [7] NCCL INFO Ring 14 : 6 -> 7 -> 0 +n116-042-186:1208:1267 [2] NCCL INFO Ring 11 : 1 -> 2 -> 3 +n116-042-186:1207:1268 [1] NCCL INFO Tree 19 : 0 -> 1 -> 2/-1/-1 +n116-042-186:1209:1269 [3] NCCL INFO Ring 07 : 2 -> 3 -> 4 +n116-042-186:1206:1263 [0] NCCL INFO Tree 19 : -1 -> 0 -> 1/-1/-1 +n116-042-186:1207:1268 [1] NCCL INFO Tree 8 : 0 -> 1 -> 2/-1/-1 +n116-042-186:1212:1266 [6] NCCL INFO Ring 12 : 5 -> 6 -> 7 +n116-042-186:1209:1269 [3] NCCL INFO Ring 08 : 2 -> 3 -> 4 +n116-042-186:1211:1270 [5] NCCL INFO Ring 11 : 4 -> 5 -> 6 +n116-042-186:1211:1270 [5] NCCL INFO Ring 12 : 4 -> 5 -> 6 +n116-042-186:1210:1265 [4] NCCL INFO Ring 13 : 3 -> 4 -> 5 +n116-042-186:1213:1264 [7] NCCL INFO Ring 15 : 6 -> 7 -> 0 +n116-042-186:1211:1270 [5] NCCL INFO Ring 13 : 4 -> 5 -> 6 +n116-042-186:1211:1270 [5] NCCL INFO Ring 14 : 4 -> 5 -> 6 +n116-042-186:1211:1270 [5] NCCL INFO Ring 15 : 4 -> 5 -> 6 +n116-042-186:1211:1270 [5] NCCL INFO Ring 16 : 4 -> 5 -> 6 +n116-042-186:1209:1269 [3] NCCL INFO Ring 09 : 2 -> 3 -> 4 +n116-042-186:1208:1267 [2] NCCL INFO Ring 12 : 1 -> 2 -> 3 +n116-042-186:1207:1268 [1] NCCL INFO Tree 20 : 0 -> 1 -> 2/-1/-1 +n116-042-186:1206:1263 [0] NCCL INFO Tree 8 : -1 -> 0 -> 1/-1/-1 +n116-042-186:1207:1268 [1] NCCL INFO Tree 9 : 0 -> 1 -> 2/-1/-1 +n116-042-186:1212:1266 [6] NCCL INFO Ring 13 : 5 -> 6 -> 7 +n116-042-186:1210:1265 [4] NCCL INFO Ring 14 : 3 -> 4 -> 5 +n116-042-186:1213:1264 [7] NCCL INFO Ring 16 : 6 -> 7 -> 0 +n116-042-186:1213:1264 [7] NCCL INFO Ring 17 : 6 -> 7 -> 0 +n116-042-186:1211:1270 [5] NCCL INFO Ring 17 : 4 -> 5 -> 6 +n116-042-186:1208:1267 [2] NCCL INFO Ring 13 : 1 -> 2 -> 3 +n116-042-186:1213:1264 [7] NCCL INFO Ring 18 : 6 -> 7 -> 0 +n116-042-186:1206:1263 [0] NCCL INFO Tree 20 : -1 -> 0 -> 1/-1/-1 +n116-042-186:1207:1268 [1] NCCL INFO Tree 21 : 0 -> 1 -> 2/-1/-1 +n116-042-186:1209:1269 [3] NCCL INFO Ring 10 : 2 -> 3 -> 4 +n116-042-186:1206:1263 [0] NCCL INFO Tree 9 : -1 -> 0 -> 1/-1/-1 +n116-042-186:1212:1266 [6] NCCL INFO Ring 14 : 5 -> 6 -> 7 +n116-042-186:1210:1265 [4] NCCL INFO Ring 15 : 3 -> 4 -> 5 +n116-042-186:1211:1270 [5] NCCL INFO Ring 18 : 4 -> 5 -> 6 +n116-042-186:1208:1267 [2] NCCL INFO Ring 14 : 1 -> 2 -> 3 +n116-042-186:1213:1264 [7] NCCL INFO Ring 19 : 6 -> 7 -> 0 +n116-042-186:1207:1268 [1] NCCL INFO Tree 10 : 0 -> 1 -> 2/-1/-1 +n116-042-186:1212:1266 [6] NCCL INFO Ring 15 : 5 -> 6 -> 7 +n116-042-186:1206:1263 [0] NCCL INFO Tree 21 : -1 -> 0 -> 1/-1/-1 +n116-042-186:1210:1265 [4] NCCL INFO Ring 16 : 3 -> 4 -> 5 +n116-042-186:1209:1269 [3] NCCL INFO Ring 11 : 2 -> 3 -> 4 +n116-042-186:1211:1270 [5] NCCL INFO Ring 19 : 4 -> 5 -> 6 +n116-042-186:1211:1270 [5] NCCL INFO Ring 20 : 4 -> 5 -> 6 +n116-042-186:1208:1267 [2] NCCL INFO Ring 15 : 1 -> 2 -> 3 +n116-042-186:1213:1264 [7] NCCL INFO Ring 20 : 6 -> 7 -> 0 +n116-042-186:1207:1268 [1] NCCL INFO Tree 22 : 0 -> 1 -> 2/-1/-1 +n116-042-186:1212:1266 [6] NCCL INFO Ring 16 : 5 -> 6 -> 7 +n116-042-186:1207:1268 [1] NCCL INFO Tree 11 : 0 -> 1 -> 2/-1/-1 +n116-042-186:1206:1263 [0] NCCL INFO Tree 10 : -1 -> 0 -> 1/-1/-1 +n116-042-186:1212:1266 [6] NCCL INFO Ring 17 : 5 -> 6 -> 7 +n116-042-186:1210:1265 [4] NCCL INFO Ring 17 : 3 -> 4 -> 5 +n116-042-186:1209:1269 [3] NCCL INFO Ring 12 : 2 -> 3 -> 4 +n116-042-186:1212:1266 [6] NCCL INFO Ring 18 : 5 -> 6 -> 7 +n116-042-186:1211:1270 [5] NCCL INFO Ring 21 : 4 -> 5 -> 6 +n116-042-186:1208:1267 [2] NCCL INFO Ring 16 : 1 -> 2 -> 3 +n116-042-186:1209:1269 [3] NCCL INFO Ring 13 : 2 -> 3 -> 4 +n116-042-186:1213:1264 [7] NCCL INFO Ring 21 : 6 -> 7 -> 0 +n116-042-186:1207:1268 [1] NCCL INFO Tree 23 : 0 -> 1 -> 2/-1/-1 +n116-042-186:1206:1263 [0] NCCL INFO Tree 22 : -1 -> 0 -> 1/-1/-1 +n116-042-186:1213:1264 [7] NCCL INFO Ring 22 : 6 -> 7 -> 0 +n116-042-186:1210:1265 [4] NCCL INFO Ring 18 : 3 -> 4 -> 5 +n116-042-186:1206:1263 [0] NCCL INFO Tree 11 : -1 -> 0 -> 1/-1/-1 +n116-042-186:1212:1266 [6] NCCL INFO Ring 19 : 5 -> 6 -> 7 +n116-042-186:1211:1270 [5] NCCL INFO Ring 22 : 4 -> 5 -> 6 +n116-042-186:1208:1267 [2] NCCL INFO Ring 17 : 1 -> 2 -> 3 +n116-042-186:1212:1266 [6] NCCL INFO Ring 20 : 5 -> 6 -> 7 +n116-042-186:1209:1269 [3] NCCL INFO Ring 14 : 2 -> 3 -> 4 +n116-042-186:1213:1264 [7] NCCL INFO Ring 23 : 6 -> 7 -> 0 +n116-042-186:1210:1265 [4] NCCL INFO Ring 19 : 3 -> 4 -> 5 +n116-042-186:1206:1263 [0] NCCL INFO Tree 23 : -1 -> 0 -> 1/-1/-1 +n116-042-186:1213:1264 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6 +n116-042-186:1207:1268 [1] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-042-186:1211:1270 [5] NCCL INFO Ring 23 : 4 -> 5 -> 6 +n116-042-186:1208:1267 [2] NCCL INFO Ring 18 : 1 -> 2 -> 3 +n116-042-186:1211:1270 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4 +n116-042-186:1212:1266 [6] NCCL INFO Ring 21 : 5 -> 6 -> 7 +n116-042-186:1209:1269 [3] NCCL INFO Ring 15 : 2 -> 3 -> 4 +n116-042-186:1206:1263 [0] NCCL INFO NCCL_MIN_NCHANNELS set by environment to 4. +n116-042-186:1210:1265 [4] NCCL INFO Ring 20 : 3 -> 4 -> 5 +n116-042-186:1213:1264 [7] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-042-186:1208:1267 [2] NCCL INFO Ring 19 : 1 -> 2 -> 3 +n116-042-186:1212:1266 [6] NCCL INFO Ring 22 : 5 -> 6 -> 7 +n116-042-186:1213:1264 [7] NCCL INFO NCCL_P2P_NVL_CHUNKSIZE set by environment to 1048576. +n116-042-186:1211:1270 [5] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-042-186:1209:1269 [3] NCCL INFO Ring 16 : 2 -> 3 -> 4 +n116-042-186:1207:1268 [1] NCCL INFO Ring 00 : 0 -> 1 -> 2 +n116-042-186:1206:1263 [0] NCCL INFO Channel 00/24 : 0 1 2 3 4 5 6 7 +n116-042-186:1210:1265 [4] NCCL INFO Ring 21 : 3 -> 4 -> 5 +n116-042-186:1208:1267 [2] NCCL INFO Ring 20 : 1 -> 2 -> 3 +n116-042-186:1213:1264 [7] NCCL INFO P2P Chunksize set to 524288 +n116-042-186:1212:1266 [6] NCCL INFO Ring 23 : 5 -> 6 -> 7 +n116-042-186:1211:1270 [5] NCCL INFO NCCL_P2P_NVL_CHUNKSIZE set by environment to 1048576. +n116-042-186:1209:1269 [3] NCCL INFO Ring 17 : 2 -> 3 -> 4 +n116-042-186:1207:1268 [1] NCCL INFO Ring 01 : 0 -> 1 -> 2 +n116-042-186:1210:1265 [4] NCCL INFO Ring 22 : 3 -> 4 -> 5 +n116-042-186:1206:1263 [0] NCCL INFO Channel 01/24 : 0 1 2 3 4 5 6 7 +n116-042-186:1207:1268 [1] NCCL INFO Ring 02 : 0 -> 1 -> 2 +n116-042-186:1208:1267 [2] NCCL INFO Ring 21 : 1 -> 2 -> 3 +n116-042-186:1212:1266 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5 +n116-042-186:1211:1270 [5] NCCL INFO P2P Chunksize set to 524288 +n116-042-186:1209:1269 [3] NCCL INFO Ring 18 : 2 -> 3 -> 4 +n116-042-186:1210:1265 [4] NCCL INFO Ring 23 : 3 -> 4 -> 5 +n116-042-186:1207:1268 [1] NCCL INFO Ring 03 : 0 -> 1 -> 2 +n116-042-186:1206:1263 [0] NCCL INFO Channel 02/24 : 0 1 2 3 4 5 6 7 +n116-042-186:1208:1267 [2] NCCL INFO Ring 22 : 1 -> 2 -> 3 +n116-042-186:1212:1266 [6] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-042-186:1212:1266 [6] NCCL INFO NCCL_P2P_NVL_CHUNKSIZE set by environment to 1048576. +n116-042-186:1210:1265 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3 +n116-042-186:1207:1268 [1] NCCL INFO Ring 04 : 0 -> 1 -> 2 +n116-042-186:1210:1265 [4] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-042-186:1209:1269 [3] NCCL INFO Ring 19 : 2 -> 3 -> 4 +n116-042-186:1206:1263 [0] NCCL INFO Channel 03/24 : 0 1 2 3 4 5 6 7 +n116-042-186:1208:1267 [2] NCCL INFO Ring 23 : 1 -> 2 -> 3 +n116-042-186:1212:1266 [6] NCCL INFO P2P Chunksize set to 524288 +n116-042-186:1207:1268 [1] NCCL INFO Ring 05 : 0 -> 1 -> 2 +n116-042-186:1210:1265 [4] NCCL INFO NCCL_P2P_NVL_CHUNKSIZE set by environment to 1048576. +n116-042-186:1208:1267 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1 +n116-042-186:1209:1269 [3] NCCL INFO Ring 20 : 2 -> 3 -> 4 +n116-042-186:1210:1265 [4] NCCL INFO P2P Chunksize set to 524288 +n116-042-186:1206:1263 [0] NCCL INFO Channel 04/24 : 0 1 2 3 4 5 6 7 +n116-042-186:1207:1268 [1] NCCL INFO Ring 06 : 0 -> 1 -> 2 +n116-042-186:1208:1267 [2] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-042-186:1209:1269 [3] NCCL INFO Ring 21 : 2 -> 3 -> 4 +n116-042-186:1206:1263 [0] NCCL INFO Channel 05/24 : 0 1 2 3 4 5 6 7 +n116-042-186:1207:1268 [1] NCCL INFO Ring 07 : 0 -> 1 -> 2 +n116-042-186:1208:1267 [2] NCCL INFO NCCL_P2P_NVL_CHUNKSIZE set by environment to 1048576. +n116-042-186:1208:1267 [2] NCCL INFO P2P Chunksize set to 524288 +n116-042-186:1209:1269 [3] NCCL INFO Ring 22 : 2 -> 3 -> 4 +n116-042-186:1207:1268 [1] NCCL INFO Ring 08 : 0 -> 1 -> 2 +n116-042-186:1206:1263 [0] NCCL INFO Channel 06/24 : 0 1 2 3 4 5 6 7 +n116-042-186:1209:1269 [3] NCCL INFO Ring 23 : 2 -> 3 -> 4 +n116-042-186:1207:1268 [1] NCCL INFO Ring 09 : 0 -> 1 -> 2 +n116-042-186:1206:1263 [0] NCCL INFO Channel 07/24 : 0 1 2 3 4 5 6 7 +n116-042-186:1209:1269 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2 +n116-042-186:1207:1268 [1] NCCL INFO Ring 10 : 0 -> 1 -> 2 +n116-042-186:1207:1268 [1] NCCL INFO Ring 11 : 0 -> 1 -> 2 +n116-042-186:1206:1263 [0] NCCL INFO Channel 08/24 : 0 1 2 3 4 5 6 7 +n116-042-186:1207:1268 [1] NCCL INFO Ring 12 : 0 -> 1 -> 2 +n116-042-186:1209:1269 [3] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-042-186:1207:1268 [1] NCCL INFO Ring 13 : 0 -> 1 -> 2 +n116-042-186:1209:1269 [3] NCCL INFO NCCL_P2P_NVL_CHUNKSIZE set by environment to 1048576. +n116-042-186:1207:1268 [1] NCCL INFO Ring 14 : 0 -> 1 -> 2 +n116-042-186:1206:1263 [0] NCCL INFO Channel 09/24 : 0 1 2 3 4 5 6 7 +n116-042-186:1209:1269 [3] NCCL INFO P2P Chunksize set to 524288 +n116-042-186:1207:1268 [1] NCCL INFO Ring 15 : 0 -> 1 -> 2 +n116-042-186:1206:1263 [0] NCCL INFO Channel 10/24 : 0 1 2 3 4 5 6 7 +n116-042-186:1207:1268 [1] NCCL INFO Ring 16 : 0 -> 1 -> 2 +n116-042-186:1207:1268 [1] NCCL INFO Ring 17 : 0 -> 1 -> 2 +n116-042-186:1206:1263 [0] NCCL INFO Channel 11/24 : 0 1 2 3 4 5 6 7 +n116-042-186:1207:1268 [1] NCCL INFO Ring 18 : 0 -> 1 -> 2 +n116-042-186:1207:1268 [1] NCCL INFO Ring 19 : 0 -> 1 -> 2 +n116-042-186:1207:1268 [1] NCCL INFO Ring 20 : 0 -> 1 -> 2 +n116-042-186:1206:1263 [0] NCCL INFO Channel 12/24 : 0 1 2 3 4 5 6 7 +n116-042-186:1207:1268 [1] NCCL INFO Ring 21 : 0 -> 1 -> 2 +n116-042-186:1207:1268 [1] NCCL INFO Ring 22 : 0 -> 1 -> 2 +n116-042-186:1206:1263 [0] NCCL INFO Channel 13/24 : 0 1 2 3 4 5 6 7 +n116-042-186:1207:1268 [1] NCCL INFO Ring 23 : 0 -> 1 -> 2 +n116-042-186:1207:1268 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0 +n116-042-186:1206:1263 [0] NCCL INFO Channel 14/24 : 0 1 2 3 4 5 6 7 +n116-042-186:1207:1268 [1] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-042-186:1207:1268 [1] NCCL INFO NCCL_P2P_NVL_CHUNKSIZE set by environment to 1048576. +n116-042-186:1206:1263 [0] NCCL INFO Channel 15/24 : 0 1 2 3 4 5 6 7 +n116-042-186:1207:1268 [1] NCCL INFO P2P Chunksize set to 524288 +n116-042-186:1206:1263 [0] NCCL INFO Channel 16/24 : 0 1 2 3 4 5 6 7 +n116-042-186:1206:1263 [0] NCCL INFO Channel 17/24 : 0 1 2 3 4 5 6 7 +n116-042-186:1206:1263 [0] NCCL INFO Channel 18/24 : 0 1 2 3 4 5 6 7 +n116-042-186:1206:1263 [0] NCCL INFO Channel 19/24 : 0 1 2 3 4 5 6 7 +n116-042-186:1206:1263 [0] NCCL INFO Channel 20/24 : 0 1 2 3 4 5 6 7 +n116-042-186:1206:1263 [0] NCCL INFO Channel 21/24 : 0 1 2 3 4 5 6 7 +n116-042-186:1206:1263 [0] NCCL INFO Channel 22/24 : 0 1 2 3 4 5 6 7 +n116-042-186:1206:1263 [0] NCCL INFO Channel 23/24 : 0 1 2 3 4 5 6 7 +n116-042-186:1206:1263 [0] NCCL INFO Ring 00 : 7 -> 0 -> 1 +n116-042-186:1206:1263 [0] NCCL INFO Ring 01 : 7 -> 0 -> 1 +n116-042-186:1206:1263 [0] NCCL INFO Ring 02 : 7 -> 0 -> 1 +n116-042-186:1206:1263 [0] NCCL INFO Ring 03 : 7 -> 0 -> 1 +n116-042-186:1206:1263 [0] NCCL INFO Ring 04 : 7 -> 0 -> 1 +n116-042-186:1206:1263 [0] NCCL INFO Ring 05 : 7 -> 0 -> 1 +n116-042-186:1206:1263 [0] NCCL INFO Ring 06 : 7 -> 0 -> 1 +n116-042-186:1206:1263 [0] NCCL INFO Ring 07 : 7 -> 0 -> 1 +n116-042-186:1206:1263 [0] NCCL INFO Ring 08 : 7 -> 0 -> 1 +n116-042-186:1206:1263 [0] NCCL INFO Ring 09 : 7 -> 0 -> 1 +n116-042-186:1206:1263 [0] NCCL INFO Ring 10 : 7 -> 0 -> 1 +n116-042-186:1206:1263 [0] NCCL INFO Ring 11 : 7 -> 0 -> 1 +n116-042-186:1206:1263 [0] NCCL INFO Ring 12 : 7 -> 0 -> 1 +n116-042-186:1206:1263 [0] NCCL INFO Ring 13 : 7 -> 0 -> 1 +n116-042-186:1206:1263 [0] NCCL INFO Ring 14 : 7 -> 0 -> 1 +n116-042-186:1206:1263 [0] NCCL INFO Ring 15 : 7 -> 0 -> 1 +n116-042-186:1206:1263 [0] NCCL INFO Ring 16 : 7 -> 0 -> 1 +n116-042-186:1206:1263 [0] NCCL INFO Ring 17 : 7 -> 0 -> 1 +n116-042-186:1206:1263 [0] NCCL INFO Ring 18 : 7 -> 0 -> 1 +n116-042-186:1206:1263 [0] NCCL INFO Ring 19 : 7 -> 0 -> 1 +n116-042-186:1206:1263 [0] NCCL INFO Ring 20 : 7 -> 0 -> 1 +n116-042-186:1206:1263 [0] NCCL INFO Ring 21 : 7 -> 0 -> 1 +n116-042-186:1206:1263 [0] NCCL INFO Ring 22 : 7 -> 0 -> 1 +n116-042-186:1206:1263 [0] NCCL INFO Ring 23 : 7 -> 0 -> 1 +n116-042-186:1206:1263 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1 +n116-042-186:1206:1263 [0] NCCL INFO NCCL_BUFFSIZE set by environment to 4194304. +n116-042-186:1206:1263 [0] NCCL INFO NCCL_P2P_NVL_CHUNKSIZE set by environment to 1048576. +n116-042-186:1206:1263 [0] NCCL INFO P2P Chunksize set to 524288 +n116-042-186:1211:1303 [5] NCCL INFO [Proxy Service] Device 5 CPU core 197 +n116-042-186:1211:1304 [5] NCCL INFO [Proxy Service UDS] Device 5 CPU core 204 +n116-042-186:1206:1263 [0] NCCL INFO Check P2P Type intraNodeP2pSupport 1 directMode 0 +n116-042-186:1206:1305 [0] NCCL INFO [Proxy Service] Device 0 CPU core 154 +n116-042-186:1206:1306 [0] NCCL INFO [Proxy Service UDS] Device 0 CPU core 124 +n116-042-186:1212:1307 [6] NCCL INFO [Proxy Service] Device 6 CPU core 169 +n116-042-186:1212:1308 [6] NCCL INFO [Proxy Service UDS] Device 6 CPU core 198 +n116-042-186:1207:1310 [1] NCCL INFO [Proxy Service UDS] Device 1 CPU core 139 +n116-042-186:1207:1309 [1] NCCL INFO [Proxy Service] Device 1 CPU core 134 +n116-042-186:1208:1311 [2] NCCL INFO [Proxy Service] Device 2 CPU core 141 +n116-042-186:1208:1312 [2] NCCL INFO [Proxy Service UDS] Device 2 CPU core 140 +n116-042-186:1213:1314 [7] NCCL INFO [Proxy Service UDS] Device 7 CPU core 175 +n116-042-186:1213:1313 [7] NCCL INFO [Proxy Service] Device 7 CPU core 164 +n116-042-186:1210:1315 [4] NCCL INFO [Proxy Service] Device 4 CPU core 86 +n116-042-186:1210:1316 [4] NCCL INFO [Proxy Service UDS] Device 4 CPU core 203 +n116-042-186:1209:1317 [3] NCCL INFO [Proxy Service] Device 3 CPU core 6 +n116-042-186:1209:1318 [3] NCCL INFO [Proxy Service UDS] Device 3 CPU core 120 +n116-042-186:1211:1270 [5] NCCL INFO NCCL_PROTO set by environment to Simple +n116-042-186:1211:1270 [5] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-042-186:1211:1270 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512 +n116-042-186:1211:1270 [5] NCCL INFO 24 coll channels, 24 collnet channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer +n116-042-186:1206:1263 [0] NCCL INFO NCCL_PROTO set by environment to Simple +n116-042-186:1206:1263 [0] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-042-186:1206:1263 [0] NCCL INFO Enabled NCCL Func/Proto/Algo Matrix: + Function | LL LL128 Simple | Tree Ring CollNetDirect CollNetChain NVLS NVLSTree PAT + Broadcast | 0 0 1 | 1 1 0 0 0 0 0 + Reduce | 0 0 1 | 1 1 0 0 0 0 0 + AllGather | 0 0 1 | 1 1 0 0 0 0 0 +ReduceScatter | 0 0 1 | 1 1 0 0 0 0 0 + AllReduce | 0 0 1 | 1 1 0 0 0 0 0 + +n116-042-186:1206:1263 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512 +n116-042-186:1206:1263 [0] NCCL INFO 24 coll channels, 24 collnet channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer +n116-042-186:1207:1268 [1] NCCL INFO NCCL_PROTO set by environment to Simple +n116-042-186:1207:1268 [1] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-042-186:1207:1268 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512 +n116-042-186:1207:1268 [1] NCCL INFO 24 coll channels, 24 collnet channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer +n116-042-186:1206:1263 [0] NCCL INFO CC Off, workFifoBytes 1048576 +n116-042-186:1209:1269 [3] NCCL INFO NCCL_PROTO set by environment to Simple +n116-042-186:1209:1269 [3] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-042-186:1209:1269 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512 +n116-042-186:1209:1269 [3] NCCL INFO 24 coll channels, 24 collnet channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer +n116-042-186:1213:1264 [7] NCCL INFO NCCL_PROTO set by environment to Simple +n116-042-186:1213:1264 [7] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-042-186:1213:1264 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512 +n116-042-186:1213:1264 [7] NCCL INFO 24 coll channels, 24 collnet channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer +n116-042-186:1208:1267 [2] NCCL INFO NCCL_PROTO set by environment to Simple +n116-042-186:1208:1267 [2] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-042-186:1208:1267 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512 +n116-042-186:1208:1267 [2] NCCL INFO 24 coll channels, 24 collnet channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer +n116-042-186:1212:1266 [6] NCCL INFO NCCL_PROTO set by environment to Simple +n116-042-186:1212:1266 [6] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-042-186:1212:1266 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512 +n116-042-186:1212:1266 [6] NCCL INFO 24 coll channels, 24 collnet channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer +n116-042-186:1210:1265 [4] NCCL INFO NCCL_PROTO set by environment to Simple +n116-042-186:1210:1265 [4] NCCL INFO NCCL_ALGO set by environment to Ring,Tree +n116-042-186:1210:1265 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512 +n116-042-186:1210:1265 [4] NCCL INFO 24 coll channels, 24 collnet channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer +n116-042-186:1207:1268 [1] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-042-186:1207:1268 [1] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-042-186:1207:1268 [1] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-042-186:1207:1268 [1] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-042-186:1207:1268 [1] NCCL INFO ncclCommInitRankConfig comm 0xb5957c0 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 5000 commId 0x5e9d7d8d37acfcd1 - Init COMPLETE +n116-042-186:1207:1268 [1] NCCL INFO Init timings - ncclCommInitRankConfig: rank 1 nranks 8 total 5.28 (kernels 0.53, alloc 3.49, bootstrap 0.01, allgathers 0.01, topo 0.70, graphs 0.01, connections 0.06, rest 0.46) +n116-042-186:1208:1267 [2] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-042-186:1208:1267 [2] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-042-186:1208:1267 [2] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-042-186:1208:1267 [2] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-042-186:1208:1267 [2] NCCL INFO ncclCommInitRankConfig comm 0xb4d5d00 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId b000 commId 0x5e9d7d8d37acfcd1 - Init COMPLETE +n116-042-186:1208:1267 [2] NCCL INFO Init timings - ncclCommInitRankConfig: rank 2 nranks 8 total 5.29 (kernels 0.59, alloc 3.46, bootstrap 0.00, allgathers 0.01, topo 0.70, graphs 0.01, connections 0.06, rest 0.46) +n116-042-186:1212:1266 [6] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-042-186:1212:1266 [6] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-042-186:1212:1266 [6] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-042-186:1212:1266 [6] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-042-186:1212:1266 [6] NCCL INFO ncclCommInitRankConfig comm 0xb67df70 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId 8b000 commId 0x5e9d7d8d37acfcd1 - Init COMPLETE +n116-042-186:1212:1266 [6] NCCL INFO Init timings - ncclCommInitRankConfig: rank 6 nranks 8 total 5.30 (kernels 0.54, alloc 3.50, bootstrap 0.01, allgathers 0.01, topo 0.70, graphs 0.01, connections 0.06, rest 0.47) +n116-042-186:1211:1270 [5] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-042-186:1211:1270 [5] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-042-186:1211:1270 [5] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-042-186:1211:1270 [5] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-042-186:1211:1270 [5] NCCL INFO ncclCommInitRankConfig comm 0xbc38ba0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId 85000 commId 0x5e9d7d8d37acfcd1 - Init COMPLETE +n116-042-186:1211:1270 [5] NCCL INFO Init timings - ncclCommInitRankConfig: rank 5 nranks 8 total 5.30 (kernels 0.53, alloc 3.49, bootstrap 0.01, allgathers 0.01, topo 0.70, graphs 0.01, connections 0.06, rest 0.48) +n116-042-186:1210:1265 [4] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-042-186:1210:1265 [4] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-042-186:1210:1265 [4] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-042-186:1210:1265 [4] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-042-186:1210:1265 [4] NCCL INFO ncclCommInitRankConfig comm 0xb3b2a80 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 84000 commId 0x5e9d7d8d37acfcd1 - Init COMPLETE +n116-042-186:1210:1265 [4] NCCL INFO Init timings - ncclCommInitRankConfig: rank 4 nranks 8 total 5.32 (kernels 0.55, alloc 3.40, bootstrap 0.11, allgathers 0.01, topo 0.70, graphs 0.01, connections 0.06, rest 0.49) +n116-042-186:1209:1269 [3] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-042-186:1209:1269 [3] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-042-186:1209:1269 [3] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-042-186:1209:1269 [3] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-042-186:1209:1269 [3] NCCL INFO ncclCommInitRankConfig comm 0xd05c320 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId c000 commId 0x5e9d7d8d37acfcd1 - Init COMPLETE +n116-042-186:1209:1269 [3] NCCL INFO Init timings - ncclCommInitRankConfig: rank 3 nranks 8 total 5.31 (kernels 0.53, alloc 3.51, bootstrap 0.00, allgathers 0.00, topo 0.70, graphs 0.01, connections 0.05, rest 0.50) +n116-042-186:1206:1263 [0] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-042-186:1206:1263 [0] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-042-186:1206:1263 [0] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-042-186:1206:1263 [0] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-042-186:1206:1263 [0] NCCL INFO ncclCommInitRankConfig comm 0xbec8fd0 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 4000 commId 0x5e9d7d8d37acfcd1 - Init COMPLETE +n116-042-186:1206:1263 [0] NCCL INFO Init timings - ncclCommInitRankConfig: rank 0 nranks 8 total 5.39 (kernels 0.61, alloc 3.50, bootstrap 0.01, allgathers 0.00, topo 0.70, graphs 0.01, connections 0.06, rest 0.49) +n116-042-186:1213:1264 [7] NCCL INFO TUNER/Plugin: Could not find: libnccl-tuner-libnccl-tuner.so.so. Using internal tuner plugin. +n116-042-186:1213:1264 [7] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v4 symbol. +n116-042-186:1213:1264 [7] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v3 symbol. +n116-042-186:1213:1264 [7] NCCL INFO TUNER/Plugin: Failed to find ncclTunerPlugin_v2 symbol, using internal tuner instead. +n116-042-186:1213:1264 [7] NCCL INFO ncclCommInitRankConfig comm 0xbf6b490 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId 8c000 commId 0x5e9d7d8d37acfcd1 - Init COMPLETE +n116-042-186:1213:1264 [7] NCCL INFO Init timings - ncclCommInitRankConfig: rank 7 nranks 8 total 5.34 (kernels 0.56, alloc 3.51, bootstrap 0.00, allgathers 0.01, topo 0.71, graphs 0.01, connections 0.06, rest 0.50) +n116-042-186:1212:1321 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/CUMEM +n116-042-186:1207:1320 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/CUMEM +n116-042-186:1212:1321 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/CUMEM +n116-042-186:1207:1320 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/CUMEM +n116-042-186:1207:1320 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/CUMEM +n116-042-186:1207:1320 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/CUMEM +n116-042-186:1207:1320 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/CUMEM +n116-042-186:1207:1320 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/CUMEM +n116-042-186:1211:1322 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/CUMEM +n116-042-186:1207:1320 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/CUMEM +n116-042-186:1211:1322 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/CUMEM +n116-042-186:1207:1320 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/CUMEM +n116-042-186:1211:1322 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/CUMEM +n116-042-186:1207:1320 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/CUMEM +n116-042-186:1211:1322 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/CUMEM +n116-042-186:1207:1320 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/CUMEM +n116-042-186:1211:1322 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/CUMEM +n116-042-186:1207:1320 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/CUMEM +n116-042-186:1212:1321 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/CUMEM +n116-042-186:1211:1322 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/CUMEM +n116-042-186:1207:1320 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/CUMEM +n116-042-186:1212:1321 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/CUMEM +n116-042-186:1211:1322 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/CUMEM +n116-042-186:1207:1320 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/CUMEM +n116-042-186:1212:1321 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/CUMEM +n116-042-186:1211:1322 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/CUMEM +n116-042-186:1207:1320 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/CUMEM +n116-042-186:1212:1321 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/CUMEM +n116-042-186:1211:1322 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/CUMEM +n116-042-186:1207:1320 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/CUMEM +n116-042-186:1208:1319 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/CUMEM +n116-042-186:1212:1321 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/CUMEM +n116-042-186:1211:1322 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/CUMEM +n116-042-186:1207:1320 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/CUMEM +n116-042-186:1208:1319 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/CUMEM +n116-042-186:1212:1321 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/CUMEM +n116-042-186:1210:1323 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/CUMEM +n116-042-186:1211:1322 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/CUMEM +n116-042-186:1207:1320 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/CUMEM +n116-042-186:1208:1319 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/CUMEM +n116-042-186:1212:1321 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/CUMEM +n116-042-186:1210:1323 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/CUMEM +n116-042-186:1211:1322 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/CUMEM +n116-042-186:1207:1320 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/CUMEM +n116-042-186:1208:1319 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/CUMEM +n116-042-186:1212:1321 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/CUMEM +n116-042-186:1210:1323 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/CUMEM +n116-042-186:1211:1322 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/CUMEM +n116-042-186:1210:1323 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/CUMEM +n116-042-186:1210:1323 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/CUMEM +n116-042-186:1207:1320 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/CUMEM +n116-042-186:1212:1321 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/CUMEM +n116-042-186:1210:1323 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/CUMEM +n116-042-186:1207:1320 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/CUMEM +n116-042-186:1212:1321 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/CUMEM +n116-042-186:1210:1323 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/CUMEM +n116-042-186:1207:1320 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/CUMEM +n116-042-186:1212:1321 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/CUMEM +n116-042-186:1210:1323 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/CUMEM +n116-042-186:1207:1320 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/CUMEM +n116-042-186:1212:1321 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/CUMEM +n116-042-186:1210:1323 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/CUMEM +n116-042-186:1207:1320 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/CUMEM +n116-042-186:1212:1321 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/CUMEM +n116-042-186:1210:1323 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/CUMEM +n116-042-186:1207:1320 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/CUMEM +n116-042-186:1212:1321 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/CUMEM +n116-042-186:1210:1323 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/CUMEM +n116-042-186:1212:1321 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/CUMEM +n116-042-186:1210:1323 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/CUMEM +n116-042-186:1212:1321 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/CUMEM +n116-042-186:1208:1319 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/CUMEM +n116-042-186:1210:1323 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/CUMEM +n116-042-186:1212:1321 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/CUMEM +n116-042-186:1208:1319 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/CUMEM +n116-042-186:1210:1323 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/CUMEM +n116-042-186:1212:1321 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/CUMEM +n116-042-186:1208:1319 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/CUMEM +n116-042-186:1210:1323 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/CUMEM +n116-042-186:1212:1321 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/CUMEM +n116-042-186:1208:1319 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/CUMEM +n116-042-186:1210:1323 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/CUMEM +n116-042-186:1212:1321 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/CUMEM +n116-042-186:1211:1322 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/CUMEM +n116-042-186:1212:1321 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/CUMEM +n116-042-186:1211:1322 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/CUMEM +n116-042-186:1208:1319 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/CUMEM +n116-042-186:1212:1321 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/CUMEM +n116-042-186:1211:1322 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/CUMEM +n116-042-186:1208:1319 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/CUMEM +n116-042-186:1211:1322 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/CUMEM +n116-042-186:1208:1319 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/CUMEM +n116-042-186:1211:1322 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/CUMEM +n116-042-186:1210:1323 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/CUMEM +n116-042-186:1208:1319 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/CUMEM +n116-042-186:1211:1322 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/CUMEM +n116-042-186:1210:1323 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/CUMEM +n116-042-186:1208:1319 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/CUMEM +n116-042-186:1211:1322 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/CUMEM +n116-042-186:1210:1323 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/CUMEM +n116-042-186:1208:1319 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/CUMEM +n116-042-186:1211:1322 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/CUMEM +n116-042-186:1210:1323 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/CUMEM +n116-042-186:1208:1319 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/CUMEM +n116-042-186:1211:1322 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/CUMEM +n116-042-186:1210:1323 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/CUMEM +n116-042-186:1208:1319 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/CUMEM +n116-042-186:1211:1322 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/CUMEM +n116-042-186:1210:1323 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/CUMEM +n116-042-186:1208:1319 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/CUMEM +n116-042-186:1211:1322 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/CUMEM +n116-042-186:1210:1323 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/CUMEM +n116-042-186:1208:1319 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/CUMEM +n116-042-186:1209:1324 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/CUMEM +n116-042-186:1213:1326 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/CUMEM +n116-042-186:1210:1323 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/CUMEM +n116-042-186:1208:1319 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/CUMEM +n116-042-186:1209:1324 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/CUMEM +n116-042-186:1208:1319 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/CUMEM +n116-042-186:1213:1326 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/CUMEM +n116-042-186:1209:1324 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/CUMEM +n116-042-186:1208:1319 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/CUMEM +n116-042-186:1213:1326 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/CUMEM +n116-042-186:1208:1319 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/CUMEM +n116-042-186:1209:1324 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/CUMEM +n116-042-186:1213:1326 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/CUMEM +n116-042-186:1208:1319 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/CUMEM +n116-042-186:1209:1324 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/CUMEM +n116-042-186:1213:1326 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/CUMEM +n116-042-186:1209:1324 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/CUMEM +n116-042-186:1209:1324 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/CUMEM +n116-042-186:1213:1326 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/CUMEM +n116-042-186:1208:1319 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/CUMEM +n116-042-186:1209:1324 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/CUMEM +n116-042-186:1213:1326 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/CUMEM +n116-042-186:1209:1324 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/CUMEM +n116-042-186:1213:1326 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/CUMEM +n116-042-186:1209:1324 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/CUMEM +n116-042-186:1213:1326 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/CUMEM +n116-042-186:1206:1325 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/CUMEM +n116-042-186:1209:1324 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/CUMEM +n116-042-186:1213:1326 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/CUMEM +n116-042-186:1206:1325 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/CUMEM +n116-042-186:1209:1324 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/CUMEM +n116-042-186:1213:1326 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/CUMEM +n116-042-186:1206:1325 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/CUMEM +n116-042-186:1209:1324 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/CUMEM +n116-042-186:1213:1326 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/CUMEM +n116-042-186:1206:1325 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/CUMEM +n116-042-186:1209:1324 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/CUMEM +n116-042-186:1213:1326 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/CUMEM +n116-042-186:1209:1324 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/CUMEM +n116-042-186:1206:1325 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/CUMEM +n116-042-186:1213:1326 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/CUMEM +n116-042-186:1209:1324 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/CUMEM +n116-042-186:1206:1325 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/CUMEM +n116-042-186:1213:1326 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/CUMEM +n116-042-186:1209:1324 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/CUMEM +n116-042-186:1213:1326 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/CUMEM +n116-042-186:1206:1325 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/CUMEM +n116-042-186:1209:1324 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/CUMEM +n116-042-186:1213:1326 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/CUMEM +n116-042-186:1206:1325 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/CUMEM +n116-042-186:1209:1324 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/CUMEM +n116-042-186:1206:1325 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/CUMEM +n116-042-186:1209:1324 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/CUMEM +n116-042-186:1206:1325 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/CUMEM +n116-042-186:1213:1326 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/CUMEM +n116-042-186:1209:1324 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/CUMEM +n116-042-186:1206:1325 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/CUMEM +n116-042-186:1213:1326 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/CUMEM +n116-042-186:1209:1324 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/CUMEM +n116-042-186:1206:1325 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/CUMEM +n116-042-186:1213:1326 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/CUMEM +n116-042-186:1209:1324 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/CUMEM +n116-042-186:1206:1325 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/CUMEM +n116-042-186:1213:1326 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/CUMEM +n116-042-186:1209:1324 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/CUMEM +n116-042-186:1206:1325 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/CUMEM +n116-042-186:1213:1326 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/CUMEM +n116-042-186:1213:1326 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/CUMEM +n116-042-186:1206:1325 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/CUMEM +n116-042-186:1213:1326 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/CUMEM +n116-042-186:1206:1325 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/CUMEM +n116-042-186:1206:1325 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/CUMEM +n116-042-186:1206:1325 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/CUMEM +n116-042-186:1206:1325 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/CUMEM +n116-042-186:1206:1325 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/CUMEM +n116-042-186:1206:1325 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/CUMEM +n116-042-186:1206:1325 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/CUMEM +n116-042-186:1206:1325 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/CUMEM +n116-042-186:1206:1325 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/CUMEM +n116-042-186:1213:1326 [7] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-042-186:1212:1321 [6] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-042-186:1209:1324 [3] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-042-186:1211:1322 [5] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-042-186:1210:1323 [4] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-042-186:1207:1320 [1] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-042-186:1208:1319 [2] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +n116-042-186:1206:1325 [0] NCCL INFO Connected all rings, use ring PXN 0 GDR 1 +Namespace(batch_size=256, epochs=10, bce_loss=False, unscale_lr=True, select_models='scaledf_v4_train_fix.pkl', select_images='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/selections/fixed_budget_500k_all_46real_91fake_20260725/scaledf_v4_fixed500000_all_46real_91fake_imgseed20260725.pkl', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.2, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.1, sched='cosine', lr=1e-05, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-06, decay_epochs=15, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=False, train_mode=True, ThreeAugment=False, src=False, reprob=0.0, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='/tmp/ScaleDF_repro_20260724/pretrained/deit_3_base_224_21k.pth', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/ScaleDF', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_all_46real_91fake_imgseed20260725_n116-042-186_20260725_1215', device='cuda', seed=0, resume='', start_epoch=0, eval=False, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +Removing key head.weight from pretrained checkpoint +Removing key head.bias from pretrained checkpoint +number of params: 85817858 +Start training for 10 epochs +Epoch: [0] [ 0/488] eta: 1:18:55 lr: 0.000001 loss: 1.1268 (1.1268) time: 9.7032 data: 4.4583 max mem: 28119 +Epoch: [0] [ 10/488] eta: 0:08:56 lr: 0.000001 loss: 0.9832 (0.9776) time: 1.1223 data: 0.4057 max mem: 28824 +Epoch: [0] [ 20/488] eta: 0:05:23 lr: 0.000001 loss: 0.8666 (0.9128) time: 0.2414 data: 0.0004 max mem: 28824 +Epoch: [0] [ 30/488] eta: 0:04:47 lr: 0.000001 loss: 0.8249 (0.8807) time: 0.3546 data: 0.0189 max mem: 28824 +Epoch: [0] [ 40/488] eta: 0:04:30 lr: 0.000001 loss: 0.8049 (0.8625) time: 0.5099 data: 0.0220 max mem: 28824 +Epoch: [0] [ 50/488] eta: 0:04:13 lr: 0.000001 loss: 0.8010 (0.8497) time: 0.5044 data: 0.0225 max mem: 28824 +Epoch: [0] [ 60/488] eta: 0:03:59 lr: 0.000001 loss: 0.7948 (0.8409) time: 0.4699 data: 0.1435 max mem: 28824 +Epoch: [0] [ 70/488] eta: 0:03:52 lr: 0.000001 loss: 0.7826 (0.8326) time: 0.5000 data: 0.2182 max mem: 28824 +Epoch: [0] [ 80/488] eta: 0:03:40 lr: 0.000001 loss: 0.7805 (0.8260) time: 0.4852 data: 0.1201 max mem: 28824 +Epoch: [0] [ 90/488] eta: 0:03:28 lr: 0.000001 loss: 0.7722 (0.8197) time: 0.4060 data: 0.0618 max mem: 28824 +Epoch: [0] [100/488] eta: 0:03:22 lr: 0.000001 loss: 0.7744 (0.8158) time: 0.4452 data: 0.0754 max mem: 28824 +Epoch: [0] [110/488] eta: 0:03:13 lr: 0.000001 loss: 0.7744 (0.8114) time: 0.4540 data: 0.0399 max mem: 28825 +Epoch: [0] [120/488] eta: 0:03:07 lr: 0.000001 loss: 0.7650 (0.8070) time: 0.4381 data: 0.0005 max mem: 28825 +Epoch: [0] [130/488] eta: 0:02:56 lr: 0.000001 loss: 0.7521 (0.8034) time: 0.3996 data: 0.0005 max mem: 28825 +Epoch: [0] [140/488] eta: 0:02:52 lr: 0.000001 loss: 0.7518 (0.7999) time: 0.4115 data: 0.0005 max mem: 28825 +Epoch: [0] [150/488] eta: 0:02:46 lr: 0.000001 loss: 0.7506 (0.7966) time: 0.4748 data: 0.0005 max mem: 28825 +Epoch: [0] [160/488] eta: 0:02:40 lr: 0.000001 loss: 0.7506 (0.7939) time: 0.4613 data: 0.0005 max mem: 28825 +Epoch: [0] [170/488] eta: 0:02:31 lr: 0.000001 loss: 0.7459 (0.7910) time: 0.3749 data: 0.0004 max mem: 28825 +Epoch: [0] [180/488] eta: 0:02:28 lr: 0.000001 loss: 0.7403 (0.7883) time: 0.4244 data: 0.0004 max mem: 28825 +Epoch: [0] [190/488] eta: 0:02:24 lr: 0.000001 loss: 0.7403 (0.7859) time: 0.5453 data: 0.0005 max mem: 28825 +Epoch: [0] [200/488] eta: 0:02:19 lr: 0.000001 loss: 0.7378 (0.7831) time: 0.4848 data: 0.0005 max mem: 28825 +Epoch: [0] [210/488] eta: 0:02:13 lr: 0.000001 loss: 0.7205 (0.7802) time: 0.4303 data: 0.0005 max mem: 28825 +Epoch: [0] [220/488] eta: 0:02:08 lr: 0.000001 loss: 0.7253 (0.7780) time: 0.4593 data: 0.0005 max mem: 28825 +Epoch: [0] [230/488] eta: 0:02:02 lr: 0.000001 loss: 0.7279 (0.7758) time: 0.4178 data: 0.0005 max mem: 28825 +Epoch: [0] [240/488] eta: 0:01:57 lr: 0.000001 loss: 0.7288 (0.7742) time: 0.3955 data: 0.0005 max mem: 28825 +Epoch: [0] [250/488] eta: 0:01:53 lr: 0.000001 loss: 0.7284 (0.7721) time: 0.5115 data: 0.0005 max mem: 28825 +Epoch: [0] [260/488] eta: 0:01:48 lr: 0.000001 loss: 0.7199 (0.7700) time: 0.5094 data: 0.0005 max mem: 28825 +Epoch: [0] [270/488] eta: 0:01:43 lr: 0.000001 loss: 0.7128 (0.7679) time: 0.4262 data: 0.0005 max mem: 28825 +Epoch: [0] [280/488] eta: 0:01:37 lr: 0.000001 loss: 0.7157 (0.7663) time: 0.3911 data: 0.0005 max mem: 28825 +Epoch: [0] [290/488] eta: 0:01:33 lr: 0.000001 loss: 0.7196 (0.7647) time: 0.4455 data: 0.0019 max mem: 28825 +Epoch: [0] [300/488] eta: 0:01:28 lr: 0.000001 loss: 0.7179 (0.7631) time: 0.4544 data: 0.0020 max mem: 28825 +Epoch: [0] [310/488] eta: 0:01:23 lr: 0.000001 loss: 0.7172 (0.7617) time: 0.4292 data: 0.0019 max mem: 28825 +Epoch: [0] [320/488] eta: 0:01:18 lr: 0.000001 loss: 0.7166 (0.7604) time: 0.4504 data: 0.0019 max mem: 28825 +Epoch: [0] [330/488] eta: 0:01:14 lr: 0.000001 loss: 0.7142 (0.7590) time: 0.5037 data: 0.0005 max mem: 28825 +Epoch: [0] [340/488] eta: 0:01:09 lr: 0.000001 loss: 0.7156 (0.7579) time: 0.4747 data: 0.0019 max mem: 28825 +Epoch: [0] [350/488] eta: 0:01:04 lr: 0.000001 loss: 0.7128 (0.7565) time: 0.4307 data: 0.0019 max mem: 28825 +Epoch: [0] [360/488] eta: 0:00:59 lr: 0.000001 loss: 0.7083 (0.7553) time: 0.4491 data: 0.0005 max mem: 28825 +Epoch: [0] [370/488] eta: 0:00:54 lr: 0.000001 loss: 0.7086 (0.7541) time: 0.3691 data: 0.0004 max mem: 28825 +Epoch: [0] [380/488] eta: 0:00:49 lr: 0.000001 loss: 0.7097 (0.7530) time: 0.3396 data: 0.0004 max mem: 28825 +Epoch: [0] [390/488] eta: 0:00:45 lr: 0.000001 loss: 0.7103 (0.7519) time: 0.4611 data: 0.0004 max mem: 28825 +Epoch: [0] [400/488] eta: 0:00:40 lr: 0.000001 loss: 0.7099 (0.7510) time: 0.4958 data: 0.0005 max mem: 28825 +Epoch: [0] [410/488] eta: 0:00:36 lr: 0.000001 loss: 0.7079 (0.7499) time: 0.4646 data: 0.0005 max mem: 28825 +Epoch: [0] [420/488] eta: 0:00:31 lr: 0.000001 loss: 0.7066 (0.7489) time: 0.4449 data: 0.0005 max mem: 28825 +Epoch: [0] [430/488] eta: 0:00:26 lr: 0.000001 loss: 0.7050 (0.7479) time: 0.5090 data: 0.0005 max mem: 28825 +Epoch: [0] [440/488] eta: 0:00:22 lr: 0.000001 loss: 0.7050 (0.7470) time: 0.4987 data: 0.0005 max mem: 28825 +Epoch: [0] [450/488] eta: 0:00:17 lr: 0.000001 loss: 0.7071 (0.7460) time: 0.4002 data: 0.0005 max mem: 28825 +Epoch: [0] [460/488] eta: 0:00:12 lr: 0.000001 loss: 0.7053 (0.7451) time: 0.4753 data: 0.0005 max mem: 28825 +Epoch: [0] [470/488] eta: 0:00:08 lr: 0.000001 loss: 0.7030 (0.7442) time: 0.4609 data: 0.0009 max mem: 28825 +Epoch: [0] [480/488] eta: 0:00:03 lr: 0.000001 loss: 0.6955 (0.7432) time: 0.3393 data: 0.0007 max mem: 28825 +Epoch: [0] [487/488] eta: 0:00:00 lr: 0.000001 loss: 0.6948 (0.7426) time: 0.2784 data: 0.0005 max mem: 28825 +Epoch: [0] Total time: 0:03:43 (0.4584 s / it) +Averaged stats: lr: 0.000001 loss: 0.6948 (0.7437) +Epoch: [1] [ 0/488] eta: 0:38:21 lr: 0.000001 loss: 0.6939 (0.6939) time: 4.7164 data: 4.5161 max mem: 28825 +Epoch: [1] [ 10/488] eta: 0:05:39 lr: 0.000001 loss: 0.6939 (0.6963) time: 0.7105 data: 0.5033 max mem: 28825 +Epoch: [1] [ 20/488] eta: 0:04:20 lr: 0.000001 loss: 0.6975 (0.6991) time: 0.3489 data: 0.1337 max mem: 28825 +Epoch: [1] [ 30/488] eta: 0:03:52 lr: 0.000001 loss: 0.6979 (0.6995) time: 0.3944 data: 0.1760 max mem: 28825 +Epoch: [1] [ 40/488] eta: 0:03:36 lr: 0.000001 loss: 0.6958 (0.6995) time: 0.4053 data: 0.1855 max mem: 28825 +Epoch: [1] [ 50/488] eta: 0:03:31 lr: 0.000001 loss: 0.7037 (0.7002) time: 0.4436 data: 0.2171 max mem: 28825 +Epoch: [1] [ 60/488] eta: 0:03:30 lr: 0.000001 loss: 0.7013 (0.6999) time: 0.5071 data: 0.2633 max mem: 28825 +Epoch: [1] [ 70/488] eta: 0:03:12 lr: 0.000001 loss: 0.6941 (0.6992) time: 0.4047 data: 0.1679 max mem: 28825 +Epoch: [1] [ 80/488] eta: 0:03:05 lr: 0.000001 loss: 0.6959 (0.6991) time: 0.3431 data: 0.1107 max mem: 28825 +Epoch: [1] [ 90/488] eta: 0:03:05 lr: 0.000001 loss: 0.6991 (0.6990) time: 0.4822 data: 0.1702 max mem: 28825 +Epoch: [1] [100/488] eta: 0:02:57 lr: 0.000001 loss: 0.6996 (0.6991) time: 0.4704 data: 0.1740 max mem: 28825 +Epoch: [1] [110/488] eta: 0:02:55 lr: 0.000001 loss: 0.7001 (0.6987) time: 0.4648 data: 0.1797 max mem: 28825 +Epoch: [1] [120/488] eta: 0:02:46 lr: 0.000001 loss: 0.6946 (0.6985) time: 0.4338 data: 0.0951 max mem: 28825 +Epoch: [1] [130/488] eta: 0:02:41 lr: 0.000001 loss: 0.6976 (0.6983) time: 0.3745 data: 0.1017 max mem: 28825 +Epoch: [1] [140/488] eta: 0:02:37 lr: 0.000001 loss: 0.6984 (0.6983) time: 0.4509 data: 0.2365 max mem: 28825 +Epoch: [1] [150/488] eta: 0:02:33 lr: 0.000001 loss: 0.6964 (0.6980) time: 0.4752 data: 0.2452 max mem: 28825 +Epoch: [1] [160/488] eta: 0:02:29 lr: 0.000001 loss: 0.6904 (0.6971) time: 0.4629 data: 0.2317 max mem: 28825 +Epoch: [1] [170/488] eta: 0:02:24 lr: 0.000001 loss: 0.6901 (0.6965) time: 0.4490 data: 0.2343 max mem: 28825 +Epoch: [1] [180/488] eta: 0:02:19 lr: 0.000001 loss: 0.6925 (0.6966) time: 0.4263 data: 0.1938 max mem: 28825 +Epoch: [1] [190/488] eta: 0:02:14 lr: 0.000001 loss: 0.6921 (0.6961) time: 0.4210 data: 0.1725 max mem: 28825 +Epoch: [1] [200/488] eta: 0:02:09 lr: 0.000001 loss: 0.6897 (0.6962) time: 0.4398 data: 0.1918 max mem: 28825 +Epoch: [1] [210/488] eta: 0:02:07 lr: 0.000001 loss: 0.6951 (0.6959) time: 0.5544 data: 0.2706 max mem: 28825 +Epoch: [1] [220/488] eta: 0:02:01 lr: 0.000001 loss: 0.6919 (0.6959) time: 0.4706 data: 0.2020 max mem: 28825 +Epoch: [1] [230/488] eta: 0:01:56 lr: 0.000001 loss: 0.6917 (0.6956) time: 0.3793 data: 0.1346 max mem: 28825 +Epoch: [1] [240/488] eta: 0:01:54 lr: 0.000001 loss: 0.6917 (0.6955) time: 0.5549 data: 0.2768 max mem: 28825 +Epoch: [1] [250/488] eta: 0:01:48 lr: 0.000001 loss: 0.6894 (0.6952) time: 0.5002 data: 0.2152 max mem: 28825 +Epoch: [1] [260/488] eta: 0:01:43 lr: 0.000001 loss: 0.6883 (0.6951) time: 0.3641 data: 0.1016 max mem: 28825 +Epoch: [1] [270/488] eta: 0:01:39 lr: 0.000001 loss: 0.6868 (0.6950) time: 0.4315 data: 0.2050 max mem: 28825 +Epoch: [1] [280/488] eta: 0:01:33 lr: 0.000001 loss: 0.6852 (0.6947) time: 0.4201 data: 0.2060 max mem: 28825 +Epoch: [1] [290/488] eta: 0:01:29 lr: 0.000001 loss: 0.6861 (0.6944) time: 0.4333 data: 0.1821 max mem: 28825 +Epoch: [1] [300/488] eta: 0:01:25 lr: 0.000001 loss: 0.6885 (0.6944) time: 0.4803 data: 0.1905 max mem: 28825 +Epoch: [1] [310/488] eta: 0:01:21 lr: 0.000001 loss: 0.6907 (0.6943) time: 0.5098 data: 0.1400 max mem: 28825 +Epoch: [1] [320/488] eta: 0:01:16 lr: 0.000001 loss: 0.6872 (0.6939) time: 0.4496 data: 0.0705 max mem: 28825 +Epoch: [1] [330/488] eta: 0:01:11 lr: 0.000001 loss: 0.6863 (0.6939) time: 0.3757 data: 0.0005 max mem: 28825 +Epoch: [1] [340/488] eta: 0:01:06 lr: 0.000001 loss: 0.6863 (0.6936) time: 0.4346 data: 0.0005 max mem: 28825 +Epoch: [1] [350/488] eta: 0:01:02 lr: 0.000001 loss: 0.6844 (0.6935) time: 0.4398 data: 0.0005 max mem: 28825 +Epoch: [1] [360/488] eta: 0:00:57 lr: 0.000001 loss: 0.6856 (0.6933) time: 0.4314 data: 0.0087 max mem: 28825 +Epoch: [1] [370/488] eta: 0:00:53 lr: 0.000001 loss: 0.6862 (0.6931) time: 0.4202 data: 0.0087 max mem: 28825 +Epoch: [1] [380/488] eta: 0:00:48 lr: 0.000001 loss: 0.6842 (0.6929) time: 0.4870 data: 0.0005 max mem: 28825 +Epoch: [1] [390/488] eta: 0:00:44 lr: 0.000001 loss: 0.6884 (0.6928) time: 0.5401 data: 0.0041 max mem: 28825 +Epoch: [1] [400/488] eta: 0:00:39 lr: 0.000001 loss: 0.6891 (0.6927) time: 0.4715 data: 0.0221 max mem: 28825 +Epoch: [1] [410/488] eta: 0:00:35 lr: 0.000001 loss: 0.6877 (0.6926) time: 0.4435 data: 0.0400 max mem: 28825 +Epoch: [1] [420/488] eta: 0:00:30 lr: 0.000001 loss: 0.6854 (0.6924) time: 0.4606 data: 0.1068 max mem: 28825 +Epoch: [1] [430/488] eta: 0:00:26 lr: 0.000001 loss: 0.6854 (0.6922) time: 0.4105 data: 0.0987 max mem: 28825 +Epoch: [1] [440/488] eta: 0:00:21 lr: 0.000001 loss: 0.6875 (0.6920) time: 0.3942 data: 0.0360 max mem: 28825 +Epoch: [1] [450/488] eta: 0:00:17 lr: 0.000001 loss: 0.6798 (0.6917) time: 0.4497 data: 0.0227 max mem: 28825 +Epoch: [1] [460/488] eta: 0:00:12 lr: 0.000001 loss: 0.6803 (0.6915) time: 0.4352 data: 0.0006 max mem: 28825 +Epoch: [1] [470/488] eta: 0:00:08 lr: 0.000001 loss: 0.6837 (0.6914) time: 0.4199 data: 0.0494 max mem: 28825 +Epoch: [1] [480/488] eta: 0:00:03 lr: 0.000001 loss: 0.6837 (0.6913) time: 0.3691 data: 0.0492 max mem: 28825 +Epoch: [1] [487/488] eta: 0:00:00 lr: 0.000001 loss: 0.6839 (0.6912) time: 0.3452 data: 0.0490 max mem: 28825 +Epoch: [1] Total time: 0:03:38 (0.4475 s / it) +Averaged stats: lr: 0.000001 loss: 0.6839 (0.6917) +Epoch: [2] [ 0/488] eta: 0:36:43 lr: 0.000003 loss: 0.6652 (0.6652) time: 4.5156 data: 4.2974 max mem: 28825 +Epoch: [2] [ 10/488] eta: 0:05:41 lr: 0.000003 loss: 0.6777 (0.6786) time: 0.7139 data: 0.5083 max mem: 28825 +Epoch: [2] [ 20/488] eta: 0:04:12 lr: 0.000003 loss: 0.6778 (0.6813) time: 0.3397 data: 0.1244 max mem: 28825 +Epoch: [2] [ 30/488] eta: 0:04:03 lr: 0.000003 loss: 0.6819 (0.6821) time: 0.4298 data: 0.2123 max mem: 28825 +Epoch: [2] [ 40/488] eta: 0:03:35 lr: 0.000003 loss: 0.6877 (0.6836) time: 0.4209 data: 0.2081 max mem: 28825 +Epoch: [2] [ 50/488] eta: 0:03:22 lr: 0.000003 loss: 0.6828 (0.6823) time: 0.3550 data: 0.1247 max mem: 28825 +Epoch: [2] [ 60/488] eta: 0:03:25 lr: 0.000003 loss: 0.6811 (0.6824) time: 0.4762 data: 0.1419 max mem: 28825 +Epoch: [2] [ 70/488] eta: 0:03:22 lr: 0.000003 loss: 0.6814 (0.6823) time: 0.5387 data: 0.0728 max mem: 28825 +Epoch: [2] [ 80/488] eta: 0:03:15 lr: 0.000003 loss: 0.6813 (0.6822) time: 0.4736 data: 0.0004 max mem: 28825 +Epoch: [2] [ 90/488] eta: 0:03:08 lr: 0.000003 loss: 0.6805 (0.6820) time: 0.4403 data: 0.0004 max mem: 28825 +Epoch: [2] [100/488] eta: 0:03:01 lr: 0.000003 loss: 0.6787 (0.6813) time: 0.4309 data: 0.0004 max mem: 28825 +Epoch: [2] [110/488] eta: 0:02:54 lr: 0.000003 loss: 0.6751 (0.6808) time: 0.4112 data: 0.0005 max mem: 28825 +Epoch: [2] [120/488] eta: 0:02:48 lr: 0.000003 loss: 0.6775 (0.6807) time: 0.4047 data: 0.0005 max mem: 28825 +Epoch: [2] [130/488] eta: 0:02:42 lr: 0.000003 loss: 0.6753 (0.6803) time: 0.4138 data: 0.0004 max mem: 28825 +Epoch: [2] [140/488] eta: 0:02:38 lr: 0.000003 loss: 0.6785 (0.6806) time: 0.4352 data: 0.0004 max mem: 28825 +Epoch: [2] [150/488] eta: 0:02:33 lr: 0.000003 loss: 0.6824 (0.6805) time: 0.4434 data: 0.0087 max mem: 28825 +Epoch: [2] [160/488] eta: 0:02:29 lr: 0.000003 loss: 0.6778 (0.6801) time: 0.4727 data: 0.0088 max mem: 28825 +Epoch: [2] [170/488] eta: 0:02:25 lr: 0.000003 loss: 0.6716 (0.6797) time: 0.4898 data: 0.0006 max mem: 28825 +Epoch: [2] [180/488] eta: 0:02:20 lr: 0.000003 loss: 0.6726 (0.6796) time: 0.4418 data: 0.0005 max mem: 28825 +Epoch: [2] [190/488] eta: 0:02:17 lr: 0.000003 loss: 0.6665 (0.6791) time: 0.4773 data: 0.0004 max mem: 28825 +Epoch: [2] [200/488] eta: 0:02:11 lr: 0.000003 loss: 0.6708 (0.6792) time: 0.4843 data: 0.0004 max mem: 28825 +Epoch: [2] [210/488] eta: 0:02:08 lr: 0.000003 loss: 0.6780 (0.6791) time: 0.4838 data: 0.0005 max mem: 28825 +Epoch: [2] [220/488] eta: 0:02:02 lr: 0.000003 loss: 0.6775 (0.6792) time: 0.4410 data: 0.0005 max mem: 28825 +Epoch: [2] [230/488] eta: 0:01:58 lr: 0.000003 loss: 0.6810 (0.6793) time: 0.4309 data: 0.0005 max mem: 28825 +Epoch: [2] [240/488] eta: 0:01:53 lr: 0.000003 loss: 0.6784 (0.6794) time: 0.4797 data: 0.0005 max mem: 28825 +Epoch: [2] [250/488] eta: 0:01:48 lr: 0.000003 loss: 0.6784 (0.6793) time: 0.4000 data: 0.0005 max mem: 28825 +Epoch: [2] [260/488] eta: 0:01:42 lr: 0.000003 loss: 0.6772 (0.6790) time: 0.3605 data: 0.0005 max mem: 28825 +Epoch: [2] [270/488] eta: 0:01:38 lr: 0.000003 loss: 0.6774 (0.6791) time: 0.4395 data: 0.0005 max mem: 28825 +Epoch: [2] [280/488] eta: 0:01:34 lr: 0.000003 loss: 0.6765 (0.6789) time: 0.4847 data: 0.0005 max mem: 28825 +Epoch: [2] [290/488] eta: 0:01:30 lr: 0.000003 loss: 0.6721 (0.6789) time: 0.4943 data: 0.0005 max mem: 28825 +Epoch: [2] [300/488] eta: 0:01:26 lr: 0.000003 loss: 0.6718 (0.6787) time: 0.5260 data: 0.0005 max mem: 28825 +Epoch: [2] [310/488] eta: 0:01:21 lr: 0.000003 loss: 0.6746 (0.6787) time: 0.4965 data: 0.0013 max mem: 28825 +Epoch: [2] [320/488] eta: 0:01:16 lr: 0.000003 loss: 0.6746 (0.6786) time: 0.4279 data: 0.0013 max mem: 28825 +Epoch: [2] [330/488] eta: 0:01:11 lr: 0.000003 loss: 0.6734 (0.6786) time: 0.3642 data: 0.0005 max mem: 28825 +Epoch: [2] [340/488] eta: 0:01:06 lr: 0.000003 loss: 0.6693 (0.6783) time: 0.3855 data: 0.0005 max mem: 28825 +Epoch: [2] [350/488] eta: 0:01:02 lr: 0.000003 loss: 0.6659 (0.6780) time: 0.4639 data: 0.0005 max mem: 28825 +Epoch: [2] [360/488] eta: 0:00:57 lr: 0.000003 loss: 0.6739 (0.6779) time: 0.4445 data: 0.0005 max mem: 28825 +Epoch: [2] [370/488] eta: 0:00:53 lr: 0.000003 loss: 0.6791 (0.6780) time: 0.3992 data: 0.0005 max mem: 28825 +Epoch: [2] [380/488] eta: 0:00:49 lr: 0.000003 loss: 0.6821 (0.6781) time: 0.4907 data: 0.0005 max mem: 28825 +Epoch: [2] [390/488] eta: 0:00:44 lr: 0.000003 loss: 0.6804 (0.6780) time: 0.5057 data: 0.0017 max mem: 28825 +Epoch: [2] [400/488] eta: 0:00:39 lr: 0.000003 loss: 0.6707 (0.6779) time: 0.4340 data: 0.0018 max mem: 28825 +Epoch: [2] [410/488] eta: 0:00:35 lr: 0.000003 loss: 0.6700 (0.6777) time: 0.4555 data: 0.0005 max mem: 28825 +Epoch: [2] [420/488] eta: 0:00:30 lr: 0.000003 loss: 0.6714 (0.6776) time: 0.4857 data: 0.0005 max mem: 28825 +Epoch: [2] [430/488] eta: 0:00:26 lr: 0.000003 loss: 0.6748 (0.6776) time: 0.4721 data: 0.0004 max mem: 28825 +Epoch: [2] [440/488] eta: 0:00:21 lr: 0.000003 loss: 0.6714 (0.6774) time: 0.4149 data: 0.0004 max mem: 28825 +Epoch: [2] [450/488] eta: 0:00:17 lr: 0.000003 loss: 0.6682 (0.6773) time: 0.4262 data: 0.0005 max mem: 28825 +Epoch: [2] [460/488] eta: 0:00:12 lr: 0.000003 loss: 0.6716 (0.6772) time: 0.4302 data: 0.0005 max mem: 28825 +Epoch: [2] [470/488] eta: 0:00:08 lr: 0.000003 loss: 0.6746 (0.6771) time: 0.3920 data: 0.0008 max mem: 28825 +Epoch: [2] [480/488] eta: 0:00:03 lr: 0.000003 loss: 0.6710 (0.6769) time: 0.3758 data: 0.0006 max mem: 28825 +Epoch: [2] [487/488] eta: 0:00:00 lr: 0.000003 loss: 0.6695 (0.6769) time: 0.3221 data: 0.0005 max mem: 28825 +Epoch: [2] Total time: 0:03:39 (0.4489 s / it) +Averaged stats: lr: 0.000003 loss: 0.6695 (0.6769) +Epoch: [3] [ 0/488] eta: 0:40:20 lr: 0.000005 loss: 0.6676 (0.6676) time: 4.9593 data: 4.7591 max mem: 28825 +Epoch: [3] [ 10/488] eta: 0:05:29 lr: 0.000005 loss: 0.6708 (0.6692) time: 0.6886 data: 0.4367 max mem: 28825 +Epoch: [3] [ 20/488] eta: 0:04:11 lr: 0.000005 loss: 0.6704 (0.6678) time: 0.3161 data: 0.0761 max mem: 28825 +Epoch: [3] [ 30/488] eta: 0:03:51 lr: 0.000005 loss: 0.6696 (0.6692) time: 0.4059 data: 0.1883 max mem: 28825 +Epoch: [3] [ 40/488] eta: 0:03:48 lr: 0.000005 loss: 0.6696 (0.6696) time: 0.4802 data: 0.2706 max mem: 28825 +Epoch: [3] [ 50/488] eta: 0:03:26 lr: 0.000005 loss: 0.6719 (0.6704) time: 0.4179 data: 0.2042 max mem: 28825 +Epoch: [3] [ 60/488] eta: 0:03:17 lr: 0.000005 loss: 0.6748 (0.6707) time: 0.3659 data: 0.1490 max mem: 28825 +Epoch: [3] [ 70/488] eta: 0:03:11 lr: 0.000005 loss: 0.6694 (0.6706) time: 0.4219 data: 0.1581 max mem: 28825 +Epoch: [3] [ 80/488] eta: 0:03:08 lr: 0.000005 loss: 0.6679 (0.6702) time: 0.4618 data: 0.1126 max mem: 28825 +Epoch: [3] [ 90/488] eta: 0:03:02 lr: 0.000005 loss: 0.6679 (0.6698) time: 0.4675 data: 0.0566 max mem: 28825 +Epoch: [3] [100/488] eta: 0:02:58 lr: 0.000005 loss: 0.6620 (0.6688) time: 0.4454 data: 0.1006 max mem: 28825 +Epoch: [3] [110/488] eta: 0:02:54 lr: 0.000005 loss: 0.6644 (0.6695) time: 0.4717 data: 0.2373 max mem: 28825 +Epoch: [3] [120/488] eta: 0:02:49 lr: 0.000005 loss: 0.6694 (0.6689) time: 0.4709 data: 0.2028 max mem: 28825 +Epoch: [3] [130/488] eta: 0:02:45 lr: 0.000005 loss: 0.6689 (0.6692) time: 0.4708 data: 0.1717 max mem: 28825 +Epoch: [3] [140/488] eta: 0:02:43 lr: 0.000005 loss: 0.6693 (0.6697) time: 0.5301 data: 0.2287 max mem: 28825 +Epoch: [3] [150/488] eta: 0:02:37 lr: 0.000005 loss: 0.6701 (0.6693) time: 0.4843 data: 0.2143 max mem: 28825 +Epoch: [3] [160/488] eta: 0:02:31 lr: 0.000005 loss: 0.6701 (0.6693) time: 0.3882 data: 0.1640 max mem: 28825 +Epoch: [3] [170/488] eta: 0:02:25 lr: 0.000005 loss: 0.6739 (0.6698) time: 0.3959 data: 0.0848 max mem: 28825 +Epoch: [3] [180/488] eta: 0:02:24 lr: 0.000005 loss: 0.6739 (0.6695) time: 0.5370 data: 0.0149 max mem: 28825 +Epoch: [3] [190/488] eta: 0:02:18 lr: 0.000005 loss: 0.6626 (0.6687) time: 0.5187 data: 0.0019 max mem: 28825 +Epoch: [3] [200/488] eta: 0:02:13 lr: 0.000005 loss: 0.6569 (0.6680) time: 0.4113 data: 0.0005 max mem: 28825 +Epoch: [3] [210/488] eta: 0:02:06 lr: 0.000005 loss: 0.6634 (0.6679) time: 0.3813 data: 0.0005 max mem: 28825 +Epoch: [3] [220/488] eta: 0:02:02 lr: 0.000005 loss: 0.6661 (0.6678) time: 0.3944 data: 0.0005 max mem: 28825 +Epoch: [3] [230/488] eta: 0:01:58 lr: 0.000005 loss: 0.6727 (0.6680) time: 0.4785 data: 0.0005 max mem: 28825 +Epoch: [3] [240/488] eta: 0:01:53 lr: 0.000005 loss: 0.6705 (0.6680) time: 0.4796 data: 0.0006 max mem: 28825 +Epoch: [3] [250/488] eta: 0:01:49 lr: 0.000005 loss: 0.6584 (0.6677) time: 0.4815 data: 0.0005 max mem: 28825 +Epoch: [3] [260/488] eta: 0:01:45 lr: 0.000005 loss: 0.6584 (0.6675) time: 0.4990 data: 0.0006 max mem: 28825 +Epoch: [3] [270/488] eta: 0:01:40 lr: 0.000005 loss: 0.6671 (0.6676) time: 0.4517 data: 0.0006 max mem: 28825 +Epoch: [3] [280/488] eta: 0:01:35 lr: 0.000005 loss: 0.6704 (0.6675) time: 0.4061 data: 0.0006 max mem: 28825 +Epoch: [3] [290/488] eta: 0:01:30 lr: 0.000005 loss: 0.6635 (0.6672) time: 0.4126 data: 0.0005 max mem: 28825 +Epoch: [3] [300/488] eta: 0:01:25 lr: 0.000005 loss: 0.6635 (0.6671) time: 0.4321 data: 0.0005 max mem: 28825 +Epoch: [3] [310/488] eta: 0:01:21 lr: 0.000005 loss: 0.6717 (0.6671) time: 0.4550 data: 0.0005 max mem: 28825 +Epoch: [3] [320/488] eta: 0:01:16 lr: 0.000005 loss: 0.6690 (0.6668) time: 0.4730 data: 0.0005 max mem: 28825 +Epoch: [3] [330/488] eta: 0:01:12 lr: 0.000005 loss: 0.6686 (0.6667) time: 0.4703 data: 0.0005 max mem: 28825 +Epoch: [3] [340/488] eta: 0:01:07 lr: 0.000005 loss: 0.6675 (0.6667) time: 0.4344 data: 0.0021 max mem: 28825 +Epoch: [3] [350/488] eta: 0:01:02 lr: 0.000005 loss: 0.6641 (0.6665) time: 0.4035 data: 0.0021 max mem: 28825 +Epoch: [3] [360/488] eta: 0:00:58 lr: 0.000005 loss: 0.6641 (0.6664) time: 0.4611 data: 0.0005 max mem: 28825 +Epoch: [3] [370/488] eta: 0:00:53 lr: 0.000005 loss: 0.6648 (0.6663) time: 0.4860 data: 0.0004 max mem: 28825 +Epoch: [3] [380/488] eta: 0:00:49 lr: 0.000005 loss: 0.6589 (0.6660) time: 0.4184 data: 0.0005 max mem: 28825 +Epoch: [3] [390/488] eta: 0:00:44 lr: 0.000005 loss: 0.6569 (0.6658) time: 0.4220 data: 0.0005 max mem: 28825 +Epoch: [3] [400/488] eta: 0:00:39 lr: 0.000005 loss: 0.6578 (0.6655) time: 0.4542 data: 0.0004 max mem: 28825 +Epoch: [3] [410/488] eta: 0:00:35 lr: 0.000005 loss: 0.6616 (0.6655) time: 0.4464 data: 0.0005 max mem: 28825 +Epoch: [3] [420/488] eta: 0:00:31 lr: 0.000005 loss: 0.6616 (0.6654) time: 0.4999 data: 0.0005 max mem: 28825 +Epoch: [3] [430/488] eta: 0:00:26 lr: 0.000005 loss: 0.6593 (0.6652) time: 0.5594 data: 0.0005 max mem: 28825 +Epoch: [3] [440/488] eta: 0:00:21 lr: 0.000005 loss: 0.6663 (0.6651) time: 0.4310 data: 0.0005 max mem: 28825 +Epoch: [3] [450/488] eta: 0:00:17 lr: 0.000005 loss: 0.6663 (0.6648) time: 0.4056 data: 0.0963 max mem: 28825 +Epoch: [3] [460/488] eta: 0:00:12 lr: 0.000005 loss: 0.6471 (0.6644) time: 0.4492 data: 0.1758 max mem: 28825 +Epoch: [3] [470/488] eta: 0:00:08 lr: 0.000005 loss: 0.6471 (0.6641) time: 0.5417 data: 0.2318 max mem: 28825 +Epoch: [3] [480/488] eta: 0:00:03 lr: 0.000005 loss: 0.6551 (0.6641) time: 0.4630 data: 0.1717 max mem: 28825 +Epoch: [3] [487/488] eta: 0:00:00 lr: 0.000005 loss: 0.6650 (0.6641) time: 0.2768 data: 0.0493 max mem: 28825 +Epoch: [3] Total time: 0:03:42 (0.4560 s / it) +Averaged stats: lr: 0.000005 loss: 0.6650 (0.6645) +Epoch: [4] [ 0/488] eta: 0:50:15 lr: 0.000006 loss: 0.6299 (0.6299) time: 6.1796 data: 5.9753 max mem: 28825 +Epoch: [4] [ 10/488] eta: 0:07:15 lr: 0.000006 loss: 0.6639 (0.6551) time: 0.9119 data: 0.6977 max mem: 28825 +Epoch: [4] [ 20/488] eta: 0:05:12 lr: 0.000006 loss: 0.6639 (0.6557) time: 0.3927 data: 0.1751 max mem: 28825 +Epoch: [4] [ 30/488] eta: 0:04:26 lr: 0.000006 loss: 0.6591 (0.6556) time: 0.4000 data: 0.1807 max mem: 28825 +Epoch: [4] [ 40/488] eta: 0:04:10 lr: 0.000006 loss: 0.6525 (0.6550) time: 0.4452 data: 0.2302 max mem: 28825 +Epoch: [4] [ 50/488] eta: 0:03:57 lr: 0.000006 loss: 0.6525 (0.6552) time: 0.4800 data: 0.2682 max mem: 28825 +Epoch: [4] [ 60/488] eta: 0:03:45 lr: 0.000006 loss: 0.6631 (0.6569) time: 0.4613 data: 0.2503 max mem: 28825 +Epoch: [4] [ 70/488] eta: 0:03:30 lr: 0.000006 loss: 0.6591 (0.6568) time: 0.4053 data: 0.1686 max mem: 28825 +Epoch: [4] [ 80/488] eta: 0:03:24 lr: 0.000006 loss: 0.6611 (0.6570) time: 0.4183 data: 0.1283 max mem: 28825 +Epoch: [4] [ 90/488] eta: 0:03:17 lr: 0.000006 loss: 0.6586 (0.6572) time: 0.4682 data: 0.1537 max mem: 28825 +Epoch: [4] [100/488] eta: 0:03:07 lr: 0.000006 loss: 0.6537 (0.6568) time: 0.4202 data: 0.0930 max mem: 28825 +Epoch: [4] [110/488] eta: 0:03:02 lr: 0.000006 loss: 0.6452 (0.6556) time: 0.4269 data: 0.0819 max mem: 28825 +Epoch: [4] [120/488] eta: 0:02:56 lr: 0.000006 loss: 0.6531 (0.6565) time: 0.4498 data: 0.1027 max mem: 28825 +Epoch: [4] [130/488] eta: 0:02:51 lr: 0.000006 loss: 0.6610 (0.6567) time: 0.4645 data: 0.1152 max mem: 28825 +Epoch: [4] [140/488] eta: 0:02:45 lr: 0.000006 loss: 0.6610 (0.6569) time: 0.4508 data: 0.0933 max mem: 28825 +Epoch: [4] [150/488] eta: 0:02:40 lr: 0.000006 loss: 0.6592 (0.6569) time: 0.4496 data: 0.0199 max mem: 28825 +Epoch: [4] [160/488] eta: 0:02:33 lr: 0.000006 loss: 0.6592 (0.6568) time: 0.4128 data: 0.0005 max mem: 28825 +Epoch: [4] [170/488] eta: 0:02:28 lr: 0.000006 loss: 0.6606 (0.6570) time: 0.4100 data: 0.0005 max mem: 28825 +Epoch: [4] [180/488] eta: 0:02:23 lr: 0.000006 loss: 0.6635 (0.6575) time: 0.4593 data: 0.0005 max mem: 28825 +Epoch: [4] [190/488] eta: 0:02:19 lr: 0.000006 loss: 0.6665 (0.6574) time: 0.4503 data: 0.0005 max mem: 28825 +Epoch: [4] [200/488] eta: 0:02:13 lr: 0.000006 loss: 0.6610 (0.6573) time: 0.4505 data: 0.0005 max mem: 28825 +Epoch: [4] [210/488] eta: 0:02:09 lr: 0.000006 loss: 0.6610 (0.6573) time: 0.4749 data: 0.0005 max mem: 28825 +Epoch: [4] [220/488] eta: 0:02:03 lr: 0.000006 loss: 0.6626 (0.6573) time: 0.4316 data: 0.0005 max mem: 28825 +Epoch: [4] [230/488] eta: 0:01:58 lr: 0.000006 loss: 0.6504 (0.6568) time: 0.3800 data: 0.0005 max mem: 28825 +Epoch: [4] [240/488] eta: 0:01:55 lr: 0.000006 loss: 0.6507 (0.6570) time: 0.5157 data: 0.0004 max mem: 28825 +Epoch: [4] [250/488] eta: 0:01:50 lr: 0.000006 loss: 0.6476 (0.6566) time: 0.5296 data: 0.0004 max mem: 28825 +Epoch: [4] [260/488] eta: 0:01:45 lr: 0.000006 loss: 0.6416 (0.6565) time: 0.4146 data: 0.0004 max mem: 28825 +Epoch: [4] [270/488] eta: 0:01:41 lr: 0.000006 loss: 0.6391 (0.6559) time: 0.4509 data: 0.0005 max mem: 28825 +Epoch: [4] [280/488] eta: 0:01:36 lr: 0.000006 loss: 0.6406 (0.6554) time: 0.4544 data: 0.0005 max mem: 28825 +Epoch: [4] [290/488] eta: 0:01:31 lr: 0.000006 loss: 0.6442 (0.6551) time: 0.4226 data: 0.0004 max mem: 28825 +Epoch: [4] [300/488] eta: 0:01:27 lr: 0.000006 loss: 0.6493 (0.6550) time: 0.5085 data: 0.0004 max mem: 28825 +Epoch: [4] [310/488] eta: 0:01:21 lr: 0.000006 loss: 0.6510 (0.6549) time: 0.4318 data: 0.0004 max mem: 28825 +Epoch: [4] [320/488] eta: 0:01:17 lr: 0.000006 loss: 0.6569 (0.6550) time: 0.4174 data: 0.0004 max mem: 28825 +Epoch: [4] [330/488] eta: 0:01:13 lr: 0.000006 loss: 0.6466 (0.6548) time: 0.5198 data: 0.0004 max mem: 28825 +Epoch: [4] [340/488] eta: 0:01:08 lr: 0.000006 loss: 0.6466 (0.6544) time: 0.4227 data: 0.0004 max mem: 28825 +Epoch: [4] [350/488] eta: 0:01:03 lr: 0.000006 loss: 0.6499 (0.6543) time: 0.4147 data: 0.0005 max mem: 28825 +Epoch: [4] [360/488] eta: 0:00:58 lr: 0.000006 loss: 0.6576 (0.6544) time: 0.3961 data: 0.0005 max mem: 28825 +Epoch: [4] [370/488] eta: 0:00:53 lr: 0.000006 loss: 0.6601 (0.6545) time: 0.3403 data: 0.0005 max mem: 28825 +Epoch: [4] [380/488] eta: 0:00:49 lr: 0.000006 loss: 0.6584 (0.6546) time: 0.5197 data: 0.0192 max mem: 28825 +Epoch: [4] [390/488] eta: 0:00:44 lr: 0.000006 loss: 0.6504 (0.6546) time: 0.4599 data: 0.0192 max mem: 28825 +Epoch: [4] [400/488] eta: 0:00:39 lr: 0.000006 loss: 0.6481 (0.6544) time: 0.3399 data: 0.0004 max mem: 28825 +Epoch: [4] [410/488] eta: 0:00:35 lr: 0.000006 loss: 0.6494 (0.6542) time: 0.4003 data: 0.0005 max mem: 28825 +Epoch: [4] [420/488] eta: 0:00:30 lr: 0.000006 loss: 0.6481 (0.6540) time: 0.4400 data: 0.0005 max mem: 28825 +Epoch: [4] [430/488] eta: 0:00:26 lr: 0.000006 loss: 0.6519 (0.6541) time: 0.4649 data: 0.0005 max mem: 28825 +Epoch: [4] [440/488] eta: 0:00:21 lr: 0.000006 loss: 0.6523 (0.6540) time: 0.4412 data: 0.0005 max mem: 28825 +Epoch: [4] [450/488] eta: 0:00:17 lr: 0.000006 loss: 0.6486 (0.6539) time: 0.4245 data: 0.0004 max mem: 28825 +Epoch: [4] [460/488] eta: 0:00:12 lr: 0.000006 loss: 0.6504 (0.6539) time: 0.4639 data: 0.0004 max mem: 28825 +Epoch: [4] [470/488] eta: 0:00:08 lr: 0.000006 loss: 0.6396 (0.6534) time: 0.5031 data: 0.0023 max mem: 28825 +Epoch: [4] [480/488] eta: 0:00:03 lr: 0.000006 loss: 0.6312 (0.6533) time: 0.3774 data: 0.0366 max mem: 28825 +Epoch: [4] [487/488] eta: 0:00:00 lr: 0.000006 loss: 0.6473 (0.6532) time: 0.3540 data: 0.0685 max mem: 28825 +Epoch: [4] Total time: 0:03:39 (0.4491 s / it) +Averaged stats: lr: 0.000006 loss: 0.6473 (0.6539) +Test: [ 0/234] eta: 0:20:32 acc1_pos: 39.583 acc5_pos: 100.000 loss: 0.8075 (0.8075) acc1: 39.5833 (39.5833) acc5: 100.0000 (100.0000) time: 5.2692 data: 4.5405 max mem: 28825 +Test: [ 10/234] eta: 0:02:16 acc1_pos: 50.781 acc5_pos: 100.000 loss: 0.8009 (0.8097) acc1: 46.6146 (41.9981) acc5: 100.0000 (100.0000) time: 0.6094 data: 0.4597 max mem: 28825 +Test: [ 20/234] eta: 0:01:19 acc1_pos: 96.354 acc5_pos: 100.000 loss: 0.7856 (0.7610) acc1: 46.6146 (45.9821) acc5: 100.0000 (100.0000) time: 0.1269 data: 0.0361 max mem: 28825 +Test: [ 30/234] eta: 0:01:02 acc1_pos: 80.990 acc5_pos: 100.000 loss: 0.2768 (0.6072) acc1: 94.0104 (61.3071) acc5: 100.0000 (100.0000) time: 0.1413 data: 0.0482 max mem: 28825 +Test: [ 40/234] eta: 0:00:56 acc1_pos: 98.958 acc5_pos: 100.000 loss: 0.2757 (0.5650) acc1: 94.2708 (65.6885) acc5: 100.0000 (100.0000) time: 0.2048 data: 0.1088 max mem: 28825 +Test: [ 50/234] eta: 0:00:50 acc1_pos: 99.219 acc5_pos: 100.000 loss: 0.1931 (0.4900) acc1: 97.3958 (72.0588) acc5: 100.0000 (100.0000) time: 0.2258 data: 0.1329 max mem: 28825 +Test: [ 60/234] eta: 0:00:46 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.1758 (0.4382) acc1: 99.2188 (76.6137) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2115 data: 0.1204 max mem: 28825 +Test: [ 70/234] eta: 0:00:43 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.1788 (0.4579) acc1: 99.7396 (75.7629) acc5: 100.0000 (100.0000) acc1_neg: 53.385 acc5_neg: 100.000 time: 0.2436 data: 0.1511 max mem: 28825 +Test: [ 80/234] eta: 0:00:39 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.7585 (0.4733) acc1: 57.0312 (75.0675) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2331 data: 0.1381 max mem: 28825 +Test: [ 90/234] eta: 0:00:35 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.3269 (0.4579) acc1: 91.1458 (76.7485) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.1639 data: 0.0659 max mem: 28825 +Test: [100/234] eta: 0:00:32 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.3395 (0.4684) acc1: 86.7188 (76.2041) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1700 data: 0.0712 max mem: 28825 +Test: [110/234] eta: 0:00:29 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.4945 (0.4754) acc1: 71.0938 (76.0769) acc5: 100.0000 (100.0000) acc1_neg: 61.979 acc5_neg: 100.000 time: 0.2219 data: 0.1234 max mem: 28825 +Test: [120/234] eta: 0:00:27 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.4761 (0.4798) acc1: 80.2083 (76.0890) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.2249 data: 0.1279 max mem: 28825 +Test: [130/234] eta: 0:00:23 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.4134 (0.4760) acc1: 87.2396 (76.8050) acc5: 100.0000 (100.0000) acc1_neg: 74.740 acc5_neg: 100.000 time: 0.1650 data: 0.0680 max mem: 28825 +Test: [140/234] eta: 0:00:20 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.4976 (0.5044) acc1: 78.1250 (75.2770) acc5: 100.0000 (100.0000) acc1_neg: 26.823 acc5_neg: 100.000 time: 0.1361 data: 0.0397 max mem: 28825 +Test: [150/234] eta: 0:00:18 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.6734 (0.5179) acc1: 65.3646 (74.6327) acc5: 100.0000 (100.0000) acc1_neg: 72.135 acc5_neg: 100.000 time: 0.1755 data: 0.0824 max mem: 28825 +Test: [160/234] eta: 0:00:16 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.4472 (0.5094) acc1: 82.8125 (75.4933) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.2046 data: 0.1132 max mem: 28825 +Test: [170/234] eta: 0:00:14 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.4275 (0.5049) acc1: 84.6354 (76.0691) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.2194 data: 0.1280 max mem: 28825 +Test: [180/234] eta: 0:00:11 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.4403 (0.5153) acc1: 84.1146 (75.2532) acc5: 100.0000 (100.0000) acc1_neg: 27.083 acc5_neg: 100.000 time: 0.1940 data: 0.1028 max mem: 28825 +Test: [190/234] eta: 0:00:09 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.6446 (0.5236) acc1: 62.2396 (74.5351) acc5: 100.0000 (100.0000) acc1_neg: 63.542 acc5_neg: 100.000 time: 0.1437 data: 0.0536 max mem: 28825 +Test: [200/234] eta: 0:00:07 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.6319 (0.5117) acc1: 65.1042 (75.1257) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1318 data: 0.0416 max mem: 28825 +Test: [210/234] eta: 0:00:04 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.0791 (0.4924) acc1: 99.7396 (76.2959) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1394 data: 0.0492 max mem: 28825 +Test: [220/234] eta: 0:00:02 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.1532 (0.4783) acc1: 99.7396 (77.2530) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.1200 data: 0.0305 max mem: 28825 +Test: [230/234] eta: 0:00:00 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.3902 (0.4929) acc1: 45.5729 (75.7001) acc5: 100.0000 (100.0000) acc1_neg: 40.365 acc5_neg: 100.000 time: 0.1115 data: 0.0221 max mem: 28825 +Test: [233/234] eta: 0:00:00 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.7904 (0.4970) acc1: 42.1875 (75.3045) acc5: 100.0000 (100.0000) acc1_neg: 47.471 acc5_neg: 100.000 time: 0.1119 data: 0.0213 max mem: 28825 +Test: Total time: 0:00:46 (0.1973 s / it) +* Overall Acc@1 75.305 Acc@5 100.000 + target == real Acc@1 72.254 Acc@5 100.000 + target == fake Acc@1 76.164 Acc@5 100.000 + loss 0.497 +Accuracy of the network on the 89729 test images: 75.3% +Max accuracy: 75.30% +Epoch: [5] [ 0/488] eta: 0:41:16 lr: 0.000008 loss: 0.6633 (0.6633) time: 5.0757 data: 4.8699 max mem: 28825 +Epoch: [5] [ 10/488] eta: 0:06:15 lr: 0.000008 loss: 0.6632 (0.6607) time: 0.7857 data: 0.5753 max mem: 28826 +Epoch: [5] [ 20/488] eta: 0:04:52 lr: 0.000008 loss: 0.6579 (0.6519) time: 0.4026 data: 0.1876 max mem: 28826 +Epoch: [5] [ 30/488] eta: 0:04:15 lr: 0.000008 loss: 0.6522 (0.6526) time: 0.4341 data: 0.2159 max mem: 28826 +Epoch: [5] [ 40/488] eta: 0:03:58 lr: 0.000008 loss: 0.6482 (0.6504) time: 0.4350 data: 0.1868 max mem: 28826 +Epoch: [5] [ 50/488] eta: 0:03:50 lr: 0.000008 loss: 0.6577 (0.6531) time: 0.4751 data: 0.1844 max mem: 28827 +Epoch: [5] [ 60/488] eta: 0:03:34 lr: 0.000008 loss: 0.6472 (0.6513) time: 0.4390 data: 0.1796 max mem: 28827 +Epoch: [5] [ 70/488] eta: 0:03:26 lr: 0.000008 loss: 0.6412 (0.6500) time: 0.4149 data: 0.1647 max mem: 28827 +Epoch: [5] [ 80/488] eta: 0:03:17 lr: 0.000008 loss: 0.6473 (0.6502) time: 0.4261 data: 0.1269 max mem: 28827 +Epoch: [5] [ 90/488] eta: 0:03:13 lr: 0.000008 loss: 0.6549 (0.6505) time: 0.4599 data: 0.0918 max mem: 28827 +Epoch: [5] [100/488] eta: 0:03:06 lr: 0.000008 loss: 0.6476 (0.6494) time: 0.4647 data: 0.0932 max mem: 28827 +Epoch: [5] [110/488] eta: 0:03:00 lr: 0.000008 loss: 0.6468 (0.6493) time: 0.4303 data: 0.1366 max mem: 28827 +Epoch: [5] [120/488] eta: 0:02:53 lr: 0.000008 loss: 0.6525 (0.6501) time: 0.4313 data: 0.1272 max mem: 28827 +Epoch: [5] [130/488] eta: 0:02:50 lr: 0.000008 loss: 0.6559 (0.6502) time: 0.4754 data: 0.0351 max mem: 28827 +Epoch: [5] [140/488] eta: 0:02:45 lr: 0.000008 loss: 0.6453 (0.6495) time: 0.5088 data: 0.0005 max mem: 28827 +Epoch: [5] [150/488] eta: 0:02:41 lr: 0.000008 loss: 0.6428 (0.6492) time: 0.4892 data: 0.0508 max mem: 28827 +Epoch: [5] [160/488] eta: 0:02:35 lr: 0.000008 loss: 0.6443 (0.6494) time: 0.4598 data: 0.0993 max mem: 28827 +Epoch: [5] [170/488] eta: 0:02:28 lr: 0.000008 loss: 0.6443 (0.6486) time: 0.3935 data: 0.0490 max mem: 28827 +Epoch: [5] [180/488] eta: 0:02:23 lr: 0.000008 loss: 0.6514 (0.6487) time: 0.4034 data: 0.0164 max mem: 28827 +Epoch: [5] [190/488] eta: 0:02:18 lr: 0.000008 loss: 0.6534 (0.6482) time: 0.4315 data: 0.0164 max mem: 28827 +Epoch: [5] [200/488] eta: 0:02:16 lr: 0.000008 loss: 0.6551 (0.6487) time: 0.5322 data: 0.0005 max mem: 28827 +Epoch: [5] [210/488] eta: 0:02:12 lr: 0.000008 loss: 0.6595 (0.6488) time: 0.5835 data: 0.0005 max mem: 28827 +Epoch: [5] [220/488] eta: 0:02:07 lr: 0.000008 loss: 0.6493 (0.6486) time: 0.4900 data: 0.0005 max mem: 28827 +Epoch: [5] [230/488] eta: 0:02:07 lr: 0.000008 loss: 0.6417 (0.6487) time: 0.6899 data: 0.0005 max mem: 28827 +Epoch: [5] [240/488] eta: 0:02:07 lr: 0.000008 loss: 0.6479 (0.6490) time: 0.9434 data: 0.0005 max mem: 28827 +Epoch: [5] [250/488] eta: 0:02:05 lr: 0.000008 loss: 0.6550 (0.6492) time: 0.9307 data: 0.0006 max mem: 28827 +Epoch: [5] [260/488] eta: 0:02:07 lr: 0.000008 loss: 0.6565 (0.6494) time: 1.0901 data: 0.0006 max mem: 28827 +Epoch: [5] [270/488] eta: 0:02:06 lr: 0.000008 loss: 0.6557 (0.6496) time: 1.2297 data: 0.0251 max mem: 28827 +Epoch: [5] [280/488] eta: 0:02:06 lr: 0.000008 loss: 0.6529 (0.6495) time: 1.2547 data: 0.2040 max mem: 28827 +Epoch: [5] [290/488] eta: 0:02:07 lr: 0.000008 loss: 0.6544 (0.6497) time: 1.4648 data: 0.4332 max mem: 28827 +Epoch: [5] [300/488] eta: 0:02:04 lr: 0.000008 loss: 0.6279 (0.6487) time: 1.4393 data: 0.5411 max mem: 28827 +Epoch: [5] [310/488] eta: 0:02:01 lr: 0.000008 loss: 0.6279 (0.6485) time: 1.2701 data: 0.4984 max mem: 28827 +Epoch: [5] [320/488] eta: 0:01:56 lr: 0.000008 loss: 0.6406 (0.6480) time: 1.2007 data: 0.3969 max mem: 28827 +Epoch: [5] [330/488] eta: 0:01:52 lr: 0.000008 loss: 0.6498 (0.6482) time: 1.1952 data: 0.5294 max mem: 28827 +Epoch: [5] [340/488] eta: 0:01:48 lr: 0.000008 loss: 0.6515 (0.6480) time: 1.2791 data: 0.7540 max mem: 28827 +Epoch: [5] [350/488] eta: 0:01:39 lr: 0.000008 loss: 0.6335 (0.6477) time: 0.9100 data: 0.4362 max mem: 28827 +Epoch: [5] [360/488] eta: 0:01:32 lr: 0.000008 loss: 0.6408 (0.6474) time: 0.6030 data: 0.0262 max mem: 28827 +Epoch: [5] [370/488] eta: 0:01:23 lr: 0.000008 loss: 0.6491 (0.6473) time: 0.4565 data: 0.0005 max mem: 28827 +Epoch: [5] [380/488] eta: 0:01:15 lr: 0.000008 loss: 0.6408 (0.6472) time: 0.3088 data: 0.0005 max mem: 28827 +Epoch: [5] [390/488] eta: 0:01:08 lr: 0.000008 loss: 0.6413 (0.6472) time: 0.4494 data: 0.0004 max mem: 28827 +Epoch: [5] [400/488] eta: 0:01:00 lr: 0.000008 loss: 0.6413 (0.6469) time: 0.4930 data: 0.0266 max mem: 28827 +Epoch: [5] [410/488] eta: 0:00:53 lr: 0.000008 loss: 0.6364 (0.6466) time: 0.4991 data: 0.0811 max mem: 28827 +Epoch: [5] [420/488] eta: 0:00:46 lr: 0.000008 loss: 0.6229 (0.6459) time: 0.4909 data: 0.1042 max mem: 28827 +Epoch: [5] [430/488] eta: 0:00:39 lr: 0.000008 loss: 0.6249 (0.6457) time: 0.4335 data: 0.0708 max mem: 28827 +Epoch: [5] [440/488] eta: 0:00:32 lr: 0.000008 loss: 0.6384 (0.6456) time: 0.4406 data: 0.0215 max mem: 28827 +Epoch: [5] [450/488] eta: 0:00:25 lr: 0.000008 loss: 0.6353 (0.6453) time: 0.5071 data: 0.0004 max mem: 28827 +Epoch: [5] [460/488] eta: 0:00:18 lr: 0.000008 loss: 0.6353 (0.6452) time: 0.4554 data: 0.0005 max mem: 28827 +Epoch: [5] [470/488] eta: 0:00:11 lr: 0.000008 loss: 0.6374 (0.6449) time: 0.4255 data: 0.0008 max mem: 28827 +Epoch: [5] [480/488] eta: 0:00:05 lr: 0.000008 loss: 0.6442 (0.6449) time: 0.4076 data: 0.0006 max mem: 28827 +Epoch: [5] [487/488] eta: 0:00:00 lr: 0.000008 loss: 0.6480 (0.6449) time: 0.3307 data: 0.0005 max mem: 28827 +Epoch: [5] Total time: 0:05:16 (0.6486 s / it) +Averaged stats: lr: 0.000008 loss: 0.6480 (0.6449) +Epoch: [6] [ 0/488] eta: 0:37:18 lr: 0.000005 loss: 0.6669 (0.6669) time: 4.5873 data: 4.3861 max mem: 28827 +Epoch: [6] [ 10/488] eta: 0:05:21 lr: 0.000005 loss: 0.6571 (0.6547) time: 0.6721 data: 0.4594 max mem: 28827 +Epoch: [6] [ 20/488] eta: 0:04:40 lr: 0.000005 loss: 0.6531 (0.6470) time: 0.3991 data: 0.1808 max mem: 28827 +Epoch: [6] [ 30/488] eta: 0:04:38 lr: 0.000005 loss: 0.6289 (0.6419) time: 0.5714 data: 0.3568 max mem: 28827 +Epoch: [6] [ 40/488] eta: 0:04:11 lr: 0.000005 loss: 0.6321 (0.6422) time: 0.5237 data: 0.2796 max mem: 28828 +Epoch: [6] [ 50/488] eta: 0:04:02 lr: 0.000005 loss: 0.6380 (0.6433) time: 0.4719 data: 0.1210 max mem: 28828 +Epoch: [6] [ 60/488] eta: 0:04:14 lr: 0.000005 loss: 0.6389 (0.6407) time: 0.6592 data: 0.2114 max mem: 28828 +Epoch: [6] [ 70/488] eta: 0:04:34 lr: 0.000005 loss: 0.6402 (0.6405) time: 0.9148 data: 0.3021 max mem: 28828 +Epoch: [6] [ 80/488] eta: 0:04:57 lr: 0.000005 loss: 0.6343 (0.6402) time: 1.1449 data: 0.3410 max mem: 28828 +Epoch: [6] [ 90/488] eta: 0:05:20 lr: 0.000005 loss: 0.6486 (0.6417) time: 1.3392 data: 0.4571 max mem: 28828 +Epoch: [6] [100/488] eta: 0:05:27 lr: 0.000005 loss: 0.6512 (0.6409) time: 1.3050 data: 0.5720 max mem: 28828 +Epoch: [6] [110/488] eta: 0:05:33 lr: 0.000005 loss: 0.6412 (0.6410) time: 1.2347 data: 0.4855 max mem: 28828 +Epoch: [6] [120/488] eta: 0:05:46 lr: 0.000005 loss: 0.6480 (0.6410) time: 1.4405 data: 0.1713 max mem: 28828 +Epoch: [6] [130/488] eta: 0:05:47 lr: 0.000005 loss: 0.6479 (0.6411) time: 1.4645 data: 0.0006 max mem: 28828 +Epoch: [6] [140/488] eta: 0:05:45 lr: 0.000005 loss: 0.6370 (0.6401) time: 1.2944 data: 0.0006 max mem: 28828 +Epoch: [6] [150/488] eta: 0:05:41 lr: 0.000005 loss: 0.6319 (0.6401) time: 1.2545 data: 0.0006 max mem: 28828 +Epoch: [6] [160/488] eta: 0:05:36 lr: 0.000005 loss: 0.6308 (0.6393) time: 1.2708 data: 0.0007 max mem: 28828 +Epoch: [6] [170/488] eta: 0:05:38 lr: 0.000005 loss: 0.6459 (0.6401) time: 1.4820 data: 0.0007 max mem: 28828 +Epoch: [6] [180/488] eta: 0:05:34 lr: 0.000005 loss: 0.6495 (0.6399) time: 1.5755 data: 0.0006 max mem: 28828 +Epoch: [6] [190/488] eta: 0:05:28 lr: 0.000005 loss: 0.6350 (0.6399) time: 1.4296 data: 0.1743 max mem: 28828 +Epoch: [6] [200/488] eta: 0:05:22 lr: 0.000005 loss: 0.6316 (0.6394) time: 1.4003 data: 0.3592 max mem: 28828 +Epoch: [6] [210/488] eta: 0:05:17 lr: 0.000005 loss: 0.6328 (0.6392) time: 1.5202 data: 0.2851 max mem: 28828 +Epoch: [6] [220/488] eta: 0:05:05 lr: 0.000005 loss: 0.6487 (0.6395) time: 1.3450 data: 0.1002 max mem: 28828 +Epoch: [6] [230/488] eta: 0:04:54 lr: 0.000005 loss: 0.6524 (0.6393) time: 1.1502 data: 0.0006 max mem: 28828 +Epoch: [6] [240/488] eta: 0:04:44 lr: 0.000005 loss: 0.6329 (0.6392) time: 1.2398 data: 0.0007 max mem: 28828 +Epoch: [6] [250/488] eta: 0:04:37 lr: 0.000005 loss: 0.6329 (0.6391) time: 1.4207 data: 0.0957 max mem: 28828 +Epoch: [6] [260/488] eta: 0:04:27 lr: 0.000005 loss: 0.6442 (0.6387) time: 1.4907 data: 0.0957 max mem: 28828 +Epoch: [6] [270/488] eta: 0:04:14 lr: 0.000005 loss: 0.6465 (0.6389) time: 1.1689 data: 0.0007 max mem: 28828 +Epoch: [6] [280/488] eta: 0:04:02 lr: 0.000005 loss: 0.6459 (0.6391) time: 1.0333 data: 0.0007 max mem: 28828 +Epoch: [6] [290/488] eta: 0:03:52 lr: 0.000005 loss: 0.6409 (0.6392) time: 1.2747 data: 0.0007 max mem: 28828 +Epoch: [6] [300/488] eta: 0:03:41 lr: 0.000005 loss: 0.6424 (0.6392) time: 1.3908 data: 0.0503 max mem: 28828 +Epoch: [6] [310/488] eta: 0:03:28 lr: 0.000005 loss: 0.6458 (0.6395) time: 1.1400 data: 0.1002 max mem: 28828 +Epoch: [6] [320/488] eta: 0:03:17 lr: 0.000005 loss: 0.6510 (0.6400) time: 1.1000 data: 0.3021 max mem: 28828 +Epoch: [6] [330/488] eta: 0:03:05 lr: 0.000005 loss: 0.6523 (0.6403) time: 1.2546 data: 0.5629 max mem: 28828 +Epoch: [6] [340/488] eta: 0:02:53 lr: 0.000005 loss: 0.6402 (0.6402) time: 1.1541 data: 0.6402 max mem: 28828 +Epoch: [6] [350/488] eta: 0:02:43 lr: 0.000005 loss: 0.6281 (0.6399) time: 1.2701 data: 0.8207 max mem: 28828 +Epoch: [6] [360/488] eta: 0:02:30 lr: 0.000005 loss: 0.6311 (0.6398) time: 1.2308 data: 0.6415 max mem: 28828 +Epoch: [6] [370/488] eta: 0:02:19 lr: 0.000005 loss: 0.6434 (0.6399) time: 1.1552 data: 0.2554 max mem: 28828 +Epoch: [6] [380/488] eta: 0:02:07 lr: 0.000005 loss: 0.6474 (0.6402) time: 1.3142 data: 0.3061 max mem: 28828 +Epoch: [6] [390/488] eta: 0:01:56 lr: 0.000005 loss: 0.6492 (0.6405) time: 1.3848 data: 0.5049 max mem: 28828 +Epoch: [6] [400/488] eta: 0:01:44 lr: 0.000005 loss: 0.6451 (0.6401) time: 1.3400 data: 0.3297 max mem: 28828 +Epoch: [6] [410/488] eta: 0:01:32 lr: 0.000005 loss: 0.6322 (0.6398) time: 1.1443 data: 0.0258 max mem: 28828 +Epoch: [6] [420/488] eta: 0:01:20 lr: 0.000005 loss: 0.6326 (0.6399) time: 1.1203 data: 0.0006 max mem: 28828 +Epoch: [6] [430/488] eta: 0:01:09 lr: 0.000005 loss: 0.6366 (0.6397) time: 1.3012 data: 0.0006 max mem: 28828 +Epoch: [6] [440/488] eta: 0:00:57 lr: 0.000005 loss: 0.6460 (0.6397) time: 1.2044 data: 0.0007 max mem: 28828 +Epoch: [6] [450/488] eta: 0:00:45 lr: 0.000005 loss: 0.6501 (0.6399) time: 1.1544 data: 0.0372 max mem: 28828 +Epoch: [6] [460/488] eta: 0:00:33 lr: 0.000005 loss: 0.6335 (0.6397) time: 1.3603 data: 0.0401 max mem: 28828 +Epoch: [6] [470/488] eta: 0:00:21 lr: 0.000005 loss: 0.6295 (0.6395) time: 1.3256 data: 0.0046 max mem: 28828 +Epoch: [6] [480/488] eta: 0:00:09 lr: 0.000005 loss: 0.6295 (0.6395) time: 0.9848 data: 0.0014 max mem: 28828 +Epoch: [6] [487/488] eta: 0:00:01 lr: 0.000005 loss: 0.6397 (0.6395) time: 0.5762 data: 0.0012 max mem: 28828 +Epoch: [6] Total time: 0:09:36 (1.1815 s / it) +Averaged stats: lr: 0.000005 loss: 0.6397 (0.6390) +Epoch: [7] [ 0/488] eta: 2:23:52 lr: 0.000004 loss: 0.6646 (0.6646) time: 17.6892 data: 17.2759 max mem: 28828 +Epoch: [7] [ 10/488] eta: 0:19:28 lr: 0.000004 loss: 0.6341 (0.6314) time: 2.4436 data: 2.0073 max mem: 28828 +Epoch: [7] [ 20/488] eta: 0:13:55 lr: 0.000004 loss: 0.6298 (0.6303) time: 0.9899 data: 0.5646 max mem: 28828 +Epoch: [7] [ 30/488] eta: 0:12:21 lr: 0.000004 loss: 0.6314 (0.6310) time: 1.1650 data: 0.7436 max mem: 28828 +Epoch: [7] [ 40/488] eta: 0:10:59 lr: 0.000004 loss: 0.6341 (0.6317) time: 1.1438 data: 0.7155 max mem: 28828 +Epoch: [7] [ 50/488] eta: 0:10:12 lr: 0.000004 loss: 0.6303 (0.6299) time: 1.0556 data: 0.6526 max mem: 28828 +Epoch: [7] [ 60/488] eta: 0:09:41 lr: 0.000004 loss: 0.6291 (0.6298) time: 1.1251 data: 0.7276 max mem: 28828 +Epoch: [7] [ 70/488] eta: 0:09:09 lr: 0.000004 loss: 0.6309 (0.6305) time: 1.0995 data: 0.6664 max mem: 28828 +Epoch: [7] [ 80/488] eta: 0:08:55 lr: 0.000004 loss: 0.6337 (0.6310) time: 1.1707 data: 0.7454 max mem: 28828 +Epoch: [7] [ 90/488] eta: 0:08:38 lr: 0.000004 loss: 0.6466 (0.6331) time: 1.2598 data: 0.8538 max mem: 28828 +Epoch: [7] [100/488] eta: 0:08:24 lr: 0.000004 loss: 0.6354 (0.6314) time: 1.2491 data: 0.8580 max mem: 28828 +Epoch: [7] [110/488] eta: 0:08:02 lr: 0.000004 loss: 0.6247 (0.6312) time: 1.1553 data: 0.5700 max mem: 28828 +Epoch: [7] [120/488] eta: 0:07:55 lr: 0.000004 loss: 0.6395 (0.6318) time: 1.2507 data: 0.3034 max mem: 28828 +Epoch: [7] [130/488] eta: 0:07:33 lr: 0.000004 loss: 0.6285 (0.6307) time: 1.2137 data: 0.3682 max mem: 28828 +Epoch: [7] [140/488] eta: 0:07:20 lr: 0.000004 loss: 0.6307 (0.6312) time: 1.1090 data: 0.3005 max mem: 28828 +Epoch: [7] [150/488] eta: 0:07:17 lr: 0.000004 loss: 0.6515 (0.6321) time: 1.4707 data: 0.2858 max mem: 28828 +Epoch: [7] [160/488] eta: 0:07:06 lr: 0.000004 loss: 0.6443 (0.6326) time: 1.5408 data: 0.1661 max mem: 28828 +Epoch: [7] [170/488] eta: 0:06:42 lr: 0.000004 loss: 0.6415 (0.6334) time: 1.0600 data: 0.0006 max mem: 28828 +Epoch: [7] [180/488] eta: 0:06:31 lr: 0.000004 loss: 0.6318 (0.6331) time: 1.0297 data: 0.0006 max mem: 28828 +Epoch: [7] [190/488] eta: 0:06:16 lr: 0.000004 loss: 0.6318 (0.6331) time: 1.2544 data: 0.0007 max mem: 28828 +Epoch: [7] [200/488] eta: 0:06:03 lr: 0.000004 loss: 0.6466 (0.6336) time: 1.1798 data: 0.0038 max mem: 28828 +Epoch: [7] [210/488] eta: 0:05:46 lr: 0.000004 loss: 0.6466 (0.6343) time: 1.0805 data: 0.0037 max mem: 28828 +Epoch: [7] [220/488] eta: 0:05:34 lr: 0.000004 loss: 0.6473 (0.6348) time: 1.1048 data: 0.0007 max mem: 28828 +Epoch: [7] [230/488] eta: 0:05:23 lr: 0.000004 loss: 0.6333 (0.6347) time: 1.3348 data: 0.0006 max mem: 28828 +Epoch: [7] [240/488] eta: 0:05:13 lr: 0.000004 loss: 0.6326 (0.6345) time: 1.4600 data: 0.0034 max mem: 28828 +Epoch: [7] [250/488] eta: 0:04:59 lr: 0.000004 loss: 0.6429 (0.6347) time: 1.3138 data: 0.0034 max mem: 28828 +Epoch: [7] [260/488] eta: 0:04:46 lr: 0.000004 loss: 0.6387 (0.6345) time: 1.1545 data: 0.0007 max mem: 28828 +Epoch: [7] [270/488] eta: 0:04:34 lr: 0.000004 loss: 0.6333 (0.6346) time: 1.2451 data: 0.0007 max mem: 28828 +Epoch: [7] [280/488] eta: 0:04:20 lr: 0.000004 loss: 0.6340 (0.6345) time: 1.2353 data: 0.0007 max mem: 28828 +Epoch: [7] [290/488] eta: 0:04:08 lr: 0.000004 loss: 0.6345 (0.6344) time: 1.1851 data: 0.0007 max mem: 28828 +Epoch: [7] [300/488] eta: 0:03:54 lr: 0.000004 loss: 0.6371 (0.6347) time: 1.1094 data: 0.0007 max mem: 28828 +Epoch: [7] [310/488] eta: 0:03:43 lr: 0.000004 loss: 0.6371 (0.6347) time: 1.3006 data: 0.0007 max mem: 28828 +Epoch: [7] [320/488] eta: 0:03:31 lr: 0.000004 loss: 0.6387 (0.6347) time: 1.4764 data: 0.0007 max mem: 28828 +Epoch: [7] [330/488] eta: 0:03:19 lr: 0.000004 loss: 0.6442 (0.6352) time: 1.4098 data: 0.0006 max mem: 28828 +Epoch: [7] [340/488] eta: 0:03:05 lr: 0.000004 loss: 0.6329 (0.6350) time: 1.1484 data: 0.0006 max mem: 28828 +Epoch: [7] [350/488] eta: 0:02:53 lr: 0.000004 loss: 0.6317 (0.6351) time: 1.1391 data: 0.0007 max mem: 28828 +Epoch: [7] [360/488] eta: 0:02:40 lr: 0.000004 loss: 0.6338 (0.6350) time: 1.2446 data: 0.0007 max mem: 28828 +Epoch: [7] [370/488] eta: 0:02:28 lr: 0.000004 loss: 0.6385 (0.6350) time: 1.2106 data: 0.0008 max mem: 28828 +Epoch: [7] [380/488] eta: 0:02:16 lr: 0.000004 loss: 0.6395 (0.6354) time: 1.3967 data: 0.0007 max mem: 28828 +Epoch: [7] [390/488] eta: 0:02:02 lr: 0.000004 loss: 0.6335 (0.6351) time: 1.1942 data: 0.0006 max mem: 28828 +Epoch: [7] [400/488] eta: 0:01:50 lr: 0.000004 loss: 0.6335 (0.6350) time: 1.2737 data: 0.0006 max mem: 28828 +Epoch: [7] [410/488] eta: 0:01:37 lr: 0.000004 loss: 0.6443 (0.6352) time: 1.2606 data: 0.0006 max mem: 28828 +Epoch: [7] [420/488] eta: 0:01:25 lr: 0.000004 loss: 0.6443 (0.6351) time: 1.0747 data: 0.0007 max mem: 28828 +Epoch: [7] [430/488] eta: 0:01:12 lr: 0.000004 loss: 0.6289 (0.6348) time: 1.2439 data: 0.0007 max mem: 28828 +Epoch: [7] [440/488] eta: 0:01:00 lr: 0.000004 loss: 0.6289 (0.6348) time: 1.2794 data: 0.0007 max mem: 28828 +Epoch: [7] [450/488] eta: 0:00:47 lr: 0.000004 loss: 0.6269 (0.6343) time: 1.1009 data: 0.0007 max mem: 28828 +Epoch: [7] [460/488] eta: 0:00:34 lr: 0.000004 loss: 0.6308 (0.6344) time: 1.1199 data: 0.0006 max mem: 28828 +Epoch: [7] [470/488] eta: 0:00:22 lr: 0.000004 loss: 0.6326 (0.6344) time: 1.2194 data: 0.0049 max mem: 28828 +Epoch: [7] [480/488] eta: 0:00:09 lr: 0.000004 loss: 0.6278 (0.6342) time: 0.8129 data: 0.0047 max mem: 28828 +Epoch: [7] [487/488] eta: 0:00:01 lr: 0.000004 loss: 0.6248 (0.6339) time: 0.8300 data: 0.0045 max mem: 28828 +Epoch: [7] Total time: 0:09:56 (1.2229 s / it) +Averaged stats: lr: 0.000004 loss: 0.6248 (0.6349) +Epoch: [8] [ 0/488] eta: 2:11:14 lr: 0.000003 loss: 0.6121 (0.6121) time: 16.1359 data: 15.9184 max mem: 28828 +Epoch: [8] [ 10/488] eta: 0:16:40 lr: 0.000003 loss: 0.6316 (0.6341) time: 2.0937 data: 1.7092 max mem: 28828 +Epoch: [8] [ 20/488] eta: 0:13:20 lr: 0.000003 loss: 0.6319 (0.6317) time: 0.9882 data: 0.5857 max mem: 28828 +Epoch: [8] [ 30/488] eta: 0:11:40 lr: 0.000003 loss: 0.6319 (0.6294) time: 1.2182 data: 0.8168 max mem: 28828 +Epoch: [8] [ 40/488] eta: 0:11:26 lr: 0.000003 loss: 0.6273 (0.6297) time: 1.3452 data: 0.9419 max mem: 28828 +Epoch: [8] [ 50/488] eta: 0:10:26 lr: 0.000003 loss: 0.6284 (0.6298) time: 1.2749 data: 0.8635 max mem: 28828 +Epoch: [8] [ 60/488] eta: 0:09:54 lr: 0.000003 loss: 0.6393 (0.6318) time: 1.0944 data: 0.6617 max mem: 28828 +Epoch: [8] [ 70/488] eta: 0:09:35 lr: 0.000003 loss: 0.6444 (0.6332) time: 1.2457 data: 0.8229 max mem: 28828 +Epoch: [8] [ 80/488] eta: 0:09:04 lr: 0.000003 loss: 0.6444 (0.6338) time: 1.1662 data: 0.6770 max mem: 28828 +Epoch: [8] [ 90/488] eta: 0:08:33 lr: 0.000003 loss: 0.6335 (0.6344) time: 0.9790 data: 0.4280 max mem: 28828 +Epoch: [8] [100/488] eta: 0:08:19 lr: 0.000003 loss: 0.6309 (0.6338) time: 1.0999 data: 0.4501 max mem: 28828 +Epoch: [8] [110/488] eta: 0:08:02 lr: 0.000003 loss: 0.6309 (0.6324) time: 1.2106 data: 0.5457 max mem: 28828 +Epoch: [8] [120/488] eta: 0:07:49 lr: 0.000003 loss: 0.6346 (0.6337) time: 1.2240 data: 0.6031 max mem: 28828 +Epoch: [8] [130/488] eta: 0:07:37 lr: 0.000003 loss: 0.6419 (0.6333) time: 1.2847 data: 0.4724 max mem: 28828 +Epoch: [8] [140/488] eta: 0:07:22 lr: 0.000003 loss: 0.6419 (0.6347) time: 1.2453 data: 0.2047 max mem: 28828 +Epoch: [8] [150/488] eta: 0:07:01 lr: 0.000003 loss: 0.6508 (0.6357) time: 1.0553 data: 0.0313 max mem: 28828 +Epoch: [8] [160/488] eta: 0:06:47 lr: 0.000003 loss: 0.6400 (0.6352) time: 1.0198 data: 0.1205 max mem: 28828 +Epoch: [8] [170/488] eta: 0:06:23 lr: 0.000003 loss: 0.6221 (0.6342) time: 0.8902 data: 0.1261 max mem: 28828 +Epoch: [8] [180/488] eta: 0:06:00 lr: 0.000003 loss: 0.6146 (0.6328) time: 0.5896 data: 0.0061 max mem: 28828 +Epoch: [8] [190/488] eta: 0:05:39 lr: 0.000003 loss: 0.6201 (0.6328) time: 0.5767 data: 0.0204 max mem: 28828 +Epoch: [8] [200/488] eta: 0:05:17 lr: 0.000003 loss: 0.6162 (0.6317) time: 0.5021 data: 0.0204 max mem: 28828 +Epoch: [8] [210/488] eta: 0:04:58 lr: 0.000003 loss: 0.6237 (0.6325) time: 0.4509 data: 0.0005 max mem: 28828 +Epoch: [8] [220/488] eta: 0:04:39 lr: 0.000003 loss: 0.6412 (0.6324) time: 0.4255 data: 0.0005 max mem: 28828 +Epoch: [8] [230/488] eta: 0:04:22 lr: 0.000003 loss: 0.6243 (0.6319) time: 0.4031 data: 0.0005 max mem: 28828 +Epoch: [8] [240/488] eta: 0:04:05 lr: 0.000003 loss: 0.6291 (0.6320) time: 0.4393 data: 0.0004 max mem: 28828 +Epoch: [8] [250/488] eta: 0:03:51 lr: 0.000003 loss: 0.6350 (0.6325) time: 0.4492 data: 0.0005 max mem: 28828 +Epoch: [8] [260/488] eta: 0:03:37 lr: 0.000003 loss: 0.6442 (0.6322) time: 0.4797 data: 0.0015 max mem: 28828 +Epoch: [8] [270/488] eta: 0:03:23 lr: 0.000003 loss: 0.6266 (0.6327) time: 0.4290 data: 0.0014 max mem: 28828 +Epoch: [8] [280/488] eta: 0:03:09 lr: 0.000003 loss: 0.6405 (0.6330) time: 0.3881 data: 0.0005 max mem: 28828 +Epoch: [8] [290/488] eta: 0:02:57 lr: 0.000003 loss: 0.6259 (0.6325) time: 0.4294 data: 0.0005 max mem: 28828 +Epoch: [8] [300/488] eta: 0:02:46 lr: 0.000003 loss: 0.6259 (0.6325) time: 0.4923 data: 0.0247 max mem: 28828 +Epoch: [8] [310/488] eta: 0:02:34 lr: 0.000003 loss: 0.6348 (0.6327) time: 0.4618 data: 0.0302 max mem: 28828 +Epoch: [8] [320/488] eta: 0:02:24 lr: 0.000003 loss: 0.6362 (0.6329) time: 0.4738 data: 0.0771 max mem: 28828 +Epoch: [8] [330/488] eta: 0:02:13 lr: 0.000003 loss: 0.6362 (0.6324) time: 0.4508 data: 0.0716 max mem: 28828 +Epoch: [8] [340/488] eta: 0:02:03 lr: 0.000003 loss: 0.6335 (0.6324) time: 0.4246 data: 0.0004 max mem: 28828 +Epoch: [8] [350/488] eta: 0:01:53 lr: 0.000003 loss: 0.6410 (0.6326) time: 0.4584 data: 0.0004 max mem: 28828 +Epoch: [8] [360/488] eta: 0:01:43 lr: 0.000003 loss: 0.6395 (0.6324) time: 0.4562 data: 0.0005 max mem: 28828 +Epoch: [8] [370/488] eta: 0:01:35 lr: 0.000003 loss: 0.6364 (0.6324) time: 0.5552 data: 0.0005 max mem: 28828 +Epoch: [8] [380/488] eta: 0:01:26 lr: 0.000003 loss: 0.6364 (0.6323) time: 0.5640 data: 0.0005 max mem: 28828 +Epoch: [8] [390/488] eta: 0:01:17 lr: 0.000003 loss: 0.6344 (0.6323) time: 0.5040 data: 0.0005 max mem: 28828 +Epoch: [8] [400/488] eta: 0:01:10 lr: 0.000003 loss: 0.6344 (0.6321) time: 0.7740 data: 0.0006 max mem: 28828 +Epoch: [8] [410/488] eta: 0:01:02 lr: 0.000003 loss: 0.6363 (0.6321) time: 1.0904 data: 0.0006 max mem: 28828 +Epoch: [8] [420/488] eta: 0:00:55 lr: 0.000003 loss: 0.6415 (0.6325) time: 1.1397 data: 0.0006 max mem: 28828 +Epoch: [8] [430/488] eta: 0:00:47 lr: 0.000003 loss: 0.6377 (0.6323) time: 1.0840 data: 0.0006 max mem: 28828 +Epoch: [8] [440/488] eta: 0:00:40 lr: 0.000003 loss: 0.6170 (0.6319) time: 1.2652 data: 0.0006 max mem: 28828 +Epoch: [8] [450/488] eta: 0:00:32 lr: 0.000003 loss: 0.6404 (0.6322) time: 1.3904 data: 0.0007 max mem: 28828 +Epoch: [8] [460/488] eta: 0:00:23 lr: 0.000003 loss: 0.6404 (0.6320) time: 1.2799 data: 0.0008 max mem: 28828 +Epoch: [8] [470/488] eta: 0:00:15 lr: 0.000003 loss: 0.6344 (0.6321) time: 1.2447 data: 0.0012 max mem: 28828 +Epoch: [8] [480/488] eta: 0:00:06 lr: 0.000003 loss: 0.6344 (0.6321) time: 1.1801 data: 0.0009 max mem: 28828 +Epoch: [8] [487/488] eta: 0:00:00 lr: 0.000003 loss: 0.6276 (0.6322) time: 0.8215 data: 0.0008 max mem: 28828 +Epoch: [8] Total time: 0:07:02 (0.8666 s / it) +Averaged stats: lr: 0.000003 loss: 0.6276 (0.6330) +Epoch: [9] [ 0/488] eta: 2:40:22 lr: 0.000002 loss: 0.6549 (0.6549) time: 19.7191 data: 19.3517 max mem: 28828 +Epoch: [9] [ 10/488] eta: 0:19:16 lr: 0.000002 loss: 0.6486 (0.6448) time: 2.4200 data: 1.9853 max mem: 28828 +Epoch: [9] [ 20/488] eta: 0:13:42 lr: 0.000002 loss: 0.6319 (0.6405) time: 0.8599 data: 0.3942 max mem: 28828 +Epoch: [9] [ 30/488] eta: 0:12:14 lr: 0.000002 loss: 0.6330 (0.6379) time: 1.1549 data: 0.6853 max mem: 28828 +Epoch: [9] [ 40/488] eta: 0:11:12 lr: 0.000002 loss: 0.6314 (0.6327) time: 1.2296 data: 0.7613 max mem: 28828 +Epoch: [9] [ 50/488] eta: 0:10:28 lr: 0.000002 loss: 0.6292 (0.6333) time: 1.1745 data: 0.7352 max mem: 28828 +Epoch: [9] [ 60/488] eta: 0:10:23 lr: 0.000002 loss: 0.6353 (0.6340) time: 1.3710 data: 0.9596 max mem: 28828 +Epoch: [9] [ 70/488] eta: 0:09:17 lr: 0.000002 loss: 0.6353 (0.6312) time: 1.0736 data: 0.7366 max mem: 28828 +Epoch: [9] [ 80/488] eta: 0:08:33 lr: 0.000002 loss: 0.6499 (0.6339) time: 0.6504 data: 0.2587 max mem: 28828 +Epoch: [9] [ 90/488] eta: 0:07:35 lr: 0.000002 loss: 0.6525 (0.6346) time: 0.4701 data: 0.0926 max mem: 28828 +Epoch: [9] [100/488] eta: 0:06:52 lr: 0.000002 loss: 0.6268 (0.6323) time: 0.2660 data: 0.0005 max mem: 28828 +Epoch: [9] [110/488] eta: 0:06:27 lr: 0.000002 loss: 0.6268 (0.6330) time: 0.4888 data: 0.0005 max mem: 28828 +Epoch: [9] [120/488] eta: 0:05:56 lr: 0.000002 loss: 0.6440 (0.6330) time: 0.5055 data: 0.0004 max mem: 28828 +Epoch: [9] [130/488] eta: 0:05:39 lr: 0.000002 loss: 0.6397 (0.6328) time: 0.5153 data: 0.0005 max mem: 28828 +Epoch: [9] [140/488] eta: 0:05:16 lr: 0.000002 loss: 0.6375 (0.6327) time: 0.5455 data: 0.0005 max mem: 28828 +Epoch: [9] [150/488] eta: 0:04:55 lr: 0.000002 loss: 0.6369 (0.6321) time: 0.3997 data: 0.0004 max mem: 28828 +Epoch: [9] [160/488] eta: 0:04:39 lr: 0.000002 loss: 0.6368 (0.6319) time: 0.4350 data: 0.0005 max mem: 28828 +Epoch: [9] [170/488] eta: 0:04:21 lr: 0.000002 loss: 0.6368 (0.6324) time: 0.4285 data: 0.0005 max mem: 28828 +Epoch: [9] [180/488] eta: 0:04:07 lr: 0.000002 loss: 0.6358 (0.6324) time: 0.4191 data: 0.0005 max mem: 28828 +Epoch: [9] [190/488] eta: 0:03:53 lr: 0.000002 loss: 0.6316 (0.6324) time: 0.4521 data: 0.0594 max mem: 28828 +Epoch: [9] [200/488] eta: 0:03:41 lr: 0.000002 loss: 0.6345 (0.6332) time: 0.4657 data: 0.0603 max mem: 28828 +Epoch: [9] [210/488] eta: 0:03:29 lr: 0.000002 loss: 0.6355 (0.6332) time: 0.4794 data: 0.0144 max mem: 28828 +Epoch: [9] [220/488] eta: 0:03:18 lr: 0.000002 loss: 0.6239 (0.6328) time: 0.4539 data: 0.0135 max mem: 28828 +Epoch: [9] [230/488] eta: 0:03:08 lr: 0.000002 loss: 0.6216 (0.6327) time: 0.4763 data: 0.0005 max mem: 28828 +Epoch: [9] [240/488] eta: 0:02:56 lr: 0.000002 loss: 0.6341 (0.6323) time: 0.3940 data: 0.0005 max mem: 28828 +Epoch: [9] [250/488] eta: 0:02:47 lr: 0.000002 loss: 0.6349 (0.6324) time: 0.3885 data: 0.0004 max mem: 28828 +Epoch: [9] [260/488] eta: 0:02:38 lr: 0.000002 loss: 0.6349 (0.6323) time: 0.4979 data: 0.0004 max mem: 28828 +Epoch: [9] [270/488] eta: 0:02:29 lr: 0.000002 loss: 0.6438 (0.6329) time: 0.4430 data: 0.0005 max mem: 28828 +Epoch: [9] [280/488] eta: 0:02:20 lr: 0.000002 loss: 0.6359 (0.6326) time: 0.3979 data: 0.0005 max mem: 28828 +Epoch: [9] [290/488] eta: 0:02:12 lr: 0.000002 loss: 0.6392 (0.6328) time: 0.4469 data: 0.0005 max mem: 28828 +Epoch: [9] [300/488] eta: 0:02:04 lr: 0.000002 loss: 0.6448 (0.6332) time: 0.4539 data: 0.0005 max mem: 28828 +Epoch: [9] [310/488] eta: 0:01:56 lr: 0.000002 loss: 0.6387 (0.6334) time: 0.4396 data: 0.0696 max mem: 28828 +Epoch: [9] [320/488] eta: 0:01:48 lr: 0.000002 loss: 0.6365 (0.6335) time: 0.4260 data: 0.0776 max mem: 28828 +Epoch: [9] [330/488] eta: 0:01:41 lr: 0.000002 loss: 0.6404 (0.6334) time: 0.4702 data: 0.0232 max mem: 28828 +Epoch: [9] [340/488] eta: 0:01:34 lr: 0.000002 loss: 0.6330 (0.6329) time: 0.4806 data: 0.0497 max mem: 28828 +Epoch: [9] [350/488] eta: 0:01:27 lr: 0.000002 loss: 0.6317 (0.6330) time: 0.4397 data: 0.1597 max mem: 28828 +Epoch: [9] [360/488] eta: 0:01:20 lr: 0.000002 loss: 0.6310 (0.6326) time: 0.4995 data: 0.2762 max mem: 28828 +Epoch: [9] [370/488] eta: 0:01:13 lr: 0.000002 loss: 0.6193 (0.6325) time: 0.5171 data: 0.2421 max mem: 28828 +Epoch: [9] [380/488] eta: 0:01:06 lr: 0.000002 loss: 0.6257 (0.6322) time: 0.4695 data: 0.0911 max mem: 28828 +Epoch: [9] [390/488] eta: 0:01:00 lr: 0.000002 loss: 0.6321 (0.6322) time: 0.4705 data: 0.0004 max mem: 28828 +Epoch: [9] [400/488] eta: 0:00:53 lr: 0.000002 loss: 0.6385 (0.6322) time: 0.4251 data: 0.0004 max mem: 28828 +Epoch: [9] [410/488] eta: 0:00:47 lr: 0.000002 loss: 0.6415 (0.6325) time: 0.4076 data: 0.0036 max mem: 28828 +Epoch: [9] [420/488] eta: 0:00:41 lr: 0.000002 loss: 0.6415 (0.6326) time: 0.4701 data: 0.0036 max mem: 28828 +Epoch: [9] [430/488] eta: 0:00:34 lr: 0.000002 loss: 0.6334 (0.6323) time: 0.4302 data: 0.0169 max mem: 28828 +Epoch: [9] [440/488] eta: 0:00:28 lr: 0.000002 loss: 0.6296 (0.6323) time: 0.4054 data: 0.0853 max mem: 28828 +Epoch: [9] [450/488] eta: 0:00:22 lr: 0.000002 loss: 0.6339 (0.6324) time: 0.4200 data: 0.1663 max mem: 28828 +Epoch: [9] [460/488] eta: 0:00:16 lr: 0.000002 loss: 0.6429 (0.6324) time: 0.4378 data: 0.1888 max mem: 28828 +Epoch: [9] [470/488] eta: 0:00:10 lr: 0.000002 loss: 0.6318 (0.6321) time: 0.4196 data: 0.0918 max mem: 28828 +Epoch: [9] [480/488] eta: 0:00:04 lr: 0.000002 loss: 0.6272 (0.6318) time: 0.3837 data: 0.0007 max mem: 28828 +Epoch: [9] [487/488] eta: 0:00:00 lr: 0.000002 loss: 0.6272 (0.6317) time: 0.3078 data: 0.0006 max mem: 28828 +Epoch: [9] Total time: 0:04:41 (0.5760 s / it) +Averaged stats: lr: 0.000002 loss: 0.6272 (0.6320) +Test: [ 0/234] eta: 0:13:44 acc1_pos: 67.969 acc5_pos: 100.000 loss: 0.6078 (0.6078) acc1: 67.9688 (67.9688) acc5: 100.0000 (100.0000) time: 3.5238 data: 3.4328 max mem: 28828 +Test: [ 10/234] eta: 0:02:06 acc1_pos: 65.365 acc5_pos: 100.000 loss: 0.6591 (0.6566) acc1: 65.3646 (61.6241) acc5: 100.0000 (100.0000) time: 0.5628 data: 0.4702 max mem: 28828 +Test: [ 20/234] eta: 0:01:26 acc1_pos: 96.354 acc5_pos: 100.000 loss: 0.6677 (0.6455) acc1: 55.7292 (60.6275) acc5: 100.0000 (100.0000) time: 0.2483 data: 0.1543 max mem: 28828 +Test: [ 30/234] eta: 0:01:07 acc1_pos: 84.635 acc5_pos: 100.000 loss: 0.2048 (0.5084) acc1: 95.5729 (71.7154) acc5: 100.0000 (100.0000) time: 0.2051 data: 0.1117 max mem: 28828 +Test: [ 40/234] eta: 0:01:00 acc1_pos: 99.219 acc5_pos: 100.000 loss: 0.2021 (0.4656) acc1: 95.8333 (75.6733) acc5: 100.0000 (100.0000) time: 0.2099 data: 0.1189 max mem: 28828 +Test: [ 50/234] eta: 0:00:53 acc1_pos: 99.219 acc5_pos: 100.000 loss: 0.1599 (0.4043) acc1: 97.9167 (80.1113) acc5: 100.0000 (100.0000) time: 0.2258 data: 0.1322 max mem: 28828 +Test: [ 60/234] eta: 0:00:46 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.1057 (0.3494) acc1: 99.2188 (83.3461) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0904 max mem: 28828 +Test: [ 70/234] eta: 0:00:41 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.0680 (0.3665) acc1: 100.0000 (82.1523) acc5: 100.0000 (100.0000) acc1_neg: 59.896 acc5_neg: 100.000 time: 0.1595 data: 0.0678 max mem: 28828 +Test: [ 80/234] eta: 0:00:38 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.6405 (0.3822) acc1: 63.2812 (81.3111) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1999 data: 0.1072 max mem: 28828 +Test: [ 90/234] eta: 0:00:35 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.2154 (0.3605) acc1: 92.7083 (82.8154) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.2343 data: 0.1416 max mem: 28828 +Test: [100/234] eta: 0:00:32 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.2233 (0.3662) acc1: 88.8021 (82.4928) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2065 data: 0.1131 max mem: 28828 +Test: [110/234] eta: 0:00:29 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.4363 (0.3692) acc1: 77.8646 (82.5052) acc5: 100.0000 (100.0000) acc1_neg: 72.135 acc5_neg: 100.000 time: 0.1854 data: 0.0914 max mem: 28828 +Test: [120/234] eta: 0:00:26 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.4326 (0.3790) acc1: 79.1667 (82.1259) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1992 data: 0.1057 max mem: 28828 +Test: [130/234] eta: 0:00:24 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.4317 (0.3824) acc1: 83.0729 (82.3851) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.2094 data: 0.1132 max mem: 28828 +Test: [140/234] eta: 0:00:21 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.4425 (0.4093) acc1: 81.7708 (81.0136) acc5: 100.0000 (100.0000) acc1_neg: 35.417 acc5_neg: 100.000 time: 0.1741 data: 0.0759 max mem: 28828 +Test: [150/234] eta: 0:00:18 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.4493 (0.4170) acc1: 76.0417 (80.5722) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.1556 data: 0.0589 max mem: 28828 +Test: [160/234] eta: 0:00:16 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.4108 (0.4161) acc1: 80.4688 (80.8521) acc5: 100.0000 (100.0000) acc1_neg: 77.344 acc5_neg: 100.000 time: 0.2048 data: 0.1109 max mem: 28828 +Test: [170/234] eta: 0:00:14 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.4775 (0.4199) acc1: 79.6875 (80.7429) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.2423 data: 0.1502 max mem: 28828 +Test: [180/234] eta: 0:00:12 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.4899 (0.4300) acc1: 76.8229 (80.0731) acc5: 100.0000 (100.0000) acc1_neg: 38.802 acc5_neg: 100.000 time: 0.2227 data: 0.1322 max mem: 28828 +Test: [190/234] eta: 0:00:09 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.7002 (0.4456) acc1: 60.9375 (78.9676) acc5: 100.0000 (100.0000) acc1_neg: 59.635 acc5_neg: 100.000 time: 0.1447 data: 0.0550 max mem: 28828 +Test: [200/234] eta: 0:00:07 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.7002 (0.4396) acc1: 61.1979 (79.2638) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.0923 data: 0.0017 max mem: 28828 +Test: [210/234] eta: 0:00:04 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.1001 (0.4241) acc1: 100.0000 (80.2231) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1206 data: 0.0298 max mem: 28828 +Test: [220/234] eta: 0:00:02 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.1370 (0.4123) acc1: 99.2188 (80.9837) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.1317 data: 0.0416 max mem: 28828 +Test: [230/234] eta: 0:00:00 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.3535 (0.4282) acc1: 52.3438 (79.6041) acc5: 100.0000 (100.0000) acc1_neg: 51.302 acc5_neg: 100.000 time: 0.1104 data: 0.0204 max mem: 28828 +Test: [233/234] eta: 0:00:00 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.7540 (0.4325) acc1: 51.3021 (79.2509) acc5: 100.0000 (100.0000) acc1_neg: 51.362 acc5_neg: 100.000 time: 0.1087 data: 0.0204 max mem: 28828 +Test: Total time: 0:00:46 (0.1985 s / it) +* Overall Acc@1 79.251 Acc@5 100.000 + target == real Acc@1 80.232 Acc@5 100.000 + target == fake Acc@1 78.974 Acc@5 100.000 + loss 0.433 +Accuracy of the network on the 89729 test images: 79.3% +n116-042-186:1209:59061 [3] NCCL INFO misc/socket.cc:64 -> 3 +n116-042-186:1209:59061 [3] NCCL INFO misc/socket.cc:80 -> 3 +n116-042-186:1209:59061 [3] NCCL INFO misc/socket.cc:829 -> 3 +n116-042-186:1209:59061 [3] NCCL INFO misc/socket.cc:64 -> 3 +n116-042-186:1209:59061 [3] NCCL INFO misc/socket.cc:80 -> 3 +n116-042-186:1209:59061 [3] NCCL INFO misc/socket.cc:829 -> 3 +n116-042-186:1209:59061 [3] NCCL INFO misc/socket.cc:64 -> 3 +n116-042-186:1209:59061 [3] NCCL INFO misc/socket.cc:80 -> 3 +n116-042-186:1209:59061 [3] NCCL INFO misc/socket.cc:829 -> 3 +n116-042-186:1209:1317 [3] NCCL INFO misc/socket.cc:881 -> 3 +n116-042-186:1212:59063 [6] NCCL INFO misc/socket.cc:64 -> 3 +n116-042-186:1212:59063 [6] NCCL INFO misc/socket.cc:80 -> 3 +n116-042-186:1212:59063 [6] NCCL INFO misc/socket.cc:829 -> 3 +n116-042-186:1212:59063 [6] NCCL INFO misc/socket.cc:64 -> 3 +n116-042-186:1212:59063 [6] NCCL INFO misc/socket.cc:80 -> 3 +n116-042-186:1212:59063 [6] NCCL INFO misc/socket.cc:829 -> 3 +n116-042-186:1212:59063 [6] NCCL INFO misc/socket.cc:64 -> 3 +n116-042-186:1212:59063 [6] NCCL INFO misc/socket.cc:80 -> 3 +n116-042-186:1212:59063 [6] NCCL INFO misc/socket.cc:829 -> 3 +n116-042-186:1212:1307 [6] NCCL INFO misc/socket.cc:881 -> 3 +n116-042-186:1210:59065 [4] NCCL INFO misc/socket.cc:64 -> 3 +n116-042-186:1210:59065 [4] NCCL INFO misc/socket.cc:80 -> 3 +n116-042-186:1210:59065 [4] NCCL INFO misc/socket.cc:829 -> 3 +n116-042-186:1209:1317 [3] NCCL INFO misc/socket.cc:881 -> 3 +n116-042-186:1210:59065 [4] NCCL INFO misc/socket.cc:64 -> 3 +n116-042-186:1210:59065 [4] NCCL INFO misc/socket.cc:80 -> 3 +n116-042-186:1210:59065 [4] NCCL INFO misc/socket.cc:829 -> 3 +n116-042-186:1210:59065 [4] NCCL INFO misc/socket.cc:64 -> 3 +n116-042-186:1210:59065 [4] NCCL INFO misc/socket.cc:80 -> 3 +n116-042-186:1210:59065 [4] NCCL INFO misc/socket.cc:829 -> 3 +n116-042-186:1210:1315 [4] NCCL INFO misc/socket.cc:881 -> 3 +n116-042-186:1211:59067 [5] NCCL INFO misc/socket.cc:64 -> 3 +n116-042-186:1210:1315 [4] NCCL INFO misc/socket.cc:881 -> 3 +n116-042-186:1211:59067 [5] NCCL INFO misc/socket.cc:80 -> 3 +n116-042-186:1211:59067 [5] NCCL INFO misc/socket.cc:829 -> 3 +n116-042-186:1211:59067 [5] NCCL INFO misc/socket.cc:64 -> 3 +n116-042-186:1211:59067 [5] NCCL INFO misc/socket.cc:80 -> 3 +n116-042-186:1211:59067 [5] NCCL INFO misc/socket.cc:829 -> 3 +n116-042-186:1211:59067 [5] NCCL INFO misc/socket.cc:64 -> 3 +n116-042-186:1211:59067 [5] NCCL INFO misc/socket.cc:80 -> 3 +n116-042-186:1211:1303 [5] NCCL INFO misc/socket.cc:881 -> 3 +n116-042-186:1212:1307 [6] NCCL INFO misc/socket.cc:881 -> 3 +n116-042-186:1211:59067 [5] NCCL INFO misc/socket.cc:829 -> 3 +n116-042-186:1208:59069 [2] NCCL INFO misc/socket.cc:64 -> 3 +n116-042-186:1208:59069 [2] NCCL INFO misc/socket.cc:80 -> 3 +n116-042-186:1208:59069 [2] NCCL INFO misc/socket.cc:829 -> 3 +n116-042-186:1208:59069 [2] NCCL INFO misc/socket.cc:64 -> 3 +n116-042-186:1208:59069 [2] NCCL INFO misc/socket.cc:80 -> 3 +n116-042-186:1208:59069 [2] NCCL INFO misc/socket.cc:829 -> 3 +n116-042-186:1208:1311 [2] NCCL INFO misc/socket.cc:881 -> 3 +n116-042-186:1208:59069 [2] NCCL INFO misc/socket.cc:64 -> 3 +n116-042-186:1208:59069 [2] NCCL INFO misc/socket.cc:80 -> 3 +n116-042-186:1208:59069 [2] NCCL INFO misc/socket.cc:829 -> 3 +n116-042-186:1213:59071 [7] NCCL INFO misc/socket.cc:64 -> 3 +n116-042-186:1213:59071 [7] NCCL INFO misc/socket.cc:80 -> 3 +n116-042-186:1213:59071 [7] NCCL INFO misc/socket.cc:829 -> 3 +n116-042-186:1213:59071 [7] NCCL INFO misc/socket.cc:64 -> 3 +n116-042-186:1213:59071 [7] NCCL INFO misc/socket.cc:80 -> 3 +n116-042-186:1213:59071 [7] NCCL INFO misc/socket.cc:829 -> 3 +n116-042-186:1213:59071 [7] NCCL INFO misc/socket.cc:64 -> 3 +n116-042-186:1213:59071 [7] NCCL INFO misc/socket.cc:80 -> 3 +n116-042-186:1213:59071 [7] NCCL INFO misc/socket.cc:829 -> 3 +n116-042-186:1213:1313 [7] NCCL INFO misc/socket.cc:881 -> 3 +n116-042-186:1209:59061 [3] NCCL INFO comm 0xd05c320 rank 3 nranks 8 cudaDev 3 busId c000 - Abort COMPLETE +n116-042-186:1207:59073 [1] NCCL INFO misc/socket.cc:64 -> 3 +n116-042-186:1207:59073 [1] NCCL INFO misc/socket.cc:80 -> 3 +n116-042-186:1207:59073 [1] NCCL INFO misc/socket.cc:829 -> 3 +n116-042-186:1207:59073 [1] NCCL INFO misc/socket.cc:64 -> 3 +n116-042-186:1207:59073 [1] NCCL INFO misc/socket.cc:80 -> 3 +n116-042-186:1207:59073 [1] NCCL INFO misc/socket.cc:829 -> 3 +n116-042-186:1207:59073 [1] NCCL INFO misc/socket.cc:64 -> 3 +n116-042-186:1207:59073 [1] NCCL INFO misc/socket.cc:80 -> 3 +n116-042-186:1207:59073 [1] NCCL INFO misc/socket.cc:829 -> 3 +n116-042-186:1207:1309 [1] NCCL INFO misc/socket.cc:881 -> 3 +n116-042-186:1208:1311 [2] NCCL INFO misc/socket.cc:881 -> 3 +n116-042-186:1208:59069 [2] NCCL INFO comm 0xb4d5d00 rank 2 nranks 8 cudaDev 2 busId b000 - Abort COMPLETE +n116-042-186:1212:59063 [6] NCCL INFO comm 0xb67df70 rank 6 nranks 8 cudaDev 6 busId 8b000 - Abort COMPLETE +n116-042-186:1210:59065 [4] NCCL INFO comm 0xb3b2a80 rank 4 nranks 8 cudaDev 4 busId 84000 - Abort COMPLETE +n116-042-186:1211:59067 [5] NCCL INFO comm 0xbc38ba0 rank 5 nranks 8 cudaDev 5 busId 85000 - Abort COMPLETE +n116-042-186:1213:59071 [7] NCCL INFO comm 0xbf6b490 rank 7 nranks 8 cudaDev 7 busId 8c000 - Abort COMPLETE +n116-042-186:1207:59073 [1] NCCL INFO comm 0xb5957c0 rank 1 nranks 8 cudaDev 1 busId 5000 - Abort COMPLETE +Max accuracy: 79.25% +Training time 0:57:36 +[rank0]:[W725 13:47:37.568514940 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +n116-042-186:1206:59075 [0] NCCL INFO misc/socket.cc:64 -> 3 +n116-042-186:1206:59075 [0] NCCL INFO misc/socket.cc:80 -> 3 +n116-042-186:1206:59075 [0] NCCL INFO misc/socket.cc:829 -> 3 +n116-042-186:1206:1305 [0] NCCL INFO misc/socket.cc:881 -> 3 +n116-042-186:1206:59075 [0] NCCL INFO misc/socket.cc:64 -> 3 +n116-042-186:1206:59075 [0] NCCL INFO misc/socket.cc:80 -> 3 +n116-042-186:1206:59075 [0] NCCL INFO misc/socket.cc:829 -> 3 +n116-042-186:1206:59075 [0] NCCL INFO misc/socket.cc:64 -> 3 +n116-042-186:1206:59075 [0] NCCL INFO misc/socket.cc:80 -> 3 +n116-042-186:1206:59075 [0] NCCL INFO misc/socket.cc:829 -> 3 +n116-042-186:1206:59075 [0] NCCL INFO comm 0xbec8fd0 rank 0 nranks 8 cudaDev 0 busId 4000 - Abort COMPLETE diff --git a/ScaleDF_rebuttal/_host_logs/all_46real_91fake_11451_20260725_1215/train_all_46real_91fake_retry1.pid b/ScaleDF_rebuttal/_host_logs/all_46real_91fake_11451_20260725_1215/train_all_46real_91fake_retry1.pid new file mode 100644 index 0000000000000000000000000000000000000000..5e8d8532ec23564e1e5c54012e0e04035eadf383 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/all_46real_91fake_11451_20260725_1215/train_all_46real_91fake_retry1.pid @@ -0,0 +1 @@ +1130 diff --git a/ScaleDF_rebuttal/_host_logs/dualnode_nccl_bw_eth0_256mb_20260725_131627/node0.log b/ScaleDF_rebuttal/_host_logs/dualnode_nccl_bw_eth0_256mb_20260725_131627/node0.log new file mode 100644 index 0000000000000000000000000000000000000000..204b5f70be76b8eeb56d176005b4412ef88f1463 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/dualnode_nccl_bw_eth0_256mb_20260725_131627/node0.log @@ -0,0 +1,24 @@ +/home/tiger/.profile: line 29: /run/bin/env: No such file or directory +/home/tiger/.profile: line 31: /tmp/m9/bin/env: No such file or directory +W0725 13:17:10.480000 41515 torch/distributed/run.py:766] +W0725 13:17:10.480000 41515 torch/distributed/run.py:766] ***************************************** +W0725 13:17:10.480000 41515 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 13:17:10.480000 41515 torch/distributed/run.py:766] ***************************************** +NCCL version 2.26.2+cuda12.2 +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +NCCL_BENCH host=n116-042-186 world=16 mb=256 iters=3 avg_s=0.126694 algbw_gbps=16.95 busbw_gbps=31.78 diff --git a/ScaleDF_rebuttal/_host_logs/dualnode_nccl_bw_eth0_256mb_20260725_131627/node0.pid b/ScaleDF_rebuttal/_host_logs/dualnode_nccl_bw_eth0_256mb_20260725_131627/node0.pid new file mode 100644 index 0000000000000000000000000000000000000000..0d19860238e3c190612a5f702f9a24b7f040cc5b --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/dualnode_nccl_bw_eth0_256mb_20260725_131627/node0.pid @@ -0,0 +1 @@ +41513 diff --git a/ScaleDF_rebuttal/_host_logs/dualnode_nccl_bw_eth0_256mb_20260725_131627/node1.log b/ScaleDF_rebuttal/_host_logs/dualnode_nccl_bw_eth0_256mb_20260725_131627/node1.log new file mode 100644 index 0000000000000000000000000000000000000000..a8942ed383fab4c95a4f7d36d22b7293475ec400 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/dualnode_nccl_bw_eth0_256mb_20260725_131627/node1.log @@ -0,0 +1,22 @@ +/home/tiger/.profile: line 29: /run/bin/env: No such file or directory +/home/tiger/.profile: line 31: /tmp/m9/bin/env: No such file or directory +W0725 13:16:42.678000 13710 torch/distributed/run.py:766] +W0725 13:16:42.678000 13710 torch/distributed/run.py:766] ***************************************** +W0725 13:16:42.678000 13710 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 13:16:42.678000 13710 torch/distributed/run.py:766] ***************************************** +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once diff --git a/ScaleDF_rebuttal/_host_logs/dualnode_nccl_bw_eth0_256mb_20260725_131627/node1.pid b/ScaleDF_rebuttal/_host_logs/dualnode_nccl_bw_eth0_256mb_20260725_131627/node1.pid new file mode 100644 index 0000000000000000000000000000000000000000..126b9d2a54a0fb15fb395774408bd8fc9f392563 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/dualnode_nccl_bw_eth0_256mb_20260725_131627/node1.pid @@ -0,0 +1 @@ +13708 diff --git a/ScaleDF_rebuttal/_host_logs/dualnode_nccl_smoke_20260725_130016/node0_eth0.log b/ScaleDF_rebuttal/_host_logs/dualnode_nccl_smoke_20260725_130016/node0_eth0.log new file mode 100644 index 0000000000000000000000000000000000000000..de5951a35869c4dedf51a32f13ded052c61dba12 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/dualnode_nccl_smoke_20260725_130016/node0_eth0.log @@ -0,0 +1,31 @@ +/home/tiger/.profile: line 29: /run/bin/env: No such file or directory +/home/tiger/.profile: line 31: /tmp/m9/bin/env: No such file or directory +W0725 13:00:33.397000 17804 torch/distributed/run.py:766] +W0725 13:00:33.397000 17804 torch/distributed/run.py:766] ***************************************** +W0725 13:00:33.397000 17804 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 13:00:33.397000 17804 torch/distributed/run.py:766] ***************************************** +start utc=2026-07-25T13:00:50Z host=n116-042-186 rank=7/16 local_rank=7 backend=nccl master=10.116.42.186:34791 +[W725 13:00:50.272920761 Utils.hpp:137] Warning: Environment variable NCCL_ASYNC_ERROR_HANDLING is deprecated; use TORCH_NCCL_ASYNC_ERROR_HANDLING instead (function operator()) +start utc=2026-07-25T13:00:50Z host=n116-042-186 rank=0/16 local_rank=0 backend=nccl master=10.116.42.186:34791 +[W725 13:00:50.936200437 Utils.hpp:137] Warning: Environment variable NCCL_ASYNC_ERROR_HANDLING is deprecated; use TORCH_NCCL_ASYNC_ERROR_HANDLING instead (function operator()) +start utc=2026-07-25T13:00:52Z host=n116-042-186 rank=5/16 local_rank=5 backend=nccl master=10.116.42.186:34791 +[W725 13:00:52.307344035 Utils.hpp:137] Warning: Environment variable NCCL_ASYNC_ERROR_HANDLING is deprecated; use TORCH_NCCL_ASYNC_ERROR_HANDLING instead (function operator()) +start utc=2026-07-25T13:00:52Z host=n116-042-186 rank=6/16 local_rank=6 backend=nccl master=10.116.42.186:34791 +[W725 13:00:52.554344072 Utils.hpp:137] Warning: Environment variable NCCL_ASYNC_ERROR_HANDLING is deprecated; use TORCH_NCCL_ASYNC_ERROR_HANDLING instead (function operator()) +start utc=2026-07-25T13:00:53Z host=n116-042-186 rank=1/16 local_rank=1 backend=nccl master=10.116.42.186:34791 +[W725 13:00:53.719705429 Utils.hpp:137] Warning: Environment variable NCCL_ASYNC_ERROR_HANDLING is deprecated; use TORCH_NCCL_ASYNC_ERROR_HANDLING instead (function operator()) +start utc=2026-07-25T13:00:53Z host=n116-042-186 rank=2/16 local_rank=2 backend=nccl master=10.116.42.186:34791 +[W725 13:00:53.808774120 Utils.hpp:137] Warning: Environment variable NCCL_ASYNC_ERROR_HANDLING is deprecated; use TORCH_NCCL_ASYNC_ERROR_HANDLING instead (function operator()) +start utc=2026-07-25T13:00:53Z host=n116-042-186 rank=4/16 local_rank=4 backend=nccl master=10.116.42.186:34791 +[W725 13:00:53.862178360 Utils.hpp:137] Warning: Environment variable NCCL_ASYNC_ERROR_HANDLING is deprecated; use TORCH_NCCL_ASYNC_ERROR_HANDLING instead (function operator()) +NCCL version 2.26.2+cuda12.2 +start utc=2026-07-25T13:00:54Z host=n116-042-186 rank=3/16 local_rank=3 backend=nccl master=10.116.42.186:34791 +[W725 13:00:54.130994423 Utils.hpp:137] Warning: Environment variable NCCL_ASYNC_ERROR_HANDLING is deprecated; use TORCH_NCCL_ASYNC_ERROR_HANDLING instead (function operator()) +done host=n116-042-186 rank=0 value=136.0 expected=136.0 elapsed=14.846s +done host=n116-042-186 rank=1 value=136.0 expected=136.0 elapsed=12.064s +done host=n116-042-186 rank=2 value=136.0 expected=136.0 elapsed=11.976s +done host=n116-042-186 rank=3 value=136.0 expected=136.0 elapsed=11.653s +done host=n116-042-186 rank=4 value=136.0 expected=136.0 elapsed=11.923s +done host=n116-042-186 rank=5 value=136.0 expected=136.0 elapsed=13.477s +done host=n116-042-186 rank=6 value=136.0 expected=136.0 elapsed=13.230s +done host=n116-042-186 rank=7 value=136.0 expected=136.0 elapsed=15.525s diff --git a/ScaleDF_rebuttal/_host_logs/dualnode_nccl_smoke_20260725_130016/node0_eth0.pid b/ScaleDF_rebuttal/_host_logs/dualnode_nccl_smoke_20260725_130016/node0_eth0.pid new file mode 100644 index 0000000000000000000000000000000000000000..5097429c4ab2afdcd2f9616af9f75ee385a0bb94 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/dualnode_nccl_smoke_20260725_130016/node0_eth0.pid @@ -0,0 +1 @@ +17802 diff --git a/ScaleDF_rebuttal/_host_logs/dualnode_nccl_smoke_20260725_130016/node1_eth0.log b/ScaleDF_rebuttal/_host_logs/dualnode_nccl_smoke_20260725_130016/node1_eth0.log new file mode 100644 index 0000000000000000000000000000000000000000..8a3eb756f40d3a0b6bc7f04d090572aab281bb2e --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/dualnode_nccl_smoke_20260725_130016/node1_eth0.log @@ -0,0 +1,30 @@ +/home/tiger/.profile: line 29: /run/bin/env: No such file or directory +/home/tiger/.profile: line 31: /tmp/m9/bin/env: No such file or directory +W0725 13:00:30.760000 12958 torch/distributed/run.py:766] +W0725 13:00:30.760000 12958 torch/distributed/run.py:766] ***************************************** +W0725 13:00:30.760000 12958 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 13:00:30.760000 12958 torch/distributed/run.py:766] ***************************************** +start utc=2026-07-25T13:00:46Z host=n116-042-185 rank=8/16 local_rank=0 backend=nccl master=10.116.42.186:34791 +[W725 13:00:46.774397589 Utils.hpp:137] Warning: Environment variable NCCL_ASYNC_ERROR_HANDLING is deprecated; use TORCH_NCCL_ASYNC_ERROR_HANDLING instead (function operator()) +start utc=2026-07-25T13:00:47Z host=n116-042-185 rank=11/16 local_rank=3 backend=nccl master=10.116.42.186:34791 +[W725 13:00:47.974587541 Utils.hpp:137] Warning: Environment variable NCCL_ASYNC_ERROR_HANDLING is deprecated; use TORCH_NCCL_ASYNC_ERROR_HANDLING instead (function operator()) +start utc=2026-07-25T13:00:47Z host=n116-042-185 rank=15/16 local_rank=7 backend=nccl master=10.116.42.186:34791 +[W725 13:00:47.065950718 Utils.hpp:137] Warning: Environment variable NCCL_ASYNC_ERROR_HANDLING is deprecated; use TORCH_NCCL_ASYNC_ERROR_HANDLING instead (function operator()) +start utc=2026-07-25T13:00:48Z host=n116-042-185 rank=10/16 local_rank=2 backend=nccl master=10.116.42.186:34791 +[W725 13:00:48.840085634 Utils.hpp:137] Warning: Environment variable NCCL_ASYNC_ERROR_HANDLING is deprecated; use TORCH_NCCL_ASYNC_ERROR_HANDLING instead (function operator()) +start utc=2026-07-25T13:00:48Z host=n116-042-185 rank=14/16 local_rank=6 backend=nccl master=10.116.42.186:34791 +[W725 13:00:48.986676957 Utils.hpp:137] Warning: Environment variable NCCL_ASYNC_ERROR_HANDLING is deprecated; use TORCH_NCCL_ASYNC_ERROR_HANDLING instead (function operator()) +start utc=2026-07-25T13:00:48Z host=n116-042-185 rank=9/16 local_rank=1 backend=nccl master=10.116.42.186:34791 +start utc=2026-07-25T13:00:48Z host=n116-042-185 rank=13/16 local_rank=5 backend=nccl master=10.116.42.186:34791 +[W725 13:00:48.034810780 Utils.hpp:137] Warning: Environment variable NCCL_ASYNC_ERROR_HANDLING is deprecated; use TORCH_NCCL_ASYNC_ERROR_HANDLING instead (function operator()) +start utc=2026-07-25T13:00:48Z host=n116-042-185 rank=12/16 local_rank=4 backend=nccl master=10.116.42.186:34791 +[W725 13:00:48.041503835 Utils.hpp:137] Warning: Environment variable NCCL_ASYNC_ERROR_HANDLING is deprecated; use TORCH_NCCL_ASYNC_ERROR_HANDLING instead (function operator()) +[W725 13:00:48.047580987 Utils.hpp:137] Warning: Environment variable NCCL_ASYNC_ERROR_HANDLING is deprecated; use TORCH_NCCL_ASYNC_ERROR_HANDLING instead (function operator()) +done host=n116-042-185 rank=8 value=136.0 expected=136.0 elapsed=19.143s +done host=n116-042-185 rank=9 value=136.0 expected=136.0 elapsed=16.872s +done host=n116-042-185 rank=10 value=136.0 expected=136.0 elapsed=17.069s +done host=n116-042-185 rank=11 value=136.0 expected=136.0 elapsed=17.937s +done host=n116-042-185 rank=12 value=136.0 expected=136.0 elapsed=16.864sdone host=n116-042-185 rank=13 value=136.0 expected=136.0 elapsed=16.871s + +done host=n116-042-185 rank=14 value=136.0 expected=136.0 elapsed=16.927s +done host=n116-042-185 rank=15 value=136.0 expected=136.0 elapsed=17.851s diff --git a/ScaleDF_rebuttal/_host_logs/dualnode_nccl_smoke_20260725_130016/node1_eth0.pid b/ScaleDF_rebuttal/_host_logs/dualnode_nccl_smoke_20260725_130016/node1_eth0.pid new file mode 100644 index 0000000000000000000000000000000000000000..3b16ae8d70d35575e645b459a8498154229fba74 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/dualnode_nccl_smoke_20260725_130016/node1_eth0.pid @@ -0,0 +1 @@ +12956 diff --git a/ScaleDF_rebuttal/_host_logs/dualnode_nccl_smoke_eth1_20260725_130230/node0_eth1.log b/ScaleDF_rebuttal/_host_logs/dualnode_nccl_smoke_eth1_20260725_130230/node0_eth1.log new file mode 100644 index 0000000000000000000000000000000000000000..23f6b30e0b93a7833887db9ef005f530cca33128 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/dualnode_nccl_smoke_eth1_20260725_130230/node0_eth1.log @@ -0,0 +1,8 @@ +/home/tiger/.profile: line 29: /run/bin/env: No such file or directory +/home/tiger/.profile: line 31: /tmp/m9/bin/env: No such file or directory +W0725 13:02:47.862000 23436 torch/distributed/run.py:766] +W0725 13:02:47.862000 23436 torch/distributed/run.py:766] ***************************************** +W0725 13:02:47.862000 23436 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 13:02:47.862000 23436 torch/distributed/run.py:766] ***************************************** +[W725 13:03:07.943072155 socket.cpp:200] [c10d] The hostname of the client socket cannot be retrieved. err=-3 +[W725 13:03:25.427091715 socket.cpp:200] [c10d] The hostname of the client socket cannot be retrieved. err=-3 diff --git a/ScaleDF_rebuttal/_host_logs/dualnode_nccl_smoke_eth1_20260725_130230/node0_eth1.pid b/ScaleDF_rebuttal/_host_logs/dualnode_nccl_smoke_eth1_20260725_130230/node0_eth1.pid new file mode 100644 index 0000000000000000000000000000000000000000..c26988f17bc8237d706fc1d8b4e035c166af8d98 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/dualnode_nccl_smoke_eth1_20260725_130230/node0_eth1.pid @@ -0,0 +1 @@ +23434 diff --git a/ScaleDF_rebuttal/_host_logs/dualnode_nccl_smoke_eth1_20260725_130230/node1_eth1.log b/ScaleDF_rebuttal/_host_logs/dualnode_nccl_smoke_eth1_20260725_130230/node1_eth1.log new file mode 100644 index 0000000000000000000000000000000000000000..32b868665514b91d9564b3be4d9181b7df8ad16a --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/dualnode_nccl_smoke_eth1_20260725_130230/node1_eth1.log @@ -0,0 +1,7 @@ +/home/tiger/.profile: line 29: /run/bin/env: No such file or directory +/home/tiger/.profile: line 31: /tmp/m9/bin/env: No such file or directory +W0725 13:02:45.332000 13209 torch/distributed/run.py:766] +W0725 13:02:45.332000 13209 torch/distributed/run.py:766] ***************************************** +W0725 13:02:45.332000 13209 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 13:02:45.332000 13209 torch/distributed/run.py:766] ***************************************** +[W725 13:03:05.535129503 socket.cpp:200] [c10d] The hostname of the client socket cannot be retrieved. err=-3 diff --git a/ScaleDF_rebuttal/_host_logs/dualnode_nccl_smoke_eth1_20260725_130230/node1_eth1.pid b/ScaleDF_rebuttal/_host_logs/dualnode_nccl_smoke_eth1_20260725_130230/node1_eth1.pid new file mode 100644 index 0000000000000000000000000000000000000000..f024fc0522489ecb2326b86ece68ad57d01e88f6 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/dualnode_nccl_smoke_eth1_20260725_130230/node1_eth1.pid @@ -0,0 +1 @@ +13207 diff --git a/ScaleDF_rebuttal/_host_logs/dualnode_nccl_smoke_store_eth0_nccl_eth1_20260725_130401/node0_store_eth0_nccl_eth1.log b/ScaleDF_rebuttal/_host_logs/dualnode_nccl_smoke_store_eth0_nccl_eth1_20260725_130401/node0_store_eth0_nccl_eth1.log new file mode 100644 index 0000000000000000000000000000000000000000..ac5bd1b71021dcc0d668f047c0b7d977d2f629f2 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/dualnode_nccl_smoke_store_eth0_nccl_eth1_20260725_130401/node0_store_eth0_nccl_eth1.log @@ -0,0 +1,52 @@ +/home/tiger/.profile: line 29: /run/bin/env: No such file or directory +/home/tiger/.profile: line 31: /tmp/m9/bin/env: No such file or directory +W0725 13:04:19.266000 23551 torch/distributed/run.py:766] +W0725 13:04:19.266000 23551 torch/distributed/run.py:766] ***************************************** +W0725 13:04:19.266000 23551 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 13:04:19.266000 23551 torch/distributed/run.py:766] ***************************************** +start utc=2026-07-25T13:04:32Z host=n116-042-186 rank=0/16 local_rank=0 backend=nccl master=10.116.42.186:34901 +start utc=2026-07-25T13:04:34Z host=n116-042-186 rank=2/16 local_rank=2 backend=nccl master=10.116.42.186:34901 +start utc=2026-07-25T13:04:35Z host=n116-042-186 rank=3/16 local_rank=3 backend=nccl master=10.116.42.186:34901 +start utc=2026-07-25T13:04:35Z host=n116-042-186 rank=4/16 local_rank=4 backend=nccl master=10.116.42.186:34901 +start utc=2026-07-25T13:04:36Z host=n116-042-186 rank=7/16 local_rank=7 backend=nccl master=10.116.42.186:34901 +NCCL version 2.26.2+cuda12.2 +start utc=2026-07-25T13:04:36Z host=n116-042-186 rank=1/16 local_rank=1 backend=nccl master=10.116.42.186:34901 +start utc=2026-07-25T13:04:36Z host=n116-042-186 rank=5/16 local_rank=5 backend=nccl master=10.116.42.186:34901 +start utc=2026-07-25T13:04:37Z host=n116-042-186 rank=6/16 local_rank=6 backend=nccl master=10.116.42.186:34901 +W0725 13:05:14.571000 23551 torch/distributed/elastic/agent/server/api.py:719] Received 15 death signal, shutting down workers +W0725 13:05:14.572000 23551 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 23616 closing signal SIGTERM +W0725 13:05:14.572000 23551 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 23617 closing signal SIGTERM +W0725 13:05:14.573000 23551 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 23618 closing signal SIGTERM +W0725 13:05:14.577000 23551 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 23619 closing signal SIGTERM +W0725 13:05:14.578000 23551 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 23620 closing signal SIGTERM +W0725 13:05:14.581000 23551 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 23621 closing signal SIGTERM +W0725 13:05:14.582000 23551 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 23622 closing signal SIGTERM +W0725 13:05:14.585000 23551 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 23623 closing signal SIGTERM +Traceback (most recent call last): + File "/home/tiger/.local/bin/torchrun", line 7, in + sys.exit(main()) + ^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 355, in wrapper + return f(*args, **kwargs) + ^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 892, in main + run(args) + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 883, in run + elastic_launch( + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 139, in __call__ + return launch_agent(self._config, self._entrypoint, list(args)) + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 261, in launch_agent + result = agent.run() + ^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/metrics/api.py", line 138, in wrapper + result = f(*args, **kwargs) + ^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/agent/server/api.py", line 711, in run + result = self._invoke_run(role) + ^^^^^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/agent/server/api.py", line 870, in _invoke_run + time.sleep(monitor_interval) + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 84, in _terminate_process_handler + raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval) +torch.distributed.elastic.multiprocessing.api.SignalException: Process 23551 got signal: 15 diff --git a/ScaleDF_rebuttal/_host_logs/dualnode_nccl_smoke_store_eth0_nccl_eth1_20260725_130401/node0_store_eth0_nccl_eth1.pid b/ScaleDF_rebuttal/_host_logs/dualnode_nccl_smoke_store_eth0_nccl_eth1_20260725_130401/node0_store_eth0_nccl_eth1.pid new file mode 100644 index 0000000000000000000000000000000000000000..6ba39ac9457a7321c4ddb5762d62d812f8477205 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/dualnode_nccl_smoke_store_eth0_nccl_eth1_20260725_130401/node0_store_eth0_nccl_eth1.pid @@ -0,0 +1 @@ +23549 diff --git a/ScaleDF_rebuttal/_host_logs/dualnode_nccl_smoke_store_eth0_nccl_eth1_20260725_130401/node1_store_eth0_nccl_eth1.log b/ScaleDF_rebuttal/_host_logs/dualnode_nccl_smoke_store_eth0_nccl_eth1_20260725_130401/node1_store_eth0_nccl_eth1.log new file mode 100644 index 0000000000000000000000000000000000000000..957231fdcdb8fa3becb1a801a1186f333768b5c0 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/dualnode_nccl_smoke_store_eth0_nccl_eth1_20260725_130401/node1_store_eth0_nccl_eth1.log @@ -0,0 +1,75 @@ +/home/tiger/.profile: line 29: /run/bin/env: No such file or directory +/home/tiger/.profile: line 31: /tmp/m9/bin/env: No such file or directory +W0725 13:04:15.989000 13319 torch/distributed/run.py:766] +W0725 13:04:15.989000 13319 torch/distributed/run.py:766] ***************************************** +W0725 13:04:15.989000 13319 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 13:04:15.989000 13319 torch/distributed/run.py:766] ***************************************** +start utc=2026-07-25T13:04:30Z host=n116-042-185 rank=8/16 local_rank=0 backend=nccl master=10.116.42.186:34901 +start utc=2026-07-25T13:04:30Z host=n116-042-185 rank=12/16 local_rank=4 backend=nccl master=10.116.42.186:34901 +start utc=2026-07-25T13:04:31Z host=n116-042-185 rank=11/16 local_rank=3 backend=nccl master=10.116.42.186:34901 +start utc=2026-07-25T13:04:31Z host=n116-042-185 rank=9/16 local_rank=1 backend=nccl master=10.116.42.186:34901 +start utc=2026-07-25T13:04:32Z host=n116-042-185 rank=14/16 local_rank=6 backend=nccl master=10.116.42.186:34901 +start utc=2026-07-25T13:04:32Z host=n116-042-185 rank=15/16 local_rank=7 backend=nccl master=10.116.42.186:34901 +start utc=2026-07-25T13:04:33Z host=n116-042-185 rank=13/16 local_rank=5 backend=nccl master=10.116.42.186:34901 +start utc=2026-07-25T13:04:33Z host=n116-042-185 rank=10/16 local_rank=2 backend=nccl master=10.116.42.186:34901 + +[2026-07-25 13:04:55] n116-042-185:13384:13430 [1] misc/socket.cc:583 NCCL WARN socketPollConnect poll() returned 1, no POLLOUT events +SMOKE_FAILED: DistBackendError: NCCL error in: /pytorch/torch/csrc/distributed/c10d/NCCLUtils.cpp:77, unhandled system error (run with NCCL_DEBUG=INFO for details), NCCL version 2.26.2 +ncclSystemError: System call (e.g. socket, malloc) or external library call failed or device error. +Last error: +socketPollConnect poll() returned 1, no POLLOUT events +[rank9]: Traceback (most recent call last): +[rank9]: File "/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/distributed_nccl_smoke.py", line 55, in +[rank9]: main() +[rank9]: File "/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/distributed_nccl_smoke.py", line 38, in main +[rank9]: dist.all_reduce(tensor, op=dist.ReduceOp.SUM) +[rank9]: File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/c10d_logger.py", line 81, in wrapper +[rank9]: return func(*args, **kwargs) +[rank9]: ^^^^^^^^^^^^^^^^^^^^^ +[rank9]: File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py", line 2810, in all_reduce +[rank9]: work = group.allreduce([tensor], opts) +[rank9]: ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ +[rank9]: torch.distributed.DistBackendError: NCCL error in: /pytorch/torch/csrc/distributed/c10d/NCCLUtils.cpp:77, unhandled system error (run with NCCL_DEBUG=INFO for details), NCCL version 2.26.2 +[rank9]: ncclSystemError: System call (e.g. socket, malloc) or external library call failed or device error. +[rank9]: Last error: +[rank9]: socketPollConnect poll() returned 1, no POLLOUT events +W0725 13:04:56.239000 13319 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 13383 closing signal SIGTERM +W0725 13:04:56.240000 13319 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 13385 closing signal SIGTERM +W0725 13:04:56.241000 13319 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 13386 closing signal SIGTERM +W0725 13:04:56.245000 13319 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 13387 closing signal SIGTERM +W0725 13:04:56.249000 13319 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 13388 closing signal SIGTERM +W0725 13:04:56.253000 13319 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 13389 closing signal SIGTERM +W0725 13:04:56.253000 13319 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 13390 closing signal SIGTERM +E0725 13:04:58.201000 13319 torch/distributed/elastic/multiprocessing/api.py:874] failed (exitcode: 1) local_rank: 1 (pid: 13384) of binary: /usr/bin/python3 +Traceback (most recent call last): + File "/home/tiger/.local/bin/torchrun", line 7, in + sys.exit(main()) + ^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 355, in wrapper + return f(*args, **kwargs) + ^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 892, in main + run(args) + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/run.py", line 883, in run + elastic_launch( + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 139, in __call__ + return launch_agent(self._config, self._entrypoint, list(args)) + ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + File "/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/launcher/api.py", line 270, in launch_agent + raise ChildFailedError( +torch.distributed.elastic.multiprocessing.errors.ChildFailedError: +============================================================ +/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/distributed_nccl_smoke.py FAILED +------------------------------------------------------------ +Failures: + +------------------------------------------------------------ +Root Cause (first observed failure): +[0]: + time : 2026-07-25_13:04:56 + host : n116-042-185.byted.org + rank : 9 (local_rank: 1) + exitcode : 1 (pid: 13384) + error_file: + traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html +============================================================ diff --git a/ScaleDF_rebuttal/_host_logs/dualnode_nccl_smoke_store_eth0_nccl_eth1_20260725_130401/node1_store_eth0_nccl_eth1.pid b/ScaleDF_rebuttal/_host_logs/dualnode_nccl_smoke_store_eth0_nccl_eth1_20260725_130401/node1_store_eth0_nccl_eth1.pid new file mode 100644 index 0000000000000000000000000000000000000000..c7ce89ced910f47521223032ca3bc81fd76c103d --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/dualnode_nccl_smoke_store_eth0_nccl_eth1_20260725_130401/node1_store_eth0_nccl_eth1.pid @@ -0,0 +1 @@ +13317 diff --git a/ScaleDF_rebuttal/_host_logs/dualnode_smoke_20260725_125809/tcp_server_29686.log b/ScaleDF_rebuttal/_host_logs/dualnode_smoke_20260725_125809/tcp_server_29686.log new file mode 100644 index 0000000000000000000000000000000000000000..75a94461bd6f610ccab33c6e56cf66362f64a5de --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/dualnode_smoke_20260725_125809/tcp_server_29686.log @@ -0,0 +1,3 @@ +Traceback (most recent call last): + File "", line 1, in +OSError: [Errno 98] Address already in use diff --git a/ScaleDF_rebuttal/_host_logs/dualnode_smoke_20260725_125809/tcp_server_29686.pid b/ScaleDF_rebuttal/_host_logs/dualnode_smoke_20260725_125809/tcp_server_29686.pid new file mode 100644 index 0000000000000000000000000000000000000000..777251e9d56e0b91982675b9c59b26d6f8a872f9 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/dualnode_smoke_20260725_125809/tcp_server_29686.pid @@ -0,0 +1 @@ +17694 diff --git a/ScaleDF_rebuttal/_host_logs/dualnode_smoke_20260725_125809/tcp_server_random.log b/ScaleDF_rebuttal/_host_logs/dualnode_smoke_20260725_125809/tcp_server_random.log new file mode 100644 index 0000000000000000000000000000000000000000..d4c5281045217d6d5967bf50c7f74276f0f9b133 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/dualnode_smoke_20260725_125809/tcp_server_random.log @@ -0,0 +1,4 @@ +Traceback (most recent call last): + File "", line 1, in + File "", line 679, in __getitem__ +KeyError: 'PORTFILE' diff --git a/ScaleDF_rebuttal/_host_logs/dualnode_smoke_20260725_125809/tcp_server_random.pid b/ScaleDF_rebuttal/_host_logs/dualnode_smoke_20260725_125809/tcp_server_random.pid new file mode 100644 index 0000000000000000000000000000000000000000..3e5d688fd0b92cf85e106b8aec0225768239dd0d --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/dualnode_smoke_20260725_125809/tcp_server_random.pid @@ -0,0 +1 @@ +17727 PORTFILE=/tcp_server_random.port diff --git a/ScaleDF_rebuttal/_host_logs/dualnode_smoke_20260725_125913/tcp_server_random.log b/ScaleDF_rebuttal/_host_logs/dualnode_smoke_20260725_125913/tcp_server_random.log new file mode 100644 index 0000000000000000000000000000000000000000..55ddf53d423fe8a8bca20aa1b83a32201d9dd2d6 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/dualnode_smoke_20260725_125913/tcp_server_random.log @@ -0,0 +1,4 @@ +listening 0.0.0.0:34043 +accepted ('10.116.42.185', 53742) hello-10.116.42.186 +accepted ('100.64.3.114', 51936) hello-100.64.3.115 +server_done accepted=2 diff --git a/ScaleDF_rebuttal/_host_logs/dualnode_smoke_20260725_125913/tcp_server_random.pid b/ScaleDF_rebuttal/_host_logs/dualnode_smoke_20260725_125913/tcp_server_random.pid new file mode 100644 index 0000000000000000000000000000000000000000..c92d5be833f17591f95633faed8d2e3cb75ef501 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/dualnode_smoke_20260725_125913/tcp_server_random.pid @@ -0,0 +1 @@ +17750 diff --git a/ScaleDF_rebuttal/_host_logs/dualnode_smoke_20260725_125913/tcp_server_random.port b/ScaleDF_rebuttal/_host_logs/dualnode_smoke_20260725_125913/tcp_server_random.port new file mode 100644 index 0000000000000000000000000000000000000000..d9c25f40ba3df406a8f821dc4fbdcd183ddf7936 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/dualnode_smoke_20260725_125913/tcp_server_random.port @@ -0,0 +1 @@ +34043 \ No newline at end of file diff --git a/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_112559/eval_queue_n111-046-081.log b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_112559/eval_queue_n111-046-081.log new file mode 100644 index 0000000000000000000000000000000000000000..3ce6f28565347b6fabdb0fa40ffc1d3b500b413e --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_112559/eval_queue_n111-046-081.log @@ -0,0 +1,5169 @@ +eval_queue_started_utc=2026-07-25T17:02:42Z +hostname=n111-046-081 +train_tag=20260725_112559 +eval_tag=20260725_112559 +real_level=16 +fake_level=32 +eval_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559 +batch_per_gpu=256 +[2026-07-25T17:02:42Z] eval_wait tag=fixed500k_real_n16_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed1_resampled +[2026-07-25T17:03:42Z] eval_wait tag=fixed500k_real_n16_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed1_resampled +[2026-07-25T17:04:42Z] eval_skip tag=fixed500k_real_n16_seed1_resampled reason=complete_after_lock output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed1_resampled +[2026-07-25T17:04:43Z] eval_wait tag=fixed500k_fake_n32_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed1_resampled +[2026-07-25T17:05:43Z] eval_wait tag=fixed500k_fake_n32_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed1_resampled +[2026-07-25T17:06:43Z] eval_wait tag=fixed500k_fake_n32_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed1_resampled +[2026-07-25T17:07:43Z] eval_wait tag=fixed500k_fake_n32_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed1_resampled +[2026-07-25T17:08:43Z] eval_wait tag=fixed500k_fake_n32_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed1_resampled +[2026-07-25T17:09:43Z] eval_wait tag=fixed500k_fake_n32_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed1_resampled +[2026-07-25T17:10:43Z] eval_wait tag=fixed500k_fake_n32_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed1_resampled +[2026-07-25T17:11:43Z] eval_wait tag=fixed500k_fake_n32_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed1_resampled +[2026-07-25T17:12:44Z] eval_wait tag=fixed500k_fake_n32_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed1_resampled +[2026-07-25T17:13:44Z] eval_wait tag=fixed500k_fake_n32_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed1_resampled +[2026-07-25T17:14:44Z] eval_wait tag=fixed500k_fake_n32_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed1_resampled +[2026-07-25T17:15:44Z] eval_start tag=fixed500k_fake_n32_seed1_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed1_resampled_n111-046-081_20260725_112559/best_checkpoint.pth +started_utc=2026-07-25T17:15:44Z +hostname=n111-046-081 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed1_resampled_n111-046-081_20260725_112559/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed1_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0725 17:15:59.558000 2562114 torch/distributed/run.py:766] +W0725 17:15:59.558000 2562114 torch/distributed/run.py:766] ***************************************** +W0725 17:15:59.558000 2562114 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:15:59.558000 2562114 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:16:33.506013947 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:16:34.745321214 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:16:34.879429642 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:16:34.189579309 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:16:34.308733221 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:16:34.310764704 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 17:16:34.312138644 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:16:34.347618135 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed1_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed1_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:20:07 acc1_pos: 59.115 acc5_pos: 100.000 loss: 0.6906 (0.6906) acc1: 59.1146 (59.1146) acc5: 100.0000 (100.0000) time: 5.1597 data: 3.3009 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/234] eta: 0:02:03 acc1_pos: 26.562 acc5_pos: 100.000 loss: 0.7892 (0.8256) acc1: 47.3958 (45.6676) acc5: 100.0000 (100.0000) time: 0.5508 data: 0.3004 max mem: 3531 +Test: [ 20/234] eta: 0:01:12 acc1_pos: 92.969 acc5_pos: 100.000 loss: 0.7892 (0.7877) acc1: 45.3125 (49.4668) acc5: 100.0000 (100.0000) time: 0.1001 data: 0.0093 max mem: 3531 +Test: [ 30/234] eta: 0:00:54 acc1_pos: 76.562 acc5_pos: 100.000 loss: 0.3036 (0.6326) acc1: 91.9271 (62.6932) acc5: 100.0000 (100.0000) time: 0.1116 data: 0.0167 max mem: 3531 +Test: [ 40/234] eta: 0:00:47 acc1_pos: 98.958 acc5_pos: 100.000 loss: 0.3012 (0.5868) acc1: 92.1875 (66.0569) acc5: 100.0000 (100.0000) time: 0.1449 data: 0.0468 max mem: 3531 +Test: [ 50/234] eta: 0:00:42 acc1_pos: 99.479 acc5_pos: 100.000 loss: 0.1703 (0.5039) acc1: 97.6562 (72.3601) acc5: 100.0000 (100.0000) time: 0.1696 data: 0.0748 max mem: 3531 +Test: [ 60/234] eta: 0:00:38 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.1703 (0.4501) acc1: 99.2188 (76.8272) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1699 data: 0.0759 max mem: 3531 +Test: [ 70/234] eta: 0:00:35 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.1759 (0.4664) acc1: 99.2188 (76.3021) acc5: 100.0000 (100.0000) acc1_neg: 58.854 acc5_neg: 100.000 time: 0.1788 data: 0.0812 max mem: 3531 +Test: [ 80/234] eta: 0:00:32 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.6863 (0.4770) acc1: 60.9375 (75.9549) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1751 data: 0.0789 max mem: 3531 +Test: [ 90/234] eta: 0:00:29 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.3343 (0.4643) acc1: 86.1979 (77.2350) acc5: 100.0000 (100.0000) acc1_neg: 65.885 acc5_neg: 100.000 time: 0.1625 data: 0.0706 max mem: 3531 +Test: [100/234] eta: 0:00:27 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.3343 (0.4725) acc1: 86.1979 (76.7688) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1715 data: 0.0786 max mem: 3531 +Test: [110/234] eta: 0:00:25 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.5463 (0.4849) acc1: 72.1354 (76.0417) acc5: 100.0000 (100.0000) acc1_neg: 54.688 acc5_neg: 100.000 time: 0.1952 data: 0.1021 max mem: 3531 +Test: [120/234] eta: 0:00:22 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.5123 (0.4925) acc1: 72.1354 (75.6909) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.1983 data: 0.1078 max mem: 3531 +Test: [130/234] eta: 0:00:20 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.3749 (0.4868) acc1: 84.3750 (76.3995) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1570 data: 0.0626 max mem: 3531 +Test: [140/234] eta: 0:00:18 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.3749 (0.5002) acc1: 83.5938 (75.9752) acc5: 100.0000 (100.0000) acc1_neg: 37.500 acc5_neg: 100.000 time: 0.1509 data: 0.0546 max mem: 3531 +Test: [150/234] eta: 0:00:16 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.5528 (0.5112) acc1: 65.8854 (75.4674) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.1999 data: 0.1052 max mem: 3531 +Test: [160/234] eta: 0:00:14 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.4854 (0.5035) acc1: 77.6042 (76.1177) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.2100 data: 0.1169 max mem: 3531 +Test: [170/234] eta: 0:00:12 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.4800 (0.5020) acc1: 78.9062 (76.3158) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1999 data: 0.1074 max mem: 3531 +Test: [180/234] eta: 0:00:10 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.5039 (0.5174) acc1: 75.7812 (75.2949) acc5: 100.0000 (100.0000) acc1_neg: 22.396 acc5_neg: 100.000 time: 0.2060 data: 0.1132 max mem: 3531 +Test: [190/234] eta: 0:00:08 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.5751 (0.5219) acc1: 69.2708 (74.9386) acc5: 100.0000 (100.0000) acc1_neg: 72.396 acc5_neg: 100.000 time: 0.2011 data: 0.1103 max mem: 3531 +Test: [200/234] eta: 0:00:06 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.5405 (0.5088) acc1: 72.3958 (75.6646) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1989 data: 0.1089 max mem: 3531 +Test: [210/234] eta: 0:00:04 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.0864 (0.4910) acc1: 99.7396 (76.7810) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2217 data: 0.1309 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.2227 (0.4796) acc1: 97.9167 (77.6749) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.2159 data: 0.1251 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.3702 (0.4874) acc1: 64.5833 (76.9920) acc5: 100.0000 (100.0000) acc1_neg: 61.198 acc5_neg: 100.000 time: 0.1717 data: 0.0819 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.6393 (0.4896) acc1: 63.0208 (76.8291) acc5: 100.0000 (100.0000) acc1_neg: 61.868 acc5_neg: 100.000 time: 0.1628 data: 0.0720 max mem: 3532 +Test: Total time: 0:00:46 (0.1972 s / it) +* Overall Acc@1 76.829 Acc@5 100.000 + target == real Acc@1 72.553 Acc@5 100.000 + target == fake Acc@1 78.034 Acc@5 100.000 + loss 0.490 + PR-AUC 0.9493 ROC-AUC 0.8420 EER 24.35% (thr=0.5303) +Accuracy of the network on the 89729 test images: 76.8% +[rank0]:[W725 17:17:32.268924205 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 17:17:53.718000 2562699 torch/distributed/run.py:766] +W0725 17:17:53.718000 2562699 torch/distributed/run.py:766] ***************************************** +W0725 17:17:53.718000 2562699 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:17:53.718000 2562699 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 4): env:// +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:18:22.835678406 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:18:22.099456071 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:18:22.560816521 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:18:23.799563614 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +| distributed init (rank 5): env:// +[rank0]:[W725 17:18:23.803739060 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:18:23.806702050 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:18:23.814304553 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank1]:[W725 17:18:23.814352865 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed1_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed1_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:31:54 acc1_pos: 26.302 acc5_pos: 100.000 loss: 1.1336 (1.1336) acc1: 26.3021 (26.3021) acc5: 100.0000 (100.0000) time: 6.8346 data: 5.3563 max mem: 3531 +Test: [ 10/2387] eta: 0:27:51 acc1_pos: 12.240 acc5_pos: 100.000 loss: 1.1401 (1.1180) acc1: 21.0938 (22.8693) acc5: 100.0000 (100.0000) time: 0.7031 data: 0.4873 max mem: 3531 +Test: [ 20/2387] eta: 0:16:24 acc1_pos: 24.219 acc5_pos: 100.000 loss: 1.1401 (1.0901) acc1: 19.7917 (24.6900) acc5: 100.0000 (100.0000) time: 0.0949 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:14:22 acc1_pos: 15.104 acc5_pos: 100.000 loss: 1.1086 (1.0862) acc1: 19.0104 (24.6724) acc5: 100.0000 (100.0000) time: 0.1805 data: 0.0853 max mem: 3531 +Test: [ 40/2387] eta: 0:12:54 acc1_pos: 17.969 acc5_pos: 100.000 loss: 1.1086 (1.0875) acc1: 18.7500 (23.0818) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1484 max mem: 3531 +Test: [ 50/2387] eta: 0:12:10 acc1_pos: 41.927 acc5_pos: 100.000 loss: 1.0200 (1.0765) acc1: 20.3125 (24.4690) acc5: 100.0000 (100.0000) time: 0.2294 data: 0.1371 max mem: 3531 +Test: [ 60/2387] eta: 0:11:50 acc1_pos: 39.844 acc5_pos: 100.000 loss: 1.0709 (1.0815) acc1: 22.6562 (24.1078) acc5: 100.0000 (100.0000) time: 0.2549 data: 0.1596 max mem: 3531 +Test: [ 70/2387] eta: 0:11:29 acc1_pos: 33.594 acc5_pos: 100.000 loss: 1.1287 (1.0915) acc1: 19.5312 (23.5329) acc5: 100.0000 (100.0000) time: 0.2598 data: 0.1630 max mem: 3531 +Test: [ 80/2387] eta: 0:11:10 acc1_pos: 36.198 acc5_pos: 100.000 loss: 1.1175 (1.0886) acc1: 16.1458 (23.2382) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1490 max mem: 3531 +Test: [ 90/2387] eta: 0:10:59 acc1_pos: 36.979 acc5_pos: 100.000 loss: 1.0483 (1.0783) acc1: 25.0000 (24.0642) acc5: 100.0000 (100.0000) time: 0.2499 data: 0.1532 max mem: 3531 +Test: [ 100/2387] eta: 0:10:43 acc1_pos: 22.135 acc5_pos: 100.000 loss: 1.0451 (1.0815) acc1: 25.0000 (23.8114) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1473 max mem: 3531 +Test: [ 110/2387] eta: 0:10:32 acc1_pos: 38.281 acc5_pos: 100.000 loss: 1.0238 (1.0707) acc1: 24.7396 (24.4369) acc5: 100.0000 (100.0000) time: 0.2360 data: 0.1390 max mem: 3531 +Test: [ 120/2387] eta: 0:10:22 acc1_pos: 21.875 acc5_pos: 100.000 loss: 1.0082 (1.0709) acc1: 25.7812 (24.6018) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1449 max mem: 3531 +Test: [ 130/2387] eta: 0:10:13 acc1_pos: 15.885 acc5_pos: 100.000 loss: 1.0660 (1.0703) acc1: 20.3125 (24.2684) acc5: 100.0000 (100.0000) time: 0.2390 data: 0.1440 max mem: 3531 +Test: [ 140/2387] eta: 0:10:07 acc1_pos: 26.302 acc5_pos: 100.000 loss: 1.0432 (1.0666) acc1: 20.5729 (24.4219) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1476 max mem: 3531 +Test: [ 150/2387] eta: 0:10:14 acc1_pos: 8.854 acc5_pos: 100.000 loss: 0.9972 (1.0626) acc1: 26.3021 (24.6896) acc5: 100.0000 (100.0000) time: 0.2913 data: 0.1940 max mem: 3531 +Test: [ 160/2387] eta: 0:09:46 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.9487 (1.0433) acc1: 29.4271 (26.3263) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2149 data: 0.1189 max mem: 3531 +Test: [ 170/2387] eta: 0:09:33 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2050 (0.9937) acc1: 97.1354 (30.5266) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1384 data: 0.0386 max mem: 3531 +Test: [ 180/2387] eta: 0:09:22 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1933 (0.9496) acc1: 98.1771 (34.2728) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1852 data: 0.0859 max mem: 3531 +Test: [ 190/2387] eta: 0:09:14 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1911 (0.9100) acc1: 98.4375 (37.6323) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1999 data: 0.1040 max mem: 3531 +Test: [ 200/2387] eta: 0:09:02 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1909 (0.8742) acc1: 98.4375 (40.6639) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1847 data: 0.0888 max mem: 3531 +Test: [ 210/2387] eta: 0:08:51 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1935 (0.8422) acc1: 98.4375 (43.4020) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1649 data: 0.0686 max mem: 3532 +Test: [ 220/2387] eta: 0:08:42 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1953 (0.8129) acc1: 98.4375 (45.8852) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1747 data: 0.0749 max mem: 3532 +Test: [ 230/2387] eta: 0:08:33 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1953 (0.7863) acc1: 98.1771 (48.1444) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1749 data: 0.0738 max mem: 3532 +Test: [ 240/2387] eta: 0:08:25 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2042 (0.7624) acc1: 98.1771 (50.2140) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1699 data: 0.0717 max mem: 3532 +Test: [ 250/2387] eta: 0:08:16 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1916 (0.7355) acc1: 98.6979 (52.1871) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1649 data: 0.0670 max mem: 3532 +Test: [ 260/2387] eta: 0:08:09 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0637 (0.7096) acc1: 100.0000 (54.0190) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1700 data: 0.0717 max mem: 3532 +Test: [ 270/2387] eta: 0:08:01 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0611 (0.6857) acc1: 100.0000 (55.7157) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1654 data: 0.0689 max mem: 3532 +Test: [ 280/2387] eta: 0:07:54 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0611 (0.6636) acc1: 100.0000 (57.2917) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1599 data: 0.0632 max mem: 3532 +Test: [ 290/2387] eta: 0:07:44 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0649 (0.6430) acc1: 100.0000 (58.7593) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1395 data: 0.0410 max mem: 3532 +Test: [ 300/2387] eta: 0:07:38 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0649 (0.6238) acc1: 100.0000 (60.1294) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1399 data: 0.0413 max mem: 3532 +Test: [ 310/2387] eta: 0:07:32 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0692 (0.6102) acc1: 100.0000 (61.3779) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1649 data: 0.0669 max mem: 3532 +Test: [ 320/2387] eta: 0:07:28 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.3646 (0.6026) acc1: 98.1771 (62.5073) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1749 data: 0.0793 max mem: 3532 +Test: [ 330/2387] eta: 0:07:23 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.3652 (0.5954) acc1: 97.6562 (63.5794) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1815 data: 0.0875 max mem: 3532 +Test: [ 340/2387] eta: 0:07:21 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.3621 (0.5885) acc1: 97.6562 (64.5826) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2000 data: 0.1069 max mem: 3532 +Test: [ 350/2387] eta: 0:07:24 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.3555 (0.5770) acc1: 98.6979 (65.5835) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2591 data: 0.1646 max mem: 3532 +Test: [ 360/2387] eta: 0:07:21 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1607 (0.5656) acc1: 99.7396 (66.5166) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2487 data: 0.1522 max mem: 3532 +Test: [ 370/2387] eta: 0:07:13 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1711 (0.5550) acc1: 98.9583 (67.3883) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1641 data: 0.0711 max mem: 3532 +Test: [ 380/2387] eta: 0:07:08 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1718 (0.5449) acc1: 98.6979 (68.2107) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1370 data: 0.0445 max mem: 3532 +Test: [ 390/2387] eta: 0:07:06 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1696 (0.5354) acc1: 98.9583 (68.9958) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1805 data: 0.0876 max mem: 3532 +Test: [ 400/2387] eta: 0:07:03 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1699 (0.5263) acc1: 98.9583 (69.7410) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2057 data: 0.1144 max mem: 3532 +Test: [ 410/2387] eta: 0:06:59 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1694 (0.5177) acc1: 98.9583 (70.4500) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1901 data: 0.0986 max mem: 3532 +Test: [ 420/2387] eta: 0:06:57 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1695 (0.5094) acc1: 98.9583 (71.1253) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1931 data: 0.0953 max mem: 3532 +Test: [ 430/2387] eta: 0:06:54 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1737 (0.5019) acc1: 98.6979 (71.7656) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2052 data: 0.1081 max mem: 3532 +Test: [ 440/2387] eta: 0:06:51 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1743 (0.4932) acc1: 98.9583 (72.3905) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1998 data: 0.1080 max mem: 3533 +Test: [ 450/2387] eta: 0:06:49 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0701 (0.4838) acc1: 100.0000 (73.0027) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1990 data: 0.1046 max mem: 3533 +Test: [ 460/2387] eta: 0:06:46 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0695 (0.4748) acc1: 100.0000 (73.5883) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1999 data: 0.1059 max mem: 3533 +Test: [ 470/2387] eta: 0:06:43 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0695 (0.4662) acc1: 100.0000 (74.1480) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1978 data: 0.1037 max mem: 3533 +Test: [ 480/2387] eta: 0:06:41 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0699 (0.4580) acc1: 100.0000 (74.6854) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.1012 max mem: 3533 +Test: [ 490/2387] eta: 0:06:38 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0706 (0.4501) acc1: 100.0000 (75.2010) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1968 data: 0.1028 max mem: 3533 +Test: [ 500/2387] eta: 0:06:36 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0704 (0.4425) acc1: 100.0000 (75.6955) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1979 data: 0.1041 max mem: 3533 +Test: [ 510/2387] eta: 0:06:33 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0704 (0.4354) acc1: 100.0000 (76.1711) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.1039 max mem: 3533 +Test: [ 520/2387] eta: 0:06:31 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0990 (0.4291) acc1: 100.0000 (76.6285) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2094 data: 0.1178 max mem: 3533 +Test: [ 530/2387] eta: 0:06:28 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1059 (0.4230) acc1: 100.0000 (77.0686) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1966 data: 0.1035 max mem: 3533 +Test: [ 540/2387] eta: 0:06:24 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1078 (0.4172) acc1: 100.0000 (77.4920) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1588 data: 0.0653 max mem: 3533 +Test: [ 550/2387] eta: 0:06:21 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1074 (0.4116) acc1: 100.0000 (77.9005) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1682 data: 0.0749 max mem: 3533 +Test: [ 560/2387] eta: 0:06:19 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1078 (0.4062) acc1: 100.0000 (78.2944) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1940 data: 0.1012 max mem: 3533 +Test: [ 570/2387] eta: 0:06:17 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1080 (0.4009) acc1: 100.0000 (78.6746) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2068 data: 0.1116 max mem: 3533 +Test: [ 580/2387] eta: 0:06:15 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1080 (0.3959) acc1: 100.0000 (79.0416) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2094 data: 0.1137 max mem: 3533 +Test: [ 590/2387] eta: 0:06:12 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1087 (0.3910) acc1: 100.0000 (79.3962) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1987 data: 0.1063 max mem: 3533 +Test: [ 600/2387] eta: 0:06:09 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1052 (0.3862) acc1: 100.0000 (79.7391) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1803 data: 0.0895 max mem: 3533 +Test: [ 610/2387] eta: 0:06:06 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1190 (0.3824) acc1: 100.0000 (80.0634) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1698 data: 0.0787 max mem: 3533 +Test: [ 620/2387] eta: 0:06:02 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1494 (0.3786) acc1: 99.7396 (80.3786) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1626 data: 0.0711 max mem: 3533 +Test: [ 630/2387] eta: 0:05:59 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1486 (0.3750) acc1: 99.7396 (80.6825) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1542 data: 0.0613 max mem: 3533 +Test: [ 640/2387] eta: 0:05:55 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1527 (0.3715) acc1: 99.4792 (80.9750) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1553 data: 0.0622 max mem: 3533 +Test: [ 650/2387] eta: 0:05:52 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1531 (0.3681) acc1: 99.4792 (81.2584) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1550 data: 0.0644 max mem: 3533 +Test: [ 660/2387] eta: 0:05:49 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1503 (0.3648) acc1: 99.4792 (81.5341) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1564 data: 0.0644 max mem: 3533 +Test: [ 670/2387] eta: 0:05:46 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1509 (0.3617) acc1: 99.4792 (81.8050) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1605 data: 0.0684 max mem: 3534 +Test: [ 680/2387] eta: 0:05:42 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1537 (0.3588) acc1: 100.0000 (82.0695) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1566 data: 0.0659 max mem: 3534 +Test: [ 690/2387] eta: 0:05:39 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1711 (0.3588) acc1: 99.7396 (82.1183) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.1556 data: 0.0641 max mem: 3534 +Test: [ 700/2387] eta: 0:05:36 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.4276 (0.3597) acc1: 80.9896 (82.1096) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.1551 data: 0.0636 max mem: 3534 +Test: [ 710/2387] eta: 0:05:33 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.4264 (0.3607) acc1: 81.2500 (82.0917) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1536 data: 0.0622 max mem: 3534 +Test: [ 720/2387] eta: 0:05:30 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.4258 (0.3616) acc1: 80.9896 (82.0822) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.1555 data: 0.0637 max mem: 3534 +Test: [ 730/2387] eta: 0:05:28 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.4327 (0.3627) acc1: 80.7292 (82.0462) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.1699 data: 0.0762 max mem: 3534 +Test: [ 740/2387] eta: 0:05:26 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.4341 (0.3636) acc1: 79.6875 (82.0225) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.1937 data: 0.1007 max mem: 3534 +Test: [ 750/2387] eta: 0:05:24 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.4253 (0.3644) acc1: 80.7292 (82.0132) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.2049 data: 0.1139 max mem: 3534 +Test: [ 760/2387] eta: 0:05:22 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.4140 (0.3648) acc1: 81.2500 (82.0141) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.2090 data: 0.1180 max mem: 3534 +Test: [ 770/2387] eta: 0:05:21 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.3595 (0.3638) acc1: 85.4167 (82.1272) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2051 data: 0.1141 max mem: 3534 +Test: [ 780/2387] eta: 0:05:18 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2252 (0.3619) acc1: 95.5729 (82.3093) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1910 data: 0.0972 max mem: 3534 +Test: [ 790/2387] eta: 0:05:15 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1965 (0.3596) acc1: 96.8750 (82.5053) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1721 data: 0.0782 max mem: 3534 +Test: [ 800/2387] eta: 0:05:13 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1712 (0.3576) acc1: 97.9167 (82.6831) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1685 data: 0.0733 max mem: 3534 +Test: [ 810/2387] eta: 0:05:11 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1881 (0.3558) acc1: 97.9167 (82.8549) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1725 data: 0.0722 max mem: 3534 +Test: [ 820/2387] eta: 0:05:08 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2083 (0.3545) acc1: 95.0521 (82.9889) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1699 data: 0.0730 max mem: 3534 +Test: [ 830/2387] eta: 0:05:05 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2704 (0.3538) acc1: 93.2292 (83.0870) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1649 data: 0.0736 max mem: 3534 +Test: [ 840/2387] eta: 0:05:03 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2330 (0.3521) acc1: 96.3542 (83.2692) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1657 data: 0.0729 max mem: 3534 +Test: [ 850/2387] eta: 0:05:01 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1959 (0.3502) acc1: 99.2188 (83.4545) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1720 data: 0.0786 max mem: 3534 +Test: [ 860/2387] eta: 0:04:58 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1782 (0.3481) acc1: 99.7396 (83.6418) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1695 data: 0.0773 max mem: 3534 +Test: [ 870/2387] eta: 0:04:56 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1788 (0.3464) acc1: 99.2188 (83.8168) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1681 data: 0.0744 max mem: 3534 +Test: [ 880/2387] eta: 0:04:53 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1946 (0.3447) acc1: 98.9583 (83.9893) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1697 data: 0.0756 max mem: 3534 +Test: [ 890/2387] eta: 0:04:51 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1881 (0.3428) acc1: 99.2188 (84.1581) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1652 data: 0.0736 max mem: 3534 +Test: [ 900/2387] eta: 0:04:48 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2116 (0.3453) acc1: 95.5729 (84.0076) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.1567 data: 0.0643 max mem: 3535 +Test: [ 910/2387] eta: 0:04:46 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.6526 (0.3487) acc1: 64.3229 (83.7930) acc5: 100.0000 (100.0000) acc1_neg: 63.802 acc5_neg: 100.000 time: 0.1542 data: 0.0612 max mem: 3535 +Test: [ 920/2387] eta: 0:04:43 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.6647 (0.3521) acc1: 63.8021 (83.5697) acc5: 100.0000 (100.0000) acc1_neg: 59.896 acc5_neg: 100.000 time: 0.1487 data: 0.0566 max mem: 3535 +Test: [ 930/2387] eta: 0:04:40 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.6542 (0.3553) acc1: 64.0625 (83.3613) acc5: 100.0000 (100.0000) acc1_neg: 58.594 acc5_neg: 100.000 time: 0.1460 data: 0.0535 max mem: 3535 +Test: [ 940/2387] eta: 0:04:38 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.6573 (0.3586) acc1: 64.0625 (83.1457) acc5: 100.0000 (100.0000) acc1_neg: 59.896 acc5_neg: 100.000 time: 0.1534 data: 0.0608 max mem: 3535 +Test: [ 950/2387] eta: 0:04:35 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.6573 (0.3617) acc1: 63.8021 (82.9390) acc5: 100.0000 (100.0000) acc1_neg: 64.844 acc5_neg: 100.000 time: 0.1536 data: 0.0616 max mem: 3535 +Test: [ 960/2387] eta: 0:04:33 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.6429 (0.3646) acc1: 64.5833 (82.7499) acc5: 100.0000 (100.0000) acc1_neg: 65.625 acc5_neg: 100.000 time: 0.1596 data: 0.0656 max mem: 3535 +Test: [ 970/2387] eta: 0:04:31 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.6516 (0.3676) acc1: 64.5833 (82.5507) acc5: 100.0000 (100.0000) acc1_neg: 65.885 acc5_neg: 100.000 time: 0.1643 data: 0.0688 max mem: 3535 +Test: [ 980/2387] eta: 0:04:29 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.6199 (0.3679) acc1: 66.9271 (82.5667) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.1849 data: 0.0897 max mem: 3535 +Test: [ 990/2387] eta: 0:04:27 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2712 (0.3667) acc1: 92.9688 (82.6801) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2049 data: 0.1084 max mem: 3535 +Test: [1000/2387] eta: 0:04:25 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2212 (0.3653) acc1: 97.3958 (82.8159) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1999 data: 0.1041 max mem: 3535 +Test: [1010/2387] eta: 0:04:23 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2294 (0.3644) acc1: 94.7917 (82.9209) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1849 data: 0.0889 max mem: 3535 +Test: [1020/2387] eta: 0:04:21 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2433 (0.3632) acc1: 95.8333 (83.0303) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1649 data: 0.0676 max mem: 3535 +Test: [1030/2387] eta: 0:04:18 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1751 (0.3612) acc1: 97.6562 (83.1798) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1599 data: 0.0631 max mem: 3535 +Test: [1040/2387] eta: 0:04:16 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1440 (0.3590) acc1: 99.4792 (83.3376) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1599 data: 0.0655 max mem: 3535 +Test: [1050/2387] eta: 0:04:14 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1309 (0.3569) acc1: 100.0000 (83.4897) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1649 data: 0.0699 max mem: 3535 +Test: [1060/2387] eta: 0:04:12 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1357 (0.3550) acc1: 99.4792 (83.6367) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1699 data: 0.0720 max mem: 3535 +Test: [1070/2387] eta: 0:04:10 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1620 (0.3534) acc1: 98.9583 (83.7715) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1650 data: 0.0689 max mem: 3535 +Test: [1080/2387] eta: 0:04:07 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1958 (0.3521) acc1: 97.6562 (83.8838) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1676 data: 0.0721 max mem: 3535 +Test: [1090/2387] eta: 0:04:05 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1775 (0.3504) acc1: 97.6562 (84.0143) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1748 data: 0.0771 max mem: 3535 +Test: [1100/2387] eta: 0:04:03 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1521 (0.3486) acc1: 98.6979 (84.1508) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1772 data: 0.0795 max mem: 3535 +Test: [1110/2387] eta: 0:04:01 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1530 (0.3471) acc1: 98.6979 (84.2756) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1702 data: 0.0742 max mem: 3535 +Test: [1120/2387] eta: 0:03:59 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1853 (0.3457) acc1: 97.6562 (84.3943) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1749 data: 0.0795 max mem: 3536 +Test: [1130/2387] eta: 0:03:57 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1813 (0.3442) acc1: 98.1771 (84.5180) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1846 data: 0.0892 max mem: 3536 +Test: [1140/2387] eta: 0:03:55 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1990 (0.3431) acc1: 97.6562 (84.6249) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1849 data: 0.0883 max mem: 3536 +Test: [1150/2387] eta: 0:03:53 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1876 (0.3417) acc1: 98.6979 (84.7418) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0824 max mem: 3536 +Test: [1160/2387] eta: 0:03:52 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1244 (0.3396) acc1: 100.0000 (84.8725) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1953 data: 0.0782 max mem: 3536 +Test: [1170/2387] eta: 0:03:49 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1069 (0.3377) acc1: 100.0000 (85.0012) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1801 data: 0.0640 max mem: 3536 +Test: [1180/2387] eta: 0:03:50 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1085 (0.3358) acc1: 100.0000 (85.1263) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2699 data: 0.1726 max mem: 3536 +Test: [1190/2387] eta: 0:03:51 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1118 (0.3340) acc1: 100.0000 (85.2501) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.4750 data: 0.3763 max mem: 3536 +Test: [1200/2387] eta: 0:03:51 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1122 (0.3321) acc1: 100.0000 (85.3724) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.4798 data: 0.3849 max mem: 3536 +Test: [1210/2387] eta: 0:03:51 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1152 (0.3304) acc1: 100.0000 (85.4926) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.3546 data: 0.2560 max mem: 3536 +Test: [1220/2387] eta: 0:03:54 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0846 (0.3284) acc1: 100.0000 (85.6114) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5249 data: 0.4124 max mem: 3536 +Test: [1230/2387] eta: 0:03:55 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0722 (0.3263) acc1: 100.0000 (85.7283) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.6502 data: 0.5251 max mem: 3536 +Test: [1240/2387] eta: 0:03:55 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0694 (0.3242) acc1: 100.0000 (85.8429) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.4596 data: 0.3280 max mem: 3536 +Test: [1250/2387] eta: 0:03:55 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0721 (0.3223) acc1: 100.0000 (85.9560) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.4346 data: 0.3047 max mem: 3536 +Test: [1260/2387] eta: 0:03:56 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0745 (0.3203) acc1: 100.0000 (86.0674) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5148 data: 0.3926 max mem: 3536 +Test: [1270/2387] eta: 0:03:57 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0745 (0.3184) acc1: 100.0000 (86.1768) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5298 data: 0.4125 max mem: 3536 +Test: [1280/2387] eta: 0:03:58 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0806 (0.3166) acc1: 100.0000 (86.2821) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.5559 data: 0.4305 max mem: 3536 +Test: [1290/2387] eta: 0:03:59 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0955 (0.3149) acc1: 99.4792 (86.3849) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.5649 data: 0.4364 max mem: 3536 +Test: [1300/2387] eta: 0:03:59 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0925 (0.3133) acc1: 99.4792 (86.4864) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5192 data: 0.3952 max mem: 3536 +Test: [1310/2387] eta: 0:03:59 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0990 (0.3117) acc1: 99.7396 (86.5853) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.5196 data: 0.3969 max mem: 3536 +Test: [1320/2387] eta: 0:03:59 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1099 (0.3103) acc1: 99.4792 (86.6831) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5096 data: 0.3817 max mem: 3536 +Test: [1330/2387] eta: 0:03:59 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1278 (0.3090) acc1: 99.4792 (86.7753) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.4799 data: 0.3459 max mem: 3536 +Test: [1340/2387] eta: 0:03:58 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1652 (0.3088) acc1: 97.9167 (86.7955) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.4296 data: 0.2990 max mem: 3536 +Test: [1350/2387] eta: 0:03:56 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0544 (0.3069) acc1: 100.0000 (86.8932) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.3521 data: 0.2396 max mem: 3537 +Test: [1360/2387] eta: 0:03:55 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0410 (0.3049) acc1: 100.0000 (86.9895) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.3575 data: 0.2402 max mem: 3537 +Test: [1370/2387] eta: 0:03:55 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0410 (0.3031) acc1: 100.0000 (87.0844) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.4248 data: 0.2903 max mem: 3537 +Test: [1380/2387] eta: 0:03:54 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0444 (0.3012) acc1: 100.0000 (87.1779) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.4668 data: 0.3447 max mem: 3537 +Test: [1390/2387] eta: 0:03:52 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0444 (0.2994) acc1: 100.0000 (87.2701) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.3763 data: 0.2638 max mem: 3537 +Test: [1400/2387] eta: 0:03:52 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0451 (0.2976) acc1: 100.0000 (87.3610) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.4228 data: 0.3162 max mem: 3537 +Test: [1410/2387] eta: 0:03:52 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0465 (0.2964) acc1: 100.0000 (87.4347) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.6000 data: 0.4772 max mem: 3537 +Test: [1420/2387] eta: 0:03:50 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1633 (0.2955) acc1: 98.9583 (87.5090) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.4295 data: 0.2911 max mem: 3537 +Test: [1430/2387] eta: 0:03:49 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1778 (0.2948) acc1: 97.9167 (87.5766) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.3640 data: 0.2410 max mem: 3537 +Test: [1440/2387] eta: 0:03:47 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1849 (0.2941) acc1: 97.9167 (87.6458) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.4171 data: 0.2994 max mem: 3537 +Test: [1450/2387] eta: 0:03:45 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1804 (0.2935) acc1: 98.1771 (87.7064) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.3142 data: 0.1821 max mem: 3537 +Test: [1460/2387] eta: 0:03:43 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1627 (0.2926) acc1: 98.1771 (87.7798) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2551 data: 0.1172 max mem: 3537 +Test: [1470/2387] eta: 0:03:41 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1343 (0.2914) acc1: 98.9583 (87.8578) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2349 data: 0.1197 max mem: 3537 +Test: [1480/2387] eta: 0:03:38 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1236 (0.2903) acc1: 99.7396 (87.9363) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2251 data: 0.1318 max mem: 3537 +Test: [1490/2387] eta: 0:03:35 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1317 (0.2893) acc1: 99.7396 (88.0105) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1999 data: 0.1081 max mem: 3537 +Test: [1500/2387] eta: 0:03:33 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1336 (0.2882) acc1: 99.7396 (88.0852) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2102 data: 0.1181 max mem: 3537 +Test: [1510/2387] eta: 0:03:30 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1290 (0.2872) acc1: 99.7396 (88.1594) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2080 data: 0.1175 max mem: 3537 +Test: [1520/2387] eta: 0:03:27 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1574 (0.2865) acc1: 99.4792 (88.2302) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2084 data: 0.1183 max mem: 3537 +Test: [1530/2387] eta: 0:03:25 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1683 (0.2857) acc1: 99.4792 (88.3015) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2038 data: 0.1117 max mem: 3537 +Test: [1540/2387] eta: 0:03:22 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1570 (0.2848) acc1: 99.7396 (88.3752) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2150 data: 0.1190 max mem: 3537 +Test: [1550/2387] eta: 0:03:20 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1535 (0.2841) acc1: 99.7396 (88.4468) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2237 data: 0.1275 max mem: 3537 +Test: [1560/2387] eta: 0:03:17 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1606 (0.2834) acc1: 99.4792 (88.5175) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2210 data: 0.1259 max mem: 3537 +Test: [1570/2387] eta: 0:03:15 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1606 (0.2826) acc1: 99.4792 (88.5871) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2100 data: 0.1144 max mem: 3537 +Test: [1580/2387] eta: 0:03:12 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1737 (0.2820) acc1: 98.9583 (88.6385) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2042 data: 0.1098 max mem: 3538 +Test: [1590/2387] eta: 0:03:10 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1845 (0.2814) acc1: 96.3542 (88.6891) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2063 data: 0.1117 max mem: 3538 +Test: [1600/2387] eta: 0:03:07 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1663 (0.2808) acc1: 97.6562 (88.7465) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2005 data: 0.1072 max mem: 3538 +Test: [1610/2387] eta: 0:03:05 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1663 (0.2802) acc1: 97.9167 (88.8021) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2035 data: 0.1095 max mem: 3538 +Test: [1620/2387] eta: 0:03:02 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1958 (0.2797) acc1: 96.8750 (88.8467) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1974 data: 0.1018 max mem: 3538 +Test: [1630/2387] eta: 0:02:59 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2354 (0.2797) acc1: 95.0521 (88.8679) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1940 data: 0.1000 max mem: 3538 +Test: [1640/2387] eta: 0:02:57 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2831 (0.2800) acc1: 91.6667 (88.8686) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.2059 data: 0.1133 max mem: 3538 +Test: [1650/2387] eta: 0:02:54 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2904 (0.2800) acc1: 91.1458 (88.8906) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.2049 data: 0.1126 max mem: 3538 +Test: [1660/2387] eta: 0:02:52 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.3125 (0.2802) acc1: 90.3646 (88.8886) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1956 data: 0.1038 max mem: 3538 +Test: [1670/2387] eta: 0:02:49 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.3123 (0.2804) acc1: 89.8438 (88.8996) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1899 data: 0.0991 max mem: 3538 +Test: [1680/2387] eta: 0:02:47 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2957 (0.2805) acc1: 91.1458 (88.9158) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1903 data: 0.0990 max mem: 3538 +Test: [1690/2387] eta: 0:02:44 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2999 (0.2807) acc1: 91.1458 (88.9251) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1908 data: 0.0993 max mem: 3538 +Test: [1700/2387] eta: 0:02:42 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2925 (0.2804) acc1: 91.4062 (88.9625) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1902 data: 0.0967 max mem: 3538 +Test: [1710/2387] eta: 0:02:39 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1097 (0.2793) acc1: 99.7396 (89.0257) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1850 data: 0.0914 max mem: 3538 +Test: [1720/2387] eta: 0:02:36 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0938 (0.2782) acc1: 100.0000 (89.0891) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1789 data: 0.0866 max mem: 3538 +Test: [1730/2387] eta: 0:02:34 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0931 (0.2772) acc1: 100.0000 (89.1516) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1817 data: 0.0858 max mem: 3538 +Test: [1740/2387] eta: 0:02:31 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0976 (0.2762) acc1: 100.0000 (89.2130) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1794 data: 0.0826 max mem: 3538 +Test: [1750/2387] eta: 0:02:29 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0982 (0.2752) acc1: 100.0000 (89.2740) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1776 data: 0.0838 max mem: 3538 +Test: [1760/2387] eta: 0:02:26 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1003 (0.2742) acc1: 100.0000 (89.3345) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1696 data: 0.0785 max mem: 3538 +Test: [1770/2387] eta: 0:02:24 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1192 (0.2735) acc1: 99.7396 (89.3888) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1710 data: 0.0802 max mem: 3538 +Test: [1780/2387] eta: 0:02:21 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1354 (0.2727) acc1: 99.4792 (89.4475) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1778 data: 0.0857 max mem: 3538 +Test: [1790/2387] eta: 0:02:18 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1244 (0.2718) acc1: 100.0000 (89.5045) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1696 data: 0.0782 max mem: 3538 +Test: [1800/2387] eta: 0:02:16 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1231 (0.2710) acc1: 99.7396 (89.5614) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1768 data: 0.0863 max mem: 3538 +Test: [1810/2387] eta: 0:02:14 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1156 (0.2701) acc1: 99.7396 (89.6177) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1856 data: 0.0941 max mem: 3539 +Test: [1820/2387] eta: 0:02:11 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1085 (0.2693) acc1: 99.7396 (89.6734) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1802 data: 0.0883 max mem: 3539 +Test: [1830/2387] eta: 0:02:09 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1190 (0.2685) acc1: 100.0000 (89.7293) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1838 data: 0.0911 max mem: 3539 +Test: [1840/2387] eta: 0:02:06 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1150 (0.2676) acc1: 100.0000 (89.7849) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1867 data: 0.0918 max mem: 3539 +Test: [1850/2387] eta: 0:02:04 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1138 (0.2668) acc1: 100.0000 (89.8397) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2142 data: 0.1206 max mem: 3539 +Test: [1860/2387] eta: 0:02:01 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1153 (0.2660) acc1: 100.0000 (89.8943) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2234 data: 0.1306 max mem: 3539 +Test: [1870/2387] eta: 0:01:59 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1163 (0.2652) acc1: 100.0000 (89.9479) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2015 data: 0.1074 max mem: 3539 +Test: [1880/2387] eta: 0:01:57 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1143 (0.2644) acc1: 100.0000 (90.0010) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2025 data: 0.1068 max mem: 3539 +Test: [1890/2387] eta: 0:01:54 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1135 (0.2636) acc1: 100.0000 (90.0538) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1780 data: 0.0843 max mem: 3539 +Test: [1900/2387] eta: 0:01:52 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1153 (0.2628) acc1: 100.0000 (90.1055) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1599 data: 0.0696 max mem: 3539 +Test: [1910/2387] eta: 0:01:49 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1293 (0.2622) acc1: 100.0000 (90.1553) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1839 data: 0.0927 max mem: 3539 +Test: [1920/2387] eta: 0:01:47 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1512 (0.2618) acc1: 99.7396 (90.1927) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2014 data: 0.0965 max mem: 3539 +Test: [1930/2387] eta: 0:01:44 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1292 (0.2610) acc1: 100.0000 (90.2435) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1815 data: 0.0690 max mem: 3539 +Test: [1940/2387] eta: 0:01:42 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1173 (0.2605) acc1: 100.0000 (90.2919) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2084 data: 0.1044 max mem: 3539 +Test: [1950/2387] eta: 0:01:41 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1402 (0.2599) acc1: 100.0000 (90.3374) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.3882 data: 0.2874 max mem: 3539 +Test: [1960/2387] eta: 0:01:39 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1478 (0.2595) acc1: 99.7396 (90.3748) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.5252 data: 0.4229 max mem: 3539 +Test: [1970/2387] eta: 0:01:37 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2088 (0.2597) acc1: 97.1354 (90.3680) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.4198 data: 0.3178 max mem: 3539 +Test: [1980/2387] eta: 0:01:35 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2178 (0.2594) acc1: 95.3125 (90.3989) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.4006 data: 0.2746 max mem: 3539 +Test: [1990/2387] eta: 0:01:34 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1476 (0.2589) acc1: 98.4375 (90.4398) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.5849 data: 0.4531 max mem: 3539 +Test: [2000/2387] eta: 0:01:32 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1472 (0.2583) acc1: 98.4375 (90.4807) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.6346 data: 0.5184 max mem: 3539 +Test: [2010/2387] eta: 0:01:30 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1478 (0.2578) acc1: 98.4375 (90.5209) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.5704 data: 0.4339 max mem: 3539 +Test: [2020/2387] eta: 0:01:28 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1474 (0.2572) acc1: 98.9583 (90.5617) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.5400 data: 0.4130 max mem: 3539 +Test: [2030/2387] eta: 0:01:26 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1453 (0.2567) acc1: 98.9583 (90.6015) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.5599 data: 0.4493 max mem: 3540 +Test: [2040/2387] eta: 0:01:24 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1465 (0.2561) acc1: 98.9583 (90.6418) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.5247 data: 0.4032 max mem: 3540 +Test: [2050/2387] eta: 0:01:22 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1467 (0.2556) acc1: 98.9583 (90.6828) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.5093 data: 0.3966 max mem: 3540 +Test: [2060/2387] eta: 0:01:20 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1455 (0.2551) acc1: 98.6979 (90.7215) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.5943 data: 0.4805 max mem: 3540 +Test: [2070/2387] eta: 0:01:18 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1447 (0.2545) acc1: 98.6979 (90.7613) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.5598 data: 0.4382 max mem: 3540 +Test: [2080/2387] eta: 0:01:16 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1439 (0.2540) acc1: 98.6979 (90.7989) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.5153 data: 0.3887 max mem: 3540 +Test: [2090/2387] eta: 0:01:14 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1458 (0.2535) acc1: 98.6979 (90.8378) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.5814 data: 0.4555 max mem: 3540 +Test: [2100/2387] eta: 0:01:12 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1465 (0.2530) acc1: 98.9583 (90.8761) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.6093 data: 0.4843 max mem: 3540 +Test: [2110/2387] eta: 0:01:10 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1436 (0.2525) acc1: 98.6979 (90.9133) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.6082 data: 0.4948 max mem: 3540 +Test: [2120/2387] eta: 0:01:08 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1485 (0.2523) acc1: 98.4375 (90.9329) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.6048 data: 0.4909 max mem: 3540 +Test: [2130/2387] eta: 0:01:06 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1675 (0.2519) acc1: 97.1354 (90.9631) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.5199 data: 0.3813 max mem: 3540 +Test: [2140/2387] eta: 0:01:04 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1699 (0.2516) acc1: 98.1771 (90.9910) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5300 data: 0.3859 max mem: 3540 +Test: [2150/2387] eta: 0:01:01 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1878 (0.2514) acc1: 97.1354 (91.0213) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.5098 data: 0.3642 max mem: 3540 +Test: [2160/2387] eta: 0:00:59 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1805 (0.2511) acc1: 97.3958 (91.0540) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.4598 data: 0.3269 max mem: 3540 +Test: [2170/2387] eta: 0:00:56 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2255 (0.2512) acc1: 95.5729 (91.0585) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.5014 data: 0.3936 max mem: 3540 +Test: [2180/2387] eta: 0:00:54 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2985 (0.2523) acc1: 91.4062 (90.9858) acc5: 100.0000 (100.0000) acc1_neg: 46.094 acc5_neg: 100.000 time: 0.5048 data: 0.3921 max mem: 3540 +Test: [2190/2387] eta: 0:00:52 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2822 (0.2526) acc1: 93.4896 (90.9809) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5484 data: 0.4089 max mem: 3540 +Test: [2200/2387] eta: 0:00:49 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1488 (0.2520) acc1: 97.1354 (91.0145) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.5599 data: 0.4352 max mem: 3540 +Test: [2210/2387] eta: 0:00:47 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1019 (0.2513) acc1: 99.7396 (91.0541) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5698 data: 0.4603 max mem: 3540 +Test: [2220/2387] eta: 0:00:44 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0984 (0.2506) acc1: 100.0000 (91.0925) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5351 data: 0.4200 max mem: 3540 +Test: [2230/2387] eta: 0:00:42 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1023 (0.2500) acc1: 100.0000 (91.1314) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.4950 data: 0.3677 max mem: 3540 +Test: [2240/2387] eta: 0:00:39 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1057 (0.2494) acc1: 99.7396 (91.1651) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.4804 data: 0.3541 max mem: 3540 +Test: [2250/2387] eta: 0:00:37 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1166 (0.2489) acc1: 99.7396 (91.2013) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.4897 data: 0.3687 max mem: 3540 +Test: [2260/2387] eta: 0:00:34 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1447 (0.2485) acc1: 99.7396 (91.2366) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5349 data: 0.4160 max mem: 3541 +Test: [2270/2387] eta: 0:00:32 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1332 (0.2479) acc1: 99.7396 (91.2733) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.4909 data: 0.3704 max mem: 3541 +Test: [2280/2387] eta: 0:00:29 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1201 (0.2474) acc1: 100.0000 (91.3074) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.5290 data: 0.4092 max mem: 3541 +Test: [2290/2387] eta: 0:00:26 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1151 (0.2468) acc1: 100.0000 (91.3437) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.5591 data: 0.4556 max mem: 3541 +Test: [2300/2387] eta: 0:00:24 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1346 (0.2464) acc1: 99.7396 (91.3782) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.5359 data: 0.4129 max mem: 3541 +Test: [2310/2387] eta: 0:00:21 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1498 (0.2461) acc1: 99.2188 (91.4065) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.5146 data: 0.3754 max mem: 3541 +Test: [2320/2387] eta: 0:00:18 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1722 (0.2459) acc1: 98.1771 (91.4286) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.4989 data: 0.3658 max mem: 3541 +Test: [2330/2387] eta: 0:00:15 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2003 (0.2459) acc1: 96.3542 (91.4450) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.5049 data: 0.3686 max mem: 3541 +Test: [2340/2387] eta: 0:00:13 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1910 (0.2456) acc1: 96.3542 (91.4723) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.4898 data: 0.3570 max mem: 3541 +Test: [2350/2387] eta: 0:00:10 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1619 (0.2455) acc1: 98.9583 (91.4923) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.4948 data: 0.3621 max mem: 3541 +Test: [2360/2387] eta: 0:00:07 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2060 (0.2455) acc1: 96.8750 (91.5081) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.5849 data: 0.4589 max mem: 3541 +Test: [2370/2387] eta: 0:00:04 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2159 (0.2456) acc1: 96.3542 (91.5066) acc5: 100.0000 (100.0000) acc1_neg: 68.229 acc5_neg: 100.000 time: 0.5799 data: 0.4434 max mem: 3541 +Test: [2380/2387] eta: 0:00:02 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2396 (0.2457) acc1: 92.4479 (91.5134) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.5199 data: 0.3709 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.3021 (0.2459) acc1: 90.8854 (91.5170) acc5: 100.0000 (100.0000) acc1_neg: 88.372 acc5_neg: 100.000 time: 0.4601 data: 0.3321 max mem: 3541 +Test: Total time: 0:11:22 (0.2860 s / it) +* Overall Acc@1 91.517 Acc@5 100.000 + target == real Acc@1 24.472 Acc@5 100.000 + target == fake Acc@1 96.225 Acc@5 100.000 + loss 0.246 + PR-AUC 0.9903 ROC-AUC 0.8769 EER 19.96% (thr=0.7710) +Accuracy of the network on the 916267 test images: 91.5% +[rank0]:[W725 17:30:19.852734274 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 17:30:50.941000 2564489 torch/distributed/run.py:766] +W0725 17:30:50.941000 2564489 torch/distributed/run.py:766] ***************************************** +W0725 17:30:50.941000 2564489 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:30:50.941000 2564489 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:31:26.758990782 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:31:26.070982001 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 17:31:27.031023981 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:31:28.720178866 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +[rank6]:[W725 17:31:28.857873806 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:31:28.869827585 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W725 17:31:28.869847660 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:31:28.046161747 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed1_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed1_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 3:15:38 acc1_pos: 26.302 acc5_pos: 100.000 loss: 1.1336 (1.1336) acc1: 26.3021 (26.3021) acc5: 100.0000 (100.0000) time: 22.7055 data: 17.0110 max mem: 3531 +Test: [ 10/517] eta: 0:18:15 acc1_pos: 12.240 acc5_pos: 100.000 loss: 1.1401 (1.1180) acc1: 21.0938 (22.8693) acc5: 100.0000 (100.0000) time: 2.1606 data: 1.5470 max mem: 3531 +Test: [ 20/517] eta: 0:10:07 acc1_pos: 24.219 acc5_pos: 100.000 loss: 1.1401 (1.0901) acc1: 19.7917 (24.6900) acc5: 100.0000 (100.0000) time: 0.1485 data: 0.0362 max mem: 3531 +Test: [ 30/517] eta: 0:09:04 acc1_pos: 15.104 acc5_pos: 100.000 loss: 1.1086 (1.0862) acc1: 19.0104 (24.6724) acc5: 100.0000 (100.0000) time: 0.5449 data: 0.4269 max mem: 3531 +Test: [ 40/517] eta: 0:07:50 acc1_pos: 17.969 acc5_pos: 100.000 loss: 1.1086 (1.0875) acc1: 18.7500 (23.0818) acc5: 100.0000 (100.0000) time: 0.7394 data: 0.6238 max mem: 3531 +Test: [ 50/517] eta: 0:07:20 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.9860 (1.0462) acc1: 20.3125 (26.8689) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.6748 data: 0.5592 max mem: 3531 +Test: [ 60/517] eta: 0:06:48 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.4359 (0.9405) acc1: 77.0833 (36.5010) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.6999 data: 0.5799 max mem: 3531 +Test: [ 70/517] eta: 0:06:29 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.4021 (0.8653) acc1: 88.8021 (43.8637) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.6849 data: 0.5556 max mem: 3531 +Test: [ 80/517] eta: 0:06:09 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.4174 (0.8127) acc1: 86.1979 (48.8265) acc5: 100.0000 (100.0000) acc1_neg: 72.396 acc5_neg: 100.000 time: 0.7048 data: 0.5895 max mem: 3531 +Test: [ 90/517] eta: 0:05:54 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3973 (0.7653) acc1: 89.3229 (53.6659) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.6799 data: 0.5692 max mem: 3531 +Test: [100/517] eta: 0:05:35 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3770 (0.7270) acc1: 90.8854 (57.2710) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.6309 data: 0.5014 max mem: 3531 +Test: [110/517] eta: 0:05:19 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3499 (0.6907) acc1: 90.8854 (60.4824) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.5855 data: 0.4726 max mem: 3531 +Test: [120/517] eta: 0:05:09 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3527 (0.6651) acc1: 91.9271 (62.9498) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.6539 data: 0.5295 max mem: 3531 +Test: [130/517] eta: 0:04:59 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3874 (0.6428) acc1: 89.5833 (64.9889) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.7147 data: 0.5722 max mem: 3531 +Test: [140/517] eta: 0:04:45 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.2529 (0.6143) acc1: 99.4792 (67.4664) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.6300 data: 0.5024 max mem: 3531 +Test: [150/517] eta: 0:04:35 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.2461 (0.5910) acc1: 100.0000 (69.5261) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.5947 data: 0.4749 max mem: 3531 +Test: [160/517] eta: 0:04:27 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.2641 (0.5701) acc1: 98.4375 (71.3315) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.6848 data: 0.5585 max mem: 3531 +Test: [170/517] eta: 0:04:18 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.2594 (0.5512) acc1: 98.4375 (72.9075) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.6947 data: 0.5588 max mem: 3531 +Test: [180/517] eta: 0:04:10 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.2606 (0.5356) acc1: 98.6979 (74.3065) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.6899 data: 0.5705 max mem: 3531 +Test: [190/517] eta: 0:04:00 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.2769 (0.5220) acc1: 98.6979 (75.5836) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.6748 data: 0.5703 max mem: 3531 +Test: [200/517] eta: 0:03:52 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.2686 (0.5086) acc1: 98.9583 (76.7322) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.6499 data: 0.5416 max mem: 3531 +Test: [210/517] eta: 0:03:43 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3083 (0.5048) acc1: 95.8333 (77.3326) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.6349 data: 0.5198 max mem: 3532 +Test: [220/517] eta: 0:03:35 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.4304 (0.5014) acc1: 88.2812 (77.8717) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.6349 data: 0.5246 max mem: 3532 +Test: [230/517] eta: 0:03:26 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.4304 (0.4987) acc1: 89.3229 (78.2896) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.6499 data: 0.5340 max mem: 3532 +Test: [240/517] eta: 0:03:19 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.4125 (0.4951) acc1: 90.1042 (78.8295) acc5: 100.0000 (100.0000) acc1_neg: 77.083 acc5_neg: 100.000 time: 0.6599 data: 0.5389 max mem: 3532 +Test: [250/517] eta: 0:03:11 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3776 (0.4880) acc1: 95.0521 (79.5101) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.6898 data: 0.5618 max mem: 3532 +Test: [260/517] eta: 0:03:04 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.2807 (0.4813) acc1: 99.4792 (80.1555) acc5: 100.0000 (100.0000) acc1_neg: 77.083 acc5_neg: 100.000 time: 0.7299 data: 0.5851 max mem: 3532 +Test: [270/517] eta: 0:02:57 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.4796 (0.4824) acc1: 79.4271 (80.0594) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.7252 data: 0.5888 max mem: 3532 +Test: [280/517] eta: 0:02:49 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.4935 (0.4810) acc1: 78.3854 (80.3177) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.6450 data: 0.5204 max mem: 3532 +Test: [290/517] eta: 0:02:42 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3589 (0.4763) acc1: 96.3542 (80.8858) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.6646 data: 0.5465 max mem: 3532 +Test: [300/517] eta: 0:02:34 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3589 (0.4734) acc1: 95.5729 (81.2985) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.7098 data: 0.6012 max mem: 3532 +Test: [310/517] eta: 0:02:27 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3970 (0.4705) acc1: 91.9271 (81.6561) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.6649 data: 0.5646 max mem: 3532 +Test: [320/517] eta: 0:02:20 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3613 (0.4651) acc1: 92.1875 (82.1059) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.7050 data: 0.5890 max mem: 3532 +Test: [330/517] eta: 0:02:12 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.2441 (0.4583) acc1: 99.4792 (82.6292) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.6902 data: 0.5637 max mem: 3532 +Test: [340/517] eta: 0:02:05 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.2314 (0.4544) acc1: 99.2188 (82.9232) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.6132 data: 0.5030 max mem: 3532 +Test: [350/517] eta: 0:01:58 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.4805 (0.4560) acc1: 83.8542 (82.8993) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.6930 data: 0.5738 max mem: 3532 +Test: [360/517] eta: 0:01:51 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.4901 (0.4551) acc1: 83.8542 (83.0130) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.7257 data: 0.6065 max mem: 3532 +Test: [370/517] eta: 0:01:44 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.2828 (0.4501) acc1: 99.2188 (83.4576) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.6755 data: 0.5771 max mem: 3532 +Test: [380/517] eta: 0:01:37 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.2941 (0.4474) acc1: 99.2188 (83.8015) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.7190 data: 0.6158 max mem: 3532 +Test: [390/517] eta: 0:01:29 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3638 (0.4471) acc1: 92.9688 (83.8562) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.6643 data: 0.5568 max mem: 3532 +Test: [400/517] eta: 0:01:22 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3037 (0.4429) acc1: 96.0938 (84.2328) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.6349 data: 0.5254 max mem: 3532 +Test: [410/517] eta: 0:01:15 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.2967 (0.4409) acc1: 98.9583 (84.5492) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.6949 data: 0.5749 max mem: 3532 +Test: [420/517] eta: 0:01:08 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3352 (0.4381) acc1: 98.9583 (84.8958) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.6899 data: 0.5798 max mem: 3532 +Test: [430/517] eta: 0:01:01 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3162 (0.4352) acc1: 99.4792 (85.1738) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.8248 data: 0.7184 max mem: 3532 +Test: [440/517] eta: 0:00:54 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3756 (0.4349) acc1: 96.3542 (85.3074) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.7603 data: 0.6310 max mem: 3533 +Test: [450/517] eta: 0:00:47 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3955 (0.4336) acc1: 93.2292 (85.5154) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.5752 data: 0.4350 max mem: 3533 +Test: [460/517] eta: 0:00:39 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3812 (0.4327) acc1: 94.7917 (85.6771) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.5445 data: 0.4194 max mem: 3533 +Test: [470/517] eta: 0:00:32 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3712 (0.4304) acc1: 95.5729 (85.9248) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.5055 data: 0.3869 max mem: 3533 +Test: [480/517] eta: 0:00:25 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3016 (0.4270) acc1: 96.3542 (86.1043) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.4749 data: 0.3531 max mem: 3533 +Test: [490/517] eta: 0:00:18 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3016 (0.4247) acc1: 95.0521 (86.2785) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.4240 data: 0.2866 max mem: 3533 +Test: [500/517] eta: 0:00:11 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.4667 (0.4277) acc1: 83.0729 (85.9250) acc5: 100.0000 (100.0000) acc1_neg: 56.510 acc5_neg: 100.000 time: 0.3605 data: 0.2153 max mem: 3533 +Test: [510/517] eta: 0:00:04 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.5331 (0.4280) acc1: 72.6562 (85.9013) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.3399 data: 0.2020 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3862 (0.4268) acc1: 86.7188 (86.0402) acc5: 100.0000 (100.0000) acc1_neg: 99.647 acc5_neg: 100.000 time: 0.3198 data: 0.1950 max mem: 3533 +Test: Total time: 0:05:45 (0.6674 s / it) +* Overall Acc@1 86.040 Acc@5 100.000 + target == real Acc@1 24.055 Acc@5 100.000 + target == fake Acc@1 92.470 Acc@5 100.000 + loss 0.427 + PR-AUC 0.9417 ROC-AUC 0.6594 EER 38.77% (thr=0.6914) +Accuracy of the network on the 198427 test images: 86.0% +[rank0]:[W725 17:37:29.223909998 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0725 17:38:06.200000 2565136 torch/distributed/run.py:766] +W0725 17:38:06.200000 2565136 torch/distributed/run.py:766] ***************************************** +W0725 17:38:06.200000 2565136 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:38:06.200000 2565136 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:38:42.470788014 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:38:44.817282533 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:38:44.108757637 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 17:38:44.338802958 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:38:44.559834767 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +[rank3]:[W725 17:38:45.668351923 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:38:45.728808263 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:38:45.950770742 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed1_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed1_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 1:43:32 acc1_pos: 24.479 acc5_pos: 100.000 loss: 0.9853 (0.9853) acc1: 24.4792 (24.4792) acc5: 100.0000 (100.0000) time: 15.9704 data: 12.3976 max mem: 3531 +Test: [ 10/389] eta: 0:09:57 acc1_pos: 82.812 acc5_pos: 100.000 loss: 0.7133 (0.8052) acc1: 48.4375 (48.6269) acc5: 100.0000 (100.0000) time: 1.5768 data: 1.1275 max mem: 3531 +Test: [ 20/389] eta: 0:05:41 acc1_pos: 6.510 acc5_pos: 100.000 loss: 0.7133 (0.8309) acc1: 45.5729 (44.7297) acc5: 100.0000 (100.0000) time: 0.1737 data: 0.0490 max mem: 3531 +Test: [ 30/389] eta: 0:05:09 acc1_pos: 48.958 acc5_pos: 100.000 loss: 0.8546 (0.8399) acc1: 22.1354 (39.4909) acc5: 100.0000 (100.0000) time: 0.4699 data: 0.3520 max mem: 3531 +Test: [ 40/389] eta: 0:04:36 acc1_pos: 7.031 acc5_pos: 100.000 loss: 0.8589 (0.8490) acc1: 23.6979 (37.6905) acc5: 100.0000 (100.0000) time: 0.6510 data: 0.5354 max mem: 3531 +Test: [ 50/389] eta: 0:04:10 acc1_pos: 90.365 acc5_pos: 100.000 loss: 0.8589 (0.8429) acc1: 25.0000 (39.7518) acc5: 100.0000 (100.0000) time: 0.5449 data: 0.4347 max mem: 3531 +Test: [ 60/389] eta: 0:03:50 acc1_pos: 3.385 acc5_pos: 100.000 loss: 0.7465 (0.8108) acc1: 48.6979 (43.8397) acc5: 100.0000 (100.0000) time: 0.5140 data: 0.3917 max mem: 3531 +Test: [ 70/389] eta: 0:03:36 acc1_pos: 53.906 acc5_pos: 100.000 loss: 0.7664 (0.8397) acc1: 47.1354 (41.2962) acc5: 100.0000 (100.0000) time: 0.5248 data: 0.3983 max mem: 3531 +Test: [ 80/389] eta: 0:03:25 acc1_pos: 77.604 acc5_pos: 100.000 loss: 1.0543 (0.8515) acc1: 10.1562 (40.5961) acc5: 100.0000 (100.0000) time: 0.5547 data: 0.4368 max mem: 3531 +Test: [ 90/389] eta: 0:03:10 acc1_pos: 13.021 acc5_pos: 100.000 loss: 0.6499 (0.8281) acc1: 57.5521 (42.5166) acc5: 100.0000 (100.0000) time: 0.4855 data: 0.3697 max mem: 3531 +Test: [100/389] eta: 0:02:58 acc1_pos: 21.354 acc5_pos: 100.000 loss: 0.7771 (0.8293) acc1: 41.4062 (41.8472) acc5: 100.0000 (100.0000) time: 0.4306 data: 0.3156 max mem: 3531 +Test: [110/389] eta: 0:02:50 acc1_pos: 99.479 acc5_pos: 100.000 loss: 0.8251 (0.8270) acc1: 34.6354 (42.1077) acc5: 100.0000 (100.0000) time: 0.4942 data: 0.3833 max mem: 3531 +Test: [120/389] eta: 0:02:39 acc1_pos: 65.885 acc5_pos: 100.000 loss: 0.7818 (0.8217) acc1: 40.6250 (42.7320) acc5: 100.0000 (100.0000) time: 0.4646 data: 0.3566 max mem: 3531 +Test: [130/389] eta: 0:02:31 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.6056 (0.8044) acc1: 65.1042 (44.5670) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.4450 data: 0.3310 max mem: 3531 +Test: [140/389] eta: 0:02:24 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.4371 (0.7782) acc1: 86.1979 (47.4217) acc5: 100.0000 (100.0000) acc1_neg: 44.792 acc5_neg: 100.000 time: 0.5103 data: 0.3997 max mem: 3531 +Test: [150/389] eta: 0:02:16 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.4091 (0.7646) acc1: 91.1458 (49.0635) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.4847 data: 0.3755 max mem: 3531 +Test: [160/389] eta: 0:02:08 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.4368 (0.7415) acc1: 86.9792 (51.5253) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.4092 data: 0.2911 max mem: 3531 +Test: [170/389] eta: 0:02:01 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.3589 (0.7210) acc1: 97.6562 (54.0753) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.4314 data: 0.3174 max mem: 3531 +Test: [180/389] eta: 0:01:54 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.3614 (0.7068) acc1: 95.0521 (55.8227) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.4644 data: 0.3470 max mem: 3531 +Test: [190/389] eta: 0:01:49 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.4510 (0.6958) acc1: 89.5833 (57.2303) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.5084 data: 0.3938 max mem: 3531 +Test: [200/389] eta: 0:01:42 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.4454 (0.6835) acc1: 90.3646 (58.7155) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.4949 data: 0.3883 max mem: 3531 +Test: [210/389] eta: 0:01:35 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.4032 (0.6694) acc1: 94.2708 (60.2019) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.3998 data: 0.2834 max mem: 3532 +Test: [220/389] eta: 0:01:30 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.2539 (0.6519) acc1: 99.4792 (61.9532) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.4637 data: 0.3274 max mem: 3532 +Test: [230/389] eta: 0:01:24 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.3202 (0.6410) acc1: 98.4375 (63.3917) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.5037 data: 0.3743 max mem: 3532 +Test: [240/389] eta: 0:01:18 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.3878 (0.6303) acc1: 96.6146 (64.6395) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.4106 data: 0.3003 max mem: 3532 +Test: [250/389] eta: 0:01:12 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.3472 (0.6183) acc1: 98.4375 (66.0141) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.3648 data: 0.2548 max mem: 3532 +Test: [260/389] eta: 0:01:06 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.3433 (0.6146) acc1: 99.2188 (66.5858) acc5: 100.0000 (100.0000) acc1_neg: 25.521 acc5_neg: 100.000 time: 0.4241 data: 0.3110 max mem: 3532 +Test: [270/389] eta: 0:01:02 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.4932 (0.6181) acc1: 75.5208 (66.0642) acc5: 100.0000 (100.0000) acc1_neg: 75.521 acc5_neg: 100.000 time: 0.5499 data: 0.4323 max mem: 3532 +Test: [280/389] eta: 0:00:55 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.6467 (0.6186) acc1: 59.3750 (65.9762) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.4309 data: 0.3206 max mem: 3532 +Test: [290/389] eta: 0:00:50 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.6055 (0.6133) acc1: 69.5312 (66.6497) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.3407 data: 0.2329 max mem: 3532 +Test: [300/389] eta: 0:00:45 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.4189 (0.6074) acc1: 92.4479 (67.5569) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.4237 data: 0.3063 max mem: 3532 +Test: [310/389] eta: 0:00:39 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.4811 (0.6079) acc1: 85.9375 (67.5250) acc5: 100.0000 (100.0000) acc1_neg: 54.948 acc5_neg: 100.000 time: 0.3240 data: 0.1947 max mem: 3532 +Test: [320/389] eta: 0:00:33 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.4719 (0.6018) acc1: 85.4167 (68.2162) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2413 data: 0.0949 max mem: 3532 +Test: [330/389] eta: 0:00:28 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.4693 (0.5981) acc1: 89.5833 (68.8208) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.2855 data: 0.1351 max mem: 3532 +Test: [340/389] eta: 0:00:23 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.4287 (0.5893) acc1: 91.1458 (69.6351) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2986 data: 0.1560 max mem: 3532 +Test: [350/389] eta: 0:00:18 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.2791 (0.5967) acc1: 97.6562 (68.9444) acc5: 100.0000 (100.0000) acc1_neg: 2.344 acc5_neg: 100.000 time: 0.2594 data: 0.1255 max mem: 3532 +Test: [360/389] eta: 0:00:13 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.6213 (0.6003) acc1: 61.7188 (68.4788) acc5: 100.0000 (100.0000) acc1_neg: 70.573 acc5_neg: 100.000 time: 0.2705 data: 0.1267 max mem: 3532 +Test: [370/389] eta: 0:00:08 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.5822 (0.5960) acc1: 68.2292 (68.8321) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.3018 data: 0.1369 max mem: 3532 +Test: [380/389] eta: 0:00:04 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.3976 (0.5914) acc1: 82.2917 (69.1403) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2574 data: 0.0716 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.3344 (0.5880) acc1: 88.8021 (69.5115) acc5: 100.0000 (100.0000) acc1_neg: 74.863 acc5_neg: 100.000 time: 0.1883 data: 0.0002 max mem: 3532 +Test: Total time: 0:02:55 (0.4506 s / it) +* Overall Acc@1 69.511 Acc@5 100.000 + target == real Acc@1 42.851 Acc@5 100.000 + target == fake Acc@1 81.904 Acc@5 100.000 + loss 0.588 + PR-AUC 0.8086 ROC-AUC 0.6924 EER 35.78% (thr=0.6147) +Accuracy of the network on the 149358 test images: 69.5% +[rank0]:[W725 17:41:57.514456233 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 17:42:17.699000 2565811 torch/distributed/run.py:766] +W0725 17:42:17.699000 2565811 torch/distributed/run.py:766] ***************************************** +W0725 17:42:17.699000 2565811 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:42:17.699000 2565811 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W725 17:42:50.075469426 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:42:51.649459031 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:42:51.995410278 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:42:51.256275699 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:42:51.526642653 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:42:51.559219262 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:42:52.620676546 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:42:52.853719500 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed1_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed1_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 1:19:43 acc1_pos: 80.208 acc5_pos: 100.000 loss: 0.4223 (0.4223) acc1: 80.2083 (80.2083) acc5: 100.0000 (100.0000) time: 14.3660 data: 11.5141 max mem: 3531 +Test: [ 10/333] eta: 0:07:30 acc1_pos: 82.812 acc5_pos: 100.000 loss: 0.3405 (0.3341) acc1: 87.5000 (87.2159) acc5: 100.0000 (100.0000) time: 1.3959 data: 1.0472 max mem: 3531 +Test: [ 20/333] eta: 0:04:03 acc1_pos: 92.708 acc5_pos: 100.000 loss: 0.3340 (0.3332) acc1: 87.2396 (86.0987) acc5: 100.0000 (100.0000) time: 0.0993 data: 0.0005 max mem: 3531 +Test: [ 30/333] eta: 0:03:45 acc1_pos: 40.365 acc5_pos: 100.000 loss: 0.4042 (0.4939) acc1: 72.1354 (73.1183) acc5: 100.0000 (100.0000) time: 0.3848 data: 0.2691 max mem: 3531 +Test: [ 40/333] eta: 0:03:32 acc1_pos: 45.573 acc5_pos: 100.000 loss: 0.8626 (0.5783) acc1: 42.4479 (66.1331) acc5: 100.0000 (100.0000) time: 0.6718 data: 0.5481 max mem: 3531 +Test: [ 50/333] eta: 0:03:24 acc1_pos: 49.740 acc5_pos: 100.000 loss: 0.8460 (0.6316) acc1: 43.7500 (61.7392) acc5: 100.0000 (100.0000) time: 0.6899 data: 0.5752 max mem: 3531 +Test: [ 60/333] eta: 0:03:07 acc1_pos: 33.333 acc5_pos: 100.000 loss: 0.8429 (0.6650) acc1: 45.8333 (59.1359) acc5: 100.0000 (100.0000) time: 0.6030 data: 0.4940 max mem: 3531 +Test: [ 70/333] eta: 0:02:55 acc1_pos: 37.500 acc5_pos: 100.000 loss: 0.8429 (0.6902) acc1: 46.6146 (57.4090) acc5: 100.0000 (100.0000) time: 0.5299 data: 0.4185 max mem: 3531 +Test: [ 80/333] eta: 0:02:46 acc1_pos: 47.135 acc5_pos: 100.000 loss: 0.8195 (0.7097) acc1: 44.5312 (55.9992) acc5: 100.0000 (100.0000) time: 0.5706 data: 0.4370 max mem: 3531 +Test: [ 90/333] eta: 0:02:38 acc1_pos: 53.646 acc5_pos: 100.000 loss: 0.7761 (0.7156) acc1: 51.3021 (55.5746) acc5: 100.0000 (100.0000) time: 0.6006 data: 0.4521 max mem: 3531 +Test: [100/333] eta: 0:02:31 acc1_pos: 58.333 acc5_pos: 100.000 loss: 0.7711 (0.7243) acc1: 53.3854 (55.0278) acc5: 100.0000 (100.0000) time: 0.6192 data: 0.4752 max mem: 3531 +Test: [110/333] eta: 0:02:27 acc1_pos: 52.083 acc5_pos: 100.000 loss: 0.7694 (0.7287) acc1: 51.5625 (54.6734) acc5: 100.0000 (100.0000) time: 0.6859 data: 0.5667 max mem: 3531 +Test: [120/333] eta: 0:02:21 acc1_pos: 42.708 acc5_pos: 100.000 loss: 0.7685 (0.7321) acc1: 51.5625 (54.3862) acc5: 100.0000 (100.0000) time: 0.7349 data: 0.6242 max mem: 3531 +Test: [130/333] eta: 0:02:15 acc1_pos: 75.260 acc5_pos: 100.000 loss: 0.7581 (0.7333) acc1: 50.7812 (54.4092) acc5: 100.0000 (100.0000) time: 0.7136 data: 0.6015 max mem: 3531 +Test: [140/333] eta: 0:02:07 acc1_pos: 67.188 acc5_pos: 100.000 loss: 0.7423 (0.7342) acc1: 52.3438 (54.2978) acc5: 100.0000 (100.0000) time: 0.6457 data: 0.5257 max mem: 3531 +Test: [150/333] eta: 0:02:01 acc1_pos: 54.948 acc5_pos: 100.000 loss: 0.7593 (0.7383) acc1: 51.3021 (53.8114) acc5: 100.0000 (100.0000) time: 0.6394 data: 0.5044 max mem: 3531 +Test: [160/333] eta: 0:01:54 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.7842 (0.7441) acc1: 48.9583 (53.5019) acc5: 100.0000 (100.0000) acc1_neg: 52.344 acc5_neg: 100.000 time: 0.6490 data: 0.5147 max mem: 3531 +Test: [170/333] eta: 0:01:47 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.8174 (0.7475) acc1: 49.2188 (53.4555) acc5: 100.0000 (100.0000) acc1_neg: 59.375 acc5_neg: 100.000 time: 0.6149 data: 0.4841 max mem: 3531 +Test: [180/333] eta: 0:01:40 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.7515 (0.7450) acc1: 56.5104 (53.8372) acc5: 100.0000 (100.0000) acc1_neg: 65.885 acc5_neg: 100.000 time: 0.6348 data: 0.5193 max mem: 3531 +Test: [190/333] eta: 0:01:32 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.6718 (0.7400) acc1: 59.1146 (54.3575) acc5: 100.0000 (100.0000) acc1_neg: 56.250 acc5_neg: 100.000 time: 0.5653 data: 0.4654 max mem: 3531 +Test: [200/333] eta: 0:01:25 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.6405 (0.7364) acc1: 59.8958 (54.7821) acc5: 100.0000 (100.0000) acc1_neg: 59.896 acc5_neg: 100.000 time: 0.5101 data: 0.3948 max mem: 3531 +Test: [210/333] eta: 0:01:19 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.7020 (0.7373) acc1: 59.6354 (54.9973) acc5: 100.0000 (100.0000) acc1_neg: 57.031 acc5_neg: 100.000 time: 0.6044 data: 0.4846 max mem: 3532 +Test: [220/333] eta: 0:01:12 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.7577 (0.7401) acc1: 58.8542 (54.9515) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.6446 data: 0.5385 max mem: 3532 +Test: [230/333] eta: 0:01:06 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.5294 (0.7297) acc1: 75.0000 (55.9918) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.6399 data: 0.5337 max mem: 3532 +Test: [240/333] eta: 0:00:59 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.5294 (0.7235) acc1: 76.5625 (56.5936) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.5860 data: 0.4800 max mem: 3532 +Test: [250/333] eta: 0:00:52 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.6050 (0.7197) acc1: 66.6667 (56.9752) acc5: 100.0000 (100.0000) acc1_neg: 58.073 acc5_neg: 100.000 time: 0.5200 data: 0.4149 max mem: 3532 +Test: [260/333] eta: 0:00:46 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.6479 (0.7180) acc1: 64.8438 (57.2148) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.6738 data: 0.5669 max mem: 3532 +Test: [270/333] eta: 0:00:40 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.6195 (0.7135) acc1: 65.3646 (57.6280) acc5: 100.0000 (100.0000) acc1_neg: 66.927 acc5_neg: 100.000 time: 0.7252 data: 0.6062 max mem: 3532 +Test: [280/333] eta: 0:00:34 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.6139 (0.7137) acc1: 65.3646 (57.5947) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.6703 data: 0.5560 max mem: 3532 +Test: [290/333] eta: 0:00:27 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.7915 (0.7165) acc1: 49.7396 (57.3821) acc5: 100.0000 (100.0000) acc1_neg: 58.073 acc5_neg: 100.000 time: 0.7094 data: 0.6018 max mem: 3532 +Test: [300/333] eta: 0:00:20 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.7844 (0.7187) acc1: 52.6042 (57.2865) acc5: 100.0000 (100.0000) acc1_neg: 59.896 acc5_neg: 100.000 time: 0.5393 data: 0.4010 max mem: 3532 +Test: [310/333] eta: 0:00:14 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.7186 (0.7143) acc1: 57.8125 (57.7263) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.4198 data: 0.2812 max mem: 3532 +Test: [320/333] eta: 0:00:08 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.5509 (0.7103) acc1: 73.1771 (58.1678) acc5: 100.0000 (100.0000) acc1_neg: 44.792 acc5_neg: 100.000 time: 0.4305 data: 0.3038 max mem: 3532 +Test: [330/333] eta: 0:00:01 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.8247 (0.7157) acc1: 50.5208 (57.8345) acc5: 100.0000 (100.0000) acc1_neg: 48.177 acc5_neg: 100.000 time: 0.3253 data: 0.1731 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.8606 (0.7168) acc1: 48.1771 (57.7638) acc5: 100.0000 (100.0000) acc1_neg: 46.197 acc5_neg: 100.000 time: 0.3449 data: 0.1731 max mem: 3532 +Test: Total time: 0:03:23 (0.6116 s / it) +* Overall Acc@1 57.764 Acc@5 100.000 + target == real Acc@1 53.651 Acc@5 100.000 + target == fake Acc@1 61.425 Acc@5 100.000 + loss 0.717 + PR-AUC 0.6121 ROC-AUC 0.6052 EER 42.53% (thr=0.5244) +Accuracy of the network on the 127843 test images: 57.8% +[rank0]:[W725 17:46:30.865996428 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 17:47:03.538000 2566841 torch/distributed/run.py:766] +W0725 17:47:03.538000 2566841 torch/distributed/run.py:766] ***************************************** +W0725 17:47:03.538000 2566841 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:47:03.538000 2566841 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:47:38.443367950 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:47:38.475671797 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 17:47:39.719900843 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:47:39.754543958 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:47:39.891730840 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +| distributed init (rank 6): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:47:39.995850231 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:47:39.001775316 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W725 17:47:39.001788965 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed1_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed1_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:20:05 acc1_pos: 94.531 acc5_pos: 100.000 loss: 0.2751 (0.2751) acc1: 94.5312 (94.5312) acc5: 100.0000 (100.0000) time: 5.1292 data: 3.1054 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/235] eta: 0:02:03 acc1_pos: 93.750 acc5_pos: 100.000 loss: 0.2744 (0.2752) acc1: 93.7500 (93.4186) acc5: 100.0000 (100.0000) time: 0.5479 data: 0.2826 max mem: 3531 +Test: [ 20/235] eta: 0:01:11 acc1_pos: 91.927 acc5_pos: 100.000 loss: 0.2675 (0.2732) acc1: 93.4896 (93.4524) acc5: 100.0000 (100.0000) time: 0.0903 data: 0.0007 max mem: 3531 +Test: [ 30/235] eta: 0:01:02 acc1_pos: 94.792 acc5_pos: 100.000 loss: 0.2648 (0.2712) acc1: 93.7500 (93.5484) acc5: 100.0000 (100.0000) time: 0.1683 data: 0.0777 max mem: 3531 +Test: [ 40/235] eta: 0:00:59 acc1_pos: 93.229 acc5_pos: 100.000 loss: 0.2678 (0.2714) acc1: 93.7500 (93.6039) acc5: 100.0000 (100.0000) time: 0.2798 data: 0.1884 max mem: 3531 +Test: [ 50/235] eta: 0:00:56 acc1_pos: 91.667 acc5_pos: 100.000 loss: 0.2715 (0.2721) acc1: 93.4896 (93.5815) acc5: 100.0000 (100.0000) time: 0.3132 data: 0.2226 max mem: 3531 +Test: [ 60/235] eta: 0:00:50 acc1_pos: 94.271 acc5_pos: 100.000 loss: 0.2756 (0.2734) acc1: 93.2292 (93.5280) acc5: 100.0000 (100.0000) time: 0.2480 data: 0.1539 max mem: 3531 +Test: [ 70/235] eta: 0:00:45 acc1_pos: 92.308 acc5_pos: 100.000 loss: 0.2957 (0.3868) acc1: 90.1042 (85.5560) acc5: 100.0000 (100.0000) acc1_neg: 26.042 acc5_neg: 100.000 time: 0.1965 data: 0.0977 max mem: 3531 +Test: [ 80/235] eta: 0:00:42 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.1831 (0.4866) acc1: 29.6875 (78.6233) acc5: 100.0000 (100.0000) acc1_neg: 28.125 acc5_neg: 100.000 time: 0.2348 data: 0.1382 max mem: 3531 +Test: [ 90/235] eta: 0:00:38 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.2004 (0.5643) acc1: 29.1667 (73.2000) acc5: 100.0000 (100.0000) acc1_neg: 29.948 acc5_neg: 100.000 time: 0.2223 data: 0.1277 max mem: 3531 +Test: [100/235] eta: 0:00:35 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.2004 (0.6248) acc1: 29.6875 (68.9485) acc5: 100.0000 (100.0000) acc1_neg: 27.604 acc5_neg: 100.000 time: 0.2004 data: 0.1014 max mem: 3531 +Test: [110/235] eta: 0:00:32 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.1733 (0.6756) acc1: 29.6875 (65.3411) acc5: 100.0000 (100.0000) acc1_neg: 30.990 acc5_neg: 100.000 time: 0.2225 data: 0.1207 max mem: 3531 +Test: [120/235] eta: 0:00:29 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.2344 (0.7460) acc1: 24.7396 (60.9483) acc5: 100.0000 (100.0000) acc1_neg: 13.281 acc5_neg: 100.000 time: 0.2213 data: 0.1241 max mem: 3531 +Test: [130/235] eta: 0:00:26 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.5439 (0.8078) acc1: 10.9375 (57.1028) acc5: 100.0000 (100.0000) acc1_neg: 11.458 acc5_neg: 100.000 time: 0.2001 data: 0.1051 max mem: 3531 +Test: [140/235] eta: 0:00:23 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.5732 (0.8635) acc1: 10.1562 (53.7622) acc5: 100.0000 (100.0000) acc1_neg: 7.812 acc5_neg: 100.000 time: 0.2130 data: 0.1147 max mem: 3531 +Test: [150/235] eta: 0:00:20 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.5732 (0.9094) acc1: 9.8958 (50.8744) acc5: 100.0000 (100.0000) acc1_neg: 11.719 acc5_neg: 100.000 time: 0.2213 data: 0.1247 max mem: 3531 +Test: [160/235] eta: 0:00:18 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.5594 (0.9501) acc1: 10.6771 (48.4019) acc5: 100.0000 (100.0000) acc1_neg: 12.760 acc5_neg: 100.000 time: 0.2111 data: 0.1122 max mem: 3531 +Test: [170/235] eta: 0:00:16 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.5072 (0.9526) acc1: 12.7604 (47.8710) acc5: 100.0000 (100.0000) acc1_neg: 38.542 acc5_neg: 100.000 time: 0.2785 data: 0.1767 max mem: 3531 +Test: [180/235] eta: 0:00:14 acc1_pos: 92.308 acc5_pos: 100.000 loss: 0.9272 (0.9508) acc1: 42.4479 (47.6275) acc5: 100.0000 (100.0000) acc1_neg: 43.750 acc5_neg: 100.000 time: 0.4537 data: 0.3563 max mem: 3531 +Test: [190/235] eta: 0:00:12 acc1_pos: 92.308 acc5_pos: 100.000 loss: 0.9193 (0.9486) acc1: 43.7500 (47.4408) acc5: 100.0000 (100.0000) acc1_neg: 45.312 acc5_neg: 100.000 time: 0.4696 data: 0.3762 max mem: 3531 +Test: [200/235] eta: 0:00:09 acc1_pos: 92.308 acc5_pos: 100.000 loss: 0.9193 (0.9478) acc1: 42.9688 (47.1743) acc5: 100.0000 (100.0000) acc1_neg: 41.146 acc5_neg: 100.000 time: 0.3249 data: 0.2247 max mem: 3531 +Test: [210/235] eta: 0:00:06 acc1_pos: 92.308 acc5_pos: 100.000 loss: 0.9263 (0.9462) acc1: 42.7083 (46.9910) acc5: 100.0000 (100.0000) acc1_neg: 42.969 acc5_neg: 100.000 time: 0.3505 data: 0.2311 max mem: 3532 +Test: [220/235] eta: 0:00:04 acc1_pos: 92.308 acc5_pos: 100.000 loss: 0.9117 (0.9447) acc1: 43.7500 (46.8691) acc5: 100.0000 (100.0000) acc1_neg: 43.490 acc5_neg: 100.000 time: 0.3948 data: 0.2684 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 92.308 acc5_pos: 100.000 loss: 0.9032 (0.9429) acc1: 44.2708 (46.7938) acc5: 100.0000 (100.0000) acc1_neg: 46.354 acc5_neg: 100.000 time: 0.3294 data: 0.2032 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 92.308 acc5_pos: 100.000 loss: 0.9084 (0.9431) acc1: 44.0104 (46.7417) acc5: 100.0000 (100.0000) acc1_neg: 52.174 acc5_neg: 100.000 time: 0.2771 data: 0.1557 max mem: 3532 +Test: Total time: 0:01:06 (0.2815 s / it) +* Overall Acc@1 46.742 Acc@5 100.000 + target == real Acc@1 93.516 Acc@5 100.000 + target == fake Acc@1 29.825 Acc@5 100.000 + loss 0.943 + PR-AUC 0.8761 ROC-AUC 0.7187 EER 34.00% (thr=0.2389) +Accuracy of the network on the 89879 test images: 46.7% +[rank0]:[W725 17:49:02.756019457 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 17:49:38.349000 2567949 torch/distributed/run.py:766] +W0725 17:49:38.349000 2567949 torch/distributed/run.py:766] ***************************************** +W0725 17:49:38.349000 2567949 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:49:38.349000 2567949 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:50:10.910338398 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:50:10.953690454 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:50:10.001485765 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 17:50:10.055268038 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:50:10.564074917 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:50:11.643482523 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank6]:[W725 17:50:11.643488229 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:50:11.667564744 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed1_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed1_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:24:37 acc1_pos: 35.417 acc5_pos: 100.000 loss: 0.8183 (0.8183) acc1: 35.4167 (35.4167) acc5: 100.0000 (100.0000) time: 5.4921 data: 3.8367 max mem: 3531 +Test: [ 10/269] eta: 0:02:36 acc1_pos: 61.458 acc5_pos: 100.000 loss: 0.6768 (0.6813) acc1: 55.7292 (57.0786) acc5: 100.0000 (100.0000) time: 0.6027 data: 0.3708 max mem: 3531 +Test: [ 20/269] eta: 0:01:30 acc1_pos: 79.167 acc5_pos: 100.000 loss: 0.6529 (0.6398) acc1: 59.3750 (62.1032) acc5: 100.0000 (100.0000) time: 0.1061 data: 0.0129 max mem: 3531 +Test: [ 30/269] eta: 0:01:07 acc1_pos: 65.104 acc5_pos: 100.000 loss: 0.5707 (0.6191) acc1: 65.1042 (64.5497) acc5: 100.0000 (100.0000) time: 0.1092 data: 0.0099 max mem: 3531 +Test: [ 40/269] eta: 0:01:01 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.5230 (0.6034) acc1: 71.6146 (67.6321) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.1700 data: 0.0737 max mem: 3531 +Test: [ 50/269] eta: 0:00:55 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.5851 (0.6083) acc1: 71.6146 (67.3356) acc5: 100.0000 (100.0000) acc1_neg: 44.271 acc5_neg: 100.000 time: 0.2025 data: 0.1112 max mem: 3531 +Test: [ 60/269] eta: 0:00:49 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.6283 (0.6168) acc1: 61.1979 (66.1629) acc5: 100.0000 (100.0000) acc1_neg: 59.896 acc5_neg: 100.000 time: 0.1672 data: 0.0751 max mem: 3531 +Test: [ 70/269] eta: 0:00:44 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.6570 (0.6238) acc1: 60.4167 (65.2472) acc5: 100.0000 (100.0000) acc1_neg: 60.938 acc5_neg: 100.000 time: 0.1435 data: 0.0504 max mem: 3531 +Test: [ 80/269] eta: 0:00:40 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.6356 (0.6213) acc1: 63.5417 (65.7118) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.1455 data: 0.0522 max mem: 3531 +Test: [ 90/269] eta: 0:00:37 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.6414 (0.6233) acc1: 63.5417 (65.4504) acc5: 100.0000 (100.0000) acc1_neg: 66.146 acc5_neg: 100.000 time: 0.1673 data: 0.0707 max mem: 3531 +Test: [100/269] eta: 0:00:34 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.6631 (0.6327) acc1: 58.3333 (64.3410) acc5: 100.0000 (100.0000) acc1_neg: 36.979 acc5_neg: 100.000 time: 0.1779 data: 0.0822 max mem: 3531 +Test: [110/269] eta: 0:00:32 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.7178 (0.6428) acc1: 52.3438 (62.9106) acc5: 100.0000 (100.0000) acc1_neg: 47.396 acc5_neg: 100.000 time: 0.1674 data: 0.0748 max mem: 3531 +Test: [120/269] eta: 0:00:29 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.7504 (0.6584) acc1: 46.0938 (61.6004) acc5: 100.0000 (100.0000) acc1_neg: 36.458 acc5_neg: 100.000 time: 0.1640 data: 0.0669 max mem: 3531 +Test: [130/269] eta: 0:00:27 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.8144 (0.6696) acc1: 43.2292 (60.3531) acc5: 100.0000 (100.0000) acc1_neg: 43.490 acc5_neg: 100.000 time: 0.1640 data: 0.0672 max mem: 3531 +Test: [140/269] eta: 0:00:25 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.8147 (0.6858) acc1: 39.0625 (58.5217) acc5: 100.0000 (100.0000) acc1_neg: 40.625 acc5_neg: 100.000 time: 0.1649 data: 0.0708 max mem: 3531 +Test: [150/269] eta: 0:00:23 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.7994 (0.6910) acc1: 40.3646 (57.7849) acc5: 100.0000 (100.0000) acc1_neg: 46.875 acc5_neg: 100.000 time: 0.1807 data: 0.0871 max mem: 3531 +Test: [160/269] eta: 0:00:21 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.7316 (0.6900) acc1: 51.8229 (57.9645) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.2050 data: 0.1132 max mem: 3531 +Test: [170/269] eta: 0:00:19 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.6659 (0.6909) acc1: 59.6354 (57.9785) acc5: 100.0000 (100.0000) acc1_neg: 71.094 acc5_neg: 100.000 time: 0.2097 data: 0.1165 max mem: 3531 +Test: [180/269] eta: 0:00:17 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.6495 (0.6882) acc1: 60.9375 (58.4844) acc5: 100.0000 (100.0000) acc1_neg: 60.938 acc5_neg: 100.000 time: 0.2214 data: 0.1164 max mem: 3531 +Test: [190/269] eta: 0:00:15 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.6349 (0.6893) acc1: 60.9375 (58.4056) acc5: 100.0000 (100.0000) acc1_neg: 47.656 acc5_neg: 100.000 time: 0.2145 data: 0.0810 max mem: 3531 +Test: [200/269] eta: 0:00:13 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.7370 (0.6937) acc1: 50.7812 (57.8993) acc5: 100.0000 (100.0000) acc1_neg: 56.771 acc5_neg: 100.000 time: 0.1789 data: 0.0536 max mem: 3531 +Test: [210/269] eta: 0:00:12 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.7379 (0.6958) acc1: 48.1771 (57.6200) acc5: 100.0000 (100.0000) acc1_neg: 53.385 acc5_neg: 100.000 time: 0.4096 data: 0.3030 max mem: 3532 +Test: [220/269] eta: 0:00:11 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.6067 (0.6943) acc1: 53.3854 (57.8490) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.5987 data: 0.4933 max mem: 3532 +Test: [230/269] eta: 0:00:09 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.6000 (0.6921) acc1: 64.8438 (58.0673) acc5: 100.0000 (100.0000) acc1_neg: 58.594 acc5_neg: 100.000 time: 0.3663 data: 0.2680 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.6597 (0.6938) acc1: 58.5938 (57.8979) acc5: 100.0000 (100.0000) acc1_neg: 61.198 acc5_neg: 100.000 time: 0.2955 data: 0.1950 max mem: 3532 +Test: [250/269] eta: 0:00:04 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.7723 (0.7013) acc1: 48.4375 (57.0053) acc5: 100.0000 (100.0000) acc1_neg: 15.365 acc5_neg: 100.000 time: 0.5435 data: 0.4359 max mem: 3532 +Test: [260/269] eta: 0:00:02 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.9100 (0.7087) acc1: 25.5208 (56.1103) acc5: 100.0000 (100.0000) acc1_neg: 56.250 acc5_neg: 100.000 time: 0.5998 data: 0.4767 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.7695 (0.7089) acc1: 37.5000 (55.9835) acc5: 100.0000 (100.0000) acc1_neg: 65.879 acc5_neg: 100.000 time: 0.4549 data: 0.3256 max mem: 3532 +Test: Total time: 0:01:13 (0.2727 s / it) +* Overall Acc@1 55.983 Acc@5 100.000 + target == real Acc@1 64.036 Acc@5 100.000 + target == fake Acc@1 54.885 Acc@5 100.000 + loss 0.709 + PR-AUC 0.9225 ROC-AUC 0.6322 EER 40.63% (thr=0.4792) +Accuracy of the network on the 103293 test images: 56.0% +[rank0]:[W725 17:51:48.040896529 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-25T17:52:03Z] eval_done tag=fixed500k_fake_n32_seed1_resampled +[2026-07-25T17:52:04Z] eval_start tag=fixed500k_real_n16_seed2_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed2_resampled_n111-046-081_20260725_112559/best_checkpoint.pth +started_utc=2026-07-25T17:52:04Z +hostname=n111-046-081 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed2_resampled_n111-046-081_20260725_112559/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed2_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0725 17:52:19.837000 2569062 torch/distributed/run.py:766] +W0725 17:52:19.837000 2569062 torch/distributed/run.py:766] ***************************************** +W0725 17:52:19.837000 2569062 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:52:19.837000 2569062 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:52:53.292311337 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:52:53.558244519 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:52:54.835057713 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:52:54.895654215 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:52:54.956447948 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:52:54.998814475 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +[rank0]:[W725 17:52:54.032638923 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:52:54.041234188 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed2_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed2_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:20:56 acc1_pos: 77.865 acc5_pos: 100.000 loss: 0.5878 (0.5878) acc1: 77.8646 (77.8646) acc5: 100.0000 (100.0000) time: 5.3702 data: 2.9393 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/234] eta: 0:02:09 acc1_pos: 79.688 acc5_pos: 100.000 loss: 0.5878 (0.7062) acc1: 70.0521 (61.1032) acc5: 100.0000 (100.0000) time: 0.5766 data: 0.2741 max mem: 3531 +Test: [ 20/234] eta: 0:01:15 acc1_pos: 94.792 acc5_pos: 100.000 loss: 0.5675 (0.6160) acc1: 70.0521 (68.3160) acc5: 100.0000 (100.0000) time: 0.1010 data: 0.0113 max mem: 3531 +Test: [ 30/234] eta: 0:01:01 acc1_pos: 38.021 acc5_pos: 100.000 loss: 0.2444 (0.5255) acc1: 89.3229 (73.6475) acc5: 100.0000 (100.0000) time: 0.1495 data: 0.0573 max mem: 3531 +Test: [ 40/234] eta: 0:00:59 acc1_pos: 97.917 acc5_pos: 100.000 loss: 0.2231 (0.5506) acc1: 92.4479 (69.9378) acc5: 100.0000 (100.0000) time: 0.2608 data: 0.1663 max mem: 3531 +Test: [ 50/234] eta: 0:00:53 acc1_pos: 97.135 acc5_pos: 100.000 loss: 0.1493 (0.4699) acc1: 96.8750 (75.2757) acc5: 100.0000 (100.0000) time: 0.2736 data: 0.1791 max mem: 3531 +Test: [ 60/234] eta: 0:00:49 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.1120 (0.4067) acc1: 97.6562 (79.2819) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2450 data: 0.1515 max mem: 3531 +Test: [ 70/234] eta: 0:00:44 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.0816 (0.4299) acc1: 99.7396 (78.2314) acc5: 100.0000 (100.0000) acc1_neg: 53.385 acc5_neg: 100.000 time: 0.2249 data: 0.1301 max mem: 3531 +Test: [ 80/234] eta: 0:00:41 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.7812 (0.4530) acc1: 59.1146 (77.2923) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2098 data: 0.1109 max mem: 3531 +Test: [ 90/234] eta: 0:00:37 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.2831 (0.4293) acc1: 92.7083 (79.0064) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.2253 data: 0.1289 max mem: 3531 +Test: [100/234] eta: 0:00:34 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.2951 (0.4328) acc1: 89.5833 (78.9423) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1999 data: 0.1064 max mem: 3531 +Test: [110/234] eta: 0:00:30 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.4540 (0.4355) acc1: 77.8646 (79.0095) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.1745 data: 0.0787 max mem: 3531 +Test: [120/234] eta: 0:00:27 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.4021 (0.4437) acc1: 81.2500 (78.7685) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1605 data: 0.0648 max mem: 3531 +Test: [130/234] eta: 0:00:26 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.4723 (0.4499) acc1: 76.3021 (78.3278) acc5: 100.0000 (100.0000) acc1_neg: 76.302 acc5_neg: 100.000 time: 0.2950 data: 0.1969 max mem: 3531 +Test: [140/234] eta: 0:00:26 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.5638 (0.4732) acc1: 70.8333 (77.3493) acc5: 100.0000 (100.0000) acc1_neg: 30.729 acc5_neg: 100.000 time: 0.5543 data: 0.4570 max mem: 3531 +Test: [150/234] eta: 0:00:23 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.5299 (0.4842) acc1: 72.3958 (76.8729) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.4872 data: 0.3946 max mem: 3531 +Test: [160/234] eta: 0:00:21 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.4055 (0.4793) acc1: 82.5521 (77.3292) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.3456 data: 0.2245 max mem: 3531 +Test: [170/234] eta: 0:00:19 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.3889 (0.4741) acc1: 84.8958 (77.7153) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.4835 data: 0.3536 max mem: 3531 +Test: [180/234] eta: 0:00:17 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.3889 (0.4862) acc1: 81.2500 (76.9898) acc5: 100.0000 (100.0000) acc1_neg: 32.812 acc5_neg: 100.000 time: 0.5991 data: 0.4754 max mem: 3531 +Test: [190/234] eta: 0:00:14 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.3652 (0.4811) acc1: 82.8125 (77.3928) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.5689 data: 0.4442 max mem: 3531 +Test: [200/234] eta: 0:00:11 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.3253 (0.4669) acc1: 87.5000 (78.2610) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.4853 data: 0.3702 max mem: 3531 +Test: [210/234] eta: 0:00:08 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.1104 (0.4519) acc1: 100.0000 (79.1975) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.5566 data: 0.4339 max mem: 3532 +Test: [220/234] eta: 0:00:05 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.2271 (0.4426) acc1: 94.2708 (79.8218) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.5744 data: 0.4426 max mem: 3532 +Test: [230/234] eta: 0:00:01 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.3782 (0.4532) acc1: 65.8854 (78.9931) acc5: 100.0000 (100.0000) acc1_neg: 58.854 acc5_neg: 100.000 time: 0.4535 data: 0.3072 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.6637 (0.4558) acc1: 61.1979 (78.7950) acc5: 100.0000 (100.0000) acc1_neg: 64.591 acc5_neg: 100.000 time: 0.4249 data: 0.2788 max mem: 3532 +Test: Total time: 0:01:25 (0.3647 s / it) +* Overall Acc@1 78.795 Acc@5 100.000 + target == real Acc@1 75.412 Acc@5 100.000 + target == fake Acc@1 79.749 Acc@5 100.000 + loss 0.456 + PR-AUC 0.9591 ROC-AUC 0.8669 EER 22.06% (thr=0.5308) +Accuracy of the network on the 89729 test images: 78.8% +[rank0]:[W725 17:54:36.757105224 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 17:55:09.433000 2570168 torch/distributed/run.py:766] +W0725 17:55:09.433000 2570168 torch/distributed/run.py:766] ***************************************** +W0725 17:55:09.433000 2570168 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:55:09.433000 2570168 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W725 17:55:41.218717333 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:55:42.656502359 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:55:42.805145905 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:55:42.440677722 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:55:43.750232320 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W725 17:55:43.750247622 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:55:43.769291024 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:55:43.852121792 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed2_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed2_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 13:52:51 acc1_pos: 60.156 acc5_pos: 100.000 loss: 0.7284 (0.7284) acc1: 60.1562 (60.1562) acc5: 100.0000 (100.0000) time: 20.9347 data: 16.3401 max mem: 3529 +Test: [ 10/2387] eta: 1:20:38 acc1_pos: 23.177 acc5_pos: 100.000 loss: 0.8606 (0.9247) acc1: 36.7188 (37.8314) acc5: 100.0000 (100.0000) time: 2.0357 data: 1.4860 max mem: 3529 +Test: [ 20/2387] eta: 0:46:45 acc1_pos: 14.583 acc5_pos: 100.000 loss: 0.9938 (0.9844) acc1: 28.6458 (31.8452) acc5: 100.0000 (100.0000) time: 0.1978 data: 0.0637 max mem: 3529 +Test: [ 30/2387] eta: 0:42:41 acc1_pos: 24.219 acc5_pos: 100.000 loss: 1.0484 (0.9775) acc1: 26.5625 (33.5938) acc5: 100.0000 (100.0000) time: 0.5649 data: 0.4471 max mem: 3529 +Test: [ 40/2387] eta: 0:39:00 acc1_pos: 56.250 acc5_pos: 100.000 loss: 0.9165 (0.9663) acc1: 34.8958 (34.4703) acc5: 100.0000 (100.0000) time: 0.7999 data: 0.6753 max mem: 3529 +Test: [ 50/2387] eta: 0:36:25 acc1_pos: 48.958 acc5_pos: 100.000 loss: 0.9165 (0.9665) acc1: 34.8958 (34.2984) acc5: 100.0000 (100.0000) time: 0.6999 data: 0.5830 max mem: 3529 +Test: [ 60/2387] eta: 0:33:29 acc1_pos: 60.156 acc5_pos: 100.000 loss: 0.8658 (0.9465) acc1: 39.8438 (35.8564) acc5: 100.0000 (100.0000) time: 0.5899 data: 0.4745 max mem: 3530 +Test: [ 70/2387] eta: 0:31:39 acc1_pos: 52.604 acc5_pos: 100.000 loss: 0.8249 (0.9287) acc1: 44.2708 (37.3900) acc5: 100.0000 (100.0000) time: 0.5262 data: 0.3954 max mem: 3530 +Test: [ 80/2387] eta: 0:30:02 acc1_pos: 98.698 acc5_pos: 100.000 loss: 0.7191 (0.8790) acc1: 49.2188 (42.1521) acc5: 100.0000 (100.0000) time: 0.5299 data: 0.4055 max mem: 3530 +Test: [ 90/2387] eta: 0:29:26 acc1_pos: 99.740 acc5_pos: 100.000 loss: 0.4065 (0.8240) acc1: 88.0208 (47.7707) acc5: 100.0000 (100.0000) time: 0.5888 data: 0.4525 max mem: 3530 +Test: [ 100/2387] eta: 0:28:42 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.3740 (0.7867) acc1: 91.1458 (51.5573) acc5: 100.0000 (100.0000) time: 0.6399 data: 0.5129 max mem: 3530 +Test: [ 110/2387] eta: 0:28:01 acc1_pos: 62.500 acc5_pos: 100.000 loss: 0.4203 (0.7570) acc1: 86.7188 (54.4576) acc5: 100.0000 (100.0000) time: 0.6002 data: 0.4981 max mem: 3530 +Test: [ 120/2387] eta: 0:27:20 acc1_pos: 89.583 acc5_pos: 100.000 loss: 0.4378 (0.7318) acc1: 84.6354 (57.0183) acc5: 100.0000 (100.0000) time: 0.5749 data: 0.4595 max mem: 3530 +Test: [ 130/2387] eta: 0:26:50 acc1_pos: 81.510 acc5_pos: 100.000 loss: 0.4367 (0.7077) acc1: 87.7604 (59.5519) acc5: 100.0000 (100.0000) time: 0.5744 data: 0.4455 max mem: 3530 +Test: [ 140/2387] eta: 0:26:27 acc1_pos: 90.365 acc5_pos: 100.000 loss: 0.4157 (0.6902) acc1: 89.0625 (61.3826) acc5: 100.0000 (100.0000) time: 0.6008 data: 0.4752 max mem: 3530 +Test: [ 150/2387] eta: 0:25:52 acc1_pos: 99.479 acc5_pos: 100.000 loss: 0.4346 (0.6723) acc1: 88.2812 (63.1726) acc5: 100.0000 (100.0000) time: 0.5647 data: 0.4441 max mem: 3530 +Test: [ 160/2387] eta: 0:25:19 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.3790 (0.6508) acc1: 89.0625 (64.9926) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.5140 data: 0.3861 max mem: 3530 +Test: [ 170/2387] eta: 0:24:45 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2528 (0.6274) acc1: 95.3125 (66.8007) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.4899 data: 0.3522 max mem: 3530 +Test: [ 180/2387] eta: 0:24:04 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2476 (0.6065) acc1: 96.3542 (68.4220) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.4299 data: 0.2954 max mem: 3530 +Test: [ 190/2387] eta: 0:23:55 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2465 (0.5877) acc1: 96.3542 (69.8735) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.5099 data: 0.3737 max mem: 3530 +Test: [ 200/2387] eta: 0:23:23 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2432 (0.5706) acc1: 96.0938 (71.1896) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.5261 data: 0.3904 max mem: 3530 +Test: [ 210/2387] eta: 0:23:03 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2432 (0.5555) acc1: 96.0938 (72.3847) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.4649 data: 0.3367 max mem: 3530 +Test: [ 220/2387] eta: 0:22:18 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2462 (0.5415) acc1: 96.0938 (73.4693) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.3737 data: 0.2515 max mem: 3530 +Test: [ 230/2387] eta: 0:21:44 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2462 (0.5290) acc1: 96.0938 (74.4431) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2798 data: 0.1211 max mem: 3530 +Test: [ 240/2387] eta: 0:21:02 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2594 (0.5297) acc1: 94.5312 (74.2674) acc5: 100.0000 (100.0000) acc1_neg: 65.625 acc5_neg: 100.000 time: 0.2648 data: 0.0779 max mem: 3530 +Test: [ 250/2387] eta: 0:20:23 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2410 (0.5155) acc1: 74.2188 (75.1639) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2007 data: 0.0665 max mem: 3530 +Test: [ 260/2387] eta: 0:19:50 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1222 (0.5002) acc1: 100.0000 (76.1145) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2159 data: 0.1232 max mem: 3530 +Test: [ 270/2387] eta: 0:19:17 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1088 (0.4859) acc1: 100.0000 (76.9959) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2241 data: 0.1308 max mem: 3530 +Test: [ 280/2387] eta: 0:18:47 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1147 (0.4727) acc1: 100.0000 (77.8136) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2145 data: 0.1191 max mem: 3530 +Test: [ 290/2387] eta: 0:18:18 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1144 (0.4604) acc1: 100.0000 (78.5751) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2121 data: 0.1195 max mem: 3531 +Test: [ 300/2387] eta: 0:17:53 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1243 (0.4502) acc1: 100.0000 (79.2835) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2250 data: 0.1339 max mem: 3531 +Test: [ 310/2387] eta: 0:17:32 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1595 (0.4435) acc1: 100.0000 (79.8910) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2576 data: 0.1646 max mem: 3531 +Test: [ 320/2387] eta: 0:17:06 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.3231 (0.4400) acc1: 97.1354 (80.4274) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2292 data: 0.1363 max mem: 3531 +Test: [ 330/2387] eta: 0:16:42 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.3292 (0.4367) acc1: 96.8750 (80.9180) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1877 data: 0.0973 max mem: 3531 +Test: [ 340/2387] eta: 0:16:20 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.3295 (0.4336) acc1: 96.6146 (81.3684) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1922 data: 0.0993 max mem: 3531 +Test: [ 350/2387] eta: 0:15:57 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.3244 (0.4275) acc1: 97.1354 (81.8866) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1872 data: 0.0937 max mem: 3531 +Test: [ 360/2387] eta: 0:15:37 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2080 (0.4215) acc1: 98.6979 (82.3350) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1802 data: 0.0886 max mem: 3531 +Test: [ 370/2387] eta: 0:15:17 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2063 (0.4157) acc1: 98.1771 (82.7669) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1852 data: 0.0932 max mem: 3531 +Test: [ 380/2387] eta: 0:14:58 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2058 (0.4103) acc1: 98.1771 (83.1713) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1874 data: 0.0964 max mem: 3531 +Test: [ 390/2387] eta: 0:14:40 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2131 (0.4052) acc1: 97.9167 (83.5418) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1795 data: 0.0888 max mem: 3531 +Test: [ 400/2387] eta: 0:14:22 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2115 (0.4004) acc1: 97.9167 (83.9003) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1727 data: 0.0818 max mem: 3531 +Test: [ 410/2387] eta: 0:14:05 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2089 (0.3957) acc1: 97.9167 (84.2426) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1760 data: 0.0857 max mem: 3531 +Test: [ 420/2387] eta: 0:13:49 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2105 (0.3913) acc1: 97.6562 (84.5674) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1769 data: 0.0850 max mem: 3531 +Test: [ 430/2387] eta: 0:13:35 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2156 (0.3909) acc1: 97.3958 (84.5943) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.1881 data: 0.0952 max mem: 3531 +Test: [ 440/2387] eta: 0:13:21 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2218 (0.3874) acc1: 82.8125 (84.7565) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1953 data: 0.1019 max mem: 3531 +Test: [ 450/2387] eta: 0:13:07 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.0869 (0.3808) acc1: 100.0000 (85.0945) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1914 data: 0.0989 max mem: 3531 +Test: [ 460/2387] eta: 0:12:54 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.0864 (0.3744) acc1: 100.0000 (85.4178) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1842 data: 0.0940 max mem: 3531 +Test: [ 470/2387] eta: 0:12:43 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.0862 (0.3683) acc1: 100.0000 (85.7274) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2029 data: 0.1106 max mem: 3531 +Test: [ 480/2387] eta: 0:12:31 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.0865 (0.3624) acc1: 100.0000 (86.0241) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2191 data: 0.1257 max mem: 3531 +Test: [ 490/2387] eta: 0:12:21 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.0872 (0.3568) acc1: 100.0000 (86.3082) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2149 data: 0.1234 max mem: 3531 +Test: [ 500/2387] eta: 0:12:10 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.0871 (0.3514) acc1: 100.0000 (86.5815) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2125 data: 0.1217 max mem: 3531 +Test: [ 510/2387] eta: 0:11:59 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.0869 (0.3472) acc1: 100.0000 (86.8395) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1978 data: 0.1067 max mem: 3531 +Test: [ 520/2387] eta: 0:11:49 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1405 (0.3446) acc1: 99.7396 (87.0846) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2090 data: 0.1182 max mem: 3532 +Test: [ 530/2387] eta: 0:11:39 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1364 (0.3407) acc1: 99.7396 (87.3264) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2119 data: 0.1195 max mem: 3532 +Test: [ 540/2387] eta: 0:11:29 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1352 (0.3369) acc1: 100.0000 (87.5597) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1988 data: 0.1044 max mem: 3532 +Test: [ 550/2387] eta: 0:11:20 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1346 (0.3332) acc1: 100.0000 (87.7850) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1979 data: 0.1048 max mem: 3532 +Test: [ 560/2387] eta: 0:11:09 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1343 (0.3296) acc1: 100.0000 (88.0023) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1843 data: 0.0924 max mem: 3532 +Test: [ 570/2387] eta: 0:11:00 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1334 (0.3262) acc1: 100.0000 (88.2115) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1828 data: 0.0909 max mem: 3532 +Test: [ 580/2387] eta: 0:10:51 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1331 (0.3229) acc1: 100.0000 (88.4135) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1953 data: 0.1018 max mem: 3532 +Test: [ 590/2387] eta: 0:10:43 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1347 (0.3197) acc1: 100.0000 (88.6086) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1941 data: 0.1016 max mem: 3532 +Test: [ 600/2387] eta: 0:10:33 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1414 (0.3186) acc1: 99.7396 (88.7878) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1805 data: 0.0899 max mem: 3532 +Test: [ 610/2387] eta: 0:10:25 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2271 (0.3169) acc1: 98.4375 (88.9282) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1700 data: 0.0789 max mem: 3532 +Test: [ 620/2387] eta: 0:10:16 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2159 (0.3153) acc1: 97.1354 (89.0558) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1665 data: 0.0747 max mem: 3532 +Test: [ 630/2387] eta: 0:10:07 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2168 (0.3137) acc1: 97.1354 (89.1917) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1610 data: 0.0696 max mem: 3532 +Test: [ 640/2387] eta: 0:09:59 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2158 (0.3123) acc1: 97.6562 (89.3193) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1687 data: 0.0758 max mem: 3532 +Test: [ 650/2387] eta: 0:09:51 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2158 (0.3108) acc1: 97.3958 (89.4429) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1786 data: 0.0852 max mem: 3532 +Test: [ 660/2387] eta: 0:09:44 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2165 (0.3094) acc1: 97.3958 (89.5640) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1838 data: 0.0904 max mem: 3532 +Test: [ 670/2387] eta: 0:09:36 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2165 (0.3080) acc1: 97.3958 (89.6846) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1826 data: 0.0873 max mem: 3532 +Test: [ 680/2387] eta: 0:09:28 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2204 (0.3099) acc1: 97.1354 (89.5355) acc5: 100.0000 (100.0000) acc1_neg: 67.969 acc5_neg: 100.000 time: 0.1698 data: 0.0756 max mem: 3532 +Test: [ 690/2387] eta: 0:09:22 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2852 (0.3103) acc1: 74.2188 (89.4612) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1772 data: 0.0840 max mem: 3532 +Test: [ 700/2387] eta: 0:09:15 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2612 (0.3096) acc1: 89.5833 (89.4663) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.2014 data: 0.1086 max mem: 3532 +Test: [ 710/2387] eta: 0:09:08 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2590 (0.3088) acc1: 89.3229 (89.4749) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.1873 data: 0.0949 max mem: 3532 +Test: [ 720/2387] eta: 0:09:01 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2553 (0.3081) acc1: 90.3646 (89.4901) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1584 data: 0.0660 max mem: 3532 +Test: [ 730/2387] eta: 0:08:55 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2542 (0.3074) acc1: 90.6250 (89.5071) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1725 data: 0.0806 max mem: 3532 +Test: [ 740/2387] eta: 0:08:49 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2513 (0.3066) acc1: 90.6250 (89.5253) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1956 data: 0.1048 max mem: 3532 +Test: [ 750/2387] eta: 0:08:42 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2507 (0.3059) acc1: 90.6250 (89.5442) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.1854 data: 0.0953 max mem: 3533 +Test: [ 760/2387] eta: 0:08:36 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2557 (0.3058) acc1: 90.1042 (89.5115) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1797 data: 0.0889 max mem: 3533 +Test: [ 770/2387] eta: 0:08:30 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.3386 (0.3063) acc1: 84.3750 (89.4665) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1795 data: 0.0870 max mem: 3533 +Test: [ 780/2387] eta: 0:08:24 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2994 (0.3054) acc1: 91.6667 (89.5437) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1748 data: 0.0797 max mem: 3533 +Test: [ 790/2387] eta: 0:08:18 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1783 (0.3039) acc1: 98.1771 (89.6544) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1800 data: 0.0853 max mem: 3533 +Test: [ 800/2387] eta: 0:08:13 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1783 (0.3027) acc1: 99.2188 (89.7638) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1846 data: 0.0900 max mem: 3533 +Test: [ 810/2387] eta: 0:08:07 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2199 (0.3020) acc1: 95.8333 (89.8222) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1863 data: 0.0933 max mem: 3533 +Test: [ 820/2387] eta: 0:08:02 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2353 (0.3018) acc1: 95.3125 (89.8482) acc5: 100.0000 (100.0000) acc1_neg: 69.531 acc5_neg: 100.000 time: 0.1850 data: 0.0928 max mem: 3533 +Test: [ 830/2387] eta: 0:07:56 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.5079 (0.3062) acc1: 69.5312 (89.4649) acc5: 100.0000 (100.0000) acc1_neg: 54.427 acc5_neg: 100.000 time: 0.1783 data: 0.0827 max mem: 3533 +Test: [ 840/2387] eta: 0:07:50 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.3521 (0.3056) acc1: 73.9583 (89.5375) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1678 data: 0.0733 max mem: 3533 +Test: [ 850/2387] eta: 0:07:45 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2266 (0.3046) acc1: 97.1354 (89.6298) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1754 data: 0.0830 max mem: 3533 +Test: [ 860/2387] eta: 0:07:40 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2068 (0.3035) acc1: 98.4375 (89.7321) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1821 data: 0.0901 max mem: 3533 +Test: [ 870/2387] eta: 0:07:35 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2119 (0.3024) acc1: 98.9583 (89.8342) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1807 data: 0.0881 max mem: 3533 +Test: [ 880/2387] eta: 0:07:30 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2274 (0.3019) acc1: 97.9167 (89.9147) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1916 data: 0.0976 max mem: 3533 +Test: [ 890/2387] eta: 0:07:25 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2842 (0.3030) acc1: 94.5312 (89.8619) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1899 data: 0.0958 max mem: 3533 +Test: [ 900/2387] eta: 0:07:20 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.4494 (0.3064) acc1: 79.6875 (89.6125) acc5: 100.0000 (100.0000) acc1_neg: 64.844 acc5_neg: 100.000 time: 0.1800 data: 0.0875 max mem: 3533 +Test: [ 910/2387] eta: 0:07:15 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.6477 (0.3103) acc1: 62.5000 (89.3206) acc5: 100.0000 (100.0000) acc1_neg: 62.240 acc5_neg: 100.000 time: 0.1800 data: 0.0892 max mem: 3533 +Test: [ 920/2387] eta: 0:07:10 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.6614 (0.3142) acc1: 61.7188 (89.0153) acc5: 100.0000 (100.0000) acc1_neg: 61.979 acc5_neg: 100.000 time: 0.1779 data: 0.0860 max mem: 3533 +Test: [ 930/2387] eta: 0:07:06 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.6701 (0.3179) acc1: 61.9792 (88.7347) acc5: 100.0000 (100.0000) acc1_neg: 61.979 acc5_neg: 100.000 time: 0.1835 data: 0.0894 max mem: 3533 +Test: [ 940/2387] eta: 0:07:01 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.6727 (0.3216) acc1: 61.9792 (88.4537) acc5: 100.0000 (100.0000) acc1_neg: 57.552 acc5_neg: 100.000 time: 0.1860 data: 0.0929 max mem: 3533 +Test: [ 950/2387] eta: 0:06:57 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.6603 (0.3252) acc1: 61.7188 (88.1709) acc5: 100.0000 (100.0000) acc1_neg: 61.719 acc5_neg: 100.000 time: 0.1856 data: 0.0943 max mem: 3533 +Test: [ 960/2387] eta: 0:06:52 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.6547 (0.3286) acc1: 63.2812 (87.9219) acc5: 100.0000 (100.0000) acc1_neg: 64.062 acc5_neg: 100.000 time: 0.1850 data: 0.0932 max mem: 3533 +Test: [ 970/2387] eta: 0:06:48 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.6547 (0.3321) acc1: 63.2812 (87.6518) acc5: 100.0000 (100.0000) acc1_neg: 65.365 acc5_neg: 100.000 time: 0.1843 data: 0.0920 max mem: 3534 +Test: [ 980/2387] eta: 0:06:44 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.6366 (0.3333) acc1: 63.5417 (87.5804) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.1866 data: 0.0950 max mem: 3534 +Test: [ 990/2387] eta: 0:06:39 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2938 (0.3325) acc1: 93.7500 (87.6298) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1869 data: 0.0960 max mem: 3534 +Test: [1000/2387] eta: 0:06:35 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1913 (0.3312) acc1: 97.9167 (87.7263) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1992 data: 0.1069 max mem: 3534 +Test: [1010/2387] eta: 0:06:31 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2258 (0.3319) acc1: 97.3958 (87.6499) acc5: 100.0000 (100.0000) acc1_neg: 70.833 acc5_neg: 100.000 time: 0.1978 data: 0.1039 max mem: 3534 +Test: [1020/2387] eta: 0:06:27 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.3638 (0.3328) acc1: 79.9479 (87.5811) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1877 data: 0.0940 max mem: 3534 +Test: [1030/2387] eta: 0:06:23 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2343 (0.3316) acc1: 97.1354 (87.6798) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1925 data: 0.1002 max mem: 3534 +Test: [1040/2387] eta: 0:06:19 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1870 (0.3303) acc1: 98.4375 (87.7707) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1983 data: 0.1069 max mem: 3534 +Test: [1050/2387] eta: 0:06:16 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1823 (0.3289) acc1: 98.9583 (87.8737) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2081 data: 0.1148 max mem: 3534 +Test: [1060/2387] eta: 0:06:12 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1952 (0.3279) acc1: 98.6979 (87.9605) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2041 data: 0.1116 max mem: 3534 +Test: [1070/2387] eta: 0:06:08 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2228 (0.3280) acc1: 95.3125 (87.9464) acc5: 100.0000 (100.0000) acc1_neg: 42.188 acc5_neg: 100.000 time: 0.1961 data: 0.1059 max mem: 3534 +Test: [1080/2387] eta: 0:06:04 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.5621 (0.3308) acc1: 65.6250 (87.6973) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.2071 data: 0.1144 max mem: 3534 +Test: [1090/2387] eta: 0:06:00 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.3202 (0.3301) acc1: 80.4688 (87.7762) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2079 data: 0.1128 max mem: 3534 +Test: [1100/2387] eta: 0:05:57 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2153 (0.3289) acc1: 98.1771 (87.8773) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1985 data: 0.1047 max mem: 3534 +Test: [1110/2387] eta: 0:05:53 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2002 (0.3278) acc1: 99.4792 (87.9789) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2075 data: 0.1146 max mem: 3534 +Test: [1120/2387] eta: 0:05:50 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2087 (0.3269) acc1: 98.4375 (88.0538) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.2144 data: 0.1208 max mem: 3534 +Test: [1130/2387] eta: 0:05:46 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2182 (0.3260) acc1: 97.3958 (88.1362) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2043 data: 0.1111 max mem: 3534 +Test: [1140/2387] eta: 0:05:43 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2578 (0.3281) acc1: 92.9688 (87.9398) acc5: 100.0000 (100.0000) acc1_neg: 53.646 acc5_neg: 100.000 time: 0.1970 data: 0.1050 max mem: 3534 +Test: [1150/2387] eta: 0:05:39 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.5016 (0.3283) acc1: 73.9583 (87.8955) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1979 data: 0.1069 max mem: 3534 +Test: [1160/2387] eta: 0:05:35 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1415 (0.3265) acc1: 100.0000 (87.9975) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1948 data: 0.1043 max mem: 3534 +Test: [1170/2387] eta: 0:05:32 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1229 (0.3247) acc1: 100.0000 (88.0989) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1923 data: 0.1016 max mem: 3534 +Test: [1180/2387] eta: 0:05:28 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1271 (0.3230) acc1: 100.0000 (88.1983) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1927 data: 0.1004 max mem: 3534 +Test: [1190/2387] eta: 0:05:25 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1362 (0.3215) acc1: 100.0000 (88.2963) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1884 data: 0.0959 max mem: 3534 +Test: [1200/2387] eta: 0:05:21 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1480 (0.3204) acc1: 100.0000 (88.3769) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1830 data: 0.0912 max mem: 3535 +Test: [1210/2387] eta: 0:05:17 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1611 (0.3195) acc1: 99.2188 (88.4466) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1740 data: 0.0825 max mem: 3535 +Test: [1220/2387] eta: 0:05:13 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1250 (0.3178) acc1: 100.0000 (88.5410) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1470 data: 0.0539 max mem: 3535 +Test: [1230/2387] eta: 0:05:10 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1036 (0.3160) acc1: 100.0000 (88.6333) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1523 data: 0.0595 max mem: 3535 +Test: [1240/2387] eta: 0:05:06 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.0982 (0.3143) acc1: 100.0000 (88.7249) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1745 data: 0.0840 max mem: 3535 +Test: [1250/2387] eta: 0:05:03 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.0978 (0.3126) acc1: 100.0000 (88.8150) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1734 data: 0.0829 max mem: 3535 +Test: [1260/2387] eta: 0:04:58 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1013 (0.3109) acc1: 100.0000 (88.9037) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1397 data: 0.0466 max mem: 3535 +Test: [1270/2387] eta: 0:04:55 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1284 (0.3098) acc1: 100.0000 (88.9889) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1260 data: 0.0328 max mem: 3535 +Test: [1280/2387] eta: 0:04:51 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1587 (0.3087) acc1: 99.7396 (89.0633) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1582 data: 0.0662 max mem: 3535 +Test: [1290/2387] eta: 0:04:48 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1444 (0.3074) acc1: 99.2188 (89.1444) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1739 data: 0.0823 max mem: 3535 +Test: [1300/2387] eta: 0:04:44 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1440 (0.3062) acc1: 99.7396 (89.2210) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1710 data: 0.0812 max mem: 3535 +Test: [1310/2387] eta: 0:04:41 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1419 (0.3050) acc1: 99.4792 (89.2967) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1540 data: 0.0631 max mem: 3535 +Test: [1320/2387] eta: 0:04:37 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1503 (0.3039) acc1: 99.7396 (89.3712) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1546 data: 0.0610 max mem: 3535 +Test: [1330/2387] eta: 0:04:34 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1784 (0.3048) acc1: 96.8750 (89.2944) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.1699 data: 0.0752 max mem: 3535 +Test: [1340/2387] eta: 0:04:31 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.5121 (0.3062) acc1: 73.9583 (89.1742) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1793 data: 0.0843 max mem: 3535 +Test: [1350/2387] eta: 0:04:28 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.0986 (0.3046) acc1: 100.0000 (89.2543) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1673 data: 0.0742 max mem: 3535 +Test: [1360/2387] eta: 0:04:24 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.0746 (0.3029) acc1: 100.0000 (89.3333) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1691 data: 0.0780 max mem: 3535 +Test: [1370/2387] eta: 0:04:21 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.0723 (0.3012) acc1: 100.0000 (89.4111) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1848 data: 0.0935 max mem: 3535 +Test: [1380/2387] eta: 0:04:18 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.0723 (0.2995) acc1: 100.0000 (89.4877) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1819 data: 0.0892 max mem: 3535 +Test: [1390/2387] eta: 0:04:15 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.0777 (0.2980) acc1: 100.0000 (89.5633) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1895 data: 0.0957 max mem: 3535 +Test: [1400/2387] eta: 0:04:12 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.0835 (0.2966) acc1: 100.0000 (89.6367) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1860 data: 0.0934 max mem: 3535 +Test: [1410/2387] eta: 0:04:09 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1107 (0.2962) acc1: 100.0000 (89.6712) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1751 data: 0.0824 max mem: 3535 +Test: [1420/2387] eta: 0:04:06 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2411 (0.2958) acc1: 96.8750 (89.7191) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1783 data: 0.0857 max mem: 3535 +Test: [1430/2387] eta: 0:04:03 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2389 (0.2957) acc1: 96.8750 (89.7457) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1988 data: 0.1053 max mem: 3536 +Test: [1440/2387] eta: 0:04:00 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2660 (0.2960) acc1: 93.4896 (89.7348) acc5: 100.0000 (100.0000) acc1_neg: 43.229 acc5_neg: 100.000 time: 0.1934 data: 0.1000 max mem: 3536 +Test: [1450/2387] eta: 0:03:57 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.3869 (0.2977) acc1: 74.7396 (89.5826) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1746 data: 0.0821 max mem: 3536 +Test: [1460/2387] eta: 0:03:54 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2124 (0.2971) acc1: 95.0521 (89.6414) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1870 data: 0.0946 max mem: 3536 +Test: [1470/2387] eta: 0:03:51 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1717 (0.2961) acc1: 99.4792 (89.7094) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1872 data: 0.0955 max mem: 3536 +Test: [1480/2387] eta: 0:03:48 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1573 (0.2952) acc1: 99.4792 (89.7718) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1857 data: 0.0948 max mem: 3536 +Test: [1490/2387] eta: 0:03:45 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1667 (0.2944) acc1: 99.2188 (89.8350) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1883 data: 0.0980 max mem: 3536 +Test: [1500/2387] eta: 0:03:42 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1738 (0.2936) acc1: 99.2188 (89.8958) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1876 data: 0.0973 max mem: 3536 +Test: [1510/2387] eta: 0:03:39 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2064 (0.2936) acc1: 98.1771 (89.9242) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1891 data: 0.0977 max mem: 3536 +Test: [1520/2387] eta: 0:03:37 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2429 (0.2930) acc1: 97.6562 (89.9768) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1896 data: 0.0974 max mem: 3536 +Test: [1530/2387] eta: 0:03:34 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1835 (0.2923) acc1: 98.6979 (90.0341) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1902 data: 0.0983 max mem: 3536 +Test: [1540/2387] eta: 0:03:31 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1814 (0.2916) acc1: 99.7396 (90.0961) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1871 data: 0.0944 max mem: 3536 +Test: [1550/2387] eta: 0:03:28 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1818 (0.2908) acc1: 99.7396 (90.1561) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1992 data: 0.1076 max mem: 3536 +Test: [1560/2387] eta: 0:03:25 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1877 (0.2903) acc1: 99.4792 (90.2128) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1929 data: 0.1021 max mem: 3536 +Test: [1570/2387] eta: 0:03:23 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2377 (0.2905) acc1: 98.1771 (90.2230) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1847 data: 0.0916 max mem: 3536 +Test: [1580/2387] eta: 0:03:20 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2825 (0.2904) acc1: 94.2708 (90.2550) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1844 data: 0.0878 max mem: 3536 +Test: [1590/2387] eta: 0:03:17 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2110 (0.2897) acc1: 97.3958 (90.3089) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1567 data: 0.0585 max mem: 3536 +Test: [1600/2387] eta: 0:03:14 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1827 (0.2891) acc1: 99.2188 (90.3634) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1805 data: 0.0843 max mem: 3536 +Test: [1610/2387] eta: 0:03:11 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1827 (0.2884) acc1: 99.2188 (90.4160) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1981 data: 0.1049 max mem: 3536 +Test: [1620/2387] eta: 0:03:08 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1949 (0.2880) acc1: 97.3958 (90.4494) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1673 data: 0.0738 max mem: 3536 +Test: [1630/2387] eta: 0:03:06 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2228 (0.2885) acc1: 94.7917 (90.4098) acc5: 100.0000 (100.0000) acc1_neg: 73.958 acc5_neg: 100.000 time: 0.1699 data: 0.0759 max mem: 3536 +Test: [1640/2387] eta: 0:03:03 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.5440 (0.2905) acc1: 71.3542 (90.2494) acc5: 100.0000 (100.0000) acc1_neg: 66.927 acc5_neg: 100.000 time: 0.1734 data: 0.0821 max mem: 3536 +Test: [1650/2387] eta: 0:03:00 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.5568 (0.2921) acc1: 66.1458 (90.1127) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.1749 data: 0.0822 max mem: 3536 +Test: [1660/2387] eta: 0:02:57 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.5339 (0.2935) acc1: 68.4896 (89.9913) acc5: 100.0000 (100.0000) acc1_neg: 72.917 acc5_neg: 100.000 time: 0.1784 data: 0.0849 max mem: 3537 +Test: [1670/2387] eta: 0:02:54 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.5171 (0.2948) acc1: 72.9167 (89.8935) acc5: 100.0000 (100.0000) acc1_neg: 77.344 acc5_neg: 100.000 time: 0.1656 data: 0.0746 max mem: 3537 +Test: [1680/2387] eta: 0:02:52 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.5066 (0.2961) acc1: 72.9167 (89.7919) acc5: 100.0000 (100.0000) acc1_neg: 72.917 acc5_neg: 100.000 time: 0.1657 data: 0.0748 max mem: 3537 +Test: [1690/2387] eta: 0:02:49 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.5258 (0.2977) acc1: 70.0521 (89.6634) acc5: 100.0000 (100.0000) acc1_neg: 70.573 acc5_neg: 100.000 time: 0.1557 data: 0.0649 max mem: 3537 +Test: [1700/2387] eta: 0:02:46 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.5471 (0.2985) acc1: 69.7917 (89.5818) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1516 data: 0.0612 max mem: 3537 +Test: [1710/2387] eta: 0:02:43 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1383 (0.2975) acc1: 99.2188 (89.6410) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1638 data: 0.0726 max mem: 3537 +Test: [1720/2387] eta: 0:02:41 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1158 (0.2964) acc1: 100.0000 (89.7008) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1738 data: 0.0794 max mem: 3537 +Test: [1730/2387] eta: 0:02:38 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1053 (0.2953) acc1: 100.0000 (89.7583) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1553 data: 0.0611 max mem: 3537 +Test: [1740/2387] eta: 0:02:35 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1053 (0.2942) acc1: 100.0000 (89.8170) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1360 data: 0.0426 max mem: 3537 +Test: [1750/2387] eta: 0:02:33 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1108 (0.2933) acc1: 100.0000 (89.8733) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1820 data: 0.0874 max mem: 3537 +Test: [1760/2387] eta: 0:02:30 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1466 (0.2928) acc1: 98.9583 (89.9137) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1895 data: 0.0974 max mem: 3537 +Test: [1770/2387] eta: 0:02:27 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1944 (0.2922) acc1: 98.4375 (89.9586) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1320 data: 0.0410 max mem: 3537 +Test: [1780/2387] eta: 0:02:24 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1595 (0.2914) acc1: 99.4792 (90.0109) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1304 data: 0.0389 max mem: 3537 +Test: [1790/2387] eta: 0:02:22 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1295 (0.2905) acc1: 99.7396 (90.0623) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1578 data: 0.0665 max mem: 3537 +Test: [1800/2387] eta: 0:02:19 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1271 (0.2896) acc1: 99.7396 (90.1162) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1791 data: 0.0885 max mem: 3537 +Test: [1810/2387] eta: 0:02:17 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1300 (0.2887) acc1: 100.0000 (90.1677) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1689 data: 0.0780 max mem: 3537 +Test: [1820/2387] eta: 0:02:14 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1396 (0.2883) acc1: 99.2188 (90.2034) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1345 data: 0.0423 max mem: 3537 +Test: [1830/2387] eta: 0:02:11 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2221 (0.2879) acc1: 98.9583 (90.2416) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1487 data: 0.0571 max mem: 3537 +Test: [1840/2387] eta: 0:02:09 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1440 (0.2871) acc1: 99.4792 (90.2932) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1997 data: 0.1078 max mem: 3537 +Test: [1850/2387] eta: 0:02:06 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1423 (0.2863) acc1: 99.7396 (90.3439) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2092 data: 0.1167 max mem: 3537 +Test: [1860/2387] eta: 0:02:04 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1423 (0.2856) acc1: 99.7396 (90.3940) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1836 data: 0.0921 max mem: 3537 +Test: [1870/2387] eta: 0:02:01 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1419 (0.2848) acc1: 99.7396 (90.4434) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1907 data: 0.0993 max mem: 3537 +Test: [1880/2387] eta: 0:01:59 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1419 (0.2840) acc1: 99.7396 (90.4928) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2074 data: 0.1169 max mem: 3538 +Test: [1890/2387] eta: 0:01:56 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1424 (0.2833) acc1: 99.7396 (90.5425) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2017 data: 0.1087 max mem: 3538 +Test: [1900/2387] eta: 0:01:54 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1444 (0.2826) acc1: 99.7396 (90.5884) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1987 data: 0.1041 max mem: 3538 +Test: [1910/2387] eta: 0:01:52 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2401 (0.2826) acc1: 96.0938 (90.6122) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.2173 data: 0.1234 max mem: 3538 +Test: [1920/2387] eta: 0:01:49 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2735 (0.2828) acc1: 95.8333 (90.6147) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2419 data: 0.1427 max mem: 3538 +Test: [1930/2387] eta: 0:01:47 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2140 (0.2824) acc1: 98.9583 (90.6537) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2380 data: 0.1412 max mem: 3538 +Test: [1940/2387] eta: 0:01:45 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2135 (0.2820) acc1: 98.9583 (90.6942) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2303 data: 0.1392 max mem: 3538 +Test: [1950/2387] eta: 0:01:42 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2302 (0.2819) acc1: 98.4375 (90.7134) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1884 data: 0.0964 max mem: 3538 +Test: [1960/2387] eta: 0:01:40 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2302 (0.2817) acc1: 96.8750 (90.7377) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1775 data: 0.0866 max mem: 3538 +Test: [1970/2387] eta: 0:01:37 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.3416 (0.2831) acc1: 89.5833 (90.6379) acc5: 100.0000 (100.0000) acc1_neg: 65.625 acc5_neg: 100.000 time: 0.2108 data: 0.1185 max mem: 3538 +Test: [1980/2387] eta: 0:01:35 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.5258 (0.2837) acc1: 66.1458 (90.5704) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2251 data: 0.1312 max mem: 3538 +Test: [1990/2387] eta: 0:01:33 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1122 (0.2829) acc1: 99.2188 (90.6157) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2183 data: 0.1267 max mem: 3538 +Test: [2000/2387] eta: 0:01:30 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1122 (0.2820) acc1: 99.4792 (90.6607) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1850 data: 0.0947 max mem: 3538 +Test: [2010/2387] eta: 0:01:28 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1127 (0.2811) acc1: 99.7396 (90.7058) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1860 data: 0.0934 max mem: 3538 +Test: [2020/2387] eta: 0:01:25 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1115 (0.2803) acc1: 99.7396 (90.7497) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1903 data: 0.0957 max mem: 3538 +Test: [2030/2387] eta: 0:01:23 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1115 (0.2795) acc1: 99.4792 (90.7935) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1951 data: 0.1011 max mem: 3538 +Test: [2040/2387] eta: 0:01:20 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1105 (0.2787) acc1: 99.7396 (90.8364) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1876 data: 0.0950 max mem: 3538 +Test: [2050/2387] eta: 0:01:18 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1105 (0.2778) acc1: 99.7396 (90.8800) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1881 data: 0.0968 max mem: 3538 +Test: [2060/2387] eta: 0:01:16 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1215 (0.2771) acc1: 99.7396 (90.9213) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1911 data: 0.0987 max mem: 3538 +Test: [2070/2387] eta: 0:01:13 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1251 (0.2764) acc1: 99.4792 (90.9636) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1857 data: 0.0930 max mem: 3538 +Test: [2080/2387] eta: 0:01:11 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1251 (0.2757) acc1: 99.7396 (91.0057) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1827 data: 0.0911 max mem: 3538 +Test: [2090/2387] eta: 0:01:08 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1252 (0.2749) acc1: 99.7396 (91.0471) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1842 data: 0.0926 max mem: 3538 +Test: [2100/2387] eta: 0:01:06 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1247 (0.2742) acc1: 99.7396 (91.0884) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1882 data: 0.0959 max mem: 3538 +Test: [2110/2387] eta: 0:01:04 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1249 (0.2735) acc1: 99.7396 (91.1289) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1864 data: 0.0934 max mem: 3539 +Test: [2120/2387] eta: 0:01:01 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1360 (0.2735) acc1: 99.2188 (91.1383) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.1927 data: 0.1004 max mem: 3539 +Test: [2130/2387] eta: 0:00:59 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2106 (0.2731) acc1: 97.6562 (91.1717) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1829 data: 0.0906 max mem: 3539 +Test: [2140/2387] eta: 0:00:57 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1858 (0.2728) acc1: 98.1771 (91.2020) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1727 data: 0.0811 max mem: 3539 +Test: [2150/2387] eta: 0:00:54 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2075 (0.2726) acc1: 97.9167 (91.2237) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1812 data: 0.0892 max mem: 3539 +Test: [2160/2387] eta: 0:00:52 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2191 (0.2725) acc1: 96.6146 (91.2404) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1883 data: 0.0928 max mem: 3539 +Test: [2170/2387] eta: 0:00:50 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2775 (0.2734) acc1: 91.9271 (91.1717) acc5: 100.0000 (100.0000) acc1_neg: 66.667 acc5_neg: 100.000 time: 0.2006 data: 0.1048 max mem: 3539 +Test: [2180/2387] eta: 0:00:47 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.6401 (0.2759) acc1: 59.1146 (90.9831) acc5: 100.0000 (100.0000) acc1_neg: 37.760 acc5_neg: 100.000 time: 0.2006 data: 0.1078 max mem: 3539 +Test: [2190/2387] eta: 0:00:45 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.6148 (0.2758) acc1: 60.4167 (90.9765) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2128 data: 0.1195 max mem: 3539 +Test: [2200/2387] eta: 0:00:43 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1734 (0.2755) acc1: 98.9583 (91.0100) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1946 data: 0.1007 max mem: 3539 +Test: [2210/2387] eta: 0:00:40 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1816 (0.2750) acc1: 99.4792 (91.0478) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1580 data: 0.0647 max mem: 3539 +Test: [2220/2387] eta: 0:00:38 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1677 (0.2745) acc1: 99.7396 (91.0853) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1730 data: 0.0807 max mem: 3539 +Test: [2230/2387] eta: 0:00:35 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1679 (0.2740) acc1: 99.4792 (91.1211) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1868 data: 0.0951 max mem: 3539 +Test: [2240/2387] eta: 0:00:33 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1774 (0.2736) acc1: 99.4792 (91.1558) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1809 data: 0.0892 max mem: 3539 +Test: [2250/2387] eta: 0:00:31 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2099 (0.2739) acc1: 97.1354 (91.1540) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1749 data: 0.0810 max mem: 3539 +Test: [2260/2387] eta: 0:00:29 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2616 (0.2738) acc1: 96.0938 (91.1753) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1810 data: 0.0846 max mem: 3539 +Test: [2270/2387] eta: 0:00:26 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1794 (0.2733) acc1: 98.1771 (91.2102) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1931 data: 0.0992 max mem: 3539 +Test: [2280/2387] eta: 0:00:24 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1537 (0.2729) acc1: 99.7396 (91.2432) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2010 data: 0.1099 max mem: 3539 +Test: [2290/2387] eta: 0:00:22 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1574 (0.2724) acc1: 99.4792 (91.2777) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1997 data: 0.1079 max mem: 3539 +Test: [2300/2387] eta: 0:00:19 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.1855 (0.2723) acc1: 98.9583 (91.2979) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1842 data: 0.0929 max mem: 3539 +Test: [2310/2387] eta: 0:00:17 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2495 (0.2726) acc1: 94.2708 (91.2696) acc5: 100.0000 (100.0000) acc1_neg: 66.667 acc5_neg: 100.000 time: 0.1698 data: 0.0779 max mem: 3539 +Test: [2320/2387] eta: 0:00:15 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.4421 (0.2736) acc1: 77.0833 (91.1996) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.1657 data: 0.0736 max mem: 3539 +Test: [2330/2387] eta: 0:00:12 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.4223 (0.2739) acc1: 80.9896 (91.1897) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1599 data: 0.0699 max mem: 3539 +Test: [2340/2387] eta: 0:00:10 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2758 (0.2739) acc1: 92.9688 (91.2049) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1510 data: 0.0608 max mem: 3540 +Test: [2350/2387] eta: 0:00:08 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.2713 (0.2739) acc1: 95.3125 (91.2217) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1352 data: 0.0448 max mem: 3540 +Test: [2360/2387] eta: 0:00:06 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.3443 (0.2745) acc1: 93.4896 (91.1848) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1292 data: 0.0391 max mem: 3540 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.3365 (0.2746) acc1: 91.4062 (91.1871) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1263 data: 0.0366 max mem: 3540 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.4089 (0.2764) acc1: 79.4271 (91.0362) acc5: 100.0000 (100.0000) acc1_neg: 51.823 acc5_neg: 100.000 time: 0.1208 data: 0.0312 max mem: 3540 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 90.868 acc5_pos: 100.000 loss: 0.5934 (0.2767) acc1: 60.6771 (90.9921) acc5: 100.0000 (100.0000) acc1_neg: 97.674 acc5_neg: 100.000 time: 0.1138 data: 0.0270 max mem: 3540 +Test: Total time: 0:08:56 (0.2249 s / it) +* Overall Acc@1 90.993 Acc@5 100.000 + target == real Acc@1 64.184 Acc@5 100.000 + target == fake Acc@1 92.875 Acc@5 100.000 + loss 0.277 + PR-AUC 0.9932 ROC-AUC 0.9144 EER 16.44% (thr=0.6689) +Accuracy of the network on the 916267 test images: 91.0% +[rank0]:[W725 18:05:00.526288440 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:05:19.025000 2571403 torch/distributed/run.py:766] +W0725 18:05:19.025000 2571403 torch/distributed/run.py:766] ***************************************** +W0725 18:05:19.025000 2571403 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:05:19.025000 2571403 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:05:47.900483691 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:05:47.044487651 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:05:47.293613337 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:05:47.463886283 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:05:47.528497649 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:05:47.547249340 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:05:47.557412279 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:05:48.571148057 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed2_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed2_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:02:06 acc1_pos: 60.156 acc5_pos: 100.000 loss: 0.7284 (0.7284) acc1: 60.1562 (60.1562) acc5: 100.0000 (100.0000) time: 7.2075 data: 5.4971 max mem: 3531 +Test: [ 10/517] eta: 0:06:13 acc1_pos: 23.177 acc5_pos: 100.000 loss: 0.8606 (0.9247) acc1: 36.7188 (37.8551) acc5: 100.0000 (100.0000) time: 0.7370 data: 0.5001 max mem: 3531 +Test: [ 20/517] eta: 0:03:35 acc1_pos: 14.583 acc5_pos: 100.000 loss: 0.9938 (0.9844) acc1: 28.6458 (31.8452) acc5: 100.0000 (100.0000) time: 0.0939 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:02:57 acc1_pos: 24.219 acc5_pos: 100.000 loss: 1.0484 (0.9775) acc1: 26.5625 (33.5938) acc5: 100.0000 (100.0000) time: 0.1594 data: 0.0642 max mem: 3531 +Test: [ 40/517] eta: 0:02:41 acc1_pos: 56.250 acc5_pos: 100.000 loss: 0.9165 (0.9663) acc1: 34.8958 (34.4893) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1470 max mem: 3531 +Test: [ 50/517] eta: 0:02:29 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.9066 (0.9488) acc1: 35.6771 (36.2132) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.2494 data: 0.1555 max mem: 3531 +Test: [ 60/517] eta: 0:02:20 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.5408 (0.8747) acc1: 67.4479 (43.1609) acc5: 100.0000 (100.0000) acc1_neg: 72.135 acc5_neg: 100.000 time: 0.2452 data: 0.1519 max mem: 3531 +Test: [ 70/517] eta: 0:02:13 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.4235 (0.8018) acc1: 87.5000 (50.0660) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2499 data: 0.1570 max mem: 3531 +Test: [ 80/517] eta: 0:02:08 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.3467 (0.7457) acc1: 94.0104 (55.3691) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.2496 data: 0.1559 max mem: 3531 +Test: [ 90/517] eta: 0:02:03 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.3931 (0.7127) acc1: 88.5417 (58.4564) acc5: 100.0000 (100.0000) acc1_neg: 78.646 acc5_neg: 100.000 time: 0.2549 data: 0.1605 max mem: 3531 +Test: [100/517] eta: 0:02:00 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.3873 (0.6773) acc1: 88.5417 (61.8760) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.2649 data: 0.1688 max mem: 3531 +Test: [110/517] eta: 0:01:55 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.3310 (0.6467) acc1: 94.0104 (64.5974) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2599 data: 0.1660 max mem: 3531 +Test: [120/517] eta: 0:01:52 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2893 (0.6159) acc1: 96.3542 (67.3834) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2554 data: 0.1645 max mem: 3531 +Test: [130/517] eta: 0:01:48 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2604 (0.5881) acc1: 98.9583 (69.7817) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2549 data: 0.1605 max mem: 3531 +Test: [140/517] eta: 0:01:44 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2476 (0.5648) acc1: 99.4792 (71.8787) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2502 data: 0.1524 max mem: 3531 +Test: [150/517] eta: 0:01:41 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2467 (0.5433) acc1: 99.4792 (73.6962) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2449 data: 0.1492 max mem: 3531 +Test: [160/517] eta: 0:01:38 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2682 (0.5284) acc1: 98.4375 (75.1100) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2541 data: 0.1582 max mem: 3531 +Test: [170/517] eta: 0:01:34 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2831 (0.5136) acc1: 97.6562 (76.3919) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2599 data: 0.1628 max mem: 3531 +Test: [180/517] eta: 0:01:32 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2848 (0.5017) acc1: 96.8750 (77.4445) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2805 data: 0.1863 max mem: 3531 +Test: [190/517] eta: 0:01:29 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.3083 (0.4919) acc1: 96.3542 (78.4427) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2649 data: 0.1711 max mem: 3531 +Test: [200/517] eta: 0:01:25 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2783 (0.4814) acc1: 96.8750 (79.3973) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2279 data: 0.1320 max mem: 3531 +Test: [210/517] eta: 0:01:23 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2586 (0.4706) acc1: 98.9583 (80.3182) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2499 data: 0.1553 max mem: 3532 +Test: [220/517] eta: 0:01:20 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2512 (0.4608) acc1: 98.6979 (81.1381) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2665 data: 0.1724 max mem: 3532 +Test: [230/517] eta: 0:01:17 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2850 (0.4559) acc1: 96.6146 (81.7280) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.2708 data: 0.1764 max mem: 3532 +Test: [240/517] eta: 0:01:14 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.3542 (0.4515) acc1: 94.7917 (82.2204) acc5: 100.0000 (100.0000) acc1_neg: 77.344 acc5_neg: 100.000 time: 0.2449 data: 0.1502 max mem: 3532 +Test: [250/517] eta: 0:01:11 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.3177 (0.4461) acc1: 97.3958 (82.8374) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2390 data: 0.1434 max mem: 3532 +Test: [260/517] eta: 0:01:08 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.3047 (0.4407) acc1: 98.6979 (83.4281) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2658 data: 0.1721 max mem: 3532 +Test: [270/517] eta: 0:01:06 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.3234 (0.4374) acc1: 97.9167 (83.8801) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2649 data: 0.1695 max mem: 3532 +Test: [280/517] eta: 0:01:03 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.3234 (0.4331) acc1: 97.9167 (84.3296) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2540 data: 0.1572 max mem: 3532 +Test: [290/517] eta: 0:01:00 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2981 (0.4283) acc1: 98.6979 (84.7947) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2499 data: 0.1568 max mem: 3532 +Test: [300/517] eta: 0:00:57 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2684 (0.4229) acc1: 98.6979 (85.2549) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2558 data: 0.1638 max mem: 3532 +Test: [310/517] eta: 0:00:55 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.3377 (0.4213) acc1: 97.3958 (85.5691) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2599 data: 0.1671 max mem: 3532 +Test: [320/517] eta: 0:00:52 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.3471 (0.4179) acc1: 95.8333 (85.8986) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2545 data: 0.1615 max mem: 3532 +Test: [330/517] eta: 0:00:49 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2691 (0.4133) acc1: 96.6146 (86.2160) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2549 data: 0.1632 max mem: 3532 +Test: [340/517] eta: 0:00:46 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2560 (0.4100) acc1: 98.4375 (86.4476) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.2595 data: 0.1672 max mem: 3532 +Test: [350/517] eta: 0:00:44 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.4095 (0.4113) acc1: 90.6250 (86.4086) acc5: 100.0000 (100.0000) acc1_neg: 66.146 acc5_neg: 100.000 time: 0.2449 data: 0.1498 max mem: 3532 +Test: [360/517] eta: 0:00:41 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.4176 (0.4120) acc1: 85.1562 (86.3595) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.2399 data: 0.1447 max mem: 3532 +Test: [370/517] eta: 0:00:38 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.3495 (0.4090) acc1: 91.1458 (86.6324) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2555 data: 0.1627 max mem: 3532 +Test: [380/517] eta: 0:00:36 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2926 (0.4059) acc1: 96.8750 (86.8965) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2599 data: 0.1681 max mem: 3532 +Test: [390/517] eta: 0:00:33 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2678 (0.4024) acc1: 97.1354 (87.1543) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2602 data: 0.1686 max mem: 3532 +Test: [400/517] eta: 0:00:30 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2589 (0.4002) acc1: 97.1354 (87.2779) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2549 data: 0.1625 max mem: 3532 +Test: [410/517] eta: 0:00:28 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.3623 (0.4002) acc1: 90.6250 (87.2535) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.2555 data: 0.1615 max mem: 3532 +Test: [420/517] eta: 0:00:25 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.3998 (0.3996) acc1: 91.4062 (87.3720) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2554 data: 0.1632 max mem: 3532 +Test: [430/517] eta: 0:00:22 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.3429 (0.3978) acc1: 92.1875 (87.4958) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.2485 data: 0.1576 max mem: 3532 +Test: [440/517] eta: 0:00:20 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.3181 (0.3965) acc1: 95.0521 (87.6966) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2544 data: 0.1596 max mem: 3533 +Test: [450/517] eta: 0:00:17 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.3226 (0.3948) acc1: 95.5729 (87.8314) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2540 data: 0.1546 max mem: 3533 +Test: [460/517] eta: 0:00:14 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.3226 (0.3936) acc1: 94.2708 (87.9497) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2390 data: 0.1398 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.3410 (0.3928) acc1: 94.5312 (88.0783) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.2302 data: 0.1340 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.3681 (0.3930) acc1: 93.2292 (88.1329) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2199 data: 0.1256 max mem: 3533 +Test: [490/517] eta: 0:00:06 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.3385 (0.3907) acc1: 96.0938 (88.3449) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1973 data: 0.1030 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.3079 (0.3896) acc1: 98.6979 (88.5230) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1739 data: 0.0820 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.3392 (0.3902) acc1: 94.0104 (88.4158) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1641 data: 0.0746 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.3279 (0.3893) acc1: 93.7500 (88.4945) acc5: 100.0000 (100.0000) acc1_neg: 91.873 acc5_neg: 100.000 time: 0.1410 data: 0.0499 max mem: 3533 +Test: Total time: 0:02:11 (0.2535 s / it) +* Overall Acc@1 88.495 Acc@5 100.000 + target == real Acc@1 33.759 Acc@5 100.000 + target == fake Acc@1 94.173 Acc@5 100.000 + loss 0.389 + PR-AUC 0.9594 ROC-AUC 0.7539 EER 31.29% (thr=0.6919) +Accuracy of the network on the 198427 test images: 88.5% +[rank0]:[W725 18:08:07.457362170 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0725 18:08:27.373000 2571991 torch/distributed/run.py:766] +W0725 18:08:27.373000 2571991 torch/distributed/run.py:766] ***************************************** +W0725 18:08:27.373000 2571991 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:08:27.373000 2571991 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:08:53.308521252 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:08:54.620149429 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 1): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:08:54.054268855 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank1]:[W725 18:08:54.054716023 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:08:54.127992598 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 4): env:// +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:08:54.310212670 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank4]:[W725 18:08:54.310251512 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W725 18:08:54.310783815 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed2_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed2_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:28:10 acc1_pos: 7.292 acc5_pos: 100.000 loss: 1.4318 (1.4318) acc1: 7.2917 (7.2917) acc5: 100.0000 (100.0000) time: 4.3448 data: 2.4941 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/389] eta: 0:03:18 acc1_pos: 80.469 acc5_pos: 100.000 loss: 1.0225 (1.0539) acc1: 36.7188 (33.5701) acc5: 100.0000 (100.0000) time: 0.5250 data: 0.2754 max mem: 3531 +Test: [ 20/389] eta: 0:02:07 acc1_pos: 0.781 acc5_pos: 100.000 loss: 1.0268 (1.0998) acc1: 21.8750 (31.3120) acc5: 100.0000 (100.0000) time: 0.1442 data: 0.0511 max mem: 3531 +Test: [ 30/389] eta: 0:01:38 acc1_pos: 12.500 acc5_pos: 100.000 loss: 1.4271 (1.2136) acc1: 2.6042 (22.7571) acc5: 100.0000 (100.0000) time: 0.1376 data: 0.0381 max mem: 3531 +Test: [ 40/389] eta: 0:01:31 acc1_pos: 4.948 acc5_pos: 100.000 loss: 1.3400 (1.2058) acc1: 4.9479 (21.4240) acc5: 100.0000 (100.0000) time: 0.1751 data: 0.0760 max mem: 3531 +Test: [ 50/389] eta: 0:01:23 acc1_pos: 8.333 acc5_pos: 100.000 loss: 1.2885 (1.2243) acc1: 8.3333 (19.2351) acc5: 100.0000 (100.0000) time: 0.2049 data: 0.1095 max mem: 3531 +Test: [ 60/389] eta: 0:01:18 acc1_pos: 17.708 acc5_pos: 100.000 loss: 1.2890 (1.2307) acc1: 7.0312 (17.6827) acc5: 100.0000 (100.0000) time: 0.1903 data: 0.0977 max mem: 3531 +Test: [ 70/389] eta: 0:01:13 acc1_pos: 30.469 acc5_pos: 100.000 loss: 1.2846 (1.2186) acc1: 9.8958 (18.0788) acc5: 100.0000 (100.0000) time: 0.1849 data: 0.0944 max mem: 3531 +Test: [ 80/389] eta: 0:01:09 acc1_pos: 74.219 acc5_pos: 100.000 loss: 1.1526 (1.1847) acc1: 16.6667 (20.2836) acc5: 100.0000 (100.0000) time: 0.1843 data: 0.0939 max mem: 3531 +Test: [ 90/389] eta: 0:01:06 acc1_pos: 32.552 acc5_pos: 100.000 loss: 0.8065 (1.1349) acc1: 44.5312 (24.1844) acc5: 100.0000 (100.0000) time: 0.1899 data: 0.0976 max mem: 3531 +Test: [100/389] eta: 0:01:02 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.1296 (1.1713) acc1: 23.9583 (22.7259) acc5: 100.0000 (100.0000) time: 0.1770 data: 0.0847 max mem: 3531 +Test: [110/389] eta: 0:00:58 acc1_pos: 94.271 acc5_pos: 100.000 loss: 1.5019 (1.1778) acc1: 2.3438 (23.0340) acc5: 100.0000 (100.0000) time: 0.1664 data: 0.0764 max mem: 3531 +Test: [120/389] eta: 0:00:56 acc1_pos: 32.031 acc5_pos: 100.000 loss: 1.1106 (1.1681) acc1: 18.2292 (23.9519) acc5: 100.0000 (100.0000) time: 0.1738 data: 0.0825 max mem: 3531 +Test: [130/389] eta: 0:00:53 acc1_pos: 52.147 acc5_pos: 100.000 loss: 0.6935 (1.1094) acc1: 54.6875 (28.5206) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1754 data: 0.0832 max mem: 3531 +Test: [140/389] eta: 0:00:51 acc1_pos: 52.147 acc5_pos: 100.000 loss: 0.3026 (1.0571) acc1: 94.7917 (32.7774) acc5: 100.0000 (100.0000) acc1_neg: 65.104 acc5_neg: 100.000 time: 0.1840 data: 0.0883 max mem: 3531 +Test: [150/389] eta: 0:00:48 acc1_pos: 52.147 acc5_pos: 100.000 loss: 0.3076 (1.0065) acc1: 95.5729 (36.9205) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1938 data: 0.0988 max mem: 3531 +Test: [160/389] eta: 0:00:46 acc1_pos: 52.147 acc5_pos: 100.000 loss: 0.2240 (0.9552) acc1: 98.9583 (40.7819) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1901 data: 0.0989 max mem: 3531 +Test: [170/389] eta: 0:00:44 acc1_pos: 52.147 acc5_pos: 100.000 loss: 0.1578 (0.9123) acc1: 99.7396 (44.1414) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1789 data: 0.0863 max mem: 3531 +Test: [180/389] eta: 0:00:41 acc1_pos: 52.147 acc5_pos: 100.000 loss: 0.1532 (0.8736) acc1: 100.0000 (47.0678) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1698 data: 0.0763 max mem: 3531 +Test: [190/389] eta: 0:00:39 acc1_pos: 52.147 acc5_pos: 100.000 loss: 0.2360 (0.8462) acc1: 96.0938 (49.1942) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1873 data: 0.0939 max mem: 3531 +Test: [200/389] eta: 0:00:37 acc1_pos: 52.147 acc5_pos: 100.000 loss: 0.3283 (0.8204) acc1: 95.8333 (51.4420) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.1867 data: 0.0949 max mem: 3531 +Test: [210/389] eta: 0:00:35 acc1_pos: 52.147 acc5_pos: 100.000 loss: 0.2886 (0.7933) acc1: 97.1354 (53.5175) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1684 data: 0.0764 max mem: 3532 +Test: [220/389] eta: 0:00:33 acc1_pos: 52.147 acc5_pos: 100.000 loss: 0.1604 (0.7639) acc1: 100.0000 (55.6160) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1789 data: 0.0860 max mem: 3532 +Test: [230/389] eta: 0:00:31 acc1_pos: 52.147 acc5_pos: 100.000 loss: 0.2050 (0.7439) acc1: 99.7396 (57.2105) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1865 data: 0.0954 max mem: 3532 +Test: [240/389] eta: 0:00:28 acc1_pos: 52.147 acc5_pos: 100.000 loss: 0.2557 (0.7262) acc1: 93.7500 (58.5765) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1696 data: 0.0789 max mem: 3532 +Test: [250/389] eta: 0:00:26 acc1_pos: 52.147 acc5_pos: 100.000 loss: 0.2354 (0.7071) acc1: 96.6146 (60.0992) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1584 data: 0.0677 max mem: 3532 +Test: [260/389] eta: 0:00:24 acc1_pos: 52.147 acc5_pos: 100.000 loss: 0.2444 (0.6914) acc1: 99.7396 (61.3127) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1762 data: 0.0853 max mem: 3532 +Test: [270/389] eta: 0:00:22 acc1_pos: 52.147 acc5_pos: 100.000 loss: 0.3068 (0.6794) acc1: 98.4375 (62.3693) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1755 data: 0.0845 max mem: 3532 +Test: [280/389] eta: 0:00:20 acc1_pos: 52.147 acc5_pos: 100.000 loss: 0.2945 (0.6657) acc1: 98.1771 (63.5695) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1681 data: 0.0770 max mem: 3532 +Test: [290/389] eta: 0:00:18 acc1_pos: 52.147 acc5_pos: 100.000 loss: 0.2649 (0.6513) acc1: 98.4375 (64.7167) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1697 data: 0.0766 max mem: 3532 +Test: [300/389] eta: 0:00:16 acc1_pos: 52.147 acc5_pos: 100.000 loss: 0.3132 (0.6403) acc1: 96.6146 (65.7392) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1799 data: 0.0878 max mem: 3532 +Test: [310/389] eta: 0:00:14 acc1_pos: 52.147 acc5_pos: 100.000 loss: 0.3562 (0.6304) acc1: 95.5729 (66.5880) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.1843 data: 0.0928 max mem: 3532 +Test: [320/389] eta: 0:00:12 acc1_pos: 52.147 acc5_pos: 100.000 loss: 0.2808 (0.6190) acc1: 95.0521 (67.5274) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1508 data: 0.0573 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 52.147 acc5_pos: 100.000 loss: 0.2417 (0.6075) acc1: 98.9583 (68.4652) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1600 data: 0.0678 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 52.147 acc5_pos: 100.000 loss: 0.1975 (0.5945) acc1: 99.7396 (69.3464) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2241 data: 0.1342 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 52.147 acc5_pos: 100.000 loss: 0.1769 (0.5963) acc1: 99.7396 (69.0126) acc5: 100.0000 (100.0000) acc1_neg: 31.250 acc5_neg: 100.000 time: 0.2526 data: 0.1588 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 52.147 acc5_pos: 100.000 loss: 0.4725 (0.5959) acc1: 78.6458 (69.0010) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.1690 data: 0.0746 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 52.147 acc5_pos: 100.000 loss: 0.2893 (0.5857) acc1: 83.0729 (69.7959) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.0906 data: 0.0004 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 52.147 acc5_pos: 100.000 loss: 0.2419 (0.5783) acc1: 99.4792 (70.4075) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1168 data: 0.0273 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 52.147 acc5_pos: 100.000 loss: 0.2555 (0.5722) acc1: 97.1354 (70.8399) acc5: 100.0000 (100.0000) acc1_neg: 78.962 acc5_neg: 100.000 time: 0.1364 data: 0.0434 max mem: 3532 +Test: Total time: 0:01:11 (0.1846 s / it) +* Overall Acc@1 70.840 Acc@5 100.000 + target == real Acc@1 24.435 Acc@5 100.000 + target == fake Acc@1 92.410 Acc@5 100.000 + loss 0.572 + PR-AUC 0.8369 ROC-AUC 0.7098 EER 35.50% (thr=0.7627) +Accuracy of the network on the 149358 test images: 70.8% +[rank0]:[W725 18:10:17.640595753 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:10:36.739000 2572560 torch/distributed/run.py:766] +W0725 18:10:36.739000 2572560 torch/distributed/run.py:766] ***************************************** +W0725 18:10:36.739000 2572560 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:10:36.739000 2572560 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:11:07.847247772 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:11:07.455380286 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:11:08.971340319 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:11:08.003623851 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:11:08.045086526 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:11:08.127579298 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank1]:[W725 18:11:08.127578885 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:11:08.149812138 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed2_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed2_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:31:15 acc1_pos: 84.115 acc5_pos: 100.000 loss: 0.3236 (0.3236) acc1: 84.1146 (84.1146) acc5: 100.0000 (100.0000) time: 5.6308 data: 3.7657 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/333] eta: 0:03:11 acc1_pos: 27.604 acc5_pos: 100.000 loss: 0.3236 (0.5897) acc1: 84.1146 (67.1638) acc5: 100.0000 (100.0000) time: 0.5937 data: 0.3426 max mem: 3531 +Test: [ 20/333] eta: 0:01:50 acc1_pos: 32.292 acc5_pos: 100.000 loss: 1.1185 (0.8885) acc1: 31.7708 (47.7307) acc5: 100.0000 (100.0000) time: 0.0900 data: 0.0004 max mem: 3531 +Test: [ 30/333] eta: 0:01:42 acc1_pos: 47.396 acc5_pos: 100.000 loss: 1.0207 (0.8654) acc1: 32.5521 (47.5554) acc5: 100.0000 (100.0000) time: 0.1969 data: 0.1065 max mem: 3531 +Test: [ 40/333] eta: 0:01:39 acc1_pos: 48.958 acc5_pos: 100.000 loss: 0.7761 (0.8408) acc1: 47.3958 (48.2025) acc5: 100.0000 (100.0000) time: 0.3275 data: 0.2335 max mem: 3531 +Test: [ 50/333] eta: 0:01:26 acc1_pos: 52.865 acc5_pos: 100.000 loss: 0.7541 (0.8167) acc1: 52.8646 (49.7447) acc5: 100.0000 (100.0000) time: 0.2536 data: 0.1592 max mem: 3531 +Test: [ 60/333] eta: 0:01:15 acc1_pos: 47.656 acc5_pos: 100.000 loss: 0.7301 (0.8047) acc1: 53.1250 (50.3586) acc5: 100.0000 (100.0000) time: 0.1459 data: 0.0509 max mem: 3531 +Test: [ 70/333] eta: 0:01:10 acc1_pos: 45.833 acc5_pos: 100.000 loss: 0.7301 (0.7927) acc1: 52.0833 (51.3351) acc5: 100.0000 (100.0000) time: 0.1714 data: 0.0769 max mem: 3531 +Test: [ 80/333] eta: 0:01:05 acc1_pos: 59.375 acc5_pos: 100.000 loss: 0.7396 (0.7899) acc1: 54.9479 (51.4564) acc5: 100.0000 (100.0000) time: 0.2077 data: 0.1167 max mem: 3531 +Test: [ 90/333] eta: 0:01:01 acc1_pos: 61.719 acc5_pos: 100.000 loss: 0.7153 (0.7790) acc1: 54.9479 (52.3924) acc5: 100.0000 (100.0000) time: 0.1958 data: 0.1017 max mem: 3531 +Test: [100/333] eta: 0:00:56 acc1_pos: 54.427 acc5_pos: 100.000 loss: 0.6895 (0.7694) acc1: 60.1562 (53.2926) acc5: 100.0000 (100.0000) time: 0.1790 data: 0.0844 max mem: 3531 +Test: [110/333] eta: 0:00:53 acc1_pos: 73.698 acc5_pos: 100.000 loss: 0.6895 (0.7608) acc1: 60.9375 (54.1526) acc5: 100.0000 (100.0000) time: 0.1883 data: 0.0929 max mem: 3531 +Test: [120/333] eta: 0:00:50 acc1_pos: 58.594 acc5_pos: 100.000 loss: 0.6915 (0.7560) acc1: 60.9375 (54.6466) acc5: 100.0000 (100.0000) time: 0.1900 data: 0.0941 max mem: 3531 +Test: [130/333] eta: 0:00:46 acc1_pos: 61.719 acc5_pos: 100.000 loss: 0.6874 (0.7497) acc1: 59.1146 (55.2700) acc5: 100.0000 (100.0000) time: 0.1794 data: 0.0837 max mem: 3531 +Test: [140/333] eta: 0:00:43 acc1_pos: 71.875 acc5_pos: 100.000 loss: 0.6309 (0.7434) acc1: 61.7188 (55.7753) acc5: 100.0000 (100.0000) time: 0.1806 data: 0.0830 max mem: 3531 +Test: [150/333] eta: 0:00:41 acc1_pos: 60.156 acc5_pos: 100.000 loss: 0.6486 (0.7397) acc1: 60.4167 (56.0534) acc5: 100.0000 (100.0000) time: 0.1911 data: 0.0960 max mem: 3531 +Test: [160/333] eta: 0:00:38 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.6672 (0.7398) acc1: 58.8542 (56.0705) acc5: 100.0000 (100.0000) acc1_neg: 47.917 acc5_neg: 100.000 time: 0.1751 data: 0.0794 max mem: 3531 +Test: [170/333] eta: 0:00:36 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.6949 (0.7386) acc1: 58.3333 (56.1282) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.1882 data: 0.0905 max mem: 3531 +Test: [180/333] eta: 0:00:33 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.6592 (0.7326) acc1: 60.9375 (56.5507) acc5: 100.0000 (100.0000) acc1_neg: 59.375 acc5_neg: 100.000 time: 0.2239 data: 0.1261 max mem: 3531 +Test: [190/333] eta: 0:00:31 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.6312 (0.7256) acc1: 66.1458 (57.1840) acc5: 100.0000 (100.0000) acc1_neg: 59.375 acc5_neg: 100.000 time: 0.1999 data: 0.1026 max mem: 3531 +Test: [200/333] eta: 0:00:28 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.6309 (0.7201) acc1: 67.4479 (57.6687) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.1606 data: 0.0651 max mem: 3531 +Test: [210/333] eta: 0:00:26 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.6346 (0.7170) acc1: 66.1458 (57.8656) acc5: 100.0000 (100.0000) acc1_neg: 43.229 acc5_neg: 100.000 time: 0.1553 data: 0.0612 max mem: 3532 +Test: [220/333] eta: 0:00:24 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.6346 (0.7157) acc1: 63.8021 (57.9963) acc5: 100.0000 (100.0000) acc1_neg: 68.750 acc5_neg: 100.000 time: 0.1908 data: 0.0956 max mem: 3532 +Test: [230/333] eta: 0:00:21 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.5309 (0.7068) acc1: 69.2708 (58.8643) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1849 data: 0.0901 max mem: 3532 +Test: [240/333] eta: 0:00:19 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.5333 (0.7046) acc1: 73.4375 (58.9222) acc5: 100.0000 (100.0000) acc1_neg: 59.375 acc5_neg: 100.000 time: 0.1703 data: 0.0769 max mem: 3532 +Test: [250/333] eta: 0:00:17 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.6554 (0.7014) acc1: 58.0729 (59.1384) acc5: 100.0000 (100.0000) acc1_neg: 71.615 acc5_neg: 100.000 time: 0.1844 data: 0.0914 max mem: 3532 +Test: [260/333] eta: 0:00:15 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.6160 (0.6980) acc1: 65.1042 (59.4458) acc5: 100.0000 (100.0000) acc1_neg: 71.615 acc5_neg: 100.000 time: 0.1734 data: 0.0802 max mem: 3532 +Test: [270/333] eta: 0:00:12 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.5686 (0.6937) acc1: 70.0521 (59.8516) acc5: 100.0000 (100.0000) acc1_neg: 71.094 acc5_neg: 100.000 time: 0.1699 data: 0.0753 max mem: 3532 +Test: [280/333] eta: 0:00:10 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.6050 (0.6949) acc1: 65.1042 (59.6873) acc5: 100.0000 (100.0000) acc1_neg: 56.510 acc5_neg: 100.000 time: 0.1650 data: 0.0704 max mem: 3532 +Test: [290/333] eta: 0:00:08 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.8242 (0.6989) acc1: 46.8750 (59.3016) acc5: 100.0000 (100.0000) acc1_neg: 54.948 acc5_neg: 100.000 time: 0.1699 data: 0.0751 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.7192 (0.6991) acc1: 52.3438 (59.2219) acc5: 100.0000 (100.0000) acc1_neg: 51.823 acc5_neg: 100.000 time: 0.1893 data: 0.0928 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.6893 (0.6956) acc1: 57.8125 (59.5299) acc5: 100.0000 (100.0000) acc1_neg: 55.990 acc5_neg: 100.000 time: 0.1849 data: 0.0900 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.6686 (0.6948) acc1: 60.9375 (59.5648) acc5: 100.0000 (100.0000) acc1_neg: 57.031 acc5_neg: 100.000 time: 0.1768 data: 0.0859 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.7317 (0.6976) acc1: 56.5104 (59.3742) acc5: 100.0000 (100.0000) acc1_neg: 55.729 acc5_neg: 100.000 time: 0.1443 data: 0.0546 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.7582 (0.6980) acc1: 55.7292 (59.3462) acc5: 100.0000 (100.0000) acc1_neg: 53.803 acc5_neg: 100.000 time: 0.1343 data: 0.0414 max mem: 3532 +Test: Total time: 0:01:06 (0.1992 s / it) +* Overall Acc@1 59.346 Acc@5 100.000 + target == real Acc@1 56.287 Acc@5 100.000 + target == fake Acc@1 62.070 Acc@5 100.000 + loss 0.698 + PR-AUC 0.6263 ROC-AUC 0.6261 EER 40.86% (thr=0.5190) +Accuracy of the network on the 127843 test images: 59.3% +[rank0]:[W725 18:12:24.527692924 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:12:43.937000 2573129 torch/distributed/run.py:766] +W0725 18:12:43.937000 2573129 torch/distributed/run.py:766] ***************************************** +W0725 18:12:43.937000 2573129 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:12:43.937000 2573129 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W725 18:13:12.226779054 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:13:13.570881666 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:13:13.589965231 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:13:13.893985841 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:13:13.966805626 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:13:13.003880110 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:13:13.049657346 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:13:13.060974070 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed2_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed2_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:22:04 acc1_pos: 93.750 acc5_pos: 100.000 loss: 0.2247 (0.2247) acc1: 93.7500 (93.7500) acc5: 100.0000 (100.0000) time: 5.6347 data: 4.2687 max mem: 3531 +Test: [ 10/235] eta: 0:02:13 acc1_pos: 94.010 acc5_pos: 100.000 loss: 0.2319 (0.2339) acc1: 94.2708 (93.9157) acc5: 100.0000 (100.0000) time: 0.5941 data: 0.3885 max mem: 3531 +Test: [ 20/235] eta: 0:01:17 acc1_pos: 94.010 acc5_pos: 100.000 loss: 0.2319 (0.2311) acc1: 94.2708 (94.4072) acc5: 100.0000 (100.0000) time: 0.0950 data: 0.0004 max mem: 3531 +Test: [ 30/235] eta: 0:00:58 acc1_pos: 97.396 acc5_pos: 100.000 loss: 0.2286 (0.2283) acc1: 94.7917 (94.6321) acc5: 100.0000 (100.0000) time: 0.1122 data: 0.0147 max mem: 3531 +Test: [ 40/235] eta: 0:00:51 acc1_pos: 92.188 acc5_pos: 100.000 loss: 0.2275 (0.2284) acc1: 94.7917 (94.5884) acc5: 100.0000 (100.0000) time: 0.1649 data: 0.0680 max mem: 3531 +Test: [ 50/235] eta: 0:00:46 acc1_pos: 94.010 acc5_pos: 100.000 loss: 0.2302 (0.2289) acc1: 94.2708 (94.5466) acc5: 100.0000 (100.0000) time: 0.2076 data: 0.1126 max mem: 3531 +Test: [ 60/235] eta: 0:00:42 acc1_pos: 95.573 acc5_pos: 100.000 loss: 0.2314 (0.2294) acc1: 94.2708 (94.4758) acc5: 100.0000 (100.0000) time: 0.1949 data: 0.1014 max mem: 3531 +Test: [ 70/235] eta: 0:00:39 acc1_pos: 95.385 acc5_pos: 100.000 loss: 0.2482 (0.3894) acc1: 91.4062 (85.1563) acc5: 100.0000 (100.0000) acc1_neg: 19.010 acc5_neg: 100.000 time: 0.1949 data: 0.0989 max mem: 3531 +Test: [ 80/235] eta: 0:00:35 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.5055 (0.5286) acc1: 19.0104 (76.9579) acc5: 100.0000 (100.0000) acc1_neg: 17.969 acc5_neg: 100.000 time: 0.1799 data: 0.0853 max mem: 3531 +Test: [ 90/235] eta: 0:00:31 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.5062 (0.6376) acc1: 18.2292 (70.5500) acc5: 100.0000 (100.0000) acc1_neg: 18.490 acc5_neg: 100.000 time: 0.1599 data: 0.0655 max mem: 3531 +Test: [100/235] eta: 0:00:29 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.5198 (0.7241) acc1: 19.0104 (65.5090) acc5: 100.0000 (100.0000) acc1_neg: 16.146 acc5_neg: 100.000 time: 0.1849 data: 0.0870 max mem: 3531 +Test: [110/235] eta: 0:00:27 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.5118 (0.7948) acc1: 19.0104 (61.3199) acc5: 100.0000 (100.0000) acc1_neg: 19.271 acc5_neg: 100.000 time: 0.2199 data: 0.1193 max mem: 3531 +Test: [120/235] eta: 0:00:25 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.5574 (0.8655) acc1: 16.1458 (57.2293) acc5: 100.0000 (100.0000) acc1_neg: 11.458 acc5_neg: 100.000 time: 0.2249 data: 0.1253 max mem: 3531 +Test: [130/235] eta: 0:00:22 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.6572 (0.9247) acc1: 11.9792 (53.8108) acc5: 100.0000 (100.0000) acc1_neg: 13.021 acc5_neg: 100.000 time: 0.1949 data: 0.0950 max mem: 3531 +Test: [140/235] eta: 0:00:20 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.6573 (0.9786) acc1: 11.9792 (50.8071) acc5: 100.0000 (100.0000) acc1_neg: 10.417 acc5_neg: 100.000 time: 0.1768 data: 0.0784 max mem: 3531 +Test: [150/235] eta: 0:00:18 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.6634 (1.0235) acc1: 11.1979 (48.2168) acc5: 100.0000 (100.0000) acc1_neg: 13.021 acc5_neg: 100.000 time: 0.1905 data: 0.0973 max mem: 3531 +Test: [160/235] eta: 0:00:15 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.6575 (1.0627) acc1: 11.4583 (45.9724) acc5: 100.0000 (100.0000) acc1_neg: 11.458 acc5_neg: 100.000 time: 0.2039 data: 0.1123 max mem: 3531 +Test: [170/235] eta: 0:00:13 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.5977 (1.0789) acc1: 12.7604 (44.7734) acc5: 100.0000 (100.0000) acc1_neg: 21.875 acc5_neg: 100.000 time: 0.2243 data: 0.1300 max mem: 3531 +Test: [180/235] eta: 0:00:11 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.2844 (1.0902) acc1: 27.0833 (43.8680) acc5: 100.0000 (100.0000) acc1_neg: 26.302 acc5_neg: 100.000 time: 0.2191 data: 0.1222 max mem: 3531 +Test: [190/235] eta: 0:00:09 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.2783 (1.1011) acc1: 27.8646 (43.0342) acc5: 100.0000 (100.0000) acc1_neg: 25.781 acc5_neg: 100.000 time: 0.1821 data: 0.0848 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.2906 (1.1119) acc1: 26.8229 (42.1953) acc5: 100.0000 (100.0000) acc1_neg: 24.479 acc5_neg: 100.000 time: 0.1849 data: 0.0909 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.2947 (1.1205) acc1: 26.8229 (41.4877) acc5: 100.0000 (100.0000) acc1_neg: 23.177 acc5_neg: 100.000 time: 0.1994 data: 0.1094 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.2965 (1.1284) acc1: 27.6042 (40.8560) acc5: 100.0000 (100.0000) acc1_neg: 27.865 acc5_neg: 100.000 time: 0.1873 data: 0.0972 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.2956 (1.1352) acc1: 28.3854 (40.3195) acc5: 100.0000 (100.0000) acc1_neg: 29.167 acc5_neg: 100.000 time: 0.1563 data: 0.0658 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.2956 (1.1391) acc1: 28.6458 (40.1495) acc5: 100.0000 (100.0000) acc1_neg: 30.435 acc5_neg: 100.000 time: 0.1401 data: 0.0530 max mem: 3532 +Test: Total time: 0:00:47 (0.2039 s / it) +* Overall Acc@1 40.150 Acc@5 100.000 + target == real Acc@1 94.475 Acc@5 100.000 + target == fake Acc@1 20.501 Acc@5 100.000 + loss 1.139 + PR-AUC 0.8507 ROC-AUC 0.6780 EER 37.26% (thr=0.1639) +Accuracy of the network on the 89879 test images: 40.1% +[rank0]:[W725 18:14:11.200200756 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:14:31.216000 2573699 torch/distributed/run.py:766] +W0725 18:14:31.216000 2573699 torch/distributed/run.py:766] ***************************************** +W0725 18:14:31.216000 2573699 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:14:31.216000 2573699 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W725 18:14:59.656121137 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:14:59.142770684 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:14:59.367170318 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:14:59.509275663 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:15:00.642918618 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:15:00.786439103 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:15:00.818933228 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:15:00.833925524 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed2_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed2_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:23:09 acc1_pos: 57.812 acc5_pos: 100.000 loss: 0.6716 (0.6716) acc1: 57.8125 (57.8125) acc5: 100.0000 (100.0000) time: 5.1662 data: 3.5771 max mem: 3531 +Test: [ 10/269] eta: 0:02:24 acc1_pos: 58.594 acc5_pos: 100.000 loss: 0.7489 (0.7785) acc1: 53.3854 (47.5616) acc5: 100.0000 (100.0000) time: 0.5589 data: 0.3331 max mem: 3531 +Test: [ 20/269] eta: 0:01:27 acc1_pos: 54.948 acc5_pos: 100.000 loss: 0.7898 (0.7692) acc1: 43.7500 (48.4995) acc5: 100.0000 (100.0000) time: 0.1125 data: 0.0216 max mem: 3531 +Test: [ 30/269] eta: 0:01:09 acc1_pos: 53.646 acc5_pos: 100.000 loss: 0.7340 (0.7480) acc1: 52.6042 (50.8485) acc5: 100.0000 (100.0000) time: 0.1454 data: 0.0513 max mem: 3531 +Test: [ 40/269] eta: 0:01:03 acc1_pos: 30.172 acc5_pos: 100.000 loss: 0.6733 (0.7115) acc1: 58.3333 (55.0877) acc5: 100.0000 (100.0000) acc1_neg: 74.219 acc5_neg: 100.000 time: 0.1974 data: 0.1014 max mem: 3531 +Test: [ 50/269] eta: 0:00:58 acc1_pos: 30.172 acc5_pos: 100.000 loss: 0.5209 (0.6769) acc1: 74.2188 (58.3589) acc5: 100.0000 (100.0000) acc1_neg: 51.042 acc5_neg: 100.000 time: 0.2238 data: 0.1299 max mem: 3531 +Test: [ 60/269] eta: 0:00:53 acc1_pos: 30.172 acc5_pos: 100.000 loss: 0.5410 (0.6666) acc1: 71.0938 (59.4262) acc5: 100.0000 (100.0000) acc1_neg: 71.094 acc5_neg: 100.000 time: 0.2125 data: 0.1194 max mem: 3531 +Test: [ 70/269] eta: 0:00:51 acc1_pos: 30.172 acc5_pos: 100.000 loss: 0.6253 (0.6658) acc1: 62.2396 (59.1183) acc5: 100.0000 (100.0000) acc1_neg: 53.646 acc5_neg: 100.000 time: 0.2377 data: 0.1421 max mem: 3531 +Test: [ 80/269] eta: 0:00:47 acc1_pos: 30.172 acc5_pos: 100.000 loss: 0.6345 (0.6546) acc1: 57.8125 (60.4360) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2362 data: 0.1420 max mem: 3531 +Test: [ 90/269] eta: 0:00:42 acc1_pos: 30.172 acc5_pos: 100.000 loss: 0.5406 (0.6400) acc1: 76.3021 (62.6774) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.1640 data: 0.0714 max mem: 3531 +Test: [100/269] eta: 0:00:40 acc1_pos: 30.172 acc5_pos: 100.000 loss: 0.5442 (0.6359) acc1: 73.9583 (62.8352) acc5: 100.0000 (100.0000) acc1_neg: 55.729 acc5_neg: 100.000 time: 0.2049 data: 0.1120 max mem: 3531 +Test: [110/269] eta: 0:00:37 acc1_pos: 30.172 acc5_pos: 100.000 loss: 0.6256 (0.6383) acc1: 57.2917 (62.4859) acc5: 100.0000 (100.0000) acc1_neg: 49.479 acc5_neg: 100.000 time: 0.2480 data: 0.1568 max mem: 3531 +Test: [120/269] eta: 0:00:34 acc1_pos: 30.172 acc5_pos: 100.000 loss: 0.6417 (0.6378) acc1: 60.9375 (62.5280) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.1958 data: 0.1044 max mem: 3531 +Test: [130/269] eta: 0:00:32 acc1_pos: 30.172 acc5_pos: 100.000 loss: 0.6468 (0.6408) acc1: 57.8125 (62.0945) acc5: 100.0000 (100.0000) acc1_neg: 52.083 acc5_neg: 100.000 time: 0.1955 data: 0.1033 max mem: 3531 +Test: [140/269] eta: 0:00:29 acc1_pos: 30.172 acc5_pos: 100.000 loss: 0.6882 (0.6395) acc1: 54.4271 (62.2488) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.1843 data: 0.0931 max mem: 3531 +Test: [150/269] eta: 0:00:26 acc1_pos: 30.172 acc5_pos: 100.000 loss: 0.5153 (0.6312) acc1: 72.6562 (63.0070) acc5: 100.0000 (100.0000) acc1_neg: 72.656 acc5_neg: 100.000 time: 0.1698 data: 0.0789 max mem: 3531 +Test: [160/269] eta: 0:00:24 acc1_pos: 30.172 acc5_pos: 100.000 loss: 0.5201 (0.6247) acc1: 72.9167 (63.7115) acc5: 100.0000 (100.0000) acc1_neg: 72.917 acc5_neg: 100.000 time: 0.1832 data: 0.0880 max mem: 3531 +Test: [170/269] eta: 0:00:21 acc1_pos: 30.172 acc5_pos: 100.000 loss: 0.5517 (0.6229) acc1: 71.8750 (64.0975) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1945 data: 0.0993 max mem: 3531 +Test: [180/269] eta: 0:00:19 acc1_pos: 30.172 acc5_pos: 100.000 loss: 0.4868 (0.6130) acc1: 76.3021 (65.3675) acc5: 100.0000 (100.0000) acc1_neg: 74.219 acc5_neg: 100.000 time: 0.2061 data: 0.1154 max mem: 3531 +Test: [190/269] eta: 0:00:17 acc1_pos: 30.172 acc5_pos: 100.000 loss: 0.4873 (0.6120) acc1: 81.2500 (65.5159) acc5: 100.0000 (100.0000) acc1_neg: 70.833 acc5_neg: 100.000 time: 0.2096 data: 0.1193 max mem: 3531 +Test: [200/269] eta: 0:00:15 acc1_pos: 30.172 acc5_pos: 100.000 loss: 0.6161 (0.6147) acc1: 60.6771 (65.1897) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.2139 data: 0.1224 max mem: 3531 +Test: [210/269] eta: 0:00:12 acc1_pos: 30.172 acc5_pos: 100.000 loss: 0.6240 (0.6153) acc1: 60.6771 (65.1819) acc5: 100.0000 (100.0000) acc1_neg: 70.573 acc5_neg: 100.000 time: 0.2041 data: 0.1099 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 30.172 acc5_pos: 100.000 loss: 0.5763 (0.6135) acc1: 69.7917 (65.4259) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.1821 data: 0.0892 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 30.172 acc5_pos: 100.000 loss: 0.4859 (0.6090) acc1: 79.4271 (66.0049) acc5: 100.0000 (100.0000) acc1_neg: 61.198 acc5_neg: 100.000 time: 0.1705 data: 0.0803 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 30.172 acc5_pos: 100.000 loss: 0.5506 (0.6099) acc1: 72.6562 (65.8476) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.1584 data: 0.0675 max mem: 3532 +Test: [250/269] eta: 0:00:03 acc1_pos: 30.172 acc5_pos: 100.000 loss: 0.6536 (0.6110) acc1: 59.6354 (65.6935) acc5: 100.0000 (100.0000) acc1_neg: 48.177 acc5_neg: 100.000 time: 0.1386 data: 0.0474 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 30.172 acc5_pos: 100.000 loss: 0.6728 (0.6153) acc1: 54.1667 (65.1630) acc5: 100.0000 (100.0000) acc1_neg: 71.094 acc5_neg: 100.000 time: 0.1283 data: 0.0361 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 30.172 acc5_pos: 100.000 loss: 0.6520 (0.6142) acc1: 57.0312 (65.2977) acc5: 100.0000 (100.0000) acc1_neg: 82.677 acc5_neg: 100.000 time: 0.1292 data: 0.0321 max mem: 3532 +Test: Total time: 0:00:54 (0.2030 s / it) +* Overall Acc@1 65.298 Acc@5 100.000 + target == real Acc@1 50.177 Acc@5 100.000 + target == fake Acc@1 67.361 Acc@5 100.000 + loss 0.614 + PR-AUC 0.9212 ROC-AUC 0.6328 EER 40.95% (thr=0.5449) +Accuracy of the network on the 103293 test images: 65.3% +[rank0]:[W725 18:16:04.300464497 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-25T18:16:11Z] eval_done tag=fixed500k_real_n16_seed2_resampled +[2026-07-25T18:16:11Z] eval_skip tag=fixed500k_fake_n32_seed2_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed2_resampled +[2026-07-25T18:16:11Z] eval_start tag=fixed500k_real_n16_seed3_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed3_resampled_n111-046-081_20260725_112559/best_checkpoint.pth +started_utc=2026-07-25T18:16:11Z +hostname=n111-046-081 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed3_resampled_n111-046-081_20260725_112559/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed3_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0725 18:16:23.992000 2574280 torch/distributed/run.py:766] +W0725 18:16:23.992000 2574280 torch/distributed/run.py:766] ***************************************** +W0725 18:16:23.992000 2574280 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:16:23.992000 2574280 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:16:50.476333173 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +[rank6]:[W725 18:16:52.586797028 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:16:52.595034072 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:16:52.629694806 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:16:52.666892288 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:16:52.670047723 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:16:52.768504784 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:16:52.778503785 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed3_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed3_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:18:04 acc1_pos: 81.250 acc5_pos: 100.000 loss: 0.4233 (0.4233) acc1: 81.2500 (81.2500) acc5: 100.0000 (100.0000) time: 4.6344 data: 2.6740 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/234] eta: 0:01:58 acc1_pos: 45.052 acc5_pos: 100.000 loss: 0.7731 (0.7585) acc1: 50.5208 (49.9527) acc5: 100.0000 (100.0000) time: 0.5284 data: 0.2689 max mem: 3531 +Test: [ 20/234] eta: 0:01:10 acc1_pos: 94.271 acc5_pos: 100.000 loss: 0.7731 (0.7064) acc1: 47.6562 (55.6052) acc5: 100.0000 (100.0000) time: 0.1157 data: 0.0245 max mem: 3531 +Test: [ 30/234] eta: 0:00:56 acc1_pos: 47.135 acc5_pos: 100.000 loss: 0.2426 (0.5818) acc1: 93.4896 (65.7930) acc5: 100.0000 (100.0000) time: 0.1381 data: 0.0452 max mem: 3531 +Test: [ 40/234] eta: 0:00:51 acc1_pos: 97.656 acc5_pos: 100.000 loss: 0.2292 (0.5828) acc1: 93.7500 (65.8219) acc5: 100.0000 (100.0000) time: 0.1999 data: 0.1060 max mem: 3531 +Test: [ 50/234] eta: 0:00:47 acc1_pos: 97.396 acc5_pos: 100.000 loss: 0.1847 (0.5029) acc1: 95.8333 (71.8903) acc5: 100.0000 (100.0000) time: 0.2275 data: 0.1330 max mem: 3531 +Test: [ 60/234] eta: 0:00:42 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.1375 (0.4346) acc1: 97.6562 (76.4600) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.1016 max mem: 3531 +Test: [ 70/234] eta: 0:00:37 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.0805 (0.4378) acc1: 100.0000 (76.4525) acc5: 100.0000 (100.0000) acc1_neg: 63.021 acc5_neg: 100.000 time: 0.1569 data: 0.0625 max mem: 3531 +Test: [ 80/234] eta: 0:00:34 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.5779 (0.4389) acc1: 67.4479 (76.7425) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0789 max mem: 3531 +Test: [ 90/234] eta: 0:00:32 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.1546 (0.4073) acc1: 95.5729 (78.9320) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.2240 data: 0.1242 max mem: 3531 +Test: [100/234] eta: 0:00:29 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.2251 (0.4037) acc1: 94.2708 (79.3085) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1999 data: 0.0999 max mem: 3531 +Test: [110/234] eta: 0:00:26 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.3408 (0.3954) acc1: 82.8125 (80.0206) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.1500 data: 0.0545 max mem: 3531 +Test: [120/234] eta: 0:00:24 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.2854 (0.3912) acc1: 87.7604 (80.6388) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1799 data: 0.0846 max mem: 3531 +Test: [130/234] eta: 0:00:22 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.3827 (0.3962) acc1: 84.1146 (80.6000) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.2248 data: 0.1299 max mem: 3531 +Test: [140/234] eta: 0:00:19 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.4498 (0.4184) acc1: 77.6042 (79.5896) acc5: 100.0000 (100.0000) acc1_neg: 41.667 acc5_neg: 100.000 time: 0.1670 data: 0.0707 max mem: 3531 +Test: [150/234] eta: 0:00:17 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.3670 (0.4205) acc1: 82.2917 (79.5720) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.1471 data: 0.0530 max mem: 3531 +Test: [160/234] eta: 0:00:15 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.3614 (0.4188) acc1: 83.0729 (79.8654) acc5: 100.0000 (100.0000) acc1_neg: 78.646 acc5_neg: 100.000 time: 0.1879 data: 0.0979 max mem: 3531 +Test: [170/234] eta: 0:00:13 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.4579 (0.4206) acc1: 79.9479 (79.9068) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.2027 data: 0.1113 max mem: 3531 +Test: [180/234] eta: 0:00:10 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.4620 (0.4281) acc1: 78.9062 (79.4084) acc5: 100.0000 (100.0000) acc1_neg: 45.312 acc5_neg: 100.000 time: 0.1899 data: 0.0968 max mem: 3531 +Test: [190/234] eta: 0:00:08 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.6600 (0.4422) acc1: 63.0208 (78.4550) acc5: 100.0000 (100.0000) acc1_neg: 63.542 acc5_neg: 100.000 time: 0.1549 data: 0.0594 max mem: 3531 +Test: [200/234] eta: 0:00:06 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.6600 (0.4361) acc1: 63.0208 (78.8324) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1652 data: 0.0688 max mem: 3531 +Test: [210/234] eta: 0:00:04 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.1107 (0.4208) acc1: 99.7396 (79.8208) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2199 data: 0.1243 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.1186 (0.4085) acc1: 99.7396 (80.6149) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.2076 data: 0.1121 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.3788 (0.4273) acc1: 51.3021 (79.0866) acc5: 100.0000 (100.0000) acc1_neg: 45.312 acc5_neg: 100.000 time: 0.1227 data: 0.0303 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.8331 (0.4325) acc1: 46.0938 (78.7103) acc5: 100.0000 (100.0000) acc1_neg: 45.525 acc5_neg: 100.000 time: 0.1221 data: 0.0303 max mem: 3532 +Test: Total time: 0:00:45 (0.1949 s / it) +* Overall Acc@1 78.710 Acc@5 100.000 + target == real Acc@1 72.056 Acc@5 100.000 + target == fake Acc@1 80.586 Acc@5 100.000 + loss 0.433 + PR-AUC 0.9600 ROC-AUC 0.8700 EER 22.92% (thr=0.5601) +Accuracy of the network on the 89729 test images: 78.7% +[rank0]:[W725 18:17:47.739313640 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:18:05.646000 2574872 torch/distributed/run.py:766] +W0725 18:18:05.646000 2574872 torch/distributed/run.py:766] ***************************************** +W0725 18:18:05.646000 2574872 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:18:05.646000 2574872 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:18:33.012017339 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:18:34.594074866 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:18:34.694552207 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:18:34.785316609 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:18:34.812818791 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:18:34.895879536 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:18:34.925744963 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W725 18:18:34.925799560 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed3_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed3_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 3:34:27 acc1_pos: 59.115 acc5_pos: 100.000 loss: 0.6576 (0.6576) acc1: 59.1146 (59.1146) acc5: 100.0000 (100.0000) time: 5.3907 data: 3.2529 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/2387] eta: 0:23:39 acc1_pos: 23.698 acc5_pos: 100.000 loss: 0.6405 (0.6988) acc1: 61.4583 (61.1269) acc5: 100.0000 (100.0000) time: 0.5971 data: 0.3209 max mem: 3531 +Test: [ 20/2387] eta: 0:14:18 acc1_pos: 48.438 acc5_pos: 100.000 loss: 0.7726 (0.7671) acc1: 53.3854 (53.6334) acc5: 100.0000 (100.0000) time: 0.1113 data: 0.0212 max mem: 3531 +Test: [ 30/2387] eta: 0:11:12 acc1_pos: 52.865 acc5_pos: 100.000 loss: 0.7497 (0.7440) acc1: 48.4375 (54.7379) acc5: 100.0000 (100.0000) time: 0.1138 data: 0.0213 max mem: 3531 +Test: [ 40/2387] eta: 0:10:49 acc1_pos: 72.135 acc5_pos: 100.000 loss: 0.6774 (0.7227) acc1: 59.1146 (57.3171) acc5: 100.0000 (100.0000) time: 0.1864 data: 0.0936 max mem: 3531 +Test: [ 50/2387] eta: 0:10:25 acc1_pos: 36.198 acc5_pos: 100.000 loss: 0.6648 (0.7344) acc1: 60.4167 (56.4696) acc5: 100.0000 (100.0000) time: 0.2400 data: 0.1485 max mem: 3531 +Test: [ 60/2387] eta: 0:10:19 acc1_pos: 52.604 acc5_pos: 100.000 loss: 0.7914 (0.7493) acc1: 50.2604 (55.1614) acc5: 100.0000 (100.0000) time: 0.2442 data: 0.1501 max mem: 3531 +Test: [ 70/2387] eta: 0:10:10 acc1_pos: 39.583 acc5_pos: 100.000 loss: 0.8485 (0.7695) acc1: 46.3542 (53.4038) acc5: 100.0000 (100.0000) time: 0.2534 data: 0.1576 max mem: 3531 +Test: [ 80/2387] eta: 0:10:04 acc1_pos: 96.094 acc5_pos: 100.000 loss: 0.8040 (0.7332) acc1: 50.5208 (56.8769) acc5: 100.0000 (100.0000) time: 0.2489 data: 0.1536 max mem: 3531 +Test: [ 90/2387] eta: 0:09:53 acc1_pos: 92.708 acc5_pos: 100.000 loss: 0.3889 (0.6951) acc1: 87.7604 (60.3537) acc5: 100.0000 (100.0000) time: 0.2401 data: 0.1472 max mem: 3531 +Test: [ 100/2387] eta: 0:09:46 acc1_pos: 98.177 acc5_pos: 100.000 loss: 0.3888 (0.6701) acc1: 85.4167 (62.6702) acc5: 100.0000 (100.0000) time: 0.2350 data: 0.1445 max mem: 3531 +Test: [ 110/2387] eta: 0:09:48 acc1_pos: 71.094 acc5_pos: 100.000 loss: 0.4658 (0.6508) acc1: 79.9479 (64.3558) acc5: 100.0000 (100.0000) time: 0.2595 data: 0.1686 max mem: 3531 +Test: [ 120/2387] eta: 0:09:42 acc1_pos: 90.625 acc5_pos: 100.000 loss: 0.4624 (0.6351) acc1: 79.9479 (65.8467) acc5: 100.0000 (100.0000) time: 0.2598 data: 0.1671 max mem: 3531 +Test: [ 130/2387] eta: 0:09:38 acc1_pos: 93.750 acc5_pos: 100.000 loss: 0.4092 (0.6154) acc1: 87.5000 (67.8077) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1478 max mem: 3531 +Test: [ 140/2387] eta: 0:09:30 acc1_pos: 90.365 acc5_pos: 100.000 loss: 0.4061 (0.6039) acc1: 89.0625 (68.8239) acc5: 100.0000 (100.0000) time: 0.2356 data: 0.1386 max mem: 3531 +Test: [ 150/2387] eta: 0:09:24 acc1_pos: 94.010 acc5_pos: 100.000 loss: 0.3986 (0.5892) acc1: 87.2396 (70.1297) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1316 max mem: 3531 +Test: [ 160/2387] eta: 0:09:19 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.3763 (0.5746) acc1: 89.3229 (71.2830) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2299 data: 0.1375 max mem: 3531 +Test: [ 170/2387] eta: 0:09:07 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.2396 (0.5542) acc1: 95.3125 (72.7172) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2049 data: 0.1112 max mem: 3531 +Test: [ 180/2387] eta: 0:09:00 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.2272 (0.5363) acc1: 95.8333 (73.9900) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1942 data: 0.0958 max mem: 3531 +Test: [ 190/2387] eta: 0:08:50 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.2274 (0.5200) acc1: 95.8333 (75.1309) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1949 data: 0.0967 max mem: 3531 +Test: [ 200/2387] eta: 0:08:43 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.2220 (0.5052) acc1: 95.8333 (76.1712) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1899 data: 0.0941 max mem: 3531 +Test: [ 210/2387] eta: 0:08:37 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.2280 (0.4924) acc1: 95.5729 (77.0821) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2049 data: 0.1082 max mem: 3532 +Test: [ 220/2387] eta: 0:08:29 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.2302 (0.4805) acc1: 95.5729 (77.9200) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1949 data: 0.0982 max mem: 3532 +Test: [ 230/2387] eta: 0:08:26 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.2289 (0.4696) acc1: 95.5729 (78.6932) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1999 data: 0.1025 max mem: 3532 +Test: [ 240/2387] eta: 0:08:17 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.2378 (0.4704) acc1: 94.0104 (78.6383) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.1899 data: 0.0894 max mem: 3532 +Test: [ 250/2387] eta: 0:08:07 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.2262 (0.4582) acc1: 80.2083 (79.3773) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1504 data: 0.0515 max mem: 3532 +Test: [ 260/2387] eta: 0:08:04 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1168 (0.4449) acc1: 100.0000 (80.1495) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1808 data: 0.0875 max mem: 3532 +Test: [ 270/2387] eta: 0:07:59 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1017 (0.4320) acc1: 100.0000 (80.8743) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2061 data: 0.1137 max mem: 3532 +Test: [ 280/2387] eta: 0:07:52 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.0978 (0.4204) acc1: 100.0000 (81.5475) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1791 data: 0.0865 max mem: 3532 +Test: [ 290/2387] eta: 0:07:46 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1093 (0.4102) acc1: 100.0000 (82.1440) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1732 data: 0.0770 max mem: 3532 +Test: [ 300/2387] eta: 0:07:41 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1333 (0.4020) acc1: 99.2188 (82.6827) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1770 data: 0.0775 max mem: 3532 +Test: [ 310/2387] eta: 0:07:36 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1759 (0.3952) acc1: 98.9583 (83.1625) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1767 data: 0.0809 max mem: 3532 +Test: [ 320/2387] eta: 0:07:33 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1771 (0.3884) acc1: 99.4792 (83.6749) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1936 data: 0.1019 max mem: 3532 +Test: [ 330/2387] eta: 0:07:29 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1767 (0.3819) acc1: 99.7396 (84.1571) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2001 data: 0.1088 max mem: 3532 +Test: [ 340/2387] eta: 0:07:25 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1739 (0.3758) acc1: 99.7396 (84.6133) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1946 data: 0.1032 max mem: 3532 +Test: [ 350/2387] eta: 0:07:20 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1674 (0.3688) acc1: 100.0000 (85.0509) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1793 data: 0.0869 max mem: 3532 +Test: [ 360/2387] eta: 0:07:15 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1360 (0.3639) acc1: 99.7396 (85.3936) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1649 data: 0.0702 max mem: 3532 +Test: [ 370/2387] eta: 0:07:12 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.2024 (0.3596) acc1: 97.3958 (85.7108) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1785 data: 0.0807 max mem: 3532 +Test: [ 380/2387] eta: 0:07:06 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.2025 (0.3555) acc1: 96.8750 (86.0031) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1698 data: 0.0705 max mem: 3532 +Test: [ 390/2387] eta: 0:07:01 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.2077 (0.3518) acc1: 96.8750 (86.2772) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1511 data: 0.0557 max mem: 3532 +Test: [ 400/2387] eta: 0:06:59 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.2085 (0.3481) acc1: 96.6146 (86.5356) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1845 data: 0.0884 max mem: 3532 +Test: [ 410/2387] eta: 0:06:55 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.2015 (0.3445) acc1: 96.8750 (86.7973) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1952 data: 0.0988 max mem: 3532 +Test: [ 420/2387] eta: 0:06:49 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.2015 (0.3412) acc1: 97.1354 (87.0367) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1450 data: 0.0524 max mem: 3532 +Test: [ 430/2387] eta: 0:06:44 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.2133 (0.3416) acc1: 96.6146 (87.0396) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.1388 data: 0.0468 max mem: 3532 +Test: [ 440/2387] eta: 0:06:40 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.2213 (0.3393) acc1: 86.1979 (87.1675) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1573 data: 0.0655 max mem: 3533 +Test: [ 450/2387] eta: 0:06:36 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1149 (0.3344) acc1: 100.0000 (87.4515) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1606 data: 0.0703 max mem: 3533 +Test: [ 460/2387] eta: 0:06:33 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1140 (0.3296) acc1: 100.0000 (87.7214) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1725 data: 0.0814 max mem: 3533 +Test: [ 470/2387] eta: 0:06:29 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1144 (0.3250) acc1: 100.0000 (87.9799) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1683 data: 0.0762 max mem: 3533 +Test: [ 480/2387] eta: 0:06:26 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1149 (0.3207) acc1: 100.0000 (88.2277) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1601 data: 0.0660 max mem: 3533 +Test: [ 490/2387] eta: 0:06:23 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1149 (0.3165) acc1: 100.0000 (88.4664) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1786 data: 0.0827 max mem: 3533 +Test: [ 500/2387] eta: 0:06:20 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1153 (0.3125) acc1: 100.0000 (88.6955) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1822 data: 0.0876 max mem: 3533 +Test: [ 510/2387] eta: 0:06:17 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1152 (0.3094) acc1: 100.0000 (88.9106) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1657 data: 0.0725 max mem: 3533 +Test: [ 520/2387] eta: 0:06:14 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1707 (0.3077) acc1: 99.2188 (89.1065) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1729 data: 0.0815 max mem: 3533 +Test: [ 530/2387] eta: 0:06:11 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1721 (0.3051) acc1: 99.2188 (89.3033) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1768 data: 0.0831 max mem: 3533 +Test: [ 540/2387] eta: 0:06:09 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1704 (0.3026) acc1: 99.4792 (89.4895) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1826 data: 0.0882 max mem: 3533 +Test: [ 550/2387] eta: 0:06:06 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1688 (0.3002) acc1: 99.4792 (89.6712) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1912 data: 0.0991 max mem: 3533 +Test: [ 560/2387] eta: 0:06:04 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1660 (0.2978) acc1: 99.4792 (89.8484) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1827 data: 0.0911 max mem: 3533 +Test: [ 570/2387] eta: 0:06:02 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1660 (0.2955) acc1: 99.7396 (90.0221) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1878 data: 0.0951 max mem: 3533 +Test: [ 580/2387] eta: 0:05:59 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1692 (0.2934) acc1: 99.7396 (90.1862) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1841 data: 0.0896 max mem: 3533 +Test: [ 590/2387] eta: 0:05:57 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1706 (0.2913) acc1: 99.4792 (90.3417) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1852 data: 0.0930 max mem: 3533 +Test: [ 600/2387] eta: 0:05:54 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1808 (0.2906) acc1: 99.2188 (90.4855) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1847 data: 0.0941 max mem: 3533 +Test: [ 610/2387] eta: 0:05:51 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.2072 (0.2890) acc1: 98.9583 (90.6263) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1747 data: 0.0838 max mem: 3533 +Test: [ 620/2387] eta: 0:05:49 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1960 (0.2876) acc1: 98.9583 (90.7621) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1814 data: 0.0885 max mem: 3533 +Test: [ 630/2387] eta: 0:05:47 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1987 (0.2861) acc1: 98.9583 (90.8904) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1883 data: 0.0953 max mem: 3533 +Test: [ 640/2387] eta: 0:05:45 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1965 (0.2847) acc1: 98.9583 (91.0105) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1912 data: 0.1007 max mem: 3533 +Test: [ 650/2387] eta: 0:05:42 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1977 (0.2834) acc1: 98.9583 (91.1314) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1845 data: 0.0927 max mem: 3533 +Test: [ 660/2387] eta: 0:05:40 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1982 (0.2821) acc1: 98.9583 (91.2491) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1844 data: 0.0909 max mem: 3533 +Test: [ 670/2387] eta: 0:05:38 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1974 (0.2809) acc1: 98.9583 (91.3616) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1934 data: 0.1013 max mem: 3534 +Test: [ 680/2387] eta: 0:05:36 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.2018 (0.2819) acc1: 98.4375 (91.3692) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1975 data: 0.1063 max mem: 3534 +Test: [ 690/2387] eta: 0:05:34 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.3475 (0.2830) acc1: 87.7604 (91.3026) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.1947 data: 0.1025 max mem: 3534 +Test: [ 700/2387] eta: 0:05:32 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.3244 (0.2835) acc1: 86.4583 (91.2287) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1926 data: 0.0996 max mem: 3534 +Test: [ 710/2387] eta: 0:05:30 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.3141 (0.2840) acc1: 86.7188 (91.1737) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1887 data: 0.0958 max mem: 3534 +Test: [ 720/2387] eta: 0:05:27 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.3141 (0.2844) acc1: 87.2396 (91.1104) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1823 data: 0.0912 max mem: 3534 +Test: [ 730/2387] eta: 0:05:25 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.3184 (0.2850) acc1: 86.1979 (91.0397) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.1812 data: 0.0909 max mem: 3534 +Test: [ 740/2387] eta: 0:05:23 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.3184 (0.2854) acc1: 86.1979 (90.9782) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.1834 data: 0.0924 max mem: 3534 +Test: [ 750/2387] eta: 0:05:20 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.3091 (0.2857) acc1: 86.4583 (90.9243) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1840 data: 0.0919 max mem: 3534 +Test: [ 760/2387] eta: 0:05:18 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.3083 (0.2862) acc1: 86.7188 (90.8652) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1863 data: 0.0944 max mem: 3534 +Test: [ 770/2387] eta: 0:05:16 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.3155 (0.2867) acc1: 87.5000 (90.8239) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1849 data: 0.0943 max mem: 3534 +Test: [ 780/2387] eta: 0:05:14 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.3045 (0.2864) acc1: 88.5417 (90.8507) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1850 data: 0.0939 max mem: 3534 +Test: [ 790/2387] eta: 0:05:12 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.2160 (0.2854) acc1: 96.6146 (90.9312) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1921 data: 0.1000 max mem: 3534 +Test: [ 800/2387] eta: 0:05:10 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.2036 (0.2847) acc1: 96.8750 (90.9904) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1854 data: 0.0940 max mem: 3534 +Test: [ 810/2387] eta: 0:05:07 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.2757 (0.2849) acc1: 92.7083 (90.9670) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1728 data: 0.0815 max mem: 3534 +Test: [ 820/2387] eta: 0:05:05 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.3287 (0.2859) acc1: 86.7188 (90.8838) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.1693 data: 0.0757 max mem: 3534 +Test: [ 830/2387] eta: 0:05:03 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.4973 (0.2898) acc1: 73.4375 (90.5680) acc5: 100.0000 (100.0000) acc1_neg: 63.542 acc5_neg: 100.000 time: 0.1862 data: 0.0921 max mem: 3534 +Test: [ 840/2387] eta: 0:05:00 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.2862 (0.2888) acc1: 74.2188 (90.6501) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1803 data: 0.0856 max mem: 3534 +Test: [ 850/2387] eta: 0:04:58 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1727 (0.2873) acc1: 99.2188 (90.7459) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1545 data: 0.0602 max mem: 3534 +Test: [ 860/2387] eta: 0:04:55 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1596 (0.2858) acc1: 99.7396 (90.8464) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1633 data: 0.0697 max mem: 3534 +Test: [ 870/2387] eta: 0:04:53 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1536 (0.2845) acc1: 99.7396 (90.9407) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1741 data: 0.0790 max mem: 3534 +Test: [ 880/2387] eta: 0:04:51 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1705 (0.2832) acc1: 99.4792 (91.0335) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1867 data: 0.0928 max mem: 3534 +Test: [ 890/2387] eta: 0:04:49 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1980 (0.2836) acc1: 97.9167 (91.0558) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1944 data: 0.1007 max mem: 3534 +Test: [ 900/2387] eta: 0:04:47 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.3748 (0.2870) acc1: 88.0208 (90.8201) acc5: 100.0000 (100.0000) acc1_neg: 65.104 acc5_neg: 100.000 time: 0.1942 data: 0.1002 max mem: 3535 +Test: [ 910/2387] eta: 0:04:45 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.6385 (0.2910) acc1: 63.8021 (90.5107) acc5: 100.0000 (100.0000) acc1_neg: 61.979 acc5_neg: 100.000 time: 0.1917 data: 0.0985 max mem: 3535 +Test: [ 920/2387] eta: 0:04:43 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.6513 (0.2949) acc1: 61.9792 (90.2043) acc5: 100.0000 (100.0000) acc1_neg: 58.073 acc5_neg: 100.000 time: 0.1922 data: 0.1004 max mem: 3535 +Test: [ 930/2387] eta: 0:04:41 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.6541 (0.2988) acc1: 61.9792 (89.9098) acc5: 100.0000 (100.0000) acc1_neg: 58.333 acc5_neg: 100.000 time: 0.1949 data: 0.1026 max mem: 3535 +Test: [ 940/2387] eta: 0:04:39 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.6644 (0.3025) acc1: 61.7188 (89.6146) acc5: 100.0000 (100.0000) acc1_neg: 60.156 acc5_neg: 100.000 time: 0.1776 data: 0.0815 max mem: 3535 +Test: [ 950/2387] eta: 0:04:37 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.6485 (0.3061) acc1: 62.2396 (89.3388) acc5: 100.0000 (100.0000) acc1_neg: 62.240 acc5_neg: 100.000 time: 0.1700 data: 0.0745 max mem: 3535 +Test: [ 960/2387] eta: 0:04:35 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.6405 (0.3096) acc1: 64.0625 (89.0850) acc5: 100.0000 (100.0000) acc1_neg: 66.927 acc5_neg: 100.000 time: 0.1831 data: 0.0903 max mem: 3535 +Test: [ 970/2387] eta: 0:04:33 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.6430 (0.3130) acc1: 63.8021 (88.8077) acc5: 100.0000 (100.0000) acc1_neg: 63.802 acc5_neg: 100.000 time: 0.1836 data: 0.0919 max mem: 3535 +Test: [ 980/2387] eta: 0:04:31 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.6224 (0.3147) acc1: 63.8021 (88.6757) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1817 data: 0.0882 max mem: 3535 +Test: [ 990/2387] eta: 0:04:29 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.3382 (0.3145) acc1: 83.0729 (88.6828) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1803 data: 0.0857 max mem: 3535 +Test: [1000/2387] eta: 0:04:27 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.2516 (0.3140) acc1: 92.7083 (88.7201) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.1812 data: 0.0891 max mem: 3535 +Test: [1010/2387] eta: 0:04:25 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.3233 (0.3156) acc1: 87.5000 (88.5970) acc5: 100.0000 (100.0000) acc1_neg: 68.490 acc5_neg: 100.000 time: 0.1901 data: 0.0990 max mem: 3535 +Test: [1020/2387] eta: 0:04:22 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.3621 (0.3172) acc1: 72.3958 (88.4738) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.1837 data: 0.0898 max mem: 3535 +Test: [1030/2387] eta: 0:04:20 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.3031 (0.3166) acc1: 89.0625 (88.5179) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1737 data: 0.0792 max mem: 3535 +Test: [1040/2387] eta: 0:04:18 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1910 (0.3153) acc1: 97.9167 (88.6152) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1784 data: 0.0871 max mem: 3535 +Test: [1050/2387] eta: 0:04:16 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1798 (0.3142) acc1: 97.9167 (88.6884) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1859 data: 0.0943 max mem: 3535 +Test: [1060/2387] eta: 0:04:14 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.2324 (0.3139) acc1: 94.5312 (88.7216) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1921 data: 0.0989 max mem: 3535 +Test: [1070/2387] eta: 0:04:12 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.2990 (0.3145) acc1: 89.8438 (88.6744) acc5: 100.0000 (100.0000) acc1_neg: 55.469 acc5_neg: 100.000 time: 0.1846 data: 0.0900 max mem: 3535 +Test: [1080/2387] eta: 0:04:10 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.5187 (0.3173) acc1: 66.9271 (88.4571) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.1868 data: 0.0929 max mem: 3535 +Test: [1090/2387] eta: 0:04:08 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.3281 (0.3168) acc1: 77.8646 (88.4963) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1942 data: 0.1002 max mem: 3535 +Test: [1100/2387] eta: 0:04:06 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.2408 (0.3158) acc1: 95.5729 (88.5731) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1855 data: 0.0917 max mem: 3535 +Test: [1110/2387] eta: 0:04:05 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1931 (0.3148) acc1: 97.6562 (88.6532) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1918 data: 0.1009 max mem: 3535 +Test: [1120/2387] eta: 0:04:02 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.2192 (0.3141) acc1: 96.0938 (88.6985) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.1847 data: 0.0911 max mem: 3536 +Test: [1130/2387] eta: 0:04:00 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.2568 (0.3138) acc1: 92.7083 (88.7284) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1740 data: 0.0796 max mem: 3536 +Test: [1140/2387] eta: 0:03:58 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.3130 (0.3161) acc1: 89.3229 (88.5437) acc5: 100.0000 (100.0000) acc1_neg: 53.646 acc5_neg: 100.000 time: 0.1784 data: 0.0871 max mem: 3536 +Test: [1150/2387] eta: 0:03:56 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.4801 (0.3161) acc1: 71.8750 (88.5283) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0890 max mem: 3536 +Test: [1160/2387] eta: 0:03:54 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1224 (0.3144) acc1: 100.0000 (88.6242) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1858 data: 0.0946 max mem: 3536 +Test: [1170/2387] eta: 0:03:53 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1125 (0.3127) acc1: 100.0000 (88.7182) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2064 data: 0.1140 max mem: 3536 +Test: [1180/2387] eta: 0:03:51 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1146 (0.3110) acc1: 100.0000 (88.8109) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1931 data: 0.1011 max mem: 3536 +Test: [1190/2387] eta: 0:03:49 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1191 (0.3095) acc1: 100.0000 (88.9020) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1741 data: 0.0829 max mem: 3536 +Test: [1200/2387] eta: 0:03:47 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1305 (0.3083) acc1: 100.0000 (88.9864) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1853 data: 0.0947 max mem: 3536 +Test: [1210/2387] eta: 0:03:45 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1638 (0.3072) acc1: 99.7396 (89.0606) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1914 data: 0.1005 max mem: 3536 +Test: [1220/2387] eta: 0:03:43 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.0914 (0.3055) acc1: 99.7396 (89.1487) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1803 data: 0.0891 max mem: 3536 +Test: [1230/2387] eta: 0:03:41 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.0843 (0.3036) acc1: 100.0000 (89.2368) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1718 data: 0.0806 max mem: 3536 +Test: [1240/2387] eta: 0:03:38 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.0703 (0.3018) acc1: 100.0000 (89.3231) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1750 data: 0.0837 max mem: 3536 +Test: [1250/2387] eta: 0:03:37 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.0742 (0.3000) acc1: 100.0000 (89.4066) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1803 data: 0.0881 max mem: 3536 +Test: [1260/2387] eta: 0:03:34 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.0864 (0.2984) acc1: 100.0000 (89.4867) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1804 data: 0.0860 max mem: 3536 +Test: [1270/2387] eta: 0:03:33 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1256 (0.2974) acc1: 99.4792 (89.5446) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1854 data: 0.0924 max mem: 3536 +Test: [1280/2387] eta: 0:03:30 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1917 (0.2969) acc1: 96.8750 (89.5931) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1751 data: 0.0833 max mem: 3536 +Test: [1290/2387] eta: 0:03:28 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1996 (0.2961) acc1: 97.6562 (89.6606) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1655 data: 0.0735 max mem: 3536 +Test: [1300/2387] eta: 0:03:26 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1982 (0.2954) acc1: 98.4375 (89.7217) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1752 data: 0.0833 max mem: 3536 +Test: [1310/2387] eta: 0:03:24 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.2026 (0.2948) acc1: 97.3958 (89.7703) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1633 data: 0.0713 max mem: 3536 +Test: [1320/2387] eta: 0:03:22 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.2290 (0.2946) acc1: 95.8333 (89.8041) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1586 data: 0.0665 max mem: 3536 +Test: [1330/2387] eta: 0:03:20 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.3045 (0.2960) acc1: 89.8438 (89.6904) acc5: 100.0000 (100.0000) acc1_neg: 60.938 acc5_neg: 100.000 time: 0.1575 data: 0.0642 max mem: 3536 +Test: [1340/2387] eta: 0:03:18 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.4957 (0.2974) acc1: 72.9167 (89.5711) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1636 data: 0.0700 max mem: 3536 +Test: [1350/2387] eta: 0:03:16 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1001 (0.2959) acc1: 98.9583 (89.6475) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1839 data: 0.0921 max mem: 3537 +Test: [1360/2387] eta: 0:03:14 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.0755 (0.2942) acc1: 100.0000 (89.7236) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1866 data: 0.0960 max mem: 3537 +Test: [1370/2387] eta: 0:03:12 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.0685 (0.2925) acc1: 100.0000 (89.7985) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1812 data: 0.0899 max mem: 3537 +Test: [1380/2387] eta: 0:03:10 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.0685 (0.2909) acc1: 100.0000 (89.8724) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1853 data: 0.0907 max mem: 3537 +Test: [1390/2387] eta: 0:03:08 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.0712 (0.2894) acc1: 100.0000 (89.9452) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1930 data: 0.0991 max mem: 3537 +Test: [1400/2387] eta: 0:03:06 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.0780 (0.2881) acc1: 100.0000 (90.0118) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1992 data: 0.1074 max mem: 3537 +Test: [1410/2387] eta: 0:03:04 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1377 (0.2884) acc1: 98.4375 (89.9901) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.1869 data: 0.0928 max mem: 3537 +Test: [1420/2387] eta: 0:03:02 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.2805 (0.2882) acc1: 91.6667 (90.0140) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1686 data: 0.0745 max mem: 3537 +Test: [1430/2387] eta: 0:03:00 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.2830 (0.2885) acc1: 91.6667 (90.0012) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.1678 data: 0.0755 max mem: 3537 +Test: [1440/2387] eta: 0:02:58 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.3592 (0.2894) acc1: 86.4583 (89.9343) acc5: 100.0000 (100.0000) acc1_neg: 52.604 acc5_neg: 100.000 time: 0.1649 data: 0.0720 max mem: 3537 +Test: [1450/2387] eta: 0:02:56 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.4837 (0.2911) acc1: 73.9583 (89.7928) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1668 data: 0.0736 max mem: 3537 +Test: [1460/2387] eta: 0:02:54 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.2167 (0.2905) acc1: 94.5312 (89.8407) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1759 data: 0.0847 max mem: 3537 +Test: [1470/2387] eta: 0:02:52 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1660 (0.2895) acc1: 98.6979 (89.9041) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1537 data: 0.0610 max mem: 3537 +Test: [1480/2387] eta: 0:02:50 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1500 (0.2887) acc1: 98.9583 (89.9565) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1462 data: 0.0505 max mem: 3537 +Test: [1490/2387] eta: 0:02:48 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1547 (0.2879) acc1: 98.4375 (90.0118) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1751 data: 0.0797 max mem: 3537 +Test: [1500/2387] eta: 0:02:46 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1818 (0.2873) acc1: 98.1771 (90.0568) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1682 data: 0.0761 max mem: 3537 +Test: [1510/2387] eta: 0:02:44 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.2139 (0.2873) acc1: 96.6146 (90.0959) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1601 data: 0.0681 max mem: 3537 +Test: [1520/2387] eta: 0:02:42 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.2139 (0.2866) acc1: 98.6979 (90.1518) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1792 data: 0.0852 max mem: 3537 +Test: [1530/2387] eta: 0:02:40 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1552 (0.2857) acc1: 99.4792 (90.2130) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1981 data: 0.1037 max mem: 3537 +Test: [1540/2387] eta: 0:02:38 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1459 (0.2848) acc1: 99.7396 (90.2728) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0932 max mem: 3537 +Test: [1550/2387] eta: 0:02:36 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1454 (0.2839) acc1: 99.7396 (90.3318) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1599 data: 0.0620 max mem: 3537 +Test: [1560/2387] eta: 0:02:35 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1472 (0.2831) acc1: 99.4792 (90.3904) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1674 data: 0.0737 max mem: 3537 +Test: [1570/2387] eta: 0:02:33 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1602 (0.2829) acc1: 98.9583 (90.4234) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1815 data: 0.0909 max mem: 3537 +Test: [1580/2387] eta: 0:02:31 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.2919 (0.2832) acc1: 94.0104 (90.4166) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1774 data: 0.0854 max mem: 3538 +Test: [1590/2387] eta: 0:02:29 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.2730 (0.2829) acc1: 95.0521 (90.4477) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1783 data: 0.0859 max mem: 3538 +Test: [1600/2387] eta: 0:02:27 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.2136 (0.2825) acc1: 96.3542 (90.4835) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1802 data: 0.0878 max mem: 3538 +Test: [1610/2387] eta: 0:02:25 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.2136 (0.2821) acc1: 97.3958 (90.5204) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1849 data: 0.0909 max mem: 3538 +Test: [1620/2387] eta: 0:02:23 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.2602 (0.2822) acc1: 91.9271 (90.5019) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1596 data: 0.0639 max mem: 3538 +Test: [1630/2387] eta: 0:02:21 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.3281 (0.2831) acc1: 84.6354 (90.4390) acc5: 100.0000 (100.0000) acc1_neg: 72.396 acc5_neg: 100.000 time: 0.1749 data: 0.0813 max mem: 3538 +Test: [1640/2387] eta: 0:02:19 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.5005 (0.2849) acc1: 72.9167 (90.2989) acc5: 100.0000 (100.0000) acc1_neg: 72.917 acc5_neg: 100.000 time: 0.2099 data: 0.1157 max mem: 3538 +Test: [1650/2387] eta: 0:02:17 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.5781 (0.2866) acc1: 66.6667 (90.1639) acc5: 100.0000 (100.0000) acc1_neg: 72.135 acc5_neg: 100.000 time: 0.1808 data: 0.0829 max mem: 3538 +Test: [1660/2387] eta: 0:02:15 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.5402 (0.2881) acc1: 69.7917 (90.0491) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.1648 data: 0.0704 max mem: 3538 +Test: [1670/2387] eta: 0:02:14 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.5097 (0.2895) acc1: 71.6146 (89.9408) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.1807 data: 0.0892 max mem: 3538 +Test: [1680/2387] eta: 0:02:12 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.5247 (0.2910) acc1: 71.6146 (89.8377) acc5: 100.0000 (100.0000) acc1_neg: 71.615 acc5_neg: 100.000 time: 0.1701 data: 0.0789 max mem: 3538 +Test: [1690/2387] eta: 0:02:10 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.5269 (0.2925) acc1: 71.6146 (89.7242) acc5: 100.0000 (100.0000) acc1_neg: 73.958 acc5_neg: 100.000 time: 0.1532 data: 0.0624 max mem: 3538 +Test: [1700/2387] eta: 0:02:08 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.5269 (0.2933) acc1: 71.6146 (89.6570) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0783 max mem: 3538 +Test: [1710/2387] eta: 0:02:06 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.0980 (0.2921) acc1: 99.4792 (89.7154) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1701 data: 0.0789 max mem: 3538 +Test: [1720/2387] eta: 0:02:04 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.0784 (0.2909) acc1: 100.0000 (89.7743) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1446 data: 0.0511 max mem: 3538 +Test: [1730/2387] eta: 0:02:02 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.0721 (0.2896) acc1: 100.0000 (89.8325) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1469 data: 0.0508 max mem: 3538 +Test: [1740/2387] eta: 0:02:00 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.0781 (0.2885) acc1: 100.0000 (89.8889) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1722 data: 0.0783 max mem: 3538 +Test: [1750/2387] eta: 0:01:58 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.0884 (0.2874) acc1: 99.7396 (89.9437) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1777 data: 0.0854 max mem: 3538 +Test: [1760/2387] eta: 0:01:56 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1274 (0.2870) acc1: 98.6979 (89.9675) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1681 data: 0.0761 max mem: 3538 +Test: [1770/2387] eta: 0:01:54 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1747 (0.2863) acc1: 96.0938 (90.0043) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1651 data: 0.0740 max mem: 3538 +Test: [1780/2387] eta: 0:01:52 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1218 (0.2854) acc1: 98.4375 (90.0518) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1763 data: 0.0844 max mem: 3538 +Test: [1790/2387] eta: 0:01:50 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1014 (0.2843) acc1: 99.7396 (90.1058) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1747 data: 0.0814 max mem: 3538 +Test: [1800/2387] eta: 0:01:48 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.0961 (0.2833) acc1: 100.0000 (90.1598) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1804 data: 0.0875 max mem: 3538 +Test: [1810/2387] eta: 0:01:47 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1064 (0.2824) acc1: 99.7396 (90.2087) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1970 data: 0.1056 max mem: 3539 +Test: [1820/2387] eta: 0:01:45 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1242 (0.2820) acc1: 98.4375 (90.2327) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2042 data: 0.1138 max mem: 3539 +Test: [1830/2387] eta: 0:01:43 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.2255 (0.2818) acc1: 93.7500 (90.2464) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2028 data: 0.1104 max mem: 3539 +Test: [1840/2387] eta: 0:01:41 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1701 (0.2812) acc1: 98.6979 (90.2937) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2144 data: 0.1194 max mem: 3539 +Test: [1850/2387] eta: 0:01:39 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1660 (0.2805) acc1: 98.9583 (90.3415) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2160 data: 0.1230 max mem: 3539 +Test: [1860/2387] eta: 0:01:38 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1674 (0.2799) acc1: 98.9583 (90.3892) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1949 data: 0.1046 max mem: 3539 +Test: [1870/2387] eta: 0:01:36 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1651 (0.2793) acc1: 99.2188 (90.4363) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1905 data: 0.1003 max mem: 3539 +Test: [1880/2387] eta: 0:01:34 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1644 (0.2787) acc1: 99.2188 (90.4825) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1657 data: 0.0740 max mem: 3539 +Test: [1890/2387] eta: 0:01:32 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1669 (0.2781) acc1: 99.2188 (90.5279) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1581 data: 0.0643 max mem: 3539 +Test: [1900/2387] eta: 0:01:30 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1687 (0.2776) acc1: 98.9583 (90.5709) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1791 data: 0.0856 max mem: 3539 +Test: [1910/2387] eta: 0:01:28 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.2349 (0.2775) acc1: 97.1354 (90.5962) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1749 data: 0.0816 max mem: 3539 +Test: [1920/2387] eta: 0:01:26 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.2642 (0.2780) acc1: 92.7083 (90.5534) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1650 data: 0.0729 max mem: 3539 +Test: [1930/2387] eta: 0:01:24 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.2581 (0.2777) acc1: 92.9688 (90.5818) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1605 data: 0.0693 max mem: 3539 +Test: [1940/2387] eta: 0:01:22 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.2155 (0.2775) acc1: 95.8333 (90.6141) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1655 data: 0.0739 max mem: 3539 +Test: [1950/2387] eta: 0:01:20 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.2451 (0.2775) acc1: 95.8333 (90.6161) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.1593 data: 0.0671 max mem: 3539 +Test: [1960/2387] eta: 0:01:19 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.2934 (0.2779) acc1: 88.8021 (90.5943) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.1530 data: 0.0605 max mem: 3539 +Test: [1970/2387] eta: 0:01:17 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.4300 (0.2793) acc1: 77.8646 (90.4690) acc5: 100.0000 (100.0000) acc1_neg: 72.135 acc5_neg: 100.000 time: 0.1573 data: 0.0665 max mem: 3539 +Test: [1980/2387] eta: 0:01:15 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.5228 (0.2800) acc1: 66.9271 (90.4116) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1536 data: 0.0629 max mem: 3539 +Test: [1990/2387] eta: 0:01:13 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1278 (0.2793) acc1: 99.4792 (90.4576) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1682 data: 0.0775 max mem: 3539 +Test: [2000/2387] eta: 0:01:11 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1266 (0.2785) acc1: 99.4792 (90.5029) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1890 data: 0.0983 max mem: 3539 +Test: [2010/2387] eta: 0:01:09 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1271 (0.2778) acc1: 99.4792 (90.5480) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1992 data: 0.1072 max mem: 3539 +Test: [2020/2387] eta: 0:01:07 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1273 (0.2770) acc1: 99.7396 (90.5930) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1771 data: 0.0836 max mem: 3539 +Test: [2030/2387] eta: 0:01:05 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1279 (0.2763) acc1: 99.7396 (90.6376) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1463 data: 0.0514 max mem: 3540 +Test: [2040/2387] eta: 0:01:03 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1272 (0.2756) acc1: 99.7396 (90.6818) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1499 data: 0.0541 max mem: 3540 +Test: [2050/2387] eta: 0:01:02 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1272 (0.2749) acc1: 99.7396 (90.7253) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1536 data: 0.0567 max mem: 3540 +Test: [2060/2387] eta: 0:01:00 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1490 (0.2743) acc1: 99.4792 (90.7685) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1599 data: 0.0628 max mem: 3540 +Test: [2070/2387] eta: 0:00:58 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1507 (0.2737) acc1: 99.7396 (90.8119) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1599 data: 0.0642 max mem: 3540 +Test: [2080/2387] eta: 0:00:56 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1504 (0.2731) acc1: 99.7396 (90.8545) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1749 data: 0.0801 max mem: 3540 +Test: [2090/2387] eta: 0:00:54 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1497 (0.2725) acc1: 99.7396 (90.8977) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0739 max mem: 3540 +Test: [2100/2387] eta: 0:00:52 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1515 (0.2719) acc1: 99.7396 (90.9396) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1499 data: 0.0529 max mem: 3540 +Test: [2110/2387] eta: 0:00:50 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1530 (0.2714) acc1: 99.7396 (90.9814) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1749 data: 0.0769 max mem: 3540 +Test: [2120/2387] eta: 0:00:49 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1560 (0.2716) acc1: 99.4792 (90.9747) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1649 data: 0.0667 max mem: 3540 +Test: [2130/2387] eta: 0:00:47 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.2455 (0.2714) acc1: 93.7500 (90.9981) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1392 data: 0.0439 max mem: 3540 +Test: [2140/2387] eta: 0:00:45 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.2115 (0.2711) acc1: 97.3958 (91.0254) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1642 data: 0.0699 max mem: 3540 +Test: [2150/2387] eta: 0:00:43 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.2236 (0.2710) acc1: 96.6146 (91.0467) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1989 data: 0.1046 max mem: 3540 +Test: [2160/2387] eta: 0:00:41 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.2495 (0.2713) acc1: 91.9271 (91.0252) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.1649 data: 0.0722 max mem: 3540 +Test: [2170/2387] eta: 0:00:39 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.3795 (0.2724) acc1: 83.3333 (90.9509) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.1509 data: 0.0554 max mem: 3540 +Test: [2180/2387] eta: 0:00:37 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.6045 (0.2744) acc1: 64.3229 (90.7990) acc5: 100.0000 (100.0000) acc1_neg: 51.042 acc5_neg: 100.000 time: 0.1814 data: 0.0841 max mem: 3540 +Test: [2190/2387] eta: 0:00:36 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.5513 (0.2745) acc1: 70.3125 (90.7982) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1663 data: 0.0724 max mem: 3540 +Test: [2200/2387] eta: 0:00:34 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1879 (0.2741) acc1: 98.4375 (90.8283) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1634 data: 0.0703 max mem: 3540 +Test: [2210/2387] eta: 0:00:32 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1732 (0.2736) acc1: 99.2188 (90.8670) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1925 data: 0.0975 max mem: 3540 +Test: [2220/2387] eta: 0:00:30 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1593 (0.2731) acc1: 99.4792 (90.9052) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2049 data: 0.1096 max mem: 3540 +Test: [2230/2387] eta: 0:00:28 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1668 (0.2726) acc1: 99.2188 (90.9390) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1668 data: 0.0723 max mem: 3540 +Test: [2240/2387] eta: 0:00:26 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1817 (0.2722) acc1: 98.9583 (90.9714) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2002 data: 0.1067 max mem: 3540 +Test: [2250/2387] eta: 0:00:25 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1948 (0.2723) acc1: 97.6562 (90.9878) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2341 data: 0.1391 max mem: 3540 +Test: [2260/2387] eta: 0:00:23 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.2881 (0.2724) acc1: 94.5312 (90.9921) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1598 data: 0.0638 max mem: 3541 +Test: [2270/2387] eta: 0:00:21 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.2462 (0.2722) acc1: 95.0521 (91.0203) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1598 data: 0.0676 max mem: 3541 +Test: [2280/2387] eta: 0:00:19 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1935 (0.2718) acc1: 98.1771 (91.0531) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1973 data: 0.1064 max mem: 3541 +Test: [2290/2387] eta: 0:00:17 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.1919 (0.2715) acc1: 98.1771 (91.0802) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1999 data: 0.1085 max mem: 3541 +Test: [2300/2387] eta: 0:00:15 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.2340 (0.2717) acc1: 93.2292 (91.0655) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1924 data: 0.1018 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.3728 (0.2725) acc1: 83.8542 (91.0159) acc5: 100.0000 (100.0000) acc1_neg: 57.812 acc5_neg: 100.000 time: 0.1835 data: 0.0915 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.5022 (0.2736) acc1: 75.0000 (90.9270) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.1777 data: 0.0844 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.4526 (0.2742) acc1: 75.0000 (90.8854) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.1815 data: 0.0874 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.3086 (0.2743) acc1: 89.8438 (90.8920) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1869 data: 0.0930 max mem: 3541 +Test: [2350/2387] eta: 0:00:06 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.2626 (0.2744) acc1: 92.1875 (90.8949) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1747 data: 0.0801 max mem: 3541 +Test: [2360/2387] eta: 0:00:04 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.3430 (0.2749) acc1: 88.8021 (90.8639) acc5: 100.0000 (100.0000) acc1_neg: 73.958 acc5_neg: 100.000 time: 0.1705 data: 0.0763 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.4476 (0.2757) acc1: 77.3438 (90.8018) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.1435 data: 0.0526 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.5361 (0.2774) acc1: 68.2292 (90.6758) acc5: 100.0000 (100.0000) acc1_neg: 63.021 acc5_neg: 100.000 time: 0.1211 data: 0.0307 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 90.411 acc5_pos: 100.000 loss: 0.6006 (0.2779) acc1: 65.1042 (90.6376) acc5: 100.0000 (100.0000) acc1_neg: 93.023 acc5_neg: 100.000 time: 0.1213 data: 0.0299 max mem: 3541 +Test: Total time: 0:07:16 (0.1827 s / it) +* Overall Acc@1 90.638 Acc@5 100.000 + target == real Acc@1 70.841 Acc@5 100.000 + target == fake Acc@1 92.028 Acc@5 100.000 + loss 0.278 + PR-AUC 0.9941 ROC-AUC 0.9261 EER 14.55% (thr=0.6367) +Accuracy of the network on the 916267 test images: 90.6% +[rank0]:[W725 18:26:19.362794439 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:26:38.714000 2575530 torch/distributed/run.py:766] +W0725 18:26:38.714000 2575530 torch/distributed/run.py:766] ***************************************** +W0725 18:26:38.714000 2575530 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:26:38.714000 2575530 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:27:07.874935810 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:27:07.951887142 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:27:08.584998113 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +[rank0]:[W725 18:27:08.712996882 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:27:08.715989643 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +| distributed init (rank 3): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:27:08.755539146 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank4]:[W725 18:27:08.755591076 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W725 18:27:08.761818732 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed3_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed3_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 0:59:58 acc1_pos: 59.115 acc5_pos: 100.000 loss: 0.6576 (0.6576) acc1: 59.1146 (59.1146) acc5: 100.0000 (100.0000) time: 6.9600 data: 5.5472 max mem: 3531 +Test: [ 10/517] eta: 0:06:03 acc1_pos: 23.698 acc5_pos: 100.000 loss: 0.6405 (0.6988) acc1: 61.4583 (61.1269) acc5: 100.0000 (100.0000) time: 0.7166 data: 0.5046 max mem: 3531 +Test: [ 20/517] eta: 0:03:30 acc1_pos: 48.438 acc5_pos: 100.000 loss: 0.7726 (0.7671) acc1: 53.3854 (53.6334) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:03:02 acc1_pos: 52.865 acc5_pos: 100.000 loss: 0.7497 (0.7440) acc1: 48.4375 (54.7379) acc5: 100.0000 (100.0000) time: 0.1854 data: 0.0895 max mem: 3531 +Test: [ 40/517] eta: 0:02:52 acc1_pos: 72.135 acc5_pos: 100.000 loss: 0.6774 (0.7227) acc1: 59.1146 (57.3171) acc5: 100.0000 (100.0000) time: 0.2951 data: 0.2017 max mem: 3531 +Test: [ 50/517] eta: 0:02:34 acc1_pos: 59.633 acc5_pos: 100.000 loss: 0.6455 (0.7224) acc1: 61.1979 (57.2457) acc5: 100.0000 (100.0000) acc1_neg: 60.156 acc5_neg: 100.000 time: 0.2617 data: 0.1682 max mem: 3531 +Test: [ 60/517] eta: 0:02:26 acc1_pos: 59.633 acc5_pos: 100.000 loss: 0.6455 (0.7144) acc1: 60.9375 (57.8467) acc5: 100.0000 (100.0000) acc1_neg: 35.417 acc5_neg: 100.000 time: 0.2357 data: 0.1436 max mem: 3531 +Test: [ 70/517] eta: 0:02:17 acc1_pos: 59.633 acc5_pos: 100.000 loss: 0.6623 (0.7085) acc1: 60.9375 (57.8565) acc5: 100.0000 (100.0000) acc1_neg: 69.531 acc5_neg: 100.000 time: 0.2536 data: 0.1595 max mem: 3531 +Test: [ 80/517] eta: 0:02:12 acc1_pos: 59.633 acc5_pos: 100.000 loss: 0.6507 (0.6991) acc1: 63.0208 (58.4973) acc5: 100.0000 (100.0000) acc1_neg: 52.344 acc5_neg: 100.000 time: 0.2495 data: 0.1561 max mem: 3531 +Test: [ 90/517] eta: 0:02:07 acc1_pos: 59.633 acc5_pos: 100.000 loss: 0.5968 (0.6849) acc1: 64.3229 (59.4923) acc5: 100.0000 (100.0000) acc1_neg: 64.062 acc5_neg: 100.000 time: 0.2639 data: 0.1684 max mem: 3531 +Test: [100/517] eta: 0:02:03 acc1_pos: 59.633 acc5_pos: 100.000 loss: 0.5420 (0.6717) acc1: 70.3125 (60.4502) acc5: 100.0000 (100.0000) acc1_neg: 61.979 acc5_neg: 100.000 time: 0.2693 data: 0.1702 max mem: 3531 +Test: [110/517] eta: 0:01:57 acc1_pos: 59.633 acc5_pos: 100.000 loss: 0.5133 (0.6574) acc1: 74.7396 (61.8971) acc5: 100.0000 (100.0000) acc1_neg: 77.083 acc5_neg: 100.000 time: 0.2399 data: 0.1441 max mem: 3531 +Test: [120/517] eta: 0:01:52 acc1_pos: 59.633 acc5_pos: 100.000 loss: 0.5743 (0.6522) acc1: 69.5312 (62.2525) acc5: 100.0000 (100.0000) acc1_neg: 64.844 acc5_neg: 100.000 time: 0.2199 data: 0.1247 max mem: 3531 +Test: [130/517] eta: 0:01:48 acc1_pos: 59.633 acc5_pos: 100.000 loss: 0.5777 (0.6449) acc1: 68.2292 (63.0864) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.2299 data: 0.1334 max mem: 3531 +Test: [140/517] eta: 0:01:44 acc1_pos: 59.633 acc5_pos: 100.000 loss: 0.5361 (0.6366) acc1: 74.7396 (64.0126) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.2399 data: 0.1426 max mem: 3531 +Test: [150/517] eta: 0:01:40 acc1_pos: 59.633 acc5_pos: 100.000 loss: 0.5099 (0.6244) acc1: 77.3438 (65.3680) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2399 data: 0.1417 max mem: 3531 +Test: [160/517] eta: 0:01:36 acc1_pos: 59.633 acc5_pos: 100.000 loss: 0.4393 (0.6124) acc1: 84.3750 (66.6731) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.2299 data: 0.1317 max mem: 3531 +Test: [170/517] eta: 0:01:33 acc1_pos: 59.633 acc5_pos: 100.000 loss: 0.4081 (0.5985) acc1: 88.8021 (68.0921) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2299 data: 0.1340 max mem: 3531 +Test: [180/517] eta: 0:01:30 acc1_pos: 59.633 acc5_pos: 100.000 loss: 0.3747 (0.5906) acc1: 90.3646 (68.7457) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.2399 data: 0.1457 max mem: 3531 +Test: [190/517] eta: 0:01:26 acc1_pos: 59.633 acc5_pos: 100.000 loss: 0.3520 (0.5781) acc1: 92.9688 (70.1012) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2352 data: 0.1410 max mem: 3531 +Test: [200/517] eta: 0:01:24 acc1_pos: 59.633 acc5_pos: 100.000 loss: 0.3755 (0.5713) acc1: 92.7083 (70.7582) acc5: 100.0000 (100.0000) acc1_neg: 74.219 acc5_neg: 100.000 time: 0.2399 data: 0.1443 max mem: 3531 +Test: [210/517] eta: 0:01:20 acc1_pos: 59.633 acc5_pos: 100.000 loss: 0.4428 (0.5653) acc1: 85.1562 (71.5245) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.2396 data: 0.1429 max mem: 3532 +Test: [220/517] eta: 0:01:18 acc1_pos: 59.633 acc5_pos: 100.000 loss: 0.4407 (0.5593) acc1: 87.5000 (72.2910) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.2399 data: 0.1443 max mem: 3532 +Test: [230/517] eta: 0:01:14 acc1_pos: 59.633 acc5_pos: 100.000 loss: 0.4530 (0.5557) acc1: 89.5833 (72.9764) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.2449 data: 0.1483 max mem: 3532 +Test: [240/517] eta: 0:01:12 acc1_pos: 59.633 acc5_pos: 100.000 loss: 0.5009 (0.5568) acc1: 80.9896 (72.8454) acc5: 100.0000 (100.0000) acc1_neg: 46.615 acc5_neg: 100.000 time: 0.2349 data: 0.1369 max mem: 3532 +Test: [250/517] eta: 0:01:09 acc1_pos: 59.633 acc5_pos: 100.000 loss: 0.5592 (0.5564) acc1: 73.9583 (73.1117) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.2449 data: 0.1484 max mem: 3532 +Test: [260/517] eta: 0:01:06 acc1_pos: 59.633 acc5_pos: 100.000 loss: 0.5282 (0.5548) acc1: 81.5104 (73.5034) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.2299 data: 0.1346 max mem: 3532 +Test: [270/517] eta: 0:01:03 acc1_pos: 59.633 acc5_pos: 100.000 loss: 0.4844 (0.5523) acc1: 82.5521 (73.8459) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.2299 data: 0.1326 max mem: 3532 +Test: [280/517] eta: 0:01:01 acc1_pos: 59.633 acc5_pos: 100.000 loss: 0.4612 (0.5490) acc1: 84.3750 (74.2827) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.2499 data: 0.1511 max mem: 3532 +Test: [290/517] eta: 0:00:58 acc1_pos: 59.633 acc5_pos: 100.000 loss: 0.4467 (0.5460) acc1: 86.9792 (74.6358) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.2399 data: 0.1422 max mem: 3532 +Test: [300/517] eta: 0:00:55 acc1_pos: 59.633 acc5_pos: 100.000 loss: 0.4738 (0.5445) acc1: 78.9062 (74.7318) acc5: 100.0000 (100.0000) acc1_neg: 59.375 acc5_neg: 100.000 time: 0.2349 data: 0.1393 max mem: 3532 +Test: [310/517] eta: 0:00:52 acc1_pos: 59.633 acc5_pos: 100.000 loss: 0.5418 (0.5464) acc1: 76.3021 (74.5713) acc5: 100.0000 (100.0000) acc1_neg: 60.156 acc5_neg: 100.000 time: 0.2399 data: 0.1448 max mem: 3532 +Test: [320/517] eta: 0:00:50 acc1_pos: 59.633 acc5_pos: 100.000 loss: 0.5564 (0.5457) acc1: 72.1354 (74.6187) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.2449 data: 0.1483 max mem: 3532 +Test: [330/517] eta: 0:00:47 acc1_pos: 59.633 acc5_pos: 100.000 loss: 0.4805 (0.5430) acc1: 80.4688 (74.9127) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.2499 data: 0.1504 max mem: 3532 +Test: [340/517] eta: 0:00:45 acc1_pos: 59.633 acc5_pos: 100.000 loss: 0.4567 (0.5446) acc1: 84.1146 (74.7694) acc5: 100.0000 (100.0000) acc1_neg: 27.604 acc5_neg: 100.000 time: 0.2449 data: 0.1450 max mem: 3532 +Test: [350/517] eta: 0:00:42 acc1_pos: 59.633 acc5_pos: 100.000 loss: 0.5187 (0.5469) acc1: 76.0417 (74.4562) acc5: 100.0000 (100.0000) acc1_neg: 66.146 acc5_neg: 100.000 time: 0.2449 data: 0.1468 max mem: 3532 +Test: [360/517] eta: 0:00:39 acc1_pos: 59.633 acc5_pos: 100.000 loss: 0.5511 (0.5486) acc1: 70.5729 (74.1459) acc5: 100.0000 (100.0000) acc1_neg: 35.417 acc5_neg: 100.000 time: 0.2449 data: 0.1487 max mem: 3532 +Test: [370/517] eta: 0:00:37 acc1_pos: 59.633 acc5_pos: 100.000 loss: 0.6558 (0.5517) acc1: 57.8125 (73.7267) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.2349 data: 0.1399 max mem: 3532 +Test: [380/517] eta: 0:00:34 acc1_pos: 59.633 acc5_pos: 100.000 loss: 0.6102 (0.5520) acc1: 61.9792 (73.6009) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.2349 data: 0.1394 max mem: 3532 +Test: [390/517] eta: 0:00:32 acc1_pos: 59.633 acc5_pos: 100.000 loss: 0.5808 (0.5543) acc1: 66.1458 (73.3196) acc5: 100.0000 (100.0000) acc1_neg: 67.969 acc5_neg: 100.000 time: 0.2399 data: 0.1413 max mem: 3532 +Test: [400/517] eta: 0:00:29 acc1_pos: 59.633 acc5_pos: 100.000 loss: 0.5899 (0.5544) acc1: 63.2812 (73.1550) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.2399 data: 0.1382 max mem: 3532 +Test: [410/517] eta: 0:00:27 acc1_pos: 59.633 acc5_pos: 100.000 loss: 0.5899 (0.5562) acc1: 63.8021 (72.9655) acc5: 100.0000 (100.0000) acc1_neg: 62.240 acc5_neg: 100.000 time: 0.2449 data: 0.1475 max mem: 3532 +Test: [420/517] eta: 0:00:24 acc1_pos: 59.633 acc5_pos: 100.000 loss: 0.6664 (0.5628) acc1: 59.1146 (72.2672) acc5: 100.0000 (100.0000) acc1_neg: 73.698 acc5_neg: 100.000 time: 0.2351 data: 0.1424 max mem: 3532 +Test: [430/517] eta: 0:00:21 acc1_pos: 59.633 acc5_pos: 100.000 loss: 0.5828 (0.5624) acc1: 62.7604 (72.2557) acc5: 100.0000 (100.0000) acc1_neg: 51.823 acc5_neg: 100.000 time: 0.2349 data: 0.1403 max mem: 3532 +Test: [440/517] eta: 0:00:19 acc1_pos: 59.633 acc5_pos: 100.000 loss: 0.6389 (0.5653) acc1: 62.7604 (71.9270) acc5: 100.0000 (100.0000) acc1_neg: 57.812 acc5_neg: 100.000 time: 0.2447 data: 0.1482 max mem: 3533 +Test: [450/517] eta: 0:00:16 acc1_pos: 59.633 acc5_pos: 100.000 loss: 0.6361 (0.5660) acc1: 62.7604 (71.8392) acc5: 100.0000 (100.0000) acc1_neg: 66.406 acc5_neg: 100.000 time: 0.2399 data: 0.1443 max mem: 3533 +Test: [460/517] eta: 0:00:14 acc1_pos: 59.633 acc5_pos: 100.000 loss: 0.5991 (0.5672) acc1: 66.4062 (71.6468) acc5: 100.0000 (100.0000) acc1_neg: 71.094 acc5_neg: 100.000 time: 0.2449 data: 0.1484 max mem: 3533 +Test: [470/517] eta: 0:00:11 acc1_pos: 59.633 acc5_pos: 100.000 loss: 0.6079 (0.5689) acc1: 62.5000 (71.3558) acc5: 100.0000 (100.0000) acc1_neg: 56.510 acc5_neg: 100.000 time: 0.2253 data: 0.1298 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 59.633 acc5_pos: 100.000 loss: 0.6079 (0.5701) acc1: 61.4583 (71.2210) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.2199 data: 0.1233 max mem: 3533 +Test: [490/517] eta: 0:00:06 acc1_pos: 59.633 acc5_pos: 100.000 loss: 0.5039 (0.5680) acc1: 74.7396 (71.4756) acc5: 100.0000 (100.0000) acc1_neg: 70.573 acc5_neg: 100.000 time: 0.2445 data: 0.1470 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 59.633 acc5_pos: 100.000 loss: 0.5766 (0.5719) acc1: 65.3646 (70.9706) acc5: 100.0000 (100.0000) acc1_neg: 44.792 acc5_neg: 100.000 time: 0.2299 data: 0.1352 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 59.633 acc5_pos: 100.000 loss: 0.6966 (0.5734) acc1: 48.9583 (70.7778) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.2062 data: 0.1125 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 59.633 acc5_pos: 100.000 loss: 0.6541 (0.5738) acc1: 61.9792 (70.7439) acc5: 100.0000 (100.0000) acc1_neg: 66.784 acc5_neg: 100.000 time: 0.1846 data: 0.0907 max mem: 3533 +Test: Total time: 0:02:08 (0.2482 s / it) +* Overall Acc@1 70.744 Acc@5 100.000 + target == real Acc@1 57.056 Acc@5 100.000 + target == fake Acc@1 72.164 Acc@5 100.000 + loss 0.574 + PR-AUC 0.9452 ROC-AUC 0.6897 EER 35.80% (thr=0.5425) +Accuracy of the network on the 198427 test images: 70.7% +[rank0]:[W725 18:29:28.508164193 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0725 18:29:50.466000 2576117 torch/distributed/run.py:766] +W0725 18:29:50.466000 2576117 torch/distributed/run.py:766] ***************************************** +W0725 18:29:50.466000 2576117 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:29:50.466000 2576117 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:30:22.059183878 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:30:22.370379810 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:30:22.419618457 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:30:22.533473963 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:30:23.761756236 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:30:23.812594830 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:30:23.859148682 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:30:23.885333831 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed3_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed3_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:41:22 acc1_pos: 0.260 acc5_pos: 100.000 loss: 1.9428 (1.9428) acc1: 0.2604 (0.2604) acc5: 100.0000 (100.0000) time: 6.3813 data: 4.9276 max mem: 3531 +Test: [ 10/389] eta: 0:04:11 acc1_pos: 42.448 acc5_pos: 100.000 loss: 1.6316 (1.5074) acc1: 13.0208 (13.0919) acc5: 100.0000 (100.0000) time: 0.6647 data: 0.4483 max mem: 3531 +Test: [ 20/389] eta: 0:02:26 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.3491 (1.5045) acc1: 11.4583 (12.8720) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0004 max mem: 3531 +Test: [ 30/389] eta: 0:02:11 acc1_pos: 0.521 acc5_pos: 100.000 loss: 1.5242 (1.5049) acc1: 0.2604 (9.2070) acc5: 100.0000 (100.0000) time: 0.2004 data: 0.1013 max mem: 3531 +Test: [ 40/389] eta: 0:01:54 acc1_pos: 0.521 acc5_pos: 100.000 loss: 1.5382 (1.5302) acc1: 0.5208 (7.4123) acc5: 100.0000 (100.0000) time: 0.2575 data: 0.1623 max mem: 3531 +Test: [ 50/389] eta: 0:01:43 acc1_pos: 10.156 acc5_pos: 100.000 loss: 1.3916 (1.4900) acc1: 3.1250 (7.1385) acc5: 100.0000 (100.0000) time: 0.2095 data: 0.1169 max mem: 3531 +Test: [ 60/389] eta: 0:01:30 acc1_pos: 1.042 acc5_pos: 100.000 loss: 1.2867 (1.4868) acc1: 4.4271 (6.7324) acc5: 100.0000 (100.0000) time: 0.1623 data: 0.0672 max mem: 3531 +Test: [ 70/389] eta: 0:01:21 acc1_pos: 23.958 acc5_pos: 100.000 loss: 1.4470 (1.4832) acc1: 7.8125 (7.8528) acc5: 100.0000 (100.0000) time: 0.1275 data: 0.0316 max mem: 3531 +Test: [ 80/389] eta: 0:01:16 acc1_pos: 42.448 acc5_pos: 100.000 loss: 1.4217 (1.4586) acc1: 15.3646 (10.0502) acc5: 100.0000 (100.0000) time: 0.1657 data: 0.0705 max mem: 3531 +Test: [ 90/389] eta: 0:01:11 acc1_pos: 11.719 acc5_pos: 100.000 loss: 1.1557 (1.4166) acc1: 22.3958 (12.2596) acc5: 100.0000 (100.0000) time: 0.1922 data: 0.0966 max mem: 3531 +Test: [100/389] eta: 0:01:07 acc1_pos: 26.302 acc5_pos: 100.000 loss: 1.3214 (1.4280) acc1: 13.5417 (12.0230) acc5: 100.0000 (100.0000) time: 0.1846 data: 0.0916 max mem: 3531 +Test: [110/389] eta: 0:01:03 acc1_pos: 62.500 acc5_pos: 100.000 loss: 1.4570 (1.4159) acc1: 9.3750 (13.6332) acc5: 100.0000 (100.0000) time: 0.1717 data: 0.0804 max mem: 3531 +Test: [120/389] eta: 0:01:00 acc1_pos: 28.125 acc5_pos: 100.000 loss: 1.2034 (1.4019) acc1: 21.6146 (15.0762) acc5: 100.0000 (100.0000) time: 0.1862 data: 0.0917 max mem: 3531 +Test: [130/389] eta: 0:00:57 acc1_pos: 8.589 acc5_pos: 100.000 loss: 0.8190 (1.3383) acc1: 50.5208 (19.1277) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1914 data: 0.0977 max mem: 3531 +Test: [140/389] eta: 0:00:54 acc1_pos: 8.589 acc5_pos: 100.000 loss: 0.2018 (1.2573) acc1: 98.9583 (24.8338) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1647 data: 0.0737 max mem: 3531 +Test: [150/389] eta: 0:00:51 acc1_pos: 8.589 acc5_pos: 100.000 loss: 0.1860 (1.1928) acc1: 100.0000 (29.3374) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1741 data: 0.0831 max mem: 3531 +Test: [160/389] eta: 0:00:48 acc1_pos: 8.589 acc5_pos: 100.000 loss: 0.1922 (1.1306) acc1: 100.0000 (33.6487) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1855 data: 0.0948 max mem: 3531 +Test: [170/389] eta: 0:00:46 acc1_pos: 8.589 acc5_pos: 100.000 loss: 0.1705 (1.0751) acc1: 100.0000 (37.5137) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1828 data: 0.0919 max mem: 3531 +Test: [180/389] eta: 0:00:43 acc1_pos: 8.589 acc5_pos: 100.000 loss: 0.1707 (1.0255) acc1: 100.0000 (40.9343) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1785 data: 0.0866 max mem: 3531 +Test: [190/389] eta: 0:00:41 acc1_pos: 8.589 acc5_pos: 100.000 loss: 0.1707 (0.9837) acc1: 100.0000 (43.8373) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1776 data: 0.0852 max mem: 3531 +Test: [200/389] eta: 0:00:39 acc1_pos: 8.589 acc5_pos: 100.000 loss: 0.1766 (0.9444) acc1: 100.0000 (46.5744) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1870 data: 0.0951 max mem: 3531 +Test: [210/389] eta: 0:00:37 acc1_pos: 8.589 acc5_pos: 100.000 loss: 0.1742 (0.9070) acc1: 100.0000 (49.1064) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1950 data: 0.1027 max mem: 3532 +Test: [220/389] eta: 0:00:35 acc1_pos: 8.589 acc5_pos: 100.000 loss: 0.1273 (0.8720) acc1: 100.0000 (51.4093) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2019 data: 0.1083 max mem: 3532 +Test: [230/389] eta: 0:00:32 acc1_pos: 8.589 acc5_pos: 100.000 loss: 0.1503 (0.8443) acc1: 100.0000 (53.4677) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2017 data: 0.1086 max mem: 3532 +Test: [240/389] eta: 0:00:30 acc1_pos: 8.589 acc5_pos: 100.000 loss: 0.2247 (0.8189) acc1: 99.7396 (55.3348) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0926 max mem: 3532 +Test: [250/389] eta: 0:00:28 acc1_pos: 8.589 acc5_pos: 100.000 loss: 0.2078 (0.7928) acc1: 100.0000 (57.1143) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0811 max mem: 3532 +Test: [260/389] eta: 0:00:26 acc1_pos: 8.589 acc5_pos: 100.000 loss: 0.1412 (0.7706) acc1: 100.0000 (58.7454) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1753 data: 0.0823 max mem: 3532 +Test: [270/389] eta: 0:00:24 acc1_pos: 8.589 acc5_pos: 100.000 loss: 0.2461 (0.7514) acc1: 100.0000 (60.2053) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1709 data: 0.0797 max mem: 3532 +Test: [280/389] eta: 0:00:21 acc1_pos: 8.589 acc5_pos: 100.000 loss: 0.2074 (0.7327) acc1: 99.7396 (61.5195) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1758 data: 0.0846 max mem: 3532 +Test: [290/389] eta: 0:00:19 acc1_pos: 8.589 acc5_pos: 100.000 loss: 0.2038 (0.7149) acc1: 99.7396 (62.7363) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1648 data: 0.0732 max mem: 3532 +Test: [300/389] eta: 0:00:17 acc1_pos: 8.589 acc5_pos: 100.000 loss: 0.1820 (0.6975) acc1: 99.7396 (63.9682) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1553 data: 0.0620 max mem: 3532 +Test: [310/389] eta: 0:00:15 acc1_pos: 8.589 acc5_pos: 100.000 loss: 0.1968 (0.6826) acc1: 99.7396 (65.0497) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1610 data: 0.0658 max mem: 3532 +Test: [320/389] eta: 0:00:13 acc1_pos: 8.589 acc5_pos: 100.000 loss: 0.1935 (0.6665) acc1: 99.7396 (66.1223) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1784 data: 0.0854 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 8.589 acc5_pos: 100.000 loss: 0.2223 (0.6545) acc1: 99.7396 (67.1120) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1792 data: 0.0867 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 8.589 acc5_pos: 100.000 loss: 0.1921 (0.6395) acc1: 99.7396 (68.0627) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1454 data: 0.0528 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 8.589 acc5_pos: 100.000 loss: 0.1568 (0.6316) acc1: 100.0000 (68.6187) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.1293 data: 0.0380 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 8.589 acc5_pos: 100.000 loss: 0.2787 (0.6231) acc1: 95.3125 (69.3033) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1244 data: 0.0319 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 8.589 acc5_pos: 100.000 loss: 0.2700 (0.6128) acc1: 97.1354 (70.0352) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1126 data: 0.0212 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 8.589 acc5_pos: 100.000 loss: 0.1839 (0.6016) acc1: 100.0000 (70.8115) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1063 data: 0.0167 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 8.589 acc5_pos: 100.000 loss: 0.1830 (0.5934) acc1: 100.0000 (71.3835) acc5: 100.0000 (100.0000) acc1_neg: 95.082 acc5_neg: 100.000 time: 0.1107 data: 0.0174 max mem: 3532 +Test: Total time: 0:01:11 (0.1848 s / it) +* Overall Acc@1 71.384 Acc@5 100.000 + target == real Acc@1 15.021 Acc@5 100.000 + target == fake Acc@1 97.582 Acc@5 100.000 + loss 0.593 + PR-AUC 0.8013 ROC-AUC 0.6958 EER 35.39% (thr=0.8154) +Accuracy of the network on the 149358 test images: 71.4% +[rank0]:[W725 18:31:45.831657118 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:32:03.968000 2576685 torch/distributed/run.py:766] +W0725 18:32:03.968000 2576685 torch/distributed/run.py:766] ***************************************** +W0725 18:32:03.968000 2576685 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:32:03.968000 2576685 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:32:31.833412830 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:32:31.164433791 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:32:31.422429278 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:32:32.576877875 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:32:32.655665906 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:32:32.683328343 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:32:32.722333031 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank6]:[W725 18:32:32.730283232 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed3_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed3_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:38:06 acc1_pos: 4.167 acc5_pos: 100.000 loss: 1.7152 (1.7152) acc1: 4.1667 (4.1667) acc5: 100.0000 (100.0000) time: 6.8652 data: 5.2848 max mem: 3531 +Test: [ 10/333] eta: 0:03:51 acc1_pos: 15.365 acc5_pos: 100.000 loss: 1.5074 (1.4998) acc1: 13.0208 (12.9498) acc5: 100.0000 (100.0000) time: 0.7162 data: 0.4819 max mem: 3531 +Test: [ 20/333] eta: 0:02:12 acc1_pos: 20.833 acc5_pos: 100.000 loss: 1.4538 (1.4611) acc1: 15.6250 (15.5134) acc5: 100.0000 (100.0000) time: 0.1006 data: 0.0010 max mem: 3531 +Test: [ 30/333] eta: 0:01:57 acc1_pos: 56.250 acc5_pos: 100.000 loss: 1.1998 (1.2680) acc1: 25.0000 (25.9157) acc5: 100.0000 (100.0000) time: 0.2062 data: 0.1092 max mem: 3531 +Test: [ 40/333] eta: 0:01:41 acc1_pos: 41.927 acc5_pos: 100.000 loss: 0.8299 (1.1616) acc1: 47.6562 (31.0340) acc5: 100.0000 (100.0000) time: 0.2649 data: 0.1695 max mem: 3531 +Test: [ 50/333] eta: 0:01:29 acc1_pos: 46.615 acc5_pos: 100.000 loss: 0.8230 (1.0999) acc1: 45.8333 (34.0074) acc5: 100.0000 (100.0000) time: 0.2037 data: 0.1087 max mem: 3531 +Test: [ 60/333] eta: 0:01:21 acc1_pos: 41.406 acc5_pos: 100.000 loss: 0.8156 (1.0587) acc1: 46.6146 (35.8393) acc5: 100.0000 (100.0000) time: 0.2049 data: 0.1094 max mem: 3531 +Test: [ 70/333] eta: 0:01:16 acc1_pos: 38.281 acc5_pos: 100.000 loss: 0.8000 (1.0275) acc1: 47.6562 (37.7091) acc5: 100.0000 (100.0000) time: 0.2302 data: 0.1340 max mem: 3531 +Test: [ 80/333] eta: 0:01:11 acc1_pos: 46.094 acc5_pos: 100.000 loss: 0.8422 (1.0055) acc1: 46.6146 (38.8985) acc5: 100.0000 (100.0000) time: 0.2309 data: 0.1365 max mem: 3531 +Test: [ 90/333] eta: 0:01:05 acc1_pos: 62.760 acc5_pos: 100.000 loss: 0.8148 (0.9731) acc1: 50.7812 (41.1344) acc5: 100.0000 (100.0000) time: 0.1952 data: 0.1013 max mem: 3531 +Test: [100/333] eta: 0:01:01 acc1_pos: 74.479 acc5_pos: 100.000 loss: 0.6148 (0.9380) acc1: 65.1042 (43.6185) acc5: 100.0000 (100.0000) time: 0.1952 data: 0.1026 max mem: 3531 +Test: [110/333] eta: 0:00:56 acc1_pos: 70.052 acc5_pos: 100.000 loss: 0.6017 (0.9088) acc1: 66.6667 (45.7700) acc5: 100.0000 (100.0000) time: 0.1806 data: 0.0883 max mem: 3531 +Test: [120/333] eta: 0:00:52 acc1_pos: 65.625 acc5_pos: 100.000 loss: 0.5818 (0.8833) acc1: 70.0521 (47.6735) acc5: 100.0000 (100.0000) time: 0.1412 data: 0.0489 max mem: 3531 +Test: [130/333] eta: 0:00:49 acc1_pos: 75.260 acc5_pos: 100.000 loss: 0.5689 (0.8603) acc1: 70.0521 (49.3579) acc5: 100.0000 (100.0000) time: 0.1737 data: 0.0821 max mem: 3531 +Test: [140/333] eta: 0:00:45 acc1_pos: 74.219 acc5_pos: 100.000 loss: 0.5892 (0.8414) acc1: 68.7500 (50.8588) acc5: 100.0000 (100.0000) time: 0.1984 data: 0.1025 max mem: 3531 +Test: [150/333] eta: 0:00:42 acc1_pos: 72.135 acc5_pos: 100.000 loss: 0.5857 (0.8260) acc1: 69.2708 (52.0075) acc5: 100.0000 (100.0000) time: 0.1902 data: 0.0936 max mem: 3531 +Test: [160/333] eta: 0:00:39 acc1_pos: 52.632 acc5_pos: 100.000 loss: 0.5857 (0.8154) acc1: 66.1458 (52.7077) acc5: 100.0000 (100.0000) acc1_neg: 50.521 acc5_neg: 100.000 time: 0.1776 data: 0.0832 max mem: 3531 +Test: [170/333] eta: 0:00:37 acc1_pos: 52.632 acc5_pos: 100.000 loss: 0.6681 (0.8082) acc1: 58.5938 (53.0382) acc5: 100.0000 (100.0000) acc1_neg: 65.365 acc5_neg: 100.000 time: 0.1753 data: 0.0810 max mem: 3531 +Test: [180/333] eta: 0:00:34 acc1_pos: 52.632 acc5_pos: 100.000 loss: 0.6335 (0.7975) acc1: 62.7604 (53.6976) acc5: 100.0000 (100.0000) acc1_neg: 66.927 acc5_neg: 100.000 time: 0.1867 data: 0.0917 max mem: 3531 +Test: [190/333] eta: 0:00:32 acc1_pos: 52.632 acc5_pos: 100.000 loss: 0.6062 (0.7834) acc1: 65.6250 (54.5852) acc5: 100.0000 (100.0000) acc1_neg: 65.625 acc5_neg: 100.000 time: 0.1894 data: 0.0953 max mem: 3531 +Test: [200/333] eta: 0:00:29 acc1_pos: 52.632 acc5_pos: 100.000 loss: 0.5332 (0.7734) acc1: 66.9271 (55.2718) acc5: 100.0000 (100.0000) acc1_neg: 64.062 acc5_neg: 100.000 time: 0.1847 data: 0.0923 max mem: 3531 +Test: [210/333] eta: 0:00:27 acc1_pos: 52.632 acc5_pos: 100.000 loss: 0.6219 (0.7669) acc1: 64.8438 (55.7218) acc5: 100.0000 (100.0000) acc1_neg: 62.240 acc5_neg: 100.000 time: 0.1804 data: 0.0871 max mem: 3532 +Test: [220/333] eta: 0:00:24 acc1_pos: 52.632 acc5_pos: 100.000 loss: 0.6308 (0.7640) acc1: 64.0625 (55.8446) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.1795 data: 0.0883 max mem: 3532 +Test: [230/333] eta: 0:00:22 acc1_pos: 52.632 acc5_pos: 100.000 loss: 0.5268 (0.7533) acc1: 71.6146 (56.7494) acc5: 100.0000 (100.0000) acc1_neg: 77.344 acc5_neg: 100.000 time: 0.1769 data: 0.0839 max mem: 3532 +Test: [240/333] eta: 0:00:19 acc1_pos: 52.632 acc5_pos: 100.000 loss: 0.5267 (0.7465) acc1: 71.8750 (57.1642) acc5: 100.0000 (100.0000) acc1_neg: 65.104 acc5_neg: 100.000 time: 0.1774 data: 0.0841 max mem: 3532 +Test: [250/333] eta: 0:00:17 acc1_pos: 52.632 acc5_pos: 100.000 loss: 0.6112 (0.7411) acc1: 65.1042 (57.4473) acc5: 100.0000 (100.0000) acc1_neg: 61.719 acc5_neg: 100.000 time: 0.1922 data: 0.1006 max mem: 3532 +Test: [260/333] eta: 0:00:15 acc1_pos: 52.632 acc5_pos: 100.000 loss: 0.6037 (0.7350) acc1: 65.1042 (57.8624) acc5: 100.0000 (100.0000) acc1_neg: 66.927 acc5_neg: 100.000 time: 0.1974 data: 0.1039 max mem: 3532 +Test: [270/333] eta: 0:00:13 acc1_pos: 52.632 acc5_pos: 100.000 loss: 0.5655 (0.7287) acc1: 69.7917 (58.3948) acc5: 100.0000 (100.0000) acc1_neg: 72.656 acc5_neg: 100.000 time: 0.1807 data: 0.0857 max mem: 3532 +Test: [280/333] eta: 0:00:11 acc1_pos: 52.632 acc5_pos: 100.000 loss: 0.5737 (0.7276) acc1: 69.5312 (58.4065) acc5: 100.0000 (100.0000) acc1_neg: 60.938 acc5_neg: 100.000 time: 0.1860 data: 0.0898 max mem: 3532 +Test: [290/333] eta: 0:00:09 acc1_pos: 52.632 acc5_pos: 100.000 loss: 0.7733 (0.7294) acc1: 51.8229 (58.1615) acc5: 100.0000 (100.0000) acc1_neg: 55.469 acc5_neg: 100.000 time: 0.1838 data: 0.0889 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 52.632 acc5_pos: 100.000 loss: 0.7149 (0.7282) acc1: 55.4688 (58.1871) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.1429 data: 0.0517 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 52.632 acc5_pos: 100.000 loss: 0.6525 (0.7221) acc1: 62.5000 (58.6624) acc5: 100.0000 (100.0000) acc1_neg: 71.354 acc5_neg: 100.000 time: 0.1250 data: 0.0354 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 52.632 acc5_pos: 100.000 loss: 0.5296 (0.7172) acc1: 71.6146 (59.0310) acc5: 100.0000 (100.0000) acc1_neg: 53.385 acc5_neg: 100.000 time: 0.1290 data: 0.0389 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 52.632 acc5_pos: 100.000 loss: 0.7348 (0.7187) acc1: 55.7292 (58.8557) acc5: 100.0000 (100.0000) acc1_neg: 53.646 acc5_neg: 100.000 time: 0.1174 data: 0.0276 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 52.632 acc5_pos: 100.000 loss: 0.7366 (0.7188) acc1: 55.4688 (58.8362) acc5: 100.0000 (100.0000) acc1_neg: 55.493 acc5_neg: 100.000 time: 0.1189 data: 0.0264 max mem: 3532 +Test: Total time: 0:01:06 (0.1990 s / it) +* Overall Acc@1 58.836 Acc@5 100.000 + target == real Acc@1 52.745 Acc@5 100.000 + target == fake Acc@1 64.258 Acc@5 100.000 + loss 0.719 + PR-AUC 0.6013 ROC-AUC 0.6083 EER 41.79% (thr=0.5425) +Accuracy of the network on the 127843 test images: 58.8% +[rank0]:[W725 18:33:46.448067033 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:34:05.914000 2577261 torch/distributed/run.py:766] +W0725 18:34:05.914000 2577261 torch/distributed/run.py:766] ***************************************** +W0725 18:34:05.914000 2577261 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:34:05.914000 2577261 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:34:33.759075403 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:34:33.998979897 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:34:33.052022677 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:34:33.445069423 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:34:33.492111559 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:34:33.505355745 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:34:34.613263597 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:34:34.639971590 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed3_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed3_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:25:22 acc1_pos: 97.396 acc5_pos: 100.000 loss: 0.1798 (0.1798) acc1: 97.3958 (97.3958) acc5: 100.0000 (100.0000) time: 6.4804 data: 4.9257 max mem: 3531 +Test: [ 10/235] eta: 0:02:32 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.1869 (0.1887) acc1: 96.6146 (96.4489) acc5: 100.0000 (100.0000) time: 0.6776 data: 0.4481 max mem: 3531 +Test: [ 20/235] eta: 0:01:26 acc1_pos: 96.615 acc5_pos: 100.000 loss: 0.1844 (0.1843) acc1: 96.3542 (96.6146) acc5: 100.0000 (100.0000) time: 0.0986 data: 0.0004 max mem: 3531 +Test: [ 30/235] eta: 0:01:04 acc1_pos: 96.094 acc5_pos: 100.000 loss: 0.1805 (0.1835) acc1: 96.6146 (96.5306) acc5: 100.0000 (100.0000) time: 0.1152 data: 0.0196 max mem: 3531 +Test: [ 40/235] eta: 0:00:53 acc1_pos: 97.396 acc5_pos: 100.000 loss: 0.1810 (0.1826) acc1: 96.8750 (96.6400) acc5: 100.0000 (100.0000) time: 0.1450 data: 0.0540 max mem: 3531 +Test: [ 50/235] eta: 0:00:46 acc1_pos: 96.094 acc5_pos: 100.000 loss: 0.1810 (0.1831) acc1: 96.8750 (96.6963) acc5: 100.0000 (100.0000) time: 0.1591 data: 0.0671 max mem: 3531 +Test: [ 60/235] eta: 0:00:41 acc1_pos: 97.135 acc5_pos: 100.000 loss: 0.1899 (0.1844) acc1: 96.0938 (96.5591) acc5: 100.0000 (100.0000) time: 0.1498 data: 0.0572 max mem: 3531 +Test: [ 70/235] eta: 0:00:38 acc1_pos: 98.462 acc5_pos: 100.000 loss: 0.1998 (0.3535) acc1: 95.0521 (86.6674) acc5: 100.0000 (100.0000) acc1_neg: 17.448 acc5_neg: 100.000 time: 0.1757 data: 0.0834 max mem: 3531 +Test: [ 80/235] eta: 0:00:35 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.5433 (0.5026) acc1: 16.4062 (77.9675) acc5: 100.0000 (100.0000) acc1_neg: 16.667 acc5_neg: 100.000 time: 0.2049 data: 0.1106 max mem: 3531 +Test: [ 90/235] eta: 0:00:31 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.5576 (0.6190) acc1: 16.4062 (71.2483) acc5: 100.0000 (100.0000) acc1_neg: 16.406 acc5_neg: 100.000 time: 0.1696 data: 0.0736 max mem: 3531 +Test: [100/235] eta: 0:00:28 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.5576 (0.7105) acc1: 16.9271 (65.9679) acc5: 100.0000 (100.0000) acc1_neg: 15.625 acc5_neg: 100.000 time: 0.1389 data: 0.0427 max mem: 3531 +Test: [110/235] eta: 0:00:25 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.5402 (0.7850) acc1: 17.1875 (61.6132) acc5: 100.0000 (100.0000) acc1_neg: 20.833 acc5_neg: 100.000 time: 0.1549 data: 0.0581 max mem: 3531 +Test: [120/235] eta: 0:00:23 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.6065 (0.8691) acc1: 13.2812 (57.2013) acc5: 100.0000 (100.0000) acc1_neg: 7.292 acc5_neg: 100.000 time: 0.1909 data: 0.0943 max mem: 3531 +Test: [130/235] eta: 0:00:21 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.8194 (0.9412) acc1: 7.5521 (53.4113) acc5: 100.0000 (100.0000) acc1_neg: 6.771 acc5_neg: 100.000 time: 0.1999 data: 0.1043 max mem: 3531 +Test: [140/235] eta: 0:00:19 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.8396 (1.0061) acc1: 7.2917 (50.1385) acc5: 100.0000 (100.0000) acc1_neg: 6.250 acc5_neg: 100.000 time: 0.1868 data: 0.0925 max mem: 3531 +Test: [150/235] eta: 0:00:17 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.8275 (1.0597) acc1: 7.2917 (47.2975) acc5: 100.0000 (100.0000) acc1_neg: 8.073 acc5_neg: 100.000 time: 0.1949 data: 0.1000 max mem: 3531 +Test: [160/235] eta: 0:00:15 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.8217 (1.1071) acc1: 7.5521 (44.8450) acc5: 100.0000 (100.0000) acc1_neg: 8.594 acc5_neg: 100.000 time: 0.2040 data: 0.1087 max mem: 3531 +Test: [170/235] eta: 0:00:13 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.7871 (1.1184) acc1: 8.5938 (43.7759) acc5: 100.0000 (100.0000) acc1_neg: 26.823 acc5_neg: 100.000 time: 0.2025 data: 0.1089 max mem: 3531 +Test: [180/235] eta: 0:00:11 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.2482 (1.1238) acc1: 28.9062 (43.1126) acc5: 100.0000 (100.0000) acc1_neg: 31.510 acc5_neg: 100.000 time: 0.1990 data: 0.1054 max mem: 3531 +Test: [190/235] eta: 0:00:09 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.2075 (1.1292) acc1: 31.5104 (42.4697) acc5: 100.0000 (100.0000) acc1_neg: 32.031 acc5_neg: 100.000 time: 0.1698 data: 0.0743 max mem: 3531 +Test: [200/235] eta: 0:00:06 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.2206 (1.1347) acc1: 29.9479 (41.8299) acc5: 100.0000 (100.0000) acc1_neg: 27.083 acc5_neg: 100.000 time: 0.1460 data: 0.0519 max mem: 3531 +Test: [210/235] eta: 0:00:04 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.2517 (1.1394) acc1: 29.1667 (41.2544) acc5: 100.0000 (100.0000) acc1_neg: 27.865 acc5_neg: 100.000 time: 0.1701 data: 0.0785 max mem: 3532 +Test: [220/235] eta: 0:00:02 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.2429 (1.1437) acc1: 28.9062 (40.7181) acc5: 100.0000 (100.0000) acc1_neg: 28.906 acc5_neg: 100.000 time: 0.2015 data: 0.1103 max mem: 3532 +Test: [230/235] eta: 0:00:00 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.2259 (1.1471) acc1: 30.2083 (40.2834) acc5: 100.0000 (100.0000) acc1_neg: 27.604 acc5_neg: 100.000 time: 0.1737 data: 0.0838 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.2259 (1.1494) acc1: 30.4688 (40.1451) acc5: 100.0000 (100.0000) acc1_neg: 34.783 acc5_neg: 100.000 time: 0.1487 data: 0.0620 max mem: 3532 +Test: Total time: 0:00:45 (0.1941 s / it) +* Overall Acc@1 40.145 Acc@5 100.000 + target == real Acc@1 96.582 Acc@5 100.000 + target == fake Acc@1 19.733 Acc@5 100.000 + loss 1.149 + PR-AUC 0.8765 ROC-AUC 0.7283 EER 33.62% (thr=0.1465) +Accuracy of the network on the 89879 test images: 40.1% +[rank0]:[W725 18:35:31.814606415 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:35:49.511000 2577814 torch/distributed/run.py:766] +W0725 18:35:49.511000 2577814 torch/distributed/run.py:766] ***************************************** +W0725 18:35:49.511000 2577814 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:35:49.511000 2577814 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:36:17.328251115 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:36:18.647966073 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:36:18.817228948 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:36:18.077200335 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:36:18.260319656 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +[rank1]:[W725 18:36:18.291892767 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:36:18.299971154 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:36:18.550131001 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed3_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed3_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:29:18 acc1_pos: 58.594 acc5_pos: 100.000 loss: 0.6862 (0.6862) acc1: 58.5938 (58.5938) acc5: 100.0000 (100.0000) time: 6.5372 data: 4.9152 max mem: 3531 +Test: [ 10/269] eta: 0:02:55 acc1_pos: 32.552 acc5_pos: 100.000 loss: 0.7477 (0.7771) acc1: 48.6979 (46.9460) acc5: 100.0000 (100.0000) time: 0.6780 data: 0.4490 max mem: 3531 +Test: [ 20/269] eta: 0:01:42 acc1_pos: 41.927 acc5_pos: 100.000 loss: 0.7383 (0.7347) acc1: 48.6979 (51.9097) acc5: 100.0000 (100.0000) time: 0.1055 data: 0.0145 max mem: 3531 +Test: [ 30/269] eta: 0:01:23 acc1_pos: 72.135 acc5_pos: 100.000 loss: 0.7128 (0.7205) acc1: 53.1250 (53.8474) acc5: 100.0000 (100.0000) time: 0.1694 data: 0.0778 max mem: 3531 +Test: [ 40/269] eta: 0:01:12 acc1_pos: 31.897 acc5_pos: 100.000 loss: 0.6566 (0.6951) acc1: 59.1146 (56.8915) acc5: 100.0000 (100.0000) acc1_neg: 59.115 acc5_neg: 100.000 time: 0.2207 data: 0.1290 max mem: 3531 +Test: [ 50/269] eta: 0:01:04 acc1_pos: 31.897 acc5_pos: 100.000 loss: 0.6145 (0.6807) acc1: 64.3229 (58.7674) acc5: 100.0000 (100.0000) acc1_neg: 46.875 acc5_neg: 100.000 time: 0.2099 data: 0.1148 max mem: 3531 +Test: [ 60/269] eta: 0:00:58 acc1_pos: 31.897 acc5_pos: 100.000 loss: 0.6695 (0.6898) acc1: 57.0312 (57.8381) acc5: 100.0000 (100.0000) acc1_neg: 37.760 acc5_neg: 100.000 time: 0.1991 data: 0.1019 max mem: 3531 +Test: [ 70/269] eta: 0:00:53 acc1_pos: 31.897 acc5_pos: 100.000 loss: 0.6789 (0.6894) acc1: 52.3438 (57.5191) acc5: 100.0000 (100.0000) acc1_neg: 55.469 acc5_neg: 100.000 time: 0.2049 data: 0.1087 max mem: 3531 +Test: [ 80/269] eta: 0:00:49 acc1_pos: 31.897 acc5_pos: 100.000 loss: 0.6482 (0.6780) acc1: 55.9896 (58.6613) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2011 data: 0.1074 max mem: 3531 +Test: [ 90/269] eta: 0:00:44 acc1_pos: 31.897 acc5_pos: 100.000 loss: 0.6130 (0.6669) acc1: 64.3229 (60.2278) acc5: 100.0000 (100.0000) acc1_neg: 61.458 acc5_neg: 100.000 time: 0.1749 data: 0.0832 max mem: 3531 +Test: [100/269] eta: 0:00:41 acc1_pos: 31.897 acc5_pos: 100.000 loss: 0.6326 (0.6665) acc1: 64.3229 (60.2903) acc5: 100.0000 (100.0000) acc1_neg: 58.854 acc5_neg: 100.000 time: 0.1991 data: 0.1072 max mem: 3531 +Test: [110/269] eta: 0:00:38 acc1_pos: 31.897 acc5_pos: 100.000 loss: 0.7006 (0.6745) acc1: 54.1667 (59.4360) acc5: 100.0000 (100.0000) acc1_neg: 53.906 acc5_neg: 100.000 time: 0.2121 data: 0.1203 max mem: 3531 +Test: [120/269] eta: 0:00:35 acc1_pos: 31.897 acc5_pos: 100.000 loss: 0.6700 (0.6682) acc1: 54.1667 (60.0594) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.1895 data: 0.0982 max mem: 3531 +Test: [130/269] eta: 0:00:32 acc1_pos: 31.897 acc5_pos: 100.000 loss: 0.5986 (0.6631) acc1: 64.5833 (60.6751) acc5: 100.0000 (100.0000) acc1_neg: 59.896 acc5_neg: 100.000 time: 0.1932 data: 0.0995 max mem: 3531 +Test: [140/269] eta: 0:00:29 acc1_pos: 31.897 acc5_pos: 100.000 loss: 0.6720 (0.6693) acc1: 58.0729 (59.9697) acc5: 100.0000 (100.0000) acc1_neg: 51.302 acc5_neg: 100.000 time: 0.1658 data: 0.0707 max mem: 3531 +Test: [150/269] eta: 0:00:27 acc1_pos: 31.897 acc5_pos: 100.000 loss: 0.7299 (0.6736) acc1: 51.3021 (59.7010) acc5: 100.0000 (100.0000) acc1_neg: 62.240 acc5_neg: 100.000 time: 0.1818 data: 0.0883 max mem: 3531 +Test: [160/269] eta: 0:00:24 acc1_pos: 31.897 acc5_pos: 100.000 loss: 0.6406 (0.6698) acc1: 64.0625 (60.1514) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.2003 data: 0.1085 max mem: 3531 +Test: [170/269] eta: 0:00:22 acc1_pos: 31.897 acc5_pos: 100.000 loss: 0.5677 (0.6632) acc1: 70.5729 (60.7928) acc5: 100.0000 (100.0000) acc1_neg: 50.781 acc5_neg: 100.000 time: 0.1823 data: 0.0899 max mem: 3531 +Test: [180/269] eta: 0:00:19 acc1_pos: 31.897 acc5_pos: 100.000 loss: 0.5342 (0.6575) acc1: 74.7396 (61.4281) acc5: 100.0000 (100.0000) acc1_neg: 48.177 acc5_neg: 100.000 time: 0.1901 data: 0.0977 max mem: 3531 +Test: [190/269] eta: 0:00:17 acc1_pos: 31.897 acc5_pos: 100.000 loss: 0.5772 (0.6561) acc1: 72.1354 (61.6356) acc5: 100.0000 (100.0000) acc1_neg: 72.656 acc5_neg: 100.000 time: 0.1699 data: 0.0784 max mem: 3531 +Test: [200/269] eta: 0:00:14 acc1_pos: 31.897 acc5_pos: 100.000 loss: 0.6147 (0.6533) acc1: 64.3229 (61.9584) acc5: 100.0000 (100.0000) acc1_neg: 75.781 acc5_neg: 100.000 time: 0.1594 data: 0.0663 max mem: 3531 +Test: [210/269] eta: 0:00:12 acc1_pos: 31.897 acc5_pos: 100.000 loss: 0.6481 (0.6569) acc1: 60.4167 (61.7632) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.1816 data: 0.0899 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 31.897 acc5_pos: 100.000 loss: 0.7115 (0.6585) acc1: 56.7708 (61.7812) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.1789 data: 0.0872 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 31.897 acc5_pos: 100.000 loss: 0.5716 (0.6528) acc1: 67.9688 (62.3580) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.1532 data: 0.0597 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 31.897 acc5_pos: 100.000 loss: 0.5783 (0.6511) acc1: 67.7083 (62.5389) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.1574 data: 0.0657 max mem: 3532 +Test: [250/269] eta: 0:00:03 acc1_pos: 31.897 acc5_pos: 100.000 loss: 0.6451 (0.6549) acc1: 61.9792 (62.2085) acc5: 100.0000 (100.0000) acc1_neg: 34.375 acc5_neg: 100.000 time: 0.1589 data: 0.0690 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 31.897 acc5_pos: 100.000 loss: 0.7451 (0.6607) acc1: 52.3438 (61.4693) acc5: 100.0000 (100.0000) acc1_neg: 52.344 acc5_neg: 100.000 time: 0.1265 data: 0.0366 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 31.897 acc5_pos: 100.000 loss: 0.7683 (0.6620) acc1: 48.9583 (61.3110) acc5: 100.0000 (100.0000) acc1_neg: 71.654 acc5_neg: 100.000 time: 0.1069 data: 0.0134 max mem: 3532 +Test: Total time: 0:00:53 (0.1989 s / it) +* Overall Acc@1 61.311 Acc@5 100.000 + target == real Acc@1 53.652 Acc@5 100.000 + target == fake Acc@1 62.356 Acc@5 100.000 + loss 0.662 + PR-AUC 0.9150 ROC-AUC 0.6159 EER 41.80% (thr=0.5229) +Accuracy of the network on the 103293 test images: 61.3% +[rank0]:[W725 18:37:21.776124769 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-25T18:37:26Z] eval_done tag=fixed500k_real_n16_seed3_resampled +[2026-07-25T18:37:27Z] eval_start tag=fixed500k_fake_n32_seed3_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed3_resampled_n111-046-081_20260725_112559/best_checkpoint.pth +started_utc=2026-07-25T18:37:27Z +hostname=n111-046-081 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed3_resampled_n111-046-081_20260725_112559/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed3_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0725 18:37:39.250000 2578400 torch/distributed/run.py:766] +W0725 18:37:39.250000 2578400 torch/distributed/run.py:766] ***************************************** +W0725 18:37:39.250000 2578400 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:37:39.250000 2578400 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:38:07.547018643 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:38:08.664627272 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:38:08.794607720 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:38:08.857880303 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 6): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:38:08.085083280 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:38:08.088798130 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +[rank1]:[W725 18:38:08.111105857 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:38:08.122123459 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed3_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed3_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:17:50 acc1_pos: 49.219 acc5_pos: 100.000 loss: 0.7417 (0.7417) acc1: 49.2188 (49.2188) acc5: 100.0000 (100.0000) time: 4.5727 data: 2.7105 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/234] eta: 0:01:51 acc1_pos: 47.917 acc5_pos: 100.000 loss: 0.7417 (0.7151) acc1: 49.2188 (55.0189) acc5: 100.0000 (100.0000) time: 0.4976 data: 0.2469 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 20/234] eta: 0:01:07 acc1_pos: 92.448 acc5_pos: 100.000 loss: 0.6918 (0.7030) acc1: 50.5208 (55.5556) acc5: 100.0000 (100.0000) time: 0.1028 data: 0.0132 max mem: 3531 +Test: [ 30/234] eta: 0:00:56 acc1_pos: 79.948 acc5_pos: 100.000 loss: 0.2544 (0.5594) acc1: 92.4479 (67.3471) acc5: 100.0000 (100.0000) time: 0.1524 data: 0.0609 max mem: 3531 +Test: [ 40/234] eta: 0:00:52 acc1_pos: 99.219 acc5_pos: 100.000 loss: 0.2439 (0.5054) acc1: 92.7083 (71.9830) acc5: 100.0000 (100.0000) time: 0.2233 data: 0.1279 max mem: 3531 +Test: [ 50/234] eta: 0:00:46 acc1_pos: 97.917 acc5_pos: 100.000 loss: 0.1497 (0.4345) acc1: 97.9167 (77.1497) acc5: 100.0000 (100.0000) time: 0.2207 data: 0.1247 max mem: 3531 +Test: [ 60/234] eta: 0:00:40 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.1438 (0.3870) acc1: 98.9583 (80.8530) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1599 data: 0.0663 max mem: 3531 +Test: [ 70/234] eta: 0:00:35 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.1468 (0.3846) acc1: 99.7396 (81.1326) acc5: 100.0000 (100.0000) acc1_neg: 73.698 acc5_neg: 100.000 time: 0.1178 data: 0.0217 max mem: 3531 +Test: [ 80/234] eta: 0:00:31 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4481 (0.3830) acc1: 76.3021 (81.4172) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1253 data: 0.0271 max mem: 3531 +Test: [ 90/234] eta: 0:00:29 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.2410 (0.3605) acc1: 93.2292 (82.9241) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.1699 data: 0.0747 max mem: 3531 +Test: [100/234] eta: 0:00:27 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.2439 (0.3661) acc1: 89.8438 (82.5959) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1945 data: 0.1002 max mem: 3531 +Test: [110/234] eta: 0:00:24 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.3634 (0.3642) acc1: 83.0729 (82.8148) acc5: 100.0000 (100.0000) acc1_neg: 78.646 acc5_neg: 100.000 time: 0.1810 data: 0.0847 max mem: 3531 +Test: [120/234] eta: 0:00:22 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.3613 (0.3770) acc1: 83.8542 (82.1539) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1371 data: 0.0403 max mem: 3531 +Test: [130/234] eta: 0:00:19 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.3891 (0.3793) acc1: 81.7708 (82.3553) acc5: 100.0000 (100.0000) acc1_neg: 79.427 acc5_neg: 100.000 time: 0.1321 data: 0.0368 max mem: 3531 +Test: [140/234] eta: 0:00:18 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4474 (0.4138) acc1: 79.4271 (80.3783) acc5: 100.0000 (100.0000) acc1_neg: 23.438 acc5_neg: 100.000 time: 0.1826 data: 0.0862 max mem: 3531 +Test: [150/234] eta: 0:00:16 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4563 (0.4216) acc1: 73.4375 (79.8462) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.2070 data: 0.1121 max mem: 3531 +Test: [160/234] eta: 0:00:14 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4182 (0.4230) acc1: 76.3021 (79.8897) acc5: 100.0000 (100.0000) acc1_neg: 71.615 acc5_neg: 100.000 time: 0.2149 data: 0.1241 max mem: 3531 +Test: [170/234] eta: 0:00:12 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.5422 (0.4293) acc1: 72.9167 (79.6144) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.2097 data: 0.1191 max mem: 3531 +Test: [180/234] eta: 0:00:10 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.5232 (0.4361) acc1: 73.9583 (79.1552) acc5: 100.0000 (100.0000) acc1_neg: 40.625 acc5_neg: 100.000 time: 0.1978 data: 0.1049 max mem: 3531 +Test: [190/234] eta: 0:00:08 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.7661 (0.4548) acc1: 56.7708 (77.9096) acc5: 100.0000 (100.0000) acc1_neg: 56.771 acc5_neg: 100.000 time: 0.2047 data: 0.1098 max mem: 3531 +Test: [200/234] eta: 0:00:06 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.7661 (0.4507) acc1: 56.7708 (78.1859) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2070 data: 0.1131 max mem: 3531 +Test: [210/234] eta: 0:00:04 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.1294 (0.4383) acc1: 99.7396 (79.0704) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.2106 data: 0.1187 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.2908 (0.4324) acc1: 92.7083 (79.6427) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.1964 data: 0.1059 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.3458 (0.4334) acc1: 80.4688 (79.6018) acc5: 100.0000 (100.0000) acc1_neg: 79.427 acc5_neg: 100.000 time: 0.1622 data: 0.0719 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4475 (0.4337) acc1: 79.4271 (79.5908) acc5: 100.0000 (100.0000) acc1_neg: 80.156 acc5_neg: 100.000 time: 0.1528 data: 0.0601 max mem: 3532 +Test: Total time: 0:00:45 (0.1943 s / it) +* Overall Acc@1 79.591 Acc@5 100.000 + target == real Acc@1 77.302 Acc@5 100.000 + target == fake Acc@1 80.236 Acc@5 100.000 + loss 0.434 + PR-AUC 0.9644 ROC-AUC 0.8846 EER 20.98% (thr=0.5210) +Accuracy of the network on the 89729 test images: 79.6% +[rank0]:[W725 18:39:05.795154922 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:39:23.296000 2578982 torch/distributed/run.py:766] +W0725 18:39:23.296000 2578982 torch/distributed/run.py:766] ***************************************** +W0725 18:39:23.296000 2578982 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:39:23.296000 2578982 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:39:49.889628702 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:39:49.151199332 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:39:49.514730851 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:39:49.559275644 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:39:50.785224854 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:39:50.837518903 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +[rank0]:[W725 18:39:50.854217845 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:39:50.859871818 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed3_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed3_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 5:02:34 acc1_pos: 22.917 acc5_pos: 100.000 loss: 1.1469 (1.1469) acc1: 22.9167 (22.9167) acc5: 100.0000 (100.0000) time: 7.6056 data: 5.7869 max mem: 3531 +Test: [ 10/2387] eta: 0:30:48 acc1_pos: 10.677 acc5_pos: 100.000 loss: 1.1989 (1.2050) acc1: 15.6250 (19.3892) acc5: 100.0000 (100.0000) time: 0.7779 data: 0.5265 max mem: 3531 +Test: [ 20/2387] eta: 0:17:57 acc1_pos: 9.896 acc5_pos: 100.000 loss: 1.1989 (1.2156) acc1: 15.6250 (18.7624) acc5: 100.0000 (100.0000) time: 0.0975 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:17:02 acc1_pos: 8.854 acc5_pos: 100.000 loss: 1.2612 (1.2387) acc1: 15.1042 (17.8595) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1453 max mem: 3531 +Test: [ 40/2387] eta: 0:15:30 acc1_pos: 10.938 acc5_pos: 100.000 loss: 1.2693 (1.2492) acc1: 13.8021 (16.6540) acc5: 100.0000 (100.0000) time: 0.3349 data: 0.2341 max mem: 3531 +Test: [ 50/2387] eta: 0:14:51 acc1_pos: 38.281 acc5_pos: 100.000 loss: 1.1524 (1.2146) acc1: 15.1042 (18.7806) acc5: 100.0000 (100.0000) time: 0.2999 data: 0.2018 max mem: 3531 +Test: [ 60/2387] eta: 0:14:39 acc1_pos: 45.833 acc5_pos: 100.000 loss: 1.0669 (1.2043) acc1: 23.4375 (19.5014) acc5: 100.0000 (100.0000) time: 0.3399 data: 0.2459 max mem: 3531 +Test: [ 70/2387] eta: 0:13:34 acc1_pos: 34.375 acc5_pos: 100.000 loss: 1.1960 (1.2061) acc1: 21.0938 (19.4286) acc5: 100.0000 (100.0000) time: 0.2749 data: 0.1795 max mem: 3531 +Test: [ 80/2387] eta: 0:13:04 acc1_pos: 28.906 acc5_pos: 100.000 loss: 1.1750 (1.1963) acc1: 16.6667 (19.8753) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1286 max mem: 3531 +Test: [ 90/2387] eta: 0:12:46 acc1_pos: 36.979 acc5_pos: 100.000 loss: 1.0541 (1.1810) acc1: 28.1250 (20.7503) acc5: 100.0000 (100.0000) time: 0.2699 data: 0.1768 max mem: 3531 +Test: [ 100/2387] eta: 0:12:35 acc1_pos: 22.135 acc5_pos: 100.000 loss: 1.0136 (1.1760) acc1: 25.7812 (21.1582) acc5: 100.0000 (100.0000) time: 0.2899 data: 0.1981 max mem: 3531 +Test: [ 110/2387] eta: 0:12:15 acc1_pos: 28.125 acc5_pos: 100.000 loss: 1.0516 (1.1642) acc1: 25.2604 (21.5888) acc5: 100.0000 (100.0000) time: 0.2749 data: 0.1805 max mem: 3531 +Test: [ 120/2387] eta: 0:11:58 acc1_pos: 22.135 acc5_pos: 100.000 loss: 1.0908 (1.1600) acc1: 25.2604 (21.8406) acc5: 100.0000 (100.0000) time: 0.2499 data: 0.1565 max mem: 3531 +Test: [ 130/2387] eta: 0:11:36 acc1_pos: 27.865 acc5_pos: 100.000 loss: 1.0924 (1.1570) acc1: 23.9583 (21.9128) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1371 max mem: 3531 +Test: [ 140/2387] eta: 0:11:19 acc1_pos: 22.396 acc5_pos: 100.000 loss: 1.0922 (1.1535) acc1: 23.1771 (22.0523) acc5: 100.0000 (100.0000) time: 0.2149 data: 0.1202 max mem: 3531 +Test: [ 150/2387] eta: 0:11:10 acc1_pos: 8.594 acc5_pos: 100.000 loss: 1.0737 (1.1463) acc1: 23.1771 (22.2234) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1459 max mem: 3531 +Test: [ 160/2387] eta: 0:10:56 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.9920 (1.1193) acc1: 27.3438 (24.2479) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2399 data: 0.1455 max mem: 3531 +Test: [ 170/2387] eta: 0:10:38 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1972 (1.0648) acc1: 97.1354 (28.5529) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1999 data: 0.1062 max mem: 3531 +Test: [ 180/2387] eta: 0:10:26 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1865 (1.0161) acc1: 97.6562 (32.3780) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1949 data: 0.1013 max mem: 3531 +Test: [ 190/2387] eta: 0:10:14 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1865 (0.9725) acc1: 97.9167 (35.8107) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2099 data: 0.1147 max mem: 3531 +Test: [ 200/2387] eta: 0:10:03 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1826 (0.9332) acc1: 97.6562 (38.8889) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2049 data: 0.1085 max mem: 3531 +Test: [ 210/2387] eta: 0:09:54 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1829 (0.8979) acc1: 97.6562 (41.6741) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2087 data: 0.1121 max mem: 3532 +Test: [ 220/2387] eta: 0:09:42 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1847 (0.8656) acc1: 97.9167 (44.2060) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1965 data: 0.1038 max mem: 3532 +Test: [ 230/2387] eta: 0:09:32 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1849 (0.8362) acc1: 97.6562 (46.5165) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1863 data: 0.0947 max mem: 3532 +Test: [ 240/2387] eta: 0:09:23 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1888 (0.8095) acc1: 97.6562 (48.6633) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1933 data: 0.0994 max mem: 3532 +Test: [ 250/2387] eta: 0:09:14 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1726 (0.7807) acc1: 98.9583 (50.6941) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1898 data: 0.0969 max mem: 3532 +Test: [ 260/2387] eta: 0:09:06 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0676 (0.7535) acc1: 100.0000 (52.5832) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1922 data: 0.0998 max mem: 3532 +Test: [ 270/2387] eta: 0:08:59 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0665 (0.7282) acc1: 100.0000 (54.3300) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1960 data: 0.1019 max mem: 3532 +Test: [ 280/2387] eta: 0:08:51 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0677 (0.7049) acc1: 100.0000 (55.9516) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1893 data: 0.0961 max mem: 3532 +Test: [ 290/2387] eta: 0:08:44 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0671 (0.6830) acc1: 100.0000 (57.4635) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1888 data: 0.0955 max mem: 3532 +Test: [ 300/2387] eta: 0:08:38 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0646 (0.6624) acc1: 100.0000 (58.8758) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1957 data: 0.1023 max mem: 3532 +Test: [ 310/2387] eta: 0:08:31 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0677 (0.6456) acc1: 100.0000 (60.1671) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1849 data: 0.0894 max mem: 3532 +Test: [ 320/2387] eta: 0:08:24 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2150 (0.6322) acc1: 98.6979 (61.3618) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1783 data: 0.0803 max mem: 3532 +Test: [ 330/2387] eta: 0:08:17 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2163 (0.6196) acc1: 98.6979 (62.4866) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1802 data: 0.0865 max mem: 3532 +Test: [ 340/2387] eta: 0:08:12 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2159 (0.6078) acc1: 98.4375 (63.5401) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1840 data: 0.0915 max mem: 3532 +Test: [ 350/2387] eta: 0:08:07 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2072 (0.5938) acc1: 99.2188 (64.5722) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1896 data: 0.0949 max mem: 3532 +Test: [ 360/2387] eta: 0:08:00 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1332 (0.5812) acc1: 99.7396 (65.5298) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1784 data: 0.0834 max mem: 3532 +Test: [ 370/2387] eta: 0:07:54 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1427 (0.5695) acc1: 99.2188 (66.4378) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1699 data: 0.0722 max mem: 3532 +Test: [ 380/2387] eta: 0:07:48 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1454 (0.5583) acc1: 99.2188 (67.2934) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1674 data: 0.0694 max mem: 3532 +Test: [ 390/2387] eta: 0:07:43 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1440 (0.5477) acc1: 99.2188 (68.1080) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1699 data: 0.0722 max mem: 3532 +Test: [ 400/2387] eta: 0:07:38 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1437 (0.5377) acc1: 99.2188 (68.8825) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1819 data: 0.0844 max mem: 3532 +Test: [ 410/2387] eta: 0:07:32 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1436 (0.5281) acc1: 98.9583 (69.6193) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1660 data: 0.0714 max mem: 3532 +Test: [ 420/2387] eta: 0:07:26 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1403 (0.5189) acc1: 99.2188 (70.3236) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1470 data: 0.0536 max mem: 3532 +Test: [ 430/2387] eta: 0:07:21 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1469 (0.5103) acc1: 98.9583 (70.9910) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1638 data: 0.0693 max mem: 3532 +Test: [ 440/2387] eta: 0:07:18 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1498 (0.5013) acc1: 99.4792 (71.6423) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0937 max mem: 3533 +Test: [ 450/2387] eta: 0:07:12 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0840 (0.4920) acc1: 100.0000 (72.2711) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0723 max mem: 3533 +Test: [ 460/2387] eta: 0:07:06 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0831 (0.4832) acc1: 100.0000 (72.8726) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1322 data: 0.0329 max mem: 3533 +Test: [ 470/2387] eta: 0:07:03 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0830 (0.4747) acc1: 100.0000 (73.4486) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1714 data: 0.0762 max mem: 3533 +Test: [ 480/2387] eta: 0:07:01 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0842 (0.4666) acc1: 100.0000 (74.0006) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2126 data: 0.1200 max mem: 3533 +Test: [ 490/2387] eta: 0:06:57 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0841 (0.4588) acc1: 100.0000 (74.5301) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1948 data: 0.1002 max mem: 3533 +Test: [ 500/2387] eta: 0:06:52 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0837 (0.4513) acc1: 100.0000 (75.0385) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1599 data: 0.0646 max mem: 3533 +Test: [ 510/2387] eta: 0:06:48 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0836 (0.4442) acc1: 100.0000 (75.5270) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1636 data: 0.0704 max mem: 3533 +Test: [ 520/2387] eta: 0:06:45 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0953 (0.4377) acc1: 100.0000 (75.9967) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1817 data: 0.0891 max mem: 3533 +Test: [ 530/2387] eta: 0:06:41 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0951 (0.4312) acc1: 100.0000 (76.4487) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1748 data: 0.0816 max mem: 3533 +Test: [ 540/2387] eta: 0:06:37 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0929 (0.4250) acc1: 100.0000 (76.8841) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1681 data: 0.0732 max mem: 3533 +Test: [ 550/2387] eta: 0:06:33 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0936 (0.4190) acc1: 100.0000 (77.3036) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1586 data: 0.0633 max mem: 3533 +Test: [ 560/2387] eta: 0:06:29 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0939 (0.4132) acc1: 100.0000 (77.7077) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1609 data: 0.0683 max mem: 3533 +Test: [ 570/2387] eta: 0:06:25 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0934 (0.4076) acc1: 100.0000 (78.0981) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1662 data: 0.0732 max mem: 3533 +Test: [ 580/2387] eta: 0:06:21 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0929 (0.4022) acc1: 100.0000 (78.4751) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1568 data: 0.0630 max mem: 3533 +Test: [ 590/2387] eta: 0:06:18 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0939 (0.3970) acc1: 100.0000 (78.8393) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1660 data: 0.0730 max mem: 3533 +Test: [ 600/2387] eta: 0:06:16 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0967 (0.3922) acc1: 100.0000 (79.1909) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1920 data: 0.0996 max mem: 3533 +Test: [ 610/2387] eta: 0:06:14 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1150 (0.3878) acc1: 100.0000 (79.5243) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1995 data: 0.1078 max mem: 3533 +Test: [ 620/2387] eta: 0:06:11 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1266 (0.3836) acc1: 99.4792 (79.8469) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1979 data: 0.1071 max mem: 3533 +Test: [ 630/2387] eta: 0:06:09 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1269 (0.3796) acc1: 99.4792 (80.1584) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1991 data: 0.1076 max mem: 3533 +Test: [ 640/2387] eta: 0:06:06 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1282 (0.3757) acc1: 99.7396 (80.4618) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2019 data: 0.1102 max mem: 3533 +Test: [ 650/2387] eta: 0:06:03 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1283 (0.3719) acc1: 99.4792 (80.7512) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1899 data: 0.0992 max mem: 3533 +Test: [ 660/2387] eta: 0:06:01 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1285 (0.3682) acc1: 99.4792 (81.0345) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1849 data: 0.0904 max mem: 3533 +Test: [ 670/2387] eta: 0:05:58 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1271 (0.3646) acc1: 99.4792 (81.3090) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1851 data: 0.0892 max mem: 3534 +Test: [ 680/2387] eta: 0:05:55 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1253 (0.3611) acc1: 99.7396 (81.5793) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1803 data: 0.0882 max mem: 3534 +Test: [ 690/2387] eta: 0:05:53 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1265 (0.3592) acc1: 99.2188 (81.7290) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.1860 data: 0.0943 max mem: 3534 +Test: [ 700/2387] eta: 0:05:50 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2690 (0.3579) acc1: 89.3229 (81.8292) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1918 data: 0.0988 max mem: 3534 +Test: [ 710/2387] eta: 0:05:48 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2690 (0.3567) acc1: 89.0625 (81.9375) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1902 data: 0.0976 max mem: 3534 +Test: [ 720/2387] eta: 0:05:45 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2668 (0.3554) acc1: 89.8438 (82.0443) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1928 data: 0.0982 max mem: 3534 +Test: [ 730/2387] eta: 0:05:43 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2686 (0.3543) acc1: 89.3229 (82.1392) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1933 data: 0.0980 max mem: 3534 +Test: [ 740/2387] eta: 0:05:41 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2730 (0.3532) acc1: 88.8021 (82.2291) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.1952 data: 0.1035 max mem: 3534 +Test: [ 750/2387] eta: 0:05:39 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2658 (0.3520) acc1: 89.0625 (82.3201) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.2021 data: 0.1110 max mem: 3534 +Test: [ 760/2387] eta: 0:05:36 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2583 (0.3503) acc1: 89.3229 (82.4439) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1998 data: 0.1088 max mem: 3534 +Test: [ 770/2387] eta: 0:05:34 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1561 (0.3475) acc1: 96.0938 (82.6423) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1953 data: 0.1038 max mem: 3534 +Test: [ 780/2387] eta: 0:05:32 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1426 (0.3448) acc1: 98.9583 (82.8558) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1974 data: 0.1047 max mem: 3534 +Test: [ 790/2387] eta: 0:05:30 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1274 (0.3424) acc1: 99.2188 (83.0021) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.2028 data: 0.1105 max mem: 3534 +Test: [ 800/2387] eta: 0:05:27 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1246 (0.3397) acc1: 99.4792 (83.2062) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2052 data: 0.1127 max mem: 3534 +Test: [ 810/2387] eta: 0:05:25 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1266 (0.3371) acc1: 99.4792 (83.4040) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1980 data: 0.1056 max mem: 3534 +Test: [ 820/2387] eta: 0:05:23 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1266 (0.3347) acc1: 99.4792 (83.5909) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2031 data: 0.1109 max mem: 3534 +Test: [ 830/2387] eta: 0:05:22 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1608 (0.3330) acc1: 98.4375 (83.7439) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2239 data: 0.1312 max mem: 3534 +Test: [ 840/2387] eta: 0:05:19 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2390 (0.3324) acc1: 96.0938 (83.8532) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.2058 data: 0.1137 max mem: 3534 +Test: [ 850/2387] eta: 0:05:17 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2960 (0.3321) acc1: 91.1458 (83.9328) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1936 data: 0.1016 max mem: 3534 +Test: [ 860/2387] eta: 0:05:15 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2909 (0.3316) acc1: 90.8854 (84.0302) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.2058 data: 0.1139 max mem: 3534 +Test: [ 870/2387] eta: 0:05:13 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2778 (0.3309) acc1: 91.1458 (84.1331) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2085 data: 0.1155 max mem: 3534 +Test: [ 880/2387] eta: 0:05:10 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2600 (0.3303) acc1: 94.0104 (84.2305) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1829 data: 0.0898 max mem: 3534 +Test: [ 890/2387] eta: 0:05:08 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2161 (0.3289) acc1: 96.8750 (84.3791) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1699 data: 0.0778 max mem: 3534 +Test: [ 900/2387] eta: 0:05:05 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2601 (0.3306) acc1: 96.0938 (84.3001) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.1791 data: 0.0838 max mem: 3535 +Test: [ 910/2387] eta: 0:05:03 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.5383 (0.3330) acc1: 72.6562 (84.1655) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.2001 data: 0.1051 max mem: 3535 +Test: [ 920/2387] eta: 0:05:01 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.5462 (0.3352) acc1: 72.3958 (84.0394) acc5: 100.0000 (100.0000) acc1_neg: 73.698 acc5_neg: 100.000 time: 0.2194 data: 0.1262 max mem: 3535 +Test: [ 930/2387] eta: 0:04:59 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.5439 (0.3374) acc1: 72.3958 (83.9096) acc5: 100.0000 (100.0000) acc1_neg: 66.146 acc5_neg: 100.000 time: 0.1951 data: 0.1002 max mem: 3535 +Test: [ 940/2387] eta: 0:04:56 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.5448 (0.3396) acc1: 71.3542 (83.7803) acc5: 100.0000 (100.0000) acc1_neg: 67.708 acc5_neg: 100.000 time: 0.1807 data: 0.0864 max mem: 3535 +Test: [ 950/2387] eta: 0:04:54 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.5259 (0.3416) acc1: 73.6979 (83.6770) acc5: 100.0000 (100.0000) acc1_neg: 74.219 acc5_neg: 100.000 time: 0.1914 data: 0.0982 max mem: 3535 +Test: [ 960/2387] eta: 0:04:52 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.5269 (0.3436) acc1: 72.9167 (83.5607) acc5: 100.0000 (100.0000) acc1_neg: 71.094 acc5_neg: 100.000 time: 0.1908 data: 0.0990 max mem: 3535 +Test: [ 970/2387] eta: 0:04:51 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.5366 (0.3457) acc1: 72.6562 (83.4444) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.2193 data: 0.1271 max mem: 3535 +Test: [ 980/2387] eta: 0:04:48 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.5090 (0.3452) acc1: 73.6979 (83.4931) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.2250 data: 0.1321 max mem: 3535 +Test: [ 990/2387] eta: 0:04:46 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1167 (0.3432) acc1: 97.9167 (83.6203) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2022 data: 0.1100 max mem: 3535 +Test: [1000/2387] eta: 0:04:44 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0829 (0.3408) acc1: 100.0000 (83.7673) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1979 data: 0.1054 max mem: 3535 +Test: [1010/2387] eta: 0:04:42 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0897 (0.3385) acc1: 99.7396 (83.9170) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2063 data: 0.1139 max mem: 3535 +Test: [1020/2387] eta: 0:04:40 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1177 (0.3367) acc1: 99.4792 (84.0516) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2237 data: 0.1310 max mem: 3535 +Test: [1030/2387] eta: 0:04:39 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1537 (0.3349) acc1: 98.9583 (84.1962) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2289 data: 0.1361 max mem: 3535 +Test: [1040/2387] eta: 0:04:37 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1275 (0.3328) acc1: 99.4792 (84.3430) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2210 data: 0.1276 max mem: 3535 +Test: [1050/2387] eta: 0:04:35 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1155 (0.3309) acc1: 100.0000 (84.4868) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2104 data: 0.1144 max mem: 3535 +Test: [1060/2387] eta: 0:04:33 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1205 (0.3290) acc1: 99.7396 (84.6268) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2167 data: 0.1198 max mem: 3535 +Test: [1070/2387] eta: 0:04:31 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1293 (0.3274) acc1: 99.2188 (84.7526) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2114 data: 0.1181 max mem: 3535 +Test: [1080/2387] eta: 0:04:28 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1642 (0.3261) acc1: 97.3958 (84.8614) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1873 data: 0.0965 max mem: 3535 +Test: [1090/2387] eta: 0:04:26 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1526 (0.3244) acc1: 97.9167 (84.9894) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1829 data: 0.0912 max mem: 3535 +Test: [1100/2387] eta: 0:04:24 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1336 (0.3227) acc1: 99.2188 (85.1189) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2119 data: 0.1179 max mem: 3535 +Test: [1110/2387] eta: 0:04:22 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1346 (0.3215) acc1: 99.2188 (85.1975) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1957 data: 0.1009 max mem: 3535 +Test: [1120/2387] eta: 0:04:19 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1463 (0.3200) acc1: 98.6979 (85.3138) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1729 data: 0.0806 max mem: 3536 +Test: [1130/2387] eta: 0:04:17 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1356 (0.3183) acc1: 98.9583 (85.4385) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1838 data: 0.0916 max mem: 3536 +Test: [1140/2387] eta: 0:04:15 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1360 (0.3169) acc1: 98.9583 (85.5527) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1801 data: 0.0878 max mem: 3536 +Test: [1150/2387] eta: 0:04:12 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1614 (0.3160) acc1: 98.4375 (85.6382) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1699 data: 0.0788 max mem: 3536 +Test: [1160/2387] eta: 0:04:10 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2033 (0.3149) acc1: 96.3542 (85.7332) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1693 data: 0.0759 max mem: 3536 +Test: [1170/2387] eta: 0:04:08 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2033 (0.3139) acc1: 96.6146 (85.8292) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1720 data: 0.0779 max mem: 3536 +Test: [1180/2387] eta: 0:04:05 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1853 (0.3129) acc1: 96.6146 (85.9075) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1669 data: 0.0734 max mem: 3536 +Test: [1190/2387] eta: 0:04:03 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1672 (0.3118) acc1: 96.6146 (86.0031) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1686 data: 0.0760 max mem: 3536 +Test: [1200/2387] eta: 0:04:01 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1399 (0.3103) acc1: 99.2188 (86.1155) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1750 data: 0.0834 max mem: 3536 +Test: [1210/2387] eta: 0:03:58 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1216 (0.3088) acc1: 99.7396 (86.2239) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1649 data: 0.0734 max mem: 3536 +Test: [1220/2387] eta: 0:03:56 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1030 (0.3069) acc1: 100.0000 (86.3334) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1679 data: 0.0762 max mem: 3536 +Test: [1230/2387] eta: 0:03:53 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0745 (0.3051) acc1: 100.0000 (86.4423) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1763 data: 0.0846 max mem: 3536 +Test: [1240/2387] eta: 0:03:51 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0780 (0.3033) acc1: 99.7396 (86.5488) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1720 data: 0.0814 max mem: 3536 +Test: [1250/2387] eta: 0:03:49 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0824 (0.3015) acc1: 99.7396 (86.6532) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1644 data: 0.0721 max mem: 3536 +Test: [1260/2387] eta: 0:03:47 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0726 (0.2997) acc1: 99.7396 (86.7572) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1683 data: 0.0758 max mem: 3536 +Test: [1270/2387] eta: 0:03:44 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0641 (0.2978) acc1: 100.0000 (86.8612) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1771 data: 0.0830 max mem: 3536 +Test: [1280/2387] eta: 0:03:42 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0740 (0.2962) acc1: 100.0000 (86.9605) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1814 data: 0.0878 max mem: 3536 +Test: [1290/2387] eta: 0:03:40 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0981 (0.2948) acc1: 99.4792 (87.0417) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1854 data: 0.0947 max mem: 3536 +Test: [1300/2387] eta: 0:03:38 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0981 (0.2934) acc1: 99.7396 (87.1325) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1782 data: 0.0859 max mem: 3536 +Test: [1310/2387] eta: 0:03:36 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0990 (0.2920) acc1: 99.7396 (87.2199) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1819 data: 0.0899 max mem: 3536 +Test: [1320/2387] eta: 0:03:33 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1050 (0.2907) acc1: 99.4792 (87.3039) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1794 data: 0.0884 max mem: 3536 +Test: [1330/2387] eta: 0:03:31 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1103 (0.2894) acc1: 99.7396 (87.3891) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1729 data: 0.0812 max mem: 3536 +Test: [1340/2387] eta: 0:03:29 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1393 (0.2890) acc1: 97.6562 (87.4208) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1806 data: 0.0872 max mem: 3536 +Test: [1350/2387] eta: 0:03:27 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0517 (0.2873) acc1: 100.0000 (87.5139) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1795 data: 0.0849 max mem: 3537 +Test: [1360/2387] eta: 0:03:25 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0488 (0.2855) acc1: 100.0000 (87.6056) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1800 data: 0.0879 max mem: 3537 +Test: [1370/2387] eta: 0:03:22 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0470 (0.2838) acc1: 100.0000 (87.6958) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1728 data: 0.0807 max mem: 3537 +Test: [1380/2387] eta: 0:03:20 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0464 (0.2821) acc1: 100.0000 (87.7849) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1739 data: 0.0815 max mem: 3537 +Test: [1390/2387] eta: 0:03:18 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0467 (0.2804) acc1: 100.0000 (87.8727) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1728 data: 0.0818 max mem: 3537 +Test: [1400/2387] eta: 0:03:16 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0452 (0.2787) acc1: 100.0000 (87.9593) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1653 data: 0.0744 max mem: 3537 +Test: [1410/2387] eta: 0:03:14 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0508 (0.2777) acc1: 100.0000 (88.0308) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1747 data: 0.0839 max mem: 3537 +Test: [1420/2387] eta: 0:03:12 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1573 (0.2770) acc1: 98.4375 (88.0819) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1749 data: 0.0837 max mem: 3537 +Test: [1430/2387] eta: 0:03:10 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1667 (0.2763) acc1: 97.9167 (88.1488) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1821 data: 0.0909 max mem: 3537 +Test: [1440/2387] eta: 0:03:08 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1667 (0.2754) acc1: 98.4375 (88.2231) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1944 data: 0.1035 max mem: 3537 +Test: [1450/2387] eta: 0:03:05 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1667 (0.2747) acc1: 98.4375 (88.2877) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1834 data: 0.0898 max mem: 3537 +Test: [1460/2387] eta: 0:03:03 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1239 (0.2737) acc1: 98.9583 (88.3606) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1797 data: 0.0848 max mem: 3537 +Test: [1470/2387] eta: 0:03:01 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1031 (0.2725) acc1: 99.7396 (88.4386) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1935 data: 0.0985 max mem: 3537 +Test: [1480/2387] eta: 0:02:59 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0963 (0.2713) acc1: 99.7396 (88.5132) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1912 data: 0.0964 max mem: 3537 +Test: [1490/2387] eta: 0:02:57 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0981 (0.2702) acc1: 99.7396 (88.5874) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1766 data: 0.0832 max mem: 3537 +Test: [1500/2387] eta: 0:02:55 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1036 (0.2691) acc1: 99.7396 (88.6603) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1801 data: 0.0881 max mem: 3537 +Test: [1510/2387] eta: 0:02:53 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1159 (0.2683) acc1: 99.7396 (88.7318) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1844 data: 0.0936 max mem: 3537 +Test: [1520/2387] eta: 0:02:51 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1574 (0.2681) acc1: 98.9583 (88.7714) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1880 data: 0.0955 max mem: 3537 +Test: [1530/2387] eta: 0:02:49 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2685 (0.2682) acc1: 92.9688 (88.7944) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1896 data: 0.0973 max mem: 3537 +Test: [1540/2387] eta: 0:02:47 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2610 (0.2681) acc1: 92.9688 (88.8337) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1997 data: 0.1091 max mem: 3537 +Test: [1550/2387] eta: 0:02:45 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2544 (0.2680) acc1: 95.8333 (88.8786) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2146 data: 0.1187 max mem: 3537 +Test: [1560/2387] eta: 0:02:43 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2481 (0.2678) acc1: 95.8333 (88.9145) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.2015 data: 0.1049 max mem: 3537 +Test: [1570/2387] eta: 0:02:41 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1917 (0.2673) acc1: 97.6562 (88.9697) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2100 data: 0.1187 max mem: 3537 +Test: [1580/2387] eta: 0:02:39 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1613 (0.2665) acc1: 99.2188 (89.0337) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2009 data: 0.1095 max mem: 3538 +Test: [1590/2387] eta: 0:02:37 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1246 (0.2656) acc1: 99.4792 (89.0998) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1870 data: 0.0954 max mem: 3538 +Test: [1600/2387] eta: 0:02:35 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1101 (0.2649) acc1: 99.7396 (89.1336) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1975 data: 0.1059 max mem: 3538 +Test: [1610/2387] eta: 0:02:33 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1187 (0.2640) acc1: 99.7396 (89.1980) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1895 data: 0.0971 max mem: 3538 +Test: [1620/2387] eta: 0:02:31 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1254 (0.2631) acc1: 99.4792 (89.2583) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1947 data: 0.1014 max mem: 3538 +Test: [1630/2387] eta: 0:02:29 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1349 (0.2625) acc1: 99.2188 (89.3130) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1914 data: 0.0961 max mem: 3538 +Test: [1640/2387] eta: 0:02:27 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1612 (0.2619) acc1: 98.1771 (89.3639) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1725 data: 0.0768 max mem: 3538 +Test: [1650/2387] eta: 0:02:25 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1612 (0.2613) acc1: 98.1771 (89.4147) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1786 data: 0.0853 max mem: 3538 +Test: [1660/2387] eta: 0:02:23 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1605 (0.2607) acc1: 98.1771 (89.4645) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1873 data: 0.0957 max mem: 3538 +Test: [1670/2387] eta: 0:02:21 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1512 (0.2601) acc1: 98.6979 (89.5177) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1824 data: 0.0906 max mem: 3538 +Test: [1680/2387] eta: 0:02:19 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1483 (0.2596) acc1: 98.4375 (89.5595) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1714 data: 0.0793 max mem: 3538 +Test: [1690/2387] eta: 0:02:17 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1585 (0.2591) acc1: 97.9167 (89.6064) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1649 data: 0.0728 max mem: 3538 +Test: [1700/2387] eta: 0:02:15 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1574 (0.2584) acc1: 98.6979 (89.6585) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1645 data: 0.0725 max mem: 3538 +Test: [1710/2387] eta: 0:02:13 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1225 (0.2574) acc1: 99.2188 (89.7165) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1699 data: 0.0780 max mem: 3538 +Test: [1720/2387] eta: 0:02:11 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0874 (0.2564) acc1: 99.7396 (89.7746) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1765 data: 0.0852 max mem: 3538 +Test: [1730/2387] eta: 0:02:09 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0884 (0.2555) acc1: 99.7396 (89.8305) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0826 max mem: 3538 +Test: [1740/2387] eta: 0:02:06 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0928 (0.2546) acc1: 100.0000 (89.8847) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1722 data: 0.0775 max mem: 3538 +Test: [1750/2387] eta: 0:02:04 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0771 (0.2536) acc1: 100.0000 (89.9412) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1710 data: 0.0780 max mem: 3538 +Test: [1760/2387] eta: 0:02:02 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0759 (0.2526) acc1: 100.0000 (89.9981) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1702 data: 0.0794 max mem: 3538 +Test: [1770/2387] eta: 0:02:00 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0889 (0.2518) acc1: 99.7396 (90.0479) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1688 data: 0.0765 max mem: 3538 +Test: [1780/2387] eta: 0:01:58 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1021 (0.2510) acc1: 99.4792 (90.1020) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1752 data: 0.0834 max mem: 3538 +Test: [1790/2387] eta: 0:01:56 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0899 (0.2500) acc1: 99.7396 (90.1555) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1749 data: 0.0817 max mem: 3538 +Test: [1800/2387] eta: 0:01:54 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0836 (0.2492) acc1: 99.7396 (90.2055) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1803 data: 0.0857 max mem: 3538 +Test: [1810/2387] eta: 0:01:52 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0869 (0.2483) acc1: 99.7396 (90.2585) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1810 data: 0.0867 max mem: 3539 +Test: [1820/2387] eta: 0:01:50 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0740 (0.2473) acc1: 100.0000 (90.3118) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1766 data: 0.0812 max mem: 3539 +Test: [1830/2387] eta: 0:01:48 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0897 (0.2467) acc1: 100.0000 (90.3626) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1800 data: 0.0871 max mem: 3539 +Test: [1840/2387] eta: 0:01:46 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1954 (0.2464) acc1: 99.2188 (90.4093) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1929 data: 0.1006 max mem: 3539 +Test: [1850/2387] eta: 0:01:44 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2018 (0.2462) acc1: 99.2188 (90.4559) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1988 data: 0.1046 max mem: 3539 +Test: [1860/2387] eta: 0:01:42 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2021 (0.2459) acc1: 98.6979 (90.5002) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1894 data: 0.0940 max mem: 3539 +Test: [1870/2387] eta: 0:01:40 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2010 (0.2457) acc1: 98.6979 (90.5451) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1899 data: 0.0938 max mem: 3539 +Test: [1880/2387] eta: 0:01:38 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2010 (0.2454) acc1: 99.2188 (90.5918) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1898 data: 0.0964 max mem: 3539 +Test: [1890/2387] eta: 0:01:36 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2003 (0.2452) acc1: 99.2188 (90.6371) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1800 data: 0.0890 max mem: 3539 +Test: [1900/2387] eta: 0:01:34 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2021 (0.2450) acc1: 98.9583 (90.6810) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1732 data: 0.0809 max mem: 3539 +Test: [1910/2387] eta: 0:01:32 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2125 (0.2449) acc1: 98.4375 (90.7193) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1758 data: 0.0815 max mem: 3539 +Test: [1920/2387] eta: 0:01:30 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2235 (0.2446) acc1: 98.1771 (90.7482) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1766 data: 0.0804 max mem: 3539 +Test: [1930/2387] eta: 0:01:28 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1332 (0.2440) acc1: 99.7396 (90.7940) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1789 data: 0.0845 max mem: 3539 +Test: [1940/2387] eta: 0:01:27 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1376 (0.2438) acc1: 99.7396 (90.8049) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1786 data: 0.0861 max mem: 3539 +Test: [1950/2387] eta: 0:01:25 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1639 (0.2435) acc1: 98.4375 (90.8392) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1949 data: 0.0994 max mem: 3539 +Test: [1960/2387] eta: 0:01:23 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1663 (0.2431) acc1: 98.6979 (90.8768) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1762 data: 0.0781 max mem: 3539 +Test: [1970/2387] eta: 0:01:20 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1938 (0.2433) acc1: 96.6146 (90.8603) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1361 data: 0.0405 max mem: 3539 +Test: [1980/2387] eta: 0:01:18 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2759 (0.2435) acc1: 91.9271 (90.8719) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1510 data: 0.0594 max mem: 3539 +Test: [1990/2387] eta: 0:01:16 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.3145 (0.2439) acc1: 92.1875 (90.8730) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1686 data: 0.0754 max mem: 3539 +Test: [2000/2387] eta: 0:01:14 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.3333 (0.2444) acc1: 90.1042 (90.8729) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1553 data: 0.0624 max mem: 3539 +Test: [2010/2387] eta: 0:01:12 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.3395 (0.2449) acc1: 90.1042 (90.8714) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1449 data: 0.0546 max mem: 3539 +Test: [2020/2387] eta: 0:01:10 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.3433 (0.2453) acc1: 90.3646 (90.8705) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1637 data: 0.0722 max mem: 3539 +Test: [2030/2387] eta: 0:01:09 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.3361 (0.2458) acc1: 90.3646 (90.8684) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1849 data: 0.0914 max mem: 3540 +Test: [2040/2387] eta: 0:01:07 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.3357 (0.2462) acc1: 89.8438 (90.8653) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1795 data: 0.0840 max mem: 3540 +Test: [2050/2387] eta: 0:01:05 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.3307 (0.2466) acc1: 90.8854 (90.8692) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1500 data: 0.0516 max mem: 3540 +Test: [2060/2387] eta: 0:01:02 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.3260 (0.2469) acc1: 91.9271 (90.8758) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1372 data: 0.0405 max mem: 3540 +Test: [2070/2387] eta: 0:01:01 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.3160 (0.2473) acc1: 92.1875 (90.8840) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1513 data: 0.0568 max mem: 3540 +Test: [2080/2387] eta: 0:00:59 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.3129 (0.2476) acc1: 92.1875 (90.8907) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1576 data: 0.0627 max mem: 3540 +Test: [2090/2387] eta: 0:00:57 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.3124 (0.2479) acc1: 92.4479 (90.8999) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1585 data: 0.0658 max mem: 3540 +Test: [2100/2387] eta: 0:00:55 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.3162 (0.2482) acc1: 92.7083 (90.9107) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1698 data: 0.0769 max mem: 3540 +Test: [2110/2387] eta: 0:00:53 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.3131 (0.2485) acc1: 92.7083 (90.9200) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1611 data: 0.0671 max mem: 3540 +Test: [2120/2387] eta: 0:00:51 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2879 (0.2483) acc1: 94.2708 (90.9484) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1649 data: 0.0716 max mem: 3540 +Test: [2130/2387] eta: 0:00:49 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1358 (0.2477) acc1: 98.6979 (90.9846) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1835 data: 0.0895 max mem: 3540 +Test: [2140/2387] eta: 0:00:47 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1353 (0.2475) acc1: 98.6979 (90.9965) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1649 data: 0.0680 max mem: 3540 +Test: [2150/2387] eta: 0:00:45 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1578 (0.2471) acc1: 98.6979 (91.0306) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1499 data: 0.0527 max mem: 3540 +Test: [2160/2387] eta: 0:00:43 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1530 (0.2467) acc1: 99.2188 (91.0650) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1658 data: 0.0717 max mem: 3540 +Test: [2170/2387] eta: 0:00:41 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1563 (0.2464) acc1: 98.6979 (91.0907) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1749 data: 0.0799 max mem: 3540 +Test: [2180/2387] eta: 0:00:39 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2301 (0.2470) acc1: 95.3125 (91.0614) acc5: 100.0000 (100.0000) acc1_neg: 67.708 acc5_neg: 100.000 time: 0.1490 data: 0.0521 max mem: 3540 +Test: [2190/2387] eta: 0:00:37 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2181 (0.2470) acc1: 95.8333 (91.0701) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1449 data: 0.0499 max mem: 3540 +Test: [2200/2387] eta: 0:00:35 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1547 (0.2465) acc1: 98.1771 (91.1051) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1599 data: 0.0655 max mem: 3540 +Test: [2210/2387] eta: 0:00:33 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1463 (0.2461) acc1: 98.6979 (91.1343) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1617 data: 0.0686 max mem: 3540 +Test: [2220/2387] eta: 0:00:31 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1534 (0.2457) acc1: 97.9167 (91.1658) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1652 data: 0.0731 max mem: 3540 +Test: [2230/2387] eta: 0:00:29 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1452 (0.2452) acc1: 98.6979 (91.1979) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1631 data: 0.0694 max mem: 3540 +Test: [2240/2387] eta: 0:00:27 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1157 (0.2447) acc1: 99.2188 (91.2292) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1647 data: 0.0707 max mem: 3540 +Test: [2250/2387] eta: 0:00:26 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1530 (0.2444) acc1: 99.2188 (91.2605) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1699 data: 0.0755 max mem: 3540 +Test: [2260/2387] eta: 0:00:24 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1530 (0.2439) acc1: 99.7396 (91.2965) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1598 data: 0.0655 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1064 (0.2432) acc1: 100.0000 (91.3347) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1500 data: 0.0572 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1048 (0.2429) acc1: 100.0000 (91.3486) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1693 data: 0.0754 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1072 (0.2423) acc1: 100.0000 (91.3859) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1610 data: 0.0680 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1118 (0.2417) acc1: 100.0000 (91.4199) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1236 data: 0.0334 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1177 (0.2413) acc1: 99.7396 (91.4527) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1164 data: 0.0260 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1455 (0.2410) acc1: 99.2188 (91.4763) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1193 data: 0.0292 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1650 (0.2408) acc1: 98.1771 (91.5013) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.1206 data: 0.0303 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1660 (0.2405) acc1: 98.1771 (91.5295) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1154 data: 0.0248 max mem: 3541 +Test: [2350/2387] eta: 0:00:06 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1682 (0.2405) acc1: 99.2188 (91.5282) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1172 data: 0.0267 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1739 (0.2404) acc1: 98.9583 (91.5495) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1291 data: 0.0392 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1947 (0.2403) acc1: 97.6562 (91.5625) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.1298 data: 0.0403 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2400 (0.2404) acc1: 94.0104 (91.5640) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1116 data: 0.0222 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2677 (0.2405) acc1: 92.7083 (91.5676) acc5: 100.0000 (100.0000) acc1_neg: 95.349 acc5_neg: 100.000 time: 0.0908 data: 0.0038 max mem: 3541 +Test: Total time: 0:07:25 (0.1868 s / it) +* Overall Acc@1 91.568 Acc@5 100.000 + target == real Acc@1 22.354 Acc@5 100.000 + target == fake Acc@1 96.428 Acc@5 100.000 + loss 0.240 + PR-AUC 0.9899 ROC-AUC 0.8713 EER 20.66% (thr=0.7983) +Accuracy of the network on the 916267 test images: 91.6% +[rank0]:[W725 18:47:33.445075932 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:47:52.736000 2579641 torch/distributed/run.py:766] +W0725 18:47:52.736000 2579641 torch/distributed/run.py:766] ***************************************** +W0725 18:47:52.736000 2579641 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:47:52.736000 2579641 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:48:22.442492702 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:48:23.666411985 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:48:23.682735989 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:48:23.686040709 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:48:23.866538631 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +[rank0]:[W725 18:48:23.874079882 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:48:23.877573115 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:48:23.913133229 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed3_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed3_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:00:59 acc1_pos: 22.917 acc5_pos: 100.000 loss: 1.1469 (1.1469) acc1: 22.9167 (22.9167) acc5: 100.0000 (100.0000) time: 7.0792 data: 5.3619 max mem: 3531 +Test: [ 10/517] eta: 0:06:08 acc1_pos: 10.677 acc5_pos: 100.000 loss: 1.1989 (1.2050) acc1: 15.6250 (19.3892) acc5: 100.0000 (100.0000) time: 0.7261 data: 0.4878 max mem: 3531 +Test: [ 20/517] eta: 0:03:32 acc1_pos: 9.896 acc5_pos: 100.000 loss: 1.1989 (1.2156) acc1: 15.6250 (18.7624) acc5: 100.0000 (100.0000) time: 0.0953 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:03:13 acc1_pos: 8.854 acc5_pos: 100.000 loss: 1.2612 (1.2387) acc1: 15.1042 (17.8595) acc5: 100.0000 (100.0000) time: 0.2159 data: 0.1212 max mem: 3531 +Test: [ 40/517] eta: 0:03:05 acc1_pos: 10.938 acc5_pos: 100.000 loss: 1.2693 (1.2492) acc1: 13.8021 (16.6540) acc5: 100.0000 (100.0000) time: 0.3467 data: 0.2568 max mem: 3531 +Test: [ 50/517] eta: 0:02:53 acc1_pos: 59.174 acc5_pos: 100.000 loss: 1.1524 (1.1823) acc1: 15.1042 (21.4767) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.3339 data: 0.2432 max mem: 3531 +Test: [ 60/517] eta: 0:02:42 acc1_pos: 59.174 acc5_pos: 100.000 loss: 0.3888 (1.0418) acc1: 88.5417 (33.2010) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2881 data: 0.1940 max mem: 3531 +Test: [ 70/517] eta: 0:02:32 acc1_pos: 59.174 acc5_pos: 100.000 loss: 0.2756 (0.9310) acc1: 97.1354 (42.2865) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2649 data: 0.1685 max mem: 3531 +Test: [ 80/517] eta: 0:02:24 acc1_pos: 59.174 acc5_pos: 100.000 loss: 0.2599 (0.8525) acc1: 97.6562 (48.8619) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2549 data: 0.1596 max mem: 3531 +Test: [ 90/517] eta: 0:02:19 acc1_pos: 59.174 acc5_pos: 100.000 loss: 0.3044 (0.7913) acc1: 96.6146 (54.1638) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2699 data: 0.1762 max mem: 3531 +Test: [100/517] eta: 0:02:13 acc1_pos: 59.174 acc5_pos: 100.000 loss: 0.2611 (0.7362) acc1: 98.6979 (58.5680) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2749 data: 0.1791 max mem: 3531 +Test: [110/517] eta: 0:02:07 acc1_pos: 59.174 acc5_pos: 100.000 loss: 0.2075 (0.6882) acc1: 99.4792 (62.2701) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2499 data: 0.1527 max mem: 3531 +Test: [120/517] eta: 0:02:02 acc1_pos: 59.174 acc5_pos: 100.000 loss: 0.2402 (0.6542) acc1: 99.2188 (65.1881) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2499 data: 0.1544 max mem: 3531 +Test: [130/517] eta: 0:01:57 acc1_pos: 59.174 acc5_pos: 100.000 loss: 0.2709 (0.6248) acc1: 97.9167 (67.6566) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1588 max mem: 3531 +Test: [140/517] eta: 0:01:53 acc1_pos: 59.174 acc5_pos: 100.000 loss: 0.2196 (0.5950) acc1: 99.7396 (69.9339) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2499 data: 0.1549 max mem: 3531 +Test: [150/517] eta: 0:01:49 acc1_pos: 59.174 acc5_pos: 100.000 loss: 0.2055 (0.5701) acc1: 99.7396 (71.8819) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2549 data: 0.1594 max mem: 3531 +Test: [160/517] eta: 0:01:45 acc1_pos: 59.174 acc5_pos: 100.000 loss: 0.2388 (0.5511) acc1: 98.6979 (73.4957) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2599 data: 0.1628 max mem: 3531 +Test: [170/517] eta: 0:01:41 acc1_pos: 59.174 acc5_pos: 100.000 loss: 0.2578 (0.5336) acc1: 98.1771 (74.9299) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2599 data: 0.1631 max mem: 3531 +Test: [180/517] eta: 0:01:37 acc1_pos: 59.174 acc5_pos: 100.000 loss: 0.2513 (0.5179) acc1: 98.1771 (76.1525) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2499 data: 0.1529 max mem: 3531 +Test: [190/517] eta: 0:01:34 acc1_pos: 59.174 acc5_pos: 100.000 loss: 0.1966 (0.5004) acc1: 99.4792 (77.3860) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2499 data: 0.1525 max mem: 3531 +Test: [200/517] eta: 0:01:30 acc1_pos: 59.174 acc5_pos: 100.000 loss: 0.1866 (0.4851) acc1: 100.0000 (78.4955) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2499 data: 0.1523 max mem: 3531 +Test: [210/517] eta: 0:01:27 acc1_pos: 59.174 acc5_pos: 100.000 loss: 0.2250 (0.4783) acc1: 96.6146 (79.0075) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.2499 data: 0.1517 max mem: 3532 +Test: [220/517] eta: 0:01:24 acc1_pos: 59.174 acc5_pos: 100.000 loss: 0.3443 (0.4723) acc1: 89.0625 (79.4719) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.2599 data: 0.1611 max mem: 3532 +Test: [230/517] eta: 0:01:20 acc1_pos: 59.174 acc5_pos: 100.000 loss: 0.3136 (0.4648) acc1: 92.7083 (80.2534) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2549 data: 0.1588 max mem: 3532 +Test: [240/517] eta: 0:01:17 acc1_pos: 59.174 acc5_pos: 100.000 loss: 0.2685 (0.4562) acc1: 99.2188 (81.0393) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2499 data: 0.1537 max mem: 3532 +Test: [250/517] eta: 0:01:14 acc1_pos: 59.174 acc5_pos: 100.000 loss: 0.2364 (0.4470) acc1: 99.4792 (81.7490) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2549 data: 0.1543 max mem: 3532 +Test: [260/517] eta: 0:01:11 acc1_pos: 59.174 acc5_pos: 100.000 loss: 0.2050 (0.4383) acc1: 99.4792 (82.4154) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2549 data: 0.1560 max mem: 3532 +Test: [270/517] eta: 0:01:08 acc1_pos: 59.174 acc5_pos: 100.000 loss: 0.2368 (0.4312) acc1: 98.6979 (82.9778) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2549 data: 0.1593 max mem: 3532 +Test: [280/517] eta: 0:01:05 acc1_pos: 59.174 acc5_pos: 100.000 loss: 0.2440 (0.4243) acc1: 97.9167 (83.5141) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2599 data: 0.1655 max mem: 3532 +Test: [290/517] eta: 0:01:02 acc1_pos: 59.174 acc5_pos: 100.000 loss: 0.2373 (0.4182) acc1: 98.4375 (84.0349) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2599 data: 0.1621 max mem: 3532 +Test: [300/517] eta: 0:00:59 acc1_pos: 59.174 acc5_pos: 100.000 loss: 0.2298 (0.4121) acc1: 99.2188 (84.5420) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2549 data: 0.1551 max mem: 3532 +Test: [310/517] eta: 0:00:57 acc1_pos: 59.174 acc5_pos: 100.000 loss: 0.2481 (0.4076) acc1: 99.4792 (85.0047) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2599 data: 0.1617 max mem: 3532 +Test: [320/517] eta: 0:00:54 acc1_pos: 59.174 acc5_pos: 100.000 loss: 0.2532 (0.4026) acc1: 99.2188 (85.4378) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1565 max mem: 3532 +Test: [330/517] eta: 0:00:51 acc1_pos: 59.174 acc5_pos: 100.000 loss: 0.2221 (0.3968) acc1: 99.4792 (85.8667) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2449 data: 0.1493 max mem: 3532 +Test: [340/517] eta: 0:00:48 acc1_pos: 59.174 acc5_pos: 100.000 loss: 0.1958 (0.3920) acc1: 99.2188 (86.2162) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2449 data: 0.1490 max mem: 3532 +Test: [350/517] eta: 0:00:45 acc1_pos: 59.174 acc5_pos: 100.000 loss: 0.2729 (0.3894) acc1: 97.6562 (86.5014) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.2449 data: 0.1473 max mem: 3532 +Test: [360/517] eta: 0:00:42 acc1_pos: 59.174 acc5_pos: 100.000 loss: 0.2963 (0.3868) acc1: 96.8750 (86.7548) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2449 data: 0.1483 max mem: 3532 +Test: [370/517] eta: 0:00:39 acc1_pos: 59.174 acc5_pos: 100.000 loss: 0.2280 (0.3817) acc1: 98.4375 (87.1034) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2449 data: 0.1501 max mem: 3532 +Test: [380/517] eta: 0:00:37 acc1_pos: 59.174 acc5_pos: 100.000 loss: 0.1900 (0.3769) acc1: 99.7396 (87.4180) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2549 data: 0.1602 max mem: 3532 +Test: [390/517] eta: 0:00:34 acc1_pos: 59.174 acc5_pos: 100.000 loss: 0.1989 (0.3733) acc1: 99.4792 (87.6778) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.3349 data: 0.2380 max mem: 3532 +Test: [400/517] eta: 0:00:31 acc1_pos: 59.174 acc5_pos: 100.000 loss: 0.2263 (0.3697) acc1: 99.7396 (87.9643) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2899 data: 0.1916 max mem: 3532 +Test: [410/517] eta: 0:00:29 acc1_pos: 59.174 acc5_pos: 100.000 loss: 0.2470 (0.3681) acc1: 99.4792 (88.1849) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2149 data: 0.1189 max mem: 3532 +Test: [420/517] eta: 0:00:26 acc1_pos: 59.174 acc5_pos: 100.000 loss: 0.2699 (0.3654) acc1: 98.9583 (88.4439) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2549 data: 0.1614 max mem: 3532 +Test: [430/517] eta: 0:00:23 acc1_pos: 59.174 acc5_pos: 100.000 loss: 0.2601 (0.3633) acc1: 98.9583 (88.6462) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2549 data: 0.1603 max mem: 3532 +Test: [440/517] eta: 0:00:20 acc1_pos: 59.174 acc5_pos: 100.000 loss: 0.2726 (0.3610) acc1: 98.1771 (88.8605) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2499 data: 0.1525 max mem: 3533 +Test: [450/517] eta: 0:00:18 acc1_pos: 59.174 acc5_pos: 100.000 loss: 0.2566 (0.3586) acc1: 98.4375 (89.0856) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2699 data: 0.1727 max mem: 3533 +Test: [460/517] eta: 0:00:15 acc1_pos: 59.174 acc5_pos: 100.000 loss: 0.2600 (0.3569) acc1: 98.9583 (89.2489) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.2599 data: 0.1645 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 59.174 acc5_pos: 100.000 loss: 0.2404 (0.3540) acc1: 99.7396 (89.4634) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2340 data: 0.1405 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 59.174 acc5_pos: 100.000 loss: 0.2175 (0.3513) acc1: 99.7396 (89.6792) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2083 data: 0.1177 max mem: 3533 +Test: [490/517] eta: 0:00:07 acc1_pos: 59.174 acc5_pos: 100.000 loss: 0.2334 (0.3499) acc1: 99.7396 (89.8316) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1707 data: 0.0812 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 59.174 acc5_pos: 100.000 loss: 0.2783 (0.3486) acc1: 98.9583 (90.0059) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1755 data: 0.0860 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 59.174 acc5_pos: 100.000 loss: 0.2627 (0.3471) acc1: 98.6979 (90.1312) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1729 data: 0.0835 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 59.174 acc5_pos: 100.000 loss: 0.2391 (0.3458) acc1: 98.4375 (90.2251) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1701 data: 0.0789 max mem: 3533 +Test: Total time: 0:02:15 (0.2612 s / it) +* Overall Acc@1 90.225 Acc@5 100.000 + target == real Acc@1 17.823 Acc@5 100.000 + target == fake Acc@1 97.736 Acc@5 100.000 + loss 0.346 + PR-AUC 0.9504 ROC-AUC 0.7211 EER 33.49% (thr=0.7681) +Accuracy of the network on the 198427 test images: 90.2% +[rank0]:[W725 18:50:46.043814769 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0725 18:51:05.852000 2580232 torch/distributed/run.py:766] +W0725 18:51:05.852000 2580232 torch/distributed/run.py:766] ***************************************** +W0725 18:51:05.852000 2580232 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:51:05.852000 2580232 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:51:32.295398472 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:51:33.630989327 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:51:33.756646865 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:51:33.924020750 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:51:33.023642448 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:51:33.116063321 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:51:33.177764089 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank6]:[W725 18:51:33.178149602 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed3_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed3_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:39:16 acc1_pos: 99.479 acc5_pos: 100.000 loss: 0.2596 (0.2596) acc1: 99.4792 (99.4792) acc5: 100.0000 (100.0000) time: 6.0588 data: 4.5993 max mem: 3531 +Test: [ 10/389] eta: 0:03:59 acc1_pos: 98.958 acc5_pos: 100.000 loss: 0.2855 (0.3146) acc1: 98.9583 (95.7860) acc5: 100.0000 (100.0000) time: 0.6328 data: 0.4185 max mem: 3531 +Test: [ 20/389] eta: 0:02:19 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.3244 (0.3266) acc1: 98.4375 (96.9618) acc5: 100.0000 (100.0000) time: 0.0948 data: 0.0004 max mem: 3531 +Test: [ 30/389] eta: 0:01:52 acc1_pos: 98.177 acc5_pos: 100.000 loss: 0.2781 (0.3065) acc1: 99.2188 (97.6226) acc5: 100.0000 (100.0000) time: 0.1397 data: 0.0407 max mem: 3531 +Test: [ 40/389] eta: 0:01:38 acc1_pos: 84.115 acc5_pos: 100.000 loss: 0.2746 (0.3145) acc1: 99.2188 (97.1862) acc5: 100.0000 (100.0000) time: 0.1803 data: 0.0839 max mem: 3531 +Test: [ 50/389] eta: 0:01:30 acc1_pos: 97.656 acc5_pos: 100.000 loss: 0.3367 (0.3191) acc1: 97.9167 (97.1814) acc5: 100.0000 (100.0000) time: 0.1899 data: 0.0963 max mem: 3531 +Test: [ 60/389] eta: 0:01:21 acc1_pos: 84.635 acc5_pos: 100.000 loss: 0.3564 (0.3135) acc1: 97.9167 (97.2080) acc5: 100.0000 (100.0000) time: 0.1747 data: 0.0826 max mem: 3531 +Test: [ 70/389] eta: 0:01:14 acc1_pos: 97.396 acc5_pos: 100.000 loss: 0.2890 (0.3083) acc1: 97.9167 (96.9227) acc5: 100.0000 (100.0000) time: 0.1550 data: 0.0637 max mem: 3531 +Test: [ 80/389] eta: 0:01:10 acc1_pos: 99.740 acc5_pos: 100.000 loss: 0.3074 (0.3097) acc1: 97.3958 (96.5953) acc5: 100.0000 (100.0000) time: 0.1705 data: 0.0781 max mem: 3531 +Test: [ 90/389] eta: 0:01:06 acc1_pos: 97.396 acc5_pos: 100.000 loss: 0.1886 (0.2994) acc1: 98.4375 (96.8464) acc5: 100.0000 (100.0000) time: 0.1806 data: 0.0872 max mem: 3531 +Test: [100/389] eta: 0:01:02 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.2789 (0.3044) acc1: 98.4375 (96.5630) acc5: 100.0000 (100.0000) time: 0.1623 data: 0.0703 max mem: 3531 +Test: [110/389] eta: 0:00:58 acc1_pos: 99.219 acc5_pos: 100.000 loss: 0.3120 (0.3103) acc1: 95.5729 (95.9671) acc5: 100.0000 (100.0000) time: 0.1573 data: 0.0659 max mem: 3531 +Test: [120/389] eta: 0:00:55 acc1_pos: 99.479 acc5_pos: 100.000 loss: 0.2973 (0.3094) acc1: 97.1354 (96.0550) acc5: 100.0000 (100.0000) time: 0.1699 data: 0.0766 max mem: 3531 +Test: [130/389] eta: 0:00:52 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4178 (0.3627) acc1: 96.0938 (91.1836) acc5: 100.0000 (100.0000) acc1_neg: 34.115 acc5_neg: 100.000 time: 0.1449 data: 0.0510 max mem: 3531 +Test: [140/389] eta: 0:00:49 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.3360 (0.4342) acc1: 2.3438 (84.8700) acc5: 100.0000 (100.0000) acc1_neg: 0.781 acc5_neg: 100.000 time: 0.1349 data: 0.0424 max mem: 3531 +Test: [150/389] eta: 0:00:47 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.2751 (0.4823) acc1: 2.6042 (80.0755) acc5: 100.0000 (100.0000) acc1_neg: 18.229 acc5_neg: 100.000 time: 0.1699 data: 0.0775 max mem: 3531 +Test: [160/389] eta: 0:00:44 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.0472 (0.5211) acc1: 7.0312 (75.9834) acc5: 100.0000 (100.0000) acc1_neg: 19.531 acc5_neg: 100.000 time: 0.1849 data: 0.0897 max mem: 3531 +Test: [170/389] eta: 0:00:42 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.9501 (0.5517) acc1: 10.1562 (72.0928) acc5: 100.0000 (100.0000) acc1_neg: 7.552 acc5_neg: 100.000 time: 0.1599 data: 0.0650 max mem: 3531 +Test: [180/389] eta: 0:00:39 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.0144 (0.5833) acc1: 6.5104 (68.3572) acc5: 100.0000 (100.0000) acc1_neg: 3.906 acc5_neg: 100.000 time: 0.1566 data: 0.0637 max mem: 3531 +Test: [190/389] eta: 0:00:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.1188 (0.6202) acc1: 3.1250 (65.2092) acc5: 100.0000 (100.0000) acc1_neg: 0.000 acc5_neg: 100.000 time: 0.1949 data: 0.1030 max mem: 3531 +Test: [200/389] eta: 0:00:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.3446 (0.6514) acc1: 0.5208 (62.5531) acc5: 100.0000 (100.0000) acc1_neg: 0.521 acc5_neg: 100.000 time: 0.1982 data: 0.1044 max mem: 3531 +Test: [210/389] eta: 0:00:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.0898 (0.6662) acc1: 1.0417 (60.9215) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.1849 data: 0.0905 max mem: 3532 +Test: [220/389] eta: 0:00:32 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.8567 (0.6733) acc1: 28.9062 (60.5015) acc5: 100.0000 (100.0000) acc1_neg: 1.042 acc5_neg: 100.000 time: 0.1913 data: 0.0996 max mem: 3532 +Test: [230/389] eta: 0:00:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.1560 (0.6965) acc1: 5.9896 (58.2409) acc5: 100.0000 (100.0000) acc1_neg: 3.646 acc5_neg: 100.000 time: 0.1849 data: 0.0930 max mem: 3532 +Test: [240/389] eta: 0:00:28 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.1428 (0.7109) acc1: 4.1667 (56.3613) acc5: 100.0000 (100.0000) acc1_neg: 21.094 acc5_neg: 100.000 time: 0.1635 data: 0.0704 max mem: 3532 +Test: [250/389] eta: 0:00:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.8710 (0.7171) acc1: 21.3542 (55.0403) acc5: 100.0000 (100.0000) acc1_neg: 26.823 acc5_neg: 100.000 time: 0.1378 data: 0.0439 max mem: 3532 +Test: [260/389] eta: 0:00:23 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.8933 (0.7340) acc1: 21.3542 (53.3086) acc5: 100.0000 (100.0000) acc1_neg: 0.000 acc5_neg: 100.000 time: 0.1357 data: 0.0413 max mem: 3532 +Test: [270/389] eta: 0:00:21 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.3260 (0.7623) acc1: 1.0417 (51.4193) acc5: 100.0000 (100.0000) acc1_neg: 1.562 acc5_neg: 100.000 time: 0.1485 data: 0.0541 max mem: 3532 +Test: [280/389] eta: 0:00:19 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.4262 (0.7860) acc1: 0.2604 (50.0213) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.1442 data: 0.0518 max mem: 3532 +Test: [290/389] eta: 0:00:17 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.0811 (0.7921) acc1: 6.7708 (49.1651) acc5: 100.0000 (100.0000) acc1_neg: 14.062 acc5_neg: 100.000 time: 0.1395 data: 0.0484 max mem: 3532 +Test: [300/389] eta: 0:00:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.0987 (0.8045) acc1: 4.9479 (47.6095) acc5: 100.0000 (100.0000) acc1_neg: 3.125 acc5_neg: 100.000 time: 0.1464 data: 0.0528 max mem: 3532 +Test: [310/389] eta: 0:00:14 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.2088 (0.8202) acc1: 1.8229 (46.3031) acc5: 100.0000 (100.0000) acc1_neg: 1.823 acc5_neg: 100.000 time: 0.1506 data: 0.0572 max mem: 3532 +Test: [320/389] eta: 0:00:12 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.2181 (0.8316) acc1: 5.7292 (45.1024) acc5: 100.0000 (100.0000) acc1_neg: 14.323 acc5_neg: 100.000 time: 0.1623 data: 0.0712 max mem: 3532 +Test: [330/389] eta: 0:00:10 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.4661 (0.8576) acc1: 0.2604 (43.7815) acc5: 100.0000 (100.0000) acc1_neg: 0.260 acc5_neg: 100.000 time: 0.1787 data: 0.0863 max mem: 3532 +Test: [340/389] eta: 0:00:08 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.1160 (0.8565) acc1: 7.2917 (43.6416) acc5: 100.0000 (100.0000) acc1_neg: 26.042 acc5_neg: 100.000 time: 0.1888 data: 0.0929 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.9007 (0.8665) acc1: 21.0938 (42.6408) acc5: 100.0000 (100.0000) acc1_neg: 1.042 acc5_neg: 100.000 time: 0.1984 data: 0.1036 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.1477 (0.8750) acc1: 5.7292 (41.6147) acc5: 100.0000 (100.0000) acc1_neg: 5.990 acc5_neg: 100.000 time: 0.1813 data: 0.0884 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.1462 (0.8782) acc1: 6.7708 (41.2638) acc5: 100.0000 (100.0000) acc1_neg: 27.344 acc5_neg: 100.000 time: 0.1551 data: 0.0621 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.9774 (0.8835) acc1: 15.3646 (40.6476) acc5: 100.0000 (100.0000) acc1_neg: 3.906 acc5_neg: 100.000 time: 0.1603 data: 0.0685 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.9940 (0.8876) acc1: 15.3646 (40.1043) acc5: 100.0000 (100.0000) acc1_neg: 9.016 acc5_neg: 100.000 time: 0.1552 data: 0.0602 max mem: 3532 +Test: Total time: 0:01:09 (0.1786 s / it) +* Overall Acc@1 40.104 Acc@5 100.000 + target == real Acc@1 96.082 Acc@5 100.000 + target == fake Acc@1 14.085 Acc@5 100.000 + loss 0.888 + PR-AUC 0.8138 ROC-AUC 0.6923 EER 35.69% (thr=0.2739) +Accuracy of the network on the 149358 test images: 40.1% +[rank0]:[W725 18:52:54.188859362 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:53:13.464000 2580805 torch/distributed/run.py:766] +W0725 18:53:13.464000 2580805 torch/distributed/run.py:766] ***************************************** +W0725 18:53:13.464000 2580805 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:53:13.464000 2580805 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:53:41.115837291 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:53:41.244826932 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:53:41.384384141 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:53:41.464601884 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:53:42.660526497 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +[rank3]:[W725 18:53:42.673976647 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:53:42.685235851 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:53:42.699146072 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed3_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed3_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:35:26 acc1_pos: 77.083 acc5_pos: 100.000 loss: 0.4674 (0.4674) acc1: 77.0833 (77.0833) acc5: 100.0000 (100.0000) time: 6.3862 data: 4.7881 max mem: 3531 +Test: [ 10/333] eta: 0:03:34 acc1_pos: 86.719 acc5_pos: 100.000 loss: 0.3241 (0.3357) acc1: 86.7188 (86.4820) acc5: 100.0000 (100.0000) time: 0.6628 data: 0.4358 max mem: 3531 +Test: [ 20/333] eta: 0:02:06 acc1_pos: 93.229 acc5_pos: 100.000 loss: 0.3131 (0.3279) acc1: 86.7188 (86.2351) acc5: 100.0000 (100.0000) time: 0.1050 data: 0.0107 max mem: 3531 +Test: [ 30/333] eta: 0:02:00 acc1_pos: 60.417 acc5_pos: 100.000 loss: 0.4168 (0.4555) acc1: 74.7396 (76.5541) acc5: 100.0000 (100.0000) time: 0.2497 data: 0.1500 max mem: 3531 +Test: [ 40/333] eta: 0:01:44 acc1_pos: 59.635 acc5_pos: 100.000 loss: 0.7407 (0.5300) acc1: 54.6875 (70.8841) acc5: 100.0000 (100.0000) time: 0.3091 data: 0.2137 max mem: 3531 +Test: [ 50/333] eta: 0:01:32 acc1_pos: 53.385 acc5_pos: 100.000 loss: 0.7417 (0.5774) acc1: 53.3854 (67.1977) acc5: 100.0000 (100.0000) time: 0.2162 data: 0.1232 max mem: 3531 +Test: [ 60/333] eta: 0:01:24 acc1_pos: 49.479 acc5_pos: 100.000 loss: 0.7571 (0.6120) acc1: 52.8646 (64.7285) acc5: 100.0000 (100.0000) time: 0.2080 data: 0.1148 max mem: 3531 +Test: [ 70/333] eta: 0:01:16 acc1_pos: 44.792 acc5_pos: 100.000 loss: 0.7778 (0.6345) acc1: 53.1250 (63.0318) acc5: 100.0000 (100.0000) time: 0.1997 data: 0.1075 max mem: 3531 +Test: [ 80/333] eta: 0:01:10 acc1_pos: 61.719 acc5_pos: 100.000 loss: 0.7477 (0.6491) acc1: 54.1667 (61.6737) acc5: 100.0000 (100.0000) time: 0.1825 data: 0.0887 max mem: 3531 +Test: [ 90/333] eta: 0:01:05 acc1_pos: 63.281 acc5_pos: 100.000 loss: 0.6989 (0.6544) acc1: 57.8125 (61.3467) acc5: 100.0000 (100.0000) time: 0.1938 data: 0.1000 max mem: 3531 +Test: [100/333] eta: 0:01:00 acc1_pos: 65.885 acc5_pos: 100.000 loss: 0.6975 (0.6613) acc1: 58.5938 (60.9091) acc5: 100.0000 (100.0000) time: 0.1914 data: 0.0971 max mem: 3531 +Test: [110/333] eta: 0:00:56 acc1_pos: 67.969 acc5_pos: 100.000 loss: 0.7195 (0.6692) acc1: 55.2083 (60.3815) acc5: 100.0000 (100.0000) time: 0.1924 data: 0.0988 max mem: 3531 +Test: [120/333] eta: 0:00:53 acc1_pos: 64.844 acc5_pos: 100.000 loss: 0.7083 (0.6714) acc1: 55.2083 (60.2961) acc5: 100.0000 (100.0000) time: 0.1945 data: 0.1029 max mem: 3531 +Test: [130/333] eta: 0:00:50 acc1_pos: 59.896 acc5_pos: 100.000 loss: 0.6545 (0.6739) acc1: 59.8958 (60.0191) acc5: 100.0000 (100.0000) time: 0.1985 data: 0.1055 max mem: 3531 +Test: [140/333] eta: 0:00:46 acc1_pos: 82.552 acc5_pos: 100.000 loss: 0.6654 (0.6744) acc1: 59.8958 (60.0842) acc5: 100.0000 (100.0000) time: 0.2022 data: 0.1083 max mem: 3531 +Test: [150/333] eta: 0:00:43 acc1_pos: 60.677 acc5_pos: 100.000 loss: 0.6812 (0.6782) acc1: 59.1146 (59.8027) acc5: 100.0000 (100.0000) time: 0.1913 data: 0.0975 max mem: 3531 +Test: [160/333] eta: 0:00:41 acc1_pos: 50.329 acc5_pos: 100.000 loss: 0.7596 (0.6862) acc1: 52.8646 (59.1906) acc5: 100.0000 (100.0000) acc1_neg: 45.573 acc5_neg: 100.000 time: 0.2037 data: 0.1082 max mem: 3531 +Test: [170/333] eta: 0:00:38 acc1_pos: 50.329 acc5_pos: 100.000 loss: 0.7595 (0.6916) acc1: 52.8646 (58.8785) acc5: 100.0000 (100.0000) acc1_neg: 57.031 acc5_neg: 100.000 time: 0.1997 data: 0.1042 max mem: 3531 +Test: [180/333] eta: 0:00:35 acc1_pos: 50.329 acc5_pos: 100.000 loss: 0.7083 (0.6914) acc1: 59.1146 (59.0254) acc5: 100.0000 (100.0000) acc1_neg: 65.365 acc5_neg: 100.000 time: 0.2013 data: 0.1060 max mem: 3531 +Test: [190/333] eta: 0:00:33 acc1_pos: 50.329 acc5_pos: 100.000 loss: 0.6531 (0.6862) acc1: 65.1042 (59.5754) acc5: 100.0000 (100.0000) acc1_neg: 66.927 acc5_neg: 100.000 time: 0.2095 data: 0.1148 max mem: 3531 +Test: [200/333] eta: 0:00:30 acc1_pos: 50.329 acc5_pos: 100.000 loss: 0.6004 (0.6830) acc1: 66.9271 (59.9490) acc5: 100.0000 (100.0000) acc1_neg: 59.635 acc5_neg: 100.000 time: 0.1976 data: 0.1015 max mem: 3531 +Test: [210/333] eta: 0:00:28 acc1_pos: 50.329 acc5_pos: 100.000 loss: 0.6864 (0.6841) acc1: 61.1979 (59.9637) acc5: 100.0000 (100.0000) acc1_neg: 54.688 acc5_neg: 100.000 time: 0.1985 data: 0.1043 max mem: 3532 +Test: [220/333] eta: 0:00:25 acc1_pos: 50.329 acc5_pos: 100.000 loss: 0.7029 (0.6884) acc1: 60.1562 (59.6884) acc5: 100.0000 (100.0000) acc1_neg: 61.198 acc5_neg: 100.000 time: 0.1991 data: 0.1041 max mem: 3532 +Test: [230/333] eta: 0:00:23 acc1_pos: 50.329 acc5_pos: 100.000 loss: 0.6101 (0.6840) acc1: 65.8854 (60.2092) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.1996 data: 0.1033 max mem: 3532 +Test: [240/333] eta: 0:00:20 acc1_pos: 50.329 acc5_pos: 100.000 loss: 0.6177 (0.6828) acc1: 66.6667 (60.3281) acc5: 100.0000 (100.0000) acc1_neg: 63.802 acc5_neg: 100.000 time: 0.1959 data: 0.1038 max mem: 3532 +Test: [250/333] eta: 0:00:18 acc1_pos: 50.329 acc5_pos: 100.000 loss: 0.6584 (0.6818) acc1: 63.8021 (60.4519) acc5: 100.0000 (100.0000) acc1_neg: 58.854 acc5_neg: 100.000 time: 0.1996 data: 0.1067 max mem: 3532 +Test: [260/333] eta: 0:00:16 acc1_pos: 50.329 acc5_pos: 100.000 loss: 0.6402 (0.6804) acc1: 64.5833 (60.6292) acc5: 100.0000 (100.0000) acc1_neg: 70.573 acc5_neg: 100.000 time: 0.2084 data: 0.1121 max mem: 3532 +Test: [270/333] eta: 0:00:14 acc1_pos: 50.329 acc5_pos: 100.000 loss: 0.6333 (0.6790) acc1: 65.1042 (60.8155) acc5: 100.0000 (100.0000) acc1_neg: 64.844 acc5_neg: 100.000 time: 0.2002 data: 0.1063 max mem: 3532 +Test: [280/333] eta: 0:00:11 acc1_pos: 50.329 acc5_pos: 100.000 loss: 0.6514 (0.6811) acc1: 64.8438 (60.6511) acc5: 100.0000 (100.0000) acc1_neg: 61.198 acc5_neg: 100.000 time: 0.1906 data: 0.1000 max mem: 3532 +Test: [290/333] eta: 0:00:09 acc1_pos: 50.329 acc5_pos: 100.000 loss: 0.8345 (0.6866) acc1: 47.9167 (60.1733) acc5: 100.0000 (100.0000) acc1_neg: 52.083 acc5_neg: 100.000 time: 0.1838 data: 0.0925 max mem: 3532 +Test: [300/333] eta: 0:00:07 acc1_pos: 50.329 acc5_pos: 100.000 loss: 0.8484 (0.6921) acc1: 46.8750 (59.7730) acc5: 100.0000 (100.0000) acc1_neg: 47.656 acc5_neg: 100.000 time: 0.1625 data: 0.0719 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 50.329 acc5_pos: 100.000 loss: 0.7731 (0.6889) acc1: 52.3438 (60.0859) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.1395 data: 0.0499 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 50.329 acc5_pos: 100.000 loss: 0.6138 (0.6877) acc1: 66.6667 (60.2114) acc5: 100.0000 (100.0000) acc1_neg: 55.208 acc5_neg: 100.000 time: 0.1303 data: 0.0408 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 50.329 acc5_pos: 100.000 loss: 0.7138 (0.6906) acc1: 59.1146 (60.0319) acc5: 100.0000 (100.0000) acc1_neg: 59.115 acc5_neg: 100.000 time: 0.1254 data: 0.0361 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 50.329 acc5_pos: 100.000 loss: 0.7770 (0.6912) acc1: 55.4688 (59.9978) acc5: 100.0000 (100.0000) acc1_neg: 51.831 acc5_neg: 100.000 time: 0.1292 data: 0.0339 max mem: 3532 +Test: Total time: 0:01:09 (0.2095 s / it) +* Overall Acc@1 59.998 Acc@5 100.000 + target == real Acc@1 59.643 Acc@5 100.000 + target == fake Acc@1 60.313 Acc@5 100.000 + loss 0.691 + PR-AUC 0.6495 ROC-AUC 0.6414 EER 40.02% (thr=0.5024) +Accuracy of the network on the 127843 test images: 60.0% +[rank0]:[W725 18:54:59.215126276 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:55:18.800000 2581373 torch/distributed/run.py:766] +W0725 18:55:18.800000 2581373 torch/distributed/run.py:766] ***************************************** +W0725 18:55:18.800000 2581373 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:55:18.800000 2581373 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:55:47.231820281 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:55:47.437159076 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:55:48.730474189 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:55:48.773680104 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:55:48.962480235 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:55:48.994241567 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +[rank0]:[W725 18:55:48.007502183 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:55:48.014867841 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed3_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed3_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:27:48 acc1_pos: 95.833 acc5_pos: 100.000 loss: 0.2097 (0.2097) acc1: 95.8333 (95.8333) acc5: 100.0000 (100.0000) time: 7.1011 data: 5.0446 max mem: 3531 +Test: [ 10/235] eta: 0:02:44 acc1_pos: 94.271 acc5_pos: 100.000 loss: 0.2170 (0.2194) acc1: 95.8333 (95.6676) acc5: 100.0000 (100.0000) time: 0.7301 data: 0.4590 max mem: 3531 +Test: [ 20/235] eta: 0:01:32 acc1_pos: 95.573 acc5_pos: 100.000 loss: 0.2160 (0.2174) acc1: 95.8333 (95.6597) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0004 max mem: 3531 +Test: [ 30/235] eta: 0:01:11 acc1_pos: 95.833 acc5_pos: 100.000 loss: 0.2160 (0.2169) acc1: 95.8333 (95.6401) acc5: 100.0000 (100.0000) time: 0.1412 data: 0.0455 max mem: 3531 +Test: [ 40/235] eta: 0:01:02 acc1_pos: 95.573 acc5_pos: 100.000 loss: 0.2186 (0.2170) acc1: 95.5729 (95.6047) acc5: 100.0000 (100.0000) time: 0.2066 data: 0.1126 max mem: 3531 +Test: [ 50/235] eta: 0:00:54 acc1_pos: 94.531 acc5_pos: 100.000 loss: 0.2231 (0.2187) acc1: 95.0521 (95.4504) acc5: 100.0000 (100.0000) time: 0.2136 data: 0.1187 max mem: 3531 +Test: [ 60/235] eta: 0:00:48 acc1_pos: 94.792 acc5_pos: 100.000 loss: 0.2250 (0.2199) acc1: 94.7917 (95.3979) acc5: 100.0000 (100.0000) time: 0.1832 data: 0.0899 max mem: 3531 +Test: [ 70/235] eta: 0:00:43 acc1_pos: 98.462 acc5_pos: 100.000 loss: 0.2378 (0.3694) acc1: 94.0104 (85.9632) acc5: 100.0000 (100.0000) acc1_neg: 20.833 acc5_neg: 100.000 time: 0.1849 data: 0.0918 max mem: 3531 +Test: [ 80/235] eta: 0:00:39 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4085 (0.5010) acc1: 20.0521 (77.7842) acc5: 100.0000 (100.0000) acc1_neg: 17.708 acc5_neg: 100.000 time: 0.1949 data: 0.0989 max mem: 3531 +Test: [ 90/235] eta: 0:00:35 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4341 (0.6052) acc1: 19.2708 (71.3256) acc5: 100.0000 (100.0000) acc1_neg: 19.271 acc5_neg: 100.000 time: 0.1850 data: 0.0878 max mem: 3531 +Test: [100/235] eta: 0:00:32 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4452 (0.6858) acc1: 19.2708 (66.2515) acc5: 100.0000 (100.0000) acc1_neg: 19.010 acc5_neg: 100.000 time: 0.1912 data: 0.0967 max mem: 3531 +Test: [110/235] eta: 0:00:29 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4084 (0.7527) acc1: 19.7917 (62.0707) acc5: 100.0000 (100.0000) acc1_neg: 21.615 acc5_neg: 100.000 time: 0.1958 data: 0.1026 max mem: 3531 +Test: [120/235] eta: 0:00:26 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4334 (0.8100) acc1: 19.0104 (58.4560) acc5: 100.0000 (100.0000) acc1_neg: 20.833 acc5_neg: 100.000 time: 0.1897 data: 0.0960 max mem: 3531 +Test: [130/235] eta: 0:00:24 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4336 (0.8574) acc1: 18.2292 (55.3773) acc5: 100.0000 (100.0000) acc1_neg: 18.490 acc5_neg: 100.000 time: 0.1889 data: 0.0945 max mem: 3531 +Test: [140/235] eta: 0:00:21 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4578 (0.9010) acc1: 17.9688 (52.6873) acc5: 100.0000 (100.0000) acc1_neg: 18.750 acc5_neg: 100.000 time: 0.1839 data: 0.0876 max mem: 3531 +Test: [150/235] eta: 0:00:19 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4509 (0.9360) acc1: 17.9688 (50.4225) acc5: 100.0000 (100.0000) acc1_neg: 22.656 acc5_neg: 100.000 time: 0.1899 data: 0.0897 max mem: 3531 +Test: [160/235] eta: 0:00:16 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4312 (0.9669) acc1: 17.9688 (48.4100) acc5: 100.0000 (100.0000) acc1_neg: 18.750 acc5_neg: 100.000 time: 0.1849 data: 0.0848 max mem: 3531 +Test: [170/235] eta: 0:00:14 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4443 (0.9950) acc1: 18.4896 (46.7349) acc5: 100.0000 (100.0000) acc1_neg: 17.708 acc5_neg: 100.000 time: 0.1649 data: 0.0683 max mem: 3531 +Test: [180/235] eta: 0:00:11 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4275 (1.0195) acc1: 19.5312 (45.2967) acc5: 100.0000 (100.0000) acc1_neg: 19.271 acc5_neg: 100.000 time: 0.1699 data: 0.0734 max mem: 3531 +Test: [190/235] eta: 0:00:09 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4275 (1.0410) acc1: 21.0938 (43.9913) acc5: 100.0000 (100.0000) acc1_neg: 21.094 acc5_neg: 100.000 time: 0.1749 data: 0.0750 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4339 (1.0620) acc1: 19.2708 (42.7407) acc5: 100.0000 (100.0000) acc1_neg: 18.490 acc5_neg: 100.000 time: 0.1799 data: 0.0800 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4595 (1.0799) acc1: 18.4896 (41.6407) acc5: 100.0000 (100.0000) acc1_neg: 18.229 acc5_neg: 100.000 time: 0.1671 data: 0.0706 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4369 (1.0961) acc1: 19.7917 (40.6851) acc5: 100.0000 (100.0000) acc1_neg: 20.573 acc5_neg: 100.000 time: 0.1410 data: 0.0482 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4259 (1.1102) acc1: 20.5729 (39.8381) acc5: 100.0000 (100.0000) acc1_neg: 21.354 acc5_neg: 100.000 time: 0.1370 data: 0.0471 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4344 (1.1168) acc1: 20.5729 (39.5587) acc5: 100.0000 (100.0000) acc1_neg: 26.087 acc5_neg: 100.000 time: 0.1273 data: 0.0407 max mem: 3532 +Test: Total time: 0:00:47 (0.2014 s / it) +* Overall Acc@1 39.559 Acc@5 100.000 + target == real Acc@1 95.417 Acc@5 100.000 + target == fake Acc@1 19.356 Acc@5 100.000 + loss 1.117 + PR-AUC 0.8592 ROC-AUC 0.6996 EER 35.46% (thr=0.1746) +Accuracy of the network on the 89879 test images: 39.6% +[rank0]:[W725 18:56:45.852843653 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:57:04.149000 2581929 torch/distributed/run.py:766] +W0725 18:57:04.149000 2581929 torch/distributed/run.py:766] ***************************************** +W0725 18:57:04.149000 2581929 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:57:04.149000 2581929 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:57:30.395864251 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:57:31.662403326 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:57:31.751587599 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +[rank4]:[W725 18:57:31.052418555 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:57:31.059800034 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:57:31.174761475 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:57:31.232669330 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:57:31.236710228 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed3_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed3_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:23:27 acc1_pos: 42.188 acc5_pos: 100.000 loss: 0.7863 (0.7863) acc1: 42.1875 (42.1875) acc5: 100.0000 (100.0000) time: 5.2317 data: 3.4163 max mem: 3531 +Test: [ 10/269] eta: 0:02:30 acc1_pos: 66.927 acc5_pos: 100.000 loss: 0.6437 (0.6371) acc1: 62.7604 (62.9261) acc5: 100.0000 (100.0000) time: 0.5830 data: 0.3365 max mem: 3531 +Test: [ 20/269] eta: 0:01:36 acc1_pos: 72.135 acc5_pos: 100.000 loss: 0.6202 (0.6076) acc1: 64.5833 (66.7659) acc5: 100.0000 (100.0000) time: 0.1460 data: 0.0514 max mem: 3531 +Test: [ 30/269] eta: 0:01:29 acc1_pos: 76.823 acc5_pos: 100.000 loss: 0.5384 (0.5818) acc1: 73.4375 (69.7245) acc5: 100.0000 (100.0000) time: 0.2593 data: 0.1602 max mem: 3531 +Test: [ 40/269] eta: 0:01:19 acc1_pos: 77.586 acc5_pos: 100.000 loss: 0.4974 (0.5579) acc1: 77.8646 (72.7388) acc5: 100.0000 (100.0000) acc1_neg: 80.729 acc5_neg: 100.000 time: 0.3071 data: 0.2116 max mem: 3531 +Test: [ 50/269] eta: 0:01:10 acc1_pos: 77.586 acc5_pos: 100.000 loss: 0.5134 (0.5702) acc1: 75.5208 (71.0478) acc5: 100.0000 (100.0000) acc1_neg: 46.615 acc5_neg: 100.000 time: 0.2425 data: 0.1498 max mem: 3531 +Test: [ 60/269] eta: 0:00:59 acc1_pos: 77.586 acc5_pos: 100.000 loss: 0.6295 (0.5948) acc1: 60.1562 (68.3871) acc5: 100.0000 (100.0000) acc1_neg: 38.542 acc5_neg: 100.000 time: 0.1577 data: 0.0615 max mem: 3531 +Test: [ 70/269] eta: 0:00:52 acc1_pos: 77.586 acc5_pos: 100.000 loss: 0.7057 (0.6032) acc1: 57.0312 (67.3562) acc5: 100.0000 (100.0000) acc1_neg: 63.281 acc5_neg: 100.000 time: 0.1084 data: 0.0106 max mem: 3531 +Test: [ 80/269] eta: 0:00:47 acc1_pos: 77.586 acc5_pos: 100.000 loss: 0.6273 (0.6040) acc1: 61.9792 (67.2389) acc5: 100.0000 (100.0000) acc1_neg: 71.615 acc5_neg: 100.000 time: 0.1554 data: 0.0605 max mem: 3531 +Test: [ 90/269] eta: 0:00:44 acc1_pos: 77.586 acc5_pos: 100.000 loss: 0.6561 (0.6237) acc1: 59.3750 (64.9611) acc5: 100.0000 (100.0000) acc1_neg: 44.271 acc5_neg: 100.000 time: 0.1880 data: 0.0956 max mem: 3531 +Test: [100/269] eta: 0:00:40 acc1_pos: 77.586 acc5_pos: 100.000 loss: 0.8069 (0.6438) acc1: 44.2708 (63.0389) acc5: 100.0000 (100.0000) acc1_neg: 34.896 acc5_neg: 100.000 time: 0.1718 data: 0.0787 max mem: 3531 +Test: [110/269] eta: 0:00:37 acc1_pos: 77.586 acc5_pos: 100.000 loss: 0.8052 (0.6568) acc1: 49.4792 (61.7962) acc5: 100.0000 (100.0000) acc1_neg: 36.979 acc5_neg: 100.000 time: 0.1733 data: 0.0790 max mem: 3531 +Test: [120/269] eta: 0:00:33 acc1_pos: 77.586 acc5_pos: 100.000 loss: 0.6732 (0.6520) acc1: 53.6458 (62.2977) acc5: 100.0000 (100.0000) acc1_neg: 66.667 acc5_neg: 100.000 time: 0.1796 data: 0.0848 max mem: 3531 +Test: [130/269] eta: 0:00:31 acc1_pos: 77.586 acc5_pos: 100.000 loss: 0.6533 (0.6597) acc1: 64.5833 (61.7665) acc5: 100.0000 (100.0000) acc1_neg: 49.479 acc5_neg: 100.000 time: 0.1787 data: 0.0848 max mem: 3531 +Test: [140/269] eta: 0:00:28 acc1_pos: 77.586 acc5_pos: 100.000 loss: 0.7799 (0.6684) acc1: 49.4792 (60.7971) acc5: 100.0000 (100.0000) acc1_neg: 38.021 acc5_neg: 100.000 time: 0.1902 data: 0.0993 max mem: 3531 +Test: [150/269] eta: 0:00:26 acc1_pos: 77.586 acc5_pos: 100.000 loss: 0.7753 (0.6752) acc1: 48.1771 (59.9390) acc5: 100.0000 (100.0000) acc1_neg: 47.396 acc5_neg: 100.000 time: 0.2011 data: 0.1108 max mem: 3531 +Test: [160/269] eta: 0:00:23 acc1_pos: 77.586 acc5_pos: 100.000 loss: 0.7255 (0.6766) acc1: 51.5625 (59.7438) acc5: 100.0000 (100.0000) acc1_neg: 68.490 acc5_neg: 100.000 time: 0.1825 data: 0.0897 max mem: 3531 +Test: [170/269] eta: 0:00:21 acc1_pos: 77.586 acc5_pos: 100.000 loss: 0.6649 (0.6759) acc1: 59.8958 (59.8258) acc5: 100.0000 (100.0000) acc1_neg: 74.219 acc5_neg: 100.000 time: 0.1824 data: 0.0888 max mem: 3531 +Test: [180/269] eta: 0:00:19 acc1_pos: 77.586 acc5_pos: 100.000 loss: 0.5893 (0.6678) acc1: 67.4479 (60.7217) acc5: 100.0000 (100.0000) acc1_neg: 70.833 acc5_neg: 100.000 time: 0.2023 data: 0.1103 max mem: 3531 +Test: [190/269] eta: 0:00:16 acc1_pos: 77.586 acc5_pos: 100.000 loss: 0.5893 (0.6687) acc1: 65.6250 (60.5121) acc5: 100.0000 (100.0000) acc1_neg: 48.698 acc5_neg: 100.000 time: 0.1901 data: 0.0992 max mem: 3531 +Test: [200/269] eta: 0:00:14 acc1_pos: 77.586 acc5_pos: 100.000 loss: 0.7337 (0.6727) acc1: 52.8646 (60.1757) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.1795 data: 0.0882 max mem: 3531 +Test: [210/269] eta: 0:00:12 acc1_pos: 77.586 acc5_pos: 100.000 loss: 0.7156 (0.6740) acc1: 52.8646 (59.9674) acc5: 100.0000 (100.0000) acc1_neg: 51.823 acc5_neg: 100.000 time: 0.1971 data: 0.1044 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 77.586 acc5_pos: 100.000 loss: 0.6691 (0.6748) acc1: 54.6875 (59.8569) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.2282 data: 0.1359 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 77.586 acc5_pos: 100.000 loss: 0.6010 (0.6710) acc1: 63.8021 (60.1732) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.2149 data: 0.1203 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 77.586 acc5_pos: 100.000 loss: 0.6758 (0.6750) acc1: 58.8542 (59.9142) acc5: 100.0000 (100.0000) acc1_neg: 53.125 acc5_neg: 100.000 time: 0.1947 data: 0.0983 max mem: 3532 +Test: [250/269] eta: 0:00:03 acc1_pos: 77.586 acc5_pos: 100.000 loss: 0.7401 (0.6805) acc1: 53.1250 (59.4331) acc5: 100.0000 (100.0000) acc1_neg: 35.938 acc5_neg: 100.000 time: 0.1662 data: 0.0735 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 77.586 acc5_pos: 100.000 loss: 0.8453 (0.6885) acc1: 44.2708 (58.6935) acc5: 100.0000 (100.0000) acc1_neg: 63.021 acc5_neg: 100.000 time: 0.1303 data: 0.0394 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 77.586 acc5_pos: 100.000 loss: 0.7206 (0.6874) acc1: 44.2708 (58.7949) acc5: 100.0000 (100.0000) acc1_neg: 75.853 acc5_neg: 100.000 time: 0.1286 data: 0.0345 max mem: 3532 +Test: Total time: 0:00:54 (0.2042 s / it) +* Overall Acc@1 58.795 Acc@5 100.000 + target == real Acc@1 69.631 Acc@5 100.000 + target == fake Acc@1 57.316 Acc@5 100.000 + loss 0.687 + PR-AUC 0.9357 ROC-AUC 0.6845 EER 36.62% (thr=0.4641) +Accuracy of the network on the 103293 test images: 58.8% +[rank0]:[W725 18:58:35.952495779 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-25T18:58:41Z] eval_done tag=fixed500k_fake_n32_seed3_resampled +[2026-07-25T18:58:41Z] eval_start tag=fixed500k_real_n16_seed4_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed4_resampled_n111-046-081_20260725_112559/best_checkpoint.pth +started_utc=2026-07-25T18:58:41Z +hostname=n111-046-081 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed4_resampled_n111-046-081_20260725_112559/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed4_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0725 18:58:54.399000 2582514 torch/distributed/run.py:766] +W0725 18:58:54.399000 2582514 torch/distributed/run.py:766] ***************************************** +W0725 18:58:54.399000 2582514 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:58:54.399000 2582514 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:59:23.708866338 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:59:23.746768826 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +[rank3]:[W725 18:59:23.104300653 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:59:23.115076709 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:59:23.187799576 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:59:23.294123593 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:59:23.308610687 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:59:23.329170848 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed4_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed4_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:25:23 acc1_pos: 0.781 acc5_pos: 100.000 loss: 1.9188 (1.9188) acc1: 0.7812 (0.7812) acc5: 100.0000 (100.0000) time: 6.5128 data: 4.6763 max mem: 3531 +Test: [ 10/234] eta: 0:02:32 acc1_pos: 3.646 acc5_pos: 100.000 loss: 1.7188 (1.6006) acc1: 10.4167 (15.1042) acc5: 100.0000 (100.0000) time: 0.6807 data: 0.4260 max mem: 3531 +Test: [ 20/234] eta: 0:01:26 acc1_pos: 87.240 acc5_pos: 100.000 loss: 1.7188 (1.6004) acc1: 5.4688 (22.6811) acc5: 100.0000 (100.0000) time: 0.0987 data: 0.0006 max mem: 3531 +Test: [ 30/234] eta: 0:01:11 acc1_pos: 35.417 acc5_pos: 100.000 loss: 0.3746 (1.2266) acc1: 85.9375 (41.1626) acc5: 100.0000 (100.0000) time: 0.1707 data: 0.0752 max mem: 3531 +Test: [ 40/234] eta: 0:01:03 acc1_pos: 93.229 acc5_pos: 100.000 loss: 0.3318 (1.1207) acc1: 86.4583 (45.1156) acc5: 100.0000 (100.0000) time: 0.2499 data: 0.1560 max mem: 3531 +Test: [ 50/234] eta: 0:00:55 acc1_pos: 94.271 acc5_pos: 100.000 loss: 0.2379 (0.9447) acc1: 92.4479 (54.6518) acc5: 100.0000 (100.0000) time: 0.2292 data: 0.1334 max mem: 3531 +Test: [ 60/234] eta: 0:00:50 acc1_pos: 91.724 acc5_pos: 100.000 loss: 0.1315 (0.8016) acc1: 95.3125 (62.0347) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2060 data: 0.1126 max mem: 3531 +Test: [ 70/234] eta: 0:00:45 acc1_pos: 91.724 acc5_pos: 100.000 loss: 0.0656 (0.7355) acc1: 100.0000 (65.0565) acc5: 100.0000 (100.0000) acc1_neg: 73.698 acc5_neg: 100.000 time: 0.2148 data: 0.1232 max mem: 3531 +Test: [ 80/234] eta: 0:00:41 acc1_pos: 91.724 acc5_pos: 100.000 loss: 0.4344 (0.6830) acc1: 76.5625 (67.5283) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2148 data: 0.1233 max mem: 3531 +Test: [ 90/234] eta: 0:00:37 acc1_pos: 91.724 acc5_pos: 100.000 loss: 0.0923 (0.6200) acc1: 96.6146 (70.8476) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.2098 data: 0.1164 max mem: 3531 +Test: [100/234] eta: 0:00:34 acc1_pos: 91.724 acc5_pos: 100.000 loss: 0.1206 (0.5836) acc1: 96.6146 (72.7104) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2049 data: 0.1114 max mem: 3531 +Test: [110/234] eta: 0:00:31 acc1_pos: 91.724 acc5_pos: 100.000 loss: 0.2073 (0.5497) acc1: 89.5833 (74.5026) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.1958 data: 0.1043 max mem: 3531 +Test: [120/234] eta: 0:00:28 acc1_pos: 91.724 acc5_pos: 100.000 loss: 0.1894 (0.5219) acc1: 92.7083 (76.0309) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1998 data: 0.1085 max mem: 3531 +Test: [130/234] eta: 0:00:25 acc1_pos: 91.724 acc5_pos: 100.000 loss: 0.1551 (0.4943) acc1: 96.8750 (77.6121) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1991 data: 0.1064 max mem: 3531 +Test: [140/234] eta: 0:00:22 acc1_pos: 91.724 acc5_pos: 100.000 loss: 0.1676 (0.5024) acc1: 95.5729 (77.1886) acc5: 100.0000 (100.0000) acc1_neg: 41.927 acc5_neg: 100.000 time: 0.1992 data: 0.1051 max mem: 3531 +Test: [150/234] eta: 0:00:20 acc1_pos: 91.724 acc5_pos: 100.000 loss: 0.2634 (0.4932) acc1: 87.5000 (77.6456) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.2101 data: 0.1170 max mem: 3531 +Test: [160/234] eta: 0:00:17 acc1_pos: 91.724 acc5_pos: 100.000 loss: 0.2060 (0.4732) acc1: 94.0104 (78.7914) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2076 data: 0.1155 max mem: 3531 +Test: [170/234] eta: 0:00:15 acc1_pos: 91.724 acc5_pos: 100.000 loss: 0.2122 (0.4585) acc1: 94.0104 (79.6723) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2064 data: 0.1143 max mem: 3531 +Test: [180/234] eta: 0:00:12 acc1_pos: 91.724 acc5_pos: 100.000 loss: 0.2493 (0.4580) acc1: 92.1875 (79.5710) acc5: 100.0000 (100.0000) acc1_neg: 46.354 acc5_neg: 100.000 time: 0.1968 data: 0.1050 max mem: 3531 +Test: [190/234] eta: 0:00:10 acc1_pos: 91.724 acc5_pos: 100.000 loss: 0.4382 (0.4583) acc1: 80.4688 (79.5730) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.1859 data: 0.0951 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 91.724 acc5_pos: 100.000 loss: 0.4179 (0.4464) acc1: 81.5104 (80.2537) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1534 data: 0.0637 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 91.724 acc5_pos: 100.000 loss: 0.0950 (0.4297) acc1: 100.0000 (81.1858) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1277 data: 0.0383 max mem: 3532 +Test: [220/234] eta: 0:00:03 acc1_pos: 91.724 acc5_pos: 100.000 loss: 0.0936 (0.4148) acc1: 99.7396 (81.9971) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1300 data: 0.0406 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 91.724 acc5_pos: 100.000 loss: 0.1891 (0.4138) acc1: 83.8542 (81.9873) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.1278 data: 0.0384 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 91.724 acc5_pos: 100.000 loss: 0.3773 (0.4135) acc1: 83.3333 (81.9757) acc5: 100.0000 (100.0000) acc1_neg: 85.992 acc5_neg: 100.000 time: 0.1263 data: 0.0360 max mem: 3532 +Test: Total time: 0:00:49 (0.2116 s / it) +* Overall Acc@1 81.976 Acc@5 100.000 + target == real Acc@1 54.924 Acc@5 100.000 + target == fake Acc@1 89.600 Acc@5 100.000 + loss 0.413 + PR-AUC 0.9425 ROC-AUC 0.8317 EER 26.20% (thr=0.8252) +Accuracy of the network on the 89729 test images: 82.0% +[rank0]:[W725 19:00:20.175261292 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 19:00:39.789000 2583087 torch/distributed/run.py:766] +W0725 19:00:39.789000 2583087 torch/distributed/run.py:766] ***************************************** +W0725 19:00:39.789000 2583087 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 19:00:39.789000 2583087 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W725 19:01:07.395597953 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 19:01:07.460632096 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 19:01:07.525106442 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 19:01:08.939422971 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 19:01:08.964394400 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 2): env:// +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 19:01:08.031928901 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W725 19:01:08.031930587 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 19:01:08.036864549 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed4_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed4_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:39:38 acc1_pos: 1.042 acc5_pos: 100.000 loss: 2.1250 (2.1250) acc1: 1.0417 (1.0417) acc5: 100.0000 (100.0000) time: 7.0291 data: 5.4199 max mem: 3531 +Test: [ 10/2387] eta: 0:28:36 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.1415 (2.1365) acc1: 1.8229 (3.6458) acc5: 100.0000 (100.0000) time: 0.7223 data: 0.4931 max mem: 3531 +Test: [ 20/2387] eta: 0:16:48 acc1_pos: 0.260 acc5_pos: 100.000 loss: 2.1582 (2.1605) acc1: 1.8229 (2.7902) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:15:01 acc1_pos: 1.302 acc5_pos: 100.000 loss: 2.2389 (2.1928) acc1: 0.7812 (2.4950) acc5: 100.0000 (100.0000) time: 0.1957 data: 0.0997 max mem: 3531 +Test: [ 40/2387] eta: 0:14:15 acc1_pos: 2.604 acc5_pos: 100.000 loss: 2.2362 (2.1866) acc1: 0.7812 (2.3501) acc5: 100.0000 (100.0000) time: 0.2999 data: 0.2050 max mem: 3531 +Test: [ 50/2387] eta: 0:14:05 acc1_pos: 1.562 acc5_pos: 100.000 loss: 2.1930 (2.1780) acc1: 1.0417 (2.2518) acc5: 100.0000 (100.0000) time: 0.3291 data: 0.2314 max mem: 3531 +Test: [ 60/2387] eta: 0:14:04 acc1_pos: 10.677 acc5_pos: 100.000 loss: 2.2085 (2.1845) acc1: 1.0417 (2.3181) acc5: 100.0000 (100.0000) time: 0.3599 data: 0.2635 max mem: 3531 +Test: [ 70/2387] eta: 0:13:24 acc1_pos: 9.115 acc5_pos: 100.000 loss: 2.2719 (2.2035) acc1: 0.5208 (2.2997) acc5: 100.0000 (100.0000) time: 0.3099 data: 0.2136 max mem: 3531 +Test: [ 80/2387] eta: 0:13:10 acc1_pos: 1.302 acc5_pos: 100.000 loss: 2.2998 (2.2148) acc1: 0.5208 (2.0930) acc5: 100.0000 (100.0000) time: 0.2801 data: 0.1843 max mem: 3531 +Test: [ 90/2387] eta: 0:12:38 acc1_pos: 1.823 acc5_pos: 100.000 loss: 2.2595 (2.2209) acc1: 0.2604 (2.0862) acc5: 100.0000 (100.0000) time: 0.2700 data: 0.1761 max mem: 3531 +Test: [ 100/2387] eta: 0:12:17 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.2224 (2.2242) acc1: 0.2604 (2.0086) acc5: 100.0000 (100.0000) time: 0.2410 data: 0.1457 max mem: 3531 +Test: [ 110/2387] eta: 0:11:56 acc1_pos: 3.646 acc5_pos: 100.000 loss: 2.2105 (2.2250) acc1: 1.0417 (1.9566) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1466 max mem: 3531 +Test: [ 120/2387] eta: 0:11:37 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.2006 (2.2227) acc1: 1.3021 (1.9219) acc5: 100.0000 (100.0000) time: 0.2337 data: 0.1367 max mem: 3531 +Test: [ 130/2387] eta: 0:11:22 acc1_pos: 3.646 acc5_pos: 100.000 loss: 2.2372 (2.2264) acc1: 0.5208 (1.8567) acc5: 100.0000 (100.0000) time: 0.2353 data: 0.1413 max mem: 3531 +Test: [ 140/2387] eta: 0:11:05 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.2736 (2.2276) acc1: 0.2604 (1.7564) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1311 max mem: 3531 +Test: [ 150/2387] eta: 0:10:55 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.2444 (2.2281) acc1: 0.2604 (1.7695) acc5: 100.0000 (100.0000) time: 0.2295 data: 0.1330 max mem: 3531 +Test: [ 160/2387] eta: 0:10:40 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.2203 (2.1716) acc1: 1.8229 (4.3219) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2299 data: 0.1322 max mem: 3531 +Test: [ 170/2387] eta: 0:10:22 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0829 (2.0493) acc1: 99.7396 (9.9111) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0948 max mem: 3531 +Test: [ 180/2387] eta: 0:10:10 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0810 (1.9406) acc1: 100.0000 (14.8855) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0906 max mem: 3531 +Test: [ 190/2387] eta: 0:09:58 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0800 (1.8431) acc1: 100.0000 (19.3390) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1999 data: 0.1051 max mem: 3531 +Test: [ 200/2387] eta: 0:09:43 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0800 (1.7555) acc1: 100.0000 (23.3494) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0835 max mem: 3531 +Test: [ 210/2387] eta: 0:09:31 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0803 (1.6761) acc1: 100.0000 (26.9784) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0726 max mem: 3532 +Test: [ 220/2387] eta: 0:09:24 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0801 (1.6038) acc1: 100.0000 (30.2802) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1949 data: 0.0988 max mem: 3532 +Test: [ 230/2387] eta: 0:09:13 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0799 (1.5378) acc1: 100.0000 (33.2984) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.1003 max mem: 3532 +Test: [ 240/2387] eta: 0:09:02 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0774 (1.4772) acc1: 100.0000 (36.0661) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0748 max mem: 3532 +Test: [ 250/2387] eta: 0:08:54 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0733 (1.4209) acc1: 100.0000 (38.6133) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0706 max mem: 3532 +Test: [ 260/2387] eta: 0:08:48 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0621 (1.3688) acc1: 100.0000 (40.9652) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.0966 max mem: 3532 +Test: [ 270/2387] eta: 0:08:41 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0612 (1.3206) acc1: 100.0000 (43.1436) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2025 data: 0.1068 max mem: 3532 +Test: [ 280/2387] eta: 0:08:27 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0618 (1.2758) acc1: 100.0000 (45.1670) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1449 data: 0.0493 max mem: 3532 +Test: [ 290/2387] eta: 0:08:20 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0611 (1.2340) acc1: 100.0000 (47.0513) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1332 data: 0.0387 max mem: 3532 +Test: [ 300/2387] eta: 0:08:16 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0600 (1.1950) acc1: 100.0000 (48.8104) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.1006 max mem: 3532 +Test: [ 310/2387] eta: 0:08:14 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0600 (1.1591) acc1: 100.0000 (50.4538) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2301 data: 0.1353 max mem: 3532 +Test: [ 320/2387] eta: 0:08:11 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0942 (1.1259) acc1: 100.0000 (51.9973) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2353 data: 0.1419 max mem: 3532 +Test: [ 330/2387] eta: 0:08:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0957 (1.0948) acc1: 100.0000 (53.4468) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1747 data: 0.0823 max mem: 3532 +Test: [ 340/2387] eta: 0:07:53 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0949 (1.0655) acc1: 100.0000 (54.8120) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1294 data: 0.0351 max mem: 3532 +Test: [ 350/2387] eta: 0:07:48 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0914 (1.0370) acc1: 100.0000 (56.0994) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1593 data: 0.0664 max mem: 3532 +Test: [ 360/2387] eta: 0:07:43 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0724 (1.0103) acc1: 100.0000 (57.3148) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0925 max mem: 3532 +Test: [ 370/2387] eta: 0:07:38 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0766 (0.9852) acc1: 100.0000 (58.4653) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1800 data: 0.0857 max mem: 3532 +Test: [ 380/2387] eta: 0:07:33 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0777 (0.9614) acc1: 100.0000 (59.5548) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0793 max mem: 3532 +Test: [ 390/2387] eta: 0:07:28 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0774 (0.9388) acc1: 100.0000 (60.5878) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1795 data: 0.0833 max mem: 3532 +Test: [ 400/2387] eta: 0:07:25 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0774 (0.9173) acc1: 100.0000 (61.5694) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1910 data: 0.0962 max mem: 3532 +Test: [ 410/2387] eta: 0:07:21 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0773 (0.8968) acc1: 100.0000 (62.5044) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1950 data: 0.1016 max mem: 3532 +Test: [ 420/2387] eta: 0:07:18 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0766 (0.8774) acc1: 100.0000 (63.3938) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1982 data: 0.1050 max mem: 3532 +Test: [ 430/2387] eta: 0:07:16 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0755 (0.8587) acc1: 100.0000 (64.2420) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2163 data: 0.1239 max mem: 3532 +Test: [ 440/2387] eta: 0:07:10 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0730 (0.8408) acc1: 100.0000 (65.0528) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1776 data: 0.0847 max mem: 3533 +Test: [ 450/2387] eta: 0:07:06 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0642 (0.8236) acc1: 100.0000 (65.8277) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1636 data: 0.0705 max mem: 3533 +Test: [ 460/2387] eta: 0:07:02 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0639 (0.8071) acc1: 100.0000 (66.5689) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1827 data: 0.0884 max mem: 3533 +Test: [ 470/2387] eta: 0:07:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0639 (0.7913) acc1: 100.0000 (67.2787) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1969 data: 0.1034 max mem: 3533 +Test: [ 480/2387] eta: 0:06:57 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0643 (0.7762) acc1: 100.0000 (67.9590) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2082 data: 0.1181 max mem: 3533 +Test: [ 490/2387] eta: 0:06:54 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0641 (0.7617) acc1: 100.0000 (68.6116) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1928 data: 0.1005 max mem: 3533 +Test: [ 500/2387] eta: 0:06:51 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0641 (0.7478) acc1: 100.0000 (69.2381) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1980 data: 0.1048 max mem: 3533 +Test: [ 510/2387] eta: 0:06:48 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0644 (0.7344) acc1: 100.0000 (69.8401) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1999 data: 0.1056 max mem: 3533 +Test: [ 520/2387] eta: 0:06:44 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0660 (0.7216) acc1: 100.0000 (70.4190) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1873 data: 0.0926 max mem: 3533 +Test: [ 530/2387] eta: 0:06:41 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0682 (0.7093) acc1: 100.0000 (70.9760) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1815 data: 0.0890 max mem: 3533 +Test: [ 540/2387] eta: 0:06:38 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0684 (0.6975) acc1: 100.0000 (71.5125) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1812 data: 0.0880 max mem: 3533 +Test: [ 550/2387] eta: 0:06:34 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0686 (0.6861) acc1: 100.0000 (72.0295) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1797 data: 0.0870 max mem: 3533 +Test: [ 560/2387] eta: 0:06:31 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0686 (0.6751) acc1: 100.0000 (72.5281) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1862 data: 0.0935 max mem: 3533 +Test: [ 570/2387] eta: 0:06:29 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0685 (0.6644) acc1: 100.0000 (73.0092) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1913 data: 0.0977 max mem: 3533 +Test: [ 580/2387] eta: 0:06:25 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0685 (0.6542) acc1: 100.0000 (73.4738) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1896 data: 0.0958 max mem: 3533 +Test: [ 590/2387] eta: 0:06:23 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0685 (0.6443) acc1: 100.0000 (73.9226) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1937 data: 0.1002 max mem: 3533 +Test: [ 600/2387] eta: 0:06:20 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0676 (0.6347) acc1: 100.0000 (74.3565) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1903 data: 0.0972 max mem: 3533 +Test: [ 610/2387] eta: 0:06:17 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0675 (0.6254) acc1: 100.0000 (74.7762) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1891 data: 0.0971 max mem: 3533 +Test: [ 620/2387] eta: 0:06:15 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0679 (0.6164) acc1: 100.0000 (75.1824) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1996 data: 0.1080 max mem: 3533 +Test: [ 630/2387] eta: 0:06:12 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0681 (0.6077) acc1: 100.0000 (75.5757) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2015 data: 0.1094 max mem: 3533 +Test: [ 640/2387] eta: 0:06:09 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0684 (0.5993) acc1: 100.0000 (75.9568) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1810 data: 0.0899 max mem: 3533 +Test: [ 650/2387] eta: 0:06:06 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0680 (0.5911) acc1: 100.0000 (76.3261) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1739 data: 0.0831 max mem: 3533 +Test: [ 660/2387] eta: 0:06:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0681 (0.5832) acc1: 100.0000 (76.6842) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1888 data: 0.0955 max mem: 3533 +Test: [ 670/2387] eta: 0:06:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0677 (0.5755) acc1: 100.0000 (77.0317) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1835 data: 0.0901 max mem: 3534 +Test: [ 680/2387] eta: 0:05:58 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0672 (0.5680) acc1: 100.0000 (77.3690) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1888 data: 0.0964 max mem: 3534 +Test: [ 690/2387] eta: 0:05:56 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0675 (0.5622) acc1: 100.0000 (77.6245) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1957 data: 0.1028 max mem: 3534 +Test: [ 700/2387] eta: 0:05:53 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2004 (0.5571) acc1: 93.2292 (77.8501) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1866 data: 0.0909 max mem: 3534 +Test: [ 710/2387] eta: 0:05:51 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2005 (0.5521) acc1: 93.7500 (78.0726) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2090 data: 0.1119 max mem: 3534 +Test: [ 720/2387] eta: 0:05:48 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1924 (0.5472) acc1: 93.7500 (78.2893) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1886 data: 0.0938 max mem: 3534 +Test: [ 730/2387] eta: 0:05:45 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2004 (0.5425) acc1: 93.4896 (78.4944) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1703 data: 0.0756 max mem: 3534 +Test: [ 740/2387] eta: 0:05:43 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2028 (0.5379) acc1: 93.2292 (78.7007) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2067 data: 0.1142 max mem: 3534 +Test: [ 750/2387] eta: 0:05:40 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1974 (0.5334) acc1: 93.7500 (78.9000) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1845 data: 0.0929 max mem: 3534 +Test: [ 760/2387] eta: 0:05:38 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1917 (0.5284) acc1: 94.2708 (79.1222) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1698 data: 0.0777 max mem: 3534 +Test: [ 770/2387] eta: 0:05:35 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1009 (0.5228) acc1: 98.4375 (79.3801) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1948 data: 0.1019 max mem: 3534 +Test: [ 780/2387] eta: 0:05:33 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0756 (0.5170) acc1: 100.0000 (79.6425) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1985 data: 0.1061 max mem: 3534 +Test: [ 790/2387] eta: 0:05:31 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0696 (0.5116) acc1: 100.0000 (79.8817) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2024 data: 0.1118 max mem: 3534 +Test: [ 800/2387] eta: 0:05:29 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0689 (0.5061) acc1: 100.0000 (80.1329) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2091 data: 0.1162 max mem: 3534 +Test: [ 810/2387] eta: 0:05:27 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0705 (0.5008) acc1: 100.0000 (80.3753) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2083 data: 0.1142 max mem: 3534 +Test: [ 820/2387] eta: 0:05:24 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0766 (0.4956) acc1: 100.0000 (80.6124) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0920 max mem: 3534 +Test: [ 830/2387] eta: 0:05:21 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0758 (0.4907) acc1: 100.0000 (80.8398) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1707 data: 0.0788 max mem: 3534 +Test: [ 840/2387] eta: 0:05:19 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0738 (0.4857) acc1: 100.0000 (81.0664) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1850 data: 0.0941 max mem: 3534 +Test: [ 850/2387] eta: 0:05:16 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0738 (0.4809) acc1: 100.0000 (81.2889) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1798 data: 0.0874 max mem: 3534 +Test: [ 860/2387] eta: 0:05:13 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0714 (0.4761) acc1: 100.0000 (81.5062) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1677 data: 0.0726 max mem: 3534 +Test: [ 870/2387] eta: 0:05:11 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0712 (0.4715) acc1: 100.0000 (81.7185) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1734 data: 0.0775 max mem: 3534 +Test: [ 880/2387] eta: 0:05:09 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0715 (0.4669) acc1: 100.0000 (81.9260) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1870 data: 0.0901 max mem: 3534 +Test: [ 890/2387] eta: 0:05:06 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0674 (0.4624) acc1: 100.0000 (82.1289) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0933 max mem: 3534 +Test: [ 900/2387] eta: 0:05:04 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0685 (0.4615) acc1: 100.0000 (82.1287) acc5: 100.0000 (100.0000) acc1_neg: 78.646 acc5_neg: 100.000 time: 0.1799 data: 0.0865 max mem: 3535 +Test: [ 910/2387] eta: 0:05:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.4494 (0.4614) acc1: 78.3854 (82.0864) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.2252 data: 0.1324 max mem: 3535 +Test: [ 920/2387] eta: 0:05:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.4556 (0.4614) acc1: 77.3438 (82.0392) acc5: 100.0000 (100.0000) acc1_neg: 75.781 acc5_neg: 100.000 time: 0.2111 data: 0.1167 max mem: 3535 +Test: [ 930/2387] eta: 0:04:57 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.4556 (0.4613) acc1: 77.6042 (81.9988) acc5: 100.0000 (100.0000) acc1_neg: 76.302 acc5_neg: 100.000 time: 0.1348 data: 0.0411 max mem: 3535 +Test: [ 940/2387] eta: 0:04:54 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.4538 (0.4612) acc1: 78.6458 (81.9590) acc5: 100.0000 (100.0000) acc1_neg: 76.302 acc5_neg: 100.000 time: 0.1489 data: 0.0541 max mem: 3535 +Test: [ 950/2387] eta: 0:04:52 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.4538 (0.4610) acc1: 78.3854 (81.9228) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.1763 data: 0.0819 max mem: 3535 +Test: [ 960/2387] eta: 0:04:49 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.4382 (0.4608) acc1: 79.1667 (81.8996) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.1793 data: 0.0870 max mem: 3535 +Test: [ 970/2387] eta: 0:04:47 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.4420 (0.4607) acc1: 79.1667 (81.8567) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.1700 data: 0.0781 max mem: 3535 +Test: [ 980/2387] eta: 0:04:44 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.4143 (0.4581) acc1: 80.2083 (81.9681) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1716 data: 0.0778 max mem: 3535 +Test: [ 990/2387] eta: 0:04:42 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0751 (0.4542) acc1: 100.0000 (82.1487) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1743 data: 0.0791 max mem: 3535 +Test: [1000/2387] eta: 0:04:39 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0691 (0.4505) acc1: 100.0000 (82.3187) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1692 data: 0.0757 max mem: 3535 +Test: [1010/2387] eta: 0:04:37 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0706 (0.4468) acc1: 100.0000 (82.4923) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1756 data: 0.0829 max mem: 3535 +Test: [1020/2387] eta: 0:04:35 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0733 (0.4432) acc1: 100.0000 (82.6559) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1810 data: 0.0884 max mem: 3535 +Test: [1030/2387] eta: 0:04:32 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0715 (0.4396) acc1: 100.0000 (82.8241) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1792 data: 0.0877 max mem: 3535 +Test: [1040/2387] eta: 0:04:30 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0698 (0.4361) acc1: 100.0000 (82.9889) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1767 data: 0.0854 max mem: 3535 +Test: [1050/2387] eta: 0:04:28 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0672 (0.4326) acc1: 100.0000 (83.1505) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1753 data: 0.0836 max mem: 3535 +Test: [1060/2387] eta: 0:04:25 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0680 (0.4292) acc1: 100.0000 (83.3090) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1764 data: 0.0854 max mem: 3535 +Test: [1070/2387] eta: 0:04:23 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0705 (0.4259) acc1: 100.0000 (83.4644) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1739 data: 0.0827 max mem: 3535 +Test: [1080/2387] eta: 0:04:21 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0750 (0.4227) acc1: 100.0000 (83.6162) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1723 data: 0.0809 max mem: 3535 +Test: [1090/2387] eta: 0:04:19 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0841 (0.4196) acc1: 100.0000 (83.7656) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1791 data: 0.0872 max mem: 3535 +Test: [1100/2387] eta: 0:04:16 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0849 (0.4166) acc1: 100.0000 (83.9131) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1772 data: 0.0856 max mem: 3535 +Test: [1110/2387] eta: 0:04:14 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0852 (0.4138) acc1: 100.0000 (84.0497) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1867 data: 0.0940 max mem: 3535 +Test: [1120/2387] eta: 0:04:12 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0827 (0.4108) acc1: 100.0000 (84.1901) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1949 data: 0.1024 max mem: 3536 +Test: [1130/2387] eta: 0:04:10 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0827 (0.4080) acc1: 100.0000 (84.3285) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1831 data: 0.0919 max mem: 3536 +Test: [1140/2387] eta: 0:04:08 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0838 (0.4052) acc1: 100.0000 (84.4649) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1775 data: 0.0866 max mem: 3536 +Test: [1150/2387] eta: 0:04:06 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0792 (0.4024) acc1: 100.0000 (84.5994) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0891 max mem: 3536 +Test: [1160/2387] eta: 0:04:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0718 (0.3995) acc1: 100.0000 (84.7321) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1828 data: 0.0917 max mem: 3536 +Test: [1170/2387] eta: 0:04:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0699 (0.3967) acc1: 100.0000 (84.8625) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1855 data: 0.0945 max mem: 3536 +Test: [1180/2387] eta: 0:03:59 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0699 (0.3939) acc1: 100.0000 (84.9907) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1911 data: 0.0971 max mem: 3536 +Test: [1190/2387] eta: 0:03:57 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0679 (0.3912) acc1: 100.0000 (85.1167) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1892 data: 0.0951 max mem: 3536 +Test: [1200/2387] eta: 0:03:55 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0667 (0.3885) acc1: 100.0000 (85.2406) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0927 max mem: 3536 +Test: [1210/2387] eta: 0:03:53 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0621 (0.3858) acc1: 100.0000 (85.3625) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1953 data: 0.1021 max mem: 3536 +Test: [1220/2387] eta: 0:03:51 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0617 (0.3832) acc1: 100.0000 (85.4824) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2008 data: 0.1079 max mem: 3536 +Test: [1230/2387] eta: 0:03:49 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0609 (0.3805) acc1: 100.0000 (85.6003) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1945 data: 0.1027 max mem: 3536 +Test: [1240/2387] eta: 0:03:47 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0609 (0.3780) acc1: 100.0000 (85.7163) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1943 data: 0.1032 max mem: 3536 +Test: [1250/2387] eta: 0:03:45 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0612 (0.3754) acc1: 100.0000 (85.8305) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2049 data: 0.1128 max mem: 3536 +Test: [1260/2387] eta: 0:03:43 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0604 (0.3729) acc1: 100.0000 (85.9429) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2058 data: 0.1134 max mem: 3536 +Test: [1270/2387] eta: 0:03:41 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0593 (0.3704) acc1: 100.0000 (86.0535) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2014 data: 0.1101 max mem: 3536 +Test: [1280/2387] eta: 0:03:39 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0597 (0.3681) acc1: 100.0000 (86.1621) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2042 data: 0.1138 max mem: 3536 +Test: [1290/2387] eta: 0:03:37 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0651 (0.3658) acc1: 100.0000 (86.2659) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2004 data: 0.1101 max mem: 3536 +Test: [1300/2387] eta: 0:03:35 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0657 (0.3635) acc1: 100.0000 (86.3701) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1933 data: 0.1031 max mem: 3536 +Test: [1310/2387] eta: 0:03:33 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0661 (0.3613) acc1: 100.0000 (86.4714) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1850 data: 0.0927 max mem: 3536 +Test: [1320/2387] eta: 0:03:31 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0705 (0.3591) acc1: 100.0000 (86.5711) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1895 data: 0.0969 max mem: 3536 +Test: [1330/2387] eta: 0:03:29 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0706 (0.3570) acc1: 100.0000 (86.6712) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1927 data: 0.0979 max mem: 3536 +Test: [1340/2387] eta: 0:03:27 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0743 (0.3553) acc1: 100.0000 (86.7510) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1984 data: 0.1035 max mem: 3536 +Test: [1350/2387] eta: 0:03:25 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0532 (0.3530) acc1: 100.0000 (86.8491) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2028 data: 0.1107 max mem: 3537 +Test: [1360/2387] eta: 0:03:23 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0519 (0.3508) acc1: 100.0000 (86.9457) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1973 data: 0.1046 max mem: 3537 +Test: [1370/2387] eta: 0:03:21 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0517 (0.3486) acc1: 100.0000 (87.0409) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1897 data: 0.0971 max mem: 3537 +Test: [1380/2387] eta: 0:03:19 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0518 (0.3465) acc1: 100.0000 (87.1347) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1863 data: 0.0926 max mem: 3537 +Test: [1390/2387] eta: 0:03:17 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0511 (0.3443) acc1: 100.0000 (87.2272) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1973 data: 0.1049 max mem: 3537 +Test: [1400/2387] eta: 0:03:15 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0490 (0.3422) acc1: 100.0000 (87.3184) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2088 data: 0.1185 max mem: 3537 +Test: [1410/2387] eta: 0:03:13 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0496 (0.3403) acc1: 100.0000 (87.4074) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2067 data: 0.1162 max mem: 3537 +Test: [1420/2387] eta: 0:03:11 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0789 (0.3386) acc1: 100.0000 (87.4894) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2095 data: 0.1191 max mem: 3537 +Test: [1430/2387] eta: 0:03:09 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0794 (0.3368) acc1: 100.0000 (87.5755) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2080 data: 0.1162 max mem: 3537 +Test: [1440/2387] eta: 0:03:07 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0801 (0.3350) acc1: 100.0000 (87.6614) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1835 data: 0.0907 max mem: 3537 +Test: [1450/2387] eta: 0:03:05 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0790 (0.3333) acc1: 100.0000 (87.7461) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1712 data: 0.0784 max mem: 3537 +Test: [1460/2387] eta: 0:03:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0713 (0.3315) acc1: 100.0000 (87.8296) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1756 data: 0.0803 max mem: 3537 +Test: [1470/2387] eta: 0:03:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0654 (0.3297) acc1: 100.0000 (87.9123) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1701 data: 0.0750 max mem: 3537 +Test: [1480/2387] eta: 0:02:59 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0643 (0.3279) acc1: 100.0000 (87.9939) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1728 data: 0.0802 max mem: 3537 +Test: [1490/2387] eta: 0:02:57 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0656 (0.3261) acc1: 100.0000 (88.0745) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1834 data: 0.0906 max mem: 3537 +Test: [1500/2387] eta: 0:02:55 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0650 (0.3244) acc1: 100.0000 (88.1536) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1878 data: 0.0958 max mem: 3537 +Test: [1510/2387] eta: 0:02:53 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0641 (0.3227) acc1: 100.0000 (88.2320) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1958 data: 0.1041 max mem: 3537 +Test: [1520/2387] eta: 0:02:51 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0657 (0.3210) acc1: 100.0000 (88.3076) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1807 data: 0.0878 max mem: 3537 +Test: [1530/2387] eta: 0:02:48 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0709 (0.3194) acc1: 100.0000 (88.3836) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1719 data: 0.0795 max mem: 3537 +Test: [1540/2387] eta: 0:02:46 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0691 (0.3178) acc1: 100.0000 (88.4590) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1854 data: 0.0944 max mem: 3537 +Test: [1550/2387] eta: 0:02:44 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0686 (0.3161) acc1: 100.0000 (88.5334) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1688 data: 0.0777 max mem: 3537 +Test: [1560/2387] eta: 0:02:42 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0678 (0.3146) acc1: 100.0000 (88.6061) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1770 data: 0.0851 max mem: 3537 +Test: [1570/2387] eta: 0:02:40 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0659 (0.3130) acc1: 100.0000 (88.6786) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1795 data: 0.0861 max mem: 3537 +Test: [1580/2387] eta: 0:02:38 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0661 (0.3115) acc1: 100.0000 (88.7494) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1649 data: 0.0688 max mem: 3538 +Test: [1590/2387] eta: 0:02:36 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0720 (0.3099) acc1: 100.0000 (88.8198) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1998 data: 0.1045 max mem: 3538 +Test: [1600/2387] eta: 0:02:34 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0740 (0.3086) acc1: 100.0000 (88.8839) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1852 data: 0.0908 max mem: 3538 +Test: [1610/2387] eta: 0:02:32 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0706 (0.3071) acc1: 100.0000 (88.9529) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1348 data: 0.0398 max mem: 3538 +Test: [1620/2387] eta: 0:02:30 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0742 (0.3058) acc1: 100.0000 (89.0188) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1467 data: 0.0529 max mem: 3538 +Test: [1630/2387] eta: 0:02:28 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0871 (0.3044) acc1: 100.0000 (89.0818) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1788 data: 0.0865 max mem: 3538 +Test: [1640/2387] eta: 0:02:26 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0809 (0.3031) acc1: 100.0000 (89.1464) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1778 data: 0.0837 max mem: 3538 +Test: [1650/2387] eta: 0:02:23 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0806 (0.3017) acc1: 100.0000 (89.2092) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1567 data: 0.0589 max mem: 3538 +Test: [1660/2387] eta: 0:02:21 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0761 (0.3003) acc1: 100.0000 (89.2735) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1501 data: 0.0558 max mem: 3538 +Test: [1670/2387] eta: 0:02:19 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0760 (0.2990) acc1: 100.0000 (89.3351) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1981 data: 0.1068 max mem: 3538 +Test: [1680/2387] eta: 0:02:17 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0756 (0.2977) acc1: 100.0000 (89.3976) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1946 data: 0.1006 max mem: 3538 +Test: [1690/2387] eta: 0:02:15 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0778 (0.2965) acc1: 100.0000 (89.4583) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1785 data: 0.0822 max mem: 3538 +Test: [1700/2387] eta: 0:02:14 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0739 (0.2951) acc1: 100.0000 (89.5201) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2178 data: 0.1238 max mem: 3538 +Test: [1710/2387] eta: 0:02:12 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0653 (0.2938) acc1: 100.0000 (89.5814) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1968 data: 0.1053 max mem: 3538 +Test: [1720/2387] eta: 0:02:10 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0622 (0.2924) acc1: 100.0000 (89.6419) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1825 data: 0.0913 max mem: 3538 +Test: [1730/2387] eta: 0:02:08 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0625 (0.2911) acc1: 100.0000 (89.7014) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2093 data: 0.1151 max mem: 3538 +Test: [1740/2387] eta: 0:02:06 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0629 (0.2898) acc1: 100.0000 (89.7606) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2163 data: 0.1217 max mem: 3538 +Test: [1750/2387] eta: 0:02:04 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0628 (0.2885) acc1: 100.0000 (89.8191) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2049 data: 0.1135 max mem: 3538 +Test: [1760/2387] eta: 0:02:02 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0609 (0.2872) acc1: 100.0000 (89.8769) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1778 data: 0.0863 max mem: 3538 +Test: [1770/2387] eta: 0:02:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0609 (0.2859) acc1: 100.0000 (89.9336) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1672 data: 0.0759 max mem: 3538 +Test: [1780/2387] eta: 0:01:58 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0627 (0.2847) acc1: 100.0000 (89.9901) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1872 data: 0.0960 max mem: 3538 +Test: [1790/2387] eta: 0:01:56 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0606 (0.2834) acc1: 100.0000 (90.0460) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1957 data: 0.1047 max mem: 3538 +Test: [1800/2387] eta: 0:01:54 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0610 (0.2822) acc1: 100.0000 (90.1013) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1946 data: 0.1031 max mem: 3538 +Test: [1810/2387] eta: 0:01:52 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0615 (0.2810) acc1: 100.0000 (90.1559) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1787 data: 0.0878 max mem: 3539 +Test: [1820/2387] eta: 0:01:50 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0588 (0.2798) acc1: 100.0000 (90.2100) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1653 data: 0.0749 max mem: 3539 +Test: [1830/2387] eta: 0:01:48 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0603 (0.2786) acc1: 100.0000 (90.2635) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1701 data: 0.0789 max mem: 3539 +Test: [1840/2387] eta: 0:01:46 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0686 (0.2774) acc1: 100.0000 (90.3163) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1679 data: 0.0767 max mem: 3539 +Test: [1850/2387] eta: 0:01:44 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0693 (0.2763) acc1: 100.0000 (90.3687) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1657 data: 0.0731 max mem: 3539 +Test: [1860/2387] eta: 0:01:42 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0694 (0.2752) acc1: 100.0000 (90.4204) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1696 data: 0.0757 max mem: 3539 +Test: [1870/2387] eta: 0:01:40 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0691 (0.2741) acc1: 100.0000 (90.4716) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1748 data: 0.0814 max mem: 3539 +Test: [1880/2387] eta: 0:01:38 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0691 (0.2730) acc1: 100.0000 (90.5223) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1816 data: 0.0894 max mem: 3539 +Test: [1890/2387] eta: 0:01:36 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0692 (0.2719) acc1: 100.0000 (90.5724) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1840 data: 0.0936 max mem: 3539 +Test: [1900/2387] eta: 0:01:34 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0695 (0.2709) acc1: 100.0000 (90.6220) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1782 data: 0.0865 max mem: 3539 +Test: [1910/2387] eta: 0:01:32 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0704 (0.2698) acc1: 100.0000 (90.6711) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2087 data: 0.1146 max mem: 3539 +Test: [1920/2387] eta: 0:01:30 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0711 (0.2688) acc1: 100.0000 (90.7196) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2027 data: 0.1076 max mem: 3539 +Test: [1930/2387] eta: 0:01:28 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0670 (0.2677) acc1: 100.0000 (90.7677) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1642 data: 0.0708 max mem: 3539 +Test: [1940/2387] eta: 0:01:26 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0670 (0.2669) acc1: 100.0000 (90.8120) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1692 data: 0.0769 max mem: 3539 +Test: [1950/2387] eta: 0:01:24 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0672 (0.2659) acc1: 100.0000 (90.8591) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1859 data: 0.0940 max mem: 3539 +Test: [1960/2387] eta: 0:01:22 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0706 (0.2650) acc1: 100.0000 (90.8999) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1879 data: 0.0966 max mem: 3539 +Test: [1970/2387] eta: 0:01:20 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0843 (0.2641) acc1: 100.0000 (90.9420) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.2048 data: 0.1125 max mem: 3539 +Test: [1980/2387] eta: 0:01:18 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0844 (0.2632) acc1: 100.0000 (90.9865) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2071 data: 0.1153 max mem: 3539 +Test: [1990/2387] eta: 0:01:16 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0847 (0.2623) acc1: 99.7396 (91.0294) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1990 data: 0.1043 max mem: 3539 +Test: [2000/2387] eta: 0:01:14 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0882 (0.2615) acc1: 99.7396 (91.0726) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1939 data: 0.0992 max mem: 3539 +Test: [2010/2387] eta: 0:01:13 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0882 (0.2606) acc1: 99.7396 (91.1148) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1831 data: 0.0917 max mem: 3539 +Test: [2020/2387] eta: 0:01:11 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0860 (0.2598) acc1: 99.7396 (91.1570) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1736 data: 0.0814 max mem: 3539 +Test: [2030/2387] eta: 0:01:09 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0866 (0.2589) acc1: 99.4792 (91.1983) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1628 data: 0.0700 max mem: 3540 +Test: [2040/2387] eta: 0:01:07 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0869 (0.2581) acc1: 99.4792 (91.2397) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1721 data: 0.0805 max mem: 3540 +Test: [2050/2387] eta: 0:01:05 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0852 (0.2572) acc1: 99.7396 (91.2812) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1749 data: 0.0821 max mem: 3540 +Test: [2060/2387] eta: 0:01:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0795 (0.2564) acc1: 99.7396 (91.3220) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1777 data: 0.0843 max mem: 3540 +Test: [2070/2387] eta: 0:01:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0770 (0.2555) acc1: 99.7396 (91.3632) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2006 data: 0.1094 max mem: 3540 +Test: [2080/2387] eta: 0:00:59 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0770 (0.2546) acc1: 100.0000 (91.4034) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2049 data: 0.1138 max mem: 3540 +Test: [2090/2387] eta: 0:00:57 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0769 (0.2538) acc1: 99.7396 (91.4439) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1540 data: 0.0600 max mem: 3540 +Test: [2100/2387] eta: 0:00:55 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0769 (0.2530) acc1: 99.7396 (91.4840) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2131 data: 0.1181 max mem: 3540 +Test: [2110/2387] eta: 0:00:53 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0780 (0.2521) acc1: 99.7396 (91.5230) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2455 data: 0.1538 max mem: 3540 +Test: [2120/2387] eta: 0:00:51 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0787 (0.2513) acc1: 99.7396 (91.5614) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1616 data: 0.0700 max mem: 3540 +Test: [2130/2387] eta: 0:00:49 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0761 (0.2505) acc1: 100.0000 (91.6010) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1602 data: 0.0681 max mem: 3540 +Test: [2140/2387] eta: 0:00:47 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0749 (0.2498) acc1: 100.0000 (91.6321) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1767 data: 0.0857 max mem: 3540 +Test: [2150/2387] eta: 0:00:45 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0737 (0.2490) acc1: 100.0000 (91.6709) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1790 data: 0.0881 max mem: 3540 +Test: [2160/2387] eta: 0:00:43 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0805 (0.2483) acc1: 100.0000 (91.7081) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1846 data: 0.0940 max mem: 3540 +Test: [2170/2387] eta: 0:00:41 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0833 (0.2475) acc1: 100.0000 (91.7457) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1801 data: 0.0881 max mem: 3540 +Test: [2180/2387] eta: 0:00:39 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0882 (0.2485) acc1: 99.4792 (91.6786) acc5: 100.0000 (100.0000) acc1_neg: 42.969 acc5_neg: 100.000 time: 0.1789 data: 0.0857 max mem: 3540 +Test: [2190/2387] eta: 0:00:37 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1267 (0.2484) acc1: 97.9167 (91.6734) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1704 data: 0.0781 max mem: 3540 +Test: [2200/2387] eta: 0:00:36 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0702 (0.2476) acc1: 100.0000 (91.7113) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1664 data: 0.0743 max mem: 3540 +Test: [2210/2387] eta: 0:00:34 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0611 (0.2468) acc1: 100.0000 (91.7488) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1653 data: 0.0728 max mem: 3540 +Test: [2220/2387] eta: 0:00:32 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0632 (0.2459) acc1: 100.0000 (91.7858) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1648 data: 0.0733 max mem: 3540 +Test: [2230/2387] eta: 0:00:30 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0637 (0.2451) acc1: 100.0000 (91.8226) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1533 data: 0.0621 max mem: 3540 +Test: [2240/2387] eta: 0:00:28 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0616 (0.2443) acc1: 100.0000 (91.8590) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1431 data: 0.0512 max mem: 3540 +Test: [2250/2387] eta: 0:00:26 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0613 (0.2435) acc1: 100.0000 (91.8950) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1509 data: 0.0602 max mem: 3540 +Test: [2260/2387] eta: 0:00:24 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0633 (0.2427) acc1: 100.0000 (91.9309) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1515 data: 0.0611 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0655 (0.2419) acc1: 100.0000 (91.9664) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1457 data: 0.0551 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0653 (0.2413) acc1: 100.0000 (91.9914) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1425 data: 0.0523 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0642 (0.2405) acc1: 100.0000 (92.0263) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1417 data: 0.0511 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0656 (0.2398) acc1: 100.0000 (92.0602) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1409 data: 0.0505 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0705 (0.2390) acc1: 100.0000 (92.0941) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1415 data: 0.0517 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0703 (0.2383) acc1: 100.0000 (92.1276) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1363 data: 0.0466 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0745 (0.2377) acc1: 100.0000 (92.1606) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1314 data: 0.0418 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0721 (0.2370) acc1: 100.0000 (92.1940) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1311 data: 0.0413 max mem: 3541 +Test: [2350/2387] eta: 0:00:07 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0736 (0.2364) acc1: 100.0000 (92.2164) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1294 data: 0.0394 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0757 (0.2358) acc1: 100.0000 (92.2492) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1246 data: 0.0349 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0800 (0.2352) acc1: 100.0000 (92.2800) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1197 data: 0.0302 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0832 (0.2345) acc1: 100.0000 (92.3123) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1153 data: 0.0260 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0854 (0.2342) acc1: 100.0000 (92.3271) acc5: 100.0000 (100.0000) acc1_neg: 97.674 acc5_neg: 100.000 time: 0.1048 data: 0.0181 max mem: 3541 +Test: Total time: 0:07:29 (0.1885 s / it) +* Overall Acc@1 92.327 Acc@5 100.000 + target == real Acc@1 1.750 Acc@5 100.000 + target == fake Acc@1 98.688 Acc@5 100.000 + loss 0.234 + PR-AUC 0.9825 ROC-AUC 0.8092 EER 24.96% (thr=0.9229) +Accuracy of the network on the 916267 test images: 92.3% +[rank0]:[W725 19:08:51.344281752 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 19:09:10.970000 2583763 torch/distributed/run.py:766] +W0725 19:09:10.970000 2583763 torch/distributed/run.py:766] ***************************************** +W0725 19:09:10.970000 2583763 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 19:09:10.970000 2583763 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 19:09:39.897172770 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 19:09:39.295024565 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 19:09:39.565960302 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 19:09:40.644501550 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 19:09:40.685826753 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 19:09:40.788096482 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 19:09:40.814884571 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 19:09:40.871131765 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed4_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed4_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 0:58:21 acc1_pos: 1.042 acc5_pos: 100.000 loss: 2.1250 (2.1250) acc1: 1.0417 (1.0417) acc5: 100.0000 (100.0000) time: 6.7720 data: 5.0943 max mem: 3531 +Test: [ 10/517] eta: 0:05:53 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.1415 (2.1365) acc1: 1.8229 (3.6458) acc5: 100.0000 (100.0000) time: 0.6975 data: 0.4635 max mem: 3531 +Test: [ 20/517] eta: 0:03:25 acc1_pos: 0.260 acc5_pos: 100.000 loss: 2.1582 (2.1605) acc1: 1.8229 (2.7902) acc5: 100.0000 (100.0000) time: 0.0945 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:02:57 acc1_pos: 1.302 acc5_pos: 100.000 loss: 2.2389 (2.1928) acc1: 0.7812 (2.4950) acc5: 100.0000 (100.0000) time: 0.1818 data: 0.0848 max mem: 3531 +Test: [ 40/517] eta: 0:02:51 acc1_pos: 2.604 acc5_pos: 100.000 loss: 2.2362 (2.1866) acc1: 0.7812 (2.3501) acc5: 100.0000 (100.0000) time: 0.3049 data: 0.2106 max mem: 3531 +Test: [ 50/517] eta: 0:02:48 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.0790 (2.0782) acc1: 1.3021 (6.9649) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.3528 data: 0.2579 max mem: 3531 +Test: [ 60/517] eta: 0:02:33 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1191 (1.7564) acc1: 99.4792 (22.1824) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2866 data: 0.1903 max mem: 3531 +Test: [ 70/517] eta: 0:02:25 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1171 (1.5257) acc1: 100.0000 (33.1279) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2346 data: 0.1388 max mem: 3531 +Test: [ 80/517] eta: 0:02:18 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1187 (1.3526) acc1: 100.0000 (41.3709) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2546 data: 0.1585 max mem: 3531 +Test: [ 90/517] eta: 0:02:11 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1019 (1.2146) acc1: 100.0000 (47.8136) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2515 data: 0.1562 max mem: 3531 +Test: [100/517] eta: 0:02:06 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0941 (1.1036) acc1: 100.0000 (52.9806) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2485 data: 0.1548 max mem: 3531 +Test: [110/517] eta: 0:02:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0934 (1.0126) acc1: 100.0000 (57.2166) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2483 data: 0.1524 max mem: 3531 +Test: [120/517] eta: 0:01:56 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0949 (0.9372) acc1: 100.0000 (60.7503) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2461 data: 0.1487 max mem: 3531 +Test: [130/517] eta: 0:01:52 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0972 (0.8733) acc1: 100.0000 (63.7405) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2450 data: 0.1497 max mem: 3531 +Test: [140/517] eta: 0:01:47 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0944 (0.8180) acc1: 100.0000 (66.3121) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2451 data: 0.1498 max mem: 3531 +Test: [150/517] eta: 0:01:44 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0931 (0.7701) acc1: 100.0000 (68.5430) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2462 data: 0.1511 max mem: 3531 +Test: [160/517] eta: 0:01:40 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0932 (0.7280) acc1: 100.0000 (70.4969) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2535 data: 0.1566 max mem: 3531 +Test: [170/517] eta: 0:01:37 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0903 (0.6906) acc1: 100.0000 (72.2222) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2507 data: 0.1519 max mem: 3531 +Test: [180/517] eta: 0:01:33 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0830 (0.6566) acc1: 100.0000 (73.7569) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2499 data: 0.1519 max mem: 3531 +Test: [190/517] eta: 0:01:30 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0853 (0.6271) acc1: 100.0000 (75.1309) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2476 data: 0.1500 max mem: 3531 +Test: [200/517] eta: 0:01:27 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0913 (0.6003) acc1: 100.0000 (76.3682) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2412 data: 0.1445 max mem: 3531 +Test: [210/517] eta: 0:01:24 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0997 (0.5769) acc1: 100.0000 (77.4882) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2555 data: 0.1602 max mem: 3532 +Test: [220/517] eta: 0:01:21 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1085 (0.5558) acc1: 100.0000 (78.5068) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2586 data: 0.1639 max mem: 3532 +Test: [230/517] eta: 0:01:18 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1153 (0.5368) acc1: 100.0000 (79.4350) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2497 data: 0.1537 max mem: 3532 +Test: [240/517] eta: 0:01:15 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1162 (0.5196) acc1: 100.0000 (80.2851) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2499 data: 0.1530 max mem: 3532 +Test: [250/517] eta: 0:01:12 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1106 (0.5032) acc1: 100.0000 (81.0684) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2463 data: 0.1494 max mem: 3532 +Test: [260/517] eta: 0:01:09 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0975 (0.4879) acc1: 100.0000 (81.7908) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2471 data: 0.1520 max mem: 3532 +Test: [270/517] eta: 0:01:06 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1230 (0.4752) acc1: 100.0000 (82.4550) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2442 data: 0.1513 max mem: 3532 +Test: [280/517] eta: 0:01:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1430 (0.4624) acc1: 99.7396 (83.0748) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2478 data: 0.1530 max mem: 3532 +Test: [290/517] eta: 0:01:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1037 (0.4500) acc1: 100.0000 (83.6546) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2458 data: 0.1488 max mem: 3532 +Test: [300/517] eta: 0:00:57 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0899 (0.4381) acc1: 100.0000 (84.1950) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2411 data: 0.1466 max mem: 3532 +Test: [310/517] eta: 0:00:55 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0888 (0.4269) acc1: 100.0000 (84.7032) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2498 data: 0.1565 max mem: 3532 +Test: [320/517] eta: 0:00:52 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0924 (0.4166) acc1: 100.0000 (85.1798) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2498 data: 0.1564 max mem: 3532 +Test: [330/517] eta: 0:00:49 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0937 (0.4067) acc1: 100.0000 (85.6275) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2548 data: 0.1612 max mem: 3532 +Test: [340/517] eta: 0:00:46 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0929 (0.3979) acc1: 100.0000 (86.0284) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2538 data: 0.1601 max mem: 3532 +Test: [350/517] eta: 0:00:44 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0861 (0.3895) acc1: 100.0000 (86.4190) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2482 data: 0.1521 max mem: 3532 +Test: [360/517] eta: 0:00:41 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0935 (0.3813) acc1: 100.0000 (86.7945) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2450 data: 0.1472 max mem: 3532 +Test: [370/517] eta: 0:00:38 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0946 (0.3735) acc1: 100.0000 (87.1504) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2400 data: 0.1435 max mem: 3532 +Test: [380/517] eta: 0:00:35 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0946 (0.3662) acc1: 100.0000 (87.4877) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2448 data: 0.1453 max mem: 3532 +Test: [390/517] eta: 0:00:33 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1045 (0.3599) acc1: 100.0000 (87.8024) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2452 data: 0.1456 max mem: 3532 +Test: [400/517] eta: 0:00:30 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1088 (0.3532) acc1: 100.0000 (88.1066) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2417 data: 0.1455 max mem: 3532 +Test: [410/517] eta: 0:00:27 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0993 (0.3471) acc1: 100.0000 (88.3959) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2447 data: 0.1502 max mem: 3532 +Test: [420/517] eta: 0:00:25 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0985 (0.3415) acc1: 100.0000 (88.6703) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2436 data: 0.1489 max mem: 3532 +Test: [430/517] eta: 0:00:22 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1207 (0.3369) acc1: 100.0000 (88.9235) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2447 data: 0.1474 max mem: 3532 +Test: [440/517] eta: 0:00:20 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1375 (0.3321) acc1: 99.7396 (89.1605) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2462 data: 0.1488 max mem: 3533 +Test: [450/517] eta: 0:00:17 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1063 (0.3271) acc1: 100.0000 (89.4009) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2499 data: 0.1537 max mem: 3533 +Test: [460/517] eta: 0:00:14 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1100 (0.3227) acc1: 100.0000 (89.6212) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2494 data: 0.1531 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1027 (0.3179) acc1: 100.0000 (89.8360) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2455 data: 0.1502 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0988 (0.3135) acc1: 100.0000 (90.0473) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2443 data: 0.1497 max mem: 3533 +Test: [490/517] eta: 0:00:06 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1095 (0.3093) acc1: 100.0000 (90.2500) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2340 data: 0.1388 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1084 (0.3055) acc1: 100.0000 (90.4446) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2067 data: 0.1143 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1034 (0.3016) acc1: 100.0000 (90.6316) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1783 data: 0.0882 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0855 (0.2990) acc1: 100.0000 (90.7356) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1828 data: 0.0882 max mem: 3533 +Test: Total time: 0:02:11 (0.2537 s / it) +* Overall Acc@1 90.736 Acc@5 100.000 + target == real Acc@1 2.306 Acc@5 100.000 + target == fake Acc@1 99.909 Acc@5 100.000 + loss 0.299 + PR-AUC 0.9275 ROC-AUC 0.6177 EER 41.78% (thr=0.9077) +Accuracy of the network on the 198427 test images: 90.7% +[rank0]:[W725 19:12:00.245873432 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0725 19:12:19.944000 2584350 torch/distributed/run.py:766] +W0725 19:12:19.944000 2584350 torch/distributed/run.py:766] ***************************************** +W0725 19:12:19.944000 2584350 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 19:12:19.944000 2584350 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 19:12:48.810212027 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 19:12:48.881793841 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 19:12:48.933769506 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 19:12:48.209935701 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +[rank2]:[W725 19:12:48.455778999 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 19:12:48.485881142 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 19:12:48.507805516 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 19:12:48.547809256 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed4_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed4_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:45:25 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.0084 (2.0084) acc1: 0.0000 (0.0000) acc5: 100.0000 (100.0000) time: 7.0059 data: 4.8343 max mem: 3531 +Test: [ 10/389] eta: 0:04:35 acc1_pos: 10.938 acc5_pos: 100.000 loss: 2.1438 (2.1051) acc1: 0.5208 (1.7992) acc5: 100.0000 (100.0000) time: 0.7274 data: 0.4398 max mem: 3531 +Test: [ 20/389] eta: 0:02:38 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.2060 (2.2164) acc1: 0.0000 (1.7361) acc5: 100.0000 (100.0000) time: 0.0997 data: 0.0003 max mem: 3531 +Test: [ 30/389] eta: 0:01:57 acc1_pos: 0.260 acc5_pos: 100.000 loss: 2.3564 (2.2648) acc1: 0.0000 (1.2433) acc5: 100.0000 (100.0000) time: 0.1066 data: 0.0105 max mem: 3531 +Test: [ 40/389] eta: 0:01:40 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.3311 (2.2624) acc1: 0.0000 (1.1560) acc5: 100.0000 (100.0000) time: 0.1399 data: 0.0474 max mem: 3531 +Test: [ 50/389] eta: 0:01:28 acc1_pos: 1.042 acc5_pos: 100.000 loss: 2.2569 (2.2557) acc1: 0.0000 (0.9906) acc5: 100.0000 (100.0000) time: 0.1583 data: 0.0643 max mem: 3531 +Test: [ 60/389] eta: 0:01:18 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.0382 (2.2376) acc1: 0.0000 (0.8922) acc5: 100.0000 (100.0000) time: 0.1399 data: 0.0452 max mem: 3531 +Test: [ 70/389] eta: 0:01:12 acc1_pos: 3.125 acc5_pos: 100.000 loss: 2.0104 (2.1946) acc1: 0.0000 (1.0673) acc5: 100.0000 (100.0000) time: 0.1399 data: 0.0433 max mem: 3531 +Test: [ 80/389] eta: 0:01:07 acc1_pos: 4.948 acc5_pos: 100.000 loss: 1.9998 (2.1729) acc1: 0.0000 (2.5784) acc5: 100.0000 (100.0000) time: 0.1599 data: 0.0615 max mem: 3531 +Test: [ 90/389] eta: 0:01:05 acc1_pos: 7.292 acc5_pos: 100.000 loss: 1.8096 (2.1204) acc1: 1.0417 (3.2910) acc5: 100.0000 (100.0000) time: 0.1874 data: 0.0927 max mem: 3531 +Test: [100/389] eta: 0:01:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.7444 (2.0981) acc1: 2.3438 (3.3442) acc5: 100.0000 (100.0000) time: 0.1933 data: 0.1019 max mem: 3531 +Test: [110/389] eta: 0:00:58 acc1_pos: 53.906 acc5_pos: 100.000 loss: 1.7627 (2.0669) acc1: 2.0833 (5.0371) acc5: 100.0000 (100.0000) time: 0.1808 data: 0.0904 max mem: 3531 +Test: [120/389] eta: 0:00:55 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.7551 (2.0345) acc1: 6.5104 (6.0348) acc5: 100.0000 (100.0000) time: 0.1646 data: 0.0732 max mem: 3531 +Test: [130/389] eta: 0:00:53 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.0478 (1.9300) acc1: 33.8542 (11.2973) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1673 data: 0.0758 max mem: 3531 +Test: [140/389] eta: 0:00:50 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1146 (1.8018) acc1: 100.0000 (17.5403) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1771 data: 0.0870 max mem: 3531 +Test: [150/389] eta: 0:00:47 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0943 (1.6941) acc1: 100.0000 (22.7477) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1703 data: 0.0803 max mem: 3531 +Test: [160/389] eta: 0:00:45 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0943 (1.6018) acc1: 100.0000 (27.0203) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1796 data: 0.0888 max mem: 3531 +Test: [170/389] eta: 0:00:43 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0873 (1.5138) acc1: 100.0000 (31.2820) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1687 data: 0.0760 max mem: 3531 +Test: [180/389] eta: 0:00:40 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0956 (1.4387) acc1: 100.0000 (34.8915) acc5: 100.0000 (100.0000) acc1_neg: 66.146 acc5_neg: 100.000 time: 0.1511 data: 0.0585 max mem: 3531 +Test: [190/389] eta: 0:00:38 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1025 (1.3754) acc1: 100.0000 (37.9608) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1572 data: 0.0662 max mem: 3531 +Test: [200/389] eta: 0:00:35 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0971 (1.3128) acc1: 100.0000 (41.0137) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1551 data: 0.0634 max mem: 3531 +Test: [210/389] eta: 0:00:33 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0884 (1.2546) acc1: 100.0000 (43.8092) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1662 data: 0.0737 max mem: 3532 +Test: [220/389] eta: 0:00:32 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0818 (1.2024) acc1: 100.0000 (46.3353) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1850 data: 0.0932 max mem: 3532 +Test: [230/389] eta: 0:00:29 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1068 (1.1575) acc1: 100.0000 (48.5739) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.1647 data: 0.0738 max mem: 3532 +Test: [240/389] eta: 0:00:27 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1407 (1.1149) acc1: 100.0000 (50.7024) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1411 data: 0.0505 max mem: 3532 +Test: [250/389] eta: 0:00:25 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1083 (1.0748) acc1: 100.0000 (52.6664) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1442 data: 0.0524 max mem: 3532 +Test: [260/389] eta: 0:00:23 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0939 (1.0413) acc1: 100.0000 (54.1876) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1473 data: 0.0554 max mem: 3532 +Test: [270/389] eta: 0:00:21 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0977 (1.0092) acc1: 99.7396 (55.8137) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1451 data: 0.0544 max mem: 3532 +Test: [280/389] eta: 0:00:19 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1060 (0.9794) acc1: 100.0000 (57.2286) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1489 data: 0.0573 max mem: 3532 +Test: [290/389] eta: 0:00:17 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1206 (0.9515) acc1: 100.0000 (58.6045) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1522 data: 0.0601 max mem: 3532 +Test: [300/389] eta: 0:00:15 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1247 (0.9244) acc1: 100.0000 (59.9625) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1463 data: 0.0552 max mem: 3532 +Test: [310/389] eta: 0:00:13 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1466 (0.9007) acc1: 100.0000 (61.1862) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1419 data: 0.0511 max mem: 3532 +Test: [320/389] eta: 0:00:12 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1454 (0.8763) acc1: 99.7396 (62.3848) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1502 data: 0.0591 max mem: 3532 +Test: [330/389] eta: 0:00:10 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1009 (0.8532) acc1: 100.0000 (63.5070) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1541 data: 0.0624 max mem: 3532 +Test: [340/389] eta: 0:00:08 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0999 (0.8309) acc1: 100.0000 (64.5474) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1516 data: 0.0594 max mem: 3532 +Test: [350/389] eta: 0:00:06 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0962 (0.8143) acc1: 100.0000 (65.3416) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1505 data: 0.0581 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1158 (0.7956) acc1: 99.7396 (66.2887) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1443 data: 0.0518 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1156 (0.7785) acc1: 100.0000 (67.1861) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1327 data: 0.0416 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0939 (0.7610) acc1: 100.0000 (68.0426) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1252 data: 0.0353 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0926 (0.7480) acc1: 100.0000 (68.6532) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1310 data: 0.0371 max mem: 3532 +Test: Total time: 0:01:06 (0.1705 s / it) +* Overall Acc@1 68.653 Acc@5 100.000 + target == real Acc@1 5.916 Acc@5 100.000 + target == fake Acc@1 97.815 Acc@5 100.000 + loss 0.748 + PR-AUC 0.7239 ROC-AUC 0.5954 EER 41.98% (thr=0.9019) +Accuracy of the network on the 149358 test images: 68.7% +[rank0]:[W725 19:14:08.377542553 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 19:14:27.363000 2584919 torch/distributed/run.py:766] +W0725 19:14:27.363000 2584919 torch/distributed/run.py:766] ***************************************** +W0725 19:14:27.363000 2584919 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 19:14:27.363000 2584919 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 19:14:55.952502623 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W725 19:14:55.385630549 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank4]:[W725 19:14:55.397970067 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +NCCL version 2.26.2+cuda12.2 +[rank1]:[W725 19:14:56.617117637 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +| distributed init (rank 7): env:// +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 19:14:56.795091359 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 19:14:56.800998446 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank6]:[W725 19:14:56.801080002 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 19:14:56.828544362 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed4_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed4_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:30:13 acc1_pos: 78.906 acc5_pos: 100.000 loss: 0.4045 (0.4045) acc1: 78.9062 (78.9062) acc5: 100.0000 (100.0000) time: 5.4456 data: 4.0567 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/333] eta: 0:03:06 acc1_pos: 32.292 acc5_pos: 100.000 loss: 0.4045 (0.7122) acc1: 80.7292 (64.2045) acc5: 100.0000 (100.0000) time: 0.5767 data: 0.3691 max mem: 3531 +Test: [ 20/333] eta: 0:01:47 acc1_pos: 29.948 acc5_pos: 100.000 loss: 1.2779 (1.0440) acc1: 30.4688 (46.2674) acc5: 100.0000 (100.0000) time: 0.0900 data: 0.0003 max mem: 3531 +Test: [ 30/333] eta: 0:01:20 acc1_pos: 11.719 acc5_pos: 100.000 loss: 1.5339 (1.2338) acc1: 19.2708 (36.5339) acc5: 100.0000 (100.0000) time: 0.0951 data: 0.0003 max mem: 3531 +Test: [ 40/333] eta: 0:01:10 acc1_pos: 11.979 acc5_pos: 100.000 loss: 1.6253 (1.3299) acc1: 14.5833 (31.4787) acc5: 100.0000 (100.0000) time: 0.1299 data: 0.0329 max mem: 3531 +Test: [ 50/333] eta: 0:01:04 acc1_pos: 18.229 acc5_pos: 100.000 loss: 1.6318 (1.3974) acc1: 13.0208 (27.9769) acc5: 100.0000 (100.0000) time: 0.1669 data: 0.0718 max mem: 3531 +Test: [ 60/333] eta: 0:00:59 acc1_pos: 16.406 acc5_pos: 100.000 loss: 1.6446 (1.4321) acc1: 13.0208 (25.9648) acc5: 100.0000 (100.0000) time: 0.1699 data: 0.0765 max mem: 3531 +Test: [ 70/333] eta: 0:00:54 acc1_pos: 13.542 acc5_pos: 100.000 loss: 1.6446 (1.4601) acc1: 13.5417 (24.7616) acc5: 100.0000 (100.0000) time: 0.1581 data: 0.0671 max mem: 3531 +Test: [ 80/333] eta: 0:00:51 acc1_pos: 19.271 acc5_pos: 100.000 loss: 1.5949 (1.4807) acc1: 13.8021 (23.6208) acc5: 100.0000 (100.0000) time: 0.1649 data: 0.0720 max mem: 3531 +Test: [ 90/333] eta: 0:00:48 acc1_pos: 19.531 acc5_pos: 100.000 loss: 1.5455 (1.4772) acc1: 18.7500 (23.2916) acc5: 100.0000 (100.0000) time: 0.1698 data: 0.0754 max mem: 3531 +Test: [100/333] eta: 0:00:45 acc1_pos: 22.135 acc5_pos: 100.000 loss: 1.3929 (1.4728) acc1: 20.0521 (23.1178) acc5: 100.0000 (100.0000) time: 0.1655 data: 0.0699 max mem: 3531 +Test: [110/333] eta: 0:00:42 acc1_pos: 26.562 acc5_pos: 100.000 loss: 1.4189 (1.4690) acc1: 21.6146 (23.2615) acc5: 100.0000 (100.0000) time: 0.1604 data: 0.0654 max mem: 3531 +Test: [120/333] eta: 0:00:40 acc1_pos: 21.875 acc5_pos: 100.000 loss: 1.4239 (1.4653) acc1: 21.8750 (23.1319) acc5: 100.0000 (100.0000) time: 0.1557 data: 0.0631 max mem: 3531 +Test: [130/333] eta: 0:00:38 acc1_pos: 33.333 acc5_pos: 100.000 loss: 1.3770 (1.4569) acc1: 22.3958 (23.5190) acc5: 100.0000 (100.0000) time: 0.1594 data: 0.0666 max mem: 3531 +Test: [140/333] eta: 0:00:35 acc1_pos: 19.792 acc5_pos: 100.000 loss: 1.3468 (1.4546) acc1: 24.4792 (23.5631) acc5: 100.0000 (100.0000) time: 0.1538 data: 0.0605 max mem: 3531 +Test: [150/333] eta: 0:00:33 acc1_pos: 20.833 acc5_pos: 100.000 loss: 1.3980 (1.4529) acc1: 22.6562 (23.6772) acc5: 100.0000 (100.0000) time: 0.1503 data: 0.0547 max mem: 3531 +Test: [160/333] eta: 0:00:31 acc1_pos: 17.434 acc5_pos: 100.000 loss: 1.3485 (1.4259) acc1: 26.5625 (25.1828) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.1555 data: 0.0602 max mem: 3531 +Test: [170/333] eta: 0:00:29 acc1_pos: 17.434 acc5_pos: 100.000 loss: 0.4641 (1.3663) acc1: 78.3854 (28.5468) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1496 data: 0.0553 max mem: 3531 +Test: [180/333] eta: 0:00:27 acc1_pos: 17.434 acc5_pos: 100.000 loss: 0.3722 (1.3100) acc1: 84.3750 (31.7435) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1492 data: 0.0524 max mem: 3531 +Test: [190/333] eta: 0:00:25 acc1_pos: 17.434 acc5_pos: 100.000 loss: 0.3153 (1.2571) acc1: 87.5000 (34.7486) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.1599 data: 0.0640 max mem: 3531 +Test: [200/333] eta: 0:00:23 acc1_pos: 17.434 acc5_pos: 100.000 loss: 0.3041 (1.2113) acc1: 87.5000 (37.3497) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1557 data: 0.0618 max mem: 3531 +Test: [210/333] eta: 0:00:21 acc1_pos: 17.434 acc5_pos: 100.000 loss: 0.3479 (1.1717) acc1: 85.6771 (39.5969) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.1549 data: 0.0596 max mem: 3532 +Test: [220/333] eta: 0:00:19 acc1_pos: 17.434 acc5_pos: 100.000 loss: 0.3657 (1.1386) acc1: 84.6354 (41.4887) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1560 data: 0.0599 max mem: 3532 +Test: [230/333] eta: 0:00:17 acc1_pos: 17.434 acc5_pos: 100.000 loss: 0.2652 (1.0997) acc1: 91.4062 (43.7421) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1649 data: 0.0723 max mem: 3532 +Test: [240/333] eta: 0:00:16 acc1_pos: 17.434 acc5_pos: 100.000 loss: 0.2477 (1.0648) acc1: 92.1875 (45.7696) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1780 data: 0.0846 max mem: 3532 +Test: [250/333] eta: 0:00:14 acc1_pos: 17.434 acc5_pos: 100.000 loss: 0.2684 (1.0343) acc1: 91.9271 (47.5079) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1649 data: 0.0666 max mem: 3532 +Test: [260/333] eta: 0:00:12 acc1_pos: 17.434 acc5_pos: 100.000 loss: 0.2988 (1.0076) acc1: 89.0625 (49.0751) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1599 data: 0.0609 max mem: 3532 +Test: [270/333] eta: 0:00:10 acc1_pos: 17.434 acc5_pos: 100.000 loss: 0.2991 (0.9812) acc1: 89.0625 (50.5756) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1603 data: 0.0643 max mem: 3532 +Test: [280/333] eta: 0:00:09 acc1_pos: 17.434 acc5_pos: 100.000 loss: 0.2955 (0.9601) acc1: 88.2812 (51.7469) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1749 data: 0.0792 max mem: 3532 +Test: [290/333] eta: 0:00:07 acc1_pos: 17.434 acc5_pos: 100.000 loss: 0.4578 (0.9435) acc1: 76.5625 (52.6149) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.2095 data: 0.1110 max mem: 3532 +Test: [300/333] eta: 0:00:05 acc1_pos: 17.434 acc5_pos: 100.000 loss: 0.4673 (0.9272) acc1: 78.1250 (53.5481) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.1749 data: 0.0742 max mem: 3532 +Test: [310/333] eta: 0:00:03 acc1_pos: 17.434 acc5_pos: 100.000 loss: 0.3864 (0.9066) acc1: 84.1146 (54.7470) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1278 data: 0.0290 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 17.434 acc5_pos: 100.000 loss: 0.2547 (0.8870) acc1: 91.6667 (55.8768) acc5: 100.0000 (100.0000) acc1_neg: 77.344 acc5_neg: 100.000 time: 0.1385 data: 0.0459 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 17.434 acc5_pos: 100.000 loss: 0.4461 (0.8754) acc1: 78.3854 (56.4860) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.1408 data: 0.0502 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 17.434 acc5_pos: 100.000 loss: 0.4829 (0.8730) acc1: 77.6042 (56.6077) acc5: 100.0000 (100.0000) acc1_neg: 75.775 acc5_neg: 100.000 time: 0.1355 data: 0.0409 max mem: 3532 +Test: Total time: 0:00:56 (0.1702 s / it) +* Overall Acc@1 56.608 Acc@5 100.000 + target == real Acc@1 23.764 Acc@5 100.000 + target == fake Acc@1 85.845 Acc@5 100.000 + loss 0.873 + PR-AUC 0.5804 ROC-AUC 0.5759 EER 44.64% (thr=0.8130) +Accuracy of the network on the 127843 test images: 56.6% +[rank0]:[W725 19:16:15.594831674 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 19:16:34.568000 2585475 torch/distributed/run.py:766] +W0725 19:16:34.568000 2585475 torch/distributed/run.py:766] ***************************************** +W0725 19:16:34.568000 2585475 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 19:16:34.568000 2585475 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 19:17:04.546248958 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 19:17:05.260963806 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 19:17:05.532425577 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 19:17:06.791428125 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 19:17:06.825899223 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +NCCL version 2.26.2+cuda12.2 +[rank3]:[W725 19:17:06.949544472 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 19:17:06.996751874 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 19:17:06.077052587 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed4_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed4_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:27:46 acc1_pos: 85.938 acc5_pos: 100.000 loss: 0.3662 (0.3662) acc1: 85.9375 (85.9375) acc5: 100.0000 (100.0000) time: 7.0903 data: 5.3052 max mem: 3531 +Test: [ 10/235] eta: 0:02:45 acc1_pos: 86.198 acc5_pos: 100.000 loss: 0.3515 (0.3543) acc1: 85.9375 (85.7244) acc5: 100.0000 (100.0000) time: 0.7374 data: 0.4855 max mem: 3531 +Test: [ 20/235] eta: 0:01:33 acc1_pos: 81.771 acc5_pos: 100.000 loss: 0.3500 (0.3496) acc1: 85.6771 (85.7143) acc5: 100.0000 (100.0000) time: 0.1000 data: 0.0020 max mem: 3531 +Test: [ 30/235] eta: 0:01:20 acc1_pos: 87.240 acc5_pos: 100.000 loss: 0.3413 (0.3489) acc1: 86.1979 (85.8199) acc5: 100.0000 (100.0000) time: 0.1994 data: 0.1051 max mem: 3531 +Test: [ 40/235] eta: 0:01:10 acc1_pos: 86.198 acc5_pos: 100.000 loss: 0.3413 (0.3485) acc1: 86.1979 (85.7724) acc5: 100.0000 (100.0000) time: 0.2898 data: 0.1981 max mem: 3531 +Test: [ 50/235] eta: 0:01:00 acc1_pos: 84.375 acc5_pos: 100.000 loss: 0.3543 (0.3506) acc1: 85.4167 (85.5750) acc5: 100.0000 (100.0000) time: 0.2250 data: 0.1333 max mem: 3531 +Test: [ 60/235] eta: 0:00:53 acc1_pos: 86.458 acc5_pos: 100.000 loss: 0.3610 (0.3519) acc1: 84.3750 (85.4124) acc5: 100.0000 (100.0000) time: 0.1949 data: 0.1013 max mem: 3531 +Test: [ 70/235] eta: 0:00:49 acc1_pos: 83.077 acc5_pos: 100.000 loss: 0.3839 (0.4808) acc1: 82.5521 (78.5395) acc5: 100.0000 (100.0000) acc1_neg: 29.688 acc5_neg: 100.000 time: 0.2343 data: 0.1405 max mem: 3531 +Test: [ 80/235] eta: 0:00:45 acc1_pos: 83.077 acc5_pos: 100.000 loss: 1.3905 (0.5956) acc1: 29.4271 (72.3733) acc5: 100.0000 (100.0000) acc1_neg: 28.125 acc5_neg: 100.000 time: 0.2463 data: 0.1556 max mem: 3531 +Test: [ 90/235] eta: 0:00:41 acc1_pos: 83.077 acc5_pos: 100.000 loss: 1.4020 (0.6854) acc1: 29.4271 (67.6597) acc5: 100.0000 (100.0000) acc1_neg: 30.990 acc5_neg: 100.000 time: 0.2307 data: 0.1406 max mem: 3531 +Test: [100/235] eta: 0:00:37 acc1_pos: 83.077 acc5_pos: 100.000 loss: 1.3953 (0.7543) acc1: 30.2083 (64.0212) acc5: 100.0000 (100.0000) acc1_neg: 30.469 acc5_neg: 100.000 time: 0.2236 data: 0.1328 max mem: 3531 +Test: [110/235] eta: 0:00:33 acc1_pos: 83.077 acc5_pos: 100.000 loss: 1.3878 (0.8126) acc1: 29.4271 (60.8953) acc5: 100.0000 (100.0000) acc1_neg: 29.167 acc5_neg: 100.000 time: 0.2103 data: 0.1184 max mem: 3531 +Test: [120/235] eta: 0:00:31 acc1_pos: 83.077 acc5_pos: 100.000 loss: 1.4622 (0.8760) acc1: 24.7396 (57.6317) acc5: 100.0000 (100.0000) acc1_neg: 22.917 acc5_neg: 100.000 time: 0.2349 data: 0.1401 max mem: 3531 +Test: [130/235] eta: 0:00:28 acc1_pos: 83.077 acc5_pos: 100.000 loss: 1.5848 (0.9295) acc1: 21.0938 (54.8605) acc5: 100.0000 (100.0000) acc1_neg: 22.656 acc5_neg: 100.000 time: 0.2437 data: 0.1462 max mem: 3531 +Test: [140/235] eta: 0:00:24 acc1_pos: 83.077 acc5_pos: 100.000 loss: 1.5961 (0.9789) acc1: 20.5729 (52.4306) acc5: 100.0000 (100.0000) acc1_neg: 15.885 acc5_neg: 100.000 time: 0.2064 data: 0.1121 max mem: 3531 +Test: [150/235] eta: 0:00:22 acc1_pos: 83.077 acc5_pos: 100.000 loss: 1.5895 (1.0183) acc1: 20.5729 (50.3604) acc5: 100.0000 (100.0000) acc1_neg: 23.177 acc5_neg: 100.000 time: 0.2149 data: 0.1215 max mem: 3531 +Test: [160/235] eta: 0:00:19 acc1_pos: 83.077 acc5_pos: 100.000 loss: 1.5773 (1.0539) acc1: 20.5729 (48.5248) acc5: 100.0000 (100.0000) acc1_neg: 21.354 acc5_neg: 100.000 time: 0.2110 data: 0.1175 max mem: 3531 +Test: [170/235] eta: 0:00:16 acc1_pos: 83.077 acc5_pos: 100.000 loss: 1.4834 (1.0465) acc1: 22.9167 (48.6659) acc5: 100.0000 (100.0000) acc1_neg: 50.000 acc5_neg: 100.000 time: 0.2049 data: 0.1122 max mem: 3531 +Test: [180/235] eta: 0:00:13 acc1_pos: 83.077 acc5_pos: 100.000 loss: 0.8648 (1.0357) acc1: 54.1667 (49.0202) acc5: 100.0000 (100.0000) acc1_neg: 52.865 acc5_neg: 100.000 time: 0.1943 data: 0.1018 max mem: 3531 +Test: [190/235] eta: 0:00:11 acc1_pos: 83.077 acc5_pos: 100.000 loss: 0.8339 (1.0259) acc1: 54.6875 (49.3142) acc5: 100.0000 (100.0000) acc1_neg: 55.208 acc5_neg: 100.000 time: 0.1669 data: 0.0766 max mem: 3531 +Test: [200/235] eta: 0:00:08 acc1_pos: 83.077 acc5_pos: 100.000 loss: 0.8509 (1.0173) acc1: 54.6875 (49.5867) acc5: 100.0000 (100.0000) acc1_neg: 51.823 acc5_neg: 100.000 time: 0.1439 data: 0.0542 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 83.077 acc5_pos: 100.000 loss: 0.8589 (1.0095) acc1: 53.6458 (49.8396) acc5: 100.0000 (100.0000) acc1_neg: 57.292 acc5_neg: 100.000 time: 0.1224 data: 0.0330 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 83.077 acc5_pos: 100.000 loss: 0.8440 (1.0021) acc1: 54.6875 (50.0483) acc5: 100.0000 (100.0000) acc1_neg: 54.948 acc5_neg: 100.000 time: 0.1153 data: 0.0259 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 83.077 acc5_pos: 100.000 loss: 0.8399 (0.9954) acc1: 54.9479 (50.2514) acc5: 100.0000 (100.0000) acc1_neg: 55.729 acc5_neg: 100.000 time: 0.1104 data: 0.0212 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 83.077 acc5_pos: 100.000 loss: 0.8399 (0.9939) acc1: 54.9479 (50.3188) acc5: 100.0000 (100.0000) acc1_neg: 47.826 acc5_neg: 100.000 time: 0.1045 data: 0.0183 max mem: 3532 +Test: Total time: 0:00:51 (0.2207 s / it) +* Overall Acc@1 50.319 Acc@5 100.000 + target == real Acc@1 85.452 Acc@5 100.000 + target == fake Acc@1 37.612 Acc@5 100.000 + loss 0.994 + PR-AUC 0.8524 ROC-AUC 0.6859 EER 36.66% (thr=0.2181) +Accuracy of the network on the 89879 test images: 50.3% +[rank0]:[W725 19:18:05.501799394 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 19:18:25.139000 2586045 torch/distributed/run.py:766] +W0725 19:18:25.139000 2586045 torch/distributed/run.py:766] ***************************************** +W0725 19:18:25.139000 2586045 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 19:18:25.139000 2586045 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 19:18:52.980816579 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 19:18:52.315271548 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 19:18:52.501825802 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +[rank5]:[W725 19:18:53.613829875 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 19:18:53.620158952 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 19:18:53.626062976 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 19:18:53.741360634 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 19:18:53.761742870 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed4_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed4_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:25:56 acc1_pos: 8.594 acc5_pos: 100.000 loss: 1.7336 (1.7336) acc1: 8.5938 (8.5938) acc5: 100.0000 (100.0000) time: 5.7875 data: 4.1197 max mem: 3531 +Test: [ 10/269] eta: 0:02:37 acc1_pos: 25.521 acc5_pos: 100.000 loss: 1.4983 (1.3975) acc1: 14.8438 (20.0047) acc5: 100.0000 (100.0000) time: 0.6080 data: 0.3749 max mem: 3531 +Test: [ 20/269] eta: 0:01:32 acc1_pos: 15.625 acc5_pos: 100.000 loss: 1.3921 (1.4023) acc1: 16.1458 (18.6384) acc5: 100.0000 (100.0000) time: 0.1027 data: 0.0099 max mem: 3531 +Test: [ 30/269] eta: 0:01:21 acc1_pos: 29.948 acc5_pos: 100.000 loss: 1.4538 (1.4460) acc1: 16.1458 (18.0276) acc5: 100.0000 (100.0000) time: 0.1970 data: 0.1017 max mem: 3531 +Test: [ 40/269] eta: 0:01:18 acc1_pos: 6.897 acc5_pos: 100.000 loss: 1.0701 (1.1791) acc1: 29.9479 (34.8514) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.3142 data: 0.2200 max mem: 3531 +Test: [ 50/269] eta: 0:01:09 acc1_pos: 6.897 acc5_pos: 100.000 loss: 0.2265 (1.0015) acc1: 96.6146 (46.3542) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.2799 data: 0.1848 max mem: 3531 +Test: [ 60/269] eta: 0:01:04 acc1_pos: 6.897 acc5_pos: 100.000 loss: 0.3084 (0.8929) acc1: 90.1042 (53.3214) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.2310 data: 0.1373 max mem: 3531 +Test: [ 70/269] eta: 0:00:56 acc1_pos: 6.897 acc5_pos: 100.000 loss: 0.3343 (0.8113) acc1: 88.5417 (58.4800) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.2049 data: 0.1131 max mem: 3531 +Test: [ 80/269] eta: 0:00:52 acc1_pos: 6.897 acc5_pos: 100.000 loss: 0.2692 (0.7446) acc1: 91.9271 (62.6897) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1820 data: 0.0886 max mem: 3531 +Test: [ 90/269] eta: 0:00:48 acc1_pos: 6.897 acc5_pos: 100.000 loss: 0.2583 (0.7058) acc1: 92.1875 (65.2787) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.2160 data: 0.1219 max mem: 3531 +Test: [100/269] eta: 0:00:44 acc1_pos: 6.897 acc5_pos: 100.000 loss: 0.3663 (0.6715) acc1: 86.4583 (67.5562) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.2178 data: 0.1237 max mem: 3531 +Test: [110/269] eta: 0:00:41 acc1_pos: 6.897 acc5_pos: 100.000 loss: 0.2908 (0.6336) acc1: 92.1875 (70.0005) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2062 data: 0.1138 max mem: 3531 +Test: [120/269] eta: 0:00:37 acc1_pos: 6.897 acc5_pos: 100.000 loss: 0.2334 (0.5994) acc1: 95.8333 (72.2602) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2017 data: 0.1099 max mem: 3531 +Test: [130/269] eta: 0:00:34 acc1_pos: 6.897 acc5_pos: 100.000 loss: 0.2342 (0.5785) acc1: 95.8333 (73.6542) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.1944 data: 0.1021 max mem: 3531 +Test: [140/269] eta: 0:00:31 acc1_pos: 6.897 acc5_pos: 100.000 loss: 0.3031 (0.5592) acc1: 93.2292 (75.0111) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.1740 data: 0.0824 max mem: 3531 +Test: [150/269] eta: 0:00:28 acc1_pos: 6.897 acc5_pos: 100.000 loss: 0.2772 (0.5374) acc1: 94.0104 (76.3331) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1754 data: 0.0836 max mem: 3531 +Test: [160/269] eta: 0:00:25 acc1_pos: 6.897 acc5_pos: 100.000 loss: 0.1999 (0.5190) acc1: 95.5729 (77.4359) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1812 data: 0.0884 max mem: 3531 +Test: [170/269] eta: 0:00:23 acc1_pos: 6.897 acc5_pos: 100.000 loss: 0.3007 (0.5084) acc1: 90.8854 (78.1676) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1843 data: 0.0903 max mem: 3531 +Test: [180/269] eta: 0:00:20 acc1_pos: 6.897 acc5_pos: 100.000 loss: 0.2809 (0.4932) acc1: 94.2708 (79.2026) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1839 data: 0.0918 max mem: 3531 +Test: [190/269] eta: 0:00:17 acc1_pos: 6.897 acc5_pos: 100.000 loss: 0.2465 (0.4839) acc1: 96.3542 (79.8525) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1831 data: 0.0909 max mem: 3531 +Test: [200/269] eta: 0:00:15 acc1_pos: 6.897 acc5_pos: 100.000 loss: 0.3377 (0.4768) acc1: 90.1042 (80.2731) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1880 data: 0.0955 max mem: 3531 +Test: [210/269] eta: 0:00:13 acc1_pos: 6.897 acc5_pos: 100.000 loss: 0.2137 (0.4619) acc1: 96.0938 (81.1340) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1849 data: 0.0924 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 6.897 acc5_pos: 100.000 loss: 0.1409 (0.4472) acc1: 98.9583 (81.9452) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1753 data: 0.0822 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 6.897 acc5_pos: 100.000 loss: 0.1379 (0.4347) acc1: 99.4792 (82.6535) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1599 data: 0.0676 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 6.897 acc5_pos: 100.000 loss: 0.2247 (0.4305) acc1: 95.5729 (82.9357) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1593 data: 0.0669 max mem: 3532 +Test: [250/269] eta: 0:00:04 acc1_pos: 6.897 acc5_pos: 100.000 loss: 0.3180 (0.4262) acc1: 90.6250 (83.3012) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1408 data: 0.0499 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 6.897 acc5_pos: 100.000 loss: 0.3341 (0.4246) acc1: 90.6250 (83.4122) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1167 data: 0.0271 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 6.897 acc5_pos: 100.000 loss: 0.3336 (0.4207) acc1: 89.8438 (83.6649) acc5: 100.0000 (100.0000) acc1_neg: 96.850 acc5_neg: 100.000 time: 0.1260 data: 0.0329 max mem: 3532 +Test: Total time: 0:00:55 (0.2066 s / it) +* Overall Acc@1 83.665 Acc@5 100.000 + target == real Acc@1 17.752 Acc@5 100.000 + target == fake Acc@1 92.660 Acc@5 100.000 + loss 0.421 + PR-AUC 0.9092 ROC-AUC 0.6097 EER 42.19% (thr=0.8086) +Accuracy of the network on the 103293 test images: 83.7% +[rank0]:[W725 19:19:57.210925826 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-25T19:20:04Z] eval_done tag=fixed500k_real_n16_seed4_resampled +[2026-07-25T19:20:04Z] eval_wait tag=fixed500k_fake_n32_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed4_resampled +[2026-07-25T19:21:04Z] eval_wait tag=fixed500k_fake_n32_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed4_resampled +[2026-07-25T19:22:04Z] eval_wait tag=fixed500k_fake_n32_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed4_resampled +[2026-07-25T19:23:04Z] eval_wait tag=fixed500k_fake_n32_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed4_resampled +[2026-07-25T19:24:04Z] eval_wait tag=fixed500k_fake_n32_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed4_resampled +[2026-07-25T19:25:04Z] eval_wait tag=fixed500k_fake_n32_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed4_resampled +[2026-07-25T19:26:04Z] eval_wait tag=fixed500k_fake_n32_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed4_resampled +[2026-07-25T19:27:04Z] eval_wait tag=fixed500k_fake_n32_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed4_resampled +[2026-07-25T19:28:04Z] eval_wait tag=fixed500k_fake_n32_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed4_resampled +[2026-07-25T19:29:04Z] eval_wait tag=fixed500k_fake_n32_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed4_resampled +[2026-07-25T19:30:04Z] eval_wait tag=fixed500k_fake_n32_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed4_resampled +[2026-07-25T19:31:04Z] eval_wait tag=fixed500k_fake_n32_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed4_resampled +[2026-07-25T19:32:04Z] eval_wait tag=fixed500k_fake_n32_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed4_resampled +[2026-07-25T19:33:04Z] eval_wait tag=fixed500k_fake_n32_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed4_resampled +[2026-07-25T19:34:04Z] eval_skip tag=fixed500k_fake_n32_seed4_resampled reason=complete_after_lock output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed4_resampled +eval_queue_done_utc=2026-07-25T19:34:04Z diff --git a/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_112559/eval_queue_n111-046-081.pid b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_112559/eval_queue_n111-046-081.pid new file mode 100644 index 0000000000000000000000000000000000000000..4621f22632b9de110a7c811e9365994a0d5cc7e9 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_112559/eval_queue_n111-046-081.pid @@ -0,0 +1 @@ +2561247 diff --git a/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_112559/eval_queue_n116-040-168.log b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_112559/eval_queue_n116-040-168.log new file mode 100644 index 0000000000000000000000000000000000000000..10414b529284381e5cdde73ac935946018cb1923 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_112559/eval_queue_n116-040-168.log @@ -0,0 +1,1136 @@ +eval_queue_started_utc=2026-07-25T17:02:40Z +hostname=n116-040-168 +train_tag=20260725_112559 +eval_tag=20260725_112559 +real_level=5 +fake_level=11 +eval_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559 +batch_per_gpu=256 +[2026-07-25T17:02:40Z] eval_wait tag=fixed500k_real_n5_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed1_resampled +[2026-07-25T17:03:40Z] eval_wait tag=fixed500k_real_n5_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed1_resampled +[2026-07-25T17:04:40Z] eval_wait tag=fixed500k_real_n5_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed1_resampled +[2026-07-25T17:05:40Z] eval_wait tag=fixed500k_real_n5_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed1_resampled +[2026-07-25T17:06:40Z] eval_wait tag=fixed500k_real_n5_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed1_resampled +[2026-07-25T17:07:40Z] eval_wait tag=fixed500k_real_n5_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed1_resampled +[2026-07-25T17:08:40Z] eval_wait tag=fixed500k_real_n5_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed1_resampled +[2026-07-25T17:09:41Z] eval_wait tag=fixed500k_real_n5_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed1_resampled +[2026-07-25T17:10:41Z] eval_skip tag=fixed500k_real_n5_seed1_resampled reason=complete_after_lock output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed1_resampled +[2026-07-25T17:10:41Z] eval_skip tag=fixed500k_fake_n11_seed1_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed1_resampled +[2026-07-25T17:10:41Z] eval_wait tag=fixed500k_real_n5_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed2_resampled +[2026-07-25T17:11:41Z] eval_wait tag=fixed500k_real_n5_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed2_resampled +[2026-07-25T17:12:41Z] eval_wait tag=fixed500k_real_n5_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed2_resampled +[2026-07-25T17:13:41Z] eval_wait tag=fixed500k_real_n5_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed2_resampled +[2026-07-25T17:14:41Z] eval_wait tag=fixed500k_real_n5_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed2_resampled +[2026-07-25T17:15:42Z] eval_wait tag=fixed500k_real_n5_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed2_resampled +[2026-07-25T17:16:42Z] eval_wait tag=fixed500k_real_n5_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed2_resampled +[2026-07-25T17:17:42Z] eval_wait tag=fixed500k_real_n5_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed2_resampled +[2026-07-25T17:18:42Z] eval_wait tag=fixed500k_real_n5_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed2_resampled +[2026-07-25T17:19:42Z] eval_wait tag=fixed500k_real_n5_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed2_resampled +[2026-07-25T17:20:42Z] eval_wait tag=fixed500k_real_n5_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed2_resampled +[2026-07-25T17:21:43Z] eval_wait tag=fixed500k_real_n5_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed2_resampled +[2026-07-25T17:22:43Z] eval_wait tag=fixed500k_real_n5_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed2_resampled +[2026-07-25T17:23:43Z] eval_wait tag=fixed500k_real_n5_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed2_resampled +[2026-07-25T17:24:43Z] eval_wait tag=fixed500k_real_n5_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed2_resampled +[2026-07-25T17:25:43Z] eval_wait tag=fixed500k_real_n5_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed2_resampled +[2026-07-25T17:26:43Z] eval_wait tag=fixed500k_real_n5_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed2_resampled +[2026-07-25T17:27:44Z] eval_wait tag=fixed500k_real_n5_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed2_resampled +[2026-07-25T17:28:44Z] eval_wait tag=fixed500k_real_n5_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed2_resampled +[2026-07-25T17:29:44Z] eval_wait tag=fixed500k_real_n5_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed2_resampled +[2026-07-25T17:30:44Z] eval_wait tag=fixed500k_real_n5_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed2_resampled +[2026-07-25T17:31:44Z] eval_wait tag=fixed500k_real_n5_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed2_resampled +[2026-07-25T17:32:44Z] eval_skip tag=fixed500k_real_n5_seed2_resampled reason=complete_after_lock output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed2_resampled +[2026-07-25T17:32:44Z] eval_skip tag=fixed500k_fake_n11_seed2_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed2_resampled +[2026-07-25T17:32:44Z] eval_wait tag=fixed500k_real_n5_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed3_resampled +[2026-07-25T17:33:45Z] eval_wait tag=fixed500k_real_n5_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed3_resampled +[2026-07-25T17:34:45Z] eval_wait tag=fixed500k_real_n5_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed3_resampled +[2026-07-25T17:35:45Z] eval_wait tag=fixed500k_real_n5_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed3_resampled +[2026-07-25T17:36:45Z] eval_wait tag=fixed500k_real_n5_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed3_resampled +[2026-07-25T17:37:45Z] eval_wait tag=fixed500k_real_n5_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed3_resampled +[2026-07-25T17:38:45Z] eval_wait tag=fixed500k_real_n5_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed3_resampled +[2026-07-25T17:39:46Z] eval_wait tag=fixed500k_real_n5_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed3_resampled +[2026-07-25T17:40:46Z] eval_wait tag=fixed500k_real_n5_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed3_resampled +[2026-07-25T17:41:46Z] eval_wait tag=fixed500k_real_n5_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed3_resampled +[2026-07-25T17:42:46Z] eval_wait tag=fixed500k_real_n5_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed3_resampled +[2026-07-25T17:43:46Z] eval_wait tag=fixed500k_real_n5_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed3_resampled +[2026-07-25T17:44:46Z] eval_wait tag=fixed500k_real_n5_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed3_resampled +[2026-07-25T17:45:46Z] eval_wait tag=fixed500k_real_n5_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed3_resampled +[2026-07-25T17:46:47Z] eval_wait tag=fixed500k_real_n5_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed3_resampled +[2026-07-25T17:47:47Z] eval_wait tag=fixed500k_real_n5_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed3_resampled +[2026-07-25T17:48:47Z] eval_wait tag=fixed500k_real_n5_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed3_resampled +[2026-07-25T17:49:47Z] eval_wait tag=fixed500k_real_n5_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed3_resampled +[2026-07-25T17:50:47Z] eval_wait tag=fixed500k_real_n5_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed3_resampled +[2026-07-25T17:51:47Z] eval_wait tag=fixed500k_real_n5_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed3_resampled +[2026-07-25T17:52:48Z] eval_wait tag=fixed500k_real_n5_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed3_resampled +[2026-07-25T17:53:48Z] eval_skip tag=fixed500k_real_n5_seed3_resampled reason=complete_after_lock output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed3_resampled +[2026-07-25T17:53:48Z] eval_wait tag=fixed500k_fake_n11_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed3_resampled +[2026-07-25T17:54:48Z] eval_wait tag=fixed500k_fake_n11_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed3_resampled +[2026-07-25T17:55:48Z] eval_wait tag=fixed500k_fake_n11_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed3_resampled +[2026-07-25T17:56:48Z] eval_wait tag=fixed500k_fake_n11_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed3_resampled +[2026-07-25T17:57:48Z] eval_wait tag=fixed500k_fake_n11_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed3_resampled +[2026-07-25T17:58:48Z] eval_wait tag=fixed500k_fake_n11_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed3_resampled +[2026-07-25T17:59:49Z] eval_wait tag=fixed500k_fake_n11_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed3_resampled +[2026-07-25T18:00:49Z] eval_wait tag=fixed500k_fake_n11_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed3_resampled +[2026-07-25T18:01:49Z] eval_wait tag=fixed500k_fake_n11_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed3_resampled +[2026-07-25T18:02:49Z] eval_wait tag=fixed500k_fake_n11_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed3_resampled +[2026-07-25T18:03:49Z] eval_wait tag=fixed500k_fake_n11_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed3_resampled +[2026-07-25T18:04:49Z] eval_wait tag=fixed500k_fake_n11_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed3_resampled +[2026-07-25T18:05:50Z] eval_wait tag=fixed500k_fake_n11_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed3_resampled +[2026-07-25T18:06:50Z] eval_wait tag=fixed500k_fake_n11_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed3_resampled +[2026-07-25T18:07:50Z] eval_wait tag=fixed500k_fake_n11_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed3_resampled +[2026-07-25T18:08:50Z] eval_wait tag=fixed500k_fake_n11_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed3_resampled +[2026-07-25T18:09:50Z] eval_wait tag=fixed500k_fake_n11_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed3_resampled +[2026-07-25T18:10:50Z] eval_wait tag=fixed500k_fake_n11_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed3_resampled +[2026-07-25T18:11:50Z] eval_wait tag=fixed500k_fake_n11_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed3_resampled +[2026-07-25T18:12:51Z] eval_wait tag=fixed500k_fake_n11_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed3_resampled +[2026-07-25T18:13:51Z] eval_wait tag=fixed500k_fake_n11_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed3_resampled +[2026-07-25T18:14:51Z] eval_wait tag=fixed500k_fake_n11_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed3_resampled +[2026-07-25T18:15:51Z] eval_skip tag=fixed500k_fake_n11_seed3_resampled reason=complete_after_lock output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed3_resampled +[2026-07-25T18:15:51Z] eval_wait tag=fixed500k_real_n5_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed4_resampled +[2026-07-25T18:16:52Z] eval_wait tag=fixed500k_real_n5_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed4_resampled +[2026-07-25T18:17:52Z] eval_wait tag=fixed500k_real_n5_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed4_resampled +[2026-07-25T18:18:52Z] eval_wait tag=fixed500k_real_n5_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed4_resampled +[2026-07-25T18:19:52Z] eval_wait tag=fixed500k_real_n5_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed4_resampled +[2026-07-25T18:20:52Z] eval_wait tag=fixed500k_real_n5_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed4_resampled +[2026-07-25T18:21:52Z] eval_wait tag=fixed500k_real_n5_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed4_resampled +[2026-07-25T18:22:53Z] eval_wait tag=fixed500k_real_n5_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed4_resampled +[2026-07-25T18:23:53Z] eval_wait tag=fixed500k_real_n5_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed4_resampled +[2026-07-25T18:24:53Z] eval_wait tag=fixed500k_real_n5_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed4_resampled +[2026-07-25T18:25:53Z] eval_wait tag=fixed500k_real_n5_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed4_resampled +[2026-07-25T18:26:53Z] eval_wait tag=fixed500k_real_n5_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed4_resampled +[2026-07-25T18:27:53Z] eval_wait tag=fixed500k_real_n5_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed4_resampled +[2026-07-25T18:28:54Z] eval_wait tag=fixed500k_real_n5_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed4_resampled +[2026-07-25T18:29:54Z] eval_wait tag=fixed500k_real_n5_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed4_resampled +[2026-07-25T18:30:54Z] eval_wait tag=fixed500k_real_n5_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed4_resampled +[2026-07-25T18:31:54Z] eval_wait tag=fixed500k_real_n5_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed4_resampled +[2026-07-25T18:32:54Z] eval_wait tag=fixed500k_real_n5_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed4_resampled +[2026-07-25T18:33:54Z] eval_wait tag=fixed500k_real_n5_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed4_resampled +[2026-07-25T18:34:54Z] eval_wait tag=fixed500k_real_n5_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed4_resampled +[2026-07-25T18:35:55Z] eval_wait tag=fixed500k_real_n5_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed4_resampled +[2026-07-25T18:36:55Z] eval_skip tag=fixed500k_real_n5_seed4_resampled reason=complete_after_wait output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed4_resampled +[2026-07-25T18:36:55Z] eval_start tag=fixed500k_fake_n11_seed4_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed4_resampled_n116-040-168_20260725_112559/best_checkpoint.pth +started_utc=2026-07-25T18:36:55Z +hostname=n116-040-168 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed4_resampled_n116-040-168_20260725_112559/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed4_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0725 18:37:08.034000 2575623 torch/distributed/run.py:766] +W0725 18:37:08.034000 2575623 torch/distributed/run.py:766] ***************************************** +W0725 18:37:08.034000 2575623 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:37:08.034000 2575623 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:37:34.129913587 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:37:34.168899631 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:37:35.809324566 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:37:35.841242522 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:37:35.850735535 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:37:35.910724987 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:37:35.951392082 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:37:35.015290447 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed4_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed4_resampled_n116-040-168_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:24:35 acc1_pos: 82.292 acc5_pos: 100.000 loss: 0.3483 (0.3483) acc1: 82.2917 (82.2917) acc5: 100.0000 (100.0000) time: 6.3035 data: 3.9319 max mem: 3531 +Test: [ 10/234] eta: 0:02:26 acc1_pos: 94.010 acc5_pos: 100.000 loss: 0.5096 (0.5261) acc1: 71.3542 (73.7926) acc5: 100.0000 (100.0000) time: 0.6550 data: 0.3578 max mem: 3531 +Test: [ 20/234] eta: 0:01:23 acc1_pos: 95.833 acc5_pos: 100.000 loss: 0.3843 (0.4253) acc1: 81.2500 (81.1384) acc5: 100.0000 (100.0000) time: 0.0950 data: 0.0024 max mem: 3531 +Test: [ 30/234] eta: 0:01:15 acc1_pos: 86.458 acc5_pos: 100.000 loss: 0.1941 (0.3544) acc1: 95.8333 (85.5007) acc5: 100.0000 (100.0000) time: 0.2165 data: 0.1221 max mem: 3531 +Test: [ 40/234] eta: 0:01:06 acc1_pos: 98.958 acc5_pos: 100.000 loss: 0.1829 (0.3316) acc1: 95.5729 (87.4937) acc5: 100.0000 (100.0000) time: 0.2933 data: 0.2013 max mem: 3531 +Test: [ 50/234] eta: 0:00:57 acc1_pos: 97.917 acc5_pos: 100.000 loss: 0.1446 (0.2945) acc1: 97.9167 (89.5731) acc5: 100.0000 (100.0000) time: 0.2232 data: 0.1304 max mem: 3531 +Test: [ 60/234] eta: 0:00:52 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.1114 (0.2612) acc1: 99.2188 (91.2227) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2166 data: 0.1177 max mem: 3531 +Test: [ 70/234] eta: 0:00:46 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.0947 (0.3151) acc1: 99.4792 (87.4963) acc5: 100.0000 (100.0000) acc1_neg: 45.052 acc5_neg: 100.000 time: 0.2049 data: 0.1043 max mem: 3531 +Test: [ 80/234] eta: 0:00:42 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.8972 (0.3608) acc1: 48.9583 (84.5968) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.2049 data: 0.1080 max mem: 3531 +Test: [ 90/234] eta: 0:00:39 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.3117 (0.3453) acc1: 86.9792 (85.5454) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.2299 data: 0.1360 max mem: 3531 +Test: [100/234] eta: 0:00:35 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.3138 (0.3551) acc1: 86.1979 (84.9190) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2030 data: 0.1093 max mem: 3531 +Test: [110/234] eta: 0:00:32 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.4091 (0.3592) acc1: 78.6458 (84.5627) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.1975 data: 0.1029 max mem: 3531 +Test: [120/234] eta: 0:00:28 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.3841 (0.3697) acc1: 79.9479 (83.9446) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.2068 data: 0.1138 max mem: 3531 +Test: [130/234] eta: 0:00:25 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.4829 (0.3915) acc1: 77.0833 (82.3414) acc5: 100.0000 (100.0000) acc1_neg: 57.031 acc5_neg: 100.000 time: 0.1942 data: 0.1027 max mem: 3531 +Test: [140/234] eta: 0:00:23 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.7777 (0.4442) acc1: 50.7812 (79.2978) acc5: 100.0000 (100.0000) acc1_neg: 25.000 acc5_neg: 100.000 time: 0.1923 data: 0.0972 max mem: 3531 +Test: [150/234] eta: 0:00:20 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.7600 (0.4486) acc1: 50.7812 (79.1098) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.2008 data: 0.1056 max mem: 3531 +Test: [160/234] eta: 0:00:17 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.4321 (0.4537) acc1: 77.3438 (78.7348) acc5: 100.0000 (100.0000) acc1_neg: 67.969 acc5_neg: 100.000 time: 0.1997 data: 0.1073 max mem: 3531 +Test: [170/234] eta: 0:00:15 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.5899 (0.4614) acc1: 69.0104 (78.2118) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.1985 data: 0.1055 max mem: 3531 +Test: [180/234] eta: 0:00:12 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.6038 (0.4758) acc1: 65.8854 (77.2833) acc5: 100.0000 (100.0000) acc1_neg: 32.292 acc5_neg: 100.000 time: 0.1892 data: 0.0933 max mem: 3531 +Test: [190/234] eta: 0:00:10 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.6717 (0.4876) acc1: 66.9271 (76.7057) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.1688 data: 0.0728 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.6700 (0.4796) acc1: 68.2292 (77.2738) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1581 data: 0.0659 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.1296 (0.4685) acc1: 99.4792 (77.9831) acc5: 100.0000 (100.0000) acc1_neg: 75.781 acc5_neg: 100.000 time: 0.1537 data: 0.0630 max mem: 3532 +Test: [220/234] eta: 0:00:03 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.4720 (0.4712) acc1: 78.9062 (77.8634) acc5: 100.0000 (100.0000) acc1_neg: 56.250 acc5_neg: 100.000 time: 0.1255 data: 0.0347 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.9122 (0.5275) acc1: 12.5000 (74.9391) acc5: 100.0000 (100.0000) acc1_neg: 8.073 acc5_neg: 100.000 time: 0.0986 data: 0.0065 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 99.310 acc5_pos: 100.000 loss: 1.7185 (0.5434) acc1: 11.7188 (74.2068) acc5: 100.0000 (100.0000) acc1_neg: 10.117 acc5_neg: 100.000 time: 0.0996 data: 0.0065 max mem: 3532 +Test: Total time: 0:00:49 (0.2113 s / it) +* Overall Acc@1 74.207 Acc@5 100.000 + target == real Acc@1 89.645 Acc@5 100.000 + target == fake Acc@1 69.856 Acc@5 100.000 + loss 0.543 + PR-AUC 0.9632 ROC-AUC 0.8748 EER 20.89% (thr=0.3440) +Accuracy of the network on the 89729 test images: 74.2% +[rank0]:[W725 18:38:32.507939720 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:38:53.709000 2576192 torch/distributed/run.py:766] +W0725 18:38:53.709000 2576192 torch/distributed/run.py:766] ***************************************** +W0725 18:38:53.709000 2576192 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:38:53.709000 2576192 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:39:25.065063297 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:39:26.226887969 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:39:26.734413875 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:39:26.750567939 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:39:26.842598335 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W725 18:39:26.843797839 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:39:26.849536684 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:39:26.886869940 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed4_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed4_resampled_n116-040-168_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:45:40 acc1_pos: 75.521 acc5_pos: 100.000 loss: 0.5186 (0.5186) acc1: 75.5208 (75.5208) acc5: 100.0000 (100.0000) time: 7.1809 data: 5.4900 max mem: 3531 +Test: [ 10/2387] eta: 0:29:07 acc1_pos: 46.615 acc5_pos: 100.000 loss: 0.6607 (0.6851) acc1: 61.9792 (61.2453) acc5: 100.0000 (100.0000) time: 0.7350 data: 0.4995 max mem: 3531 +Test: [ 20/2387] eta: 0:17:03 acc1_pos: 48.698 acc5_pos: 100.000 loss: 0.6715 (0.7170) acc1: 58.0729 (57.7381) acc5: 100.0000 (100.0000) time: 0.0952 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:15:57 acc1_pos: 44.531 acc5_pos: 100.000 loss: 0.7224 (0.7087) acc1: 54.1667 (58.1233) acc5: 100.0000 (100.0000) time: 0.2255 data: 0.1303 max mem: 3531 +Test: [ 40/2387] eta: 0:15:20 acc1_pos: 71.615 acc5_pos: 100.000 loss: 0.6437 (0.6839) acc1: 61.9792 (60.3659) acc5: 100.0000 (100.0000) time: 0.3499 data: 0.2560 max mem: 3531 +Test: [ 50/2387] eta: 0:14:44 acc1_pos: 52.865 acc5_pos: 100.000 loss: 0.5902 (0.6741) acc1: 69.7917 (61.4226) acc5: 100.0000 (100.0000) time: 0.3350 data: 0.2397 max mem: 3531 +Test: [ 60/2387] eta: 0:14:06 acc1_pos: 70.052 acc5_pos: 100.000 loss: 0.5948 (0.6607) acc1: 70.0521 (63.0294) acc5: 100.0000 (100.0000) time: 0.3049 data: 0.2118 max mem: 3531 +Test: [ 70/2387] eta: 0:13:41 acc1_pos: 63.802 acc5_pos: 100.000 loss: 0.6304 (0.6650) acc1: 66.6667 (62.9291) acc5: 100.0000 (100.0000) time: 0.2942 data: 0.2006 max mem: 3531 +Test: [ 80/2387] eta: 0:12:52 acc1_pos: 91.927 acc5_pos: 100.000 loss: 0.6097 (0.6407) acc1: 68.4896 (65.0945) acc5: 100.0000 (100.0000) time: 0.2463 data: 0.1511 max mem: 3531 +Test: [ 90/2387] eta: 0:12:24 acc1_pos: 95.052 acc5_pos: 100.000 loss: 0.4192 (0.6161) acc1: 82.0312 (66.9843) acc5: 100.0000 (100.0000) time: 0.2149 data: 0.1210 max mem: 3531 +Test: [ 100/2387] eta: 0:11:59 acc1_pos: 86.719 acc5_pos: 100.000 loss: 0.3888 (0.6006) acc1: 82.0312 (68.3581) acc5: 100.0000 (100.0000) time: 0.2335 data: 0.1387 max mem: 3531 +Test: [ 110/2387] eta: 0:11:39 acc1_pos: 59.115 acc5_pos: 100.000 loss: 0.4234 (0.5877) acc1: 78.3854 (69.2779) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1335 max mem: 3531 +Test: [ 120/2387] eta: 0:11:29 acc1_pos: 79.948 acc5_pos: 100.000 loss: 0.4626 (0.5786) acc1: 77.8646 (70.2006) acc5: 100.0000 (100.0000) time: 0.2510 data: 0.1567 max mem: 3531 +Test: [ 130/2387] eta: 0:11:10 acc1_pos: 72.917 acc5_pos: 100.000 loss: 0.4728 (0.5699) acc1: 77.8646 (70.9387) acc5: 100.0000 (100.0000) time: 0.2422 data: 0.1498 max mem: 3531 +Test: [ 140/2387] eta: 0:11:00 acc1_pos: 82.552 acc5_pos: 100.000 loss: 0.4598 (0.5644) acc1: 79.4271 (71.5075) acc5: 100.0000 (100.0000) time: 0.2338 data: 0.1393 max mem: 3531 +Test: [ 150/2387] eta: 0:10:42 acc1_pos: 90.365 acc5_pos: 100.000 loss: 0.4382 (0.5529) acc1: 83.0729 (72.5269) acc5: 100.0000 (100.0000) time: 0.2226 data: 0.1271 max mem: 3531 +Test: [ 160/2387] eta: 0:10:26 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.3956 (0.5440) acc1: 86.9792 (73.3065) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.1904 data: 0.0954 max mem: 3531 +Test: [ 170/2387] eta: 0:10:08 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.3172 (0.5305) acc1: 90.1042 (74.2919) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1800 data: 0.0840 max mem: 3531 +Test: [ 180/2387] eta: 0:09:54 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.3132 (0.5185) acc1: 90.3646 (75.1439) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1743 data: 0.0753 max mem: 3531 +Test: [ 190/2387] eta: 0:09:45 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.3098 (0.5075) acc1: 89.3229 (75.9530) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1949 data: 0.0922 max mem: 3531 +Test: [ 200/2387] eta: 0:09:31 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.3049 (0.4976) acc1: 89.8438 (76.6623) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1899 data: 0.0897 max mem: 3531 +Test: [ 210/2387] eta: 0:09:22 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.3068 (0.4890) acc1: 89.8438 (77.2771) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1810 data: 0.0854 max mem: 3532 +Test: [ 220/2387] eta: 0:09:12 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.3143 (0.4813) acc1: 89.8438 (77.8316) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1901 data: 0.0979 max mem: 3532 +Test: [ 230/2387] eta: 0:09:07 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.3136 (0.4738) acc1: 90.1042 (78.3922) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.2051 data: 0.1126 max mem: 3532 +Test: [ 240/2387] eta: 0:08:58 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.3243 (0.4730) acc1: 89.5833 (78.4719) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.2049 data: 0.1108 max mem: 3532 +Test: [ 250/2387] eta: 0:08:55 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.2811 (0.4609) acc1: 80.9896 (79.2082) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2136 data: 0.1199 max mem: 3532 +Test: [ 260/2387] eta: 0:08:45 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.1265 (0.4480) acc1: 99.7396 (79.9938) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2014 data: 0.1081 max mem: 3532 +Test: [ 270/2387] eta: 0:08:37 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.1113 (0.4357) acc1: 99.7396 (80.7234) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1699 data: 0.0760 max mem: 3532 +Test: [ 280/2387] eta: 0:08:30 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.1142 (0.4247) acc1: 99.7396 (81.3862) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1782 data: 0.0823 max mem: 3532 +Test: [ 290/2387] eta: 0:08:24 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.1261 (0.4144) acc1: 99.7396 (81.9964) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1862 data: 0.0916 max mem: 3532 +Test: [ 300/2387] eta: 0:08:19 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.1174 (0.4049) acc1: 100.0000 (82.5858) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1999 data: 0.1060 max mem: 3532 +Test: [ 310/2387] eta: 0:08:12 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.1360 (0.3992) acc1: 100.0000 (83.0495) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1906 data: 0.0939 max mem: 3532 +Test: [ 320/2387] eta: 0:08:07 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.3181 (0.3969) acc1: 95.5729 (83.4250) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1857 data: 0.0918 max mem: 3532 +Test: [ 330/2387] eta: 0:08:03 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.3222 (0.3945) acc1: 95.0521 (83.7928) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1996 data: 0.1075 max mem: 3532 +Test: [ 340/2387] eta: 0:07:58 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.3168 (0.3923) acc1: 95.3125 (84.1222) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1991 data: 0.1058 max mem: 3532 +Test: [ 350/2387] eta: 0:07:53 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.3089 (0.3875) acc1: 95.5729 (84.5442) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1883 data: 0.0944 max mem: 3532 +Test: [ 360/2387] eta: 0:07:47 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.2290 (0.3834) acc1: 96.0938 (84.8280) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1763 data: 0.0825 max mem: 3532 +Test: [ 370/2387] eta: 0:07:43 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.2404 (0.3797) acc1: 94.5312 (85.1008) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1805 data: 0.0887 max mem: 3532 +Test: [ 380/2387] eta: 0:07:38 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.2440 (0.3761) acc1: 94.5312 (85.3538) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1857 data: 0.0933 max mem: 3532 +Test: [ 390/2387] eta: 0:07:33 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.2478 (0.3728) acc1: 94.0104 (85.5778) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1794 data: 0.0850 max mem: 3532 +Test: [ 400/2387] eta: 0:07:29 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.2478 (0.3697) acc1: 94.0104 (85.7888) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1812 data: 0.0864 max mem: 3532 +Test: [ 410/2387] eta: 0:07:24 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.2412 (0.3665) acc1: 94.5312 (86.0110) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1820 data: 0.0882 max mem: 3532 +Test: [ 420/2387] eta: 0:07:19 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.2392 (0.3635) acc1: 95.0521 (86.2220) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1750 data: 0.0810 max mem: 3532 +Test: [ 430/2387] eta: 0:07:16 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.2502 (0.3628) acc1: 93.7500 (86.2759) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.1826 data: 0.0878 max mem: 3532 +Test: [ 440/2387] eta: 0:07:12 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.2636 (0.3589) acc1: 88.5417 (86.4660) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1953 data: 0.1034 max mem: 3533 +Test: [ 450/2387] eta: 0:07:08 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.0760 (0.3526) acc1: 100.0000 (86.7661) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1879 data: 0.0974 max mem: 3533 +Test: [ 460/2387] eta: 0:07:05 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.0749 (0.3465) acc1: 100.0000 (87.0532) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1886 data: 0.0952 max mem: 3533 +Test: [ 470/2387] eta: 0:07:02 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.0751 (0.3408) acc1: 100.0000 (87.3280) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1996 data: 0.1039 max mem: 3533 +Test: [ 480/2387] eta: 0:06:58 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.0755 (0.3353) acc1: 100.0000 (87.5915) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1859 data: 0.0917 max mem: 3533 +Test: [ 490/2387] eta: 0:06:55 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.0755 (0.3300) acc1: 100.0000 (87.8442) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1926 data: 0.0997 max mem: 3533 +Test: [ 500/2387] eta: 0:06:52 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.0758 (0.3249) acc1: 100.0000 (88.0868) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1984 data: 0.1054 max mem: 3533 +Test: [ 510/2387] eta: 0:06:49 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.0763 (0.3202) acc1: 100.0000 (88.3200) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1868 data: 0.0948 max mem: 3533 +Test: [ 520/2387] eta: 0:06:45 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.0860 (0.3160) acc1: 100.0000 (88.5437) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1826 data: 0.0897 max mem: 3533 +Test: [ 530/2387] eta: 0:06:41 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.0844 (0.3116) acc1: 100.0000 (88.7594) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1729 data: 0.0807 max mem: 3533 +Test: [ 540/2387] eta: 0:06:38 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.0829 (0.3074) acc1: 100.0000 (88.9672) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1748 data: 0.0842 max mem: 3533 +Test: [ 550/2387] eta: 0:06:35 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.0829 (0.3033) acc1: 100.0000 (89.1674) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1855 data: 0.0936 max mem: 3533 +Test: [ 560/2387] eta: 0:06:31 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.0825 (0.2994) acc1: 100.0000 (89.3605) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1830 data: 0.0912 max mem: 3533 +Test: [ 570/2387] eta: 0:06:28 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.0826 (0.2956) acc1: 100.0000 (89.5468) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1814 data: 0.0902 max mem: 3533 +Test: [ 580/2387] eta: 0:06:25 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.0832 (0.2919) acc1: 100.0000 (89.7268) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1892 data: 0.0963 max mem: 3533 +Test: [ 590/2387] eta: 0:06:22 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.0831 (0.2884) acc1: 100.0000 (89.9006) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1862 data: 0.0934 max mem: 3533 +Test: [ 600/2387] eta: 0:06:19 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.0865 (0.2855) acc1: 100.0000 (90.0665) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1828 data: 0.0921 max mem: 3533 +Test: [ 610/2387] eta: 0:06:16 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.1531 (0.2850) acc1: 97.6562 (90.1681) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1851 data: 0.0929 max mem: 3533 +Test: [ 620/2387] eta: 0:06:14 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.2615 (0.2847) acc1: 96.0938 (90.2648) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1971 data: 0.1018 max mem: 3533 +Test: [ 630/2387] eta: 0:06:11 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.2615 (0.2843) acc1: 96.6146 (90.3666) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1962 data: 0.1010 max mem: 3533 +Test: [ 640/2387] eta: 0:06:09 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.2622 (0.2840) acc1: 96.3542 (90.4536) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1903 data: 0.0976 max mem: 3533 +Test: [ 650/2387] eta: 0:06:06 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.2659 (0.2837) acc1: 95.5729 (90.5406) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1979 data: 0.1062 max mem: 3533 +Test: [ 660/2387] eta: 0:06:03 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.2655 (0.2834) acc1: 96.0938 (90.6270) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1844 data: 0.0928 max mem: 3533 +Test: [ 670/2387] eta: 0:06:00 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.2627 (0.2831) acc1: 96.3542 (90.7104) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1789 data: 0.0869 max mem: 3534 +Test: [ 680/2387] eta: 0:05:58 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.2662 (0.2831) acc1: 95.5729 (90.7619) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1899 data: 0.0978 max mem: 3534 +Test: [ 690/2387] eta: 0:05:55 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.2924 (0.2863) acc1: 91.9271 (90.5342) acc5: 100.0000 (100.0000) acc1_neg: 65.885 acc5_neg: 100.000 time: 0.1899 data: 0.0987 max mem: 3534 +Test: [ 700/2387] eta: 0:05:52 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.5696 (0.2904) acc1: 70.3125 (90.2431) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.1849 data: 0.0915 max mem: 3534 +Test: [ 710/2387] eta: 0:05:50 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.5805 (0.2945) acc1: 69.7917 (89.9485) acc5: 100.0000 (100.0000) acc1_neg: 69.531 acc5_neg: 100.000 time: 0.1822 data: 0.0872 max mem: 3534 +Test: [ 720/2387] eta: 0:05:47 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.5805 (0.2984) acc1: 69.2708 (89.6614) acc5: 100.0000 (100.0000) acc1_neg: 69.271 acc5_neg: 100.000 time: 0.1824 data: 0.0898 max mem: 3534 +Test: [ 730/2387] eta: 0:05:45 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.5763 (0.3023) acc1: 69.2708 (89.3892) acc5: 100.0000 (100.0000) acc1_neg: 72.396 acc5_neg: 100.000 time: 0.1882 data: 0.0960 max mem: 3534 +Test: [ 740/2387] eta: 0:05:43 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.5694 (0.3059) acc1: 69.5312 (89.1268) acc5: 100.0000 (100.0000) acc1_neg: 67.708 acc5_neg: 100.000 time: 0.2025 data: 0.1092 max mem: 3534 +Test: [ 750/2387] eta: 0:05:40 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.5694 (0.3094) acc1: 69.5312 (88.8742) acc5: 100.0000 (100.0000) acc1_neg: 69.010 acc5_neg: 100.000 time: 0.1993 data: 0.1053 max mem: 3534 +Test: [ 760/2387] eta: 0:05:37 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.5706 (0.3128) acc1: 69.5312 (88.6358) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.1756 data: 0.0816 max mem: 3534 +Test: [ 770/2387] eta: 0:05:34 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.5178 (0.3147) acc1: 73.9583 (88.5113) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1682 data: 0.0762 max mem: 3534 +Test: [ 780/2387] eta: 0:05:32 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.4011 (0.3157) acc1: 80.9896 (88.4396) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1794 data: 0.0850 max mem: 3534 +Test: [ 790/2387] eta: 0:05:29 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.3064 (0.3157) acc1: 90.1042 (88.4426) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.1785 data: 0.0844 max mem: 3534 +Test: [ 800/2387] eta: 0:05:26 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.3062 (0.3157) acc1: 90.1042 (88.4666) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.1712 data: 0.0805 max mem: 3534 +Test: [ 810/2387] eta: 0:05:24 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.3297 (0.3165) acc1: 86.9792 (88.4235) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.1730 data: 0.0758 max mem: 3534 +Test: [ 820/2387] eta: 0:05:21 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.3886 (0.3178) acc1: 84.1146 (88.3444) acc5: 100.0000 (100.0000) acc1_neg: 59.115 acc5_neg: 100.000 time: 0.1808 data: 0.0831 max mem: 3534 +Test: [ 830/2387] eta: 0:05:18 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.4922 (0.3229) acc1: 63.8021 (87.9569) acc5: 100.0000 (100.0000) acc1_neg: 57.812 acc5_neg: 100.000 time: 0.1720 data: 0.0801 max mem: 3534 +Test: [ 840/2387] eta: 0:05:15 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.5671 (0.3244) acc1: 65.3646 (87.8546) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.1643 data: 0.0730 max mem: 3534 +Test: [ 850/2387] eta: 0:05:13 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.4586 (0.3263) acc1: 77.3438 (87.7228) acc5: 100.0000 (100.0000) acc1_neg: 76.042 acc5_neg: 100.000 time: 0.1878 data: 0.0973 max mem: 3534 +Test: [ 860/2387] eta: 0:05:11 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.4503 (0.3276) acc1: 79.9479 (87.6428) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1929 data: 0.1021 max mem: 3534 +Test: [ 870/2387] eta: 0:05:09 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.4143 (0.3286) acc1: 80.4688 (87.5559) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.1820 data: 0.0901 max mem: 3534 +Test: [ 880/2387] eta: 0:05:06 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.4143 (0.3298) acc1: 78.3854 (87.4743) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.1902 data: 0.0965 max mem: 3534 +Test: [ 890/2387] eta: 0:05:04 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.4508 (0.3313) acc1: 78.1250 (87.3740) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.1968 data: 0.1034 max mem: 3534 +Test: [ 900/2387] eta: 0:05:02 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.5373 (0.3368) acc1: 71.0938 (87.0000) acc5: 100.0000 (100.0000) acc1_neg: 53.125 acc5_neg: 100.000 time: 0.1890 data: 0.0960 max mem: 3535 +Test: [ 910/2387] eta: 0:05:00 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.8892 (0.3430) acc1: 48.9583 (86.5855) acc5: 100.0000 (100.0000) acc1_neg: 50.521 acc5_neg: 100.000 time: 0.1909 data: 0.0980 max mem: 3535 +Test: [ 920/2387] eta: 0:04:58 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.9049 (0.3492) acc1: 47.6562 (86.1646) acc5: 100.0000 (100.0000) acc1_neg: 46.875 acc5_neg: 100.000 time: 0.2122 data: 0.1210 max mem: 3535 +Test: [ 930/2387] eta: 0:04:56 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.9049 (0.3550) acc1: 47.6562 (85.7711) acc5: 100.0000 (100.0000) acc1_neg: 44.792 acc5_neg: 100.000 time: 0.2151 data: 0.1229 max mem: 3535 +Test: [ 940/2387] eta: 0:04:54 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.8981 (0.3608) acc1: 49.2188 (85.3735) acc5: 100.0000 (100.0000) acc1_neg: 44.531 acc5_neg: 100.000 time: 0.2078 data: 0.1162 max mem: 3535 +Test: [ 950/2387] eta: 0:04:52 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.8951 (0.3664) acc1: 49.2188 (84.9920) acc5: 100.0000 (100.0000) acc1_neg: 51.042 acc5_neg: 100.000 time: 0.1938 data: 0.1030 max mem: 3535 +Test: [ 960/2387] eta: 0:04:49 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.8841 (0.3718) acc1: 49.2188 (84.6232) acc5: 100.0000 (100.0000) acc1_neg: 51.042 acc5_neg: 100.000 time: 0.1830 data: 0.0910 max mem: 3535 +Test: [ 970/2387] eta: 0:04:47 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.8894 (0.3773) acc1: 48.4375 (84.2503) acc5: 100.0000 (100.0000) acc1_neg: 48.438 acc5_neg: 100.000 time: 0.1922 data: 0.0982 max mem: 3535 +Test: [ 980/2387] eta: 0:04:45 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.8708 (0.3798) acc1: 49.2188 (84.0724) acc5: 100.0000 (100.0000) acc1_neg: 57.812 acc5_neg: 100.000 time: 0.2013 data: 0.1053 max mem: 3535 +Test: [ 990/2387] eta: 0:04:43 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.4076 (0.3794) acc1: 80.4688 (84.1038) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1902 data: 0.0949 max mem: 3535 +Test: [1000/2387] eta: 0:04:41 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.2566 (0.3783) acc1: 94.2708 (84.1835) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1801 data: 0.0870 max mem: 3535 +Test: [1010/2387] eta: 0:04:38 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.2918 (0.3787) acc1: 92.4479 (84.1695) acc5: 100.0000 (100.0000) acc1_neg: 70.573 acc5_neg: 100.000 time: 0.1823 data: 0.0880 max mem: 3535 +Test: [1020/2387] eta: 0:04:36 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.3602 (0.3791) acc1: 85.6771 (84.1424) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1783 data: 0.0836 max mem: 3535 +Test: [1030/2387] eta: 0:04:34 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.3408 (0.3789) acc1: 89.3229 (84.1517) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1782 data: 0.0861 max mem: 3535 +Test: [1040/2387] eta: 0:04:31 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.2649 (0.3776) acc1: 92.4479 (84.2572) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1815 data: 0.0896 max mem: 3535 +Test: [1050/2387] eta: 0:04:29 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.2365 (0.3763) acc1: 95.5729 (84.3537) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1841 data: 0.0930 max mem: 3535 +Test: [1060/2387] eta: 0:04:27 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.2496 (0.3755) acc1: 92.7083 (84.4234) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1852 data: 0.0939 max mem: 3535 +Test: [1070/2387] eta: 0:04:25 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.3020 (0.3754) acc1: 91.4062 (84.4341) acc5: 100.0000 (100.0000) acc1_neg: 64.323 acc5_neg: 100.000 time: 0.1789 data: 0.0867 max mem: 3535 +Test: [1080/2387] eta: 0:04:22 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.4762 (0.3777) acc1: 76.0417 (84.2666) acc5: 100.0000 (100.0000) acc1_neg: 77.344 acc5_neg: 100.000 time: 0.1795 data: 0.0868 max mem: 3535 +Test: [1090/2387] eta: 0:04:20 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.5039 (0.3784) acc1: 75.5208 (84.2055) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.1760 data: 0.0833 max mem: 3535 +Test: [1100/2387] eta: 0:04:18 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.3913 (0.3786) acc1: 82.5521 (84.1960) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1813 data: 0.0892 max mem: 3535 +Test: [1110/2387] eta: 0:04:16 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.3496 (0.3785) acc1: 84.3750 (84.1868) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1923 data: 0.0979 max mem: 3535 +Test: [1120/2387] eta: 0:04:13 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.3978 (0.3790) acc1: 82.8125 (84.1538) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.1820 data: 0.0878 max mem: 3536 +Test: [1130/2387] eta: 0:04:11 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.4081 (0.3794) acc1: 79.1667 (84.1282) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.1704 data: 0.0795 max mem: 3536 +Test: [1140/2387] eta: 0:04:09 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.4858 (0.3816) acc1: 76.5625 (83.9646) acc5: 100.0000 (100.0000) acc1_neg: 61.458 acc5_neg: 100.000 time: 0.1770 data: 0.0869 max mem: 3536 +Test: [1150/2387] eta: 0:04:07 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.5849 (0.3824) acc1: 69.0104 (83.8958) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.1959 data: 0.1049 max mem: 3536 +Test: [1160/2387] eta: 0:04:05 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.3033 (0.3816) acc1: 90.3646 (83.9695) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1943 data: 0.0998 max mem: 3536 +Test: [1170/2387] eta: 0:04:03 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.2758 (0.3808) acc1: 92.4479 (84.0372) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1833 data: 0.0892 max mem: 3536 +Test: [1180/2387] eta: 0:04:00 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.2853 (0.3801) acc1: 92.1875 (84.0912) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.1818 data: 0.0888 max mem: 3536 +Test: [1190/2387] eta: 0:03:58 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.2681 (0.3791) acc1: 93.2292 (84.1673) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1918 data: 0.0985 max mem: 3536 +Test: [1200/2387] eta: 0:03:56 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.2323 (0.3781) acc1: 94.0104 (84.2436) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1833 data: 0.0909 max mem: 3536 +Test: [1210/2387] eta: 0:03:54 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.2353 (0.3770) acc1: 94.5312 (84.3288) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1708 data: 0.0785 max mem: 3536 +Test: [1220/2387] eta: 0:03:52 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.1756 (0.3751) acc1: 97.9167 (84.4452) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1800 data: 0.0888 max mem: 3536 +Test: [1230/2387] eta: 0:03:50 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.1394 (0.3731) acc1: 99.2188 (84.5601) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1801 data: 0.0881 max mem: 3536 +Test: [1240/2387] eta: 0:03:47 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.1333 (0.3712) acc1: 99.2188 (84.6726) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1761 data: 0.0843 max mem: 3536 +Test: [1250/2387] eta: 0:03:45 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.1348 (0.3694) acc1: 98.9583 (84.7840) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1819 data: 0.0891 max mem: 3536 +Test: [1260/2387] eta: 0:03:43 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.1282 (0.3675) acc1: 98.9583 (84.8981) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1883 data: 0.0940 max mem: 3536 +Test: [1270/2387] eta: 0:03:41 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.1455 (0.3659) acc1: 99.2188 (85.0042) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1849 data: 0.0924 max mem: 3536 +Test: [1280/2387] eta: 0:03:39 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.2497 (0.3657) acc1: 94.0104 (85.0276) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.1859 data: 0.0942 max mem: 3536 +Test: [1290/2387] eta: 0:03:37 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.3732 (0.3661) acc1: 84.3750 (85.0042) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1849 data: 0.0924 max mem: 3536 +Test: [1300/2387] eta: 0:03:35 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.3770 (0.3660) acc1: 83.5938 (85.0089) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.1798 data: 0.0867 max mem: 3536 +Test: [1310/2387] eta: 0:03:32 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.3600 (0.3660) acc1: 84.8958 (85.0136) acc5: 100.0000 (100.0000) acc1_neg: 80.729 acc5_neg: 100.000 time: 0.1676 data: 0.0761 max mem: 3536 +Test: [1320/2387] eta: 0:03:30 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.3811 (0.3663) acc1: 83.8542 (84.9944) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1642 data: 0.0727 max mem: 3536 +Test: [1330/2387] eta: 0:03:28 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.4578 (0.3676) acc1: 76.0417 (84.8929) acc5: 100.0000 (100.0000) acc1_neg: 68.229 acc5_neg: 100.000 time: 0.1772 data: 0.0850 max mem: 3536 +Test: [1340/2387] eta: 0:03:26 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.5184 (0.3684) acc1: 69.2708 (84.8308) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1773 data: 0.0859 max mem: 3536 +Test: [1350/2387] eta: 0:03:24 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.1008 (0.3664) acc1: 100.0000 (84.9431) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1674 data: 0.0760 max mem: 3537 +Test: [1360/2387] eta: 0:03:22 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.0868 (0.3643) acc1: 100.0000 (85.0537) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1697 data: 0.0777 max mem: 3537 +Test: [1370/2387] eta: 0:03:19 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.0831 (0.3622) acc1: 100.0000 (85.1627) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1724 data: 0.0809 max mem: 3537 +Test: [1380/2387] eta: 0:03:17 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.0850 (0.3603) acc1: 100.0000 (85.2701) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1625 data: 0.0704 max mem: 3537 +Test: [1390/2387] eta: 0:03:15 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.0875 (0.3583) acc1: 100.0000 (85.3760) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1721 data: 0.0790 max mem: 3537 +Test: [1400/2387] eta: 0:03:13 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.0873 (0.3564) acc1: 100.0000 (85.4799) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1821 data: 0.0904 max mem: 3537 +Test: [1410/2387] eta: 0:03:11 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.0938 (0.3565) acc1: 100.0000 (85.4591) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.1736 data: 0.0830 max mem: 3537 +Test: [1420/2387] eta: 0:03:09 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.3439 (0.3565) acc1: 85.6771 (85.4559) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1734 data: 0.0811 max mem: 3537 +Test: [1430/2387] eta: 0:03:07 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.3310 (0.3567) acc1: 85.9375 (85.4374) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1795 data: 0.0859 max mem: 3537 +Test: [1440/2387] eta: 0:03:05 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.3917 (0.3572) acc1: 79.9479 (85.3957) acc5: 100.0000 (100.0000) acc1_neg: 57.292 acc5_neg: 100.000 time: 0.1866 data: 0.0933 max mem: 3537 +Test: [1450/2387] eta: 0:03:03 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.4749 (0.3589) acc1: 69.5312 (85.2666) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1825 data: 0.0882 max mem: 3537 +Test: [1460/2387] eta: 0:03:01 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.2795 (0.3581) acc1: 88.0208 (85.3247) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1698 data: 0.0760 max mem: 3537 +Test: [1470/2387] eta: 0:02:58 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.2158 (0.3571) acc1: 95.5729 (85.3979) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1679 data: 0.0760 max mem: 3537 +Test: [1480/2387] eta: 0:02:56 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.2056 (0.3561) acc1: 95.3125 (85.4620) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1644 data: 0.0735 max mem: 3537 +Test: [1490/2387] eta: 0:02:54 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.2049 (0.3551) acc1: 95.3125 (85.5360) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1655 data: 0.0736 max mem: 3537 +Test: [1500/2387] eta: 0:02:52 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.2190 (0.3541) acc1: 95.3125 (85.6006) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.1696 data: 0.0777 max mem: 3537 +Test: [1510/2387] eta: 0:02:50 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.2345 (0.3537) acc1: 94.5312 (85.6542) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1600 data: 0.0660 max mem: 3537 +Test: [1520/2387] eta: 0:02:48 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.3241 (0.3539) acc1: 88.8021 (85.6434) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1571 data: 0.0623 max mem: 3537 +Test: [1530/2387] eta: 0:02:46 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.4017 (0.3545) acc1: 83.0729 (85.6094) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.1595 data: 0.0680 max mem: 3537 +Test: [1540/2387] eta: 0:02:43 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.4043 (0.3548) acc1: 82.2917 (85.5973) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.1585 data: 0.0656 max mem: 3537 +Test: [1550/2387] eta: 0:02:41 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.3813 (0.3550) acc1: 83.3333 (85.5886) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.1601 data: 0.0666 max mem: 3537 +Test: [1560/2387] eta: 0:02:39 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.3781 (0.3552) acc1: 84.6354 (85.5837) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.1613 data: 0.0697 max mem: 3537 +Test: [1570/2387] eta: 0:02:37 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.3845 (0.3554) acc1: 84.3750 (85.5791) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.1593 data: 0.0679 max mem: 3537 +Test: [1580/2387] eta: 0:02:35 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.3698 (0.3554) acc1: 86.4583 (85.5827) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.1550 data: 0.0602 max mem: 3538 +Test: [1590/2387] eta: 0:02:33 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.2519 (0.3546) acc1: 92.7083 (85.6463) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1705 data: 0.0760 max mem: 3538 +Test: [1600/2387] eta: 0:02:31 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.1982 (0.3538) acc1: 97.9167 (85.6929) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1876 data: 0.0962 max mem: 3538 +Test: [1610/2387] eta: 0:02:29 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.2023 (0.3531) acc1: 96.6146 (85.7480) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1838 data: 0.0926 max mem: 3538 +Test: [1620/2387] eta: 0:02:27 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.2425 (0.3526) acc1: 92.7083 (85.7799) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1680 data: 0.0766 max mem: 3538 +Test: [1630/2387] eta: 0:02:25 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.3041 (0.3534) acc1: 89.5833 (85.7290) acc5: 100.0000 (100.0000) acc1_neg: 63.281 acc5_neg: 100.000 time: 0.1642 data: 0.0728 max mem: 3538 +Test: [1640/2387] eta: 0:02:23 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.6269 (0.3555) acc1: 64.0625 (85.5876) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.1698 data: 0.0797 max mem: 3538 +Test: [1650/2387] eta: 0:02:21 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.7005 (0.3576) acc1: 59.3750 (85.4376) acc5: 100.0000 (100.0000) acc1_neg: 62.760 acc5_neg: 100.000 time: 0.1760 data: 0.0830 max mem: 3538 +Test: [1660/2387] eta: 0:02:19 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.6821 (0.3594) acc1: 62.7604 (85.3121) acc5: 100.0000 (100.0000) acc1_neg: 76.302 acc5_neg: 100.000 time: 0.1623 data: 0.0692 max mem: 3538 +Test: [1670/2387] eta: 0:02:17 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.5944 (0.3608) acc1: 65.8854 (85.2142) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.1443 data: 0.0526 max mem: 3538 +Test: [1680/2387] eta: 0:02:15 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.5603 (0.3622) acc1: 69.7917 (85.1107) acc5: 100.0000 (100.0000) acc1_neg: 69.792 acc5_neg: 100.000 time: 0.1546 data: 0.0587 max mem: 3538 +Test: [1690/2387] eta: 0:02:13 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.6168 (0.3638) acc1: 66.6667 (84.9981) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.1641 data: 0.0685 max mem: 3538 +Test: [1700/2387] eta: 0:02:11 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.6173 (0.3647) acc1: 66.1458 (84.9356) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1671 data: 0.0741 max mem: 3538 +Test: [1710/2387] eta: 0:02:09 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.1964 (0.3635) acc1: 96.8750 (85.0114) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1641 data: 0.0699 max mem: 3538 +Test: [1720/2387] eta: 0:02:07 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.1500 (0.3623) acc1: 98.4375 (85.0883) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1632 data: 0.0689 max mem: 3538 +Test: [1730/2387] eta: 0:02:05 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.1457 (0.3611) acc1: 98.6979 (85.1612) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1659 data: 0.0737 max mem: 3538 +Test: [1740/2387] eta: 0:02:03 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.1551 (0.3600) acc1: 98.4375 (85.2351) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1669 data: 0.0751 max mem: 3538 +Test: [1750/2387] eta: 0:02:01 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.1480 (0.3587) acc1: 99.2188 (85.3152) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1651 data: 0.0733 max mem: 3538 +Test: [1760/2387] eta: 0:01:59 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.1561 (0.3579) acc1: 98.4375 (85.3757) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1610 data: 0.0689 max mem: 3538 +Test: [1770/2387] eta: 0:01:57 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.2012 (0.3573) acc1: 95.3125 (85.4189) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1604 data: 0.0687 max mem: 3538 +Test: [1780/2387] eta: 0:01:55 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.2235 (0.3565) acc1: 95.5729 (85.4759) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1585 data: 0.0659 max mem: 3538 +Test: [1790/2387] eta: 0:01:53 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.2032 (0.3555) acc1: 97.3958 (85.5468) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1691 data: 0.0746 max mem: 3538 +Test: [1800/2387] eta: 0:01:51 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.1782 (0.3545) acc1: 97.6562 (85.6075) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1615 data: 0.0682 max mem: 3538 +Test: [1810/2387] eta: 0:01:49 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.1923 (0.3536) acc1: 97.1354 (85.6712) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1660 data: 0.0741 max mem: 3539 +Test: [1820/2387] eta: 0:01:47 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.1834 (0.3528) acc1: 97.1354 (85.7266) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1921 data: 0.0995 max mem: 3539 +Test: [1830/2387] eta: 0:01:45 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.1842 (0.3521) acc1: 95.8333 (85.7803) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1879 data: 0.0957 max mem: 3539 +Test: [1840/2387] eta: 0:01:43 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.1697 (0.3511) acc1: 98.9583 (85.8545) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1888 data: 0.0981 max mem: 3539 +Test: [1850/2387] eta: 0:01:41 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.1686 (0.3501) acc1: 99.4792 (85.9278) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1759 data: 0.0838 max mem: 3539 +Test: [1860/2387] eta: 0:01:39 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.1680 (0.3491) acc1: 99.4792 (85.9992) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1641 data: 0.0720 max mem: 3539 +Test: [1870/2387] eta: 0:01:37 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.1667 (0.3482) acc1: 99.4792 (86.0714) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1670 data: 0.0761 max mem: 3539 +Test: [1880/2387] eta: 0:01:35 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.1667 (0.3472) acc1: 99.4792 (86.1417) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1619 data: 0.0711 max mem: 3539 +Test: [1890/2387] eta: 0:01:33 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.1664 (0.3462) acc1: 99.4792 (86.2121) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1669 data: 0.0764 max mem: 3539 +Test: [1900/2387] eta: 0:01:31 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.1680 (0.3454) acc1: 99.4792 (86.2793) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1723 data: 0.0816 max mem: 3539 +Test: [1910/2387] eta: 0:01:29 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.2476 (0.3450) acc1: 96.6146 (86.3221) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.1649 data: 0.0741 max mem: 3539 +Test: [1920/2387] eta: 0:01:27 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.3042 (0.3457) acc1: 89.8438 (86.2715) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.1587 data: 0.0664 max mem: 3539 +Test: [1930/2387] eta: 0:01:26 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.3168 (0.3455) acc1: 87.2396 (86.3041) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1613 data: 0.0693 max mem: 3539 +Test: [1940/2387] eta: 0:01:24 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.2821 (0.3455) acc1: 93.4896 (86.3048) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1600 data: 0.0696 max mem: 3539 +Test: [1950/2387] eta: 0:01:22 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.3058 (0.3457) acc1: 92.1875 (86.3076) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.1609 data: 0.0706 max mem: 3539 +Test: [1960/2387] eta: 0:01:20 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.3979 (0.3462) acc1: 81.5104 (86.2641) acc5: 100.0000 (100.0000) acc1_neg: 60.417 acc5_neg: 100.000 time: 0.1599 data: 0.0686 max mem: 3539 +Test: [1970/2387] eta: 0:01:18 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.4985 (0.3480) acc1: 69.0104 (86.1247) acc5: 100.0000 (100.0000) acc1_neg: 69.010 acc5_neg: 100.000 time: 0.1525 data: 0.0607 max mem: 3539 +Test: [1980/2387] eta: 0:01:16 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.5968 (0.3485) acc1: 63.2812 (86.0751) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1584 data: 0.0668 max mem: 3539 +Test: [1990/2387] eta: 0:01:14 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.1709 (0.3476) acc1: 97.6562 (86.1347) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1667 data: 0.0752 max mem: 3539 +Test: [2000/2387] eta: 0:01:12 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.1709 (0.3467) acc1: 98.1771 (86.1947) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1642 data: 0.0725 max mem: 3539 +Test: [2010/2387] eta: 0:01:10 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.1752 (0.3459) acc1: 97.6562 (86.2511) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1642 data: 0.0726 max mem: 3539 +Test: [2020/2387] eta: 0:01:08 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.1761 (0.3450) acc1: 97.9167 (86.3102) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1643 data: 0.0716 max mem: 3539 +Test: [2030/2387] eta: 0:01:06 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.1720 (0.3442) acc1: 98.1771 (86.3679) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1642 data: 0.0721 max mem: 3540 +Test: [2040/2387] eta: 0:01:04 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.1713 (0.3433) acc1: 98.1771 (86.4268) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1678 data: 0.0780 max mem: 3540 +Test: [2050/2387] eta: 0:01:02 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.1709 (0.3425) acc1: 98.1771 (86.4840) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1651 data: 0.0751 max mem: 3540 +Test: [2060/2387] eta: 0:01:01 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.1811 (0.3417) acc1: 97.9167 (86.5374) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1614 data: 0.0711 max mem: 3540 +Test: [2070/2387] eta: 0:00:59 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.1809 (0.3409) acc1: 98.1771 (86.5951) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1643 data: 0.0732 max mem: 3540 +Test: [2080/2387] eta: 0:00:57 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.1792 (0.3402) acc1: 98.4375 (86.6506) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1634 data: 0.0712 max mem: 3540 +Test: [2090/2387] eta: 0:00:55 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.1763 (0.3394) acc1: 98.1771 (86.7047) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1570 data: 0.0657 max mem: 3540 +Test: [2100/2387] eta: 0:00:53 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.1776 (0.3386) acc1: 98.1771 (86.7588) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1558 data: 0.0658 max mem: 3540 +Test: [2110/2387] eta: 0:00:51 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.1794 (0.3379) acc1: 98.1771 (86.8118) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1578 data: 0.0654 max mem: 3540 +Test: [2120/2387] eta: 0:00:49 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.1953 (0.3379) acc1: 96.3542 (86.8128) acc5: 100.0000 (100.0000) acc1_neg: 66.146 acc5_neg: 100.000 time: 0.1619 data: 0.0683 max mem: 3540 +Test: [2130/2387] eta: 0:00:47 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.2400 (0.3373) acc1: 95.8333 (86.8615) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1661 data: 0.0746 max mem: 3540 +Test: [2140/2387] eta: 0:00:45 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.2016 (0.3369) acc1: 96.8750 (86.8834) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1781 data: 0.0873 max mem: 3540 +Test: [2150/2387] eta: 0:00:44 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.2427 (0.3366) acc1: 94.0104 (86.9097) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1789 data: 0.0866 max mem: 3540 +Test: [2160/2387] eta: 0:00:42 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.2803 (0.3365) acc1: 91.9271 (86.9227) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1722 data: 0.0784 max mem: 3540 +Test: [2170/2387] eta: 0:00:40 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.3382 (0.3373) acc1: 86.7188 (86.8682) acc5: 100.0000 (100.0000) acc1_neg: 65.104 acc5_neg: 100.000 time: 0.1719 data: 0.0797 max mem: 3540 +Test: [2180/2387] eta: 0:00:38 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.6035 (0.3394) acc1: 66.4062 (86.7205) acc5: 100.0000 (100.0000) acc1_neg: 33.333 acc5_neg: 100.000 time: 0.1685 data: 0.0757 max mem: 3540 +Test: [2190/2387] eta: 0:00:36 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.5136 (0.3401) acc1: 70.3125 (86.6826) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1676 data: 0.0738 max mem: 3540 +Test: [2200/2387] eta: 0:00:34 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.3236 (0.3399) acc1: 88.2812 (86.7060) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1679 data: 0.0747 max mem: 3540 +Test: [2210/2387] eta: 0:00:32 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.3013 (0.3397) acc1: 90.3646 (86.7153) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1648 data: 0.0710 max mem: 3540 +Test: [2220/2387] eta: 0:00:30 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.2934 (0.3395) acc1: 91.9271 (86.7408) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1661 data: 0.0728 max mem: 3540 +Test: [2230/2387] eta: 0:00:29 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.2731 (0.3391) acc1: 93.2292 (86.7717) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1689 data: 0.0765 max mem: 3540 +Test: [2240/2387] eta: 0:00:27 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.2477 (0.3389) acc1: 92.7083 (86.7914) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1704 data: 0.0774 max mem: 3540 +Test: [2250/2387] eta: 0:00:25 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.3111 (0.3391) acc1: 88.8021 (86.7834) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1779 data: 0.0860 max mem: 3540 +Test: [2260/2387] eta: 0:00:23 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.3111 (0.3390) acc1: 88.2812 (86.7976) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1748 data: 0.0840 max mem: 3541 +Test: [2270/2387] eta: 0:00:21 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.2279 (0.3386) acc1: 96.0938 (86.8299) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1725 data: 0.0804 max mem: 3541 +Test: [2280/2387] eta: 0:00:19 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.1904 (0.3381) acc1: 97.1354 (86.8588) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1737 data: 0.0801 max mem: 3541 +Test: [2290/2387] eta: 0:00:17 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.1802 (0.3376) acc1: 97.6562 (86.8958) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1744 data: 0.0818 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.2323 (0.3372) acc1: 94.5312 (86.9246) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1801 data: 0.0873 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.2902 (0.3375) acc1: 89.0625 (86.9084) acc5: 100.0000 (100.0000) acc1_neg: 55.469 acc5_neg: 100.000 time: 0.2048 data: 0.1105 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.4471 (0.3383) acc1: 77.8646 (86.8481) acc5: 100.0000 (100.0000) acc1_neg: 72.396 acc5_neg: 100.000 time: 0.2033 data: 0.1102 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.5042 (0.3392) acc1: 74.7396 (86.7837) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.1724 data: 0.0804 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.4760 (0.3398) acc1: 76.0417 (86.7430) acc5: 100.0000 (100.0000) acc1_neg: 76.042 acc5_neg: 100.000 time: 0.1677 data: 0.0763 max mem: 3541 +Test: [2350/2387] eta: 0:00:06 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.4388 (0.3402) acc1: 79.4271 (86.7063) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.1786 data: 0.0877 max mem: 3541 +Test: [2360/2387] eta: 0:00:04 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.4625 (0.3413) acc1: 76.3021 (86.6319) acc5: 100.0000 (100.0000) acc1_neg: 72.656 acc5_neg: 100.000 time: 0.1871 data: 0.0950 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.5595 (0.3422) acc1: 68.2292 (86.5577) acc5: 100.0000 (100.0000) acc1_neg: 65.365 acc5_neg: 100.000 time: 0.1770 data: 0.0838 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.6723 (0.3442) acc1: 57.5521 (86.4182) acc5: 100.0000 (100.0000) acc1_neg: 46.094 acc5_neg: 100.000 time: 0.1568 data: 0.0641 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 73.973 acc5_pos: 100.000 loss: 0.7673 (0.3453) acc1: 53.3854 (86.3452) acc5: 100.0000 (100.0000) acc1_neg: 65.116 acc5_neg: 100.000 time: 0.1316 data: 0.0420 max mem: 3541 +Test: Total time: 0:07:20 (0.1846 s / it) +* Overall Acc@1 86.345 Acc@5 100.000 + target == real Acc@1 73.039 Acc@5 100.000 + target == fake Acc@1 87.280 Acc@5 100.000 + loss 0.345 + PR-AUC 0.9916 ROC-AUC 0.8968 EER 17.86% (thr=0.5884) +Accuracy of the network on the 916267 test images: 86.3% +[rank0]:[W725 18:47:14.801856765 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:47:38.022000 2576867 torch/distributed/run.py:766] +W0725 18:47:38.022000 2576867 torch/distributed/run.py:766] ***************************************** +W0725 18:47:38.022000 2576867 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:47:38.022000 2576867 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:48:11.329264006 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:48:11.816122076 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:48:11.024587595 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +NCCL version 2.26.2+cuda12.2 +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:48:11.160633680 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:48:12.239967234 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:48:12.314912357 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:48:12.513931045 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W725 18:48:12.515395933 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed4_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed4_resampled_n116-040-168_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 0:50:18 acc1_pos: 75.521 acc5_pos: 100.000 loss: 0.5186 (0.5186) acc1: 75.5208 (75.5208) acc5: 100.0000 (100.0000) time: 5.8385 data: 3.5728 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/517] eta: 0:05:10 acc1_pos: 46.615 acc5_pos: 100.000 loss: 0.6607 (0.6851) acc1: 61.9792 (61.2453) acc5: 100.0000 (100.0000) time: 0.6125 data: 0.3252 max mem: 3531 +Test: [ 20/517] eta: 0:03:01 acc1_pos: 48.698 acc5_pos: 100.000 loss: 0.6715 (0.7170) acc1: 58.0729 (57.7381) acc5: 100.0000 (100.0000) time: 0.0915 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:02:41 acc1_pos: 44.531 acc5_pos: 100.000 loss: 0.7224 (0.7087) acc1: 54.1667 (58.1233) acc5: 100.0000 (100.0000) time: 0.1766 data: 0.0834 max mem: 3531 +Test: [ 40/517] eta: 0:02:41 acc1_pos: 71.615 acc5_pos: 100.000 loss: 0.6437 (0.6839) acc1: 61.9792 (60.3659) acc5: 100.0000 (100.0000) time: 0.3099 data: 0.2142 max mem: 3531 +Test: [ 50/517] eta: 0:02:44 acc1_pos: 72.018 acc5_pos: 100.000 loss: 0.6005 (0.6871) acc1: 67.1875 (60.1614) acc5: 100.0000 (100.0000) acc1_neg: 38.021 acc5_neg: 100.000 time: 0.3861 data: 0.2898 max mem: 3531 +Test: [ 60/517] eta: 0:02:33 acc1_pos: 72.018 acc5_pos: 100.000 loss: 0.8531 (0.7242) acc1: 41.1458 (56.5019) acc5: 100.0000 (100.0000) acc1_neg: 41.146 acc5_neg: 100.000 time: 0.3289 data: 0.2348 max mem: 3531 +Test: [ 70/517] eta: 0:02:21 acc1_pos: 72.018 acc5_pos: 100.000 loss: 0.7799 (0.7176) acc1: 47.9167 (56.7011) acc5: 100.0000 (100.0000) acc1_neg: 47.917 acc5_neg: 100.000 time: 0.2237 data: 0.1313 max mem: 3531 +Test: [ 80/517] eta: 0:02:13 acc1_pos: 72.018 acc5_pos: 100.000 loss: 0.7199 (0.7206) acc1: 52.6042 (55.9606) acc5: 100.0000 (100.0000) acc1_neg: 50.260 acc5_neg: 100.000 time: 0.2159 data: 0.1224 max mem: 3531 +Test: [ 90/517] eta: 0:02:07 acc1_pos: 72.018 acc5_pos: 100.000 loss: 0.7876 (0.7328) acc1: 50.2604 (55.3428) acc5: 100.0000 (100.0000) acc1_neg: 52.083 acc5_neg: 100.000 time: 0.2349 data: 0.1402 max mem: 3531 +Test: [100/517] eta: 0:02:01 acc1_pos: 72.018 acc5_pos: 100.000 loss: 0.8534 (0.7483) acc1: 44.0104 (53.4086) acc5: 100.0000 (100.0000) acc1_neg: 57.031 acc5_neg: 100.000 time: 0.2355 data: 0.1422 max mem: 3531 +Test: [110/517] eta: 0:01:56 acc1_pos: 72.018 acc5_pos: 100.000 loss: 0.6370 (0.7341) acc1: 57.0312 (54.3590) acc5: 100.0000 (100.0000) acc1_neg: 72.135 acc5_neg: 100.000 time: 0.2299 data: 0.1365 max mem: 3531 +Test: [120/517] eta: 0:01:52 acc1_pos: 72.018 acc5_pos: 100.000 loss: 0.6608 (0.7348) acc1: 56.2500 (54.0203) acc5: 100.0000 (100.0000) acc1_neg: 48.177 acc5_neg: 100.000 time: 0.2343 data: 0.1387 max mem: 3531 +Test: [130/517] eta: 0:01:47 acc1_pos: 72.018 acc5_pos: 100.000 loss: 0.7326 (0.7348) acc1: 49.7396 (53.7611) acc5: 100.0000 (100.0000) acc1_neg: 49.740 acc5_neg: 100.000 time: 0.2313 data: 0.1355 max mem: 3531 +Test: [140/517] eta: 0:01:44 acc1_pos: 72.018 acc5_pos: 100.000 loss: 0.7628 (0.7391) acc1: 46.0938 (52.9883) acc5: 100.0000 (100.0000) acc1_neg: 49.740 acc5_neg: 100.000 time: 0.2404 data: 0.1484 max mem: 3531 +Test: [150/517] eta: 0:01:40 acc1_pos: 72.018 acc5_pos: 100.000 loss: 0.7736 (0.7359) acc1: 46.0938 (53.2233) acc5: 100.0000 (100.0000) acc1_neg: 62.240 acc5_neg: 100.000 time: 0.2486 data: 0.1567 max mem: 3531 +Test: [160/517] eta: 0:01:36 acc1_pos: 72.018 acc5_pos: 100.000 loss: 0.5978 (0.7271) acc1: 65.8854 (54.2799) acc5: 100.0000 (100.0000) acc1_neg: 77.083 acc5_neg: 100.000 time: 0.2344 data: 0.1403 max mem: 3531 +Test: [170/517] eta: 0:01:33 acc1_pos: 72.018 acc5_pos: 100.000 loss: 0.5634 (0.7164) acc1: 73.9583 (55.4215) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.2349 data: 0.1387 max mem: 3531 +Test: [180/517] eta: 0:01:30 acc1_pos: 72.018 acc5_pos: 100.000 loss: 0.5388 (0.7190) acc1: 69.2708 (55.0055) acc5: 100.0000 (100.0000) acc1_neg: 46.354 acc5_neg: 100.000 time: 0.2362 data: 0.1397 max mem: 3531 +Test: [190/517] eta: 0:01:26 acc1_pos: 72.018 acc5_pos: 100.000 loss: 0.5645 (0.7099) acc1: 69.2708 (55.9323) acc5: 100.0000 (100.0000) acc1_neg: 63.281 acc5_neg: 100.000 time: 0.2261 data: 0.1324 max mem: 3531 +Test: [200/517] eta: 0:01:23 acc1_pos: 72.018 acc5_pos: 100.000 loss: 0.6065 (0.7069) acc1: 64.8438 (55.9935) acc5: 100.0000 (100.0000) acc1_neg: 44.531 acc5_neg: 100.000 time: 0.2286 data: 0.1356 max mem: 3531 +Test: [210/517] eta: 0:01:20 acc1_pos: 72.018 acc5_pos: 100.000 loss: 0.6891 (0.7069) acc1: 62.7604 (56.4191) acc5: 100.0000 (100.0000) acc1_neg: 62.760 acc5_neg: 100.000 time: 0.2387 data: 0.1433 max mem: 3532 +Test: [220/517] eta: 0:01:17 acc1_pos: 72.018 acc5_pos: 100.000 loss: 0.6891 (0.7060) acc1: 67.1875 (56.8840) acc5: 100.0000 (100.0000) acc1_neg: 69.531 acc5_neg: 100.000 time: 0.2399 data: 0.1431 max mem: 3532 +Test: [230/517] eta: 0:01:14 acc1_pos: 72.018 acc5_pos: 100.000 loss: 0.7297 (0.7117) acc1: 59.3750 (56.3210) acc5: 100.0000 (100.0000) acc1_neg: 62.240 acc5_neg: 100.000 time: 0.2354 data: 0.1373 max mem: 3532 +Test: [240/517] eta: 0:01:11 acc1_pos: 72.018 acc5_pos: 100.000 loss: 0.6565 (0.7078) acc1: 55.7292 (56.5720) acc5: 100.0000 (100.0000) acc1_neg: 54.688 acc5_neg: 100.000 time: 0.2349 data: 0.1372 max mem: 3532 +Test: [250/517] eta: 0:01:08 acc1_pos: 72.018 acc5_pos: 100.000 loss: 0.6196 (0.7043) acc1: 62.2396 (56.9099) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.2355 data: 0.1399 max mem: 3532 +Test: [260/517] eta: 0:01:06 acc1_pos: 72.018 acc5_pos: 100.000 loss: 0.6194 (0.7010) acc1: 65.6250 (57.2538) acc5: 100.0000 (100.0000) acc1_neg: 62.240 acc5_neg: 100.000 time: 0.2450 data: 0.1511 max mem: 3532 +Test: [270/517] eta: 0:01:03 acc1_pos: 72.018 acc5_pos: 100.000 loss: 0.6346 (0.6986) acc1: 62.7604 (57.4147) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.2540 data: 0.1616 max mem: 3532 +Test: [280/517] eta: 0:01:00 acc1_pos: 72.018 acc5_pos: 100.000 loss: 0.6193 (0.6965) acc1: 61.1979 (57.6077) acc5: 100.0000 (100.0000) acc1_neg: 65.625 acc5_neg: 100.000 time: 0.2448 data: 0.1499 max mem: 3532 +Test: [290/517] eta: 0:00:58 acc1_pos: 72.018 acc5_pos: 100.000 loss: 0.6476 (0.6955) acc1: 61.1979 (57.7087) acc5: 100.0000 (100.0000) acc1_neg: 52.083 acc5_neg: 100.000 time: 0.2397 data: 0.1437 max mem: 3532 +Test: [300/517] eta: 0:00:55 acc1_pos: 72.018 acc5_pos: 100.000 loss: 0.6284 (0.6915) acc1: 62.5000 (58.0487) acc5: 100.0000 (100.0000) acc1_neg: 36.979 acc5_neg: 100.000 time: 0.2399 data: 0.1440 max mem: 3532 +Test: [310/517] eta: 0:00:52 acc1_pos: 72.018 acc5_pos: 100.000 loss: 0.8434 (0.6989) acc1: 44.0104 (57.4231) acc5: 100.0000 (100.0000) acc1_neg: 41.927 acc5_neg: 100.000 time: 0.2399 data: 0.1420 max mem: 3532 +Test: [320/517] eta: 0:00:50 acc1_pos: 72.018 acc5_pos: 100.000 loss: 0.8907 (0.6993) acc1: 41.9271 (57.4783) acc5: 100.0000 (100.0000) acc1_neg: 68.750 acc5_neg: 100.000 time: 0.2549 data: 0.1571 max mem: 3532 +Test: [330/517] eta: 0:00:47 acc1_pos: 72.018 acc5_pos: 100.000 loss: 0.5546 (0.6945) acc1: 69.7917 (57.9211) acc5: 100.0000 (100.0000) acc1_neg: 68.229 acc5_neg: 100.000 time: 0.2455 data: 0.1497 max mem: 3532 +Test: [340/517] eta: 0:00:44 acc1_pos: 72.018 acc5_pos: 100.000 loss: 0.6190 (0.7000) acc1: 66.1458 (57.5429) acc5: 100.0000 (100.0000) acc1_neg: 15.365 acc5_neg: 100.000 time: 0.2264 data: 0.1322 max mem: 3532 +Test: [350/517] eta: 0:00:42 acc1_pos: 72.018 acc5_pos: 100.000 loss: 0.8457 (0.7081) acc1: 42.4479 (56.8339) acc5: 100.0000 (100.0000) acc1_neg: 26.302 acc5_neg: 100.000 time: 0.2349 data: 0.1397 max mem: 3532 +Test: [360/517] eta: 0:00:39 acc1_pos: 72.018 acc5_pos: 100.000 loss: 0.8765 (0.7130) acc1: 39.0625 (56.3972) acc5: 100.0000 (100.0000) acc1_neg: 39.583 acc5_neg: 100.000 time: 0.2384 data: 0.1401 max mem: 3532 +Test: [370/517] eta: 0:00:37 acc1_pos: 72.018 acc5_pos: 100.000 loss: 0.8131 (0.7138) acc1: 45.3125 (56.3588) acc5: 100.0000 (100.0000) acc1_neg: 40.104 acc5_neg: 100.000 time: 0.2348 data: 0.1375 max mem: 3532 +Test: [380/517] eta: 0:00:34 acc1_pos: 72.018 acc5_pos: 100.000 loss: 0.7621 (0.7168) acc1: 49.2188 (56.0750) acc5: 100.0000 (100.0000) acc1_neg: 71.615 acc5_neg: 100.000 time: 0.2399 data: 0.1443 max mem: 3532 +Test: [390/517] eta: 0:00:32 acc1_pos: 72.018 acc5_pos: 100.000 loss: 0.6532 (0.7136) acc1: 61.1979 (56.3432) acc5: 100.0000 (100.0000) acc1_neg: 62.760 acc5_neg: 100.000 time: 0.2494 data: 0.1538 max mem: 3532 +Test: [400/517] eta: 0:00:29 acc1_pos: 72.018 acc5_pos: 100.000 loss: 0.6911 (0.7181) acc1: 57.8125 (55.9422) acc5: 100.0000 (100.0000) acc1_neg: 21.875 acc5_neg: 100.000 time: 0.2449 data: 0.1488 max mem: 3532 +Test: [410/517] eta: 0:00:26 acc1_pos: 72.018 acc5_pos: 100.000 loss: 0.9283 (0.7286) acc1: 32.5521 (55.0658) acc5: 100.0000 (100.0000) acc1_neg: 8.594 acc5_neg: 100.000 time: 0.2399 data: 0.1413 max mem: 3532 +Test: [420/517] eta: 0:00:24 acc1_pos: 72.018 acc5_pos: 100.000 loss: 0.8803 (0.7295) acc1: 30.7292 (54.8428) acc5: 100.0000 (100.0000) acc1_neg: 70.833 acc5_neg: 100.000 time: 0.2451 data: 0.1479 max mem: 3532 +Test: [430/517] eta: 0:00:21 acc1_pos: 72.018 acc5_pos: 100.000 loss: 0.6061 (0.7278) acc1: 55.9896 (55.0555) acc5: 100.0000 (100.0000) acc1_neg: 24.479 acc5_neg: 100.000 time: 0.2499 data: 0.1526 max mem: 3532 +Test: [440/517] eta: 0:00:19 acc1_pos: 72.018 acc5_pos: 100.000 loss: 0.9799 (0.7351) acc1: 39.0625 (54.5074) acc5: 100.0000 (100.0000) acc1_neg: 33.333 acc5_neg: 100.000 time: 0.2446 data: 0.1474 max mem: 3533 +Test: [450/517] eta: 0:00:16 acc1_pos: 72.018 acc5_pos: 100.000 loss: 0.9058 (0.7369) acc1: 35.1562 (54.2510) acc5: 100.0000 (100.0000) acc1_neg: 64.844 acc5_neg: 100.000 time: 0.2404 data: 0.1457 max mem: 3533 +Test: [460/517] eta: 0:00:14 acc1_pos: 72.018 acc5_pos: 100.000 loss: 0.8532 (0.7402) acc1: 35.9375 (53.9814) acc5: 100.0000 (100.0000) acc1_neg: 34.375 acc5_neg: 100.000 time: 0.2468 data: 0.1527 max mem: 3533 +Test: [470/517] eta: 0:00:11 acc1_pos: 72.018 acc5_pos: 100.000 loss: 0.8172 (0.7421) acc1: 39.0625 (53.7354) acc5: 100.0000 (100.0000) acc1_neg: 38.542 acc5_neg: 100.000 time: 0.2544 data: 0.1619 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 72.018 acc5_pos: 100.000 loss: 0.7859 (0.7415) acc1: 44.7917 (53.6816) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.2493 data: 0.1552 max mem: 3533 +Test: [490/517] eta: 0:00:06 acc1_pos: 72.018 acc5_pos: 100.000 loss: 0.7037 (0.7409) acc1: 51.3021 (53.6983) acc5: 100.0000 (100.0000) acc1_neg: 63.802 acc5_neg: 100.000 time: 0.2413 data: 0.1465 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 72.018 acc5_pos: 100.000 loss: 0.7149 (0.7410) acc1: 52.3438 (53.6318) acc5: 100.0000 (100.0000) acc1_neg: 52.865 acc5_neg: 100.000 time: 0.2452 data: 0.1532 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 72.018 acc5_pos: 100.000 loss: 0.7375 (0.7428) acc1: 49.4792 (53.5796) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.2254 data: 0.1344 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 72.018 acc5_pos: 100.000 loss: 0.6522 (0.7414) acc1: 59.6354 (53.6968) acc5: 100.0000 (100.0000) acc1_neg: 61.131 acc5_neg: 100.000 time: 0.1736 data: 0.0813 max mem: 3533 +Test: Total time: 0:02:08 (0.2494 s / it) +* Overall Acc@1 53.697 Acc@5 100.000 + target == real Acc@1 61.335 Acc@5 100.000 + target == fake Acc@1 52.904 Acc@5 100.000 + loss 0.741 + PR-AUC 0.9287 ROC-AUC 0.6018 EER 42.71% (thr=0.4727) +Accuracy of the network on the 198427 test images: 53.7% +[rank0]:[W725 18:50:37.557177448 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0725 18:50:56.752000 2577455 torch/distributed/run.py:766] +W0725 18:50:56.752000 2577455 torch/distributed/run.py:766] ***************************************** +W0725 18:50:56.752000 2577455 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:50:56.752000 2577455 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:51:24.507684758 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:51:24.704661116 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:51:24.936102505 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:51:24.974720659 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:51:24.016438767 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:51:24.054696481 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:51:24.089339928 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank4]:[W725 18:51:24.091239877 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed4_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed4_resampled_n116-040-168_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:40:06 acc1_pos: 80.208 acc5_pos: 100.000 loss: 0.4535 (0.4535) acc1: 80.2083 (80.2083) acc5: 100.0000 (100.0000) time: 6.1864 data: 4.6128 max mem: 3531 +Test: [ 10/389] eta: 0:04:05 acc1_pos: 71.875 acc5_pos: 100.000 loss: 0.5380 (0.5557) acc1: 77.0833 (70.8807) acc5: 100.0000 (100.0000) time: 0.6482 data: 0.4197 max mem: 3531 +Test: [ 20/389] eta: 0:02:22 acc1_pos: 51.302 acc5_pos: 100.000 loss: 0.6740 (0.6487) acc1: 57.0312 (61.2227) acc5: 100.0000 (100.0000) time: 0.0971 data: 0.0010 max mem: 3531 +Test: [ 30/389] eta: 0:02:02 acc1_pos: 77.344 acc5_pos: 100.000 loss: 0.6429 (0.6170) acc1: 63.0208 (65.8098) acc5: 100.0000 (100.0000) time: 0.1706 data: 0.0745 max mem: 3531 +Test: [ 40/389] eta: 0:01:46 acc1_pos: 27.865 acc5_pos: 100.000 loss: 0.5538 (0.6499) acc1: 71.6146 (63.1034) acc5: 100.0000 (100.0000) time: 0.2206 data: 0.1280 max mem: 3531 +Test: [ 50/389] eta: 0:01:37 acc1_pos: 54.427 acc5_pos: 100.000 loss: 0.7731 (0.6709) acc1: 51.5625 (61.2694) acc5: 100.0000 (100.0000) time: 0.2047 data: 0.1128 max mem: 3531 +Test: [ 60/389] eta: 0:01:27 acc1_pos: 28.906 acc5_pos: 100.000 loss: 0.7710 (0.6599) acc1: 51.8229 (62.0816) acc5: 100.0000 (100.0000) time: 0.1809 data: 0.0869 max mem: 3531 +Test: [ 70/389] eta: 0:01:20 acc1_pos: 75.521 acc5_pos: 100.000 loss: 0.7398 (0.6616) acc1: 51.8229 (61.6894) acc5: 100.0000 (100.0000) time: 0.1650 data: 0.0718 max mem: 3531 +Test: [ 80/389] eta: 0:01:15 acc1_pos: 89.062 acc5_pos: 100.000 loss: 0.5795 (0.6495) acc1: 64.3229 (63.2555) acc5: 100.0000 (100.0000) time: 0.1833 data: 0.0916 max mem: 3531 +Test: [ 90/389] eta: 0:01:11 acc1_pos: 70.833 acc5_pos: 100.000 loss: 0.4814 (0.6265) acc1: 81.5104 (65.6164) acc5: 100.0000 (100.0000) time: 0.1858 data: 0.0947 max mem: 3531 +Test: [100/389] eta: 0:01:07 acc1_pos: 30.469 acc5_pos: 100.000 loss: 0.4899 (0.6346) acc1: 77.3438 (64.9159) acc5: 100.0000 (100.0000) time: 0.1860 data: 0.0960 max mem: 3531 +Test: [110/389] eta: 0:01:03 acc1_pos: 91.146 acc5_pos: 100.000 loss: 0.6919 (0.6368) acc1: 54.1667 (64.4825) acc5: 100.0000 (100.0000) time: 0.1905 data: 0.0991 max mem: 3531 +Test: [120/389] eta: 0:01:00 acc1_pos: 79.167 acc5_pos: 100.000 loss: 0.4836 (0.6297) acc1: 75.0000 (65.4033) acc5: 100.0000 (100.0000) time: 0.1861 data: 0.0944 max mem: 3531 +Test: [130/389] eta: 0:00:57 acc1_pos: 65.644 acc5_pos: 100.000 loss: 0.7356 (0.6483) acc1: 52.3438 (63.1620) acc5: 100.0000 (100.0000) acc1_neg: 43.750 acc5_neg: 100.000 time: 0.1892 data: 0.0981 max mem: 3531 +Test: [140/389] eta: 0:00:54 acc1_pos: 65.644 acc5_pos: 100.000 loss: 0.8353 (0.6704) acc1: 37.5000 (60.9985) acc5: 100.0000 (100.0000) acc1_neg: 2.865 acc5_neg: 100.000 time: 0.1768 data: 0.0853 max mem: 3531 +Test: [150/389] eta: 0:00:51 acc1_pos: 65.644 acc5_pos: 100.000 loss: 0.8147 (0.6809) acc1: 38.8021 (60.0769) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.1639 data: 0.0729 max mem: 3531 +Test: [160/389] eta: 0:00:48 acc1_pos: 65.644 acc5_pos: 100.000 loss: 0.7552 (0.6852) acc1: 45.5729 (59.8150) acc5: 100.0000 (100.0000) acc1_neg: 61.719 acc5_neg: 100.000 time: 0.1706 data: 0.0797 max mem: 3531 +Test: [170/389] eta: 0:00:46 acc1_pos: 65.644 acc5_pos: 100.000 loss: 0.7427 (0.6919) acc1: 45.8333 (58.7887) acc5: 100.0000 (100.0000) acc1_neg: 52.865 acc5_neg: 100.000 time: 0.1870 data: 0.0959 max mem: 3531 +Test: [180/389] eta: 0:00:43 acc1_pos: 65.644 acc5_pos: 100.000 loss: 0.7396 (0.6894) acc1: 47.3958 (58.9707) acc5: 100.0000 (100.0000) acc1_neg: 42.708 acc5_neg: 100.000 time: 0.1847 data: 0.0936 max mem: 3531 +Test: [190/389] eta: 0:00:41 acc1_pos: 65.644 acc5_pos: 100.000 loss: 0.6235 (0.6984) acc1: 61.7188 (58.2311) acc5: 100.0000 (100.0000) acc1_neg: 5.990 acc5_neg: 100.000 time: 0.1949 data: 0.1032 max mem: 3531 +Test: [200/389] eta: 0:00:39 acc1_pos: 65.644 acc5_pos: 100.000 loss: 0.7057 (0.7057) acc1: 39.5833 (57.4005) acc5: 100.0000 (100.0000) acc1_neg: 60.938 acc5_neg: 100.000 time: 0.1864 data: 0.0952 max mem: 3531 +Test: [210/389] eta: 0:00:37 acc1_pos: 65.644 acc5_pos: 100.000 loss: 0.6507 (0.6963) acc1: 60.1562 (58.2593) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1749 data: 0.0841 max mem: 3532 +Test: [220/389] eta: 0:00:34 acc1_pos: 65.644 acc5_pos: 100.000 loss: 0.4581 (0.6894) acc1: 78.6458 (58.8942) acc5: 100.0000 (100.0000) acc1_neg: 20.833 acc5_neg: 100.000 time: 0.1958 data: 0.1047 max mem: 3532 +Test: [230/389] eta: 0:00:32 acc1_pos: 65.644 acc5_pos: 100.000 loss: 0.6062 (0.6929) acc1: 62.7604 (58.5543) acc5: 100.0000 (100.0000) acc1_neg: 38.281 acc5_neg: 100.000 time: 0.1995 data: 0.1072 max mem: 3532 +Test: [240/389] eta: 0:00:30 acc1_pos: 65.644 acc5_pos: 100.000 loss: 0.6406 (0.6915) acc1: 61.7188 (58.6953) acc5: 100.0000 (100.0000) acc1_neg: 31.771 acc5_neg: 100.000 time: 0.2162 data: 0.1238 max mem: 3532 +Test: [250/389] eta: 0:00:28 acc1_pos: 65.644 acc5_pos: 100.000 loss: 0.5889 (0.6904) acc1: 66.4062 (58.7940) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.2020 data: 0.1100 max mem: 3532 +Test: [260/389] eta: 0:00:26 acc1_pos: 65.644 acc5_pos: 100.000 loss: 0.5686 (0.6942) acc1: 64.8438 (58.3164) acc5: 100.0000 (100.0000) acc1_neg: 26.562 acc5_neg: 100.000 time: 0.1857 data: 0.0938 max mem: 3532 +Test: [270/389] eta: 0:00:24 acc1_pos: 65.644 acc5_pos: 100.000 loss: 0.8080 (0.6996) acc1: 30.7292 (57.6722) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1655 data: 0.0744 max mem: 3532 +Test: [280/389] eta: 0:00:22 acc1_pos: 65.644 acc5_pos: 100.000 loss: 0.7674 (0.6986) acc1: 39.3229 (57.8449) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1604 data: 0.0688 max mem: 3532 +Test: [290/389] eta: 0:00:19 acc1_pos: 65.644 acc5_pos: 100.000 loss: 0.6786 (0.6995) acc1: 58.5938 (57.7248) acc5: 100.0000 (100.0000) acc1_neg: 58.594 acc5_neg: 100.000 time: 0.1645 data: 0.0738 max mem: 3532 +Test: [300/389] eta: 0:00:17 acc1_pos: 65.644 acc5_pos: 100.000 loss: 0.7280 (0.7034) acc1: 51.0417 (57.2424) acc5: 100.0000 (100.0000) acc1_neg: 27.344 acc5_neg: 100.000 time: 0.1525 data: 0.0619 max mem: 3532 +Test: [310/389] eta: 0:00:15 acc1_pos: 65.644 acc5_pos: 100.000 loss: 0.9335 (0.7148) acc1: 32.0312 (56.2885) acc5: 100.0000 (100.0000) acc1_neg: 7.031 acc5_neg: 100.000 time: 0.1608 data: 0.0693 max mem: 3532 +Test: [320/389] eta: 0:00:13 acc1_pos: 65.644 acc5_pos: 100.000 loss: 1.0016 (0.7223) acc1: 27.6042 (55.5012) acc5: 100.0000 (100.0000) acc1_neg: 70.833 acc5_neg: 100.000 time: 0.1643 data: 0.0723 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 65.644 acc5_pos: 100.000 loss: 1.2239 (0.7384) acc1: 13.0208 (54.2509) acc5: 100.0000 (100.0000) acc1_neg: 9.896 acc5_neg: 100.000 time: 0.1807 data: 0.0889 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 65.644 acc5_pos: 100.000 loss: 0.7708 (0.7323) acc1: 46.8750 (54.9349) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.1943 data: 0.1038 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 65.644 acc5_pos: 100.000 loss: 0.5001 (0.7306) acc1: 80.2083 (55.1987) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1768 data: 0.0866 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 65.644 acc5_pos: 100.000 loss: 0.4203 (0.7218) acc1: 84.6354 (56.0487) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1609 data: 0.0709 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 65.644 acc5_pos: 100.000 loss: 0.4267 (0.7195) acc1: 84.3750 (56.3062) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1574 data: 0.0670 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 65.644 acc5_pos: 100.000 loss: 0.6653 (0.7196) acc1: 61.1979 (56.3088) acc5: 100.0000 (100.0000) acc1_neg: 53.646 acc5_neg: 100.000 time: 0.1390 data: 0.0487 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 65.644 acc5_pos: 100.000 loss: 0.5971 (0.7188) acc1: 63.0208 (56.3739) acc5: 100.0000 (100.0000) acc1_neg: 70.219 acc5_neg: 100.000 time: 0.1368 data: 0.0438 max mem: 3532 +Test: Total time: 0:01:14 (0.1910 s / it) +* Overall Acc@1 56.374 Acc@5 100.000 + target == real Acc@1 64.965 Acc@5 100.000 + target == fake Acc@1 52.381 Acc@5 100.000 + loss 0.719 + PR-AUC 0.7715 ROC-AUC 0.6204 EER 41.47% (thr=0.4583) +Accuracy of the network on the 149358 test images: 56.4% +[rank0]:[W725 18:52:47.332093281 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:53:05.667000 2578023 torch/distributed/run.py:766] +W0725 18:53:05.667000 2578023 torch/distributed/run.py:766] ***************************************** +W0725 18:53:05.667000 2578023 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:53:05.667000 2578023 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:53:36.373348171 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:53:36.579694176 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:53:36.585514767 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +[rank0]:[W725 18:53:36.661111058 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:53:36.669021920 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:53:36.725596269 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:53:36.779495202 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:53:36.906519860 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed4_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed4_resampled_n116-040-168_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:35:18 acc1_pos: 81.510 acc5_pos: 100.000 loss: 0.4499 (0.4499) acc1: 81.5104 (81.5104) acc5: 100.0000 (100.0000) time: 6.3607 data: 4.6116 max mem: 3531 +Test: [ 10/333] eta: 0:03:33 acc1_pos: 86.198 acc5_pos: 100.000 loss: 0.3492 (0.3405) acc1: 83.5938 (85.7008) acc5: 100.0000 (100.0000) time: 0.6599 data: 0.4195 max mem: 3531 +Test: [ 20/333] eta: 0:02:02 acc1_pos: 90.365 acc5_pos: 100.000 loss: 0.3479 (0.3411) acc1: 86.1979 (86.0739) acc5: 100.0000 (100.0000) time: 0.0929 data: 0.0003 max mem: 3531 +Test: [ 30/333] eta: 0:01:53 acc1_pos: 66.927 acc5_pos: 100.000 loss: 0.4299 (0.4257) acc1: 78.6458 (80.3679) acc5: 100.0000 (100.0000) time: 0.2194 data: 0.1244 max mem: 3531 +Test: [ 40/333] eta: 0:01:42 acc1_pos: 74.219 acc5_pos: 100.000 loss: 0.5968 (0.4674) acc1: 66.9271 (77.4454) acc5: 100.0000 (100.0000) time: 0.3053 data: 0.2132 max mem: 3531 +Test: [ 50/333] eta: 0:01:30 acc1_pos: 55.208 acc5_pos: 100.000 loss: 0.5990 (0.4919) acc1: 65.6250 (75.6944) acc5: 100.0000 (100.0000) time: 0.2374 data: 0.1471 max mem: 3531 +Test: [ 60/333] eta: 0:01:20 acc1_pos: 60.417 acc5_pos: 100.000 loss: 0.5991 (0.5154) acc1: 68.2292 (74.0138) acc5: 100.0000 (100.0000) time: 0.1845 data: 0.0919 max mem: 3531 +Test: [ 70/333] eta: 0:01:12 acc1_pos: 74.479 acc5_pos: 100.000 loss: 0.5826 (0.5240) acc1: 70.0521 (73.5072) acc5: 100.0000 (100.0000) time: 0.1611 data: 0.0660 max mem: 3531 +Test: [ 80/333] eta: 0:01:06 acc1_pos: 75.000 acc5_pos: 100.000 loss: 0.5624 (0.5311) acc1: 70.5729 (73.0421) acc5: 100.0000 (100.0000) time: 0.1584 data: 0.0646 max mem: 3531 +Test: [ 90/333] eta: 0:01:00 acc1_pos: 76.042 acc5_pos: 100.000 loss: 0.5507 (0.5327) acc1: 71.6146 (73.0912) acc5: 100.0000 (100.0000) time: 0.1571 data: 0.0641 max mem: 3531 +Test: [100/333] eta: 0:00:56 acc1_pos: 79.427 acc5_pos: 100.000 loss: 0.5200 (0.5347) acc1: 73.6979 (72.8367) acc5: 100.0000 (100.0000) time: 0.1615 data: 0.0666 max mem: 3531 +Test: [110/333] eta: 0:00:51 acc1_pos: 72.917 acc5_pos: 100.000 loss: 0.5352 (0.5368) acc1: 72.9167 (72.6633) acc5: 100.0000 (100.0000) time: 0.1491 data: 0.0553 max mem: 3531 +Test: [120/333] eta: 0:00:48 acc1_pos: 72.396 acc5_pos: 100.000 loss: 0.5352 (0.5375) acc1: 72.3958 (72.5831) acc5: 100.0000 (100.0000) time: 0.1560 data: 0.0649 max mem: 3531 +Test: [130/333] eta: 0:00:45 acc1_pos: 76.823 acc5_pos: 100.000 loss: 0.5224 (0.5378) acc1: 72.6562 (72.5270) acc5: 100.0000 (100.0000) time: 0.1655 data: 0.0730 max mem: 3531 +Test: [140/333] eta: 0:00:41 acc1_pos: 81.771 acc5_pos: 100.000 loss: 0.5224 (0.5346) acc1: 74.2188 (72.8206) acc5: 100.0000 (100.0000) time: 0.1550 data: 0.0608 max mem: 3531 +Test: [150/333] eta: 0:00:39 acc1_pos: 80.990 acc5_pos: 100.000 loss: 0.5173 (0.5361) acc1: 74.2188 (72.6752) acc5: 100.0000 (100.0000) time: 0.1640 data: 0.0722 max mem: 3531 +Test: [160/333] eta: 0:00:36 acc1_pos: 70.066 acc5_pos: 100.000 loss: 0.5485 (0.5524) acc1: 71.3542 (71.5596) acc5: 100.0000 (100.0000) acc1_neg: 29.688 acc5_neg: 100.000 time: 0.1694 data: 0.0769 max mem: 3531 +Test: [170/333] eta: 0:00:34 acc1_pos: 70.066 acc5_pos: 100.000 loss: 1.0662 (0.5847) acc1: 31.5104 (69.1993) acc5: 100.0000 (100.0000) acc1_neg: 34.115 acc5_neg: 100.000 time: 0.1892 data: 0.0942 max mem: 3531 +Test: [180/333] eta: 0:00:32 acc1_pos: 70.066 acc5_pos: 100.000 loss: 1.0198 (0.6060) acc1: 34.1146 (67.4450) acc5: 100.0000 (100.0000) acc1_neg: 46.094 acc5_neg: 100.000 time: 0.1944 data: 0.0989 max mem: 3531 +Test: [190/333] eta: 0:00:29 acc1_pos: 70.066 acc5_pos: 100.000 loss: 0.9546 (0.6196) acc1: 40.1042 (66.2495) acc5: 100.0000 (100.0000) acc1_neg: 53.125 acc5_neg: 100.000 time: 0.1844 data: 0.0874 max mem: 3531 +Test: [200/333] eta: 0:00:27 acc1_pos: 70.066 acc5_pos: 100.000 loss: 0.8926 (0.6363) acc1: 40.1042 (64.9500) acc5: 100.0000 (100.0000) acc1_neg: 38.542 acc5_neg: 100.000 time: 0.1862 data: 0.0886 max mem: 3531 +Test: [210/333] eta: 0:00:25 acc1_pos: 70.066 acc5_pos: 100.000 loss: 1.0359 (0.6557) acc1: 35.1562 (63.5528) acc5: 100.0000 (100.0000) acc1_neg: 48.958 acc5_neg: 100.000 time: 0.1749 data: 0.0770 max mem: 3532 +Test: [220/333] eta: 0:00:23 acc1_pos: 70.066 acc5_pos: 100.000 loss: 1.0359 (0.6692) acc1: 34.8958 (62.3386) acc5: 100.0000 (100.0000) acc1_neg: 29.688 acc5_neg: 100.000 time: 0.1691 data: 0.0708 max mem: 3532 +Test: [230/333] eta: 0:00:20 acc1_pos: 70.066 acc5_pos: 100.000 loss: 0.9345 (0.6802) acc1: 37.5000 (61.4392) acc5: 100.0000 (100.0000) acc1_neg: 33.594 acc5_neg: 100.000 time: 0.1720 data: 0.0762 max mem: 3532 +Test: [240/333] eta: 0:00:18 acc1_pos: 70.066 acc5_pos: 100.000 loss: 0.9420 (0.6924) acc1: 37.5000 (60.3853) acc5: 100.0000 (100.0000) acc1_neg: 34.375 acc5_neg: 100.000 time: 0.1741 data: 0.0807 max mem: 3532 +Test: [250/333] eta: 0:00:16 acc1_pos: 70.066 acc5_pos: 100.000 loss: 0.9674 (0.7022) acc1: 38.0208 (59.5887) acc5: 100.0000 (100.0000) acc1_neg: 38.281 acc5_neg: 100.000 time: 0.1752 data: 0.0805 max mem: 3532 +Test: [260/333] eta: 0:00:14 acc1_pos: 70.066 acc5_pos: 100.000 loss: 0.9058 (0.7096) acc1: 42.1875 (59.0527) acc5: 100.0000 (100.0000) acc1_neg: 52.083 acc5_neg: 100.000 time: 0.1550 data: 0.0604 max mem: 3532 +Test: [270/333] eta: 0:00:12 acc1_pos: 70.066 acc5_pos: 100.000 loss: 0.9979 (0.7222) acc1: 35.6771 (58.0864) acc5: 100.0000 (100.0000) acc1_neg: 31.250 acc5_neg: 100.000 time: 0.1584 data: 0.0639 max mem: 3532 +Test: [280/333] eta: 0:00:10 acc1_pos: 70.066 acc5_pos: 100.000 loss: 1.0536 (0.7334) acc1: 32.0312 (57.2314) acc5: 100.0000 (100.0000) acc1_neg: 45.573 acc5_neg: 100.000 time: 0.1763 data: 0.0808 max mem: 3532 +Test: [290/333] eta: 0:00:08 acc1_pos: 70.066 acc5_pos: 100.000 loss: 1.0596 (0.7446) acc1: 33.0729 (56.3995) acc5: 100.0000 (100.0000) acc1_neg: 36.719 acc5_neg: 100.000 time: 0.1851 data: 0.0912 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 70.066 acc5_pos: 100.000 loss: 1.0784 (0.7552) acc1: 34.1146 (55.6608) acc5: 100.0000 (100.0000) acc1_neg: 34.896 acc5_neg: 100.000 time: 0.1884 data: 0.0947 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 70.066 acc5_pos: 100.000 loss: 0.9590 (0.7567) acc1: 39.5833 (55.4821) acc5: 100.0000 (100.0000) acc1_neg: 46.875 acc5_neg: 100.000 time: 0.1902 data: 0.0948 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 70.066 acc5_pos: 100.000 loss: 0.7898 (0.7595) acc1: 48.4375 (55.2108) acc5: 100.0000 (100.0000) acc1_neg: 35.938 acc5_neg: 100.000 time: 0.2205 data: 0.1252 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 70.066 acc5_pos: 100.000 loss: 1.0463 (0.7695) acc1: 36.9792 (54.5758) acc5: 100.0000 (100.0000) acc1_neg: 36.458 acc5_neg: 100.000 time: 0.1834 data: 0.0878 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 70.066 acc5_pos: 100.000 loss: 1.0655 (0.7714) acc1: 36.4583 (54.4660) acc5: 100.0000 (100.0000) acc1_neg: 34.648 acc5_neg: 100.000 time: 0.1955 data: 0.0878 max mem: 3532 +Test: Total time: 0:01:05 (0.1961 s / it) +* Overall Acc@1 54.466 Acc@5 100.000 + target == real Acc@1 72.726 Acc@5 100.000 + target == fake Acc@1 38.211 Acc@5 100.000 + loss 0.771 + PR-AUC 0.6048 ROC-AUC 0.5961 EER 43.37% (thr=0.3740) +Accuracy of the network on the 127843 test images: 54.5% +[rank0]:[W725 18:54:55.337389443 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:55:17.387000 2578598 torch/distributed/run.py:766] +W0725 18:55:17.387000 2578598 torch/distributed/run.py:766] ***************************************** +W0725 18:55:17.387000 2578598 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:55:17.387000 2578598 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:55:51.347574945 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:55:51.537505135 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:55:51.657695933 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:55:51.802922589 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:55:51.834461329 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:55:51.034743020 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 7): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:55:51.115260903 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W725 18:55:51.116152029 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed4_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed4_resampled_n116-040-168_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:26:35 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.1812 (0.1812) acc1: 96.8750 (96.8750) acc5: 100.0000 (100.0000) time: 6.7875 data: 5.2002 max mem: 3531 +Test: [ 10/235] eta: 0:02:37 acc1_pos: 95.573 acc5_pos: 100.000 loss: 0.1880 (0.1889) acc1: 95.8333 (95.5256) acc5: 100.0000 (100.0000) time: 0.7001 data: 0.4731 max mem: 3531 +Test: [ 20/235] eta: 0:01:29 acc1_pos: 95.573 acc5_pos: 100.000 loss: 0.1877 (0.1853) acc1: 95.5729 (95.6349) acc5: 100.0000 (100.0000) time: 0.0956 data: 0.0004 max mem: 3531 +Test: [ 30/235] eta: 0:01:16 acc1_pos: 96.615 acc5_pos: 100.000 loss: 0.1802 (0.1840) acc1: 95.5729 (95.6905) acc5: 100.0000 (100.0000) time: 0.1899 data: 0.0909 max mem: 3531 +Test: [ 40/235] eta: 0:01:06 acc1_pos: 94.531 acc5_pos: 100.000 loss: 0.1822 (0.1844) acc1: 95.5729 (95.6491) acc5: 100.0000 (100.0000) time: 0.2649 data: 0.1671 max mem: 3531 +Test: [ 50/235] eta: 0:00:58 acc1_pos: 95.573 acc5_pos: 100.000 loss: 0.1835 (0.1857) acc1: 95.5729 (95.5627) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1279 max mem: 3531 +Test: [ 60/235] eta: 0:00:51 acc1_pos: 95.052 acc5_pos: 100.000 loss: 0.1893 (0.1875) acc1: 95.3125 (95.4875) acc5: 100.0000 (100.0000) time: 0.2049 data: 0.1088 max mem: 3531 +Test: [ 70/235] eta: 0:00:46 acc1_pos: 96.923 acc5_pos: 100.000 loss: 0.2185 (0.3596) acc1: 93.4896 (86.1209) acc5: 100.0000 (100.0000) acc1_neg: 16.927 acc5_neg: 100.000 time: 0.1999 data: 0.1045 max mem: 3531 +Test: [ 80/235] eta: 0:00:42 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.5586 (0.5101) acc1: 21.0938 (78.0575) acc5: 100.0000 (100.0000) acc1_neg: 17.969 acc5_neg: 100.000 time: 0.2149 data: 0.1212 max mem: 3531 +Test: [ 90/235] eta: 0:00:38 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.5728 (0.6294) acc1: 19.7917 (71.6403) acc5: 100.0000 (100.0000) acc1_neg: 20.833 acc5_neg: 100.000 time: 0.2110 data: 0.1174 max mem: 3531 +Test: [100/235] eta: 0:00:33 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.5790 (0.7231) acc1: 20.3125 (66.6203) acc5: 100.0000 (100.0000) acc1_neg: 19.271 acc5_neg: 100.000 time: 0.1506 data: 0.0562 max mem: 3531 +Test: [110/235] eta: 0:00:30 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.5707 (0.7990) acc1: 20.8333 (62.5774) acc5: 100.0000 (100.0000) acc1_neg: 21.875 acc5_neg: 100.000 time: 0.1617 data: 0.0698 max mem: 3531 +Test: [120/235] eta: 0:00:28 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.6078 (0.8814) acc1: 19.2708 (58.3957) acc5: 100.0000 (100.0000) acc1_neg: 12.240 acc5_neg: 100.000 time: 0.2052 data: 0.1130 max mem: 3531 +Test: [130/235] eta: 0:00:25 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.8064 (0.9516) acc1: 10.9375 (54.8505) acc5: 100.0000 (100.0000) acc1_neg: 10.938 acc5_neg: 100.000 time: 0.2070 data: 0.1111 max mem: 3531 +Test: [140/235] eta: 0:00:22 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.8259 (1.0145) acc1: 10.9375 (51.7712) acc5: 100.0000 (100.0000) acc1_neg: 8.854 acc5_neg: 100.000 time: 0.2065 data: 0.1114 max mem: 3531 +Test: [150/235] eta: 0:00:20 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.8259 (1.0672) acc1: 10.9375 (49.0998) acc5: 100.0000 (100.0000) acc1_neg: 11.979 acc5_neg: 100.000 time: 0.2062 data: 0.1142 max mem: 3531 +Test: [160/235] eta: 0:00:17 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.8077 (1.1136) acc1: 11.1979 (46.7877) acc5: 100.0000 (100.0000) acc1_neg: 13.021 acc5_neg: 100.000 time: 0.2129 data: 0.1210 max mem: 3531 +Test: [170/235] eta: 0:00:15 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.7713 (1.1375) acc1: 13.8021 (45.3658) acc5: 100.0000 (100.0000) acc1_neg: 21.094 acc5_neg: 100.000 time: 0.1996 data: 0.1077 max mem: 3531 +Test: [180/235] eta: 0:00:12 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.4751 (1.1565) acc1: 23.6979 (44.2320) acc5: 100.0000 (100.0000) acc1_neg: 22.135 acc5_neg: 100.000 time: 0.1843 data: 0.0904 max mem: 3531 +Test: [190/235] eta: 0:00:10 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.4667 (1.1727) acc1: 25.0000 (43.2046) acc5: 100.0000 (100.0000) acc1_neg: 25.521 acc5_neg: 100.000 time: 0.1789 data: 0.0852 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.4689 (1.1883) acc1: 25.2604 (42.2639) acc5: 100.0000 (100.0000) acc1_neg: 25.781 acc5_neg: 100.000 time: 0.1758 data: 0.0836 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.4876 (1.2028) acc1: 23.4375 (41.3828) acc5: 100.0000 (100.0000) acc1_neg: 22.917 acc5_neg: 100.000 time: 0.1823 data: 0.0901 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.4758 (1.2152) acc1: 23.4375 (40.6168) acc5: 100.0000 (100.0000) acc1_neg: 24.740 acc5_neg: 100.000 time: 0.1627 data: 0.0721 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.4677 (1.2266) acc1: 24.7396 (39.9238) acc5: 100.0000 (100.0000) acc1_neg: 23.958 acc5_neg: 100.000 time: 0.1333 data: 0.0425 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.4737 (1.2313) acc1: 24.7396 (39.7212) acc5: 100.0000 (100.0000) acc1_neg: 30.435 acc5_neg: 100.000 time: 0.1145 data: 0.0272 max mem: 3532 +Test: Total time: 0:00:50 (0.2133 s / it) +* Overall Acc@1 39.721 Acc@5 100.000 + target == real Acc@1 95.493 Acc@5 100.000 + target == fake Acc@1 19.550 Acc@5 100.000 + loss 1.231 + PR-AUC 0.8608 ROC-AUC 0.7016 EER 36.03% (thr=0.1312) +Accuracy of the network on the 89879 test images: 39.7% +[rank0]:[W725 18:56:50.483701099 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:57:13.809000 2579150 torch/distributed/run.py:766] +W0725 18:57:13.809000 2579150 torch/distributed/run.py:766] ***************************************** +W0725 18:57:13.809000 2579150 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:57:13.809000 2579150 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W725 18:57:50.287882263 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:57:50.990674485 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:57:50.044768853 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:57:51.656527887 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:57:51.792026543 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:57:51.857756260 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:57:51.923355179 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:57:51.966267429 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed4_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed4_resampled_n116-040-168_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:26:46 acc1_pos: 80.469 acc5_pos: 100.000 loss: 0.4358 (0.4358) acc1: 80.4688 (80.4688) acc5: 100.0000 (100.0000) time: 5.9717 data: 4.4683 max mem: 3531 +Test: [ 10/269] eta: 0:02:43 acc1_pos: 90.104 acc5_pos: 100.000 loss: 0.3937 (0.3749) acc1: 88.8021 (87.0028) acc5: 100.0000 (100.0000) time: 0.6301 data: 0.4066 max mem: 3531 +Test: [ 20/269] eta: 0:01:35 acc1_pos: 87.760 acc5_pos: 100.000 loss: 0.3575 (0.3750) acc1: 90.1042 (87.5248) acc5: 100.0000 (100.0000) time: 0.1060 data: 0.0103 max mem: 3531 +Test: [ 30/269] eta: 0:01:17 acc1_pos: 99.740 acc5_pos: 100.000 loss: 0.3575 (0.3687) acc1: 90.1042 (88.1720) acc5: 100.0000 (100.0000) time: 0.1580 data: 0.0598 max mem: 3531 +Test: [ 40/269] eta: 0:01:08 acc1_pos: 98.276 acc5_pos: 100.000 loss: 0.4702 (0.5087) acc1: 78.9062 (76.3021) acc5: 100.0000 (100.0000) acc1_neg: 36.198 acc5_neg: 100.000 time: 0.2065 data: 0.1098 max mem: 3531 +Test: [ 50/269] eta: 0:01:00 acc1_pos: 98.276 acc5_pos: 100.000 loss: 1.2366 (0.6953) acc1: 16.4062 (63.1536) acc5: 100.0000 (100.0000) acc1_neg: 1.823 acc5_neg: 100.000 time: 0.2048 data: 0.1126 max mem: 3531 +Test: [ 60/269] eta: 0:00:55 acc1_pos: 98.276 acc5_pos: 100.000 loss: 1.4692 (0.8301) acc1: 4.9479 (53.7013) acc5: 100.0000 (100.0000) acc1_neg: 3.906 acc5_neg: 100.000 time: 0.1992 data: 0.1077 max mem: 3531 +Test: [ 70/269] eta: 0:00:51 acc1_pos: 98.276 acc5_pos: 100.000 loss: 1.4704 (0.9213) acc1: 6.7708 (47.8286) acc5: 100.0000 (100.0000) acc1_neg: 14.844 acc5_neg: 100.000 time: 0.2158 data: 0.1246 max mem: 3531 +Test: [ 80/269] eta: 0:00:48 acc1_pos: 98.276 acc5_pos: 100.000 loss: 1.4036 (0.9653) acc1: 13.8021 (44.4027) acc5: 100.0000 (100.0000) acc1_neg: 34.375 acc5_neg: 100.000 time: 0.2250 data: 0.1346 max mem: 3531 +Test: [ 90/269] eta: 0:00:45 acc1_pos: 98.276 acc5_pos: 100.000 loss: 1.2036 (0.9923) acc1: 29.1667 (42.4536) acc5: 100.0000 (100.0000) acc1_neg: 10.417 acc5_neg: 100.000 time: 0.2188 data: 0.1276 max mem: 3531 +Test: [100/269] eta: 0:00:41 acc1_pos: 98.276 acc5_pos: 100.000 loss: 1.3043 (1.0282) acc1: 15.6250 (39.8025) acc5: 100.0000 (100.0000) acc1_neg: 9.635 acc5_neg: 100.000 time: 0.1946 data: 0.1011 max mem: 3531 +Test: [110/269] eta: 0:00:37 acc1_pos: 98.276 acc5_pos: 100.000 loss: 1.3043 (1.0513) acc1: 9.6354 (36.8900) acc5: 100.0000 (100.0000) acc1_neg: 7.031 acc5_neg: 100.000 time: 0.1658 data: 0.0705 max mem: 3531 +Test: [120/269] eta: 0:00:34 acc1_pos: 98.276 acc5_pos: 100.000 loss: 1.2159 (1.0540) acc1: 10.1562 (36.1979) acc5: 100.0000 (100.0000) acc1_neg: 29.427 acc5_neg: 100.000 time: 0.1776 data: 0.0833 max mem: 3531 +Test: [130/269] eta: 0:00:31 acc1_pos: 98.276 acc5_pos: 100.000 loss: 1.1372 (1.0717) acc1: 21.3542 (34.3869) acc5: 100.0000 (100.0000) acc1_neg: 5.990 acc5_neg: 100.000 time: 0.1759 data: 0.0833 max mem: 3531 +Test: [140/269] eta: 0:00:29 acc1_pos: 98.276 acc5_pos: 100.000 loss: 1.2630 (1.0890) acc1: 9.6354 (33.0563) acc5: 100.0000 (100.0000) acc1_neg: 20.573 acc5_neg: 100.000 time: 0.1813 data: 0.0888 max mem: 3531 +Test: [150/269] eta: 0:00:26 acc1_pos: 98.276 acc5_pos: 100.000 loss: 1.1985 (1.0925) acc1: 20.5729 (32.5435) acc5: 100.0000 (100.0000) acc1_neg: 30.208 acc5_neg: 100.000 time: 0.1944 data: 0.1011 max mem: 3531 +Test: [160/269] eta: 0:00:23 acc1_pos: 98.276 acc5_pos: 100.000 loss: 1.0474 (1.0915) acc1: 28.6458 (32.2205) acc5: 100.0000 (100.0000) acc1_neg: 30.729 acc5_neg: 100.000 time: 0.1799 data: 0.0851 max mem: 3531 +Test: [170/269] eta: 0:00:21 acc1_pos: 98.276 acc5_pos: 100.000 loss: 1.0554 (1.0951) acc1: 25.5208 (31.5850) acc5: 100.0000 (100.0000) acc1_neg: 33.073 acc5_neg: 100.000 time: 0.1836 data: 0.0888 max mem: 3531 +Test: [180/269] eta: 0:00:19 acc1_pos: 98.276 acc5_pos: 100.000 loss: 1.0932 (1.0943) acc1: 20.8333 (31.0889) acc5: 100.0000 (100.0000) acc1_neg: 24.740 acc5_neg: 100.000 time: 0.1828 data: 0.0895 max mem: 3531 +Test: [190/269] eta: 0:00:16 acc1_pos: 98.276 acc5_pos: 100.000 loss: 1.0957 (1.0965) acc1: 20.8333 (30.5328) acc5: 100.0000 (100.0000) acc1_neg: 15.625 acc5_neg: 100.000 time: 0.1802 data: 0.0870 max mem: 3531 +Test: [200/269] eta: 0:00:14 acc1_pos: 98.276 acc5_pos: 100.000 loss: 1.0502 (1.0967) acc1: 22.9167 (30.2381) acc5: 100.0000 (100.0000) acc1_neg: 16.406 acc5_neg: 100.000 time: 0.1820 data: 0.0872 max mem: 3531 +Test: [210/269] eta: 0:00:12 acc1_pos: 98.276 acc5_pos: 100.000 loss: 1.0272 (1.0953) acc1: 25.5208 (30.2441) acc5: 100.0000 (100.0000) acc1_neg: 29.427 acc5_neg: 100.000 time: 0.1896 data: 0.0947 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 98.276 acc5_pos: 100.000 loss: 1.0191 (1.0910) acc1: 28.1250 (30.2590) acc5: 100.0000 (100.0000) acc1_neg: 42.448 acc5_neg: 100.000 time: 0.1965 data: 0.1037 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 98.276 acc5_pos: 100.000 loss: 0.9754 (1.0830) acc1: 32.8125 (30.5871) acc5: 100.0000 (100.0000) acc1_neg: 24.740 acc5_neg: 100.000 time: 0.1867 data: 0.0955 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 98.276 acc5_pos: 100.000 loss: 1.0704 (1.0967) acc1: 16.4062 (29.7005) acc5: 100.0000 (100.0000) acc1_neg: 15.365 acc5_neg: 100.000 time: 0.1814 data: 0.0897 max mem: 3532 +Test: [250/269] eta: 0:00:03 acc1_pos: 98.276 acc5_pos: 100.000 loss: 1.3019 (1.1032) acc1: 11.1979 (29.4375) acc5: 100.0000 (100.0000) acc1_neg: 11.719 acc5_neg: 100.000 time: 0.1693 data: 0.0781 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 98.276 acc5_pos: 100.000 loss: 1.4977 (1.1203) acc1: 10.1562 (28.5840) acc5: 100.0000 (100.0000) acc1_neg: 5.990 acc5_neg: 100.000 time: 0.1403 data: 0.0505 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 98.276 acc5_pos: 100.000 loss: 1.5218 (1.1296) acc1: 6.2500 (27.9709) acc5: 100.0000 (100.0000) acc1_neg: 15.748 acc5_neg: 100.000 time: 0.1267 data: 0.0336 max mem: 3532 +Test: Total time: 0:00:54 (0.2012 s / it) +* Overall Acc@1 27.971 Acc@5 100.000 + target == real Acc@1 88.278 Acc@5 100.000 + target == fake Acc@1 19.741 Acc@5 100.000 + loss 1.130 + PR-AUC 0.9099 ROC-AUC 0.5886 EER 44.32% (thr=0.2754) +Accuracy of the network on the 103293 test images: 28.0% +[rank0]:[W725 18:58:55.216524914 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-25T18:59:01Z] eval_done tag=fixed500k_fake_n11_seed4_resampled +eval_queue_done_utc=2026-07-25T18:59:01Z diff --git a/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_112559/eval_queue_n116-040-168.pid b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_112559/eval_queue_n116-040-168.pid new file mode 100644 index 0000000000000000000000000000000000000000..c0683c8813fca586c0cb6ab12456c9c1fa660711 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_112559/eval_queue_n116-040-168.pid @@ -0,0 +1 @@ +2556782 diff --git a/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_112559/eval_queue_n116-040-213.log b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_112559/eval_queue_n116-040-213.log new file mode 100644 index 0000000000000000000000000000000000000000..fb7b8dcf789d20d5f9a17a45996de1bd8526c2fe --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_112559/eval_queue_n116-040-213.log @@ -0,0 +1,3187 @@ +eval_queue_started_utc=2026-07-25T17:02:43Z +hostname=n116-040-213 +train_tag=20260725_112559 +eval_tag=20260725_112559 +real_level=32 +fake_level=64 +eval_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559 +batch_per_gpu=256 +[2026-07-25T17:02:43Z] eval_wait tag=fixed500k_real_n32_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed1_resampled +[2026-07-25T17:03:43Z] eval_wait tag=fixed500k_real_n32_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed1_resampled +[2026-07-25T17:04:44Z] eval_wait tag=fixed500k_real_n32_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed1_resampled +[2026-07-25T17:05:44Z] eval_wait tag=fixed500k_real_n32_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed1_resampled +[2026-07-25T17:06:44Z] eval_wait tag=fixed500k_real_n32_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed1_resampled +[2026-07-25T17:07:44Z] eval_wait tag=fixed500k_real_n32_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed1_resampled +[2026-07-25T17:08:44Z] eval_wait tag=fixed500k_real_n32_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed1_resampled +[2026-07-25T17:09:44Z] eval_wait tag=fixed500k_real_n32_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed1_resampled +[2026-07-25T17:10:44Z] eval_skip tag=fixed500k_real_n32_seed1_resampled reason=complete_after_lock output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed1_resampled +[2026-07-25T17:10:44Z] eval_wait tag=fixed500k_fake_n64_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed1_resampled +[2026-07-25T17:11:44Z] eval_wait tag=fixed500k_fake_n64_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed1_resampled +[2026-07-25T17:12:44Z] eval_wait tag=fixed500k_fake_n64_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed1_resampled +[2026-07-25T17:13:45Z] eval_wait tag=fixed500k_fake_n64_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed1_resampled +[2026-07-25T17:14:45Z] eval_wait tag=fixed500k_fake_n64_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed1_resampled +[2026-07-25T17:15:45Z] eval_wait tag=fixed500k_fake_n64_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed1_resampled +[2026-07-25T17:16:45Z] eval_wait tag=fixed500k_fake_n64_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed1_resampled +[2026-07-25T17:17:45Z] eval_wait tag=fixed500k_fake_n64_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed1_resampled +[2026-07-25T17:18:45Z] eval_wait tag=fixed500k_fake_n64_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed1_resampled +[2026-07-25T17:19:46Z] eval_wait tag=fixed500k_fake_n64_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed1_resampled +[2026-07-25T17:20:46Z] eval_wait tag=fixed500k_fake_n64_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed1_resampled +[2026-07-25T17:21:46Z] eval_wait tag=fixed500k_fake_n64_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed1_resampled +[2026-07-25T17:22:46Z] eval_wait tag=fixed500k_fake_n64_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed1_resampled +[2026-07-25T17:23:46Z] eval_wait tag=fixed500k_fake_n64_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed1_resampled +[2026-07-25T17:24:46Z] eval_wait tag=fixed500k_fake_n64_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed1_resampled +[2026-07-25T17:25:47Z] eval_wait tag=fixed500k_fake_n64_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed1_resampled +[2026-07-25T17:26:47Z] eval_wait tag=fixed500k_fake_n64_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed1_resampled +[2026-07-25T17:27:47Z] eval_wait tag=fixed500k_fake_n64_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed1_resampled +[2026-07-25T17:28:47Z] eval_wait tag=fixed500k_fake_n64_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed1_resampled +[2026-07-25T17:29:47Z] eval_wait tag=fixed500k_fake_n64_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed1_resampled +[2026-07-25T17:30:47Z] eval_wait tag=fixed500k_fake_n64_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed1_resampled +[2026-07-25T17:31:48Z] eval_wait tag=fixed500k_fake_n64_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed1_resampled +[2026-07-25T17:32:48Z] eval_skip tag=fixed500k_fake_n64_seed1_resampled reason=complete_after_lock output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed1_resampled +[2026-07-25T17:32:48Z] eval_wait tag=fixed500k_real_n32_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed2_resampled +[2026-07-25T17:33:48Z] eval_wait tag=fixed500k_real_n32_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed2_resampled +[2026-07-25T17:34:48Z] eval_wait tag=fixed500k_real_n32_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed2_resampled +[2026-07-25T17:35:49Z] eval_wait tag=fixed500k_real_n32_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed2_resampled +[2026-07-25T17:36:49Z] eval_wait tag=fixed500k_real_n32_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed2_resampled +[2026-07-25T17:37:49Z] eval_wait tag=fixed500k_real_n32_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed2_resampled +[2026-07-25T17:38:49Z] eval_wait tag=fixed500k_real_n32_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed2_resampled +[2026-07-25T17:39:49Z] eval_wait tag=fixed500k_real_n32_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed2_resampled +[2026-07-25T17:40:49Z] eval_wait tag=fixed500k_real_n32_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed2_resampled +[2026-07-25T17:41:50Z] eval_wait tag=fixed500k_real_n32_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed2_resampled +[2026-07-25T17:42:50Z] eval_wait tag=fixed500k_real_n32_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed2_resampled +[2026-07-25T17:43:50Z] eval_wait tag=fixed500k_real_n32_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed2_resampled +[2026-07-25T17:44:50Z] eval_wait tag=fixed500k_real_n32_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed2_resampled +[2026-07-25T17:45:50Z] eval_wait tag=fixed500k_real_n32_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed2_resampled +[2026-07-25T17:46:50Z] eval_wait tag=fixed500k_real_n32_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed2_resampled +[2026-07-25T17:47:51Z] eval_wait tag=fixed500k_real_n32_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed2_resampled +[2026-07-25T17:48:51Z] eval_wait tag=fixed500k_real_n32_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed2_resampled +[2026-07-25T17:49:51Z] eval_wait tag=fixed500k_real_n32_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed2_resampled +[2026-07-25T17:50:51Z] eval_wait tag=fixed500k_real_n32_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed2_resampled +[2026-07-25T17:51:51Z] eval_wait tag=fixed500k_real_n32_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed2_resampled +[2026-07-25T17:52:51Z] eval_wait tag=fixed500k_real_n32_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed2_resampled +[2026-07-25T17:53:51Z] eval_skip tag=fixed500k_real_n32_seed2_resampled reason=complete_after_wait output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed2_resampled +[2026-07-25T17:53:52Z] eval_start tag=fixed500k_fake_n64_seed2_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed2_resampled_n116-040-213_20260725_112559/best_checkpoint.pth +started_utc=2026-07-25T17:53:52Z +hostname=n116-040-213 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed2_resampled_n116-040-213_20260725_112559/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed2_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0725 17:54:09.150000 599489 torch/distributed/run.py:766] +W0725 17:54:09.150000 599489 torch/distributed/run.py:766] ***************************************** +W0725 17:54:09.150000 599489 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:54:09.150000 599489 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:54:45.457251831 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:54:45.925138619 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:54:46.525347851 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:54:46.547481063 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:54:46.579157801 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 17:54:46.597272143 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:54:46.610557446 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:54:46.632784970 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed2_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed2_resampled_n116-040-213_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:18:30 acc1_pos: 67.448 acc5_pos: 100.000 loss: 0.6068 (0.6068) acc1: 67.4479 (67.4479) acc5: 100.0000 (100.0000) time: 4.7466 data: 2.5678 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/234] eta: 0:02:09 acc1_pos: 55.990 acc5_pos: 100.000 loss: 0.6068 (0.6561) acc1: 67.4479 (62.7131) acc5: 100.0000 (100.0000) time: 0.5771 data: 0.2847 max mem: 3531 +Test: [ 20/234] eta: 0:01:17 acc1_pos: 97.135 acc5_pos: 100.000 loss: 0.6731 (0.6663) acc1: 55.9896 (59.5486) acc5: 100.0000 (100.0000) time: 0.1449 data: 0.0479 max mem: 3531 +Test: [ 30/234] eta: 0:01:04 acc1_pos: 77.344 acc5_pos: 100.000 loss: 0.1830 (0.5214) acc1: 96.8750 (70.7577) acc5: 100.0000 (100.0000) time: 0.1708 data: 0.0792 max mem: 3531 +Test: [ 40/234] eta: 0:00:58 acc1_pos: 98.958 acc5_pos: 100.000 loss: 0.1790 (0.4818) acc1: 97.1354 (73.8821) acc5: 100.0000 (100.0000) time: 0.2404 data: 0.1482 max mem: 3531 +Test: [ 50/234] eta: 0:00:52 acc1_pos: 98.698 acc5_pos: 100.000 loss: 0.1407 (0.4135) acc1: 97.3958 (78.6050) acc5: 100.0000 (100.0000) time: 0.2389 data: 0.1462 max mem: 3531 +Test: [ 60/234] eta: 0:00:45 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.1167 (0.3616) acc1: 98.6979 (82.0910) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1744 data: 0.0794 max mem: 3531 +Test: [ 70/234] eta: 0:00:39 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.0964 (0.3773) acc1: 100.0000 (81.3527) acc5: 100.0000 (100.0000) acc1_neg: 63.281 acc5_neg: 100.000 time: 0.1349 data: 0.0392 max mem: 3531 +Test: [ 80/234] eta: 0:00:36 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.6427 (0.3917) acc1: 66.1458 (80.8288) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1699 data: 0.0738 max mem: 3531 +Test: [ 90/234] eta: 0:00:34 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.2299 (0.3720) acc1: 90.6250 (82.2974) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.2149 data: 0.1167 max mem: 3531 +Test: [100/234] eta: 0:00:31 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.2706 (0.3824) acc1: 86.9792 (81.7579) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2049 data: 0.1054 max mem: 3531 +Test: [110/234] eta: 0:00:28 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.4864 (0.3878) acc1: 73.4375 (81.6441) acc5: 100.0000 (100.0000) acc1_neg: 65.625 acc5_neg: 100.000 time: 0.1949 data: 0.0961 max mem: 3531 +Test: [120/234] eta: 0:00:25 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.4472 (0.4010) acc1: 77.8646 (81.0628) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.1999 data: 0.1011 max mem: 3531 +Test: [130/234] eta: 0:00:23 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.4472 (0.4043) acc1: 80.7292 (81.1625) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.1799 data: 0.0819 max mem: 3531 +Test: [140/234] eta: 0:00:20 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.4509 (0.4262) acc1: 77.6042 (80.1659) acc5: 100.0000 (100.0000) acc1_neg: 41.406 acc5_neg: 100.000 time: 0.1499 data: 0.0509 max mem: 3531 +Test: [150/234] eta: 0:00:18 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.5086 (0.4364) acc1: 74.4792 (79.5995) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.1749 data: 0.0781 max mem: 3531 +Test: [160/234] eta: 0:00:15 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.4690 (0.4356) acc1: 77.3438 (79.7926) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.2168 data: 0.1249 max mem: 3531 +Test: [170/234] eta: 0:00:13 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.4987 (0.4400) acc1: 75.7812 (79.6266) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.2199 data: 0.1268 max mem: 3531 +Test: [180/234] eta: 0:00:11 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.5330 (0.4518) acc1: 75.0000 (78.9796) acc5: 100.0000 (100.0000) acc1_neg: 37.500 acc5_neg: 100.000 time: 0.1838 data: 0.0889 max mem: 3531 +Test: [190/234] eta: 0:00:09 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.7069 (0.4668) acc1: 62.2396 (78.0037) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.1748 data: 0.0813 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.7036 (0.4605) acc1: 62.2396 (78.4010) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2090 data: 0.1129 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.1208 (0.4456) acc1: 99.7396 (79.3518) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2149 data: 0.1172 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.1992 (0.4360) acc1: 95.0521 (79.9550) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.1723 data: 0.0788 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.4815 (0.4633) acc1: 38.5417 (77.8916) acc5: 100.0000 (100.0000) acc1_neg: 34.115 acc5_neg: 100.000 time: 0.1214 data: 0.0303 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 96.552 acc5_pos: 100.000 loss: 1.0403 (0.4708) acc1: 33.5938 (77.3719) acc5: 100.0000 (100.0000) acc1_neg: 33.074 acc5_neg: 100.000 time: 0.1004 data: 0.0091 max mem: 3532 +Test: Total time: 0:00:47 (0.2034 s / it) +* Overall Acc@1 77.372 Acc@5 100.000 + target == real Acc@1 78.737 Acc@5 100.000 + target == fake Acc@1 76.987 Acc@5 100.000 + loss 0.471 + PR-AUC 0.9621 ROC-AUC 0.8742 EER 22.36% (thr=0.4885) +Accuracy of the network on the 89729 test images: 77.4% +[rank0]:[W725 17:55:43.770759377 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 17:56:06.647000 600058 torch/distributed/run.py:766] +W0725 17:56:06.647000 600058 torch/distributed/run.py:766] ***************************************** +W0725 17:56:06.647000 600058 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:56:06.647000 600058 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:56:44.582175405 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 17:56:44.692558646 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:56:44.780811494 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:56:44.901491939 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:56:45.225865852 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W725 17:56:45.226509956 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:56:45.257627687 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank4]:[W725 17:56:45.257647007 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed2_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed2_resampled_n116-040-213_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 5:12:31 acc1_pos: 47.135 acc5_pos: 100.000 loss: 0.7997 (0.7997) acc1: 47.1354 (47.1354) acc5: 100.0000 (100.0000) time: 7.8556 data: 6.2265 max mem: 3531 +Test: [ 10/2387] eta: 0:31:39 acc1_pos: 25.521 acc5_pos: 100.000 loss: 0.9191 (0.9184) acc1: 42.7083 (39.7017) acc5: 100.0000 (100.0000) time: 0.7989 data: 0.5665 max mem: 3531 +Test: [ 20/2387] eta: 0:18:23 acc1_pos: 26.302 acc5_pos: 100.000 loss: 0.9191 (0.9289) acc1: 38.0208 (38.6161) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:17:53 acc1_pos: 21.615 acc5_pos: 100.000 loss: 0.9576 (0.9455) acc1: 29.4271 (37.6260) acc5: 100.0000 (100.0000) time: 0.2663 data: 0.1704 max mem: 3531 +Test: [ 40/2387] eta: 0:16:26 acc1_pos: 28.906 acc5_pos: 100.000 loss: 1.0188 (0.9611) acc1: 28.9062 (35.2325) acc5: 100.0000 (100.0000) time: 0.3727 data: 0.2809 max mem: 3531 +Test: [ 50/2387] eta: 0:15:08 acc1_pos: 40.365 acc5_pos: 100.000 loss: 0.9188 (0.9590) acc1: 29.4271 (35.2686) acc5: 100.0000 (100.0000) time: 0.2861 data: 0.1950 max mem: 3531 +Test: [ 60/2387] eta: 0:14:29 acc1_pos: 57.292 acc5_pos: 100.000 loss: 0.9466 (0.9617) acc1: 30.7292 (34.8531) acc5: 100.0000 (100.0000) time: 0.2782 data: 0.1873 max mem: 3531 +Test: [ 70/2387] eta: 0:13:51 acc1_pos: 32.292 acc5_pos: 100.000 loss: 1.0428 (0.9793) acc1: 28.9062 (33.5681) acc5: 100.0000 (100.0000) time: 0.2826 data: 0.1893 max mem: 3531 +Test: [ 80/2387] eta: 0:13:15 acc1_pos: 51.823 acc5_pos: 100.000 loss: 1.0128 (0.9687) acc1: 28.9062 (34.2271) acc5: 100.0000 (100.0000) time: 0.2555 data: 0.1609 max mem: 3531 +Test: [ 90/2387] eta: 0:12:49 acc1_pos: 50.000 acc5_pos: 100.000 loss: 0.8561 (0.9496) acc1: 40.8854 (35.5140) acc5: 100.0000 (100.0000) time: 0.2503 data: 0.1564 max mem: 3531 +Test: [ 100/2387] eta: 0:12:24 acc1_pos: 41.927 acc5_pos: 100.000 loss: 0.8186 (0.9461) acc1: 41.9271 (35.7209) acc5: 100.0000 (100.0000) time: 0.2486 data: 0.1541 max mem: 3531 +Test: [ 110/2387] eta: 0:12:08 acc1_pos: 37.500 acc5_pos: 100.000 loss: 0.8186 (0.9326) acc1: 41.6667 (36.8290) acc5: 100.0000 (100.0000) time: 0.2518 data: 0.1540 max mem: 3531 +Test: [ 120/2387] eta: 0:11:49 acc1_pos: 33.854 acc5_pos: 100.000 loss: 0.8349 (0.9272) acc1: 41.6667 (37.3085) acc5: 100.0000 (100.0000) time: 0.2497 data: 0.1531 max mem: 3531 +Test: [ 130/2387] eta: 0:11:34 acc1_pos: 40.104 acc5_pos: 100.000 loss: 0.8508 (0.9207) acc1: 40.1042 (37.5060) acc5: 100.0000 (100.0000) time: 0.2382 data: 0.1436 max mem: 3531 +Test: [ 140/2387] eta: 0:11:19 acc1_pos: 45.573 acc5_pos: 100.000 loss: 0.8409 (0.9153) acc1: 41.4062 (37.8657) acc5: 100.0000 (100.0000) time: 0.2373 data: 0.1434 max mem: 3531 +Test: [ 150/2387] eta: 0:11:08 acc1_pos: 26.302 acc5_pos: 100.000 loss: 0.7956 (0.9077) acc1: 43.7500 (38.2519) acc5: 100.0000 (100.0000) time: 0.2420 data: 0.1496 max mem: 3531 +Test: [ 160/2387] eta: 0:10:52 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.7480 (0.8901) acc1: 47.1354 (39.8001) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2291 data: 0.1316 max mem: 3531 +Test: [ 170/2387] eta: 0:10:39 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.2655 (0.8528) acc1: 94.2708 (43.0373) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2089 data: 0.1095 max mem: 3531 +Test: [ 180/2387] eta: 0:10:19 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.2488 (0.8194) acc1: 95.3125 (45.9499) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1801 data: 0.0836 max mem: 3531 +Test: [ 190/2387] eta: 0:10:07 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.2463 (0.7894) acc1: 95.5729 (48.5507) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1741 data: 0.0798 max mem: 3531 +Test: [ 200/2387] eta: 0:09:55 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.2444 (0.7623) acc1: 95.3125 (50.9108) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1983 data: 0.1014 max mem: 3531 +Test: [ 210/2387] eta: 0:09:45 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.2489 (0.7383) acc1: 95.3125 (52.9942) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1987 data: 0.1016 max mem: 3532 +Test: [ 220/2387] eta: 0:09:35 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.2508 (0.7162) acc1: 95.3125 (54.9220) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1961 data: 0.1036 max mem: 3532 +Test: [ 230/2387] eta: 0:09:26 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.2528 (0.6962) acc1: 95.5729 (56.6705) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1925 data: 0.0996 max mem: 3532 +Test: [ 240/2387] eta: 0:09:17 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.2597 (0.6784) acc1: 95.3125 (58.2750) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1933 data: 0.0981 max mem: 3532 +Test: [ 250/2387] eta: 0:09:09 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.2355 (0.6561) acc1: 96.3542 (59.9166) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1985 data: 0.1044 max mem: 3532 +Test: [ 260/2387] eta: 0:09:03 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.0956 (0.6344) acc1: 100.0000 (61.4523) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2049 data: 0.1111 max mem: 3532 +Test: [ 270/2387] eta: 0:08:57 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.0930 (0.6145) acc1: 100.0000 (62.8719) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2087 data: 0.1152 max mem: 3532 +Test: [ 280/2387] eta: 0:08:47 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.0942 (0.5961) acc1: 100.0000 (64.1932) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1866 data: 0.0955 max mem: 3532 +Test: [ 290/2387] eta: 0:08:42 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.0897 (0.5786) acc1: 100.0000 (65.4236) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1854 data: 0.0918 max mem: 3532 +Test: [ 300/2387] eta: 0:08:37 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.0887 (0.5624) acc1: 100.0000 (66.5724) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2106 data: 0.1152 max mem: 3532 +Test: [ 310/2387] eta: 0:08:32 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1001 (0.5515) acc1: 100.0000 (67.5601) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2157 data: 0.1210 max mem: 3532 +Test: [ 320/2387] eta: 0:08:28 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.3728 (0.5462) acc1: 94.7917 (68.3971) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2200 data: 0.1266 max mem: 3532 +Test: [ 330/2387] eta: 0:08:24 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.3794 (0.5411) acc1: 94.2708 (69.1906) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2186 data: 0.1247 max mem: 3532 +Test: [ 340/2387] eta: 0:08:19 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.3731 (0.5362) acc1: 94.5312 (69.9414) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2094 data: 0.1142 max mem: 3532 +Test: [ 350/2387] eta: 0:08:16 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.3654 (0.5264) acc1: 96.0938 (70.7799) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2173 data: 0.1247 max mem: 3532 +Test: [ 360/2387] eta: 0:08:11 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1934 (0.5174) acc1: 98.6979 (71.5309) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2178 data: 0.1271 max mem: 3532 +Test: [ 370/2387] eta: 0:08:06 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1997 (0.5088) acc1: 97.6562 (72.2414) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1974 data: 0.1048 max mem: 3532 +Test: [ 380/2387] eta: 0:08:01 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.2002 (0.5008) acc1: 97.6562 (72.9085) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1949 data: 0.1013 max mem: 3532 +Test: [ 390/2387] eta: 0:07:56 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1999 (0.4932) acc1: 97.6562 (73.5454) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1962 data: 0.1044 max mem: 3532 +Test: [ 400/2387] eta: 0:07:53 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.2000 (0.4859) acc1: 97.6562 (74.1460) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2027 data: 0.1119 max mem: 3532 +Test: [ 410/2387] eta: 0:07:49 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1985 (0.4790) acc1: 97.6562 (74.7206) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2152 data: 0.1245 max mem: 3532 +Test: [ 420/2387] eta: 0:07:46 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1985 (0.4723) acc1: 97.9167 (75.2753) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2187 data: 0.1278 max mem: 3532 +Test: [ 430/2387] eta: 0:07:42 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.2054 (0.4665) acc1: 97.9167 (75.7915) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2088 data: 0.1180 max mem: 3532 +Test: [ 440/2387] eta: 0:07:41 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.2088 (0.4592) acc1: 97.6562 (76.3180) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2312 data: 0.1401 max mem: 3533 +Test: [ 450/2387] eta: 0:07:36 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.0925 (0.4511) acc1: 100.0000 (76.8431) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2258 data: 0.1331 max mem: 3533 +Test: [ 460/2387] eta: 0:07:34 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.0924 (0.4433) acc1: 100.0000 (77.3454) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2064 data: 0.1120 max mem: 3533 +Test: [ 470/2387] eta: 0:07:29 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.0922 (0.4358) acc1: 100.0000 (77.8264) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2007 data: 0.1062 max mem: 3533 +Test: [ 480/2387] eta: 0:07:26 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.0928 (0.4287) acc1: 100.0000 (78.2874) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1950 data: 0.1006 max mem: 3533 +Test: [ 490/2387] eta: 0:07:23 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.0928 (0.4219) acc1: 100.0000 (78.7296) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2170 data: 0.1234 max mem: 3533 +Test: [ 500/2387] eta: 0:07:17 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.0926 (0.4153) acc1: 100.0000 (79.1542) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1823 data: 0.0898 max mem: 3533 +Test: [ 510/2387] eta: 0:07:14 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.0927 (0.4093) acc1: 100.0000 (79.5621) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1715 data: 0.0801 max mem: 3533 +Test: [ 520/2387] eta: 0:07:10 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1318 (0.4040) acc1: 100.0000 (79.9544) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1925 data: 0.1004 max mem: 3533 +Test: [ 530/2387] eta: 0:07:05 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1281 (0.3988) acc1: 100.0000 (80.3319) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1815 data: 0.0888 max mem: 3533 +Test: [ 540/2387] eta: 0:07:02 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1268 (0.3938) acc1: 100.0000 (80.6955) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1920 data: 0.1008 max mem: 3533 +Test: [ 550/2387] eta: 0:06:59 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1269 (0.3889) acc1: 100.0000 (81.0458) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1988 data: 0.1069 max mem: 3533 +Test: [ 560/2387] eta: 0:06:55 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1277 (0.3843) acc1: 100.0000 (81.3837) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1934 data: 0.0993 max mem: 3533 +Test: [ 570/2387] eta: 0:06:53 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1270 (0.3798) acc1: 100.0000 (81.7097) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2040 data: 0.1104 max mem: 3533 +Test: [ 580/2387] eta: 0:06:50 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1258 (0.3754) acc1: 100.0000 (82.0245) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2048 data: 0.1131 max mem: 3533 +Test: [ 590/2387] eta: 0:06:48 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1266 (0.3712) acc1: 100.0000 (82.3287) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2203 data: 0.1287 max mem: 3533 +Test: [ 600/2387] eta: 0:06:45 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1306 (0.3674) acc1: 100.0000 (82.6223) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2260 data: 0.1340 max mem: 3533 +Test: [ 610/2387] eta: 0:06:41 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1584 (0.3646) acc1: 99.7396 (82.8867) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1861 data: 0.0943 max mem: 3533 +Test: [ 620/2387] eta: 0:06:38 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1923 (0.3618) acc1: 98.6979 (83.1409) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1849 data: 0.0907 max mem: 3533 +Test: [ 630/2387] eta: 0:06:35 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1930 (0.3591) acc1: 98.6979 (83.3849) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1981 data: 0.1037 max mem: 3533 +Test: [ 640/2387] eta: 0:06:32 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1961 (0.3566) acc1: 98.4375 (83.6198) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2001 data: 0.1085 max mem: 3533 +Test: [ 650/2387] eta: 0:06:29 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1961 (0.3541) acc1: 98.4375 (83.8522) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1942 data: 0.1025 max mem: 3533 +Test: [ 660/2387] eta: 0:06:24 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1942 (0.3517) acc1: 98.9583 (84.0772) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1631 data: 0.0681 max mem: 3533 +Test: [ 670/2387] eta: 0:06:21 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1918 (0.3493) acc1: 98.6979 (84.2943) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1556 data: 0.0606 max mem: 3534 +Test: [ 680/2387] eta: 0:06:17 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1949 (0.3471) acc1: 99.2188 (84.5161) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1705 data: 0.0780 max mem: 3534 +Test: [ 690/2387] eta: 0:06:14 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.2027 (0.3486) acc1: 98.9583 (84.4281) acc5: 100.0000 (100.0000) acc1_neg: 67.708 acc5_neg: 100.000 time: 0.1808 data: 0.0869 max mem: 3534 +Test: [ 700/2387] eta: 0:06:10 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.5298 (0.3511) acc1: 71.0938 (84.2357) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.1700 data: 0.0748 max mem: 3534 +Test: [ 710/2387] eta: 0:06:07 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.5298 (0.3537) acc1: 71.0938 (84.0402) acc5: 100.0000 (100.0000) acc1_neg: 72.656 acc5_neg: 100.000 time: 0.1593 data: 0.0652 max mem: 3534 +Test: [ 720/2387] eta: 0:06:04 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.5286 (0.3561) acc1: 71.0938 (83.8589) acc5: 100.0000 (100.0000) acc1_neg: 71.094 acc5_neg: 100.000 time: 0.1713 data: 0.0771 max mem: 3534 +Test: [ 730/2387] eta: 0:06:01 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.5333 (0.3587) acc1: 70.8333 (83.6832) acc5: 100.0000 (100.0000) acc1_neg: 69.792 acc5_neg: 100.000 time: 0.1791 data: 0.0846 max mem: 3534 +Test: [ 740/2387] eta: 0:05:58 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.5330 (0.3610) acc1: 70.8333 (83.5154) acc5: 100.0000 (100.0000) acc1_neg: 68.490 acc5_neg: 100.000 time: 0.1864 data: 0.0940 max mem: 3534 +Test: [ 750/2387] eta: 0:05:54 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.5224 (0.3632) acc1: 70.8333 (83.3580) acc5: 100.0000 (100.0000) acc1_neg: 74.219 acc5_neg: 100.000 time: 0.1722 data: 0.0812 max mem: 3534 +Test: [ 760/2387] eta: 0:05:51 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.5165 (0.3646) acc1: 71.6146 (83.2598) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1477 data: 0.0559 max mem: 3534 +Test: [ 770/2387] eta: 0:05:48 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.3528 (0.3636) acc1: 84.6354 (83.3482) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1726 data: 0.0809 max mem: 3534 +Test: [ 780/2387] eta: 0:05:46 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.2145 (0.3614) acc1: 95.5729 (83.5294) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2088 data: 0.1173 max mem: 3534 +Test: [ 790/2387] eta: 0:05:43 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1869 (0.3594) acc1: 97.6562 (83.6774) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.1856 data: 0.0933 max mem: 3534 +Test: [ 800/2387] eta: 0:05:40 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1869 (0.3571) acc1: 98.4375 (83.8616) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1619 data: 0.0701 max mem: 3534 +Test: [ 810/2387] eta: 0:05:36 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1885 (0.3551) acc1: 98.4375 (84.0301) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1685 data: 0.0757 max mem: 3534 +Test: [ 820/2387] eta: 0:05:34 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1866 (0.3532) acc1: 97.6562 (84.1774) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1744 data: 0.0812 max mem: 3534 +Test: [ 830/2387] eta: 0:05:31 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.2526 (0.3527) acc1: 93.2292 (84.2462) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1860 data: 0.0944 max mem: 3534 +Test: [ 840/2387] eta: 0:05:29 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.3059 (0.3519) acc1: 91.4062 (84.3499) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1932 data: 0.1008 max mem: 3534 +Test: [ 850/2387] eta: 0:05:26 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.2989 (0.3514) acc1: 92.4479 (84.4365) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1800 data: 0.0848 max mem: 3534 +Test: [ 860/2387] eta: 0:05:22 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.2788 (0.3503) acc1: 94.0104 (84.5649) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1399 data: 0.0432 max mem: 3534 +Test: [ 870/2387] eta: 0:05:19 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.2554 (0.3493) acc1: 95.5729 (84.6776) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1390 data: 0.0436 max mem: 3534 +Test: [ 880/2387] eta: 0:05:16 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.2554 (0.3485) acc1: 95.0521 (84.7832) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1513 data: 0.0585 max mem: 3534 +Test: [ 890/2387] eta: 0:05:13 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.2245 (0.3471) acc1: 96.8750 (84.9268) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1655 data: 0.0743 max mem: 3534 +Test: [ 900/2387] eta: 0:05:10 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.2606 (0.3507) acc1: 94.5312 (84.6701) acc5: 100.0000 (100.0000) acc1_neg: 54.948 acc5_neg: 100.000 time: 0.1752 data: 0.0822 max mem: 3535 +Test: [ 910/2387] eta: 0:05:08 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.7753 (0.3554) acc1: 52.3438 (84.3213) acc5: 100.0000 (100.0000) acc1_neg: 50.000 acc5_neg: 100.000 time: 0.1655 data: 0.0720 max mem: 3535 +Test: [ 920/2387] eta: 0:05:05 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.7887 (0.3602) acc1: 52.6042 (83.9814) acc5: 100.0000 (100.0000) acc1_neg: 50.260 acc5_neg: 100.000 time: 0.1694 data: 0.0775 max mem: 3535 +Test: [ 930/2387] eta: 0:05:02 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.7900 (0.3647) acc1: 52.6042 (83.6489) acc5: 100.0000 (100.0000) acc1_neg: 47.135 acc5_neg: 100.000 time: 0.1784 data: 0.0853 max mem: 3535 +Test: [ 940/2387] eta: 0:05:00 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.7900 (0.3691) acc1: 52.6042 (83.3306) acc5: 100.0000 (100.0000) acc1_neg: 53.906 acc5_neg: 100.000 time: 0.1837 data: 0.0880 max mem: 3535 +Test: [ 950/2387] eta: 0:04:57 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.7821 (0.3735) acc1: 53.1250 (83.0047) acc5: 100.0000 (100.0000) acc1_neg: 54.688 acc5_neg: 100.000 time: 0.1799 data: 0.0834 max mem: 3535 +Test: [ 960/2387] eta: 0:04:55 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.7673 (0.3776) acc1: 53.3854 (82.7049) acc5: 100.0000 (100.0000) acc1_neg: 55.990 acc5_neg: 100.000 time: 0.1749 data: 0.0774 max mem: 3535 +Test: [ 970/2387] eta: 0:04:52 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.7758 (0.3819) acc1: 53.1250 (82.3912) acc5: 100.0000 (100.0000) acc1_neg: 54.688 acc5_neg: 100.000 time: 0.1649 data: 0.0672 max mem: 3535 +Test: [ 980/2387] eta: 0:04:49 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.7627 (0.3823) acc1: 54.4271 (82.3710) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.1651 data: 0.0694 max mem: 3535 +Test: [ 990/2387] eta: 0:04:47 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1929 (0.3806) acc1: 98.1771 (82.4919) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1799 data: 0.0842 max mem: 3535 +Test: [1000/2387] eta: 0:04:45 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1579 (0.3785) acc1: 99.2188 (82.6447) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1755 data: 0.0798 max mem: 3535 +Test: [1010/2387] eta: 0:04:42 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1739 (0.3768) acc1: 98.4375 (82.7731) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1749 data: 0.0792 max mem: 3535 +Test: [1020/2387] eta: 0:04:40 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.2062 (0.3754) acc1: 96.3542 (82.8842) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1842 data: 0.0866 max mem: 3535 +Test: [1030/2387] eta: 0:04:37 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1874 (0.3734) acc1: 98.4375 (83.0320) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1649 data: 0.0670 max mem: 3535 +Test: [1040/2387] eta: 0:04:34 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1556 (0.3713) acc1: 99.4792 (83.1890) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1549 data: 0.0584 max mem: 3535 +Test: [1050/2387] eta: 0:04:33 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1459 (0.3692) acc1: 99.4792 (83.3423) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1949 data: 0.0980 max mem: 3535 +Test: [1060/2387] eta: 0:04:30 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1501 (0.3672) acc1: 99.7396 (83.4912) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2049 data: 0.1046 max mem: 3535 +Test: [1070/2387] eta: 0:04:28 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1557 (0.3656) acc1: 98.4375 (83.6176) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1799 data: 0.0791 max mem: 3535 +Test: [1080/2387] eta: 0:04:25 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.2205 (0.3645) acc1: 95.8333 (83.7128) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1699 data: 0.0697 max mem: 3535 +Test: [1090/2387] eta: 0:04:23 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.2070 (0.3629) acc1: 95.8333 (83.8391) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1525 data: 0.0526 max mem: 3535 +Test: [1100/2387] eta: 0:04:20 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1711 (0.3612) acc1: 98.4375 (83.9672) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1567 data: 0.0612 max mem: 3535 +Test: [1110/2387] eta: 0:04:18 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1744 (0.3599) acc1: 97.6562 (84.0644) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1873 data: 0.0947 max mem: 3535 +Test: [1120/2387] eta: 0:04:16 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1779 (0.3584) acc1: 97.3958 (84.1864) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1961 data: 0.1041 max mem: 3536 +Test: [1130/2387] eta: 0:04:14 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1779 (0.3568) acc1: 97.6562 (84.3101) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1802 data: 0.0851 max mem: 3536 +Test: [1140/2387] eta: 0:04:12 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1906 (0.3556) acc1: 97.6562 (84.4163) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2037 data: 0.1075 max mem: 3536 +Test: [1150/2387] eta: 0:04:10 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.2350 (0.3547) acc1: 95.3125 (84.5114) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2168 data: 0.1215 max mem: 3536 +Test: [1160/2387] eta: 0:04:08 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.2099 (0.3534) acc1: 94.5312 (84.6069) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1844 data: 0.0907 max mem: 3536 +Test: [1170/2387] eta: 0:04:06 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1974 (0.3521) acc1: 95.3125 (84.7086) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1808 data: 0.0899 max mem: 3536 +Test: [1180/2387] eta: 0:04:03 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.2061 (0.3509) acc1: 96.0938 (84.8043) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1796 data: 0.0877 max mem: 3536 +Test: [1190/2387] eta: 0:04:01 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1668 (0.3496) acc1: 97.6562 (84.9096) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1819 data: 0.0900 max mem: 3536 +Test: [1200/2387] eta: 0:03:59 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1631 (0.3480) acc1: 98.6979 (85.0288) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1819 data: 0.0877 max mem: 3536 +Test: [1210/2387] eta: 0:03:56 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1479 (0.3464) acc1: 99.7396 (85.1502) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1716 data: 0.0772 max mem: 3536 +Test: [1220/2387] eta: 0:03:54 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1200 (0.3445) acc1: 99.7396 (85.2691) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1744 data: 0.0829 max mem: 3536 +Test: [1230/2387] eta: 0:03:52 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1096 (0.3425) acc1: 99.7396 (85.3868) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1821 data: 0.0904 max mem: 3536 +Test: [1240/2387] eta: 0:03:50 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1084 (0.3406) acc1: 100.0000 (85.5033) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1911 data: 0.0987 max mem: 3536 +Test: [1250/2387] eta: 0:03:48 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1012 (0.3387) acc1: 100.0000 (85.6186) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1864 data: 0.0948 max mem: 3536 +Test: [1260/2387] eta: 0:03:45 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.0998 (0.3368) acc1: 100.0000 (85.7312) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1727 data: 0.0816 max mem: 3536 +Test: [1270/2387] eta: 0:03:43 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.0975 (0.3350) acc1: 100.0000 (85.8433) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1712 data: 0.0795 max mem: 3536 +Test: [1280/2387] eta: 0:03:41 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1021 (0.3332) acc1: 100.0000 (85.9491) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1849 data: 0.0934 max mem: 3536 +Test: [1290/2387] eta: 0:03:39 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1219 (0.3317) acc1: 99.2188 (86.0420) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1819 data: 0.0899 max mem: 3536 +Test: [1300/2387] eta: 0:03:37 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1211 (0.3301) acc1: 99.2188 (86.1413) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1798 data: 0.0872 max mem: 3536 +Test: [1310/2387] eta: 0:03:35 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1211 (0.3286) acc1: 99.4792 (86.2359) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1877 data: 0.0956 max mem: 3536 +Test: [1320/2387] eta: 0:03:33 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1267 (0.3271) acc1: 99.4792 (86.3282) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1959 data: 0.1040 max mem: 3536 +Test: [1330/2387] eta: 0:03:31 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1374 (0.3259) acc1: 99.4792 (86.4176) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1954 data: 0.1010 max mem: 3536 +Test: [1340/2387] eta: 0:03:28 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1822 (0.3258) acc1: 96.8750 (86.4284) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1836 data: 0.0896 max mem: 3536 +Test: [1350/2387] eta: 0:03:26 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.0809 (0.3239) acc1: 100.0000 (86.5289) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1844 data: 0.0927 max mem: 3537 +Test: [1360/2387] eta: 0:03:24 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.0738 (0.3221) acc1: 100.0000 (86.6279) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1889 data: 0.0963 max mem: 3537 +Test: [1370/2387] eta: 0:03:22 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.0731 (0.3203) acc1: 100.0000 (86.7254) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1901 data: 0.0974 max mem: 3537 +Test: [1380/2387] eta: 0:03:20 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.0737 (0.3185) acc1: 100.0000 (86.8215) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.1018 max mem: 3537 +Test: [1390/2387] eta: 0:03:18 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.0737 (0.3168) acc1: 100.0000 (86.9163) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1742 data: 0.0807 max mem: 3537 +Test: [1400/2387] eta: 0:03:16 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.0740 (0.3150) acc1: 100.0000 (87.0097) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1954 data: 0.1029 max mem: 3537 +Test: [1410/2387] eta: 0:03:14 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.0760 (0.3142) acc1: 100.0000 (87.0709) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1997 data: 0.1078 max mem: 3537 +Test: [1420/2387] eta: 0:03:12 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.2016 (0.3137) acc1: 96.8750 (87.1150) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.2020 data: 0.1056 max mem: 3537 +Test: [1430/2387] eta: 0:03:10 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.2283 (0.3132) acc1: 94.2708 (87.1619) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2050 data: 0.1093 max mem: 3537 +Test: [1440/2387] eta: 0:03:08 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.2327 (0.3125) acc1: 95.3125 (87.2284) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1987 data: 0.1076 max mem: 3537 +Test: [1450/2387] eta: 0:03:06 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.2071 (0.3120) acc1: 96.0938 (87.2778) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2163 data: 0.1249 max mem: 3537 +Test: [1460/2387] eta: 0:03:05 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1811 (0.3112) acc1: 96.3542 (87.3414) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2249 data: 0.1335 max mem: 3537 +Test: [1470/2387] eta: 0:03:03 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1576 (0.3101) acc1: 98.4375 (87.4184) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2196 data: 0.1264 max mem: 3537 +Test: [1480/2387] eta: 0:03:01 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1461 (0.3090) acc1: 98.9583 (87.4889) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1970 data: 0.1024 max mem: 3537 +Test: [1490/2387] eta: 0:02:58 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1461 (0.3080) acc1: 98.6979 (87.5627) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1928 data: 0.0980 max mem: 3537 +Test: [1500/2387] eta: 0:02:57 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1428 (0.3070) acc1: 98.9583 (87.6339) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1969 data: 0.1012 max mem: 3537 +Test: [1510/2387] eta: 0:02:54 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1540 (0.3060) acc1: 99.4792 (87.7125) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1960 data: 0.1008 max mem: 3537 +Test: [1520/2387] eta: 0:02:52 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1728 (0.3055) acc1: 98.4375 (87.7573) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1857 data: 0.0920 max mem: 3537 +Test: [1530/2387] eta: 0:02:50 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.2413 (0.3053) acc1: 94.2708 (87.7900) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1938 data: 0.1019 max mem: 3537 +Test: [1540/2387] eta: 0:02:49 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.2336 (0.3049) acc1: 96.0938 (87.8481) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2095 data: 0.1183 max mem: 3537 +Test: [1550/2387] eta: 0:02:46 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.2258 (0.3044) acc1: 97.3958 (87.8993) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1950 data: 0.1036 max mem: 3537 +Test: [1560/2387] eta: 0:02:44 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.2198 (0.3040) acc1: 97.3958 (87.9483) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.1739 data: 0.0818 max mem: 3537 +Test: [1570/2387] eta: 0:02:42 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1993 (0.3033) acc1: 98.4375 (88.0125) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1809 data: 0.0877 max mem: 3537 +Test: [1580/2387] eta: 0:02:40 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1823 (0.3026) acc1: 98.4375 (88.0726) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1879 data: 0.0956 max mem: 3538 +Test: [1590/2387] eta: 0:02:38 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1821 (0.3017) acc1: 98.1771 (88.1398) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2048 data: 0.1134 max mem: 3538 +Test: [1600/2387] eta: 0:02:36 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1641 (0.3010) acc1: 98.6979 (88.1880) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1941 data: 0.1019 max mem: 3538 +Test: [1610/2387] eta: 0:02:34 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1598 (0.3002) acc1: 99.2188 (88.2538) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1756 data: 0.0818 max mem: 3538 +Test: [1620/2387] eta: 0:02:32 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1720 (0.2995) acc1: 97.6562 (88.3036) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1726 data: 0.0794 max mem: 3538 +Test: [1630/2387] eta: 0:02:30 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1935 (0.2990) acc1: 96.6146 (88.3515) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1666 data: 0.0732 max mem: 3538 +Test: [1640/2387] eta: 0:02:28 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.2269 (0.2987) acc1: 96.3542 (88.3863) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1656 data: 0.0717 max mem: 3538 +Test: [1650/2387] eta: 0:02:26 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.2376 (0.2982) acc1: 95.8333 (88.4300) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1813 data: 0.0884 max mem: 3538 +Test: [1660/2387] eta: 0:02:24 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.2342 (0.2979) acc1: 96.0938 (88.4666) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1792 data: 0.0850 max mem: 3538 +Test: [1670/2387] eta: 0:02:21 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.2342 (0.2975) acc1: 95.8333 (88.5089) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1549 data: 0.0574 max mem: 3538 +Test: [1680/2387] eta: 0:02:19 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.2329 (0.2972) acc1: 95.8333 (88.5440) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1768 data: 0.0807 max mem: 3538 +Test: [1690/2387] eta: 0:02:17 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.2329 (0.2969) acc1: 96.0938 (88.5816) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2149 data: 0.1200 max mem: 3538 +Test: [1700/2387] eta: 0:02:16 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.2287 (0.2964) acc1: 96.3542 (88.6300) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2230 data: 0.1274 max mem: 3538 +Test: [1710/2387] eta: 0:02:13 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1384 (0.2954) acc1: 98.9583 (88.6927) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1861 data: 0.0910 max mem: 3538 +Test: [1720/2387] eta: 0:02:11 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1246 (0.2944) acc1: 99.7396 (88.7544) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1626 data: 0.0670 max mem: 3538 +Test: [1730/2387] eta: 0:02:09 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1246 (0.2934) acc1: 99.7396 (88.8131) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1743 data: 0.0814 max mem: 3538 +Test: [1740/2387] eta: 0:02:07 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1162 (0.2924) acc1: 99.7396 (88.8755) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1856 data: 0.0929 max mem: 3538 +Test: [1750/2387] eta: 0:02:05 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1172 (0.2915) acc1: 99.7396 (88.9356) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1899 data: 0.0970 max mem: 3538 +Test: [1760/2387] eta: 0:02:03 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1192 (0.2905) acc1: 100.0000 (88.9974) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1883 data: 0.0973 max mem: 3538 +Test: [1770/2387] eta: 0:02:01 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1320 (0.2898) acc1: 99.2188 (89.0471) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.2043 data: 0.1129 max mem: 3538 +Test: [1780/2387] eta: 0:01:59 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1631 (0.2891) acc1: 98.1771 (89.0974) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2081 data: 0.1160 max mem: 3538 +Test: [1790/2387] eta: 0:01:57 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1631 (0.2883) acc1: 98.4375 (89.1515) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1920 data: 0.1004 max mem: 3538 +Test: [1800/2387] eta: 0:01:55 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1530 (0.2876) acc1: 98.6979 (89.2048) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1852 data: 0.0932 max mem: 3538 +Test: [1810/2387] eta: 0:01:53 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1407 (0.2867) acc1: 99.2188 (89.2608) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1928 data: 0.0979 max mem: 3539 +Test: [1820/2387] eta: 0:01:51 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1244 (0.2858) acc1: 99.4792 (89.3182) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1969 data: 0.1023 max mem: 3539 +Test: [1830/2387] eta: 0:01:49 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1244 (0.2850) acc1: 100.0000 (89.3753) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1914 data: 0.0988 max mem: 3539 +Test: [1840/2387] eta: 0:01:47 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1168 (0.2841) acc1: 100.0000 (89.4330) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1940 data: 0.1024 max mem: 3539 +Test: [1850/2387] eta: 0:01:46 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1163 (0.2832) acc1: 100.0000 (89.4899) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2000 data: 0.1090 max mem: 3539 +Test: [1860/2387] eta: 0:01:43 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1158 (0.2823) acc1: 100.0000 (89.5464) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1910 data: 0.0998 max mem: 3539 +Test: [1870/2387] eta: 0:01:41 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1158 (0.2814) acc1: 100.0000 (89.6021) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1859 data: 0.0945 max mem: 3539 +Test: [1880/2387] eta: 0:01:40 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1153 (0.2805) acc1: 100.0000 (89.6574) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1988 data: 0.1077 max mem: 3539 +Test: [1890/2387] eta: 0:01:38 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1151 (0.2796) acc1: 100.0000 (89.7121) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1988 data: 0.1075 max mem: 3539 +Test: [1900/2387] eta: 0:01:36 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1169 (0.2788) acc1: 100.0000 (89.7662) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1987 data: 0.1064 max mem: 3539 +Test: [1910/2387] eta: 0:01:34 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1421 (0.2781) acc1: 100.0000 (89.8162) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2003 data: 0.1057 max mem: 3539 +Test: [1920/2387] eta: 0:01:32 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1482 (0.2780) acc1: 99.4792 (89.8280) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1819 data: 0.0877 max mem: 3539 +Test: [1930/2387] eta: 0:01:30 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1683 (0.2774) acc1: 99.2188 (89.8783) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1680 data: 0.0762 max mem: 3539 +Test: [1940/2387] eta: 0:01:28 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1699 (0.2772) acc1: 99.2188 (89.8840) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1825 data: 0.0886 max mem: 3539 +Test: [1950/2387] eta: 0:01:25 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1879 (0.2768) acc1: 98.1771 (89.9162) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1688 data: 0.0729 max mem: 3539 +Test: [1960/2387] eta: 0:01:24 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1879 (0.2764) acc1: 98.4375 (89.9491) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1667 data: 0.0728 max mem: 3539 +Test: [1970/2387] eta: 0:01:21 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.2235 (0.2765) acc1: 92.9688 (89.9362) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.1810 data: 0.0886 max mem: 3539 +Test: [1980/2387] eta: 0:01:19 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.2524 (0.2763) acc1: 92.1875 (89.9601) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1692 data: 0.0746 max mem: 3539 +Test: [1990/2387] eta: 0:01:17 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1465 (0.2756) acc1: 98.1771 (90.0011) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1749 data: 0.0779 max mem: 3539 +Test: [2000/2387] eta: 0:01:15 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1465 (0.2750) acc1: 98.1771 (90.0420) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1710 data: 0.0732 max mem: 3539 +Test: [2010/2387] eta: 0:01:13 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1472 (0.2744) acc1: 98.1771 (90.0825) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1765 data: 0.0821 max mem: 3539 +Test: [2020/2387] eta: 0:01:11 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1463 (0.2737) acc1: 98.1771 (90.1240) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1864 data: 0.0951 max mem: 3539 +Test: [2030/2387] eta: 0:01:09 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1455 (0.2731) acc1: 98.4375 (90.1631) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1817 data: 0.0861 max mem: 3540 +Test: [2040/2387] eta: 0:01:08 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1468 (0.2725) acc1: 98.4375 (90.2039) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1962 data: 0.0990 max mem: 3540 +Test: [2050/2387] eta: 0:01:06 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1444 (0.2719) acc1: 98.4375 (90.2447) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1937 data: 0.0994 max mem: 3540 +Test: [2060/2387] eta: 0:01:04 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1449 (0.2713) acc1: 98.4375 (90.2836) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1881 data: 0.0961 max mem: 3540 +Test: [2070/2387] eta: 0:01:02 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1445 (0.2706) acc1: 98.4375 (90.3243) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1941 data: 0.1033 max mem: 3540 +Test: [2080/2387] eta: 0:01:00 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1405 (0.2700) acc1: 98.6979 (90.3653) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1981 data: 0.1057 max mem: 3540 +Test: [2090/2387] eta: 0:00:58 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1395 (0.2694) acc1: 98.6979 (90.4049) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1785 data: 0.0829 max mem: 3540 +Test: [2100/2387] eta: 0:00:56 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1422 (0.2688) acc1: 98.6979 (90.4453) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1759 data: 0.0798 max mem: 3540 +Test: [2110/2387] eta: 0:00:54 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1423 (0.2682) acc1: 98.6979 (90.4849) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1907 data: 0.0976 max mem: 3540 +Test: [2120/2387] eta: 0:00:52 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1500 (0.2680) acc1: 98.4375 (90.5054) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.1699 data: 0.0780 max mem: 3540 +Test: [2130/2387] eta: 0:00:50 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1659 (0.2674) acc1: 98.4375 (90.5450) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1693 data: 0.0768 max mem: 3540 +Test: [2140/2387] eta: 0:00:48 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1642 (0.2672) acc1: 98.9583 (90.5649) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1775 data: 0.0820 max mem: 3540 +Test: [2150/2387] eta: 0:00:46 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1866 (0.2668) acc1: 98.1771 (90.5964) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1699 data: 0.0747 max mem: 3540 +Test: [2160/2387] eta: 0:00:44 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1806 (0.2664) acc1: 98.1771 (90.6311) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1802 data: 0.0885 max mem: 3540 +Test: [2170/2387] eta: 0:00:42 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1918 (0.2663) acc1: 97.6562 (90.6503) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1851 data: 0.0931 max mem: 3540 +Test: [2180/2387] eta: 0:00:40 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.2814 (0.2673) acc1: 89.8438 (90.5899) acc5: 100.0000 (100.0000) acc1_neg: 54.948 acc5_neg: 100.000 time: 0.1749 data: 0.0823 max mem: 3540 +Test: [2190/2387] eta: 0:00:38 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.2651 (0.2674) acc1: 89.3229 (90.5819) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1795 data: 0.0848 max mem: 3540 +Test: [2200/2387] eta: 0:00:36 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1660 (0.2669) acc1: 96.8750 (90.6160) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1759 data: 0.0797 max mem: 3540 +Test: [2210/2387] eta: 0:00:34 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1588 (0.2665) acc1: 98.9583 (90.6508) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1799 data: 0.0862 max mem: 3540 +Test: [2220/2387] eta: 0:00:32 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1588 (0.2660) acc1: 98.9583 (90.6835) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1706 data: 0.0781 max mem: 3540 +Test: [2230/2387] eta: 0:00:30 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1562 (0.2655) acc1: 98.9583 (90.7190) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1552 data: 0.0633 max mem: 3540 +Test: [2240/2387] eta: 0:00:28 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1439 (0.2650) acc1: 98.1771 (90.7498) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1640 data: 0.0717 max mem: 3540 +Test: [2250/2387] eta: 0:00:26 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1505 (0.2646) acc1: 98.1771 (90.7837) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1704 data: 0.0772 max mem: 3540 +Test: [2260/2387] eta: 0:00:24 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1689 (0.2642) acc1: 98.9583 (90.8193) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1569 data: 0.0638 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1531 (0.2636) acc1: 99.7396 (90.8592) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1440 data: 0.0513 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1382 (0.2633) acc1: 100.0000 (90.8812) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1668 data: 0.0737 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1289 (0.2627) acc1: 100.0000 (90.9195) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1917 data: 0.0983 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1492 (0.2622) acc1: 99.7396 (90.9531) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1685 data: 0.0771 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1614 (0.2619) acc1: 98.9583 (90.9809) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1514 data: 0.0608 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.1991 (0.2619) acc1: 96.8750 (90.9937) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1492 data: 0.0587 max mem: 3541 +Test: [2330/2387] eta: 0:00:11 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.2577 (0.2620) acc1: 94.2708 (90.9987) acc5: 100.0000 (100.0000) acc1_neg: 80.729 acc5_neg: 100.000 time: 0.1418 data: 0.0517 max mem: 3541 +Test: [2340/2387] eta: 0:00:09 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.2367 (0.2619) acc1: 95.0521 (91.0186) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1447 data: 0.0546 max mem: 3541 +Test: [2350/2387] eta: 0:00:07 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.2254 (0.2621) acc1: 95.8333 (91.0150) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1422 data: 0.0520 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.2792 (0.2622) acc1: 91.4062 (91.0161) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1361 data: 0.0460 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.2612 (0.2623) acc1: 91.4062 (91.0175) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.1311 data: 0.0411 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.2978 (0.2626) acc1: 89.8438 (91.0077) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1272 data: 0.0372 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 42.922 acc5_pos: 100.000 loss: 0.3119 (0.2627) acc1: 89.8438 (91.0074) acc5: 100.0000 (100.0000) acc1_neg: 90.698 acc5_neg: 100.000 time: 0.1214 data: 0.0341 max mem: 3541 +Test: Total time: 0:07:37 (0.1917 s / it) +* Overall Acc@1 91.007 Acc@5 100.000 + target == real Acc@1 38.450 Acc@5 100.000 + target == fake Acc@1 94.698 Acc@5 100.000 + loss 0.263 + PR-AUC 0.9919 ROC-AUC 0.8943 EER 17.98% (thr=0.7471) +Accuracy of the network on the 916267 test images: 91.0% +[rank0]:[W725 18:04:36.561910429 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:04:58.614000 600735 torch/distributed/run.py:766] +W0725 18:04:58.614000 600735 torch/distributed/run.py:766] ***************************************** +W0725 18:04:58.614000 600735 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:04:58.614000 600735 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:05:35.934149577 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:05:36.529611919 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:05:36.658586798 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:05:36.785489705 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:05:36.061182751 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:05:36.097677594 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W725 18:05:36.098473680 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:05:36.129211697 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed2_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed2_resampled_n116-040-213_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:00:16 acc1_pos: 47.135 acc5_pos: 100.000 loss: 0.7997 (0.7997) acc1: 47.1354 (47.1354) acc5: 100.0000 (100.0000) time: 6.9960 data: 5.1913 max mem: 3531 +Test: [ 10/517] eta: 0:06:05 acc1_pos: 25.521 acc5_pos: 100.000 loss: 0.9191 (0.9184) acc1: 42.7083 (39.7017) acc5: 100.0000 (100.0000) time: 0.7205 data: 0.4723 max mem: 3531 +Test: [ 20/517] eta: 0:03:31 acc1_pos: 26.302 acc5_pos: 100.000 loss: 0.9191 (0.9289) acc1: 38.0208 (38.6161) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0019 max mem: 3531 +Test: [ 30/517] eta: 0:03:18 acc1_pos: 21.615 acc5_pos: 100.000 loss: 0.9576 (0.9455) acc1: 29.4271 (37.6260) acc5: 100.0000 (100.0000) time: 0.2351 data: 0.1406 max mem: 3531 +Test: [ 40/517] eta: 0:03:07 acc1_pos: 28.906 acc5_pos: 100.000 loss: 1.0188 (0.9611) acc1: 28.9062 (35.2325) acc5: 100.0000 (100.0000) time: 0.3599 data: 0.2657 max mem: 3531 +Test: [ 50/517] eta: 0:02:56 acc1_pos: 61.927 acc5_pos: 100.000 loss: 0.9150 (0.9368) acc1: 29.4271 (37.2600) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.3347 data: 0.2406 max mem: 3531 +Test: [ 60/517] eta: 0:02:50 acc1_pos: 61.927 acc5_pos: 100.000 loss: 0.5061 (0.8548) acc1: 76.0417 (44.8600) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.3349 data: 0.2392 max mem: 3531 +Test: [ 70/517] eta: 0:02:35 acc1_pos: 61.927 acc5_pos: 100.000 loss: 0.4032 (0.7888) acc1: 88.5417 (51.2764) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.2699 data: 0.1724 max mem: 3531 +Test: [ 80/517] eta: 0:02:26 acc1_pos: 61.927 acc5_pos: 100.000 loss: 0.4032 (0.7454) acc1: 88.5417 (55.5491) acc5: 100.0000 (100.0000) acc1_neg: 78.646 acc5_neg: 100.000 time: 0.2199 data: 0.1232 max mem: 3531 +Test: [ 90/517] eta: 0:02:19 acc1_pos: 61.927 acc5_pos: 100.000 loss: 0.4201 (0.7067) acc1: 89.8438 (59.5639) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.2499 data: 0.1540 max mem: 3531 +Test: [100/517] eta: 0:02:12 acc1_pos: 61.927 acc5_pos: 100.000 loss: 0.3554 (0.6692) acc1: 93.2292 (63.1214) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2449 data: 0.1492 max mem: 3531 +Test: [110/517] eta: 0:02:06 acc1_pos: 61.927 acc5_pos: 100.000 loss: 0.2919 (0.6342) acc1: 97.3958 (66.2983) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2449 data: 0.1485 max mem: 3531 +Test: [120/517] eta: 0:02:01 acc1_pos: 61.927 acc5_pos: 100.000 loss: 0.3109 (0.6100) acc1: 96.3542 (68.5262) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2399 data: 0.1421 max mem: 3531 +Test: [130/517] eta: 0:01:55 acc1_pos: 61.927 acc5_pos: 100.000 loss: 0.3403 (0.5882) acc1: 93.4896 (70.4695) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2249 data: 0.1281 max mem: 3531 +Test: [140/517] eta: 0:01:51 acc1_pos: 61.927 acc5_pos: 100.000 loss: 0.3033 (0.5669) acc1: 96.8750 (72.3755) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2399 data: 0.1460 max mem: 3531 +Test: [150/517] eta: 0:01:47 acc1_pos: 61.927 acc5_pos: 100.000 loss: 0.2770 (0.5472) acc1: 97.9167 (74.0601) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2499 data: 0.1536 max mem: 3531 +Test: [160/517] eta: 0:01:43 acc1_pos: 61.927 acc5_pos: 100.000 loss: 0.2920 (0.5332) acc1: 95.8333 (75.2637) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2399 data: 0.1435 max mem: 3531 +Test: [170/517] eta: 0:01:39 acc1_pos: 61.927 acc5_pos: 100.000 loss: 0.3164 (0.5196) acc1: 94.0104 (76.4133) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2399 data: 0.1437 max mem: 3531 +Test: [180/517] eta: 0:01:35 acc1_pos: 61.927 acc5_pos: 100.000 loss: 0.2727 (0.5053) acc1: 95.0521 (77.5294) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2349 data: 0.1369 max mem: 3531 +Test: [190/517] eta: 0:01:31 acc1_pos: 61.927 acc5_pos: 100.000 loss: 0.2171 (0.4895) acc1: 98.9583 (78.6813) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2299 data: 0.1347 max mem: 3531 +Test: [200/517] eta: 0:01:28 acc1_pos: 61.927 acc5_pos: 100.000 loss: 0.2161 (0.4771) acc1: 99.4792 (79.6797) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2399 data: 0.1445 max mem: 3531 +Test: [210/517] eta: 0:01:24 acc1_pos: 61.927 acc5_pos: 100.000 loss: 0.2819 (0.4751) acc1: 95.3125 (79.9862) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.2449 data: 0.1446 max mem: 3532 +Test: [220/517] eta: 0:01:21 acc1_pos: 61.927 acc5_pos: 100.000 loss: 0.4297 (0.4733) acc1: 86.1979 (80.2696) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.2499 data: 0.1518 max mem: 3532 +Test: [230/517] eta: 0:01:18 acc1_pos: 61.927 acc5_pos: 100.000 loss: 0.4102 (0.4694) acc1: 88.2812 (80.7765) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2499 data: 0.1531 max mem: 3532 +Test: [240/517] eta: 0:01:15 acc1_pos: 61.927 acc5_pos: 100.000 loss: 0.3368 (0.4634) acc1: 95.8333 (81.4423) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.2549 data: 0.1563 max mem: 3532 +Test: [250/517] eta: 0:01:12 acc1_pos: 61.927 acc5_pos: 100.000 loss: 0.3127 (0.4564) acc1: 97.9167 (82.1132) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2408 data: 0.1459 max mem: 3532 +Test: [260/517] eta: 0:01:09 acc1_pos: 61.927 acc5_pos: 100.000 loss: 0.2717 (0.4498) acc1: 98.9583 (82.7407) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2299 data: 0.1370 max mem: 3532 +Test: [270/517] eta: 0:01:06 acc1_pos: 61.927 acc5_pos: 100.000 loss: 0.3622 (0.4483) acc1: 96.8750 (83.1162) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2353 data: 0.1411 max mem: 3532 +Test: [280/517] eta: 0:01:03 acc1_pos: 61.927 acc5_pos: 100.000 loss: 0.3726 (0.4450) acc1: 94.2708 (83.4510) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.2356 data: 0.1413 max mem: 3532 +Test: [290/517] eta: 0:01:00 acc1_pos: 61.927 acc5_pos: 100.000 loss: 0.3123 (0.4400) acc1: 96.8750 (83.9392) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2453 data: 0.1489 max mem: 3532 +Test: [300/517] eta: 0:00:57 acc1_pos: 61.927 acc5_pos: 100.000 loss: 0.2983 (0.4350) acc1: 98.1771 (84.4226) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2455 data: 0.1497 max mem: 3532 +Test: [310/517] eta: 0:00:55 acc1_pos: 61.927 acc5_pos: 100.000 loss: 0.3084 (0.4323) acc1: 97.1354 (84.7702) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.2548 data: 0.1612 max mem: 3532 +Test: [320/517] eta: 0:00:52 acc1_pos: 61.927 acc5_pos: 100.000 loss: 0.3336 (0.4280) acc1: 96.6146 (85.1433) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2644 data: 0.1712 max mem: 3532 +Test: [330/517] eta: 0:00:49 acc1_pos: 61.927 acc5_pos: 100.000 loss: 0.2611 (0.4226) acc1: 98.6979 (85.5599) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2485 data: 0.1545 max mem: 3532 +Test: [340/517] eta: 0:00:46 acc1_pos: 61.927 acc5_pos: 100.000 loss: 0.2395 (0.4194) acc1: 98.9583 (85.7756) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.2342 data: 0.1369 max mem: 3532 +Test: [350/517] eta: 0:00:44 acc1_pos: 61.927 acc5_pos: 100.000 loss: 0.3944 (0.4197) acc1: 88.2812 (85.7824) acc5: 100.0000 (100.0000) acc1_neg: 71.615 acc5_neg: 100.000 time: 0.2320 data: 0.1346 max mem: 3532 +Test: [360/517] eta: 0:00:41 acc1_pos: 61.927 acc5_pos: 100.000 loss: 0.4036 (0.4190) acc1: 88.2812 (85.8661) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.2266 data: 0.1301 max mem: 3532 +Test: [370/517] eta: 0:00:38 acc1_pos: 61.927 acc5_pos: 100.000 loss: 0.3217 (0.4143) acc1: 95.3125 (86.2190) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2328 data: 0.1360 max mem: 3532 +Test: [380/517] eta: 0:00:35 acc1_pos: 61.927 acc5_pos: 100.000 loss: 0.2562 (0.4109) acc1: 98.4375 (86.4823) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2432 data: 0.1465 max mem: 3532 +Test: [390/517] eta: 0:00:33 acc1_pos: 61.927 acc5_pos: 100.000 loss: 0.2820 (0.4077) acc1: 96.8750 (86.7361) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2406 data: 0.1463 max mem: 3532 +Test: [400/517] eta: 0:00:30 acc1_pos: 61.927 acc5_pos: 100.000 loss: 0.2789 (0.4045) acc1: 96.8750 (86.9759) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2568 data: 0.1634 max mem: 3532 +Test: [410/517] eta: 0:00:27 acc1_pos: 61.927 acc5_pos: 100.000 loss: 0.3162 (0.4033) acc1: 96.3542 (87.1483) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.2692 data: 0.1698 max mem: 3532 +Test: [420/517] eta: 0:00:25 acc1_pos: 61.927 acc5_pos: 100.000 loss: 0.3146 (0.4011) acc1: 97.1354 (87.3849) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2539 data: 0.1534 max mem: 3532 +Test: [430/517] eta: 0:00:22 acc1_pos: 61.927 acc5_pos: 100.000 loss: 0.3146 (0.3997) acc1: 96.0938 (87.5357) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.2606 data: 0.1659 max mem: 3532 +Test: [440/517] eta: 0:00:20 acc1_pos: 61.927 acc5_pos: 100.000 loss: 0.3454 (0.3985) acc1: 95.0521 (87.6807) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.2444 data: 0.1528 max mem: 3533 +Test: [450/517] eta: 0:00:17 acc1_pos: 61.927 acc5_pos: 100.000 loss: 0.3168 (0.3964) acc1: 95.0521 (87.8615) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2306 data: 0.1402 max mem: 3533 +Test: [460/517] eta: 0:00:14 acc1_pos: 61.927 acc5_pos: 100.000 loss: 0.3168 (0.3951) acc1: 95.3125 (87.9892) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.2395 data: 0.1484 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 61.927 acc5_pos: 100.000 loss: 0.2999 (0.3928) acc1: 96.3542 (88.1889) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2235 data: 0.1288 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 61.927 acc5_pos: 100.000 loss: 0.2756 (0.3903) acc1: 98.1771 (88.4096) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2310 data: 0.1332 max mem: 3533 +Test: [490/517] eta: 0:00:06 acc1_pos: 61.927 acc5_pos: 100.000 loss: 0.2689 (0.3876) acc1: 99.2188 (88.6085) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2599 data: 0.1652 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 61.927 acc5_pos: 100.000 loss: 0.3113 (0.3881) acc1: 95.8333 (88.6181) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.2594 data: 0.1680 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 61.927 acc5_pos: 100.000 loss: 0.3619 (0.3871) acc1: 93.7500 (88.6803) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2106 data: 0.1197 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 61.927 acc5_pos: 100.000 loss: 0.2704 (0.3857) acc1: 97.1354 (88.7843) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1810 data: 0.0890 max mem: 3533 +Test: Total time: 0:02:12 (0.2559 s / it) +* Overall Acc@1 88.784 Acc@5 100.000 + target == real Acc@1 35.088 Acc@5 100.000 + target == fake Acc@1 94.355 Acc@5 100.000 + loss 0.386 + PR-AUC 0.9600 ROC-AUC 0.7569 EER 31.16% (thr=0.6978) +Accuracy of the network on the 198427 test images: 88.8% +[rank0]:[W725 18:08:03.013455119 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0725 18:08:27.834000 601329 torch/distributed/run.py:766] +W0725 18:08:27.834000 601329 torch/distributed/run.py:766] ***************************************** +W0725 18:08:27.834000 601329 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:08:27.834000 601329 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W725 18:09:04.345323041 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +[rank7]:[W725 18:09:04.423021183 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:09:04.433661697 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:09:04.513773454 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:09:04.978320801 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:09:04.050226808 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W725 18:09:04.050307000 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:09:04.082872887 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed2_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed2_resampled_n116-040-213_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:33:04 acc1_pos: 62.500 acc5_pos: 100.000 loss: 0.6751 (0.6751) acc1: 62.5000 (62.5000) acc5: 100.0000 (100.0000) time: 5.1019 data: 3.0207 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/389] eta: 0:03:26 acc1_pos: 89.062 acc5_pos: 100.000 loss: 0.6615 (0.6102) acc1: 62.5000 (65.0568) acc5: 100.0000 (100.0000) time: 0.5459 data: 0.2750 max mem: 3531 +Test: [ 20/389] eta: 0:02:01 acc1_pos: 27.344 acc5_pos: 100.000 loss: 0.6615 (0.6824) acc1: 54.9479 (57.4033) acc5: 100.0000 (100.0000) time: 0.0909 data: 0.0004 max mem: 3531 +Test: [ 30/389] eta: 0:01:42 acc1_pos: 67.188 acc5_pos: 100.000 loss: 0.6503 (0.6698) acc1: 55.9896 (59.6942) acc5: 100.0000 (100.0000) time: 0.1414 data: 0.0470 max mem: 3531 +Test: [ 40/389] eta: 0:01:48 acc1_pos: 10.938 acc5_pos: 100.000 loss: 0.6755 (0.7012) acc1: 55.9896 (56.1611) acc5: 100.0000 (100.0000) time: 0.2915 data: 0.1938 max mem: 3531 +Test: [ 50/389] eta: 0:01:38 acc1_pos: 83.333 acc5_pos: 100.000 loss: 0.6080 (0.6816) acc1: 59.1146 (58.5682) acc5: 100.0000 (100.0000) time: 0.2959 data: 0.2018 max mem: 3531 +Test: [ 60/389] eta: 0:01:28 acc1_pos: 48.438 acc5_pos: 100.000 loss: 0.4946 (0.6540) acc1: 79.1667 (61.7188) acc5: 100.0000 (100.0000) time: 0.1802 data: 0.0884 max mem: 3531 +Test: [ 70/389] eta: 0:01:24 acc1_pos: 65.885 acc5_pos: 100.000 loss: 0.5047 (0.6486) acc1: 76.0417 (61.9572) acc5: 100.0000 (100.0000) time: 0.1996 data: 0.1057 max mem: 3531 +Test: [ 80/389] eta: 0:01:21 acc1_pos: 90.104 acc5_pos: 100.000 loss: 0.6789 (0.6584) acc1: 57.2917 (60.9021) acc5: 100.0000 (100.0000) time: 0.2473 data: 0.1533 max mem: 3531 +Test: [ 90/389] eta: 0:01:13 acc1_pos: 74.479 acc5_pos: 100.000 loss: 0.4696 (0.6364) acc1: 74.4792 (62.8148) acc5: 100.0000 (100.0000) time: 0.1746 data: 0.0828 max mem: 3531 +Test: [100/389] eta: 0:01:09 acc1_pos: 76.042 acc5_pos: 100.000 loss: 0.5651 (0.6506) acc1: 66.1458 (61.7806) acc5: 100.0000 (100.0000) time: 0.1507 data: 0.0601 max mem: 3531 +Test: [110/389] eta: 0:01:06 acc1_pos: 95.052 acc5_pos: 100.000 loss: 0.6394 (0.6642) acc1: 63.0208 (61.2660) acc5: 100.0000 (100.0000) time: 0.2060 data: 0.1136 max mem: 3531 +Test: [120/389] eta: 0:01:03 acc1_pos: 75.781 acc5_pos: 100.000 loss: 0.5470 (0.6544) acc1: 71.6146 (62.2848) acc5: 100.0000 (100.0000) time: 0.2065 data: 0.1142 max mem: 3531 +Test: [130/389] eta: 0:00:58 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.5470 (0.6526) acc1: 71.0938 (62.3151) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1636 data: 0.0716 max mem: 3531 +Test: [140/389] eta: 0:00:56 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.5714 (0.6470) acc1: 70.0521 (62.8066) acc5: 100.0000 (100.0000) acc1_neg: 28.385 acc5_neg: 100.000 time: 0.1682 data: 0.0748 max mem: 3531 +Test: [150/389] eta: 0:00:53 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.5868 (0.6477) acc1: 70.0521 (62.6466) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2107 data: 0.1165 max mem: 3531 +Test: [160/389] eta: 0:00:50 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.6198 (0.6370) acc1: 57.5521 (63.4931) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1904 data: 0.0955 max mem: 3531 +Test: [170/389] eta: 0:00:47 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.2333 (0.6227) acc1: 97.9167 (64.6458) acc5: 100.0000 (100.0000) acc1_neg: 66.406 acc5_neg: 100.000 time: 0.1682 data: 0.0725 max mem: 3531 +Test: [180/389] eta: 0:00:45 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.5654 (0.6233) acc1: 69.0104 (64.7171) acc5: 100.0000 (100.0000) acc1_neg: 48.177 acc5_neg: 100.000 time: 0.1765 data: 0.0822 max mem: 3531 +Test: [190/389] eta: 0:00:42 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.6052 (0.6263) acc1: 67.4479 (64.4047) acc5: 100.0000 (100.0000) acc1_neg: 17.448 acc5_neg: 100.000 time: 0.1728 data: 0.0813 max mem: 3531 +Test: [200/389] eta: 0:00:40 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.7514 (0.6309) acc1: 47.3958 (63.8086) acc5: 100.0000 (100.0000) acc1_neg: 37.760 acc5_neg: 100.000 time: 0.1878 data: 0.0959 max mem: 3531 +Test: [210/389] eta: 0:00:38 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.4118 (0.6219) acc1: 52.3438 (64.5068) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2121 data: 0.1182 max mem: 3532 +Test: [220/389] eta: 0:00:35 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.2774 (0.6093) acc1: 97.9167 (65.6592) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.1904 data: 0.0974 max mem: 3532 +Test: [230/389] eta: 0:00:33 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.4965 (0.6107) acc1: 71.8750 (65.4311) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1966 data: 0.1046 max mem: 3532 +Test: [240/389] eta: 0:00:31 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.5453 (0.6077) acc1: 68.4896 (65.6877) acc5: 100.0000 (100.0000) acc1_neg: 70.312 acc5_neg: 100.000 time: 0.1994 data: 0.1048 max mem: 3532 +Test: [250/389] eta: 0:00:29 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.4129 (0.5974) acc1: 84.3750 (66.6760) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1895 data: 0.0928 max mem: 3532 +Test: [260/389] eta: 0:00:26 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.3032 (0.5947) acc1: 94.5312 (67.1566) acc5: 100.0000 (100.0000) acc1_neg: 17.969 acc5_neg: 100.000 time: 0.1805 data: 0.0867 max mem: 3532 +Test: [270/389] eta: 0:00:24 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.5121 (0.6036) acc1: 76.0417 (66.3361) acc5: 100.0000 (100.0000) acc1_neg: 76.042 acc5_neg: 100.000 time: 0.1815 data: 0.0908 max mem: 3532 +Test: [280/389] eta: 0:00:22 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.6685 (0.6067) acc1: 55.7292 (66.0550) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2101 data: 0.1158 max mem: 3532 +Test: [290/389] eta: 0:00:20 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.5780 (0.6034) acc1: 69.0104 (66.4483) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.2128 data: 0.1186 max mem: 3532 +Test: [300/389] eta: 0:00:18 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.5271 (0.6002) acc1: 80.2083 (66.9020) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.1660 data: 0.0751 max mem: 3532 +Test: [310/389] eta: 0:00:16 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.5504 (0.6070) acc1: 69.7917 (66.3426) acc5: 100.0000 (100.0000) acc1_neg: 7.292 acc5_neg: 100.000 time: 0.1607 data: 0.0677 max mem: 3532 +Test: [320/389] eta: 0:00:13 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.6438 (0.6045) acc1: 59.6354 (66.5628) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1537 data: 0.0607 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.8122 (0.6124) acc1: 43.2292 (65.5841) acc5: 100.0000 (100.0000) acc1_neg: 53.385 acc5_neg: 100.000 time: 0.1135 data: 0.0226 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.4922 (0.6021) acc1: 66.6667 (66.4887) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1184 data: 0.0278 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.2418 (0.6099) acc1: 98.6979 (66.0167) acc5: 100.0000 (100.0000) acc1_neg: 20.052 acc5_neg: 100.000 time: 0.1263 data: 0.0358 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.5506 (0.6116) acc1: 71.0938 (65.8717) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.1209 data: 0.0306 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.5442 (0.6109) acc1: 72.1354 (65.8995) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1126 data: 0.0224 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.5084 (0.6089) acc1: 70.0521 (66.0392) acc5: 100.0000 (100.0000) acc1_neg: 70.052 acc5_neg: 100.000 time: 0.1051 data: 0.0149 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.4974 (0.6060) acc1: 75.2604 (66.3272) acc5: 100.0000 (100.0000) acc1_neg: 82.240 acc5_neg: 100.000 time: 0.1084 data: 0.0120 max mem: 3532 +Test: Total time: 0:01:12 (0.1874 s / it) +* Overall Acc@1 66.327 Acc@5 100.000 + target == real Acc@1 62.004 Acc@5 100.000 + target == fake Acc@1 68.337 Acc@5 100.000 + loss 0.606 + PR-AUC 0.8430 ROC-AUC 0.7196 EER 34.59% (thr=0.5264) +Accuracy of the network on the 149358 test images: 66.3% +[rank0]:[W725 18:10:28.166187800 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:10:48.858000 601897 torch/distributed/run.py:766] +W0725 18:10:48.858000 601897 torch/distributed/run.py:766] ***************************************** +W0725 18:10:48.858000 601897 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:10:48.858000 601897 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:11:22.077852963 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:11:22.120490673 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:11:23.702735964 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:11:23.738505421 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:11:24.172866616 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:11:24.206189218 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank4]:[W725 18:11:24.207278483 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:11:24.249569638 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed2_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed2_resampled_n116-040-213_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:27:25 acc1_pos: 88.542 acc5_pos: 100.000 loss: 0.3347 (0.3347) acc1: 88.5417 (88.5417) acc5: 100.0000 (100.0000) time: 4.9412 data: 3.1738 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/333] eta: 0:02:51 acc1_pos: 88.802 acc5_pos: 100.000 loss: 0.2969 (0.2863) acc1: 88.8021 (88.6364) acc5: 100.0000 (100.0000) time: 0.5313 data: 0.2889 max mem: 3531 +Test: [ 20/333] eta: 0:01:42 acc1_pos: 93.229 acc5_pos: 100.000 loss: 0.2736 (0.2832) acc1: 89.5833 (88.9509) acc5: 100.0000 (100.0000) time: 0.0952 data: 0.0052 max mem: 3531 +Test: [ 30/333] eta: 0:01:21 acc1_pos: 60.938 acc5_pos: 100.000 loss: 0.3890 (0.4259) acc1: 75.2604 (78.5282) acc5: 100.0000 (100.0000) time: 0.1248 data: 0.0335 max mem: 3531 +Test: [ 40/333] eta: 0:01:19 acc1_pos: 57.031 acc5_pos: 100.000 loss: 0.7484 (0.5043) acc1: 54.4271 (72.4339) acc5: 100.0000 (100.0000) time: 0.2136 data: 0.1198 max mem: 3531 +Test: [ 50/333] eta: 0:01:17 acc1_pos: 57.031 acc5_pos: 100.000 loss: 0.7414 (0.5503) acc1: 54.4271 (69.1483) acc5: 100.0000 (100.0000) time: 0.2836 data: 0.1904 max mem: 3531 +Test: [ 60/333] eta: 0:01:10 acc1_pos: 47.396 acc5_pos: 100.000 loss: 0.7262 (0.5841) acc1: 54.1667 (66.7094) acc5: 100.0000 (100.0000) time: 0.2352 data: 0.1430 max mem: 3531 +Test: [ 70/333] eta: 0:01:03 acc1_pos: 50.000 acc5_pos: 100.000 loss: 0.7320 (0.6082) acc1: 53.9062 (65.1408) acc5: 100.0000 (100.0000) time: 0.1550 data: 0.0611 max mem: 3531 +Test: [ 80/333] eta: 0:00:59 acc1_pos: 55.990 acc5_pos: 100.000 loss: 0.7449 (0.6244) acc1: 55.2083 (63.7088) acc5: 100.0000 (100.0000) time: 0.1650 data: 0.0717 max mem: 3531 +Test: [ 90/333] eta: 0:00:56 acc1_pos: 70.052 acc5_pos: 100.000 loss: 0.6728 (0.6228) acc1: 60.1562 (64.0224) acc5: 100.0000 (100.0000) time: 0.2048 data: 0.1112 max mem: 3531 +Test: [100/333] eta: 0:00:53 acc1_pos: 71.875 acc5_pos: 100.000 loss: 0.6138 (0.6243) acc1: 65.8854 (63.9439) acc5: 100.0000 (100.0000) time: 0.2060 data: 0.1102 max mem: 3531 +Test: [110/333] eta: 0:00:50 acc1_pos: 71.094 acc5_pos: 100.000 loss: 0.6292 (0.6247) acc1: 62.7604 (63.9616) acc5: 100.0000 (100.0000) time: 0.2000 data: 0.1050 max mem: 3531 +Test: [120/333] eta: 0:00:48 acc1_pos: 71.875 acc5_pos: 100.000 loss: 0.6018 (0.6246) acc1: 64.5833 (64.0152) acc5: 100.0000 (100.0000) time: 0.2098 data: 0.1173 max mem: 3531 +Test: [130/333] eta: 0:00:45 acc1_pos: 82.031 acc5_pos: 100.000 loss: 0.5800 (0.6222) acc1: 68.4896 (64.3567) acc5: 100.0000 (100.0000) time: 0.2048 data: 0.1134 max mem: 3531 +Test: [140/333] eta: 0:00:43 acc1_pos: 81.250 acc5_pos: 100.000 loss: 0.5796 (0.6197) acc1: 71.3542 (64.6110) acc5: 100.0000 (100.0000) time: 0.2140 data: 0.1206 max mem: 3531 +Test: [150/333] eta: 0:00:40 acc1_pos: 74.219 acc5_pos: 100.000 loss: 0.6016 (0.6203) acc1: 67.7083 (64.5851) acc5: 100.0000 (100.0000) time: 0.2060 data: 0.1117 max mem: 3531 +Test: [160/333] eta: 0:00:38 acc1_pos: 57.237 acc5_pos: 100.000 loss: 0.6252 (0.6302) acc1: 64.3229 (63.9800) acc5: 100.0000 (100.0000) acc1_neg: 44.271 acc5_neg: 100.000 time: 0.2005 data: 0.1083 max mem: 3531 +Test: [170/333] eta: 0:00:35 acc1_pos: 57.237 acc5_pos: 100.000 loss: 0.7873 (0.6429) acc1: 51.5625 (63.1351) acc5: 100.0000 (100.0000) acc1_neg: 51.042 acc5_neg: 100.000 time: 0.1941 data: 0.1007 max mem: 3531 +Test: [180/333] eta: 0:00:32 acc1_pos: 57.237 acc5_pos: 100.000 loss: 0.7591 (0.6470) acc1: 54.6875 (62.8511) acc5: 100.0000 (100.0000) acc1_neg: 53.906 acc5_neg: 100.000 time: 0.1287 data: 0.0314 max mem: 3531 +Test: [190/333] eta: 0:00:30 acc1_pos: 57.237 acc5_pos: 100.000 loss: 0.7040 (0.6485) acc1: 58.5938 (62.7931) acc5: 100.0000 (100.0000) acc1_neg: 51.562 acc5_neg: 100.000 time: 0.1747 data: 0.0773 max mem: 3531 +Test: [200/333] eta: 0:00:28 acc1_pos: 57.237 acc5_pos: 100.000 loss: 0.6748 (0.6504) acc1: 60.1562 (62.8200) acc5: 100.0000 (100.0000) acc1_neg: 60.156 acc5_neg: 100.000 time: 0.2548 data: 0.1564 max mem: 3531 +Test: [210/333] eta: 0:00:26 acc1_pos: 57.237 acc5_pos: 100.000 loss: 0.7385 (0.6568) acc1: 59.3750 (62.5457) acc5: 100.0000 (100.0000) acc1_neg: 54.167 acc5_neg: 100.000 time: 0.2414 data: 0.1442 max mem: 3532 +Test: [220/333] eta: 0:00:24 acc1_pos: 57.237 acc5_pos: 100.000 loss: 0.7933 (0.6660) acc1: 54.9479 (61.9709) acc5: 100.0000 (100.0000) acc1_neg: 67.969 acc5_neg: 100.000 time: 0.1917 data: 0.0993 max mem: 3532 +Test: [230/333] eta: 0:00:21 acc1_pos: 57.237 acc5_pos: 100.000 loss: 0.5398 (0.6593) acc1: 71.6146 (62.6172) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.1784 data: 0.0859 max mem: 3532 +Test: [240/333] eta: 0:00:19 acc1_pos: 57.237 acc5_pos: 100.000 loss: 0.5355 (0.6580) acc1: 71.6146 (62.7042) acc5: 100.0000 (100.0000) acc1_neg: 63.802 acc5_neg: 100.000 time: 0.1991 data: 0.1067 max mem: 3532 +Test: [250/333] eta: 0:00:17 acc1_pos: 57.237 acc5_pos: 100.000 loss: 0.6474 (0.6588) acc1: 63.8021 (62.6971) acc5: 100.0000 (100.0000) acc1_neg: 53.646 acc5_neg: 100.000 time: 0.2075 data: 0.1164 max mem: 3532 +Test: [260/333] eta: 0:00:15 acc1_pos: 57.237 acc5_pos: 100.000 loss: 0.6779 (0.6605) acc1: 61.7188 (62.6267) acc5: 100.0000 (100.0000) acc1_neg: 63.802 acc5_neg: 100.000 time: 0.2039 data: 0.1105 max mem: 3532 +Test: [270/333] eta: 0:00:13 acc1_pos: 57.237 acc5_pos: 100.000 loss: 0.6398 (0.6587) acc1: 65.3646 (62.8527) acc5: 100.0000 (100.0000) acc1_neg: 65.625 acc5_neg: 100.000 time: 0.2024 data: 0.1093 max mem: 3532 +Test: [280/333] eta: 0:00:11 acc1_pos: 57.237 acc5_pos: 100.000 loss: 0.6209 (0.6629) acc1: 65.3646 (62.5343) acc5: 100.0000 (100.0000) acc1_neg: 58.333 acc5_neg: 100.000 time: 0.1878 data: 0.0968 max mem: 3532 +Test: [290/333] eta: 0:00:08 acc1_pos: 57.237 acc5_pos: 100.000 loss: 0.9004 (0.6709) acc1: 44.7917 (61.9085) acc5: 100.0000 (100.0000) acc1_neg: 47.917 acc5_neg: 100.000 time: 0.1651 data: 0.0738 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 57.237 acc5_pos: 100.000 loss: 0.8978 (0.6776) acc1: 45.0521 (61.4488) acc5: 100.0000 (100.0000) acc1_neg: 47.917 acc5_neg: 100.000 time: 0.1616 data: 0.0707 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 57.237 acc5_pos: 100.000 loss: 0.7804 (0.6754) acc1: 53.1250 (61.6467) acc5: 100.0000 (100.0000) acc1_neg: 64.323 acc5_neg: 100.000 time: 0.1469 data: 0.0559 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 57.237 acc5_pos: 100.000 loss: 0.5962 (0.6744) acc1: 69.2708 (61.7780) acc5: 100.0000 (100.0000) acc1_neg: 46.615 acc5_neg: 100.000 time: 0.1353 data: 0.0441 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 57.237 acc5_pos: 100.000 loss: 0.8580 (0.6824) acc1: 48.1771 (61.2892) acc5: 100.0000 (100.0000) acc1_neg: 47.656 acc5_neg: 100.000 time: 0.1333 data: 0.0426 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 57.237 acc5_pos: 100.000 loss: 0.9143 (0.6839) acc1: 47.1354 (61.1946) acc5: 100.0000 (100.0000) acc1_neg: 45.915 acc5_neg: 100.000 time: 0.1357 data: 0.0414 max mem: 3532 +Test: Total time: 0:01:06 (0.2005 s / it) +* Overall Acc@1 61.195 Acc@5 100.000 + target == real Acc@1 64.648 Acc@5 100.000 + target == fake Acc@1 58.121 Acc@5 100.000 + loss 0.684 + PR-AUC 0.6666 ROC-AUC 0.6585 EER 38.53% (thr=0.4758) +Accuracy of the network on the 127843 test images: 61.2% +[rank0]:[W725 18:12:41.460497785 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:13:02.061000 602465 torch/distributed/run.py:766] +W0725 18:13:02.061000 602465 torch/distributed/run.py:766] ***************************************** +W0725 18:13:02.061000 602465 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:13:02.061000 602465 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:13:32.128684775 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:13:33.527975719 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:13:33.608407871 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +| distributed init (rank 5): env:// +[rank0]:[W725 18:13:33.055217245 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:13:33.058078551 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank6]:[W725 18:13:33.058133909 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:13:33.091472897 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W725 18:13:33.091511832 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed2_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed2_resampled_n116-040-213_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:22:52 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.1799 (0.1799) acc1: 96.8750 (96.8750) acc5: 100.0000 (100.0000) time: 5.8406 data: 4.4257 max mem: 3531 +Test: [ 10/235] eta: 0:02:18 acc1_pos: 95.833 acc5_pos: 100.000 loss: 0.1857 (0.1862) acc1: 96.8750 (96.5436) acc5: 100.0000 (100.0000) time: 0.6143 data: 0.4028 max mem: 3531 +Test: [ 20/235] eta: 0:01:24 acc1_pos: 96.354 acc5_pos: 100.000 loss: 0.1813 (0.1836) acc1: 96.6146 (96.6394) acc5: 100.0000 (100.0000) time: 0.1226 data: 0.0314 max mem: 3531 +Test: [ 30/235] eta: 0:01:12 acc1_pos: 97.396 acc5_pos: 100.000 loss: 0.1782 (0.1822) acc1: 96.6146 (96.7742) acc5: 100.0000 (100.0000) time: 0.2118 data: 0.1206 max mem: 3531 +Test: [ 40/235] eta: 0:01:05 acc1_pos: 96.094 acc5_pos: 100.000 loss: 0.1782 (0.1821) acc1: 96.6146 (96.7607) acc5: 100.0000 (100.0000) time: 0.2718 data: 0.1809 max mem: 3531 +Test: [ 50/235] eta: 0:00:58 acc1_pos: 94.792 acc5_pos: 100.000 loss: 0.1825 (0.1827) acc1: 96.6146 (96.7269) acc5: 100.0000 (100.0000) time: 0.2571 data: 0.1641 max mem: 3531 +Test: [ 60/235] eta: 0:00:51 acc1_pos: 96.615 acc5_pos: 100.000 loss: 0.1863 (0.1838) acc1: 96.6146 (96.7000) acc5: 100.0000 (100.0000) time: 0.2079 data: 0.1135 max mem: 3531 +Test: [ 70/235] eta: 0:00:47 acc1_pos: 98.462 acc5_pos: 100.000 loss: 0.1986 (0.3472) acc1: 94.7917 (87.0672) acc5: 100.0000 (100.0000) acc1_neg: 18.229 acc5_neg: 100.000 time: 0.2045 data: 0.1120 max mem: 3531 +Test: [ 80/235] eta: 0:00:43 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4960 (0.4911) acc1: 20.3125 (78.7648) acc5: 100.0000 (100.0000) acc1_neg: 16.927 acc5_neg: 100.000 time: 0.2399 data: 0.1475 max mem: 3531 +Test: [ 90/235] eta: 0:00:39 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.5073 (0.6032) acc1: 19.7917 (72.2070) acc5: 100.0000 (100.0000) acc1_neg: 17.188 acc5_neg: 100.000 time: 0.2132 data: 0.1180 max mem: 3531 +Test: [100/235] eta: 0:00:35 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.5076 (0.6909) acc1: 19.0104 (67.0663) acc5: 100.0000 (100.0000) acc1_neg: 19.010 acc5_neg: 100.000 time: 0.2052 data: 0.1100 max mem: 3531 +Test: [110/235] eta: 0:00:32 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4959 (0.7636) acc1: 19.5312 (62.7839) acc5: 100.0000 (100.0000) acc1_neg: 22.396 acc5_neg: 100.000 time: 0.2064 data: 0.1134 max mem: 3531 +Test: [120/235] eta: 0:00:29 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.5440 (0.8381) acc1: 15.6250 (58.5894) acc5: 100.0000 (100.0000) acc1_neg: 11.719 acc5_neg: 100.000 time: 0.1916 data: 0.0991 max mem: 3531 +Test: [130/235] eta: 0:00:26 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.6703 (0.9017) acc1: 11.7188 (55.0414) acc5: 100.0000 (100.0000) acc1_neg: 12.500 acc5_neg: 100.000 time: 0.1952 data: 0.1030 max mem: 3531 +Test: [140/235] eta: 0:00:23 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.6885 (0.9593) acc1: 11.1979 (51.9245) acc5: 100.0000 (100.0000) acc1_neg: 8.854 acc5_neg: 100.000 time: 0.1802 data: 0.0887 max mem: 3531 +Test: [150/235] eta: 0:00:20 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.6903 (1.0063) acc1: 11.7188 (49.2808) acc5: 100.0000 (100.0000) acc1_neg: 12.240 acc5_neg: 100.000 time: 0.1753 data: 0.0842 max mem: 3531 +Test: [160/235] eta: 0:00:17 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.6730 (1.0479) acc1: 11.9792 (47.0335) acc5: 100.0000 (100.0000) acc1_neg: 13.021 acc5_neg: 100.000 time: 0.1854 data: 0.0922 max mem: 3531 +Test: [170/235] eta: 0:00:15 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.6163 (1.0647) acc1: 14.8438 (45.7526) acc5: 100.0000 (100.0000) acc1_neg: 23.438 acc5_neg: 100.000 time: 0.2076 data: 0.1123 max mem: 3531 +Test: [180/235] eta: 0:00:12 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.3037 (1.0773) acc1: 26.3021 (44.7010) acc5: 100.0000 (100.0000) acc1_neg: 27.604 acc5_neg: 100.000 time: 0.2149 data: 0.1193 max mem: 3531 +Test: [190/235] eta: 0:00:10 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.2844 (1.0886) acc1: 26.3021 (43.7432) acc5: 100.0000 (100.0000) acc1_neg: 26.562 acc5_neg: 100.000 time: 0.1771 data: 0.0834 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.2960 (1.0998) acc1: 26.0417 (42.7900) acc5: 100.0000 (100.0000) acc1_neg: 23.438 acc5_neg: 100.000 time: 0.1423 data: 0.0507 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.3051 (1.1089) acc1: 25.2604 (41.9913) acc5: 100.0000 (100.0000) acc1_neg: 26.042 acc5_neg: 100.000 time: 0.1586 data: 0.0677 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.3006 (1.1171) acc1: 26.0417 (41.2755) acc5: 100.0000 (100.0000) acc1_neg: 23.698 acc5_neg: 100.000 time: 0.1769 data: 0.0863 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.2841 (1.1241) acc1: 26.5625 (40.6678) acc5: 100.0000 (100.0000) acc1_neg: 28.125 acc5_neg: 100.000 time: 0.1440 data: 0.0533 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.2863 (1.1277) acc1: 26.5625 (40.4711) acc5: 100.0000 (100.0000) acc1_neg: 34.783 acc5_neg: 100.000 time: 0.1226 data: 0.0348 max mem: 3532 +Test: Total time: 0:00:50 (0.2138 s / it) +* Overall Acc@1 40.471 Acc@5 100.000 + target == real Acc@1 96.708 Acc@5 100.000 + target == fake Acc@1 20.132 Acc@5 100.000 + loss 1.128 + PR-AUC 0.8816 ROC-AUC 0.7356 EER 32.72% (thr=0.1548) +Accuracy of the network on the 89879 test images: 40.5% +[rank0]:[W725 18:14:32.765364766 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:14:53.631000 603036 torch/distributed/run.py:766] +W0725 18:14:53.631000 603036 torch/distributed/run.py:766] ***************************************** +W0725 18:14:53.631000 603036 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:14:53.631000 603036 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:15:26.362493466 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:15:26.846380375 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:15:26.957036087 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:15:26.005242688 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:15:26.124499768 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:15:26.170694757 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:15:27.228663615 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:15:27.423093834 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed2_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed2_resampled_n116-040-213_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:27:48 acc1_pos: 59.375 acc5_pos: 100.000 loss: 0.6646 (0.6646) acc1: 59.3750 (59.3750) acc5: 100.0000 (100.0000) time: 6.2036 data: 4.8813 max mem: 3531 +Test: [ 10/269] eta: 0:02:48 acc1_pos: 69.792 acc5_pos: 100.000 loss: 0.6050 (0.5963) acc1: 66.6667 (67.9924) acc5: 100.0000 (100.0000) time: 0.6494 data: 0.4442 max mem: 3531 +Test: [ 20/269] eta: 0:01:36 acc1_pos: 82.031 acc5_pos: 100.000 loss: 0.5940 (0.5787) acc1: 67.4479 (70.0025) acc5: 100.0000 (100.0000) time: 0.0969 data: 0.0004 max mem: 3531 +Test: [ 30/269] eta: 0:01:23 acc1_pos: 68.229 acc5_pos: 100.000 loss: 0.5356 (0.5651) acc1: 76.3021 (71.7490) acc5: 100.0000 (100.0000) time: 0.1849 data: 0.0864 max mem: 3531 +Test: [ 40/269] eta: 0:01:13 acc1_pos: 58.621 acc5_pos: 100.000 loss: 0.5380 (0.5601) acc1: 75.5208 (72.5483) acc5: 100.0000 (100.0000) acc1_neg: 72.135 acc5_neg: 100.000 time: 0.2549 data: 0.1575 max mem: 3531 +Test: [ 50/269] eta: 0:01:05 acc1_pos: 58.621 acc5_pos: 100.000 loss: 0.5562 (0.5722) acc1: 70.8333 (71.4614) acc5: 100.0000 (100.0000) acc1_neg: 45.833 acc5_neg: 100.000 time: 0.2199 data: 0.1224 max mem: 3531 +Test: [ 60/269] eta: 0:00:56 acc1_pos: 58.621 acc5_pos: 100.000 loss: 0.6440 (0.5939) acc1: 63.5417 (69.1855) acc5: 100.0000 (100.0000) acc1_neg: 42.969 acc5_neg: 100.000 time: 0.1649 data: 0.0664 max mem: 3531 +Test: [ 70/269] eta: 0:00:49 acc1_pos: 58.621 acc5_pos: 100.000 loss: 0.6951 (0.6021) acc1: 56.2500 (67.7597) acc5: 100.0000 (100.0000) acc1_neg: 63.802 acc5_neg: 100.000 time: 0.1150 data: 0.0160 max mem: 3531 +Test: [ 80/269] eta: 0:00:44 acc1_pos: 58.621 acc5_pos: 100.000 loss: 0.6343 (0.6047) acc1: 60.6771 (67.2422) acc5: 100.0000 (100.0000) acc1_neg: 60.677 acc5_neg: 100.000 time: 0.1249 data: 0.0282 max mem: 3531 +Test: [ 90/269] eta: 0:00:40 acc1_pos: 58.621 acc5_pos: 100.000 loss: 0.6594 (0.6178) acc1: 56.5104 (65.6450) acc5: 100.0000 (100.0000) acc1_neg: 51.302 acc5_neg: 100.000 time: 0.1449 data: 0.0517 max mem: 3531 +Test: [100/269] eta: 0:00:37 acc1_pos: 58.621 acc5_pos: 100.000 loss: 0.7290 (0.6282) acc1: 52.0833 (64.5576) acc5: 100.0000 (100.0000) acc1_neg: 49.479 acc5_neg: 100.000 time: 0.1699 data: 0.0740 max mem: 3531 +Test: [110/269] eta: 0:00:34 acc1_pos: 58.621 acc5_pos: 100.000 loss: 0.7341 (0.6405) acc1: 51.8229 (63.1921) acc5: 100.0000 (100.0000) acc1_neg: 44.271 acc5_neg: 100.000 time: 0.1799 data: 0.0830 max mem: 3531 +Test: [120/269] eta: 0:00:31 acc1_pos: 58.621 acc5_pos: 100.000 loss: 0.7283 (0.6433) acc1: 52.3438 (62.8788) acc5: 100.0000 (100.0000) acc1_neg: 50.260 acc5_neg: 100.000 time: 0.1649 data: 0.0705 max mem: 3531 +Test: [130/269] eta: 0:00:29 acc1_pos: 58.621 acc5_pos: 100.000 loss: 0.7220 (0.6508) acc1: 53.1250 (62.0607) acc5: 100.0000 (100.0000) acc1_neg: 41.927 acc5_neg: 100.000 time: 0.1699 data: 0.0738 max mem: 3531 +Test: [140/269] eta: 0:00:26 acc1_pos: 58.621 acc5_pos: 100.000 loss: 0.7759 (0.6604) acc1: 45.0521 (60.9393) acc5: 100.0000 (100.0000) acc1_neg: 45.052 acc5_neg: 100.000 time: 0.1549 data: 0.0598 max mem: 3531 +Test: [150/269] eta: 0:00:24 acc1_pos: 58.621 acc5_pos: 100.000 loss: 0.7338 (0.6600) acc1: 48.1771 (60.8754) acc5: 100.0000 (100.0000) acc1_neg: 65.885 acc5_neg: 100.000 time: 0.1716 data: 0.0795 max mem: 3531 +Test: [160/269] eta: 0:00:22 acc1_pos: 58.621 acc5_pos: 100.000 loss: 0.6194 (0.6586) acc1: 62.7604 (60.9812) acc5: 100.0000 (100.0000) acc1_neg: 62.760 acc5_neg: 100.000 time: 0.2269 data: 0.1352 max mem: 3531 +Test: [170/269] eta: 0:00:20 acc1_pos: 58.621 acc5_pos: 100.000 loss: 0.6712 (0.6620) acc1: 55.4688 (60.6192) acc5: 100.0000 (100.0000) acc1_neg: 59.635 acc5_neg: 100.000 time: 0.2082 data: 0.1134 max mem: 3531 +Test: [180/269] eta: 0:00:18 acc1_pos: 58.621 acc5_pos: 100.000 loss: 0.6526 (0.6599) acc1: 59.6354 (60.8454) acc5: 100.0000 (100.0000) acc1_neg: 57.292 acc5_neg: 100.000 time: 0.1831 data: 0.0892 max mem: 3531 +Test: [190/269] eta: 0:00:16 acc1_pos: 58.621 acc5_pos: 100.000 loss: 0.6433 (0.6597) acc1: 62.5000 (60.9116) acc5: 100.0000 (100.0000) acc1_neg: 56.510 acc5_neg: 100.000 time: 0.1999 data: 0.1065 max mem: 3531 +Test: [200/269] eta: 0:00:14 acc1_pos: 58.621 acc5_pos: 100.000 loss: 0.6802 (0.6651) acc1: 56.2500 (60.4102) acc5: 100.0000 (100.0000) acc1_neg: 65.104 acc5_neg: 100.000 time: 0.1997 data: 0.1057 max mem: 3531 +Test: [210/269] eta: 0:00:12 acc1_pos: 58.621 acc5_pos: 100.000 loss: 0.6787 (0.6651) acc1: 58.0729 (60.3525) acc5: 100.0000 (100.0000) acc1_neg: 66.927 acc5_neg: 100.000 time: 0.2159 data: 0.1245 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 58.621 acc5_pos: 100.000 loss: 0.6403 (0.6643) acc1: 58.8542 (60.4662) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.2209 data: 0.1296 max mem: 3532 +Test: [230/269] eta: 0:00:07 acc1_pos: 58.621 acc5_pos: 100.000 loss: 0.5721 (0.6600) acc1: 66.9271 (60.8732) acc5: 100.0000 (100.0000) acc1_neg: 70.052 acc5_neg: 100.000 time: 0.1954 data: 0.1039 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 58.621 acc5_pos: 100.000 loss: 0.6384 (0.6629) acc1: 61.1979 (60.6317) acc5: 100.0000 (100.0000) acc1_neg: 56.771 acc5_neg: 100.000 time: 0.2239 data: 0.1324 max mem: 3532 +Test: [250/269] eta: 0:00:03 acc1_pos: 58.621 acc5_pos: 100.000 loss: 0.7199 (0.6676) acc1: 51.0417 (60.1106) acc5: 100.0000 (100.0000) acc1_neg: 22.135 acc5_neg: 100.000 time: 0.2234 data: 0.1279 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 58.621 acc5_pos: 100.000 loss: 0.8283 (0.6752) acc1: 39.0625 (59.2652) acc5: 100.0000 (100.0000) acc1_neg: 48.177 acc5_neg: 100.000 time: 0.1542 data: 0.0576 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 58.621 acc5_pos: 100.000 loss: 0.7404 (0.6758) acc1: 40.6250 (59.1967) acc5: 100.0000 (100.0000) acc1_neg: 68.766 acc5_neg: 100.000 time: 0.1344 data: 0.0372 max mem: 3532 +Test: Total time: 0:00:54 (0.2022 s / it) +* Overall Acc@1 59.197 Acc@5 100.000 + target == real Acc@1 71.533 Acc@5 100.000 + target == fake Acc@1 57.513 Acc@5 100.000 + loss 0.676 + PR-AUC 0.9406 ROC-AUC 0.7012 EER 35.58% (thr=0.4622) +Accuracy of the network on the 103293 test images: 59.2% +[rank0]:[W725 18:16:32.244075420 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-25T18:16:38Z] eval_done tag=fixed500k_fake_n64_seed2_resampled +[2026-07-25T18:16:38Z] eval_start tag=fixed500k_real_n32_seed3_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n32_classseed3_resampled_n116-040-213_20260725_112559/best_checkpoint.pth +started_utc=2026-07-25T18:16:38Z +hostname=n116-040-213 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n32_classseed3_resampled_n116-040-213_20260725_112559/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed3_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0725 18:16:52.392000 603622 torch/distributed/run.py:766] +W0725 18:16:52.392000 603622 torch/distributed/run.py:766] ***************************************** +W0725 18:16:52.392000 603622 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:16:52.392000 603622 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:17:22.103889464 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:17:23.538424302 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:17:23.651597475 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:17:23.163441591 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +[rank4]:[W725 18:17:24.204701661 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:17:24.207667027 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:17:24.210435158 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:17:24.243354879 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed3_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n32_classseed3_resampled_n116-040-213_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:19:18 acc1_pos: 61.719 acc5_pos: 100.000 loss: 0.6004 (0.6004) acc1: 61.7188 (61.7188) acc5: 100.0000 (100.0000) time: 4.9507 data: 2.7865 max mem: 3531 +Test: [ 10/234] eta: 0:02:04 acc1_pos: 28.385 acc5_pos: 100.000 loss: 0.7620 (0.7425) acc1: 49.7396 (49.1004) acc5: 100.0000 (100.0000) time: 0.5559 data: 0.2774 max mem: 3531 +Test: [ 20/234] eta: 0:01:14 acc1_pos: 94.271 acc5_pos: 100.000 loss: 0.7291 (0.6727) acc1: 53.6458 (57.3289) acc5: 100.0000 (100.0000) time: 0.1172 data: 0.0251 max mem: 3531 +Test: [ 30/234] eta: 0:00:58 acc1_pos: 63.281 acc5_pos: 100.000 loss: 0.2442 (0.5501) acc1: 94.0104 (67.7587) acc5: 100.0000 (100.0000) time: 0.1391 data: 0.0443 max mem: 3531 +Test: [ 40/234] eta: 0:00:51 acc1_pos: 98.958 acc5_pos: 100.000 loss: 0.2442 (0.5323) acc1: 94.0104 (69.0930) acc5: 100.0000 (100.0000) time: 0.1844 data: 0.0891 max mem: 3531 +Test: [ 50/234] eta: 0:00:47 acc1_pos: 98.177 acc5_pos: 100.000 loss: 0.1484 (0.4555) acc1: 96.8750 (74.6170) acc5: 100.0000 (100.0000) time: 0.2109 data: 0.1172 max mem: 3531 +Test: [ 60/234] eta: 0:00:43 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.1162 (0.3941) acc1: 98.1771 (78.7440) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2204 data: 0.1254 max mem: 3531 +Test: [ 70/234] eta: 0:00:39 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.0790 (0.4108) acc1: 100.0000 (77.9783) acc5: 100.0000 (100.0000) acc1_neg: 59.115 acc5_neg: 100.000 time: 0.2033 data: 0.1073 max mem: 3531 +Test: [ 80/234] eta: 0:00:35 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.6575 (0.4264) acc1: 63.0208 (77.4820) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1644 data: 0.0714 max mem: 3531 +Test: [ 90/234] eta: 0:00:32 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.2723 (0.4034) acc1: 91.4062 (79.2926) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.1749 data: 0.0797 max mem: 3531 +Test: [100/234] eta: 0:00:29 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.3023 (0.4080) acc1: 88.0208 (79.2157) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1949 data: 0.0981 max mem: 3531 +Test: [110/234] eta: 0:00:27 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.4130 (0.4065) acc1: 78.6458 (79.5561) acc5: 100.0000 (100.0000) acc1_neg: 78.646 acc5_neg: 100.000 time: 0.1799 data: 0.0825 max mem: 3531 +Test: [120/234] eta: 0:00:24 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.3513 (0.4127) acc1: 83.8542 (79.5046) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1749 data: 0.0790 max mem: 3531 +Test: [130/234] eta: 0:00:21 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.3954 (0.4151) acc1: 83.8542 (79.6835) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.1713 data: 0.0773 max mem: 3531 +Test: [140/234] eta: 0:00:19 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.4921 (0.4365) acc1: 78.3854 (78.8231) acc5: 100.0000 (100.0000) acc1_neg: 35.938 acc5_neg: 100.000 time: 0.1717 data: 0.0776 max mem: 3531 +Test: [150/234] eta: 0:00:17 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.4924 (0.4466) acc1: 73.1771 (78.3182) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.1745 data: 0.0812 max mem: 3531 +Test: [160/234] eta: 0:00:15 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.4368 (0.4466) acc1: 79.1667 (78.4922) acc5: 100.0000 (100.0000) acc1_neg: 72.656 acc5_neg: 100.000 time: 0.1660 data: 0.0736 max mem: 3531 +Test: [170/234] eta: 0:00:13 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.4857 (0.4485) acc1: 78.9062 (78.5560) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1839 data: 0.0911 max mem: 3531 +Test: [180/234] eta: 0:00:10 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.5009 (0.4591) acc1: 76.5625 (77.9595) acc5: 100.0000 (100.0000) acc1_neg: 36.198 acc5_neg: 100.000 time: 0.1825 data: 0.0881 max mem: 3531 +Test: [190/234] eta: 0:00:08 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.6873 (0.4736) acc1: 59.6354 (76.8925) acc5: 100.0000 (100.0000) acc1_neg: 57.552 acc5_neg: 100.000 time: 0.1649 data: 0.0708 max mem: 3531 +Test: [200/234] eta: 0:00:06 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.6873 (0.4669) acc1: 59.6354 (77.3023) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1994 data: 0.1032 max mem: 3531 +Test: [210/234] eta: 0:00:04 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.1217 (0.4510) acc1: 99.7396 (78.3546) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2074 data: 0.1113 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.1476 (0.4384) acc1: 98.6979 (79.2044) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1598 data: 0.0638 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.3410 (0.4506) acc1: 62.2396 (78.2231) acc5: 100.0000 (100.0000) acc1_neg: 55.990 acc5_neg: 100.000 time: 0.1294 data: 0.0353 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.7096 (0.4540) acc1: 56.5104 (77.9570) acc5: 100.0000 (100.0000) acc1_neg: 56.420 acc5_neg: 100.000 time: 0.1125 data: 0.0189 max mem: 3532 +Test: Total time: 0:00:45 (0.1939 s / it) +* Overall Acc@1 77.957 Acc@5 100.000 + target == real Acc@1 74.773 Acc@5 100.000 + target == fake Acc@1 78.854 Acc@5 100.000 + loss 0.454 + PR-AUC 0.9607 ROC-AUC 0.8691 EER 22.70% (thr=0.5239) +Accuracy of the network on the 89729 test images: 78.0% +[rank0]:[W725 18:18:21.313142459 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:18:40.016000 604192 torch/distributed/run.py:766] +W0725 18:18:40.016000 604192 torch/distributed/run.py:766] ***************************************** +W0725 18:18:40.016000 604192 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:18:40.016000 604192 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:19:09.861613605 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:19:10.233694856 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:19:10.317905250 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:19:10.322219234 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:19:10.490824553 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:19:10.711802475 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:19:10.717793241 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:19:10.732574743 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed3_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n32_classseed3_resampled_n116-040-213_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 3:47:15 acc1_pos: 64.062 acc5_pos: 100.000 loss: 0.6097 (0.6097) acc1: 64.0625 (64.0625) acc5: 100.0000 (100.0000) time: 5.7126 data: 2.9299 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/2387] eta: 0:25:31 acc1_pos: 28.385 acc5_pos: 100.000 loss: 0.7532 (0.7666) acc1: 53.3854 (52.4621) acc5: 100.0000 (100.0000) time: 0.6445 data: 0.3067 max mem: 3531 +Test: [ 20/2387] eta: 0:16:08 acc1_pos: 38.802 acc5_pos: 100.000 loss: 0.7905 (0.7987) acc1: 52.6042 (48.8715) acc5: 100.0000 (100.0000) time: 0.1438 data: 0.0495 max mem: 3531 +Test: [ 30/2387] eta: 0:14:36 acc1_pos: 36.198 acc5_pos: 100.000 loss: 0.7735 (0.7944) acc1: 47.6562 (49.8068) acc5: 100.0000 (100.0000) time: 0.2222 data: 0.1285 max mem: 3531 +Test: [ 40/2387] eta: 0:14:19 acc1_pos: 47.656 acc5_pos: 100.000 loss: 0.7618 (0.7923) acc1: 51.5625 (49.6888) acc5: 100.0000 (100.0000) time: 0.3215 data: 0.2301 max mem: 3531 +Test: [ 50/2387] eta: 0:12:53 acc1_pos: 41.927 acc5_pos: 100.000 loss: 0.7618 (0.7923) acc1: 48.9583 (49.5813) acc5: 100.0000 (100.0000) time: 0.2677 data: 0.1752 max mem: 3531 +Test: [ 60/2387] eta: 0:12:11 acc1_pos: 50.260 acc5_pos: 100.000 loss: 0.8291 (0.8060) acc1: 39.0625 (47.8953) acc5: 100.0000 (100.0000) time: 0.2084 data: 0.1119 max mem: 3531 +Test: [ 70/2387] eta: 0:11:28 acc1_pos: 39.323 acc5_pos: 100.000 loss: 0.9160 (0.8293) acc1: 36.4583 (45.6903) acc5: 100.0000 (100.0000) time: 0.2110 data: 0.1129 max mem: 3531 +Test: [ 80/2387] eta: 0:11:07 acc1_pos: 73.177 acc5_pos: 100.000 loss: 0.8587 (0.8170) acc1: 42.9688 (47.0808) acc5: 100.0000 (100.0000) time: 0.2121 data: 0.1163 max mem: 3531 +Test: [ 90/2387] eta: 0:10:45 acc1_pos: 62.760 acc5_pos: 100.000 loss: 0.6754 (0.7977) acc1: 59.1146 (48.6836) acc5: 100.0000 (100.0000) time: 0.2238 data: 0.1295 max mem: 3531 +Test: [ 100/2387] eta: 0:10:26 acc1_pos: 54.948 acc5_pos: 100.000 loss: 0.6754 (0.7913) acc1: 59.1146 (49.3992) acc5: 100.0000 (100.0000) time: 0.2127 data: 0.1171 max mem: 3531 +Test: [ 110/2387] eta: 0:10:11 acc1_pos: 46.354 acc5_pos: 100.000 loss: 0.7032 (0.7811) acc1: 54.9479 (50.3590) acc5: 100.0000 (100.0000) time: 0.2110 data: 0.1148 max mem: 3531 +Test: [ 120/2387] eta: 0:10:02 acc1_pos: 41.927 acc5_pos: 100.000 loss: 0.7065 (0.7778) acc1: 54.4271 (50.4778) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1277 max mem: 3531 +Test: [ 130/2387] eta: 0:09:48 acc1_pos: 49.479 acc5_pos: 100.000 loss: 0.7338 (0.7715) acc1: 55.4688 (51.0417) acc5: 100.0000 (100.0000) time: 0.2188 data: 0.1200 max mem: 3531 +Test: [ 140/2387] eta: 0:09:39 acc1_pos: 69.010 acc5_pos: 100.000 loss: 0.6899 (0.7668) acc1: 57.8125 (51.4942) acc5: 100.0000 (100.0000) time: 0.2106 data: 0.1142 max mem: 3531 +Test: [ 150/2387] eta: 0:09:33 acc1_pos: 65.885 acc5_pos: 100.000 loss: 0.6453 (0.7577) acc1: 63.0208 (52.4058) acc5: 100.0000 (100.0000) time: 0.2268 data: 0.1332 max mem: 3531 +Test: [ 160/2387] eta: 0:09:19 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.5986 (0.7422) acc1: 66.9271 (53.8545) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.2044 data: 0.1125 max mem: 3531 +Test: [ 170/2387] eta: 0:09:07 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.3183 (0.7167) acc1: 90.6250 (56.0871) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1755 data: 0.0825 max mem: 3531 +Test: [ 180/2387] eta: 0:08:52 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.3008 (0.6936) acc1: 91.9271 (58.0830) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1597 data: 0.0658 max mem: 3531 +Test: [ 190/2387] eta: 0:08:39 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2957 (0.6728) acc1: 91.6667 (59.8836) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1475 data: 0.0530 max mem: 3531 +Test: [ 200/2387] eta: 0:08:32 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2948 (0.6543) acc1: 92.4479 (61.4868) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1699 data: 0.0726 max mem: 3531 +Test: [ 210/2387] eta: 0:08:27 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2992 (0.6378) acc1: 91.9271 (62.9110) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.2003 data: 0.1039 max mem: 3532 +Test: [ 220/2387] eta: 0:08:23 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.3077 (0.6229) acc1: 91.6667 (64.2298) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.2149 data: 0.1175 max mem: 3532 +Test: [ 230/2387] eta: 0:08:20 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.3077 (0.6092) acc1: 91.6667 (65.4119) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.2201 data: 0.1222 max mem: 3532 +Test: [ 240/2387] eta: 0:08:12 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.3187 (0.5984) acc1: 90.6250 (66.4171) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1962 data: 0.1005 max mem: 3532 +Test: [ 250/2387] eta: 0:08:06 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2817 (0.5800) acc1: 90.8854 (67.7000) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1766 data: 0.0822 max mem: 3532 +Test: [ 260/2387] eta: 0:08:01 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1001 (0.5618) acc1: 100.0000 (68.9346) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0943 max mem: 3532 +Test: [ 270/2387] eta: 0:07:57 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1006 (0.5449) acc1: 100.0000 (70.0761) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1994 data: 0.1029 max mem: 3532 +Test: [ 280/2387] eta: 0:07:52 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1034 (0.5294) acc1: 100.0000 (71.1364) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1957 data: 0.1004 max mem: 3532 +Test: [ 290/2387] eta: 0:07:49 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.0910 (0.5144) acc1: 100.0000 (72.1256) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2035 data: 0.1065 max mem: 3532 +Test: [ 300/2387] eta: 0:07:44 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.0864 (0.5004) acc1: 100.0000 (73.0508) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2028 data: 0.1072 max mem: 3532 +Test: [ 310/2387] eta: 0:07:41 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1046 (0.4906) acc1: 100.0000 (73.8754) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1996 data: 0.1077 max mem: 3532 +Test: [ 320/2387] eta: 0:07:37 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.3115 (0.4852) acc1: 98.4375 (74.6333) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2024 data: 0.1110 max mem: 3532 +Test: [ 330/2387] eta: 0:07:35 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.3123 (0.4799) acc1: 98.1771 (75.3470) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2062 data: 0.1145 max mem: 3532 +Test: [ 340/2387] eta: 0:07:32 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.3093 (0.4749) acc1: 98.4375 (76.0195) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2132 data: 0.1216 max mem: 3532 +Test: [ 350/2387] eta: 0:07:29 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.3039 (0.4669) acc1: 98.6979 (76.6960) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2108 data: 0.1157 max mem: 3532 +Test: [ 360/2387] eta: 0:07:26 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2154 (0.4606) acc1: 98.4375 (77.2363) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2055 data: 0.1105 max mem: 3532 +Test: [ 370/2387] eta: 0:07:21 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2455 (0.4547) acc1: 96.0938 (77.7481) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1888 data: 0.0978 max mem: 3532 +Test: [ 380/2387] eta: 0:07:18 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2425 (0.4491) acc1: 96.3542 (78.2282) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1841 data: 0.0929 max mem: 3532 +Test: [ 390/2387] eta: 0:07:13 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2425 (0.4440) acc1: 95.5729 (78.6745) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1785 data: 0.0863 max mem: 3532 +Test: [ 400/2387] eta: 0:07:10 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2458 (0.4390) acc1: 95.5729 (79.0920) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1853 data: 0.0926 max mem: 3532 +Test: [ 410/2387] eta: 0:07:06 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2438 (0.4342) acc1: 96.0938 (79.5082) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1918 data: 0.0996 max mem: 3532 +Test: [ 420/2387] eta: 0:07:01 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2404 (0.4297) acc1: 96.0938 (79.8960) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1708 data: 0.0792 max mem: 3532 +Test: [ 430/2387] eta: 0:07:02 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2495 (0.4260) acc1: 95.3125 (80.2337) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.2180 data: 0.1256 max mem: 3532 +Test: [ 440/2387] eta: 0:06:57 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2562 (0.4203) acc1: 95.0521 (80.6347) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2149 data: 0.1194 max mem: 3533 +Test: [ 450/2387] eta: 0:06:54 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.0998 (0.4132) acc1: 100.0000 (81.0641) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1698 data: 0.0725 max mem: 3533 +Test: [ 460/2387] eta: 0:06:50 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.0994 (0.4064) acc1: 100.0000 (81.4748) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1787 data: 0.0818 max mem: 3533 +Test: [ 470/2387] eta: 0:06:47 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.0995 (0.3999) acc1: 100.0000 (81.8681) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1824 data: 0.0870 max mem: 3533 +Test: [ 480/2387] eta: 0:06:43 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1006 (0.3937) acc1: 100.0000 (82.2451) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1854 data: 0.0931 max mem: 3533 +Test: [ 490/2387] eta: 0:06:38 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1004 (0.3877) acc1: 100.0000 (82.6067) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1554 data: 0.0642 max mem: 3533 +Test: [ 500/2387] eta: 0:06:34 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1002 (0.3820) acc1: 100.0000 (82.9539) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1487 data: 0.0540 max mem: 3533 +Test: [ 510/2387] eta: 0:06:31 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1002 (0.3767) acc1: 100.0000 (83.2875) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1740 data: 0.0791 max mem: 3533 +Test: [ 520/2387] eta: 0:06:28 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1382 (0.3722) acc1: 100.0000 (83.6077) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1765 data: 0.0848 max mem: 3533 +Test: [ 530/2387] eta: 0:06:24 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1366 (0.3677) acc1: 100.0000 (83.9165) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1598 data: 0.0691 max mem: 3533 +Test: [ 540/2387] eta: 0:06:20 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1342 (0.3634) acc1: 100.0000 (84.2133) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1589 data: 0.0674 max mem: 3533 +Test: [ 550/2387] eta: 0:06:17 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1350 (0.3592) acc1: 100.0000 (84.4993) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1684 data: 0.0767 max mem: 3533 +Test: [ 560/2387] eta: 0:06:15 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1351 (0.3552) acc1: 100.0000 (84.7742) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1819 data: 0.0907 max mem: 3533 +Test: [ 570/2387] eta: 0:06:11 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1344 (0.3514) acc1: 100.0000 (85.0404) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1756 data: 0.0836 max mem: 3533 +Test: [ 580/2387] eta: 0:06:08 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1343 (0.3476) acc1: 100.0000 (85.2979) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1673 data: 0.0747 max mem: 3533 +Test: [ 590/2387] eta: 0:06:05 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1353 (0.3440) acc1: 100.0000 (85.5458) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1736 data: 0.0819 max mem: 3533 +Test: [ 600/2387] eta: 0:06:02 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1396 (0.3408) acc1: 100.0000 (85.7819) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1712 data: 0.0803 max mem: 3533 +Test: [ 610/2387] eta: 0:06:00 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1855 (0.3398) acc1: 97.1354 (85.9388) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1799 data: 0.0829 max mem: 3533 +Test: [ 620/2387] eta: 0:05:58 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2778 (0.3388) acc1: 95.5729 (86.0847) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1935 data: 0.0934 max mem: 3533 +Test: [ 630/2387] eta: 0:05:55 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2820 (0.3379) acc1: 95.3125 (86.2272) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1949 data: 0.0970 max mem: 3533 +Test: [ 640/2387] eta: 0:05:53 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2820 (0.3371) acc1: 95.3125 (86.3629) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1918 data: 0.0950 max mem: 3533 +Test: [ 650/2387] eta: 0:05:50 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2842 (0.3363) acc1: 95.3125 (86.4999) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1849 data: 0.0896 max mem: 3533 +Test: [ 660/2387] eta: 0:05:48 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2842 (0.3354) acc1: 95.3125 (86.6352) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1930 data: 0.0962 max mem: 3533 +Test: [ 670/2387] eta: 0:05:45 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2827 (0.3346) acc1: 95.0521 (86.7607) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1774 data: 0.0812 max mem: 3534 +Test: [ 680/2387] eta: 0:05:43 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2870 (0.3343) acc1: 95.5729 (86.8966) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1749 data: 0.0774 max mem: 3534 +Test: [ 690/2387] eta: 0:05:40 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.3174 (0.3341) acc1: 94.7917 (86.9343) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.1924 data: 0.0941 max mem: 3534 +Test: [ 700/2387] eta: 0:05:39 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.3174 (0.3339) acc1: 87.5000 (86.9416) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.1970 data: 0.1036 max mem: 3534 +Test: [ 710/2387] eta: 0:05:37 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.3125 (0.3336) acc1: 87.5000 (86.9477) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.2046 data: 0.1137 max mem: 3534 +Test: [ 720/2387] eta: 0:05:35 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.3102 (0.3333) acc1: 87.5000 (86.9589) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.2065 data: 0.1147 max mem: 3534 +Test: [ 730/2387] eta: 0:05:33 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.3180 (0.3331) acc1: 87.2396 (86.9589) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.2049 data: 0.1118 max mem: 3534 +Test: [ 740/2387] eta: 0:05:31 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.3117 (0.3328) acc1: 86.9792 (86.9704) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1925 data: 0.1003 max mem: 3534 +Test: [ 750/2387] eta: 0:05:28 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.3055 (0.3324) acc1: 88.0208 (86.9795) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1952 data: 0.1041 max mem: 3534 +Test: [ 760/2387] eta: 0:05:26 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2969 (0.3319) acc1: 88.0208 (87.0107) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1919 data: 0.0987 max mem: 3534 +Test: [ 770/2387] eta: 0:05:24 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2657 (0.3309) acc1: 90.3646 (87.0846) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1906 data: 0.0975 max mem: 3534 +Test: [ 780/2387] eta: 0:05:22 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2465 (0.3297) acc1: 93.4896 (87.1836) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1918 data: 0.1000 max mem: 3534 +Test: [ 790/2387] eta: 0:05:19 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2054 (0.3282) acc1: 95.8333 (87.2784) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1841 data: 0.0919 max mem: 3534 +Test: [ 800/2387] eta: 0:05:17 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2042 (0.3268) acc1: 96.0938 (87.3875) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1852 data: 0.0921 max mem: 3534 +Test: [ 810/2387] eta: 0:05:15 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2332 (0.3258) acc1: 95.3125 (87.4650) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1954 data: 0.1018 max mem: 3534 +Test: [ 820/2387] eta: 0:05:13 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2403 (0.3249) acc1: 93.7500 (87.5447) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1980 data: 0.1047 max mem: 3534 +Test: [ 830/2387] eta: 0:05:11 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2851 (0.3253) acc1: 90.8854 (87.5248) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.2058 data: 0.1126 max mem: 3534 +Test: [ 840/2387] eta: 0:05:09 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.3808 (0.3266) acc1: 85.1562 (87.4656) acc5: 100.0000 (100.0000) acc1_neg: 76.302 acc5_neg: 100.000 time: 0.1913 data: 0.0992 max mem: 3534 +Test: [ 850/2387] eta: 0:05:07 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.4316 (0.3281) acc1: 83.0729 (87.3825) acc5: 100.0000 (100.0000) acc1_neg: 71.354 acc5_neg: 100.000 time: 0.1845 data: 0.0922 max mem: 3534 +Test: [ 860/2387] eta: 0:05:05 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.4087 (0.3290) acc1: 85.1562 (87.3485) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.2052 data: 0.1104 max mem: 3534 +Test: [ 870/2387] eta: 0:05:03 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.3851 (0.3298) acc1: 84.3750 (87.3045) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.2037 data: 0.1089 max mem: 3534 +Test: [ 880/2387] eta: 0:05:01 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.3729 (0.3308) acc1: 84.3750 (87.2615) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.1944 data: 0.1001 max mem: 3534 +Test: [ 890/2387] eta: 0:04:59 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.3572 (0.3310) acc1: 87.2396 (87.2922) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1903 data: 0.0976 max mem: 3534 +Test: [ 900/2387] eta: 0:04:57 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.3942 (0.3355) acc1: 84.3750 (86.9841) acc5: 100.0000 (100.0000) acc1_neg: 55.729 acc5_neg: 100.000 time: 0.1944 data: 0.1039 max mem: 3535 +Test: [ 910/2387] eta: 0:04:55 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.8163 (0.3409) acc1: 51.5625 (86.5881) acc5: 100.0000 (100.0000) acc1_neg: 50.781 acc5_neg: 100.000 time: 0.2041 data: 0.1130 max mem: 3535 +Test: [ 920/2387] eta: 0:04:53 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.8290 (0.3463) acc1: 50.7812 (86.1965) acc5: 100.0000 (100.0000) acc1_neg: 48.438 acc5_neg: 100.000 time: 0.2258 data: 0.1323 max mem: 3535 +Test: [ 930/2387] eta: 0:04:51 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.8331 (0.3515) acc1: 51.0417 (85.8217) acc5: 100.0000 (100.0000) acc1_neg: 47.135 acc5_neg: 100.000 time: 0.2089 data: 0.1145 max mem: 3535 +Test: [ 940/2387] eta: 0:04:49 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.8359 (0.3566) acc1: 51.0417 (85.4573) acc5: 100.0000 (100.0000) acc1_neg: 49.740 acc5_neg: 100.000 time: 0.1963 data: 0.1042 max mem: 3535 +Test: [ 950/2387] eta: 0:04:47 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.8259 (0.3614) acc1: 51.0417 (85.0968) acc5: 100.0000 (100.0000) acc1_neg: 50.521 acc5_neg: 100.000 time: 0.2155 data: 0.1244 max mem: 3535 +Test: [ 960/2387] eta: 0:04:45 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.8062 (0.3661) acc1: 51.8229 (84.7511) acc5: 100.0000 (100.0000) acc1_neg: 52.344 acc5_neg: 100.000 time: 0.2039 data: 0.1096 max mem: 3535 +Test: [ 970/2387] eta: 0:04:43 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.8110 (0.3709) acc1: 51.0417 (84.3946) acc5: 100.0000 (100.0000) acc1_neg: 53.125 acc5_neg: 100.000 time: 0.1973 data: 0.1028 max mem: 3535 +Test: [ 980/2387] eta: 0:04:41 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.7892 (0.3722) acc1: 51.0417 (84.3190) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.1935 data: 0.0986 max mem: 3535 +Test: [ 990/2387] eta: 0:04:39 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2810 (0.3709) acc1: 90.1042 (84.4123) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1881 data: 0.0933 max mem: 3535 +Test: [1000/2387] eta: 0:04:37 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2215 (0.3693) acc1: 96.8750 (84.5215) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1829 data: 0.0911 max mem: 3535 +Test: [1010/2387] eta: 0:04:34 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2376 (0.3684) acc1: 94.7917 (84.5947) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1723 data: 0.0798 max mem: 3535 +Test: [1020/2387] eta: 0:04:32 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2533 (0.3676) acc1: 93.7500 (84.6533) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1681 data: 0.0755 max mem: 3535 +Test: [1030/2387] eta: 0:04:29 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2278 (0.3661) acc1: 96.3542 (84.7751) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1673 data: 0.0753 max mem: 3535 +Test: [1040/2387] eta: 0:04:27 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1744 (0.3642) acc1: 98.9583 (84.9098) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1732 data: 0.0813 max mem: 3535 +Test: [1050/2387] eta: 0:04:25 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1740 (0.3625) acc1: 98.9583 (85.0398) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1748 data: 0.0821 max mem: 3535 +Test: [1060/2387] eta: 0:04:23 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1830 (0.3610) acc1: 98.4375 (85.1484) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1712 data: 0.0780 max mem: 3535 +Test: [1070/2387] eta: 0:04:20 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2063 (0.3598) acc1: 96.6146 (85.2418) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1705 data: 0.0779 max mem: 3535 +Test: [1080/2387] eta: 0:04:18 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2616 (0.3596) acc1: 91.4062 (85.2733) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1660 data: 0.0735 max mem: 3535 +Test: [1090/2387] eta: 0:04:15 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.3068 (0.3588) acc1: 89.8438 (85.3377) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1646 data: 0.0727 max mem: 3535 +Test: [1100/2387] eta: 0:04:13 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2540 (0.3578) acc1: 94.0104 (85.4207) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1622 data: 0.0696 max mem: 3535 +Test: [1110/2387] eta: 0:04:11 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2540 (0.3572) acc1: 94.0104 (85.4680) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1567 data: 0.0638 max mem: 3535 +Test: [1120/2387] eta: 0:04:08 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2801 (0.3567) acc1: 91.9271 (85.5005) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1599 data: 0.0649 max mem: 3536 +Test: [1130/2387] eta: 0:04:06 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2560 (0.3558) acc1: 93.4896 (85.5684) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1640 data: 0.0680 max mem: 3536 +Test: [1140/2387] eta: 0:04:04 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2899 (0.3554) acc1: 92.4479 (85.6173) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1702 data: 0.0776 max mem: 3536 +Test: [1150/2387] eta: 0:04:01 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.3041 (0.3552) acc1: 89.5833 (85.6477) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.1728 data: 0.0801 max mem: 3536 +Test: [1160/2387] eta: 0:03:59 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2808 (0.3546) acc1: 90.8854 (85.7029) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.1808 data: 0.0881 max mem: 3536 +Test: [1170/2387] eta: 0:03:57 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2764 (0.3540) acc1: 92.7083 (85.7680) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1917 data: 0.1004 max mem: 3536 +Test: [1180/2387] eta: 0:03:55 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.3008 (0.3536) acc1: 91.9271 (85.8169) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1896 data: 0.0978 max mem: 3536 +Test: [1190/2387] eta: 0:03:54 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2481 (0.3528) acc1: 94.0104 (85.8811) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1931 data: 0.0982 max mem: 3536 +Test: [1200/2387] eta: 0:03:51 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2287 (0.3518) acc1: 95.5729 (85.9620) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1925 data: 0.0963 max mem: 3536 +Test: [1210/2387] eta: 0:03:49 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2216 (0.3507) acc1: 96.8750 (86.0506) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1810 data: 0.0869 max mem: 3536 +Test: [1220/2387] eta: 0:03:47 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1498 (0.3488) acc1: 98.6979 (86.1568) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1725 data: 0.0792 max mem: 3536 +Test: [1230/2387] eta: 0:03:45 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1115 (0.3469) acc1: 99.4792 (86.2637) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1609 data: 0.0685 max mem: 3536 +Test: [1240/2387] eta: 0:03:42 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1115 (0.3451) acc1: 99.4792 (86.3698) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1543 data: 0.0600 max mem: 3536 +Test: [1250/2387] eta: 0:03:40 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1079 (0.3432) acc1: 99.7396 (86.4744) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1678 data: 0.0723 max mem: 3536 +Test: [1260/2387] eta: 0:03:38 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.0959 (0.3413) acc1: 99.7396 (86.5779) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1793 data: 0.0826 max mem: 3536 +Test: [1270/2387] eta: 0:03:36 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.0943 (0.3394) acc1: 99.7396 (86.6823) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1675 data: 0.0719 max mem: 3536 +Test: [1280/2387] eta: 0:03:34 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1173 (0.3378) acc1: 99.7396 (86.7781) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1706 data: 0.0789 max mem: 3536 +Test: [1290/2387] eta: 0:03:32 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1501 (0.3364) acc1: 98.6979 (86.8604) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1833 data: 0.0899 max mem: 3536 +Test: [1300/2387] eta: 0:03:30 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1534 (0.3350) acc1: 98.6979 (86.9465) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1768 data: 0.0812 max mem: 3536 +Test: [1310/2387] eta: 0:03:28 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1577 (0.3338) acc1: 98.4375 (87.0288) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1766 data: 0.0830 max mem: 3536 +Test: [1320/2387] eta: 0:03:26 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1716 (0.3326) acc1: 98.4375 (87.1089) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1753 data: 0.0825 max mem: 3536 +Test: [1330/2387] eta: 0:03:24 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1889 (0.3319) acc1: 96.3542 (87.1662) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1670 data: 0.0730 max mem: 3536 +Test: [1340/2387] eta: 0:03:21 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2708 (0.3319) acc1: 92.7083 (87.1635) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1632 data: 0.0713 max mem: 3536 +Test: [1350/2387] eta: 0:03:19 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.0786 (0.3300) acc1: 100.0000 (87.2585) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1759 data: 0.0834 max mem: 3537 +Test: [1360/2387] eta: 0:03:17 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.0641 (0.3280) acc1: 100.0000 (87.3521) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1739 data: 0.0804 max mem: 3537 +Test: [1370/2387] eta: 0:03:15 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.0651 (0.3261) acc1: 100.0000 (87.4443) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1646 data: 0.0721 max mem: 3537 +Test: [1380/2387] eta: 0:03:13 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.0651 (0.3242) acc1: 100.0000 (87.5353) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1654 data: 0.0737 max mem: 3537 +Test: [1390/2387] eta: 0:03:11 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.0599 (0.3224) acc1: 100.0000 (87.6249) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1613 data: 0.0694 max mem: 3537 +Test: [1400/2387] eta: 0:03:09 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.0601 (0.3205) acc1: 100.0000 (87.7132) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1743 data: 0.0827 max mem: 3537 +Test: [1410/2387] eta: 0:03:07 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.0705 (0.3198) acc1: 100.0000 (87.7551) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1879 data: 0.0957 max mem: 3537 +Test: [1420/2387] eta: 0:03:05 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2618 (0.3195) acc1: 92.9688 (87.7826) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1745 data: 0.0770 max mem: 3537 +Test: [1430/2387] eta: 0:03:03 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2745 (0.3193) acc1: 90.1042 (87.7981) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1512 data: 0.0523 max mem: 3537 +Test: [1440/2387] eta: 0:03:01 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2694 (0.3190) acc1: 90.1042 (87.8320) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.1749 data: 0.0806 max mem: 3537 +Test: [1450/2387] eta: 0:02:59 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2827 (0.3190) acc1: 90.1042 (87.8383) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1847 data: 0.0901 max mem: 3537 +Test: [1460/2387] eta: 0:02:56 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2580 (0.3185) acc1: 93.7500 (87.8891) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1559 data: 0.0592 max mem: 3537 +Test: [1470/2387] eta: 0:02:54 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1925 (0.3176) acc1: 97.3958 (87.9587) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1614 data: 0.0671 max mem: 3537 +Test: [1480/2387] eta: 0:02:52 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1901 (0.3168) acc1: 97.6562 (88.0166) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1715 data: 0.0801 max mem: 3537 +Test: [1490/2387] eta: 0:02:51 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2032 (0.3160) acc1: 96.8750 (88.0788) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1838 data: 0.0888 max mem: 3537 +Test: [1500/2387] eta: 0:02:49 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1985 (0.3153) acc1: 97.3958 (88.1412) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2088 data: 0.1100 max mem: 3537 +Test: [1510/2387] eta: 0:02:47 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2314 (0.3148) acc1: 96.8750 (88.1909) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.2247 data: 0.1292 max mem: 3537 +Test: [1520/2387] eta: 0:02:45 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2640 (0.3152) acc1: 93.4896 (88.1830) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.2051 data: 0.1108 max mem: 3537 +Test: [1530/2387] eta: 0:02:44 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.3734 (0.3158) acc1: 85.4167 (88.1556) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.2087 data: 0.1145 max mem: 3537 +Test: [1540/2387] eta: 0:02:41 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.3733 (0.3162) acc1: 85.9375 (88.1461) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1944 data: 0.1012 max mem: 3537 +Test: [1550/2387] eta: 0:02:40 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.3713 (0.3165) acc1: 86.9792 (88.1451) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1761 data: 0.0814 max mem: 3537 +Test: [1560/2387] eta: 0:02:38 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.3350 (0.3168) acc1: 86.9792 (88.1406) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.2012 data: 0.1078 max mem: 3537 +Test: [1570/2387] eta: 0:02:36 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.3139 (0.3168) acc1: 91.4062 (88.1657) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1995 data: 0.1070 max mem: 3537 +Test: [1580/2387] eta: 0:02:34 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2949 (0.3166) acc1: 91.9271 (88.1895) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1945 data: 0.1023 max mem: 3538 +Test: [1590/2387] eta: 0:02:32 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2372 (0.3159) acc1: 94.2708 (88.2403) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1988 data: 0.1071 max mem: 3538 +Test: [1600/2387] eta: 0:02:30 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1958 (0.3153) acc1: 97.1354 (88.2734) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1999 data: 0.1081 max mem: 3538 +Test: [1610/2387] eta: 0:02:28 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2142 (0.3148) acc1: 96.0938 (88.3168) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1953 data: 0.1044 max mem: 3538 +Test: [1620/2387] eta: 0:02:26 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2189 (0.3144) acc1: 93.2292 (88.3413) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1945 data: 0.1031 max mem: 3538 +Test: [1630/2387] eta: 0:02:24 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2595 (0.3142) acc1: 92.7083 (88.3598) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1863 data: 0.0947 max mem: 3538 +Test: [1640/2387] eta: 0:02:22 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.3164 (0.3144) acc1: 89.3229 (88.3587) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1901 data: 0.0991 max mem: 3538 +Test: [1650/2387] eta: 0:02:21 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.3390 (0.3144) acc1: 87.7604 (88.3639) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1982 data: 0.1073 max mem: 3538 +Test: [1660/2387] eta: 0:02:19 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.3220 (0.3145) acc1: 88.2812 (88.3543) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.1993 data: 0.1064 max mem: 3538 +Test: [1670/2387] eta: 0:02:17 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.3220 (0.3145) acc1: 88.8021 (88.3640) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1964 data: 0.1009 max mem: 3538 +Test: [1680/2387] eta: 0:02:15 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.3044 (0.3145) acc1: 89.8438 (88.3708) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.1918 data: 0.0955 max mem: 3538 +Test: [1690/2387] eta: 0:02:13 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.3254 (0.3147) acc1: 88.8021 (88.3686) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1940 data: 0.0986 max mem: 3538 +Test: [1700/2387] eta: 0:02:11 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.3319 (0.3145) acc1: 88.5417 (88.3864) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1921 data: 0.0988 max mem: 3538 +Test: [1710/2387] eta: 0:02:09 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1794 (0.3135) acc1: 97.3958 (88.4443) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1765 data: 0.0845 max mem: 3538 +Test: [1720/2387] eta: 0:02:07 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1338 (0.3124) acc1: 98.9583 (88.5050) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1556 data: 0.0645 max mem: 3538 +Test: [1730/2387] eta: 0:02:05 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1339 (0.3115) acc1: 98.6979 (88.5623) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1596 data: 0.0689 max mem: 3538 +Test: [1740/2387] eta: 0:02:03 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1331 (0.3104) acc1: 98.6979 (88.6221) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1735 data: 0.0822 max mem: 3538 +Test: [1750/2387] eta: 0:02:01 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1152 (0.3094) acc1: 99.7396 (88.6828) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1876 data: 0.0934 max mem: 3538 +Test: [1760/2387] eta: 0:01:59 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1196 (0.3083) acc1: 99.2188 (88.7400) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1749 data: 0.0790 max mem: 3538 +Test: [1770/2387] eta: 0:01:57 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1445 (0.3077) acc1: 97.6562 (88.7790) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1588 data: 0.0651 max mem: 3538 +Test: [1780/2387] eta: 0:01:55 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1826 (0.3069) acc1: 96.8750 (88.8281) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1838 data: 0.0898 max mem: 3538 +Test: [1790/2387] eta: 0:01:53 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1698 (0.3062) acc1: 97.6562 (88.8755) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1775 data: 0.0837 max mem: 3538 +Test: [1800/2387] eta: 0:01:51 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1649 (0.3054) acc1: 97.1354 (88.9218) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1754 data: 0.0835 max mem: 3538 +Test: [1810/2387] eta: 0:01:49 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1546 (0.3045) acc1: 97.9167 (88.9742) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1727 data: 0.0807 max mem: 3539 +Test: [1820/2387] eta: 0:01:47 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1413 (0.3036) acc1: 98.9583 (89.0266) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1651 data: 0.0724 max mem: 3539 +Test: [1830/2387] eta: 0:01:45 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1376 (0.3027) acc1: 98.9583 (89.0801) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1726 data: 0.0787 max mem: 3539 +Test: [1840/2387] eta: 0:01:43 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1334 (0.3018) acc1: 100.0000 (89.1395) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1648 data: 0.0720 max mem: 3539 +Test: [1850/2387] eta: 0:01:41 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1328 (0.3009) acc1: 100.0000 (89.1978) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1677 data: 0.0769 max mem: 3539 +Test: [1860/2387] eta: 0:01:39 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1314 (0.3000) acc1: 100.0000 (89.2559) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1659 data: 0.0745 max mem: 3539 +Test: [1870/2387] eta: 0:01:37 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1312 (0.2991) acc1: 100.0000 (89.3133) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1608 data: 0.0680 max mem: 3539 +Test: [1880/2387] eta: 0:01:36 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1314 (0.2982) acc1: 100.0000 (89.3700) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1661 data: 0.0715 max mem: 3539 +Test: [1890/2387] eta: 0:01:34 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1322 (0.2973) acc1: 100.0000 (89.4261) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1787 data: 0.0853 max mem: 3539 +Test: [1900/2387] eta: 0:01:32 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1345 (0.2965) acc1: 100.0000 (89.4813) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1930 data: 0.1016 max mem: 3539 +Test: [1910/2387] eta: 0:01:30 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1660 (0.2959) acc1: 100.0000 (89.5320) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1847 data: 0.0934 max mem: 3539 +Test: [1920/2387] eta: 0:01:28 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1843 (0.2959) acc1: 98.1771 (89.5310) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1595 data: 0.0672 max mem: 3539 +Test: [1930/2387] eta: 0:01:26 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2035 (0.2954) acc1: 95.5729 (89.5736) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1573 data: 0.0627 max mem: 3539 +Test: [1940/2387] eta: 0:01:24 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2035 (0.2952) acc1: 98.1771 (89.5855) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1801 data: 0.0863 max mem: 3539 +Test: [1950/2387] eta: 0:01:22 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2654 (0.2952) acc1: 94.5312 (89.6011) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1921 data: 0.0998 max mem: 3539 +Test: [1960/2387] eta: 0:01:20 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2658 (0.2950) acc1: 93.4896 (89.6164) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1791 data: 0.0856 max mem: 3539 +Test: [1970/2387] eta: 0:01:18 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.3104 (0.2957) acc1: 89.0625 (89.5580) acc5: 100.0000 (100.0000) acc1_neg: 72.135 acc5_neg: 100.000 time: 0.1773 data: 0.0832 max mem: 3539 +Test: [1980/2387] eta: 0:01:16 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.3405 (0.2956) acc1: 82.8125 (89.5653) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1976 data: 0.1021 max mem: 3539 +Test: [1990/2387] eta: 0:01:15 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1488 (0.2949) acc1: 98.9583 (89.6128) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1989 data: 0.1036 max mem: 3539 +Test: [2000/2387] eta: 0:01:13 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1479 (0.2941) acc1: 99.2188 (89.6612) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2121 data: 0.1196 max mem: 3539 +Test: [2010/2387] eta: 0:01:11 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1472 (0.2934) acc1: 99.2188 (89.7086) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2349 data: 0.1393 max mem: 3539 +Test: [2020/2387] eta: 0:01:09 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1480 (0.2927) acc1: 99.2188 (89.7559) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2173 data: 0.1214 max mem: 3539 +Test: [2030/2387] eta: 0:01:07 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1475 (0.2920) acc1: 99.2188 (89.8031) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2134 data: 0.1210 max mem: 3540 +Test: [2040/2387] eta: 0:01:05 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1475 (0.2913) acc1: 99.2188 (89.8492) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2303 data: 0.1385 max mem: 3540 +Test: [2050/2387] eta: 0:01:03 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1447 (0.2906) acc1: 99.4792 (89.8966) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0875 max mem: 3540 +Test: [2060/2387] eta: 0:01:01 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1456 (0.2899) acc1: 99.4792 (89.9434) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1374 data: 0.0450 max mem: 3540 +Test: [2070/2387] eta: 0:01:00 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1439 (0.2892) acc1: 99.7396 (89.9907) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1672 data: 0.0729 max mem: 3540 +Test: [2080/2387] eta: 0:00:58 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1438 (0.2885) acc1: 99.7396 (90.0373) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1968 data: 0.1025 max mem: 3540 +Test: [2090/2387] eta: 0:00:56 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1424 (0.2878) acc1: 99.7396 (90.0837) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1946 data: 0.1007 max mem: 3540 +Test: [2100/2387] eta: 0:00:54 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1433 (0.2871) acc1: 99.7396 (90.1296) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1987 data: 0.1043 max mem: 3540 +Test: [2110/2387] eta: 0:00:52 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1453 (0.2864) acc1: 99.4792 (90.1745) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2080 data: 0.1138 max mem: 3540 +Test: [2120/2387] eta: 0:00:50 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1521 (0.2863) acc1: 99.2188 (90.1915) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2007 data: 0.1052 max mem: 3540 +Test: [2130/2387] eta: 0:00:48 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2106 (0.2858) acc1: 97.3958 (90.2266) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1998 data: 0.1049 max mem: 3540 +Test: [2140/2387] eta: 0:00:46 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1842 (0.2856) acc1: 98.6979 (90.2454) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1934 data: 0.0999 max mem: 3540 +Test: [2150/2387] eta: 0:00:44 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2148 (0.2853) acc1: 96.3542 (90.2691) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.2012 data: 0.1077 max mem: 3540 +Test: [2160/2387] eta: 0:00:43 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2169 (0.2851) acc1: 94.5312 (90.2889) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2058 data: 0.1127 max mem: 3540 +Test: [2170/2387] eta: 0:00:41 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2694 (0.2852) acc1: 91.6667 (90.2851) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.2035 data: 0.1112 max mem: 3540 +Test: [2180/2387] eta: 0:00:39 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.3458 (0.2865) acc1: 86.7188 (90.2021) acc5: 100.0000 (100.0000) acc1_neg: 51.302 acc5_neg: 100.000 time: 0.2047 data: 0.1125 max mem: 3540 +Test: [2190/2387] eta: 0:00:37 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.3344 (0.2865) acc1: 86.7188 (90.2039) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1912 data: 0.0985 max mem: 3540 +Test: [2200/2387] eta: 0:00:35 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2062 (0.2863) acc1: 96.8750 (90.2298) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1758 data: 0.0834 max mem: 3540 +Test: [2210/2387] eta: 0:00:33 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2082 (0.2860) acc1: 96.6146 (90.2560) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1669 data: 0.0753 max mem: 3540 +Test: [2220/2387] eta: 0:00:31 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2068 (0.2857) acc1: 96.6146 (90.2837) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1741 data: 0.0817 max mem: 3540 +Test: [2230/2387] eta: 0:00:29 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2026 (0.2854) acc1: 96.8750 (90.3125) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1786 data: 0.0853 max mem: 3540 +Test: [2240/2387] eta: 0:00:27 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1785 (0.2850) acc1: 98.1771 (90.3389) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1767 data: 0.0840 max mem: 3540 +Test: [2250/2387] eta: 0:00:26 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2408 (0.2850) acc1: 96.6146 (90.3542) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.2138 data: 0.1221 max mem: 3540 +Test: [2260/2387] eta: 0:00:24 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2408 (0.2848) acc1: 96.0938 (90.3800) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2302 data: 0.1376 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1947 (0.2843) acc1: 98.1771 (90.4171) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2203 data: 0.1281 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1688 (0.2839) acc1: 98.9583 (90.4398) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1831 data: 0.0890 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.1553 (0.2834) acc1: 98.9583 (90.4750) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1704 data: 0.0733 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2017 (0.2831) acc1: 97.3958 (90.4992) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1911 data: 0.0965 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2177 (0.2829) acc1: 95.5729 (90.5150) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1967 data: 0.1034 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2792 (0.2831) acc1: 91.6667 (90.5076) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1818 data: 0.0892 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.3067 (0.2834) acc1: 88.8021 (90.4971) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.1663 data: 0.0751 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2830 (0.2832) acc1: 91.9271 (90.5144) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1693 data: 0.0785 max mem: 3541 +Test: [2350/2387] eta: 0:00:07 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.2449 (0.2834) acc1: 95.0521 (90.5119) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1677 data: 0.0774 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.3205 (0.2837) acc1: 89.5833 (90.4956) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1610 data: 0.0707 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.3396 (0.2841) acc1: 86.7188 (90.4687) acc5: 100.0000 (100.0000) acc1_neg: 69.792 acc5_neg: 100.000 time: 0.1491 data: 0.0589 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.3876 (0.2846) acc1: 84.6354 (90.4335) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.1535 data: 0.0627 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 65.297 acc5_pos: 100.000 loss: 0.3997 (0.2847) acc1: 83.5938 (90.4294) acc5: 100.0000 (100.0000) acc1_neg: 95.349 acc5_neg: 100.000 time: 0.1422 data: 0.0542 max mem: 3541 +Test: Total time: 0:07:31 (0.1892 s / it) +* Overall Acc@1 90.429 Acc@5 100.000 + target == real Acc@1 52.905 Acc@5 100.000 + target == fake Acc@1 93.065 Acc@5 100.000 + loss 0.285 + PR-AUC 0.9921 ROC-AUC 0.8993 EER 17.33% (thr=0.6675) +Accuracy of the network on the 916267 test images: 90.4% +[rank0]:[W725 18:27:01.233425818 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:27:20.464000 604868 torch/distributed/run.py:766] +W0725 18:27:20.464000 604868 torch/distributed/run.py:766] ***************************************** +W0725 18:27:20.464000 604868 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:27:20.464000 604868 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:27:49.071974754 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:27:50.407875797 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:27:50.589323904 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:27:50.909720678 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:27:50.913153166 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:27:50.035620973 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:27:50.070053948 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:27:50.136878181 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed3_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n32_classseed3_resampled_n116-040-213_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 0:48:07 acc1_pos: 64.062 acc5_pos: 100.000 loss: 0.6097 (0.6097) acc1: 64.0625 (64.0625) acc5: 100.0000 (100.0000) time: 5.5850 data: 3.3255 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/517] eta: 0:04:59 acc1_pos: 28.385 acc5_pos: 100.000 loss: 0.7532 (0.7666) acc1: 53.3854 (52.4621) acc5: 100.0000 (100.0000) time: 0.5899 data: 0.3028 max mem: 3531 +Test: [ 20/517] eta: 0:02:55 acc1_pos: 38.802 acc5_pos: 100.000 loss: 0.7905 (0.7987) acc1: 52.6042 (48.8715) acc5: 100.0000 (100.0000) time: 0.0916 data: 0.0017 max mem: 3531 +Test: [ 30/517] eta: 0:02:29 acc1_pos: 36.198 acc5_pos: 100.000 loss: 0.7735 (0.7944) acc1: 47.6562 (49.8068) acc5: 100.0000 (100.0000) time: 0.1511 data: 0.0601 max mem: 3531 +Test: [ 40/517] eta: 0:02:19 acc1_pos: 47.656 acc5_pos: 100.000 loss: 0.7618 (0.7923) acc1: 51.5625 (49.6888) acc5: 100.0000 (100.0000) time: 0.2307 data: 0.1351 max mem: 3531 +Test: [ 50/517] eta: 0:02:17 acc1_pos: 84.404 acc5_pos: 100.000 loss: 0.7562 (0.7834) acc1: 55.4688 (50.4289) acc5: 100.0000 (100.0000) acc1_neg: 59.115 acc5_neg: 100.000 time: 0.2727 data: 0.1777 max mem: 3531 +Test: [ 60/517] eta: 0:02:09 acc1_pos: 84.404 acc5_pos: 100.000 loss: 0.6627 (0.7582) acc1: 59.1146 (52.8987) acc5: 100.0000 (100.0000) acc1_neg: 74.219 acc5_neg: 100.000 time: 0.2654 data: 0.1745 max mem: 3531 +Test: [ 70/517] eta: 0:02:07 acc1_pos: 84.404 acc5_pos: 100.000 loss: 0.5644 (0.7273) acc1: 70.3125 (55.7879) acc5: 100.0000 (100.0000) acc1_neg: 71.354 acc5_neg: 100.000 time: 0.2626 data: 0.1722 max mem: 3531 +Test: [ 80/517] eta: 0:02:06 acc1_pos: 84.404 acc5_pos: 100.000 loss: 0.5644 (0.7096) acc1: 69.2708 (57.1663) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.3047 data: 0.2128 max mem: 3531 +Test: [ 90/517] eta: 0:02:01 acc1_pos: 84.404 acc5_pos: 100.000 loss: 0.5880 (0.6924) acc1: 62.7604 (58.6252) acc5: 100.0000 (100.0000) acc1_neg: 58.854 acc5_neg: 100.000 time: 0.2884 data: 0.1927 max mem: 3531 +Test: [100/517] eta: 0:01:56 acc1_pos: 84.404 acc5_pos: 100.000 loss: 0.5044 (0.6724) acc1: 78.9062 (60.8034) acc5: 100.0000 (100.0000) acc1_neg: 75.781 acc5_neg: 100.000 time: 0.2431 data: 0.1467 max mem: 3531 +Test: [110/517] eta: 0:01:53 acc1_pos: 84.404 acc5_pos: 100.000 loss: 0.4483 (0.6494) acc1: 84.6354 (63.1616) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.2472 data: 0.1528 max mem: 3531 +Test: [120/517] eta: 0:01:50 acc1_pos: 84.404 acc5_pos: 100.000 loss: 0.4670 (0.6374) acc1: 80.7292 (64.2584) acc5: 100.0000 (100.0000) acc1_neg: 75.521 acc5_neg: 100.000 time: 0.2668 data: 0.1732 max mem: 3531 +Test: [130/517] eta: 0:01:46 acc1_pos: 84.404 acc5_pos: 100.000 loss: 0.5012 (0.6252) acc1: 78.1250 (65.5514) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2586 data: 0.1667 max mem: 3531 +Test: [140/517] eta: 0:01:43 acc1_pos: 84.404 acc5_pos: 100.000 loss: 0.3944 (0.6082) acc1: 90.6250 (67.5236) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2501 data: 0.1575 max mem: 3531 +Test: [150/517] eta: 0:01:39 acc1_pos: 84.404 acc5_pos: 100.000 loss: 0.3706 (0.5921) acc1: 93.4896 (69.3088) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2513 data: 0.1574 max mem: 3531 +Test: [160/517] eta: 0:01:36 acc1_pos: 84.404 acc5_pos: 100.000 loss: 0.3833 (0.5802) acc1: 91.9271 (70.5842) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2547 data: 0.1632 max mem: 3531 +Test: [170/517] eta: 0:01:33 acc1_pos: 84.404 acc5_pos: 100.000 loss: 0.3926 (0.5682) acc1: 90.3646 (71.8400) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2451 data: 0.1540 max mem: 3531 +Test: [180/517] eta: 0:01:30 acc1_pos: 84.404 acc5_pos: 100.000 loss: 0.3870 (0.5596) acc1: 90.3646 (72.7469) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.2424 data: 0.1471 max mem: 3531 +Test: [190/517] eta: 0:01:27 acc1_pos: 84.404 acc5_pos: 100.000 loss: 0.3004 (0.5454) acc1: 96.3542 (74.0388) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2493 data: 0.1509 max mem: 3531 +Test: [200/517] eta: 0:01:24 acc1_pos: 84.404 acc5_pos: 100.000 loss: 0.2992 (0.5341) acc1: 96.8750 (75.1788) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2372 data: 0.1407 max mem: 3531 +Test: [210/517] eta: 0:01:21 acc1_pos: 84.404 acc5_pos: 100.000 loss: 0.3555 (0.5336) acc1: 91.9271 (75.2703) acc5: 100.0000 (100.0000) acc1_neg: 76.302 acc5_neg: 100.000 time: 0.2353 data: 0.1420 max mem: 3532 +Test: [220/517] eta: 0:01:17 acc1_pos: 84.404 acc5_pos: 100.000 loss: 0.5408 (0.5337) acc1: 75.2604 (75.3005) acc5: 100.0000 (100.0000) acc1_neg: 73.958 acc5_neg: 100.000 time: 0.2330 data: 0.1412 max mem: 3532 +Test: [230/517] eta: 0:01:15 acc1_pos: 84.404 acc5_pos: 100.000 loss: 0.5425 (0.5346) acc1: 73.9583 (75.2976) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.2395 data: 0.1471 max mem: 3532 +Test: [240/517] eta: 0:01:11 acc1_pos: 84.404 acc5_pos: 100.000 loss: 0.5403 (0.5339) acc1: 76.8229 (75.4819) acc5: 100.0000 (100.0000) acc1_neg: 59.115 acc5_neg: 100.000 time: 0.2205 data: 0.1269 max mem: 3532 +Test: [250/517] eta: 0:01:09 acc1_pos: 84.404 acc5_pos: 100.000 loss: 0.4758 (0.5313) acc1: 84.6354 (75.9327) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.2230 data: 0.1296 max mem: 3532 +Test: [260/517] eta: 0:01:06 acc1_pos: 84.404 acc5_pos: 100.000 loss: 0.4624 (0.5286) acc1: 86.9792 (76.4308) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.2540 data: 0.1603 max mem: 3532 +Test: [270/517] eta: 0:01:03 acc1_pos: 84.404 acc5_pos: 100.000 loss: 0.4705 (0.5273) acc1: 84.3750 (76.6788) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.2465 data: 0.1520 max mem: 3532 +Test: [280/517] eta: 0:01:00 acc1_pos: 84.404 acc5_pos: 100.000 loss: 0.4894 (0.5249) acc1: 83.5938 (77.0277) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.2374 data: 0.1414 max mem: 3532 +Test: [290/517] eta: 0:00:58 acc1_pos: 84.404 acc5_pos: 100.000 loss: 0.4535 (0.5226) acc1: 86.9792 (77.4189) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.2283 data: 0.1320 max mem: 3532 +Test: [300/517] eta: 0:00:55 acc1_pos: 84.404 acc5_pos: 100.000 loss: 0.4310 (0.5199) acc1: 89.5833 (77.7599) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.2239 data: 0.1271 max mem: 3532 +Test: [310/517] eta: 0:00:52 acc1_pos: 84.404 acc5_pos: 100.000 loss: 0.5410 (0.5229) acc1: 75.0000 (77.3538) acc5: 100.0000 (100.0000) acc1_neg: 52.865 acc5_neg: 100.000 time: 0.2280 data: 0.1290 max mem: 3532 +Test: [320/517] eta: 0:00:49 acc1_pos: 84.404 acc5_pos: 100.000 loss: 0.5551 (0.5213) acc1: 70.8333 (77.4809) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.2335 data: 0.1368 max mem: 3532 +Test: [330/517] eta: 0:00:47 acc1_pos: 84.404 acc5_pos: 100.000 loss: 0.4062 (0.5175) acc1: 86.1979 (77.8284) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2318 data: 0.1380 max mem: 3532 +Test: [340/517] eta: 0:00:44 acc1_pos: 84.404 acc5_pos: 100.000 loss: 0.4174 (0.5176) acc1: 89.3229 (77.7920) acc5: 100.0000 (100.0000) acc1_neg: 51.302 acc5_neg: 100.000 time: 0.2410 data: 0.1459 max mem: 3532 +Test: [350/517] eta: 0:00:42 acc1_pos: 84.404 acc5_pos: 100.000 loss: 0.6567 (0.5224) acc1: 59.1146 (77.1449) acc5: 100.0000 (100.0000) acc1_neg: 41.667 acc5_neg: 100.000 time: 0.2449 data: 0.1485 max mem: 3532 +Test: [360/517] eta: 0:00:39 acc1_pos: 84.404 acc5_pos: 100.000 loss: 0.6665 (0.5244) acc1: 59.1146 (76.8878) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.2598 data: 0.1625 max mem: 3532 +Test: [370/517] eta: 0:00:37 acc1_pos: 84.404 acc5_pos: 100.000 loss: 0.4494 (0.5204) acc1: 84.8958 (77.2995) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2649 data: 0.1646 max mem: 3532 +Test: [380/517] eta: 0:00:34 acc1_pos: 84.404 acc5_pos: 100.000 loss: 0.3768 (0.5179) acc1: 91.1458 (77.5461) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.2497 data: 0.1497 max mem: 3532 +Test: [390/517] eta: 0:00:31 acc1_pos: 84.404 acc5_pos: 100.000 loss: 0.4321 (0.5158) acc1: 82.0312 (77.6388) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.2249 data: 0.1284 max mem: 3532 +Test: [400/517] eta: 0:00:29 acc1_pos: 84.404 acc5_pos: 100.000 loss: 0.4490 (0.5139) acc1: 83.3333 (77.8652) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.2450 data: 0.1464 max mem: 3532 +Test: [410/517] eta: 0:00:26 acc1_pos: 84.404 acc5_pos: 100.000 loss: 0.4584 (0.5131) acc1: 86.1979 (78.0838) acc5: 100.0000 (100.0000) acc1_neg: 80.729 acc5_neg: 100.000 time: 0.2611 data: 0.1634 max mem: 3532 +Test: [420/517] eta: 0:00:24 acc1_pos: 84.404 acc5_pos: 100.000 loss: 0.5237 (0.5139) acc1: 80.9896 (78.0069) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.2150 data: 0.1176 max mem: 3532 +Test: [430/517] eta: 0:00:21 acc1_pos: 84.404 acc5_pos: 100.000 loss: 0.5059 (0.5131) acc1: 78.1250 (78.0755) acc5: 100.0000 (100.0000) acc1_neg: 66.667 acc5_neg: 100.000 time: 0.2292 data: 0.1308 max mem: 3532 +Test: [440/517] eta: 0:00:19 acc1_pos: 84.404 acc5_pos: 100.000 loss: 0.5047 (0.5132) acc1: 79.6875 (78.0848) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.2353 data: 0.1407 max mem: 3533 +Test: [450/517] eta: 0:00:16 acc1_pos: 84.404 acc5_pos: 100.000 loss: 0.4689 (0.5118) acc1: 84.6354 (78.2590) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.2250 data: 0.1310 max mem: 3533 +Test: [460/517] eta: 0:00:14 acc1_pos: 84.404 acc5_pos: 100.000 loss: 0.4794 (0.5120) acc1: 82.0312 (78.2668) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.2484 data: 0.1534 max mem: 3533 +Test: [470/517] eta: 0:00:11 acc1_pos: 84.404 acc5_pos: 100.000 loss: 0.4817 (0.5105) acc1: 82.0312 (78.4739) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.2459 data: 0.1524 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 84.404 acc5_pos: 100.000 loss: 0.4384 (0.5092) acc1: 86.9792 (78.6464) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2356 data: 0.1433 max mem: 3533 +Test: [490/517] eta: 0:00:06 acc1_pos: 84.404 acc5_pos: 100.000 loss: 0.4384 (0.5074) acc1: 86.7188 (78.8304) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.2350 data: 0.1374 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 84.404 acc5_pos: 100.000 loss: 0.4829 (0.5089) acc1: 80.2083 (78.5824) acc5: 100.0000 (100.0000) acc1_neg: 72.656 acc5_neg: 100.000 time: 0.2392 data: 0.1387 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 84.404 acc5_pos: 100.000 loss: 0.5512 (0.5079) acc1: 73.1771 (78.7029) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.2218 data: 0.1256 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 84.404 acc5_pos: 100.000 loss: 0.5043 (0.5077) acc1: 81.7708 (78.7826) acc5: 100.0000 (100.0000) acc1_neg: 83.746 acc5_neg: 100.000 time: 0.2074 data: 0.1106 max mem: 3533 +Test: Total time: 0:02:08 (0.2479 s / it) +* Overall Acc@1 78.783 Acc@5 100.000 + target == real Acc@1 49.957 Acc@5 100.000 + target == fake Acc@1 81.773 Acc@5 100.000 + loss 0.508 + PR-AUC 0.9510 ROC-AUC 0.7140 EER 34.49% (thr=0.5815) +Accuracy of the network on the 198427 test images: 78.8% +[rank0]:[W725 18:30:14.828822674 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0725 18:30:34.990000 605455 torch/distributed/run.py:766] +W0725 18:30:34.990000 605455 torch/distributed/run.py:766] ***************************************** +W0725 18:30:34.990000 605455 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:30:34.990000 605455 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:31:03.683347262 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:31:03.036843230 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:31:03.157231965 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:31:04.244623252 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:31:04.534670866 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:31:04.541664998 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:31:04.550914220 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:31:04.609513051 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed3_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n32_classseed3_resampled_n116-040-213_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:44:28 acc1_pos: 66.146 acc5_pos: 100.000 loss: 0.6103 (0.6103) acc1: 66.1458 (66.1458) acc5: 100.0000 (100.0000) time: 6.8611 data: 4.8925 max mem: 3531 +Test: [ 10/389] eta: 0:04:27 acc1_pos: 79.688 acc5_pos: 100.000 loss: 0.5706 (0.5785) acc1: 66.9271 (65.8381) acc5: 100.0000 (100.0000) time: 0.7061 data: 0.4452 max mem: 3531 +Test: [ 20/389] eta: 0:02:33 acc1_pos: 58.333 acc5_pos: 100.000 loss: 0.6160 (0.6075) acc1: 66.9271 (65.2406) acc5: 100.0000 (100.0000) time: 0.0949 data: 0.0005 max mem: 3531 +Test: [ 30/389] eta: 0:02:08 acc1_pos: 77.344 acc5_pos: 100.000 loss: 0.6160 (0.6033) acc1: 69.2708 (67.6075) acc5: 100.0000 (100.0000) time: 0.1663 data: 0.0700 max mem: 3531 +Test: [ 40/389] eta: 0:01:56 acc1_pos: 50.521 acc5_pos: 100.000 loss: 0.6161 (0.6167) acc1: 63.5417 (66.6730) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1500 max mem: 3531 +Test: [ 50/389] eta: 0:01:41 acc1_pos: 40.365 acc5_pos: 100.000 loss: 0.6502 (0.6303) acc1: 61.9792 (65.7527) acc5: 100.0000 (100.0000) time: 0.2106 data: 0.1154 max mem: 3531 +Test: [ 60/389] eta: 0:01:34 acc1_pos: 72.656 acc5_pos: 100.000 loss: 0.6502 (0.6336) acc1: 52.3438 (65.3945) acc5: 100.0000 (100.0000) time: 0.1899 data: 0.0952 max mem: 3531 +Test: [ 70/389] eta: 0:01:28 acc1_pos: 78.385 acc5_pos: 100.000 loss: 0.4885 (0.6132) acc1: 75.0000 (67.3122) acc5: 100.0000 (100.0000) time: 0.2233 data: 0.1298 max mem: 3531 +Test: [ 80/389] eta: 0:01:22 acc1_pos: 95.052 acc5_pos: 100.000 loss: 0.5058 (0.6047) acc1: 75.0000 (68.2420) acc5: 100.0000 (100.0000) time: 0.2051 data: 0.1134 max mem: 3531 +Test: [ 90/389] eta: 0:01:17 acc1_pos: 84.375 acc5_pos: 100.000 loss: 0.3183 (0.5762) acc1: 86.9792 (70.6616) acc5: 100.0000 (100.0000) time: 0.1891 data: 0.0962 max mem: 3531 +Test: [100/389] eta: 0:01:13 acc1_pos: 42.969 acc5_pos: 100.000 loss: 0.5918 (0.5948) acc1: 63.5417 (68.3375) acc5: 100.0000 (100.0000) time: 0.1999 data: 0.1060 max mem: 3531 +Test: [110/389] eta: 0:01:09 acc1_pos: 93.229 acc5_pos: 100.000 loss: 0.7744 (0.6056) acc1: 48.6979 (66.8403) acc5: 100.0000 (100.0000) time: 0.1968 data: 0.1044 max mem: 3531 +Test: [120/389] eta: 0:01:05 acc1_pos: 83.854 acc5_pos: 100.000 loss: 0.5379 (0.6050) acc1: 65.6250 (66.8604) acc5: 100.0000 (100.0000) time: 0.1876 data: 0.0955 max mem: 3531 +Test: [130/389] eta: 0:01:01 acc1_pos: 71.779 acc5_pos: 100.000 loss: 0.5677 (0.6099) acc1: 63.0208 (66.4778) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.1849 data: 0.0927 max mem: 3531 +Test: [140/389] eta: 0:00:58 acc1_pos: 71.779 acc5_pos: 100.000 loss: 0.7372 (0.6178) acc1: 49.7396 (65.4348) acc5: 100.0000 (100.0000) acc1_neg: 49.740 acc5_neg: 100.000 time: 0.1774 data: 0.0850 max mem: 3531 +Test: [150/389] eta: 0:00:54 acc1_pos: 71.779 acc5_pos: 100.000 loss: 0.7372 (0.6257) acc1: 49.4792 (64.7713) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1763 data: 0.0841 max mem: 3531 +Test: [160/389] eta: 0:00:52 acc1_pos: 71.779 acc5_pos: 100.000 loss: 0.5965 (0.6168) acc1: 61.4583 (65.6978) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1914 data: 0.1001 max mem: 3531 +Test: [170/389] eta: 0:00:49 acc1_pos: 71.779 acc5_pos: 100.000 loss: 0.3360 (0.6072) acc1: 95.5729 (66.7108) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.1984 data: 0.1074 max mem: 3531 +Test: [180/389] eta: 0:00:46 acc1_pos: 71.779 acc5_pos: 100.000 loss: 0.3174 (0.5967) acc1: 90.1042 (67.8162) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.1942 data: 0.1032 max mem: 3531 +Test: [190/389] eta: 0:00:44 acc1_pos: 71.779 acc5_pos: 100.000 loss: 0.5091 (0.6047) acc1: 79.4271 (67.0321) acc5: 100.0000 (100.0000) acc1_neg: 2.083 acc5_neg: 100.000 time: 0.1883 data: 0.0963 max mem: 3531 +Test: [200/389] eta: 0:00:41 acc1_pos: 71.779 acc5_pos: 100.000 loss: 0.8334 (0.6147) acc1: 32.2917 (65.7377) acc5: 100.0000 (100.0000) acc1_neg: 27.604 acc5_neg: 100.000 time: 0.1908 data: 0.0981 max mem: 3531 +Test: [210/389] eta: 0:00:39 acc1_pos: 71.779 acc5_pos: 100.000 loss: 0.4776 (0.6076) acc1: 32.2917 (66.2174) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1874 data: 0.0949 max mem: 3532 +Test: [220/389] eta: 0:00:36 acc1_pos: 71.779 acc5_pos: 100.000 loss: 0.3151 (0.5972) acc1: 96.3542 (66.9000) acc5: 100.0000 (100.0000) acc1_neg: 38.542 acc5_neg: 100.000 time: 0.1734 data: 0.0809 max mem: 3532 +Test: [230/389] eta: 0:00:33 acc1_pos: 71.779 acc5_pos: 100.000 loss: 0.4668 (0.5997) acc1: 68.7500 (66.4931) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1586 data: 0.0666 max mem: 3532 +Test: [240/389] eta: 0:00:31 acc1_pos: 71.779 acc5_pos: 100.000 loss: 0.5386 (0.5974) acc1: 68.7500 (66.8536) acc5: 100.0000 (100.0000) acc1_neg: 67.708 acc5_neg: 100.000 time: 0.1570 data: 0.0653 max mem: 3532 +Test: [250/389] eta: 0:00:29 acc1_pos: 71.779 acc5_pos: 100.000 loss: 0.4466 (0.5903) acc1: 79.4271 (67.7529) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1584 data: 0.0659 max mem: 3532 +Test: [260/389] eta: 0:00:26 acc1_pos: 71.779 acc5_pos: 100.000 loss: 0.5337 (0.5909) acc1: 77.0833 (67.7293) acc5: 100.0000 (100.0000) acc1_neg: 24.219 acc5_neg: 100.000 time: 0.1504 data: 0.0585 max mem: 3532 +Test: [270/389] eta: 0:00:24 acc1_pos: 71.779 acc5_pos: 100.000 loss: 0.6424 (0.6002) acc1: 50.7812 (66.7580) acc5: 100.0000 (100.0000) acc1_neg: 63.542 acc5_neg: 100.000 time: 0.1481 data: 0.0570 max mem: 3532 +Test: [280/389] eta: 0:00:22 acc1_pos: 71.779 acc5_pos: 100.000 loss: 0.7433 (0.6054) acc1: 43.4896 (66.2357) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1518 data: 0.0599 max mem: 3532 +Test: [290/389] eta: 0:00:19 acc1_pos: 71.779 acc5_pos: 100.000 loss: 0.6063 (0.6005) acc1: 66.4062 (66.7114) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1596 data: 0.0676 max mem: 3532 +Test: [300/389] eta: 0:00:17 acc1_pos: 71.779 acc5_pos: 100.000 loss: 0.5092 (0.5977) acc1: 80.2083 (67.2074) acc5: 100.0000 (100.0000) acc1_neg: 69.271 acc5_neg: 100.000 time: 0.1647 data: 0.0702 max mem: 3532 +Test: [310/389] eta: 0:00:15 acc1_pos: 71.779 acc5_pos: 100.000 loss: 0.5525 (0.6015) acc1: 76.8229 (66.8392) acc5: 100.0000 (100.0000) acc1_neg: 18.750 acc5_neg: 100.000 time: 0.1790 data: 0.0849 max mem: 3532 +Test: [320/389] eta: 0:00:13 acc1_pos: 71.779 acc5_pos: 100.000 loss: 0.6395 (0.6013) acc1: 61.4583 (66.8135) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1905 data: 0.0991 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 71.779 acc5_pos: 100.000 loss: 0.8990 (0.6199) acc1: 31.7708 (65.2143) acc5: 100.0000 (100.0000) acc1_neg: 31.771 acc5_neg: 100.000 time: 0.1828 data: 0.0899 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 71.779 acc5_pos: 100.000 loss: 0.4573 (0.6127) acc1: 66.1458 (66.0458) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.1649 data: 0.0726 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 71.779 acc5_pos: 100.000 loss: 0.4255 (0.6188) acc1: 89.3229 (65.5018) acc5: 100.0000 (100.0000) acc1_neg: 16.406 acc5_neg: 100.000 time: 0.1481 data: 0.0572 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 71.779 acc5_pos: 100.000 loss: 0.6357 (0.6215) acc1: 60.4167 (65.1200) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.1383 data: 0.0477 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 71.779 acc5_pos: 100.000 loss: 0.6297 (0.6195) acc1: 62.2396 (65.2242) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1306 data: 0.0399 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 71.779 acc5_pos: 100.000 loss: 0.4969 (0.6189) acc1: 70.5729 (65.3523) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1291 data: 0.0386 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 71.779 acc5_pos: 100.000 loss: 0.4837 (0.6163) acc1: 75.2604 (65.6436) acc5: 100.0000 (100.0000) acc1_neg: 69.945 acc5_neg: 100.000 time: 0.1312 data: 0.0303 max mem: 3532 +Test: Total time: 0:01:14 (0.1905 s / it) +* Overall Acc@1 65.644 Acc@5 100.000 + target == real Acc@1 66.834 Acc@5 100.000 + target == fake Acc@1 65.090 Acc@5 100.000 + loss 0.616 + PR-AUC 0.8402 ROC-AUC 0.7213 EER 34.14% (thr=0.4941) +Accuracy of the network on the 149358 test images: 65.6% +[rank0]:[W725 18:32:26.018496010 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:32:45.951000 606024 torch/distributed/run.py:766] +W0725 18:32:45.951000 606024 torch/distributed/run.py:766] ***************************************** +W0725 18:32:45.951000 606024 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:32:45.951000 606024 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:33:14.974233609 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank4]:[W725 18:33:14.979017988 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:33:15.664888154 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:33:15.788497582 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:33:15.828378896 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:33:15.947197694 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:33:15.984049006 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:33:15.019556333 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed3_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n32_classseed3_resampled_n116-040-213_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:30:27 acc1_pos: 5.990 acc5_pos: 100.000 loss: 1.5778 (1.5778) acc1: 5.9896 (5.9896) acc5: 100.0000 (100.0000) time: 5.4877 data: 4.0475 max mem: 3531 +Test: [ 10/333] eta: 0:03:07 acc1_pos: 32.031 acc5_pos: 100.000 loss: 1.3130 (1.2819) acc1: 20.8333 (23.2481) acc5: 100.0000 (100.0000) time: 0.5811 data: 0.3683 max mem: 3531 +Test: [ 20/333] eta: 0:01:49 acc1_pos: 39.323 acc5_pos: 100.000 loss: 1.1678 (1.2090) acc1: 28.1250 (27.0833) acc5: 100.0000 (100.0000) time: 0.0918 data: 0.0004 max mem: 3531 +Test: [ 30/333] eta: 0:01:21 acc1_pos: 58.073 acc5_pos: 100.000 loss: 0.9435 (1.0793) acc1: 40.6250 (34.4086) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0004 max mem: 3531 +Test: [ 40/333] eta: 0:01:12 acc1_pos: 53.125 acc5_pos: 100.000 loss: 0.7603 (1.0019) acc1: 49.4792 (37.9954) acc5: 100.0000 (100.0000) time: 0.1404 data: 0.0440 max mem: 3531 +Test: [ 50/333] eta: 0:01:09 acc1_pos: 48.438 acc5_pos: 100.000 loss: 0.7582 (0.9531) acc1: 51.0417 (40.7578) acc5: 100.0000 (100.0000) time: 0.2099 data: 0.1166 max mem: 3531 +Test: [ 60/333] eta: 0:01:02 acc1_pos: 41.927 acc5_pos: 100.000 loss: 0.7595 (0.9276) acc1: 48.4375 (41.8417) acc5: 100.0000 (100.0000) time: 0.1895 data: 0.0941 max mem: 3531 +Test: [ 70/333] eta: 0:00:59 acc1_pos: 36.719 acc5_pos: 100.000 loss: 0.7620 (0.9062) acc1: 48.9583 (43.1485) acc5: 100.0000 (100.0000) time: 0.1799 data: 0.0828 max mem: 3531 +Test: [ 80/333] eta: 0:00:55 acc1_pos: 58.333 acc5_pos: 100.000 loss: 0.7553 (0.8900) acc1: 50.5208 (43.8368) acc5: 100.0000 (100.0000) time: 0.1948 data: 0.0985 max mem: 3531 +Test: [ 90/333] eta: 0:00:52 acc1_pos: 69.010 acc5_pos: 100.000 loss: 0.7150 (0.8644) acc1: 52.8646 (45.8219) acc5: 100.0000 (100.0000) time: 0.1699 data: 0.0751 max mem: 3531 +Test: [100/333] eta: 0:00:49 acc1_pos: 68.490 acc5_pos: 100.000 loss: 0.6320 (0.8409) acc1: 63.5417 (47.5196) acc5: 100.0000 (100.0000) time: 0.1920 data: 0.0980 max mem: 3531 +Test: [110/333] eta: 0:00:47 acc1_pos: 71.354 acc5_pos: 100.000 loss: 0.6366 (0.8230) acc1: 61.9792 (48.8645) acc5: 100.0000 (100.0000) time: 0.1999 data: 0.1072 max mem: 3531 +Test: [120/333] eta: 0:00:44 acc1_pos: 64.323 acc5_pos: 100.000 loss: 0.6119 (0.8056) acc1: 64.3229 (50.2109) acc5: 100.0000 (100.0000) time: 0.1793 data: 0.0849 max mem: 3531 +Test: [130/333] eta: 0:00:41 acc1_pos: 75.521 acc5_pos: 100.000 loss: 0.5946 (0.7907) acc1: 66.6667 (51.3955) acc5: 100.0000 (100.0000) time: 0.1749 data: 0.0785 max mem: 3531 +Test: [140/333] eta: 0:00:39 acc1_pos: 78.385 acc5_pos: 100.000 loss: 0.5625 (0.7750) acc1: 70.8333 (52.7852) acc5: 100.0000 (100.0000) time: 0.1897 data: 0.0933 max mem: 3531 +Test: [150/333] eta: 0:00:37 acc1_pos: 67.708 acc5_pos: 100.000 loss: 0.5879 (0.7655) acc1: 67.7083 (53.4958) acc5: 100.0000 (100.0000) time: 0.1899 data: 0.0933 max mem: 3531 +Test: [160/333] eta: 0:00:34 acc1_pos: 53.289 acc5_pos: 100.000 loss: 0.6085 (0.7593) acc1: 64.0625 (53.9289) acc5: 100.0000 (100.0000) acc1_neg: 53.125 acc5_neg: 100.000 time: 0.1737 data: 0.0769 max mem: 3531 +Test: [170/333] eta: 0:00:32 acc1_pos: 53.289 acc5_pos: 100.000 loss: 0.6582 (0.7544) acc1: 60.4167 (54.2382) acc5: 100.0000 (100.0000) acc1_neg: 66.667 acc5_neg: 100.000 time: 0.1749 data: 0.0763 max mem: 3531 +Test: [180/333] eta: 0:00:31 acc1_pos: 53.289 acc5_pos: 100.000 loss: 0.6200 (0.7457) acc1: 63.2812 (54.9378) acc5: 100.0000 (100.0000) acc1_neg: 72.396 acc5_neg: 100.000 time: 0.2223 data: 0.1240 max mem: 3531 +Test: [190/333] eta: 0:00:29 acc1_pos: 53.289 acc5_pos: 100.000 loss: 0.5549 (0.7336) acc1: 67.9688 (55.9691) acc5: 100.0000 (100.0000) acc1_neg: 68.490 acc5_neg: 100.000 time: 0.2306 data: 0.1355 max mem: 3531 +Test: [200/333] eta: 0:00:26 acc1_pos: 53.289 acc5_pos: 100.000 loss: 0.5404 (0.7254) acc1: 69.5312 (56.6892) acc5: 100.0000 (100.0000) acc1_neg: 68.490 acc5_neg: 100.000 time: 0.1475 data: 0.0521 max mem: 3531 +Test: [210/333] eta: 0:00:24 acc1_pos: 53.289 acc5_pos: 100.000 loss: 0.5978 (0.7196) acc1: 68.2292 (57.1695) acc5: 100.0000 (100.0000) acc1_neg: 70.573 acc5_neg: 100.000 time: 0.1442 data: 0.0379 max mem: 3532 +Test: [220/333] eta: 0:00:22 acc1_pos: 53.289 acc5_pos: 100.000 loss: 0.6038 (0.7170) acc1: 68.2292 (57.4731) acc5: 100.0000 (100.0000) acc1_neg: 71.615 acc5_neg: 100.000 time: 0.2050 data: 0.1007 max mem: 3532 +Test: [230/333] eta: 0:00:20 acc1_pos: 53.289 acc5_pos: 100.000 loss: 0.5540 (0.7097) acc1: 71.8750 (58.2296) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.2101 data: 0.1171 max mem: 3532 +Test: [240/333] eta: 0:00:18 acc1_pos: 53.289 acc5_pos: 100.000 loss: 0.5540 (0.7060) acc1: 68.4896 (58.5365) acc5: 100.0000 (100.0000) acc1_neg: 67.969 acc5_neg: 100.000 time: 0.1919 data: 0.0988 max mem: 3532 +Test: [250/333] eta: 0:00:16 acc1_pos: 53.289 acc5_pos: 100.000 loss: 0.6122 (0.7018) acc1: 64.5833 (58.8583) acc5: 100.0000 (100.0000) acc1_neg: 64.062 acc5_neg: 100.000 time: 0.1912 data: 0.0984 max mem: 3532 +Test: [260/333] eta: 0:00:14 acc1_pos: 53.289 acc5_pos: 100.000 loss: 0.5713 (0.6964) acc1: 67.1875 (59.3052) acc5: 100.0000 (100.0000) acc1_neg: 71.354 acc5_neg: 100.000 time: 0.1879 data: 0.0967 max mem: 3532 +Test: [270/333] eta: 0:00:12 acc1_pos: 53.289 acc5_pos: 100.000 loss: 0.5428 (0.6909) acc1: 72.6562 (59.8161) acc5: 100.0000 (100.0000) acc1_neg: 70.052 acc5_neg: 100.000 time: 0.1842 data: 0.0928 max mem: 3532 +Test: [280/333] eta: 0:00:10 acc1_pos: 53.289 acc5_pos: 100.000 loss: 0.5529 (0.6904) acc1: 71.3542 (59.7920) acc5: 100.0000 (100.0000) acc1_neg: 60.156 acc5_neg: 100.000 time: 0.1962 data: 0.1033 max mem: 3532 +Test: [290/333] eta: 0:00:08 acc1_pos: 53.289 acc5_pos: 100.000 loss: 0.7304 (0.6930) acc1: 52.3438 (59.4305) acc5: 100.0000 (100.0000) acc1_neg: 52.344 acc5_neg: 100.000 time: 0.1892 data: 0.0943 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 53.289 acc5_pos: 100.000 loss: 0.7266 (0.6929) acc1: 54.4271 (59.4036) acc5: 100.0000 (100.0000) acc1_neg: 58.594 acc5_neg: 100.000 time: 0.1889 data: 0.0948 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 53.289 acc5_pos: 100.000 loss: 0.6533 (0.6879) acc1: 62.7604 (59.8875) acc5: 100.0000 (100.0000) acc1_neg: 68.490 acc5_neg: 100.000 time: 0.2201 data: 0.1265 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 53.289 acc5_pos: 100.000 loss: 0.5808 (0.6852) acc1: 70.3125 (60.1595) acc5: 100.0000 (100.0000) acc1_neg: 57.031 acc5_neg: 100.000 time: 0.1989 data: 0.1054 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 53.289 acc5_pos: 100.000 loss: 0.7057 (0.6870) acc1: 56.7708 (59.9745) acc5: 100.0000 (100.0000) acc1_neg: 53.906 acc5_neg: 100.000 time: 0.1294 data: 0.0379 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 53.289 acc5_pos: 100.000 loss: 0.7223 (0.6872) acc1: 56.7708 (59.9509) acc5: 100.0000 (100.0000) acc1_neg: 56.901 acc5_neg: 100.000 time: 0.1322 data: 0.0379 max mem: 3532 +Test: Total time: 0:01:04 (0.1944 s / it) +* Overall Acc@1 59.951 Acc@5 100.000 + target == real Acc@1 54.016 Acc@5 100.000 + target == fake Acc@1 65.234 Acc@5 100.000 + loss 0.687 + PR-AUC 0.6261 ROC-AUC 0.6301 EER 40.54% (thr=0.5332) +Accuracy of the network on the 127843 test images: 60.0% +[rank0]:[W725 18:34:36.377162928 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:34:55.017000 606593 torch/distributed/run.py:766] +W0725 18:34:55.017000 606593 torch/distributed/run.py:766] ***************************************** +W0725 18:34:55.017000 606593 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:34:55.017000 606593 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 1): env:// +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:35:23.548301807 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:35:23.769746243 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:35:23.826680335 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:35:23.889624642 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:35:23.963367361 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +[rank0]:[W725 18:35:24.202125549 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:35:24.206225144 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:35:24.215703025 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed3_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n32_classseed3_resampled_n116-040-213_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:24:10 acc1_pos: 97.135 acc5_pos: 100.000 loss: 0.1989 (0.1989) acc1: 97.1354 (97.1354) acc5: 100.0000 (100.0000) time: 6.1722 data: 4.5056 max mem: 3531 +Test: [ 10/235] eta: 0:02:28 acc1_pos: 96.615 acc5_pos: 100.000 loss: 0.2041 (0.2091) acc1: 96.3542 (96.1648) acc5: 100.0000 (100.0000) time: 0.6585 data: 0.4224 max mem: 3531 +Test: [ 20/235] eta: 0:01:27 acc1_pos: 96.354 acc5_pos: 100.000 loss: 0.2039 (0.2069) acc1: 96.3542 (96.1558) acc5: 100.0000 (100.0000) time: 0.1187 data: 0.0250 max mem: 3531 +Test: [ 30/235] eta: 0:01:17 acc1_pos: 96.354 acc5_pos: 100.000 loss: 0.2028 (0.2058) acc1: 96.3542 (96.2366) acc5: 100.0000 (100.0000) time: 0.2261 data: 0.1322 max mem: 3531 +Test: [ 40/235] eta: 0:01:05 acc1_pos: 95.833 acc5_pos: 100.000 loss: 0.2057 (0.2055) acc1: 96.0938 (96.1573) acc5: 100.0000 (100.0000) time: 0.2623 data: 0.1687 max mem: 3531 +Test: [ 50/235] eta: 0:00:58 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.2061 (0.2065) acc1: 95.8333 (96.0733) acc5: 100.0000 (100.0000) time: 0.2220 data: 0.1273 max mem: 3531 +Test: [ 60/235] eta: 0:00:52 acc1_pos: 95.573 acc5_pos: 100.000 loss: 0.2111 (0.2075) acc1: 95.8333 (96.0212) acc5: 100.0000 (100.0000) time: 0.2265 data: 0.1329 max mem: 3531 +Test: [ 70/235] eta: 0:00:47 acc1_pos: 98.462 acc5_pos: 100.000 loss: 0.2250 (0.3572) acc1: 94.7917 (86.5500) acc5: 100.0000 (100.0000) acc1_neg: 16.927 acc5_neg: 100.000 time: 0.2105 data: 0.1184 max mem: 3531 +Test: [ 80/235] eta: 0:00:43 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4034 (0.4868) acc1: 19.2708 (78.2375) acc5: 100.0000 (100.0000) acc1_neg: 16.667 acc5_neg: 100.000 time: 0.2112 data: 0.1191 max mem: 3531 +Test: [ 90/235] eta: 0:00:39 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4080 (0.5889) acc1: 19.0104 (71.7834) acc5: 100.0000 (100.0000) acc1_neg: 16.146 acc5_neg: 100.000 time: 0.2220 data: 0.1297 max mem: 3531 +Test: [100/235] eta: 0:00:36 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4145 (0.6693) acc1: 19.0104 (66.6228) acc5: 100.0000 (100.0000) acc1_neg: 16.927 acc5_neg: 100.000 time: 0.2379 data: 0.1460 max mem: 3531 +Test: [110/235] eta: 0:00:33 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.3915 (0.7342) acc1: 19.7917 (62.4062) acc5: 100.0000 (100.0000) acc1_neg: 21.094 acc5_neg: 100.000 time: 0.2352 data: 0.1421 max mem: 3531 +Test: [120/235] eta: 0:00:29 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4509 (0.8029) acc1: 16.1458 (58.2709) acc5: 100.0000 (100.0000) acc1_neg: 13.021 acc5_neg: 100.000 time: 0.2127 data: 0.1187 max mem: 3531 +Test: [130/235] eta: 0:00:27 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.5763 (0.8614) acc1: 11.9792 (54.6935) acc5: 100.0000 (100.0000) acc1_neg: 11.458 acc5_neg: 100.000 time: 0.2087 data: 0.1146 max mem: 3531 +Test: [140/235] eta: 0:00:24 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.5857 (0.9143) acc1: 11.4583 (51.6327) acc5: 100.0000 (100.0000) acc1_neg: 10.156 acc5_neg: 100.000 time: 0.2050 data: 0.1113 max mem: 3531 +Test: [150/235] eta: 0:00:21 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.5858 (0.9579) acc1: 11.4583 (49.0049) acc5: 100.0000 (100.0000) acc1_neg: 12.240 acc5_neg: 100.000 time: 0.1932 data: 0.0987 max mem: 3531 +Test: [160/235] eta: 0:00:18 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.5701 (0.9959) acc1: 11.7188 (46.7197) acc5: 100.0000 (100.0000) acc1_neg: 12.240 acc5_neg: 100.000 time: 0.1942 data: 0.0973 max mem: 3531 +Test: [170/235] eta: 0:00:15 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.5233 (1.0129) acc1: 13.8021 (45.4663) acc5: 100.0000 (100.0000) acc1_neg: 23.438 acc5_neg: 100.000 time: 0.1925 data: 0.0983 max mem: 3531 +Test: [180/235] eta: 0:00:13 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.2518 (1.0256) acc1: 27.3438 (44.5140) acc5: 100.0000 (100.0000) acc1_neg: 28.125 acc5_neg: 100.000 time: 0.1745 data: 0.0834 max mem: 3531 +Test: [190/235] eta: 0:00:10 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.2246 (1.0369) acc1: 27.6042 (43.6382) acc5: 100.0000 (100.0000) acc1_neg: 27.604 acc5_neg: 100.000 time: 0.1733 data: 0.0827 max mem: 3531 +Test: [200/235] eta: 0:00:08 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.2397 (1.0481) acc1: 26.8229 (42.7809) acc5: 100.0000 (100.0000) acc1_neg: 25.000 acc5_neg: 100.000 time: 0.1829 data: 0.0924 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.2528 (1.0576) acc1: 26.8229 (42.0517) acc5: 100.0000 (100.0000) acc1_neg: 28.385 acc5_neg: 100.000 time: 0.1562 data: 0.0658 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.2466 (1.0663) acc1: 27.8646 (41.3980) acc5: 100.0000 (100.0000) acc1_neg: 28.125 acc5_neg: 100.000 time: 0.1347 data: 0.0434 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.2376 (1.0733) acc1: 27.8646 (40.8561) acc5: 100.0000 (100.0000) acc1_neg: 27.865 acc5_neg: 100.000 time: 0.1250 data: 0.0338 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.2376 (1.0776) acc1: 27.8646 (40.6658) acc5: 100.0000 (100.0000) acc1_neg: 26.087 acc5_neg: 100.000 time: 0.1173 data: 0.0305 max mem: 3532 +Test: Total time: 0:00:51 (0.2174 s / it) +* Overall Acc@1 40.666 Acc@5 100.000 + target == real Acc@1 96.033 Acc@5 100.000 + target == fake Acc@1 20.641 Acc@5 100.000 + loss 1.078 + PR-AUC 0.8764 ROC-AUC 0.7308 EER 33.21% (thr=0.1754) +Accuracy of the network on the 89879 test images: 40.7% +[rank0]:[W725 18:36:23.053061963 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:36:43.911000 607149 torch/distributed/run.py:766] +W0725 18:36:43.911000 607149 torch/distributed/run.py:766] ***************************************** +W0725 18:36:43.911000 607149 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:36:43.911000 607149 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:37:11.464807523 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:37:11.066528945 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:37:12.318675153 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:37:12.495500391 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:37:12.659191750 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +[rank2]:[W725 18:37:12.796414481 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:37:12.799093522 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:37:12.803699294 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed3_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n32_classseed3_resampled_n116-040-213_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:22:55 acc1_pos: 31.250 acc5_pos: 100.000 loss: 0.9255 (0.9255) acc1: 31.2500 (31.2500) acc5: 100.0000 (100.0000) time: 5.1151 data: 2.5128 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/269] eta: 0:02:21 acc1_pos: 42.708 acc5_pos: 100.000 loss: 0.8064 (0.8027) acc1: 42.9688 (44.3182) acc5: 100.0000 (100.0000) time: 0.5471 data: 0.2288 max mem: 3531 +Test: [ 20/269] eta: 0:01:25 acc1_pos: 50.781 acc5_pos: 100.000 loss: 0.7427 (0.7582) acc1: 45.5729 (49.7892) acc5: 100.0000 (100.0000) time: 0.1043 data: 0.0143 max mem: 3531 +Test: [ 30/269] eta: 0:01:04 acc1_pos: 56.510 acc5_pos: 100.000 loss: 0.7221 (0.7349) acc1: 50.7812 (51.7893) acc5: 100.0000 (100.0000) time: 0.1143 data: 0.0228 max mem: 3531 +Test: [ 40/269] eta: 0:01:00 acc1_pos: 33.621 acc5_pos: 100.000 loss: 0.5351 (0.6756) acc1: 70.5729 (58.9240) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.1780 data: 0.0826 max mem: 3531 +Test: [ 50/269] eta: 0:00:57 acc1_pos: 33.621 acc5_pos: 100.000 loss: 0.5093 (0.6473) acc1: 80.2083 (62.3060) acc5: 100.0000 (100.0000) acc1_neg: 54.688 acc5_neg: 100.000 time: 0.2578 data: 0.1594 max mem: 3531 +Test: [ 60/269] eta: 0:00:52 acc1_pos: 33.621 acc5_pos: 100.000 loss: 0.5665 (0.6456) acc1: 69.0104 (62.6238) acc5: 100.0000 (100.0000) acc1_neg: 52.083 acc5_neg: 100.000 time: 0.2299 data: 0.1326 max mem: 3531 +Test: [ 70/269] eta: 0:00:48 acc1_pos: 33.621 acc5_pos: 100.000 loss: 0.6624 (0.6498) acc1: 58.8542 (61.8141) acc5: 100.0000 (100.0000) acc1_neg: 61.458 acc5_neg: 100.000 time: 0.1999 data: 0.1029 max mem: 3531 +Test: [ 80/269] eta: 0:00:45 acc1_pos: 33.621 acc5_pos: 100.000 loss: 0.6473 (0.6413) acc1: 61.4583 (62.6511) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2011 data: 0.1040 max mem: 3531 +Test: [ 90/269] eta: 0:00:42 acc1_pos: 33.621 acc5_pos: 100.000 loss: 0.6048 (0.6339) acc1: 65.8854 (63.7477) acc5: 100.0000 (100.0000) acc1_neg: 69.531 acc5_neg: 100.000 time: 0.2054 data: 0.1106 max mem: 3531 +Test: [100/269] eta: 0:00:39 acc1_pos: 33.621 acc5_pos: 100.000 loss: 0.6083 (0.6361) acc1: 65.1042 (63.6293) acc5: 100.0000 (100.0000) acc1_neg: 59.375 acc5_neg: 100.000 time: 0.2237 data: 0.1293 max mem: 3531 +Test: [110/269] eta: 0:00:35 acc1_pos: 33.621 acc5_pos: 100.000 loss: 0.6180 (0.6342) acc1: 63.5417 (63.6261) acc5: 100.0000 (100.0000) acc1_neg: 54.688 acc5_neg: 100.000 time: 0.1644 data: 0.0673 max mem: 3531 +Test: [120/269] eta: 0:00:32 acc1_pos: 33.621 acc5_pos: 100.000 loss: 0.5819 (0.6241) acc1: 67.4479 (64.6522) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.1361 data: 0.0390 max mem: 3531 +Test: [130/269] eta: 0:00:30 acc1_pos: 33.621 acc5_pos: 100.000 loss: 0.5565 (0.6242) acc1: 67.4479 (64.5953) acc5: 100.0000 (100.0000) acc1_neg: 48.438 acc5_neg: 100.000 time: 0.1751 data: 0.0820 max mem: 3531 +Test: [140/269] eta: 0:00:27 acc1_pos: 33.621 acc5_pos: 100.000 loss: 0.6730 (0.6327) acc1: 54.9479 (63.5657) acc5: 100.0000 (100.0000) acc1_neg: 54.948 acc5_neg: 100.000 time: 0.1938 data: 0.1007 max mem: 3531 +Test: [150/269] eta: 0:00:25 acc1_pos: 33.621 acc5_pos: 100.000 loss: 0.6984 (0.6335) acc1: 53.3854 (63.4261) acc5: 100.0000 (100.0000) acc1_neg: 70.573 acc5_neg: 100.000 time: 0.1949 data: 0.0975 max mem: 3531 +Test: [160/269] eta: 0:00:23 acc1_pos: 33.621 acc5_pos: 100.000 loss: 0.5805 (0.6287) acc1: 70.5729 (63.9590) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.1825 data: 0.0834 max mem: 3531 +Test: [170/269] eta: 0:00:20 acc1_pos: 33.621 acc5_pos: 100.000 loss: 0.5293 (0.6231) acc1: 72.9167 (64.6001) acc5: 100.0000 (100.0000) acc1_neg: 72.656 acc5_neg: 100.000 time: 0.1972 data: 0.1029 max mem: 3531 +Test: [180/269] eta: 0:00:18 acc1_pos: 33.621 acc5_pos: 100.000 loss: 0.4810 (0.6145) acc1: 80.2083 (65.5833) acc5: 100.0000 (100.0000) acc1_neg: 59.896 acc5_neg: 100.000 time: 0.1797 data: 0.0887 max mem: 3531 +Test: [190/269] eta: 0:00:16 acc1_pos: 33.621 acc5_pos: 100.000 loss: 0.4903 (0.6128) acc1: 78.1250 (65.7109) acc5: 100.0000 (100.0000) acc1_neg: 70.312 acc5_neg: 100.000 time: 0.1728 data: 0.0793 max mem: 3531 +Test: [200/269] eta: 0:00:14 acc1_pos: 33.621 acc5_pos: 100.000 loss: 0.5898 (0.6135) acc1: 63.5417 (65.5745) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.1924 data: 0.0995 max mem: 3531 +Test: [210/269] eta: 0:00:12 acc1_pos: 33.621 acc5_pos: 100.000 loss: 0.6206 (0.6132) acc1: 61.9792 (65.6053) acc5: 100.0000 (100.0000) acc1_neg: 72.396 acc5_neg: 100.000 time: 0.1906 data: 0.0993 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 33.621 acc5_pos: 100.000 loss: 0.6002 (0.6135) acc1: 65.6250 (65.6568) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1874 data: 0.0959 max mem: 3532 +Test: [230/269] eta: 0:00:07 acc1_pos: 33.621 acc5_pos: 100.000 loss: 0.5146 (0.6090) acc1: 73.1771 (66.1684) acc5: 100.0000 (100.0000) acc1_neg: 74.219 acc5_neg: 100.000 time: 0.1845 data: 0.0920 max mem: 3532 +Test: [240/269] eta: 0:00:05 acc1_pos: 33.621 acc5_pos: 100.000 loss: 0.5589 (0.6104) acc1: 70.8333 (66.1264) acc5: 100.0000 (100.0000) acc1_neg: 68.229 acc5_neg: 100.000 time: 0.1875 data: 0.0948 max mem: 3532 +Test: [250/269] eta: 0:00:03 acc1_pos: 33.621 acc5_pos: 100.000 loss: 0.6357 (0.6137) acc1: 62.5000 (65.7786) acc5: 100.0000 (100.0000) acc1_neg: 46.354 acc5_neg: 100.000 time: 0.1877 data: 0.0963 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 33.621 acc5_pos: 100.000 loss: 0.6889 (0.6197) acc1: 53.1250 (65.0952) acc5: 100.0000 (100.0000) acc1_neg: 61.198 acc5_neg: 100.000 time: 0.1698 data: 0.0791 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 33.621 acc5_pos: 100.000 loss: 0.7254 (0.6202) acc1: 50.2604 (65.0499) acc5: 100.0000 (100.0000) acc1_neg: 83.727 acc5_neg: 100.000 time: 0.1534 data: 0.0509 max mem: 3532 +Test: Total time: 0:00:53 (0.2000 s / it) +* Overall Acc@1 65.050 Acc@5 100.000 + target == real Acc@1 51.411 Acc@5 100.000 + target == fake Acc@1 66.911 Acc@5 100.000 + loss 0.620 + PR-AUC 0.9222 ROC-AUC 0.6359 EER 40.50% (thr=0.5410) +Accuracy of the network on the 103293 test images: 65.0% +[rank0]:[W725 18:38:21.201806832 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-25T18:38:27Z] eval_done tag=fixed500k_real_n32_seed3_resampled +[2026-07-25T18:38:27Z] eval_start tag=fixed500k_fake_n64_seed3_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed3_resampled_n116-040-213_20260725_112559/best_checkpoint.pth +started_utc=2026-07-25T18:38:27Z +hostname=n116-040-213 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed3_resampled_n116-040-213_20260725_112559/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed3_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0725 18:38:40.322000 607734 torch/distributed/run.py:766] +W0725 18:38:40.322000 607734 torch/distributed/run.py:766] ***************************************** +W0725 18:38:40.322000 607734 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:38:40.322000 607734 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:39:09.514093888 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:39:09.904233009 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:39:10.290859268 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:39:10.443477298 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:39:10.530448241 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +[rank0]:[W725 18:39:10.549330305 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:39:10.556650166 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:39:10.586713262 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed3_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed3_resampled_n116-040-213_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:20:42 acc1_pos: 51.302 acc5_pos: 100.000 loss: 0.7578 (0.7578) acc1: 51.3021 (51.3021) acc5: 100.0000 (100.0000) time: 5.3087 data: 3.5200 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/234] eta: 0:02:06 acc1_pos: 37.500 acc5_pos: 100.000 loss: 0.7578 (0.7164) acc1: 51.3021 (55.5871) acc5: 100.0000 (100.0000) time: 0.5654 data: 0.3203 max mem: 3531 +Test: [ 20/234] eta: 0:01:13 acc1_pos: 94.271 acc5_pos: 100.000 loss: 0.7579 (0.7386) acc1: 48.1771 (51.4633) acc5: 100.0000 (100.0000) time: 0.0955 data: 0.0004 max mem: 3531 +Test: [ 30/234] eta: 0:01:12 acc1_pos: 76.042 acc5_pos: 100.000 loss: 0.2360 (0.5832) acc1: 93.7500 (64.6925) acc5: 100.0000 (100.0000) time: 0.2364 data: 0.1410 max mem: 3531 +Test: [ 40/234] eta: 0:01:03 acc1_pos: 98.698 acc5_pos: 100.000 loss: 0.2343 (0.5361) acc1: 94.2708 (68.8389) acc5: 100.0000 (100.0000) time: 0.3152 data: 0.2235 max mem: 3531 +Test: [ 50/234] eta: 0:00:56 acc1_pos: 98.958 acc5_pos: 100.000 loss: 0.1616 (0.4616) acc1: 96.8750 (74.4894) acc5: 100.0000 (100.0000) time: 0.2334 data: 0.1413 max mem: 3531 +Test: [ 60/234] eta: 0:00:50 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.1177 (0.3991) acc1: 98.9583 (78.6544) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2068 data: 0.1154 max mem: 3531 +Test: [ 70/234] eta: 0:00:46 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.0782 (0.3996) acc1: 100.0000 (78.7595) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.2349 data: 0.1432 max mem: 3531 +Test: [ 80/234] eta: 0:00:42 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.5185 (0.4012) acc1: 71.0938 (79.0799) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2233 data: 0.1290 max mem: 3531 +Test: [ 90/234] eta: 0:00:38 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.2021 (0.3759) acc1: 94.0104 (80.8923) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.2056 data: 0.1121 max mem: 3531 +Test: [100/234] eta: 0:00:35 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.2023 (0.3765) acc1: 89.3229 (80.9819) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2293 data: 0.1358 max mem: 3531 +Test: [110/234] eta: 0:00:31 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4082 (0.3779) acc1: 78.6458 (81.1327) acc5: 100.0000 (100.0000) acc1_neg: 72.656 acc5_neg: 100.000 time: 0.2045 data: 0.1091 max mem: 3531 +Test: [120/234] eta: 0:00:28 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4082 (0.3851) acc1: 79.9479 (80.9896) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1951 data: 0.0987 max mem: 3531 +Test: [130/234] eta: 0:00:26 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.3728 (0.3844) acc1: 86.7188 (81.5283) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.2163 data: 0.1199 max mem: 3531 +Test: [140/234] eta: 0:00:22 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4114 (0.4067) acc1: 83.0729 (80.4466) acc5: 100.0000 (100.0000) acc1_neg: 41.406 acc5_neg: 100.000 time: 0.1864 data: 0.0918 max mem: 3531 +Test: [150/234] eta: 0:00:20 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4171 (0.4122) acc1: 77.0833 (80.1187) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.1947 data: 0.1022 max mem: 3531 +Test: [160/234] eta: 0:00:17 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.3815 (0.4106) acc1: 80.9896 (80.4477) acc5: 100.0000 (100.0000) acc1_neg: 77.344 acc5_neg: 100.000 time: 0.2358 data: 0.1453 max mem: 3531 +Test: [170/234] eta: 0:00:15 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4671 (0.4142) acc1: 79.1667 (80.4611) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.2047 data: 0.1133 max mem: 3531 +Test: [180/234] eta: 0:00:12 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4905 (0.4228) acc1: 78.3854 (79.9148) acc5: 100.0000 (100.0000) acc1_neg: 41.667 acc5_neg: 100.000 time: 0.1702 data: 0.0780 max mem: 3531 +Test: [190/234] eta: 0:00:10 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.6996 (0.4387) acc1: 60.1562 (78.7917) acc5: 100.0000 (100.0000) acc1_neg: 59.635 acc5_neg: 100.000 time: 0.1925 data: 0.0999 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.6996 (0.4336) acc1: 60.1562 (79.0954) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1980 data: 0.1057 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.1147 (0.4184) acc1: 100.0000 (80.0750) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1569 data: 0.0662 max mem: 3532 +Test: [220/234] eta: 0:00:03 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.1112 (0.4055) acc1: 99.7396 (80.8753) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.1302 data: 0.0402 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.3472 (0.4222) acc1: 53.3854 (79.5139) acc5: 100.0000 (100.0000) acc1_neg: 52.083 acc5_neg: 100.000 time: 0.1230 data: 0.0331 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.7765 (0.4269) acc1: 50.5208 (79.1639) acc5: 100.0000 (100.0000) acc1_neg: 50.973 acc5_neg: 100.000 time: 0.1184 data: 0.0272 max mem: 3532 +Test: Total time: 0:00:50 (0.2166 s / it) +* Overall Acc@1 79.164 Acc@5 100.000 + target == real Acc@1 74.662 Acc@5 100.000 + target == fake Acc@1 80.433 Acc@5 100.000 + loss 0.427 + PR-AUC 0.9644 ROC-AUC 0.8806 EER 21.96% (thr=0.5396) +Accuracy of the network on the 89729 test images: 79.2% +[rank0]:[W725 18:40:10.197769516 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:40:28.738000 608307 torch/distributed/run.py:766] +W0725 18:40:28.738000 608307 torch/distributed/run.py:766] ***************************************** +W0725 18:40:28.738000 608307 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:40:28.738000 608307 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:40:57.877913519 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:40:58.225447560 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:40:58.243974518 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:40:58.680142815 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:40:58.702952901 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:40:58.726398130 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +[rank0]:[W725 18:40:58.733840440 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:40:58.744343739 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed3_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed3_resampled_n116-040-213_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:03:27 acc1_pos: 32.031 acc5_pos: 100.000 loss: 1.0122 (1.0122) acc1: 32.0312 (32.0312) acc5: 100.0000 (100.0000) time: 6.1194 data: 4.1340 max mem: 3531 +Test: [ 10/2387] eta: 0:25:17 acc1_pos: 9.115 acc5_pos: 100.000 loss: 1.1581 (1.1258) acc1: 19.7917 (22.7746) acc5: 100.0000 (100.0000) time: 0.6383 data: 0.3761 max mem: 3531 +Test: [ 20/2387] eta: 0:14:53 acc1_pos: 18.750 acc5_pos: 100.000 loss: 1.1143 (1.1141) acc1: 19.7917 (23.0903) acc5: 100.0000 (100.0000) time: 0.0902 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:11:40 acc1_pos: 7.031 acc5_pos: 100.000 loss: 1.1358 (1.1352) acc1: 18.7500 (21.6482) acc5: 100.0000 (100.0000) time: 0.1094 data: 0.0194 max mem: 3531 +Test: [ 40/2387] eta: 0:10:57 acc1_pos: 18.750 acc5_pos: 100.000 loss: 1.1993 (1.1562) acc1: 14.3229 (19.0485) acc5: 100.0000 (100.0000) time: 0.1782 data: 0.0875 max mem: 3531 +Test: [ 50/2387] eta: 0:10:46 acc1_pos: 31.771 acc5_pos: 100.000 loss: 1.1880 (1.1451) acc1: 15.8854 (19.6436) acc5: 100.0000 (100.0000) time: 0.2450 data: 0.1530 max mem: 3531 +Test: [ 60/2387] eta: 0:10:32 acc1_pos: 38.542 acc5_pos: 100.000 loss: 1.1172 (1.1464) acc1: 16.4062 (19.4245) acc5: 100.0000 (100.0000) time: 0.2551 data: 0.1636 max mem: 3531 +Test: [ 70/2387] eta: 0:10:25 acc1_pos: 27.865 acc5_pos: 100.000 loss: 1.2267 (1.1616) acc1: 14.3229 (18.9040) acc5: 100.0000 (100.0000) time: 0.2527 data: 0.1602 max mem: 3531 +Test: [ 80/2387] eta: 0:10:19 acc1_pos: 34.896 acc5_pos: 100.000 loss: 1.1305 (1.1513) acc1: 15.8854 (19.1197) acc5: 100.0000 (100.0000) time: 0.2587 data: 0.1584 max mem: 3531 +Test: [ 90/2387] eta: 0:10:10 acc1_pos: 38.542 acc5_pos: 100.000 loss: 1.0039 (1.1301) acc1: 25.5208 (20.4642) acc5: 100.0000 (100.0000) time: 0.2503 data: 0.1514 max mem: 3531 +Test: [ 100/2387] eta: 0:10:06 acc1_pos: 25.781 acc5_pos: 100.000 loss: 0.9483 (1.1214) acc1: 25.7812 (20.8359) acc5: 100.0000 (100.0000) time: 0.2502 data: 0.1573 max mem: 3531 +Test: [ 110/2387] eta: 0:10:00 acc1_pos: 24.740 acc5_pos: 100.000 loss: 0.9855 (1.1081) acc1: 25.5208 (21.5114) acc5: 100.0000 (100.0000) time: 0.2546 data: 0.1626 max mem: 3531 +Test: [ 120/2387] eta: 0:09:55 acc1_pos: 25.260 acc5_pos: 100.000 loss: 0.9924 (1.1009) acc1: 25.2604 (22.0622) acc5: 100.0000 (100.0000) time: 0.2503 data: 0.1569 max mem: 3531 +Test: [ 130/2387] eta: 0:09:50 acc1_pos: 23.698 acc5_pos: 100.000 loss: 1.0014 (1.0941) acc1: 23.6979 (22.1116) acc5: 100.0000 (100.0000) time: 0.2510 data: 0.1543 max mem: 3531 +Test: [ 140/2387] eta: 0:09:52 acc1_pos: 24.740 acc5_pos: 100.000 loss: 1.0012 (1.0876) acc1: 23.6979 (22.3995) acc5: 100.0000 (100.0000) time: 0.2700 data: 0.1739 max mem: 3531 +Test: [ 150/2387] eta: 0:09:59 acc1_pos: 9.375 acc5_pos: 100.000 loss: 0.9875 (1.0797) acc1: 25.0000 (22.6442) acc5: 100.0000 (100.0000) time: 0.3090 data: 0.2128 max mem: 3531 +Test: [ 160/2387] eta: 0:09:34 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.9436 (1.0544) acc1: 26.3021 (24.6167) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2193 data: 0.1198 max mem: 3531 +Test: [ 170/2387] eta: 0:09:26 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1997 (1.0037) acc1: 97.3958 (28.9306) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1622 data: 0.0621 max mem: 3531 +Test: [ 180/2387] eta: 0:09:15 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1888 (0.9586) acc1: 98.6979 (32.7823) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2008 data: 0.1045 max mem: 3531 +Test: [ 190/2387] eta: 0:09:07 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1857 (0.9181) acc1: 98.6979 (36.2306) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1929 data: 0.0975 max mem: 3531 +Test: [ 200/2387] eta: 0:08:57 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1831 (0.8816) acc1: 98.6979 (39.3398) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1906 data: 0.0935 max mem: 3531 +Test: [ 210/2387] eta: 0:08:49 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1879 (0.8489) acc1: 98.6979 (42.1443) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1900 data: 0.0936 max mem: 3532 +Test: [ 220/2387] eta: 0:08:40 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1909 (0.8190) acc1: 98.6979 (44.7021) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1833 data: 0.0870 max mem: 3532 +Test: [ 230/2387] eta: 0:08:36 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1898 (0.7917) acc1: 98.6979 (47.0306) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1961 data: 0.0982 max mem: 3532 +Test: [ 240/2387] eta: 0:08:28 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1944 (0.7674) acc1: 98.4375 (49.1593) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2008 data: 0.1041 max mem: 3532 +Test: [ 250/2387] eta: 0:08:23 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1806 (0.7404) acc1: 98.9583 (51.1776) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1932 data: 0.0969 max mem: 3532 +Test: [ 260/2387] eta: 0:08:16 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.0728 (0.7148) acc1: 100.0000 (53.0482) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1942 data: 0.0963 max mem: 3532 +Test: [ 270/2387] eta: 0:08:05 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.0699 (0.6910) acc1: 100.0000 (54.7798) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1477 data: 0.0496 max mem: 3532 +Test: [ 280/2387] eta: 0:07:56 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.0707 (0.6691) acc1: 100.0000 (56.3890) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1268 data: 0.0316 max mem: 3532 +Test: [ 290/2387] eta: 0:07:55 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.0746 (0.6486) acc1: 100.0000 (57.8877) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1920 data: 0.0989 max mem: 3532 +Test: [ 300/2387] eta: 0:07:52 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.0731 (0.6295) acc1: 100.0000 (59.2868) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2299 data: 0.1340 max mem: 3532 +Test: [ 310/2387] eta: 0:07:49 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.0770 (0.6142) acc1: 100.0000 (60.5850) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2174 data: 0.1220 max mem: 3532 +Test: [ 320/2387] eta: 0:07:41 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.2442 (0.6028) acc1: 99.7396 (61.8023) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1798 data: 0.0867 max mem: 3532 +Test: [ 330/2387] eta: 0:07:32 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.2449 (0.5919) acc1: 99.7396 (62.9508) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1285 data: 0.0354 max mem: 3532 +Test: [ 340/2387] eta: 0:07:29 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.2426 (0.5817) acc1: 99.7396 (64.0312) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1630 data: 0.0680 max mem: 3532 +Test: [ 350/2387] eta: 0:07:27 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.2380 (0.5688) acc1: 100.0000 (65.0545) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2089 data: 0.1063 max mem: 3532 +Test: [ 360/2387] eta: 0:07:22 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1479 (0.5572) acc1: 100.0000 (66.0088) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1905 data: 0.0890 max mem: 3532 +Test: [ 370/2387] eta: 0:07:18 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1549 (0.5464) acc1: 99.4792 (66.9088) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1855 data: 0.0922 max mem: 3532 +Test: [ 380/2387] eta: 0:07:14 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1571 (0.5361) acc1: 99.4792 (67.7596) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1848 data: 0.0928 max mem: 3532 +Test: [ 390/2387] eta: 0:07:12 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1547 (0.5264) acc1: 99.4792 (68.5722) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1993 data: 0.1053 max mem: 3532 +Test: [ 400/2387] eta: 0:07:07 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1547 (0.5172) acc1: 99.4792 (69.3403) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1948 data: 0.1003 max mem: 3532 +Test: [ 410/2387] eta: 0:07:02 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1543 (0.5084) acc1: 99.2188 (70.0705) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1558 data: 0.0615 max mem: 3532 +Test: [ 420/2387] eta: 0:06:56 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1539 (0.5000) acc1: 99.2188 (70.7690) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1393 data: 0.0442 max mem: 3532 +Test: [ 430/2387] eta: 0:06:53 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1594 (0.4923) acc1: 99.4792 (71.4291) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1590 data: 0.0654 max mem: 3532 +Test: [ 440/2387] eta: 0:06:50 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1613 (0.4838) acc1: 99.4792 (72.0687) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1885 data: 0.0962 max mem: 3533 +Test: [ 450/2387] eta: 0:06:45 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.0779 (0.4748) acc1: 100.0000 (72.6880) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1719 data: 0.0794 max mem: 3533 +Test: [ 460/2387] eta: 0:06:41 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.0770 (0.4662) acc1: 100.0000 (73.2805) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1537 data: 0.0625 max mem: 3533 +Test: [ 470/2387] eta: 0:06:37 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.0772 (0.4579) acc1: 100.0000 (73.8478) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1605 data: 0.0686 max mem: 3533 +Test: [ 480/2387] eta: 0:06:34 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.0782 (0.4500) acc1: 100.0000 (74.3915) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1732 data: 0.0798 max mem: 3533 +Test: [ 490/2387] eta: 0:06:30 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.0781 (0.4424) acc1: 100.0000 (74.9130) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0828 max mem: 3533 +Test: [ 500/2387] eta: 0:06:27 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.0781 (0.4352) acc1: 100.0000 (75.4138) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1717 data: 0.0780 max mem: 3533 +Test: [ 510/2387] eta: 0:06:24 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.0777 (0.4283) acc1: 100.0000 (75.8949) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1733 data: 0.0781 max mem: 3533 +Test: [ 520/2387] eta: 0:06:20 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1028 (0.4222) acc1: 100.0000 (76.3576) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1637 data: 0.0696 max mem: 3533 +Test: [ 530/2387] eta: 0:06:17 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1002 (0.4161) acc1: 100.0000 (76.8028) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1596 data: 0.0663 max mem: 3533 +Test: [ 540/2387] eta: 0:06:13 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.0996 (0.4103) acc1: 100.0000 (77.2316) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1616 data: 0.0701 max mem: 3533 +Test: [ 550/2387] eta: 0:06:09 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.0995 (0.4046) acc1: 100.0000 (77.6448) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1564 data: 0.0655 max mem: 3533 +Test: [ 560/2387] eta: 0:06:06 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1000 (0.3992) acc1: 100.0000 (78.0433) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1594 data: 0.0677 max mem: 3533 +Test: [ 570/2387] eta: 0:06:03 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.0997 (0.3940) acc1: 100.0000 (78.4278) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1682 data: 0.0756 max mem: 3533 +Test: [ 580/2387] eta: 0:06:00 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.0995 (0.3889) acc1: 100.0000 (78.7991) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1738 data: 0.0817 max mem: 3533 +Test: [ 590/2387] eta: 0:05:57 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1000 (0.3840) acc1: 100.0000 (79.1579) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1707 data: 0.0790 max mem: 3533 +Test: [ 600/2387] eta: 0:05:55 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1036 (0.3795) acc1: 100.0000 (79.5046) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1805 data: 0.0862 max mem: 3533 +Test: [ 610/2387] eta: 0:05:52 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1224 (0.3757) acc1: 100.0000 (79.8371) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1809 data: 0.0860 max mem: 3533 +Test: [ 620/2387] eta: 0:05:50 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1457 (0.3720) acc1: 99.7396 (80.1572) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1768 data: 0.0829 max mem: 3533 +Test: [ 630/2387] eta: 0:05:48 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1467 (0.3685) acc1: 99.7396 (80.4683) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1840 data: 0.0908 max mem: 3533 +Test: [ 640/2387] eta: 0:05:46 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1484 (0.3651) acc1: 99.7396 (80.7657) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2052 data: 0.1139 max mem: 3533 +Test: [ 650/2387] eta: 0:05:43 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1498 (0.3617) acc1: 99.4792 (81.0544) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1792 data: 0.0863 max mem: 3533 +Test: [ 660/2387] eta: 0:05:41 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1474 (0.3585) acc1: 99.7396 (81.3367) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1646 data: 0.0683 max mem: 3533 +Test: [ 670/2387] eta: 0:05:40 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1470 (0.3554) acc1: 99.7396 (81.6113) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2300 data: 0.1331 max mem: 3534 +Test: [ 680/2387] eta: 0:05:38 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1450 (0.3522) acc1: 100.0000 (81.8798) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2250 data: 0.1320 max mem: 3534 +Test: [ 690/2387] eta: 0:05:34 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1478 (0.3527) acc1: 99.4792 (81.8782) acc5: 100.0000 (100.0000) acc1_neg: 71.094 acc5_neg: 100.000 time: 0.1431 data: 0.0483 max mem: 3534 +Test: [ 700/2387] eta: 0:05:30 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.4677 (0.3544) acc1: 76.3021 (81.7950) acc5: 100.0000 (100.0000) acc1_neg: 78.646 acc5_neg: 100.000 time: 0.1149 data: 0.0174 max mem: 3534 +Test: [ 710/2387] eta: 0:05:28 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.4685 (0.3561) acc1: 76.0417 (81.7111) acc5: 100.0000 (100.0000) acc1_neg: 76.042 acc5_neg: 100.000 time: 0.1616 data: 0.0682 max mem: 3534 +Test: [ 720/2387] eta: 0:05:25 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.4666 (0.3576) acc1: 76.0417 (81.6433) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.1799 data: 0.0883 max mem: 3534 +Test: [ 730/2387] eta: 0:05:23 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.4718 (0.3593) acc1: 76.5625 (81.5685) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.1732 data: 0.0805 max mem: 3534 +Test: [ 740/2387] eta: 0:05:21 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.4718 (0.3608) acc1: 75.7812 (81.4957) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.1857 data: 0.0912 max mem: 3534 +Test: [ 750/2387] eta: 0:05:19 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.4670 (0.3621) acc1: 76.8229 (81.4421) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.1818 data: 0.0886 max mem: 3534 +Test: [ 760/2387] eta: 0:05:16 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.4593 (0.3627) acc1: 77.6042 (81.4399) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1713 data: 0.0780 max mem: 3534 +Test: [ 770/2387] eta: 0:05:13 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.3008 (0.3609) acc1: 88.8021 (81.5942) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1634 data: 0.0702 max mem: 3534 +Test: [ 780/2387] eta: 0:05:11 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1322 (0.3578) acc1: 98.1771 (81.8249) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1601 data: 0.0689 max mem: 3534 +Test: [ 790/2387] eta: 0:05:08 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1215 (0.3551) acc1: 99.7396 (82.0105) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1630 data: 0.0714 max mem: 3534 +Test: [ 800/2387] eta: 0:05:06 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1215 (0.3521) acc1: 99.7396 (82.2318) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1649 data: 0.0724 max mem: 3534 +Test: [ 810/2387] eta: 0:05:03 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1238 (0.3493) acc1: 100.0000 (82.4468) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1684 data: 0.0757 max mem: 3534 +Test: [ 820/2387] eta: 0:05:02 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1280 (0.3468) acc1: 99.7396 (82.6476) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1925 data: 0.1011 max mem: 3534 +Test: [ 830/2387] eta: 0:05:00 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1564 (0.3450) acc1: 98.6979 (82.8216) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1948 data: 0.1034 max mem: 3534 +Test: [ 840/2387] eta: 0:04:58 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1791 (0.3433) acc1: 98.1771 (83.0067) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1771 data: 0.0859 max mem: 3534 +Test: [ 850/2387] eta: 0:04:55 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1984 (0.3417) acc1: 98.4375 (83.1751) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1717 data: 0.0809 max mem: 3534 +Test: [ 860/2387] eta: 0:04:54 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1912 (0.3398) acc1: 98.9583 (83.3600) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1886 data: 0.0967 max mem: 3534 +Test: [ 870/2387] eta: 0:04:51 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1832 (0.3381) acc1: 99.4792 (83.5393) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1861 data: 0.0932 max mem: 3534 +Test: [ 880/2387] eta: 0:04:49 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1832 (0.3364) acc1: 99.4792 (83.7067) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1620 data: 0.0695 max mem: 3534 +Test: [ 890/2387] eta: 0:04:47 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1678 (0.3345) acc1: 99.4792 (83.8843) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1802 data: 0.0889 max mem: 3534 +Test: [ 900/2387] eta: 0:04:44 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1841 (0.3366) acc1: 97.6562 (83.7773) acc5: 100.0000 (100.0000) acc1_neg: 69.792 acc5_neg: 100.000 time: 0.1824 data: 0.0898 max mem: 3535 +Test: [ 910/2387] eta: 0:04:43 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.5946 (0.3394) acc1: 67.7083 (83.6018) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.1885 data: 0.0947 max mem: 3535 +Test: [ 920/2387] eta: 0:04:40 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.6091 (0.3423) acc1: 67.4479 (83.4224) acc5: 100.0000 (100.0000) acc1_neg: 64.583 acc5_neg: 100.000 time: 0.1856 data: 0.0927 max mem: 3535 +Test: [ 930/2387] eta: 0:04:38 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.6101 (0.3452) acc1: 67.4479 (83.2533) acc5: 100.0000 (100.0000) acc1_neg: 63.281 acc5_neg: 100.000 time: 0.1717 data: 0.0797 max mem: 3535 +Test: [ 940/2387] eta: 0:04:36 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.6168 (0.3479) acc1: 66.4062 (83.0796) acc5: 100.0000 (100.0000) acc1_neg: 64.323 acc5_neg: 100.000 time: 0.1787 data: 0.0849 max mem: 3535 +Test: [ 950/2387] eta: 0:04:34 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.5923 (0.3505) acc1: 66.9271 (82.9226) acc5: 100.0000 (100.0000) acc1_neg: 66.927 acc5_neg: 100.000 time: 0.1762 data: 0.0828 max mem: 3535 +Test: [ 960/2387] eta: 0:04:32 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.5889 (0.3530) acc1: 68.4896 (82.7756) acc5: 100.0000 (100.0000) acc1_neg: 69.271 acc5_neg: 100.000 time: 0.1715 data: 0.0803 max mem: 3535 +Test: [ 970/2387] eta: 0:04:29 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.5977 (0.3557) acc1: 68.2292 (82.6095) acc5: 100.0000 (100.0000) acc1_neg: 68.229 acc5_neg: 100.000 time: 0.1638 data: 0.0724 max mem: 3535 +Test: [ 980/2387] eta: 0:04:27 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.5824 (0.3555) acc1: 68.2292 (82.6418) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.1683 data: 0.0746 max mem: 3535 +Test: [ 990/2387] eta: 0:04:25 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1666 (0.3536) acc1: 98.4375 (82.7762) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1811 data: 0.0876 max mem: 3535 +Test: [1000/2387] eta: 0:04:23 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1294 (0.3515) acc1: 99.4792 (82.9293) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1862 data: 0.0945 max mem: 3535 +Test: [1010/2387] eta: 0:04:22 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1415 (0.3496) acc1: 99.4792 (83.0786) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1929 data: 0.0987 max mem: 3535 +Test: [1020/2387] eta: 0:04:19 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1539 (0.3480) acc1: 98.6979 (83.2150) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1786 data: 0.0829 max mem: 3535 +Test: [1030/2387] eta: 0:04:17 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1432 (0.3460) acc1: 99.4792 (83.3692) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1771 data: 0.0838 max mem: 3535 +Test: [1040/2387] eta: 0:04:15 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1268 (0.3438) acc1: 100.0000 (83.5277) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1861 data: 0.0951 max mem: 3535 +Test: [1050/2387] eta: 0:04:14 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1229 (0.3417) acc1: 100.0000 (83.6810) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1934 data: 0.1027 max mem: 3535 +Test: [1060/2387] eta: 0:04:12 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1248 (0.3397) acc1: 100.0000 (83.8316) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1866 data: 0.0951 max mem: 3535 +Test: [1070/2387] eta: 0:04:09 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1255 (0.3380) acc1: 99.7396 (83.9736) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1692 data: 0.0760 max mem: 3535 +Test: [1080/2387] eta: 0:04:07 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1695 (0.3366) acc1: 98.4375 (84.0941) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1779 data: 0.0849 max mem: 3535 +Test: [1090/2387] eta: 0:04:06 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1681 (0.3349) acc1: 98.4375 (84.2296) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1899 data: 0.0960 max mem: 3535 +Test: [1100/2387] eta: 0:04:04 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1471 (0.3333) acc1: 98.9583 (84.3603) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1941 data: 0.0990 max mem: 3535 +Test: [1110/2387] eta: 0:04:02 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1538 (0.3320) acc1: 98.9583 (84.4615) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1849 data: 0.0916 max mem: 3535 +Test: [1120/2387] eta: 0:04:00 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1658 (0.3306) acc1: 98.9583 (84.5871) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1849 data: 0.0934 max mem: 3536 +Test: [1130/2387] eta: 0:03:58 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1577 (0.3291) acc1: 98.9583 (84.7153) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1853 data: 0.0935 max mem: 3536 +Test: [1140/2387] eta: 0:03:56 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1689 (0.3278) acc1: 98.9583 (84.8347) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1956 data: 0.1031 max mem: 3536 +Test: [1150/2387] eta: 0:03:55 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1762 (0.3265) acc1: 99.4792 (84.9508) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2145 data: 0.1188 max mem: 3536 +Test: [1160/2387] eta: 0:03:52 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1348 (0.3249) acc1: 99.4792 (85.0728) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1741 data: 0.0763 max mem: 3536 +Test: [1170/2387] eta: 0:03:51 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1396 (0.3233) acc1: 99.7396 (85.1974) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2118 data: 0.1178 max mem: 3536 +Test: [1180/2387] eta: 0:03:49 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1440 (0.3218) acc1: 99.7396 (85.3161) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2299 data: 0.1368 max mem: 3536 +Test: [1190/2387] eta: 0:03:46 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1260 (0.3202) acc1: 99.7396 (85.4374) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1330 data: 0.0358 max mem: 3536 +Test: [1200/2387] eta: 0:03:44 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1202 (0.3186) acc1: 100.0000 (85.5587) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1327 data: 0.0350 max mem: 3536 +Test: [1210/2387] eta: 0:03:42 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1140 (0.3168) acc1: 100.0000 (85.6779) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1806 data: 0.0864 max mem: 3536 +Test: [1220/2387] eta: 0:03:40 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.0843 (0.3149) acc1: 100.0000 (85.7942) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1833 data: 0.0910 max mem: 3536 +Test: [1230/2387] eta: 0:03:38 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.0737 (0.3129) acc1: 100.0000 (85.9094) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1754 data: 0.0840 max mem: 3536 +Test: [1240/2387] eta: 0:03:36 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.0719 (0.3110) acc1: 100.0000 (86.0229) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1746 data: 0.0840 max mem: 3536 +Test: [1250/2387] eta: 0:03:34 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.0722 (0.3091) acc1: 100.0000 (86.1342) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1787 data: 0.0873 max mem: 3536 +Test: [1260/2387] eta: 0:03:32 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.0719 (0.3072) acc1: 100.0000 (86.2442) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1904 data: 0.0985 max mem: 3536 +Test: [1270/2387] eta: 0:03:31 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.0707 (0.3054) acc1: 100.0000 (86.3522) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1909 data: 0.0996 max mem: 3536 +Test: [1280/2387] eta: 0:03:29 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.0817 (0.3037) acc1: 100.0000 (86.4567) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1858 data: 0.0947 max mem: 3536 +Test: [1290/2387] eta: 0:03:27 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.0992 (0.3022) acc1: 99.7396 (86.5475) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1871 data: 0.0959 max mem: 3536 +Test: [1300/2387] eta: 0:03:25 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.0992 (0.3007) acc1: 99.7396 (86.6449) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1876 data: 0.0956 max mem: 3536 +Test: [1310/2387] eta: 0:03:23 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1049 (0.2993) acc1: 100.0000 (86.7396) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1894 data: 0.0973 max mem: 3536 +Test: [1320/2387] eta: 0:03:21 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1112 (0.2979) acc1: 99.7396 (86.8305) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1979 data: 0.1046 max mem: 3536 +Test: [1330/2387] eta: 0:03:19 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1200 (0.2967) acc1: 99.7396 (86.9226) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2086 data: 0.1147 max mem: 3536 +Test: [1340/2387] eta: 0:03:18 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1575 (0.2964) acc1: 98.6979 (86.9574) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2024 data: 0.1108 max mem: 3536 +Test: [1350/2387] eta: 0:03:16 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.0552 (0.2946) acc1: 100.0000 (87.0540) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1991 data: 0.1075 max mem: 3537 +Test: [1360/2387] eta: 0:03:14 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.0493 (0.2928) acc1: 100.0000 (87.1491) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2167 data: 0.1247 max mem: 3537 +Test: [1370/2387] eta: 0:03:12 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.0452 (0.2910) acc1: 100.0000 (87.2428) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2014 data: 0.1093 max mem: 3537 +Test: [1380/2387] eta: 0:03:10 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.0450 (0.2892) acc1: 100.0000 (87.3352) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1926 data: 0.0989 max mem: 3537 +Test: [1390/2387] eta: 0:03:09 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.0482 (0.2875) acc1: 100.0000 (87.4262) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2015 data: 0.1083 max mem: 3537 +Test: [1400/2387] eta: 0:03:07 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.0479 (0.2857) acc1: 100.0000 (87.5160) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2017 data: 0.1101 max mem: 3537 +Test: [1410/2387] eta: 0:03:05 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.0516 (0.2847) acc1: 100.0000 (87.5899) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2013 data: 0.1093 max mem: 3537 +Test: [1420/2387] eta: 0:03:03 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1555 (0.2840) acc1: 98.4375 (87.6484) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2010 data: 0.1070 max mem: 3537 +Test: [1430/2387] eta: 0:03:02 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1700 (0.2833) acc1: 97.9167 (87.7173) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2360 data: 0.1404 max mem: 3537 +Test: [1440/2387] eta: 0:03:00 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1703 (0.2825) acc1: 97.9167 (87.7948) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2023 data: 0.1077 max mem: 3537 +Test: [1450/2387] eta: 0:02:58 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1703 (0.2818) acc1: 98.9583 (87.8654) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1888 data: 0.0946 max mem: 3537 +Test: [1460/2387] eta: 0:02:56 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1421 (0.2808) acc1: 98.9583 (87.9428) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2147 data: 0.1185 max mem: 3537 +Test: [1470/2387] eta: 0:02:54 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1087 (0.2796) acc1: 99.7396 (88.0244) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1779 data: 0.0833 max mem: 3537 +Test: [1480/2387] eta: 0:02:52 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1074 (0.2784) acc1: 100.0000 (88.1021) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1599 data: 0.0662 max mem: 3537 +Test: [1490/2387] eta: 0:02:50 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1105 (0.2774) acc1: 99.7396 (88.1805) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1537 data: 0.0606 max mem: 3537 +Test: [1500/2387] eta: 0:02:48 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1152 (0.2763) acc1: 99.7396 (88.2568) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1680 data: 0.0748 max mem: 3537 +Test: [1510/2387] eta: 0:02:46 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1180 (0.2753) acc1: 100.0000 (88.3343) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1733 data: 0.0783 max mem: 3537 +Test: [1520/2387] eta: 0:02:44 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1333 (0.2746) acc1: 99.7396 (88.4042) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1689 data: 0.0765 max mem: 3537 +Test: [1530/2387] eta: 0:02:42 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1780 (0.2740) acc1: 99.2188 (88.4706) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1797 data: 0.0880 max mem: 3537 +Test: [1540/2387] eta: 0:02:40 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1596 (0.2733) acc1: 99.4792 (88.5393) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1778 data: 0.0850 max mem: 3537 +Test: [1550/2387] eta: 0:02:38 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1574 (0.2726) acc1: 99.7396 (88.6088) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1871 data: 0.0903 max mem: 3537 +Test: [1560/2387] eta: 0:02:36 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1576 (0.2719) acc1: 99.7396 (88.6755) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2018 data: 0.1047 max mem: 3537 +Test: [1570/2387] eta: 0:02:34 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1461 (0.2711) acc1: 99.7396 (88.7447) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1888 data: 0.0946 max mem: 3537 +Test: [1580/2387] eta: 0:02:32 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1379 (0.2702) acc1: 99.7396 (88.8115) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1905 data: 0.0976 max mem: 3538 +Test: [1590/2387] eta: 0:02:31 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1298 (0.2693) acc1: 99.2188 (88.8797) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1988 data: 0.1062 max mem: 3538 +Test: [1600/2387] eta: 0:02:29 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1103 (0.2685) acc1: 99.4792 (88.9246) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1974 data: 0.0997 max mem: 3538 +Test: [1610/2387] eta: 0:02:27 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1132 (0.2676) acc1: 99.7396 (88.9922) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1919 data: 0.0927 max mem: 3538 +Test: [1620/2387] eta: 0:02:25 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1243 (0.2668) acc1: 99.7396 (89.0533) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1862 data: 0.0922 max mem: 3538 +Test: [1630/2387] eta: 0:02:23 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1505 (0.2662) acc1: 98.9583 (89.1106) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1885 data: 0.0951 max mem: 3538 +Test: [1640/2387] eta: 0:02:21 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1750 (0.2657) acc1: 98.1771 (89.1587) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1835 data: 0.0885 max mem: 3538 +Test: [1650/2387] eta: 0:02:19 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1750 (0.2652) acc1: 97.3958 (89.2095) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1823 data: 0.0884 max mem: 3538 +Test: [1660/2387] eta: 0:02:17 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1779 (0.2647) acc1: 98.1771 (89.2590) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1855 data: 0.0930 max mem: 3538 +Test: [1670/2387] eta: 0:02:15 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1784 (0.2642) acc1: 98.1771 (89.3103) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1792 data: 0.0875 max mem: 3538 +Test: [1680/2387] eta: 0:02:13 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1686 (0.2638) acc1: 98.1771 (89.3559) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1793 data: 0.0875 max mem: 3538 +Test: [1690/2387] eta: 0:02:11 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1716 (0.2633) acc1: 98.1771 (89.4028) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1812 data: 0.0889 max mem: 3538 +Test: [1700/2387] eta: 0:02:10 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1747 (0.2627) acc1: 98.1771 (89.4564) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0887 max mem: 3538 +Test: [1710/2387] eta: 0:02:08 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1005 (0.2617) acc1: 99.7396 (89.5173) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1813 data: 0.0856 max mem: 3538 +Test: [1720/2387] eta: 0:02:06 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.0879 (0.2606) acc1: 100.0000 (89.5777) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1804 data: 0.0887 max mem: 3538 +Test: [1730/2387] eta: 0:02:04 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.0843 (0.2596) acc1: 100.0000 (89.6373) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1755 data: 0.0822 max mem: 3538 +Test: [1740/2387] eta: 0:02:02 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.0841 (0.2586) acc1: 100.0000 (89.6961) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1763 data: 0.0787 max mem: 3538 +Test: [1750/2387] eta: 0:02:00 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.0808 (0.2577) acc1: 100.0000 (89.7548) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1753 data: 0.0796 max mem: 3538 +Test: [1760/2387] eta: 0:01:58 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.0829 (0.2567) acc1: 100.0000 (89.8128) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1649 data: 0.0725 max mem: 3538 +Test: [1770/2387] eta: 0:01:56 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1012 (0.2559) acc1: 100.0000 (89.8661) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1961 data: 0.1025 max mem: 3538 +Test: [1780/2387] eta: 0:01:54 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1081 (0.2550) acc1: 100.0000 (89.9224) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1905 data: 0.0981 max mem: 3538 +Test: [1790/2387] eta: 0:01:52 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1051 (0.2542) acc1: 100.0000 (89.9775) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1579 data: 0.0647 max mem: 3538 +Test: [1800/2387] eta: 0:01:50 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1019 (0.2533) acc1: 99.7396 (90.0317) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1612 data: 0.0678 max mem: 3538 +Test: [1810/2387] eta: 0:01:48 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.0980 (0.2524) acc1: 100.0000 (90.0862) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1677 data: 0.0746 max mem: 3539 +Test: [1820/2387] eta: 0:01:46 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.0869 (0.2515) acc1: 100.0000 (90.1405) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1772 data: 0.0835 max mem: 3539 +Test: [1830/2387] eta: 0:01:44 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.0961 (0.2508) acc1: 100.0000 (90.1942) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1823 data: 0.0901 max mem: 3539 +Test: [1840/2387] eta: 0:01:42 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1368 (0.2501) acc1: 100.0000 (90.2475) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1795 data: 0.0884 max mem: 3539 +Test: [1850/2387] eta: 0:01:41 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1378 (0.2495) acc1: 100.0000 (90.3000) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1916 data: 0.1010 max mem: 3539 +Test: [1860/2387] eta: 0:01:39 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1373 (0.2489) acc1: 100.0000 (90.3521) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2041 data: 0.1119 max mem: 3539 +Test: [1870/2387] eta: 0:01:37 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1371 (0.2483) acc1: 100.0000 (90.4036) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2013 data: 0.1083 max mem: 3539 +Test: [1880/2387] eta: 0:01:35 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1371 (0.2478) acc1: 100.0000 (90.4546) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1956 data: 0.1035 max mem: 3539 +Test: [1890/2387] eta: 0:01:33 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1367 (0.2472) acc1: 100.0000 (90.5051) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1969 data: 0.1039 max mem: 3539 +Test: [1900/2387] eta: 0:01:31 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1385 (0.2466) acc1: 100.0000 (90.5550) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2018 data: 0.1088 max mem: 3539 +Test: [1910/2387] eta: 0:01:30 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1554 (0.2462) acc1: 100.0000 (90.6032) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2244 data: 0.1315 max mem: 3539 +Test: [1920/2387] eta: 0:01:28 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1594 (0.2459) acc1: 100.0000 (90.6314) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2072 data: 0.1140 max mem: 3539 +Test: [1930/2387] eta: 0:01:26 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1305 (0.2452) acc1: 100.0000 (90.6795) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1676 data: 0.0740 max mem: 3539 +Test: [1940/2387] eta: 0:01:24 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1363 (0.2450) acc1: 100.0000 (90.6905) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1623 data: 0.0696 max mem: 3539 +Test: [1950/2387] eta: 0:01:22 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1571 (0.2446) acc1: 99.2188 (90.7307) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1722 data: 0.0800 max mem: 3539 +Test: [1960/2387] eta: 0:01:20 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1571 (0.2442) acc1: 100.0000 (90.7674) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1775 data: 0.0817 max mem: 3539 +Test: [1970/2387] eta: 0:01:18 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1865 (0.2443) acc1: 96.6146 (90.7641) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1749 data: 0.0796 max mem: 3539 +Test: [1980/2387] eta: 0:01:16 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.2017 (0.2441) acc1: 96.6146 (90.7964) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1706 data: 0.0787 max mem: 3539 +Test: [1990/2387] eta: 0:01:14 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1614 (0.2437) acc1: 99.2188 (90.8387) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1675 data: 0.0750 max mem: 3539 +Test: [2000/2387] eta: 0:01:12 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1617 (0.2433) acc1: 99.2188 (90.8806) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1692 data: 0.0747 max mem: 3539 +Test: [2010/2387] eta: 0:01:10 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1621 (0.2429) acc1: 99.2188 (90.9223) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1682 data: 0.0740 max mem: 3539 +Test: [2020/2387] eta: 0:01:08 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1608 (0.2425) acc1: 99.2188 (90.9634) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1730 data: 0.0808 max mem: 3539 +Test: [2030/2387] eta: 0:01:07 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1617 (0.2421) acc1: 99.2188 (91.0042) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1804 data: 0.0882 max mem: 3540 +Test: [2040/2387] eta: 0:01:05 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1617 (0.2417) acc1: 99.2188 (91.0452) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1989 data: 0.1043 max mem: 3540 +Test: [2050/2387] eta: 0:01:03 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1602 (0.2413) acc1: 99.2188 (91.0859) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1885 data: 0.0946 max mem: 3540 +Test: [2060/2387] eta: 0:01:01 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1636 (0.2409) acc1: 99.2188 (91.1252) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1693 data: 0.0772 max mem: 3540 +Test: [2070/2387] eta: 0:00:59 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1614 (0.2405) acc1: 99.2188 (91.1648) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1719 data: 0.0796 max mem: 3540 +Test: [2080/2387] eta: 0:00:57 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1612 (0.2402) acc1: 99.2188 (91.2036) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1743 data: 0.0821 max mem: 3540 +Test: [2090/2387] eta: 0:00:55 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1621 (0.2398) acc1: 99.2188 (91.2429) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1798 data: 0.0847 max mem: 3540 +Test: [2100/2387] eta: 0:00:53 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1618 (0.2394) acc1: 99.4792 (91.2814) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1840 data: 0.0873 max mem: 3540 +Test: [2110/2387] eta: 0:00:51 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1623 (0.2391) acc1: 99.2188 (91.3190) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1830 data: 0.0878 max mem: 3540 +Test: [2120/2387] eta: 0:00:50 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1627 (0.2387) acc1: 98.9583 (91.3510) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1999 data: 0.1060 max mem: 3540 +Test: [2130/2387] eta: 0:00:48 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1357 (0.2382) acc1: 98.9583 (91.3879) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0941 max mem: 3540 +Test: [2140/2387] eta: 0:00:46 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1352 (0.2379) acc1: 99.2188 (91.4076) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1323 data: 0.0339 max mem: 3540 +Test: [2150/2387] eta: 0:00:44 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1490 (0.2375) acc1: 99.2188 (91.4438) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1403 data: 0.0461 max mem: 3540 +Test: [2160/2387] eta: 0:00:42 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1484 (0.2371) acc1: 99.2188 (91.4777) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1698 data: 0.0779 max mem: 3540 +Test: [2170/2387] eta: 0:00:40 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1590 (0.2369) acc1: 98.6979 (91.5047) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1805 data: 0.0875 max mem: 3540 +Test: [2180/2387] eta: 0:00:38 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.2136 (0.2377) acc1: 95.5729 (91.4643) acc5: 100.0000 (100.0000) acc1_neg: 61.979 acc5_neg: 100.000 time: 0.1976 data: 0.1021 max mem: 3540 +Test: [2190/2387] eta: 0:00:36 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.2372 (0.2379) acc1: 94.5312 (91.4587) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1702 data: 0.0739 max mem: 3540 +Test: [2200/2387] eta: 0:00:34 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1475 (0.2374) acc1: 98.4375 (91.4937) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1602 data: 0.0669 max mem: 3540 +Test: [2210/2387] eta: 0:00:33 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1248 (0.2369) acc1: 99.7396 (91.5316) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1760 data: 0.0831 max mem: 3540 +Test: [2220/2387] eta: 0:00:31 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1248 (0.2364) acc1: 100.0000 (91.5669) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1760 data: 0.0837 max mem: 3540 +Test: [2230/2387] eta: 0:00:29 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1275 (0.2359) acc1: 100.0000 (91.6038) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1775 data: 0.0854 max mem: 3540 +Test: [2240/2387] eta: 0:00:27 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1205 (0.2354) acc1: 99.7396 (91.6383) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1867 data: 0.0946 max mem: 3540 +Test: [2250/2387] eta: 0:00:25 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1312 (0.2350) acc1: 99.7396 (91.6725) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1979 data: 0.1063 max mem: 3540 +Test: [2260/2387] eta: 0:00:23 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1370 (0.2346) acc1: 100.0000 (91.7076) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1969 data: 0.1057 max mem: 3541 +Test: [2270/2387] eta: 0:00:21 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1181 (0.2340) acc1: 100.0000 (91.7434) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1974 data: 0.1060 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1091 (0.2336) acc1: 100.0000 (91.7627) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2124 data: 0.1199 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1123 (0.2331) acc1: 100.0000 (91.7982) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2044 data: 0.1116 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1173 (0.2326) acc1: 100.0000 (91.8310) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1934 data: 0.1011 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1269 (0.2323) acc1: 100.0000 (91.8622) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2189 data: 0.1271 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1545 (0.2321) acc1: 99.2188 (91.8857) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1887 data: 0.0961 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1765 (0.2319) acc1: 97.9167 (91.9064) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1414 data: 0.0476 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1732 (0.2317) acc1: 98.1771 (91.9313) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1471 data: 0.0550 max mem: 3541 +Test: [2350/2387] eta: 0:00:06 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1734 (0.2317) acc1: 98.9583 (91.9379) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1544 data: 0.0639 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.1930 (0.2316) acc1: 98.4375 (91.9564) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1528 data: 0.0624 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.2023 (0.2316) acc1: 97.3958 (91.9700) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.1423 data: 0.0523 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.2197 (0.2316) acc1: 97.3958 (91.9838) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1361 data: 0.0454 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 23.744 acc5_pos: 100.000 loss: 0.2659 (0.2319) acc1: 92.9688 (91.9817) acc5: 100.0000 (100.0000) acc1_neg: 93.023 acc5_neg: 100.000 time: 0.1342 data: 0.0430 max mem: 3541 +Test: Total time: 0:07:24 (0.1862 s / it) +* Overall Acc@1 91.982 Acc@5 100.000 + target == real Acc@1 22.730 Acc@5 100.000 + target == fake Acc@1 96.845 Acc@5 100.000 + loss 0.232 + PR-AUC 0.9927 ROC-AUC 0.9034 EER 16.61% (thr=0.7876) +Accuracy of the network on the 916267 test images: 92.0% +[rank0]:[W725 18:48:50.557508282 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:49:10.563000 608978 torch/distributed/run.py:766] +W0725 18:49:10.563000 608978 torch/distributed/run.py:766] ***************************************** +W0725 18:49:10.563000 608978 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:49:10.563000 608978 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:49:40.841371689 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:49:41.464379507 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:49:41.507376826 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:49:41.637764399 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:49:41.876881650 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +[rank2]:[W725 18:49:41.906878833 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:49:41.916289675 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:49:41.964564448 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed3_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed3_resampled_n116-040-213_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:04:29 acc1_pos: 32.031 acc5_pos: 100.000 loss: 1.0122 (1.0122) acc1: 32.0312 (32.0312) acc5: 100.0000 (100.0000) time: 7.4854 data: 5.8209 max mem: 3531 +Test: [ 10/517] eta: 0:06:29 acc1_pos: 9.115 acc5_pos: 100.000 loss: 1.1581 (1.1258) acc1: 19.7917 (22.7746) acc5: 100.0000 (100.0000) time: 0.7678 data: 0.5296 max mem: 3531 +Test: [ 20/517] eta: 0:03:43 acc1_pos: 18.750 acc5_pos: 100.000 loss: 1.1143 (1.1141) acc1: 19.7917 (23.0903) acc5: 100.0000 (100.0000) time: 0.0976 data: 0.0005 max mem: 3531 +Test: [ 30/517] eta: 0:03:15 acc1_pos: 7.031 acc5_pos: 100.000 loss: 1.1358 (1.1352) acc1: 18.7500 (21.6482) acc5: 100.0000 (100.0000) time: 0.1996 data: 0.1038 max mem: 3531 +Test: [ 40/517] eta: 0:03:05 acc1_pos: 18.750 acc5_pos: 100.000 loss: 1.1993 (1.1562) acc1: 14.3229 (19.0485) acc5: 100.0000 (100.0000) time: 0.3272 data: 0.2342 max mem: 3531 +Test: [ 50/517] eta: 0:02:54 acc1_pos: 36.697 acc5_pos: 100.000 loss: 1.1880 (1.1094) acc1: 15.8854 (22.8248) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.3317 data: 0.2393 max mem: 3531 +Test: [ 60/517] eta: 0:02:38 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.3518 (0.9811) acc1: 86.9792 (34.4775) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2576 data: 0.1649 max mem: 3531 +Test: [ 70/517] eta: 0:02:31 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2979 (0.8844) acc1: 97.1354 (43.3025) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2534 data: 0.1560 max mem: 3531 +Test: [ 80/517] eta: 0:02:30 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2979 (0.8158) acc1: 96.0938 (49.6110) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.3399 data: 0.2427 max mem: 3531 +Test: [ 90/517] eta: 0:02:20 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2991 (0.7570) acc1: 98.4375 (55.0424) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2946 data: 0.1980 max mem: 3531 +Test: [100/517] eta: 0:02:16 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2466 (0.7057) acc1: 99.2188 (59.4059) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1554 max mem: 3531 +Test: [110/517] eta: 0:02:09 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2235 (0.6622) acc1: 99.7396 (63.0091) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2599 data: 0.1587 max mem: 3531 +Test: [120/517] eta: 0:02:03 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2561 (0.6308) acc1: 98.9583 (65.9435) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2316 data: 0.1313 max mem: 3531 +Test: [130/517] eta: 0:01:58 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2860 (0.6035) acc1: 98.6979 (68.4101) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2499 data: 0.1535 max mem: 3531 +Test: [140/517] eta: 0:01:54 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2276 (0.5759) acc1: 99.7396 (70.6486) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2538 data: 0.1574 max mem: 3531 +Test: [150/517] eta: 0:01:49 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2192 (0.5526) acc1: 100.0000 (72.5631) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2510 data: 0.1540 max mem: 3531 +Test: [160/517] eta: 0:01:45 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2432 (0.5338) acc1: 99.7396 (74.1767) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2462 data: 0.1503 max mem: 3531 +Test: [170/517] eta: 0:01:41 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2443 (0.5163) acc1: 99.2188 (75.6107) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2457 data: 0.1510 max mem: 3531 +Test: [180/517] eta: 0:01:38 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2322 (0.5000) acc1: 99.4792 (76.9021) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2530 data: 0.1569 max mem: 3531 +Test: [190/517] eta: 0:01:34 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.1923 (0.4837) acc1: 99.7396 (78.1046) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2447 data: 0.1467 max mem: 3531 +Test: [200/517] eta: 0:01:30 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.1930 (0.4698) acc1: 100.0000 (79.1874) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2399 data: 0.1437 max mem: 3531 +Test: [210/517] eta: 0:01:27 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2149 (0.4620) acc1: 99.2188 (79.9985) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2482 data: 0.1520 max mem: 3532 +Test: [220/517] eta: 0:01:24 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2978 (0.4549) acc1: 96.0938 (80.7103) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2567 data: 0.1591 max mem: 3532 +Test: [230/517] eta: 0:01:21 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2923 (0.4478) acc1: 97.9167 (81.5025) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2767 data: 0.1809 max mem: 3532 +Test: [240/517] eta: 0:01:18 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2883 (0.4414) acc1: 99.2188 (82.2214) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2702 data: 0.1762 max mem: 3532 +Test: [250/517] eta: 0:01:15 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2677 (0.4338) acc1: 99.4792 (82.8955) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2502 data: 0.1565 max mem: 3532 +Test: [260/517] eta: 0:01:11 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2353 (0.4265) acc1: 99.7396 (83.5399) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2528 data: 0.1588 max mem: 3532 +Test: [270/517] eta: 0:01:08 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2685 (0.4216) acc1: 99.7396 (84.1261) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2492 data: 0.1531 max mem: 3532 +Test: [280/517] eta: 0:01:05 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2820 (0.4163) acc1: 99.4792 (84.6595) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2458 data: 0.1497 max mem: 3532 +Test: [290/517] eta: 0:01:02 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2684 (0.4112) acc1: 99.2188 (85.1670) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2435 data: 0.1482 max mem: 3532 +Test: [300/517] eta: 0:00:59 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2743 (0.4067) acc1: 99.4792 (85.6312) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2407 data: 0.1435 max mem: 3532 +Test: [310/517] eta: 0:00:56 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2794 (0.4030) acc1: 99.4792 (86.0698) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2399 data: 0.1427 max mem: 3532 +Test: [320/517] eta: 0:00:53 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2672 (0.3982) acc1: 99.7396 (86.4875) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2383 data: 0.1417 max mem: 3532 +Test: [330/517] eta: 0:00:50 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2340 (0.3926) acc1: 100.0000 (86.8934) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2449 data: 0.1474 max mem: 3532 +Test: [340/517] eta: 0:00:48 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2054 (0.3882) acc1: 100.0000 (87.2304) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2498 data: 0.1520 max mem: 3532 +Test: [350/517] eta: 0:00:45 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2797 (0.3861) acc1: 98.6979 (87.5260) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2499 data: 0.1494 max mem: 3532 +Test: [360/517] eta: 0:00:42 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2872 (0.3833) acc1: 98.1771 (87.8015) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2453 data: 0.1451 max mem: 3532 +Test: [370/517] eta: 0:00:39 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2333 (0.3784) acc1: 99.4792 (88.1282) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2410 data: 0.1438 max mem: 3532 +Test: [380/517] eta: 0:00:36 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2042 (0.3744) acc1: 100.0000 (88.4221) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2495 data: 0.1509 max mem: 3532 +Test: [390/517] eta: 0:00:34 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2333 (0.3714) acc1: 99.4792 (88.6882) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2455 data: 0.1465 max mem: 3532 +Test: [400/517] eta: 0:00:31 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2482 (0.3679) acc1: 99.2188 (88.9631) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2521 data: 0.1561 max mem: 3532 +Test: [410/517] eta: 0:00:28 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2482 (0.3653) acc1: 100.0000 (89.2291) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2535 data: 0.1597 max mem: 3532 +Test: [420/517] eta: 0:00:25 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2518 (0.3628) acc1: 100.0000 (89.4640) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2377 data: 0.1401 max mem: 3532 +Test: [430/517] eta: 0:00:23 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2526 (0.3606) acc1: 98.4375 (89.6619) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2450 data: 0.1471 max mem: 3532 +Test: [440/517] eta: 0:00:20 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2751 (0.3584) acc1: 98.9583 (89.8821) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2500 data: 0.1567 max mem: 3533 +Test: [450/517] eta: 0:00:17 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2455 (0.3559) acc1: 99.7396 (90.1007) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2498 data: 0.1551 max mem: 3533 +Test: [460/517] eta: 0:00:15 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2559 (0.3541) acc1: 99.7396 (90.2900) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2498 data: 0.1552 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2539 (0.3517) acc1: 99.7396 (90.4918) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2402 data: 0.1480 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2380 (0.3493) acc1: 100.0000 (90.6862) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2365 data: 0.1434 max mem: 3533 +Test: [490/517] eta: 0:00:07 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2485 (0.3473) acc1: 99.7396 (90.8631) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2167 data: 0.1245 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2842 (0.3476) acc1: 99.2188 (90.9530) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1901 data: 0.0998 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.3153 (0.3460) acc1: 98.6979 (91.1122) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1772 data: 0.0870 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2366 (0.3445) acc1: 99.7396 (91.2094) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1780 data: 0.0861 max mem: 3533 +Test: Total time: 0:02:13 (0.2589 s / it) +* Overall Acc@1 91.209 Acc@5 100.000 + target == real Acc@1 19.239 Acc@5 100.000 + target == fake Acc@1 98.676 Acc@5 100.000 + loss 0.345 + PR-AUC 0.9583 ROC-AUC 0.7550 EER 31.18% (thr=0.7529) +Accuracy of the network on the 198427 test images: 91.2% +[rank0]:[W725 18:52:05.040871333 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0725 18:52:26.022000 609570 torch/distributed/run.py:766] +W0725 18:52:26.022000 609570 torch/distributed/run.py:766] ***************************************** +W0725 18:52:26.022000 609570 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:52:26.022000 609570 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:52:57.616468031 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W725 18:52:57.616505116 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:52:57.726212585 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:52:57.125484380 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W725 18:52:57.125484275 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:52:57.166564228 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:52:58.177799377 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:52:58.185775698 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed3_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed3_resampled_n116-040-213_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:30:22 acc1_pos: 30.469 acc5_pos: 100.000 loss: 0.8685 (0.8685) acc1: 30.4688 (30.4688) acc5: 100.0000 (100.0000) time: 4.6851 data: 2.8249 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/389] eta: 0:03:14 acc1_pos: 73.958 acc5_pos: 100.000 loss: 0.7138 (0.7131) acc1: 46.3542 (51.1837) acc5: 100.0000 (100.0000) time: 0.5121 data: 0.2612 max mem: 3531 +Test: [ 20/389] eta: 0:01:56 acc1_pos: 11.719 acc5_pos: 100.000 loss: 0.7138 (0.7725) acc1: 46.3542 (45.3745) acc5: 100.0000 (100.0000) time: 0.0979 data: 0.0079 max mem: 3531 +Test: [ 30/389] eta: 0:01:39 acc1_pos: 48.438 acc5_pos: 100.000 loss: 0.7781 (0.7953) acc1: 35.9375 (41.4735) acc5: 100.0000 (100.0000) time: 0.1466 data: 0.0555 max mem: 3531 +Test: [ 40/389] eta: 0:01:45 acc1_pos: 5.990 acc5_pos: 100.000 loss: 0.7659 (0.7969) acc1: 35.9375 (41.9779) acc5: 100.0000 (100.0000) time: 0.2900 data: 0.1981 max mem: 3531 +Test: [ 50/389] eta: 0:01:35 acc1_pos: 31.771 acc5_pos: 100.000 loss: 0.7482 (0.7967) acc1: 40.1042 (42.1569) acc5: 100.0000 (100.0000) time: 0.2927 data: 0.2009 max mem: 3531 +Test: [ 60/389] eta: 0:01:23 acc1_pos: 27.604 acc5_pos: 100.000 loss: 0.8483 (0.7915) acc1: 30.4688 (42.5461) acc5: 100.0000 (100.0000) time: 0.1488 data: 0.0531 max mem: 3531 +Test: [ 70/389] eta: 0:01:18 acc1_pos: 62.760 acc5_pos: 100.000 loss: 0.7206 (0.7754) acc1: 45.3125 (44.2562) acc5: 100.0000 (100.0000) time: 0.1501 data: 0.0547 max mem: 3531 +Test: [ 80/389] eta: 0:01:14 acc1_pos: 86.719 acc5_pos: 100.000 loss: 0.7206 (0.7674) acc1: 50.7812 (45.5311) acc5: 100.0000 (100.0000) time: 0.2017 data: 0.1106 max mem: 3531 +Test: [ 90/389] eta: 0:01:10 acc1_pos: 60.938 acc5_pos: 100.000 loss: 0.4204 (0.7314) acc1: 80.7292 (49.3790) acc5: 100.0000 (100.0000) time: 0.2002 data: 0.1076 max mem: 3531 +Test: [100/389] eta: 0:01:07 acc1_pos: 10.156 acc5_pos: 100.000 loss: 0.6669 (0.7483) acc1: 51.0417 (47.7259) acc5: 100.0000 (100.0000) time: 0.2081 data: 0.1096 max mem: 3531 +Test: [110/389] eta: 0:01:03 acc1_pos: 98.958 acc5_pos: 100.000 loss: 0.9026 (0.7622) acc1: 40.1042 (47.1894) acc5: 100.0000 (100.0000) time: 0.1944 data: 0.0946 max mem: 3531 +Test: [120/389] eta: 0:01:01 acc1_pos: 62.240 acc5_pos: 100.000 loss: 0.6980 (0.7560) acc1: 57.0312 (48.1211) acc5: 100.0000 (100.0000) time: 0.1949 data: 0.0997 max mem: 3531 +Test: [130/389] eta: 0:00:58 acc1_pos: 42.945 acc5_pos: 100.000 loss: 0.6336 (0.7430) acc1: 62.2396 (49.6402) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.2151 data: 0.1197 max mem: 3531 +Test: [140/389] eta: 0:00:55 acc1_pos: 42.945 acc5_pos: 100.000 loss: 0.4540 (0.7241) acc1: 78.9062 (51.8580) acc5: 100.0000 (100.0000) acc1_neg: 50.781 acc5_neg: 100.000 time: 0.1899 data: 0.0933 max mem: 3531 +Test: [150/389] eta: 0:00:53 acc1_pos: 42.945 acc5_pos: 100.000 loss: 0.4540 (0.7143) acc1: 84.6354 (53.1940) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2002 data: 0.1059 max mem: 3531 +Test: [160/389] eta: 0:00:49 acc1_pos: 42.945 acc5_pos: 100.000 loss: 0.5135 (0.6954) acc1: 82.8125 (55.0304) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1876 data: 0.0933 max mem: 3531 +Test: [170/389] eta: 0:00:46 acc1_pos: 42.945 acc5_pos: 100.000 loss: 0.2625 (0.6782) acc1: 99.4792 (56.8150) acc5: 100.0000 (100.0000) acc1_neg: 64.062 acc5_neg: 100.000 time: 0.1504 data: 0.0566 max mem: 3531 +Test: [180/389] eta: 0:00:44 acc1_pos: 42.945 acc5_pos: 100.000 loss: 0.3065 (0.6645) acc1: 93.2292 (58.4657) acc5: 100.0000 (100.0000) acc1_neg: 69.531 acc5_neg: 100.000 time: 0.1735 data: 0.0819 max mem: 3531 +Test: [190/389] eta: 0:00:41 acc1_pos: 42.945 acc5_pos: 100.000 loss: 0.4619 (0.6602) acc1: 82.0312 (58.9687) acc5: 100.0000 (100.0000) acc1_neg: 46.875 acc5_neg: 100.000 time: 0.1739 data: 0.0820 max mem: 3531 +Test: [200/389] eta: 0:00:39 acc1_pos: 42.945 acc5_pos: 100.000 loss: 0.5059 (0.6545) acc1: 75.0000 (59.7585) acc5: 100.0000 (100.0000) acc1_neg: 71.354 acc5_neg: 100.000 time: 0.1660 data: 0.0744 max mem: 3531 +Test: [210/389] eta: 0:00:37 acc1_pos: 42.945 acc5_pos: 100.000 loss: 0.3259 (0.6400) acc1: 85.4167 (61.0844) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1827 data: 0.0912 max mem: 3532 +Test: [220/389] eta: 0:00:34 acc1_pos: 42.945 acc5_pos: 100.000 loss: 0.2537 (0.6211) acc1: 100.0000 (62.8052) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1835 data: 0.0900 max mem: 3532 +Test: [230/389] eta: 0:00:32 acc1_pos: 42.945 acc5_pos: 100.000 loss: 0.3774 (0.6152) acc1: 97.1354 (63.5789) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1837 data: 0.0901 max mem: 3532 +Test: [240/389] eta: 0:00:30 acc1_pos: 42.945 acc5_pos: 100.000 loss: 0.4339 (0.6072) acc1: 89.5833 (64.6568) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1841 data: 0.0914 max mem: 3532 +Test: [250/389] eta: 0:00:28 acc1_pos: 42.945 acc5_pos: 100.000 loss: 0.3742 (0.5960) acc1: 93.2292 (65.9477) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1835 data: 0.0876 max mem: 3532 +Test: [260/389] eta: 0:00:26 acc1_pos: 42.945 acc5_pos: 100.000 loss: 0.3614 (0.5902) acc1: 98.9583 (66.7395) acc5: 100.0000 (100.0000) acc1_neg: 41.927 acc5_neg: 100.000 time: 0.1899 data: 0.0947 max mem: 3532 +Test: [270/389] eta: 0:00:24 acc1_pos: 42.945 acc5_pos: 100.000 loss: 0.4914 (0.5921) acc1: 86.1979 (66.5389) acc5: 100.0000 (100.0000) acc1_neg: 78.646 acc5_neg: 100.000 time: 0.1947 data: 0.0990 max mem: 3532 +Test: [280/389] eta: 0:00:22 acc1_pos: 42.945 acc5_pos: 100.000 loss: 0.5727 (0.5933) acc1: 68.2292 (66.5026) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1910 data: 0.0958 max mem: 3532 +Test: [290/389] eta: 0:00:20 acc1_pos: 42.945 acc5_pos: 100.000 loss: 0.4428 (0.5863) acc1: 85.6771 (67.3182) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1999 data: 0.1082 max mem: 3532 +Test: [300/389] eta: 0:00:17 acc1_pos: 42.945 acc5_pos: 100.000 loss: 0.4786 (0.5846) acc1: 86.7188 (67.7663) acc5: 100.0000 (100.0000) acc1_neg: 72.135 acc5_neg: 100.000 time: 0.1937 data: 0.0986 max mem: 3532 +Test: [310/389] eta: 0:00:15 acc1_pos: 42.945 acc5_pos: 100.000 loss: 0.5725 (0.5882) acc1: 74.4792 (67.5191) acc5: 100.0000 (100.0000) acc1_neg: 29.688 acc5_neg: 100.000 time: 0.1599 data: 0.0655 max mem: 3532 +Test: [320/389] eta: 0:00:13 acc1_pos: 42.945 acc5_pos: 100.000 loss: 0.5236 (0.5844) acc1: 74.7396 (67.9971) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1524 data: 0.0607 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 42.945 acc5_pos: 100.000 loss: 0.6045 (0.5851) acc1: 73.6979 (68.1308) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.1649 data: 0.0727 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 42.945 acc5_pos: 100.000 loss: 0.3553 (0.5759) acc1: 80.4688 (68.9783) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1608 data: 0.0685 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 42.945 acc5_pos: 100.000 loss: 0.2739 (0.5800) acc1: 98.9583 (68.5868) acc5: 100.0000 (100.0000) acc1_neg: 30.469 acc5_neg: 100.000 time: 0.1344 data: 0.0428 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 42.945 acc5_pos: 100.000 loss: 0.5419 (0.5806) acc1: 70.8333 (68.4514) acc5: 100.0000 (100.0000) acc1_neg: 77.344 acc5_neg: 100.000 time: 0.1302 data: 0.0391 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 42.945 acc5_pos: 100.000 loss: 0.5200 (0.5773) acc1: 74.7396 (68.7458) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1251 data: 0.0340 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 42.945 acc5_pos: 100.000 loss: 0.4280 (0.5736) acc1: 83.5938 (69.1232) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1094 data: 0.0189 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 42.945 acc5_pos: 100.000 loss: 0.4224 (0.5701) acc1: 89.8438 (69.5363) acc5: 100.0000 (100.0000) acc1_neg: 78.142 acc5_neg: 100.000 time: 0.1122 data: 0.0179 max mem: 3532 +Test: Total time: 0:01:12 (0.1874 s / it) +* Overall Acc@1 69.536 Acc@5 100.000 + target == real Acc@1 48.007 Acc@5 100.000 + target == fake Acc@1 79.544 Acc@5 100.000 + loss 0.570 + PR-AUC 0.8483 ROC-AUC 0.7290 EER 33.99% (thr=0.5908) +Accuracy of the network on the 149358 test images: 69.5% +[rank0]:[W725 18:54:20.445082345 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:54:39.302000 610139 torch/distributed/run.py:766] +W0725 18:54:39.302000 610139 torch/distributed/run.py:766] ***************************************** +W0725 18:54:39.302000 610139 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:54:39.302000 610139 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:55:08.084858911 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:55:09.266070455 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +[rank2]:[W725 18:55:09.339746149 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:55:09.372982746 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:55:09.435133372 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +[rank0]:[W725 18:55:09.565812783 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:55:09.569571397 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:55:09.576909524 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed3_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed3_resampled_n116-040-213_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:30:40 acc1_pos: 82.292 acc5_pos: 100.000 loss: 0.4416 (0.4416) acc1: 82.2917 (82.2917) acc5: 100.0000 (100.0000) time: 5.5284 data: 4.0299 max mem: 3531 +Test: [ 10/333] eta: 0:03:08 acc1_pos: 86.198 acc5_pos: 100.000 loss: 0.3257 (0.3373) acc1: 86.1979 (85.7718) acc5: 100.0000 (100.0000) time: 0.5847 data: 0.3667 max mem: 3531 +Test: [ 20/333] eta: 0:01:50 acc1_pos: 96.354 acc5_pos: 100.000 loss: 0.3116 (0.3169) acc1: 87.2396 (87.3884) acc5: 100.0000 (100.0000) time: 0.0934 data: 0.0004 max mem: 3531 +Test: [ 30/333] eta: 0:01:29 acc1_pos: 53.646 acc5_pos: 100.000 loss: 0.4034 (0.4657) acc1: 76.3021 (75.5376) acc5: 100.0000 (100.0000) time: 0.1344 data: 0.0388 max mem: 3531 +Test: [ 40/333] eta: 0:01:26 acc1_pos: 50.521 acc5_pos: 100.000 loss: 0.8046 (0.5505) acc1: 47.1354 (68.2292) acc5: 100.0000 (100.0000) time: 0.2349 data: 0.1419 max mem: 3531 +Test: [ 50/333] eta: 0:01:19 acc1_pos: 52.083 acc5_pos: 100.000 loss: 0.7954 (0.6023) acc1: 46.8750 (63.9502) acc5: 100.0000 (100.0000) time: 0.2637 data: 0.1713 max mem: 3531 +Test: [ 60/333] eta: 0:01:16 acc1_pos: 43.490 acc5_pos: 100.000 loss: 0.7920 (0.6357) acc1: 47.3958 (61.2748) acc5: 100.0000 (100.0000) time: 0.2450 data: 0.1498 max mem: 3531 +Test: [ 70/333] eta: 0:01:09 acc1_pos: 45.052 acc5_pos: 100.000 loss: 0.7932 (0.6602) acc1: 47.6562 (59.5107) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1274 max mem: 3531 +Test: [ 80/333] eta: 0:01:02 acc1_pos: 54.167 acc5_pos: 100.000 loss: 0.7892 (0.6780) acc1: 48.1771 (57.8929) acc5: 100.0000 (100.0000) time: 0.1559 data: 0.0575 max mem: 3531 +Test: [ 90/333] eta: 0:01:01 acc1_pos: 60.156 acc5_pos: 100.000 loss: 0.7224 (0.6809) acc1: 53.1250 (57.8154) acc5: 100.0000 (100.0000) time: 0.2002 data: 0.1051 max mem: 3531 +Test: [100/333] eta: 0:00:57 acc1_pos: 65.104 acc5_pos: 100.000 loss: 0.6948 (0.6846) acc1: 58.5938 (57.7094) acc5: 100.0000 (100.0000) time: 0.2388 data: 0.1451 max mem: 3531 +Test: [110/333] eta: 0:00:52 acc1_pos: 64.583 acc5_pos: 100.000 loss: 0.7192 (0.6885) acc1: 54.6875 (57.5005) acc5: 100.0000 (100.0000) time: 0.1604 data: 0.0638 max mem: 3531 +Test: [120/333] eta: 0:00:49 acc1_pos: 64.323 acc5_pos: 100.000 loss: 0.6709 (0.6885) acc1: 56.2500 (57.4832) acc5: 100.0000 (100.0000) time: 0.1722 data: 0.0740 max mem: 3531 +Test: [130/333] eta: 0:00:47 acc1_pos: 72.396 acc5_pos: 100.000 loss: 0.6518 (0.6881) acc1: 60.4167 (57.6415) acc5: 100.0000 (100.0000) time: 0.2168 data: 0.1217 max mem: 3531 +Test: [140/333] eta: 0:00:43 acc1_pos: 78.125 acc5_pos: 100.000 loss: 0.6648 (0.6868) acc1: 59.1146 (57.7866) acc5: 100.0000 (100.0000) time: 0.1676 data: 0.0762 max mem: 3531 +Test: [150/333] eta: 0:00:40 acc1_pos: 58.073 acc5_pos: 100.000 loss: 0.6923 (0.6890) acc1: 58.0729 (57.6331) acc5: 100.0000 (100.0000) time: 0.1571 data: 0.0643 max mem: 3531 +Test: [160/333] eta: 0:00:37 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.7157 (0.6943) acc1: 55.4688 (57.3402) acc5: 100.0000 (100.0000) acc1_neg: 48.698 acc5_neg: 100.000 time: 0.1829 data: 0.0888 max mem: 3531 +Test: [170/333] eta: 0:00:35 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.7250 (0.6991) acc1: 57.5521 (57.1927) acc5: 100.0000 (100.0000) acc1_neg: 58.333 acc5_neg: 100.000 time: 0.2049 data: 0.1112 max mem: 3531 +Test: [180/333] eta: 0:00:33 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.7102 (0.6978) acc1: 60.1562 (57.4773) acc5: 100.0000 (100.0000) acc1_neg: 60.417 acc5_neg: 100.000 time: 0.2118 data: 0.1171 max mem: 3531 +Test: [190/333] eta: 0:00:30 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.6417 (0.6920) acc1: 63.5417 (58.1220) acc5: 100.0000 (100.0000) acc1_neg: 69.792 acc5_neg: 100.000 time: 0.1598 data: 0.0630 max mem: 3531 +Test: [200/333] eta: 0:00:28 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.5846 (0.6879) acc1: 69.0104 (58.7013) acc5: 100.0000 (100.0000) acc1_neg: 64.583 acc5_neg: 100.000 time: 0.1655 data: 0.0710 max mem: 3531 +Test: [210/333] eta: 0:00:26 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.6644 (0.6887) acc1: 63.8021 (58.8443) acc5: 100.0000 (100.0000) acc1_neg: 54.948 acc5_neg: 100.000 time: 0.1866 data: 0.0945 max mem: 3532 +Test: [220/333] eta: 0:00:23 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.7019 (0.6918) acc1: 61.7188 (58.7835) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.1746 data: 0.0825 max mem: 3532 +Test: [230/333] eta: 0:00:21 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.5104 (0.6827) acc1: 76.8229 (59.7425) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.1937 data: 0.1005 max mem: 3532 +Test: [240/333] eta: 0:00:19 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.5082 (0.6786) acc1: 76.8229 (60.1843) acc5: 100.0000 (100.0000) acc1_neg: 70.573 acc5_neg: 100.000 time: 0.1852 data: 0.0908 max mem: 3532 +Test: [250/333] eta: 0:00:17 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.5993 (0.6757) acc1: 69.5312 (60.5194) acc5: 100.0000 (100.0000) acc1_neg: 61.719 acc5_neg: 100.000 time: 0.1824 data: 0.0873 max mem: 3532 +Test: [260/333] eta: 0:00:15 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.6033 (0.6739) acc1: 67.9688 (60.7709) acc5: 100.0000 (100.0000) acc1_neg: 70.573 acc5_neg: 100.000 time: 0.1903 data: 0.0955 max mem: 3532 +Test: [270/333] eta: 0:00:12 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.5859 (0.6702) acc1: 69.7917 (61.1787) acc5: 100.0000 (100.0000) acc1_neg: 69.010 acc5_neg: 100.000 time: 0.1793 data: 0.0868 max mem: 3532 +Test: [280/333] eta: 0:00:10 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.5833 (0.6712) acc1: 69.7917 (61.1349) acc5: 100.0000 (100.0000) acc1_neg: 64.583 acc5_neg: 100.000 time: 0.1859 data: 0.0942 max mem: 3532 +Test: [290/333] eta: 0:00:08 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.8027 (0.6760) acc1: 50.2604 (60.7299) acc5: 100.0000 (100.0000) acc1_neg: 53.125 acc5_neg: 100.000 time: 0.1989 data: 0.1073 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.8200 (0.6806) acc1: 50.2604 (60.4279) acc5: 100.0000 (100.0000) acc1_neg: 54.167 acc5_neg: 100.000 time: 0.2040 data: 0.1129 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.7331 (0.6766) acc1: 56.2500 (60.8437) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.1785 data: 0.0876 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.5364 (0.6736) acc1: 75.0000 (61.1841) acc5: 100.0000 (100.0000) acc1_neg: 53.646 acc5_neg: 100.000 time: 0.1566 data: 0.0656 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.7621 (0.6784) acc1: 54.6875 (60.9060) acc5: 100.0000 (100.0000) acc1_neg: 54.688 acc5_neg: 100.000 time: 0.1355 data: 0.0436 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.8146 (0.6794) acc1: 53.6458 (60.8543) acc5: 100.0000 (100.0000) acc1_neg: 52.113 acc5_neg: 100.000 time: 0.1387 data: 0.0435 max mem: 3532 +Test: Total time: 0:01:06 (0.1999 s / it) +* Overall Acc@1 60.854 Acc@5 100.000 + target == real Acc@1 57.680 Acc@5 100.000 + target == fake Acc@1 63.680 Acc@5 100.000 + loss 0.679 + PR-AUC 0.6592 ROC-AUC 0.6489 EER 39.33% (thr=0.5215) +Accuracy of the network on the 127843 test images: 60.9% +[rank0]:[W725 18:56:25.885952994 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:56:45.663000 610709 torch/distributed/run.py:766] +W0725 18:56:45.663000 610709 torch/distributed/run.py:766] ***************************************** +W0725 18:56:45.663000 610709 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:56:45.663000 610709 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:57:15.577137725 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:57:15.820195200 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:57:15.082309587 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:57:16.289441046 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:57:16.464071419 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:57:16.596267002 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:57:16.620685264 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:57:16.656387896 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed3_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed3_resampled_n116-040-213_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:22:30 acc1_pos: 95.573 acc5_pos: 100.000 loss: 0.2032 (0.2032) acc1: 95.5729 (95.5729) acc5: 100.0000 (100.0000) time: 5.7483 data: 3.8331 max mem: 3531 +Test: [ 10/235] eta: 0:02:16 acc1_pos: 94.531 acc5_pos: 100.000 loss: 0.2138 (0.2161) acc1: 95.3125 (95.3125) acc5: 100.0000 (100.0000) time: 0.6049 data: 0.3488 max mem: 3531 +Test: [ 20/235] eta: 0:01:17 acc1_pos: 95.573 acc5_pos: 100.000 loss: 0.2123 (0.2130) acc1: 95.5729 (95.6225) acc5: 100.0000 (100.0000) time: 0.0918 data: 0.0005 max mem: 3531 +Test: [ 30/235] eta: 0:01:13 acc1_pos: 96.354 acc5_pos: 100.000 loss: 0.2082 (0.2116) acc1: 95.8333 (95.5813) acc5: 100.0000 (100.0000) time: 0.2239 data: 0.1312 max mem: 3531 +Test: [ 40/235] eta: 0:01:04 acc1_pos: 95.052 acc5_pos: 100.000 loss: 0.2090 (0.2112) acc1: 95.8333 (95.5729) acc5: 100.0000 (100.0000) time: 0.2976 data: 0.2059 max mem: 3531 +Test: [ 50/235] eta: 0:00:56 acc1_pos: 93.229 acc5_pos: 100.000 loss: 0.2105 (0.2124) acc1: 95.0521 (95.4657) acc5: 100.0000 (100.0000) time: 0.2280 data: 0.1372 max mem: 3531 +Test: [ 60/235] eta: 0:00:50 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.2137 (0.2133) acc1: 95.0521 (95.4278) acc5: 100.0000 (100.0000) time: 0.2029 data: 0.1106 max mem: 3531 +Test: [ 70/235] eta: 0:00:45 acc1_pos: 98.462 acc5_pos: 100.000 loss: 0.2315 (0.3669) acc1: 94.2708 (86.0989) acc5: 100.0000 (100.0000) acc1_neg: 20.312 acc5_neg: 100.000 time: 0.1993 data: 0.1056 max mem: 3531 +Test: [ 80/235] eta: 0:00:41 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4388 (0.5013) acc1: 20.5729 (77.9257) acc5: 100.0000 (100.0000) acc1_neg: 18.229 acc5_neg: 100.000 time: 0.2044 data: 0.1108 max mem: 3531 +Test: [ 90/235] eta: 0:00:36 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4496 (0.6069) acc1: 19.5312 (71.5488) acc5: 100.0000 (100.0000) acc1_neg: 18.750 acc5_neg: 100.000 time: 0.1763 data: 0.0824 max mem: 3531 +Test: [100/235] eta: 0:00:34 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4601 (0.6891) acc1: 19.5312 (66.5506) acc5: 100.0000 (100.0000) acc1_neg: 19.792 acc5_neg: 100.000 time: 0.1952 data: 0.1000 max mem: 3531 +Test: [110/235] eta: 0:00:30 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4400 (0.7569) acc1: 20.3125 (62.3897) acc5: 100.0000 (100.0000) acc1_neg: 24.219 acc5_neg: 100.000 time: 0.2135 data: 0.1182 max mem: 3531 +Test: [120/235] eta: 0:00:27 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4966 (0.8282) acc1: 16.4062 (58.2365) acc5: 100.0000 (100.0000) acc1_neg: 10.677 acc5_neg: 100.000 time: 0.1899 data: 0.0968 max mem: 3531 +Test: [130/235] eta: 0:00:25 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.6239 (0.8885) acc1: 11.7188 (54.6736) acc5: 100.0000 (100.0000) acc1_neg: 11.979 acc5_neg: 100.000 time: 0.2015 data: 0.1108 max mem: 3531 +Test: [140/235] eta: 0:00:22 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.6414 (0.9432) acc1: 11.1979 (51.5865) acc5: 100.0000 (100.0000) acc1_neg: 8.073 acc5_neg: 100.000 time: 0.2125 data: 0.1213 max mem: 3531 +Test: [150/235] eta: 0:00:20 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.6458 (0.9882) acc1: 10.9375 (48.9169) acc5: 100.0000 (100.0000) acc1_neg: 12.760 acc5_neg: 100.000 time: 0.2033 data: 0.1090 max mem: 3531 +Test: [160/235] eta: 0:00:17 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.6167 (1.0276) acc1: 11.9792 (46.6372) acc5: 100.0000 (100.0000) acc1_neg: 11.979 acc5_neg: 100.000 time: 0.1963 data: 0.1007 max mem: 3531 +Test: [170/235] eta: 0:00:15 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.5725 (1.0396) acc1: 13.2812 (45.4937) acc5: 100.0000 (100.0000) acc1_neg: 26.823 acc5_neg: 100.000 time: 0.1984 data: 0.1055 max mem: 3531 +Test: [180/235] eta: 0:00:12 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.1902 (1.0474) acc1: 28.9062 (44.5859) acc5: 100.0000 (100.0000) acc1_neg: 28.125 acc5_neg: 100.000 time: 0.2145 data: 0.1238 max mem: 3531 +Test: [190/235] eta: 0:00:10 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.1684 (1.0543) acc1: 29.4271 (43.8114) acc5: 100.0000 (100.0000) acc1_neg: 28.906 acc5_neg: 100.000 time: 0.2028 data: 0.1108 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.1848 (1.0617) acc1: 28.9062 (43.0491) acc5: 100.0000 (100.0000) acc1_neg: 27.604 acc5_neg: 100.000 time: 0.1763 data: 0.0815 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.1963 (1.0672) acc1: 28.6458 (42.4492) acc5: 100.0000 (100.0000) acc1_neg: 30.729 acc5_neg: 100.000 time: 0.1760 data: 0.0824 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.1776 (1.0722) acc1: 30.7292 (41.8929) acc5: 100.0000 (100.0000) acc1_neg: 30.729 acc5_neg: 100.000 time: 0.1485 data: 0.0579 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.1651 (1.0764) acc1: 29.6875 (41.3668) acc5: 100.0000 (100.0000) acc1_neg: 29.167 acc5_neg: 100.000 time: 0.1282 data: 0.0379 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.1695 (1.0794) acc1: 29.4271 (41.1987) acc5: 100.0000 (100.0000) acc1_neg: 34.783 acc5_neg: 100.000 time: 0.1221 data: 0.0351 max mem: 3532 +Test: Total time: 0:00:50 (0.2136 s / it) +* Overall Acc@1 41.199 Acc@5 100.000 + target == real Acc@1 95.447 Acc@5 100.000 + target == fake Acc@1 21.578 Acc@5 100.000 + loss 1.079 + PR-AUC 0.8733 ROC-AUC 0.7207 EER 33.87% (thr=0.1758) +Accuracy of the network on the 89879 test images: 41.2% +[rank0]:[W725 18:58:16.598841048 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:58:36.165000 611279 torch/distributed/run.py:766] +W0725 18:58:36.165000 611279 torch/distributed/run.py:766] ***************************************** +W0725 18:58:36.165000 611279 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:58:36.165000 611279 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:59:06.347816257 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:59:06.624302268 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:59:06.801336045 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:59:06.021755383 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:59:06.120873591 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:59:06.151909553 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:59:07.195249991 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:59:07.224314444 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed3_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed3_resampled_n116-040-213_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:23:28 acc1_pos: 42.188 acc5_pos: 100.000 loss: 0.7742 (0.7742) acc1: 42.1875 (42.1875) acc5: 100.0000 (100.0000) time: 5.2356 data: 3.3017 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/269] eta: 0:02:24 acc1_pos: 54.427 acc5_pos: 100.000 loss: 0.6998 (0.6902) acc1: 54.1667 (56.4157) acc5: 100.0000 (100.0000) time: 0.5580 data: 0.3005 max mem: 3531 +Test: [ 20/269] eta: 0:01:25 acc1_pos: 67.708 acc5_pos: 100.000 loss: 0.6716 (0.6766) acc1: 54.4271 (58.3457) acc5: 100.0000 (100.0000) time: 0.0977 data: 0.0067 max mem: 3531 +Test: [ 30/269] eta: 0:01:08 acc1_pos: 65.365 acc5_pos: 100.000 loss: 0.6393 (0.6668) acc1: 59.1146 (59.8202) acc5: 100.0000 (100.0000) time: 0.1385 data: 0.0452 max mem: 3531 +Test: [ 40/269] eta: 0:01:05 acc1_pos: 40.517 acc5_pos: 100.000 loss: 0.5573 (0.6314) acc1: 71.3542 (64.1832) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.2259 data: 0.1328 max mem: 3531 +Test: [ 50/269] eta: 0:01:00 acc1_pos: 40.517 acc5_pos: 100.000 loss: 0.4825 (0.6100) acc1: 77.3438 (66.3194) acc5: 100.0000 (100.0000) acc1_neg: 53.125 acc5_neg: 100.000 time: 0.2563 data: 0.1643 max mem: 3531 +Test: [ 60/269] eta: 0:00:55 acc1_pos: 40.517 acc5_pos: 100.000 loss: 0.5123 (0.6111) acc1: 72.3958 (66.1245) acc5: 100.0000 (100.0000) acc1_neg: 46.875 acc5_neg: 100.000 time: 0.2240 data: 0.1312 max mem: 3531 +Test: [ 70/269] eta: 0:00:51 acc1_pos: 40.517 acc5_pos: 100.000 loss: 0.6299 (0.6127) acc1: 62.5000 (65.7350) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.2086 data: 0.1153 max mem: 3531 +Test: [ 80/269] eta: 0:00:46 acc1_pos: 40.517 acc5_pos: 100.000 loss: 0.5960 (0.6084) acc1: 64.8438 (66.0269) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.1899 data: 0.0971 max mem: 3531 +Test: [ 90/269] eta: 0:00:43 acc1_pos: 40.517 acc5_pos: 100.000 loss: 0.6146 (0.6157) acc1: 62.7604 (65.1557) acc5: 100.0000 (100.0000) acc1_neg: 61.719 acc5_neg: 100.000 time: 0.1948 data: 0.1016 max mem: 3531 +Test: [100/269] eta: 0:00:40 acc1_pos: 40.517 acc5_pos: 100.000 loss: 0.6596 (0.6247) acc1: 60.1562 (64.3513) acc5: 100.0000 (100.0000) acc1_neg: 51.302 acc5_neg: 100.000 time: 0.2148 data: 0.1207 max mem: 3531 +Test: [110/269] eta: 0:00:37 acc1_pos: 40.517 acc5_pos: 100.000 loss: 0.7142 (0.6345) acc1: 52.3438 (63.3141) acc5: 100.0000 (100.0000) acc1_neg: 53.385 acc5_neg: 100.000 time: 0.1938 data: 0.0999 max mem: 3531 +Test: [120/269] eta: 0:00:34 acc1_pos: 40.517 acc5_pos: 100.000 loss: 0.6624 (0.6299) acc1: 59.1146 (63.7935) acc5: 100.0000 (100.0000) acc1_neg: 70.573 acc5_neg: 100.000 time: 0.1804 data: 0.0849 max mem: 3531 +Test: [130/269] eta: 0:00:31 acc1_pos: 40.517 acc5_pos: 100.000 loss: 0.5845 (0.6309) acc1: 66.4062 (63.5834) acc5: 100.0000 (100.0000) acc1_neg: 51.302 acc5_neg: 100.000 time: 0.1851 data: 0.0902 max mem: 3531 +Test: [140/269] eta: 0:00:28 acc1_pos: 40.517 acc5_pos: 100.000 loss: 0.6842 (0.6362) acc1: 55.4688 (62.9119) acc5: 100.0000 (100.0000) acc1_neg: 58.594 acc5_neg: 100.000 time: 0.1894 data: 0.0954 max mem: 3531 +Test: [150/269] eta: 0:00:26 acc1_pos: 40.517 acc5_pos: 100.000 loss: 0.6643 (0.6339) acc1: 57.5521 (63.0381) acc5: 100.0000 (100.0000) acc1_neg: 71.354 acc5_neg: 100.000 time: 0.1912 data: 0.0966 max mem: 3531 +Test: [160/269] eta: 0:00:24 acc1_pos: 40.517 acc5_pos: 100.000 loss: 0.5563 (0.6299) acc1: 67.4479 (63.4980) acc5: 100.0000 (100.0000) acc1_neg: 72.656 acc5_neg: 100.000 time: 0.2149 data: 0.1196 max mem: 3531 +Test: [170/269] eta: 0:00:21 acc1_pos: 40.517 acc5_pos: 100.000 loss: 0.6007 (0.6311) acc1: 65.8854 (63.4686) acc5: 100.0000 (100.0000) acc1_neg: 69.271 acc5_neg: 100.000 time: 0.2035 data: 0.1074 max mem: 3531 +Test: [180/269] eta: 0:00:19 acc1_pos: 40.517 acc5_pos: 100.000 loss: 0.5986 (0.6278) acc1: 65.8854 (63.8899) acc5: 100.0000 (100.0000) acc1_neg: 60.677 acc5_neg: 100.000 time: 0.1699 data: 0.0763 max mem: 3531 +Test: [190/269] eta: 0:00:17 acc1_pos: 40.517 acc5_pos: 100.000 loss: 0.5897 (0.6282) acc1: 68.4896 (63.8621) acc5: 100.0000 (100.0000) acc1_neg: 58.594 acc5_neg: 100.000 time: 0.1871 data: 0.0933 max mem: 3531 +Test: [200/269] eta: 0:00:14 acc1_pos: 40.517 acc5_pos: 100.000 loss: 0.6490 (0.6295) acc1: 58.5938 (63.7179) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1816 data: 0.0878 max mem: 3531 +Test: [210/269] eta: 0:00:12 acc1_pos: 40.517 acc5_pos: 100.000 loss: 0.6003 (0.6268) acc1: 66.6667 (63.9638) acc5: 100.0000 (100.0000) acc1_neg: 70.312 acc5_neg: 100.000 time: 0.1728 data: 0.0806 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 40.517 acc5_pos: 100.000 loss: 0.5498 (0.6225) acc1: 68.2292 (64.3500) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1982 data: 0.1033 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 40.517 acc5_pos: 100.000 loss: 0.4955 (0.6167) acc1: 77.3438 (64.9610) acc5: 100.0000 (100.0000) acc1_neg: 75.781 acc5_neg: 100.000 time: 0.1799 data: 0.0847 max mem: 3532 +Test: [240/269] eta: 0:00:05 acc1_pos: 40.517 acc5_pos: 100.000 loss: 0.5461 (0.6186) acc1: 72.6562 (64.8654) acc5: 100.0000 (100.0000) acc1_neg: 63.802 acc5_neg: 100.000 time: 0.1399 data: 0.0455 max mem: 3532 +Test: [250/269] eta: 0:00:03 acc1_pos: 40.517 acc5_pos: 100.000 loss: 0.6465 (0.6221) acc1: 58.0729 (64.5045) acc5: 100.0000 (100.0000) acc1_neg: 43.229 acc5_neg: 100.000 time: 0.1429 data: 0.0481 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 40.517 acc5_pos: 100.000 loss: 0.7538 (0.6283) acc1: 46.3542 (63.7781) acc5: 100.0000 (100.0000) acc1_neg: 61.458 acc5_neg: 100.000 time: 0.1382 data: 0.0464 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 40.517 acc5_pos: 100.000 loss: 0.6856 (0.6275) acc1: 53.3854 (63.9646) acc5: 100.0000 (100.0000) acc1_neg: 89.239 acc5_neg: 100.000 time: 0.1206 data: 0.0251 max mem: 3532 +Test: Total time: 0:00:53 (0.1994 s / it) +* Overall Acc@1 63.965 Acc@5 100.000 + target == real Acc@1 59.279 Acc@5 100.000 + target == fake Acc@1 64.604 Acc@5 100.000 + loss 0.627 + PR-AUC 0.9336 ROC-AUC 0.6734 EER 37.79% (thr=0.5142) +Accuracy of the network on the 103293 test images: 64.0% +[rank0]:[W725 19:00:10.597397357 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-25T19:00:18Z] eval_done tag=fixed500k_fake_n64_seed3_resampled +[2026-07-25T19:00:18Z] eval_wait tag=fixed500k_real_n32_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed4_resampled +[2026-07-25T19:01:18Z] eval_wait tag=fixed500k_real_n32_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed4_resampled +[2026-07-25T19:02:18Z] eval_wait tag=fixed500k_real_n32_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed4_resampled +[2026-07-25T19:03:18Z] eval_wait tag=fixed500k_real_n32_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed4_resampled +[2026-07-25T19:04:18Z] eval_wait tag=fixed500k_real_n32_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed4_resampled +[2026-07-25T19:05:19Z] eval_wait tag=fixed500k_real_n32_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed4_resampled +[2026-07-25T19:06:19Z] eval_wait tag=fixed500k_real_n32_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed4_resampled +[2026-07-25T19:07:19Z] eval_wait tag=fixed500k_real_n32_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed4_resampled +[2026-07-25T19:08:19Z] eval_wait tag=fixed500k_real_n32_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed4_resampled +[2026-07-25T19:09:19Z] eval_wait tag=fixed500k_real_n32_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed4_resampled +[2026-07-25T19:10:19Z] eval_wait tag=fixed500k_real_n32_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed4_resampled +[2026-07-25T19:11:20Z] eval_wait tag=fixed500k_real_n32_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed4_resampled +[2026-07-25T19:12:20Z] eval_wait tag=fixed500k_real_n32_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed4_resampled +[2026-07-25T19:13:20Z] eval_wait tag=fixed500k_real_n32_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed4_resampled +[2026-07-25T19:14:20Z] eval_wait tag=fixed500k_real_n32_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed4_resampled +[2026-07-25T19:15:20Z] eval_wait tag=fixed500k_real_n32_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed4_resampled +[2026-07-25T19:16:20Z] eval_wait tag=fixed500k_real_n32_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed4_resampled +[2026-07-25T19:17:21Z] eval_wait tag=fixed500k_real_n32_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed4_resampled +[2026-07-25T19:18:21Z] eval_wait tag=fixed500k_real_n32_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed4_resampled +[2026-07-25T19:19:21Z] eval_wait tag=fixed500k_real_n32_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed4_resampled +[2026-07-25T19:20:21Z] eval_wait tag=fixed500k_real_n32_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed4_resampled +[2026-07-25T19:21:21Z] eval_wait tag=fixed500k_real_n32_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed4_resampled +[2026-07-25T19:22:21Z] eval_skip tag=fixed500k_real_n32_seed4_resampled reason=complete_after_lock output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed4_resampled +[2026-07-25T19:22:22Z] eval_wait tag=fixed500k_fake_n64_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed4_resampled +[2026-07-25T19:23:22Z] eval_wait tag=fixed500k_fake_n64_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed4_resampled +[2026-07-25T19:24:22Z] eval_wait tag=fixed500k_fake_n64_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed4_resampled +[2026-07-25T19:25:22Z] eval_wait tag=fixed500k_fake_n64_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed4_resampled +[2026-07-25T19:26:22Z] eval_wait tag=fixed500k_fake_n64_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed4_resampled +[2026-07-25T19:27:23Z] eval_wait tag=fixed500k_fake_n64_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed4_resampled +[2026-07-25T19:28:23Z] eval_wait tag=fixed500k_fake_n64_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed4_resampled +[2026-07-25T19:29:23Z] eval_wait tag=fixed500k_fake_n64_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed4_resampled +[2026-07-25T19:30:23Z] eval_wait tag=fixed500k_fake_n64_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed4_resampled +[2026-07-25T19:31:23Z] eval_wait tag=fixed500k_fake_n64_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed4_resampled +[2026-07-25T19:32:23Z] eval_wait tag=fixed500k_fake_n64_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed4_resampled +[2026-07-25T19:33:24Z] eval_wait tag=fixed500k_fake_n64_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed4_resampled +[2026-07-25T19:34:24Z] eval_wait tag=fixed500k_fake_n64_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed4_resampled +[2026-07-25T19:35:24Z] eval_wait tag=fixed500k_fake_n64_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed4_resampled +[2026-07-25T19:36:24Z] eval_wait tag=fixed500k_fake_n64_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed4_resampled +[2026-07-25T19:37:24Z] eval_wait tag=fixed500k_fake_n64_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed4_resampled +[2026-07-25T19:38:24Z] eval_wait tag=fixed500k_fake_n64_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed4_resampled +[2026-07-25T19:39:25Z] eval_wait tag=fixed500k_fake_n64_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed4_resampled +[2026-07-25T19:40:25Z] eval_wait tag=fixed500k_fake_n64_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed4_resampled +[2026-07-25T19:41:25Z] eval_wait tag=fixed500k_fake_n64_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed4_resampled +[2026-07-25T19:42:25Z] eval_wait tag=fixed500k_fake_n64_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed4_resampled +[2026-07-25T19:43:25Z] eval_wait tag=fixed500k_fake_n64_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed4_resampled +[2026-07-25T19:44:26Z] eval_skip tag=fixed500k_fake_n64_seed4_resampled reason=complete_after_lock output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed4_resampled +eval_queue_done_utc=2026-07-25T19:44:26Z diff --git a/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_112559/eval_queue_n116-040-213.pid b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_112559/eval_queue_n116-040-213.pid new file mode 100644 index 0000000000000000000000000000000000000000..063b206be535d655c424626647d7bff374923351 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_112559/eval_queue_n116-040-213.pid @@ -0,0 +1 @@ +588894 diff --git a/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_112559/eval_queue_n116-041-023.log b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_112559/eval_queue_n116-041-023.log new file mode 100644 index 0000000000000000000000000000000000000000..967b4a540c183a23fa547af76a4831d3a9006721 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_112559/eval_queue_n116-041-023.log @@ -0,0 +1,3182 @@ +eval_queue_started_utc=2026-07-25T17:02:42Z +hostname=n116-041-023 +train_tag=20260725_112559 +eval_tag=20260725_112559 +real_level=23 +fake_level=45 +eval_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559 +batch_per_gpu=256 +[2026-07-25T17:02:42Z] eval_start tag=fixed500k_real_n23_seed1_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed1_resampled_n116-041-023_20260725_112559/best_checkpoint.pth +started_utc=2026-07-25T17:02:43Z +hostname=n116-041-023 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed1_resampled_n116-041-023_20260725_112559/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed1_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0725 17:02:56.365000 588837 torch/distributed/run.py:766] +W0725 17:02:56.365000 588837 torch/distributed/run.py:766] ***************************************** +W0725 17:02:56.365000 588837 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:02:56.365000 588837 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:03:27.395597400 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:03:28.810855995 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:03:28.213562477 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +[rank0]:[W725 17:03:29.594429265 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:03:29.602577523 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:03:29.618358004 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:03:29.722965181 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:03:29.823173746 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed1_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed1_resampled_n116-041-023_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:25:28 acc1_pos: 89.323 acc5_pos: 100.000 loss: 0.4877 (0.4877) acc1: 89.3229 (89.3229) acc5: 100.0000 (100.0000) time: 6.5330 data: 5.0526 max mem: 3531 +Test: [ 10/234] eta: 0:02:33 acc1_pos: 70.833 acc5_pos: 100.000 loss: 0.5182 (0.5563) acc1: 77.6042 (70.8333) acc5: 100.0000 (100.0000) time: 0.6832 data: 0.4596 max mem: 3531 +Test: [ 20/234] eta: 0:01:26 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.6496 (0.6454) acc1: 66.9271 (64.9058) acc5: 100.0000 (100.0000) time: 0.0991 data: 0.0003 max mem: 3531 +Test: [ 30/234] eta: 0:01:06 acc1_pos: 82.552 acc5_pos: 100.000 loss: 0.2484 (0.5300) acc1: 82.5521 (73.4375) acc5: 100.0000 (100.0000) time: 0.1261 data: 0.0308 max mem: 3531 +Test: [ 40/234] eta: 0:00:57 acc1_pos: 87.500 acc5_pos: 100.000 loss: 0.3226 (0.5070) acc1: 86.1979 (74.8730) acc5: 100.0000 (100.0000) time: 0.1799 data: 0.0879 max mem: 3531 +Test: [ 50/234] eta: 0:00:51 acc1_pos: 89.323 acc5_pos: 100.000 loss: 0.3234 (0.4683) acc1: 86.4583 (77.3897) acc5: 100.0000 (100.0000) time: 0.2137 data: 0.1190 max mem: 3531 +Test: [ 60/234] eta: 0:00:46 acc1_pos: 82.069 acc5_pos: 100.000 loss: 0.1974 (0.4038) acc1: 91.1458 (80.9682) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2099 data: 0.1155 max mem: 3531 +Test: [ 70/234] eta: 0:00:42 acc1_pos: 82.069 acc5_pos: 100.000 loss: 0.0637 (0.3929) acc1: 99.7396 (81.5178) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.1949 data: 0.1022 max mem: 3531 +Test: [ 80/234] eta: 0:00:38 acc1_pos: 82.069 acc5_pos: 100.000 loss: 0.3908 (0.3848) acc1: 79.6875 (82.1277) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0931 max mem: 3531 +Test: [ 90/234] eta: 0:00:34 acc1_pos: 82.069 acc5_pos: 100.000 loss: 0.1864 (0.3614) acc1: 93.2292 (83.5279) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1849 data: 0.0904 max mem: 3531 +Test: [100/234] eta: 0:00:31 acc1_pos: 82.069 acc5_pos: 100.000 loss: 0.2091 (0.3592) acc1: 91.1458 (83.6840) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1699 data: 0.0745 max mem: 3531 +Test: [110/234] eta: 0:00:27 acc1_pos: 82.069 acc5_pos: 100.000 loss: 0.3662 (0.3594) acc1: 81.7708 (83.6970) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.1449 data: 0.0513 max mem: 3531 +Test: [120/234] eta: 0:00:25 acc1_pos: 82.069 acc5_pos: 100.000 loss: 0.4144 (0.3620) acc1: 81.7708 (83.7358) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.1749 data: 0.0798 max mem: 3531 +Test: [130/234] eta: 0:00:23 acc1_pos: 82.069 acc5_pos: 100.000 loss: 0.4270 (0.3678) acc1: 82.8125 (83.6435) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.2099 data: 0.1153 max mem: 3531 +Test: [140/234] eta: 0:00:20 acc1_pos: 82.069 acc5_pos: 100.000 loss: 0.4473 (0.3910) acc1: 80.7292 (82.4653) acc5: 100.0000 (100.0000) acc1_neg: 40.625 acc5_neg: 100.000 time: 0.1849 data: 0.0881 max mem: 3531 +Test: [150/234] eta: 0:00:18 acc1_pos: 82.069 acc5_pos: 100.000 loss: 0.3381 (0.3939) acc1: 84.1146 (82.3710) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.1762 data: 0.0799 max mem: 3531 +Test: [160/234] eta: 0:00:15 acc1_pos: 82.069 acc5_pos: 100.000 loss: 0.3151 (0.3936) acc1: 86.4583 (82.4405) acc5: 100.0000 (100.0000) acc1_neg: 72.135 acc5_neg: 100.000 time: 0.2099 data: 0.1152 max mem: 3531 +Test: [170/234] eta: 0:00:13 acc1_pos: 82.069 acc5_pos: 100.000 loss: 0.5182 (0.4019) acc1: 75.7812 (81.9384) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.1986 data: 0.1024 max mem: 3531 +Test: [180/234] eta: 0:00:11 acc1_pos: 82.069 acc5_pos: 100.000 loss: 0.5192 (0.4095) acc1: 75.2604 (81.3593) acc5: 100.0000 (100.0000) acc1_neg: 59.115 acc5_neg: 100.000 time: 0.1849 data: 0.0917 max mem: 3531 +Test: [190/234] eta: 0:00:09 acc1_pos: 82.069 acc5_pos: 100.000 loss: 0.5891 (0.4199) acc1: 68.2292 (80.6705) acc5: 100.0000 (100.0000) acc1_neg: 70.573 acc5_neg: 100.000 time: 0.1949 data: 0.1034 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 82.069 acc5_pos: 100.000 loss: 0.5840 (0.4133) acc1: 70.0521 (81.0595) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2252 data: 0.1319 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 82.069 acc5_pos: 100.000 loss: 0.1007 (0.3986) acc1: 100.0000 (81.9375) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2102 data: 0.1155 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 82.069 acc5_pos: 100.000 loss: 0.1045 (0.3858) acc1: 99.2188 (82.6829) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1654 data: 0.0728 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 82.069 acc5_pos: 100.000 loss: 0.1715 (0.3834) acc1: 89.8438 (82.9421) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1567 data: 0.0668 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 82.069 acc5_pos: 100.000 loss: 0.3177 (0.3826) acc1: 89.5833 (83.0088) acc5: 100.0000 (100.0000) acc1_neg: 91.051 acc5_neg: 100.000 time: 0.1360 data: 0.0453 max mem: 3532 +Test: Total time: 0:00:48 (0.2068 s / it) +* Overall Acc@1 83.009 Acc@5 100.000 + target == real Acc@1 77.424 Acc@5 100.000 + target == fake Acc@1 84.583 Acc@5 100.000 + loss 0.383 + PR-AUC 0.9701 ROC-AUC 0.8993 EER 18.08% (thr=0.5469) +Accuracy of the network on the 89729 test images: 83.0% +[rank0]:[W725 17:04:26.674261143 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 17:04:47.160000 589439 torch/distributed/run.py:766] +W0725 17:04:47.160000 589439 torch/distributed/run.py:766] ***************************************** +W0725 17:04:47.160000 589439 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:04:47.160000 589439 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:05:17.726664727 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:05:17.931831638 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:05:17.998716655 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:05:17.065338307 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:05:17.077806777 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:05:17.284030287 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W725 17:05:17.284090212 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 17:05:17.295285234 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed1_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed1_resampled_n116-041-023_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:31:28 acc1_pos: 48.698 acc5_pos: 100.000 loss: 0.7107 (0.7107) acc1: 48.6979 (48.6979) acc5: 100.0000 (100.0000) time: 6.8238 data: 5.2176 max mem: 3531 +Test: [ 10/2387] eta: 0:28:05 acc1_pos: 10.156 acc5_pos: 100.000 loss: 0.7956 (0.8509) acc1: 46.0938 (42.4479) acc5: 100.0000 (100.0000) time: 0.7090 data: 0.4747 max mem: 3531 +Test: [ 20/2387] eta: 0:16:28 acc1_pos: 18.750 acc5_pos: 100.000 loss: 0.8952 (0.9256) acc1: 34.8958 (35.4415) acc5: 100.0000 (100.0000) time: 0.0974 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:13:22 acc1_pos: 26.823 acc5_pos: 100.000 loss: 0.8964 (0.9164) acc1: 26.8229 (35.4755) acc5: 100.0000 (100.0000) time: 0.1378 data: 0.0433 max mem: 3531 +Test: [ 40/2387] eta: 0:12:27 acc1_pos: 38.281 acc5_pos: 100.000 loss: 0.8738 (0.9063) acc1: 36.4583 (35.7342) acc5: 100.0000 (100.0000) time: 0.2146 data: 0.1228 max mem: 3531 +Test: [ 50/2387] eta: 0:11:52 acc1_pos: 38.542 acc5_pos: 100.000 loss: 0.8746 (0.8992) acc1: 36.4583 (36.8924) acc5: 100.0000 (100.0000) time: 0.2493 data: 0.1539 max mem: 3531 +Test: [ 60/2387] eta: 0:11:20 acc1_pos: 53.646 acc5_pos: 100.000 loss: 0.8885 (0.9086) acc1: 38.5417 (37.2865) acc5: 100.0000 (100.0000) time: 0.2389 data: 0.1415 max mem: 3531 +Test: [ 70/2387] eta: 0:11:07 acc1_pos: 37.760 acc5_pos: 100.000 loss: 0.9973 (0.9199) acc1: 35.1562 (36.9425) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1519 max mem: 3531 +Test: [ 80/2387] eta: 0:10:56 acc1_pos: 91.927 acc5_pos: 100.000 loss: 0.8111 (0.8817) acc1: 40.8854 (40.7022) acc5: 100.0000 (100.0000) time: 0.2612 data: 0.1708 max mem: 3531 +Test: [ 90/2387] eta: 0:10:53 acc1_pos: 67.448 acc5_pos: 100.000 loss: 0.5253 (0.8398) acc1: 74.7396 (44.8146) acc5: 100.0000 (100.0000) time: 0.2724 data: 0.1820 max mem: 3531 +Test: [ 100/2387] eta: 0:10:37 acc1_pos: 71.615 acc5_pos: 100.000 loss: 0.4932 (0.8113) acc1: 78.3854 (47.7929) acc5: 100.0000 (100.0000) time: 0.2536 data: 0.1622 max mem: 3531 +Test: [ 110/2387] eta: 0:10:34 acc1_pos: 70.573 acc5_pos: 100.000 loss: 0.4884 (0.7834) acc1: 78.3854 (50.6874) acc5: 100.0000 (100.0000) time: 0.2524 data: 0.1580 max mem: 3531 +Test: [ 120/2387] eta: 0:10:22 acc1_pos: 72.656 acc5_pos: 100.000 loss: 0.5320 (0.7630) acc1: 76.5625 (52.5783) acc5: 100.0000 (100.0000) time: 0.2549 data: 0.1553 max mem: 3531 +Test: [ 130/2387] eta: 0:10:19 acc1_pos: 69.010 acc5_pos: 100.000 loss: 0.5143 (0.7416) acc1: 76.5625 (54.7929) acc5: 100.0000 (100.0000) time: 0.2501 data: 0.1516 max mem: 3531 +Test: [ 140/2387] eta: 0:10:09 acc1_pos: 80.469 acc5_pos: 100.000 loss: 0.4724 (0.7236) acc1: 82.0312 (56.7025) acc5: 100.0000 (100.0000) time: 0.2499 data: 0.1543 max mem: 3531 +Test: [ 150/2387] eta: 0:10:11 acc1_pos: 78.125 acc5_pos: 100.000 loss: 0.4771 (0.7087) acc1: 80.9896 (58.3282) acc5: 100.0000 (100.0000) time: 0.2661 data: 0.1706 max mem: 3531 +Test: [ 160/2387] eta: 0:09:45 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.4303 (0.6863) acc1: 84.6354 (60.2533) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2053 data: 0.1115 max mem: 3531 +Test: [ 170/2387] eta: 0:09:32 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1783 (0.6561) acc1: 97.3958 (62.4726) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1435 data: 0.0481 max mem: 3531 +Test: [ 180/2387] eta: 0:09:23 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1681 (0.6294) acc1: 98.1771 (64.4553) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1945 data: 0.0961 max mem: 3531 +Test: [ 190/2387] eta: 0:09:12 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1681 (0.6052) acc1: 98.6979 (66.2385) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1949 data: 0.0964 max mem: 3531 +Test: [ 200/2387] eta: 0:09:05 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1678 (0.5834) acc1: 98.6979 (67.8483) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1949 data: 0.0970 max mem: 3531 +Test: [ 210/2387] eta: 0:09:01 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1695 (0.5640) acc1: 98.1771 (69.2856) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2249 data: 0.1292 max mem: 3532 +Test: [ 220/2387] eta: 0:08:49 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1732 (0.5463) acc1: 98.1771 (70.5930) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1949 data: 0.0999 max mem: 3532 +Test: [ 230/2387] eta: 0:08:41 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1693 (0.5300) acc1: 98.1771 (71.7859) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1666 data: 0.0720 max mem: 3532 +Test: [ 240/2387] eta: 0:08:38 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1760 (0.5210) acc1: 97.3958 (72.6238) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.2072 data: 0.1147 max mem: 3532 +Test: [ 250/2387] eta: 0:08:32 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1676 (0.5051) acc1: 92.7083 (73.6689) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2160 data: 0.1236 max mem: 3532 +Test: [ 260/2387] eta: 0:08:26 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.0875 (0.4891) acc1: 100.0000 (74.6777) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2011 data: 0.1085 max mem: 3532 +Test: [ 270/2387] eta: 0:08:20 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.0803 (0.4740) acc1: 100.0000 (75.6121) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1922 data: 0.1004 max mem: 3532 +Test: [ 280/2387] eta: 0:08:14 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.0775 (0.4601) acc1: 100.0000 (76.4800) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1870 data: 0.0955 max mem: 3532 +Test: [ 290/2387] eta: 0:08:10 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.0845 (0.4474) acc1: 100.0000 (77.2883) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2051 data: 0.1135 max mem: 3532 +Test: [ 300/2387] eta: 0:08:02 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.0953 (0.4360) acc1: 100.0000 (78.0419) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1814 data: 0.0881 max mem: 3532 +Test: [ 310/2387] eta: 0:07:58 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1029 (0.4270) acc1: 100.0000 (78.7371) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1745 data: 0.0786 max mem: 3532 +Test: [ 320/2387] eta: 0:07:56 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.2007 (0.4201) acc1: 99.7396 (79.3825) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2183 data: 0.1233 max mem: 3532 +Test: [ 330/2387] eta: 0:07:51 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.2053 (0.4137) acc1: 99.4792 (79.9936) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2156 data: 0.1206 max mem: 3532 +Test: [ 340/2387] eta: 0:07:48 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.2042 (0.4075) acc1: 99.4792 (80.5673) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2065 data: 0.1121 max mem: 3532 +Test: [ 350/2387] eta: 0:07:43 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1909 (0.3997) acc1: 100.0000 (81.1202) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2008 data: 0.1084 max mem: 3532 +Test: [ 360/2387] eta: 0:07:41 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1408 (0.3928) acc1: 99.7396 (81.6157) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2059 data: 0.1137 max mem: 3532 +Test: [ 370/2387] eta: 0:07:37 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1516 (0.3863) acc1: 98.9583 (82.0825) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2102 data: 0.1176 max mem: 3532 +Test: [ 380/2387] eta: 0:07:34 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1520 (0.3801) acc1: 98.6979 (82.5172) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2018 data: 0.1081 max mem: 3532 +Test: [ 390/2387] eta: 0:07:31 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1517 (0.3744) acc1: 98.4375 (82.9317) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2109 data: 0.1182 max mem: 3532 +Test: [ 400/2387] eta: 0:07:27 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1506 (0.3688) acc1: 98.6979 (83.3262) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2001 data: 0.1090 max mem: 3532 +Test: [ 410/2387] eta: 0:07:22 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1502 (0.3635) acc1: 98.9583 (83.7097) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1844 data: 0.0930 max mem: 3532 +Test: [ 420/2387] eta: 0:07:19 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1510 (0.3585) acc1: 98.9583 (84.0676) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1981 data: 0.1061 max mem: 3532 +Test: [ 430/2387] eta: 0:07:15 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1536 (0.3557) acc1: 98.6979 (84.3188) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1892 data: 0.0976 max mem: 3532 +Test: [ 440/2387] eta: 0:07:12 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1577 (0.3512) acc1: 95.3125 (84.6148) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1864 data: 0.0944 max mem: 3533 +Test: [ 450/2387] eta: 0:07:08 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.0721 (0.3450) acc1: 100.0000 (84.9559) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1942 data: 0.1014 max mem: 3533 +Test: [ 460/2387] eta: 0:07:04 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.0712 (0.3391) acc1: 100.0000 (85.2822) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1822 data: 0.0896 max mem: 3533 +Test: [ 470/2387] eta: 0:07:01 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.0711 (0.3334) acc1: 100.0000 (85.5947) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1909 data: 0.0993 max mem: 3533 +Test: [ 480/2387] eta: 0:06:58 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.0714 (0.3280) acc1: 100.0000 (85.8942) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2026 data: 0.1090 max mem: 3533 +Test: [ 490/2387] eta: 0:06:53 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.0722 (0.3228) acc1: 100.0000 (86.1815) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1782 data: 0.0837 max mem: 3533 +Test: [ 500/2387] eta: 0:06:50 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.0722 (0.3178) acc1: 100.0000 (86.4573) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1672 data: 0.0752 max mem: 3533 +Test: [ 510/2387] eta: 0:06:46 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.0729 (0.3135) acc1: 100.0000 (86.7223) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1844 data: 0.0916 max mem: 3533 +Test: [ 520/2387] eta: 0:06:44 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1070 (0.3104) acc1: 100.0000 (86.9747) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1926 data: 0.0986 max mem: 3533 +Test: [ 530/2387] eta: 0:06:40 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1070 (0.3066) acc1: 100.0000 (87.2200) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1922 data: 0.0985 max mem: 3533 +Test: [ 540/2387] eta: 0:06:38 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1054 (0.3029) acc1: 100.0000 (87.4562) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1914 data: 0.0986 max mem: 3533 +Test: [ 550/2387] eta: 0:06:35 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1054 (0.2993) acc1: 100.0000 (87.6839) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2010 data: 0.1088 max mem: 3533 +Test: [ 560/2387] eta: 0:06:32 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1053 (0.2958) acc1: 100.0000 (87.9034) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1941 data: 0.1022 max mem: 3533 +Test: [ 570/2387] eta: 0:06:29 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1053 (0.2925) acc1: 100.0000 (88.1148) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1939 data: 0.1026 max mem: 3533 +Test: [ 580/2387] eta: 0:06:26 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1050 (0.2893) acc1: 100.0000 (88.3194) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1943 data: 0.1033 max mem: 3533 +Test: [ 590/2387] eta: 0:06:24 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1058 (0.2862) acc1: 100.0000 (88.5170) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1943 data: 0.1019 max mem: 3533 +Test: [ 600/2387] eta: 0:06:21 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1114 (0.2843) acc1: 100.0000 (88.7055) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2009 data: 0.1082 max mem: 3533 +Test: [ 610/2387] eta: 0:06:19 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1653 (0.2822) acc1: 99.7396 (88.8822) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2098 data: 0.1180 max mem: 3533 +Test: [ 620/2387] eta: 0:06:17 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1588 (0.2803) acc1: 99.4792 (89.0537) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2100 data: 0.1178 max mem: 3533 +Test: [ 630/2387] eta: 0:06:14 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1601 (0.2783) acc1: 99.4792 (89.2181) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1895 data: 0.0946 max mem: 3533 +Test: [ 640/2387] eta: 0:06:13 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1593 (0.2765) acc1: 99.4792 (89.3729) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2209 data: 0.1264 max mem: 3533 +Test: [ 650/2387] eta: 0:06:10 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1593 (0.2747) acc1: 99.4792 (89.5273) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2311 data: 0.1385 max mem: 3533 +Test: [ 660/2387] eta: 0:06:07 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1619 (0.2730) acc1: 99.4792 (89.6787) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1824 data: 0.0882 max mem: 3533 +Test: [ 670/2387] eta: 0:06:03 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1612 (0.2713) acc1: 99.7396 (89.8251) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1592 data: 0.0644 max mem: 3534 +Test: [ 680/2387] eta: 0:06:01 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1621 (0.2705) acc1: 99.7396 (89.9707) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1720 data: 0.0794 max mem: 3534 +Test: [ 690/2387] eta: 0:06:00 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.2458 (0.2713) acc1: 99.2188 (89.9538) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.2280 data: 0.1356 max mem: 3534 +Test: [ 700/2387] eta: 0:05:57 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.3513 (0.2725) acc1: 85.4167 (89.8898) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.2178 data: 0.1234 max mem: 3534 +Test: [ 710/2387] eta: 0:05:53 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.3554 (0.2737) acc1: 85.9375 (89.8210) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.1446 data: 0.0511 max mem: 3534 +Test: [ 720/2387] eta: 0:05:52 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.3533 (0.2748) acc1: 86.1979 (89.7614) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.2099 data: 0.1157 max mem: 3534 +Test: [ 730/2387] eta: 0:05:50 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.3566 (0.2760) acc1: 83.8542 (89.6820) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.2413 data: 0.1445 max mem: 3534 +Test: [ 740/2387] eta: 0:05:45 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.3564 (0.2771) acc1: 84.3750 (89.6079) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.1422 data: 0.0456 max mem: 3534 +Test: [ 750/2387] eta: 0:05:43 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.3484 (0.2780) acc1: 84.6354 (89.5535) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.1571 data: 0.0627 max mem: 3534 +Test: [ 760/2387] eta: 0:05:41 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.3481 (0.2791) acc1: 84.6354 (89.4838) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.2074 data: 0.1132 max mem: 3534 +Test: [ 770/2387] eta: 0:05:39 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.3471 (0.2795) acc1: 84.6354 (89.4746) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2111 data: 0.1173 max mem: 3534 +Test: [ 780/2387] eta: 0:05:36 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.2461 (0.2785) acc1: 96.6146 (89.5877) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1949 data: 0.1031 max mem: 3534 +Test: [ 790/2387] eta: 0:05:34 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1732 (0.2773) acc1: 98.1771 (89.6699) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1949 data: 0.1037 max mem: 3534 +Test: [ 800/2387] eta: 0:05:32 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1582 (0.2759) acc1: 98.6979 (89.7911) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2022 data: 0.1108 max mem: 3534 +Test: [ 810/2387] eta: 0:05:29 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1959 (0.2752) acc1: 98.6979 (89.8781) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1868 data: 0.0952 max mem: 3534 +Test: [ 820/2387] eta: 0:05:27 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.2232 (0.2751) acc1: 96.6146 (89.9104) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.2046 data: 0.1093 max mem: 3534 +Test: [ 830/2387] eta: 0:05:25 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.3145 (0.2768) acc1: 84.1146 (89.7967) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.2042 data: 0.1059 max mem: 3534 +Test: [ 840/2387] eta: 0:05:23 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.2155 (0.2754) acc1: 92.9688 (89.9119) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2060 data: 0.1086 max mem: 3534 +Test: [ 850/2387] eta: 0:05:21 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1463 (0.2738) acc1: 100.0000 (90.0292) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2212 data: 0.1275 max mem: 3534 +Test: [ 860/2387] eta: 0:05:19 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1239 (0.2720) acc1: 100.0000 (90.1441) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2031 data: 0.1090 max mem: 3534 +Test: [ 870/2387] eta: 0:05:16 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1159 (0.2704) acc1: 100.0000 (90.2552) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1952 data: 0.1009 max mem: 3534 +Test: [ 880/2387] eta: 0:05:14 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1358 (0.2689) acc1: 99.7396 (90.3637) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2038 data: 0.1117 max mem: 3534 +Test: [ 890/2387] eta: 0:05:12 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1676 (0.2683) acc1: 99.4792 (90.4549) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1988 data: 0.1069 max mem: 3534 +Test: [ 900/2387] eta: 0:05:09 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.2507 (0.2697) acc1: 96.6146 (90.3946) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.1901 data: 0.0989 max mem: 3535 +Test: [ 910/2387] eta: 0:05:07 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.4292 (0.2714) acc1: 82.2917 (90.3091) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1834 data: 0.0915 max mem: 3535 +Test: [ 920/2387] eta: 0:05:04 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.4341 (0.2733) acc1: 82.0312 (90.2045) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.1795 data: 0.0874 max mem: 3535 +Test: [ 930/2387] eta: 0:05:02 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.4356 (0.2750) acc1: 81.2500 (90.1109) acc5: 100.0000 (100.0000) acc1_neg: 78.646 acc5_neg: 100.000 time: 0.1770 data: 0.0858 max mem: 3535 +Test: [ 940/2387] eta: 0:04:59 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.4363 (0.2767) acc1: 81.2500 (90.0167) acc5: 100.0000 (100.0000) acc1_neg: 76.302 acc5_neg: 100.000 time: 0.1748 data: 0.0835 max mem: 3535 +Test: [ 950/2387] eta: 0:04:57 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.4338 (0.2783) acc1: 82.0312 (89.9344) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.1840 data: 0.0919 max mem: 3535 +Test: [ 960/2387] eta: 0:04:54 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.4242 (0.2799) acc1: 82.2917 (89.8584) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.1729 data: 0.0818 max mem: 3535 +Test: [ 970/2387] eta: 0:04:51 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.4325 (0.2815) acc1: 82.8125 (89.7775) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.1663 data: 0.0749 max mem: 3535 +Test: [ 980/2387] eta: 0:04:49 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.4260 (0.2817) acc1: 83.0729 (89.7954) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1782 data: 0.0864 max mem: 3535 +Test: [ 990/2387] eta: 0:04:47 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.2439 (0.2818) acc1: 95.0521 (89.8198) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1821 data: 0.0897 max mem: 3535 +Test: [1000/2387] eta: 0:04:44 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.2439 (0.2816) acc1: 94.5312 (89.8609) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.1756 data: 0.0818 max mem: 3535 +Test: [1010/2387] eta: 0:04:42 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.3194 (0.2826) acc1: 89.0625 (89.8043) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.1866 data: 0.0945 max mem: 3535 +Test: [1020/2387] eta: 0:04:40 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.3685 (0.2837) acc1: 83.0729 (89.7476) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.2184 data: 0.1272 max mem: 3535 +Test: [1030/2387] eta: 0:04:38 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.2811 (0.2832) acc1: 92.4479 (89.8069) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1942 data: 0.1025 max mem: 3535 +Test: [1040/2387] eta: 0:04:35 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1898 (0.2823) acc1: 98.1771 (89.8913) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1718 data: 0.0802 max mem: 3535 +Test: [1050/2387] eta: 0:04:33 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1810 (0.2814) acc1: 98.9583 (89.9733) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1639 data: 0.0722 max mem: 3535 +Test: [1060/2387] eta: 0:04:30 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.2315 (0.2811) acc1: 97.1354 (90.0298) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1628 data: 0.0717 max mem: 3535 +Test: [1070/2387] eta: 0:04:28 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.2522 (0.2815) acc1: 95.0521 (90.0283) acc5: 100.0000 (100.0000) acc1_neg: 75.781 acc5_neg: 100.000 time: 0.1715 data: 0.0809 max mem: 3535 +Test: [1080/2387] eta: 0:04:25 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.3687 (0.2827) acc1: 84.8958 (89.9596) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1687 data: 0.0775 max mem: 3535 +Test: [1090/2387] eta: 0:04:23 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.2538 (0.2820) acc1: 91.9271 (90.0297) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1878 data: 0.0965 max mem: 3535 +Test: [1100/2387] eta: 0:04:21 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1909 (0.2810) acc1: 98.6979 (90.1068) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2153 data: 0.1244 max mem: 3535 +Test: [1110/2387] eta: 0:04:19 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1736 (0.2801) acc1: 98.6979 (90.1764) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2044 data: 0.1098 max mem: 3535 +Test: [1120/2387] eta: 0:04:17 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1800 (0.2792) acc1: 99.2188 (90.2549) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1944 data: 0.0976 max mem: 3536 +Test: [1130/2387] eta: 0:04:15 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1891 (0.2786) acc1: 98.4375 (90.3137) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1964 data: 0.1031 max mem: 3536 +Test: [1140/2387] eta: 0:04:13 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.2705 (0.2794) acc1: 91.9271 (90.2685) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.1772 data: 0.0858 max mem: 3536 +Test: [1150/2387] eta: 0:04:10 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.3326 (0.2790) acc1: 85.9375 (90.2836) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1733 data: 0.0819 max mem: 3536 +Test: [1160/2387] eta: 0:04:08 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.0995 (0.2774) acc1: 100.0000 (90.3673) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1795 data: 0.0880 max mem: 3536 +Test: [1170/2387] eta: 0:04:06 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.0878 (0.2758) acc1: 100.0000 (90.4495) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1852 data: 0.0940 max mem: 3536 +Test: [1180/2387] eta: 0:04:04 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.0851 (0.2741) acc1: 100.0000 (90.5302) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2030 data: 0.1086 max mem: 3536 +Test: [1190/2387] eta: 0:04:02 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.0881 (0.2727) acc1: 100.0000 (90.6097) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2270 data: 0.1324 max mem: 3536 +Test: [1200/2387] eta: 0:04:00 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1003 (0.2713) acc1: 100.0000 (90.6874) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2311 data: 0.1395 max mem: 3536 +Test: [1210/2387] eta: 0:03:58 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1090 (0.2702) acc1: 100.0000 (90.7637) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2032 data: 0.1119 max mem: 3536 +Test: [1220/2387] eta: 0:03:56 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.0936 (0.2687) acc1: 100.0000 (90.8391) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2041 data: 0.1135 max mem: 3536 +Test: [1230/2387] eta: 0:03:54 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.0833 (0.2672) acc1: 100.0000 (90.9136) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2052 data: 0.1146 max mem: 3536 +Test: [1240/2387] eta: 0:03:52 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.0776 (0.2656) acc1: 100.0000 (90.9868) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1795 data: 0.0885 max mem: 3536 +Test: [1250/2387] eta: 0:03:49 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.0779 (0.2642) acc1: 100.0000 (91.0588) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1641 data: 0.0722 max mem: 3536 +Test: [1260/2387] eta: 0:03:47 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.0864 (0.2628) acc1: 100.0000 (91.1295) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1579 data: 0.0665 max mem: 3536 +Test: [1270/2387] eta: 0:03:45 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1014 (0.2617) acc1: 100.0000 (91.1987) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1697 data: 0.0783 max mem: 3536 +Test: [1280/2387] eta: 0:03:43 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1469 (0.2610) acc1: 99.7396 (91.2599) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1769 data: 0.0850 max mem: 3536 +Test: [1290/2387] eta: 0:03:41 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1595 (0.2603) acc1: 98.9583 (91.3181) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1877 data: 0.0939 max mem: 3536 +Test: [1300/2387] eta: 0:03:38 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1664 (0.2595) acc1: 98.9583 (91.3782) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1888 data: 0.0932 max mem: 3536 +Test: [1310/2387] eta: 0:03:36 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1562 (0.2588) acc1: 99.4792 (91.4372) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1771 data: 0.0831 max mem: 3536 +Test: [1320/2387] eta: 0:03:34 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1789 (0.2582) acc1: 99.2188 (91.4918) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1863 data: 0.0907 max mem: 3536 +Test: [1330/2387] eta: 0:03:32 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1997 (0.2586) acc1: 97.3958 (91.4867) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.2000 data: 0.1046 max mem: 3536 +Test: [1340/2387] eta: 0:03:30 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.3381 (0.2597) acc1: 86.4583 (91.4132) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1995 data: 0.1071 max mem: 3536 +Test: [1350/2387] eta: 0:03:28 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1165 (0.2586) acc1: 100.0000 (91.4768) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1848 data: 0.0905 max mem: 3537 +Test: [1360/2387] eta: 0:03:26 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.0882 (0.2573) acc1: 100.0000 (91.5394) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1721 data: 0.0790 max mem: 3537 +Test: [1370/2387] eta: 0:03:24 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.0776 (0.2559) acc1: 100.0000 (91.6011) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0924 max mem: 3537 +Test: [1380/2387] eta: 0:03:22 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.0768 (0.2547) acc1: 100.0000 (91.6620) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2352 data: 0.1405 max mem: 3537 +Test: [1390/2387] eta: 0:03:20 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.0827 (0.2535) acc1: 100.0000 (91.7219) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2255 data: 0.1327 max mem: 3537 +Test: [1400/2387] eta: 0:03:18 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.0978 (0.2525) acc1: 100.0000 (91.7801) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1944 data: 0.1005 max mem: 3537 +Test: [1410/2387] eta: 0:03:16 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1363 (0.2523) acc1: 99.7396 (91.8073) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2170 data: 0.1226 max mem: 3537 +Test: [1420/2387] eta: 0:03:14 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.2133 (0.2520) acc1: 96.8750 (91.8406) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2052 data: 0.1143 max mem: 3537 +Test: [1430/2387] eta: 0:03:12 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.2136 (0.2517) acc1: 96.8750 (91.8727) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1888 data: 0.0976 max mem: 3537 +Test: [1440/2387] eta: 0:03:10 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.2218 (0.2519) acc1: 95.3125 (91.8777) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.1903 data: 0.0988 max mem: 3537 +Test: [1450/2387] eta: 0:03:08 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.2897 (0.2525) acc1: 90.1042 (91.8325) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1906 data: 0.0996 max mem: 3537 +Test: [1460/2387] eta: 0:03:06 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1737 (0.2518) acc1: 96.6146 (91.8782) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1847 data: 0.0915 max mem: 3537 +Test: [1470/2387] eta: 0:03:04 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1209 (0.2509) acc1: 99.4792 (91.9301) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1836 data: 0.0897 max mem: 3537 +Test: [1480/2387] eta: 0:03:01 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1093 (0.2499) acc1: 99.7396 (91.9814) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1856 data: 0.0928 max mem: 3537 +Test: [1490/2387] eta: 0:02:59 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1113 (0.2491) acc1: 99.7396 (92.0321) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1939 data: 0.0989 max mem: 3537 +Test: [1500/2387] eta: 0:02:57 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1258 (0.2483) acc1: 99.7396 (92.0811) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1978 data: 0.1026 max mem: 3537 +Test: [1510/2387] eta: 0:02:55 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1391 (0.2477) acc1: 99.4792 (92.1299) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1967 data: 0.1013 max mem: 3537 +Test: [1520/2387] eta: 0:02:53 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1427 (0.2471) acc1: 99.7396 (92.1795) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1995 data: 0.1048 max mem: 3537 +Test: [1530/2387] eta: 0:02:51 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1234 (0.2462) acc1: 100.0000 (92.2304) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2002 data: 0.1081 max mem: 3537 +Test: [1540/2387] eta: 0:02:49 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1188 (0.2454) acc1: 100.0000 (92.2806) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1969 data: 0.1055 max mem: 3537 +Test: [1550/2387] eta: 0:02:47 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1061 (0.2445) acc1: 100.0000 (92.3299) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1978 data: 0.1046 max mem: 3537 +Test: [1560/2387] eta: 0:02:46 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1119 (0.2437) acc1: 100.0000 (92.3782) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2221 data: 0.1276 max mem: 3537 +Test: [1570/2387] eta: 0:02:43 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1408 (0.2433) acc1: 99.7396 (92.4231) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1949 data: 0.1013 max mem: 3537 +Test: [1580/2387] eta: 0:02:41 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.2035 (0.2432) acc1: 98.4375 (92.4499) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1802 data: 0.0873 max mem: 3538 +Test: [1590/2387] eta: 0:02:39 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.2035 (0.2428) acc1: 98.4375 (92.4921) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2149 data: 0.1235 max mem: 3538 +Test: [1600/2387] eta: 0:02:37 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1582 (0.2424) acc1: 98.9583 (92.5250) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2006 data: 0.1090 max mem: 3538 +Test: [1610/2387] eta: 0:02:35 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1564 (0.2419) acc1: 99.7396 (92.5671) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1825 data: 0.0892 max mem: 3538 +Test: [1620/2387] eta: 0:02:33 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1927 (0.2418) acc1: 96.3542 (92.5816) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1789 data: 0.0842 max mem: 3538 +Test: [1630/2387] eta: 0:02:31 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.2337 (0.2421) acc1: 93.7500 (92.5718) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1849 data: 0.0921 max mem: 3538 +Test: [1640/2387] eta: 0:02:29 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.3273 (0.2429) acc1: 86.4583 (92.5306) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.1838 data: 0.0904 max mem: 3538 +Test: [1650/2387] eta: 0:02:27 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.3698 (0.2436) acc1: 84.8958 (92.4888) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1822 data: 0.0884 max mem: 3538 +Test: [1660/2387] eta: 0:02:25 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.3379 (0.2442) acc1: 86.4583 (92.4471) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.1862 data: 0.0943 max mem: 3538 +Test: [1670/2387] eta: 0:02:23 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.3448 (0.2448) acc1: 85.9375 (92.4111) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.1806 data: 0.0886 max mem: 3538 +Test: [1680/2387] eta: 0:02:21 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.3448 (0.2454) acc1: 85.9375 (92.3771) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.1805 data: 0.0895 max mem: 3538 +Test: [1690/2387] eta: 0:02:19 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.3401 (0.2461) acc1: 86.9792 (92.3412) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1777 data: 0.0868 max mem: 3538 +Test: [1700/2387] eta: 0:02:16 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.3388 (0.2463) acc1: 85.1562 (92.3231) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1710 data: 0.0792 max mem: 3538 +Test: [1710/2387] eta: 0:02:14 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1055 (0.2454) acc1: 100.0000 (92.3679) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1728 data: 0.0813 max mem: 3538 +Test: [1720/2387] eta: 0:02:12 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.0862 (0.2445) acc1: 100.0000 (92.4122) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1733 data: 0.0822 max mem: 3538 +Test: [1730/2387] eta: 0:02:10 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.0794 (0.2436) acc1: 100.0000 (92.4560) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1735 data: 0.0812 max mem: 3538 +Test: [1740/2387] eta: 0:02:08 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.0826 (0.2427) acc1: 100.0000 (92.4994) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1766 data: 0.0836 max mem: 3538 +Test: [1750/2387] eta: 0:02:06 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.0944 (0.2419) acc1: 100.0000 (92.5422) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1728 data: 0.0805 max mem: 3538 +Test: [1760/2387] eta: 0:02:04 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1092 (0.2413) acc1: 100.0000 (92.5760) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1740 data: 0.0828 max mem: 3538 +Test: [1770/2387] eta: 0:02:02 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1502 (0.2408) acc1: 98.9583 (92.6126) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1734 data: 0.0820 max mem: 3538 +Test: [1780/2387] eta: 0:02:00 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1324 (0.2402) acc1: 100.0000 (92.6535) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1855 data: 0.0933 max mem: 3538 +Test: [1790/2387] eta: 0:01:58 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1124 (0.2395) acc1: 100.0000 (92.6941) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2047 data: 0.1119 max mem: 3538 +Test: [1800/2387] eta: 0:01:56 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1054 (0.2388) acc1: 100.0000 (92.7345) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1782 data: 0.0852 max mem: 3538 +Test: [1810/2387] eta: 0:01:54 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1066 (0.2381) acc1: 100.0000 (92.7733) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1616 data: 0.0700 max mem: 3539 +Test: [1820/2387] eta: 0:01:52 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1337 (0.2377) acc1: 99.7396 (92.8039) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1801 data: 0.0890 max mem: 3539 +Test: [1830/2387] eta: 0:01:50 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1671 (0.2374) acc1: 98.9583 (92.8329) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1928 data: 0.1002 max mem: 3539 +Test: [1840/2387] eta: 0:01:48 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1496 (0.2370) acc1: 100.0000 (92.8714) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1880 data: 0.0950 max mem: 3539 +Test: [1850/2387] eta: 0:01:46 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1488 (0.2365) acc1: 100.0000 (92.9092) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1796 data: 0.0867 max mem: 3539 +Test: [1860/2387] eta: 0:01:44 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1492 (0.2360) acc1: 100.0000 (92.9469) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1828 data: 0.0896 max mem: 3539 +Test: [1870/2387] eta: 0:01:42 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1492 (0.2356) acc1: 100.0000 (92.9842) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1822 data: 0.0897 max mem: 3539 +Test: [1880/2387] eta: 0:01:40 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1478 (0.2351) acc1: 100.0000 (93.0211) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1817 data: 0.0890 max mem: 3539 +Test: [1890/2387] eta: 0:01:38 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1496 (0.2346) acc1: 100.0000 (93.0570) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1795 data: 0.0872 max mem: 3539 +Test: [1900/2387] eta: 0:01:36 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1504 (0.2342) acc1: 100.0000 (93.0919) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1850 data: 0.0926 max mem: 3539 +Test: [1910/2387] eta: 0:01:34 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1990 (0.2343) acc1: 98.6979 (93.1188) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1859 data: 0.0935 max mem: 3539 +Test: [1920/2387] eta: 0:01:32 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.2458 (0.2345) acc1: 97.6562 (93.1131) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1796 data: 0.0867 max mem: 3539 +Test: [1930/2387] eta: 0:01:30 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1881 (0.2340) acc1: 98.9583 (93.1460) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1836 data: 0.0895 max mem: 3539 +Test: [1940/2387] eta: 0:01:28 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1652 (0.2339) acc1: 99.4792 (93.1450) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1872 data: 0.0928 max mem: 3539 +Test: [1950/2387] eta: 0:01:26 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1805 (0.2336) acc1: 99.4792 (93.1704) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1792 data: 0.0846 max mem: 3539 +Test: [1960/2387] eta: 0:01:24 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1913 (0.2336) acc1: 98.6979 (93.1848) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1997 data: 0.1050 max mem: 3539 +Test: [1970/2387] eta: 0:01:22 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.3035 (0.2347) acc1: 90.6250 (93.1149) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1962 data: 0.1016 max mem: 3539 +Test: [1980/2387] eta: 0:01:20 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.3631 (0.2349) acc1: 84.6354 (93.0946) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1857 data: 0.0931 max mem: 3539 +Test: [1990/2387] eta: 0:01:18 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.0913 (0.2341) acc1: 99.7396 (93.1279) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1848 data: 0.0910 max mem: 3539 +Test: [2000/2387] eta: 0:01:16 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.0904 (0.2334) acc1: 99.7396 (93.1612) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1862 data: 0.0920 max mem: 3539 +Test: [2010/2387] eta: 0:01:14 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.0896 (0.2327) acc1: 99.7396 (93.1942) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2046 data: 0.1136 max mem: 3539 +Test: [2020/2387] eta: 0:01:12 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.0909 (0.2320) acc1: 99.7396 (93.2262) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2048 data: 0.1140 max mem: 3539 +Test: [2030/2387] eta: 0:01:10 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.0898 (0.2313) acc1: 99.7396 (93.2585) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1975 data: 0.1053 max mem: 3540 +Test: [2040/2387] eta: 0:01:08 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.0893 (0.2306) acc1: 99.7396 (93.2903) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1979 data: 0.1022 max mem: 3540 +Test: [2050/2387] eta: 0:01:06 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.0893 (0.2299) acc1: 99.7396 (93.3222) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2050 data: 0.1093 max mem: 3540 +Test: [2060/2387] eta: 0:01:04 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1063 (0.2293) acc1: 99.7396 (93.3532) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1971 data: 0.1054 max mem: 3540 +Test: [2070/2387] eta: 0:01:02 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1088 (0.2288) acc1: 99.7396 (93.3846) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1975 data: 0.1066 max mem: 3540 +Test: [2080/2387] eta: 0:01:00 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1083 (0.2282) acc1: 99.7396 (93.4155) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2006 data: 0.1084 max mem: 3540 +Test: [2090/2387] eta: 0:00:58 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1082 (0.2276) acc1: 99.7396 (93.4455) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2044 data: 0.1123 max mem: 3540 +Test: [2100/2387] eta: 0:00:56 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1090 (0.2270) acc1: 99.7396 (93.4758) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2144 data: 0.1214 max mem: 3540 +Test: [2110/2387] eta: 0:00:54 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1092 (0.2265) acc1: 99.7396 (93.5050) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2143 data: 0.1203 max mem: 3540 +Test: [2120/2387] eta: 0:00:52 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1178 (0.2267) acc1: 99.2188 (93.5057) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2107 data: 0.1180 max mem: 3540 +Test: [2130/2387] eta: 0:00:50 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.2089 (0.2265) acc1: 96.6146 (93.5289) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1702 data: 0.0780 max mem: 3540 +Test: [2140/2387] eta: 0:00:48 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1835 (0.2264) acc1: 98.6979 (93.5398) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1569 data: 0.0635 max mem: 3540 +Test: [2150/2387] eta: 0:00:46 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1899 (0.2262) acc1: 98.6979 (93.5631) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1839 data: 0.0912 max mem: 3540 +Test: [2160/2387] eta: 0:00:44 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.2233 (0.2264) acc1: 97.9167 (93.5708) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2007 data: 0.1088 max mem: 3540 +Test: [2170/2387] eta: 0:00:42 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.2813 (0.2271) acc1: 90.6250 (93.5269) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.2096 data: 0.1168 max mem: 3540 +Test: [2180/2387] eta: 0:00:40 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.4000 (0.2286) acc1: 82.2917 (93.4184) acc5: 100.0000 (100.0000) acc1_neg: 53.906 acc5_neg: 100.000 time: 0.1969 data: 0.1055 max mem: 3540 +Test: [2190/2387] eta: 0:00:38 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.3638 (0.2286) acc1: 83.8542 (93.4164) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1817 data: 0.0895 max mem: 3540 +Test: [2200/2387] eta: 0:00:36 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1505 (0.2283) acc1: 99.2188 (93.4379) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1840 data: 0.0914 max mem: 3540 +Test: [2210/2387] eta: 0:00:34 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1410 (0.2279) acc1: 99.7396 (93.4666) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1922 data: 0.1009 max mem: 3540 +Test: [2220/2387] eta: 0:00:32 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1316 (0.2274) acc1: 100.0000 (93.4947) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1839 data: 0.0923 max mem: 3540 +Test: [2230/2387] eta: 0:00:30 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1316 (0.2270) acc1: 100.0000 (93.5231) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2118 data: 0.1201 max mem: 3540 +Test: [2240/2387] eta: 0:00:28 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1360 (0.2267) acc1: 99.7396 (93.5497) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2208 data: 0.1293 max mem: 3540 +Test: [2250/2387] eta: 0:00:26 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1515 (0.2265) acc1: 99.4792 (93.5744) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1550 data: 0.0629 max mem: 3540 +Test: [2260/2387] eta: 0:00:24 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1999 (0.2265) acc1: 98.6979 (93.5931) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1437 data: 0.0498 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1972 (0.2262) acc1: 98.6979 (93.6166) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1668 data: 0.0744 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1534 (0.2260) acc1: 99.4792 (93.6285) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1519 data: 0.0612 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.1534 (0.2257) acc1: 100.0000 (93.6541) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1504 data: 0.0582 max mem: 3541 +Test: [2300/2387] eta: 0:00:17 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.2006 (0.2257) acc1: 98.1771 (93.6690) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1509 data: 0.0595 max mem: 3541 +Test: [2310/2387] eta: 0:00:15 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.2510 (0.2261) acc1: 95.0521 (93.6538) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.1437 data: 0.0534 max mem: 3541 +Test: [2320/2387] eta: 0:00:13 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.3163 (0.2267) acc1: 88.2812 (93.6157) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1339 data: 0.0436 max mem: 3541 +Test: [2330/2387] eta: 0:00:11 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.3163 (0.2270) acc1: 90.6250 (93.6136) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1286 data: 0.0386 max mem: 3541 +Test: [2340/2387] eta: 0:00:09 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.2354 (0.2269) acc1: 94.5312 (93.6214) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1276 data: 0.0376 max mem: 3541 +Test: [2350/2387] eta: 0:00:07 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.2082 (0.2270) acc1: 96.8750 (93.6180) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1252 data: 0.0349 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.2339 (0.2273) acc1: 95.0521 (93.6123) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1208 data: 0.0303 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.2905 (0.2278) acc1: 91.6667 (93.5887) acc5: 100.0000 (100.0000) acc1_neg: 69.531 acc5_neg: 100.000 time: 0.1154 data: 0.0253 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.3943 (0.2288) acc1: 81.7708 (93.5239) acc5: 100.0000 (100.0000) acc1_neg: 78.646 acc5_neg: 100.000 time: 0.1115 data: 0.0218 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 77.626 acc5_pos: 100.000 loss: 0.4072 (0.2289) acc1: 80.4688 (93.5100) acc5: 100.0000 (100.0000) acc1_neg: 97.674 acc5_neg: 100.000 time: 0.1059 data: 0.0189 max mem: 3541 +Test: Total time: 0:07:41 (0.1931 s / it) +* Overall Acc@1 93.510 Acc@5 100.000 + target == real Acc@1 59.215 Acc@5 100.000 + target == fake Acc@1 95.918 Acc@5 100.000 + loss 0.229 + PR-AUC 0.9952 ROC-AUC 0.9383 EER 13.21% (thr=0.7012) +Accuracy of the network on the 916267 test images: 93.5% +[rank0]:[W725 17:13:13.826148843 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 17:13:33.564000 590149 torch/distributed/run.py:766] +W0725 17:13:33.564000 590149 torch/distributed/run.py:766] ***************************************** +W0725 17:13:33.564000 590149 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:13:33.564000 590149 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 3): env:// +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:14:03.684676765 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:14:03.973142763 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:14:03.061797912 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:14:03.379343182 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +| distributed init (rank 6): env:// +[rank0]:[W725 17:14:04.495721087 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:14:04.497227386 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:14:04.508866210 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank4]:[W725 17:14:04.508891753 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed1_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed1_resampled_n116-041-023_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:04:46 acc1_pos: 48.698 acc5_pos: 100.000 loss: 0.7107 (0.7107) acc1: 48.6979 (48.6979) acc5: 100.0000 (100.0000) time: 7.5184 data: 5.8638 max mem: 3531 +Test: [ 10/517] eta: 0:06:28 acc1_pos: 10.156 acc5_pos: 100.000 loss: 0.7956 (0.8509) acc1: 46.0938 (42.4479) acc5: 100.0000 (100.0000) time: 0.7656 data: 0.5335 max mem: 3531 +Test: [ 20/517] eta: 0:03:42 acc1_pos: 18.750 acc5_pos: 100.000 loss: 0.8952 (0.9256) acc1: 34.8958 (35.4415) acc5: 100.0000 (100.0000) time: 0.0947 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:03:06 acc1_pos: 26.823 acc5_pos: 100.000 loss: 0.8964 (0.9164) acc1: 26.8229 (35.4755) acc5: 100.0000 (100.0000) time: 0.1714 data: 0.0764 max mem: 3531 +Test: [ 40/517] eta: 0:02:46 acc1_pos: 38.281 acc5_pos: 100.000 loss: 0.8738 (0.9063) acc1: 36.4583 (35.7342) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1515 max mem: 3531 +Test: [ 50/517] eta: 0:02:36 acc1_pos: 81.193 acc5_pos: 100.000 loss: 0.8485 (0.8804) acc1: 36.7188 (39.2310) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.2598 data: 0.1642 max mem: 3531 +Test: [ 60/517] eta: 0:02:26 acc1_pos: 81.193 acc5_pos: 100.000 loss: 0.6158 (0.8302) acc1: 68.4896 (44.6038) acc5: 100.0000 (100.0000) acc1_neg: 45.833 acc5_neg: 100.000 time: 0.2599 data: 0.1649 max mem: 3531 +Test: [ 70/517] eta: 0:02:19 acc1_pos: 81.193 acc5_pos: 100.000 loss: 0.5247 (0.7860) acc1: 75.5208 (49.8093) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2581 data: 0.1611 max mem: 3531 +Test: [ 80/517] eta: 0:02:13 acc1_pos: 81.193 acc5_pos: 100.000 loss: 0.4856 (0.7496) acc1: 86.7188 (54.1345) acc5: 100.0000 (100.0000) acc1_neg: 54.948 acc5_neg: 100.000 time: 0.2659 data: 0.1705 max mem: 3531 +Test: [ 90/517] eta: 0:02:08 acc1_pos: 81.193 acc5_pos: 100.000 loss: 0.4328 (0.7121) acc1: 89.0625 (58.0100) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.2599 data: 0.1677 max mem: 3531 +Test: [100/517] eta: 0:02:03 acc1_pos: 81.193 acc5_pos: 100.000 loss: 0.4088 (0.6844) acc1: 89.0625 (61.3268) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2545 data: 0.1624 max mem: 3531 +Test: [110/517] eta: 0:01:58 acc1_pos: 81.193 acc5_pos: 100.000 loss: 0.3925 (0.6555) acc1: 94.5312 (64.5083) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2503 data: 0.1587 max mem: 3531 +Test: [120/517] eta: 0:01:54 acc1_pos: 81.193 acc5_pos: 100.000 loss: 0.4194 (0.6389) acc1: 92.7083 (66.4945) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.2494 data: 0.1531 max mem: 3531 +Test: [130/517] eta: 0:01:52 acc1_pos: 81.193 acc5_pos: 100.000 loss: 0.4363 (0.6239) acc1: 90.6250 (68.1477) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2795 data: 0.1829 max mem: 3531 +Test: [140/517] eta: 0:01:48 acc1_pos: 81.193 acc5_pos: 100.000 loss: 0.4044 (0.6056) acc1: 94.5312 (70.2479) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2799 data: 0.1852 max mem: 3531 +Test: [150/517] eta: 0:01:44 acc1_pos: 81.193 acc5_pos: 100.000 loss: 0.3599 (0.5884) acc1: 98.6979 (72.0285) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2499 data: 0.1542 max mem: 3531 +Test: [160/517] eta: 0:01:40 acc1_pos: 81.193 acc5_pos: 100.000 loss: 0.3288 (0.5722) acc1: 95.3125 (73.3793) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2449 data: 0.1480 max mem: 3531 +Test: [170/517] eta: 0:01:36 acc1_pos: 81.193 acc5_pos: 100.000 loss: 0.3218 (0.5569) acc1: 93.7500 (74.5949) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2249 data: 0.1275 max mem: 3531 +Test: [180/517] eta: 0:01:33 acc1_pos: 81.193 acc5_pos: 100.000 loss: 0.2932 (0.5435) acc1: 95.5729 (75.7079) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2299 data: 0.1356 max mem: 3531 +Test: [190/517] eta: 0:01:30 acc1_pos: 81.193 acc5_pos: 100.000 loss: 0.2745 (0.5293) acc1: 98.6979 (76.9211) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2599 data: 0.1630 max mem: 3531 +Test: [200/517] eta: 0:01:26 acc1_pos: 81.193 acc5_pos: 100.000 loss: 0.2960 (0.5205) acc1: 98.6979 (77.9333) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2549 data: 0.1554 max mem: 3531 +Test: [210/517] eta: 0:01:23 acc1_pos: 81.193 acc5_pos: 100.000 loss: 0.3774 (0.5148) acc1: 97.9167 (78.8125) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2453 data: 0.1503 max mem: 3532 +Test: [220/517] eta: 0:01:20 acc1_pos: 81.193 acc5_pos: 100.000 loss: 0.3904 (0.5100) acc1: 96.6146 (79.5284) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2549 data: 0.1606 max mem: 3532 +Test: [230/517] eta: 0:01:17 acc1_pos: 81.193 acc5_pos: 100.000 loss: 0.3874 (0.5049) acc1: 97.6562 (80.2004) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.2498 data: 0.1552 max mem: 3532 +Test: [240/517] eta: 0:01:16 acc1_pos: 81.193 acc5_pos: 100.000 loss: 0.3693 (0.5003) acc1: 97.6562 (80.7270) acc5: 100.0000 (100.0000) acc1_neg: 71.615 acc5_neg: 100.000 time: 0.3099 data: 0.2138 max mem: 3532 +Test: [250/517] eta: 0:01:13 acc1_pos: 81.193 acc5_pos: 100.000 loss: 0.4116 (0.4978) acc1: 96.0938 (81.1743) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.3046 data: 0.2059 max mem: 3532 +Test: [260/517] eta: 0:01:10 acc1_pos: 81.193 acc5_pos: 100.000 loss: 0.4192 (0.4951) acc1: 96.0938 (81.6381) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2653 data: 0.1689 max mem: 3532 +Test: [270/517] eta: 0:01:07 acc1_pos: 81.193 acc5_pos: 100.000 loss: 0.4154 (0.4924) acc1: 95.8333 (82.1370) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2599 data: 0.1648 max mem: 3532 +Test: [280/517] eta: 0:01:04 acc1_pos: 81.193 acc5_pos: 100.000 loss: 0.3868 (0.4885) acc1: 95.0521 (82.4214) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2251 data: 0.1300 max mem: 3532 +Test: [290/517] eta: 0:01:01 acc1_pos: 81.193 acc5_pos: 100.000 loss: 0.3359 (0.4830) acc1: 96.3542 (82.9548) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2605 data: 0.1670 max mem: 3532 +Test: [300/517] eta: 0:00:58 acc1_pos: 81.193 acc5_pos: 100.000 loss: 0.3039 (0.4768) acc1: 99.2188 (83.5046) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2643 data: 0.1712 max mem: 3532 +Test: [310/517] eta: 0:00:55 acc1_pos: 81.193 acc5_pos: 100.000 loss: 0.3319 (0.4743) acc1: 99.2188 (83.9513) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2493 data: 0.1554 max mem: 3532 +Test: [320/517] eta: 0:00:53 acc1_pos: 81.193 acc5_pos: 100.000 loss: 0.4072 (0.4725) acc1: 97.6562 (84.2801) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.2699 data: 0.1759 max mem: 3532 +Test: [330/517] eta: 0:00:50 acc1_pos: 81.193 acc5_pos: 100.000 loss: 0.3904 (0.4697) acc1: 97.6562 (84.6984) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.3049 data: 0.2103 max mem: 3532 +Test: [340/517] eta: 0:00:48 acc1_pos: 81.193 acc5_pos: 100.000 loss: 0.3572 (0.4678) acc1: 99.4792 (84.7683) acc5: 100.0000 (100.0000) acc1_neg: 56.510 acc5_neg: 100.000 time: 0.2799 data: 0.1841 max mem: 3532 +Test: [350/517] eta: 0:00:45 acc1_pos: 81.193 acc5_pos: 100.000 loss: 0.4101 (0.4676) acc1: 92.7083 (84.7660) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.2399 data: 0.1452 max mem: 3532 +Test: [360/517] eta: 0:00:42 acc1_pos: 81.193 acc5_pos: 100.000 loss: 0.4334 (0.4668) acc1: 87.2396 (84.8186) acc5: 100.0000 (100.0000) acc1_neg: 74.740 acc5_neg: 100.000 time: 0.2499 data: 0.1562 max mem: 3532 +Test: [370/517] eta: 0:00:39 acc1_pos: 81.193 acc5_pos: 100.000 loss: 0.4433 (0.4666) acc1: 86.7188 (84.9127) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.2549 data: 0.1579 max mem: 3532 +Test: [380/517] eta: 0:00:36 acc1_pos: 81.193 acc5_pos: 100.000 loss: 0.4570 (0.4663) acc1: 86.7188 (84.9300) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.2599 data: 0.1604 max mem: 3532 +Test: [390/517] eta: 0:00:34 acc1_pos: 81.193 acc5_pos: 100.000 loss: 0.4781 (0.4672) acc1: 83.5938 (84.8472) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.2866 data: 0.1896 max mem: 3532 +Test: [400/517] eta: 0:00:31 acc1_pos: 81.193 acc5_pos: 100.000 loss: 0.4534 (0.4660) acc1: 84.6354 (84.9699) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.3049 data: 0.2081 max mem: 3532 +Test: [410/517] eta: 0:00:28 acc1_pos: 81.193 acc5_pos: 100.000 loss: 0.4477 (0.4664) acc1: 87.2396 (84.8813) acc5: 100.0000 (100.0000) acc1_neg: 71.615 acc5_neg: 100.000 time: 0.2732 data: 0.1761 max mem: 3532 +Test: [420/517] eta: 0:00:26 acc1_pos: 81.193 acc5_pos: 100.000 loss: 0.4936 (0.4676) acc1: 82.8125 (84.8173) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2599 data: 0.1645 max mem: 3532 +Test: [430/517] eta: 0:00:23 acc1_pos: 81.193 acc5_pos: 100.000 loss: 0.4948 (0.4685) acc1: 83.0729 (84.7502) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.2699 data: 0.1749 max mem: 3532 +Test: [440/517] eta: 0:00:20 acc1_pos: 81.193 acc5_pos: 100.000 loss: 0.4437 (0.4675) acc1: 89.8438 (84.9006) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.2649 data: 0.1709 max mem: 3533 +Test: [450/517] eta: 0:00:18 acc1_pos: 81.193 acc5_pos: 100.000 loss: 0.4437 (0.4669) acc1: 91.6667 (85.0575) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.2649 data: 0.1700 max mem: 3533 +Test: [460/517] eta: 0:00:15 acc1_pos: 81.193 acc5_pos: 100.000 loss: 0.4502 (0.4663) acc1: 91.1458 (85.1772) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.2749 data: 0.1755 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 81.193 acc5_pos: 100.000 loss: 0.4397 (0.4658) acc1: 89.5833 (85.2276) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2599 data: 0.1601 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 81.193 acc5_pos: 100.000 loss: 0.4198 (0.4645) acc1: 89.5833 (85.3885) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2499 data: 0.1518 max mem: 3533 +Test: [490/517] eta: 0:00:07 acc1_pos: 81.193 acc5_pos: 100.000 loss: 0.3710 (0.4622) acc1: 95.3125 (85.6256) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2334 data: 0.1387 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 81.193 acc5_pos: 100.000 loss: 0.4222 (0.4641) acc1: 93.7500 (85.4130) acc5: 100.0000 (100.0000) acc1_neg: 60.677 acc5_neg: 100.000 time: 0.1960 data: 0.1060 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 81.193 acc5_pos: 100.000 loss: 0.4731 (0.4651) acc1: 86.4583 (85.3260) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1731 data: 0.0834 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 81.193 acc5_pos: 100.000 loss: 0.4637 (0.4652) acc1: 86.9258 (85.3200) acc5: 100.0000 (100.0000) acc1_neg: 86.926 acc5_neg: 100.000 time: 0.1658 data: 0.0744 max mem: 3533 +Test: Total time: 0:02:16 (0.2647 s / it) +* Overall Acc@1 85.320 Acc@5 100.000 + target == real Acc@1 36.869 Acc@5 100.000 + target == fake Acc@1 90.346 Acc@5 100.000 + loss 0.465 + PR-AUC 0.9442 ROC-AUC 0.6963 EER 35.94% (thr=0.6362) +Accuracy of the network on the 198427 test images: 85.3% +[rank0]:[W725 17:16:29.482897916 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0725 17:16:49.017000 590738 torch/distributed/run.py:766] +W0725 17:16:49.017000 590738 torch/distributed/run.py:766] ***************************************** +W0725 17:16:49.017000 590738 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:16:49.017000 590738 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:17:16.395874651 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:17:17.095979104 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:17:18.490908763 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:17:18.533835103 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:17:18.710242208 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 17:17:18.743955843 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:17:18.757970718 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank4]:[W725 17:17:18.758055295 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed1_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed1_resampled_n116-041-023_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:29:58 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.6785 (2.6785) acc1: 0.0000 (0.0000) acc5: 100.0000 (100.0000) time: 4.6236 data: 2.4726 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/389] eta: 0:03:31 acc1_pos: 2.083 acc5_pos: 100.000 loss: 2.3009 (2.3868) acc1: 0.0000 (0.1894) acc5: 100.0000 (100.0000) time: 0.5573 data: 0.2683 max mem: 3531 +Test: [ 20/389] eta: 0:02:08 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.2959 (2.3372) acc1: 0.0000 (0.8185) acc5: 100.0000 (100.0000) time: 0.1356 data: 0.0387 max mem: 3531 +Test: [ 30/389] eta: 0:01:51 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.2040 (2.3134) acc1: 0.0000 (0.5880) acc5: 100.0000 (100.0000) time: 0.1738 data: 0.0809 max mem: 3531 +Test: [ 40/389] eta: 0:01:42 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.2369 (2.3342) acc1: 0.0000 (0.4510) acc5: 100.0000 (100.0000) time: 0.2345 data: 0.1409 max mem: 3531 +Test: [ 50/389] eta: 0:01:37 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.0951 (2.2749) acc1: 0.0000 (0.3625) acc5: 100.0000 (100.0000) time: 0.2496 data: 0.1554 max mem: 3531 +Test: [ 60/389] eta: 0:01:28 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.0815 (2.2918) acc1: 0.0000 (0.3031) acc5: 100.0000 (100.0000) time: 0.2145 data: 0.1183 max mem: 3531 +Test: [ 70/389] eta: 0:01:23 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.2700 (2.2852) acc1: 0.0000 (0.5172) acc5: 100.0000 (100.0000) time: 0.2049 data: 0.1095 max mem: 3531 +Test: [ 80/389] eta: 0:01:18 acc1_pos: 34.896 acc5_pos: 100.000 loss: 2.2881 (2.2634) acc1: 0.0000 (2.1734) acc5: 100.0000 (100.0000) time: 0.2092 data: 0.1143 max mem: 3531 +Test: [ 90/389] eta: 0:01:13 acc1_pos: 2.083 acc5_pos: 100.000 loss: 1.8016 (2.1858) acc1: 2.3438 (3.5228) acc5: 100.0000 (100.0000) time: 0.1799 data: 0.0857 max mem: 3531 +Test: [100/389] eta: 0:01:09 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.0446 (2.2080) acc1: 0.5208 (3.2952) acc5: 100.0000 (100.0000) time: 0.1948 data: 0.1024 max mem: 3531 +Test: [110/389] eta: 0:01:05 acc1_pos: 73.698 acc5_pos: 100.000 loss: 2.3485 (2.1952) acc1: 0.0000 (4.5280) acc5: 100.0000 (100.0000) time: 0.1861 data: 0.0948 max mem: 3531 +Test: [120/389] eta: 0:01:02 acc1_pos: 1.562 acc5_pos: 100.000 loss: 2.2750 (2.1809) acc1: 0.2604 (5.2578) acc5: 100.0000 (100.0000) time: 0.1847 data: 0.0930 max mem: 3531 +Test: [130/389] eta: 0:00:59 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.1960 (2.0617) acc1: 22.6562 (10.6393) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1940 data: 0.1017 max mem: 3531 +Test: [140/389] eta: 0:00:56 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0696 (1.9216) acc1: 100.0000 (16.9585) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1894 data: 0.0978 max mem: 3531 +Test: [150/389] eta: 0:00:53 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0696 (1.8038) acc1: 100.0000 (22.3406) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2041 data: 0.1110 max mem: 3531 +Test: [160/389] eta: 0:00:50 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0954 (1.6962) acc1: 100.0000 (27.1642) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1760 data: 0.0823 max mem: 3531 +Test: [170/389] eta: 0:00:47 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0660 (1.6011) acc1: 100.0000 (31.4221) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1727 data: 0.0804 max mem: 3531 +Test: [180/389] eta: 0:00:45 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0661 (1.5173) acc1: 100.0000 (35.1865) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1862 data: 0.0951 max mem: 3531 +Test: [190/389] eta: 0:00:42 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0767 (1.4438) acc1: 100.0000 (38.4271) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1738 data: 0.0825 max mem: 3531 +Test: [200/389] eta: 0:00:40 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0589 (1.3753) acc1: 100.0000 (41.4879) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1766 data: 0.0848 max mem: 3531 +Test: [210/389] eta: 0:00:37 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0603 (1.3131) acc1: 100.0000 (44.2610) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1773 data: 0.0848 max mem: 3532 +Test: [220/389] eta: 0:00:35 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0473 (1.2559) acc1: 100.0000 (46.7831) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1977 data: 0.1043 max mem: 3532 +Test: [230/389] eta: 0:00:33 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0551 (1.2058) acc1: 100.0000 (49.0857) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1937 data: 0.1015 max mem: 3532 +Test: [240/389] eta: 0:00:31 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0844 (1.1594) acc1: 100.0000 (51.1983) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1810 data: 0.0879 max mem: 3532 +Test: [250/389] eta: 0:00:28 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0610 (1.1157) acc1: 100.0000 (53.1426) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1860 data: 0.0929 max mem: 3532 +Test: [260/389] eta: 0:00:26 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0496 (1.0763) acc1: 100.0000 (54.9379) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1842 data: 0.0924 max mem: 3532 +Test: [270/389] eta: 0:00:24 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0688 (1.0395) acc1: 100.0000 (56.6007) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1826 data: 0.0916 max mem: 3532 +Test: [280/389] eta: 0:00:22 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0680 (1.0051) acc1: 100.0000 (58.1452) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1873 data: 0.0956 max mem: 3532 +Test: [290/389] eta: 0:00:20 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0823 (0.9739) acc1: 100.0000 (59.5781) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1821 data: 0.0893 max mem: 3532 +Test: [300/389] eta: 0:00:18 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0682 (0.9437) acc1: 100.0000 (60.9211) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0780 max mem: 3532 +Test: [310/389] eta: 0:00:15 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0661 (0.9155) acc1: 100.0000 (62.1776) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1759 data: 0.0850 max mem: 3532 +Test: [320/389] eta: 0:00:13 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0661 (0.8890) acc1: 100.0000 (63.3551) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1847 data: 0.0940 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0729 (0.8644) acc1: 100.0000 (64.4622) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1839 data: 0.0912 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0715 (0.8409) acc1: 100.0000 (65.5028) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1721 data: 0.0792 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0595 (0.8277) acc1: 100.0000 (65.7474) acc5: 100.0000 (100.0000) acc1_neg: 27.344 acc5_neg: 100.000 time: 0.1465 data: 0.0558 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2590 (0.8159) acc1: 95.0521 (66.1213) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1450 data: 0.0547 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2030 (0.7966) acc1: 96.8750 (67.0275) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1456 data: 0.0555 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0608 (0.7772) acc1: 100.0000 (67.8929) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1133 data: 0.0236 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0608 (0.7629) acc1: 100.0000 (68.5487) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1209 data: 0.0265 max mem: 3532 +Test: Total time: 0:01:14 (0.1917 s / it) +* Overall Acc@1 68.549 Acc@5 100.000 + target == real Acc@1 5.155 Acc@5 100.000 + target == fake Acc@1 98.016 Acc@5 100.000 + loss 0.763 + PR-AUC 0.8306 ROC-AUC 0.7149 EER 33.78% (thr=0.9253) +Accuracy of the network on the 149358 test images: 68.5% +[rank0]:[W725 17:18:42.850516208 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 17:19:01.943000 591309 torch/distributed/run.py:766] +W0725 17:19:01.943000 591309 torch/distributed/run.py:766] ***************************************** +W0725 17:19:01.943000 591309 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:19:01.943000 591309 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:19:31.010467780 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:19:31.317830337 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:19:31.322719117 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 17:19:32.419443307 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:19:32.461287715 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +NCCL version 2.26.2+cuda12.2 +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:19:32.552611127 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:19:32.613138057 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:19:32.637373930 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed1_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed1_resampled_n116-041-023_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:34:28 acc1_pos: 91.406 acc5_pos: 100.000 loss: 0.2703 (0.2703) acc1: 91.4062 (91.4062) acc5: 100.0000 (100.0000) time: 6.2106 data: 4.6869 max mem: 3531 +Test: [ 10/333] eta: 0:03:28 acc1_pos: 90.625 acc5_pos: 100.000 loss: 0.1995 (0.2114) acc1: 95.8333 (94.6733) acc5: 100.0000 (100.0000) time: 0.6465 data: 0.4265 max mem: 3531 +Test: [ 20/333] eta: 0:02:00 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.2138 (0.2216) acc1: 95.8333 (94.5065) acc5: 100.0000 (100.0000) time: 0.0949 data: 0.0004 max mem: 3531 +Test: [ 30/333] eta: 0:01:44 acc1_pos: 60.156 acc5_pos: 100.000 loss: 0.2854 (0.3768) acc1: 81.5104 (82.5017) acc5: 100.0000 (100.0000) time: 0.1800 data: 0.0821 max mem: 3531 +Test: [ 40/333] eta: 0:01:34 acc1_pos: 60.417 acc5_pos: 100.000 loss: 0.7185 (0.4609) acc1: 56.5104 (75.9972) acc5: 100.0000 (100.0000) time: 0.2554 data: 0.1605 max mem: 3531 +Test: [ 50/333] eta: 0:01:24 acc1_pos: 61.719 acc5_pos: 100.000 loss: 0.7297 (0.5181) acc1: 56.5104 (71.4716) acc5: 100.0000 (100.0000) time: 0.2274 data: 0.1343 max mem: 3531 +Test: [ 60/333] eta: 0:01:15 acc1_pos: 43.750 acc5_pos: 100.000 loss: 0.7316 (0.5570) acc1: 52.6042 (68.1993) acc5: 100.0000 (100.0000) time: 0.1806 data: 0.0892 max mem: 3531 +Test: [ 70/333] eta: 0:01:09 acc1_pos: 48.958 acc5_pos: 100.000 loss: 0.7385 (0.5844) acc1: 54.1667 (66.5933) acc5: 100.0000 (100.0000) time: 0.1784 data: 0.0874 max mem: 3531 +Test: [ 80/333] eta: 0:01:05 acc1_pos: 55.729 acc5_pos: 100.000 loss: 0.7385 (0.6046) acc1: 55.7292 (65.0881) acc5: 100.0000 (100.0000) time: 0.2043 data: 0.1132 max mem: 3531 +Test: [ 90/333] eta: 0:01:01 acc1_pos: 59.375 acc5_pos: 100.000 loss: 0.7158 (0.6162) acc1: 55.7292 (64.2142) acc5: 100.0000 (100.0000) time: 0.2049 data: 0.1110 max mem: 3531 +Test: [100/333] eta: 0:00:57 acc1_pos: 68.229 acc5_pos: 100.000 loss: 0.7129 (0.6272) acc1: 57.0312 (63.5391) acc5: 100.0000 (100.0000) time: 0.1954 data: 0.1014 max mem: 3531 +Test: [110/333] eta: 0:00:53 acc1_pos: 53.906 acc5_pos: 100.000 loss: 0.7129 (0.6335) acc1: 57.5521 (63.0912) acc5: 100.0000 (100.0000) time: 0.1907 data: 0.1002 max mem: 3531 +Test: [120/333] eta: 0:00:50 acc1_pos: 52.344 acc5_pos: 100.000 loss: 0.6834 (0.6386) acc1: 57.5521 (62.7626) acc5: 100.0000 (100.0000) time: 0.1968 data: 0.1041 max mem: 3531 +Test: [130/333] eta: 0:00:47 acc1_pos: 70.833 acc5_pos: 100.000 loss: 0.6705 (0.6421) acc1: 59.3750 (62.4066) acc5: 100.0000 (100.0000) time: 0.2030 data: 0.1098 max mem: 3531 +Test: [140/333] eta: 0:00:44 acc1_pos: 68.490 acc5_pos: 100.000 loss: 0.6673 (0.6446) acc1: 60.4167 (62.3818) acc5: 100.0000 (100.0000) time: 0.2044 data: 0.1097 max mem: 3531 +Test: [150/333] eta: 0:00:42 acc1_pos: 58.073 acc5_pos: 100.000 loss: 0.6985 (0.6490) acc1: 58.0729 (62.0326) acc5: 100.0000 (100.0000) time: 0.2111 data: 0.1169 max mem: 3531 +Test: [160/333] eta: 0:00:39 acc1_pos: 44.737 acc5_pos: 100.000 loss: 0.7106 (0.6592) acc1: 55.4688 (61.2206) acc5: 100.0000 (100.0000) acc1_neg: 41.146 acc5_neg: 100.000 time: 0.2161 data: 0.1243 max mem: 3531 +Test: [170/333] eta: 0:00:37 acc1_pos: 44.737 acc5_pos: 100.000 loss: 0.7872 (0.6684) acc1: 53.1250 (60.6512) acc5: 100.0000 (100.0000) acc1_neg: 55.469 acc5_neg: 100.000 time: 0.2056 data: 0.1130 max mem: 3531 +Test: [180/333] eta: 0:00:34 acc1_pos: 44.737 acc5_pos: 100.000 loss: 0.7475 (0.6714) acc1: 55.4688 (60.5605) acc5: 100.0000 (100.0000) acc1_neg: 52.865 acc5_neg: 100.000 time: 0.1960 data: 0.1040 max mem: 3531 +Test: [190/333] eta: 0:00:32 acc1_pos: 44.737 acc5_pos: 100.000 loss: 0.6970 (0.6726) acc1: 59.6354 (60.6416) acc5: 100.0000 (100.0000) acc1_neg: 60.938 acc5_neg: 100.000 time: 0.1843 data: 0.0930 max mem: 3531 +Test: [200/333] eta: 0:00:29 acc1_pos: 44.737 acc5_pos: 100.000 loss: 0.6913 (0.6731) acc1: 60.1562 (60.8351) acc5: 100.0000 (100.0000) acc1_neg: 57.031 acc5_neg: 100.000 time: 0.1858 data: 0.0946 max mem: 3531 +Test: [210/333] eta: 0:00:27 acc1_pos: 44.737 acc5_pos: 100.000 loss: 0.7356 (0.6778) acc1: 58.0729 (60.6401) acc5: 100.0000 (100.0000) acc1_neg: 49.740 acc5_neg: 100.000 time: 0.2007 data: 0.1082 max mem: 3532 +Test: [220/333] eta: 0:00:25 acc1_pos: 44.737 acc5_pos: 100.000 loss: 0.7607 (0.6825) acc1: 56.7708 (60.4143) acc5: 100.0000 (100.0000) acc1_neg: 77.344 acc5_neg: 100.000 time: 0.2149 data: 0.1230 max mem: 3532 +Test: [230/333] eta: 0:00:22 acc1_pos: 44.737 acc5_pos: 100.000 loss: 0.5114 (0.6734) acc1: 76.5625 (61.3129) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.2028 data: 0.1119 max mem: 3532 +Test: [240/333] eta: 0:00:20 acc1_pos: 44.737 acc5_pos: 100.000 loss: 0.5114 (0.6696) acc1: 76.5625 (61.6474) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.1821 data: 0.0894 max mem: 3532 +Test: [250/333] eta: 0:00:18 acc1_pos: 44.737 acc5_pos: 100.000 loss: 0.6172 (0.6704) acc1: 66.1458 (61.6358) acc5: 100.0000 (100.0000) acc1_neg: 50.521 acc5_neg: 100.000 time: 0.1825 data: 0.0898 max mem: 3532 +Test: [260/333] eta: 0:00:15 acc1_pos: 44.737 acc5_pos: 100.000 loss: 0.7420 (0.6731) acc1: 58.5938 (61.5481) acc5: 100.0000 (100.0000) acc1_neg: 58.594 acc5_neg: 100.000 time: 0.1819 data: 0.0906 max mem: 3532 +Test: [270/333] eta: 0:00:13 acc1_pos: 44.737 acc5_pos: 100.000 loss: 0.6635 (0.6719) acc1: 62.5000 (61.7236) acc5: 100.0000 (100.0000) acc1_neg: 70.573 acc5_neg: 100.000 time: 0.1847 data: 0.0923 max mem: 3532 +Test: [280/333] eta: 0:00:11 acc1_pos: 44.737 acc5_pos: 100.000 loss: 0.6454 (0.6745) acc1: 63.2812 (61.5186) acc5: 100.0000 (100.0000) acc1_neg: 66.667 acc5_neg: 100.000 time: 0.1862 data: 0.0944 max mem: 3532 +Test: [290/333] eta: 0:00:09 acc1_pos: 44.737 acc5_pos: 100.000 loss: 0.8162 (0.6798) acc1: 50.5208 (61.0941) acc5: 100.0000 (100.0000) acc1_neg: 54.948 acc5_neg: 100.000 time: 0.1793 data: 0.0888 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 44.737 acc5_pos: 100.000 loss: 0.8382 (0.6854) acc1: 48.9583 (60.7082) acc5: 100.0000 (100.0000) acc1_neg: 55.208 acc5_neg: 100.000 time: 0.1681 data: 0.0767 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 44.737 acc5_pos: 100.000 loss: 0.7535 (0.6830) acc1: 55.2083 (60.9626) acc5: 100.0000 (100.0000) acc1_neg: 67.969 acc5_neg: 100.000 time: 0.1492 data: 0.0579 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 44.737 acc5_pos: 100.000 loss: 0.5918 (0.6814) acc1: 69.0104 (61.1225) acc5: 100.0000 (100.0000) acc1_neg: 43.490 acc5_neg: 100.000 time: 0.1351 data: 0.0443 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 44.737 acc5_pos: 100.000 loss: 0.8440 (0.6888) acc1: 47.3958 (60.6204) acc5: 100.0000 (100.0000) acc1_neg: 47.396 acc5_neg: 100.000 time: 0.1314 data: 0.0408 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 44.737 acc5_pos: 100.000 loss: 0.8878 (0.6901) acc1: 46.1972 (60.5430) acc5: 100.0000 (100.0000) acc1_neg: 46.197 acc5_neg: 100.000 time: 0.1235 data: 0.0300 max mem: 3532 +Test: Total time: 0:01:07 (0.2040 s / it) +* Overall Acc@1 60.543 Acc@5 100.000 + target == real Acc@1 61.773 Acc@5 100.000 + target == fake Acc@1 59.449 Acc@5 100.000 + loss 0.690 + PR-AUC 0.6526 ROC-AUC 0.6468 EER 39.39% (thr=0.4910) +Accuracy of the network on the 127843 test images: 60.5% +[rank0]:[W725 17:20:50.647951293 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 17:21:10.898000 591901 torch/distributed/run.py:766] +W0725 17:21:10.898000 591901 torch/distributed/run.py:766] ***************************************** +W0725 17:21:10.898000 591901 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:21:10.898000 591901 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:21:40.491847725 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:21:40.687284387 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:21:40.006080639 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:21:40.197386521 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:21:41.451780348 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:21:41.476504794 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:21:41.485397367 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 17:21:41.495539210 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed1_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed1_resampled_n116-041-023_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:23:54 acc1_pos: 92.448 acc5_pos: 100.000 loss: 0.2607 (0.2607) acc1: 92.4479 (92.4479) acc5: 100.0000 (100.0000) time: 6.1029 data: 4.5594 max mem: 3531 +Test: [ 10/235] eta: 0:02:23 acc1_pos: 90.885 acc5_pos: 100.000 loss: 0.2620 (0.2638) acc1: 92.4479 (92.4006) acc5: 100.0000 (100.0000) time: 0.6381 data: 0.4149 max mem: 3531 +Test: [ 20/235] eta: 0:01:22 acc1_pos: 92.188 acc5_pos: 100.000 loss: 0.2554 (0.2571) acc1: 93.2292 (93.1052) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0004 max mem: 3531 +Test: [ 30/235] eta: 0:01:09 acc1_pos: 93.229 acc5_pos: 100.000 loss: 0.2507 (0.2565) acc1: 93.4896 (93.1956) acc5: 100.0000 (100.0000) time: 0.1770 data: 0.0819 max mem: 3531 +Test: [ 40/235] eta: 0:01:01 acc1_pos: 93.750 acc5_pos: 100.000 loss: 0.2507 (0.2553) acc1: 93.4896 (93.2800) acc5: 100.0000 (100.0000) time: 0.2466 data: 0.1545 max mem: 3531 +Test: [ 50/235] eta: 0:00:54 acc1_pos: 89.583 acc5_pos: 100.000 loss: 0.2546 (0.2564) acc1: 93.4896 (93.1883) acc5: 100.0000 (100.0000) time: 0.2277 data: 0.1309 max mem: 3531 +Test: [ 60/235] eta: 0:00:49 acc1_pos: 93.750 acc5_pos: 100.000 loss: 0.2570 (0.2568) acc1: 93.2292 (93.2505) acc5: 100.0000 (100.0000) time: 0.2151 data: 0.1178 max mem: 3531 +Test: [ 70/235] eta: 0:00:45 acc1_pos: 95.385 acc5_pos: 100.000 loss: 0.2759 (0.3882) acc1: 91.9271 (84.5694) acc5: 100.0000 (100.0000) acc1_neg: 20.312 acc5_neg: 100.000 time: 0.2163 data: 0.1231 max mem: 3531 +Test: [ 80/235] eta: 0:00:41 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.3032 (0.5034) acc1: 24.2188 (77.0383) acc5: 100.0000 (100.0000) acc1_neg: 20.833 acc5_neg: 100.000 time: 0.2280 data: 0.1351 max mem: 3531 +Test: [ 90/235] eta: 0:00:38 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.3245 (0.5933) acc1: 23.4375 (71.1310) acc5: 100.0000 (100.0000) acc1_neg: 23.958 acc5_neg: 100.000 time: 0.2241 data: 0.1324 max mem: 3531 +Test: [100/235] eta: 0:00:34 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.3245 (0.6639) acc1: 22.9167 (66.4088) acc5: 100.0000 (100.0000) acc1_neg: 22.917 acc5_neg: 100.000 time: 0.2149 data: 0.1225 max mem: 3531 +Test: [110/235] eta: 0:00:31 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.3062 (0.7222) acc1: 23.1771 (62.6243) acc5: 100.0000 (100.0000) acc1_neg: 23.698 acc5_neg: 100.000 time: 0.2143 data: 0.1207 max mem: 3531 +Test: [120/235] eta: 0:00:28 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.3567 (0.7873) acc1: 21.3542 (58.5077) acc5: 100.0000 (100.0000) acc1_neg: 11.979 acc5_neg: 100.000 time: 0.2099 data: 0.1161 max mem: 3531 +Test: [130/235] eta: 0:00:25 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.5215 (0.8432) acc1: 12.2396 (55.0294) acc5: 100.0000 (100.0000) acc1_neg: 12.760 acc5_neg: 100.000 time: 0.2074 data: 0.1128 max mem: 3531 +Test: [140/235] eta: 0:00:23 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.5437 (0.8934) acc1: 11.9792 (51.9577) acc5: 100.0000 (100.0000) acc1_neg: 8.854 acc5_neg: 100.000 time: 0.2062 data: 0.1133 max mem: 3531 +Test: [150/235] eta: 0:00:20 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.5390 (0.9355) acc1: 11.7188 (49.3222) acc5: 100.0000 (100.0000) acc1_neg: 12.240 acc5_neg: 100.000 time: 0.2124 data: 0.1166 max mem: 3531 +Test: [160/235] eta: 0:00:18 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.5303 (0.9728) acc1: 11.9792 (46.9915) acc5: 100.0000 (100.0000) acc1_neg: 11.979 acc5_neg: 100.000 time: 0.2194 data: 0.1235 max mem: 3531 +Test: [170/235] eta: 0:00:15 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.4769 (0.9845) acc1: 13.5417 (46.1760) acc5: 100.0000 (100.0000) acc1_neg: 30.729 acc5_neg: 100.000 time: 0.2165 data: 0.1228 max mem: 3531 +Test: [180/235] eta: 0:00:13 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.1336 (0.9928) acc1: 35.4167 (45.5830) acc5: 100.0000 (100.0000) acc1_neg: 37.500 acc5_neg: 100.000 time: 0.2289 data: 0.1331 max mem: 3531 +Test: [190/235] eta: 0:00:10 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.1260 (1.0002) acc1: 35.4167 (44.9989) acc5: 100.0000 (100.0000) acc1_neg: 34.375 acc5_neg: 100.000 time: 0.1993 data: 0.1052 max mem: 3531 +Test: [200/235] eta: 0:00:08 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.1383 (1.0082) acc1: 33.5938 (44.4030) acc5: 100.0000 (100.0000) acc1_neg: 34.896 acc5_neg: 100.000 time: 0.1605 data: 0.0686 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.1537 (1.0144) acc1: 33.5938 (43.9808) acc5: 100.0000 (100.0000) acc1_neg: 35.677 acc5_neg: 100.000 time: 0.1546 data: 0.0630 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.1303 (1.0194) acc1: 35.4167 (43.6062) acc5: 100.0000 (100.0000) acc1_neg: 38.021 acc5_neg: 100.000 time: 0.1250 data: 0.0345 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.1186 (1.0237) acc1: 36.1979 (43.2878) acc5: 100.0000 (100.0000) acc1_neg: 37.500 acc5_neg: 100.000 time: 0.1100 data: 0.0197 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.1173 (1.0268) acc1: 36.1979 (43.1914) acc5: 100.0000 (100.0000) acc1_neg: 30.435 acc5_neg: 100.000 time: 0.1089 data: 0.0219 max mem: 3532 +Test: Total time: 0:00:50 (0.2158 s / it) +* Overall Acc@1 43.191 Acc@5 100.000 + target == real Acc@1 93.235 Acc@5 100.000 + target == fake Acc@1 25.092 Acc@5 100.000 + loss 1.027 + PR-AUC 0.8623 ROC-AUC 0.7067 EER 34.95% (thr=0.1960) +Accuracy of the network on the 89879 test images: 43.2% +[rank0]:[W725 17:22:41.669405163 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 17:22:59.419000 592452 torch/distributed/run.py:766] +W0725 17:22:59.419000 592452 torch/distributed/run.py:766] ***************************************** +W0725 17:22:59.419000 592452 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:22:59.419000 592452 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:23:29.214336944 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:23:29.316586244 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 17:23:29.379878121 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:23:30.447316505 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +NCCL version 2.26.2+cuda12.2 +[rank5]:[W725 17:23:30.622762641 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:23:30.765708243 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:23:30.786154721 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W725 17:23:30.789692748 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed1_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed1_resampled_n116-041-023_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:21:38 acc1_pos: 47.396 acc5_pos: 100.000 loss: 0.7556 (0.7556) acc1: 47.3958 (47.3958) acc5: 100.0000 (100.0000) time: 4.8262 data: 2.8157 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/269] eta: 0:02:21 acc1_pos: 49.219 acc5_pos: 100.000 loss: 0.7272 (0.6800) acc1: 49.2188 (55.6345) acc5: 100.0000 (100.0000) time: 0.5452 data: 0.2809 max mem: 3531 +Test: [ 20/269] eta: 0:01:25 acc1_pos: 69.792 acc5_pos: 100.000 loss: 0.6451 (0.6588) acc1: 58.0729 (58.3333) acc5: 100.0000 (100.0000) time: 0.1188 data: 0.0284 max mem: 3531 +Test: [ 30/269] eta: 0:01:14 acc1_pos: 61.979 acc5_pos: 100.000 loss: 0.6417 (0.6577) acc1: 58.8542 (58.9718) acc5: 100.0000 (100.0000) time: 0.1858 data: 0.0931 max mem: 3531 +Test: [ 40/269] eta: 0:01:10 acc1_pos: 28.448 acc5_pos: 100.000 loss: 0.5977 (0.6347) acc1: 64.8438 (62.3984) acc5: 100.0000 (100.0000) acc1_neg: 79.427 acc5_neg: 100.000 time: 0.2728 data: 0.1795 max mem: 3531 +Test: [ 50/269] eta: 0:01:01 acc1_pos: 28.448 acc5_pos: 100.000 loss: 0.5318 (0.6115) acc1: 77.8646 (65.7169) acc5: 100.0000 (100.0000) acc1_neg: 55.729 acc5_neg: 100.000 time: 0.2311 data: 0.1392 max mem: 3531 +Test: [ 60/269] eta: 0:00:55 acc1_pos: 28.448 acc5_pos: 100.000 loss: 0.5417 (0.6083) acc1: 74.4792 (66.1117) acc5: 100.0000 (100.0000) acc1_neg: 61.198 acc5_neg: 100.000 time: 0.1731 data: 0.0791 max mem: 3531 +Test: [ 70/269] eta: 0:00:52 acc1_pos: 28.448 acc5_pos: 100.000 loss: 0.5358 (0.5976) acc1: 71.3542 (67.8771) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.2136 data: 0.1201 max mem: 3531 +Test: [ 80/269] eta: 0:00:49 acc1_pos: 28.448 acc5_pos: 100.000 loss: 0.5021 (0.5864) acc1: 81.5104 (69.5634) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.2450 data: 0.1530 max mem: 3531 +Test: [ 90/269] eta: 0:00:45 acc1_pos: 28.448 acc5_pos: 100.000 loss: 0.5788 (0.6015) acc1: 67.1875 (67.9974) acc5: 100.0000 (100.0000) acc1_neg: 66.406 acc5_neg: 100.000 time: 0.2196 data: 0.1272 max mem: 3531 +Test: [100/269] eta: 0:00:41 acc1_pos: 28.448 acc5_pos: 100.000 loss: 0.6765 (0.6077) acc1: 56.7708 (67.2184) acc5: 100.0000 (100.0000) acc1_neg: 54.167 acc5_neg: 100.000 time: 0.1846 data: 0.0913 max mem: 3531 +Test: [110/269] eta: 0:00:38 acc1_pos: 28.448 acc5_pos: 100.000 loss: 0.6413 (0.6101) acc1: 62.5000 (67.0913) acc5: 100.0000 (100.0000) acc1_neg: 65.625 acc5_neg: 100.000 time: 0.1812 data: 0.0883 max mem: 3531 +Test: [120/269] eta: 0:00:35 acc1_pos: 28.448 acc5_pos: 100.000 loss: 0.6316 (0.6145) acc1: 65.6250 (66.7398) acc5: 100.0000 (100.0000) acc1_neg: 54.948 acc5_neg: 100.000 time: 0.2044 data: 0.1136 max mem: 3531 +Test: [130/269] eta: 0:00:32 acc1_pos: 28.448 acc5_pos: 100.000 loss: 0.6814 (0.6205) acc1: 59.6354 (66.2850) acc5: 100.0000 (100.0000) acc1_neg: 53.906 acc5_neg: 100.000 time: 0.2150 data: 0.1244 max mem: 3531 +Test: [140/269] eta: 0:00:30 acc1_pos: 28.448 acc5_pos: 100.000 loss: 0.6745 (0.6247) acc1: 58.8542 (65.6601) acc5: 100.0000 (100.0000) acc1_neg: 59.115 acc5_neg: 100.000 time: 0.2066 data: 0.1156 max mem: 3531 +Test: [150/269] eta: 0:00:27 acc1_pos: 28.448 acc5_pos: 100.000 loss: 0.6469 (0.6219) acc1: 62.5000 (65.8147) acc5: 100.0000 (100.0000) acc1_neg: 73.958 acc5_neg: 100.000 time: 0.2151 data: 0.1205 max mem: 3531 +Test: [160/269] eta: 0:00:25 acc1_pos: 28.448 acc5_pos: 100.000 loss: 0.6061 (0.6186) acc1: 69.0104 (66.1345) acc5: 100.0000 (100.0000) acc1_neg: 69.271 acc5_neg: 100.000 time: 0.2200 data: 0.1251 max mem: 3531 +Test: [170/269] eta: 0:00:22 acc1_pos: 28.448 acc5_pos: 100.000 loss: 0.6096 (0.6209) acc1: 69.0104 (65.9996) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.2030 data: 0.1113 max mem: 3531 +Test: [180/269] eta: 0:00:20 acc1_pos: 28.448 acc5_pos: 100.000 loss: 0.6177 (0.6190) acc1: 65.8854 (66.3070) acc5: 100.0000 (100.0000) acc1_neg: 58.594 acc5_neg: 100.000 time: 0.2038 data: 0.1123 max mem: 3531 +Test: [190/269] eta: 0:00:18 acc1_pos: 28.448 acc5_pos: 100.000 loss: 0.6015 (0.6211) acc1: 65.6250 (66.0668) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.2229 data: 0.1312 max mem: 3531 +Test: [200/269] eta: 0:00:15 acc1_pos: 28.448 acc5_pos: 100.000 loss: 0.7055 (0.6249) acc1: 59.1146 (65.7494) acc5: 100.0000 (100.0000) acc1_neg: 63.021 acc5_neg: 100.000 time: 0.2044 data: 0.1132 max mem: 3531 +Test: [210/269] eta: 0:00:13 acc1_pos: 28.448 acc5_pos: 100.000 loss: 0.6664 (0.6227) acc1: 60.6771 (65.9755) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1833 data: 0.0916 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 28.448 acc5_pos: 100.000 loss: 0.5172 (0.6190) acc1: 75.5208 (66.3356) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.1992 data: 0.1050 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 28.448 acc5_pos: 100.000 loss: 0.5329 (0.6156) acc1: 75.7812 (66.6746) acc5: 100.0000 (100.0000) acc1_neg: 63.802 acc5_neg: 100.000 time: 0.2059 data: 0.1105 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 28.448 acc5_pos: 100.000 loss: 0.5540 (0.6174) acc1: 70.0521 (66.6191) acc5: 100.0000 (100.0000) acc1_neg: 57.812 acc5_neg: 100.000 time: 0.1960 data: 0.1030 max mem: 3532 +Test: [250/269] eta: 0:00:04 acc1_pos: 28.448 acc5_pos: 100.000 loss: 0.6366 (0.6184) acc1: 67.1875 (66.5162) acc5: 100.0000 (100.0000) acc1_neg: 51.823 acc5_neg: 100.000 time: 0.1728 data: 0.0823 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 28.448 acc5_pos: 100.000 loss: 0.6817 (0.6218) acc1: 58.0729 (66.0740) acc5: 100.0000 (100.0000) acc1_neg: 69.531 acc5_neg: 100.000 time: 0.1403 data: 0.0502 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 28.448 acc5_pos: 100.000 loss: 0.6181 (0.6207) acc1: 65.3646 (66.2717) acc5: 100.0000 (100.0000) acc1_neg: 82.152 acc5_neg: 100.000 time: 0.1342 data: 0.0406 max mem: 3532 +Test: Total time: 0:00:57 (0.2131 s / it) +* Overall Acc@1 66.272 Acc@5 100.000 + target == real Acc@1 58.231 Acc@5 100.000 + target == fake Acc@1 67.369 Acc@5 100.000 + loss 0.621 + PR-AUC 0.9341 ROC-AUC 0.6845 EER 37.21% (thr=0.5234) +Accuracy of the network on the 103293 test images: 66.3% +[rank0]:[W725 17:24:37.521067436 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-25T17:24:43Z] eval_done tag=fixed500k_real_n23_seed1_resampled +[2026-07-25T17:24:44Z] eval_start tag=fixed500k_fake_n45_seed1_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed1_resampled_n116-041-023_20260725_112559/best_checkpoint.pth +started_utc=2026-07-25T17:24:44Z +hostname=n116-041-023 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed1_resampled_n116-041-023_20260725_112559/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed1_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0725 17:24:56.500000 593059 torch/distributed/run.py:766] +W0725 17:24:56.500000 593059 torch/distributed/run.py:766] ***************************************** +W0725 17:24:56.500000 593059 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:24:56.500000 593059 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:25:24.463394537 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:25:24.853636842 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:25:24.004452186 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:25:25.427057059 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:25:25.479438107 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:25:25.544042200 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:25:25.700160960 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 17:25:25.990390325 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed1_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed1_resampled_n116-041-023_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:18:29 acc1_pos: 68.750 acc5_pos: 100.000 loss: 0.5842 (0.5842) acc1: 68.7500 (68.7500) acc5: 100.0000 (100.0000) time: 4.7395 data: 2.8144 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/234] eta: 0:01:54 acc1_pos: 72.917 acc5_pos: 100.000 loss: 0.6124 (0.6675) acc1: 68.7500 (61.8608) acc5: 100.0000 (100.0000) time: 0.5127 data: 0.2562 max mem: 3531 +Test: [ 20/234] eta: 0:01:08 acc1_pos: 96.094 acc5_pos: 100.000 loss: 0.6247 (0.6473) acc1: 65.3646 (62.5992) acc5: 100.0000 (100.0000) time: 0.0985 data: 0.0088 max mem: 3531 +Test: [ 30/234] eta: 0:00:52 acc1_pos: 65.625 acc5_pos: 100.000 loss: 0.2420 (0.5303) acc1: 94.5312 (71.5558) acc5: 100.0000 (100.0000) time: 0.1175 data: 0.0276 max mem: 3531 +Test: [ 40/234] eta: 0:00:51 acc1_pos: 98.958 acc5_pos: 100.000 loss: 0.2311 (0.5188) acc1: 94.7917 (71.7099) acc5: 100.0000 (100.0000) time: 0.2042 data: 0.1132 max mem: 3531 +Test: [ 50/234] eta: 0:00:47 acc1_pos: 99.219 acc5_pos: 100.000 loss: 0.1651 (0.4483) acc1: 97.3958 (76.8587) acc5: 100.0000 (100.0000) time: 0.2582 data: 0.1673 max mem: 3531 +Test: [ 60/234] eta: 0:00:43 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.1220 (0.3895) acc1: 98.6979 (80.6267) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2264 data: 0.1357 max mem: 3531 +Test: [ 70/234] eta: 0:00:41 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.0873 (0.4036) acc1: 100.0000 (79.6655) acc5: 100.0000 (100.0000) acc1_neg: 58.333 acc5_neg: 100.000 time: 0.2347 data: 0.1434 max mem: 3531 +Test: [ 80/234] eta: 0:00:36 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.6447 (0.4154) acc1: 62.5000 (79.0734) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2022 data: 0.1084 max mem: 3531 +Test: [ 90/234] eta: 0:00:32 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.2201 (0.3895) acc1: 92.1875 (80.8751) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1457 data: 0.0519 max mem: 3531 +Test: [100/234] eta: 0:00:29 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.2350 (0.3897) acc1: 92.1875 (80.9638) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1552 data: 0.0636 max mem: 3531 +Test: [110/234] eta: 0:00:26 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.4263 (0.3895) acc1: 79.1667 (81.1655) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.1697 data: 0.0759 max mem: 3531 +Test: [120/234] eta: 0:00:24 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.3559 (0.3948) acc1: 84.8958 (81.0907) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1837 data: 0.0862 max mem: 3531 +Test: [130/234] eta: 0:00:22 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.3916 (0.3962) acc1: 86.7188 (81.4508) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.1795 data: 0.0841 max mem: 3531 +Test: [140/234] eta: 0:00:19 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.4224 (0.4147) acc1: 83.5938 (80.5833) acc5: 100.0000 (100.0000) acc1_neg: 45.052 acc5_neg: 100.000 time: 0.1705 data: 0.0789 max mem: 3531 +Test: [150/234] eta: 0:00:17 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.4259 (0.4199) acc1: 79.9479 (80.3135) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.1949 data: 0.1025 max mem: 3531 +Test: [160/234] eta: 0:00:15 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.4020 (0.4191) acc1: 79.9479 (80.5351) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.1993 data: 0.1054 max mem: 3531 +Test: [170/234] eta: 0:00:13 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.4987 (0.4245) acc1: 76.3021 (80.3134) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.2007 data: 0.1073 max mem: 3531 +Test: [180/234] eta: 0:00:11 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.5197 (0.4336) acc1: 75.5208 (79.7206) acc5: 100.0000 (100.0000) acc1_neg: 39.323 acc5_neg: 100.000 time: 0.1759 data: 0.0825 max mem: 3531 +Test: [190/234] eta: 0:00:08 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.6185 (0.4443) acc1: 66.6667 (79.0017) acc5: 100.0000 (100.0000) acc1_neg: 68.229 acc5_neg: 100.000 time: 0.1691 data: 0.0768 max mem: 3531 +Test: [200/234] eta: 0:00:06 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.6099 (0.4368) acc1: 68.2292 (79.4297) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1890 data: 0.0959 max mem: 3531 +Test: [210/234] eta: 0:00:04 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.1016 (0.4217) acc1: 99.7396 (80.3799) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1914 data: 0.0938 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.1485 (0.4103) acc1: 99.2188 (81.1758) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1798 data: 0.0828 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.2821 (0.4165) acc1: 72.3958 (80.7235) acc5: 100.0000 (100.0000) acc1_neg: 70.052 acc5_neg: 100.000 time: 0.1494 data: 0.0564 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.5446 (0.4181) acc1: 71.8750 (80.6116) acc5: 100.0000 (100.0000) acc1_neg: 72.763 acc5_neg: 100.000 time: 0.1493 data: 0.0563 max mem: 3532 +Test: Total time: 0:00:46 (0.1978 s / it) +* Overall Acc@1 80.612 Acc@5 100.000 + target == real Acc@1 77.003 Acc@5 100.000 + target == fake Acc@1 81.629 Acc@5 100.000 + loss 0.418 + PR-AUC 0.9667 ROC-AUC 0.8883 EER 20.27% (thr=0.5273) +Accuracy of the network on the 89729 test images: 80.6% +[rank0]:[W725 17:26:23.955254014 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 17:26:42.810000 593632 torch/distributed/run.py:766] +W0725 17:26:42.810000 593632 torch/distributed/run.py:766] ***************************************** +W0725 17:26:42.810000 593632 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:26:42.810000 593632 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:27:09.989611849 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:27:09.028727172 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 17:27:09.142181420 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:27:09.199339323 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:27:10.696160415 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:27:10.711534373 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W725 17:27:10.716598979 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:27:10.722839536 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed1_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed1_resampled_n116-041-023_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:31:58 acc1_pos: 36.198 acc5_pos: 100.000 loss: 0.8702 (0.8702) acc1: 36.1979 (36.1979) acc5: 100.0000 (100.0000) time: 6.8365 data: 5.1487 max mem: 3531 +Test: [ 10/2387] eta: 0:27:52 acc1_pos: 14.323 acc5_pos: 100.000 loss: 0.9851 (0.9962) acc1: 29.6875 (28.4091) acc5: 100.0000 (100.0000) time: 0.7035 data: 0.4685 max mem: 3531 +Test: [ 20/2387] eta: 0:16:24 acc1_pos: 17.969 acc5_pos: 100.000 loss: 1.0381 (1.0198) acc1: 19.5312 (26.8105) acc5: 100.0000 (100.0000) time: 0.0951 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:14:16 acc1_pos: 18.750 acc5_pos: 100.000 loss: 1.0301 (1.0186) acc1: 18.7500 (26.9489) acc5: 100.0000 (100.0000) time: 0.1765 data: 0.0791 max mem: 3531 +Test: [ 40/2387] eta: 0:13:01 acc1_pos: 25.781 acc5_pos: 100.000 loss: 0.9761 (1.0163) acc1: 24.2188 (25.7114) acc5: 100.0000 (100.0000) time: 0.2454 data: 0.1521 max mem: 3531 +Test: [ 50/2387] eta: 0:12:33 acc1_pos: 35.156 acc5_pos: 100.000 loss: 0.9524 (1.0015) acc1: 26.8229 (27.4203) acc5: 100.0000 (100.0000) time: 0.2583 data: 0.1662 max mem: 3531 +Test: [ 60/2387] eta: 0:11:54 acc1_pos: 46.615 acc5_pos: 100.000 loss: 0.9395 (0.9985) acc1: 33.3333 (28.1848) acc5: 100.0000 (100.0000) time: 0.2544 data: 0.1598 max mem: 3531 +Test: [ 70/2387] eta: 0:11:36 acc1_pos: 34.635 acc5_pos: 100.000 loss: 1.0091 (1.0067) acc1: 27.6042 (27.7949) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1497 max mem: 3531 +Test: [ 80/2387] eta: 0:11:21 acc1_pos: 66.667 acc5_pos: 100.000 loss: 0.9573 (0.9821) acc1: 34.6354 (29.9929) acc5: 100.0000 (100.0000) time: 0.2599 data: 0.1647 max mem: 3531 +Test: [ 90/2387] eta: 0:11:07 acc1_pos: 55.729 acc5_pos: 100.000 loss: 0.7441 (0.9486) acc1: 50.2604 (32.8497) acc5: 100.0000 (100.0000) time: 0.2550 data: 0.1584 max mem: 3531 +Test: [ 100/2387] eta: 0:10:55 acc1_pos: 52.083 acc5_pos: 100.000 loss: 0.6875 (0.9313) acc1: 52.0833 (34.6019) acc5: 100.0000 (100.0000) time: 0.2504 data: 0.1550 max mem: 3531 +Test: [ 110/2387] eta: 0:10:42 acc1_pos: 57.292 acc5_pos: 100.000 loss: 0.7049 (0.9098) acc1: 55.7292 (36.7985) acc5: 100.0000 (100.0000) time: 0.2448 data: 0.1529 max mem: 3531 +Test: [ 120/2387] eta: 0:10:38 acc1_pos: 39.583 acc5_pos: 100.000 loss: 0.7049 (0.8980) acc1: 55.7292 (37.8573) acc5: 100.0000 (100.0000) time: 0.2561 data: 0.1651 max mem: 3531 +Test: [ 130/2387] eta: 0:10:31 acc1_pos: 52.604 acc5_pos: 100.000 loss: 0.7091 (0.8839) acc1: 52.6042 (39.2653) acc5: 100.0000 (100.0000) time: 0.2649 data: 0.1718 max mem: 3531 +Test: [ 140/2387] eta: 0:10:28 acc1_pos: 53.125 acc5_pos: 100.000 loss: 0.7126 (0.8726) acc1: 52.8646 (40.1060) acc5: 100.0000 (100.0000) time: 0.2683 data: 0.1717 max mem: 3531 +Test: [ 150/2387] eta: 0:10:18 acc1_pos: 41.146 acc5_pos: 100.000 loss: 0.7187 (0.8601) acc1: 52.8646 (41.2338) acc5: 100.0000 (100.0000) time: 0.2575 data: 0.1628 max mem: 3531 +Test: [ 160/2387] eta: 0:10:15 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.6383 (0.8397) acc1: 61.7188 (43.2001) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2549 data: 0.1623 max mem: 3531 +Test: [ 170/2387] eta: 0:10:00 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.2265 (0.8032) acc1: 96.8750 (46.3922) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2274 data: 0.1314 max mem: 3531 +Test: [ 180/2387] eta: 0:09:54 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.2108 (0.7707) acc1: 97.9167 (49.2490) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2099 data: 0.1115 max mem: 3531 +Test: [ 190/2387] eta: 0:09:44 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.2122 (0.7415) acc1: 97.9167 (51.8025) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2249 data: 0.1260 max mem: 3531 +Test: [ 200/2387] eta: 0:09:33 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.2085 (0.7151) acc1: 98.1771 (54.1071) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2000 data: 0.1038 max mem: 3531 +Test: [ 210/2387] eta: 0:09:25 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.2155 (0.6916) acc1: 97.9167 (56.1797) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1999 data: 0.1031 max mem: 3532 +Test: [ 220/2387] eta: 0:09:17 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.2172 (0.6701) acc1: 97.9167 (58.0706) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2049 data: 0.1056 max mem: 3532 +Test: [ 230/2387] eta: 0:09:08 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.2179 (0.6504) acc1: 97.9167 (59.7989) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1949 data: 0.0969 max mem: 3532 +Test: [ 240/2387] eta: 0:09:00 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.2246 (0.6352) acc1: 96.6146 (61.2595) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1899 data: 0.0919 max mem: 3532 +Test: [ 250/2387] eta: 0:08:50 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.2039 (0.6141) acc1: 95.5729 (62.7801) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0842 max mem: 3532 +Test: [ 260/2387] eta: 0:08:44 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.0757 (0.5934) acc1: 100.0000 (64.2062) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0896 max mem: 3532 +Test: [ 270/2387] eta: 0:08:39 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.0732 (0.5742) acc1: 100.0000 (65.5270) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2099 data: 0.1150 max mem: 3532 +Test: [ 280/2387] eta: 0:08:31 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.0760 (0.5565) acc1: 100.0000 (66.7538) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.0998 max mem: 3532 +Test: [ 290/2387] eta: 0:08:24 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.0767 (0.5401) acc1: 100.0000 (67.8963) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1766 data: 0.0816 max mem: 3532 +Test: [ 300/2387] eta: 0:08:22 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.0767 (0.5248) acc1: 100.0000 (68.9628) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2149 data: 0.1230 max mem: 3532 +Test: [ 310/2387] eta: 0:08:18 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.0842 (0.5159) acc1: 100.0000 (69.8419) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2290 data: 0.1368 max mem: 3532 +Test: [ 320/2387] eta: 0:08:13 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.4482 (0.5139) acc1: 92.4479 (70.5307) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.2053 data: 0.1104 max mem: 3532 +Test: [ 330/2387] eta: 0:08:09 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.4507 (0.5120) acc1: 91.6667 (71.1748) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2048 data: 0.1083 max mem: 3532 +Test: [ 340/2387] eta: 0:08:05 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.4473 (0.5101) acc1: 91.6667 (71.7879) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.2095 data: 0.1165 max mem: 3532 +Test: [ 350/2387] eta: 0:08:00 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.4344 (0.5017) acc1: 94.2708 (72.5709) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2014 data: 0.1107 max mem: 3532 +Test: [ 360/2387] eta: 0:07:59 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1812 (0.4927) acc1: 99.7396 (73.3141) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2315 data: 0.1409 max mem: 3532 +Test: [ 370/2387] eta: 0:07:54 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1772 (0.4842) acc1: 99.2188 (74.0075) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2254 data: 0.1294 max mem: 3532 +Test: [ 380/2387] eta: 0:07:52 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1772 (0.4762) acc1: 99.2188 (74.6685) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2133 data: 0.1168 max mem: 3532 +Test: [ 390/2387] eta: 0:07:49 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1765 (0.4685) acc1: 98.9583 (75.2884) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2323 data: 0.1378 max mem: 3532 +Test: [ 400/2387] eta: 0:07:45 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1765 (0.4612) acc1: 98.9583 (75.8832) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2114 data: 0.1167 max mem: 3532 +Test: [ 410/2387] eta: 0:07:42 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1750 (0.4543) acc1: 99.2188 (76.4510) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2099 data: 0.1178 max mem: 3532 +Test: [ 420/2387] eta: 0:07:38 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1719 (0.4476) acc1: 99.2188 (76.9930) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2052 data: 0.1137 max mem: 3532 +Test: [ 430/2387] eta: 0:07:35 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1787 (0.4423) acc1: 98.9583 (77.4676) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2055 data: 0.1141 max mem: 3532 +Test: [ 440/2387] eta: 0:07:33 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1851 (0.4355) acc1: 97.3958 (77.9526) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2342 data: 0.1412 max mem: 3533 +Test: [ 450/2387] eta: 0:07:27 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.0820 (0.4277) acc1: 100.0000 (78.4414) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2008 data: 0.1085 max mem: 3533 +Test: [ 460/2387] eta: 0:07:24 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.0818 (0.4202) acc1: 100.0000 (78.9091) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1842 data: 0.0928 max mem: 3533 +Test: [ 470/2387] eta: 0:07:18 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.0824 (0.4130) acc1: 100.0000 (79.3569) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1804 data: 0.0880 max mem: 3533 +Test: [ 480/2387] eta: 0:07:14 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.0833 (0.4062) acc1: 100.0000 (79.7860) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1595 data: 0.0633 max mem: 3533 +Test: [ 490/2387] eta: 0:07:10 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.0825 (0.3996) acc1: 100.0000 (80.1977) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1733 data: 0.0768 max mem: 3533 +Test: [ 500/2387] eta: 0:07:05 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.0825 (0.3932) acc1: 100.0000 (80.5930) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0805 max mem: 3533 +Test: [ 510/2387] eta: 0:07:01 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.0825 (0.3876) acc1: 100.0000 (80.9728) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1774 data: 0.0835 max mem: 3533 +Test: [ 520/2387] eta: 0:06:57 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1301 (0.3831) acc1: 100.0000 (81.3380) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1756 data: 0.0839 max mem: 3533 +Test: [ 530/2387] eta: 0:06:53 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1288 (0.3783) acc1: 100.0000 (81.6894) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1727 data: 0.0816 max mem: 3533 +Test: [ 540/2387] eta: 0:06:49 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1276 (0.3737) acc1: 100.0000 (82.0279) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1692 data: 0.0776 max mem: 3533 +Test: [ 550/2387] eta: 0:06:45 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1268 (0.3692) acc1: 100.0000 (82.3541) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1652 data: 0.0725 max mem: 3533 +Test: [ 560/2387] eta: 0:06:41 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1269 (0.3649) acc1: 100.0000 (82.6686) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1671 data: 0.0746 max mem: 3533 +Test: [ 570/2387] eta: 0:06:39 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1269 (0.3607) acc1: 100.0000 (82.9721) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1938 data: 0.1006 max mem: 3533 +Test: [ 580/2387] eta: 0:06:36 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1269 (0.3567) acc1: 100.0000 (83.2652) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2046 data: 0.1119 max mem: 3533 +Test: [ 590/2387] eta: 0:06:33 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1274 (0.3528) acc1: 100.0000 (83.5484) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0987 max mem: 3533 +Test: [ 600/2387] eta: 0:06:30 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1322 (0.3497) acc1: 100.0000 (83.8221) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1999 data: 0.1087 max mem: 3533 +Test: [ 610/2387] eta: 0:06:27 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1478 (0.3463) acc1: 100.0000 (84.0856) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1982 data: 0.1066 max mem: 3533 +Test: [ 620/2387] eta: 0:06:24 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1400 (0.3430) acc1: 100.0000 (84.3402) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1898 data: 0.0982 max mem: 3533 +Test: [ 630/2387] eta: 0:06:22 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1400 (0.3398) acc1: 100.0000 (84.5875) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2016 data: 0.1079 max mem: 3533 +Test: [ 640/2387] eta: 0:06:19 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1423 (0.3367) acc1: 100.0000 (84.8247) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2019 data: 0.1064 max mem: 3533 +Test: [ 650/2387] eta: 0:06:17 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1431 (0.3338) acc1: 99.7396 (85.0562) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2119 data: 0.1188 max mem: 3533 +Test: [ 660/2387] eta: 0:06:14 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1413 (0.3308) acc1: 100.0000 (85.2807) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2049 data: 0.1135 max mem: 3533 +Test: [ 670/2387] eta: 0:06:12 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1410 (0.3280) acc1: 100.0000 (85.5001) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2046 data: 0.1130 max mem: 3534 +Test: [ 680/2387] eta: 0:06:09 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1430 (0.3259) acc1: 100.0000 (85.7130) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1954 data: 0.1037 max mem: 3534 +Test: [ 690/2387] eta: 0:06:06 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1990 (0.3261) acc1: 100.0000 (85.7276) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.1877 data: 0.0957 max mem: 3534 +Test: [ 700/2387] eta: 0:06:04 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.3840 (0.3269) acc1: 82.5521 (85.6808) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.2009 data: 0.1090 max mem: 3534 +Test: [ 710/2387] eta: 0:06:01 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.3918 (0.3278) acc1: 82.5521 (85.6298) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1960 data: 0.1028 max mem: 3534 +Test: [ 720/2387] eta: 0:05:58 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.3901 (0.3286) acc1: 82.2917 (85.5857) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.1987 data: 0.1046 max mem: 3534 +Test: [ 730/2387] eta: 0:05:56 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.3916 (0.3296) acc1: 81.5104 (85.5264) acc5: 100.0000 (100.0000) acc1_neg: 80.729 acc5_neg: 100.000 time: 0.2000 data: 0.1079 max mem: 3534 +Test: [ 740/2387] eta: 0:05:54 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.3889 (0.3304) acc1: 81.7708 (85.4817) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.2058 data: 0.1146 max mem: 3534 +Test: [ 750/2387] eta: 0:05:51 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.3822 (0.3311) acc1: 82.2917 (85.4375) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.2058 data: 0.1141 max mem: 3534 +Test: [ 760/2387] eta: 0:05:49 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.3773 (0.3315) acc1: 82.8125 (85.4239) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1937 data: 0.1016 max mem: 3534 +Test: [ 770/2387] eta: 0:05:46 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.3055 (0.3303) acc1: 88.8021 (85.5224) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1828 data: 0.0901 max mem: 3534 +Test: [ 780/2387] eta: 0:05:43 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1566 (0.3278) acc1: 98.9583 (85.7004) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1859 data: 0.0930 max mem: 3534 +Test: [ 790/2387] eta: 0:05:40 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1279 (0.3255) acc1: 99.4792 (85.8496) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1756 data: 0.0837 max mem: 3534 +Test: [ 800/2387] eta: 0:05:37 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1253 (0.3231) acc1: 99.7396 (86.0217) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1639 data: 0.0723 max mem: 3534 +Test: [ 810/2387] eta: 0:05:34 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1384 (0.3210) acc1: 99.4792 (86.1764) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1651 data: 0.0718 max mem: 3534 +Test: [ 820/2387] eta: 0:05:31 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1541 (0.3192) acc1: 98.9583 (86.3127) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1699 data: 0.0765 max mem: 3534 +Test: [ 830/2387] eta: 0:05:28 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.2178 (0.3190) acc1: 94.5312 (86.3649) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1784 data: 0.0849 max mem: 3534 +Test: [ 840/2387] eta: 0:05:25 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.2337 (0.3177) acc1: 95.5729 (86.5100) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1749 data: 0.0813 max mem: 3534 +Test: [ 850/2387] eta: 0:05:23 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1994 (0.3163) acc1: 99.2188 (86.6542) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1711 data: 0.0785 max mem: 3534 +Test: [ 860/2387] eta: 0:05:20 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1856 (0.3147) acc1: 99.4792 (86.8031) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1849 data: 0.0926 max mem: 3534 +Test: [ 870/2387] eta: 0:05:17 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1927 (0.3133) acc1: 99.7396 (86.9427) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1837 data: 0.0902 max mem: 3534 +Test: [ 880/2387] eta: 0:05:14 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1972 (0.3121) acc1: 98.9583 (87.0782) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1653 data: 0.0728 max mem: 3534 +Test: [ 890/2387] eta: 0:05:12 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.2087 (0.3110) acc1: 99.2188 (87.2071) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1873 data: 0.0967 max mem: 3534 +Test: [ 900/2387] eta: 0:05:09 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.2385 (0.3141) acc1: 94.7917 (87.0040) acc5: 100.0000 (100.0000) acc1_neg: 66.667 acc5_neg: 100.000 time: 0.1812 data: 0.0900 max mem: 3535 +Test: [ 910/2387] eta: 0:05:07 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.6708 (0.3180) acc1: 61.9792 (86.7356) acc5: 100.0000 (100.0000) acc1_neg: 60.938 acc5_neg: 100.000 time: 0.1755 data: 0.0826 max mem: 3535 +Test: [ 920/2387] eta: 0:05:05 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.6808 (0.3219) acc1: 61.9792 (86.4643) acc5: 100.0000 (100.0000) acc1_neg: 58.854 acc5_neg: 100.000 time: 0.2047 data: 0.1081 max mem: 3535 +Test: [ 930/2387] eta: 0:05:03 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.6808 (0.3258) acc1: 61.9792 (86.2010) acc5: 100.0000 (100.0000) acc1_neg: 60.156 acc5_neg: 100.000 time: 0.2118 data: 0.1158 max mem: 3535 +Test: [ 940/2387] eta: 0:05:01 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.6824 (0.3295) acc1: 61.9792 (85.9422) acc5: 100.0000 (100.0000) acc1_neg: 61.198 acc5_neg: 100.000 time: 0.2048 data: 0.1119 max mem: 3535 +Test: [ 950/2387] eta: 0:04:59 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.6727 (0.3331) acc1: 61.1979 (85.6872) acc5: 100.0000 (100.0000) acc1_neg: 63.281 acc5_neg: 100.000 time: 0.2098 data: 0.1170 max mem: 3535 +Test: [ 960/2387] eta: 0:04:57 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.6669 (0.3366) acc1: 61.9792 (85.4543) acc5: 100.0000 (100.0000) acc1_neg: 64.583 acc5_neg: 100.000 time: 0.2284 data: 0.1347 max mem: 3535 +Test: [ 970/2387] eta: 0:04:55 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.6653 (0.3401) acc1: 61.9792 (85.2056) acc5: 100.0000 (100.0000) acc1_neg: 65.104 acc5_neg: 100.000 time: 0.2249 data: 0.1308 max mem: 3535 +Test: [ 980/2387] eta: 0:04:54 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.6532 (0.3404) acc1: 62.7604 (85.2003) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.2249 data: 0.1304 max mem: 3535 +Test: [ 990/2387] eta: 0:04:52 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.2049 (0.3392) acc1: 95.0521 (85.3005) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2214 data: 0.1274 max mem: 3535 +Test: [1000/2387] eta: 0:04:49 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1897 (0.3378) acc1: 98.1771 (85.4151) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1949 data: 0.1020 max mem: 3535 +Test: [1010/2387] eta: 0:04:46 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.2076 (0.3371) acc1: 96.0938 (85.4847) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1756 data: 0.0836 max mem: 3535 +Test: [1020/2387] eta: 0:04:44 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.2439 (0.3363) acc1: 94.7917 (85.5519) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1726 data: 0.0804 max mem: 3535 +Test: [1030/2387] eta: 0:04:41 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1648 (0.3345) acc1: 98.4375 (85.6824) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1733 data: 0.0811 max mem: 3535 +Test: [1040/2387] eta: 0:04:39 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1411 (0.3326) acc1: 99.7396 (85.8172) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1974 data: 0.1060 max mem: 3535 +Test: [1050/2387] eta: 0:04:37 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1355 (0.3307) acc1: 99.7396 (85.9484) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2146 data: 0.1236 max mem: 3535 +Test: [1060/2387] eta: 0:04:35 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1441 (0.3290) acc1: 99.7396 (86.0713) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2015 data: 0.1105 max mem: 3535 +Test: [1070/2387] eta: 0:04:33 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1645 (0.3278) acc1: 98.4375 (86.1707) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.1961 data: 0.1027 max mem: 3535 +Test: [1080/2387] eta: 0:04:30 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.2221 (0.3272) acc1: 94.2708 (86.2350) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1830 data: 0.0862 max mem: 3535 +Test: [1090/2387] eta: 0:04:28 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.2071 (0.3258) acc1: 96.3542 (86.3404) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1593 data: 0.0614 max mem: 3535 +Test: [1100/2387] eta: 0:04:25 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1699 (0.3244) acc1: 98.4375 (86.4477) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1549 data: 0.0595 max mem: 3535 +Test: [1110/2387] eta: 0:04:23 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1665 (0.3232) acc1: 98.4375 (86.5355) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1939 data: 0.0970 max mem: 3535 +Test: [1120/2387] eta: 0:04:21 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1706 (0.3220) acc1: 98.1771 (86.6298) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2099 data: 0.1114 max mem: 3536 +Test: [1130/2387] eta: 0:04:18 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1885 (0.3208) acc1: 97.6562 (86.7264) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1715 data: 0.0768 max mem: 3536 +Test: [1140/2387] eta: 0:04:16 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.2204 (0.3201) acc1: 97.3958 (86.7979) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1799 data: 0.0858 max mem: 3536 +Test: [1150/2387] eta: 0:04:14 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.2246 (0.3192) acc1: 97.3958 (86.8857) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2005 data: 0.1067 max mem: 3536 +Test: [1160/2387] eta: 0:04:12 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1513 (0.3177) acc1: 100.0000 (86.9960) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2009 data: 0.1096 max mem: 3536 +Test: [1170/2387] eta: 0:04:10 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1320 (0.3161) acc1: 100.0000 (87.1062) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2029 data: 0.1115 max mem: 3536 +Test: [1180/2387] eta: 0:04:08 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1359 (0.3147) acc1: 100.0000 (87.2136) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2043 data: 0.1121 max mem: 3536 +Test: [1190/2387] eta: 0:04:06 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1437 (0.3133) acc1: 100.0000 (87.3190) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2001 data: 0.1060 max mem: 3536 +Test: [1200/2387] eta: 0:04:04 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1437 (0.3119) acc1: 100.0000 (87.4204) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1965 data: 0.1016 max mem: 3536 +Test: [1210/2387] eta: 0:04:01 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1408 (0.3105) acc1: 100.0000 (87.5239) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2014 data: 0.1088 max mem: 3536 +Test: [1220/2387] eta: 0:03:59 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1015 (0.3087) acc1: 100.0000 (87.6252) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1959 data: 0.1048 max mem: 3536 +Test: [1230/2387] eta: 0:03:57 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.0816 (0.3069) acc1: 100.0000 (87.7257) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1946 data: 0.1031 max mem: 3536 +Test: [1240/2387] eta: 0:03:55 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.0798 (0.3051) acc1: 100.0000 (87.8246) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1991 data: 0.1068 max mem: 3536 +Test: [1250/2387] eta: 0:03:53 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.0849 (0.3033) acc1: 100.0000 (87.9217) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1931 data: 0.1009 max mem: 3536 +Test: [1260/2387] eta: 0:03:51 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.0859 (0.3016) acc1: 100.0000 (88.0175) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1885 data: 0.0965 max mem: 3536 +Test: [1270/2387] eta: 0:03:49 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.0861 (0.3000) acc1: 100.0000 (88.1116) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2061 data: 0.1147 max mem: 3536 +Test: [1280/2387] eta: 0:03:47 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.0970 (0.2984) acc1: 100.0000 (88.2007) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2099 data: 0.1175 max mem: 3536 +Test: [1290/2387] eta: 0:03:45 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1000 (0.2969) acc1: 99.4792 (88.2815) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2167 data: 0.1239 max mem: 3536 +Test: [1300/2387] eta: 0:03:42 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.0999 (0.2954) acc1: 99.7396 (88.3659) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2001 data: 0.1052 max mem: 3536 +Test: [1310/2387] eta: 0:03:40 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1009 (0.2941) acc1: 99.7396 (88.4471) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1718 data: 0.0739 max mem: 3536 +Test: [1320/2387] eta: 0:03:38 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1243 (0.2928) acc1: 99.4792 (88.5259) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1788 data: 0.0821 max mem: 3536 +Test: [1330/2387] eta: 0:03:36 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1353 (0.2919) acc1: 98.9583 (88.5927) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1862 data: 0.0918 max mem: 3536 +Test: [1340/2387] eta: 0:03:34 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.2016 (0.2919) acc1: 95.8333 (88.5995) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1966 data: 0.1039 max mem: 3536 +Test: [1350/2387] eta: 0:03:32 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.0663 (0.2902) acc1: 100.0000 (88.6839) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1947 data: 0.1027 max mem: 3537 +Test: [1360/2387] eta: 0:03:29 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.0546 (0.2884) acc1: 100.0000 (88.7671) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1888 data: 0.0966 max mem: 3537 +Test: [1370/2387] eta: 0:03:27 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.0530 (0.2867) acc1: 100.0000 (88.8490) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1906 data: 0.0987 max mem: 3537 +Test: [1380/2387] eta: 0:03:25 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.0530 (0.2850) acc1: 100.0000 (88.9297) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1897 data: 0.0983 max mem: 3537 +Test: [1390/2387] eta: 0:03:23 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.0535 (0.2834) acc1: 100.0000 (89.0093) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1862 data: 0.0955 max mem: 3537 +Test: [1400/2387] eta: 0:03:21 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.0578 (0.2818) acc1: 100.0000 (89.0878) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1864 data: 0.0954 max mem: 3537 +Test: [1410/2387] eta: 0:03:19 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.0605 (0.2808) acc1: 100.0000 (89.1509) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1876 data: 0.0948 max mem: 3537 +Test: [1420/2387] eta: 0:03:17 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1603 (0.2800) acc1: 98.9583 (89.2098) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1950 data: 0.1013 max mem: 3537 +Test: [1430/2387] eta: 0:03:15 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1603 (0.2792) acc1: 97.9167 (89.2701) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2051 data: 0.1120 max mem: 3537 +Test: [1440/2387] eta: 0:03:12 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1708 (0.2786) acc1: 98.1771 (89.3291) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.2050 data: 0.1128 max mem: 3537 +Test: [1450/2387] eta: 0:03:10 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1836 (0.2783) acc1: 98.1771 (89.3635) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1965 data: 0.1028 max mem: 3537 +Test: [1460/2387] eta: 0:03:08 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1333 (0.2773) acc1: 98.9583 (89.4316) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1837 data: 0.0888 max mem: 3537 +Test: [1470/2387] eta: 0:03:06 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1048 (0.2761) acc1: 99.7396 (89.5030) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1700 data: 0.0759 max mem: 3537 +Test: [1480/2387] eta: 0:03:04 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.0974 (0.2749) acc1: 100.0000 (89.5723) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1648 data: 0.0719 max mem: 3537 +Test: [1490/2387] eta: 0:03:01 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1077 (0.2738) acc1: 100.0000 (89.6411) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1771 data: 0.0836 max mem: 3537 +Test: [1500/2387] eta: 0:02:59 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1077 (0.2728) acc1: 99.7396 (89.7084) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1761 data: 0.0832 max mem: 3537 +Test: [1510/2387] eta: 0:02:57 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1244 (0.2719) acc1: 100.0000 (89.7748) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1621 data: 0.0695 max mem: 3537 +Test: [1520/2387] eta: 0:02:55 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1645 (0.2713) acc1: 99.7396 (89.8372) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1632 data: 0.0707 max mem: 3537 +Test: [1530/2387] eta: 0:02:53 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1730 (0.2707) acc1: 99.7396 (89.9012) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1727 data: 0.0807 max mem: 3537 +Test: [1540/2387] eta: 0:02:50 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1532 (0.2699) acc1: 100.0000 (89.9647) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1794 data: 0.0829 max mem: 3537 +Test: [1550/2387] eta: 0:02:48 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1546 (0.2692) acc1: 99.7396 (90.0251) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1681 data: 0.0707 max mem: 3537 +Test: [1560/2387] eta: 0:02:46 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1718 (0.2686) acc1: 99.4792 (90.0853) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1652 data: 0.0708 max mem: 3537 +Test: [1570/2387] eta: 0:02:44 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1718 (0.2681) acc1: 99.4792 (90.1436) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1722 data: 0.0791 max mem: 3537 +Test: [1580/2387] eta: 0:02:42 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1739 (0.2675) acc1: 99.2188 (90.1989) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1727 data: 0.0793 max mem: 3538 +Test: [1590/2387] eta: 0:02:40 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1561 (0.2667) acc1: 98.9583 (90.2562) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1743 data: 0.0794 max mem: 3538 +Test: [1600/2387] eta: 0:02:37 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1285 (0.2660) acc1: 99.4792 (90.2995) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1769 data: 0.0829 max mem: 3538 +Test: [1610/2387] eta: 0:02:35 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1287 (0.2653) acc1: 99.7396 (90.3550) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1757 data: 0.0812 max mem: 3538 +Test: [1620/2387] eta: 0:02:33 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1604 (0.2647) acc1: 98.4375 (90.3948) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1763 data: 0.0799 max mem: 3538 +Test: [1630/2387] eta: 0:02:31 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1888 (0.2645) acc1: 96.6146 (90.4249) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1852 data: 0.0887 max mem: 3538 +Test: [1640/2387] eta: 0:02:29 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.2566 (0.2646) acc1: 93.7500 (90.4320) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1860 data: 0.0903 max mem: 3538 +Test: [1650/2387] eta: 0:02:27 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.2730 (0.2646) acc1: 91.4062 (90.4436) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1896 data: 0.0941 max mem: 3538 +Test: [1660/2387] eta: 0:02:25 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.2666 (0.2647) acc1: 91.6667 (90.4467) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1924 data: 0.0969 max mem: 3538 +Test: [1670/2387] eta: 0:02:23 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.2675 (0.2646) acc1: 91.9271 (90.4639) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1921 data: 0.0999 max mem: 3538 +Test: [1680/2387] eta: 0:02:21 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.2516 (0.2646) acc1: 93.7500 (90.4840) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1949 data: 0.1020 max mem: 3538 +Test: [1690/2387] eta: 0:02:19 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.2526 (0.2646) acc1: 94.0104 (90.4932) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1846 data: 0.0917 max mem: 3538 +Test: [1700/2387] eta: 0:02:17 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.2526 (0.2644) acc1: 92.9688 (90.5155) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1772 data: 0.0857 max mem: 3538 +Test: [1710/2387] eta: 0:02:15 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1143 (0.2635) acc1: 99.4792 (90.5702) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1793 data: 0.0848 max mem: 3538 +Test: [1720/2387] eta: 0:02:13 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1037 (0.2625) acc1: 100.0000 (90.6245) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1732 data: 0.0795 max mem: 3538 +Test: [1730/2387] eta: 0:02:10 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1018 (0.2616) acc1: 100.0000 (90.6781) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1607 data: 0.0681 max mem: 3538 +Test: [1740/2387] eta: 0:02:08 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.0992 (0.2607) acc1: 100.0000 (90.7314) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1643 data: 0.0715 max mem: 3538 +Test: [1750/2387] eta: 0:02:06 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.0993 (0.2598) acc1: 100.0000 (90.7838) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1669 data: 0.0743 max mem: 3538 +Test: [1760/2387] eta: 0:02:04 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1090 (0.2591) acc1: 100.0000 (90.8334) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1563 data: 0.0633 max mem: 3538 +Test: [1770/2387] eta: 0:02:02 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1247 (0.2584) acc1: 99.4792 (90.8804) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1574 data: 0.0657 max mem: 3538 +Test: [1780/2387] eta: 0:02:00 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1247 (0.2576) acc1: 99.7396 (90.9312) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1558 data: 0.0632 max mem: 3538 +Test: [1790/2387] eta: 0:01:58 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1152 (0.2568) acc1: 100.0000 (90.9812) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1473 data: 0.0506 max mem: 3538 +Test: [1800/2387] eta: 0:01:55 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1138 (0.2560) acc1: 100.0000 (91.0304) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1476 data: 0.0524 max mem: 3538 +Test: [1810/2387] eta: 0:01:53 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1136 (0.2552) acc1: 100.0000 (91.0785) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1515 data: 0.0600 max mem: 3539 +Test: [1820/2387] eta: 0:01:51 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1157 (0.2544) acc1: 100.0000 (91.1262) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1554 data: 0.0622 max mem: 3539 +Test: [1830/2387] eta: 0:01:49 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1190 (0.2537) acc1: 100.0000 (91.1737) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1588 data: 0.0665 max mem: 3539 +Test: [1840/2387] eta: 0:01:47 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1115 (0.2530) acc1: 100.0000 (91.2217) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1611 data: 0.0707 max mem: 3539 +Test: [1850/2387] eta: 0:01:45 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1109 (0.2522) acc1: 100.0000 (91.2689) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1759 data: 0.0849 max mem: 3539 +Test: [1860/2387] eta: 0:01:43 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1110 (0.2514) acc1: 100.0000 (91.3159) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1706 data: 0.0779 max mem: 3539 +Test: [1870/2387] eta: 0:01:41 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1115 (0.2507) acc1: 100.0000 (91.3623) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1558 data: 0.0637 max mem: 3539 +Test: [1880/2387] eta: 0:01:39 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1098 (0.2499) acc1: 100.0000 (91.4082) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1650 data: 0.0724 max mem: 3539 +Test: [1890/2387] eta: 0:01:37 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1098 (0.2492) acc1: 100.0000 (91.4536) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1738 data: 0.0800 max mem: 3539 +Test: [1900/2387] eta: 0:01:35 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1118 (0.2485) acc1: 100.0000 (91.4986) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1829 data: 0.0903 max mem: 3539 +Test: [1910/2387] eta: 0:01:33 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1511 (0.2480) acc1: 100.0000 (91.5416) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1795 data: 0.0869 max mem: 3539 +Test: [1920/2387] eta: 0:01:31 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1573 (0.2478) acc1: 99.7396 (91.5621) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1694 data: 0.0766 max mem: 3539 +Test: [1930/2387] eta: 0:01:29 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1455 (0.2471) acc1: 100.0000 (91.6054) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1647 data: 0.0714 max mem: 3539 +Test: [1940/2387] eta: 0:01:27 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1245 (0.2467) acc1: 100.0000 (91.6439) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1714 data: 0.0786 max mem: 3539 +Test: [1950/2387] eta: 0:01:25 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1504 (0.2463) acc1: 98.9583 (91.6748) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1798 data: 0.0873 max mem: 3539 +Test: [1960/2387] eta: 0:01:23 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1732 (0.2461) acc1: 98.9583 (91.7023) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1784 data: 0.0821 max mem: 3539 +Test: [1970/2387] eta: 0:01:21 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.2132 (0.2464) acc1: 93.4896 (91.6746) acc5: 100.0000 (100.0000) acc1_neg: 77.083 acc5_neg: 100.000 time: 0.1796 data: 0.0846 max mem: 3539 +Test: [1980/2387] eta: 0:01:19 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.2757 (0.2463) acc1: 91.4062 (91.6873) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1804 data: 0.0882 max mem: 3539 +Test: [1990/2387] eta: 0:01:17 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1225 (0.2457) acc1: 98.9583 (91.7247) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1753 data: 0.0834 max mem: 3539 +Test: [2000/2387] eta: 0:01:15 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1225 (0.2451) acc1: 99.2188 (91.7628) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1946 data: 0.1007 max mem: 3539 +Test: [2010/2387] eta: 0:01:13 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1233 (0.2445) acc1: 99.4792 (91.8000) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2191 data: 0.1221 max mem: 3539 +Test: [2020/2387] eta: 0:01:11 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1220 (0.2439) acc1: 99.4792 (91.8377) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2146 data: 0.1192 max mem: 3539 +Test: [2030/2387] eta: 0:01:09 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1220 (0.2433) acc1: 99.4792 (91.8744) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2399 data: 0.1461 max mem: 3540 +Test: [2040/2387] eta: 0:01:07 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1227 (0.2427) acc1: 99.2188 (91.9109) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2206 data: 0.1269 max mem: 3540 +Test: [2050/2387] eta: 0:01:05 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1227 (0.2421) acc1: 99.2188 (91.9475) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1919 data: 0.0984 max mem: 3540 +Test: [2060/2387] eta: 0:01:04 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1258 (0.2415) acc1: 99.2188 (91.9829) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2210 data: 0.1287 max mem: 3540 +Test: [2070/2387] eta: 0:01:02 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1250 (0.2410) acc1: 99.4792 (92.0186) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2402 data: 0.1482 max mem: 3540 +Test: [2080/2387] eta: 0:01:00 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1242 (0.2404) acc1: 99.4792 (92.0537) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2231 data: 0.1313 max mem: 3540 +Test: [2090/2387] eta: 0:00:58 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1241 (0.2398) acc1: 99.2188 (92.0890) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2151 data: 0.1234 max mem: 3540 +Test: [2100/2387] eta: 0:00:56 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1229 (0.2393) acc1: 99.2188 (92.1237) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2149 data: 0.1221 max mem: 3540 +Test: [2110/2387] eta: 0:00:54 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1230 (0.2387) acc1: 99.4792 (92.1580) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2061 data: 0.1142 max mem: 3540 +Test: [2120/2387] eta: 0:00:52 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1315 (0.2384) acc1: 99.2188 (92.1868) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2059 data: 0.1146 max mem: 3540 +Test: [2130/2387] eta: 0:00:50 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1389 (0.2380) acc1: 99.2188 (92.2179) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1973 data: 0.1050 max mem: 3540 +Test: [2140/2387] eta: 0:00:48 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1370 (0.2377) acc1: 99.7396 (92.2400) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2012 data: 0.1093 max mem: 3540 +Test: [2150/2387] eta: 0:00:46 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1565 (0.2373) acc1: 98.6979 (92.2704) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2042 data: 0.1126 max mem: 3540 +Test: [2160/2387] eta: 0:00:44 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1613 (0.2370) acc1: 98.6979 (92.2982) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2198 data: 0.1276 max mem: 3540 +Test: [2170/2387] eta: 0:00:42 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1940 (0.2371) acc1: 97.3958 (92.3042) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.2244 data: 0.1328 max mem: 3540 +Test: [2180/2387] eta: 0:00:40 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.2918 (0.2382) acc1: 90.1042 (92.2363) acc5: 100.0000 (100.0000) acc1_neg: 57.812 acc5_neg: 100.000 time: 0.2087 data: 0.1181 max mem: 3540 +Test: [2190/2387] eta: 0:00:38 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.2575 (0.2383) acc1: 93.7500 (92.2342) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2158 data: 0.1250 max mem: 3540 +Test: [2200/2387] eta: 0:00:36 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1395 (0.2378) acc1: 99.2188 (92.2652) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2092 data: 0.1181 max mem: 3540 +Test: [2210/2387] eta: 0:00:34 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1173 (0.2373) acc1: 99.7396 (92.2995) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1710 data: 0.0802 max mem: 3540 +Test: [2220/2387] eta: 0:00:32 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1164 (0.2368) acc1: 100.0000 (92.3320) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1577 data: 0.0664 max mem: 3540 +Test: [2230/2387] eta: 0:00:30 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1209 (0.2362) acc1: 99.7396 (92.3652) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1593 data: 0.0679 max mem: 3540 +Test: [2240/2387] eta: 0:00:28 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1220 (0.2358) acc1: 99.7396 (92.3959) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1617 data: 0.0706 max mem: 3540 +Test: [2250/2387] eta: 0:00:26 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1452 (0.2355) acc1: 99.7396 (92.4232) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1595 data: 0.0680 max mem: 3540 +Test: [2260/2387] eta: 0:00:24 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1610 (0.2352) acc1: 99.7396 (92.4554) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1549 data: 0.0608 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1387 (0.2346) acc1: 100.0000 (92.4874) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1554 data: 0.0622 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1124 (0.2343) acc1: 100.0000 (92.5083) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1499 data: 0.0549 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1172 (0.2338) acc1: 100.0000 (92.5409) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1393 data: 0.0443 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1334 (0.2334) acc1: 99.7396 (92.5669) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1366 data: 0.0450 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1573 (0.2333) acc1: 98.9583 (92.5874) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1380 data: 0.0459 max mem: 3541 +Test: [2320/2387] eta: 0:00:13 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.2050 (0.2333) acc1: 96.6146 (92.5939) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1359 data: 0.0448 max mem: 3541 +Test: [2330/2387] eta: 0:00:11 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1995 (0.2331) acc1: 96.6146 (92.6161) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1292 data: 0.0387 max mem: 3541 +Test: [2340/2387] eta: 0:00:09 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1605 (0.2329) acc1: 99.7396 (92.6383) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1217 data: 0.0316 max mem: 3541 +Test: [2350/2387] eta: 0:00:07 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1552 (0.2328) acc1: 98.9583 (92.6467) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1233 data: 0.0327 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.1913 (0.2327) acc1: 97.9167 (92.6587) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1238 data: 0.0336 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.2003 (0.2328) acc1: 97.1354 (92.6603) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.1193 data: 0.0296 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.3007 (0.2332) acc1: 90.8854 (92.6430) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1166 data: 0.0264 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 61.644 acc5_pos: 100.000 loss: 0.3169 (0.2332) acc1: 88.8021 (92.6430) acc5: 100.0000 (100.0000) acc1_neg: 95.349 acc5_neg: 100.000 time: 0.1084 data: 0.0210 max mem: 3541 +Test: Total time: 0:07:39 (0.1924 s / it) +* Overall Acc@1 92.643 Acc@5 100.000 + target == real Acc@1 41.829 Acc@5 100.000 + target == fake Acc@1 96.211 Acc@5 100.000 + loss 0.233 + PR-AUC 0.9940 ROC-AUC 0.9215 EER 14.85% (thr=0.7339) +Accuracy of the network on the 916267 test images: 92.6% +[rank0]:[W725 17:35:08.285213869 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 17:35:27.893000 594326 torch/distributed/run.py:766] +W0725 17:35:27.893000 594326 torch/distributed/run.py:766] ***************************************** +W0725 17:35:27.893000 594326 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:35:27.893000 594326 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:35:54.136925075 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:35:55.531301181 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:35:55.834610476 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W725 17:35:55.908512038 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W725 17:35:55.910167230 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:35:55.955853622 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:35:55.102844062 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:35:55.108271987 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed1_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed1_resampled_n116-041-023_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:08:06 acc1_pos: 36.198 acc5_pos: 100.000 loss: 0.8702 (0.8702) acc1: 36.1979 (36.1979) acc5: 100.0000 (100.0000) time: 7.9038 data: 6.3016 max mem: 3531 +Test: [ 10/517] eta: 0:06:48 acc1_pos: 14.323 acc5_pos: 100.000 loss: 0.9851 (0.9962) acc1: 29.6875 (28.4091) acc5: 100.0000 (100.0000) time: 0.8056 data: 0.5732 max mem: 3531 +Test: [ 20/517] eta: 0:03:53 acc1_pos: 17.969 acc5_pos: 100.000 loss: 1.0381 (1.0198) acc1: 19.5312 (26.8105) acc5: 100.0000 (100.0000) time: 0.0978 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:03:11 acc1_pos: 18.750 acc5_pos: 100.000 loss: 1.0301 (1.0186) acc1: 18.7500 (26.9489) acc5: 100.0000 (100.0000) time: 0.1649 data: 0.0668 max mem: 3531 +Test: [ 40/517] eta: 0:02:51 acc1_pos: 25.781 acc5_pos: 100.000 loss: 0.9761 (1.0163) acc1: 24.2188 (25.7114) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1501 max mem: 3531 +Test: [ 50/517] eta: 0:02:38 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.9319 (0.9757) acc1: 26.8229 (30.0347) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.2561 data: 0.1637 max mem: 3531 +Test: [ 60/517] eta: 0:02:28 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.4594 (0.8848) acc1: 82.0312 (39.7925) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.2499 data: 0.1570 max mem: 3531 +Test: [ 70/517] eta: 0:02:20 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3884 (0.8117) acc1: 94.2708 (47.6452) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2490 data: 0.1552 max mem: 3531 +Test: [ 80/517] eta: 0:02:12 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3848 (0.7590) acc1: 95.5729 (53.5301) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2399 data: 0.1470 max mem: 3531 +Test: [ 90/517] eta: 0:02:07 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3835 (0.7159) acc1: 94.7917 (58.1445) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.2408 data: 0.1472 max mem: 3531 +Test: [100/517] eta: 0:02:02 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3352 (0.6754) acc1: 96.8750 (62.0514) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2550 data: 0.1620 max mem: 3531 +Test: [110/517] eta: 0:01:58 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.2586 (0.6365) acc1: 99.4792 (65.4326) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2587 data: 0.1663 max mem: 3531 +Test: [120/517] eta: 0:01:53 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.2839 (0.6148) acc1: 98.1771 (67.7040) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.2500 data: 0.1568 max mem: 3531 +Test: [130/517] eta: 0:01:50 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3889 (0.5968) acc1: 91.6667 (69.5372) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2599 data: 0.1673 max mem: 3531 +Test: [140/517] eta: 0:01:47 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3048 (0.5743) acc1: 98.6979 (71.6718) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2697 data: 0.1753 max mem: 3531 +Test: [150/517] eta: 0:01:43 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.2731 (0.5538) acc1: 100.0000 (73.5220) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2499 data: 0.1550 max mem: 3531 +Test: [160/517] eta: 0:01:39 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.2834 (0.5381) acc1: 98.9583 (75.0663) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2449 data: 0.1510 max mem: 3531 +Test: [170/517] eta: 0:01:35 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3010 (0.5234) acc1: 98.1771 (76.4209) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2410 data: 0.1456 max mem: 3531 +Test: [180/517] eta: 0:01:32 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3009 (0.5106) acc1: 97.9167 (77.4977) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2312 data: 0.1374 max mem: 3531 +Test: [190/517] eta: 0:01:28 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.2183 (0.4948) acc1: 99.2188 (78.6554) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2337 data: 0.1403 max mem: 3531 +Test: [200/517] eta: 0:01:25 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.2133 (0.4815) acc1: 100.0000 (79.7043) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2386 data: 0.1453 max mem: 3531 +Test: [210/517] eta: 0:01:22 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.2552 (0.4754) acc1: 98.6979 (80.4490) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2449 data: 0.1537 max mem: 3532 +Test: [220/517] eta: 0:01:19 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3611 (0.4707) acc1: 94.2708 (81.0355) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2499 data: 0.1585 max mem: 3532 +Test: [230/517] eta: 0:01:16 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3718 (0.4661) acc1: 94.5312 (81.6558) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2408 data: 0.1482 max mem: 3532 +Test: [240/517] eta: 0:01:13 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3133 (0.4582) acc1: 98.4375 (82.3965) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2419 data: 0.1483 max mem: 3532 +Test: [250/517] eta: 0:01:10 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.2815 (0.4522) acc1: 99.4792 (83.0594) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2458 data: 0.1510 max mem: 3532 +Test: [260/517] eta: 0:01:07 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3073 (0.4476) acc1: 99.2188 (83.6297) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.2429 data: 0.1481 max mem: 3532 +Test: [270/517] eta: 0:01:05 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3855 (0.4462) acc1: 96.3542 (83.9829) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2492 data: 0.1552 max mem: 3532 +Test: [280/517] eta: 0:01:02 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3886 (0.4434) acc1: 93.4896 (84.3157) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2449 data: 0.1493 max mem: 3532 +Test: [290/517] eta: 0:00:59 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3274 (0.4393) acc1: 96.0938 (84.7750) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2344 data: 0.1401 max mem: 3532 +Test: [300/517] eta: 0:00:56 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3049 (0.4346) acc1: 98.6979 (85.2514) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2399 data: 0.1443 max mem: 3532 +Test: [310/517] eta: 0:00:54 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3332 (0.4327) acc1: 98.4375 (85.6093) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2650 data: 0.1678 max mem: 3532 +Test: [320/517] eta: 0:00:51 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3523 (0.4294) acc1: 97.9167 (85.9878) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2649 data: 0.1712 max mem: 3532 +Test: [330/517] eta: 0:00:48 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3052 (0.4249) acc1: 98.4375 (86.3757) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2514 data: 0.1597 max mem: 3532 +Test: [340/517] eta: 0:00:46 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.2732 (0.4218) acc1: 99.2188 (86.5744) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.2499 data: 0.1572 max mem: 3532 +Test: [350/517] eta: 0:00:43 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3255 (0.4205) acc1: 94.5312 (86.7388) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2425 data: 0.1461 max mem: 3532 +Test: [360/517] eta: 0:00:40 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3530 (0.4189) acc1: 93.4896 (86.9236) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2171 data: 0.1188 max mem: 3532 +Test: [370/517] eta: 0:00:37 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3297 (0.4153) acc1: 95.5729 (87.2480) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2149 data: 0.1204 max mem: 3532 +Test: [380/517] eta: 0:00:35 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.2920 (0.4128) acc1: 97.9167 (87.4638) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2428 data: 0.1446 max mem: 3532 +Test: [390/517] eta: 0:00:32 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3163 (0.4111) acc1: 96.6146 (87.6259) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2499 data: 0.1475 max mem: 3532 +Test: [400/517] eta: 0:00:30 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3123 (0.4082) acc1: 97.1354 (87.8624) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2402 data: 0.1403 max mem: 3532 +Test: [410/517] eta: 0:00:27 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3150 (0.4066) acc1: 97.9167 (88.0943) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2319 data: 0.1353 max mem: 3532 +Test: [420/517] eta: 0:00:24 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3794 (0.4060) acc1: 96.8750 (88.2788) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2408 data: 0.1468 max mem: 3532 +Test: [430/517] eta: 0:00:22 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3562 (0.4044) acc1: 96.8750 (88.4577) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.2479 data: 0.1518 max mem: 3532 +Test: [440/517] eta: 0:00:19 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3562 (0.4040) acc1: 95.0521 (88.6173) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2287 data: 0.1322 max mem: 3533 +Test: [450/517] eta: 0:00:17 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3793 (0.4032) acc1: 96.0938 (88.7871) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2249 data: 0.1311 max mem: 3533 +Test: [460/517] eta: 0:00:14 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3706 (0.4027) acc1: 96.0938 (88.9100) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.2499 data: 0.1552 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3494 (0.4012) acc1: 96.8750 (89.1045) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2749 data: 0.1768 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3237 (0.3995) acc1: 98.9583 (89.2915) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2648 data: 0.1675 max mem: 3533 +Test: [490/517] eta: 0:00:06 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3237 (0.3980) acc1: 97.3958 (89.4343) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2398 data: 0.1440 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.4056 (0.3997) acc1: 92.1875 (89.2735) acc5: 100.0000 (100.0000) acc1_neg: 75.521 acc5_neg: 100.000 time: 0.2449 data: 0.1493 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.4339 (0.3993) acc1: 89.5833 (89.3122) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2499 data: 0.1541 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3202 (0.3980) acc1: 94.2708 (89.4258) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2286 data: 0.1313 max mem: 3533 +Test: Total time: 0:02:12 (0.2554 s / it) +* Overall Acc@1 89.426 Acc@5 100.000 + target == real Acc@1 26.965 Acc@5 100.000 + target == fake Acc@1 95.905 Acc@5 100.000 + loss 0.398 + PR-AUC 0.9548 ROC-AUC 0.7333 EER 32.91% (thr=0.6870) +Accuracy of the network on the 198427 test images: 89.4% +[rank0]:[W725 17:38:23.543558064 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0725 17:38:43.184000 594933 torch/distributed/run.py:766] +W0725 17:38:43.184000 594933 torch/distributed/run.py:766] ***************************************** +W0725 17:38:43.184000 594933 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:38:43.184000 594933 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:39:12.610172112 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:39:12.202242729 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:39:13.580934524 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:39:13.686542324 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 17:39:13.809412675 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:39:13.830420462 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:39:13.837604071 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +NCCL version 2.26.2+cuda12.2 +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:39:13.923640877 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed1_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed1_resampled_n116-041-023_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:35:43 acc1_pos: 19.792 acc5_pos: 100.000 loss: 1.0480 (1.0480) acc1: 19.7917 (19.7917) acc5: 100.0000 (100.0000) time: 5.5109 data: 3.4447 max mem: 3531 +Test: [ 10/389] eta: 0:03:42 acc1_pos: 79.427 acc5_pos: 100.000 loss: 0.8073 (0.7890) acc1: 38.0208 (47.6326) acc5: 100.0000 (100.0000) time: 0.5873 data: 0.3180 max mem: 3531 +Test: [ 20/389] eta: 0:02:21 acc1_pos: 2.604 acc5_pos: 100.000 loss: 0.8073 (0.8527) acc1: 35.9375 (40.5382) acc5: 100.0000 (100.0000) time: 0.1260 data: 0.0347 max mem: 3531 +Test: [ 30/389] eta: 0:02:10 acc1_pos: 32.812 acc5_pos: 100.000 loss: 1.0595 (0.9057) acc1: 12.7604 (32.8041) acc5: 100.0000 (100.0000) time: 0.2390 data: 0.1415 max mem: 3531 +Test: [ 40/389] eta: 0:01:56 acc1_pos: 10.677 acc5_pos: 100.000 loss: 1.0204 (0.8972) acc1: 16.9271 (33.1047) acc5: 100.0000 (100.0000) time: 0.2844 data: 0.1878 max mem: 3531 +Test: [ 50/389] eta: 0:01:44 acc1_pos: 45.052 acc5_pos: 100.000 loss: 0.8007 (0.8769) acc1: 39.0625 (34.8346) acc5: 100.0000 (100.0000) time: 0.2244 data: 0.1315 max mem: 3531 +Test: [ 60/389] eta: 0:01:37 acc1_pos: 37.240 acc5_pos: 100.000 loss: 0.8002 (0.8599) acc1: 40.3646 (36.3516) acc5: 100.0000 (100.0000) time: 0.2174 data: 0.1239 max mem: 3531 +Test: [ 70/389] eta: 0:01:29 acc1_pos: 58.333 acc5_pos: 100.000 loss: 0.8002 (0.8524) acc1: 40.1042 (37.3093) acc5: 100.0000 (100.0000) time: 0.2115 data: 0.1191 max mem: 3531 +Test: [ 80/389] eta: 0:01:24 acc1_pos: 87.500 acc5_pos: 100.000 loss: 0.8024 (0.8614) acc1: 34.1146 (37.4646) acc5: 100.0000 (100.0000) time: 0.2007 data: 0.1082 max mem: 3531 +Test: [ 90/389] eta: 0:01:18 acc1_pos: 64.323 acc5_pos: 100.000 loss: 0.4946 (0.8190) acc1: 74.4792 (41.9586) acc5: 100.0000 (100.0000) time: 0.2022 data: 0.1102 max mem: 3531 +Test: [100/389] eta: 0:01:14 acc1_pos: 20.833 acc5_pos: 100.000 loss: 0.7358 (0.8275) acc1: 45.3125 (40.9189) acc5: 100.0000 (100.0000) time: 0.1891 data: 0.0977 max mem: 3531 +Test: [110/389] eta: 0:01:09 acc1_pos: 99.219 acc5_pos: 100.000 loss: 0.8906 (0.8322) acc1: 28.9062 (41.0285) acc5: 100.0000 (100.0000) time: 0.1810 data: 0.0880 max mem: 3531 +Test: [120/389] eta: 0:01:05 acc1_pos: 57.031 acc5_pos: 100.000 loss: 0.7092 (0.8226) acc1: 55.2083 (42.1122) acc5: 100.0000 (100.0000) time: 0.1785 data: 0.0835 max mem: 3531 +Test: [130/389] eta: 0:01:02 acc1_pos: 17.791 acc5_pos: 100.000 loss: 0.6195 (0.8021) acc1: 63.8021 (44.2609) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1999 data: 0.1063 max mem: 3531 +Test: [140/389] eta: 0:00:58 acc1_pos: 17.791 acc5_pos: 100.000 loss: 0.3641 (0.7727) acc1: 84.6354 (47.2906) acc5: 100.0000 (100.0000) acc1_neg: 64.323 acc5_neg: 100.000 time: 0.1950 data: 0.1032 max mem: 3531 +Test: [150/389] eta: 0:00:55 acc1_pos: 17.791 acc5_pos: 100.000 loss: 0.3663 (0.7558) acc1: 89.8438 (49.0894) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1811 data: 0.0900 max mem: 3531 +Test: [160/389] eta: 0:00:53 acc1_pos: 17.791 acc5_pos: 100.000 loss: 0.3885 (0.7278) acc1: 88.0208 (51.8003) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2107 data: 0.1189 max mem: 3531 +Test: [170/389] eta: 0:00:50 acc1_pos: 17.791 acc5_pos: 100.000 loss: 0.2080 (0.7046) acc1: 99.7396 (54.1088) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.2044 data: 0.1109 max mem: 3531 +Test: [180/389] eta: 0:00:47 acc1_pos: 17.791 acc5_pos: 100.000 loss: 0.2879 (0.6874) acc1: 92.9688 (56.0817) acc5: 100.0000 (100.0000) acc1_neg: 71.354 acc5_neg: 100.000 time: 0.2005 data: 0.1076 max mem: 3531 +Test: [190/389] eta: 0:00:45 acc1_pos: 17.791 acc5_pos: 100.000 loss: 0.3974 (0.6743) acc1: 90.8854 (57.6693) acc5: 100.0000 (100.0000) acc1_neg: 60.417 acc5_neg: 100.000 time: 0.2095 data: 0.1180 max mem: 3531 +Test: [200/389] eta: 0:00:42 acc1_pos: 17.791 acc5_pos: 100.000 loss: 0.4261 (0.6649) acc1: 88.8021 (58.7985) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.1945 data: 0.1036 max mem: 3531 +Test: [210/389] eta: 0:00:40 acc1_pos: 17.791 acc5_pos: 100.000 loss: 0.2622 (0.6486) acc1: 95.8333 (60.4068) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1969 data: 0.1045 max mem: 3532 +Test: [220/389] eta: 0:00:37 acc1_pos: 17.791 acc5_pos: 100.000 loss: 0.2184 (0.6294) acc1: 100.0000 (62.1701) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2002 data: 0.1075 max mem: 3532 +Test: [230/389] eta: 0:00:35 acc1_pos: 17.791 acc5_pos: 100.000 loss: 0.2840 (0.6188) acc1: 97.6562 (63.3151) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.2077 data: 0.1155 max mem: 3532 +Test: [240/389] eta: 0:00:32 acc1_pos: 17.791 acc5_pos: 100.000 loss: 0.3678 (0.6085) acc1: 94.0104 (64.5596) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1951 data: 0.1031 max mem: 3532 +Test: [250/389] eta: 0:00:30 acc1_pos: 17.791 acc5_pos: 100.000 loss: 0.3147 (0.5949) acc1: 98.6979 (65.9508) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1788 data: 0.0872 max mem: 3532 +Test: [260/389] eta: 0:00:28 acc1_pos: 17.791 acc5_pos: 100.000 loss: 0.2553 (0.5856) acc1: 99.7396 (66.8293) acc5: 100.0000 (100.0000) acc1_neg: 63.281 acc5_neg: 100.000 time: 0.1849 data: 0.0928 max mem: 3532 +Test: [270/389] eta: 0:00:25 acc1_pos: 17.791 acc5_pos: 100.000 loss: 0.3832 (0.5840) acc1: 91.9271 (67.1049) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1760 data: 0.0832 max mem: 3532 +Test: [280/389] eta: 0:00:23 acc1_pos: 17.791 acc5_pos: 100.000 loss: 0.5004 (0.5821) acc1: 73.1771 (67.3441) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1698 data: 0.0778 max mem: 3532 +Test: [290/389] eta: 0:00:21 acc1_pos: 17.791 acc5_pos: 100.000 loss: 0.4728 (0.5751) acc1: 87.5000 (68.1093) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1756 data: 0.0850 max mem: 3532 +Test: [300/389] eta: 0:00:18 acc1_pos: 17.791 acc5_pos: 100.000 loss: 0.4712 (0.5720) acc1: 88.8021 (68.6687) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.1736 data: 0.0824 max mem: 3532 +Test: [310/389] eta: 0:00:16 acc1_pos: 17.791 acc5_pos: 100.000 loss: 0.5074 (0.5738) acc1: 78.9062 (68.5716) acc5: 100.0000 (100.0000) acc1_neg: 36.979 acc5_neg: 100.000 time: 0.1767 data: 0.0859 max mem: 3532 +Test: [320/389] eta: 0:00:14 acc1_pos: 17.791 acc5_pos: 100.000 loss: 0.4593 (0.5684) acc1: 78.1250 (69.1735) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1765 data: 0.0857 max mem: 3532 +Test: [330/389] eta: 0:00:12 acc1_pos: 17.791 acc5_pos: 100.000 loss: 0.4081 (0.5644) acc1: 95.8333 (69.9191) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.1607 data: 0.0698 max mem: 3532 +Test: [340/389] eta: 0:00:10 acc1_pos: 17.791 acc5_pos: 100.000 loss: 0.3085 (0.5548) acc1: 97.6562 (70.7157) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1388 data: 0.0480 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 17.791 acc5_pos: 100.000 loss: 0.2070 (0.5597) acc1: 99.4792 (70.1945) acc5: 100.0000 (100.0000) acc1_neg: 23.177 acc5_neg: 100.000 time: 0.1326 data: 0.0422 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 17.791 acc5_pos: 100.000 loss: 0.4987 (0.5608) acc1: 79.4271 (70.1228) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.1393 data: 0.0495 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 17.791 acc5_pos: 100.000 loss: 0.4881 (0.5555) acc1: 81.7708 (70.6003) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1361 data: 0.0463 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 17.791 acc5_pos: 100.000 loss: 0.3774 (0.5506) acc1: 90.8854 (71.0958) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1364 data: 0.0466 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 17.791 acc5_pos: 100.000 loss: 0.3618 (0.5466) acc1: 94.2708 (71.5228) acc5: 100.0000 (100.0000) acc1_neg: 96.175 acc5_neg: 100.000 time: 0.1374 data: 0.0420 max mem: 3532 +Test: Total time: 0:01:17 (0.1981 s / it) +* Overall Acc@1 71.523 Acc@5 100.000 + target == real Acc@1 41.720 Acc@5 100.000 + target == fake Acc@1 85.376 Acc@5 100.000 + loss 0.547 + PR-AUC 0.8624 ROC-AUC 0.7483 EER 32.57% (thr=0.6328) +Accuracy of the network on the 149358 test images: 71.5% +[rank0]:[W725 17:40:38.954664831 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 17:40:58.516000 595501 torch/distributed/run.py:766] +W0725 17:40:58.516000 595501 torch/distributed/run.py:766] ***************************************** +W0725 17:40:58.516000 595501 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:40:58.516000 595501 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W725 17:41:29.849615065 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:41:29.203865634 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:41:29.346535280 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:41:30.537434678 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:41:30.568824594 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:41:30.639631272 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:41:30.854172524 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:41:30.873751734 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed1_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed1_resampled_n116-041-023_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:34:40 acc1_pos: 84.375 acc5_pos: 100.000 loss: 0.3988 (0.3988) acc1: 84.3750 (84.3750) acc5: 100.0000 (100.0000) time: 6.2488 data: 4.7515 max mem: 3531 +Test: [ 10/333] eta: 0:03:30 acc1_pos: 88.542 acc5_pos: 100.000 loss: 0.2985 (0.3059) acc1: 88.0208 (88.6837) acc5: 100.0000 (100.0000) time: 0.6522 data: 0.4323 max mem: 3531 +Test: [ 20/333] eta: 0:02:01 acc1_pos: 96.094 acc5_pos: 100.000 loss: 0.2892 (0.3027) acc1: 88.2812 (89.1245) acc5: 100.0000 (100.0000) time: 0.0962 data: 0.0004 max mem: 3531 +Test: [ 30/333] eta: 0:01:38 acc1_pos: 57.812 acc5_pos: 100.000 loss: 0.3703 (0.4381) acc1: 82.2917 (78.6458) acc5: 100.0000 (100.0000) time: 0.1449 data: 0.0481 max mem: 3531 +Test: [ 40/333] eta: 0:01:31 acc1_pos: 61.458 acc5_pos: 100.000 loss: 0.7398 (0.5127) acc1: 53.3854 (71.9449) acc5: 100.0000 (100.0000) time: 0.2349 data: 0.1406 max mem: 3531 +Test: [ 50/333] eta: 0:01:24 acc1_pos: 53.385 acc5_pos: 100.000 loss: 0.7378 (0.5579) acc1: 53.3854 (68.2445) acc5: 100.0000 (100.0000) time: 0.2599 data: 0.1614 max mem: 3531 +Test: [ 60/333] eta: 0:01:18 acc1_pos: 47.917 acc5_pos: 100.000 loss: 0.7276 (0.5896) acc1: 54.1667 (65.6677) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1289 max mem: 3531 +Test: [ 70/333] eta: 0:01:11 acc1_pos: 44.531 acc5_pos: 100.000 loss: 0.7219 (0.6123) acc1: 53.3854 (64.0075) acc5: 100.0000 (100.0000) time: 0.2004 data: 0.1034 max mem: 3531 +Test: [ 80/333] eta: 0:01:06 acc1_pos: 60.938 acc5_pos: 100.000 loss: 0.7319 (0.6278) acc1: 53.3854 (62.5932) acc5: 100.0000 (100.0000) time: 0.1949 data: 0.1004 max mem: 3531 +Test: [ 90/333] eta: 0:01:03 acc1_pos: 66.146 acc5_pos: 100.000 loss: 0.6752 (0.6309) acc1: 57.8125 (62.4313) acc5: 100.0000 (100.0000) time: 0.2244 data: 0.1294 max mem: 3531 +Test: [100/333] eta: 0:00:57 acc1_pos: 67.969 acc5_pos: 100.000 loss: 0.6456 (0.6346) acc1: 63.0208 (62.3453) acc5: 100.0000 (100.0000) time: 0.1749 data: 0.0783 max mem: 3531 +Test: [110/333] eta: 0:00:55 acc1_pos: 71.094 acc5_pos: 100.000 loss: 0.6753 (0.6374) acc1: 59.6354 (62.2630) acc5: 100.0000 (100.0000) time: 0.2054 data: 0.1096 max mem: 3531 +Test: [120/333] eta: 0:00:53 acc1_pos: 66.406 acc5_pos: 100.000 loss: 0.6287 (0.6388) acc1: 59.6354 (62.2525) acc5: 100.0000 (100.0000) time: 0.2649 data: 0.1704 max mem: 3531 +Test: [130/333] eta: 0:00:48 acc1_pos: 76.042 acc5_pos: 100.000 loss: 0.6249 (0.6383) acc1: 66.1458 (62.3787) acc5: 100.0000 (100.0000) time: 0.1824 data: 0.0861 max mem: 3531 +Test: [140/333] eta: 0:00:45 acc1_pos: 79.427 acc5_pos: 100.000 loss: 0.6077 (0.6362) acc1: 66.6667 (62.7198) acc5: 100.0000 (100.0000) time: 0.1698 data: 0.0750 max mem: 3531 +Test: [150/333] eta: 0:00:42 acc1_pos: 65.365 acc5_pos: 100.000 loss: 0.6219 (0.6380) acc1: 64.3229 (62.6259) acc5: 100.0000 (100.0000) time: 0.1774 data: 0.0819 max mem: 3531 +Test: [160/333] eta: 0:00:40 acc1_pos: 48.684 acc5_pos: 100.000 loss: 0.6639 (0.6459) acc1: 58.3333 (61.9808) acc5: 100.0000 (100.0000) acc1_neg: 45.312 acc5_neg: 100.000 time: 0.1849 data: 0.0887 max mem: 3531 +Test: [170/333] eta: 0:00:37 acc1_pos: 48.684 acc5_pos: 100.000 loss: 0.7548 (0.6550) acc1: 53.6458 (61.3015) acc5: 100.0000 (100.0000) acc1_neg: 54.167 acc5_neg: 100.000 time: 0.1994 data: 0.1074 max mem: 3531 +Test: [180/333] eta: 0:00:34 acc1_pos: 48.684 acc5_pos: 100.000 loss: 0.7449 (0.6590) acc1: 53.9062 (60.9763) acc5: 100.0000 (100.0000) acc1_neg: 50.000 acc5_neg: 100.000 time: 0.1918 data: 0.1006 max mem: 3531 +Test: [190/333] eta: 0:00:32 acc1_pos: 48.684 acc5_pos: 100.000 loss: 0.7167 (0.6597) acc1: 55.4688 (61.0098) acc5: 100.0000 (100.0000) acc1_neg: 54.688 acc5_neg: 100.000 time: 0.2198 data: 0.1254 max mem: 3531 +Test: [200/333] eta: 0:00:30 acc1_pos: 48.684 acc5_pos: 100.000 loss: 0.6717 (0.6599) acc1: 61.7188 (61.1383) acc5: 100.0000 (100.0000) acc1_neg: 61.719 acc5_neg: 100.000 time: 0.2157 data: 0.1215 max mem: 3531 +Test: [210/333] eta: 0:00:27 acc1_pos: 48.684 acc5_pos: 100.000 loss: 0.7051 (0.6643) acc1: 58.8542 (60.8869) acc5: 100.0000 (100.0000) acc1_neg: 49.740 acc5_neg: 100.000 time: 0.2099 data: 0.1165 max mem: 3532 +Test: [220/333] eta: 0:00:25 acc1_pos: 48.684 acc5_pos: 100.000 loss: 0.7557 (0.6714) acc1: 51.3021 (60.3200) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.2129 data: 0.1190 max mem: 3532 +Test: [230/333] eta: 0:00:23 acc1_pos: 48.684 acc5_pos: 100.000 loss: 0.5510 (0.6648) acc1: 71.0938 (61.0254) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.2248 data: 0.1289 max mem: 3532 +Test: [240/333] eta: 0:00:20 acc1_pos: 48.684 acc5_pos: 100.000 loss: 0.5499 (0.6626) acc1: 71.6146 (61.2087) acc5: 100.0000 (100.0000) acc1_neg: 66.146 acc5_neg: 100.000 time: 0.2341 data: 0.1352 max mem: 3532 +Test: [250/333] eta: 0:00:18 acc1_pos: 48.684 acc5_pos: 100.000 loss: 0.6337 (0.6635) acc1: 63.8021 (61.1886) acc5: 100.0000 (100.0000) acc1_neg: 50.521 acc5_neg: 100.000 time: 0.1949 data: 0.0963 max mem: 3532 +Test: [260/333] eta: 0:00:16 acc1_pos: 48.684 acc5_pos: 100.000 loss: 0.7330 (0.6664) acc1: 56.2500 (61.0193) acc5: 100.0000 (100.0000) acc1_neg: 62.760 acc5_neg: 100.000 time: 0.1953 data: 0.0983 max mem: 3532 +Test: [270/333] eta: 0:00:14 acc1_pos: 48.684 acc5_pos: 100.000 loss: 0.6677 (0.6650) acc1: 61.4583 (61.2191) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.2252 data: 0.1323 max mem: 3532 +Test: [280/333] eta: 0:00:11 acc1_pos: 48.684 acc5_pos: 100.000 loss: 0.6447 (0.6676) acc1: 63.0208 (60.9718) acc5: 100.0000 (100.0000) acc1_neg: 57.812 acc5_neg: 100.000 time: 0.2146 data: 0.1228 max mem: 3532 +Test: [290/333] eta: 0:00:09 acc1_pos: 48.684 acc5_pos: 100.000 loss: 0.8286 (0.6734) acc1: 46.0938 (60.4328) acc5: 100.0000 (100.0000) acc1_neg: 48.698 acc5_neg: 100.000 time: 0.1761 data: 0.0826 max mem: 3532 +Test: [300/333] eta: 0:00:07 acc1_pos: 48.684 acc5_pos: 100.000 loss: 0.8443 (0.6798) acc1: 44.2708 (59.8863) acc5: 100.0000 (100.0000) acc1_neg: 48.698 acc5_neg: 100.000 time: 0.1556 data: 0.0637 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 48.684 acc5_pos: 100.000 loss: 0.7824 (0.6776) acc1: 49.7396 (60.1094) acc5: 100.0000 (100.0000) acc1_neg: 66.927 acc5_neg: 100.000 time: 0.1277 data: 0.0376 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 48.684 acc5_pos: 100.000 loss: 0.5985 (0.6765) acc1: 66.9271 (60.2512) acc5: 100.0000 (100.0000) acc1_neg: 45.833 acc5_neg: 100.000 time: 0.1091 data: 0.0183 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 48.684 acc5_pos: 100.000 loss: 0.8238 (0.6828) acc1: 46.6146 (59.7715) acc5: 100.0000 (100.0000) acc1_neg: 46.615 acc5_neg: 100.000 time: 0.1267 data: 0.0360 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 48.684 acc5_pos: 100.000 loss: 0.8472 (0.6840) acc1: 46.3542 (59.6912) acc5: 100.0000 (100.0000) acc1_neg: 46.761 acc5_neg: 100.000 time: 0.1329 data: 0.0397 max mem: 3532 +Test: Total time: 0:01:09 (0.2092 s / it) +* Overall Acc@1 59.691 Acc@5 100.000 + target == real Acc@1 62.560 Acc@5 100.000 + target == fake Acc@1 57.138 Acc@5 100.000 + loss 0.684 + PR-AUC 0.6532 ROC-AUC 0.6419 EER 40.11% (thr=0.4834) +Accuracy of the network on the 127843 test images: 59.7% +[rank0]:[W725 17:42:49.809383155 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 17:43:08.610000 596070 torch/distributed/run.py:766] +W0725 17:43:08.610000 596070 torch/distributed/run.py:766] ***************************************** +W0725 17:43:08.610000 596070 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:43:08.610000 596070 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:43:38.082466772 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:43:39.441592909 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 17:43:39.446957111 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:43:39.885628743 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:43:39.967097333 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:43:39.990482713 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:43:39.001836385 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:43:39.121102982 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed1_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed1_resampled_n116-041-023_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:22:34 acc1_pos: 93.750 acc5_pos: 100.000 loss: 0.2355 (0.2355) acc1: 93.7500 (93.7500) acc5: 100.0000 (100.0000) time: 5.7642 data: 4.1397 max mem: 3531 +Test: [ 10/235] eta: 0:02:16 acc1_pos: 94.792 acc5_pos: 100.000 loss: 0.2410 (0.2413) acc1: 94.5312 (94.5076) acc5: 100.0000 (100.0000) time: 0.6082 data: 0.3767 max mem: 3531 +Test: [ 20/235] eta: 0:01:22 acc1_pos: 93.750 acc5_pos: 100.000 loss: 0.2366 (0.2373) acc1: 94.7917 (94.7917) acc5: 100.0000 (100.0000) time: 0.1163 data: 0.0210 max mem: 3531 +Test: [ 30/235] eta: 0:01:06 acc1_pos: 95.573 acc5_pos: 100.000 loss: 0.2315 (0.2353) acc1: 95.0521 (94.8169) acc5: 100.0000 (100.0000) time: 0.1668 data: 0.0704 max mem: 3531 +Test: [ 40/235] eta: 0:01:00 acc1_pos: 94.010 acc5_pos: 100.000 loss: 0.2316 (0.2351) acc1: 94.7917 (94.8107) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1344 max mem: 3531 +Test: [ 50/235] eta: 0:00:53 acc1_pos: 91.406 acc5_pos: 100.000 loss: 0.2383 (0.2367) acc1: 94.5312 (94.6589) acc5: 100.0000 (100.0000) time: 0.2380 data: 0.1435 max mem: 3531 +Test: [ 60/235] eta: 0:00:49 acc1_pos: 94.010 acc5_pos: 100.000 loss: 0.2427 (0.2376) acc1: 94.0104 (94.6081) acc5: 100.0000 (100.0000) time: 0.2349 data: 0.1407 max mem: 3531 +Test: [ 70/235] eta: 0:00:44 acc1_pos: 95.385 acc5_pos: 100.000 loss: 0.2559 (0.3770) acc1: 93.2292 (85.5414) acc5: 100.0000 (100.0000) acc1_neg: 20.312 acc5_neg: 100.000 time: 0.2249 data: 0.1302 max mem: 3531 +Test: [ 80/235] eta: 0:00:40 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.3508 (0.4981) acc1: 22.9167 (77.7713) acc5: 100.0000 (100.0000) acc1_neg: 19.531 acc5_neg: 100.000 time: 0.1771 data: 0.0824 max mem: 3531 +Test: [ 90/235] eta: 0:00:36 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.3655 (0.5931) acc1: 21.8750 (71.5574) acc5: 100.0000 (100.0000) acc1_neg: 20.312 acc5_neg: 100.000 time: 0.1949 data: 0.0977 max mem: 3531 +Test: [100/235] eta: 0:00:33 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.3725 (0.6677) acc1: 20.8333 (66.6692) acc5: 100.0000 (100.0000) acc1_neg: 19.531 acc5_neg: 100.000 time: 0.2176 data: 0.1178 max mem: 3531 +Test: [110/235] eta: 0:00:30 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.3329 (0.7290) acc1: 21.6146 (62.6408) acc5: 100.0000 (100.0000) acc1_neg: 23.698 acc5_neg: 100.000 time: 0.2059 data: 0.1102 max mem: 3531 +Test: [120/235] eta: 0:00:27 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.4147 (0.7902) acc1: 18.7500 (58.7035) acc5: 100.0000 (100.0000) acc1_neg: 16.667 acc5_neg: 100.000 time: 0.1849 data: 0.0932 max mem: 3531 +Test: [130/235] eta: 0:00:24 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.4731 (0.8419) acc1: 14.5833 (55.3753) acc5: 100.0000 (100.0000) acc1_neg: 14.062 acc5_neg: 100.000 time: 0.1817 data: 0.0885 max mem: 3531 +Test: [140/235] eta: 0:00:22 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.4958 (0.8891) acc1: 14.5833 (52.4361) acc5: 100.0000 (100.0000) acc1_neg: 9.375 acc5_neg: 100.000 time: 0.2049 data: 0.1128 max mem: 3531 +Test: [150/235] eta: 0:00:19 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.4933 (0.9279) acc1: 14.5833 (49.9448) acc5: 100.0000 (100.0000) acc1_neg: 16.406 acc5_neg: 100.000 time: 0.1871 data: 0.0917 max mem: 3531 +Test: [160/235] eta: 0:00:16 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.4741 (0.9622) acc1: 15.1042 (47.7420) acc5: 100.0000 (100.0000) acc1_neg: 15.365 acc5_neg: 100.000 time: 0.1614 data: 0.0659 max mem: 3531 +Test: [170/235] eta: 0:00:14 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.4301 (0.9735) acc1: 15.3646 (46.7821) acc5: 100.0000 (100.0000) acc1_neg: 29.427 acc5_neg: 100.000 time: 0.1899 data: 0.0969 max mem: 3531 +Test: [180/235] eta: 0:00:12 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.1193 (0.9813) acc1: 33.5938 (46.0866) acc5: 100.0000 (100.0000) acc1_neg: 36.198 acc5_neg: 100.000 time: 0.2334 data: 0.1392 max mem: 3531 +Test: [190/235] eta: 0:00:10 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.1039 (0.9883) acc1: 33.8542 (45.4516) acc5: 100.0000 (100.0000) acc1_neg: 34.635 acc5_neg: 100.000 time: 0.2149 data: 0.1205 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.1151 (0.9954) acc1: 32.8125 (44.7891) acc5: 100.0000 (100.0000) acc1_neg: 30.469 acc5_neg: 100.000 time: 0.1670 data: 0.0722 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.1191 (1.0007) acc1: 31.2500 (44.2684) acc5: 100.0000 (100.0000) acc1_neg: 33.854 acc5_neg: 100.000 time: 0.1850 data: 0.0890 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.1068 (1.0058) acc1: 34.3750 (43.8360) acc5: 100.0000 (100.0000) acc1_neg: 33.594 acc5_neg: 100.000 time: 0.2182 data: 0.1243 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.0992 (1.0099) acc1: 35.6771 (43.4851) acc5: 100.0000 (100.0000) acc1_neg: 36.979 acc5_neg: 100.000 time: 0.1788 data: 0.0888 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.1002 (1.0126) acc1: 35.1562 (43.3583) acc5: 100.0000 (100.0000) acc1_neg: 34.783 acc5_neg: 100.000 time: 0.1531 data: 0.0647 max mem: 3532 +Test: Total time: 0:00:50 (0.2151 s / it) +* Overall Acc@1 43.358 Acc@5 100.000 + target == real Acc@1 94.588 Acc@5 100.000 + target == fake Acc@1 24.830 Acc@5 100.000 + loss 1.013 + PR-AUC 0.8753 ROC-AUC 0.7246 EER 33.72% (thr=0.1992) +Accuracy of the network on the 89879 test images: 43.4% +[rank0]:[W725 17:44:41.857571543 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 17:45:01.173000 596640 torch/distributed/run.py:766] +W0725 17:45:01.173000 596640 torch/distributed/run.py:766] ***************************************** +W0725 17:45:01.173000 596640 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:45:01.173000 596640 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:45:30.629981939 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank1]:[W725 17:45:30.629982494 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:45:30.688013425 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:45:30.835998569 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 17:45:30.036607748 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +NCCL version 2.26.2+cuda12.2 +[rank4]:[W725 17:45:30.143147702 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:45:30.168944522 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:45:30.210609270 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed1_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed1_resampled_n116-041-023_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:22:52 acc1_pos: 55.469 acc5_pos: 100.000 loss: 0.6903 (0.6903) acc1: 55.4688 (55.4688) acc5: 100.0000 (100.0000) time: 5.1026 data: 3.0660 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/269] eta: 0:02:38 acc1_pos: 66.146 acc5_pos: 100.000 loss: 0.6268 (0.6328) acc1: 64.0625 (63.7311) acc5: 100.0000 (100.0000) time: 0.6132 data: 0.3457 max mem: 3531 +Test: [ 20/269] eta: 0:01:31 acc1_pos: 75.260 acc5_pos: 100.000 loss: 0.6114 (0.6155) acc1: 65.3646 (66.2946) acc5: 100.0000 (100.0000) time: 0.1321 data: 0.0370 max mem: 3531 +Test: [ 30/269] eta: 0:01:07 acc1_pos: 80.990 acc5_pos: 100.000 loss: 0.5616 (0.6007) acc1: 70.8333 (68.0612) acc5: 100.0000 (100.0000) time: 0.0999 data: 0.0004 max mem: 3531 +Test: [ 40/269] eta: 0:00:58 acc1_pos: 42.241 acc5_pos: 100.000 loss: 0.5685 (0.5932) acc1: 73.6979 (69.6837) acc5: 100.0000 (100.0000) acc1_neg: 75.521 acc5_neg: 100.000 time: 0.1405 data: 0.0444 max mem: 3531 +Test: [ 50/269] eta: 0:00:54 acc1_pos: 42.241 acc5_pos: 100.000 loss: 0.5850 (0.6018) acc1: 73.6979 (68.9747) acc5: 100.0000 (100.0000) acc1_neg: 39.323 acc5_neg: 100.000 time: 0.1999 data: 0.1070 max mem: 3531 +Test: [ 60/269] eta: 0:00:50 acc1_pos: 42.241 acc5_pos: 100.000 loss: 0.6533 (0.6182) acc1: 59.8958 (67.0765) acc5: 100.0000 (100.0000) acc1_neg: 51.823 acc5_neg: 100.000 time: 0.2043 data: 0.1079 max mem: 3531 +Test: [ 70/269] eta: 0:00:46 acc1_pos: 42.241 acc5_pos: 100.000 loss: 0.6802 (0.6229) acc1: 53.6458 (66.3439) acc5: 100.0000 (100.0000) acc1_neg: 66.927 acc5_neg: 100.000 time: 0.1952 data: 0.0979 max mem: 3531 +Test: [ 80/269] eta: 0:00:43 acc1_pos: 42.241 acc5_pos: 100.000 loss: 0.6242 (0.6208) acc1: 66.9271 (66.6570) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.1899 data: 0.0949 max mem: 3531 +Test: [ 90/269] eta: 0:00:39 acc1_pos: 42.241 acc5_pos: 100.000 loss: 0.6591 (0.6326) acc1: 57.5521 (64.8438) acc5: 100.0000 (100.0000) acc1_neg: 55.729 acc5_neg: 100.000 time: 0.1810 data: 0.0872 max mem: 3531 +Test: [100/269] eta: 0:00:37 acc1_pos: 42.241 acc5_pos: 100.000 loss: 0.7073 (0.6455) acc1: 48.6979 (63.0569) acc5: 100.0000 (100.0000) acc1_neg: 32.812 acc5_neg: 100.000 time: 0.1981 data: 0.1052 max mem: 3531 +Test: [110/269] eta: 0:00:34 acc1_pos: 42.241 acc5_pos: 100.000 loss: 0.7687 (0.6581) acc1: 41.9271 (61.2026) acc5: 100.0000 (100.0000) acc1_neg: 35.938 acc5_neg: 100.000 time: 0.1884 data: 0.0954 max mem: 3531 +Test: [120/269] eta: 0:00:30 acc1_pos: 42.241 acc5_pos: 100.000 loss: 0.7251 (0.6617) acc1: 47.6562 (60.7696) acc5: 100.0000 (100.0000) acc1_neg: 50.260 acc5_neg: 100.000 time: 0.1328 data: 0.0384 max mem: 3531 +Test: [130/269] eta: 0:00:28 acc1_pos: 42.241 acc5_pos: 100.000 loss: 0.7222 (0.6671) acc1: 50.5208 (59.9018) acc5: 100.0000 (100.0000) acc1_neg: 41.146 acc5_neg: 100.000 time: 0.1513 data: 0.0553 max mem: 3531 +Test: [140/269] eta: 0:00:26 acc1_pos: 42.241 acc5_pos: 100.000 loss: 0.7505 (0.6739) acc1: 45.0521 (58.8689) acc5: 100.0000 (100.0000) acc1_neg: 44.271 acc5_neg: 100.000 time: 0.1738 data: 0.0757 max mem: 3531 +Test: [150/269] eta: 0:00:23 acc1_pos: 42.241 acc5_pos: 100.000 loss: 0.7332 (0.6728) acc1: 46.6146 (58.8352) acc5: 100.0000 (100.0000) acc1_neg: 70.833 acc5_neg: 100.000 time: 0.1478 data: 0.0477 max mem: 3531 +Test: [160/269] eta: 0:00:21 acc1_pos: 42.241 acc5_pos: 100.000 loss: 0.6075 (0.6691) acc1: 66.9271 (59.3685) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.1591 data: 0.0615 max mem: 3531 +Test: [170/269] eta: 0:00:19 acc1_pos: 42.241 acc5_pos: 100.000 loss: 0.6174 (0.6705) acc1: 61.9792 (59.2197) acc5: 100.0000 (100.0000) acc1_neg: 60.677 acc5_neg: 100.000 time: 0.1712 data: 0.0728 max mem: 3531 +Test: [180/269] eta: 0:00:17 acc1_pos: 42.241 acc5_pos: 100.000 loss: 0.6172 (0.6661) acc1: 61.1979 (59.8153) acc5: 100.0000 (100.0000) acc1_neg: 57.552 acc5_neg: 100.000 time: 0.1749 data: 0.0787 max mem: 3531 +Test: [190/269] eta: 0:00:15 acc1_pos: 42.241 acc5_pos: 100.000 loss: 0.6129 (0.6688) acc1: 61.4583 (59.3600) acc5: 100.0000 (100.0000) acc1_neg: 66.406 acc5_neg: 100.000 time: 0.1636 data: 0.0724 max mem: 3531 +Test: [200/269] eta: 0:00:13 acc1_pos: 42.241 acc5_pos: 100.000 loss: 0.6889 (0.6704) acc1: 53.3854 (59.1120) acc5: 100.0000 (100.0000) acc1_neg: 53.385 acc5_neg: 100.000 time: 0.1499 data: 0.0567 max mem: 3531 +Test: [210/269] eta: 0:00:11 acc1_pos: 42.241 acc5_pos: 100.000 loss: 0.6815 (0.6717) acc1: 53.3854 (58.9109) acc5: 100.0000 (100.0000) acc1_neg: 65.365 acc5_neg: 100.000 time: 0.1619 data: 0.0665 max mem: 3532 +Test: [220/269] eta: 0:00:09 acc1_pos: 42.241 acc5_pos: 100.000 loss: 0.6281 (0.6704) acc1: 61.4583 (59.1558) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1649 data: 0.0700 max mem: 3532 +Test: [230/269] eta: 0:00:07 acc1_pos: 42.241 acc5_pos: 100.000 loss: 0.5556 (0.6663) acc1: 72.3958 (59.6241) acc5: 100.0000 (100.0000) acc1_neg: 63.802 acc5_neg: 100.000 time: 0.1530 data: 0.0572 max mem: 3532 +Test: [240/269] eta: 0:00:05 acc1_pos: 42.241 acc5_pos: 100.000 loss: 0.6497 (0.6691) acc1: 60.6771 (59.3577) acc5: 100.0000 (100.0000) acc1_neg: 58.594 acc5_neg: 100.000 time: 0.1599 data: 0.0635 max mem: 3532 +Test: [250/269] eta: 0:00:03 acc1_pos: 42.241 acc5_pos: 100.000 loss: 0.7107 (0.6727) acc1: 52.3438 (58.8241) acc5: 100.0000 (100.0000) acc1_neg: 20.833 acc5_neg: 100.000 time: 0.1749 data: 0.0771 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 42.241 acc5_pos: 100.000 loss: 0.8125 (0.6803) acc1: 32.8125 (57.7626) acc5: 100.0000 (100.0000) acc1_neg: 43.490 acc5_neg: 100.000 time: 0.1708 data: 0.0736 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 42.241 acc5_pos: 100.000 loss: 0.8258 (0.6827) acc1: 34.3750 (57.4579) acc5: 100.0000 (100.0000) acc1_neg: 74.278 acc5_neg: 100.000 time: 0.1523 data: 0.0557 max mem: 3532 +Test: Total time: 0:00:49 (0.1844 s / it) +* Overall Acc@1 57.458 Acc@5 100.000 + target == real Acc@1 67.543 Acc@5 100.000 + target == fake Acc@1 56.082 Acc@5 100.000 + loss 0.683 + PR-AUC 0.9322 ROC-AUC 0.6665 EER 38.36% (thr=0.4768) +Accuracy of the network on the 103293 test images: 57.5% +[rank0]:[W725 17:46:34.749394269 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-25T17:46:40Z] eval_done tag=fixed500k_fake_n45_seed1_resampled +[2026-07-25T17:46:40Z] eval_start tag=fixed500k_real_n23_seed2_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed2_resampled_n116-041-023_20260725_112559/best_checkpoint.pth +started_utc=2026-07-25T17:46:40Z +hostname=n116-041-023 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed2_resampled_n116-041-023_20260725_112559/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed2_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0725 17:46:53.684000 597214 torch/distributed/run.py:766] +W0725 17:46:53.684000 597214 torch/distributed/run.py:766] ***************************************** +W0725 17:46:53.684000 597214 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:46:53.684000 597214 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:47:23.911500451 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +[rank4]:[W725 17:47:24.671047416 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:47:24.675380925 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:47:24.082311768 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 17:47:24.197235805 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:47:24.301437011 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W725 17:47:24.301474598 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:47:25.452711460 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed2_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed2_resampled_n116-041-023_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:24:06 acc1_pos: 23.438 acc5_pos: 100.000 loss: 1.2418 (1.2418) acc1: 23.4375 (23.4375) acc5: 100.0000 (100.0000) time: 6.1825 data: 3.5617 max mem: 3531 +Test: [ 10/234] eta: 0:02:25 acc1_pos: 10.677 acc5_pos: 100.000 loss: 1.1738 (1.1376) acc1: 33.5938 (29.9953) acc5: 100.0000 (100.0000) time: 0.6512 data: 0.3294 max mem: 3531 +Test: [ 20/234] eta: 0:01:28 acc1_pos: 92.448 acc5_pos: 100.000 loss: 1.1738 (1.1312) acc1: 21.3542 (33.8418) acc5: 100.0000 (100.0000) time: 0.1241 data: 0.0321 max mem: 3531 +Test: [ 30/234] eta: 0:01:19 acc1_pos: 51.823 acc5_pos: 100.000 loss: 0.2908 (0.8781) acc1: 90.8854 (50.7224) acc5: 100.0000 (100.0000) time: 0.2446 data: 0.1517 max mem: 3531 +Test: [ 40/234] eta: 0:01:15 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.2710 (0.7983) acc1: 91.9271 (55.0940) acc5: 100.0000 (100.0000) time: 0.3595 data: 0.2670 max mem: 3531 +Test: [ 50/234] eta: 0:01:00 acc1_pos: 99.740 acc5_pos: 100.000 loss: 0.1467 (0.6694) acc1: 97.9167 (63.6080) acc5: 100.0000 (100.0000) time: 0.2418 data: 0.1477 max mem: 3531 +Test: [ 60/234] eta: 0:00:55 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.1129 (0.5732) acc1: 99.4792 (69.5569) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1861 data: 0.0907 max mem: 3531 +Test: [ 70/234] eta: 0:00:50 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.0806 (0.5618) acc1: 100.0000 (70.3272) acc5: 100.0000 (100.0000) acc1_neg: 58.854 acc5_neg: 100.000 time: 0.2402 data: 0.1483 max mem: 3531 +Test: [ 80/234] eta: 0:00:43 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.6721 (0.5514) acc1: 65.1042 (71.0230) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1598 data: 0.0679 max mem: 3531 +Test: [ 90/234] eta: 0:00:39 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.1262 (0.5070) acc1: 96.6146 (73.8811) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1734 data: 0.0805 max mem: 3531 +Test: [100/234] eta: 0:00:36 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.1512 (0.4911) acc1: 94.7917 (74.9716) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2519 data: 0.1591 max mem: 3531 +Test: [110/234] eta: 0:00:32 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.3444 (0.4769) acc1: 83.3333 (76.0112) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.1791 data: 0.0857 max mem: 3531 +Test: [120/234] eta: 0:00:29 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.2649 (0.4724) acc1: 87.7604 (76.5453) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1574 data: 0.0648 max mem: 3531 +Test: [130/234] eta: 0:00:25 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.2602 (0.4595) acc1: 93.2292 (77.6161) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1649 data: 0.0710 max mem: 3531 +Test: [140/234] eta: 0:00:22 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.3508 (0.4753) acc1: 85.4167 (76.8414) acc5: 100.0000 (100.0000) acc1_neg: 34.115 acc5_neg: 100.000 time: 0.1303 data: 0.0354 max mem: 3531 +Test: [150/234] eta: 0:00:19 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4023 (0.4749) acc1: 79.4271 (76.9143) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.1811 data: 0.0873 max mem: 3531 +Test: [160/234] eta: 0:00:17 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.3254 (0.4677) acc1: 80.4688 (77.4262) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.2239 data: 0.1290 max mem: 3531 +Test: [170/234] eta: 0:00:14 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4999 (0.4675) acc1: 76.8229 (77.6011) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2104 data: 0.1157 max mem: 3531 +Test: [180/234] eta: 0:00:12 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.5200 (0.4725) acc1: 75.7812 (77.2776) acc5: 100.0000 (100.0000) acc1_neg: 37.760 acc5_neg: 100.000 time: 0.1802 data: 0.0886 max mem: 3531 +Test: [190/234] eta: 0:00:10 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.7820 (0.4904) acc1: 56.2500 (76.0553) acc5: 100.0000 (100.0000) acc1_neg: 54.167 acc5_neg: 100.000 time: 0.1945 data: 0.1036 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.7820 (0.4842) acc1: 56.2500 (76.4265) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2106 data: 0.1172 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.1143 (0.4679) acc1: 99.7396 (77.4400) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1974 data: 0.1044 max mem: 3532 +Test: [220/234] eta: 0:00:03 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.1953 (0.4570) acc1: 94.2708 (78.1014) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.1664 data: 0.0760 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4476 (0.4783) acc1: 44.0104 (76.3787) acc5: 100.0000 (100.0000) acc1_neg: 38.802 acc5_neg: 100.000 time: 0.1466 data: 0.0567 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.9252 (0.4843) acc1: 39.3229 (75.9353) acc5: 100.0000 (100.0000) acc1_neg: 38.132 acc5_neg: 100.000 time: 0.1554 data: 0.0555 max mem: 3532 +Test: Total time: 0:00:51 (0.2199 s / it) +* Overall Acc@1 75.935 Acc@5 100.000 + target == real Acc@1 63.860 Acc@5 100.000 + target == fake Acc@1 79.339 Acc@5 100.000 + loss 0.484 + PR-AUC 0.9434 ROC-AUC 0.8241 EER 27.86% (thr=0.6436) +Accuracy of the network on the 89729 test images: 75.9% +[rank0]:[W725 17:48:27.808918384 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 17:48:46.773000 597795 torch/distributed/run.py:766] +W0725 17:48:46.773000 597795 torch/distributed/run.py:766] ***************************************** +W0725 17:48:46.773000 597795 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:48:46.773000 597795 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:49:15.352985317 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 17:49:16.221133204 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:49:16.254034678 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:49:16.407285983 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:49:17.430402901 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:49:17.541814165 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:49:17.627623699 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:49:17.656513777 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed2_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed2_resampled_n116-041-023_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 5:05:03 acc1_pos: 20.833 acc5_pos: 100.000 loss: 1.2313 (1.2313) acc1: 20.8333 (20.8333) acc5: 100.0000 (100.0000) time: 7.6678 data: 5.7434 max mem: 3531 +Test: [ 10/2387] eta: 0:31:01 acc1_pos: 11.458 acc5_pos: 100.000 loss: 1.2634 (1.3039) acc1: 22.3958 (20.1231) acc5: 100.0000 (100.0000) time: 0.7833 data: 0.5226 max mem: 3531 +Test: [ 20/2387] eta: 0:18:03 acc1_pos: 14.323 acc5_pos: 100.000 loss: 1.2679 (1.3312) acc1: 17.4479 (19.7297) acc5: 100.0000 (100.0000) time: 0.0974 data: 0.0005 max mem: 3531 +Test: [ 30/2387] eta: 0:14:45 acc1_pos: 5.990 acc5_pos: 100.000 loss: 1.3719 (1.3712) acc1: 12.2396 (17.1455) acc5: 100.0000 (100.0000) time: 0.1516 data: 0.0561 max mem: 3531 +Test: [ 40/2387] eta: 0:13:47 acc1_pos: 7.031 acc5_pos: 100.000 loss: 1.4548 (1.3930) acc1: 10.6771 (15.2820) acc5: 100.0000 (100.0000) time: 0.2416 data: 0.1490 max mem: 3531 +Test: [ 50/2387] eta: 0:12:55 acc1_pos: 27.083 acc5_pos: 100.000 loss: 1.4516 (1.3910) acc1: 11.1979 (15.5076) acc5: 100.0000 (100.0000) time: 0.2643 data: 0.1715 max mem: 3531 +Test: [ 60/2387] eta: 0:12:12 acc1_pos: 26.823 acc5_pos: 100.000 loss: 1.4648 (1.4113) acc1: 11.7188 (15.2194) acc5: 100.0000 (100.0000) time: 0.2382 data: 0.1435 max mem: 3531 +Test: [ 70/2387] eta: 0:11:48 acc1_pos: 15.104 acc5_pos: 100.000 loss: 1.5587 (1.4509) acc1: 10.4167 (14.1249) acc5: 100.0000 (100.0000) time: 0.2388 data: 0.1442 max mem: 3531 +Test: [ 80/2387] eta: 0:11:57 acc1_pos: 11.198 acc5_pos: 100.000 loss: 1.6420 (1.4721) acc1: 7.8125 (13.6060) acc5: 100.0000 (100.0000) time: 0.2999 data: 0.2055 max mem: 3531 +Test: [ 90/2387] eta: 0:11:29 acc1_pos: 14.844 acc5_pos: 100.000 loss: 1.5463 (1.4860) acc1: 9.1146 (13.2183) acc5: 100.0000 (100.0000) time: 0.2798 data: 0.1845 max mem: 3531 +Test: [ 100/2387] eta: 0:11:03 acc1_pos: 9.635 acc5_pos: 100.000 loss: 1.6201 (1.5107) acc1: 9.3750 (12.6882) acc5: 100.0000 (100.0000) time: 0.2049 data: 0.1114 max mem: 3531 +Test: [ 110/2387] eta: 0:10:44 acc1_pos: 14.844 acc5_pos: 100.000 loss: 1.6272 (1.5136) acc1: 9.6354 (12.5962) acc5: 100.0000 (100.0000) time: 0.2049 data: 0.1108 max mem: 3531 +Test: [ 120/2387] eta: 0:10:35 acc1_pos: 13.802 acc5_pos: 100.000 loss: 1.4959 (1.5171) acc1: 11.4583 (12.6184) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1348 max mem: 3531 +Test: [ 130/2387] eta: 0:10:27 acc1_pos: 19.792 acc5_pos: 100.000 loss: 1.5418 (1.5241) acc1: 11.4583 (12.4920) acc5: 100.0000 (100.0000) time: 0.2515 data: 0.1565 max mem: 3531 +Test: [ 140/2387] eta: 0:10:19 acc1_pos: 11.198 acc5_pos: 100.000 loss: 1.5552 (1.5261) acc1: 11.1979 (12.3541) acc5: 100.0000 (100.0000) time: 0.2472 data: 0.1540 max mem: 3531 +Test: [ 150/2387] eta: 0:10:17 acc1_pos: 1.302 acc5_pos: 100.000 loss: 1.5519 (1.5276) acc1: 11.1979 (12.3655) acc5: 100.0000 (100.0000) time: 0.2623 data: 0.1711 max mem: 3531 +Test: [ 160/2387] eta: 0:09:58 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.4886 (1.4974) acc1: 14.0625 (14.2259) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2226 data: 0.1299 max mem: 3531 +Test: [ 170/2387] eta: 0:09:47 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1317 (1.4172) acc1: 98.6979 (19.2099) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1810 data: 0.0832 max mem: 3531 +Test: [ 180/2387] eta: 0:09:31 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1263 (1.3459) acc1: 99.4792 (23.6461) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0784 max mem: 3531 +Test: [ 190/2387] eta: 0:09:18 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1259 (1.2820) acc1: 99.4792 (27.6123) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1613 data: 0.0641 max mem: 3531 +Test: [ 200/2387] eta: 0:09:08 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1243 (1.2244) acc1: 99.4792 (31.1852) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1774 data: 0.0849 max mem: 3531 +Test: [ 210/2387] eta: 0:08:58 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1258 (1.1724) acc1: 99.4792 (34.4207) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1833 data: 0.0896 max mem: 3532 +Test: [ 220/2387] eta: 0:08:51 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1266 (1.1250) acc1: 99.4792 (37.3598) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1894 data: 0.0952 max mem: 3532 +Test: [ 230/2387] eta: 0:08:46 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1255 (1.0818) acc1: 99.2188 (40.0377) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2103 data: 0.1144 max mem: 3532 +Test: [ 240/2387] eta: 0:08:37 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1239 (1.0420) acc1: 99.4792 (42.5160) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1939 data: 0.0989 max mem: 3532 +Test: [ 250/2387] eta: 0:08:33 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1188 (1.0040) acc1: 100.0000 (44.8020) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1945 data: 0.0994 max mem: 3532 +Test: [ 260/2387] eta: 0:08:27 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0791 (0.9685) acc1: 100.0000 (46.9169) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2055 data: 0.1087 max mem: 3532 +Test: [ 270/2387] eta: 0:08:19 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0767 (0.9356) acc1: 100.0000 (48.8757) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0914 max mem: 3532 +Test: [ 280/2387] eta: 0:08:11 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0767 (0.9051) acc1: 100.0000 (50.6951) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1655 data: 0.0708 max mem: 3532 +Test: [ 290/2387] eta: 0:08:11 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0773 (0.8767) acc1: 100.0000 (52.3894) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2108 data: 0.1129 max mem: 3532 +Test: [ 300/2387] eta: 0:08:10 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0756 (0.8501) acc1: 100.0000 (53.9711) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2602 data: 0.1644 max mem: 3532 +Test: [ 310/2387] eta: 0:08:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0746 (0.8267) acc1: 100.0000 (55.4311) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1939 data: 0.1008 max mem: 3532 +Test: [ 320/2387] eta: 0:07:55 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1704 (0.8064) acc1: 99.7396 (56.8000) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1530 data: 0.0578 max mem: 3532 +Test: [ 330/2387] eta: 0:07:54 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1749 (0.7873) acc1: 99.7396 (58.0949) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2111 data: 0.1162 max mem: 3532 +Test: [ 340/2387] eta: 0:07:50 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1732 (0.7692) acc1: 99.7396 (59.3101) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2292 data: 0.1363 max mem: 3532 +Test: [ 350/2387] eta: 0:07:46 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1636 (0.7500) acc1: 99.7396 (60.4679) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1998 data: 0.1065 max mem: 3532 +Test: [ 360/2387] eta: 0:07:42 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1094 (0.7323) acc1: 100.0000 (61.5536) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1963 data: 0.1030 max mem: 3532 +Test: [ 370/2387] eta: 0:07:39 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1143 (0.7157) acc1: 99.4792 (62.5786) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2144 data: 0.1224 max mem: 3532 +Test: [ 380/2387] eta: 0:07:30 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1143 (0.7000) acc1: 99.7396 (63.5519) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1634 data: 0.0688 max mem: 3532 +Test: [ 390/2387] eta: 0:07:23 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1151 (0.6850) acc1: 99.7396 (64.4794) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1147 data: 0.0187 max mem: 3532 +Test: [ 400/2387] eta: 0:07:19 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1151 (0.6708) acc1: 99.7396 (65.3581) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1551 data: 0.0626 max mem: 3532 +Test: [ 410/2387] eta: 0:07:15 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1133 (0.6572) acc1: 99.7396 (66.1959) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1871 data: 0.0957 max mem: 3532 +Test: [ 420/2387] eta: 0:07:12 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1133 (0.6443) acc1: 99.7396 (66.9914) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1961 data: 0.1020 max mem: 3532 +Test: [ 430/2387] eta: 0:07:10 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1148 (0.6321) acc1: 99.7396 (67.7512) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2040 data: 0.1101 max mem: 3532 +Test: [ 440/2387] eta: 0:07:04 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1133 (0.6199) acc1: 100.0000 (68.4795) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1796 data: 0.0882 max mem: 3533 +Test: [ 450/2387] eta: 0:07:02 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0802 (0.6079) acc1: 100.0000 (69.1784) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1830 data: 0.0904 max mem: 3533 +Test: [ 460/2387] eta: 0:06:59 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0792 (0.5965) acc1: 100.0000 (69.8470) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2037 data: 0.1096 max mem: 3533 +Test: [ 470/2387] eta: 0:06:55 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0799 (0.5855) acc1: 100.0000 (70.4872) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1919 data: 0.0961 max mem: 3533 +Test: [ 480/2387] eta: 0:06:53 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0799 (0.5750) acc1: 100.0000 (71.1008) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1999 data: 0.1049 max mem: 3533 +Test: [ 490/2387] eta: 0:06:50 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0799 (0.5649) acc1: 100.0000 (71.6894) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2036 data: 0.1114 max mem: 3533 +Test: [ 500/2387] eta: 0:06:48 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0803 (0.5552) acc1: 100.0000 (72.2545) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2072 data: 0.1140 max mem: 3533 +Test: [ 510/2387] eta: 0:06:45 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0804 (0.5461) acc1: 100.0000 (72.7974) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2046 data: 0.1118 max mem: 3533 +Test: [ 520/2387] eta: 0:06:42 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0949 (0.5376) acc1: 100.0000 (73.3195) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1991 data: 0.1081 max mem: 3533 +Test: [ 530/2387] eta: 0:06:40 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0948 (0.5292) acc1: 100.0000 (73.8220) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2007 data: 0.1085 max mem: 3533 +Test: [ 540/2387] eta: 0:06:37 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0942 (0.5212) acc1: 100.0000 (74.3059) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1989 data: 0.1066 max mem: 3533 +Test: [ 550/2387] eta: 0:06:34 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0944 (0.5134) acc1: 100.0000 (74.7722) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1986 data: 0.1063 max mem: 3533 +Test: [ 560/2387] eta: 0:06:31 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0947 (0.5060) acc1: 100.0000 (75.2219) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1943 data: 0.1001 max mem: 3533 +Test: [ 570/2387] eta: 0:06:28 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0946 (0.4988) acc1: 100.0000 (75.6558) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1861 data: 0.0911 max mem: 3533 +Test: [ 580/2387] eta: 0:06:25 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0937 (0.4918) acc1: 100.0000 (76.0748) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1918 data: 0.0978 max mem: 3533 +Test: [ 590/2387] eta: 0:06:23 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0938 (0.4851) acc1: 100.0000 (76.4797) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2009 data: 0.1088 max mem: 3533 +Test: [ 600/2387] eta: 0:06:20 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0974 (0.4788) acc1: 100.0000 (76.8710) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1954 data: 0.1033 max mem: 3533 +Test: [ 610/2387] eta: 0:06:16 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0954 (0.4724) acc1: 100.0000 (77.2496) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1705 data: 0.0784 max mem: 3533 +Test: [ 620/2387] eta: 0:06:12 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0921 (0.4663) acc1: 100.0000 (77.6151) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1529 data: 0.0612 max mem: 3533 +Test: [ 630/2387] eta: 0:06:09 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0928 (0.4604) acc1: 100.0000 (77.9690) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1673 data: 0.0750 max mem: 3533 +Test: [ 640/2387] eta: 0:06:07 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0933 (0.4547) acc1: 100.0000 (78.3119) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1897 data: 0.0964 max mem: 3533 +Test: [ 650/2387] eta: 0:06:04 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0917 (0.4491) acc1: 100.0000 (78.6450) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1950 data: 0.1025 max mem: 3533 +Test: [ 660/2387] eta: 0:06:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0917 (0.4437) acc1: 100.0000 (78.9681) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1752 data: 0.0833 max mem: 3533 +Test: [ 670/2387] eta: 0:05:58 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0916 (0.4385) acc1: 100.0000 (79.2815) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1646 data: 0.0687 max mem: 3534 +Test: [ 680/2387] eta: 0:05:54 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0897 (0.4333) acc1: 100.0000 (79.5858) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1545 data: 0.0584 max mem: 3534 +Test: [ 690/2387] eta: 0:05:51 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0909 (0.4312) acc1: 100.0000 (79.7022) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.1471 data: 0.0528 max mem: 3534 +Test: [ 700/2387] eta: 0:05:47 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.3611 (0.4302) acc1: 83.0729 (79.7495) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.1510 data: 0.0574 max mem: 3534 +Test: [ 710/2387] eta: 0:05:44 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.3599 (0.4292) acc1: 83.0729 (79.7996) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.1532 data: 0.0604 max mem: 3534 +Test: [ 720/2387] eta: 0:05:41 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.3572 (0.4282) acc1: 83.5938 (79.8529) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.1544 data: 0.0604 max mem: 3534 +Test: [ 730/2387] eta: 0:05:38 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.3629 (0.4274) acc1: 83.3333 (79.8902) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.1560 data: 0.0629 max mem: 3534 +Test: [ 740/2387] eta: 0:05:36 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.3650 (0.4265) acc1: 82.5521 (79.9261) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.1892 data: 0.0963 max mem: 3534 +Test: [ 750/2387] eta: 0:05:34 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.3665 (0.4256) acc1: 83.0729 (79.9739) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.2035 data: 0.1108 max mem: 3534 +Test: [ 760/2387] eta: 0:05:31 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.3511 (0.4239) acc1: 83.5938 (80.0783) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1951 data: 0.1032 max mem: 3534 +Test: [ 770/2387] eta: 0:05:29 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1925 (0.4207) acc1: 94.5312 (80.2931) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2011 data: 0.1089 max mem: 3534 +Test: [ 780/2387] eta: 0:05:27 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1616 (0.4172) acc1: 97.3958 (80.5168) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1999 data: 0.1076 max mem: 3534 +Test: [ 790/2387] eta: 0:05:25 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1381 (0.4141) acc1: 98.9583 (80.7091) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.2039 data: 0.1116 max mem: 3534 +Test: [ 800/2387] eta: 0:05:23 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1208 (0.4104) acc1: 100.0000 (80.9467) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2021 data: 0.1097 max mem: 3534 +Test: [ 810/2387] eta: 0:05:21 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1199 (0.4070) acc1: 100.0000 (81.1784) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1995 data: 0.1067 max mem: 3534 +Test: [ 820/2387] eta: 0:05:20 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1327 (0.4038) acc1: 100.0000 (81.3953) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2288 data: 0.1349 max mem: 3534 +Test: [ 830/2387] eta: 0:05:17 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1396 (0.4009) acc1: 99.4792 (81.5994) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2099 data: 0.1134 max mem: 3534 +Test: [ 840/2387] eta: 0:05:15 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1186 (0.3974) acc1: 99.7396 (81.8148) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1934 data: 0.0973 max mem: 3534 +Test: [ 850/2387] eta: 0:05:13 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0994 (0.3939) acc1: 100.0000 (82.0285) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1999 data: 0.1041 max mem: 3534 +Test: [ 860/2387] eta: 0:05:11 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0958 (0.3904) acc1: 100.0000 (82.2354) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2034 data: 0.1090 max mem: 3534 +Test: [ 870/2387] eta: 0:05:08 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0942 (0.3870) acc1: 100.0000 (82.4379) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1918 data: 0.1007 max mem: 3534 +Test: [ 880/2387] eta: 0:05:05 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0982 (0.3838) acc1: 100.0000 (82.6351) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1524 data: 0.0618 max mem: 3534 +Test: [ 890/2387] eta: 0:05:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0977 (0.3806) acc1: 100.0000 (82.8300) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1690 data: 0.0782 max mem: 3534 +Test: [ 900/2387] eta: 0:05:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1028 (0.3828) acc1: 100.0000 (82.6582) acc5: 100.0000 (100.0000) acc1_neg: 59.635 acc5_neg: 100.000 time: 0.1951 data: 0.1034 max mem: 3535 +Test: [ 910/2387] eta: 0:04:59 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.6890 (0.3862) acc1: 59.6354 (82.4120) acc5: 100.0000 (100.0000) acc1_neg: 58.854 acc5_neg: 100.000 time: 0.1859 data: 0.0917 max mem: 3535 +Test: [ 920/2387] eta: 0:04:57 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.6969 (0.3896) acc1: 58.8542 (82.1624) acc5: 100.0000 (100.0000) acc1_neg: 57.552 acc5_neg: 100.000 time: 0.1860 data: 0.0893 max mem: 3535 +Test: [ 930/2387] eta: 0:04:54 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.6969 (0.3928) acc1: 59.3750 (81.9199) acc5: 100.0000 (100.0000) acc1_neg: 53.906 acc5_neg: 100.000 time: 0.1950 data: 0.0995 max mem: 3535 +Test: [ 940/2387] eta: 0:04:52 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.6955 (0.3960) acc1: 59.1146 (81.6787) acc5: 100.0000 (100.0000) acc1_neg: 59.115 acc5_neg: 100.000 time: 0.1884 data: 0.0955 max mem: 3535 +Test: [ 950/2387] eta: 0:04:50 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.6950 (0.3991) acc1: 59.1146 (81.4458) acc5: 100.0000 (100.0000) acc1_neg: 60.417 acc5_neg: 100.000 time: 0.1845 data: 0.0925 max mem: 3535 +Test: [ 960/2387] eta: 0:04:47 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.6775 (0.4020) acc1: 60.4167 (81.2305) acc5: 100.0000 (100.0000) acc1_neg: 63.542 acc5_neg: 100.000 time: 0.1846 data: 0.0935 max mem: 3535 +Test: [ 970/2387] eta: 0:04:45 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.6874 (0.4051) acc1: 59.3750 (81.0057) acc5: 100.0000 (100.0000) acc1_neg: 59.635 acc5_neg: 100.000 time: 0.1748 data: 0.0836 max mem: 3535 +Test: [ 980/2387] eta: 0:04:43 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.6629 (0.4042) acc1: 61.9792 (81.0541) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1711 data: 0.0797 max mem: 3535 +Test: [ 990/2387] eta: 0:04:40 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1512 (0.4017) acc1: 98.9583 (81.2290) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1855 data: 0.0941 max mem: 3535 +Test: [1000/2387] eta: 0:04:38 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1406 (0.3993) acc1: 99.4792 (81.3954) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1741 data: 0.0830 max mem: 3535 +Test: [1010/2387] eta: 0:04:35 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1489 (0.3968) acc1: 99.4792 (81.5679) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1680 data: 0.0767 max mem: 3535 +Test: [1020/2387] eta: 0:04:34 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1489 (0.3946) acc1: 99.4792 (81.7270) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1931 data: 0.0993 max mem: 3535 +Test: [1030/2387] eta: 0:04:31 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1382 (0.3921) acc1: 99.4792 (81.8969) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1964 data: 0.1030 max mem: 3535 +Test: [1040/2387] eta: 0:04:29 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1237 (0.3895) acc1: 100.0000 (82.0698) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1820 data: 0.0889 max mem: 3535 +Test: [1050/2387] eta: 0:04:27 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1153 (0.3869) acc1: 100.0000 (82.2364) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1943 data: 0.1016 max mem: 3535 +Test: [1060/2387] eta: 0:04:25 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1168 (0.3844) acc1: 100.0000 (82.4011) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1933 data: 0.1026 max mem: 3535 +Test: [1070/2387] eta: 0:04:23 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1285 (0.3821) acc1: 100.0000 (82.5582) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1961 data: 0.1044 max mem: 3535 +Test: [1080/2387] eta: 0:04:21 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1402 (0.3800) acc1: 99.7396 (82.7089) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1996 data: 0.1082 max mem: 3535 +Test: [1090/2387] eta: 0:04:19 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1493 (0.3783) acc1: 99.4792 (82.8337) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1970 data: 0.1026 max mem: 3535 +Test: [1100/2387] eta: 0:04:17 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1571 (0.3764) acc1: 97.3958 (82.9608) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2133 data: 0.1186 max mem: 3535 +Test: [1110/2387] eta: 0:04:15 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1615 (0.3749) acc1: 97.3958 (83.0560) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2110 data: 0.1197 max mem: 3535 +Test: [1120/2387] eta: 0:04:13 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1524 (0.3728) acc1: 98.4375 (83.2005) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2131 data: 0.1207 max mem: 3536 +Test: [1130/2387] eta: 0:04:12 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1474 (0.3711) acc1: 99.4792 (83.3384) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2142 data: 0.1213 max mem: 3536 +Test: [1140/2387] eta: 0:04:10 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1435 (0.3690) acc1: 99.7396 (83.4817) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2210 data: 0.1277 max mem: 3536 +Test: [1150/2387] eta: 0:04:08 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1246 (0.3669) acc1: 100.0000 (83.6166) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2158 data: 0.1225 max mem: 3536 +Test: [1160/2387] eta: 0:04:06 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0758 (0.3643) acc1: 100.0000 (83.7577) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1980 data: 0.1066 max mem: 3536 +Test: [1170/2387] eta: 0:04:04 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0730 (0.3618) acc1: 100.0000 (83.8964) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2036 data: 0.1126 max mem: 3536 +Test: [1180/2387] eta: 0:04:02 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0713 (0.3594) acc1: 100.0000 (84.0328) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2042 data: 0.1119 max mem: 3536 +Test: [1190/2387] eta: 0:04:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0719 (0.3570) acc1: 100.0000 (84.1668) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2106 data: 0.1168 max mem: 3536 +Test: [1200/2387] eta: 0:03:58 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0726 (0.3546) acc1: 100.0000 (84.2987) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2107 data: 0.1182 max mem: 3536 +Test: [1210/2387] eta: 0:03:56 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0733 (0.3523) acc1: 100.0000 (84.4283) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1904 data: 0.0996 max mem: 3536 +Test: [1220/2387] eta: 0:03:54 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0749 (0.3501) acc1: 100.0000 (84.5554) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1898 data: 0.0977 max mem: 3536 +Test: [1230/2387] eta: 0:03:52 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0753 (0.3478) acc1: 100.0000 (84.6809) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1950 data: 0.1031 max mem: 3536 +Test: [1240/2387] eta: 0:03:49 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0735 (0.3456) acc1: 100.0000 (84.8043) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1816 data: 0.0906 max mem: 3536 +Test: [1250/2387] eta: 0:03:48 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0750 (0.3435) acc1: 100.0000 (84.9258) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1887 data: 0.0966 max mem: 3536 +Test: [1260/2387] eta: 0:03:46 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0768 (0.3413) acc1: 100.0000 (85.0454) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2121 data: 0.1180 max mem: 3536 +Test: [1270/2387] eta: 0:03:44 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0727 (0.3392) acc1: 100.0000 (85.1630) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2206 data: 0.1262 max mem: 3536 +Test: [1280/2387] eta: 0:03:42 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0746 (0.3375) acc1: 100.0000 (85.2772) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2155 data: 0.1212 max mem: 3536 +Test: [1290/2387] eta: 0:03:40 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1159 (0.3359) acc1: 100.0000 (85.3759) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2127 data: 0.1188 max mem: 3536 +Test: [1300/2387] eta: 0:03:38 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1176 (0.3342) acc1: 99.4792 (85.4813) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2246 data: 0.1307 max mem: 3536 +Test: [1310/2387] eta: 0:03:36 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1160 (0.3326) acc1: 99.7396 (85.5845) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2013 data: 0.1081 max mem: 3536 +Test: [1320/2387] eta: 0:03:34 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1190 (0.3311) acc1: 99.7396 (85.6836) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1808 data: 0.0875 max mem: 3536 +Test: [1330/2387] eta: 0:03:32 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1151 (0.3295) acc1: 100.0000 (85.7902) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1939 data: 0.0983 max mem: 3536 +Test: [1340/2387] eta: 0:03:30 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1145 (0.3292) acc1: 99.7396 (85.8031) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1997 data: 0.1056 max mem: 3536 +Test: [1350/2387] eta: 0:03:28 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0752 (0.3273) acc1: 100.0000 (85.9082) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1956 data: 0.1025 max mem: 3537 +Test: [1360/2387] eta: 0:03:26 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0651 (0.3254) acc1: 100.0000 (86.0117) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1997 data: 0.1056 max mem: 3537 +Test: [1370/2387] eta: 0:03:24 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0639 (0.3235) acc1: 100.0000 (86.1138) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2144 data: 0.1207 max mem: 3537 +Test: [1380/2387] eta: 0:03:22 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0633 (0.3216) acc1: 100.0000 (86.2143) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2253 data: 0.1324 max mem: 3537 +Test: [1390/2387] eta: 0:03:20 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0636 (0.3198) acc1: 100.0000 (86.3134) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2087 data: 0.1172 max mem: 3537 +Test: [1400/2387] eta: 0:03:18 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0644 (0.3179) acc1: 100.0000 (86.4111) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2044 data: 0.1134 max mem: 3537 +Test: [1410/2387] eta: 0:03:16 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0653 (0.3166) acc1: 100.0000 (86.4945) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2187 data: 0.1269 max mem: 3537 +Test: [1420/2387] eta: 0:03:14 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1503 (0.3158) acc1: 98.6979 (86.5481) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2056 data: 0.1143 max mem: 3537 +Test: [1430/2387] eta: 0:03:13 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1466 (0.3146) acc1: 98.6979 (86.6321) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2222 data: 0.1315 max mem: 3537 +Test: [1440/2387] eta: 0:03:10 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1425 (0.3135) acc1: 99.2188 (86.7218) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2003 data: 0.1089 max mem: 3537 +Test: [1450/2387] eta: 0:03:08 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1341 (0.3123) acc1: 99.7396 (86.8106) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1651 data: 0.0741 max mem: 3537 +Test: [1460/2387] eta: 0:03:06 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1163 (0.3109) acc1: 99.7396 (86.8961) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1930 data: 0.1020 max mem: 3537 +Test: [1470/2387] eta: 0:03:04 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0978 (0.3094) acc1: 100.0000 (86.9850) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2100 data: 0.1171 max mem: 3537 +Test: [1480/2387] eta: 0:03:02 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0916 (0.3080) acc1: 100.0000 (87.0713) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2049 data: 0.1128 max mem: 3537 +Test: [1490/2387] eta: 0:03:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0897 (0.3065) acc1: 100.0000 (87.1568) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1994 data: 0.1089 max mem: 3537 +Test: [1500/2387] eta: 0:02:58 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0959 (0.3051) acc1: 100.0000 (87.2415) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1942 data: 0.1034 max mem: 3537 +Test: [1510/2387] eta: 0:02:56 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0893 (0.3037) acc1: 100.0000 (87.3259) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1858 data: 0.0948 max mem: 3537 +Test: [1520/2387] eta: 0:02:54 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0937 (0.3024) acc1: 100.0000 (87.4075) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1856 data: 0.0935 max mem: 3537 +Test: [1530/2387] eta: 0:02:52 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0898 (0.3010) acc1: 100.0000 (87.4898) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1795 data: 0.0878 max mem: 3537 +Test: [1540/2387] eta: 0:02:49 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0867 (0.2996) acc1: 100.0000 (87.5710) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1738 data: 0.0824 max mem: 3537 +Test: [1550/2387] eta: 0:02:47 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0863 (0.2982) acc1: 100.0000 (87.6509) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1753 data: 0.0827 max mem: 3537 +Test: [1560/2387] eta: 0:02:45 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0904 (0.2969) acc1: 100.0000 (87.7294) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1637 data: 0.0719 max mem: 3537 +Test: [1570/2387] eta: 0:02:43 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0886 (0.2956) acc1: 100.0000 (87.8075) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1632 data: 0.0717 max mem: 3537 +Test: [1580/2387] eta: 0:02:41 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0921 (0.2946) acc1: 100.0000 (87.8744) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1620 data: 0.0701 max mem: 3538 +Test: [1590/2387] eta: 0:02:38 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1484 (0.2937) acc1: 98.9583 (87.9300) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1549 data: 0.0622 max mem: 3538 +Test: [1600/2387] eta: 0:02:36 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1444 (0.2930) acc1: 97.6562 (87.9719) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1639 data: 0.0717 max mem: 3538 +Test: [1610/2387] eta: 0:02:34 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1257 (0.2919) acc1: 99.4792 (88.0444) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1699 data: 0.0757 max mem: 3538 +Test: [1620/2387] eta: 0:02:32 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1328 (0.2911) acc1: 99.4792 (88.1050) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1624 data: 0.0685 max mem: 3538 +Test: [1630/2387] eta: 0:02:30 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1434 (0.2902) acc1: 99.2188 (88.1714) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1609 data: 0.0699 max mem: 3538 +Test: [1640/2387] eta: 0:02:28 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1393 (0.2894) acc1: 99.2188 (88.2322) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1780 data: 0.0868 max mem: 3538 +Test: [1650/2387] eta: 0:02:26 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1364 (0.2884) acc1: 99.4792 (88.2988) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1744 data: 0.0834 max mem: 3538 +Test: [1660/2387] eta: 0:02:24 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1364 (0.2876) acc1: 99.4792 (88.3611) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1703 data: 0.0756 max mem: 3538 +Test: [1670/2387] eta: 0:02:22 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1373 (0.2867) acc1: 99.2188 (88.4224) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1816 data: 0.0870 max mem: 3538 +Test: [1680/2387] eta: 0:02:19 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1414 (0.2860) acc1: 98.9583 (88.4803) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1691 data: 0.0783 max mem: 3538 +Test: [1690/2387] eta: 0:02:17 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1449 (0.2852) acc1: 99.7396 (88.5427) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1627 data: 0.0704 max mem: 3538 +Test: [1700/2387] eta: 0:02:15 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1408 (0.2842) acc1: 99.7396 (88.6086) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1674 data: 0.0746 max mem: 3538 +Test: [1710/2387] eta: 0:02:13 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0831 (0.2830) acc1: 100.0000 (88.6750) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1549 data: 0.0634 max mem: 3538 +Test: [1720/2387] eta: 0:02:11 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0759 (0.2818) acc1: 100.0000 (88.7403) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1534 data: 0.0618 max mem: 3538 +Test: [1730/2387] eta: 0:02:09 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0738 (0.2806) acc1: 100.0000 (88.8051) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0786 max mem: 3538 +Test: [1740/2387] eta: 0:02:07 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0753 (0.2794) acc1: 100.0000 (88.8691) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1690 data: 0.0777 max mem: 3538 +Test: [1750/2387] eta: 0:02:05 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0767 (0.2783) acc1: 100.0000 (88.9325) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1576 data: 0.0664 max mem: 3538 +Test: [1760/2387] eta: 0:02:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0755 (0.2771) acc1: 100.0000 (88.9954) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1584 data: 0.0666 max mem: 3538 +Test: [1770/2387] eta: 0:02:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0791 (0.2761) acc1: 100.0000 (89.0551) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1622 data: 0.0701 max mem: 3538 +Test: [1780/2387] eta: 0:01:58 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0821 (0.2750) acc1: 100.0000 (89.1166) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1662 data: 0.0731 max mem: 3538 +Test: [1790/2387] eta: 0:01:56 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0816 (0.2739) acc1: 100.0000 (89.1769) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1703 data: 0.0771 max mem: 3538 +Test: [1800/2387] eta: 0:01:54 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0798 (0.2728) acc1: 100.0000 (89.2367) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1798 data: 0.0859 max mem: 3538 +Test: [1810/2387] eta: 0:01:52 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0784 (0.2718) acc1: 100.0000 (89.2960) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0806 max mem: 3539 +Test: [1820/2387] eta: 0:01:50 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0775 (0.2707) acc1: 100.0000 (89.3548) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1698 data: 0.0751 max mem: 3539 +Test: [1830/2387] eta: 0:01:48 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0776 (0.2697) acc1: 100.0000 (89.4129) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1855 data: 0.0917 max mem: 3539 +Test: [1840/2387] eta: 0:01:46 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0990 (0.2688) acc1: 100.0000 (89.4705) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1971 data: 0.1063 max mem: 3539 +Test: [1850/2387] eta: 0:01:45 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1001 (0.2678) acc1: 100.0000 (89.5273) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2189 data: 0.1279 max mem: 3539 +Test: [1860/2387] eta: 0:01:43 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0997 (0.2669) acc1: 100.0000 (89.5836) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2029 data: 0.1113 max mem: 3539 +Test: [1870/2387] eta: 0:01:41 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0991 (0.2660) acc1: 100.0000 (89.6393) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1878 data: 0.0952 max mem: 3539 +Test: [1880/2387] eta: 0:01:39 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0993 (0.2652) acc1: 100.0000 (89.6944) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1957 data: 0.1022 max mem: 3539 +Test: [1890/2387] eta: 0:01:37 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0996 (0.2643) acc1: 100.0000 (89.7489) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1818 data: 0.0888 max mem: 3539 +Test: [1900/2387] eta: 0:01:35 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1003 (0.2634) acc1: 100.0000 (89.8028) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1872 data: 0.0947 max mem: 3539 +Test: [1910/2387] eta: 0:01:33 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1094 (0.2626) acc1: 100.0000 (89.8560) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1914 data: 0.0983 max mem: 3539 +Test: [1920/2387] eta: 0:01:31 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1240 (0.2623) acc1: 100.0000 (89.8813) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1850 data: 0.0909 max mem: 3539 +Test: [1930/2387] eta: 0:01:29 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1285 (0.2615) acc1: 100.0000 (89.9326) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1939 data: 0.1011 max mem: 3539 +Test: [1940/2387] eta: 0:01:27 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1266 (0.2612) acc1: 100.0000 (89.9487) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1843 data: 0.0925 max mem: 3539 +Test: [1950/2387] eta: 0:01:25 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1230 (0.2604) acc1: 100.0000 (89.9997) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1794 data: 0.0873 max mem: 3539 +Test: [1960/2387] eta: 0:01:23 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1230 (0.2600) acc1: 100.0000 (90.0375) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1783 data: 0.0853 max mem: 3539 +Test: [1970/2387] eta: 0:01:21 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1630 (0.2595) acc1: 99.7396 (90.0828) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1809 data: 0.0887 max mem: 3539 +Test: [1980/2387] eta: 0:01:19 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1339 (0.2589) acc1: 99.2188 (90.1241) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1839 data: 0.0916 max mem: 3539 +Test: [1990/2387] eta: 0:01:17 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1218 (0.2582) acc1: 99.2188 (90.1697) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1588 data: 0.0651 max mem: 3539 +Test: [2000/2387] eta: 0:01:15 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1190 (0.2576) acc1: 99.2188 (90.2152) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1510 data: 0.0592 max mem: 3539 +Test: [2010/2387] eta: 0:01:13 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1182 (0.2569) acc1: 99.2188 (90.2593) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1615 data: 0.0713 max mem: 3539 +Test: [2020/2387] eta: 0:01:11 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1203 (0.2562) acc1: 99.2188 (90.3039) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1687 data: 0.0778 max mem: 3539 +Test: [2030/2387] eta: 0:01:09 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1205 (0.2555) acc1: 99.2188 (90.3477) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1791 data: 0.0871 max mem: 3540 +Test: [2040/2387] eta: 0:01:07 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1205 (0.2549) acc1: 99.2188 (90.3920) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1913 data: 0.1000 max mem: 3540 +Test: [2050/2387] eta: 0:01:05 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1154 (0.2542) acc1: 99.4792 (90.4359) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1740 data: 0.0835 max mem: 3540 +Test: [2060/2387] eta: 0:01:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1199 (0.2536) acc1: 99.2188 (90.4783) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1740 data: 0.0830 max mem: 3540 +Test: [2070/2387] eta: 0:01:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1189 (0.2529) acc1: 99.4792 (90.5225) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1980 data: 0.1060 max mem: 3540 +Test: [2080/2387] eta: 0:00:59 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1161 (0.2522) acc1: 99.4792 (90.5648) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1850 data: 0.0924 max mem: 3540 +Test: [2090/2387] eta: 0:00:57 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1171 (0.2516) acc1: 99.2188 (90.6068) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1658 data: 0.0739 max mem: 3540 +Test: [2100/2387] eta: 0:00:55 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1171 (0.2510) acc1: 99.4792 (90.6490) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1706 data: 0.0790 max mem: 3540 +Test: [2110/2387] eta: 0:00:53 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1163 (0.2503) acc1: 99.4792 (90.6906) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1771 data: 0.0845 max mem: 3540 +Test: [2120/2387] eta: 0:00:51 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1262 (0.2499) acc1: 99.4792 (90.7276) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1708 data: 0.0784 max mem: 3540 +Test: [2130/2387] eta: 0:00:49 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1378 (0.2494) acc1: 99.7396 (90.7685) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1673 data: 0.0760 max mem: 3540 +Test: [2140/2387] eta: 0:00:47 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1463 (0.2491) acc1: 99.2188 (90.7837) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1742 data: 0.0822 max mem: 3540 +Test: [2150/2387] eta: 0:00:45 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1285 (0.2485) acc1: 99.7396 (90.8257) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1739 data: 0.0810 max mem: 3540 +Test: [2160/2387] eta: 0:00:43 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1334 (0.2480) acc1: 100.0000 (90.8659) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1735 data: 0.0811 max mem: 3540 +Test: [2170/2387] eta: 0:00:41 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1489 (0.2476) acc1: 100.0000 (90.9041) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1686 data: 0.0760 max mem: 3540 +Test: [2180/2387] eta: 0:00:39 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1664 (0.2481) acc1: 98.1771 (90.8729) acc5: 100.0000 (100.0000) acc1_neg: 65.365 acc5_neg: 100.000 time: 0.1630 data: 0.0703 max mem: 3540 +Test: [2190/2387] eta: 0:00:37 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2385 (0.2483) acc1: 93.4896 (90.8664) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1738 data: 0.0795 max mem: 3540 +Test: [2200/2387] eta: 0:00:36 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1324 (0.2477) acc1: 98.9583 (90.9064) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1857 data: 0.0919 max mem: 3540 +Test: [2210/2387] eta: 0:00:34 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0989 (0.2470) acc1: 100.0000 (90.9474) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1838 data: 0.0919 max mem: 3540 +Test: [2220/2387] eta: 0:00:32 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0951 (0.2464) acc1: 100.0000 (90.9879) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1800 data: 0.0869 max mem: 3540 +Test: [2230/2387] eta: 0:00:30 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1002 (0.2457) acc1: 100.0000 (91.0274) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1807 data: 0.0877 max mem: 3540 +Test: [2240/2387] eta: 0:00:28 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1034 (0.2451) acc1: 100.0000 (91.0654) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1830 data: 0.0912 max mem: 3540 +Test: [2250/2387] eta: 0:00:26 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0958 (0.2444) acc1: 100.0000 (91.1051) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2000 data: 0.1082 max mem: 3540 +Test: [2260/2387] eta: 0:00:24 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0941 (0.2439) acc1: 100.0000 (91.1440) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1973 data: 0.1051 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1266 (0.2434) acc1: 100.0000 (91.1825) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1900 data: 0.0986 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1273 (0.2431) acc1: 100.0000 (91.2001) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1849 data: 0.0940 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1193 (0.2425) acc1: 100.0000 (91.2380) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1681 data: 0.0761 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1199 (0.2420) acc1: 100.0000 (91.2734) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1576 data: 0.0657 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1288 (0.2416) acc1: 100.0000 (91.3102) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1627 data: 0.0709 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1369 (0.2413) acc1: 100.0000 (91.3387) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1592 data: 0.0675 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1665 (0.2412) acc1: 99.4792 (91.3553) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1532 data: 0.0616 max mem: 3541 +Test: [2340/2387] eta: 0:00:09 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1712 (0.2409) acc1: 98.4375 (91.3833) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1630 data: 0.0718 max mem: 3541 +Test: [2350/2387] eta: 0:00:07 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1539 (0.2408) acc1: 98.9583 (91.3905) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1605 data: 0.0696 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1539 (0.2406) acc1: 98.6979 (91.4200) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1632 data: 0.0710 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1695 (0.2404) acc1: 98.4375 (91.4450) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1573 data: 0.0658 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1848 (0.2402) acc1: 98.6979 (91.4725) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1462 data: 0.0557 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1965 (0.2403) acc1: 98.1771 (91.4760) acc5: 100.0000 (100.0000) acc1_neg: 90.698 acc5_neg: 100.000 time: 0.1417 data: 0.0537 max mem: 3541 +Test: Total time: 0:07:36 (0.1911 s / it) +* Overall Acc@1 91.476 Acc@5 100.000 + target == real Acc@1 12.092 Acc@5 100.000 + target == fake Acc@1 97.051 Acc@5 100.000 + loss 0.240 + PR-AUC 0.9858 ROC-AUC 0.8353 EER 23.10% (thr=0.8701) +Accuracy of the network on the 916267 test images: 91.5% +[rank0]:[W725 17:57:18.228655040 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 17:57:38.361000 598471 torch/distributed/run.py:766] +W0725 17:57:38.361000 598471 torch/distributed/run.py:766] ***************************************** +W0725 17:57:38.361000 598471 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:57:38.361000 598471 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 4): env:// +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:58:05.059480970 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank4]:[W725 17:58:05.059481440 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:58:05.381693671 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:58:06.437390431 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:58:06.534173935 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:58:06.592123083 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +[rank0]:[W725 17:58:06.618079217 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:58:06.624187125 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed2_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed2_resampled_n116-041-023_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 0:54:59 acc1_pos: 20.833 acc5_pos: 100.000 loss: 1.2313 (1.2313) acc1: 20.8333 (20.8333) acc5: 100.0000 (100.0000) time: 6.3826 data: 4.6537 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/517] eta: 0:05:35 acc1_pos: 11.458 acc5_pos: 100.000 loss: 1.2634 (1.3039) acc1: 22.3958 (20.1231) acc5: 100.0000 (100.0000) time: 0.6620 data: 0.4234 max mem: 3531 +Test: [ 20/517] eta: 0:03:14 acc1_pos: 14.323 acc5_pos: 100.000 loss: 1.2679 (1.3312) acc1: 17.4479 (19.7297) acc5: 100.0000 (100.0000) time: 0.0909 data: 0.0003 max mem: 3531 +Test: [ 30/517] eta: 0:02:32 acc1_pos: 5.990 acc5_pos: 100.000 loss: 1.3719 (1.3712) acc1: 12.2396 (17.1455) acc5: 100.0000 (100.0000) time: 0.1229 data: 0.0316 max mem: 3531 +Test: [ 40/517] eta: 0:02:23 acc1_pos: 7.031 acc5_pos: 100.000 loss: 1.4548 (1.3930) acc1: 10.6771 (15.2820) acc5: 100.0000 (100.0000) time: 0.2049 data: 0.1083 max mem: 3531 +Test: [ 50/517] eta: 0:02:17 acc1_pos: 20.642 acc5_pos: 100.000 loss: 1.4029 (1.3385) acc1: 11.4583 (18.9389) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2630 data: 0.1649 max mem: 3531 +Test: [ 60/517] eta: 0:02:11 acc1_pos: 20.642 acc5_pos: 100.000 loss: 0.3110 (1.1662) acc1: 92.7083 (31.5958) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2614 data: 0.1670 max mem: 3531 +Test: [ 70/517] eta: 0:02:10 acc1_pos: 20.642 acc5_pos: 100.000 loss: 0.2908 (1.0492) acc1: 95.3125 (39.7374) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.2849 data: 0.1913 max mem: 3531 +Test: [ 80/517] eta: 0:02:06 acc1_pos: 20.642 acc5_pos: 100.000 loss: 0.3761 (0.9683) acc1: 86.4583 (45.2964) acc5: 100.0000 (100.0000) acc1_neg: 79.427 acc5_neg: 100.000 time: 0.2934 data: 0.2005 max mem: 3531 +Test: [ 90/517] eta: 0:02:02 acc1_pos: 20.642 acc5_pos: 100.000 loss: 0.2792 (0.8870) acc1: 95.3125 (51.1704) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2699 data: 0.1753 max mem: 3531 +Test: [100/517] eta: 0:01:57 acc1_pos: 20.642 acc5_pos: 100.000 loss: 0.2259 (0.8225) acc1: 98.9583 (55.7627) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2499 data: 0.1522 max mem: 3531 +Test: [110/517] eta: 0:01:54 acc1_pos: 20.642 acc5_pos: 100.000 loss: 0.2403 (0.7726) acc1: 95.0521 (58.9574) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2499 data: 0.1514 max mem: 3531 +Test: [120/517] eta: 0:01:49 acc1_pos: 20.642 acc5_pos: 100.000 loss: 0.2403 (0.7292) acc1: 95.3125 (62.1750) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2449 data: 0.1488 max mem: 3531 +Test: [130/517] eta: 0:01:46 acc1_pos: 20.642 acc5_pos: 100.000 loss: 0.2360 (0.6913) acc1: 99.4792 (64.9750) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2449 data: 0.1485 max mem: 3531 +Test: [140/517] eta: 0:01:41 acc1_pos: 20.642 acc5_pos: 100.000 loss: 0.2458 (0.6602) acc1: 96.8750 (67.1155) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2361 data: 0.1412 max mem: 3531 +Test: [150/517] eta: 0:01:37 acc1_pos: 20.642 acc5_pos: 100.000 loss: 0.2410 (0.6318) acc1: 96.6146 (69.0949) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2115 data: 0.1203 max mem: 3531 +Test: [160/517] eta: 0:01:35 acc1_pos: 20.642 acc5_pos: 100.000 loss: 0.2522 (0.6103) acc1: 95.0521 (70.5519) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.2587 data: 0.1656 max mem: 3531 +Test: [170/517] eta: 0:01:33 acc1_pos: 20.642 acc5_pos: 100.000 loss: 0.2699 (0.5894) acc1: 93.2292 (71.9237) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2883 data: 0.1923 max mem: 3531 +Test: [180/517] eta: 0:01:29 acc1_pos: 20.642 acc5_pos: 100.000 loss: 0.2129 (0.5669) acc1: 98.1771 (73.4231) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2549 data: 0.1573 max mem: 3531 +Test: [190/517] eta: 0:01:26 acc1_pos: 20.642 acc5_pos: 100.000 loss: 0.1612 (0.5457) acc1: 99.4792 (74.7955) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2249 data: 0.1279 max mem: 3531 +Test: [200/517] eta: 0:01:23 acc1_pos: 20.642 acc5_pos: 100.000 loss: 0.1548 (0.5265) acc1: 100.0000 (76.0404) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2299 data: 0.1345 max mem: 3531 +Test: [210/517] eta: 0:01:20 acc1_pos: 20.642 acc5_pos: 100.000 loss: 0.1823 (0.5174) acc1: 98.4375 (76.7945) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2299 data: 0.1348 max mem: 3532 +Test: [220/517] eta: 0:01:18 acc1_pos: 20.642 acc5_pos: 100.000 loss: 0.3446 (0.5097) acc1: 90.8854 (77.4274) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.2749 data: 0.1762 max mem: 3532 +Test: [230/517] eta: 0:01:17 acc1_pos: 20.642 acc5_pos: 100.000 loss: 0.3151 (0.4999) acc1: 94.5312 (78.2648) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.3699 data: 0.2695 max mem: 3532 +Test: [240/517] eta: 0:01:14 acc1_pos: 20.642 acc5_pos: 100.000 loss: 0.3043 (0.4928) acc1: 96.3542 (79.0024) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.3299 data: 0.2294 max mem: 3532 +Test: [250/517] eta: 0:01:11 acc1_pos: 20.642 acc5_pos: 100.000 loss: 0.3032 (0.4831) acc1: 97.3958 (79.7788) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2549 data: 0.1547 max mem: 3532 +Test: [260/517] eta: 0:01:08 acc1_pos: 20.642 acc5_pos: 100.000 loss: 0.2208 (0.4732) acc1: 99.4792 (80.5366) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2449 data: 0.1485 max mem: 3532 +Test: [270/517] eta: 0:01:06 acc1_pos: 20.642 acc5_pos: 100.000 loss: 0.2403 (0.4652) acc1: 99.2188 (81.1895) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2799 data: 0.1854 max mem: 3532 +Test: [280/517] eta: 0:01:04 acc1_pos: 20.642 acc5_pos: 100.000 loss: 0.2473 (0.4571) acc1: 98.1771 (81.7884) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.3349 data: 0.2401 max mem: 3532 +Test: [290/517] eta: 0:01:01 acc1_pos: 20.642 acc5_pos: 100.000 loss: 0.2362 (0.4494) acc1: 98.1771 (82.3633) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.3099 data: 0.2123 max mem: 3532 +Test: [300/517] eta: 0:00:58 acc1_pos: 20.642 acc5_pos: 100.000 loss: 0.2319 (0.4422) acc1: 98.6979 (82.9007) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1578 max mem: 3532 +Test: [310/517] eta: 0:00:56 acc1_pos: 20.642 acc5_pos: 100.000 loss: 0.2083 (0.4344) acc1: 99.7396 (83.4472) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2407 data: 0.1459 max mem: 3532 +Test: [320/517] eta: 0:00:53 acc1_pos: 20.642 acc5_pos: 100.000 loss: 0.1932 (0.4268) acc1: 100.0000 (83.9580) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2861 data: 0.1910 max mem: 3532 +Test: [330/517] eta: 0:00:51 acc1_pos: 20.642 acc5_pos: 100.000 loss: 0.1819 (0.4193) acc1: 100.0000 (84.4395) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.3240 data: 0.2300 max mem: 3532 +Test: [340/517] eta: 0:00:48 acc1_pos: 20.642 acc5_pos: 100.000 loss: 0.1769 (0.4131) acc1: 100.0000 (84.8256) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.2911 data: 0.1964 max mem: 3532 +Test: [350/517] eta: 0:00:45 acc1_pos: 20.642 acc5_pos: 100.000 loss: 0.2125 (0.4080) acc1: 99.4792 (85.2275) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2314 data: 0.1372 max mem: 3532 +Test: [360/517] eta: 0:00:42 acc1_pos: 20.642 acc5_pos: 100.000 loss: 0.2253 (0.4029) acc1: 98.9583 (85.6028) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2430 data: 0.1503 max mem: 3532 +Test: [370/517] eta: 0:00:39 acc1_pos: 20.642 acc5_pos: 100.000 loss: 0.1978 (0.3971) acc1: 99.7396 (85.9859) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2684 data: 0.1760 max mem: 3532 +Test: [380/517] eta: 0:00:37 acc1_pos: 20.642 acc5_pos: 100.000 loss: 0.1920 (0.3925) acc1: 99.7396 (86.2936) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2894 data: 0.1944 max mem: 3532 +Test: [390/517] eta: 0:00:34 acc1_pos: 20.642 acc5_pos: 100.000 loss: 0.2293 (0.3884) acc1: 97.1354 (86.5356) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2652 data: 0.1675 max mem: 3532 +Test: [400/517] eta: 0:00:31 acc1_pos: 20.642 acc5_pos: 100.000 loss: 0.2105 (0.3838) acc1: 98.6979 (86.8525) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2449 data: 0.1482 max mem: 3532 +Test: [410/517] eta: 0:00:28 acc1_pos: 20.642 acc5_pos: 100.000 loss: 0.2074 (0.3797) acc1: 99.7396 (87.1667) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2396 data: 0.1415 max mem: 3532 +Test: [420/517] eta: 0:00:26 acc1_pos: 20.642 acc5_pos: 100.000 loss: 0.2165 (0.3767) acc1: 100.0000 (87.4487) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2149 data: 0.1162 max mem: 3532 +Test: [430/517] eta: 0:00:23 acc1_pos: 20.642 acc5_pos: 100.000 loss: 0.2850 (0.3753) acc1: 98.6979 (87.6511) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.2449 data: 0.1498 max mem: 3532 +Test: [440/517] eta: 0:00:20 acc1_pos: 20.642 acc5_pos: 100.000 loss: 0.2850 (0.3727) acc1: 97.3958 (87.8313) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2463 data: 0.1524 max mem: 3533 +Test: [450/517] eta: 0:00:17 acc1_pos: 20.642 acc5_pos: 100.000 loss: 0.2298 (0.3696) acc1: 99.2188 (88.0780) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2171 data: 0.1227 max mem: 3533 +Test: [460/517] eta: 0:00:15 acc1_pos: 20.642 acc5_pos: 100.000 loss: 0.2318 (0.3670) acc1: 98.6979 (88.2400) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2246 data: 0.1304 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 20.642 acc5_pos: 100.000 loss: 0.1877 (0.3631) acc1: 98.9583 (88.4676) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2433 data: 0.1492 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 20.642 acc5_pos: 100.000 loss: 0.1792 (0.3600) acc1: 100.0000 (88.7041) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2391 data: 0.1469 max mem: 3533 +Test: [490/517] eta: 0:00:07 acc1_pos: 20.642 acc5_pos: 100.000 loss: 0.2235 (0.3572) acc1: 100.0000 (88.9145) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2293 data: 0.1349 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 20.642 acc5_pos: 100.000 loss: 0.2270 (0.3544) acc1: 98.9583 (89.1171) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2190 data: 0.1239 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 20.642 acc5_pos: 100.000 loss: 0.2280 (0.3520) acc1: 99.2188 (89.3071) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2186 data: 0.1272 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 20.642 acc5_pos: 100.000 loss: 0.2269 (0.3507) acc1: 99.7396 (89.4223) acc5: 100.0000 (100.0000) acc1_neg: 99.647 acc5_neg: 100.000 time: 0.2058 data: 0.1127 max mem: 3533 +Test: Total time: 0:02:14 (0.2608 s / it) +* Overall Acc@1 89.422 Acc@5 100.000 + target == real Acc@1 15.282 Acc@5 100.000 + target == fake Acc@1 97.114 Acc@5 100.000 + loss 0.351 + PR-AUC 0.9366 ROC-AUC 0.6500 EER 39.73% (thr=0.8018) +Accuracy of the network on the 198427 test images: 89.4% +[rank0]:[W725 18:00:35.705007765 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0725 18:00:55.060000 599060 torch/distributed/run.py:766] +W0725 18:00:55.060000 599060 torch/distributed/run.py:766] ***************************************** +W0725 18:00:55.060000 599060 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:00:55.060000 599060 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W725 18:01:21.603729024 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:01:21.676526183 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:01:21.253588933 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:01:21.310151882 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:01:21.394102047 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:01:22.429395018 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:01:22.475797962 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:01:22.551038544 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed2_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed2_resampled_n116-041-023_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:30:28 acc1_pos: 14.844 acc5_pos: 100.000 loss: 1.7777 (1.7777) acc1: 14.8438 (14.8438) acc5: 100.0000 (100.0000) time: 4.7015 data: 2.5829 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/389] eta: 0:03:21 acc1_pos: 40.104 acc5_pos: 100.000 loss: 1.4425 (1.3945) acc1: 14.8438 (16.3352) acc5: 100.0000 (100.0000) time: 0.5312 data: 0.2572 max mem: 3531 +Test: [ 20/389] eta: 0:02:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.4425 (1.4874) acc1: 16.9271 (15.7614) acc5: 100.0000 (100.0000) time: 0.1155 data: 0.0259 max mem: 3531 +Test: [ 30/389] eta: 0:01:40 acc1_pos: 0.260 acc5_pos: 100.000 loss: 1.8482 (1.5981) acc1: 0.2604 (11.1055) acc5: 100.0000 (100.0000) time: 0.1405 data: 0.0505 max mem: 3531 +Test: [ 40/389] eta: 0:01:39 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.8110 (1.5973) acc1: 0.5208 (10.4611) acc5: 100.0000 (100.0000) time: 0.2354 data: 0.1430 max mem: 3531 +Test: [ 50/389] eta: 0:01:32 acc1_pos: 81.510 acc5_pos: 100.000 loss: 1.4609 (1.5146) acc1: 3.1250 (15.4973) acc5: 100.0000 (100.0000) time: 0.2663 data: 0.1735 max mem: 3531 +Test: [ 60/389] eta: 0:01:26 acc1_pos: 5.469 acc5_pos: 100.000 loss: 1.2356 (1.4855) acc1: 7.0312 (19.6593) acc5: 100.0000 (100.0000) time: 0.2126 data: 0.1218 max mem: 3531 +Test: [ 70/389] eta: 0:01:22 acc1_pos: 16.667 acc5_pos: 100.000 loss: 1.3173 (1.4594) acc1: 8.3333 (19.4909) acc5: 100.0000 (100.0000) time: 0.2235 data: 0.1298 max mem: 3531 +Test: [ 80/389] eta: 0:01:17 acc1_pos: 41.667 acc5_pos: 100.000 loss: 1.4174 (1.4572) acc1: 8.3333 (20.1100) acc5: 100.0000 (100.0000) time: 0.2204 data: 0.1268 max mem: 3531 +Test: [ 90/389] eta: 0:01:13 acc1_pos: 18.490 acc5_pos: 100.000 loss: 1.0995 (1.4092) acc1: 21.0938 (21.8063) acc5: 100.0000 (100.0000) time: 0.1957 data: 0.1052 max mem: 3531 +Test: [100/389] eta: 0:01:09 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.3137 (1.4330) acc1: 17.1875 (20.6064) acc5: 100.0000 (100.0000) time: 0.1921 data: 0.0993 max mem: 3531 +Test: [110/389] eta: 0:01:06 acc1_pos: 90.625 acc5_pos: 100.000 loss: 1.5631 (1.4416) acc1: 5.9896 (21.6873) acc5: 100.0000 (100.0000) time: 0.1963 data: 0.1033 max mem: 3531 +Test: [120/389] eta: 0:01:02 acc1_pos: 14.583 acc5_pos: 100.000 loss: 1.4292 (1.4239) acc1: 32.0312 (22.7832) acc5: 100.0000 (100.0000) time: 0.1896 data: 0.0979 max mem: 3531 +Test: [130/389] eta: 0:00:59 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.9062 (1.3588) acc1: 40.6250 (26.6579) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1877 data: 0.0950 max mem: 3531 +Test: [140/389] eta: 0:00:55 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1566 (1.2788) acc1: 97.6562 (31.3054) acc5: 100.0000 (100.0000) acc1_neg: 64.323 acc5_neg: 100.000 time: 0.1833 data: 0.0893 max mem: 3531 +Test: [150/389] eta: 0:00:53 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1806 (1.2162) acc1: 98.9583 (34.8475) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1761 data: 0.0836 max mem: 3531 +Test: [160/389] eta: 0:00:50 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1806 (1.1507) acc1: 99.4792 (38.7228) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1887 data: 0.0968 max mem: 3531 +Test: [170/389] eta: 0:00:47 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0981 (1.0912) acc1: 100.0000 (42.2880) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1928 data: 0.1001 max mem: 3531 +Test: [180/389] eta: 0:00:45 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1146 (1.0416) acc1: 100.0000 (45.2909) acc5: 100.0000 (100.0000) acc1_neg: 66.927 acc5_neg: 100.000 time: 0.2013 data: 0.1086 max mem: 3531 +Test: [190/389] eta: 0:00:42 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1613 (1.0030) acc1: 100.0000 (47.6153) acc5: 100.0000 (100.0000) acc1_neg: 80.729 acc5_neg: 100.000 time: 0.1879 data: 0.0957 max mem: 3531 +Test: [200/389] eta: 0:00:40 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1431 (0.9652) acc1: 99.7396 (49.8471) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0917 max mem: 3531 +Test: [210/389] eta: 0:00:38 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1295 (0.9256) acc1: 100.0000 (52.2240) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1999 data: 0.1056 max mem: 3532 +Test: [220/389] eta: 0:00:36 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1094 (0.8892) acc1: 100.0000 (54.3658) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2114 data: 0.1178 max mem: 3532 +Test: [230/389] eta: 0:00:33 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1094 (0.8608) acc1: 100.0000 (56.1328) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.2067 data: 0.1133 max mem: 3532 +Test: [240/389] eta: 0:00:31 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1994 (0.8355) acc1: 98.6979 (57.7142) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1953 data: 0.1022 max mem: 3532 +Test: [250/389] eta: 0:00:29 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1632 (0.8067) acc1: 100.0000 (59.3989) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2013 data: 0.1084 max mem: 3532 +Test: [260/389] eta: 0:00:27 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1149 (0.7848) acc1: 100.0000 (60.5913) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1888 data: 0.0973 max mem: 3532 +Test: [270/389] eta: 0:00:24 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1485 (0.7652) acc1: 97.9167 (61.8687) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1789 data: 0.0882 max mem: 3532 +Test: [280/389] eta: 0:00:22 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1742 (0.7452) acc1: 98.6979 (63.0662) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1917 data: 0.1010 max mem: 3532 +Test: [290/389] eta: 0:00:20 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1960 (0.7294) acc1: 98.9583 (64.1055) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1911 data: 0.0982 max mem: 3532 +Test: [300/389] eta: 0:00:18 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1674 (0.7101) acc1: 100.0000 (65.2919) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1826 data: 0.0896 max mem: 3532 +Test: [310/389] eta: 0:00:16 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1682 (0.6961) acc1: 99.4792 (66.2145) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.1778 data: 0.0868 max mem: 3532 +Test: [320/389] eta: 0:00:14 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2045 (0.6797) acc1: 98.9583 (67.2305) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1620 data: 0.0699 max mem: 3532 +Test: [330/389] eta: 0:00:12 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1410 (0.6636) acc1: 100.0000 (68.1930) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1572 data: 0.0646 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1277 (0.6483) acc1: 100.0000 (69.0639) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1634 data: 0.0719 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1369 (0.6498) acc1: 99.7396 (68.5816) acc5: 100.0000 (100.0000) acc1_neg: 5.469 acc5_neg: 100.000 time: 0.1558 data: 0.0656 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.4849 (0.6494) acc1: 69.5312 (68.3136) acc5: 100.0000 (100.0000) acc1_neg: 75.781 acc5_neg: 100.000 time: 0.1489 data: 0.0585 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.4334 (0.6379) acc1: 76.5625 (69.0497) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1434 data: 0.0529 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1686 (0.6256) acc1: 99.7396 (69.8286) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1287 data: 0.0388 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1621 (0.6170) acc1: 99.7396 (70.3370) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1236 data: 0.0305 max mem: 3532 +Test: Total time: 0:01:15 (0.1951 s / it) +* Overall Acc@1 70.337 Acc@5 100.000 + target == real Acc@1 22.422 Acc@5 100.000 + target == fake Acc@1 92.609 Acc@5 100.000 + loss 0.617 + PR-AUC 0.8125 ROC-AUC 0.6982 EER 34.78% (thr=0.8472) +Accuracy of the network on the 149358 test images: 70.3% +[rank0]:[W725 18:02:47.359387995 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:03:07.277000 599625 torch/distributed/run.py:766] +W0725 18:03:07.277000 599625 torch/distributed/run.py:766] ***************************************** +W0725 18:03:07.277000 599625 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:03:07.277000 599625 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:03:35.498428111 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:03:35.545526184 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:03:35.715731240 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:03:35.965632456 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:03:35.985820974 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:03:35.119905141 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:03:35.290972717 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:03:35.317461268 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed2_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed2_resampled_n116-041-023_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:28:59 acc1_pos: 9.375 acc5_pos: 100.000 loss: 1.7832 (1.7832) acc1: 9.3750 (9.3750) acc5: 100.0000 (100.0000) time: 5.2242 data: 2.6969 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/333] eta: 0:03:08 acc1_pos: 82.812 acc5_pos: 100.000 loss: 1.4482 (1.1282) acc1: 34.8958 (44.6023) acc5: 100.0000 (100.0000) time: 0.5833 data: 0.2718 max mem: 3531 +Test: [ 20/333] eta: 0:01:57 acc1_pos: 92.708 acc5_pos: 100.000 loss: 0.3591 (0.7485) acc1: 85.4167 (65.2406) acc5: 100.0000 (100.0000) time: 0.1335 data: 0.0417 max mem: 3531 +Test: [ 30/333] eta: 0:01:40 acc1_pos: 41.667 acc5_pos: 100.000 loss: 0.4500 (0.8499) acc1: 74.7396 (55.6452) acc5: 100.0000 (100.0000) time: 0.1930 data: 0.0995 max mem: 3531 +Test: [ 40/333] eta: 0:01:40 acc1_pos: 34.115 acc5_pos: 100.000 loss: 1.0978 (0.9100) acc1: 33.0729 (49.8984) acc5: 100.0000 (100.0000) time: 0.3086 data: 0.2163 max mem: 3531 +Test: [ 50/333] eta: 0:01:25 acc1_pos: 29.948 acc5_pos: 100.000 loss: 1.1089 (0.9526) acc1: 29.9479 (46.0989) acc5: 100.0000 (100.0000) time: 0.2578 data: 0.1670 max mem: 3531 +Test: [ 60/333] eta: 0:01:16 acc1_pos: 19.792 acc5_pos: 100.000 loss: 1.1323 (0.9824) acc1: 29.9479 (43.4170) acc5: 100.0000 (100.0000) time: 0.1513 data: 0.0611 max mem: 3531 +Test: [ 70/333] eta: 0:01:12 acc1_pos: 21.354 acc5_pos: 100.000 loss: 1.1242 (1.0031) acc1: 30.7292 (41.5970) acc5: 100.0000 (100.0000) time: 0.2122 data: 0.1217 max mem: 3531 +Test: [ 80/333] eta: 0:01:07 acc1_pos: 34.375 acc5_pos: 100.000 loss: 1.1029 (1.0190) acc1: 31.2500 (40.2713) acc5: 100.0000 (100.0000) time: 0.2237 data: 0.1295 max mem: 3531 +Test: [ 90/333] eta: 0:01:00 acc1_pos: 44.531 acc5_pos: 100.000 loss: 1.0277 (1.0090) acc1: 35.4167 (40.5535) acc5: 100.0000 (100.0000) time: 0.1568 data: 0.0588 max mem: 3531 +Test: [100/333] eta: 0:00:56 acc1_pos: 49.740 acc5_pos: 100.000 loss: 0.8960 (1.0018) acc1: 42.9688 (40.9421) acc5: 100.0000 (100.0000) time: 0.1449 data: 0.0487 max mem: 3531 +Test: [110/333] eta: 0:00:53 acc1_pos: 41.927 acc5_pos: 100.000 loss: 0.9450 (0.9971) acc1: 41.6667 (41.0074) acc5: 100.0000 (100.0000) time: 0.1851 data: 0.0925 max mem: 3531 +Test: [120/333] eta: 0:00:51 acc1_pos: 46.875 acc5_pos: 100.000 loss: 0.9450 (0.9885) acc1: 41.9271 (41.4321) acc5: 100.0000 (100.0000) time: 0.2260 data: 0.1350 max mem: 3531 +Test: [130/333] eta: 0:00:47 acc1_pos: 59.375 acc5_pos: 100.000 loss: 0.8098 (0.9775) acc1: 46.8750 (42.0185) acc5: 100.0000 (100.0000) time: 0.2205 data: 0.1270 max mem: 3531 +Test: [140/333] eta: 0:00:44 acc1_pos: 69.271 acc5_pos: 100.000 loss: 0.8027 (0.9688) acc1: 50.0000 (42.4294) acc5: 100.0000 (100.0000) time: 0.1665 data: 0.0725 max mem: 3531 +Test: [150/333] eta: 0:00:42 acc1_pos: 52.344 acc5_pos: 100.000 loss: 0.8995 (0.9662) acc1: 41.1458 (42.4721) acc5: 100.0000 (100.0000) time: 0.2278 data: 0.1364 max mem: 3531 +Test: [160/333] eta: 0:00:39 acc1_pos: 38.487 acc5_pos: 100.000 loss: 0.8995 (0.9587) acc1: 42.7083 (42.9186) acc5: 100.0000 (100.0000) acc1_neg: 64.583 acc5_neg: 100.000 time: 0.2435 data: 0.1497 max mem: 3531 +Test: [170/333] eta: 0:00:36 acc1_pos: 38.487 acc5_pos: 100.000 loss: 0.7155 (0.9377) acc1: 59.8958 (44.4673) acc5: 100.0000 (100.0000) acc1_neg: 74.219 acc5_neg: 100.000 time: 0.1454 data: 0.0501 max mem: 3531 +Test: [180/333] eta: 0:00:34 acc1_pos: 38.487 acc5_pos: 100.000 loss: 0.5447 (0.9144) acc1: 72.3958 (46.1772) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.1786 data: 0.0834 max mem: 3531 +Test: [190/333] eta: 0:00:31 acc1_pos: 38.487 acc5_pos: 100.000 loss: 0.4930 (0.8905) acc1: 76.5625 (47.8949) acc5: 100.0000 (100.0000) acc1_neg: 73.958 acc5_neg: 100.000 time: 0.2013 data: 0.1072 max mem: 3531 +Test: [200/333] eta: 0:00:29 acc1_pos: 38.487 acc5_pos: 100.000 loss: 0.4856 (0.8712) acc1: 75.7812 (49.3107) acc5: 100.0000 (100.0000) acc1_neg: 72.396 acc5_neg: 100.000 time: 0.1749 data: 0.0816 max mem: 3531 +Test: [210/333] eta: 0:00:27 acc1_pos: 38.487 acc5_pos: 100.000 loss: 0.5352 (0.8566) acc1: 73.9583 (50.4480) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.2035 data: 0.1106 max mem: 3532 +Test: [220/333] eta: 0:00:24 acc1_pos: 38.487 acc5_pos: 100.000 loss: 0.5440 (0.8451) acc1: 73.1771 (51.3799) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.2149 data: 0.1227 max mem: 3532 +Test: [230/333] eta: 0:00:22 acc1_pos: 38.487 acc5_pos: 100.000 loss: 0.3959 (0.8240) acc1: 85.1562 (52.9683) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.2053 data: 0.1121 max mem: 3532 +Test: [240/333] eta: 0:00:20 acc1_pos: 38.487 acc5_pos: 100.000 loss: 0.3848 (0.8068) acc1: 86.7188 (54.2801) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.1998 data: 0.1076 max mem: 3532 +Test: [250/333] eta: 0:00:17 acc1_pos: 38.487 acc5_pos: 100.000 loss: 0.4254 (0.7941) acc1: 81.7708 (55.1917) acc5: 100.0000 (100.0000) acc1_neg: 64.062 acc5_neg: 100.000 time: 0.1882 data: 0.0948 max mem: 3532 +Test: [260/333] eta: 0:00:15 acc1_pos: 38.487 acc5_pos: 100.000 loss: 0.5040 (0.7833) acc1: 77.0833 (56.0065) acc5: 100.0000 (100.0000) acc1_neg: 72.396 acc5_neg: 100.000 time: 0.1873 data: 0.0938 max mem: 3532 +Test: [270/333] eta: 0:00:13 acc1_pos: 38.487 acc5_pos: 100.000 loss: 0.4765 (0.7710) acc1: 78.3854 (56.9236) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.2096 data: 0.1162 max mem: 3532 +Test: [280/333] eta: 0:00:11 acc1_pos: 38.487 acc5_pos: 100.000 loss: 0.4741 (0.7635) acc1: 78.3854 (57.4418) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.2037 data: 0.1100 max mem: 3532 +Test: [290/333] eta: 0:00:09 acc1_pos: 38.487 acc5_pos: 100.000 loss: 0.6513 (0.7603) acc1: 65.1042 (57.6648) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.1802 data: 0.0886 max mem: 3532 +Test: [300/333] eta: 0:00:07 acc1_pos: 38.487 acc5_pos: 100.000 loss: 0.6544 (0.7567) acc1: 65.3646 (57.9604) acc5: 100.0000 (100.0000) acc1_neg: 69.271 acc5_neg: 100.000 time: 0.1912 data: 0.1002 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 38.487 acc5_pos: 100.000 loss: 0.5924 (0.7467) acc1: 70.0521 (58.7160) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1716 data: 0.0807 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 38.487 acc5_pos: 100.000 loss: 0.4025 (0.7369) acc1: 84.3750 (59.4715) acc5: 100.0000 (100.0000) acc1_neg: 59.635 acc5_neg: 100.000 time: 0.1357 data: 0.0449 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 38.487 acc5_pos: 100.000 loss: 0.6755 (0.7370) acc1: 62.2396 (59.4836) acc5: 100.0000 (100.0000) acc1_neg: 62.240 acc5_neg: 100.000 time: 0.1318 data: 0.0416 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 38.487 acc5_pos: 100.000 loss: 0.7311 (0.7371) acc1: 60.4167 (59.4894) acc5: 100.0000 (100.0000) acc1_neg: 61.127 acc5_neg: 100.000 time: 0.1283 data: 0.0349 max mem: 3532 +Test: Total time: 0:01:08 (0.2063 s / it) +* Overall Acc@1 59.489 Acc@5 100.000 + target == real Acc@1 42.466 Acc@5 100.000 + target == fake Acc@1 74.643 Acc@5 100.000 + loss 0.737 + PR-AUC 0.6194 ROC-AUC 0.6202 EER 41.41% (thr=0.6509) +Accuracy of the network on the 127843 test images: 59.5% +[rank0]:[W725 18:04:56.515400252 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:05:14.443000 600188 torch/distributed/run.py:766] +W0725 18:05:14.443000 600188 torch/distributed/run.py:766] ***************************************** +W0725 18:05:14.443000 600188 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:05:14.443000 600188 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W725 18:05:44.272320883 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:05:45.280536626 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:05:45.369903092 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:05:46.477816739 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:05:46.507665395 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:05:46.524774352 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank1]:[W725 18:05:46.524854065 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:05:46.760422975 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed2_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed2_resampled_n116-041-023_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:19:40 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.2268 (0.2268) acc1: 95.3125 (95.3125) acc5: 100.0000 (100.0000) time: 5.0218 data: 3.1462 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/235] eta: 0:02:11 acc1_pos: 96.354 acc5_pos: 100.000 loss: 0.2298 (0.2286) acc1: 95.3125 (95.0047) acc5: 100.0000 (100.0000) time: 0.5846 data: 0.3324 max mem: 3531 +Test: [ 20/235] eta: 0:01:17 acc1_pos: 96.094 acc5_pos: 100.000 loss: 0.2258 (0.2254) acc1: 95.0521 (95.1141) acc5: 100.0000 (100.0000) time: 0.1280 data: 0.0371 max mem: 3531 +Test: [ 30/235] eta: 0:01:07 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.2213 (0.2228) acc1: 95.3125 (95.2621) acc5: 100.0000 (100.0000) time: 0.1875 data: 0.0932 max mem: 3531 +Test: [ 40/235] eta: 0:01:02 acc1_pos: 93.750 acc5_pos: 100.000 loss: 0.2197 (0.2234) acc1: 95.3125 (95.2109) acc5: 100.0000 (100.0000) time: 0.2777 data: 0.1833 max mem: 3531 +Test: [ 50/235] eta: 0:00:57 acc1_pos: 93.490 acc5_pos: 100.000 loss: 0.2254 (0.2245) acc1: 95.0521 (95.2053) acc5: 100.0000 (100.0000) time: 0.2870 data: 0.1929 max mem: 3531 +Test: [ 60/235] eta: 0:00:51 acc1_pos: 94.531 acc5_pos: 100.000 loss: 0.2308 (0.2262) acc1: 94.7917 (95.1204) acc5: 100.0000 (100.0000) time: 0.2333 data: 0.1399 max mem: 3531 +Test: [ 70/235] eta: 0:00:44 acc1_pos: 98.462 acc5_pos: 100.000 loss: 0.2480 (0.3541) acc1: 94.0104 (86.3923) acc5: 100.0000 (100.0000) acc1_neg: 23.177 acc5_neg: 100.000 time: 0.1689 data: 0.0773 max mem: 3531 +Test: [ 80/235] eta: 0:00:40 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.2440 (0.4653) acc1: 25.5208 (78.8162) acc5: 100.0000 (100.0000) acc1_neg: 24.219 acc5_neg: 100.000 time: 0.1605 data: 0.0682 max mem: 3531 +Test: [ 90/235] eta: 0:00:36 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.2581 (0.5525) acc1: 24.7396 (72.9567) acc5: 100.0000 (100.0000) acc1_neg: 23.698 acc5_neg: 100.000 time: 0.1760 data: 0.0805 max mem: 3531 +Test: [100/235] eta: 0:00:33 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.2562 (0.6218) acc1: 26.0417 (68.2936) acc5: 100.0000 (100.0000) acc1_neg: 21.354 acc5_neg: 100.000 time: 0.1880 data: 0.0921 max mem: 3531 +Test: [110/235] eta: 0:00:30 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.2483 (0.6774) acc1: 25.2604 (64.4215) acc5: 100.0000 (100.0000) acc1_neg: 25.260 acc5_neg: 100.000 time: 0.2040 data: 0.1093 max mem: 3531 +Test: [120/235] eta: 0:00:27 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.2966 (0.7452) acc1: 21.8750 (60.4081) acc5: 100.0000 (100.0000) acc1_neg: 16.406 acc5_neg: 100.000 time: 0.2099 data: 0.1157 max mem: 3531 +Test: [130/235] eta: 0:00:24 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.5214 (0.8048) acc1: 15.6250 (56.9756) acc5: 100.0000 (100.0000) acc1_neg: 15.885 acc5_neg: 100.000 time: 0.2023 data: 0.1096 max mem: 3531 +Test: [140/235] eta: 0:00:22 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.5473 (0.8577) acc1: 14.8438 (53.9746) acc5: 100.0000 (100.0000) acc1_neg: 12.760 acc5_neg: 100.000 time: 0.2140 data: 0.1211 max mem: 3531 +Test: [150/235] eta: 0:00:19 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.5389 (0.9014) acc1: 14.8438 (51.4211) acc5: 100.0000 (100.0000) acc1_neg: 14.583 acc5_neg: 100.000 time: 0.2209 data: 0.1287 max mem: 3531 +Test: [160/235] eta: 0:00:17 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.5249 (0.9403) acc1: 15.3646 (49.2026) acc5: 100.0000 (100.0000) acc1_neg: 17.188 acc5_neg: 100.000 time: 0.2096 data: 0.1165 max mem: 3531 +Test: [170/235] eta: 0:00:15 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4873 (0.9615) acc1: 17.7083 (47.7674) acc5: 100.0000 (100.0000) acc1_neg: 23.177 acc5_neg: 100.000 time: 0.2107 data: 0.1185 max mem: 3531 +Test: [180/235] eta: 0:00:12 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.2777 (0.9789) acc1: 25.0000 (46.5441) acc5: 100.0000 (100.0000) acc1_neg: 24.740 acc5_neg: 100.000 time: 0.2299 data: 0.1384 max mem: 3531 +Test: [190/235] eta: 0:00:10 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.2676 (0.9946) acc1: 24.7396 (45.4079) acc5: 100.0000 (100.0000) acc1_neg: 26.823 acc5_neg: 100.000 time: 0.2260 data: 0.1337 max mem: 3531 +Test: [200/235] eta: 0:00:08 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.2985 (1.0099) acc1: 23.6979 (44.3045) acc5: 100.0000 (100.0000) acc1_neg: 22.917 acc5_neg: 100.000 time: 0.1987 data: 0.1066 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.2925 (1.0222) acc1: 23.6979 (43.3760) acc5: 100.0000 (100.0000) acc1_neg: 23.177 acc5_neg: 100.000 time: 0.1683 data: 0.0768 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.2697 (1.0336) acc1: 24.4792 (42.5339) acc5: 100.0000 (100.0000) acc1_neg: 23.177 acc5_neg: 100.000 time: 0.1434 data: 0.0530 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.2697 (1.0437) acc1: 25.0000 (41.8234) acc5: 100.0000 (100.0000) acc1_neg: 23.958 acc5_neg: 100.000 time: 0.1374 data: 0.0474 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.2779 (1.0492) acc1: 25.2604 (41.5993) acc5: 100.0000 (100.0000) acc1_neg: 26.087 acc5_neg: 100.000 time: 0.1232 data: 0.0365 max mem: 3532 +Test: Total time: 0:00:50 (0.2160 s / it) +* Overall Acc@1 41.599 Acc@5 100.000 + target == real Acc@1 95.116 Acc@5 100.000 + target == fake Acc@1 22.243 Acc@5 100.000 + loss 1.049 + PR-AUC 0.8730 ROC-AUC 0.7265 EER 33.30% (thr=0.1862) +Accuracy of the network on the 89879 test images: 41.6% +[rank0]:[W725 18:06:44.840443515 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:07:02.972000 600757 torch/distributed/run.py:766] +W0725 18:07:02.972000 600757 torch/distributed/run.py:766] ***************************************** +W0725 18:07:02.972000 600757 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:07:02.972000 600757 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:07:30.158598281 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:07:31.495732174 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:07:31.500432121 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:07:31.637110731 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:07:31.789028624 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:07:31.801934514 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:07:31.844567986 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W725 18:07:31.847883412 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed2_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed2_resampled_n116-041-023_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:21:25 acc1_pos: 35.677 acc5_pos: 100.000 loss: 0.9446 (0.9446) acc1: 35.6771 (35.6771) acc5: 100.0000 (100.0000) time: 4.7792 data: 2.5494 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/269] eta: 0:02:26 acc1_pos: 45.052 acc5_pos: 100.000 loss: 0.7170 (0.7190) acc1: 58.0729 (56.4394) acc5: 100.0000 (100.0000) time: 0.5638 data: 0.2795 max mem: 3531 +Test: [ 20/269] eta: 0:01:27 acc1_pos: 58.854 acc5_pos: 100.000 loss: 0.6617 (0.6792) acc1: 58.8542 (59.4866) acc5: 100.0000 (100.0000) time: 0.1321 data: 0.0420 max mem: 3531 +Test: [ 30/269] eta: 0:01:15 acc1_pos: 64.583 acc5_pos: 100.000 loss: 0.6263 (0.6696) acc1: 61.9792 (60.8031) acc5: 100.0000 (100.0000) time: 0.1820 data: 0.0876 max mem: 3531 +Test: [ 40/269] eta: 0:01:10 acc1_pos: 72.414 acc5_pos: 100.000 loss: 0.5398 (0.6194) acc1: 70.5729 (65.8537) acc5: 100.0000 (100.0000) acc1_neg: 69.531 acc5_neg: 100.000 time: 0.2598 data: 0.1644 max mem: 3531 +Test: [ 50/269] eta: 0:01:02 acc1_pos: 72.414 acc5_pos: 100.000 loss: 0.5277 (0.6152) acc1: 73.6979 (66.1050) acc5: 100.0000 (100.0000) acc1_neg: 45.312 acc5_neg: 100.000 time: 0.2308 data: 0.1390 max mem: 3531 +Test: [ 60/269] eta: 0:00:56 acc1_pos: 72.414 acc5_pos: 100.000 loss: 0.6171 (0.6340) acc1: 62.2396 (64.1906) acc5: 100.0000 (100.0000) acc1_neg: 35.156 acc5_neg: 100.000 time: 0.1902 data: 0.0995 max mem: 3531 +Test: [ 70/269] eta: 0:00:52 acc1_pos: 72.414 acc5_pos: 100.000 loss: 0.6690 (0.6373) acc1: 52.3438 (63.5637) acc5: 100.0000 (100.0000) acc1_neg: 63.281 acc5_neg: 100.000 time: 0.2142 data: 0.1236 max mem: 3531 +Test: [ 80/269] eta: 0:00:48 acc1_pos: 72.414 acc5_pos: 100.000 loss: 0.6258 (0.6328) acc1: 63.2812 (64.2297) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.2119 data: 0.1199 max mem: 3531 +Test: [ 90/269] eta: 0:00:44 acc1_pos: 72.414 acc5_pos: 100.000 loss: 0.6258 (0.6380) acc1: 61.7188 (63.7076) acc5: 100.0000 (100.0000) acc1_neg: 47.135 acc5_neg: 100.000 time: 0.1899 data: 0.0971 max mem: 3531 +Test: [100/269] eta: 0:00:41 acc1_pos: 72.414 acc5_pos: 100.000 loss: 0.7001 (0.6473) acc1: 55.9896 (62.7682) acc5: 100.0000 (100.0000) acc1_neg: 49.479 acc5_neg: 100.000 time: 0.1928 data: 0.0980 max mem: 3531 +Test: [110/269] eta: 0:00:37 acc1_pos: 72.414 acc5_pos: 100.000 loss: 0.7034 (0.6520) acc1: 55.9896 (62.1598) acc5: 100.0000 (100.0000) acc1_neg: 47.396 acc5_neg: 100.000 time: 0.1935 data: 0.0988 max mem: 3531 +Test: [120/269] eta: 0:00:34 acc1_pos: 72.414 acc5_pos: 100.000 loss: 0.6431 (0.6448) acc1: 59.1146 (62.7776) acc5: 100.0000 (100.0000) acc1_neg: 48.958 acc5_neg: 100.000 time: 0.1898 data: 0.0978 max mem: 3531 +Test: [130/269] eta: 0:00:32 acc1_pos: 72.414 acc5_pos: 100.000 loss: 0.6106 (0.6463) acc1: 64.5833 (62.7465) acc5: 100.0000 (100.0000) acc1_neg: 53.125 acc5_neg: 100.000 time: 0.1938 data: 0.1008 max mem: 3531 +Test: [140/269] eta: 0:00:29 acc1_pos: 72.414 acc5_pos: 100.000 loss: 0.7272 (0.6579) acc1: 54.9479 (61.6929) acc5: 100.0000 (100.0000) acc1_neg: 34.115 acc5_neg: 100.000 time: 0.2041 data: 0.1114 max mem: 3531 +Test: [150/269] eta: 0:00:26 acc1_pos: 72.414 acc5_pos: 100.000 loss: 0.7375 (0.6630) acc1: 49.4792 (61.1755) acc5: 100.0000 (100.0000) acc1_neg: 70.573 acc5_neg: 100.000 time: 0.1916 data: 0.0998 max mem: 3531 +Test: [160/269] eta: 0:00:24 acc1_pos: 72.414 acc5_pos: 100.000 loss: 0.6148 (0.6616) acc1: 65.3646 (61.3160) acc5: 100.0000 (100.0000) acc1_neg: 65.885 acc5_neg: 100.000 time: 0.1753 data: 0.0827 max mem: 3531 +Test: [170/269] eta: 0:00:21 acc1_pos: 72.414 acc5_pos: 100.000 loss: 0.5970 (0.6596) acc1: 65.8854 (61.5710) acc5: 100.0000 (100.0000) acc1_neg: 73.698 acc5_neg: 100.000 time: 0.1895 data: 0.0974 max mem: 3531 +Test: [180/269] eta: 0:00:19 acc1_pos: 72.414 acc5_pos: 100.000 loss: 0.5682 (0.6500) acc1: 73.6979 (62.5432) acc5: 100.0000 (100.0000) acc1_neg: 64.323 acc5_neg: 100.000 time: 0.1892 data: 0.0955 max mem: 3531 +Test: [190/269] eta: 0:00:17 acc1_pos: 72.414 acc5_pos: 100.000 loss: 0.5537 (0.6471) acc1: 72.3958 (62.8599) acc5: 100.0000 (100.0000) acc1_neg: 68.750 acc5_neg: 100.000 time: 0.1860 data: 0.0924 max mem: 3531 +Test: [200/269] eta: 0:00:15 acc1_pos: 72.414 acc5_pos: 100.000 loss: 0.5910 (0.6462) acc1: 68.7500 (62.9897) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.2015 data: 0.1100 max mem: 3531 +Test: [210/269] eta: 0:00:12 acc1_pos: 72.414 acc5_pos: 100.000 loss: 0.6471 (0.6465) acc1: 62.2396 (62.9381) acc5: 100.0000 (100.0000) acc1_neg: 68.750 acc5_neg: 100.000 time: 0.2187 data: 0.1265 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 72.414 acc5_pos: 100.000 loss: 0.6043 (0.6468) acc1: 59.1146 (62.8335) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.2091 data: 0.1161 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 72.414 acc5_pos: 100.000 loss: 0.4729 (0.6388) acc1: 76.0417 (63.5552) acc5: 100.0000 (100.0000) acc1_neg: 76.042 acc5_neg: 100.000 time: 0.2091 data: 0.1167 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 72.414 acc5_pos: 100.000 loss: 0.6025 (0.6435) acc1: 65.6250 (63.1646) acc5: 100.0000 (100.0000) acc1_neg: 50.260 acc5_neg: 100.000 time: 0.2165 data: 0.1236 max mem: 3532 +Test: [250/269] eta: 0:00:04 acc1_pos: 72.414 acc5_pos: 100.000 loss: 0.7216 (0.6514) acc1: 50.2604 (62.4108) acc5: 100.0000 (100.0000) acc1_neg: 17.188 acc5_neg: 100.000 time: 0.1917 data: 0.0994 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 72.414 acc5_pos: 100.000 loss: 0.8490 (0.6616) acc1: 40.6250 (61.4783) acc5: 100.0000 (100.0000) acc1_neg: 49.740 acc5_neg: 100.000 time: 0.1646 data: 0.0746 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 72.414 acc5_pos: 100.000 loss: 0.8034 (0.6625) acc1: 43.4896 (61.2946) acc5: 100.0000 (100.0000) acc1_neg: 62.205 acc5_neg: 100.000 time: 0.1378 data: 0.0441 max mem: 3532 +Test: Total time: 0:00:56 (0.2110 s / it) +* Overall Acc@1 61.295 Acc@5 100.000 + target == real Acc@1 60.924 Acc@5 100.000 + target == fake Acc@1 61.345 Acc@5 100.000 + loss 0.663 + PR-AUC 0.9261 ROC-AUC 0.6571 EER 38.83% (thr=0.5015) +Accuracy of the network on the 103293 test images: 61.3% +[rank0]:[W725 18:08:37.093987382 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-25T18:08:44Z] eval_done tag=fixed500k_real_n23_seed2_resampled +[2026-07-25T18:08:45Z] eval_wait tag=fixed500k_fake_n45_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed2_resampled +[2026-07-25T18:09:45Z] eval_wait tag=fixed500k_fake_n45_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed2_resampled +[2026-07-25T18:10:45Z] eval_skip tag=fixed500k_fake_n45_seed2_resampled reason=complete_after_lock output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed2_resampled +[2026-07-25T18:10:45Z] eval_wait tag=fixed500k_real_n23_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed3_resampled +[2026-07-25T18:11:45Z] eval_wait tag=fixed500k_real_n23_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed3_resampled +[2026-07-25T18:12:45Z] eval_wait tag=fixed500k_real_n23_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed3_resampled +[2026-07-25T18:13:45Z] eval_wait tag=fixed500k_real_n23_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed3_resampled +[2026-07-25T18:14:46Z] eval_wait tag=fixed500k_real_n23_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed3_resampled +[2026-07-25T18:15:46Z] eval_wait tag=fixed500k_real_n23_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed3_resampled +[2026-07-25T18:16:46Z] eval_wait tag=fixed500k_real_n23_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed3_resampled +[2026-07-25T18:17:46Z] eval_wait tag=fixed500k_real_n23_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed3_resampled +[2026-07-25T18:18:46Z] eval_wait tag=fixed500k_real_n23_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed3_resampled +[2026-07-25T18:19:46Z] eval_wait tag=fixed500k_real_n23_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed3_resampled +[2026-07-25T18:20:47Z] eval_wait tag=fixed500k_real_n23_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed3_resampled +[2026-07-25T18:21:47Z] eval_wait tag=fixed500k_real_n23_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed3_resampled +[2026-07-25T18:22:47Z] eval_wait tag=fixed500k_real_n23_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed3_resampled +[2026-07-25T18:23:47Z] eval_wait tag=fixed500k_real_n23_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed3_resampled +[2026-07-25T18:24:47Z] eval_wait tag=fixed500k_real_n23_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed3_resampled +[2026-07-25T18:25:47Z] eval_wait tag=fixed500k_real_n23_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed3_resampled +[2026-07-25T18:26:47Z] eval_wait tag=fixed500k_real_n23_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed3_resampled +[2026-07-25T18:27:48Z] eval_wait tag=fixed500k_real_n23_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed3_resampled +[2026-07-25T18:28:48Z] eval_wait tag=fixed500k_real_n23_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed3_resampled +[2026-07-25T18:29:48Z] eval_wait tag=fixed500k_real_n23_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed3_resampled +[2026-07-25T18:30:48Z] eval_wait tag=fixed500k_real_n23_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed3_resampled +[2026-07-25T18:31:48Z] eval_wait tag=fixed500k_real_n23_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed3_resampled +[2026-07-25T18:32:48Z] eval_skip tag=fixed500k_real_n23_seed3_resampled reason=complete_after_lock output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed3_resampled +[2026-07-25T18:32:49Z] eval_wait tag=fixed500k_fake_n45_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed3_resampled +[2026-07-25T18:33:49Z] eval_wait tag=fixed500k_fake_n45_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed3_resampled +[2026-07-25T18:34:49Z] eval_wait tag=fixed500k_fake_n45_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed3_resampled +[2026-07-25T18:35:49Z] eval_wait tag=fixed500k_fake_n45_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed3_resampled +[2026-07-25T18:36:49Z] eval_wait tag=fixed500k_fake_n45_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed3_resampled +[2026-07-25T18:37:49Z] eval_wait tag=fixed500k_fake_n45_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed3_resampled +[2026-07-25T18:38:50Z] eval_wait tag=fixed500k_fake_n45_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed3_resampled +[2026-07-25T18:39:50Z] eval_wait tag=fixed500k_fake_n45_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed3_resampled +[2026-07-25T18:40:50Z] eval_wait tag=fixed500k_fake_n45_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed3_resampled +[2026-07-25T18:41:50Z] eval_wait tag=fixed500k_fake_n45_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed3_resampled +[2026-07-25T18:42:50Z] eval_wait tag=fixed500k_fake_n45_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed3_resampled +[2026-07-25T18:43:50Z] eval_wait tag=fixed500k_fake_n45_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed3_resampled +[2026-07-25T18:44:51Z] eval_wait tag=fixed500k_fake_n45_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed3_resampled +[2026-07-25T18:45:51Z] eval_wait tag=fixed500k_fake_n45_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed3_resampled +[2026-07-25T18:46:51Z] eval_wait tag=fixed500k_fake_n45_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed3_resampled +[2026-07-25T18:47:51Z] eval_wait tag=fixed500k_fake_n45_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed3_resampled +[2026-07-25T18:48:51Z] eval_wait tag=fixed500k_fake_n45_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed3_resampled +[2026-07-25T18:49:51Z] eval_wait tag=fixed500k_fake_n45_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed3_resampled +[2026-07-25T18:50:51Z] eval_wait tag=fixed500k_fake_n45_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed3_resampled +[2026-07-25T18:51:52Z] eval_wait tag=fixed500k_fake_n45_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed3_resampled +[2026-07-25T18:52:52Z] eval_wait tag=fixed500k_fake_n45_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed3_resampled +[2026-07-25T18:53:52Z] eval_wait tag=fixed500k_fake_n45_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed3_resampled +[2026-07-25T18:54:52Z] eval_skip tag=fixed500k_fake_n45_seed3_resampled reason=complete_after_lock output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed3_resampled +[2026-07-25T18:54:52Z] eval_wait tag=fixed500k_real_n23_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed4_resampled +[2026-07-25T18:55:52Z] eval_wait tag=fixed500k_real_n23_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed4_resampled +[2026-07-25T18:56:53Z] eval_wait tag=fixed500k_real_n23_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed4_resampled +[2026-07-25T18:57:53Z] eval_wait tag=fixed500k_real_n23_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed4_resampled +[2026-07-25T18:58:53Z] eval_wait tag=fixed500k_real_n23_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed4_resampled +[2026-07-25T18:59:53Z] eval_wait tag=fixed500k_real_n23_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed4_resampled +[2026-07-25T19:00:53Z] eval_wait tag=fixed500k_real_n23_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed4_resampled +[2026-07-25T19:01:54Z] eval_wait tag=fixed500k_real_n23_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed4_resampled +[2026-07-25T19:02:54Z] eval_wait tag=fixed500k_real_n23_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed4_resampled +[2026-07-25T19:03:54Z] eval_wait tag=fixed500k_real_n23_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed4_resampled +[2026-07-25T19:04:54Z] eval_wait tag=fixed500k_real_n23_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed4_resampled +[2026-07-25T19:05:54Z] eval_wait tag=fixed500k_real_n23_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed4_resampled +[2026-07-25T19:06:54Z] eval_wait tag=fixed500k_real_n23_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed4_resampled +[2026-07-25T19:07:55Z] eval_wait tag=fixed500k_real_n23_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed4_resampled +[2026-07-25T19:08:55Z] eval_wait tag=fixed500k_real_n23_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed4_resampled +[2026-07-25T19:09:55Z] eval_wait tag=fixed500k_real_n23_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed4_resampled +[2026-07-25T19:10:55Z] eval_wait tag=fixed500k_real_n23_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed4_resampled +[2026-07-25T19:11:55Z] eval_wait tag=fixed500k_real_n23_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed4_resampled +[2026-07-25T19:12:55Z] eval_wait tag=fixed500k_real_n23_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed4_resampled +[2026-07-25T19:13:56Z] eval_wait tag=fixed500k_real_n23_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed4_resampled +[2026-07-25T19:14:56Z] eval_wait tag=fixed500k_real_n23_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed4_resampled +[2026-07-25T19:15:56Z] eval_wait tag=fixed500k_real_n23_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed4_resampled +[2026-07-25T19:16:56Z] eval_skip tag=fixed500k_real_n23_seed4_resampled reason=complete_after_lock output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed4_resampled +[2026-07-25T19:16:56Z] eval_wait tag=fixed500k_fake_n45_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed4_resampled +[2026-07-25T19:17:56Z] eval_wait tag=fixed500k_fake_n45_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed4_resampled +[2026-07-25T19:18:57Z] eval_wait tag=fixed500k_fake_n45_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed4_resampled +[2026-07-25T19:19:57Z] eval_wait tag=fixed500k_fake_n45_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed4_resampled +[2026-07-25T19:20:57Z] eval_wait tag=fixed500k_fake_n45_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed4_resampled +[2026-07-25T19:21:57Z] eval_wait tag=fixed500k_fake_n45_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed4_resampled +[2026-07-25T19:22:57Z] eval_wait tag=fixed500k_fake_n45_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed4_resampled +[2026-07-25T19:23:57Z] eval_wait tag=fixed500k_fake_n45_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed4_resampled +[2026-07-25T19:24:58Z] eval_wait tag=fixed500k_fake_n45_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed4_resampled +[2026-07-25T19:25:58Z] eval_wait tag=fixed500k_fake_n45_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed4_resampled +[2026-07-25T19:26:58Z] eval_wait tag=fixed500k_fake_n45_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed4_resampled +[2026-07-25T19:27:58Z] eval_wait tag=fixed500k_fake_n45_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed4_resampled +[2026-07-25T19:28:58Z] eval_wait tag=fixed500k_fake_n45_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed4_resampled +[2026-07-25T19:29:58Z] eval_wait tag=fixed500k_fake_n45_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed4_resampled +[2026-07-25T19:30:59Z] eval_wait tag=fixed500k_fake_n45_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed4_resampled +[2026-07-25T19:31:59Z] eval_wait tag=fixed500k_fake_n45_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed4_resampled +[2026-07-25T19:32:59Z] eval_wait tag=fixed500k_fake_n45_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed4_resampled +[2026-07-25T19:33:59Z] eval_wait tag=fixed500k_fake_n45_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed4_resampled +[2026-07-25T19:34:59Z] eval_wait tag=fixed500k_fake_n45_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed4_resampled +[2026-07-25T19:35:59Z] eval_wait tag=fixed500k_fake_n45_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed4_resampled +[2026-07-25T19:37:00Z] eval_wait tag=fixed500k_fake_n45_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed4_resampled +[2026-07-25T19:38:00Z] eval_wait tag=fixed500k_fake_n45_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed4_resampled +[2026-07-25T19:39:00Z] eval_skip tag=fixed500k_fake_n45_seed4_resampled reason=complete_after_lock output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed4_resampled +eval_queue_done_utc=2026-07-25T19:39:00Z diff --git a/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_112559/eval_queue_n116-041-023.pid b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_112559/eval_queue_n116-041-023.pid new file mode 100644 index 0000000000000000000000000000000000000000..1c32df3d6b711c141086b418f74cc3992aca680e --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_112559/eval_queue_n116-041-023.pid @@ -0,0 +1 @@ +588816 diff --git a/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_112559/eval_queue_n116-041-153.log b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_112559/eval_queue_n116-041-153.log new file mode 100644 index 0000000000000000000000000000000000000000..db7d799436f274bc12a5639021f0d983e676056a --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_112559/eval_queue_n116-041-153.log @@ -0,0 +1,4155 @@ +eval_queue_started_utc=2026-07-25T17:02:40Z +hostname=n116-041-153 +train_tag=20260725_112559 +eval_tag=20260725_112559 +real_level=8 +fake_level=16 +eval_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559 +batch_per_gpu=256 +[2026-07-25T17:02:41Z] eval_wait tag=fixed500k_real_n8_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed1_resampled +[2026-07-25T17:03:41Z] eval_wait tag=fixed500k_real_n8_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed1_resampled +[2026-07-25T17:04:41Z] eval_wait tag=fixed500k_real_n8_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed1_resampled +[2026-07-25T17:05:41Z] eval_wait tag=fixed500k_real_n8_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed1_resampled +[2026-07-25T17:06:41Z] eval_wait tag=fixed500k_real_n8_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed1_resampled +[2026-07-25T17:07:41Z] eval_wait tag=fixed500k_real_n8_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed1_resampled +[2026-07-25T17:08:41Z] eval_wait tag=fixed500k_real_n8_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed1_resampled +[2026-07-25T17:09:41Z] eval_wait tag=fixed500k_real_n8_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed1_resampled +[2026-07-25T17:10:42Z] eval_wait tag=fixed500k_real_n8_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed1_resampled +[2026-07-25T17:11:42Z] eval_wait tag=fixed500k_real_n8_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed1_resampled +[2026-07-25T17:12:42Z] eval_wait tag=fixed500k_real_n8_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed1_resampled +[2026-07-25T17:13:42Z] eval_wait tag=fixed500k_real_n8_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed1_resampled +[2026-07-25T17:14:42Z] eval_wait tag=fixed500k_real_n8_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed1_resampled +[2026-07-25T17:15:42Z] eval_wait tag=fixed500k_real_n8_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed1_resampled +[2026-07-25T17:16:42Z] eval_wait tag=fixed500k_real_n8_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed1_resampled +[2026-07-25T17:17:43Z] eval_wait tag=fixed500k_real_n8_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed1_resampled +[2026-07-25T17:18:43Z] eval_skip tag=fixed500k_real_n8_seed1_resampled reason=complete_after_lock output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed1_resampled +[2026-07-25T17:18:43Z] eval_skip tag=fixed500k_fake_n16_seed1_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n16_seed1_resampled +[2026-07-25T17:18:43Z] eval_wait tag=fixed500k_real_n8_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed2_resampled +[2026-07-25T17:19:43Z] eval_wait tag=fixed500k_real_n8_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed2_resampled +[2026-07-25T17:20:43Z] eval_wait tag=fixed500k_real_n8_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed2_resampled +[2026-07-25T17:21:43Z] eval_wait tag=fixed500k_real_n8_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed2_resampled +[2026-07-25T17:22:43Z] eval_wait tag=fixed500k_real_n8_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed2_resampled +[2026-07-25T17:23:44Z] eval_wait tag=fixed500k_real_n8_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed2_resampled +[2026-07-25T17:24:44Z] eval_wait tag=fixed500k_real_n8_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed2_resampled +[2026-07-25T17:25:44Z] eval_wait tag=fixed500k_real_n8_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed2_resampled +[2026-07-25T17:26:44Z] eval_wait tag=fixed500k_real_n8_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed2_resampled +[2026-07-25T17:27:44Z] eval_wait tag=fixed500k_real_n8_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed2_resampled +[2026-07-25T17:28:44Z] eval_wait tag=fixed500k_real_n8_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed2_resampled +[2026-07-25T17:29:45Z] eval_wait tag=fixed500k_real_n8_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed2_resampled +[2026-07-25T17:30:45Z] eval_wait tag=fixed500k_real_n8_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed2_resampled +[2026-07-25T17:31:45Z] eval_wait tag=fixed500k_real_n8_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed2_resampled +[2026-07-25T17:32:45Z] eval_wait tag=fixed500k_real_n8_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed2_resampled +[2026-07-25T17:33:45Z] eval_wait tag=fixed500k_real_n8_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed2_resampled +[2026-07-25T17:34:45Z] eval_wait tag=fixed500k_real_n8_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed2_resampled +[2026-07-25T17:35:45Z] eval_wait tag=fixed500k_real_n8_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed2_resampled +[2026-07-25T17:36:46Z] eval_wait tag=fixed500k_real_n8_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed2_resampled +[2026-07-25T17:37:46Z] eval_wait tag=fixed500k_real_n8_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed2_resampled +[2026-07-25T17:38:46Z] eval_wait tag=fixed500k_real_n8_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed2_resampled +[2026-07-25T17:39:46Z] eval_wait tag=fixed500k_real_n8_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed2_resampled +[2026-07-25T17:40:46Z] eval_skip tag=fixed500k_real_n8_seed2_resampled reason=complete_after_wait output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed2_resampled +[2026-07-25T17:40:47Z] eval_skip tag=fixed500k_fake_n16_seed2_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n16_seed2_resampled +[2026-07-25T17:40:47Z] eval_start tag=fixed500k_real_n8_seed3_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed3_resampled_n116-041-153_20260725_112559/best_checkpoint.pth +started_utc=2026-07-25T17:40:47Z +hostname=n116-041-153 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed3_resampled_n116-041-153_20260725_112559/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed3_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0725 17:41:00.543000 597722 torch/distributed/run.py:766] +W0725 17:41:00.543000 597722 torch/distributed/run.py:766] ***************************************** +W0725 17:41:00.543000 597722 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:41:00.543000 597722 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:41:29.094621358 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:41:29.469329775 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W725 17:41:30.926538490 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W725 17:41:30.926668514 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:41:30.216367851 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:41:30.293956587 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:41:30.296353080 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:41:30.324886527 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed3_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed3_resampled_n116-041-153_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:19:15 acc1_pos: 81.771 acc5_pos: 100.000 loss: 0.4723 (0.4723) acc1: 81.7708 (81.7708) acc5: 100.0000 (100.0000) time: 4.9396 data: 2.8856 max mem: 3531 +Test: [ 10/234] eta: 0:02:06 acc1_pos: 88.021 acc5_pos: 100.000 loss: 0.4723 (0.4858) acc1: 81.7708 (77.8409) acc5: 100.0000 (100.0000) time: 0.5637 data: 0.2954 max mem: 3531 +Test: [ 20/234] eta: 0:01:14 acc1_pos: 92.188 acc5_pos: 100.000 loss: 0.3540 (0.4303) acc1: 85.9375 (82.3289) acc5: 100.0000 (100.0000) time: 0.1194 data: 0.0272 max mem: 3531 +Test: [ 30/234] eta: 0:01:02 acc1_pos: 93.750 acc5_pos: 100.000 loss: 0.2825 (0.3814) acc1: 92.1875 (85.6771) acc5: 100.0000 (100.0000) time: 0.1639 data: 0.0699 max mem: 3531 +Test: [ 40/234] eta: 0:00:58 acc1_pos: 98.958 acc5_pos: 100.000 loss: 0.2661 (0.3738) acc1: 92.4479 (86.4075) acc5: 100.0000 (100.0000) time: 0.2526 data: 0.1608 max mem: 3531 +Test: [ 50/234] eta: 0:00:51 acc1_pos: 97.135 acc5_pos: 100.000 loss: 0.1995 (0.3381) acc1: 96.6146 (88.5672) acc5: 100.0000 (100.0000) time: 0.2401 data: 0.1489 max mem: 3531 +Test: [ 60/234] eta: 0:00:46 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.1623 (0.3022) acc1: 98.1771 (90.3560) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1942 data: 0.1008 max mem: 3531 +Test: [ 70/234] eta: 0:00:42 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.1181 (0.3119) acc1: 99.2188 (89.2276) acc5: 100.0000 (100.0000) acc1_neg: 73.698 acc5_neg: 100.000 time: 0.1998 data: 0.1043 max mem: 3531 +Test: [ 80/234] eta: 0:00:38 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.4659 (0.3208) acc1: 74.7396 (88.4613) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1932 data: 0.0986 max mem: 3531 +Test: [ 90/234] eta: 0:00:35 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.2261 (0.3048) acc1: 93.7500 (89.1970) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.1935 data: 0.1011 max mem: 3531 +Test: [100/234] eta: 0:00:32 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.2806 (0.3168) acc1: 90.3646 (88.1601) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1999 data: 0.1070 max mem: 3531 +Test: [110/234] eta: 0:00:28 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.3779 (0.3195) acc1: 80.7292 (87.8543) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.1863 data: 0.0911 max mem: 3531 +Test: [120/234] eta: 0:00:26 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.3368 (0.3402) acc1: 84.1146 (86.2905) acc5: 100.0000 (100.0000) acc1_neg: 40.885 acc5_neg: 100.000 time: 0.1900 data: 0.0943 max mem: 3531 +Test: [130/234] eta: 0:00:23 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.5874 (0.3661) acc1: 67.4479 (84.3730) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.1921 data: 0.0976 max mem: 3531 +Test: [140/234] eta: 0:00:21 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.5879 (0.3864) acc1: 65.3646 (83.1967) acc5: 100.0000 (100.0000) acc1_neg: 41.667 acc5_neg: 100.000 time: 0.1827 data: 0.0888 max mem: 3531 +Test: [150/234] eta: 0:00:18 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.4736 (0.3939) acc1: 76.8229 (82.5831) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.1849 data: 0.0922 max mem: 3531 +Test: [160/234] eta: 0:00:16 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.4147 (0.4018) acc1: 76.8229 (82.0102) acc5: 100.0000 (100.0000) acc1_neg: 59.115 acc5_neg: 100.000 time: 0.1943 data: 0.0977 max mem: 3531 +Test: [170/234] eta: 0:00:13 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.6289 (0.4157) acc1: 63.8021 (80.9637) acc5: 100.0000 (100.0000) acc1_neg: 74.740 acc5_neg: 100.000 time: 0.1957 data: 0.0985 max mem: 3531 +Test: [180/234] eta: 0:00:11 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.6167 (0.4262) acc1: 64.5833 (80.1119) acc5: 100.0000 (100.0000) acc1_neg: 38.802 acc5_neg: 100.000 time: 0.1854 data: 0.0919 max mem: 3531 +Test: [190/234] eta: 0:00:09 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.5178 (0.4319) acc1: 74.4792 (79.7748) acc5: 100.0000 (100.0000) acc1_neg: 78.646 acc5_neg: 100.000 time: 0.1902 data: 0.0974 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.5068 (0.4236) acc1: 76.5625 (80.3211) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1954 data: 0.1045 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.1193 (0.4108) acc1: 99.4792 (81.1624) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1967 data: 0.1057 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.2146 (0.4036) acc1: 95.8333 (81.7131) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.1460 data: 0.0536 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.4669 (0.4269) acc1: 44.2708 (79.8746) acc5: 100.0000 (100.0000) acc1_neg: 38.802 acc5_neg: 100.000 time: 0.1252 data: 0.0336 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.9134 (0.4333) acc1: 40.8560 (79.4325) acc5: 100.0000 (100.0000) acc1_neg: 40.856 acc5_neg: 100.000 time: 0.1258 data: 0.0335 max mem: 3532 +Test: Total time: 0:00:48 (0.2054 s / it) +* Overall Acc@1 79.433 Acc@5 100.000 + target == real Acc@1 88.626 Acc@5 100.000 + target == fake Acc@1 76.841 Acc@5 100.000 + loss 0.433 + PR-AUC 0.9724 ROC-AUC 0.9064 EER 17.75% (thr=0.4146) +Accuracy of the network on the 89729 test images: 79.4% +[rank0]:[W725 17:42:27.112423388 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 17:42:46.958000 598308 torch/distributed/run.py:766] +W0725 17:42:46.958000 598308 torch/distributed/run.py:766] ***************************************** +W0725 17:42:46.958000 598308 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:42:46.958000 598308 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:43:17.646285769 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 17:43:17.864979702 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:43:17.221192519 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:43:17.273605252 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:43:17.289990936 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:43:17.441031695 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:43:18.509306136 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W725 17:43:18.509306927 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed3_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed3_resampled_n116-041-153_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 3:47:54 acc1_pos: 61.198 acc5_pos: 100.000 loss: 0.7257 (0.7257) acc1: 61.1979 (61.1979) acc5: 100.0000 (100.0000) time: 5.7288 data: 3.7793 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/2387] eta: 0:24:19 acc1_pos: 48.958 acc5_pos: 100.000 loss: 0.9768 (1.0116) acc1: 35.1562 (34.5407) acc5: 100.0000 (100.0000) time: 0.6141 data: 0.3555 max mem: 3531 +Test: [ 20/2387] eta: 0:14:51 acc1_pos: 44.531 acc5_pos: 100.000 loss: 0.9211 (0.9203) acc1: 37.7604 (40.6126) acc5: 100.0000 (100.0000) time: 0.1092 data: 0.0196 max mem: 3531 +Test: [ 30/2387] eta: 0:11:54 acc1_pos: 4.167 acc5_pos: 100.000 loss: 0.9211 (0.9803) acc1: 33.3333 (37.1052) acc5: 100.0000 (100.0000) time: 0.1324 data: 0.0415 max mem: 3531 +Test: [ 40/2387] eta: 0:12:01 acc1_pos: 48.698 acc5_pos: 100.000 loss: 1.0589 (0.9977) acc1: 22.3958 (35.4865) acc5: 100.0000 (100.0000) time: 0.2349 data: 0.1435 max mem: 3531 +Test: [ 50/2387] eta: 0:13:11 acc1_pos: 64.844 acc5_pos: 100.000 loss: 0.9441 (0.9709) acc1: 32.8125 (37.4234) acc5: 100.0000 (100.0000) time: 0.3935 data: 0.3008 max mem: 3531 +Test: [ 60/2387] eta: 0:11:44 acc1_pos: 73.438 acc5_pos: 100.000 loss: 0.7327 (0.9254) acc1: 50.7812 (40.7872) acc5: 100.0000 (100.0000) time: 0.2932 data: 0.1988 max mem: 3531 +Test: [ 70/2387] eta: 0:11:30 acc1_pos: 68.490 acc5_pos: 100.000 loss: 0.7097 (0.8992) acc1: 57.2917 (42.9504) acc5: 100.0000 (100.0000) time: 0.1949 data: 0.0987 max mem: 3531 +Test: [ 80/2387] eta: 0:11:11 acc1_pos: 52.865 acc5_pos: 100.000 loss: 0.7097 (0.8814) acc1: 56.5104 (44.1390) acc5: 100.0000 (100.0000) time: 0.2547 data: 0.1567 max mem: 3531 +Test: [ 90/2387] eta: 0:10:55 acc1_pos: 53.125 acc5_pos: 100.000 loss: 0.6961 (0.8607) acc1: 53.1250 (45.7360) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1417 max mem: 3531 +Test: [ 100/2387] eta: 0:10:42 acc1_pos: 65.365 acc5_pos: 100.000 loss: 0.7239 (0.8510) acc1: 57.8125 (46.8879) acc5: 100.0000 (100.0000) time: 0.2411 data: 0.1445 max mem: 3531 +Test: [ 110/2387] eta: 0:10:33 acc1_pos: 52.604 acc5_pos: 100.000 loss: 0.7439 (0.8418) acc1: 54.9479 (47.5366) acc5: 100.0000 (100.0000) time: 0.2458 data: 0.1518 max mem: 3531 +Test: [ 120/2387] eta: 0:10:25 acc1_pos: 68.490 acc5_pos: 100.000 loss: 0.7217 (0.8289) acc1: 54.9479 (48.5968) acc5: 100.0000 (100.0000) time: 0.2486 data: 0.1539 max mem: 3531 +Test: [ 130/2387] eta: 0:10:12 acc1_pos: 61.198 acc5_pos: 100.000 loss: 0.7409 (0.8252) acc1: 53.9062 (48.8490) acc5: 100.0000 (100.0000) time: 0.2339 data: 0.1368 max mem: 3531 +Test: [ 140/2387] eta: 0:10:02 acc1_pos: 68.229 acc5_pos: 100.000 loss: 0.8075 (0.8231) acc1: 48.6979 (48.9306) acc5: 100.0000 (100.0000) time: 0.2211 data: 0.1237 max mem: 3531 +Test: [ 150/2387] eta: 0:09:56 acc1_pos: 44.531 acc5_pos: 100.000 loss: 0.7785 (0.8171) acc1: 45.8333 (49.3412) acc5: 100.0000 (100.0000) time: 0.2349 data: 0.1390 max mem: 3531 +Test: [ 160/2387] eta: 0:09:42 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.6676 (0.7987) acc1: 57.5521 (50.6567) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2147 data: 0.1195 max mem: 3531 +Test: [ 170/2387] eta: 0:09:29 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2135 (0.7640) acc1: 94.2708 (53.2514) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1850 data: 0.0913 max mem: 3531 +Test: [ 180/2387] eta: 0:09:20 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2089 (0.7333) acc1: 94.7917 (55.5680) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1938 data: 0.0984 max mem: 3531 +Test: [ 190/2387] eta: 0:09:11 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2053 (0.7057) acc1: 94.7917 (57.6148) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1997 data: 0.1033 max mem: 3531 +Test: [ 200/2387] eta: 0:09:01 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2021 (0.6807) acc1: 94.2708 (59.4605) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1899 data: 0.0945 max mem: 3531 +Test: [ 210/2387] eta: 0:08:56 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2056 (0.6584) acc1: 95.0521 (61.1473) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2016 data: 0.1077 max mem: 3532 +Test: [ 220/2387] eta: 0:08:49 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2080 (0.6380) acc1: 95.0521 (62.6803) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2099 data: 0.1137 max mem: 3532 +Test: [ 230/2387] eta: 0:08:42 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2110 (0.6194) acc1: 94.7917 (64.0839) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1982 data: 0.1011 max mem: 3532 +Test: [ 240/2387] eta: 0:08:36 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2153 (0.6036) acc1: 94.5312 (65.3268) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2039 data: 0.1081 max mem: 3532 +Test: [ 250/2387] eta: 0:08:28 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2037 (0.5838) acc1: 94.5312 (66.6687) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1914 data: 0.0957 max mem: 3532 +Test: [ 260/2387] eta: 0:08:24 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0873 (0.5651) acc1: 99.7396 (67.9278) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1952 data: 0.1029 max mem: 3532 +Test: [ 270/2387] eta: 0:08:19 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0840 (0.5474) acc1: 99.7396 (69.0969) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2074 data: 0.1163 max mem: 3532 +Test: [ 280/2387] eta: 0:08:15 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0842 (0.5310) acc1: 99.7396 (70.1930) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2045 data: 0.1133 max mem: 3532 +Test: [ 290/2387] eta: 0:08:10 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0842 (0.5156) acc1: 100.0000 (71.2137) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2069 data: 0.1157 max mem: 3532 +Test: [ 300/2387] eta: 0:08:06 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0916 (0.5020) acc1: 100.0000 (72.1441) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2049 data: 0.1119 max mem: 3532 +Test: [ 310/2387] eta: 0:08:01 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1017 (0.4899) acc1: 99.7396 (73.0096) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2000 data: 0.1073 max mem: 3532 +Test: [ 320/2387] eta: 0:07:57 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1439 (0.4792) acc1: 99.7396 (73.8375) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2012 data: 0.1102 max mem: 3532 +Test: [ 330/2387] eta: 0:07:54 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1443 (0.4689) acc1: 99.7396 (74.6247) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2133 data: 0.1214 max mem: 3532 +Test: [ 340/2387] eta: 0:07:50 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1400 (0.4594) acc1: 100.0000 (75.3658) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2144 data: 0.1208 max mem: 3532 +Test: [ 350/2387] eta: 0:07:46 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1375 (0.4486) acc1: 100.0000 (76.0669) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2063 data: 0.1134 max mem: 3532 +Test: [ 360/2387] eta: 0:07:44 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0923 (0.4408) acc1: 99.7396 (76.6455) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2113 data: 0.1192 max mem: 3532 +Test: [ 370/2387] eta: 0:07:39 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1768 (0.4338) acc1: 96.6146 (77.1830) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2034 data: 0.1107 max mem: 3532 +Test: [ 380/2387] eta: 0:07:36 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1800 (0.4271) acc1: 96.6146 (77.6889) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1982 data: 0.1043 max mem: 3532 +Test: [ 390/2387] eta: 0:07:32 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1800 (0.4208) acc1: 96.6146 (78.1776) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1994 data: 0.1035 max mem: 3532 +Test: [ 400/2387] eta: 0:07:28 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1787 (0.4148) acc1: 96.6146 (78.6361) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1918 data: 0.0981 max mem: 3532 +Test: [ 410/2387] eta: 0:07:24 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1833 (0.4091) acc1: 96.3542 (79.0666) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1998 data: 0.1076 max mem: 3532 +Test: [ 420/2387] eta: 0:07:21 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1784 (0.4036) acc1: 96.6146 (79.4920) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1975 data: 0.1033 max mem: 3532 +Test: [ 430/2387] eta: 0:07:17 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1795 (0.3990) acc1: 96.6146 (79.8621) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1949 data: 0.1014 max mem: 3532 +Test: [ 440/2387] eta: 0:07:14 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1879 (0.3929) acc1: 96.0938 (80.2780) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1958 data: 0.1029 max mem: 3533 +Test: [ 450/2387] eta: 0:07:10 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0751 (0.3858) acc1: 100.0000 (80.7153) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1939 data: 0.0999 max mem: 3533 +Test: [ 460/2387] eta: 0:07:08 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0743 (0.3791) acc1: 100.0000 (81.1336) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2067 data: 0.1142 max mem: 3533 +Test: [ 470/2387] eta: 0:07:04 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0748 (0.3726) acc1: 100.0000 (81.5342) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2044 data: 0.1136 max mem: 3533 +Test: [ 480/2387] eta: 0:07:02 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0748 (0.3664) acc1: 100.0000 (81.9181) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2051 data: 0.1147 max mem: 3533 +Test: [ 490/2387] eta: 0:06:58 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0757 (0.3605) acc1: 100.0000 (82.2858) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1964 data: 0.1046 max mem: 3533 +Test: [ 500/2387] eta: 0:06:56 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0757 (0.3548) acc1: 100.0000 (82.6394) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1987 data: 0.1041 max mem: 3533 +Test: [ 510/2387] eta: 0:06:53 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0757 (0.3496) acc1: 100.0000 (82.9786) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2126 data: 0.1185 max mem: 3533 +Test: [ 520/2387] eta: 0:06:50 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1190 (0.3452) acc1: 100.0000 (83.3043) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1978 data: 0.1052 max mem: 3533 +Test: [ 530/2387] eta: 0:06:47 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1207 (0.3410) acc1: 100.0000 (83.6178) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1989 data: 0.1016 max mem: 3533 +Test: [ 540/2387] eta: 0:06:42 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1213 (0.3369) acc1: 100.0000 (83.9172) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1700 data: 0.0743 max mem: 3533 +Test: [ 550/2387] eta: 0:06:39 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1223 (0.3330) acc1: 100.0000 (84.2072) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1653 data: 0.0738 max mem: 3533 +Test: [ 560/2387] eta: 0:06:35 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1226 (0.3293) acc1: 99.7396 (84.4841) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1849 data: 0.0922 max mem: 3533 +Test: [ 570/2387] eta: 0:06:31 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1226 (0.3256) acc1: 99.7396 (84.7540) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1637 data: 0.0711 max mem: 3533 +Test: [ 580/2387] eta: 0:06:29 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1204 (0.3221) acc1: 100.0000 (85.0155) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1810 data: 0.0894 max mem: 3533 +Test: [ 590/2387] eta: 0:06:25 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1204 (0.3187) acc1: 100.0000 (85.2686) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1893 data: 0.0982 max mem: 3533 +Test: [ 600/2387] eta: 0:06:25 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1219 (0.3154) acc1: 100.0000 (85.5137) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2215 data: 0.1305 max mem: 3533 +Test: [ 610/2387] eta: 0:06:23 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1191 (0.3122) acc1: 100.0000 (85.7461) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2477 data: 0.1573 max mem: 3533 +Test: [ 620/2387] eta: 0:06:19 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1177 (0.3091) acc1: 99.7396 (85.9710) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1835 data: 0.0931 max mem: 3533 +Test: [ 630/2387] eta: 0:06:15 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1204 (0.3061) acc1: 99.4792 (86.1851) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1523 data: 0.0593 max mem: 3533 +Test: [ 640/2387] eta: 0:06:13 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1213 (0.3033) acc1: 99.4792 (86.3937) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1850 data: 0.0920 max mem: 3533 +Test: [ 650/2387] eta: 0:06:10 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1213 (0.3005) acc1: 99.7396 (86.5975) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1942 data: 0.1027 max mem: 3533 +Test: [ 660/2387] eta: 0:06:07 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1206 (0.2977) acc1: 99.7396 (86.7944) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1833 data: 0.0888 max mem: 3533 +Test: [ 670/2387] eta: 0:06:05 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1194 (0.2951) acc1: 99.7396 (86.9881) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1976 data: 0.1036 max mem: 3534 +Test: [ 680/2387] eta: 0:06:02 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1217 (0.2927) acc1: 100.0000 (87.1780) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2061 data: 0.1141 max mem: 3534 +Test: [ 690/2387] eta: 0:06:00 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1372 (0.2950) acc1: 99.7396 (87.0519) acc5: 100.0000 (100.0000) acc1_neg: 66.406 acc5_neg: 100.000 time: 0.1981 data: 0.1064 max mem: 3534 +Test: [ 700/2387] eta: 0:05:57 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.5506 (0.2987) acc1: 71.6146 (86.8161) acc5: 100.0000 (100.0000) acc1_neg: 72.656 acc5_neg: 100.000 time: 0.1904 data: 0.0990 max mem: 3534 +Test: [ 710/2387] eta: 0:05:55 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.5598 (0.3024) acc1: 70.8333 (86.5825) acc5: 100.0000 (100.0000) acc1_neg: 68.750 acc5_neg: 100.000 time: 0.1941 data: 0.1023 max mem: 3534 +Test: [ 720/2387] eta: 0:05:52 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.5553 (0.3059) acc1: 70.8333 (86.3800) acc5: 100.0000 (100.0000) acc1_neg: 72.396 acc5_neg: 100.000 time: 0.1931 data: 0.0997 max mem: 3534 +Test: [ 730/2387] eta: 0:05:49 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.5589 (0.3096) acc1: 70.5729 (86.1534) acc5: 100.0000 (100.0000) acc1_neg: 71.094 acc5_neg: 100.000 time: 0.1909 data: 0.0926 max mem: 3534 +Test: [ 740/2387] eta: 0:05:47 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.5595 (0.3130) acc1: 70.5729 (85.9473) acc5: 100.0000 (100.0000) acc1_neg: 70.052 acc5_neg: 100.000 time: 0.1956 data: 0.0984 max mem: 3534 +Test: [ 750/2387] eta: 0:05:46 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.5469 (0.3161) acc1: 71.6146 (85.7596) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.2288 data: 0.1356 max mem: 3534 +Test: [ 760/2387] eta: 0:05:44 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.5406 (0.3182) acc1: 71.8750 (85.6374) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.2303 data: 0.1357 max mem: 3534 +Test: [ 770/2387] eta: 0:05:41 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.3231 (0.3174) acc1: 85.4167 (85.6953) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1917 data: 0.0983 max mem: 3534 +Test: [ 780/2387] eta: 0:05:38 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1909 (0.3154) acc1: 95.0521 (85.8465) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1844 data: 0.0932 max mem: 3534 +Test: [ 790/2387] eta: 0:05:36 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1674 (0.3139) acc1: 96.6146 (85.9480) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1982 data: 0.1065 max mem: 3534 +Test: [ 800/2387] eta: 0:05:34 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1855 (0.3123) acc1: 96.0938 (86.0819) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2112 data: 0.1192 max mem: 3534 +Test: [ 810/2387] eta: 0:05:31 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1439 (0.3102) acc1: 97.6562 (86.2287) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1903 data: 0.0993 max mem: 3534 +Test: [ 820/2387] eta: 0:05:28 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1514 (0.3088) acc1: 97.9167 (86.3492) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1706 data: 0.0799 max mem: 3534 +Test: [ 830/2387] eta: 0:05:26 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2449 (0.3090) acc1: 91.6667 (86.3465) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.1732 data: 0.0822 max mem: 3534 +Test: [ 840/2387] eta: 0:05:23 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2132 (0.3072) acc1: 93.7500 (86.4843) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1691 data: 0.0757 max mem: 3534 +Test: [ 850/2387] eta: 0:05:20 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1647 (0.3057) acc1: 98.6979 (86.6184) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1762 data: 0.0817 max mem: 3534 +Test: [ 860/2387] eta: 0:05:17 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1640 (0.3039) acc1: 98.4375 (86.7535) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1781 data: 0.0840 max mem: 3534 +Test: [ 870/2387] eta: 0:05:15 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1458 (0.3022) acc1: 98.6979 (86.8847) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1703 data: 0.0751 max mem: 3534 +Test: [ 880/2387] eta: 0:05:12 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1360 (0.3004) acc1: 99.4792 (87.0217) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1671 data: 0.0740 max mem: 3534 +Test: [ 890/2387] eta: 0:05:09 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1449 (0.2989) acc1: 99.4792 (87.1501) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1660 data: 0.0726 max mem: 3534 +Test: [ 900/2387] eta: 0:05:06 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1743 (0.3025) acc1: 96.0938 (86.8838) acc5: 100.0000 (100.0000) acc1_neg: 59.896 acc5_neg: 100.000 time: 0.1685 data: 0.0733 max mem: 3535 +Test: [ 910/2387] eta: 0:05:04 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.7294 (0.3072) acc1: 54.6875 (86.5238) acc5: 100.0000 (100.0000) acc1_neg: 54.688 acc5_neg: 100.000 time: 0.1609 data: 0.0672 max mem: 3535 +Test: [ 920/2387] eta: 0:05:01 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.7337 (0.3119) acc1: 54.6875 (86.1795) acc5: 100.0000 (100.0000) acc1_neg: 53.125 acc5_neg: 100.000 time: 0.1589 data: 0.0640 max mem: 3535 +Test: [ 930/2387] eta: 0:04:58 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.7378 (0.3165) acc1: 54.9479 (85.8416) acc5: 100.0000 (100.0000) acc1_neg: 50.781 acc5_neg: 100.000 time: 0.1672 data: 0.0698 max mem: 3535 +Test: [ 940/2387] eta: 0:04:55 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.7430 (0.3210) acc1: 54.4271 (85.5088) acc5: 100.0000 (100.0000) acc1_neg: 50.521 acc5_neg: 100.000 time: 0.1649 data: 0.0677 max mem: 3535 +Test: [ 950/2387] eta: 0:04:53 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.7392 (0.3253) acc1: 54.4271 (85.1845) acc5: 100.0000 (100.0000) acc1_neg: 55.990 acc5_neg: 100.000 time: 0.1603 data: 0.0661 max mem: 3535 +Test: [ 960/2387] eta: 0:04:50 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.7223 (0.3294) acc1: 54.9479 (84.8720) acc5: 100.0000 (100.0000) acc1_neg: 54.948 acc5_neg: 100.000 time: 0.1649 data: 0.0738 max mem: 3535 +Test: [ 970/2387] eta: 0:04:48 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.7269 (0.3336) acc1: 54.9479 (84.5601) acc5: 100.0000 (100.0000) acc1_neg: 57.031 acc5_neg: 100.000 time: 0.1659 data: 0.0749 max mem: 3535 +Test: [ 980/2387] eta: 0:04:45 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.7102 (0.3343) acc1: 55.9896 (84.5093) acc5: 100.0000 (100.0000) acc1_neg: 72.656 acc5_neg: 100.000 time: 0.1805 data: 0.0889 max mem: 3535 +Test: [ 990/2387] eta: 0:04:43 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1654 (0.3331) acc1: 95.8333 (84.5915) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2036 data: 0.1100 max mem: 3535 +Test: [1000/2387] eta: 0:04:41 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1290 (0.3313) acc1: 98.4375 (84.7252) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2024 data: 0.1063 max mem: 3535 +Test: [1010/2387] eta: 0:04:39 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1342 (0.3299) acc1: 98.4375 (84.8304) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.2052 data: 0.1117 max mem: 3535 +Test: [1020/2387] eta: 0:04:38 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1726 (0.3285) acc1: 96.3542 (84.9328) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2119 data: 0.1208 max mem: 3535 +Test: [1030/2387] eta: 0:04:35 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1336 (0.3266) acc1: 98.4375 (85.0686) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2006 data: 0.1080 max mem: 3535 +Test: [1040/2387] eta: 0:04:33 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1193 (0.3247) acc1: 99.4792 (85.1990) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2015 data: 0.1087 max mem: 3535 +Test: [1050/2387] eta: 0:04:31 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1161 (0.3227) acc1: 99.4792 (85.3359) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1837 data: 0.0893 max mem: 3535 +Test: [1060/2387] eta: 0:04:28 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1132 (0.3208) acc1: 99.7396 (85.4694) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1632 data: 0.0675 max mem: 3535 +Test: [1070/2387] eta: 0:04:26 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1250 (0.3194) acc1: 99.2188 (85.5679) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.1728 data: 0.0800 max mem: 3535 +Test: [1080/2387] eta: 0:04:23 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2636 (0.3192) acc1: 91.1458 (85.6012) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1704 data: 0.0793 max mem: 3535 +Test: [1090/2387] eta: 0:04:21 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2289 (0.3179) acc1: 91.9271 (85.6940) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1672 data: 0.0730 max mem: 3535 +Test: [1100/2387] eta: 0:04:19 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1721 (0.3168) acc1: 96.0938 (85.7738) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1697 data: 0.0755 max mem: 3535 +Test: [1110/2387] eta: 0:04:16 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2044 (0.3160) acc1: 93.7500 (85.8386) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1678 data: 0.0751 max mem: 3535 +Test: [1120/2387] eta: 0:04:14 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2122 (0.3150) acc1: 93.7500 (85.9161) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1667 data: 0.0720 max mem: 3536 +Test: [1130/2387] eta: 0:04:11 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1892 (0.3138) acc1: 94.2708 (86.0054) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1644 data: 0.0710 max mem: 3536 +Test: [1140/2387] eta: 0:04:09 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2170 (0.3135) acc1: 94.2708 (86.0448) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1658 data: 0.0740 max mem: 3536 +Test: [1150/2387] eta: 0:04:07 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2371 (0.3125) acc1: 93.2292 (86.1174) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1702 data: 0.0791 max mem: 3536 +Test: [1160/2387] eta: 0:04:04 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1459 (0.3110) acc1: 97.6562 (86.2203) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1690 data: 0.0785 max mem: 3536 +Test: [1170/2387] eta: 0:04:02 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1322 (0.3095) acc1: 98.4375 (86.3256) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1645 data: 0.0741 max mem: 3536 +Test: [1180/2387] eta: 0:04:00 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1355 (0.3081) acc1: 98.1771 (86.4217) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1591 data: 0.0673 max mem: 3536 +Test: [1190/2387] eta: 0:03:57 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1258 (0.3066) acc1: 98.4375 (86.5231) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1591 data: 0.0644 max mem: 3536 +Test: [1200/2387] eta: 0:03:55 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1116 (0.3050) acc1: 99.7396 (86.6277) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1606 data: 0.0673 max mem: 3536 +Test: [1210/2387] eta: 0:03:53 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1082 (0.3033) acc1: 99.7396 (86.7338) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1694 data: 0.0792 max mem: 3536 +Test: [1220/2387] eta: 0:03:51 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0925 (0.3016) acc1: 99.7396 (86.8333) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1777 data: 0.0861 max mem: 3536 +Test: [1230/2387] eta: 0:03:48 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0825 (0.2998) acc1: 99.7396 (86.9366) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1684 data: 0.0748 max mem: 3536 +Test: [1240/2387] eta: 0:03:46 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0767 (0.2981) acc1: 100.0000 (87.0383) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1834 data: 0.0877 max mem: 3536 +Test: [1250/2387] eta: 0:03:44 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0791 (0.2963) acc1: 99.7396 (87.1392) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1955 data: 0.1011 max mem: 3536 +Test: [1260/2387] eta: 0:03:42 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0807 (0.2946) acc1: 99.7396 (87.2379) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1864 data: 0.0913 max mem: 3536 +Test: [1270/2387] eta: 0:03:40 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0944 (0.2932) acc1: 99.4792 (87.3281) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1895 data: 0.0947 max mem: 3536 +Test: [1280/2387] eta: 0:03:38 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1054 (0.2917) acc1: 99.2188 (87.4213) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1813 data: 0.0904 max mem: 3536 +Test: [1290/2387] eta: 0:03:36 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1054 (0.2904) acc1: 99.2188 (87.5061) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1754 data: 0.0828 max mem: 3536 +Test: [1300/2387] eta: 0:03:34 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1190 (0.2891) acc1: 98.4375 (87.5875) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1907 data: 0.0982 max mem: 3536 +Test: [1310/2387] eta: 0:03:32 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1194 (0.2879) acc1: 98.4375 (87.6700) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1956 data: 0.1028 max mem: 3536 +Test: [1320/2387] eta: 0:03:30 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1194 (0.2866) acc1: 98.9583 (87.7551) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1933 data: 0.0998 max mem: 3536 +Test: [1330/2387] eta: 0:03:28 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1403 (0.2860) acc1: 98.9583 (87.8087) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1947 data: 0.1021 max mem: 3536 +Test: [1340/2387] eta: 0:03:26 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2151 (0.2857) acc1: 94.5312 (87.8362) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2060 data: 0.1143 max mem: 3536 +Test: [1350/2387] eta: 0:03:24 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0647 (0.2840) acc1: 100.0000 (87.9262) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2060 data: 0.1147 max mem: 3537 +Test: [1360/2387] eta: 0:03:22 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0530 (0.2823) acc1: 100.0000 (88.0149) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1909 data: 0.0987 max mem: 3537 +Test: [1370/2387] eta: 0:03:20 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0511 (0.2807) acc1: 100.0000 (88.1021) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1855 data: 0.0934 max mem: 3537 +Test: [1380/2387] eta: 0:03:18 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0512 (0.2790) acc1: 100.0000 (88.1883) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1728 data: 0.0814 max mem: 3537 +Test: [1390/2387] eta: 0:03:16 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0521 (0.2774) acc1: 100.0000 (88.2732) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1759 data: 0.0827 max mem: 3537 +Test: [1400/2387] eta: 0:03:14 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0575 (0.2758) acc1: 100.0000 (88.3563) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1860 data: 0.0910 max mem: 3537 +Test: [1410/2387] eta: 0:03:12 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0619 (0.2751) acc1: 100.0000 (88.4080) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1820 data: 0.0888 max mem: 3537 +Test: [1420/2387] eta: 0:03:09 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2283 (0.2750) acc1: 91.9271 (88.4138) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1744 data: 0.0818 max mem: 3537 +Test: [1430/2387] eta: 0:03:08 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2539 (0.2748) acc1: 90.8854 (88.4396) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1914 data: 0.0989 max mem: 3537 +Test: [1440/2387] eta: 0:03:05 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2459 (0.2746) acc1: 92.1875 (88.4565) acc5: 100.0000 (100.0000) acc1_neg: 77.344 acc5_neg: 100.000 time: 0.1896 data: 0.0983 max mem: 3537 +Test: [1450/2387] eta: 0:03:03 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2570 (0.2747) acc1: 92.4479 (88.4677) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1778 data: 0.0862 max mem: 3537 +Test: [1460/2387] eta: 0:03:01 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1728 (0.2738) acc1: 94.0104 (88.5279) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1797 data: 0.0869 max mem: 3537 +Test: [1470/2387] eta: 0:02:59 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1481 (0.2729) acc1: 97.6562 (88.5861) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1761 data: 0.0826 max mem: 3537 +Test: [1480/2387] eta: 0:02:57 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1440 (0.2721) acc1: 97.3958 (88.6461) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1891 data: 0.0948 max mem: 3537 +Test: [1490/2387] eta: 0:02:55 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1414 (0.2712) acc1: 97.9167 (88.7085) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1988 data: 0.1052 max mem: 3537 +Test: [1500/2387] eta: 0:02:53 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1328 (0.2703) acc1: 97.9167 (88.7712) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1938 data: 0.0996 max mem: 3537 +Test: [1510/2387] eta: 0:02:51 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1356 (0.2695) acc1: 98.4375 (88.8388) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1948 data: 0.1013 max mem: 3537 +Test: [1520/2387] eta: 0:02:49 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1445 (0.2687) acc1: 99.2188 (88.8966) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1961 data: 0.1032 max mem: 3537 +Test: [1530/2387] eta: 0:02:47 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1546 (0.2680) acc1: 98.6979 (88.9577) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1956 data: 0.1009 max mem: 3537 +Test: [1540/2387] eta: 0:02:45 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1555 (0.2673) acc1: 98.4375 (89.0138) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1994 data: 0.1070 max mem: 3537 +Test: [1550/2387] eta: 0:02:43 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1460 (0.2665) acc1: 98.4375 (89.0732) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1932 data: 0.1024 max mem: 3537 +Test: [1560/2387] eta: 0:02:42 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1265 (0.2656) acc1: 99.2188 (89.1367) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1918 data: 0.1002 max mem: 3537 +Test: [1570/2387] eta: 0:02:40 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1342 (0.2649) acc1: 99.2188 (89.1983) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2015 data: 0.1088 max mem: 3537 +Test: [1580/2387] eta: 0:02:38 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1413 (0.2642) acc1: 99.4792 (89.2527) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1999 data: 0.1039 max mem: 3538 +Test: [1590/2387] eta: 0:02:36 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1456 (0.2635) acc1: 98.4375 (89.3033) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2030 data: 0.1055 max mem: 3538 +Test: [1600/2387] eta: 0:02:34 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1566 (0.2630) acc1: 97.3958 (89.3463) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2085 data: 0.1137 max mem: 3538 +Test: [1610/2387] eta: 0:02:32 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1502 (0.2623) acc1: 97.3958 (89.3981) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2048 data: 0.1101 max mem: 3538 +Test: [1620/2387] eta: 0:02:30 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1452 (0.2616) acc1: 97.6562 (89.4434) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1988 data: 0.1041 max mem: 3538 +Test: [1630/2387] eta: 0:02:28 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1686 (0.2615) acc1: 94.7917 (89.4613) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.1963 data: 0.1023 max mem: 3538 +Test: [1640/2387] eta: 0:02:26 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2591 (0.2618) acc1: 90.3646 (89.4530) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1864 data: 0.0942 max mem: 3538 +Test: [1650/2387] eta: 0:02:24 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2903 (0.2619) acc1: 87.7604 (89.4433) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1811 data: 0.0903 max mem: 3538 +Test: [1660/2387] eta: 0:02:22 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2829 (0.2621) acc1: 87.7604 (89.4391) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.1910 data: 0.0981 max mem: 3538 +Test: [1670/2387] eta: 0:02:20 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2944 (0.2624) acc1: 88.8021 (89.4345) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1773 data: 0.0851 max mem: 3538 +Test: [1680/2387] eta: 0:02:18 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2944 (0.2626) acc1: 89.0625 (89.4312) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1658 data: 0.0752 max mem: 3538 +Test: [1690/2387] eta: 0:02:16 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2794 (0.2628) acc1: 90.3646 (89.4275) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1526 data: 0.0613 max mem: 3538 +Test: [1700/2387] eta: 0:02:14 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2632 (0.2626) acc1: 90.8854 (89.4403) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1627 data: 0.0707 max mem: 3538 +Test: [1710/2387] eta: 0:02:12 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1152 (0.2617) acc1: 97.3958 (89.4931) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1971 data: 0.1047 max mem: 3538 +Test: [1720/2387] eta: 0:02:10 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0957 (0.2608) acc1: 98.9583 (89.5463) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1885 data: 0.0937 max mem: 3538 +Test: [1730/2387] eta: 0:02:08 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0931 (0.2598) acc1: 99.2188 (89.5997) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1909 data: 0.0970 max mem: 3538 +Test: [1740/2387] eta: 0:02:06 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0925 (0.2589) acc1: 99.4792 (89.6565) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1902 data: 0.0992 max mem: 3538 +Test: [1750/2387] eta: 0:02:04 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1001 (0.2580) acc1: 99.2188 (89.7057) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1689 data: 0.0778 max mem: 3538 +Test: [1760/2387] eta: 0:02:02 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1152 (0.2573) acc1: 98.1771 (89.7482) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1696 data: 0.0765 max mem: 3538 +Test: [1770/2387] eta: 0:02:00 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1248 (0.2565) acc1: 97.6562 (89.7942) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1749 data: 0.0822 max mem: 3538 +Test: [1780/2387] eta: 0:01:58 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1189 (0.2558) acc1: 98.9583 (89.8388) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1822 data: 0.0899 max mem: 3538 +Test: [1790/2387] eta: 0:01:56 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1027 (0.2550) acc1: 98.6979 (89.8852) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1867 data: 0.0941 max mem: 3538 +Test: [1800/2387] eta: 0:01:54 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1031 (0.2542) acc1: 98.6979 (89.9338) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1843 data: 0.0936 max mem: 3538 +Test: [1810/2387] eta: 0:01:52 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1011 (0.2533) acc1: 99.2188 (89.9858) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1934 data: 0.1021 max mem: 3539 +Test: [1820/2387] eta: 0:01:50 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1024 (0.2526) acc1: 98.4375 (90.0275) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1855 data: 0.0937 max mem: 3539 +Test: [1830/2387] eta: 0:01:48 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1254 (0.2520) acc1: 97.9167 (90.0726) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1846 data: 0.0930 max mem: 3539 +Test: [1840/2387] eta: 0:01:46 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1276 (0.2513) acc1: 99.7396 (90.1252) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1930 data: 0.1005 max mem: 3539 +Test: [1850/2387] eta: 0:01:44 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1268 (0.2506) acc1: 99.7396 (90.1775) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1826 data: 0.0889 max mem: 3539 +Test: [1860/2387] eta: 0:01:42 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1293 (0.2500) acc1: 99.7396 (90.2286) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1864 data: 0.0907 max mem: 3539 +Test: [1870/2387] eta: 0:01:40 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1282 (0.2493) acc1: 99.7396 (90.2794) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2010 data: 0.1056 max mem: 3539 +Test: [1880/2387] eta: 0:01:38 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1260 (0.2486) acc1: 99.7396 (90.3300) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1871 data: 0.0915 max mem: 3539 +Test: [1890/2387] eta: 0:01:36 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1268 (0.2480) acc1: 99.7396 (90.3801) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1774 data: 0.0805 max mem: 3539 +Test: [1900/2387] eta: 0:01:34 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1274 (0.2474) acc1: 99.7396 (90.4297) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2174 data: 0.1240 max mem: 3539 +Test: [1910/2387] eta: 0:01:32 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1551 (0.2470) acc1: 99.7396 (90.4714) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.2182 data: 0.1247 max mem: 3539 +Test: [1920/2387] eta: 0:01:30 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1683 (0.2467) acc1: 99.2188 (90.4974) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1821 data: 0.0859 max mem: 3539 +Test: [1930/2387] eta: 0:01:28 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1826 (0.2464) acc1: 94.7917 (90.5212) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1660 data: 0.0713 max mem: 3539 +Test: [1940/2387] eta: 0:01:26 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2079 (0.2463) acc1: 93.4896 (90.5157) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2168 data: 0.1248 max mem: 3539 +Test: [1950/2387] eta: 0:01:24 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2186 (0.2462) acc1: 92.1875 (90.5298) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2044 data: 0.1125 max mem: 3539 +Test: [1960/2387] eta: 0:01:22 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2115 (0.2460) acc1: 95.3125 (90.5524) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.1455 data: 0.0539 max mem: 3539 +Test: [1970/2387] eta: 0:01:20 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2524 (0.2469) acc1: 90.8854 (90.4971) acc5: 100.0000 (100.0000) acc1_neg: 75.521 acc5_neg: 100.000 time: 0.1814 data: 0.0902 max mem: 3539 +Test: [1980/2387] eta: 0:01:18 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2853 (0.2468) acc1: 92.1875 (90.5218) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1825 data: 0.0893 max mem: 3539 +Test: [1990/2387] eta: 0:01:16 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1357 (0.2462) acc1: 98.9583 (90.5643) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1673 data: 0.0739 max mem: 3539 +Test: [2000/2387] eta: 0:01:15 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1360 (0.2457) acc1: 98.9583 (90.6073) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1999 data: 0.1077 max mem: 3539 +Test: [2010/2387] eta: 0:01:13 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1374 (0.2452) acc1: 99.2188 (90.6495) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2246 data: 0.1320 max mem: 3539 +Test: [2020/2387] eta: 0:01:11 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1382 (0.2446) acc1: 99.2188 (90.6905) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1948 data: 0.0990 max mem: 3539 +Test: [2030/2387] eta: 0:01:09 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1387 (0.2441) acc1: 99.2188 (90.7323) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1849 data: 0.0883 max mem: 3540 +Test: [2040/2387] eta: 0:01:07 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1376 (0.2436) acc1: 99.2188 (90.7747) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2049 data: 0.1117 max mem: 3540 +Test: [2050/2387] eta: 0:01:05 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1376 (0.2431) acc1: 99.4792 (90.8167) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1854 data: 0.0925 max mem: 3540 +Test: [2060/2387] eta: 0:01:03 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1501 (0.2427) acc1: 99.2188 (90.8566) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1749 data: 0.0814 max mem: 3540 +Test: [2070/2387] eta: 0:01:01 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1534 (0.2422) acc1: 99.2188 (90.8981) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1743 data: 0.0823 max mem: 3540 +Test: [2080/2387] eta: 0:00:59 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1533 (0.2418) acc1: 99.2188 (90.9361) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1727 data: 0.0810 max mem: 3540 +Test: [2090/2387] eta: 0:00:57 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1527 (0.2414) acc1: 98.9583 (90.9750) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1775 data: 0.0855 max mem: 3540 +Test: [2100/2387] eta: 0:00:55 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1527 (0.2409) acc1: 99.2188 (91.0140) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1751 data: 0.0840 max mem: 3540 +Test: [2110/2387] eta: 0:00:53 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1535 (0.2405) acc1: 99.2188 (91.0531) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1700 data: 0.0792 max mem: 3540 +Test: [2120/2387] eta: 0:00:51 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1523 (0.2402) acc1: 99.2188 (91.0809) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1771 data: 0.0838 max mem: 3540 +Test: [2130/2387] eta: 0:00:49 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1635 (0.2398) acc1: 97.9167 (91.1135) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1844 data: 0.0910 max mem: 3540 +Test: [2140/2387] eta: 0:00:47 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1639 (0.2396) acc1: 97.9167 (91.1295) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1846 data: 0.0923 max mem: 3540 +Test: [2150/2387] eta: 0:00:45 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1713 (0.2393) acc1: 97.3958 (91.1600) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1784 data: 0.0864 max mem: 3540 +Test: [2160/2387] eta: 0:00:43 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1498 (0.2389) acc1: 97.6562 (91.1903) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1767 data: 0.0834 max mem: 3540 +Test: [2170/2387] eta: 0:00:41 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1835 (0.2391) acc1: 95.5729 (91.1804) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1942 data: 0.1007 max mem: 3540 +Test: [2180/2387] eta: 0:00:39 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.3107 (0.2399) acc1: 87.2396 (91.1353) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.1932 data: 0.1010 max mem: 3540 +Test: [2190/2387] eta: 0:00:38 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2921 (0.2400) acc1: 87.7604 (91.1374) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1856 data: 0.0932 max mem: 3540 +Test: [2200/2387] eta: 0:00:36 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1306 (0.2394) acc1: 96.3542 (91.1713) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1864 data: 0.0935 max mem: 3540 +Test: [2210/2387] eta: 0:00:34 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1221 (0.2389) acc1: 99.4792 (91.2064) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1809 data: 0.0871 max mem: 3540 +Test: [2220/2387] eta: 0:00:32 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1283 (0.2385) acc1: 98.9583 (91.2402) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1943 data: 0.0997 max mem: 3540 +Test: [2230/2387] eta: 0:00:30 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1277 (0.2380) acc1: 99.2188 (91.2760) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1925 data: 0.0983 max mem: 3540 +Test: [2240/2387] eta: 0:00:28 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1132 (0.2375) acc1: 99.4792 (91.3083) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1924 data: 0.0996 max mem: 3540 +Test: [2250/2387] eta: 0:00:26 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1330 (0.2371) acc1: 99.4792 (91.3389) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2043 data: 0.1125 max mem: 3540 +Test: [2260/2387] eta: 0:00:24 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1414 (0.2367) acc1: 99.7396 (91.3702) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1955 data: 0.1042 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1305 (0.2362) acc1: 99.7396 (91.4064) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1923 data: 0.1015 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1458 (0.2360) acc1: 99.2188 (91.4222) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1893 data: 0.0971 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1574 (0.2356) acc1: 97.3958 (91.4517) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1907 data: 0.0976 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1356 (0.2352) acc1: 98.4375 (91.4822) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2049 data: 0.1094 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1603 (0.2351) acc1: 97.3958 (91.4955) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2025 data: 0.1042 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2140 (0.2351) acc1: 94.7917 (91.5023) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2012 data: 0.1058 max mem: 3541 +Test: [2330/2387] eta: 0:00:11 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2271 (0.2350) acc1: 94.2708 (91.5170) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1770 data: 0.0854 max mem: 3541 +Test: [2340/2387] eta: 0:00:09 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2359 (0.2352) acc1: 93.4896 (91.5138) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1472 data: 0.0563 max mem: 3541 +Test: [2350/2387] eta: 0:00:07 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2694 (0.2355) acc1: 89.8438 (91.4981) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1471 data: 0.0559 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2463 (0.2356) acc1: 91.9271 (91.4975) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1313 data: 0.0410 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2348 (0.2357) acc1: 92.9688 (91.4963) acc5: 100.0000 (100.0000) acc1_neg: 67.969 acc5_neg: 100.000 time: 0.1180 data: 0.0277 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2850 (0.2362) acc1: 89.5833 (91.4705) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1156 data: 0.0253 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.3444 (0.2366) acc1: 86.0465 (91.4563) acc5: 100.0000 (100.0000) acc1_neg: 86.047 acc5_neg: 100.000 time: 0.1092 data: 0.0218 max mem: 3541 +Test: Total time: 0:07:37 (0.1917 s / it) +* Overall Acc@1 91.456 Acc@5 100.000 + target == real Acc@1 49.537 Acc@5 100.000 + target == fake Acc@1 94.400 Acc@5 100.000 + loss 0.237 + PR-AUC 0.9936 ROC-AUC 0.9158 EER 15.47% (thr=0.7476) +Accuracy of the network on the 916267 test images: 91.5% +[rank0]:[W725 17:51:11.851496621 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 17:51:31.099000 598972 torch/distributed/run.py:766] +W0725 17:51:31.099000 598972 torch/distributed/run.py:766] ***************************************** +W0725 17:51:31.099000 598972 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:51:31.099000 598972 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:52:00.025762459 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:52:00.395916855 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 17:52:01.522009369 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:52:01.735131583 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:52:01.762380696 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +[rank5]:[W725 17:52:01.855599251 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +[rank3]:[W725 17:52:01.874615249 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:52:01.879927671 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed3_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed3_resampled_n116-041-153_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:05:01 acc1_pos: 61.198 acc5_pos: 100.000 loss: 0.7257 (0.7257) acc1: 61.1979 (61.1979) acc5: 100.0000 (100.0000) time: 7.5463 data: 5.6689 max mem: 3531 +Test: [ 10/517] eta: 0:06:31 acc1_pos: 48.958 acc5_pos: 100.000 loss: 0.9768 (1.0116) acc1: 35.1562 (34.5407) acc5: 100.0000 (100.0000) time: 0.7714 data: 0.5158 max mem: 3531 +Test: [ 20/517] eta: 0:03:44 acc1_pos: 44.531 acc5_pos: 100.000 loss: 0.9211 (0.9203) acc1: 37.7604 (40.6126) acc5: 100.0000 (100.0000) time: 0.0969 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:03:02 acc1_pos: 4.167 acc5_pos: 100.000 loss: 0.9211 (0.9803) acc1: 33.3333 (37.1052) acc5: 100.0000 (100.0000) time: 0.1553 data: 0.0589 max mem: 3531 +Test: [ 40/517] eta: 0:02:47 acc1_pos: 48.698 acc5_pos: 100.000 loss: 1.0589 (0.9977) acc1: 22.3958 (35.4865) acc5: 100.0000 (100.0000) time: 0.2458 data: 0.1538 max mem: 3531 +Test: [ 50/517] eta: 0:02:34 acc1_pos: 92.202 acc5_pos: 100.000 loss: 0.9441 (0.9663) acc1: 32.8125 (37.5817) acc5: 100.0000 (100.0000) acc1_neg: 55.729 acc5_neg: 100.000 time: 0.2646 data: 0.1729 max mem: 3531 +Test: [ 60/517] eta: 0:02:28 acc1_pos: 92.202 acc5_pos: 100.000 loss: 0.6878 (0.9025) acc1: 55.4688 (42.3412) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2741 data: 0.1794 max mem: 3531 +Test: [ 70/517] eta: 0:02:20 acc1_pos: 92.202 acc5_pos: 100.000 loss: 0.2727 (0.8089) acc1: 95.5729 (50.1687) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2699 data: 0.1740 max mem: 3531 +Test: [ 80/517] eta: 0:02:14 acc1_pos: 92.202 acc5_pos: 100.000 loss: 0.2382 (0.7416) acc1: 98.6979 (56.1085) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2498 data: 0.1529 max mem: 3531 +Test: [ 90/517] eta: 0:02:08 acc1_pos: 92.202 acc5_pos: 100.000 loss: 0.2858 (0.6985) acc1: 94.5312 (59.8214) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.2549 data: 0.1591 max mem: 3531 +Test: [100/517] eta: 0:02:03 acc1_pos: 92.202 acc5_pos: 100.000 loss: 0.3394 (0.6649) acc1: 90.1042 (62.7785) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.2549 data: 0.1603 max mem: 3531 +Test: [110/517] eta: 0:01:59 acc1_pos: 92.202 acc5_pos: 100.000 loss: 0.3575 (0.6370) acc1: 90.1042 (65.1558) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.2549 data: 0.1594 max mem: 3531 +Test: [120/517] eta: 0:01:55 acc1_pos: 92.202 acc5_pos: 100.000 loss: 0.3583 (0.6145) acc1: 89.5833 (67.0820) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.2649 data: 0.1697 max mem: 3531 +Test: [130/517] eta: 0:01:51 acc1_pos: 92.202 acc5_pos: 100.000 loss: 0.3519 (0.5936) acc1: 90.1042 (68.9070) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.2599 data: 0.1639 max mem: 3531 +Test: [140/517] eta: 0:01:47 acc1_pos: 92.202 acc5_pos: 100.000 loss: 0.3859 (0.5808) acc1: 84.6354 (69.8619) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.2399 data: 0.1457 max mem: 3531 +Test: [150/517] eta: 0:01:43 acc1_pos: 92.202 acc5_pos: 100.000 loss: 0.4017 (0.5672) acc1: 83.5938 (70.9644) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.2549 data: 0.1630 max mem: 3531 +Test: [160/517] eta: 0:01:40 acc1_pos: 92.202 acc5_pos: 100.000 loss: 0.3729 (0.5549) acc1: 88.2812 (72.1807) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.2549 data: 0.1635 max mem: 3531 +Test: [170/517] eta: 0:01:37 acc1_pos: 92.202 acc5_pos: 100.000 loss: 0.3539 (0.5430) acc1: 90.8854 (73.2228) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.2749 data: 0.1795 max mem: 3531 +Test: [180/517] eta: 0:01:33 acc1_pos: 92.202 acc5_pos: 100.000 loss: 0.3889 (0.5375) acc1: 87.5000 (73.6965) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.2649 data: 0.1667 max mem: 3531 +Test: [190/517] eta: 0:01:30 acc1_pos: 92.202 acc5_pos: 100.000 loss: 0.3841 (0.5266) acc1: 89.8438 (74.8078) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.2399 data: 0.1454 max mem: 3531 +Test: [200/517] eta: 0:01:27 acc1_pos: 92.202 acc5_pos: 100.000 loss: 0.3600 (0.5203) acc1: 91.6667 (75.3537) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.2449 data: 0.1523 max mem: 3531 +Test: [210/517] eta: 0:01:24 acc1_pos: 92.202 acc5_pos: 100.000 loss: 0.4162 (0.5165) acc1: 82.5521 (75.6270) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.2449 data: 0.1518 max mem: 3532 +Test: [220/517] eta: 0:01:21 acc1_pos: 92.202 acc5_pos: 100.000 loss: 0.4121 (0.5112) acc1: 83.3333 (76.0900) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.2599 data: 0.1657 max mem: 3532 +Test: [230/517] eta: 0:01:18 acc1_pos: 92.202 acc5_pos: 100.000 loss: 0.4184 (0.5104) acc1: 83.5938 (76.2367) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.2549 data: 0.1609 max mem: 3532 +Test: [240/517] eta: 0:01:15 acc1_pos: 92.202 acc5_pos: 100.000 loss: 0.4286 (0.5051) acc1: 83.5938 (76.6922) acc5: 100.0000 (100.0000) acc1_neg: 75.521 acc5_neg: 100.000 time: 0.2499 data: 0.1550 max mem: 3532 +Test: [250/517] eta: 0:01:12 acc1_pos: 92.202 acc5_pos: 100.000 loss: 0.3632 (0.4995) acc1: 91.1458 (77.2691) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2499 data: 0.1541 max mem: 3532 +Test: [260/517] eta: 0:01:09 acc1_pos: 92.202 acc5_pos: 100.000 loss: 0.3504 (0.4942) acc1: 92.7083 (77.8167) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.2499 data: 0.1565 max mem: 3532 +Test: [270/517] eta: 0:01:06 acc1_pos: 92.202 acc5_pos: 100.000 loss: 0.3661 (0.4903) acc1: 87.7604 (78.1000) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.2599 data: 0.1648 max mem: 3532 +Test: [280/517] eta: 0:01:03 acc1_pos: 92.202 acc5_pos: 100.000 loss: 0.4018 (0.4879) acc1: 84.1146 (78.2510) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.2649 data: 0.1683 max mem: 3532 +Test: [290/517] eta: 0:01:01 acc1_pos: 92.202 acc5_pos: 100.000 loss: 0.4252 (0.4859) acc1: 82.8125 (78.4919) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.2598 data: 0.1627 max mem: 3532 +Test: [300/517] eta: 0:00:58 acc1_pos: 92.202 acc5_pos: 100.000 loss: 0.4093 (0.4840) acc1: 85.1562 (78.6796) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.2599 data: 0.1639 max mem: 3532 +Test: [310/517] eta: 0:00:55 acc1_pos: 92.202 acc5_pos: 100.000 loss: 0.4548 (0.4838) acc1: 80.4688 (78.6818) acc5: 100.0000 (100.0000) acc1_neg: 73.958 acc5_neg: 100.000 time: 0.2749 data: 0.1805 max mem: 3532 +Test: [320/517] eta: 0:00:53 acc1_pos: 92.202 acc5_pos: 100.000 loss: 0.4548 (0.4805) acc1: 80.4688 (78.9793) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2749 data: 0.1777 max mem: 3532 +Test: [330/517] eta: 0:00:50 acc1_pos: 92.202 acc5_pos: 100.000 loss: 0.3109 (0.4749) acc1: 94.0104 (79.4845) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2599 data: 0.1621 max mem: 3532 +Test: [340/517] eta: 0:00:47 acc1_pos: 92.202 acc5_pos: 100.000 loss: 0.2789 (0.4687) acc1: 96.0938 (79.9846) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2549 data: 0.1579 max mem: 3532 +Test: [350/517] eta: 0:00:44 acc1_pos: 92.202 acc5_pos: 100.000 loss: 0.2892 (0.4698) acc1: 93.7500 (79.8107) acc5: 100.0000 (100.0000) acc1_neg: 48.958 acc5_neg: 100.000 time: 0.2499 data: 0.1521 max mem: 3532 +Test: [360/517] eta: 0:00:41 acc1_pos: 92.202 acc5_pos: 100.000 loss: 0.4742 (0.4691) acc1: 76.0417 (79.7928) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2453 data: 0.1495 max mem: 3532 +Test: [370/517] eta: 0:00:39 acc1_pos: 92.202 acc5_pos: 100.000 loss: 0.2395 (0.4621) acc1: 96.0938 (80.2961) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2499 data: 0.1542 max mem: 3532 +Test: [380/517] eta: 0:00:36 acc1_pos: 92.202 acc5_pos: 100.000 loss: 0.1795 (0.4543) acc1: 99.2188 (80.7907) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2547 data: 0.1584 max mem: 3532 +Test: [390/517] eta: 0:00:33 acc1_pos: 92.202 acc5_pos: 100.000 loss: 0.1591 (0.4469) acc1: 99.7396 (81.2653) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2497 data: 0.1555 max mem: 3532 +Test: [400/517] eta: 0:00:31 acc1_pos: 92.202 acc5_pos: 100.000 loss: 0.1610 (0.4410) acc1: 99.4792 (81.6604) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2546 data: 0.1586 max mem: 3532 +Test: [410/517] eta: 0:00:28 acc1_pos: 92.202 acc5_pos: 100.000 loss: 0.2472 (0.4367) acc1: 96.0938 (81.9882) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.2899 data: 0.1943 max mem: 3532 +Test: [420/517] eta: 0:00:26 acc1_pos: 92.202 acc5_pos: 100.000 loss: 0.3116 (0.4343) acc1: 93.2292 (82.2620) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.3199 data: 0.2241 max mem: 3532 +Test: [430/517] eta: 0:00:23 acc1_pos: 92.202 acc5_pos: 100.000 loss: 0.3309 (0.4322) acc1: 90.6250 (82.4173) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.3549 data: 0.2591 max mem: 3532 +Test: [440/517] eta: 0:00:20 acc1_pos: 92.202 acc5_pos: 100.000 loss: 0.3309 (0.4302) acc1: 90.3646 (82.6377) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2949 data: 0.1988 max mem: 3533 +Test: [450/517] eta: 0:00:18 acc1_pos: 92.202 acc5_pos: 100.000 loss: 0.3142 (0.4276) acc1: 92.9688 (82.8431) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2449 data: 0.1470 max mem: 3533 +Test: [460/517] eta: 0:00:15 acc1_pos: 92.202 acc5_pos: 100.000 loss: 0.3310 (0.4261) acc1: 90.8854 (83.0051) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.2800 data: 0.1846 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 92.202 acc5_pos: 100.000 loss: 0.3447 (0.4241) acc1: 91.1458 (83.2106) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.2549 data: 0.1567 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 92.202 acc5_pos: 100.000 loss: 0.3360 (0.4227) acc1: 92.4479 (83.3804) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2498 data: 0.1488 max mem: 3533 +Test: [490/517] eta: 0:00:07 acc1_pos: 92.202 acc5_pos: 100.000 loss: 0.3173 (0.4201) acc1: 92.7083 (83.6160) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.2524 data: 0.1578 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 92.202 acc5_pos: 100.000 loss: 0.3305 (0.4193) acc1: 92.4479 (83.7138) acc5: 100.0000 (100.0000) acc1_neg: 77.083 acc5_neg: 100.000 time: 0.2198 data: 0.1302 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 92.202 acc5_pos: 100.000 loss: 0.3734 (0.4196) acc1: 87.5000 (83.7120) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1850 data: 0.0956 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 92.202 acc5_pos: 100.000 loss: 0.3734 (0.4189) acc1: 86.9792 (83.7678) acc5: 100.0000 (100.0000) acc1_neg: 77.739 acc5_neg: 100.000 time: 0.1798 data: 0.0888 max mem: 3533 +Test: Total time: 0:02:17 (0.2661 s / it) +* Overall Acc@1 83.768 Acc@5 100.000 + target == real Acc@1 36.402 Acc@5 100.000 + target == fake Acc@1 88.682 Acc@5 100.000 + loss 0.419 + PR-AUC 0.9545 ROC-AUC 0.7156 EER 34.31% (thr=0.6924) +Accuracy of the network on the 198427 test images: 83.8% +[rank0]:[W725 17:54:27.587919972 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0725 17:54:46.681000 599560 torch/distributed/run.py:766] +W0725 17:54:46.681000 599560 torch/distributed/run.py:766] ***************************************** +W0725 17:54:46.681000 599560 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:54:46.681000 599560 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:55:13.392589989 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W725 17:55:13.392621776 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:55:13.453698834 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W725 17:55:14.014704966 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank1]:[W725 17:55:14.015896839 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:55:14.021510704 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:55:14.209463520 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank4]:[W725 17:55:14.209517733 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed3_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed3_resampled_n116-041-153_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:36:56 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.9478 (1.9478) acc1: 0.0000 (0.0000) acc5: 100.0000 (100.0000) time: 5.6971 data: 4.2023 max mem: 3531 +Test: [ 10/389] eta: 0:03:47 acc1_pos: 66.927 acc5_pos: 100.000 loss: 1.4817 (1.4308) acc1: 10.6771 (15.1752) acc5: 100.0000 (100.0000) time: 0.6014 data: 0.3824 max mem: 3531 +Test: [ 20/389] eta: 0:02:15 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.3136 (1.3962) acc1: 13.0208 (22.7307) acc5: 100.0000 (100.0000) time: 0.1009 data: 0.0069 max mem: 3531 +Test: [ 30/389] eta: 0:01:54 acc1_pos: 1.042 acc5_pos: 100.000 loss: 1.5604 (1.4683) acc1: 0.5208 (16.5155) acc5: 100.0000 (100.0000) time: 0.1649 data: 0.0703 max mem: 3531 +Test: [ 40/389] eta: 0:01:43 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.5822 (1.5076) acc1: 0.2604 (15.0597) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1319 max mem: 3531 +Test: [ 50/389] eta: 0:01:33 acc1_pos: 42.188 acc5_pos: 100.000 loss: 1.2897 (1.4537) acc1: 3.1250 (16.7228) acc5: 100.0000 (100.0000) time: 0.2055 data: 0.1106 max mem: 3531 +Test: [ 60/389] eta: 0:01:25 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.9729 (1.4393) acc1: 25.7812 (17.1491) acc5: 100.0000 (100.0000) time: 0.1863 data: 0.0925 max mem: 3531 +Test: [ 70/389] eta: 0:01:19 acc1_pos: 32.292 acc5_pos: 100.000 loss: 1.4626 (1.4581) acc1: 3.9062 (16.1678) acc5: 100.0000 (100.0000) time: 0.1802 data: 0.0883 max mem: 3531 +Test: [ 80/389] eta: 0:01:12 acc1_pos: 50.521 acc5_pos: 100.000 loss: 1.3643 (1.4115) acc1: 14.3229 (18.5410) acc5: 100.0000 (100.0000) time: 0.1549 data: 0.0618 max mem: 3531 +Test: [ 90/389] eta: 0:01:09 acc1_pos: 25.781 acc5_pos: 100.000 loss: 1.1110 (1.3733) acc1: 30.9896 (20.5329) acc5: 100.0000 (100.0000) time: 0.1740 data: 0.0795 max mem: 3531 +Test: [100/389] eta: 0:01:05 acc1_pos: 74.479 acc5_pos: 100.000 loss: 1.2837 (1.3858) acc1: 22.1354 (20.5987) acc5: 100.0000 (100.0000) time: 0.1855 data: 0.0900 max mem: 3531 +Test: [110/389] eta: 0:01:00 acc1_pos: 95.312 acc5_pos: 100.000 loss: 1.4206 (1.3597) acc1: 18.4896 (23.4047) acc5: 100.0000 (100.0000) time: 0.1529 data: 0.0597 max mem: 3531 +Test: [120/389] eta: 0:00:57 acc1_pos: 40.365 acc5_pos: 100.000 loss: 0.9772 (1.3466) acc1: 40.3646 (24.2144) acc5: 100.0000 (100.0000) time: 0.1460 data: 0.0549 max mem: 3531 +Test: [130/389] eta: 0:00:54 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.7977 (1.2904) acc1: 51.3021 (27.5962) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1628 data: 0.0712 max mem: 3531 +Test: [140/389] eta: 0:00:52 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1772 (1.2143) acc1: 99.2188 (32.2991) acc5: 100.0000 (100.0000) acc1_neg: 74.740 acc5_neg: 100.000 time: 0.1925 data: 0.1002 max mem: 3531 +Test: [150/389] eta: 0:00:50 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1772 (1.1569) acc1: 99.7396 (35.8789) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2109 data: 0.1190 max mem: 3531 +Test: [160/389] eta: 0:00:47 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2469 (1.1010) acc1: 99.7396 (39.5122) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2053 data: 0.1146 max mem: 3531 +Test: [170/389] eta: 0:00:45 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2175 (1.0489) acc1: 100.0000 (43.0403) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1989 data: 0.1077 max mem: 3531 +Test: [180/389] eta: 0:00:43 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2158 (1.0035) acc1: 100.0000 (45.9945) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1993 data: 0.1076 max mem: 3531 +Test: [190/389] eta: 0:00:40 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1706 (0.9652) acc1: 98.9583 (48.4443) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1756 data: 0.0838 max mem: 3531 +Test: [200/389] eta: 0:00:38 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1958 (0.9283) acc1: 98.9583 (50.7955) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1613 data: 0.0703 max mem: 3531 +Test: [210/389] eta: 0:00:36 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1903 (0.8937) acc1: 100.0000 (53.0559) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1736 data: 0.0825 max mem: 3532 +Test: [220/389] eta: 0:00:33 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1570 (0.8599) acc1: 100.0000 (55.1777) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0827 max mem: 3532 +Test: [230/389] eta: 0:00:31 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1585 (0.8314) acc1: 100.0000 (57.1158) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1761 data: 0.0849 max mem: 3532 +Test: [240/389] eta: 0:00:29 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2200 (0.8125) acc1: 100.0000 (58.2058) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1698 data: 0.0789 max mem: 3532 +Test: [250/389] eta: 0:00:27 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1678 (0.7857) acc1: 100.0000 (59.8709) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1761 data: 0.0841 max mem: 3532 +Test: [260/389] eta: 0:00:25 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1436 (0.7664) acc1: 100.0000 (61.1730) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1785 data: 0.0857 max mem: 3532 +Test: [270/389] eta: 0:00:23 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2677 (0.7462) acc1: 99.2188 (62.5653) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1748 data: 0.0824 max mem: 3532 +Test: [280/389] eta: 0:00:21 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1554 (0.7254) acc1: 100.0000 (63.8957) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1820 data: 0.0903 max mem: 3532 +Test: [290/389] eta: 0:00:19 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1559 (0.7085) acc1: 100.0000 (64.9914) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1801 data: 0.0860 max mem: 3532 +Test: [300/389] eta: 0:00:17 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2427 (0.6946) acc1: 98.9583 (66.1130) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1796 data: 0.0856 max mem: 3532 +Test: [310/389] eta: 0:00:15 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2693 (0.6796) acc1: 98.4375 (67.1004) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1721 data: 0.0800 max mem: 3532 +Test: [320/389] eta: 0:00:13 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1593 (0.6632) acc1: 100.0000 (68.1213) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1727 data: 0.0811 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1740 (0.6488) acc1: 100.0000 (69.0812) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1849 data: 0.0936 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1912 (0.6350) acc1: 100.0000 (69.9834) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1832 data: 0.0921 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2291 (0.6323) acc1: 100.0000 (69.9282) acc5: 100.0000 (100.0000) acc1_neg: 24.479 acc5_neg: 100.000 time: 0.1821 data: 0.0914 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.7462 (0.6372) acc1: 46.8750 (69.1157) acc5: 100.0000 (100.0000) acc1_neg: 48.177 acc5_neg: 100.000 time: 0.1871 data: 0.0958 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.7103 (0.6281) acc1: 48.6979 (69.6534) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1737 data: 0.0825 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1734 (0.6174) acc1: 98.1771 (70.3330) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1510 data: 0.0605 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1734 (0.6089) acc1: 100.0000 (70.8579) acc5: 100.0000 (100.0000) acc1_neg: 72.678 acc5_neg: 100.000 time: 0.1401 data: 0.0467 max mem: 3532 +Test: Total time: 0:01:13 (0.1894 s / it) +* Overall Acc@1 70.858 Acc@5 100.000 + target == real Acc@1 24.273 Acc@5 100.000 + target == fake Acc@1 92.512 Acc@5 100.000 + loss 0.609 + PR-AUC 0.7864 ROC-AUC 0.6645 EER 38.57% (thr=0.8091) +Accuracy of the network on the 149358 test images: 70.9% +[rank0]:[W725 17:56:37.543861688 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 17:56:56.200000 600130 torch/distributed/run.py:766] +W0725 17:56:56.200000 600130 torch/distributed/run.py:766] ***************************************** +W0725 17:56:56.200000 600130 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:56:56.200000 600130 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:57:25.715589813 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:57:25.976952189 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:57:25.304424955 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:57:25.419708618 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 17:57:25.442086229 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:57:26.516853388 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:57:26.555313111 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:57:26.675745347 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed3_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed3_resampled_n116-041-153_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:33:08 acc1_pos: 26.562 acc5_pos: 100.000 loss: 1.0835 (1.0835) acc1: 26.5625 (26.5625) acc5: 100.0000 (100.0000) time: 5.9720 data: 4.4977 max mem: 3531 +Test: [ 10/333] eta: 0:03:21 acc1_pos: 25.000 acc5_pos: 100.000 loss: 1.0080 (1.0268) acc1: 30.9896 (32.2680) acc5: 100.0000 (100.0000) time: 0.6248 data: 0.4093 max mem: 3531 +Test: [ 20/333] eta: 0:01:56 acc1_pos: 10.938 acc5_pos: 100.000 loss: 1.1817 (1.1435) acc1: 24.2188 (25.4092) acc5: 100.0000 (100.0000) time: 0.0934 data: 0.0004 max mem: 3531 +Test: [ 30/333] eta: 0:01:30 acc1_pos: 87.760 acc5_pos: 100.000 loss: 0.9932 (0.9410) acc1: 24.7396 (41.7255) acc5: 100.0000 (100.0000) time: 0.1207 data: 0.0234 max mem: 3531 +Test: [ 40/333] eta: 0:01:21 acc1_pos: 81.510 acc5_pos: 100.000 loss: 0.4299 (0.8248) acc1: 81.2500 (50.7749) acc5: 100.0000 (100.0000) time: 0.1799 data: 0.0845 max mem: 3531 +Test: [ 50/333] eta: 0:01:14 acc1_pos: 73.177 acc5_pos: 100.000 loss: 0.4635 (0.7527) acc1: 77.6042 (56.3215) acc5: 100.0000 (100.0000) time: 0.2076 data: 0.1139 max mem: 3531 +Test: [ 60/333] eta: 0:01:05 acc1_pos: 82.292 acc5_pos: 100.000 loss: 0.4506 (0.7035) acc1: 81.2500 (60.2630) acc5: 100.0000 (100.0000) time: 0.1649 data: 0.0690 max mem: 3531 +Test: [ 70/333] eta: 0:01:01 acc1_pos: 69.792 acc5_pos: 100.000 loss: 0.4506 (0.6669) acc1: 82.2917 (63.1932) acc5: 100.0000 (100.0000) time: 0.1600 data: 0.0645 max mem: 3531 +Test: [ 80/333] eta: 0:00:58 acc1_pos: 75.000 acc5_pos: 100.000 loss: 0.4313 (0.6389) acc1: 82.0312 (65.4964) acc5: 100.0000 (100.0000) time: 0.2049 data: 0.1092 max mem: 3531 +Test: [ 90/333] eta: 0:00:55 acc1_pos: 77.865 acc5_pos: 100.000 loss: 0.4491 (0.6219) acc1: 80.4688 (66.8040) acc5: 100.0000 (100.0000) time: 0.2049 data: 0.1075 max mem: 3531 +Test: [100/333] eta: 0:00:51 acc1_pos: 82.031 acc5_pos: 100.000 loss: 0.5042 (0.6125) acc1: 76.0417 (67.5511) acc5: 100.0000 (100.0000) time: 0.1799 data: 0.0827 max mem: 3531 +Test: [110/333] eta: 0:00:49 acc1_pos: 82.292 acc5_pos: 100.000 loss: 0.5002 (0.6018) acc1: 76.3021 (68.3371) acc5: 100.0000 (100.0000) time: 0.1849 data: 0.0870 max mem: 3531 +Test: [120/333] eta: 0:00:45 acc1_pos: 68.750 acc5_pos: 100.000 loss: 0.4972 (0.5945) acc1: 73.4375 (68.7952) acc5: 100.0000 (100.0000) time: 0.1799 data: 0.0840 max mem: 3531 +Test: [130/333] eta: 0:00:43 acc1_pos: 86.198 acc5_pos: 100.000 loss: 0.5171 (0.5883) acc1: 73.1771 (69.2112) acc5: 100.0000 (100.0000) time: 0.1699 data: 0.0767 max mem: 3531 +Test: [140/333] eta: 0:00:40 acc1_pos: 77.865 acc5_pos: 100.000 loss: 0.5254 (0.5842) acc1: 75.7812 (69.5922) acc5: 100.0000 (100.0000) time: 0.1799 data: 0.0874 max mem: 3531 +Test: [150/333] eta: 0:00:38 acc1_pos: 66.667 acc5_pos: 100.000 loss: 0.5194 (0.5799) acc1: 75.7812 (69.8727) acc5: 100.0000 (100.0000) time: 0.1850 data: 0.0899 max mem: 3531 +Test: [160/333] eta: 0:00:35 acc1_pos: 74.342 acc5_pos: 100.000 loss: 0.5173 (0.5895) acc1: 75.7812 (69.2854) acc5: 100.0000 (100.0000) acc1_neg: 28.125 acc5_neg: 100.000 time: 0.1798 data: 0.0845 max mem: 3531 +Test: [170/333] eta: 0:00:33 acc1_pos: 74.342 acc5_pos: 100.000 loss: 0.9595 (0.6125) acc1: 39.3229 (67.4799) acc5: 100.0000 (100.0000) acc1_neg: 42.969 acc5_neg: 100.000 time: 0.1954 data: 0.1025 max mem: 3531 +Test: [180/333] eta: 0:00:31 acc1_pos: 74.342 acc5_pos: 100.000 loss: 0.9385 (0.6280) acc1: 39.8438 (66.1415) acc5: 100.0000 (100.0000) acc1_neg: 39.844 acc5_neg: 100.000 time: 0.2049 data: 0.1110 max mem: 3531 +Test: [190/333] eta: 0:00:29 acc1_pos: 74.342 acc5_pos: 100.000 loss: 0.8821 (0.6398) acc1: 42.9688 (65.0946) acc5: 100.0000 (100.0000) acc1_neg: 36.458 acc5_neg: 100.000 time: 0.1792 data: 0.0851 max mem: 3531 +Test: [200/333] eta: 0:00:27 acc1_pos: 74.342 acc5_pos: 100.000 loss: 0.8217 (0.6495) acc1: 48.1771 (64.3216) acc5: 100.0000 (100.0000) acc1_neg: 46.354 acc5_neg: 100.000 time: 0.1870 data: 0.0951 max mem: 3531 +Test: [210/333] eta: 0:00:25 acc1_pos: 74.342 acc5_pos: 100.000 loss: 0.8622 (0.6603) acc1: 46.8750 (63.4479) acc5: 100.0000 (100.0000) acc1_neg: 57.031 acc5_neg: 100.000 time: 0.2157 data: 0.1253 max mem: 3532 +Test: [220/333] eta: 0:00:23 acc1_pos: 74.342 acc5_pos: 100.000 loss: 0.8769 (0.6718) acc1: 42.7083 (62.3362) acc5: 100.0000 (100.0000) acc1_neg: 42.708 acc5_neg: 100.000 time: 0.2000 data: 0.1091 max mem: 3532 +Test: [230/333] eta: 0:00:21 acc1_pos: 74.342 acc5_pos: 100.000 loss: 0.8875 (0.6821) acc1: 39.5833 (61.4741) acc5: 100.0000 (100.0000) acc1_neg: 45.312 acc5_neg: 100.000 time: 0.2042 data: 0.1113 max mem: 3532 +Test: [240/333] eta: 0:00:19 acc1_pos: 74.342 acc5_pos: 100.000 loss: 0.9616 (0.7004) acc1: 35.6771 (60.0687) acc5: 100.0000 (100.0000) acc1_neg: 20.312 acc5_neg: 100.000 time: 0.2127 data: 0.1184 max mem: 3532 +Test: [250/333] eta: 0:00:16 acc1_pos: 74.342 acc5_pos: 100.000 loss: 1.0968 (0.7105) acc1: 28.6458 (59.2733) acc5: 100.0000 (100.0000) acc1_neg: 54.167 acc5_neg: 100.000 time: 0.1852 data: 0.0929 max mem: 3532 +Test: [260/333] eta: 0:00:14 acc1_pos: 74.342 acc5_pos: 100.000 loss: 0.7831 (0.7133) acc1: 48.6979 (59.0288) acc5: 100.0000 (100.0000) acc1_neg: 59.896 acc5_neg: 100.000 time: 0.2052 data: 0.1148 max mem: 3532 +Test: [270/333] eta: 0:00:12 acc1_pos: 74.342 acc5_pos: 100.000 loss: 0.7642 (0.7152) acc1: 55.9896 (58.9224) acc5: 100.0000 (100.0000) acc1_neg: 56.510 acc5_neg: 100.000 time: 0.1984 data: 0.1068 max mem: 3532 +Test: [280/333] eta: 0:00:10 acc1_pos: 74.342 acc5_pos: 100.000 loss: 0.7877 (0.7252) acc1: 53.9062 (58.2481) acc5: 100.0000 (100.0000) acc1_neg: 34.375 acc5_neg: 100.000 time: 0.1749 data: 0.0829 max mem: 3532 +Test: [290/333] eta: 0:00:08 acc1_pos: 74.342 acc5_pos: 100.000 loss: 1.1072 (0.7382) acc1: 32.2917 (57.3301) acc5: 100.0000 (100.0000) acc1_neg: 33.073 acc5_neg: 100.000 time: 0.1901 data: 0.0992 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 74.342 acc5_pos: 100.000 loss: 1.0212 (0.7463) acc1: 33.0729 (56.6791) acc5: 100.0000 (100.0000) acc1_neg: 40.104 acc5_neg: 100.000 time: 0.1905 data: 0.0995 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 74.342 acc5_pos: 100.000 loss: 0.9652 (0.7495) acc1: 40.1042 (56.4032) acc5: 100.0000 (100.0000) acc1_neg: 30.729 acc5_neg: 100.000 time: 0.1771 data: 0.0852 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 74.342 acc5_pos: 100.000 loss: 1.0096 (0.7580) acc1: 34.3750 (55.7381) acc5: 100.0000 (100.0000) acc1_neg: 49.219 acc5_neg: 100.000 time: 0.1638 data: 0.0721 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 74.342 acc5_pos: 100.000 loss: 0.9286 (0.7623) acc1: 40.1042 (55.3515) acc5: 100.0000 (100.0000) acc1_neg: 40.104 acc5_neg: 100.000 time: 0.1555 data: 0.0648 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 74.342 acc5_pos: 100.000 loss: 0.9248 (0.7630) acc1: 42.4479 (55.2920) acc5: 100.0000 (100.0000) acc1_neg: 45.070 acc5_neg: 100.000 time: 0.1613 data: 0.0614 max mem: 3532 +Test: Total time: 0:01:06 (0.1988 s / it) +* Overall Acc@1 55.292 Acc@5 100.000 + target == real Acc@1 70.298 Acc@5 100.000 + target == fake Acc@1 41.934 Acc@5 100.000 + loss 0.763 + PR-AUC 0.6046 ROC-AUC 0.5991 EER 42.82% (thr=0.3882) +Accuracy of the network on the 127843 test images: 55.3% +[rank0]:[W725 17:58:43.409787157 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 17:59:03.170000 600698 torch/distributed/run.py:766] +W0725 17:59:03.170000 600698 torch/distributed/run.py:766] ***************************************** +W0725 17:59:03.170000 600698 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:59:03.170000 600698 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W725 17:59:32.109531307 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:59:32.152486362 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:59:32.384501839 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:59:33.707408691 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:59:33.739413260 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:59:33.751907483 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:59:33.782174085 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank4]:[W725 17:59:33.782283098 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed3_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed3_resampled_n116-041-153_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:21:17 acc1_pos: 84.375 acc5_pos: 100.000 loss: 0.3592 (0.3592) acc1: 84.3750 (84.3750) acc5: 100.0000 (100.0000) time: 5.4376 data: 3.6007 max mem: 3531 +Test: [ 10/235] eta: 0:02:19 acc1_pos: 84.115 acc5_pos: 100.000 loss: 0.3568 (0.3577) acc1: 85.1562 (85.0852) acc5: 100.0000 (100.0000) time: 0.6185 data: 0.3696 max mem: 3531 +Test: [ 20/235] eta: 0:01:22 acc1_pos: 80.729 acc5_pos: 100.000 loss: 0.3552 (0.3541) acc1: 85.1562 (85.4043) acc5: 100.0000 (100.0000) time: 0.1316 data: 0.0387 max mem: 3531 +Test: [ 30/235] eta: 0:01:04 acc1_pos: 86.979 acc5_pos: 100.000 loss: 0.3432 (0.3503) acc1: 86.1979 (85.7359) acc5: 100.0000 (100.0000) time: 0.1483 data: 0.0531 max mem: 3531 +Test: [ 40/235] eta: 0:00:58 acc1_pos: 87.240 acc5_pos: 100.000 loss: 0.3430 (0.3483) acc1: 86.1979 (85.7533) acc5: 100.0000 (100.0000) time: 0.2067 data: 0.1142 max mem: 3531 +Test: [ 50/235] eta: 0:00:52 acc1_pos: 83.854 acc5_pos: 100.000 loss: 0.3429 (0.3491) acc1: 85.6771 (85.8098) acc5: 100.0000 (100.0000) time: 0.2325 data: 0.1419 max mem: 3531 +Test: [ 60/235] eta: 0:00:48 acc1_pos: 87.240 acc5_pos: 100.000 loss: 0.3448 (0.3488) acc1: 86.4583 (85.8521) acc5: 100.0000 (100.0000) time: 0.2281 data: 0.1315 max mem: 3531 +Test: [ 70/235] eta: 0:00:44 acc1_pos: 80.000 acc5_pos: 100.000 loss: 0.3683 (0.4560) acc1: 83.3333 (79.2290) acc5: 100.0000 (100.0000) acc1_neg: 30.729 acc5_neg: 100.000 time: 0.2322 data: 0.1337 max mem: 3531 +Test: [ 80/235] eta: 0:00:40 acc1_pos: 80.000 acc5_pos: 100.000 loss: 1.2033 (0.5504) acc1: 33.5938 (73.5693) acc5: 100.0000 (100.0000) acc1_neg: 30.208 acc5_neg: 100.000 time: 0.2254 data: 0.1322 max mem: 3531 +Test: [ 90/235] eta: 0:00:36 acc1_pos: 80.000 acc5_pos: 100.000 loss: 1.2209 (0.6254) acc1: 33.0729 (69.0591) acc5: 100.0000 (100.0000) acc1_neg: 33.073 acc5_neg: 100.000 time: 0.2049 data: 0.1114 max mem: 3531 +Test: [100/235] eta: 0:00:34 acc1_pos: 80.000 acc5_pos: 100.000 loss: 1.2177 (0.6822) acc1: 32.0312 (65.6121) acc5: 100.0000 (100.0000) acc1_neg: 33.854 acc5_neg: 100.000 time: 0.2420 data: 0.1474 max mem: 3531 +Test: [110/235] eta: 0:00:31 acc1_pos: 80.000 acc5_pos: 100.000 loss: 1.2121 (0.7299) acc1: 33.5938 (62.6900) acc5: 100.0000 (100.0000) acc1_neg: 31.510 acc5_neg: 100.000 time: 0.2349 data: 0.1425 max mem: 3531 +Test: [120/235] eta: 0:00:27 acc1_pos: 80.000 acc5_pos: 100.000 loss: 1.2525 (0.7863) acc1: 28.3854 (59.4719) acc5: 100.0000 (100.0000) acc1_neg: 21.875 acc5_neg: 100.000 time: 0.1573 data: 0.0623 max mem: 3531 +Test: [130/235] eta: 0:00:24 acc1_pos: 80.000 acc5_pos: 100.000 loss: 1.4115 (0.8339) acc1: 22.3958 (56.6834) acc5: 100.0000 (100.0000) acc1_neg: 22.656 acc5_neg: 100.000 time: 0.1653 data: 0.0690 max mem: 3531 +Test: [140/235] eta: 0:00:22 acc1_pos: 80.000 acc5_pos: 100.000 loss: 1.4402 (0.8788) acc1: 21.6146 (54.1704) acc5: 100.0000 (100.0000) acc1_neg: 20.573 acc5_neg: 100.000 time: 0.1999 data: 0.1045 max mem: 3531 +Test: [150/235] eta: 0:00:19 acc1_pos: 80.000 acc5_pos: 100.000 loss: 1.4402 (0.9147) acc1: 21.3542 (52.0764) acc5: 100.0000 (100.0000) acc1_neg: 20.833 acc5_neg: 100.000 time: 0.2145 data: 0.1166 max mem: 3531 +Test: [160/235] eta: 0:00:17 acc1_pos: 80.000 acc5_pos: 100.000 loss: 1.4244 (0.9465) acc1: 22.1354 (50.2167) acc5: 100.0000 (100.0000) acc1_neg: 23.958 acc5_neg: 100.000 time: 0.1999 data: 0.1021 max mem: 3531 +Test: [170/235] eta: 0:00:14 acc1_pos: 80.000 acc5_pos: 100.000 loss: 1.3825 (0.9463) acc1: 23.9583 (49.9437) acc5: 100.0000 (100.0000) acc1_neg: 45.052 acc5_neg: 100.000 time: 0.1613 data: 0.0645 max mem: 3531 +Test: [180/235] eta: 0:00:12 acc1_pos: 80.000 acc5_pos: 100.000 loss: 0.8808 (0.9433) acc1: 48.4375 (49.8202) acc5: 100.0000 (100.0000) acc1_neg: 45.052 acc5_neg: 100.000 time: 0.1817 data: 0.0867 max mem: 3531 +Test: [190/235] eta: 0:00:10 acc1_pos: 80.000 acc5_pos: 100.000 loss: 0.8808 (0.9408) acc1: 48.1771 (49.7082) acc5: 100.0000 (100.0000) acc1_neg: 50.000 acc5_neg: 100.000 time: 0.2312 data: 0.1390 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 80.000 acc5_pos: 100.000 loss: 0.8974 (0.9390) acc1: 46.6146 (49.5712) acc5: 100.0000 (100.0000) acc1_neg: 48.958 acc5_neg: 100.000 time: 0.2062 data: 0.1149 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 80.000 acc5_pos: 100.000 loss: 0.9052 (0.9365) acc1: 46.6146 (49.5014) acc5: 100.0000 (100.0000) acc1_neg: 50.000 acc5_neg: 100.000 time: 0.1561 data: 0.0649 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 80.000 acc5_pos: 100.000 loss: 0.8974 (0.9343) acc1: 47.9167 (49.4450) acc5: 100.0000 (100.0000) acc1_neg: 46.094 acc5_neg: 100.000 time: 0.1415 data: 0.0516 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 80.000 acc5_pos: 100.000 loss: 0.8856 (0.9318) acc1: 48.9583 (49.4792) acc5: 100.0000 (100.0000) acc1_neg: 51.562 acc5_neg: 100.000 time: 0.1429 data: 0.0533 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 80.000 acc5_pos: 100.000 loss: 0.8856 (0.9315) acc1: 48.4375 (49.4431) acc5: 100.0000 (100.0000) acc1_neg: 39.130 acc5_neg: 100.000 time: 0.1303 data: 0.0434 max mem: 3532 +Test: Total time: 0:00:49 (0.2116 s / it) +* Overall Acc@1 49.443 Acc@5 100.000 + target == real Acc@1 85.812 Acc@5 100.000 + target == fake Acc@1 36.289 Acc@5 100.000 + loss 0.931 + PR-AUC 0.8526 ROC-AUC 0.6920 EER 36.38% (thr=0.2494) +Accuracy of the network on the 89879 test images: 49.4% +[rank0]:[W725 18:00:34.269793034 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:00:54.358000 601267 torch/distributed/run.py:766] +W0725 18:00:54.358000 601267 torch/distributed/run.py:766] ***************************************** +W0725 18:00:54.358000 601267 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:00:54.358000 601267 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:01:24.232779633 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:01:24.467391707 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:01:25.575463710 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:01:25.881624735 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:01:25.915113475 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:01:25.993304144 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:01:25.011119847 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:01:25.038160840 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed3_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed3_resampled_n116-041-153_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:30:06 acc1_pos: 54.167 acc5_pos: 100.000 loss: 0.7278 (0.7278) acc1: 54.1667 (54.1667) acc5: 100.0000 (100.0000) time: 6.7156 data: 4.9842 max mem: 3531 +Test: [ 10/269] eta: 0:03:00 acc1_pos: 76.562 acc5_pos: 100.000 loss: 0.6099 (0.6091) acc1: 67.7083 (67.2822) acc5: 100.0000 (100.0000) time: 0.6952 data: 0.4535 max mem: 3531 +Test: [ 20/269] eta: 0:01:42 acc1_pos: 70.833 acc5_pos: 100.000 loss: 0.6146 (0.6393) acc1: 67.7083 (64.9058) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0006 max mem: 3531 +Test: [ 30/269] eta: 0:01:26 acc1_pos: 51.823 acc5_pos: 100.000 loss: 0.6209 (0.6236) acc1: 66.9271 (66.3727) acc5: 100.0000 (100.0000) time: 0.1768 data: 0.0796 max mem: 3531 +Test: [ 40/269] eta: 0:01:14 acc1_pos: 27.586 acc5_pos: 100.000 loss: 0.6687 (0.6504) acc1: 57.2917 (63.1479) acc5: 100.0000 (100.0000) acc1_neg: 60.156 acc5_neg: 100.000 time: 0.2355 data: 0.1403 max mem: 3531 +Test: [ 50/269] eta: 0:01:05 acc1_pos: 27.586 acc5_pos: 100.000 loss: 0.7261 (0.6676) acc1: 51.3021 (61.2081) acc5: 100.0000 (100.0000) acc1_neg: 40.104 acc5_neg: 100.000 time: 0.2036 data: 0.1104 max mem: 3531 +Test: [ 60/269] eta: 0:00:56 acc1_pos: 27.586 acc5_pos: 100.000 loss: 0.7560 (0.6898) acc1: 50.7812 (59.1615) acc5: 100.0000 (100.0000) acc1_neg: 37.500 acc5_neg: 100.000 time: 0.1596 data: 0.0674 max mem: 3531 +Test: [ 70/269] eta: 0:00:51 acc1_pos: 27.586 acc5_pos: 100.000 loss: 0.7886 (0.7023) acc1: 49.7396 (57.9592) acc5: 100.0000 (100.0000) acc1_neg: 53.125 acc5_neg: 100.000 time: 0.1553 data: 0.0608 max mem: 3531 +Test: [ 80/269] eta: 0:00:47 acc1_pos: 27.586 acc5_pos: 100.000 loss: 0.7425 (0.7052) acc1: 50.0000 (57.3077) acc5: 100.0000 (100.0000) acc1_neg: 49.479 acc5_neg: 100.000 time: 0.1905 data: 0.0976 max mem: 3531 +Test: [ 90/269] eta: 0:00:44 acc1_pos: 27.586 acc5_pos: 100.000 loss: 0.7206 (0.7080) acc1: 55.7292 (56.9397) acc5: 100.0000 (100.0000) acc1_neg: 51.823 acc5_neg: 100.000 time: 0.2014 data: 0.1111 max mem: 3531 +Test: [100/269] eta: 0:00:40 acc1_pos: 27.586 acc5_pos: 100.000 loss: 0.7427 (0.7180) acc1: 51.8229 (56.1159) acc5: 100.0000 (100.0000) acc1_neg: 45.833 acc5_neg: 100.000 time: 0.1951 data: 0.1048 max mem: 3531 +Test: [110/269] eta: 0:00:37 acc1_pos: 27.586 acc5_pos: 100.000 loss: 0.8239 (0.7310) acc1: 46.0938 (55.2670) acc5: 100.0000 (100.0000) acc1_neg: 53.385 acc5_neg: 100.000 time: 0.1877 data: 0.0959 max mem: 3531 +Test: [120/269] eta: 0:00:34 acc1_pos: 27.586 acc5_pos: 100.000 loss: 0.7807 (0.7324) acc1: 52.6042 (55.0900) acc5: 100.0000 (100.0000) acc1_neg: 35.677 acc5_neg: 100.000 time: 0.1801 data: 0.0878 max mem: 3531 +Test: [130/269] eta: 0:00:31 acc1_pos: 27.586 acc5_pos: 100.000 loss: 0.7732 (0.7413) acc1: 50.2604 (54.1349) acc5: 100.0000 (100.0000) acc1_neg: 35.417 acc5_neg: 100.000 time: 0.1909 data: 0.0964 max mem: 3531 +Test: [140/269] eta: 0:00:29 acc1_pos: 27.586 acc5_pos: 100.000 loss: 0.8256 (0.7478) acc1: 41.1458 (53.3762) acc5: 100.0000 (100.0000) acc1_neg: 55.729 acc5_neg: 100.000 time: 0.2001 data: 0.1061 max mem: 3531 +Test: [150/269] eta: 0:00:26 acc1_pos: 27.586 acc5_pos: 100.000 loss: 0.7945 (0.7520) acc1: 45.3125 (52.9249) acc5: 100.0000 (100.0000) acc1_neg: 52.344 acc5_neg: 100.000 time: 0.1902 data: 0.0993 max mem: 3531 +Test: [160/269] eta: 0:00:24 acc1_pos: 27.586 acc5_pos: 100.000 loss: 0.7525 (0.7473) acc1: 52.3438 (53.3110) acc5: 100.0000 (100.0000) acc1_neg: 72.917 acc5_neg: 100.000 time: 0.1809 data: 0.0899 max mem: 3531 +Test: [170/269] eta: 0:00:21 acc1_pos: 27.586 acc5_pos: 100.000 loss: 0.6355 (0.7406) acc1: 58.5938 (53.7981) acc5: 100.0000 (100.0000) acc1_neg: 50.000 acc5_neg: 100.000 time: 0.1860 data: 0.0945 max mem: 3531 +Test: [180/269] eta: 0:00:19 acc1_pos: 27.586 acc5_pos: 100.000 loss: 0.6088 (0.7351) acc1: 63.5417 (54.2429) acc5: 100.0000 (100.0000) acc1_neg: 50.260 acc5_neg: 100.000 time: 0.1981 data: 0.1069 max mem: 3531 +Test: [190/269] eta: 0:00:17 acc1_pos: 27.586 acc5_pos: 100.000 loss: 0.6944 (0.7358) acc1: 52.6042 (54.0181) acc5: 100.0000 (100.0000) acc1_neg: 48.438 acc5_neg: 100.000 time: 0.2122 data: 0.1205 max mem: 3531 +Test: [200/269] eta: 0:00:14 acc1_pos: 27.586 acc5_pos: 100.000 loss: 0.7687 (0.7377) acc1: 50.0000 (54.0695) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1958 data: 0.1038 max mem: 3531 +Test: [210/269] eta: 0:00:12 acc1_pos: 27.586 acc5_pos: 100.000 loss: 0.8508 (0.7426) acc1: 47.9167 (53.7421) acc5: 100.0000 (100.0000) acc1_neg: 42.969 acc5_neg: 100.000 time: 0.1702 data: 0.0783 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 27.586 acc5_pos: 100.000 loss: 0.8508 (0.7489) acc1: 42.9688 (53.2782) acc5: 100.0000 (100.0000) acc1_neg: 51.562 acc5_neg: 100.000 time: 0.1833 data: 0.0907 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 27.586 acc5_pos: 100.000 loss: 0.8638 (0.7516) acc1: 36.9792 (52.8578) acc5: 100.0000 (100.0000) acc1_neg: 60.677 acc5_neg: 100.000 time: 0.1916 data: 0.0966 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 27.586 acc5_pos: 100.000 loss: 0.7441 (0.7497) acc1: 54.4271 (53.0245) acc5: 100.0000 (100.0000) acc1_neg: 69.271 acc5_neg: 100.000 time: 0.2003 data: 0.1054 max mem: 3532 +Test: [250/269] eta: 0:00:03 acc1_pos: 27.586 acc5_pos: 100.000 loss: 0.7321 (0.7490) acc1: 54.4271 (53.0960) acc5: 100.0000 (100.0000) acc1_neg: 49.479 acc5_neg: 100.000 time: 0.1795 data: 0.0879 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 27.586 acc5_pos: 100.000 loss: 0.7509 (0.7517) acc1: 49.4792 (52.7778) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.1405 data: 0.0500 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 27.586 acc5_pos: 100.000 loss: 0.7076 (0.7479) acc1: 51.3021 (53.1062) acc5: 100.0000 (100.0000) acc1_neg: 78.215 acc5_neg: 100.000 time: 0.1279 data: 0.0341 max mem: 3532 +Test: Total time: 0:00:55 (0.2046 s / it) +* Overall Acc@1 53.106 Acc@5 100.000 + target == real Acc@1 65.519 Acc@5 100.000 + target == fake Acc@1 51.412 Acc@5 100.000 + loss 0.748 + PR-AUC 0.9178 ROC-AUC 0.6194 EER 41.54% (thr=0.4517) +Accuracy of the network on the 103293 test images: 53.1% +[rank0]:[W725 18:02:29.669038028 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-25T18:02:35Z] eval_done tag=fixed500k_real_n8_seed3_resampled +[2026-07-25T18:02:36Z] eval_start tag=fixed500k_fake_n16_seed3_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed3_resampled_n116-041-153_20260725_112559/best_checkpoint.pth +started_utc=2026-07-25T18:02:36Z +hostname=n116-041-153 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed3_resampled_n116-041-153_20260725_112559/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n16_seed3_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0725 18:02:49.248000 601840 torch/distributed/run.py:766] +W0725 18:02:49.248000 601840 torch/distributed/run.py:766] ***************************************** +W0725 18:02:49.248000 601840 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:02:49.248000 601840 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W725 18:03:19.203702862 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank6]:[W725 18:03:19.204055841 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:03:20.605698028 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:03:20.815119288 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:03:20.849606698 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:03:20.862237999 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:03:20.887292341 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:03:20.951426956 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n16_seed3_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed3_resampled_n116-041-153_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:27:00 acc1_pos: 10.938 acc5_pos: 100.000 loss: 1.3012 (1.3012) acc1: 10.9375 (10.9375) acc5: 100.0000 (100.0000) time: 6.9244 data: 5.4275 max mem: 3531 +Test: [ 10/234] eta: 0:02:46 acc1_pos: 13.802 acc5_pos: 100.000 loss: 1.0682 (1.0629) acc1: 22.6562 (25.1184) acc5: 100.0000 (100.0000) time: 0.7442 data: 0.5214 max mem: 3531 +Test: [ 20/234] eta: 0:01:33 acc1_pos: 89.323 acc5_pos: 100.000 loss: 1.0059 (0.9535) acc1: 24.4792 (34.8214) acc5: 100.0000 (100.0000) time: 0.1130 data: 0.0156 max mem: 3531 +Test: [ 30/234] eta: 0:01:16 acc1_pos: 59.375 acc5_pos: 100.000 loss: 0.3084 (0.7506) acc1: 89.0625 (51.4869) acc5: 100.0000 (100.0000) time: 0.1749 data: 0.0800 max mem: 3531 +Test: [ 40/234] eta: 0:01:07 acc1_pos: 99.479 acc5_pos: 100.000 loss: 0.2952 (0.6740) acc1: 89.5833 (57.4505) acc5: 100.0000 (100.0000) time: 0.2499 data: 0.1587 max mem: 3531 +Test: [ 50/234] eta: 0:00:57 acc1_pos: 98.698 acc5_pos: 100.000 loss: 0.1232 (0.5653) acc1: 98.4375 (65.5586) acc5: 100.0000 (100.0000) time: 0.2162 data: 0.1246 max mem: 3531 +Test: [ 60/234] eta: 0:00:50 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.1291 (0.4959) acc1: 98.6979 (70.9486) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1771 data: 0.0835 max mem: 3531 +Test: [ 70/234] eta: 0:00:46 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.1519 (0.5032) acc1: 97.9167 (71.3102) acc5: 100.0000 (100.0000) acc1_neg: 60.677 acc5_neg: 100.000 time: 0.1965 data: 0.0998 max mem: 3531 +Test: [ 80/234] eta: 0:00:41 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.7038 (0.5045) acc1: 61.9792 (71.8236) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2048 data: 0.1110 max mem: 3531 +Test: [ 90/234] eta: 0:00:37 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.1546 (0.4634) acc1: 97.9167 (74.6623) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.2000 data: 0.1095 max mem: 3531 +Test: [100/234] eta: 0:00:35 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.1546 (0.4448) acc1: 95.8333 (76.0803) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2307 data: 0.1369 max mem: 3531 +Test: [110/234] eta: 0:00:31 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.2486 (0.4287) acc1: 88.5417 (77.4141) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.2257 data: 0.1283 max mem: 3531 +Test: [120/234] eta: 0:00:28 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.2477 (0.4307) acc1: 92.4479 (77.7419) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1873 data: 0.0923 max mem: 3531 +Test: [130/234] eta: 0:00:25 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.2588 (0.4232) acc1: 94.0104 (78.7552) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.1611 data: 0.0685 max mem: 3531 +Test: [140/234] eta: 0:00:22 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.4043 (0.4574) acc1: 83.0729 (76.8783) acc5: 100.0000 (100.0000) acc1_neg: 21.354 acc5_neg: 100.000 time: 0.1595 data: 0.0669 max mem: 3531 +Test: [150/234] eta: 0:00:19 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.4188 (0.4565) acc1: 80.7292 (76.9885) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1770 data: 0.0843 max mem: 3531 +Test: [160/234] eta: 0:00:17 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.3403 (0.4563) acc1: 82.5521 (77.2095) acc5: 100.0000 (100.0000) acc1_neg: 70.833 acc5_neg: 100.000 time: 0.2006 data: 0.1088 max mem: 3531 +Test: [170/234] eta: 0:00:14 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.6189 (0.4626) acc1: 68.4896 (76.9920) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1853 data: 0.0931 max mem: 3531 +Test: [180/234] eta: 0:00:12 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.5951 (0.4696) acc1: 68.4896 (76.6416) acc5: 100.0000 (100.0000) acc1_neg: 37.500 acc5_neg: 100.000 time: 0.1606 data: 0.0687 max mem: 3531 +Test: [190/234] eta: 0:00:09 acc1_pos: 99.310 acc5_pos: 100.000 loss: 1.1300 (0.5154) acc1: 35.4167 (74.0006) acc5: 100.0000 (100.0000) acc1_neg: 26.042 acc5_neg: 100.000 time: 0.1874 data: 0.0938 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 99.310 acc5_pos: 100.000 loss: 1.3314 (0.5225) acc1: 26.0417 (73.8327) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2137 data: 0.1199 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.2591 (0.5153) acc1: 96.6146 (74.4347) acc5: 100.0000 (100.0000) acc1_neg: 65.365 acc5_neg: 100.000 time: 0.2198 data: 0.1274 max mem: 3532 +Test: [220/234] eta: 0:00:03 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.5870 (0.5198) acc1: 67.9688 (74.0538) acc5: 100.0000 (100.0000) acc1_neg: 61.198 acc5_neg: 100.000 time: 0.1795 data: 0.0856 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.7030 (0.5429) acc1: 38.8021 (72.3733) acc5: 100.0000 (100.0000) acc1_neg: 36.719 acc5_neg: 100.000 time: 0.1435 data: 0.0520 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 99.310 acc5_pos: 100.000 loss: 1.0239 (0.5492) acc1: 36.7188 (71.9589) acc5: 100.0000 (100.0000) acc1_neg: 36.187 acc5_neg: 100.000 time: 0.1478 data: 0.0568 max mem: 3532 +Test: Total time: 0:00:50 (0.2157 s / it) +* Overall Acc@1 71.959 Acc@5 100.000 + target == real Acc@1 65.807 Acc@5 100.000 + target == fake Acc@1 73.693 Acc@5 100.000 + loss 0.549 + PR-AUC 0.9377 ROC-AUC 0.8060 EER 29.38% (thr=0.5552) +Accuracy of the network on the 89729 test images: 72.0% +[rank0]:[W725 18:04:18.771821009 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:04:38.228000 602422 torch/distributed/run.py:766] +W0725 18:04:38.228000 602422 torch/distributed/run.py:766] ***************************************** +W0725 18:04:38.228000 602422 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:04:38.228000 602422 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:05:08.099883940 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:05:08.235956935 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:05:08.315998725 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:05:08.348945472 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank4]:[W725 18:05:08.349143956 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:05:09.575889302 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:05:09.598762334 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank6]:[W725 18:05:09.603379271 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n16_seed3_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed3_resampled_n116-041-153_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 3:36:49 acc1_pos: 1.823 acc5_pos: 100.000 loss: 1.4775 (1.4775) acc1: 1.8229 (1.8229) acc5: 100.0000 (100.0000) time: 5.4501 data: 3.6687 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/2387] eta: 0:23:39 acc1_pos: 0.521 acc5_pos: 100.000 loss: 1.6182 (1.5914) acc1: 1.5625 (3.0303) acc5: 100.0000 (100.0000) time: 0.5973 data: 0.3540 max mem: 3531 +Test: [ 20/2387] eta: 0:14:10 acc1_pos: 0.521 acc5_pos: 100.000 loss: 1.5616 (1.5452) acc1: 1.8229 (3.5838) acc5: 100.0000 (100.0000) time: 0.1047 data: 0.0151 max mem: 3531 +Test: [ 30/2387] eta: 0:12:18 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.5616 (1.5961) acc1: 1.8229 (2.8562) acc5: 100.0000 (100.0000) time: 0.1574 data: 0.0647 max mem: 3531 +Test: [ 40/2387] eta: 0:11:35 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.6748 (1.6270) acc1: 0.2604 (2.2866) acc5: 100.0000 (100.0000) time: 0.2302 data: 0.1371 max mem: 3531 +Test: [ 50/2387] eta: 0:11:03 acc1_pos: 8.854 acc5_pos: 100.000 loss: 1.6159 (1.5874) acc1: 2.0833 (2.7778) acc5: 100.0000 (100.0000) time: 0.2376 data: 0.1468 max mem: 3531 +Test: [ 60/2387] eta: 0:10:39 acc1_pos: 7.812 acc5_pos: 100.000 loss: 1.4936 (1.5746) acc1: 4.6875 (3.2702) acc5: 100.0000 (100.0000) time: 0.2303 data: 0.1398 max mem: 3531 +Test: [ 70/2387] eta: 0:10:17 acc1_pos: 10.938 acc5_pos: 100.000 loss: 1.5962 (1.5868) acc1: 2.8646 (3.2864) acc5: 100.0000 (100.0000) time: 0.2222 data: 0.1316 max mem: 3531 +Test: [ 80/2387] eta: 0:10:01 acc1_pos: 1.302 acc5_pos: 100.000 loss: 1.6285 (1.5883) acc1: 1.3021 (3.1732) acc5: 100.0000 (100.0000) time: 0.2184 data: 0.1258 max mem: 3531 +Test: [ 90/2387] eta: 0:09:44 acc1_pos: 12.240 acc5_pos: 100.000 loss: 1.5686 (1.5849) acc1: 2.8646 (3.4369) acc5: 100.0000 (100.0000) time: 0.2117 data: 0.1198 max mem: 3531 +Test: [ 100/2387] eta: 0:09:24 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.6027 (1.5878) acc1: 2.8646 (3.3932) acc5: 100.0000 (100.0000) time: 0.1896 data: 0.0955 max mem: 3531 +Test: [ 110/2387] eta: 0:09:07 acc1_pos: 12.500 acc5_pos: 100.000 loss: 1.5964 (1.5832) acc1: 1.8229 (3.4675) acc5: 100.0000 (100.0000) time: 0.1781 data: 0.0820 max mem: 3531 +Test: [ 120/2387] eta: 0:09:09 acc1_pos: 10.417 acc5_pos: 100.000 loss: 1.5422 (1.5807) acc1: 3.1250 (3.6179) acc5: 100.0000 (100.0000) time: 0.2199 data: 0.1257 max mem: 3531 +Test: [ 130/2387] eta: 0:09:04 acc1_pos: 7.812 acc5_pos: 100.000 loss: 1.5585 (1.5787) acc1: 3.3854 (3.6339) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1509 max mem: 3531 +Test: [ 140/2387] eta: 0:09:01 acc1_pos: 3.125 acc5_pos: 100.000 loss: 1.5917 (1.5805) acc1: 2.3438 (3.6754) acc5: 100.0000 (100.0000) time: 0.2349 data: 0.1398 max mem: 3531 +Test: [ 150/2387] eta: 0:08:57 acc1_pos: 0.521 acc5_pos: 100.000 loss: 1.5855 (1.5743) acc1: 3.3854 (3.8028) acc5: 100.0000 (100.0000) time: 0.2349 data: 0.1398 max mem: 3531 +Test: [ 160/2387] eta: 0:08:49 acc1_pos: 0.457 acc5_pos: 100.000 loss: 1.4635 (1.5399) acc1: 5.7292 (6.0510) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2149 data: 0.1205 max mem: 3531 +Test: [ 170/2387] eta: 0:08:39 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1598 (1.4587) acc1: 98.9583 (11.5269) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1863 data: 0.0901 max mem: 3531 +Test: [ 180/2387] eta: 0:08:30 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1510 (1.3865) acc1: 99.7396 (16.3962) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1799 data: 0.0853 max mem: 3531 +Test: [ 190/2387] eta: 0:08:20 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1496 (1.3217) acc1: 99.4792 (20.7433) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1735 data: 0.0793 max mem: 3531 +Test: [ 200/2387] eta: 0:08:12 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1457 (1.2633) acc1: 99.7396 (24.6735) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1699 data: 0.0708 max mem: 3531 +Test: [ 210/2387] eta: 0:08:05 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1489 (1.2106) acc1: 99.7396 (28.2089) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0831 max mem: 3532 +Test: [ 220/2387] eta: 0:08:00 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1509 (1.1626) acc1: 99.7396 (31.4385) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1849 data: 0.0893 max mem: 3532 +Test: [ 230/2387] eta: 0:07:59 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1504 (1.1187) acc1: 99.7396 (34.3942) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2149 data: 0.1156 max mem: 3532 +Test: [ 240/2387] eta: 0:07:57 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1438 (1.0781) acc1: 99.7396 (37.1067) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2299 data: 0.1320 max mem: 3532 +Test: [ 250/2387] eta: 0:07:57 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1299 (1.0387) acc1: 100.0000 (39.6103) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2313 data: 0.1370 max mem: 3532 +Test: [ 260/2387] eta: 0:07:57 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0807 (1.0018) acc1: 100.0000 (41.9231) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2476 data: 0.1565 max mem: 3532 +Test: [ 270/2387] eta: 0:07:52 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0722 (0.9675) acc1: 100.0000 (44.0662) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2235 data: 0.1308 max mem: 3532 +Test: [ 280/2387] eta: 0:07:53 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0752 (0.9360) acc1: 100.0000 (46.0567) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2272 data: 0.1305 max mem: 3532 +Test: [ 290/2387] eta: 0:07:51 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0829 (0.9067) acc1: 100.0000 (47.9104) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2499 data: 0.1539 max mem: 3532 +Test: [ 300/2387] eta: 0:07:50 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0833 (0.8794) acc1: 100.0000 (49.6401) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2360 data: 0.1433 max mem: 3532 +Test: [ 310/2387] eta: 0:07:45 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0842 (0.8571) acc1: 100.0000 (51.2041) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2074 data: 0.1158 max mem: 3532 +Test: [ 320/2387] eta: 0:07:45 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.2860 (0.8394) acc1: 98.6979 (52.6772) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2241 data: 0.1319 max mem: 3532 +Test: [ 330/2387] eta: 0:07:45 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.2860 (0.8227) acc1: 98.4375 (54.0620) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2644 data: 0.1698 max mem: 3532 +Test: [ 340/2387] eta: 0:07:42 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.2859 (0.8071) acc1: 98.1771 (55.3664) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2402 data: 0.1448 max mem: 3532 +Test: [ 350/2387] eta: 0:07:40 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.2760 (0.7883) acc1: 99.4792 (56.6314) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2182 data: 0.1249 max mem: 3532 +Test: [ 360/2387] eta: 0:07:35 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1273 (0.7700) acc1: 100.0000 (57.8298) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1967 data: 0.1035 max mem: 3532 +Test: [ 370/2387] eta: 0:07:31 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1296 (0.7528) acc1: 100.0000 (58.9651) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1845 data: 0.0916 max mem: 3532 +Test: [ 380/2387] eta: 0:07:28 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1286 (0.7364) acc1: 100.0000 (60.0401) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2089 data: 0.1181 max mem: 3532 +Test: [ 390/2387] eta: 0:07:25 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1289 (0.7208) acc1: 100.0000 (61.0581) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2160 data: 0.1229 max mem: 3532 +Test: [ 400/2387] eta: 0:07:22 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1289 (0.7061) acc1: 100.0000 (62.0285) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2035 data: 0.1100 max mem: 3532 +Test: [ 410/2387] eta: 0:07:19 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1270 (0.6921) acc1: 100.0000 (62.9499) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1989 data: 0.1071 max mem: 3532 +Test: [ 420/2387] eta: 0:07:15 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1263 (0.6786) acc1: 100.0000 (63.8281) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1956 data: 0.1026 max mem: 3532 +Test: [ 430/2387] eta: 0:07:12 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1228 (0.6657) acc1: 100.0000 (64.6643) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1959 data: 0.1041 max mem: 3532 +Test: [ 440/2387] eta: 0:07:09 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1189 (0.6524) acc1: 100.0000 (65.4632) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2004 data: 0.1101 max mem: 3533 +Test: [ 450/2387] eta: 0:07:06 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0530 (0.6392) acc1: 100.0000 (66.2290) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2057 data: 0.1154 max mem: 3533 +Test: [ 460/2387] eta: 0:07:04 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0525 (0.6264) acc1: 100.0000 (66.9615) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2158 data: 0.1244 max mem: 3533 +Test: [ 470/2387] eta: 0:07:01 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0525 (0.6143) acc1: 100.0000 (67.6630) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2141 data: 0.1230 max mem: 3533 +Test: [ 480/2387] eta: 0:07:00 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0533 (0.6026) acc1: 100.0000 (68.3353) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2315 data: 0.1401 max mem: 3533 +Test: [ 490/2387] eta: 0:06:55 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0534 (0.5914) acc1: 100.0000 (68.9802) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2009 data: 0.1052 max mem: 3533 +Test: [ 500/2387] eta: 0:06:53 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0534 (0.5807) acc1: 100.0000 (69.5993) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1875 data: 0.0910 max mem: 3533 +Test: [ 510/2387] eta: 0:06:51 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0535 (0.5704) acc1: 100.0000 (70.1943) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2227 data: 0.1285 max mem: 3533 +Test: [ 520/2387] eta: 0:06:48 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0612 (0.5607) acc1: 100.0000 (70.7664) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2019 data: 0.1055 max mem: 3533 +Test: [ 530/2387] eta: 0:06:45 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0616 (0.5513) acc1: 100.0000 (71.3169) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1978 data: 0.1030 max mem: 3533 +Test: [ 540/2387] eta: 0:06:42 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0614 (0.5422) acc1: 100.0000 (71.8471) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2049 data: 0.1143 max mem: 3533 +Test: [ 550/2387] eta: 0:06:39 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0610 (0.5335) acc1: 100.0000 (72.3580) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1928 data: 0.1008 max mem: 3533 +Test: [ 560/2387] eta: 0:06:36 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0611 (0.5250) acc1: 100.0000 (72.8508) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1886 data: 0.0963 max mem: 3533 +Test: [ 570/2387] eta: 0:06:33 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0612 (0.5169) acc1: 100.0000 (73.3262) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1866 data: 0.0927 max mem: 3533 +Test: [ 580/2387] eta: 0:06:30 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0613 (0.5091) acc1: 100.0000 (73.7853) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1807 data: 0.0869 max mem: 3533 +Test: [ 590/2387] eta: 0:06:27 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0618 (0.5015) acc1: 100.0000 (74.2289) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1886 data: 0.0962 max mem: 3533 +Test: [ 600/2387] eta: 0:06:24 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0620 (0.4942) acc1: 100.0000 (74.6577) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1911 data: 0.0985 max mem: 3533 +Test: [ 610/2387] eta: 0:06:20 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0701 (0.4876) acc1: 100.0000 (75.0712) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1693 data: 0.0786 max mem: 3533 +Test: [ 620/2387] eta: 0:06:17 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0881 (0.4811) acc1: 100.0000 (75.4722) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1718 data: 0.0807 max mem: 3533 +Test: [ 630/2387] eta: 0:06:14 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0881 (0.4749) acc1: 100.0000 (75.8609) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1820 data: 0.0901 max mem: 3533 +Test: [ 640/2387] eta: 0:06:10 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0884 (0.4689) acc1: 100.0000 (76.2371) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1689 data: 0.0773 max mem: 3533 +Test: [ 650/2387] eta: 0:06:07 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0884 (0.4631) acc1: 100.0000 (76.6009) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1729 data: 0.0814 max mem: 3533 +Test: [ 660/2387] eta: 0:06:04 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0883 (0.4574) acc1: 100.0000 (76.9549) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1773 data: 0.0863 max mem: 3533 +Test: [ 670/2387] eta: 0:06:01 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0877 (0.4519) acc1: 100.0000 (77.2980) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1768 data: 0.0865 max mem: 3534 +Test: [ 680/2387] eta: 0:05:58 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0871 (0.4465) acc1: 100.0000 (77.6313) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1810 data: 0.0906 max mem: 3534 +Test: [ 690/2387] eta: 0:05:56 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0872 (0.4448) acc1: 100.0000 (77.7489) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.1836 data: 0.0926 max mem: 3534 +Test: [ 700/2387] eta: 0:05:53 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.4126 (0.4444) acc1: 80.7292 (77.7780) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.1869 data: 0.0942 max mem: 3534 +Test: [ 710/2387] eta: 0:05:50 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.4145 (0.4441) acc1: 80.4688 (77.8118) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.1785 data: 0.0859 max mem: 3534 +Test: [ 720/2387] eta: 0:05:47 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.4130 (0.4435) acc1: 80.9896 (77.8588) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.1772 data: 0.0860 max mem: 3534 +Test: [ 730/2387] eta: 0:05:45 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.4138 (0.4433) acc1: 80.9896 (77.8927) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.1799 data: 0.0874 max mem: 3534 +Test: [ 740/2387] eta: 0:05:42 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.4187 (0.4430) acc1: 80.2083 (77.9187) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.1721 data: 0.0790 max mem: 3534 +Test: [ 750/2387] eta: 0:05:40 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.4071 (0.4425) acc1: 80.2083 (77.9620) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.1926 data: 0.1007 max mem: 3534 +Test: [ 760/2387] eta: 0:05:37 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.4056 (0.4410) acc1: 81.5104 (78.0692) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1899 data: 0.0990 max mem: 3534 +Test: [ 770/2387] eta: 0:05:34 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.2138 (0.4373) acc1: 93.2292 (78.3057) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1632 data: 0.0721 max mem: 3534 +Test: [ 780/2387] eta: 0:05:31 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0988 (0.4330) acc1: 99.7396 (78.5808) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0756 max mem: 3534 +Test: [ 790/2387] eta: 0:05:28 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0915 (0.4291) acc1: 100.0000 (78.8108) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1739 data: 0.0781 max mem: 3534 +Test: [ 800/2387] eta: 0:05:25 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0885 (0.4249) acc1: 100.0000 (79.0724) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1649 data: 0.0708 max mem: 3534 +Test: [ 810/2387] eta: 0:05:23 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0885 (0.4208) acc1: 100.0000 (79.3291) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0765 max mem: 3534 +Test: [ 820/2387] eta: 0:05:20 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0940 (0.4170) acc1: 100.0000 (79.5758) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1749 data: 0.0800 max mem: 3534 +Test: [ 830/2387] eta: 0:05:17 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1048 (0.4135) acc1: 99.7396 (79.8016) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1718 data: 0.0753 max mem: 3534 +Test: [ 840/2387] eta: 0:05:15 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1495 (0.4107) acc1: 98.9583 (80.0250) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1655 data: 0.0698 max mem: 3534 +Test: [ 850/2387] eta: 0:05:12 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1804 (0.4081) acc1: 98.4375 (80.2414) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1630 data: 0.0673 max mem: 3534 +Test: [ 860/2387] eta: 0:05:09 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1804 (0.4054) acc1: 98.9583 (80.4588) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1766 data: 0.0819 max mem: 3534 +Test: [ 870/2387] eta: 0:05:07 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1773 (0.4028) acc1: 99.7396 (80.6795) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1699 data: 0.0771 max mem: 3534 +Test: [ 880/2387] eta: 0:05:05 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1753 (0.4003) acc1: 99.7396 (80.8888) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1777 data: 0.0857 max mem: 3534 +Test: [ 890/2387] eta: 0:05:02 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1620 (0.3974) acc1: 99.7396 (81.0968) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1901 data: 0.0971 max mem: 3534 +Test: [ 900/2387] eta: 0:04:59 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1633 (0.4012) acc1: 95.5729 (80.8315) acc5: 100.0000 (100.0000) acc1_neg: 51.823 acc5_neg: 100.000 time: 0.1716 data: 0.0776 max mem: 3535 +Test: [ 910/2387] eta: 0:04:57 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.8696 (0.4063) acc1: 47.3958 (80.4639) acc5: 100.0000 (100.0000) acc1_neg: 47.656 acc5_neg: 100.000 time: 0.1668 data: 0.0747 max mem: 3535 +Test: [ 920/2387] eta: 0:04:55 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.8795 (0.4115) acc1: 47.3958 (80.1048) acc5: 100.0000 (100.0000) acc1_neg: 43.750 acc5_neg: 100.000 time: 0.1932 data: 0.1019 max mem: 3535 +Test: [ 930/2387] eta: 0:04:54 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.8724 (0.4163) acc1: 48.4375 (79.7619) acc5: 100.0000 (100.0000) acc1_neg: 47.135 acc5_neg: 100.000 time: 0.2326 data: 0.1384 max mem: 3535 +Test: [ 940/2387] eta: 0:04:52 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.8613 (0.4212) acc1: 48.6979 (79.4296) acc5: 100.0000 (100.0000) acc1_neg: 43.750 acc5_neg: 100.000 time: 0.2347 data: 0.1373 max mem: 3535 +Test: [ 950/2387] eta: 0:04:50 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.8571 (0.4258) acc1: 48.6979 (79.0914) acc5: 100.0000 (100.0000) acc1_neg: 49.740 acc5_neg: 100.000 time: 0.1972 data: 0.1031 max mem: 3535 +Test: [ 960/2387] eta: 0:04:47 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.8519 (0.4303) acc1: 47.6562 (78.7651) acc5: 100.0000 (100.0000) acc1_neg: 47.917 acc5_neg: 100.000 time: 0.1757 data: 0.0853 max mem: 3535 +Test: [ 970/2387] eta: 0:04:45 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.8605 (0.4349) acc1: 46.8750 (78.4329) acc5: 100.0000 (100.0000) acc1_neg: 46.094 acc5_neg: 100.000 time: 0.1740 data: 0.0816 max mem: 3535 +Test: [ 980/2387] eta: 0:04:43 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.8270 (0.4341) acc1: 48.9583 (78.4674) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1749 data: 0.0820 max mem: 3535 +Test: [ 990/2387] eta: 0:04:40 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1213 (0.4311) acc1: 99.4792 (78.6718) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1803 data: 0.0878 max mem: 3535 +Test: [1000/2387] eta: 0:04:38 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1182 (0.4281) acc1: 99.7396 (78.8688) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1714 data: 0.0795 max mem: 3535 +Test: [1010/2387] eta: 0:04:35 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1184 (0.4251) acc1: 99.4792 (79.0678) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1658 data: 0.0754 max mem: 3535 +Test: [1020/2387] eta: 0:04:33 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1219 (0.4223) acc1: 99.4792 (79.2526) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1727 data: 0.0818 max mem: 3535 +Test: [1030/2387] eta: 0:04:30 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1240 (0.4195) acc1: 99.4792 (79.4374) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1600 data: 0.0681 max mem: 3535 +Test: [1040/2387] eta: 0:04:28 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1071 (0.4165) acc1: 100.0000 (79.6347) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1565 data: 0.0639 max mem: 3535 +Test: [1050/2387] eta: 0:04:25 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0999 (0.4135) acc1: 100.0000 (79.8280) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1633 data: 0.0716 max mem: 3535 +Test: [1060/2387] eta: 0:04:23 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1047 (0.4106) acc1: 100.0000 (80.0171) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1583 data: 0.0664 max mem: 3535 +Test: [1070/2387] eta: 0:04:20 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1076 (0.4079) acc1: 100.0000 (80.2008) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1539 data: 0.0619 max mem: 3535 +Test: [1080/2387] eta: 0:04:18 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1206 (0.4054) acc1: 99.7396 (80.3712) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1656 data: 0.0735 max mem: 3535 +Test: [1090/2387] eta: 0:04:16 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1206 (0.4028) acc1: 99.4792 (80.5442) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1769 data: 0.0848 max mem: 3535 +Test: [1100/2387] eta: 0:04:13 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1190 (0.4002) acc1: 99.7396 (80.7169) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1709 data: 0.0780 max mem: 3535 +Test: [1110/2387] eta: 0:04:11 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1256 (0.3980) acc1: 99.7396 (80.8588) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1755 data: 0.0808 max mem: 3535 +Test: [1120/2387] eta: 0:04:09 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1240 (0.3956) acc1: 99.7396 (81.0237) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1632 data: 0.0706 max mem: 3536 +Test: [1130/2387] eta: 0:04:07 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1190 (0.3932) acc1: 99.7396 (81.1888) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1702 data: 0.0787 max mem: 3536 +Test: [1140/2387] eta: 0:04:05 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1162 (0.3908) acc1: 99.7396 (81.3486) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1899 data: 0.0950 max mem: 3536 +Test: [1150/2387] eta: 0:04:02 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1198 (0.3888) acc1: 99.7396 (81.4950) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1750 data: 0.0802 max mem: 3536 +Test: [1160/2387] eta: 0:04:00 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1275 (0.3865) acc1: 100.0000 (81.6529) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1768 data: 0.0855 max mem: 3536 +Test: [1170/2387] eta: 0:03:58 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1204 (0.3842) acc1: 100.0000 (81.8078) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1993 data: 0.1086 max mem: 3536 +Test: [1180/2387] eta: 0:03:56 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1292 (0.3822) acc1: 100.0000 (81.9556) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1987 data: 0.1080 max mem: 3536 +Test: [1190/2387] eta: 0:03:54 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1260 (0.3801) acc1: 100.0000 (82.1065) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1897 data: 0.0988 max mem: 3536 +Test: [1200/2387] eta: 0:03:52 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1123 (0.3778) acc1: 100.0000 (82.2552) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1947 data: 0.1034 max mem: 3536 +Test: [1210/2387] eta: 0:03:50 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1059 (0.3756) acc1: 100.0000 (82.4018) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1925 data: 0.0993 max mem: 3536 +Test: [1220/2387] eta: 0:03:48 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0949 (0.3732) acc1: 100.0000 (82.5455) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1853 data: 0.0924 max mem: 3536 +Test: [1230/2387] eta: 0:03:47 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0803 (0.3708) acc1: 100.0000 (82.6873) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2026 data: 0.1103 max mem: 3536 +Test: [1240/2387] eta: 0:03:45 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0816 (0.3685) acc1: 100.0000 (82.8266) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2203 data: 0.1279 max mem: 3536 +Test: [1250/2387] eta: 0:03:43 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0851 (0.3662) acc1: 100.0000 (82.9634) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2127 data: 0.1214 max mem: 3536 +Test: [1260/2387] eta: 0:03:41 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0819 (0.3640) acc1: 100.0000 (83.0985) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2136 data: 0.1216 max mem: 3536 +Test: [1270/2387] eta: 0:03:40 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0790 (0.3617) acc1: 100.0000 (83.2315) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2343 data: 0.1405 max mem: 3536 +Test: [1280/2387] eta: 0:03:38 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0827 (0.3596) acc1: 100.0000 (83.3616) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2132 data: 0.1183 max mem: 3536 +Test: [1290/2387] eta: 0:03:36 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0963 (0.3577) acc1: 99.7396 (83.4784) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1957 data: 0.1010 max mem: 3536 +Test: [1300/2387] eta: 0:03:33 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0981 (0.3557) acc1: 99.7396 (83.6010) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1880 data: 0.0950 max mem: 3536 +Test: [1310/2387] eta: 0:03:31 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1008 (0.3539) acc1: 100.0000 (83.7207) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1779 data: 0.0863 max mem: 3536 +Test: [1320/2387] eta: 0:03:29 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1048 (0.3520) acc1: 100.0000 (83.8356) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1747 data: 0.0833 max mem: 3536 +Test: [1330/2387] eta: 0:03:27 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1110 (0.3502) acc1: 99.7396 (83.9518) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1745 data: 0.0834 max mem: 3536 +Test: [1340/2387] eta: 0:03:25 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1225 (0.3490) acc1: 98.6979 (84.0305) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2035 data: 0.1105 max mem: 3536 +Test: [1350/2387] eta: 0:03:23 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0548 (0.3468) acc1: 100.0000 (84.1487) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1961 data: 0.1032 max mem: 3537 +Test: [1360/2387] eta: 0:03:21 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0459 (0.3446) acc1: 100.0000 (84.2652) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1859 data: 0.0934 max mem: 3537 +Test: [1370/2387] eta: 0:03:19 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0455 (0.3424) acc1: 100.0000 (84.3799) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1953 data: 0.1031 max mem: 3537 +Test: [1380/2387] eta: 0:03:17 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0462 (0.3403) acc1: 100.0000 (84.4930) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1923 data: 0.1016 max mem: 3537 +Test: [1390/2387] eta: 0:03:15 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0474 (0.3382) acc1: 100.0000 (84.6045) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1869 data: 0.0939 max mem: 3537 +Test: [1400/2387] eta: 0:03:13 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0485 (0.3361) acc1: 100.0000 (84.7144) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1817 data: 0.0889 max mem: 3537 +Test: [1410/2387] eta: 0:03:11 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0492 (0.3344) acc1: 100.0000 (84.8203) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1768 data: 0.0862 max mem: 3537 +Test: [1420/2387] eta: 0:03:09 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1044 (0.3330) acc1: 100.0000 (84.9070) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1857 data: 0.0941 max mem: 3537 +Test: [1430/2387] eta: 0:03:07 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1102 (0.3314) acc1: 100.0000 (85.0119) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1826 data: 0.0915 max mem: 3537 +Test: [1440/2387] eta: 0:03:05 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1060 (0.3298) acc1: 100.0000 (85.1147) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1798 data: 0.0897 max mem: 3537 +Test: [1450/2387] eta: 0:03:03 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1058 (0.3284) acc1: 100.0000 (85.2097) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1818 data: 0.0912 max mem: 3537 +Test: [1460/2387] eta: 0:03:01 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1006 (0.3268) acc1: 100.0000 (85.3094) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1740 data: 0.0829 max mem: 3537 +Test: [1470/2387] eta: 0:02:59 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0870 (0.3251) acc1: 100.0000 (85.4092) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1715 data: 0.0805 max mem: 3537 +Test: [1480/2387] eta: 0:02:56 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0859 (0.3235) acc1: 100.0000 (85.5076) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1665 data: 0.0740 max mem: 3537 +Test: [1490/2387] eta: 0:02:54 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0910 (0.3220) acc1: 100.0000 (85.6046) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1696 data: 0.0767 max mem: 3537 +Test: [1500/2387] eta: 0:02:52 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0954 (0.3205) acc1: 100.0000 (85.6995) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1718 data: 0.0777 max mem: 3537 +Test: [1510/2387] eta: 0:02:50 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0999 (0.3190) acc1: 100.0000 (85.7938) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1653 data: 0.0721 max mem: 3537 +Test: [1520/2387] eta: 0:02:48 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1117 (0.3180) acc1: 100.0000 (85.8808) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1699 data: 0.0792 max mem: 3537 +Test: [1530/2387] eta: 0:02:46 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1509 (0.3169) acc1: 99.2188 (85.9674) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1712 data: 0.0779 max mem: 3537 +Test: [1540/2387] eta: 0:02:44 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1528 (0.3159) acc1: 99.4792 (86.0555) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1904 data: 0.0967 max mem: 3537 +Test: [1550/2387] eta: 0:02:42 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1578 (0.3149) acc1: 99.7396 (86.1430) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1872 data: 0.0941 max mem: 3537 +Test: [1560/2387] eta: 0:02:40 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1578 (0.3139) acc1: 99.7396 (86.2288) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1652 data: 0.0721 max mem: 3537 +Test: [1570/2387] eta: 0:02:38 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1491 (0.3128) acc1: 99.7396 (86.3138) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1919 data: 0.1004 max mem: 3537 +Test: [1580/2387] eta: 0:02:36 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1164 (0.3115) acc1: 100.0000 (86.3997) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2159 data: 0.1251 max mem: 3538 +Test: [1590/2387] eta: 0:02:34 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0835 (0.3100) acc1: 100.0000 (86.4850) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2028 data: 0.1121 max mem: 3538 +Test: [1600/2387] eta: 0:02:33 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0811 (0.3088) acc1: 100.0000 (86.5491) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2002 data: 0.1087 max mem: 3538 +Test: [1610/2387] eta: 0:02:31 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0941 (0.3075) acc1: 100.0000 (86.6321) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2047 data: 0.1129 max mem: 3538 +Test: [1620/2387] eta: 0:02:29 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0943 (0.3062) acc1: 100.0000 (86.7130) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1945 data: 0.1017 max mem: 3538 +Test: [1630/2387] eta: 0:02:27 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0973 (0.3050) acc1: 99.7396 (86.7885) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1916 data: 0.0997 max mem: 3538 +Test: [1640/2387] eta: 0:02:25 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1054 (0.3038) acc1: 99.4792 (86.8635) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1786 data: 0.0882 max mem: 3538 +Test: [1650/2387] eta: 0:02:23 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1034 (0.3026) acc1: 99.4792 (86.9385) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1681 data: 0.0768 max mem: 3538 +Test: [1660/2387] eta: 0:02:20 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1027 (0.3014) acc1: 99.7396 (87.0149) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1666 data: 0.0751 max mem: 3538 +Test: [1670/2387] eta: 0:02:18 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1018 (0.3002) acc1: 99.7396 (87.0879) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1732 data: 0.0810 max mem: 3538 +Test: [1680/2387] eta: 0:02:16 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1003 (0.2991) acc1: 99.7396 (87.1561) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1782 data: 0.0850 max mem: 3538 +Test: [1690/2387] eta: 0:02:14 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1021 (0.2980) acc1: 99.7396 (87.2231) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1731 data: 0.0816 max mem: 3538 +Test: [1700/2387] eta: 0:02:12 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1055 (0.2969) acc1: 99.7396 (87.2956) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1708 data: 0.0788 max mem: 3538 +Test: [1710/2387] eta: 0:02:10 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0972 (0.2957) acc1: 100.0000 (87.3694) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1691 data: 0.0746 max mem: 3538 +Test: [1720/2387] eta: 0:02:08 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0920 (0.2946) acc1: 100.0000 (87.4425) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1659 data: 0.0733 max mem: 3538 +Test: [1730/2387] eta: 0:02:06 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0917 (0.2934) acc1: 100.0000 (87.5140) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1851 data: 0.0909 max mem: 3538 +Test: [1740/2387] eta: 0:02:05 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0917 (0.2923) acc1: 100.0000 (87.5845) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2025 data: 0.1048 max mem: 3538 +Test: [1750/2387] eta: 0:02:03 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0916 (0.2912) acc1: 100.0000 (87.6554) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2098 data: 0.1147 max mem: 3538 +Test: [1760/2387] eta: 0:02:01 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0894 (0.2900) acc1: 100.0000 (87.7255) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2174 data: 0.1247 max mem: 3538 +Test: [1770/2387] eta: 0:01:59 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0905 (0.2889) acc1: 100.0000 (87.7935) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2094 data: 0.1175 max mem: 3538 +Test: [1780/2387] eta: 0:01:57 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0938 (0.2878) acc1: 100.0000 (87.8616) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1985 data: 0.1069 max mem: 3538 +Test: [1790/2387] eta: 0:01:55 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0839 (0.2867) acc1: 100.0000 (87.9289) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1996 data: 0.1058 max mem: 3538 +Test: [1800/2387] eta: 0:01:53 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0838 (0.2856) acc1: 100.0000 (87.9948) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1967 data: 0.1030 max mem: 3538 +Test: [1810/2387] eta: 0:01:51 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0862 (0.2845) acc1: 100.0000 (88.0611) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1821 data: 0.0903 max mem: 3539 +Test: [1820/2387] eta: 0:01:49 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0856 (0.2834) acc1: 100.0000 (88.1267) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1927 data: 0.1012 max mem: 3539 +Test: [1830/2387] eta: 0:01:47 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0846 (0.2824) acc1: 100.0000 (88.1915) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1897 data: 0.0989 max mem: 3539 +Test: [1840/2387] eta: 0:01:45 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1028 (0.2814) acc1: 100.0000 (88.2556) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1867 data: 0.0964 max mem: 3539 +Test: [1850/2387] eta: 0:01:44 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1049 (0.2804) acc1: 100.0000 (88.3188) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2178 data: 0.1259 max mem: 3539 +Test: [1860/2387] eta: 0:01:42 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1046 (0.2795) acc1: 100.0000 (88.3814) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2094 data: 0.1162 max mem: 3539 +Test: [1870/2387] eta: 0:01:40 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1040 (0.2786) acc1: 100.0000 (88.4433) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1655 data: 0.0739 max mem: 3539 +Test: [1880/2387] eta: 0:01:38 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1038 (0.2776) acc1: 100.0000 (88.5047) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1659 data: 0.0752 max mem: 3539 +Test: [1890/2387] eta: 0:01:36 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1041 (0.2767) acc1: 100.0000 (88.5655) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1804 data: 0.0894 max mem: 3539 +Test: [1900/2387] eta: 0:01:34 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1047 (0.2758) acc1: 100.0000 (88.6255) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1823 data: 0.0896 max mem: 3539 +Test: [1910/2387] eta: 0:01:32 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1117 (0.2750) acc1: 100.0000 (88.6850) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1751 data: 0.0819 max mem: 3539 +Test: [1920/2387] eta: 0:01:30 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1200 (0.2742) acc1: 100.0000 (88.7438) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1683 data: 0.0762 max mem: 3539 +Test: [1930/2387] eta: 0:01:28 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1064 (0.2733) acc1: 100.0000 (88.8021) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1857 data: 0.0924 max mem: 3539 +Test: [1940/2387] eta: 0:01:26 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1005 (0.2727) acc1: 100.0000 (88.8254) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1810 data: 0.0870 max mem: 3539 +Test: [1950/2387] eta: 0:01:24 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1150 (0.2719) acc1: 100.0000 (88.8826) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1726 data: 0.0792 max mem: 3539 +Test: [1960/2387] eta: 0:01:22 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1238 (0.2712) acc1: 100.0000 (88.9350) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1798 data: 0.0878 max mem: 3539 +Test: [1970/2387] eta: 0:01:20 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1395 (0.2707) acc1: 100.0000 (88.9815) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1852 data: 0.0949 max mem: 3539 +Test: [1980/2387] eta: 0:01:18 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1790 (0.2703) acc1: 97.9167 (89.0175) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1974 data: 0.1066 max mem: 3539 +Test: [1990/2387] eta: 0:01:16 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.2127 (0.2701) acc1: 95.3125 (89.0486) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1854 data: 0.0945 max mem: 3539 +Test: [2000/2387] eta: 0:01:14 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.2261 (0.2699) acc1: 95.0521 (89.0779) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1725 data: 0.0823 max mem: 3539 +Test: [2010/2387] eta: 0:01:12 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.2295 (0.2697) acc1: 94.7917 (89.1055) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1769 data: 0.0863 max mem: 3539 +Test: [2020/2387] eta: 0:01:10 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.2281 (0.2695) acc1: 94.7917 (89.1348) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1797 data: 0.0890 max mem: 3539 +Test: [2030/2387] eta: 0:01:08 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.2281 (0.2693) acc1: 94.5312 (89.1607) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1985 data: 0.1078 max mem: 3540 +Test: [2040/2387] eta: 0:01:06 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.2286 (0.2691) acc1: 94.5312 (89.1896) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2225 data: 0.1298 max mem: 3540 +Test: [2050/2387] eta: 0:01:04 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.2201 (0.2688) acc1: 95.8333 (89.2229) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2044 data: 0.1104 max mem: 3540 +Test: [2060/2387] eta: 0:01:03 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.2114 (0.2685) acc1: 96.0938 (89.2565) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1856 data: 0.0934 max mem: 3540 +Test: [2070/2387] eta: 0:01:01 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.2057 (0.2682) acc1: 96.0938 (89.2900) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1927 data: 0.1010 max mem: 3540 +Test: [2080/2387] eta: 0:00:59 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1982 (0.2679) acc1: 96.3542 (89.3240) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1982 data: 0.1063 max mem: 3540 +Test: [2090/2387] eta: 0:00:57 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1982 (0.2676) acc1: 96.3542 (89.3589) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1960 data: 0.1038 max mem: 3540 +Test: [2100/2387] eta: 0:00:55 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.2011 (0.2673) acc1: 96.3542 (89.3923) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1884 data: 0.0961 max mem: 3540 +Test: [2110/2387] eta: 0:00:53 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.2005 (0.2670) acc1: 96.3542 (89.4259) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1911 data: 0.1002 max mem: 3540 +Test: [2120/2387] eta: 0:00:51 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1777 (0.2663) acc1: 97.3958 (89.4722) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2005 data: 0.1097 max mem: 3540 +Test: [2130/2387] eta: 0:00:49 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0972 (0.2654) acc1: 100.0000 (89.5215) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2045 data: 0.1102 max mem: 3540 +Test: [2140/2387] eta: 0:00:47 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0965 (0.2649) acc1: 100.0000 (89.5533) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1991 data: 0.1005 max mem: 3540 +Test: [2150/2387] eta: 0:00:45 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0982 (0.2641) acc1: 100.0000 (89.6015) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1862 data: 0.0914 max mem: 3540 +Test: [2160/2387] eta: 0:00:43 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1002 (0.2633) acc1: 100.0000 (89.6476) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0833 max mem: 3540 +Test: [2170/2387] eta: 0:00:41 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1057 (0.2627) acc1: 100.0000 (89.6927) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1789 data: 0.0867 max mem: 3540 +Test: [2180/2387] eta: 0:00:39 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1242 (0.2627) acc1: 98.9583 (89.6957) acc5: 100.0000 (100.0000) acc1_neg: 77.344 acc5_neg: 100.000 time: 0.1751 data: 0.0829 max mem: 3540 +Test: [2190/2387] eta: 0:00:37 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.3325 (0.2631) acc1: 85.4167 (89.6737) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1638 data: 0.0714 max mem: 3540 +Test: [2200/2387] eta: 0:00:35 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1121 (0.2623) acc1: 100.0000 (89.7206) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1766 data: 0.0849 max mem: 3540 +Test: [2210/2387] eta: 0:00:34 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0844 (0.2615) acc1: 100.0000 (89.7665) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1981 data: 0.1073 max mem: 3540 +Test: [2220/2387] eta: 0:00:32 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0946 (0.2608) acc1: 100.0000 (89.8113) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1997 data: 0.1091 max mem: 3540 +Test: [2230/2387] eta: 0:00:30 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0946 (0.2600) acc1: 100.0000 (89.8564) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2045 data: 0.1133 max mem: 3540 +Test: [2240/2387] eta: 0:00:28 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0870 (0.2593) acc1: 100.0000 (89.9014) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2132 data: 0.1190 max mem: 3540 +Test: [2250/2387] eta: 0:00:26 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0961 (0.2586) acc1: 100.0000 (89.9443) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1943 data: 0.0978 max mem: 3540 +Test: [2260/2387] eta: 0:00:24 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1053 (0.2579) acc1: 100.0000 (89.9885) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1961 data: 0.1023 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0817 (0.2571) acc1: 100.0000 (90.0326) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2031 data: 0.1119 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0756 (0.2565) acc1: 100.0000 (90.0598) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1755 data: 0.0813 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0826 (0.2558) acc1: 100.0000 (90.1029) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1871 data: 0.0927 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0838 (0.2551) acc1: 100.0000 (90.1439) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2138 data: 0.1224 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0879 (0.2544) acc1: 100.0000 (90.1846) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2241 data: 0.1315 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1026 (0.2538) acc1: 100.0000 (90.2213) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2098 data: 0.1168 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1041 (0.2532) acc1: 100.0000 (90.2615) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1696 data: 0.0773 max mem: 3541 +Test: [2340/2387] eta: 0:00:09 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1071 (0.2526) acc1: 100.0000 (90.3021) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1705 data: 0.0787 max mem: 3541 +Test: [2350/2387] eta: 0:00:07 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1047 (0.2522) acc1: 100.0000 (90.3226) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1893 data: 0.0973 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1047 (0.2516) acc1: 100.0000 (90.3636) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1947 data: 0.1026 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1182 (0.2512) acc1: 100.0000 (90.4020) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0988 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1289 (0.2507) acc1: 100.0000 (90.4393) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1717 data: 0.0813 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1522 (0.2508) acc1: 98.9583 (90.4398) acc5: 100.0000 (100.0000) acc1_neg: 86.047 acc5_neg: 100.000 time: 0.1373 data: 0.0497 max mem: 3541 +Test: Total time: 0:07:39 (0.1924 s / it) +* Overall Acc@1 90.440 Acc@5 100.000 + target == real Acc@1 3.766 Acc@5 100.000 + target == fake Acc@1 96.527 Acc@5 100.000 + loss 0.251 + PR-AUC 0.9865 ROC-AUC 0.8288 EER 23.66% (thr=0.8540) +Accuracy of the network on the 916267 test images: 90.4% +[rank0]:[W725 18:13:08.649349852 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:13:27.005000 603097 torch/distributed/run.py:766] +W0725 18:13:27.005000 603097 torch/distributed/run.py:766] ***************************************** +W0725 18:13:27.005000 603097 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:13:27.005000 603097 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:13:58.565615448 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:13:58.901888825 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:13:58.066066143 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:13:58.088701787 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:13:59.512119306 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:13:59.512122851 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:13:59.529141513 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:13:59.545648227 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n16_seed3_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed3_resampled_n116-041-153_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 0:59:57 acc1_pos: 1.823 acc5_pos: 100.000 loss: 1.4775 (1.4775) acc1: 1.8229 (1.8229) acc5: 100.0000 (100.0000) time: 6.9590 data: 5.5673 max mem: 3531 +Test: [ 10/517] eta: 0:06:02 acc1_pos: 0.521 acc5_pos: 100.000 loss: 1.6182 (1.5914) acc1: 1.5625 (3.0303) acc5: 100.0000 (100.0000) time: 0.7146 data: 0.5065 max mem: 3531 +Test: [ 20/517] eta: 0:03:29 acc1_pos: 0.521 acc5_pos: 100.000 loss: 1.5616 (1.5452) acc1: 1.8229 (3.5838) acc5: 100.0000 (100.0000) time: 0.0942 data: 0.0005 max mem: 3531 +Test: [ 30/517] eta: 0:03:07 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.5616 (1.5961) acc1: 1.8229 (2.8562) acc5: 100.0000 (100.0000) time: 0.2040 data: 0.1097 max mem: 3531 +Test: [ 40/517] eta: 0:03:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.6748 (1.6270) acc1: 0.2604 (2.2866) acc5: 100.0000 (100.0000) time: 0.3449 data: 0.2535 max mem: 3531 +Test: [ 50/517] eta: 0:02:50 acc1_pos: 3.670 acc5_pos: 100.000 loss: 1.6145 (1.5271) acc1: 2.0833 (7.2815) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.3356 data: 0.2424 max mem: 3531 +Test: [ 60/517] eta: 0:02:37 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.2954 (1.3173) acc1: 97.6562 (22.3062) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2655 data: 0.1722 max mem: 3531 +Test: [ 70/517] eta: 0:02:26 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.2529 (1.1622) acc1: 99.4792 (33.2123) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2293 data: 0.1362 max mem: 3531 +Test: [ 80/517] eta: 0:02:19 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.2569 (1.0531) acc1: 100.0000 (41.2648) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2443 data: 0.1505 max mem: 3531 +Test: [ 90/517] eta: 0:02:13 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.2161 (0.9600) acc1: 100.0000 (47.7192) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2599 data: 0.1659 max mem: 3531 +Test: [100/517] eta: 0:02:09 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1988 (0.8831) acc1: 100.0000 (52.8904) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2749 data: 0.1794 max mem: 3531 +Test: [110/517] eta: 0:02:04 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1751 (0.8189) acc1: 100.0000 (57.1040) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2699 data: 0.1734 max mem: 3531 +Test: [120/517] eta: 0:01:59 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.2169 (0.7705) acc1: 100.0000 (60.6426) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2449 data: 0.1474 max mem: 3531 +Test: [130/517] eta: 0:01:54 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.2340 (0.7294) acc1: 100.0000 (63.6371) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2449 data: 0.1481 max mem: 3531 +Test: [140/517] eta: 0:01:50 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.2110 (0.6920) acc1: 100.0000 (66.2142) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2449 data: 0.1505 max mem: 3531 +Test: [150/517] eta: 0:01:46 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1940 (0.6586) acc1: 100.0000 (68.4413) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2549 data: 0.1581 max mem: 3531 +Test: [160/517] eta: 0:01:42 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.2039 (0.6312) acc1: 100.0000 (70.4015) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2599 data: 0.1615 max mem: 3531 +Test: [170/517] eta: 0:01:39 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.2120 (0.6065) acc1: 100.0000 (72.1248) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2499 data: 0.1540 max mem: 3531 +Test: [180/517] eta: 0:01:35 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.2097 (0.5850) acc1: 99.7396 (73.6490) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2549 data: 0.1600 max mem: 3531 +Test: [190/517] eta: 0:01:32 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1825 (0.5629) acc1: 100.0000 (75.0273) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2649 data: 0.1697 max mem: 3531 +Test: [200/517] eta: 0:01:29 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1632 (0.5431) acc1: 100.0000 (76.2684) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1604 max mem: 3531 +Test: [210/517] eta: 0:01:26 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1897 (0.5290) acc1: 100.0000 (77.3697) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2499 data: 0.1561 max mem: 3532 +Test: [220/517] eta: 0:01:26 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.2504 (0.5164) acc1: 99.4792 (78.3595) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.4049 data: 0.3088 max mem: 3532 +Test: [230/517] eta: 0:01:21 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.2566 (0.5053) acc1: 99.4792 (79.2749) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.3299 data: 0.2308 max mem: 3532 +Test: [240/517] eta: 0:01:18 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.2376 (0.4937) acc1: 99.7396 (80.1316) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1716 data: 0.0746 max mem: 3532 +Test: [250/517] eta: 0:01:15 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.2251 (0.4832) acc1: 99.7396 (80.9087) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2504 data: 0.1581 max mem: 3532 +Test: [260/517] eta: 0:01:12 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.2251 (0.4734) acc1: 99.7396 (81.6292) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2504 data: 0.1592 max mem: 3532 +Test: [270/517] eta: 0:01:08 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.2446 (0.4654) acc1: 99.7396 (82.2859) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2455 data: 0.1544 max mem: 3532 +Test: [280/517] eta: 0:01:05 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.2490 (0.4577) acc1: 99.4792 (82.8412) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2477 data: 0.1533 max mem: 3532 +Test: [290/517] eta: 0:01:02 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.2529 (0.4506) acc1: 99.2188 (83.3852) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2495 data: 0.1533 max mem: 3532 +Test: [300/517] eta: 0:00:59 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.2260 (0.4428) acc1: 99.7396 (83.9355) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2417 data: 0.1488 max mem: 3532 +Test: [310/517] eta: 0:00:56 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.2518 (0.4371) acc1: 100.0000 (84.4412) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2441 data: 0.1521 max mem: 3532 +Test: [320/517] eta: 0:00:54 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.2562 (0.4305) acc1: 100.0000 (84.9161) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2547 data: 0.1615 max mem: 3532 +Test: [330/517] eta: 0:00:51 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1923 (0.4232) acc1: 100.0000 (85.3718) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2849 data: 0.1911 max mem: 3532 +Test: [340/517] eta: 0:00:48 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1967 (0.4167) acc1: 100.0000 (85.7985) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2933 data: 0.1982 max mem: 3532 +Test: [350/517] eta: 0:00:45 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.2041 (0.4111) acc1: 100.0000 (86.1994) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2700 data: 0.1742 max mem: 3532 +Test: [360/517] eta: 0:00:43 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.2232 (0.4054) acc1: 100.0000 (86.5774) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2599 data: 0.1656 max mem: 3532 +Test: [370/517] eta: 0:00:40 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1470 (0.3978) acc1: 100.0000 (86.9392) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2697 data: 0.1759 max mem: 3532 +Test: [380/517] eta: 0:00:37 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1256 (0.3908) acc1: 100.0000 (87.2820) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2899 data: 0.1953 max mem: 3532 +Test: [390/517] eta: 0:00:35 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1314 (0.3842) acc1: 100.0000 (87.6072) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.3162 data: 0.2196 max mem: 3532 +Test: [400/517] eta: 0:00:32 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1389 (0.3781) acc1: 100.0000 (87.9163) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.3299 data: 0.2348 max mem: 3532 +Test: [410/517] eta: 0:00:29 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1437 (0.3725) acc1: 100.0000 (88.2103) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2836 data: 0.1895 max mem: 3532 +Test: [420/517] eta: 0:00:27 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1556 (0.3686) acc1: 100.0000 (88.4903) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.3349 data: 0.2395 max mem: 3532 +Test: [430/517] eta: 0:00:24 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.2159 (0.3652) acc1: 100.0000 (88.7562) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.3207 data: 0.2243 max mem: 3532 +Test: [440/517] eta: 0:00:21 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.2089 (0.3615) acc1: 100.0000 (89.0046) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2399 data: 0.1430 max mem: 3533 +Test: [450/517] eta: 0:00:18 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1931 (0.3577) acc1: 100.0000 (89.2479) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2641 data: 0.1666 max mem: 3533 +Test: [460/517] eta: 0:00:15 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1784 (0.3538) acc1: 100.0000 (89.4811) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1574 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1692 (0.3497) acc1: 100.0000 (89.7044) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2289 data: 0.1341 max mem: 3533 +Test: [480/517] eta: 0:00:10 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1664 (0.3461) acc1: 100.0000 (89.9185) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2078 data: 0.1162 max mem: 3533 +Test: [490/517] eta: 0:00:07 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1877 (0.3449) acc1: 100.0000 (90.0898) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1843 data: 0.0941 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.2631 (0.3434) acc1: 99.4792 (90.2830) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1624 data: 0.0722 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.2294 (0.3407) acc1: 99.7396 (90.4670) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1582 data: 0.0688 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.2073 (0.3391) acc1: 100.0000 (90.5713) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1424 data: 0.0513 max mem: 3533 +Test: Total time: 0:02:18 (0.2673 s / it) +* Overall Acc@1 90.571 Acc@5 100.000 + target == real Acc@1 2.692 Acc@5 100.000 + target == fake Acc@1 99.688 Acc@5 100.000 + loss 0.339 + PR-AUC 0.9309 ROC-AUC 0.6133 EER 42.16% (thr=0.8174) +Accuracy of the network on the 198427 test images: 90.6% +[rank0]:[W725 18:16:25.402382636 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0725 18:16:46.180000 603687 torch/distributed/run.py:766] +W0725 18:16:46.180000 603687 torch/distributed/run.py:766] ***************************************** +W0725 18:16:46.180000 603687 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:16:46.180000 603687 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W725 18:17:14.648140524 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:17:14.031379382 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:17:14.415740053 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:17:14.465902158 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:17:15.569516372 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +| distributed init (rank 2): env:// +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:17:15.817461478 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W725 18:17:15.817477253 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank4]:[W725 18:17:15.829676641 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n16_seed3_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed3_resampled_n116-041-153_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:41:46 acc1_pos: 94.271 acc5_pos: 100.000 loss: 0.4498 (0.4498) acc1: 94.2708 (94.2708) acc5: 100.0000 (100.0000) time: 6.4427 data: 4.8126 max mem: 3531 +Test: [ 10/389] eta: 0:04:13 acc1_pos: 90.625 acc5_pos: 100.000 loss: 0.4390 (0.4341) acc1: 93.4896 (85.2746) acc5: 100.0000 (100.0000) time: 0.6677 data: 0.4379 max mem: 3531 +Test: [ 20/389] eta: 0:02:26 acc1_pos: 86.198 acc5_pos: 100.000 loss: 0.4718 (0.4795) acc1: 80.9896 (83.3953) acc5: 100.0000 (100.0000) time: 0.0940 data: 0.0004 max mem: 3531 +Test: [ 30/389] eta: 0:02:06 acc1_pos: 80.729 acc5_pos: 100.000 loss: 0.5031 (0.4613) acc1: 86.1979 (85.9795) acc5: 100.0000 (100.0000) time: 0.1789 data: 0.0807 max mem: 3531 +Test: [ 40/389] eta: 0:01:58 acc1_pos: 70.312 acc5_pos: 100.000 loss: 0.4504 (0.4678) acc1: 87.7604 (85.7978) acc5: 100.0000 (100.0000) time: 0.2799 data: 0.1845 max mem: 3531 +Test: [ 50/389] eta: 0:01:45 acc1_pos: 97.396 acc5_pos: 100.000 loss: 0.4304 (0.4517) acc1: 92.4479 (87.2804) acc5: 100.0000 (100.0000) time: 0.2498 data: 0.1572 max mem: 3531 +Test: [ 60/389] eta: 0:01:34 acc1_pos: 77.604 acc5_pos: 100.000 loss: 0.3038 (0.4243) acc1: 96.6146 (88.7509) acc5: 100.0000 (100.0000) time: 0.1815 data: 0.0866 max mem: 3531 +Test: [ 70/389] eta: 0:01:27 acc1_pos: 92.188 acc5_pos: 100.000 loss: 0.3038 (0.4199) acc1: 96.6146 (89.0808) acc5: 100.0000 (100.0000) time: 0.1749 data: 0.0780 max mem: 3531 +Test: [ 80/389] eta: 0:01:20 acc1_pos: 99.740 acc5_pos: 100.000 loss: 0.4398 (0.4239) acc1: 91.6667 (89.0336) acc5: 100.0000 (100.0000) time: 0.1833 data: 0.0850 max mem: 3531 +Test: [ 90/389] eta: 0:01:14 acc1_pos: 81.771 acc5_pos: 100.000 loss: 0.2825 (0.4102) acc1: 92.4479 (89.7779) acc5: 100.0000 (100.0000) time: 0.1631 data: 0.0660 max mem: 3531 +Test: [100/389] eta: 0:01:10 acc1_pos: 85.938 acc5_pos: 100.000 loss: 0.3916 (0.4177) acc1: 94.0104 (89.2404) acc5: 100.0000 (100.0000) time: 0.1649 data: 0.0686 max mem: 3531 +Test: [110/389] eta: 0:01:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4568 (0.4226) acc1: 82.8125 (88.0185) acc5: 100.0000 (100.0000) time: 0.1668 data: 0.0700 max mem: 3531 +Test: [120/389] eta: 0:01:01 acc1_pos: 95.833 acc5_pos: 100.000 loss: 0.3984 (0.4223) acc1: 89.0625 (87.6765) acc5: 100.0000 (100.0000) time: 0.1464 data: 0.0527 max mem: 3531 +Test: [130/389] eta: 0:00:57 acc1_pos: 99.386 acc5_pos: 100.000 loss: 0.5033 (0.4506) acc1: 77.3438 (84.7209) acc5: 100.0000 (100.0000) acc1_neg: 44.010 acc5_neg: 100.000 time: 0.1485 data: 0.0559 max mem: 3531 +Test: [140/389] eta: 0:00:54 acc1_pos: 99.386 acc5_pos: 100.000 loss: 0.9294 (0.4985) acc1: 16.6667 (79.5915) acc5: 100.0000 (100.0000) acc1_neg: 0.260 acc5_neg: 100.000 time: 0.1661 data: 0.0726 max mem: 3531 +Test: [150/389] eta: 0:00:50 acc1_pos: 99.386 acc5_pos: 100.000 loss: 0.9770 (0.5285) acc1: 12.2396 (76.2848) acc5: 100.0000 (100.0000) acc1_neg: 68.229 acc5_neg: 100.000 time: 0.1614 data: 0.0693 max mem: 3531 +Test: [160/389] eta: 0:00:48 acc1_pos: 99.386 acc5_pos: 100.000 loss: 0.9561 (0.5447) acc1: 16.4062 (74.6733) acc5: 100.0000 (100.0000) acc1_neg: 79.427 acc5_neg: 100.000 time: 0.1522 data: 0.0590 max mem: 3531 +Test: [170/389] eta: 0:00:45 acc1_pos: 99.386 acc5_pos: 100.000 loss: 0.6506 (0.5583) acc1: 58.0729 (73.1375) acc5: 100.0000 (100.0000) acc1_neg: 30.208 acc5_neg: 100.000 time: 0.1572 data: 0.0636 max mem: 3531 +Test: [180/389] eta: 0:00:42 acc1_pos: 99.386 acc5_pos: 100.000 loss: 0.7275 (0.5716) acc1: 58.0729 (71.6894) acc5: 100.0000 (100.0000) acc1_neg: 51.042 acc5_neg: 100.000 time: 0.1549 data: 0.0628 max mem: 3531 +Test: [190/389] eta: 0:00:40 acc1_pos: 99.386 acc5_pos: 100.000 loss: 0.8797 (0.5917) acc1: 31.2500 (69.4958) acc5: 100.0000 (100.0000) acc1_neg: 0.260 acc5_neg: 100.000 time: 0.1522 data: 0.0583 max mem: 3531 +Test: [200/389] eta: 0:00:37 acc1_pos: 99.386 acc5_pos: 100.000 loss: 0.9424 (0.6111) acc1: 14.8438 (67.1137) acc5: 100.0000 (100.0000) acc1_neg: 13.542 acc5_neg: 100.000 time: 0.1599 data: 0.0646 max mem: 3531 +Test: [210/389] eta: 0:00:35 acc1_pos: 99.386 acc5_pos: 100.000 loss: 0.9141 (0.6171) acc1: 14.8438 (66.0483) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1661 data: 0.0725 max mem: 3532 +Test: [220/389] eta: 0:00:32 acc1_pos: 99.386 acc5_pos: 100.000 loss: 0.6891 (0.6139) acc1: 46.6146 (66.2648) acc5: 100.0000 (100.0000) acc1_neg: 21.875 acc5_neg: 100.000 time: 0.1515 data: 0.0603 max mem: 3532 +Test: [230/389] eta: 0:00:30 acc1_pos: 99.386 acc5_pos: 100.000 loss: 0.7108 (0.6236) acc1: 45.8333 (65.0794) acc5: 100.0000 (100.0000) acc1_neg: 5.729 acc5_neg: 100.000 time: 0.1610 data: 0.0702 max mem: 3532 +Test: [240/389] eta: 0:00:28 acc1_pos: 99.386 acc5_pos: 100.000 loss: 0.8185 (0.6327) acc1: 35.6771 (63.9274) acc5: 100.0000 (100.0000) acc1_neg: 49.740 acc5_neg: 100.000 time: 0.1819 data: 0.0908 max mem: 3532 +Test: [250/389] eta: 0:00:26 acc1_pos: 99.386 acc5_pos: 100.000 loss: 0.6985 (0.6313) acc1: 49.7396 (64.2472) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.1776 data: 0.0864 max mem: 3532 +Test: [260/389] eta: 0:00:24 acc1_pos: 99.386 acc5_pos: 100.000 loss: 0.6422 (0.6401) acc1: 49.2188 (63.0318) acc5: 100.0000 (100.0000) acc1_neg: 1.562 acc5_neg: 100.000 time: 0.1730 data: 0.0815 max mem: 3532 +Test: [270/389] eta: 0:00:22 acc1_pos: 99.386 acc5_pos: 100.000 loss: 0.8959 (0.6566) acc1: 12.2396 (61.2517) acc5: 100.0000 (100.0000) acc1_neg: 35.156 acc5_neg: 100.000 time: 0.1831 data: 0.0917 max mem: 3532 +Test: [280/389] eta: 0:00:20 acc1_pos: 99.386 acc5_pos: 100.000 loss: 0.9612 (0.6696) acc1: 21.0938 (59.9626) acc5: 100.0000 (100.0000) acc1_neg: 77.344 acc5_neg: 100.000 time: 0.1756 data: 0.0841 max mem: 3532 +Test: [290/389] eta: 0:00:18 acc1_pos: 99.386 acc5_pos: 100.000 loss: 0.8938 (0.6754) acc1: 26.0417 (59.1611) acc5: 100.0000 (100.0000) acc1_neg: 63.542 acc5_neg: 100.000 time: 0.1744 data: 0.0832 max mem: 3532 +Test: [300/389] eta: 0:00:17 acc1_pos: 99.386 acc5_pos: 100.000 loss: 0.8020 (0.6818) acc1: 27.6042 (58.0115) acc5: 100.0000 (100.0000) acc1_neg: 20.052 acc5_neg: 100.000 time: 0.1935 data: 0.0993 max mem: 3532 +Test: [310/389] eta: 0:00:15 acc1_pos: 99.386 acc5_pos: 100.000 loss: 0.8986 (0.6923) acc1: 21.8750 (56.8077) acc5: 100.0000 (100.0000) acc1_neg: 2.344 acc5_neg: 100.000 time: 0.1777 data: 0.0823 max mem: 3532 +Test: [320/389] eta: 0:00:13 acc1_pos: 99.386 acc5_pos: 100.000 loss: 0.9164 (0.6996) acc1: 19.2708 (55.7316) acc5: 100.0000 (100.0000) acc1_neg: 52.083 acc5_neg: 100.000 time: 0.1693 data: 0.0758 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 99.386 acc5_pos: 100.000 loss: 1.1686 (0.7137) acc1: 3.1250 (54.2123) acc5: 100.0000 (100.0000) acc1_neg: 1.042 acc5_neg: 100.000 time: 0.1872 data: 0.0937 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 99.386 acc5_pos: 100.000 loss: 0.8103 (0.7110) acc1: 38.0208 (54.7669) acc5: 100.0000 (100.0000) acc1_neg: 64.323 acc5_neg: 100.000 time: 0.1970 data: 0.1046 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 99.386 acc5_pos: 100.000 loss: 0.6765 (0.7240) acc1: 59.8958 (53.9471) acc5: 100.0000 (100.0000) acc1_neg: 0.781 acc5_neg: 100.000 time: 0.1914 data: 0.1001 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 99.386 acc5_pos: 100.000 loss: 1.1576 (0.7388) acc1: 21.3542 (52.9353) acc5: 100.0000 (100.0000) acc1_neg: 21.875 acc5_neg: 100.000 time: 0.1749 data: 0.0841 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 99.386 acc5_pos: 100.000 loss: 1.1235 (0.7407) acc1: 22.3958 (52.7860) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1704 data: 0.0793 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 99.386 acc5_pos: 100.000 loss: 0.7589 (0.7424) acc1: 45.3125 (52.6288) acc5: 100.0000 (100.0000) acc1_neg: 42.188 acc5_neg: 100.000 time: 0.1500 data: 0.0595 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 99.386 acc5_pos: 100.000 loss: 0.7219 (0.7427) acc1: 46.3542 (52.5255) acc5: 100.0000 (100.0000) acc1_neg: 23.224 acc5_neg: 100.000 time: 0.1303 data: 0.0369 max mem: 3532 +Test: Total time: 0:01:12 (0.1869 s / it) +* Overall Acc@1 52.525 Acc@5 100.000 + target == real Acc@1 87.678 Acc@5 100.000 + target == fake Acc@1 36.186 Acc@5 100.000 + loss 0.743 + PR-AUC 0.8257 ROC-AUC 0.7054 EER 35.31% (thr=0.3694) +Accuracy of the network on the 149358 test images: 52.5% +[rank0]:[W725 18:18:38.020381929 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:18:58.042000 604255 torch/distributed/run.py:766] +W0725 18:18:58.042000 604255 torch/distributed/run.py:766] ***************************************** +W0725 18:18:58.042000 604255 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:18:58.042000 604255 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:19:26.123053370 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:19:26.165035568 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:19:27.610177515 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:19:27.693267482 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:19:27.743281865 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +NCCL version 2.26.2+cuda12.2 +[rank7]:[W725 18:19:27.801872305 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:19:27.926612239 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W725 18:19:27.926784394 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n16_seed3_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed3_resampled_n116-041-153_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:33:55 acc1_pos: 59.375 acc5_pos: 100.000 loss: 0.7723 (0.7723) acc1: 59.3750 (59.3750) acc5: 100.0000 (100.0000) time: 6.1128 data: 4.4810 max mem: 3531 +Test: [ 10/333] eta: 0:03:26 acc1_pos: 79.427 acc5_pos: 100.000 loss: 0.5596 (0.5786) acc1: 69.2708 (69.6023) acc5: 100.0000 (100.0000) time: 0.6393 data: 0.4078 max mem: 3531 +Test: [ 20/333] eta: 0:01:59 acc1_pos: 91.406 acc5_pos: 100.000 loss: 0.5472 (0.5315) acc1: 73.4375 (73.0159) acc5: 100.0000 (100.0000) time: 0.0959 data: 0.0004 max mem: 3531 +Test: [ 30/333] eta: 0:01:35 acc1_pos: 39.583 acc5_pos: 100.000 loss: 0.6345 (0.6726) acc1: 65.1042 (61.6683) acc5: 100.0000 (100.0000) time: 0.1354 data: 0.0389 max mem: 3531 +Test: [ 40/333] eta: 0:01:26 acc1_pos: 32.552 acc5_pos: 100.000 loss: 0.9762 (0.7486) acc1: 34.3750 (55.0940) acc5: 100.0000 (100.0000) time: 0.2007 data: 0.1081 max mem: 3531 +Test: [ 50/333] eta: 0:01:18 acc1_pos: 35.938 acc5_pos: 100.000 loss: 0.9934 (0.8023) acc1: 34.3750 (50.7200) acc5: 100.0000 (100.0000) time: 0.2194 data: 0.1255 max mem: 3531 +Test: [ 60/333] eta: 0:01:09 acc1_pos: 28.906 acc5_pos: 100.000 loss: 1.0476 (0.8409) acc1: 30.2083 (47.4983) acc5: 100.0000 (100.0000) time: 0.1741 data: 0.0777 max mem: 3531 +Test: [ 70/333] eta: 0:01:03 acc1_pos: 28.906 acc5_pos: 100.000 loss: 1.0285 (0.8664) acc1: 28.9062 (45.2942) acc5: 100.0000 (100.0000) time: 0.1499 data: 0.0538 max mem: 3531 +Test: [ 80/333] eta: 0:00:59 acc1_pos: 40.365 acc5_pos: 100.000 loss: 0.9750 (0.8796) acc1: 32.5521 (44.0008) acc5: 100.0000 (100.0000) time: 0.1749 data: 0.0778 max mem: 3531 +Test: [ 90/333] eta: 0:00:56 acc1_pos: 47.917 acc5_pos: 100.000 loss: 0.9442 (0.8845) acc1: 34.8958 (43.1490) acc5: 100.0000 (100.0000) time: 0.1999 data: 0.1037 max mem: 3531 +Test: [100/333] eta: 0:00:52 acc1_pos: 40.885 acc5_pos: 100.000 loss: 0.8914 (0.8877) acc1: 37.2396 (42.6929) acc5: 100.0000 (100.0000) time: 0.1921 data: 0.0979 max mem: 3531 +Test: [110/333] eta: 0:00:49 acc1_pos: 35.677 acc5_pos: 100.000 loss: 0.8914 (0.8935) acc1: 38.0208 (42.1030) acc5: 100.0000 (100.0000) time: 0.1749 data: 0.0813 max mem: 3531 +Test: [120/333] eta: 0:00:47 acc1_pos: 40.885 acc5_pos: 100.000 loss: 0.8870 (0.8943) acc1: 37.5000 (41.8496) acc5: 100.0000 (100.0000) time: 0.1928 data: 0.0982 max mem: 3531 +Test: [130/333] eta: 0:00:44 acc1_pos: 47.656 acc5_pos: 100.000 loss: 0.8534 (0.8926) acc1: 40.8854 (41.7700) acc5: 100.0000 (100.0000) time: 0.2149 data: 0.1170 max mem: 3531 +Test: [140/333] eta: 0:00:42 acc1_pos: 63.802 acc5_pos: 100.000 loss: 0.8572 (0.8907) acc1: 40.8854 (41.6279) acc5: 100.0000 (100.0000) time: 0.2158 data: 0.1197 max mem: 3531 +Test: [150/333] eta: 0:00:40 acc1_pos: 39.062 acc5_pos: 100.000 loss: 0.8885 (0.8923) acc1: 39.0625 (41.4908) acc5: 100.0000 (100.0000) time: 0.2156 data: 0.1226 max mem: 3531 +Test: [160/333] eta: 0:00:37 acc1_pos: 31.908 acc5_pos: 100.000 loss: 0.8747 (0.8891) acc1: 40.6250 (41.8171) acc5: 100.0000 (100.0000) acc1_neg: 55.990 acc5_neg: 100.000 time: 0.1790 data: 0.0836 max mem: 3531 +Test: [170/333] eta: 0:00:34 acc1_pos: 31.908 acc5_pos: 100.000 loss: 0.7592 (0.8747) acc1: 57.2917 (43.2459) acc5: 100.0000 (100.0000) acc1_neg: 71.354 acc5_neg: 100.000 time: 0.1597 data: 0.0655 max mem: 3531 +Test: [180/333] eta: 0:00:32 acc1_pos: 31.908 acc5_pos: 100.000 loss: 0.6187 (0.8599) acc1: 67.1875 (44.6636) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1801 data: 0.0896 max mem: 3531 +Test: [190/333] eta: 0:00:30 acc1_pos: 31.908 acc5_pos: 100.000 loss: 0.5106 (0.8381) acc1: 76.3021 (46.6964) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.1854 data: 0.0950 max mem: 3531 +Test: [200/333] eta: 0:00:27 acc1_pos: 31.908 acc5_pos: 100.000 loss: 0.4592 (0.8225) acc1: 80.9896 (48.1434) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.1950 data: 0.1049 max mem: 3531 +Test: [210/333] eta: 0:00:25 acc1_pos: 31.908 acc5_pos: 100.000 loss: 0.5879 (0.8119) acc1: 70.3125 (49.1299) acc5: 100.0000 (100.0000) acc1_neg: 60.417 acc5_neg: 100.000 time: 0.1953 data: 0.1041 max mem: 3532 +Test: [220/333] eta: 0:00:23 acc1_pos: 31.908 acc5_pos: 100.000 loss: 0.5908 (0.8073) acc1: 70.0521 (49.6701) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.1922 data: 0.0990 max mem: 3532 +Test: [230/333] eta: 0:00:21 acc1_pos: 31.908 acc5_pos: 100.000 loss: 0.5243 (0.7932) acc1: 76.3021 (51.0372) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.1936 data: 0.0994 max mem: 3532 +Test: [240/333] eta: 0:00:19 acc1_pos: 31.908 acc5_pos: 100.000 loss: 0.4911 (0.7806) acc1: 80.9896 (52.1860) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.1910 data: 0.0955 max mem: 3532 +Test: [250/333] eta: 0:00:17 acc1_pos: 31.908 acc5_pos: 100.000 loss: 0.4918 (0.7704) acc1: 78.6458 (53.1717) acc5: 100.0000 (100.0000) acc1_neg: 70.312 acc5_neg: 100.000 time: 0.1878 data: 0.0908 max mem: 3532 +Test: [260/333] eta: 0:00:14 acc1_pos: 31.908 acc5_pos: 100.000 loss: 0.5294 (0.7616) acc1: 75.5208 (53.9831) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.1912 data: 0.0946 max mem: 3532 +Test: [270/333] eta: 0:00:12 acc1_pos: 31.908 acc5_pos: 100.000 loss: 0.5510 (0.7543) acc1: 73.9583 (54.6721) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.1893 data: 0.0955 max mem: 3532 +Test: [280/333] eta: 0:00:10 acc1_pos: 31.908 acc5_pos: 100.000 loss: 0.5528 (0.7480) acc1: 72.1354 (55.2380) acc5: 100.0000 (100.0000) acc1_neg: 75.781 acc5_neg: 100.000 time: 0.1924 data: 0.1002 max mem: 3532 +Test: [290/333] eta: 0:00:08 acc1_pos: 31.908 acc5_pos: 100.000 loss: 0.6220 (0.7444) acc1: 67.9688 (55.5761) acc5: 100.0000 (100.0000) acc1_neg: 59.115 acc5_neg: 100.000 time: 0.2075 data: 0.1146 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 31.908 acc5_pos: 100.000 loss: 0.6878 (0.7445) acc1: 61.1979 (55.6738) acc5: 100.0000 (100.0000) acc1_neg: 60.156 acc5_neg: 100.000 time: 0.2097 data: 0.1135 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 31.908 acc5_pos: 100.000 loss: 0.6929 (0.7374) acc1: 61.1979 (56.3388) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1948 data: 0.0987 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 31.908 acc5_pos: 100.000 loss: 0.4760 (0.7301) acc1: 79.9479 (57.0207) acc5: 100.0000 (100.0000) acc1_neg: 61.198 acc5_neg: 100.000 time: 0.1682 data: 0.0765 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 31.908 acc5_pos: 100.000 loss: 0.6781 (0.7304) acc1: 64.3229 (57.1327) acc5: 100.0000 (100.0000) acc1_neg: 64.323 acc5_neg: 100.000 time: 0.1381 data: 0.0481 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 31.908 acc5_pos: 100.000 loss: 0.7149 (0.7306) acc1: 61.1979 (57.1474) acc5: 100.0000 (100.0000) acc1_neg: 59.437 acc5_neg: 100.000 time: 0.1294 data: 0.0359 max mem: 3532 +Test: Total time: 0:01:06 (0.2004 s / it) +* Overall Acc@1 57.147 Acc@5 100.000 + target == real Acc@1 41.395 Acc@5 100.000 + target == fake Acc@1 71.170 Acc@5 100.000 + loss 0.731 + PR-AUC 0.6075 ROC-AUC 0.5935 EER 43.83% (thr=0.6060) +Accuracy of the network on the 127843 test images: 57.1% +[rank0]:[W725 18:20:44.033302045 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:21:04.684000 604814 torch/distributed/run.py:766] +W0725 18:21:04.684000 604814 torch/distributed/run.py:766] ***************************************** +W0725 18:21:04.684000 604814 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:21:04.684000 604814 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:21:34.680629024 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W725 18:21:34.683863630 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:21:34.730823518 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:21:34.155598282 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:21:34.221072059 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:21:34.221433386 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:21:34.238781343 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:21:34.265405981 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n16_seed3_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed3_resampled_n116-041-153_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:22:19 acc1_pos: 94.792 acc5_pos: 100.000 loss: 0.2265 (0.2265) acc1: 94.7917 (94.7917) acc5: 100.0000 (100.0000) time: 5.7007 data: 4.2505 max mem: 3531 +Test: [ 10/235] eta: 0:02:14 acc1_pos: 92.708 acc5_pos: 100.000 loss: 0.2443 (0.2433) acc1: 93.2292 (93.5606) acc5: 100.0000 (100.0000) time: 0.5998 data: 0.3867 max mem: 3531 +Test: [ 20/235] eta: 0:01:16 acc1_pos: 91.406 acc5_pos: 100.000 loss: 0.2390 (0.2391) acc1: 93.7500 (93.8120) acc5: 100.0000 (100.0000) time: 0.0899 data: 0.0004 max mem: 3531 +Test: [ 30/235] eta: 0:00:58 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.2315 (0.2371) acc1: 94.0104 (93.8760) acc5: 100.0000 (100.0000) time: 0.1141 data: 0.0240 max mem: 3531 +Test: [ 40/235] eta: 0:00:54 acc1_pos: 92.969 acc5_pos: 100.000 loss: 0.2299 (0.2363) acc1: 94.0104 (93.8580) acc5: 100.0000 (100.0000) time: 0.1931 data: 0.1005 max mem: 3531 +Test: [ 50/235] eta: 0:00:49 acc1_pos: 92.708 acc5_pos: 100.000 loss: 0.2359 (0.2375) acc1: 94.0104 (93.7653) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1500 max mem: 3531 +Test: [ 60/235] eta: 0:00:45 acc1_pos: 92.708 acc5_pos: 100.000 loss: 0.2400 (0.2384) acc1: 93.7500 (93.7415) acc5: 100.0000 (100.0000) time: 0.2260 data: 0.1322 max mem: 3531 +Test: [ 70/235] eta: 0:00:41 acc1_pos: 92.308 acc5_pos: 100.000 loss: 0.2549 (0.4051) acc1: 92.1875 (83.4837) acc5: 100.0000 (100.0000) acc1_neg: 10.938 acc5_neg: 100.000 time: 0.1990 data: 0.1069 max mem: 3531 +Test: [ 80/235] eta: 0:00:38 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.5712 (0.5509) acc1: 11.4583 (74.5531) acc5: 100.0000 (100.0000) acc1_neg: 9.115 acc5_neg: 100.000 time: 0.2053 data: 0.1117 max mem: 3531 +Test: [ 90/235] eta: 0:00:35 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.5860 (0.6640) acc1: 11.7188 (67.6998) acc5: 100.0000 (100.0000) acc1_neg: 9.896 acc5_neg: 100.000 time: 0.2199 data: 0.1253 max mem: 3531 +Test: [100/235] eta: 0:00:32 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.5861 (0.7550) acc1: 11.9792 (62.2009) acc5: 100.0000 (100.0000) acc1_neg: 10.677 acc5_neg: 100.000 time: 0.2098 data: 0.1151 max mem: 3531 +Test: [110/235] eta: 0:00:29 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.5810 (0.8291) acc1: 12.2396 (57.7327) acc5: 100.0000 (100.0000) acc1_neg: 12.240 acc5_neg: 100.000 time: 0.1805 data: 0.0855 max mem: 3531 +Test: [120/235] eta: 0:00:26 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.5300 (0.8773) acc1: 13.8021 (54.6552) acc5: 100.0000 (100.0000) acc1_neg: 23.698 acc5_neg: 100.000 time: 0.1736 data: 0.0776 max mem: 3531 +Test: [130/235] eta: 0:00:23 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.3793 (0.9158) acc1: 21.3542 (52.1092) acc5: 100.0000 (100.0000) acc1_neg: 19.271 acc5_neg: 100.000 time: 0.1921 data: 0.0943 max mem: 3531 +Test: [140/235] eta: 0:00:21 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.3970 (0.9514) acc1: 21.3542 (49.9003) acc5: 100.0000 (100.0000) acc1_neg: 19.010 acc5_neg: 100.000 time: 0.1957 data: 0.1002 max mem: 3531 +Test: [150/235] eta: 0:00:18 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.3970 (0.9801) acc1: 20.8333 (48.0270) acc5: 100.0000 (100.0000) acc1_neg: 25.000 acc5_neg: 100.000 time: 0.2003 data: 0.1073 max mem: 3531 +Test: [160/235] eta: 0:00:16 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.3813 (1.0055) acc1: 20.8333 (46.3364) acc5: 100.0000 (100.0000) acc1_neg: 20.573 acc5_neg: 100.000 time: 0.2054 data: 0.1114 max mem: 3531 +Test: [170/235] eta: 0:00:14 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.4242 (1.0445) acc1: 17.1875 (44.1810) acc5: 100.0000 (100.0000) acc1_neg: 6.510 acc5_neg: 100.000 time: 0.2036 data: 0.1098 max mem: 3531 +Test: [180/235] eta: 0:00:12 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.6939 (1.0815) acc1: 8.8542 (42.2292) acc5: 100.0000 (100.0000) acc1_neg: 7.292 acc5_neg: 100.000 time: 0.1948 data: 0.0983 max mem: 3531 +Test: [190/235] eta: 0:00:09 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.6939 (1.1140) acc1: 8.5938 (40.4450) acc5: 100.0000 (100.0000) acc1_neg: 10.156 acc5_neg: 100.000 time: 0.1999 data: 0.1040 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.7078 (1.1437) acc1: 8.5938 (38.8630) acc5: 100.0000 (100.0000) acc1_neg: 8.854 acc5_neg: 100.000 time: 0.2204 data: 0.1293 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.7095 (1.1699) acc1: 8.5938 (37.4334) acc5: 100.0000 (100.0000) acc1_neg: 9.375 acc5_neg: 100.000 time: 0.1989 data: 0.1071 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.6942 (1.1938) acc1: 8.8542 (36.1590) acc5: 100.0000 (100.0000) acc1_neg: 9.635 acc5_neg: 100.000 time: 0.1521 data: 0.0609 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.6915 (1.2149) acc1: 8.8542 (34.9928) acc5: 100.0000 (100.0000) acc1_neg: 8.594 acc5_neg: 100.000 time: 0.1381 data: 0.0474 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.7026 (1.2242) acc1: 8.5938 (34.6432) acc5: 100.0000 (100.0000) acc1_neg: 8.696 acc5_neg: 100.000 time: 0.1264 data: 0.0389 max mem: 3532 +Test: Total time: 0:00:48 (0.2084 s / it) +* Overall Acc@1 34.643 Acc@5 100.000 + target == real Acc@1 93.746 Acc@5 100.000 + target == fake Acc@1 13.267 Acc@5 100.000 + loss 1.224 + PR-AUC 0.8151 ROC-AUC 0.6393 EER 39.79% (thr=0.1667) +Accuracy of the network on the 89879 test images: 34.6% +[rank0]:[W725 18:22:34.968376899 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:22:55.299000 605384 torch/distributed/run.py:766] +W0725 18:22:55.299000 605384 torch/distributed/run.py:766] ***************************************** +W0725 18:22:55.299000 605384 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:22:55.299000 605384 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W725 18:23:23.614512620 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:23:23.810344534 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:23:23.946704129 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:23:23.408180753 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:23:24.620708331 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 6): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:23:24.653356046 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W725 18:23:24.653950108 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank6]:[W725 18:23:24.654054651 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n16_seed3_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed3_resampled_n116-041-153_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:21:21 acc1_pos: 58.073 acc5_pos: 100.000 loss: 0.6858 (0.6858) acc1: 58.0729 (58.0729) acc5: 100.0000 (100.0000) time: 4.7637 data: 2.8364 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/269] eta: 0:02:25 acc1_pos: 82.552 acc5_pos: 100.000 loss: 0.6335 (0.5986) acc1: 66.1458 (68.5843) acc5: 100.0000 (100.0000) time: 0.5605 data: 0.3039 max mem: 3531 +Test: [ 20/269] eta: 0:01:25 acc1_pos: 74.740 acc5_pos: 100.000 loss: 0.5327 (0.5660) acc1: 73.4375 (71.9494) acc5: 100.0000 (100.0000) time: 0.1244 data: 0.0342 max mem: 3531 +Test: [ 30/269] eta: 0:01:03 acc1_pos: 72.396 acc5_pos: 100.000 loss: 0.5327 (0.5625) acc1: 73.4375 (72.3370) acc5: 100.0000 (100.0000) time: 0.1033 data: 0.0099 max mem: 3531 +Test: [ 40/269] eta: 0:00:59 acc1_pos: 70.690 acc5_pos: 100.000 loss: 0.6193 (0.5861) acc1: 67.1875 (70.0521) acc5: 100.0000 (100.0000) acc1_neg: 52.083 acc5_neg: 100.000 time: 0.1737 data: 0.0785 max mem: 3531 +Test: [ 50/269] eta: 0:00:55 acc1_pos: 70.690 acc5_pos: 100.000 loss: 0.6574 (0.6141) acc1: 54.6875 (66.4726) acc5: 100.0000 (100.0000) acc1_neg: 41.667 acc5_neg: 100.000 time: 0.2402 data: 0.1474 max mem: 3531 +Test: [ 60/269] eta: 0:00:52 acc1_pos: 70.690 acc5_pos: 100.000 loss: 0.7557 (0.6491) acc1: 45.5729 (62.8031) acc5: 100.0000 (100.0000) acc1_neg: 21.615 acc5_neg: 100.000 time: 0.2342 data: 0.1422 max mem: 3531 +Test: [ 70/269] eta: 0:00:49 acc1_pos: 70.690 acc5_pos: 100.000 loss: 0.8473 (0.6791) acc1: 38.5417 (59.5070) acc5: 100.0000 (100.0000) acc1_neg: 38.542 acc5_neg: 100.000 time: 0.2237 data: 0.1317 max mem: 3531 +Test: [ 80/269] eta: 0:00:45 acc1_pos: 70.690 acc5_pos: 100.000 loss: 0.8242 (0.6914) acc1: 38.5417 (57.7514) acc5: 100.0000 (100.0000) acc1_neg: 48.698 acc5_neg: 100.000 time: 0.2049 data: 0.1123 max mem: 3531 +Test: [ 90/269] eta: 0:00:41 acc1_pos: 70.690 acc5_pos: 100.000 loss: 0.7525 (0.7043) acc1: 46.3542 (56.4990) acc5: 100.0000 (100.0000) acc1_neg: 48.698 acc5_neg: 100.000 time: 0.1853 data: 0.0922 max mem: 3531 +Test: [100/269] eta: 0:00:38 acc1_pos: 70.690 acc5_pos: 100.000 loss: 0.8196 (0.7201) acc1: 46.8750 (55.4868) acc5: 100.0000 (100.0000) acc1_neg: 46.094 acc5_neg: 100.000 time: 0.1768 data: 0.0839 max mem: 3531 +Test: [110/269] eta: 0:00:34 acc1_pos: 70.690 acc5_pos: 100.000 loss: 0.8497 (0.7283) acc1: 48.6979 (55.1145) acc5: 100.0000 (100.0000) acc1_neg: 39.323 acc5_neg: 100.000 time: 0.1600 data: 0.0680 max mem: 3531 +Test: [120/269] eta: 0:00:32 acc1_pos: 70.690 acc5_pos: 100.000 loss: 0.7575 (0.7247) acc1: 53.1250 (55.3870) acc5: 100.0000 (100.0000) acc1_neg: 64.844 acc5_neg: 100.000 time: 0.1741 data: 0.0804 max mem: 3531 +Test: [130/269] eta: 0:00:30 acc1_pos: 70.690 acc5_pos: 100.000 loss: 0.7263 (0.7330) acc1: 51.5625 (54.5205) acc5: 100.0000 (100.0000) acc1_neg: 30.469 acc5_neg: 100.000 time: 0.2394 data: 0.1432 max mem: 3531 +Test: [140/269] eta: 0:00:28 acc1_pos: 70.690 acc5_pos: 100.000 loss: 0.9317 (0.7493) acc1: 33.3333 (52.5968) acc5: 100.0000 (100.0000) acc1_neg: 19.010 acc5_neg: 100.000 time: 0.2287 data: 0.1337 max mem: 3531 +Test: [150/269] eta: 0:00:25 acc1_pos: 70.690 acc5_pos: 100.000 loss: 0.9315 (0.7592) acc1: 28.1250 (51.4263) acc5: 100.0000 (100.0000) acc1_neg: 50.781 acc5_neg: 100.000 time: 0.1608 data: 0.0661 max mem: 3531 +Test: [160/269] eta: 0:00:23 acc1_pos: 70.690 acc5_pos: 100.000 loss: 0.8321 (0.7626) acc1: 42.9688 (51.1792) acc5: 100.0000 (100.0000) acc1_neg: 43.750 acc5_neg: 100.000 time: 0.1777 data: 0.0845 max mem: 3531 +Test: [170/269] eta: 0:00:21 acc1_pos: 70.690 acc5_pos: 100.000 loss: 0.8225 (0.7682) acc1: 43.7500 (50.6335) acc5: 100.0000 (100.0000) acc1_neg: 29.948 acc5_neg: 100.000 time: 0.2090 data: 0.1135 max mem: 3531 +Test: [180/269] eta: 0:00:18 acc1_pos: 70.690 acc5_pos: 100.000 loss: 0.7622 (0.7648) acc1: 49.7396 (51.0992) acc5: 100.0000 (100.0000) acc1_neg: 43.490 acc5_neg: 100.000 time: 0.1974 data: 0.1014 max mem: 3531 +Test: [190/269] eta: 0:00:16 acc1_pos: 70.690 acc5_pos: 100.000 loss: 0.7263 (0.7660) acc1: 53.9062 (51.0035) acc5: 100.0000 (100.0000) acc1_neg: 53.906 acc5_neg: 100.000 time: 0.2054 data: 0.1140 max mem: 3531 +Test: [200/269] eta: 0:00:14 acc1_pos: 70.690 acc5_pos: 100.000 loss: 0.7838 (0.7671) acc1: 46.0938 (50.9069) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.1975 data: 0.1060 max mem: 3531 +Test: [210/269] eta: 0:00:12 acc1_pos: 70.690 acc5_pos: 100.000 loss: 0.7736 (0.7657) acc1: 45.5729 (50.8343) acc5: 100.0000 (100.0000) acc1_neg: 43.750 acc5_neg: 100.000 time: 0.1646 data: 0.0732 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 70.690 acc5_pos: 100.000 loss: 0.7611 (0.7642) acc1: 45.5729 (50.8413) acc5: 100.0000 (100.0000) acc1_neg: 79.427 acc5_neg: 100.000 time: 0.1621 data: 0.0690 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 70.690 acc5_pos: 100.000 loss: 0.6838 (0.7600) acc1: 55.4688 (51.1803) acc5: 100.0000 (100.0000) acc1_neg: 64.844 acc5_neg: 100.000 time: 0.1848 data: 0.0915 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 70.690 acc5_pos: 100.000 loss: 0.7323 (0.7652) acc1: 51.5625 (50.8536) acc5: 100.0000 (100.0000) acc1_neg: 42.448 acc5_neg: 100.000 time: 0.2115 data: 0.1196 max mem: 3532 +Test: [250/269] eta: 0:00:03 acc1_pos: 70.690 acc5_pos: 100.000 loss: 0.8588 (0.7707) acc1: 40.1042 (50.3673) acc5: 100.0000 (100.0000) acc1_neg: 25.260 acc5_neg: 100.000 time: 0.2098 data: 0.1155 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 70.690 acc5_pos: 100.000 loss: 0.9133 (0.7772) acc1: 35.6771 (49.6119) acc5: 100.0000 (100.0000) acc1_neg: 46.094 acc5_neg: 100.000 time: 0.1432 data: 0.0457 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 70.690 acc5_pos: 100.000 loss: 0.8585 (0.7769) acc1: 36.1979 (49.5242) acc5: 100.0000 (100.0000) acc1_neg: 66.929 acc5_neg: 100.000 time: 0.1060 data: 0.0046 max mem: 3532 +Test: Total time: 0:00:54 (0.2009 s / it) +* Overall Acc@1 49.524 Acc@5 100.000 + target == real Acc@1 72.162 Acc@5 100.000 + target == fake Acc@1 46.435 Acc@5 100.000 + loss 0.777 + PR-AUC 0.9216 ROC-AUC 0.6310 EER 40.49% (thr=0.4314) +Accuracy of the network on the 103293 test images: 49.5% +[rank0]:[W725 18:24:29.240429144 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-25T18:24:37Z] eval_done tag=fixed500k_fake_n16_seed3_resampled +[2026-07-25T18:24:37Z] eval_start tag=fixed500k_real_n8_seed4_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed4_resampled_n116-041-153_20260725_112559/best_checkpoint.pth +started_utc=2026-07-25T18:24:38Z +hostname=n116-041-153 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed4_resampled_n116-041-153_20260725_112559/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed4_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0725 18:24:50.992000 605969 torch/distributed/run.py:766] +W0725 18:24:50.992000 605969 torch/distributed/run.py:766] ***************************************** +W0725 18:24:50.992000 605969 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:24:50.992000 605969 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:25:20.088968461 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:25:21.547972305 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:25:21.741438565 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +NCCL version 2.26.2+cuda12.2 +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:25:21.961898371 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +[rank1]:[W725 18:25:21.073851155 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank4]:[W725 18:25:21.073850963 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:25:21.080913662 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:25:21.117464638 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed4_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed4_resampled_n116-041-153_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:22:25 acc1_pos: 82.031 acc5_pos: 100.000 loss: 0.4147 (0.4147) acc1: 82.0312 (82.0312) acc5: 100.0000 (100.0000) time: 5.7488 data: 4.1547 max mem: 3531 +Test: [ 10/234] eta: 0:02:16 acc1_pos: 60.417 acc5_pos: 100.000 loss: 0.5280 (0.5682) acc1: 72.3958 (69.1051) acc5: 100.0000 (100.0000) time: 0.6082 data: 0.3786 max mem: 3531 +Test: [ 20/234] eta: 0:01:20 acc1_pos: 52.344 acc5_pos: 100.000 loss: 0.6433 (0.6761) acc1: 60.4167 (62.0908) acc5: 100.0000 (100.0000) time: 0.1070 data: 0.0109 max mem: 3531 +Test: [ 30/234] eta: 0:01:09 acc1_pos: 36.979 acc5_pos: 100.000 loss: 0.8286 (0.7293) acc1: 52.3438 (58.8374) acc5: 100.0000 (100.0000) time: 0.1920 data: 0.0962 max mem: 3531 +Test: [ 40/234] eta: 0:01:04 acc1_pos: 60.938 acc5_pos: 100.000 loss: 0.8409 (0.7963) acc1: 50.7812 (54.8590) acc5: 100.0000 (100.0000) time: 0.2879 data: 0.1968 max mem: 3531 +Test: [ 50/234] eta: 0:00:53 acc1_pos: 59.115 acc5_pos: 100.000 loss: 0.7147 (0.7716) acc1: 61.4583 (56.8730) acc5: 100.0000 (100.0000) time: 0.2179 data: 0.1265 max mem: 3531 +Test: [ 60/234] eta: 0:00:46 acc1_pos: 60.000 acc5_pos: 100.000 loss: 0.3289 (0.6616) acc1: 72.1354 (63.6697) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1273 data: 0.0325 max mem: 3531 +Test: [ 70/234] eta: 0:00:43 acc1_pos: 60.000 acc5_pos: 100.000 loss: 0.0781 (0.5984) acc1: 99.7396 (67.8550) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1901 data: 0.0968 max mem: 3531 +Test: [ 80/234] eta: 0:00:39 acc1_pos: 60.000 acc5_pos: 100.000 loss: 0.2597 (0.5553) acc1: 91.6667 (70.9652) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2396 data: 0.1460 max mem: 3531 +Test: [ 90/234] eta: 0:00:36 acc1_pos: 60.000 acc5_pos: 100.000 loss: 0.1963 (0.5101) acc1: 95.5729 (73.8353) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2147 data: 0.1199 max mem: 3531 +Test: [100/234] eta: 0:00:32 acc1_pos: 60.000 acc5_pos: 100.000 loss: 0.2135 (0.4895) acc1: 93.4896 (75.2475) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1899 data: 0.0946 max mem: 3531 +Test: [110/234] eta: 0:00:29 acc1_pos: 60.000 acc5_pos: 100.000 loss: 0.3014 (0.4705) acc1: 88.5417 (76.6000) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1757 data: 0.0808 max mem: 3531 +Test: [120/234] eta: 0:00:26 acc1_pos: 60.000 acc5_pos: 100.000 loss: 0.3194 (0.4622) acc1: 87.2396 (77.2663) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.1814 data: 0.0902 max mem: 3531 +Test: [130/234] eta: 0:00:24 acc1_pos: 60.000 acc5_pos: 100.000 loss: 0.3921 (0.4610) acc1: 83.8542 (77.4451) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.2048 data: 0.1110 max mem: 3531 +Test: [140/234] eta: 0:00:21 acc1_pos: 60.000 acc5_pos: 100.000 loss: 0.4714 (0.4604) acc1: 76.5625 (77.5931) acc5: 100.0000 (100.0000) acc1_neg: 60.417 acc5_neg: 100.000 time: 0.2134 data: 0.1177 max mem: 3531 +Test: [150/234] eta: 0:00:19 acc1_pos: 60.000 acc5_pos: 100.000 loss: 0.3644 (0.4543) acc1: 83.8542 (77.9836) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.1913 data: 0.0961 max mem: 3531 +Test: [160/234] eta: 0:00:16 acc1_pos: 60.000 acc5_pos: 100.000 loss: 0.4154 (0.4598) acc1: 78.3854 (77.6850) acc5: 100.0000 (100.0000) acc1_neg: 61.979 acc5_neg: 100.000 time: 0.1950 data: 0.0981 max mem: 3531 +Test: [170/234] eta: 0:00:14 acc1_pos: 60.000 acc5_pos: 100.000 loss: 0.6309 (0.4719) acc1: 65.8854 (76.8732) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.2080 data: 0.1081 max mem: 3531 +Test: [180/234] eta: 0:00:12 acc1_pos: 60.000 acc5_pos: 100.000 loss: 0.4990 (0.4645) acc1: 70.5729 (77.3740) acc5: 100.0000 (100.0000) acc1_neg: 70.573 acc5_neg: 100.000 time: 0.2063 data: 0.1098 max mem: 3531 +Test: [190/234] eta: 0:00:09 acc1_pos: 60.000 acc5_pos: 100.000 loss: 0.3465 (0.4586) acc1: 85.9375 (77.8141) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.2145 data: 0.1222 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 60.000 acc5_pos: 100.000 loss: 0.3280 (0.4464) acc1: 86.7188 (78.6303) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1984 data: 0.1057 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 60.000 acc5_pos: 100.000 loss: 0.1196 (0.4309) acc1: 99.7396 (79.6307) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1742 data: 0.0817 max mem: 3532 +Test: [220/234] eta: 0:00:03 acc1_pos: 60.000 acc5_pos: 100.000 loss: 0.1181 (0.4169) acc1: 99.4792 (80.4970) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1668 data: 0.0755 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 60.000 acc5_pos: 100.000 loss: 0.1705 (0.4117) acc1: 92.1875 (80.9017) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1420 data: 0.0522 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 60.000 acc5_pos: 100.000 loss: 0.2860 (0.4102) acc1: 90.3646 (80.9995) acc5: 100.0000 (100.0000) acc1_neg: 89.105 acc5_neg: 100.000 time: 0.1188 data: 0.0280 max mem: 3532 +Test: Total time: 0:00:49 (0.2125 s / it) +* Overall Acc@1 80.999 Acc@5 100.000 + target == real Acc@1 56.896 Acc@5 100.000 + target == fake Acc@1 87.793 Acc@5 100.000 + loss 0.410 + PR-AUC 0.9466 ROC-AUC 0.8406 EER 23.69% (thr=0.6851) +Accuracy of the network on the 89729 test images: 81.0% +[rank0]:[W725 18:26:18.085755808 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:26:36.624000 606538 torch/distributed/run.py:766] +W0725 18:26:36.624000 606538 torch/distributed/run.py:766] ***************************************** +W0725 18:26:36.624000 606538 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:26:36.624000 606538 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:27:04.065943307 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:27:04.124360294 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:27:05.825735193 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:27:05.853165038 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:27:05.860246760 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:27:05.895954700 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:27:05.955450817 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank1]:[W725 18:27:05.955491866 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed4_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed4_resampled_n116-041-153_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:09:35 acc1_pos: 55.208 acc5_pos: 100.000 loss: 0.7369 (0.7369) acc1: 55.2083 (55.2083) acc5: 100.0000 (100.0000) time: 6.2737 data: 4.5322 max mem: 3531 +Test: [ 10/2387] eta: 0:25:50 acc1_pos: 36.979 acc5_pos: 100.000 loss: 0.9778 (0.9490) acc1: 46.8750 (44.7443) acc5: 100.0000 (100.0000) time: 0.6521 data: 0.4125 max mem: 3531 +Test: [ 20/2387] eta: 0:15:09 acc1_pos: 60.677 acc5_pos: 100.000 loss: 0.9214 (0.9087) acc1: 46.8750 (49.0079) acc5: 100.0000 (100.0000) time: 0.0900 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:12:50 acc1_pos: 16.667 acc5_pos: 100.000 loss: 0.8208 (0.9462) acc1: 42.4479 (45.9173) acc5: 100.0000 (100.0000) time: 0.1478 data: 0.0535 max mem: 3531 +Test: [ 40/2387] eta: 0:13:06 acc1_pos: 37.240 acc5_pos: 100.000 loss: 1.0031 (0.9682) acc1: 35.6771 (43.0132) acc5: 100.0000 (100.0000) time: 0.2837 data: 0.1836 max mem: 3531 +Test: [ 50/2387] eta: 0:12:37 acc1_pos: 47.917 acc5_pos: 100.000 loss: 0.9039 (0.9406) acc1: 42.9688 (44.9397) acc5: 100.0000 (100.0000) time: 0.3199 data: 0.2232 max mem: 3531 +Test: [ 60/2387] eta: 0:12:43 acc1_pos: 37.760 acc5_pos: 100.000 loss: 1.0019 (0.9754) acc1: 39.5833 (43.1779) acc5: 100.0000 (100.0000) time: 0.3140 data: 0.2214 max mem: 3531 +Test: [ 70/2387] eta: 0:11:48 acc1_pos: 28.906 acc5_pos: 100.000 loss: 1.1243 (1.0027) acc1: 32.0312 (41.4356) acc5: 100.0000 (100.0000) time: 0.2599 data: 0.1668 max mem: 3531 +Test: [ 80/2387] eta: 0:11:32 acc1_pos: 40.625 acc5_pos: 100.000 loss: 1.0347 (0.9949) acc1: 34.3750 (41.6442) acc5: 100.0000 (100.0000) time: 0.2149 data: 0.1209 max mem: 3531 +Test: [ 90/2387] eta: 0:11:09 acc1_pos: 35.677 acc5_pos: 100.000 loss: 0.9429 (0.9891) acc1: 40.3646 (41.5636) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1454 max mem: 3531 +Test: [ 100/2387] eta: 0:10:52 acc1_pos: 50.521 acc5_pos: 100.000 loss: 0.9305 (0.9834) acc1: 42.9688 (41.8085) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1291 max mem: 3531 +Test: [ 110/2387] eta: 0:10:42 acc1_pos: 48.177 acc5_pos: 100.000 loss: 0.8393 (0.9671) acc1: 44.2708 (42.3564) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1424 max mem: 3531 +Test: [ 120/2387] eta: 0:10:28 acc1_pos: 51.562 acc5_pos: 100.000 loss: 0.8276 (0.9585) acc1: 42.4479 (42.7643) acc5: 100.0000 (100.0000) time: 0.2349 data: 0.1397 max mem: 3531 +Test: [ 130/2387] eta: 0:10:18 acc1_pos: 48.698 acc5_pos: 100.000 loss: 0.8621 (0.9520) acc1: 42.1875 (42.8614) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1351 max mem: 3531 +Test: [ 140/2387] eta: 0:10:12 acc1_pos: 59.896 acc5_pos: 100.000 loss: 0.8495 (0.9410) acc1: 45.5729 (43.2624) acc5: 100.0000 (100.0000) time: 0.2443 data: 0.1447 max mem: 3531 +Test: [ 150/2387] eta: 0:10:08 acc1_pos: 33.594 acc5_pos: 100.000 loss: 0.8008 (0.9332) acc1: 50.0000 (43.8793) acc5: 100.0000 (100.0000) time: 0.2593 data: 0.1595 max mem: 3531 +Test: [ 160/2387] eta: 0:09:57 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.6729 (0.9078) acc1: 55.7292 (45.6409) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2389 data: 0.1420 max mem: 3531 +Test: [ 170/2387] eta: 0:09:36 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2767 (0.8707) acc1: 92.1875 (48.4070) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1699 data: 0.0751 max mem: 3531 +Test: [ 180/2387] eta: 0:09:27 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2733 (0.8375) acc1: 93.4896 (50.9050) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1659 data: 0.0718 max mem: 3531 +Test: [ 190/2387] eta: 0:09:22 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2697 (0.8077) acc1: 93.7500 (53.1086) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.2199 data: 0.1249 max mem: 3531 +Test: [ 200/2387] eta: 0:09:20 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2697 (0.7808) acc1: 92.7083 (55.0904) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2499 data: 0.1540 max mem: 3531 +Test: [ 210/2387] eta: 0:09:12 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2709 (0.7569) acc1: 92.7083 (56.8683) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2303 data: 0.1342 max mem: 3532 +Test: [ 220/2387] eta: 0:09:20 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2719 (0.7348) acc1: 93.4896 (58.5325) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2845 data: 0.1887 max mem: 3532 +Test: [ 230/2387] eta: 0:09:03 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2729 (0.7148) acc1: 93.4896 (60.0153) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.2341 data: 0.1368 max mem: 3532 +Test: [ 240/2387] eta: 0:08:57 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2747 (0.6966) acc1: 93.2292 (61.4000) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1558 data: 0.0581 max mem: 3532 +Test: [ 250/2387] eta: 0:08:49 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2600 (0.6742) acc1: 95.0521 (62.8766) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2019 data: 0.1079 max mem: 3532 +Test: [ 260/2387] eta: 0:08:42 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1146 (0.6523) acc1: 99.7396 (64.2820) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1894 data: 0.0987 max mem: 3532 +Test: [ 270/2387] eta: 0:08:33 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1014 (0.6320) acc1: 99.7396 (65.5904) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1780 data: 0.0869 max mem: 3532 +Test: [ 280/2387] eta: 0:08:26 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1022 (0.6135) acc1: 99.7396 (66.7927) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1746 data: 0.0780 max mem: 3532 +Test: [ 290/2387] eta: 0:08:19 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1040 (0.5961) acc1: 100.0000 (67.9070) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1748 data: 0.0764 max mem: 3532 +Test: [ 300/2387] eta: 0:08:10 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.0934 (0.5790) acc1: 100.0000 (68.9689) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1608 data: 0.0650 max mem: 3532 +Test: [ 310/2387] eta: 0:08:06 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.0903 (0.5660) acc1: 99.7396 (69.8922) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1762 data: 0.0809 max mem: 3532 +Test: [ 320/2387] eta: 0:08:00 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2808 (0.5575) acc1: 95.3125 (70.6605) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1907 data: 0.0960 max mem: 3532 +Test: [ 330/2387] eta: 0:07:57 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2870 (0.5494) acc1: 95.0521 (71.4006) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2045 data: 0.1103 max mem: 3532 +Test: [ 340/2387] eta: 0:07:55 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2888 (0.5417) acc1: 95.3125 (72.1079) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2339 data: 0.1399 max mem: 3532 +Test: [ 350/2387] eta: 0:07:55 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2657 (0.5298) acc1: 96.6146 (72.8840) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2542 data: 0.1603 max mem: 3532 +Test: [ 360/2387] eta: 0:07:48 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1685 (0.5213) acc1: 95.8333 (73.5140) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2092 data: 0.1140 max mem: 3532 +Test: [ 370/2387] eta: 0:07:39 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2345 (0.5137) acc1: 95.0521 (74.0798) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1297 data: 0.0322 max mem: 3532 +Test: [ 380/2387] eta: 0:07:37 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2350 (0.5064) acc1: 95.0521 (74.6364) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1703 data: 0.0739 max mem: 3532 +Test: [ 390/2387] eta: 0:07:35 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2360 (0.4996) acc1: 94.5312 (75.1459) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2316 data: 0.1356 max mem: 3532 +Test: [ 400/2387] eta: 0:07:28 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2392 (0.4932) acc1: 94.5312 (75.6345) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1857 data: 0.0921 max mem: 3532 +Test: [ 410/2387] eta: 0:07:23 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2387 (0.4869) acc1: 95.3125 (76.1120) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1502 data: 0.0594 max mem: 3532 +Test: [ 420/2387] eta: 0:07:19 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2364 (0.4810) acc1: 94.7917 (76.5439) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1740 data: 0.0790 max mem: 3532 +Test: [ 430/2387] eta: 0:07:15 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2382 (0.4754) acc1: 94.7917 (76.9806) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1887 data: 0.0946 max mem: 3532 +Test: [ 440/2387] eta: 0:07:13 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2278 (0.4680) acc1: 96.3542 (77.4613) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2095 data: 0.1189 max mem: 3533 +Test: [ 450/2387] eta: 0:07:10 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.0958 (0.4597) acc1: 100.0000 (77.9610) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2149 data: 0.1225 max mem: 3533 +Test: [ 460/2387] eta: 0:07:07 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.0949 (0.4518) acc1: 100.0000 (78.4391) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2056 data: 0.1113 max mem: 3533 +Test: [ 470/2387] eta: 0:07:06 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.0953 (0.4443) acc1: 100.0000 (78.8969) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2240 data: 0.1295 max mem: 3533 +Test: [ 480/2387] eta: 0:07:02 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.0955 (0.4370) acc1: 100.0000 (79.3350) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2092 data: 0.1162 max mem: 3533 +Test: [ 490/2387] eta: 0:06:57 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.0944 (0.4301) acc1: 100.0000 (79.7554) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1625 data: 0.0706 max mem: 3533 +Test: [ 500/2387] eta: 0:06:52 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.0951 (0.4234) acc1: 100.0000 (80.1595) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1502 data: 0.0583 max mem: 3533 +Test: [ 510/2387] eta: 0:06:48 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.0958 (0.4172) acc1: 100.0000 (80.5462) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1545 data: 0.0611 max mem: 3533 +Test: [ 520/2387] eta: 0:06:43 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1277 (0.4119) acc1: 100.0000 (80.9171) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1551 data: 0.0606 max mem: 3533 +Test: [ 530/2387] eta: 0:06:38 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1581 (0.4072) acc1: 99.7396 (81.2677) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1442 data: 0.0515 max mem: 3533 +Test: [ 540/2387] eta: 0:06:35 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1615 (0.4026) acc1: 99.7396 (81.6096) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1518 data: 0.0596 max mem: 3533 +Test: [ 550/2387] eta: 0:06:31 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1615 (0.3983) acc1: 99.7396 (81.9391) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1683 data: 0.0722 max mem: 3533 +Test: [ 560/2387] eta: 0:06:27 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1630 (0.3940) acc1: 99.7396 (82.2569) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1586 data: 0.0625 max mem: 3533 +Test: [ 570/2387] eta: 0:06:24 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1603 (0.3900) acc1: 99.7396 (82.5626) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1749 data: 0.0810 max mem: 3533 +Test: [ 580/2387] eta: 0:06:23 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1612 (0.3860) acc1: 99.7396 (82.8587) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2189 data: 0.1237 max mem: 3533 +Test: [ 590/2387] eta: 0:06:22 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1619 (0.3822) acc1: 99.7396 (83.1461) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2462 data: 0.1525 max mem: 3533 +Test: [ 600/2387] eta: 0:06:18 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1478 (0.3781) acc1: 100.0000 (83.4252) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2099 data: 0.1178 max mem: 3533 +Test: [ 610/2387] eta: 0:06:13 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1478 (0.3749) acc1: 99.4792 (83.6816) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1336 data: 0.0377 max mem: 3533 +Test: [ 620/2387] eta: 0:06:10 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1849 (0.3718) acc1: 98.9583 (83.9246) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1300 data: 0.0329 max mem: 3533 +Test: [ 630/2387] eta: 0:06:07 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1849 (0.3688) acc1: 98.9583 (84.1629) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1755 data: 0.0829 max mem: 3533 +Test: [ 640/2387] eta: 0:06:04 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1867 (0.3660) acc1: 98.9583 (84.3917) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1848 data: 0.0919 max mem: 3533 +Test: [ 650/2387] eta: 0:06:01 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1858 (0.3632) acc1: 98.6979 (84.6138) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1811 data: 0.0855 max mem: 3533 +Test: [ 660/2387] eta: 0:05:58 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1848 (0.3605) acc1: 98.9583 (84.8332) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1754 data: 0.0818 max mem: 3533 +Test: [ 670/2387] eta: 0:05:55 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1827 (0.3578) acc1: 98.9583 (85.0437) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1732 data: 0.0805 max mem: 3534 +Test: [ 680/2387] eta: 0:05:52 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1775 (0.3549) acc1: 99.4792 (85.2606) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1751 data: 0.0819 max mem: 3534 +Test: [ 690/2387] eta: 0:05:50 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1730 (0.3527) acc1: 99.2188 (85.4084) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1761 data: 0.0835 max mem: 3534 +Test: [ 700/2387] eta: 0:05:47 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2261 (0.3509) acc1: 93.7500 (85.5199) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1892 data: 0.0967 max mem: 3534 +Test: [ 710/2387] eta: 0:05:45 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2281 (0.3492) acc1: 93.4896 (85.6302) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1941 data: 0.1028 max mem: 3534 +Test: [ 720/2387] eta: 0:05:43 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2242 (0.3475) acc1: 93.7500 (85.7486) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1976 data: 0.1061 max mem: 3534 +Test: [ 730/2387] eta: 0:05:40 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2242 (0.3458) acc1: 93.7500 (85.8531) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1958 data: 0.1042 max mem: 3534 +Test: [ 740/2387] eta: 0:05:38 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2227 (0.3442) acc1: 93.4896 (85.9603) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1873 data: 0.0943 max mem: 3534 +Test: [ 750/2387] eta: 0:05:36 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2267 (0.3426) acc1: 92.9688 (86.0603) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1973 data: 0.1030 max mem: 3534 +Test: [ 760/2387] eta: 0:05:33 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2271 (0.3405) acc1: 93.2292 (86.1825) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1960 data: 0.1039 max mem: 3534 +Test: [ 770/2387] eta: 0:05:31 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1363 (0.3381) acc1: 97.9167 (86.3300) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1965 data: 0.1054 max mem: 3534 +Test: [ 780/2387] eta: 0:05:29 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1444 (0.3362) acc1: 97.9167 (86.4590) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2054 data: 0.1150 max mem: 3534 +Test: [ 790/2387] eta: 0:05:27 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1613 (0.3349) acc1: 96.8750 (86.5239) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.1912 data: 0.1005 max mem: 3534 +Test: [ 800/2387] eta: 0:05:24 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1615 (0.3333) acc1: 96.8750 (86.6342) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1846 data: 0.0937 max mem: 3534 +Test: [ 810/2387] eta: 0:05:22 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1835 (0.3317) acc1: 97.3958 (86.7615) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1892 data: 0.0970 max mem: 3534 +Test: [ 820/2387] eta: 0:05:19 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1835 (0.3299) acc1: 96.8750 (86.8834) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1833 data: 0.0906 max mem: 3534 +Test: [ 830/2387] eta: 0:05:17 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1940 (0.3288) acc1: 95.5729 (86.9597) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1849 data: 0.0929 max mem: 3534 +Test: [ 840/2387] eta: 0:05:15 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3077 (0.3296) acc1: 88.2812 (86.9262) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.1854 data: 0.0934 max mem: 3534 +Test: [ 850/2387] eta: 0:05:12 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3788 (0.3299) acc1: 85.9375 (86.9327) acc5: 100.0000 (100.0000) acc1_neg: 76.042 acc5_neg: 100.000 time: 0.1799 data: 0.0890 max mem: 3534 +Test: [ 860/2387] eta: 0:05:10 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3457 (0.3300) acc1: 89.3229 (86.9695) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1829 data: 0.0926 max mem: 3534 +Test: [ 870/2387] eta: 0:05:07 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3335 (0.3302) acc1: 89.3229 (86.9756) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1744 data: 0.0841 max mem: 3534 +Test: [ 880/2387] eta: 0:05:05 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3439 (0.3305) acc1: 86.1979 (86.9703) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.1782 data: 0.0878 max mem: 3534 +Test: [ 890/2387] eta: 0:05:03 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2566 (0.3293) acc1: 94.5312 (87.0768) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1857 data: 0.0951 max mem: 3534 +Test: [ 900/2387] eta: 0:05:00 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2518 (0.3292) acc1: 91.6667 (87.1035) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1899 data: 0.0984 max mem: 3535 +Test: [ 910/2387] eta: 0:04:58 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3305 (0.3293) acc1: 88.2812 (87.1147) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1914 data: 0.0997 max mem: 3535 +Test: [ 920/2387] eta: 0:04:56 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3347 (0.3294) acc1: 87.7604 (87.1200) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.2042 data: 0.1114 max mem: 3535 +Test: [ 930/2387] eta: 0:04:54 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3406 (0.3295) acc1: 86.7188 (87.1204) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.2121 data: 0.1193 max mem: 3535 +Test: [ 940/2387] eta: 0:04:52 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3415 (0.3296) acc1: 86.7188 (87.1178) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.1923 data: 0.1009 max mem: 3535 +Test: [ 950/2387] eta: 0:04:50 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3317 (0.3296) acc1: 87.5000 (87.1287) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1813 data: 0.0873 max mem: 3535 +Test: [ 960/2387] eta: 0:04:47 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3287 (0.3296) acc1: 87.7604 (87.1369) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1714 data: 0.0763 max mem: 3535 +Test: [ 970/2387] eta: 0:04:45 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3297 (0.3297) acc1: 87.7604 (87.1395) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.1687 data: 0.0735 max mem: 3535 +Test: [ 980/2387] eta: 0:04:42 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3274 (0.3292) acc1: 87.5000 (87.1671) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1699 data: 0.0754 max mem: 3535 +Test: [ 990/2387] eta: 0:04:40 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2908 (0.3293) acc1: 89.8438 (87.1692) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1717 data: 0.0791 max mem: 3535 +Test: [1000/2387] eta: 0:04:37 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2951 (0.3293) acc1: 89.0625 (87.1818) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1623 data: 0.0701 max mem: 3535 +Test: [1010/2387] eta: 0:04:35 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2951 (0.3292) acc1: 89.0625 (87.2007) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1583 data: 0.0671 max mem: 3535 +Test: [1020/2387] eta: 0:04:32 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2966 (0.3289) acc1: 91.1458 (87.2258) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.1679 data: 0.0766 max mem: 3535 +Test: [1030/2387] eta: 0:04:30 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2813 (0.3286) acc1: 92.4479 (87.2679) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1644 data: 0.0732 max mem: 3535 +Test: [1040/2387] eta: 0:04:27 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2697 (0.3280) acc1: 93.4896 (87.3294) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.1517 data: 0.0598 max mem: 3535 +Test: [1050/2387] eta: 0:04:25 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2872 (0.3280) acc1: 94.0104 (87.3627) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1599 data: 0.0643 max mem: 3535 +Test: [1060/2387] eta: 0:04:23 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2916 (0.3279) acc1: 92.9688 (87.3886) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1827 data: 0.0857 max mem: 3535 +Test: [1070/2387] eta: 0:04:20 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2748 (0.3274) acc1: 92.9688 (87.4375) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1699 data: 0.0725 max mem: 3535 +Test: [1080/2387] eta: 0:04:18 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2663 (0.3270) acc1: 92.4479 (87.4636) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1599 data: 0.0638 max mem: 3535 +Test: [1090/2387] eta: 0:04:16 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2893 (0.3274) acc1: 89.8438 (87.4339) acc5: 100.0000 (100.0000) acc1_neg: 73.958 acc5_neg: 100.000 time: 0.1849 data: 0.0888 max mem: 3535 +Test: [1100/2387] eta: 0:04:14 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3527 (0.3276) acc1: 84.3750 (87.4179) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1899 data: 0.0942 max mem: 3535 +Test: [1110/2387] eta: 0:04:11 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3602 (0.3290) acc1: 81.7708 (87.3137) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.1699 data: 0.0747 max mem: 3535 +Test: [1120/2387] eta: 0:04:09 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3549 (0.3291) acc1: 85.1562 (87.3102) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1649 data: 0.0693 max mem: 3536 +Test: [1130/2387] eta: 0:04:07 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3226 (0.3297) acc1: 88.2812 (87.2702) acc5: 100.0000 (100.0000) acc1_neg: 71.354 acc5_neg: 100.000 time: 0.1649 data: 0.0690 max mem: 3536 +Test: [1140/2387] eta: 0:04:04 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3352 (0.3298) acc1: 88.8021 (87.2709) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.1649 data: 0.0678 max mem: 3536 +Test: [1150/2387] eta: 0:04:02 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3073 (0.3296) acc1: 89.0625 (87.2952) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1799 data: 0.0839 max mem: 3536 +Test: [1160/2387] eta: 0:04:00 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2406 (0.3284) acc1: 95.3125 (87.3769) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1699 data: 0.0746 max mem: 3536 +Test: [1170/2387] eta: 0:03:57 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1744 (0.3272) acc1: 98.9583 (87.4642) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1449 data: 0.0495 max mem: 3536 +Test: [1180/2387] eta: 0:03:55 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1889 (0.3261) acc1: 97.3958 (87.5448) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1599 data: 0.0638 max mem: 3536 +Test: [1190/2387] eta: 0:03:53 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1889 (0.3250) acc1: 96.8750 (87.6155) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1733 data: 0.0767 max mem: 3536 +Test: [1200/2387] eta: 0:03:51 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1673 (0.3237) acc1: 98.1771 (87.6995) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1683 data: 0.0722 max mem: 3536 +Test: [1210/2387] eta: 0:03:48 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1396 (0.3222) acc1: 98.9583 (87.7942) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1649 data: 0.0700 max mem: 3536 +Test: [1220/2387] eta: 0:03:46 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1540 (0.3207) acc1: 99.2188 (87.8799) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1449 data: 0.0498 max mem: 3536 +Test: [1230/2387] eta: 0:03:44 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1297 (0.3191) acc1: 98.9583 (87.9701) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1499 data: 0.0535 max mem: 3536 +Test: [1240/2387] eta: 0:03:42 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1266 (0.3176) acc1: 98.9583 (88.0580) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1699 data: 0.0716 max mem: 3536 +Test: [1250/2387] eta: 0:03:39 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1370 (0.3163) acc1: 98.9583 (88.1384) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1699 data: 0.0720 max mem: 3536 +Test: [1260/2387] eta: 0:03:37 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1271 (0.3148) acc1: 98.9583 (88.2234) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1549 data: 0.0588 max mem: 3536 +Test: [1270/2387] eta: 0:03:35 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1118 (0.3131) acc1: 99.4792 (88.3134) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1549 data: 0.0590 max mem: 3536 +Test: [1280/2387] eta: 0:03:33 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1071 (0.3116) acc1: 99.4792 (88.3988) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1648 data: 0.0690 max mem: 3536 +Test: [1290/2387] eta: 0:03:31 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1323 (0.3105) acc1: 98.1771 (88.4571) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1672 data: 0.0718 max mem: 3536 +Test: [1300/2387] eta: 0:03:29 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1431 (0.3094) acc1: 96.3542 (88.5150) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1930 data: 0.0992 max mem: 3536 +Test: [1310/2387] eta: 0:03:27 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1495 (0.3085) acc1: 97.1354 (88.5756) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2125 data: 0.1203 max mem: 3536 +Test: [1320/2387] eta: 0:03:25 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1512 (0.3074) acc1: 98.1771 (88.6387) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1926 data: 0.0995 max mem: 3536 +Test: [1330/2387] eta: 0:03:23 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1547 (0.3063) acc1: 98.4375 (88.7076) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1879 data: 0.0950 max mem: 3536 +Test: [1340/2387] eta: 0:03:21 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1837 (0.3065) acc1: 96.8750 (88.6811) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2001 data: 0.1082 max mem: 3536 +Test: [1350/2387] eta: 0:03:19 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.0879 (0.3048) acc1: 100.0000 (88.7649) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2022 data: 0.1087 max mem: 3537 +Test: [1360/2387] eta: 0:03:17 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.0668 (0.3030) acc1: 100.0000 (88.8474) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1895 data: 0.0954 max mem: 3537 +Test: [1370/2387] eta: 0:03:15 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.0668 (0.3014) acc1: 100.0000 (88.9284) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1662 data: 0.0746 max mem: 3537 +Test: [1380/2387] eta: 0:03:13 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.0681 (0.2997) acc1: 100.0000 (89.0086) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1692 data: 0.0746 max mem: 3537 +Test: [1390/2387] eta: 0:03:11 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.0663 (0.2980) acc1: 100.0000 (89.0876) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1724 data: 0.0776 max mem: 3537 +Test: [1400/2387] eta: 0:03:09 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.0597 (0.2963) acc1: 100.0000 (89.1655) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1601 data: 0.0684 max mem: 3537 +Test: [1410/2387] eta: 0:03:07 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.0628 (0.2958) acc1: 100.0000 (89.1963) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1799 data: 0.0881 max mem: 3537 +Test: [1420/2387] eta: 0:03:05 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2669 (0.2962) acc1: 91.6667 (89.1644) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.2003 data: 0.1087 max mem: 3537 +Test: [1430/2387] eta: 0:03:03 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3051 (0.2965) acc1: 87.7604 (89.1582) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1904 data: 0.0990 max mem: 3537 +Test: [1440/2387] eta: 0:03:01 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2980 (0.2966) acc1: 90.6250 (89.1619) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2012 data: 0.1109 max mem: 3537 +Test: [1450/2387] eta: 0:02:59 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2406 (0.2962) acc1: 94.2708 (89.1901) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2017 data: 0.1113 max mem: 3537 +Test: [1460/2387] eta: 0:02:58 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2045 (0.2957) acc1: 95.5729 (89.2297) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1923 data: 0.1004 max mem: 3537 +Test: [1470/2387] eta: 0:02:56 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1788 (0.2947) acc1: 96.6146 (89.2923) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1927 data: 0.0990 max mem: 3537 +Test: [1480/2387] eta: 0:02:54 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1661 (0.2939) acc1: 98.6979 (89.3453) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1857 data: 0.0926 max mem: 3537 +Test: [1490/2387] eta: 0:02:52 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1651 (0.2931) acc1: 98.1771 (89.4029) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1835 data: 0.0894 max mem: 3537 +Test: [1500/2387] eta: 0:02:50 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1645 (0.2923) acc1: 98.1771 (89.4565) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1841 data: 0.0894 max mem: 3537 +Test: [1510/2387] eta: 0:02:48 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1433 (0.2912) acc1: 99.2188 (89.5246) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1786 data: 0.0869 max mem: 3537 +Test: [1520/2387] eta: 0:02:46 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1474 (0.2912) acc1: 98.9583 (89.5402) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1767 data: 0.0861 max mem: 3537 +Test: [1530/2387] eta: 0:02:44 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2815 (0.2910) acc1: 91.9271 (89.5646) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1895 data: 0.0983 max mem: 3537 +Test: [1540/2387] eta: 0:02:42 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2539 (0.2908) acc1: 93.4896 (89.5945) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1833 data: 0.0914 max mem: 3537 +Test: [1550/2387] eta: 0:02:40 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2555 (0.2907) acc1: 94.2708 (89.6214) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1779 data: 0.0864 max mem: 3537 +Test: [1560/2387] eta: 0:02:38 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2555 (0.2905) acc1: 94.2708 (89.6432) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.1765 data: 0.0854 max mem: 3537 +Test: [1570/2387] eta: 0:02:36 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1936 (0.2897) acc1: 96.6146 (89.6962) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1829 data: 0.0924 max mem: 3537 +Test: [1580/2387] eta: 0:02:34 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1916 (0.2893) acc1: 97.1354 (89.7322) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1890 data: 0.0990 max mem: 3538 +Test: [1590/2387] eta: 0:02:32 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2115 (0.2888) acc1: 96.3542 (89.7616) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1744 data: 0.0846 max mem: 3538 +Test: [1600/2387] eta: 0:02:30 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2303 (0.2889) acc1: 92.7083 (89.7525) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.1768 data: 0.0865 max mem: 3538 +Test: [1610/2387] eta: 0:02:28 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2337 (0.2885) acc1: 93.7500 (89.7854) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1809 data: 0.0880 max mem: 3538 +Test: [1620/2387] eta: 0:02:26 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2337 (0.2884) acc1: 93.7500 (89.7920) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1800 data: 0.0853 max mem: 3538 +Test: [1630/2387] eta: 0:02:24 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2349 (0.2883) acc1: 93.4896 (89.8035) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1741 data: 0.0815 max mem: 3538 +Test: [1640/2387] eta: 0:02:22 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2235 (0.2878) acc1: 96.0938 (89.8469) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1739 data: 0.0825 max mem: 3538 +Test: [1650/2387] eta: 0:02:20 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1899 (0.2873) acc1: 96.3542 (89.8788) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1885 data: 0.0963 max mem: 3538 +Test: [1660/2387] eta: 0:02:18 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1947 (0.2871) acc1: 94.5312 (89.8938) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1856 data: 0.0926 max mem: 3538 +Test: [1670/2387] eta: 0:02:16 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2303 (0.2867) acc1: 95.5729 (89.9243) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1748 data: 0.0820 max mem: 3538 +Test: [1680/2387] eta: 0:02:14 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2381 (0.2865) acc1: 95.5729 (89.9485) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2009 data: 0.1081 max mem: 3538 +Test: [1690/2387] eta: 0:02:13 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2543 (0.2863) acc1: 91.9271 (89.9656) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.2093 data: 0.1155 max mem: 3538 +Test: [1700/2387] eta: 0:02:11 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2471 (0.2859) acc1: 93.4896 (89.9978) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1963 data: 0.1017 max mem: 3538 +Test: [1710/2387] eta: 0:02:09 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1843 (0.2852) acc1: 97.3958 (90.0409) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1926 data: 0.0983 max mem: 3538 +Test: [1720/2387] eta: 0:02:07 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1301 (0.2843) acc1: 98.9583 (90.0899) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1912 data: 0.0992 max mem: 3538 +Test: [1730/2387] eta: 0:02:05 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1298 (0.2834) acc1: 99.2188 (90.1367) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1990 data: 0.1066 max mem: 3538 +Test: [1740/2387] eta: 0:02:03 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1373 (0.2827) acc1: 98.6979 (90.1793) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.2047 data: 0.1087 max mem: 3538 +Test: [1750/2387] eta: 0:02:01 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1424 (0.2819) acc1: 98.6979 (90.2267) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1937 data: 0.0980 max mem: 3538 +Test: [1760/2387] eta: 0:01:59 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1294 (0.2810) acc1: 99.4792 (90.2748) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1900 data: 0.0962 max mem: 3538 +Test: [1770/2387] eta: 0:01:57 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1623 (0.2806) acc1: 97.6562 (90.2992) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1938 data: 0.0988 max mem: 3538 +Test: [1780/2387] eta: 0:01:56 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2021 (0.2801) acc1: 96.6146 (90.3372) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2060 data: 0.1112 max mem: 3538 +Test: [1790/2387] eta: 0:01:54 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1820 (0.2795) acc1: 97.3958 (90.3765) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2120 data: 0.1197 max mem: 3538 +Test: [1800/2387] eta: 0:01:52 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1751 (0.2791) acc1: 97.6562 (90.4043) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2031 data: 0.1122 max mem: 3538 +Test: [1810/2387] eta: 0:01:50 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1689 (0.2785) acc1: 97.3958 (90.4431) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1980 data: 0.1066 max mem: 3539 +Test: [1820/2387] eta: 0:01:48 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1376 (0.2777) acc1: 99.2188 (90.4874) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1853 data: 0.0936 max mem: 3539 +Test: [1830/2387] eta: 0:01:46 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1159 (0.2769) acc1: 99.7396 (90.5351) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1755 data: 0.0832 max mem: 3539 +Test: [1840/2387] eta: 0:01:44 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1125 (0.2760) acc1: 100.0000 (90.5862) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1744 data: 0.0802 max mem: 3539 +Test: [1850/2387] eta: 0:01:42 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1117 (0.2751) acc1: 100.0000 (90.6367) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1722 data: 0.0789 max mem: 3539 +Test: [1860/2387] eta: 0:01:40 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1121 (0.2743) acc1: 100.0000 (90.6866) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1682 data: 0.0766 max mem: 3539 +Test: [1870/2387] eta: 0:01:38 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1125 (0.2734) acc1: 100.0000 (90.7355) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1876 data: 0.0943 max mem: 3539 +Test: [1880/2387] eta: 0:01:36 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1126 (0.2725) acc1: 100.0000 (90.7845) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2011 data: 0.1077 max mem: 3539 +Test: [1890/2387] eta: 0:01:34 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1128 (0.2717) acc1: 100.0000 (90.8329) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1941 data: 0.1017 max mem: 3539 +Test: [1900/2387] eta: 0:01:33 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1147 (0.2709) acc1: 100.0000 (90.8810) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1976 data: 0.1064 max mem: 3539 +Test: [1910/2387] eta: 0:01:31 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1200 (0.2702) acc1: 100.0000 (90.9177) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1973 data: 0.1059 max mem: 3539 +Test: [1920/2387] eta: 0:01:29 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1918 (0.2711) acc1: 92.7083 (90.8751) acc5: 100.0000 (100.0000) acc1_neg: 72.917 acc5_neg: 100.000 time: 0.1922 data: 0.0987 max mem: 3539 +Test: [1930/2387] eta: 0:01:27 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3147 (0.2710) acc1: 89.5833 (90.8876) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.1943 data: 0.1017 max mem: 3539 +Test: [1940/2387] eta: 0:01:25 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3036 (0.2721) acc1: 86.4583 (90.8103) acc5: 100.0000 (100.0000) acc1_neg: 66.667 acc5_neg: 100.000 time: 0.1998 data: 0.1077 max mem: 3539 +Test: [1950/2387] eta: 0:01:23 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3334 (0.2724) acc1: 86.1979 (90.7957) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.2056 data: 0.1121 max mem: 3539 +Test: [1960/2387] eta: 0:01:21 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3341 (0.2731) acc1: 84.6354 (90.7401) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2018 data: 0.1082 max mem: 3539 +Test: [1970/2387] eta: 0:01:19 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3471 (0.2736) acc1: 82.8125 (90.6856) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.2026 data: 0.1092 max mem: 3539 +Test: [1980/2387] eta: 0:01:17 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2558 (0.2734) acc1: 88.8021 (90.6915) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1993 data: 0.1072 max mem: 3539 +Test: [1990/2387] eta: 0:01:15 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.0891 (0.2725) acc1: 100.0000 (90.7380) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1925 data: 0.1020 max mem: 3539 +Test: [2000/2387] eta: 0:01:14 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.0885 (0.2716) acc1: 100.0000 (90.7839) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1973 data: 0.1039 max mem: 3539 +Test: [2010/2387] eta: 0:01:12 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.0878 (0.2706) acc1: 100.0000 (90.8290) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1996 data: 0.1059 max mem: 3539 +Test: [2020/2387] eta: 0:01:10 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.0879 (0.2697) acc1: 100.0000 (90.8740) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2075 data: 0.1151 max mem: 3539 +Test: [2030/2387] eta: 0:01:08 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.0879 (0.2689) acc1: 100.0000 (90.9185) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2090 data: 0.1164 max mem: 3540 +Test: [2040/2387] eta: 0:01:06 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.0877 (0.2680) acc1: 100.0000 (90.9622) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1944 data: 0.1023 max mem: 3540 +Test: [2050/2387] eta: 0:01:04 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.0875 (0.2671) acc1: 100.0000 (91.0060) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1832 data: 0.0907 max mem: 3540 +Test: [2060/2387] eta: 0:01:02 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.0869 (0.2662) acc1: 100.0000 (91.0497) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1841 data: 0.0900 max mem: 3540 +Test: [2070/2387] eta: 0:01:00 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.0865 (0.2653) acc1: 100.0000 (91.0929) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1901 data: 0.0964 max mem: 3540 +Test: [2080/2387] eta: 0:00:58 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.0865 (0.2645) acc1: 100.0000 (91.1356) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1878 data: 0.0974 max mem: 3540 +Test: [2090/2387] eta: 0:00:56 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.0866 (0.2636) acc1: 100.0000 (91.1773) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1897 data: 0.0980 max mem: 3540 +Test: [2100/2387] eta: 0:00:54 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.0870 (0.2628) acc1: 100.0000 (91.2193) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1986 data: 0.1060 max mem: 3540 +Test: [2110/2387] eta: 0:00:53 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.0874 (0.2620) acc1: 100.0000 (91.2608) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1999 data: 0.1076 max mem: 3540 +Test: [2120/2387] eta: 0:00:51 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.0899 (0.2620) acc1: 99.7396 (91.2639) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1951 data: 0.1036 max mem: 3540 +Test: [2130/2387] eta: 0:00:49 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2719 (0.2621) acc1: 93.2292 (91.2702) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1844 data: 0.0940 max mem: 3540 +Test: [2140/2387] eta: 0:00:47 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2905 (0.2626) acc1: 88.0208 (91.2264) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1706 data: 0.0796 max mem: 3540 +Test: [2150/2387] eta: 0:00:45 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3054 (0.2628) acc1: 87.7604 (91.2268) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.1726 data: 0.0817 max mem: 3540 +Test: [2160/2387] eta: 0:00:43 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2817 (0.2629) acc1: 93.2292 (91.2346) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2055 data: 0.1152 max mem: 3540 +Test: [2170/2387] eta: 0:00:41 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2970 (0.2630) acc1: 92.4479 (91.2298) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1981 data: 0.1076 max mem: 3540 +Test: [2180/2387] eta: 0:00:39 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3176 (0.2636) acc1: 86.9792 (91.1951) acc5: 100.0000 (100.0000) acc1_neg: 77.083 acc5_neg: 100.000 time: 0.1619 data: 0.0698 max mem: 3540 +Test: [2190/2387] eta: 0:00:37 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2872 (0.2632) acc1: 91.6667 (91.2126) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1699 data: 0.0779 max mem: 3540 +Test: [2200/2387] eta: 0:00:35 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2140 (0.2634) acc1: 95.0521 (91.2089) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1899 data: 0.0977 max mem: 3540 +Test: [2210/2387] eta: 0:00:33 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2525 (0.2632) acc1: 95.0521 (91.2270) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1877 data: 0.0947 max mem: 3540 +Test: [2220/2387] eta: 0:00:31 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2364 (0.2632) acc1: 96.0938 (91.2473) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1799 data: 0.0870 max mem: 3540 +Test: [2230/2387] eta: 0:00:30 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2400 (0.2631) acc1: 95.8333 (91.2654) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1836 data: 0.0912 max mem: 3540 +Test: [2240/2387] eta: 0:00:28 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2400 (0.2630) acc1: 94.2708 (91.2780) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1899 data: 0.0989 max mem: 3540 +Test: [2250/2387] eta: 0:00:26 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2292 (0.2626) acc1: 97.6562 (91.3080) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1959 data: 0.1033 max mem: 3540 +Test: [2260/2387] eta: 0:00:24 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1885 (0.2625) acc1: 98.1771 (91.3233) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1902 data: 0.0975 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2085 (0.2623) acc1: 95.5729 (91.3418) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1790 data: 0.0868 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2359 (0.2627) acc1: 93.2292 (91.3151) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.1766 data: 0.0838 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2605 (0.2626) acc1: 91.1458 (91.3250) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1806 data: 0.0887 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2403 (0.2627) acc1: 93.7500 (91.3273) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1850 data: 0.0922 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2403 (0.2626) acc1: 94.5312 (91.3381) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1849 data: 0.0924 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2568 (0.2628) acc1: 93.4896 (91.3342) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1747 data: 0.0834 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3267 (0.2634) acc1: 88.8021 (91.2931) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.1873 data: 0.0954 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3600 (0.2638) acc1: 83.8542 (91.2712) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.1904 data: 0.0980 max mem: 3541 +Test: [2350/2387] eta: 0:00:07 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3776 (0.2650) acc1: 82.5521 (91.1864) acc5: 100.0000 (100.0000) acc1_neg: 77.083 acc5_neg: 100.000 time: 0.1799 data: 0.0880 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.4374 (0.2657) acc1: 77.0833 (91.1390) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1686 data: 0.0764 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.4272 (0.2666) acc1: 78.6458 (91.0719) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.1460 data: 0.0540 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3416 (0.2669) acc1: 85.4167 (91.0607) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.1421 data: 0.0514 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3416 (0.2668) acc1: 85.4167 (91.0554) acc5: 100.0000 (100.0000) acc1_neg: 97.674 acc5_neg: 100.000 time: 0.1359 data: 0.0481 max mem: 3541 +Test: Total time: 0:07:34 (0.1902 s / it) +* Overall Acc@1 91.055 Acc@5 100.000 + target == real Acc@1 44.171 Acc@5 100.000 + target == fake Acc@1 94.348 Acc@5 100.000 + loss 0.267 + PR-AUC 0.9870 ROC-AUC 0.8591 EER 21.92% (thr=0.7539) +Accuracy of the network on the 916267 test images: 91.1% +[rank0]:[W725 18:35:04.143167662 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:35:28.233000 607216 torch/distributed/run.py:766] +W0725 18:35:28.233000 607216 torch/distributed/run.py:766] ***************************************** +W0725 18:35:28.233000 607216 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:35:28.233000 607216 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W725 18:36:01.452634885 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:36:02.741251931 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:36:02.780841065 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:36:02.852777861 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:36:02.177723732 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:36:02.198965159 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:36:02.244796744 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:36:02.247367067 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed4_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed4_resampled_n116-041-153_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:07:22 acc1_pos: 55.208 acc5_pos: 100.000 loss: 0.7369 (0.7369) acc1: 55.2083 (55.2083) acc5: 100.0000 (100.0000) time: 7.8196 data: 6.3899 max mem: 3531 +Test: [ 10/517] eta: 0:06:45 acc1_pos: 36.979 acc5_pos: 100.000 loss: 0.9778 (0.9490) acc1: 46.8750 (44.7443) acc5: 100.0000 (100.0000) time: 0.8007 data: 0.5813 max mem: 3531 +Test: [ 20/517] eta: 0:03:52 acc1_pos: 60.677 acc5_pos: 100.000 loss: 0.9214 (0.9087) acc1: 46.8750 (49.0079) acc5: 100.0000 (100.0000) time: 0.0993 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:03:25 acc1_pos: 16.667 acc5_pos: 100.000 loss: 0.8208 (0.9462) acc1: 42.4479 (45.9173) acc5: 100.0000 (100.0000) time: 0.2149 data: 0.1166 max mem: 3531 +Test: [ 40/517] eta: 0:03:19 acc1_pos: 37.240 acc5_pos: 100.000 loss: 1.0031 (0.9682) acc1: 35.6771 (43.0132) acc5: 100.0000 (100.0000) time: 0.3650 data: 0.2710 max mem: 3531 +Test: [ 50/517] eta: 0:03:03 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.8216 (0.9297) acc1: 43.2292 (45.2104) acc5: 100.0000 (100.0000) acc1_neg: 48.958 acc5_neg: 100.000 time: 0.3449 data: 0.2517 max mem: 3531 +Test: [ 60/517] eta: 0:02:48 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.6521 (0.8728) acc1: 58.0729 (49.0949) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.2723 data: 0.1783 max mem: 3531 +Test: [ 70/517] eta: 0:02:36 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.6206 (0.8357) acc1: 63.0208 (51.6872) acc5: 100.0000 (100.0000) acc1_neg: 60.417 acc5_neg: 100.000 time: 0.2449 data: 0.1495 max mem: 3531 +Test: [ 80/517] eta: 0:02:27 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.7285 (0.8299) acc1: 53.3854 (51.7393) acc5: 100.0000 (100.0000) acc1_neg: 47.656 acc5_neg: 100.000 time: 0.2439 data: 0.1484 max mem: 3531 +Test: [ 90/517] eta: 0:02:21 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.5127 (0.7830) acc1: 67.7083 (55.6118) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.2602 data: 0.1674 max mem: 3531 +Test: [100/517] eta: 0:02:18 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.4094 (0.7483) acc1: 84.8958 (58.3076) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.3034 data: 0.2101 max mem: 3531 +Test: [110/517] eta: 0:02:11 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.4154 (0.7196) acc1: 82.8125 (60.3275) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.2846 data: 0.1903 max mem: 3531 +Test: [120/517] eta: 0:02:06 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.4310 (0.6985) acc1: 79.6875 (61.7489) acc5: 100.0000 (100.0000) acc1_neg: 69.271 acc5_neg: 100.000 time: 0.2499 data: 0.1537 max mem: 3531 +Test: [130/517] eta: 0:02:01 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.4432 (0.6804) acc1: 78.3854 (62.9453) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.2599 data: 0.1640 max mem: 3531 +Test: [140/517] eta: 0:01:56 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.4283 (0.6616) acc1: 78.3854 (64.1696) acc5: 100.0000 (100.0000) acc1_neg: 75.781 acc5_neg: 100.000 time: 0.2449 data: 0.1503 max mem: 3531 +Test: [150/517] eta: 0:01:51 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.4039 (0.6427) acc1: 81.2500 (65.5457) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.2449 data: 0.1484 max mem: 3531 +Test: [160/517] eta: 0:01:46 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.5127 (0.6405) acc1: 73.6979 (65.7010) acc5: 100.0000 (100.0000) acc1_neg: 67.708 acc5_neg: 100.000 time: 0.2349 data: 0.1386 max mem: 3531 +Test: [170/517] eta: 0:01:42 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.6254 (0.6364) acc1: 65.8854 (65.8946) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.2299 data: 0.1343 max mem: 3531 +Test: [180/517] eta: 0:01:39 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.4125 (0.6192) acc1: 80.7292 (67.0638) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.2549 data: 0.1587 max mem: 3531 +Test: [190/517] eta: 0:01:34 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.2649 (0.6010) acc1: 90.1042 (68.3860) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.2399 data: 0.1448 max mem: 3531 +Test: [200/517] eta: 0:01:31 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.2394 (0.5832) acc1: 95.3125 (69.7774) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2448 data: 0.1508 max mem: 3531 +Test: [210/517] eta: 0:01:28 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.3113 (0.5837) acc1: 82.5521 (69.6843) acc5: 100.0000 (100.0000) acc1_neg: 69.531 acc5_neg: 100.000 time: 0.2649 data: 0.1692 max mem: 3532 +Test: [220/517] eta: 0:01:24 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.6367 (0.5882) acc1: 64.3229 (69.2744) acc5: 100.0000 (100.0000) acc1_neg: 53.906 acc5_neg: 100.000 time: 0.2310 data: 0.1361 max mem: 3532 +Test: [230/517] eta: 0:01:21 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.6460 (0.5928) acc1: 60.4167 (68.8830) acc5: 100.0000 (100.0000) acc1_neg: 39.323 acc5_neg: 100.000 time: 0.2449 data: 0.1536 max mem: 3532 +Test: [240/517] eta: 0:01:18 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.7737 (0.6060) acc1: 50.2604 (67.6716) acc5: 100.0000 (100.0000) acc1_neg: 17.448 acc5_neg: 100.000 time: 0.2738 data: 0.1811 max mem: 3532 +Test: [250/517] eta: 0:01:14 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.8294 (0.6060) acc1: 49.2188 (67.6980) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.2365 data: 0.1417 max mem: 3532 +Test: [260/517] eta: 0:01:11 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.4968 (0.6030) acc1: 76.0417 (67.9887) acc5: 100.0000 (100.0000) acc1_neg: 76.042 acc5_neg: 100.000 time: 0.2349 data: 0.1412 max mem: 3532 +Test: [270/517] eta: 0:01:08 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.5280 (0.6004) acc1: 71.6146 (68.1139) acc5: 100.0000 (100.0000) acc1_neg: 61.979 acc5_neg: 100.000 time: 0.2483 data: 0.1546 max mem: 3532 +Test: [280/517] eta: 0:01:05 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.5280 (0.5984) acc1: 71.3542 (68.3163) acc5: 100.0000 (100.0000) acc1_neg: 63.542 acc5_neg: 100.000 time: 0.2351 data: 0.1408 max mem: 3532 +Test: [290/517] eta: 0:01:02 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.6223 (0.6022) acc1: 64.5833 (67.8721) acc5: 100.0000 (100.0000) acc1_neg: 44.531 acc5_neg: 100.000 time: 0.2399 data: 0.1453 max mem: 3532 +Test: [300/517] eta: 0:00:59 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.6184 (0.6010) acc1: 61.7188 (67.8364) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.2408 data: 0.1441 max mem: 3532 +Test: [310/517] eta: 0:00:56 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.5012 (0.5975) acc1: 70.8333 (68.1287) acc5: 100.0000 (100.0000) acc1_neg: 68.490 acc5_neg: 100.000 time: 0.2499 data: 0.1517 max mem: 3532 +Test: [320/517] eta: 0:00:53 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.4585 (0.5938) acc1: 79.4271 (68.4352) acc5: 100.0000 (100.0000) acc1_neg: 66.927 acc5_neg: 100.000 time: 0.2755 data: 0.1803 max mem: 3532 +Test: [330/517] eta: 0:00:51 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.4555 (0.5892) acc1: 80.4688 (68.8090) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.3107 data: 0.2197 max mem: 3532 +Test: [340/517] eta: 0:00:48 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.3922 (0.5834) acc1: 85.6771 (69.3823) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2809 data: 0.1883 max mem: 3532 +Test: [350/517] eta: 0:00:45 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.4024 (0.5813) acc1: 86.7188 (69.5624) acc5: 100.0000 (100.0000) acc1_neg: 51.562 acc5_neg: 100.000 time: 0.2365 data: 0.1437 max mem: 3532 +Test: [360/517] eta: 0:00:42 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.5221 (0.5779) acc1: 75.5208 (69.7960) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2530 data: 0.1617 max mem: 3532 +Test: [370/517] eta: 0:00:40 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.2542 (0.5683) acc1: 96.8750 (70.5778) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2693 data: 0.1750 max mem: 3532 +Test: [380/517] eta: 0:00:37 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.2154 (0.5593) acc1: 98.1771 (71.2660) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2554 data: 0.1598 max mem: 3532 +Test: [390/517] eta: 0:00:34 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.2687 (0.5532) acc1: 92.7083 (71.6439) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2303 data: 0.1367 max mem: 3532 +Test: [400/517] eta: 0:00:31 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.2918 (0.5461) acc1: 90.8854 (72.2322) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.2334 data: 0.1396 max mem: 3532 +Test: [410/517] eta: 0:00:28 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.2628 (0.5393) acc1: 95.8333 (72.8235) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2475 data: 0.1510 max mem: 3532 +Test: [420/517] eta: 0:00:26 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.3267 (0.5401) acc1: 83.0729 (72.7923) acc5: 100.0000 (100.0000) acc1_neg: 59.375 acc5_neg: 100.000 time: 0.2999 data: 0.2050 max mem: 3532 +Test: [430/517] eta: 0:00:23 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.6538 (0.5448) acc1: 59.8958 (72.3505) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.3003 data: 0.2060 max mem: 3532 +Test: [440/517] eta: 0:00:20 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.5227 (0.5424) acc1: 67.9688 (72.5069) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.2499 data: 0.1510 max mem: 3533 +Test: [450/517] eta: 0:00:18 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.4049 (0.5400) acc1: 80.9896 (72.7764) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.2595 data: 0.1616 max mem: 3533 +Test: [460/517] eta: 0:00:15 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.3908 (0.5361) acc1: 88.2812 (73.1065) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.2399 data: 0.1459 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.3194 (0.5313) acc1: 91.1458 (73.5370) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2349 data: 0.1425 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.3004 (0.5276) acc1: 92.4479 (73.8961) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2499 data: 0.1582 max mem: 3533 +Test: [490/517] eta: 0:00:07 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.4013 (0.5269) acc1: 84.3750 (73.9366) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.2399 data: 0.1446 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.4828 (0.5260) acc1: 74.4792 (73.9999) acc5: 100.0000 (100.0000) acc1_neg: 72.917 acc5_neg: 100.000 time: 0.2349 data: 0.1378 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.4609 (0.5237) acc1: 77.8646 (74.1973) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.2026 data: 0.1098 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.4349 (0.5232) acc1: 79.1667 (74.2752) acc5: 100.0000 (100.0000) acc1_neg: 69.258 acc5_neg: 100.000 time: 0.1751 data: 0.0814 max mem: 3533 +Test: Total time: 0:02:17 (0.2653 s / it) +* Overall Acc@1 74.275 Acc@5 100.000 + target == real Acc@1 44.922 Acc@5 100.000 + target == fake Acc@1 77.320 Acc@5 100.000 + loss 0.523 + PR-AUC 0.9371 ROC-AUC 0.6510 EER 39.24% (thr=0.6309) +Accuracy of the network on the 198427 test images: 74.3% +[rank0]:[W725 18:38:30.593738810 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0725 18:38:53.316000 607804 torch/distributed/run.py:766] +W0725 18:38:53.316000 607804 torch/distributed/run.py:766] ***************************************** +W0725 18:38:53.316000 607804 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:38:53.316000 607804 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:39:29.973281323 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:39:29.172240957 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:39:29.355879606 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:39:29.439242531 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:39:30.603379535 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:39:30.620504293 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:39:30.787440099 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:39:30.815576883 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed4_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed4_resampled_n116-041-153_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:40:14 acc1_pos: 27.083 acc5_pos: 100.000 loss: 1.0417 (1.0417) acc1: 27.0833 (27.0833) acc5: 100.0000 (100.0000) time: 6.2061 data: 4.5432 max mem: 3531 +Test: [ 10/389] eta: 0:04:04 acc1_pos: 92.969 acc5_pos: 100.000 loss: 0.4252 (0.6020) acc1: 82.8125 (70.0047) acc5: 100.0000 (100.0000) time: 0.6461 data: 0.4134 max mem: 3531 +Test: [ 20/389] eta: 0:02:23 acc1_pos: 41.927 acc5_pos: 100.000 loss: 0.5913 (0.6445) acc1: 69.0104 (64.9678) acc5: 100.0000 (100.0000) time: 0.0986 data: 0.0056 max mem: 3531 +Test: [ 30/389] eta: 0:02:12 acc1_pos: 63.021 acc5_pos: 100.000 loss: 0.8153 (0.7793) acc1: 41.9271 (51.1173) acc5: 100.0000 (100.0000) time: 0.2185 data: 0.1231 max mem: 3531 +Test: [ 40/389] eta: 0:01:54 acc1_pos: 76.302 acc5_pos: 100.000 loss: 0.9543 (0.8042) acc1: 21.6146 (49.7015) acc5: 100.0000 (100.0000) time: 0.2649 data: 0.1694 max mem: 3531 +Test: [ 50/389] eta: 0:01:39 acc1_pos: 5.469 acc5_pos: 100.000 loss: 0.7270 (0.8008) acc1: 49.7396 (50.1889) acc5: 100.0000 (100.0000) time: 0.1758 data: 0.0791 max mem: 3531 +Test: [ 60/389] eta: 0:01:30 acc1_pos: 60.156 acc5_pos: 100.000 loss: 0.8625 (0.8401) acc1: 32.2917 (45.6028) acc5: 100.0000 (100.0000) time: 0.1649 data: 0.0710 max mem: 3531 +Test: [ 70/389] eta: 0:01:25 acc1_pos: 35.938 acc5_pos: 100.000 loss: 0.9054 (0.8460) acc1: 32.2917 (45.3345) acc5: 100.0000 (100.0000) time: 0.2000 data: 0.1056 max mem: 3531 +Test: [ 80/389] eta: 0:01:18 acc1_pos: 95.573 acc5_pos: 100.000 loss: 0.6689 (0.8304) acc1: 60.1562 (47.1644) acc5: 100.0000 (100.0000) time: 0.1961 data: 0.1008 max mem: 3531 +Test: [ 90/389] eta: 0:01:14 acc1_pos: 88.542 acc5_pos: 100.000 loss: 0.4852 (0.7844) acc1: 75.0000 (51.6169) acc5: 100.0000 (100.0000) time: 0.1889 data: 0.0966 max mem: 3531 +Test: [100/389] eta: 0:01:10 acc1_pos: 3.646 acc5_pos: 100.000 loss: 0.7142 (0.8302) acc1: 40.6250 (47.9528) acc5: 100.0000 (100.0000) time: 0.1889 data: 0.0952 max mem: 3531 +Test: [110/389] eta: 0:01:05 acc1_pos: 96.875 acc5_pos: 100.000 loss: 1.0552 (0.8272) acc1: 16.4062 (47.8275) acc5: 100.0000 (100.0000) time: 0.1574 data: 0.0623 max mem: 3531 +Test: [120/389] eta: 0:01:02 acc1_pos: 83.073 acc5_pos: 100.000 loss: 0.7369 (0.8290) acc1: 39.0625 (48.4310) acc5: 100.0000 (100.0000) time: 0.1820 data: 0.0892 max mem: 3531 +Test: [130/389] eta: 0:00:58 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.4790 (0.8083) acc1: 76.3021 (50.0835) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1926 data: 0.1020 max mem: 3531 +Test: [140/389] eta: 0:00:55 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.3439 (0.7805) acc1: 91.4062 (52.4213) acc5: 100.0000 (100.0000) acc1_neg: 38.281 acc5_neg: 100.000 time: 0.1656 data: 0.0741 max mem: 3531 +Test: [150/389] eta: 0:00:53 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.3801 (0.7661) acc1: 88.5417 (53.6148) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1916 data: 0.0977 max mem: 3531 +Test: [160/389] eta: 0:00:50 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.3047 (0.7365) acc1: 94.0104 (56.1449) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1975 data: 0.1050 max mem: 3531 +Test: [170/389] eta: 0:00:47 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.2325 (0.7091) acc1: 99.2188 (58.3927) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1751 data: 0.0838 max mem: 3531 +Test: [180/389] eta: 0:00:45 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.1979 (0.6823) acc1: 99.2188 (60.3951) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1862 data: 0.0945 max mem: 3531 +Test: [190/389] eta: 0:00:42 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.2915 (0.6751) acc1: 92.4479 (60.9757) acc5: 100.0000 (100.0000) acc1_neg: 30.469 acc5_neg: 100.000 time: 0.1630 data: 0.0709 max mem: 3531 +Test: [200/389] eta: 0:00:39 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.5539 (0.6720) acc1: 69.7917 (61.1979) acc5: 100.0000 (100.0000) acc1_neg: 63.802 acc5_neg: 100.000 time: 0.1548 data: 0.0619 max mem: 3531 +Test: [210/389] eta: 0:00:37 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.5355 (0.6615) acc1: 70.8333 (62.1075) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1663 data: 0.0728 max mem: 3532 +Test: [220/389] eta: 0:00:34 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.3775 (0.6417) acc1: 94.2708 (63.7337) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1586 data: 0.0664 max mem: 3532 +Test: [230/389] eta: 0:00:32 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.3204 (0.6324) acc1: 94.0104 (64.4864) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.1756 data: 0.0845 max mem: 3532 +Test: [240/389] eta: 0:00:30 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.3804 (0.6191) acc1: 92.4479 (65.6801) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1907 data: 0.0988 max mem: 3532 +Test: [250/389] eta: 0:00:28 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.2490 (0.6039) acc1: 97.1354 (66.9572) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2003 data: 0.1089 max mem: 3532 +Test: [260/389] eta: 0:00:26 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.2803 (0.5937) acc1: 98.6979 (67.8919) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1908 data: 0.0996 max mem: 3532 +Test: [270/389] eta: 0:00:24 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.4183 (0.5882) acc1: 91.1458 (68.4560) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1872 data: 0.0925 max mem: 3532 +Test: [280/389] eta: 0:00:22 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.3431 (0.5797) acc1: 90.6250 (69.2477) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1903 data: 0.0957 max mem: 3532 +Test: [290/389] eta: 0:00:20 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.2697 (0.5691) acc1: 97.9167 (70.1246) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2049 data: 0.1132 max mem: 3532 +Test: [300/389] eta: 0:00:18 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.2181 (0.5573) acc1: 99.7396 (71.0981) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2102 data: 0.1189 max mem: 3532 +Test: [310/389] eta: 0:00:15 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.2497 (0.5514) acc1: 98.1771 (71.5995) acc5: 100.0000 (100.0000) acc1_neg: 58.854 acc5_neg: 100.000 time: 0.1960 data: 0.1050 max mem: 3532 +Test: [320/389] eta: 0:00:13 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.3247 (0.5453) acc1: 94.7917 (72.2279) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1976 data: 0.1047 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.4919 (0.5473) acc1: 84.3750 (71.9254) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.1849 data: 0.0908 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.3259 (0.5381) acc1: 79.9479 (72.6753) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1514 data: 0.0593 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.2751 (0.5380) acc1: 98.6979 (72.6778) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.1321 data: 0.0421 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.4390 (0.5361) acc1: 88.2812 (73.0494) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1371 data: 0.0474 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.4289 (0.5309) acc1: 90.1042 (73.5554) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1336 data: 0.0440 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.3373 (0.5261) acc1: 96.0938 (73.9542) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1218 data: 0.0322 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.2364 (0.5206) acc1: 98.4375 (74.4245) acc5: 100.0000 (100.0000) acc1_neg: 94.809 acc5_neg: 100.000 time: 0.1220 data: 0.0250 max mem: 3532 +Test: Total time: 0:01:14 (0.1914 s / it) +* Overall Acc@1 74.425 Acc@5 100.000 + target == real Acc@1 47.731 Acc@5 100.000 + target == fake Acc@1 86.832 Acc@5 100.000 + loss 0.521 + PR-AUC 0.8627 ROC-AUC 0.7694 EER 31.12% (thr=0.6509) +Accuracy of the network on the 149358 test images: 74.4% +[rank0]:[W725 18:40:53.670808649 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:41:15.070000 608374 torch/distributed/run.py:766] +W0725 18:41:15.070000 608374 torch/distributed/run.py:766] ***************************************** +W0725 18:41:15.070000 608374 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:41:15.070000 608374 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W725 18:41:52.992784343 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank1]:[W725 18:41:52.996181731 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:41:52.315341231 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:41:53.539424385 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:41:53.568139861 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W725 18:41:53.578204891 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:41:53.750761608 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:41:53.790631624 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed4_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed4_resampled_n116-041-153_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:33:15 acc1_pos: 94.531 acc5_pos: 100.000 loss: 0.2505 (0.2505) acc1: 94.5312 (94.5312) acc5: 100.0000 (100.0000) time: 5.9913 data: 4.6960 max mem: 3531 +Test: [ 10/333] eta: 0:03:22 acc1_pos: 17.969 acc5_pos: 100.000 loss: 0.2840 (0.6969) acc1: 89.8438 (64.7017) acc5: 100.0000 (100.0000) time: 0.6266 data: 0.4273 max mem: 3531 +Test: [ 20/333] eta: 0:01:57 acc1_pos: 13.542 acc5_pos: 100.000 loss: 1.4113 (1.0709) acc1: 16.9271 (40.4886) acc5: 100.0000 (100.0000) time: 0.0942 data: 0.0004 max mem: 3531 +Test: [ 30/333] eta: 0:01:31 acc1_pos: 63.281 acc5_pos: 100.000 loss: 1.1341 (1.0175) acc1: 19.5312 (42.8679) acc5: 100.0000 (100.0000) time: 0.1241 data: 0.0250 max mem: 3531 +Test: [ 40/333] eta: 0:01:22 acc1_pos: 58.333 acc5_pos: 100.000 loss: 0.8344 (0.9690) acc1: 49.4792 (45.1855) acc5: 100.0000 (100.0000) time: 0.1849 data: 0.0856 max mem: 3531 +Test: [ 50/333] eta: 0:01:15 acc1_pos: 49.740 acc5_pos: 100.000 loss: 0.8344 (0.9438) acc1: 50.0000 (45.6904) acc5: 100.0000 (100.0000) time: 0.2149 data: 0.1192 max mem: 3531 +Test: [ 60/333] eta: 0:01:09 acc1_pos: 54.167 acc5_pos: 100.000 loss: 0.8242 (0.9284) acc1: 46.3542 (45.9358) acc5: 100.0000 (100.0000) time: 0.1970 data: 0.1055 max mem: 3531 +Test: [ 70/333] eta: 0:01:05 acc1_pos: 40.885 acc5_pos: 100.000 loss: 0.8374 (0.9253) acc1: 44.7917 (46.0351) acc5: 100.0000 (100.0000) time: 0.1964 data: 0.1060 max mem: 3531 +Test: [ 80/333] eta: 0:00:59 acc1_pos: 38.281 acc5_pos: 100.000 loss: 0.8520 (0.9154) acc1: 45.0521 (46.2545) acc5: 100.0000 (100.0000) time: 0.1838 data: 0.0903 max mem: 3531 +Test: [ 90/333] eta: 0:00:56 acc1_pos: 59.115 acc5_pos: 100.000 loss: 0.7656 (0.8915) acc1: 52.6042 (47.5761) acc5: 100.0000 (100.0000) time: 0.1688 data: 0.0739 max mem: 3531 +Test: [100/333] eta: 0:00:52 acc1_pos: 54.688 acc5_pos: 100.000 loss: 0.7111 (0.8743) acc1: 56.7708 (48.4813) acc5: 100.0000 (100.0000) time: 0.1794 data: 0.0870 max mem: 3531 +Test: [110/333] eta: 0:00:49 acc1_pos: 45.052 acc5_pos: 100.000 loss: 0.7583 (0.8660) acc1: 54.6875 (48.8926) acc5: 100.0000 (100.0000) time: 0.1845 data: 0.0929 max mem: 3531 +Test: [120/333] eta: 0:00:46 acc1_pos: 49.219 acc5_pos: 100.000 loss: 0.7177 (0.8506) acc1: 54.9479 (49.9182) acc5: 100.0000 (100.0000) time: 0.1695 data: 0.0770 max mem: 3531 +Test: [130/333] eta: 0:00:43 acc1_pos: 72.135 acc5_pos: 100.000 loss: 0.6529 (0.8366) acc1: 63.0208 (50.8926) acc5: 100.0000 (100.0000) time: 0.1855 data: 0.0937 max mem: 3531 +Test: [140/333] eta: 0:00:42 acc1_pos: 58.594 acc5_pos: 100.000 loss: 0.6454 (0.8230) acc1: 64.8438 (51.9910) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1381 max mem: 3531 +Test: [150/333] eta: 0:00:39 acc1_pos: 60.677 acc5_pos: 100.000 loss: 0.7232 (0.8179) acc1: 59.8958 (52.2437) acc5: 100.0000 (100.0000) time: 0.1966 data: 0.1026 max mem: 3531 +Test: [160/333] eta: 0:00:37 acc1_pos: 57.566 acc5_pos: 100.000 loss: 0.7673 (0.8132) acc1: 54.4271 (52.5427) acc5: 100.0000 (100.0000) acc1_neg: 51.302 acc5_neg: 100.000 time: 0.2126 data: 0.1154 max mem: 3531 +Test: [170/333] eta: 0:00:34 acc1_pos: 57.566 acc5_pos: 100.000 loss: 0.6726 (0.8059) acc1: 58.3333 (53.0519) acc5: 100.0000 (100.0000) acc1_neg: 66.406 acc5_neg: 100.000 time: 0.2100 data: 0.1136 max mem: 3531 +Test: [180/333] eta: 0:00:32 acc1_pos: 57.566 acc5_pos: 100.000 loss: 0.6176 (0.7923) acc1: 65.6250 (54.0357) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1672 data: 0.0736 max mem: 3531 +Test: [190/333] eta: 0:00:30 acc1_pos: 57.566 acc5_pos: 100.000 loss: 0.5661 (0.7789) acc1: 69.0104 (55.0297) acc5: 100.0000 (100.0000) acc1_neg: 68.490 acc5_neg: 100.000 time: 0.1946 data: 0.1020 max mem: 3531 +Test: [200/333] eta: 0:00:28 acc1_pos: 57.566 acc5_pos: 100.000 loss: 0.5693 (0.7680) acc1: 69.5312 (55.8471) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.2116 data: 0.1181 max mem: 3531 +Test: [210/333] eta: 0:00:25 acc1_pos: 57.566 acc5_pos: 100.000 loss: 0.5723 (0.7586) acc1: 69.5312 (56.5820) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.2006 data: 0.1075 max mem: 3532 +Test: [220/333] eta: 0:00:23 acc1_pos: 57.566 acc5_pos: 100.000 loss: 0.5912 (0.7527) acc1: 69.2708 (56.9818) acc5: 100.0000 (100.0000) acc1_neg: 70.573 acc5_neg: 100.000 time: 0.1881 data: 0.0949 max mem: 3532 +Test: [230/333] eta: 0:00:21 acc1_pos: 57.566 acc5_pos: 100.000 loss: 0.5181 (0.7407) acc1: 73.1771 (57.9343) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.1972 data: 0.1034 max mem: 3532 +Test: [240/333] eta: 0:00:19 acc1_pos: 57.566 acc5_pos: 100.000 loss: 0.5039 (0.7356) acc1: 73.6979 (58.2674) acc5: 100.0000 (100.0000) acc1_neg: 66.667 acc5_neg: 100.000 time: 0.1998 data: 0.1060 max mem: 3532 +Test: [250/333] eta: 0:00:17 acc1_pos: 57.566 acc5_pos: 100.000 loss: 0.6286 (0.7318) acc1: 65.6250 (58.5045) acc5: 100.0000 (100.0000) acc1_neg: 68.750 acc5_neg: 100.000 time: 0.1816 data: 0.0880 max mem: 3532 +Test: [260/333] eta: 0:00:15 acc1_pos: 57.566 acc5_pos: 100.000 loss: 0.6105 (0.7269) acc1: 66.6667 (58.9011) acc5: 100.0000 (100.0000) acc1_neg: 66.667 acc5_neg: 100.000 time: 0.1680 data: 0.0744 max mem: 3532 +Test: [270/333] eta: 0:00:12 acc1_pos: 57.566 acc5_pos: 100.000 loss: 0.5499 (0.7195) acc1: 70.5729 (59.4797) acc5: 100.0000 (100.0000) acc1_neg: 73.698 acc5_neg: 100.000 time: 0.1849 data: 0.0918 max mem: 3532 +Test: [280/333] eta: 0:00:10 acc1_pos: 57.566 acc5_pos: 100.000 loss: 0.5405 (0.7167) acc1: 71.0938 (59.6808) acc5: 100.0000 (100.0000) acc1_neg: 68.750 acc5_neg: 100.000 time: 0.1973 data: 0.1068 max mem: 3532 +Test: [290/333] eta: 0:00:08 acc1_pos: 57.566 acc5_pos: 100.000 loss: 0.7335 (0.7180) acc1: 56.2500 (59.5459) acc5: 100.0000 (100.0000) acc1_neg: 53.385 acc5_neg: 100.000 time: 0.1730 data: 0.0819 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 57.566 acc5_pos: 100.000 loss: 0.7398 (0.7188) acc1: 54.6875 (59.4390) acc5: 100.0000 (100.0000) acc1_neg: 61.198 acc5_neg: 100.000 time: 0.1753 data: 0.0837 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 57.566 acc5_pos: 100.000 loss: 0.7045 (0.7143) acc1: 61.1979 (59.8238) acc5: 100.0000 (100.0000) acc1_neg: 66.406 acc5_neg: 100.000 time: 0.1949 data: 0.1042 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 57.566 acc5_pos: 100.000 loss: 0.6377 (0.7121) acc1: 66.4062 (59.9721) acc5: 100.0000 (100.0000) acc1_neg: 66.406 acc5_neg: 100.000 time: 0.1658 data: 0.0759 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 57.566 acc5_pos: 100.000 loss: 0.6443 (0.7102) acc1: 64.0625 (60.1224) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.1417 data: 0.0516 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 57.566 acc5_pos: 100.000 loss: 0.6431 (0.7097) acc1: 65.1042 (60.1488) acc5: 100.0000 (100.0000) acc1_neg: 65.634 acc5_neg: 100.000 time: 0.1423 data: 0.0483 max mem: 3532 +Test: Total time: 0:01:06 (0.1993 s / it) +* Overall Acc@1 60.149 Acc@5 100.000 + target == real Acc@1 52.578 Acc@5 100.000 + target == fake Acc@1 66.888 Acc@5 100.000 + loss 0.710 + PR-AUC 0.6303 ROC-AUC 0.6341 EER 40.59% (thr=0.5596) +Accuracy of the network on the 127843 test images: 60.1% +[rank0]:[W725 18:43:10.449602262 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:43:29.698000 608945 torch/distributed/run.py:766] +W0725 18:43:29.698000 608945 torch/distributed/run.py:766] ***************************************** +W0725 18:43:29.698000 608945 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:43:29.698000 608945 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:44:00.901248336 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:44:00.210022318 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:44:00.236597928 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:44:00.313365387 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:44:00.470486734 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:44:00.470726697 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:44:01.616419760 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:44:01.688147763 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed4_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed4_resampled_n116-041-153_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:23:30 acc1_pos: 40.625 acc5_pos: 100.000 loss: 1.0045 (1.0045) acc1: 40.6250 (40.6250) acc5: 100.0000 (100.0000) time: 6.0009 data: 4.6424 max mem: 3531 +Test: [ 10/235] eta: 0:02:21 acc1_pos: 46.615 acc5_pos: 100.000 loss: 0.9921 (0.9962) acc1: 43.2292 (43.0634) acc5: 100.0000 (100.0000) time: 0.6276 data: 0.4225 max mem: 3531 +Test: [ 20/235] eta: 0:01:20 acc1_pos: 41.667 acc5_pos: 100.000 loss: 0.9816 (0.9913) acc1: 43.7500 (43.5144) acc5: 100.0000 (100.0000) time: 0.0940 data: 0.0025 max mem: 3531 +Test: [ 30/235] eta: 0:01:05 acc1_pos: 48.177 acc5_pos: 100.000 loss: 0.9754 (0.9875) acc1: 44.7917 (44.0188) acc5: 100.0000 (100.0000) time: 0.1500 data: 0.0562 max mem: 3531 +Test: [ 40/235] eta: 0:00:59 acc1_pos: 44.010 acc5_pos: 100.000 loss: 0.9814 (0.9885) acc1: 44.0104 (43.9215) acc5: 100.0000 (100.0000) time: 0.2303 data: 0.1381 max mem: 3531 +Test: [ 50/235] eta: 0:00:52 acc1_pos: 40.885 acc5_pos: 100.000 loss: 0.9915 (0.9895) acc1: 44.0104 (43.9185) acc5: 100.0000 (100.0000) time: 0.2288 data: 0.1383 max mem: 3531 +Test: [ 60/235] eta: 0:00:47 acc1_pos: 48.698 acc5_pos: 100.000 loss: 0.9737 (0.9858) acc1: 44.2708 (44.0702) acc5: 100.0000 (100.0000) time: 0.1995 data: 0.1027 max mem: 3531 +Test: [ 70/235] eta: 0:00:43 acc1_pos: 35.385 acc5_pos: 100.000 loss: 0.9156 (0.9298) acc1: 47.1354 (47.9093) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.2120 data: 0.1120 max mem: 3531 +Test: [ 80/235] eta: 0:00:40 acc1_pos: 35.385 acc5_pos: 100.000 loss: 0.5312 (0.8805) acc1: 74.7396 (51.1638) acc5: 100.0000 (100.0000) acc1_neg: 76.042 acc5_neg: 100.000 time: 0.2332 data: 0.1390 max mem: 3531 +Test: [ 90/235] eta: 0:00:36 acc1_pos: 35.385 acc5_pos: 100.000 loss: 0.5292 (0.8429) acc1: 74.4792 (53.6487) acc5: 100.0000 (100.0000) acc1_neg: 74.740 acc5_neg: 100.000 time: 0.2094 data: 0.1187 max mem: 3531 +Test: [100/235] eta: 0:00:33 acc1_pos: 35.385 acc5_pos: 100.000 loss: 0.5217 (0.8097) acc1: 74.4792 (55.8091) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.1845 data: 0.0944 max mem: 3531 +Test: [110/235] eta: 0:00:30 acc1_pos: 35.385 acc5_pos: 100.000 loss: 0.5230 (0.7850) acc1: 74.4792 (57.4911) acc5: 100.0000 (100.0000) acc1_neg: 73.958 acc5_neg: 100.000 time: 0.1992 data: 0.1085 max mem: 3531 +Test: [120/235] eta: 0:00:27 acc1_pos: 35.385 acc5_pos: 100.000 loss: 0.5550 (0.7726) acc1: 70.8333 (58.2817) acc5: 100.0000 (100.0000) acc1_neg: 66.927 acc5_neg: 100.000 time: 0.2042 data: 0.1132 max mem: 3531 +Test: [130/235] eta: 0:00:24 acc1_pos: 35.385 acc5_pos: 100.000 loss: 0.6388 (0.7623) acc1: 67.4479 (58.9734) acc5: 100.0000 (100.0000) acc1_neg: 68.229 acc5_neg: 100.000 time: 0.2038 data: 0.1093 max mem: 3531 +Test: [140/235] eta: 0:00:22 acc1_pos: 35.385 acc5_pos: 100.000 loss: 0.6428 (0.7549) acc1: 66.4062 (59.4710) acc5: 100.0000 (100.0000) acc1_neg: 61.719 acc5_neg: 100.000 time: 0.2026 data: 0.1069 max mem: 3531 +Test: [150/235] eta: 0:00:19 acc1_pos: 35.385 acc5_pos: 100.000 loss: 0.6454 (0.7470) acc1: 66.4062 (59.9769) acc5: 100.0000 (100.0000) acc1_neg: 65.104 acc5_neg: 100.000 time: 0.2118 data: 0.1197 max mem: 3531 +Test: [160/235] eta: 0:00:17 acc1_pos: 35.385 acc5_pos: 100.000 loss: 0.6475 (0.7414) acc1: 66.4062 (60.3115) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.2055 data: 0.1127 max mem: 3531 +Test: [170/235] eta: 0:00:14 acc1_pos: 35.385 acc5_pos: 100.000 loss: 0.6267 (0.7272) acc1: 67.1875 (61.2725) acc5: 100.0000 (100.0000) acc1_neg: 71.354 acc5_neg: 100.000 time: 0.1949 data: 0.1023 max mem: 3531 +Test: [180/235] eta: 0:00:12 acc1_pos: 35.385 acc5_pos: 100.000 loss: 0.4712 (0.7134) acc1: 78.1250 (62.1907) acc5: 100.0000 (100.0000) acc1_neg: 72.656 acc5_neg: 100.000 time: 0.2057 data: 0.1136 max mem: 3531 +Test: [190/235] eta: 0:00:10 acc1_pos: 35.385 acc5_pos: 100.000 loss: 0.4683 (0.7005) acc1: 78.3854 (63.0399) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.1987 data: 0.1070 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 35.385 acc5_pos: 100.000 loss: 0.4745 (0.6897) acc1: 77.3438 (63.7334) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.1837 data: 0.0929 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 35.385 acc5_pos: 100.000 loss: 0.4843 (0.6801) acc1: 76.8229 (64.3649) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.1871 data: 0.0934 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 35.385 acc5_pos: 100.000 loss: 0.4773 (0.6707) acc1: 78.1250 (64.9781) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.1786 data: 0.0855 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 35.385 acc5_pos: 100.000 loss: 0.4505 (0.6610) acc1: 78.6458 (65.6216) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.1430 data: 0.0530 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 35.385 acc5_pos: 100.000 loss: 0.4505 (0.6577) acc1: 78.3854 (65.7595) acc5: 100.0000 (100.0000) acc1_neg: 69.565 acc5_neg: 100.000 time: 0.1170 data: 0.0302 max mem: 3532 +Test: Total time: 0:00:50 (0.2134 s / it) +* Overall Acc@1 65.760 Acc@5 100.000 + target == real Acc@1 44.050 Acc@5 100.000 + target == fake Acc@1 73.611 Acc@5 100.000 + loss 0.658 + PR-AUC 0.8279 ROC-AUC 0.6423 EER 39.34% (thr=0.6523) +Accuracy of the network on the 89879 test images: 65.8% +[rank0]:[W725 18:44:59.080307565 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:45:19.229000 609515 torch/distributed/run.py:766] +W0725 18:45:19.229000 609515 torch/distributed/run.py:766] ***************************************** +W0725 18:45:19.229000 609515 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:45:19.229000 609515 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:45:48.720636087 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:45:48.758524591 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:45:48.808694105 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:45:48.061126545 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:45:48.198016879 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:45:48.300824901 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W725 18:45:48.300856682 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:45:48.397647364 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed4_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed4_resampled_n116-041-153_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:21:07 acc1_pos: 14.844 acc5_pos: 100.000 loss: 1.4851 (1.4851) acc1: 14.8438 (14.8438) acc5: 100.0000 (100.0000) time: 4.7120 data: 2.7074 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/269] eta: 0:02:24 acc1_pos: 56.250 acc5_pos: 100.000 loss: 1.0508 (1.0504) acc1: 29.9479 (34.3040) acc5: 100.0000 (100.0000) time: 0.5562 data: 0.2919 max mem: 3531 +Test: [ 20/269] eta: 0:01:27 acc1_pos: 44.010 acc5_pos: 100.000 loss: 0.8751 (0.9569) acc1: 41.1458 (40.5506) acc5: 100.0000 (100.0000) time: 0.1330 data: 0.0398 max mem: 3531 +Test: [ 30/269] eta: 0:01:10 acc1_pos: 28.906 acc5_pos: 100.000 loss: 0.9337 (0.9818) acc1: 37.2396 (39.2053) acc5: 100.0000 (100.0000) time: 0.1487 data: 0.0533 max mem: 3531 +Test: [ 40/269] eta: 0:01:03 acc1_pos: 31.034 acc5_pos: 100.000 loss: 0.6806 (0.8541) acc1: 45.8333 (49.1679) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.2021 data: 0.1064 max mem: 3531 +Test: [ 50/269] eta: 0:00:57 acc1_pos: 31.034 acc5_pos: 100.000 loss: 0.5164 (0.7992) acc1: 76.5625 (53.0790) acc5: 100.0000 (100.0000) acc1_neg: 45.312 acc5_neg: 100.000 time: 0.2191 data: 0.1226 max mem: 3531 +Test: [ 60/269] eta: 0:00:52 acc1_pos: 31.034 acc5_pos: 100.000 loss: 0.5839 (0.7722) acc1: 64.5833 (54.4484) acc5: 100.0000 (100.0000) acc1_neg: 70.052 acc5_neg: 100.000 time: 0.1977 data: 0.0993 max mem: 3531 +Test: [ 70/269] eta: 0:00:49 acc1_pos: 31.034 acc5_pos: 100.000 loss: 0.4825 (0.7246) acc1: 76.8229 (58.4654) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.2067 data: 0.1099 max mem: 3531 +Test: [ 80/269] eta: 0:00:45 acc1_pos: 31.034 acc5_pos: 100.000 loss: 0.4158 (0.6825) acc1: 84.1146 (62.0306) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2057 data: 0.1104 max mem: 3531 +Test: [ 90/269] eta: 0:00:41 acc1_pos: 31.034 acc5_pos: 100.000 loss: 0.4453 (0.6639) acc1: 80.7292 (63.3986) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.1829 data: 0.0836 max mem: 3531 +Test: [100/269] eta: 0:00:39 acc1_pos: 31.034 acc5_pos: 100.000 loss: 0.4453 (0.6411) acc1: 80.4688 (65.4136) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.2041 data: 0.1033 max mem: 3531 +Test: [110/269] eta: 0:00:36 acc1_pos: 31.034 acc5_pos: 100.000 loss: 0.3904 (0.6162) acc1: 86.7188 (67.6192) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2249 data: 0.1253 max mem: 3531 +Test: [120/269] eta: 0:00:34 acc1_pos: 31.034 acc5_pos: 100.000 loss: 0.3911 (0.6060) acc1: 85.4167 (68.3712) acc5: 100.0000 (100.0000) acc1_neg: 61.719 acc5_neg: 100.000 time: 0.2199 data: 0.1239 max mem: 3531 +Test: [130/269] eta: 0:00:32 acc1_pos: 31.034 acc5_pos: 100.000 loss: 0.5731 (0.6074) acc1: 65.1042 (67.9250) acc5: 100.0000 (100.0000) acc1_neg: 56.250 acc5_neg: 100.000 time: 0.2227 data: 0.1294 max mem: 3531 +Test: [140/269] eta: 0:00:29 acc1_pos: 31.034 acc5_pos: 100.000 loss: 0.6357 (0.6085) acc1: 63.2812 (67.6640) acc5: 100.0000 (100.0000) acc1_neg: 57.552 acc5_neg: 100.000 time: 0.2269 data: 0.1353 max mem: 3531 +Test: [150/269] eta: 0:00:27 acc1_pos: 31.034 acc5_pos: 100.000 loss: 0.6031 (0.6049) acc1: 64.5833 (67.8773) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.2287 data: 0.1351 max mem: 3531 +Test: [160/269] eta: 0:00:25 acc1_pos: 31.034 acc5_pos: 100.000 loss: 0.5631 (0.6007) acc1: 72.1354 (68.2098) acc5: 100.0000 (100.0000) acc1_neg: 72.396 acc5_neg: 100.000 time: 0.2434 data: 0.1494 max mem: 3531 +Test: [170/269] eta: 0:00:22 acc1_pos: 31.034 acc5_pos: 100.000 loss: 0.5631 (0.6009) acc1: 72.1354 (68.1530) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.2190 data: 0.1253 max mem: 3531 +Test: [180/269] eta: 0:00:20 acc1_pos: 31.034 acc5_pos: 100.000 loss: 0.4586 (0.5901) acc1: 82.5521 (69.1370) acc5: 100.0000 (100.0000) acc1_neg: 75.781 acc5_neg: 100.000 time: 0.1812 data: 0.0865 max mem: 3531 +Test: [190/269] eta: 0:00:17 acc1_pos: 31.034 acc5_pos: 100.000 loss: 0.4586 (0.5869) acc1: 81.7708 (69.3417) acc5: 100.0000 (100.0000) acc1_neg: 66.927 acc5_neg: 100.000 time: 0.1792 data: 0.0833 max mem: 3531 +Test: [200/269] eta: 0:00:15 acc1_pos: 31.034 acc5_pos: 100.000 loss: 0.5646 (0.5873) acc1: 65.3646 (69.1892) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1791 data: 0.0817 max mem: 3531 +Test: [210/269] eta: 0:00:13 acc1_pos: 31.034 acc5_pos: 100.000 loss: 0.4950 (0.5793) acc1: 80.7292 (69.9114) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.1949 data: 0.0991 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 31.034 acc5_pos: 100.000 loss: 0.3933 (0.5722) acc1: 83.8542 (70.5470) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.2040 data: 0.1086 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 31.034 acc5_pos: 100.000 loss: 0.3873 (0.5638) acc1: 86.1979 (71.2798) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1916 data: 0.0988 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 31.034 acc5_pos: 100.000 loss: 0.4495 (0.5649) acc1: 80.7292 (71.1240) acc5: 100.0000 (100.0000) acc1_neg: 57.552 acc5_neg: 100.000 time: 0.1783 data: 0.0875 max mem: 3532 +Test: [250/269] eta: 0:00:04 acc1_pos: 31.034 acc5_pos: 100.000 loss: 0.5827 (0.5648) acc1: 67.4479 (71.0419) acc5: 100.0000 (100.0000) acc1_neg: 60.938 acc5_neg: 100.000 time: 0.1613 data: 0.0710 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 31.034 acc5_pos: 100.000 loss: 0.5470 (0.5674) acc1: 67.7083 (70.7505) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.1412 data: 0.0503 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 31.034 acc5_pos: 100.000 loss: 0.5620 (0.5665) acc1: 64.5833 (70.7880) acc5: 100.0000 (100.0000) acc1_neg: 87.139 acc5_neg: 100.000 time: 0.1357 data: 0.0415 max mem: 3532 +Test: Total time: 0:00:56 (0.2093 s / it) +* Overall Acc@1 70.788 Acc@5 100.000 + target == real Acc@1 39.028 Acc@5 100.000 + target == fake Acc@1 75.122 Acc@5 100.000 + loss 0.567 + PR-AUC 0.9073 ROC-AUC 0.6033 EER 42.45% (thr=0.6328) +Accuracy of the network on the 103293 test images: 70.8% +[rank0]:[W725 18:46:54.536152954 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-25T18:47:00Z] eval_done tag=fixed500k_real_n8_seed4_resampled +[2026-07-25T18:47:00Z] eval_start tag=fixed500k_fake_n16_seed4_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed4_resampled_n116-041-153_20260725_112559/best_checkpoint.pth +started_utc=2026-07-25T18:47:00Z +hostname=n116-041-153 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed4_resampled_n116-041-153_20260725_112559/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n16_seed4_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0725 18:47:16.783000 610109 torch/distributed/run.py:766] +W0725 18:47:16.783000 610109 torch/distributed/run.py:766] ***************************************** +W0725 18:47:16.783000 610109 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:47:16.783000 610109 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:47:52.103485474 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:47:53.877462502 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:47:53.916714587 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:47:53.027491737 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:47:53.080903317 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:47:53.185492697 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:47:53.309984971 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank6]:[W725 18:47:53.310034540 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n16_seed4_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed4_resampled_n116-041-153_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:24:34 acc1_pos: 54.688 acc5_pos: 100.000 loss: 0.6860 (0.6860) acc1: 54.6875 (54.6875) acc5: 100.0000 (100.0000) time: 6.3011 data: 4.8601 max mem: 3531 +Test: [ 10/234] eta: 0:02:30 acc1_pos: 31.771 acc5_pos: 100.000 loss: 0.6277 (0.6610) acc1: 64.0625 (59.7775) acc5: 100.0000 (100.0000) time: 0.6698 data: 0.4539 max mem: 3531 +Test: [ 20/234] eta: 0:01:25 acc1_pos: 94.792 acc5_pos: 100.000 loss: 0.6277 (0.6653) acc1: 59.6354 (59.2758) acc5: 100.0000 (100.0000) time: 0.1033 data: 0.0069 max mem: 3531 +Test: [ 30/234] eta: 0:01:17 acc1_pos: 82.812 acc5_pos: 100.000 loss: 0.2128 (0.5200) acc1: 94.7917 (70.7829) acc5: 100.0000 (100.0000) time: 0.2199 data: 0.1201 max mem: 3531 +Test: [ 40/234] eta: 0:01:05 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1845 (0.4721) acc1: 96.0938 (75.1080) acc5: 100.0000 (100.0000) time: 0.2789 data: 0.1840 max mem: 3531 +Test: [ 50/234] eta: 0:00:56 acc1_pos: 97.656 acc5_pos: 100.000 loss: 0.1451 (0.4064) acc1: 96.8750 (79.4373) acc5: 100.0000 (100.0000) time: 0.2000 data: 0.1101 max mem: 3531 +Test: [ 60/234] eta: 0:00:50 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.1144 (0.3551) acc1: 98.6979 (82.7527) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1859 data: 0.0929 max mem: 3531 +Test: [ 70/234] eta: 0:00:45 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.0939 (0.3500) acc1: 99.4792 (83.0876) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.2056 data: 0.1104 max mem: 3531 +Test: [ 80/234] eta: 0:00:41 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4109 (0.3515) acc1: 79.4271 (83.3173) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2149 data: 0.1183 max mem: 3531 +Test: [ 90/234] eta: 0:00:37 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.2808 (0.3389) acc1: 89.0625 (84.2205) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.1898 data: 0.0912 max mem: 3531 +Test: [100/234] eta: 0:00:34 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.3387 (0.3625) acc1: 82.2917 (82.9981) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1975 data: 0.1024 max mem: 3531 +Test: [110/234] eta: 0:00:31 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.5208 (0.3714) acc1: 74.4792 (82.6483) acc5: 100.0000 (100.0000) acc1_neg: 69.271 acc5_neg: 100.000 time: 0.2291 data: 0.1351 max mem: 3531 +Test: [120/234] eta: 0:00:28 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4943 (0.3918) acc1: 76.3021 (81.6718) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.2292 data: 0.1353 max mem: 3531 +Test: [130/234] eta: 0:00:25 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4890 (0.4005) acc1: 77.0833 (81.3494) acc5: 100.0000 (100.0000) acc1_neg: 66.667 acc5_neg: 100.000 time: 0.2131 data: 0.1219 max mem: 3531 +Test: [140/234] eta: 0:00:23 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.5493 (0.4393) acc1: 70.8333 (79.3274) acc5: 100.0000 (100.0000) acc1_neg: 28.125 acc5_neg: 100.000 time: 0.2033 data: 0.1120 max mem: 3531 +Test: [150/234] eta: 0:00:20 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.6558 (0.4578) acc1: 65.6250 (78.2526) acc5: 100.0000 (100.0000) acc1_neg: 66.667 acc5_neg: 100.000 time: 0.2070 data: 0.1129 max mem: 3531 +Test: [160/234] eta: 0:00:17 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.5823 (0.4637) acc1: 66.6667 (77.8759) acc5: 100.0000 (100.0000) acc1_neg: 59.896 acc5_neg: 100.000 time: 0.2107 data: 0.1175 max mem: 3531 +Test: [170/234] eta: 0:00:15 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.6679 (0.4744) acc1: 61.9792 (77.1412) acc5: 100.0000 (100.0000) acc1_neg: 71.615 acc5_neg: 100.000 time: 0.2051 data: 0.1146 max mem: 3531 +Test: [180/234] eta: 0:00:12 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.6679 (0.4852) acc1: 63.8021 (76.5884) acc5: 100.0000 (100.0000) acc1_neg: 34.375 acc5_neg: 100.000 time: 0.1999 data: 0.1067 max mem: 3531 +Test: [190/234] eta: 0:00:10 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.7556 (0.5016) acc1: 61.1979 (75.6872) acc5: 100.0000 (100.0000) acc1_neg: 58.073 acc5_neg: 100.000 time: 0.1997 data: 0.1061 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.7556 (0.4952) acc1: 61.4583 (76.1712) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1884 data: 0.0976 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.1453 (0.4827) acc1: 98.9583 (76.9760) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.1620 data: 0.0717 max mem: 3532 +Test: [220/234] eta: 0:00:03 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.3945 (0.4801) acc1: 80.7292 (77.0716) acc5: 100.0000 (100.0000) acc1_neg: 73.958 acc5_neg: 100.000 time: 0.1397 data: 0.0494 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4983 (0.4917) acc1: 60.9375 (76.1815) acc5: 100.0000 (100.0000) acc1_neg: 57.031 acc5_neg: 100.000 time: 0.1338 data: 0.0438 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.6969 (0.4946) acc1: 60.1562 (76.0022) acc5: 100.0000 (100.0000) acc1_neg: 60.311 acc5_neg: 100.000 time: 0.1410 data: 0.0405 max mem: 3532 +Test: Total time: 0:00:51 (0.2196 s / it) +* Overall Acc@1 76.002 Acc@5 100.000 + target == real Acc@1 79.573 Acc@5 100.000 + target == fake Acc@1 74.996 Acc@5 100.000 + loss 0.495 + PR-AUC 0.9579 ROC-AUC 0.8637 EER 23.02% (thr=0.4685) +Accuracy of the network on the 89729 test images: 76.0% +[rank0]:[W725 18:48:54.797387559 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:49:16.599000 610660 torch/distributed/run.py:766] +W0725 18:49:16.599000 610660 torch/distributed/run.py:766] ***************************************** +W0725 18:49:16.599000 610660 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:49:16.599000 610660 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:49:53.247216367 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:49:54.866331762 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:49:54.972413461 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:49:54.080924530 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:49:54.425652299 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +| distributed init (rank 5): env:// +[rank0]:[W725 18:49:54.446715504 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:49:54.450691371 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:49:54.455520401 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n16_seed4_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed4_resampled_n116-041-153_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:22:04 acc1_pos: 28.646 acc5_pos: 100.000 loss: 1.0921 (1.0921) acc1: 28.6458 (28.6458) acc5: 100.0000 (100.0000) time: 6.5878 data: 5.2767 max mem: 3531 +Test: [ 10/2387] eta: 0:26:57 acc1_pos: 14.062 acc5_pos: 100.000 loss: 1.0386 (1.0226) acc1: 22.1354 (26.4915) acc5: 100.0000 (100.0000) time: 0.6806 data: 0.4801 max mem: 3531 +Test: [ 20/2387] eta: 0:15:48 acc1_pos: 8.854 acc5_pos: 100.000 loss: 1.0386 (1.0652) acc1: 20.3125 (24.1071) acc5: 100.0000 (100.0000) time: 0.0915 data: 0.0005 max mem: 3531 +Test: [ 30/2387] eta: 0:13:14 acc1_pos: 9.635 acc5_pos: 100.000 loss: 1.1432 (1.0965) acc1: 17.9688 (21.6314) acc5: 100.0000 (100.0000) time: 0.1480 data: 0.0567 max mem: 3531 +Test: [ 40/2387] eta: 0:12:20 acc1_pos: 13.802 acc5_pos: 100.000 loss: 1.2234 (1.1239) acc1: 12.7604 (19.2518) acc5: 100.0000 (100.0000) time: 0.2262 data: 0.1334 max mem: 3531 +Test: [ 50/2387] eta: 0:11:37 acc1_pos: 30.208 acc5_pos: 100.000 loss: 1.1458 (1.1142) acc1: 13.8021 (20.3227) acc5: 100.0000 (100.0000) time: 0.2385 data: 0.1421 max mem: 3531 +Test: [ 60/2387] eta: 0:11:00 acc1_pos: 44.271 acc5_pos: 100.000 loss: 1.0639 (1.1125) acc1: 21.8750 (20.9699) acc5: 100.0000 (100.0000) time: 0.2187 data: 0.1215 max mem: 3531 +Test: [ 70/2387] eta: 0:10:40 acc1_pos: 19.531 acc5_pos: 100.000 loss: 1.1423 (1.1253) acc1: 18.7500 (20.5766) acc5: 100.0000 (100.0000) time: 0.2199 data: 0.1230 max mem: 3531 +Test: [ 80/2387] eta: 0:10:26 acc1_pos: 37.760 acc5_pos: 100.000 loss: 1.1155 (1.1169) acc1: 19.2708 (21.2802) acc5: 100.0000 (100.0000) time: 0.2348 data: 0.1414 max mem: 3531 +Test: [ 90/2387] eta: 0:10:18 acc1_pos: 37.240 acc5_pos: 100.000 loss: 1.0155 (1.0980) acc1: 31.7708 (22.6992) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1499 max mem: 3531 +Test: [ 100/2387] eta: 0:10:09 acc1_pos: 30.469 acc5_pos: 100.000 loss: 0.9343 (1.0910) acc1: 33.5938 (23.0404) acc5: 100.0000 (100.0000) time: 0.2451 data: 0.1485 max mem: 3531 +Test: [ 110/2387] eta: 0:09:59 acc1_pos: 42.188 acc5_pos: 100.000 loss: 0.9002 (1.0729) acc1: 30.4688 (24.4463) acc5: 100.0000 (100.0000) time: 0.2352 data: 0.1433 max mem: 3531 +Test: [ 120/2387] eta: 0:09:50 acc1_pos: 26.302 acc5_pos: 100.000 loss: 0.9310 (1.0633) acc1: 34.6354 (25.3207) acc5: 100.0000 (100.0000) time: 0.2304 data: 0.1391 max mem: 3531 +Test: [ 130/2387] eta: 0:09:42 acc1_pos: 33.073 acc5_pos: 100.000 loss: 0.9790 (1.0576) acc1: 31.5104 (25.6500) acc5: 100.0000 (100.0000) time: 0.2295 data: 0.1353 max mem: 3531 +Test: [ 140/2387] eta: 0:09:34 acc1_pos: 37.760 acc5_pos: 100.000 loss: 0.9731 (1.0498) acc1: 30.2083 (26.1026) acc5: 100.0000 (100.0000) time: 0.2252 data: 0.1273 max mem: 3531 +Test: [ 150/2387] eta: 0:09:30 acc1_pos: 9.115 acc5_pos: 100.000 loss: 0.8960 (1.0381) acc1: 37.2396 (26.9126) acc5: 100.0000 (100.0000) time: 0.2348 data: 0.1363 max mem: 3531 +Test: [ 160/2387] eta: 0:09:30 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.8456 (1.0172) acc1: 42.4479 (28.6685) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2619 data: 0.1642 max mem: 3531 +Test: [ 170/2387] eta: 0:09:17 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2844 (0.9738) acc1: 93.4896 (32.5551) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2249 data: 0.1274 max mem: 3531 +Test: [ 180/2387] eta: 0:09:00 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2707 (0.9349) acc1: 95.3125 (36.0382) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1528 data: 0.0571 max mem: 3531 +Test: [ 190/2387] eta: 0:08:58 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2688 (0.9001) acc1: 95.3125 (39.1320) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1924 data: 0.0992 max mem: 3531 +Test: [ 200/2387] eta: 0:09:03 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2678 (0.8687) acc1: 95.0521 (41.9206) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2797 data: 0.1886 max mem: 3531 +Test: [ 210/2387] eta: 0:08:55 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2723 (0.8406) acc1: 95.0521 (44.4375) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2529 data: 0.1594 max mem: 3532 +Test: [ 220/2387] eta: 0:08:43 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2753 (0.8150) acc1: 95.0521 (46.7254) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1740 data: 0.0767 max mem: 3532 +Test: [ 230/2387] eta: 0:08:39 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2751 (0.7915) acc1: 95.0521 (48.8197) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1844 data: 0.0880 max mem: 3532 +Test: [ 240/2387] eta: 0:08:33 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2808 (0.7705) acc1: 94.7917 (50.7197) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2099 data: 0.1130 max mem: 3532 +Test: [ 250/2387] eta: 0:08:22 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2474 (0.7460) acc1: 95.8333 (52.6270) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0731 max mem: 3532 +Test: [ 260/2387] eta: 0:08:15 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1412 (0.7227) acc1: 99.4792 (54.4191) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1600 data: 0.0656 max mem: 3532 +Test: [ 270/2387] eta: 0:08:07 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1317 (0.7008) acc1: 99.4792 (56.0818) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1699 data: 0.0751 max mem: 3532 +Test: [ 280/2387] eta: 0:08:04 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1308 (0.6806) acc1: 99.7396 (57.6281) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1915 data: 0.0975 max mem: 3532 +Test: [ 290/2387] eta: 0:07:59 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1239 (0.6613) acc1: 99.7396 (59.0761) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2076 data: 0.1160 max mem: 3532 +Test: [ 300/2387] eta: 0:07:49 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1122 (0.6429) acc1: 100.0000 (60.4296) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1578 data: 0.0638 max mem: 3532 +Test: [ 310/2387] eta: 0:07:45 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1159 (0.6331) acc1: 100.0000 (61.1954) acc5: 100.0000 (100.0000) acc1_neg: 67.969 acc5_neg: 100.000 time: 0.1568 data: 0.0574 max mem: 3532 +Test: [ 320/2387] eta: 0:07:41 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.6137 (0.6328) acc1: 66.4062 (61.3196) acc5: 100.0000 (100.0000) acc1_neg: 64.062 acc5_neg: 100.000 time: 0.1949 data: 0.0958 max mem: 3532 +Test: [ 330/2387] eta: 0:07:41 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.6199 (0.6324) acc1: 64.5833 (61.4560) acc5: 100.0000 (100.0000) acc1_neg: 66.406 acc5_neg: 100.000 time: 0.2257 data: 0.1313 max mem: 3532 +Test: [ 340/2387] eta: 0:07:35 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.6166 (0.6318) acc1: 66.9271 (61.6340) acc5: 100.0000 (100.0000) acc1_neg: 65.365 acc5_neg: 100.000 time: 0.2114 data: 0.1168 max mem: 3532 +Test: [ 350/2387] eta: 0:07:33 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.5942 (0.6247) acc1: 70.0521 (62.5223) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1941 data: 0.0980 max mem: 3532 +Test: [ 360/2387] eta: 0:07:27 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.3380 (0.6136) acc1: 97.1354 (63.4905) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1907 data: 0.0938 max mem: 3532 +Test: [ 370/2387] eta: 0:07:21 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2145 (0.6029) acc1: 97.9167 (64.4191) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1559 data: 0.0612 max mem: 3532 +Test: [ 380/2387] eta: 0:07:17 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2152 (0.5928) acc1: 97.9167 (65.2914) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1648 data: 0.0719 max mem: 3532 +Test: [ 390/2387] eta: 0:07:15 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2147 (0.5830) acc1: 98.1771 (66.1305) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2056 data: 0.1110 max mem: 3532 +Test: [ 400/2387] eta: 0:07:10 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2147 (0.5739) acc1: 98.1771 (66.9206) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1926 data: 0.0992 max mem: 3532 +Test: [ 410/2387] eta: 0:07:06 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2117 (0.5651) acc1: 97.6562 (67.6703) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1632 data: 0.0709 max mem: 3532 +Test: [ 420/2387] eta: 0:07:01 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2100 (0.5567) acc1: 97.9167 (68.3875) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1703 data: 0.0789 max mem: 3532 +Test: [ 430/2387] eta: 0:06:59 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2198 (0.5491) acc1: 97.6562 (69.0485) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1949 data: 0.1004 max mem: 3532 +Test: [ 440/2387] eta: 0:06:57 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2189 (0.5396) acc1: 97.6562 (69.7226) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2094 data: 0.1129 max mem: 3533 +Test: [ 450/2387] eta: 0:06:54 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.0681 (0.5291) acc1: 100.0000 (70.3939) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1999 data: 0.1058 max mem: 3533 +Test: [ 460/2387] eta: 0:06:51 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.0679 (0.5191) acc1: 100.0000 (71.0361) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.1022 max mem: 3533 +Test: [ 470/2387] eta: 0:06:48 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.0677 (0.5095) acc1: 100.0000 (71.6511) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1924 data: 0.1011 max mem: 3533 +Test: [ 480/2387] eta: 0:06:45 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.0677 (0.5004) acc1: 100.0000 (72.2405) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2004 data: 0.1100 max mem: 3533 +Test: [ 490/2387] eta: 0:06:42 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.0681 (0.4916) acc1: 100.0000 (72.8058) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1998 data: 0.1090 max mem: 3533 +Test: [ 500/2387] eta: 0:06:39 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.0682 (0.4831) acc1: 100.0000 (73.3486) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1907 data: 0.1003 max mem: 3533 +Test: [ 510/2387] eta: 0:06:37 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.0681 (0.4753) acc1: 100.0000 (73.8702) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1948 data: 0.1042 max mem: 3533 +Test: [ 520/2387] eta: 0:06:37 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1193 (0.4686) acc1: 100.0000 (74.3717) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2327 data: 0.1410 max mem: 3533 +Test: [ 530/2387] eta: 0:06:35 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1202 (0.4620) acc1: 100.0000 (74.8543) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2391 data: 0.1476 max mem: 3533 +Test: [ 540/2387] eta: 0:06:29 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1191 (0.4557) acc1: 100.0000 (75.3187) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1611 data: 0.0692 max mem: 3533 +Test: [ 550/2387] eta: 0:06:26 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1200 (0.4496) acc1: 100.0000 (75.7661) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1534 data: 0.0617 max mem: 3533 +Test: [ 560/2387] eta: 0:06:24 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1203 (0.4437) acc1: 100.0000 (76.1981) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1995 data: 0.1079 max mem: 3533 +Test: [ 570/2387] eta: 0:06:21 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1200 (0.4380) acc1: 100.0000 (76.6149) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1979 data: 0.1059 max mem: 3533 +Test: [ 580/2387] eta: 0:06:19 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1190 (0.4326) acc1: 100.0000 (77.0174) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1955 data: 0.1037 max mem: 3533 +Test: [ 590/2387] eta: 0:06:16 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1192 (0.4272) acc1: 100.0000 (77.4063) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1919 data: 0.0989 max mem: 3533 +Test: [ 600/2387] eta: 0:06:13 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1210 (0.4223) acc1: 100.0000 (77.7805) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1770 data: 0.0844 max mem: 3533 +Test: [ 610/2387] eta: 0:06:10 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1681 (0.4196) acc1: 97.1354 (78.0739) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1749 data: 0.0831 max mem: 3533 +Test: [ 620/2387] eta: 0:06:07 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2555 (0.4170) acc1: 95.5729 (78.3624) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1793 data: 0.0878 max mem: 3533 +Test: [ 630/2387] eta: 0:06:04 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2520 (0.4145) acc1: 95.8333 (78.6392) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1749 data: 0.0830 max mem: 3533 +Test: [ 640/2387] eta: 0:06:01 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2638 (0.4122) acc1: 95.5729 (78.8945) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1754 data: 0.0837 max mem: 3533 +Test: [ 650/2387] eta: 0:05:58 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2634 (0.4099) acc1: 95.0521 (79.1507) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1707 data: 0.0797 max mem: 3533 +Test: [ 660/2387] eta: 0:05:55 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2588 (0.4076) acc1: 95.3125 (79.4046) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1680 data: 0.0774 max mem: 3533 +Test: [ 670/2387] eta: 0:05:52 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2574 (0.4054) acc1: 95.3125 (79.6433) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1680 data: 0.0770 max mem: 3534 +Test: [ 680/2387] eta: 0:05:49 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2526 (0.4026) acc1: 95.8333 (79.9204) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1666 data: 0.0754 max mem: 3534 +Test: [ 690/2387] eta: 0:05:46 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2580 (0.4023) acc1: 95.0521 (79.9720) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.1734 data: 0.0811 max mem: 3534 +Test: [ 700/2387] eta: 0:05:43 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.4363 (0.4028) acc1: 77.6042 (79.9397) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.1825 data: 0.0873 max mem: 3534 +Test: [ 710/2387] eta: 0:05:41 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.4399 (0.4034) acc1: 77.6042 (79.9040) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.1849 data: 0.0904 max mem: 3534 +Test: [ 720/2387] eta: 0:05:38 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.4387 (0.4038) acc1: 77.6042 (79.8825) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.1806 data: 0.0876 max mem: 3534 +Test: [ 730/2387] eta: 0:05:36 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.4424 (0.4044) acc1: 78.3854 (79.8546) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.1779 data: 0.0858 max mem: 3534 +Test: [ 740/2387] eta: 0:05:33 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.4445 (0.4050) acc1: 76.5625 (79.8126) acc5: 100.0000 (100.0000) acc1_neg: 74.219 acc5_neg: 100.000 time: 0.1800 data: 0.0889 max mem: 3534 +Test: [ 750/2387] eta: 0:05:31 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.4367 (0.4054) acc1: 77.3438 (79.7915) acc5: 100.0000 (100.0000) acc1_neg: 78.646 acc5_neg: 100.000 time: 0.1834 data: 0.0916 max mem: 3534 +Test: [ 760/2387] eta: 0:05:28 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.4262 (0.4049) acc1: 78.6458 (79.8405) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1817 data: 0.0906 max mem: 3534 +Test: [ 770/2387] eta: 0:05:26 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2716 (0.4026) acc1: 90.1042 (80.0280) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1809 data: 0.0905 max mem: 3534 +Test: [ 780/2387] eta: 0:05:23 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1490 (0.3993) acc1: 99.7396 (80.2827) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1838 data: 0.0924 max mem: 3534 +Test: [ 790/2387] eta: 0:05:21 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1426 (0.3965) acc1: 100.0000 (80.4740) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1839 data: 0.0921 max mem: 3534 +Test: [ 800/2387] eta: 0:05:19 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1377 (0.3932) acc1: 99.7396 (80.7129) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1927 data: 0.1018 max mem: 3534 +Test: [ 810/2387] eta: 0:05:17 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1379 (0.3903) acc1: 100.0000 (80.9424) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1952 data: 0.1033 max mem: 3534 +Test: [ 820/2387] eta: 0:05:14 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1429 (0.3874) acc1: 99.7396 (81.1555) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1822 data: 0.0899 max mem: 3534 +Test: [ 830/2387] eta: 0:05:12 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1996 (0.3858) acc1: 97.1354 (81.3117) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1833 data: 0.0893 max mem: 3534 +Test: [ 840/2387] eta: 0:05:10 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.3408 (0.3865) acc1: 86.4583 (81.3067) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.2022 data: 0.1084 max mem: 3534 +Test: [ 850/2387] eta: 0:05:08 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.4426 (0.3878) acc1: 78.1250 (81.2549) acc5: 100.0000 (100.0000) acc1_neg: 69.792 acc5_neg: 100.000 time: 0.1995 data: 0.1090 max mem: 3534 +Test: [ 860/2387] eta: 0:05:06 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.4344 (0.3882) acc1: 81.5104 (81.2739) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.1915 data: 0.1014 max mem: 3534 +Test: [ 870/2387] eta: 0:05:04 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.4052 (0.3885) acc1: 85.4167 (81.3134) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.1874 data: 0.0968 max mem: 3534 +Test: [ 880/2387] eta: 0:05:02 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.4126 (0.3893) acc1: 83.5938 (81.3044) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.2084 data: 0.1157 max mem: 3534 +Test: [ 890/2387] eta: 0:05:00 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.3744 (0.3884) acc1: 85.4167 (81.4233) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.2018 data: 0.1060 max mem: 3534 +Test: [ 900/2387] eta: 0:04:56 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.3650 (0.3924) acc1: 83.5938 (81.1939) acc5: 100.0000 (100.0000) acc1_neg: 52.604 acc5_neg: 100.000 time: 0.1402 data: 0.0457 max mem: 3535 +Test: [ 910/2387] eta: 0:04:55 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.8304 (0.3972) acc1: 52.0833 (80.8755) acc5: 100.0000 (100.0000) acc1_neg: 51.823 acc5_neg: 100.000 time: 0.1735 data: 0.0816 max mem: 3535 +Test: [ 920/2387] eta: 0:04:52 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.8430 (0.4022) acc1: 51.8229 (80.5621) acc5: 100.0000 (100.0000) acc1_neg: 49.740 acc5_neg: 100.000 time: 0.2066 data: 0.1149 max mem: 3535 +Test: [ 930/2387] eta: 0:04:50 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.8422 (0.4069) acc1: 51.8229 (80.2620) acc5: 100.0000 (100.0000) acc1_neg: 47.135 acc5_neg: 100.000 time: 0.1760 data: 0.0853 max mem: 3535 +Test: [ 940/2387] eta: 0:04:48 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.8371 (0.4115) acc1: 51.8229 (79.9723) acc5: 100.0000 (100.0000) acc1_neg: 49.740 acc5_neg: 100.000 time: 0.1928 data: 0.1008 max mem: 3535 +Test: [ 950/2387] eta: 0:04:46 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.8266 (0.4160) acc1: 53.1250 (79.6815) acc5: 100.0000 (100.0000) acc1_neg: 53.646 acc5_neg: 100.000 time: 0.2083 data: 0.1140 max mem: 3535 +Test: [ 960/2387] eta: 0:04:44 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.8282 (0.4204) acc1: 52.8646 (79.3992) acc5: 100.0000 (100.0000) acc1_neg: 49.740 acc5_neg: 100.000 time: 0.1899 data: 0.0957 max mem: 3535 +Test: [ 970/2387] eta: 0:04:42 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.8332 (0.4247) acc1: 51.8229 (79.1117) acc5: 100.0000 (100.0000) acc1_neg: 52.083 acc5_neg: 100.000 time: 0.1891 data: 0.0968 max mem: 3535 +Test: [ 980/2387] eta: 0:04:40 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.8156 (0.4245) acc1: 53.6458 (79.1465) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.1978 data: 0.1066 max mem: 3535 +Test: [ 990/2387] eta: 0:04:38 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1351 (0.4217) acc1: 99.4792 (79.3359) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1852 data: 0.0943 max mem: 3535 +Test: [1000/2387] eta: 0:04:35 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1041 (0.4186) acc1: 100.0000 (79.5259) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1772 data: 0.0860 max mem: 3535 +Test: [1010/2387] eta: 0:04:33 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1227 (0.4159) acc1: 99.7396 (79.7071) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1769 data: 0.0838 max mem: 3535 +Test: [1020/2387] eta: 0:04:31 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1438 (0.4138) acc1: 98.6979 (79.8609) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1817 data: 0.0890 max mem: 3535 +Test: [1030/2387] eta: 0:04:29 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1454 (0.4113) acc1: 98.6979 (80.0383) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2026 data: 0.1105 max mem: 3535 +Test: [1040/2387] eta: 0:04:27 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1320 (0.4086) acc1: 99.7396 (80.2251) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2078 data: 0.1154 max mem: 3535 +Test: [1050/2387] eta: 0:04:26 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1299 (0.4060) acc1: 99.7396 (80.4085) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2114 data: 0.1177 max mem: 3535 +Test: [1060/2387] eta: 0:04:23 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1288 (0.4035) acc1: 99.7396 (80.5829) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2049 data: 0.1112 max mem: 3535 +Test: [1070/2387] eta: 0:04:21 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1441 (0.4014) acc1: 99.2188 (80.7408) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1934 data: 0.0999 max mem: 3535 +Test: [1080/2387] eta: 0:04:19 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2074 (0.3999) acc1: 97.3958 (80.8764) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2006 data: 0.1055 max mem: 3535 +Test: [1090/2387] eta: 0:04:18 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1310 (0.3972) acc1: 99.4792 (81.0505) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2062 data: 0.1131 max mem: 3535 +Test: [1100/2387] eta: 0:04:16 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1063 (0.3947) acc1: 100.0000 (81.2190) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2116 data: 0.1163 max mem: 3535 +Test: [1110/2387] eta: 0:04:14 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1095 (0.3927) acc1: 99.7396 (81.3419) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2005 data: 0.1046 max mem: 3535 +Test: [1120/2387] eta: 0:04:11 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1225 (0.3903) acc1: 99.4792 (81.5016) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1802 data: 0.0874 max mem: 3536 +Test: [1130/2387] eta: 0:04:10 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1215 (0.3879) acc1: 100.0000 (81.6633) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1930 data: 0.1007 max mem: 3536 +Test: [1140/2387] eta: 0:04:07 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1189 (0.3856) acc1: 100.0000 (81.8163) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1888 data: 0.0958 max mem: 3536 +Test: [1150/2387] eta: 0:04:05 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1601 (0.3856) acc1: 97.1354 (81.8265) acc5: 100.0000 (100.0000) acc1_neg: 63.021 acc5_neg: 100.000 time: 0.1772 data: 0.0844 max mem: 3536 +Test: [1160/2387] eta: 0:04:03 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.3673 (0.3856) acc1: 84.8958 (81.8478) acc5: 100.0000 (100.0000) acc1_neg: 74.740 acc5_neg: 100.000 time: 0.1755 data: 0.0836 max mem: 3536 +Test: [1170/2387] eta: 0:04:01 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.4002 (0.3858) acc1: 84.6354 (81.8656) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1827 data: 0.0904 max mem: 3536 +Test: [1180/2387] eta: 0:03:59 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.4289 (0.3861) acc1: 84.6354 (81.8795) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.1909 data: 0.0982 max mem: 3536 +Test: [1190/2387] eta: 0:03:56 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.3800 (0.3862) acc1: 85.6771 (81.8935) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1808 data: 0.0888 max mem: 3536 +Test: [1200/2387] eta: 0:03:54 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.3199 (0.3853) acc1: 91.1458 (81.9831) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1837 data: 0.0925 max mem: 3536 +Test: [1210/2387] eta: 0:03:52 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2740 (0.3845) acc1: 92.9688 (82.0728) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1795 data: 0.0883 max mem: 3536 +Test: [1220/2387] eta: 0:03:50 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2428 (0.3829) acc1: 96.0938 (82.1914) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1757 data: 0.0853 max mem: 3536 +Test: [1230/2387] eta: 0:03:48 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1852 (0.3813) acc1: 96.6146 (82.3126) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1824 data: 0.0916 max mem: 3536 +Test: [1240/2387] eta: 0:03:46 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1824 (0.3797) acc1: 97.3958 (82.4388) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1884 data: 0.0975 max mem: 3536 +Test: [1250/2387] eta: 0:03:44 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1777 (0.3781) acc1: 98.1771 (82.5671) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1875 data: 0.0959 max mem: 3536 +Test: [1260/2387] eta: 0:03:42 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1645 (0.3764) acc1: 98.9583 (82.6929) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1846 data: 0.0918 max mem: 3536 +Test: [1270/2387] eta: 0:03:40 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1428 (0.3745) acc1: 99.4792 (82.8264) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1903 data: 0.0978 max mem: 3536 +Test: [1280/2387] eta: 0:03:38 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1265 (0.3725) acc1: 99.7396 (82.9564) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1892 data: 0.0965 max mem: 3536 +Test: [1290/2387] eta: 0:03:36 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1168 (0.3706) acc1: 99.4792 (83.0683) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2039 data: 0.1105 max mem: 3536 +Test: [1300/2387] eta: 0:03:33 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1169 (0.3687) acc1: 99.2188 (83.1870) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1859 data: 0.0894 max mem: 3536 +Test: [1310/2387] eta: 0:03:32 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1240 (0.3670) acc1: 99.2188 (83.3023) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1751 data: 0.0791 max mem: 3536 +Test: [1320/2387] eta: 0:03:30 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1346 (0.3652) acc1: 99.2188 (83.4146) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1992 data: 0.1073 max mem: 3536 +Test: [1330/2387] eta: 0:03:27 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1503 (0.3637) acc1: 99.2188 (83.5253) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1797 data: 0.0868 max mem: 3536 +Test: [1340/2387] eta: 0:03:25 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1872 (0.3636) acc1: 96.8750 (83.5390) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1754 data: 0.0824 max mem: 3536 +Test: [1350/2387] eta: 0:03:23 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1105 (0.3616) acc1: 100.0000 (83.6608) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1905 data: 0.0987 max mem: 3537 +Test: [1360/2387] eta: 0:03:22 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.0992 (0.3597) acc1: 100.0000 (83.7809) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2137 data: 0.1218 max mem: 3537 +Test: [1370/2387] eta: 0:03:20 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.0990 (0.3578) acc1: 100.0000 (83.8992) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2149 data: 0.1236 max mem: 3537 +Test: [1380/2387] eta: 0:03:18 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.0991 (0.3560) acc1: 100.0000 (84.0158) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2055 data: 0.1115 max mem: 3537 +Test: [1390/2387] eta: 0:03:16 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.0998 (0.3541) acc1: 100.0000 (84.1303) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2019 data: 0.1057 max mem: 3537 +Test: [1400/2387] eta: 0:03:14 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.0864 (0.3522) acc1: 100.0000 (84.2436) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1712 data: 0.0780 max mem: 3537 +Test: [1410/2387] eta: 0:03:12 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.0918 (0.3510) acc1: 100.0000 (84.3410) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1735 data: 0.0820 max mem: 3537 +Test: [1420/2387] eta: 0:03:09 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1993 (0.3501) acc1: 98.4375 (84.4153) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1760 data: 0.0848 max mem: 3537 +Test: [1430/2387] eta: 0:03:07 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2111 (0.3492) acc1: 96.8750 (84.5028) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1653 data: 0.0735 max mem: 3537 +Test: [1440/2387] eta: 0:03:05 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2011 (0.3481) acc1: 98.1771 (84.6040) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1840 data: 0.0903 max mem: 3537 +Test: [1450/2387] eta: 0:03:03 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1996 (0.3473) acc1: 98.4375 (84.6878) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1795 data: 0.0869 max mem: 3537 +Test: [1460/2387] eta: 0:03:01 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2103 (0.3463) acc1: 97.3958 (84.7655) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1762 data: 0.0856 max mem: 3537 +Test: [1470/2387] eta: 0:02:59 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1922 (0.3452) acc1: 98.6979 (84.8606) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1856 data: 0.0945 max mem: 3537 +Test: [1480/2387] eta: 0:02:57 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1653 (0.3441) acc1: 98.6979 (84.9503) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1841 data: 0.0912 max mem: 3537 +Test: [1490/2387] eta: 0:02:55 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1784 (0.3429) acc1: 98.4375 (85.0413) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1781 data: 0.0855 max mem: 3537 +Test: [1500/2387] eta: 0:02:53 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1784 (0.3419) acc1: 98.6979 (85.1304) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1927 data: 0.1002 max mem: 3537 +Test: [1510/2387] eta: 0:02:51 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1920 (0.3409) acc1: 98.6979 (85.2142) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1999 data: 0.1072 max mem: 3537 +Test: [1520/2387] eta: 0:02:49 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2358 (0.3412) acc1: 92.7083 (85.2088) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1970 data: 0.1043 max mem: 3537 +Test: [1530/2387] eta: 0:02:47 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.4176 (0.3420) acc1: 84.6354 (85.1779) acc5: 100.0000 (100.0000) acc1_neg: 75.781 acc5_neg: 100.000 time: 0.1976 data: 0.1044 max mem: 3537 +Test: [1540/2387] eta: 0:02:45 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.4018 (0.3423) acc1: 84.8958 (85.1860) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.1989 data: 0.1060 max mem: 3537 +Test: [1550/2387] eta: 0:02:43 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.3748 (0.3425) acc1: 87.2396 (85.2028) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1991 data: 0.1057 max mem: 3537 +Test: [1560/2387] eta: 0:02:41 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.3873 (0.3428) acc1: 86.7188 (85.2015) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.1835 data: 0.0887 max mem: 3537 +Test: [1570/2387] eta: 0:02:39 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.3286 (0.3425) acc1: 88.5417 (85.2506) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1682 data: 0.0739 max mem: 3537 +Test: [1580/2387] eta: 0:02:37 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2172 (0.3415) acc1: 97.3958 (85.3323) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1659 data: 0.0728 max mem: 3538 +Test: [1590/2387] eta: 0:02:35 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1425 (0.3402) acc1: 100.0000 (85.4231) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1746 data: 0.0792 max mem: 3538 +Test: [1600/2387] eta: 0:02:33 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1366 (0.3391) acc1: 100.0000 (85.4821) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1713 data: 0.0757 max mem: 3538 +Test: [1610/2387] eta: 0:02:31 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1363 (0.3379) acc1: 99.7396 (85.5702) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1654 data: 0.0713 max mem: 3538 +Test: [1620/2387] eta: 0:02:29 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1364 (0.3367) acc1: 99.7396 (85.6536) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1648 data: 0.0697 max mem: 3538 +Test: [1630/2387] eta: 0:02:27 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1476 (0.3356) acc1: 99.2188 (85.7290) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1697 data: 0.0753 max mem: 3538 +Test: [1640/2387] eta: 0:02:25 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1640 (0.3345) acc1: 98.9583 (85.8036) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1941 data: 0.1023 max mem: 3538 +Test: [1650/2387] eta: 0:02:23 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1514 (0.3334) acc1: 98.6979 (85.8815) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1868 data: 0.0957 max mem: 3538 +Test: [1660/2387] eta: 0:02:21 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1496 (0.3324) acc1: 98.6979 (85.9566) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1606 data: 0.0701 max mem: 3538 +Test: [1670/2387] eta: 0:02:19 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1496 (0.3314) acc1: 98.9583 (86.0324) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1598 data: 0.0693 max mem: 3538 +Test: [1680/2387] eta: 0:02:16 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1394 (0.3304) acc1: 99.2188 (86.0986) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1610 data: 0.0690 max mem: 3538 +Test: [1690/2387] eta: 0:02:14 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1560 (0.3295) acc1: 97.6562 (86.1611) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1626 data: 0.0706 max mem: 3538 +Test: [1700/2387] eta: 0:02:12 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1650 (0.3286) acc1: 97.6562 (86.2287) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1730 data: 0.0820 max mem: 3538 +Test: [1710/2387] eta: 0:02:10 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1638 (0.3277) acc1: 98.1771 (86.2921) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1903 data: 0.0992 max mem: 3538 +Test: [1720/2387] eta: 0:02:08 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1743 (0.3268) acc1: 97.9167 (86.3588) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1611 data: 0.0678 max mem: 3538 +Test: [1730/2387] eta: 0:02:06 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1754 (0.3260) acc1: 97.9167 (86.4222) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1605 data: 0.0676 max mem: 3538 +Test: [1740/2387] eta: 0:02:04 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1611 (0.3251) acc1: 98.4375 (86.4911) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1749 data: 0.0840 max mem: 3538 +Test: [1750/2387] eta: 0:02:02 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1522 (0.3242) acc1: 99.2188 (86.5587) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1653 data: 0.0742 max mem: 3538 +Test: [1760/2387] eta: 0:02:00 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1493 (0.3232) acc1: 99.4792 (86.6321) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1735 data: 0.0822 max mem: 3538 +Test: [1770/2387] eta: 0:01:58 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1570 (0.3226) acc1: 98.4375 (86.6823) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1955 data: 0.1045 max mem: 3538 +Test: [1780/2387] eta: 0:01:57 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2192 (0.3222) acc1: 95.8333 (86.7350) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2108 data: 0.1203 max mem: 3538 +Test: [1790/2387] eta: 0:01:55 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2149 (0.3216) acc1: 96.0938 (86.7883) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2185 data: 0.1257 max mem: 3538 +Test: [1800/2387] eta: 0:01:53 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2109 (0.3210) acc1: 96.3542 (86.8414) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2376 data: 0.1437 max mem: 3538 +Test: [1810/2387] eta: 0:01:51 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2035 (0.3202) acc1: 97.1354 (86.9008) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2164 data: 0.1226 max mem: 3539 +Test: [1820/2387] eta: 0:01:49 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1707 (0.3194) acc1: 98.1771 (86.9630) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1869 data: 0.0937 max mem: 3539 +Test: [1830/2387] eta: 0:01:47 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1516 (0.3185) acc1: 99.4792 (87.0315) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1876 data: 0.0948 max mem: 3539 +Test: [1840/2387] eta: 0:01:45 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1361 (0.3175) acc1: 100.0000 (87.1017) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1911 data: 0.0977 max mem: 3539 +Test: [1850/2387] eta: 0:01:43 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1351 (0.3165) acc1: 100.0000 (87.1708) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0866 max mem: 3539 +Test: [1860/2387] eta: 0:01:41 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1347 (0.3155) acc1: 100.0000 (87.2394) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1784 data: 0.0860 max mem: 3539 +Test: [1870/2387] eta: 0:01:39 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1334 (0.3146) acc1: 100.0000 (87.3071) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1955 data: 0.1036 max mem: 3539 +Test: [1880/2387] eta: 0:01:37 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1332 (0.3136) acc1: 100.0000 (87.3742) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1940 data: 0.1024 max mem: 3539 +Test: [1890/2387] eta: 0:01:35 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1334 (0.3126) acc1: 100.0000 (87.4409) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1842 data: 0.0917 max mem: 3539 +Test: [1900/2387] eta: 0:01:33 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1349 (0.3117) acc1: 100.0000 (87.5069) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1740 data: 0.0814 max mem: 3539 +Test: [1910/2387] eta: 0:01:31 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1654 (0.3110) acc1: 99.7396 (87.5683) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1731 data: 0.0818 max mem: 3539 +Test: [1920/2387] eta: 0:01:29 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1816 (0.3108) acc1: 99.4792 (87.6070) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1749 data: 0.0830 max mem: 3539 +Test: [1930/2387] eta: 0:01:28 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1982 (0.3101) acc1: 99.7396 (87.6672) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1770 data: 0.0859 max mem: 3539 +Test: [1940/2387] eta: 0:01:26 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2002 (0.3099) acc1: 98.9583 (87.6843) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1827 data: 0.0927 max mem: 3539 +Test: [1950/2387] eta: 0:01:24 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2293 (0.3095) acc1: 96.6146 (87.7276) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1722 data: 0.0806 max mem: 3539 +Test: [1960/2387] eta: 0:01:22 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2192 (0.3092) acc1: 98.6979 (87.7777) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1830 data: 0.0912 max mem: 3539 +Test: [1970/2387] eta: 0:01:20 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2473 (0.3092) acc1: 97.3958 (87.7960) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1946 data: 0.1035 max mem: 3539 +Test: [1980/2387] eta: 0:01:18 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2809 (0.3091) acc1: 95.5729 (87.8181) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2021 data: 0.1088 max mem: 3539 +Test: [1990/2387] eta: 0:01:16 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2411 (0.3087) acc1: 95.8333 (87.8545) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2104 data: 0.1158 max mem: 3539 +Test: [2000/2387] eta: 0:01:14 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2425 (0.3084) acc1: 95.8333 (87.8923) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2083 data: 0.1152 max mem: 3539 +Test: [2010/2387] eta: 0:01:12 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2454 (0.3081) acc1: 95.5729 (87.9297) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2031 data: 0.1117 max mem: 3539 +Test: [2020/2387] eta: 0:01:10 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2455 (0.3078) acc1: 95.5729 (87.9683) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1806 data: 0.0899 max mem: 3539 +Test: [2030/2387] eta: 0:01:08 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2448 (0.3075) acc1: 95.3125 (88.0025) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1800 data: 0.0875 max mem: 3540 +Test: [2040/2387] eta: 0:01:06 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2441 (0.3072) acc1: 95.3125 (88.0398) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1927 data: 0.0991 max mem: 3540 +Test: [2050/2387] eta: 0:01:04 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2368 (0.3068) acc1: 95.8333 (88.0773) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1716 data: 0.0795 max mem: 3540 +Test: [2060/2387] eta: 0:01:02 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2317 (0.3064) acc1: 95.8333 (88.1152) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1601 data: 0.0693 max mem: 3540 +Test: [2070/2387] eta: 0:01:00 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2219 (0.3060) acc1: 96.0938 (88.1561) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1872 data: 0.0963 max mem: 3540 +Test: [2080/2387] eta: 0:00:59 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2202 (0.3056) acc1: 96.0938 (88.1939) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2164 data: 0.1245 max mem: 3540 +Test: [2090/2387] eta: 0:00:57 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2198 (0.3052) acc1: 96.0938 (88.2322) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1947 data: 0.1024 max mem: 3540 +Test: [2100/2387] eta: 0:00:55 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2206 (0.3048) acc1: 96.0938 (88.2706) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1432 data: 0.0464 max mem: 3540 +Test: [2110/2387] eta: 0:00:53 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2187 (0.3044) acc1: 96.3542 (88.3114) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1414 data: 0.0441 max mem: 3540 +Test: [2120/2387] eta: 0:00:51 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2001 (0.3038) acc1: 97.6562 (88.3607) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1755 data: 0.0824 max mem: 3540 +Test: [2130/2387] eta: 0:00:49 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1538 (0.3030) acc1: 99.7396 (88.4140) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1837 data: 0.0917 max mem: 3540 +Test: [2140/2387] eta: 0:00:47 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1430 (0.3024) acc1: 99.7396 (88.4433) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1793 data: 0.0844 max mem: 3540 +Test: [2150/2387] eta: 0:00:45 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1520 (0.3017) acc1: 99.7396 (88.4947) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1766 data: 0.0817 max mem: 3540 +Test: [2160/2387] eta: 0:00:43 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1436 (0.3011) acc1: 99.7396 (88.5436) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1810 data: 0.0887 max mem: 3540 +Test: [2170/2387] eta: 0:00:41 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1672 (0.3006) acc1: 99.2188 (88.5857) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1897 data: 0.0971 max mem: 3540 +Test: [2180/2387] eta: 0:00:39 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2242 (0.3010) acc1: 95.5729 (88.5591) acc5: 100.0000 (100.0000) acc1_neg: 65.365 acc5_neg: 100.000 time: 0.1897 data: 0.0964 max mem: 3540 +Test: [2190/2387] eta: 0:00:37 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2808 (0.3012) acc1: 90.3646 (88.5537) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1887 data: 0.0954 max mem: 3540 +Test: [2200/2387] eta: 0:00:35 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2189 (0.3008) acc1: 96.6146 (88.5953) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1939 data: 0.1017 max mem: 3540 +Test: [2210/2387] eta: 0:00:33 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2262 (0.3004) acc1: 96.8750 (88.6291) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1994 data: 0.1080 max mem: 3540 +Test: [2220/2387] eta: 0:00:32 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2262 (0.3001) acc1: 96.8750 (88.6696) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1999 data: 0.1096 max mem: 3540 +Test: [2230/2387] eta: 0:00:30 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1988 (0.2996) acc1: 98.6979 (88.7095) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1845 data: 0.0938 max mem: 3540 +Test: [2240/2387] eta: 0:00:28 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1882 (0.2992) acc1: 98.4375 (88.7465) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1990 data: 0.1081 max mem: 3540 +Test: [2250/2387] eta: 0:00:26 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2051 (0.2990) acc1: 96.8750 (88.7764) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.2038 data: 0.1113 max mem: 3540 +Test: [2260/2387] eta: 0:00:24 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2051 (0.2984) acc1: 98.4375 (88.8201) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1654 data: 0.0701 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1214 (0.2976) acc1: 100.0000 (88.8691) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.1002 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1109 (0.2970) acc1: 100.0000 (88.8971) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2156 data: 0.1235 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1099 (0.2962) acc1: 100.0000 (88.9455) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1998 data: 0.1057 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1148 (0.2954) acc1: 100.0000 (88.9907) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2188 data: 0.1222 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1287 (0.2948) acc1: 100.0000 (89.0328) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2158 data: 0.1222 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.1445 (0.2945) acc1: 98.6979 (89.0599) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2121 data: 0.1206 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2645 (0.2944) acc1: 96.6146 (89.0850) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.2014 data: 0.1092 max mem: 3541 +Test: [2340/2387] eta: 0:00:09 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2525 (0.2941) acc1: 96.6146 (89.1179) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1821 data: 0.0887 max mem: 3541 +Test: [2350/2387] eta: 0:00:07 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2347 (0.2941) acc1: 97.1354 (89.1235) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1904 data: 0.0983 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2508 (0.2942) acc1: 97.1354 (89.1329) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1595 data: 0.0693 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2505 (0.2940) acc1: 98.1771 (89.1603) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1372 data: 0.0472 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2734 (0.2941) acc1: 92.1875 (89.1689) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.1397 data: 0.0498 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 43.836 acc5_pos: 100.000 loss: 0.2789 (0.2939) acc1: 92.1875 (89.1794) acc5: 100.0000 (100.0000) acc1_neg: 95.349 acc5_neg: 100.000 time: 0.1266 data: 0.0394 max mem: 3541 +Test: Total time: 0:07:36 (0.1914 s / it) +* Overall Acc@1 89.179 Acc@5 100.000 + target == real Acc@1 27.209 Acc@5 100.000 + target == fake Acc@1 93.531 Acc@5 100.000 + loss 0.294 + PR-AUC 0.9870 ROC-AUC 0.8396 EER 23.26% (thr=0.7539) +Accuracy of the network on the 916267 test images: 89.2% +[rank0]:[W725 18:57:59.342992431 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:58:23.465000 611342 torch/distributed/run.py:766] +W0725 18:58:23.465000 611342 torch/distributed/run.py:766] ***************************************** +W0725 18:58:23.465000 611342 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:58:23.465000 611342 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:58:59.418299349 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:59:00.495657223 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:59:00.729063904 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:59:00.771130918 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 7): env:// +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:59:00.060909934 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:59:00.069346867 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:59:00.118090611 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:59:00.168045222 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n16_seed4_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed4_resampled_n116-041-153_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 0:51:10 acc1_pos: 28.646 acc5_pos: 100.000 loss: 1.0921 (1.0921) acc1: 28.6458 (28.6458) acc5: 100.0000 (100.0000) time: 5.9388 data: 3.9082 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/517] eta: 0:05:15 acc1_pos: 14.062 acc5_pos: 100.000 loss: 1.0386 (1.0226) acc1: 22.1354 (26.4915) acc5: 100.0000 (100.0000) time: 0.6218 data: 0.3557 max mem: 3531 +Test: [ 20/517] eta: 0:03:07 acc1_pos: 8.854 acc5_pos: 100.000 loss: 1.0386 (1.0652) acc1: 20.3125 (24.1071) acc5: 100.0000 (100.0000) time: 0.0999 data: 0.0102 max mem: 3531 +Test: [ 30/517] eta: 0:02:33 acc1_pos: 9.635 acc5_pos: 100.000 loss: 1.1432 (1.0965) acc1: 17.9688 (21.6314) acc5: 100.0000 (100.0000) time: 0.1458 data: 0.0533 max mem: 3531 +Test: [ 40/517] eta: 0:02:33 acc1_pos: 13.802 acc5_pos: 100.000 loss: 1.2234 (1.1239) acc1: 12.7604 (19.2518) acc5: 100.0000 (100.0000) time: 0.2629 data: 0.1690 max mem: 3531 +Test: [ 50/517] eta: 0:02:31 acc1_pos: 26.606 acc5_pos: 100.000 loss: 1.0789 (1.0767) acc1: 13.8021 (23.5856) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.3420 data: 0.2477 max mem: 3531 +Test: [ 60/517] eta: 0:02:23 acc1_pos: 26.606 acc5_pos: 100.000 loss: 0.3216 (0.9473) acc1: 92.4479 (35.3270) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.2949 data: 0.1977 max mem: 3531 +Test: [ 70/517] eta: 0:02:16 acc1_pos: 26.606 acc5_pos: 100.000 loss: 0.2822 (0.8556) acc1: 96.0938 (43.7793) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2549 data: 0.1568 max mem: 3531 +Test: [ 80/517] eta: 0:02:12 acc1_pos: 26.606 acc5_pos: 100.000 loss: 0.2837 (0.7909) acc1: 95.5729 (49.8939) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2699 data: 0.1728 max mem: 3531 +Test: [ 90/517] eta: 0:02:08 acc1_pos: 26.606 acc5_pos: 100.000 loss: 0.2955 (0.7370) acc1: 96.3542 (55.0796) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2799 data: 0.1832 max mem: 3531 +Test: [100/517] eta: 0:02:04 acc1_pos: 26.606 acc5_pos: 100.000 loss: 0.2857 (0.6893) acc1: 98.6979 (59.3879) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2849 data: 0.1871 max mem: 3531 +Test: [110/517] eta: 0:01:59 acc1_pos: 26.606 acc5_pos: 100.000 loss: 0.2330 (0.6462) acc1: 99.2188 (62.9575) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2599 data: 0.1624 max mem: 3531 +Test: [120/517] eta: 0:01:54 acc1_pos: 26.606 acc5_pos: 100.000 loss: 0.2248 (0.6131) acc1: 98.6979 (65.8876) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2399 data: 0.1452 max mem: 3531 +Test: [130/517] eta: 0:01:50 acc1_pos: 26.606 acc5_pos: 100.000 loss: 0.2374 (0.5848) acc1: 98.4375 (68.3942) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2499 data: 0.1548 max mem: 3531 +Test: [140/517] eta: 0:01:46 acc1_pos: 26.606 acc5_pos: 100.000 loss: 0.2377 (0.5617) acc1: 99.2188 (70.5471) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2349 data: 0.1383 max mem: 3531 +Test: [150/517] eta: 0:01:42 acc1_pos: 26.606 acc5_pos: 100.000 loss: 0.2377 (0.5401) acc1: 99.4792 (72.4338) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2399 data: 0.1454 max mem: 3531 +Test: [160/517] eta: 0:01:39 acc1_pos: 26.606 acc5_pos: 100.000 loss: 0.2596 (0.5237) acc1: 98.6979 (73.9664) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2502 data: 0.1542 max mem: 3531 +Test: [170/517] eta: 0:01:35 acc1_pos: 26.606 acc5_pos: 100.000 loss: 0.2683 (0.5081) acc1: 98.1771 (75.3472) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2499 data: 0.1522 max mem: 3531 +Test: [180/517] eta: 0:01:32 acc1_pos: 26.606 acc5_pos: 100.000 loss: 0.2305 (0.4920) acc1: 98.4375 (76.6287) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2546 data: 0.1594 max mem: 3531 +Test: [190/517] eta: 0:01:29 acc1_pos: 26.606 acc5_pos: 100.000 loss: 0.1610 (0.4733) acc1: 99.7396 (77.8496) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2399 data: 0.1448 max mem: 3531 +Test: [200/517] eta: 0:01:26 acc1_pos: 26.606 acc5_pos: 100.000 loss: 0.1538 (0.4598) acc1: 100.0000 (78.9153) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2449 data: 0.1507 max mem: 3531 +Test: [210/517] eta: 0:01:23 acc1_pos: 26.606 acc5_pos: 100.000 loss: 0.2618 (0.4531) acc1: 97.6562 (79.7011) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2599 data: 0.1645 max mem: 3532 +Test: [220/517] eta: 0:01:20 acc1_pos: 26.606 acc5_pos: 100.000 loss: 0.3175 (0.4468) acc1: 95.0521 (80.4016) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2549 data: 0.1591 max mem: 3532 +Test: [230/517] eta: 0:01:17 acc1_pos: 26.606 acc5_pos: 100.000 loss: 0.2983 (0.4398) acc1: 97.1354 (81.1756) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2599 data: 0.1653 max mem: 3532 +Test: [240/517] eta: 0:01:14 acc1_pos: 26.606 acc5_pos: 100.000 loss: 0.2249 (0.4297) acc1: 99.2188 (81.9308) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2653 data: 0.1685 max mem: 3532 +Test: [250/517] eta: 0:01:12 acc1_pos: 26.606 acc5_pos: 100.000 loss: 0.2231 (0.4227) acc1: 99.4792 (82.6060) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2649 data: 0.1687 max mem: 3532 +Test: [260/517] eta: 0:01:09 acc1_pos: 26.606 acc5_pos: 100.000 loss: 0.2495 (0.4160) acc1: 99.4792 (83.2595) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2795 data: 0.1831 max mem: 3532 +Test: [270/517] eta: 0:01:06 acc1_pos: 26.606 acc5_pos: 100.000 loss: 0.2358 (0.4094) acc1: 99.7396 (83.8705) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2699 data: 0.1714 max mem: 3532 +Test: [280/517] eta: 0:01:03 acc1_pos: 26.606 acc5_pos: 100.000 loss: 0.2260 (0.4024) acc1: 100.0000 (84.4213) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1588 max mem: 3532 +Test: [290/517] eta: 0:01:01 acc1_pos: 26.606 acc5_pos: 100.000 loss: 0.1965 (0.3957) acc1: 99.7396 (84.9442) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1588 max mem: 3532 +Test: [300/517] eta: 0:00:58 acc1_pos: 26.606 acc5_pos: 100.000 loss: 0.1945 (0.3891) acc1: 99.4792 (85.4245) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2549 data: 0.1577 max mem: 3532 +Test: [310/517] eta: 0:00:55 acc1_pos: 26.606 acc5_pos: 100.000 loss: 0.2278 (0.3858) acc1: 98.6979 (85.8303) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2549 data: 0.1597 max mem: 3532 +Test: [320/517] eta: 0:00:52 acc1_pos: 26.606 acc5_pos: 100.000 loss: 0.2714 (0.3814) acc1: 98.6979 (86.2296) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2499 data: 0.1566 max mem: 3532 +Test: [330/517] eta: 0:00:49 acc1_pos: 26.606 acc5_pos: 100.000 loss: 0.2063 (0.3758) acc1: 99.4792 (86.6369) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2549 data: 0.1595 max mem: 3532 +Test: [340/517] eta: 0:00:47 acc1_pos: 26.606 acc5_pos: 100.000 loss: 0.1974 (0.3714) acc1: 99.7396 (87.0013) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2549 data: 0.1560 max mem: 3532 +Test: [350/517] eta: 0:00:44 acc1_pos: 26.606 acc5_pos: 100.000 loss: 0.2610 (0.3692) acc1: 98.1771 (87.2693) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2449 data: 0.1461 max mem: 3532 +Test: [360/517] eta: 0:00:41 acc1_pos: 26.606 acc5_pos: 100.000 loss: 0.2829 (0.3671) acc1: 96.8750 (87.5166) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2299 data: 0.1340 max mem: 3532 +Test: [370/517] eta: 0:00:38 acc1_pos: 26.606 acc5_pos: 100.000 loss: 0.2586 (0.3630) acc1: 97.3958 (87.8348) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2303 data: 0.1344 max mem: 3532 +Test: [380/517] eta: 0:00:36 acc1_pos: 26.606 acc5_pos: 100.000 loss: 0.1960 (0.3586) acc1: 99.7396 (88.1418) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2548 data: 0.1547 max mem: 3532 +Test: [390/517] eta: 0:00:33 acc1_pos: 26.606 acc5_pos: 100.000 loss: 0.2009 (0.3549) acc1: 100.0000 (88.4371) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2444 data: 0.1424 max mem: 3532 +Test: [400/517] eta: 0:00:30 acc1_pos: 26.606 acc5_pos: 100.000 loss: 0.2311 (0.3523) acc1: 99.7396 (88.6878) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2349 data: 0.1351 max mem: 3532 +Test: [410/517] eta: 0:00:28 acc1_pos: 26.606 acc5_pos: 100.000 loss: 0.2699 (0.3505) acc1: 99.2188 (88.9370) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2399 data: 0.1434 max mem: 3532 +Test: [420/517] eta: 0:00:25 acc1_pos: 26.606 acc5_pos: 100.000 loss: 0.2437 (0.3480) acc1: 99.2188 (89.1837) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2499 data: 0.1531 max mem: 3532 +Test: [430/517] eta: 0:00:22 acc1_pos: 26.606 acc5_pos: 100.000 loss: 0.2585 (0.3465) acc1: 99.2188 (89.4021) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2549 data: 0.1570 max mem: 3532 +Test: [440/517] eta: 0:00:20 acc1_pos: 26.606 acc5_pos: 100.000 loss: 0.2698 (0.3442) acc1: 98.6979 (89.5975) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2449 data: 0.1478 max mem: 3533 +Test: [450/517] eta: 0:00:17 acc1_pos: 26.606 acc5_pos: 100.000 loss: 0.2336 (0.3417) acc1: 98.9583 (89.8201) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2499 data: 0.1527 max mem: 3533 +Test: [460/517] eta: 0:00:14 acc1_pos: 26.606 acc5_pos: 100.000 loss: 0.2414 (0.3397) acc1: 99.4792 (90.0279) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2599 data: 0.1612 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 26.606 acc5_pos: 100.000 loss: 0.2425 (0.3379) acc1: 99.7396 (90.2131) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2749 data: 0.1762 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 26.606 acc5_pos: 100.000 loss: 0.2201 (0.3355) acc1: 100.0000 (90.4025) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2649 data: 0.1684 max mem: 3533 +Test: [490/517] eta: 0:00:07 acc1_pos: 26.606 acc5_pos: 100.000 loss: 0.2162 (0.3334) acc1: 99.7396 (90.5889) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2399 data: 0.1460 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 26.606 acc5_pos: 100.000 loss: 0.2576 (0.3329) acc1: 98.9583 (90.7071) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2699 data: 0.1751 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 26.606 acc5_pos: 100.000 loss: 0.2800 (0.3307) acc1: 98.6979 (90.8589) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2446 data: 0.1513 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 26.606 acc5_pos: 100.000 loss: 0.1672 (0.3289) acc1: 99.4792 (90.9559) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1766 data: 0.0832 max mem: 3533 +Test: Total time: 0:02:14 (0.2602 s / it) +* Overall Acc@1 90.956 Acc@5 100.000 + target == real Acc@1 19.957 Acc@5 100.000 + target == fake Acc@1 98.321 Acc@5 100.000 + loss 0.329 + PR-AUC 0.9694 ROC-AUC 0.7923 EER 28.47% (thr=0.7495) +Accuracy of the network on the 198427 test images: 91.0% +[rank0]:[W725 19:01:25.327106690 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0725 19:01:48.129000 611930 torch/distributed/run.py:766] +W0725 19:01:48.129000 611930 torch/distributed/run.py:766] ***************************************** +W0725 19:01:48.129000 611930 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 19:01:48.129000 611930 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 19:02:24.458783581 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 19:02:25.497041291 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 19:02:25.744058412 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 19:02:25.145067282 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +[rank1]:[W725 19:02:25.200728113 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 19:02:25.208646805 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 19:02:25.287069573 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 19:02:25.415594873 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n16_seed4_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed4_resampled_n116-041-153_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:39:44 acc1_pos: 80.990 acc5_pos: 100.000 loss: 0.5211 (0.5211) acc1: 80.9896 (80.9896) acc5: 100.0000 (100.0000) time: 6.1286 data: 4.3323 max mem: 3531 +Test: [ 10/389] eta: 0:04:04 acc1_pos: 95.573 acc5_pos: 100.000 loss: 0.4237 (0.3687) acc1: 87.7604 (87.8314) acc5: 100.0000 (100.0000) time: 0.6442 data: 0.3943 max mem: 3531 +Test: [ 20/389] eta: 0:02:34 acc1_pos: 58.073 acc5_pos: 100.000 loss: 0.4213 (0.4389) acc1: 87.7604 (80.8408) acc5: 100.0000 (100.0000) time: 0.1328 data: 0.0383 max mem: 3531 +Test: [ 30/389] eta: 0:02:14 acc1_pos: 95.052 acc5_pos: 100.000 loss: 0.3496 (0.4054) acc1: 95.0521 (85.9879) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1311 max mem: 3531 +Test: [ 40/389] eta: 0:02:02 acc1_pos: 48.177 acc5_pos: 100.000 loss: 0.3664 (0.4367) acc1: 91.4062 (83.5493) acc5: 100.0000 (100.0000) time: 0.2799 data: 0.1876 max mem: 3531 +Test: [ 50/389] eta: 0:01:50 acc1_pos: 97.396 acc5_pos: 100.000 loss: 0.3953 (0.4176) acc1: 90.3646 (85.5341) acc5: 100.0000 (100.0000) time: 0.2499 data: 0.1547 max mem: 3531 +Test: [ 60/389] eta: 0:01:36 acc1_pos: 98.958 acc5_pos: 100.000 loss: 0.2873 (0.3995) acc1: 97.3958 (87.3250) acc5: 100.0000 (100.0000) time: 0.1749 data: 0.0766 max mem: 3531 +Test: [ 70/389] eta: 0:01:32 acc1_pos: 85.156 acc5_pos: 100.000 loss: 0.3348 (0.3955) acc1: 95.8333 (88.0649) acc5: 100.0000 (100.0000) time: 0.1949 data: 0.1005 max mem: 3531 +Test: [ 80/389] eta: 0:01:25 acc1_pos: 91.927 acc5_pos: 100.000 loss: 0.3556 (0.4064) acc1: 88.2812 (86.6062) acc5: 100.0000 (100.0000) time: 0.2322 data: 0.1411 max mem: 3531 +Test: [ 90/389] eta: 0:01:18 acc1_pos: 89.323 acc5_pos: 100.000 loss: 0.3104 (0.3938) acc1: 89.3229 (87.3998) acc5: 100.0000 (100.0000) time: 0.1749 data: 0.0842 max mem: 3531 +Test: [100/389] eta: 0:01:14 acc1_pos: 92.448 acc5_pos: 100.000 loss: 0.3180 (0.4035) acc1: 92.4479 (86.5640) acc5: 100.0000 (100.0000) time: 0.1837 data: 0.0882 max mem: 3531 +Test: [110/389] eta: 0:01:10 acc1_pos: 69.010 acc5_pos: 100.000 loss: 0.4461 (0.4306) acc1: 82.2917 (84.1568) acc5: 100.0000 (100.0000) time: 0.2102 data: 0.1149 max mem: 3531 +Test: [120/389] eta: 0:01:06 acc1_pos: 74.479 acc5_pos: 100.000 loss: 0.4190 (0.4241) acc1: 86.7188 (84.5945) acc5: 100.0000 (100.0000) time: 0.1799 data: 0.0889 max mem: 3531 +Test: [130/389] eta: 0:01:03 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.4351 (0.4461) acc1: 78.1250 (82.3771) acc5: 100.0000 (100.0000) acc1_neg: 44.271 acc5_neg: 100.000 time: 0.1848 data: 0.0916 max mem: 3531 +Test: [140/389] eta: 0:00:59 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.8154 (0.4786) acc1: 33.8542 (78.7031) acc5: 100.0000 (100.0000) acc1_neg: 44.792 acc5_neg: 100.000 time: 0.1938 data: 0.1011 max mem: 3531 +Test: [150/389] eta: 0:00:55 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.8938 (0.5005) acc1: 31.2500 (76.8264) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1621 data: 0.0697 max mem: 3531 +Test: [160/389] eta: 0:00:52 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.5321 (0.5127) acc1: 71.8750 (75.9851) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1549 data: 0.0610 max mem: 3531 +Test: [170/389] eta: 0:00:49 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.5964 (0.5273) acc1: 70.0521 (74.6954) acc5: 100.0000 (100.0000) acc1_neg: 27.344 acc5_neg: 100.000 time: 0.1725 data: 0.0792 max mem: 3531 +Test: [180/389] eta: 0:00:46 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.6203 (0.5418) acc1: 61.4583 (73.1569) acc5: 100.0000 (100.0000) acc1_neg: 42.708 acc5_neg: 100.000 time: 0.1701 data: 0.0783 max mem: 3531 +Test: [190/389] eta: 0:00:43 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.7646 (0.5630) acc1: 42.7083 (71.1687) acc5: 100.0000 (100.0000) acc1_neg: 0.260 acc5_neg: 100.000 time: 0.1630 data: 0.0709 max mem: 3531 +Test: [200/389] eta: 0:00:41 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.9396 (0.5810) acc1: 25.2604 (69.3227) acc5: 100.0000 (100.0000) acc1_neg: 26.562 acc5_neg: 100.000 time: 0.1763 data: 0.0822 max mem: 3531 +Test: [210/389] eta: 0:00:38 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.7778 (0.5858) acc1: 26.8229 (68.7315) acc5: 100.0000 (100.0000) acc1_neg: 80.729 acc5_neg: 100.000 time: 0.1842 data: 0.0904 max mem: 3532 +Test: [220/389] eta: 0:00:36 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.5330 (0.5904) acc1: 70.5729 (68.3058) acc5: 100.0000 (100.0000) acc1_neg: 20.573 acc5_neg: 100.000 time: 0.1870 data: 0.0953 max mem: 3532 +Test: [230/389] eta: 0:00:33 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.7963 (0.6039) acc1: 49.7396 (66.9203) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.1939 data: 0.1033 max mem: 3532 +Test: [240/389] eta: 0:00:31 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.7310 (0.6056) acc1: 46.3542 (66.8460) acc5: 100.0000 (100.0000) acc1_neg: 46.354 acc5_neg: 100.000 time: 0.1937 data: 0.1030 max mem: 3532 +Test: [250/389] eta: 0:00:29 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.5156 (0.6031) acc1: 77.0833 (67.2684) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1889 data: 0.0970 max mem: 3532 +Test: [260/389] eta: 0:00:27 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.4484 (0.6058) acc1: 87.5000 (67.5367) acc5: 100.0000 (100.0000) acc1_neg: 11.458 acc5_neg: 100.000 time: 0.1782 data: 0.0859 max mem: 3532 +Test: [270/389] eta: 0:00:24 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.7617 (0.6293) acc1: 44.2708 (65.8258) acc5: 100.0000 (100.0000) acc1_neg: 44.271 acc5_neg: 100.000 time: 0.1703 data: 0.0791 max mem: 3532 +Test: [280/389] eta: 0:00:22 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.9503 (0.6413) acc1: 28.9062 (64.8159) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1721 data: 0.0811 max mem: 3532 +Test: [290/389] eta: 0:00:20 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.8107 (0.6417) acc1: 40.8854 (64.5628) acc5: 100.0000 (100.0000) acc1_neg: 21.354 acc5_neg: 100.000 time: 0.1761 data: 0.0846 max mem: 3532 +Test: [300/389] eta: 0:00:18 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.9348 (0.6567) acc1: 21.3542 (63.0528) acc5: 100.0000 (100.0000) acc1_neg: 24.219 acc5_neg: 100.000 time: 0.1840 data: 0.0921 max mem: 3532 +Test: [310/389] eta: 0:00:16 acc1_pos: 67.485 acc5_pos: 100.000 loss: 1.0981 (0.6705) acc1: 13.0208 (62.0537) acc5: 100.0000 (100.0000) acc1_neg: 0.000 acc5_neg: 100.000 time: 0.1850 data: 0.0929 max mem: 3532 +Test: [320/389] eta: 0:00:14 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.8790 (0.6750) acc1: 40.6250 (61.7050) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1727 data: 0.0818 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 67.485 acc5_pos: 100.000 loss: 1.0691 (0.6870) acc1: 10.6771 (60.2806) acc5: 100.0000 (100.0000) acc1_neg: 7.292 acc5_neg: 100.000 time: 0.1679 data: 0.0776 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.5834 (0.6806) acc1: 59.3750 (61.0353) acc5: 100.0000 (100.0000) acc1_neg: 69.531 acc5_neg: 100.000 time: 0.1668 data: 0.0765 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.4797 (0.6899) acc1: 71.0938 (60.5020) acc5: 100.0000 (100.0000) acc1_neg: 9.115 acc5_neg: 100.000 time: 0.1549 data: 0.0646 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 67.485 acc5_pos: 100.000 loss: 1.1069 (0.7024) acc1: 20.8333 (59.3375) acc5: 100.0000 (100.0000) acc1_neg: 23.177 acc5_neg: 100.000 time: 0.1380 data: 0.0479 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 67.485 acc5_pos: 100.000 loss: 1.0771 (0.7041) acc1: 22.3958 (59.2852) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.1284 data: 0.0387 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.7275 (0.7054) acc1: 53.9062 (59.0879) acc5: 100.0000 (100.0000) acc1_neg: 41.927 acc5_neg: 100.000 time: 0.1261 data: 0.0365 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.6843 (0.7048) acc1: 52.1858 (59.0755) acc5: 100.0000 (100.0000) acc1_neg: 52.186 acc5_neg: 100.000 time: 0.1281 data: 0.0349 max mem: 3532 +Test: Total time: 0:01:15 (0.1935 s / it) +* Overall Acc@1 59.076 Acc@5 100.000 + target == real Acc@1 83.893 Acc@5 100.000 + target == fake Acc@1 47.540 Acc@5 100.000 + loss 0.705 + PR-AUC 0.8423 ROC-AUC 0.7317 EER 32.48% (thr=0.3667) +Accuracy of the network on the 149358 test images: 59.1% +[rank0]:[W725 19:03:51.870417537 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 19:04:13.875000 612501 torch/distributed/run.py:766] +W0725 19:04:13.875000 612501 torch/distributed/run.py:766] ***************************************** +W0725 19:04:13.875000 612501 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 19:04:13.875000 612501 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W725 19:04:50.555079277 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 19:04:50.871804285 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 19:04:50.899111498 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 19:04:50.131439586 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 19:04:50.241743006 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 19:04:50.252266799 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 19:04:50.275466447 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 19:04:50.290320661 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n16_seed4_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed4_resampled_n116-041-153_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:41:29 acc1_pos: 82.292 acc5_pos: 100.000 loss: 0.4079 (0.4079) acc1: 82.2917 (82.2917) acc5: 100.0000 (100.0000) time: 7.4749 data: 5.4870 max mem: 3531 +Test: [ 10/333] eta: 0:04:07 acc1_pos: 91.406 acc5_pos: 100.000 loss: 0.3200 (0.3216) acc1: 86.4583 (86.6477) acc5: 100.0000 (100.0000) time: 0.7661 data: 0.4994 max mem: 3531 +Test: [ 20/333] eta: 0:02:20 acc1_pos: 95.052 acc5_pos: 100.000 loss: 0.2888 (0.3034) acc1: 88.0208 (87.9340) acc5: 100.0000 (100.0000) time: 0.0975 data: 0.0005 max mem: 3531 +Test: [ 30/333] eta: 0:01:58 acc1_pos: 55.729 acc5_pos: 100.000 loss: 0.3727 (0.4479) acc1: 75.5208 (76.8145) acc5: 100.0000 (100.0000) time: 0.1849 data: 0.0877 max mem: 3531 +Test: [ 40/333] eta: 0:01:43 acc1_pos: 45.573 acc5_pos: 100.000 loss: 0.7565 (0.5266) acc1: 49.4792 (70.5031) acc5: 100.0000 (100.0000) time: 0.2549 data: 0.1618 max mem: 3531 +Test: [ 50/333] eta: 0:01:31 acc1_pos: 53.125 acc5_pos: 100.000 loss: 0.7593 (0.5763) acc1: 48.9583 (66.2684) acc5: 100.0000 (100.0000) time: 0.2199 data: 0.1263 max mem: 3531 +Test: [ 60/333] eta: 0:01:23 acc1_pos: 52.083 acc5_pos: 100.000 loss: 0.7811 (0.6164) acc1: 48.6979 (63.0038) acc5: 100.0000 (100.0000) time: 0.2121 data: 0.1191 max mem: 3531 +Test: [ 70/333] eta: 0:01:16 acc1_pos: 46.875 acc5_pos: 100.000 loss: 0.7811 (0.6395) acc1: 48.4375 (61.5170) acc5: 100.0000 (100.0000) time: 0.2099 data: 0.1162 max mem: 3531 +Test: [ 80/333] eta: 0:01:10 acc1_pos: 57.031 acc5_pos: 100.000 loss: 0.7538 (0.6548) acc1: 52.8646 (60.2013) acc5: 100.0000 (100.0000) time: 0.1827 data: 0.0849 max mem: 3531 +Test: [ 90/333] eta: 0:01:03 acc1_pos: 56.250 acc5_pos: 100.000 loss: 0.7208 (0.6602) acc1: 54.6875 (59.8272) acc5: 100.0000 (100.0000) time: 0.1452 data: 0.0477 max mem: 3531 +Test: [100/333] eta: 0:00:59 acc1_pos: 57.031 acc5_pos: 100.000 loss: 0.6891 (0.6631) acc1: 56.5104 (59.5374) acc5: 100.0000 (100.0000) time: 0.1649 data: 0.0679 max mem: 3531 +Test: [110/333] eta: 0:00:55 acc1_pos: 56.771 acc5_pos: 100.000 loss: 0.6891 (0.6666) acc1: 55.7292 (59.0606) acc5: 100.0000 (100.0000) time: 0.1913 data: 0.0965 max mem: 3531 +Test: [120/333] eta: 0:00:52 acc1_pos: 61.198 acc5_pos: 100.000 loss: 0.6729 (0.6660) acc1: 56.7708 (59.0845) acc5: 100.0000 (100.0000) time: 0.1999 data: 0.1077 max mem: 3531 +Test: [130/333] eta: 0:00:48 acc1_pos: 71.354 acc5_pos: 100.000 loss: 0.6429 (0.6653) acc1: 59.1146 (59.1782) acc5: 100.0000 (100.0000) time: 0.2048 data: 0.1116 max mem: 3531 +Test: [140/333] eta: 0:00:45 acc1_pos: 76.823 acc5_pos: 100.000 loss: 0.6263 (0.6628) acc1: 59.1146 (59.3436) acc5: 100.0000 (100.0000) time: 0.1848 data: 0.0907 max mem: 3531 +Test: [150/333] eta: 0:00:43 acc1_pos: 65.365 acc5_pos: 100.000 loss: 0.6584 (0.6654) acc1: 55.9896 (59.0973) acc5: 100.0000 (100.0000) time: 0.2084 data: 0.1118 max mem: 3531 +Test: [160/333] eta: 0:00:40 acc1_pos: 50.658 acc5_pos: 100.000 loss: 0.6895 (0.6715) acc1: 54.1667 (58.8703) acc5: 100.0000 (100.0000) acc1_neg: 47.917 acc5_neg: 100.000 time: 0.2199 data: 0.1212 max mem: 3531 +Test: [170/333] eta: 0:00:37 acc1_pos: 50.658 acc5_pos: 100.000 loss: 0.6141 (0.6689) acc1: 61.4583 (59.2912) acc5: 100.0000 (100.0000) acc1_neg: 73.698 acc5_neg: 100.000 time: 0.1915 data: 0.0960 max mem: 3531 +Test: [180/333] eta: 0:00:34 acc1_pos: 50.658 acc5_pos: 100.000 loss: 0.5263 (0.6599) acc1: 73.6979 (60.2325) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.1727 data: 0.0809 max mem: 3531 +Test: [190/333] eta: 0:00:32 acc1_pos: 50.658 acc5_pos: 100.000 loss: 0.4930 (0.6510) acc1: 76.5625 (61.1788) acc5: 100.0000 (100.0000) acc1_neg: 77.344 acc5_neg: 100.000 time: 0.1986 data: 0.1056 max mem: 3531 +Test: [200/333] eta: 0:00:30 acc1_pos: 50.658 acc5_pos: 100.000 loss: 0.5039 (0.6460) acc1: 74.4792 (61.7382) acc5: 100.0000 (100.0000) acc1_neg: 67.969 acc5_neg: 100.000 time: 0.2121 data: 0.1144 max mem: 3531 +Test: [210/333] eta: 0:00:27 acc1_pos: 50.658 acc5_pos: 100.000 loss: 0.5858 (0.6439) acc1: 68.2292 (62.0680) acc5: 100.0000 (100.0000) acc1_neg: 61.198 acc5_neg: 100.000 time: 0.2113 data: 0.1116 max mem: 3532 +Test: [220/333] eta: 0:00:25 acc1_pos: 50.658 acc5_pos: 100.000 loss: 0.5897 (0.6423) acc1: 67.9688 (62.2042) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.2268 data: 0.1306 max mem: 3532 +Test: [230/333] eta: 0:00:23 acc1_pos: 50.658 acc5_pos: 100.000 loss: 0.4736 (0.6342) acc1: 77.6042 (63.0299) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.2260 data: 0.1345 max mem: 3532 +Test: [240/333] eta: 0:00:20 acc1_pos: 50.658 acc5_pos: 100.000 loss: 0.4736 (0.6293) acc1: 77.6042 (63.5427) acc5: 100.0000 (100.0000) acc1_neg: 73.698 acc5_neg: 100.000 time: 0.1739 data: 0.0839 max mem: 3532 +Test: [250/333] eta: 0:00:18 acc1_pos: 50.658 acc5_pos: 100.000 loss: 0.5294 (0.6261) acc1: 73.4375 (63.8934) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.1827 data: 0.0903 max mem: 3532 +Test: [260/333] eta: 0:00:16 acc1_pos: 50.658 acc5_pos: 100.000 loss: 0.5530 (0.6243) acc1: 69.7917 (64.1333) acc5: 100.0000 (100.0000) acc1_neg: 68.750 acc5_neg: 100.000 time: 0.2214 data: 0.1274 max mem: 3532 +Test: [270/333] eta: 0:00:13 acc1_pos: 50.658 acc5_pos: 100.000 loss: 0.5189 (0.6201) acc1: 73.4375 (64.5362) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.2143 data: 0.1207 max mem: 3532 +Test: [280/333] eta: 0:00:11 acc1_pos: 50.658 acc5_pos: 100.000 loss: 0.5183 (0.6197) acc1: 73.4375 (64.6223) acc5: 100.0000 (100.0000) acc1_neg: 71.615 acc5_neg: 100.000 time: 0.1948 data: 0.1015 max mem: 3532 +Test: [290/333] eta: 0:00:09 acc1_pos: 50.658 acc5_pos: 100.000 loss: 0.6898 (0.6228) acc1: 58.3333 (64.4491) acc5: 100.0000 (100.0000) acc1_neg: 62.760 acc5_neg: 100.000 time: 0.1929 data: 0.1010 max mem: 3532 +Test: [300/333] eta: 0:00:07 acc1_pos: 50.658 acc5_pos: 100.000 loss: 0.7030 (0.6251) acc1: 59.6354 (64.3766) acc5: 100.0000 (100.0000) acc1_neg: 59.635 acc5_neg: 100.000 time: 0.1986 data: 0.1078 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 50.658 acc5_pos: 100.000 loss: 0.6195 (0.6213) acc1: 67.4479 (64.7466) acc5: 100.0000 (100.0000) acc1_neg: 69.010 acc5_neg: 100.000 time: 0.1534 data: 0.0630 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 50.658 acc5_pos: 100.000 loss: 0.5772 (0.6207) acc1: 70.8333 (64.8397) acc5: 100.0000 (100.0000) acc1_neg: 59.115 acc5_neg: 100.000 time: 0.1317 data: 0.0416 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 50.658 acc5_pos: 100.000 loss: 0.6834 (0.6245) acc1: 60.1562 (64.6164) acc5: 100.0000 (100.0000) acc1_neg: 60.156 acc5_neg: 100.000 time: 0.1243 data: 0.0346 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 50.658 acc5_pos: 100.000 loss: 0.7025 (0.6250) acc1: 59.1146 (64.5933) acc5: 100.0000 (100.0000) acc1_neg: 58.310 acc5_neg: 100.000 time: 0.1259 data: 0.0333 max mem: 3532 +Test: Total time: 0:01:10 (0.2105 s / it) +* Overall Acc@1 64.593 Acc@5 100.000 + target == real Acc@1 59.165 Acc@5 100.000 + target == fake Acc@1 69.426 Acc@5 100.000 + loss 0.625 + PR-AUC 0.7372 ROC-AUC 0.7099 EER 35.60% (thr=0.5366) +Accuracy of the network on the 127843 test images: 64.6% +[rank0]:[W725 19:06:11.796661390 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 19:06:33.584000 613085 torch/distributed/run.py:766] +W0725 19:06:33.584000 613085 torch/distributed/run.py:766] ***************************************** +W0725 19:06:33.584000 613085 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 19:06:33.584000 613085 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 19:07:11.054719594 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 19:07:11.203446435 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 19:07:11.272057352 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 19:07:11.297833790 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W725 19:07:11.314250112 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +| distributed init (rank 5): env:// +[rank0]:[W725 19:07:12.622171397 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 19:07:12.630110241 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W725 19:07:12.630130431 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n16_seed4_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed4_resampled_n116-041-153_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:23:19 acc1_pos: 93.490 acc5_pos: 100.000 loss: 0.2214 (0.2214) acc1: 93.4896 (93.4896) acc5: 100.0000 (100.0000) time: 5.9547 data: 3.8326 max mem: 3531 +Test: [ 10/235] eta: 0:02:20 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.2130 (0.2134) acc1: 94.0104 (93.9631) acc5: 100.0000 (100.0000) time: 0.6233 data: 0.3488 max mem: 3531 +Test: [ 20/235] eta: 0:01:32 acc1_pos: 94.792 acc5_pos: 100.000 loss: 0.2040 (0.2051) acc1: 94.5312 (94.5809) acc5: 100.0000 (100.0000) time: 0.1532 data: 0.0605 max mem: 3531 +Test: [ 30/235] eta: 0:01:15 acc1_pos: 94.010 acc5_pos: 100.000 loss: 0.1997 (0.2041) acc1: 94.7917 (94.5481) acc5: 100.0000 (100.0000) time: 0.2286 data: 0.1316 max mem: 3531 +Test: [ 40/235] eta: 0:01:06 acc1_pos: 95.052 acc5_pos: 100.000 loss: 0.2006 (0.2040) acc1: 94.5312 (94.5249) acc5: 100.0000 (100.0000) time: 0.2531 data: 0.1567 max mem: 3531 +Test: [ 50/235] eta: 0:00:59 acc1_pos: 93.750 acc5_pos: 100.000 loss: 0.2012 (0.2050) acc1: 94.5312 (94.4802) acc5: 100.0000 (100.0000) time: 0.2494 data: 0.1572 max mem: 3531 +Test: [ 60/235] eta: 0:00:54 acc1_pos: 95.573 acc5_pos: 100.000 loss: 0.2095 (0.2057) acc1: 94.0104 (94.4843) acc5: 100.0000 (100.0000) time: 0.2467 data: 0.1537 max mem: 3531 +Test: [ 70/235] eta: 0:00:47 acc1_pos: 93.846 acc5_pos: 100.000 loss: 0.2274 (0.3939) acc1: 92.9688 (84.7014) acc5: 100.0000 (100.0000) acc1_neg: 13.802 acc5_neg: 100.000 time: 0.1949 data: 0.0981 max mem: 3531 +Test: [ 80/235] eta: 0:00:43 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.7161 (0.5594) acc1: 16.4062 (76.2024) acc5: 100.0000 (100.0000) acc1_neg: 14.323 acc5_neg: 100.000 time: 0.1764 data: 0.0818 max mem: 3531 +Test: [ 90/235] eta: 0:00:39 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.7272 (0.6893) acc1: 16.1458 (69.6085) acc5: 100.0000 (100.0000) acc1_neg: 16.406 acc5_neg: 100.000 time: 0.2153 data: 0.1192 max mem: 3531 +Test: [100/235] eta: 0:00:35 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.7263 (0.7916) acc1: 16.4062 (64.3255) acc5: 100.0000 (100.0000) acc1_neg: 13.802 acc5_neg: 100.000 time: 0.2006 data: 0.1046 max mem: 3531 +Test: [110/235] eta: 0:00:32 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.7203 (0.8761) acc1: 15.6250 (60.0202) acc5: 100.0000 (100.0000) acc1_neg: 18.490 acc5_neg: 100.000 time: 0.2145 data: 0.1221 max mem: 3531 +Test: [120/235] eta: 0:00:29 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.6650 (0.9193) acc1: 19.5312 (57.2658) acc5: 100.0000 (100.0000) acc1_neg: 31.510 acc5_neg: 100.000 time: 0.2127 data: 0.1171 max mem: 3531 +Test: [130/235] eta: 0:00:25 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.3402 (0.9506) acc1: 28.3854 (55.1666) acc5: 100.0000 (100.0000) acc1_neg: 26.823 acc5_neg: 100.000 time: 0.1655 data: 0.0685 max mem: 3531 +Test: [140/235] eta: 0:00:23 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.3486 (0.9810) acc1: 29.1667 (53.2543) acc5: 100.0000 (100.0000) acc1_neg: 26.302 acc5_neg: 100.000 time: 0.1749 data: 0.0792 max mem: 3531 +Test: [150/235] eta: 0:00:20 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.3471 (1.0037) acc1: 28.6458 (51.7005) acc5: 100.0000 (100.0000) acc1_neg: 31.510 acc5_neg: 100.000 time: 0.2093 data: 0.1143 max mem: 3531 +Test: [160/235] eta: 0:00:17 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.3337 (1.0243) acc1: 29.1667 (50.3445) acc5: 100.0000 (100.0000) acc1_neg: 29.688 acc5_neg: 100.000 time: 0.2063 data: 0.1122 max mem: 3531 +Test: [170/235] eta: 0:00:15 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.3663 (1.0540) acc1: 27.6042 (48.6263) acc5: 100.0000 (100.0000) acc1_neg: 17.448 acc5_neg: 100.000 time: 0.1899 data: 0.0968 max mem: 3531 +Test: [180/235] eta: 0:00:12 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.5652 (1.0815) acc1: 19.5312 (47.0419) acc5: 100.0000 (100.0000) acc1_neg: 18.229 acc5_neg: 100.000 time: 0.1952 data: 0.1024 max mem: 3531 +Test: [190/235] eta: 0:00:10 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.5454 (1.1055) acc1: 20.0521 (45.6261) acc5: 100.0000 (100.0000) acc1_neg: 17.708 acc5_neg: 100.000 time: 0.2099 data: 0.1160 max mem: 3531 +Test: [200/235] eta: 0:00:08 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.5454 (1.1283) acc1: 19.7917 (44.2980) acc5: 100.0000 (100.0000) acc1_neg: 17.708 acc5_neg: 100.000 time: 0.1959 data: 0.1021 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.5600 (1.1481) acc1: 18.7500 (43.1206) acc5: 100.0000 (100.0000) acc1_neg: 19.792 acc5_neg: 100.000 time: 0.1791 data: 0.0873 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.5404 (1.1663) acc1: 19.0104 (42.0543) acc5: 100.0000 (100.0000) acc1_neg: 18.490 acc5_neg: 100.000 time: 0.1583 data: 0.0674 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.5459 (1.1825) acc1: 19.7917 (41.0928) acc5: 100.0000 (100.0000) acc1_neg: 19.792 acc5_neg: 100.000 time: 0.1362 data: 0.0461 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.5459 (1.1903) acc1: 19.7917 (40.8149) acc5: 100.0000 (100.0000) acc1_neg: 17.391 acc5_neg: 100.000 time: 0.1285 data: 0.0419 max mem: 3532 +Test: Total time: 0:00:51 (0.2175 s / it) +* Overall Acc@1 40.815 Acc@5 100.000 + target == real Acc@1 94.487 Acc@5 100.000 + target == fake Acc@1 21.403 Acc@5 100.000 + loss 1.190 + PR-AUC 0.8588 ROC-AUC 0.7028 EER 35.51% (thr=0.1456) +Accuracy of the network on the 89879 test images: 40.8% +[rank0]:[W725 19:08:11.299401866 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 19:08:33.294000 613643 torch/distributed/run.py:766] +W0725 19:08:33.294000 613643 torch/distributed/run.py:766] ***************************************** +W0725 19:08:33.294000 613643 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 19:08:33.294000 613643 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 19:09:05.373604124 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 19:09:06.547492941 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 19:09:06.742328393 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +[rank5]:[W725 19:09:06.979681126 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 19:09:06.982815748 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 19:09:06.988067613 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 19:09:06.996706298 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 19:09:06.026419035 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n16_seed4_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed4_resampled_n116-041-153_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:24:19 acc1_pos: 53.385 acc5_pos: 100.000 loss: 0.7517 (0.7517) acc1: 53.3854 (53.3854) acc5: 100.0000 (100.0000) time: 5.4254 data: 3.2986 max mem: 3531 +Test: [ 10/269] eta: 0:02:36 acc1_pos: 63.802 acc5_pos: 100.000 loss: 0.6464 (0.6676) acc1: 60.4167 (59.3277) acc5: 100.0000 (100.0000) time: 0.6026 data: 0.3191 max mem: 3531 +Test: [ 20/269] eta: 0:01:41 acc1_pos: 53.906 acc5_pos: 100.000 loss: 0.6450 (0.6436) acc1: 60.4167 (61.4087) acc5: 100.0000 (100.0000) time: 0.1579 data: 0.0612 max mem: 3531 +Test: [ 30/269] eta: 0:01:26 acc1_pos: 58.854 acc5_pos: 100.000 loss: 0.5770 (0.6325) acc1: 59.6354 (62.8276) acc5: 100.0000 (100.0000) time: 0.2292 data: 0.1336 max mem: 3531 +Test: [ 40/269] eta: 0:01:19 acc1_pos: 83.621 acc5_pos: 100.000 loss: 0.5494 (0.6089) acc1: 71.3542 (65.4726) acc5: 100.0000 (100.0000) acc1_neg: 63.021 acc5_neg: 100.000 time: 0.2811 data: 0.1869 max mem: 3531 +Test: [ 50/269] eta: 0:01:09 acc1_pos: 83.621 acc5_pos: 100.000 loss: 0.5224 (0.5939) acc1: 75.0000 (67.2692) acc5: 100.0000 (100.0000) acc1_neg: 57.292 acc5_neg: 100.000 time: 0.2533 data: 0.1609 max mem: 3531 +Test: [ 60/269] eta: 0:01:01 acc1_pos: 83.621 acc5_pos: 100.000 loss: 0.5528 (0.5985) acc1: 72.9167 (66.9228) acc5: 100.0000 (100.0000) acc1_neg: 47.656 acc5_neg: 100.000 time: 0.1858 data: 0.0911 max mem: 3531 +Test: [ 70/269] eta: 0:00:58 acc1_pos: 83.621 acc5_pos: 100.000 loss: 0.5867 (0.5965) acc1: 64.8438 (67.0371) acc5: 100.0000 (100.0000) acc1_neg: 64.844 acc5_neg: 100.000 time: 0.2349 data: 0.1370 max mem: 3531 +Test: [ 80/269] eta: 0:00:54 acc1_pos: 83.621 acc5_pos: 100.000 loss: 0.5393 (0.5895) acc1: 71.6146 (67.7373) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.2678 data: 0.1693 max mem: 3531 +Test: [ 90/269] eta: 0:00:50 acc1_pos: 83.621 acc5_pos: 100.000 loss: 0.6018 (0.5968) acc1: 65.3646 (66.9271) acc5: 100.0000 (100.0000) acc1_neg: 63.021 acc5_neg: 100.000 time: 0.2275 data: 0.1321 max mem: 3531 +Test: [100/269] eta: 0:00:45 acc1_pos: 83.621 acc5_pos: 100.000 loss: 0.6279 (0.6013) acc1: 63.0208 (66.7337) acc5: 100.0000 (100.0000) acc1_neg: 63.802 acc5_neg: 100.000 time: 0.1950 data: 0.1025 max mem: 3531 +Test: [110/269] eta: 0:00:42 acc1_pos: 83.621 acc5_pos: 100.000 loss: 0.6347 (0.6076) acc1: 60.4167 (66.0238) acc5: 100.0000 (100.0000) acc1_neg: 50.521 acc5_neg: 100.000 time: 0.2023 data: 0.1105 max mem: 3531 +Test: [120/269] eta: 0:00:39 acc1_pos: 83.621 acc5_pos: 100.000 loss: 0.6658 (0.6061) acc1: 60.4167 (66.2298) acc5: 100.0000 (100.0000) acc1_neg: 62.240 acc5_neg: 100.000 time: 0.2362 data: 0.1446 max mem: 3531 +Test: [130/269] eta: 0:00:36 acc1_pos: 83.621 acc5_pos: 100.000 loss: 0.6443 (0.6076) acc1: 62.2396 (66.1896) acc5: 100.0000 (100.0000) acc1_neg: 56.510 acc5_neg: 100.000 time: 0.2314 data: 0.1384 max mem: 3531 +Test: [140/269] eta: 0:00:33 acc1_pos: 83.621 acc5_pos: 100.000 loss: 0.6709 (0.6194) acc1: 58.8542 (65.3997) acc5: 100.0000 (100.0000) acc1_neg: 36.458 acc5_neg: 100.000 time: 0.2286 data: 0.1340 max mem: 3531 +Test: [150/269] eta: 0:00:30 acc1_pos: 83.621 acc5_pos: 100.000 loss: 0.7661 (0.6287) acc1: 49.4792 (64.7541) acc5: 100.0000 (100.0000) acc1_neg: 64.062 acc5_neg: 100.000 time: 0.2147 data: 0.1227 max mem: 3531 +Test: [160/269] eta: 0:00:27 acc1_pos: 83.621 acc5_pos: 100.000 loss: 0.5963 (0.6278) acc1: 64.0625 (64.8810) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.2082 data: 0.1173 max mem: 3531 +Test: [170/269] eta: 0:00:24 acc1_pos: 83.621 acc5_pos: 100.000 loss: 0.5584 (0.6212) acc1: 72.6562 (65.5519) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.2031 data: 0.1109 max mem: 3531 +Test: [180/269] eta: 0:00:21 acc1_pos: 83.621 acc5_pos: 100.000 loss: 0.4360 (0.6098) acc1: 82.5521 (66.5401) acc5: 100.0000 (100.0000) acc1_neg: 70.573 acc5_neg: 100.000 time: 0.2047 data: 0.1124 max mem: 3531 +Test: [190/269] eta: 0:00:19 acc1_pos: 83.621 acc5_pos: 100.000 loss: 0.4680 (0.6089) acc1: 75.0000 (66.5344) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.2054 data: 0.1131 max mem: 3531 +Test: [200/269] eta: 0:00:16 acc1_pos: 83.621 acc5_pos: 100.000 loss: 0.5784 (0.6077) acc1: 69.0104 (66.7107) acc5: 100.0000 (100.0000) acc1_neg: 76.302 acc5_neg: 100.000 time: 0.2049 data: 0.1124 max mem: 3531 +Test: [210/269] eta: 0:00:14 acc1_pos: 83.621 acc5_pos: 100.000 loss: 0.6573 (0.6136) acc1: 60.9375 (66.1310) acc5: 100.0000 (100.0000) acc1_neg: 58.594 acc5_neg: 100.000 time: 0.1942 data: 0.1029 max mem: 3532 +Test: [220/269] eta: 0:00:11 acc1_pos: 83.621 acc5_pos: 100.000 loss: 0.6915 (0.6154) acc1: 57.5521 (66.0575) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1612 data: 0.0702 max mem: 3532 +Test: [230/269] eta: 0:00:09 acc1_pos: 83.621 acc5_pos: 100.000 loss: 0.4779 (0.6084) acc1: 80.7292 (66.7253) acc5: 100.0000 (100.0000) acc1_neg: 62.240 acc5_neg: 100.000 time: 0.1433 data: 0.0527 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 83.621 acc5_pos: 100.000 loss: 0.5126 (0.6099) acc1: 71.6146 (66.6375) acc5: 100.0000 (100.0000) acc1_neg: 64.583 acc5_neg: 100.000 time: 0.1249 data: 0.0352 max mem: 3532 +Test: [250/269] eta: 0:00:04 acc1_pos: 83.621 acc5_pos: 100.000 loss: 0.6479 (0.6141) acc1: 63.0208 (66.1811) acc5: 100.0000 (100.0000) acc1_neg: 30.208 acc5_neg: 100.000 time: 0.1116 data: 0.0221 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 83.621 acc5_pos: 100.000 loss: 0.7322 (0.6194) acc1: 56.7708 (65.5572) acc5: 100.0000 (100.0000) acc1_neg: 58.333 acc5_neg: 100.000 time: 0.1131 data: 0.0236 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 83.621 acc5_pos: 100.000 loss: 0.6330 (0.6194) acc1: 61.1979 (65.5156) acc5: 100.0000 (100.0000) acc1_neg: 63.780 acc5_neg: 100.000 time: 0.1079 data: 0.0148 max mem: 3532 +Test: Total time: 0:00:57 (0.2148 s / it) +* Overall Acc@1 65.516 Acc@5 100.000 + target == real Acc@1 63.286 Acc@5 100.000 + target == fake Acc@1 65.820 Acc@5 100.000 + loss 0.619 + PR-AUC 0.9442 ROC-AUC 0.7069 EER 35.42% (thr=0.5078) +Accuracy of the network on the 103293 test images: 65.5% +[rank0]:[W725 19:10:12.600914377 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-25T19:10:19Z] eval_done tag=fixed500k_fake_n16_seed4_resampled +eval_queue_done_utc=2026-07-25T19:10:19Z diff --git a/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_112559/eval_queue_n116-041-153.pid b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_112559/eval_queue_n116-041-153.pid new file mode 100644 index 0000000000000000000000000000000000000000..31315f016c2000b75a20a94f12c27da07e6b3e71 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_112559/eval_queue_n116-041-153.pid @@ -0,0 +1 @@ +590465 diff --git a/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_112559/eval_queue_n116-041-188.log b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_112559/eval_queue_n116-041-188.log new file mode 100644 index 0000000000000000000000000000000000000000..76488d5fa7e43d6f047b5fe77cfa4cfc68ded920 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_112559/eval_queue_n116-041-188.log @@ -0,0 +1,2141 @@ +eval_queue_started_utc=2026-07-25T17:02:41Z +hostname=n116-041-188 +train_tag=20260725_112559 +eval_tag=20260725_112559 +real_level=11 +fake_level=22 +eval_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559 +batch_per_gpu=256 +[2026-07-25T17:02:42Z] eval_wait tag=fixed500k_real_n11_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed1_resampled +[2026-07-25T17:03:42Z] eval_wait tag=fixed500k_real_n11_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed1_resampled +[2026-07-25T17:04:42Z] eval_wait tag=fixed500k_real_n11_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed1_resampled +[2026-07-25T17:05:42Z] eval_wait tag=fixed500k_real_n11_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed1_resampled +[2026-07-25T17:06:42Z] eval_wait tag=fixed500k_real_n11_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed1_resampled +[2026-07-25T17:07:42Z] eval_wait tag=fixed500k_real_n11_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed1_resampled +[2026-07-25T17:08:42Z] eval_wait tag=fixed500k_real_n11_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed1_resampled +[2026-07-25T17:09:42Z] eval_wait tag=fixed500k_real_n11_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed1_resampled +[2026-07-25T17:10:42Z] eval_wait tag=fixed500k_real_n11_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed1_resampled +[2026-07-25T17:11:42Z] eval_skip tag=fixed500k_real_n11_seed1_resampled reason=complete_after_lock output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed1_resampled +[2026-07-25T17:11:42Z] eval_skip tag=fixed500k_fake_n22_seed1_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n22_seed1_resampled +[2026-07-25T17:11:42Z] eval_wait tag=fixed500k_real_n11_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed2_resampled +[2026-07-25T17:12:43Z] eval_wait tag=fixed500k_real_n11_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed2_resampled +[2026-07-25T17:13:43Z] eval_wait tag=fixed500k_real_n11_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed2_resampled +[2026-07-25T17:14:43Z] eval_wait tag=fixed500k_real_n11_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed2_resampled +[2026-07-25T17:15:43Z] eval_wait tag=fixed500k_real_n11_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed2_resampled +[2026-07-25T17:16:43Z] eval_wait tag=fixed500k_real_n11_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed2_resampled +[2026-07-25T17:17:44Z] eval_wait tag=fixed500k_real_n11_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed2_resampled +[2026-07-25T17:18:44Z] eval_wait tag=fixed500k_real_n11_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed2_resampled +[2026-07-25T17:19:44Z] eval_wait tag=fixed500k_real_n11_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed2_resampled +[2026-07-25T17:20:44Z] eval_wait tag=fixed500k_real_n11_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed2_resampled +[2026-07-25T17:21:44Z] eval_wait tag=fixed500k_real_n11_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed2_resampled +[2026-07-25T17:22:44Z] eval_wait tag=fixed500k_real_n11_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed2_resampled +[2026-07-25T17:23:45Z] eval_wait tag=fixed500k_real_n11_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed2_resampled +[2026-07-25T17:24:45Z] eval_wait tag=fixed500k_real_n11_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed2_resampled +[2026-07-25T17:25:45Z] eval_wait tag=fixed500k_real_n11_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed2_resampled +[2026-07-25T17:26:45Z] eval_wait tag=fixed500k_real_n11_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed2_resampled +[2026-07-25T17:27:45Z] eval_wait tag=fixed500k_real_n11_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed2_resampled +[2026-07-25T17:28:45Z] eval_wait tag=fixed500k_real_n11_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed2_resampled +[2026-07-25T17:29:45Z] eval_wait tag=fixed500k_real_n11_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed2_resampled +[2026-07-25T17:30:46Z] eval_wait tag=fixed500k_real_n11_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed2_resampled +[2026-07-25T17:31:46Z] eval_wait tag=fixed500k_real_n11_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed2_resampled +[2026-07-25T17:32:46Z] eval_wait tag=fixed500k_real_n11_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed2_resampled +[2026-07-25T17:33:46Z] eval_skip tag=fixed500k_real_n11_seed2_resampled reason=complete_after_lock output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed2_resampled +[2026-07-25T17:33:46Z] eval_skip tag=fixed500k_fake_n22_seed2_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n22_seed2_resampled +[2026-07-25T17:33:47Z] eval_wait tag=fixed500k_real_n11_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed3_resampled +[2026-07-25T17:34:47Z] eval_wait tag=fixed500k_real_n11_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed3_resampled +[2026-07-25T17:35:47Z] eval_wait tag=fixed500k_real_n11_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed3_resampled +[2026-07-25T17:36:47Z] eval_wait tag=fixed500k_real_n11_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed3_resampled +[2026-07-25T17:37:47Z] eval_wait tag=fixed500k_real_n11_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed3_resampled +[2026-07-25T17:38:47Z] eval_wait tag=fixed500k_real_n11_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed3_resampled +[2026-07-25T17:39:47Z] eval_wait tag=fixed500k_real_n11_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed3_resampled +[2026-07-25T17:40:48Z] eval_wait tag=fixed500k_real_n11_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed3_resampled +[2026-07-25T17:41:48Z] eval_wait tag=fixed500k_real_n11_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed3_resampled +[2026-07-25T17:42:48Z] eval_wait tag=fixed500k_real_n11_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed3_resampled +[2026-07-25T17:43:48Z] eval_wait tag=fixed500k_real_n11_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed3_resampled +[2026-07-25T17:44:48Z] eval_wait tag=fixed500k_real_n11_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed3_resampled +[2026-07-25T17:45:48Z] eval_wait tag=fixed500k_real_n11_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed3_resampled +[2026-07-25T17:46:49Z] eval_wait tag=fixed500k_real_n11_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed3_resampled +[2026-07-25T17:47:49Z] eval_wait tag=fixed500k_real_n11_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed3_resampled +[2026-07-25T17:48:49Z] eval_wait tag=fixed500k_real_n11_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed3_resampled +[2026-07-25T17:49:49Z] eval_wait tag=fixed500k_real_n11_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed3_resampled +[2026-07-25T17:50:49Z] eval_wait tag=fixed500k_real_n11_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed3_resampled +[2026-07-25T17:51:49Z] eval_wait tag=fixed500k_real_n11_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed3_resampled +[2026-07-25T17:52:50Z] eval_wait tag=fixed500k_real_n11_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed3_resampled +[2026-07-25T17:53:50Z] eval_wait tag=fixed500k_real_n11_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed3_resampled +[2026-07-25T17:54:50Z] eval_wait tag=fixed500k_real_n11_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed3_resampled +[2026-07-25T17:55:50Z] eval_skip tag=fixed500k_real_n11_seed3_resampled reason=complete_after_lock output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed3_resampled +[2026-07-25T17:55:50Z] eval_wait tag=fixed500k_fake_n22_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n22_seed3_resampled +[2026-07-25T17:56:50Z] eval_wait tag=fixed500k_fake_n22_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n22_seed3_resampled +[2026-07-25T17:57:51Z] eval_wait tag=fixed500k_fake_n22_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n22_seed3_resampled +[2026-07-25T17:58:51Z] eval_wait tag=fixed500k_fake_n22_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n22_seed3_resampled +[2026-07-25T17:59:51Z] eval_wait tag=fixed500k_fake_n22_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n22_seed3_resampled +[2026-07-25T18:00:51Z] eval_wait tag=fixed500k_fake_n22_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n22_seed3_resampled +[2026-07-25T18:01:51Z] eval_wait tag=fixed500k_fake_n22_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n22_seed3_resampled +[2026-07-25T18:02:51Z] eval_wait tag=fixed500k_fake_n22_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n22_seed3_resampled +[2026-07-25T18:03:52Z] eval_wait tag=fixed500k_fake_n22_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n22_seed3_resampled +[2026-07-25T18:04:52Z] eval_wait tag=fixed500k_fake_n22_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n22_seed3_resampled +[2026-07-25T18:05:52Z] eval_wait tag=fixed500k_fake_n22_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n22_seed3_resampled +[2026-07-25T18:06:52Z] eval_wait tag=fixed500k_fake_n22_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n22_seed3_resampled +[2026-07-25T18:07:52Z] eval_wait tag=fixed500k_fake_n22_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n22_seed3_resampled +[2026-07-25T18:08:52Z] eval_wait tag=fixed500k_fake_n22_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n22_seed3_resampled +[2026-07-25T18:09:53Z] eval_wait tag=fixed500k_fake_n22_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n22_seed3_resampled +[2026-07-25T18:10:53Z] eval_wait tag=fixed500k_fake_n22_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n22_seed3_resampled +[2026-07-25T18:11:53Z] eval_wait tag=fixed500k_fake_n22_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n22_seed3_resampled +[2026-07-25T18:12:53Z] eval_wait tag=fixed500k_fake_n22_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n22_seed3_resampled +[2026-07-25T18:13:53Z] eval_wait tag=fixed500k_fake_n22_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n22_seed3_resampled +[2026-07-25T18:14:53Z] eval_wait tag=fixed500k_fake_n22_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n22_seed3_resampled +[2026-07-25T18:15:53Z] eval_wait tag=fixed500k_fake_n22_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n22_seed3_resampled +[2026-07-25T18:16:54Z] eval_skip tag=fixed500k_fake_n22_seed3_resampled reason=complete_after_wait output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n22_seed3_resampled +[2026-07-25T18:16:54Z] eval_start tag=fixed500k_real_n11_seed4_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n11_classseed4_resampled_n116-041-188_20260725_112559/best_checkpoint.pth +started_utc=2026-07-25T18:16:54Z +hostname=n116-041-188 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n11_classseed4_resampled_n116-041-188_20260725_112559/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed4_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0725 18:17:07.713000 603389 torch/distributed/run.py:766] +W0725 18:17:07.713000 603389 torch/distributed/run.py:766] ***************************************** +W0725 18:17:07.713000 603389 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:17:07.713000 603389 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:17:36.532125495 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W725 18:17:37.129565820 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W725 18:17:37.133407482 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +[rank6]:[W725 18:17:37.357797486 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:17:37.361906388 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:17:37.451305868 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:17:37.457555863 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:17:37.480611546 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed4_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n11_classseed4_resampled_n116-041-188_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:29:23 acc1_pos: 71.094 acc5_pos: 100.000 loss: 0.5819 (0.5819) acc1: 71.0938 (71.0938) acc5: 100.0000 (100.0000) time: 7.5375 data: 5.4164 max mem: 3531 +Test: [ 10/234] eta: 0:03:07 acc1_pos: 36.458 acc5_pos: 100.000 loss: 0.6268 (0.7440) acc1: 60.1562 (53.1960) acc5: 100.0000 (100.0000) time: 0.8383 data: 0.5588 max mem: 3531 +Test: [ 20/234] eta: 0:01:43 acc1_pos: 88.802 acc5_pos: 100.000 loss: 0.6256 (0.6735) acc1: 61.9792 (60.8755) acc5: 100.0000 (100.0000) time: 0.1333 data: 0.0367 max mem: 3531 +Test: [ 30/234] eta: 0:01:28 acc1_pos: 64.323 acc5_pos: 100.000 loss: 0.3106 (0.5634) acc1: 88.8021 (69.4388) acc5: 100.0000 (100.0000) time: 0.2091 data: 0.1136 max mem: 3531 +Test: [ 40/234] eta: 0:01:15 acc1_pos: 97.656 acc5_pos: 100.000 loss: 0.2905 (0.5506) acc1: 90.8854 (69.9187) acc5: 100.0000 (100.0000) time: 0.2899 data: 0.1944 max mem: 3531 +Test: [ 50/234] eta: 0:01:05 acc1_pos: 96.354 acc5_pos: 100.000 loss: 0.1750 (0.4759) acc1: 96.3542 (75.2196) acc5: 100.0000 (100.0000) time: 0.2360 data: 0.1416 max mem: 3531 +Test: [ 60/234] eta: 0:00:57 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.1465 (0.4161) acc1: 97.9167 (79.2222) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2099 data: 0.1178 max mem: 3531 +Test: [ 70/234] eta: 0:00:51 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.1153 (0.4171) acc1: 99.4792 (79.2950) acc5: 100.0000 (100.0000) acc1_neg: 68.750 acc5_neg: 100.000 time: 0.2067 data: 0.1113 max mem: 3531 +Test: [ 80/234] eta: 0:00:45 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.5441 (0.4281) acc1: 70.3125 (78.7712) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.1910 data: 0.0970 max mem: 3531 +Test: [ 90/234] eta: 0:00:40 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.3738 (0.4096) acc1: 81.2500 (79.9422) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.1772 data: 0.0867 max mem: 3531 +Test: [100/234] eta: 0:00:36 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.3507 (0.4160) acc1: 83.3333 (79.6024) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1763 data: 0.0854 max mem: 3531 +Test: [110/234] eta: 0:00:32 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.4326 (0.4171) acc1: 78.3854 (79.6617) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.1750 data: 0.0842 max mem: 3531 +Test: [120/234] eta: 0:00:29 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.4116 (0.4279) acc1: 79.9479 (79.1172) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1679 data: 0.0775 max mem: 3531 +Test: [130/234] eta: 0:00:26 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.4166 (0.4299) acc1: 80.7292 (79.1905) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.2047 data: 0.1108 max mem: 3531 +Test: [140/234] eta: 0:00:23 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.4404 (0.4477) acc1: 76.8229 (78.3688) acc5: 100.0000 (100.0000) acc1_neg: 36.198 acc5_neg: 100.000 time: 0.2244 data: 0.1273 max mem: 3531 +Test: [150/234] eta: 0:00:20 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.5822 (0.4604) acc1: 68.7500 (77.5317) acc5: 100.0000 (100.0000) acc1_neg: 76.042 acc5_neg: 100.000 time: 0.1649 data: 0.0675 max mem: 3531 +Test: [160/234] eta: 0:00:17 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.4984 (0.4612) acc1: 73.6979 (77.5540) acc5: 100.0000 (100.0000) acc1_neg: 77.083 acc5_neg: 100.000 time: 0.1749 data: 0.0794 max mem: 3531 +Test: [170/234] eta: 0:00:15 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.4234 (0.4582) acc1: 81.5104 (77.9027) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.2199 data: 0.1252 max mem: 3531 +Test: [180/234] eta: 0:00:12 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.4234 (0.4679) acc1: 81.5104 (77.2934) acc5: 100.0000 (100.0000) acc1_neg: 33.854 acc5_neg: 100.000 time: 0.2170 data: 0.1213 max mem: 3531 +Test: [190/234] eta: 0:00:10 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.2799 (0.4596) acc1: 90.1042 (77.9314) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2177 data: 0.1245 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.2535 (0.4450) acc1: 92.7083 (78.8765) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1961 data: 0.1053 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.1052 (0.4323) acc1: 100.0000 (79.7011) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1788 data: 0.0884 max mem: 3532 +Test: [220/234] eta: 0:00:03 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.3032 (0.4269) acc1: 90.8854 (80.1871) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1563 data: 0.0656 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.3424 (0.4244) acc1: 86.4583 (80.3786) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.1293 data: 0.0385 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.3526 (0.4237) acc1: 85.6771 (80.4132) acc5: 100.0000 (100.0000) acc1_neg: 84.047 acc5_neg: 100.000 time: 0.1333 data: 0.0415 max mem: 3532 +Test: Total time: 0:00:52 (0.2227 s / it) +* Overall Acc@1 80.413 Acc@5 100.000 + target == real Acc@1 75.366 Acc@5 100.000 + target == fake Acc@1 81.836 Acc@5 100.000 + loss 0.424 + PR-AUC 0.9635 ROC-AUC 0.8797 EER 20.73% (thr=0.5396) +Accuracy of the network on the 89729 test images: 80.4% +[rank0]:[W725 18:18:36.865746113 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:18:56.509000 603968 torch/distributed/run.py:766] +W0725 18:18:56.509000 603968 torch/distributed/run.py:766] ***************************************** +W0725 18:18:56.509000 603968 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:18:56.509000 603968 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 3): env:// +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:19:26.607153211 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:19:27.998561258 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank1]:[W725 18:19:27.000908947 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:19:27.105214363 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:19:27.253792023 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W725 18:19:27.255761176 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:19:27.402549553 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:19:27.402554846 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed4_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n11_classseed4_resampled_n116-041-188_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:08:05 acc1_pos: 32.031 acc5_pos: 100.000 loss: 1.0930 (1.0930) acc1: 32.0312 (32.0312) acc5: 100.0000 (100.0000) time: 6.2361 data: 4.5236 max mem: 3531 +Test: [ 10/2387] eta: 0:25:50 acc1_pos: 18.229 acc5_pos: 100.000 loss: 1.1794 (1.2340) acc1: 18.2292 (18.7500) acc5: 100.0000 (100.0000) time: 0.6522 data: 0.4115 max mem: 3531 +Test: [ 20/2387] eta: 0:15:20 acc1_pos: 33.073 acc5_pos: 100.000 loss: 1.1726 (1.1719) acc1: 18.4896 (21.5278) acc5: 100.0000 (100.0000) time: 0.0967 data: 0.0003 max mem: 3531 +Test: [ 30/2387] eta: 0:13:38 acc1_pos: 2.865 acc5_pos: 100.000 loss: 1.1726 (1.2253) acc1: 17.7083 (19.2204) acc5: 100.0000 (100.0000) time: 0.1797 data: 0.0827 max mem: 3531 +Test: [ 40/2387] eta: 0:13:36 acc1_pos: 20.052 acc5_pos: 100.000 loss: 1.3087 (1.2551) acc1: 11.7188 (17.3653) acc5: 100.0000 (100.0000) time: 0.3049 data: 0.2064 max mem: 3531 +Test: [ 50/2387] eta: 0:13:20 acc1_pos: 47.917 acc5_pos: 100.000 loss: 1.2782 (1.2368) acc1: 11.9792 (18.0709) acc5: 100.0000 (100.0000) time: 0.3349 data: 0.2382 max mem: 3531 +Test: [ 60/2387] eta: 0:12:49 acc1_pos: 47.396 acc5_pos: 100.000 loss: 1.1188 (1.2050) acc1: 23.9583 (20.2357) acc5: 100.0000 (100.0000) time: 0.2949 data: 0.2029 max mem: 3531 +Test: [ 70/2387] eta: 0:12:42 acc1_pos: 44.792 acc5_pos: 100.000 loss: 1.0487 (1.1794) acc1: 27.3438 (21.4532) acc5: 100.0000 (100.0000) time: 0.2949 data: 0.2032 max mem: 3531 +Test: [ 80/2387] eta: 0:12:39 acc1_pos: 46.875 acc5_pos: 100.000 loss: 0.9962 (1.1610) acc1: 28.1250 (22.4955) acc5: 100.0000 (100.0000) time: 0.3249 data: 0.2317 max mem: 3531 +Test: [ 90/2387] eta: 0:12:03 acc1_pos: 40.104 acc5_pos: 100.000 loss: 0.9599 (1.1329) acc1: 30.9896 (23.9955) acc5: 100.0000 (100.0000) time: 0.2649 data: 0.1698 max mem: 3531 +Test: [ 100/2387] eta: 0:11:47 acc1_pos: 50.000 acc5_pos: 100.000 loss: 0.9564 (1.1199) acc1: 30.9896 (24.7860) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1367 max mem: 3531 +Test: [ 110/2387] eta: 0:11:32 acc1_pos: 39.583 acc5_pos: 100.000 loss: 0.9296 (1.1045) acc1: 29.1667 (25.4997) acc5: 100.0000 (100.0000) time: 0.2549 data: 0.1612 max mem: 3531 +Test: [ 120/2387] eta: 0:11:19 acc1_pos: 40.885 acc5_pos: 100.000 loss: 0.9209 (1.0922) acc1: 32.2917 (26.4764) acc5: 100.0000 (100.0000) time: 0.2499 data: 0.1528 max mem: 3531 +Test: [ 130/2387] eta: 0:11:07 acc1_pos: 36.458 acc5_pos: 100.000 loss: 0.9899 (1.0866) acc1: 31.2500 (26.4909) acc5: 100.0000 (100.0000) time: 0.2499 data: 0.1522 max mem: 3531 +Test: [ 140/2387] eta: 0:11:00 acc1_pos: 42.188 acc5_pos: 100.000 loss: 1.0299 (1.0804) acc1: 29.1667 (26.9430) acc5: 100.0000 (100.0000) time: 0.2599 data: 0.1605 max mem: 3531 +Test: [ 150/2387] eta: 0:10:46 acc1_pos: 20.312 acc5_pos: 100.000 loss: 0.9256 (1.0698) acc1: 32.0312 (27.5231) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1475 max mem: 3531 +Test: [ 160/2387] eta: 0:10:48 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.9088 (1.0429) acc1: 35.4167 (29.6357) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2699 data: 0.1750 max mem: 3531 +Test: [ 170/2387] eta: 0:10:23 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2844 (0.9982) acc1: 92.4479 (33.3745) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2199 data: 0.1200 max mem: 3531 +Test: [ 180/2387] eta: 0:10:12 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2740 (0.9580) acc1: 94.2708 (36.7303) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1699 data: 0.0718 max mem: 3531 +Test: [ 190/2387] eta: 0:10:02 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2736 (0.9223) acc1: 94.2708 (39.7456) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2149 data: 0.1212 max mem: 3531 +Test: [ 200/2387] eta: 0:09:53 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2730 (0.8900) acc1: 94.2708 (42.4479) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2149 data: 0.1204 max mem: 3531 +Test: [ 210/2387] eta: 0:09:44 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2742 (0.8610) acc1: 93.7500 (44.8867) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2149 data: 0.1201 max mem: 3532 +Test: [ 220/2387] eta: 0:09:29 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2777 (0.8344) acc1: 93.4896 (47.0954) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1749 data: 0.0792 max mem: 3532 +Test: [ 230/2387] eta: 0:09:16 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2777 (0.8105) acc1: 93.4896 (49.1071) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1455 data: 0.0499 max mem: 3532 +Test: [ 240/2387] eta: 0:09:15 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2836 (0.7888) acc1: 93.4896 (50.9639) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.2149 data: 0.1215 max mem: 3532 +Test: [ 250/2387] eta: 0:09:06 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2725 (0.7639) acc1: 95.5729 (52.8210) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2299 data: 0.1363 max mem: 3532 +Test: [ 260/2387] eta: 0:08:58 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1398 (0.7400) acc1: 98.9583 (54.5887) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1877 data: 0.0933 max mem: 3532 +Test: [ 270/2387] eta: 0:08:52 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1398 (0.7179) acc1: 99.2188 (56.2068) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1970 data: 0.1039 max mem: 3532 +Test: [ 280/2387] eta: 0:08:46 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1409 (0.6974) acc1: 98.6979 (57.7180) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2084 data: 0.1159 max mem: 3532 +Test: [ 290/2387] eta: 0:08:36 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1169 (0.6774) acc1: 99.4792 (59.1549) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1821 data: 0.0896 max mem: 3532 +Test: [ 300/2387] eta: 0:08:25 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1034 (0.6585) acc1: 100.0000 (60.5041) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1312 data: 0.0368 max mem: 3532 +Test: [ 310/2387] eta: 0:08:20 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1055 (0.6477) acc1: 100.0000 (61.2892) acc5: 100.0000 (100.0000) acc1_neg: 69.010 acc5_neg: 100.000 time: 0.1619 data: 0.0656 max mem: 3532 +Test: [ 320/2387] eta: 0:08:14 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.6148 (0.6470) acc1: 65.3646 (61.3772) acc5: 100.0000 (100.0000) acc1_neg: 62.760 acc5_neg: 100.000 time: 0.1983 data: 0.1052 max mem: 3532 +Test: [ 330/2387] eta: 0:08:08 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.6198 (0.6461) acc1: 65.1042 (61.5087) acc5: 100.0000 (100.0000) acc1_neg: 69.010 acc5_neg: 100.000 time: 0.1841 data: 0.0922 max mem: 3532 +Test: [ 340/2387] eta: 0:08:04 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.6178 (0.6454) acc1: 65.6250 (61.6157) acc5: 100.0000 (100.0000) acc1_neg: 61.979 acc5_neg: 100.000 time: 0.1938 data: 0.0990 max mem: 3532 +Test: [ 350/2387] eta: 0:07:59 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.5998 (0.6347) acc1: 67.7083 (62.5742) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1962 data: 0.1025 max mem: 3532 +Test: [ 360/2387] eta: 0:07:55 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2240 (0.6233) acc1: 97.1354 (63.5164) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1959 data: 0.1047 max mem: 3532 +Test: [ 370/2387] eta: 0:07:50 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2247 (0.6126) acc1: 96.6146 (64.4156) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2014 data: 0.1095 max mem: 3532 +Test: [ 380/2387] eta: 0:07:48 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2247 (0.6023) acc1: 96.6146 (65.2634) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2100 data: 0.1158 max mem: 3532 +Test: [ 390/2387] eta: 0:07:41 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2239 (0.5927) acc1: 96.3542 (66.0526) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1839 data: 0.0896 max mem: 3532 +Test: [ 400/2387] eta: 0:07:36 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2248 (0.5835) acc1: 96.3542 (66.8154) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1638 data: 0.0708 max mem: 3532 +Test: [ 410/2387] eta: 0:07:32 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2232 (0.5747) acc1: 96.3542 (67.5398) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1929 data: 0.0979 max mem: 3532 +Test: [ 420/2387] eta: 0:07:28 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2201 (0.5662) acc1: 96.6146 (68.2422) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1949 data: 0.1004 max mem: 3532 +Test: [ 430/2387] eta: 0:07:24 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2218 (0.5584) acc1: 96.8750 (68.9041) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1905 data: 0.0948 max mem: 3532 +Test: [ 440/2387] eta: 0:07:21 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2218 (0.5488) acc1: 97.1354 (69.5814) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1955 data: 0.1002 max mem: 3533 +Test: [ 450/2387] eta: 0:07:16 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.0818 (0.5385) acc1: 100.0000 (70.2559) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1865 data: 0.0951 max mem: 3533 +Test: [ 460/2387] eta: 0:07:13 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.0819 (0.5286) acc1: 100.0000 (70.9011) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1843 data: 0.0900 max mem: 3533 +Test: [ 470/2387] eta: 0:07:09 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.0819 (0.5191) acc1: 100.0000 (71.5189) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1885 data: 0.0930 max mem: 3533 +Test: [ 480/2387] eta: 0:07:05 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.0827 (0.5100) acc1: 100.0000 (72.1111) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1869 data: 0.0943 max mem: 3533 +Test: [ 490/2387] eta: 0:07:02 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.0821 (0.5013) acc1: 100.0000 (72.6785) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1891 data: 0.0977 max mem: 3533 +Test: [ 500/2387] eta: 0:06:58 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.0818 (0.4930) acc1: 100.0000 (73.2239) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1844 data: 0.0919 max mem: 3533 +Test: [ 510/2387] eta: 0:06:54 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.0824 (0.4852) acc1: 100.0000 (73.7479) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1755 data: 0.0815 max mem: 3533 +Test: [ 520/2387] eta: 0:06:50 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1343 (0.4787) acc1: 100.0000 (74.2512) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1748 data: 0.0819 max mem: 3533 +Test: [ 530/2387] eta: 0:06:46 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1479 (0.4725) acc1: 100.0000 (74.7362) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1800 data: 0.0885 max mem: 3533 +Test: [ 540/2387] eta: 0:06:43 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1507 (0.4666) acc1: 100.0000 (75.2022) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1762 data: 0.0841 max mem: 3533 +Test: [ 550/2387] eta: 0:06:39 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1505 (0.4608) acc1: 100.0000 (75.6508) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1793 data: 0.0874 max mem: 3533 +Test: [ 560/2387] eta: 0:06:36 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1519 (0.4553) acc1: 100.0000 (76.0830) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1852 data: 0.0940 max mem: 3533 +Test: [ 570/2387] eta: 0:06:33 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1525 (0.4500) acc1: 100.0000 (76.5000) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1806 data: 0.0879 max mem: 3533 +Test: [ 580/2387] eta: 0:06:29 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1493 (0.4449) acc1: 100.0000 (76.9036) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1737 data: 0.0795 max mem: 3533 +Test: [ 590/2387] eta: 0:06:26 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1491 (0.4399) acc1: 100.0000 (77.2935) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1837 data: 0.0889 max mem: 3533 +Test: [ 600/2387] eta: 0:06:23 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1479 (0.4349) acc1: 100.0000 (77.6674) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1910 data: 0.0950 max mem: 3533 +Test: [ 610/2387] eta: 0:06:20 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1770 (0.4320) acc1: 96.0938 (77.9362) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1846 data: 0.0906 max mem: 3533 +Test: [ 620/2387] eta: 0:06:18 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2537 (0.4292) acc1: 94.2708 (78.1921) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1889 data: 0.0958 max mem: 3533 +Test: [ 630/2387] eta: 0:06:15 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2537 (0.4266) acc1: 93.7500 (78.4424) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1902 data: 0.0950 max mem: 3533 +Test: [ 640/2387] eta: 0:06:12 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2658 (0.4241) acc1: 93.7500 (78.6718) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1831 data: 0.0865 max mem: 3533 +Test: [ 650/2387] eta: 0:06:09 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2693 (0.4217) acc1: 93.2292 (78.8967) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1847 data: 0.0891 max mem: 3533 +Test: [ 660/2387] eta: 0:06:06 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2608 (0.4192) acc1: 93.4896 (79.1281) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1899 data: 0.0957 max mem: 3533 +Test: [ 670/2387] eta: 0:06:03 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2575 (0.4169) acc1: 94.7917 (79.3576) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1827 data: 0.0890 max mem: 3534 +Test: [ 680/2387] eta: 0:06:00 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2529 (0.4143) acc1: 95.0521 (79.6064) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1849 data: 0.0908 max mem: 3534 +Test: [ 690/2387] eta: 0:05:58 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2349 (0.4117) acc1: 95.5729 (79.8269) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1921 data: 0.0971 max mem: 3534 +Test: [ 700/2387] eta: 0:05:56 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2273 (0.4090) acc1: 94.5312 (80.0367) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2013 data: 0.1081 max mem: 3534 +Test: [ 710/2387] eta: 0:05:53 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2242 (0.4064) acc1: 94.2708 (80.2398) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1956 data: 0.1014 max mem: 3534 +Test: [ 720/2387] eta: 0:05:51 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2231 (0.4039) acc1: 94.0104 (80.4323) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2038 data: 0.1098 max mem: 3534 +Test: [ 730/2387] eta: 0:05:48 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2244 (0.4015) acc1: 93.7500 (80.6127) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2018 data: 0.1093 max mem: 3534 +Test: [ 740/2387] eta: 0:05:45 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2168 (0.3990) acc1: 94.2708 (80.8044) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1696 data: 0.0755 max mem: 3534 +Test: [ 750/2387] eta: 0:05:43 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2168 (0.3966) acc1: 94.5312 (80.9882) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1873 data: 0.0938 max mem: 3534 +Test: [ 760/2387] eta: 0:05:42 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2182 (0.3940) acc1: 95.0521 (81.1843) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2249 data: 0.1305 max mem: 3534 +Test: [ 770/2387] eta: 0:05:39 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1731 (0.3909) acc1: 96.8750 (81.4060) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2249 data: 0.1284 max mem: 3534 +Test: [ 780/2387] eta: 0:05:35 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1539 (0.3880) acc1: 98.4375 (81.6278) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1562 data: 0.0594 max mem: 3534 +Test: [ 790/2387] eta: 0:05:32 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1364 (0.3851) acc1: 99.7396 (81.8498) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1414 data: 0.0456 max mem: 3534 +Test: [ 800/2387] eta: 0:05:30 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1585 (0.3825) acc1: 99.4792 (82.0621) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1808 data: 0.0872 max mem: 3534 +Test: [ 810/2387] eta: 0:05:27 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1585 (0.3799) acc1: 98.9583 (82.2615) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1834 data: 0.0840 max mem: 3534 +Test: [ 820/2387] eta: 0:05:25 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1461 (0.3771) acc1: 98.9583 (82.4595) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1840 data: 0.0848 max mem: 3534 +Test: [ 830/2387] eta: 0:05:22 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1646 (0.3755) acc1: 98.4375 (82.6232) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1898 data: 0.0968 max mem: 3534 +Test: [ 840/2387] eta: 0:05:20 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2926 (0.3765) acc1: 91.4062 (82.5496) acc5: 100.0000 (100.0000) acc1_neg: 69.531 acc5_neg: 100.000 time: 0.1795 data: 0.0834 max mem: 3534 +Test: [ 850/2387] eta: 0:05:17 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.5170 (0.3784) acc1: 71.8750 (82.4104) acc5: 100.0000 (100.0000) acc1_neg: 71.615 acc5_neg: 100.000 time: 0.1724 data: 0.0760 max mem: 3534 +Test: [ 860/2387] eta: 0:05:15 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.5191 (0.3800) acc1: 71.8750 (82.3053) acc5: 100.0000 (100.0000) acc1_neg: 70.312 acc5_neg: 100.000 time: 0.1866 data: 0.0920 max mem: 3534 +Test: [ 870/2387] eta: 0:05:12 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.4752 (0.3809) acc1: 76.8229 (82.2549) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1874 data: 0.0930 max mem: 3534 +Test: [ 880/2387] eta: 0:05:10 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.4712 (0.3821) acc1: 74.2188 (82.1655) acc5: 100.0000 (100.0000) acc1_neg: 74.219 acc5_neg: 100.000 time: 0.1833 data: 0.0912 max mem: 3534 +Test: [ 890/2387] eta: 0:05:07 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.3820 (0.3815) acc1: 84.6354 (82.2440) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1812 data: 0.0879 max mem: 3534 +Test: [ 900/2387] eta: 0:05:05 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.3870 (0.3820) acc1: 83.5938 (82.2269) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.1839 data: 0.0897 max mem: 3535 +Test: [ 910/2387] eta: 0:05:04 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.4575 (0.3830) acc1: 78.1250 (82.1819) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.2195 data: 0.1269 max mem: 3535 +Test: [ 920/2387] eta: 0:05:01 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.4715 (0.3839) acc1: 77.3438 (82.1316) acc5: 100.0000 (100.0000) acc1_neg: 77.083 acc5_neg: 100.000 time: 0.2226 data: 0.1284 max mem: 3535 +Test: [ 930/2387] eta: 0:04:58 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.4713 (0.3847) acc1: 77.3438 (82.0850) acc5: 100.0000 (100.0000) acc1_neg: 77.344 acc5_neg: 100.000 time: 0.1561 data: 0.0627 max mem: 3535 +Test: [ 940/2387] eta: 0:04:55 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.4659 (0.3856) acc1: 77.3438 (82.0335) acc5: 100.0000 (100.0000) acc1_neg: 74.740 acc5_neg: 100.000 time: 0.1409 data: 0.0499 max mem: 3535 +Test: [ 950/2387] eta: 0:04:53 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.4640 (0.3864) acc1: 78.3854 (82.0025) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.1729 data: 0.0792 max mem: 3535 +Test: [ 960/2387] eta: 0:04:51 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.4507 (0.3870) acc1: 79.4271 (81.9697) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.1861 data: 0.0919 max mem: 3535 +Test: [ 970/2387] eta: 0:04:48 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.4507 (0.3878) acc1: 78.1250 (81.9275) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.1907 data: 0.0961 max mem: 3535 +Test: [ 980/2387] eta: 0:04:46 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.4268 (0.3871) acc1: 80.4688 (82.0013) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.1799 data: 0.0829 max mem: 3535 +Test: [ 990/2387] eta: 0:04:43 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2089 (0.3851) acc1: 97.9167 (82.1316) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1663 data: 0.0709 max mem: 3535 +Test: [1000/2387] eta: 0:04:41 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1482 (0.3829) acc1: 99.7396 (82.3029) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1689 data: 0.0748 max mem: 3535 +Test: [1010/2387] eta: 0:04:38 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1606 (0.3808) acc1: 99.2188 (82.4604) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1666 data: 0.0727 max mem: 3535 +Test: [1020/2387] eta: 0:04:36 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1899 (0.3789) acc1: 98.4375 (82.6095) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1700 data: 0.0775 max mem: 3535 +Test: [1030/2387] eta: 0:04:33 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1707 (0.3768) acc1: 99.2188 (82.7701) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1753 data: 0.0817 max mem: 3535 +Test: [1040/2387] eta: 0:04:31 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1613 (0.3748) acc1: 98.9583 (82.9181) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1569 data: 0.0632 max mem: 3535 +Test: [1050/2387] eta: 0:04:28 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1587 (0.3728) acc1: 98.9583 (83.0640) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1551 data: 0.0628 max mem: 3535 +Test: [1060/2387] eta: 0:04:26 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1482 (0.3709) acc1: 99.2188 (83.2104) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1685 data: 0.0758 max mem: 3535 +Test: [1070/2387] eta: 0:04:23 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1480 (0.3690) acc1: 99.2188 (83.3491) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1611 data: 0.0681 max mem: 3535 +Test: [1080/2387] eta: 0:04:21 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2031 (0.3677) acc1: 98.1771 (83.4709) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1571 data: 0.0632 max mem: 3535 +Test: [1090/2387] eta: 0:04:18 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2135 (0.3660) acc1: 97.6562 (83.6014) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1582 data: 0.0658 max mem: 3535 +Test: [1100/2387] eta: 0:04:16 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1841 (0.3643) acc1: 98.4375 (83.7404) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1647 data: 0.0696 max mem: 3535 +Test: [1110/2387] eta: 0:04:14 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1914 (0.3629) acc1: 98.4375 (83.8692) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1820 data: 0.0856 max mem: 3535 +Test: [1120/2387] eta: 0:04:11 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2093 (0.3616) acc1: 97.6562 (83.9880) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1722 data: 0.0778 max mem: 3536 +Test: [1130/2387] eta: 0:04:09 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1824 (0.3600) acc1: 98.6979 (84.1203) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1676 data: 0.0733 max mem: 3536 +Test: [1140/2387] eta: 0:04:07 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1903 (0.3590) acc1: 98.6979 (84.2330) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1775 data: 0.0851 max mem: 3536 +Test: [1150/2387] eta: 0:04:05 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2573 (0.3586) acc1: 97.1354 (84.2782) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.1919 data: 0.0989 max mem: 3536 +Test: [1160/2387] eta: 0:04:03 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2920 (0.3581) acc1: 90.1042 (84.3308) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.2026 data: 0.1093 max mem: 3536 +Test: [1170/2387] eta: 0:04:01 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.3202 (0.3578) acc1: 90.1042 (84.3803) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1838 data: 0.0900 max mem: 3536 +Test: [1180/2387] eta: 0:03:58 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.3245 (0.3575) acc1: 90.1042 (84.4319) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1661 data: 0.0716 max mem: 3536 +Test: [1190/2387] eta: 0:03:56 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2979 (0.3567) acc1: 92.7083 (84.5152) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1646 data: 0.0701 max mem: 3536 +Test: [1200/2387] eta: 0:03:54 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2039 (0.3556) acc1: 96.8750 (84.6022) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1555 data: 0.0597 max mem: 3536 +Test: [1210/2387] eta: 0:03:52 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2021 (0.3543) acc1: 96.8750 (84.6961) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1692 data: 0.0747 max mem: 3536 +Test: [1220/2387] eta: 0:03:50 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1710 (0.3527) acc1: 96.8750 (84.7984) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1933 data: 0.1019 max mem: 3536 +Test: [1230/2387] eta: 0:03:47 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1461 (0.3510) acc1: 97.3958 (84.9018) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1720 data: 0.0811 max mem: 3536 +Test: [1240/2387] eta: 0:03:45 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1470 (0.3495) acc1: 96.8750 (84.9938) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1705 data: 0.0792 max mem: 3536 +Test: [1250/2387] eta: 0:03:43 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1312 (0.3478) acc1: 97.3958 (85.0990) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1790 data: 0.0869 max mem: 3536 +Test: [1260/2387] eta: 0:03:41 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1247 (0.3461) acc1: 98.1771 (85.2052) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1766 data: 0.0838 max mem: 3536 +Test: [1270/2387] eta: 0:03:38 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1097 (0.3441) acc1: 100.0000 (85.3208) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1650 data: 0.0736 max mem: 3536 +Test: [1280/2387] eta: 0:03:36 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.0946 (0.3422) acc1: 100.0000 (85.4319) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1491 data: 0.0586 max mem: 3536 +Test: [1290/2387] eta: 0:03:34 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1029 (0.3404) acc1: 99.7396 (85.5417) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1543 data: 0.0634 max mem: 3536 +Test: [1300/2387] eta: 0:03:32 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1070 (0.3386) acc1: 99.7396 (85.6485) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1566 data: 0.0634 max mem: 3536 +Test: [1310/2387] eta: 0:03:29 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1164 (0.3370) acc1: 99.2188 (85.7486) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1553 data: 0.0613 max mem: 3536 +Test: [1320/2387] eta: 0:03:27 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1083 (0.3352) acc1: 99.7396 (85.8549) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1580 data: 0.0645 max mem: 3536 +Test: [1330/2387] eta: 0:03:25 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1090 (0.3339) acc1: 99.7396 (85.9502) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1660 data: 0.0725 max mem: 3536 +Test: [1340/2387] eta: 0:03:23 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1905 (0.3332) acc1: 97.6562 (86.0037) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1738 data: 0.0810 max mem: 3536 +Test: [1350/2387] eta: 0:03:21 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1001 (0.3314) acc1: 100.0000 (86.1073) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1753 data: 0.0829 max mem: 3537 +Test: [1360/2387] eta: 0:03:19 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.0747 (0.3295) acc1: 100.0000 (86.2094) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1753 data: 0.0834 max mem: 3537 +Test: [1370/2387] eta: 0:03:17 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.0868 (0.3278) acc1: 100.0000 (86.3100) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1788 data: 0.0869 max mem: 3537 +Test: [1380/2387] eta: 0:03:15 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.0824 (0.3260) acc1: 100.0000 (86.4091) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1795 data: 0.0861 max mem: 3537 +Test: [1390/2387] eta: 0:03:13 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.0739 (0.3242) acc1: 100.0000 (86.5068) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1895 data: 0.0952 max mem: 3537 +Test: [1400/2387] eta: 0:03:11 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.0645 (0.3224) acc1: 100.0000 (86.6009) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0920 max mem: 3537 +Test: [1410/2387] eta: 0:03:08 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.0614 (0.3213) acc1: 100.0000 (86.6804) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1690 data: 0.0759 max mem: 3537 +Test: [1420/2387] eta: 0:03:07 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2066 (0.3206) acc1: 97.6562 (86.7457) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1803 data: 0.0860 max mem: 3537 +Test: [1430/2387] eta: 0:03:04 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2274 (0.3201) acc1: 96.6146 (86.8043) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1859 data: 0.0927 max mem: 3537 +Test: [1440/2387] eta: 0:03:03 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2170 (0.3193) acc1: 96.8750 (86.8823) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1903 data: 0.0990 max mem: 3537 +Test: [1450/2387] eta: 0:03:01 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2163 (0.3187) acc1: 97.1354 (86.9465) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1911 data: 0.0997 max mem: 3537 +Test: [1460/2387] eta: 0:02:59 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2302 (0.3180) acc1: 97.1354 (87.0129) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1893 data: 0.0962 max mem: 3537 +Test: [1470/2387] eta: 0:02:57 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1900 (0.3172) acc1: 97.1354 (87.0833) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2016 data: 0.1074 max mem: 3537 +Test: [1480/2387] eta: 0:02:55 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1958 (0.3165) acc1: 96.6146 (87.1448) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1892 data: 0.0960 max mem: 3537 +Test: [1490/2387] eta: 0:02:53 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2156 (0.3156) acc1: 97.3958 (87.2185) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1848 data: 0.0914 max mem: 3537 +Test: [1500/2387] eta: 0:02:51 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1764 (0.3147) acc1: 98.6979 (87.2939) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1895 data: 0.0967 max mem: 3537 +Test: [1510/2387] eta: 0:02:49 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1702 (0.3138) acc1: 99.4792 (87.3678) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1905 data: 0.0985 max mem: 3537 +Test: [1520/2387] eta: 0:02:47 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2304 (0.3141) acc1: 96.6146 (87.3485) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1807 data: 0.0867 max mem: 3537 +Test: [1530/2387] eta: 0:02:45 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.3905 (0.3150) acc1: 82.8125 (87.2896) acc5: 100.0000 (100.0000) acc1_neg: 68.750 acc5_neg: 100.000 time: 0.1701 data: 0.0761 max mem: 3537 +Test: [1540/2387] eta: 0:02:43 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.4300 (0.3157) acc1: 79.6875 (87.2470) acc5: 100.0000 (100.0000) acc1_neg: 70.312 acc5_neg: 100.000 time: 0.1779 data: 0.0842 max mem: 3537 +Test: [1550/2387] eta: 0:02:41 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.4220 (0.3162) acc1: 81.7708 (87.2250) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1745 data: 0.0797 max mem: 3537 +Test: [1560/2387] eta: 0:02:39 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.4030 (0.3167) acc1: 81.7708 (87.1937) acc5: 100.0000 (100.0000) acc1_neg: 75.521 acc5_neg: 100.000 time: 0.1767 data: 0.0835 max mem: 3537 +Test: [1570/2387] eta: 0:02:37 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.3029 (0.3165) acc1: 90.3646 (87.2240) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1945 data: 0.1007 max mem: 3537 +Test: [1580/2387] eta: 0:02:35 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2060 (0.3156) acc1: 96.3542 (87.2936) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1908 data: 0.0958 max mem: 3538 +Test: [1590/2387] eta: 0:02:33 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1440 (0.3145) acc1: 99.4792 (87.3709) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1826 data: 0.0908 max mem: 3538 +Test: [1600/2387] eta: 0:02:31 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1526 (0.3136) acc1: 99.4792 (87.4452) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1922 data: 0.1012 max mem: 3538 +Test: [1610/2387] eta: 0:02:29 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1659 (0.3127) acc1: 99.2188 (87.5168) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1872 data: 0.0912 max mem: 3538 +Test: [1620/2387] eta: 0:02:27 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1494 (0.3117) acc1: 99.2188 (87.5871) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1858 data: 0.0900 max mem: 3538 +Test: [1630/2387] eta: 0:02:25 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1625 (0.3109) acc1: 98.9583 (87.6526) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1910 data: 0.1000 max mem: 3538 +Test: [1640/2387] eta: 0:02:23 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1966 (0.3104) acc1: 97.6562 (87.7104) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1865 data: 0.0951 max mem: 3538 +Test: [1650/2387] eta: 0:02:21 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2233 (0.3098) acc1: 96.3542 (87.7592) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1837 data: 0.0908 max mem: 3538 +Test: [1660/2387] eta: 0:02:19 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2159 (0.3093) acc1: 96.3542 (87.8125) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1902 data: 0.0976 max mem: 3538 +Test: [1670/2387] eta: 0:02:17 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2136 (0.3087) acc1: 97.3958 (87.8718) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1925 data: 0.1002 max mem: 3538 +Test: [1680/2387] eta: 0:02:16 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2027 (0.3082) acc1: 97.6562 (87.9174) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.2034 data: 0.1108 max mem: 3538 +Test: [1690/2387] eta: 0:02:14 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2198 (0.3078) acc1: 96.6146 (87.9643) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1936 data: 0.1014 max mem: 3538 +Test: [1700/2387] eta: 0:02:12 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2204 (0.3073) acc1: 96.6146 (88.0074) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1640 data: 0.0697 max mem: 3538 +Test: [1710/2387] eta: 0:02:10 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2160 (0.3067) acc1: 96.6146 (88.0389) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.1746 data: 0.0813 max mem: 3538 +Test: [1720/2387] eta: 0:02:08 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1740 (0.3060) acc1: 96.6146 (88.0858) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1919 data: 0.1007 max mem: 3538 +Test: [1730/2387] eta: 0:02:06 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1787 (0.3056) acc1: 96.3542 (88.1141) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2020 data: 0.1101 max mem: 3538 +Test: [1740/2387] eta: 0:02:04 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1787 (0.3047) acc1: 95.0521 (88.1610) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.2125 data: 0.1181 max mem: 3538 +Test: [1750/2387] eta: 0:02:02 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1609 (0.3040) acc1: 97.3958 (88.2145) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2027 data: 0.1061 max mem: 3538 +Test: [1760/2387] eta: 0:02:00 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1428 (0.3030) acc1: 98.9583 (88.2767) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2009 data: 0.1069 max mem: 3538 +Test: [1770/2387] eta: 0:01:58 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1492 (0.3026) acc1: 98.6979 (88.2939) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2089 data: 0.1179 max mem: 3538 +Test: [1780/2387] eta: 0:01:56 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2004 (0.3023) acc1: 94.7917 (88.3172) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1946 data: 0.1026 max mem: 3538 +Test: [1790/2387] eta: 0:01:55 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2163 (0.3020) acc1: 94.2708 (88.3385) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2149 data: 0.1220 max mem: 3538 +Test: [1800/2387] eta: 0:01:53 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2156 (0.3016) acc1: 94.2708 (88.3599) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2243 data: 0.1308 max mem: 3538 +Test: [1810/2387] eta: 0:01:51 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1956 (0.3010) acc1: 95.0521 (88.4013) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1917 data: 0.0979 max mem: 3539 +Test: [1820/2387] eta: 0:01:49 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1554 (0.3002) acc1: 97.9167 (88.4550) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1964 data: 0.1046 max mem: 3539 +Test: [1830/2387] eta: 0:01:47 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1440 (0.2994) acc1: 99.2188 (88.5138) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2080 data: 0.1128 max mem: 3539 +Test: [1840/2387] eta: 0:01:45 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1439 (0.2985) acc1: 99.4792 (88.5743) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1995 data: 0.1038 max mem: 3539 +Test: [1850/2387] eta: 0:01:43 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1436 (0.2977) acc1: 99.7396 (88.6344) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1773 data: 0.0839 max mem: 3539 +Test: [1860/2387] eta: 0:01:41 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1429 (0.2969) acc1: 99.7396 (88.6936) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1706 data: 0.0774 max mem: 3539 +Test: [1870/2387] eta: 0:01:39 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1437 (0.2961) acc1: 99.7396 (88.7511) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1735 data: 0.0787 max mem: 3539 +Test: [1880/2387] eta: 0:01:37 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1445 (0.2953) acc1: 99.7396 (88.8094) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1753 data: 0.0793 max mem: 3539 +Test: [1890/2387] eta: 0:01:35 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1426 (0.2945) acc1: 99.7396 (88.8669) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1824 data: 0.0903 max mem: 3539 +Test: [1900/2387] eta: 0:01:33 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1467 (0.2937) acc1: 99.7396 (88.9236) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1735 data: 0.0830 max mem: 3539 +Test: [1910/2387] eta: 0:01:31 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1520 (0.2931) acc1: 99.7396 (88.9758) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1326 data: 0.0396 max mem: 3539 +Test: [1920/2387] eta: 0:01:29 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1635 (0.2927) acc1: 99.7396 (89.0225) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1159 data: 0.0208 max mem: 3539 +Test: [1930/2387] eta: 0:01:27 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1910 (0.2921) acc1: 99.2188 (89.0694) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1341 data: 0.0405 max mem: 3539 +Test: [1940/2387] eta: 0:01:25 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1910 (0.2918) acc1: 98.9583 (89.1128) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1453 data: 0.0524 max mem: 3539 +Test: [1950/2387] eta: 0:01:23 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2469 (0.2917) acc1: 96.3542 (89.1386) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1554 data: 0.0622 max mem: 3539 +Test: [1960/2387] eta: 0:01:21 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2160 (0.2912) acc1: 96.8750 (89.1840) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1542 data: 0.0596 max mem: 3539 +Test: [1970/2387] eta: 0:01:19 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2088 (0.2910) acc1: 98.9583 (89.2217) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1551 data: 0.0609 max mem: 3539 +Test: [1980/2387] eta: 0:01:17 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2271 (0.2906) acc1: 98.9583 (89.2577) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1580 data: 0.0639 max mem: 3539 +Test: [1990/2387] eta: 0:01:15 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1339 (0.2898) acc1: 98.9583 (89.3068) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1552 data: 0.0589 max mem: 3539 +Test: [2000/2387] eta: 0:01:13 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1314 (0.2890) acc1: 99.2188 (89.3575) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1599 data: 0.0657 max mem: 3539 +Test: [2010/2387] eta: 0:01:11 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1314 (0.2882) acc1: 99.2188 (89.4066) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1702 data: 0.0771 max mem: 3539 +Test: [2020/2387] eta: 0:01:09 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1301 (0.2874) acc1: 99.2188 (89.4558) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1749 data: 0.0812 max mem: 3539 +Test: [2030/2387] eta: 0:01:07 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1299 (0.2867) acc1: 99.4792 (89.5051) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1682 data: 0.0734 max mem: 3540 +Test: [2040/2387] eta: 0:01:05 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1318 (0.2859) acc1: 99.2188 (89.5536) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1649 data: 0.0688 max mem: 3540 +Test: [2050/2387] eta: 0:01:03 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1278 (0.2851) acc1: 99.4792 (89.6024) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1761 data: 0.0797 max mem: 3540 +Test: [2060/2387] eta: 0:01:02 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1264 (0.2844) acc1: 99.4792 (89.6495) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1849 data: 0.0896 max mem: 3540 +Test: [2070/2387] eta: 0:01:00 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1256 (0.2836) acc1: 99.4792 (89.6976) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1787 data: 0.0822 max mem: 3540 +Test: [2080/2387] eta: 0:00:58 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1233 (0.2828) acc1: 99.7396 (89.7459) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1658 data: 0.0671 max mem: 3540 +Test: [2090/2387] eta: 0:00:56 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1247 (0.2821) acc1: 99.4792 (89.7923) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1699 data: 0.0708 max mem: 3540 +Test: [2100/2387] eta: 0:00:54 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1249 (0.2813) acc1: 99.4792 (89.8385) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1645 data: 0.0656 max mem: 3540 +Test: [2110/2387] eta: 0:00:52 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1249 (0.2806) acc1: 99.4792 (89.8837) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1621 data: 0.0654 max mem: 3540 +Test: [2120/2387] eta: 0:00:50 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1297 (0.2801) acc1: 99.4792 (89.9243) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1744 data: 0.0798 max mem: 3540 +Test: [2130/2387] eta: 0:00:48 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1460 (0.2795) acc1: 99.4792 (89.9678) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1805 data: 0.0864 max mem: 3540 +Test: [2140/2387] eta: 0:00:46 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1509 (0.2790) acc1: 99.7396 (90.0100) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1649 data: 0.0687 max mem: 3540 +Test: [2150/2387] eta: 0:00:44 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1737 (0.2786) acc1: 98.9583 (90.0451) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1447 data: 0.0482 max mem: 3540 +Test: [2160/2387] eta: 0:00:42 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1570 (0.2780) acc1: 99.2188 (90.0868) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1599 data: 0.0635 max mem: 3540 +Test: [2170/2387] eta: 0:00:40 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1570 (0.2776) acc1: 99.2188 (90.1223) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1779 data: 0.0815 max mem: 3540 +Test: [2180/2387] eta: 0:00:39 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2220 (0.2781) acc1: 95.8333 (90.0959) acc5: 100.0000 (100.0000) acc1_neg: 66.146 acc5_neg: 100.000 time: 0.1821 data: 0.0885 max mem: 3540 +Test: [2190/2387] eta: 0:00:37 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2216 (0.2778) acc1: 96.0938 (90.1172) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1763 data: 0.0817 max mem: 3540 +Test: [2200/2387] eta: 0:00:35 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1720 (0.2774) acc1: 99.2188 (90.1495) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1927 data: 0.0970 max mem: 3540 +Test: [2210/2387] eta: 0:00:33 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2081 (0.2772) acc1: 96.3542 (90.1712) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1895 data: 0.0953 max mem: 3540 +Test: [2220/2387] eta: 0:00:31 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2461 (0.2770) acc1: 96.3542 (90.1984) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1719 data: 0.0787 max mem: 3540 +Test: [2230/2387] eta: 0:00:29 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1958 (0.2766) acc1: 96.8750 (90.2267) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1898 data: 0.0965 max mem: 3540 +Test: [2240/2387] eta: 0:00:27 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1776 (0.2762) acc1: 98.4375 (90.2609) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1957 data: 0.1024 max mem: 3540 +Test: [2250/2387] eta: 0:00:25 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2035 (0.2761) acc1: 97.3958 (90.2785) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1911 data: 0.0962 max mem: 3540 +Test: [2260/2387] eta: 0:00:23 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2020 (0.2756) acc1: 97.6562 (90.3148) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1922 data: 0.0984 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1242 (0.2749) acc1: 100.0000 (90.3566) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1887 data: 0.0969 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1388 (0.2744) acc1: 100.0000 (90.3966) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1883 data: 0.0958 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1425 (0.2737) acc1: 100.0000 (90.4374) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1891 data: 0.0972 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1379 (0.2732) acc1: 100.0000 (90.4763) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1816 data: 0.0901 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1447 (0.2727) acc1: 99.7396 (90.5129) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1947 data: 0.1014 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1643 (0.2723) acc1: 98.9583 (90.5417) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1853 data: 0.0911 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.1993 (0.2722) acc1: 97.3958 (90.5609) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1530 data: 0.0608 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2251 (0.2720) acc1: 97.3958 (90.5882) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1474 data: 0.0552 max mem: 3541 +Test: [2350/2387] eta: 0:00:06 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2343 (0.2721) acc1: 96.3542 (90.5999) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1426 data: 0.0511 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2971 (0.2723) acc1: 90.6250 (90.5872) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1364 data: 0.0462 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2390 (0.2721) acc1: 96.8750 (90.6175) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1444 data: 0.0543 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2390 (0.2720) acc1: 97.9167 (90.6380) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1364 data: 0.0460 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 47.945 acc5_pos: 100.000 loss: 0.2440 (0.2719) acc1: 96.6146 (90.6465) acc5: 100.0000 (100.0000) acc1_neg: 97.674 acc5_neg: 100.000 time: 0.1302 data: 0.0423 max mem: 3541 +Test: Total time: 0:07:26 (0.1872 s / it) +* Overall Acc@1 90.647 Acc@5 100.000 + target == real Acc@1 27.830 Acc@5 100.000 + target == fake Acc@1 95.058 Acc@5 100.000 + loss 0.272 + PR-AUC 0.9873 ROC-AUC 0.8469 EER 22.85% (thr=0.7661) +Accuracy of the network on the 916267 test images: 90.6% +[rank0]:[W725 18:27:22.240651140 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:27:42.251000 604631 torch/distributed/run.py:766] +W0725 18:27:42.251000 604631 torch/distributed/run.py:766] ***************************************** +W0725 18:27:42.251000 604631 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:27:42.251000 604631 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:28:10.373418196 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:28:10.562677587 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:28:10.884744633 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:28:11.948624721 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:28:11.965416546 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:28:11.004723426 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:28:11.200423327 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:28:11.293145987 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed4_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n11_classseed4_resampled_n116-041-188_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:11:17 acc1_pos: 32.031 acc5_pos: 100.000 loss: 1.0930 (1.0930) acc1: 32.0312 (32.0312) acc5: 100.0000 (100.0000) time: 8.2733 data: 6.2590 max mem: 3531 +Test: [ 10/517] eta: 0:07:03 acc1_pos: 18.229 acc5_pos: 100.000 loss: 1.1794 (1.2340) acc1: 18.2292 (18.7500) acc5: 100.0000 (100.0000) time: 0.8343 data: 0.5695 max mem: 3531 +Test: [ 20/517] eta: 0:04:00 acc1_pos: 33.073 acc5_pos: 100.000 loss: 1.1726 (1.1719) acc1: 18.4896 (21.5278) acc5: 100.0000 (100.0000) time: 0.0948 data: 0.0005 max mem: 3531 +Test: [ 30/517] eta: 0:03:23 acc1_pos: 2.865 acc5_pos: 100.000 loss: 1.1726 (1.2253) acc1: 17.7083 (19.2204) acc5: 100.0000 (100.0000) time: 0.1895 data: 0.0941 max mem: 3531 +Test: [ 40/517] eta: 0:02:55 acc1_pos: 20.052 acc5_pos: 100.000 loss: 1.3087 (1.2551) acc1: 11.7188 (17.3653) acc5: 100.0000 (100.0000) time: 0.2464 data: 0.1549 max mem: 3531 +Test: [ 50/517] eta: 0:02:42 acc1_pos: 23.394 acc5_pos: 100.000 loss: 1.2782 (1.2089) acc1: 11.9792 (20.5933) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.2399 data: 0.1466 max mem: 3531 +Test: [ 60/517] eta: 0:02:28 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.3987 (1.0629) acc1: 88.2812 (32.5222) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2392 data: 0.1432 max mem: 3531 +Test: [ 70/517] eta: 0:02:20 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2535 (0.9482) acc1: 96.6146 (41.7437) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2263 data: 0.1312 max mem: 3531 +Test: [ 80/517] eta: 0:02:13 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2591 (0.8654) acc1: 97.6562 (48.5757) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2441 data: 0.1494 max mem: 3531 +Test: [ 90/517] eta: 0:02:07 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2997 (0.8073) acc1: 96.3542 (53.6830) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2462 data: 0.1507 max mem: 3531 +Test: [100/517] eta: 0:02:03 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.3011 (0.7565) acc1: 96.3542 (57.8821) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2549 data: 0.1599 max mem: 3531 +Test: [110/517] eta: 0:01:58 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2755 (0.7127) acc1: 94.2708 (61.0407) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2522 data: 0.1589 max mem: 3531 +Test: [120/517] eta: 0:01:54 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2346 (0.6712) acc1: 96.0938 (64.2261) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2499 data: 0.1545 max mem: 3531 +Test: [130/517] eta: 0:01:50 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2075 (0.6352) acc1: 100.0000 (66.9450) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2549 data: 0.1562 max mem: 3531 +Test: [140/517] eta: 0:01:47 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2164 (0.6061) acc1: 98.9583 (69.1083) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2599 data: 0.1600 max mem: 3531 +Test: [150/517] eta: 0:01:42 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2268 (0.5824) acc1: 97.3958 (71.0127) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2499 data: 0.1529 max mem: 3531 +Test: [160/517] eta: 0:01:39 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2461 (0.5614) acc1: 98.4375 (72.7436) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2299 data: 0.1356 max mem: 3531 +Test: [170/517] eta: 0:01:36 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2409 (0.5422) acc1: 98.4375 (74.2325) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2649 data: 0.1690 max mem: 3531 +Test: [180/517] eta: 0:01:33 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2411 (0.5271) acc1: 97.9167 (75.3597) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.2699 data: 0.1748 max mem: 3531 +Test: [190/517] eta: 0:01:29 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2717 (0.5146) acc1: 94.2708 (76.4043) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.2349 data: 0.1386 max mem: 3531 +Test: [200/517] eta: 0:01:26 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2497 (0.5013) acc1: 96.0938 (77.4513) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2349 data: 0.1377 max mem: 3531 +Test: [210/517] eta: 0:01:23 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2470 (0.4901) acc1: 98.9583 (78.4632) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2549 data: 0.1565 max mem: 3532 +Test: [220/517] eta: 0:01:20 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2660 (0.4801) acc1: 98.4375 (79.3552) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2549 data: 0.1554 max mem: 3532 +Test: [230/517] eta: 0:01:17 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2888 (0.4758) acc1: 97.1354 (79.7777) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.2349 data: 0.1399 max mem: 3532 +Test: [240/517] eta: 0:01:14 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.3656 (0.4706) acc1: 91.9271 (80.3023) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.2449 data: 0.1513 max mem: 3532 +Test: [250/517] eta: 0:01:11 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.3054 (0.4632) acc1: 96.6146 (81.0083) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2699 data: 0.1739 max mem: 3532 +Test: [260/517] eta: 0:01:08 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2733 (0.4562) acc1: 98.4375 (81.6381) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.2549 data: 0.1591 max mem: 3532 +Test: [270/517] eta: 0:01:05 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2963 (0.4507) acc1: 93.4896 (81.9649) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.2349 data: 0.1422 max mem: 3532 +Test: [280/517] eta: 0:01:03 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2956 (0.4445) acc1: 91.4062 (82.4362) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2449 data: 0.1529 max mem: 3532 +Test: [290/517] eta: 0:01:00 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2930 (0.4420) acc1: 91.9271 (82.6720) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.2599 data: 0.1670 max mem: 3532 +Test: [300/517] eta: 0:00:57 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.3464 (0.4385) acc1: 91.6667 (83.0210) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2499 data: 0.1567 max mem: 3532 +Test: [310/517] eta: 0:00:54 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.3339 (0.4352) acc1: 95.3125 (83.4732) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2349 data: 0.1381 max mem: 3532 +Test: [320/517] eta: 0:00:51 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.3070 (0.4298) acc1: 98.1771 (83.9191) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2449 data: 0.1485 max mem: 3532 +Test: [330/517] eta: 0:00:49 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2359 (0.4237) acc1: 97.3958 (84.3105) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2899 data: 0.1976 max mem: 3532 +Test: [340/517] eta: 0:00:46 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2211 (0.4179) acc1: 97.3958 (84.7248) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2749 data: 0.1796 max mem: 3532 +Test: [350/517] eta: 0:00:44 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2543 (0.4177) acc1: 97.3958 (84.7437) acc5: 100.0000 (100.0000) acc1_neg: 56.250 acc5_neg: 100.000 time: 0.2499 data: 0.1525 max mem: 3532 +Test: [360/517] eta: 0:00:41 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.4114 (0.4150) acc1: 85.6771 (84.9413) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2599 data: 0.1632 max mem: 3532 +Test: [370/517] eta: 0:00:38 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.1627 (0.4076) acc1: 100.0000 (85.3472) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2399 data: 0.1423 max mem: 3532 +Test: [380/517] eta: 0:00:36 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.1344 (0.4005) acc1: 100.0000 (85.7311) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2449 data: 0.1448 max mem: 3532 +Test: [390/517] eta: 0:00:33 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.1502 (0.3942) acc1: 100.0000 (86.0927) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2499 data: 0.1495 max mem: 3532 +Test: [400/517] eta: 0:00:30 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.1562 (0.3886) acc1: 100.0000 (86.4285) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2450 data: 0.1477 max mem: 3532 +Test: [410/517] eta: 0:00:27 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2101 (0.3843) acc1: 99.4792 (86.6966) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2399 data: 0.1431 max mem: 3532 +Test: [420/517] eta: 0:00:25 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2399 (0.3812) acc1: 97.6562 (86.9359) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.2448 data: 0.1477 max mem: 3532 +Test: [430/517] eta: 0:00:22 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2541 (0.3786) acc1: 93.4896 (87.0752) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2449 data: 0.1481 max mem: 3532 +Test: [440/517] eta: 0:00:20 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2305 (0.3748) acc1: 97.9167 (87.3589) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2449 data: 0.1485 max mem: 3533 +Test: [450/517] eta: 0:00:17 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2159 (0.3714) acc1: 98.6979 (87.5860) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2449 data: 0.1478 max mem: 3533 +Test: [460/517] eta: 0:00:14 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2081 (0.3679) acc1: 98.1771 (87.8107) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2449 data: 0.1467 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2081 (0.3648) acc1: 98.4375 (88.0441) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2449 data: 0.1484 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2369 (0.3623) acc1: 99.2188 (88.2769) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2449 data: 0.1480 max mem: 3533 +Test: [490/517] eta: 0:00:06 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2302 (0.3595) acc1: 99.2188 (88.4839) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2471 data: 0.1488 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2425 (0.3577) acc1: 98.9583 (88.6929) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2414 data: 0.1466 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2732 (0.3561) acc1: 98.4375 (88.8265) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1956 data: 0.1034 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2617 (0.3549) acc1: 96.8750 (88.9113) acc5: 100.0000 (100.0000) acc1_neg: 91.519 acc5_neg: 100.000 time: 0.1376 data: 0.0449 max mem: 3533 +Test: Total time: 0:02:12 (0.2556 s / it) +* Overall Acc@1 88.911 Acc@5 100.000 + target == real Acc@1 16.987 Acc@5 100.000 + target == fake Acc@1 96.373 Acc@5 100.000 + loss 0.355 + PR-AUC 0.9529 ROC-AUC 0.7046 EER 35.11% (thr=0.7646) +Accuracy of the network on the 198427 test images: 88.9% +[rank0]:[W725 18:30:34.140039054 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0725 18:30:54.160000 605218 torch/distributed/run.py:766] +W0725 18:30:54.160000 605218 torch/distributed/run.py:766] ***************************************** +W0725 18:30:54.160000 605218 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:30:54.160000 605218 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:31:23.438130162 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:31:23.492702999 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:31:24.015820596 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:31:24.124234166 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:31:24.178400384 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:31:24.242080141 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:31:24.286271040 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:31:24.526231766 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed4_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n11_classseed4_resampled_n116-041-188_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:41:38 acc1_pos: 53.906 acc5_pos: 100.000 loss: 0.6630 (0.6630) acc1: 53.9062 (53.9062) acc5: 100.0000 (100.0000) time: 6.4241 data: 4.7920 max mem: 3531 +Test: [ 10/389] eta: 0:04:13 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2684 (0.3729) acc1: 98.6979 (86.5767) acc5: 100.0000 (100.0000) time: 0.6683 data: 0.4361 max mem: 3531 +Test: [ 20/389] eta: 0:02:26 acc1_pos: 83.854 acc5_pos: 100.000 loss: 0.2997 (0.3437) acc1: 99.2188 (91.2946) acc5: 100.0000 (100.0000) time: 0.0956 data: 0.0010 max mem: 3531 +Test: [ 30/389] eta: 0:01:58 acc1_pos: 97.396 acc5_pos: 100.000 loss: 0.3497 (0.4060) acc1: 93.7500 (86.4667) acc5: 100.0000 (100.0000) time: 0.1442 data: 0.0452 max mem: 3531 +Test: [ 40/389] eta: 0:01:42 acc1_pos: 77.865 acc5_pos: 100.000 loss: 0.4378 (0.4091) acc1: 89.3229 (87.2713) acc5: 100.0000 (100.0000) time: 0.1861 data: 0.0889 max mem: 3531 +Test: [ 50/389] eta: 0:01:31 acc1_pos: 89.844 acc5_pos: 100.000 loss: 0.4365 (0.4092) acc1: 89.8438 (88.1230) acc5: 100.0000 (100.0000) time: 0.1760 data: 0.0842 max mem: 3531 +Test: [ 60/389] eta: 0:01:24 acc1_pos: 91.406 acc5_pos: 100.000 loss: 0.4690 (0.4094) acc1: 89.8438 (88.8448) acc5: 100.0000 (100.0000) time: 0.1809 data: 0.0899 max mem: 3531 +Test: [ 70/389] eta: 0:01:17 acc1_pos: 72.917 acc5_pos: 100.000 loss: 0.4886 (0.4274) acc1: 86.9792 (87.0122) acc5: 100.0000 (100.0000) time: 0.1789 data: 0.0877 max mem: 3531 +Test: [ 80/389] eta: 0:01:11 acc1_pos: 98.958 acc5_pos: 100.000 loss: 0.4481 (0.4265) acc1: 85.6771 (87.4775) acc5: 100.0000 (100.0000) time: 0.1538 data: 0.0622 max mem: 3531 +Test: [ 90/389] eta: 0:01:07 acc1_pos: 84.896 acc5_pos: 100.000 loss: 0.2820 (0.4149) acc1: 95.3125 (88.0351) acc5: 100.0000 (100.0000) time: 0.1665 data: 0.0720 max mem: 3531 +Test: [100/389] eta: 0:01:04 acc1_pos: 60.156 acc5_pos: 100.000 loss: 0.4167 (0.4317) acc1: 88.5417 (86.4506) acc5: 100.0000 (100.0000) time: 0.1837 data: 0.0901 max mem: 3531 +Test: [110/389] eta: 0:01:00 acc1_pos: 93.229 acc5_pos: 100.000 loss: 0.4672 (0.4364) acc1: 83.8542 (85.9586) acc5: 100.0000 (100.0000) time: 0.1795 data: 0.0845 max mem: 3531 +Test: [120/389] eta: 0:00:57 acc1_pos: 96.615 acc5_pos: 100.000 loss: 0.3749 (0.4318) acc1: 88.2812 (86.2237) acc5: 100.0000 (100.0000) time: 0.1771 data: 0.0799 max mem: 3531 +Test: [130/389] eta: 0:00:54 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.5270 (0.4632) acc1: 70.3125 (82.9278) acc5: 100.0000 (100.0000) acc1_neg: 69.010 acc5_neg: 100.000 time: 0.1686 data: 0.0742 max mem: 3531 +Test: [140/389] eta: 0:00:51 acc1_pos: 67.485 acc5_pos: 100.000 loss: 1.0672 (0.5134) acc1: 13.2812 (77.9828) acc5: 100.0000 (100.0000) acc1_neg: 0.000 acc5_neg: 100.000 time: 0.1697 data: 0.0767 max mem: 3531 +Test: [150/389] eta: 0:00:49 acc1_pos: 67.485 acc5_pos: 100.000 loss: 1.0613 (0.5359) acc1: 13.8021 (75.4553) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.1762 data: 0.0844 max mem: 3531 +Test: [160/389] eta: 0:00:46 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.8569 (0.5539) acc1: 29.1667 (73.6251) acc5: 100.0000 (100.0000) acc1_neg: 36.979 acc5_neg: 100.000 time: 0.1558 data: 0.0641 max mem: 3531 +Test: [170/389] eta: 0:00:43 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.6193 (0.5663) acc1: 65.6250 (72.5923) acc5: 100.0000 (100.0000) acc1_neg: 17.969 acc5_neg: 100.000 time: 0.1536 data: 0.0623 max mem: 3531 +Test: [180/389] eta: 0:00:41 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.6046 (0.5744) acc1: 66.9271 (71.8203) acc5: 100.0000 (100.0000) acc1_neg: 59.896 acc5_neg: 100.000 time: 0.1671 data: 0.0757 max mem: 3531 +Test: [190/389] eta: 0:00:39 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.6819 (0.5985) acc1: 55.2083 (69.6281) acc5: 100.0000 (100.0000) acc1_neg: 0.521 acc5_neg: 100.000 time: 0.1606 data: 0.0678 max mem: 3531 +Test: [200/389] eta: 0:00:36 acc1_pos: 67.485 acc5_pos: 100.000 loss: 1.0508 (0.6283) acc1: 3.3854 (67.3857) acc5: 100.0000 (100.0000) acc1_neg: 0.000 acc5_neg: 100.000 time: 0.1558 data: 0.0640 max mem: 3531 +Test: [210/389] eta: 0:00:34 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.7424 (0.6423) acc1: 20.3125 (66.7296) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1494 data: 0.0592 max mem: 3532 +Test: [220/389] eta: 0:00:32 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.5230 (0.6421) acc1: 69.0104 (66.8328) acc5: 100.0000 (100.0000) acc1_neg: 15.885 acc5_neg: 100.000 time: 0.1489 data: 0.0576 max mem: 3532 +Test: [230/389] eta: 0:00:29 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.5582 (0.6529) acc1: 66.1458 (65.8606) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1496 data: 0.0562 max mem: 3532 +Test: [240/389] eta: 0:00:27 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.6216 (0.6531) acc1: 51.0417 (65.6239) acc5: 100.0000 (100.0000) acc1_neg: 42.708 acc5_neg: 100.000 time: 0.1400 data: 0.0467 max mem: 3532 +Test: [250/389] eta: 0:00:25 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.7459 (0.6571) acc1: 43.4896 (64.7296) acc5: 100.0000 (100.0000) acc1_neg: 44.792 acc5_neg: 100.000 time: 0.1400 data: 0.0485 max mem: 3532 +Test: [260/389] eta: 0:00:23 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.7703 (0.6679) acc1: 40.1042 (63.1675) acc5: 100.0000 (100.0000) acc1_neg: 3.646 acc5_neg: 100.000 time: 0.1548 data: 0.0613 max mem: 3532 +Test: [270/389] eta: 0:00:21 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.9047 (0.6853) acc1: 21.3542 (61.6794) acc5: 100.0000 (100.0000) acc1_neg: 16.406 acc5_neg: 100.000 time: 0.1451 data: 0.0491 max mem: 3532 +Test: [280/389] eta: 0:00:19 acc1_pos: 67.485 acc5_pos: 100.000 loss: 1.1722 (0.7011) acc1: 7.2917 (60.2211) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1349 data: 0.0414 max mem: 3532 +Test: [290/389] eta: 0:00:17 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.8525 (0.6968) acc1: 43.4896 (60.6511) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1396 data: 0.0472 max mem: 3532 +Test: [300/389] eta: 0:00:15 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.8525 (0.7103) acc1: 43.4896 (59.1475) acc5: 100.0000 (100.0000) acc1_neg: 5.729 acc5_neg: 100.000 time: 0.1456 data: 0.0539 max mem: 3532 +Test: [310/389] eta: 0:00:13 acc1_pos: 67.485 acc5_pos: 100.000 loss: 1.1968 (0.7203) acc1: 5.7292 (58.1960) acc5: 100.0000 (100.0000) acc1_neg: 1.562 acc5_neg: 100.000 time: 0.1499 data: 0.0564 max mem: 3532 +Test: [320/389] eta: 0:00:12 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.8996 (0.7240) acc1: 22.3958 (57.6592) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1592 data: 0.0627 max mem: 3532 +Test: [330/389] eta: 0:00:10 acc1_pos: 67.485 acc5_pos: 100.000 loss: 1.0529 (0.7335) acc1: 8.3333 (56.4050) acc5: 100.0000 (100.0000) acc1_neg: 28.906 acc5_neg: 100.000 time: 0.1613 data: 0.0674 max mem: 3532 +Test: [340/389] eta: 0:00:08 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.7187 (0.7301) acc1: 48.6979 (56.8136) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.1673 data: 0.0754 max mem: 3532 +Test: [350/389] eta: 0:00:06 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.6599 (0.7396) acc1: 61.1979 (56.3628) acc5: 100.0000 (100.0000) acc1_neg: 0.260 acc5_neg: 100.000 time: 0.1745 data: 0.0818 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 67.485 acc5_pos: 100.000 loss: 1.2429 (0.7549) acc1: 3.6458 (54.8909) acc5: 100.0000 (100.0000) acc1_neg: 4.688 acc5_neg: 100.000 time: 0.1775 data: 0.0831 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 67.485 acc5_pos: 100.000 loss: 1.2323 (0.7557) acc1: 4.6875 (54.6938) acc5: 100.0000 (100.0000) acc1_neg: 3.646 acc5_neg: 100.000 time: 0.1694 data: 0.0748 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 67.485 acc5_pos: 100.000 loss: 1.1030 (0.7668) acc1: 4.4271 (53.7613) acc5: 100.0000 (100.0000) acc1_neg: 52.865 acc5_neg: 100.000 time: 0.1450 data: 0.0528 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.7841 (0.7651) acc1: 39.0625 (53.9255) acc5: 100.0000 (100.0000) acc1_neg: 55.738 acc5_neg: 100.000 time: 0.1452 data: 0.0513 max mem: 3532 +Test: Total time: 0:01:07 (0.1741 s / it) +* Overall Acc@1 53.925 Acc@5 100.000 + target == real Acc@1 86.011 Acc@5 100.000 + target == fake Acc@1 39.011 Acc@5 100.000 + loss 0.765 + PR-AUC 0.8158 ROC-AUC 0.6692 EER 39.44% (thr=0.3511) +Accuracy of the network on the 149358 test images: 53.9% +[rank0]:[W725 18:32:48.820252740 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:33:08.499000 605787 torch/distributed/run.py:766] +W0725 18:33:08.499000 605787 torch/distributed/run.py:766] ***************************************** +W0725 18:33:08.499000 605787 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:33:08.499000 605787 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:33:37.102629359 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:33:37.376165971 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:33:37.716492622 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:33:37.864311034 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:33:37.888369229 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:33:38.083108475 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:33:38.149423499 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W725 18:33:38.149460802 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed4_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n11_classseed4_resampled_n116-041-188_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:30:57 acc1_pos: 95.573 acc5_pos: 100.000 loss: 0.2344 (0.2344) acc1: 95.5729 (95.5729) acc5: 100.0000 (100.0000) time: 5.5777 data: 3.7734 max mem: 3531 +Test: [ 10/333] eta: 0:03:10 acc1_pos: 18.229 acc5_pos: 100.000 loss: 0.2344 (0.6427) acc1: 92.1875 (66.3352) acc5: 100.0000 (100.0000) time: 0.5890 data: 0.3434 max mem: 3531 +Test: [ 20/333] eta: 0:01:52 acc1_pos: 21.875 acc5_pos: 100.000 loss: 1.3102 (1.0255) acc1: 18.4896 (41.9643) acc5: 100.0000 (100.0000) time: 0.0983 data: 0.0081 max mem: 3531 +Test: [ 30/333] eta: 0:01:27 acc1_pos: 54.688 acc5_pos: 100.000 loss: 1.2060 (0.9654) acc1: 22.1354 (43.9180) acc5: 100.0000 (100.0000) time: 0.1243 data: 0.0306 max mem: 3531 +Test: [ 40/333] eta: 0:01:27 acc1_pos: 50.000 acc5_pos: 100.000 loss: 0.7761 (0.9281) acc1: 50.0000 (45.2744) acc5: 100.0000 (100.0000) time: 0.2349 data: 0.1375 max mem: 3531 +Test: [ 50/333] eta: 0:01:18 acc1_pos: 51.302 acc5_pos: 100.000 loss: 0.7761 (0.8970) acc1: 51.3021 (46.9312) acc5: 100.0000 (100.0000) time: 0.2589 data: 0.1629 max mem: 3531 +Test: [ 60/333] eta: 0:01:11 acc1_pos: 48.958 acc5_pos: 100.000 loss: 0.7951 (0.8823) acc1: 51.3021 (47.4257) acc5: 100.0000 (100.0000) time: 0.1876 data: 0.0930 max mem: 3531 +Test: [ 70/333] eta: 0:01:04 acc1_pos: 34.635 acc5_pos: 100.000 loss: 0.7951 (0.8679) acc1: 48.9583 (48.1001) acc5: 100.0000 (100.0000) time: 0.1652 data: 0.0707 max mem: 3531 +Test: [ 80/333] eta: 0:00:59 acc1_pos: 58.854 acc5_pos: 100.000 loss: 0.7752 (0.8549) acc1: 52.0833 (48.8747) acc5: 100.0000 (100.0000) time: 0.1601 data: 0.0671 max mem: 3531 +Test: [ 90/333] eta: 0:00:56 acc1_pos: 58.073 acc5_pos: 100.000 loss: 0.7701 (0.8471) acc1: 54.4271 (49.2617) acc5: 100.0000 (100.0000) time: 0.1869 data: 0.0955 max mem: 3531 +Test: [100/333] eta: 0:00:53 acc1_pos: 52.865 acc5_pos: 100.000 loss: 0.7651 (0.8404) acc1: 52.8646 (49.6416) acc5: 100.0000 (100.0000) time: 0.1970 data: 0.1057 max mem: 3531 +Test: [110/333] eta: 0:00:49 acc1_pos: 63.281 acc5_pos: 100.000 loss: 0.8017 (0.8398) acc1: 48.9583 (49.3994) acc5: 100.0000 (100.0000) time: 0.1787 data: 0.0872 max mem: 3531 +Test: [120/333] eta: 0:00:46 acc1_pos: 45.833 acc5_pos: 100.000 loss: 0.8307 (0.8351) acc1: 48.4375 (49.5825) acc5: 100.0000 (100.0000) time: 0.1676 data: 0.0758 max mem: 3531 +Test: [130/333] eta: 0:00:43 acc1_pos: 61.979 acc5_pos: 100.000 loss: 0.7912 (0.8319) acc1: 50.2604 (49.7356) acc5: 100.0000 (100.0000) time: 0.1707 data: 0.0787 max mem: 3531 +Test: [140/333] eta: 0:00:40 acc1_pos: 62.500 acc5_pos: 100.000 loss: 0.7926 (0.8284) acc1: 50.5208 (49.8098) acc5: 100.0000 (100.0000) time: 0.1680 data: 0.0756 max mem: 3531 +Test: [150/333] eta: 0:00:38 acc1_pos: 41.667 acc5_pos: 100.000 loss: 0.7741 (0.8283) acc1: 50.0000 (49.7310) acc5: 100.0000 (100.0000) time: 0.1854 data: 0.0925 max mem: 3531 +Test: [160/333] eta: 0:00:36 acc1_pos: 49.013 acc5_pos: 100.000 loss: 0.7735 (0.8257) acc1: 52.3438 (49.9499) acc5: 100.0000 (100.0000) acc1_neg: 52.344 acc5_neg: 100.000 time: 0.1999 data: 0.1079 max mem: 3531 +Test: [170/333] eta: 0:00:33 acc1_pos: 49.013 acc5_pos: 100.000 loss: 0.7290 (0.8166) acc1: 55.9896 (50.7082) acc5: 100.0000 (100.0000) acc1_neg: 69.010 acc5_neg: 100.000 time: 0.1879 data: 0.0966 max mem: 3531 +Test: [180/333] eta: 0:00:31 acc1_pos: 49.013 acc5_pos: 100.000 loss: 0.6039 (0.8032) acc1: 66.9271 (51.8258) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.1791 data: 0.0876 max mem: 3531 +Test: [190/333] eta: 0:00:29 acc1_pos: 49.013 acc5_pos: 100.000 loss: 0.5581 (0.7888) acc1: 72.1354 (53.1795) acc5: 100.0000 (100.0000) acc1_neg: 72.656 acc5_neg: 100.000 time: 0.1855 data: 0.0914 max mem: 3531 +Test: [200/333] eta: 0:00:27 acc1_pos: 49.013 acc5_pos: 100.000 loss: 0.5581 (0.7776) acc1: 72.6562 (54.1524) acc5: 100.0000 (100.0000) acc1_neg: 70.833 acc5_neg: 100.000 time: 0.1914 data: 0.0969 max mem: 3531 +Test: [210/333] eta: 0:00:25 acc1_pos: 49.013 acc5_pos: 100.000 loss: 0.5864 (0.7694) acc1: 69.2708 (54.8196) acc5: 100.0000 (100.0000) acc1_neg: 57.552 acc5_neg: 100.000 time: 0.1812 data: 0.0887 max mem: 3532 +Test: [220/333] eta: 0:00:22 acc1_pos: 49.013 acc5_pos: 100.000 loss: 0.5952 (0.7634) acc1: 69.2708 (55.2932) acc5: 100.0000 (100.0000) acc1_neg: 70.052 acc5_neg: 100.000 time: 0.1704 data: 0.0742 max mem: 3532 +Test: [230/333] eta: 0:00:20 acc1_pos: 49.013 acc5_pos: 100.000 loss: 0.5427 (0.7536) acc1: 70.3125 (56.0572) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.1746 data: 0.0771 max mem: 3532 +Test: [240/333] eta: 0:00:18 acc1_pos: 49.013 acc5_pos: 100.000 loss: 0.5691 (0.7505) acc1: 68.2292 (56.1787) acc5: 100.0000 (100.0000) acc1_neg: 61.198 acc5_neg: 100.000 time: 0.1798 data: 0.0871 max mem: 3532 +Test: [250/333] eta: 0:00:16 acc1_pos: 49.013 acc5_pos: 100.000 loss: 0.6731 (0.7445) acc1: 60.9375 (56.6567) acc5: 100.0000 (100.0000) acc1_neg: 75.781 acc5_neg: 100.000 time: 0.1778 data: 0.0860 max mem: 3532 +Test: [260/333] eta: 0:00:14 acc1_pos: 49.013 acc5_pos: 100.000 loss: 0.5570 (0.7370) acc1: 70.8333 (57.3465) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.1654 data: 0.0725 max mem: 3532 +Test: [270/333] eta: 0:00:12 acc1_pos: 49.013 acc5_pos: 100.000 loss: 0.5179 (0.7288) acc1: 76.5625 (58.0614) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.1666 data: 0.0737 max mem: 3532 +Test: [280/333] eta: 0:00:10 acc1_pos: 49.013 acc5_pos: 100.000 loss: 0.5369 (0.7269) acc1: 71.8750 (58.1851) acc5: 100.0000 (100.0000) acc1_neg: 54.427 acc5_neg: 100.000 time: 0.1689 data: 0.0767 max mem: 3532 +Test: [290/333] eta: 0:00:08 acc1_pos: 49.013 acc5_pos: 100.000 loss: 0.7649 (0.7280) acc1: 54.4271 (58.0514) acc5: 100.0000 (100.0000) acc1_neg: 61.719 acc5_neg: 100.000 time: 0.1642 data: 0.0718 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 49.013 acc5_pos: 100.000 loss: 0.6820 (0.7253) acc1: 61.7188 (58.2529) acc5: 100.0000 (100.0000) acc1_neg: 61.719 acc5_neg: 100.000 time: 0.1866 data: 0.0922 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 49.013 acc5_pos: 100.000 loss: 0.6464 (0.7206) acc1: 63.8021 (58.6725) acc5: 100.0000 (100.0000) acc1_neg: 61.198 acc5_neg: 100.000 time: 0.1991 data: 0.1037 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 49.013 acc5_pos: 100.000 loss: 0.6540 (0.7186) acc1: 63.2812 (58.8201) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.1945 data: 0.1011 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 49.013 acc5_pos: 100.000 loss: 0.6667 (0.7177) acc1: 62.5000 (58.9281) acc5: 100.0000 (100.0000) acc1_neg: 60.938 acc5_neg: 100.000 time: 0.1860 data: 0.0951 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 49.013 acc5_pos: 100.000 loss: 0.6626 (0.7173) acc1: 63.2812 (58.9575) acc5: 100.0000 (100.0000) acc1_neg: 63.944 acc5_neg: 100.000 time: 0.1753 data: 0.0813 max mem: 3532 +Test: Total time: 0:01:04 (0.1945 s / it) +* Overall Acc@1 58.957 Acc@5 100.000 + target == real Acc@1 49.844 Acc@5 100.000 + target == fake Acc@1 67.070 Acc@5 100.000 + loss 0.717 + PR-AUC 0.6205 ROC-AUC 0.6135 EER 41.44% (thr=0.5610) +Accuracy of the network on the 127843 test images: 59.0% +[rank0]:[W725 18:34:53.870458036 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:35:13.220000 606355 torch/distributed/run.py:766] +W0725 18:35:13.220000 606355 torch/distributed/run.py:766] ***************************************** +W0725 18:35:13.220000 606355 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:35:13.220000 606355 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:35:41.494688164 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:35:41.762062517 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:35:42.022029217 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:35:42.186758305 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:35:42.243667059 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:35:42.496984640 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:35:42.506417787 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:35:42.534493975 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed4_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n11_classseed4_resampled_n116-041-188_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:18:55 acc1_pos: 85.938 acc5_pos: 100.000 loss: 0.3501 (0.3501) acc1: 85.9375 (85.9375) acc5: 100.0000 (100.0000) time: 4.8327 data: 2.7761 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/235] eta: 0:02:12 acc1_pos: 88.281 acc5_pos: 100.000 loss: 0.3635 (0.3630) acc1: 86.1979 (86.3400) acc5: 100.0000 (100.0000) time: 0.5886 data: 0.3141 max mem: 3531 +Test: [ 20/235] eta: 0:01:17 acc1_pos: 84.896 acc5_pos: 100.000 loss: 0.3576 (0.3600) acc1: 86.1979 (86.3591) acc5: 100.0000 (100.0000) time: 0.1353 data: 0.0407 max mem: 3531 +Test: [ 30/235] eta: 0:01:04 acc1_pos: 89.844 acc5_pos: 100.000 loss: 0.3524 (0.3556) acc1: 87.2396 (86.7944) acc5: 100.0000 (100.0000) time: 0.1671 data: 0.0724 max mem: 3531 +Test: [ 40/235] eta: 0:00:59 acc1_pos: 86.979 acc5_pos: 100.000 loss: 0.3513 (0.3557) acc1: 86.9792 (86.7315) acc5: 100.0000 (100.0000) time: 0.2441 data: 0.1486 max mem: 3531 +Test: [ 50/235] eta: 0:00:51 acc1_pos: 84.896 acc5_pos: 100.000 loss: 0.3525 (0.3561) acc1: 86.7188 (86.7494) acc5: 100.0000 (100.0000) time: 0.2149 data: 0.1205 max mem: 3531 +Test: [ 60/235] eta: 0:00:45 acc1_pos: 89.323 acc5_pos: 100.000 loss: 0.3592 (0.3569) acc1: 86.9792 (86.8084) acc5: 100.0000 (100.0000) time: 0.1746 data: 0.0810 max mem: 3531 +Test: [ 70/235] eta: 0:00:41 acc1_pos: 86.154 acc5_pos: 100.000 loss: 0.3797 (0.4464) acc1: 84.1146 (80.5531) acc5: 100.0000 (100.0000) acc1_neg: 33.333 acc5_neg: 100.000 time: 0.1957 data: 0.0992 max mem: 3531 +Test: [ 80/235] eta: 0:00:38 acc1_pos: 86.154 acc5_pos: 100.000 loss: 1.0748 (0.5239) acc1: 36.4583 (75.0418) acc5: 100.0000 (100.0000) acc1_neg: 34.375 acc5_neg: 100.000 time: 0.2169 data: 0.1212 max mem: 3531 +Test: [ 90/235] eta: 0:00:35 acc1_pos: 86.154 acc5_pos: 100.000 loss: 1.0846 (0.5863) acc1: 36.4583 (70.7189) acc5: 100.0000 (100.0000) acc1_neg: 32.292 acc5_neg: 100.000 time: 0.2257 data: 0.1339 max mem: 3531 +Test: [100/235] eta: 0:00:33 acc1_pos: 86.154 acc5_pos: 100.000 loss: 1.0891 (0.6353) acc1: 36.4583 (67.2623) acc5: 100.0000 (100.0000) acc1_neg: 28.646 acc5_neg: 100.000 time: 0.2279 data: 0.1367 max mem: 3531 +Test: [110/235] eta: 0:00:30 acc1_pos: 86.154 acc5_pos: 100.000 loss: 1.0777 (0.6737) acc1: 36.1979 (64.5223) acc5: 100.0000 (100.0000) acc1_neg: 35.677 acc5_neg: 100.000 time: 0.2134 data: 0.1212 max mem: 3531 +Test: [120/235] eta: 0:00:27 acc1_pos: 86.154 acc5_pos: 100.000 loss: 1.1153 (0.7158) acc1: 31.2500 (61.4928) acc5: 100.0000 (100.0000) acc1_neg: 27.865 acc5_neg: 100.000 time: 0.2159 data: 0.1234 max mem: 3531 +Test: [130/235] eta: 0:00:24 acc1_pos: 86.154 acc5_pos: 100.000 loss: 1.1856 (0.7519) acc1: 27.6042 (58.8939) acc5: 100.0000 (100.0000) acc1_neg: 29.948 acc5_neg: 100.000 time: 0.2158 data: 0.1245 max mem: 3531 +Test: [140/235] eta: 0:00:22 acc1_pos: 86.154 acc5_pos: 100.000 loss: 1.2110 (0.7855) acc1: 26.5625 (56.6138) acc5: 100.0000 (100.0000) acc1_neg: 22.917 acc5_neg: 100.000 time: 0.2089 data: 0.1165 max mem: 3531 +Test: [150/235] eta: 0:00:20 acc1_pos: 86.154 acc5_pos: 100.000 loss: 1.2110 (0.8132) acc1: 26.8229 (54.6616) acc5: 100.0000 (100.0000) acc1_neg: 27.604 acc5_neg: 100.000 time: 0.2306 data: 0.1380 max mem: 3531 +Test: [160/235] eta: 0:00:17 acc1_pos: 86.154 acc5_pos: 100.000 loss: 1.1950 (0.8366) acc1: 28.9062 (53.0813) acc5: 100.0000 (100.0000) acc1_neg: 28.906 acc5_neg: 100.000 time: 0.2402 data: 0.1487 max mem: 3531 +Test: [170/235] eta: 0:00:15 acc1_pos: 86.154 acc5_pos: 100.000 loss: 1.1568 (0.8437) acc1: 30.7292 (52.5082) acc5: 100.0000 (100.0000) acc1_neg: 41.146 acc5_neg: 100.000 time: 0.2053 data: 0.1128 max mem: 3531 +Test: [180/235] eta: 0:00:12 acc1_pos: 86.154 acc5_pos: 100.000 loss: 0.9204 (0.8468) acc1: 46.3542 (52.1927) acc5: 100.0000 (100.0000) acc1_neg: 43.490 acc5_neg: 100.000 time: 0.1899 data: 0.0977 max mem: 3531 +Test: [190/235] eta: 0:00:10 acc1_pos: 86.154 acc5_pos: 100.000 loss: 0.8832 (0.8495) acc1: 47.3958 (51.9374) acc5: 100.0000 (100.0000) acc1_neg: 51.042 acc5_neg: 100.000 time: 0.1982 data: 0.1060 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 86.154 acc5_pos: 100.000 loss: 0.9155 (0.8532) acc1: 46.0938 (51.6325) acc5: 100.0000 (100.0000) acc1_neg: 47.135 acc5_neg: 100.000 time: 0.1658 data: 0.0734 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 86.154 acc5_pos: 100.000 loss: 0.9216 (0.8559) acc1: 46.0938 (51.4304) acc5: 100.0000 (100.0000) acc1_neg: 45.573 acc5_neg: 100.000 time: 0.1479 data: 0.0574 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 86.154 acc5_pos: 100.000 loss: 0.9080 (0.8583) acc1: 46.8750 (51.1972) acc5: 100.0000 (100.0000) acc1_neg: 45.052 acc5_neg: 100.000 time: 0.1328 data: 0.0429 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 86.154 acc5_pos: 100.000 loss: 0.9073 (0.8603) acc1: 46.6146 (51.0248) acc5: 100.0000 (100.0000) acc1_neg: 50.260 acc5_neg: 100.000 time: 0.1127 data: 0.0226 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 86.154 acc5_pos: 100.000 loss: 0.9073 (0.8614) acc1: 46.6146 (50.9763) acc5: 100.0000 (100.0000) acc1_neg: 47.826 acc5_neg: 100.000 time: 0.1121 data: 0.0255 max mem: 3532 +Test: Total time: 0:00:49 (0.2108 s / it) +* Overall Acc@1 50.976 Acc@5 100.000 + target == real Acc@1 86.793 Acc@5 100.000 + target == fake Acc@1 38.022 Acc@5 100.000 + loss 0.861 + PR-AUC 0.8638 ROC-AUC 0.7083 EER 34.89% (thr=0.2861) +Accuracy of the network on the 89879 test images: 51.0% +[rank0]:[W725 18:36:40.904784645 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:37:00.388000 606911 torch/distributed/run.py:766] +W0725 18:37:00.388000 606911 torch/distributed/run.py:766] ***************************************** +W0725 18:37:00.388000 606911 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:37:00.388000 606911 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W725 18:37:31.027694529 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +NCCL version 2.26.2+cuda12.2 +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:37:31.647965500 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:37:32.995243091 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:37:32.195899867 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:37:32.295666554 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:37:32.327360839 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W725 18:37:32.327435694 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:37:32.451473577 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed4_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n11_classseed4_resampled_n116-041-188_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:22:54 acc1_pos: 30.990 acc5_pos: 100.000 loss: 0.9535 (0.9535) acc1: 30.9896 (30.9896) acc5: 100.0000 (100.0000) time: 5.1083 data: 3.3780 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/269] eta: 0:02:21 acc1_pos: 48.698 acc5_pos: 100.000 loss: 0.9551 (0.9181) acc1: 30.9896 (36.2216) acc5: 100.0000 (100.0000) time: 0.5462 data: 0.3074 max mem: 3531 +Test: [ 20/269] eta: 0:01:22 acc1_pos: 52.604 acc5_pos: 100.000 loss: 0.8588 (0.8616) acc1: 44.5312 (41.5179) acc5: 100.0000 (100.0000) time: 0.0918 data: 0.0021 max mem: 3531 +Test: [ 30/269] eta: 0:01:00 acc1_pos: 26.042 acc5_pos: 100.000 loss: 0.7485 (0.8258) acc1: 49.2188 (45.3797) acc5: 100.0000 (100.0000) time: 0.0933 data: 0.0021 max mem: 3531 +Test: [ 40/269] eta: 0:00:50 acc1_pos: 15.517 acc5_pos: 100.000 loss: 0.5732 (0.7470) acc1: 71.0938 (53.1822) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.1069 data: 0.0141 max mem: 3531 +Test: [ 50/269] eta: 0:00:46 acc1_pos: 15.517 acc5_pos: 100.000 loss: 0.4000 (0.6868) acc1: 85.4167 (59.0278) acc5: 100.0000 (100.0000) acc1_neg: 72.135 acc5_neg: 100.000 time: 0.1469 data: 0.0541 max mem: 3531 +Test: [ 60/269] eta: 0:00:42 acc1_pos: 15.517 acc5_pos: 100.000 loss: 0.4510 (0.6505) acc1: 82.8125 (62.8629) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.1624 data: 0.0707 max mem: 3531 +Test: [ 70/269] eta: 0:00:38 acc1_pos: 15.517 acc5_pos: 100.000 loss: 0.4581 (0.6226) acc1: 84.6354 (66.0798) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1500 data: 0.0595 max mem: 3531 +Test: [ 80/269] eta: 0:00:36 acc1_pos: 15.517 acc5_pos: 100.000 loss: 0.4111 (0.5936) acc1: 89.5833 (69.3384) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1571 data: 0.0664 max mem: 3531 +Test: [ 90/269] eta: 0:00:33 acc1_pos: 15.517 acc5_pos: 100.000 loss: 0.4111 (0.5786) acc1: 89.8438 (71.0480) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.1729 data: 0.0815 max mem: 3531 +Test: [100/269] eta: 0:00:32 acc1_pos: 15.517 acc5_pos: 100.000 loss: 0.5171 (0.5742) acc1: 78.9062 (71.4109) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.2009 data: 0.1080 max mem: 3531 +Test: [110/269] eta: 0:00:30 acc1_pos: 15.517 acc5_pos: 100.000 loss: 0.4941 (0.5646) acc1: 77.0833 (72.3489) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.2159 data: 0.1223 max mem: 3531 +Test: [120/269] eta: 0:00:28 acc1_pos: 15.517 acc5_pos: 100.000 loss: 0.4895 (0.5632) acc1: 77.0833 (72.1419) acc5: 100.0000 (100.0000) acc1_neg: 56.771 acc5_neg: 100.000 time: 0.1948 data: 0.1023 max mem: 3531 +Test: [130/269] eta: 0:00:27 acc1_pos: 15.517 acc5_pos: 100.000 loss: 0.6048 (0.5720) acc1: 63.2812 (71.1474) acc5: 100.0000 (100.0000) acc1_neg: 49.479 acc5_neg: 100.000 time: 0.1937 data: 0.0991 max mem: 3531 +Test: [140/269] eta: 0:00:24 acc1_pos: 15.517 acc5_pos: 100.000 loss: 0.6993 (0.5833) acc1: 58.0729 (69.9782) acc5: 100.0000 (100.0000) acc1_neg: 58.854 acc5_neg: 100.000 time: 0.1852 data: 0.0882 max mem: 3531 +Test: [150/269] eta: 0:00:23 acc1_pos: 15.517 acc5_pos: 100.000 loss: 0.7093 (0.5913) acc1: 54.9479 (69.1208) acc5: 100.0000 (100.0000) acc1_neg: 53.385 acc5_neg: 100.000 time: 0.1951 data: 0.1005 max mem: 3531 +Test: [160/269] eta: 0:00:21 acc1_pos: 15.517 acc5_pos: 100.000 loss: 0.6326 (0.5905) acc1: 59.1146 (69.0460) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.2337 data: 0.1375 max mem: 3531 +Test: [170/269] eta: 0:00:19 acc1_pos: 15.517 acc5_pos: 100.000 loss: 0.5436 (0.5872) acc1: 69.5312 (69.3211) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.2198 data: 0.1208 max mem: 3531 +Test: [180/269] eta: 0:00:17 acc1_pos: 15.517 acc5_pos: 100.000 loss: 0.4654 (0.5780) acc1: 81.2500 (70.3744) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.1755 data: 0.0782 max mem: 3531 +Test: [190/269] eta: 0:00:15 acc1_pos: 15.517 acc5_pos: 100.000 loss: 0.4552 (0.5776) acc1: 81.5104 (70.3439) acc5: 100.0000 (100.0000) acc1_neg: 52.604 acc5_neg: 100.000 time: 0.1599 data: 0.0650 max mem: 3531 +Test: [200/269] eta: 0:00:13 acc1_pos: 15.517 acc5_pos: 100.000 loss: 0.6196 (0.5840) acc1: 61.1979 (69.6401) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1892 data: 0.0934 max mem: 3531 +Test: [210/269] eta: 0:00:11 acc1_pos: 15.517 acc5_pos: 100.000 loss: 0.7320 (0.5898) acc1: 51.0417 (68.8734) acc5: 100.0000 (100.0000) acc1_neg: 50.260 acc5_neg: 100.000 time: 0.2210 data: 0.1255 max mem: 3532 +Test: [220/269] eta: 0:00:09 acc1_pos: 15.517 acc5_pos: 100.000 loss: 0.7028 (0.5959) acc1: 50.7812 (68.1467) acc5: 100.0000 (100.0000) acc1_neg: 71.615 acc5_neg: 100.000 time: 0.2049 data: 0.1117 max mem: 3532 +Test: [230/269] eta: 0:00:07 acc1_pos: 15.517 acc5_pos: 100.000 loss: 0.6380 (0.5951) acc1: 59.8958 (68.0713) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.1895 data: 0.0965 max mem: 3532 +Test: [240/269] eta: 0:00:05 acc1_pos: 15.517 acc5_pos: 100.000 loss: 0.4586 (0.5911) acc1: 75.2604 (68.4096) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.2108 data: 0.1193 max mem: 3532 +Test: [250/269] eta: 0:00:03 acc1_pos: 15.517 acc5_pos: 100.000 loss: 0.5293 (0.5922) acc1: 67.1875 (68.1742) acc5: 100.0000 (100.0000) acc1_neg: 58.333 acc5_neg: 100.000 time: 0.1998 data: 0.1084 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 15.517 acc5_pos: 100.000 loss: 0.6184 (0.5933) acc1: 60.6771 (68.0546) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.1739 data: 0.0816 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 15.517 acc5_pos: 100.000 loss: 0.4871 (0.5889) acc1: 84.6354 (68.5913) acc5: 100.0000 (100.0000) acc1_neg: 89.764 acc5_neg: 100.000 time: 0.1720 data: 0.0770 max mem: 3532 +Test: Total time: 0:00:52 (0.1946 s / it) +* Overall Acc@1 68.591 Acc@5 100.000 + target == real Acc@1 44.373 Acc@5 100.000 + target == fake Acc@1 71.896 Acc@5 100.000 + loss 0.589 + PR-AUC 0.9193 ROC-AUC 0.6226 EER 41.19% (thr=0.5771) +Accuracy of the network on the 103293 test images: 68.6% +[rank0]:[W725 18:38:37.213059189 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-25T18:38:43Z] eval_done tag=fixed500k_real_n11_seed4_resampled +[2026-07-25T18:38:44Z] eval_start tag=fixed500k_fake_n22_seed4_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed4_resampled_n116-041-188_20260725_112559/best_checkpoint.pth +started_utc=2026-07-25T18:38:44Z +hostname=n116-041-188 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed4_resampled_n116-041-188_20260725_112559/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n22_seed4_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0725 18:38:57.552000 607499 torch/distributed/run.py:766] +W0725 18:38:57.552000 607499 torch/distributed/run.py:766] ***************************************** +W0725 18:38:57.552000 607499 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:38:57.552000 607499 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:39:27.539629890 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:39:27.693417158 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank1]:[W725 18:39:27.702179115 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:39:27.834941248 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:39:28.126191842 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:39:28.277555012 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:39:28.343131113 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:39:28.436442824 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n22_seed4_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed4_resampled_n116-041-188_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:18:24 acc1_pos: 91.146 acc5_pos: 100.000 loss: 0.3418 (0.3418) acc1: 91.1458 (91.1458) acc5: 100.0000 (100.0000) time: 4.7222 data: 2.5072 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/234] eta: 0:01:55 acc1_pos: 90.885 acc5_pos: 100.000 loss: 0.3442 (0.3603) acc1: 91.4062 (89.3229) acc5: 100.0000 (100.0000) time: 0.5158 data: 0.2328 max mem: 3531 +Test: [ 20/234] eta: 0:01:10 acc1_pos: 95.052 acc5_pos: 100.000 loss: 0.3698 (0.4330) acc1: 86.4583 (82.7505) acc5: 100.0000 (100.0000) time: 0.1104 data: 0.0206 max mem: 3531 +Test: [ 30/234] eta: 0:00:55 acc1_pos: 94.010 acc5_pos: 100.000 loss: 0.2038 (0.3514) acc1: 95.0521 (87.2984) acc5: 100.0000 (100.0000) time: 0.1356 data: 0.0456 max mem: 3531 +Test: [ 40/234] eta: 0:00:52 acc1_pos: 99.219 acc5_pos: 100.000 loss: 0.1852 (0.3133) acc1: 97.1354 (89.5897) acc5: 100.0000 (100.0000) time: 0.2060 data: 0.1108 max mem: 3531 +Test: [ 50/234] eta: 0:00:47 acc1_pos: 99.740 acc5_pos: 100.000 loss: 0.1316 (0.2764) acc1: 98.4375 (91.3143) acc5: 100.0000 (100.0000) time: 0.2425 data: 0.1461 max mem: 3531 +Test: [ 60/234] eta: 0:00:44 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.1165 (0.2490) acc1: 98.9583 (92.6785) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2241 data: 0.1301 max mem: 3531 +Test: [ 70/234] eta: 0:00:43 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.1132 (0.2649) acc1: 99.4792 (91.4026) acc5: 100.0000 (100.0000) acc1_neg: 75.521 acc5_neg: 100.000 time: 0.2703 data: 0.1765 max mem: 3531 +Test: [ 80/234] eta: 0:00:38 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.4736 (0.2856) acc1: 76.0417 (90.0141) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.2312 data: 0.1373 max mem: 3531 +Test: [ 90/234] eta: 0:00:34 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.3590 (0.2878) acc1: 85.6771 (89.7121) acc5: 100.0000 (100.0000) acc1_neg: 74.219 acc5_neg: 100.000 time: 0.1561 data: 0.0624 max mem: 3531 +Test: [100/234] eta: 0:00:31 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.4423 (0.3275) acc1: 78.1250 (87.2112) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.1638 data: 0.0722 max mem: 3531 +Test: [110/234] eta: 0:00:28 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.7335 (0.3669) acc1: 59.8958 (84.6941) acc5: 100.0000 (100.0000) acc1_neg: 45.052 acc5_neg: 100.000 time: 0.1833 data: 0.0859 max mem: 3531 +Test: [120/234] eta: 0:00:25 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.8425 (0.4098) acc1: 52.0833 (81.8010) acc5: 100.0000 (100.0000) acc1_neg: 43.229 acc5_neg: 100.000 time: 0.1955 data: 0.0968 max mem: 3531 +Test: [130/234] eta: 0:00:23 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.7443 (0.4316) acc1: 55.9896 (80.0950) acc5: 100.0000 (100.0000) acc1_neg: 72.917 acc5_neg: 100.000 time: 0.1899 data: 0.0972 max mem: 3531 +Test: [140/234] eta: 0:00:20 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.6992 (0.4554) acc1: 59.8958 (79.0577) acc5: 100.0000 (100.0000) acc1_neg: 41.146 acc5_neg: 100.000 time: 0.1890 data: 0.0950 max mem: 3531 +Test: [150/234] eta: 0:00:18 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.7068 (0.4771) acc1: 61.7188 (77.6318) acc5: 100.0000 (100.0000) acc1_neg: 66.146 acc5_neg: 100.000 time: 0.1812 data: 0.0868 max mem: 3531 +Test: [160/234] eta: 0:00:16 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.7068 (0.4898) acc1: 61.1979 (76.6515) acc5: 100.0000 (100.0000) acc1_neg: 53.906 acc5_neg: 100.000 time: 0.2321 data: 0.1382 max mem: 3531 +Test: [170/234] eta: 0:00:14 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.8102 (0.5084) acc1: 52.3438 (75.2163) acc5: 100.0000 (100.0000) acc1_neg: 48.698 acc5_neg: 100.000 time: 0.2297 data: 0.1356 max mem: 3531 +Test: [180/234] eta: 0:00:11 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.7986 (0.5199) acc1: 52.3438 (74.5310) acc5: 100.0000 (100.0000) acc1_neg: 39.844 acc5_neg: 100.000 time: 0.1397 data: 0.0464 max mem: 3531 +Test: [190/234] eta: 0:00:09 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.8175 (0.5377) acc1: 54.6875 (73.3693) acc5: 100.0000 (100.0000) acc1_neg: 54.688 acc5_neg: 100.000 time: 0.1488 data: 0.0535 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.8175 (0.5297) acc1: 54.9479 (73.8599) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1904 data: 0.0968 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.1138 (0.5103) acc1: 99.7396 (75.0753) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1820 data: 0.0909 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.1307 (0.4946) acc1: 99.2188 (76.0688) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.1690 data: 0.0784 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.3839 (0.5109) acc1: 48.9583 (74.7509) acc5: 100.0000 (100.0000) acc1_neg: 45.833 acc5_neg: 100.000 time: 0.1584 data: 0.0683 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.8593 (0.5156) acc1: 46.3542 (74.3851) acc5: 100.0000 (100.0000) acc1_neg: 43.969 acc5_neg: 100.000 time: 0.1546 data: 0.0545 max mem: 3532 +Test: Total time: 0:00:47 (0.2043 s / it) +* Overall Acc@1 74.385 Acc@5 100.000 + target == real Acc@1 91.358 Acc@5 100.000 + target == fake Acc@1 69.601 Acc@5 100.000 + loss 0.516 + PR-AUC 0.9676 ROC-AUC 0.8944 EER 18.68% (thr=0.3257) +Accuracy of the network on the 89729 test images: 74.4% +[rank0]:[W725 18:40:27.460242023 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:40:47.502000 608080 torch/distributed/run.py:766] +W0725 18:40:47.502000 608080 torch/distributed/run.py:766] ***************************************** +W0725 18:40:47.502000 608080 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:40:47.502000 608080 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:41:16.463866188 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:41:16.576241766 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:41:16.735227227 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:41:16.786648669 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:41:17.945680969 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:41:17.033858318 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:41:17.041054066 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:41:17.061544680 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n22_seed4_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed4_resampled_n116-041-188_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 3:53:41 acc1_pos: 67.188 acc5_pos: 100.000 loss: 0.5531 (0.5531) acc1: 67.1875 (67.1875) acc5: 100.0000 (100.0000) time: 5.8740 data: 3.9802 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/2387] eta: 0:24:23 acc1_pos: 44.792 acc5_pos: 100.000 loss: 0.7477 (0.7271) acc1: 54.4271 (58.8778) acc5: 100.0000 (100.0000) time: 0.6157 data: 0.3621 max mem: 3531 +Test: [ 20/2387] eta: 0:14:26 acc1_pos: 54.427 acc5_pos: 100.000 loss: 0.7477 (0.7244) acc1: 54.4271 (60.5407) acc5: 100.0000 (100.0000) time: 0.0907 data: 0.0011 max mem: 3531 +Test: [ 30/2387] eta: 0:11:36 acc1_pos: 49.219 acc5_pos: 100.000 loss: 0.7750 (0.7322) acc1: 55.7292 (59.9210) acc5: 100.0000 (100.0000) time: 0.1197 data: 0.0280 max mem: 3531 +Test: [ 40/2387] eta: 0:11:36 acc1_pos: 41.146 acc5_pos: 100.000 loss: 0.8271 (0.7538) acc1: 49.2188 (57.4949) acc5: 100.0000 (100.0000) time: 0.2239 data: 0.1282 max mem: 3531 +Test: [ 50/2387] eta: 0:11:25 acc1_pos: 60.417 acc5_pos: 100.000 loss: 0.7178 (0.7387) acc1: 50.5208 (58.3640) acc5: 100.0000 (100.0000) time: 0.2899 data: 0.1927 max mem: 3531 +Test: [ 60/2387] eta: 0:11:48 acc1_pos: 71.354 acc5_pos: 100.000 loss: 0.7258 (0.7457) acc1: 56.5104 (58.0003) acc5: 100.0000 (100.0000) time: 0.3199 data: 0.2229 max mem: 3531 +Test: [ 70/2387] eta: 0:11:01 acc1_pos: 60.156 acc5_pos: 100.000 loss: 0.7971 (0.7541) acc1: 52.6042 (57.2220) acc5: 100.0000 (100.0000) time: 0.2649 data: 0.1682 max mem: 3531 +Test: [ 80/2387] eta: 0:10:45 acc1_pos: 87.760 acc5_pos: 100.000 loss: 0.7167 (0.7294) acc1: 60.1562 (59.0021) acc5: 100.0000 (100.0000) time: 0.2049 data: 0.1088 max mem: 3531 +Test: [ 90/2387] eta: 0:10:30 acc1_pos: 71.354 acc5_pos: 100.000 loss: 0.4543 (0.6980) acc1: 72.1354 (61.2380) acc5: 100.0000 (100.0000) time: 0.2349 data: 0.1403 max mem: 3531 +Test: [ 100/2387] eta: 0:10:17 acc1_pos: 83.854 acc5_pos: 100.000 loss: 0.4540 (0.6771) acc1: 77.8646 (62.9280) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1344 max mem: 3531 +Test: [ 110/2387] eta: 0:10:04 acc1_pos: 74.219 acc5_pos: 100.000 loss: 0.4343 (0.6540) acc1: 81.2500 (64.7429) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1288 max mem: 3531 +Test: [ 120/2387] eta: 0:09:53 acc1_pos: 72.656 acc5_pos: 100.000 loss: 0.4126 (0.6405) acc1: 82.0312 (65.6637) acc5: 100.0000 (100.0000) time: 0.2199 data: 0.1256 max mem: 3531 +Test: [ 130/2387] eta: 0:09:43 acc1_pos: 77.344 acc5_pos: 100.000 loss: 0.4412 (0.6248) acc1: 82.0312 (66.9927) acc5: 100.0000 (100.0000) time: 0.2199 data: 0.1248 max mem: 3531 +Test: [ 140/2387] eta: 0:09:40 acc1_pos: 88.281 acc5_pos: 100.000 loss: 0.4382 (0.6114) acc1: 81.2500 (67.8838) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1452 max mem: 3531 +Test: [ 150/2387] eta: 0:09:25 acc1_pos: 67.448 acc5_pos: 100.000 loss: 0.3946 (0.5993) acc1: 79.6875 (68.8397) acc5: 100.0000 (100.0000) time: 0.2149 data: 0.1200 max mem: 3531 +Test: [ 160/2387] eta: 0:09:22 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.3958 (0.5888) acc1: 82.0312 (69.6024) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.2099 data: 0.1132 max mem: 3531 +Test: [ 170/2387] eta: 0:09:15 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.3061 (0.5718) acc1: 88.0208 (70.7937) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.2349 data: 0.1384 max mem: 3531 +Test: [ 180/2387] eta: 0:08:58 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2948 (0.5563) acc1: 90.3646 (71.8793) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1749 data: 0.0778 max mem: 3531 +Test: [ 190/2387] eta: 0:08:51 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2942 (0.5423) acc1: 90.3646 (72.8485) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1658 data: 0.0710 max mem: 3531 +Test: [ 200/2387] eta: 0:08:55 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2895 (0.5297) acc1: 90.6250 (73.7174) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.2514 data: 0.1593 max mem: 3531 +Test: [ 210/2387] eta: 0:08:49 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2943 (0.5190) acc1: 89.3229 (74.4310) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.2558 data: 0.1634 max mem: 3532 +Test: [ 220/2387] eta: 0:08:32 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.3018 (0.5092) acc1: 89.0625 (75.0896) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1534 data: 0.0594 max mem: 3532 +Test: [ 230/2387] eta: 0:08:25 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2992 (0.5001) acc1: 89.5833 (75.7384) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1410 data: 0.0475 max mem: 3532 +Test: [ 240/2387] eta: 0:08:20 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.3188 (0.4964) acc1: 88.0208 (75.9963) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.1961 data: 0.1056 max mem: 3532 +Test: [ 250/2387] eta: 0:08:11 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2661 (0.4824) acc1: 86.9792 (76.8468) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1835 data: 0.0934 max mem: 3532 +Test: [ 260/2387] eta: 0:08:06 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1053 (0.4679) acc1: 99.7396 (77.7059) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1787 data: 0.0875 max mem: 3532 +Test: [ 270/2387] eta: 0:08:00 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1028 (0.4546) acc1: 99.7396 (78.5017) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1858 data: 0.0928 max mem: 3532 +Test: [ 280/2387] eta: 0:07:59 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1052 (0.4423) acc1: 99.7396 (79.2566) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2071 data: 0.1137 max mem: 3532 +Test: [ 290/2387] eta: 0:07:48 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1005 (0.4306) acc1: 100.0000 (79.9631) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1772 data: 0.0855 max mem: 3532 +Test: [ 300/2387] eta: 0:07:42 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1005 (0.4195) acc1: 100.0000 (80.6228) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1386 data: 0.0480 max mem: 3532 +Test: [ 310/2387] eta: 0:07:38 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1060 (0.4128) acc1: 99.7396 (81.1537) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1826 data: 0.0917 max mem: 3532 +Test: [ 320/2387] eta: 0:07:36 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.3404 (0.4107) acc1: 94.7917 (81.5688) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2115 data: 0.1179 max mem: 3532 +Test: [ 330/2387] eta: 0:07:32 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.3451 (0.4088) acc1: 94.2708 (81.9479) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2044 data: 0.1089 max mem: 3532 +Test: [ 340/2387] eta: 0:07:27 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.3446 (0.4068) acc1: 94.0104 (82.2940) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1845 data: 0.0917 max mem: 3532 +Test: [ 350/2387] eta: 0:07:28 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.3354 (0.4002) acc1: 94.5312 (82.7791) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2250 data: 0.1321 max mem: 3532 +Test: [ 360/2387] eta: 0:07:22 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2026 (0.3955) acc1: 95.5729 (83.0974) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2128 data: 0.1197 max mem: 3532 +Test: [ 370/2387] eta: 0:07:17 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2407 (0.3913) acc1: 93.7500 (83.3888) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1582 data: 0.0646 max mem: 3532 +Test: [ 380/2387] eta: 0:07:13 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2417 (0.3875) acc1: 93.7500 (83.6505) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1698 data: 0.0750 max mem: 3532 +Test: [ 390/2387] eta: 0:07:12 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2377 (0.3838) acc1: 94.2708 (83.9035) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2096 data: 0.1140 max mem: 3532 +Test: [ 400/2387] eta: 0:07:07 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2420 (0.3803) acc1: 93.7500 (84.1367) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2049 data: 0.1103 max mem: 3532 +Test: [ 410/2387] eta: 0:07:02 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2376 (0.3769) acc1: 93.4896 (84.3775) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1612 data: 0.0694 max mem: 3532 +Test: [ 420/2387] eta: 0:07:05 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2376 (0.3736) acc1: 93.4896 (84.5977) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2352 data: 0.1438 max mem: 3532 +Test: [ 430/2387] eta: 0:07:01 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2476 (0.3722) acc1: 92.7083 (84.6777) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.2446 data: 0.1525 max mem: 3532 +Test: [ 440/2387] eta: 0:06:54 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2683 (0.3679) acc1: 88.2812 (84.9012) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1445 data: 0.0493 max mem: 3533 +Test: [ 450/2387] eta: 0:06:50 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.0698 (0.3613) acc1: 100.0000 (85.2359) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1362 data: 0.0405 max mem: 3533 +Test: [ 460/2387] eta: 0:06:46 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.0693 (0.3549) acc1: 100.0000 (85.5562) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1701 data: 0.0779 max mem: 3533 +Test: [ 470/2387] eta: 0:06:43 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.0696 (0.3489) acc1: 100.0000 (85.8629) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1826 data: 0.0906 max mem: 3533 +Test: [ 480/2387] eta: 0:06:41 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.0696 (0.3431) acc1: 100.0000 (86.1568) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2012 data: 0.1068 max mem: 3533 +Test: [ 490/2387] eta: 0:06:37 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.0699 (0.3375) acc1: 100.0000 (86.4387) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1828 data: 0.0878 max mem: 3533 +Test: [ 500/2387] eta: 0:06:33 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.0701 (0.3322) acc1: 100.0000 (86.7094) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1540 data: 0.0596 max mem: 3533 +Test: [ 510/2387] eta: 0:06:30 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.0701 (0.3278) acc1: 100.0000 (86.9664) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1793 data: 0.0869 max mem: 3533 +Test: [ 520/2387] eta: 0:06:27 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1337 (0.3248) acc1: 99.7396 (87.2111) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1792 data: 0.0850 max mem: 3533 +Test: [ 530/2387] eta: 0:06:22 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1280 (0.3211) acc1: 100.0000 (87.4505) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1576 data: 0.0623 max mem: 3533 +Test: [ 540/2387] eta: 0:06:21 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1277 (0.3176) acc1: 100.0000 (87.6795) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0853 max mem: 3533 +Test: [ 550/2387] eta: 0:06:17 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1278 (0.3141) acc1: 99.7396 (87.8984) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1856 data: 0.0915 max mem: 3533 +Test: [ 560/2387] eta: 0:06:13 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1281 (0.3108) acc1: 99.7396 (88.1127) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1599 data: 0.0660 max mem: 3533 +Test: [ 570/2387] eta: 0:06:10 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1284 (0.3076) acc1: 100.0000 (88.3200) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1565 data: 0.0624 max mem: 3533 +Test: [ 580/2387] eta: 0:06:07 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1281 (0.3045) acc1: 100.0000 (88.5193) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1605 data: 0.0647 max mem: 3533 +Test: [ 590/2387] eta: 0:06:04 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1279 (0.3016) acc1: 100.0000 (88.7118) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1727 data: 0.0763 max mem: 3533 +Test: [ 600/2387] eta: 0:06:01 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1378 (0.2999) acc1: 99.7396 (88.8866) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1743 data: 0.0815 max mem: 3533 +Test: [ 610/2387] eta: 0:05:57 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2338 (0.2996) acc1: 92.9688 (88.9133) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1620 data: 0.0694 max mem: 3533 +Test: [ 620/2387] eta: 0:05:54 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2772 (0.2993) acc1: 90.6250 (88.9451) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1613 data: 0.0689 max mem: 3533 +Test: [ 630/2387] eta: 0:05:52 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2772 (0.2990) acc1: 91.4062 (88.9808) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1728 data: 0.0807 max mem: 3533 +Test: [ 640/2387] eta: 0:05:49 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2879 (0.2989) acc1: 91.1458 (89.0024) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1808 data: 0.0894 max mem: 3533 +Test: [ 650/2387] eta: 0:05:47 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2911 (0.2987) acc1: 90.1042 (89.0261) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1951 data: 0.1019 max mem: 3533 +Test: [ 660/2387] eta: 0:05:44 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2841 (0.2984) acc1: 91.1458 (89.0601) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1799 data: 0.0861 max mem: 3533 +Test: [ 670/2387] eta: 0:05:41 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2820 (0.2982) acc1: 91.1458 (89.0819) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1601 data: 0.0688 max mem: 3534 +Test: [ 680/2387] eta: 0:05:39 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2782 (0.2978) acc1: 91.4062 (89.1692) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1683 data: 0.0778 max mem: 3534 +Test: [ 690/2387] eta: 0:05:36 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2854 (0.3005) acc1: 90.6250 (88.9943) acc5: 100.0000 (100.0000) acc1_neg: 67.969 acc5_neg: 100.000 time: 0.1648 data: 0.0743 max mem: 3534 +Test: [ 700/2387] eta: 0:05:33 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.5547 (0.3042) acc1: 70.5729 (88.7334) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.1672 data: 0.0757 max mem: 3534 +Test: [ 710/2387] eta: 0:05:31 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.5678 (0.3079) acc1: 70.0521 (88.4662) acc5: 100.0000 (100.0000) acc1_neg: 70.573 acc5_neg: 100.000 time: 0.1730 data: 0.0807 max mem: 3534 +Test: [ 720/2387] eta: 0:05:28 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.5583 (0.3114) acc1: 70.0521 (88.2213) acc5: 100.0000 (100.0000) acc1_neg: 72.656 acc5_neg: 100.000 time: 0.1783 data: 0.0849 max mem: 3534 +Test: [ 730/2387] eta: 0:05:26 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.5626 (0.3149) acc1: 70.0521 (87.9695) acc5: 100.0000 (100.0000) acc1_neg: 68.750 acc5_neg: 100.000 time: 0.1873 data: 0.0929 max mem: 3534 +Test: [ 740/2387] eta: 0:05:24 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.5626 (0.3183) acc1: 70.3125 (87.7355) acc5: 100.0000 (100.0000) acc1_neg: 66.667 acc5_neg: 100.000 time: 0.1843 data: 0.0921 max mem: 3534 +Test: [ 750/2387] eta: 0:05:22 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.5512 (0.3214) acc1: 70.8333 (87.5090) acc5: 100.0000 (100.0000) acc1_neg: 71.354 acc5_neg: 100.000 time: 0.1944 data: 0.1013 max mem: 3534 +Test: [ 760/2387] eta: 0:05:19 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.5382 (0.3237) acc1: 70.8333 (87.3453) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1842 data: 0.0903 max mem: 3534 +Test: [ 770/2387] eta: 0:05:17 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.4045 (0.3232) acc1: 82.0312 (87.3656) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1699 data: 0.0754 max mem: 3534 +Test: [ 780/2387] eta: 0:05:15 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1200 (0.3203) acc1: 98.4375 (87.5223) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1896 data: 0.0942 max mem: 3534 +Test: [ 790/2387] eta: 0:05:13 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.0910 (0.3180) acc1: 99.7396 (87.6119) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1847 data: 0.0923 max mem: 3534 +Test: [ 800/2387] eta: 0:05:10 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.0881 (0.3152) acc1: 99.7396 (87.7601) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1707 data: 0.0794 max mem: 3534 +Test: [ 810/2387] eta: 0:05:08 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1026 (0.3127) acc1: 99.4792 (87.8962) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1772 data: 0.0842 max mem: 3534 +Test: [ 820/2387] eta: 0:05:05 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1159 (0.3105) acc1: 98.6979 (88.0208) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1716 data: 0.0786 max mem: 3534 +Test: [ 830/2387] eta: 0:05:03 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1989 (0.3105) acc1: 93.4896 (88.0164) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1690 data: 0.0744 max mem: 3534 +Test: [ 840/2387] eta: 0:05:00 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.3711 (0.3115) acc1: 85.6771 (87.9509) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.1693 data: 0.0753 max mem: 3534 +Test: [ 850/2387] eta: 0:04:58 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.3793 (0.3128) acc1: 82.5521 (87.8840) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.1641 data: 0.0733 max mem: 3534 +Test: [ 860/2387] eta: 0:04:56 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.3975 (0.3136) acc1: 82.8125 (87.8400) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1695 data: 0.0745 max mem: 3534 +Test: [ 870/2387] eta: 0:04:53 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.3665 (0.3143) acc1: 85.4167 (87.8166) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1656 data: 0.0685 max mem: 3534 +Test: [ 880/2387] eta: 0:04:51 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.3385 (0.3149) acc1: 88.2812 (87.8021) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1685 data: 0.0751 max mem: 3534 +Test: [ 890/2387] eta: 0:04:49 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.3343 (0.3152) acc1: 89.0625 (87.8078) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.1714 data: 0.0802 max mem: 3534 +Test: [ 900/2387] eta: 0:04:46 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.3805 (0.3196) acc1: 81.5104 (87.5023) acc5: 100.0000 (100.0000) acc1_neg: 53.385 acc5_neg: 100.000 time: 0.1660 data: 0.0746 max mem: 3535 +Test: [ 910/2387] eta: 0:04:44 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.7793 (0.3247) acc1: 53.3854 (87.1244) acc5: 100.0000 (100.0000) acc1_neg: 56.250 acc5_neg: 100.000 time: 0.1644 data: 0.0705 max mem: 3535 +Test: [ 920/2387] eta: 0:04:42 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.7973 (0.3299) acc1: 52.0833 (86.7465) acc5: 100.0000 (100.0000) acc1_neg: 51.302 acc5_neg: 100.000 time: 0.1686 data: 0.0750 max mem: 3535 +Test: [ 930/2387] eta: 0:04:39 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.7982 (0.3349) acc1: 53.1250 (86.3948) acc5: 100.0000 (100.0000) acc1_neg: 50.781 acc5_neg: 100.000 time: 0.1725 data: 0.0808 max mem: 3535 +Test: [ 940/2387] eta: 0:04:37 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.8026 (0.3399) acc1: 53.1250 (86.0377) acc5: 100.0000 (100.0000) acc1_neg: 51.302 acc5_neg: 100.000 time: 0.1720 data: 0.0781 max mem: 3535 +Test: [ 950/2387] eta: 0:04:35 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.8026 (0.3447) acc1: 52.6042 (85.6872) acc5: 100.0000 (100.0000) acc1_neg: 53.125 acc5_neg: 100.000 time: 0.1713 data: 0.0781 max mem: 3535 +Test: [ 960/2387] eta: 0:04:33 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.7820 (0.3494) acc1: 53.6458 (85.3519) acc5: 100.0000 (100.0000) acc1_neg: 54.167 acc5_neg: 100.000 time: 0.1682 data: 0.0776 max mem: 3535 +Test: [ 970/2387] eta: 0:04:31 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.7972 (0.3541) acc1: 51.8229 (84.9980) acc5: 100.0000 (100.0000) acc1_neg: 51.823 acc5_neg: 100.000 time: 0.1753 data: 0.0841 max mem: 3535 +Test: [ 980/2387] eta: 0:04:28 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.7931 (0.3550) acc1: 52.3438 (84.9256) acc5: 100.0000 (100.0000) acc1_neg: 75.521 acc5_neg: 100.000 time: 0.1780 data: 0.0859 max mem: 3535 +Test: [ 990/2387] eta: 0:04:26 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1696 (0.3532) acc1: 96.0938 (85.0120) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1730 data: 0.0795 max mem: 3535 +Test: [1000/2387] eta: 0:04:24 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1133 (0.3510) acc1: 98.6979 (85.1326) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1667 data: 0.0740 max mem: 3535 +Test: [1010/2387] eta: 0:04:22 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1237 (0.3493) acc1: 98.1771 (85.2317) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.1626 data: 0.0713 max mem: 3535 +Test: [1020/2387] eta: 0:04:19 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1885 (0.3484) acc1: 91.4062 (85.2817) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1626 data: 0.0701 max mem: 3535 +Test: [1030/2387] eta: 0:04:17 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1885 (0.3468) acc1: 94.5312 (85.3727) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1645 data: 0.0728 max mem: 3535 +Test: [1040/2387] eta: 0:04:15 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1792 (0.3452) acc1: 95.8333 (85.4690) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1673 data: 0.0764 max mem: 3535 +Test: [1050/2387] eta: 0:04:13 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1758 (0.3437) acc1: 96.0938 (85.5631) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1678 data: 0.0766 max mem: 3535 +Test: [1060/2387] eta: 0:04:11 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1758 (0.3423) acc1: 96.0938 (85.6518) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1697 data: 0.0780 max mem: 3535 +Test: [1070/2387] eta: 0:04:09 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1791 (0.3415) acc1: 95.8333 (85.7036) acc5: 100.0000 (100.0000) acc1_neg: 70.833 acc5_neg: 100.000 time: 0.1732 data: 0.0790 max mem: 3535 +Test: [1080/2387] eta: 0:04:06 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2719 (0.3418) acc1: 88.2812 (85.6756) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1756 data: 0.0821 max mem: 3535 +Test: [1090/2387] eta: 0:04:04 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2141 (0.3401) acc1: 92.4479 (85.7709) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1738 data: 0.0821 max mem: 3535 +Test: [1100/2387] eta: 0:04:02 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1494 (0.3387) acc1: 96.8750 (85.8521) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1661 data: 0.0721 max mem: 3535 +Test: [1110/2387] eta: 0:04:00 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1780 (0.3376) acc1: 95.0521 (85.9012) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1684 data: 0.0745 max mem: 3535 +Test: [1120/2387] eta: 0:03:58 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1805 (0.3362) acc1: 95.3125 (85.9861) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1734 data: 0.0821 max mem: 3536 +Test: [1130/2387] eta: 0:03:56 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1660 (0.3348) acc1: 96.6146 (86.0757) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1705 data: 0.0771 max mem: 3536 +Test: [1140/2387] eta: 0:03:54 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2326 (0.3344) acc1: 92.4479 (86.1069) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.1764 data: 0.0825 max mem: 3536 +Test: [1150/2387] eta: 0:03:52 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.3144 (0.3345) acc1: 88.5417 (86.1210) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.1764 data: 0.0856 max mem: 3536 +Test: [1160/2387] eta: 0:03:50 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2834 (0.3340) acc1: 89.8438 (86.1654) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1899 data: 0.0973 max mem: 3536 +Test: [1170/2387] eta: 0:03:48 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2697 (0.3336) acc1: 90.3646 (86.2037) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1920 data: 0.0984 max mem: 3536 +Test: [1180/2387] eta: 0:03:46 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2743 (0.3333) acc1: 90.3646 (86.2363) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1681 data: 0.0761 max mem: 3536 +Test: [1190/2387] eta: 0:03:44 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2743 (0.3328) acc1: 92.9688 (86.2939) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1655 data: 0.0716 max mem: 3536 +Test: [1200/2387] eta: 0:03:42 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2406 (0.3320) acc1: 95.3125 (86.3649) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1850 data: 0.0902 max mem: 3536 +Test: [1210/2387] eta: 0:03:40 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2383 (0.3313) acc1: 95.5729 (86.4377) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1951 data: 0.1012 max mem: 3536 +Test: [1220/2387] eta: 0:03:39 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1369 (0.3295) acc1: 97.1354 (86.5347) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1964 data: 0.1041 max mem: 3536 +Test: [1230/2387] eta: 0:03:37 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1077 (0.3277) acc1: 99.4792 (86.6396) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1966 data: 0.1050 max mem: 3536 +Test: [1240/2387] eta: 0:03:35 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1071 (0.3260) acc1: 99.4792 (86.7389) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1880 data: 0.0941 max mem: 3536 +Test: [1250/2387] eta: 0:03:33 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1130 (0.3244) acc1: 99.2188 (86.8368) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1847 data: 0.0916 max mem: 3536 +Test: [1260/2387] eta: 0:03:31 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1057 (0.3227) acc1: 99.4792 (86.9354) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1782 data: 0.0873 max mem: 3536 +Test: [1270/2387] eta: 0:03:29 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1028 (0.3210) acc1: 99.7396 (87.0355) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1642 data: 0.0730 max mem: 3536 +Test: [1280/2387] eta: 0:03:27 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1041 (0.3193) acc1: 99.4792 (87.1282) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1641 data: 0.0723 max mem: 3536 +Test: [1290/2387] eta: 0:03:25 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1061 (0.3177) acc1: 98.4375 (87.1994) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1718 data: 0.0803 max mem: 3536 +Test: [1300/2387] eta: 0:03:23 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1076 (0.3161) acc1: 97.9167 (87.2768) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1907 data: 0.0999 max mem: 3536 +Test: [1310/2387] eta: 0:03:21 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1036 (0.3147) acc1: 97.9167 (87.3544) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1724 data: 0.0808 max mem: 3536 +Test: [1320/2387] eta: 0:03:19 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1188 (0.3132) acc1: 98.4375 (87.4308) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1517 data: 0.0600 max mem: 3536 +Test: [1330/2387] eta: 0:03:17 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1328 (0.3122) acc1: 96.3542 (87.4916) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1718 data: 0.0787 max mem: 3536 +Test: [1340/2387] eta: 0:03:15 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2125 (0.3131) acc1: 93.4896 (87.4390) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1798 data: 0.0847 max mem: 3536 +Test: [1350/2387] eta: 0:03:13 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.0775 (0.3113) acc1: 100.0000 (87.5320) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1789 data: 0.0862 max mem: 3537 +Test: [1360/2387] eta: 0:03:11 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.0736 (0.3095) acc1: 100.0000 (87.6236) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0883 max mem: 3537 +Test: [1370/2387] eta: 0:03:09 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.0671 (0.3078) acc1: 100.0000 (87.7137) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1788 data: 0.0862 max mem: 3537 +Test: [1380/2387] eta: 0:03:07 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.0709 (0.3061) acc1: 100.0000 (87.8027) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0770 max mem: 3537 +Test: [1390/2387] eta: 0:03:05 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.0753 (0.3044) acc1: 100.0000 (87.8902) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1772 data: 0.0847 max mem: 3537 +Test: [1400/2387] eta: 0:03:03 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.0785 (0.3029) acc1: 100.0000 (87.9762) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1798 data: 0.0884 max mem: 3537 +Test: [1410/2387] eta: 0:03:01 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.0897 (0.3024) acc1: 100.0000 (87.9963) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.1689 data: 0.0781 max mem: 3537 +Test: [1420/2387] eta: 0:03:00 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2633 (0.3024) acc1: 89.3229 (87.9862) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.1849 data: 0.0920 max mem: 3537 +Test: [1430/2387] eta: 0:02:58 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.3014 (0.3025) acc1: 87.5000 (87.9722) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.1913 data: 0.0966 max mem: 3537 +Test: [1440/2387] eta: 0:02:56 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.3201 (0.3024) acc1: 85.1562 (87.9749) acc5: 100.0000 (100.0000) acc1_neg: 72.656 acc5_neg: 100.000 time: 0.1782 data: 0.0854 max mem: 3537 +Test: [1450/2387] eta: 0:02:54 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2692 (0.3027) acc1: 86.9792 (87.9607) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1688 data: 0.0766 max mem: 3537 +Test: [1460/2387] eta: 0:02:52 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2335 (0.3022) acc1: 92.1875 (87.9962) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1727 data: 0.0807 max mem: 3537 +Test: [1470/2387] eta: 0:02:50 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1870 (0.3013) acc1: 95.3125 (88.0477) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1704 data: 0.0789 max mem: 3537 +Test: [1480/2387] eta: 0:02:48 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1817 (0.3007) acc1: 95.8333 (88.0910) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1706 data: 0.0793 max mem: 3537 +Test: [1490/2387] eta: 0:02:46 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1852 (0.2999) acc1: 96.6146 (88.1462) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1727 data: 0.0811 max mem: 3537 +Test: [1500/2387] eta: 0:02:44 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1821 (0.2991) acc1: 97.1354 (88.2006) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1731 data: 0.0812 max mem: 3537 +Test: [1510/2387] eta: 0:02:42 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2059 (0.2986) acc1: 97.3958 (88.2594) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1765 data: 0.0841 max mem: 3537 +Test: [1520/2387] eta: 0:02:40 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2601 (0.2989) acc1: 94.7917 (88.2487) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1676 data: 0.0753 max mem: 3537 +Test: [1530/2387] eta: 0:02:38 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.3605 (0.2995) acc1: 83.5938 (88.2166) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.1738 data: 0.0813 max mem: 3537 +Test: [1540/2387] eta: 0:02:36 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.3540 (0.2998) acc1: 86.4583 (88.2115) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.1725 data: 0.0796 max mem: 3537 +Test: [1550/2387] eta: 0:02:35 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.3457 (0.3001) acc1: 88.5417 (88.2164) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1788 data: 0.0867 max mem: 3537 +Test: [1560/2387] eta: 0:02:33 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.3060 (0.3002) acc1: 89.0625 (88.2257) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.1897 data: 0.0975 max mem: 3537 +Test: [1570/2387] eta: 0:02:31 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.3010 (0.3003) acc1: 91.6667 (88.2440) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1771 data: 0.0836 max mem: 3537 +Test: [1580/2387] eta: 0:02:29 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2600 (0.2995) acc1: 93.4896 (88.2964) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1771 data: 0.0838 max mem: 3538 +Test: [1590/2387] eta: 0:02:27 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1067 (0.2982) acc1: 98.9583 (88.3670) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2108 data: 0.1186 max mem: 3538 +Test: [1600/2387] eta: 0:02:25 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.0904 (0.2972) acc1: 99.4792 (88.4062) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1981 data: 0.1063 max mem: 3538 +Test: [1610/2387] eta: 0:02:24 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.0907 (0.2960) acc1: 99.4792 (88.4730) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1799 data: 0.0881 max mem: 3538 +Test: [1620/2387] eta: 0:02:22 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1096 (0.2949) acc1: 98.9583 (88.5322) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1944 data: 0.1013 max mem: 3538 +Test: [1630/2387] eta: 0:02:20 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1333 (0.2941) acc1: 97.9167 (88.5813) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1874 data: 0.0949 max mem: 3538 +Test: [1640/2387] eta: 0:02:18 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1903 (0.2936) acc1: 95.3125 (88.6163) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1853 data: 0.0915 max mem: 3538 +Test: [1650/2387] eta: 0:02:16 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2092 (0.2931) acc1: 94.5312 (88.6519) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2055 data: 0.1106 max mem: 3538 +Test: [1660/2387] eta: 0:02:15 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2026 (0.2926) acc1: 94.0104 (88.6779) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2245 data: 0.1313 max mem: 3538 +Test: [1670/2387] eta: 0:02:13 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2025 (0.2920) acc1: 94.5312 (88.7165) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2210 data: 0.1278 max mem: 3538 +Test: [1680/2387] eta: 0:02:11 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1939 (0.2916) acc1: 94.7917 (88.7420) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1832 data: 0.0908 max mem: 3538 +Test: [1690/2387] eta: 0:02:09 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1939 (0.2911) acc1: 95.3125 (88.7688) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1824 data: 0.0917 max mem: 3538 +Test: [1700/2387] eta: 0:02:07 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2017 (0.2906) acc1: 95.3125 (88.8047) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1877 data: 0.0939 max mem: 3538 +Test: [1710/2387] eta: 0:02:05 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1652 (0.2897) acc1: 96.3542 (88.8552) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1851 data: 0.0908 max mem: 3538 +Test: [1720/2387] eta: 0:02:04 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1341 (0.2889) acc1: 97.9167 (88.9082) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2016 data: 0.1095 max mem: 3538 +Test: [1730/2387] eta: 0:02:02 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1397 (0.2881) acc1: 97.3958 (88.9548) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1803 data: 0.0877 max mem: 3538 +Test: [1740/2387] eta: 0:02:00 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1379 (0.2873) acc1: 97.3958 (89.0075) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1880 data: 0.0948 max mem: 3538 +Test: [1750/2387] eta: 0:01:58 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1309 (0.2864) acc1: 98.4375 (89.0622) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2006 data: 0.1069 max mem: 3538 +Test: [1760/2387] eta: 0:01:56 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1522 (0.2857) acc1: 97.9167 (89.1023) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2058 data: 0.1133 max mem: 3538 +Test: [1770/2387] eta: 0:01:54 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1976 (0.2854) acc1: 94.7917 (89.1238) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.2002 data: 0.1082 max mem: 3538 +Test: [1780/2387] eta: 0:01:53 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2332 (0.2851) acc1: 91.6667 (89.1447) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1745 data: 0.0829 max mem: 3538 +Test: [1790/2387] eta: 0:01:51 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2120 (0.2847) acc1: 93.4896 (89.1708) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1659 data: 0.0746 max mem: 3538 +Test: [1800/2387] eta: 0:01:49 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2120 (0.2843) acc1: 94.0104 (89.1967) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1791 data: 0.0857 max mem: 3538 +Test: [1810/2387] eta: 0:01:47 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1906 (0.2837) acc1: 95.5729 (89.2395) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1836 data: 0.0909 max mem: 3539 +Test: [1820/2387] eta: 0:01:45 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1644 (0.2831) acc1: 97.1354 (89.2774) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1750 data: 0.0837 max mem: 3539 +Test: [1830/2387] eta: 0:01:43 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1803 (0.2826) acc1: 96.6146 (89.3185) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1752 data: 0.0825 max mem: 3539 +Test: [1840/2387] eta: 0:01:41 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1788 (0.2820) acc1: 98.1771 (89.3697) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1796 data: 0.0859 max mem: 3539 +Test: [1850/2387] eta: 0:01:39 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1745 (0.2814) acc1: 98.9583 (89.4194) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1747 data: 0.0804 max mem: 3539 +Test: [1860/2387] eta: 0:01:37 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1737 (0.2809) acc1: 98.6979 (89.4683) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1651 data: 0.0721 max mem: 3539 +Test: [1870/2387] eta: 0:01:35 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1775 (0.2804) acc1: 98.1771 (89.5137) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1750 data: 0.0832 max mem: 3539 +Test: [1880/2387] eta: 0:01:34 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1796 (0.2798) acc1: 98.1771 (89.5605) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1772 data: 0.0850 max mem: 3539 +Test: [1890/2387] eta: 0:01:32 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1788 (0.2793) acc1: 98.6979 (89.6074) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1851 data: 0.0938 max mem: 3539 +Test: [1900/2387] eta: 0:01:30 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1804 (0.2788) acc1: 98.4375 (89.6490) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1839 data: 0.0931 max mem: 3539 +Test: [1910/2387] eta: 0:01:28 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2677 (0.2791) acc1: 91.9271 (89.6487) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1710 data: 0.0798 max mem: 3539 +Test: [1920/2387] eta: 0:01:26 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.3358 (0.2794) acc1: 89.0625 (89.6320) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1797 data: 0.0891 max mem: 3539 +Test: [1930/2387] eta: 0:01:24 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2223 (0.2788) acc1: 92.7083 (89.6632) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1839 data: 0.0924 max mem: 3539 +Test: [1940/2387] eta: 0:01:22 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2197 (0.2791) acc1: 93.4896 (89.6338) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1780 data: 0.0857 max mem: 3539 +Test: [1950/2387] eta: 0:01:20 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.3084 (0.2793) acc1: 88.0208 (89.6279) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1704 data: 0.0780 max mem: 3539 +Test: [1960/2387] eta: 0:01:19 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2532 (0.2793) acc1: 91.4062 (89.6322) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1858 data: 0.0935 max mem: 3539 +Test: [1970/2387] eta: 0:01:17 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.3257 (0.2803) acc1: 87.5000 (89.5607) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.1976 data: 0.1049 max mem: 3539 +Test: [1980/2387] eta: 0:01:15 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.3488 (0.2805) acc1: 84.8958 (89.5564) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1811 data: 0.0868 max mem: 3539 +Test: [1990/2387] eta: 0:01:13 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2037 (0.2801) acc1: 96.6146 (89.5905) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1715 data: 0.0768 max mem: 3539 +Test: [2000/2387] eta: 0:01:11 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2037 (0.2798) acc1: 96.8750 (89.6268) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1847 data: 0.0919 max mem: 3539 +Test: [2010/2387] eta: 0:01:09 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2064 (0.2794) acc1: 96.8750 (89.6606) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1813 data: 0.0894 max mem: 3539 +Test: [2020/2387] eta: 0:01:08 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2064 (0.2790) acc1: 96.8750 (89.6976) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1847 data: 0.0934 max mem: 3539 +Test: [2030/2387] eta: 0:01:06 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1983 (0.2787) acc1: 96.8750 (89.7317) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1842 data: 0.0935 max mem: 3540 +Test: [2040/2387] eta: 0:01:04 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2040 (0.2783) acc1: 96.6146 (89.7671) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1731 data: 0.0820 max mem: 3540 +Test: [2050/2387] eta: 0:01:02 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2040 (0.2780) acc1: 96.6146 (89.7982) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1931 data: 0.1016 max mem: 3540 +Test: [2060/2387] eta: 0:01:00 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2411 (0.2779) acc1: 94.7917 (89.8205) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1779 data: 0.0866 max mem: 3540 +Test: [2070/2387] eta: 0:00:58 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2510 (0.2777) acc1: 94.2708 (89.8440) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1670 data: 0.0752 max mem: 3540 +Test: [2080/2387] eta: 0:00:56 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2459 (0.2776) acc1: 94.2708 (89.8670) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1793 data: 0.0868 max mem: 3540 +Test: [2090/2387] eta: 0:00:54 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2422 (0.2774) acc1: 94.7917 (89.8911) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1799 data: 0.0888 max mem: 3540 +Test: [2100/2387] eta: 0:00:53 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2434 (0.2773) acc1: 94.7917 (89.9151) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1782 data: 0.0873 max mem: 3540 +Test: [2110/2387] eta: 0:00:51 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2427 (0.2771) acc1: 94.5312 (89.9375) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1673 data: 0.0752 max mem: 3540 +Test: [2120/2387] eta: 0:00:49 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2409 (0.2769) acc1: 94.5312 (89.9502) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.1599 data: 0.0681 max mem: 3540 +Test: [2130/2387] eta: 0:00:47 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1586 (0.2761) acc1: 96.3542 (89.9906) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1651 data: 0.0738 max mem: 3540 +Test: [2140/2387] eta: 0:00:45 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1308 (0.2758) acc1: 96.8750 (89.9996) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1676 data: 0.0762 max mem: 3540 +Test: [2150/2387] eta: 0:00:43 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1714 (0.2753) acc1: 95.5729 (90.0257) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1605 data: 0.0693 max mem: 3540 +Test: [2160/2387] eta: 0:00:41 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1738 (0.2749) acc1: 96.0938 (90.0543) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1588 data: 0.0676 max mem: 3540 +Test: [2170/2387] eta: 0:00:40 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1962 (0.2749) acc1: 95.0521 (90.0515) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.1657 data: 0.0743 max mem: 3540 +Test: [2180/2387] eta: 0:00:38 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.3304 (0.2760) acc1: 84.6354 (89.9750) acc5: 100.0000 (100.0000) acc1_neg: 58.073 acc5_neg: 100.000 time: 0.1697 data: 0.0761 max mem: 3540 +Test: [2190/2387] eta: 0:00:36 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.3185 (0.2764) acc1: 86.7188 (89.9617) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1683 data: 0.0734 max mem: 3540 +Test: [2200/2387] eta: 0:00:34 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2330 (0.2761) acc1: 92.7083 (89.9825) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1658 data: 0.0735 max mem: 3540 +Test: [2210/2387] eta: 0:00:32 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2130 (0.2758) acc1: 94.5312 (90.0055) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1705 data: 0.0803 max mem: 3540 +Test: [2220/2387] eta: 0:00:30 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2163 (0.2756) acc1: 94.2708 (90.0241) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1719 data: 0.0785 max mem: 3540 +Test: [2230/2387] eta: 0:00:28 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2054 (0.2752) acc1: 94.7917 (90.0499) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1693 data: 0.0747 max mem: 3540 +Test: [2240/2387] eta: 0:00:27 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1868 (0.2748) acc1: 94.7917 (90.0720) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1754 data: 0.0841 max mem: 3540 +Test: [2250/2387] eta: 0:00:25 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2209 (0.2748) acc1: 94.2708 (90.0850) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2004 data: 0.1103 max mem: 3540 +Test: [2260/2387] eta: 0:00:23 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2309 (0.2746) acc1: 93.4896 (90.1023) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2014 data: 0.1084 max mem: 3541 +Test: [2270/2387] eta: 0:00:21 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1445 (0.2739) acc1: 97.3958 (90.1429) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1761 data: 0.0819 max mem: 3541 +Test: [2280/2387] eta: 0:00:19 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1004 (0.2734) acc1: 98.9583 (90.1581) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1879 data: 0.0946 max mem: 3541 +Test: [2290/2387] eta: 0:00:17 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1108 (0.2727) acc1: 98.9583 (90.1970) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1981 data: 0.1037 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1277 (0.2723) acc1: 98.1771 (90.2229) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1918 data: 0.0976 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.1746 (0.2720) acc1: 95.8333 (90.2419) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2099 data: 0.1172 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.2367 (0.2723) acc1: 92.1875 (90.2215) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.1889 data: 0.0972 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.3874 (0.2729) acc1: 82.0312 (90.1765) acc5: 100.0000 (100.0000) acc1_neg: 71.354 acc5_neg: 100.000 time: 0.1602 data: 0.0675 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.3855 (0.2731) acc1: 82.0312 (90.1589) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1672 data: 0.0746 max mem: 3541 +Test: [2350/2387] eta: 0:00:06 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.3652 (0.2740) acc1: 82.2917 (90.0942) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.1874 data: 0.0938 max mem: 3541 +Test: [2360/2387] eta: 0:00:04 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.4347 (0.2748) acc1: 78.9062 (90.0425) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.2017 data: 0.1080 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.4042 (0.2754) acc1: 78.9062 (89.9984) acc5: 100.0000 (100.0000) acc1_neg: 71.615 acc5_neg: 100.000 time: 0.2048 data: 0.1113 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.4956 (0.2765) acc1: 73.1771 (89.9264) acc5: 100.0000 (100.0000) acc1_neg: 63.542 acc5_neg: 100.000 time: 0.1922 data: 0.0974 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 84.018 acc5_pos: 100.000 loss: 0.5321 (0.2771) acc1: 71.6146 (89.8916) acc5: 100.0000 (100.0000) acc1_neg: 81.395 acc5_neg: 100.000 time: 0.1817 data: 0.0914 max mem: 3541 +Test: Total time: 0:07:20 (0.1845 s / it) +* Overall Acc@1 89.892 Acc@5 100.000 + target == real Acc@1 69.391 Acc@5 100.000 + target == fake Acc@1 91.331 Acc@5 100.000 + loss 0.277 + PR-AUC 0.9932 ROC-AUC 0.9153 EER 16.22% (thr=0.6597) +Accuracy of the network on the 916267 test images: 89.9% +[rank0]:[W725 18:49:09.386010378 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:49:29.270000 608755 torch/distributed/run.py:766] +W0725 18:49:29.270000 608755 torch/distributed/run.py:766] ***************************************** +W0725 18:49:29.270000 608755 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:49:29.270000 608755 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W725 18:49:56.230377745 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:49:56.345896738 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:49:56.433696476 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:49:56.804716091 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:49:56.818519232 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +[rank6]:[W725 18:49:57.188875386 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:49:57.198437981 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:49:57.226961381 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n22_seed4_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed4_resampled_n116-041-188_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 0:54:58 acc1_pos: 67.188 acc5_pos: 100.000 loss: 0.5531 (0.5531) acc1: 67.1875 (67.1875) acc5: 100.0000 (100.0000) time: 6.3802 data: 4.8434 max mem: 3531 +Test: [ 10/517] eta: 0:05:35 acc1_pos: 44.792 acc5_pos: 100.000 loss: 0.7477 (0.7271) acc1: 54.4271 (58.8778) acc5: 100.0000 (100.0000) time: 0.6620 data: 0.4407 max mem: 3531 +Test: [ 20/517] eta: 0:03:15 acc1_pos: 54.427 acc5_pos: 100.000 loss: 0.7477 (0.7244) acc1: 54.4271 (60.5407) acc5: 100.0000 (100.0000) time: 0.0948 data: 0.0036 max mem: 3531 +Test: [ 30/517] eta: 0:02:38 acc1_pos: 49.219 acc5_pos: 100.000 loss: 0.7750 (0.7322) acc1: 55.7292 (59.9210) acc5: 100.0000 (100.0000) time: 0.1406 data: 0.0483 max mem: 3531 +Test: [ 40/517] eta: 0:02:27 acc1_pos: 41.146 acc5_pos: 100.000 loss: 0.8271 (0.7538) acc1: 49.2188 (57.4949) acc5: 100.0000 (100.0000) time: 0.2204 data: 0.1293 max mem: 3531 +Test: [ 50/517] eta: 0:02:17 acc1_pos: 78.899 acc5_pos: 100.000 loss: 0.7178 (0.7338) acc1: 50.5208 (58.6295) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.2458 data: 0.1549 max mem: 3531 +Test: [ 60/517] eta: 0:02:09 acc1_pos: 78.899 acc5_pos: 100.000 loss: 0.5621 (0.6962) acc1: 68.7500 (61.3943) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.2314 data: 0.1400 max mem: 3531 +Test: [ 70/517] eta: 0:02:06 acc1_pos: 78.899 acc5_pos: 100.000 loss: 0.4632 (0.6614) acc1: 78.3854 (63.9965) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.2540 data: 0.1632 max mem: 3531 +Test: [ 80/517] eta: 0:02:01 acc1_pos: 78.899 acc5_pos: 100.000 loss: 0.4632 (0.6453) acc1: 76.5625 (65.0527) acc5: 100.0000 (100.0000) acc1_neg: 64.583 acc5_neg: 100.000 time: 0.2639 data: 0.1734 max mem: 3531 +Test: [ 90/517] eta: 0:01:56 acc1_pos: 78.899 acc5_pos: 100.000 loss: 0.4014 (0.6152) acc1: 82.8125 (67.5767) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.2395 data: 0.1486 max mem: 3531 +Test: [100/517] eta: 0:01:52 acc1_pos: 78.899 acc5_pos: 100.000 loss: 0.3192 (0.5845) acc1: 90.3646 (69.9773) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2386 data: 0.1444 max mem: 3531 +Test: [110/517] eta: 0:01:49 acc1_pos: 78.899 acc5_pos: 100.000 loss: 0.2535 (0.5523) acc1: 93.7500 (72.3137) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2491 data: 0.1524 max mem: 3531 +Test: [120/517] eta: 0:01:45 acc1_pos: 78.899 acc5_pos: 100.000 loss: 0.2337 (0.5281) acc1: 95.8333 (74.1327) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2452 data: 0.1516 max mem: 3531 +Test: [130/517] eta: 0:01:43 acc1_pos: 78.899 acc5_pos: 100.000 loss: 0.2787 (0.5084) acc1: 92.9688 (75.6381) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.2549 data: 0.1581 max mem: 3531 +Test: [140/517] eta: 0:01:40 acc1_pos: 78.899 acc5_pos: 100.000 loss: 0.2809 (0.4937) acc1: 92.9688 (76.7731) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.2746 data: 0.1742 max mem: 3531 +Test: [150/517] eta: 0:01:37 acc1_pos: 78.899 acc5_pos: 100.000 loss: 0.2984 (0.4824) acc1: 91.4062 (77.6266) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.2649 data: 0.1686 max mem: 3531 +Test: [160/517] eta: 0:01:35 acc1_pos: 78.899 acc5_pos: 100.000 loss: 0.3063 (0.4724) acc1: 89.8438 (78.2706) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.2549 data: 0.1582 max mem: 3531 +Test: [170/517] eta: 0:01:31 acc1_pos: 78.899 acc5_pos: 100.000 loss: 0.3063 (0.4633) acc1: 88.0208 (78.8240) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.2454 data: 0.1502 max mem: 3531 +Test: [180/517] eta: 0:01:28 acc1_pos: 78.899 acc5_pos: 100.000 loss: 0.3088 (0.4543) acc1: 87.7604 (79.4084) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.2449 data: 0.1530 max mem: 3531 +Test: [190/517] eta: 0:01:25 acc1_pos: 78.899 acc5_pos: 100.000 loss: 0.2088 (0.4402) acc1: 94.2708 (80.3133) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2494 data: 0.1562 max mem: 3531 +Test: [200/517] eta: 0:01:23 acc1_pos: 78.899 acc5_pos: 100.000 loss: 0.2278 (0.4315) acc1: 95.5729 (80.9468) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.2749 data: 0.1784 max mem: 3531 +Test: [210/517] eta: 0:01:21 acc1_pos: 78.899 acc5_pos: 100.000 loss: 0.3329 (0.4315) acc1: 89.0625 (81.0106) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.2799 data: 0.1822 max mem: 3532 +Test: [220/517] eta: 0:01:18 acc1_pos: 78.899 acc5_pos: 100.000 loss: 0.4168 (0.4315) acc1: 82.8125 (81.0190) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.2399 data: 0.1441 max mem: 3532 +Test: [230/517] eta: 0:01:15 acc1_pos: 78.899 acc5_pos: 100.000 loss: 0.4210 (0.4307) acc1: 82.8125 (81.1993) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.2399 data: 0.1450 max mem: 3532 +Test: [240/517] eta: 0:01:12 acc1_pos: 78.899 acc5_pos: 100.000 loss: 0.3168 (0.4235) acc1: 91.6667 (81.7503) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.2499 data: 0.1550 max mem: 3532 +Test: [250/517] eta: 0:01:09 acc1_pos: 78.899 acc5_pos: 100.000 loss: 0.2943 (0.4206) acc1: 94.5312 (82.0634) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2449 data: 0.1505 max mem: 3532 +Test: [260/517] eta: 0:01:06 acc1_pos: 78.899 acc5_pos: 100.000 loss: 0.3593 (0.4194) acc1: 87.7604 (82.2278) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.2404 data: 0.1480 max mem: 3532 +Test: [270/517] eta: 0:01:04 acc1_pos: 78.899 acc5_pos: 100.000 loss: 0.3881 (0.4181) acc1: 87.5000 (82.3839) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2449 data: 0.1497 max mem: 3532 +Test: [280/517] eta: 0:01:01 acc1_pos: 78.899 acc5_pos: 100.000 loss: 0.3359 (0.4147) acc1: 89.5833 (82.6892) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2447 data: 0.1493 max mem: 3532 +Test: [290/517] eta: 0:00:58 acc1_pos: 78.899 acc5_pos: 100.000 loss: 0.2874 (0.4118) acc1: 90.8854 (82.9369) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.2559 data: 0.1648 max mem: 3532 +Test: [300/517] eta: 0:00:56 acc1_pos: 78.899 acc5_pos: 100.000 loss: 0.2761 (0.4072) acc1: 93.4896 (83.3238) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.2503 data: 0.1592 max mem: 3532 +Test: [310/517] eta: 0:00:53 acc1_pos: 78.899 acc5_pos: 100.000 loss: 0.3255 (0.4081) acc1: 87.7604 (83.2513) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.2347 data: 0.1439 max mem: 3532 +Test: [320/517] eta: 0:00:50 acc1_pos: 78.899 acc5_pos: 100.000 loss: 0.4039 (0.4058) acc1: 83.5938 (83.4088) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2442 data: 0.1516 max mem: 3532 +Test: [330/517] eta: 0:00:48 acc1_pos: 78.899 acc5_pos: 100.000 loss: 0.2833 (0.4015) acc1: 92.7083 (83.7228) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.2591 data: 0.1621 max mem: 3532 +Test: [340/517] eta: 0:00:45 acc1_pos: 78.899 acc5_pos: 100.000 loss: 0.2503 (0.3993) acc1: 93.4896 (83.9130) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.2555 data: 0.1570 max mem: 3532 +Test: [350/517] eta: 0:00:42 acc1_pos: 78.899 acc5_pos: 100.000 loss: 0.4117 (0.4018) acc1: 84.3750 (83.6976) acc5: 100.0000 (100.0000) acc1_neg: 50.000 acc5_neg: 100.000 time: 0.2449 data: 0.1503 max mem: 3532 +Test: [360/517] eta: 0:00:40 acc1_pos: 78.899 acc5_pos: 100.000 loss: 0.4730 (0.4045) acc1: 76.5625 (83.4906) acc5: 100.0000 (100.0000) acc1_neg: 68.750 acc5_neg: 100.000 time: 0.2447 data: 0.1527 max mem: 3532 +Test: [370/517] eta: 0:00:37 acc1_pos: 78.899 acc5_pos: 100.000 loss: 0.3755 (0.4024) acc1: 87.5000 (83.7089) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2449 data: 0.1520 max mem: 3532 +Test: [380/517] eta: 0:00:35 acc1_pos: 78.899 acc5_pos: 100.000 loss: 0.3197 (0.4010) acc1: 91.6667 (83.8644) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2449 data: 0.1519 max mem: 3532 +Test: [390/517] eta: 0:00:32 acc1_pos: 78.899 acc5_pos: 100.000 loss: 0.3203 (0.3987) acc1: 92.1875 (84.0546) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.2799 data: 0.1867 max mem: 3532 +Test: [400/517] eta: 0:00:30 acc1_pos: 78.899 acc5_pos: 100.000 loss: 0.3509 (0.3984) acc1: 89.3229 (84.1308) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.2947 data: 0.2005 max mem: 3532 +Test: [410/517] eta: 0:00:27 acc1_pos: 78.899 acc5_pos: 100.000 loss: 0.4302 (0.4000) acc1: 82.2917 (83.9974) acc5: 100.0000 (100.0000) acc1_neg: 68.229 acc5_neg: 100.000 time: 0.2649 data: 0.1701 max mem: 3532 +Test: [420/517] eta: 0:00:24 acc1_pos: 78.899 acc5_pos: 100.000 loss: 0.3657 (0.3985) acc1: 82.2917 (84.1919) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.2447 data: 0.1492 max mem: 3532 +Test: [430/517] eta: 0:00:22 acc1_pos: 78.899 acc5_pos: 100.000 loss: 0.3375 (0.3985) acc1: 87.7604 (84.1998) acc5: 100.0000 (100.0000) acc1_neg: 78.646 acc5_neg: 100.000 time: 0.2353 data: 0.1398 max mem: 3532 +Test: [440/517] eta: 0:00:19 acc1_pos: 78.899 acc5_pos: 100.000 loss: 0.3691 (0.3971) acc1: 87.5000 (84.3266) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.2449 data: 0.1493 max mem: 3533 +Test: [450/517] eta: 0:00:17 acc1_pos: 78.899 acc5_pos: 100.000 loss: 0.3001 (0.3954) acc1: 90.8854 (84.4691) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.2503 data: 0.1546 max mem: 3533 +Test: [460/517] eta: 0:00:14 acc1_pos: 78.899 acc5_pos: 100.000 loss: 0.3780 (0.3961) acc1: 86.4583 (84.4620) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.2456 data: 0.1514 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 78.899 acc5_pos: 100.000 loss: 0.3831 (0.3949) acc1: 86.4583 (84.6011) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.2391 data: 0.1434 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 78.899 acc5_pos: 100.000 loss: 0.3187 (0.3927) acc1: 93.2292 (84.8222) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2342 data: 0.1372 max mem: 3533 +Test: [490/517] eta: 0:00:06 acc1_pos: 78.899 acc5_pos: 100.000 loss: 0.2780 (0.3910) acc1: 94.0104 (84.9839) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2319 data: 0.1353 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 78.899 acc5_pos: 100.000 loss: 0.3450 (0.3916) acc1: 91.4062 (84.9561) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.2399 data: 0.1417 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 78.899 acc5_pos: 100.000 loss: 0.3450 (0.3911) acc1: 86.1979 (84.9667) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2107 data: 0.1156 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 78.899 acc5_pos: 100.000 loss: 0.3234 (0.3898) acc1: 89.0459 (85.0479) acc5: 100.0000 (100.0000) acc1_neg: 89.046 acc5_neg: 100.000 time: 0.1471 data: 0.0530 max mem: 3533 +Test: Total time: 0:02:10 (0.2522 s / it) +* Overall Acc@1 85.048 Acc@5 100.000 + target == real Acc@1 58.043 Acc@5 100.000 + target == fake Acc@1 87.849 Acc@5 100.000 + loss 0.390 + PR-AUC 0.9725 ROC-AUC 0.8194 EER 25.78% (thr=0.6436) +Accuracy of the network on the 198427 test images: 85.0% +[rank0]:[W725 18:52:20.760870598 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0725 18:52:40.013000 609332 torch/distributed/run.py:766] +W0725 18:52:40.013000 609332 torch/distributed/run.py:766] ***************************************** +W0725 18:52:40.013000 609332 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:52:40.013000 609332 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W725 18:53:08.066375713 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:53:08.173151017 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:53:08.308551942 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:53:08.538059582 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:53:08.608417887 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:53:08.710832842 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:53:08.788225614 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:53:08.813974848 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n22_seed4_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed4_resampled_n116-041-188_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:31:56 acc1_pos: 62.760 acc5_pos: 100.000 loss: 0.6472 (0.6472) acc1: 62.7604 (62.7604) acc5: 100.0000 (100.0000) time: 4.9274 data: 3.1640 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/389] eta: 0:03:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2477 (0.3415) acc1: 98.4375 (88.8494) acc5: 100.0000 (100.0000) time: 0.5345 data: 0.2927 max mem: 3531 +Test: [ 20/389] eta: 0:02:00 acc1_pos: 72.135 acc5_pos: 100.000 loss: 0.2477 (0.3672) acc1: 96.6146 (86.1979) acc5: 100.0000 (100.0000) time: 0.0979 data: 0.0082 max mem: 3531 +Test: [ 30/389] eta: 0:01:30 acc1_pos: 98.698 acc5_pos: 100.000 loss: 0.3569 (0.3565) acc1: 92.1875 (88.4997) acc5: 100.0000 (100.0000) time: 0.0967 data: 0.0057 max mem: 3531 +Test: [ 40/389] eta: 0:01:18 acc1_pos: 77.865 acc5_pos: 100.000 loss: 0.3575 (0.3671) acc1: 92.1875 (87.6715) acc5: 100.0000 (100.0000) time: 0.1176 data: 0.0257 max mem: 3531 +Test: [ 50/389] eta: 0:01:11 acc1_pos: 95.573 acc5_pos: 100.000 loss: 0.3367 (0.3529) acc1: 95.5729 (89.2157) acc5: 100.0000 (100.0000) time: 0.1500 data: 0.0564 max mem: 3531 +Test: [ 60/389] eta: 0:01:09 acc1_pos: 98.698 acc5_pos: 100.000 loss: 0.3410 (0.3599) acc1: 95.5729 (89.3613) acc5: 100.0000 (100.0000) time: 0.1868 data: 0.0905 max mem: 3531 +Test: [ 70/389] eta: 0:01:07 acc1_pos: 81.510 acc5_pos: 100.000 loss: 0.3513 (0.3670) acc1: 90.6250 (89.0918) acc5: 100.0000 (100.0000) time: 0.2124 data: 0.1157 max mem: 3531 +Test: [ 80/389] eta: 0:01:02 acc1_pos: 93.750 acc5_pos: 100.000 loss: 0.4209 (0.3789) acc1: 86.1979 (88.4613) acc5: 100.0000 (100.0000) time: 0.1768 data: 0.0802 max mem: 3531 +Test: [ 90/389] eta: 0:01:00 acc1_pos: 99.219 acc5_pos: 100.000 loss: 0.2480 (0.3615) acc1: 93.7500 (89.3716) acc5: 100.0000 (100.0000) time: 0.1640 data: 0.0701 max mem: 3531 +Test: [100/389] eta: 0:00:58 acc1_pos: 90.885 acc5_pos: 100.000 loss: 0.3175 (0.3711) acc1: 92.4479 (88.6010) acc5: 100.0000 (100.0000) time: 0.2003 data: 0.1099 max mem: 3531 +Test: [110/389] eta: 0:00:56 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4523 (0.3916) acc1: 82.8125 (86.5827) acc5: 100.0000 (100.0000) time: 0.2036 data: 0.1131 max mem: 3531 +Test: [120/389] eta: 0:00:53 acc1_pos: 87.760 acc5_pos: 100.000 loss: 0.3871 (0.3897) acc1: 87.5000 (86.5659) acc5: 100.0000 (100.0000) time: 0.1699 data: 0.0761 max mem: 3531 +Test: [130/389] eta: 0:00:50 acc1_pos: 61.963 acc5_pos: 100.000 loss: 0.5336 (0.4143) acc1: 70.8333 (84.3213) acc5: 100.0000 (100.0000) acc1_neg: 70.833 acc5_neg: 100.000 time: 0.1664 data: 0.0719 max mem: 3531 +Test: [140/389] eta: 0:00:48 acc1_pos: 61.963 acc5_pos: 100.000 loss: 0.8935 (0.4520) acc1: 34.6354 (80.8991) acc5: 100.0000 (100.0000) acc1_neg: 20.052 acc5_neg: 100.000 time: 0.1810 data: 0.0898 max mem: 3531 +Test: [150/389] eta: 0:00:46 acc1_pos: 61.963 acc5_pos: 100.000 loss: 0.9559 (0.4832) acc1: 30.7292 (78.2681) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1862 data: 0.0939 max mem: 3531 +Test: [160/389] eta: 0:00:45 acc1_pos: 61.963 acc5_pos: 100.000 loss: 0.7855 (0.4924) acc1: 45.8333 (77.7449) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2085 data: 0.1154 max mem: 3531 +Test: [170/389] eta: 0:00:42 acc1_pos: 61.963 acc5_pos: 100.000 loss: 0.4033 (0.5094) acc1: 89.5833 (76.8351) acc5: 100.0000 (100.0000) acc1_neg: 2.083 acc5_neg: 100.000 time: 0.2048 data: 0.1134 max mem: 3531 +Test: [180/389] eta: 0:00:41 acc1_pos: 61.963 acc5_pos: 100.000 loss: 0.5984 (0.5325) acc1: 66.6667 (74.9252) acc5: 100.0000 (100.0000) acc1_neg: 45.833 acc5_neg: 100.000 time: 0.1958 data: 0.1043 max mem: 3531 +Test: [190/389] eta: 0:00:39 acc1_pos: 61.963 acc5_pos: 100.000 loss: 0.8784 (0.5602) acc1: 32.0312 (72.7831) acc5: 100.0000 (100.0000) acc1_neg: 0.000 acc5_neg: 100.000 time: 0.2003 data: 0.1089 max mem: 3531 +Test: [200/389] eta: 0:00:36 acc1_pos: 61.963 acc5_pos: 100.000 loss: 0.9790 (0.5829) acc1: 29.9479 (70.9538) acc5: 100.0000 (100.0000) acc1_neg: 8.854 acc5_neg: 100.000 time: 0.1856 data: 0.0944 max mem: 3531 +Test: [210/389] eta: 0:00:35 acc1_pos: 61.963 acc5_pos: 100.000 loss: 0.6384 (0.5907) acc1: 38.2812 (70.3137) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1969 data: 0.1047 max mem: 3532 +Test: [220/389] eta: 0:00:32 acc1_pos: 61.963 acc5_pos: 100.000 loss: 0.4989 (0.5853) acc1: 75.7812 (70.9653) acc5: 100.0000 (100.0000) acc1_neg: 79.427 acc5_neg: 100.000 time: 0.1900 data: 0.0979 max mem: 3532 +Test: [230/389] eta: 0:00:30 acc1_pos: 61.963 acc5_pos: 100.000 loss: 0.5064 (0.5942) acc1: 76.8229 (70.0972) acc5: 100.0000 (100.0000) acc1_neg: 70.052 acc5_neg: 100.000 time: 0.1714 data: 0.0781 max mem: 3532 +Test: [240/389] eta: 0:00:28 acc1_pos: 61.963 acc5_pos: 100.000 loss: 0.7835 (0.6026) acc1: 45.5729 (69.2384) acc5: 100.0000 (100.0000) acc1_neg: 32.031 acc5_neg: 100.000 time: 0.1853 data: 0.0920 max mem: 3532 +Test: [250/389] eta: 0:00:26 acc1_pos: 61.963 acc5_pos: 100.000 loss: 0.6124 (0.5985) acc1: 55.4688 (69.4752) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1780 data: 0.0865 max mem: 3532 +Test: [260/389] eta: 0:00:24 acc1_pos: 61.963 acc5_pos: 100.000 loss: 0.3082 (0.5956) acc1: 83.8542 (69.7428) acc5: 100.0000 (100.0000) acc1_neg: 41.146 acc5_neg: 100.000 time: 0.1659 data: 0.0743 max mem: 3532 +Test: [270/389] eta: 0:00:22 acc1_pos: 61.963 acc5_pos: 100.000 loss: 0.7350 (0.6081) acc1: 48.9583 (68.5809) acc5: 100.0000 (100.0000) acc1_neg: 52.083 acc5_neg: 100.000 time: 0.1658 data: 0.0745 max mem: 3532 +Test: [280/389] eta: 0:00:20 acc1_pos: 61.963 acc5_pos: 100.000 loss: 0.8112 (0.6199) acc1: 43.2292 (67.6499) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1617 data: 0.0698 max mem: 3532 +Test: [290/389] eta: 0:00:18 acc1_pos: 61.963 acc5_pos: 100.000 loss: 0.7075 (0.6201) acc1: 50.0000 (67.4783) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1541 data: 0.0618 max mem: 3532 +Test: [300/389] eta: 0:00:16 acc1_pos: 61.963 acc5_pos: 100.000 loss: 0.8814 (0.6402) acc1: 32.0312 (65.9313) acc5: 100.0000 (100.0000) acc1_neg: 15.365 acc5_neg: 100.000 time: 0.1549 data: 0.0632 max mem: 3532 +Test: [310/389] eta: 0:00:14 acc1_pos: 61.963 acc5_pos: 100.000 loss: 1.3090 (0.6594) acc1: 10.1562 (64.7885) acc5: 100.0000 (100.0000) acc1_neg: 3.906 acc5_neg: 100.000 time: 0.1603 data: 0.0686 max mem: 3532 +Test: [320/389] eta: 0:00:12 acc1_pos: 61.963 acc5_pos: 100.000 loss: 0.9390 (0.6632) acc1: 37.5000 (64.4373) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.1633 data: 0.0706 max mem: 3532 +Test: [330/389] eta: 0:00:10 acc1_pos: 61.963 acc5_pos: 100.000 loss: 0.9493 (0.6720) acc1: 22.9167 (63.2584) acc5: 100.0000 (100.0000) acc1_neg: 27.865 acc5_neg: 100.000 time: 0.1703 data: 0.0779 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 61.963 acc5_pos: 100.000 loss: 0.6188 (0.6647) acc1: 64.3229 (63.9059) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1699 data: 0.0789 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 61.963 acc5_pos: 100.000 loss: 0.4317 (0.6801) acc1: 76.3021 (63.3161) acc5: 100.0000 (100.0000) acc1_neg: 1.042 acc5_neg: 100.000 time: 0.1710 data: 0.0798 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 61.963 acc5_pos: 100.000 loss: 1.3714 (0.7029) acc1: 17.4479 (61.9655) acc5: 100.0000 (100.0000) acc1_neg: 19.010 acc5_neg: 100.000 time: 0.1684 data: 0.0770 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 61.963 acc5_pos: 100.000 loss: 1.3425 (0.7037) acc1: 19.2708 (61.8802) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1577 data: 0.0663 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 61.963 acc5_pos: 100.000 loss: 0.6754 (0.7028) acc1: 56.2500 (61.7953) acc5: 100.0000 (100.0000) acc1_neg: 59.115 acc5_neg: 100.000 time: 0.1415 data: 0.0508 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 61.963 acc5_pos: 100.000 loss: 0.6511 (0.7030) acc1: 59.1146 (61.7095) acc5: 100.0000 (100.0000) acc1_neg: 45.902 acc5_neg: 100.000 time: 0.1396 data: 0.0368 max mem: 3532 +Test: Total time: 0:01:10 (0.1821 s / it) +* Overall Acc@1 61.709 Acc@5 100.000 + target == real Acc@1 86.243 Acc@5 100.000 + target == fake Acc@1 50.306 Acc@5 100.000 + loss 0.703 + PR-AUC 0.8604 ROC-AUC 0.7474 EER 32.11% (thr=0.3396) +Accuracy of the network on the 149358 test images: 61.7% +[rank0]:[W725 18:54:30.695970031 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:54:49.445000 609903 torch/distributed/run.py:766] +W0725 18:54:49.445000 609903 torch/distributed/run.py:766] ***************************************** +W0725 18:54:49.445000 609903 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:54:49.445000 609903 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:55:16.510159207 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:55:16.637611822 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:55:17.055786648 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:55:17.086531187 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:55:17.444535629 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:55:17.491697528 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:55:17.511882648 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W725 18:55:17.511922144 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n22_seed4_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed4_resampled_n116-041-188_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:36:27 acc1_pos: 92.969 acc5_pos: 100.000 loss: 0.2695 (0.2695) acc1: 92.9688 (92.9688) acc5: 100.0000 (100.0000) time: 6.5692 data: 4.9222 max mem: 3531 +Test: [ 10/333] eta: 0:03:41 acc1_pos: 92.708 acc5_pos: 100.000 loss: 0.2163 (0.2140) acc1: 93.4896 (93.5606) acc5: 100.0000 (100.0000) time: 0.6846 data: 0.4478 max mem: 3531 +Test: [ 20/333] eta: 0:02:07 acc1_pos: 98.958 acc5_pos: 100.000 loss: 0.1909 (0.2096) acc1: 93.4896 (93.6632) acc5: 100.0000 (100.0000) time: 0.0980 data: 0.0004 max mem: 3531 +Test: [ 30/333] eta: 0:01:46 acc1_pos: 81.250 acc5_pos: 100.000 loss: 0.2729 (0.2933) acc1: 86.9792 (88.6005) acc5: 100.0000 (100.0000) time: 0.1661 data: 0.0681 max mem: 3531 +Test: [ 40/333] eta: 0:01:38 acc1_pos: 75.781 acc5_pos: 100.000 loss: 0.4798 (0.3428) acc1: 77.3438 (85.7279) acc5: 100.0000 (100.0000) time: 0.2649 data: 0.1686 max mem: 3531 +Test: [ 50/333] eta: 0:01:32 acc1_pos: 79.948 acc5_pos: 100.000 loss: 0.4757 (0.3694) acc1: 77.3438 (84.2269) acc5: 100.0000 (100.0000) time: 0.2937 data: 0.1986 max mem: 3531 +Test: [ 60/333] eta: 0:01:23 acc1_pos: 82.292 acc5_pos: 100.000 loss: 0.4602 (0.3901) acc1: 77.8646 (82.9406) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1492 max mem: 3531 +Test: [ 70/333] eta: 0:01:17 acc1_pos: 73.698 acc5_pos: 100.000 loss: 0.4948 (0.4078) acc1: 76.8229 (81.9322) acc5: 100.0000 (100.0000) time: 0.2102 data: 0.1142 max mem: 3531 +Test: [ 80/333] eta: 0:01:11 acc1_pos: 84.375 acc5_pos: 100.000 loss: 0.4836 (0.4178) acc1: 75.5208 (81.2725) acc5: 100.0000 (100.0000) time: 0.2160 data: 0.1214 max mem: 3531 +Test: [ 90/333] eta: 0:01:05 acc1_pos: 81.771 acc5_pos: 100.000 loss: 0.4422 (0.4204) acc1: 81.7708 (81.3931) acc5: 100.0000 (100.0000) time: 0.1796 data: 0.0846 max mem: 3531 +Test: [100/333] eta: 0:00:59 acc1_pos: 90.885 acc5_pos: 100.000 loss: 0.4196 (0.4225) acc1: 82.8125 (81.4408) acc5: 100.0000 (100.0000) time: 0.1438 data: 0.0476 max mem: 3531 +Test: [110/333] eta: 0:00:55 acc1_pos: 89.062 acc5_pos: 100.000 loss: 0.4620 (0.4254) acc1: 80.2083 (81.2594) acc5: 100.0000 (100.0000) time: 0.1659 data: 0.0692 max mem: 3531 +Test: [120/333] eta: 0:00:54 acc1_pos: 84.115 acc5_pos: 100.000 loss: 0.4167 (0.4257) acc1: 80.4688 (81.3146) acc5: 100.0000 (100.0000) time: 0.2417 data: 0.1481 max mem: 3531 +Test: [130/333] eta: 0:00:50 acc1_pos: 94.271 acc5_pos: 100.000 loss: 0.3978 (0.4261) acc1: 82.5521 (81.4090) acc5: 100.0000 (100.0000) time: 0.2388 data: 0.1469 max mem: 3531 +Test: [140/333] eta: 0:00:46 acc1_pos: 89.062 acc5_pos: 100.000 loss: 0.4066 (0.4254) acc1: 84.8958 (81.5603) acc5: 100.0000 (100.0000) time: 0.1432 data: 0.0466 max mem: 3531 +Test: [150/333] eta: 0:00:42 acc1_pos: 82.031 acc5_pos: 100.000 loss: 0.4149 (0.4270) acc1: 82.5521 (81.5190) acc5: 100.0000 (100.0000) time: 0.1350 data: 0.0384 max mem: 3531 +Test: [160/333] eta: 0:00:40 acc1_pos: 68.421 acc5_pos: 100.000 loss: 0.4306 (0.4491) acc1: 81.5104 (80.0692) acc5: 100.0000 (100.0000) acc1_neg: 28.646 acc5_neg: 100.000 time: 0.1913 data: 0.0992 max mem: 3531 +Test: [170/333] eta: 0:00:37 acc1_pos: 68.421 acc5_pos: 100.000 loss: 0.8523 (0.4801) acc1: 47.6562 (77.8387) acc5: 100.0000 (100.0000) acc1_neg: 49.219 acc5_neg: 100.000 time: 0.2147 data: 0.1225 max mem: 3531 +Test: [180/333] eta: 0:00:34 acc1_pos: 68.421 acc5_pos: 100.000 loss: 0.8335 (0.4933) acc1: 49.7396 (76.6992) acc5: 100.0000 (100.0000) acc1_neg: 41.927 acc5_neg: 100.000 time: 0.1943 data: 0.1003 max mem: 3531 +Test: [190/333] eta: 0:00:32 acc1_pos: 68.421 acc5_pos: 100.000 loss: 0.6948 (0.5061) acc1: 58.0729 (75.6763) acc5: 100.0000 (100.0000) acc1_neg: 55.729 acc5_neg: 100.000 time: 0.1799 data: 0.0873 max mem: 3531 +Test: [200/333] eta: 0:00:29 acc1_pos: 68.421 acc5_pos: 100.000 loss: 0.7706 (0.5194) acc1: 55.4688 (74.6230) acc5: 100.0000 (100.0000) acc1_neg: 50.521 acc5_neg: 100.000 time: 0.1891 data: 0.0980 max mem: 3531 +Test: [210/333] eta: 0:00:27 acc1_pos: 68.421 acc5_pos: 100.000 loss: 0.8325 (0.5357) acc1: 50.5208 (73.4202) acc5: 100.0000 (100.0000) acc1_neg: 39.844 acc5_neg: 100.000 time: 0.1864 data: 0.0953 max mem: 3532 +Test: [220/333] eta: 0:00:25 acc1_pos: 68.421 acc5_pos: 100.000 loss: 0.8623 (0.5507) acc1: 49.2188 (72.2155) acc5: 100.0000 (100.0000) acc1_neg: 57.812 acc5_neg: 100.000 time: 0.2018 data: 0.1108 max mem: 3532 +Test: [230/333] eta: 0:00:22 acc1_pos: 68.421 acc5_pos: 100.000 loss: 0.6615 (0.5548) acc1: 55.7292 (71.7983) acc5: 100.0000 (100.0000) acc1_neg: 63.281 acc5_neg: 100.000 time: 0.2183 data: 0.1270 max mem: 3532 +Test: [240/333] eta: 0:00:20 acc1_pos: 68.421 acc5_pos: 100.000 loss: 0.6935 (0.5693) acc1: 54.9479 (70.4724) acc5: 100.0000 (100.0000) acc1_neg: 34.635 acc5_neg: 100.000 time: 0.1986 data: 0.1058 max mem: 3532 +Test: [250/333] eta: 0:00:18 acc1_pos: 68.421 acc5_pos: 100.000 loss: 0.9197 (0.5822) acc1: 40.1042 (69.4877) acc5: 100.0000 (100.0000) acc1_neg: 44.792 acc5_neg: 100.000 time: 0.1958 data: 0.1033 max mem: 3532 +Test: [260/333] eta: 0:00:15 acc1_pos: 68.421 acc5_pos: 100.000 loss: 0.8572 (0.5921) acc1: 47.9167 (68.7540) acc5: 100.0000 (100.0000) acc1_neg: 52.865 acc5_neg: 100.000 time: 0.2041 data: 0.1119 max mem: 3532 +Test: [270/333] eta: 0:00:13 acc1_pos: 68.421 acc5_pos: 100.000 loss: 0.7470 (0.5951) acc1: 55.9896 (68.5261) acc5: 100.0000 (100.0000) acc1_neg: 64.062 acc5_neg: 100.000 time: 0.1942 data: 0.1019 max mem: 3532 +Test: [280/333] eta: 0:00:11 acc1_pos: 68.421 acc5_pos: 100.000 loss: 0.7074 (0.6061) acc1: 59.3750 (67.6639) acc5: 100.0000 (100.0000) acc1_neg: 43.750 acc5_neg: 100.000 time: 0.1816 data: 0.0896 max mem: 3532 +Test: [290/333] eta: 0:00:09 acc1_pos: 68.421 acc5_pos: 100.000 loss: 1.0518 (0.6217) acc1: 35.4167 (66.5396) acc5: 100.0000 (100.0000) acc1_neg: 39.583 acc5_neg: 100.000 time: 0.1431 data: 0.0504 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 68.421 acc5_pos: 100.000 loss: 0.9972 (0.6333) acc1: 36.4583 (65.6778) acc5: 100.0000 (100.0000) acc1_neg: 38.542 acc5_neg: 100.000 time: 0.1254 data: 0.0335 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 68.421 acc5_pos: 100.000 loss: 0.9173 (0.6365) acc1: 42.9688 (65.3696) acc5: 100.0000 (100.0000) acc1_neg: 38.802 acc5_neg: 100.000 time: 0.1326 data: 0.0420 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 68.421 acc5_pos: 100.000 loss: 0.8838 (0.6456) acc1: 41.1458 (64.5704) acc5: 100.0000 (100.0000) acc1_neg: 44.010 acc5_neg: 100.000 time: 0.1193 data: 0.0294 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 68.421 acc5_pos: 100.000 loss: 0.9750 (0.6568) acc1: 39.5833 (63.8257) acc5: 100.0000 (100.0000) acc1_neg: 40.885 acc5_neg: 100.000 time: 0.1032 data: 0.0134 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 68.421 acc5_pos: 100.000 loss: 0.9888 (0.6587) acc1: 40.1042 (63.7031) acc5: 100.0000 (100.0000) acc1_neg: 42.254 acc5_neg: 100.000 time: 0.1119 data: 0.0189 max mem: 3532 +Test: Total time: 0:01:07 (0.2019 s / it) +* Overall Acc@1 63.703 Acc@5 100.000 + target == real Acc@1 81.552 Acc@5 100.000 + target == fake Acc@1 47.814 Acc@5 100.000 + loss 0.659 + PR-AUC 0.7478 ROC-AUC 0.7245 EER 33.93% (thr=0.3547) +Accuracy of the network on the 127843 test images: 63.7% +[rank0]:[W725 18:56:33.964028700 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:56:52.189000 610471 torch/distributed/run.py:766] +W0725 18:56:52.189000 610471 torch/distributed/run.py:766] ***************************************** +W0725 18:56:52.189000 610471 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:56:52.189000 610471 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +[rank0]:[W725 18:57:20.648430432 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:57:20.652157461 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:57:21.203810810 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:57:21.233040283 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:57:21.341358387 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:57:21.430706501 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:57:21.475441664 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:57:21.539943562 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n22_seed4_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed4_resampled_n116-041-188_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:24:48 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.1726 (0.1726) acc1: 96.8750 (96.8750) acc5: 100.0000 (100.0000) time: 6.3360 data: 4.6770 max mem: 3531 +Test: [ 10/235] eta: 0:02:29 acc1_pos: 96.615 acc5_pos: 100.000 loss: 0.1781 (0.1791) acc1: 96.6146 (96.6856) acc5: 100.0000 (100.0000) time: 0.6629 data: 0.4298 max mem: 3531 +Test: [ 20/235] eta: 0:01:24 acc1_pos: 97.135 acc5_pos: 100.000 loss: 0.1741 (0.1751) acc1: 96.8750 (97.0114) acc5: 100.0000 (100.0000) time: 0.0953 data: 0.0028 max mem: 3531 +Test: [ 30/235] eta: 0:01:07 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1675 (0.1737) acc1: 97.3958 (97.0598) acc5: 100.0000 (100.0000) time: 0.1424 data: 0.0473 max mem: 3531 +Test: [ 40/235] eta: 0:00:58 acc1_pos: 97.656 acc5_pos: 100.000 loss: 0.1705 (0.1742) acc1: 97.3958 (97.0211) acc5: 100.0000 (100.0000) time: 0.1999 data: 0.1043 max mem: 3531 +Test: [ 50/235] eta: 0:00:51 acc1_pos: 96.094 acc5_pos: 100.000 loss: 0.1757 (0.1753) acc1: 96.6146 (96.9618) acc5: 100.0000 (100.0000) time: 0.2049 data: 0.1097 max mem: 3531 +Test: [ 60/235] eta: 0:00:46 acc1_pos: 97.396 acc5_pos: 100.000 loss: 0.1785 (0.1763) acc1: 96.6146 (96.9134) acc5: 100.0000 (100.0000) time: 0.1914 data: 0.0985 max mem: 3531 +Test: [ 70/235] eta: 0:00:41 acc1_pos: 98.462 acc5_pos: 100.000 loss: 0.1943 (0.3306) acc1: 95.8333 (87.8484) acc5: 100.0000 (100.0000) acc1_neg: 23.698 acc5_neg: 100.000 time: 0.1786 data: 0.0855 max mem: 3531 +Test: [ 80/235] eta: 0:00:38 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4045 (0.4662) acc1: 24.4792 (79.9608) acc5: 100.0000 (100.0000) acc1_neg: 22.656 acc5_neg: 100.000 time: 0.1987 data: 0.1043 max mem: 3531 +Test: [ 90/235] eta: 0:00:35 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4277 (0.5738) acc1: 22.6562 (73.6865) acc5: 100.0000 (100.0000) acc1_neg: 22.135 acc5_neg: 100.000 time: 0.2087 data: 0.1149 max mem: 3531 +Test: [100/235] eta: 0:00:31 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4277 (0.6570) acc1: 22.3958 (68.7758) acc5: 100.0000 (100.0000) acc1_neg: 24.219 acc5_neg: 100.000 time: 0.1895 data: 0.0961 max mem: 3531 +Test: [110/235] eta: 0:00:29 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4172 (0.7265) acc1: 23.1771 (64.6866) acc5: 100.0000 (100.0000) acc1_neg: 23.177 acc5_neg: 100.000 time: 0.1938 data: 0.1011 max mem: 3531 +Test: [120/235] eta: 0:00:26 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4745 (0.7992) acc1: 20.5729 (60.6599) acc5: 100.0000 (100.0000) acc1_neg: 17.969 acc5_neg: 100.000 time: 0.1919 data: 0.1000 max mem: 3531 +Test: [130/235] eta: 0:00:23 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.5964 (0.8599) acc1: 15.8854 (57.3076) acc5: 100.0000 (100.0000) acc1_neg: 13.802 acc5_neg: 100.000 time: 0.1740 data: 0.0811 max mem: 3531 +Test: [140/235] eta: 0:00:20 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.6078 (0.9144) acc1: 15.6250 (54.3458) acc5: 100.0000 (100.0000) acc1_neg: 14.062 acc5_neg: 100.000 time: 0.1729 data: 0.0787 max mem: 3531 +Test: [150/235] eta: 0:00:18 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.6133 (0.9594) acc1: 15.6250 (51.8367) acc5: 100.0000 (100.0000) acc1_neg: 16.406 acc5_neg: 100.000 time: 0.1893 data: 0.0948 max mem: 3531 +Test: [160/235] eta: 0:00:16 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.5968 (0.9990) acc1: 16.1458 (49.6070) acc5: 100.0000 (100.0000) acc1_neg: 15.625 acc5_neg: 100.000 time: 0.1855 data: 0.0910 max mem: 3531 +Test: [170/235] eta: 0:00:13 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.5509 (1.0224) acc1: 17.9688 (48.0857) acc5: 100.0000 (100.0000) acc1_neg: 22.656 acc5_neg: 100.000 time: 0.1699 data: 0.0762 max mem: 3531 +Test: [180/235] eta: 0:00:11 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.3831 (1.0416) acc1: 23.6979 (46.7901) acc5: 100.0000 (100.0000) acc1_neg: 23.698 acc5_neg: 100.000 time: 0.1743 data: 0.0815 max mem: 3531 +Test: [190/235] eta: 0:00:09 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.3604 (1.0590) acc1: 24.2188 (45.6343) acc5: 100.0000 (100.0000) acc1_neg: 23.438 acc5_neg: 100.000 time: 0.1807 data: 0.0891 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.3865 (1.0763) acc1: 23.4375 (44.4717) acc5: 100.0000 (100.0000) acc1_neg: 20.833 acc5_neg: 100.000 time: 0.1999 data: 0.1074 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.3965 (1.0907) acc1: 22.1354 (43.5118) acc5: 100.0000 (100.0000) acc1_neg: 22.656 acc5_neg: 100.000 time: 0.2145 data: 0.1204 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.3560 (1.1029) acc1: 23.9583 (42.6636) acc5: 100.0000 (100.0000) acc1_neg: 25.000 acc5_neg: 100.000 time: 0.1772 data: 0.0838 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.3560 (1.1144) acc1: 25.0000 (41.9136) acc5: 100.0000 (100.0000) acc1_neg: 23.958 acc5_neg: 100.000 time: 0.1348 data: 0.0439 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.3685 (1.1201) acc1: 25.0000 (41.6771) acc5: 100.0000 (100.0000) acc1_neg: 13.043 acc5_neg: 100.000 time: 0.1071 data: 0.0196 max mem: 3532 +Test: Total time: 0:00:47 (0.2025 s / it) +* Overall Acc@1 41.677 Acc@5 100.000 + target == real Acc@1 96.917 Acc@5 100.000 + target == fake Acc@1 21.698 Acc@5 100.000 + loss 1.120 + PR-AUC 0.8886 ROC-AUC 0.7474 EER 31.82% (thr=0.1478) +Accuracy of the network on the 89879 test images: 41.7% +[rank0]:[W725 18:58:21.961063370 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:58:39.879000 611039 torch/distributed/run.py:766] +W0725 18:58:39.879000 611039 torch/distributed/run.py:766] ***************************************** +W0725 18:58:39.879000 611039 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:58:39.879000 611039 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:59:08.006596681 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:59:08.017320014 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:59:08.126830475 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:59:08.370406795 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:59:08.443311478 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:59:08.475224447 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +[rank0]:[W725 18:59:08.477072652 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:59:08.484661538 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n22_seed4_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed4_resampled_n116-041-188_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:22:31 acc1_pos: 97.396 acc5_pos: 100.000 loss: 0.2605 (0.2605) acc1: 97.3958 (97.3958) acc5: 100.0000 (100.0000) time: 5.0240 data: 3.3317 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/269] eta: 0:02:19 acc1_pos: 89.323 acc5_pos: 100.000 loss: 0.3761 (0.3424) acc1: 89.5833 (91.4773) acc5: 100.0000 (100.0000) time: 0.5388 data: 0.3032 max mem: 3531 +Test: [ 20/269] eta: 0:01:21 acc1_pos: 95.052 acc5_pos: 100.000 loss: 0.3569 (0.3638) acc1: 90.1042 (89.7942) acc5: 100.0000 (100.0000) time: 0.0940 data: 0.0004 max mem: 3531 +Test: [ 30/269] eta: 0:01:03 acc1_pos: 94.010 acc5_pos: 100.000 loss: 0.3275 (0.3565) acc1: 91.6667 (89.9866) acc5: 100.0000 (100.0000) time: 0.1138 data: 0.0168 max mem: 3531 +Test: [ 40/269] eta: 0:00:56 acc1_pos: 92.241 acc5_pos: 100.000 loss: 0.4547 (0.5074) acc1: 75.2604 (77.0198) acc5: 100.0000 (100.0000) acc1_neg: 23.438 acc5_neg: 100.000 time: 0.1564 data: 0.0608 max mem: 3531 +Test: [ 50/269] eta: 0:00:52 acc1_pos: 92.241 acc5_pos: 100.000 loss: 0.9789 (0.5877) acc1: 30.2083 (69.7968) acc5: 100.0000 (100.0000) acc1_neg: 25.260 acc5_neg: 100.000 time: 0.2049 data: 0.1127 max mem: 3531 +Test: [ 60/269] eta: 0:00:50 acc1_pos: 92.241 acc5_pos: 100.000 loss: 0.9876 (0.6680) acc1: 29.6875 (62.7818) acc5: 100.0000 (100.0000) acc1_neg: 10.938 acc5_neg: 100.000 time: 0.2334 data: 0.1398 max mem: 3531 +Test: [ 70/269] eta: 0:00:45 acc1_pos: 92.241 acc5_pos: 100.000 loss: 1.0351 (0.7157) acc1: 28.9062 (58.4397) acc5: 100.0000 (100.0000) acc1_neg: 29.688 acc5_neg: 100.000 time: 0.2049 data: 0.1095 max mem: 3531 +Test: [ 80/269] eta: 0:00:41 acc1_pos: 92.241 acc5_pos: 100.000 loss: 0.9525 (0.7392) acc1: 35.4167 (55.9285) acc5: 100.0000 (100.0000) acc1_neg: 41.927 acc5_neg: 100.000 time: 0.1570 data: 0.0645 max mem: 3531 +Test: [ 90/269] eta: 0:00:39 acc1_pos: 92.241 acc5_pos: 100.000 loss: 0.9771 (0.7692) acc1: 32.2917 (53.0105) acc5: 100.0000 (100.0000) acc1_neg: 15.885 acc5_neg: 100.000 time: 0.1911 data: 0.0999 max mem: 3531 +Test: [100/269] eta: 0:00:37 acc1_pos: 92.241 acc5_pos: 100.000 loss: 1.0090 (0.7980) acc1: 31.7708 (51.0417) acc5: 100.0000 (100.0000) acc1_neg: 39.062 acc5_neg: 100.000 time: 0.2155 data: 0.1248 max mem: 3531 +Test: [110/269] eta: 0:00:34 acc1_pos: 92.241 acc5_pos: 100.000 loss: 1.1834 (0.8373) acc1: 19.0104 (48.0997) acc5: 100.0000 (100.0000) acc1_neg: 14.844 acc5_neg: 100.000 time: 0.1958 data: 0.1036 max mem: 3531 +Test: [120/269] eta: 0:00:32 acc1_pos: 92.241 acc5_pos: 100.000 loss: 1.0499 (0.8434) acc1: 28.1250 (47.8456) acc5: 100.0000 (100.0000) acc1_neg: 36.719 acc5_neg: 100.000 time: 0.2073 data: 0.1146 max mem: 3531 +Test: [130/269] eta: 0:00:30 acc1_pos: 92.241 acc5_pos: 100.000 loss: 0.9820 (0.8531) acc1: 39.3229 (47.0599) acc5: 100.0000 (100.0000) acc1_neg: 39.323 acc5_neg: 100.000 time: 0.2290 data: 0.1379 max mem: 3531 +Test: [140/269] eta: 0:00:27 acc1_pos: 92.241 acc5_pos: 100.000 loss: 0.9564 (0.8594) acc1: 39.3229 (46.6848) acc5: 100.0000 (100.0000) acc1_neg: 37.500 acc5_neg: 100.000 time: 0.2049 data: 0.1137 max mem: 3531 +Test: [150/269] eta: 0:00:25 acc1_pos: 92.241 acc5_pos: 100.000 loss: 0.8808 (0.8561) acc1: 43.2292 (47.0216) acc5: 100.0000 (100.0000) acc1_neg: 53.906 acc5_neg: 100.000 time: 0.1960 data: 0.1046 max mem: 3531 +Test: [160/269] eta: 0:00:23 acc1_pos: 92.241 acc5_pos: 100.000 loss: 0.8808 (0.8608) acc1: 41.4062 (46.5078) acc5: 100.0000 (100.0000) acc1_neg: 28.125 acc5_neg: 100.000 time: 0.2384 data: 0.1461 max mem: 3531 +Test: [170/269] eta: 0:00:21 acc1_pos: 92.241 acc5_pos: 100.000 loss: 1.0058 (0.8738) acc1: 30.9896 (45.5090) acc5: 100.0000 (100.0000) acc1_neg: 36.198 acc5_neg: 100.000 time: 0.2279 data: 0.1352 max mem: 3531 +Test: [180/269] eta: 0:00:19 acc1_pos: 92.241 acc5_pos: 100.000 loss: 0.9654 (0.8778) acc1: 33.3333 (45.1528) acc5: 100.0000 (100.0000) acc1_neg: 38.021 acc5_neg: 100.000 time: 0.1987 data: 0.1066 max mem: 3531 +Test: [190/269] eta: 0:00:17 acc1_pos: 92.241 acc5_pos: 100.000 loss: 0.9485 (0.8826) acc1: 36.4583 (44.7140) acc5: 100.0000 (100.0000) acc1_neg: 36.458 acc5_neg: 100.000 time: 0.2247 data: 0.1321 max mem: 3531 +Test: [200/269] eta: 0:00:14 acc1_pos: 92.241 acc5_pos: 100.000 loss: 1.0707 (0.8972) acc1: 32.2917 (43.8200) acc5: 100.0000 (100.0000) acc1_neg: 44.271 acc5_neg: 100.000 time: 0.2225 data: 0.1264 max mem: 3531 +Test: [210/269] eta: 0:00:12 acc1_pos: 92.241 acc5_pos: 100.000 loss: 1.0707 (0.9004) acc1: 32.2917 (43.5180) acc5: 100.0000 (100.0000) acc1_neg: 47.917 acc5_neg: 100.000 time: 0.2069 data: 0.1115 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 92.241 acc5_pos: 100.000 loss: 0.9098 (0.8967) acc1: 38.8021 (43.8372) acc5: 100.0000 (100.0000) acc1_neg: 77.083 acc5_neg: 100.000 time: 0.2086 data: 0.1151 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 92.241 acc5_pos: 100.000 loss: 0.7121 (0.8876) acc1: 53.6458 (44.5143) acc5: 100.0000 (100.0000) acc1_neg: 47.656 acc5_neg: 100.000 time: 0.1981 data: 0.1029 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 92.241 acc5_pos: 100.000 loss: 0.8614 (0.8966) acc1: 39.3229 (43.8473) acc5: 100.0000 (100.0000) acc1_neg: 29.167 acc5_neg: 100.000 time: 0.1837 data: 0.0911 max mem: 3532 +Test: [250/269] eta: 0:00:04 acc1_pos: 92.241 acc5_pos: 100.000 loss: 1.0730 (0.9026) acc1: 29.1667 (43.5881) acc5: 100.0000 (100.0000) acc1_neg: 14.323 acc5_neg: 100.000 time: 0.1612 data: 0.0706 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 92.241 acc5_pos: 100.000 loss: 1.2056 (0.9158) acc1: 22.1354 (42.6594) acc5: 100.0000 (100.0000) acc1_neg: 27.083 acc5_neg: 100.000 time: 0.1573 data: 0.0670 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 92.241 acc5_pos: 100.000 loss: 1.1012 (0.9190) acc1: 22.1354 (42.3978) acc5: 100.0000 (100.0000) acc1_neg: 41.470 acc5_neg: 100.000 time: 0.1402 data: 0.0468 max mem: 3532 +Test: Total time: 0:00:55 (0.2069 s / it) +* Overall Acc@1 42.398 Acc@5 100.000 + target == real Acc@1 90.044 Acc@5 100.000 + target == fake Acc@1 35.895 Acc@5 100.000 + loss 0.919 + PR-AUC 0.9455 ROC-AUC 0.7113 EER 34.82% (thr=0.2981) +Accuracy of the network on the 103293 test images: 42.4% +[rank0]:[W725 19:00:12.170674041 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-25T19:00:17Z] eval_done tag=fixed500k_fake_n22_seed4_resampled +eval_queue_done_utc=2026-07-25T19:00:17Z diff --git a/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_112559/eval_queue_n116-041-188.pid b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_112559/eval_queue_n116-041-188.pid new file mode 100644 index 0000000000000000000000000000000000000000..70b3d99f89a450a2a5e4a7c18b623ac21cb347d3 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_112559/eval_queue_n116-041-188.pid @@ -0,0 +1 @@ +588536 diff --git a/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_112559/manual_recovery.log b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_112559/manual_recovery.log new file mode 100644 index 0000000000000000000000000000000000000000..919e8b717d55a2d5ef5b5f6458f4c50f8d1fd6a2 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_112559/manual_recovery.log @@ -0,0 +1 @@ +2026-07-25T17:15:06Z removing stale lock /mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed1_resampled/.eval.lock accuracy_lines=4 diff --git a/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_1958_seed5_9/eval_queue_n111-046-081.log b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_1958_seed5_9/eval_queue_n111-046-081.log new file mode 100644 index 0000000000000000000000000000000000000000..0fe4c5d8133fc09db88145235d897e0ed93f39ea --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_1958_seed5_9/eval_queue_n111-046-081.log @@ -0,0 +1,10270 @@ +eval_queue_started_utc=2026-07-26T02:59:38Z +hostname=n111-046-081 +train_tag=20260725_1958_seed5_9 +eval_tag=20260725_1958_seed5_9 +real_level=16 +fake_level=32 +eval_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9 +batch_per_gpu=256 +seeds=5 6 7 8 9 +[2026-07-26T02:59:38Z] eval_start tag=fixed500k_real_n16_seed5_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed5_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T02:59:38Z +hostname=n111-046-081 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed5_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n16_seed5_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 02:59:51.232000 3138788 torch/distributed/run.py:766] +W0726 02:59:51.232000 3138788 torch/distributed/run.py:766] ***************************************** +W0726 02:59:51.232000 3138788 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 02:59:51.232000 3138788 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:00:22.397466820 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:00:23.614475154 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:00:23.783085565 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:00:23.794887036 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +[rank6]:[W726 03:00:23.879801945 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:00:23.886448829 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:00:23.906228318 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:00:23.946217373 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n16_seed5_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed5_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:17:07 acc1_pos: 68.750 acc5_pos: 100.000 loss: 0.5270 (0.5270) acc1: 68.7500 (68.7500) acc5: 100.0000 (100.0000) time: 4.3914 data: 2.4409 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/234] eta: 0:02:01 acc1_pos: 99.740 acc5_pos: 100.000 loss: 0.3233 (0.3262) acc1: 92.9688 (89.6307) acc5: 100.0000 (100.0000) time: 0.5442 data: 0.2739 max mem: 3531 +Test: [ 20/234] eta: 0:01:10 acc1_pos: 72.396 acc5_pos: 100.000 loss: 0.4232 (0.4216) acc1: 81.5104 (81.7088) acc5: 100.0000 (100.0000) time: 0.1261 data: 0.0289 max mem: 3531 +Test: [ 30/234] eta: 0:00:52 acc1_pos: 85.417 acc5_pos: 100.000 loss: 0.5091 (0.4369) acc1: 74.4792 (80.8468) acc5: 100.0000 (100.0000) time: 0.1025 data: 0.0078 max mem: 3531 +Test: [ 40/234] eta: 0:00:48 acc1_pos: 90.625 acc5_pos: 100.000 loss: 0.4134 (0.4226) acc1: 83.5938 (82.0948) acc5: 100.0000 (100.0000) time: 0.1618 data: 0.0681 max mem: 3531 +Test: [ 50/234] eta: 0:00:43 acc1_pos: 90.365 acc5_pos: 100.000 loss: 0.2870 (0.3925) acc1: 90.6250 (83.9154) acc5: 100.0000 (100.0000) time: 0.1987 data: 0.1011 max mem: 3531 +Test: [ 60/234] eta: 0:00:40 acc1_pos: 89.655 acc5_pos: 100.000 loss: 0.1560 (0.3432) acc1: 92.9688 (86.4711) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1941 data: 0.0932 max mem: 3531 +Test: [ 70/234] eta: 0:00:36 acc1_pos: 89.655 acc5_pos: 100.000 loss: 0.0868 (0.3279) acc1: 99.7396 (87.2873) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1913 data: 0.0981 max mem: 3531 +Test: [ 80/234] eta: 0:00:33 acc1_pos: 89.655 acc5_pos: 100.000 loss: 0.2786 (0.3175) acc1: 89.8438 (87.9501) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1899 data: 0.0980 max mem: 3531 +Test: [ 90/234] eta: 0:00:31 acc1_pos: 89.655 acc5_pos: 100.000 loss: 0.1406 (0.2950) acc1: 98.1771 (89.1197) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1985 data: 0.1033 max mem: 3531 +Test: [100/234] eta: 0:00:28 acc1_pos: 89.655 acc5_pos: 100.000 loss: 0.2211 (0.2951) acc1: 92.9688 (89.0573) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1889 data: 0.0939 max mem: 3531 +Test: [110/234] eta: 0:00:25 acc1_pos: 89.655 acc5_pos: 100.000 loss: 0.2876 (0.2943) acc1: 88.8021 (89.1587) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1599 data: 0.0650 max mem: 3531 +Test: [120/234] eta: 0:00:23 acc1_pos: 89.655 acc5_pos: 100.000 loss: 0.3041 (0.3097) acc1: 88.0208 (88.1758) acc5: 100.0000 (100.0000) acc1_neg: 66.927 acc5_neg: 100.000 time: 0.1563 data: 0.0623 max mem: 3531 +Test: [130/234] eta: 0:00:21 acc1_pos: 89.655 acc5_pos: 100.000 loss: 0.5078 (0.3394) acc1: 68.2292 (85.8043) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.1799 data: 0.0852 max mem: 3531 +Test: [140/234] eta: 0:00:19 acc1_pos: 89.655 acc5_pos: 100.000 loss: 0.5640 (0.3538) acc1: 68.2292 (85.0602) acc5: 100.0000 (100.0000) acc1_neg: 46.875 acc5_neg: 100.000 time: 0.1935 data: 0.0978 max mem: 3531 +Test: [150/234] eta: 0:00:16 acc1_pos: 89.655 acc5_pos: 100.000 loss: 0.3448 (0.3557) acc1: 85.4167 (84.9838) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1749 data: 0.0789 max mem: 3531 +Test: [160/234] eta: 0:00:14 acc1_pos: 89.655 acc5_pos: 100.000 loss: 0.3387 (0.3677) acc1: 85.4167 (84.0790) acc5: 100.0000 (100.0000) acc1_neg: 55.469 acc5_neg: 100.000 time: 0.1299 data: 0.0324 max mem: 3531 +Test: [170/234] eta: 0:00:12 acc1_pos: 89.655 acc5_pos: 100.000 loss: 0.7333 (0.3909) acc1: 55.4688 (82.3145) acc5: 100.0000 (100.0000) acc1_neg: 43.229 acc5_neg: 100.000 time: 0.1799 data: 0.0842 max mem: 3531 +Test: [180/234] eta: 0:00:10 acc1_pos: 89.655 acc5_pos: 100.000 loss: 0.7127 (0.3982) acc1: 58.5938 (81.9104) acc5: 100.0000 (100.0000) acc1_neg: 58.594 acc5_neg: 100.000 time: 0.2099 data: 0.1155 max mem: 3531 +Test: [190/234] eta: 0:00:08 acc1_pos: 89.655 acc5_pos: 100.000 loss: 0.4805 (0.4035) acc1: 77.3438 (81.6236) acc5: 100.0000 (100.0000) acc1_neg: 78.646 acc5_neg: 100.000 time: 0.1699 data: 0.0758 max mem: 3531 +Test: [200/234] eta: 0:00:06 acc1_pos: 89.655 acc5_pos: 100.000 loss: 0.4698 (0.3959) acc1: 77.8646 (82.0973) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1699 data: 0.0719 max mem: 3531 +Test: [210/234] eta: 0:00:04 acc1_pos: 89.655 acc5_pos: 100.000 loss: 0.1092 (0.3819) acc1: 99.4792 (82.9310) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1749 data: 0.0741 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 89.655 acc5_pos: 100.000 loss: 0.0865 (0.3689) acc1: 99.7396 (83.6762) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1699 data: 0.0711 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 89.655 acc5_pos: 100.000 loss: 0.1790 (0.3684) acc1: 90.6250 (83.8474) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.1754 data: 0.0800 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 89.655 acc5_pos: 100.000 loss: 0.3476 (0.3684) acc1: 88.5417 (83.8804) acc5: 100.0000 (100.0000) acc1_neg: 89.105 acc5_neg: 100.000 time: 0.1449 data: 0.0492 max mem: 3532 +Test: Total time: 0:00:44 (0.1905 s / it) +* Overall Acc@1 83.880 Acc@5 100.000 + target == real Acc@1 83.958 Acc@5 100.000 + target == fake Acc@1 83.859 Acc@5 100.000 + loss 0.368 + PR-AUC 0.9755 ROC-AUC 0.9181 EER 16.09% (thr=0.4990) +Accuracy of the network on the 89729 test images: 83.9% +[rank0]:[W726 03:01:19.496947315 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:01:38.038000 3139336 torch/distributed/run.py:766] +W0726 03:01:38.038000 3139336 torch/distributed/run.py:766] ***************************************** +W0726 03:01:38.038000 3139336 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:01:38.038000 3139336 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:02:06.280718782 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:02:07.950437605 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:02:07.005551855 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +[rank6]:[W726 03:02:07.056675325 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:02:07.064582869 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:02:07.260457212 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +[rank0]:[W726 03:02:08.574617265 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:02:08.581677599 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n16_seed5_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed5_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 3:58:46 acc1_pos: 61.979 acc5_pos: 100.000 loss: 0.6331 (0.6331) acc1: 61.9792 (61.9792) acc5: 100.0000 (100.0000) time: 6.0018 data: 4.2259 max mem: 3531 +Test: [ 10/2387] eta: 0:24:51 acc1_pos: 56.510 acc5_pos: 100.000 loss: 0.7408 (0.7674) acc1: 59.1146 (52.4621) acc5: 100.0000 (100.0000) time: 0.6273 data: 0.3846 max mem: 3531 +Test: [ 20/2387] eta: 0:15:11 acc1_pos: 71.354 acc5_pos: 100.000 loss: 0.6248 (0.6971) acc1: 64.0625 (60.7267) acc5: 100.0000 (100.0000) time: 0.1042 data: 0.0126 max mem: 3531 +Test: [ 30/2387] eta: 0:11:53 acc1_pos: 17.969 acc5_pos: 100.000 loss: 0.6600 (0.7463) acc1: 54.1667 (54.3347) acc5: 100.0000 (100.0000) time: 0.1242 data: 0.0290 max mem: 3531 +Test: [ 40/2387] eta: 0:11:20 acc1_pos: 55.729 acc5_pos: 100.000 loss: 0.8749 (0.7776) acc1: 35.9375 (50.7495) acc5: 100.0000 (100.0000) time: 0.1901 data: 0.0961 max mem: 3531 +Test: [ 50/2387] eta: 0:10:54 acc1_pos: 58.854 acc5_pos: 100.000 loss: 0.7502 (0.7555) acc1: 52.3438 (53.0280) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1537 max mem: 3531 +Test: [ 60/2387] eta: 0:10:29 acc1_pos: 68.490 acc5_pos: 100.000 loss: 0.7028 (0.7472) acc1: 58.8542 (54.1154) acc5: 100.0000 (100.0000) time: 0.2302 data: 0.1389 max mem: 3531 +Test: [ 70/2387] eta: 0:10:13 acc1_pos: 58.333 acc5_pos: 100.000 loss: 0.6812 (0.7466) acc1: 59.1146 (54.5298) acc5: 100.0000 (100.0000) time: 0.2255 data: 0.1336 max mem: 3531 +Test: [ 80/2387] eta: 0:10:06 acc1_pos: 35.938 acc5_pos: 100.000 loss: 0.7348 (0.7563) acc1: 52.6042 (53.5847) acc5: 100.0000 (100.0000) time: 0.2395 data: 0.1469 max mem: 3531 +Test: [ 90/2387] eta: 0:09:55 acc1_pos: 36.458 acc5_pos: 100.000 loss: 0.8177 (0.7624) acc1: 45.3125 (52.7444) acc5: 100.0000 (100.0000) time: 0.2394 data: 0.1458 max mem: 3531 +Test: [ 100/2387] eta: 0:09:46 acc1_pos: 63.802 acc5_pos: 100.000 loss: 0.8494 (0.7761) acc1: 42.9688 (51.7456) acc5: 100.0000 (100.0000) time: 0.2303 data: 0.1384 max mem: 3531 +Test: [ 110/2387] eta: 0:09:38 acc1_pos: 67.708 acc5_pos: 100.000 loss: 0.8169 (0.7742) acc1: 43.7500 (51.6822) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1382 max mem: 3531 +Test: [ 120/2387] eta: 0:09:29 acc1_pos: 65.104 acc5_pos: 100.000 loss: 0.7363 (0.7682) acc1: 53.1250 (52.0726) acc5: 100.0000 (100.0000) time: 0.2245 data: 0.1291 max mem: 3531 +Test: [ 130/2387] eta: 0:09:26 acc1_pos: 63.802 acc5_pos: 100.000 loss: 0.7363 (0.7745) acc1: 51.0417 (51.7434) acc5: 100.0000 (100.0000) time: 0.2349 data: 0.1393 max mem: 3531 +Test: [ 140/2387] eta: 0:09:23 acc1_pos: 74.479 acc5_pos: 100.000 loss: 0.7812 (0.7750) acc1: 51.0417 (52.0963) acc5: 100.0000 (100.0000) time: 0.2499 data: 0.1563 max mem: 3531 +Test: [ 150/2387] eta: 0:09:21 acc1_pos: 28.906 acc5_pos: 100.000 loss: 0.7757 (0.7793) acc1: 51.0417 (51.7039) acc5: 100.0000 (100.0000) time: 0.2499 data: 0.1538 max mem: 3531 +Test: [ 160/2387] eta: 0:09:15 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.7295 (0.7659) acc1: 52.6042 (52.7756) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2404 data: 0.1459 max mem: 3531 +Test: [ 170/2387] eta: 0:08:58 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2285 (0.7342) acc1: 92.9688 (55.1505) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1849 data: 0.0920 max mem: 3531 +Test: [ 180/2387] eta: 0:08:50 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2285 (0.7065) acc1: 93.2292 (57.2413) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1644 data: 0.0712 max mem: 3531 +Test: [ 190/2387] eta: 0:08:45 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2213 (0.6812) acc1: 93.2292 (59.1132) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.2049 data: 0.1111 max mem: 3531 +Test: [ 200/2387] eta: 0:08:37 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2213 (0.6584) acc1: 92.9688 (60.8144) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.2049 data: 0.1085 max mem: 3531 +Test: [ 210/2387] eta: 0:08:32 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2259 (0.6380) acc1: 92.7083 (62.3519) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1999 data: 0.1025 max mem: 3532 +Test: [ 220/2387] eta: 0:08:28 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2283 (0.6194) acc1: 92.9688 (63.7526) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2149 data: 0.1191 max mem: 3532 +Test: [ 230/2387] eta: 0:08:19 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2278 (0.6023) acc1: 92.9688 (65.0207) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1899 data: 0.0947 max mem: 3532 +Test: [ 240/2387] eta: 0:08:09 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2277 (0.5868) acc1: 93.4896 (66.2226) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1521 data: 0.0568 max mem: 3532 +Test: [ 250/2387] eta: 0:08:05 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2084 (0.5691) acc1: 95.5729 (67.5029) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1749 data: 0.0787 max mem: 3532 +Test: [ 260/2387] eta: 0:08:01 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1273 (0.5522) acc1: 98.6979 (68.6951) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2086 data: 0.1084 max mem: 3532 +Test: [ 270/2387] eta: 0:07:54 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1251 (0.5363) acc1: 98.4375 (69.7946) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1899 data: 0.0913 max mem: 3532 +Test: [ 280/2387] eta: 0:07:46 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1252 (0.5218) acc1: 98.6979 (70.8074) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1560 data: 0.0611 max mem: 3532 +Test: [ 290/2387] eta: 0:07:42 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1313 (0.5083) acc1: 98.9583 (71.7712) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1723 data: 0.0793 max mem: 3532 +Test: [ 300/2387] eta: 0:07:38 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1172 (0.4953) acc1: 99.2188 (72.6813) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1988 data: 0.1064 max mem: 3532 +Test: [ 310/2387] eta: 0:07:37 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1114 (0.4832) acc1: 99.4792 (73.5271) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2126 data: 0.1208 max mem: 3532 +Test: [ 320/2387] eta: 0:07:34 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1117 (0.4717) acc1: 99.7396 (74.3453) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2191 data: 0.1285 max mem: 3532 +Test: [ 330/2387] eta: 0:07:30 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1115 (0.4607) acc1: 100.0000 (75.1188) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2005 data: 0.1087 max mem: 3532 +Test: [ 340/2387] eta: 0:07:25 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1101 (0.4505) acc1: 100.0000 (75.8469) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1848 data: 0.0924 max mem: 3532 +Test: [ 350/2387] eta: 0:07:20 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1068 (0.4401) acc1: 100.0000 (76.5351) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1746 data: 0.0836 max mem: 3532 +Test: [ 360/2387] eta: 0:07:16 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.0906 (0.4331) acc1: 100.0000 (77.0603) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1726 data: 0.0787 max mem: 3532 +Test: [ 370/2387] eta: 0:07:11 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1951 (0.4269) acc1: 94.5312 (77.5284) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1703 data: 0.0757 max mem: 3532 +Test: [ 380/2387] eta: 0:07:07 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1966 (0.4210) acc1: 94.2708 (77.9698) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1688 data: 0.0777 max mem: 3532 +Test: [ 390/2387] eta: 0:07:04 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1987 (0.4153) acc1: 94.7917 (78.4101) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1907 data: 0.1004 max mem: 3532 +Test: [ 400/2387] eta: 0:07:03 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1997 (0.4100) acc1: 95.0521 (78.8179) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2182 data: 0.1249 max mem: 3532 +Test: [ 410/2387] eta: 0:06:56 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1997 (0.4049) acc1: 94.7917 (79.2022) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1683 data: 0.0710 max mem: 3532 +Test: [ 420/2387] eta: 0:06:52 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1970 (0.3999) acc1: 94.7917 (79.5811) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1467 data: 0.0504 max mem: 3532 +Test: [ 430/2387] eta: 0:06:49 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2024 (0.3955) acc1: 95.0521 (79.9376) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1860 data: 0.0935 max mem: 3532 +Test: [ 440/2387] eta: 0:06:49 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2062 (0.3898) acc1: 95.3125 (80.3430) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2238 data: 0.1330 max mem: 3533 +Test: [ 450/2387] eta: 0:06:46 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.0929 (0.3832) acc1: 100.0000 (80.7782) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2238 data: 0.1327 max mem: 3533 +Test: [ 460/2387] eta: 0:06:41 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.0915 (0.3768) acc1: 100.0000 (81.1946) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1613 data: 0.0675 max mem: 3533 +Test: [ 470/2387] eta: 0:06:39 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.0922 (0.3708) acc1: 100.0000 (81.5939) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1701 data: 0.0745 max mem: 3533 +Test: [ 480/2387] eta: 0:06:38 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.0924 (0.3650) acc1: 100.0000 (81.9760) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2228 data: 0.1272 max mem: 3533 +Test: [ 490/2387] eta: 0:06:34 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.0921 (0.3595) acc1: 100.0000 (82.3421) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1974 data: 0.1032 max mem: 3533 +Test: [ 500/2387] eta: 0:06:31 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.0920 (0.3541) acc1: 100.0000 (82.6945) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1760 data: 0.0841 max mem: 3533 +Test: [ 510/2387] eta: 0:06:29 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.0927 (0.3492) acc1: 100.0000 (83.0311) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1934 data: 0.1027 max mem: 3533 +Test: [ 520/2387] eta: 0:06:26 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1161 (0.3448) acc1: 100.0000 (83.3523) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1854 data: 0.0933 max mem: 3533 +Test: [ 530/2387] eta: 0:06:23 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1241 (0.3407) acc1: 99.7396 (83.6590) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1837 data: 0.0895 max mem: 3533 +Test: [ 540/2387] eta: 0:06:20 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1279 (0.3368) acc1: 99.4792 (83.9524) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1891 data: 0.0968 max mem: 3533 +Test: [ 550/2387] eta: 0:06:17 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1285 (0.3330) acc1: 99.4792 (84.2323) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1759 data: 0.0823 max mem: 3533 +Test: [ 560/2387] eta: 0:06:16 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1291 (0.3294) acc1: 99.4792 (84.5013) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1951 data: 0.1014 max mem: 3533 +Test: [ 570/2387] eta: 0:06:14 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1274 (0.3259) acc1: 99.4792 (84.7636) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2189 data: 0.1267 max mem: 3533 +Test: [ 580/2387] eta: 0:06:10 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1277 (0.3225) acc1: 99.4792 (85.0164) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1787 data: 0.0829 max mem: 3533 +Test: [ 590/2387] eta: 0:06:06 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1305 (0.3193) acc1: 99.4792 (85.2602) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1565 data: 0.0610 max mem: 3533 +Test: [ 600/2387] eta: 0:06:04 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1221 (0.3159) acc1: 99.7396 (85.5025) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1751 data: 0.0821 max mem: 3533 +Test: [ 610/2387] eta: 0:06:01 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1221 (0.3128) acc1: 99.4792 (85.7265) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1790 data: 0.0862 max mem: 3533 +Test: [ 620/2387] eta: 0:05:58 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1281 (0.3099) acc1: 98.9583 (85.9430) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1813 data: 0.0885 max mem: 3533 +Test: [ 630/2387] eta: 0:05:55 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1299 (0.3071) acc1: 98.9583 (86.1488) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1794 data: 0.0867 max mem: 3533 +Test: [ 640/2387] eta: 0:05:53 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1299 (0.3043) acc1: 98.9583 (86.3495) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1692 data: 0.0776 max mem: 3533 +Test: [ 650/2387] eta: 0:05:50 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1297 (0.3016) acc1: 99.2188 (86.5495) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1771 data: 0.0861 max mem: 3533 +Test: [ 660/2387] eta: 0:05:48 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1283 (0.2990) acc1: 99.2188 (86.7420) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1842 data: 0.0905 max mem: 3533 +Test: [ 670/2387] eta: 0:05:45 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1279 (0.2965) acc1: 99.2188 (86.9283) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1794 data: 0.0854 max mem: 3534 +Test: [ 680/2387] eta: 0:05:42 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1254 (0.2938) acc1: 99.4792 (87.1176) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1748 data: 0.0828 max mem: 3534 +Test: [ 690/2387] eta: 0:05:40 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1272 (0.2934) acc1: 99.4792 (87.1729) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1783 data: 0.0863 max mem: 3534 +Test: [ 700/2387] eta: 0:05:37 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.3133 (0.2937) acc1: 87.7604 (87.1720) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1823 data: 0.0912 max mem: 3534 +Test: [ 710/2387] eta: 0:05:35 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.3141 (0.2941) acc1: 87.5000 (87.1784) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.1812 data: 0.0887 max mem: 3534 +Test: [ 720/2387] eta: 0:05:33 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.3151 (0.2943) acc1: 87.7604 (87.1926) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1850 data: 0.0915 max mem: 3534 +Test: [ 730/2387] eta: 0:05:30 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.3184 (0.2948) acc1: 87.7604 (87.1879) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1734 data: 0.0814 max mem: 3534 +Test: [ 740/2387] eta: 0:05:27 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.3204 (0.2952) acc1: 86.9792 (87.1844) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.1631 data: 0.0720 max mem: 3534 +Test: [ 750/2387] eta: 0:05:24 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.3149 (0.2954) acc1: 87.5000 (87.1897) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.1666 data: 0.0756 max mem: 3534 +Test: [ 760/2387] eta: 0:05:22 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.3073 (0.2951) acc1: 88.0208 (87.2297) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1704 data: 0.0782 max mem: 3534 +Test: [ 770/2387] eta: 0:05:19 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2205 (0.2940) acc1: 94.0104 (87.3328) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1682 data: 0.0761 max mem: 3534 +Test: [ 780/2387] eta: 0:05:16 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2148 (0.2932) acc1: 94.7917 (87.4036) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1660 data: 0.0746 max mem: 3534 +Test: [ 790/2387] eta: 0:05:14 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2260 (0.2924) acc1: 91.9271 (87.4628) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1675 data: 0.0758 max mem: 3534 +Test: [ 800/2387] eta: 0:05:11 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2238 (0.2916) acc1: 91.6667 (87.5257) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1575 data: 0.0653 max mem: 3534 +Test: [ 810/2387] eta: 0:05:09 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2069 (0.2906) acc1: 94.0104 (87.6172) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1625 data: 0.0696 max mem: 3534 +Test: [ 820/2387] eta: 0:05:06 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1958 (0.2895) acc1: 96.0938 (87.7297) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1719 data: 0.0786 max mem: 3534 +Test: [ 830/2387] eta: 0:05:04 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2135 (0.2890) acc1: 94.5312 (87.7864) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1726 data: 0.0780 max mem: 3534 +Test: [ 840/2387] eta: 0:05:02 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1800 (0.2874) acc1: 96.3542 (87.9162) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1828 data: 0.0868 max mem: 3534 +Test: [ 850/2387] eta: 0:05:00 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1557 (0.2859) acc1: 98.9583 (88.0398) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1880 data: 0.0939 max mem: 3534 +Test: [ 860/2387] eta: 0:04:58 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1440 (0.2841) acc1: 99.2188 (88.1742) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1949 data: 0.1023 max mem: 3534 +Test: [ 870/2387] eta: 0:04:57 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1346 (0.2824) acc1: 99.4792 (88.2968) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2203 data: 0.1278 max mem: 3534 +Test: [ 880/2387] eta: 0:04:54 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1308 (0.2807) acc1: 99.4792 (88.4231) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1999 data: 0.1074 max mem: 3534 +Test: [ 890/2387] eta: 0:04:52 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1256 (0.2790) acc1: 99.7396 (88.5458) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1885 data: 0.0942 max mem: 3534 +Test: [ 900/2387] eta: 0:04:50 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1363 (0.2785) acc1: 99.2188 (88.6252) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1909 data: 0.0968 max mem: 3535 +Test: [ 910/2387] eta: 0:04:48 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2599 (0.2783) acc1: 95.3125 (88.6972) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1699 data: 0.0774 max mem: 3535 +Test: [ 920/2387] eta: 0:04:45 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2613 (0.2781) acc1: 95.0521 (88.7619) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1750 data: 0.0821 max mem: 3535 +Test: [ 930/2387] eta: 0:04:43 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2613 (0.2780) acc1: 94.7917 (88.8242) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1774 data: 0.0846 max mem: 3535 +Test: [ 940/2387] eta: 0:04:41 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2643 (0.2778) acc1: 94.7917 (88.8859) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1738 data: 0.0820 max mem: 3535 +Test: [ 950/2387] eta: 0:04:39 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2612 (0.2776) acc1: 95.3125 (88.9563) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1760 data: 0.0816 max mem: 3535 +Test: [ 960/2387] eta: 0:04:36 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2525 (0.2774) acc1: 95.5729 (89.0232) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1736 data: 0.0783 max mem: 3535 +Test: [ 970/2387] eta: 0:04:34 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2552 (0.2772) acc1: 95.5729 (89.0923) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1676 data: 0.0739 max mem: 3535 +Test: [ 980/2387] eta: 0:04:32 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2552 (0.2767) acc1: 95.3125 (89.1451) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1703 data: 0.0765 max mem: 3535 +Test: [ 990/2387] eta: 0:04:30 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2212 (0.2768) acc1: 93.4896 (89.1547) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1773 data: 0.0839 max mem: 3535 +Test: [1000/2387] eta: 0:04:27 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2212 (0.2762) acc1: 93.2292 (89.2032) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1724 data: 0.0789 max mem: 3535 +Test: [1010/2387] eta: 0:04:26 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2309 (0.2758) acc1: 93.4896 (89.2449) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1980 data: 0.1025 max mem: 3535 +Test: [1020/2387] eta: 0:04:24 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2210 (0.2753) acc1: 94.7917 (89.2895) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2025 data: 0.1087 max mem: 3535 +Test: [1030/2387] eta: 0:04:21 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2184 (0.2747) acc1: 95.0521 (89.3328) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1526 data: 0.0603 max mem: 3535 +Test: [1040/2387] eta: 0:04:19 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2034 (0.2741) acc1: 95.3125 (89.3847) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1519 data: 0.0598 max mem: 3535 +Test: [1050/2387] eta: 0:04:16 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2034 (0.2736) acc1: 95.0521 (89.4307) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1668 data: 0.0766 max mem: 3535 +Test: [1060/2387] eta: 0:04:14 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2114 (0.2731) acc1: 94.2708 (89.4778) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1735 data: 0.0815 max mem: 3535 +Test: [1070/2387] eta: 0:04:12 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2029 (0.2726) acc1: 95.0521 (89.5225) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.1829 data: 0.0902 max mem: 3535 +Test: [1080/2387] eta: 0:04:10 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2483 (0.2726) acc1: 91.9271 (89.5279) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1702 data: 0.0775 max mem: 3535 +Test: [1090/2387] eta: 0:04:08 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2864 (0.2731) acc1: 89.0625 (89.4919) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.1664 data: 0.0734 max mem: 3535 +Test: [1100/2387] eta: 0:04:06 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.3226 (0.2739) acc1: 84.1146 (89.4353) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1752 data: 0.0839 max mem: 3535 +Test: [1110/2387] eta: 0:04:04 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.3526 (0.2748) acc1: 82.0312 (89.3703) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.1908 data: 0.0974 max mem: 3535 +Test: [1120/2387] eta: 0:04:02 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.3378 (0.2752) acc1: 82.5521 (89.3317) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1752 data: 0.0801 max mem: 3536 +Test: [1130/2387] eta: 0:03:59 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.3009 (0.2756) acc1: 85.1562 (89.3033) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1434 data: 0.0513 max mem: 3536 +Test: [1140/2387] eta: 0:03:57 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2871 (0.2757) acc1: 88.0208 (89.3056) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1547 data: 0.0646 max mem: 3536 +Test: [1150/2387] eta: 0:03:55 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2465 (0.2749) acc1: 93.7500 (89.3603) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1647 data: 0.0741 max mem: 3536 +Test: [1160/2387] eta: 0:03:53 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1086 (0.2735) acc1: 100.0000 (89.4519) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1659 data: 0.0745 max mem: 3536 +Test: [1170/2387] eta: 0:03:51 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1044 (0.2721) acc1: 100.0000 (89.5417) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0842 max mem: 3536 +Test: [1180/2387] eta: 0:03:49 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1050 (0.2707) acc1: 100.0000 (89.6288) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1706 data: 0.0797 max mem: 3536 +Test: [1190/2387] eta: 0:03:47 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1060 (0.2693) acc1: 100.0000 (89.7158) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1721 data: 0.0791 max mem: 3536 +Test: [1200/2387] eta: 0:03:44 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1050 (0.2679) acc1: 100.0000 (89.8008) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1759 data: 0.0830 max mem: 3536 +Test: [1210/2387] eta: 0:03:43 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1020 (0.2666) acc1: 100.0000 (89.8797) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1767 data: 0.0849 max mem: 3536 +Test: [1220/2387] eta: 0:03:40 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1443 (0.2658) acc1: 97.9167 (89.9331) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1782 data: 0.0873 max mem: 3536 +Test: [1230/2387] eta: 0:03:38 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1553 (0.2648) acc1: 97.1354 (89.9889) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1724 data: 0.0808 max mem: 3536 +Test: [1240/2387] eta: 0:03:36 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1485 (0.2639) acc1: 97.1354 (90.0465) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1719 data: 0.0798 max mem: 3536 +Test: [1250/2387] eta: 0:03:34 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1483 (0.2630) acc1: 97.1354 (90.1052) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1695 data: 0.0791 max mem: 3536 +Test: [1260/2387] eta: 0:03:32 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1442 (0.2620) acc1: 97.3958 (90.1645) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1676 data: 0.0753 max mem: 3536 +Test: [1270/2387] eta: 0:03:30 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1407 (0.2610) acc1: 97.6562 (90.2273) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1731 data: 0.0789 max mem: 3536 +Test: [1280/2387] eta: 0:03:28 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1345 (0.2600) acc1: 98.6979 (90.2932) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1685 data: 0.0760 max mem: 3536 +Test: [1290/2387] eta: 0:03:26 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1415 (0.2592) acc1: 97.9167 (90.3452) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1677 data: 0.0762 max mem: 3536 +Test: [1300/2387] eta: 0:03:24 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1659 (0.2586) acc1: 95.8333 (90.3886) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1740 data: 0.0804 max mem: 3536 +Test: [1310/2387] eta: 0:03:22 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1614 (0.2579) acc1: 96.3542 (90.4383) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1689 data: 0.0749 max mem: 3536 +Test: [1320/2387] eta: 0:03:20 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1607 (0.2571) acc1: 98.1771 (90.5002) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1672 data: 0.0735 max mem: 3536 +Test: [1330/2387] eta: 0:03:18 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1721 (0.2565) acc1: 98.1771 (90.5467) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1627 data: 0.0690 max mem: 3536 +Test: [1340/2387] eta: 0:03:16 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1876 (0.2567) acc1: 95.8333 (90.5401) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1663 data: 0.0725 max mem: 3536 +Test: [1350/2387] eta: 0:03:14 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.0704 (0.2553) acc1: 100.0000 (90.6102) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1830 data: 0.0898 max mem: 3537 +Test: [1360/2387] eta: 0:03:12 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.0587 (0.2538) acc1: 100.0000 (90.6792) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1844 data: 0.0917 max mem: 3537 +Test: [1370/2387] eta: 0:03:10 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.0575 (0.2524) acc1: 100.0000 (90.7468) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1760 data: 0.0819 max mem: 3537 +Test: [1380/2387] eta: 0:03:08 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.0600 (0.2510) acc1: 100.0000 (90.8138) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0807 max mem: 3537 +Test: [1390/2387] eta: 0:03:07 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.0607 (0.2497) acc1: 100.0000 (90.8798) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2078 data: 0.1119 max mem: 3537 +Test: [1400/2387] eta: 0:03:05 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.0643 (0.2484) acc1: 100.0000 (90.9449) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1952 data: 0.0982 max mem: 3537 +Test: [1410/2387] eta: 0:03:02 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.0705 (0.2481) acc1: 100.0000 (90.9511) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.1471 data: 0.0533 max mem: 3537 +Test: [1420/2387] eta: 0:03:00 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2777 (0.2489) acc1: 87.7604 (90.8997) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1642 data: 0.0733 max mem: 3537 +Test: [1430/2387] eta: 0:02:59 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.3220 (0.2493) acc1: 85.1562 (90.8701) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1948 data: 0.1035 max mem: 3537 +Test: [1440/2387] eta: 0:02:57 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.3088 (0.2496) acc1: 87.2396 (90.8509) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.2043 data: 0.1132 max mem: 3537 +Test: [1450/2387] eta: 0:02:55 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2536 (0.2494) acc1: 91.9271 (90.8743) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1846 data: 0.0924 max mem: 3537 +Test: [1460/2387] eta: 0:02:53 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1693 (0.2489) acc1: 96.3542 (90.9109) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1705 data: 0.0780 max mem: 3537 +Test: [1470/2387] eta: 0:02:51 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1571 (0.2483) acc1: 96.6146 (90.9437) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1786 data: 0.0876 max mem: 3537 +Test: [1480/2387] eta: 0:02:49 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1608 (0.2477) acc1: 96.3542 (90.9809) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1848 data: 0.0935 max mem: 3537 +Test: [1490/2387] eta: 0:02:47 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1501 (0.2470) acc1: 97.1354 (91.0243) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1893 data: 0.0940 max mem: 3537 +Test: [1500/2387] eta: 0:02:46 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1365 (0.2463) acc1: 98.4375 (91.0756) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1921 data: 0.0973 max mem: 3537 +Test: [1510/2387] eta: 0:02:44 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1160 (0.2454) acc1: 99.4792 (91.1314) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1969 data: 0.1034 max mem: 3537 +Test: [1520/2387] eta: 0:02:42 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1128 (0.2446) acc1: 99.4792 (91.1840) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1871 data: 0.0916 max mem: 3537 +Test: [1530/2387] eta: 0:02:40 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1305 (0.2439) acc1: 99.2188 (91.2367) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1778 data: 0.0844 max mem: 3537 +Test: [1540/2387] eta: 0:02:38 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1276 (0.2431) acc1: 99.4792 (91.2910) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1931 data: 0.1006 max mem: 3537 +Test: [1550/2387] eta: 0:02:36 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1141 (0.2423) acc1: 99.7396 (91.3448) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2077 data: 0.1143 max mem: 3537 +Test: [1560/2387] eta: 0:02:34 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1131 (0.2414) acc1: 99.7396 (91.3976) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1921 data: 0.0993 max mem: 3537 +Test: [1570/2387] eta: 0:02:33 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1093 (0.2406) acc1: 99.7396 (91.4490) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2299 data: 0.1375 max mem: 3537 +Test: [1580/2387] eta: 0:02:31 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1188 (0.2402) acc1: 99.2188 (91.4837) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2403 data: 0.1471 max mem: 3538 +Test: [1590/2387] eta: 0:02:29 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1990 (0.2403) acc1: 95.8333 (91.4804) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1471 data: 0.0519 max mem: 3538 +Test: [1600/2387] eta: 0:02:27 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2444 (0.2402) acc1: 91.6667 (91.4887) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1408 data: 0.0456 max mem: 3538 +Test: [1610/2387] eta: 0:02:25 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2273 (0.2401) acc1: 92.1875 (91.4942) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2023 data: 0.1096 max mem: 3538 +Test: [1620/2387] eta: 0:02:23 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2273 (0.2400) acc1: 92.4479 (91.4991) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2081 data: 0.1152 max mem: 3538 +Test: [1630/2387] eta: 0:02:21 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2182 (0.2399) acc1: 93.4896 (91.5164) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1841 data: 0.0903 max mem: 3538 +Test: [1640/2387] eta: 0:02:20 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2158 (0.2398) acc1: 94.7917 (91.5346) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1731 data: 0.0801 max mem: 3538 +Test: [1650/2387] eta: 0:02:18 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2175 (0.2396) acc1: 94.7917 (91.5539) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1858 data: 0.0921 max mem: 3538 +Test: [1660/2387] eta: 0:02:16 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2175 (0.2397) acc1: 94.2708 (91.5619) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1877 data: 0.0937 max mem: 3538 +Test: [1670/2387] eta: 0:02:14 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2281 (0.2396) acc1: 94.0104 (91.5743) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1823 data: 0.0901 max mem: 3538 +Test: [1680/2387] eta: 0:02:12 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2281 (0.2397) acc1: 93.7500 (91.5777) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1687 data: 0.0747 max mem: 3538 +Test: [1690/2387] eta: 0:02:10 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2347 (0.2396) acc1: 92.7083 (91.5914) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1849 data: 0.0911 max mem: 3538 +Test: [1700/2387] eta: 0:02:08 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2022 (0.2394) acc1: 95.0521 (91.6139) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2049 data: 0.1107 max mem: 3538 +Test: [1710/2387] eta: 0:02:06 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1858 (0.2391) acc1: 95.0521 (91.6349) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1554 data: 0.0587 max mem: 3538 +Test: [1720/2387] eta: 0:02:04 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1758 (0.2387) acc1: 94.5312 (91.6508) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1360 data: 0.0413 max mem: 3538 +Test: [1730/2387] eta: 0:02:02 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1646 (0.2383) acc1: 95.5729 (91.6767) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1599 data: 0.0660 max mem: 3538 +Test: [1740/2387] eta: 0:02:00 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1619 (0.2379) acc1: 95.8333 (91.6964) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1745 data: 0.0799 max mem: 3538 +Test: [1750/2387] eta: 0:01:58 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1689 (0.2376) acc1: 95.5729 (91.7196) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1707 data: 0.0778 max mem: 3538 +Test: [1760/2387] eta: 0:01:56 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1583 (0.2371) acc1: 96.6146 (91.7490) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1594 data: 0.0675 max mem: 3538 +Test: [1770/2387] eta: 0:01:54 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1554 (0.2367) acc1: 96.6146 (91.7728) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1492 data: 0.0569 max mem: 3538 +Test: [1780/2387] eta: 0:01:52 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1693 (0.2365) acc1: 95.3125 (91.7864) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1653 data: 0.0712 max mem: 3538 +Test: [1790/2387] eta: 0:01:51 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1627 (0.2360) acc1: 95.3125 (91.8137) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1896 data: 0.0953 max mem: 3538 +Test: [1800/2387] eta: 0:01:49 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1541 (0.2357) acc1: 96.8750 (91.8312) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1949 data: 0.1019 max mem: 3538 +Test: [1810/2387] eta: 0:01:47 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1464 (0.2352) acc1: 97.3958 (91.8654) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2148 data: 0.1226 max mem: 3539 +Test: [1820/2387] eta: 0:01:45 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1331 (0.2346) acc1: 98.1771 (91.9008) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2074 data: 0.1149 max mem: 3539 +Test: [1830/2387] eta: 0:01:43 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1274 (0.2341) acc1: 98.9583 (91.9375) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1918 data: 0.0964 max mem: 3539 +Test: [1840/2387] eta: 0:01:42 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1254 (0.2335) acc1: 99.7396 (91.9798) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1917 data: 0.0963 max mem: 3539 +Test: [1850/2387] eta: 0:01:40 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1262 (0.2329) acc1: 99.7396 (92.0218) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1986 data: 0.1065 max mem: 3539 +Test: [1860/2387] eta: 0:01:38 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1258 (0.2324) acc1: 99.7396 (92.0624) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2065 data: 0.1160 max mem: 3539 +Test: [1870/2387] eta: 0:01:36 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1249 (0.2318) acc1: 99.7396 (92.1039) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1939 data: 0.1020 max mem: 3539 +Test: [1880/2387] eta: 0:01:34 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1243 (0.2312) acc1: 99.7396 (92.1442) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1897 data: 0.0975 max mem: 3539 +Test: [1890/2387] eta: 0:01:32 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1247 (0.2307) acc1: 99.7396 (92.1843) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1924 data: 0.1000 max mem: 3539 +Test: [1900/2387] eta: 0:01:30 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1259 (0.2301) acc1: 99.7396 (92.2230) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1914 data: 0.0970 max mem: 3539 +Test: [1910/2387] eta: 0:01:29 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1846 (0.2300) acc1: 98.1771 (92.2483) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1899 data: 0.0974 max mem: 3539 +Test: [1920/2387] eta: 0:01:27 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2071 (0.2306) acc1: 95.5729 (92.2104) acc5: 100.0000 (100.0000) acc1_neg: 77.083 acc5_neg: 100.000 time: 0.1876 data: 0.0964 max mem: 3539 +Test: [1930/2387] eta: 0:01:25 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.3004 (0.2308) acc1: 88.2812 (92.2038) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1984 data: 0.1046 max mem: 3539 +Test: [1940/2387] eta: 0:01:23 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.3004 (0.2313) acc1: 88.2812 (92.1808) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.2096 data: 0.1161 max mem: 3539 +Test: [1950/2387] eta: 0:01:21 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.3022 (0.2318) acc1: 87.2396 (92.1544) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.2111 data: 0.1201 max mem: 3539 +Test: [1960/2387] eta: 0:01:19 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2938 (0.2324) acc1: 87.2396 (92.1029) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1973 data: 0.1061 max mem: 3539 +Test: [1970/2387] eta: 0:01:18 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.3218 (0.2331) acc1: 84.8958 (92.0572) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.1975 data: 0.1026 max mem: 3539 +Test: [1980/2387] eta: 0:01:16 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2686 (0.2329) acc1: 91.9271 (92.0713) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2021 data: 0.1036 max mem: 3539 +Test: [1990/2387] eta: 0:01:14 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1112 (0.2323) acc1: 98.9583 (92.1060) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1999 data: 0.1040 max mem: 3539 +Test: [2000/2387] eta: 0:01:12 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1089 (0.2317) acc1: 98.9583 (92.1407) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2068 data: 0.1140 max mem: 3539 +Test: [2010/2387] eta: 0:01:10 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1092 (0.2311) acc1: 98.9583 (92.1739) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2017 data: 0.1088 max mem: 3539 +Test: [2020/2387] eta: 0:01:08 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1111 (0.2305) acc1: 98.9583 (92.2082) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1898 data: 0.0975 max mem: 3539 +Test: [2030/2387] eta: 0:01:06 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1111 (0.2299) acc1: 98.9583 (92.2414) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1731 data: 0.0822 max mem: 3540 +Test: [2040/2387] eta: 0:01:05 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1078 (0.2293) acc1: 99.2188 (92.2758) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1647 data: 0.0740 max mem: 3540 +Test: [2050/2387] eta: 0:01:03 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1090 (0.2288) acc1: 99.2188 (92.3083) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1599 data: 0.0685 max mem: 3540 +Test: [2060/2387] eta: 0:01:01 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1172 (0.2283) acc1: 98.6979 (92.3375) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1596 data: 0.0679 max mem: 3540 +Test: [2070/2387] eta: 0:00:59 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1212 (0.2277) acc1: 98.6979 (92.3696) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1649 data: 0.0727 max mem: 3540 +Test: [2080/2387] eta: 0:00:57 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1187 (0.2272) acc1: 98.9583 (92.4004) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1761 data: 0.0844 max mem: 3540 +Test: [2090/2387] eta: 0:00:55 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1187 (0.2267) acc1: 98.9583 (92.4315) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1831 data: 0.0910 max mem: 3540 +Test: [2100/2387] eta: 0:00:53 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1160 (0.2262) acc1: 98.6979 (92.4616) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2024 data: 0.1096 max mem: 3540 +Test: [2110/2387] eta: 0:00:51 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1160 (0.2257) acc1: 98.6979 (92.4912) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1917 data: 0.0954 max mem: 3540 +Test: [2120/2387] eta: 0:00:49 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1269 (0.2258) acc1: 97.9167 (92.4823) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.1465 data: 0.0483 max mem: 3540 +Test: [2130/2387] eta: 0:00:48 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2776 (0.2262) acc1: 89.3229 (92.4565) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.1766 data: 0.0823 max mem: 3540 +Test: [2140/2387] eta: 0:00:46 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2990 (0.2266) acc1: 88.0208 (92.4333) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1833 data: 0.0908 max mem: 3540 +Test: [2150/2387] eta: 0:00:44 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.3092 (0.2270) acc1: 87.2396 (92.4081) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.1657 data: 0.0741 max mem: 3540 +Test: [2160/2387] eta: 0:00:42 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2794 (0.2272) acc1: 88.8021 (92.4075) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1903 data: 0.0974 max mem: 3540 +Test: [2170/2387] eta: 0:00:40 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2538 (0.2274) acc1: 90.6250 (92.3953) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1922 data: 0.0994 max mem: 3540 +Test: [2180/2387] eta: 0:00:38 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2997 (0.2280) acc1: 88.5417 (92.3676) acc5: 100.0000 (100.0000) acc1_neg: 79.427 acc5_neg: 100.000 time: 0.1995 data: 0.1067 max mem: 3540 +Test: [2190/2387] eta: 0:00:36 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2798 (0.2279) acc1: 90.6250 (92.3821) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1907 data: 0.0977 max mem: 3540 +Test: [2200/2387] eta: 0:00:34 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1671 (0.2276) acc1: 98.1771 (92.4079) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2006 data: 0.1099 max mem: 3540 +Test: [2210/2387] eta: 0:00:33 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1632 (0.2274) acc1: 98.4375 (92.4279) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1990 data: 0.1044 max mem: 3540 +Test: [2220/2387] eta: 0:00:31 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1696 (0.2272) acc1: 98.4375 (92.4490) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1864 data: 0.0915 max mem: 3540 +Test: [2230/2387] eta: 0:00:29 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1587 (0.2269) acc1: 98.4375 (92.4721) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2099 data: 0.1164 max mem: 3540 +Test: [2240/2387] eta: 0:00:27 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1451 (0.2265) acc1: 98.6979 (92.5001) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1950 data: 0.1010 max mem: 3540 +Test: [2250/2387] eta: 0:00:25 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1388 (0.2260) acc1: 99.4792 (92.5302) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1899 data: 0.0971 max mem: 3540 +Test: [2260/2387] eta: 0:00:23 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1317 (0.2258) acc1: 99.2188 (92.5474) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1874 data: 0.0927 max mem: 3541 +Test: [2270/2387] eta: 0:00:21 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2120 (0.2260) acc1: 94.2708 (92.5401) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1949 data: 0.1001 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2621 (0.2262) acc1: 90.1042 (92.5281) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2149 data: 0.1194 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2621 (0.2263) acc1: 90.6250 (92.5229) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1599 data: 0.0604 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2330 (0.2264) acc1: 92.4479 (92.5211) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1499 data: 0.0506 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2330 (0.2263) acc1: 94.5312 (92.5302) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2054 data: 0.1116 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2312 (0.2265) acc1: 94.2708 (92.5286) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2089 data: 0.1179 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2625 (0.2270) acc1: 90.3646 (92.4933) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1756 data: 0.0846 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.3608 (0.2276) acc1: 79.9479 (92.4496) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.1515 data: 0.0610 max mem: 3541 +Test: [2350/2387] eta: 0:00:06 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.3926 (0.2282) acc1: 81.2500 (92.4074) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.1510 data: 0.0614 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.3555 (0.2288) acc1: 82.8125 (92.3736) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1478 data: 0.0583 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.3310 (0.2292) acc1: 83.3333 (92.3385) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.1423 data: 0.0528 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.3051 (0.2296) acc1: 86.1979 (92.3227) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.1407 data: 0.0512 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.3038 (0.2297) acc1: 90.1042 (92.3235) acc5: 100.0000 (100.0000) acc1_neg: 97.674 acc5_neg: 100.000 time: 0.1334 data: 0.0465 max mem: 3541 +Test: Total time: 0:07:24 (0.1862 s / it) +* Overall Acc@1 92.324 Acc@5 100.000 + target == real Acc@1 51.568 Acc@5 100.000 + target == fake Acc@1 95.186 Acc@5 100.000 + loss 0.230 + PR-AUC 0.9945 ROC-AUC 0.9277 EER 14.47% (thr=0.7544) +Accuracy of the network on the 916267 test images: 92.3% +[rank0]:[W726 03:09:49.999589512 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:10:09.376000 3139966 torch/distributed/run.py:766] +W0726 03:10:09.376000 3139966 torch/distributed/run.py:766] ***************************************** +W0726 03:10:09.376000 3139966 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:10:09.376000 3139966 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:10:37.235365184 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:10:37.446872464 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:10:38.973242720 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:10:38.052537080 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:10:38.100609941 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:10:38.222964890 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:10:38.257459220 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank6]:[W726 03:10:38.257481123 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n16_seed5_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed5_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:05:05 acc1_pos: 61.979 acc5_pos: 100.000 loss: 0.6331 (0.6331) acc1: 61.9792 (61.9792) acc5: 100.0000 (100.0000) time: 7.5534 data: 6.0416 max mem: 3531 +Test: [ 10/517] eta: 0:06:32 acc1_pos: 56.510 acc5_pos: 100.000 loss: 0.7408 (0.7674) acc1: 59.1146 (52.4621) acc5: 100.0000 (100.0000) time: 0.7746 data: 0.5496 max mem: 3531 +Test: [ 20/517] eta: 0:03:45 acc1_pos: 71.354 acc5_pos: 100.000 loss: 0.6248 (0.6971) acc1: 64.0625 (60.7267) acc5: 100.0000 (100.0000) time: 0.0983 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:03:22 acc1_pos: 17.969 acc5_pos: 100.000 loss: 0.6600 (0.7463) acc1: 54.1667 (54.3347) acc5: 100.0000 (100.0000) time: 0.2199 data: 0.1247 max mem: 3531 +Test: [ 40/517] eta: 0:03:12 acc1_pos: 55.729 acc5_pos: 100.000 loss: 0.8749 (0.7776) acc1: 35.9375 (50.7495) acc5: 100.0000 (100.0000) time: 0.3499 data: 0.2582 max mem: 3531 +Test: [ 50/517] eta: 0:02:54 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.7436 (0.7571) acc1: 52.3438 (52.6654) acc5: 100.0000 (100.0000) acc1_neg: 59.115 acc5_neg: 100.000 time: 0.3049 data: 0.2114 max mem: 3531 +Test: [ 60/517] eta: 0:02:41 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.6637 (0.7303) acc1: 59.1146 (54.9308) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.2549 data: 0.1597 max mem: 3531 +Test: [ 70/517] eta: 0:02:32 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.4565 (0.6880) acc1: 77.0833 (59.1109) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.2599 data: 0.1632 max mem: 3531 +Test: [ 80/517] eta: 0:02:24 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.4694 (0.6681) acc1: 80.7292 (61.1658) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.2602 data: 0.1645 max mem: 3531 +Test: [ 90/517] eta: 0:02:20 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.5171 (0.6509) acc1: 77.3438 (63.0695) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.2849 data: 0.1904 max mem: 3531 +Test: [100/517] eta: 0:02:12 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.5082 (0.6379) acc1: 76.5625 (64.0831) acc5: 100.0000 (100.0000) acc1_neg: 65.625 acc5_neg: 100.000 time: 0.2652 data: 0.1667 max mem: 3531 +Test: [110/517] eta: 0:02:05 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.5359 (0.6331) acc1: 71.3542 (64.2385) acc5: 100.0000 (100.0000) acc1_neg: 74.219 acc5_neg: 100.000 time: 0.2149 data: 0.1159 max mem: 3531 +Test: [120/517] eta: 0:02:00 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.5361 (0.6246) acc1: 71.0938 (64.8588) acc5: 100.0000 (100.0000) acc1_neg: 66.667 acc5_neg: 100.000 time: 0.2252 data: 0.1302 max mem: 3531 +Test: [130/517] eta: 0:01:55 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.5050 (0.6159) acc1: 74.7396 (65.6886) acc5: 100.0000 (100.0000) acc1_neg: 70.052 acc5_neg: 100.000 time: 0.2409 data: 0.1476 max mem: 3531 +Test: [140/517] eta: 0:01:51 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.6045 (0.6217) acc1: 65.8854 (65.2704) acc5: 100.0000 (100.0000) acc1_neg: 54.167 acc5_neg: 100.000 time: 0.2540 data: 0.1579 max mem: 3531 +Test: [150/517] eta: 0:01:47 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.6771 (0.6209) acc1: 62.7604 (65.4284) acc5: 100.0000 (100.0000) acc1_neg: 64.323 acc5_neg: 100.000 time: 0.2595 data: 0.1631 max mem: 3531 +Test: [160/517] eta: 0:01:43 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.6366 (0.6232) acc1: 64.3229 (65.1689) acc5: 100.0000 (100.0000) acc1_neg: 66.146 acc5_neg: 100.000 time: 0.2449 data: 0.1479 max mem: 3531 +Test: [170/517] eta: 0:01:39 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.6575 (0.6236) acc1: 61.9792 (64.9976) acc5: 100.0000 (100.0000) acc1_neg: 69.792 acc5_neg: 100.000 time: 0.2392 data: 0.1400 max mem: 3531 +Test: [180/517] eta: 0:01:35 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.5804 (0.6197) acc1: 67.4479 (65.3286) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.2449 data: 0.1452 max mem: 3531 +Test: [190/517] eta: 0:01:31 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.3569 (0.6009) acc1: 80.9896 (66.8085) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.2360 data: 0.1376 max mem: 3531 +Test: [200/517] eta: 0:01:28 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.3153 (0.5956) acc1: 88.8021 (67.1357) acc5: 100.0000 (100.0000) acc1_neg: 61.458 acc5_neg: 100.000 time: 0.2349 data: 0.1400 max mem: 3531 +Test: [210/517] eta: 0:01:24 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.6289 (0.6015) acc1: 57.5521 (66.2075) acc5: 100.0000 (100.0000) acc1_neg: 48.698 acc5_neg: 100.000 time: 0.2338 data: 0.1381 max mem: 3532 +Test: [220/517] eta: 0:01:21 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.7333 (0.6080) acc1: 45.0521 (65.3127) acc5: 100.0000 (100.0000) acc1_neg: 48.177 acc5_neg: 100.000 time: 0.2254 data: 0.1299 max mem: 3532 +Test: [230/517] eta: 0:01:18 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.7336 (0.6135) acc1: 46.6146 (64.6465) acc5: 100.0000 (100.0000) acc1_neg: 53.385 acc5_neg: 100.000 time: 0.2349 data: 0.1410 max mem: 3532 +Test: [240/517] eta: 0:01:15 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.6796 (0.6150) acc1: 52.8646 (64.5120) acc5: 100.0000 (100.0000) acc1_neg: 33.594 acc5_neg: 100.000 time: 0.2350 data: 0.1421 max mem: 3532 +Test: [250/517] eta: 0:01:12 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.5889 (0.6145) acc1: 65.6250 (64.5802) acc5: 100.0000 (100.0000) acc1_neg: 73.958 acc5_neg: 100.000 time: 0.2400 data: 0.1488 max mem: 3532 +Test: [260/517] eta: 0:01:09 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.5885 (0.6151) acc1: 64.8438 (64.4866) acc5: 100.0000 (100.0000) acc1_neg: 56.250 acc5_neg: 100.000 time: 0.2344 data: 0.1412 max mem: 3532 +Test: [270/517] eta: 0:01:06 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.6596 (0.6180) acc1: 59.8958 (64.2364) acc5: 100.0000 (100.0000) acc1_neg: 61.458 acc5_neg: 100.000 time: 0.2299 data: 0.1330 max mem: 3532 +Test: [280/517] eta: 0:01:03 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.6746 (0.6207) acc1: 54.6875 (63.9207) acc5: 100.0000 (100.0000) acc1_neg: 47.396 acc5_neg: 100.000 time: 0.2349 data: 0.1348 max mem: 3532 +Test: [290/517] eta: 0:01:00 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.6922 (0.6255) acc1: 54.6875 (63.3537) acc5: 100.0000 (100.0000) acc1_neg: 39.323 acc5_neg: 100.000 time: 0.2349 data: 0.1337 max mem: 3532 +Test: [300/517] eta: 0:00:57 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.6834 (0.6247) acc1: 57.2917 (63.4526) acc5: 100.0000 (100.0000) acc1_neg: 66.927 acc5_neg: 100.000 time: 0.2399 data: 0.1436 max mem: 3532 +Test: [310/517] eta: 0:00:54 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.5851 (0.6237) acc1: 66.1458 (63.4546) acc5: 100.0000 (100.0000) acc1_neg: 55.208 acc5_neg: 100.000 time: 0.2399 data: 0.1470 max mem: 3532 +Test: [320/517] eta: 0:00:51 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.5313 (0.6193) acc1: 69.5312 (63.8783) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.2451 data: 0.1498 max mem: 3532 +Test: [330/517] eta: 0:00:48 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.4351 (0.6128) acc1: 84.1146 (64.5629) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2353 data: 0.1386 max mem: 3532 +Test: [340/517] eta: 0:00:46 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.3801 (0.6063) acc1: 89.8438 (65.3073) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.2247 data: 0.1309 max mem: 3532 +Test: [350/517] eta: 0:00:43 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.4487 (0.6069) acc1: 83.8542 (65.2592) acc5: 100.0000 (100.0000) acc1_neg: 44.792 acc5_neg: 100.000 time: 0.2545 data: 0.1622 max mem: 3532 +Test: [360/517] eta: 0:00:40 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.6469 (0.6056) acc1: 57.2917 (65.4324) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.2599 data: 0.1655 max mem: 3532 +Test: [370/517] eta: 0:00:38 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.3402 (0.5977) acc1: 96.3542 (66.3101) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2303 data: 0.1367 max mem: 3532 +Test: [380/517] eta: 0:00:35 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.2957 (0.5896) acc1: 98.4375 (67.1431) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2149 data: 0.1224 max mem: 3532 +Test: [390/517] eta: 0:00:32 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.2880 (0.5818) acc1: 98.9583 (67.9301) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.2296 data: 0.1356 max mem: 3532 +Test: [400/517] eta: 0:00:30 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.2953 (0.5766) acc1: 98.1771 (68.4857) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2449 data: 0.1502 max mem: 3532 +Test: [410/517] eta: 0:00:27 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.3912 (0.5720) acc1: 87.5000 (68.9686) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.2449 data: 0.1495 max mem: 3532 +Test: [420/517] eta: 0:00:24 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.3735 (0.5692) acc1: 88.5417 (69.2745) acc5: 100.0000 (100.0000) acc1_neg: 67.708 acc5_neg: 100.000 time: 0.2399 data: 0.1438 max mem: 3532 +Test: [430/517] eta: 0:00:22 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.6042 (0.5720) acc1: 66.9271 (68.9911) acc5: 100.0000 (100.0000) acc1_neg: 68.750 acc5_neg: 100.000 time: 0.2299 data: 0.1346 max mem: 3532 +Test: [440/517] eta: 0:00:19 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.5614 (0.5694) acc1: 69.7917 (69.3110) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.2349 data: 0.1404 max mem: 3533 +Test: [450/517] eta: 0:00:17 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.4354 (0.5667) acc1: 84.1146 (69.6219) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.2399 data: 0.1452 max mem: 3533 +Test: [460/517] eta: 0:00:14 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.4354 (0.5640) acc1: 85.4167 (69.9538) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.2399 data: 0.1443 max mem: 3533 +Test: [470/517] eta: 0:00:11 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.4020 (0.5604) acc1: 86.9792 (70.3518) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.2399 data: 0.1424 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.4020 (0.5576) acc1: 88.2812 (70.6937) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2399 data: 0.1421 max mem: 3533 +Test: [490/517] eta: 0:00:06 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.4442 (0.5592) acc1: 82.8125 (70.4870) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.2299 data: 0.1316 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.4500 (0.5564) acc1: 80.2083 (70.7231) acc5: 100.0000 (100.0000) acc1_neg: 72.135 acc5_neg: 100.000 time: 0.2131 data: 0.1175 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.4202 (0.5538) acc1: 84.1146 (71.0081) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1874 data: 0.0969 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.4202 (0.5526) acc1: 84.3750 (71.1063) acc5: 100.0000 (100.0000) acc1_neg: 71.025 acc5_neg: 100.000 time: 0.1713 data: 0.0800 max mem: 3533 +Test: Total time: 0:02:09 (0.2514 s / it) +* Overall Acc@1 71.106 Acc@5 100.000 + target == real Acc@1 52.306 Acc@5 100.000 + target == fake Acc@1 73.057 Acc@5 100.000 + loss 0.553 + PR-AUC 0.9508 ROC-AUC 0.6896 EER 36.92% (thr=0.5737) +Accuracy of the network on the 198427 test images: 71.1% +[rank0]:[W726 03:13:00.209689925 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 03:13:20.167000 3140528 torch/distributed/run.py:766] +W0726 03:13:20.167000 3140528 torch/distributed/run.py:766] ***************************************** +W0726 03:13:20.167000 3140528 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:13:20.167000 3140528 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:13:46.448064912 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:13:47.874961055 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:13:47.888054811 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:13:47.113242196 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:13:47.258627569 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:13:47.273626608 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:13:47.291795213 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W726 03:13:47.291945153 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n16_seed5_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed5_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:50:23 acc1_pos: 21.615 acc5_pos: 100.000 loss: 1.0495 (1.0495) acc1: 21.6146 (21.6146) acc5: 100.0000 (100.0000) time: 7.7735 data: 5.5845 max mem: 3531 +Test: [ 10/389] eta: 0:05:01 acc1_pos: 70.833 acc5_pos: 100.000 loss: 1.0069 (0.9501) acc1: 34.6354 (37.6894) acc5: 100.0000 (100.0000) time: 0.7957 data: 0.5081 max mem: 3531 +Test: [ 20/389] eta: 0:02:51 acc1_pos: 4.948 acc5_pos: 100.000 loss: 1.0069 (1.1200) acc1: 34.6354 (35.3051) acc5: 100.0000 (100.0000) time: 0.0989 data: 0.0004 max mem: 3531 +Test: [ 30/389] eta: 0:02:18 acc1_pos: 1.042 acc5_pos: 100.000 loss: 1.1757 (1.1391) acc1: 14.5833 (30.0487) acc5: 100.0000 (100.0000) time: 0.1609 data: 0.0662 max mem: 3531 +Test: [ 40/389] eta: 0:01:59 acc1_pos: 6.250 acc5_pos: 100.000 loss: 1.1757 (1.0999) acc1: 14.5833 (31.7899) acc5: 100.0000 (100.0000) time: 0.2156 data: 0.1233 max mem: 3531 +Test: [ 50/389] eta: 0:01:48 acc1_pos: 27.083 acc5_pos: 100.000 loss: 0.9067 (1.0456) acc1: 42.1875 (34.5486) acc5: 100.0000 (100.0000) time: 0.2139 data: 0.1171 max mem: 3531 +Test: [ 60/389] eta: 0:01:36 acc1_pos: 23.177 acc5_pos: 100.000 loss: 1.0072 (1.0509) acc1: 23.6979 (32.6503) acc5: 100.0000 (100.0000) time: 0.1892 data: 0.0934 max mem: 3531 +Test: [ 70/389] eta: 0:01:28 acc1_pos: 72.656 acc5_pos: 100.000 loss: 1.0072 (1.0154) acc1: 23.6979 (35.0829) acc5: 100.0000 (100.0000) time: 0.1699 data: 0.0764 max mem: 3531 +Test: [ 80/389] eta: 0:01:22 acc1_pos: 63.281 acc5_pos: 100.000 loss: 0.8750 (1.0167) acc1: 42.9688 (35.2495) acc5: 100.0000 (100.0000) time: 0.1903 data: 0.0983 max mem: 3531 +Test: [ 90/389] eta: 0:01:18 acc1_pos: 41.146 acc5_pos: 100.000 loss: 0.7178 (0.9740) acc1: 51.8229 (38.1525) acc5: 100.0000 (100.0000) time: 0.2099 data: 0.1171 max mem: 3531 +Test: [100/389] eta: 0:01:13 acc1_pos: 84.115 acc5_pos: 100.000 loss: 0.9292 (1.0008) acc1: 34.8958 (36.7600) acc5: 100.0000 (100.0000) time: 0.2019 data: 0.1066 max mem: 3531 +Test: [110/389] eta: 0:01:09 acc1_pos: 99.740 acc5_pos: 100.000 loss: 1.2179 (1.0045) acc1: 26.5625 (38.1100) acc5: 100.0000 (100.0000) time: 0.1851 data: 0.0912 max mem: 3531 +Test: [120/389] eta: 0:01:05 acc1_pos: 65.885 acc5_pos: 100.000 loss: 0.7284 (0.9908) acc1: 54.6875 (39.3982) acc5: 100.0000 (100.0000) time: 0.1881 data: 0.0965 max mem: 3531 +Test: [130/389] eta: 0:01:02 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.7284 (0.9778) acc1: 54.6875 (40.5872) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1958 data: 0.1052 max mem: 3531 +Test: [140/389] eta: 0:00:58 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.3414 (0.9412) acc1: 84.1146 (43.4009) acc5: 100.0000 (100.0000) acc1_neg: 22.656 acc5_neg: 100.000 time: 0.1842 data: 0.0910 max mem: 3531 +Test: [150/389] eta: 0:00:55 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.4334 (0.9264) acc1: 73.4375 (44.4088) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1793 data: 0.0842 max mem: 3531 +Test: [160/389] eta: 0:00:52 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.3452 (0.8862) acc1: 86.7188 (47.4702) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1802 data: 0.0863 max mem: 3531 +Test: [170/389] eta: 0:00:49 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2741 (0.8501) acc1: 97.3958 (50.3670) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1712 data: 0.0789 max mem: 3531 +Test: [180/389] eta: 0:00:46 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2773 (0.8242) acc1: 97.1354 (52.2287) acc5: 100.0000 (100.0000) acc1_neg: 79.427 acc5_neg: 100.000 time: 0.1667 data: 0.0724 max mem: 3531 +Test: [190/389] eta: 0:00:43 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.3064 (0.7968) acc1: 92.1875 (54.3248) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1647 data: 0.0693 max mem: 3531 +Test: [200/389] eta: 0:00:40 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2789 (0.7747) acc1: 93.4896 (56.1632) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1667 data: 0.0738 max mem: 3531 +Test: [210/389] eta: 0:00:38 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2361 (0.7490) acc1: 98.6979 (58.1211) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1592 data: 0.0667 max mem: 3532 +Test: [220/389] eta: 0:00:35 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1909 (0.7231) acc1: 99.2188 (59.9984) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1649 data: 0.0714 max mem: 3532 +Test: [230/389] eta: 0:00:33 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2067 (0.7080) acc1: 99.4792 (61.1596) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.1708 data: 0.0765 max mem: 3532 +Test: [240/389] eta: 0:00:30 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.3739 (0.6952) acc1: 92.7083 (62.1510) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1712 data: 0.0785 max mem: 3532 +Test: [250/389] eta: 0:00:28 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2312 (0.6763) acc1: 98.4375 (63.6299) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1696 data: 0.0766 max mem: 3532 +Test: [260/389] eta: 0:00:26 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2225 (0.6651) acc1: 98.6979 (64.4746) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1646 data: 0.0708 max mem: 3532 +Test: [270/389] eta: 0:00:24 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.3290 (0.6577) acc1: 92.9688 (64.9725) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1676 data: 0.0749 max mem: 3532 +Test: [280/389] eta: 0:00:22 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.3008 (0.6458) acc1: 96.8750 (65.9447) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1717 data: 0.0794 max mem: 3532 +Test: [290/389] eta: 0:00:19 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.3008 (0.6375) acc1: 95.5729 (66.4940) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1733 data: 0.0813 max mem: 3532 +Test: [300/389] eta: 0:00:17 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.3561 (0.6276) acc1: 93.2292 (67.3813) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1731 data: 0.0806 max mem: 3532 +Test: [310/389] eta: 0:00:15 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.3784 (0.6247) acc1: 90.6250 (67.5727) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1763 data: 0.0844 max mem: 3532 +Test: [320/389] eta: 0:00:13 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.4103 (0.6147) acc1: 88.5417 (68.3557) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1696 data: 0.0788 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2721 (0.6060) acc1: 95.3125 (69.1025) acc5: 100.0000 (100.0000) acc1_neg: 79.427 acc5_neg: 100.000 time: 0.1697 data: 0.0786 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2662 (0.5962) acc1: 97.1354 (69.8215) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1855 data: 0.0924 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2451 (0.5989) acc1: 84.8958 (69.4615) acc5: 100.0000 (100.0000) acc1_neg: 15.104 acc5_neg: 100.000 time: 0.1509 data: 0.0590 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.6892 (0.6032) acc1: 56.2500 (68.8878) acc5: 100.0000 (100.0000) acc1_neg: 61.979 acc5_neg: 100.000 time: 0.1169 data: 0.0272 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.6601 (0.5985) acc1: 60.9375 (69.0911) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1208 data: 0.0312 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.4042 (0.5973) acc1: 78.3854 (69.0186) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.1128 data: 0.0233 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2384 (0.5927) acc1: 88.8021 (69.3455) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1128 data: 0.0200 max mem: 3532 +Test: Total time: 0:01:13 (0.1880 s / it) +* Overall Acc@1 69.345 Acc@5 100.000 + target == real Acc@1 38.764 Acc@5 100.000 + target == fake Acc@1 83.561 Acc@5 100.000 + loss 0.593 + PR-AUC 0.8104 ROC-AUC 0.6944 EER 35.70% (thr=0.7158) +Accuracy of the network on the 149358 test images: 69.3% +[rank0]:[W726 03:15:08.152205290 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:15:27.194000 3141087 torch/distributed/run.py:766] +W0726 03:15:27.194000 3141087 torch/distributed/run.py:766] ***************************************** +W0726 03:15:27.194000 3141087 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:15:27.194000 3141087 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:15:55.936773063 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:15:55.026546650 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:15:55.398821739 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:15:55.452095952 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +NCCL version 2.26.2+cuda12.2 +[rank1]:[W726 03:15:56.636678867 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +| distributed init (rank 3): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:15:56.780056353 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:15:56.785192560 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W726 03:15:56.785243133 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n16_seed5_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed5_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:26:21 acc1_pos: 93.490 acc5_pos: 100.000 loss: 0.2467 (0.2467) acc1: 93.4896 (93.4896) acc5: 100.0000 (100.0000) time: 4.7481 data: 2.6518 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/333] eta: 0:02:48 acc1_pos: 47.135 acc5_pos: 100.000 loss: 0.2467 (0.4580) acc1: 91.6667 (76.7282) acc5: 100.0000 (100.0000) time: 0.5220 data: 0.2501 max mem: 3531 +Test: [ 20/333] eta: 0:01:42 acc1_pos: 51.562 acc5_pos: 100.000 loss: 0.8535 (0.7074) acc1: 47.1354 (59.7346) acc5: 100.0000 (100.0000) time: 0.1052 data: 0.0157 max mem: 3531 +Test: [ 30/333] eta: 0:01:19 acc1_pos: 86.979 acc5_pos: 100.000 loss: 0.7871 (0.6419) acc1: 48.4375 (65.2386) acc5: 100.0000 (100.0000) time: 0.1192 data: 0.0289 max mem: 3531 +Test: [ 40/333] eta: 0:01:15 acc1_pos: 78.906 acc5_pos: 100.000 loss: 0.4653 (0.6017) acc1: 79.6875 (68.7373) acc5: 100.0000 (100.0000) time: 0.1876 data: 0.0954 max mem: 3531 +Test: [ 50/333] eta: 0:01:19 acc1_pos: 75.521 acc5_pos: 100.000 loss: 0.4692 (0.5754) acc1: 79.6875 (71.0325) acc5: 100.0000 (100.0000) time: 0.3089 data: 0.2154 max mem: 3531 +Test: [ 60/333] eta: 0:01:11 acc1_pos: 80.469 acc5_pos: 100.000 loss: 0.4252 (0.5517) acc1: 82.0312 (73.0789) acc5: 100.0000 (100.0000) time: 0.2649 data: 0.1687 max mem: 3531 +Test: [ 70/333] eta: 0:01:02 acc1_pos: 73.958 acc5_pos: 100.000 loss: 0.4187 (0.5403) acc1: 80.4688 (73.8556) acc5: 100.0000 (100.0000) time: 0.1299 data: 0.0307 max mem: 3531 +Test: [ 80/333] eta: 0:00:58 acc1_pos: 84.375 acc5_pos: 100.000 loss: 0.4474 (0.5288) acc1: 79.1667 (74.7107) acc5: 100.0000 (100.0000) time: 0.1349 data: 0.0392 max mem: 3531 +Test: [ 90/333] eta: 0:00:55 acc1_pos: 78.646 acc5_pos: 100.000 loss: 0.4288 (0.5192) acc1: 79.1667 (75.3520) acc5: 100.0000 (100.0000) time: 0.1949 data: 0.1024 max mem: 3531 +Test: [100/333] eta: 0:00:52 acc1_pos: 81.250 acc5_pos: 100.000 loss: 0.4712 (0.5171) acc1: 77.8646 (75.5054) acc5: 100.0000 (100.0000) time: 0.2099 data: 0.1170 max mem: 3531 +Test: [110/333] eta: 0:00:49 acc1_pos: 82.552 acc5_pos: 100.000 loss: 0.4873 (0.5146) acc1: 76.5625 (75.5795) acc5: 100.0000 (100.0000) time: 0.1999 data: 0.1071 max mem: 3531 +Test: [120/333] eta: 0:00:46 acc1_pos: 77.344 acc5_pos: 100.000 loss: 0.4716 (0.5113) acc1: 77.3438 (75.7705) acc5: 100.0000 (100.0000) time: 0.1949 data: 0.1002 max mem: 3531 +Test: [130/333] eta: 0:00:43 acc1_pos: 89.583 acc5_pos: 100.000 loss: 0.4720 (0.5076) acc1: 77.3438 (76.0079) acc5: 100.0000 (100.0000) time: 0.1749 data: 0.0789 max mem: 3531 +Test: [140/333] eta: 0:00:41 acc1_pos: 80.469 acc5_pos: 100.000 loss: 0.4612 (0.5038) acc1: 80.4688 (76.3593) acc5: 100.0000 (100.0000) time: 0.1749 data: 0.0812 max mem: 3531 +Test: [150/333] eta: 0:00:38 acc1_pos: 81.510 acc5_pos: 100.000 loss: 0.4395 (0.5018) acc1: 81.2500 (76.4676) acc5: 100.0000 (100.0000) time: 0.1899 data: 0.0966 max mem: 3531 +Test: [160/333] eta: 0:00:36 acc1_pos: 74.342 acc5_pos: 100.000 loss: 0.4777 (0.5168) acc1: 77.3438 (75.2992) acc5: 100.0000 (100.0000) acc1_neg: 29.427 acc5_neg: 100.000 time: 0.1799 data: 0.0833 max mem: 3531 +Test: [170/333] eta: 0:00:33 acc1_pos: 74.342 acc5_pos: 100.000 loss: 0.9794 (0.5462) acc1: 34.8958 (72.8695) acc5: 100.0000 (100.0000) acc1_neg: 39.844 acc5_neg: 100.000 time: 0.1699 data: 0.0733 max mem: 3531 +Test: [180/333] eta: 0:00:31 acc1_pos: 74.342 acc5_pos: 100.000 loss: 0.9604 (0.5653) acc1: 35.4167 (70.9830) acc5: 100.0000 (100.0000) acc1_neg: 52.344 acc5_neg: 100.000 time: 0.1799 data: 0.0851 max mem: 3531 +Test: [190/333] eta: 0:00:29 acc1_pos: 74.342 acc5_pos: 100.000 loss: 0.9048 (0.5821) acc1: 36.7188 (69.3281) acc5: 100.0000 (100.0000) acc1_neg: 25.781 acc5_neg: 100.000 time: 0.1749 data: 0.0783 max mem: 3531 +Test: [200/333] eta: 0:00:26 acc1_pos: 74.342 acc5_pos: 100.000 loss: 0.8904 (0.5962) acc1: 42.4479 (68.1126) acc5: 100.0000 (100.0000) acc1_neg: 45.312 acc5_neg: 100.000 time: 0.1599 data: 0.0629 max mem: 3531 +Test: [210/333] eta: 0:00:24 acc1_pos: 74.342 acc5_pos: 100.000 loss: 0.8835 (0.6104) acc1: 42.9688 (66.8938) acc5: 100.0000 (100.0000) acc1_neg: 44.792 acc5_neg: 100.000 time: 0.1457 data: 0.0509 max mem: 3532 +Test: [220/333] eta: 0:00:22 acc1_pos: 74.342 acc5_pos: 100.000 loss: 0.9053 (0.6294) acc1: 40.8854 (65.2962) acc5: 100.0000 (100.0000) acc1_neg: 46.875 acc5_neg: 100.000 time: 0.1449 data: 0.0520 max mem: 3532 +Test: [230/333] eta: 0:00:20 acc1_pos: 74.342 acc5_pos: 100.000 loss: 0.8716 (0.6349) acc1: 41.6667 (64.6544) acc5: 100.0000 (100.0000) acc1_neg: 48.958 acc5_neg: 100.000 time: 0.1621 data: 0.0694 max mem: 3532 +Test: [240/333] eta: 0:00:18 acc1_pos: 74.342 acc5_pos: 100.000 loss: 0.7991 (0.6503) acc1: 41.6667 (63.2532) acc5: 100.0000 (100.0000) acc1_neg: 27.344 acc5_neg: 100.000 time: 0.1805 data: 0.0881 max mem: 3532 +Test: [250/333] eta: 0:00:16 acc1_pos: 74.342 acc5_pos: 100.000 loss: 1.0292 (0.6628) acc1: 30.7292 (62.2199) acc5: 100.0000 (100.0000) acc1_neg: 48.958 acc5_neg: 100.000 time: 0.1770 data: 0.0845 max mem: 3532 +Test: [260/333] eta: 0:00:14 acc1_pos: 74.342 acc5_pos: 100.000 loss: 0.9150 (0.6722) acc1: 38.2812 (61.4144) acc5: 100.0000 (100.0000) acc1_neg: 40.104 acc5_neg: 100.000 time: 0.1604 data: 0.0666 max mem: 3532 +Test: [270/333] eta: 0:00:12 acc1_pos: 74.342 acc5_pos: 100.000 loss: 0.8535 (0.6776) acc1: 45.8333 (60.9490) acc5: 100.0000 (100.0000) acc1_neg: 49.740 acc5_neg: 100.000 time: 0.1598 data: 0.0665 max mem: 3532 +Test: [280/333] eta: 0:00:10 acc1_pos: 74.342 acc5_pos: 100.000 loss: 0.8530 (0.6886) acc1: 45.8333 (60.0747) acc5: 100.0000 (100.0000) acc1_neg: 30.990 acc5_neg: 100.000 time: 0.1590 data: 0.0650 max mem: 3532 +Test: [290/333] eta: 0:00:08 acc1_pos: 74.342 acc5_pos: 100.000 loss: 1.0922 (0.7035) acc1: 28.6458 (58.9732) acc5: 100.0000 (100.0000) acc1_neg: 22.135 acc5_neg: 100.000 time: 0.1568 data: 0.0604 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 74.342 acc5_pos: 100.000 loss: 1.1177 (0.7162) acc1: 28.6458 (57.9855) acc5: 100.0000 (100.0000) acc1_neg: 31.250 acc5_neg: 100.000 time: 0.1572 data: 0.0631 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 74.342 acc5_pos: 100.000 loss: 1.0213 (0.7211) acc1: 32.0312 (57.5035) acc5: 100.0000 (100.0000) acc1_neg: 33.594 acc5_neg: 100.000 time: 0.1580 data: 0.0654 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 74.342 acc5_pos: 100.000 loss: 0.9691 (0.7291) acc1: 34.8958 (56.7919) acc5: 100.0000 (100.0000) acc1_neg: 45.833 acc5_neg: 100.000 time: 0.1470 data: 0.0514 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 74.342 acc5_pos: 100.000 loss: 0.9540 (0.7360) acc1: 36.4583 (56.2886) acc5: 100.0000 (100.0000) acc1_neg: 39.844 acc5_neg: 100.000 time: 0.1249 data: 0.0308 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 74.342 acc5_pos: 100.000 loss: 0.9534 (0.7372) acc1: 38.2812 (56.2049) acc5: 100.0000 (100.0000) acc1_neg: 43.944 acc5_neg: 100.000 time: 0.1289 data: 0.0308 max mem: 3532 +Test: Total time: 0:01:01 (0.1838 s / it) +* Overall Acc@1 56.205 Acc@5 100.000 + target == real Acc@1 76.555 Acc@5 100.000 + target == fake Acc@1 38.090 Acc@5 100.000 + loss 0.737 + PR-AUC 0.6333 ROC-AUC 0.6355 EER 40.49% (thr=0.3596) +Accuracy of the network on the 127843 test images: 56.2% +[rank0]:[W726 03:17:11.709219817 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:17:30.011000 3141646 torch/distributed/run.py:766] +W0726 03:17:30.011000 3141646 torch/distributed/run.py:766] ***************************************** +W0726 03:17:30.011000 3141646 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:17:30.011000 3141646 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:17:57.428408250 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:17:58.636328537 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:17:58.653987631 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:17:58.735304765 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W726 03:17:58.735381592 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:17:58.066243031 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:17:58.068905257 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:17:58.103846661 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n16_seed5_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed5_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:23:50 acc1_pos: 83.594 acc5_pos: 100.000 loss: 0.3701 (0.3701) acc1: 83.5938 (83.5938) acc5: 100.0000 (100.0000) time: 6.0883 data: 4.7298 max mem: 3531 +Test: [ 10/235] eta: 0:02:23 acc1_pos: 83.594 acc5_pos: 100.000 loss: 0.3835 (0.3800) acc1: 83.5938 (83.3570) acc5: 100.0000 (100.0000) time: 0.6363 data: 0.4303 max mem: 3531 +Test: [ 20/235] eta: 0:01:21 acc1_pos: 77.344 acc5_pos: 100.000 loss: 0.3756 (0.3783) acc1: 83.5938 (83.7426) acc5: 100.0000 (100.0000) time: 0.0955 data: 0.0004 max mem: 3531 +Test: [ 30/235] eta: 0:01:07 acc1_pos: 84.896 acc5_pos: 100.000 loss: 0.3620 (0.3744) acc1: 83.8542 (83.7954) acc5: 100.0000 (100.0000) time: 0.1603 data: 0.0614 max mem: 3531 +Test: [ 40/235] eta: 0:01:01 acc1_pos: 83.854 acc5_pos: 100.000 loss: 0.3696 (0.3759) acc1: 83.8542 (83.6954) acc5: 100.0000 (100.0000) time: 0.2450 data: 0.1502 max mem: 3531 +Test: [ 50/235] eta: 0:00:55 acc1_pos: 81.250 acc5_pos: 100.000 loss: 0.3805 (0.3773) acc1: 83.8542 (83.7061) acc5: 100.0000 (100.0000) time: 0.2500 data: 0.1585 max mem: 3531 +Test: [ 60/235] eta: 0:00:49 acc1_pos: 85.156 acc5_pos: 100.000 loss: 0.3777 (0.3776) acc1: 84.1146 (83.8072) acc5: 100.0000 (100.0000) time: 0.2160 data: 0.1238 max mem: 3531 +Test: [ 70/235] eta: 0:00:44 acc1_pos: 87.692 acc5_pos: 100.000 loss: 0.4008 (0.4749) acc1: 80.7292 (77.7032) acc5: 100.0000 (100.0000) acc1_neg: 31.250 acc5_neg: 100.000 time: 0.1901 data: 0.0981 max mem: 3531 +Test: [ 80/235] eta: 0:00:40 acc1_pos: 87.692 acc5_pos: 100.000 loss: 1.1578 (0.5601) acc1: 35.4167 (72.5341) acc5: 100.0000 (100.0000) acc1_neg: 34.896 acc5_neg: 100.000 time: 0.1887 data: 0.0955 max mem: 3531 +Test: [ 90/235] eta: 0:00:36 acc1_pos: 87.692 acc5_pos: 100.000 loss: 1.1653 (0.6261) acc1: 34.8958 (68.4095) acc5: 100.0000 (100.0000) acc1_neg: 36.198 acc5_neg: 100.000 time: 0.1846 data: 0.0903 max mem: 3531 +Test: [100/235] eta: 0:00:32 acc1_pos: 87.692 acc5_pos: 100.000 loss: 1.1537 (0.6768) acc1: 35.1562 (65.2924) acc5: 100.0000 (100.0000) acc1_neg: 34.115 acc5_neg: 100.000 time: 0.1703 data: 0.0777 max mem: 3531 +Test: [110/235] eta: 0:00:29 acc1_pos: 87.692 acc5_pos: 100.000 loss: 1.1521 (0.7204) acc1: 36.4583 (62.7346) acc5: 100.0000 (100.0000) acc1_neg: 40.104 acc5_neg: 100.000 time: 0.1846 data: 0.0902 max mem: 3531 +Test: [120/235] eta: 0:00:26 acc1_pos: 87.692 acc5_pos: 100.000 loss: 1.2045 (0.7742) acc1: 32.2917 (59.7602) acc5: 100.0000 (100.0000) acc1_neg: 27.604 acc5_neg: 100.000 time: 0.1949 data: 0.0981 max mem: 3531 +Test: [130/235] eta: 0:00:24 acc1_pos: 87.692 acc5_pos: 100.000 loss: 1.3858 (0.8206) acc1: 25.2604 (57.1346) acc5: 100.0000 (100.0000) acc1_neg: 23.698 acc5_neg: 100.000 time: 0.1817 data: 0.0882 max mem: 3531 +Test: [140/235] eta: 0:00:21 acc1_pos: 87.692 acc5_pos: 100.000 loss: 1.4165 (0.8637) acc1: 23.9583 (54.7337) acc5: 100.0000 (100.0000) acc1_neg: 21.354 acc5_neg: 100.000 time: 0.1800 data: 0.0894 max mem: 3531 +Test: [150/235] eta: 0:00:18 acc1_pos: 87.692 acc5_pos: 100.000 loss: 1.4165 (0.8987) acc1: 24.7396 (52.8059) acc5: 100.0000 (100.0000) acc1_neg: 26.302 acc5_neg: 100.000 time: 0.1853 data: 0.0950 max mem: 3531 +Test: [160/235] eta: 0:00:16 acc1_pos: 87.692 acc5_pos: 100.000 loss: 1.4005 (0.9308) acc1: 26.0417 (51.0627) acc5: 100.0000 (100.0000) acc1_neg: 26.562 acc5_neg: 100.000 time: 0.1899 data: 0.0995 max mem: 3531 +Test: [170/235] eta: 0:00:14 acc1_pos: 87.692 acc5_pos: 100.000 loss: 1.3177 (0.9218) acc1: 26.8229 (51.3843) acc5: 100.0000 (100.0000) acc1_neg: 56.771 acc5_neg: 100.000 time: 0.1954 data: 0.1045 max mem: 3531 +Test: [180/235] eta: 0:00:11 acc1_pos: 87.692 acc5_pos: 100.000 loss: 0.7117 (0.9106) acc1: 59.6354 (51.8646) acc5: 100.0000 (100.0000) acc1_neg: 59.375 acc5_neg: 100.000 time: 0.1847 data: 0.0905 max mem: 3531 +Test: [190/235] eta: 0:00:09 acc1_pos: 87.692 acc5_pos: 100.000 loss: 0.7146 (0.9005) acc1: 59.8958 (52.2688) acc5: 100.0000 (100.0000) acc1_neg: 60.938 acc5_neg: 100.000 time: 0.1647 data: 0.0705 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 87.692 acc5_pos: 100.000 loss: 0.7240 (0.8918) acc1: 59.6354 (52.6068) acc5: 100.0000 (100.0000) acc1_neg: 60.156 acc5_neg: 100.000 time: 0.1748 data: 0.0816 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 87.692 acc5_pos: 100.000 loss: 0.7155 (0.8829) acc1: 60.1562 (52.9843) acc5: 100.0000 (100.0000) acc1_neg: 60.417 acc5_neg: 100.000 time: 0.1875 data: 0.0917 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 87.692 acc5_pos: 100.000 loss: 0.7155 (0.8753) acc1: 60.4167 (53.3088) acc5: 100.0000 (100.0000) acc1_neg: 61.458 acc5_neg: 100.000 time: 0.1448 data: 0.0504 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 87.692 acc5_pos: 100.000 loss: 0.7189 (0.8688) acc1: 59.6354 (53.5669) acc5: 100.0000 (100.0000) acc1_neg: 60.417 acc5_neg: 100.000 time: 0.1158 data: 0.0243 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 87.692 acc5_pos: 100.000 loss: 0.7189 (0.8669) acc1: 59.6354 (53.6532) acc5: 100.0000 (100.0000) acc1_neg: 47.826 acc5_neg: 100.000 time: 0.1077 data: 0.0207 max mem: 3532 +Test: Total time: 0:00:47 (0.2023 s / it) +* Overall Acc@1 53.653 Acc@5 100.000 + target == real Acc@1 83.768 Acc@5 100.000 + target == fake Acc@1 42.761 Acc@5 100.000 + loss 0.867 + PR-AUC 0.8638 ROC-AUC 0.7037 EER 34.75% (thr=0.2825) +Accuracy of the network on the 89879 test images: 53.7% +[rank0]:[W726 03:18:58.814851272 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:19:18.311000 3142196 torch/distributed/run.py:766] +W0726 03:19:18.311000 3142196 torch/distributed/run.py:766] ***************************************** +W0726 03:19:18.311000 3142196 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:19:18.311000 3142196 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 03:19:44.458905407 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:19:45.596038614 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:19:45.650029032 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:19:45.023591473 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:19:45.076034605 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:19:45.103275916 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:19:45.229316133 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:19:45.235971616 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n16_seed5_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed5_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:38:04 acc1_pos: 65.625 acc5_pos: 100.000 loss: 0.6258 (0.6258) acc1: 65.6250 (65.6250) acc5: 100.0000 (100.0000) time: 8.4911 data: 5.3329 max mem: 3531 +Test: [ 10/269] eta: 0:03:42 acc1_pos: 84.896 acc5_pos: 100.000 loss: 0.4691 (0.5180) acc1: 78.3854 (76.7519) acc5: 100.0000 (100.0000) time: 0.8575 data: 0.4852 max mem: 3531 +Test: [ 20/269] eta: 0:02:03 acc1_pos: 88.542 acc5_pos: 100.000 loss: 0.4235 (0.4872) acc1: 82.8125 (79.4023) acc5: 100.0000 (100.0000) time: 0.0970 data: 0.0004 max mem: 3531 +Test: [ 30/269] eta: 0:01:39 acc1_pos: 60.417 acc5_pos: 100.000 loss: 0.4170 (0.4838) acc1: 85.1562 (80.4940) acc5: 100.0000 (100.0000) time: 0.1714 data: 0.0762 max mem: 3531 +Test: [ 40/269] eta: 0:01:25 acc1_pos: 84.483 acc5_pos: 100.000 loss: 0.5745 (0.5340) acc1: 68.7500 (74.2060) acc5: 100.0000 (100.0000) acc1_neg: 32.812 acc5_neg: 100.000 time: 0.2399 data: 0.1467 max mem: 3531 +Test: [ 50/269] eta: 0:01:13 acc1_pos: 84.483 acc5_pos: 100.000 loss: 0.7352 (0.5833) acc1: 41.6667 (67.7185) acc5: 100.0000 (100.0000) acc1_neg: 27.604 acc5_neg: 100.000 time: 0.2134 data: 0.1194 max mem: 3531 +Test: [ 60/269] eta: 0:01:05 acc1_pos: 84.483 acc5_pos: 100.000 loss: 0.7830 (0.6247) acc1: 39.8438 (62.6067) acc5: 100.0000 (100.0000) acc1_neg: 40.365 acc5_neg: 100.000 time: 0.1949 data: 0.1003 max mem: 3531 +Test: [ 70/269] eta: 0:00:58 acc1_pos: 84.483 acc5_pos: 100.000 loss: 0.7306 (0.6313) acc1: 49.4792 (61.7664) acc5: 100.0000 (100.0000) acc1_neg: 47.917 acc5_neg: 100.000 time: 0.1799 data: 0.0841 max mem: 3531 +Test: [ 80/269] eta: 0:00:52 acc1_pos: 84.483 acc5_pos: 100.000 loss: 0.7063 (0.6446) acc1: 51.5625 (60.1627) acc5: 100.0000 (100.0000) acc1_neg: 35.417 acc5_neg: 100.000 time: 0.1761 data: 0.0824 max mem: 3531 +Test: [ 90/269] eta: 0:00:48 acc1_pos: 84.483 acc5_pos: 100.000 loss: 0.8045 (0.6820) acc1: 35.4167 (56.5476) acc5: 100.0000 (100.0000) acc1_neg: 54.167 acc5_neg: 100.000 time: 0.1999 data: 0.1083 max mem: 3531 +Test: [100/269] eta: 0:00:45 acc1_pos: 84.483 acc5_pos: 100.000 loss: 0.9083 (0.6976) acc1: 32.2917 (54.9557) acc5: 100.0000 (100.0000) acc1_neg: 38.281 acc5_neg: 100.000 time: 0.2192 data: 0.1277 max mem: 3531 +Test: [110/269] eta: 0:00:41 acc1_pos: 84.483 acc5_pos: 100.000 loss: 0.7904 (0.7025) acc1: 42.7083 (54.4717) acc5: 100.0000 (100.0000) acc1_neg: 40.885 acc5_neg: 100.000 time: 0.2209 data: 0.1287 max mem: 3531 +Test: [120/269] eta: 0:00:38 acc1_pos: 84.483 acc5_pos: 100.000 loss: 0.7799 (0.7136) acc1: 46.0938 (53.3295) acc5: 100.0000 (100.0000) acc1_neg: 28.385 acc5_neg: 100.000 time: 0.2168 data: 0.1252 max mem: 3531 +Test: [130/269] eta: 0:00:35 acc1_pos: 84.483 acc5_pos: 100.000 loss: 0.9102 (0.7357) acc1: 32.2917 (51.5605) acc5: 100.0000 (100.0000) acc1_neg: 21.875 acc5_neg: 100.000 time: 0.2291 data: 0.1380 max mem: 3531 +Test: [140/269] eta: 0:00:32 acc1_pos: 84.483 acc5_pos: 100.000 loss: 1.0280 (0.7590) acc1: 25.2604 (49.3443) acc5: 100.0000 (100.0000) acc1_neg: 11.458 acc5_neg: 100.000 time: 0.1902 data: 0.0965 max mem: 3531 +Test: [150/269] eta: 0:00:29 acc1_pos: 84.483 acc5_pos: 100.000 loss: 1.0608 (0.7801) acc1: 15.8854 (47.4441) acc5: 100.0000 (100.0000) acc1_neg: 29.948 acc5_neg: 100.000 time: 0.1547 data: 0.0608 max mem: 3531 +Test: [160/269] eta: 0:00:26 acc1_pos: 84.483 acc5_pos: 100.000 loss: 1.0277 (0.7904) acc1: 27.3438 (46.4868) acc5: 100.0000 (100.0000) acc1_neg: 27.344 acc5_neg: 100.000 time: 0.1737 data: 0.0807 max mem: 3531 +Test: [170/269] eta: 0:00:23 acc1_pos: 84.483 acc5_pos: 100.000 loss: 0.9109 (0.7992) acc1: 29.1667 (45.4206) acc5: 100.0000 (100.0000) acc1_neg: 41.146 acc5_neg: 100.000 time: 0.1816 data: 0.0892 max mem: 3531 +Test: [180/269] eta: 0:00:20 acc1_pos: 84.483 acc5_pos: 100.000 loss: 0.8129 (0.7964) acc1: 41.1458 (45.6549) acc5: 100.0000 (100.0000) acc1_neg: 31.250 acc5_neg: 100.000 time: 0.1986 data: 0.1056 max mem: 3531 +Test: [190/269] eta: 0:00:18 acc1_pos: 84.483 acc5_pos: 100.000 loss: 0.7813 (0.7993) acc1: 43.4896 (45.3520) acc5: 100.0000 (100.0000) acc1_neg: 38.021 acc5_neg: 100.000 time: 0.1836 data: 0.0891 max mem: 3531 +Test: [200/269] eta: 0:00:15 acc1_pos: 84.483 acc5_pos: 100.000 loss: 0.9062 (0.8069) acc1: 37.2396 (45.0145) acc5: 100.0000 (100.0000) acc1_neg: 64.844 acc5_neg: 100.000 time: 0.1620 data: 0.0698 max mem: 3531 +Test: [210/269] eta: 0:00:13 acc1_pos: 84.483 acc5_pos: 100.000 loss: 0.9168 (0.8095) acc1: 34.3750 (44.6917) acc5: 100.0000 (100.0000) acc1_neg: 34.375 acc5_neg: 100.000 time: 0.1759 data: 0.0849 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 84.483 acc5_pos: 100.000 loss: 0.8786 (0.8158) acc1: 34.3750 (44.0953) acc5: 100.0000 (100.0000) acc1_neg: 33.073 acc5_neg: 100.000 time: 0.1732 data: 0.0829 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 84.483 acc5_pos: 100.000 loss: 0.9150 (0.8198) acc1: 27.6042 (43.5989) acc5: 100.0000 (100.0000) acc1_neg: 52.083 acc5_neg: 100.000 time: 0.1547 data: 0.0645 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 84.483 acc5_pos: 100.000 loss: 0.8431 (0.8198) acc1: 38.5417 (43.5868) acc5: 100.0000 (100.0000) acc1_neg: 37.760 acc5_neg: 100.000 time: 0.1458 data: 0.0540 max mem: 3532 +Test: [250/269] eta: 0:00:04 acc1_pos: 84.483 acc5_pos: 100.000 loss: 0.8581 (0.8231) acc1: 38.5417 (43.2800) acc5: 100.0000 (100.0000) acc1_neg: 18.490 acc5_neg: 100.000 time: 0.1274 data: 0.0358 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 84.483 acc5_pos: 100.000 loss: 0.8873 (0.8271) acc1: 35.6771 (42.9258) acc5: 100.0000 (100.0000) acc1_neg: 58.854 acc5_neg: 100.000 time: 0.1097 data: 0.0198 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 84.483 acc5_pos: 100.000 loss: 0.7969 (0.8243) acc1: 39.8438 (43.1326) acc5: 100.0000 (100.0000) acc1_neg: 74.278 acc5_neg: 100.000 time: 0.1203 data: 0.0271 max mem: 3532 +Test: Total time: 0:00:55 (0.2067 s / it) +* Overall Acc@1 43.133 Acc@5 100.000 + target == real Acc@1 80.168 Acc@5 100.000 + target == fake Acc@1 38.078 Acc@5 100.000 + loss 0.824 + PR-AUC 0.9273 ROC-AUC 0.6578 EER 38.85% (thr=0.3755) +Accuracy of the network on the 103293 test images: 43.1% +[rank0]:[W726 03:20:48.450758087 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T03:20:54Z] eval_done tag=fixed500k_real_n16_seed5_resampled +[2026-07-26T03:20:55Z] eval_start tag=fixed500k_fake_n32_seed5_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed5_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T03:20:55Z +hostname=n111-046-081 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed5_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n32_seed5_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 03:21:07.817000 3142774 torch/distributed/run.py:766] +W0726 03:21:07.817000 3142774 torch/distributed/run.py:766] ***************************************** +W0726 03:21:07.817000 3142774 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:21:07.817000 3142774 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:21:36.062437229 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:21:36.485048761 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:21:37.757992736 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:21:37.884169835 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:21:37.894354007 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +[rank0]:[W726 03:21:37.926385069 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:21:37.933229173 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:21:37.959813144 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n32_seed5_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed5_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:17:40 acc1_pos: 76.562 acc5_pos: 100.000 loss: 0.5468 (0.5468) acc1: 76.5625 (76.5625) acc5: 100.0000 (100.0000) time: 4.5310 data: 2.6620 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/234] eta: 0:01:56 acc1_pos: 54.688 acc5_pos: 100.000 loss: 0.5468 (0.5909) acc1: 76.5625 (68.3949) acc5: 100.0000 (100.0000) time: 0.5205 data: 0.2692 max mem: 3531 +Test: [ 20/234] eta: 0:01:12 acc1_pos: 96.354 acc5_pos: 100.000 loss: 0.6107 (0.5853) acc1: 60.9375 (66.2574) acc5: 100.0000 (100.0000) time: 0.1307 data: 0.0394 max mem: 3531 +Test: [ 30/234] eta: 0:01:04 acc1_pos: 78.385 acc5_pos: 100.000 loss: 0.2229 (0.4750) acc1: 95.3125 (74.9412) acc5: 100.0000 (100.0000) time: 0.2017 data: 0.1068 max mem: 3531 +Test: [ 40/234] eta: 0:00:59 acc1_pos: 99.479 acc5_pos: 100.000 loss: 0.2124 (0.4519) acc1: 95.5729 (75.9718) acc5: 100.0000 (100.0000) time: 0.2674 data: 0.1742 max mem: 3531 +Test: [ 50/234] eta: 0:00:52 acc1_pos: 99.740 acc5_pos: 100.000 loss: 0.1098 (0.3840) acc1: 99.4792 (80.5964) acc5: 100.0000 (100.0000) time: 0.2348 data: 0.1452 max mem: 3531 +Test: [ 60/234] eta: 0:00:44 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.0851 (0.3334) acc1: 99.7396 (83.7432) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1489 data: 0.0580 max mem: 3531 +Test: [ 70/234] eta: 0:00:39 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.0813 (0.3843) acc1: 99.7396 (80.2340) acc5: 100.0000 (100.0000) acc1_neg: 35.417 acc5_neg: 100.000 time: 0.1392 data: 0.0480 max mem: 3531 +Test: [ 80/234] eta: 0:00:36 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.9659 (0.4240) acc1: 40.6250 (77.7328) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1850 data: 0.0906 max mem: 3531 +Test: [ 90/234] eta: 0:00:33 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.1899 (0.3980) acc1: 93.2292 (79.5988) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.1899 data: 0.0945 max mem: 3531 +Test: [100/234] eta: 0:00:29 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.2449 (0.4028) acc1: 89.5833 (79.4606) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1603 data: 0.0670 max mem: 3531 +Test: [110/234] eta: 0:00:26 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.4806 (0.4096) acc1: 73.4375 (79.3145) acc5: 100.0000 (100.0000) acc1_neg: 70.833 acc5_neg: 100.000 time: 0.1563 data: 0.0624 max mem: 3531 +Test: [120/234] eta: 0:00:24 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.4806 (0.4256) acc1: 74.7396 (78.6609) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.1787 data: 0.0846 max mem: 3531 +Test: [130/234] eta: 0:00:21 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.4766 (0.4298) acc1: 81.5104 (78.8864) acc5: 100.0000 (100.0000) acc1_neg: 66.146 acc5_neg: 100.000 time: 0.1735 data: 0.0773 max mem: 3531 +Test: [140/234] eta: 0:00:19 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.5181 (0.4685) acc1: 75.7812 (76.7361) acc5: 100.0000 (100.0000) acc1_neg: 23.438 acc5_neg: 100.000 time: 0.1609 data: 0.0630 max mem: 3531 +Test: [150/234] eta: 0:00:17 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.5528 (0.4793) acc1: 67.7083 (76.2228) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.1674 data: 0.0723 max mem: 3531 +Test: [160/234] eta: 0:00:15 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.4546 (0.4797) acc1: 71.0938 (76.3328) acc5: 100.0000 (100.0000) acc1_neg: 65.625 acc5_neg: 100.000 time: 0.1784 data: 0.0863 max mem: 3531 +Test: [170/234] eta: 0:00:12 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.6107 (0.4870) acc1: 66.6667 (75.9107) acc5: 100.0000 (100.0000) acc1_neg: 71.094 acc5_neg: 100.000 time: 0.1694 data: 0.0761 max mem: 3531 +Test: [180/234] eta: 0:00:10 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.6115 (0.5016) acc1: 66.6667 (75.0633) acc5: 100.0000 (100.0000) acc1_neg: 22.917 acc5_neg: 100.000 time: 0.1647 data: 0.0724 max mem: 3531 +Test: [190/234] eta: 0:00:08 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.8000 (0.5198) acc1: 57.5521 (74.0470) acc5: 100.0000 (100.0000) acc1_neg: 56.510 acc5_neg: 100.000 time: 0.1738 data: 0.0801 max mem: 3531 +Test: [200/234] eta: 0:00:06 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.7978 (0.5118) acc1: 58.3333 (74.5634) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1720 data: 0.0757 max mem: 3531 +Test: [210/234] eta: 0:00:04 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.1003 (0.4941) acc1: 99.7396 (75.7097) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1607 data: 0.0680 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.2006 (0.4839) acc1: 96.0938 (76.4529) acc5: 100.0000 (100.0000) acc1_neg: 65.104 acc5_neg: 100.000 time: 0.1741 data: 0.0826 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.7634 (0.5417) acc1: 5.2083 (73.2955) acc5: 100.0000 (100.0000) acc1_neg: 4.948 acc5_neg: 100.000 time: 0.1545 data: 0.0627 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 98.621 acc5_pos: 100.000 loss: 1.7685 (0.5577) acc1: 4.4271 (72.5039) acc5: 100.0000 (100.0000) acc1_neg: 5.058 acc5_neg: 100.000 time: 0.1555 data: 0.0626 max mem: 3532 +Test: Total time: 0:00:44 (0.1905 s / it) +* Overall Acc@1 72.504 Acc@5 100.000 + target == real Acc@1 80.729 Acc@5 100.000 + target == fake Acc@1 70.186 Acc@5 100.000 + loss 0.558 + PR-AUC 0.9529 ROC-AUC 0.8406 EER 26.08% (thr=0.4380) +Accuracy of the network on the 89729 test images: 72.5% +[rank0]:[W726 03:22:34.260365928 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:22:53.122000 3143333 torch/distributed/run.py:766] +W0726 03:22:53.122000 3143333 torch/distributed/run.py:766] ***************************************** +W0726 03:22:53.122000 3143333 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:22:53.122000 3143333 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:23:22.004164092 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:23:22.217786168 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:23:22.452968041 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:23:23.661098745 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:23:23.676856176 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +| distributed init (rank 1): env:// +[rank0]:[W726 03:23:23.776742440 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:23:23.777406146 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:23:23.780861740 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n32_seed5_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed5_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:47:35 acc1_pos: 39.583 acc5_pos: 100.000 loss: 0.8342 (0.8342) acc1: 39.5833 (39.5833) acc5: 100.0000 (100.0000) time: 7.2290 data: 5.4916 max mem: 3531 +Test: [ 10/2387] eta: 0:29:28 acc1_pos: 11.458 acc5_pos: 100.000 loss: 0.9668 (0.9813) acc1: 27.6042 (28.8116) acc5: 100.0000 (100.0000) time: 0.7439 data: 0.4996 max mem: 3531 +Test: [ 20/2387] eta: 0:17:14 acc1_pos: 19.531 acc5_pos: 100.000 loss: 0.9668 (0.9904) acc1: 25.0000 (26.9097) acc5: 100.0000 (100.0000) time: 0.0977 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:15:56 acc1_pos: 12.760 acc5_pos: 100.000 loss: 1.0178 (1.0114) acc1: 21.8750 (23.8827) acc5: 100.0000 (100.0000) time: 0.2199 data: 0.1245 max mem: 3531 +Test: [ 40/2387] eta: 0:14:29 acc1_pos: 21.875 acc5_pos: 100.000 loss: 1.0930 (1.0328) acc1: 14.0625 (21.4685) acc5: 100.0000 (100.0000) time: 0.2999 data: 0.2083 max mem: 3531 +Test: [ 50/2387] eta: 0:13:39 acc1_pos: 36.458 acc5_pos: 100.000 loss: 1.0204 (1.0120) acc1: 19.2708 (23.5345) acc5: 100.0000 (100.0000) time: 0.2649 data: 0.1681 max mem: 3531 +Test: [ 60/2387] eta: 0:12:38 acc1_pos: 55.208 acc5_pos: 100.000 loss: 0.8876 (0.9968) acc1: 30.4688 (24.8634) acc5: 100.0000 (100.0000) time: 0.2349 data: 0.1347 max mem: 3531 +Test: [ 70/2387] eta: 0:12:03 acc1_pos: 30.469 acc5_pos: 100.000 loss: 0.9664 (0.9973) acc1: 27.8646 (24.7322) acc5: 100.0000 (100.0000) time: 0.2149 data: 0.1194 max mem: 3531 +Test: [ 80/2387] eta: 0:11:37 acc1_pos: 70.052 acc5_pos: 100.000 loss: 0.9602 (0.9708) acc1: 30.4688 (27.7006) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1369 max mem: 3531 +Test: [ 90/2387] eta: 0:11:20 acc1_pos: 53.385 acc5_pos: 100.000 loss: 0.7165 (0.9362) acc1: 53.3854 (30.8980) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1452 max mem: 3531 +Test: [ 100/2387] eta: 0:11:02 acc1_pos: 62.500 acc5_pos: 100.000 loss: 0.6404 (0.9145) acc1: 57.8125 (33.1606) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1431 max mem: 3531 +Test: [ 110/2387] eta: 0:10:43 acc1_pos: 62.500 acc5_pos: 100.000 loss: 0.6491 (0.8906) acc1: 60.4167 (35.7099) acc5: 100.0000 (100.0000) time: 0.2199 data: 0.1241 max mem: 3531 +Test: [ 120/2387] eta: 0:10:30 acc1_pos: 37.760 acc5_pos: 100.000 loss: 0.6683 (0.8758) acc1: 57.8125 (36.9770) acc5: 100.0000 (100.0000) time: 0.2199 data: 0.1241 max mem: 3531 +Test: [ 130/2387] eta: 0:10:18 acc1_pos: 51.302 acc5_pos: 100.000 loss: 0.6883 (0.8615) acc1: 52.6042 (38.4363) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1289 max mem: 3531 +Test: [ 140/2387] eta: 0:10:08 acc1_pos: 59.635 acc5_pos: 100.000 loss: 0.6807 (0.8491) acc1: 56.2500 (39.5981) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1312 max mem: 3531 +Test: [ 150/2387] eta: 0:10:02 acc1_pos: 38.281 acc5_pos: 100.000 loss: 0.6747 (0.8368) acc1: 56.2500 (40.6716) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1462 max mem: 3531 +Test: [ 160/2387] eta: 0:09:48 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.6738 (0.8264) acc1: 49.4792 (41.7492) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2199 data: 0.1236 max mem: 3531 +Test: [ 170/2387] eta: 0:09:37 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2645 (0.7927) acc1: 96.0938 (44.9973) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1949 data: 0.0974 max mem: 3531 +Test: [ 180/2387] eta: 0:09:26 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2511 (0.7628) acc1: 97.1354 (47.8778) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1948 data: 0.0980 max mem: 3531 +Test: [ 190/2387] eta: 0:09:16 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2464 (0.7358) acc1: 97.1354 (50.4581) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1899 data: 0.0932 max mem: 3531 +Test: [ 200/2387] eta: 0:09:05 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2449 (0.7116) acc1: 97.3958 (52.7920) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1849 data: 0.0881 max mem: 3531 +Test: [ 210/2387] eta: 0:08:52 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2522 (0.6900) acc1: 97.3958 (54.8961) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1599 data: 0.0620 max mem: 3532 +Test: [ 220/2387] eta: 0:08:43 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2546 (0.6703) acc1: 97.3958 (56.8227) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1599 data: 0.0610 max mem: 3532 +Test: [ 230/2387] eta: 0:08:34 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2546 (0.6522) acc1: 97.3958 (58.5780) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1750 data: 0.0776 max mem: 3532 +Test: [ 240/2387] eta: 0:08:27 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2581 (0.6389) acc1: 96.6146 (59.9920) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1805 data: 0.0876 max mem: 3532 +Test: [ 250/2387] eta: 0:08:21 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2428 (0.6181) acc1: 93.2292 (61.5559) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1948 data: 0.1003 max mem: 3532 +Test: [ 260/2387] eta: 0:08:12 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.0897 (0.5977) acc1: 100.0000 (63.0288) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1703 data: 0.0753 max mem: 3532 +Test: [ 270/2387] eta: 0:08:05 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.0828 (0.5787) acc1: 100.0000 (64.3921) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1600 data: 0.0689 max mem: 3532 +Test: [ 280/2387] eta: 0:07:59 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.0847 (0.5613) acc1: 100.0000 (65.6593) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1793 data: 0.0864 max mem: 3532 +Test: [ 290/2387] eta: 0:07:53 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.0847 (0.5447) acc1: 100.0000 (66.8394) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1797 data: 0.0839 max mem: 3532 +Test: [ 300/2387] eta: 0:07:47 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.0820 (0.5296) acc1: 100.0000 (67.9411) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1766 data: 0.0834 max mem: 3532 +Test: [ 310/2387] eta: 0:07:43 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.0929 (0.5196) acc1: 100.0000 (68.9267) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1800 data: 0.0894 max mem: 3532 +Test: [ 320/2387] eta: 0:07:39 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.3585 (0.5147) acc1: 97.9167 (69.8209) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1940 data: 0.1020 max mem: 3532 +Test: [ 330/2387] eta: 0:07:37 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.3585 (0.5100) acc1: 97.6562 (70.6713) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2149 data: 0.1205 max mem: 3532 +Test: [ 340/2387] eta: 0:07:35 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.3554 (0.5055) acc1: 98.1771 (71.4794) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2259 data: 0.1326 max mem: 3532 +Test: [ 350/2387] eta: 0:07:28 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.3495 (0.4963) acc1: 98.9583 (72.2838) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0882 max mem: 3532 +Test: [ 360/2387] eta: 0:07:21 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2007 (0.4883) acc1: 99.2188 (73.0148) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1377 data: 0.0426 max mem: 3532 +Test: [ 370/2387] eta: 0:07:17 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2092 (0.4808) acc1: 98.4375 (73.6986) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1627 data: 0.0680 max mem: 3532 +Test: [ 380/2387] eta: 0:07:13 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2092 (0.4737) acc1: 98.4375 (74.3479) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1799 data: 0.0870 max mem: 3532 +Test: [ 390/2387] eta: 0:07:08 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2085 (0.4670) acc1: 98.6979 (74.9680) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1740 data: 0.0811 max mem: 3532 +Test: [ 400/2387] eta: 0:07:03 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2098 (0.4605) acc1: 98.6979 (75.5643) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1660 data: 0.0748 max mem: 3532 +Test: [ 410/2387] eta: 0:07:00 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2066 (0.4544) acc1: 98.6979 (76.1209) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1748 data: 0.0836 max mem: 3532 +Test: [ 420/2387] eta: 0:06:56 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2052 (0.4485) acc1: 98.6979 (76.6522) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1838 data: 0.0923 max mem: 3532 +Test: [ 430/2387] eta: 0:06:52 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2124 (0.4442) acc1: 98.1771 (77.1154) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1706 data: 0.0793 max mem: 3532 +Test: [ 440/2387] eta: 0:06:49 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2192 (0.4379) acc1: 96.0938 (77.5870) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1712 data: 0.0762 max mem: 3533 +Test: [ 450/2387] eta: 0:06:45 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.0841 (0.4300) acc1: 100.0000 (78.0840) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1741 data: 0.0792 max mem: 3533 +Test: [ 460/2387] eta: 0:06:42 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.0839 (0.4225) acc1: 100.0000 (78.5594) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1841 data: 0.0926 max mem: 3533 +Test: [ 470/2387] eta: 0:06:40 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.0835 (0.4153) acc1: 100.0000 (79.0146) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1962 data: 0.1037 max mem: 3533 +Test: [ 480/2387] eta: 0:06:37 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.0837 (0.4084) acc1: 100.0000 (79.4509) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1974 data: 0.1057 max mem: 3533 +Test: [ 490/2387] eta: 0:06:34 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.0845 (0.4018) acc1: 100.0000 (79.8694) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1881 data: 0.0972 max mem: 3533 +Test: [ 500/2387] eta: 0:06:31 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.0840 (0.3955) acc1: 100.0000 (80.2712) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1847 data: 0.0936 max mem: 3533 +Test: [ 510/2387] eta: 0:06:28 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.0840 (0.3896) acc1: 100.0000 (80.6573) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0994 max mem: 3533 +Test: [ 520/2387] eta: 0:06:25 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1174 (0.3845) acc1: 100.0000 (81.0286) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1837 data: 0.0932 max mem: 3533 +Test: [ 530/2387] eta: 0:06:22 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1121 (0.3793) acc1: 100.0000 (81.3858) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1705 data: 0.0800 max mem: 3533 +Test: [ 540/2387] eta: 0:06:18 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1109 (0.3744) acc1: 100.0000 (81.7299) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1634 data: 0.0725 max mem: 3533 +Test: [ 550/2387] eta: 0:06:16 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1109 (0.3696) acc1: 100.0000 (82.0615) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1786 data: 0.0873 max mem: 3533 +Test: [ 560/2387] eta: 0:06:12 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1108 (0.3650) acc1: 100.0000 (82.3813) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1779 data: 0.0876 max mem: 3533 +Test: [ 570/2387] eta: 0:06:09 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1105 (0.3605) acc1: 100.0000 (82.6898) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1619 data: 0.0711 max mem: 3533 +Test: [ 580/2387] eta: 0:06:05 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1102 (0.3562) acc1: 100.0000 (82.9878) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1572 data: 0.0655 max mem: 3533 +Test: [ 590/2387] eta: 0:06:02 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1111 (0.3521) acc1: 100.0000 (83.2756) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1627 data: 0.0710 max mem: 3533 +Test: [ 600/2387] eta: 0:05:59 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1154 (0.3484) acc1: 100.0000 (83.5522) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1652 data: 0.0735 max mem: 3533 +Test: [ 610/2387] eta: 0:05:57 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1562 (0.3464) acc1: 98.4375 (83.7868) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1815 data: 0.0895 max mem: 3533 +Test: [ 620/2387] eta: 0:05:54 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2251 (0.3444) acc1: 97.9167 (84.0156) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1743 data: 0.0808 max mem: 3533 +Test: [ 630/2387] eta: 0:05:52 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2242 (0.3426) acc1: 97.9167 (84.2363) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1781 data: 0.0848 max mem: 3533 +Test: [ 640/2387] eta: 0:05:48 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2263 (0.3408) acc1: 97.6562 (84.4473) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1711 data: 0.0757 max mem: 3533 +Test: [ 650/2387] eta: 0:05:44 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2271 (0.3391) acc1: 97.9167 (84.6554) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1365 data: 0.0407 max mem: 3533 +Test: [ 660/2387] eta: 0:05:42 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2281 (0.3374) acc1: 97.9167 (84.8572) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1549 data: 0.0626 max mem: 3533 +Test: [ 670/2387] eta: 0:05:39 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2260 (0.3357) acc1: 98.1771 (85.0604) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1691 data: 0.0772 max mem: 3534 +Test: [ 680/2387] eta: 0:05:36 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2313 (0.3348) acc1: 98.4375 (85.2526) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1687 data: 0.0761 max mem: 3534 +Test: [ 690/2387] eta: 0:05:33 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2907 (0.3375) acc1: 97.1354 (85.1031) acc5: 100.0000 (100.0000) acc1_neg: 63.802 acc5_neg: 100.000 time: 0.1650 data: 0.0726 max mem: 3534 +Test: [ 700/2387] eta: 0:05:30 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.5999 (0.3412) acc1: 66.9271 (84.8461) acc5: 100.0000 (100.0000) acc1_neg: 71.615 acc5_neg: 100.000 time: 0.1579 data: 0.0658 max mem: 3534 +Test: [ 710/2387] eta: 0:05:28 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.5999 (0.3449) acc1: 66.6667 (84.5856) acc5: 100.0000 (100.0000) acc1_neg: 68.750 acc5_neg: 100.000 time: 0.1640 data: 0.0721 max mem: 3534 +Test: [ 720/2387] eta: 0:05:25 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.5984 (0.3483) acc1: 66.6667 (84.3447) acc5: 100.0000 (100.0000) acc1_neg: 67.969 acc5_neg: 100.000 time: 0.1724 data: 0.0814 max mem: 3534 +Test: [ 730/2387] eta: 0:05:23 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.6054 (0.3519) acc1: 66.1458 (84.0943) acc5: 100.0000 (100.0000) acc1_neg: 64.844 acc5_neg: 100.000 time: 0.1776 data: 0.0830 max mem: 3534 +Test: [ 740/2387] eta: 0:05:21 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.5992 (0.3553) acc1: 65.6250 (83.8654) acc5: 100.0000 (100.0000) acc1_neg: 64.062 acc5_neg: 100.000 time: 0.1844 data: 0.0880 max mem: 3534 +Test: [ 750/2387] eta: 0:05:19 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.5926 (0.3584) acc1: 66.4062 (83.6472) acc5: 100.0000 (100.0000) acc1_neg: 70.573 acc5_neg: 100.000 time: 0.1893 data: 0.0948 max mem: 3534 +Test: [ 760/2387] eta: 0:05:17 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.5819 (0.3611) acc1: 67.7083 (83.4771) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.1849 data: 0.0893 max mem: 3534 +Test: [ 770/2387] eta: 0:05:14 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.4926 (0.3609) acc1: 75.2604 (83.5073) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1678 data: 0.0732 max mem: 3534 +Test: [ 780/2387] eta: 0:05:12 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1840 (0.3585) acc1: 97.1354 (83.6881) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1774 data: 0.0836 max mem: 3534 +Test: [ 790/2387] eta: 0:05:10 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1505 (0.3562) acc1: 98.9583 (83.8361) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1825 data: 0.0899 max mem: 3534 +Test: [ 800/2387] eta: 0:05:07 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1361 (0.3535) acc1: 99.7396 (84.0366) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1681 data: 0.0779 max mem: 3534 +Test: [ 810/2387] eta: 0:05:05 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1416 (0.3510) acc1: 100.0000 (84.2283) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1793 data: 0.0866 max mem: 3534 +Test: [ 820/2387] eta: 0:05:03 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1553 (0.3489) acc1: 99.7396 (84.4013) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1821 data: 0.0886 max mem: 3534 +Test: [ 830/2387] eta: 0:05:00 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2244 (0.3481) acc1: 96.3542 (84.5057) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1732 data: 0.0816 max mem: 3534 +Test: [ 840/2387] eta: 0:04:58 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2573 (0.3469) acc1: 96.3542 (84.6642) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1796 data: 0.0879 max mem: 3534 +Test: [ 850/2387] eta: 0:04:56 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2309 (0.3455) acc1: 98.4375 (84.8132) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1795 data: 0.0875 max mem: 3534 +Test: [ 860/2387] eta: 0:04:54 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2142 (0.3438) acc1: 98.9583 (84.9751) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1682 data: 0.0753 max mem: 3534 +Test: [ 870/2387] eta: 0:04:51 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1955 (0.3423) acc1: 99.7396 (85.1326) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1629 data: 0.0706 max mem: 3534 +Test: [ 880/2387] eta: 0:04:49 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2016 (0.3409) acc1: 99.2188 (85.2777) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1713 data: 0.0803 max mem: 3534 +Test: [ 890/2387] eta: 0:04:47 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2042 (0.3395) acc1: 99.2188 (85.4327) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1849 data: 0.0933 max mem: 3534 +Test: [ 900/2387] eta: 0:04:45 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2542 (0.3447) acc1: 95.3125 (85.0450) acc5: 100.0000 (100.0000) acc1_neg: 41.146 acc5_neg: 100.000 time: 0.1879 data: 0.0955 max mem: 3535 +Test: [ 910/2387] eta: 0:04:43 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.9474 (0.3514) acc1: 38.5417 (84.5317) acc5: 100.0000 (100.0000) acc1_neg: 39.844 acc5_neg: 100.000 time: 0.1919 data: 0.0980 max mem: 3535 +Test: [ 920/2387] eta: 0:04:41 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.9519 (0.3580) acc1: 38.2812 (84.0193) acc5: 100.0000 (100.0000) acc1_neg: 33.073 acc5_neg: 100.000 time: 0.1996 data: 0.1063 max mem: 3535 +Test: [ 930/2387] eta: 0:04:40 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.9494 (0.3643) acc1: 37.7604 (83.5386) acc5: 100.0000 (100.0000) acc1_neg: 36.458 acc5_neg: 100.000 time: 0.1999 data: 0.1057 max mem: 3535 +Test: [ 940/2387] eta: 0:04:38 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.9590 (0.3706) acc1: 38.0208 (83.0668) acc5: 100.0000 (100.0000) acc1_neg: 40.365 acc5_neg: 100.000 time: 0.1917 data: 0.0965 max mem: 3535 +Test: [ 950/2387] eta: 0:04:35 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.9590 (0.3767) acc1: 37.7604 (82.5880) acc5: 100.0000 (100.0000) acc1_neg: 40.365 acc5_neg: 100.000 time: 0.1807 data: 0.0889 max mem: 3535 +Test: [ 960/2387] eta: 0:04:34 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.9394 (0.3826) acc1: 37.7604 (82.1353) acc5: 100.0000 (100.0000) acc1_neg: 42.188 acc5_neg: 100.000 time: 0.1881 data: 0.0956 max mem: 3535 +Test: [ 970/2387] eta: 0:04:31 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.9479 (0.3885) acc1: 38.5417 (81.6880) acc5: 100.0000 (100.0000) acc1_neg: 41.146 acc5_neg: 100.000 time: 0.1850 data: 0.0918 max mem: 3535 +Test: [ 980/2387] eta: 0:04:29 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.9266 (0.3894) acc1: 39.8438 (81.6325) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.1799 data: 0.0889 max mem: 3535 +Test: [ 990/2387] eta: 0:04:27 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.3369 (0.3889) acc1: 88.2812 (81.7101) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1849 data: 0.0936 max mem: 3535 +Test: [1000/2387] eta: 0:04:25 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.3414 (0.3887) acc1: 88.0208 (81.7615) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1811 data: 0.0895 max mem: 3535 +Test: [1010/2387] eta: 0:04:23 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.3738 (0.3887) acc1: 84.6354 (81.7917) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.1890 data: 0.0959 max mem: 3535 +Test: [1020/2387] eta: 0:04:21 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.3820 (0.3880) acc1: 84.6354 (81.8517) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1901 data: 0.0911 max mem: 3535 +Test: [1030/2387] eta: 0:04:19 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1886 (0.3860) acc1: 97.3958 (82.0110) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1872 data: 0.0885 max mem: 3535 +Test: [1040/2387] eta: 0:04:17 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1615 (0.3837) acc1: 99.2188 (82.1781) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1836 data: 0.0897 max mem: 3535 +Test: [1050/2387] eta: 0:04:15 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1502 (0.3815) acc1: 99.7396 (82.3435) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1740 data: 0.0814 max mem: 3535 +Test: [1060/2387] eta: 0:04:13 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1510 (0.3794) acc1: 99.7396 (82.5045) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1757 data: 0.0821 max mem: 3535 +Test: [1070/2387] eta: 0:04:11 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1646 (0.3779) acc1: 99.2188 (82.6425) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1908 data: 0.0973 max mem: 3535 +Test: [1080/2387] eta: 0:04:09 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2605 (0.3771) acc1: 95.0521 (82.7424) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1811 data: 0.0896 max mem: 3535 +Test: [1090/2387] eta: 0:04:07 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2756 (0.3758) acc1: 92.9688 (82.8481) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1825 data: 0.0894 max mem: 3535 +Test: [1100/2387] eta: 0:04:05 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2345 (0.3746) acc1: 95.5729 (82.9587) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1758 data: 0.0800 max mem: 3535 +Test: [1110/2387] eta: 0:04:03 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2268 (0.3736) acc1: 95.8333 (83.0389) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1550 data: 0.0615 max mem: 3535 +Test: [1120/2387] eta: 0:04:01 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2199 (0.3722) acc1: 96.8750 (83.1640) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1874 data: 0.0963 max mem: 3536 +Test: [1130/2387] eta: 0:03:59 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2199 (0.3711) acc1: 97.1354 (83.2691) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2010 data: 0.1099 max mem: 3536 +Test: [1140/2387] eta: 0:03:57 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2751 (0.3705) acc1: 93.7500 (83.3505) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1973 data: 0.1049 max mem: 3536 +Test: [1150/2387] eta: 0:03:56 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2758 (0.3696) acc1: 94.0104 (83.4582) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2060 data: 0.1115 max mem: 3536 +Test: [1160/2387] eta: 0:03:54 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2045 (0.3679) acc1: 98.4375 (83.5832) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1943 data: 0.1010 max mem: 3536 +Test: [1170/2387] eta: 0:03:52 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1781 (0.3663) acc1: 99.2188 (83.7181) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1784 data: 0.0869 max mem: 3536 +Test: [1180/2387] eta: 0:03:50 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1785 (0.3647) acc1: 99.2188 (83.8484) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1805 data: 0.0878 max mem: 3536 +Test: [1190/2387] eta: 0:03:47 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1659 (0.3631) acc1: 99.4792 (83.9806) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1770 data: 0.0838 max mem: 3536 +Test: [1200/2387] eta: 0:03:46 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1617 (0.3615) acc1: 100.0000 (84.1133) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1772 data: 0.0825 max mem: 3536 +Test: [1210/2387] eta: 0:03:44 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1777 (0.3599) acc1: 100.0000 (84.2430) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1859 data: 0.0910 max mem: 3536 +Test: [1220/2387] eta: 0:03:42 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1106 (0.3578) acc1: 100.0000 (84.3709) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1887 data: 0.0946 max mem: 3536 +Test: [1230/2387] eta: 0:03:40 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.0917 (0.3556) acc1: 100.0000 (84.4979) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1838 data: 0.0913 max mem: 3536 +Test: [1240/2387] eta: 0:03:38 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.0848 (0.3534) acc1: 100.0000 (84.6226) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1867 data: 0.0947 max mem: 3536 +Test: [1250/2387] eta: 0:03:35 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.0856 (0.3514) acc1: 100.0000 (84.7449) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1628 data: 0.0695 max mem: 3536 +Test: [1260/2387] eta: 0:03:33 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.0896 (0.3493) acc1: 100.0000 (84.8657) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1431 data: 0.0511 max mem: 3536 +Test: [1270/2387] eta: 0:03:31 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.0909 (0.3473) acc1: 100.0000 (84.9846) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1798 data: 0.0878 max mem: 3536 +Test: [1280/2387] eta: 0:03:29 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.0979 (0.3455) acc1: 100.0000 (85.0977) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1711 data: 0.0770 max mem: 3536 +Test: [1290/2387] eta: 0:03:27 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1044 (0.3437) acc1: 99.7396 (85.1946) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1561 data: 0.0609 max mem: 3536 +Test: [1300/2387] eta: 0:03:25 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1048 (0.3420) acc1: 99.7396 (85.3004) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1702 data: 0.0752 max mem: 3536 +Test: [1310/2387] eta: 0:03:23 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1110 (0.3404) acc1: 99.7396 (85.4018) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1788 data: 0.0857 max mem: 3536 +Test: [1320/2387] eta: 0:03:21 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1237 (0.3388) acc1: 99.4792 (85.4985) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1802 data: 0.0887 max mem: 3536 +Test: [1330/2387] eta: 0:03:19 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1461 (0.3378) acc1: 98.9583 (85.5881) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1819 data: 0.0892 max mem: 3536 +Test: [1340/2387] eta: 0:03:17 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2270 (0.3382) acc1: 95.8333 (85.5590) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1872 data: 0.0954 max mem: 3536 +Test: [1350/2387] eta: 0:03:15 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.0601 (0.3361) acc1: 100.0000 (85.6659) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1740 data: 0.0834 max mem: 3537 +Test: [1360/2387] eta: 0:03:13 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.0401 (0.3338) acc1: 100.0000 (85.7712) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1673 data: 0.0761 max mem: 3537 +Test: [1370/2387] eta: 0:03:11 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.0358 (0.3317) acc1: 100.0000 (85.8750) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1681 data: 0.0764 max mem: 3537 +Test: [1380/2387] eta: 0:03:09 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.0406 (0.3296) acc1: 100.0000 (85.9773) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1795 data: 0.0845 max mem: 3537 +Test: [1390/2387] eta: 0:03:07 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.0438 (0.3276) acc1: 100.0000 (86.0781) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1860 data: 0.0913 max mem: 3537 +Test: [1400/2387] eta: 0:03:05 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.0500 (0.3256) acc1: 100.0000 (86.1775) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1852 data: 0.0922 max mem: 3537 +Test: [1410/2387] eta: 0:03:04 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.0675 (0.3246) acc1: 100.0000 (86.2557) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1897 data: 0.0948 max mem: 3537 +Test: [1420/2387] eta: 0:03:02 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2036 (0.3241) acc1: 96.3542 (86.3029) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1896 data: 0.0959 max mem: 3537 +Test: [1430/2387] eta: 0:03:00 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2104 (0.3233) acc1: 97.1354 (86.3804) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1990 data: 0.1068 max mem: 3537 +Test: [1440/2387] eta: 0:02:58 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2104 (0.3226) acc1: 98.1771 (86.4600) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1949 data: 0.1005 max mem: 3537 +Test: [1450/2387] eta: 0:02:56 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2270 (0.3220) acc1: 97.6562 (86.5271) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2088 data: 0.1146 max mem: 3537 +Test: [1460/2387] eta: 0:02:55 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1740 (0.3209) acc1: 97.6562 (86.6091) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2299 data: 0.1383 max mem: 3537 +Test: [1470/2387] eta: 0:02:53 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1276 (0.3194) acc1: 99.4792 (86.6982) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2035 data: 0.1106 max mem: 3537 +Test: [1480/2387] eta: 0:02:51 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1140 (0.3181) acc1: 99.7396 (86.7828) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2084 data: 0.1149 max mem: 3537 +Test: [1490/2387] eta: 0:02:49 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1220 (0.3168) acc1: 99.7396 (86.8677) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1930 data: 0.1000 max mem: 3537 +Test: [1500/2387] eta: 0:02:47 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1262 (0.3156) acc1: 99.4792 (86.9511) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1885 data: 0.0969 max mem: 3537 +Test: [1510/2387] eta: 0:02:45 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1342 (0.3145) acc1: 99.7396 (87.0367) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1998 data: 0.1100 max mem: 3537 +Test: [1520/2387] eta: 0:02:43 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1825 (0.3138) acc1: 99.7396 (87.1120) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1700 data: 0.0794 max mem: 3537 +Test: [1530/2387] eta: 0:02:41 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1996 (0.3130) acc1: 99.2188 (87.1887) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1702 data: 0.0773 max mem: 3537 +Test: [1540/2387] eta: 0:02:39 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1703 (0.3121) acc1: 99.7396 (87.2649) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1736 data: 0.0783 max mem: 3537 +Test: [1550/2387] eta: 0:02:37 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1665 (0.3113) acc1: 99.7396 (87.3407) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1725 data: 0.0781 max mem: 3537 +Test: [1560/2387] eta: 0:02:35 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1665 (0.3104) acc1: 99.4792 (87.4131) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1703 data: 0.0785 max mem: 3537 +Test: [1570/2387] eta: 0:02:33 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1640 (0.3096) acc1: 99.7396 (87.4897) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1708 data: 0.0798 max mem: 3537 +Test: [1580/2387] eta: 0:02:32 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1734 (0.3088) acc1: 99.4792 (87.5626) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1814 data: 0.0890 max mem: 3538 +Test: [1590/2387] eta: 0:02:30 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1750 (0.3079) acc1: 99.2188 (87.6252) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1831 data: 0.0905 max mem: 3538 +Test: [1600/2387] eta: 0:02:28 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1544 (0.3072) acc1: 99.4792 (87.6680) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1934 data: 0.0984 max mem: 3538 +Test: [1610/2387] eta: 0:02:26 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1416 (0.3062) acc1: 99.7396 (87.7422) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2058 data: 0.1112 max mem: 3538 +Test: [1620/2387] eta: 0:02:24 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1561 (0.3053) acc1: 99.7396 (87.8104) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2049 data: 0.1140 max mem: 3538 +Test: [1630/2387] eta: 0:02:22 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1807 (0.3048) acc1: 97.9167 (87.8650) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2025 data: 0.1105 max mem: 3538 +Test: [1640/2387] eta: 0:02:21 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2497 (0.3046) acc1: 96.3542 (87.9018) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2080 data: 0.1141 max mem: 3538 +Test: [1650/2387] eta: 0:02:19 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2435 (0.3042) acc1: 95.8333 (87.9532) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1949 data: 0.1016 max mem: 3538 +Test: [1660/2387] eta: 0:02:17 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2388 (0.3039) acc1: 96.0938 (87.9994) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1855 data: 0.0941 max mem: 3538 +Test: [1670/2387] eta: 0:02:15 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2406 (0.3036) acc1: 96.3542 (88.0422) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1957 data: 0.1037 max mem: 3538 +Test: [1680/2387] eta: 0:02:13 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2406 (0.3033) acc1: 95.5729 (88.0870) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1931 data: 0.1009 max mem: 3538 +Test: [1690/2387] eta: 0:02:11 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2362 (0.3030) acc1: 96.6146 (88.1326) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1891 data: 0.0977 max mem: 3538 +Test: [1700/2387] eta: 0:02:09 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2358 (0.3025) acc1: 96.6146 (88.1833) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1900 data: 0.0986 max mem: 3538 +Test: [1710/2387] eta: 0:02:07 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1317 (0.3014) acc1: 99.4792 (88.2505) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1917 data: 0.1007 max mem: 3538 +Test: [1720/2387] eta: 0:02:06 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1184 (0.3003) acc1: 99.7396 (88.3167) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1993 data: 0.1087 max mem: 3538 +Test: [1730/2387] eta: 0:02:04 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1139 (0.2993) acc1: 99.7396 (88.3804) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2098 data: 0.1193 max mem: 3538 +Test: [1740/2387] eta: 0:02:02 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1069 (0.2982) acc1: 100.0000 (88.4459) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2014 data: 0.1109 max mem: 3538 +Test: [1750/2387] eta: 0:02:00 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1069 (0.2971) acc1: 100.0000 (88.5107) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1946 data: 0.1038 max mem: 3538 +Test: [1760/2387] eta: 0:01:58 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1215 (0.2962) acc1: 99.7396 (88.5723) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1891 data: 0.0980 max mem: 3538 +Test: [1770/2387] eta: 0:01:56 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1541 (0.2955) acc1: 99.4792 (88.6289) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1754 data: 0.0843 max mem: 3538 +Test: [1780/2387] eta: 0:01:54 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1558 (0.2947) acc1: 99.2188 (88.6889) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1755 data: 0.0831 max mem: 3538 +Test: [1790/2387] eta: 0:01:52 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1454 (0.2938) acc1: 99.2188 (88.7477) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1701 data: 0.0777 max mem: 3538 +Test: [1800/2387] eta: 0:01:50 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1342 (0.2929) acc1: 99.7396 (88.8084) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1680 data: 0.0765 max mem: 3538 +Test: [1810/2387] eta: 0:01:48 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1270 (0.2919) acc1: 100.0000 (88.8691) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1766 data: 0.0835 max mem: 3539 +Test: [1820/2387] eta: 0:01:47 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1281 (0.2911) acc1: 100.0000 (88.9285) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1822 data: 0.0894 max mem: 3539 +Test: [1830/2387] eta: 0:01:45 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1412 (0.2903) acc1: 99.7396 (88.9864) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1711 data: 0.0791 max mem: 3539 +Test: [1840/2387] eta: 0:01:43 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.0994 (0.2892) acc1: 100.0000 (89.0461) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1550 data: 0.0630 max mem: 3539 +Test: [1850/2387] eta: 0:01:41 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.0977 (0.2882) acc1: 100.0000 (89.1053) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1617 data: 0.0703 max mem: 3539 +Test: [1860/2387] eta: 0:01:39 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.0977 (0.2872) acc1: 100.0000 (89.1638) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1690 data: 0.0768 max mem: 3539 +Test: [1870/2387] eta: 0:01:37 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.0980 (0.2862) acc1: 100.0000 (89.2216) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1732 data: 0.0813 max mem: 3539 +Test: [1880/2387] eta: 0:01:35 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.0980 (0.2852) acc1: 100.0000 (89.2789) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1715 data: 0.0804 max mem: 3539 +Test: [1890/2387] eta: 0:01:33 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.0986 (0.2842) acc1: 100.0000 (89.3356) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1797 data: 0.0888 max mem: 3539 +Test: [1900/2387] eta: 0:01:31 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.0998 (0.2832) acc1: 100.0000 (89.3917) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1866 data: 0.0943 max mem: 3539 +Test: [1910/2387] eta: 0:01:29 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1276 (0.2825) acc1: 100.0000 (89.4445) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1741 data: 0.0805 max mem: 3539 +Test: [1920/2387] eta: 0:01:27 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1526 (0.2824) acc1: 99.4792 (89.4559) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1865 data: 0.0944 max mem: 3539 +Test: [1930/2387] eta: 0:01:25 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1854 (0.2819) acc1: 98.1771 (89.5038) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1857 data: 0.0952 max mem: 3539 +Test: [1940/2387] eta: 0:01:23 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1909 (0.2818) acc1: 98.9583 (89.5154) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1647 data: 0.0732 max mem: 3539 +Test: [1950/2387] eta: 0:01:22 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1928 (0.2814) acc1: 98.9583 (89.5545) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1841 data: 0.0923 max mem: 3539 +Test: [1960/2387] eta: 0:01:20 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1960 (0.2810) acc1: 98.6979 (89.5945) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1774 data: 0.0861 max mem: 3539 +Test: [1970/2387] eta: 0:01:18 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2508 (0.2814) acc1: 92.7083 (89.5662) acc5: 100.0000 (100.0000) acc1_neg: 80.729 acc5_neg: 100.000 time: 0.1613 data: 0.0697 max mem: 3539 +Test: [1980/2387] eta: 0:01:16 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2767 (0.2811) acc1: 89.8438 (89.5919) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1699 data: 0.0789 max mem: 3539 +Test: [1990/2387] eta: 0:01:14 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1161 (0.2803) acc1: 99.2188 (89.6406) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1646 data: 0.0732 max mem: 3539 +Test: [2000/2387] eta: 0:01:12 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1158 (0.2794) acc1: 99.2188 (89.6882) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1617 data: 0.0708 max mem: 3539 +Test: [2010/2387] eta: 0:01:10 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1158 (0.2786) acc1: 99.2188 (89.7359) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1593 data: 0.0691 max mem: 3539 +Test: [2020/2387] eta: 0:01:08 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1137 (0.2778) acc1: 99.2188 (89.7829) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1553 data: 0.0650 max mem: 3539 +Test: [2030/2387] eta: 0:01:06 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1140 (0.2770) acc1: 99.2188 (89.8291) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1666 data: 0.0740 max mem: 3540 +Test: [2040/2387] eta: 0:01:04 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1139 (0.2762) acc1: 98.9583 (89.8754) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1652 data: 0.0728 max mem: 3540 +Test: [2050/2387] eta: 0:01:02 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1139 (0.2755) acc1: 99.2188 (89.9213) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1529 data: 0.0604 max mem: 3540 +Test: [2060/2387] eta: 0:01:00 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1255 (0.2748) acc1: 99.2188 (89.9647) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1613 data: 0.0676 max mem: 3540 +Test: [2070/2387] eta: 0:00:59 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1277 (0.2741) acc1: 98.9583 (90.0086) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0764 max mem: 3540 +Test: [2080/2387] eta: 0:00:57 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1271 (0.2734) acc1: 98.9583 (90.0510) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1991 data: 0.1032 max mem: 3540 +Test: [2090/2387] eta: 0:00:55 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1271 (0.2726) acc1: 98.9583 (90.0940) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2028 data: 0.1067 max mem: 3540 +Test: [2100/2387] eta: 0:00:53 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1256 (0.2720) acc1: 99.2188 (90.1371) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1561 data: 0.0610 max mem: 3540 +Test: [2110/2387] eta: 0:00:51 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1264 (0.2713) acc1: 99.2188 (90.1802) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1566 data: 0.0633 max mem: 3540 +Test: [2120/2387] eta: 0:00:49 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1365 (0.2708) acc1: 99.2188 (90.2197) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1804 data: 0.0897 max mem: 3540 +Test: [2130/2387] eta: 0:00:47 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1551 (0.2703) acc1: 99.2188 (90.2612) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1803 data: 0.0847 max mem: 3540 +Test: [2140/2387] eta: 0:00:45 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1383 (0.2699) acc1: 99.4792 (90.2836) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1784 data: 0.0823 max mem: 3540 +Test: [2150/2387] eta: 0:00:44 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1554 (0.2694) acc1: 99.4792 (90.3267) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1868 data: 0.0948 max mem: 3540 +Test: [2160/2387] eta: 0:00:42 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1615 (0.2689) acc1: 99.7396 (90.3657) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1891 data: 0.0954 max mem: 3540 +Test: [2170/2387] eta: 0:00:40 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1953 (0.2688) acc1: 97.9167 (90.3931) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2009 data: 0.1082 max mem: 3540 +Test: [2180/2387] eta: 0:00:38 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2737 (0.2697) acc1: 94.0104 (90.3395) acc5: 100.0000 (100.0000) acc1_neg: 55.729 acc5_neg: 100.000 time: 0.2199 data: 0.1286 max mem: 3540 +Test: [2190/2387] eta: 0:00:36 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.6092 (0.2712) acc1: 64.0625 (90.2365) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2149 data: 0.1214 max mem: 3540 +Test: [2200/2387] eta: 0:00:34 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1874 (0.2707) acc1: 95.0521 (90.2719) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1770 data: 0.0820 max mem: 3540 +Test: [2210/2387] eta: 0:00:33 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1518 (0.2701) acc1: 99.4792 (90.3136) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1758 data: 0.0806 max mem: 3540 +Test: [2220/2387] eta: 0:00:31 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1501 (0.2696) acc1: 99.7396 (90.3523) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1946 data: 0.1012 max mem: 3540 +Test: [2230/2387] eta: 0:00:29 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1524 (0.2691) acc1: 99.7396 (90.3940) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1851 data: 0.0941 max mem: 3540 +Test: [2240/2387] eta: 0:00:27 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1445 (0.2685) acc1: 99.7396 (90.4323) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1802 data: 0.0877 max mem: 3540 +Test: [2250/2387] eta: 0:00:25 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1532 (0.2681) acc1: 99.7396 (90.4697) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1693 data: 0.0752 max mem: 3540 +Test: [2260/2387] eta: 0:00:23 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1650 (0.2677) acc1: 99.7396 (90.5099) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1646 data: 0.0712 max mem: 3541 +Test: [2270/2387] eta: 0:00:21 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1495 (0.2671) acc1: 99.7396 (90.5463) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0777 max mem: 3541 +Test: [2280/2387] eta: 0:00:19 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1390 (0.2667) acc1: 100.0000 (90.5670) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1834 data: 0.0906 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1291 (0.2661) acc1: 100.0000 (90.6080) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1848 data: 0.0922 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1418 (0.2656) acc1: 100.0000 (90.6454) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1823 data: 0.0911 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.1623 (0.2654) acc1: 99.7396 (90.6785) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1765 data: 0.0856 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2145 (0.2653) acc1: 98.1771 (90.6984) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1500 data: 0.0589 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2335 (0.2653) acc1: 97.1354 (90.7162) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1469 data: 0.0549 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2191 (0.2651) acc1: 97.6562 (90.7411) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1635 data: 0.0726 max mem: 3541 +Test: [2350/2387] eta: 0:00:06 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2002 (0.2652) acc1: 98.4375 (90.7460) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1489 data: 0.0589 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2302 (0.2651) acc1: 98.4375 (90.7677) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1474 data: 0.0567 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2495 (0.2651) acc1: 96.6146 (90.7749) acc5: 100.0000 (100.0000) acc1_neg: 76.042 acc5_neg: 100.000 time: 0.1462 data: 0.0550 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.2747 (0.2654) acc1: 94.7917 (90.7754) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1359 data: 0.0450 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 54.338 acc5_pos: 100.000 loss: 0.3675 (0.2678) acc1: 86.1979 (90.6773) acc5: 100.0000 (100.0000) acc1_neg: 6.977 acc5_neg: 100.000 time: 0.1277 data: 0.0389 max mem: 3541 +Test: Total time: 0:07:21 (0.1850 s / it) +* Overall Acc@1 90.677 Acc@5 100.000 + target == real Acc@1 41.152 Acc@5 100.000 + target == fake Acc@1 94.155 Acc@5 100.000 + loss 0.268 + PR-AUC 0.9924 ROC-AUC 0.9001 EER 16.76% (thr=0.7036) +Accuracy of the network on the 916267 test images: 90.7% +[rank0]:[W726 03:31:14.454588388 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:31:33.823000 3143951 torch/distributed/run.py:766] +W0726 03:31:33.823000 3143951 torch/distributed/run.py:766] ***************************************** +W0726 03:31:33.823000 3143951 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:31:33.823000 3143951 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 03:32:02.960599514 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:32:02.385696117 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:32:03.675522574 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:32:03.703334033 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:32:03.804318912 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:32:03.832157520 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:32:03.920226312 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:32:03.961774517 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n32_seed5_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed5_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:02:35 acc1_pos: 39.583 acc5_pos: 100.000 loss: 0.8342 (0.8342) acc1: 39.5833 (39.5833) acc5: 100.0000 (100.0000) time: 7.2635 data: 5.8261 max mem: 3531 +Test: [ 10/517] eta: 0:06:16 acc1_pos: 11.458 acc5_pos: 100.000 loss: 0.9668 (0.9813) acc1: 27.6042 (28.8116) acc5: 100.0000 (100.0000) time: 0.7421 data: 0.5301 max mem: 3531 +Test: [ 20/517] eta: 0:03:35 acc1_pos: 19.531 acc5_pos: 100.000 loss: 0.9668 (0.9904) acc1: 25.0000 (26.9097) acc5: 100.0000 (100.0000) time: 0.0913 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:03:08 acc1_pos: 12.760 acc5_pos: 100.000 loss: 1.0178 (1.0114) acc1: 21.8750 (23.8827) acc5: 100.0000 (100.0000) time: 0.1921 data: 0.0977 max mem: 3531 +Test: [ 40/517] eta: 0:02:57 acc1_pos: 21.875 acc5_pos: 100.000 loss: 1.0930 (1.0328) acc1: 14.0625 (21.4685) acc5: 100.0000 (100.0000) time: 0.3099 data: 0.2160 max mem: 3531 +Test: [ 50/517] eta: 0:02:57 acc1_pos: 40.826 acc5_pos: 100.000 loss: 1.0204 (0.9899) acc1: 19.2708 (25.9702) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.3691 data: 0.2742 max mem: 3531 +Test: [ 60/517] eta: 0:02:41 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.4662 (0.8969) acc1: 81.2500 (36.2662) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.3105 data: 0.2144 max mem: 3531 +Test: [ 70/517] eta: 0:02:31 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.3668 (0.8185) acc1: 93.7500 (44.6596) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2349 data: 0.1410 max mem: 3531 +Test: [ 80/517] eta: 0:02:23 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.3442 (0.7614) acc1: 96.3542 (50.8488) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.2544 data: 0.1615 max mem: 3531 +Test: [ 90/517] eta: 0:02:16 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.3464 (0.7151) acc1: 97.3958 (55.9524) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2549 data: 0.1628 max mem: 3531 +Test: [100/517] eta: 0:02:10 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.3380 (0.6750) acc1: 97.3958 (60.0609) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2459 data: 0.1543 max mem: 3531 +Test: [110/517] eta: 0:02:03 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.3050 (0.6429) acc1: 97.1354 (63.0631) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2299 data: 0.1377 max mem: 3531 +Test: [120/517] eta: 0:01:58 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.3468 (0.6199) acc1: 95.3125 (65.6788) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2293 data: 0.1354 max mem: 3531 +Test: [130/517] eta: 0:01:54 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.3491 (0.5984) acc1: 95.8333 (68.0602) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2449 data: 0.1517 max mem: 3531 +Test: [140/517] eta: 0:01:49 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.3171 (0.5785) acc1: 98.9583 (70.2774) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2495 data: 0.1571 max mem: 3531 +Test: [150/517] eta: 0:01:45 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.3033 (0.5587) acc1: 99.4792 (72.1354) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2399 data: 0.1443 max mem: 3531 +Test: [160/517] eta: 0:01:41 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.2963 (0.5434) acc1: 97.1354 (73.5734) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2397 data: 0.1450 max mem: 3531 +Test: [170/517] eta: 0:01:38 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.2946 (0.5280) acc1: 96.0938 (74.9376) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2447 data: 0.1517 max mem: 3531 +Test: [180/517] eta: 0:01:34 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.2614 (0.5147) acc1: 95.8333 (76.1352) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2449 data: 0.1499 max mem: 3531 +Test: [190/517] eta: 0:01:31 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.1823 (0.4962) acc1: 99.7396 (77.3833) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2401 data: 0.1468 max mem: 3531 +Test: [200/517] eta: 0:01:27 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.1712 (0.4851) acc1: 100.0000 (78.4645) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2405 data: 0.1494 max mem: 3531 +Test: [210/517] eta: 0:01:24 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.3316 (0.4795) acc1: 97.1354 (79.1000) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2448 data: 0.1524 max mem: 3532 +Test: [220/517] eta: 0:01:21 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.3673 (0.4742) acc1: 91.1458 (79.6580) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.2444 data: 0.1498 max mem: 3532 +Test: [230/517] eta: 0:01:17 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.3673 (0.4693) acc1: 94.7917 (80.4000) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2349 data: 0.1403 max mem: 3532 +Test: [240/517] eta: 0:01:14 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.3115 (0.4608) acc1: 98.4375 (81.1430) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2349 data: 0.1390 max mem: 3532 +Test: [250/517] eta: 0:01:11 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.3024 (0.4550) acc1: 98.4375 (81.8331) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2349 data: 0.1350 max mem: 3532 +Test: [260/517] eta: 0:01:08 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.3150 (0.4501) acc1: 98.9583 (82.4563) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2249 data: 0.1246 max mem: 3532 +Test: [270/517] eta: 0:01:05 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.3235 (0.4459) acc1: 96.6146 (82.9278) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2249 data: 0.1273 max mem: 3532 +Test: [280/517] eta: 0:01:02 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.3231 (0.4411) acc1: 96.3542 (83.4344) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2399 data: 0.1417 max mem: 3532 +Test: [290/517] eta: 0:00:59 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.2837 (0.4358) acc1: 99.4792 (83.9732) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2449 data: 0.1466 max mem: 3532 +Test: [300/517] eta: 0:00:57 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.2905 (0.4311) acc1: 99.4792 (84.4806) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2399 data: 0.1416 max mem: 3532 +Test: [310/517] eta: 0:00:54 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.3045 (0.4282) acc1: 99.2188 (84.9561) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2449 data: 0.1457 max mem: 3532 +Test: [320/517] eta: 0:00:51 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.3324 (0.4244) acc1: 99.2188 (85.3818) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2465 data: 0.1487 max mem: 3532 +Test: [330/517] eta: 0:00:48 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.2719 (0.4195) acc1: 99.4792 (85.8053) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2449 data: 0.1508 max mem: 3532 +Test: [340/517] eta: 0:00:46 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.2497 (0.4161) acc1: 99.7396 (86.1513) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.2483 data: 0.1537 max mem: 3532 +Test: [350/517] eta: 0:00:43 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.3258 (0.4141) acc1: 96.8750 (86.4339) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.2566 data: 0.1605 max mem: 3532 +Test: [360/517] eta: 0:00:41 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.3257 (0.4112) acc1: 96.8750 (86.7346) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2608 data: 0.1668 max mem: 3532 +Test: [370/517] eta: 0:00:38 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.2706 (0.4068) acc1: 99.4792 (87.0873) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2492 data: 0.1561 max mem: 3532 +Test: [380/517] eta: 0:00:35 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.2641 (0.4036) acc1: 100.0000 (87.4098) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2403 data: 0.1464 max mem: 3532 +Test: [390/517] eta: 0:00:32 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.2869 (0.4007) acc1: 99.4792 (87.6845) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2389 data: 0.1436 max mem: 3532 +Test: [400/517] eta: 0:00:30 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.2853 (0.3978) acc1: 99.4792 (87.9793) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2386 data: 0.1429 max mem: 3532 +Test: [410/517] eta: 0:00:27 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.3052 (0.3958) acc1: 99.7396 (88.2559) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2399 data: 0.1442 max mem: 3532 +Test: [420/517] eta: 0:00:25 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.3018 (0.3937) acc1: 99.2188 (88.4928) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2413 data: 0.1461 max mem: 3532 +Test: [430/517] eta: 0:00:22 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.3007 (0.3915) acc1: 98.4375 (88.6849) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2451 data: 0.1519 max mem: 3532 +Test: [440/517] eta: 0:00:19 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.2914 (0.3896) acc1: 98.4375 (88.9060) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2440 data: 0.1511 max mem: 3533 +Test: [450/517] eta: 0:00:17 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.2819 (0.3873) acc1: 99.4792 (89.1474) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2409 data: 0.1467 max mem: 3533 +Test: [460/517] eta: 0:00:14 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.2963 (0.3860) acc1: 99.4792 (89.3218) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.2444 data: 0.1499 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.2937 (0.3839) acc1: 99.2188 (89.5308) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2386 data: 0.1440 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.2894 (0.3816) acc1: 99.7396 (89.7441) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2316 data: 0.1369 max mem: 3533 +Test: [490/517] eta: 0:00:06 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.2927 (0.3800) acc1: 99.7396 (89.9000) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2349 data: 0.1388 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.3479 (0.3807) acc1: 96.6146 (89.8879) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.2182 data: 0.1220 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.3683 (0.3803) acc1: 90.6250 (89.9436) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1895 data: 0.0974 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.3261 (0.3791) acc1: 97.6562 (90.0437) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1574 data: 0.0660 max mem: 3533 +Test: Total time: 0:02:10 (0.2516 s / it) +* Overall Acc@1 90.044 Acc@5 100.000 + target == real Acc@1 22.794 Acc@5 100.000 + target == fake Acc@1 97.020 Acc@5 100.000 + loss 0.379 + PR-AUC 0.9622 ROC-AUC 0.7596 EER 30.83% (thr=0.6982) +Accuracy of the network on the 198427 test images: 90.0% +[rank0]:[W726 03:34:28.037236537 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 03:34:48.378000 3144524 torch/distributed/run.py:766] +W0726 03:34:48.378000 3144524 torch/distributed/run.py:766] ***************************************** +W0726 03:34:48.378000 3144524 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:34:48.378000 3144524 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:35:17.189420935 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:35:17.283376668 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:35:18.922241263 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:35:18.979091659 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:35:18.040764350 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:35:18.053192649 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:35:18.105881350 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:35:18.351659227 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n32_seed5_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed5_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:30:59 acc1_pos: 36.198 acc5_pos: 100.000 loss: 0.8683 (0.8683) acc1: 36.1979 (36.1979) acc5: 100.0000 (100.0000) time: 4.7810 data: 3.3363 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/389] eta: 0:03:15 acc1_pos: 73.698 acc5_pos: 100.000 loss: 0.7873 (0.7405) acc1: 39.0625 (50.1894) acc5: 100.0000 (100.0000) time: 0.5162 data: 0.3036 max mem: 3531 +Test: [ 20/389] eta: 0:02:03 acc1_pos: 10.938 acc5_pos: 100.000 loss: 0.7527 (0.8005) acc1: 39.0625 (43.3036) acc5: 100.0000 (100.0000) time: 0.1125 data: 0.0230 max mem: 3531 +Test: [ 30/389] eta: 0:01:34 acc1_pos: 50.000 acc5_pos: 100.000 loss: 0.7636 (0.8189) acc1: 35.1562 (39.1381) acc5: 100.0000 (100.0000) time: 0.1221 data: 0.0307 max mem: 3531 +Test: [ 40/389] eta: 0:01:29 acc1_pos: 9.375 acc5_pos: 100.000 loss: 0.7626 (0.8072) acc1: 35.6771 (40.6250) acc5: 100.0000 (100.0000) time: 0.1745 data: 0.0802 max mem: 3531 +Test: [ 50/389] eta: 0:01:22 acc1_pos: 60.938 acc5_pos: 100.000 loss: 0.6838 (0.7767) acc1: 58.0729 (45.3227) acc5: 100.0000 (100.0000) time: 0.2182 data: 0.1233 max mem: 3531 +Test: [ 60/389] eta: 0:01:17 acc1_pos: 72.917 acc5_pos: 100.000 loss: 0.6826 (0.7572) acc1: 60.9375 (48.0063) acc5: 100.0000 (100.0000) time: 0.1881 data: 0.0929 max mem: 3531 +Test: [ 70/389] eta: 0:01:11 acc1_pos: 52.344 acc5_pos: 100.000 loss: 0.6333 (0.7420) acc1: 57.8125 (49.5525) acc5: 100.0000 (100.0000) time: 0.1749 data: 0.0806 max mem: 3531 +Test: [ 80/389] eta: 0:01:07 acc1_pos: 84.635 acc5_pos: 100.000 loss: 0.6772 (0.7427) acc1: 52.3438 (49.8103) acc5: 100.0000 (100.0000) time: 0.1756 data: 0.0837 max mem: 3531 +Test: [ 90/389] eta: 0:01:04 acc1_pos: 65.104 acc5_pos: 100.000 loss: 0.5194 (0.7182) acc1: 74.7396 (52.5040) acc5: 100.0000 (100.0000) time: 0.1799 data: 0.0857 max mem: 3531 +Test: [100/389] eta: 0:01:01 acc1_pos: 27.344 acc5_pos: 100.000 loss: 0.7143 (0.7316) acc1: 51.3021 (50.6265) acc5: 100.0000 (100.0000) time: 0.1781 data: 0.0841 max mem: 3531 +Test: [110/389] eta: 0:00:57 acc1_pos: 95.833 acc5_pos: 100.000 loss: 0.8654 (0.7465) acc1: 27.3438 (49.8710) acc5: 100.0000 (100.0000) time: 0.1704 data: 0.0790 max mem: 3531 +Test: [120/389] eta: 0:00:54 acc1_pos: 50.781 acc5_pos: 100.000 loss: 0.7258 (0.7353) acc1: 50.7812 (51.0115) acc5: 100.0000 (100.0000) time: 0.1550 data: 0.0638 max mem: 3531 +Test: [130/389] eta: 0:00:51 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.5143 (0.7175) acc1: 72.1354 (52.9163) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1635 data: 0.0710 max mem: 3531 +Test: [140/389] eta: 0:00:49 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.4073 (0.7009) acc1: 87.2396 (54.7448) acc5: 100.0000 (100.0000) acc1_neg: 38.021 acc5_neg: 100.000 time: 0.1756 data: 0.0827 max mem: 3531 +Test: [150/389] eta: 0:00:46 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.4230 (0.6884) acc1: 83.0729 (56.0206) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1561 data: 0.0653 max mem: 3531 +Test: [160/389] eta: 0:00:44 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.4974 (0.6688) acc1: 80.4688 (57.6685) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1502 data: 0.0594 max mem: 3531 +Test: [170/389] eta: 0:00:41 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.2196 (0.6494) acc1: 100.0000 (59.4542) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.1585 data: 0.0670 max mem: 3531 +Test: [180/389] eta: 0:00:39 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.2801 (0.6370) acc1: 93.4896 (60.8684) acc5: 100.0000 (100.0000) acc1_neg: 69.271 acc5_neg: 100.000 time: 0.1507 data: 0.0598 max mem: 3531 +Test: [190/389] eta: 0:00:37 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.4610 (0.6302) acc1: 85.6771 (61.6151) acc5: 100.0000 (100.0000) acc1_neg: 61.979 acc5_neg: 100.000 time: 0.1506 data: 0.0596 max mem: 3531 +Test: [200/389] eta: 0:00:34 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.4222 (0.6203) acc1: 95.0521 (63.0066) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1509 data: 0.0595 max mem: 3531 +Test: [210/389] eta: 0:00:32 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.3128 (0.6051) acc1: 99.2188 (64.6376) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1545 data: 0.0637 max mem: 3532 +Test: [220/389] eta: 0:00:30 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.2317 (0.5899) acc1: 100.0000 (66.1105) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1577 data: 0.0662 max mem: 3532 +Test: [230/389] eta: 0:00:28 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.4328 (0.5874) acc1: 90.8854 (66.4480) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1520 data: 0.0600 max mem: 3532 +Test: [240/389] eta: 0:00:26 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.4547 (0.5794) acc1: 86.9792 (67.3896) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1508 data: 0.0590 max mem: 3532 +Test: [250/389] eta: 0:00:24 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.2887 (0.5673) acc1: 95.3125 (68.5695) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1548 data: 0.0622 max mem: 3532 +Test: [260/389] eta: 0:00:22 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.2836 (0.5596) acc1: 99.2188 (69.3487) acc5: 100.0000 (100.0000) acc1_neg: 66.146 acc5_neg: 100.000 time: 0.1556 data: 0.0614 max mem: 3532 +Test: [270/389] eta: 0:00:21 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.4098 (0.5581) acc1: 96.0938 (69.8167) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1582 data: 0.0655 max mem: 3532 +Test: [280/389] eta: 0:00:19 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.5352 (0.5584) acc1: 78.1250 (69.8482) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1645 data: 0.0723 max mem: 3532 +Test: [290/389] eta: 0:00:17 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.5261 (0.5536) acc1: 79.4271 (70.4888) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1607 data: 0.0682 max mem: 3532 +Test: [300/389] eta: 0:00:15 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.5524 (0.5542) acc1: 73.9583 (70.5409) acc5: 100.0000 (100.0000) acc1_neg: 73.958 acc5_neg: 100.000 time: 0.1601 data: 0.0676 max mem: 3532 +Test: [310/389] eta: 0:00:13 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.6009 (0.5564) acc1: 69.0104 (70.3301) acc5: 100.0000 (100.0000) acc1_neg: 18.490 acc5_neg: 100.000 time: 0.1617 data: 0.0689 max mem: 3532 +Test: [320/389] eta: 0:00:12 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.4911 (0.5525) acc1: 79.9479 (70.8049) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1695 data: 0.0783 max mem: 3532 +Test: [330/389] eta: 0:00:10 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.4998 (0.5509) acc1: 90.3646 (71.3652) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1766 data: 0.0855 max mem: 3532 +Test: [340/389] eta: 0:00:08 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.3155 (0.5417) acc1: 92.4479 (72.1209) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1793 data: 0.0876 max mem: 3532 +Test: [350/389] eta: 0:00:06 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.2135 (0.5507) acc1: 99.2188 (71.3987) acc5: 100.0000 (100.0000) acc1_neg: 0.781 acc5_neg: 100.000 time: 0.1820 data: 0.0905 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.9358 (0.5652) acc1: 27.0833 (70.0506) acc5: 100.0000 (100.0000) acc1_neg: 29.427 acc5_neg: 100.000 time: 0.1737 data: 0.0807 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.7072 (0.5621) acc1: 32.2917 (70.3455) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1632 data: 0.0699 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.4275 (0.5591) acc1: 80.2083 (70.6727) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.1505 data: 0.0596 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.4019 (0.5554) acc1: 90.1042 (71.0635) acc5: 100.0000 (100.0000) acc1_neg: 79.781 acc5_neg: 100.000 time: 0.1386 data: 0.0449 max mem: 3532 +Test: Total time: 0:01:07 (0.1736 s / it) +* Overall Acc@1 71.063 Acc@5 100.000 + target == real Acc@1 50.769 Acc@5 100.000 + target == fake Acc@1 80.497 Acc@5 100.000 + loss 0.555 + PR-AUC 0.8723 ROC-AUC 0.7534 EER 31.23% (thr=0.5825) +Accuracy of the network on the 149358 test images: 71.1% +[rank0]:[W726 03:36:41.513869049 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:37:01.464000 3145083 torch/distributed/run.py:766] +W0726 03:37:01.464000 3145083 torch/distributed/run.py:766] ***************************************** +W0726 03:37:01.464000 3145083 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:37:01.464000 3145083 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:37:30.997532099 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:37:30.035716206 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:37:30.273403289 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 1): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:37:30.440962604 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:37:30.447329254 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank1]:[W726 03:37:30.447346329 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:37:30.475166675 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:37:30.517333875 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n32_seed5_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed5_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:25:37 acc1_pos: 84.375 acc5_pos: 100.000 loss: 0.3685 (0.3685) acc1: 84.3750 (84.3750) acc5: 100.0000 (100.0000) time: 4.6180 data: 2.5799 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/333] eta: 0:03:02 acc1_pos: 85.417 acc5_pos: 100.000 loss: 0.3130 (0.2946) acc1: 88.8021 (88.7784) acc5: 100.0000 (100.0000) time: 0.5660 data: 0.2990 max mem: 3531 +Test: [ 20/333] eta: 0:01:52 acc1_pos: 96.615 acc5_pos: 100.000 loss: 0.2796 (0.2882) acc1: 89.5833 (89.3353) acc5: 100.0000 (100.0000) time: 0.1461 data: 0.0525 max mem: 3531 +Test: [ 30/333] eta: 0:01:25 acc1_pos: 63.281 acc5_pos: 100.000 loss: 0.3569 (0.4105) acc1: 80.2083 (80.0991) acc5: 100.0000 (100.0000) time: 0.1284 data: 0.0311 max mem: 3531 +Test: [ 40/333] eta: 0:01:22 acc1_pos: 60.417 acc5_pos: 100.000 loss: 0.6683 (0.4845) acc1: 58.5938 (74.2442) acc5: 100.0000 (100.0000) time: 0.1975 data: 0.1007 max mem: 3531 +Test: [ 50/333] eta: 0:01:17 acc1_pos: 65.885 acc5_pos: 100.000 loss: 0.6786 (0.5253) acc1: 57.5521 (71.0121) acc5: 100.0000 (100.0000) time: 0.2544 data: 0.1609 max mem: 3531 +Test: [ 60/333] eta: 0:01:10 acc1_pos: 53.906 acc5_pos: 100.000 loss: 0.6630 (0.5512) acc1: 60.9375 (69.3306) acc5: 100.0000 (100.0000) time: 0.2172 data: 0.1251 max mem: 3531 +Test: [ 70/333] eta: 0:01:04 acc1_pos: 54.427 acc5_pos: 100.000 loss: 0.6809 (0.5703) acc1: 58.3333 (67.8074) acc5: 100.0000 (100.0000) time: 0.1790 data: 0.0847 max mem: 3531 +Test: [ 80/333] eta: 0:01:01 acc1_pos: 66.667 acc5_pos: 100.000 loss: 0.6796 (0.5853) acc1: 55.4688 (66.6763) acc5: 100.0000 (100.0000) time: 0.1849 data: 0.0881 max mem: 3531 +Test: [ 90/333] eta: 0:00:58 acc1_pos: 67.448 acc5_pos: 100.000 loss: 0.6230 (0.5879) acc1: 64.3229 (66.5379) acc5: 100.0000 (100.0000) time: 0.2139 data: 0.1181 max mem: 3531 +Test: [100/333] eta: 0:00:55 acc1_pos: 75.521 acc5_pos: 100.000 loss: 0.6045 (0.5907) acc1: 64.3229 (66.3985) acc5: 100.0000 (100.0000) time: 0.2199 data: 0.1241 max mem: 3531 +Test: [110/333] eta: 0:00:51 acc1_pos: 68.229 acc5_pos: 100.000 loss: 0.6045 (0.5930) acc1: 63.5417 (66.1529) acc5: 100.0000 (100.0000) time: 0.1946 data: 0.0977 max mem: 3531 +Test: [120/333] eta: 0:00:48 acc1_pos: 68.490 acc5_pos: 100.000 loss: 0.5936 (0.5939) acc1: 63.5417 (65.9844) acc5: 100.0000 (100.0000) time: 0.1899 data: 0.0964 max mem: 3531 +Test: [130/333] eta: 0:00:46 acc1_pos: 75.260 acc5_pos: 100.000 loss: 0.5936 (0.5940) acc1: 65.3646 (66.0623) acc5: 100.0000 (100.0000) time: 0.2080 data: 0.1163 max mem: 3531 +Test: [140/333] eta: 0:00:43 acc1_pos: 81.510 acc5_pos: 100.000 loss: 0.5991 (0.5929) acc1: 65.8854 (66.1662) acc5: 100.0000 (100.0000) time: 0.2085 data: 0.1167 max mem: 3531 +Test: [150/333] eta: 0:00:41 acc1_pos: 65.625 acc5_pos: 100.000 loss: 0.6112 (0.5949) acc1: 65.6250 (65.9716) acc5: 100.0000 (100.0000) time: 0.2084 data: 0.1166 max mem: 3531 +Test: [160/333] eta: 0:00:38 acc1_pos: 56.908 acc5_pos: 100.000 loss: 0.6159 (0.6061) acc1: 61.9792 (65.2336) acc5: 100.0000 (100.0000) acc1_neg: 42.448 acc5_neg: 100.000 time: 0.2120 data: 0.1193 max mem: 3531 +Test: [170/333] eta: 0:00:36 acc1_pos: 56.908 acc5_pos: 100.000 loss: 0.7871 (0.6194) acc1: 51.3021 (64.2315) acc5: 100.0000 (100.0000) acc1_neg: 52.344 acc5_neg: 100.000 time: 0.2083 data: 0.1141 max mem: 3531 +Test: [180/333] eta: 0:00:33 acc1_pos: 56.908 acc5_pos: 100.000 loss: 0.7638 (0.6254) acc1: 52.0833 (63.7532) acc5: 100.0000 (100.0000) acc1_neg: 51.042 acc5_neg: 100.000 time: 0.1993 data: 0.1059 max mem: 3531 +Test: [190/333] eta: 0:00:32 acc1_pos: 56.908 acc5_pos: 100.000 loss: 0.7047 (0.6280) acc1: 55.7292 (63.5867) acc5: 100.0000 (100.0000) acc1_neg: 58.854 acc5_neg: 100.000 time: 0.2400 data: 0.1484 max mem: 3531 +Test: [200/333] eta: 0:00:29 acc1_pos: 56.908 acc5_pos: 100.000 loss: 0.6859 (0.6305) acc1: 58.3333 (63.5456) acc5: 100.0000 (100.0000) acc1_neg: 58.073 acc5_neg: 100.000 time: 0.2580 data: 0.1662 max mem: 3531 +Test: [210/333] eta: 0:00:28 acc1_pos: 56.908 acc5_pos: 100.000 loss: 0.7226 (0.6372) acc1: 57.8125 (63.1344) acc5: 100.0000 (100.0000) acc1_neg: 46.615 acc5_neg: 100.000 time: 0.2623 data: 0.1690 max mem: 3532 +Test: [220/333] eta: 0:00:25 acc1_pos: 56.908 acc5_pos: 100.000 loss: 0.7847 (0.6460) acc1: 55.2083 (62.4918) acc5: 100.0000 (100.0000) acc1_neg: 67.969 acc5_neg: 100.000 time: 0.2153 data: 0.1224 max mem: 3532 +Test: [230/333] eta: 0:00:23 acc1_pos: 56.908 acc5_pos: 100.000 loss: 0.5359 (0.6397) acc1: 70.0521 (63.1877) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.2123 data: 0.1170 max mem: 3532 +Test: [240/333] eta: 0:00:20 acc1_pos: 56.908 acc5_pos: 100.000 loss: 0.5359 (0.6396) acc1: 76.5625 (63.2099) acc5: 100.0000 (100.0000) acc1_neg: 62.760 acc5_neg: 100.000 time: 0.2276 data: 0.1288 max mem: 3532 +Test: [250/333] eta: 0:00:18 acc1_pos: 56.908 acc5_pos: 100.000 loss: 0.6722 (0.6429) acc1: 59.3750 (62.9939) acc5: 100.0000 (100.0000) acc1_neg: 53.906 acc5_neg: 100.000 time: 0.1368 data: 0.0400 max mem: 3532 +Test: [260/333] eta: 0:00:15 acc1_pos: 56.908 acc5_pos: 100.000 loss: 0.7247 (0.6478) acc1: 54.9479 (62.6606) acc5: 100.0000 (100.0000) acc1_neg: 56.250 acc5_neg: 100.000 time: 0.1465 data: 0.0536 max mem: 3532 +Test: [270/333] eta: 0:00:13 acc1_pos: 56.908 acc5_pos: 100.000 loss: 0.7148 (0.6488) acc1: 57.0312 (62.6336) acc5: 100.0000 (100.0000) acc1_neg: 63.281 acc5_neg: 100.000 time: 0.1701 data: 0.0779 max mem: 3532 +Test: [280/333] eta: 0:00:11 acc1_pos: 56.908 acc5_pos: 100.000 loss: 0.6839 (0.6537) acc1: 57.0312 (62.1979) acc5: 100.0000 (100.0000) acc1_neg: 57.812 acc5_neg: 100.000 time: 0.1465 data: 0.0549 max mem: 3532 +Test: [290/333] eta: 0:00:09 acc1_pos: 56.908 acc5_pos: 100.000 loss: 0.8788 (0.6620) acc1: 43.4896 (61.5120) acc5: 100.0000 (100.0000) acc1_neg: 43.750 acc5_neg: 100.000 time: 0.1437 data: 0.0538 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 56.908 acc5_pos: 100.000 loss: 0.9056 (0.6704) acc1: 40.6250 (60.8449) acc5: 100.0000 (100.0000) acc1_neg: 44.531 acc5_neg: 100.000 time: 0.1386 data: 0.0487 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 56.908 acc5_pos: 100.000 loss: 0.8139 (0.6697) acc1: 48.1771 (60.9568) acc5: 100.0000 (100.0000) acc1_neg: 63.802 acc5_neg: 100.000 time: 0.1343 data: 0.0447 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 56.908 acc5_pos: 100.000 loss: 0.6305 (0.6704) acc1: 64.3229 (60.9448) acc5: 100.0000 (100.0000) acc1_neg: 35.938 acc5_neg: 100.000 time: 0.1282 data: 0.0386 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 56.908 acc5_pos: 100.000 loss: 0.9830 (0.6817) acc1: 38.5417 (60.2027) acc5: 100.0000 (100.0000) acc1_neg: 38.021 acc5_neg: 100.000 time: 0.1206 data: 0.0311 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 56.908 acc5_pos: 100.000 loss: 1.0056 (0.6838) acc1: 38.0208 (60.0784) acc5: 100.0000 (100.0000) acc1_neg: 37.746 acc5_neg: 100.000 time: 0.1251 data: 0.0303 max mem: 3532 +Test: Total time: 0:01:07 (0.2013 s / it) +* Overall Acc@1 60.078 Acc@5 100.000 + target == real Acc@1 65.945 Acc@5 100.000 + target == fake Acc@1 54.856 Acc@5 100.000 + loss 0.684 + PR-AUC 0.6617 ROC-AUC 0.6497 EER 39.52% (thr=0.4658) +Accuracy of the network on the 127843 test images: 60.1% +[rank0]:[W726 03:38:45.510063659 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:39:04.998000 3145644 torch/distributed/run.py:766] +W0726 03:39:04.998000 3145644 torch/distributed/run.py:766] ***************************************** +W0726 03:39:04.998000 3145644 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:39:04.998000 3145644 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:39:32.202549575 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:39:32.550464294 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:39:33.595761804 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:39:33.770105343 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:39:33.048282106 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:39:33.088431472 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W726 03:39:33.088475789 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:39:33.195953358 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n32_seed5_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed5_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:19:45 acc1_pos: 97.656 acc5_pos: 100.000 loss: 0.1655 (0.1655) acc1: 97.6562 (97.6562) acc5: 100.0000 (100.0000) time: 5.0443 data: 3.2557 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/235] eta: 0:02:08 acc1_pos: 96.094 acc5_pos: 100.000 loss: 0.1694 (0.1704) acc1: 97.1354 (97.0644) acc5: 100.0000 (100.0000) time: 0.5726 data: 0.3287 max mem: 3531 +Test: [ 20/235] eta: 0:01:16 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.1680 (0.1667) acc1: 97.1354 (97.2842) acc5: 100.0000 (100.0000) time: 0.1222 data: 0.0317 max mem: 3531 +Test: [ 30/235] eta: 0:00:59 acc1_pos: 97.917 acc5_pos: 100.000 loss: 0.1615 (0.1657) acc1: 97.6562 (97.3118) acc5: 100.0000 (100.0000) time: 0.1349 data: 0.0407 max mem: 3531 +Test: [ 40/235] eta: 0:00:56 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.1617 (0.1659) acc1: 97.6562 (97.3069) acc5: 100.0000 (100.0000) time: 0.2201 data: 0.1267 max mem: 3531 +Test: [ 50/235] eta: 0:00:50 acc1_pos: 95.833 acc5_pos: 100.000 loss: 0.1669 (0.1671) acc1: 96.8750 (97.1814) acc5: 100.0000 (100.0000) time: 0.2494 data: 0.1570 max mem: 3531 +Test: [ 60/235] eta: 0:00:46 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.1741 (0.1691) acc1: 96.6146 (97.1098) acc5: 100.0000 (100.0000) time: 0.2147 data: 0.1189 max mem: 3531 +Test: [ 70/235] eta: 0:00:42 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1871 (0.3453) acc1: 95.0521 (87.0599) acc5: 100.0000 (100.0000) acc1_neg: 14.583 acc5_neg: 100.000 time: 0.2221 data: 0.1282 max mem: 3531 +Test: [ 80/235] eta: 0:00:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.5771 (0.4982) acc1: 16.6667 (78.3886) acc5: 100.0000 (100.0000) acc1_neg: 16.146 acc5_neg: 100.000 time: 0.2052 data: 0.1147 max mem: 3531 +Test: [ 90/235] eta: 0:00:35 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.5808 (0.6172) acc1: 16.6667 (71.5946) acc5: 100.0000 (100.0000) acc1_neg: 16.667 acc5_neg: 100.000 time: 0.1877 data: 0.0940 max mem: 3531 +Test: [100/235] eta: 0:00:31 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.5754 (0.7110) acc1: 15.8854 (66.1716) acc5: 100.0000 (100.0000) acc1_neg: 15.104 acc5_neg: 100.000 time: 0.1810 data: 0.0864 max mem: 3531 +Test: [110/235] eta: 0:00:29 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.5700 (0.7886) acc1: 16.6667 (61.7657) acc5: 100.0000 (100.0000) acc1_neg: 17.969 acc5_neg: 100.000 time: 0.1899 data: 0.0981 max mem: 3531 +Test: [120/235] eta: 0:00:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.5776 (0.8539) acc1: 15.1042 (57.8383) acc5: 100.0000 (100.0000) acc1_neg: 15.104 acc5_neg: 100.000 time: 0.1885 data: 0.0943 max mem: 3531 +Test: [130/235] eta: 0:00:23 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.5678 (0.9069) acc1: 14.5833 (54.5543) acc5: 100.0000 (100.0000) acc1_neg: 12.500 acc5_neg: 100.000 time: 0.1808 data: 0.0846 max mem: 3531 +Test: [140/235] eta: 0:00:21 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.5718 (0.9555) acc1: 14.3229 (51.6604) acc5: 100.0000 (100.0000) acc1_neg: 9.375 acc5_neg: 100.000 time: 0.1949 data: 0.1007 max mem: 3531 +Test: [150/235] eta: 0:00:18 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.5749 (0.9954) acc1: 14.3229 (49.1687) acc5: 100.0000 (100.0000) acc1_neg: 14.323 acc5_neg: 100.000 time: 0.1800 data: 0.0876 max mem: 3531 +Test: [160/235] eta: 0:00:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.5501 (1.0305) acc1: 14.8438 (47.0578) acc5: 100.0000 (100.0000) acc1_neg: 14.844 acc5_neg: 100.000 time: 0.1799 data: 0.0847 max mem: 3531 +Test: [170/235] eta: 0:00:14 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.4992 (1.0475) acc1: 16.1458 (45.6856) acc5: 100.0000 (100.0000) acc1_neg: 23.438 acc5_neg: 100.000 time: 0.1989 data: 0.1032 max mem: 3531 +Test: [180/235] eta: 0:00:11 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.3037 (1.0614) acc1: 24.7396 (44.5571) acc5: 100.0000 (100.0000) acc1_neg: 25.781 acc5_neg: 100.000 time: 0.1809 data: 0.0853 max mem: 3531 +Test: [190/235] eta: 0:00:09 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.2904 (1.0735) acc1: 25.2604 (43.5755) acc5: 100.0000 (100.0000) acc1_neg: 26.562 acc5_neg: 100.000 time: 0.1824 data: 0.0880 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.3003 (1.0854) acc1: 24.2188 (42.5827) acc5: 100.0000 (100.0000) acc1_neg: 26.042 acc5_neg: 100.000 time: 0.1871 data: 0.0960 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.2987 (1.0947) acc1: 23.6979 (41.7728) acc5: 100.0000 (100.0000) acc1_neg: 23.438 acc5_neg: 100.000 time: 0.1542 data: 0.0633 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.2868 (1.1037) acc1: 25.7812 (40.9915) acc5: 100.0000 (100.0000) acc1_neg: 25.781 acc5_neg: 100.000 time: 0.1349 data: 0.0450 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.3009 (1.1120) acc1: 24.7396 (40.3341) acc5: 100.0000 (100.0000) acc1_neg: 23.958 acc5_neg: 100.000 time: 0.1255 data: 0.0354 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.3022 (1.1158) acc1: 25.7812 (40.1440) acc5: 100.0000 (100.0000) acc1_neg: 26.087 acc5_neg: 100.000 time: 0.1117 data: 0.0244 max mem: 3532 +Test: Total time: 0:00:46 (0.1998 s / it) +* Overall Acc@1 40.144 Acc@5 100.000 + target == real Acc@1 97.131 Acc@5 100.000 + target == fake Acc@1 19.533 Acc@5 100.000 + loss 1.116 + PR-AUC 0.8940 ROC-AUC 0.7678 EER 30.39% (thr=0.1488) +Accuracy of the network on the 89879 test images: 40.1% +[rank0]:[W726 03:40:29.819469529 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:40:49.247000 3146192 torch/distributed/run.py:766] +W0726 03:40:49.247000 3146192 torch/distributed/run.py:766] ***************************************** +W0726 03:40:49.247000 3146192 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:40:49.247000 3146192 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:41:18.987131225 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:41:19.596216476 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:41:19.814706745 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:41:19.166845393 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:41:19.213791241 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:41:19.286191169 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:41:19.309763759 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:41:19.355150369 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n32_seed5_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed5_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:24:06 acc1_pos: 83.073 acc5_pos: 100.000 loss: 0.4436 (0.4436) acc1: 83.0729 (83.0729) acc5: 100.0000 (100.0000) time: 5.3763 data: 4.0161 max mem: 3531 +Test: [ 10/269] eta: 0:02:27 acc1_pos: 89.062 acc5_pos: 100.000 loss: 0.4436 (0.4472) acc1: 87.2396 (85.5587) acc5: 100.0000 (100.0000) time: 0.5706 data: 0.3655 max mem: 3531 +Test: [ 20/269] eta: 0:01:27 acc1_pos: 94.010 acc5_pos: 100.000 loss: 0.4325 (0.4351) acc1: 87.7604 (87.3512) acc5: 100.0000 (100.0000) time: 0.1012 data: 0.0074 max mem: 3531 +Test: [ 30/269] eta: 0:01:04 acc1_pos: 91.667 acc5_pos: 100.000 loss: 0.4085 (0.4289) acc1: 88.8021 (88.0376) acc5: 100.0000 (100.0000) time: 0.1061 data: 0.0074 max mem: 3531 +Test: [ 40/269] eta: 0:00:58 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.5026 (0.5103) acc1: 77.3438 (78.8237) acc5: 100.0000 (100.0000) acc1_neg: 32.031 acc5_neg: 100.000 time: 0.1499 data: 0.0537 max mem: 3531 +Test: [ 50/269] eta: 0:00:52 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.7342 (0.5555) acc1: 53.6458 (73.8205) acc5: 100.0000 (100.0000) acc1_neg: 44.531 acc5_neg: 100.000 time: 0.1931 data: 0.1003 max mem: 3531 +Test: [ 60/269] eta: 0:00:48 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.7384 (0.5955) acc1: 50.2604 (68.7287) acc5: 100.0000 (100.0000) acc1_neg: 23.958 acc5_neg: 100.000 time: 0.1860 data: 0.0930 max mem: 3531 +Test: [ 70/269] eta: 0:00:44 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.7666 (0.6178) acc1: 46.0938 (65.6800) acc5: 100.0000 (100.0000) acc1_neg: 46.094 acc5_neg: 100.000 time: 0.1835 data: 0.0900 max mem: 3531 +Test: [ 80/269] eta: 0:00:41 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.7315 (0.6307) acc1: 47.3958 (63.7828) acc5: 100.0000 (100.0000) acc1_neg: 52.083 acc5_neg: 100.000 time: 0.1749 data: 0.0827 max mem: 3531 +Test: [ 90/269] eta: 0:00:37 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.7465 (0.6527) acc1: 42.4479 (60.8717) acc5: 100.0000 (100.0000) acc1_neg: 36.458 acc5_neg: 100.000 time: 0.1681 data: 0.0768 max mem: 3531 +Test: [100/269] eta: 0:00:34 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.8499 (0.6750) acc1: 36.1979 (58.5396) acc5: 100.0000 (100.0000) acc1_neg: 23.177 acc5_neg: 100.000 time: 0.1618 data: 0.0699 max mem: 3531 +Test: [110/269] eta: 0:00:32 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.8927 (0.6968) acc1: 31.2500 (56.0412) acc5: 100.0000 (100.0000) acc1_neg: 24.740 acc5_neg: 100.000 time: 0.1621 data: 0.0708 max mem: 3531 +Test: [120/269] eta: 0:00:30 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.8763 (0.7094) acc1: 33.0729 (54.7973) acc5: 100.0000 (100.0000) acc1_neg: 33.073 acc5_neg: 100.000 time: 0.1769 data: 0.0862 max mem: 3531 +Test: [130/269] eta: 0:00:27 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.8771 (0.7241) acc1: 33.8542 (53.1131) acc5: 100.0000 (100.0000) acc1_neg: 35.417 acc5_neg: 100.000 time: 0.1835 data: 0.0921 max mem: 3531 +Test: [140/269] eta: 0:00:25 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.8844 (0.7369) acc1: 33.5938 (51.7453) acc5: 100.0000 (100.0000) acc1_neg: 28.906 acc5_neg: 100.000 time: 0.1814 data: 0.0891 max mem: 3531 +Test: [150/269] eta: 0:00:23 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.8782 (0.7436) acc1: 34.1146 (51.0192) acc5: 100.0000 (100.0000) acc1_neg: 48.698 acc5_neg: 100.000 time: 0.1975 data: 0.1045 max mem: 3531 +Test: [160/269] eta: 0:00:21 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.8358 (0.7506) acc1: 38.8021 (50.1601) acc5: 100.0000 (100.0000) acc1_neg: 28.125 acc5_neg: 100.000 time: 0.2034 data: 0.1109 max mem: 3531 +Test: [170/269] eta: 0:00:19 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.8832 (0.7612) acc1: 31.5104 (49.1502) acc5: 100.0000 (100.0000) acc1_neg: 21.875 acc5_neg: 100.000 time: 0.1773 data: 0.0838 max mem: 3531 +Test: [180/269] eta: 0:00:17 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.9404 (0.7700) acc1: 32.5521 (48.1713) acc5: 100.0000 (100.0000) acc1_neg: 32.031 acc5_neg: 100.000 time: 0.1801 data: 0.0852 max mem: 3531 +Test: [190/269] eta: 0:00:15 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.9072 (0.7775) acc1: 32.5521 (47.3754) acc5: 100.0000 (100.0000) acc1_neg: 34.635 acc5_neg: 100.000 time: 0.2091 data: 0.1158 max mem: 3531 +Test: [200/269] eta: 0:00:13 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.9522 (0.7881) acc1: 27.3438 (46.3645) acc5: 100.0000 (100.0000) acc1_neg: 50.781 acc5_neg: 100.000 time: 0.1947 data: 0.0995 max mem: 3531 +Test: [210/269] eta: 0:00:11 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.9630 (0.7944) acc1: 25.5208 (45.6136) acc5: 100.0000 (100.0000) acc1_neg: 33.073 acc5_neg: 100.000 time: 0.1715 data: 0.0764 max mem: 3532 +Test: [220/269] eta: 0:00:09 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.8827 (0.7983) acc1: 32.2917 (45.2548) acc5: 100.0000 (100.0000) acc1_neg: 63.542 acc5_neg: 100.000 time: 0.1712 data: 0.0789 max mem: 3532 +Test: [230/269] eta: 0:00:07 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.8025 (0.7963) acc1: 41.6667 (45.4252) acc5: 100.0000 (100.0000) acc1_neg: 52.083 acc5_neg: 100.000 time: 0.1840 data: 0.0919 max mem: 3532 +Test: [240/269] eta: 0:00:05 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.8025 (0.7982) acc1: 42.1875 (45.2779) acc5: 100.0000 (100.0000) acc1_neg: 47.656 acc5_neg: 100.000 time: 0.1921 data: 0.0994 max mem: 3532 +Test: [250/269] eta: 0:00:03 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.8333 (0.8041) acc1: 37.7604 (44.5873) acc5: 100.0000 (100.0000) acc1_neg: 11.458 acc5_neg: 100.000 time: 0.1994 data: 0.1067 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 87.931 acc5_pos: 100.000 loss: 1.0057 (0.8103) acc1: 15.3646 (43.7839) acc5: 100.0000 (100.0000) acc1_neg: 38.802 acc5_neg: 100.000 time: 0.1603 data: 0.0696 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.8408 (0.8101) acc1: 26.0417 (43.6448) acc5: 100.0000 (100.0000) acc1_neg: 32.546 acc5_neg: 100.000 time: 0.1369 data: 0.0426 max mem: 3532 +Test: Total time: 0:00:51 (0.1910 s / it) +* Overall Acc@1 43.645 Acc@5 100.000 + target == real Acc@1 87.810 Acc@5 100.000 + target == fake Acc@1 37.617 Acc@5 100.000 + loss 0.810 + PR-AUC 0.9429 ROC-AUC 0.7122 EER 35.11% (thr=0.3760) +Accuracy of the network on the 103293 test images: 43.6% +[rank0]:[W726 03:42:23.512121991 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T03:42:30Z] eval_done tag=fixed500k_fake_n32_seed5_resampled +[2026-07-26T03:42:31Z] eval_start tag=fixed500k_real_n16_seed6_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed6_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T03:42:31Z +hostname=n111-046-081 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed6_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n16_seed6_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 03:42:43.554000 3146769 torch/distributed/run.py:766] +W0726 03:42:43.554000 3146769 torch/distributed/run.py:766] ***************************************** +W0726 03:42:43.554000 3146769 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:42:43.554000 3146769 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:43:10.690898476 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:43:10.983705375 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +[rank2]:[W726 03:43:10.308956000 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:43:10.314968147 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:43:10.439559800 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:43:10.458565244 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +[rank0]:[W726 03:43:10.479512759 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:43:10.490175013 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n16_seed6_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed6_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:20:17 acc1_pos: 26.042 acc5_pos: 100.000 loss: 0.9490 (0.9490) acc1: 26.0417 (26.0417) acc5: 100.0000 (100.0000) time: 5.2018 data: 3.6118 max mem: 3531 +Test: [ 10/234] eta: 0:02:04 acc1_pos: 72.135 acc5_pos: 100.000 loss: 0.6417 (0.7371) acc1: 57.2917 (55.4924) acc5: 100.0000 (100.0000) time: 0.5558 data: 0.3287 max mem: 3531 +Test: [ 20/234] eta: 0:01:12 acc1_pos: 94.010 acc5_pos: 100.000 loss: 0.7144 (0.7807) acc1: 52.3438 (51.0913) acc5: 100.0000 (100.0000) time: 0.0935 data: 0.0004 max mem: 3531 +Test: [ 30/234] eta: 0:00:53 acc1_pos: 71.615 acc5_pos: 100.000 loss: 0.2367 (0.6137) acc1: 94.0104 (64.1885) acc5: 100.0000 (100.0000) time: 0.1043 data: 0.0102 max mem: 3531 +Test: [ 40/234] eta: 0:00:50 acc1_pos: 96.094 acc5_pos: 100.000 loss: 0.2366 (0.5716) acc1: 93.7500 (67.4987) acc5: 100.0000 (100.0000) time: 0.1749 data: 0.0776 max mem: 3531 +Test: [ 50/234] eta: 0:00:45 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.2069 (0.4979) acc1: 93.4896 (72.8299) acc5: 100.0000 (100.0000) time: 0.2185 data: 0.1182 max mem: 3531 +Test: [ 60/234] eta: 0:00:40 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.1562 (0.4360) acc1: 96.0938 (77.2370) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0774 max mem: 3531 +Test: [ 70/234] eta: 0:00:36 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.1180 (0.4311) acc1: 99.7396 (77.8096) acc5: 100.0000 (100.0000) acc1_neg: 71.615 acc5_neg: 100.000 time: 0.1699 data: 0.0759 max mem: 3531 +Test: [ 80/234] eta: 0:00:34 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.5095 (0.4311) acc1: 73.1771 (78.2150) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2049 data: 0.1113 max mem: 3531 +Test: [ 90/234] eta: 0:00:31 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.2969 (0.4100) acc1: 88.2812 (79.7848) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.1899 data: 0.0937 max mem: 3531 +Test: [100/234] eta: 0:00:29 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.3191 (0.4124) acc1: 84.6354 (79.7855) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1849 data: 0.0886 max mem: 3531 +Test: [110/234] eta: 0:00:26 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.4311 (0.4101) acc1: 79.4271 (80.1168) acc5: 100.0000 (100.0000) acc1_neg: 74.219 acc5_neg: 100.000 time: 0.1799 data: 0.0852 max mem: 3531 +Test: [120/234] eta: 0:00:23 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.4071 (0.4156) acc1: 81.5104 (80.0039) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.1649 data: 0.0708 max mem: 3531 +Test: [130/234] eta: 0:00:21 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.3770 (0.4120) acc1: 86.7188 (80.6019) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1799 data: 0.0835 max mem: 3531 +Test: [140/234] eta: 0:00:19 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.3909 (0.4343) acc1: 84.3750 (79.5970) acc5: 100.0000 (100.0000) acc1_neg: 41.146 acc5_neg: 100.000 time: 0.1849 data: 0.0863 max mem: 3531 +Test: [150/234] eta: 0:00:16 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.4640 (0.4415) acc1: 76.8229 (79.2236) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.1699 data: 0.0716 max mem: 3531 +Test: [160/234] eta: 0:00:14 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.3618 (0.4338) acc1: 83.0729 (79.9398) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1599 data: 0.0612 max mem: 3531 +Test: [170/234] eta: 0:00:12 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.3338 (0.4287) acc1: 90.1042 (80.4870) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.1649 data: 0.0658 max mem: 3531 +Test: [180/234] eta: 0:00:10 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.3576 (0.4365) acc1: 87.2396 (80.0227) acc5: 100.0000 (100.0000) acc1_neg: 44.531 acc5_neg: 100.000 time: 0.1649 data: 0.0662 max mem: 3531 +Test: [190/234] eta: 0:00:08 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.3640 (0.4340) acc1: 84.8958 (80.2411) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1649 data: 0.0688 max mem: 3531 +Test: [200/234] eta: 0:00:06 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.3471 (0.4224) acc1: 86.7188 (80.9494) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1649 data: 0.0706 max mem: 3531 +Test: [210/234] eta: 0:00:04 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.1047 (0.4094) acc1: 99.7396 (81.8042) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1699 data: 0.0732 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.2273 (0.4024) acc1: 97.6562 (82.4095) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.1649 data: 0.0664 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.4250 (0.4193) acc1: 57.2917 (81.0730) acc5: 100.0000 (100.0000) acc1_neg: 48.958 acc5_neg: 100.000 time: 0.1532 data: 0.0589 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.7733 (0.4241) acc1: 53.1250 (80.7342) acc5: 100.0000 (100.0000) acc1_neg: 53.307 acc5_neg: 100.000 time: 0.1252 data: 0.0285 max mem: 3532 +Test: Total time: 0:00:43 (0.1877 s / it) +* Overall Acc@1 80.734 Acc@5 100.000 + target == real Acc@1 72.984 Acc@5 100.000 + target == fake Acc@1 82.919 Acc@5 100.000 + loss 0.424 + PR-AUC 0.9621 ROC-AUC 0.8773 EER 21.19% (thr=0.5737) +Accuracy of the network on the 89729 test images: 80.7% +[rank0]:[W726 03:44:06.191807461 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:44:25.309000 3147318 torch/distributed/run.py:766] +W0726 03:44:25.309000 3147318 torch/distributed/run.py:766] ***************************************** +W0726 03:44:25.309000 3147318 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:44:25.309000 3147318 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:44:53.561897412 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:44:54.948365370 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:44:54.182525459 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:44:54.305978838 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:44:54.561583939 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W726 03:44:54.561947636 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:44:54.566115738 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:44:55.602172437 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n16_seed6_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed6_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:34:18 acc1_pos: 9.115 acc5_pos: 100.000 loss: 1.3653 (1.3653) acc1: 9.1146 (9.1146) acc5: 100.0000 (100.0000) time: 6.8952 data: 5.3812 max mem: 3531 +Test: [ 10/2387] eta: 0:28:15 acc1_pos: 2.865 acc5_pos: 100.000 loss: 1.6165 (1.6183) acc1: 2.8646 (4.4744) acc5: 100.0000 (100.0000) time: 0.7132 data: 0.4896 max mem: 3531 +Test: [ 20/2387] eta: 0:16:36 acc1_pos: 0.521 acc5_pos: 100.000 loss: 1.6134 (1.5915) acc1: 1.5625 (4.5139) acc5: 100.0000 (100.0000) time: 0.0975 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:13:29 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.6849 (1.6284) acc1: 1.0417 (3.7466) acc5: 100.0000 (100.0000) time: 0.1399 data: 0.0445 max mem: 3531 +Test: [ 40/2387] eta: 0:12:15 acc1_pos: 1.042 acc5_pos: 100.000 loss: 1.7250 (1.6387) acc1: 1.0417 (3.4172) acc5: 100.0000 (100.0000) time: 0.1999 data: 0.1077 max mem: 3531 +Test: [ 50/2387] eta: 0:11:24 acc1_pos: 25.781 acc5_pos: 100.000 loss: 1.5814 (1.6064) acc1: 1.5625 (4.7232) acc5: 100.0000 (100.0000) time: 0.2149 data: 0.1199 max mem: 3531 +Test: [ 60/2387] eta: 0:10:50 acc1_pos: 27.344 acc5_pos: 100.000 loss: 1.4043 (1.5671) acc1: 9.1146 (6.0280) acc5: 100.0000 (100.0000) time: 0.2099 data: 0.1151 max mem: 3531 +Test: [ 70/2387] eta: 0:10:28 acc1_pos: 28.646 acc5_pos: 100.000 loss: 1.3985 (1.5429) acc1: 9.6354 (6.8259) acc5: 100.0000 (100.0000) time: 0.2156 data: 0.1234 max mem: 3531 +Test: [ 80/2387] eta: 0:10:22 acc1_pos: 15.625 acc5_pos: 100.000 loss: 1.3486 (1.5219) acc1: 9.6354 (7.4653) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1466 max mem: 3531 +Test: [ 90/2387] eta: 0:10:19 acc1_pos: 19.010 acc5_pos: 100.000 loss: 1.3105 (1.4962) acc1: 14.0625 (8.5794) acc5: 100.0000 (100.0000) time: 0.2645 data: 0.1698 max mem: 3531 +Test: [ 100/2387] eta: 0:10:15 acc1_pos: 14.323 acc5_pos: 100.000 loss: 1.3190 (1.4834) acc1: 14.3229 (9.0140) acc5: 100.0000 (100.0000) time: 0.2661 data: 0.1720 max mem: 3531 +Test: [ 110/2387] eta: 0:10:06 acc1_pos: 20.052 acc5_pos: 100.000 loss: 1.3302 (1.4675) acc1: 11.7188 (9.5252) acc5: 100.0000 (100.0000) time: 0.2499 data: 0.1565 max mem: 3531 +Test: [ 120/2387] eta: 0:10:00 acc1_pos: 23.438 acc5_pos: 100.000 loss: 1.2526 (1.4497) acc1: 16.4062 (10.2122) acc5: 100.0000 (100.0000) time: 0.2438 data: 0.1477 max mem: 3531 +Test: [ 130/2387] eta: 0:09:53 acc1_pos: 22.135 acc5_pos: 100.000 loss: 1.2669 (1.4431) acc1: 17.4479 (10.4763) acc5: 100.0000 (100.0000) time: 0.2446 data: 0.1470 max mem: 3531 +Test: [ 140/2387] eta: 0:09:48 acc1_pos: 26.042 acc5_pos: 100.000 loss: 1.3722 (1.4353) acc1: 13.8021 (10.8138) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1472 max mem: 3531 +Test: [ 150/2387] eta: 0:09:59 acc1_pos: 7.031 acc5_pos: 100.000 loss: 1.3106 (1.4254) acc1: 13.0208 (11.1082) acc5: 100.0000 (100.0000) time: 0.2999 data: 0.2007 max mem: 3531 +Test: [ 160/2387] eta: 0:09:33 acc1_pos: 5.479 acc5_pos: 100.000 loss: 1.2191 (1.3900) acc1: 15.8854 (13.4074) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2249 data: 0.1255 max mem: 3531 +Test: [ 170/2387] eta: 0:09:16 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1868 (1.3191) acc1: 98.1771 (18.4150) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1249 data: 0.0266 max mem: 3531 +Test: [ 180/2387] eta: 0:09:05 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1736 (1.2557) acc1: 99.2188 (22.8677) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1652 data: 0.0711 max mem: 3531 +Test: [ 190/2387] eta: 0:08:58 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1703 (1.1989) acc1: 99.2188 (26.8543) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1949 data: 0.1025 max mem: 3531 +Test: [ 200/2387] eta: 0:08:50 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1716 (1.1480) acc1: 98.9583 (30.4312) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1995 data: 0.1038 max mem: 3531 +Test: [ 210/2387] eta: 0:08:40 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1753 (1.1019) acc1: 98.6979 (33.6629) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1799 data: 0.0816 max mem: 3532 +Test: [ 220/2387] eta: 0:08:30 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1738 (1.0599) acc1: 98.9583 (36.6257) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1649 data: 0.0683 max mem: 3532 +Test: [ 230/2387] eta: 0:08:22 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1738 (1.0216) acc1: 99.2188 (39.3229) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1699 data: 0.0746 max mem: 3532 +Test: [ 240/2387] eta: 0:08:16 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1799 (0.9870) acc1: 98.9583 (41.7963) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1860 data: 0.0923 max mem: 3532 +Test: [ 250/2387] eta: 0:08:08 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1644 (0.9514) acc1: 99.4792 (44.1090) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1800 data: 0.0892 max mem: 3532 +Test: [ 260/2387] eta: 0:08:00 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.0816 (0.9181) acc1: 100.0000 (46.2504) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1591 data: 0.0683 max mem: 3532 +Test: [ 270/2387] eta: 0:07:53 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.0801 (0.8873) acc1: 100.0000 (48.2319) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1598 data: 0.0685 max mem: 3532 +Test: [ 280/2387] eta: 0:07:47 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.0800 (0.8586) acc1: 100.0000 (50.0741) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1704 data: 0.0774 max mem: 3532 +Test: [ 290/2387] eta: 0:07:40 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.0748 (0.8317) acc1: 100.0000 (51.7898) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1652 data: 0.0724 max mem: 3532 +Test: [ 300/2387] eta: 0:07:35 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.0750 (0.8068) acc1: 100.0000 (53.3880) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1667 data: 0.0751 max mem: 3532 +Test: [ 310/2387] eta: 0:07:28 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.0795 (0.7872) acc1: 100.0000 (54.7880) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1647 data: 0.0709 max mem: 3532 +Test: [ 320/2387] eta: 0:07:25 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.3444 (0.7736) acc1: 94.0104 (55.9863) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1724 data: 0.0774 max mem: 3532 +Test: [ 330/2387] eta: 0:07:25 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.3456 (0.7605) acc1: 93.4896 (57.1335) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2249 data: 0.1296 max mem: 3532 +Test: [ 340/2387] eta: 0:07:22 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.3440 (0.7483) acc1: 93.4896 (58.2012) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2307 data: 0.1349 max mem: 3532 +Test: [ 350/2387] eta: 0:07:15 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.3279 (0.7316) acc1: 94.2708 (59.3691) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1656 data: 0.0704 max mem: 3532 +Test: [ 360/2387] eta: 0:07:12 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1483 (0.7156) acc1: 99.4792 (60.4758) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1650 data: 0.0695 max mem: 3532 +Test: [ 370/2387] eta: 0:07:09 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1526 (0.7004) acc1: 99.4792 (61.5327) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1941 data: 0.1004 max mem: 3532 +Test: [ 380/2387] eta: 0:07:04 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1515 (0.6860) acc1: 99.4792 (62.5260) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1748 data: 0.0837 max mem: 3532 +Test: [ 390/2387] eta: 0:07:00 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1505 (0.6723) acc1: 99.4792 (63.4717) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1757 data: 0.0847 max mem: 3532 +Test: [ 400/2387] eta: 0:06:56 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1513 (0.6594) acc1: 99.4792 (64.3690) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1768 data: 0.0857 max mem: 3532 +Test: [ 410/2387] eta: 0:06:52 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1515 (0.6471) acc1: 99.2188 (65.2208) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1708 data: 0.0800 max mem: 3532 +Test: [ 420/2387] eta: 0:06:49 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1506 (0.6352) acc1: 99.2188 (66.0326) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1748 data: 0.0824 max mem: 3532 +Test: [ 430/2387] eta: 0:06:45 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1528 (0.6242) acc1: 99.4792 (66.8093) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1738 data: 0.0814 max mem: 3532 +Test: [ 440/2387] eta: 0:06:41 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1544 (0.6129) acc1: 99.4792 (67.5536) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1705 data: 0.0804 max mem: 3533 +Test: [ 450/2387] eta: 0:06:37 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.0978 (0.6015) acc1: 100.0000 (68.2731) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1654 data: 0.0740 max mem: 3533 +Test: [ 460/2387] eta: 0:06:33 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.0968 (0.5906) acc1: 100.0000 (68.9613) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1636 data: 0.0722 max mem: 3533 +Test: [ 470/2387] eta: 0:06:30 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.0967 (0.5801) acc1: 100.0000 (69.6203) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1664 data: 0.0760 max mem: 3533 +Test: [ 480/2387] eta: 0:06:28 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.0973 (0.5701) acc1: 100.0000 (70.2519) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1880 data: 0.0966 max mem: 3533 +Test: [ 490/2387] eta: 0:06:25 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.0974 (0.5604) acc1: 100.0000 (70.8577) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1946 data: 0.1034 max mem: 3533 +Test: [ 500/2387] eta: 0:06:23 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.0979 (0.5512) acc1: 100.0000 (71.4394) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1862 data: 0.0940 max mem: 3533 +Test: [ 510/2387] eta: 0:06:20 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.0979 (0.5426) acc1: 100.0000 (71.9983) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1832 data: 0.0906 max mem: 3533 +Test: [ 520/2387] eta: 0:06:17 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1366 (0.5349) acc1: 100.0000 (72.5353) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1808 data: 0.0895 max mem: 3533 +Test: [ 530/2387] eta: 0:06:15 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1348 (0.5273) acc1: 100.0000 (73.0525) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1871 data: 0.0943 max mem: 3533 +Test: [ 540/2387] eta: 0:06:12 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1333 (0.5200) acc1: 100.0000 (73.5506) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1878 data: 0.0950 max mem: 3533 +Test: [ 550/2387] eta: 0:06:09 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1331 (0.5130) acc1: 100.0000 (74.0306) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1824 data: 0.0905 max mem: 3533 +Test: [ 560/2387] eta: 0:06:07 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1335 (0.5063) acc1: 100.0000 (74.4926) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1811 data: 0.0887 max mem: 3533 +Test: [ 570/2387] eta: 0:06:04 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1329 (0.4997) acc1: 100.0000 (74.9389) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1777 data: 0.0847 max mem: 3533 +Test: [ 580/2387] eta: 0:06:01 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1323 (0.4934) acc1: 100.0000 (75.3702) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1815 data: 0.0881 max mem: 3533 +Test: [ 590/2387] eta: 0:05:59 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1332 (0.4873) acc1: 100.0000 (75.7870) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1898 data: 0.0966 max mem: 3533 +Test: [ 600/2387] eta: 0:05:57 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1375 (0.4816) acc1: 100.0000 (76.1894) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1923 data: 0.1003 max mem: 3533 +Test: [ 610/2387] eta: 0:05:54 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1405 (0.4760) acc1: 100.0000 (76.5770) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1832 data: 0.0912 max mem: 3533 +Test: [ 620/2387] eta: 0:05:51 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1330 (0.4705) acc1: 100.0000 (76.9525) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1697 data: 0.0758 max mem: 3533 +Test: [ 630/2387] eta: 0:05:48 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1335 (0.4651) acc1: 100.0000 (77.3165) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1662 data: 0.0733 max mem: 3533 +Test: [ 640/2387] eta: 0:05:45 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1357 (0.4600) acc1: 100.0000 (77.6675) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1575 data: 0.0667 max mem: 3533 +Test: [ 650/2387] eta: 0:05:42 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1364 (0.4550) acc1: 99.7396 (78.0078) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1634 data: 0.0714 max mem: 3533 +Test: [ 660/2387] eta: 0:05:41 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1344 (0.4501) acc1: 100.0000 (78.3393) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1973 data: 0.1050 max mem: 3533 +Test: [ 670/2387] eta: 0:05:37 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1342 (0.4454) acc1: 100.0000 (78.6606) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1740 data: 0.0791 max mem: 3534 +Test: [ 680/2387] eta: 0:05:36 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1353 (0.4410) acc1: 100.0000 (78.9739) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1829 data: 0.0872 max mem: 3534 +Test: [ 690/2387] eta: 0:05:34 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1467 (0.4388) acc1: 100.0000 (79.1116) acc5: 100.0000 (100.0000) acc1_neg: 80.729 acc5_neg: 100.000 time: 0.2054 data: 0.1139 max mem: 3534 +Test: [ 700/2387] eta: 0:05:30 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.3343 (0.4374) acc1: 84.8958 (79.1878) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.1436 data: 0.0506 max mem: 3534 +Test: [ 710/2387] eta: 0:05:27 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.3387 (0.4361) acc1: 84.6354 (79.2645) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1354 data: 0.0410 max mem: 3534 +Test: [ 720/2387] eta: 0:05:24 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.3387 (0.4348) acc1: 85.1562 (79.3400) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.1587 data: 0.0648 max mem: 3534 +Test: [ 730/2387] eta: 0:05:22 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.3418 (0.4336) acc1: 84.3750 (79.4032) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.1689 data: 0.0708 max mem: 3534 +Test: [ 740/2387] eta: 0:05:19 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.3447 (0.4325) acc1: 83.3333 (79.4552) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.1661 data: 0.0670 max mem: 3534 +Test: [ 750/2387] eta: 0:05:16 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.3418 (0.4313) acc1: 83.8542 (79.5256) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.1581 data: 0.0628 max mem: 3534 +Test: [ 760/2387] eta: 0:05:14 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.3337 (0.4293) acc1: 85.9375 (79.6519) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1595 data: 0.0674 max mem: 3534 +Test: [ 770/2387] eta: 0:05:11 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1917 (0.4257) acc1: 95.0521 (79.8824) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1581 data: 0.0660 max mem: 3534 +Test: [ 780/2387] eta: 0:05:08 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1042 (0.4216) acc1: 100.0000 (80.1380) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1602 data: 0.0663 max mem: 3534 +Test: [ 790/2387] eta: 0:05:06 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.0993 (0.4179) acc1: 100.0000 (80.3637) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1548 data: 0.0599 max mem: 3534 +Test: [ 800/2387] eta: 0:05:03 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.0955 (0.4138) acc1: 100.0000 (80.6069) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1584 data: 0.0632 max mem: 3534 +Test: [ 810/2387] eta: 0:05:00 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.0957 (0.4099) acc1: 100.0000 (80.8444) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1567 data: 0.0606 max mem: 3534 +Test: [ 820/2387] eta: 0:04:57 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1016 (0.4064) acc1: 100.0000 (81.0724) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1359 data: 0.0412 max mem: 3534 +Test: [ 830/2387] eta: 0:04:54 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1263 (0.4037) acc1: 99.4792 (81.2644) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1324 data: 0.0399 max mem: 3534 +Test: [ 840/2387] eta: 0:04:52 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1620 (0.4011) acc1: 97.9167 (81.4627) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1589 data: 0.0657 max mem: 3534 +Test: [ 850/2387] eta: 0:04:50 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1913 (0.3988) acc1: 97.9167 (81.6478) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1715 data: 0.0788 max mem: 3534 +Test: [ 860/2387] eta: 0:04:48 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1906 (0.3964) acc1: 98.6979 (81.8492) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1649 data: 0.0712 max mem: 3534 +Test: [ 870/2387] eta: 0:04:45 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1862 (0.3939) acc1: 99.4792 (82.0399) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1645 data: 0.0713 max mem: 3534 +Test: [ 880/2387] eta: 0:04:43 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1749 (0.3917) acc1: 98.9583 (82.2207) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1549 data: 0.0633 max mem: 3534 +Test: [ 890/2387] eta: 0:04:40 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1729 (0.3891) acc1: 98.9583 (82.4141) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1492 data: 0.0560 max mem: 3534 +Test: [ 900/2387] eta: 0:04:38 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1821 (0.3910) acc1: 97.9167 (82.2766) acc5: 100.0000 (100.0000) acc1_neg: 66.406 acc5_neg: 100.000 time: 0.1699 data: 0.0763 max mem: 3535 +Test: [ 910/2387] eta: 0:04:37 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.6498 (0.3938) acc1: 63.0208 (82.0638) acc5: 100.0000 (100.0000) acc1_neg: 61.198 acc5_neg: 100.000 time: 0.1888 data: 0.0939 max mem: 3535 +Test: [ 920/2387] eta: 0:04:35 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.6619 (0.3967) acc1: 62.7604 (81.8557) acc5: 100.0000 (100.0000) acc1_neg: 61.979 acc5_neg: 100.000 time: 0.1982 data: 0.1039 max mem: 3535 +Test: [ 930/2387] eta: 0:04:33 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.6508 (0.3994) acc1: 63.0208 (81.6542) acc5: 100.0000 (100.0000) acc1_neg: 59.115 acc5_neg: 100.000 time: 0.1834 data: 0.0913 max mem: 3535 +Test: [ 940/2387] eta: 0:04:31 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.6556 (0.4021) acc1: 62.5000 (81.4498) acc5: 100.0000 (100.0000) acc1_neg: 57.812 acc5_neg: 100.000 time: 0.1766 data: 0.0842 max mem: 3535 +Test: [ 950/2387] eta: 0:04:29 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.6512 (0.4046) acc1: 62.5000 (81.2571) acc5: 100.0000 (100.0000) acc1_neg: 64.062 acc5_neg: 100.000 time: 0.1776 data: 0.0828 max mem: 3535 +Test: [ 960/2387] eta: 0:04:27 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.6321 (0.4071) acc1: 63.0208 (81.0747) acc5: 100.0000 (100.0000) acc1_neg: 61.719 acc5_neg: 100.000 time: 0.1699 data: 0.0752 max mem: 3535 +Test: [ 970/2387] eta: 0:04:25 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.6396 (0.4096) acc1: 63.0208 (80.8812) acc5: 100.0000 (100.0000) acc1_neg: 66.146 acc5_neg: 100.000 time: 0.1787 data: 0.0848 max mem: 3535 +Test: [ 980/2387] eta: 0:04:23 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.6237 (0.4089) acc1: 63.8021 (80.9230) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.1804 data: 0.0867 max mem: 3535 +Test: [ 990/2387] eta: 0:04:21 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1341 (0.4063) acc1: 99.7396 (81.0892) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1799 data: 0.0887 max mem: 3535 +Test: [1000/2387] eta: 0:04:19 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.0999 (0.4033) acc1: 100.0000 (81.2690) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1808 data: 0.0900 max mem: 3535 +Test: [1010/2387] eta: 0:04:17 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1073 (0.4006) acc1: 99.7396 (81.4463) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1814 data: 0.0909 max mem: 3535 +Test: [1020/2387] eta: 0:04:16 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1184 (0.3981) acc1: 99.7396 (81.6079) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2094 data: 0.1188 max mem: 3535 +Test: [1030/2387] eta: 0:04:13 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1093 (0.3953) acc1: 100.0000 (81.7837) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1884 data: 0.0965 max mem: 3535 +Test: [1040/2387] eta: 0:04:11 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1013 (0.3926) acc1: 100.0000 (81.9555) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1640 data: 0.0712 max mem: 3535 +Test: [1050/2387] eta: 0:04:09 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.0981 (0.3898) acc1: 100.0000 (82.1269) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1815 data: 0.0852 max mem: 3535 +Test: [1060/2387] eta: 0:04:08 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.0981 (0.3871) acc1: 100.0000 (82.2951) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2037 data: 0.1068 max mem: 3535 +Test: [1070/2387] eta: 0:04:06 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1075 (0.3846) acc1: 100.0000 (82.4563) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2087 data: 0.1135 max mem: 3535 +Test: [1080/2387] eta: 0:04:03 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1384 (0.3827) acc1: 99.2188 (82.5967) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1591 data: 0.0631 max mem: 3535 +Test: [1090/2387] eta: 0:04:01 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1326 (0.3802) acc1: 99.4792 (82.7555) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1462 data: 0.0513 max mem: 3535 +Test: [1100/2387] eta: 0:03:59 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1117 (0.3779) acc1: 100.0000 (82.9095) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1752 data: 0.0767 max mem: 3535 +Test: [1110/2387] eta: 0:03:58 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1223 (0.3757) acc1: 99.7396 (83.0521) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1910 data: 0.0942 max mem: 3535 +Test: [1120/2387] eta: 0:03:56 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1223 (0.3734) acc1: 99.7396 (83.1993) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1929 data: 0.1010 max mem: 3536 +Test: [1130/2387] eta: 0:03:54 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1240 (0.3712) acc1: 99.7396 (83.3453) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1886 data: 0.0933 max mem: 3536 +Test: [1140/2387] eta: 0:03:52 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1357 (0.3693) acc1: 99.7396 (83.4824) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1802 data: 0.0833 max mem: 3536 +Test: [1150/2387] eta: 0:03:50 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1473 (0.3677) acc1: 99.4792 (83.6042) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1758 data: 0.0804 max mem: 3536 +Test: [1160/2387] eta: 0:03:48 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1545 (0.3660) acc1: 99.4792 (83.7344) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1703 data: 0.0761 max mem: 3536 +Test: [1170/2387] eta: 0:03:46 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1610 (0.3642) acc1: 99.7396 (83.8682) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1681 data: 0.0757 max mem: 3536 +Test: [1180/2387] eta: 0:03:44 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1706 (0.3627) acc1: 99.4792 (83.9926) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1717 data: 0.0803 max mem: 3536 +Test: [1190/2387] eta: 0:03:42 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1645 (0.3610) acc1: 99.4792 (84.1211) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1732 data: 0.0823 max mem: 3536 +Test: [1200/2387] eta: 0:03:40 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1362 (0.3592) acc1: 100.0000 (84.2523) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1665 data: 0.0740 max mem: 3536 +Test: [1210/2387] eta: 0:03:38 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1327 (0.3573) acc1: 100.0000 (84.3812) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1614 data: 0.0686 max mem: 3536 +Test: [1220/2387] eta: 0:03:36 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.0950 (0.3550) acc1: 100.0000 (84.5085) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1711 data: 0.0793 max mem: 3536 +Test: [1230/2387] eta: 0:03:34 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.0818 (0.3528) acc1: 100.0000 (84.6339) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0868 max mem: 3536 +Test: [1240/2387] eta: 0:03:32 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.0820 (0.3506) acc1: 100.0000 (84.7573) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1785 data: 0.0843 max mem: 3536 +Test: [1250/2387] eta: 0:03:30 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.0795 (0.3485) acc1: 100.0000 (84.8779) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1767 data: 0.0841 max mem: 3536 +Test: [1260/2387] eta: 0:03:28 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.0763 (0.3463) acc1: 100.0000 (84.9974) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1657 data: 0.0744 max mem: 3536 +Test: [1270/2387] eta: 0:03:26 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.0776 (0.3442) acc1: 100.0000 (85.1153) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1839 data: 0.0923 max mem: 3536 +Test: [1280/2387] eta: 0:03:25 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.0813 (0.3422) acc1: 100.0000 (85.2311) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2140 data: 0.1209 max mem: 3536 +Test: [1290/2387] eta: 0:03:23 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.0897 (0.3403) acc1: 100.0000 (85.3402) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2089 data: 0.1109 max mem: 3536 +Test: [1300/2387] eta: 0:03:21 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.0920 (0.3385) acc1: 99.7396 (85.4503) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1710 data: 0.0725 max mem: 3536 +Test: [1310/2387] eta: 0:03:19 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.0920 (0.3366) acc1: 99.7396 (85.5579) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1599 data: 0.0664 max mem: 3536 +Test: [1320/2387] eta: 0:03:17 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.0931 (0.3349) acc1: 100.0000 (85.6629) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1859 data: 0.0950 max mem: 3536 +Test: [1330/2387] eta: 0:03:15 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1032 (0.3333) acc1: 99.7396 (85.7640) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1915 data: 0.1003 max mem: 3536 +Test: [1340/2387] eta: 0:03:14 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1357 (0.3321) acc1: 99.2188 (85.8451) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1908 data: 0.0957 max mem: 3536 +Test: [1350/2387] eta: 0:03:11 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.0780 (0.3302) acc1: 100.0000 (85.9498) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1757 data: 0.0794 max mem: 3537 +Test: [1360/2387] eta: 0:03:09 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.0723 (0.3283) acc1: 100.0000 (86.0531) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1589 data: 0.0666 max mem: 3537 +Test: [1370/2387] eta: 0:03:07 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.0718 (0.3264) acc1: 100.0000 (86.1548) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1593 data: 0.0686 max mem: 3537 +Test: [1380/2387] eta: 0:03:05 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.0693 (0.3246) acc1: 100.0000 (86.2551) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1536 data: 0.0627 max mem: 3537 +Test: [1390/2387] eta: 0:03:03 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.0685 (0.3227) acc1: 100.0000 (86.3539) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1502 data: 0.0595 max mem: 3537 +Test: [1400/2387] eta: 0:03:01 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.0665 (0.3209) acc1: 100.0000 (86.4513) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1541 data: 0.0636 max mem: 3537 +Test: [1410/2387] eta: 0:02:59 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.0695 (0.3194) acc1: 100.0000 (86.5445) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1647 data: 0.0725 max mem: 3537 +Test: [1420/2387] eta: 0:02:57 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1179 (0.3182) acc1: 100.0000 (86.6258) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1709 data: 0.0781 max mem: 3537 +Test: [1430/2387] eta: 0:02:55 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1307 (0.3168) acc1: 99.7396 (86.7147) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1660 data: 0.0745 max mem: 3537 +Test: [1440/2387] eta: 0:02:53 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1307 (0.3156) acc1: 99.7396 (86.8013) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1721 data: 0.0812 max mem: 3537 +Test: [1450/2387] eta: 0:02:51 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1510 (0.3146) acc1: 99.4792 (86.8763) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1728 data: 0.0816 max mem: 3537 +Test: [1460/2387] eta: 0:02:50 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1429 (0.3133) acc1: 99.4792 (86.9624) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1696 data: 0.0756 max mem: 3537 +Test: [1470/2387] eta: 0:02:48 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1108 (0.3120) acc1: 100.0000 (87.0502) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1803 data: 0.0871 max mem: 3537 +Test: [1480/2387] eta: 0:02:46 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1110 (0.3107) acc1: 100.0000 (87.1339) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1874 data: 0.0965 max mem: 3537 +Test: [1490/2387] eta: 0:02:44 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1125 (0.3093) acc1: 99.7396 (87.2183) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1807 data: 0.0876 max mem: 3537 +Test: [1500/2387] eta: 0:02:42 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1125 (0.3081) acc1: 99.7396 (87.3015) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1694 data: 0.0748 max mem: 3537 +Test: [1510/2387] eta: 0:02:40 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1305 (0.3070) acc1: 100.0000 (87.3844) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1658 data: 0.0727 max mem: 3537 +Test: [1520/2387] eta: 0:02:38 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1490 (0.3061) acc1: 100.0000 (87.4553) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1699 data: 0.0787 max mem: 3537 +Test: [1530/2387] eta: 0:02:36 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1846 (0.3054) acc1: 98.9583 (87.5262) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1621 data: 0.0717 max mem: 3537 +Test: [1540/2387] eta: 0:02:34 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1846 (0.3045) acc1: 99.2188 (87.5999) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1613 data: 0.0704 max mem: 3537 +Test: [1550/2387] eta: 0:02:33 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1735 (0.3037) acc1: 99.2188 (87.6698) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1755 data: 0.0844 max mem: 3537 +Test: [1560/2387] eta: 0:02:31 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1655 (0.3029) acc1: 99.2188 (87.7374) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1842 data: 0.0937 max mem: 3537 +Test: [1570/2387] eta: 0:02:29 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1492 (0.3020) acc1: 99.7396 (87.8115) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1812 data: 0.0892 max mem: 3537 +Test: [1580/2387] eta: 0:02:27 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1370 (0.3008) acc1: 99.7396 (87.8879) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1898 data: 0.0981 max mem: 3538 +Test: [1590/2387] eta: 0:02:25 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1000 (0.2995) acc1: 100.0000 (87.9639) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1794 data: 0.0869 max mem: 3538 +Test: [1600/2387] eta: 0:02:23 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1000 (0.2984) acc1: 100.0000 (88.0317) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1748 data: 0.0786 max mem: 3538 +Test: [1610/2387] eta: 0:02:22 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.0968 (0.2972) acc1: 100.0000 (88.1047) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1904 data: 0.0961 max mem: 3538 +Test: [1620/2387] eta: 0:02:20 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.0990 (0.2960) acc1: 100.0000 (88.1751) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1868 data: 0.0951 max mem: 3538 +Test: [1630/2387] eta: 0:02:18 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1149 (0.2951) acc1: 99.7396 (88.2388) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1928 data: 0.0993 max mem: 3538 +Test: [1640/2387] eta: 0:02:16 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1652 (0.2944) acc1: 98.9583 (88.2992) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1879 data: 0.0920 max mem: 3538 +Test: [1650/2387] eta: 0:02:14 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1665 (0.2937) acc1: 98.6979 (88.3582) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1701 data: 0.0766 max mem: 3538 +Test: [1660/2387] eta: 0:02:12 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1665 (0.2930) acc1: 98.6979 (88.4173) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1763 data: 0.0863 max mem: 3538 +Test: [1670/2387] eta: 0:02:11 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1660 (0.2922) acc1: 99.2188 (88.4789) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1950 data: 0.1040 max mem: 3538 +Test: [1680/2387] eta: 0:02:09 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1660 (0.2916) acc1: 98.6979 (88.5282) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1930 data: 0.1017 max mem: 3538 +Test: [1690/2387] eta: 0:02:07 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1798 (0.2910) acc1: 97.9167 (88.5791) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1911 data: 0.0987 max mem: 3538 +Test: [1700/2387] eta: 0:02:05 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1615 (0.2901) acc1: 99.2188 (88.6415) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1918 data: 0.0979 max mem: 3538 +Test: [1710/2387] eta: 0:02:04 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1087 (0.2890) acc1: 99.7396 (88.7073) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1955 data: 0.1025 max mem: 3538 +Test: [1720/2387] eta: 0:02:02 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.0903 (0.2878) acc1: 100.0000 (88.7723) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1891 data: 0.0952 max mem: 3538 +Test: [1730/2387] eta: 0:02:00 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.0921 (0.2867) acc1: 100.0000 (88.8355) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1858 data: 0.0918 max mem: 3538 +Test: [1740/2387] eta: 0:01:58 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.0947 (0.2856) acc1: 100.0000 (88.8983) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1794 data: 0.0860 max mem: 3538 +Test: [1750/2387] eta: 0:01:56 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.0838 (0.2844) acc1: 100.0000 (88.9615) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1846 data: 0.0919 max mem: 3538 +Test: [1760/2387] eta: 0:01:54 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.0848 (0.2833) acc1: 100.0000 (89.0239) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1843 data: 0.0927 max mem: 3538 +Test: [1770/2387] eta: 0:01:53 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.0911 (0.2823) acc1: 100.0000 (89.0822) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1836 data: 0.0915 max mem: 3538 +Test: [1780/2387] eta: 0:01:51 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.0954 (0.2813) acc1: 100.0000 (89.1429) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1887 data: 0.0958 max mem: 3538 +Test: [1790/2387] eta: 0:01:49 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.0953 (0.2802) acc1: 100.0000 (89.2028) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1841 data: 0.0909 max mem: 3538 +Test: [1800/2387] eta: 0:01:47 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.0955 (0.2792) acc1: 100.0000 (89.2606) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2103 data: 0.1186 max mem: 3538 +Test: [1810/2387] eta: 0:01:45 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.0857 (0.2781) acc1: 100.0000 (89.3192) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1992 data: 0.1065 max mem: 3539 +Test: [1820/2387] eta: 0:01:44 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.0857 (0.2771) acc1: 100.0000 (89.3775) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1796 data: 0.0860 max mem: 3539 +Test: [1830/2387] eta: 0:01:42 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.0945 (0.2762) acc1: 100.0000 (89.4354) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1996 data: 0.1076 max mem: 3539 +Test: [1840/2387] eta: 0:01:40 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1366 (0.2754) acc1: 100.0000 (89.4928) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1817 data: 0.0890 max mem: 3539 +Test: [1850/2387] eta: 0:01:38 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1400 (0.2747) acc1: 100.0000 (89.5496) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1460 data: 0.0529 max mem: 3539 +Test: [1860/2387] eta: 0:01:36 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1389 (0.2740) acc1: 100.0000 (89.6057) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1530 data: 0.0605 max mem: 3539 +Test: [1870/2387] eta: 0:01:34 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1389 (0.2733) acc1: 100.0000 (89.6613) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1708 data: 0.0773 max mem: 3539 +Test: [1880/2387] eta: 0:01:32 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1387 (0.2725) acc1: 100.0000 (89.7162) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1549 data: 0.0613 max mem: 3539 +Test: [1890/2387] eta: 0:01:30 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1381 (0.2718) acc1: 100.0000 (89.7705) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1482 data: 0.0552 max mem: 3539 +Test: [1900/2387] eta: 0:01:28 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1398 (0.2712) acc1: 100.0000 (89.8243) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1703 data: 0.0778 max mem: 3539 +Test: [1910/2387] eta: 0:01:27 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1436 (0.2706) acc1: 100.0000 (89.8771) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1719 data: 0.0777 max mem: 3539 +Test: [1920/2387] eta: 0:01:25 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1371 (0.2698) acc1: 100.0000 (89.9298) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1567 data: 0.0630 max mem: 3539 +Test: [1930/2387] eta: 0:01:23 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1022 (0.2689) acc1: 100.0000 (89.9817) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1723 data: 0.0795 max mem: 3539 +Test: [1940/2387] eta: 0:01:21 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1022 (0.2682) acc1: 100.0000 (90.0296) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1890 data: 0.0960 max mem: 3539 +Test: [1950/2387] eta: 0:01:19 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1002 (0.2674) acc1: 100.0000 (90.0807) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1797 data: 0.0865 max mem: 3539 +Test: [1960/2387] eta: 0:01:17 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1057 (0.2667) acc1: 100.0000 (90.1259) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1616 data: 0.0668 max mem: 3539 +Test: [1970/2387] eta: 0:01:16 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1300 (0.2664) acc1: 99.2188 (90.1524) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1662 data: 0.0716 max mem: 3539 +Test: [1980/2387] eta: 0:01:14 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1699 (0.2660) acc1: 97.6562 (90.1880) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1770 data: 0.0851 max mem: 3539 +Test: [1990/2387] eta: 0:01:12 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1711 (0.2655) acc1: 99.2188 (90.2340) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1699 data: 0.0784 max mem: 3539 +Test: [2000/2387] eta: 0:01:10 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1743 (0.2651) acc1: 99.4792 (90.2793) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1800 data: 0.0859 max mem: 3539 +Test: [2010/2387] eta: 0:01:08 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1747 (0.2646) acc1: 99.4792 (90.3250) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1893 data: 0.0967 max mem: 3539 +Test: [2020/2387] eta: 0:01:06 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1748 (0.2642) acc1: 99.2188 (90.3690) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1877 data: 0.0967 max mem: 3539 +Test: [2030/2387] eta: 0:01:05 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1734 (0.2637) acc1: 99.4792 (90.4136) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1814 data: 0.0886 max mem: 3540 +Test: [2040/2387] eta: 0:01:03 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1723 (0.2633) acc1: 99.4792 (90.4576) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1736 data: 0.0797 max mem: 3540 +Test: [2050/2387] eta: 0:01:01 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1715 (0.2628) acc1: 99.4792 (90.5020) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1706 data: 0.0780 max mem: 3540 +Test: [2060/2387] eta: 0:00:59 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1767 (0.2624) acc1: 99.2188 (90.5432) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1699 data: 0.0787 max mem: 3540 +Test: [2070/2387] eta: 0:00:57 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1811 (0.2620) acc1: 99.2188 (90.5856) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1688 data: 0.0771 max mem: 3540 +Test: [2080/2387] eta: 0:00:55 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1765 (0.2616) acc1: 99.2188 (90.6265) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1698 data: 0.0764 max mem: 3540 +Test: [2090/2387] eta: 0:00:54 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1774 (0.2612) acc1: 99.2188 (90.6677) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1741 data: 0.0818 max mem: 3540 +Test: [2100/2387] eta: 0:00:52 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1780 (0.2608) acc1: 99.2188 (90.7083) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1788 data: 0.0880 max mem: 3540 +Test: [2110/2387] eta: 0:00:50 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1780 (0.2604) acc1: 99.2188 (90.7498) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1794 data: 0.0882 max mem: 3540 +Test: [2120/2387] eta: 0:00:48 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1669 (0.2598) acc1: 99.7396 (90.7914) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1793 data: 0.0882 max mem: 3540 +Test: [2130/2387] eta: 0:00:46 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1140 (0.2591) acc1: 100.0000 (90.8345) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1764 data: 0.0830 max mem: 3540 +Test: [2140/2387] eta: 0:00:44 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1140 (0.2586) acc1: 100.0000 (90.8649) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1763 data: 0.0830 max mem: 3540 +Test: [2150/2387] eta: 0:00:43 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1195 (0.2580) acc1: 99.7396 (90.9065) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1781 data: 0.0873 max mem: 3540 +Test: [2160/2387] eta: 0:00:41 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1156 (0.2573) acc1: 100.0000 (90.9464) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1760 data: 0.0849 max mem: 3540 +Test: [2170/2387] eta: 0:00:39 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1384 (0.2569) acc1: 100.0000 (90.9799) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1817 data: 0.0901 max mem: 3540 +Test: [2180/2387] eta: 0:00:37 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1839 (0.2573) acc1: 97.1354 (90.9579) acc5: 100.0000 (100.0000) acc1_neg: 69.531 acc5_neg: 100.000 time: 0.1987 data: 0.1062 max mem: 3540 +Test: [2190/2387] eta: 0:00:35 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1859 (0.2571) acc1: 96.3542 (90.9697) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2027 data: 0.1079 max mem: 3540 +Test: [2200/2387] eta: 0:00:34 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1056 (0.2564) acc1: 99.4792 (91.0102) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1908 data: 0.0974 max mem: 3540 +Test: [2210/2387] eta: 0:00:32 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1005 (0.2558) acc1: 100.0000 (91.0498) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.1004 max mem: 3540 +Test: [2220/2387] eta: 0:00:30 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1111 (0.2551) acc1: 100.0000 (91.0873) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1950 data: 0.1003 max mem: 3540 +Test: [2230/2387] eta: 0:00:28 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1030 (0.2544) acc1: 100.0000 (91.1255) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.1026 max mem: 3540 +Test: [2240/2387] eta: 0:00:26 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.0931 (0.2538) acc1: 100.0000 (91.1606) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1990 data: 0.1074 max mem: 3540 +Test: [2250/2387] eta: 0:00:24 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1213 (0.2533) acc1: 99.7396 (91.1960) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1829 data: 0.0910 max mem: 3540 +Test: [2260/2387] eta: 0:00:23 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1320 (0.2527) acc1: 100.0000 (91.2348) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1754 data: 0.0835 max mem: 3541 +Test: [2270/2387] eta: 0:00:21 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1037 (0.2520) acc1: 100.0000 (91.2732) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1733 data: 0.0823 max mem: 3541 +Test: [2280/2387] eta: 0:00:19 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1014 (0.2515) acc1: 100.0000 (91.3002) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1679 data: 0.0749 max mem: 3541 +Test: [2290/2387] eta: 0:00:17 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1012 (0.2508) acc1: 100.0000 (91.3382) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1649 data: 0.0706 max mem: 3541 +Test: [2300/2387] eta: 0:00:15 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1041 (0.2502) acc1: 100.0000 (91.3737) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1657 data: 0.0732 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1120 (0.2497) acc1: 100.0000 (91.4081) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1711 data: 0.0798 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1252 (0.2493) acc1: 99.7396 (91.4374) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1767 data: 0.0833 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1331 (0.2488) acc1: 100.0000 (91.4717) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1844 data: 0.0911 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1264 (0.2483) acc1: 100.0000 (91.5065) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1935 data: 0.1003 max mem: 3541 +Test: [2350/2387] eta: 0:00:06 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1298 (0.2480) acc1: 99.7396 (91.5283) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1900 data: 0.0968 max mem: 3541 +Test: [2360/2387] eta: 0:00:04 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1383 (0.2475) acc1: 100.0000 (91.5635) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1836 data: 0.0904 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1396 (0.2472) acc1: 100.0000 (91.5921) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1942 data: 0.1018 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1640 (0.2470) acc1: 99.4792 (91.6175) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1969 data: 0.1065 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 5.479 acc5_pos: 100.000 loss: 0.1881 (0.2468) acc1: 97.6562 (91.6270) acc5: 100.0000 (100.0000) acc1_neg: 95.349 acc5_neg: 100.000 time: 0.1734 data: 0.0849 max mem: 3541 +Test: Total time: 0:07:15 (0.1823 s / it) +* Overall Acc@1 91.627 Acc@5 100.000 + target == real Acc@1 11.096 Acc@5 100.000 + target == fake Acc@1 97.282 Acc@5 100.000 + loss 0.247 + PR-AUC 0.9870 ROC-AUC 0.8395 EER 23.80% (thr=0.8418) +Accuracy of the network on the 916267 test images: 91.6% +[rank0]:[W726 03:52:47.921629159 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:53:06.721000 3147948 torch/distributed/run.py:766] +W0726 03:53:06.721000 3147948 torch/distributed/run.py:766] ***************************************** +W0726 03:53:06.721000 3147948 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:53:06.721000 3147948 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 5): env:// +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:53:33.544539525 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:53:34.758084887 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:53:34.244428915 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:53:34.343242933 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:53:34.447843781 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +| distributed init (rank 3): env:// +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:53:34.485857849 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:53:34.490434637 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank1]:[W726 03:53:34.493647581 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n16_seed6_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed6_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 0:55:14 acc1_pos: 9.115 acc5_pos: 100.000 loss: 1.3653 (1.3653) acc1: 9.1146 (9.1146) acc5: 100.0000 (100.0000) time: 6.4108 data: 4.6606 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/517] eta: 0:05:36 acc1_pos: 2.865 acc5_pos: 100.000 loss: 1.6165 (1.6183) acc1: 2.8646 (4.4744) acc5: 100.0000 (100.0000) time: 0.6644 data: 0.4240 max mem: 3531 +Test: [ 20/517] eta: 0:03:14 acc1_pos: 0.521 acc5_pos: 100.000 loss: 1.6134 (1.5915) acc1: 1.5625 (4.5139) acc5: 100.0000 (100.0000) time: 0.0905 data: 0.0003 max mem: 3531 +Test: [ 30/517] eta: 0:02:40 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.6849 (1.6284) acc1: 1.0417 (3.7466) acc5: 100.0000 (100.0000) time: 0.1444 data: 0.0529 max mem: 3531 +Test: [ 40/517] eta: 0:02:33 acc1_pos: 1.042 acc5_pos: 100.000 loss: 1.7250 (1.6387) acc1: 1.0417 (3.4172) acc5: 100.0000 (100.0000) time: 0.2468 data: 0.1527 max mem: 3531 +Test: [ 50/517] eta: 0:02:24 acc1_pos: 9.174 acc5_pos: 100.000 loss: 1.5814 (1.5557) acc1: 1.5625 (8.5580) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2779 data: 0.1807 max mem: 3531 +Test: [ 60/517] eta: 0:02:14 acc1_pos: 9.174 acc5_pos: 100.000 loss: 0.2566 (1.3381) acc1: 98.6979 (23.4332) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2399 data: 0.1425 max mem: 3531 +Test: [ 70/517] eta: 0:02:08 acc1_pos: 9.174 acc5_pos: 100.000 loss: 0.1683 (1.1717) acc1: 99.7396 (34.1953) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2299 data: 0.1330 max mem: 3531 +Test: [ 80/517] eta: 0:02:01 acc1_pos: 9.174 acc5_pos: 100.000 loss: 0.1558 (1.0471) acc1: 100.0000 (42.3193) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2299 data: 0.1334 max mem: 3531 +Test: [ 90/517] eta: 0:01:57 acc1_pos: 9.174 acc5_pos: 100.000 loss: 0.1672 (0.9516) acc1: 100.0000 (48.6579) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2299 data: 0.1356 max mem: 3531 +Test: [100/517] eta: 0:01:52 acc1_pos: 9.174 acc5_pos: 100.000 loss: 0.1654 (0.8726) acc1: 100.0000 (53.7387) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2399 data: 0.1444 max mem: 3531 +Test: [110/517] eta: 0:01:49 acc1_pos: 9.174 acc5_pos: 100.000 loss: 0.1603 (0.8092) acc1: 100.0000 (57.8852) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2399 data: 0.1416 max mem: 3531 +Test: [120/517] eta: 0:01:45 acc1_pos: 9.174 acc5_pos: 100.000 loss: 0.1781 (0.7584) acc1: 100.0000 (61.3400) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2399 data: 0.1418 max mem: 3531 +Test: [130/517] eta: 0:01:42 acc1_pos: 9.174 acc5_pos: 100.000 loss: 0.1973 (0.7151) acc1: 99.7396 (64.2673) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2399 data: 0.1416 max mem: 3531 +Test: [140/517] eta: 0:01:38 acc1_pos: 9.174 acc5_pos: 100.000 loss: 0.1811 (0.6769) acc1: 100.0000 (66.7978) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2299 data: 0.1333 max mem: 3531 +Test: [150/517] eta: 0:01:35 acc1_pos: 9.174 acc5_pos: 100.000 loss: 0.1760 (0.6435) acc1: 100.0000 (68.9966) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2299 data: 0.1328 max mem: 3531 +Test: [160/517] eta: 0:01:31 acc1_pos: 9.174 acc5_pos: 100.000 loss: 0.1775 (0.6147) acc1: 100.0000 (70.9191) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2356 data: 0.1367 max mem: 3531 +Test: [170/517] eta: 0:01:29 acc1_pos: 9.174 acc5_pos: 100.000 loss: 0.1772 (0.5886) acc1: 100.0000 (72.6182) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2454 data: 0.1491 max mem: 3531 +Test: [180/517] eta: 0:01:26 acc1_pos: 9.174 acc5_pos: 100.000 loss: 0.1664 (0.5652) acc1: 100.0000 (74.1209) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2408 data: 0.1447 max mem: 3531 +Test: [190/517] eta: 0:01:23 acc1_pos: 9.174 acc5_pos: 100.000 loss: 0.1783 (0.5461) acc1: 100.0000 (75.4758) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2256 data: 0.1305 max mem: 3531 +Test: [200/517] eta: 0:01:20 acc1_pos: 9.174 acc5_pos: 100.000 loss: 0.1919 (0.5275) acc1: 100.0000 (76.6947) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2309 data: 0.1359 max mem: 3531 +Test: [210/517] eta: 0:01:17 acc1_pos: 9.174 acc5_pos: 100.000 loss: 0.1831 (0.5116) acc1: 100.0000 (77.7930) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2402 data: 0.1429 max mem: 3532 +Test: [220/517] eta: 0:01:14 acc1_pos: 9.174 acc5_pos: 100.000 loss: 0.1873 (0.4968) acc1: 100.0000 (78.7943) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2450 data: 0.1497 max mem: 3532 +Test: [230/517] eta: 0:01:12 acc1_pos: 9.174 acc5_pos: 100.000 loss: 0.1913 (0.4841) acc1: 100.0000 (79.6999) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2447 data: 0.1513 max mem: 3532 +Test: [240/517] eta: 0:01:09 acc1_pos: 9.174 acc5_pos: 100.000 loss: 0.1943 (0.4720) acc1: 100.0000 (80.5422) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2397 data: 0.1458 max mem: 3532 +Test: [250/517] eta: 0:01:07 acc1_pos: 9.174 acc5_pos: 100.000 loss: 0.1813 (0.4601) acc1: 100.0000 (81.3174) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2436 data: 0.1487 max mem: 3532 +Test: [260/517] eta: 0:01:05 acc1_pos: 9.174 acc5_pos: 100.000 loss: 0.1600 (0.4487) acc1: 100.0000 (82.0303) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2730 data: 0.1791 max mem: 3532 +Test: [270/517] eta: 0:01:02 acc1_pos: 9.174 acc5_pos: 100.000 loss: 0.1793 (0.4393) acc1: 100.0000 (82.6924) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1611 max mem: 3532 +Test: [280/517] eta: 0:00:59 acc1_pos: 9.174 acc5_pos: 100.000 loss: 0.2011 (0.4312) acc1: 100.0000 (83.2462) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2359 data: 0.1390 max mem: 3532 +Test: [290/517] eta: 0:00:57 acc1_pos: 9.174 acc5_pos: 100.000 loss: 0.2115 (0.4237) acc1: 100.0000 (83.7987) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2449 data: 0.1462 max mem: 3532 +Test: [300/517] eta: 0:00:54 acc1_pos: 9.174 acc5_pos: 100.000 loss: 0.1928 (0.4158) acc1: 100.0000 (84.3361) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2383 data: 0.1376 max mem: 3532 +Test: [310/517] eta: 0:00:51 acc1_pos: 9.174 acc5_pos: 100.000 loss: 0.1953 (0.4090) acc1: 100.0000 (84.8372) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2399 data: 0.1359 max mem: 3532 +Test: [320/517] eta: 0:00:49 acc1_pos: 9.174 acc5_pos: 100.000 loss: 0.1982 (0.4017) acc1: 100.0000 (85.3088) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2349 data: 0.1326 max mem: 3532 +Test: [330/517] eta: 0:00:46 acc1_pos: 9.174 acc5_pos: 100.000 loss: 0.1504 (0.3940) acc1: 100.0000 (85.7526) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2399 data: 0.1422 max mem: 3532 +Test: [340/517] eta: 0:00:44 acc1_pos: 9.174 acc5_pos: 100.000 loss: 0.1427 (0.3868) acc1: 100.0000 (86.1697) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2372 data: 0.1417 max mem: 3532 +Test: [350/517] eta: 0:00:41 acc1_pos: 9.174 acc5_pos: 100.000 loss: 0.1866 (0.3817) acc1: 100.0000 (86.5607) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2249 data: 0.1297 max mem: 3532 +Test: [360/517] eta: 0:00:38 acc1_pos: 9.174 acc5_pos: 100.000 loss: 0.1936 (0.3763) acc1: 100.0000 (86.9258) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2293 data: 0.1354 max mem: 3532 +Test: [370/517] eta: 0:00:36 acc1_pos: 9.174 acc5_pos: 100.000 loss: 0.1488 (0.3699) acc1: 100.0000 (87.2782) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2375 data: 0.1423 max mem: 3532 +Test: [380/517] eta: 0:00:33 acc1_pos: 9.174 acc5_pos: 100.000 loss: 0.1291 (0.3634) acc1: 100.0000 (87.6121) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2356 data: 0.1396 max mem: 3532 +Test: [390/517] eta: 0:00:31 acc1_pos: 9.174 acc5_pos: 100.000 loss: 0.1283 (0.3575) acc1: 100.0000 (87.9289) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2240 data: 0.1297 max mem: 3532 +Test: [400/517] eta: 0:00:28 acc1_pos: 9.174 acc5_pos: 100.000 loss: 0.1319 (0.3520) acc1: 100.0000 (88.2299) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2201 data: 0.1258 max mem: 3532 +Test: [410/517] eta: 0:00:26 acc1_pos: 9.174 acc5_pos: 100.000 loss: 0.1418 (0.3471) acc1: 100.0000 (88.5163) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2243 data: 0.1301 max mem: 3532 +Test: [420/517] eta: 0:00:23 acc1_pos: 9.174 acc5_pos: 100.000 loss: 0.1698 (0.3430) acc1: 100.0000 (88.7891) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2300 data: 0.1359 max mem: 3532 +Test: [430/517] eta: 0:00:21 acc1_pos: 9.174 acc5_pos: 100.000 loss: 0.1713 (0.3391) acc1: 100.0000 (89.0486) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2343 data: 0.1392 max mem: 3532 +Test: [440/517] eta: 0:00:18 acc1_pos: 9.174 acc5_pos: 100.000 loss: 0.1761 (0.3355) acc1: 100.0000 (89.2916) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2322 data: 0.1358 max mem: 3533 +Test: [450/517] eta: 0:00:16 acc1_pos: 9.174 acc5_pos: 100.000 loss: 0.1808 (0.3320) acc1: 100.0000 (89.5285) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2346 data: 0.1372 max mem: 3533 +Test: [460/517] eta: 0:00:13 acc1_pos: 9.174 acc5_pos: 100.000 loss: 0.1793 (0.3287) acc1: 100.0000 (89.7534) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2308 data: 0.1338 max mem: 3533 +Test: [470/517] eta: 0:00:11 acc1_pos: 9.174 acc5_pos: 100.000 loss: 0.1729 (0.3253) acc1: 100.0000 (89.9670) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2268 data: 0.1332 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 9.174 acc5_pos: 100.000 loss: 0.1812 (0.3225) acc1: 100.0000 (90.1724) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2304 data: 0.1382 max mem: 3533 +Test: [490/517] eta: 0:00:06 acc1_pos: 9.174 acc5_pos: 100.000 loss: 0.1707 (0.3193) acc1: 100.0000 (90.3683) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2230 data: 0.1296 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 9.174 acc5_pos: 100.000 loss: 0.1843 (0.3171) acc1: 100.0000 (90.5559) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2296 data: 0.1347 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 9.174 acc5_pos: 100.000 loss: 0.2054 (0.3148) acc1: 100.0000 (90.7305) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2403 data: 0.1460 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 9.174 acc5_pos: 100.000 loss: 0.1986 (0.3134) acc1: 100.0000 (90.8324) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2157 data: 0.1200 max mem: 3533 +Test: Total time: 0:02:05 (0.2433 s / it) +* Overall Acc@1 90.832 Acc@5 100.000 + target == real Acc@1 3.989 Acc@5 100.000 + target == fake Acc@1 99.841 Acc@5 100.000 + loss 0.313 + PR-AUC 0.9435 ROC-AUC 0.6723 EER 38.15% (thr=0.8389) +Accuracy of the network on the 198427 test images: 90.8% +[rank0]:[W726 03:55:55.488693053 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 03:56:15.395000 3148511 torch/distributed/run.py:766] +W0726 03:56:15.395000 3148511 torch/distributed/run.py:766] ***************************************** +W0726 03:56:15.395000 3148511 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:56:15.395000 3148511 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:56:44.848593752 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:56:44.507615000 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:56:45.712350300 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:56:45.831118205 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +| distributed init (rank 7): env:// +[rank0]:[W726 03:56:45.856777443 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:56:45.859456934 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:56:45.868220068 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W726 03:56:45.868245881 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n16_seed6_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed6_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:44:18 acc1_pos: 14.844 acc5_pos: 100.000 loss: 1.0788 (1.0788) acc1: 14.8438 (14.8438) acc5: 100.0000 (100.0000) time: 6.8330 data: 4.7408 max mem: 3531 +Test: [ 10/389] eta: 0:04:29 acc1_pos: 86.719 acc5_pos: 100.000 loss: 0.8265 (0.8638) acc1: 37.5000 (42.5426) acc5: 100.0000 (100.0000) time: 0.7099 data: 0.4313 max mem: 3531 +Test: [ 20/389] eta: 0:02:34 acc1_pos: 7.292 acc5_pos: 100.000 loss: 0.7513 (0.8566) acc1: 37.5000 (43.3532) acc5: 100.0000 (100.0000) time: 0.0988 data: 0.0007 max mem: 3531 +Test: [ 30/389] eta: 0:02:03 acc1_pos: 21.354 acc5_pos: 100.000 loss: 1.1981 (0.9900) acc1: 7.2917 (31.7792) acc5: 100.0000 (100.0000) time: 0.1412 data: 0.0454 max mem: 3531 +Test: [ 40/389] eta: 0:01:48 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.2203 (1.0407) acc1: 6.2500 (28.2584) acc5: 100.0000 (100.0000) time: 0.1956 data: 0.1041 max mem: 3531 +Test: [ 50/389] eta: 0:01:35 acc1_pos: 30.208 acc5_pos: 100.000 loss: 1.1726 (1.0787) acc1: 11.4583 (25.0766) acc5: 100.0000 (100.0000) time: 0.1886 data: 0.0972 max mem: 3531 +Test: [ 60/389] eta: 0:01:28 acc1_pos: 1.562 acc5_pos: 100.000 loss: 0.8754 (1.0497) acc1: 24.4792 (25.2561) acc5: 100.0000 (100.0000) time: 0.1816 data: 0.0872 max mem: 3531 +Test: [ 70/389] eta: 0:01:21 acc1_pos: 34.896 acc5_pos: 100.000 loss: 0.8759 (1.0535) acc1: 26.0417 (24.7653) acc5: 100.0000 (100.0000) time: 0.1862 data: 0.0927 max mem: 3531 +Test: [ 80/389] eta: 0:01:15 acc1_pos: 68.229 acc5_pos: 100.000 loss: 0.8199 (1.0244) acc1: 29.6875 (27.4852) acc5: 100.0000 (100.0000) time: 0.1781 data: 0.0873 max mem: 3531 +Test: [ 90/389] eta: 0:01:10 acc1_pos: 29.948 acc5_pos: 100.000 loss: 0.7865 (0.9864) acc1: 45.0521 (30.7978) acc5: 100.0000 (100.0000) time: 0.1696 data: 0.0789 max mem: 3531 +Test: [100/389] eta: 0:01:06 acc1_pos: 28.646 acc5_pos: 100.000 loss: 0.8742 (0.9866) acc1: 34.1146 (30.3527) acc5: 100.0000 (100.0000) time: 0.1607 data: 0.0697 max mem: 3531 +Test: [110/389] eta: 0:01:01 acc1_pos: 92.188 acc5_pos: 100.000 loss: 0.9031 (0.9759) acc1: 25.2604 (31.3955) acc5: 100.0000 (100.0000) time: 0.1511 data: 0.0595 max mem: 3531 +Test: [120/389] eta: 0:00:58 acc1_pos: 51.562 acc5_pos: 100.000 loss: 0.8190 (0.9673) acc1: 39.8438 (31.9538) acc5: 100.0000 (100.0000) time: 0.1510 data: 0.0604 max mem: 3531 +Test: [130/389] eta: 0:00:55 acc1_pos: 3.681 acc5_pos: 100.000 loss: 0.7202 (0.9368) acc1: 55.4688 (34.7328) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1680 data: 0.0775 max mem: 3531 +Test: [140/389] eta: 0:00:52 acc1_pos: 3.681 acc5_pos: 100.000 loss: 0.3231 (0.9064) acc1: 94.5312 (37.9414) acc5: 100.0000 (100.0000) acc1_neg: 6.771 acc5_neg: 100.000 time: 0.1796 data: 0.0883 max mem: 3531 +Test: [150/389] eta: 0:00:49 acc1_pos: 3.681 acc5_pos: 100.000 loss: 0.3382 (0.8744) acc1: 98.1771 (41.1700) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1770 data: 0.0855 max mem: 3531 +Test: [160/389] eta: 0:00:47 acc1_pos: 3.681 acc5_pos: 100.000 loss: 0.3382 (0.8473) acc1: 98.1771 (43.9473) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1743 data: 0.0825 max mem: 3531 +Test: [170/389] eta: 0:00:44 acc1_pos: 3.681 acc5_pos: 100.000 loss: 0.2582 (0.8157) acc1: 100.0000 (47.0882) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1795 data: 0.0861 max mem: 3531 +Test: [180/389] eta: 0:00:42 acc1_pos: 3.681 acc5_pos: 100.000 loss: 0.2847 (0.7900) acc1: 100.0000 (49.6792) acc5: 100.0000 (100.0000) acc1_neg: 72.135 acc5_neg: 100.000 time: 0.1783 data: 0.0852 max mem: 3531 +Test: [190/389] eta: 0:00:40 acc1_pos: 3.681 acc5_pos: 100.000 loss: 0.3259 (0.7777) acc1: 94.5312 (50.9517) acc5: 100.0000 (100.0000) acc1_neg: 48.698 acc5_neg: 100.000 time: 0.1932 data: 0.1024 max mem: 3531 +Test: [200/389] eta: 0:00:38 acc1_pos: 3.681 acc5_pos: 100.000 loss: 0.3975 (0.7620) acc1: 87.2396 (52.6456) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1907 data: 0.1001 max mem: 3531 +Test: [210/389] eta: 0:00:36 acc1_pos: 3.681 acc5_pos: 100.000 loss: 0.3722 (0.7419) acc1: 90.8854 (54.6097) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1855 data: 0.0944 max mem: 3532 +Test: [220/389] eta: 0:00:34 acc1_pos: 3.681 acc5_pos: 100.000 loss: 0.2588 (0.7213) acc1: 100.0000 (56.6118) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2012 data: 0.1082 max mem: 3532 +Test: [230/389] eta: 0:00:31 acc1_pos: 3.681 acc5_pos: 100.000 loss: 0.2975 (0.7043) acc1: 99.7396 (58.3187) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1863 data: 0.0924 max mem: 3532 +Test: [240/389] eta: 0:00:29 acc1_pos: 3.681 acc5_pos: 100.000 loss: 0.3286 (0.6908) acc1: 97.1354 (59.7791) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1778 data: 0.0833 max mem: 3532 +Test: [250/389] eta: 0:00:27 acc1_pos: 3.681 acc5_pos: 100.000 loss: 0.3425 (0.6763) acc1: 96.0938 (61.2975) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1833 data: 0.0904 max mem: 3532 +Test: [260/389] eta: 0:00:25 acc1_pos: 3.681 acc5_pos: 100.000 loss: 0.3562 (0.6696) acc1: 97.3958 (62.1129) acc5: 100.0000 (100.0000) acc1_neg: 26.823 acc5_neg: 100.000 time: 0.1679 data: 0.0779 max mem: 3532 +Test: [270/389] eta: 0:00:23 acc1_pos: 3.681 acc5_pos: 100.000 loss: 0.4148 (0.6649) acc1: 96.6146 (62.4923) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1597 data: 0.0695 max mem: 3532 +Test: [280/389] eta: 0:00:21 acc1_pos: 3.681 acc5_pos: 100.000 loss: 0.3136 (0.6576) acc1: 93.4896 (63.1413) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1654 data: 0.0733 max mem: 3532 +Test: [290/389] eta: 0:00:19 acc1_pos: 3.681 acc5_pos: 100.000 loss: 0.2746 (0.6470) acc1: 96.0938 (64.1386) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1649 data: 0.0715 max mem: 3532 +Test: [300/389] eta: 0:00:17 acc1_pos: 3.681 acc5_pos: 100.000 loss: 0.4880 (0.6434) acc1: 85.1562 (64.7270) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.1684 data: 0.0770 max mem: 3532 +Test: [310/389] eta: 0:00:15 acc1_pos: 3.681 acc5_pos: 100.000 loss: 0.5430 (0.6374) acc1: 84.8958 (65.3520) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1625 data: 0.0721 max mem: 3532 +Test: [320/389] eta: 0:00:13 acc1_pos: 3.681 acc5_pos: 100.000 loss: 0.3613 (0.6290) acc1: 95.8333 (66.1832) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1587 data: 0.0679 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 3.681 acc5_pos: 100.000 loss: 0.3922 (0.6221) acc1: 97.6562 (67.0726) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1643 data: 0.0730 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 3.681 acc5_pos: 100.000 loss: 0.3922 (0.6131) acc1: 97.6562 (67.8168) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1624 data: 0.0712 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 3.681 acc5_pos: 100.000 loss: 0.3075 (0.6140) acc1: 89.0625 (67.6586) acc5: 100.0000 (100.0000) acc1_neg: 40.885 acc5_neg: 100.000 time: 0.1563 data: 0.0647 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 3.681 acc5_pos: 100.000 loss: 0.6280 (0.6153) acc1: 58.5938 (67.4133) acc5: 100.0000 (100.0000) acc1_neg: 66.146 acc5_neg: 100.000 time: 0.1503 data: 0.0577 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 3.681 acc5_pos: 100.000 loss: 0.6047 (0.6084) acc1: 66.1458 (68.0768) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1347 data: 0.0430 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 3.681 acc5_pos: 100.000 loss: 0.3232 (0.6034) acc1: 95.0521 (68.4629) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1201 data: 0.0300 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 3.681 acc5_pos: 100.000 loss: 0.3124 (0.5981) acc1: 95.3125 (68.9156) acc5: 100.0000 (100.0000) acc1_neg: 71.858 acc5_neg: 100.000 time: 0.1208 data: 0.0277 max mem: 3532 +Test: Total time: 0:01:11 (0.1831 s / it) +* Overall Acc@1 68.916 Acc@5 100.000 + target == real Acc@1 31.915 Acc@5 100.000 + target == fake Acc@1 86.115 Acc@5 100.000 + loss 0.598 + PR-AUC 0.7903 ROC-AUC 0.6665 EER 37.47% (thr=0.6743) +Accuracy of the network on the 149358 test images: 68.9% +[rank0]:[W726 03:58:05.298833253 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:58:25.516000 3149069 torch/distributed/run.py:766] +W0726 03:58:25.516000 3149069 torch/distributed/run.py:766] ***************************************** +W0726 03:58:25.516000 3149069 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:58:25.516000 3149069 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:58:54.597674634 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:58:54.730825311 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:58:54.881472728 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:58:54.918562305 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:58:54.082569496 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:58:54.115704409 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:58:54.158307204 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:58:54.169808575 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n16_seed6_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed6_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:30:35 acc1_pos: 10.156 acc5_pos: 100.000 loss: 1.2753 (1.2753) acc1: 10.1562 (10.1562) acc5: 100.0000 (100.0000) time: 5.5121 data: 3.4807 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/333] eta: 0:03:08 acc1_pos: 16.927 acc5_pos: 100.000 loss: 1.1721 (1.1900) acc1: 14.5833 (14.7491) acc5: 100.0000 (100.0000) time: 0.5827 data: 0.3167 max mem: 3531 +Test: [ 20/333] eta: 0:01:56 acc1_pos: 10.677 acc5_pos: 100.000 loss: 1.2741 (1.2602) acc1: 14.0625 (13.7525) acc5: 100.0000 (100.0000) time: 0.1165 data: 0.0263 max mem: 3531 +Test: [ 30/333] eta: 0:01:44 acc1_pos: 54.688 acc5_pos: 100.000 loss: 1.2506 (1.1552) acc1: 19.5312 (22.6731) acc5: 100.0000 (100.0000) time: 0.2116 data: 0.1184 max mem: 3531 +Test: [ 40/333] eta: 0:01:34 acc1_pos: 43.229 acc5_pos: 100.000 loss: 0.8707 (1.0853) acc1: 42.9688 (27.9281) acc5: 100.0000 (100.0000) time: 0.2665 data: 0.1735 max mem: 3531 +Test: [ 50/333] eta: 0:01:25 acc1_pos: 49.740 acc5_pos: 100.000 loss: 0.8486 (1.0414) acc1: 46.0938 (31.4951) acc5: 100.0000 (100.0000) time: 0.2400 data: 0.1497 max mem: 3531 +Test: [ 60/333] eta: 0:01:22 acc1_pos: 37.500 acc5_pos: 100.000 loss: 0.8645 (1.0118) acc1: 45.5729 (33.7901) acc5: 100.0000 (100.0000) time: 0.2606 data: 0.1695 max mem: 3531 +Test: [ 70/333] eta: 0:01:17 acc1_pos: 41.667 acc5_pos: 100.000 loss: 0.8681 (0.9893) acc1: 44.5312 (35.5707) acc5: 100.0000 (100.0000) time: 0.2699 data: 0.1778 max mem: 3531 +Test: [ 80/333] eta: 0:01:11 acc1_pos: 48.698 acc5_pos: 100.000 loss: 0.8459 (0.9709) acc1: 45.5729 (36.8281) acc5: 100.0000 (100.0000) time: 0.2295 data: 0.1377 max mem: 3531 +Test: [ 90/333] eta: 0:01:07 acc1_pos: 41.667 acc5_pos: 100.000 loss: 0.8613 (0.9617) acc1: 45.5729 (37.5200) acc5: 100.0000 (100.0000) time: 0.2149 data: 0.1241 max mem: 3531 +Test: [100/333] eta: 0:01:02 acc1_pos: 35.156 acc5_pos: 100.000 loss: 0.9115 (0.9636) acc1: 41.1458 (37.6702) acc5: 100.0000 (100.0000) time: 0.2090 data: 0.1178 max mem: 3531 +Test: [110/333] eta: 0:00:58 acc1_pos: 48.177 acc5_pos: 100.000 loss: 0.9728 (0.9626) acc1: 39.5833 (37.9763) acc5: 100.0000 (100.0000) time: 0.2051 data: 0.1136 max mem: 3531 +Test: [120/333] eta: 0:00:55 acc1_pos: 39.844 acc5_pos: 100.000 loss: 0.9792 (0.9641) acc1: 39.8438 (38.0187) acc5: 100.0000 (100.0000) time: 0.2190 data: 0.1281 max mem: 3531 +Test: [130/333] eta: 0:00:52 acc1_pos: 53.125 acc5_pos: 100.000 loss: 0.9228 (0.9621) acc1: 40.8854 (38.2475) acc5: 100.0000 (100.0000) time: 0.2226 data: 0.1301 max mem: 3531 +Test: [140/333] eta: 0:00:49 acc1_pos: 44.010 acc5_pos: 100.000 loss: 0.9228 (0.9623) acc1: 40.3646 (38.2129) acc5: 100.0000 (100.0000) time: 0.2201 data: 0.1283 max mem: 3531 +Test: [150/333] eta: 0:00:45 acc1_pos: 36.458 acc5_pos: 100.000 loss: 0.9562 (0.9641) acc1: 36.4583 (38.1295) acc5: 100.0000 (100.0000) time: 0.2086 data: 0.1178 max mem: 3531 +Test: [160/333] eta: 0:00:43 acc1_pos: 33.553 acc5_pos: 100.000 loss: 0.9163 (0.9554) acc1: 41.9271 (38.8894) acc5: 100.0000 (100.0000) acc1_neg: 66.406 acc5_neg: 100.000 time: 0.2123 data: 0.1184 max mem: 3531 +Test: [170/333] eta: 0:00:40 acc1_pos: 33.553 acc5_pos: 100.000 loss: 0.6083 (0.9330) acc1: 66.4062 (40.7788) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.2146 data: 0.1208 max mem: 3531 +Test: [180/333] eta: 0:00:37 acc1_pos: 33.553 acc5_pos: 100.000 loss: 0.5443 (0.9099) acc1: 73.1771 (42.7644) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1950 data: 0.1015 max mem: 3531 +Test: [190/333] eta: 0:00:34 acc1_pos: 33.553 acc5_pos: 100.000 loss: 0.4681 (0.8854) acc1: 80.4688 (44.8912) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.1910 data: 0.0980 max mem: 3531 +Test: [200/333] eta: 0:00:31 acc1_pos: 33.553 acc5_pos: 100.000 loss: 0.4303 (0.8642) acc1: 83.3333 (46.6820) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.1873 data: 0.0945 max mem: 3531 +Test: [210/333] eta: 0:00:28 acc1_pos: 33.553 acc5_pos: 100.000 loss: 0.4911 (0.8464) acc1: 77.6042 (48.1709) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1882 data: 0.0946 max mem: 3532 +Test: [220/333] eta: 0:00:26 acc1_pos: 33.553 acc5_pos: 100.000 loss: 0.4953 (0.8308) acc1: 77.0833 (49.3908) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.1913 data: 0.0993 max mem: 3532 +Test: [230/333] eta: 0:00:23 acc1_pos: 33.553 acc5_pos: 100.000 loss: 0.4914 (0.8159) acc1: 77.3438 (50.7091) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.1914 data: 0.0990 max mem: 3532 +Test: [240/333] eta: 0:00:21 acc1_pos: 33.553 acc5_pos: 100.000 loss: 0.5162 (0.8071) acc1: 74.4792 (51.3983) acc5: 100.0000 (100.0000) acc1_neg: 74.219 acc5_neg: 100.000 time: 0.1834 data: 0.0910 max mem: 3532 +Test: [250/333] eta: 0:00:18 acc1_pos: 33.553 acc5_pos: 100.000 loss: 0.5881 (0.7959) acc1: 69.7917 (52.2888) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.1787 data: 0.0868 max mem: 3532 +Test: [260/333] eta: 0:00:16 acc1_pos: 33.553 acc5_pos: 100.000 loss: 0.4195 (0.7805) acc1: 82.2917 (53.5810) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.1772 data: 0.0855 max mem: 3532 +Test: [270/333] eta: 0:00:14 acc1_pos: 33.553 acc5_pos: 100.000 loss: 0.3992 (0.7670) acc1: 84.3750 (54.7211) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.1593 data: 0.0678 max mem: 3532 +Test: [280/333] eta: 0:00:11 acc1_pos: 33.553 acc5_pos: 100.000 loss: 0.4310 (0.7588) acc1: 82.8125 (55.4039) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.1509 data: 0.0604 max mem: 3532 +Test: [290/333] eta: 0:00:09 acc1_pos: 33.553 acc5_pos: 100.000 loss: 0.6165 (0.7544) acc1: 66.6667 (55.7140) acc5: 100.0000 (100.0000) acc1_neg: 65.365 acc5_neg: 100.000 time: 0.1507 data: 0.0606 max mem: 3532 +Test: [300/333] eta: 0:00:07 acc1_pos: 33.553 acc5_pos: 100.000 loss: 0.6165 (0.7491) acc1: 66.6667 (56.1557) acc5: 100.0000 (100.0000) acc1_neg: 70.052 acc5_neg: 100.000 time: 0.1317 data: 0.0406 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 33.553 acc5_pos: 100.000 loss: 0.5421 (0.7395) acc1: 72.6562 (56.9676) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.1253 data: 0.0342 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 33.553 acc5_pos: 100.000 loss: 0.5099 (0.7326) acc1: 76.5625 (57.5578) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.1389 data: 0.0486 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 33.553 acc5_pos: 100.000 loss: 0.5237 (0.7264) acc1: 75.2604 (58.0769) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.1286 data: 0.0387 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 33.553 acc5_pos: 100.000 loss: 0.5223 (0.7251) acc1: 75.2604 (58.1721) acc5: 100.0000 (100.0000) acc1_neg: 75.493 acc5_neg: 100.000 time: 0.1299 data: 0.0369 max mem: 3532 +Test: Total time: 0:01:08 (0.2066 s / it) +* Overall Acc@1 58.172 Acc@5 100.000 + target == real Acc@1 38.144 Acc@5 100.000 + target == fake Acc@1 76.001 Acc@5 100.000 + loss 0.725 + PR-AUC 0.6035 ROC-AUC 0.6017 EER 42.79% (thr=0.6377) +Accuracy of the network on the 127843 test images: 58.2% +[rank0]:[W726 04:00:12.310171939 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:00:31.777000 3149628 torch/distributed/run.py:766] +W0726 04:00:31.777000 3149628 torch/distributed/run.py:766] ***************************************** +W0726 04:00:31.777000 3149628 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:00:31.777000 3149628 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:00:59.351566935 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank1]:[W726 04:00:59.352903434 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:00:59.534238726 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 04:01:00.575738073 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank4]:[W726 04:01:00.575770307 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:01:00.730958206 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:01:00.763630230 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:01:00.903238582 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n16_seed6_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed6_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:23:47 acc1_pos: 86.198 acc5_pos: 100.000 loss: 0.3531 (0.3531) acc1: 86.1979 (86.1979) acc5: 100.0000 (100.0000) time: 6.0751 data: 4.4228 max mem: 3531 +Test: [ 10/235] eta: 0:02:24 acc1_pos: 85.677 acc5_pos: 100.000 loss: 0.3579 (0.3577) acc1: 85.9375 (85.8665) acc5: 100.0000 (100.0000) time: 0.6405 data: 0.4025 max mem: 3531 +Test: [ 20/235] eta: 0:01:22 acc1_pos: 80.990 acc5_pos: 100.000 loss: 0.3540 (0.3554) acc1: 85.6771 (85.8383) acc5: 100.0000 (100.0000) time: 0.0969 data: 0.0004 max mem: 3531 +Test: [ 30/235] eta: 0:01:06 acc1_pos: 89.844 acc5_pos: 100.000 loss: 0.3487 (0.3532) acc1: 85.9375 (85.9375) acc5: 100.0000 (100.0000) time: 0.1533 data: 0.0571 max mem: 3531 +Test: [ 40/235] eta: 0:00:58 acc1_pos: 87.760 acc5_pos: 100.000 loss: 0.3481 (0.3517) acc1: 86.4583 (85.9883) acc5: 100.0000 (100.0000) time: 0.2162 data: 0.1211 max mem: 3531 +Test: [ 50/235] eta: 0:00:53 acc1_pos: 83.594 acc5_pos: 100.000 loss: 0.3481 (0.3529) acc1: 86.4583 (85.9426) acc5: 100.0000 (100.0000) time: 0.2372 data: 0.1451 max mem: 3531 +Test: [ 60/235] eta: 0:00:48 acc1_pos: 87.240 acc5_pos: 100.000 loss: 0.3506 (0.3528) acc1: 85.4167 (86.0656) acc5: 100.0000 (100.0000) time: 0.2303 data: 0.1390 max mem: 3531 +Test: [ 70/235] eta: 0:00:43 acc1_pos: 86.154 acc5_pos: 100.000 loss: 0.3816 (0.4499) acc1: 84.3750 (80.0103) acc5: 100.0000 (100.0000) acc1_neg: 35.417 acc5_neg: 100.000 time: 0.1904 data: 0.0980 max mem: 3531 +Test: [ 80/235] eta: 0:00:40 acc1_pos: 86.154 acc5_pos: 100.000 loss: 1.1274 (0.5344) acc1: 37.5000 (74.7782) acc5: 100.0000 (100.0000) acc1_neg: 35.156 acc5_neg: 100.000 time: 0.1995 data: 0.1069 max mem: 3531 +Test: [ 90/235] eta: 0:00:36 acc1_pos: 86.154 acc5_pos: 100.000 loss: 1.1366 (0.6006) acc1: 36.9792 (70.6759) acc5: 100.0000 (100.0000) acc1_neg: 35.938 acc5_neg: 100.000 time: 0.1986 data: 0.1054 max mem: 3531 +Test: [100/235] eta: 0:00:32 acc1_pos: 86.154 acc5_pos: 100.000 loss: 1.1313 (0.6518) acc1: 37.2396 (67.4299) acc5: 100.0000 (100.0000) acc1_neg: 35.677 acc5_neg: 100.000 time: 0.1796 data: 0.0860 max mem: 3531 +Test: [110/235] eta: 0:00:30 acc1_pos: 86.154 acc5_pos: 100.000 loss: 1.1158 (0.6949) acc1: 38.5417 (64.8461) acc5: 100.0000 (100.0000) acc1_neg: 40.885 acc5_neg: 100.000 time: 0.2084 data: 0.1163 max mem: 3531 +Test: [120/235] eta: 0:00:27 acc1_pos: 86.154 acc5_pos: 100.000 loss: 1.1792 (0.7439) acc1: 34.6354 (61.9189) acc5: 100.0000 (100.0000) acc1_neg: 28.385 acc5_neg: 100.000 time: 0.1949 data: 0.1010 max mem: 3531 +Test: [130/235] eta: 0:00:24 acc1_pos: 86.154 acc5_pos: 100.000 loss: 1.2765 (0.7847) acc1: 29.1667 (59.4684) acc5: 100.0000 (100.0000) acc1_neg: 30.990 acc5_neg: 100.000 time: 0.1716 data: 0.0768 max mem: 3531 +Test: [140/235] eta: 0:00:21 acc1_pos: 86.154 acc5_pos: 100.000 loss: 1.3219 (0.8238) acc1: 28.3854 (57.1624) acc5: 100.0000 (100.0000) acc1_neg: 25.000 acc5_neg: 100.000 time: 0.1815 data: 0.0898 max mem: 3531 +Test: [150/235] eta: 0:00:18 acc1_pos: 86.154 acc5_pos: 100.000 loss: 1.3219 (0.8550) acc1: 27.8646 (55.2359) acc5: 100.0000 (100.0000) acc1_neg: 28.906 acc5_neg: 100.000 time: 0.1718 data: 0.0786 max mem: 3531 +Test: [160/235] eta: 0:00:16 acc1_pos: 86.154 acc5_pos: 100.000 loss: 1.2960 (0.8831) acc1: 28.3854 (53.5941) acc5: 100.0000 (100.0000) acc1_neg: 30.469 acc5_neg: 100.000 time: 0.1644 data: 0.0701 max mem: 3531 +Test: [170/235] eta: 0:00:14 acc1_pos: 86.154 acc5_pos: 100.000 loss: 1.2386 (0.8750) acc1: 30.4688 (53.9763) acc5: 100.0000 (100.0000) acc1_neg: 60.156 acc5_neg: 100.000 time: 0.1721 data: 0.0771 max mem: 3531 +Test: [180/235] eta: 0:00:11 acc1_pos: 86.154 acc5_pos: 100.000 loss: 0.6938 (0.8647) acc1: 63.8021 (54.5264) acc5: 100.0000 (100.0000) acc1_neg: 64.323 acc5_neg: 100.000 time: 0.1738 data: 0.0796 max mem: 3531 +Test: [190/235] eta: 0:00:09 acc1_pos: 86.154 acc5_pos: 100.000 loss: 0.6938 (0.8558) acc1: 63.8021 (54.9806) acc5: 100.0000 (100.0000) acc1_neg: 64.062 acc5_neg: 100.000 time: 0.1661 data: 0.0752 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 86.154 acc5_pos: 100.000 loss: 0.6938 (0.8478) acc1: 63.2812 (55.3742) acc5: 100.0000 (100.0000) acc1_neg: 59.896 acc5_neg: 100.000 time: 0.1621 data: 0.0699 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 86.154 acc5_pos: 100.000 loss: 0.6967 (0.8402) acc1: 63.8021 (55.7909) acc5: 100.0000 (100.0000) acc1_neg: 65.625 acc5_neg: 100.000 time: 0.1712 data: 0.0777 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 86.154 acc5_pos: 100.000 loss: 0.6885 (0.8330) acc1: 64.5833 (56.1758) acc5: 100.0000 (100.0000) acc1_neg: 61.979 acc5_neg: 100.000 time: 0.1651 data: 0.0739 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 86.154 acc5_pos: 100.000 loss: 0.6714 (0.8258) acc1: 64.8438 (56.5521) acc5: 100.0000 (100.0000) acc1_neg: 65.885 acc5_neg: 100.000 time: 0.1313 data: 0.0412 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 86.154 acc5_pos: 100.000 loss: 0.6714 (0.8240) acc1: 64.5833 (56.6428) acc5: 100.0000 (100.0000) acc1_neg: 47.826 acc5_neg: 100.000 time: 0.1281 data: 0.0411 max mem: 3532 +Test: Total time: 0:00:46 (0.1997 s / it) +* Overall Acc@1 56.643 Acc@5 100.000 + target == real Acc@1 86.055 Acc@5 100.000 + target == fake Acc@1 46.005 Acc@5 100.000 + loss 0.824 + PR-AUC 0.8815 ROC-AUC 0.7323 EER 32.88% (thr=0.2827) +Accuracy of the network on the 89879 test images: 56.6% +[rank0]:[W726 04:01:58.234647065 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:02:17.048000 3150187 torch/distributed/run.py:766] +W0726 04:02:17.048000 3150187 torch/distributed/run.py:766] ***************************************** +W0726 04:02:17.048000 3150187 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:02:17.048000 3150187 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:02:45.163778885 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:02:45.388540156 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:02:45.444528160 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:02:46.903899802 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:02:46.960154426 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:02:46.029269592 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:02:46.035019497 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:02:46.133293204 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n16_seed6_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed6_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:30:49 acc1_pos: 19.271 acc5_pos: 100.000 loss: 1.0758 (1.0758) acc1: 19.2708 (19.2708) acc5: 100.0000 (100.0000) time: 6.8744 data: 5.0155 max mem: 3531 +Test: [ 10/269] eta: 0:03:03 acc1_pos: 37.500 acc5_pos: 100.000 loss: 0.9139 (0.9057) acc1: 35.6771 (36.2453) acc5: 100.0000 (100.0000) time: 0.7089 data: 0.4564 max mem: 3531 +Test: [ 20/269] eta: 0:01:44 acc1_pos: 37.760 acc5_pos: 100.000 loss: 0.9592 (0.9604) acc1: 30.7292 (32.5025) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0004 max mem: 3531 +Test: [ 30/269] eta: 0:01:22 acc1_pos: 41.667 acc5_pos: 100.000 loss: 1.0083 (0.9607) acc1: 27.3438 (33.3921) acc5: 100.0000 (100.0000) time: 0.1462 data: 0.0470 max mem: 3531 +Test: [ 40/269] eta: 0:01:14 acc1_pos: 2.586 acc5_pos: 100.000 loss: 0.6244 (0.8405) acc1: 49.2188 (45.4586) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2249 data: 0.1300 max mem: 3531 +Test: [ 50/269] eta: 0:01:07 acc1_pos: 2.586 acc5_pos: 100.000 loss: 0.4268 (0.7732) acc1: 82.2917 (51.8025) acc5: 100.0000 (100.0000) acc1_neg: 59.375 acc5_neg: 100.000 time: 0.2536 data: 0.1595 max mem: 3531 +Test: [ 60/269] eta: 0:01:01 acc1_pos: 2.586 acc5_pos: 100.000 loss: 0.5204 (0.7375) acc1: 73.9583 (55.0632) acc5: 100.0000 (100.0000) acc1_neg: 69.792 acc5_neg: 100.000 time: 0.2267 data: 0.1307 max mem: 3531 +Test: [ 70/269] eta: 0:00:55 acc1_pos: 2.586 acc5_pos: 100.000 loss: 0.5243 (0.7067) acc1: 73.1771 (57.9996) acc5: 100.0000 (100.0000) acc1_neg: 79.427 acc5_neg: 100.000 time: 0.2027 data: 0.1102 max mem: 3531 +Test: [ 80/269] eta: 0:00:51 acc1_pos: 2.586 acc5_pos: 100.000 loss: 0.5019 (0.6786) acc1: 76.5625 (60.6803) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.2039 data: 0.1136 max mem: 3531 +Test: [ 90/269] eta: 0:00:46 acc1_pos: 2.586 acc5_pos: 100.000 loss: 0.4952 (0.6619) acc1: 79.1667 (62.4342) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.1983 data: 0.1076 max mem: 3531 +Test: [100/269] eta: 0:00:43 acc1_pos: 2.586 acc5_pos: 100.000 loss: 0.5202 (0.6495) acc1: 75.7812 (63.6474) acc5: 100.0000 (100.0000) acc1_neg: 72.656 acc5_neg: 100.000 time: 0.1913 data: 0.0993 max mem: 3531 +Test: [110/269] eta: 0:00:39 acc1_pos: 2.586 acc5_pos: 100.000 loss: 0.4837 (0.6339) acc1: 78.1250 (65.2919) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.1896 data: 0.0959 max mem: 3531 +Test: [120/269] eta: 0:00:36 acc1_pos: 2.586 acc5_pos: 100.000 loss: 0.4258 (0.6147) acc1: 86.7188 (67.2349) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1943 data: 0.0994 max mem: 3531 +Test: [130/269] eta: 0:00:33 acc1_pos: 2.586 acc5_pos: 100.000 loss: 0.4384 (0.6111) acc1: 81.5104 (67.5791) acc5: 100.0000 (100.0000) acc1_neg: 57.812 acc5_neg: 100.000 time: 0.2150 data: 0.1218 max mem: 3531 +Test: [140/269] eta: 0:00:31 acc1_pos: 2.586 acc5_pos: 100.000 loss: 0.5157 (0.6021) acc1: 78.9062 (68.4767) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.2132 data: 0.1205 max mem: 3531 +Test: [150/269] eta: 0:00:28 acc1_pos: 2.586 acc5_pos: 100.000 loss: 0.3906 (0.5882) acc1: 89.8438 (69.8503) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.2138 data: 0.1195 max mem: 3531 +Test: [160/269] eta: 0:00:26 acc1_pos: 2.586 acc5_pos: 100.000 loss: 0.3772 (0.5757) acc1: 91.4062 (71.1423) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2507 data: 0.1559 max mem: 3531 +Test: [170/269] eta: 0:00:23 acc1_pos: 2.586 acc5_pos: 100.000 loss: 0.3954 (0.5675) acc1: 88.2812 (72.1324) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.2257 data: 0.1307 max mem: 3531 +Test: [180/269] eta: 0:00:20 acc1_pos: 2.586 acc5_pos: 100.000 loss: 0.4331 (0.5588) acc1: 88.0208 (73.1296) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.1813 data: 0.0860 max mem: 3531 +Test: [190/269] eta: 0:00:18 acc1_pos: 2.586 acc5_pos: 100.000 loss: 0.4195 (0.5548) acc1: 88.8021 (73.5684) acc5: 100.0000 (100.0000) acc1_neg: 77.344 acc5_neg: 100.000 time: 0.1941 data: 0.1001 max mem: 3531 +Test: [200/269] eta: 0:00:15 acc1_pos: 2.586 acc5_pos: 100.000 loss: 0.5084 (0.5542) acc1: 73.9583 (73.4880) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1882 data: 0.0944 max mem: 3531 +Test: [210/269] eta: 0:00:13 acc1_pos: 2.586 acc5_pos: 100.000 loss: 0.4249 (0.5472) acc1: 80.9896 (74.0990) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1699 data: 0.0745 max mem: 3532 +Test: [220/269] eta: 0:00:11 acc1_pos: 2.586 acc5_pos: 100.000 loss: 0.4369 (0.5429) acc1: 81.7708 (74.5428) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.1817 data: 0.0871 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 2.586 acc5_pos: 100.000 loss: 0.4481 (0.5419) acc1: 81.5104 (74.5727) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1770 data: 0.0847 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 2.586 acc5_pos: 100.000 loss: 0.4487 (0.5388) acc1: 80.7292 (74.8476) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1478 data: 0.0580 max mem: 3532 +Test: [250/269] eta: 0:00:04 acc1_pos: 2.586 acc5_pos: 100.000 loss: 0.4678 (0.5346) acc1: 80.7292 (75.2303) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.1160 data: 0.0266 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 2.586 acc5_pos: 100.000 loss: 0.4784 (0.5349) acc1: 79.9479 (75.1986) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1057 data: 0.0164 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 2.586 acc5_pos: 100.000 loss: 0.4676 (0.5317) acc1: 80.2083 (75.5540) acc5: 100.0000 (100.0000) acc1_neg: 97.375 acc5_neg: 100.000 time: 0.1062 data: 0.0133 max mem: 3532 +Test: Total time: 0:00:55 (0.2080 s / it) +* Overall Acc@1 75.554 Acc@5 100.000 + target == real Acc@1 32.530 Acc@5 100.000 + target == fake Acc@1 81.426 Acc@5 100.000 + loss 0.532 + PR-AUC 0.9086 ROC-AUC 0.6070 EER 42.33% (thr=0.6431) +Accuracy of the network on the 103293 test images: 75.6% +[rank0]:[W726 04:03:50.614659892 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T04:03:56Z] eval_done tag=fixed500k_real_n16_seed6_resampled +[2026-07-26T04:03:56Z] eval_start tag=fixed500k_fake_n32_seed6_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed6_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T04:03:56Z +hostname=n111-046-081 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed6_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n32_seed6_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 04:04:09.516000 3150750 torch/distributed/run.py:766] +W0726 04:04:09.516000 3150750 torch/distributed/run.py:766] ***************************************** +W0726 04:04:09.516000 3150750 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:04:09.516000 3150750 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 04:04:38.530796215 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +NCCL version 2.26.2+cuda12.2 +[rank2]:[W726 04:04:39.859840909 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +[rank5]:[W726 04:04:39.930455798 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:04:39.931736258 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:04:39.048779179 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:04:39.248678475 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:04:39.299717342 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank4]:[W726 04:04:39.299752811 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n32_seed6_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed6_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:25:52 acc1_pos: 43.750 acc5_pos: 100.000 loss: 0.7959 (0.7959) acc1: 43.7500 (43.7500) acc5: 100.0000 (100.0000) time: 6.6355 data: 5.1682 max mem: 3531 +Test: [ 10/234] eta: 0:02:33 acc1_pos: 44.271 acc5_pos: 100.000 loss: 0.5735 (0.6298) acc1: 66.4062 (63.3049) acc5: 100.0000 (100.0000) time: 0.6857 data: 0.4702 max mem: 3531 +Test: [ 20/234] eta: 0:01:27 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.5926 (0.6593) acc1: 59.3750 (59.6602) acc5: 100.0000 (100.0000) time: 0.0953 data: 0.0004 max mem: 3531 +Test: [ 30/234] eta: 0:01:10 acc1_pos: 95.833 acc5_pos: 100.000 loss: 0.1979 (0.5050) acc1: 96.8750 (71.9254) acc5: 100.0000 (100.0000) time: 0.1599 data: 0.0633 max mem: 3531 +Test: [ 40/234] eta: 0:01:01 acc1_pos: 99.479 acc5_pos: 100.000 loss: 0.1680 (0.4370) acc1: 97.6562 (78.0488) acc5: 100.0000 (100.0000) time: 0.2199 data: 0.1255 max mem: 3531 +Test: [ 50/234] eta: 0:00:54 acc1_pos: 99.740 acc5_pos: 100.000 loss: 0.1334 (0.3767) acc1: 98.4375 (82.1436) acc5: 100.0000 (100.0000) time: 0.2149 data: 0.1209 max mem: 3531 +Test: [ 60/234] eta: 0:00:48 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.1221 (0.3343) acc1: 99.2188 (85.0410) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1999 data: 0.1079 max mem: 3531 +Test: [ 70/234] eta: 0:00:43 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.1188 (0.3469) acc1: 99.4792 (84.1586) acc5: 100.0000 (100.0000) acc1_neg: 66.927 acc5_neg: 100.000 time: 0.1949 data: 0.1038 max mem: 3531 +Test: [ 80/234] eta: 0:00:39 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.5474 (0.3603) acc1: 69.2708 (83.4652) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1915 data: 0.1011 max mem: 3531 +Test: [ 90/234] eta: 0:00:35 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.2777 (0.3471) acc1: 89.0625 (84.4237) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.1749 data: 0.0829 max mem: 3531 +Test: [100/234] eta: 0:00:32 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.3086 (0.3677) acc1: 88.2812 (83.0420) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1783 data: 0.0859 max mem: 3531 +Test: [110/234] eta: 0:00:29 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.5958 (0.3810) acc1: 66.1458 (82.1673) acc5: 100.0000 (100.0000) acc1_neg: 63.542 acc5_neg: 100.000 time: 0.1874 data: 0.0960 max mem: 3531 +Test: [120/234] eta: 0:00:26 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.5516 (0.4076) acc1: 67.4479 (80.3332) acc5: 100.0000 (100.0000) acc1_neg: 39.583 acc5_neg: 100.000 time: 0.1849 data: 0.0909 max mem: 3531 +Test: [130/234] eta: 0:00:23 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.5163 (0.4149) acc1: 70.3125 (80.0990) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.1938 data: 0.0997 max mem: 3531 +Test: [140/234] eta: 0:00:21 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.5163 (0.4420) acc1: 70.5729 (78.6662) acc5: 100.0000 (100.0000) acc1_neg: 31.771 acc5_neg: 100.000 time: 0.1912 data: 0.0987 max mem: 3531 +Test: [150/234] eta: 0:00:18 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.6691 (0.4601) acc1: 62.2396 (77.4541) acc5: 100.0000 (100.0000) acc1_neg: 69.792 acc5_neg: 100.000 time: 0.1885 data: 0.0943 max mem: 3531 +Test: [160/234] eta: 0:00:16 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.6017 (0.4626) acc1: 65.6250 (77.4165) acc5: 100.0000 (100.0000) acc1_neg: 65.625 acc5_neg: 100.000 time: 0.1865 data: 0.0929 max mem: 3531 +Test: [170/234] eta: 0:00:14 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.5997 (0.4704) acc1: 67.7083 (76.8854) acc5: 100.0000 (100.0000) acc1_neg: 69.792 acc5_neg: 100.000 time: 0.1801 data: 0.0890 max mem: 3531 +Test: [180/234] eta: 0:00:11 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.6195 (0.4807) acc1: 65.1042 (76.2100) acc5: 100.0000 (100.0000) acc1_neg: 33.854 acc5_neg: 100.000 time: 0.1920 data: 0.1001 max mem: 3531 +Test: [190/234] eta: 0:00:09 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.6720 (0.4930) acc1: 62.5000 (75.3981) acc5: 100.0000 (100.0000) acc1_neg: 64.844 acc5_neg: 100.000 time: 0.1915 data: 0.0997 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.6718 (0.4844) acc1: 63.2812 (75.9069) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1845 data: 0.0936 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.1065 (0.4686) acc1: 99.7396 (76.9291) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1699 data: 0.0791 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.2278 (0.4603) acc1: 93.4896 (77.5134) acc5: 100.0000 (100.0000) acc1_neg: 64.583 acc5_neg: 100.000 time: 0.1374 data: 0.0475 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.6584 (0.5051) acc1: 15.6250 (74.6765) acc5: 100.0000 (100.0000) acc1_neg: 13.021 acc5_neg: 100.000 time: 0.1247 data: 0.0349 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 99.310 acc5_pos: 100.000 loss: 1.4702 (0.5177) acc1: 13.0208 (73.9761) acc5: 100.0000 (100.0000) acc1_neg: 15.175 acc5_neg: 100.000 time: 0.1257 data: 0.0348 max mem: 3532 +Test: Total time: 0:00:47 (0.2036 s / it) +* Overall Acc@1 73.976 Acc@5 100.000 + target == real Acc@1 82.270 Acc@5 100.000 + target == fake Acc@1 71.639 Acc@5 100.000 + loss 0.518 + PR-AUC 0.9582 ROC-AUC 0.8637 EER 23.14% (thr=0.4119) +Accuracy of the network on the 89729 test images: 74.0% +[rank0]:[W726 04:05:36.863471104 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:05:55.848000 3151312 torch/distributed/run.py:766] +W0726 04:05:55.848000 3151312 torch/distributed/run.py:766] ***************************************** +W0726 04:05:55.848000 3151312 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:05:55.848000 3151312 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:06:25.212253926 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W726 04:06:25.212255538 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:06:25.344993957 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:06:26.753832718 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:06:26.804983416 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:06:26.816412607 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:06:26.829584647 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:06:26.844969254 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n32_seed6_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed6_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:41:29 acc1_pos: 46.875 acc5_pos: 100.000 loss: 0.8150 (0.8150) acc1: 46.8750 (46.8750) acc5: 100.0000 (100.0000) time: 7.0754 data: 5.6273 max mem: 3531 +Test: [ 10/2387] eta: 0:28:54 acc1_pos: 18.490 acc5_pos: 100.000 loss: 0.9472 (0.9562) acc1: 35.4167 (34.1619) acc5: 100.0000 (100.0000) time: 0.7295 data: 0.5120 max mem: 3531 +Test: [ 20/2387] eta: 0:16:57 acc1_pos: 36.719 acc5_pos: 100.000 loss: 0.9435 (0.9377) acc1: 35.4167 (36.9048) acc5: 100.0000 (100.0000) time: 0.0974 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:15:21 acc1_pos: 20.312 acc5_pos: 100.000 loss: 0.9704 (0.9646) acc1: 27.8646 (34.5430) acc5: 100.0000 (100.0000) time: 0.2049 data: 0.1092 max mem: 3531 +Test: [ 40/2387] eta: 0:14:18 acc1_pos: 29.167 acc5_pos: 100.000 loss: 1.0681 (0.9998) acc1: 25.7812 (31.6438) acc5: 100.0000 (100.0000) time: 0.2983 data: 0.2058 max mem: 3531 +Test: [ 50/2387] eta: 0:13:05 acc1_pos: 41.146 acc5_pos: 100.000 loss: 1.0413 (0.9876) acc1: 28.6458 (32.4040) acc5: 100.0000 (100.0000) time: 0.2506 data: 0.1591 max mem: 3531 +Test: [ 60/2387] eta: 0:12:21 acc1_pos: 62.760 acc5_pos: 100.000 loss: 0.9384 (0.9905) acc1: 34.6354 (32.4197) acc5: 100.0000 (100.0000) time: 0.2223 data: 0.1311 max mem: 3531 +Test: [ 70/2387] eta: 0:11:53 acc1_pos: 37.760 acc5_pos: 100.000 loss: 1.0422 (1.0024) acc1: 26.8229 (31.6461) acc5: 100.0000 (100.0000) time: 0.2364 data: 0.1443 max mem: 3531 +Test: [ 80/2387] eta: 0:11:38 acc1_pos: 24.479 acc5_pos: 100.000 loss: 1.0873 (1.0164) acc1: 24.2188 (30.5266) acc5: 100.0000 (100.0000) time: 0.2542 data: 0.1614 max mem: 3531 +Test: [ 90/2387] eta: 0:11:20 acc1_pos: 32.292 acc5_pos: 100.000 loss: 1.0873 (1.0220) acc1: 24.2188 (30.3629) acc5: 100.0000 (100.0000) time: 0.2539 data: 0.1590 max mem: 3531 +Test: [ 100/2387] eta: 0:11:10 acc1_pos: 34.375 acc5_pos: 100.000 loss: 1.0860 (1.0352) acc1: 21.0938 (29.3394) acc5: 100.0000 (100.0000) time: 0.2549 data: 0.1585 max mem: 3531 +Test: [ 110/2387] eta: 0:10:57 acc1_pos: 32.552 acc5_pos: 100.000 loss: 1.0569 (1.0348) acc1: 26.3021 (29.5069) acc5: 100.0000 (100.0000) time: 0.2546 data: 0.1612 max mem: 3531 +Test: [ 120/2387] eta: 0:10:43 acc1_pos: 27.344 acc5_pos: 100.000 loss: 1.0474 (1.0368) acc1: 30.2083 (29.6143) acc5: 100.0000 (100.0000) time: 0.2356 data: 0.1458 max mem: 3531 +Test: [ 130/2387] eta: 0:10:35 acc1_pos: 39.062 acc5_pos: 100.000 loss: 1.0681 (1.0402) acc1: 25.7812 (29.3714) acc5: 100.0000 (100.0000) time: 0.2418 data: 0.1518 max mem: 3531 +Test: [ 140/2387] eta: 0:10:26 acc1_pos: 27.865 acc5_pos: 100.000 loss: 1.0623 (1.0422) acc1: 26.8229 (29.2923) acc5: 100.0000 (100.0000) time: 0.2493 data: 0.1546 max mem: 3531 +Test: [ 150/2387] eta: 0:10:19 acc1_pos: 8.073 acc5_pos: 100.000 loss: 1.0441 (1.0419) acc1: 29.6875 (29.2512) acc5: 100.0000 (100.0000) time: 0.2481 data: 0.1519 max mem: 3531 +Test: [ 160/2387] eta: 0:10:05 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.9884 (1.0299) acc1: 30.9896 (30.1647) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2219 data: 0.1301 max mem: 3531 +Test: [ 170/2387] eta: 0:09:56 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2801 (0.9849) acc1: 92.9688 (33.9349) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2101 data: 0.1198 max mem: 3531 +Test: [ 180/2387] eta: 0:09:45 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2611 (0.9449) acc1: 94.2708 (37.2899) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2132 data: 0.1215 max mem: 3531 +Test: [ 190/2387] eta: 0:09:38 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2561 (0.9089) acc1: 94.7917 (40.3114) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2139 data: 0.1203 max mem: 3531 +Test: [ 200/2387] eta: 0:09:29 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2546 (0.8764) acc1: 95.0521 (43.0089) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2146 data: 0.1200 max mem: 3531 +Test: [ 210/2387] eta: 0:09:14 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2633 (0.8475) acc1: 94.2708 (45.4273) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1704 data: 0.0764 max mem: 3532 +Test: [ 220/2387] eta: 0:09:07 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2642 (0.8212) acc1: 94.0104 (47.6303) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1765 data: 0.0834 max mem: 3532 +Test: [ 230/2387] eta: 0:08:59 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2651 (0.7971) acc1: 94.2708 (49.6426) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.2021 data: 0.1084 max mem: 3532 +Test: [ 240/2387] eta: 0:08:49 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2425 (0.7737) acc1: 95.5729 (51.5992) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1805 data: 0.0875 max mem: 3532 +Test: [ 250/2387] eta: 0:08:42 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2129 (0.7473) acc1: 97.9167 (53.5037) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1788 data: 0.0871 max mem: 3532 +Test: [ 260/2387] eta: 0:08:34 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.0949 (0.7223) acc1: 100.0000 (55.2852) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1848 data: 0.0936 max mem: 3532 +Test: [ 270/2387] eta: 0:08:27 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.0903 (0.6990) acc1: 100.0000 (56.9323) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1809 data: 0.0900 max mem: 3532 +Test: [ 280/2387] eta: 0:08:20 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.0906 (0.6775) acc1: 100.0000 (58.4631) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1791 data: 0.0883 max mem: 3532 +Test: [ 290/2387] eta: 0:08:14 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.0855 (0.6569) acc1: 100.0000 (59.8905) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1850 data: 0.0895 max mem: 3532 +Test: [ 300/2387] eta: 0:08:09 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.0742 (0.6377) acc1: 100.0000 (61.2230) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1919 data: 0.0967 max mem: 3532 +Test: [ 310/2387] eta: 0:08:04 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.0803 (0.6226) acc1: 100.0000 (62.4238) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1948 data: 0.1028 max mem: 3532 +Test: [ 320/2387] eta: 0:07:59 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2738 (0.6119) acc1: 98.4375 (63.5409) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1930 data: 0.1003 max mem: 3532 +Test: [ 330/2387] eta: 0:07:55 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2761 (0.6018) acc1: 98.4375 (64.5896) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1974 data: 0.1037 max mem: 3532 +Test: [ 340/2387] eta: 0:07:50 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2732 (0.5922) acc1: 98.1771 (65.5799) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1987 data: 0.1039 max mem: 3532 +Test: [ 350/2387] eta: 0:07:45 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2682 (0.5794) acc1: 99.2188 (66.5532) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0946 max mem: 3532 +Test: [ 360/2387] eta: 0:07:41 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1648 (0.5690) acc1: 97.9167 (67.3996) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1859 data: 0.0909 max mem: 3532 +Test: [ 370/2387] eta: 0:07:35 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2128 (0.5595) acc1: 97.1354 (68.1941) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1749 data: 0.0785 max mem: 3532 +Test: [ 380/2387] eta: 0:07:30 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2121 (0.5505) acc1: 96.8750 (68.9462) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1663 data: 0.0699 max mem: 3532 +Test: [ 390/2387] eta: 0:07:24 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2144 (0.5419) acc1: 97.1354 (69.6618) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1573 data: 0.0656 max mem: 3532 +Test: [ 400/2387] eta: 0:07:18 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2126 (0.5337) acc1: 97.1354 (70.3430) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1576 data: 0.0659 max mem: 3532 +Test: [ 410/2387] eta: 0:07:15 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2097 (0.5259) acc1: 96.8750 (70.9911) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1757 data: 0.0825 max mem: 3532 +Test: [ 420/2387] eta: 0:07:09 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2119 (0.5185) acc1: 96.8750 (71.6004) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1674 data: 0.0761 max mem: 3532 +Test: [ 430/2387] eta: 0:07:04 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2101 (0.5113) acc1: 97.1354 (72.2001) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1520 data: 0.0619 max mem: 3532 +Test: [ 440/2387] eta: 0:07:00 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1979 (0.5027) acc1: 98.1771 (72.8127) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1632 data: 0.0710 max mem: 3533 +Test: [ 450/2387] eta: 0:06:55 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.0929 (0.4936) acc1: 100.0000 (73.4156) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1627 data: 0.0706 max mem: 3533 +Test: [ 460/2387] eta: 0:06:52 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.0929 (0.4850) acc1: 100.0000 (73.9922) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1698 data: 0.0787 max mem: 3533 +Test: [ 470/2387] eta: 0:06:47 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.0929 (0.4767) acc1: 100.0000 (74.5444) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1719 data: 0.0787 max mem: 3533 +Test: [ 480/2387] eta: 0:06:44 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.0940 (0.4687) acc1: 100.0000 (75.0736) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0776 max mem: 3533 +Test: [ 490/2387] eta: 0:06:40 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.0937 (0.4611) acc1: 100.0000 (75.5813) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1760 data: 0.0857 max mem: 3533 +Test: [ 500/2387] eta: 0:06:37 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.0938 (0.4537) acc1: 100.0000 (76.0687) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1746 data: 0.0834 max mem: 3533 +Test: [ 510/2387] eta: 0:06:34 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.0939 (0.4468) acc1: 100.0000 (76.5370) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1788 data: 0.0863 max mem: 3533 +Test: [ 520/2387] eta: 0:06:30 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1139 (0.4405) acc1: 100.0000 (76.9874) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1746 data: 0.0810 max mem: 3533 +Test: [ 530/2387] eta: 0:06:26 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1254 (0.4346) acc1: 100.0000 (77.4207) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1667 data: 0.0732 max mem: 3533 +Test: [ 540/2387] eta: 0:06:24 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1269 (0.4290) acc1: 100.0000 (77.8381) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0813 max mem: 3533 +Test: [ 550/2387] eta: 0:06:21 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1280 (0.4235) acc1: 100.0000 (78.2398) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1897 data: 0.0965 max mem: 3533 +Test: [ 560/2387] eta: 0:06:18 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1290 (0.4183) acc1: 100.0000 (78.6263) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1857 data: 0.0923 max mem: 3533 +Test: [ 570/2387] eta: 0:06:16 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1290 (0.4132) acc1: 100.0000 (79.0002) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1888 data: 0.0951 max mem: 3533 +Test: [ 580/2387] eta: 0:06:13 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1261 (0.4083) acc1: 100.0000 (79.3607) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1891 data: 0.0947 max mem: 3533 +Test: [ 590/2387] eta: 0:06:10 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1272 (0.4036) acc1: 100.0000 (79.7095) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1759 data: 0.0816 max mem: 3533 +Test: [ 600/2387] eta: 0:06:07 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1233 (0.3988) acc1: 100.0000 (80.0458) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1799 data: 0.0875 max mem: 3533 +Test: [ 610/2387] eta: 0:06:04 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1513 (0.3959) acc1: 98.4375 (80.3311) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1792 data: 0.0860 max mem: 3533 +Test: [ 620/2387] eta: 0:06:01 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2172 (0.3930) acc1: 97.6562 (80.6105) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1754 data: 0.0818 max mem: 3533 +Test: [ 630/2387] eta: 0:05:59 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2187 (0.3902) acc1: 97.6562 (80.8786) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1842 data: 0.0905 max mem: 3533 +Test: [ 640/2387] eta: 0:05:56 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2238 (0.3877) acc1: 97.1354 (81.1289) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1894 data: 0.0952 max mem: 3533 +Test: [ 650/2387] eta: 0:05:54 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2228 (0.3851) acc1: 96.8750 (81.3792) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1915 data: 0.0967 max mem: 3533 +Test: [ 660/2387] eta: 0:05:51 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2209 (0.3826) acc1: 97.3958 (81.6219) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1830 data: 0.0890 max mem: 3533 +Test: [ 670/2387] eta: 0:05:48 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2176 (0.3802) acc1: 97.3958 (81.8566) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1649 data: 0.0721 max mem: 3534 +Test: [ 680/2387] eta: 0:05:45 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2148 (0.3776) acc1: 98.1771 (82.1096) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1670 data: 0.0745 max mem: 3534 +Test: [ 690/2387] eta: 0:05:43 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2166 (0.3773) acc1: 96.8750 (82.1537) acc5: 100.0000 (100.0000) acc1_neg: 77.083 acc5_neg: 100.000 time: 0.1791 data: 0.0865 max mem: 3534 +Test: [ 700/2387] eta: 0:05:40 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.4011 (0.3777) acc1: 81.2500 (82.1371) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.1852 data: 0.0903 max mem: 3534 +Test: [ 710/2387] eta: 0:05:37 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.4115 (0.3781) acc1: 81.2500 (82.1203) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1770 data: 0.0827 max mem: 3534 +Test: [ 720/2387] eta: 0:05:35 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.4105 (0.3785) acc1: 80.9896 (82.1111) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.1671 data: 0.0741 max mem: 3534 +Test: [ 730/2387] eta: 0:05:33 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.4097 (0.3790) acc1: 80.4688 (82.0847) acc5: 100.0000 (100.0000) acc1_neg: 79.427 acc5_neg: 100.000 time: 0.1886 data: 0.0964 max mem: 3534 +Test: [ 740/2387] eta: 0:05:30 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.4068 (0.3794) acc1: 80.2083 (82.0667) acc5: 100.0000 (100.0000) acc1_neg: 77.344 acc5_neg: 100.000 time: 0.1842 data: 0.0921 max mem: 3534 +Test: [ 750/2387] eta: 0:05:27 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.4024 (0.3796) acc1: 80.2083 (82.0489) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.1648 data: 0.0714 max mem: 3534 +Test: [ 760/2387] eta: 0:05:24 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.3875 (0.3793) acc1: 80.7292 (82.0720) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1655 data: 0.0731 max mem: 3534 +Test: [ 770/2387] eta: 0:05:22 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2748 (0.3773) acc1: 89.5833 (82.2170) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1695 data: 0.0786 max mem: 3534 +Test: [ 780/2387] eta: 0:05:19 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1513 (0.3742) acc1: 97.6562 (82.4324) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1706 data: 0.0794 max mem: 3534 +Test: [ 790/2387] eta: 0:05:16 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1361 (0.3715) acc1: 98.6979 (82.6031) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1628 data: 0.0718 max mem: 3534 +Test: [ 800/2387] eta: 0:05:14 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1311 (0.3685) acc1: 99.2188 (82.8112) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1667 data: 0.0768 max mem: 3534 +Test: [ 810/2387] eta: 0:05:12 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1311 (0.3656) acc1: 99.4792 (83.0154) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1811 data: 0.0891 max mem: 3534 +Test: [ 820/2387] eta: 0:05:09 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1329 (0.3629) acc1: 99.4792 (83.2017) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1765 data: 0.0841 max mem: 3534 +Test: [ 830/2387] eta: 0:05:06 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1563 (0.3608) acc1: 97.9167 (83.3647) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1600 data: 0.0681 max mem: 3534 +Test: [ 840/2387] eta: 0:05:03 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2439 (0.3602) acc1: 95.0521 (83.4640) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1533 data: 0.0602 max mem: 3534 +Test: [ 850/2387] eta: 0:05:01 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.3062 (0.3597) acc1: 92.7083 (83.5555) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1608 data: 0.0693 max mem: 3534 +Test: [ 860/2387] eta: 0:04:58 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2887 (0.3588) acc1: 94.0104 (83.6869) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1620 data: 0.0712 max mem: 3534 +Test: [ 870/2387] eta: 0:04:56 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2730 (0.3579) acc1: 95.8333 (83.8036) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1660 data: 0.0741 max mem: 3534 +Test: [ 880/2387] eta: 0:04:54 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2666 (0.3572) acc1: 94.7917 (83.9086) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1840 data: 0.0911 max mem: 3534 +Test: [ 890/2387] eta: 0:04:52 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2403 (0.3555) acc1: 96.8750 (84.0696) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1815 data: 0.0891 max mem: 3534 +Test: [ 900/2387] eta: 0:04:49 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2506 (0.3601) acc1: 94.7917 (83.7007) acc5: 100.0000 (100.0000) acc1_neg: 39.583 acc5_neg: 100.000 time: 0.1700 data: 0.0783 max mem: 3535 +Test: [ 910/2387] eta: 0:04:47 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.9055 (0.3661) acc1: 39.5833 (83.2116) acc5: 100.0000 (100.0000) acc1_neg: 40.365 acc5_neg: 100.000 time: 0.1726 data: 0.0812 max mem: 3535 +Test: [ 920/2387] eta: 0:04:45 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.9047 (0.3720) acc1: 39.0625 (82.7322) acc5: 100.0000 (100.0000) acc1_neg: 36.458 acc5_neg: 100.000 time: 0.1686 data: 0.0782 max mem: 3535 +Test: [ 930/2387] eta: 0:04:42 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.9030 (0.3777) acc1: 39.0625 (82.2760) acc5: 100.0000 (100.0000) acc1_neg: 35.417 acc5_neg: 100.000 time: 0.1589 data: 0.0684 max mem: 3535 +Test: [ 940/2387] eta: 0:04:40 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.9165 (0.3834) acc1: 39.0625 (81.8187) acc5: 100.0000 (100.0000) acc1_neg: 39.062 acc5_neg: 100.000 time: 0.1599 data: 0.0691 max mem: 3535 +Test: [ 950/2387] eta: 0:04:37 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.9110 (0.3888) acc1: 39.0625 (81.3688) acc5: 100.0000 (100.0000) acc1_neg: 40.365 acc5_neg: 100.000 time: 0.1604 data: 0.0689 max mem: 3535 +Test: [ 960/2387] eta: 0:04:35 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.8929 (0.3941) acc1: 39.5833 (80.9389) acc5: 100.0000 (100.0000) acc1_neg: 41.406 acc5_neg: 100.000 time: 0.1549 data: 0.0628 max mem: 3535 +Test: [ 970/2387] eta: 0:04:32 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.9006 (0.3995) acc1: 39.3229 (80.4996) acc5: 100.0000 (100.0000) acc1_neg: 35.417 acc5_neg: 100.000 time: 0.1514 data: 0.0590 max mem: 3535 +Test: [ 980/2387] eta: 0:04:29 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.8880 (0.4002) acc1: 40.1042 (80.4435) acc5: 100.0000 (100.0000) acc1_neg: 77.344 acc5_neg: 100.000 time: 0.1462 data: 0.0550 max mem: 3535 +Test: [ 990/2387] eta: 0:04:27 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1913 (0.3978) acc1: 97.3958 (80.5944) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1492 data: 0.0588 max mem: 3535 +Test: [1000/2387] eta: 0:04:25 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1004 (0.3950) acc1: 99.7396 (80.7718) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1686 data: 0.0775 max mem: 3535 +Test: [1010/2387] eta: 0:04:22 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1091 (0.3924) acc1: 99.7396 (80.9484) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1639 data: 0.0707 max mem: 3535 +Test: [1020/2387] eta: 0:04:20 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1320 (0.3903) acc1: 98.9583 (81.1023) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1381 data: 0.0449 max mem: 3535 +Test: [1030/2387] eta: 0:04:17 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1405 (0.3880) acc1: 97.9167 (81.2619) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1393 data: 0.0475 max mem: 3535 +Test: [1040/2387] eta: 0:04:15 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1346 (0.3855) acc1: 99.4792 (81.4351) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1668 data: 0.0753 max mem: 3535 +Test: [1050/2387] eta: 0:04:13 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1282 (0.3832) acc1: 99.4792 (81.5984) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1649 data: 0.0703 max mem: 3535 +Test: [1060/2387] eta: 0:04:10 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1294 (0.3809) acc1: 98.9583 (81.7618) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1464 data: 0.0499 max mem: 3535 +Test: [1070/2387] eta: 0:04:08 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1335 (0.3788) acc1: 98.9583 (81.9136) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1559 data: 0.0634 max mem: 3535 +Test: [1080/2387] eta: 0:04:06 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1724 (0.3772) acc1: 97.1354 (82.0433) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1652 data: 0.0742 max mem: 3535 +Test: [1090/2387] eta: 0:04:04 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1427 (0.3748) acc1: 98.6979 (82.2014) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1639 data: 0.0724 max mem: 3535 +Test: [1100/2387] eta: 0:04:02 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1263 (0.3727) acc1: 99.4792 (82.3527) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1609 data: 0.0689 max mem: 3535 +Test: [1110/2387] eta: 0:03:59 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1263 (0.3709) acc1: 99.2188 (82.4686) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1606 data: 0.0689 max mem: 3535 +Test: [1120/2387] eta: 0:03:58 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1297 (0.3688) acc1: 99.7396 (82.6183) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1897 data: 0.0984 max mem: 3536 +Test: [1130/2387] eta: 0:03:56 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1271 (0.3666) acc1: 99.7396 (82.7658) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1864 data: 0.0948 max mem: 3536 +Test: [1140/2387] eta: 0:03:54 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1271 (0.3646) acc1: 99.7396 (82.9109) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1672 data: 0.0736 max mem: 3536 +Test: [1150/2387] eta: 0:03:51 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1571 (0.3636) acc1: 98.6979 (83.0148) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1648 data: 0.0712 max mem: 3536 +Test: [1160/2387] eta: 0:03:49 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2162 (0.3626) acc1: 95.5729 (83.1095) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1611 data: 0.0670 max mem: 3536 +Test: [1170/2387] eta: 0:03:47 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2407 (0.3617) acc1: 94.5312 (83.2092) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1679 data: 0.0727 max mem: 3536 +Test: [1180/2387] eta: 0:03:45 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2774 (0.3609) acc1: 93.7500 (83.3007) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1675 data: 0.0730 max mem: 3536 +Test: [1190/2387] eta: 0:03:43 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2042 (0.3597) acc1: 96.6146 (83.4166) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1756 data: 0.0824 max mem: 3536 +Test: [1200/2387] eta: 0:03:41 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1714 (0.3581) acc1: 98.9583 (83.5445) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1714 data: 0.0806 max mem: 3536 +Test: [1210/2387] eta: 0:03:39 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1631 (0.3566) acc1: 99.7396 (83.6768) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1729 data: 0.0820 max mem: 3536 +Test: [1220/2387] eta: 0:03:37 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1452 (0.3547) acc1: 99.4792 (83.8045) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1830 data: 0.0918 max mem: 3536 +Test: [1230/2387] eta: 0:03:35 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1142 (0.3527) acc1: 99.4792 (83.9333) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1798 data: 0.0871 max mem: 3536 +Test: [1240/2387] eta: 0:03:34 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1114 (0.3508) acc1: 99.7396 (84.0596) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1846 data: 0.0922 max mem: 3536 +Test: [1250/2387] eta: 0:03:32 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1123 (0.3489) acc1: 99.7396 (84.1847) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1844 data: 0.0934 max mem: 3536 +Test: [1260/2387] eta: 0:03:30 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1051 (0.3469) acc1: 99.7396 (84.3069) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1770 data: 0.0851 max mem: 3536 +Test: [1270/2387] eta: 0:03:28 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.0896 (0.3448) acc1: 100.0000 (84.4301) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1804 data: 0.0889 max mem: 3536 +Test: [1280/2387] eta: 0:03:26 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.0883 (0.3429) acc1: 100.0000 (84.5470) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1840 data: 0.0920 max mem: 3536 +Test: [1290/2387] eta: 0:03:24 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1058 (0.3412) acc1: 99.2188 (84.6461) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1905 data: 0.0982 max mem: 3536 +Test: [1300/2387] eta: 0:03:22 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1062 (0.3394) acc1: 99.2188 (84.7543) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1791 data: 0.0875 max mem: 3536 +Test: [1310/2387] eta: 0:03:20 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1095 (0.3377) acc1: 99.4792 (84.8589) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1698 data: 0.0766 max mem: 3536 +Test: [1320/2387] eta: 0:03:18 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1124 (0.3361) acc1: 99.2188 (84.9599) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1753 data: 0.0811 max mem: 3536 +Test: [1330/2387] eta: 0:03:16 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1145 (0.3345) acc1: 99.4792 (85.0680) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1740 data: 0.0792 max mem: 3536 +Test: [1340/2387] eta: 0:03:14 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1393 (0.3340) acc1: 98.6979 (85.1005) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1808 data: 0.0865 max mem: 3536 +Test: [1350/2387] eta: 0:03:12 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.0976 (0.3322) acc1: 100.0000 (85.2106) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1765 data: 0.0841 max mem: 3537 +Test: [1360/2387] eta: 0:03:10 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.0750 (0.3303) acc1: 100.0000 (85.3193) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1785 data: 0.0877 max mem: 3537 +Test: [1370/2387] eta: 0:03:09 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.0729 (0.3284) acc1: 100.0000 (85.4262) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1833 data: 0.0905 max mem: 3537 +Test: [1380/2387] eta: 0:03:07 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.0721 (0.3266) acc1: 100.0000 (85.5317) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1800 data: 0.0854 max mem: 3537 +Test: [1390/2387] eta: 0:03:05 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.0727 (0.3248) acc1: 100.0000 (85.6355) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1731 data: 0.0786 max mem: 3537 +Test: [1400/2387] eta: 0:03:03 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.0669 (0.3229) acc1: 100.0000 (85.7381) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1731 data: 0.0787 max mem: 3537 +Test: [1410/2387] eta: 0:03:01 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.0705 (0.3219) acc1: 100.0000 (85.8105) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1918 data: 0.0991 max mem: 3537 +Test: [1420/2387] eta: 0:02:59 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2208 (0.3214) acc1: 94.5312 (85.8543) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1921 data: 0.0980 max mem: 3537 +Test: [1430/2387] eta: 0:02:57 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2330 (0.3208) acc1: 92.9688 (85.9033) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1802 data: 0.0867 max mem: 3537 +Test: [1440/2387] eta: 0:02:55 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2115 (0.3199) acc1: 95.0521 (85.9814) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1818 data: 0.0902 max mem: 3537 +Test: [1450/2387] eta: 0:02:53 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1833 (0.3191) acc1: 97.6562 (86.0565) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1802 data: 0.0883 max mem: 3537 +Test: [1460/2387] eta: 0:02:52 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1700 (0.3181) acc1: 97.6562 (86.1298) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1845 data: 0.0908 max mem: 3537 +Test: [1470/2387] eta: 0:02:50 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1523 (0.3170) acc1: 98.9583 (86.2176) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1956 data: 0.1016 max mem: 3537 +Test: [1480/2387] eta: 0:02:48 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1447 (0.3159) acc1: 98.9583 (86.2959) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1999 data: 0.1071 max mem: 3537 +Test: [1490/2387] eta: 0:02:46 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1437 (0.3147) acc1: 98.6979 (86.3780) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1937 data: 0.1000 max mem: 3537 +Test: [1500/2387] eta: 0:02:44 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1437 (0.3136) acc1: 98.9583 (86.4616) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1837 data: 0.0906 max mem: 3537 +Test: [1510/2387] eta: 0:02:43 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1457 (0.3125) acc1: 99.4792 (86.5488) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2260 data: 0.1347 max mem: 3537 +Test: [1520/2387] eta: 0:02:41 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1749 (0.3122) acc1: 98.6979 (86.5921) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.2258 data: 0.1342 max mem: 3537 +Test: [1530/2387] eta: 0:02:39 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2803 (0.3121) acc1: 91.1458 (86.6320) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.2012 data: 0.1092 max mem: 3537 +Test: [1540/2387] eta: 0:02:38 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2641 (0.3117) acc1: 94.7917 (86.6926) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2080 data: 0.1173 max mem: 3537 +Test: [1550/2387] eta: 0:02:36 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2507 (0.3113) acc1: 96.6146 (86.7520) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1937 data: 0.1019 max mem: 3537 +Test: [1560/2387] eta: 0:02:34 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2345 (0.3109) acc1: 96.6146 (86.8055) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.1956 data: 0.1026 max mem: 3537 +Test: [1570/2387] eta: 0:02:32 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2067 (0.3102) acc1: 98.1771 (86.8772) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1845 data: 0.0917 max mem: 3537 +Test: [1580/2387] eta: 0:02:30 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1706 (0.3092) acc1: 98.6979 (86.9485) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1867 data: 0.0940 max mem: 3538 +Test: [1590/2387] eta: 0:02:28 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1515 (0.3082) acc1: 98.4375 (87.0188) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1967 data: 0.1051 max mem: 3538 +Test: [1600/2387] eta: 0:02:26 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1329 (0.3073) acc1: 98.6979 (87.0686) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1829 data: 0.0911 max mem: 3538 +Test: [1610/2387] eta: 0:02:25 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1311 (0.3063) acc1: 99.2188 (87.1439) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1833 data: 0.0895 max mem: 3538 +Test: [1620/2387] eta: 0:02:23 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1341 (0.3053) acc1: 99.2188 (87.2128) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1745 data: 0.0816 max mem: 3538 +Test: [1630/2387] eta: 0:02:21 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1534 (0.3044) acc1: 98.1771 (87.2763) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1629 data: 0.0694 max mem: 3538 +Test: [1640/2387] eta: 0:02:19 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1659 (0.3036) acc1: 98.1771 (87.3400) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1776 data: 0.0848 max mem: 3538 +Test: [1650/2387] eta: 0:02:17 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1679 (0.3027) acc1: 98.1771 (87.4046) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1721 data: 0.0799 max mem: 3538 +Test: [1660/2387] eta: 0:02:15 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1639 (0.3020) acc1: 98.1771 (87.4646) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1796 data: 0.0861 max mem: 3538 +Test: [1670/2387] eta: 0:02:13 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1639 (0.3012) acc1: 98.1771 (87.5251) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1991 data: 0.1074 max mem: 3538 +Test: [1680/2387] eta: 0:02:11 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1691 (0.3006) acc1: 98.1771 (87.5768) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1774 data: 0.0857 max mem: 3538 +Test: [1690/2387] eta: 0:02:09 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1710 (0.2999) acc1: 98.1771 (87.6329) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1666 data: 0.0732 max mem: 3538 +Test: [1700/2387] eta: 0:02:08 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1647 (0.2990) acc1: 98.1771 (87.6955) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1906 data: 0.0977 max mem: 3538 +Test: [1710/2387] eta: 0:02:06 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1494 (0.2981) acc1: 98.6979 (87.7610) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1989 data: 0.1075 max mem: 3538 +Test: [1720/2387] eta: 0:02:04 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1297 (0.2971) acc1: 99.2188 (87.8293) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1717 data: 0.0801 max mem: 3538 +Test: [1730/2387] eta: 0:02:02 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1272 (0.2962) acc1: 99.7396 (87.8922) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2040 data: 0.1117 max mem: 3538 +Test: [1740/2387] eta: 0:02:00 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1225 (0.2952) acc1: 99.7396 (87.9594) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1993 data: 0.1057 max mem: 3538 +Test: [1750/2387] eta: 0:01:58 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1149 (0.2942) acc1: 100.0000 (88.0253) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1662 data: 0.0733 max mem: 3538 +Test: [1760/2387] eta: 0:01:56 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1037 (0.2931) acc1: 100.0000 (88.0921) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.1013 max mem: 3538 +Test: [1770/2387] eta: 0:01:55 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1238 (0.2924) acc1: 99.4792 (88.1454) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1936 data: 0.0982 max mem: 3538 +Test: [1780/2387] eta: 0:01:53 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1691 (0.2917) acc1: 97.9167 (88.2005) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1932 data: 0.0987 max mem: 3538 +Test: [1790/2387] eta: 0:01:51 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1612 (0.2909) acc1: 97.9167 (88.2552) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1957 data: 0.1004 max mem: 3538 +Test: [1800/2387] eta: 0:01:49 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1546 (0.2902) acc1: 98.1771 (88.3109) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1949 data: 0.1015 max mem: 3538 +Test: [1810/2387] eta: 0:01:47 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1419 (0.2893) acc1: 98.9583 (88.3724) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2048 data: 0.1135 max mem: 3539 +Test: [1820/2387] eta: 0:01:45 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1144 (0.2884) acc1: 99.4792 (88.4324) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1946 data: 0.1031 max mem: 3539 +Test: [1830/2387] eta: 0:01:43 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1216 (0.2875) acc1: 99.7396 (88.4947) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1658 data: 0.0731 max mem: 3539 +Test: [1840/2387] eta: 0:01:41 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1549 (0.2868) acc1: 99.7396 (88.5564) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1624 data: 0.0692 max mem: 3539 +Test: [1850/2387] eta: 0:01:40 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1578 (0.2861) acc1: 99.7396 (88.6171) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1636 data: 0.0713 max mem: 3539 +Test: [1860/2387] eta: 0:01:38 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1562 (0.2854) acc1: 100.0000 (88.6777) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1623 data: 0.0702 max mem: 3539 +Test: [1870/2387] eta: 0:01:36 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1563 (0.2847) acc1: 99.7396 (88.7358) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1705 data: 0.0761 max mem: 3539 +Test: [1880/2387] eta: 0:01:34 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1569 (0.2841) acc1: 99.7396 (88.7941) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1702 data: 0.0761 max mem: 3539 +Test: [1890/2387] eta: 0:01:32 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1566 (0.2834) acc1: 100.0000 (88.8530) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1617 data: 0.0706 max mem: 3539 +Test: [1900/2387] eta: 0:01:30 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1576 (0.2827) acc1: 100.0000 (88.9110) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1646 data: 0.0734 max mem: 3539 +Test: [1910/2387] eta: 0:01:28 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1585 (0.2821) acc1: 100.0000 (88.9662) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1732 data: 0.0820 max mem: 3539 +Test: [1920/2387] eta: 0:01:26 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1585 (0.2819) acc1: 99.4792 (88.9896) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1713 data: 0.0805 max mem: 3539 +Test: [1930/2387] eta: 0:01:24 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1645 (0.2812) acc1: 98.1771 (89.0384) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1849 data: 0.0943 max mem: 3539 +Test: [1940/2387] eta: 0:01:22 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1691 (0.2810) acc1: 98.1771 (89.0512) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1761 data: 0.0839 max mem: 3539 +Test: [1950/2387] eta: 0:01:21 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1956 (0.2806) acc1: 97.9167 (89.0900) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1657 data: 0.0717 max mem: 3539 +Test: [1960/2387] eta: 0:01:19 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1906 (0.2802) acc1: 98.1771 (89.1280) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1873 data: 0.0947 max mem: 3539 +Test: [1970/2387] eta: 0:01:17 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1981 (0.2802) acc1: 97.3958 (89.1295) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.1788 data: 0.0863 max mem: 3539 +Test: [1980/2387] eta: 0:01:15 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2208 (0.2799) acc1: 95.8333 (89.1612) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1699 data: 0.0739 max mem: 3539 +Test: [1990/2387] eta: 0:01:13 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2298 (0.2797) acc1: 96.3542 (89.1954) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1840 data: 0.0881 max mem: 3539 +Test: [2000/2387] eta: 0:01:11 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2323 (0.2795) acc1: 96.0938 (89.2301) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1837 data: 0.0885 max mem: 3539 +Test: [2010/2387] eta: 0:01:09 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2371 (0.2793) acc1: 95.8333 (89.2632) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1881 data: 0.0920 max mem: 3539 +Test: [2020/2387] eta: 0:01:08 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2379 (0.2790) acc1: 96.0938 (89.2979) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1861 data: 0.0918 max mem: 3539 +Test: [2030/2387] eta: 0:01:06 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2319 (0.2788) acc1: 96.0938 (89.3297) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1873 data: 0.0929 max mem: 3540 +Test: [2040/2387] eta: 0:01:04 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2294 (0.2786) acc1: 96.0938 (89.3640) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2214 data: 0.1270 max mem: 3540 +Test: [2050/2387] eta: 0:01:02 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2271 (0.2783) acc1: 96.6146 (89.4009) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2193 data: 0.1256 max mem: 3540 +Test: [2060/2387] eta: 0:01:00 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2181 (0.2780) acc1: 96.8750 (89.4376) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1862 data: 0.0930 max mem: 3540 +Test: [2070/2387] eta: 0:00:58 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2115 (0.2777) acc1: 97.1354 (89.4778) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1926 data: 0.1011 max mem: 3540 +Test: [2080/2387] eta: 0:00:57 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2065 (0.2773) acc1: 97.6562 (89.5165) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2043 data: 0.1128 max mem: 3540 +Test: [2090/2387] eta: 0:00:55 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2050 (0.2770) acc1: 97.3958 (89.5547) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2040 data: 0.1119 max mem: 3540 +Test: [2100/2387] eta: 0:00:53 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2055 (0.2766) acc1: 97.3958 (89.5935) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2002 data: 0.1081 max mem: 3540 +Test: [2110/2387] eta: 0:00:51 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2061 (0.2763) acc1: 97.3958 (89.6321) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1993 data: 0.1072 max mem: 3540 +Test: [2120/2387] eta: 0:00:49 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2026 (0.2759) acc1: 97.1354 (89.6635) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2024 data: 0.1097 max mem: 3540 +Test: [2130/2387] eta: 0:00:47 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1534 (0.2753) acc1: 98.1771 (89.7062) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2036 data: 0.1109 max mem: 3540 +Test: [2140/2387] eta: 0:00:45 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1465 (0.2749) acc1: 98.1771 (89.7309) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1908 data: 0.0984 max mem: 3540 +Test: [2150/2387] eta: 0:00:44 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1696 (0.2744) acc1: 97.6562 (89.7692) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1699 data: 0.0783 max mem: 3540 +Test: [2160/2387] eta: 0:00:42 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1618 (0.2738) acc1: 98.6979 (89.8109) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1644 data: 0.0725 max mem: 3540 +Test: [2170/2387] eta: 0:00:40 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1473 (0.2734) acc1: 99.2188 (89.8478) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1605 data: 0.0689 max mem: 3540 +Test: [2180/2387] eta: 0:00:38 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1946 (0.2742) acc1: 96.3542 (89.7883) acc5: 100.0000 (100.0000) acc1_neg: 51.042 acc5_neg: 100.000 time: 0.1606 data: 0.0676 max mem: 3540 +Test: [2190/2387] eta: 0:00:36 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.4918 (0.2754) acc1: 74.2188 (89.7098) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0800 max mem: 3540 +Test: [2200/2387] eta: 0:00:34 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1781 (0.2748) acc1: 94.5312 (89.7482) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1737 data: 0.0792 max mem: 3540 +Test: [2210/2387] eta: 0:00:32 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1505 (0.2743) acc1: 98.6979 (89.7871) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1393 data: 0.0439 max mem: 3540 +Test: [2220/2387] eta: 0:00:30 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1515 (0.2738) acc1: 98.9583 (89.8269) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1399 data: 0.0453 max mem: 3540 +Test: [2230/2387] eta: 0:00:29 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1512 (0.2732) acc1: 98.9583 (89.8642) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0748 max mem: 3540 +Test: [2240/2387] eta: 0:00:27 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1471 (0.2727) acc1: 98.4375 (89.8993) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1699 data: 0.0739 max mem: 3540 +Test: [2250/2387] eta: 0:00:25 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1485 (0.2722) acc1: 98.6979 (89.9373) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1699 data: 0.0747 max mem: 3540 +Test: [2260/2387] eta: 0:00:23 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1457 (0.2717) acc1: 98.9583 (89.9771) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1650 data: 0.0699 max mem: 3541 +Test: [2270/2387] eta: 0:00:21 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1325 (0.2710) acc1: 99.7396 (90.0195) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1849 data: 0.0894 max mem: 3541 +Test: [2280/2387] eta: 0:00:19 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1149 (0.2705) acc1: 99.7396 (90.0465) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2105 data: 0.1152 max mem: 3541 +Test: [2290/2387] eta: 0:00:17 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1139 (0.2698) acc1: 99.7396 (90.0891) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1699 data: 0.0754 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1194 (0.2692) acc1: 100.0000 (90.1283) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1246 data: 0.0306 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1196 (0.2686) acc1: 99.7396 (90.1647) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1313 data: 0.0389 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.1409 (0.2683) acc1: 98.4375 (90.1862) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1552 data: 0.0639 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2216 (0.2684) acc1: 95.3125 (90.1932) acc5: 100.0000 (100.0000) acc1_neg: 78.646 acc5_neg: 100.000 time: 0.1551 data: 0.0627 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2201 (0.2681) acc1: 94.0104 (90.2143) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1397 data: 0.0468 max mem: 3541 +Test: [2350/2387] eta: 0:00:06 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2383 (0.2683) acc1: 94.0104 (90.2060) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1233 data: 0.0286 max mem: 3541 +Test: [2360/2387] eta: 0:00:04 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2552 (0.2683) acc1: 90.3646 (90.2129) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1344 data: 0.0384 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2511 (0.2683) acc1: 94.0104 (90.2211) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.1480 data: 0.0529 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2476 (0.2682) acc1: 94.7917 (90.2380) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1336 data: 0.0410 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 30.594 acc5_pos: 100.000 loss: 0.2799 (0.2698) acc1: 92.1875 (90.1563) acc5: 100.0000 (100.0000) acc1_neg: 11.628 acc5_neg: 100.000 time: 0.1029 data: 0.0154 max mem: 3541 +Test: Total time: 0:07:17 (0.1831 s / it) +* Overall Acc@1 90.156 Acc@5 100.000 + target == real Acc@1 29.030 Acc@5 100.000 + target == fake Acc@1 94.449 Acc@5 100.000 + loss 0.270 + PR-AUC 0.9890 ROC-AUC 0.8602 EER 21.80% (thr=0.7729) +Accuracy of the network on the 916267 test images: 90.2% +[rank0]:[W726 04:14:08.033650321 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:14:27.619000 3151932 torch/distributed/run.py:766] +W0726 04:14:27.619000 3151932 torch/distributed/run.py:766] ***************************************** +W0726 04:14:27.619000 3151932 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:14:27.619000 3151932 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:14:56.115487762 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:14:56.239557640 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:14:57.568535354 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:14:57.816792809 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:14:57.835755320 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:14:57.878311028 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:14:57.912672130 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:14:57.010849055 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n32_seed6_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed6_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:01:48 acc1_pos: 46.875 acc5_pos: 100.000 loss: 0.8150 (0.8150) acc1: 46.8750 (46.8750) acc5: 100.0000 (100.0000) time: 7.1731 data: 5.5461 max mem: 3531 +Test: [ 10/517] eta: 0:06:11 acc1_pos: 18.490 acc5_pos: 100.000 loss: 0.9472 (0.9562) acc1: 35.4167 (34.1619) acc5: 100.0000 (100.0000) time: 0.7336 data: 0.5045 max mem: 3531 +Test: [ 20/517] eta: 0:03:33 acc1_pos: 36.719 acc5_pos: 100.000 loss: 0.9435 (0.9377) acc1: 35.4167 (36.9048) acc5: 100.0000 (100.0000) time: 0.0916 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:02:50 acc1_pos: 20.312 acc5_pos: 100.000 loss: 0.9704 (0.9646) acc1: 27.8646 (34.5430) acc5: 100.0000 (100.0000) time: 0.1387 data: 0.0461 max mem: 3531 +Test: [ 40/517] eta: 0:02:38 acc1_pos: 29.167 acc5_pos: 100.000 loss: 1.0681 (0.9998) acc1: 25.7812 (31.6438) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1372 max mem: 3531 +Test: [ 50/517] eta: 0:02:25 acc1_pos: 36.239 acc5_pos: 100.000 loss: 1.0413 (0.9643) acc1: 28.6458 (34.6252) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2505 data: 0.1558 max mem: 3531 +Test: [ 60/517] eta: 0:02:17 acc1_pos: 36.239 acc5_pos: 100.000 loss: 0.4240 (0.8677) acc1: 82.8125 (43.5963) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.2399 data: 0.1449 max mem: 3531 +Test: [ 70/517] eta: 0:02:11 acc1_pos: 36.239 acc5_pos: 100.000 loss: 0.3497 (0.7953) acc1: 91.1458 (50.3815) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.2483 data: 0.1553 max mem: 3531 +Test: [ 80/517] eta: 0:02:05 acc1_pos: 36.239 acc5_pos: 100.000 loss: 0.3515 (0.7441) acc1: 91.6667 (55.2823) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.2399 data: 0.1445 max mem: 3531 +Test: [ 90/517] eta: 0:01:59 acc1_pos: 36.239 acc5_pos: 100.000 loss: 0.3145 (0.6952) acc1: 95.0521 (59.8186) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2352 data: 0.1398 max mem: 3531 +Test: [100/517] eta: 0:01:55 acc1_pos: 36.239 acc5_pos: 100.000 loss: 0.2804 (0.6527) acc1: 97.3958 (63.6190) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2402 data: 0.1465 max mem: 3531 +Test: [110/517] eta: 0:01:51 acc1_pos: 36.239 acc5_pos: 100.000 loss: 0.2555 (0.6166) acc1: 98.6979 (66.7488) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2439 data: 0.1498 max mem: 3531 +Test: [120/517] eta: 0:01:47 acc1_pos: 36.239 acc5_pos: 100.000 loss: 0.2856 (0.5924) acc1: 96.8750 (69.0061) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2348 data: 0.1421 max mem: 3531 +Test: [130/517] eta: 0:01:43 acc1_pos: 36.239 acc5_pos: 100.000 loss: 0.3233 (0.5708) acc1: 94.2708 (70.9168) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2395 data: 0.1445 max mem: 3531 +Test: [140/517] eta: 0:01:40 acc1_pos: 36.239 acc5_pos: 100.000 loss: 0.2929 (0.5509) acc1: 97.6562 (72.8890) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2462 data: 0.1516 max mem: 3531 +Test: [150/517] eta: 0:01:37 acc1_pos: 36.239 acc5_pos: 100.000 loss: 0.2861 (0.5340) acc1: 98.4375 (74.4102) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.2416 data: 0.1494 max mem: 3531 +Test: [160/517] eta: 0:01:34 acc1_pos: 36.239 acc5_pos: 100.000 loss: 0.2855 (0.5201) acc1: 94.0104 (75.5613) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2533 data: 0.1596 max mem: 3531 +Test: [170/517] eta: 0:01:31 acc1_pos: 36.239 acc5_pos: 100.000 loss: 0.2855 (0.5063) acc1: 93.7500 (76.6828) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2532 data: 0.1598 max mem: 3531 +Test: [180/517] eta: 0:01:28 acc1_pos: 36.239 acc5_pos: 100.000 loss: 0.2739 (0.4927) acc1: 95.8333 (77.7783) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2399 data: 0.1453 max mem: 3531 +Test: [190/517] eta: 0:01:25 acc1_pos: 36.239 acc5_pos: 100.000 loss: 0.2176 (0.4771) acc1: 98.9583 (78.9281) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2399 data: 0.1454 max mem: 3531 +Test: [200/517] eta: 0:01:21 acc1_pos: 36.239 acc5_pos: 100.000 loss: 0.2176 (0.4658) acc1: 99.4792 (79.9103) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2249 data: 0.1311 max mem: 3531 +Test: [210/517] eta: 0:01:19 acc1_pos: 36.239 acc5_pos: 100.000 loss: 0.2951 (0.4602) acc1: 97.3958 (80.5663) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2311 data: 0.1364 max mem: 3532 +Test: [220/517] eta: 0:01:16 acc1_pos: 36.239 acc5_pos: 100.000 loss: 0.3362 (0.4545) acc1: 94.2708 (81.1852) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2568 data: 0.1630 max mem: 3532 +Test: [230/517] eta: 0:01:13 acc1_pos: 36.239 acc5_pos: 100.000 loss: 0.3486 (0.4511) acc1: 94.0104 (81.6987) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2437 data: 0.1487 max mem: 3532 +Test: [240/517] eta: 0:01:11 acc1_pos: 36.239 acc5_pos: 100.000 loss: 0.3520 (0.4459) acc1: 94.7917 (82.2863) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.2380 data: 0.1419 max mem: 3532 +Test: [250/517] eta: 0:01:08 acc1_pos: 36.239 acc5_pos: 100.000 loss: 0.3058 (0.4399) acc1: 96.6146 (82.8914) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2399 data: 0.1464 max mem: 3532 +Test: [260/517] eta: 0:01:05 acc1_pos: 36.239 acc5_pos: 100.000 loss: 0.2802 (0.4341) acc1: 98.6979 (83.4740) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2399 data: 0.1452 max mem: 3532 +Test: [270/517] eta: 0:01:03 acc1_pos: 36.239 acc5_pos: 100.000 loss: 0.3157 (0.4313) acc1: 97.6562 (83.8715) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2549 data: 0.1557 max mem: 3532 +Test: [280/517] eta: 0:01:00 acc1_pos: 36.239 acc5_pos: 100.000 loss: 0.3403 (0.4275) acc1: 95.0521 (84.2749) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2499 data: 0.1523 max mem: 3532 +Test: [290/517] eta: 0:00:57 acc1_pos: 36.239 acc5_pos: 100.000 loss: 0.3263 (0.4240) acc1: 94.5312 (84.6372) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2499 data: 0.1523 max mem: 3532 +Test: [300/517] eta: 0:00:55 acc1_pos: 36.239 acc5_pos: 100.000 loss: 0.3207 (0.4203) acc1: 96.0938 (85.0472) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2314 data: 0.1367 max mem: 3532 +Test: [310/517] eta: 0:00:52 acc1_pos: 36.239 acc5_pos: 100.000 loss: 0.3267 (0.4182) acc1: 95.3125 (85.3179) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.2199 data: 0.1282 max mem: 3532 +Test: [320/517] eta: 0:00:49 acc1_pos: 36.239 acc5_pos: 100.000 loss: 0.3267 (0.4140) acc1: 94.7917 (85.6860) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2386 data: 0.1448 max mem: 3532 +Test: [330/517] eta: 0:00:47 acc1_pos: 36.239 acc5_pos: 100.000 loss: 0.2452 (0.4085) acc1: 98.9583 (86.0949) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2499 data: 0.1563 max mem: 3532 +Test: [340/517] eta: 0:00:44 acc1_pos: 36.239 acc5_pos: 100.000 loss: 0.2290 (0.4045) acc1: 99.7396 (86.3789) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.2502 data: 0.1583 max mem: 3532 +Test: [350/517] eta: 0:00:42 acc1_pos: 36.239 acc5_pos: 100.000 loss: 0.3245 (0.4032) acc1: 94.7917 (86.5629) acc5: 100.0000 (100.0000) acc1_neg: 77.344 acc5_neg: 100.000 time: 0.2401 data: 0.1474 max mem: 3532 +Test: [360/517] eta: 0:00:39 acc1_pos: 36.239 acc5_pos: 100.000 loss: 0.3287 (0.4013) acc1: 94.7917 (86.7469) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2395 data: 0.1446 max mem: 3532 +Test: [370/517] eta: 0:00:36 acc1_pos: 36.239 acc5_pos: 100.000 loss: 0.2686 (0.3969) acc1: 98.6979 (87.0964) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2403 data: 0.1480 max mem: 3532 +Test: [380/517] eta: 0:00:34 acc1_pos: 36.239 acc5_pos: 100.000 loss: 0.2349 (0.3929) acc1: 99.4792 (87.3954) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2414 data: 0.1516 max mem: 3532 +Test: [390/517] eta: 0:00:31 acc1_pos: 36.239 acc5_pos: 100.000 loss: 0.2407 (0.3898) acc1: 98.1771 (87.6219) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2360 data: 0.1439 max mem: 3532 +Test: [400/517] eta: 0:00:29 acc1_pos: 36.239 acc5_pos: 100.000 loss: 0.2646 (0.3868) acc1: 97.1354 (87.8845) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2384 data: 0.1444 max mem: 3532 +Test: [410/517] eta: 0:00:26 acc1_pos: 36.239 acc5_pos: 100.000 loss: 0.3068 (0.3855) acc1: 97.1354 (88.0665) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.2432 data: 0.1474 max mem: 3532 +Test: [420/517] eta: 0:00:24 acc1_pos: 36.239 acc5_pos: 100.000 loss: 0.2781 (0.3824) acc1: 97.1354 (88.3208) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2299 data: 0.1318 max mem: 3532 +Test: [430/517] eta: 0:00:21 acc1_pos: 36.239 acc5_pos: 100.000 loss: 0.2649 (0.3802) acc1: 98.1771 (88.5266) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2264 data: 0.1311 max mem: 3532 +Test: [440/517] eta: 0:00:19 acc1_pos: 36.239 acc5_pos: 100.000 loss: 0.2813 (0.3777) acc1: 97.6562 (88.7230) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2410 data: 0.1493 max mem: 3533 +Test: [450/517] eta: 0:00:16 acc1_pos: 36.239 acc5_pos: 100.000 loss: 0.2773 (0.3755) acc1: 97.9167 (88.9187) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2484 data: 0.1539 max mem: 3533 +Test: [460/517] eta: 0:00:14 acc1_pos: 36.239 acc5_pos: 100.000 loss: 0.3000 (0.3745) acc1: 96.6146 (89.0365) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2440 data: 0.1470 max mem: 3533 +Test: [470/517] eta: 0:00:11 acc1_pos: 36.239 acc5_pos: 100.000 loss: 0.2968 (0.3720) acc1: 96.6146 (89.2140) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2469 data: 0.1528 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 36.239 acc5_pos: 100.000 loss: 0.2606 (0.3698) acc1: 98.6979 (89.3993) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2414 data: 0.1505 max mem: 3533 +Test: [490/517] eta: 0:00:06 acc1_pos: 36.239 acc5_pos: 100.000 loss: 0.2734 (0.3679) acc1: 98.6979 (89.5579) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2380 data: 0.1442 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 36.239 acc5_pos: 100.000 loss: 0.3292 (0.3685) acc1: 92.7083 (89.5797) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.2432 data: 0.1458 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 36.239 acc5_pos: 100.000 loss: 0.3734 (0.3670) acc1: 92.4479 (89.7041) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2235 data: 0.1298 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 36.239 acc5_pos: 100.000 loss: 0.2454 (0.3653) acc1: 97.3958 (89.8043) acc5: 100.0000 (100.0000) acc1_neg: 99.293 acc5_neg: 100.000 time: 0.1891 data: 0.0955 max mem: 3533 +Test: Total time: 0:02:08 (0.2477 s / it) +* Overall Acc@1 89.804 Acc@5 100.000 + target == real Acc@1 31.759 Acc@5 100.000 + target == fake Acc@1 95.826 Acc@5 100.000 + loss 0.365 + PR-AUC 0.9630 ROC-AUC 0.7734 EER 29.64% (thr=0.7188) +Accuracy of the network on the 198427 test images: 89.8% +[rank0]:[W726 04:17:20.605094570 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 04:17:40.083000 3152504 torch/distributed/run.py:766] +W0726 04:17:40.083000 3152504 torch/distributed/run.py:766] ***************************************** +W0726 04:17:40.083000 3152504 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:17:40.083000 3152504 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 04:18:08.547842183 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:18:09.193100907 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W726 04:18:09.193123957 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +[rank4]:[W726 04:18:09.230396786 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:18:09.238621243 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:18:09.285708084 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:18:09.471205956 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank1]:[W726 04:18:09.471206821 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n32_seed6_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed6_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:33:38 acc1_pos: 65.365 acc5_pos: 100.000 loss: 0.5973 (0.5973) acc1: 65.3646 (65.3646) acc5: 100.0000 (100.0000) time: 5.1890 data: 3.4687 max mem: 3531 +Test: [ 10/389] eta: 0:03:29 acc1_pos: 99.479 acc5_pos: 100.000 loss: 0.3505 (0.4386) acc1: 90.1042 (80.6345) acc5: 100.0000 (100.0000) time: 0.5535 data: 0.3158 max mem: 3531 +Test: [ 20/389] eta: 0:02:07 acc1_pos: 33.333 acc5_pos: 100.000 loss: 0.3799 (0.5093) acc1: 84.8958 (71.0565) acc5: 100.0000 (100.0000) time: 0.1032 data: 0.0101 max mem: 3531 +Test: [ 30/389] eta: 0:01:47 acc1_pos: 78.906 acc5_pos: 100.000 loss: 0.5099 (0.5329) acc1: 78.9062 (72.0346) acc5: 100.0000 (100.0000) time: 0.1605 data: 0.0652 max mem: 3531 +Test: [ 40/389] eta: 0:01:44 acc1_pos: 57.552 acc5_pos: 100.000 loss: 0.5225 (0.5441) acc1: 77.0833 (71.6590) acc5: 100.0000 (100.0000) time: 0.2521 data: 0.1589 max mem: 3531 +Test: [ 50/389] eta: 0:01:34 acc1_pos: 88.802 acc5_pos: 100.000 loss: 0.4539 (0.5310) acc1: 85.9375 (74.0809) acc5: 100.0000 (100.0000) time: 0.2476 data: 0.1543 max mem: 3531 +Test: [ 60/389] eta: 0:01:27 acc1_pos: 79.427 acc5_pos: 100.000 loss: 0.4346 (0.5275) acc1: 87.7604 (74.9616) acc5: 100.0000 (100.0000) time: 0.1978 data: 0.1035 max mem: 3531 +Test: [ 70/389] eta: 0:01:22 acc1_pos: 89.323 acc5_pos: 100.000 loss: 0.4415 (0.5167) acc1: 82.8125 (76.2214) acc5: 100.0000 (100.0000) time: 0.2099 data: 0.1146 max mem: 3531 +Test: [ 80/389] eta: 0:01:18 acc1_pos: 91.406 acc5_pos: 100.000 loss: 0.4600 (0.5113) acc1: 82.8125 (76.9901) acc5: 100.0000 (100.0000) time: 0.2167 data: 0.1223 max mem: 3531 +Test: [ 90/389] eta: 0:01:15 acc1_pos: 88.281 acc5_pos: 100.000 loss: 0.3247 (0.4932) acc1: 88.2812 (78.7002) acc5: 100.0000 (100.0000) time: 0.2231 data: 0.1306 max mem: 3531 +Test: [100/389] eta: 0:01:09 acc1_pos: 91.146 acc5_pos: 100.000 loss: 0.4319 (0.5012) acc1: 87.5000 (78.0322) acc5: 100.0000 (100.0000) time: 0.1856 data: 0.0939 max mem: 3531 +Test: [110/389] eta: 0:01:05 acc1_pos: 95.052 acc5_pos: 100.000 loss: 0.5137 (0.5147) acc1: 80.2083 (76.7314) acc5: 100.0000 (100.0000) time: 0.1530 data: 0.0623 max mem: 3531 +Test: [120/389] eta: 0:01:01 acc1_pos: 87.500 acc5_pos: 100.000 loss: 0.4470 (0.5110) acc1: 80.7292 (77.0209) acc5: 100.0000 (100.0000) time: 0.1640 data: 0.0718 max mem: 3531 +Test: [130/389] eta: 0:00:57 acc1_pos: 47.239 acc5_pos: 100.000 loss: 0.5564 (0.5237) acc1: 71.8750 (75.7455) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.1539 data: 0.0598 max mem: 3531 +Test: [140/389] eta: 0:00:53 acc1_pos: 47.239 acc5_pos: 100.000 loss: 0.5834 (0.5282) acc1: 64.3229 (75.4322) acc5: 100.0000 (100.0000) acc1_neg: 24.740 acc5_neg: 100.000 time: 0.1384 data: 0.0451 max mem: 3531 +Test: [150/389] eta: 0:00:50 acc1_pos: 47.239 acc5_pos: 100.000 loss: 0.5339 (0.5510) acc1: 69.0104 (73.6117) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1401 data: 0.0469 max mem: 3531 +Test: [160/389] eta: 0:00:47 acc1_pos: 47.239 acc5_pos: 100.000 loss: 0.6129 (0.5559) acc1: 57.0312 (73.4262) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1646 data: 0.0723 max mem: 3531 +Test: [170/389] eta: 0:00:45 acc1_pos: 47.239 acc5_pos: 100.000 loss: 0.3769 (0.5528) acc1: 95.0521 (73.7802) acc5: 100.0000 (100.0000) acc1_neg: 42.708 acc5_neg: 100.000 time: 0.1694 data: 0.0760 max mem: 3531 +Test: [180/389] eta: 0:00:43 acc1_pos: 47.239 acc5_pos: 100.000 loss: 0.4812 (0.5620) acc1: 80.7292 (72.8634) acc5: 100.0000 (100.0000) acc1_neg: 57.292 acc5_neg: 100.000 time: 0.1900 data: 0.0967 max mem: 3531 +Test: [190/389] eta: 0:00:40 acc1_pos: 47.239 acc5_pos: 100.000 loss: 0.6833 (0.5741) acc1: 57.2917 (71.5737) acc5: 100.0000 (100.0000) acc1_neg: 17.188 acc5_neg: 100.000 time: 0.2049 data: 0.1134 max mem: 3531 +Test: [200/389] eta: 0:00:38 acc1_pos: 47.239 acc5_pos: 100.000 loss: 0.6833 (0.5806) acc1: 52.6042 (71.0238) acc5: 100.0000 (100.0000) acc1_neg: 52.604 acc5_neg: 100.000 time: 0.1723 data: 0.0800 max mem: 3531 +Test: [210/389] eta: 0:00:36 acc1_pos: 47.239 acc5_pos: 100.000 loss: 0.4899 (0.5758) acc1: 75.7812 (71.4504) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1870 data: 0.0950 max mem: 3532 +Test: [220/389] eta: 0:00:34 acc1_pos: 47.239 acc5_pos: 100.000 loss: 0.3430 (0.5673) acc1: 95.0521 (72.3888) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1941 data: 0.1027 max mem: 3532 +Test: [230/389] eta: 0:00:32 acc1_pos: 47.239 acc5_pos: 100.000 loss: 0.4961 (0.5759) acc1: 76.0417 (71.3339) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.1844 data: 0.0925 max mem: 3532 +Test: [240/389] eta: 0:00:30 acc1_pos: 47.239 acc5_pos: 100.000 loss: 0.6703 (0.5827) acc1: 47.1354 (70.6831) acc5: 100.0000 (100.0000) acc1_neg: 47.135 acc5_neg: 100.000 time: 0.1981 data: 0.1052 max mem: 3532 +Test: [250/389] eta: 0:00:28 acc1_pos: 47.239 acc5_pos: 100.000 loss: 0.5633 (0.5761) acc1: 71.8750 (71.0917) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1956 data: 0.1024 max mem: 3532 +Test: [260/389] eta: 0:00:26 acc1_pos: 47.239 acc5_pos: 100.000 loss: 0.4474 (0.5767) acc1: 78.6458 (71.0229) acc5: 100.0000 (100.0000) acc1_neg: 18.490 acc5_neg: 100.000 time: 0.1967 data: 0.1047 max mem: 3532 +Test: [270/389] eta: 0:00:24 acc1_pos: 47.239 acc5_pos: 100.000 loss: 0.6955 (0.5893) acc1: 54.4271 (69.6562) acc5: 100.0000 (100.0000) acc1_neg: 46.615 acc5_neg: 100.000 time: 0.2026 data: 0.1100 max mem: 3532 +Test: [280/389] eta: 0:00:22 acc1_pos: 47.239 acc5_pos: 100.000 loss: 0.8059 (0.5969) acc1: 42.1875 (68.6925) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2041 data: 0.1091 max mem: 3532 +Test: [290/389] eta: 0:00:20 acc1_pos: 47.239 acc5_pos: 100.000 loss: 0.7638 (0.5981) acc1: 43.2292 (68.3026) acc5: 100.0000 (100.0000) acc1_neg: 41.927 acc5_neg: 100.000 time: 0.2048 data: 0.1098 max mem: 3532 +Test: [300/389] eta: 0:00:17 acc1_pos: 47.239 acc5_pos: 100.000 loss: 0.6048 (0.5998) acc1: 66.9271 (68.1513) acc5: 100.0000 (100.0000) acc1_neg: 73.698 acc5_neg: 100.000 time: 0.1857 data: 0.0934 max mem: 3532 +Test: [310/389] eta: 0:00:15 acc1_pos: 47.239 acc5_pos: 100.000 loss: 0.7037 (0.6086) acc1: 47.1354 (67.3659) acc5: 100.0000 (100.0000) acc1_neg: 10.938 acc5_neg: 100.000 time: 0.1603 data: 0.0695 max mem: 3532 +Test: [320/389] eta: 0:00:13 acc1_pos: 47.239 acc5_pos: 100.000 loss: 0.6448 (0.6095) acc1: 59.1146 (67.3246) acc5: 100.0000 (100.0000) acc1_neg: 80.729 acc5_neg: 100.000 time: 0.1599 data: 0.0688 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 47.239 acc5_pos: 100.000 loss: 0.8917 (0.6198) acc1: 17.4479 (65.9192) acc5: 100.0000 (100.0000) acc1_neg: 50.521 acc5_neg: 100.000 time: 0.1644 data: 0.0719 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 47.239 acc5_pos: 100.000 loss: 0.5547 (0.6137) acc1: 66.9271 (66.7285) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1566 data: 0.0634 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 47.239 acc5_pos: 100.000 loss: 0.4439 (0.6251) acc1: 91.4062 (66.0605) acc5: 100.0000 (100.0000) acc1_neg: 1.042 acc5_neg: 100.000 time: 0.1470 data: 0.0553 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 47.239 acc5_pos: 100.000 loss: 1.0109 (0.6380) acc1: 24.2188 (64.8387) acc5: 100.0000 (100.0000) acc1_neg: 28.906 acc5_neg: 100.000 time: 0.1379 data: 0.0475 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 47.239 acc5_pos: 100.000 loss: 0.9502 (0.6379) acc1: 28.9062 (64.6760) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1255 data: 0.0353 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 47.239 acc5_pos: 100.000 loss: 0.6081 (0.6398) acc1: 60.6771 (64.5457) acc5: 100.0000 (100.0000) acc1_neg: 65.885 acc5_neg: 100.000 time: 0.1189 data: 0.0291 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 47.239 acc5_pos: 100.000 loss: 0.5250 (0.6382) acc1: 69.9454 (64.6835) acc5: 100.0000 (100.0000) acc1_neg: 69.945 acc5_neg: 100.000 time: 0.1129 data: 0.0201 max mem: 3532 +Test: Total time: 0:01:12 (0.1876 s / it) +* Overall Acc@1 64.684 Acc@5 100.000 + target == real Acc@1 76.774 Acc@5 100.000 + target == fake Acc@1 59.064 Acc@5 100.000 + loss 0.638 + PR-AUC 0.8546 ROC-AUC 0.7394 EER 32.44% (thr=0.4385) +Accuracy of the network on the 149358 test images: 64.7% +[rank0]:[W726 04:19:32.623372039 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:19:50.515000 3153062 torch/distributed/run.py:766] +W0726 04:19:50.515000 3153062 torch/distributed/run.py:766] ***************************************** +W0726 04:19:50.515000 3153062 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:19:50.515000 3153062 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:20:22.987988916 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:20:22.284530589 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:20:22.407553677 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:20:22.510374895 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:20:23.604791972 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:20:23.740206372 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:20:23.777919014 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:20:23.804621318 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n32_seed6_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed6_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:31:54 acc1_pos: 84.375 acc5_pos: 100.000 loss: 0.3666 (0.3666) acc1: 84.3750 (84.3750) acc5: 100.0000 (100.0000) time: 5.7479 data: 4.0571 max mem: 3531 +Test: [ 10/333] eta: 0:03:15 acc1_pos: 89.323 acc5_pos: 100.000 loss: 0.2796 (0.2795) acc1: 89.3229 (89.2756) acc5: 100.0000 (100.0000) time: 0.6048 data: 0.3691 max mem: 3531 +Test: [ 20/333] eta: 0:01:53 acc1_pos: 96.615 acc5_pos: 100.000 loss: 0.2522 (0.2645) acc1: 91.4062 (90.5010) acc5: 100.0000 (100.0000) time: 0.0918 data: 0.0005 max mem: 3531 +Test: [ 30/333] eta: 0:01:29 acc1_pos: 64.844 acc5_pos: 100.000 loss: 0.3402 (0.4007) acc1: 77.8646 (80.1831) acc5: 100.0000 (100.0000) time: 0.1246 data: 0.0306 max mem: 3531 +Test: [ 40/333] eta: 0:01:20 acc1_pos: 51.562 acc5_pos: 100.000 loss: 0.7196 (0.4840) acc1: 55.7292 (73.8821) acc5: 100.0000 (100.0000) time: 0.1867 data: 0.0937 max mem: 3531 +Test: [ 50/333] eta: 0:01:13 acc1_pos: 59.375 acc5_pos: 100.000 loss: 0.7100 (0.5301) acc1: 54.9479 (70.3585) acc5: 100.0000 (100.0000) time: 0.2082 data: 0.1151 max mem: 3531 +Test: [ 60/333] eta: 0:01:07 acc1_pos: 53.385 acc5_pos: 100.000 loss: 0.6952 (0.5626) acc1: 55.9896 (67.8834) acc5: 100.0000 (100.0000) time: 0.1844 data: 0.0912 max mem: 3531 +Test: [ 70/333] eta: 0:01:02 acc1_pos: 49.479 acc5_pos: 100.000 loss: 0.7237 (0.5878) acc1: 54.4271 (66.1678) acc5: 100.0000 (100.0000) time: 0.1736 data: 0.0796 max mem: 3531 +Test: [ 80/333] eta: 0:00:57 acc1_pos: 57.552 acc5_pos: 100.000 loss: 0.7237 (0.6038) acc1: 55.4688 (64.7987) acc5: 100.0000 (100.0000) time: 0.1696 data: 0.0735 max mem: 3531 +Test: [ 90/333] eta: 0:00:53 acc1_pos: 67.188 acc5_pos: 100.000 loss: 0.6402 (0.6064) acc1: 60.4167 (64.8695) acc5: 100.0000 (100.0000) time: 0.1649 data: 0.0717 max mem: 3531 +Test: [100/333] eta: 0:00:51 acc1_pos: 71.094 acc5_pos: 100.000 loss: 0.6338 (0.6120) acc1: 60.9375 (64.4596) acc5: 100.0000 (100.0000) time: 0.1862 data: 0.0911 max mem: 3531 +Test: [110/333] eta: 0:00:48 acc1_pos: 64.844 acc5_pos: 100.000 loss: 0.6568 (0.6157) acc1: 59.8958 (64.1024) acc5: 100.0000 (100.0000) time: 0.1999 data: 0.0986 max mem: 3531 +Test: [120/333] eta: 0:00:45 acc1_pos: 67.448 acc5_pos: 100.000 loss: 0.6038 (0.6177) acc1: 63.0208 (63.9915) acc5: 100.0000 (100.0000) time: 0.1827 data: 0.0849 max mem: 3531 +Test: [130/333] eta: 0:00:43 acc1_pos: 76.823 acc5_pos: 100.000 loss: 0.5917 (0.6178) acc1: 66.9271 (64.3050) acc5: 100.0000 (100.0000) time: 0.1949 data: 0.1035 max mem: 3531 +Test: [140/333] eta: 0:00:40 acc1_pos: 77.083 acc5_pos: 100.000 loss: 0.6180 (0.6181) acc1: 66.9271 (64.3783) acc5: 100.0000 (100.0000) time: 0.1907 data: 0.0977 max mem: 3531 +Test: [150/333] eta: 0:00:37 acc1_pos: 71.094 acc5_pos: 100.000 loss: 0.6263 (0.6203) acc1: 64.3229 (64.2539) acc5: 100.0000 (100.0000) time: 0.1458 data: 0.0529 max mem: 3531 +Test: [160/333] eta: 0:00:35 acc1_pos: 57.237 acc5_pos: 100.000 loss: 0.6646 (0.6298) acc1: 59.8958 (63.6597) acc5: 100.0000 (100.0000) acc1_neg: 45.052 acc5_neg: 100.000 time: 0.1891 data: 0.0976 max mem: 3531 +Test: [170/333] eta: 0:00:33 acc1_pos: 57.237 acc5_pos: 100.000 loss: 0.7864 (0.6416) acc1: 52.8646 (62.8670) acc5: 100.0000 (100.0000) acc1_neg: 54.427 acc5_neg: 100.000 time: 0.2054 data: 0.1131 max mem: 3531 +Test: [180/333] eta: 0:00:30 acc1_pos: 57.237 acc5_pos: 100.000 loss: 0.7478 (0.6437) acc1: 54.9479 (62.7921) acc5: 100.0000 (100.0000) acc1_neg: 54.688 acc5_neg: 100.000 time: 0.1649 data: 0.0728 max mem: 3531 +Test: [190/333] eta: 0:00:28 acc1_pos: 57.237 acc5_pos: 100.000 loss: 0.6524 (0.6429) acc1: 63.5417 (62.9977) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.1647 data: 0.0715 max mem: 3531 +Test: [200/333] eta: 0:00:26 acc1_pos: 57.237 acc5_pos: 100.000 loss: 0.6429 (0.6445) acc1: 63.2812 (63.0182) acc5: 100.0000 (100.0000) acc1_neg: 57.812 acc5_neg: 100.000 time: 0.1753 data: 0.0809 max mem: 3531 +Test: [210/333] eta: 0:00:24 acc1_pos: 57.237 acc5_pos: 100.000 loss: 0.7251 (0.6497) acc1: 58.0729 (62.7394) acc5: 100.0000 (100.0000) acc1_neg: 53.385 acc5_neg: 100.000 time: 0.1802 data: 0.0853 max mem: 3532 +Test: [220/333] eta: 0:00:22 acc1_pos: 57.237 acc5_pos: 100.000 loss: 0.7365 (0.6551) acc1: 56.7708 (62.4375) acc5: 100.0000 (100.0000) acc1_neg: 67.708 acc5_neg: 100.000 time: 0.1755 data: 0.0786 max mem: 3532 +Test: [230/333] eta: 0:00:20 acc1_pos: 57.237 acc5_pos: 100.000 loss: 0.5821 (0.6500) acc1: 67.7083 (62.9645) acc5: 100.0000 (100.0000) acc1_neg: 71.094 acc5_neg: 100.000 time: 0.1798 data: 0.0849 max mem: 3532 +Test: [240/333] eta: 0:00:18 acc1_pos: 57.237 acc5_pos: 100.000 loss: 0.5604 (0.6505) acc1: 71.0938 (62.9052) acc5: 100.0000 (100.0000) acc1_neg: 60.417 acc5_neg: 100.000 time: 0.1855 data: 0.0923 max mem: 3532 +Test: [250/333] eta: 0:00:16 acc1_pos: 57.237 acc5_pos: 100.000 loss: 0.6702 (0.6508) acc1: 61.4583 (62.9036) acc5: 100.0000 (100.0000) acc1_neg: 61.458 acc5_neg: 100.000 time: 0.1845 data: 0.0911 max mem: 3532 +Test: [260/333] eta: 0:00:14 acc1_pos: 57.237 acc5_pos: 100.000 loss: 0.6552 (0.6503) acc1: 64.5833 (62.9969) acc5: 100.0000 (100.0000) acc1_neg: 68.750 acc5_neg: 100.000 time: 0.1783 data: 0.0826 max mem: 3532 +Test: [270/333] eta: 0:00:12 acc1_pos: 57.237 acc5_pos: 100.000 loss: 0.6137 (0.6482) acc1: 68.2292 (63.2284) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.1739 data: 0.0768 max mem: 3532 +Test: [280/333] eta: 0:00:10 acc1_pos: 57.237 acc5_pos: 100.000 loss: 0.6137 (0.6523) acc1: 66.1458 (62.8855) acc5: 100.0000 (100.0000) acc1_neg: 57.031 acc5_neg: 100.000 time: 0.1645 data: 0.0710 max mem: 3532 +Test: [290/333] eta: 0:00:08 acc1_pos: 57.237 acc5_pos: 100.000 loss: 0.8831 (0.6601) acc1: 43.7500 (62.2503) acc5: 100.0000 (100.0000) acc1_neg: 53.125 acc5_neg: 100.000 time: 0.1652 data: 0.0741 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 57.237 acc5_pos: 100.000 loss: 0.8186 (0.6648) acc1: 51.3021 (61.9290) acc5: 100.0000 (100.0000) acc1_neg: 52.604 acc5_neg: 100.000 time: 0.1787 data: 0.0875 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 57.237 acc5_pos: 100.000 loss: 0.7128 (0.6630) acc1: 57.8125 (62.0805) acc5: 100.0000 (100.0000) acc1_neg: 58.073 acc5_neg: 100.000 time: 0.1763 data: 0.0847 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 57.237 acc5_pos: 100.000 loss: 0.6298 (0.6630) acc1: 64.5833 (62.1122) acc5: 100.0000 (100.0000) acc1_neg: 51.562 acc5_neg: 100.000 time: 0.1718 data: 0.0795 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 57.237 acc5_pos: 100.000 loss: 0.8432 (0.6701) acc1: 51.5625 (61.6653) acc5: 100.0000 (100.0000) acc1_neg: 50.521 acc5_neg: 100.000 time: 0.1645 data: 0.0725 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 57.237 acc5_pos: 100.000 loss: 0.8664 (0.6715) acc1: 50.2604 (61.5857) acc5: 100.0000 (100.0000) acc1_neg: 45.352 acc5_neg: 100.000 time: 0.1619 data: 0.0667 max mem: 3532 +Test: Total time: 0:01:03 (0.1904 s / it) +* Overall Acc@1 61.586 Acc@5 100.000 + target == real Acc@1 64.292 Acc@5 100.000 + target == fake Acc@1 59.176 Acc@5 100.000 + loss 0.671 + PR-AUC 0.6796 ROC-AUC 0.6668 EER 38.25% (thr=0.4800) +Accuracy of the network on the 127843 test images: 61.6% +[rank0]:[W726 04:21:37.001304975 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:21:56.442000 3153610 torch/distributed/run.py:766] +W0726 04:21:56.442000 3153610 torch/distributed/run.py:766] ***************************************** +W0726 04:21:56.442000 3153610 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:21:56.442000 3153610 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:22:26.735950408 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:22:26.957595543 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:22:26.322204565 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:22:26.378054563 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:22:26.400459313 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +[rank0]:[W726 04:22:26.427346641 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:22:26.440580299 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank4]:[W726 04:22:26.440623369 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n32_seed6_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed6_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:20:44 acc1_pos: 97.917 acc5_pos: 100.000 loss: 0.1758 (0.1758) acc1: 97.9167 (97.9167) acc5: 100.0000 (100.0000) time: 5.2972 data: 3.7056 max mem: 3531 +Test: [ 10/235] eta: 0:02:06 acc1_pos: 97.135 acc5_pos: 100.000 loss: 0.1767 (0.1784) acc1: 97.6562 (97.4195) acc5: 100.0000 (100.0000) time: 0.5631 data: 0.3372 max mem: 3531 +Test: [ 20/235] eta: 0:01:13 acc1_pos: 97.656 acc5_pos: 100.000 loss: 0.1739 (0.1758) acc1: 97.6562 (97.4454) acc5: 100.0000 (100.0000) time: 0.0919 data: 0.0012 max mem: 3531 +Test: [ 30/235] eta: 0:00:55 acc1_pos: 97.656 acc5_pos: 100.000 loss: 0.1733 (0.1750) acc1: 97.6562 (97.5554) acc5: 100.0000 (100.0000) time: 0.1070 data: 0.0140 max mem: 3531 +Test: [ 40/235] eta: 0:00:48 acc1_pos: 97.656 acc5_pos: 100.000 loss: 0.1733 (0.1752) acc1: 97.6562 (97.5673) acc5: 100.0000 (100.0000) time: 0.1499 data: 0.0557 max mem: 3531 +Test: [ 50/235] eta: 0:00:43 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.1760 (0.1763) acc1: 97.1354 (97.4724) acc5: 100.0000 (100.0000) time: 0.1849 data: 0.0915 max mem: 3531 +Test: [ 60/235] eta: 0:00:40 acc1_pos: 97.396 acc5_pos: 100.000 loss: 0.1824 (0.1777) acc1: 96.8750 (97.4172) acc5: 100.0000 (100.0000) time: 0.1999 data: 0.1075 max mem: 3531 +Test: [ 70/235] eta: 0:00:37 acc1_pos: 98.462 acc5_pos: 100.000 loss: 0.1959 (0.3349) acc1: 96.3542 (88.0062) acc5: 100.0000 (100.0000) acc1_neg: 20.573 acc5_neg: 100.000 time: 0.2049 data: 0.1106 max mem: 3531 +Test: [ 80/235] eta: 0:00:34 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4267 (0.4706) acc1: 22.9167 (79.9318) acc5: 100.0000 (100.0000) acc1_neg: 21.354 acc5_neg: 100.000 time: 0.1899 data: 0.0912 max mem: 3531 +Test: [ 90/235] eta: 0:00:31 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4267 (0.5774) acc1: 21.6146 (73.5148) acc5: 100.0000 (100.0000) acc1_neg: 21.615 acc5_neg: 100.000 time: 0.1800 data: 0.0839 max mem: 3531 +Test: [100/235] eta: 0:00:29 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4255 (0.6604) acc1: 21.6146 (68.5205) acc5: 100.0000 (100.0000) acc1_neg: 21.875 acc5_neg: 100.000 time: 0.1999 data: 0.1071 max mem: 3531 +Test: [110/235] eta: 0:00:27 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4207 (0.7289) acc1: 22.9167 (64.3722) acc5: 100.0000 (100.0000) acc1_neg: 25.260 acc5_neg: 100.000 time: 0.2199 data: 0.1235 max mem: 3531 +Test: [120/235] eta: 0:00:24 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4578 (0.8097) acc1: 19.5312 (59.8765) acc5: 100.0000 (100.0000) acc1_neg: 9.896 acc5_neg: 100.000 time: 0.1951 data: 0.1005 max mem: 3531 +Test: [130/235] eta: 0:00:22 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.7170 (0.8788) acc1: 8.5938 (55.9598) acc5: 100.0000 (100.0000) acc1_neg: 7.292 acc5_neg: 100.000 time: 0.1949 data: 0.1029 max mem: 3531 +Test: [140/235] eta: 0:00:20 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.7409 (0.9412) acc1: 8.0729 (52.5525) acc5: 100.0000 (100.0000) acc1_neg: 6.510 acc5_neg: 100.000 time: 0.2247 data: 0.1313 max mem: 3531 +Test: [150/235] eta: 0:00:18 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.7279 (0.9923) acc1: 8.3333 (49.6637) acc5: 100.0000 (100.0000) acc1_neg: 10.677 acc5_neg: 100.000 time: 0.2099 data: 0.1152 max mem: 3531 +Test: [160/235] eta: 0:00:15 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.7115 (1.0372) acc1: 8.3333 (47.1273) acc5: 100.0000 (100.0000) acc1_neg: 9.635 acc5_neg: 100.000 time: 0.1805 data: 0.0852 max mem: 3531 +Test: [170/235] eta: 0:00:13 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.6697 (1.0493) acc1: 9.6354 (46.0359) acc5: 100.0000 (100.0000) acc1_neg: 30.208 acc5_neg: 100.000 time: 0.1909 data: 0.0977 max mem: 3531 +Test: [180/235] eta: 0:00:11 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.1890 (1.0568) acc1: 30.2083 (45.2046) acc5: 100.0000 (100.0000) acc1_neg: 29.688 acc5_neg: 100.000 time: 0.2093 data: 0.1180 max mem: 3531 +Test: [190/235] eta: 0:00:09 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.1846 (1.0634) acc1: 31.2500 (44.4740) acc5: 100.0000 (100.0000) acc1_neg: 32.552 acc5_neg: 100.000 time: 0.1909 data: 0.0998 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.1851 (1.0703) acc1: 30.2083 (43.7500) acc5: 100.0000 (100.0000) acc1_neg: 30.469 acc5_neg: 100.000 time: 0.1884 data: 0.0971 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.1848 (1.0755) acc1: 30.4688 (43.1712) acc5: 100.0000 (100.0000) acc1_neg: 33.854 acc5_neg: 100.000 time: 0.1943 data: 0.1027 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.1795 (1.0803) acc1: 29.9479 (42.6247) acc5: 100.0000 (100.0000) acc1_neg: 29.948 acc5_neg: 100.000 time: 0.1656 data: 0.0746 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.1756 (1.0844) acc1: 32.5521 (42.2078) acc5: 100.0000 (100.0000) acc1_neg: 29.948 acc5_neg: 100.000 time: 0.1365 data: 0.0457 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.1758 (1.0874) acc1: 32.0312 (42.0376) acc5: 100.0000 (100.0000) acc1_neg: 26.087 acc5_neg: 100.000 time: 0.1191 data: 0.0316 max mem: 3532 +Test: Total time: 0:00:47 (0.2005 s / it) +* Overall Acc@1 42.038 Acc@5 100.000 + target == real Acc@1 97.420 Acc@5 100.000 + target == fake Acc@1 22.007 Acc@5 100.000 + loss 1.087 + PR-AUC 0.8937 ROC-AUC 0.7559 EER 31.39% (thr=0.1554) +Accuracy of the network on the 89879 test images: 42.0% +[rank0]:[W726 04:23:24.417377380 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:23:44.328000 3154168 torch/distributed/run.py:766] +W0726 04:23:44.328000 3154168 torch/distributed/run.py:766] ***************************************** +W0726 04:23:44.328000 3154168 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:23:44.328000 3154168 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:24:13.285234893 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:24:13.358550021 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:24:14.704771631 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:24:14.764401629 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:24:14.184894994 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:24:14.263073818 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:24:14.268710866 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:24:14.294331546 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n32_seed6_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed6_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:22:23 acc1_pos: 75.000 acc5_pos: 100.000 loss: 0.5259 (0.5259) acc1: 75.0000 (75.0000) acc5: 100.0000 (100.0000) time: 4.9956 data: 3.2565 max mem: 3531 +Test: [ 10/269] eta: 0:02:18 acc1_pos: 88.281 acc5_pos: 100.000 loss: 0.5259 (0.5005) acc1: 75.5208 (78.8589) acc5: 100.0000 (100.0000) time: 0.5359 data: 0.2964 max mem: 3531 +Test: [ 20/269] eta: 0:01:24 acc1_pos: 86.719 acc5_pos: 100.000 loss: 0.4904 (0.5001) acc1: 77.0833 (78.6582) acc5: 100.0000 (100.0000) time: 0.1047 data: 0.0129 max mem: 3531 +Test: [ 30/269] eta: 0:01:06 acc1_pos: 78.385 acc5_pos: 100.000 loss: 0.4890 (0.4883) acc1: 78.9062 (79.9815) acc5: 100.0000 (100.0000) time: 0.1397 data: 0.0422 max mem: 3531 +Test: [ 40/269] eta: 0:01:04 acc1_pos: 86.207 acc5_pos: 100.000 loss: 0.5331 (0.5422) acc1: 73.4375 (74.3267) acc5: 100.0000 (100.0000) acc1_neg: 52.865 acc5_neg: 100.000 time: 0.2206 data: 0.1238 max mem: 3531 +Test: [ 50/269] eta: 0:00:57 acc1_pos: 86.207 acc5_pos: 100.000 loss: 0.6942 (0.5763) acc1: 58.3333 (70.5780) acc5: 100.0000 (100.0000) acc1_neg: 43.490 acc5_neg: 100.000 time: 0.2399 data: 0.1489 max mem: 3531 +Test: [ 60/269] eta: 0:00:52 acc1_pos: 86.207 acc5_pos: 100.000 loss: 0.7371 (0.6143) acc1: 52.8646 (66.6880) acc5: 100.0000 (100.0000) acc1_neg: 27.344 acc5_neg: 100.000 time: 0.1843 data: 0.0886 max mem: 3531 +Test: [ 70/269] eta: 0:00:47 acc1_pos: 86.207 acc5_pos: 100.000 loss: 0.7864 (0.6386) acc1: 44.5312 (63.5930) acc5: 100.0000 (100.0000) acc1_neg: 48.177 acc5_neg: 100.000 time: 0.1672 data: 0.0690 max mem: 3531 +Test: [ 80/269] eta: 0:00:43 acc1_pos: 86.207 acc5_pos: 100.000 loss: 0.7393 (0.6470) acc1: 48.1771 (62.2460) acc5: 100.0000 (100.0000) acc1_neg: 64.062 acc5_neg: 100.000 time: 0.1699 data: 0.0746 max mem: 3531 +Test: [ 90/269] eta: 0:00:39 acc1_pos: 86.207 acc5_pos: 100.000 loss: 0.7393 (0.6628) acc1: 48.6979 (60.1648) acc5: 100.0000 (100.0000) acc1_neg: 28.906 acc5_neg: 100.000 time: 0.1631 data: 0.0689 max mem: 3531 +Test: [100/269] eta: 0:00:36 acc1_pos: 86.207 acc5_pos: 100.000 loss: 0.8150 (0.6818) acc1: 42.1875 (58.5731) acc5: 100.0000 (100.0000) acc1_neg: 38.281 acc5_neg: 100.000 time: 0.1622 data: 0.0679 max mem: 3531 +Test: [110/269] eta: 0:00:33 acc1_pos: 86.207 acc5_pos: 100.000 loss: 0.8500 (0.6996) acc1: 41.9271 (56.6981) acc5: 100.0000 (100.0000) acc1_neg: 32.031 acc5_neg: 100.000 time: 0.1761 data: 0.0831 max mem: 3531 +Test: [120/269] eta: 0:00:31 acc1_pos: 86.207 acc5_pos: 100.000 loss: 0.7871 (0.6987) acc1: 45.8333 (56.9043) acc5: 100.0000 (100.0000) acc1_neg: 52.865 acc5_neg: 100.000 time: 0.1928 data: 0.1026 max mem: 3531 +Test: [130/269] eta: 0:00:29 acc1_pos: 86.207 acc5_pos: 100.000 loss: 0.7402 (0.7077) acc1: 48.1771 (56.0492) acc5: 100.0000 (100.0000) acc1_neg: 44.010 acc5_neg: 100.000 time: 0.1932 data: 0.1010 max mem: 3531 +Test: [140/269] eta: 0:00:26 acc1_pos: 86.207 acc5_pos: 100.000 loss: 0.7719 (0.7120) acc1: 48.1771 (55.5611) acc5: 100.0000 (100.0000) acc1_neg: 49.479 acc5_neg: 100.000 time: 0.1699 data: 0.0772 max mem: 3531 +Test: [150/269] eta: 0:00:24 acc1_pos: 86.207 acc5_pos: 100.000 loss: 0.7245 (0.7101) acc1: 50.7812 (55.8551) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.1603 data: 0.0667 max mem: 3531 +Test: [160/269] eta: 0:00:21 acc1_pos: 86.207 acc5_pos: 100.000 loss: 0.7245 (0.7142) acc1: 52.8646 (55.4962) acc5: 100.0000 (100.0000) acc1_neg: 41.667 acc5_neg: 100.000 time: 0.1671 data: 0.0710 max mem: 3531 +Test: [170/269] eta: 0:00:19 acc1_pos: 86.207 acc5_pos: 100.000 loss: 0.8559 (0.7236) acc1: 42.1875 (54.7286) acc5: 100.0000 (100.0000) acc1_neg: 39.583 acc5_neg: 100.000 time: 0.1620 data: 0.0655 max mem: 3531 +Test: [180/269] eta: 0:00:17 acc1_pos: 86.207 acc5_pos: 100.000 loss: 0.8095 (0.7246) acc1: 44.0104 (54.5638) acc5: 100.0000 (100.0000) acc1_neg: 45.573 acc5_neg: 100.000 time: 0.1589 data: 0.0656 max mem: 3531 +Test: [190/269] eta: 0:00:15 acc1_pos: 86.207 acc5_pos: 100.000 loss: 0.7883 (0.7301) acc1: 45.5729 (54.0576) acc5: 100.0000 (100.0000) acc1_neg: 45.573 acc5_neg: 100.000 time: 0.1724 data: 0.0795 max mem: 3531 +Test: [200/269] eta: 0:00:13 acc1_pos: 86.207 acc5_pos: 100.000 loss: 0.8797 (0.7406) acc1: 41.1458 (53.4334) acc5: 100.0000 (100.0000) acc1_neg: 66.406 acc5_neg: 100.000 time: 0.1658 data: 0.0695 max mem: 3531 +Test: [210/269] eta: 0:00:11 acc1_pos: 86.207 acc5_pos: 100.000 loss: 0.8078 (0.7424) acc1: 41.1458 (53.2743) acc5: 100.0000 (100.0000) acc1_neg: 55.990 acc5_neg: 100.000 time: 0.1499 data: 0.0553 max mem: 3532 +Test: [220/269] eta: 0:00:09 acc1_pos: 86.207 acc5_pos: 100.000 loss: 0.7757 (0.7436) acc1: 48.1771 (53.2747) acc5: 100.0000 (100.0000) acc1_neg: 62.760 acc5_neg: 100.000 time: 0.1476 data: 0.0560 max mem: 3532 +Test: [230/269] eta: 0:00:07 acc1_pos: 86.207 acc5_pos: 100.000 loss: 0.7076 (0.7398) acc1: 55.4688 (53.5804) acc5: 100.0000 (100.0000) acc1_neg: 70.052 acc5_neg: 100.000 time: 0.1756 data: 0.0841 max mem: 3532 +Test: [240/269] eta: 0:00:05 acc1_pos: 86.207 acc5_pos: 100.000 loss: 0.7199 (0.7439) acc1: 50.5208 (53.1985) acc5: 100.0000 (100.0000) acc1_neg: 50.000 acc5_neg: 100.000 time: 0.1961 data: 0.1047 max mem: 3532 +Test: [250/269] eta: 0:00:03 acc1_pos: 86.207 acc5_pos: 100.000 loss: 0.7833 (0.7468) acc1: 45.5729 (52.9175) acc5: 100.0000 (100.0000) acc1_neg: 22.396 acc5_neg: 100.000 time: 0.1911 data: 0.0997 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 86.207 acc5_pos: 100.000 loss: 0.8741 (0.7534) acc1: 35.1562 (52.1203) acc5: 100.0000 (100.0000) acc1_neg: 41.146 acc5_neg: 100.000 time: 0.1724 data: 0.0818 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 86.207 acc5_pos: 100.000 loss: 0.8618 (0.7546) acc1: 35.1562 (51.9028) acc5: 100.0000 (100.0000) acc1_neg: 36.483 acc5_neg: 100.000 time: 0.1704 data: 0.0769 max mem: 3532 +Test: Total time: 0:00:50 (0.1889 s / it) +* Overall Acc@1 51.903 Acc@5 100.000 + target == real Acc@1 80.023 Acc@5 100.000 + target == fake Acc@1 48.065 Acc@5 100.000 + loss 0.755 + PR-AUC 0.9417 ROC-AUC 0.7007 EER 35.65% (thr=0.4038) +Accuracy of the network on the 103293 test images: 51.9% +[rank0]:[W726 04:25:18.452045035 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T04:25:25Z] eval_done tag=fixed500k_fake_n32_seed6_resampled +[2026-07-26T04:25:26Z] eval_start tag=fixed500k_real_n16_seed7_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed7_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T04:25:26Z +hostname=n111-046-081 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed7_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n16_seed7_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 04:25:38.521000 3154745 torch/distributed/run.py:766] +W0726 04:25:38.521000 3154745 torch/distributed/run.py:766] ***************************************** +W0726 04:25:38.521000 3154745 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:25:38.521000 3154745 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:26:07.340796752 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:26:08.598852917 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:26:08.637286840 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:26:08.673398628 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +[rank0]:[W726 04:26:08.036987004 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:26:08.042261557 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:26:08.165750099 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:26:08.190498331 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n16_seed7_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed7_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:19:34 acc1_pos: 94.792 acc5_pos: 100.000 loss: 0.2878 (0.2878) acc1: 94.7917 (94.7917) acc5: 100.0000 (100.0000) time: 5.0195 data: 2.9919 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/234] eta: 0:02:01 acc1_pos: 94.010 acc5_pos: 100.000 loss: 0.3921 (0.4855) acc1: 91.6667 (79.3087) acc5: 100.0000 (100.0000) time: 0.5420 data: 0.2756 max mem: 3531 +Test: [ 20/234] eta: 0:01:11 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.3921 (0.4665) acc1: 86.4583 (79.9355) acc5: 100.0000 (100.0000) time: 0.1006 data: 0.0108 max mem: 3531 +Test: [ 30/234] eta: 0:00:56 acc1_pos: 72.917 acc5_pos: 100.000 loss: 0.2304 (0.4016) acc1: 94.5312 (83.8542) acc5: 100.0000 (100.0000) time: 0.1276 data: 0.0364 max mem: 3531 +Test: [ 40/234] eta: 0:00:51 acc1_pos: 95.833 acc5_pos: 100.000 loss: 0.2252 (0.4114) acc1: 94.5312 (82.7109) acc5: 100.0000 (100.0000) time: 0.1934 data: 0.0994 max mem: 3531 +Test: [ 50/234] eta: 0:00:48 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.2126 (0.3705) acc1: 94.7917 (85.3094) acc5: 100.0000 (100.0000) time: 0.2414 data: 0.1451 max mem: 3531 +Test: [ 60/234] eta: 0:00:43 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.1667 (0.3304) acc1: 97.3958 (87.6708) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2189 data: 0.1229 max mem: 3531 +Test: [ 70/234] eta: 0:00:39 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.1232 (0.3454) acc1: 99.7396 (86.3226) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.1949 data: 0.1025 max mem: 3531 +Test: [ 80/234] eta: 0:00:36 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.5477 (0.3608) acc1: 70.3125 (85.3106) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.2005 data: 0.1100 max mem: 3531 +Test: [ 90/234] eta: 0:00:33 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.2855 (0.3421) acc1: 91.1458 (86.3811) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.2101 data: 0.1194 max mem: 3531 +Test: [100/234] eta: 0:00:30 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.2765 (0.3530) acc1: 86.4583 (85.4579) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1976 data: 0.1058 max mem: 3531 +Test: [110/234] eta: 0:00:28 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.3973 (0.3537) acc1: 82.8125 (85.3580) acc5: 100.0000 (100.0000) acc1_neg: 75.781 acc5_neg: 100.000 time: 0.1948 data: 0.1035 max mem: 3531 +Test: [120/234] eta: 0:00:25 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.3859 (0.3702) acc1: 82.8125 (84.1878) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.1999 data: 0.1073 max mem: 3531 +Test: [130/234] eta: 0:00:23 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.5417 (0.3870) acc1: 70.5729 (82.9994) acc5: 100.0000 (100.0000) acc1_neg: 72.396 acc5_neg: 100.000 time: 0.1948 data: 0.1001 max mem: 3531 +Test: [140/234] eta: 0:00:20 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.5788 (0.4094) acc1: 68.7500 (81.7598) acc5: 100.0000 (100.0000) acc1_neg: 51.562 acc5_neg: 100.000 time: 0.1797 data: 0.0864 max mem: 3531 +Test: [150/234] eta: 0:00:18 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.4855 (0.4168) acc1: 73.1771 (81.1258) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.1707 data: 0.0767 max mem: 3531 +Test: [160/234] eta: 0:00:15 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.5376 (0.4272) acc1: 71.8750 (80.3151) acc5: 100.0000 (100.0000) acc1_neg: 61.719 acc5_neg: 100.000 time: 0.1875 data: 0.0915 max mem: 3531 +Test: [170/234] eta: 0:00:13 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.6281 (0.4393) acc1: 63.2812 (79.3311) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.1899 data: 0.0942 max mem: 3531 +Test: [180/234] eta: 0:00:11 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.6281 (0.4474) acc1: 64.3229 (78.7250) acc5: 100.0000 (100.0000) acc1_neg: 48.177 acc5_neg: 100.000 time: 0.1923 data: 0.0978 max mem: 3531 +Test: [190/234] eta: 0:00:09 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.6750 (0.4610) acc1: 61.7188 (77.7282) acc5: 100.0000 (100.0000) acc1_neg: 57.552 acc5_neg: 100.000 time: 0.1846 data: 0.0914 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.6750 (0.4542) acc1: 61.7188 (78.1120) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1649 data: 0.0729 max mem: 3531 +Test: [210/234] eta: 0:00:04 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.1084 (0.4393) acc1: 99.7396 (79.1111) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1544 data: 0.0630 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.2002 (0.4297) acc1: 98.1771 (79.8690) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.1378 data: 0.0473 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.3911 (0.4454) acc1: 53.3854 (78.5579) acc5: 100.0000 (100.0000) acc1_neg: 50.000 acc5_neg: 100.000 time: 0.1239 data: 0.0339 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.7745 (0.4498) acc1: 50.5208 (78.2289) acc5: 100.0000 (100.0000) acc1_neg: 49.027 acc5_neg: 100.000 time: 0.1232 data: 0.0291 max mem: 3532 +Test: Total time: 0:00:46 (0.1979 s / it) +* Overall Acc@1 78.229 Acc@5 100.000 + target == real Acc@1 85.372 Acc@5 100.000 + target == fake Acc@1 76.216 Acc@5 100.000 + loss 0.450 + PR-AUC 0.9673 ROC-AUC 0.8930 EER 19.56% (thr=0.4417) +Accuracy of the network on the 89729 test images: 78.2% +[rank0]:[W726 04:27:05.641932713 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:27:24.440000 3155293 torch/distributed/run.py:766] +W0726 04:27:24.440000 3155293 torch/distributed/run.py:766] ***************************************** +W0726 04:27:24.440000 3155293 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:27:24.440000 3155293 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:27:51.371072681 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:27:51.472091448 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:27:52.617806505 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:27:52.650393184 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:27:52.727917109 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:27:52.948173712 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +[rank1]:[W726 04:27:52.978549897 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:27:52.982865827 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n16_seed7_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed7_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:38:15 acc1_pos: 71.615 acc5_pos: 100.000 loss: 0.5537 (0.5537) acc1: 71.6146 (71.6146) acc5: 100.0000 (100.0000) time: 6.9941 data: 5.5610 max mem: 3531 +Test: [ 10/2387] eta: 0:28:33 acc1_pos: 54.167 acc5_pos: 100.000 loss: 0.6994 (0.6622) acc1: 61.9792 (61.3163) acc5: 100.0000 (100.0000) time: 0.7210 data: 0.5059 max mem: 3531 +Test: [ 20/2387] eta: 0:16:46 acc1_pos: 69.531 acc5_pos: 100.000 loss: 0.6681 (0.6576) acc1: 61.9792 (62.2520) acc5: 100.0000 (100.0000) time: 0.0968 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:14:06 acc1_pos: 34.635 acc5_pos: 100.000 loss: 0.6835 (0.6740) acc1: 59.3750 (60.5511) acc5: 100.0000 (100.0000) time: 0.1599 data: 0.0602 max mem: 3531 +Test: [ 40/2387] eta: 0:13:01 acc1_pos: 65.104 acc5_pos: 100.000 loss: 0.7523 (0.6970) acc1: 50.7812 (58.4604) acc5: 100.0000 (100.0000) time: 0.2360 data: 0.1406 max mem: 3531 +Test: [ 50/2387] eta: 0:12:11 acc1_pos: 65.885 acc5_pos: 100.000 loss: 0.6971 (0.6873) acc1: 54.9479 (59.3290) acc5: 100.0000 (100.0000) time: 0.2418 data: 0.1493 max mem: 3531 +Test: [ 60/2387] eta: 0:11:47 acc1_pos: 64.844 acc5_pos: 100.000 loss: 0.6361 (0.6847) acc1: 61.9792 (59.8105) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1488 max mem: 3531 +Test: [ 70/2387] eta: 0:11:32 acc1_pos: 67.188 acc5_pos: 100.000 loss: 0.6756 (0.6849) acc1: 59.1146 (59.8995) acc5: 100.0000 (100.0000) time: 0.2629 data: 0.1628 max mem: 3531 +Test: [ 80/2387] eta: 0:11:07 acc1_pos: 97.656 acc5_pos: 100.000 loss: 0.5868 (0.6532) acc1: 67.1875 (63.0273) acc5: 100.0000 (100.0000) time: 0.2439 data: 0.1457 max mem: 3531 +Test: [ 90/2387] eta: 0:10:54 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.3463 (0.6181) acc1: 91.4062 (66.3977) acc5: 100.0000 (100.0000) time: 0.2360 data: 0.1412 max mem: 3531 +Test: [ 100/2387] eta: 0:10:36 acc1_pos: 96.615 acc5_pos: 100.000 loss: 0.3337 (0.5957) acc1: 91.4062 (68.5824) acc5: 100.0000 (100.0000) time: 0.2353 data: 0.1424 max mem: 3531 +Test: [ 110/2387] eta: 0:10:24 acc1_pos: 82.031 acc5_pos: 100.000 loss: 0.3610 (0.5748) acc1: 89.5833 (70.5682) acc5: 100.0000 (100.0000) time: 0.2238 data: 0.1308 max mem: 3531 +Test: [ 120/2387] eta: 0:10:14 acc1_pos: 94.010 acc5_pos: 100.000 loss: 0.3467 (0.5591) acc1: 93.7500 (72.1225) acc5: 100.0000 (100.0000) time: 0.2346 data: 0.1408 max mem: 3531 +Test: [ 130/2387] eta: 0:10:08 acc1_pos: 91.146 acc5_pos: 100.000 loss: 0.3439 (0.5431) acc1: 91.1458 (73.6442) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1510 max mem: 3531 +Test: [ 140/2387] eta: 0:10:01 acc1_pos: 97.917 acc5_pos: 100.000 loss: 0.3343 (0.5309) acc1: 91.1458 (74.8467) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1490 max mem: 3531 +Test: [ 150/2387] eta: 0:09:51 acc1_pos: 98.698 acc5_pos: 100.000 loss: 0.3313 (0.5179) acc1: 92.9688 (76.1296) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1345 max mem: 3531 +Test: [ 160/2387] eta: 0:09:41 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.3270 (0.5068) acc1: 93.2292 (77.0930) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.2154 data: 0.1223 max mem: 3531 +Test: [ 170/2387] eta: 0:09:29 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.3314 (0.4966) acc1: 90.3646 (77.8966) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1999 data: 0.1065 max mem: 3531 +Test: [ 180/2387] eta: 0:09:17 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.3355 (0.4876) acc1: 90.1042 (78.6084) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1844 data: 0.0899 max mem: 3531 +Test: [ 190/2387] eta: 0:09:08 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.3355 (0.4792) acc1: 90.8854 (79.2744) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1899 data: 0.0974 max mem: 3531 +Test: [ 200/2387] eta: 0:09:01 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.3292 (0.4717) acc1: 91.1458 (79.8520) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.2049 data: 0.1121 max mem: 3531 +Test: [ 210/2387] eta: 0:08:51 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.3292 (0.4653) acc1: 90.8854 (80.3638) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1901 data: 0.0969 max mem: 3532 +Test: [ 220/2387] eta: 0:08:45 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.3333 (0.4594) acc1: 90.6250 (80.8117) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1899 data: 0.0977 max mem: 3532 +Test: [ 230/2387] eta: 0:08:41 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.3353 (0.4543) acc1: 90.1042 (81.2184) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.2165 data: 0.1251 max mem: 3532 +Test: [ 240/2387] eta: 0:08:37 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.3515 (0.4583) acc1: 87.2396 (80.8037) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.2249 data: 0.1303 max mem: 3532 +Test: [ 250/2387] eta: 0:08:30 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.3190 (0.4474) acc1: 73.4375 (81.4202) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2081 data: 0.1102 max mem: 3532 +Test: [ 260/2387] eta: 0:08:17 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1401 (0.4353) acc1: 99.4792 (82.1091) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1449 data: 0.0469 max mem: 3532 +Test: [ 270/2387] eta: 0:08:14 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1248 (0.4237) acc1: 99.7396 (82.7539) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1616 data: 0.0667 max mem: 3532 +Test: [ 280/2387] eta: 0:08:16 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1239 (0.4131) acc1: 99.7396 (83.3519) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2624 data: 0.1717 max mem: 3532 +Test: [ 290/2387] eta: 0:08:11 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1211 (0.4029) acc1: 99.7396 (83.9114) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2490 data: 0.1578 max mem: 3532 +Test: [ 300/2387] eta: 0:08:04 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1246 (0.3942) acc1: 99.7396 (84.4312) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1823 data: 0.0888 max mem: 3532 +Test: [ 310/2387] eta: 0:08:02 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1459 (0.3892) acc1: 99.4792 (84.7995) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.2045 data: 0.1101 max mem: 3532 +Test: [ 320/2387] eta: 0:07:59 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.3432 (0.3881) acc1: 92.4479 (85.0338) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.2249 data: 0.1319 max mem: 3532 +Test: [ 330/2387] eta: 0:07:59 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.3488 (0.3869) acc1: 92.4479 (85.2743) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2445 data: 0.1514 max mem: 3532 +Test: [ 340/2387] eta: 0:07:53 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.3476 (0.3858) acc1: 92.7083 (85.4930) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2249 data: 0.1296 max mem: 3532 +Test: [ 350/2387] eta: 0:07:51 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.3395 (0.3808) acc1: 93.7500 (85.8759) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2064 data: 0.1106 max mem: 3532 +Test: [ 360/2387] eta: 0:07:44 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2398 (0.3775) acc1: 96.3542 (86.1258) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1956 data: 0.1010 max mem: 3532 +Test: [ 370/2387] eta: 0:07:37 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2678 (0.3747) acc1: 94.5312 (86.3636) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1416 data: 0.0478 max mem: 3532 +Test: [ 380/2387] eta: 0:07:33 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2691 (0.3720) acc1: 94.7917 (86.5896) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1598 data: 0.0652 max mem: 3532 +Test: [ 390/2387] eta: 0:07:28 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2738 (0.3695) acc1: 94.7917 (86.7854) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1779 data: 0.0818 max mem: 3532 +Test: [ 400/2387] eta: 0:07:23 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2730 (0.3671) acc1: 94.0104 (86.9766) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1699 data: 0.0734 max mem: 3532 +Test: [ 410/2387] eta: 0:07:18 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2706 (0.3648) acc1: 94.7917 (87.1693) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1749 data: 0.0806 max mem: 3532 +Test: [ 420/2387] eta: 0:07:14 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2723 (0.3626) acc1: 94.5312 (87.3355) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1805 data: 0.0879 max mem: 3532 +Test: [ 430/2387] eta: 0:07:09 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2786 (0.3632) acc1: 93.4896 (87.2740) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.1719 data: 0.0786 max mem: 3532 +Test: [ 440/2387] eta: 0:07:05 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2886 (0.3606) acc1: 82.0312 (87.3801) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1687 data: 0.0753 max mem: 3533 +Test: [ 450/2387] eta: 0:07:01 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1117 (0.3551) acc1: 100.0000 (87.6599) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0775 max mem: 3533 +Test: [ 460/2387] eta: 0:06:57 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1107 (0.3498) acc1: 100.0000 (87.9271) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1701 data: 0.0790 max mem: 3533 +Test: [ 470/2387] eta: 0:06:52 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1118 (0.3448) acc1: 100.0000 (88.1834) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1700 data: 0.0783 max mem: 3533 +Test: [ 480/2387] eta: 0:06:48 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1122 (0.3400) acc1: 100.0000 (88.4291) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1666 data: 0.0742 max mem: 3533 +Test: [ 490/2387] eta: 0:06:44 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1123 (0.3353) acc1: 100.0000 (88.6642) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1680 data: 0.0751 max mem: 3533 +Test: [ 500/2387] eta: 0:06:41 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1129 (0.3309) acc1: 100.0000 (88.8899) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1689 data: 0.0769 max mem: 3533 +Test: [ 510/2387] eta: 0:06:37 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1140 (0.3276) acc1: 100.0000 (89.0982) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1697 data: 0.0790 max mem: 3533 +Test: [ 520/2387] eta: 0:06:34 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1959 (0.3263) acc1: 99.4792 (89.2839) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1741 data: 0.0820 max mem: 3533 +Test: [ 530/2387] eta: 0:06:30 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1949 (0.3238) acc1: 99.4792 (89.4789) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1716 data: 0.0782 max mem: 3533 +Test: [ 540/2387] eta: 0:06:26 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1902 (0.3213) acc1: 99.4792 (89.6661) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1684 data: 0.0756 max mem: 3533 +Test: [ 550/2387] eta: 0:06:23 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1914 (0.3190) acc1: 99.7396 (89.8452) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1694 data: 0.0756 max mem: 3533 +Test: [ 560/2387] eta: 0:06:20 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1916 (0.3167) acc1: 99.7396 (90.0215) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1764 data: 0.0819 max mem: 3533 +Test: [ 570/2387] eta: 0:06:17 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1902 (0.3145) acc1: 99.7396 (90.1917) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1754 data: 0.0835 max mem: 3533 +Test: [ 580/2387] eta: 0:06:13 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1897 (0.3124) acc1: 99.7396 (90.3561) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1608 data: 0.0697 max mem: 3533 +Test: [ 590/2387] eta: 0:06:09 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1909 (0.3104) acc1: 99.7396 (90.5100) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1560 data: 0.0649 max mem: 3533 +Test: [ 600/2387] eta: 0:06:06 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2022 (0.3101) acc1: 98.6979 (90.6345) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1592 data: 0.0688 max mem: 3533 +Test: [ 610/2387] eta: 0:06:03 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.3084 (0.3105) acc1: 92.1875 (90.6276) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1633 data: 0.0725 max mem: 3533 +Test: [ 620/2387] eta: 0:06:00 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.3349 (0.3109) acc1: 90.3646 (90.6187) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.1650 data: 0.0735 max mem: 3533 +Test: [ 630/2387] eta: 0:05:57 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.3362 (0.3113) acc1: 90.3646 (90.6147) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1729 data: 0.0809 max mem: 3533 +Test: [ 640/2387] eta: 0:05:54 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.3364 (0.3117) acc1: 90.3646 (90.5986) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1777 data: 0.0828 max mem: 3533 +Test: [ 650/2387] eta: 0:05:51 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.3364 (0.3122) acc1: 90.3646 (90.5930) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1734 data: 0.0789 max mem: 3533 +Test: [ 660/2387] eta: 0:05:49 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.3362 (0.3125) acc1: 90.3646 (90.5955) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1877 data: 0.0966 max mem: 3533 +Test: [ 670/2387] eta: 0:05:46 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.3325 (0.3128) acc1: 90.8854 (90.5994) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.1811 data: 0.0902 max mem: 3534 +Test: [ 680/2387] eta: 0:05:44 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.3436 (0.3149) acc1: 90.3646 (90.4254) acc5: 100.0000 (100.0000) acc1_neg: 70.833 acc5_neg: 100.000 time: 0.1695 data: 0.0788 max mem: 3534 +Test: [ 690/2387] eta: 0:05:41 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.4871 (0.3173) acc1: 74.4792 (90.2014) acc5: 100.0000 (100.0000) acc1_neg: 72.396 acc5_neg: 100.000 time: 0.1749 data: 0.0837 max mem: 3534 +Test: [ 700/2387] eta: 0:05:38 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.4676 (0.3194) acc1: 75.7812 (89.9957) acc5: 100.0000 (100.0000) acc1_neg: 77.083 acc5_neg: 100.000 time: 0.1720 data: 0.0811 max mem: 3534 +Test: [ 710/2387] eta: 0:05:36 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.4645 (0.3214) acc1: 75.7812 (89.7972) acc5: 100.0000 (100.0000) acc1_neg: 76.042 acc5_neg: 100.000 time: 0.1799 data: 0.0872 max mem: 3534 +Test: [ 720/2387] eta: 0:05:33 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.4625 (0.3233) acc1: 76.0417 (89.6086) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.1827 data: 0.0893 max mem: 3534 +Test: [ 730/2387] eta: 0:05:31 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.4625 (0.3253) acc1: 75.5208 (89.4155) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.1751 data: 0.0840 max mem: 3534 +Test: [ 740/2387] eta: 0:05:28 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.4563 (0.3270) acc1: 76.3021 (89.2417) acc5: 100.0000 (100.0000) acc1_neg: 74.219 acc5_neg: 100.000 time: 0.1809 data: 0.0898 max mem: 3534 +Test: [ 750/2387] eta: 0:05:26 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.4524 (0.3287) acc1: 77.3438 (89.0795) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.1824 data: 0.0917 max mem: 3534 +Test: [ 760/2387] eta: 0:05:24 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.4441 (0.3301) acc1: 77.6042 (88.9407) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.1812 data: 0.0892 max mem: 3534 +Test: [ 770/2387] eta: 0:05:21 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.3887 (0.3300) acc1: 82.5521 (88.9274) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1835 data: 0.0890 max mem: 3534 +Test: [ 780/2387] eta: 0:05:19 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2560 (0.3286) acc1: 94.7917 (89.0305) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1754 data: 0.0810 max mem: 3534 +Test: [ 790/2387] eta: 0:05:16 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2186 (0.3273) acc1: 95.5729 (89.0981) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1712 data: 0.0777 max mem: 3534 +Test: [ 800/2387] eta: 0:05:14 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2156 (0.3260) acc1: 96.3542 (89.1968) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1868 data: 0.0934 max mem: 3534 +Test: [ 810/2387] eta: 0:05:12 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2109 (0.3246) acc1: 97.1354 (89.2847) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1937 data: 0.1022 max mem: 3534 +Test: [ 820/2387] eta: 0:05:10 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2208 (0.3237) acc1: 96.6146 (89.3642) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.1865 data: 0.0964 max mem: 3534 +Test: [ 830/2387] eta: 0:05:08 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.3591 (0.3256) acc1: 84.1146 (89.2079) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.1950 data: 0.1046 max mem: 3534 +Test: [ 840/2387] eta: 0:05:06 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.4685 (0.3274) acc1: 76.5625 (89.0845) acc5: 100.0000 (100.0000) acc1_neg: 75.781 acc5_neg: 100.000 time: 0.1961 data: 0.1042 max mem: 3534 +Test: [ 850/2387] eta: 0:05:05 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.4603 (0.3292) acc1: 77.0833 (88.9572) acc5: 100.0000 (100.0000) acc1_neg: 68.750 acc5_neg: 100.000 time: 0.2149 data: 0.1226 max mem: 3534 +Test: [ 860/2387] eta: 0:05:03 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.4448 (0.3304) acc1: 78.6458 (88.8635) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.2221 data: 0.1306 max mem: 3534 +Test: [ 870/2387] eta: 0:05:01 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.4319 (0.3316) acc1: 78.6458 (88.7770) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.2044 data: 0.1113 max mem: 3534 +Test: [ 880/2387] eta: 0:04:58 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.4135 (0.3329) acc1: 81.2500 (88.6992) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1719 data: 0.0796 max mem: 3534 +Test: [ 890/2387] eta: 0:04:56 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.4501 (0.3348) acc1: 79.1667 (88.5630) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.1769 data: 0.0841 max mem: 3534 +Test: [ 900/2387] eta: 0:04:54 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.5675 (0.3399) acc1: 68.2292 (88.1428) acc5: 100.0000 (100.0000) acc1_neg: 48.177 acc5_neg: 100.000 time: 0.2147 data: 0.1206 max mem: 3535 +Test: [ 910/2387] eta: 0:04:52 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.8489 (0.3455) acc1: 45.5729 (87.6684) acc5: 100.0000 (100.0000) acc1_neg: 42.188 acc5_neg: 100.000 time: 0.2047 data: 0.1130 max mem: 3535 +Test: [ 920/2387] eta: 0:04:50 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.8658 (0.3512) acc1: 44.2708 (87.2079) acc5: 100.0000 (100.0000) acc1_neg: 42.969 acc5_neg: 100.000 time: 0.1930 data: 0.1018 max mem: 3535 +Test: [ 930/2387] eta: 0:04:48 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.8678 (0.3566) acc1: 44.7917 (86.7655) acc5: 100.0000 (100.0000) acc1_neg: 44.792 acc5_neg: 100.000 time: 0.1992 data: 0.1085 max mem: 3535 +Test: [ 940/2387] eta: 0:04:47 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.8654 (0.3619) acc1: 46.0938 (86.3272) acc5: 100.0000 (100.0000) acc1_neg: 40.885 acc5_neg: 100.000 time: 0.2072 data: 0.1159 max mem: 3535 +Test: [ 950/2387] eta: 0:04:45 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.8513 (0.3670) acc1: 45.8333 (85.8992) acc5: 100.0000 (100.0000) acc1_neg: 47.917 acc5_neg: 100.000 time: 0.2178 data: 0.1246 max mem: 3535 +Test: [ 960/2387] eta: 0:04:43 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.8439 (0.3720) acc1: 45.8333 (85.4917) acc5: 100.0000 (100.0000) acc1_neg: 46.354 acc5_neg: 100.000 time: 0.1985 data: 0.1066 max mem: 3535 +Test: [ 970/2387] eta: 0:04:40 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.8466 (0.3770) acc1: 45.5729 (85.0694) acc5: 100.0000 (100.0000) acc1_neg: 44.792 acc5_neg: 100.000 time: 0.1739 data: 0.0832 max mem: 3535 +Test: [ 980/2387] eta: 0:04:38 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.8358 (0.3781) acc1: 46.8750 (84.9853) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.1694 data: 0.0783 max mem: 3535 +Test: [ 990/2387] eta: 0:04:35 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.3199 (0.3770) acc1: 88.5417 (85.0632) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1664 data: 0.0751 max mem: 3535 +Test: [1000/2387] eta: 0:04:33 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2388 (0.3757) acc1: 96.0938 (85.1547) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.1739 data: 0.0806 max mem: 3535 +Test: [1010/2387] eta: 0:04:31 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2741 (0.3759) acc1: 92.9688 (85.1457) acc5: 100.0000 (100.0000) acc1_neg: 76.302 acc5_neg: 100.000 time: 0.1790 data: 0.0865 max mem: 3535 +Test: [1020/2387] eta: 0:04:28 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.3224 (0.3760) acc1: 81.7708 (85.1302) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1665 data: 0.0760 max mem: 3535 +Test: [1030/2387] eta: 0:04:26 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2631 (0.3747) acc1: 94.5312 (85.2277) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1621 data: 0.0702 max mem: 3535 +Test: [1040/2387] eta: 0:04:24 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2087 (0.3731) acc1: 96.8750 (85.3391) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1713 data: 0.0776 max mem: 3535 +Test: [1050/2387] eta: 0:04:22 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1975 (0.3716) acc1: 97.3958 (85.4462) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1853 data: 0.0932 max mem: 3535 +Test: [1060/2387] eta: 0:04:20 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2177 (0.3704) acc1: 96.8750 (85.5337) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2116 data: 0.1208 max mem: 3535 +Test: [1070/2387] eta: 0:04:18 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2674 (0.3701) acc1: 93.2292 (85.5514) acc5: 100.0000 (100.0000) acc1_neg: 57.292 acc5_neg: 100.000 time: 0.2129 data: 0.1208 max mem: 3535 +Test: [1080/2387] eta: 0:04:17 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.4300 (0.3715) acc1: 77.3438 (85.4398) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.2030 data: 0.1112 max mem: 3535 +Test: [1090/2387] eta: 0:04:15 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.3616 (0.3708) acc1: 83.5938 (85.4909) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.2306 data: 0.1393 max mem: 3535 +Test: [1100/2387] eta: 0:04:13 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2671 (0.3700) acc1: 91.6667 (85.5588) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2224 data: 0.1314 max mem: 3535 +Test: [1110/2387] eta: 0:04:11 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2847 (0.3694) acc1: 91.9271 (85.6014) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1970 data: 0.1035 max mem: 3535 +Test: [1120/2387] eta: 0:04:10 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2913 (0.3689) acc1: 90.3646 (85.6390) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.2105 data: 0.1170 max mem: 3536 +Test: [1130/2387] eta: 0:04:08 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2865 (0.3680) acc1: 92.7083 (85.7128) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2156 data: 0.1251 max mem: 3536 +Test: [1140/2387] eta: 0:04:06 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.3351 (0.3695) acc1: 84.3750 (85.6056) acc5: 100.0000 (100.0000) acc1_neg: 67.969 acc5_neg: 100.000 time: 0.2188 data: 0.1281 max mem: 3536 +Test: [1150/2387] eta: 0:04:04 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.4803 (0.3701) acc1: 75.0000 (85.5619) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.2117 data: 0.1207 max mem: 3536 +Test: [1160/2387] eta: 0:04:02 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.3715 (0.3698) acc1: 82.8125 (85.5865) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.1951 data: 0.1028 max mem: 3536 +Test: [1170/2387] eta: 0:04:00 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.3575 (0.3697) acc1: 87.7604 (85.5979) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2090 data: 0.1172 max mem: 3536 +Test: [1180/2387] eta: 0:03:58 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.3385 (0.3695) acc1: 87.7604 (85.6114) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.2088 data: 0.1184 max mem: 3536 +Test: [1190/2387] eta: 0:03:56 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2939 (0.3691) acc1: 91.6667 (85.6395) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1986 data: 0.1078 max mem: 3536 +Test: [1200/2387] eta: 0:03:55 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2939 (0.3688) acc1: 89.5833 (85.6541) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.2045 data: 0.1129 max mem: 3536 +Test: [1210/2387] eta: 0:03:53 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.3121 (0.3682) acc1: 89.5833 (85.6969) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1995 data: 0.1066 max mem: 3536 +Test: [1220/2387] eta: 0:03:50 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1478 (0.3662) acc1: 98.4375 (85.8085) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1889 data: 0.0966 max mem: 3536 +Test: [1230/2387] eta: 0:03:48 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1265 (0.3642) acc1: 99.4792 (85.9191) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1930 data: 0.1023 max mem: 3536 +Test: [1240/2387] eta: 0:03:46 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1154 (0.3623) acc1: 99.4792 (86.0275) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2009 data: 0.1098 max mem: 3536 +Test: [1250/2387] eta: 0:03:44 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1143 (0.3603) acc1: 99.4792 (86.1355) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1944 data: 0.1024 max mem: 3536 +Test: [1260/2387] eta: 0:03:43 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1125 (0.3583) acc1: 99.7396 (86.2423) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2018 data: 0.1100 max mem: 3536 +Test: [1270/2387] eta: 0:03:41 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1247 (0.3566) acc1: 99.4792 (86.3454) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2026 data: 0.1117 max mem: 3536 +Test: [1280/2387] eta: 0:03:39 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1365 (0.3550) acc1: 99.2188 (86.4396) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2004 data: 0.1097 max mem: 3536 +Test: [1290/2387] eta: 0:03:37 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1487 (0.3535) acc1: 98.1771 (86.5277) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2019 data: 0.1112 max mem: 3536 +Test: [1300/2387] eta: 0:03:35 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1626 (0.3521) acc1: 97.6562 (86.6139) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2056 data: 0.1148 max mem: 3536 +Test: [1310/2387] eta: 0:03:33 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1714 (0.3508) acc1: 97.6562 (86.6937) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2164 data: 0.1247 max mem: 3536 +Test: [1320/2387] eta: 0:03:31 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1718 (0.3494) acc1: 97.3958 (86.7783) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2112 data: 0.1203 max mem: 3536 +Test: [1330/2387] eta: 0:03:29 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1874 (0.3493) acc1: 96.0938 (86.7882) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.2056 data: 0.1138 max mem: 3536 +Test: [1340/2387] eta: 0:03:27 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.3844 (0.3497) acc1: 85.4167 (86.7473) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2143 data: 0.1214 max mem: 3536 +Test: [1350/2387] eta: 0:03:25 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.0953 (0.3478) acc1: 100.0000 (86.8454) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2142 data: 0.1218 max mem: 3537 +Test: [1360/2387] eta: 0:03:24 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.0758 (0.3458) acc1: 100.0000 (86.9420) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2149 data: 0.1217 max mem: 3537 +Test: [1370/2387] eta: 0:03:22 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.0739 (0.3438) acc1: 100.0000 (87.0373) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2068 data: 0.1139 max mem: 3537 +Test: [1380/2387] eta: 0:03:20 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.0715 (0.3418) acc1: 100.0000 (87.1312) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1880 data: 0.0960 max mem: 3537 +Test: [1390/2387] eta: 0:03:17 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.0746 (0.3399) acc1: 100.0000 (87.2237) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1716 data: 0.0796 max mem: 3537 +Test: [1400/2387] eta: 0:03:15 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.0803 (0.3381) acc1: 100.0000 (87.3143) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1636 data: 0.0723 max mem: 3537 +Test: [1410/2387] eta: 0:03:13 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.0914 (0.3376) acc1: 100.0000 (87.3461) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1623 data: 0.0705 max mem: 3537 +Test: [1420/2387] eta: 0:03:11 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.3015 (0.3374) acc1: 91.4062 (87.3693) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1642 data: 0.0708 max mem: 3537 +Test: [1430/2387] eta: 0:03:08 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.3034 (0.3374) acc1: 90.1042 (87.3724) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1658 data: 0.0738 max mem: 3537 +Test: [1440/2387] eta: 0:03:06 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.3388 (0.3375) acc1: 87.2396 (87.3733) acc5: 100.0000 (100.0000) acc1_neg: 61.979 acc5_neg: 100.000 time: 0.1688 data: 0.0770 max mem: 3537 +Test: [1450/2387] eta: 0:03:04 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.3491 (0.3382) acc1: 84.3750 (87.3238) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1771 data: 0.0845 max mem: 3537 +Test: [1460/2387] eta: 0:03:02 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2548 (0.3376) acc1: 92.4479 (87.3740) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1737 data: 0.0819 max mem: 3537 +Test: [1470/2387] eta: 0:03:00 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2152 (0.3367) acc1: 96.6146 (87.4414) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1705 data: 0.0791 max mem: 3537 +Test: [1480/2387] eta: 0:02:58 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2076 (0.3359) acc1: 96.8750 (87.4967) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1842 data: 0.0926 max mem: 3537 +Test: [1490/2387] eta: 0:02:56 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2106 (0.3351) acc1: 96.8750 (87.5599) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1885 data: 0.0942 max mem: 3537 +Test: [1500/2387] eta: 0:02:54 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2083 (0.3343) acc1: 97.6562 (87.6142) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.1630 data: 0.0680 max mem: 3537 +Test: [1510/2387] eta: 0:02:52 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2691 (0.3345) acc1: 94.7917 (87.6201) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1586 data: 0.0660 max mem: 3537 +Test: [1520/2387] eta: 0:02:49 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.3848 (0.3350) acc1: 86.7188 (87.6007) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1638 data: 0.0732 max mem: 3537 +Test: [1530/2387] eta: 0:02:47 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.4010 (0.3355) acc1: 84.8958 (87.5684) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.1665 data: 0.0758 max mem: 3537 +Test: [1540/2387] eta: 0:02:45 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.3766 (0.3358) acc1: 84.8958 (87.5580) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.1731 data: 0.0813 max mem: 3537 +Test: [1550/2387] eta: 0:02:43 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.3600 (0.3360) acc1: 87.2396 (87.5589) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1704 data: 0.0770 max mem: 3537 +Test: [1560/2387] eta: 0:02:41 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.3424 (0.3362) acc1: 88.8021 (87.5589) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.1670 data: 0.0735 max mem: 3537 +Test: [1570/2387] eta: 0:02:39 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.3549 (0.3367) acc1: 87.5000 (87.5408) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1718 data: 0.0796 max mem: 3537 +Test: [1580/2387] eta: 0:02:37 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.3432 (0.3365) acc1: 89.0625 (87.5619) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1927 data: 0.1002 max mem: 3538 +Test: [1590/2387] eta: 0:02:35 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2310 (0.3357) acc1: 94.2708 (87.6201) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1823 data: 0.0884 max mem: 3538 +Test: [1600/2387] eta: 0:02:33 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2196 (0.3351) acc1: 96.3542 (87.6640) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1880 data: 0.0948 max mem: 3538 +Test: [1610/2387] eta: 0:02:31 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2178 (0.3344) acc1: 96.0938 (87.7190) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1974 data: 0.1055 max mem: 3538 +Test: [1620/2387] eta: 0:02:29 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2217 (0.3338) acc1: 95.3125 (87.7574) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1738 data: 0.0825 max mem: 3538 +Test: [1630/2387] eta: 0:02:27 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2456 (0.3338) acc1: 93.2292 (87.7619) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.1602 data: 0.0701 max mem: 3538 +Test: [1640/2387] eta: 0:02:25 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.4115 (0.3346) acc1: 82.0312 (87.7076) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.1661 data: 0.0738 max mem: 3538 +Test: [1650/2387] eta: 0:02:23 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.4504 (0.3353) acc1: 77.0833 (87.6604) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1787 data: 0.0863 max mem: 3538 +Test: [1660/2387] eta: 0:02:21 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.4384 (0.3359) acc1: 78.6458 (87.6176) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1732 data: 0.0821 max mem: 3538 +Test: [1670/2387] eta: 0:02:19 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.4215 (0.3364) acc1: 81.2500 (87.5873) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1710 data: 0.0798 max mem: 3538 +Test: [1680/2387] eta: 0:02:17 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.4301 (0.3369) acc1: 81.2500 (87.5533) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.1747 data: 0.0842 max mem: 3538 +Test: [1690/2387] eta: 0:02:15 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.4398 (0.3375) acc1: 80.7292 (87.5111) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1772 data: 0.0850 max mem: 3538 +Test: [1700/2387] eta: 0:02:13 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.4398 (0.3377) acc1: 79.9479 (87.4934) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1751 data: 0.0826 max mem: 3538 +Test: [1710/2387] eta: 0:02:11 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1774 (0.3367) acc1: 97.1354 (87.5545) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1628 data: 0.0722 max mem: 3538 +Test: [1720/2387] eta: 0:02:09 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1454 (0.3356) acc1: 98.6979 (87.6164) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1664 data: 0.0749 max mem: 3538 +Test: [1730/2387] eta: 0:02:07 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1346 (0.3345) acc1: 98.6979 (87.6768) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1753 data: 0.0830 max mem: 3538 +Test: [1740/2387] eta: 0:02:05 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1346 (0.3334) acc1: 99.2188 (87.7432) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1772 data: 0.0859 max mem: 3538 +Test: [1750/2387] eta: 0:02:02 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1485 (0.3323) acc1: 99.2188 (87.8077) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1666 data: 0.0736 max mem: 3538 +Test: [1760/2387] eta: 0:02:00 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1559 (0.3316) acc1: 98.6979 (87.8543) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1509 data: 0.0566 max mem: 3538 +Test: [1770/2387] eta: 0:01:58 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2054 (0.3310) acc1: 95.5729 (87.8939) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1534 data: 0.0596 max mem: 3538 +Test: [1780/2387] eta: 0:01:56 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2118 (0.3303) acc1: 95.3125 (87.9419) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1587 data: 0.0658 max mem: 3538 +Test: [1790/2387] eta: 0:01:54 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1949 (0.3295) acc1: 97.3958 (87.9936) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1470 data: 0.0555 max mem: 3538 +Test: [1800/2387] eta: 0:01:52 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1851 (0.3287) acc1: 97.1354 (88.0444) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1435 data: 0.0526 max mem: 3538 +Test: [1810/2387] eta: 0:01:50 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1729 (0.3278) acc1: 97.6562 (88.1015) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1562 data: 0.0658 max mem: 3539 +Test: [1820/2387] eta: 0:01:48 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1784 (0.3272) acc1: 97.6562 (88.1417) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1519 data: 0.0610 max mem: 3539 +Test: [1830/2387] eta: 0:01:46 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1985 (0.3266) acc1: 97.6562 (88.1894) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1507 data: 0.0598 max mem: 3539 +Test: [1840/2387] eta: 0:01:44 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1954 (0.3259) acc1: 98.9583 (88.2484) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1543 data: 0.0635 max mem: 3539 +Test: [1850/2387] eta: 0:01:42 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1931 (0.3252) acc1: 99.2188 (88.3074) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1532 data: 0.0626 max mem: 3539 +Test: [1860/2387] eta: 0:01:40 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1931 (0.3245) acc1: 98.9583 (88.3644) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1618 data: 0.0713 max mem: 3539 +Test: [1870/2387] eta: 0:01:38 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1954 (0.3238) acc1: 98.6979 (88.4207) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1716 data: 0.0806 max mem: 3539 +Test: [1880/2387] eta: 0:01:36 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1943 (0.3231) acc1: 98.9583 (88.4783) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1630 data: 0.0720 max mem: 3539 +Test: [1890/2387] eta: 0:01:34 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1937 (0.3224) acc1: 99.2188 (88.5346) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1547 data: 0.0616 max mem: 3539 +Test: [1900/2387] eta: 0:01:32 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1948 (0.3218) acc1: 98.9583 (88.5891) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1649 data: 0.0720 max mem: 3539 +Test: [1910/2387] eta: 0:01:30 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2839 (0.3219) acc1: 93.2292 (88.6089) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1724 data: 0.0810 max mem: 3539 +Test: [1920/2387] eta: 0:01:28 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.3315 (0.3222) acc1: 91.9271 (88.5845) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1626 data: 0.0696 max mem: 3539 +Test: [1930/2387] eta: 0:01:26 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2695 (0.3219) acc1: 92.7083 (88.6153) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1610 data: 0.0690 max mem: 3539 +Test: [1940/2387] eta: 0:01:24 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2746 (0.3220) acc1: 94.0104 (88.6050) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1622 data: 0.0716 max mem: 3539 +Test: [1950/2387] eta: 0:01:22 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.3492 (0.3223) acc1: 87.2396 (88.5973) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1691 data: 0.0769 max mem: 3539 +Test: [1960/2387] eta: 0:01:20 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.3374 (0.3224) acc1: 89.0625 (88.5959) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1807 data: 0.0876 max mem: 3539 +Test: [1970/2387] eta: 0:01:19 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.4078 (0.3235) acc1: 81.7708 (88.5027) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.1922 data: 0.1000 max mem: 3539 +Test: [1980/2387] eta: 0:01:17 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.5015 (0.3238) acc1: 74.2188 (88.4791) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2140 data: 0.1222 max mem: 3539 +Test: [1990/2387] eta: 0:01:15 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1989 (0.3231) acc1: 97.3958 (88.5251) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1898 data: 0.0936 max mem: 3539 +Test: [2000/2387] eta: 0:01:13 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1954 (0.3225) acc1: 97.6562 (88.5703) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1649 data: 0.0693 max mem: 3539 +Test: [2010/2387] eta: 0:01:11 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2002 (0.3219) acc1: 97.6562 (88.6141) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1755 data: 0.0839 max mem: 3539 +Test: [2020/2387] eta: 0:01:09 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2025 (0.3213) acc1: 97.1354 (88.6574) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1749 data: 0.0824 max mem: 3539 +Test: [2030/2387] eta: 0:01:07 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1980 (0.3207) acc1: 97.3958 (88.7008) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1697 data: 0.0772 max mem: 3540 +Test: [2040/2387] eta: 0:01:05 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1972 (0.3201) acc1: 97.6562 (88.7453) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1661 data: 0.0739 max mem: 3540 +Test: [2050/2387] eta: 0:01:03 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1990 (0.3196) acc1: 97.3958 (88.7841) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1680 data: 0.0754 max mem: 3540 +Test: [2060/2387] eta: 0:01:01 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2539 (0.3194) acc1: 94.7917 (88.8102) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1547 data: 0.0622 max mem: 3540 +Test: [2070/2387] eta: 0:00:59 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2636 (0.3191) acc1: 94.2708 (88.8380) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1491 data: 0.0579 max mem: 3540 +Test: [2080/2387] eta: 0:00:57 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2570 (0.3189) acc1: 94.5312 (88.8650) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1569 data: 0.0668 max mem: 3540 +Test: [2090/2387] eta: 0:00:55 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2627 (0.3186) acc1: 94.5312 (88.8933) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1655 data: 0.0744 max mem: 3540 +Test: [2100/2387] eta: 0:00:54 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2627 (0.3183) acc1: 94.7917 (88.9210) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1743 data: 0.0807 max mem: 3540 +Test: [2110/2387] eta: 0:00:52 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2612 (0.3180) acc1: 95.0521 (88.9490) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1645 data: 0.0707 max mem: 3540 +Test: [2120/2387] eta: 0:00:50 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2592 (0.3178) acc1: 95.0521 (88.9680) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.1552 data: 0.0640 max mem: 3540 +Test: [2130/2387] eta: 0:00:48 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2220 (0.3172) acc1: 96.3542 (89.0115) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1556 data: 0.0650 max mem: 3540 +Test: [2140/2387] eta: 0:00:46 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2160 (0.3169) acc1: 96.6146 (89.0355) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1584 data: 0.0654 max mem: 3540 +Test: [2150/2387] eta: 0:00:44 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2364 (0.3165) acc1: 94.7917 (89.0655) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1689 data: 0.0747 max mem: 3540 +Test: [2160/2387] eta: 0:00:42 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2050 (0.3160) acc1: 95.5729 (89.1005) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1587 data: 0.0649 max mem: 3540 +Test: [2170/2387] eta: 0:00:40 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2523 (0.3165) acc1: 92.1875 (89.0692) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.1465 data: 0.0524 max mem: 3540 +Test: [2180/2387] eta: 0:00:38 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.4790 (0.3177) acc1: 73.1771 (88.9726) acc5: 100.0000 (100.0000) acc1_neg: 54.167 acc5_neg: 100.000 time: 0.1749 data: 0.0806 max mem: 3540 +Test: [2190/2387] eta: 0:00:36 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.4579 (0.3176) acc1: 77.6042 (88.9797) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1770 data: 0.0832 max mem: 3540 +Test: [2200/2387] eta: 0:00:34 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2298 (0.3174) acc1: 95.8333 (89.0065) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1499 data: 0.0574 max mem: 3540 +Test: [2210/2387] eta: 0:00:33 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2563 (0.3171) acc1: 96.6146 (89.0385) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1542 data: 0.0626 max mem: 3540 +Test: [2220/2387] eta: 0:00:31 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2559 (0.3169) acc1: 96.8750 (89.0672) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1822 data: 0.0911 max mem: 3540 +Test: [2230/2387] eta: 0:00:29 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2559 (0.3165) acc1: 96.6146 (89.0966) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1818 data: 0.0911 max mem: 3540 +Test: [2240/2387] eta: 0:00:27 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2558 (0.3163) acc1: 96.6146 (89.1258) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1687 data: 0.0769 max mem: 3540 +Test: [2250/2387] eta: 0:00:25 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2764 (0.3166) acc1: 92.4479 (89.1067) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.1749 data: 0.0835 max mem: 3540 +Test: [2260/2387] eta: 0:00:23 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.3508 (0.3165) acc1: 89.3229 (89.1223) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1842 data: 0.0937 max mem: 3541 +Test: [2270/2387] eta: 0:00:21 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2055 (0.3159) acc1: 97.6562 (89.1656) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1850 data: 0.0940 max mem: 3541 +Test: [2280/2387] eta: 0:00:19 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1820 (0.3155) acc1: 97.9167 (89.1906) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1679 data: 0.0771 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1839 (0.3149) acc1: 97.3958 (89.2287) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1636 data: 0.0730 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.1918 (0.3145) acc1: 98.4375 (89.2628) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1655 data: 0.0746 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.2510 (0.3145) acc1: 95.5729 (89.2677) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1624 data: 0.0712 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.3749 (0.3148) acc1: 85.1562 (89.2436) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.1630 data: 0.0717 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.3880 (0.3152) acc1: 84.8958 (89.2286) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1571 data: 0.0660 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.3356 (0.3152) acc1: 87.5000 (89.2323) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1447 data: 0.0536 max mem: 3541 +Test: [2350/2387] eta: 0:00:06 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.3323 (0.3156) acc1: 87.7604 (89.2124) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1405 data: 0.0498 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.3954 (0.3161) acc1: 82.2917 (89.1798) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1458 data: 0.0559 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.3878 (0.3164) acc1: 83.5938 (89.1592) acc5: 100.0000 (100.0000) acc1_neg: 71.094 acc5_neg: 100.000 time: 0.1475 data: 0.0579 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.5089 (0.3176) acc1: 74.7396 (89.0647) acc5: 100.0000 (100.0000) acc1_neg: 71.354 acc5_neg: 100.000 time: 0.1487 data: 0.0587 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 95.434 acc5_pos: 100.000 loss: 0.5314 (0.3178) acc1: 71.3542 (89.0398) acc5: 100.0000 (100.0000) acc1_neg: 93.023 acc5_neg: 100.000 time: 0.1373 data: 0.0500 max mem: 3541 +Test: Total time: 0:07:22 (0.1853 s / it) +* Overall Acc@1 89.040 Acc@5 100.000 + target == real Acc@1 76.851 Acc@5 100.000 + target == fake Acc@1 89.896 Acc@5 100.000 + loss 0.318 + PR-AUC 0.9942 ROC-AUC 0.9262 EER 15.08% (thr=0.5806) +Accuracy of the network on the 916267 test images: 89.0% +[rank0]:[W726 04:35:41.726759559 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:36:00.110000 3155911 torch/distributed/run.py:766] +W0726 04:36:00.110000 3155911 torch/distributed/run.py:766] ***************************************** +W0726 04:36:00.110000 3155911 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:36:00.110000 3155911 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:36:26.169539253 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:36:26.195979511 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:36:27.781512201 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 04:36:27.856949376 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W726 04:36:27.859418623 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:36:27.908218294 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:36:27.045990576 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:36:27.066555375 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n16_seed7_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed7_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:02:03 acc1_pos: 71.615 acc5_pos: 100.000 loss: 0.5537 (0.5537) acc1: 71.6146 (71.6146) acc5: 100.0000 (100.0000) time: 7.2026 data: 5.4929 max mem: 3531 +Test: [ 10/517] eta: 0:06:13 acc1_pos: 54.167 acc5_pos: 100.000 loss: 0.6994 (0.6622) acc1: 61.9792 (61.3163) acc5: 100.0000 (100.0000) time: 0.7367 data: 0.4997 max mem: 3531 +Test: [ 20/517] eta: 0:03:35 acc1_pos: 69.531 acc5_pos: 100.000 loss: 0.6681 (0.6576) acc1: 61.9792 (62.2520) acc5: 100.0000 (100.0000) time: 0.0948 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:02:51 acc1_pos: 34.635 acc5_pos: 100.000 loss: 0.6835 (0.6740) acc1: 59.3750 (60.5511) acc5: 100.0000 (100.0000) time: 0.1402 data: 0.0410 max mem: 3531 +Test: [ 40/517] eta: 0:02:40 acc1_pos: 65.104 acc5_pos: 100.000 loss: 0.7523 (0.6970) acc1: 50.7812 (58.4604) acc5: 100.0000 (100.0000) time: 0.2352 data: 0.1396 max mem: 3531 +Test: [ 50/517] eta: 0:02:26 acc1_pos: 90.367 acc5_pos: 100.000 loss: 0.6971 (0.6891) acc1: 54.9479 (59.2116) acc5: 100.0000 (100.0000) acc1_neg: 62.760 acc5_neg: 100.000 time: 0.2547 data: 0.1634 max mem: 3531 +Test: [ 60/517] eta: 0:02:18 acc1_pos: 90.367 acc5_pos: 100.000 loss: 0.6290 (0.6798) acc1: 63.5417 (60.0794) acc5: 100.0000 (100.0000) acc1_neg: 71.094 acc5_neg: 100.000 time: 0.2347 data: 0.1403 max mem: 3531 +Test: [ 70/517] eta: 0:02:12 acc1_pos: 90.367 acc5_pos: 100.000 loss: 0.5512 (0.6551) acc1: 72.3958 (62.7201) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.2496 data: 0.1543 max mem: 3531 +Test: [ 80/517] eta: 0:02:05 acc1_pos: 90.367 acc5_pos: 100.000 loss: 0.4989 (0.6429) acc1: 73.4375 (63.7442) acc5: 100.0000 (100.0000) acc1_neg: 64.062 acc5_neg: 100.000 time: 0.2349 data: 0.1416 max mem: 3531 +Test: [ 90/517] eta: 0:02:01 acc1_pos: 90.367 acc5_pos: 100.000 loss: 0.5323 (0.6296) acc1: 70.0521 (65.1700) acc5: 100.0000 (100.0000) acc1_neg: 72.135 acc5_neg: 100.000 time: 0.2454 data: 0.1510 max mem: 3531 +Test: [100/517] eta: 0:01:55 acc1_pos: 90.367 acc5_pos: 100.000 loss: 0.5033 (0.6182) acc1: 79.1667 (66.5120) acc5: 100.0000 (100.0000) acc1_neg: 74.219 acc5_neg: 100.000 time: 0.2450 data: 0.1516 max mem: 3531 +Test: [110/517] eta: 0:01:52 acc1_pos: 90.367 acc5_pos: 100.000 loss: 0.4735 (0.6029) acc1: 82.8125 (67.8843) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.2344 data: 0.1392 max mem: 3531 +Test: [120/517] eta: 0:01:47 acc1_pos: 90.367 acc5_pos: 100.000 loss: 0.4352 (0.5897) acc1: 86.4583 (69.4344) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.2403 data: 0.1445 max mem: 3531 +Test: [130/517] eta: 0:01:44 acc1_pos: 90.367 acc5_pos: 100.000 loss: 0.4314 (0.5758) acc1: 87.7604 (71.0500) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2399 data: 0.1462 max mem: 3531 +Test: [140/517] eta: 0:01:41 acc1_pos: 90.367 acc5_pos: 100.000 loss: 0.4154 (0.5658) acc1: 90.1042 (72.3293) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.2444 data: 0.1516 max mem: 3531 +Test: [150/517] eta: 0:01:37 acc1_pos: 90.367 acc5_pos: 100.000 loss: 0.4320 (0.5565) acc1: 87.5000 (73.4479) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.2399 data: 0.1475 max mem: 3531 +Test: [160/517] eta: 0:01:34 acc1_pos: 90.367 acc5_pos: 100.000 loss: 0.4450 (0.5507) acc1: 82.5521 (73.8451) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.2399 data: 0.1449 max mem: 3531 +Test: [170/517] eta: 0:01:31 acc1_pos: 90.367 acc5_pos: 100.000 loss: 0.4307 (0.5435) acc1: 82.0312 (74.4259) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.2499 data: 0.1550 max mem: 3531 +Test: [180/517] eta: 0:01:28 acc1_pos: 90.367 acc5_pos: 100.000 loss: 0.4366 (0.5399) acc1: 83.0729 (74.9007) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.2399 data: 0.1456 max mem: 3531 +Test: [190/517] eta: 0:01:25 acc1_pos: 90.367 acc5_pos: 100.000 loss: 0.3554 (0.5288) acc1: 92.4479 (75.9817) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2249 data: 0.1281 max mem: 3531 +Test: [200/517] eta: 0:01:21 acc1_pos: 90.367 acc5_pos: 100.000 loss: 0.3553 (0.5232) acc1: 92.4479 (76.5703) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.2304 data: 0.1339 max mem: 3531 +Test: [210/517] eta: 0:01:19 acc1_pos: 90.367 acc5_pos: 100.000 loss: 0.4670 (0.5228) acc1: 82.0312 (76.6452) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.2399 data: 0.1456 max mem: 3532 +Test: [220/517] eta: 0:01:16 acc1_pos: 90.367 acc5_pos: 100.000 loss: 0.5257 (0.5226) acc1: 76.5625 (76.6485) acc5: 100.0000 (100.0000) acc1_neg: 74.740 acc5_neg: 100.000 time: 0.2444 data: 0.1495 max mem: 3532 +Test: [230/517] eta: 0:01:13 acc1_pos: 90.367 acc5_pos: 100.000 loss: 0.5488 (0.5258) acc1: 71.0938 (76.2232) acc5: 100.0000 (100.0000) acc1_neg: 65.104 acc5_neg: 100.000 time: 0.2399 data: 0.1451 max mem: 3532 +Test: [240/517] eta: 0:01:10 acc1_pos: 90.367 acc5_pos: 100.000 loss: 0.5780 (0.5267) acc1: 69.2708 (76.0427) acc5: 100.0000 (100.0000) acc1_neg: 34.115 acc5_neg: 100.000 time: 0.2399 data: 0.1428 max mem: 3532 +Test: [250/517] eta: 0:01:08 acc1_pos: 90.367 acc5_pos: 100.000 loss: 0.4791 (0.5238) acc1: 76.8229 (76.4598) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.2399 data: 0.1434 max mem: 3532 +Test: [260/517] eta: 0:01:05 acc1_pos: 90.367 acc5_pos: 100.000 loss: 0.4149 (0.5199) acc1: 90.8854 (76.9856) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.2399 data: 0.1452 max mem: 3532 +Test: [270/517] eta: 0:01:02 acc1_pos: 90.367 acc5_pos: 100.000 loss: 0.4558 (0.5183) acc1: 86.4583 (77.2544) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.2499 data: 0.1557 max mem: 3532 +Test: [280/517] eta: 0:01:00 acc1_pos: 90.367 acc5_pos: 100.000 loss: 0.4724 (0.5168) acc1: 85.4167 (77.4624) acc5: 100.0000 (100.0000) acc1_neg: 73.958 acc5_neg: 100.000 time: 0.2499 data: 0.1556 max mem: 3532 +Test: [290/517] eta: 0:00:57 acc1_pos: 90.367 acc5_pos: 100.000 loss: 0.4764 (0.5169) acc1: 82.8125 (77.5165) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.2449 data: 0.1481 max mem: 3532 +Test: [300/517] eta: 0:00:55 acc1_pos: 90.367 acc5_pos: 100.000 loss: 0.5007 (0.5168) acc1: 79.1667 (77.5523) acc5: 100.0000 (100.0000) acc1_neg: 62.760 acc5_neg: 100.000 time: 0.2449 data: 0.1495 max mem: 3532 +Test: [310/517] eta: 0:00:52 acc1_pos: 90.367 acc5_pos: 100.000 loss: 0.5731 (0.5204) acc1: 70.8333 (77.0708) acc5: 100.0000 (100.0000) acc1_neg: 56.250 acc5_neg: 100.000 time: 0.2399 data: 0.1451 max mem: 3532 +Test: [320/517] eta: 0:00:49 acc1_pos: 90.367 acc5_pos: 100.000 loss: 0.5903 (0.5191) acc1: 65.8854 (77.1612) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.2400 data: 0.1448 max mem: 3532 +Test: [330/517] eta: 0:00:47 acc1_pos: 90.367 acc5_pos: 100.000 loss: 0.3843 (0.5150) acc1: 89.5833 (77.5585) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.2399 data: 0.1465 max mem: 3532 +Test: [340/517] eta: 0:00:44 acc1_pos: 90.367 acc5_pos: 100.000 loss: 0.3669 (0.5123) acc1: 92.4479 (77.8264) acc5: 100.0000 (100.0000) acc1_neg: 69.010 acc5_neg: 100.000 time: 0.2448 data: 0.1528 max mem: 3532 +Test: [350/517] eta: 0:00:42 acc1_pos: 90.367 acc5_pos: 100.000 loss: 0.5361 (0.5162) acc1: 74.7396 (77.3475) acc5: 100.0000 (100.0000) acc1_neg: 33.594 acc5_neg: 100.000 time: 0.2399 data: 0.1454 max mem: 3532 +Test: [360/517] eta: 0:00:39 acc1_pos: 90.367 acc5_pos: 100.000 loss: 0.6115 (0.5170) acc1: 64.3229 (77.2334) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.2349 data: 0.1390 max mem: 3532 +Test: [370/517] eta: 0:00:36 acc1_pos: 90.367 acc5_pos: 100.000 loss: 0.3674 (0.5122) acc1: 93.4896 (77.7305) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2399 data: 0.1415 max mem: 3532 +Test: [380/517] eta: 0:00:34 acc1_pos: 90.367 acc5_pos: 100.000 loss: 0.3152 (0.5069) acc1: 96.6146 (78.1961) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2349 data: 0.1369 max mem: 3532 +Test: [390/517] eta: 0:00:31 acc1_pos: 90.367 acc5_pos: 100.000 loss: 0.3057 (0.5021) acc1: 96.6146 (78.6385) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2449 data: 0.1515 max mem: 3532 +Test: [400/517] eta: 0:00:29 acc1_pos: 90.367 acc5_pos: 100.000 loss: 0.3048 (0.4983) acc1: 96.0938 (78.9641) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2449 data: 0.1473 max mem: 3532 +Test: [410/517] eta: 0:00:26 acc1_pos: 90.367 acc5_pos: 100.000 loss: 0.3878 (0.4959) acc1: 92.7083 (79.1825) acc5: 100.0000 (100.0000) acc1_neg: 79.427 acc5_neg: 100.000 time: 0.2299 data: 0.1312 max mem: 3532 +Test: [420/517] eta: 0:00:24 acc1_pos: 90.367 acc5_pos: 100.000 loss: 0.4718 (0.4964) acc1: 81.7708 (79.1091) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.2399 data: 0.1453 max mem: 3532 +Test: [430/517] eta: 0:00:21 acc1_pos: 90.367 acc5_pos: 100.000 loss: 0.4543 (0.4949) acc1: 81.2500 (79.2198) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.2499 data: 0.1567 max mem: 3532 +Test: [440/517] eta: 0:00:19 acc1_pos: 90.367 acc5_pos: 100.000 loss: 0.4297 (0.4934) acc1: 84.3750 (79.3899) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.2499 data: 0.1574 max mem: 3533 +Test: [450/517] eta: 0:00:16 acc1_pos: 90.367 acc5_pos: 100.000 loss: 0.3932 (0.4906) acc1: 89.3229 (79.6944) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2449 data: 0.1499 max mem: 3533 +Test: [460/517] eta: 0:00:14 acc1_pos: 90.367 acc5_pos: 100.000 loss: 0.3669 (0.4884) acc1: 93.2292 (79.9524) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.2299 data: 0.1314 max mem: 3533 +Test: [470/517] eta: 0:00:11 acc1_pos: 90.367 acc5_pos: 100.000 loss: 0.3858 (0.4872) acc1: 91.6667 (80.0906) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.2250 data: 0.1277 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 90.367 acc5_pos: 100.000 loss: 0.4142 (0.4853) acc1: 89.0625 (80.3426) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2299 data: 0.1363 max mem: 3533 +Test: [490/517] eta: 0:00:06 acc1_pos: 90.367 acc5_pos: 100.000 loss: 0.4314 (0.4841) acc1: 88.2812 (80.5011) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.2338 data: 0.1410 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 90.367 acc5_pos: 100.000 loss: 0.4616 (0.4850) acc1: 86.1979 (80.4173) acc5: 100.0000 (100.0000) acc1_neg: 65.885 acc5_neg: 100.000 time: 0.2389 data: 0.1456 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 90.367 acc5_pos: 100.000 loss: 0.4996 (0.4849) acc1: 77.8646 (80.4698) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.2201 data: 0.1274 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 90.367 acc5_pos: 100.000 loss: 0.4947 (0.4855) acc1: 78.6458 (80.4210) acc5: 100.0000 (100.0000) acc1_neg: 68.551 acc5_neg: 100.000 time: 0.1617 data: 0.0684 max mem: 3533 +Test: Total time: 0:02:07 (0.2461 s / it) +* Overall Acc@1 80.421 Acc@5 100.000 + target == real Acc@1 58.890 Acc@5 100.000 + target == fake Acc@1 82.655 Acc@5 100.000 + loss 0.485 + PR-AUC 0.9653 ROC-AUC 0.7820 EER 29.10% (thr=0.5762) +Accuracy of the network on the 198427 test images: 80.4% +[rank0]:[W726 04:38:50.713571269 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 04:39:09.333000 3156485 torch/distributed/run.py:766] +W0726 04:39:09.333000 3156485 torch/distributed/run.py:766] ***************************************** +W0726 04:39:09.333000 3156485 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:39:09.333000 3156485 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:39:37.174684621 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:39:37.450977833 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:39:38.645939459 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:39:38.717246327 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:39:38.818341866 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +[rank2]:[W726 04:39:38.004278320 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:39:38.004330595 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:39:38.006161254 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n16_seed7_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed7_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:32:24 acc1_pos: 53.646 acc5_pos: 100.000 loss: 0.7039 (0.7039) acc1: 53.6458 (53.6458) acc5: 100.0000 (100.0000) time: 4.9979 data: 3.4599 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/389] eta: 0:03:23 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2557 (0.3551) acc1: 99.7396 (88.9915) acc5: 100.0000 (100.0000) time: 0.5359 data: 0.3148 max mem: 3531 +Test: [ 20/389] eta: 0:01:59 acc1_pos: 96.354 acc5_pos: 100.000 loss: 0.2804 (0.3213) acc1: 98.6979 (93.5764) acc5: 100.0000 (100.0000) time: 0.0898 data: 0.0003 max mem: 3531 +Test: [ 30/389] eta: 0:01:29 acc1_pos: 99.740 acc5_pos: 100.000 loss: 0.2880 (0.3084) acc1: 98.4375 (94.6573) acc5: 100.0000 (100.0000) time: 0.0915 data: 0.0003 max mem: 3531 +Test: [ 40/389] eta: 0:01:15 acc1_pos: 86.198 acc5_pos: 100.000 loss: 0.3361 (0.3212) acc1: 95.0521 (93.8262) acc5: 100.0000 (100.0000) time: 0.1016 data: 0.0061 max mem: 3531 +Test: [ 50/389] eta: 0:01:09 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2465 (0.3024) acc1: 96.8750 (94.7661) acc5: 100.0000 (100.0000) time: 0.1349 data: 0.0379 max mem: 3531 +Test: [ 60/389] eta: 0:01:04 acc1_pos: 97.135 acc5_pos: 100.000 loss: 0.1957 (0.2948) acc1: 99.7396 (95.4064) acc5: 100.0000 (100.0000) time: 0.1549 data: 0.0599 max mem: 3531 +Test: [ 70/389] eta: 0:00:59 acc1_pos: 86.198 acc5_pos: 100.000 loss: 0.3346 (0.3182) acc1: 96.8750 (93.7793) acc5: 100.0000 (100.0000) time: 0.1425 data: 0.0503 max mem: 3531 +Test: [ 80/389] eta: 0:00:55 acc1_pos: 99.479 acc5_pos: 100.000 loss: 0.4134 (0.3311) acc1: 87.2396 (93.1327) acc5: 100.0000 (100.0000) time: 0.1314 data: 0.0410 max mem: 3531 +Test: [ 90/389] eta: 0:00:52 acc1_pos: 98.958 acc5_pos: 100.000 loss: 0.3003 (0.3247) acc1: 97.1354 (93.7386) acc5: 100.0000 (100.0000) time: 0.1294 data: 0.0363 max mem: 3531 +Test: [100/389] eta: 0:00:50 acc1_pos: 98.177 acc5_pos: 100.000 loss: 0.3003 (0.3285) acc1: 98.1771 (93.6056) acc5: 100.0000 (100.0000) time: 0.1584 data: 0.0648 max mem: 3531 +Test: [110/389] eta: 0:00:48 acc1_pos: 91.406 acc5_pos: 100.000 loss: 0.3219 (0.3400) acc1: 92.9688 (92.4808) acc5: 100.0000 (100.0000) time: 0.1779 data: 0.0839 max mem: 3531 +Test: [120/389] eta: 0:00:46 acc1_pos: 97.917 acc5_pos: 100.000 loss: 0.3218 (0.3371) acc1: 94.5312 (92.7514) acc5: 100.0000 (100.0000) time: 0.1617 data: 0.0672 max mem: 3531 +Test: [130/389] eta: 0:00:45 acc1_pos: 69.325 acc5_pos: 100.000 loss: 0.4322 (0.3847) acc1: 88.2812 (88.9154) acc5: 100.0000 (100.0000) acc1_neg: 50.521 acc5_neg: 100.000 time: 0.1727 data: 0.0800 max mem: 3531 +Test: [140/389] eta: 0:00:43 acc1_pos: 69.325 acc5_pos: 100.000 loss: 0.9113 (0.4330) acc1: 29.9479 (84.1608) acc5: 100.0000 (100.0000) acc1_neg: 0.000 acc5_neg: 100.000 time: 0.1908 data: 0.0977 max mem: 3531 +Test: [150/389] eta: 0:00:41 acc1_pos: 69.325 acc5_pos: 100.000 loss: 1.0734 (0.4808) acc1: 16.9271 (80.1566) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.1771 data: 0.0852 max mem: 3531 +Test: [160/389] eta: 0:00:40 acc1_pos: 69.325 acc5_pos: 100.000 loss: 1.0621 (0.5082) acc1: 9.3750 (77.0397) acc5: 100.0000 (100.0000) acc1_neg: 22.396 acc5_neg: 100.000 time: 0.1780 data: 0.0868 max mem: 3531 +Test: [170/389] eta: 0:00:38 acc1_pos: 69.325 acc5_pos: 100.000 loss: 0.7638 (0.5317) acc1: 41.4062 (74.3924) acc5: 100.0000 (100.0000) acc1_neg: 10.938 acc5_neg: 100.000 time: 0.1860 data: 0.0944 max mem: 3531 +Test: [180/389] eta: 0:00:36 acc1_pos: 69.325 acc5_pos: 100.000 loss: 0.8382 (0.5624) acc1: 30.2083 (71.4707) acc5: 100.0000 (100.0000) acc1_neg: 27.083 acc5_neg: 100.000 time: 0.1804 data: 0.0893 max mem: 3531 +Test: [190/389] eta: 0:00:35 acc1_pos: 69.325 acc5_pos: 100.000 loss: 1.0613 (0.5891) acc1: 15.8854 (69.1263) acc5: 100.0000 (100.0000) acc1_neg: 0.000 acc5_neg: 100.000 time: 0.1796 data: 0.0891 max mem: 3531 +Test: [200/389] eta: 0:00:33 acc1_pos: 69.325 acc5_pos: 100.000 loss: 1.2161 (0.6161) acc1: 8.5938 (66.8791) acc5: 100.0000 (100.0000) acc1_neg: 2.083 acc5_neg: 100.000 time: 0.1847 data: 0.0946 max mem: 3531 +Test: [210/389] eta: 0:00:31 acc1_pos: 69.325 acc5_pos: 100.000 loss: 0.8064 (0.6305) acc1: 16.6667 (65.3263) acc5: 100.0000 (100.0000) acc1_neg: 38.802 acc5_neg: 100.000 time: 0.1645 data: 0.0743 max mem: 3532 +Test: [220/389] eta: 0:00:29 acc1_pos: 69.325 acc5_pos: 100.000 loss: 0.7880 (0.6417) acc1: 35.9375 (63.7703) acc5: 100.0000 (100.0000) acc1_neg: 12.240 acc5_neg: 100.000 time: 0.1552 data: 0.0651 max mem: 3532 +Test: [230/389] eta: 0:00:27 acc1_pos: 69.325 acc5_pos: 100.000 loss: 0.7965 (0.6575) acc1: 35.9375 (62.8021) acc5: 100.0000 (100.0000) acc1_neg: 71.094 acc5_neg: 100.000 time: 0.1745 data: 0.0833 max mem: 3532 +Test: [240/389] eta: 0:00:26 acc1_pos: 69.325 acc5_pos: 100.000 loss: 1.0094 (0.6725) acc1: 21.8750 (61.4519) acc5: 100.0000 (100.0000) acc1_neg: 15.365 acc5_neg: 100.000 time: 0.1725 data: 0.0797 max mem: 3532 +Test: [250/389] eta: 0:00:24 acc1_pos: 69.325 acc5_pos: 100.000 loss: 0.9232 (0.6819) acc1: 26.8229 (60.0452) acc5: 100.0000 (100.0000) acc1_neg: 32.031 acc5_neg: 100.000 time: 0.1662 data: 0.0732 max mem: 3532 +Test: [260/389] eta: 0:00:22 acc1_pos: 69.325 acc5_pos: 100.000 loss: 0.9104 (0.6929) acc1: 27.0833 (58.6895) acc5: 100.0000 (100.0000) acc1_neg: 11.458 acc5_neg: 100.000 time: 0.1659 data: 0.0721 max mem: 3532 +Test: [270/389] eta: 0:00:20 acc1_pos: 69.325 acc5_pos: 100.000 loss: 1.0493 (0.7156) acc1: 10.6771 (56.8583) acc5: 100.0000 (100.0000) acc1_neg: 8.073 acc5_neg: 100.000 time: 0.1667 data: 0.0721 max mem: 3532 +Test: [280/389] eta: 0:00:18 acc1_pos: 69.325 acc5_pos: 100.000 loss: 1.1705 (0.7313) acc1: 5.2083 (55.5105) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1649 data: 0.0717 max mem: 3532 +Test: [290/389] eta: 0:00:17 acc1_pos: 69.325 acc5_pos: 100.000 loss: 0.9929 (0.7317) acc1: 15.3646 (55.4410) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.1576 data: 0.0653 max mem: 3532 +Test: [300/389] eta: 0:00:15 acc1_pos: 69.325 acc5_pos: 100.000 loss: 0.8536 (0.7371) acc1: 24.4792 (54.3561) acc5: 100.0000 (100.0000) acc1_neg: 24.479 acc5_neg: 100.000 time: 0.1676 data: 0.0744 max mem: 3532 +Test: [310/389] eta: 0:00:13 acc1_pos: 69.325 acc5_pos: 100.000 loss: 0.9778 (0.7485) acc1: 20.5729 (53.2431) acc5: 100.0000 (100.0000) acc1_neg: 1.042 acc5_neg: 100.000 time: 0.1816 data: 0.0887 max mem: 3532 +Test: [320/389] eta: 0:00:11 acc1_pos: 69.325 acc5_pos: 100.000 loss: 1.0132 (0.7545) acc1: 20.3125 (52.7981) acc5: 100.0000 (100.0000) acc1_neg: 64.844 acc5_neg: 100.000 time: 0.1679 data: 0.0765 max mem: 3532 +Test: [330/389] eta: 0:00:10 acc1_pos: 69.325 acc5_pos: 100.000 loss: 1.2693 (0.7709) acc1: 3.9062 (51.3878) acc5: 100.0000 (100.0000) acc1_neg: 7.031 acc5_neg: 100.000 time: 0.1533 data: 0.0614 max mem: 3532 +Test: [340/389] eta: 0:00:08 acc1_pos: 69.325 acc5_pos: 100.000 loss: 0.8474 (0.7692) acc1: 31.7708 (51.5572) acc5: 100.0000 (100.0000) acc1_neg: 69.792 acc5_neg: 100.000 time: 0.1598 data: 0.0680 max mem: 3532 +Test: [350/389] eta: 0:00:06 acc1_pos: 69.325 acc5_pos: 100.000 loss: 0.7937 (0.7771) acc1: 42.1875 (51.1003) acc5: 100.0000 (100.0000) acc1_neg: 5.990 acc5_neg: 100.000 time: 0.1703 data: 0.0794 max mem: 3532 +Test: [360/389] eta: 0:00:04 acc1_pos: 69.325 acc5_pos: 100.000 loss: 1.1674 (0.7879) acc1: 6.7708 (49.8543) acc5: 100.0000 (100.0000) acc1_neg: 7.031 acc5_neg: 100.000 time: 0.1549 data: 0.0641 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 69.325 acc5_pos: 100.000 loss: 1.1674 (0.7931) acc1: 7.0312 (49.4321) acc5: 100.0000 (100.0000) acc1_neg: 18.490 acc5_neg: 100.000 time: 0.1536 data: 0.0634 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 69.325 acc5_pos: 100.000 loss: 1.0278 (0.8019) acc1: 20.5729 (48.6740) acc5: 100.0000 (100.0000) acc1_neg: 34.896 acc5_neg: 100.000 time: 0.1667 data: 0.0761 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 69.325 acc5_pos: 100.000 loss: 0.9282 (0.8015) acc1: 28.9062 (48.6154) acc5: 100.0000 (100.0000) acc1_neg: 45.902 acc5_neg: 100.000 time: 0.1377 data: 0.0431 max mem: 3532 +Test: Total time: 0:01:06 (0.1704 s / it) +* Overall Acc@1 48.615 Acc@5 100.000 + target == real Acc@1 92.480 Acc@5 100.000 + target == fake Acc@1 28.226 Acc@5 100.000 + loss 0.801 + PR-AUC 0.8366 ROC-AUC 0.7334 EER 32.53% (thr=0.2900) +Accuracy of the network on the 149358 test images: 48.6% +[rank0]:[W726 04:41:02.615920497 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:41:20.795000 3157045 torch/distributed/run.py:766] +W0726 04:41:20.795000 3157045 torch/distributed/run.py:766] ***************************************** +W0726 04:41:20.795000 3157045 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:41:20.795000 3157045 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:41:48.180290772 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:41:49.155419662 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W726 04:41:49.155831557 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:41:49.191528919 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:41:49.244507513 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +| distributed init (rank 4): env:// +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:41:49.460571751 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank4]:[W726 04:41:49.460594554 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W726 04:41:49.460616335 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n16_seed7_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed7_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:36:07 acc1_pos: 6.250 acc5_pos: 100.000 loss: 1.3854 (1.3854) acc1: 6.2500 (6.2500) acc5: 100.0000 (100.0000) time: 6.5091 data: 4.6943 max mem: 3531 +Test: [ 10/333] eta: 0:03:37 acc1_pos: 15.625 acc5_pos: 100.000 loss: 1.2192 (1.2280) acc1: 18.4896 (18.9867) acc5: 100.0000 (100.0000) time: 0.6742 data: 0.4271 max mem: 3531 +Test: [ 20/333] eta: 0:02:05 acc1_pos: 15.885 acc5_pos: 100.000 loss: 1.2179 (1.2271) acc1: 19.5312 (20.3993) acc5: 100.0000 (100.0000) time: 0.0953 data: 0.0004 max mem: 3531 +Test: [ 30/333] eta: 0:01:33 acc1_pos: 73.958 acc5_pos: 100.000 loss: 1.0012 (1.0309) acc1: 30.7292 (35.1815) acc5: 100.0000 (100.0000) time: 0.1070 data: 0.0109 max mem: 3531 +Test: [ 40/333] eta: 0:01:20 acc1_pos: 71.354 acc5_pos: 100.000 loss: 0.5606 (0.9181) acc1: 68.4896 (43.6039) acc5: 100.0000 (100.0000) time: 0.1449 data: 0.0499 max mem: 3531 +Test: [ 50/333] eta: 0:01:11 acc1_pos: 67.448 acc5_pos: 100.000 loss: 0.5606 (0.8508) acc1: 71.3542 (48.8409) acc5: 100.0000 (100.0000) time: 0.1629 data: 0.0675 max mem: 3531 +Test: [ 60/333] eta: 0:01:04 acc1_pos: 70.312 acc5_pos: 100.000 loss: 0.5750 (0.8068) acc1: 70.3125 (52.3736) acc5: 100.0000 (100.0000) time: 0.1503 data: 0.0588 max mem: 3531 +Test: [ 70/333] eta: 0:00:58 acc1_pos: 58.854 acc5_pos: 100.000 loss: 0.5750 (0.7732) acc1: 70.3125 (55.0543) acc5: 100.0000 (100.0000) time: 0.1511 data: 0.0599 max mem: 3531 +Test: [ 80/333] eta: 0:00:56 acc1_pos: 62.760 acc5_pos: 100.000 loss: 0.5721 (0.7492) acc1: 69.0104 (56.8512) acc5: 100.0000 (100.0000) time: 0.1807 data: 0.0853 max mem: 3531 +Test: [ 90/333] eta: 0:00:53 acc1_pos: 82.812 acc5_pos: 100.000 loss: 0.5389 (0.7228) acc1: 72.3958 (59.1203) acc5: 100.0000 (100.0000) time: 0.2042 data: 0.1096 max mem: 3531 +Test: [100/333] eta: 0:00:49 acc1_pos: 80.729 acc5_pos: 100.000 loss: 0.5081 (0.6984) acc1: 79.1667 (61.1103) acc5: 100.0000 (100.0000) time: 0.1837 data: 0.0924 max mem: 3531 +Test: [110/333] eta: 0:00:47 acc1_pos: 75.781 acc5_pos: 100.000 loss: 0.4972 (0.6799) acc1: 78.1250 (62.6032) acc5: 100.0000 (100.0000) time: 0.1844 data: 0.0930 max mem: 3531 +Test: [120/333] eta: 0:00:45 acc1_pos: 78.906 acc5_pos: 100.000 loss: 0.4590 (0.6633) acc1: 78.9062 (63.9592) acc5: 100.0000 (100.0000) time: 0.2011 data: 0.1079 max mem: 3531 +Test: [130/333] eta: 0:00:43 acc1_pos: 84.896 acc5_pos: 100.000 loss: 0.4332 (0.6486) acc1: 80.2083 (65.1360) acc5: 100.0000 (100.0000) time: 0.2103 data: 0.1161 max mem: 3531 +Test: [140/333] eta: 0:00:40 acc1_pos: 84.375 acc5_pos: 100.000 loss: 0.4192 (0.6332) acc1: 83.5938 (66.4450) acc5: 100.0000 (100.0000) time: 0.2011 data: 0.1072 max mem: 3531 +Test: [150/333] eta: 0:00:37 acc1_pos: 77.604 acc5_pos: 100.000 loss: 0.4423 (0.6238) acc1: 82.5521 (67.2479) acc5: 100.0000 (100.0000) time: 0.1745 data: 0.0812 max mem: 3531 +Test: [160/333] eta: 0:00:35 acc1_pos: 75.658 acc5_pos: 100.000 loss: 0.4718 (0.6259) acc1: 77.6042 (67.0856) acc5: 100.0000 (100.0000) acc1_neg: 34.115 acc5_neg: 100.000 time: 0.1676 data: 0.0735 max mem: 3531 +Test: [170/333] eta: 0:00:33 acc1_pos: 75.658 acc5_pos: 100.000 loss: 0.8045 (0.6391) acc1: 47.6562 (65.8093) acc5: 100.0000 (100.0000) acc1_neg: 50.521 acc5_neg: 100.000 time: 0.1749 data: 0.0787 max mem: 3531 +Test: [180/333] eta: 0:00:31 acc1_pos: 75.658 acc5_pos: 100.000 loss: 0.7761 (0.6433) acc1: 49.7396 (65.2121) acc5: 100.0000 (100.0000) acc1_neg: 57.812 acc5_neg: 100.000 time: 0.1874 data: 0.0945 max mem: 3531 +Test: [190/333] eta: 0:00:29 acc1_pos: 75.658 acc5_pos: 100.000 loss: 0.6939 (0.6438) acc1: 57.0312 (64.9024) acc5: 100.0000 (100.0000) acc1_neg: 57.812 acc5_neg: 100.000 time: 0.2001 data: 0.1102 max mem: 3531 +Test: [200/333] eta: 0:00:26 acc1_pos: 75.658 acc5_pos: 100.000 loss: 0.6860 (0.6457) acc1: 57.8125 (64.5730) acc5: 100.0000 (100.0000) acc1_neg: 55.469 acc5_neg: 100.000 time: 0.1825 data: 0.0926 max mem: 3531 +Test: [210/333] eta: 0:00:24 acc1_pos: 75.658 acc5_pos: 100.000 loss: 0.7121 (0.6490) acc1: 56.2500 (64.1884) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.1722 data: 0.0800 max mem: 3532 +Test: [220/333] eta: 0:00:22 acc1_pos: 75.658 acc5_pos: 100.000 loss: 0.7240 (0.6551) acc1: 54.9479 (63.5605) acc5: 100.0000 (100.0000) acc1_neg: 52.083 acc5_neg: 100.000 time: 0.2060 data: 0.1116 max mem: 3532 +Test: [230/333] eta: 0:00:20 acc1_pos: 75.658 acc5_pos: 100.000 loss: 0.6852 (0.6552) acc1: 60.1562 (63.5552) acc5: 100.0000 (100.0000) acc1_neg: 68.750 acc5_neg: 100.000 time: 0.2079 data: 0.1151 max mem: 3532 +Test: [240/333] eta: 0:00:18 acc1_pos: 75.658 acc5_pos: 100.000 loss: 0.6960 (0.6618) acc1: 58.8542 (62.7907) acc5: 100.0000 (100.0000) acc1_neg: 38.021 acc5_neg: 100.000 time: 0.1903 data: 0.0993 max mem: 3532 +Test: [250/333] eta: 0:00:16 acc1_pos: 75.658 acc5_pos: 100.000 loss: 0.8203 (0.6661) acc1: 42.7083 (62.2759) acc5: 100.0000 (100.0000) acc1_neg: 53.906 acc5_neg: 100.000 time: 0.1882 data: 0.0972 max mem: 3532 +Test: [260/333] eta: 0:00:14 acc1_pos: 75.658 acc5_pos: 100.000 loss: 0.7064 (0.6669) acc1: 54.9479 (62.1707) acc5: 100.0000 (100.0000) acc1_neg: 61.458 acc5_neg: 100.000 time: 0.1832 data: 0.0911 max mem: 3532 +Test: [270/333] eta: 0:00:12 acc1_pos: 75.658 acc5_pos: 100.000 loss: 0.6701 (0.6667) acc1: 61.4583 (62.1839) acc5: 100.0000 (100.0000) acc1_neg: 58.854 acc5_neg: 100.000 time: 0.1722 data: 0.0796 max mem: 3532 +Test: [280/333] eta: 0:00:10 acc1_pos: 75.658 acc5_pos: 100.000 loss: 0.6816 (0.6736) acc1: 58.8542 (61.5519) acc5: 100.0000 (100.0000) acc1_neg: 42.188 acc5_neg: 100.000 time: 0.1657 data: 0.0745 max mem: 3532 +Test: [290/333] eta: 0:00:08 acc1_pos: 75.658 acc5_pos: 100.000 loss: 0.9610 (0.6834) acc1: 37.2396 (60.6816) acc5: 100.0000 (100.0000) acc1_neg: 39.844 acc5_neg: 100.000 time: 0.1790 data: 0.0882 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 75.658 acc5_pos: 100.000 loss: 0.8709 (0.6887) acc1: 41.1458 (60.1606) acc5: 100.0000 (100.0000) acc1_neg: 41.927 acc5_neg: 100.000 time: 0.1741 data: 0.0793 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 75.658 acc5_pos: 100.000 loss: 0.7749 (0.6875) acc1: 48.6979 (60.2165) acc5: 100.0000 (100.0000) acc1_neg: 48.698 acc5_neg: 100.000 time: 0.1649 data: 0.0696 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 75.658 acc5_pos: 100.000 loss: 0.7516 (0.6900) acc1: 52.0833 (59.9048) acc5: 100.0000 (100.0000) acc1_neg: 52.083 acc5_neg: 100.000 time: 0.1934 data: 0.1019 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 75.658 acc5_pos: 100.000 loss: 0.7838 (0.6940) acc1: 48.1771 (59.5213) acc5: 100.0000 (100.0000) acc1_neg: 44.271 acc5_neg: 100.000 time: 0.1792 data: 0.0888 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 75.658 acc5_pos: 100.000 loss: 0.7850 (0.6945) acc1: 48.1771 (59.4784) acc5: 100.0000 (100.0000) acc1_neg: 53.521 acc5_neg: 100.000 time: 0.1716 data: 0.0774 max mem: 3532 +Test: Total time: 0:01:04 (0.1937 s / it) +* Overall Acc@1 59.478 Acc@5 100.000 + target == real Acc@1 67.901 Acc@5 100.000 + target == fake Acc@1 51.980 Acc@5 100.000 + loss 0.695 + PR-AUC 0.6335 ROC-AUC 0.6459 EER 39.14% (thr=0.4424) +Accuracy of the network on the 127843 test images: 59.5% +[rank0]:[W726 04:43:06.172482186 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:43:25.387000 3157605 torch/distributed/run.py:766] +W0726 04:43:25.387000 3157605 torch/distributed/run.py:766] ***************************************** +W0726 04:43:25.387000 3157605 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:43:25.387000 3157605 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:43:53.046998059 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:43:53.463399042 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:43:53.489059183 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:43:53.503412289 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:43:54.961187368 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:43:54.038748938 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:43:54.077816537 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:43:54.144491272 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n16_seed7_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed7_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:20:57 acc1_pos: 96.354 acc5_pos: 100.000 loss: 0.1787 (0.1787) acc1: 96.3542 (96.3542) acc5: 100.0000 (100.0000) time: 5.3507 data: 3.7746 max mem: 3531 +Test: [ 10/235] eta: 0:02:07 acc1_pos: 96.354 acc5_pos: 100.000 loss: 0.1834 (0.1857) acc1: 96.6146 (96.4015) acc5: 100.0000 (100.0000) time: 0.5681 data: 0.3436 max mem: 3531 +Test: [ 20/235] eta: 0:01:13 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.1785 (0.1816) acc1: 96.8750 (96.7014) acc5: 100.0000 (100.0000) time: 0.0913 data: 0.0004 max mem: 3531 +Test: [ 30/235] eta: 0:00:54 acc1_pos: 97.656 acc5_pos: 100.000 loss: 0.1767 (0.1806) acc1: 96.8750 (96.7070) acc5: 100.0000 (100.0000) time: 0.0961 data: 0.0003 max mem: 3531 +Test: [ 40/235] eta: 0:00:46 acc1_pos: 96.354 acc5_pos: 100.000 loss: 0.1759 (0.1802) acc1: 96.6146 (96.6972) acc5: 100.0000 (100.0000) time: 0.1258 data: 0.0290 max mem: 3531 +Test: [ 50/235] eta: 0:00:41 acc1_pos: 95.052 acc5_pos: 100.000 loss: 0.1777 (0.1807) acc1: 96.6146 (96.6605) acc5: 100.0000 (100.0000) time: 0.1652 data: 0.0723 max mem: 3531 +Test: [ 60/235] eta: 0:00:37 acc1_pos: 96.354 acc5_pos: 100.000 loss: 0.1864 (0.1818) acc1: 96.3542 (96.5890) acc5: 100.0000 (100.0000) time: 0.1688 data: 0.0776 max mem: 3531 +Test: [ 70/235] eta: 0:00:34 acc1_pos: 95.385 acc5_pos: 100.000 loss: 0.2018 (0.3556) acc1: 94.7917 (86.8471) acc5: 100.0000 (100.0000) acc1_neg: 17.969 acc5_neg: 100.000 time: 0.1647 data: 0.0733 max mem: 3531 +Test: [ 80/235] eta: 0:00:30 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.5785 (0.5095) acc1: 17.9688 (78.2375) acc5: 100.0000 (100.0000) acc1_neg: 16.146 acc5_neg: 100.000 time: 0.1561 data: 0.0636 max mem: 3531 +Test: [ 90/235] eta: 0:00:28 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.5915 (0.6286) acc1: 17.4479 (71.5516) acc5: 100.0000 (100.0000) acc1_neg: 17.708 acc5_neg: 100.000 time: 0.1499 data: 0.0556 max mem: 3531 +Test: [100/235] eta: 0:00:25 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.5848 (0.7225) acc1: 17.7083 (66.2722) acc5: 100.0000 (100.0000) acc1_neg: 14.844 acc5_neg: 100.000 time: 0.1586 data: 0.0644 max mem: 3531 +Test: [110/235] eta: 0:00:23 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.5792 (0.7993) acc1: 17.4479 (61.9252) acc5: 100.0000 (100.0000) acc1_neg: 20.052 acc5_neg: 100.000 time: 0.1520 data: 0.0584 max mem: 3531 +Test: [120/235] eta: 0:00:21 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.6126 (0.8703) acc1: 16.4062 (58.0299) acc5: 100.0000 (100.0000) acc1_neg: 16.927 acc5_neg: 100.000 time: 0.1458 data: 0.0511 max mem: 3531 +Test: [130/235] eta: 0:00:19 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.6570 (0.9296) acc1: 14.3229 (54.7054) acc5: 100.0000 (100.0000) acc1_neg: 14.062 acc5_neg: 100.000 time: 0.1578 data: 0.0631 max mem: 3531 +Test: [140/235] eta: 0:00:17 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.6869 (0.9845) acc1: 13.5417 (51.7564) acc5: 100.0000 (100.0000) acc1_neg: 9.375 acc5_neg: 100.000 time: 0.1590 data: 0.0672 max mem: 3531 +Test: [150/235] eta: 0:00:15 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.6714 (1.0286) acc1: 13.5417 (49.2722) acc5: 100.0000 (100.0000) acc1_neg: 16.146 acc5_neg: 100.000 time: 0.1661 data: 0.0738 max mem: 3531 +Test: [160/235] eta: 0:00:13 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.6551 (1.0675) acc1: 14.0625 (47.1176) acc5: 100.0000 (100.0000) acc1_neg: 15.365 acc5_neg: 100.000 time: 0.1649 data: 0.0723 max mem: 3531 +Test: [170/235] eta: 0:00:11 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.6167 (1.0786) acc1: 16.1458 (45.9034) acc5: 100.0000 (100.0000) acc1_neg: 23.958 acc5_neg: 100.000 time: 0.1507 data: 0.0586 max mem: 3531 +Test: [180/235] eta: 0:00:09 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.2166 (1.0862) acc1: 28.3854 (45.0046) acc5: 100.0000 (100.0000) acc1_neg: 28.385 acc5_neg: 100.000 time: 0.1549 data: 0.0621 max mem: 3531 +Test: [190/235] eta: 0:00:07 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.2167 (1.0929) acc1: 28.3854 (44.1468) acc5: 100.0000 (100.0000) acc1_neg: 27.344 acc5_neg: 100.000 time: 0.1579 data: 0.0658 max mem: 3531 +Test: [200/235] eta: 0:00:06 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.2194 (1.1001) acc1: 27.3438 (43.3017) acc5: 100.0000 (100.0000) acc1_neg: 26.823 acc5_neg: 100.000 time: 0.1649 data: 0.0728 max mem: 3531 +Test: [210/235] eta: 0:00:04 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.2273 (1.1054) acc1: 27.0833 (42.6059) acc5: 100.0000 (100.0000) acc1_neg: 29.427 acc5_neg: 100.000 time: 0.1772 data: 0.0854 max mem: 3532 +Test: [220/235] eta: 0:00:02 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.2155 (1.1101) acc1: 29.1667 (42.0013) acc5: 100.0000 (100.0000) acc1_neg: 28.385 acc5_neg: 100.000 time: 0.1706 data: 0.0798 max mem: 3532 +Test: [230/235] eta: 0:00:00 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.2100 (1.1147) acc1: 29.1667 (41.4153) acc5: 100.0000 (100.0000) acc1_neg: 29.688 acc5_neg: 100.000 time: 0.1577 data: 0.0664 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.2146 (1.1182) acc1: 29.1667 (41.2310) acc5: 100.0000 (100.0000) acc1_neg: 30.435 acc5_neg: 100.000 time: 0.1508 data: 0.0626 max mem: 3532 +Test: Total time: 0:00:40 (0.1736 s / it) +* Overall Acc@1 41.231 Acc@5 100.000 + target == real Acc@1 96.594 Acc@5 100.000 + target == fake Acc@1 21.207 Acc@5 100.000 + loss 1.118 + PR-AUC 0.8849 ROC-AUC 0.7467 EER 32.03% (thr=0.1575) +Accuracy of the network on the 89879 test images: 41.2% +[rank0]:[W726 04:44:50.964977982 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:45:10.623000 3158151 torch/distributed/run.py:766] +W0726 04:45:10.623000 3158151 torch/distributed/run.py:766] ***************************************** +W0726 04:45:10.623000 3158151 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:45:10.623000 3158151 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:45:38.390323951 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:45:38.423071230 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:45:39.858730099 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:45:39.141575697 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:45:39.141622170 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:45:39.171217675 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:45:39.262874038 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:45:39.301203300 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n16_seed7_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed7_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:23:44 acc1_pos: 93.229 acc5_pos: 100.000 loss: 0.3935 (0.3935) acc1: 93.2292 (93.2292) acc5: 100.0000 (100.0000) time: 5.2939 data: 3.6610 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/269] eta: 0:02:26 acc1_pos: 86.198 acc5_pos: 100.000 loss: 0.4563 (0.4475) acc1: 88.2812 (87.0265) acc5: 100.0000 (100.0000) time: 0.5639 data: 0.3337 max mem: 3531 +Test: [ 20/269] eta: 0:01:24 acc1_pos: 90.365 acc5_pos: 100.000 loss: 0.4354 (0.4221) acc1: 88.2812 (88.4053) acc5: 100.0000 (100.0000) time: 0.0909 data: 0.0008 max mem: 3531 +Test: [ 30/269] eta: 0:01:13 acc1_pos: 97.135 acc5_pos: 100.000 loss: 0.3733 (0.4059) acc1: 93.2292 (89.9110) acc5: 100.0000 (100.0000) time: 0.1655 data: 0.0712 max mem: 3531 +Test: [ 40/269] eta: 0:01:11 acc1_pos: 74.138 acc5_pos: 100.000 loss: 0.4835 (0.5175) acc1: 79.1667 (77.1024) acc5: 100.0000 (100.0000) acc1_neg: 15.104 acc5_neg: 100.000 time: 0.2853 data: 0.1897 max mem: 3531 +Test: [ 50/269] eta: 0:01:04 acc1_pos: 74.138 acc5_pos: 100.000 loss: 0.8682 (0.5949) acc1: 32.8125 (68.0913) acc5: 100.0000 (100.0000) acc1_neg: 10.938 acc5_neg: 100.000 time: 0.2762 data: 0.1830 max mem: 3531 +Test: [ 60/269] eta: 0:00:59 acc1_pos: 74.138 acc5_pos: 100.000 loss: 0.9759 (0.6676) acc1: 22.3958 (60.3911) acc5: 100.0000 (100.0000) acc1_neg: 9.896 acc5_neg: 100.000 time: 0.2290 data: 0.1362 max mem: 3531 +Test: [ 70/269] eta: 0:00:56 acc1_pos: 74.138 acc5_pos: 100.000 loss: 0.9389 (0.7051) acc1: 21.8750 (55.5641) acc5: 100.0000 (100.0000) acc1_neg: 26.302 acc5_neg: 100.000 time: 0.2480 data: 0.1524 max mem: 3531 +Test: [ 80/269] eta: 0:00:51 acc1_pos: 74.138 acc5_pos: 100.000 loss: 0.8656 (0.7219) acc1: 26.3021 (52.9257) acc5: 100.0000 (100.0000) acc1_neg: 51.042 acc5_neg: 100.000 time: 0.2327 data: 0.1379 max mem: 3531 +Test: [ 90/269] eta: 0:00:47 acc1_pos: 74.138 acc5_pos: 100.000 loss: 0.8222 (0.7370) acc1: 34.8958 (50.9587) acc5: 100.0000 (100.0000) acc1_neg: 22.917 acc5_neg: 100.000 time: 0.2172 data: 0.1235 max mem: 3531 +Test: [100/269] eta: 0:00:44 acc1_pos: 74.138 acc5_pos: 100.000 loss: 0.9102 (0.7550) acc1: 31.5104 (48.9068) acc5: 100.0000 (100.0000) acc1_neg: 21.615 acc5_neg: 100.000 time: 0.2127 data: 0.1162 max mem: 3531 +Test: [110/269] eta: 0:00:40 acc1_pos: 74.138 acc5_pos: 100.000 loss: 0.9244 (0.7720) acc1: 28.9062 (47.1495) acc5: 100.0000 (100.0000) acc1_neg: 28.906 acc5_neg: 100.000 time: 0.2026 data: 0.1068 max mem: 3531 +Test: [120/269] eta: 0:00:37 acc1_pos: 74.138 acc5_pos: 100.000 loss: 0.9433 (0.7907) acc1: 26.3021 (45.6353) acc5: 100.0000 (100.0000) acc1_neg: 20.052 acc5_neg: 100.000 time: 0.2112 data: 0.1174 max mem: 3531 +Test: [130/269] eta: 0:00:34 acc1_pos: 74.138 acc5_pos: 100.000 loss: 1.0031 (0.8131) acc1: 22.9167 (43.8156) acc5: 100.0000 (100.0000) acc1_neg: 14.583 acc5_neg: 100.000 time: 0.2026 data: 0.1109 max mem: 3531 +Test: [140/269] eta: 0:00:31 acc1_pos: 74.138 acc5_pos: 100.000 loss: 1.1189 (0.8411) acc1: 20.0521 (42.0065) acc5: 100.0000 (100.0000) acc1_neg: 24.479 acc5_neg: 100.000 time: 0.2048 data: 0.1125 max mem: 3531 +Test: [150/269] eta: 0:00:28 acc1_pos: 74.138 acc5_pos: 100.000 loss: 1.1497 (0.8622) acc1: 16.6667 (40.4025) acc5: 100.0000 (100.0000) acc1_neg: 20.052 acc5_neg: 100.000 time: 0.2101 data: 0.1158 max mem: 3531 +Test: [160/269] eta: 0:00:26 acc1_pos: 74.138 acc5_pos: 100.000 loss: 1.1181 (0.8728) acc1: 17.9688 (39.4135) acc5: 100.0000 (100.0000) acc1_neg: 32.031 acc5_neg: 100.000 time: 0.2085 data: 0.1135 max mem: 3531 +Test: [170/269] eta: 0:00:23 acc1_pos: 74.138 acc5_pos: 100.000 loss: 0.9138 (0.8740) acc1: 28.3854 (38.9498) acc5: 100.0000 (100.0000) acc1_neg: 40.885 acc5_neg: 100.000 time: 0.2091 data: 0.1157 max mem: 3531 +Test: [180/269] eta: 0:00:21 acc1_pos: 74.138 acc5_pos: 100.000 loss: 0.8088 (0.8674) acc1: 38.0208 (39.5387) acc5: 100.0000 (100.0000) acc1_neg: 34.635 acc5_neg: 100.000 time: 0.2043 data: 0.1107 max mem: 3531 +Test: [190/269] eta: 0:00:18 acc1_pos: 74.138 acc5_pos: 100.000 loss: 0.8001 (0.8692) acc1: 40.8854 (39.1893) acc5: 100.0000 (100.0000) acc1_neg: 47.656 acc5_neg: 100.000 time: 0.2079 data: 0.1125 max mem: 3531 +Test: [200/269] eta: 0:00:16 acc1_pos: 74.138 acc5_pos: 100.000 loss: 0.8914 (0.8712) acc1: 34.6354 (38.8902) acc5: 100.0000 (100.0000) acc1_neg: 34.635 acc5_neg: 100.000 time: 0.2073 data: 0.1136 max mem: 3531 +Test: [210/269] eta: 0:00:13 acc1_pos: 74.138 acc5_pos: 100.000 loss: 0.9892 (0.8848) acc1: 20.3125 (37.6975) acc5: 100.0000 (100.0000) acc1_neg: 16.146 acc5_neg: 100.000 time: 0.1864 data: 0.0952 max mem: 3532 +Test: [220/269] eta: 0:00:11 acc1_pos: 74.138 acc5_pos: 100.000 loss: 1.1105 (0.8959) acc1: 12.7604 (36.7777) acc5: 100.0000 (100.0000) acc1_neg: 34.375 acc5_neg: 100.000 time: 0.1579 data: 0.0678 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 74.138 acc5_pos: 100.000 loss: 1.0167 (0.8977) acc1: 15.1042 (36.4245) acc5: 100.0000 (100.0000) acc1_neg: 33.594 acc5_neg: 100.000 time: 0.1458 data: 0.0561 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 74.138 acc5_pos: 100.000 loss: 0.9041 (0.9003) acc1: 24.2188 (35.9688) acc5: 100.0000 (100.0000) acc1_neg: 30.208 acc5_neg: 100.000 time: 0.1500 data: 0.0601 max mem: 3532 +Test: [250/269] eta: 0:00:04 acc1_pos: 74.138 acc5_pos: 100.000 loss: 0.9728 (0.9093) acc1: 24.2188 (35.3440) acc5: 100.0000 (100.0000) acc1_neg: 3.646 acc5_neg: 100.000 time: 0.1437 data: 0.0539 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 74.138 acc5_pos: 100.000 loss: 1.1234 (0.9185) acc1: 9.8958 (34.4548) acc5: 100.0000 (100.0000) acc1_neg: 26.823 acc5_neg: 100.000 time: 0.1275 data: 0.0380 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 74.138 acc5_pos: 100.000 loss: 0.9983 (0.9177) acc1: 17.9688 (34.4438) acc5: 100.0000 (100.0000) acc1_neg: 61.680 acc5_neg: 100.000 time: 0.1230 data: 0.0300 max mem: 3532 +Test: Total time: 0:00:56 (0.2113 s / it) +* Overall Acc@1 34.444 Acc@5 100.000 + target == real Acc@1 89.592 Acc@5 100.000 + target == fake Acc@1 26.917 Acc@5 100.000 + loss 0.918 + PR-AUC 0.9298 ROC-AUC 0.6567 EER 38.97% (thr=0.3406) +Accuracy of the network on the 103293 test images: 34.4% +[rank0]:[W726 04:46:43.308224170 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T04:46:50Z] eval_done tag=fixed500k_real_n16_seed7_resampled +[2026-07-26T04:46:51Z] eval_start tag=fixed500k_fake_n32_seed7_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed7_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T04:46:51Z +hostname=n111-046-081 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed7_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n32_seed7_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 04:47:03.580000 3158728 torch/distributed/run.py:766] +W0726 04:47:03.580000 3158728 torch/distributed/run.py:766] ***************************************** +W0726 04:47:03.580000 3158728 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:47:03.580000 3158728 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:47:31.445460479 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:47:31.504025153 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:47:31.544053995 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:47:31.555837162 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:47:32.780610491 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +[rank0]:[W726 04:47:32.783847744 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:47:32.794485145 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank6]:[W726 04:47:32.794603373 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n32_seed7_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed7_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:17:44 acc1_pos: 77.604 acc5_pos: 100.000 loss: 0.4496 (0.4496) acc1: 77.6042 (77.6042) acc5: 100.0000 (100.0000) time: 4.5481 data: 2.7604 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/234] eta: 0:01:51 acc1_pos: 56.510 acc5_pos: 100.000 loss: 0.6189 (0.6352) acc1: 65.1042 (62.7131) acc5: 100.0000 (100.0000) time: 0.4990 data: 0.2553 max mem: 3531 +Test: [ 20/234] eta: 0:01:06 acc1_pos: 95.833 acc5_pos: 100.000 loss: 0.6186 (0.5955) acc1: 69.0104 (66.9023) acc5: 100.0000 (100.0000) time: 0.0972 data: 0.0078 max mem: 3531 +Test: [ 30/234] eta: 0:00:48 acc1_pos: 75.521 acc5_pos: 100.000 loss: 0.2427 (0.4864) acc1: 93.4896 (74.9076) acc5: 100.0000 (100.0000) time: 0.0967 data: 0.0056 max mem: 3531 +Test: [ 40/234] eta: 0:00:46 acc1_pos: 98.958 acc5_pos: 100.000 loss: 0.2423 (0.4591) acc1: 93.7500 (75.9972) acc5: 100.0000 (100.0000) time: 0.1671 data: 0.0753 max mem: 3531 +Test: [ 50/234] eta: 0:00:43 acc1_pos: 99.479 acc5_pos: 100.000 loss: 0.1022 (0.3889) acc1: 98.6979 (80.4943) acc5: 100.0000 (100.0000) time: 0.2349 data: 0.1395 max mem: 3531 +Test: [ 60/234] eta: 0:00:40 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.1019 (0.3424) acc1: 99.4792 (83.6578) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2209 data: 0.1255 max mem: 3531 +Test: [ 70/234] eta: 0:00:37 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.1071 (0.4058) acc1: 99.7396 (80.0176) acc5: 100.0000 (100.0000) acc1_neg: 34.375 acc5_neg: 100.000 time: 0.2107 data: 0.1192 max mem: 3531 +Test: [ 80/234] eta: 0:00:35 acc1_pos: 98.621 acc5_pos: 100.000 loss: 1.0587 (0.4540) acc1: 38.0208 (77.3502) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2090 data: 0.1166 max mem: 3531 +Test: [ 90/234] eta: 0:00:32 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.2435 (0.4247) acc1: 90.6250 (79.2611) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.1998 data: 0.1068 max mem: 3531 +Test: [100/234] eta: 0:00:29 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.2456 (0.4204) acc1: 90.6250 (79.5715) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1795 data: 0.0869 max mem: 3531 +Test: [110/234] eta: 0:00:26 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.3540 (0.4159) acc1: 82.8125 (79.9901) acc5: 100.0000 (100.0000) acc1_neg: 79.427 acc5_neg: 100.000 time: 0.1697 data: 0.0772 max mem: 3531 +Test: [120/234] eta: 0:00:23 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.3536 (0.4242) acc1: 84.6354 (79.8059) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1648 data: 0.0727 max mem: 3531 +Test: [130/234] eta: 0:00:21 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.4052 (0.4267) acc1: 82.8125 (79.8823) acc5: 100.0000 (100.0000) acc1_neg: 78.646 acc5_neg: 100.000 time: 0.1594 data: 0.0667 max mem: 3531 +Test: [140/234] eta: 0:00:19 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.4944 (0.4543) acc1: 76.8229 (78.5406) acc5: 100.0000 (100.0000) acc1_neg: 34.635 acc5_neg: 100.000 time: 0.1612 data: 0.0666 max mem: 3531 +Test: [150/234] eta: 0:00:16 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.4663 (0.4586) acc1: 75.2604 (78.3716) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1652 data: 0.0727 max mem: 3531 +Test: [160/234] eta: 0:00:14 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.4219 (0.4570) acc1: 78.9062 (78.5714) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.1688 data: 0.0755 max mem: 3531 +Test: [170/234] eta: 0:00:12 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.5217 (0.4614) acc1: 75.2604 (78.4387) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.1646 data: 0.0710 max mem: 3531 +Test: [180/234] eta: 0:00:10 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.5475 (0.4787) acc1: 72.9167 (77.4401) acc5: 100.0000 (100.0000) acc1_neg: 26.562 acc5_neg: 100.000 time: 0.1697 data: 0.0753 max mem: 3531 +Test: [190/234] eta: 0:00:08 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.7502 (0.4942) acc1: 58.3333 (76.4248) acc5: 100.0000 (100.0000) acc1_neg: 58.333 acc5_neg: 100.000 time: 0.1849 data: 0.0882 max mem: 3531 +Test: [200/234] eta: 0:00:06 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.7321 (0.4862) acc1: 59.8958 (76.8462) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1819 data: 0.0884 max mem: 3531 +Test: [210/234] eta: 0:00:04 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.0960 (0.4706) acc1: 99.7396 (77.8078) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.2050 data: 0.1123 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.2704 (0.4638) acc1: 91.9271 (78.3041) acc5: 100.0000 (100.0000) acc1_neg: 67.708 acc5_neg: 100.000 time: 0.2233 data: 0.1269 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.6487 (0.5071) acc1: 18.7500 (75.6527) acc5: 100.0000 (100.0000) acc1_neg: 17.188 acc5_neg: 100.000 time: 0.1764 data: 0.0805 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 98.621 acc5_pos: 100.000 loss: 1.4385 (0.5191) acc1: 18.4896 (74.9981) acc5: 100.0000 (100.0000) acc1_neg: 23.735 acc5_neg: 100.000 time: 0.1560 data: 0.0617 max mem: 3532 +Test: Total time: 0:00:45 (0.1934 s / it) +* Overall Acc@1 74.998 Acc@5 100.000 + target == real Acc@1 80.628 Acc@5 100.000 + target == fake Acc@1 73.411 Acc@5 100.000 + loss 0.519 + PR-AUC 0.9576 ROC-AUC 0.8587 EER 23.83% (thr=0.4492) +Accuracy of the network on the 89729 test images: 75.0% +[rank0]:[W726 04:48:28.395339024 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:48:47.510000 3159274 torch/distributed/run.py:766] +W0726 04:48:47.510000 3159274 torch/distributed/run.py:766] ***************************************** +W0726 04:48:47.510000 3159274 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:48:47.510000 3159274 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:49:14.624809648 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:49:14.246631528 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:49:14.421341830 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:49:15.743353078 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +[rank1]:[W726 04:49:15.762177974 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:49:15.772332147 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:49:15.777347520 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:49:15.803749859 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n32_seed7_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed7_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:15:23 acc1_pos: 43.490 acc5_pos: 100.000 loss: 0.8134 (0.8134) acc1: 43.4896 (43.4896) acc5: 100.0000 (100.0000) time: 6.4197 data: 4.7413 max mem: 3531 +Test: [ 10/2387] eta: 0:26:23 acc1_pos: 20.312 acc5_pos: 100.000 loss: 1.0574 (1.0040) acc1: 26.0417 (30.5871) acc5: 100.0000 (100.0000) time: 0.6661 data: 0.4314 max mem: 3531 +Test: [ 20/2387] eta: 0:15:35 acc1_pos: 30.208 acc5_pos: 100.000 loss: 1.0574 (1.0274) acc1: 25.7812 (28.7946) acc5: 100.0000 (100.0000) time: 0.0940 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:13:03 acc1_pos: 8.073 acc5_pos: 100.000 loss: 1.0492 (1.0408) acc1: 29.1667 (28.2762) acc5: 100.0000 (100.0000) time: 0.1486 data: 0.0523 max mem: 3531 +Test: [ 40/2387] eta: 0:12:06 acc1_pos: 23.698 acc5_pos: 100.000 loss: 1.0488 (1.0461) acc1: 23.6979 (27.1786) acc5: 100.0000 (100.0000) time: 0.2199 data: 0.1253 max mem: 3531 +Test: [ 50/2387] eta: 0:11:27 acc1_pos: 36.198 acc5_pos: 100.000 loss: 0.9938 (1.0347) acc1: 28.1250 (28.3241) acc5: 100.0000 (100.0000) time: 0.2349 data: 0.1419 max mem: 3531 +Test: [ 60/2387] eta: 0:11:04 acc1_pos: 42.188 acc5_pos: 100.000 loss: 0.9732 (1.0307) acc1: 32.0312 (28.8294) acc5: 100.0000 (100.0000) time: 0.2361 data: 0.1421 max mem: 3531 +Test: [ 70/2387] eta: 0:10:47 acc1_pos: 31.771 acc5_pos: 100.000 loss: 1.0354 (1.0406) acc1: 24.4792 (28.0223) acc5: 100.0000 (100.0000) time: 0.2417 data: 0.1482 max mem: 3531 +Test: [ 80/2387] eta: 0:10:32 acc1_pos: 88.802 acc5_pos: 100.000 loss: 0.9634 (0.9919) acc1: 31.7708 (32.1181) acc5: 100.0000 (100.0000) time: 0.2402 data: 0.1496 max mem: 3531 +Test: [ 90/2387] eta: 0:10:24 acc1_pos: 70.833 acc5_pos: 100.000 loss: 0.5345 (0.9366) acc1: 68.2292 (37.1652) acc5: 100.0000 (100.0000) time: 0.2446 data: 0.1538 max mem: 3531 +Test: [ 100/2387] eta: 0:10:14 acc1_pos: 67.969 acc5_pos: 100.000 loss: 0.5058 (0.9016) acc1: 72.9167 (40.4548) acc5: 100.0000 (100.0000) time: 0.2448 data: 0.1532 max mem: 3531 +Test: [ 110/2387] eta: 0:10:05 acc1_pos: 72.656 acc5_pos: 100.000 loss: 0.5659 (0.8679) acc1: 72.6562 (43.7078) acc5: 100.0000 (100.0000) time: 0.2385 data: 0.1466 max mem: 3531 +Test: [ 120/2387] eta: 0:09:57 acc1_pos: 62.760 acc5_pos: 100.000 loss: 0.5716 (0.8465) acc1: 69.5312 (45.4954) acc5: 100.0000 (100.0000) time: 0.2384 data: 0.1451 max mem: 3531 +Test: [ 130/2387] eta: 0:09:49 acc1_pos: 64.323 acc5_pos: 100.000 loss: 0.5671 (0.8232) acc1: 69.5312 (47.6900) acc5: 100.0000 (100.0000) time: 0.2367 data: 0.1429 max mem: 3531 +Test: [ 140/2387] eta: 0:09:44 acc1_pos: 74.740 acc5_pos: 100.000 loss: 0.5296 (0.8043) acc1: 75.0000 (49.4699) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1428 max mem: 3531 +Test: [ 150/2387] eta: 0:09:37 acc1_pos: 76.562 acc5_pos: 100.000 loss: 0.5179 (0.7834) acc1: 76.3021 (51.4297) acc5: 100.0000 (100.0000) time: 0.2380 data: 0.1392 max mem: 3531 +Test: [ 160/2387] eta: 0:09:27 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.4957 (0.7629) acc1: 76.5625 (53.2609) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2165 data: 0.1233 max mem: 3531 +Test: [ 170/2387] eta: 0:09:17 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2484 (0.7319) acc1: 95.8333 (55.8221) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2013 data: 0.1081 max mem: 3531 +Test: [ 180/2387] eta: 0:09:09 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2325 (0.7042) acc1: 97.1354 (58.1017) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2033 data: 0.1088 max mem: 3531 +Test: [ 190/2387] eta: 0:09:02 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2318 (0.6793) acc1: 96.8750 (60.1372) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2085 data: 0.1127 max mem: 3531 +Test: [ 200/2387] eta: 0:08:54 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2253 (0.6569) acc1: 96.8750 (61.9623) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1999 data: 0.1043 max mem: 3531 +Test: [ 210/2387] eta: 0:08:46 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2295 (0.6370) acc1: 96.6146 (63.5923) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1899 data: 0.0979 max mem: 3532 +Test: [ 220/2387] eta: 0:08:36 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2334 (0.6187) acc1: 96.8750 (65.0912) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1799 data: 0.0852 max mem: 3532 +Test: [ 230/2387] eta: 0:08:27 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2345 (0.6020) acc1: 96.8750 (66.4559) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1678 data: 0.0720 max mem: 3532 +Test: [ 240/2387] eta: 0:08:20 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2433 (0.5924) acc1: 95.0521 (67.3431) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.1770 data: 0.0837 max mem: 3532 +Test: [ 250/2387] eta: 0:08:14 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2201 (0.5732) acc1: 88.2812 (68.5819) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1870 data: 0.0947 max mem: 3532 +Test: [ 260/2387] eta: 0:08:07 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0767 (0.5540) acc1: 100.0000 (69.7857) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1788 data: 0.0864 max mem: 3532 +Test: [ 270/2387] eta: 0:08:01 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0681 (0.5361) acc1: 100.0000 (70.8996) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1801 data: 0.0862 max mem: 3532 +Test: [ 280/2387] eta: 0:07:56 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0703 (0.5197) acc1: 100.0000 (71.9343) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1854 data: 0.0917 max mem: 3532 +Test: [ 290/2387] eta: 0:07:50 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0737 (0.5044) acc1: 100.0000 (72.8979) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1813 data: 0.0889 max mem: 3532 +Test: [ 300/2387] eta: 0:07:45 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0737 (0.4903) acc1: 100.0000 (73.7965) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1833 data: 0.0879 max mem: 3532 +Test: [ 310/2387] eta: 0:07:41 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0888 (0.4820) acc1: 100.0000 (74.5369) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1933 data: 0.0953 max mem: 3532 +Test: [ 320/2387] eta: 0:07:38 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3983 (0.4796) acc1: 94.0104 (75.1290) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.2018 data: 0.1030 max mem: 3532 +Test: [ 330/2387] eta: 0:07:35 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.4042 (0.4771) acc1: 93.2292 (75.6916) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2063 data: 0.1084 max mem: 3532 +Test: [ 340/2387] eta: 0:07:32 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3994 (0.4749) acc1: 93.4896 (76.2280) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2105 data: 0.1146 max mem: 3532 +Test: [ 350/2387] eta: 0:07:28 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3916 (0.4683) acc1: 95.0521 (76.8786) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2035 data: 0.1100 max mem: 3532 +Test: [ 360/2387] eta: 0:07:25 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2115 (0.4607) acc1: 99.2188 (77.4779) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1925 data: 0.0985 max mem: 3532 +Test: [ 370/2387] eta: 0:07:22 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1874 (0.4534) acc1: 98.4375 (78.0429) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1999 data: 0.1058 max mem: 3532 +Test: [ 380/2387] eta: 0:07:17 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1874 (0.4464) acc1: 98.6979 (78.5843) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1898 data: 0.0969 max mem: 3532 +Test: [ 390/2387] eta: 0:07:14 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1885 (0.4398) acc1: 98.6979 (79.0934) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1900 data: 0.0935 max mem: 3532 +Test: [ 400/2387] eta: 0:07:12 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1876 (0.4336) acc1: 98.4375 (79.5849) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2063 data: 0.1111 max mem: 3532 +Test: [ 410/2387] eta: 0:07:09 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1859 (0.4276) acc1: 98.6979 (80.0430) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2052 data: 0.1118 max mem: 3532 +Test: [ 420/2387] eta: 0:07:07 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1859 (0.4218) acc1: 98.6979 (80.4836) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2134 data: 0.1181 max mem: 3532 +Test: [ 430/2387] eta: 0:07:03 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1916 (0.4181) acc1: 98.1771 (80.7902) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.2053 data: 0.1113 max mem: 3532 +Test: [ 440/2387] eta: 0:07:00 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1955 (0.4126) acc1: 93.4896 (81.1567) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1917 data: 0.0993 max mem: 3533 +Test: [ 450/2387] eta: 0:06:58 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0928 (0.4055) acc1: 100.0000 (81.5745) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1999 data: 0.1079 max mem: 3533 +Test: [ 460/2387] eta: 0:06:55 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0926 (0.3988) acc1: 100.0000 (81.9742) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1994 data: 0.1053 max mem: 3533 +Test: [ 470/2387] eta: 0:06:51 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0924 (0.3923) acc1: 100.0000 (82.3564) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1872 data: 0.0887 max mem: 3533 +Test: [ 480/2387] eta: 0:06:49 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0936 (0.3861) acc1: 100.0000 (82.7232) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1994 data: 0.1018 max mem: 3533 +Test: [ 490/2387] eta: 0:06:44 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0936 (0.3801) acc1: 100.0000 (83.0750) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1818 data: 0.0876 max mem: 3533 +Test: [ 500/2387] eta: 0:06:42 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0933 (0.3744) acc1: 100.0000 (83.4123) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1840 data: 0.0907 max mem: 3533 +Test: [ 510/2387] eta: 0:06:40 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0951 (0.3692) acc1: 100.0000 (83.7339) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2108 data: 0.1179 max mem: 3533 +Test: [ 520/2387] eta: 0:06:36 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1243 (0.3649) acc1: 100.0000 (84.0431) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0866 max mem: 3533 +Test: [ 530/2387] eta: 0:06:34 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1229 (0.3603) acc1: 100.0000 (84.3436) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1989 data: 0.1030 max mem: 3533 +Test: [ 540/2387] eta: 0:06:33 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1211 (0.3559) acc1: 100.0000 (84.6330) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2299 data: 0.1343 max mem: 3533 +Test: [ 550/2387] eta: 0:06:28 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1212 (0.3516) acc1: 100.0000 (84.9119) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0855 max mem: 3533 +Test: [ 560/2387] eta: 0:06:26 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1212 (0.3475) acc1: 100.0000 (85.1804) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1727 data: 0.0758 max mem: 3533 +Test: [ 570/2387] eta: 0:06:24 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1203 (0.3436) acc1: 100.0000 (85.4399) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2163 data: 0.1218 max mem: 3533 +Test: [ 580/2387] eta: 0:06:22 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1203 (0.3397) acc1: 100.0000 (85.6905) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2172 data: 0.1259 max mem: 3533 +Test: [ 590/2387] eta: 0:06:20 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1215 (0.3361) acc1: 100.0000 (85.9322) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2118 data: 0.1201 max mem: 3533 +Test: [ 600/2387] eta: 0:06:17 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1285 (0.3331) acc1: 100.0000 (86.1607) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2077 data: 0.1166 max mem: 3533 +Test: [ 610/2387] eta: 0:06:14 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1678 (0.3308) acc1: 98.9583 (86.3654) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1828 data: 0.0907 max mem: 3533 +Test: [ 620/2387] eta: 0:06:10 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1922 (0.3286) acc1: 98.6979 (86.5669) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1604 data: 0.0670 max mem: 3533 +Test: [ 630/2387] eta: 0:06:07 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1926 (0.3264) acc1: 98.9583 (86.7604) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1609 data: 0.0694 max mem: 3533 +Test: [ 640/2387] eta: 0:06:04 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1957 (0.3244) acc1: 98.6979 (86.9438) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1646 data: 0.0741 max mem: 3533 +Test: [ 650/2387] eta: 0:06:00 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1961 (0.3224) acc1: 98.4375 (87.1220) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1603 data: 0.0687 max mem: 3533 +Test: [ 660/2387] eta: 0:05:58 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1951 (0.3205) acc1: 98.9583 (87.3026) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1891 data: 0.0980 max mem: 3533 +Test: [ 670/2387] eta: 0:05:56 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1941 (0.3186) acc1: 98.9583 (87.4767) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2125 data: 0.1210 max mem: 3534 +Test: [ 680/2387] eta: 0:05:54 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1983 (0.3182) acc1: 98.6979 (87.5941) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1930 data: 0.1005 max mem: 3534 +Test: [ 690/2387] eta: 0:05:52 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3403 (0.3205) acc1: 92.9688 (87.4555) acc5: 100.0000 (100.0000) acc1_neg: 67.969 acc5_neg: 100.000 time: 0.1948 data: 0.1031 max mem: 3534 +Test: [ 700/2387] eta: 0:05:49 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.5117 (0.3233) acc1: 73.9583 (87.2511) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.1912 data: 0.0992 max mem: 3534 +Test: [ 710/2387] eta: 0:05:46 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.5148 (0.3261) acc1: 72.6562 (87.0436) acc5: 100.0000 (100.0000) acc1_neg: 72.656 acc5_neg: 100.000 time: 0.1880 data: 0.0961 max mem: 3534 +Test: [ 720/2387] eta: 0:05:44 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.5148 (0.3287) acc1: 72.9167 (86.8571) acc5: 100.0000 (100.0000) acc1_neg: 73.698 acc5_neg: 100.000 time: 0.1983 data: 0.1077 max mem: 3534 +Test: [ 730/2387] eta: 0:05:43 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.5260 (0.3315) acc1: 71.6146 (86.6553) acc5: 100.0000 (100.0000) acc1_neg: 71.615 acc5_neg: 100.000 time: 0.2267 data: 0.1357 max mem: 3534 +Test: [ 740/2387] eta: 0:05:40 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.5267 (0.3341) acc1: 71.8750 (86.4808) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.2162 data: 0.1224 max mem: 3534 +Test: [ 750/2387] eta: 0:05:38 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.5130 (0.3365) acc1: 73.1771 (86.3068) acc5: 100.0000 (100.0000) acc1_neg: 72.917 acc5_neg: 100.000 time: 0.1756 data: 0.0822 max mem: 3534 +Test: [ 760/2387] eta: 0:05:36 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.5319 (0.3393) acc1: 72.9167 (86.1175) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.1910 data: 0.0983 max mem: 3534 +Test: [ 770/2387] eta: 0:05:34 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.5408 (0.3400) acc1: 72.9167 (86.0800) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2098 data: 0.1166 max mem: 3534 +Test: [ 780/2387] eta: 0:05:31 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2074 (0.3381) acc1: 95.0521 (86.1989) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2031 data: 0.1120 max mem: 3534 +Test: [ 790/2387] eta: 0:05:29 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1583 (0.3360) acc1: 96.3542 (86.3164) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1844 data: 0.0932 max mem: 3534 +Test: [ 800/2387] eta: 0:05:27 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1349 (0.3336) acc1: 98.4375 (86.4632) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1913 data: 0.0999 max mem: 3534 +Test: [ 810/2387] eta: 0:05:24 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1479 (0.3317) acc1: 97.6562 (86.5855) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1876 data: 0.0948 max mem: 3534 +Test: [ 820/2387] eta: 0:05:22 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1804 (0.3304) acc1: 96.0938 (86.6712) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.1888 data: 0.0962 max mem: 3534 +Test: [ 830/2387] eta: 0:05:20 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2823 (0.3315) acc1: 88.2812 (86.5937) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.2051 data: 0.1147 max mem: 3534 +Test: [ 840/2387] eta: 0:05:18 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3038 (0.3307) acc1: 88.2812 (86.6872) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2040 data: 0.1124 max mem: 3534 +Test: [ 850/2387] eta: 0:05:15 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2391 (0.3296) acc1: 95.0521 (86.7916) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1971 data: 0.1049 max mem: 3534 +Test: [ 860/2387] eta: 0:05:13 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2149 (0.3283) acc1: 97.6562 (86.9175) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1869 data: 0.0946 max mem: 3534 +Test: [ 870/2387] eta: 0:05:10 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2149 (0.3272) acc1: 97.6562 (87.0318) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1849 data: 0.0930 max mem: 3534 +Test: [ 880/2387] eta: 0:05:08 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2364 (0.3262) acc1: 96.8750 (87.1394) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1961 data: 0.1054 max mem: 3534 +Test: [ 890/2387] eta: 0:05:06 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2631 (0.3257) acc1: 95.3125 (87.2241) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1898 data: 0.0992 max mem: 3534 +Test: [ 900/2387] eta: 0:05:03 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3195 (0.3317) acc1: 90.3646 (86.8156) acc5: 100.0000 (100.0000) acc1_neg: 45.573 acc5_neg: 100.000 time: 0.1768 data: 0.0842 max mem: 3535 +Test: [ 910/2387] eta: 0:05:01 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.9964 (0.3390) acc1: 39.3229 (86.3020) acc5: 100.0000 (100.0000) acc1_neg: 38.802 acc5_neg: 100.000 time: 0.1810 data: 0.0881 max mem: 3535 +Test: [ 920/2387] eta: 0:04:58 acc1_pos: 69.406 acc5_pos: 100.000 loss: 1.0057 (0.3463) acc1: 38.8021 (85.7826) acc5: 100.0000 (100.0000) acc1_neg: 35.417 acc5_neg: 100.000 time: 0.1799 data: 0.0885 max mem: 3535 +Test: [ 930/2387] eta: 0:04:56 acc1_pos: 69.406 acc5_pos: 100.000 loss: 1.0041 (0.3534) acc1: 39.3229 (85.2863) acc5: 100.0000 (100.0000) acc1_neg: 35.677 acc5_neg: 100.000 time: 0.1767 data: 0.0860 max mem: 3535 +Test: [ 940/2387] eta: 0:04:54 acc1_pos: 69.406 acc5_pos: 100.000 loss: 1.0020 (0.3603) acc1: 39.8438 (84.8070) acc5: 100.0000 (100.0000) acc1_neg: 41.406 acc5_neg: 100.000 time: 0.1776 data: 0.0871 max mem: 3535 +Test: [ 950/2387] eta: 0:04:51 acc1_pos: 69.406 acc5_pos: 100.000 loss: 1.0002 (0.3670) acc1: 39.0625 (84.3230) acc5: 100.0000 (100.0000) acc1_neg: 39.062 acc5_neg: 100.000 time: 0.1785 data: 0.0857 max mem: 3535 +Test: [ 960/2387] eta: 0:04:49 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.9899 (0.3735) acc1: 39.5833 (83.8639) acc5: 100.0000 (100.0000) acc1_neg: 41.927 acc5_neg: 100.000 time: 0.1774 data: 0.0836 max mem: 3535 +Test: [ 970/2387] eta: 0:04:46 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.9922 (0.3800) acc1: 39.5833 (83.4007) acc5: 100.0000 (100.0000) acc1_neg: 41.667 acc5_neg: 100.000 time: 0.1735 data: 0.0816 max mem: 3535 +Test: [ 980/2387] eta: 0:04:44 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.9738 (0.3816) acc1: 40.1042 (83.2946) acc5: 100.0000 (100.0000) acc1_neg: 80.729 acc5_neg: 100.000 time: 0.1706 data: 0.0802 max mem: 3535 +Test: [ 990/2387] eta: 0:04:41 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2939 (0.3805) acc1: 88.8021 (83.3683) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1599 data: 0.0668 max mem: 3535 +Test: [1000/2387] eta: 0:04:39 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2361 (0.3792) acc1: 91.1458 (83.4486) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.1599 data: 0.0661 max mem: 3535 +Test: [1010/2387] eta: 0:04:36 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2648 (0.3791) acc1: 90.8854 (83.4706) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.1699 data: 0.0761 max mem: 3535 +Test: [1020/2387] eta: 0:04:34 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3080 (0.3787) acc1: 86.4583 (83.4894) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1669 data: 0.0721 max mem: 3535 +Test: [1030/2387] eta: 0:04:31 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1886 (0.3769) acc1: 96.3542 (83.6150) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1711 data: 0.0789 max mem: 3535 +Test: [1040/2387] eta: 0:04:29 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1451 (0.3746) acc1: 98.9583 (83.7644) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1722 data: 0.0802 max mem: 3535 +Test: [1050/2387] eta: 0:04:26 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1361 (0.3724) acc1: 99.2188 (83.9072) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1658 data: 0.0737 max mem: 3535 +Test: [1060/2387] eta: 0:04:24 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1519 (0.3705) acc1: 98.4375 (84.0353) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1720 data: 0.0810 max mem: 3535 +Test: [1070/2387] eta: 0:04:22 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1716 (0.3692) acc1: 97.1354 (84.1248) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.1728 data: 0.0815 max mem: 3535 +Test: [1080/2387] eta: 0:04:19 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2952 (0.3694) acc1: 86.7188 (84.1184) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1597 data: 0.0676 max mem: 3535 +Test: [1090/2387] eta: 0:04:17 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2952 (0.3681) acc1: 89.8438 (84.2082) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1628 data: 0.0712 max mem: 3535 +Test: [1100/2387] eta: 0:04:15 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2052 (0.3668) acc1: 94.0104 (84.2943) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1878 data: 0.0958 max mem: 3535 +Test: [1110/2387] eta: 0:04:12 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2013 (0.3655) acc1: 95.8333 (84.3813) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1720 data: 0.0777 max mem: 3535 +Test: [1120/2387] eta: 0:04:10 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2057 (0.3644) acc1: 94.7917 (84.4472) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1573 data: 0.0633 max mem: 3536 +Test: [1130/2387] eta: 0:04:08 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2146 (0.3632) acc1: 94.0104 (84.5320) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1801 data: 0.0884 max mem: 3536 +Test: [1140/2387] eta: 0:04:06 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2582 (0.3634) acc1: 89.3229 (84.5293) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.1806 data: 0.0900 max mem: 3536 +Test: [1150/2387] eta: 0:04:04 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3583 (0.3628) acc1: 84.1146 (84.5843) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1751 data: 0.0834 max mem: 3536 +Test: [1160/2387] eta: 0:04:01 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2101 (0.3613) acc1: 96.8750 (84.6955) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1697 data: 0.0760 max mem: 3536 +Test: [1170/2387] eta: 0:03:59 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1786 (0.3597) acc1: 98.9583 (84.8164) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1656 data: 0.0732 max mem: 3536 +Test: [1180/2387] eta: 0:03:57 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1740 (0.3583) acc1: 99.2188 (84.9311) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1693 data: 0.0774 max mem: 3536 +Test: [1190/2387] eta: 0:03:55 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1590 (0.3568) acc1: 99.2188 (85.0456) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1736 data: 0.0784 max mem: 3536 +Test: [1200/2387] eta: 0:03:52 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1585 (0.3553) acc1: 99.2188 (85.1643) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1705 data: 0.0746 max mem: 3536 +Test: [1210/2387] eta: 0:03:50 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1714 (0.3538) acc1: 99.4792 (85.2762) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1728 data: 0.0805 max mem: 3536 +Test: [1220/2387] eta: 0:03:48 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0921 (0.3515) acc1: 99.7396 (85.3956) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1775 data: 0.0857 max mem: 3536 +Test: [1230/2387] eta: 0:03:46 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0720 (0.3493) acc1: 100.0000 (85.5142) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1739 data: 0.0798 max mem: 3536 +Test: [1240/2387] eta: 0:03:44 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0707 (0.3471) acc1: 100.0000 (85.6301) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1829 data: 0.0901 max mem: 3536 +Test: [1250/2387] eta: 0:03:42 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0724 (0.3449) acc1: 100.0000 (85.7443) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1816 data: 0.0900 max mem: 3536 +Test: [1260/2387] eta: 0:03:40 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0746 (0.3427) acc1: 100.0000 (85.8563) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1874 data: 0.0938 max mem: 3536 +Test: [1270/2387] eta: 0:03:38 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0779 (0.3408) acc1: 100.0000 (85.9668) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1941 data: 0.0987 max mem: 3536 +Test: [1280/2387] eta: 0:03:36 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1019 (0.3391) acc1: 99.7396 (86.0700) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2065 data: 0.1119 max mem: 3536 +Test: [1290/2387] eta: 0:03:34 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1261 (0.3375) acc1: 98.9583 (86.1626) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2092 data: 0.1176 max mem: 3536 +Test: [1300/2387] eta: 0:03:33 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1314 (0.3359) acc1: 99.2188 (86.2598) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2338 data: 0.1393 max mem: 3536 +Test: [1310/2387] eta: 0:03:31 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1314 (0.3345) acc1: 99.2188 (86.3531) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2185 data: 0.1232 max mem: 3536 +Test: [1320/2387] eta: 0:03:29 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1436 (0.3331) acc1: 98.4375 (86.4414) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2044 data: 0.1134 max mem: 3536 +Test: [1330/2387] eta: 0:03:27 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1716 (0.3325) acc1: 95.8333 (86.4859) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.1942 data: 0.1005 max mem: 3536 +Test: [1340/2387] eta: 0:03:25 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2655 (0.3327) acc1: 90.3646 (86.4663) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1809 data: 0.0857 max mem: 3536 +Test: [1350/2387] eta: 0:03:23 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0606 (0.3306) acc1: 100.0000 (86.5665) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2105 data: 0.1177 max mem: 3537 +Test: [1360/2387] eta: 0:03:20 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0527 (0.3286) acc1: 100.0000 (86.6652) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1543 data: 0.0594 max mem: 3537 +Test: [1370/2387] eta: 0:03:18 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0515 (0.3266) acc1: 100.0000 (86.7624) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1570 data: 0.0609 max mem: 3537 +Test: [1380/2387] eta: 0:03:17 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0488 (0.3246) acc1: 100.0000 (86.8583) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2055 data: 0.1131 max mem: 3537 +Test: [1390/2387] eta: 0:03:14 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0513 (0.3227) acc1: 100.0000 (86.9528) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1885 data: 0.0975 max mem: 3537 +Test: [1400/2387] eta: 0:03:12 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0539 (0.3207) acc1: 100.0000 (87.0459) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1639 data: 0.0712 max mem: 3537 +Test: [1410/2387] eta: 0:03:10 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0588 (0.3196) acc1: 100.0000 (87.1169) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1785 data: 0.0861 max mem: 3537 +Test: [1420/2387] eta: 0:03:08 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1642 (0.3187) acc1: 96.3542 (87.1725) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1727 data: 0.0811 max mem: 3537 +Test: [1430/2387] eta: 0:03:06 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1722 (0.3177) acc1: 96.3542 (87.2347) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1717 data: 0.0799 max mem: 3537 +Test: [1440/2387] eta: 0:03:04 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1662 (0.3170) acc1: 97.6562 (87.2981) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.1894 data: 0.0985 max mem: 3537 +Test: [1450/2387] eta: 0:03:02 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2437 (0.3170) acc1: 94.2708 (87.3080) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1898 data: 0.0998 max mem: 3537 +Test: [1460/2387] eta: 0:03:00 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1581 (0.3158) acc1: 97.6562 (87.3833) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1630 data: 0.0726 max mem: 3537 +Test: [1470/2387] eta: 0:02:58 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1211 (0.3144) acc1: 99.4792 (87.4678) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1464 data: 0.0559 max mem: 3537 +Test: [1480/2387] eta: 0:02:56 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1141 (0.3132) acc1: 99.7396 (87.5422) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1863 data: 0.0946 max mem: 3537 +Test: [1490/2387] eta: 0:02:53 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1269 (0.3119) acc1: 99.4792 (87.6223) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1775 data: 0.0859 max mem: 3537 +Test: [1500/2387] eta: 0:02:51 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1311 (0.3107) acc1: 99.4792 (87.6978) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1610 data: 0.0702 max mem: 3537 +Test: [1510/2387] eta: 0:02:49 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1496 (0.3100) acc1: 98.4375 (87.7649) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1821 data: 0.0911 max mem: 3537 +Test: [1520/2387] eta: 0:02:47 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2124 (0.3095) acc1: 98.1771 (87.8246) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1690 data: 0.0781 max mem: 3537 +Test: [1530/2387] eta: 0:02:45 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2134 (0.3089) acc1: 97.3958 (87.8800) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1820 data: 0.0900 max mem: 3537 +Test: [1540/2387] eta: 0:02:43 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1927 (0.3082) acc1: 98.6979 (87.9482) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2018 data: 0.1078 max mem: 3537 +Test: [1550/2387] eta: 0:02:42 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1927 (0.3075) acc1: 99.2188 (88.0141) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1959 data: 0.1006 max mem: 3537 +Test: [1560/2387] eta: 0:02:40 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2088 (0.3069) acc1: 98.1771 (88.0731) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1857 data: 0.0927 max mem: 3537 +Test: [1570/2387] eta: 0:02:38 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2138 (0.3064) acc1: 96.8750 (88.1263) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1756 data: 0.0828 max mem: 3537 +Test: [1580/2387] eta: 0:02:36 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2138 (0.3058) acc1: 96.3542 (88.1767) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.2046 data: 0.1102 max mem: 3538 +Test: [1590/2387] eta: 0:02:34 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1818 (0.3049) acc1: 96.3542 (88.2353) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2332 data: 0.1391 max mem: 3538 +Test: [1600/2387] eta: 0:02:32 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1327 (0.3040) acc1: 98.4375 (88.2839) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2366 data: 0.1443 max mem: 3538 +Test: [1610/2387] eta: 0:02:30 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1342 (0.3031) acc1: 98.4375 (88.3424) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2097 data: 0.1168 max mem: 3538 +Test: [1620/2387] eta: 0:02:29 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1608 (0.3022) acc1: 96.3542 (88.3899) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2029 data: 0.1069 max mem: 3538 +Test: [1630/2387] eta: 0:02:27 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1892 (0.3022) acc1: 94.5312 (88.3935) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.2215 data: 0.1270 max mem: 3538 +Test: [1640/2387] eta: 0:02:25 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3661 (0.3030) acc1: 83.5938 (88.3516) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.2076 data: 0.1154 max mem: 3538 +Test: [1650/2387] eta: 0:02:23 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3985 (0.3035) acc1: 79.9479 (88.3142) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.2087 data: 0.1147 max mem: 3538 +Test: [1660/2387] eta: 0:02:21 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3673 (0.3040) acc1: 83.3333 (88.2809) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1872 data: 0.0930 max mem: 3538 +Test: [1670/2387] eta: 0:02:19 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3673 (0.3044) acc1: 84.1146 (88.2644) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1844 data: 0.0895 max mem: 3538 +Test: [1680/2387] eta: 0:02:17 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3648 (0.3047) acc1: 85.1562 (88.2448) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1977 data: 0.1015 max mem: 3538 +Test: [1690/2387] eta: 0:02:15 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3589 (0.3052) acc1: 83.8542 (88.2113) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.1599 data: 0.0639 max mem: 3538 +Test: [1700/2387] eta: 0:02:13 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3503 (0.3051) acc1: 85.4167 (88.2168) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1677 data: 0.0742 max mem: 3538 +Test: [1710/2387] eta: 0:02:11 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1049 (0.3039) acc1: 99.2188 (88.2840) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0981 max mem: 3538 +Test: [1720/2387] eta: 0:02:09 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0854 (0.3026) acc1: 99.7396 (88.3503) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1791 data: 0.0865 max mem: 3538 +Test: [1730/2387] eta: 0:02:07 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0849 (0.3014) acc1: 99.7396 (88.4159) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1777 data: 0.0837 max mem: 3538 +Test: [1740/2387] eta: 0:02:05 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0849 (0.3001) acc1: 99.7396 (88.4806) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1805 data: 0.0865 max mem: 3538 +Test: [1750/2387] eta: 0:02:03 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0820 (0.2989) acc1: 100.0000 (88.5443) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1722 data: 0.0806 max mem: 3538 +Test: [1760/2387] eta: 0:02:01 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0883 (0.2979) acc1: 99.4792 (88.6026) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1992 data: 0.1077 max mem: 3538 +Test: [1770/2387] eta: 0:01:59 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1363 (0.2970) acc1: 98.1771 (88.6546) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1910 data: 0.0966 max mem: 3538 +Test: [1780/2387] eta: 0:01:57 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1179 (0.2960) acc1: 98.9583 (88.7155) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1579 data: 0.0635 max mem: 3538 +Test: [1790/2387] eta: 0:01:55 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1025 (0.2949) acc1: 99.7396 (88.7758) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1753 data: 0.0842 max mem: 3538 +Test: [1800/2387] eta: 0:01:53 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0958 (0.2938) acc1: 99.7396 (88.8343) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1821 data: 0.0894 max mem: 3538 +Test: [1810/2387] eta: 0:01:51 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1006 (0.2927) acc1: 99.4792 (88.8933) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1812 data: 0.0890 max mem: 3539 +Test: [1820/2387] eta: 0:01:49 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1006 (0.2918) acc1: 99.4792 (88.9461) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1721 data: 0.0816 max mem: 3539 +Test: [1830/2387] eta: 0:01:47 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1242 (0.2910) acc1: 98.9583 (88.9978) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1877 data: 0.0952 max mem: 3539 +Test: [1840/2387] eta: 0:01:45 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1077 (0.2900) acc1: 100.0000 (89.0576) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1848 data: 0.0910 max mem: 3539 +Test: [1850/2387] eta: 0:01:43 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1071 (0.2890) acc1: 100.0000 (89.1165) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1651 data: 0.0728 max mem: 3539 +Test: [1860/2387] eta: 0:01:41 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1071 (0.2880) acc1: 100.0000 (89.1750) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1846 data: 0.0940 max mem: 3539 +Test: [1870/2387] eta: 0:01:39 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1073 (0.2870) acc1: 100.0000 (89.2327) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1890 data: 0.0960 max mem: 3539 +Test: [1880/2387] eta: 0:01:37 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1066 (0.2861) acc1: 100.0000 (89.2900) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1796 data: 0.0863 max mem: 3539 +Test: [1890/2387] eta: 0:01:35 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1066 (0.2851) acc1: 100.0000 (89.3466) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0827 max mem: 3539 +Test: [1900/2387] eta: 0:01:33 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1073 (0.2842) acc1: 100.0000 (89.4026) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1741 data: 0.0801 max mem: 3539 +Test: [1910/2387] eta: 0:01:31 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1314 (0.2835) acc1: 100.0000 (89.4558) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1659 data: 0.0729 max mem: 3539 +Test: [1920/2387] eta: 0:01:29 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1413 (0.2832) acc1: 100.0000 (89.4761) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1617 data: 0.0684 max mem: 3539 +Test: [1930/2387] eta: 0:01:27 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1495 (0.2825) acc1: 98.1771 (89.5214) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1639 data: 0.0700 max mem: 3539 +Test: [1940/2387] eta: 0:01:25 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1311 (0.2819) acc1: 99.2188 (89.5391) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1672 data: 0.0747 max mem: 3539 +Test: [1950/2387] eta: 0:01:23 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1540 (0.2814) acc1: 98.6979 (89.5751) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1710 data: 0.0792 max mem: 3539 +Test: [1960/2387] eta: 0:01:21 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1667 (0.2810) acc1: 98.4375 (89.6084) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1655 data: 0.0742 max mem: 3539 +Test: [1970/2387] eta: 0:01:19 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2420 (0.2813) acc1: 91.4062 (89.5732) acc5: 100.0000 (100.0000) acc1_neg: 73.698 acc5_neg: 100.000 time: 0.1721 data: 0.0798 max mem: 3539 +Test: [1980/2387] eta: 0:01:18 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3294 (0.2813) acc1: 87.5000 (89.5753) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1950 data: 0.1017 max mem: 3539 +Test: [1990/2387] eta: 0:01:16 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1548 (0.2807) acc1: 98.1771 (89.6185) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1989 data: 0.1064 max mem: 3539 +Test: [2000/2387] eta: 0:01:14 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1548 (0.2800) acc1: 98.1771 (89.6609) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2023 data: 0.1110 max mem: 3539 +Test: [2010/2387] eta: 0:01:12 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1566 (0.2794) acc1: 98.1771 (89.7023) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2100 data: 0.1191 max mem: 3539 +Test: [2020/2387] eta: 0:01:10 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1584 (0.2788) acc1: 97.9167 (89.7430) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1789 data: 0.0874 max mem: 3539 +Test: [2030/2387] eta: 0:01:08 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1553 (0.2783) acc1: 97.9167 (89.7836) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1825 data: 0.0905 max mem: 3540 +Test: [2040/2387] eta: 0:01:06 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1547 (0.2776) acc1: 98.1771 (89.8245) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2027 data: 0.1084 max mem: 3540 +Test: [2050/2387] eta: 0:01:04 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1541 (0.2771) acc1: 98.1771 (89.8661) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1666 data: 0.0700 max mem: 3540 +Test: [2060/2387] eta: 0:01:02 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1608 (0.2765) acc1: 97.9167 (89.9048) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1799 data: 0.0853 max mem: 3540 +Test: [2070/2387] eta: 0:01:00 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1608 (0.2759) acc1: 97.9167 (89.9461) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2005 data: 0.1083 max mem: 3540 +Test: [2080/2387] eta: 0:00:58 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1568 (0.2754) acc1: 98.1771 (89.9852) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1899 data: 0.0980 max mem: 3540 +Test: [2090/2387] eta: 0:00:56 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1580 (0.2748) acc1: 98.1771 (90.0246) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1903 data: 0.0973 max mem: 3540 +Test: [2100/2387] eta: 0:00:55 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1558 (0.2742) acc1: 98.1771 (90.0640) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1888 data: 0.0961 max mem: 3540 +Test: [2110/2387] eta: 0:00:53 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1561 (0.2737) acc1: 98.4375 (90.1045) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1896 data: 0.0969 max mem: 3540 +Test: [2120/2387] eta: 0:00:51 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1616 (0.2734) acc1: 98.1771 (90.1312) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1860 data: 0.0921 max mem: 3540 +Test: [2130/2387] eta: 0:00:49 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1712 (0.2729) acc1: 97.6562 (90.1670) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1824 data: 0.0896 max mem: 3540 +Test: [2140/2387] eta: 0:00:47 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1530 (0.2725) acc1: 98.4375 (90.1944) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1958 data: 0.1017 max mem: 3540 +Test: [2150/2387] eta: 0:00:45 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1661 (0.2720) acc1: 97.6562 (90.2271) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1925 data: 0.0980 max mem: 3540 +Test: [2160/2387] eta: 0:00:43 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1661 (0.2716) acc1: 97.3958 (90.2595) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1815 data: 0.0895 max mem: 3540 +Test: [2170/2387] eta: 0:00:41 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2100 (0.2719) acc1: 95.0521 (90.2480) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.1743 data: 0.0830 max mem: 3540 +Test: [2180/2387] eta: 0:00:39 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3837 (0.2734) acc1: 82.0312 (90.1386) acc5: 100.0000 (100.0000) acc1_neg: 42.969 acc5_neg: 100.000 time: 0.1814 data: 0.0879 max mem: 3540 +Test: [2190/2387] eta: 0:00:37 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3837 (0.2739) acc1: 80.4688 (90.1046) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1934 data: 0.0990 max mem: 3540 +Test: [2200/2387] eta: 0:00:35 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1607 (0.2733) acc1: 97.9167 (90.1464) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1800 data: 0.0873 max mem: 3540 +Test: [2210/2387] eta: 0:00:33 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1239 (0.2726) acc1: 99.4792 (90.1899) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1846 data: 0.0926 max mem: 3540 +Test: [2220/2387] eta: 0:00:31 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1211 (0.2719) acc1: 99.7396 (90.2297) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1900 data: 0.0983 max mem: 3540 +Test: [2230/2387] eta: 0:00:30 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1194 (0.2713) acc1: 99.7396 (90.2697) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1890 data: 0.0955 max mem: 3540 +Test: [2240/2387] eta: 0:00:28 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1130 (0.2706) acc1: 99.7396 (90.3078) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1899 data: 0.0964 max mem: 3540 +Test: [2250/2387] eta: 0:00:26 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1625 (0.2704) acc1: 97.9167 (90.3351) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1897 data: 0.0957 max mem: 3540 +Test: [2260/2387] eta: 0:00:24 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1760 (0.2700) acc1: 98.1771 (90.3710) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1863 data: 0.0923 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1498 (0.2694) acc1: 99.7396 (90.4076) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1878 data: 0.0964 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1192 (0.2689) acc1: 100.0000 (90.4341) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1865 data: 0.0913 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1096 (0.2683) acc1: 100.0000 (90.4743) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1872 data: 0.0912 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1375 (0.2678) acc1: 99.7396 (90.5099) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1731 data: 0.0807 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1589 (0.2676) acc1: 97.9167 (90.5288) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1416 data: 0.0497 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2583 (0.2676) acc1: 94.0104 (90.5309) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1299 data: 0.0388 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2597 (0.2676) acc1: 92.9688 (90.5440) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.1307 data: 0.0406 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1900 (0.2673) acc1: 96.8750 (90.5696) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1202 data: 0.0306 max mem: 3541 +Test: [2350/2387] eta: 0:00:07 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1675 (0.2671) acc1: 98.1771 (90.5822) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1179 data: 0.0284 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1965 (0.2670) acc1: 95.8333 (90.5931) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1264 data: 0.0369 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2318 (0.2670) acc1: 94.0104 (90.5934) acc5: 100.0000 (100.0000) acc1_neg: 78.646 acc5_neg: 100.000 time: 0.1198 data: 0.0304 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3361 (0.2677) acc1: 82.0312 (90.5494) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.1079 data: 0.0185 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3917 (0.2680) acc1: 80.9896 (90.5284) acc5: 100.0000 (100.0000) acc1_neg: 81.395 acc5_neg: 100.000 time: 0.1055 data: 0.0188 max mem: 3541 +Test: Total time: 0:07:30 (0.1886 s / it) +* Overall Acc@1 90.528 Acc@5 100.000 + target == real Acc@1 52.353 Acc@5 100.000 + target == fake Acc@1 93.209 Acc@5 100.000 + loss 0.268 + PR-AUC 0.9921 ROC-AUC 0.8982 EER 17.85% (thr=0.7144) +Accuracy of the network on the 916267 test images: 90.5% +[rank0]:[W726 04:57:01.473815591 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:57:21.045000 3159906 torch/distributed/run.py:766] +W0726 04:57:21.045000 3159906 torch/distributed/run.py:766] ***************************************** +W0726 04:57:21.045000 3159906 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:57:21.045000 3159906 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:57:48.934053154 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:57:48.150615568 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:57:48.531755517 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:57:49.594202005 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:57:49.654506957 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +| distributed init (rank 5): env:// +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:57:49.901414375 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W726 04:57:49.901509519 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:57:49.905134403 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n32_seed7_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed7_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 0:53:51 acc1_pos: 43.490 acc5_pos: 100.000 loss: 0.8134 (0.8134) acc1: 43.4896 (43.4896) acc5: 100.0000 (100.0000) time: 6.2503 data: 4.4265 max mem: 3531 +Test: [ 10/517] eta: 0:05:29 acc1_pos: 20.312 acc5_pos: 100.000 loss: 1.0574 (1.0040) acc1: 26.0417 (30.5871) acc5: 100.0000 (100.0000) time: 0.6500 data: 0.4028 max mem: 3531 +Test: [ 20/517] eta: 0:03:14 acc1_pos: 30.208 acc5_pos: 100.000 loss: 1.0574 (1.0274) acc1: 25.7812 (28.7946) acc5: 100.0000 (100.0000) time: 0.0978 data: 0.0063 max mem: 3531 +Test: [ 30/517] eta: 0:02:51 acc1_pos: 8.073 acc5_pos: 100.000 loss: 1.0492 (1.0408) acc1: 29.1667 (28.2762) acc5: 100.0000 (100.0000) time: 0.1878 data: 0.0931 max mem: 3531 +Test: [ 40/517] eta: 0:02:52 acc1_pos: 23.698 acc5_pos: 100.000 loss: 1.0488 (1.0461) acc1: 23.6979 (27.1786) acc5: 100.0000 (100.0000) time: 0.3299 data: 0.2349 max mem: 3531 +Test: [ 50/517] eta: 0:02:46 acc1_pos: 41.284 acc5_pos: 100.000 loss: 0.9841 (1.0111) acc1: 28.1250 (30.4126) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.3658 data: 0.2701 max mem: 3531 +Test: [ 60/517] eta: 0:02:32 acc1_pos: 41.284 acc5_pos: 100.000 loss: 0.5841 (0.9283) acc1: 65.8854 (38.2087) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.2804 data: 0.1837 max mem: 3531 +Test: [ 70/517] eta: 0:02:25 acc1_pos: 41.284 acc5_pos: 100.000 loss: 0.4577 (0.8566) acc1: 80.2083 (45.0778) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.2450 data: 0.1486 max mem: 3531 +Test: [ 80/517] eta: 0:02:19 acc1_pos: 41.284 acc5_pos: 100.000 loss: 0.4577 (0.8109) acc1: 83.8542 (49.2509) acc5: 100.0000 (100.0000) acc1_neg: 66.927 acc5_neg: 100.000 time: 0.2721 data: 0.1759 max mem: 3531 +Test: [ 90/517] eta: 0:02:13 acc1_pos: 41.284 acc5_pos: 100.000 loss: 0.5095 (0.7790) acc1: 78.9062 (52.5670) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.2612 data: 0.1670 max mem: 3531 +Test: [100/517] eta: 0:02:08 acc1_pos: 41.284 acc5_pos: 100.000 loss: 0.4566 (0.7438) acc1: 85.4167 (56.1366) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.2631 data: 0.1699 max mem: 3531 +Test: [110/517] eta: 0:02:02 acc1_pos: 41.284 acc5_pos: 100.000 loss: 0.3766 (0.7101) acc1: 89.8438 (59.0160) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2550 data: 0.1599 max mem: 3531 +Test: [120/517] eta: 0:01:58 acc1_pos: 41.284 acc5_pos: 100.000 loss: 0.4345 (0.6918) acc1: 84.3750 (60.6426) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.2466 data: 0.1512 max mem: 3531 +Test: [130/517] eta: 0:01:54 acc1_pos: 41.284 acc5_pos: 100.000 loss: 0.4904 (0.6746) acc1: 78.1250 (62.1959) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2574 data: 0.1637 max mem: 3531 +Test: [140/517] eta: 0:01:50 acc1_pos: 41.284 acc5_pos: 100.000 loss: 0.3297 (0.6486) acc1: 96.6146 (64.7773) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2573 data: 0.1613 max mem: 3531 +Test: [150/517] eta: 0:01:46 acc1_pos: 41.284 acc5_pos: 100.000 loss: 0.3052 (0.6253) acc1: 98.6979 (67.0168) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2599 data: 0.1607 max mem: 3531 +Test: [160/517] eta: 0:01:42 acc1_pos: 41.284 acc5_pos: 100.000 loss: 0.3061 (0.6064) acc1: 97.1354 (68.6352) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2568 data: 0.1578 max mem: 3531 +Test: [170/517] eta: 0:01:39 acc1_pos: 41.284 acc5_pos: 100.000 loss: 0.2980 (0.5879) acc1: 94.5312 (70.1968) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2549 data: 0.1582 max mem: 3531 +Test: [180/517] eta: 0:01:35 acc1_pos: 41.284 acc5_pos: 100.000 loss: 0.2959 (0.5751) acc1: 94.2708 (71.3196) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.2541 data: 0.1574 max mem: 3531 +Test: [190/517] eta: 0:01:31 acc1_pos: 41.284 acc5_pos: 100.000 loss: 0.2661 (0.5588) acc1: 97.3958 (72.7258) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2366 data: 0.1392 max mem: 3531 +Test: [200/517] eta: 0:01:28 acc1_pos: 41.284 acc5_pos: 100.000 loss: 0.2661 (0.5446) acc1: 98.6979 (74.0076) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2252 data: 0.1295 max mem: 3531 +Test: [210/517] eta: 0:01:24 acc1_pos: 41.284 acc5_pos: 100.000 loss: 0.2847 (0.5380) acc1: 95.3125 (74.5730) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.2333 data: 0.1360 max mem: 3532 +Test: [220/517] eta: 0:01:21 acc1_pos: 41.284 acc5_pos: 100.000 loss: 0.4059 (0.5321) acc1: 85.4167 (75.0825) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.2384 data: 0.1396 max mem: 3532 +Test: [230/517] eta: 0:01:18 acc1_pos: 41.284 acc5_pos: 100.000 loss: 0.4219 (0.5275) acc1: 85.9375 (75.5986) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2399 data: 0.1410 max mem: 3532 +Test: [240/517] eta: 0:01:15 acc1_pos: 41.284 acc5_pos: 100.000 loss: 0.3931 (0.5212) acc1: 89.0625 (76.2967) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.2408 data: 0.1419 max mem: 3532 +Test: [250/517] eta: 0:01:12 acc1_pos: 41.284 acc5_pos: 100.000 loss: 0.3448 (0.5139) acc1: 94.7917 (77.0916) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2311 data: 0.1335 max mem: 3532 +Test: [260/517] eta: 0:01:09 acc1_pos: 41.284 acc5_pos: 100.000 loss: 0.3144 (0.5066) acc1: 98.6979 (77.8855) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.2363 data: 0.1399 max mem: 3532 +Test: [270/517] eta: 0:01:06 acc1_pos: 41.284 acc5_pos: 100.000 loss: 0.3855 (0.5038) acc1: 91.9271 (78.2912) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.2513 data: 0.1564 max mem: 3532 +Test: [280/517] eta: 0:01:03 acc1_pos: 41.284 acc5_pos: 100.000 loss: 0.4244 (0.5001) acc1: 89.8438 (78.6894) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2262 data: 0.1321 max mem: 3532 +Test: [290/517] eta: 0:01:00 acc1_pos: 41.284 acc5_pos: 100.000 loss: 0.3631 (0.4951) acc1: 93.2292 (79.2374) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2445 data: 0.1480 max mem: 3532 +Test: [300/517] eta: 0:00:57 acc1_pos: 41.284 acc5_pos: 100.000 loss: 0.3319 (0.4902) acc1: 95.0521 (79.7325) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.2663 data: 0.1678 max mem: 3532 +Test: [310/517] eta: 0:00:54 acc1_pos: 41.284 acc5_pos: 100.000 loss: 0.4127 (0.4894) acc1: 87.2396 (79.8382) acc5: 100.0000 (100.0000) acc1_neg: 75.521 acc5_neg: 100.000 time: 0.2246 data: 0.1261 max mem: 3532 +Test: [320/517] eta: 0:00:51 acc1_pos: 41.284 acc5_pos: 100.000 loss: 0.4449 (0.4856) acc1: 85.6771 (80.1881) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.2299 data: 0.1347 max mem: 3532 +Test: [330/517] eta: 0:00:49 acc1_pos: 41.284 acc5_pos: 100.000 loss: 0.3058 (0.4798) acc1: 94.7917 (80.6662) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2340 data: 0.1388 max mem: 3532 +Test: [340/517] eta: 0:00:46 acc1_pos: 41.284 acc5_pos: 100.000 loss: 0.2840 (0.4773) acc1: 97.3958 (80.8124) acc5: 100.0000 (100.0000) acc1_neg: 54.688 acc5_neg: 100.000 time: 0.2200 data: 0.1241 max mem: 3532 +Test: [350/517] eta: 0:00:43 acc1_pos: 41.284 acc5_pos: 100.000 loss: 0.4321 (0.4782) acc1: 83.8542 (80.6698) acc5: 100.0000 (100.0000) acc1_neg: 76.302 acc5_neg: 100.000 time: 0.2299 data: 0.1361 max mem: 3532 +Test: [360/517] eta: 0:00:40 acc1_pos: 41.284 acc5_pos: 100.000 loss: 0.4524 (0.4766) acc1: 81.7708 (80.8222) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.2397 data: 0.1430 max mem: 3532 +Test: [370/517] eta: 0:00:38 acc1_pos: 41.284 acc5_pos: 100.000 loss: 0.3190 (0.4707) acc1: 95.0521 (81.3125) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2403 data: 0.1418 max mem: 3532 +Test: [380/517] eta: 0:00:35 acc1_pos: 41.284 acc5_pos: 100.000 loss: 0.2847 (0.4668) acc1: 98.4375 (81.6649) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2499 data: 0.1548 max mem: 3532 +Test: [390/517] eta: 0:00:32 acc1_pos: 41.284 acc5_pos: 100.000 loss: 0.3018 (0.4629) acc1: 94.2708 (81.9513) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2542 data: 0.1609 max mem: 3532 +Test: [400/517] eta: 0:00:30 acc1_pos: 41.284 acc5_pos: 100.000 loss: 0.2949 (0.4585) acc1: 95.3125 (82.3326) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2399 data: 0.1442 max mem: 3532 +Test: [410/517] eta: 0:00:27 acc1_pos: 41.284 acc5_pos: 100.000 loss: 0.2971 (0.4561) acc1: 95.3125 (82.6427) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2450 data: 0.1496 max mem: 3532 +Test: [420/517] eta: 0:00:25 acc1_pos: 41.284 acc5_pos: 100.000 loss: 0.3942 (0.4551) acc1: 92.7083 (82.7840) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2449 data: 0.1499 max mem: 3532 +Test: [430/517] eta: 0:00:22 acc1_pos: 41.284 acc5_pos: 100.000 loss: 0.3597 (0.4524) acc1: 91.1458 (83.0342) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.2450 data: 0.1500 max mem: 3532 +Test: [440/517] eta: 0:00:19 acc1_pos: 41.284 acc5_pos: 100.000 loss: 0.4094 (0.4523) acc1: 90.1042 (83.0918) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.2549 data: 0.1619 max mem: 3533 +Test: [450/517] eta: 0:00:17 acc1_pos: 41.284 acc5_pos: 100.000 loss: 0.3783 (0.4499) acc1: 92.7083 (83.4020) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2640 data: 0.1669 max mem: 3533 +Test: [460/517] eta: 0:00:14 acc1_pos: 41.284 acc5_pos: 100.000 loss: 0.3422 (0.4477) acc1: 97.1354 (83.6446) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.2649 data: 0.1632 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 41.284 acc5_pos: 100.000 loss: 0.3638 (0.4458) acc1: 94.5312 (83.8542) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2549 data: 0.1549 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 41.284 acc5_pos: 100.000 loss: 0.3286 (0.4427) acc1: 95.5729 (84.1563) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2499 data: 0.1518 max mem: 3533 +Test: [490/517] eta: 0:00:06 acc1_pos: 41.284 acc5_pos: 100.000 loss: 0.3259 (0.4410) acc1: 96.6146 (84.3145) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.2449 data: 0.1491 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 41.284 acc5_pos: 100.000 loss: 0.4606 (0.4437) acc1: 84.1146 (84.0247) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.2114 data: 0.1197 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 41.284 acc5_pos: 100.000 loss: 0.4685 (0.4429) acc1: 81.2500 (84.1329) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1787 data: 0.0891 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 41.284 acc5_pos: 100.000 loss: 0.3858 (0.4420) acc1: 90.3646 (84.2688) acc5: 100.0000 (100.0000) acc1_neg: 96.113 acc5_neg: 100.000 time: 0.1688 data: 0.0777 max mem: 3533 +Test: Total time: 0:02:11 (0.2542 s / it) +* Overall Acc@1 84.269 Acc@5 100.000 + target == real Acc@1 28.145 Acc@5 100.000 + target == fake Acc@1 90.091 Acc@5 100.000 + loss 0.442 + PR-AUC 0.9409 ROC-AUC 0.6622 EER 38.14% (thr=0.6831) +Accuracy of the network on the 198427 test images: 84.3% +[rank0]:[W726 05:00:11.724726786 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 05:00:31.165000 3160472 torch/distributed/run.py:766] +W0726 05:00:31.165000 3160472 torch/distributed/run.py:766] ***************************************** +W0726 05:00:31.165000 3160472 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:00:31.165000 3160472 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:01:01.994153442 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:01:01.060173373 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:01:01.224216730 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:01:01.470036939 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +NCCL version 2.26.2+cuda12.2 +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:01:01.498187377 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:01:02.615158462 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank1]:[W726 05:01:02.615276837 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:01:02.715982221 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n32_seed7_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed7_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:44:41 acc1_pos: 66.146 acc5_pos: 100.000 loss: 0.6235 (0.6235) acc1: 66.1458 (66.1458) acc5: 100.0000 (100.0000) time: 6.8923 data: 4.2977 max mem: 3531 +Test: [ 10/389] eta: 0:04:29 acc1_pos: 76.823 acc5_pos: 100.000 loss: 0.5459 (0.6061) acc1: 72.6562 (64.6307) acc5: 100.0000 (100.0000) time: 0.7099 data: 0.3912 max mem: 3531 +Test: [ 20/389] eta: 0:02:34 acc1_pos: 54.948 acc5_pos: 100.000 loss: 0.6981 (0.6442) acc1: 54.9479 (60.5035) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0005 max mem: 3531 +Test: [ 30/389] eta: 0:02:07 acc1_pos: 78.385 acc5_pos: 100.000 loss: 0.6244 (0.6240) acc1: 68.2292 (64.7513) acc5: 100.0000 (100.0000) time: 0.1611 data: 0.0652 max mem: 3531 +Test: [ 40/389] eta: 0:01:50 acc1_pos: 33.854 acc5_pos: 100.000 loss: 0.5748 (0.6185) acc1: 75.0000 (66.4380) acc5: 100.0000 (100.0000) time: 0.2101 data: 0.1177 max mem: 3531 +Test: [ 50/389] eta: 0:01:39 acc1_pos: 81.771 acc5_pos: 100.000 loss: 0.4971 (0.6003) acc1: 81.7708 (68.8368) acc5: 100.0000 (100.0000) time: 0.1987 data: 0.1063 max mem: 3531 +Test: [ 60/389] eta: 0:01:31 acc1_pos: 77.604 acc5_pos: 100.000 loss: 0.4810 (0.5702) acc1: 83.5938 (71.9604) acc5: 100.0000 (100.0000) time: 0.1947 data: 0.1013 max mem: 3531 +Test: [ 70/389] eta: 0:01:25 acc1_pos: 80.469 acc5_pos: 100.000 loss: 0.4781 (0.5675) acc1: 83.5938 (72.4288) acc5: 100.0000 (100.0000) time: 0.1964 data: 0.1026 max mem: 3531 +Test: [ 80/389] eta: 0:01:19 acc1_pos: 98.177 acc5_pos: 100.000 loss: 0.5806 (0.5732) acc1: 66.6667 (71.8879) acc5: 100.0000 (100.0000) time: 0.1919 data: 0.0999 max mem: 3531 +Test: [ 90/389] eta: 0:01:14 acc1_pos: 74.740 acc5_pos: 100.000 loss: 0.4454 (0.5574) acc1: 74.7396 (73.0941) acc5: 100.0000 (100.0000) time: 0.1801 data: 0.0886 max mem: 3531 +Test: [100/389] eta: 0:01:10 acc1_pos: 60.417 acc5_pos: 100.000 loss: 0.6023 (0.5696) acc1: 63.5417 (71.6790) acc5: 100.0000 (100.0000) time: 0.1879 data: 0.0952 max mem: 3531 +Test: [110/389] eta: 0:01:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6801 (0.5778) acc1: 58.8542 (70.0849) acc5: 100.0000 (100.0000) time: 0.1940 data: 0.1020 max mem: 3531 +Test: [120/389] eta: 0:01:03 acc1_pos: 75.521 acc5_pos: 100.000 loss: 0.6023 (0.5747) acc1: 68.2292 (70.4997) acc5: 100.0000 (100.0000) time: 0.1949 data: 0.1020 max mem: 3531 +Test: [130/389] eta: 0:00:59 acc1_pos: 82.209 acc5_pos: 100.000 loss: 0.5986 (0.5831) acc1: 68.2292 (69.6605) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.1813 data: 0.0879 max mem: 3531 +Test: [140/389] eta: 0:00:56 acc1_pos: 82.209 acc5_pos: 100.000 loss: 0.5986 (0.5915) acc1: 67.1875 (68.9956) acc5: 100.0000 (100.0000) acc1_neg: 3.125 acc5_neg: 100.000 time: 0.1778 data: 0.0862 max mem: 3531 +Test: [150/389] eta: 0:00:53 acc1_pos: 82.209 acc5_pos: 100.000 loss: 0.5986 (0.6034) acc1: 40.1042 (67.5204) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1887 data: 0.0961 max mem: 3531 +Test: [160/389] eta: 0:00:51 acc1_pos: 82.209 acc5_pos: 100.000 loss: 0.7063 (0.5981) acc1: 50.7812 (68.0868) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2033 data: 0.1108 max mem: 3531 +Test: [170/389] eta: 0:00:48 acc1_pos: 82.209 acc5_pos: 100.000 loss: 0.3971 (0.5956) acc1: 91.4062 (68.2566) acc5: 100.0000 (100.0000) acc1_neg: 38.802 acc5_neg: 100.000 time: 0.2162 data: 0.1225 max mem: 3531 +Test: [180/389] eta: 0:00:46 acc1_pos: 82.209 acc5_pos: 100.000 loss: 0.6411 (0.6007) acc1: 60.1562 (67.6263) acc5: 100.0000 (100.0000) acc1_neg: 46.094 acc5_neg: 100.000 time: 0.2059 data: 0.1108 max mem: 3531 +Test: [190/389] eta: 0:00:43 acc1_pos: 82.209 acc5_pos: 100.000 loss: 0.6627 (0.6062) acc1: 54.9479 (66.9039) acc5: 100.0000 (100.0000) acc1_neg: 8.073 acc5_neg: 100.000 time: 0.1995 data: 0.1069 max mem: 3531 +Test: [200/389] eta: 0:00:41 acc1_pos: 82.209 acc5_pos: 100.000 loss: 0.7998 (0.6173) acc1: 32.2917 (65.6030) acc5: 100.0000 (100.0000) acc1_neg: 27.083 acc5_neg: 100.000 time: 0.1976 data: 0.1030 max mem: 3531 +Test: [210/389] eta: 0:00:38 acc1_pos: 82.209 acc5_pos: 100.000 loss: 0.6436 (0.6128) acc1: 32.0312 (66.0274) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1794 data: 0.0839 max mem: 3532 +Test: [220/389] eta: 0:00:36 acc1_pos: 82.209 acc5_pos: 100.000 loss: 0.3604 (0.6058) acc1: 95.8333 (66.7892) acc5: 100.0000 (100.0000) acc1_neg: 57.031 acc5_neg: 100.000 time: 0.1711 data: 0.0792 max mem: 3532 +Test: [230/389] eta: 0:00:34 acc1_pos: 82.209 acc5_pos: 100.000 loss: 0.5173 (0.6079) acc1: 69.0104 (66.3533) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.1848 data: 0.0921 max mem: 3532 +Test: [240/389] eta: 0:00:31 acc1_pos: 82.209 acc5_pos: 100.000 loss: 0.6472 (0.6097) acc1: 52.3438 (66.1469) acc5: 100.0000 (100.0000) acc1_neg: 43.490 acc5_neg: 100.000 time: 0.1837 data: 0.0898 max mem: 3532 +Test: [250/389] eta: 0:00:29 acc1_pos: 82.209 acc5_pos: 100.000 loss: 0.5727 (0.6079) acc1: 71.0938 (66.5142) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1891 data: 0.0953 max mem: 3532 +Test: [260/389] eta: 0:00:27 acc1_pos: 82.209 acc5_pos: 100.000 loss: 0.5871 (0.6100) acc1: 73.1771 (66.3643) acc5: 100.0000 (100.0000) acc1_neg: 25.000 acc5_neg: 100.000 time: 0.1950 data: 0.1018 max mem: 3532 +Test: [270/389] eta: 0:00:24 acc1_pos: 82.209 acc5_pos: 100.000 loss: 0.6962 (0.6178) acc1: 44.5312 (65.3156) acc5: 100.0000 (100.0000) acc1_neg: 37.500 acc5_neg: 100.000 time: 0.1794 data: 0.0857 max mem: 3532 +Test: [280/389] eta: 0:00:22 acc1_pos: 82.209 acc5_pos: 100.000 loss: 0.8432 (0.6259) acc1: 31.2500 (64.2599) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1762 data: 0.0811 max mem: 3532 +Test: [290/389] eta: 0:00:20 acc1_pos: 82.209 acc5_pos: 100.000 loss: 0.8288 (0.6261) acc1: 36.4583 (64.3292) acc5: 100.0000 (100.0000) acc1_neg: 75.781 acc5_neg: 100.000 time: 0.1759 data: 0.0821 max mem: 3532 +Test: [300/389] eta: 0:00:18 acc1_pos: 82.209 acc5_pos: 100.000 loss: 0.6178 (0.6277) acc1: 67.9688 (64.1975) acc5: 100.0000 (100.0000) acc1_neg: 57.552 acc5_neg: 100.000 time: 0.1791 data: 0.0872 max mem: 3532 +Test: [310/389] eta: 0:00:16 acc1_pos: 82.209 acc5_pos: 100.000 loss: 0.6970 (0.6340) acc1: 54.1667 (63.5316) acc5: 100.0000 (100.0000) acc1_neg: 15.104 acc5_neg: 100.000 time: 0.1956 data: 0.1040 max mem: 3532 +Test: [320/389] eta: 0:00:14 acc1_pos: 82.209 acc5_pos: 100.000 loss: 0.6869 (0.6337) acc1: 47.9167 (63.5766) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1819 data: 0.0904 max mem: 3532 +Test: [330/389] eta: 0:00:12 acc1_pos: 82.209 acc5_pos: 100.000 loss: 0.8428 (0.6406) acc1: 28.1250 (62.5480) acc5: 100.0000 (100.0000) acc1_neg: 36.198 acc5_neg: 100.000 time: 0.1645 data: 0.0739 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 82.209 acc5_pos: 100.000 loss: 0.5216 (0.6336) acc1: 65.1042 (63.4294) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1628 data: 0.0718 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 82.209 acc5_pos: 100.000 loss: 0.4173 (0.6440) acc1: 92.4479 (62.8717) acc5: 100.0000 (100.0000) acc1_neg: 3.906 acc5_neg: 100.000 time: 0.1497 data: 0.0591 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 82.209 acc5_pos: 100.000 loss: 0.8544 (0.6521) acc1: 40.6250 (62.0376) acc5: 100.0000 (100.0000) acc1_neg: 40.885 acc5_neg: 100.000 time: 0.1389 data: 0.0494 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 82.209 acc5_pos: 100.000 loss: 0.8450 (0.6533) acc1: 40.8854 (61.9469) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1295 data: 0.0400 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 82.209 acc5_pos: 100.000 loss: 0.6410 (0.6527) acc1: 63.0208 (62.0796) acc5: 100.0000 (100.0000) acc1_neg: 44.271 acc5_neg: 100.000 time: 0.1250 data: 0.0356 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 82.209 acc5_pos: 100.000 loss: 0.5690 (0.6518) acc1: 63.8021 (62.1915) acc5: 100.0000 (100.0000) acc1_neg: 63.388 acc5_neg: 100.000 time: 0.1251 data: 0.0324 max mem: 3532 +Test: Total time: 0:01:15 (0.1942 s / it) +* Overall Acc@1 62.192 Acc@5 100.000 + target == real Acc@1 70.602 Acc@5 100.000 + target == fake Acc@1 58.282 Acc@5 100.000 + loss 0.652 + PR-AUC 0.8241 ROC-AUC 0.6990 EER 35.90% (thr=0.4663) +Accuracy of the network on the 149358 test images: 62.2% +[rank0]:[W726 05:02:26.241184461 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:02:45.786000 3161027 torch/distributed/run.py:766] +W0726 05:02:45.786000 3161027 torch/distributed/run.py:766] ***************************************** +W0726 05:02:45.786000 3161027 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:02:45.786000 3161027 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 05:03:12.297767564 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +[rank3]:[W726 05:03:13.795341023 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:03:13.840371541 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:03:13.928521477 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 5): env:// +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:03:13.337057782 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:03:13.342207592 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +[rank7]:[W726 05:03:13.354731612 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:03:13.381250289 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n32_seed7_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed7_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:26:41 acc1_pos: 77.344 acc5_pos: 100.000 loss: 0.4669 (0.4669) acc1: 77.3438 (77.3438) acc5: 100.0000 (100.0000) time: 4.8094 data: 2.7758 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/333] eta: 0:02:47 acc1_pos: 83.073 acc5_pos: 100.000 loss: 0.3528 (0.3568) acc1: 83.0729 (82.9072) acc5: 100.0000 (100.0000) time: 0.5187 data: 0.2526 max mem: 3531 +Test: [ 20/333] eta: 0:01:41 acc1_pos: 92.448 acc5_pos: 100.000 loss: 0.3462 (0.3458) acc1: 83.3333 (83.8666) acc5: 100.0000 (100.0000) time: 0.0987 data: 0.0092 max mem: 3531 +Test: [ 30/333] eta: 0:01:21 acc1_pos: 54.948 acc5_pos: 100.000 loss: 0.4241 (0.4728) acc1: 74.7396 (75.0168) acc5: 100.0000 (100.0000) time: 0.1339 data: 0.0439 max mem: 3531 +Test: [ 40/333] eta: 0:01:18 acc1_pos: 55.208 acc5_pos: 100.000 loss: 0.7527 (0.5414) acc1: 54.9479 (69.4423) acc5: 100.0000 (100.0000) time: 0.2079 data: 0.1154 max mem: 3531 +Test: [ 50/333] eta: 0:01:13 acc1_pos: 58.333 acc5_pos: 100.000 loss: 0.7440 (0.5801) acc1: 55.2083 (66.6667) acc5: 100.0000 (100.0000) time: 0.2407 data: 0.1477 max mem: 3531 +Test: [ 60/333] eta: 0:01:08 acc1_pos: 44.271 acc5_pos: 100.000 loss: 0.7207 (0.6080) acc1: 55.9896 (64.6602) acc5: 100.0000 (100.0000) time: 0.2219 data: 0.1277 max mem: 3531 +Test: [ 70/333] eta: 0:01:05 acc1_pos: 58.073 acc5_pos: 100.000 loss: 0.7127 (0.6272) acc1: 56.5104 (63.3289) acc5: 100.0000 (100.0000) time: 0.2225 data: 0.1263 max mem: 3531 +Test: [ 80/333] eta: 0:01:01 acc1_pos: 55.469 acc5_pos: 100.000 loss: 0.7474 (0.6421) acc1: 55.4688 (62.0081) acc5: 100.0000 (100.0000) time: 0.2180 data: 0.1234 max mem: 3531 +Test: [ 90/333] eta: 0:00:57 acc1_pos: 66.146 acc5_pos: 100.000 loss: 0.6737 (0.6402) acc1: 57.0312 (62.1509) acc5: 100.0000 (100.0000) time: 0.1944 data: 0.0987 max mem: 3531 +Test: [100/333] eta: 0:00:53 acc1_pos: 74.479 acc5_pos: 100.000 loss: 0.6145 (0.6381) acc1: 66.1458 (62.4562) acc5: 100.0000 (100.0000) time: 0.1768 data: 0.0806 max mem: 3531 +Test: [110/333] eta: 0:00:50 acc1_pos: 67.969 acc5_pos: 100.000 loss: 0.6145 (0.6373) acc1: 63.5417 (62.5023) acc5: 100.0000 (100.0000) time: 0.1848 data: 0.0882 max mem: 3531 +Test: [120/333] eta: 0:00:47 acc1_pos: 64.844 acc5_pos: 100.000 loss: 0.6122 (0.6349) acc1: 63.0208 (62.7475) acc5: 100.0000 (100.0000) time: 0.1780 data: 0.0823 max mem: 3531 +Test: [130/333] eta: 0:00:44 acc1_pos: 73.698 acc5_pos: 100.000 loss: 0.5794 (0.6318) acc1: 64.8438 (63.0705) acc5: 100.0000 (100.0000) time: 0.1649 data: 0.0708 max mem: 3531 +Test: [140/333] eta: 0:00:41 acc1_pos: 82.031 acc5_pos: 100.000 loss: 0.5557 (0.6262) acc1: 69.2708 (63.5491) acc5: 100.0000 (100.0000) time: 0.1747 data: 0.0794 max mem: 3531 +Test: [150/333] eta: 0:00:38 acc1_pos: 73.958 acc5_pos: 100.000 loss: 0.5744 (0.6257) acc1: 66.9271 (63.5865) acc5: 100.0000 (100.0000) time: 0.1814 data: 0.0882 max mem: 3531 +Test: [160/333] eta: 0:00:36 acc1_pos: 60.197 acc5_pos: 100.000 loss: 0.6312 (0.6349) acc1: 60.1562 (63.0985) acc5: 100.0000 (100.0000) acc1_neg: 38.802 acc5_neg: 100.000 time: 0.1806 data: 0.0901 max mem: 3531 +Test: [170/333] eta: 0:00:34 acc1_pos: 60.197 acc5_pos: 100.000 loss: 0.9212 (0.6539) acc1: 43.2292 (61.8330) acc5: 100.0000 (100.0000) acc1_neg: 46.094 acc5_neg: 100.000 time: 0.2084 data: 0.1134 max mem: 3531 +Test: [180/333] eta: 0:00:32 acc1_pos: 60.197 acc5_pos: 100.000 loss: 0.8994 (0.6650) acc1: 43.2292 (60.9404) acc5: 100.0000 (100.0000) acc1_neg: 48.958 acc5_neg: 100.000 time: 0.2161 data: 0.1203 max mem: 3531 +Test: [190/333] eta: 0:00:30 acc1_pos: 60.197 acc5_pos: 100.000 loss: 0.7828 (0.6688) acc1: 46.6146 (60.6444) acc5: 100.0000 (100.0000) acc1_neg: 51.302 acc5_neg: 100.000 time: 0.1916 data: 0.1001 max mem: 3531 +Test: [200/333] eta: 0:00:27 acc1_pos: 60.197 acc5_pos: 100.000 loss: 0.7618 (0.6744) acc1: 53.6458 (60.3467) acc5: 100.0000 (100.0000) acc1_neg: 53.125 acc5_neg: 100.000 time: 0.1988 data: 0.1080 max mem: 3531 +Test: [210/333] eta: 0:00:25 acc1_pos: 60.197 acc5_pos: 100.000 loss: 0.8346 (0.6837) acc1: 51.5625 (59.8168) acc5: 100.0000 (100.0000) acc1_neg: 48.958 acc5_neg: 100.000 time: 0.1842 data: 0.0933 max mem: 3532 +Test: [220/333] eta: 0:00:23 acc1_pos: 60.197 acc5_pos: 100.000 loss: 0.8602 (0.6937) acc1: 49.4792 (59.1134) acc5: 100.0000 (100.0000) acc1_neg: 60.156 acc5_neg: 100.000 time: 0.1823 data: 0.0910 max mem: 3532 +Test: [230/333] eta: 0:00:21 acc1_pos: 60.197 acc5_pos: 100.000 loss: 0.6021 (0.6883) acc1: 67.1875 (59.6816) acc5: 100.0000 (100.0000) acc1_neg: 71.094 acc5_neg: 100.000 time: 0.2142 data: 0.1216 max mem: 3532 +Test: [240/333] eta: 0:00:19 acc1_pos: 60.197 acc5_pos: 100.000 loss: 0.5940 (0.6875) acc1: 69.7917 (59.7392) acc5: 100.0000 (100.0000) acc1_neg: 60.156 acc5_neg: 100.000 time: 0.1866 data: 0.0910 max mem: 3532 +Test: [250/333] eta: 0:00:17 acc1_pos: 60.197 acc5_pos: 100.000 loss: 0.6867 (0.6890) acc1: 58.3333 (59.6520) acc5: 100.0000 (100.0000) acc1_neg: 48.438 acc5_neg: 100.000 time: 0.1946 data: 0.0982 max mem: 3532 +Test: [260/333] eta: 0:00:15 acc1_pos: 60.197 acc5_pos: 100.000 loss: 0.7319 (0.6925) acc1: 57.5521 (59.4508) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.2049 data: 0.1094 max mem: 3532 +Test: [270/333] eta: 0:00:12 acc1_pos: 60.197 acc5_pos: 100.000 loss: 0.7625 (0.6954) acc1: 53.9062 (59.2337) acc5: 100.0000 (100.0000) acc1_neg: 54.167 acc5_neg: 100.000 time: 0.1749 data: 0.0806 max mem: 3532 +Test: [280/333] eta: 0:00:10 acc1_pos: 60.197 acc5_pos: 100.000 loss: 0.7707 (0.7012) acc1: 51.8229 (58.7615) acc5: 100.0000 (100.0000) acc1_neg: 54.688 acc5_neg: 100.000 time: 0.2049 data: 0.1102 max mem: 3532 +Test: [290/333] eta: 0:00:08 acc1_pos: 60.197 acc5_pos: 100.000 loss: 0.9258 (0.7098) acc1: 40.8854 (58.1150) acc5: 100.0000 (100.0000) acc1_neg: 40.885 acc5_neg: 100.000 time: 0.2113 data: 0.1173 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 60.197 acc5_pos: 100.000 loss: 0.9548 (0.7182) acc1: 38.8021 (57.5010) acc5: 100.0000 (100.0000) acc1_neg: 43.490 acc5_neg: 100.000 time: 0.1702 data: 0.0790 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 60.197 acc5_pos: 100.000 loss: 0.8706 (0.7167) acc1: 44.5312 (57.6568) acc5: 100.0000 (100.0000) acc1_neg: 70.312 acc5_neg: 100.000 time: 0.1423 data: 0.0524 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 60.197 acc5_pos: 100.000 loss: 0.6104 (0.7149) acc1: 66.4062 (57.8782) acc5: 100.0000 (100.0000) acc1_neg: 37.760 acc5_neg: 100.000 time: 0.1336 data: 0.0438 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 60.197 acc5_pos: 100.000 loss: 0.9941 (0.7255) acc1: 41.1458 (57.2932) acc5: 100.0000 (100.0000) acc1_neg: 42.969 acc5_neg: 100.000 time: 0.1258 data: 0.0354 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 60.197 acc5_pos: 100.000 loss: 1.0280 (0.7274) acc1: 39.7183 (57.1905) acc5: 100.0000 (100.0000) acc1_neg: 39.718 acc5_neg: 100.000 time: 0.1289 data: 0.0354 max mem: 3532 +Test: Total time: 0:01:05 (0.1967 s / it) +* Overall Acc@1 57.190 Acc@5 100.000 + target == real Acc@1 63.822 Acc@5 100.000 + target == fake Acc@1 51.287 Acc@5 100.000 + loss 0.727 + PR-AUC 0.6166 ROC-AUC 0.6120 EER 42.23% (thr=0.4565) +Accuracy of the network on the 127843 test images: 57.2% +[rank0]:[W726 05:04:28.117664920 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:04:48.320000 3161586 torch/distributed/run.py:766] +W0726 05:04:48.320000 3161586 torch/distributed/run.py:766] ***************************************** +W0726 05:04:48.320000 3161586 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:04:48.320000 3161586 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:05:16.401134257 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:05:17.842046127 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:05:17.068198480 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:05:17.302313931 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:05:17.318070889 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:05:17.401994619 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:05:17.448491777 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:05:18.628889670 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n32_seed7_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed7_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:17:48 acc1_pos: 95.573 acc5_pos: 100.000 loss: 0.1934 (0.1934) acc1: 95.5729 (95.5729) acc5: 100.0000 (100.0000) time: 4.5488 data: 2.6394 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/235] eta: 0:01:56 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.2071 (0.2050) acc1: 95.5729 (95.2888) acc5: 100.0000 (100.0000) time: 0.5188 data: 0.2638 max mem: 3531 +Test: [ 20/235] eta: 0:01:07 acc1_pos: 95.573 acc5_pos: 100.000 loss: 0.1987 (0.1999) acc1: 95.5729 (95.6473) acc5: 100.0000 (100.0000) time: 0.1029 data: 0.0133 max mem: 3531 +Test: [ 30/235] eta: 0:00:49 acc1_pos: 97.135 acc5_pos: 100.000 loss: 0.1927 (0.1983) acc1: 95.8333 (95.7325) acc5: 100.0000 (100.0000) time: 0.0915 data: 0.0004 max mem: 3531 +Test: [ 40/235] eta: 0:00:43 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.1927 (0.1978) acc1: 95.8333 (95.7444) acc5: 100.0000 (100.0000) time: 0.1264 data: 0.0317 max mem: 3531 +Test: [ 50/235] eta: 0:00:38 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.1982 (0.1988) acc1: 95.5729 (95.7006) acc5: 100.0000 (100.0000) time: 0.1599 data: 0.0655 max mem: 3531 +Test: [ 60/235] eta: 0:00:35 acc1_pos: 94.010 acc5_pos: 100.000 loss: 0.2038 (0.2002) acc1: 95.5729 (95.6668) acc5: 100.0000 (100.0000) time: 0.1649 data: 0.0711 max mem: 3531 +Test: [ 70/235] eta: 0:00:32 acc1_pos: 98.462 acc5_pos: 100.000 loss: 0.2176 (0.3759) acc1: 94.0104 (85.7614) acc5: 100.0000 (100.0000) acc1_neg: 14.844 acc5_neg: 100.000 time: 0.1699 data: 0.0738 max mem: 3531 +Test: [ 80/235] eta: 0:00:29 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.6067 (0.5303) acc1: 15.8854 (77.0544) acc5: 100.0000 (100.0000) acc1_neg: 14.844 acc5_neg: 100.000 time: 0.1512 data: 0.0566 max mem: 3531 +Test: [ 90/235] eta: 0:00:27 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.6308 (0.6492) acc1: 14.8438 (70.2524) acc5: 100.0000 (100.0000) acc1_neg: 14.323 acc5_neg: 100.000 time: 0.1499 data: 0.0570 max mem: 3531 +Test: [100/235] eta: 0:00:25 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.6119 (0.7440) acc1: 15.1042 (64.9237) acc5: 100.0000 (100.0000) acc1_neg: 14.844 acc5_neg: 100.000 time: 0.1794 data: 0.0842 max mem: 3531 +Test: [110/235] eta: 0:00:23 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.5898 (0.8205) acc1: 16.6667 (60.5903) acc5: 100.0000 (100.0000) acc1_neg: 18.229 acc5_neg: 100.000 time: 0.1904 data: 0.0942 max mem: 3531 +Test: [120/235] eta: 0:00:21 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.6422 (0.8985) acc1: 13.8021 (56.4114) acc5: 100.0000 (100.0000) acc1_neg: 10.417 acc5_neg: 100.000 time: 0.1591 data: 0.0631 max mem: 3531 +Test: [130/235] eta: 0:00:19 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.7717 (0.9654) acc1: 9.3750 (52.8069) acc5: 100.0000 (100.0000) acc1_neg: 10.156 acc5_neg: 100.000 time: 0.1663 data: 0.0704 max mem: 3531 +Test: [140/235] eta: 0:00:17 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.7851 (1.0249) acc1: 9.1146 (49.7156) acc5: 100.0000 (100.0000) acc1_neg: 7.552 acc5_neg: 100.000 time: 0.1663 data: 0.0725 max mem: 3531 +Test: [150/235] eta: 0:00:15 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.7942 (1.0749) acc1: 9.1146 (47.0302) acc5: 100.0000 (100.0000) acc1_neg: 9.115 acc5_neg: 100.000 time: 0.1389 data: 0.0456 max mem: 3531 +Test: [160/235] eta: 0:00:13 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.7840 (1.1183) acc1: 9.1146 (44.7124) acc5: 100.0000 (100.0000) acc1_neg: 10.677 acc5_neg: 100.000 time: 0.1744 data: 0.0800 max mem: 3531 +Test: [170/235] eta: 0:00:11 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.7306 (1.1410) acc1: 11.4583 (43.2703) acc5: 100.0000 (100.0000) acc1_neg: 17.969 acc5_neg: 100.000 time: 0.1745 data: 0.0795 max mem: 3531 +Test: [180/235] eta: 0:00:09 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4652 (1.1587) acc1: 22.1354 (42.1487) acc5: 100.0000 (100.0000) acc1_neg: 22.396 acc5_neg: 100.000 time: 0.1754 data: 0.0801 max mem: 3531 +Test: [190/235] eta: 0:00:08 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4446 (1.1744) acc1: 22.3958 (41.0927) acc5: 100.0000 (100.0000) acc1_neg: 22.135 acc5_neg: 100.000 time: 0.1949 data: 0.1001 max mem: 3531 +Test: [200/235] eta: 0:00:06 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4517 (1.1898) acc1: 21.3542 (40.1094) acc5: 100.0000 (100.0000) acc1_neg: 23.698 acc5_neg: 100.000 time: 0.1703 data: 0.0756 max mem: 3531 +Test: [210/235] eta: 0:00:04 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4650 (1.2024) acc1: 20.8333 (39.2266) acc5: 100.0000 (100.0000) acc1_neg: 19.531 acc5_neg: 100.000 time: 0.1361 data: 0.0409 max mem: 3532 +Test: [220/235] eta: 0:00:02 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4573 (1.2138) acc1: 21.3542 (38.4545) acc5: 100.0000 (100.0000) acc1_neg: 21.615 acc5_neg: 100.000 time: 0.1633 data: 0.0709 max mem: 3532 +Test: [230/235] eta: 0:00:00 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4458 (1.2237) acc1: 22.1354 (37.7920) acc5: 100.0000 (100.0000) acc1_neg: 21.875 acc5_neg: 100.000 time: 0.1654 data: 0.0729 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4579 (1.2291) acc1: 22.1354 (37.5750) acc5: 100.0000 (100.0000) acc1_neg: 30.435 acc5_neg: 100.000 time: 0.1571 data: 0.0675 max mem: 3532 +Test: Total time: 0:00:40 (0.1744 s / it) +* Overall Acc@1 37.575 Acc@5 100.000 + target == real Acc@1 95.669 Acc@5 100.000 + target == fake Acc@1 16.564 Acc@5 100.000 + loss 1.229 + PR-AUC 0.8476 ROC-AUC 0.6742 EER 37.48% (thr=0.1471) +Accuracy of the network on the 89879 test images: 37.6% +[rank0]:[W726 05:06:15.247682898 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:06:34.805000 3162148 torch/distributed/run.py:766] +W0726 05:06:34.805000 3162148 torch/distributed/run.py:766] ***************************************** +W0726 05:06:34.805000 3162148 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:06:34.805000 3162148 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:07:02.486282578 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:07:03.904231492 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:07:03.917100459 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:07:03.079393548 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +[rank0]:[W726 05:07:03.545114475 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:07:03.552267715 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:07:04.583478043 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:07:04.728956994 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n32_seed7_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed7_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:27:22 acc1_pos: 77.604 acc5_pos: 100.000 loss: 0.5252 (0.5252) acc1: 77.6042 (77.6042) acc5: 100.0000 (100.0000) time: 6.1059 data: 4.7269 max mem: 3531 +Test: [ 10/269] eta: 0:02:45 acc1_pos: 92.969 acc5_pos: 100.000 loss: 0.5252 (0.5183) acc1: 76.8229 (77.3674) acc5: 100.0000 (100.0000) time: 0.6398 data: 0.4300 max mem: 3531 +Test: [ 20/269] eta: 0:01:35 acc1_pos: 79.948 acc5_pos: 100.000 loss: 0.5121 (0.5058) acc1: 77.8646 (79.1295) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0003 max mem: 3531 +Test: [ 30/269] eta: 0:01:15 acc1_pos: 89.323 acc5_pos: 100.000 loss: 0.5051 (0.4984) acc1: 79.9479 (79.7547) acc5: 100.0000 (100.0000) time: 0.1368 data: 0.0388 max mem: 3531 +Test: [ 40/269] eta: 0:01:06 acc1_pos: 64.655 acc5_pos: 100.000 loss: 0.6125 (0.5465) acc1: 66.9271 (74.2950) acc5: 100.0000 (100.0000) acc1_neg: 51.042 acc5_neg: 100.000 time: 0.1958 data: 0.1015 max mem: 3531 +Test: [ 50/269] eta: 0:01:00 acc1_pos: 64.655 acc5_pos: 100.000 loss: 0.7451 (0.6051) acc1: 47.9167 (67.5143) acc5: 100.0000 (100.0000) acc1_neg: 18.750 acc5_neg: 100.000 time: 0.2131 data: 0.1204 max mem: 3531 +Test: [ 60/269] eta: 0:00:55 acc1_pos: 64.655 acc5_pos: 100.000 loss: 0.8671 (0.6551) acc1: 32.5521 (61.8084) acc5: 100.0000 (100.0000) acc1_neg: 31.771 acc5_neg: 100.000 time: 0.2140 data: 0.1188 max mem: 3531 +Test: [ 70/269] eta: 0:00:50 acc1_pos: 64.655 acc5_pos: 100.000 loss: 0.8671 (0.6760) acc1: 36.9792 (59.3310) acc5: 100.0000 (100.0000) acc1_neg: 51.823 acc5_neg: 100.000 time: 0.2049 data: 0.1087 max mem: 3531 +Test: [ 80/269] eta: 0:00:47 acc1_pos: 64.655 acc5_pos: 100.000 loss: 0.7552 (0.6810) acc1: 48.6979 (58.7127) acc5: 100.0000 (100.0000) acc1_neg: 66.146 acc5_neg: 100.000 time: 0.1999 data: 0.1071 max mem: 3531 +Test: [ 90/269] eta: 0:00:43 acc1_pos: 64.655 acc5_pos: 100.000 loss: 0.7715 (0.6964) acc1: 46.0938 (56.9397) acc5: 100.0000 (100.0000) acc1_neg: 41.406 acc5_neg: 100.000 time: 0.2112 data: 0.1196 max mem: 3531 +Test: [100/269] eta: 0:00:41 acc1_pos: 64.655 acc5_pos: 100.000 loss: 0.8085 (0.7094) acc1: 40.6250 (55.1697) acc5: 100.0000 (100.0000) acc1_neg: 21.354 acc5_neg: 100.000 time: 0.2153 data: 0.1240 max mem: 3531 +Test: [110/269] eta: 0:00:38 acc1_pos: 64.655 acc5_pos: 100.000 loss: 0.8466 (0.7239) acc1: 37.2396 (53.3057) acc5: 100.0000 (100.0000) acc1_neg: 30.208 acc5_neg: 100.000 time: 0.2287 data: 0.1385 max mem: 3531 +Test: [120/269] eta: 0:00:35 acc1_pos: 64.655 acc5_pos: 100.000 loss: 0.8493 (0.7319) acc1: 37.2396 (52.6171) acc5: 100.0000 (100.0000) acc1_neg: 47.396 acc5_neg: 100.000 time: 0.2269 data: 0.1363 max mem: 3531 +Test: [130/269] eta: 0:00:33 acc1_pos: 64.655 acc5_pos: 100.000 loss: 0.8845 (0.7471) acc1: 39.5833 (51.1729) acc5: 100.0000 (100.0000) acc1_neg: 34.635 acc5_neg: 100.000 time: 0.2126 data: 0.1191 max mem: 3531 +Test: [140/269] eta: 0:00:30 acc1_pos: 64.655 acc5_pos: 100.000 loss: 0.9065 (0.7600) acc1: 34.3750 (49.8135) acc5: 100.0000 (100.0000) acc1_neg: 34.375 acc5_neg: 100.000 time: 0.2047 data: 0.1108 max mem: 3531 +Test: [150/269] eta: 0:00:28 acc1_pos: 64.655 acc5_pos: 100.000 loss: 0.8637 (0.7585) acc1: 34.8958 (49.9655) acc5: 100.0000 (100.0000) acc1_neg: 68.229 acc5_neg: 100.000 time: 0.2321 data: 0.1395 max mem: 3531 +Test: [160/269] eta: 0:00:25 acc1_pos: 64.655 acc5_pos: 100.000 loss: 0.7231 (0.7551) acc1: 49.2188 (50.3170) acc5: 100.0000 (100.0000) acc1_neg: 47.917 acc5_neg: 100.000 time: 0.2176 data: 0.1232 max mem: 3531 +Test: [170/269] eta: 0:00:22 acc1_pos: 64.655 acc5_pos: 100.000 loss: 0.7665 (0.7609) acc1: 47.1354 (49.7183) acc5: 100.0000 (100.0000) acc1_neg: 27.865 acc5_neg: 100.000 time: 0.1900 data: 0.0971 max mem: 3531 +Test: [180/269] eta: 0:00:20 acc1_pos: 64.655 acc5_pos: 100.000 loss: 0.7889 (0.7616) acc1: 41.4062 (49.5712) acc5: 100.0000 (100.0000) acc1_neg: 29.948 acc5_neg: 100.000 time: 0.2037 data: 0.1108 max mem: 3531 +Test: [190/269] eta: 0:00:18 acc1_pos: 64.655 acc5_pos: 100.000 loss: 0.7802 (0.7656) acc1: 44.5312 (49.0306) acc5: 100.0000 (100.0000) acc1_neg: 57.292 acc5_neg: 100.000 time: 0.2045 data: 0.1091 max mem: 3531 +Test: [200/269] eta: 0:00:15 acc1_pos: 64.655 acc5_pos: 100.000 loss: 0.8313 (0.7700) acc1: 39.0625 (48.6448) acc5: 100.0000 (100.0000) acc1_neg: 35.156 acc5_neg: 100.000 time: 0.2049 data: 0.1084 max mem: 3531 +Test: [210/269] eta: 0:00:13 acc1_pos: 64.655 acc5_pos: 100.000 loss: 0.7988 (0.7714) acc1: 42.4479 (48.5634) acc5: 100.0000 (100.0000) acc1_neg: 47.917 acc5_neg: 100.000 time: 0.2103 data: 0.1151 max mem: 3532 +Test: [220/269] eta: 0:00:11 acc1_pos: 64.655 acc5_pos: 100.000 loss: 0.7242 (0.7700) acc1: 50.2604 (48.8051) acc5: 100.0000 (100.0000) acc1_neg: 72.656 acc5_neg: 100.000 time: 0.2262 data: 0.1345 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 64.655 acc5_pos: 100.000 loss: 0.6639 (0.7658) acc1: 53.9062 (49.2142) acc5: 100.0000 (100.0000) acc1_neg: 51.562 acc5_neg: 100.000 time: 0.1892 data: 0.0991 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 64.655 acc5_pos: 100.000 loss: 0.7722 (0.7727) acc1: 39.0625 (48.3910) acc5: 100.0000 (100.0000) acc1_neg: 33.594 acc5_neg: 100.000 time: 0.1251 data: 0.0354 max mem: 3532 +Test: [250/269] eta: 0:00:04 acc1_pos: 64.655 acc5_pos: 100.000 loss: 0.9489 (0.7796) acc1: 26.0417 (47.6967) acc5: 100.0000 (100.0000) acc1_neg: 10.156 acc5_neg: 100.000 time: 0.1093 data: 0.0196 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 64.655 acc5_pos: 100.000 loss: 1.0578 (0.7902) acc1: 15.3646 (46.6086) acc5: 100.0000 (100.0000) acc1_neg: 38.021 acc5_neg: 100.000 time: 0.1185 data: 0.0289 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 64.655 acc5_pos: 100.000 loss: 0.9678 (0.7911) acc1: 23.4375 (46.4368) acc5: 100.0000 (100.0000) acc1_neg: 67.454 acc5_neg: 100.000 time: 0.1208 data: 0.0275 max mem: 3532 +Test: Total time: 0:00:56 (0.2093 s / it) +* Overall Acc@1 46.437 Acc@5 100.000 + target == real Acc@1 79.208 Acc@5 100.000 + target == fake Acc@1 41.964 Acc@5 100.000 + loss 0.791 + PR-AUC 0.9290 ROC-AUC 0.6649 EER 38.18% (thr=0.4031) +Accuracy of the network on the 103293 test images: 46.4% +[rank0]:[W726 05:08:09.708336165 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T05:08:15Z] eval_done tag=fixed500k_fake_n32_seed7_resampled +[2026-07-26T05:08:15Z] eval_start tag=fixed500k_real_n16_seed8_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed8_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T05:08:15Z +hostname=n111-046-081 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed8_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n16_seed8_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 05:08:28.752000 3162713 torch/distributed/run.py:766] +W0726 05:08:28.752000 3162713 torch/distributed/run.py:766] ***************************************** +W0726 05:08:28.752000 3162713 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:08:28.752000 3162713 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:08:57.742836874 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:08:57.764246823 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:08:57.246713881 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:08:57.264422388 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:08:57.475833158 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:08:57.513558004 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:08:58.615302698 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W726 05:08:58.615515849 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n16_seed8_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed8_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:27:31 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.9713 (1.9713) acc1: 0.0000 (0.0000) acc5: 100.0000 (100.0000) time: 7.0563 data: 5.3837 max mem: 3531 +Test: [ 10/234] eta: 0:02:42 acc1_pos: 1.562 acc5_pos: 100.000 loss: 1.8749 (1.7400) acc1: 1.5625 (7.2680) acc5: 100.0000 (100.0000) time: 0.7239 data: 0.4899 max mem: 3531 +Test: [ 20/234] eta: 0:01:31 acc1_pos: 96.094 acc5_pos: 100.000 loss: 1.7258 (1.5194) acc1: 4.4271 (20.8705) acc5: 100.0000 (100.0000) time: 0.0952 data: 0.0004 max mem: 3531 +Test: [ 30/234] eta: 0:01:14 acc1_pos: 50.521 acc5_pos: 100.000 loss: 0.2236 (1.1305) acc1: 95.0521 (42.5571) acc5: 100.0000 (100.0000) time: 0.1699 data: 0.0689 max mem: 3531 +Test: [ 40/234] eta: 0:01:03 acc1_pos: 90.885 acc5_pos: 100.000 loss: 0.2737 (1.0055) acc1: 89.5833 (47.9294) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1247 max mem: 3531 +Test: [ 50/234] eta: 0:00:55 acc1_pos: 92.188 acc5_pos: 100.000 loss: 0.2785 (0.8601) acc1: 90.6250 (56.5462) acc5: 100.0000 (100.0000) time: 0.2049 data: 0.1098 max mem: 3531 +Test: [ 60/234] eta: 0:00:50 acc1_pos: 90.345 acc5_pos: 100.000 loss: 0.1729 (0.7355) acc1: 93.7500 (63.6057) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2099 data: 0.1158 max mem: 3531 +Test: [ 70/234] eta: 0:00:44 acc1_pos: 90.345 acc5_pos: 100.000 loss: 0.0955 (0.6991) acc1: 100.0000 (65.2435) acc5: 100.0000 (100.0000) acc1_neg: 60.156 acc5_neg: 100.000 time: 0.1999 data: 0.1057 max mem: 3531 +Test: [ 80/234] eta: 0:00:41 acc1_pos: 90.345 acc5_pos: 100.000 loss: 0.6203 (0.6678) acc1: 64.8438 (66.6763) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2049 data: 0.1125 max mem: 3531 +Test: [ 90/234] eta: 0:00:37 acc1_pos: 90.345 acc5_pos: 100.000 loss: 0.1682 (0.6120) acc1: 93.2292 (69.8747) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2199 data: 0.1255 max mem: 3531 +Test: [100/234] eta: 0:00:33 acc1_pos: 90.345 acc5_pos: 100.000 loss: 0.2138 (0.5860) acc1: 91.4062 (71.3310) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1700 data: 0.0751 max mem: 3531 +Test: [110/234] eta: 0:00:30 acc1_pos: 90.345 acc5_pos: 100.000 loss: 0.3173 (0.5579) acc1: 86.1979 (72.9964) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.1664 data: 0.0743 max mem: 3531 +Test: [120/234] eta: 0:00:27 acc1_pos: 90.345 acc5_pos: 100.000 loss: 0.2841 (0.5395) acc1: 89.0625 (74.0401) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1948 data: 0.1010 max mem: 3531 +Test: [130/234] eta: 0:00:24 acc1_pos: 90.345 acc5_pos: 100.000 loss: 0.1982 (0.5176) acc1: 94.0104 (75.4632) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1894 data: 0.0944 max mem: 3531 +Test: [140/234] eta: 0:00:21 acc1_pos: 90.345 acc5_pos: 100.000 loss: 0.2385 (0.5227) acc1: 86.1979 (75.1644) acc5: 100.0000 (100.0000) acc1_neg: 59.375 acc5_neg: 100.000 time: 0.1799 data: 0.0876 max mem: 3531 +Test: [150/234] eta: 0:00:19 acc1_pos: 90.345 acc5_pos: 100.000 loss: 0.3477 (0.5127) acc1: 82.8125 (75.7140) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.1715 data: 0.0787 max mem: 3531 +Test: [160/234] eta: 0:00:16 acc1_pos: 90.345 acc5_pos: 100.000 loss: 0.3101 (0.4960) acc1: 89.0625 (76.8164) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.1849 data: 0.0915 max mem: 3531 +Test: [170/234] eta: 0:00:14 acc1_pos: 90.345 acc5_pos: 100.000 loss: 0.2880 (0.4837) acc1: 91.1458 (77.6697) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1891 data: 0.0974 max mem: 3531 +Test: [180/234] eta: 0:00:11 acc1_pos: 90.345 acc5_pos: 100.000 loss: 0.3310 (0.4826) acc1: 87.5000 (77.6142) acc5: 100.0000 (100.0000) acc1_neg: 56.510 acc5_neg: 100.000 time: 0.1766 data: 0.0862 max mem: 3531 +Test: [190/234] eta: 0:00:09 acc1_pos: 90.345 acc5_pos: 100.000 loss: 0.6056 (0.4898) acc1: 64.0625 (76.9293) acc5: 100.0000 (100.0000) acc1_neg: 61.979 acc5_neg: 100.000 time: 0.1731 data: 0.0826 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 90.345 acc5_pos: 100.000 loss: 0.5896 (0.4801) acc1: 66.1458 (77.4539) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1656 data: 0.0744 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 90.345 acc5_pos: 100.000 loss: 0.1113 (0.4629) acc1: 99.7396 (78.4940) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1425 data: 0.0516 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 90.345 acc5_pos: 100.000 loss: 0.1319 (0.4500) acc1: 98.6979 (79.2692) acc5: 100.0000 (100.0000) acc1_neg: 73.698 acc5_neg: 100.000 time: 0.1271 data: 0.0372 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 90.345 acc5_pos: 100.000 loss: 0.4932 (0.4827) acc1: 23.9583 (76.7327) acc5: 100.0000 (100.0000) acc1_neg: 19.271 acc5_neg: 100.000 time: 0.1241 data: 0.0343 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 90.345 acc5_pos: 100.000 loss: 1.1763 (0.4916) acc1: 22.5681 (76.1047) acc5: 100.0000 (100.0000) acc1_neg: 22.568 acc5_neg: 100.000 time: 0.1246 data: 0.0326 max mem: 3532 +Test: Total time: 0:00:47 (0.2032 s / it) +* Overall Acc@1 76.105 Acc@5 100.000 + target == real Acc@1 56.795 Acc@5 100.000 + target == fake Acc@1 81.547 Acc@5 100.000 + loss 0.492 + PR-AUC 0.9178 ROC-AUC 0.7851 EER 30.18% (thr=0.7285) +Accuracy of the network on the 89729 test images: 76.1% +[rank0]:[W726 05:09:56.056127906 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:10:15.556000 3163272 torch/distributed/run.py:766] +W0726 05:10:15.556000 3163272 torch/distributed/run.py:766] ***************************************** +W0726 05:10:15.556000 3163272 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:10:15.556000 3163272 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:10:42.672272494 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:10:42.095027631 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:10:42.195960562 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:10:42.195971713 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:10:42.555383764 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +| distributed init (rank 3): env:// +[rank7]:[W726 05:10:43.586519853 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:10:43.595660138 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank6]:[W726 05:10:43.596786625 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n16_seed8_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed8_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:38:41 acc1_pos: 2.604 acc5_pos: 100.000 loss: 1.7590 (1.7590) acc1: 2.6042 (2.6042) acc5: 100.0000 (100.0000) time: 7.0051 data: 5.1624 max mem: 3531 +Test: [ 10/2387] eta: 0:28:28 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.9687 (1.9141) acc1: 2.6042 (2.7699) acc5: 100.0000 (100.0000) time: 0.7186 data: 0.4696 max mem: 3531 +Test: [ 20/2387] eta: 0:16:42 acc1_pos: 1.042 acc5_pos: 100.000 loss: 1.9745 (1.9465) acc1: 2.6042 (2.7902) acc5: 100.0000 (100.0000) time: 0.0946 data: 0.0003 max mem: 3531 +Test: [ 30/2387] eta: 0:14:21 acc1_pos: 5.469 acc5_pos: 100.000 loss: 1.9904 (1.9655) acc1: 1.0417 (2.2177) acc5: 100.0000 (100.0000) time: 0.1712 data: 0.0766 max mem: 3531 +Test: [ 40/2387] eta: 0:13:21 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.0138 (1.9825) acc1: 0.0000 (1.7721) acc5: 100.0000 (100.0000) time: 0.2550 data: 0.1612 max mem: 3531 +Test: [ 50/2387] eta: 0:13:00 acc1_pos: 7.812 acc5_pos: 100.000 loss: 1.9676 (1.9469) acc1: 0.5208 (2.2518) acc5: 100.0000 (100.0000) time: 0.2851 data: 0.1877 max mem: 3531 +Test: [ 60/2387] eta: 0:12:12 acc1_pos: 1.562 acc5_pos: 100.000 loss: 1.9624 (1.9593) acc1: 1.8229 (2.2199) acc5: 100.0000 (100.0000) time: 0.2603 data: 0.1660 max mem: 3531 +Test: [ 70/2387] eta: 0:11:38 acc1_pos: 6.510 acc5_pos: 100.000 loss: 2.0725 (1.9878) acc1: 1.0417 (2.1347) acc5: 100.0000 (100.0000) time: 0.2181 data: 0.1238 max mem: 3531 +Test: [ 80/2387] eta: 0:11:15 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.1255 (2.0023) acc1: 0.2604 (1.9579) acc5: 100.0000 (100.0000) time: 0.2258 data: 0.1296 max mem: 3531 +Test: [ 90/2387] eta: 0:10:56 acc1_pos: 2.083 acc5_pos: 100.000 loss: 2.0755 (2.0085) acc1: 0.2604 (1.9689) acc5: 100.0000 (100.0000) time: 0.2311 data: 0.1378 max mem: 3531 +Test: [ 100/2387] eta: 0:10:45 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.0857 (2.0195) acc1: 0.2604 (1.8255) acc5: 100.0000 (100.0000) time: 0.2391 data: 0.1466 max mem: 3531 +Test: [ 110/2387] eta: 0:10:29 acc1_pos: 0.521 acc5_pos: 100.000 loss: 2.0972 (2.0232) acc1: 0.5208 (1.8956) acc5: 100.0000 (100.0000) time: 0.2343 data: 0.1405 max mem: 3531 +Test: [ 120/2387] eta: 0:10:16 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.0611 (2.0266) acc1: 0.5208 (1.8939) acc5: 100.0000 (100.0000) time: 0.2202 data: 0.1237 max mem: 3531 +Test: [ 130/2387] eta: 0:10:09 acc1_pos: 4.948 acc5_pos: 100.000 loss: 2.0858 (2.0326) acc1: 0.5208 (1.8706) acc5: 100.0000 (100.0000) time: 0.2343 data: 0.1380 max mem: 3531 +Test: [ 140/2387] eta: 0:10:02 acc1_pos: 2.083 acc5_pos: 100.000 loss: 2.1025 (2.0372) acc1: 0.2604 (1.8358) acc5: 100.0000 (100.0000) time: 0.2440 data: 0.1476 max mem: 3531 +Test: [ 150/2387] eta: 0:09:54 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.0595 (2.0384) acc1: 0.5208 (1.9574) acc5: 100.0000 (100.0000) time: 0.2355 data: 0.1378 max mem: 3531 +Test: [ 160/2387] eta: 0:09:42 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.0284 (1.9978) acc1: 0.5208 (4.0292) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2154 data: 0.1190 max mem: 3531 +Test: [ 170/2387] eta: 0:09:32 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1191 (1.8877) acc1: 99.7396 (9.6308) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2011 data: 0.1048 max mem: 3531 +Test: [ 180/2387] eta: 0:09:24 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1149 (1.7898) acc1: 99.7396 (14.6107) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2106 data: 0.1149 max mem: 3531 +Test: [ 190/2387] eta: 0:09:11 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1153 (1.7022) acc1: 99.7396 (19.0622) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1897 data: 0.0948 max mem: 3531 +Test: [ 200/2387] eta: 0:08:57 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1146 (1.6231) acc1: 99.7396 (23.0825) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1548 data: 0.0605 max mem: 3531 +Test: [ 210/2387] eta: 0:08:50 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1146 (1.5518) acc1: 99.7396 (26.7032) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1738 data: 0.0820 max mem: 3532 +Test: [ 220/2387] eta: 0:08:45 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1169 (1.4868) acc1: 99.4792 (30.0033) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2100 data: 0.1172 max mem: 3532 +Test: [ 230/2387] eta: 0:08:39 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1149 (1.4274) acc1: 99.7396 (33.0233) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2099 data: 0.1163 max mem: 3532 +Test: [ 240/2387] eta: 0:08:34 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1061 (1.3725) acc1: 99.7396 (35.7992) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2056 data: 0.1129 max mem: 3532 +Test: [ 250/2387] eta: 0:08:28 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0990 (1.3214) acc1: 100.0000 (38.3539) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2070 data: 0.1124 max mem: 3532 +Test: [ 260/2387] eta: 0:08:23 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0866 (1.2740) acc1: 100.0000 (40.7158) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2053 data: 0.1113 max mem: 3532 +Test: [ 270/2387] eta: 0:08:17 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0835 (1.2301) acc1: 100.0000 (42.9024) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1984 data: 0.1071 max mem: 3532 +Test: [ 280/2387] eta: 0:08:16 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0839 (1.1894) acc1: 100.0000 (44.9344) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2226 data: 0.1313 max mem: 3532 +Test: [ 290/2387] eta: 0:08:09 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0839 (1.1513) acc1: 100.0000 (46.8267) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2102 data: 0.1171 max mem: 3532 +Test: [ 300/2387] eta: 0:08:10 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0823 (1.1158) acc1: 100.0000 (48.5932) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2219 data: 0.1281 max mem: 3532 +Test: [ 310/2387] eta: 0:08:05 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0843 (1.0855) acc1: 100.0000 (50.1708) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2434 data: 0.1512 max mem: 3532 +Test: [ 320/2387] eta: 0:08:04 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2643 (1.0602) acc1: 96.0938 (51.5836) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2234 data: 0.1321 max mem: 3532 +Test: [ 330/2387] eta: 0:08:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2713 (1.0363) acc1: 95.5729 (52.9220) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2252 data: 0.1336 max mem: 3532 +Test: [ 340/2387] eta: 0:07:54 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2713 (1.0138) acc1: 95.5729 (54.1758) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1893 data: 0.0960 max mem: 3532 +Test: [ 350/2387] eta: 0:07:50 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2600 (0.9877) acc1: 96.3542 (55.4636) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1861 data: 0.0913 max mem: 3532 +Test: [ 360/2387] eta: 0:07:46 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1034 (0.9632) acc1: 100.0000 (56.6951) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2016 data: 0.1086 max mem: 3532 +Test: [ 370/2387] eta: 0:07:43 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1066 (0.9402) acc1: 100.0000 (57.8581) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2096 data: 0.1172 max mem: 3532 +Test: [ 380/2387] eta: 0:07:39 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1081 (0.9183) acc1: 99.7396 (58.9587) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2141 data: 0.1199 max mem: 3532 +Test: [ 390/2387] eta: 0:07:33 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1076 (0.8976) acc1: 99.7396 (60.0051) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1800 data: 0.0868 max mem: 3532 +Test: [ 400/2387] eta: 0:07:28 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1071 (0.8779) acc1: 100.0000 (60.9998) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1610 data: 0.0704 max mem: 3532 +Test: [ 410/2387] eta: 0:07:23 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1068 (0.8591) acc1: 100.0000 (61.9456) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1690 data: 0.0763 max mem: 3532 +Test: [ 420/2387] eta: 0:07:18 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1050 (0.8412) acc1: 99.7396 (62.8452) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1691 data: 0.0758 max mem: 3532 +Test: [ 430/2387] eta: 0:07:14 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1027 (0.8240) acc1: 99.7396 (63.7048) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0880 max mem: 3532 +Test: [ 440/2387] eta: 0:07:12 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0968 (0.8073) acc1: 100.0000 (64.5261) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2088 data: 0.1164 max mem: 3533 +Test: [ 450/2387] eta: 0:07:09 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0793 (0.7912) acc1: 100.0000 (65.3126) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2110 data: 0.1187 max mem: 3533 +Test: [ 460/2387] eta: 0:07:06 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0789 (0.7757) acc1: 100.0000 (66.0651) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1960 data: 0.1014 max mem: 3533 +Test: [ 470/2387] eta: 0:07:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0789 (0.7609) acc1: 100.0000 (66.7855) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2045 data: 0.1091 max mem: 3533 +Test: [ 480/2387] eta: 0:07:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0789 (0.7467) acc1: 100.0000 (67.4761) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2056 data: 0.1125 max mem: 3533 +Test: [ 490/2387] eta: 0:06:57 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0790 (0.7332) acc1: 100.0000 (68.1385) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2021 data: 0.1102 max mem: 3533 +Test: [ 500/2387] eta: 0:06:54 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0794 (0.7201) acc1: 100.0000 (68.7744) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2008 data: 0.1063 max mem: 3533 +Test: [ 510/2387] eta: 0:06:51 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0795 (0.7076) acc1: 100.0000 (69.3855) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1966 data: 0.1019 max mem: 3533 +Test: [ 520/2387] eta: 0:06:48 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0943 (0.6959) acc1: 100.0000 (69.9731) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1945 data: 0.1028 max mem: 3533 +Test: [ 530/2387] eta: 0:06:44 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0956 (0.6846) acc1: 100.0000 (70.5386) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1875 data: 0.0957 max mem: 3533 +Test: [ 540/2387] eta: 0:06:41 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0958 (0.6737) acc1: 100.0000 (71.0832) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1828 data: 0.0914 max mem: 3533 +Test: [ 550/2387] eta: 0:06:38 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0960 (0.6632) acc1: 100.0000 (71.6080) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1893 data: 0.0990 max mem: 3533 +Test: [ 560/2387] eta: 0:06:36 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0957 (0.6531) acc1: 100.0000 (72.1141) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2009 data: 0.1104 max mem: 3533 +Test: [ 570/2387] eta: 0:06:33 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0954 (0.6433) acc1: 100.0000 (72.6024) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2029 data: 0.1117 max mem: 3533 +Test: [ 580/2387] eta: 0:06:30 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0954 (0.6339) acc1: 100.0000 (73.0740) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.1029 max mem: 3533 +Test: [ 590/2387] eta: 0:06:27 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0954 (0.6248) acc1: 100.0000 (73.5296) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1970 data: 0.1055 max mem: 3533 +Test: [ 600/2387] eta: 0:06:25 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0947 (0.6160) acc1: 100.0000 (73.9700) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1970 data: 0.1043 max mem: 3533 +Test: [ 610/2387] eta: 0:06:21 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0916 (0.6074) acc1: 100.0000 (74.3956) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1877 data: 0.0941 max mem: 3533 +Test: [ 620/2387] eta: 0:06:18 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0886 (0.5990) acc1: 100.0000 (74.8075) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1829 data: 0.0909 max mem: 3533 +Test: [ 630/2387] eta: 0:06:15 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0887 (0.5909) acc1: 100.0000 (75.2064) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1762 data: 0.0845 max mem: 3533 +Test: [ 640/2387] eta: 0:06:12 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0889 (0.5831) acc1: 100.0000 (75.5927) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1762 data: 0.0847 max mem: 3533 +Test: [ 650/2387] eta: 0:06:09 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0888 (0.5755) acc1: 100.0000 (75.9665) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1763 data: 0.0843 max mem: 3533 +Test: [ 660/2387] eta: 0:06:06 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0890 (0.5682) acc1: 100.0000 (76.3293) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1741 data: 0.0802 max mem: 3533 +Test: [ 670/2387] eta: 0:06:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0892 (0.5610) acc1: 100.0000 (76.6809) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1823 data: 0.0889 max mem: 3534 +Test: [ 680/2387] eta: 0:06:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0875 (0.5540) acc1: 100.0000 (77.0233) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1747 data: 0.0838 max mem: 3534 +Test: [ 690/2387] eta: 0:05:56 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0878 (0.5489) acc1: 100.0000 (77.2733) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1651 data: 0.0744 max mem: 3534 +Test: [ 700/2387] eta: 0:05:53 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2419 (0.5446) acc1: 92.7083 (77.4860) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1710 data: 0.0774 max mem: 3534 +Test: [ 710/2387] eta: 0:05:51 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2466 (0.5404) acc1: 92.4479 (77.7020) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1767 data: 0.0824 max mem: 3534 +Test: [ 720/2387] eta: 0:05:48 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2452 (0.5363) acc1: 92.7083 (77.9054) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1785 data: 0.0859 max mem: 3534 +Test: [ 730/2387] eta: 0:05:45 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2452 (0.5323) acc1: 92.1875 (78.0965) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1706 data: 0.0774 max mem: 3534 +Test: [ 740/2387] eta: 0:05:42 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2449 (0.5284) acc1: 91.9271 (78.2874) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1705 data: 0.0777 max mem: 3534 +Test: [ 750/2387] eta: 0:05:39 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2420 (0.5245) acc1: 92.4479 (78.4842) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1781 data: 0.0863 max mem: 3534 +Test: [ 760/2387] eta: 0:05:37 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2342 (0.5204) acc1: 92.7083 (78.6910) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1847 data: 0.0918 max mem: 3534 +Test: [ 770/2387] eta: 0:05:34 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1593 (0.5155) acc1: 96.8750 (78.9458) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1803 data: 0.0876 max mem: 3534 +Test: [ 780/2387] eta: 0:05:31 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1286 (0.5103) acc1: 99.7396 (79.2110) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1739 data: 0.0810 max mem: 3534 +Test: [ 790/2387] eta: 0:05:29 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0952 (0.5057) acc1: 100.0000 (79.4245) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.1837 data: 0.0914 max mem: 3534 +Test: [ 800/2387] eta: 0:05:26 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0941 (0.5006) acc1: 100.0000 (79.6787) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1813 data: 0.0906 max mem: 3534 +Test: [ 810/2387] eta: 0:05:23 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0993 (0.4957) acc1: 100.0000 (79.9254) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1714 data: 0.0797 max mem: 3534 +Test: [ 820/2387] eta: 0:05:21 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0998 (0.4909) acc1: 100.0000 (80.1639) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1741 data: 0.0818 max mem: 3534 +Test: [ 830/2387] eta: 0:05:18 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1035 (0.4865) acc1: 99.7396 (80.3804) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1784 data: 0.0869 max mem: 3534 +Test: [ 840/2387] eta: 0:05:16 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1265 (0.4823) acc1: 98.9583 (80.6022) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1762 data: 0.0856 max mem: 3534 +Test: [ 850/2387] eta: 0:05:13 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1298 (0.4782) acc1: 99.2188 (80.8201) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1779 data: 0.0872 max mem: 3534 +Test: [ 860/2387] eta: 0:05:11 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1156 (0.4739) acc1: 99.7396 (81.0395) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1843 data: 0.0910 max mem: 3534 +Test: [ 870/2387] eta: 0:05:08 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1153 (0.4699) acc1: 99.7396 (81.2506) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1829 data: 0.0897 max mem: 3534 +Test: [ 880/2387] eta: 0:05:06 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1199 (0.4661) acc1: 99.4792 (81.4504) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1780 data: 0.0875 max mem: 3534 +Test: [ 890/2387] eta: 0:05:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1067 (0.4619) acc1: 99.7396 (81.6580) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1753 data: 0.0853 max mem: 3534 +Test: [ 900/2387] eta: 0:05:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1045 (0.4659) acc1: 99.7396 (81.2968) acc5: 100.0000 (100.0000) acc1_neg: 36.458 acc5_neg: 100.000 time: 0.1678 data: 0.0771 max mem: 3535 +Test: [ 910/2387] eta: 0:04:58 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.9897 (0.4717) acc1: 38.0208 (80.8161) acc5: 100.0000 (100.0000) acc1_neg: 41.406 acc5_neg: 100.000 time: 0.1740 data: 0.0829 max mem: 3535 +Test: [ 920/2387] eta: 0:04:56 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.0028 (0.4776) acc1: 38.0208 (80.3370) acc5: 100.0000 (100.0000) acc1_neg: 38.802 acc5_neg: 100.000 time: 0.1791 data: 0.0869 max mem: 3535 +Test: [ 930/2387] eta: 0:04:53 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.0039 (0.4832) acc1: 37.2396 (79.8724) acc5: 100.0000 (100.0000) acc1_neg: 35.677 acc5_neg: 100.000 time: 0.1761 data: 0.0828 max mem: 3535 +Test: [ 940/2387] eta: 0:04:51 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.0117 (0.4887) acc1: 36.4583 (79.4155) acc5: 100.0000 (100.0000) acc1_neg: 37.240 acc5_neg: 100.000 time: 0.1760 data: 0.0842 max mem: 3535 +Test: [ 950/2387] eta: 0:04:48 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.0090 (0.4941) acc1: 36.9792 (78.9687) acc5: 100.0000 (100.0000) acc1_neg: 40.885 acc5_neg: 100.000 time: 0.1699 data: 0.0777 max mem: 3535 +Test: [ 960/2387] eta: 0:04:46 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.9851 (0.4993) acc1: 37.2396 (78.5418) acc5: 100.0000 (100.0000) acc1_neg: 39.583 acc5_neg: 100.000 time: 0.1693 data: 0.0757 max mem: 3535 +Test: [ 970/2387] eta: 0:04:44 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.9955 (0.5046) acc1: 36.9792 (78.1060) acc5: 100.0000 (100.0000) acc1_neg: 37.760 acc5_neg: 100.000 time: 0.1735 data: 0.0820 max mem: 3535 +Test: [ 980/2387] eta: 0:04:41 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.9775 (0.5037) acc1: 37.7604 (78.1046) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1695 data: 0.0779 max mem: 3535 +Test: [ 990/2387] eta: 0:04:39 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1029 (0.4997) acc1: 99.7396 (78.3237) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1675 data: 0.0752 max mem: 3535 +Test: [1000/2387] eta: 0:04:36 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0935 (0.4958) acc1: 100.0000 (78.5256) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1753 data: 0.0833 max mem: 3535 +Test: [1010/2387] eta: 0:04:34 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0967 (0.4920) acc1: 99.7396 (78.7267) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1844 data: 0.0925 max mem: 3535 +Test: [1020/2387] eta: 0:04:32 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1062 (0.4883) acc1: 99.7396 (78.9274) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1905 data: 0.0990 max mem: 3535 +Test: [1030/2387] eta: 0:04:30 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1062 (0.4846) acc1: 100.0000 (79.1295) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1836 data: 0.0904 max mem: 3535 +Test: [1040/2387] eta: 0:04:28 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0962 (0.4809) acc1: 100.0000 (79.3295) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1735 data: 0.0802 max mem: 3535 +Test: [1050/2387] eta: 0:04:25 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0933 (0.4772) acc1: 100.0000 (79.5255) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1788 data: 0.0860 max mem: 3535 +Test: [1060/2387] eta: 0:04:24 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0934 (0.4736) acc1: 100.0000 (79.7184) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2006 data: 0.1063 max mem: 3535 +Test: [1070/2387] eta: 0:04:22 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0962 (0.4702) acc1: 100.0000 (79.9076) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2056 data: 0.1119 max mem: 3535 +Test: [1080/2387] eta: 0:04:19 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1015 (0.4669) acc1: 100.0000 (80.0855) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1899 data: 0.0986 max mem: 3535 +Test: [1090/2387] eta: 0:04:17 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1172 (0.4638) acc1: 99.7396 (80.2654) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1804 data: 0.0894 max mem: 3535 +Test: [1100/2387] eta: 0:04:15 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1158 (0.4606) acc1: 100.0000 (80.4427) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1738 data: 0.0827 max mem: 3535 +Test: [1110/2387] eta: 0:04:13 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1165 (0.4580) acc1: 100.0000 (80.5792) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1887 data: 0.0974 max mem: 3535 +Test: [1120/2387] eta: 0:04:11 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1157 (0.4549) acc1: 99.7396 (80.7496) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2057 data: 0.1128 max mem: 3536 +Test: [1130/2387] eta: 0:04:09 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1115 (0.4519) acc1: 99.7396 (80.9159) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2112 data: 0.1174 max mem: 3536 +Test: [1140/2387] eta: 0:04:07 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1074 (0.4489) acc1: 100.0000 (81.0809) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2147 data: 0.1226 max mem: 3536 +Test: [1150/2387] eta: 0:04:05 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1105 (0.4461) acc1: 100.0000 (81.2410) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2043 data: 0.1132 max mem: 3536 +Test: [1160/2387] eta: 0:04:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1075 (0.4432) acc1: 100.0000 (81.3998) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1968 data: 0.1040 max mem: 3536 +Test: [1170/2387] eta: 0:04:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1063 (0.4403) acc1: 100.0000 (81.5567) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1993 data: 0.1063 max mem: 3536 +Test: [1180/2387] eta: 0:03:59 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1099 (0.4376) acc1: 99.7396 (81.7056) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1944 data: 0.1013 max mem: 3536 +Test: [1190/2387] eta: 0:03:57 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1099 (0.4349) acc1: 99.7396 (81.8576) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1998 data: 0.1068 max mem: 3536 +Test: [1200/2387] eta: 0:03:55 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0921 (0.4320) acc1: 100.0000 (82.0083) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2064 data: 0.1142 max mem: 3536 +Test: [1210/2387] eta: 0:03:54 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0838 (0.4291) acc1: 100.0000 (82.1568) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2012 data: 0.1071 max mem: 3536 +Test: [1220/2387] eta: 0:03:52 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0873 (0.4264) acc1: 100.0000 (82.3023) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2027 data: 0.1077 max mem: 3536 +Test: [1230/2387] eta: 0:03:50 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0877 (0.4236) acc1: 100.0000 (82.4455) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2013 data: 0.1088 max mem: 3536 +Test: [1240/2387] eta: 0:03:48 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0855 (0.4209) acc1: 100.0000 (82.5867) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2092 data: 0.1180 max mem: 3536 +Test: [1250/2387] eta: 0:03:46 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0862 (0.4182) acc1: 100.0000 (82.7255) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1944 data: 0.1027 max mem: 3536 +Test: [1260/2387] eta: 0:03:44 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0848 (0.4156) acc1: 100.0000 (82.8621) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1940 data: 0.0992 max mem: 3536 +Test: [1270/2387] eta: 0:03:42 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0783 (0.4129) acc1: 100.0000 (82.9969) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2055 data: 0.1110 max mem: 3536 +Test: [1280/2387] eta: 0:03:40 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0826 (0.4104) acc1: 100.0000 (83.1290) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2053 data: 0.1141 max mem: 3536 +Test: [1290/2387] eta: 0:03:38 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0879 (0.4081) acc1: 100.0000 (83.2452) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2121 data: 0.1213 max mem: 3536 +Test: [1300/2387] eta: 0:03:36 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0873 (0.4057) acc1: 100.0000 (83.3692) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2064 data: 0.1157 max mem: 3536 +Test: [1310/2387] eta: 0:03:34 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0920 (0.4033) acc1: 100.0000 (83.4901) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2054 data: 0.1139 max mem: 3536 +Test: [1320/2387] eta: 0:03:32 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0966 (0.4011) acc1: 100.0000 (83.6044) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1995 data: 0.1083 max mem: 3536 +Test: [1330/2387] eta: 0:03:30 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0941 (0.3988) acc1: 99.7396 (83.7248) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1856 data: 0.0923 max mem: 3536 +Test: [1340/2387] eta: 0:03:28 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1032 (0.3975) acc1: 99.7396 (83.7975) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1835 data: 0.0879 max mem: 3536 +Test: [1350/2387] eta: 0:03:26 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0659 (0.3950) acc1: 100.0000 (83.9174) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1953 data: 0.1017 max mem: 3537 +Test: [1360/2387] eta: 0:03:24 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0597 (0.3926) acc1: 100.0000 (84.0356) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2077 data: 0.1165 max mem: 3537 +Test: [1370/2387] eta: 0:03:22 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0600 (0.3901) acc1: 100.0000 (84.1520) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1893 data: 0.0984 max mem: 3537 +Test: [1380/2387] eta: 0:03:20 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0594 (0.3877) acc1: 100.0000 (84.2668) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1944 data: 0.1025 max mem: 3537 +Test: [1390/2387] eta: 0:03:18 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0561 (0.3854) acc1: 100.0000 (84.3799) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2444 data: 0.1527 max mem: 3537 +Test: [1400/2387] eta: 0:03:16 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0540 (0.3830) acc1: 100.0000 (84.4914) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2166 data: 0.1237 max mem: 3537 +Test: [1410/2387] eta: 0:03:14 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0552 (0.3809) acc1: 100.0000 (84.5987) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1922 data: 0.0985 max mem: 3537 +Test: [1420/2387] eta: 0:03:12 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1004 (0.3793) acc1: 100.0000 (84.6838) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2069 data: 0.1156 max mem: 3537 +Test: [1430/2387] eta: 0:03:10 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1026 (0.3774) acc1: 100.0000 (84.7885) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2028 data: 0.1128 max mem: 3537 +Test: [1440/2387] eta: 0:03:08 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0999 (0.3755) acc1: 100.0000 (84.8929) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2008 data: 0.1106 max mem: 3537 +Test: [1450/2387] eta: 0:03:06 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0965 (0.3737) acc1: 100.0000 (84.9928) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1892 data: 0.0982 max mem: 3537 +Test: [1460/2387] eta: 0:03:04 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1014 (0.3718) acc1: 99.4792 (85.0917) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1979 data: 0.1057 max mem: 3537 +Test: [1470/2387] eta: 0:03:02 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0896 (0.3699) acc1: 100.0000 (85.1931) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1808 data: 0.0887 max mem: 3537 +Test: [1480/2387] eta: 0:03:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0846 (0.3680) acc1: 100.0000 (85.2925) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1737 data: 0.0825 max mem: 3537 +Test: [1490/2387] eta: 0:02:58 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0907 (0.3661) acc1: 100.0000 (85.3898) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1848 data: 0.0918 max mem: 3537 +Test: [1500/2387] eta: 0:02:56 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0911 (0.3643) acc1: 100.0000 (85.4861) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1718 data: 0.0774 max mem: 3537 +Test: [1510/2387] eta: 0:02:54 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0907 (0.3625) acc1: 100.0000 (85.5818) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1703 data: 0.0779 max mem: 3537 +Test: [1520/2387] eta: 0:02:52 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0951 (0.3609) acc1: 99.7396 (85.6704) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1916 data: 0.0998 max mem: 3537 +Test: [1530/2387] eta: 0:02:50 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1228 (0.3594) acc1: 99.2188 (85.7592) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1948 data: 0.1025 max mem: 3537 +Test: [1540/2387] eta: 0:02:48 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1132 (0.3578) acc1: 99.7396 (85.8500) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1973 data: 0.1058 max mem: 3537 +Test: [1550/2387] eta: 0:02:46 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1118 (0.3562) acc1: 99.7396 (85.9363) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1939 data: 0.1013 max mem: 3537 +Test: [1560/2387] eta: 0:02:44 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1142 (0.3547) acc1: 99.4792 (86.0214) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1918 data: 0.0988 max mem: 3537 +Test: [1570/2387] eta: 0:02:41 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1066 (0.3531) acc1: 100.0000 (86.1101) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1904 data: 0.0988 max mem: 3537 +Test: [1580/2387] eta: 0:02:39 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0909 (0.3515) acc1: 100.0000 (86.1964) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1725 data: 0.0819 max mem: 3538 +Test: [1590/2387] eta: 0:02:37 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0929 (0.3499) acc1: 100.0000 (86.2820) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1812 data: 0.0900 max mem: 3538 +Test: [1600/2387] eta: 0:02:35 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0985 (0.3486) acc1: 100.0000 (86.3414) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1845 data: 0.0921 max mem: 3538 +Test: [1610/2387] eta: 0:02:33 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0970 (0.3470) acc1: 100.0000 (86.4257) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1745 data: 0.0824 max mem: 3538 +Test: [1620/2387] eta: 0:02:31 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0969 (0.3455) acc1: 100.0000 (86.5069) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1766 data: 0.0830 max mem: 3538 +Test: [1630/2387] eta: 0:02:29 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0995 (0.3440) acc1: 99.7396 (86.5840) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1708 data: 0.0777 max mem: 3538 +Test: [1640/2387] eta: 0:02:27 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0995 (0.3426) acc1: 99.7396 (86.6604) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1649 data: 0.0738 max mem: 3538 +Test: [1650/2387] eta: 0:02:25 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0987 (0.3411) acc1: 99.7396 (86.7377) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1777 data: 0.0846 max mem: 3538 +Test: [1660/2387] eta: 0:02:23 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1054 (0.3397) acc1: 99.7396 (86.8152) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1900 data: 0.0949 max mem: 3538 +Test: [1670/2387] eta: 0:02:21 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0985 (0.3383) acc1: 100.0000 (86.8902) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.1016 max mem: 3538 +Test: [1680/2387] eta: 0:02:19 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0985 (0.3370) acc1: 99.4792 (86.9595) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1835 data: 0.0910 max mem: 3538 +Test: [1690/2387] eta: 0:02:17 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1059 (0.3357) acc1: 99.2188 (87.0288) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1819 data: 0.0878 max mem: 3538 +Test: [1700/2387] eta: 0:02:15 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1059 (0.3343) acc1: 99.4792 (87.1018) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1834 data: 0.0903 max mem: 3538 +Test: [1710/2387] eta: 0:02:13 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0969 (0.3329) acc1: 100.0000 (87.1766) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1966 data: 0.1039 max mem: 3538 +Test: [1720/2387] eta: 0:02:11 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0916 (0.3315) acc1: 100.0000 (87.2503) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1932 data: 0.0991 max mem: 3538 +Test: [1730/2387] eta: 0:02:09 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0875 (0.3301) acc1: 100.0000 (87.3226) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1877 data: 0.0930 max mem: 3538 +Test: [1740/2387] eta: 0:02:07 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0871 (0.3287) acc1: 100.0000 (87.3944) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2006 data: 0.1075 max mem: 3538 +Test: [1750/2387] eta: 0:02:05 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0845 (0.3273) acc1: 100.0000 (87.4661) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2033 data: 0.1121 max mem: 3538 +Test: [1760/2387] eta: 0:02:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0773 (0.3259) acc1: 100.0000 (87.5373) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2005 data: 0.1082 max mem: 3538 +Test: [1770/2387] eta: 0:02:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0806 (0.3246) acc1: 100.0000 (87.6062) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1934 data: 0.1014 max mem: 3538 +Test: [1780/2387] eta: 0:01:59 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0877 (0.3232) acc1: 100.0000 (87.6758) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1869 data: 0.0952 max mem: 3538 +Test: [1790/2387] eta: 0:01:57 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0812 (0.3219) acc1: 100.0000 (87.7431) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1964 data: 0.1029 max mem: 3538 +Test: [1800/2387] eta: 0:01:55 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0826 (0.3206) acc1: 100.0000 (87.8106) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2005 data: 0.1077 max mem: 3538 +Test: [1810/2387] eta: 0:01:53 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0815 (0.3193) acc1: 100.0000 (87.8775) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2049 data: 0.1108 max mem: 3539 +Test: [1820/2387] eta: 0:01:51 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0753 (0.3179) acc1: 100.0000 (87.9439) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2100 data: 0.1158 max mem: 3539 +Test: [1830/2387] eta: 0:01:49 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0737 (0.3166) acc1: 100.0000 (88.0097) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2049 data: 0.1123 max mem: 3539 +Test: [1840/2387] eta: 0:01:47 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1049 (0.3155) acc1: 100.0000 (88.0749) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2063 data: 0.1143 max mem: 3539 +Test: [1850/2387] eta: 0:01:45 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1057 (0.3144) acc1: 100.0000 (88.1393) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2063 data: 0.1150 max mem: 3539 +Test: [1860/2387] eta: 0:01:43 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1057 (0.3132) acc1: 100.0000 (88.2030) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2097 data: 0.1181 max mem: 3539 +Test: [1870/2387] eta: 0:01:42 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1056 (0.3121) acc1: 100.0000 (88.2658) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2114 data: 0.1192 max mem: 3539 +Test: [1880/2387] eta: 0:01:40 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1053 (0.3110) acc1: 100.0000 (88.3282) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2102 data: 0.1175 max mem: 3539 +Test: [1890/2387] eta: 0:01:38 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1054 (0.3099) acc1: 100.0000 (88.3899) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2135 data: 0.1215 max mem: 3539 +Test: [1900/2387] eta: 0:01:36 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1058 (0.3089) acc1: 100.0000 (88.4510) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2046 data: 0.1119 max mem: 3539 +Test: [1910/2387] eta: 0:01:34 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1076 (0.3078) acc1: 100.0000 (88.5114) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2037 data: 0.1118 max mem: 3539 +Test: [1920/2387] eta: 0:01:32 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1087 (0.3068) acc1: 100.0000 (88.5711) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2083 data: 0.1178 max mem: 3539 +Test: [1930/2387] eta: 0:01:30 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0934 (0.3057) acc1: 100.0000 (88.6303) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1945 data: 0.1019 max mem: 3539 +Test: [1940/2387] eta: 0:01:28 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0912 (0.3049) acc1: 100.0000 (88.6545) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1858 data: 0.0931 max mem: 3539 +Test: [1950/2387] eta: 0:01:26 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0944 (0.3038) acc1: 100.0000 (88.7127) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0983 max mem: 3539 +Test: [1960/2387] eta: 0:01:24 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0944 (0.3028) acc1: 100.0000 (88.7660) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1795 data: 0.0879 max mem: 3539 +Test: [1970/2387] eta: 0:01:22 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1177 (0.3020) acc1: 100.0000 (88.8185) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1708 data: 0.0801 max mem: 3539 +Test: [1980/2387] eta: 0:01:20 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1289 (0.3011) acc1: 99.7396 (88.8678) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1784 data: 0.0876 max mem: 3539 +Test: [1990/2387] eta: 0:01:18 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1333 (0.3003) acc1: 99.2188 (88.9189) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1867 data: 0.0948 max mem: 3539 +Test: [2000/2387] eta: 0:01:16 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1351 (0.2995) acc1: 99.2188 (88.9696) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1976 data: 0.1058 max mem: 3539 +Test: [2010/2387] eta: 0:01:14 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1381 (0.2987) acc1: 99.2188 (89.0190) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2009 data: 0.1094 max mem: 3539 +Test: [2020/2387] eta: 0:01:12 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1381 (0.2979) acc1: 98.9583 (89.0686) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1983 data: 0.1066 max mem: 3539 +Test: [2030/2387] eta: 0:01:10 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1353 (0.2971) acc1: 98.9583 (89.1164) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1875 data: 0.0958 max mem: 3540 +Test: [2040/2387] eta: 0:01:08 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1346 (0.2963) acc1: 98.9583 (89.1657) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1761 data: 0.0836 max mem: 3540 +Test: [2050/2387] eta: 0:01:06 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1333 (0.2955) acc1: 99.2188 (89.2154) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1799 data: 0.0865 max mem: 3540 +Test: [2060/2387] eta: 0:01:04 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1254 (0.2946) acc1: 99.2188 (89.2640) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1833 data: 0.0895 max mem: 3540 +Test: [2070/2387] eta: 0:01:02 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1192 (0.2938) acc1: 99.4792 (89.3139) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1686 data: 0.0751 max mem: 3540 +Test: [2080/2387] eta: 0:01:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1192 (0.2930) acc1: 99.4792 (89.3621) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1666 data: 0.0717 max mem: 3540 +Test: [2090/2387] eta: 0:00:58 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1182 (0.2921) acc1: 99.4792 (89.4103) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1713 data: 0.0764 max mem: 3540 +Test: [2100/2387] eta: 0:00:56 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1189 (0.2913) acc1: 99.4792 (89.4585) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1651 data: 0.0726 max mem: 3540 +Test: [2110/2387] eta: 0:00:54 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1204 (0.2905) acc1: 99.4792 (89.5054) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1745 data: 0.0814 max mem: 3540 +Test: [2120/2387] eta: 0:00:52 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1211 (0.2897) acc1: 99.4792 (89.5529) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1786 data: 0.0859 max mem: 3540 +Test: [2130/2387] eta: 0:00:50 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1090 (0.2888) acc1: 100.0000 (89.6017) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1667 data: 0.0754 max mem: 3540 +Test: [2140/2387] eta: 0:00:48 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1077 (0.2883) acc1: 100.0000 (89.6274) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1708 data: 0.0780 max mem: 3540 +Test: [2150/2387] eta: 0:00:46 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1108 (0.2875) acc1: 100.0000 (89.6753) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1831 data: 0.0897 max mem: 3540 +Test: [2160/2387] eta: 0:00:44 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1108 (0.2866) acc1: 100.0000 (89.7201) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1854 data: 0.0912 max mem: 3540 +Test: [2170/2387] eta: 0:00:42 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1122 (0.2859) acc1: 99.7396 (89.7634) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1841 data: 0.0890 max mem: 3540 +Test: [2180/2387] eta: 0:00:40 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1328 (0.2869) acc1: 98.4375 (89.6945) acc5: 100.0000 (100.0000) acc1_neg: 41.146 acc5_neg: 100.000 time: 0.1791 data: 0.0853 max mem: 3540 +Test: [2190/2387] eta: 0:00:38 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.3210 (0.2875) acc1: 86.7188 (89.6567) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1900 data: 0.0976 max mem: 3540 +Test: [2200/2387] eta: 0:00:36 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1138 (0.2866) acc1: 100.0000 (89.7037) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2009 data: 0.1077 max mem: 3540 +Test: [2210/2387] eta: 0:00:34 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0920 (0.2857) acc1: 100.0000 (89.7502) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2005 data: 0.1078 max mem: 3540 +Test: [2220/2387] eta: 0:00:32 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0883 (0.2849) acc1: 100.0000 (89.7944) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1921 data: 0.1014 max mem: 3540 +Test: [2230/2387] eta: 0:00:30 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0888 (0.2840) acc1: 100.0000 (89.8392) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1701 data: 0.0796 max mem: 3540 +Test: [2240/2387] eta: 0:00:28 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0902 (0.2832) acc1: 100.0000 (89.8843) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1630 data: 0.0723 max mem: 3540 +Test: [2250/2387] eta: 0:00:26 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0902 (0.2823) acc1: 100.0000 (89.9286) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1541 data: 0.0636 max mem: 3540 +Test: [2260/2387] eta: 0:00:24 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0932 (0.2815) acc1: 100.0000 (89.9731) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1448 data: 0.0546 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0947 (0.2807) acc1: 100.0000 (90.0172) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1472 data: 0.0560 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0966 (0.2802) acc1: 100.0000 (90.0410) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1524 data: 0.0613 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0971 (0.2793) acc1: 100.0000 (90.0845) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1530 data: 0.0629 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0980 (0.2786) acc1: 100.0000 (90.1247) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1495 data: 0.0597 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0996 (0.2778) acc1: 100.0000 (90.1649) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1418 data: 0.0514 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1068 (0.2772) acc1: 99.7396 (90.2039) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1319 data: 0.0409 max mem: 3541 +Test: [2330/2387] eta: 0:00:11 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1129 (0.2765) acc1: 100.0000 (90.2453) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1286 data: 0.0383 max mem: 3541 +Test: [2340/2387] eta: 0:00:09 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1027 (0.2757) acc1: 100.0000 (90.2866) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1254 data: 0.0355 max mem: 3541 +Test: [2350/2387] eta: 0:00:07 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1027 (0.2754) acc1: 100.0000 (90.3036) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1219 data: 0.0319 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1055 (0.2746) acc1: 100.0000 (90.3446) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1210 data: 0.0314 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1086 (0.2740) acc1: 100.0000 (90.3825) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1200 data: 0.0306 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1125 (0.2733) acc1: 100.0000 (90.4208) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1187 data: 0.0293 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1289 (0.2733) acc1: 99.2188 (90.4258) acc5: 100.0000 (100.0000) acc1_neg: 90.698 acc5_neg: 100.000 time: 0.1129 data: 0.0258 max mem: 3541 +Test: Total time: 0:07:37 (0.1916 s / it) +* Overall Acc@1 90.426 Acc@5 100.000 + target == real Acc@1 1.888 Acc@5 100.000 + target == fake Acc@1 96.643 Acc@5 100.000 + loss 0.273 + PR-AUC 0.9707 ROC-AUC 0.7088 EER 33.31% (thr=0.8994) +Accuracy of the network on the 916267 test images: 90.4% +[rank0]:[W726 05:18:34.930215941 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:18:54.038000 3163890 torch/distributed/run.py:766] +W0726 05:18:54.038000 3163890 torch/distributed/run.py:766] ***************************************** +W0726 05:18:54.038000 3163890 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:18:54.038000 3163890 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:19:21.365470205 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:19:22.770246639 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:19:22.220946878 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:19:22.342099152 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:19:22.545711365 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:19:23.576699481 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank6]:[W726 05:19:23.576810024 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:19:23.599656131 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n16_seed8_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed8_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 0:59:40 acc1_pos: 2.604 acc5_pos: 100.000 loss: 1.7590 (1.7590) acc1: 2.6042 (2.6042) acc5: 100.0000 (100.0000) time: 6.9251 data: 5.2541 max mem: 3531 +Test: [ 10/517] eta: 0:06:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.9687 (1.9141) acc1: 2.6042 (2.7699) acc5: 100.0000 (100.0000) time: 0.7113 data: 0.4780 max mem: 3531 +Test: [ 20/517] eta: 0:03:27 acc1_pos: 1.042 acc5_pos: 100.000 loss: 1.9745 (1.9465) acc1: 2.6042 (2.7902) acc5: 100.0000 (100.0000) time: 0.0927 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:02:57 acc1_pos: 5.469 acc5_pos: 100.000 loss: 1.9904 (1.9655) acc1: 1.0417 (2.2177) acc5: 100.0000 (100.0000) time: 0.1727 data: 0.0794 max mem: 3531 +Test: [ 40/517] eta: 0:02:37 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.0138 (1.9825) acc1: 0.0000 (1.7721) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1487 max mem: 3531 +Test: [ 50/517] eta: 0:02:25 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.8953 (1.8620) acc1: 0.7812 (6.8525) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2310 data: 0.1400 max mem: 3531 +Test: [ 60/517] eta: 0:02:16 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1707 (1.5830) acc1: 99.4792 (22.0799) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2349 data: 0.1405 max mem: 3531 +Test: [ 70/517] eta: 0:02:10 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1470 (1.3795) acc1: 100.0000 (33.0509) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2390 data: 0.1452 max mem: 3531 +Test: [ 80/517] eta: 0:02:05 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1359 (1.2266) acc1: 100.0000 (41.3098) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2499 data: 0.1589 max mem: 3531 +Test: [ 90/517] eta: 0:02:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1209 (1.1047) acc1: 100.0000 (47.7593) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2498 data: 0.1555 max mem: 3531 +Test: [100/517] eta: 0:01:55 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1194 (1.0074) acc1: 100.0000 (52.9316) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2399 data: 0.1422 max mem: 3531 +Test: [110/517] eta: 0:01:52 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1228 (0.9289) acc1: 100.0000 (57.1720) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2499 data: 0.1504 max mem: 3531 +Test: [120/517] eta: 0:01:47 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1248 (0.8626) acc1: 100.0000 (60.7072) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2349 data: 0.1374 max mem: 3531 +Test: [130/517] eta: 0:01:43 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1199 (0.8057) acc1: 100.0000 (63.7047) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2249 data: 0.1307 max mem: 3531 +Test: [140/517] eta: 0:01:40 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1185 (0.7572) acc1: 100.0000 (66.2770) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2449 data: 0.1504 max mem: 3531 +Test: [150/517] eta: 0:01:37 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1216 (0.7152) acc1: 100.0000 (68.5034) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2399 data: 0.1420 max mem: 3531 +Test: [160/517] eta: 0:01:33 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1219 (0.6784) acc1: 100.0000 (70.4597) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2299 data: 0.1340 max mem: 3531 +Test: [170/517] eta: 0:01:31 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1206 (0.6457) acc1: 100.0000 (72.1857) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2499 data: 0.1586 max mem: 3531 +Test: [180/517] eta: 0:01:28 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1187 (0.6164) acc1: 100.0000 (73.7137) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1624 max mem: 3531 +Test: [190/517] eta: 0:01:24 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1113 (0.5898) acc1: 100.0000 (75.0900) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2301 data: 0.1371 max mem: 3531 +Test: [200/517] eta: 0:01:21 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1042 (0.5657) acc1: 100.0000 (76.3293) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2299 data: 0.1345 max mem: 3531 +Test: [210/517] eta: 0:01:18 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1181 (0.5462) acc1: 100.0000 (77.4450) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2349 data: 0.1390 max mem: 3532 +Test: [220/517] eta: 0:01:16 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1529 (0.5284) acc1: 100.0000 (78.4644) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2356 data: 0.1421 max mem: 3532 +Test: [230/517] eta: 0:01:13 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1430 (0.5113) acc1: 100.0000 (79.3966) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2347 data: 0.1376 max mem: 3532 +Test: [240/517] eta: 0:01:10 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1342 (0.4961) acc1: 100.0000 (80.2505) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2242 data: 0.1288 max mem: 3532 +Test: [250/517] eta: 0:01:07 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1342 (0.4812) acc1: 100.0000 (81.0373) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2249 data: 0.1325 max mem: 3532 +Test: [260/517] eta: 0:01:05 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1155 (0.4672) acc1: 100.0000 (81.7638) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2450 data: 0.1520 max mem: 3532 +Test: [270/517] eta: 0:01:02 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1195 (0.4545) acc1: 100.0000 (82.4368) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2509 data: 0.1554 max mem: 3532 +Test: [280/517] eta: 0:00:59 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1206 (0.4426) acc1: 100.0000 (83.0609) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2350 data: 0.1401 max mem: 3532 +Test: [290/517] eta: 0:00:56 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1234 (0.4318) acc1: 100.0000 (83.6430) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2243 data: 0.1325 max mem: 3532 +Test: [300/517] eta: 0:00:54 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1222 (0.4213) acc1: 100.0000 (84.1864) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2407 data: 0.1486 max mem: 3532 +Test: [310/517] eta: 0:00:51 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1128 (0.4114) acc1: 100.0000 (84.6949) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2395 data: 0.1474 max mem: 3532 +Test: [320/517] eta: 0:00:49 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1119 (0.4021) acc1: 100.0000 (85.1717) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2389 data: 0.1449 max mem: 3532 +Test: [330/517] eta: 0:00:46 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1128 (0.3933) acc1: 100.0000 (85.6197) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2449 data: 0.1508 max mem: 3532 +Test: [340/517] eta: 0:00:44 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1114 (0.3851) acc1: 100.0000 (86.0414) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2363 data: 0.1418 max mem: 3532 +Test: [350/517] eta: 0:00:41 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1129 (0.3775) acc1: 100.0000 (86.4390) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2349 data: 0.1403 max mem: 3532 +Test: [360/517] eta: 0:00:39 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1185 (0.3704) acc1: 100.0000 (86.8147) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2441 data: 0.1490 max mem: 3532 +Test: [370/517] eta: 0:00:36 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1198 (0.3637) acc1: 100.0000 (87.1701) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2449 data: 0.1492 max mem: 3532 +Test: [380/517] eta: 0:00:34 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1190 (0.3572) acc1: 100.0000 (87.5068) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2300 data: 0.1359 max mem: 3532 +Test: [390/517] eta: 0:00:31 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1201 (0.3519) acc1: 100.0000 (87.8064) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2449 data: 0.1506 max mem: 3532 +Test: [400/517] eta: 0:00:29 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1285 (0.3464) acc1: 100.0000 (88.1105) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2441 data: 0.1491 max mem: 3532 +Test: [410/517] eta: 0:00:26 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1268 (0.3411) acc1: 100.0000 (88.3997) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2299 data: 0.1369 max mem: 3532 +Test: [420/517] eta: 0:00:24 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1259 (0.3359) acc1: 100.0000 (88.6753) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2399 data: 0.1487 max mem: 3532 +Test: [430/517] eta: 0:00:21 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1314 (0.3313) acc1: 100.0000 (88.9380) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2449 data: 0.1518 max mem: 3532 +Test: [440/517] eta: 0:00:19 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1293 (0.3265) acc1: 100.0000 (89.1889) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2449 data: 0.1501 max mem: 3533 +Test: [450/517] eta: 0:00:16 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1171 (0.3222) acc1: 100.0000 (89.4286) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2299 data: 0.1329 max mem: 3533 +Test: [460/517] eta: 0:00:14 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1289 (0.3182) acc1: 100.0000 (89.6579) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2273 data: 0.1291 max mem: 3533 +Test: [470/517] eta: 0:00:11 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1126 (0.3138) acc1: 100.0000 (89.8775) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2449 data: 0.1475 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1078 (0.3098) acc1: 100.0000 (90.0879) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2358 data: 0.1408 max mem: 3533 +Test: [490/517] eta: 0:00:06 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1376 (0.3073) acc1: 100.0000 (90.2760) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2474 data: 0.1549 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1108 (0.3033) acc1: 100.0000 (90.4701) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2504 data: 0.1550 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1142 (0.3002) acc1: 100.0000 (90.6556) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2103 data: 0.1164 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1203 (0.2981) acc1: 100.0000 (90.7593) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1935 data: 0.1000 max mem: 3533 +Test: Total time: 0:02:06 (0.2455 s / it) +* Overall Acc@1 90.759 Acc@5 100.000 + target == real Acc@1 2.188 Acc@5 100.000 + target == fake Acc@1 99.948 Acc@5 100.000 + loss 0.298 + PR-AUC 0.9341 ROC-AUC 0.6510 EER 39.33% (thr=0.8838) +Accuracy of the network on the 198427 test images: 90.8% +[rank0]:[W726 05:21:43.968016546 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 05:22:02.535000 3164463 torch/distributed/run.py:766] +W0726 05:22:02.535000 3164463 torch/distributed/run.py:766] ***************************************** +W0726 05:22:02.535000 3164463 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:22:02.535000 3164463 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:22:30.034046243 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:22:30.132330067 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:22:30.355497197 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:22:30.485472790 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:22:30.535063753 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:22:31.591177729 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:22:31.615380619 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W726 05:22:31.620252783 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n16_seed8_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed8_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:42:09 acc1_pos: 24.219 acc5_pos: 100.000 loss: 1.1712 (1.1712) acc1: 24.2188 (24.2188) acc5: 100.0000 (100.0000) time: 6.5030 data: 4.9158 max mem: 3531 +Test: [ 10/389] eta: 0:04:33 acc1_pos: 48.958 acc5_pos: 100.000 loss: 1.0247 (1.0266) acc1: 35.9375 (32.6231) acc5: 100.0000 (100.0000) time: 0.7207 data: 0.4908 max mem: 3531 +Test: [ 20/389] eta: 0:02:36 acc1_pos: 4.948 acc5_pos: 100.000 loss: 1.0819 (1.2648) acc1: 16.6667 (23.2267) acc5: 100.0000 (100.0000) time: 0.1212 data: 0.0244 max mem: 3531 +Test: [ 30/389] eta: 0:02:14 acc1_pos: 3.646 acc5_pos: 100.000 loss: 1.2093 (1.2259) acc1: 10.9375 (18.9600) acc5: 100.0000 (100.0000) time: 0.1831 data: 0.0866 max mem: 3531 +Test: [ 40/389] eta: 0:01:58 acc1_pos: 16.406 acc5_pos: 100.000 loss: 1.0755 (1.2043) acc1: 10.9375 (18.4007) acc5: 100.0000 (100.0000) time: 0.2473 data: 0.1554 max mem: 3531 +Test: [ 50/389] eta: 0:01:46 acc1_pos: 16.667 acc5_pos: 100.000 loss: 1.2572 (1.2165) acc1: 10.6771 (17.0088) acc5: 100.0000 (100.0000) time: 0.2217 data: 0.1285 max mem: 3531 +Test: [ 60/389] eta: 0:01:35 acc1_pos: 21.615 acc5_pos: 100.000 loss: 1.3280 (1.2431) acc1: 11.1979 (15.8214) acc5: 100.0000 (100.0000) time: 0.1875 data: 0.0916 max mem: 3531 +Test: [ 70/389] eta: 0:01:27 acc1_pos: 38.021 acc5_pos: 100.000 loss: 1.0436 (1.1814) acc1: 19.0104 (19.6266) acc5: 100.0000 (100.0000) time: 0.1699 data: 0.0735 max mem: 3531 +Test: [ 80/389] eta: 0:01:21 acc1_pos: 48.958 acc5_pos: 100.000 loss: 0.9211 (1.1822) acc1: 28.3854 (20.4347) acc5: 100.0000 (100.0000) time: 0.1899 data: 0.0937 max mem: 3531 +Test: [ 90/389] eta: 0:01:16 acc1_pos: 34.115 acc5_pos: 100.000 loss: 0.8216 (1.1345) acc1: 37.7604 (23.6207) acc5: 100.0000 (100.0000) time: 0.1899 data: 0.0944 max mem: 3531 +Test: [100/389] eta: 0:01:11 acc1_pos: 56.250 acc5_pos: 100.000 loss: 0.7574 (1.1242) acc1: 43.4896 (24.6416) acc5: 100.0000 (100.0000) time: 0.1805 data: 0.0862 max mem: 3531 +Test: [110/389] eta: 0:01:06 acc1_pos: 21.615 acc5_pos: 100.000 loss: 0.9377 (1.1382) acc1: 32.2917 (24.8240) acc5: 100.0000 (100.0000) time: 0.1680 data: 0.0767 max mem: 3531 +Test: [120/389] eta: 0:01:03 acc1_pos: 7.292 acc5_pos: 100.000 loss: 1.0679 (1.1314) acc1: 27.8646 (25.3637) acc5: 100.0000 (100.0000) time: 0.1700 data: 0.0796 max mem: 3531 +Test: [130/389] eta: 0:00:59 acc1_pos: 12.883 acc5_pos: 100.000 loss: 0.7475 (1.0884) acc1: 34.3750 (28.9063) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1851 data: 0.0943 max mem: 3531 +Test: [140/389] eta: 0:00:56 acc1_pos: 12.883 acc5_pos: 100.000 loss: 0.2910 (1.0312) acc1: 95.8333 (33.4996) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1747 data: 0.0839 max mem: 3531 +Test: [150/389] eta: 0:00:53 acc1_pos: 12.883 acc5_pos: 100.000 loss: 0.2910 (0.9897) acc1: 96.6146 (37.0361) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1665 data: 0.0733 max mem: 3531 +Test: [160/389] eta: 0:00:50 acc1_pos: 12.883 acc5_pos: 100.000 loss: 0.3402 (0.9479) acc1: 90.1042 (40.1640) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1681 data: 0.0742 max mem: 3531 +Test: [170/389] eta: 0:00:47 acc1_pos: 12.883 acc5_pos: 100.000 loss: 0.1492 (0.9059) acc1: 100.0000 (43.4606) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1614 data: 0.0683 max mem: 3531 +Test: [180/389] eta: 0:00:44 acc1_pos: 12.883 acc5_pos: 100.000 loss: 0.2096 (0.8752) acc1: 98.6979 (46.0276) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.1660 data: 0.0738 max mem: 3531 +Test: [190/389] eta: 0:00:42 acc1_pos: 12.883 acc5_pos: 100.000 loss: 0.3472 (0.8496) acc1: 92.1875 (48.2057) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.1737 data: 0.0832 max mem: 3531 +Test: [200/389] eta: 0:00:39 acc1_pos: 12.883 acc5_pos: 100.000 loss: 0.2519 (0.8205) acc1: 97.9167 (50.5195) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1752 data: 0.0838 max mem: 3531 +Test: [210/389] eta: 0:00:36 acc1_pos: 12.883 acc5_pos: 100.000 loss: 0.1360 (0.7880) acc1: 100.0000 (52.8584) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1586 data: 0.0670 max mem: 3532 +Test: [220/389] eta: 0:00:34 acc1_pos: 12.883 acc5_pos: 100.000 loss: 0.1265 (0.7592) acc1: 100.0000 (54.9774) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1559 data: 0.0650 max mem: 3532 +Test: [230/389] eta: 0:00:32 acc1_pos: 12.883 acc5_pos: 100.000 loss: 0.2294 (0.7439) acc1: 98.6979 (56.2444) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.1719 data: 0.0816 max mem: 3532 +Test: [240/389] eta: 0:00:30 acc1_pos: 12.883 acc5_pos: 100.000 loss: 0.3136 (0.7281) acc1: 96.6146 (57.5726) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1862 data: 0.0955 max mem: 3532 +Test: [250/389] eta: 0:00:28 acc1_pos: 12.883 acc5_pos: 100.000 loss: 0.2347 (0.7081) acc1: 98.4375 (59.2380) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1866 data: 0.0955 max mem: 3532 +Test: [260/389] eta: 0:00:25 acc1_pos: 12.883 acc5_pos: 100.000 loss: 0.1711 (0.6922) acc1: 100.0000 (60.4386) acc5: 100.0000 (100.0000) acc1_neg: 23.958 acc5_neg: 100.000 time: 0.1768 data: 0.0855 max mem: 3532 +Test: [270/389] eta: 0:00:23 acc1_pos: 12.883 acc5_pos: 100.000 loss: 0.3533 (0.6935) acc1: 95.5729 (60.3974) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1873 data: 0.0955 max mem: 3532 +Test: [280/389] eta: 0:00:21 acc1_pos: 12.883 acc5_pos: 100.000 loss: 0.3969 (0.6865) acc1: 83.3333 (61.1052) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1954 data: 0.1014 max mem: 3532 +Test: [290/389] eta: 0:00:19 acc1_pos: 12.883 acc5_pos: 100.000 loss: 0.3515 (0.6729) acc1: 94.0104 (62.2566) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1957 data: 0.1010 max mem: 3532 +Test: [300/389] eta: 0:00:17 acc1_pos: 12.883 acc5_pos: 100.000 loss: 0.2443 (0.6587) acc1: 99.2188 (63.4664) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1883 data: 0.0959 max mem: 3532 +Test: [310/389] eta: 0:00:15 acc1_pos: 12.883 acc5_pos: 100.000 loss: 0.2938 (0.6526) acc1: 95.3125 (63.9863) acc5: 100.0000 (100.0000) acc1_neg: 38.021 acc5_neg: 100.000 time: 0.1723 data: 0.0810 max mem: 3532 +Test: [320/389] eta: 0:00:13 acc1_pos: 12.883 acc5_pos: 100.000 loss: 0.3894 (0.6434) acc1: 85.1562 (64.7334) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1656 data: 0.0752 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 12.883 acc5_pos: 100.000 loss: 0.3863 (0.6362) acc1: 92.9688 (65.5416) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.1646 data: 0.0746 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 12.883 acc5_pos: 100.000 loss: 0.2111 (0.6234) acc1: 94.7917 (66.4689) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1503 data: 0.0604 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 12.883 acc5_pos: 100.000 loss: 0.1782 (0.6211) acc1: 99.7396 (66.6362) acc5: 100.0000 (100.0000) acc1_neg: 77.344 acc5_neg: 100.000 time: 0.1434 data: 0.0528 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 12.883 acc5_pos: 100.000 loss: 0.2924 (0.6134) acc1: 89.0625 (67.3678) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1385 data: 0.0479 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 12.883 acc5_pos: 100.000 loss: 0.2901 (0.6072) acc1: 97.3958 (67.7961) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1324 data: 0.0424 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 12.883 acc5_pos: 100.000 loss: 0.3026 (0.5987) acc1: 96.8750 (68.4711) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.1282 data: 0.0383 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 12.883 acc5_pos: 100.000 loss: 0.2301 (0.5918) acc1: 98.1771 (69.0268) acc5: 100.0000 (100.0000) acc1_neg: 97.814 acc5_neg: 100.000 time: 0.1285 data: 0.0353 max mem: 3532 +Test: Total time: 0:01:13 (0.1879 s / it) +* Overall Acc@1 69.027 Acc@5 100.000 + target == real Acc@1 25.024 Acc@5 100.000 + target == fake Acc@1 89.480 Acc@5 100.000 + loss 0.592 + PR-AUC 0.8229 ROC-AUC 0.6921 EER 35.66% (thr=0.7427) +Accuracy of the network on the 149358 test images: 69.0% +[rank0]:[W726 05:23:52.684371430 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:24:10.084000 3165021 torch/distributed/run.py:766] +W0726 05:24:10.084000 3165021 torch/distributed/run.py:766] ***************************************** +W0726 05:24:10.084000 3165021 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:24:10.084000 3165021 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:24:37.970975997 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:24:37.473605756 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:24:38.826554074 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W726 05:24:38.829299828 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:24:38.996951465 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:24:38.145717823 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:24:38.162162900 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W726 05:24:38.162318308 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n16_seed8_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed8_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:36:05 acc1_pos: 86.458 acc5_pos: 100.000 loss: 0.3011 (0.3011) acc1: 86.4583 (86.4583) acc5: 100.0000 (100.0000) time: 6.5027 data: 4.9573 max mem: 3531 +Test: [ 10/333] eta: 0:03:38 acc1_pos: 83.854 acc5_pos: 100.000 loss: 0.2487 (0.2651) acc1: 89.5833 (89.9621) acc5: 100.0000 (100.0000) time: 0.6753 data: 0.4512 max mem: 3531 +Test: [ 20/333] eta: 0:02:05 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.2740 (0.2852) acc1: 89.5833 (89.7321) acc5: 100.0000 (100.0000) time: 0.0952 data: 0.0005 max mem: 3531 +Test: [ 30/333] eta: 0:01:39 acc1_pos: 21.094 acc5_pos: 100.000 loss: 0.4271 (0.6404) acc1: 64.8438 (67.9435) acc5: 100.0000 (100.0000) time: 0.1395 data: 0.0439 max mem: 3531 +Test: [ 40/333] eta: 0:01:28 acc1_pos: 10.938 acc5_pos: 100.000 loss: 1.4641 (0.8428) acc1: 17.4479 (55.3354) acc5: 100.0000 (100.0000) time: 0.1999 data: 0.1057 max mem: 3531 +Test: [ 50/333] eta: 0:01:21 acc1_pos: 20.573 acc5_pos: 100.000 loss: 1.5012 (0.9682) acc1: 14.8438 (47.7788) acc5: 100.0000 (100.0000) time: 0.2244 data: 0.1318 max mem: 3531 +Test: [ 60/333] eta: 0:01:14 acc1_pos: 9.896 acc5_pos: 100.000 loss: 1.5026 (1.0608) acc1: 16.9271 (42.7468) acc5: 100.0000 (100.0000) time: 0.2159 data: 0.1244 max mem: 3531 +Test: [ 70/333] eta: 0:01:08 acc1_pos: 11.979 acc5_pos: 100.000 loss: 1.4494 (1.1154) acc1: 18.7500 (39.3046) acc5: 100.0000 (100.0000) time: 0.1910 data: 0.0998 max mem: 3531 +Test: [ 80/333] eta: 0:01:03 acc1_pos: 17.188 acc5_pos: 100.000 loss: 1.4437 (1.1623) acc1: 17.1875 (36.6159) acc5: 100.0000 (100.0000) time: 0.1793 data: 0.0886 max mem: 3531 +Test: [ 90/333] eta: 0:00:59 acc1_pos: 12.760 acc5_pos: 100.000 loss: 1.4325 (1.1864) acc1: 17.1875 (34.4093) acc5: 100.0000 (100.0000) time: 0.1837 data: 0.0873 max mem: 3531 +Test: [100/333] eta: 0:00:55 acc1_pos: 18.750 acc5_pos: 100.000 loss: 1.4022 (1.2063) acc1: 17.9688 (32.8409) acc5: 100.0000 (100.0000) time: 0.1967 data: 0.0985 max mem: 3531 +Test: [110/333] eta: 0:00:52 acc1_pos: 22.135 acc5_pos: 100.000 loss: 1.4044 (1.2255) acc1: 16.6667 (31.4119) acc5: 100.0000 (100.0000) time: 0.1958 data: 0.0992 max mem: 3531 +Test: [120/333] eta: 0:00:49 acc1_pos: 19.531 acc5_pos: 100.000 loss: 1.3915 (1.2364) acc1: 19.0104 (30.5053) acc5: 100.0000 (100.0000) time: 0.1903 data: 0.0926 max mem: 3531 +Test: [130/333] eta: 0:00:46 acc1_pos: 23.958 acc5_pos: 100.000 loss: 1.2870 (1.2426) acc1: 19.5312 (29.6974) acc5: 100.0000 (100.0000) time: 0.1940 data: 0.0983 max mem: 3531 +Test: [140/333] eta: 0:00:43 acc1_pos: 46.875 acc5_pos: 100.000 loss: 1.3232 (1.2490) acc1: 22.1354 (29.1500) acc5: 100.0000 (100.0000) time: 0.1790 data: 0.0867 max mem: 3531 +Test: [150/333] eta: 0:00:40 acc1_pos: 21.094 acc5_pos: 100.000 loss: 1.3548 (1.2581) acc1: 17.4479 (28.4044) acc5: 100.0000 (100.0000) time: 0.1755 data: 0.0848 max mem: 3531 +Test: [160/333] eta: 0:00:38 acc1_pos: 14.145 acc5_pos: 100.000 loss: 1.3272 (1.2450) acc1: 21.6146 (29.2508) acc5: 100.0000 (100.0000) acc1_neg: 72.917 acc5_neg: 100.000 time: 0.1933 data: 0.1014 max mem: 3531 +Test: [170/333] eta: 0:00:35 acc1_pos: 14.145 acc5_pos: 100.000 loss: 0.5809 (1.2018) acc1: 72.9167 (32.1531) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.1919 data: 0.0997 max mem: 3531 +Test: [180/333] eta: 0:00:32 acc1_pos: 14.145 acc5_pos: 100.000 loss: 0.4570 (1.1590) acc1: 80.4688 (34.8930) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.1749 data: 0.0807 max mem: 3531 +Test: [190/333] eta: 0:00:30 acc1_pos: 14.145 acc5_pos: 100.000 loss: 0.4159 (1.1198) acc1: 82.8125 (37.5068) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1987 data: 0.1047 max mem: 3531 +Test: [200/333] eta: 0:00:28 acc1_pos: 14.145 acc5_pos: 100.000 loss: 0.4411 (1.0868) acc1: 81.7708 (39.6857) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.2222 data: 0.1302 max mem: 3531 +Test: [210/333] eta: 0:00:26 acc1_pos: 14.145 acc5_pos: 100.000 loss: 0.4753 (1.0591) acc1: 79.4271 (41.4791) acc5: 100.0000 (100.0000) acc1_neg: 65.365 acc5_neg: 100.000 time: 0.1985 data: 0.1048 max mem: 3532 +Test: [220/333] eta: 0:00:24 acc1_pos: 14.145 acc5_pos: 100.000 loss: 0.4679 (1.0326) acc1: 79.4271 (43.2798) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1794 data: 0.0860 max mem: 3532 +Test: [230/333] eta: 0:00:21 acc1_pos: 14.145 acc5_pos: 100.000 loss: 0.3187 (1.0004) acc1: 91.4062 (45.4095) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1776 data: 0.0865 max mem: 3532 +Test: [240/333] eta: 0:00:19 acc1_pos: 14.145 acc5_pos: 100.000 loss: 0.2926 (0.9715) acc1: 91.6667 (47.2953) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1889 data: 0.0985 max mem: 3532 +Test: [250/333] eta: 0:00:17 acc1_pos: 14.145 acc5_pos: 100.000 loss: 0.3194 (0.9483) acc1: 89.3229 (48.8276) acc5: 100.0000 (100.0000) acc1_neg: 75.781 acc5_neg: 100.000 time: 0.1772 data: 0.0866 max mem: 3532 +Test: [260/333] eta: 0:00:15 acc1_pos: 14.145 acc5_pos: 100.000 loss: 0.4113 (0.9291) acc1: 82.5521 (50.0858) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.1705 data: 0.0778 max mem: 3532 +Test: [270/333] eta: 0:00:13 acc1_pos: 14.145 acc5_pos: 100.000 loss: 0.4106 (0.9085) acc1: 83.0729 (51.4136) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.1959 data: 0.1011 max mem: 3532 +Test: [280/333] eta: 0:00:10 acc1_pos: 14.145 acc5_pos: 100.000 loss: 0.3826 (0.8916) acc1: 84.3750 (52.5226) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.2041 data: 0.1089 max mem: 3532 +Test: [290/333] eta: 0:00:08 acc1_pos: 14.145 acc5_pos: 100.000 loss: 0.4674 (0.8782) acc1: 79.9479 (53.4033) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.2013 data: 0.1085 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 14.145 acc5_pos: 100.000 loss: 0.5081 (0.8657) acc1: 78.3854 (54.2489) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.1844 data: 0.0936 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 14.145 acc5_pos: 100.000 loss: 0.4195 (0.8504) acc1: 81.2500 (55.2201) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1550 data: 0.0646 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 14.145 acc5_pos: 100.000 loss: 0.3667 (0.8360) acc1: 86.1979 (56.1502) acc5: 100.0000 (100.0000) acc1_neg: 69.010 acc5_neg: 100.000 time: 0.1347 data: 0.0451 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 14.145 acc5_pos: 100.000 loss: 0.5684 (0.8299) acc1: 72.1354 (56.5757) acc5: 100.0000 (100.0000) acc1_neg: 71.615 acc5_neg: 100.000 time: 0.1187 data: 0.0293 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 14.145 acc5_pos: 100.000 loss: 0.6051 (0.8287) acc1: 71.3542 (56.6554) acc5: 100.0000 (100.0000) acc1_neg: 69.859 acc5_neg: 100.000 time: 0.1314 data: 0.0347 max mem: 3532 +Test: Total time: 0:01:06 (0.1996 s / it) +* Overall Acc@1 56.655 Acc@5 100.000 + target == real Acc@1 28.119 Acc@5 100.000 + target == fake Acc@1 82.058 Acc@5 100.000 + loss 0.829 + PR-AUC 0.5815 ROC-AUC 0.5783 EER 44.35% (thr=0.7568) +Accuracy of the network on the 127843 test images: 56.7% +[rank0]:[W726 05:25:52.234530033 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:26:11.325000 3165569 torch/distributed/run.py:766] +W0726 05:26:11.325000 3165569 torch/distributed/run.py:766] ***************************************** +W0726 05:26:11.325000 3165569 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:26:11.325000 3165569 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:26:37.560979142 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W726 05:26:37.560995053 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:26:38.723773482 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:26:38.096362893 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +NCCL version 2.26.2+cuda12.2 +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:26:38.237910922 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:26:38.299977933 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:26:38.311735662 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:26:38.338670732 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n16_seed8_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed8_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:25:14 acc1_pos: 92.188 acc5_pos: 100.000 loss: 0.3090 (0.3090) acc1: 92.1875 (92.1875) acc5: 100.0000 (100.0000) time: 6.4451 data: 5.0019 max mem: 3531 +Test: [ 10/235] eta: 0:02:30 acc1_pos: 92.969 acc5_pos: 100.000 loss: 0.2986 (0.2969) acc1: 92.1875 (92.4006) acc5: 100.0000 (100.0000) time: 0.6679 data: 0.4551 max mem: 3531 +Test: [ 20/235] eta: 0:01:25 acc1_pos: 92.448 acc5_pos: 100.000 loss: 0.2907 (0.2948) acc1: 92.4479 (92.7207) acc5: 100.0000 (100.0000) time: 0.0948 data: 0.0004 max mem: 3531 +Test: [ 30/235] eta: 0:01:20 acc1_pos: 93.490 acc5_pos: 100.000 loss: 0.2849 (0.2916) acc1: 93.4896 (92.9436) acc5: 100.0000 (100.0000) time: 0.2397 data: 0.1409 max mem: 3531 +Test: [ 40/235] eta: 0:01:14 acc1_pos: 90.625 acc5_pos: 100.000 loss: 0.2880 (0.2920) acc1: 92.9688 (92.9243) acc5: 100.0000 (100.0000) time: 0.3616 data: 0.2675 max mem: 3531 +Test: [ 50/235] eta: 0:01:06 acc1_pos: 90.365 acc5_pos: 100.000 loss: 0.2965 (0.2928) acc1: 92.4479 (92.7543) acc5: 100.0000 (100.0000) time: 0.3069 data: 0.2170 max mem: 3531 +Test: [ 60/235] eta: 0:00:57 acc1_pos: 92.708 acc5_pos: 100.000 loss: 0.2923 (0.2935) acc1: 92.1875 (92.6955) acc5: 100.0000 (100.0000) time: 0.2200 data: 0.1292 max mem: 3531 +Test: [ 70/235] eta: 0:00:52 acc1_pos: 92.308 acc5_pos: 100.000 loss: 0.3148 (0.4091) acc1: 90.3646 (84.1733) acc5: 100.0000 (100.0000) acc1_neg: 20.573 acc5_neg: 100.000 time: 0.2158 data: 0.1250 max mem: 3531 +Test: [ 80/235] eta: 0:00:47 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.2149 (0.5091) acc1: 25.0000 (76.8229) acc5: 100.0000 (100.0000) acc1_neg: 23.438 acc5_neg: 100.000 time: 0.2532 data: 0.1594 max mem: 3531 +Test: [ 90/235] eta: 0:00:43 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.2246 (0.5895) acc1: 24.4792 (71.0279) acc5: 100.0000 (100.0000) acc1_neg: 20.312 acc5_neg: 100.000 time: 0.2371 data: 0.1409 max mem: 3531 +Test: [100/235] eta: 0:00:39 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.2339 (0.6520) acc1: 24.4792 (66.4320) acc5: 100.0000 (100.0000) acc1_neg: 19.792 acc5_neg: 100.000 time: 0.2107 data: 0.1150 max mem: 3531 +Test: [110/235] eta: 0:00:34 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.2148 (0.7025) acc1: 24.4792 (62.6431) acc5: 100.0000 (100.0000) acc1_neg: 23.177 acc5_neg: 100.000 time: 0.1719 data: 0.0776 max mem: 3531 +Test: [120/235] eta: 0:00:31 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.2495 (0.7515) acc1: 22.1354 (59.1060) acc5: 100.0000 (100.0000) acc1_neg: 20.833 acc5_neg: 100.000 time: 0.1766 data: 0.0854 max mem: 3531 +Test: [130/235] eta: 0:00:27 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.3030 (0.7936) acc1: 19.5312 (56.1029) acc5: 100.0000 (100.0000) acc1_neg: 19.792 acc5_neg: 100.000 time: 0.2031 data: 0.1109 max mem: 3531 +Test: [140/235] eta: 0:00:24 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.3243 (0.8325) acc1: 19.2708 (53.4741) acc5: 100.0000 (100.0000) acc1_neg: 14.062 acc5_neg: 100.000 time: 0.1990 data: 0.1057 max mem: 3531 +Test: [150/235] eta: 0:00:22 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.3224 (0.8647) acc1: 18.2292 (51.1607) acc5: 100.0000 (100.0000) acc1_neg: 21.094 acc5_neg: 100.000 time: 0.2125 data: 0.1209 max mem: 3531 +Test: [160/235] eta: 0:00:18 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.3139 (0.8918) acc1: 18.4896 (49.2511) acc5: 100.0000 (100.0000) acc1_neg: 22.656 acc5_neg: 100.000 time: 0.1716 data: 0.0813 max mem: 3531 +Test: [170/235] eta: 0:00:15 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.2694 (0.9067) acc1: 23.9583 (48.0766) acc5: 100.0000 (100.0000) acc1_neg: 28.906 acc5_neg: 100.000 time: 0.1295 data: 0.0396 max mem: 3531 +Test: [180/235] eta: 0:00:13 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.1240 (0.9188) acc1: 30.4688 (47.2002) acc5: 100.0000 (100.0000) acc1_neg: 30.729 acc5_neg: 100.000 time: 0.1571 data: 0.0672 max mem: 3531 +Test: [190/235] eta: 0:00:10 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.1254 (0.9302) acc1: 31.7708 (46.3160) acc5: 100.0000 (100.0000) acc1_neg: 28.646 acc5_neg: 100.000 time: 0.1632 data: 0.0734 max mem: 3531 +Test: [200/235] eta: 0:00:08 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.1419 (0.9414) acc1: 29.9479 (45.5043) acc5: 100.0000 (100.0000) acc1_neg: 27.865 acc5_neg: 100.000 time: 0.1580 data: 0.0682 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.1382 (0.9499) acc1: 31.5104 (44.8484) acc5: 100.0000 (100.0000) acc1_neg: 30.208 acc5_neg: 100.000 time: 0.1521 data: 0.0624 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.1309 (0.9581) acc1: 31.7708 (44.2213) acc5: 100.0000 (100.0000) acc1_neg: 28.646 acc5_neg: 100.000 time: 0.1355 data: 0.0460 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.1221 (0.9651) acc1: 31.7708 (43.6824) acc5: 100.0000 (100.0000) acc1_neg: 31.771 acc5_neg: 100.000 time: 0.1232 data: 0.0338 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.1291 (0.9695) acc1: 31.5104 (43.5096) acc5: 100.0000 (100.0000) acc1_neg: 26.087 acc5_neg: 100.000 time: 0.1141 data: 0.0279 max mem: 3532 +Test: Total time: 0:00:50 (0.2167 s / it) +* Overall Acc@1 43.510 Acc@5 100.000 + target == real Acc@1 92.711 Acc@5 100.000 + target == fake Acc@1 25.714 Acc@5 100.000 + loss 0.969 + PR-AUC 0.8567 ROC-AUC 0.6948 EER 35.86% (thr=0.2313) +Accuracy of the network on the 89879 test images: 43.5% +[rank0]:[W726 05:27:36.556029609 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:27:56.039000 3166129 torch/distributed/run.py:766] +W0726 05:27:56.039000 3166129 torch/distributed/run.py:766] ***************************************** +W0726 05:27:56.039000 3166129 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:27:56.039000 3166129 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:28:23.710242749 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:28:23.875927017 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:28:23.966447191 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:28:23.280287375 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:28:23.384986105 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:28:23.404232617 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:28:23.433827073 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:28:23.454743451 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n16_seed8_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed8_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:30:46 acc1_pos: 3.385 acc5_pos: 100.000 loss: 1.5687 (1.5687) acc1: 3.3854 (3.3854) acc5: 100.0000 (100.0000) time: 6.8650 data: 5.1467 max mem: 3531 +Test: [ 10/269] eta: 0:03:03 acc1_pos: 15.885 acc5_pos: 100.000 loss: 1.4056 (1.3988) acc1: 11.4583 (13.8021) acc5: 100.0000 (100.0000) time: 0.7095 data: 0.4683 max mem: 3531 +Test: [ 20/269] eta: 0:01:44 acc1_pos: 19.792 acc5_pos: 100.000 loss: 1.2977 (1.2643) acc1: 15.8854 (21.2426) acc5: 100.0000 (100.0000) time: 0.0969 data: 0.0004 max mem: 3531 +Test: [ 30/269] eta: 0:01:25 acc1_pos: 1.823 acc5_pos: 100.000 loss: 1.0345 (1.2284) acc1: 23.1771 (24.0507) acc5: 100.0000 (100.0000) time: 0.1651 data: 0.0681 max mem: 3531 +Test: [ 40/269] eta: 0:01:15 acc1_pos: 3.448 acc5_pos: 100.000 loss: 0.6263 (1.0274) acc1: 63.2812 (39.0371) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2316 data: 0.1392 max mem: 3531 +Test: [ 50/269] eta: 0:01:06 acc1_pos: 3.448 acc5_pos: 100.000 loss: 0.2062 (0.8665) acc1: 98.1771 (50.6025) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2159 data: 0.1249 max mem: 3531 +Test: [ 60/269] eta: 0:00:59 acc1_pos: 3.448 acc5_pos: 100.000 loss: 0.2098 (0.7627) acc1: 97.3958 (58.2437) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2033 data: 0.1106 max mem: 3531 +Test: [ 70/269] eta: 0:00:54 acc1_pos: 3.448 acc5_pos: 100.000 loss: 0.2247 (0.6834) acc1: 97.3958 (63.8828) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2005 data: 0.1054 max mem: 3531 +Test: [ 80/269] eta: 0:00:50 acc1_pos: 3.448 acc5_pos: 100.000 loss: 0.1921 (0.6238) acc1: 98.9583 (68.2163) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2098 data: 0.1113 max mem: 3531 +Test: [ 90/269] eta: 0:00:46 acc1_pos: 3.448 acc5_pos: 100.000 loss: 0.2679 (0.6015) acc1: 97.3958 (69.9119) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.2146 data: 0.1187 max mem: 3531 +Test: [100/269] eta: 0:00:42 acc1_pos: 3.448 acc5_pos: 100.000 loss: 0.3290 (0.5750) acc1: 89.5833 (71.9317) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1801 data: 0.0881 max mem: 3531 +Test: [110/269] eta: 0:00:39 acc1_pos: 3.448 acc5_pos: 100.000 loss: 0.3159 (0.5508) acc1: 93.4896 (74.0897) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1785 data: 0.0841 max mem: 3531 +Test: [120/269] eta: 0:00:35 acc1_pos: 3.448 acc5_pos: 100.000 loss: 0.2864 (0.5283) acc1: 96.3542 (75.8759) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1961 data: 0.1012 max mem: 3531 +Test: [130/269] eta: 0:00:33 acc1_pos: 3.448 acc5_pos: 100.000 loss: 0.3060 (0.5133) acc1: 94.7917 (76.9919) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.1955 data: 0.1034 max mem: 3531 +Test: [140/269] eta: 0:00:30 acc1_pos: 3.448 acc5_pos: 100.000 loss: 0.3514 (0.5100) acc1: 86.9792 (77.0704) acc5: 100.0000 (100.0000) acc1_neg: 59.115 acc5_neg: 100.000 time: 0.2050 data: 0.1144 max mem: 3531 +Test: [150/269] eta: 0:00:27 acc1_pos: 3.448 acc5_pos: 100.000 loss: 0.4271 (0.5085) acc1: 73.6979 (76.9868) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.2043 data: 0.1129 max mem: 3531 +Test: [160/269] eta: 0:00:25 acc1_pos: 3.448 acc5_pos: 100.000 loss: 0.3611 (0.4964) acc1: 86.7188 (77.9066) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1840 data: 0.0911 max mem: 3531 +Test: [170/269] eta: 0:00:22 acc1_pos: 3.448 acc5_pos: 100.000 loss: 0.2651 (0.4825) acc1: 98.1771 (79.1286) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1729 data: 0.0792 max mem: 3531 +Test: [180/269] eta: 0:00:19 acc1_pos: 3.448 acc5_pos: 100.000 loss: 0.2511 (0.4699) acc1: 98.6979 (80.1580) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1803 data: 0.0871 max mem: 3531 +Test: [190/269] eta: 0:00:17 acc1_pos: 3.448 acc5_pos: 100.000 loss: 0.2561 (0.4622) acc1: 97.1354 (80.8273) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1920 data: 0.0997 max mem: 3531 +Test: [200/269] eta: 0:00:15 acc1_pos: 3.448 acc5_pos: 100.000 loss: 0.3234 (0.4560) acc1: 92.1875 (81.2759) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1939 data: 0.1031 max mem: 3531 +Test: [210/269] eta: 0:00:12 acc1_pos: 3.448 acc5_pos: 100.000 loss: 0.3635 (0.4546) acc1: 88.8021 (81.3598) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1853 data: 0.0926 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 3.448 acc5_pos: 100.000 loss: 0.4036 (0.4524) acc1: 82.8125 (81.3631) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1902 data: 0.0956 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 3.448 acc5_pos: 100.000 loss: 0.2328 (0.4430) acc1: 94.5312 (81.9625) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1846 data: 0.0923 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 3.448 acc5_pos: 100.000 loss: 0.2528 (0.4422) acc1: 96.8750 (81.9999) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.1443 data: 0.0533 max mem: 3532 +Test: [250/269] eta: 0:00:03 acc1_pos: 3.448 acc5_pos: 100.000 loss: 0.3846 (0.4429) acc1: 83.0729 (81.8870) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.1270 data: 0.0366 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 3.448 acc5_pos: 100.000 loss: 0.4085 (0.4418) acc1: 83.0729 (81.8646) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1293 data: 0.0398 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 3.448 acc5_pos: 100.000 loss: 0.2507 (0.4355) acc1: 98.4375 (82.3657) acc5: 100.0000 (100.0000) acc1_neg: 99.738 acc5_neg: 100.000 time: 0.1222 data: 0.0292 max mem: 3532 +Test: Total time: 0:00:54 (0.2041 s / it) +* Overall Acc@1 82.366 Acc@5 100.000 + target == real Acc@1 23.218 Acc@5 100.000 + target == fake Acc@1 90.438 Acc@5 100.000 + loss 0.436 + PR-AUC 0.9190 ROC-AUC 0.6411 EER 39.62% (thr=0.7539) +Accuracy of the network on the 103293 test images: 82.4% +[rank0]:[W726 05:29:26.695430931 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T05:29:32Z] eval_done tag=fixed500k_real_n16_seed8_resampled +[2026-07-26T05:29:32Z] eval_start tag=fixed500k_fake_n32_seed8_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed8_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T05:29:32Z +hostname=n111-046-081 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed8_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n32_seed8_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 05:29:44.753000 3166697 torch/distributed/run.py:766] +W0726 05:29:44.753000 3166697 torch/distributed/run.py:766] ***************************************** +W0726 05:29:44.753000 3166697 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:29:44.753000 3166697 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:30:11.145173683 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:30:11.277574628 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:30:12.711325271 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:30:12.734792077 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:30:12.743681610 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:30:12.811477595 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:30:12.826548503 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:30:12.848477384 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n32_seed8_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed8_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:20:12 acc1_pos: 51.562 acc5_pos: 100.000 loss: 0.6966 (0.6966) acc1: 51.5625 (51.5625) acc5: 100.0000 (100.0000) time: 5.1835 data: 2.7223 max mem: 3531 +Test: [ 10/234] eta: 0:02:03 acc1_pos: 53.906 acc5_pos: 100.000 loss: 0.7613 (0.7817) acc1: 51.5625 (50.2604) acc5: 100.0000 (100.0000) time: 0.5531 data: 0.2478 max mem: 3531 +Test: [ 20/234] eta: 0:01:22 acc1_pos: 93.490 acc5_pos: 100.000 loss: 0.8204 (0.7818) acc1: 42.4479 (49.3180) acc5: 100.0000 (100.0000) time: 0.1455 data: 0.0551 max mem: 3531 +Test: [ 30/234] eta: 0:01:15 acc1_pos: 41.927 acc5_pos: 100.000 loss: 0.2845 (0.6455) acc1: 92.7083 (60.9879) acc5: 100.0000 (100.0000) time: 0.2699 data: 0.1783 max mem: 3531 +Test: [ 40/234] eta: 0:01:08 acc1_pos: 99.479 acc5_pos: 100.000 loss: 0.2823 (0.6471) acc1: 93.2292 (61.0772) acc5: 100.0000 (100.0000) time: 0.3181 data: 0.2258 max mem: 3531 +Test: [ 50/234] eta: 0:01:00 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1448 (0.5478) acc1: 98.1771 (68.4079) acc5: 100.0000 (100.0000) time: 0.2636 data: 0.1697 max mem: 3531 +Test: [ 60/234] eta: 0:00:53 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.1228 (0.4751) acc1: 98.9583 (73.5784) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2105 data: 0.1155 max mem: 3531 +Test: [ 70/234] eta: 0:00:47 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.1035 (0.4861) acc1: 100.0000 (73.2284) acc5: 100.0000 (100.0000) acc1_neg: 57.292 acc5_neg: 100.000 time: 0.1927 data: 0.1001 max mem: 3531 +Test: [ 80/234] eta: 0:00:42 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.7389 (0.4941) acc1: 59.3750 (73.0646) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1843 data: 0.0919 max mem: 3531 +Test: [ 90/234] eta: 0:00:38 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.1840 (0.4570) acc1: 96.0938 (75.6696) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1971 data: 0.1020 max mem: 3531 +Test: [100/234] eta: 0:00:35 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.1998 (0.4511) acc1: 95.3125 (76.3021) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2199 data: 0.1240 max mem: 3531 +Test: [110/234] eta: 0:00:31 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.3329 (0.4397) acc1: 83.0729 (77.2593) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.2019 data: 0.1044 max mem: 3531 +Test: [120/234] eta: 0:00:28 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.2790 (0.4423) acc1: 88.0208 (77.4514) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1779 data: 0.0820 max mem: 3531 +Test: [130/234] eta: 0:00:25 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.3014 (0.4355) acc1: 90.6250 (78.2522) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.1952 data: 0.1008 max mem: 3531 +Test: [140/234] eta: 0:00:22 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.3666 (0.4535) acc1: 84.1146 (77.4934) acc5: 100.0000 (100.0000) acc1_neg: 37.760 acc5_neg: 100.000 time: 0.2003 data: 0.1055 max mem: 3531 +Test: [150/234] eta: 0:00:20 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.4609 (0.4577) acc1: 75.2604 (77.3903) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.1927 data: 0.0995 max mem: 3531 +Test: [160/234] eta: 0:00:17 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.4169 (0.4518) acc1: 84.1146 (77.9455) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.1966 data: 0.1046 max mem: 3531 +Test: [170/234] eta: 0:00:15 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.4169 (0.4505) acc1: 82.5521 (78.2149) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.2085 data: 0.1156 max mem: 3531 +Test: [180/234] eta: 0:00:12 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.4512 (0.4613) acc1: 80.4688 (77.6516) acc5: 100.0000 (100.0000) acc1_neg: 30.729 acc5_neg: 100.000 time: 0.2200 data: 0.1259 max mem: 3531 +Test: [190/234] eta: 0:00:10 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.5211 (0.4666) acc1: 73.1771 (77.3328) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.1750 data: 0.0816 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.5124 (0.4567) acc1: 74.2188 (77.9747) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1407 data: 0.0487 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.1170 (0.4433) acc1: 99.7396 (78.8791) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1371 data: 0.0459 max mem: 3532 +Test: [220/234] eta: 0:00:03 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.2737 (0.4366) acc1: 92.1875 (79.4448) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.1267 data: 0.0356 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.3612 (0.4425) acc1: 71.0938 (78.9209) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.1161 data: 0.0263 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.5600 (0.4442) acc1: 68.0934 (78.7850) acc5: 100.0000 (100.0000) acc1_neg: 68.093 acc5_neg: 100.000 time: 0.1125 data: 0.0217 max mem: 3532 +Test: Total time: 0:00:49 (0.2119 s / it) +* Overall Acc@1 78.785 Acc@5 100.000 + target == real Acc@1 68.630 Acc@5 100.000 + target == fake Acc@1 81.647 Acc@5 100.000 + loss 0.444 + PR-AUC 0.9567 ROC-AUC 0.8593 EER 23.68% (thr=0.5938) +Accuracy of the network on the 89729 test images: 78.8% +[rank0]:[W726 05:31:11.996520172 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:31:29.322000 3167251 torch/distributed/run.py:766] +W0726 05:31:29.322000 3167251 torch/distributed/run.py:766] ***************************************** +W0726 05:31:29.322000 3167251 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:31:29.322000 3167251 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:31:56.347178128 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:31:56.477768040 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:31:56.516741446 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:31:57.942955596 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +[rank4]:[W726 05:31:57.964517214 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:31:57.975168922 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:31:57.173530164 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:31:57.197662371 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n32_seed8_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed8_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:58:29 acc1_pos: 24.479 acc5_pos: 100.000 loss: 1.1024 (1.1024) acc1: 24.4792 (24.4792) acc5: 100.0000 (100.0000) time: 7.5030 data: 5.1031 max mem: 3531 +Test: [ 10/2387] eta: 0:30:20 acc1_pos: 1.562 acc5_pos: 100.000 loss: 1.3249 (1.2779) acc1: 15.1042 (15.7197) acc5: 100.0000 (100.0000) time: 0.7658 data: 0.4643 max mem: 3531 +Test: [ 20/2387] eta: 0:17:42 acc1_pos: 7.031 acc5_pos: 100.000 loss: 1.3283 (1.2913) acc1: 12.5000 (15.6994) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:13:58 acc1_pos: 4.427 acc5_pos: 100.000 loss: 1.3573 (1.3204) acc1: 11.7188 (14.8017) acc5: 100.0000 (100.0000) time: 0.1304 data: 0.0353 max mem: 3531 +Test: [ 40/2387] eta: 0:12:43 acc1_pos: 7.031 acc5_pos: 100.000 loss: 1.3803 (1.3443) acc1: 7.0312 (12.6588) acc5: 100.0000 (100.0000) time: 0.1957 data: 0.1044 max mem: 3531 +Test: [ 50/2387] eta: 0:11:51 acc1_pos: 28.906 acc5_pos: 100.000 loss: 1.3254 (1.3255) acc1: 8.3333 (13.2915) acc5: 100.0000 (100.0000) time: 0.2250 data: 0.1339 max mem: 3531 +Test: [ 60/2387] eta: 0:11:23 acc1_pos: 23.958 acc5_pos: 100.000 loss: 1.2109 (1.3065) acc1: 17.1875 (14.2930) acc5: 100.0000 (100.0000) time: 0.2291 data: 0.1338 max mem: 3531 +Test: [ 70/2387] eta: 0:11:04 acc1_pos: 22.396 acc5_pos: 100.000 loss: 1.2692 (1.3072) acc1: 15.8854 (13.9048) acc5: 100.0000 (100.0000) time: 0.2414 data: 0.1428 max mem: 3531 +Test: [ 80/2387] eta: 0:10:55 acc1_pos: 53.906 acc5_pos: 100.000 loss: 1.1675 (1.2709) acc1: 15.8854 (16.1490) acc5: 100.0000 (100.0000) time: 0.2553 data: 0.1575 max mem: 3531 +Test: [ 90/2387] eta: 0:10:34 acc1_pos: 45.573 acc5_pos: 100.000 loss: 0.9093 (1.2245) acc1: 37.7604 (18.6813) acc5: 100.0000 (100.0000) time: 0.2394 data: 0.1441 max mem: 3531 +Test: [ 100/2387] eta: 0:10:21 acc1_pos: 45.312 acc5_pos: 100.000 loss: 0.8531 (1.1948) acc1: 40.1042 (20.5935) acc5: 100.0000 (100.0000) time: 0.2215 data: 0.1306 max mem: 3531 +Test: [ 110/2387] eta: 0:10:19 acc1_pos: 41.146 acc5_pos: 100.000 loss: 0.8531 (1.1641) acc1: 35.6771 (22.4568) acc5: 100.0000 (100.0000) time: 0.2532 data: 0.1610 max mem: 3531 +Test: [ 120/2387] eta: 0:10:14 acc1_pos: 26.823 acc5_pos: 100.000 loss: 0.8618 (1.1430) acc1: 37.5000 (23.7108) acc5: 100.0000 (100.0000) time: 0.2679 data: 0.1736 max mem: 3531 +Test: [ 130/2387] eta: 0:10:10 acc1_pos: 35.156 acc5_pos: 100.000 loss: 0.8929 (1.1240) acc1: 35.1562 (24.5925) acc5: 100.0000 (100.0000) time: 0.2599 data: 0.1662 max mem: 3531 +Test: [ 140/2387] eta: 0:10:00 acc1_pos: 31.510 acc5_pos: 100.000 loss: 0.8929 (1.1081) acc1: 35.1562 (25.3084) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1487 max mem: 3531 +Test: [ 150/2387] eta: 0:09:52 acc1_pos: 16.406 acc5_pos: 100.000 loss: 0.8811 (1.0915) acc1: 35.9375 (26.0503) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1335 max mem: 3531 +Test: [ 160/2387] eta: 0:09:45 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.8053 (1.0618) acc1: 38.8021 (28.3078) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2299 data: 0.1343 max mem: 3531 +Test: [ 170/2387] eta: 0:09:29 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1904 (1.0103) acc1: 98.4375 (32.4348) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1949 data: 0.0979 max mem: 3531 +Test: [ 180/2387] eta: 0:09:18 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1817 (0.9645) acc1: 98.9583 (36.1015) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1749 data: 0.0802 max mem: 3531 +Test: [ 190/2387] eta: 0:09:08 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1797 (0.9234) acc1: 98.6979 (39.3720) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1899 data: 0.0950 max mem: 3531 +Test: [ 200/2387] eta: 0:08:59 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1764 (0.8862) acc1: 98.6979 (42.3313) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1899 data: 0.0917 max mem: 3531 +Test: [ 210/2387] eta: 0:08:52 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1799 (0.8529) acc1: 98.6979 (44.9941) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1949 data: 0.0967 max mem: 3532 +Test: [ 220/2387] eta: 0:08:43 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1820 (0.8226) acc1: 98.1771 (47.4088) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1899 data: 0.0946 max mem: 3532 +Test: [ 230/2387] eta: 0:08:38 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1821 (0.7949) acc1: 98.4375 (49.6190) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1949 data: 0.0997 max mem: 3532 +Test: [ 240/2387] eta: 0:08:29 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1910 (0.7729) acc1: 97.6562 (51.5150) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1899 data: 0.0928 max mem: 3532 +Test: [ 250/2387] eta: 0:08:23 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1690 (0.7464) acc1: 96.3542 (53.4134) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0800 max mem: 3532 +Test: [ 260/2387] eta: 0:08:16 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.0823 (0.7209) acc1: 100.0000 (55.1984) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0872 max mem: 3532 +Test: [ 270/2387] eta: 0:08:10 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.0814 (0.6973) acc1: 100.0000 (56.8506) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0894 max mem: 3532 +Test: [ 280/2387] eta: 0:08:03 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.0809 (0.6754) acc1: 100.0000 (58.3862) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1800 data: 0.0844 max mem: 3532 +Test: [ 290/2387] eta: 0:07:59 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.0807 (0.6550) acc1: 100.0000 (59.8162) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0893 max mem: 3532 +Test: [ 300/2387] eta: 0:07:54 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.0816 (0.6361) acc1: 100.0000 (61.1512) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1947 data: 0.0991 max mem: 3532 +Test: [ 310/2387] eta: 0:07:43 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.0856 (0.6235) acc1: 100.0000 (62.2036) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1449 data: 0.0491 max mem: 3532 +Test: [ 320/2387] eta: 0:07:39 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.4246 (0.6176) acc1: 88.2812 (62.9989) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.1499 data: 0.0535 max mem: 3532 +Test: [ 330/2387] eta: 0:07:36 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.4291 (0.6119) acc1: 87.7604 (63.7557) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1999 data: 0.1052 max mem: 3532 +Test: [ 340/2387] eta: 0:07:33 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.4278 (0.6065) acc1: 87.7604 (64.4657) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.2049 data: 0.1063 max mem: 3532 +Test: [ 350/2387] eta: 0:07:28 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.4173 (0.5954) acc1: 90.6250 (65.4314) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1949 data: 0.0928 max mem: 3532 +Test: [ 360/2387] eta: 0:07:25 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1774 (0.5831) acc1: 99.7396 (66.3752) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1899 data: 0.0897 max mem: 3532 +Test: [ 370/2387] eta: 0:07:20 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1483 (0.5714) acc1: 99.4792 (67.2640) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1903 data: 0.0946 max mem: 3532 +Test: [ 380/2387] eta: 0:07:16 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1489 (0.5603) acc1: 99.4792 (68.1061) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1799 data: 0.0849 max mem: 3532 +Test: [ 390/2387] eta: 0:07:12 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1493 (0.5498) acc1: 99.4792 (68.9085) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1816 data: 0.0847 max mem: 3532 +Test: [ 400/2387] eta: 0:07:09 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1498 (0.5398) acc1: 99.4792 (69.6735) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1884 data: 0.0931 max mem: 3532 +Test: [ 410/2387] eta: 0:07:05 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1481 (0.5303) acc1: 99.4792 (70.3968) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1855 data: 0.0932 max mem: 3532 +Test: [ 420/2387] eta: 0:07:02 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1474 (0.5212) acc1: 99.4792 (71.0882) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1836 data: 0.0927 max mem: 3532 +Test: [ 430/2387] eta: 0:06:58 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1506 (0.5135) acc1: 99.2188 (71.7173) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1886 data: 0.0971 max mem: 3532 +Test: [ 440/2387] eta: 0:06:55 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1523 (0.5053) acc1: 98.1771 (72.3397) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1856 data: 0.0931 max mem: 3533 +Test: [ 450/2387] eta: 0:06:53 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1096 (0.4965) acc1: 100.0000 (72.9530) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2015 data: 0.1071 max mem: 3533 +Test: [ 460/2387] eta: 0:06:50 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1093 (0.4881) acc1: 100.0000 (73.5397) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2133 data: 0.1192 max mem: 3533 +Test: [ 470/2387] eta: 0:06:48 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1096 (0.4801) acc1: 100.0000 (74.1015) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2086 data: 0.1163 max mem: 3533 +Test: [ 480/2387] eta: 0:06:46 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1098 (0.4724) acc1: 100.0000 (74.6400) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2086 data: 0.1176 max mem: 3533 +Test: [ 490/2387] eta: 0:06:42 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1100 (0.4650) acc1: 100.0000 (75.1565) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1941 data: 0.1032 max mem: 3533 +Test: [ 500/2387] eta: 0:06:39 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1099 (0.4579) acc1: 100.0000 (75.6523) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1768 data: 0.0817 max mem: 3533 +Test: [ 510/2387] eta: 0:06:35 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1096 (0.4514) acc1: 100.0000 (76.1288) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1712 data: 0.0767 max mem: 3533 +Test: [ 520/2387] eta: 0:06:33 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1310 (0.4456) acc1: 100.0000 (76.5870) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1880 data: 0.0945 max mem: 3533 +Test: [ 530/2387] eta: 0:06:32 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1303 (0.4397) acc1: 100.0000 (77.0279) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2226 data: 0.1262 max mem: 3533 +Test: [ 540/2387] eta: 0:06:29 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1287 (0.4339) acc1: 100.0000 (77.4525) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2099 data: 0.1154 max mem: 3533 +Test: [ 550/2387] eta: 0:06:24 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1285 (0.4284) acc1: 100.0000 (77.8617) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1615 data: 0.0674 max mem: 3533 +Test: [ 560/2387] eta: 0:06:21 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1290 (0.4230) acc1: 100.0000 (78.2564) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1609 data: 0.0675 max mem: 3533 +Test: [ 570/2387] eta: 0:06:19 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1290 (0.4179) acc1: 100.0000 (78.6372) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.1041 max mem: 3533 +Test: [ 580/2387] eta: 0:06:17 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1292 (0.4129) acc1: 100.0000 (79.0049) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2044 data: 0.1117 max mem: 3533 +Test: [ 590/2387] eta: 0:06:14 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1294 (0.4081) acc1: 100.0000 (79.3601) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1940 data: 0.1010 max mem: 3533 +Test: [ 600/2387] eta: 0:06:12 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1317 (0.4040) acc1: 100.0000 (79.7035) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1920 data: 0.0995 max mem: 3533 +Test: [ 610/2387] eta: 0:06:09 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1440 (0.3996) acc1: 100.0000 (80.0344) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1919 data: 0.0995 max mem: 3533 +Test: [ 620/2387] eta: 0:06:07 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1364 (0.3954) acc1: 100.0000 (80.3547) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1944 data: 0.1018 max mem: 3533 +Test: [ 630/2387] eta: 0:06:05 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1365 (0.3913) acc1: 100.0000 (80.6652) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2044 data: 0.1114 max mem: 3533 +Test: [ 640/2387] eta: 0:06:03 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1369 (0.3873) acc1: 100.0000 (80.9656) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2148 data: 0.1224 max mem: 3533 +Test: [ 650/2387] eta: 0:06:00 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1369 (0.3835) acc1: 100.0000 (81.2560) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1950 data: 0.1007 max mem: 3533 +Test: [ 660/2387] eta: 0:05:57 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1371 (0.3798) acc1: 100.0000 (81.5388) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1801 data: 0.0863 max mem: 3533 +Test: [ 670/2387] eta: 0:05:55 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1371 (0.3761) acc1: 100.0000 (81.8120) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1916 data: 0.0987 max mem: 3534 +Test: [ 680/2387] eta: 0:05:52 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1380 (0.3730) acc1: 100.0000 (82.0791) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1887 data: 0.0963 max mem: 3534 +Test: [ 690/2387] eta: 0:05:50 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1791 (0.3711) acc1: 100.0000 (82.2480) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1908 data: 0.0965 max mem: 3534 +Test: [ 700/2387] eta: 0:05:47 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.2520 (0.3694) acc1: 90.8854 (82.3760) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1940 data: 0.0991 max mem: 3534 +Test: [ 710/2387] eta: 0:05:44 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.2519 (0.3678) acc1: 91.4062 (82.5052) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1799 data: 0.0857 max mem: 3534 +Test: [ 720/2387] eta: 0:05:43 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.2510 (0.3662) acc1: 92.1875 (82.6359) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1908 data: 0.0956 max mem: 3534 +Test: [ 730/2387] eta: 0:05:40 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.2544 (0.3647) acc1: 91.4062 (82.7498) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1949 data: 0.1022 max mem: 3534 +Test: [ 740/2387] eta: 0:05:37 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.2544 (0.3632) acc1: 90.8854 (82.8635) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.1812 data: 0.0878 max mem: 3534 +Test: [ 750/2387] eta: 0:05:35 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.2529 (0.3617) acc1: 91.1458 (82.9803) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1849 data: 0.0900 max mem: 3534 +Test: [ 760/2387] eta: 0:05:32 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.2492 (0.3601) acc1: 91.9271 (83.1051) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1794 data: 0.0864 max mem: 3534 +Test: [ 770/2387] eta: 0:05:29 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.2079 (0.3578) acc1: 94.7917 (83.2830) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1656 data: 0.0722 max mem: 3534 +Test: [ 780/2387] eta: 0:05:26 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1750 (0.3552) acc1: 98.4375 (83.4881) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1559 data: 0.0622 max mem: 3534 +Test: [ 790/2387] eta: 0:05:24 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1449 (0.3528) acc1: 99.4792 (83.6497) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.1692 data: 0.0763 max mem: 3534 +Test: [ 800/2387] eta: 0:05:21 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1313 (0.3501) acc1: 100.0000 (83.8460) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0918 max mem: 3534 +Test: [ 810/2387] eta: 0:05:18 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1361 (0.3476) acc1: 100.0000 (84.0350) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1699 data: 0.0781 max mem: 3534 +Test: [ 820/2387] eta: 0:05:16 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1432 (0.3453) acc1: 100.0000 (84.2078) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1677 data: 0.0739 max mem: 3534 +Test: [ 830/2387] eta: 0:05:14 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1991 (0.3444) acc1: 96.6146 (84.3227) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1849 data: 0.0900 max mem: 3534 +Test: [ 840/2387] eta: 0:05:11 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.2312 (0.3430) acc1: 96.0938 (84.4698) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1797 data: 0.0871 max mem: 3534 +Test: [ 850/2387] eta: 0:05:09 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.2266 (0.3415) acc1: 96.0938 (84.6082) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1758 data: 0.0837 max mem: 3534 +Test: [ 860/2387] eta: 0:05:06 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.2051 (0.3399) acc1: 97.6562 (84.7655) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1871 data: 0.0927 max mem: 3534 +Test: [ 870/2387] eta: 0:05:04 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.2042 (0.3383) acc1: 98.4375 (84.9162) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1953 data: 0.1016 max mem: 3534 +Test: [ 880/2387] eta: 0:05:02 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.2042 (0.3369) acc1: 97.9167 (85.0581) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1876 data: 0.0958 max mem: 3534 +Test: [ 890/2387] eta: 0:05:00 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1944 (0.3353) acc1: 98.9583 (85.2071) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1749 data: 0.0827 max mem: 3534 +Test: [ 900/2387] eta: 0:04:58 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.2236 (0.3382) acc1: 94.5312 (85.0062) acc5: 100.0000 (100.0000) acc1_neg: 60.417 acc5_neg: 100.000 time: 0.1899 data: 0.0971 max mem: 3535 +Test: [ 910/2387] eta: 0:04:56 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.6906 (0.3420) acc1: 59.6354 (84.7269) acc5: 100.0000 (100.0000) acc1_neg: 58.594 acc5_neg: 100.000 time: 0.2102 data: 0.1176 max mem: 3535 +Test: [ 920/2387] eta: 0:04:53 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.6953 (0.3459) acc1: 58.8542 (84.4505) acc5: 100.0000 (100.0000) acc1_neg: 57.031 acc5_neg: 100.000 time: 0.1761 data: 0.0833 max mem: 3535 +Test: [ 930/2387] eta: 0:04:50 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.6942 (0.3495) acc1: 59.1146 (84.1828) acc5: 100.0000 (100.0000) acc1_neg: 55.990 acc5_neg: 100.000 time: 0.1482 data: 0.0555 max mem: 3535 +Test: [ 940/2387] eta: 0:04:48 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.6894 (0.3531) acc1: 59.3750 (83.9283) acc5: 100.0000 (100.0000) acc1_neg: 58.073 acc5_neg: 100.000 time: 0.1753 data: 0.0822 max mem: 3535 +Test: [ 950/2387] eta: 0:04:46 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.6724 (0.3565) acc1: 59.3750 (83.6756) acc5: 100.0000 (100.0000) acc1_neg: 61.198 acc5_neg: 100.000 time: 0.1899 data: 0.0941 max mem: 3535 +Test: [ 960/2387] eta: 0:04:44 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.6724 (0.3599) acc1: 59.8958 (83.4271) acc5: 100.0000 (100.0000) acc1_neg: 58.594 acc5_neg: 100.000 time: 0.1945 data: 0.1010 max mem: 3535 +Test: [ 970/2387] eta: 0:04:42 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.6819 (0.3632) acc1: 59.1146 (83.1810) acc5: 100.0000 (100.0000) acc1_neg: 60.417 acc5_neg: 100.000 time: 0.1999 data: 0.1092 max mem: 3535 +Test: [ 980/2387] eta: 0:04:40 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.6666 (0.3632) acc1: 60.4167 (83.1884) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.2057 data: 0.1146 max mem: 3535 +Test: [ 990/2387] eta: 0:04:38 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.2077 (0.3618) acc1: 97.3958 (83.3210) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2009 data: 0.1099 max mem: 3535 +Test: [1000/2387] eta: 0:04:36 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1954 (0.3603) acc1: 98.4375 (83.4525) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1929 data: 0.0989 max mem: 3535 +Test: [1010/2387] eta: 0:04:34 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.2265 (0.3593) acc1: 95.8333 (83.5551) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1889 data: 0.0935 max mem: 3535 +Test: [1020/2387] eta: 0:04:32 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.2371 (0.3581) acc1: 95.3125 (83.6746) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1864 data: 0.0930 max mem: 3535 +Test: [1030/2387] eta: 0:04:30 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1784 (0.3563) acc1: 98.6979 (83.8201) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2018 data: 0.1095 max mem: 3535 +Test: [1040/2387] eta: 0:04:28 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1438 (0.3543) acc1: 99.4792 (83.9660) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2063 data: 0.1150 max mem: 3535 +Test: [1050/2387] eta: 0:04:26 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1357 (0.3522) acc1: 99.7396 (84.1121) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1884 data: 0.0974 max mem: 3535 +Test: [1060/2387] eta: 0:04:23 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1373 (0.3504) acc1: 99.7396 (84.2550) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1797 data: 0.0890 max mem: 3535 +Test: [1070/2387] eta: 0:04:21 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1529 (0.3488) acc1: 99.2188 (84.3748) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1770 data: 0.0852 max mem: 3535 +Test: [1080/2387] eta: 0:04:19 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.2042 (0.3480) acc1: 96.8750 (84.4692) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1799 data: 0.0854 max mem: 3535 +Test: [1090/2387] eta: 0:04:17 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1969 (0.3463) acc1: 97.9167 (84.6008) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1891 data: 0.0953 max mem: 3535 +Test: [1100/2387] eta: 0:04:15 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1552 (0.3447) acc1: 99.2188 (84.7314) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1808 data: 0.0895 max mem: 3535 +Test: [1110/2387] eta: 0:04:12 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1569 (0.3433) acc1: 99.2188 (84.8220) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1790 data: 0.0882 max mem: 3535 +Test: [1120/2387] eta: 0:04:10 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1601 (0.3417) acc1: 98.9583 (84.9451) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1828 data: 0.0920 max mem: 3536 +Test: [1130/2387] eta: 0:04:08 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1605 (0.3402) acc1: 98.9583 (85.0727) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1791 data: 0.0868 max mem: 3536 +Test: [1140/2387] eta: 0:04:06 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1941 (0.3393) acc1: 98.4375 (85.1683) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1764 data: 0.0837 max mem: 3536 +Test: [1150/2387] eta: 0:04:04 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.2298 (0.3383) acc1: 97.6562 (85.2678) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1751 data: 0.0832 max mem: 3536 +Test: [1160/2387] eta: 0:04:01 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1607 (0.3368) acc1: 98.6979 (85.3826) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1773 data: 0.0863 max mem: 3536 +Test: [1170/2387] eta: 0:03:59 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1496 (0.3352) acc1: 99.4792 (85.5012) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1683 data: 0.0777 max mem: 3536 +Test: [1180/2387] eta: 0:03:57 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1460 (0.3336) acc1: 99.4792 (85.6109) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1694 data: 0.0778 max mem: 3536 +Test: [1190/2387] eta: 0:03:55 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1353 (0.3321) acc1: 99.4792 (85.7263) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1812 data: 0.0894 max mem: 3536 +Test: [1200/2387] eta: 0:03:53 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1317 (0.3305) acc1: 99.7396 (85.8419) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0888 max mem: 3536 +Test: [1210/2387] eta: 0:03:50 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1309 (0.3289) acc1: 100.0000 (85.9549) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0787 max mem: 3536 +Test: [1220/2387] eta: 0:03:48 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.0963 (0.3269) acc1: 100.0000 (86.0685) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0785 max mem: 3536 +Test: [1230/2387] eta: 0:03:46 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.0789 (0.3249) acc1: 100.0000 (86.1812) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1788 data: 0.0865 max mem: 3536 +Test: [1240/2387] eta: 0:03:44 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.0801 (0.3229) acc1: 100.0000 (86.2919) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1715 data: 0.0789 max mem: 3536 +Test: [1250/2387] eta: 0:03:42 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.0801 (0.3209) acc1: 100.0000 (86.4009) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1739 data: 0.0808 max mem: 3536 +Test: [1260/2387] eta: 0:03:40 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.0767 (0.3190) acc1: 100.0000 (86.5087) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1796 data: 0.0865 max mem: 3536 +Test: [1270/2387] eta: 0:03:37 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.0764 (0.3172) acc1: 100.0000 (86.6149) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1736 data: 0.0791 max mem: 3536 +Test: [1280/2387] eta: 0:03:35 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.0947 (0.3156) acc1: 100.0000 (86.7177) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1712 data: 0.0767 max mem: 3536 +Test: [1290/2387] eta: 0:03:33 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1171 (0.3142) acc1: 100.0000 (86.8049) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0836 max mem: 3536 +Test: [1300/2387] eta: 0:03:31 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1170 (0.3126) acc1: 100.0000 (86.9007) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1805 data: 0.0891 max mem: 3536 +Test: [1310/2387] eta: 0:03:29 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1145 (0.3112) acc1: 100.0000 (86.9929) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1844 data: 0.0931 max mem: 3536 +Test: [1320/2387] eta: 0:03:27 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1236 (0.3099) acc1: 99.7396 (87.0803) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1825 data: 0.0906 max mem: 3536 +Test: [1330/2387] eta: 0:03:25 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1273 (0.3087) acc1: 99.7396 (87.1672) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1793 data: 0.0873 max mem: 3536 +Test: [1340/2387] eta: 0:03:23 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1682 (0.3084) acc1: 98.6979 (87.2011) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1840 data: 0.0911 max mem: 3536 +Test: [1350/2387] eta: 0:03:21 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.0711 (0.3066) acc1: 100.0000 (87.2959) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1856 data: 0.0934 max mem: 3537 +Test: [1360/2387] eta: 0:03:19 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.0656 (0.3049) acc1: 100.0000 (87.3892) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1872 data: 0.0940 max mem: 3537 +Test: [1370/2387] eta: 0:03:17 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.0651 (0.3031) acc1: 100.0000 (87.4812) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1906 data: 0.0969 max mem: 3537 +Test: [1380/2387] eta: 0:03:15 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.0639 (0.3014) acc1: 100.0000 (87.5718) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2076 data: 0.1127 max mem: 3537 +Test: [1390/2387] eta: 0:03:13 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.0643 (0.2997) acc1: 100.0000 (87.6612) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1794 data: 0.0812 max mem: 3537 +Test: [1400/2387] eta: 0:03:11 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.0638 (0.2980) acc1: 100.0000 (87.7493) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1752 data: 0.0794 max mem: 3537 +Test: [1410/2387] eta: 0:03:09 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.0659 (0.2968) acc1: 100.0000 (87.8274) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2084 data: 0.1122 max mem: 3537 +Test: [1420/2387] eta: 0:03:07 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1392 (0.2959) acc1: 99.4792 (87.8911) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2052 data: 0.1100 max mem: 3537 +Test: [1430/2387] eta: 0:03:05 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1474 (0.2949) acc1: 99.4792 (87.9690) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1872 data: 0.0956 max mem: 3537 +Test: [1440/2387] eta: 0:03:04 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1474 (0.2940) acc1: 99.4792 (88.0440) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1947 data: 0.1031 max mem: 3537 +Test: [1450/2387] eta: 0:03:01 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1708 (0.2934) acc1: 99.2188 (88.1043) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1891 data: 0.0971 max mem: 3537 +Test: [1460/2387] eta: 0:02:59 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1148 (0.2922) acc1: 99.7396 (88.1843) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1673 data: 0.0742 max mem: 3537 +Test: [1470/2387] eta: 0:02:57 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1040 (0.2909) acc1: 100.0000 (88.2644) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1853 data: 0.0922 max mem: 3537 +Test: [1480/2387] eta: 0:02:55 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.0991 (0.2896) acc1: 100.0000 (88.3431) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1896 data: 0.0956 max mem: 3537 +Test: [1490/2387] eta: 0:02:53 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1013 (0.2883) acc1: 100.0000 (88.4203) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1905 data: 0.0957 max mem: 3537 +Test: [1500/2387] eta: 0:02:51 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1032 (0.2871) acc1: 100.0000 (88.4964) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1890 data: 0.0972 max mem: 3537 +Test: [1510/2387] eta: 0:02:50 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1222 (0.2862) acc1: 100.0000 (88.5684) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1948 data: 0.1047 max mem: 3537 +Test: [1520/2387] eta: 0:02:48 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1585 (0.2855) acc1: 99.2188 (88.6334) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1906 data: 0.0995 max mem: 3537 +Test: [1530/2387] eta: 0:02:46 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1853 (0.2850) acc1: 98.6979 (88.6932) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1847 data: 0.0916 max mem: 3537 +Test: [1540/2387] eta: 0:02:44 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1743 (0.2842) acc1: 98.9583 (88.7581) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1805 data: 0.0883 max mem: 3537 +Test: [1550/2387] eta: 0:02:42 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1700 (0.2835) acc1: 99.2188 (88.8249) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1747 data: 0.0837 max mem: 3537 +Test: [1560/2387] eta: 0:02:40 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1718 (0.2829) acc1: 99.2188 (88.8863) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1821 data: 0.0902 max mem: 3537 +Test: [1570/2387] eta: 0:02:37 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1688 (0.2822) acc1: 99.2188 (88.9493) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1681 data: 0.0762 max mem: 3537 +Test: [1580/2387] eta: 0:02:35 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1508 (0.2813) acc1: 99.4792 (89.0165) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1745 data: 0.0808 max mem: 3538 +Test: [1590/2387] eta: 0:02:33 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1312 (0.2803) acc1: 100.0000 (89.0849) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1894 data: 0.0942 max mem: 3538 +Test: [1600/2387] eta: 0:02:31 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1222 (0.2796) acc1: 100.0000 (89.1251) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1761 data: 0.0821 max mem: 3538 +Test: [1610/2387] eta: 0:02:29 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1180 (0.2786) acc1: 100.0000 (89.1918) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1694 data: 0.0770 max mem: 3538 +Test: [1620/2387] eta: 0:02:27 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1233 (0.2777) acc1: 100.0000 (89.2538) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1748 data: 0.0825 max mem: 3538 +Test: [1630/2387] eta: 0:02:25 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1386 (0.2770) acc1: 99.7396 (89.3084) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1903 data: 0.0980 max mem: 3538 +Test: [1640/2387] eta: 0:02:24 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1915 (0.2767) acc1: 98.1771 (89.3526) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1984 data: 0.1076 max mem: 3538 +Test: [1650/2387] eta: 0:02:22 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.2038 (0.2763) acc1: 97.3958 (89.3996) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1902 data: 0.0960 max mem: 3538 +Test: [1660/2387] eta: 0:02:20 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.2038 (0.2759) acc1: 97.6562 (89.4458) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1957 data: 0.1000 max mem: 3538 +Test: [1670/2387] eta: 0:02:18 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.2048 (0.2755) acc1: 97.9167 (89.4950) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1909 data: 0.0995 max mem: 3538 +Test: [1680/2387] eta: 0:02:16 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.2034 (0.2751) acc1: 97.6562 (89.5381) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1786 data: 0.0883 max mem: 3538 +Test: [1690/2387] eta: 0:02:14 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.2133 (0.2748) acc1: 97.6562 (89.5819) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1877 data: 0.0936 max mem: 3538 +Test: [1700/2387] eta: 0:02:12 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1923 (0.2742) acc1: 98.4375 (89.6314) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2003 data: 0.1064 max mem: 3538 +Test: [1710/2387] eta: 0:02:10 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1061 (0.2731) acc1: 99.7396 (89.6908) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1910 data: 0.1002 max mem: 3538 +Test: [1720/2387] eta: 0:02:08 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.0861 (0.2720) acc1: 100.0000 (89.7502) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1713 data: 0.0797 max mem: 3538 +Test: [1730/2387] eta: 0:02:06 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.0824 (0.2710) acc1: 100.0000 (89.8076) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1736 data: 0.0797 max mem: 3538 +Test: [1740/2387] eta: 0:02:04 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.0824 (0.2699) acc1: 100.0000 (89.8650) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2182 data: 0.1243 max mem: 3538 +Test: [1750/2387] eta: 0:02:02 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.0821 (0.2689) acc1: 100.0000 (89.9224) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2070 data: 0.1156 max mem: 3538 +Test: [1760/2387] eta: 0:02:00 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.0878 (0.2679) acc1: 100.0000 (89.9797) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1418 data: 0.0501 max mem: 3538 +Test: [1770/2387] eta: 0:01:58 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.0944 (0.2670) acc1: 100.0000 (90.0311) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1585 data: 0.0663 max mem: 3538 +Test: [1780/2387] eta: 0:01:56 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1009 (0.2660) acc1: 100.0000 (90.0871) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0985 max mem: 3538 +Test: [1790/2387] eta: 0:01:54 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.0957 (0.2651) acc1: 100.0000 (90.1412) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1818 data: 0.0890 max mem: 3538 +Test: [1800/2387] eta: 0:01:52 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.0897 (0.2641) acc1: 99.7396 (90.1945) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1777 data: 0.0838 max mem: 3538 +Test: [1810/2387] eta: 0:01:50 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.0903 (0.2631) acc1: 100.0000 (90.2481) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1782 data: 0.0867 max mem: 3539 +Test: [1820/2387] eta: 0:01:48 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.0904 (0.2622) acc1: 100.0000 (90.3015) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1672 data: 0.0761 max mem: 3539 +Test: [1830/2387] eta: 0:01:46 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.0967 (0.2614) acc1: 100.0000 (90.3543) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1722 data: 0.0799 max mem: 3539 +Test: [1840/2387] eta: 0:01:44 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1538 (0.2609) acc1: 100.0000 (90.4065) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1814 data: 0.0901 max mem: 3539 +Test: [1850/2387] eta: 0:01:42 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1563 (0.2603) acc1: 100.0000 (90.4580) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1716 data: 0.0812 max mem: 3539 +Test: [1860/2387] eta: 0:01:40 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1565 (0.2597) acc1: 100.0000 (90.5090) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1661 data: 0.0734 max mem: 3539 +Test: [1870/2387] eta: 0:01:38 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1559 (0.2592) acc1: 100.0000 (90.5590) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1722 data: 0.0792 max mem: 3539 +Test: [1880/2387] eta: 0:01:36 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1542 (0.2586) acc1: 100.0000 (90.6092) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1724 data: 0.0785 max mem: 3539 +Test: [1890/2387] eta: 0:01:35 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1543 (0.2581) acc1: 100.0000 (90.6586) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1742 data: 0.0787 max mem: 3539 +Test: [1900/2387] eta: 0:01:33 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1562 (0.2576) acc1: 100.0000 (90.7075) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1780 data: 0.0857 max mem: 3539 +Test: [1910/2387] eta: 0:01:31 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1654 (0.2574) acc1: 100.0000 (90.7527) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1807 data: 0.0886 max mem: 3539 +Test: [1920/2387] eta: 0:01:29 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.2134 (0.2569) acc1: 99.4792 (90.7890) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1759 data: 0.0825 max mem: 3539 +Test: [1930/2387] eta: 0:01:27 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1241 (0.2562) acc1: 100.0000 (90.8367) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1725 data: 0.0790 max mem: 3539 +Test: [1940/2387] eta: 0:01:25 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1241 (0.2558) acc1: 100.0000 (90.8488) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1737 data: 0.0811 max mem: 3539 +Test: [1950/2387] eta: 0:01:23 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1296 (0.2552) acc1: 99.7396 (90.8936) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1721 data: 0.0809 max mem: 3539 +Test: [1960/2387] eta: 0:01:21 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1299 (0.2546) acc1: 100.0000 (90.9336) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1757 data: 0.0847 max mem: 3539 +Test: [1970/2387] eta: 0:01:19 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1671 (0.2545) acc1: 99.7396 (90.9504) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1755 data: 0.0806 max mem: 3539 +Test: [1980/2387] eta: 0:01:17 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.2249 (0.2545) acc1: 95.8333 (90.9744) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1750 data: 0.0796 max mem: 3539 +Test: [1990/2387] eta: 0:01:15 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.2390 (0.2544) acc1: 95.3125 (90.9937) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1824 data: 0.0909 max mem: 3539 +Test: [2000/2387] eta: 0:01:13 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.2437 (0.2544) acc1: 94.7917 (91.0127) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1809 data: 0.0895 max mem: 3539 +Test: [2010/2387] eta: 0:01:11 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.2469 (0.2544) acc1: 94.5312 (91.0294) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1869 data: 0.0953 max mem: 3539 +Test: [2020/2387] eta: 0:01:09 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.2495 (0.2544) acc1: 94.2708 (91.0464) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1809 data: 0.0895 max mem: 3539 +Test: [2030/2387] eta: 0:01:07 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.2464 (0.2543) acc1: 94.0104 (91.0607) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1687 data: 0.0777 max mem: 3540 +Test: [2040/2387] eta: 0:01:05 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.2464 (0.2543) acc1: 94.0104 (91.0773) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1779 data: 0.0856 max mem: 3540 +Test: [2050/2387] eta: 0:01:04 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.2483 (0.2543) acc1: 94.0104 (91.0942) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1988 data: 0.1061 max mem: 3540 +Test: [2060/2387] eta: 0:01:02 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.2558 (0.2544) acc1: 94.0104 (91.1091) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1840 data: 0.0910 max mem: 3540 +Test: [2070/2387] eta: 0:01:00 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.2647 (0.2544) acc1: 94.2708 (91.1268) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1715 data: 0.0788 max mem: 3540 +Test: [2080/2387] eta: 0:00:58 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.2584 (0.2544) acc1: 94.7917 (91.1437) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.2000 data: 0.1082 max mem: 3540 +Test: [2090/2387] eta: 0:00:56 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.2606 (0.2544) acc1: 94.7917 (91.1615) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1904 data: 0.0971 max mem: 3540 +Test: [2100/2387] eta: 0:00:54 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.2630 (0.2545) acc1: 94.5312 (91.1778) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1825 data: 0.0900 max mem: 3540 +Test: [2110/2387] eta: 0:00:52 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.2598 (0.2545) acc1: 94.5312 (91.1954) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1736 data: 0.0824 max mem: 3540 +Test: [2120/2387] eta: 0:00:50 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.2500 (0.2543) acc1: 95.0521 (91.2223) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1666 data: 0.0746 max mem: 3540 +Test: [2130/2387] eta: 0:00:48 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1646 (0.2538) acc1: 99.2188 (91.2580) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1801 data: 0.0870 max mem: 3540 +Test: [2140/2387] eta: 0:00:46 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1636 (0.2537) acc1: 99.2188 (91.2712) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1761 data: 0.0841 max mem: 3540 +Test: [2150/2387] eta: 0:00:44 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1785 (0.2533) acc1: 98.4375 (91.3062) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1765 data: 0.0856 max mem: 3540 +Test: [2160/2387] eta: 0:00:43 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1573 (0.2529) acc1: 99.2188 (91.3391) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1810 data: 0.0890 max mem: 3540 +Test: [2170/2387] eta: 0:00:41 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1832 (0.2529) acc1: 97.9167 (91.3547) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1830 data: 0.0906 max mem: 3540 +Test: [2180/2387] eta: 0:00:39 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.2628 (0.2537) acc1: 92.4479 (91.3138) acc5: 100.0000 (100.0000) acc1_neg: 65.104 acc5_neg: 100.000 time: 0.1846 data: 0.0923 max mem: 3540 +Test: [2190/2387] eta: 0:00:37 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.2511 (0.2535) acc1: 95.5729 (91.3196) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1766 data: 0.0841 max mem: 3540 +Test: [2200/2387] eta: 0:00:35 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1279 (0.2529) acc1: 100.0000 (91.3588) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1688 data: 0.0767 max mem: 3540 +Test: [2210/2387] eta: 0:00:33 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1203 (0.2523) acc1: 100.0000 (91.3977) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1680 data: 0.0759 max mem: 3540 +Test: [2220/2387] eta: 0:00:31 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1215 (0.2518) acc1: 100.0000 (91.4336) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1675 data: 0.0757 max mem: 3540 +Test: [2230/2387] eta: 0:00:29 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1203 (0.2512) acc1: 100.0000 (91.4702) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1632 data: 0.0716 max mem: 3540 +Test: [2240/2387] eta: 0:00:27 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1098 (0.2506) acc1: 99.7396 (91.5068) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1605 data: 0.0680 max mem: 3540 +Test: [2250/2387] eta: 0:00:25 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1287 (0.2502) acc1: 99.7396 (91.5384) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1612 data: 0.0682 max mem: 3540 +Test: [2260/2387] eta: 0:00:23 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1524 (0.2498) acc1: 99.4792 (91.5737) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1662 data: 0.0734 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1489 (0.2493) acc1: 99.7396 (91.6094) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1653 data: 0.0731 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1462 (0.2490) acc1: 100.0000 (91.6283) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1560 data: 0.0646 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1337 (0.2485) acc1: 100.0000 (91.6636) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1593 data: 0.0685 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1402 (0.2481) acc1: 100.0000 (91.6955) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1630 data: 0.0727 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1564 (0.2478) acc1: 99.7396 (91.7219) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1539 data: 0.0632 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1957 (0.2477) acc1: 98.6979 (91.7443) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1551 data: 0.0630 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1945 (0.2474) acc1: 98.4375 (91.7718) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1615 data: 0.0696 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1627 (0.2471) acc1: 99.2188 (91.7997) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1433 data: 0.0528 max mem: 3541 +Test: [2350/2387] eta: 0:00:06 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1435 (0.2470) acc1: 99.4792 (91.8126) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1363 data: 0.0460 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1724 (0.2468) acc1: 99.7396 (91.8340) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1410 data: 0.0510 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.1834 (0.2465) acc1: 99.4792 (91.8603) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1365 data: 0.0469 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.2229 (0.2467) acc1: 97.9167 (91.8673) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1396 data: 0.0453 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 41.096 acc5_pos: 100.000 loss: 0.2383 (0.2465) acc1: 96.0938 (91.8762) acc5: 100.0000 (100.0000) acc1_neg: 95.349 acc5_neg: 100.000 time: 0.1312 data: 0.0395 max mem: 3541 +Test: Total time: 0:07:25 (0.1868 s / it) +* Overall Acc@1 91.876 Acc@5 100.000 + target == real Acc@1 26.484 Acc@5 100.000 + target == fake Acc@1 96.468 Acc@5 100.000 + loss 0.247 + PR-AUC 0.9906 ROC-AUC 0.8825 EER 19.28% (thr=0.7939) +Accuracy of the network on the 916267 test images: 91.9% +[rank0]:[W726 05:39:42.396537102 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:40:03.212000 3167869 torch/distributed/run.py:766] +W0726 05:40:03.212000 3167869 torch/distributed/run.py:766] ***************************************** +W0726 05:40:03.212000 3167869 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:40:03.212000 3167869 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:40:32.032320451 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:40:32.214723481 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:40:33.910272198 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:40:33.987565528 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:40:33.043912941 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:40:33.050279500 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:40:33.133322409 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:40:33.146279500 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n32_seed8_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed8_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 0:51:27 acc1_pos: 24.479 acc5_pos: 100.000 loss: 1.1024 (1.1024) acc1: 24.4792 (24.4792) acc5: 100.0000 (100.0000) time: 5.9723 data: 3.8523 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/517] eta: 0:05:16 acc1_pos: 1.562 acc5_pos: 100.000 loss: 1.3249 (1.2779) acc1: 15.1042 (15.7197) acc5: 100.0000 (100.0000) time: 0.6245 data: 0.3505 max mem: 3531 +Test: [ 20/517] eta: 0:03:07 acc1_pos: 7.031 acc5_pos: 100.000 loss: 1.3283 (1.2913) acc1: 12.5000 (15.6994) acc5: 100.0000 (100.0000) time: 0.0972 data: 0.0064 max mem: 3531 +Test: [ 30/517] eta: 0:02:29 acc1_pos: 4.427 acc5_pos: 100.000 loss: 1.3573 (1.3204) acc1: 11.7188 (14.8017) acc5: 100.0000 (100.0000) time: 0.1327 data: 0.0403 max mem: 3531 +Test: [ 40/517] eta: 0:02:28 acc1_pos: 7.031 acc5_pos: 100.000 loss: 1.3803 (1.3443) acc1: 7.0312 (12.6588) acc5: 100.0000 (100.0000) time: 0.2410 data: 0.1499 max mem: 3531 +Test: [ 50/517] eta: 0:02:33 acc1_pos: 32.110 acc5_pos: 100.000 loss: 1.3254 (1.2829) acc1: 8.3333 (16.8709) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.3645 data: 0.2732 max mem: 3531 +Test: [ 60/517] eta: 0:02:20 acc1_pos: 32.110 acc5_pos: 100.000 loss: 0.3750 (1.1236) acc1: 87.2396 (29.5338) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2990 data: 0.2048 max mem: 3531 +Test: [ 70/517] eta: 0:02:12 acc1_pos: 32.110 acc5_pos: 100.000 loss: 0.2603 (1.0005) acc1: 98.1771 (39.2532) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2099 data: 0.1144 max mem: 3531 +Test: [ 80/517] eta: 0:02:05 acc1_pos: 32.110 acc5_pos: 100.000 loss: 0.2554 (0.9151) acc1: 97.1354 (46.1002) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.2258 data: 0.1302 max mem: 3531 +Test: [ 90/517] eta: 0:02:01 acc1_pos: 32.110 acc5_pos: 100.000 loss: 0.2555 (0.8421) acc1: 98.1771 (51.8000) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2399 data: 0.1452 max mem: 3531 +Test: [100/517] eta: 0:01:56 acc1_pos: 32.110 acc5_pos: 100.000 loss: 0.2324 (0.7811) acc1: 98.6979 (56.4692) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2437 data: 0.1483 max mem: 3531 +Test: [110/517] eta: 0:01:51 acc1_pos: 32.110 acc5_pos: 100.000 loss: 0.1993 (0.7283) acc1: 100.0000 (60.3838) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2299 data: 0.1331 max mem: 3531 +Test: [120/517] eta: 0:01:47 acc1_pos: 32.110 acc5_pos: 100.000 loss: 0.2356 (0.6897) acc1: 99.7396 (63.5718) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2349 data: 0.1356 max mem: 3531 +Test: [130/517] eta: 0:01:43 acc1_pos: 32.110 acc5_pos: 100.000 loss: 0.2620 (0.6577) acc1: 98.9583 (66.1955) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2355 data: 0.1363 max mem: 3531 +Test: [140/517] eta: 0:01:40 acc1_pos: 32.110 acc5_pos: 100.000 loss: 0.2000 (0.6242) acc1: 100.0000 (68.5930) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2352 data: 0.1400 max mem: 3531 +Test: [150/517] eta: 0:01:37 acc1_pos: 32.110 acc5_pos: 100.000 loss: 0.1912 (0.5962) acc1: 100.0000 (70.6729) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2493 data: 0.1539 max mem: 3531 +Test: [160/517] eta: 0:01:34 acc1_pos: 32.110 acc5_pos: 100.000 loss: 0.1990 (0.5718) acc1: 100.0000 (72.4848) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2496 data: 0.1516 max mem: 3531 +Test: [170/517] eta: 0:01:30 acc1_pos: 32.110 acc5_pos: 100.000 loss: 0.2005 (0.5504) acc1: 99.7396 (74.0771) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2358 data: 0.1383 max mem: 3531 +Test: [180/517] eta: 0:01:27 acc1_pos: 32.110 acc5_pos: 100.000 loss: 0.1942 (0.5305) acc1: 99.7396 (75.4892) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2304 data: 0.1358 max mem: 3531 +Test: [190/517] eta: 0:01:24 acc1_pos: 32.110 acc5_pos: 100.000 loss: 0.2013 (0.5136) acc1: 99.7396 (76.7643) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2243 data: 0.1302 max mem: 3531 +Test: [200/517] eta: 0:01:21 acc1_pos: 32.110 acc5_pos: 100.000 loss: 0.1997 (0.4972) acc1: 100.0000 (77.9112) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2344 data: 0.1393 max mem: 3531 +Test: [210/517] eta: 0:01:19 acc1_pos: 32.110 acc5_pos: 100.000 loss: 0.2431 (0.4872) acc1: 97.6562 (78.7187) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2496 data: 0.1527 max mem: 3532 +Test: [220/517] eta: 0:01:15 acc1_pos: 32.110 acc5_pos: 100.000 loss: 0.2966 (0.4793) acc1: 94.2708 (79.3316) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.2354 data: 0.1385 max mem: 3532 +Test: [230/517] eta: 0:01:13 acc1_pos: 32.110 acc5_pos: 100.000 loss: 0.2828 (0.4695) acc1: 94.7917 (80.1553) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2349 data: 0.1395 max mem: 3532 +Test: [240/517] eta: 0:01:10 acc1_pos: 32.110 acc5_pos: 100.000 loss: 0.2406 (0.4604) acc1: 99.7396 (80.9637) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2394 data: 0.1434 max mem: 3532 +Test: [250/517] eta: 0:01:07 acc1_pos: 32.110 acc5_pos: 100.000 loss: 0.2326 (0.4510) acc1: 99.7396 (81.6816) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2299 data: 0.1338 max mem: 3532 +Test: [260/517] eta: 0:01:04 acc1_pos: 32.110 acc5_pos: 100.000 loss: 0.2102 (0.4423) acc1: 99.7396 (82.3745) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2351 data: 0.1395 max mem: 3532 +Test: [270/517] eta: 0:01:02 acc1_pos: 32.110 acc5_pos: 100.000 loss: 0.2519 (0.4358) acc1: 99.7396 (83.0114) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2358 data: 0.1404 max mem: 3532 +Test: [280/517] eta: 0:00:59 acc1_pos: 32.110 acc5_pos: 100.000 loss: 0.2584 (0.4299) acc1: 99.4792 (83.5363) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2306 data: 0.1360 max mem: 3532 +Test: [290/517] eta: 0:00:56 acc1_pos: 32.110 acc5_pos: 100.000 loss: 0.2478 (0.4231) acc1: 99.2188 (84.0680) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2340 data: 0.1382 max mem: 3532 +Test: [300/517] eta: 0:00:54 acc1_pos: 32.110 acc5_pos: 100.000 loss: 0.2151 (0.4162) acc1: 99.7396 (84.5861) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2342 data: 0.1375 max mem: 3532 +Test: [310/517] eta: 0:00:51 acc1_pos: 32.110 acc5_pos: 100.000 loss: 0.2133 (0.4097) acc1: 100.0000 (85.0792) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2349 data: 0.1391 max mem: 3532 +Test: [320/517] eta: 0:00:49 acc1_pos: 32.110 acc5_pos: 100.000 loss: 0.2158 (0.4039) acc1: 100.0000 (85.5416) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2407 data: 0.1449 max mem: 3532 +Test: [330/517] eta: 0:00:46 acc1_pos: 32.110 acc5_pos: 100.000 loss: 0.2111 (0.3983) acc1: 100.0000 (85.9713) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2459 data: 0.1532 max mem: 3532 +Test: [340/517] eta: 0:00:44 acc1_pos: 32.110 acc5_pos: 100.000 loss: 0.2048 (0.3930) acc1: 100.0000 (86.3438) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2389 data: 0.1485 max mem: 3532 +Test: [350/517] eta: 0:00:41 acc1_pos: 32.110 acc5_pos: 100.000 loss: 0.2571 (0.3899) acc1: 98.1771 (86.6364) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.2339 data: 0.1408 max mem: 3532 +Test: [360/517] eta: 0:00:38 acc1_pos: 32.110 acc5_pos: 100.000 loss: 0.2748 (0.3867) acc1: 97.1354 (86.8883) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2349 data: 0.1398 max mem: 3532 +Test: [370/517] eta: 0:00:36 acc1_pos: 32.110 acc5_pos: 100.000 loss: 0.2032 (0.3809) acc1: 100.0000 (87.2410) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2349 data: 0.1405 max mem: 3532 +Test: [380/517] eta: 0:00:33 acc1_pos: 32.110 acc5_pos: 100.000 loss: 0.1743 (0.3759) acc1: 100.0000 (87.5731) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2355 data: 0.1427 max mem: 3532 +Test: [390/517] eta: 0:00:31 acc1_pos: 32.110 acc5_pos: 100.000 loss: 0.2007 (0.3718) acc1: 100.0000 (87.8790) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2403 data: 0.1491 max mem: 3532 +Test: [400/517] eta: 0:00:28 acc1_pos: 32.110 acc5_pos: 100.000 loss: 0.1876 (0.3670) acc1: 100.0000 (88.1780) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2452 data: 0.1528 max mem: 3532 +Test: [410/517] eta: 0:00:26 acc1_pos: 32.110 acc5_pos: 100.000 loss: 0.1876 (0.3634) acc1: 100.0000 (88.4644) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2515 data: 0.1582 max mem: 3532 +Test: [420/517] eta: 0:00:24 acc1_pos: 32.110 acc5_pos: 100.000 loss: 0.2346 (0.3605) acc1: 100.0000 (88.7378) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2497 data: 0.1567 max mem: 3532 +Test: [430/517] eta: 0:00:21 acc1_pos: 32.110 acc5_pos: 100.000 loss: 0.2346 (0.3580) acc1: 100.0000 (88.9852) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2328 data: 0.1393 max mem: 3532 +Test: [440/517] eta: 0:00:18 acc1_pos: 32.110 acc5_pos: 100.000 loss: 0.2538 (0.3554) acc1: 99.7396 (89.2172) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2301 data: 0.1366 max mem: 3533 +Test: [450/517] eta: 0:00:16 acc1_pos: 32.110 acc5_pos: 100.000 loss: 0.2196 (0.3523) acc1: 100.0000 (89.4563) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2355 data: 0.1429 max mem: 3533 +Test: [460/517] eta: 0:00:14 acc1_pos: 32.110 acc5_pos: 100.000 loss: 0.2088 (0.3492) acc1: 100.0000 (89.6771) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2395 data: 0.1463 max mem: 3533 +Test: [470/517] eta: 0:00:11 acc1_pos: 32.110 acc5_pos: 100.000 loss: 0.1926 (0.3457) acc1: 100.0000 (89.8946) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2443 data: 0.1509 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 32.110 acc5_pos: 100.000 loss: 0.1836 (0.3428) acc1: 100.0000 (90.1042) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2445 data: 0.1495 max mem: 3533 +Test: [490/517] eta: 0:00:06 acc1_pos: 32.110 acc5_pos: 100.000 loss: 0.2146 (0.3401) acc1: 100.0000 (90.2972) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2499 data: 0.1520 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 32.110 acc5_pos: 100.000 loss: 0.2230 (0.3382) acc1: 100.0000 (90.4821) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2561 data: 0.1607 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 32.110 acc5_pos: 100.000 loss: 0.2557 (0.3368) acc1: 99.7396 (90.6337) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2401 data: 0.1488 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 32.110 acc5_pos: 100.000 loss: 0.2222 (0.3354) acc1: 100.0000 (90.7377) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2050 data: 0.1125 max mem: 3533 +Test: Total time: 0:02:06 (0.2454 s / it) +* Overall Acc@1 90.738 Acc@5 100.000 + target == real Acc@1 12.820 Acc@5 100.000 + target == fake Acc@1 98.821 Acc@5 100.000 + loss 0.335 + PR-AUC 0.9463 ROC-AUC 0.7016 EER 35.20% (thr=0.7925) +Accuracy of the network on the 198427 test images: 90.7% +[rank0]:[W726 05:42:51.971405920 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 05:43:12.112000 3168442 torch/distributed/run.py:766] +W0726 05:43:12.112000 3168442 torch/distributed/run.py:766] ***************************************** +W0726 05:43:12.112000 3168442 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:43:12.112000 3168442 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 05:43:41.949144802 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:43:41.221489941 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:43:41.324456649 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:43:41.473274467 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +| distributed init (rank 4): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:43:42.637757029 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:43:42.671829595 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W726 05:43:42.671873722 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank4]:[W726 05:43:42.674645026 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n32_seed8_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed8_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:35:39 acc1_pos: 68.229 acc5_pos: 100.000 loss: 0.5814 (0.5814) acc1: 68.2292 (68.2292) acc5: 100.0000 (100.0000) time: 5.4995 data: 4.0046 max mem: 3531 +Test: [ 10/389] eta: 0:03:40 acc1_pos: 98.958 acc5_pos: 100.000 loss: 0.2998 (0.4167) acc1: 90.6250 (82.6231) acc5: 100.0000 (100.0000) time: 0.5818 data: 0.3645 max mem: 3531 +Test: [ 20/389] eta: 0:02:17 acc1_pos: 91.146 acc5_pos: 100.000 loss: 0.3406 (0.3965) acc1: 91.4062 (87.6488) acc5: 100.0000 (100.0000) time: 0.1159 data: 0.0221 max mem: 3531 +Test: [ 30/389] eta: 0:01:46 acc1_pos: 93.490 acc5_pos: 100.000 loss: 0.3449 (0.3795) acc1: 94.5312 (89.9530) acc5: 100.0000 (100.0000) time: 0.1377 data: 0.0388 max mem: 3531 +Test: [ 40/389] eta: 0:01:35 acc1_pos: 67.188 acc5_pos: 100.000 loss: 0.3556 (0.3955) acc1: 93.4896 (88.7703) acc5: 100.0000 (100.0000) time: 0.1699 data: 0.0735 max mem: 3531 +Test: [ 50/389] eta: 0:01:25 acc1_pos: 96.615 acc5_pos: 100.000 loss: 0.3502 (0.3788) acc1: 93.4896 (89.8438) acc5: 100.0000 (100.0000) time: 0.1837 data: 0.0918 max mem: 3531 +Test: [ 60/389] eta: 0:01:18 acc1_pos: 87.240 acc5_pos: 100.000 loss: 0.2707 (0.3627) acc1: 96.8750 (90.9452) acc5: 100.0000 (100.0000) time: 0.1631 data: 0.0727 max mem: 3531 +Test: [ 70/389] eta: 0:01:11 acc1_pos: 91.406 acc5_pos: 100.000 loss: 0.2853 (0.3656) acc1: 94.7917 (90.5810) acc5: 100.0000 (100.0000) time: 0.1596 data: 0.0693 max mem: 3531 +Test: [ 80/389] eta: 0:01:07 acc1_pos: 97.656 acc5_pos: 100.000 loss: 0.3925 (0.3742) acc1: 89.3229 (89.9306) acc5: 100.0000 (100.0000) time: 0.1671 data: 0.0754 max mem: 3531 +Test: [ 90/389] eta: 0:01:05 acc1_pos: 75.521 acc5_pos: 100.000 loss: 0.2322 (0.3656) acc1: 87.2396 (90.1471) acc5: 100.0000 (100.0000) time: 0.1946 data: 0.1019 max mem: 3531 +Test: [100/389] eta: 0:01:01 acc1_pos: 79.167 acc5_pos: 100.000 loss: 0.4000 (0.3790) acc1: 89.5833 (89.1192) acc5: 100.0000 (100.0000) time: 0.1846 data: 0.0888 max mem: 3531 +Test: [110/389] eta: 0:00:59 acc1_pos: 96.094 acc5_pos: 100.000 loss: 0.4883 (0.4010) acc1: 77.6042 (86.6484) acc5: 100.0000 (100.0000) time: 0.1799 data: 0.0809 max mem: 3531 +Test: [120/389] eta: 0:00:56 acc1_pos: 88.021 acc5_pos: 100.000 loss: 0.4468 (0.3998) acc1: 81.5104 (86.7596) acc5: 100.0000 (100.0000) time: 0.1970 data: 0.1000 max mem: 3531 +Test: [130/389] eta: 0:00:52 acc1_pos: 85.276 acc5_pos: 100.000 loss: 0.4470 (0.4334) acc1: 81.5104 (83.9396) acc5: 100.0000 (100.0000) acc1_neg: 53.906 acc5_neg: 100.000 time: 0.1608 data: 0.0674 max mem: 3531 +Test: [140/389] eta: 0:00:50 acc1_pos: 85.276 acc5_pos: 100.000 loss: 0.7888 (0.4711) acc1: 43.2292 (80.2841) acc5: 100.0000 (100.0000) acc1_neg: 0.781 acc5_neg: 100.000 time: 0.1443 data: 0.0510 max mem: 3531 +Test: [150/389] eta: 0:00:47 acc1_pos: 85.276 acc5_pos: 100.000 loss: 0.9411 (0.5123) acc1: 23.4375 (76.9333) acc5: 100.0000 (100.0000) acc1_neg: 76.302 acc5_neg: 100.000 time: 0.1694 data: 0.0764 max mem: 3531 +Test: [160/389] eta: 0:00:44 acc1_pos: 85.276 acc5_pos: 100.000 loss: 0.9504 (0.5287) acc1: 23.4375 (75.3397) acc5: 100.0000 (100.0000) acc1_neg: 64.323 acc5_neg: 100.000 time: 0.1645 data: 0.0730 max mem: 3531 +Test: [170/389] eta: 0:00:42 acc1_pos: 85.276 acc5_pos: 100.000 loss: 0.6304 (0.5403) acc1: 62.7604 (74.1655) acc5: 100.0000 (100.0000) acc1_neg: 28.125 acc5_neg: 100.000 time: 0.1594 data: 0.0659 max mem: 3531 +Test: [180/389] eta: 0:00:40 acc1_pos: 85.276 acc5_pos: 100.000 loss: 0.8186 (0.5671) acc1: 36.7188 (71.3671) acc5: 100.0000 (100.0000) acc1_neg: 26.042 acc5_neg: 100.000 time: 0.1723 data: 0.0790 max mem: 3531 +Test: [190/389] eta: 0:00:38 acc1_pos: 85.276 acc5_pos: 100.000 loss: 0.9250 (0.5889) acc1: 26.0417 (69.3472) acc5: 100.0000 (100.0000) acc1_neg: 0.260 acc5_neg: 100.000 time: 0.1718 data: 0.0813 max mem: 3531 +Test: [200/389] eta: 0:00:36 acc1_pos: 85.276 acc5_pos: 100.000 loss: 1.1099 (0.6129) acc1: 19.5312 (67.1344) acc5: 100.0000 (100.0000) acc1_neg: 6.510 acc5_neg: 100.000 time: 0.1674 data: 0.0770 max mem: 3531 +Test: [210/389] eta: 0:00:33 acc1_pos: 85.276 acc5_pos: 100.000 loss: 0.9499 (0.6196) acc1: 20.8333 (66.6580) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1662 data: 0.0751 max mem: 3532 +Test: [220/389] eta: 0:00:31 acc1_pos: 85.276 acc5_pos: 100.000 loss: 0.4994 (0.6161) acc1: 67.1875 (67.1298) acc5: 100.0000 (100.0000) acc1_neg: 49.219 acc5_neg: 100.000 time: 0.1627 data: 0.0714 max mem: 3532 +Test: [230/389] eta: 0:00:29 acc1_pos: 85.276 acc5_pos: 100.000 loss: 0.7326 (0.6301) acc1: 49.2188 (65.8922) acc5: 100.0000 (100.0000) acc1_neg: 52.344 acc5_neg: 100.000 time: 0.1692 data: 0.0766 max mem: 3532 +Test: [240/389] eta: 0:00:27 acc1_pos: 85.276 acc5_pos: 100.000 loss: 0.9412 (0.6416) acc1: 28.1250 (64.5077) acc5: 100.0000 (100.0000) acc1_neg: 61.979 acc5_neg: 100.000 time: 0.1724 data: 0.0799 max mem: 3532 +Test: [250/389] eta: 0:00:25 acc1_pos: 85.276 acc5_pos: 100.000 loss: 0.6616 (0.6399) acc1: 61.1979 (64.7483) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.1685 data: 0.0776 max mem: 3532 +Test: [260/389] eta: 0:00:24 acc1_pos: 85.276 acc5_pos: 100.000 loss: 0.6522 (0.6505) acc1: 62.7604 (63.5806) acc5: 100.0000 (100.0000) acc1_neg: 2.344 acc5_neg: 100.000 time: 0.1739 data: 0.0834 max mem: 3532 +Test: [270/389] eta: 0:00:22 acc1_pos: 85.276 acc5_pos: 100.000 loss: 1.0417 (0.6740) acc1: 17.1875 (61.6765) acc5: 100.0000 (100.0000) acc1_neg: 11.979 acc5_neg: 100.000 time: 0.1873 data: 0.0962 max mem: 3532 +Test: [280/389] eta: 0:00:20 acc1_pos: 85.276 acc5_pos: 100.000 loss: 1.1703 (0.6929) acc1: 9.8958 (60.2341) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1869 data: 0.0952 max mem: 3532 +Test: [290/389] eta: 0:00:18 acc1_pos: 85.276 acc5_pos: 100.000 loss: 0.9870 (0.6960) acc1: 16.4062 (59.8663) acc5: 100.0000 (100.0000) acc1_neg: 68.229 acc5_neg: 100.000 time: 0.1793 data: 0.0886 max mem: 3532 +Test: [300/389] eta: 0:00:16 acc1_pos: 85.276 acc5_pos: 100.000 loss: 0.8709 (0.7027) acc1: 30.7292 (58.8931) acc5: 100.0000 (100.0000) acc1_neg: 19.271 acc5_neg: 100.000 time: 0.1830 data: 0.0921 max mem: 3532 +Test: [310/389] eta: 0:00:14 acc1_pos: 85.276 acc5_pos: 100.000 loss: 0.9250 (0.7162) acc1: 20.5729 (57.6794) acc5: 100.0000 (100.0000) acc1_neg: 3.125 acc5_neg: 100.000 time: 0.1854 data: 0.0939 max mem: 3532 +Test: [320/389] eta: 0:00:12 acc1_pos: 85.276 acc5_pos: 100.000 loss: 0.9816 (0.7215) acc1: 20.5729 (57.2057) acc5: 100.0000 (100.0000) acc1_neg: 68.490 acc5_neg: 100.000 time: 0.1850 data: 0.0928 max mem: 3532 +Test: [330/389] eta: 0:00:10 acc1_pos: 85.276 acc5_pos: 100.000 loss: 1.1319 (0.7340) acc1: 5.2083 (55.7213) acc5: 100.0000 (100.0000) acc1_neg: 3.906 acc5_neg: 100.000 time: 0.1822 data: 0.0900 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 85.276 acc5_pos: 100.000 loss: 0.7055 (0.7287) acc1: 32.8125 (56.3768) acc5: 100.0000 (100.0000) acc1_neg: 60.417 acc5_neg: 100.000 time: 0.1695 data: 0.0783 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 85.276 acc5_pos: 100.000 loss: 0.6370 (0.7454) acc1: 53.6458 (55.5993) acc5: 100.0000 (100.0000) acc1_neg: 0.000 acc5_neg: 100.000 time: 0.1743 data: 0.0834 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 85.276 acc5_pos: 100.000 loss: 1.4708 (0.7673) acc1: 1.8229 (54.1017) acc5: 100.0000 (100.0000) acc1_neg: 2.604 acc5_neg: 100.000 time: 0.1875 data: 0.0971 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 85.276 acc5_pos: 100.000 loss: 1.3646 (0.7701) acc1: 2.6042 (53.8122) acc5: 100.0000 (100.0000) acc1_neg: 57.031 acc5_neg: 100.000 time: 0.1507 data: 0.0603 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 85.276 acc5_pos: 100.000 loss: 0.8437 (0.7739) acc1: 36.4583 (53.3095) acc5: 100.0000 (100.0000) acc1_neg: 12.240 acc5_neg: 100.000 time: 0.1104 data: 0.0199 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 85.276 acc5_pos: 100.000 loss: 0.7766 (0.7754) acc1: 49.4792 (53.2258) acc5: 100.0000 (100.0000) acc1_neg: 53.552 acc5_neg: 100.000 time: 0.1128 data: 0.0172 max mem: 3532 +Test: Total time: 0:01:10 (0.1806 s / it) +* Overall Acc@1 53.226 Acc@5 100.000 + target == real Acc@1 86.731 Acc@5 100.000 + target == fake Acc@1 37.652 Acc@5 100.000 + loss 0.775 + PR-AUC 0.8138 ROC-AUC 0.7012 EER 35.02% (thr=0.3286) +Accuracy of the network on the 149358 test images: 53.2% +[rank0]:[W726 05:45:03.342161290 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:45:22.582000 3169002 torch/distributed/run.py:766] +W0726 05:45:22.582000 3169002 torch/distributed/run.py:766] ***************************************** +W0726 05:45:22.582000 3169002 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:45:22.582000 3169002 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 05:45:51.871716082 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:45:51.258371227 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:45:52.669331198 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:45:52.784313222 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:45:52.838687331 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:45:52.853259653 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:45:52.863792741 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:45:52.886092114 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n32_seed8_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed8_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:34:59 acc1_pos: 79.688 acc5_pos: 100.000 loss: 0.4367 (0.4367) acc1: 79.6875 (79.6875) acc5: 100.0000 (100.0000) time: 6.3056 data: 4.6437 max mem: 3531 +Test: [ 10/333] eta: 0:03:32 acc1_pos: 82.292 acc5_pos: 100.000 loss: 0.3726 (0.3570) acc1: 82.2917 (83.2623) acc5: 100.0000 (100.0000) time: 0.6585 data: 0.4248 max mem: 3531 +Test: [ 20/333] eta: 0:02:02 acc1_pos: 93.229 acc5_pos: 100.000 loss: 0.3713 (0.3691) acc1: 82.2917 (82.9489) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0017 max mem: 3531 +Test: [ 30/333] eta: 0:01:42 acc1_pos: 55.990 acc5_pos: 100.000 loss: 0.4659 (0.5228) acc1: 71.0938 (71.8162) acc5: 100.0000 (100.0000) time: 0.1645 data: 0.0679 max mem: 3531 +Test: [ 40/333] eta: 0:01:30 acc1_pos: 52.083 acc5_pos: 100.000 loss: 0.8771 (0.6083) acc1: 43.7500 (65.2121) acc5: 100.0000 (100.0000) time: 0.2203 data: 0.1263 max mem: 3531 +Test: [ 50/333] eta: 0:01:22 acc1_pos: 49.219 acc5_pos: 100.000 loss: 0.8771 (0.6619) acc1: 46.0938 (61.3000) acc5: 100.0000 (100.0000) time: 0.2143 data: 0.1216 max mem: 3531 +Test: [ 60/333] eta: 0:01:16 acc1_pos: 39.844 acc5_pos: 100.000 loss: 0.8743 (0.6980) acc1: 45.0521 (58.4016) acc5: 100.0000 (100.0000) time: 0.2245 data: 0.1288 max mem: 3531 +Test: [ 70/333] eta: 0:01:08 acc1_pos: 44.792 acc5_pos: 100.000 loss: 0.8743 (0.7218) acc1: 44.0104 (56.8112) acc5: 100.0000 (100.0000) time: 0.1862 data: 0.0892 max mem: 3531 +Test: [ 80/333] eta: 0:01:04 acc1_pos: 45.573 acc5_pos: 100.000 loss: 0.8522 (0.7395) acc1: 44.7917 (55.1826) acc5: 100.0000 (100.0000) time: 0.1751 data: 0.0814 max mem: 3531 +Test: [ 90/333] eta: 0:01:00 acc1_pos: 53.125 acc5_pos: 100.000 loss: 0.8034 (0.7443) acc1: 46.6146 (54.6160) acc5: 100.0000 (100.0000) time: 0.2036 data: 0.1099 max mem: 3531 +Test: [100/333] eta: 0:00:56 acc1_pos: 52.604 acc5_pos: 100.000 loss: 0.7633 (0.7498) acc1: 50.7812 (54.3394) acc5: 100.0000 (100.0000) time: 0.1899 data: 0.0937 max mem: 3531 +Test: [110/333] eta: 0:00:52 acc1_pos: 59.115 acc5_pos: 100.000 loss: 0.7947 (0.7571) acc1: 50.7812 (53.8523) acc5: 100.0000 (100.0000) time: 0.1749 data: 0.0779 max mem: 3531 +Test: [120/333] eta: 0:00:49 acc1_pos: 51.823 acc5_pos: 100.000 loss: 0.7895 (0.7596) acc1: 46.6146 (53.5942) acc5: 100.0000 (100.0000) time: 0.1821 data: 0.0870 max mem: 3531 +Test: [130/333] eta: 0:00:46 acc1_pos: 68.490 acc5_pos: 100.000 loss: 0.7263 (0.7596) acc1: 51.8229 (53.6836) acc5: 100.0000 (100.0000) time: 0.2010 data: 0.1088 max mem: 3531 +Test: [140/333] eta: 0:00:44 acc1_pos: 79.167 acc5_pos: 100.000 loss: 0.7247 (0.7583) acc1: 54.4271 (53.7807) acc5: 100.0000 (100.0000) time: 0.2193 data: 0.1280 max mem: 3531 +Test: [150/333] eta: 0:00:41 acc1_pos: 53.906 acc5_pos: 100.000 loss: 0.7483 (0.7609) acc1: 53.6458 (53.5527) acc5: 100.0000 (100.0000) time: 0.2138 data: 0.1209 max mem: 3531 +Test: [160/333] eta: 0:00:39 acc1_pos: 43.421 acc5_pos: 100.000 loss: 0.7711 (0.7620) acc1: 52.3438 (53.4792) acc5: 100.0000 (100.0000) acc1_neg: 56.250 acc5_neg: 100.000 time: 0.1982 data: 0.1024 max mem: 3531 +Test: [170/333] eta: 0:00:36 acc1_pos: 43.421 acc5_pos: 100.000 loss: 0.7218 (0.7574) acc1: 57.0312 (53.9184) acc5: 100.0000 (100.0000) acc1_neg: 63.802 acc5_neg: 100.000 time: 0.1849 data: 0.0880 max mem: 3531 +Test: [180/333] eta: 0:00:34 acc1_pos: 43.421 acc5_pos: 100.000 loss: 0.6420 (0.7504) acc1: 63.0208 (54.4803) acc5: 100.0000 (100.0000) acc1_neg: 69.531 acc5_neg: 100.000 time: 0.2049 data: 0.1087 max mem: 3531 +Test: [190/333] eta: 0:00:31 acc1_pos: 43.421 acc5_pos: 100.000 loss: 0.5720 (0.7405) acc1: 64.5833 (55.4019) acc5: 100.0000 (100.0000) acc1_neg: 68.750 acc5_neg: 100.000 time: 0.2089 data: 0.1132 max mem: 3531 +Test: [200/333] eta: 0:00:29 acc1_pos: 43.421 acc5_pos: 100.000 loss: 0.5690 (0.7314) acc1: 71.8750 (56.2889) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.1667 data: 0.0710 max mem: 3531 +Test: [210/333] eta: 0:00:26 acc1_pos: 43.421 acc5_pos: 100.000 loss: 0.5930 (0.7262) acc1: 69.2708 (56.8449) acc5: 100.0000 (100.0000) acc1_neg: 63.281 acc5_neg: 100.000 time: 0.1664 data: 0.0734 max mem: 3532 +Test: [220/333] eta: 0:00:24 acc1_pos: 43.421 acc5_pos: 100.000 loss: 0.6394 (0.7262) acc1: 65.8854 (56.9111) acc5: 100.0000 (100.0000) acc1_neg: 73.698 acc5_neg: 100.000 time: 0.1872 data: 0.0945 max mem: 3532 +Test: [230/333] eta: 0:00:22 acc1_pos: 43.421 acc5_pos: 100.000 loss: 0.5078 (0.7151) acc1: 73.6979 (57.9636) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1885 data: 0.0945 max mem: 3532 +Test: [240/333] eta: 0:00:19 acc1_pos: 43.421 acc5_pos: 100.000 loss: 0.4918 (0.7072) acc1: 77.3438 (58.6553) acc5: 100.0000 (100.0000) acc1_neg: 75.781 acc5_neg: 100.000 time: 0.1899 data: 0.0971 max mem: 3532 +Test: [250/333] eta: 0:00:17 acc1_pos: 43.421 acc5_pos: 100.000 loss: 0.5278 (0.7026) acc1: 72.6562 (59.0492) acc5: 100.0000 (100.0000) acc1_neg: 57.812 acc5_neg: 100.000 time: 0.1850 data: 0.0928 max mem: 3532 +Test: [260/333] eta: 0:00:15 acc1_pos: 43.421 acc5_pos: 100.000 loss: 0.6200 (0.7002) acc1: 65.8854 (59.2762) acc5: 100.0000 (100.0000) acc1_neg: 66.667 acc5_neg: 100.000 time: 0.1849 data: 0.0932 max mem: 3532 +Test: [270/333] eta: 0:00:13 acc1_pos: 43.421 acc5_pos: 100.000 loss: 0.5882 (0.6953) acc1: 68.4896 (59.7507) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.1998 data: 0.1059 max mem: 3532 +Test: [280/333] eta: 0:00:11 acc1_pos: 43.421 acc5_pos: 100.000 loss: 0.5821 (0.6937) acc1: 69.7917 (59.8773) acc5: 100.0000 (100.0000) acc1_neg: 69.010 acc5_neg: 100.000 time: 0.2049 data: 0.1078 max mem: 3532 +Test: [290/333] eta: 0:00:08 acc1_pos: 43.421 acc5_pos: 100.000 loss: 0.7229 (0.6954) acc1: 56.5104 (59.6766) acc5: 100.0000 (100.0000) acc1_neg: 52.865 acc5_neg: 100.000 time: 0.1621 data: 0.0646 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 43.421 acc5_pos: 100.000 loss: 0.7637 (0.6985) acc1: 52.8646 (59.4070) acc5: 100.0000 (100.0000) acc1_neg: 54.948 acc5_neg: 100.000 time: 0.1244 data: 0.0297 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 43.421 acc5_pos: 100.000 loss: 0.7236 (0.6933) acc1: 55.2083 (59.8632) acc5: 100.0000 (100.0000) acc1_neg: 73.698 acc5_neg: 100.000 time: 0.1344 data: 0.0431 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 43.421 acc5_pos: 100.000 loss: 0.5159 (0.6889) acc1: 74.4792 (60.2504) acc5: 100.0000 (100.0000) acc1_neg: 57.292 acc5_neg: 100.000 time: 0.1395 data: 0.0493 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 43.421 acc5_pos: 100.000 loss: 0.6666 (0.6907) acc1: 57.2917 (60.1067) acc5: 100.0000 (100.0000) acc1_neg: 56.771 acc5_neg: 100.000 time: 0.1303 data: 0.0404 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 43.421 acc5_pos: 100.000 loss: 0.7299 (0.6910) acc1: 56.7708 (60.0839) acc5: 100.0000 (100.0000) acc1_neg: 56.901 acc5_neg: 100.000 time: 0.1298 data: 0.0367 max mem: 3532 +Test: Total time: 0:01:06 (0.1984 s / it) +* Overall Acc@1 60.084 Acc@5 100.000 + target == real Acc@1 53.496 Acc@5 100.000 + target == fake Acc@1 65.948 Acc@5 100.000 + loss 0.691 + PR-AUC 0.6346 ROC-AUC 0.6338 EER 40.23% (thr=0.5435) +Accuracy of the network on the 127843 test images: 60.1% +[rank0]:[W726 05:47:06.350784804 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:47:25.976000 3169560 torch/distributed/run.py:766] +W0726 05:47:25.976000 3169560 torch/distributed/run.py:766] ***************************************** +W0726 05:47:25.976000 3169560 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:47:25.976000 3169560 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 05:47:53.827265924 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:47:53.838478504 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:47:53.072310724 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:47:53.308706318 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:47:53.382442464 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:47:53.475923278 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:47:53.493293775 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:47:54.570706920 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n32_seed8_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed8_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:18:23 acc1_pos: 93.490 acc5_pos: 100.000 loss: 0.2650 (0.2650) acc1: 93.4896 (93.4896) acc5: 100.0000 (100.0000) time: 4.6965 data: 3.0614 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/235] eta: 0:02:00 acc1_pos: 92.448 acc5_pos: 100.000 loss: 0.2766 (0.2739) acc1: 93.4896 (93.2292) acc5: 100.0000 (100.0000) time: 0.5347 data: 0.2949 max mem: 3531 +Test: [ 20/235] eta: 0:01:09 acc1_pos: 93.229 acc5_pos: 100.000 loss: 0.2706 (0.2684) acc1: 93.7500 (93.7252) acc5: 100.0000 (100.0000) time: 0.1048 data: 0.0099 max mem: 3531 +Test: [ 30/235] eta: 0:00:50 acc1_pos: 95.052 acc5_pos: 100.000 loss: 0.2576 (0.2674) acc1: 94.0104 (93.8424) acc5: 100.0000 (100.0000) time: 0.0914 data: 0.0010 max mem: 3531 +Test: [ 40/235] eta: 0:00:44 acc1_pos: 94.010 acc5_pos: 100.000 loss: 0.2606 (0.2665) acc1: 94.0104 (93.8389) acc5: 100.0000 (100.0000) time: 0.1283 data: 0.0349 max mem: 3531 +Test: [ 50/235] eta: 0:00:41 acc1_pos: 91.406 acc5_pos: 100.000 loss: 0.2634 (0.2679) acc1: 93.4896 (93.7245) acc5: 100.0000 (100.0000) time: 0.1806 data: 0.0857 max mem: 3531 +Test: [ 60/235] eta: 0:00:37 acc1_pos: 92.708 acc5_pos: 100.000 loss: 0.2786 (0.2693) acc1: 93.2292 (93.6561) acc5: 100.0000 (100.0000) time: 0.1895 data: 0.0947 max mem: 3531 +Test: [ 70/235] eta: 0:00:34 acc1_pos: 92.308 acc5_pos: 100.000 loss: 0.2989 (0.3915) acc1: 91.1458 (84.7895) acc5: 100.0000 (100.0000) acc1_neg: 21.094 acc5_neg: 100.000 time: 0.1764 data: 0.0817 max mem: 3531 +Test: [ 80/235] eta: 0:00:31 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.2528 (0.4989) acc1: 22.3958 (77.0126) acc5: 100.0000 (100.0000) acc1_neg: 19.792 acc5_neg: 100.000 time: 0.1745 data: 0.0805 max mem: 3531 +Test: [ 90/235] eta: 0:00:29 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.2722 (0.5843) acc1: 21.8750 (70.9135) acc5: 100.0000 (100.0000) acc1_neg: 21.094 acc5_neg: 100.000 time: 0.1824 data: 0.0877 max mem: 3531 +Test: [100/235] eta: 0:00:27 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.2792 (0.6510) acc1: 21.3542 (66.0736) acc5: 100.0000 (100.0000) acc1_neg: 19.271 acc5_neg: 100.000 time: 0.1931 data: 0.0953 max mem: 3531 +Test: [110/235] eta: 0:00:24 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.2604 (0.7061) acc1: 22.1354 (62.0941) acc5: 100.0000 (100.0000) acc1_neg: 22.396 acc5_neg: 100.000 time: 0.1749 data: 0.0754 max mem: 3531 +Test: [120/235] eta: 0:00:22 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.3093 (0.7571) acc1: 19.7917 (58.4754) acc5: 100.0000 (100.0000) acc1_neg: 20.312 acc5_neg: 100.000 time: 0.1674 data: 0.0684 max mem: 3531 +Test: [130/235] eta: 0:00:20 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.3193 (0.8001) acc1: 18.2292 (55.4628) acc5: 100.0000 (100.0000) acc1_neg: 18.229 acc5_neg: 100.000 time: 0.2020 data: 0.1060 max mem: 3531 +Test: [140/235] eta: 0:00:18 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.3392 (0.8394) acc1: 18.2292 (52.8110) acc5: 100.0000 (100.0000) acc1_neg: 15.104 acc5_neg: 100.000 time: 0.1999 data: 0.1059 max mem: 3531 +Test: [150/235] eta: 0:00:16 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.3342 (0.8711) acc1: 17.9688 (50.5743) acc5: 100.0000 (100.0000) acc1_neg: 20.833 acc5_neg: 100.000 time: 0.1482 data: 0.0544 max mem: 3531 +Test: [160/235] eta: 0:00:14 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.3154 (0.8991) acc1: 18.4896 (48.6154) acc5: 100.0000 (100.0000) acc1_neg: 19.792 acc5_neg: 100.000 time: 0.1380 data: 0.0438 max mem: 3531 +Test: [170/235] eta: 0:00:12 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.3106 (0.9220) acc1: 20.5729 (47.0897) acc5: 100.0000 (100.0000) acc1_neg: 20.052 acc5_neg: 100.000 time: 0.1745 data: 0.0779 max mem: 3531 +Test: [180/235] eta: 0:00:10 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.2895 (0.9419) acc1: 22.1354 (45.7758) acc5: 100.0000 (100.0000) acc1_neg: 22.656 acc5_neg: 100.000 time: 0.2067 data: 0.1116 max mem: 3531 +Test: [190/235] eta: 0:00:08 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.2773 (0.9597) acc1: 22.6562 (44.5749) acc5: 100.0000 (100.0000) acc1_neg: 22.917 acc5_neg: 100.000 time: 0.1950 data: 0.1029 max mem: 3531 +Test: [200/235] eta: 0:00:06 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.2890 (0.9765) acc1: 21.6146 (43.4442) acc5: 100.0000 (100.0000) acc1_neg: 20.573 acc5_neg: 100.000 time: 0.1726 data: 0.0799 max mem: 3531 +Test: [210/235] eta: 0:00:04 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.2993 (0.9912) acc1: 21.6146 (42.4529) acc5: 100.0000 (100.0000) acc1_neg: 21.875 acc5_neg: 100.000 time: 0.1847 data: 0.0900 max mem: 3532 +Test: [220/235] eta: 0:00:02 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.2763 (1.0041) acc1: 21.8750 (41.5724) acc5: 100.0000 (100.0000) acc1_neg: 21.354 acc5_neg: 100.000 time: 0.1728 data: 0.0761 max mem: 3532 +Test: [230/235] eta: 0:00:00 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.2721 (1.0155) acc1: 23.4375 (40.8212) acc5: 100.0000 (100.0000) acc1_neg: 23.177 acc5_neg: 100.000 time: 0.1557 data: 0.0606 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.2865 (1.0207) acc1: 23.1771 (40.5790) acc5: 100.0000 (100.0000) acc1_neg: 13.043 acc5_neg: 100.000 time: 0.1452 data: 0.0554 max mem: 3532 +Test: Total time: 0:00:43 (0.1870 s / it) +* Overall Acc@1 40.579 Acc@5 100.000 + target == real Acc@1 93.679 Acc@5 100.000 + target == fake Acc@1 21.374 Acc@5 100.000 + loss 1.021 + PR-AUC 0.8520 ROC-AUC 0.6916 EER 35.93% (thr=0.2126) +Accuracy of the network on the 89879 test images: 40.6% +[rank0]:[W726 05:48:50.155313765 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:49:08.763000 3170109 torch/distributed/run.py:766] +W0726 05:49:08.763000 3170109 torch/distributed/run.py:766] ***************************************** +W0726 05:49:08.763000 3170109 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:49:08.763000 3170109 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:49:36.420072878 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:49:36.531196437 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:49:37.662666515 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:49:37.777201488 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:49:37.054693984 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:49:37.061760479 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +[rank0]:[W726 05:49:37.080327746 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:49:37.083944737 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n32_seed8_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed8_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:21:32 acc1_pos: 29.688 acc5_pos: 100.000 loss: 0.9372 (0.9372) acc1: 29.6875 (29.6875) acc5: 100.0000 (100.0000) time: 4.8061 data: 3.2683 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/269] eta: 0:02:14 acc1_pos: 78.906 acc5_pos: 100.000 loss: 0.7200 (0.7182) acc1: 51.0417 (53.7169) acc5: 100.0000 (100.0000) time: 0.5186 data: 0.2975 max mem: 3531 +Test: [ 20/269] eta: 0:01:20 acc1_pos: 65.885 acc5_pos: 100.000 loss: 0.7200 (0.7161) acc1: 51.0417 (54.0179) acc5: 100.0000 (100.0000) time: 0.1002 data: 0.0106 max mem: 3531 +Test: [ 30/269] eta: 0:01:01 acc1_pos: 49.219 acc5_pos: 100.000 loss: 0.7274 (0.6969) acc1: 50.7812 (56.3760) acc5: 100.0000 (100.0000) time: 0.1124 data: 0.0225 max mem: 3531 +Test: [ 40/269] eta: 0:00:55 acc1_pos: 54.310 acc5_pos: 100.000 loss: 0.5194 (0.6409) acc1: 75.7812 (62.8811) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.1533 data: 0.0632 max mem: 3531 +Test: [ 50/269] eta: 0:00:51 acc1_pos: 54.310 acc5_pos: 100.000 loss: 0.4834 (0.6257) acc1: 78.3854 (64.5272) acc5: 100.0000 (100.0000) acc1_neg: 49.740 acc5_neg: 100.000 time: 0.2042 data: 0.1133 max mem: 3531 +Test: [ 60/269] eta: 0:00:47 acc1_pos: 54.310 acc5_pos: 100.000 loss: 0.5880 (0.6247) acc1: 70.3125 (64.7328) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.1955 data: 0.1042 max mem: 3531 +Test: [ 70/269] eta: 0:00:41 acc1_pos: 54.310 acc5_pos: 100.000 loss: 0.5799 (0.6149) acc1: 72.1354 (66.0725) acc5: 100.0000 (100.0000) acc1_neg: 79.427 acc5_neg: 100.000 time: 0.1449 data: 0.0522 max mem: 3531 +Test: [ 80/269] eta: 0:00:38 acc1_pos: 54.310 acc5_pos: 100.000 loss: 0.5192 (0.6018) acc1: 75.7812 (67.8241) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.1340 data: 0.0403 max mem: 3531 +Test: [ 90/269] eta: 0:00:36 acc1_pos: 54.310 acc5_pos: 100.000 loss: 0.5483 (0.6056) acc1: 70.5729 (67.3306) acc5: 100.0000 (100.0000) acc1_neg: 70.573 acc5_neg: 100.000 time: 0.1808 data: 0.0890 max mem: 3531 +Test: [100/269] eta: 0:00:34 acc1_pos: 54.310 acc5_pos: 100.000 loss: 0.6648 (0.6094) acc1: 61.4583 (66.6744) acc5: 100.0000 (100.0000) acc1_neg: 52.083 acc5_neg: 100.000 time: 0.2207 data: 0.1301 max mem: 3531 +Test: [110/269] eta: 0:00:32 acc1_pos: 54.310 acc5_pos: 100.000 loss: 0.6648 (0.6138) acc1: 59.3750 (66.0942) acc5: 100.0000 (100.0000) acc1_neg: 49.219 acc5_neg: 100.000 time: 0.2090 data: 0.1179 max mem: 3531 +Test: [120/269] eta: 0:00:29 acc1_pos: 54.310 acc5_pos: 100.000 loss: 0.6187 (0.6123) acc1: 62.7604 (66.2319) acc5: 100.0000 (100.0000) acc1_neg: 55.729 acc5_neg: 100.000 time: 0.1485 data: 0.0550 max mem: 3531 +Test: [130/269] eta: 0:00:27 acc1_pos: 54.310 acc5_pos: 100.000 loss: 0.5911 (0.6160) acc1: 65.1042 (65.7085) acc5: 100.0000 (100.0000) acc1_neg: 58.854 acc5_neg: 100.000 time: 0.1555 data: 0.0628 max mem: 3531 +Test: [140/269] eta: 0:00:25 acc1_pos: 54.310 acc5_pos: 100.000 loss: 0.6702 (0.6178) acc1: 57.8125 (65.4865) acc5: 100.0000 (100.0000) acc1_neg: 57.812 acc5_neg: 100.000 time: 0.2011 data: 0.1079 max mem: 3531 +Test: [150/269] eta: 0:00:23 acc1_pos: 54.310 acc5_pos: 100.000 loss: 0.6022 (0.6125) acc1: 61.9792 (65.8561) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1893 data: 0.0952 max mem: 3531 +Test: [160/269] eta: 0:00:21 acc1_pos: 54.310 acc5_pos: 100.000 loss: 0.4640 (0.6052) acc1: 78.3854 (66.6893) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.1827 data: 0.0907 max mem: 3531 +Test: [170/269] eta: 0:00:19 acc1_pos: 54.310 acc5_pos: 100.000 loss: 0.5631 (0.6055) acc1: 73.4375 (66.6773) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.2076 data: 0.1147 max mem: 3531 +Test: [180/269] eta: 0:00:17 acc1_pos: 54.310 acc5_pos: 100.000 loss: 0.5405 (0.5999) acc1: 75.2604 (67.4551) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.1949 data: 0.0970 max mem: 3531 +Test: [190/269] eta: 0:00:15 acc1_pos: 54.310 acc5_pos: 100.000 loss: 0.5334 (0.5986) acc1: 73.9583 (67.5774) acc5: 100.0000 (100.0000) acc1_neg: 72.917 acc5_neg: 100.000 time: 0.1790 data: 0.0805 max mem: 3531 +Test: [200/269] eta: 0:00:13 acc1_pos: 54.310 acc5_pos: 100.000 loss: 0.6380 (0.6005) acc1: 63.0208 (67.3507) acc5: 100.0000 (100.0000) acc1_neg: 63.021 acc5_neg: 100.000 time: 0.1873 data: 0.0922 max mem: 3531 +Test: [210/269] eta: 0:00:11 acc1_pos: 54.310 acc5_pos: 100.000 loss: 0.5718 (0.5973) acc1: 67.1875 (67.6725) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.1882 data: 0.0947 max mem: 3532 +Test: [220/269] eta: 0:00:09 acc1_pos: 54.310 acc5_pos: 100.000 loss: 0.5067 (0.5935) acc1: 77.3438 (68.1090) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1977 data: 0.1061 max mem: 3532 +Test: [230/269] eta: 0:00:07 acc1_pos: 54.310 acc5_pos: 100.000 loss: 0.4819 (0.5886) acc1: 80.7292 (68.6497) acc5: 100.0000 (100.0000) acc1_neg: 70.833 acc5_neg: 100.000 time: 0.1915 data: 0.1000 max mem: 3532 +Test: [240/269] eta: 0:00:05 acc1_pos: 54.310 acc5_pos: 100.000 loss: 0.5743 (0.5922) acc1: 70.8333 (68.2173) acc5: 100.0000 (100.0000) acc1_neg: 54.427 acc5_neg: 100.000 time: 0.1901 data: 0.0948 max mem: 3532 +Test: [250/269] eta: 0:00:03 acc1_pos: 54.310 acc5_pos: 100.000 loss: 0.6677 (0.5966) acc1: 55.9896 (67.6741) acc5: 100.0000 (100.0000) acc1_neg: 42.188 acc5_neg: 100.000 time: 0.1912 data: 0.0950 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 54.310 acc5_pos: 100.000 loss: 0.7445 (0.6030) acc1: 46.3542 (66.9161) acc5: 100.0000 (100.0000) acc1_neg: 70.052 acc5_neg: 100.000 time: 0.1808 data: 0.0889 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 54.310 acc5_pos: 100.000 loss: 0.6448 (0.6023) acc1: 58.8542 (67.0413) acc5: 100.0000 (100.0000) acc1_neg: 82.940 acc5_neg: 100.000 time: 0.1583 data: 0.0643 max mem: 3532 +Test: Total time: 0:00:51 (0.1926 s / it) +* Overall Acc@1 67.041 Acc@5 100.000 + target == real Acc@1 56.256 Acc@5 100.000 + target == fake Acc@1 68.513 Acc@5 100.000 + loss 0.602 + PR-AUC 0.9340 ROC-AUC 0.6783 EER 37.19% (thr=0.5283) +Accuracy of the network on the 103293 test images: 67.0% +[rank0]:[W726 05:50:41.282214126 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T05:50:48Z] eval_done tag=fixed500k_fake_n32_seed8_resampled +[2026-07-26T05:50:48Z] eval_start tag=fixed500k_real_n16_seed9_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed9_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T05:50:48Z +hostname=n111-046-081 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed9_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n16_seed9_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 05:51:00.798000 3170685 torch/distributed/run.py:766] +W0726 05:51:00.798000 3170685 torch/distributed/run.py:766] ***************************************** +W0726 05:51:00.798000 3170685 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:51:00.798000 3170685 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:51:27.834075876 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:51:27.134671898 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:51:27.510763598 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:51:27.565878353 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:51:28.627891798 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W726 05:51:28.628079668 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:51:28.674117348 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:51:28.692259685 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n16_seed9_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed9_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:23:30 acc1_pos: 28.125 acc5_pos: 100.000 loss: 1.0879 (1.0879) acc1: 28.1250 (28.1250) acc5: 100.0000 (100.0000) time: 6.0258 data: 4.4025 max mem: 3531 +Test: [ 10/234] eta: 0:02:21 acc1_pos: 11.458 acc5_pos: 100.000 loss: 1.1022 (1.0908) acc1: 27.6042 (25.1657) acc5: 100.0000 (100.0000) time: 0.6295 data: 0.4006 max mem: 3531 +Test: [ 20/234] eta: 0:01:20 acc1_pos: 95.573 acc5_pos: 100.000 loss: 1.1524 (1.0254) acc1: 19.5312 (32.8001) acc5: 100.0000 (100.0000) time: 0.0942 data: 0.0004 max mem: 3531 +Test: [ 30/234] eta: 0:01:15 acc1_pos: 39.844 acc5_pos: 100.000 loss: 0.2555 (0.8047) acc1: 94.0104 (49.7648) acc5: 100.0000 (100.0000) time: 0.2246 data: 0.1300 max mem: 3531 +Test: [ 40/234] eta: 0:01:08 acc1_pos: 99.219 acc5_pos: 100.000 loss: 0.2433 (0.7488) acc1: 94.0104 (53.2584) acc5: 100.0000 (100.0000) time: 0.3316 data: 0.2401 max mem: 3531 +Test: [ 50/234] eta: 0:01:00 acc1_pos: 99.479 acc5_pos: 100.000 loss: 0.1275 (0.6265) acc1: 98.9583 (62.2447) acc5: 100.0000 (100.0000) time: 0.2652 data: 0.1743 max mem: 3531 +Test: [ 60/234] eta: 0:00:52 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.1168 (0.5407) acc1: 99.4792 (68.4085) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1995 data: 0.1094 max mem: 3531 +Test: [ 70/234] eta: 0:00:48 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.1037 (0.5364) acc1: 100.0000 (69.3772) acc5: 100.0000 (100.0000) acc1_neg: 62.240 acc5_neg: 100.000 time: 0.2115 data: 0.1209 max mem: 3531 +Test: [ 80/234] eta: 0:00:44 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.6690 (0.5294) acc1: 65.8854 (70.3639) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2437 data: 0.1526 max mem: 3531 +Test: [ 90/234] eta: 0:00:40 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.1815 (0.4919) acc1: 92.4479 (73.0283) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.2352 data: 0.1439 max mem: 3531 +Test: [100/234] eta: 0:00:37 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.2111 (0.4867) acc1: 91.6667 (73.6824) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2249 data: 0.1337 max mem: 3531 +Test: [110/234] eta: 0:00:34 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.4027 (0.4769) acc1: 80.4688 (74.6246) acc5: 100.0000 (100.0000) acc1_neg: 76.302 acc5_neg: 100.000 time: 0.2400 data: 0.1489 max mem: 3531 +Test: [120/234] eta: 0:00:30 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.3308 (0.4739) acc1: 82.5521 (75.2066) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2257 data: 0.1349 max mem: 3531 +Test: [130/234] eta: 0:00:27 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.2931 (0.4632) acc1: 90.1042 (76.2981) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.2192 data: 0.1283 max mem: 3531 +Test: [140/234] eta: 0:00:24 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.3842 (0.4865) acc1: 83.5938 (75.2142) acc5: 100.0000 (100.0000) acc1_neg: 21.875 acc5_neg: 100.000 time: 0.2390 data: 0.1446 max mem: 3531 +Test: [150/234] eta: 0:00:21 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.5097 (0.4922) acc1: 72.1354 (75.0397) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.2110 data: 0.1168 max mem: 3531 +Test: [160/234] eta: 0:00:18 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.3688 (0.4828) acc1: 82.5521 (75.7877) acc5: 100.0000 (100.0000) acc1_neg: 78.646 acc5_neg: 100.000 time: 0.1865 data: 0.0954 max mem: 3531 +Test: [170/234] eta: 0:00:15 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.4516 (0.4793) acc1: 81.7708 (76.2381) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1778 data: 0.0852 max mem: 3531 +Test: [180/234] eta: 0:00:13 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.4805 (0.4917) acc1: 78.1250 (75.5784) acc5: 100.0000 (100.0000) acc1_neg: 26.562 acc5_neg: 100.000 time: 0.1789 data: 0.0865 max mem: 3531 +Test: [190/234] eta: 0:00:10 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.6576 (0.5029) acc1: 61.7188 (74.8214) acc5: 100.0000 (100.0000) acc1_neg: 60.417 acc5_neg: 100.000 time: 0.1673 data: 0.0764 max mem: 3531 +Test: [200/234] eta: 0:00:08 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.6499 (0.4936) acc1: 63.5417 (75.4081) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1419 data: 0.0518 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.1174 (0.4776) acc1: 99.7396 (76.4798) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1320 data: 0.0422 max mem: 3532 +Test: [220/234] eta: 0:00:03 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.2323 (0.4676) acc1: 95.0521 (77.2000) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.1219 data: 0.0323 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.3978 (0.4789) acc1: 58.0729 (76.1691) acc5: 100.0000 (100.0000) acc1_neg: 51.302 acc5_neg: 100.000 time: 0.1204 data: 0.0308 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.7098 (0.4820) acc1: 55.2083 (75.9041) acc5: 100.0000 (100.0000) acc1_neg: 51.362 acc5_neg: 100.000 time: 0.1182 data: 0.0277 max mem: 3532 +Test: Total time: 0:00:51 (0.2199 s / it) +* Overall Acc@1 75.904 Acc@5 100.000 + target == real Acc@1 62.502 Acc@5 100.000 + target == fake Acc@1 79.681 Acc@5 100.000 + loss 0.482 + PR-AUC 0.9459 ROC-AUC 0.8289 EER 27.62% (thr=0.6265) +Accuracy of the network on the 89729 test images: 75.9% +[rank0]:[W726 05:52:26.477130118 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:52:44.910000 3171234 torch/distributed/run.py:766] +W0726 05:52:44.910000 3171234 torch/distributed/run.py:766] ***************************************** +W0726 05:52:44.910000 3171234 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:52:44.910000 3171234 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:53:12.512245025 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:53:13.578994535 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:53:13.604545407 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:53:13.777941750 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +[rank0]:[W726 05:53:13.969399135 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:53:13.978462394 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:53:13.092043766 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:53:13.116510157 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n16_seed9_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed9_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:09:06 acc1_pos: 22.396 acc5_pos: 100.000 loss: 1.1914 (1.1914) acc1: 22.3958 (22.3958) acc5: 100.0000 (100.0000) time: 6.2618 data: 4.9243 max mem: 3531 +Test: [ 10/2387] eta: 0:25:47 acc1_pos: 7.812 acc5_pos: 100.000 loss: 1.2718 (1.2665) acc1: 14.0625 (16.9034) acc5: 100.0000 (100.0000) time: 0.6510 data: 0.4482 max mem: 3531 +Test: [ 20/2387] eta: 0:15:16 acc1_pos: 5.469 acc5_pos: 100.000 loss: 1.3039 (1.3113) acc1: 12.2396 (15.8730) acc5: 100.0000 (100.0000) time: 0.0934 data: 0.0005 max mem: 3531 +Test: [ 30/2387] eta: 0:12:59 acc1_pos: 9.115 acc5_pos: 100.000 loss: 1.3672 (1.3263) acc1: 10.1562 (14.5161) acc5: 100.0000 (100.0000) time: 0.1548 data: 0.0586 max mem: 3531 +Test: [ 40/2387] eta: 0:12:08 acc1_pos: 7.812 acc5_pos: 100.000 loss: 1.3427 (1.3234) acc1: 11.1979 (13.7259) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1347 max mem: 3531 +Test: [ 50/2387] eta: 0:11:24 acc1_pos: 14.583 acc5_pos: 100.000 loss: 1.3045 (1.3145) acc1: 11.4583 (14.3229) acc5: 100.0000 (100.0000) time: 0.2336 data: 0.1391 max mem: 3531 +Test: [ 60/2387] eta: 0:11:04 acc1_pos: 17.969 acc5_pos: 100.000 loss: 1.4019 (1.3451) acc1: 10.6771 (13.6697) acc5: 100.0000 (100.0000) time: 0.2349 data: 0.1402 max mem: 3531 +Test: [ 70/2387] eta: 0:10:48 acc1_pos: 15.625 acc5_pos: 100.000 loss: 1.5943 (1.3880) acc1: 7.0312 (12.4707) acc5: 100.0000 (100.0000) time: 0.2465 data: 0.1533 max mem: 3531 +Test: [ 80/2387] eta: 0:10:47 acc1_pos: 10.938 acc5_pos: 100.000 loss: 1.5740 (1.3988) acc1: 5.9896 (12.0660) acc5: 100.0000 (100.0000) time: 0.2649 data: 0.1709 max mem: 3531 +Test: [ 90/2387] eta: 0:10:34 acc1_pos: 12.240 acc5_pos: 100.000 loss: 1.4995 (1.4103) acc1: 7.0312 (11.6358) acc5: 100.0000 (100.0000) time: 0.2633 data: 0.1688 max mem: 3531 +Test: [ 100/2387] eta: 0:10:21 acc1_pos: 5.469 acc5_pos: 100.000 loss: 1.5198 (1.4260) acc1: 5.9896 (11.1876) acc5: 100.0000 (100.0000) time: 0.2357 data: 0.1440 max mem: 3531 +Test: [ 110/2387] eta: 0:10:07 acc1_pos: 18.490 acc5_pos: 100.000 loss: 1.4802 (1.4278) acc1: 8.3333 (11.0947) acc5: 100.0000 (100.0000) time: 0.2247 data: 0.1320 max mem: 3531 +Test: [ 120/2387] eta: 0:09:58 acc1_pos: 9.896 acc5_pos: 100.000 loss: 1.4307 (1.4314) acc1: 9.8958 (10.9978) acc5: 100.0000 (100.0000) time: 0.2254 data: 0.1300 max mem: 3531 +Test: [ 130/2387] eta: 0:09:54 acc1_pos: 23.958 acc5_pos: 100.000 loss: 1.4307 (1.4356) acc1: 8.5938 (10.9295) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1510 max mem: 3531 +Test: [ 140/2387] eta: 0:09:47 acc1_pos: 8.073 acc5_pos: 100.000 loss: 1.4744 (1.4383) acc1: 7.5521 (10.7362) acc5: 100.0000 (100.0000) time: 0.2467 data: 0.1536 max mem: 3531 +Test: [ 150/2387] eta: 0:09:46 acc1_pos: 1.823 acc5_pos: 100.000 loss: 1.4619 (1.4372) acc1: 9.1146 (10.8806) acc5: 100.0000 (100.0000) time: 0.2539 data: 0.1603 max mem: 3531 +Test: [ 160/2387] eta: 0:09:33 acc1_pos: 9.589 acc5_pos: 100.000 loss: 1.3512 (1.4097) acc1: 12.5000 (12.6892) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2274 data: 0.1359 max mem: 3531 +Test: [ 170/2387] eta: 0:09:21 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1632 (1.3364) acc1: 98.4375 (17.7464) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1872 data: 0.0953 max mem: 3531 +Test: [ 180/2387] eta: 0:09:13 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1562 (1.2712) acc1: 98.9583 (22.2390) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1992 data: 0.1047 max mem: 3531 +Test: [ 190/2387] eta: 0:09:05 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1547 (1.2127) acc1: 99.2188 (26.2612) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2036 data: 0.1090 max mem: 3531 +Test: [ 200/2387] eta: 0:08:56 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1537 (1.1599) acc1: 99.2188 (29.8922) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1960 data: 0.1030 max mem: 3531 +Test: [ 210/2387] eta: 0:08:48 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1549 (1.1125) acc1: 98.9583 (33.1642) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1911 data: 0.0980 max mem: 3532 +Test: [ 220/2387] eta: 0:08:42 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1584 (1.0692) acc1: 98.9583 (36.1414) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1953 data: 0.1016 max mem: 3532 +Test: [ 230/2387] eta: 0:08:33 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1567 (1.0296) acc1: 98.9583 (38.8652) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1888 data: 0.0959 max mem: 3532 +Test: [ 240/2387] eta: 0:08:25 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1494 (0.9929) acc1: 99.2188 (41.3857) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1735 data: 0.0777 max mem: 3532 +Test: [ 250/2387] eta: 0:08:17 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1332 (0.9570) acc1: 99.7396 (43.7189) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1727 data: 0.0768 max mem: 3532 +Test: [ 260/2387] eta: 0:08:12 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0818 (0.9235) acc1: 100.0000 (45.8752) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1858 data: 0.0941 max mem: 3532 +Test: [ 270/2387] eta: 0:08:05 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0804 (0.8923) acc1: 100.0000 (47.8715) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1870 data: 0.0961 max mem: 3532 +Test: [ 280/2387] eta: 0:08:00 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0795 (0.8634) acc1: 100.0000 (49.7266) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1817 data: 0.0906 max mem: 3532 +Test: [ 290/2387] eta: 0:07:55 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0790 (0.8365) acc1: 100.0000 (51.4542) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0979 max mem: 3532 +Test: [ 300/2387] eta: 0:07:48 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0762 (0.8112) acc1: 100.0000 (53.0670) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1785 data: 0.0857 max mem: 3532 +Test: [ 310/2387] eta: 0:07:46 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0778 (0.7892) acc1: 100.0000 (54.5552) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1899 data: 0.0973 max mem: 3532 +Test: [ 320/2387] eta: 0:07:42 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1715 (0.7701) acc1: 99.4792 (55.9506) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2086 data: 0.1171 max mem: 3532 +Test: [ 330/2387] eta: 0:07:39 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1756 (0.7521) acc1: 99.4792 (57.2633) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2069 data: 0.1160 max mem: 3532 +Test: [ 340/2387] eta: 0:07:34 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1752 (0.7352) acc1: 99.4792 (58.4968) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1979 data: 0.1064 max mem: 3532 +Test: [ 350/2387] eta: 0:07:29 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1644 (0.7169) acc1: 99.7396 (59.6770) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1746 data: 0.0832 max mem: 3532 +Test: [ 360/2387] eta: 0:07:25 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1090 (0.7008) acc1: 99.7396 (60.7759) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1778 data: 0.0847 max mem: 3532 +Test: [ 370/2387] eta: 0:07:21 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1396 (0.6857) acc1: 99.4792 (61.8149) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1859 data: 0.0922 max mem: 3532 +Test: [ 380/2387] eta: 0:07:16 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1409 (0.6715) acc1: 99.2188 (62.7987) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1791 data: 0.0881 max mem: 3532 +Test: [ 390/2387] eta: 0:07:12 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1409 (0.6579) acc1: 99.2188 (63.7315) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1772 data: 0.0845 max mem: 3532 +Test: [ 400/2387] eta: 0:07:10 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1414 (0.6450) acc1: 99.2188 (64.6229) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1948 data: 0.1023 max mem: 3532 +Test: [ 410/2387] eta: 0:07:06 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1399 (0.6327) acc1: 99.4792 (65.4729) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1970 data: 0.1060 max mem: 3532 +Test: [ 420/2387] eta: 0:07:04 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1378 (0.6210) acc1: 99.4792 (66.2807) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2015 data: 0.1084 max mem: 3532 +Test: [ 430/2387] eta: 0:07:01 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1366 (0.6098) acc1: 99.7396 (67.0576) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2054 data: 0.1113 max mem: 3532 +Test: [ 440/2387] eta: 0:06:57 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1280 (0.5983) acc1: 100.0000 (67.8028) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1888 data: 0.0953 max mem: 3533 +Test: [ 450/2387] eta: 0:06:53 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0897 (0.5870) acc1: 100.0000 (68.5167) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1782 data: 0.0841 max mem: 3533 +Test: [ 460/2387] eta: 0:06:49 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0889 (0.5762) acc1: 100.0000 (69.1997) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1713 data: 0.0769 max mem: 3533 +Test: [ 470/2387] eta: 0:06:46 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0889 (0.5659) acc1: 100.0000 (69.8536) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1745 data: 0.0801 max mem: 3533 +Test: [ 480/2387] eta: 0:06:42 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0893 (0.5560) acc1: 100.0000 (70.4803) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1763 data: 0.0819 max mem: 3533 +Test: [ 490/2387] eta: 0:06:38 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0892 (0.5465) acc1: 100.0000 (71.0816) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1718 data: 0.0765 max mem: 3533 +Test: [ 500/2387] eta: 0:06:34 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0895 (0.5374) acc1: 100.0000 (71.6588) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1634 data: 0.0697 max mem: 3533 +Test: [ 510/2387] eta: 0:06:31 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0902 (0.5287) acc1: 100.0000 (72.2134) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1675 data: 0.0732 max mem: 3533 +Test: [ 520/2387] eta: 0:06:27 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1100 (0.5207) acc1: 100.0000 (72.7467) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1720 data: 0.0765 max mem: 3533 +Test: [ 530/2387] eta: 0:06:24 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1110 (0.5130) acc1: 100.0000 (73.2600) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1691 data: 0.0769 max mem: 3533 +Test: [ 540/2387] eta: 0:06:21 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1092 (0.5055) acc1: 100.0000 (73.7542) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1757 data: 0.0844 max mem: 3533 +Test: [ 550/2387] eta: 0:06:18 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1103 (0.4984) acc1: 100.0000 (74.2306) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1798 data: 0.0849 max mem: 3533 +Test: [ 560/2387] eta: 0:06:15 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1104 (0.4915) acc1: 100.0000 (74.6899) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1793 data: 0.0847 max mem: 3533 +Test: [ 570/2387] eta: 0:06:13 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1103 (0.4848) acc1: 100.0000 (75.1327) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1855 data: 0.0948 max mem: 3533 +Test: [ 580/2387] eta: 0:06:10 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1099 (0.4783) acc1: 100.0000 (75.5607) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1847 data: 0.0929 max mem: 3533 +Test: [ 590/2387] eta: 0:06:07 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1099 (0.4721) acc1: 100.0000 (75.9743) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1820 data: 0.0889 max mem: 3533 +Test: [ 600/2387] eta: 0:06:05 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1115 (0.4661) acc1: 100.0000 (76.3740) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1861 data: 0.0941 max mem: 3533 +Test: [ 610/2387] eta: 0:06:02 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1151 (0.4604) acc1: 100.0000 (76.7603) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1923 data: 0.1006 max mem: 3533 +Test: [ 620/2387] eta: 0:06:00 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1168 (0.4549) acc1: 100.0000 (77.1332) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1954 data: 0.1025 max mem: 3533 +Test: [ 630/2387] eta: 0:05:58 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1179 (0.4495) acc1: 100.0000 (77.4948) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1961 data: 0.1040 max mem: 3533 +Test: [ 640/2387] eta: 0:05:56 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1186 (0.4444) acc1: 100.0000 (77.8459) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2009 data: 0.1099 max mem: 3533 +Test: [ 650/2387] eta: 0:05:53 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1171 (0.4394) acc1: 100.0000 (78.1858) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1990 data: 0.1081 max mem: 3533 +Test: [ 660/2387] eta: 0:05:51 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1177 (0.4345) acc1: 100.0000 (78.5158) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1901 data: 0.0996 max mem: 3533 +Test: [ 670/2387] eta: 0:05:49 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1177 (0.4298) acc1: 100.0000 (78.8360) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1946 data: 0.1041 max mem: 3534 +Test: [ 680/2387] eta: 0:05:47 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1152 (0.4251) acc1: 100.0000 (79.1468) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2121 data: 0.1217 max mem: 3534 +Test: [ 690/2387] eta: 0:05:45 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1159 (0.4220) acc1: 100.0000 (79.3510) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1976 data: 0.1046 max mem: 3534 +Test: [ 700/2387] eta: 0:05:42 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2481 (0.4196) acc1: 90.6250 (79.5129) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1766 data: 0.0806 max mem: 3534 +Test: [ 710/2387] eta: 0:05:39 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2469 (0.4172) acc1: 90.6250 (79.6739) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1787 data: 0.0843 max mem: 3534 +Test: [ 720/2387] eta: 0:05:37 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2442 (0.4148) acc1: 91.1458 (79.8291) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1746 data: 0.0828 max mem: 3534 +Test: [ 730/2387] eta: 0:05:34 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2485 (0.4126) acc1: 90.3646 (79.9775) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1690 data: 0.0763 max mem: 3534 +Test: [ 740/2387] eta: 0:05:31 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2550 (0.4104) acc1: 90.6250 (80.1247) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.1705 data: 0.0782 max mem: 3534 +Test: [ 750/2387] eta: 0:05:29 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2439 (0.4081) acc1: 90.8854 (80.2721) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.1796 data: 0.0888 max mem: 3534 +Test: [ 760/2387] eta: 0:05:26 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2366 (0.4056) acc1: 91.4062 (80.4417) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1777 data: 0.0843 max mem: 3534 +Test: [ 770/2387] eta: 0:05:23 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1683 (0.4024) acc1: 96.3542 (80.6620) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1700 data: 0.0770 max mem: 3534 +Test: [ 780/2387] eta: 0:05:21 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1492 (0.3990) acc1: 98.6979 (80.8992) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1804 data: 0.0874 max mem: 3534 +Test: [ 790/2387] eta: 0:05:19 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1202 (0.3958) acc1: 99.4792 (81.1091) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1795 data: 0.0854 max mem: 3534 +Test: [ 800/2387] eta: 0:05:16 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1167 (0.3924) acc1: 100.0000 (81.3414) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1695 data: 0.0746 max mem: 3534 +Test: [ 810/2387] eta: 0:05:13 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1239 (0.3892) acc1: 100.0000 (81.5640) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1701 data: 0.0764 max mem: 3534 +Test: [ 820/2387] eta: 0:05:11 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1317 (0.3862) acc1: 99.4792 (81.7775) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1717 data: 0.0783 max mem: 3534 +Test: [ 830/2387] eta: 0:05:09 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1380 (0.3834) acc1: 99.4792 (81.9817) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1803 data: 0.0851 max mem: 3534 +Test: [ 840/2387] eta: 0:05:06 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1380 (0.3806) acc1: 98.9583 (82.1867) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1759 data: 0.0836 max mem: 3534 +Test: [ 850/2387] eta: 0:05:04 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1346 (0.3777) acc1: 99.7396 (82.3853) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1663 data: 0.0762 max mem: 3534 +Test: [ 860/2387] eta: 0:05:01 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1248 (0.3747) acc1: 99.7396 (82.5838) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1682 data: 0.0772 max mem: 3534 +Test: [ 870/2387] eta: 0:04:59 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1240 (0.3719) acc1: 99.7396 (82.7766) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1729 data: 0.0802 max mem: 3534 +Test: [ 880/2387] eta: 0:04:57 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1308 (0.3692) acc1: 99.7396 (82.9677) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1797 data: 0.0872 max mem: 3534 +Test: [ 890/2387] eta: 0:04:54 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1167 (0.3663) acc1: 99.7396 (83.1583) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0808 max mem: 3534 +Test: [ 900/2387] eta: 0:04:52 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1172 (0.3709) acc1: 99.7396 (82.7911) acc5: 100.0000 (100.0000) acc1_neg: 39.323 acc5_neg: 100.000 time: 0.1642 data: 0.0679 max mem: 3535 +Test: [ 910/2387] eta: 0:04:49 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.9425 (0.3772) acc1: 38.0208 (82.3008) acc5: 100.0000 (100.0000) acc1_neg: 36.979 acc5_neg: 100.000 time: 0.1600 data: 0.0654 max mem: 3535 +Test: [ 920/2387] eta: 0:04:46 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.9584 (0.3835) acc1: 37.7604 (81.8223) acc5: 100.0000 (100.0000) acc1_neg: 38.542 acc5_neg: 100.000 time: 0.1571 data: 0.0655 max mem: 3535 +Test: [ 930/2387] eta: 0:04:44 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.9480 (0.3895) acc1: 38.5417 (81.3692) acc5: 100.0000 (100.0000) acc1_neg: 37.500 acc5_neg: 100.000 time: 0.1648 data: 0.0716 max mem: 3535 +Test: [ 940/2387] eta: 0:04:42 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.9480 (0.3955) acc1: 39.8438 (80.9251) acc5: 100.0000 (100.0000) acc1_neg: 35.677 acc5_neg: 100.000 time: 0.1675 data: 0.0711 max mem: 3535 +Test: [ 950/2387] eta: 0:04:39 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.9487 (0.4012) acc1: 38.5417 (80.4792) acc5: 100.0000 (100.0000) acc1_neg: 39.583 acc5_neg: 100.000 time: 0.1556 data: 0.0594 max mem: 3535 +Test: [ 960/2387] eta: 0:04:37 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.9294 (0.4068) acc1: 38.5417 (80.0531) acc5: 100.0000 (100.0000) acc1_neg: 39.323 acc5_neg: 100.000 time: 0.1620 data: 0.0692 max mem: 3535 +Test: [ 970/2387] eta: 0:04:34 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.9432 (0.4125) acc1: 38.2812 (79.6137) acc5: 100.0000 (100.0000) acc1_neg: 38.802 acc5_neg: 100.000 time: 0.1742 data: 0.0814 max mem: 3535 +Test: [ 980/2387] eta: 0:04:32 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.9353 (0.4126) acc1: 38.8021 (79.5943) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1727 data: 0.0765 max mem: 3535 +Test: [ 990/2387] eta: 0:04:30 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1755 (0.4100) acc1: 98.1771 (79.7874) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1763 data: 0.0822 max mem: 3535 +Test: [1000/2387] eta: 0:04:28 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1365 (0.4073) acc1: 99.2188 (79.9755) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1899 data: 0.0974 max mem: 3535 +Test: [1010/2387] eta: 0:04:26 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1426 (0.4047) acc1: 99.2188 (80.1617) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1801 data: 0.0864 max mem: 3535 +Test: [1020/2387] eta: 0:04:24 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1465 (0.4024) acc1: 99.2188 (80.3382) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1711 data: 0.0790 max mem: 3535 +Test: [1030/2387] eta: 0:04:22 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1380 (0.3999) acc1: 99.4792 (80.5195) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1797 data: 0.0870 max mem: 3535 +Test: [1040/2387] eta: 0:04:20 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1277 (0.3972) acc1: 100.0000 (80.7039) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1887 data: 0.0936 max mem: 3535 +Test: [1050/2387] eta: 0:04:18 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1267 (0.3947) acc1: 100.0000 (80.8828) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1837 data: 0.0878 max mem: 3535 +Test: [1060/2387] eta: 0:04:15 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1315 (0.3924) acc1: 99.7396 (81.0568) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1549 data: 0.0576 max mem: 3535 +Test: [1070/2387] eta: 0:04:13 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1358 (0.3901) acc1: 99.4792 (81.2247) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1547 data: 0.0586 max mem: 3535 +Test: [1080/2387] eta: 0:04:11 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1440 (0.3879) acc1: 99.4792 (81.3868) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1849 data: 0.0922 max mem: 3535 +Test: [1090/2387] eta: 0:04:09 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1682 (0.3862) acc1: 98.1771 (81.5224) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.2070 data: 0.1145 max mem: 3535 +Test: [1100/2387] eta: 0:04:07 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1721 (0.3842) acc1: 97.6562 (81.6691) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1999 data: 0.1058 max mem: 3535 +Test: [1110/2387] eta: 0:04:05 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1740 (0.3826) acc1: 97.3958 (81.7945) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1979 data: 0.1047 max mem: 3535 +Test: [1120/2387] eta: 0:04:04 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1779 (0.3807) acc1: 97.6562 (81.9413) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2049 data: 0.1120 max mem: 3536 +Test: [1130/2387] eta: 0:04:02 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1661 (0.3789) acc1: 98.6979 (82.0826) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1917 data: 0.0991 max mem: 3536 +Test: [1140/2387] eta: 0:04:00 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1354 (0.3769) acc1: 99.2188 (82.2337) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1987 data: 0.1068 max mem: 3536 +Test: [1150/2387] eta: 0:03:58 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1307 (0.3749) acc1: 99.7396 (82.3804) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2131 data: 0.1214 max mem: 3536 +Test: [1160/2387] eta: 0:03:56 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1135 (0.3726) acc1: 100.0000 (82.5299) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1964 data: 0.1041 max mem: 3536 +Test: [1170/2387] eta: 0:03:54 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1073 (0.3703) acc1: 100.0000 (82.6788) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1850 data: 0.0910 max mem: 3536 +Test: [1180/2387] eta: 0:03:52 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1020 (0.3681) acc1: 100.0000 (82.8209) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1818 data: 0.0894 max mem: 3536 +Test: [1190/2387] eta: 0:03:50 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1092 (0.3660) acc1: 100.0000 (82.9649) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1747 data: 0.0817 max mem: 3536 +Test: [1200/2387] eta: 0:03:48 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0969 (0.3637) acc1: 100.0000 (83.1067) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1827 data: 0.0889 max mem: 3536 +Test: [1210/2387] eta: 0:03:46 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0925 (0.3614) acc1: 100.0000 (83.2462) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1818 data: 0.0902 max mem: 3536 +Test: [1220/2387] eta: 0:03:44 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0855 (0.3591) acc1: 100.0000 (83.3826) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1820 data: 0.0907 max mem: 3536 +Test: [1230/2387] eta: 0:03:42 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0748 (0.3568) acc1: 100.0000 (83.5176) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1887 data: 0.0973 max mem: 3536 +Test: [1240/2387] eta: 0:03:40 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0739 (0.3545) acc1: 100.0000 (83.6504) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1893 data: 0.0974 max mem: 3536 +Test: [1250/2387] eta: 0:03:38 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0746 (0.3523) acc1: 100.0000 (83.7809) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1995 data: 0.1074 max mem: 3536 +Test: [1260/2387] eta: 0:03:36 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0753 (0.3501) acc1: 100.0000 (83.9095) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1999 data: 0.1081 max mem: 3536 +Test: [1270/2387] eta: 0:03:34 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0711 (0.3479) acc1: 100.0000 (84.0361) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1965 data: 0.1054 max mem: 3536 +Test: [1280/2387] eta: 0:03:32 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0783 (0.3461) acc1: 100.0000 (84.1595) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1892 data: 0.0986 max mem: 3536 +Test: [1290/2387] eta: 0:03:31 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1109 (0.3443) acc1: 99.7396 (84.2729) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1886 data: 0.0975 max mem: 3536 +Test: [1300/2387] eta: 0:03:29 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1169 (0.3426) acc1: 99.7396 (84.3890) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2003 data: 0.1091 max mem: 3536 +Test: [1310/2387] eta: 0:03:27 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1170 (0.3410) acc1: 99.7396 (84.5025) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2011 data: 0.1080 max mem: 3536 +Test: [1320/2387] eta: 0:03:25 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1208 (0.3394) acc1: 99.7396 (84.6131) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1945 data: 0.1007 max mem: 3536 +Test: [1330/2387] eta: 0:03:23 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1181 (0.3377) acc1: 99.7396 (84.7264) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1981 data: 0.1061 max mem: 3536 +Test: [1340/2387] eta: 0:03:21 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1210 (0.3370) acc1: 99.7396 (84.7783) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1966 data: 0.1053 max mem: 3536 +Test: [1350/2387] eta: 0:03:19 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0768 (0.3350) acc1: 100.0000 (84.8910) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0992 max mem: 3537 +Test: [1360/2387] eta: 0:03:17 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0696 (0.3330) acc1: 100.0000 (85.0020) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1876 data: 0.0959 max mem: 3537 +Test: [1370/2387] eta: 0:03:15 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0699 (0.3311) acc1: 100.0000 (85.1114) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1871 data: 0.0923 max mem: 3537 +Test: [1380/2387] eta: 0:03:13 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0677 (0.3292) acc1: 100.0000 (85.2192) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1897 data: 0.0915 max mem: 3537 +Test: [1390/2387] eta: 0:03:11 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0674 (0.3274) acc1: 100.0000 (85.3255) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1843 data: 0.0888 max mem: 3537 +Test: [1400/2387] eta: 0:03:09 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0622 (0.3254) acc1: 100.0000 (85.4302) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1697 data: 0.0777 max mem: 3537 +Test: [1410/2387] eta: 0:03:07 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0631 (0.3241) acc1: 100.0000 (85.5233) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1822 data: 0.0899 max mem: 3537 +Test: [1420/2387] eta: 0:03:05 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1437 (0.3231) acc1: 98.6979 (85.6030) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1947 data: 0.1037 max mem: 3537 +Test: [1430/2387] eta: 0:03:03 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1569 (0.3219) acc1: 98.6979 (85.6940) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1817 data: 0.0916 max mem: 3537 +Test: [1440/2387] eta: 0:03:01 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1620 (0.3208) acc1: 99.2188 (85.7886) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1751 data: 0.0844 max mem: 3537 +Test: [1450/2387] eta: 0:02:59 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1502 (0.3196) acc1: 99.7396 (85.8822) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1755 data: 0.0837 max mem: 3537 +Test: [1460/2387] eta: 0:02:57 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1240 (0.3183) acc1: 99.7396 (85.9753) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1853 data: 0.0932 max mem: 3537 +Test: [1470/2387] eta: 0:02:55 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1072 (0.3168) acc1: 100.0000 (86.0696) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1895 data: 0.0982 max mem: 3537 +Test: [1480/2387] eta: 0:02:53 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0997 (0.3154) acc1: 100.0000 (86.1605) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1840 data: 0.0914 max mem: 3537 +Test: [1490/2387] eta: 0:02:51 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1023 (0.3140) acc1: 100.0000 (86.2514) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1778 data: 0.0816 max mem: 3537 +Test: [1500/2387] eta: 0:02:49 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1087 (0.3127) acc1: 100.0000 (86.3412) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1817 data: 0.0864 max mem: 3537 +Test: [1510/2387] eta: 0:02:48 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1084 (0.3113) acc1: 100.0000 (86.4316) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1870 data: 0.0953 max mem: 3537 +Test: [1520/2387] eta: 0:02:46 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1128 (0.3101) acc1: 100.0000 (86.5165) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2325 data: 0.1413 max mem: 3537 +Test: [1530/2387] eta: 0:02:44 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1233 (0.3089) acc1: 99.7396 (86.6022) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1878 data: 0.0934 max mem: 3537 +Test: [1540/2387] eta: 0:02:42 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1158 (0.3076) acc1: 100.0000 (86.6878) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1465 data: 0.0481 max mem: 3537 +Test: [1550/2387] eta: 0:02:40 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1086 (0.3064) acc1: 100.0000 (86.7720) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1903 data: 0.0949 max mem: 3537 +Test: [1560/2387] eta: 0:02:38 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1103 (0.3051) acc1: 99.7396 (86.8550) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1806 data: 0.0869 max mem: 3537 +Test: [1570/2387] eta: 0:02:36 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1051 (0.3038) acc1: 100.0000 (86.9387) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1700 data: 0.0753 max mem: 3537 +Test: [1580/2387] eta: 0:02:34 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1092 (0.3028) acc1: 100.0000 (87.0147) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1750 data: 0.0817 max mem: 3538 +Test: [1590/2387] eta: 0:02:32 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1294 (0.3018) acc1: 99.4792 (87.0905) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1805 data: 0.0872 max mem: 3538 +Test: [1600/2387] eta: 0:02:30 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1268 (0.3009) acc1: 99.4792 (87.1529) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1577 data: 0.0656 max mem: 3538 +Test: [1610/2387] eta: 0:02:28 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1242 (0.2998) acc1: 99.4792 (87.2284) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1511 data: 0.0584 max mem: 3538 +Test: [1620/2387] eta: 0:02:26 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1323 (0.2989) acc1: 99.2188 (87.2932) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1648 data: 0.0710 max mem: 3538 +Test: [1630/2387] eta: 0:02:24 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1520 (0.2980) acc1: 98.1771 (87.3616) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1926 data: 0.0987 max mem: 3538 +Test: [1640/2387] eta: 0:02:22 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1405 (0.2970) acc1: 99.4792 (87.4332) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2099 data: 0.1153 max mem: 3538 +Test: [1650/2387] eta: 0:02:20 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1334 (0.2961) acc1: 99.4792 (87.5028) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1699 data: 0.0767 max mem: 3538 +Test: [1660/2387] eta: 0:02:18 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1334 (0.2952) acc1: 99.2188 (87.5679) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1759 data: 0.0837 max mem: 3538 +Test: [1670/2387] eta: 0:02:16 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1351 (0.2943) acc1: 99.2188 (87.6342) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1953 data: 0.1028 max mem: 3538 +Test: [1680/2387] eta: 0:02:14 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1440 (0.2935) acc1: 99.2188 (87.6954) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1896 data: 0.0985 max mem: 3538 +Test: [1690/2387] eta: 0:02:12 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1486 (0.2927) acc1: 98.9583 (87.7570) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2057 data: 0.1142 max mem: 3538 +Test: [1700/2387] eta: 0:02:10 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1340 (0.2917) acc1: 99.2188 (87.8258) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1973 data: 0.1057 max mem: 3538 +Test: [1710/2387] eta: 0:02:09 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0861 (0.2904) acc1: 100.0000 (87.8966) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1850 data: 0.0949 max mem: 3538 +Test: [1720/2387] eta: 0:02:07 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0747 (0.2892) acc1: 100.0000 (87.9662) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1878 data: 0.0970 max mem: 3538 +Test: [1730/2387] eta: 0:02:05 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0711 (0.2879) acc1: 100.0000 (88.0350) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1868 data: 0.0917 max mem: 3538 +Test: [1740/2387] eta: 0:02:03 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0723 (0.2867) acc1: 100.0000 (88.1027) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1889 data: 0.0930 max mem: 3538 +Test: [1750/2387] eta: 0:02:01 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0738 (0.2855) acc1: 100.0000 (88.1705) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1927 data: 0.1005 max mem: 3538 +Test: [1760/2387] eta: 0:01:59 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0700 (0.2843) acc1: 100.0000 (88.2376) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1862 data: 0.0934 max mem: 3538 +Test: [1770/2387] eta: 0:01:57 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0801 (0.2832) acc1: 100.0000 (88.3023) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1862 data: 0.0916 max mem: 3538 +Test: [1780/2387] eta: 0:01:55 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0858 (0.2821) acc1: 100.0000 (88.3677) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1910 data: 0.0967 max mem: 3538 +Test: [1790/2387] eta: 0:01:53 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0801 (0.2810) acc1: 100.0000 (88.4323) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1889 data: 0.0958 max mem: 3538 +Test: [1800/2387] eta: 0:01:51 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0801 (0.2799) acc1: 100.0000 (88.4960) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1848 data: 0.0921 max mem: 3538 +Test: [1810/2387] eta: 0:01:49 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0802 (0.2787) acc1: 100.0000 (88.5594) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1813 data: 0.0896 max mem: 3539 +Test: [1820/2387] eta: 0:01:47 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0727 (0.2776) acc1: 100.0000 (88.6220) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1829 data: 0.0903 max mem: 3539 +Test: [1830/2387] eta: 0:01:46 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0748 (0.2766) acc1: 100.0000 (88.6842) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1871 data: 0.0903 max mem: 3539 +Test: [1840/2387] eta: 0:01:44 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1182 (0.2758) acc1: 100.0000 (88.7456) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1998 data: 0.1046 max mem: 3539 +Test: [1850/2387] eta: 0:01:42 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1200 (0.2749) acc1: 100.0000 (88.8063) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2012 data: 0.1084 max mem: 3539 +Test: [1860/2387] eta: 0:01:40 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1199 (0.2741) acc1: 100.0000 (88.8665) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1896 data: 0.0963 max mem: 3539 +Test: [1870/2387] eta: 0:01:38 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1187 (0.2733) acc1: 100.0000 (88.9260) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1845 data: 0.0928 max mem: 3539 +Test: [1880/2387] eta: 0:01:36 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1186 (0.2724) acc1: 100.0000 (88.9847) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1825 data: 0.0879 max mem: 3539 +Test: [1890/2387] eta: 0:01:34 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1189 (0.2716) acc1: 100.0000 (89.0429) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1857 data: 0.0909 max mem: 3539 +Test: [1900/2387] eta: 0:01:32 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1199 (0.2708) acc1: 100.0000 (89.1006) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1873 data: 0.0960 max mem: 3539 +Test: [1910/2387] eta: 0:01:30 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1225 (0.2701) acc1: 100.0000 (89.1572) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1844 data: 0.0930 max mem: 3539 +Test: [1920/2387] eta: 0:01:28 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1249 (0.2696) acc1: 100.0000 (89.1925) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1803 data: 0.0879 max mem: 3539 +Test: [1930/2387] eta: 0:01:26 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1213 (0.2687) acc1: 100.0000 (89.2483) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0878 max mem: 3539 +Test: [1940/2387] eta: 0:01:25 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1147 (0.2682) acc1: 100.0000 (89.2982) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1845 data: 0.0912 max mem: 3539 +Test: [1950/2387] eta: 0:01:23 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1201 (0.2674) acc1: 100.0000 (89.3524) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1803 data: 0.0874 max mem: 3539 +Test: [1960/2387] eta: 0:01:21 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1236 (0.2670) acc1: 100.0000 (89.3928) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1726 data: 0.0815 max mem: 3539 +Test: [1970/2387] eta: 0:01:19 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1774 (0.2667) acc1: 98.9583 (89.4170) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1733 data: 0.0798 max mem: 3539 +Test: [1980/2387] eta: 0:01:17 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1513 (0.2661) acc1: 99.4792 (89.4624) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1833 data: 0.0890 max mem: 3539 +Test: [1990/2387] eta: 0:01:15 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1476 (0.2656) acc1: 99.4792 (89.5119) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1827 data: 0.0904 max mem: 3539 +Test: [2000/2387] eta: 0:01:13 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1495 (0.2650) acc1: 99.2188 (89.5612) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1708 data: 0.0801 max mem: 3539 +Test: [2010/2387] eta: 0:01:11 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1521 (0.2644) acc1: 99.2188 (89.6090) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1784 data: 0.0861 max mem: 3539 +Test: [2020/2387] eta: 0:01:09 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1521 (0.2639) acc1: 99.2188 (89.6568) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1793 data: 0.0864 max mem: 3539 +Test: [2030/2387] eta: 0:01:07 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1501 (0.2633) acc1: 99.2188 (89.7036) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1759 data: 0.0822 max mem: 3540 +Test: [2040/2387] eta: 0:01:05 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1488 (0.2628) acc1: 99.2188 (89.7509) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1990 data: 0.1059 max mem: 3540 +Test: [2050/2387] eta: 0:01:03 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1466 (0.2622) acc1: 99.4792 (89.7987) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1930 data: 0.1017 max mem: 3540 +Test: [2060/2387] eta: 0:01:01 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1449 (0.2616) acc1: 99.4792 (89.8440) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1660 data: 0.0748 max mem: 3540 +Test: [2070/2387] eta: 0:01:00 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1414 (0.2610) acc1: 99.2188 (89.8909) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1719 data: 0.0815 max mem: 3540 +Test: [2080/2387] eta: 0:00:58 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1389 (0.2604) acc1: 99.4792 (89.9371) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1812 data: 0.0899 max mem: 3540 +Test: [2090/2387] eta: 0:00:56 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1394 (0.2599) acc1: 99.4792 (89.9831) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1885 data: 0.0955 max mem: 3540 +Test: [2100/2387] eta: 0:00:54 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1401 (0.2593) acc1: 99.4792 (90.0283) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1929 data: 0.0970 max mem: 3540 +Test: [2110/2387] eta: 0:00:52 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1405 (0.2587) acc1: 99.4792 (90.0730) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2012 data: 0.1066 max mem: 3540 +Test: [2120/2387] eta: 0:00:50 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1453 (0.2583) acc1: 99.4792 (90.1125) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1803 data: 0.0882 max mem: 3540 +Test: [2130/2387] eta: 0:00:48 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1464 (0.2578) acc1: 99.7396 (90.1568) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1639 data: 0.0715 max mem: 3540 +Test: [2140/2387] eta: 0:00:46 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1443 (0.2574) acc1: 99.7396 (90.1853) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1958 data: 0.1051 max mem: 3540 +Test: [2150/2387] eta: 0:00:44 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1434 (0.2569) acc1: 99.7396 (90.2291) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1910 data: 0.1010 max mem: 3540 +Test: [2160/2387] eta: 0:00:42 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1468 (0.2564) acc1: 99.4792 (90.2676) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1805 data: 0.0890 max mem: 3540 +Test: [2170/2387] eta: 0:00:41 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1563 (0.2560) acc1: 99.2188 (90.3070) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1863 data: 0.0941 max mem: 3540 +Test: [2180/2387] eta: 0:00:39 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1740 (0.2569) acc1: 98.1771 (90.2564) acc5: 100.0000 (100.0000) acc1_neg: 53.646 acc5_neg: 100.000 time: 0.1790 data: 0.0885 max mem: 3540 +Test: [2190/2387] eta: 0:00:37 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2570 (0.2573) acc1: 90.6250 (90.2373) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1774 data: 0.0860 max mem: 3540 +Test: [2200/2387] eta: 0:00:35 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1375 (0.2567) acc1: 99.2188 (90.2803) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1700 data: 0.0780 max mem: 3540 +Test: [2210/2387] eta: 0:00:33 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0991 (0.2560) acc1: 100.0000 (90.3241) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1712 data: 0.0779 max mem: 3540 +Test: [2220/2387] eta: 0:00:31 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0996 (0.2553) acc1: 100.0000 (90.3662) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1697 data: 0.0758 max mem: 3540 +Test: [2230/2387] eta: 0:00:29 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1033 (0.2546) acc1: 100.0000 (90.4077) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1506 data: 0.0589 max mem: 3540 +Test: [2240/2387] eta: 0:00:27 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1039 (0.2540) acc1: 100.0000 (90.4484) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1491 data: 0.0582 max mem: 3540 +Test: [2250/2387] eta: 0:00:25 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1006 (0.2533) acc1: 100.0000 (90.4903) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1585 data: 0.0683 max mem: 3540 +Test: [2260/2387] eta: 0:00:23 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1063 (0.2527) acc1: 100.0000 (90.5318) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1603 data: 0.0700 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1223 (0.2522) acc1: 100.0000 (90.5726) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1631 data: 0.0723 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1301 (0.2518) acc1: 100.0000 (90.6006) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1635 data: 0.0723 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1282 (0.2512) acc1: 100.0000 (90.6413) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1616 data: 0.0706 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1282 (0.2507) acc1: 100.0000 (90.6785) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1665 data: 0.0750 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1322 (0.2503) acc1: 100.0000 (90.7170) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1660 data: 0.0747 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1395 (0.2499) acc1: 99.7396 (90.7491) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1608 data: 0.0709 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1643 (0.2497) acc1: 99.7396 (90.7801) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1486 data: 0.0589 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1642 (0.2492) acc1: 99.7396 (90.8150) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1334 data: 0.0438 max mem: 3541 +Test: [2350/2387] eta: 0:00:06 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1531 (0.2490) acc1: 99.7396 (90.8357) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1277 data: 0.0381 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1662 (0.2487) acc1: 98.9583 (90.8659) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1272 data: 0.0377 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1927 (0.2486) acc1: 97.9167 (90.8901) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1238 data: 0.0343 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1762 (0.2483) acc1: 98.9583 (90.9230) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1200 data: 0.0292 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2032 (0.2482) acc1: 97.6562 (90.9324) acc5: 100.0000 (100.0000) acc1_neg: 90.698 acc5_neg: 100.000 time: 0.1135 data: 0.0248 max mem: 3541 +Test: Total time: 0:07:24 (0.1861 s / it) +* Overall Acc@1 90.932 Acc@5 100.000 + target == real Acc@1 10.728 Acc@5 100.000 + target == fake Acc@1 96.565 Acc@5 100.000 + loss 0.248 + PR-AUC 0.9871 ROC-AUC 0.8435 EER 22.48% (thr=0.8506) +Accuracy of the network on the 916267 test images: 90.9% +[rank0]:[W726 06:00:55.891121715 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 06:01:14.302000 3171855 torch/distributed/run.py:766] +W0726 06:01:14.302000 3171855 torch/distributed/run.py:766] ***************************************** +W0726 06:01:14.302000 3171855 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:01:14.302000 3171855 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:01:41.466428691 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:01:42.673368206 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:01:42.749779980 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:01:42.845434667 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:01:42.932221243 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:01:42.988510358 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 06:01:42.014250039 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:01:42.048685606 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n16_seed9_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed9_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:03:08 acc1_pos: 22.396 acc5_pos: 100.000 loss: 1.1914 (1.1914) acc1: 22.3958 (22.3958) acc5: 100.0000 (100.0000) time: 7.3280 data: 5.7115 max mem: 3531 +Test: [ 10/517] eta: 0:06:19 acc1_pos: 7.812 acc5_pos: 100.000 loss: 1.2718 (1.2665) acc1: 14.0625 (16.9034) acc5: 100.0000 (100.0000) time: 0.7481 data: 0.5196 max mem: 3531 +Test: [ 20/517] eta: 0:03:38 acc1_pos: 5.469 acc5_pos: 100.000 loss: 1.3039 (1.3113) acc1: 12.2396 (15.8730) acc5: 100.0000 (100.0000) time: 0.0950 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:02:58 acc1_pos: 9.115 acc5_pos: 100.000 loss: 1.3672 (1.3263) acc1: 10.1562 (14.5161) acc5: 100.0000 (100.0000) time: 0.1566 data: 0.0613 max mem: 3531 +Test: [ 40/517] eta: 0:02:40 acc1_pos: 7.812 acc5_pos: 100.000 loss: 1.3427 (1.3234) acc1: 11.1979 (13.7259) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1369 max mem: 3531 +Test: [ 50/517] eta: 0:02:29 acc1_pos: 28.440 acc5_pos: 100.000 loss: 1.2054 (1.2646) acc1: 11.4583 (18.0351) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2483 data: 0.1530 max mem: 3531 +Test: [ 60/517] eta: 0:02:24 acc1_pos: 28.440 acc5_pos: 100.000 loss: 0.3455 (1.1111) acc1: 86.9792 (30.4303) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2749 data: 0.1818 max mem: 3531 +Test: [ 70/517] eta: 0:02:16 acc1_pos: 28.440 acc5_pos: 100.000 loss: 0.3199 (1.0023) acc1: 93.7500 (39.0442) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.2649 data: 0.1739 max mem: 3531 +Test: [ 80/517] eta: 0:02:09 acc1_pos: 28.440 acc5_pos: 100.000 loss: 0.3385 (0.9237) acc1: 91.6667 (45.3511) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.2349 data: 0.1404 max mem: 3531 +Test: [ 90/517] eta: 0:02:04 acc1_pos: 28.440 acc5_pos: 100.000 loss: 0.2850 (0.8500) acc1: 97.9167 (51.1848) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2454 data: 0.1516 max mem: 3531 +Test: [100/517] eta: 0:02:00 acc1_pos: 28.440 acc5_pos: 100.000 loss: 0.2458 (0.7908) acc1: 98.1771 (55.7240) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.2599 data: 0.1670 max mem: 3531 +Test: [110/517] eta: 0:01:57 acc1_pos: 28.440 acc5_pos: 100.000 loss: 0.2458 (0.7448) acc1: 94.2708 (59.0114) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2744 data: 0.1792 max mem: 3531 +Test: [120/517] eta: 0:01:54 acc1_pos: 28.440 acc5_pos: 100.000 loss: 0.2591 (0.7057) acc1: 94.2708 (62.2202) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2799 data: 0.1850 max mem: 3531 +Test: [130/517] eta: 0:01:49 acc1_pos: 28.440 acc5_pos: 100.000 loss: 0.2524 (0.6708) acc1: 99.4792 (65.0525) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2599 data: 0.1653 max mem: 3531 +Test: [140/517] eta: 0:01:45 acc1_pos: 28.440 acc5_pos: 100.000 loss: 0.2422 (0.6402) acc1: 99.2188 (67.4258) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2401 data: 0.1466 max mem: 3531 +Test: [150/517] eta: 0:01:42 acc1_pos: 28.440 acc5_pos: 100.000 loss: 0.2244 (0.6125) acc1: 99.2188 (69.5123) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2510 data: 0.1592 max mem: 3531 +Test: [160/517] eta: 0:01:39 acc1_pos: 28.440 acc5_pos: 100.000 loss: 0.2280 (0.5892) acc1: 99.2188 (71.3396) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2597 data: 0.1664 max mem: 3531 +Test: [170/517] eta: 0:01:36 acc1_pos: 28.440 acc5_pos: 100.000 loss: 0.2287 (0.5677) acc1: 99.2188 (72.9410) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2588 data: 0.1637 max mem: 3531 +Test: [180/517] eta: 0:01:32 acc1_pos: 28.440 acc5_pos: 100.000 loss: 0.2162 (0.5478) acc1: 98.6979 (74.3540) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2549 data: 0.1603 max mem: 3531 +Test: [190/517] eta: 0:01:29 acc1_pos: 28.440 acc5_pos: 100.000 loss: 0.1854 (0.5285) acc1: 99.2188 (75.6831) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2399 data: 0.1456 max mem: 3531 +Test: [200/517] eta: 0:01:26 acc1_pos: 28.440 acc5_pos: 100.000 loss: 0.1763 (0.5110) acc1: 100.0000 (76.8786) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2449 data: 0.1490 max mem: 3531 +Test: [210/517] eta: 0:01:23 acc1_pos: 28.440 acc5_pos: 100.000 loss: 0.1947 (0.4999) acc1: 98.9583 (77.8510) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2599 data: 0.1644 max mem: 3532 +Test: [220/517] eta: 0:01:20 acc1_pos: 28.440 acc5_pos: 100.000 loss: 0.2788 (0.4904) acc1: 97.6562 (78.6812) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2612 data: 0.1681 max mem: 3532 +Test: [230/517] eta: 0:01:17 acc1_pos: 28.440 acc5_pos: 100.000 loss: 0.2894 (0.4823) acc1: 97.1354 (79.5116) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2599 data: 0.1674 max mem: 3532 +Test: [240/517] eta: 0:01:14 acc1_pos: 28.440 acc5_pos: 100.000 loss: 0.3137 (0.4769) acc1: 97.1354 (80.1608) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.2588 data: 0.1656 max mem: 3532 +Test: [250/517] eta: 0:01:12 acc1_pos: 28.440 acc5_pos: 100.000 loss: 0.3104 (0.4692) acc1: 97.9167 (80.8412) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2649 data: 0.1700 max mem: 3532 +Test: [260/517] eta: 0:01:09 acc1_pos: 28.440 acc5_pos: 100.000 loss: 0.2446 (0.4608) acc1: 99.4792 (81.5543) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2650 data: 0.1683 max mem: 3532 +Test: [270/517] eta: 0:01:06 acc1_pos: 28.440 acc5_pos: 100.000 loss: 0.2595 (0.4539) acc1: 99.4792 (82.2071) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2599 data: 0.1632 max mem: 3532 +Test: [280/517] eta: 0:01:03 acc1_pos: 28.440 acc5_pos: 100.000 loss: 0.2750 (0.4474) acc1: 98.9583 (82.7856) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2646 data: 0.1679 max mem: 3532 +Test: [290/517] eta: 0:01:01 acc1_pos: 28.440 acc5_pos: 100.000 loss: 0.2659 (0.4412) acc1: 98.9583 (83.3360) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2598 data: 0.1623 max mem: 3532 +Test: [300/517] eta: 0:00:58 acc1_pos: 28.440 acc5_pos: 100.000 loss: 0.2625 (0.4348) acc1: 99.2188 (83.8351) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2555 data: 0.1594 max mem: 3532 +Test: [310/517] eta: 0:00:55 acc1_pos: 28.440 acc5_pos: 100.000 loss: 0.2343 (0.4283) acc1: 99.4792 (84.3474) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2649 data: 0.1696 max mem: 3532 +Test: [320/517] eta: 0:00:52 acc1_pos: 28.440 acc5_pos: 100.000 loss: 0.2343 (0.4224) acc1: 100.0000 (84.8244) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2693 data: 0.1735 max mem: 3532 +Test: [330/517] eta: 0:00:50 acc1_pos: 28.440 acc5_pos: 100.000 loss: 0.2362 (0.4163) acc1: 100.0000 (85.2821) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2605 data: 0.1653 max mem: 3532 +Test: [340/517] eta: 0:00:47 acc1_pos: 28.440 acc5_pos: 100.000 loss: 0.2140 (0.4110) acc1: 100.0000 (85.6549) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2551 data: 0.1604 max mem: 3532 +Test: [350/517] eta: 0:00:44 acc1_pos: 28.440 acc5_pos: 100.000 loss: 0.2518 (0.4071) acc1: 98.4375 (86.0080) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2560 data: 0.1612 max mem: 3532 +Test: [360/517] eta: 0:00:41 acc1_pos: 28.440 acc5_pos: 100.000 loss: 0.2562 (0.4030) acc1: 98.1771 (86.3400) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2447 data: 0.1501 max mem: 3532 +Test: [370/517] eta: 0:00:39 acc1_pos: 28.440 acc5_pos: 100.000 loss: 0.2334 (0.3980) acc1: 99.4792 (86.7019) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2290 data: 0.1339 max mem: 3532 +Test: [380/517] eta: 0:00:36 acc1_pos: 28.440 acc5_pos: 100.000 loss: 0.2206 (0.3936) acc1: 99.7396 (87.0079) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2299 data: 0.1346 max mem: 3532 +Test: [390/517] eta: 0:00:33 acc1_pos: 28.440 acc5_pos: 100.000 loss: 0.2506 (0.3906) acc1: 97.1354 (87.2129) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2346 data: 0.1399 max mem: 3532 +Test: [400/517] eta: 0:00:30 acc1_pos: 28.440 acc5_pos: 100.000 loss: 0.2640 (0.3870) acc1: 96.6146 (87.4799) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2299 data: 0.1358 max mem: 3532 +Test: [410/517] eta: 0:00:28 acc1_pos: 28.440 acc5_pos: 100.000 loss: 0.2488 (0.3838) acc1: 98.6979 (87.7547) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2393 data: 0.1433 max mem: 3532 +Test: [420/517] eta: 0:00:25 acc1_pos: 28.440 acc5_pos: 100.000 loss: 0.2577 (0.3814) acc1: 98.9583 (87.9936) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2399 data: 0.1411 max mem: 3532 +Test: [430/517] eta: 0:00:22 acc1_pos: 28.440 acc5_pos: 100.000 loss: 0.2914 (0.3802) acc1: 96.6146 (88.1350) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.2349 data: 0.1374 max mem: 3532 +Test: [440/517] eta: 0:00:20 acc1_pos: 28.440 acc5_pos: 100.000 loss: 0.3165 (0.3783) acc1: 93.7500 (88.2730) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2351 data: 0.1397 max mem: 3533 +Test: [450/517] eta: 0:00:17 acc1_pos: 28.440 acc5_pos: 100.000 loss: 0.2833 (0.3763) acc1: 96.8750 (88.4822) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2349 data: 0.1391 max mem: 3533 +Test: [460/517] eta: 0:00:14 acc1_pos: 28.440 acc5_pos: 100.000 loss: 0.3097 (0.3754) acc1: 95.8333 (88.5688) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.2403 data: 0.1443 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 28.440 acc5_pos: 100.000 loss: 0.2442 (0.3725) acc1: 98.1771 (88.7728) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2459 data: 0.1507 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 28.440 acc5_pos: 100.000 loss: 0.2302 (0.3697) acc1: 99.4792 (88.9986) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2493 data: 0.1529 max mem: 3533 +Test: [490/517] eta: 0:00:06 acc1_pos: 28.440 acc5_pos: 100.000 loss: 0.2631 (0.3677) acc1: 99.4792 (89.1834) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2268 data: 0.1307 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 28.440 acc5_pos: 100.000 loss: 0.2625 (0.3653) acc1: 99.4792 (89.3926) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1919 data: 0.0998 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 28.440 acc5_pos: 100.000 loss: 0.2546 (0.3635) acc1: 99.7396 (89.5670) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1800 data: 0.0904 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 28.440 acc5_pos: 100.000 loss: 0.2633 (0.3625) acc1: 99.4792 (89.6602) acc5: 100.0000 (100.0000) acc1_neg: 96.113 acc5_neg: 100.000 time: 0.1859 data: 0.0867 max mem: 3533 +Test: Total time: 0:02:11 (0.2540 s / it) +* Overall Acc@1 89.660 Acc@5 100.000 + target == real Acc@1 14.375 Acc@5 100.000 + target == fake Acc@1 97.470 Acc@5 100.000 + loss 0.362 + PR-AUC 0.9371 ROC-AUC 0.6580 EER 38.62% (thr=0.7749) +Accuracy of the network on the 198427 test images: 89.7% +[rank0]:[W726 06:04:03.076380126 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 06:04:22.618000 3172431 torch/distributed/run.py:766] +W0726 06:04:22.618000 3172431 torch/distributed/run.py:766] ***************************************** +W0726 06:04:22.618000 3172431 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:04:22.618000 3172431 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:04:49.806244518 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:04:49.247744303 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:04:49.435322250 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:04:50.607587147 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:04:50.665670283 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W726 06:04:50.665672744 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:04:50.676377453 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 06:04:50.689129170 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n16_seed9_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed9_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:31:14 acc1_pos: 28.125 acc5_pos: 100.000 loss: 1.0864 (1.0864) acc1: 28.1250 (28.1250) acc5: 100.0000 (100.0000) time: 4.8186 data: 2.8811 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/389] eta: 0:03:17 acc1_pos: 57.292 acc5_pos: 100.000 loss: 1.0580 (0.9795) acc1: 29.4271 (32.1733) acc5: 100.0000 (100.0000) time: 0.5198 data: 0.2623 max mem: 3531 +Test: [ 20/389] eta: 0:02:03 acc1_pos: 1.823 acc5_pos: 100.000 loss: 1.0156 (1.0999) acc1: 27.3438 (27.9638) acc5: 100.0000 (100.0000) time: 0.1098 data: 0.0156 max mem: 3531 +Test: [ 30/389] eta: 0:01:42 acc1_pos: 2.604 acc5_pos: 100.000 loss: 1.1162 (1.1217) acc1: 3.9062 (21.2282) acc5: 100.0000 (100.0000) time: 0.1548 data: 0.0517 max mem: 3531 +Test: [ 40/389] eta: 0:01:35 acc1_pos: 1.562 acc5_pos: 100.000 loss: 1.1126 (1.1165) acc1: 4.1667 (20.7571) acc5: 100.0000 (100.0000) time: 0.2113 data: 0.1098 max mem: 3531 +Test: [ 50/389] eta: 0:01:29 acc1_pos: 80.469 acc5_pos: 100.000 loss: 0.9152 (1.0518) acc1: 27.3438 (25.8170) acc5: 100.0000 (100.0000) time: 0.2304 data: 0.1368 max mem: 3531 +Test: [ 60/389] eta: 0:01:23 acc1_pos: 50.000 acc5_pos: 100.000 loss: 0.7485 (1.0414) acc1: 43.4896 (28.9148) acc5: 100.0000 (100.0000) time: 0.2093 data: 0.1178 max mem: 3531 +Test: [ 70/389] eta: 0:01:19 acc1_pos: 62.240 acc5_pos: 100.000 loss: 0.7485 (0.9943) acc1: 55.4688 (32.5337) acc5: 100.0000 (100.0000) time: 0.2161 data: 0.1245 max mem: 3531 +Test: [ 80/389] eta: 0:01:15 acc1_pos: 59.635 acc5_pos: 100.000 loss: 0.7660 (0.9872) acc1: 45.8333 (32.9475) acc5: 100.0000 (100.0000) time: 0.2245 data: 0.1324 max mem: 3531 +Test: [ 90/389] eta: 0:01:12 acc1_pos: 26.562 acc5_pos: 100.000 loss: 0.7883 (0.9556) acc1: 44.2708 (35.4024) acc5: 100.0000 (100.0000) time: 0.2155 data: 0.1241 max mem: 3531 +Test: [100/389] eta: 0:01:09 acc1_pos: 10.677 acc5_pos: 100.000 loss: 0.9746 (0.9720) acc1: 29.9479 (33.6969) acc5: 100.0000 (100.0000) time: 0.2147 data: 0.1245 max mem: 3531 +Test: [110/389] eta: 0:01:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.1278 (0.9806) acc1: 10.6771 (34.1803) acc5: 100.0000 (100.0000) time: 0.2000 data: 0.1097 max mem: 3531 +Test: [120/389] eta: 0:01:01 acc1_pos: 42.448 acc5_pos: 100.000 loss: 0.9391 (0.9678) acc1: 36.4583 (35.2639) acc5: 100.0000 (100.0000) time: 0.1759 data: 0.0857 max mem: 3531 +Test: [130/389] eta: 0:00:57 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.8162 (0.9407) acc1: 45.3125 (37.6332) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1588 data: 0.0682 max mem: 3531 +Test: [140/389] eta: 0:00:54 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.3113 (0.8998) acc1: 89.5833 (40.9556) acc5: 100.0000 (100.0000) acc1_neg: 29.427 acc5_neg: 100.000 time: 0.1578 data: 0.0656 max mem: 3531 +Test: [150/389] eta: 0:00:51 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.3172 (0.8779) acc1: 97.9167 (43.2309) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1505 data: 0.0587 max mem: 3531 +Test: [160/389] eta: 0:00:48 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.3920 (0.8485) acc1: 91.4062 (45.8042) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1459 data: 0.0535 max mem: 3531 +Test: [170/389] eta: 0:00:45 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.2455 (0.8152) acc1: 99.7396 (48.9126) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1544 data: 0.0598 max mem: 3531 +Test: [180/389] eta: 0:00:42 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.2459 (0.7887) acc1: 99.4792 (51.3611) acc5: 100.0000 (100.0000) acc1_neg: 66.406 acc5_neg: 100.000 time: 0.1532 data: 0.0601 max mem: 3531 +Test: [190/389] eta: 0:00:39 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.2989 (0.7723) acc1: 96.3542 (52.8782) acc5: 100.0000 (100.0000) acc1_neg: 22.917 acc5_neg: 100.000 time: 0.1504 data: 0.0597 max mem: 3531 +Test: [200/389] eta: 0:00:37 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.2731 (0.7552) acc1: 97.1354 (54.3027) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1546 data: 0.0623 max mem: 3531 +Test: [210/389] eta: 0:00:35 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.2688 (0.7309) acc1: 99.7396 (56.4351) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1619 data: 0.0667 max mem: 3532 +Test: [220/389] eta: 0:00:33 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.1763 (0.7071) acc1: 100.0000 (58.2956) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1646 data: 0.0708 max mem: 3532 +Test: [230/389] eta: 0:00:31 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.2312 (0.6947) acc1: 98.6979 (59.5452) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1738 data: 0.0811 max mem: 3532 +Test: [240/389] eta: 0:00:28 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.4052 (0.6864) acc1: 87.5000 (60.4145) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1751 data: 0.0819 max mem: 3532 +Test: [250/389] eta: 0:00:26 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.3391 (0.6690) acc1: 96.0938 (61.9211) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1714 data: 0.0785 max mem: 3532 +Test: [260/389] eta: 0:00:24 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.2658 (0.6567) acc1: 99.4792 (62.9300) acc5: 100.0000 (100.0000) acc1_neg: 70.833 acc5_neg: 100.000 time: 0.1667 data: 0.0749 max mem: 3532 +Test: [270/389] eta: 0:00:22 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.3229 (0.6488) acc1: 97.9167 (63.6358) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1703 data: 0.0795 max mem: 3532 +Test: [280/389] eta: 0:00:20 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.3245 (0.6396) acc1: 84.1146 (64.3618) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1828 data: 0.0911 max mem: 3532 +Test: [290/389] eta: 0:00:18 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.3753 (0.6330) acc1: 90.3646 (65.0192) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1779 data: 0.0849 max mem: 3532 +Test: [300/389] eta: 0:00:16 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.3753 (0.6243) acc1: 96.6146 (66.1000) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1682 data: 0.0735 max mem: 3532 +Test: [310/389] eta: 0:00:14 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.3817 (0.6185) acc1: 92.4479 (66.7152) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1703 data: 0.0774 max mem: 3532 +Test: [320/389] eta: 0:00:13 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.3647 (0.6092) acc1: 92.4479 (67.6012) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1771 data: 0.0864 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.2990 (0.6008) acc1: 100.0000 (68.5336) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.1806 data: 0.0891 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.2990 (0.5911) acc1: 100.0000 (69.3564) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1776 data: 0.0860 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.3369 (0.6016) acc1: 96.3542 (68.7359) acc5: 100.0000 (100.0000) acc1_neg: 0.260 acc5_neg: 100.000 time: 0.1787 data: 0.0864 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.9060 (0.6153) acc1: 43.4896 (67.7884) acc5: 100.0000 (100.0000) acc1_neg: 43.750 acc5_neg: 100.000 time: 0.1879 data: 0.0910 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.8777 (0.6114) acc1: 46.3542 (68.0572) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1675 data: 0.0720 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.4089 (0.6050) acc1: 88.2812 (68.5757) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1427 data: 0.0519 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.3572 (0.5995) acc1: 94.7917 (69.0810) acc5: 100.0000 (100.0000) acc1_neg: 82.787 acc5_neg: 100.000 time: 0.1370 data: 0.0433 max mem: 3532 +Test: Total time: 0:01:12 (0.1854 s / it) +* Overall Acc@1 69.081 Acc@5 100.000 + target == real Acc@1 34.890 Acc@5 100.000 + target == fake Acc@1 84.974 Acc@5 100.000 + loss 0.599 + PR-AUC 0.7983 ROC-AUC 0.6813 EER 35.71% (thr=0.6909) +Accuracy of the network on the 149358 test images: 69.1% +[rank0]:[W726 06:06:11.999712039 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 06:06:28.918000 3172990 torch/distributed/run.py:766] +W0726 06:06:28.918000 3172990 torch/distributed/run.py:766] ***************************************** +W0726 06:06:28.918000 3172990 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:06:28.918000 3172990 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:06:56.483132708 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:06:56.553860812 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:06:57.779694100 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 06:06:57.785320037 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:06:57.817210428 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:06:57.228690170 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:06:57.270351513 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:06:57.301312315 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n16_seed9_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed9_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:29:55 acc1_pos: 1.302 acc5_pos: 100.000 loss: 2.0118 (2.0118) acc1: 1.3021 (1.3021) acc5: 100.0000 (100.0000) time: 5.3928 data: 4.0243 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/333] eta: 0:03:04 acc1_pos: 20.312 acc5_pos: 100.000 loss: 1.7642 (1.6570) acc1: 7.5521 (11.7424) acc5: 100.0000 (100.0000) time: 0.5720 data: 0.3663 max mem: 3531 +Test: [ 20/333] eta: 0:01:50 acc1_pos: 30.469 acc5_pos: 100.000 loss: 1.3964 (1.5102) acc1: 16.4062 (16.7783) acc5: 100.0000 (100.0000) time: 0.1019 data: 0.0114 max mem: 3531 +Test: [ 30/333] eta: 0:01:26 acc1_pos: 45.052 acc5_pos: 100.000 loss: 1.2320 (1.4042) acc1: 24.7396 (20.0941) acc5: 100.0000 (100.0000) time: 0.1261 data: 0.0329 max mem: 3531 +Test: [ 40/333] eta: 0:01:20 acc1_pos: 32.812 acc5_pos: 100.000 loss: 1.1163 (1.3346) acc1: 26.3021 (22.5038) acc5: 100.0000 (100.0000) time: 0.1922 data: 0.0963 max mem: 3531 +Test: [ 50/333] eta: 0:01:16 acc1_pos: 31.510 acc5_pos: 100.000 loss: 1.1031 (1.3002) acc1: 28.9062 (23.2435) acc5: 100.0000 (100.0000) time: 0.2481 data: 0.1509 max mem: 3531 +Test: [ 60/333] eta: 0:01:09 acc1_pos: 18.490 acc5_pos: 100.000 loss: 1.1455 (1.2771) acc1: 26.0417 (23.6467) acc5: 100.0000 (100.0000) time: 0.2108 data: 0.1162 max mem: 3531 +Test: [ 70/333] eta: 0:01:03 acc1_pos: 23.698 acc5_pos: 100.000 loss: 1.1516 (1.2598) acc1: 26.0417 (24.3068) acc5: 100.0000 (100.0000) time: 0.1749 data: 0.0770 max mem: 3531 +Test: [ 80/333] eta: 0:01:00 acc1_pos: 28.906 acc5_pos: 100.000 loss: 1.1314 (1.2419) acc1: 29.1667 (24.9293) acc5: 100.0000 (100.0000) time: 0.1892 data: 0.0924 max mem: 3531 +Test: [ 90/333] eta: 0:00:56 acc1_pos: 39.062 acc5_pos: 100.000 loss: 1.0577 (1.2167) acc1: 30.7292 (26.0960) acc5: 100.0000 (100.0000) time: 0.2001 data: 0.1081 max mem: 3531 +Test: [100/333] eta: 0:00:53 acc1_pos: 35.938 acc5_pos: 100.000 loss: 0.9797 (1.1964) acc1: 36.9792 (27.2123) acc5: 100.0000 (100.0000) time: 0.1897 data: 0.0973 max mem: 3531 +Test: [110/333] eta: 0:00:50 acc1_pos: 32.812 acc5_pos: 100.000 loss: 1.0169 (1.1813) acc1: 36.4583 (27.9866) acc5: 100.0000 (100.0000) time: 0.1911 data: 0.0985 max mem: 3531 +Test: [120/333] eta: 0:00:47 acc1_pos: 39.323 acc5_pos: 100.000 loss: 1.0059 (1.1622) acc1: 35.9375 (28.9945) acc5: 100.0000 (100.0000) time: 0.2003 data: 0.1062 max mem: 3531 +Test: [130/333] eta: 0:00:44 acc1_pos: 55.208 acc5_pos: 100.000 loss: 0.9061 (1.1453) acc1: 39.3229 (29.9181) acc5: 100.0000 (100.0000) time: 0.1954 data: 0.1022 max mem: 3531 +Test: [140/333] eta: 0:00:42 acc1_pos: 45.312 acc5_pos: 100.000 loss: 0.8905 (1.1300) acc1: 42.9688 (30.7864) acc5: 100.0000 (100.0000) time: 0.1969 data: 0.1025 max mem: 3531 +Test: [150/333] eta: 0:00:39 acc1_pos: 35.677 acc5_pos: 100.000 loss: 0.9380 (1.1211) acc1: 38.0208 (31.1948) acc5: 100.0000 (100.0000) time: 0.1982 data: 0.1050 max mem: 3531 +Test: [160/333] eta: 0:00:37 acc1_pos: 28.618 acc5_pos: 100.000 loss: 0.9136 (1.1002) acc1: 39.3229 (32.5262) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.1902 data: 0.0982 max mem: 3531 +Test: [170/333] eta: 0:00:35 acc1_pos: 28.618 acc5_pos: 100.000 loss: 0.5483 (1.0634) acc1: 68.4896 (35.2004) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.1910 data: 0.0991 max mem: 3531 +Test: [180/333] eta: 0:00:32 acc1_pos: 28.618 acc5_pos: 100.000 loss: 0.4285 (1.0274) acc1: 82.0312 (37.9230) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.2075 data: 0.1149 max mem: 3531 +Test: [190/333] eta: 0:00:30 acc1_pos: 28.618 acc5_pos: 100.000 loss: 0.3732 (0.9919) acc1: 84.6354 (40.6032) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.2185 data: 0.1232 max mem: 3531 +Test: [200/333] eta: 0:00:28 acc1_pos: 28.618 acc5_pos: 100.000 loss: 0.3713 (0.9620) acc1: 86.9792 (42.8185) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.1748 data: 0.0780 max mem: 3531 +Test: [210/333] eta: 0:00:25 acc1_pos: 28.618 acc5_pos: 100.000 loss: 0.4100 (0.9356) acc1: 83.0729 (44.7448) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1650 data: 0.0709 max mem: 3532 +Test: [220/333] eta: 0:00:23 acc1_pos: 28.618 acc5_pos: 100.000 loss: 0.4142 (0.9136) acc1: 83.0729 (46.3695) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.2076 data: 0.1169 max mem: 3532 +Test: [230/333] eta: 0:00:21 acc1_pos: 28.618 acc5_pos: 100.000 loss: 0.3872 (0.8903) acc1: 86.1979 (48.1568) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.2157 data: 0.1251 max mem: 3532 +Test: [240/333] eta: 0:00:19 acc1_pos: 28.618 acc5_pos: 100.000 loss: 0.3932 (0.8708) acc1: 85.6771 (49.6380) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.1991 data: 0.1082 max mem: 3532 +Test: [250/333] eta: 0:00:17 acc1_pos: 28.618 acc5_pos: 100.000 loss: 0.4188 (0.8527) acc1: 82.5521 (50.9275) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.2060 data: 0.1151 max mem: 3532 +Test: [260/333] eta: 0:00:15 acc1_pos: 28.618 acc5_pos: 100.000 loss: 0.3974 (0.8346) acc1: 85.4167 (52.3178) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.1949 data: 0.1040 max mem: 3532 +Test: [270/333] eta: 0:00:13 acc1_pos: 28.618 acc5_pos: 100.000 loss: 0.3749 (0.8178) acc1: 86.1979 (53.5699) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1840 data: 0.0926 max mem: 3532 +Test: [280/333] eta: 0:00:11 acc1_pos: 28.618 acc5_pos: 100.000 loss: 0.3862 (0.8054) acc1: 83.3333 (54.4456) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.2048 data: 0.1128 max mem: 3532 +Test: [290/333] eta: 0:00:08 acc1_pos: 28.618 acc5_pos: 100.000 loss: 0.5126 (0.7961) acc1: 73.1771 (55.0634) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.2094 data: 0.1178 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 28.618 acc5_pos: 100.000 loss: 0.5097 (0.7863) acc1: 73.9583 (55.7612) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.2102 data: 0.1186 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 28.618 acc5_pos: 100.000 loss: 0.4564 (0.7732) acc1: 79.9479 (56.7290) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.1789 data: 0.0871 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 28.618 acc5_pos: 100.000 loss: 0.4010 (0.7624) acc1: 84.3750 (57.5140) acc5: 100.0000 (100.0000) acc1_neg: 71.615 acc5_neg: 100.000 time: 0.1399 data: 0.0490 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 28.618 acc5_pos: 100.000 loss: 0.4863 (0.7553) acc1: 74.7396 (57.9809) acc5: 100.0000 (100.0000) acc1_neg: 73.958 acc5_neg: 100.000 time: 0.1257 data: 0.0353 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 28.618 acc5_pos: 100.000 loss: 0.5133 (0.7538) acc1: 74.7396 (58.0845) acc5: 100.0000 (100.0000) acc1_neg: 74.930 acc5_neg: 100.000 time: 0.1327 data: 0.0395 max mem: 3532 +Test: Total time: 0:01:07 (0.2025 s / it) +* Overall Acc@1 58.085 Acc@5 100.000 + target == real Acc@1 31.544 Acc@5 100.000 + target == fake Acc@1 81.711 Acc@5 100.000 + loss 0.754 + PR-AUC 0.5982 ROC-AUC 0.6015 EER 42.33% (thr=0.6919) +Accuracy of the network on the 127843 test images: 58.1% +[rank0]:[W726 06:08:15.274064373 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 06:08:35.422000 3173548 torch/distributed/run.py:766] +W0726 06:08:35.422000 3173548 torch/distributed/run.py:766] ***************************************** +W0726 06:08:35.422000 3173548 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:08:35.422000 3173548 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 06:09:03.527143801 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:09:04.160279472 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:09:04.428635799 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:09:04.460532506 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:09:04.472535987 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 1): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:09:05.671733205 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank1]:[W726 06:09:05.671733227 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:09:05.677685616 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n16_seed9_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed9_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:21:38 acc1_pos: 94.531 acc5_pos: 100.000 loss: 0.2524 (0.2524) acc1: 94.5312 (94.5312) acc5: 100.0000 (100.0000) time: 5.5267 data: 3.6004 max mem: 3531 +Test: [ 10/235] eta: 0:02:13 acc1_pos: 95.833 acc5_pos: 100.000 loss: 0.2477 (0.2515) acc1: 94.2708 (93.7737) acc5: 100.0000 (100.0000) time: 0.5936 data: 0.3276 max mem: 3531 +Test: [ 20/235] eta: 0:01:16 acc1_pos: 93.490 acc5_pos: 100.000 loss: 0.2442 (0.2494) acc1: 94.0104 (93.9112) acc5: 100.0000 (100.0000) time: 0.0952 data: 0.0004 max mem: 3531 +Test: [ 30/235] eta: 0:01:05 acc1_pos: 94.792 acc5_pos: 100.000 loss: 0.2455 (0.2487) acc1: 93.7500 (93.8172) acc5: 100.0000 (100.0000) time: 0.1712 data: 0.0773 max mem: 3531 +Test: [ 40/235] eta: 0:01:03 acc1_pos: 90.885 acc5_pos: 100.000 loss: 0.2455 (0.2487) acc1: 93.4896 (93.7182) acc5: 100.0000 (100.0000) time: 0.2957 data: 0.2006 max mem: 3531 +Test: [ 50/235] eta: 0:00:53 acc1_pos: 91.146 acc5_pos: 100.000 loss: 0.2484 (0.2504) acc1: 93.4896 (93.5815) acc5: 100.0000 (100.0000) time: 0.2437 data: 0.1500 max mem: 3531 +Test: [ 60/235] eta: 0:00:50 acc1_pos: 93.229 acc5_pos: 100.000 loss: 0.2533 (0.2518) acc1: 92.9688 (93.4896) acc5: 100.0000 (100.0000) time: 0.2112 data: 0.1177 max mem: 3531 +Test: [ 70/235] eta: 0:00:46 acc1_pos: 93.846 acc5_pos: 100.000 loss: 0.2702 (0.3911) acc1: 91.6667 (84.4630) acc5: 100.0000 (100.0000) acc1_neg: 17.969 acc5_neg: 100.000 time: 0.2649 data: 0.1729 max mem: 3531 +Test: [ 80/235] eta: 0:00:41 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.3653 (0.5118) acc1: 21.0938 (76.5593) acc5: 100.0000 (100.0000) acc1_neg: 17.969 acc5_neg: 100.000 time: 0.1978 data: 0.1016 max mem: 3531 +Test: [ 90/235] eta: 0:00:37 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.3737 (0.6063) acc1: 20.3125 (70.5157) acc5: 100.0000 (100.0000) acc1_neg: 18.490 acc5_neg: 100.000 time: 0.1557 data: 0.0581 max mem: 3531 +Test: [100/235] eta: 0:00:33 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.3675 (0.6806) acc1: 21.0938 (65.6895) acc5: 100.0000 (100.0000) acc1_neg: 19.531 acc5_neg: 100.000 time: 0.1649 data: 0.0706 max mem: 3531 +Test: [110/235] eta: 0:00:29 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.3394 (0.7414) acc1: 21.0938 (61.7141) acc5: 100.0000 (100.0000) acc1_neg: 20.312 acc5_neg: 100.000 time: 0.1641 data: 0.0685 max mem: 3531 +Test: [120/235] eta: 0:00:26 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.3930 (0.7993) acc1: 19.2708 (58.0901) acc5: 100.0000 (100.0000) acc1_neg: 17.969 acc5_neg: 100.000 time: 0.1649 data: 0.0688 max mem: 3531 +Test: [130/235] eta: 0:00:24 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.4421 (0.8479) acc1: 17.7083 (54.9757) acc5: 100.0000 (100.0000) acc1_neg: 17.188 acc5_neg: 100.000 time: 0.1817 data: 0.0888 max mem: 3531 +Test: [140/235] eta: 0:00:21 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.4699 (0.8926) acc1: 16.1458 (52.2145) acc5: 100.0000 (100.0000) acc1_neg: 12.240 acc5_neg: 100.000 time: 0.2057 data: 0.1136 max mem: 3531 +Test: [150/235] eta: 0:00:19 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.4463 (0.9282) acc1: 16.1458 (49.8689) acc5: 100.0000 (100.0000) acc1_neg: 19.792 acc5_neg: 100.000 time: 0.1986 data: 0.1067 max mem: 3531 +Test: [160/235] eta: 0:00:16 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.4367 (0.9601) acc1: 17.1875 (47.8568) acc5: 100.0000 (100.0000) acc1_neg: 15.365 acc5_neg: 100.000 time: 0.1952 data: 0.1014 max mem: 3531 +Test: [170/235] eta: 0:00:14 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.3875 (0.9755) acc1: 19.2708 (46.6481) acc5: 100.0000 (100.0000) acc1_neg: 24.740 acc5_neg: 100.000 time: 0.1947 data: 0.0986 max mem: 3531 +Test: [180/235] eta: 0:00:12 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.2105 (0.9885) acc1: 27.6042 (45.6161) acc5: 100.0000 (100.0000) acc1_neg: 27.604 acc5_neg: 100.000 time: 0.1958 data: 0.1031 max mem: 3531 +Test: [190/235] eta: 0:00:09 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.2097 (0.9997) acc1: 28.6458 (44.7480) acc5: 100.0000 (100.0000) acc1_neg: 25.260 acc5_neg: 100.000 time: 0.1896 data: 0.0982 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.2073 (1.0105) acc1: 28.1250 (43.8938) acc5: 100.0000 (100.0000) acc1_neg: 26.823 acc5_neg: 100.000 time: 0.1729 data: 0.0806 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.2131 (1.0194) acc1: 27.0833 (43.1489) acc5: 100.0000 (100.0000) acc1_neg: 26.823 acc5_neg: 100.000 time: 0.1585 data: 0.0678 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.1949 (1.0275) acc1: 28.3854 (42.5021) acc5: 100.0000 (100.0000) acc1_neg: 27.344 acc5_neg: 100.000 time: 0.1469 data: 0.0571 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.1854 (1.0343) acc1: 28.3854 (41.9203) acc5: 100.0000 (100.0000) acc1_neg: 28.385 acc5_neg: 100.000 time: 0.1433 data: 0.0525 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.1949 (1.0383) acc1: 28.3854 (41.7305) acc5: 100.0000 (100.0000) acc1_neg: 21.739 acc5_neg: 100.000 time: 0.1278 data: 0.0401 max mem: 3532 +Test: Total time: 0:00:48 (0.2062 s / it) +* Overall Acc@1 41.731 Acc@5 100.000 + target == real Acc@1 93.503 Acc@5 100.000 + target == fake Acc@1 23.005 Acc@5 100.000 + loss 1.038 + PR-AUC 0.8604 ROC-AUC 0.7075 EER 34.87% (thr=0.1930) +Accuracy of the network on the 89879 test images: 41.7% +[rank0]:[W726 06:10:03.801415973 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 06:10:22.288000 3174094 torch/distributed/run.py:766] +W0726 06:10:22.288000 3174094 torch/distributed/run.py:766] ***************************************** +W0726 06:10:22.288000 3174094 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:10:22.288000 3174094 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:10:50.694476978 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:10:50.801136806 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 06:10:50.816147177 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:10:50.930460055 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:10:50.089845564 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:10:50.122185591 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +[rank5]:[W726 06:10:50.232910790 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:10:50.244832131 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n16_seed9_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed9_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:23:51 acc1_pos: 11.458 acc5_pos: 100.000 loss: 1.4398 (1.4398) acc1: 11.4583 (11.4583) acc5: 100.0000 (100.0000) time: 5.3213 data: 3.4644 max mem: 3531 +Test: [ 10/269] eta: 0:02:33 acc1_pos: 35.938 acc5_pos: 100.000 loss: 1.0727 (1.0689) acc1: 27.6042 (32.0549) acc5: 100.0000 (100.0000) time: 0.5943 data: 0.3441 max mem: 3531 +Test: [ 20/269] eta: 0:01:32 acc1_pos: 34.635 acc5_pos: 100.000 loss: 1.0007 (1.0166) acc1: 35.9375 (36.0615) acc5: 100.0000 (100.0000) time: 0.1249 data: 0.0324 max mem: 3531 +Test: [ 30/269] eta: 0:01:09 acc1_pos: 24.219 acc5_pos: 100.000 loss: 1.0007 (1.0175) acc1: 34.8958 (36.2987) acc5: 100.0000 (100.0000) time: 0.1240 data: 0.0279 max mem: 3531 +Test: [ 40/269] eta: 0:01:03 acc1_pos: 44.828 acc5_pos: 100.000 loss: 0.5719 (0.8665) acc1: 66.9271 (48.4375) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1802 data: 0.0867 max mem: 3531 +Test: [ 50/269] eta: 0:00:58 acc1_pos: 44.828 acc5_pos: 100.000 loss: 0.3185 (0.7715) acc1: 91.6667 (56.1121) acc5: 100.0000 (100.0000) acc1_neg: 70.312 acc5_neg: 100.000 time: 0.2249 data: 0.1334 max mem: 3531 +Test: [ 60/269] eta: 0:00:53 acc1_pos: 44.828 acc5_pos: 100.000 loss: 0.4140 (0.7220) acc1: 81.7708 (59.7507) acc5: 100.0000 (100.0000) acc1_neg: 67.969 acc5_neg: 100.000 time: 0.2046 data: 0.1088 max mem: 3531 +Test: [ 70/269] eta: 0:00:50 acc1_pos: 44.828 acc5_pos: 100.000 loss: 0.4660 (0.6870) acc1: 77.0833 (62.6174) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.2299 data: 0.1344 max mem: 3531 +Test: [ 80/269] eta: 0:00:47 acc1_pos: 44.828 acc5_pos: 100.000 loss: 0.4502 (0.6542) acc1: 81.5104 (65.4835) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2349 data: 0.1410 max mem: 3531 +Test: [ 90/269] eta: 0:00:41 acc1_pos: 44.828 acc5_pos: 100.000 loss: 0.4502 (0.6379) acc1: 81.5104 (66.7182) acc5: 100.0000 (100.0000) acc1_neg: 71.615 acc5_neg: 100.000 time: 0.1610 data: 0.0645 max mem: 3531 +Test: [100/269] eta: 0:00:39 acc1_pos: 44.828 acc5_pos: 100.000 loss: 0.5062 (0.6266) acc1: 73.4375 (67.2339) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.1749 data: 0.0777 max mem: 3531 +Test: [110/269] eta: 0:00:36 acc1_pos: 44.828 acc5_pos: 100.000 loss: 0.4951 (0.6146) acc1: 73.4375 (67.8726) acc5: 100.0000 (100.0000) acc1_neg: 64.323 acc5_neg: 100.000 time: 0.2138 data: 0.1173 max mem: 3531 +Test: [120/269] eta: 0:00:33 acc1_pos: 44.828 acc5_pos: 100.000 loss: 0.4486 (0.5979) acc1: 78.1250 (69.0599) acc5: 100.0000 (100.0000) acc1_neg: 71.094 acc5_neg: 100.000 time: 0.1805 data: 0.0849 max mem: 3531 +Test: [130/269] eta: 0:00:30 acc1_pos: 44.828 acc5_pos: 100.000 loss: 0.4730 (0.5916) acc1: 74.7396 (69.2470) acc5: 100.0000 (100.0000) acc1_neg: 58.594 acc5_neg: 100.000 time: 0.1753 data: 0.0804 max mem: 3531 +Test: [140/269] eta: 0:00:28 acc1_pos: 44.828 acc5_pos: 100.000 loss: 0.5710 (0.5934) acc1: 65.6250 (68.8276) acc5: 100.0000 (100.0000) acc1_neg: 56.771 acc5_neg: 100.000 time: 0.1757 data: 0.0820 max mem: 3531 +Test: [150/269] eta: 0:00:25 acc1_pos: 44.828 acc5_pos: 100.000 loss: 0.6064 (0.5915) acc1: 65.3646 (68.8052) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.1796 data: 0.0862 max mem: 3531 +Test: [160/269] eta: 0:00:23 acc1_pos: 44.828 acc5_pos: 100.000 loss: 0.4586 (0.5828) acc1: 79.9479 (69.5798) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1986 data: 0.1042 max mem: 3531 +Test: [170/269] eta: 0:00:21 acc1_pos: 44.828 acc5_pos: 100.000 loss: 0.4164 (0.5725) acc1: 84.3750 (70.5805) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1911 data: 0.0965 max mem: 3531 +Test: [180/269] eta: 0:00:18 acc1_pos: 44.828 acc5_pos: 100.000 loss: 0.3719 (0.5588) acc1: 88.5417 (71.6721) acc5: 100.0000 (100.0000) acc1_neg: 75.781 acc5_neg: 100.000 time: 0.1557 data: 0.0620 max mem: 3531 +Test: [190/269] eta: 0:00:16 acc1_pos: 44.828 acc5_pos: 100.000 loss: 0.3660 (0.5548) acc1: 86.9792 (71.8682) acc5: 100.0000 (100.0000) acc1_neg: 65.625 acc5_neg: 100.000 time: 0.1637 data: 0.0713 max mem: 3531 +Test: [200/269] eta: 0:00:14 acc1_pos: 44.828 acc5_pos: 100.000 loss: 0.5489 (0.5567) acc1: 67.4479 (71.6340) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1892 data: 0.0964 max mem: 3531 +Test: [210/269] eta: 0:00:12 acc1_pos: 44.828 acc5_pos: 100.000 loss: 0.4862 (0.5515) acc1: 75.5208 (72.0404) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1949 data: 0.1011 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 44.828 acc5_pos: 100.000 loss: 0.4498 (0.5471) acc1: 78.1250 (72.3298) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1732 data: 0.0781 max mem: 3532 +Test: [230/269] eta: 0:00:07 acc1_pos: 44.828 acc5_pos: 100.000 loss: 0.3475 (0.5379) acc1: 87.2396 (73.1218) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1672 data: 0.0738 max mem: 3532 +Test: [240/269] eta: 0:00:05 acc1_pos: 44.828 acc5_pos: 100.000 loss: 0.4124 (0.5378) acc1: 86.1979 (73.1306) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.2020 data: 0.1084 max mem: 3532 +Test: [250/269] eta: 0:00:03 acc1_pos: 44.828 acc5_pos: 100.000 loss: 0.5729 (0.5413) acc1: 67.1875 (72.7320) acc5: 100.0000 (100.0000) acc1_neg: 58.073 acc5_neg: 100.000 time: 0.1905 data: 0.0964 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 44.828 acc5_pos: 100.000 loss: 0.5806 (0.5438) acc1: 67.1875 (72.4477) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.1494 data: 0.0583 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 44.828 acc5_pos: 100.000 loss: 0.5014 (0.5421) acc1: 75.5208 (72.6584) acc5: 100.0000 (100.0000) acc1_neg: 78.215 acc5_neg: 100.000 time: 0.1225 data: 0.0288 max mem: 3532 +Test: Total time: 0:00:53 (0.1980 s / it) +* Overall Acc@1 72.658 Acc@5 100.000 + target == real Acc@1 36.109 Acc@5 100.000 + target == fake Acc@1 77.646 Acc@5 100.000 + loss 0.542 + PR-AUC 0.9072 ROC-AUC 0.5990 EER 42.90% (thr=0.6436) +Accuracy of the network on the 103293 test images: 72.7% +[rank0]:[W726 06:11:53.822883082 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T06:11:59Z] eval_done tag=fixed500k_real_n16_seed9_resampled +[2026-07-26T06:12:00Z] eval_start tag=fixed500k_fake_n32_seed9_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed9_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T06:12:00Z +hostname=n111-046-081 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed9_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n32_seed9_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 06:12:12.556000 3174669 torch/distributed/run.py:766] +W0726 06:12:12.556000 3174669 torch/distributed/run.py:766] ***************************************** +W0726 06:12:12.556000 3174669 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:12:12.556000 3174669 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 06:12:41.661841005 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:12:41.761627231 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:12:41.785486830 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:12:41.089813297 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:12:41.344308799 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +| distributed init (rank 6): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:12:41.373979887 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:12:41.382491016 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W726 06:12:41.382506296 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n32_seed9_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed9_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:19:22 acc1_pos: 29.688 acc5_pos: 100.000 loss: 0.8596 (0.8596) acc1: 29.6875 (29.6875) acc5: 100.0000 (100.0000) time: 4.9674 data: 3.1253 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/234] eta: 0:02:01 acc1_pos: 8.594 acc5_pos: 100.000 loss: 0.8596 (0.8943) acc1: 35.4167 (36.5057) acc5: 100.0000 (100.0000) time: 0.5423 data: 0.2928 max mem: 3531 +Test: [ 20/234] eta: 0:01:11 acc1_pos: 93.750 acc5_pos: 100.000 loss: 0.9576 (0.8403) acc1: 35.4167 (40.3770) acc5: 100.0000 (100.0000) time: 0.1027 data: 0.0125 max mem: 3531 +Test: [ 30/234] eta: 0:00:54 acc1_pos: 54.427 acc5_pos: 100.000 loss: 0.2274 (0.6565) acc1: 93.7500 (56.1492) acc5: 100.0000 (100.0000) time: 0.1179 data: 0.0243 max mem: 3531 +Test: [ 40/234] eta: 0:00:52 acc1_pos: 96.615 acc5_pos: 100.000 loss: 0.1954 (0.6264) acc1: 95.0521 (58.9939) acc5: 100.0000 (100.0000) time: 0.2059 data: 0.1110 max mem: 3531 +Test: [ 50/234] eta: 0:00:50 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.1596 (0.5336) acc1: 96.0938 (66.4267) acc5: 100.0000 (100.0000) time: 0.2890 data: 0.1966 max mem: 3531 +Test: [ 60/234] eta: 0:00:43 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.1223 (0.4603) acc1: 98.1771 (71.8878) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1982 data: 0.1023 max mem: 3531 +Test: [ 70/234] eta: 0:00:38 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.0842 (0.4628) acc1: 99.7396 (72.4509) acc5: 100.0000 (100.0000) acc1_neg: 61.719 acc5_neg: 100.000 time: 0.1313 data: 0.0345 max mem: 3531 +Test: [ 80/234] eta: 0:00:36 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.6428 (0.4650) acc1: 65.8854 (73.0806) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1951 data: 0.1020 max mem: 3531 +Test: [ 90/234] eta: 0:00:33 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.2011 (0.4329) acc1: 94.5312 (75.5666) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.2161 data: 0.1247 max mem: 3531 +Test: [100/234] eta: 0:00:30 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.2085 (0.4317) acc1: 92.4479 (76.1577) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2065 data: 0.1159 max mem: 3531 +Test: [110/234] eta: 0:00:27 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.4237 (0.4256) acc1: 80.9896 (77.0693) acc5: 100.0000 (100.0000) acc1_neg: 75.521 acc5_neg: 100.000 time: 0.1991 data: 0.1077 max mem: 3531 +Test: [120/234] eta: 0:00:25 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.3169 (0.4240) acc1: 89.8438 (77.6795) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1912 data: 0.0979 max mem: 3531 +Test: [130/234] eta: 0:00:22 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.3231 (0.4200) acc1: 91.4062 (78.4749) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.1949 data: 0.1018 max mem: 3531 +Test: [140/234] eta: 0:00:20 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.4733 (0.4610) acc1: 77.0833 (76.4868) acc5: 100.0000 (100.0000) acc1_neg: 26.823 acc5_neg: 100.000 time: 0.1883 data: 0.0969 max mem: 3531 +Test: [150/234] eta: 0:00:18 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.5194 (0.4683) acc1: 72.9167 (76.3055) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.1899 data: 0.0971 max mem: 3531 +Test: [160/234] eta: 0:00:15 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.4233 (0.4619) acc1: 80.2083 (77.0122) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.1942 data: 0.1008 max mem: 3531 +Test: [170/234] eta: 0:00:13 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.4371 (0.4597) acc1: 82.0312 (77.4229) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1805 data: 0.0856 max mem: 3531 +Test: [180/234] eta: 0:00:11 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.4583 (0.4667) acc1: 80.2083 (77.0977) acc5: 100.0000 (100.0000) acc1_neg: 43.229 acc5_neg: 100.000 time: 0.1857 data: 0.0919 max mem: 3531 +Test: [190/234] eta: 0:00:09 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.7427 (0.4824) acc1: 58.3333 (76.0458) acc5: 100.0000 (100.0000) acc1_neg: 57.812 acc5_neg: 100.000 time: 0.1905 data: 0.0993 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.7427 (0.4761) acc1: 58.3333 (76.4835) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1742 data: 0.0825 max mem: 3531 +Test: [210/234] eta: 0:00:04 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.1335 (0.4601) acc1: 99.7396 (77.5610) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1608 data: 0.0698 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.1497 (0.4472) acc1: 98.1771 (78.4219) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.1462 data: 0.0561 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.3306 (0.4582) acc1: 60.9375 (77.5174) acc5: 100.0000 (100.0000) acc1_neg: 56.250 acc5_neg: 100.000 time: 0.1343 data: 0.0449 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.6837 (0.4611) acc1: 58.8542 (77.3095) acc5: 100.0000 (100.0000) acc1_neg: 62.646 acc5_neg: 100.000 time: 0.1359 data: 0.0455 max mem: 3532 +Test: Total time: 0:00:46 (0.1993 s / it) +* Overall Acc@1 77.309 Acc@5 100.000 + target == real Acc@1 66.638 Acc@5 100.000 + target == fake Acc@1 80.317 Acc@5 100.000 + loss 0.461 + PR-AUC 0.9554 ROC-AUC 0.8533 EER 24.95% (thr=0.5903) +Accuracy of the network on the 89729 test images: 77.3% +[rank0]:[W726 06:13:37.567119917 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 06:13:56.312000 3175216 torch/distributed/run.py:766] +W0726 06:13:56.312000 3175216 torch/distributed/run.py:766] ***************************************** +W0726 06:13:56.312000 3175216 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:13:56.312000 3175216 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:14:23.408760830 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:14:23.469528616 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:14:24.668245945 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:14:24.781527755 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:14:24.051297144 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 06:14:24.085010492 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:14:24.267037781 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:14:24.287804246 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n32_seed9_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed9_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:37:13 acc1_pos: 16.927 acc5_pos: 100.000 loss: 1.1719 (1.1719) acc1: 16.9271 (16.9271) acc5: 100.0000 (100.0000) time: 6.9683 data: 5.4041 max mem: 3531 +Test: [ 10/2387] eta: 0:28:26 acc1_pos: 1.562 acc5_pos: 100.000 loss: 1.2280 (1.2356) acc1: 8.5938 (10.6771) acc5: 100.0000 (100.0000) time: 0.7180 data: 0.4916 max mem: 3531 +Test: [ 20/2387] eta: 0:16:42 acc1_pos: 2.344 acc5_pos: 100.000 loss: 1.2907 (1.2934) acc1: 7.5521 (9.5486) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:14:20 acc1_pos: 0.781 acc5_pos: 100.000 loss: 1.3491 (1.3098) acc1: 7.0312 (8.6442) acc5: 100.0000 (100.0000) time: 0.1708 data: 0.0748 max mem: 3531 +Test: [ 40/2387] eta: 0:13:15 acc1_pos: 7.292 acc5_pos: 100.000 loss: 1.3222 (1.3149) acc1: 2.0833 (7.6156) acc5: 100.0000 (100.0000) time: 0.2499 data: 0.1565 max mem: 3531 +Test: [ 50/2387] eta: 0:12:12 acc1_pos: 21.354 acc5_pos: 100.000 loss: 1.2890 (1.3025) acc1: 7.2917 (8.5989) acc5: 100.0000 (100.0000) time: 0.2340 data: 0.1379 max mem: 3531 +Test: [ 60/2387] eta: 0:11:22 acc1_pos: 23.958 acc5_pos: 100.000 loss: 1.2579 (1.3022) acc1: 7.5521 (8.8883) acc5: 100.0000 (100.0000) time: 0.1999 data: 0.1044 max mem: 3531 +Test: [ 70/2387] eta: 0:11:02 acc1_pos: 15.885 acc5_pos: 100.000 loss: 1.3319 (1.3168) acc1: 7.0312 (8.7588) acc5: 100.0000 (100.0000) time: 0.2150 data: 0.1231 max mem: 3531 +Test: [ 80/2387] eta: 0:10:49 acc1_pos: 15.104 acc5_pos: 100.000 loss: 1.3240 (1.3073) acc1: 7.2917 (8.9217) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1522 max mem: 3531 +Test: [ 90/2387] eta: 0:10:33 acc1_pos: 21.875 acc5_pos: 100.000 loss: 1.1375 (1.2887) acc1: 15.1042 (9.8958) acc5: 100.0000 (100.0000) time: 0.2398 data: 0.1470 max mem: 3531 +Test: [ 100/2387] eta: 0:10:24 acc1_pos: 10.417 acc5_pos: 100.000 loss: 1.1185 (1.2805) acc1: 15.6250 (10.1330) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1482 max mem: 3531 +Test: [ 110/2387] eta: 0:10:19 acc1_pos: 15.885 acc5_pos: 100.000 loss: 1.1185 (1.2663) acc1: 13.5417 (10.5316) acc5: 100.0000 (100.0000) time: 0.2557 data: 0.1624 max mem: 3531 +Test: [ 120/2387] eta: 0:10:12 acc1_pos: 12.760 acc5_pos: 100.000 loss: 1.1319 (1.2598) acc1: 14.5833 (10.9310) acc5: 100.0000 (100.0000) time: 0.2559 data: 0.1635 max mem: 3531 +Test: [ 130/2387] eta: 0:10:05 acc1_pos: 16.927 acc5_pos: 100.000 loss: 1.1803 (1.2532) acc1: 12.7604 (10.9494) acc5: 100.0000 (100.0000) time: 0.2469 data: 0.1523 max mem: 3531 +Test: [ 140/2387] eta: 0:09:56 acc1_pos: 21.615 acc5_pos: 100.000 loss: 1.1510 (1.2477) acc1: 11.4583 (11.1259) acc5: 100.0000 (100.0000) time: 0.2354 data: 0.1402 max mem: 3531 +Test: [ 150/2387] eta: 0:09:52 acc1_pos: 4.948 acc5_pos: 100.000 loss: 1.1136 (1.2374) acc1: 13.2812 (11.6101) acc5: 100.0000 (100.0000) time: 0.2425 data: 0.1505 max mem: 3531 +Test: [ 160/2387] eta: 0:09:36 acc1_pos: 16.438 acc5_pos: 100.000 loss: 1.0415 (1.2108) acc1: 20.8333 (13.6694) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2135 data: 0.1181 max mem: 3531 +Test: [ 170/2387] eta: 0:09:21 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2163 (1.1520) acc1: 98.6979 (18.6769) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1651 data: 0.0692 max mem: 3531 +Test: [ 180/2387] eta: 0:09:13 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2063 (1.0997) acc1: 99.4792 (23.1296) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1849 data: 0.0890 max mem: 3531 +Test: [ 190/2387] eta: 0:09:03 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2045 (1.0527) acc1: 99.2188 (27.1174) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1993 data: 0.1016 max mem: 3531 +Test: [ 200/2387] eta: 0:08:58 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2005 (1.0104) acc1: 99.2188 (30.7071) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2049 data: 0.1069 max mem: 3531 +Test: [ 210/2387] eta: 0:08:50 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2040 (0.9724) acc1: 99.2188 (33.9541) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2049 data: 0.1089 max mem: 3532 +Test: [ 220/2387] eta: 0:08:35 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2054 (0.9376) acc1: 99.2188 (36.9096) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1550 data: 0.0619 max mem: 3532 +Test: [ 230/2387] eta: 0:08:26 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2055 (0.9058) acc1: 99.2188 (39.6104) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1449 data: 0.0521 max mem: 3532 +Test: [ 240/2387] eta: 0:08:24 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2068 (0.8780) acc1: 98.9583 (42.0578) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2048 data: 0.1086 max mem: 3532 +Test: [ 250/2387] eta: 0:08:18 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2001 (0.8478) acc1: 99.2188 (44.3569) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2149 data: 0.1179 max mem: 3532 +Test: [ 260/2387] eta: 0:08:06 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1005 (0.8190) acc1: 100.0000 (46.4889) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1503 data: 0.0542 max mem: 3532 +Test: [ 270/2387] eta: 0:08:01 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.0948 (0.7923) acc1: 100.0000 (48.4634) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1517 data: 0.0562 max mem: 3532 +Test: [ 280/2387] eta: 0:07:56 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.0970 (0.7677) acc1: 100.0000 (50.2975) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1958 data: 0.1031 max mem: 3532 +Test: [ 290/2387] eta: 0:07:51 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1022 (0.7449) acc1: 100.0000 (52.0055) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1942 data: 0.1017 max mem: 3532 +Test: [ 300/2387] eta: 0:07:45 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1022 (0.7234) acc1: 100.0000 (53.6000) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1736 data: 0.0816 max mem: 3532 +Test: [ 310/2387] eta: 0:07:40 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1037 (0.7065) acc1: 100.0000 (55.0635) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1741 data: 0.0794 max mem: 3532 +Test: [ 320/2387] eta: 0:07:35 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.3060 (0.6942) acc1: 98.6979 (56.4074) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1860 data: 0.0873 max mem: 3532 +Test: [ 330/2387] eta: 0:07:32 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.3076 (0.6826) acc1: 98.6979 (57.6929) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1896 data: 0.0952 max mem: 3532 +Test: [ 340/2387] eta: 0:07:28 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.3076 (0.6716) acc1: 98.9583 (58.8901) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1988 data: 0.1032 max mem: 3532 +Test: [ 350/2387] eta: 0:07:23 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2997 (0.6578) acc1: 99.4792 (60.0561) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0809 max mem: 3532 +Test: [ 360/2387] eta: 0:07:18 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1742 (0.6445) acc1: 100.0000 (61.1590) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0732 max mem: 3532 +Test: [ 370/2387] eta: 0:07:13 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1772 (0.6319) acc1: 100.0000 (62.2024) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1682 data: 0.0749 max mem: 3532 +Test: [ 380/2387] eta: 0:07:08 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1765 (0.6200) acc1: 100.0000 (63.1890) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1607 data: 0.0708 max mem: 3532 +Test: [ 390/2387] eta: 0:07:04 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1766 (0.6087) acc1: 99.7396 (64.1218) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1648 data: 0.0730 max mem: 3532 +Test: [ 400/2387] eta: 0:07:00 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1772 (0.5980) acc1: 99.4792 (65.0087) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1743 data: 0.0812 max mem: 3532 +Test: [ 410/2387] eta: 0:06:56 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1755 (0.5877) acc1: 99.7396 (65.8550) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1767 data: 0.0849 max mem: 3532 +Test: [ 420/2387] eta: 0:06:51 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1738 (0.5779) acc1: 99.7396 (66.6586) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1647 data: 0.0728 max mem: 3532 +Test: [ 430/2387] eta: 0:06:48 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1775 (0.5690) acc1: 99.7396 (67.4219) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1649 data: 0.0728 max mem: 3532 +Test: [ 440/2387] eta: 0:06:44 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1800 (0.5594) acc1: 99.7396 (68.1542) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1752 data: 0.0829 max mem: 3533 +Test: [ 450/2387] eta: 0:06:41 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1032 (0.5493) acc1: 100.0000 (68.8603) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1823 data: 0.0893 max mem: 3533 +Test: [ 460/2387] eta: 0:06:37 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1022 (0.5396) acc1: 100.0000 (69.5358) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1714 data: 0.0796 max mem: 3533 +Test: [ 470/2387] eta: 0:06:35 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1022 (0.5303) acc1: 100.0000 (70.1826) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1775 data: 0.0842 max mem: 3533 +Test: [ 480/2387] eta: 0:06:30 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1026 (0.5215) acc1: 100.0000 (70.8025) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1731 data: 0.0776 max mem: 3533 +Test: [ 490/2387] eta: 0:06:27 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1024 (0.5129) acc1: 100.0000 (71.3971) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1649 data: 0.0697 max mem: 3533 +Test: [ 500/2387] eta: 0:06:25 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1024 (0.5048) acc1: 100.0000 (71.9680) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.1018 max mem: 3533 +Test: [ 510/2387] eta: 0:06:24 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1036 (0.4971) acc1: 100.0000 (72.5166) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2134 data: 0.1217 max mem: 3533 +Test: [ 520/2387] eta: 0:06:22 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1326 (0.4902) acc1: 100.0000 (73.0441) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2110 data: 0.1189 max mem: 3533 +Test: [ 530/2387] eta: 0:06:20 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1323 (0.4835) acc1: 100.0000 (73.5518) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2064 data: 0.1129 max mem: 3533 +Test: [ 540/2387] eta: 0:06:17 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1320 (0.4770) acc1: 100.0000 (74.0406) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2008 data: 0.1080 max mem: 3533 +Test: [ 550/2387] eta: 0:06:15 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1311 (0.4707) acc1: 100.0000 (74.5118) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.1022 max mem: 3533 +Test: [ 560/2387] eta: 0:06:13 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1311 (0.4646) acc1: 100.0000 (74.9661) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1943 data: 0.1019 max mem: 3533 +Test: [ 570/2387] eta: 0:06:09 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1305 (0.4588) acc1: 100.0000 (75.4045) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1810 data: 0.0897 max mem: 3533 +Test: [ 580/2387] eta: 0:06:06 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1305 (0.4532) acc1: 100.0000 (75.8279) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1704 data: 0.0770 max mem: 3533 +Test: [ 590/2387] eta: 0:06:03 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1315 (0.4477) acc1: 100.0000 (76.2369) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1663 data: 0.0736 max mem: 3533 +Test: [ 600/2387] eta: 0:06:00 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1354 (0.4426) acc1: 100.0000 (76.6323) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1705 data: 0.0790 max mem: 3533 +Test: [ 610/2387] eta: 0:05:57 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1439 (0.4378) acc1: 100.0000 (77.0147) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0779 max mem: 3533 +Test: [ 620/2387] eta: 0:05:54 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1445 (0.4331) acc1: 100.0000 (77.3848) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1618 data: 0.0704 max mem: 3533 +Test: [ 630/2387] eta: 0:05:51 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1445 (0.4285) acc1: 100.0000 (77.7432) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1631 data: 0.0699 max mem: 3533 +Test: [ 640/2387] eta: 0:05:48 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1462 (0.4241) acc1: 100.0000 (78.0905) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1649 data: 0.0717 max mem: 3533 +Test: [ 650/2387] eta: 0:05:45 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1464 (0.4199) acc1: 100.0000 (78.4270) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1709 data: 0.0794 max mem: 3533 +Test: [ 660/2387] eta: 0:05:43 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1464 (0.4157) acc1: 100.0000 (78.7534) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1759 data: 0.0835 max mem: 3533 +Test: [ 670/2387] eta: 0:05:41 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1458 (0.4117) acc1: 100.0000 (79.0700) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1842 data: 0.0923 max mem: 3534 +Test: [ 680/2387] eta: 0:05:38 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1477 (0.4080) acc1: 100.0000 (79.3774) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0993 max mem: 3534 +Test: [ 690/2387] eta: 0:05:36 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1632 (0.4063) acc1: 100.0000 (79.5281) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.1842 data: 0.0937 max mem: 3534 +Test: [ 700/2387] eta: 0:05:34 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.3449 (0.4055) acc1: 86.1979 (79.6154) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1812 data: 0.0891 max mem: 3534 +Test: [ 710/2387] eta: 0:05:31 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.3449 (0.4046) acc1: 85.9375 (79.7076) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1803 data: 0.0864 max mem: 3534 +Test: [ 720/2387] eta: 0:05:28 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.3391 (0.4037) acc1: 86.4583 (79.8009) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.1692 data: 0.0760 max mem: 3534 +Test: [ 730/2387] eta: 0:05:26 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.3410 (0.4029) acc1: 85.6771 (79.8777) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1703 data: 0.0785 max mem: 3534 +Test: [ 740/2387] eta: 0:05:24 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.3429 (0.4021) acc1: 85.6771 (79.9592) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.1782 data: 0.0858 max mem: 3534 +Test: [ 750/2387] eta: 0:05:22 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.3375 (0.4013) acc1: 85.9375 (80.0412) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1837 data: 0.0905 max mem: 3534 +Test: [ 760/2387] eta: 0:05:19 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.3335 (0.3998) acc1: 86.1979 (80.1632) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1899 data: 0.0972 max mem: 3534 +Test: [ 770/2387] eta: 0:05:17 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2244 (0.3973) acc1: 93.2292 (80.3759) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1855 data: 0.0934 max mem: 3534 +Test: [ 780/2387] eta: 0:05:16 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1839 (0.3943) acc1: 98.4375 (80.6108) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2067 data: 0.1152 max mem: 3534 +Test: [ 790/2387] eta: 0:05:14 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1444 (0.3914) acc1: 99.7396 (80.8227) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.2200 data: 0.1292 max mem: 3534 +Test: [ 800/2387] eta: 0:05:12 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1371 (0.3882) acc1: 100.0000 (81.0585) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2001 data: 0.1068 max mem: 3534 +Test: [ 810/2387] eta: 0:05:10 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1486 (0.3854) acc1: 100.0000 (81.2831) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1864 data: 0.0934 max mem: 3534 +Test: [ 820/2387] eta: 0:05:08 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1607 (0.3827) acc1: 100.0000 (81.5006) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1982 data: 0.1071 max mem: 3534 +Test: [ 830/2387] eta: 0:05:06 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1799 (0.3808) acc1: 99.4792 (81.6878) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1971 data: 0.1047 max mem: 3534 +Test: [ 840/2387] eta: 0:05:04 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2214 (0.3789) acc1: 99.2188 (81.8990) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1810 data: 0.0877 max mem: 3534 +Test: [ 850/2387] eta: 0:05:02 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2210 (0.3769) acc1: 99.7396 (82.1050) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1921 data: 0.0973 max mem: 3534 +Test: [ 860/2387] eta: 0:05:00 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1945 (0.3747) acc1: 99.7396 (82.3086) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2016 data: 0.1057 max mem: 3534 +Test: [ 870/2387] eta: 0:04:58 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1876 (0.3727) acc1: 99.7396 (82.5090) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2000 data: 0.1061 max mem: 3534 +Test: [ 880/2387] eta: 0:04:56 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2085 (0.3709) acc1: 99.7396 (82.6984) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2025 data: 0.1088 max mem: 3534 +Test: [ 890/2387] eta: 0:04:54 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2085 (0.3689) acc1: 99.7396 (82.8900) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2076 data: 0.1146 max mem: 3534 +Test: [ 900/2387] eta: 0:04:52 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2170 (0.3740) acc1: 98.4375 (82.5717) acc5: 100.0000 (100.0000) acc1_neg: 45.573 acc5_neg: 100.000 time: 0.1965 data: 0.1057 max mem: 3535 +Test: [ 910/2387] eta: 0:04:50 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.9693 (0.3806) acc1: 42.9688 (82.1404) acc5: 100.0000 (100.0000) acc1_neg: 42.969 acc5_neg: 100.000 time: 0.1708 data: 0.0803 max mem: 3535 +Test: [ 920/2387] eta: 0:04:47 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.9721 (0.3870) acc1: 42.9688 (81.7177) acc5: 100.0000 (100.0000) acc1_neg: 40.885 acc5_neg: 100.000 time: 0.1528 data: 0.0622 max mem: 3535 +Test: [ 930/2387] eta: 0:04:44 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.9717 (0.3933) acc1: 43.7500 (81.3157) acc5: 100.0000 (100.0000) acc1_neg: 41.667 acc5_neg: 100.000 time: 0.1547 data: 0.0636 max mem: 3535 +Test: [ 940/2387] eta: 0:04:42 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.9819 (0.3994) acc1: 42.7083 (80.9138) acc5: 100.0000 (100.0000) acc1_neg: 41.146 acc5_neg: 100.000 time: 0.1624 data: 0.0692 max mem: 3535 +Test: [ 950/2387] eta: 0:04:40 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.9559 (0.4053) acc1: 42.7083 (80.5180) acc5: 100.0000 (100.0000) acc1_neg: 45.573 acc5_neg: 100.000 time: 0.1742 data: 0.0773 max mem: 3535 +Test: [ 960/2387] eta: 0:04:37 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.9513 (0.4111) acc1: 43.2292 (80.1357) acc5: 100.0000 (100.0000) acc1_neg: 48.698 acc5_neg: 100.000 time: 0.1691 data: 0.0731 max mem: 3535 +Test: [ 970/2387] eta: 0:04:35 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.9600 (0.4169) acc1: 42.7083 (79.7478) acc5: 100.0000 (100.0000) acc1_neg: 44.271 acc5_neg: 100.000 time: 0.1613 data: 0.0690 max mem: 3535 +Test: [ 980/2387] eta: 0:04:33 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.9442 (0.4171) acc1: 44.0104 (79.7478) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1695 data: 0.0770 max mem: 3535 +Test: [ 990/2387] eta: 0:04:30 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1616 (0.4148) acc1: 99.4792 (79.9227) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1685 data: 0.0755 max mem: 3535 +Test: [1000/2387] eta: 0:04:28 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1392 (0.4121) acc1: 100.0000 (80.1131) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1654 data: 0.0740 max mem: 3535 +Test: [1010/2387] eta: 0:04:26 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1522 (0.4097) acc1: 99.7396 (80.2995) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1708 data: 0.0803 max mem: 3535 +Test: [1020/2387] eta: 0:04:24 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1785 (0.4078) acc1: 99.2188 (80.4677) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1736 data: 0.0806 max mem: 3535 +Test: [1030/2387] eta: 0:04:21 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1888 (0.4056) acc1: 99.2188 (80.6501) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1618 data: 0.0688 max mem: 3535 +Test: [1040/2387] eta: 0:04:19 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1647 (0.4033) acc1: 100.0000 (80.8327) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1645 data: 0.0739 max mem: 3535 +Test: [1050/2387] eta: 0:04:17 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1571 (0.4009) acc1: 100.0000 (81.0129) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1730 data: 0.0825 max mem: 3535 +Test: [1060/2387] eta: 0:04:15 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1665 (0.3988) acc1: 100.0000 (81.1886) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1754 data: 0.0839 max mem: 3535 +Test: [1070/2387] eta: 0:04:13 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1816 (0.3968) acc1: 100.0000 (81.3587) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1840 data: 0.0914 max mem: 3535 +Test: [1080/2387] eta: 0:04:11 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1957 (0.3951) acc1: 99.4792 (81.5075) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1971 data: 0.1055 max mem: 3535 +Test: [1090/2387] eta: 0:04:09 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1949 (0.3933) acc1: 99.2188 (81.6601) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2106 data: 0.1195 max mem: 3535 +Test: [1100/2387] eta: 0:04:08 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1792 (0.3913) acc1: 99.4792 (81.8236) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2120 data: 0.1207 max mem: 3535 +Test: [1110/2387] eta: 0:04:06 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1741 (0.3895) acc1: 99.7396 (81.9626) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2092 data: 0.1176 max mem: 3535 +Test: [1120/2387] eta: 0:04:04 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1742 (0.3877) acc1: 98.6979 (82.1144) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2076 data: 0.1147 max mem: 3536 +Test: [1130/2387] eta: 0:04:02 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1786 (0.3859) acc1: 99.7396 (82.2693) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2049 data: 0.1128 max mem: 3536 +Test: [1140/2387] eta: 0:04:00 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1884 (0.3842) acc1: 99.7396 (82.4179) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1989 data: 0.1076 max mem: 3536 +Test: [1150/2387] eta: 0:03:58 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1939 (0.3827) acc1: 99.7396 (82.5564) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1905 data: 0.0990 max mem: 3536 +Test: [1160/2387] eta: 0:03:56 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1644 (0.3807) acc1: 100.0000 (82.7060) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.1041 max mem: 3536 +Test: [1170/2387] eta: 0:03:55 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1518 (0.3788) acc1: 100.0000 (82.8536) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2061 data: 0.1155 max mem: 3536 +Test: [1180/2387] eta: 0:03:53 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1507 (0.3769) acc1: 100.0000 (82.9982) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1982 data: 0.1037 max mem: 3536 +Test: [1190/2387] eta: 0:03:51 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1626 (0.3752) acc1: 100.0000 (83.1403) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1931 data: 0.0973 max mem: 3536 +Test: [1200/2387] eta: 0:03:49 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1630 (0.3733) acc1: 100.0000 (83.2806) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2004 data: 0.1082 max mem: 3536 +Test: [1210/2387] eta: 0:03:48 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1496 (0.3715) acc1: 100.0000 (83.4187) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2349 data: 0.1410 max mem: 3536 +Test: [1220/2387] eta: 0:03:46 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1179 (0.3694) acc1: 100.0000 (83.5541) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2194 data: 0.1227 max mem: 3536 +Test: [1230/2387] eta: 0:03:44 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.0995 (0.3672) acc1: 100.0000 (83.6877) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1730 data: 0.0777 max mem: 3536 +Test: [1240/2387] eta: 0:03:42 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.0972 (0.3650) acc1: 100.0000 (83.8191) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1815 data: 0.0895 max mem: 3536 +Test: [1250/2387] eta: 0:03:40 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1008 (0.3629) acc1: 100.0000 (83.9483) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1937 data: 0.1001 max mem: 3536 +Test: [1260/2387] eta: 0:03:38 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1059 (0.3609) acc1: 100.0000 (84.0756) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2021 data: 0.1080 max mem: 3536 +Test: [1270/2387] eta: 0:03:36 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1031 (0.3588) acc1: 100.0000 (84.2008) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2006 data: 0.1092 max mem: 3536 +Test: [1280/2387] eta: 0:03:34 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1038 (0.3569) acc1: 100.0000 (84.3226) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1990 data: 0.1080 max mem: 3536 +Test: [1290/2387] eta: 0:03:32 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1144 (0.3551) acc1: 100.0000 (84.4377) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2157 data: 0.1242 max mem: 3536 +Test: [1300/2387] eta: 0:03:30 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1090 (0.3532) acc1: 100.0000 (84.5542) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1996 data: 0.1076 max mem: 3536 +Test: [1310/2387] eta: 0:03:29 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1145 (0.3515) acc1: 100.0000 (84.6684) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1978 data: 0.1065 max mem: 3536 +Test: [1320/2387] eta: 0:03:26 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1196 (0.3498) acc1: 100.0000 (84.7795) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2001 data: 0.1088 max mem: 3536 +Test: [1330/2387] eta: 0:03:24 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1397 (0.3483) acc1: 99.7396 (84.8882) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1815 data: 0.0902 max mem: 3536 +Test: [1340/2387] eta: 0:03:23 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1714 (0.3479) acc1: 98.9583 (84.9180) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1957 data: 0.1025 max mem: 3536 +Test: [1350/2387] eta: 0:03:21 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.0824 (0.3459) acc1: 100.0000 (85.0296) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1983 data: 0.1051 max mem: 3537 +Test: [1360/2387] eta: 0:03:19 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.0698 (0.3439) acc1: 100.0000 (85.1396) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1935 data: 0.1024 max mem: 3537 +Test: [1370/2387] eta: 0:03:17 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.0658 (0.3418) acc1: 100.0000 (85.2480) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1936 data: 0.1013 max mem: 3537 +Test: [1380/2387] eta: 0:03:15 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.0683 (0.3399) acc1: 100.0000 (85.3548) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1930 data: 0.0995 max mem: 3537 +Test: [1390/2387] eta: 0:03:13 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.0746 (0.3380) acc1: 100.0000 (85.4601) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1838 data: 0.0916 max mem: 3537 +Test: [1400/2387] eta: 0:03:11 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.0716 (0.3361) acc1: 100.0000 (85.5639) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1817 data: 0.0897 max mem: 3537 +Test: [1410/2387] eta: 0:03:09 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.0758 (0.3347) acc1: 100.0000 (85.6618) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1959 data: 0.1027 max mem: 3537 +Test: [1420/2387] eta: 0:03:07 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1469 (0.3335) acc1: 100.0000 (85.7511) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2096 data: 0.1177 max mem: 3537 +Test: [1430/2387] eta: 0:03:05 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1528 (0.3322) acc1: 100.0000 (85.8487) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2070 data: 0.1161 max mem: 3537 +Test: [1440/2387] eta: 0:03:03 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1532 (0.3310) acc1: 100.0000 (85.9455) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2036 data: 0.1126 max mem: 3537 +Test: [1450/2387] eta: 0:03:02 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1606 (0.3299) acc1: 100.0000 (86.0341) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2138 data: 0.1232 max mem: 3537 +Test: [1460/2387] eta: 0:03:00 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1441 (0.3286) acc1: 99.4792 (86.1268) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2021 data: 0.1119 max mem: 3537 +Test: [1470/2387] eta: 0:02:58 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1204 (0.3271) acc1: 100.0000 (86.2206) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1860 data: 0.0942 max mem: 3537 +Test: [1480/2387] eta: 0:02:56 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1108 (0.3256) acc1: 100.0000 (86.3129) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1928 data: 0.1001 max mem: 3537 +Test: [1490/2387] eta: 0:02:54 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1126 (0.3243) acc1: 100.0000 (86.4047) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1997 data: 0.1083 max mem: 3537 +Test: [1500/2387] eta: 0:02:52 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1213 (0.3229) acc1: 100.0000 (86.4942) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2007 data: 0.1088 max mem: 3537 +Test: [1510/2387] eta: 0:02:50 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1342 (0.3218) acc1: 100.0000 (86.5836) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2030 data: 0.1086 max mem: 3537 +Test: [1520/2387] eta: 0:02:48 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1660 (0.3210) acc1: 100.0000 (86.6684) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2013 data: 0.1074 max mem: 3537 +Test: [1530/2387] eta: 0:02:46 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1867 (0.3201) acc1: 99.7396 (86.7533) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1963 data: 0.1056 max mem: 3537 +Test: [1540/2387] eta: 0:02:44 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1735 (0.3191) acc1: 100.0000 (86.8389) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2045 data: 0.1137 max mem: 3537 +Test: [1550/2387] eta: 0:02:42 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1699 (0.3182) acc1: 100.0000 (86.9224) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2085 data: 0.1159 max mem: 3537 +Test: [1560/2387] eta: 0:02:41 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1868 (0.3174) acc1: 99.7396 (87.0040) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2059 data: 0.1138 max mem: 3537 +Test: [1570/2387] eta: 0:02:39 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1775 (0.3165) acc1: 100.0000 (87.0858) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2081 data: 0.1177 max mem: 3537 +Test: [1580/2387] eta: 0:02:37 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1623 (0.3156) acc1: 100.0000 (87.1663) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2040 data: 0.1138 max mem: 3538 +Test: [1590/2387] eta: 0:02:35 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1488 (0.3145) acc1: 100.0000 (87.2463) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2080 data: 0.1178 max mem: 3538 +Test: [1600/2387] eta: 0:02:33 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1344 (0.3134) acc1: 100.0000 (87.3157) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2135 data: 0.1221 max mem: 3538 +Test: [1610/2387] eta: 0:02:31 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1224 (0.3122) acc1: 100.0000 (87.3935) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2073 data: 0.1149 max mem: 3538 +Test: [1620/2387] eta: 0:02:29 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1430 (0.3113) acc1: 100.0000 (87.4667) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1939 data: 0.1006 max mem: 3538 +Test: [1630/2387] eta: 0:02:27 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1567 (0.3104) acc1: 99.4792 (87.5370) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1821 data: 0.0888 max mem: 3538 +Test: [1640/2387] eta: 0:02:25 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1538 (0.3094) acc1: 99.4792 (87.6058) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1799 data: 0.0873 max mem: 3538 +Test: [1650/2387] eta: 0:02:23 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1554 (0.3085) acc1: 99.4792 (87.6757) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1767 data: 0.0856 max mem: 3538 +Test: [1660/2387] eta: 0:02:21 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1540 (0.3076) acc1: 99.7396 (87.7463) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1709 data: 0.0808 max mem: 3538 +Test: [1670/2387] eta: 0:02:19 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1482 (0.3067) acc1: 99.7396 (87.8147) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1676 data: 0.0758 max mem: 3538 +Test: [1680/2387] eta: 0:02:17 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1466 (0.3058) acc1: 99.7396 (87.8789) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1650 data: 0.0734 max mem: 3538 +Test: [1690/2387] eta: 0:02:15 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1612 (0.3051) acc1: 99.4792 (87.9401) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1713 data: 0.0811 max mem: 3538 +Test: [1700/2387] eta: 0:02:13 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1637 (0.3042) acc1: 99.4792 (88.0078) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1738 data: 0.0823 max mem: 3538 +Test: [1710/2387] eta: 0:02:11 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1260 (0.3031) acc1: 100.0000 (88.0773) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1596 data: 0.0682 max mem: 3538 +Test: [1720/2387] eta: 0:02:09 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1162 (0.3020) acc1: 100.0000 (88.1463) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1568 data: 0.0667 max mem: 3538 +Test: [1730/2387] eta: 0:02:07 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1152 (0.3010) acc1: 100.0000 (88.2142) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1645 data: 0.0740 max mem: 3538 +Test: [1740/2387] eta: 0:02:05 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1120 (0.2999) acc1: 100.0000 (88.2813) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1697 data: 0.0779 max mem: 3538 +Test: [1750/2387] eta: 0:02:02 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1172 (0.2989) acc1: 100.0000 (88.3482) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1608 data: 0.0686 max mem: 3538 +Test: [1760/2387] eta: 0:02:00 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1186 (0.2978) acc1: 100.0000 (88.4143) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1509 data: 0.0590 max mem: 3538 +Test: [1770/2387] eta: 0:01:58 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1328 (0.2970) acc1: 100.0000 (88.4771) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1547 data: 0.0622 max mem: 3538 +Test: [1780/2387] eta: 0:01:56 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1397 (0.2961) acc1: 100.0000 (88.5418) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1627 data: 0.0709 max mem: 3538 +Test: [1790/2387] eta: 0:01:54 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1264 (0.2951) acc1: 100.0000 (88.6055) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1653 data: 0.0751 max mem: 3538 +Test: [1800/2387] eta: 0:01:52 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1202 (0.2942) acc1: 100.0000 (88.6680) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1649 data: 0.0744 max mem: 3538 +Test: [1810/2387] eta: 0:01:50 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1291 (0.2933) acc1: 100.0000 (88.7303) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1582 data: 0.0676 max mem: 3539 +Test: [1820/2387] eta: 0:01:48 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1248 (0.2923) acc1: 100.0000 (88.7921) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1545 data: 0.0636 max mem: 3539 +Test: [1830/2387] eta: 0:01:46 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1250 (0.2914) acc1: 100.0000 (88.8533) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1670 data: 0.0750 max mem: 3539 +Test: [1840/2387] eta: 0:01:44 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1291 (0.2905) acc1: 100.0000 (88.9138) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1707 data: 0.0787 max mem: 3539 +Test: [1850/2387] eta: 0:01:42 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1291 (0.2897) acc1: 100.0000 (88.9737) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1653 data: 0.0740 max mem: 3539 +Test: [1860/2387] eta: 0:01:40 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1300 (0.2888) acc1: 100.0000 (89.0330) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1591 data: 0.0670 max mem: 3539 +Test: [1870/2387] eta: 0:01:38 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1300 (0.2880) acc1: 100.0000 (89.0916) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1519 data: 0.0597 max mem: 3539 +Test: [1880/2387] eta: 0:01:36 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1284 (0.2871) acc1: 100.0000 (89.1496) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1517 data: 0.0602 max mem: 3539 +Test: [1890/2387] eta: 0:01:34 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1293 (0.2863) acc1: 100.0000 (89.2070) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1658 data: 0.0742 max mem: 3539 +Test: [1900/2387] eta: 0:01:32 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1304 (0.2855) acc1: 100.0000 (89.2636) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1790 data: 0.0875 max mem: 3539 +Test: [1910/2387] eta: 0:01:30 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1459 (0.2848) acc1: 100.0000 (89.3191) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1704 data: 0.0789 max mem: 3539 +Test: [1920/2387] eta: 0:01:28 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1530 (0.2843) acc1: 100.0000 (89.3691) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1541 data: 0.0633 max mem: 3539 +Test: [1930/2387] eta: 0:01:26 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1516 (0.2835) acc1: 100.0000 (89.4242) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1486 data: 0.0579 max mem: 3539 +Test: [1940/2387] eta: 0:01:24 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1441 (0.2830) acc1: 100.0000 (89.4755) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1510 data: 0.0602 max mem: 3539 +Test: [1950/2387] eta: 0:01:22 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1453 (0.2823) acc1: 100.0000 (89.5294) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1478 data: 0.0575 max mem: 3539 +Test: [1960/2387] eta: 0:01:20 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1509 (0.2818) acc1: 100.0000 (89.5748) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1417 data: 0.0514 max mem: 3539 +Test: [1970/2387] eta: 0:01:18 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1975 (0.2816) acc1: 99.2188 (89.6096) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1522 data: 0.0614 max mem: 3539 +Test: [1980/2387] eta: 0:01:16 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1975 (0.2811) acc1: 98.4375 (89.6497) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1542 data: 0.0637 max mem: 3539 +Test: [1990/2387] eta: 0:01:14 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1374 (0.2804) acc1: 99.2188 (89.6990) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1415 data: 0.0508 max mem: 3539 +Test: [2000/2387] eta: 0:01:12 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1373 (0.2797) acc1: 99.4792 (89.7471) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1428 data: 0.0521 max mem: 3539 +Test: [2010/2387] eta: 0:01:11 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1365 (0.2790) acc1: 99.2188 (89.7949) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1581 data: 0.0670 max mem: 3539 +Test: [2020/2387] eta: 0:01:09 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1365 (0.2783) acc1: 99.4792 (89.8422) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1652 data: 0.0737 max mem: 3539 +Test: [2030/2387] eta: 0:01:07 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1388 (0.2776) acc1: 99.2188 (89.8881) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1474 data: 0.0564 max mem: 3540 +Test: [2040/2387] eta: 0:01:05 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1383 (0.2769) acc1: 99.4792 (89.9354) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1447 data: 0.0542 max mem: 3540 +Test: [2050/2387] eta: 0:01:03 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1359 (0.2762) acc1: 99.4792 (89.9825) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1624 data: 0.0678 max mem: 3540 +Test: [2060/2387] eta: 0:01:01 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1396 (0.2756) acc1: 99.4792 (90.0282) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1596 data: 0.0633 max mem: 3540 +Test: [2070/2387] eta: 0:00:59 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1402 (0.2749) acc1: 99.4792 (90.0740) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1579 data: 0.0646 max mem: 3540 +Test: [2080/2387] eta: 0:00:57 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1373 (0.2742) acc1: 99.4792 (90.1188) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1622 data: 0.0688 max mem: 3540 +Test: [2090/2387] eta: 0:00:55 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1373 (0.2736) acc1: 99.4792 (90.1639) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1600 data: 0.0678 max mem: 3540 +Test: [2100/2387] eta: 0:00:53 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1373 (0.2729) acc1: 99.7396 (90.2092) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1626 data: 0.0712 max mem: 3540 +Test: [2110/2387] eta: 0:00:51 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1392 (0.2723) acc1: 99.4792 (90.2524) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1568 data: 0.0645 max mem: 3540 +Test: [2120/2387] eta: 0:00:49 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1483 (0.2720) acc1: 99.4792 (90.2905) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1577 data: 0.0659 max mem: 3540 +Test: [2130/2387] eta: 0:00:47 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1735 (0.2715) acc1: 99.7396 (90.3332) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1606 data: 0.0692 max mem: 3540 +Test: [2140/2387] eta: 0:00:46 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1610 (0.2711) acc1: 99.2188 (90.3630) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1521 data: 0.0605 max mem: 3540 +Test: [2150/2387] eta: 0:00:44 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1611 (0.2706) acc1: 99.7396 (90.4064) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1501 data: 0.0591 max mem: 3540 +Test: [2160/2387] eta: 0:00:42 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1668 (0.2701) acc1: 100.0000 (90.4476) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1549 data: 0.0635 max mem: 3540 +Test: [2170/2387] eta: 0:00:40 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1900 (0.2698) acc1: 99.7396 (90.4871) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1559 data: 0.0626 max mem: 3540 +Test: [2180/2387] eta: 0:00:38 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2187 (0.2712) acc1: 97.6562 (90.4014) acc5: 100.0000 (100.0000) acc1_neg: 34.115 acc5_neg: 100.000 time: 0.1516 data: 0.0595 max mem: 3540 +Test: [2190/2387] eta: 0:00:36 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2538 (0.2715) acc1: 95.0521 (90.3894) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1498 data: 0.0588 max mem: 3540 +Test: [2200/2387] eta: 0:00:34 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1627 (0.2710) acc1: 99.7396 (90.4325) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1504 data: 0.0589 max mem: 3540 +Test: [2210/2387] eta: 0:00:32 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1382 (0.2704) acc1: 100.0000 (90.4757) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1585 data: 0.0680 max mem: 3540 +Test: [2220/2387] eta: 0:00:30 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1336 (0.2698) acc1: 100.0000 (90.5183) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1831 data: 0.0925 max mem: 3540 +Test: [2230/2387] eta: 0:00:29 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1381 (0.2692) acc1: 100.0000 (90.5607) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1898 data: 0.0967 max mem: 3540 +Test: [2240/2387] eta: 0:00:27 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1381 (0.2686) acc1: 100.0000 (90.6022) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1795 data: 0.0859 max mem: 3540 +Test: [2250/2387] eta: 0:00:25 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1514 (0.2682) acc1: 100.0000 (90.6424) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1835 data: 0.0901 max mem: 3540 +Test: [2260/2387] eta: 0:00:23 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1664 (0.2677) acc1: 100.0000 (90.6825) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1749 data: 0.0823 max mem: 3541 +Test: [2270/2387] eta: 0:00:21 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1549 (0.2672) acc1: 100.0000 (90.7234) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1761 data: 0.0850 max mem: 3541 +Test: [2280/2387] eta: 0:00:19 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1410 (0.2667) acc1: 100.0000 (90.7523) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1778 data: 0.0866 max mem: 3541 +Test: [2290/2387] eta: 0:00:17 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1355 (0.2661) acc1: 100.0000 (90.7922) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0790 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1407 (0.2657) acc1: 100.0000 (90.8313) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1719 data: 0.0805 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1584 (0.2653) acc1: 100.0000 (90.8685) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1588 data: 0.0676 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1710 (0.2650) acc1: 99.7396 (90.8994) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1432 data: 0.0532 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2021 (0.2648) acc1: 99.4792 (90.9316) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1360 data: 0.0464 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1851 (0.2644) acc1: 99.7396 (90.9675) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1317 data: 0.0421 max mem: 3541 +Test: [2350/2387] eta: 0:00:06 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1654 (0.2641) acc1: 100.0000 (90.9906) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1304 data: 0.0408 max mem: 3541 +Test: [2360/2387] eta: 0:00:04 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1686 (0.2638) acc1: 100.0000 (91.0214) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1286 data: 0.0390 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1897 (0.2636) acc1: 99.7396 (91.0546) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1276 data: 0.0381 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.1941 (0.2634) acc1: 98.9583 (91.0832) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1294 data: 0.0394 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 16.438 acc5_pos: 100.000 loss: 0.2372 (0.2634) acc1: 97.3958 (91.0907) acc5: 100.0000 (100.0000) acc1_neg: 95.349 acc5_neg: 100.000 time: 0.1273 data: 0.0397 max mem: 3541 +Test: Total time: 0:07:17 (0.1835 s / it) +* Overall Acc@1 91.091 Acc@5 100.000 + target == real Acc@1 11.608 Acc@5 100.000 + target == fake Acc@1 96.672 Acc@5 100.000 + loss 0.263 + PR-AUC 0.9892 ROC-AUC 0.8665 EER 19.79% (thr=0.7979) +Accuracy of the network on the 916267 test images: 91.1% +[rank0]:[W726 06:22:12.289942983 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 06:22:31.717000 3175846 torch/distributed/run.py:766] +W0726 06:22:31.717000 3175846 torch/distributed/run.py:766] ***************************************** +W0726 06:22:31.717000 3175846 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:22:31.717000 3175846 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 06:22:59.765828154 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:22:59.090478308 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:22:59.430699629 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:23:00.753704218 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:23:00.925860569 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:23:00.929402898 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:23:00.942119538 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:23:00.957319829 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n32_seed9_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed9_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:10:20 acc1_pos: 16.927 acc5_pos: 100.000 loss: 1.1719 (1.1719) acc1: 16.9271 (16.9271) acc5: 100.0000 (100.0000) time: 8.1642 data: 5.9459 max mem: 3531 +Test: [ 10/517] eta: 0:06:57 acc1_pos: 1.562 acc5_pos: 100.000 loss: 1.2280 (1.2356) acc1: 8.5938 (10.6771) acc5: 100.0000 (100.0000) time: 0.8240 data: 0.5409 max mem: 3531 +Test: [ 20/517] eta: 0:03:57 acc1_pos: 2.344 acc5_pos: 100.000 loss: 1.2907 (1.2934) acc1: 7.5521 (9.5486) acc5: 100.0000 (100.0000) time: 0.0934 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:03:09 acc1_pos: 0.781 acc5_pos: 100.000 loss: 1.3491 (1.3098) acc1: 7.0312 (8.6442) acc5: 100.0000 (100.0000) time: 0.1484 data: 0.0519 max mem: 3531 +Test: [ 40/517] eta: 0:02:49 acc1_pos: 7.292 acc5_pos: 100.000 loss: 1.3222 (1.3149) acc1: 2.0833 (7.6156) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1304 max mem: 3531 +Test: [ 50/517] eta: 0:02:34 acc1_pos: 33.028 acc5_pos: 100.000 loss: 1.2819 (1.2553) acc1: 7.2917 (12.5715) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2406 data: 0.1479 max mem: 3531 +Test: [ 60/517] eta: 0:02:24 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2792 (1.0913) acc1: 97.6562 (26.7717) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2404 data: 0.1489 max mem: 3531 +Test: [ 70/517] eta: 0:02:18 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2474 (0.9714) acc1: 99.7396 (37.0525) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2544 data: 0.1600 max mem: 3531 +Test: [ 80/517] eta: 0:02:11 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2420 (0.8831) acc1: 99.7396 (44.7113) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2556 data: 0.1595 max mem: 3531 +Test: [ 90/517] eta: 0:02:06 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2308 (0.8098) acc1: 99.7396 (50.7641) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2547 data: 0.1604 max mem: 3531 +Test: [100/517] eta: 0:02:02 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.1985 (0.7503) acc1: 100.0000 (55.6338) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2550 data: 0.1598 max mem: 3531 +Test: [110/517] eta: 0:01:57 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.1824 (0.6985) acc1: 100.0000 (59.6284) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2558 data: 0.1613 max mem: 3531 +Test: [120/517] eta: 0:01:53 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.1872 (0.6577) acc1: 100.0000 (62.9649) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2487 data: 0.1529 max mem: 3531 +Test: [130/517] eta: 0:01:49 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2073 (0.6237) acc1: 100.0000 (65.7900) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2390 data: 0.1422 max mem: 3531 +Test: [140/517] eta: 0:01:45 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2100 (0.5943) acc1: 100.0000 (68.2162) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2449 data: 0.1484 max mem: 3531 +Test: [150/517] eta: 0:01:40 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2103 (0.5690) acc1: 100.0000 (70.3194) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2299 data: 0.1360 max mem: 3531 +Test: [160/517] eta: 0:01:37 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2103 (0.5468) acc1: 100.0000 (72.1629) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2210 data: 0.1301 max mem: 3531 +Test: [170/517] eta: 0:01:33 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2073 (0.5274) acc1: 100.0000 (73.7893) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2349 data: 0.1399 max mem: 3531 +Test: [180/517] eta: 0:01:30 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2051 (0.5091) acc1: 100.0000 (75.2374) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2439 data: 0.1453 max mem: 3531 +Test: [190/517] eta: 0:01:27 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.1681 (0.4910) acc1: 100.0000 (76.5339) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2349 data: 0.1384 max mem: 3531 +Test: [200/517] eta: 0:01:23 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.1681 (0.4753) acc1: 100.0000 (77.6975) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2259 data: 0.1301 max mem: 3531 +Test: [210/517] eta: 0:01:20 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2133 (0.4646) acc1: 99.2188 (78.6483) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2299 data: 0.1356 max mem: 3532 +Test: [220/517] eta: 0:01:17 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2552 (0.4556) acc1: 96.8750 (79.4483) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2238 data: 0.1313 max mem: 3532 +Test: [230/517] eta: 0:01:14 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2504 (0.4461) acc1: 97.6562 (80.3075) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2249 data: 0.1315 max mem: 3532 +Test: [240/517] eta: 0:01:11 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.1938 (0.4346) acc1: 100.0000 (81.1225) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2399 data: 0.1474 max mem: 3532 +Test: [250/517] eta: 0:01:09 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.1775 (0.4254) acc1: 100.0000 (81.8652) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2449 data: 0.1507 max mem: 3532 +Test: [260/517] eta: 0:01:06 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2054 (0.4169) acc1: 99.7396 (82.5521) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2260 data: 0.1292 max mem: 3532 +Test: [270/517] eta: 0:01:03 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.1855 (0.4081) acc1: 100.0000 (83.1959) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2159 data: 0.1224 max mem: 3532 +Test: [280/517] eta: 0:01:00 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.1710 (0.3998) acc1: 100.0000 (83.7921) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2288 data: 0.1365 max mem: 3532 +Test: [290/517] eta: 0:00:57 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.1703 (0.3922) acc1: 100.0000 (84.3482) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2389 data: 0.1442 max mem: 3532 +Test: [300/517] eta: 0:00:55 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.1703 (0.3848) acc1: 100.0000 (84.8681) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2353 data: 0.1413 max mem: 3532 +Test: [310/517] eta: 0:00:52 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.1947 (0.3792) acc1: 100.0000 (85.3539) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2355 data: 0.1427 max mem: 3532 +Test: [320/517] eta: 0:00:49 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2076 (0.3738) acc1: 100.0000 (85.8101) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2295 data: 0.1364 max mem: 3532 +Test: [330/517] eta: 0:00:47 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.1916 (0.3681) acc1: 100.0000 (86.2388) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2293 data: 0.1324 max mem: 3532 +Test: [340/517] eta: 0:00:44 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.1762 (0.3631) acc1: 100.0000 (86.6340) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2375 data: 0.1418 max mem: 3532 +Test: [350/517] eta: 0:00:41 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.1972 (0.3587) acc1: 100.0000 (87.0081) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2412 data: 0.1503 max mem: 3532 +Test: [360/517] eta: 0:00:39 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2013 (0.3546) acc1: 100.0000 (87.3608) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2451 data: 0.1526 max mem: 3532 +Test: [370/517] eta: 0:00:37 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2013 (0.3504) acc1: 100.0000 (87.7015) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2651 data: 0.1721 max mem: 3532 +Test: [380/517] eta: 0:00:34 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2078 (0.3471) acc1: 100.0000 (88.0215) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2571 data: 0.1628 max mem: 3532 +Test: [390/517] eta: 0:00:31 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2196 (0.3439) acc1: 100.0000 (88.3265) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2348 data: 0.1397 max mem: 3532 +Test: [400/517] eta: 0:00:29 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2103 (0.3405) acc1: 100.0000 (88.6176) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2501 data: 0.1560 max mem: 3532 +Test: [410/517] eta: 0:00:26 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2213 (0.3381) acc1: 100.0000 (88.8933) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2450 data: 0.1506 max mem: 3532 +Test: [420/517] eta: 0:00:24 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2377 (0.3356) acc1: 100.0000 (89.1565) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2372 data: 0.1442 max mem: 3532 +Test: [430/517] eta: 0:00:21 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2307 (0.3336) acc1: 100.0000 (89.4021) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2385 data: 0.1450 max mem: 3532 +Test: [440/517] eta: 0:00:19 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2298 (0.3310) acc1: 100.0000 (89.6406) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2433 data: 0.1493 max mem: 3533 +Test: [450/517] eta: 0:00:16 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2209 (0.3285) acc1: 100.0000 (89.8703) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2508 data: 0.1547 max mem: 3533 +Test: [460/517] eta: 0:00:14 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2180 (0.3262) acc1: 100.0000 (90.0895) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2592 data: 0.1617 max mem: 3533 +Test: [470/517] eta: 0:00:11 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2130 (0.3237) acc1: 100.0000 (90.2993) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2546 data: 0.1562 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2059 (0.3213) acc1: 100.0000 (90.5005) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2498 data: 0.1532 max mem: 3533 +Test: [490/517] eta: 0:00:06 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2059 (0.3190) acc1: 100.0000 (90.6833) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2517 data: 0.1572 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2050 (0.3167) acc1: 100.0000 (90.8662) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2279 data: 0.1336 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2074 (0.3151) acc1: 100.0000 (91.0312) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1948 data: 0.1044 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.1825 (0.3136) acc1: 100.0000 (91.1307) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1851 data: 0.0931 max mem: 3533 +Test: Total time: 0:02:07 (0.2476 s / it) +* Overall Acc@1 91.131 Acc@5 100.000 + target == real Acc@1 8.225 Acc@5 100.000 + target == fake Acc@1 99.731 Acc@5 100.000 + loss 0.314 + PR-AUC 0.9657 ROC-AUC 0.7835 EER 29.06% (thr=0.7876) +Accuracy of the network on the 198427 test images: 91.1% +[rank0]:[W726 06:25:23.064968202 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 06:25:43.287000 3176408 torch/distributed/run.py:766] +W0726 06:25:43.287000 3176408 torch/distributed/run.py:766] ***************************************** +W0726 06:25:43.287000 3176408 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:25:43.287000 3176408 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:26:11.737786023 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:26:11.943170092 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:26:11.142697735 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +[rank0]:[W726 06:26:11.394960024 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:26:11.400442812 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:26:11.433549557 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank4]:[W726 06:26:11.441767885 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:26:11.555687575 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n32_seed9_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed9_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:27:53 acc1_pos: 20.312 acc5_pos: 100.000 loss: 0.9087 (0.9087) acc1: 20.3125 (20.3125) acc5: 100.0000 (100.0000) time: 4.3016 data: 2.3710 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/389] eta: 0:03:08 acc1_pos: 44.792 acc5_pos: 100.000 loss: 0.8791 (0.8623) acc1: 36.1979 (35.3693) acc5: 100.0000 (100.0000) time: 0.4976 data: 0.2408 max mem: 3531 +Test: [ 20/389] eta: 0:02:00 acc1_pos: 9.115 acc5_pos: 100.000 loss: 0.9587 (0.9397) acc1: 18.7500 (24.0699) acc5: 100.0000 (100.0000) time: 0.1286 data: 0.0375 max mem: 3531 +Test: [ 30/389] eta: 0:01:36 acc1_pos: 21.875 acc5_pos: 100.000 loss: 0.9469 (0.9332) acc1: 11.7188 (23.2359) acc5: 100.0000 (100.0000) time: 0.1439 data: 0.0502 max mem: 3531 +Test: [ 40/389] eta: 0:01:26 acc1_pos: 2.344 acc5_pos: 100.000 loss: 0.9469 (0.9557) acc1: 13.5417 (20.9413) acc5: 100.0000 (100.0000) time: 0.1638 data: 0.0666 max mem: 3531 +Test: [ 50/389] eta: 0:01:22 acc1_pos: 79.688 acc5_pos: 100.000 loss: 0.8545 (0.9141) acc1: 19.5312 (26.0417) acc5: 100.0000 (100.0000) time: 0.1999 data: 0.1040 max mem: 3531 +Test: [ 60/389] eta: 0:01:15 acc1_pos: 9.896 acc5_pos: 100.000 loss: 0.6550 (0.8687) acc1: 60.1562 (32.2106) acc5: 100.0000 (100.0000) time: 0.1949 data: 0.1012 max mem: 3531 +Test: [ 70/389] eta: 0:01:08 acc1_pos: 32.812 acc5_pos: 100.000 loss: 0.6576 (0.8616) acc1: 59.3750 (32.9189) acc5: 100.0000 (100.0000) time: 0.1449 data: 0.0495 max mem: 3531 +Test: [ 80/389] eta: 0:01:03 acc1_pos: 79.427 acc5_pos: 100.000 loss: 0.8636 (0.8556) acc1: 29.9479 (34.4875) acc5: 100.0000 (100.0000) time: 0.1299 data: 0.0348 max mem: 3531 +Test: [ 90/389] eta: 0:01:00 acc1_pos: 20.833 acc5_pos: 100.000 loss: 0.5842 (0.8308) acc1: 60.6771 (37.3827) acc5: 100.0000 (100.0000) time: 0.1649 data: 0.0698 max mem: 3531 +Test: [100/389] eta: 0:00:58 acc1_pos: 13.021 acc5_pos: 100.000 loss: 0.8457 (0.8458) acc1: 28.3854 (35.7390) acc5: 100.0000 (100.0000) time: 0.1849 data: 0.0907 max mem: 3531 +Test: [110/389] eta: 0:00:54 acc1_pos: 97.135 acc5_pos: 100.000 loss: 0.9147 (0.8462) acc1: 24.4792 (36.3035) acc5: 100.0000 (100.0000) time: 0.1649 data: 0.0715 max mem: 3531 +Test: [120/389] eta: 0:00:52 acc1_pos: 33.073 acc5_pos: 100.000 loss: 0.7987 (0.8428) acc1: 41.4062 (36.6693) acc5: 100.0000 (100.0000) time: 0.1699 data: 0.0764 max mem: 3531 +Test: [130/389] eta: 0:00:50 acc1_pos: 10.429 acc5_pos: 100.000 loss: 0.6035 (0.8198) acc1: 54.1667 (39.3249) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.1855 data: 0.0933 max mem: 3531 +Test: [140/389] eta: 0:00:48 acc1_pos: 10.429 acc5_pos: 100.000 loss: 0.3484 (0.7901) acc1: 92.4479 (42.8893) acc5: 100.0000 (100.0000) acc1_neg: 61.458 acc5_neg: 100.000 time: 0.1853 data: 0.0939 max mem: 3531 +Test: [150/389] eta: 0:00:46 acc1_pos: 10.429 acc5_pos: 100.000 loss: 0.3484 (0.7640) acc1: 97.1354 (46.0782) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1863 data: 0.0944 max mem: 3531 +Test: [160/389] eta: 0:00:44 acc1_pos: 10.429 acc5_pos: 100.000 loss: 0.3824 (0.7366) acc1: 96.3542 (48.9680) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1898 data: 0.0977 max mem: 3531 +Test: [170/389] eta: 0:00:42 acc1_pos: 10.429 acc5_pos: 100.000 loss: 0.2099 (0.7104) acc1: 100.0000 (51.7163) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1997 data: 0.1081 max mem: 3531 +Test: [180/389] eta: 0:00:40 acc1_pos: 10.429 acc5_pos: 100.000 loss: 0.2468 (0.6895) acc1: 98.4375 (54.1998) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1923 data: 0.1008 max mem: 3531 +Test: [190/389] eta: 0:00:38 acc1_pos: 10.429 acc5_pos: 100.000 loss: 0.3388 (0.6751) acc1: 95.5729 (55.8410) acc5: 100.0000 (100.0000) acc1_neg: 75.521 acc5_neg: 100.000 time: 0.1857 data: 0.0946 max mem: 3531 +Test: [200/389] eta: 0:00:36 acc1_pos: 10.429 acc5_pos: 100.000 loss: 0.3046 (0.6593) acc1: 99.2188 (57.8086) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1902 data: 0.0994 max mem: 3531 +Test: [210/389] eta: 0:00:34 acc1_pos: 10.429 acc5_pos: 100.000 loss: 0.2930 (0.6385) acc1: 100.0000 (59.8045) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1923 data: 0.1012 max mem: 3532 +Test: [220/389] eta: 0:00:32 acc1_pos: 10.429 acc5_pos: 100.000 loss: 0.2007 (0.6185) acc1: 100.0000 (61.6139) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1882 data: 0.0978 max mem: 3532 +Test: [230/389] eta: 0:00:30 acc1_pos: 10.429 acc5_pos: 100.000 loss: 0.3372 (0.6096) acc1: 99.2188 (62.8585) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1905 data: 0.0993 max mem: 3532 +Test: [240/389] eta: 0:00:28 acc1_pos: 10.429 acc5_pos: 100.000 loss: 0.3441 (0.5971) acc1: 95.5729 (64.2484) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1895 data: 0.0985 max mem: 3532 +Test: [250/389] eta: 0:00:26 acc1_pos: 10.429 acc5_pos: 100.000 loss: 0.2730 (0.5836) acc1: 99.7396 (65.6686) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1816 data: 0.0910 max mem: 3532 +Test: [260/389] eta: 0:00:24 acc1_pos: 10.429 acc5_pos: 100.000 loss: 0.2854 (0.5755) acc1: 99.7396 (66.5170) acc5: 100.0000 (100.0000) acc1_neg: 65.104 acc5_neg: 100.000 time: 0.1748 data: 0.0842 max mem: 3532 +Test: [270/389] eta: 0:00:22 acc1_pos: 10.429 acc5_pos: 100.000 loss: 0.3679 (0.5719) acc1: 94.2708 (67.1289) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1677 data: 0.0776 max mem: 3532 +Test: [280/389] eta: 0:00:20 acc1_pos: 10.429 acc5_pos: 100.000 loss: 0.3994 (0.5664) acc1: 89.8438 (67.6472) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1891 data: 0.0982 max mem: 3532 +Test: [290/389] eta: 0:00:18 acc1_pos: 10.429 acc5_pos: 100.000 loss: 0.2900 (0.5580) acc1: 98.4375 (68.4950) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1872 data: 0.0950 max mem: 3532 +Test: [300/389] eta: 0:00:16 acc1_pos: 10.429 acc5_pos: 100.000 loss: 0.3605 (0.5517) acc1: 98.6979 (69.4793) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1765 data: 0.0845 max mem: 3532 +Test: [310/389] eta: 0:00:15 acc1_pos: 10.429 acc5_pos: 100.000 loss: 0.3680 (0.5477) acc1: 96.6146 (70.0261) acc5: 100.0000 (100.0000) acc1_neg: 38.542 acc5_neg: 100.000 time: 0.1843 data: 0.0924 max mem: 3532 +Test: [320/389] eta: 0:00:13 acc1_pos: 10.429 acc5_pos: 100.000 loss: 0.4030 (0.5434) acc1: 92.1875 (70.5867) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1917 data: 0.1002 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 10.429 acc5_pos: 100.000 loss: 0.5609 (0.5436) acc1: 82.2917 (70.9443) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.1909 data: 0.1000 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 10.429 acc5_pos: 100.000 loss: 0.2955 (0.5358) acc1: 83.8542 (71.7162) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1787 data: 0.0874 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 10.429 acc5_pos: 100.000 loss: 0.2575 (0.5370) acc1: 99.7396 (71.5315) acc5: 100.0000 (100.0000) acc1_neg: 59.635 acc5_neg: 100.000 time: 0.1847 data: 0.0931 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 10.429 acc5_pos: 100.000 loss: 0.3568 (0.5340) acc1: 94.2708 (71.9305) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1844 data: 0.0934 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 10.429 acc5_pos: 100.000 loss: 0.3450 (0.5279) acc1: 94.7917 (72.5453) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1552 data: 0.0655 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 10.429 acc5_pos: 100.000 loss: 0.3221 (0.5240) acc1: 95.0521 (72.8675) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1335 data: 0.0438 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 10.429 acc5_pos: 100.000 loss: 0.2973 (0.5195) acc1: 95.0521 (73.3319) acc5: 100.0000 (100.0000) acc1_neg: 98.907 acc5_neg: 100.000 time: 0.1288 data: 0.0353 max mem: 3532 +Test: Total time: 0:01:12 (0.1856 s / it) +* Overall Acc@1 73.332 Acc@5 100.000 + target == real Acc@1 36.097 Acc@5 100.000 + target == fake Acc@1 90.640 Acc@5 100.000 + loss 0.520 + PR-AUC 0.8889 ROC-AUC 0.7890 EER 28.66% (thr=0.6533) +Accuracy of the network on the 149358 test images: 73.3% +[rank0]:[W726 06:27:35.986263242 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 06:27:53.894000 3176967 torch/distributed/run.py:766] +W0726 06:27:53.894000 3176967 torch/distributed/run.py:766] ***************************************** +W0726 06:27:53.894000 3176967 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:27:53.894000 3176967 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:28:22.293671741 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:28:23.616642810 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:28:23.803327261 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:28:23.926615020 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 06:28:23.049718786 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:28:23.078062265 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +NCCL version 2.26.2+cuda12.2 +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:28:23.188695434 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W726 06:28:23.188740222 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n32_seed9_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed9_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:29:12 acc1_pos: 72.396 acc5_pos: 100.000 loss: 0.5436 (0.5436) acc1: 72.3958 (72.3958) acc5: 100.0000 (100.0000) time: 5.2614 data: 3.1889 max mem: 3531 +Test: [ 10/333] eta: 0:03:00 acc1_pos: 81.510 acc5_pos: 100.000 loss: 0.4293 (0.4198) acc1: 75.5208 (78.0303) acc5: 100.0000 (100.0000) time: 0.5600 data: 0.2902 max mem: 3531 +Test: [ 20/333] eta: 0:01:45 acc1_pos: 88.281 acc5_pos: 100.000 loss: 0.4229 (0.4071) acc1: 77.0833 (78.6458) acc5: 100.0000 (100.0000) time: 0.0900 data: 0.0004 max mem: 3531 +Test: [ 30/333] eta: 0:01:31 acc1_pos: 49.740 acc5_pos: 100.000 loss: 0.5016 (0.5666) acc1: 70.0521 (66.2466) acc5: 100.0000 (100.0000) time: 0.1611 data: 0.0672 max mem: 3531 +Test: [ 40/333] eta: 0:01:28 acc1_pos: 36.719 acc5_pos: 100.000 loss: 0.9147 (0.6581) acc1: 36.7188 (58.8161) acc5: 100.0000 (100.0000) time: 0.2696 data: 0.1726 max mem: 3531 +Test: [ 50/333] eta: 0:01:22 acc1_pos: 40.365 acc5_pos: 100.000 loss: 0.9147 (0.7135) acc1: 36.9792 (54.5701) acc5: 100.0000 (100.0000) time: 0.2750 data: 0.1823 max mem: 3531 +Test: [ 60/333] eta: 0:01:15 acc1_pos: 25.521 acc5_pos: 100.000 loss: 0.9222 (0.7498) acc1: 36.9792 (51.6180) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1273 max mem: 3531 +Test: [ 70/333] eta: 0:01:13 acc1_pos: 29.427 acc5_pos: 100.000 loss: 0.9175 (0.7773) acc1: 35.1562 (49.4902) acc5: 100.0000 (100.0000) time: 0.2484 data: 0.1484 max mem: 3531 +Test: [ 80/333] eta: 0:01:09 acc1_pos: 45.312 acc5_pos: 100.000 loss: 0.8899 (0.7929) acc1: 39.0625 (48.1803) acc5: 100.0000 (100.0000) time: 0.2702 data: 0.1775 max mem: 3531 +Test: [ 90/333] eta: 0:01:04 acc1_pos: 47.656 acc5_pos: 100.000 loss: 0.8631 (0.7999) acc1: 40.8854 (47.5847) acc5: 100.0000 (100.0000) time: 0.2199 data: 0.1273 max mem: 3531 +Test: [100/333] eta: 0:01:00 acc1_pos: 48.177 acc5_pos: 100.000 loss: 0.8361 (0.8053) acc1: 41.9271 (47.1638) acc5: 100.0000 (100.0000) time: 0.1857 data: 0.0906 max mem: 3531 +Test: [110/333] eta: 0:00:55 acc1_pos: 51.042 acc5_pos: 100.000 loss: 0.8352 (0.8080) acc1: 42.1875 (46.9923) acc5: 100.0000 (100.0000) time: 0.1759 data: 0.0802 max mem: 3531 +Test: [120/333] eta: 0:00:51 acc1_pos: 45.833 acc5_pos: 100.000 loss: 0.8029 (0.8080) acc1: 41.9271 (46.8277) acc5: 100.0000 (100.0000) time: 0.1686 data: 0.0734 max mem: 3531 +Test: [130/333] eta: 0:00:47 acc1_pos: 58.854 acc5_pos: 100.000 loss: 0.7773 (0.8065) acc1: 45.8333 (46.9784) acc5: 100.0000 (100.0000) time: 0.1600 data: 0.0647 max mem: 3531 +Test: [140/333] eta: 0:00:44 acc1_pos: 63.542 acc5_pos: 100.000 loss: 0.7750 (0.8065) acc1: 46.8750 (46.8048) acc5: 100.0000 (100.0000) time: 0.1650 data: 0.0700 max mem: 3531 +Test: [150/333] eta: 0:00:41 acc1_pos: 39.844 acc5_pos: 100.000 loss: 0.8331 (0.8093) acc1: 42.4479 (46.5387) acc5: 100.0000 (100.0000) time: 0.1741 data: 0.0810 max mem: 3531 +Test: [160/333] eta: 0:00:39 acc1_pos: 36.184 acc5_pos: 100.000 loss: 0.8208 (0.8100) acc1: 43.7500 (46.6744) acc5: 100.0000 (100.0000) acc1_neg: 59.635 acc5_neg: 100.000 time: 0.1948 data: 0.0999 max mem: 3531 +Test: [170/333] eta: 0:00:36 acc1_pos: 36.184 acc5_pos: 100.000 loss: 0.7502 (0.7974) acc1: 57.8125 (48.0842) acc5: 100.0000 (100.0000) acc1_neg: 75.781 acc5_neg: 100.000 time: 0.1993 data: 0.1032 max mem: 3531 +Test: [180/333] eta: 0:00:33 acc1_pos: 36.184 acc5_pos: 100.000 loss: 0.5364 (0.7827) acc1: 75.5208 (49.6518) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.1723 data: 0.0794 max mem: 3531 +Test: [190/333] eta: 0:00:31 acc1_pos: 36.184 acc5_pos: 100.000 loss: 0.5101 (0.7671) acc1: 75.7812 (51.2421) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.1599 data: 0.0669 max mem: 3531 +Test: [200/333] eta: 0:00:28 acc1_pos: 36.184 acc5_pos: 100.000 loss: 0.4876 (0.7528) acc1: 77.3438 (52.6612) acc5: 100.0000 (100.0000) acc1_neg: 78.646 acc5_neg: 100.000 time: 0.1681 data: 0.0756 max mem: 3531 +Test: [210/333] eta: 0:00:26 acc1_pos: 36.184 acc5_pos: 100.000 loss: 0.5110 (0.7445) acc1: 77.3438 (53.6532) acc5: 100.0000 (100.0000) acc1_neg: 60.417 acc5_neg: 100.000 time: 0.1776 data: 0.0873 max mem: 3532 +Test: [220/333] eta: 0:00:23 acc1_pos: 36.184 acc5_pos: 100.000 loss: 0.5998 (0.7435) acc1: 65.6250 (53.9392) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1693 data: 0.0758 max mem: 3532 +Test: [230/333] eta: 0:00:21 acc1_pos: 36.184 acc5_pos: 100.000 loss: 0.3892 (0.7268) acc1: 85.1562 (55.4631) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1630 data: 0.0684 max mem: 3532 +Test: [240/333] eta: 0:00:19 acc1_pos: 36.184 acc5_pos: 100.000 loss: 0.3865 (0.7142) acc1: 85.9375 (56.6639) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.1701 data: 0.0774 max mem: 3532 +Test: [250/333] eta: 0:00:17 acc1_pos: 36.184 acc5_pos: 100.000 loss: 0.4412 (0.7061) acc1: 82.2917 (57.4795) acc5: 100.0000 (100.0000) acc1_neg: 69.271 acc5_neg: 100.000 time: 0.1817 data: 0.0900 max mem: 3532 +Test: [260/333] eta: 0:00:15 acc1_pos: 36.184 acc5_pos: 100.000 loss: 0.5585 (0.7017) acc1: 74.2188 (58.0570) acc5: 100.0000 (100.0000) acc1_neg: 76.042 acc5_neg: 100.000 time: 0.1750 data: 0.0844 max mem: 3532 +Test: [270/333] eta: 0:00:12 acc1_pos: 36.184 acc5_pos: 100.000 loss: 0.4981 (0.6930) acc1: 76.0417 (58.8667) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.1773 data: 0.0863 max mem: 3532 +Test: [280/333] eta: 0:00:10 acc1_pos: 36.184 acc5_pos: 100.000 loss: 0.4737 (0.6894) acc1: 77.0833 (59.2870) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.1797 data: 0.0889 max mem: 3532 +Test: [290/333] eta: 0:00:08 acc1_pos: 36.184 acc5_pos: 100.000 loss: 0.6892 (0.6900) acc1: 61.4583 (59.3231) acc5: 100.0000 (100.0000) acc1_neg: 61.198 acc5_neg: 100.000 time: 0.1653 data: 0.0744 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 36.184 acc5_pos: 100.000 loss: 0.7153 (0.6916) acc1: 60.1562 (59.3672) acc5: 100.0000 (100.0000) acc1_neg: 64.583 acc5_neg: 100.000 time: 0.1699 data: 0.0784 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 36.184 acc5_pos: 100.000 loss: 0.6530 (0.6845) acc1: 65.6250 (60.0357) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.1699 data: 0.0779 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 36.184 acc5_pos: 100.000 loss: 0.4536 (0.6789) acc1: 82.5521 (60.6227) acc5: 100.0000 (100.0000) acc1_neg: 56.771 acc5_neg: 100.000 time: 0.1549 data: 0.0630 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 36.184 acc5_pos: 100.000 loss: 0.7278 (0.6829) acc1: 60.6771 (60.4875) acc5: 100.0000 (100.0000) acc1_neg: 56.771 acc5_neg: 100.000 time: 0.1345 data: 0.0441 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 36.184 acc5_pos: 100.000 loss: 0.7793 (0.6837) acc1: 56.9014 (60.4671) acc5: 100.0000 (100.0000) acc1_neg: 56.901 acc5_neg: 100.000 time: 0.1356 data: 0.0424 max mem: 3532 +Test: Total time: 0:01:05 (0.1966 s / it) +* Overall Acc@1 60.467 Acc@5 100.000 + target == real Acc@1 46.409 Acc@5 100.000 + target == fake Acc@1 72.981 Acc@5 100.000 + loss 0.684 + PR-AUC 0.6788 ROC-AUC 0.6526 EER 39.48% (thr=0.5986) +Accuracy of the network on the 127843 test images: 60.5% +[rank0]:[W726 06:29:40.405775446 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 06:29:59.614000 3177526 torch/distributed/run.py:766] +W0726 06:29:59.614000 3177526 torch/distributed/run.py:766] ***************************************** +W0726 06:29:59.614000 3177526 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:29:59.614000 3177526 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:30:27.294724302 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:30:27.512013438 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:30:28.609131452 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:30:28.809606734 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:30:28.853984966 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +| distributed init (rank 2): env:// +[rank0]:[W726 06:30:28.943248632 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:30:28.953605989 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank4]:[W726 06:30:28.953694699 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n32_seed9_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed9_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:24:34 acc1_pos: 95.833 acc5_pos: 100.000 loss: 0.1955 (0.1955) acc1: 95.8333 (95.8333) acc5: 100.0000 (100.0000) time: 6.2745 data: 4.6344 max mem: 3531 +Test: [ 10/235] eta: 0:02:28 acc1_pos: 94.792 acc5_pos: 100.000 loss: 0.1955 (0.1931) acc1: 96.0938 (95.9517) acc5: 100.0000 (100.0000) time: 0.6579 data: 0.4217 max mem: 3531 +Test: [ 20/235] eta: 0:01:24 acc1_pos: 96.094 acc5_pos: 100.000 loss: 0.1894 (0.1900) acc1: 96.0938 (96.1186) acc5: 100.0000 (100.0000) time: 0.0981 data: 0.0004 max mem: 3531 +Test: [ 30/235] eta: 0:01:01 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.1888 (0.1884) acc1: 96.0938 (96.1274) acc5: 100.0000 (100.0000) time: 0.1055 data: 0.0063 max mem: 3531 +Test: [ 40/235] eta: 0:00:52 acc1_pos: 95.573 acc5_pos: 100.000 loss: 0.1877 (0.1886) acc1: 96.0938 (96.1636) acc5: 100.0000 (100.0000) time: 0.1399 data: 0.0426 max mem: 3531 +Test: [ 50/235] eta: 0:00:45 acc1_pos: 94.792 acc5_pos: 100.000 loss: 0.1892 (0.1900) acc1: 96.0938 (96.0121) acc5: 100.0000 (100.0000) time: 0.1593 data: 0.0628 max mem: 3531 +Test: [ 60/235] eta: 0:00:39 acc1_pos: 95.052 acc5_pos: 100.000 loss: 0.1973 (0.1912) acc1: 95.5729 (95.9657) acc5: 100.0000 (100.0000) time: 0.1352 data: 0.0406 max mem: 3531 +Test: [ 70/235] eta: 0:00:35 acc1_pos: 96.923 acc5_pos: 100.000 loss: 0.2057 (0.3552) acc1: 94.2708 (86.3630) acc5: 100.0000 (100.0000) acc1_neg: 18.750 acc5_neg: 100.000 time: 0.1365 data: 0.0428 max mem: 3531 +Test: [ 80/235] eta: 0:00:32 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.5016 (0.4987) acc1: 18.4896 (78.0028) acc5: 100.0000 (100.0000) acc1_neg: 16.667 acc5_neg: 100.000 time: 0.1546 data: 0.0598 max mem: 3531 +Test: [ 90/235] eta: 0:00:29 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.5268 (0.6107) acc1: 18.4896 (71.5631) acc5: 100.0000 (100.0000) acc1_neg: 19.010 acc5_neg: 100.000 time: 0.1738 data: 0.0804 max mem: 3531 +Test: [100/235] eta: 0:00:27 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.5268 (0.6986) acc1: 19.2708 (66.5197) acc5: 100.0000 (100.0000) acc1_neg: 17.708 acc5_neg: 100.000 time: 0.1799 data: 0.0872 max mem: 3531 +Test: [110/235] eta: 0:00:24 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.4910 (0.7714) acc1: 19.7917 (62.3311) acc5: 100.0000 (100.0000) acc1_neg: 19.792 acc5_neg: 100.000 time: 0.1594 data: 0.0665 max mem: 3531 +Test: [120/235] eta: 0:00:22 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.5407 (0.8355) acc1: 17.1875 (58.4065) acc5: 100.0000 (100.0000) acc1_neg: 17.448 acc5_neg: 100.000 time: 0.1506 data: 0.0558 max mem: 3531 +Test: [130/235] eta: 0:00:20 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.5500 (0.8899) acc1: 15.8854 (55.1626) acc5: 100.0000 (100.0000) acc1_neg: 16.927 acc5_neg: 100.000 time: 0.1553 data: 0.0580 max mem: 3531 +Test: [140/235] eta: 0:00:17 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.5698 (0.9390) acc1: 15.1042 (52.2810) acc5: 100.0000 (100.0000) acc1_neg: 12.500 acc5_neg: 100.000 time: 0.1551 data: 0.0599 max mem: 3531 +Test: [150/235] eta: 0:00:15 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.5576 (0.9786) acc1: 14.8438 (49.8241) acc5: 100.0000 (100.0000) acc1_neg: 17.188 acc5_neg: 100.000 time: 0.1596 data: 0.0652 max mem: 3531 +Test: [160/235] eta: 0:00:13 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.5458 (1.0143) acc1: 15.6250 (47.7048) acc5: 100.0000 (100.0000) acc1_neg: 17.969 acc5_neg: 100.000 time: 0.1640 data: 0.0678 max mem: 3531 +Test: [170/235] eta: 0:00:11 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.5012 (1.0259) acc1: 18.2292 (46.7516) acc5: 100.0000 (100.0000) acc1_neg: 28.906 acc5_neg: 100.000 time: 0.1600 data: 0.0641 max mem: 3531 +Test: [180/235] eta: 0:00:10 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.1822 (1.0332) acc1: 32.8125 (46.0002) acc5: 100.0000 (100.0000) acc1_neg: 29.688 acc5_neg: 100.000 time: 0.1549 data: 0.0604 max mem: 3531 +Test: [190/235] eta: 0:00:08 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.1441 (1.0388) acc1: 33.0729 (45.3970) acc5: 100.0000 (100.0000) acc1_neg: 36.458 acc5_neg: 100.000 time: 0.1598 data: 0.0650 max mem: 3531 +Test: [200/235] eta: 0:00:06 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.1659 (1.0453) acc1: 33.0729 (44.7230) acc5: 100.0000 (100.0000) acc1_neg: 29.948 acc5_neg: 100.000 time: 0.1699 data: 0.0735 max mem: 3531 +Test: [210/235] eta: 0:00:04 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.1659 (1.0499) acc1: 32.0312 (44.1980) acc5: 100.0000 (100.0000) acc1_neg: 32.552 acc5_neg: 100.000 time: 0.1660 data: 0.0698 max mem: 3532 +Test: [220/235] eta: 0:00:02 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.1543 (1.0553) acc1: 32.5521 (43.6734) acc5: 100.0000 (100.0000) acc1_neg: 32.292 acc5_neg: 100.000 time: 0.1699 data: 0.0733 max mem: 3532 +Test: [230/235] eta: 0:00:00 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.1456 (1.0589) acc1: 34.1146 (43.3261) acc5: 100.0000 (100.0000) acc1_neg: 36.458 acc5_neg: 100.000 time: 0.1556 data: 0.0590 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.1456 (1.0622) acc1: 34.1146 (43.1792) acc5: 100.0000 (100.0000) acc1_neg: 30.435 acc5_neg: 100.000 time: 0.1536 data: 0.0590 max mem: 3532 +Test: Total time: 0:00:41 (0.1774 s / it) +* Overall Acc@1 43.179 Acc@5 100.000 + target == real Acc@1 95.941 Acc@5 100.000 + target == fake Acc@1 24.096 Acc@5 100.000 + loss 1.062 + PR-AUC 0.8911 ROC-AUC 0.7597 EER 30.91% (thr=0.1614) +Accuracy of the network on the 89879 test images: 43.2% +[rank0]:[W726 06:31:27.726880939 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 06:31:45.226000 3178077 torch/distributed/run.py:766] +W0726 06:31:45.226000 3178077 torch/distributed/run.py:766] ***************************************** +W0726 06:31:45.226000 3178077 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:31:45.226000 3178077 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:32:12.003462039 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:32:13.660133102 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:32:13.857368785 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 06:32:13.863148176 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:32:13.921341698 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:32:13.949983685 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +[rank3]:[W726 06:32:13.996650499 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:32:13.006273847 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n32_seed9_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed9_resampled_n111-046-081_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:23:01 acc1_pos: 52.083 acc5_pos: 100.000 loss: 0.7851 (0.7851) acc1: 52.0833 (52.0833) acc5: 100.0000 (100.0000) time: 5.1375 data: 3.2601 max mem: 3531 +Test: [ 10/269] eta: 0:02:23 acc1_pos: 56.250 acc5_pos: 100.000 loss: 0.6840 (0.6693) acc1: 56.2500 (58.9725) acc5: 100.0000 (100.0000) time: 0.5550 data: 0.3030 max mem: 3531 +Test: [ 20/269] eta: 0:01:26 acc1_pos: 69.271 acc5_pos: 100.000 loss: 0.6474 (0.6362) acc1: 59.3750 (62.9340) acc5: 100.0000 (100.0000) time: 0.1090 data: 0.0173 max mem: 3531 +Test: [ 30/269] eta: 0:01:18 acc1_pos: 65.885 acc5_pos: 100.000 loss: 0.5824 (0.6257) acc1: 66.6667 (64.6841) acc5: 100.0000 (100.0000) time: 0.2007 data: 0.1073 max mem: 3531 +Test: [ 40/269] eta: 0:01:10 acc1_pos: 27.586 acc5_pos: 100.000 loss: 0.5354 (0.5927) acc1: 73.1771 (67.9688) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.2639 data: 0.1723 max mem: 3531 +Test: [ 50/269] eta: 0:01:02 acc1_pos: 27.586 acc5_pos: 100.000 loss: 0.4473 (0.5695) acc1: 85.4167 (70.5474) acc5: 100.0000 (100.0000) acc1_neg: 61.979 acc5_neg: 100.000 time: 0.2247 data: 0.1339 max mem: 3531 +Test: [ 60/269] eta: 0:00:56 acc1_pos: 27.586 acc5_pos: 100.000 loss: 0.4865 (0.5674) acc1: 77.8646 (70.7266) acc5: 100.0000 (100.0000) acc1_neg: 58.594 acc5_neg: 100.000 time: 0.1973 data: 0.1047 max mem: 3531 +Test: [ 70/269] eta: 0:00:52 acc1_pos: 27.586 acc5_pos: 100.000 loss: 0.5564 (0.5622) acc1: 73.6979 (71.6806) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.2077 data: 0.1128 max mem: 3531 +Test: [ 80/269] eta: 0:00:47 acc1_pos: 27.586 acc5_pos: 100.000 loss: 0.4718 (0.5497) acc1: 82.0312 (73.2960) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.1985 data: 0.1031 max mem: 3531 +Test: [ 90/269] eta: 0:00:43 acc1_pos: 27.586 acc5_pos: 100.000 loss: 0.5054 (0.5592) acc1: 78.9062 (72.3987) acc5: 100.0000 (100.0000) acc1_neg: 65.885 acc5_neg: 100.000 time: 0.1748 data: 0.0804 max mem: 3531 +Test: [100/269] eta: 0:00:40 acc1_pos: 27.586 acc5_pos: 100.000 loss: 0.6101 (0.5680) acc1: 63.0208 (71.2330) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.1899 data: 0.0945 max mem: 3531 +Test: [110/269] eta: 0:00:37 acc1_pos: 27.586 acc5_pos: 100.000 loss: 0.6291 (0.5730) acc1: 62.5000 (70.6668) acc5: 100.0000 (100.0000) acc1_neg: 58.333 acc5_neg: 100.000 time: 0.1823 data: 0.0869 max mem: 3531 +Test: [120/269] eta: 0:00:34 acc1_pos: 27.586 acc5_pos: 100.000 loss: 0.5918 (0.5733) acc1: 67.7083 (70.6074) acc5: 100.0000 (100.0000) acc1_neg: 60.417 acc5_neg: 100.000 time: 0.1705 data: 0.0758 max mem: 3531 +Test: [130/269] eta: 0:00:31 acc1_pos: 27.586 acc5_pos: 100.000 loss: 0.6385 (0.5806) acc1: 60.4167 (69.7857) acc5: 100.0000 (100.0000) acc1_neg: 48.958 acc5_neg: 100.000 time: 0.1822 data: 0.0895 max mem: 3531 +Test: [140/269] eta: 0:00:28 acc1_pos: 27.586 acc5_pos: 100.000 loss: 0.7090 (0.5950) acc1: 52.8646 (68.0888) acc5: 100.0000 (100.0000) acc1_neg: 25.781 acc5_neg: 100.000 time: 0.1871 data: 0.0963 max mem: 3531 +Test: [150/269] eta: 0:00:26 acc1_pos: 27.586 acc5_pos: 100.000 loss: 0.7740 (0.6047) acc1: 44.5312 (66.9254) acc5: 100.0000 (100.0000) acc1_neg: 52.865 acc5_neg: 100.000 time: 0.1926 data: 0.0994 max mem: 3531 +Test: [160/269] eta: 0:00:23 acc1_pos: 27.586 acc5_pos: 100.000 loss: 0.6148 (0.6039) acc1: 60.1562 (67.0209) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.1870 data: 0.0946 max mem: 3531 +Test: [170/269] eta: 0:00:21 acc1_pos: 27.586 acc5_pos: 100.000 loss: 0.5266 (0.5996) acc1: 72.6562 (67.6215) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.1803 data: 0.0863 max mem: 3531 +Test: [180/269] eta: 0:00:19 acc1_pos: 27.586 acc5_pos: 100.000 loss: 0.4315 (0.5878) acc1: 85.1562 (68.7745) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1816 data: 0.0844 max mem: 3531 +Test: [190/269] eta: 0:00:16 acc1_pos: 27.586 acc5_pos: 100.000 loss: 0.4315 (0.5860) acc1: 83.0729 (68.8795) acc5: 100.0000 (100.0000) acc1_neg: 72.135 acc5_neg: 100.000 time: 0.1895 data: 0.0942 max mem: 3531 +Test: [200/269] eta: 0:00:14 acc1_pos: 27.586 acc5_pos: 100.000 loss: 0.5895 (0.5884) acc1: 67.4479 (68.7565) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.1834 data: 0.0908 max mem: 3531 +Test: [210/269] eta: 0:00:12 acc1_pos: 27.586 acc5_pos: 100.000 loss: 0.6001 (0.5890) acc1: 65.6250 (68.6007) acc5: 100.0000 (100.0000) acc1_neg: 66.667 acc5_neg: 100.000 time: 0.1824 data: 0.0912 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 27.586 acc5_pos: 100.000 loss: 0.5774 (0.5879) acc1: 66.6667 (68.6428) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1897 data: 0.0955 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 27.586 acc5_pos: 100.000 loss: 0.4704 (0.5828) acc1: 78.3854 (69.1062) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.1680 data: 0.0740 max mem: 3532 +Test: [240/269] eta: 0:00:05 acc1_pos: 27.586 acc5_pos: 100.000 loss: 0.5228 (0.5833) acc1: 75.7812 (69.0893) acc5: 100.0000 (100.0000) acc1_neg: 66.406 acc5_neg: 100.000 time: 0.1492 data: 0.0584 max mem: 3532 +Test: [250/269] eta: 0:00:03 acc1_pos: 27.586 acc5_pos: 100.000 loss: 0.6085 (0.5860) acc1: 63.2812 (68.7448) acc5: 100.0000 (100.0000) acc1_neg: 35.938 acc5_neg: 100.000 time: 0.1371 data: 0.0465 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 27.586 acc5_pos: 100.000 loss: 0.6914 (0.5905) acc1: 54.6875 (68.1364) acc5: 100.0000 (100.0000) acc1_neg: 54.688 acc5_neg: 100.000 time: 0.1265 data: 0.0365 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 27.586 acc5_pos: 100.000 loss: 0.6165 (0.5883) acc1: 58.8542 (68.4103) acc5: 100.0000 (100.0000) acc1_neg: 94.488 acc5_neg: 100.000 time: 0.1260 data: 0.0329 max mem: 3532 +Test: Total time: 0:00:53 (0.1975 s / it) +* Overall Acc@1 68.410 Acc@5 100.000 + target == real Acc@1 63.576 Acc@5 100.000 + target == fake Acc@1 69.070 Acc@5 100.000 + loss 0.588 + PR-AUC 0.9469 ROC-AUC 0.7289 EER 33.57% (thr=0.5161) +Accuracy of the network on the 103293 test images: 68.4% +[rank0]:[W726 06:33:16.076077529 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T06:33:22Z] eval_done tag=fixed500k_fake_n32_seed9_resampled +eval_queue_done_utc=2026-07-26T06:33:22Z diff --git a/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_1958_seed5_9/eval_queue_n111-046-081.pid b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_1958_seed5_9/eval_queue_n111-046-081.pid new file mode 100644 index 0000000000000000000000000000000000000000..1959d6ebfce26ea7dc1895f494eff5ee31258be9 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_1958_seed5_9/eval_queue_n111-046-081.pid @@ -0,0 +1 @@ +3138767 diff --git a/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_1958_seed5_9/eval_queue_n116-040-168.log b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_1958_seed5_9/eval_queue_n116-040-168.log new file mode 100644 index 0000000000000000000000000000000000000000..48545c47a858d4105243c0455fc215d4d6178ab4 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_1958_seed5_9/eval_queue_n116-040-168.log @@ -0,0 +1,10270 @@ +eval_queue_started_utc=2026-07-26T02:59:37Z +hostname=n116-040-168 +train_tag=20260725_1958_seed5_9 +eval_tag=20260725_1958_seed5_9 +real_level=5 +fake_level=11 +eval_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9 +batch_per_gpu=256 +seeds=5 6 7 8 9 +[2026-07-26T02:59:37Z] eval_start tag=fixed500k_real_n5_seed5_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed5_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T02:59:37Z +hostname=n116-040-168 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed5_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n5_seed5_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 02:59:49.697000 3131966 torch/distributed/run.py:766] +W0726 02:59:49.697000 3131966 torch/distributed/run.py:766] ***************************************** +W0726 02:59:49.697000 3131966 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 02:59:49.697000 3131966 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:00:18.507668313 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:00:18.609531026 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:00:18.781082741 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:00:18.030079744 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:00:19.341336343 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:00:19.420169816 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:00:19.425063101 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:00:19.437197440 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n5_seed5_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed5_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:25:20 acc1_pos: 10.417 acc5_pos: 100.000 loss: 1.8890 (1.8890) acc1: 10.4167 (10.4167) acc5: 100.0000 (100.0000) time: 6.4984 data: 4.9254 max mem: 3531 +Test: [ 10/234] eta: 0:02:46 acc1_pos: 0.781 acc5_pos: 100.000 loss: 2.2219 (2.1737) acc1: 1.0417 (4.4744) acc5: 100.0000 (100.0000) time: 0.7446 data: 0.5139 max mem: 3531 +Test: [ 20/234] eta: 0:01:33 acc1_pos: 75.260 acc5_pos: 100.000 loss: 2.1422 (1.8949) acc1: 1.5625 (16.3566) acc5: 100.0000 (100.0000) time: 0.1339 data: 0.0366 max mem: 3531 +Test: [ 30/234] eta: 0:01:16 acc1_pos: 73.438 acc5_pos: 100.000 loss: 0.5297 (1.4633) acc1: 71.8750 (34.8034) acc5: 100.0000 (100.0000) time: 0.1692 data: 0.0733 max mem: 3531 +Test: [ 40/234] eta: 0:01:07 acc1_pos: 73.958 acc5_pos: 100.000 loss: 0.5220 (1.2447) acc1: 72.9167 (43.6928) acc5: 100.0000 (100.0000) time: 0.2552 data: 0.1617 max mem: 3531 +Test: [ 50/234] eta: 0:00:56 acc1_pos: 76.823 acc5_pos: 100.000 loss: 0.5527 (1.1134) acc1: 70.3125 (49.1524) acc5: 100.0000 (100.0000) time: 0.2069 data: 0.1155 max mem: 3531 +Test: [ 60/234] eta: 0:00:50 acc1_pos: 67.586 acc5_pos: 100.000 loss: 0.2938 (0.9522) acc1: 76.8229 (57.2532) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1696 data: 0.0786 max mem: 3531 +Test: [ 70/234] eta: 0:00:46 acc1_pos: 67.586 acc5_pos: 100.000 loss: 0.1191 (0.8534) acc1: 99.2188 (62.1626) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.2279 data: 0.1319 max mem: 3531 +Test: [ 80/234] eta: 0:00:42 acc1_pos: 67.586 acc5_pos: 100.000 loss: 0.2900 (0.7774) acc1: 90.1042 (65.9851) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2257 data: 0.1286 max mem: 3531 +Test: [ 90/234] eta: 0:00:38 acc1_pos: 67.586 acc5_pos: 100.000 loss: 0.1409 (0.7064) acc1: 97.3958 (69.5284) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2099 data: 0.1142 max mem: 3531 +Test: [100/234] eta: 0:00:35 acc1_pos: 67.586 acc5_pos: 100.000 loss: 0.1505 (0.6581) acc1: 97.1354 (71.8647) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2141 data: 0.1169 max mem: 3531 +Test: [110/234] eta: 0:00:30 acc1_pos: 67.586 acc5_pos: 100.000 loss: 0.1945 (0.6139) acc1: 94.0104 (74.0451) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1549 data: 0.0574 max mem: 3531 +Test: [120/234] eta: 0:00:27 acc1_pos: 67.586 acc5_pos: 100.000 loss: 0.1574 (0.5840) acc1: 95.5729 (75.4649) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1169 data: 0.0200 max mem: 3531 +Test: [130/234] eta: 0:00:24 acc1_pos: 67.586 acc5_pos: 100.000 loss: 0.1475 (0.5530) acc1: 96.8750 (76.9680) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1715 data: 0.0779 max mem: 3531 +Test: [140/234] eta: 0:00:21 acc1_pos: 67.586 acc5_pos: 100.000 loss: 0.1841 (0.5434) acc1: 90.6250 (77.2606) acc5: 100.0000 (100.0000) acc1_neg: 73.958 acc5_neg: 100.000 time: 0.1966 data: 0.1025 max mem: 3531 +Test: [150/234] eta: 0:00:19 acc1_pos: 67.586 acc5_pos: 100.000 loss: 0.2643 (0.5237) acc1: 90.1042 (78.1802) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1733 data: 0.0762 max mem: 3531 +Test: [160/234] eta: 0:00:16 acc1_pos: 67.586 acc5_pos: 100.000 loss: 0.1752 (0.5010) acc1: 95.5729 (79.3397) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1512 data: 0.0545 max mem: 3531 +Test: [170/234] eta: 0:00:13 acc1_pos: 67.586 acc5_pos: 100.000 loss: 0.1752 (0.4823) acc1: 96.0938 (80.3028) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1427 data: 0.0485 max mem: 3531 +Test: [180/234] eta: 0:00:11 acc1_pos: 67.586 acc5_pos: 100.000 loss: 0.1940 (0.4700) acc1: 94.5312 (80.8630) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1599 data: 0.0688 max mem: 3531 +Test: [190/234] eta: 0:00:09 acc1_pos: 67.586 acc5_pos: 100.000 loss: 0.3743 (0.4873) acc1: 72.1354 (79.6193) acc5: 100.0000 (100.0000) acc1_neg: 53.385 acc5_neg: 100.000 time: 0.1687 data: 0.0780 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 67.586 acc5_pos: 100.000 loss: 0.7783 (0.4809) acc1: 58.8542 (79.8495) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1661 data: 0.0734 max mem: 3531 +Test: [210/234] eta: 0:00:04 acc1_pos: 67.586 acc5_pos: 100.000 loss: 0.0954 (0.4639) acc1: 99.7396 (80.7810) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1499 data: 0.0562 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 67.586 acc5_pos: 100.000 loss: 0.1813 (0.4514) acc1: 98.4375 (81.5446) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1449 data: 0.0529 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 67.586 acc5_pos: 100.000 loss: 0.1955 (0.4411) acc1: 96.3542 (82.1068) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1406 data: 0.0500 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 67.586 acc5_pos: 100.000 loss: 0.2081 (0.4382) acc1: 95.0521 (82.2399) acc5: 100.0000 (100.0000) acc1_neg: 94.163 acc5_neg: 100.000 time: 0.1329 data: 0.0413 max mem: 3532 +Test: Total time: 0:00:47 (0.2013 s / it) +* Overall Acc@1 82.240 Acc@5 100.000 + target == real Acc@1 49.288 Acc@5 100.000 + target == fake Acc@1 91.527 Acc@5 100.000 + loss 0.438 + PR-AUC 0.8841 ROC-AUC 0.7620 EER 29.96% (thr=0.8550) +Accuracy of the network on the 89729 test images: 82.2% +[rank0]:[W726 03:01:16.097938524 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:01:35.881000 3132513 torch/distributed/run.py:766] +W0726 03:01:35.881000 3132513 torch/distributed/run.py:766] ***************************************** +W0726 03:01:35.881000 3132513 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:01:35.881000 3132513 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 7): env:// +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:02:03.581036695 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:02:03.595138668 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:02:03.727169541 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:02:03.990019072 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:02:04.204629597 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:02:04.249628094 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +[rank0]:[W726 03:02:04.273235110 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:02:04.279600694 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n5_seed5_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed5_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:34:24 acc1_pos: 13.021 acc5_pos: 100.000 loss: 1.7283 (1.7283) acc1: 13.0208 (13.0208) acc5: 100.0000 (100.0000) time: 6.8974 data: 5.5554 max mem: 3531 +Test: [ 10/2387] eta: 0:28:31 acc1_pos: 9.635 acc5_pos: 100.000 loss: 1.6385 (1.5941) acc1: 14.0625 (17.9214) acc5: 100.0000 (100.0000) time: 0.7201 data: 0.5142 max mem: 3531 +Test: [ 20/2387] eta: 0:16:43 acc1_pos: 4.948 acc5_pos: 100.000 loss: 1.6003 (1.6290) acc1: 14.5833 (16.8775) acc5: 100.0000 (100.0000) time: 0.1001 data: 0.0052 max mem: 3531 +Test: [ 30/2387] eta: 0:15:05 acc1_pos: 6.510 acc5_pos: 100.000 loss: 1.6944 (1.6754) acc1: 12.2396 (15.5914) acc5: 100.0000 (100.0000) time: 0.1991 data: 0.1054 max mem: 3531 +Test: [ 40/2387] eta: 0:14:41 acc1_pos: 2.865 acc5_pos: 100.000 loss: 2.0061 (1.7780) acc1: 6.5104 (12.5635) acc5: 100.0000 (100.0000) time: 0.3249 data: 0.2316 max mem: 3531 +Test: [ 50/2387] eta: 0:13:49 acc1_pos: 9.115 acc5_pos: 100.000 loss: 2.0664 (1.8240) acc1: 3.1250 (11.1060) acc5: 100.0000 (100.0000) time: 0.3097 data: 0.2125 max mem: 3531 +Test: [ 60/2387] eta: 0:13:11 acc1_pos: 9.896 acc5_pos: 100.000 loss: 2.0839 (1.8705) acc1: 5.2083 (10.3740) acc5: 100.0000 (100.0000) time: 0.2672 data: 0.1702 max mem: 3531 +Test: [ 70/2387] eta: 0:12:25 acc1_pos: 11.979 acc5_pos: 100.000 loss: 2.1352 (1.9147) acc1: 3.1250 (9.3933) acc5: 100.0000 (100.0000) time: 0.2370 data: 0.1427 max mem: 3531 +Test: [ 80/2387] eta: 0:11:58 acc1_pos: 3.646 acc5_pos: 100.000 loss: 2.2153 (1.9476) acc1: 2.3438 (8.6741) acc5: 100.0000 (100.0000) time: 0.2235 data: 0.1312 max mem: 3531 +Test: [ 90/2387] eta: 0:11:33 acc1_pos: 9.896 acc5_pos: 100.000 loss: 2.2153 (1.9771) acc1: 3.3854 (8.2446) acc5: 100.0000 (100.0000) time: 0.2329 data: 0.1390 max mem: 3531 +Test: [ 100/2387] eta: 0:11:14 acc1_pos: 0.521 acc5_pos: 100.000 loss: 2.3264 (2.0131) acc1: 1.3021 (7.5676) acc5: 100.0000 (100.0000) time: 0.2297 data: 0.1339 max mem: 3531 +Test: [ 110/2387] eta: 0:11:00 acc1_pos: 1.042 acc5_pos: 100.000 loss: 2.3385 (2.0392) acc1: 0.2604 (6.9820) acc5: 100.0000 (100.0000) time: 0.2358 data: 0.1399 max mem: 3531 +Test: [ 120/2387] eta: 0:10:46 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.3385 (2.0630) acc1: 0.0000 (6.5018) acc5: 100.0000 (100.0000) time: 0.2363 data: 0.1410 max mem: 3531 +Test: [ 130/2387] eta: 0:10:33 acc1_pos: 0.260 acc5_pos: 100.000 loss: 2.3468 (2.0815) acc1: 0.2604 (6.1625) acc5: 100.0000 (100.0000) time: 0.2290 data: 0.1350 max mem: 3531 +Test: [ 140/2387] eta: 0:10:23 acc1_pos: 1.302 acc5_pos: 100.000 loss: 2.2967 (2.0948) acc1: 0.2604 (5.8751) acc5: 100.0000 (100.0000) time: 0.2298 data: 0.1334 max mem: 3531 +Test: [ 150/2387] eta: 0:10:20 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.2324 (2.1038) acc1: 1.0417 (5.7430) acc5: 100.0000 (100.0000) time: 0.2562 data: 0.1598 max mem: 3531 +Test: [ 160/2387] eta: 0:09:56 acc1_pos: 0.457 acc5_pos: 100.000 loss: 2.2324 (2.0592) acc1: 1.8229 (8.1231) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1994 data: 0.1077 max mem: 3531 +Test: [ 170/2387] eta: 0:09:49 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1079 (1.9448) acc1: 99.2188 (13.4610) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1761 data: 0.0834 max mem: 3531 +Test: [ 180/2387] eta: 0:09:37 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1027 (1.8431) acc1: 99.4792 (18.2133) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2105 data: 0.1169 max mem: 3531 +Test: [ 190/2387] eta: 0:09:26 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1022 (1.7519) acc1: 99.4792 (22.4790) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1925 data: 0.0987 max mem: 3531 +Test: [ 200/2387] eta: 0:09:16 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1007 (1.6698) acc1: 99.7396 (26.3137) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1895 data: 0.0933 max mem: 3531 +Test: [ 210/2387] eta: 0:09:03 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1030 (1.5956) acc1: 99.4792 (29.7801) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1715 data: 0.0756 max mem: 3532 +Test: [ 220/2387] eta: 0:08:56 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1052 (1.5282) acc1: 99.4792 (32.9221) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1794 data: 0.0855 max mem: 3532 +Test: [ 230/2387] eta: 0:08:44 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1028 (1.4665) acc1: 99.2188 (35.8022) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1744 data: 0.0828 max mem: 3532 +Test: [ 240/2387] eta: 0:08:36 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0967 (1.4094) acc1: 99.7396 (38.4574) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1642 data: 0.0737 max mem: 3532 +Test: [ 250/2387] eta: 0:08:30 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0881 (1.3567) acc1: 100.0000 (40.8865) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1894 data: 0.0953 max mem: 3532 +Test: [ 260/2387] eta: 0:08:26 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0772 (1.3079) acc1: 100.0000 (43.1344) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2098 data: 0.1136 max mem: 3532 +Test: [ 270/2387] eta: 0:08:20 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0772 (1.2626) acc1: 100.0000 (45.2279) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2043 data: 0.1092 max mem: 3532 +Test: [ 280/2387] eta: 0:08:17 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0790 (1.2206) acc1: 100.0000 (47.1595) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2097 data: 0.1160 max mem: 3532 +Test: [ 290/2387] eta: 0:08:12 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0759 (1.1815) acc1: 100.0000 (48.9539) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2158 data: 0.1228 max mem: 3532 +Test: [ 300/2387] eta: 0:08:04 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0734 (1.1447) acc1: 100.0000 (50.6489) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1756 data: 0.0820 max mem: 3532 +Test: [ 310/2387] eta: 0:08:00 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0741 (1.1111) acc1: 100.0000 (52.2248) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0877 max mem: 3532 +Test: [ 320/2387] eta: 0:07:56 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1265 (1.0805) acc1: 100.0000 (53.7083) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2042 data: 0.1125 max mem: 3532 +Test: [ 330/2387] eta: 0:07:52 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1265 (1.0517) acc1: 100.0000 (55.1037) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2001 data: 0.1077 max mem: 3532 +Test: [ 340/2387] eta: 0:07:48 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1270 (1.0246) acc1: 100.0000 (56.4180) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2049 data: 0.1127 max mem: 3532 +Test: [ 350/2387] eta: 0:07:44 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1218 (0.9981) acc1: 100.0000 (57.6582) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1998 data: 0.1078 max mem: 3532 +Test: [ 360/2387] eta: 0:07:40 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0918 (0.9730) acc1: 100.0000 (58.8275) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1970 data: 0.1046 max mem: 3532 +Test: [ 370/2387] eta: 0:07:35 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0899 (0.9492) acc1: 100.0000 (59.9337) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1893 data: 0.0964 max mem: 3532 +Test: [ 380/2387] eta: 0:07:32 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0899 (0.9267) acc1: 100.0000 (60.9785) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1904 data: 0.0957 max mem: 3532 +Test: [ 390/2387] eta: 0:07:27 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0911 (0.9054) acc1: 99.7396 (61.9712) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1948 data: 0.1005 max mem: 3532 +Test: [ 400/2387] eta: 0:07:22 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0914 (0.8851) acc1: 99.7396 (62.9143) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1784 data: 0.0868 max mem: 3532 +Test: [ 410/2387] eta: 0:07:18 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0916 (0.8658) acc1: 99.7396 (63.8091) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1702 data: 0.0770 max mem: 3532 +Test: [ 420/2387] eta: 0:07:13 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0910 (0.8474) acc1: 99.7396 (64.6650) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1689 data: 0.0760 max mem: 3532 +Test: [ 430/2387] eta: 0:07:09 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0897 (0.8297) acc1: 100.0000 (65.4836) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1794 data: 0.0888 max mem: 3532 +Test: [ 440/2387] eta: 0:07:04 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0839 (0.8128) acc1: 100.0000 (66.2663) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1723 data: 0.0805 max mem: 3533 +Test: [ 450/2387] eta: 0:07:00 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0858 (0.7967) acc1: 100.0000 (67.0143) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1649 data: 0.0708 max mem: 3533 +Test: [ 460/2387] eta: 0:06:57 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0862 (0.7813) acc1: 100.0000 (67.7298) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1780 data: 0.0847 max mem: 3533 +Test: [ 470/2387] eta: 0:06:53 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0861 (0.7665) acc1: 100.0000 (68.4149) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1874 data: 0.0968 max mem: 3533 +Test: [ 480/2387] eta: 0:06:50 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0862 (0.7524) acc1: 100.0000 (69.0716) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1844 data: 0.0917 max mem: 3533 +Test: [ 490/2387] eta: 0:06:46 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0861 (0.7388) acc1: 100.0000 (69.7015) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1725 data: 0.0779 max mem: 3533 +Test: [ 500/2387] eta: 0:06:42 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0862 (0.7258) acc1: 100.0000 (70.3063) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0816 max mem: 3533 +Test: [ 510/2387] eta: 0:06:39 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0866 (0.7134) acc1: 100.0000 (70.8874) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1771 data: 0.0845 max mem: 3533 +Test: [ 520/2387] eta: 0:06:35 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0963 (0.7016) acc1: 100.0000 (71.4461) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1719 data: 0.0784 max mem: 3533 +Test: [ 530/2387] eta: 0:06:31 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0961 (0.6902) acc1: 100.0000 (71.9839) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1693 data: 0.0762 max mem: 3533 +Test: [ 540/2387] eta: 0:06:30 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0954 (0.6792) acc1: 100.0000 (72.5017) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1988 data: 0.1070 max mem: 3533 +Test: [ 550/2387] eta: 0:06:29 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0954 (0.6686) acc1: 100.0000 (73.0008) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2390 data: 0.1477 max mem: 3533 +Test: [ 560/2387] eta: 0:06:27 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0954 (0.6584) acc1: 100.0000 (73.4816) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2280 data: 0.1371 max mem: 3533 +Test: [ 570/2387] eta: 0:06:25 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0952 (0.6486) acc1: 100.0000 (73.9460) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2158 data: 0.1241 max mem: 3533 +Test: [ 580/2387] eta: 0:06:22 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0949 (0.6390) acc1: 100.0000 (74.3945) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1965 data: 0.1045 max mem: 3533 +Test: [ 590/2387] eta: 0:06:21 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0950 (0.6298) acc1: 100.0000 (74.8277) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2086 data: 0.1167 max mem: 3533 +Test: [ 600/2387] eta: 0:06:18 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0967 (0.6211) acc1: 100.0000 (75.2466) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2263 data: 0.1344 max mem: 3533 +Test: [ 610/2387] eta: 0:06:17 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1044 (0.6126) acc1: 100.0000 (75.6461) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2171 data: 0.1239 max mem: 3533 +Test: [ 620/2387] eta: 0:06:14 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1041 (0.6044) acc1: 99.7396 (76.0337) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2050 data: 0.1121 max mem: 3533 +Test: [ 630/2387] eta: 0:06:11 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1041 (0.5965) acc1: 99.4792 (76.4040) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1904 data: 0.0995 max mem: 3533 +Test: [ 640/2387] eta: 0:06:09 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1044 (0.5889) acc1: 99.4792 (76.7648) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2079 data: 0.1173 max mem: 3533 +Test: [ 650/2387] eta: 0:06:06 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1046 (0.5815) acc1: 99.4792 (77.1125) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2017 data: 0.1095 max mem: 3533 +Test: [ 660/2387] eta: 0:06:03 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1047 (0.5743) acc1: 99.4792 (77.4521) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1806 data: 0.0884 max mem: 3533 +Test: [ 670/2387] eta: 0:06:00 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1044 (0.5673) acc1: 99.4792 (77.7811) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1714 data: 0.0807 max mem: 3534 +Test: [ 680/2387] eta: 0:05:57 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1005 (0.5603) acc1: 100.0000 (78.1070) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1695 data: 0.0779 max mem: 3534 +Test: [ 690/2387] eta: 0:05:54 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1004 (0.5563) acc1: 100.0000 (78.2467) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.1683 data: 0.0752 max mem: 3534 +Test: [ 700/2387] eta: 0:05:51 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.3503 (0.5536) acc1: 83.5938 (78.3078) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1692 data: 0.0752 max mem: 3534 +Test: [ 710/2387] eta: 0:05:48 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.3632 (0.5509) acc1: 82.8125 (78.3726) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.1725 data: 0.0798 max mem: 3534 +Test: [ 720/2387] eta: 0:05:46 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.3657 (0.5483) acc1: 82.8125 (78.4338) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.1879 data: 0.0969 max mem: 3534 +Test: [ 730/2387] eta: 0:05:44 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.3672 (0.5459) acc1: 83.3333 (78.5008) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.2125 data: 0.1206 max mem: 3534 +Test: [ 740/2387] eta: 0:05:41 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.3562 (0.5433) acc1: 83.8542 (78.5675) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.1993 data: 0.1075 max mem: 3534 +Test: [ 750/2387] eta: 0:05:39 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.3531 (0.5408) acc1: 83.8542 (78.6417) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.1828 data: 0.0915 max mem: 3534 +Test: [ 760/2387] eta: 0:05:37 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.3531 (0.5375) acc1: 83.5938 (78.7543) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1983 data: 0.1061 max mem: 3534 +Test: [ 770/2387] eta: 0:05:35 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1772 (0.5325) acc1: 94.0104 (78.9822) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2077 data: 0.1159 max mem: 3534 +Test: [ 780/2387] eta: 0:05:33 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1175 (0.5271) acc1: 98.9583 (79.2417) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2065 data: 0.1147 max mem: 3534 +Test: [ 790/2387] eta: 0:05:30 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1027 (0.5217) acc1: 99.7396 (79.4975) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1970 data: 0.1054 max mem: 3534 +Test: [ 800/2387] eta: 0:05:28 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0951 (0.5167) acc1: 99.7396 (79.7431) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1941 data: 0.1037 max mem: 3534 +Test: [ 810/2387] eta: 0:05:26 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0940 (0.5116) acc1: 99.7396 (79.9803) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2002 data: 0.1077 max mem: 3534 +Test: [ 820/2387] eta: 0:05:24 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0857 (0.5064) acc1: 100.0000 (80.2229) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2069 data: 0.1126 max mem: 3534 +Test: [ 830/2387] eta: 0:05:21 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0869 (0.5014) acc1: 100.0000 (80.4537) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1988 data: 0.1065 max mem: 3534 +Test: [ 840/2387] eta: 0:05:18 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1048 (0.4971) acc1: 99.2188 (80.6595) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1746 data: 0.0831 max mem: 3534 +Test: [ 850/2387] eta: 0:05:16 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1323 (0.4931) acc1: 98.1771 (80.8522) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1676 data: 0.0757 max mem: 3534 +Test: [ 860/2387] eta: 0:05:13 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1521 (0.4895) acc1: 96.6146 (81.0135) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1724 data: 0.0799 max mem: 3534 +Test: [ 870/2387] eta: 0:05:10 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1481 (0.4855) acc1: 97.1354 (81.1968) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1673 data: 0.0738 max mem: 3534 +Test: [ 880/2387] eta: 0:05:08 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1342 (0.4817) acc1: 98.4375 (81.3848) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1668 data: 0.0731 max mem: 3534 +Test: [ 890/2387] eta: 0:05:06 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0959 (0.4773) acc1: 99.7396 (81.5911) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1838 data: 0.0912 max mem: 3534 +Test: [ 900/2387] eta: 0:05:03 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1055 (0.4747) acc1: 98.6979 (81.7246) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.1970 data: 0.1058 max mem: 3535 +Test: [ 910/2387] eta: 0:05:02 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.2679 (0.4724) acc1: 92.4479 (81.8392) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.2068 data: 0.1146 max mem: 3535 +Test: [ 920/2387] eta: 0:04:59 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.2684 (0.4703) acc1: 92.4479 (81.9478) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1973 data: 0.1057 max mem: 3535 +Test: [ 930/2387] eta: 0:04:57 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.2679 (0.4681) acc1: 92.4479 (82.0617) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1889 data: 0.0989 max mem: 3535 +Test: [ 940/2387] eta: 0:04:55 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.2658 (0.4661) acc1: 91.9271 (82.1569) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.2034 data: 0.1131 max mem: 3535 +Test: [ 950/2387] eta: 0:04:52 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.2672 (0.4640) acc1: 91.4062 (82.2621) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1897 data: 0.0977 max mem: 3535 +Test: [ 960/2387] eta: 0:04:50 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.2644 (0.4620) acc1: 92.1875 (82.3656) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1655 data: 0.0734 max mem: 3535 +Test: [ 970/2387] eta: 0:04:48 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.2700 (0.4601) acc1: 91.6667 (82.4585) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1828 data: 0.0897 max mem: 3535 +Test: [ 980/2387] eta: 0:04:46 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.2622 (0.4571) acc1: 92.4479 (82.6031) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.2094 data: 0.1143 max mem: 3535 +Test: [ 990/2387] eta: 0:04:44 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1058 (0.4535) acc1: 99.7396 (82.7710) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2058 data: 0.1132 max mem: 3535 +Test: [1000/2387] eta: 0:04:42 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0964 (0.4501) acc1: 100.0000 (82.9384) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1951 data: 0.1044 max mem: 3535 +Test: [1010/2387] eta: 0:04:39 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0964 (0.4465) acc1: 99.7396 (83.1018) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1954 data: 0.1044 max mem: 3535 +Test: [1020/2387] eta: 0:04:37 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0862 (0.4432) acc1: 99.7396 (83.2599) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2027 data: 0.1103 max mem: 3535 +Test: [1030/2387] eta: 0:04:35 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0998 (0.4399) acc1: 100.0000 (83.4200) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1967 data: 0.1040 max mem: 3535 +Test: [1040/2387] eta: 0:04:33 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0959 (0.4366) acc1: 100.0000 (83.5735) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1971 data: 0.1056 max mem: 3535 +Test: [1050/2387] eta: 0:04:31 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0959 (0.4334) acc1: 99.7396 (83.7226) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2016 data: 0.1103 max mem: 3535 +Test: [1060/2387] eta: 0:04:29 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1003 (0.4303) acc1: 99.2188 (83.8704) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1999 data: 0.1080 max mem: 3535 +Test: [1070/2387] eta: 0:04:27 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0877 (0.4271) acc1: 100.0000 (84.0202) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2077 data: 0.1158 max mem: 3535 +Test: [1080/2387] eta: 0:04:26 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0877 (0.4240) acc1: 100.0000 (84.1642) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2337 data: 0.1421 max mem: 3535 +Test: [1090/2387] eta: 0:04:23 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1070 (0.4214) acc1: 99.4792 (84.3001) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2140 data: 0.1212 max mem: 3535 +Test: [1100/2387] eta: 0:04:21 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1209 (0.4186) acc1: 99.4792 (84.4382) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1901 data: 0.0976 max mem: 3535 +Test: [1110/2387] eta: 0:04:20 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1286 (0.4161) acc1: 99.4792 (84.5681) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2203 data: 0.1293 max mem: 3535 +Test: [1120/2387] eta: 0:04:18 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1298 (0.4138) acc1: 98.9583 (84.6842) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2305 data: 0.1380 max mem: 3536 +Test: [1130/2387] eta: 0:04:16 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1151 (0.4110) acc1: 99.7396 (84.8180) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2056 data: 0.1131 max mem: 3536 +Test: [1140/2387] eta: 0:04:14 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0914 (0.4082) acc1: 100.0000 (84.9502) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1980 data: 0.1065 max mem: 3536 +Test: [1150/2387] eta: 0:04:11 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0914 (0.4056) acc1: 100.0000 (85.0743) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1797 data: 0.0872 max mem: 3536 +Test: [1160/2387] eta: 0:04:09 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0935 (0.4029) acc1: 100.0000 (85.2025) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1824 data: 0.0897 max mem: 3536 +Test: [1170/2387] eta: 0:04:07 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0948 (0.4003) acc1: 100.0000 (85.3270) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1898 data: 0.0975 max mem: 3536 +Test: [1180/2387] eta: 0:04:05 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0960 (0.3977) acc1: 100.0000 (85.4508) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1947 data: 0.1019 max mem: 3536 +Test: [1190/2387] eta: 0:04:03 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0890 (0.3950) acc1: 100.0000 (85.5728) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2321 data: 0.1376 max mem: 3536 +Test: [1200/2387] eta: 0:04:01 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0855 (0.3925) acc1: 100.0000 (85.6916) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2015 data: 0.1068 max mem: 3536 +Test: [1210/2387] eta: 0:03:59 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0817 (0.3899) acc1: 100.0000 (85.8070) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1948 data: 0.1019 max mem: 3536 +Test: [1220/2387] eta: 0:03:57 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0817 (0.3875) acc1: 100.0000 (85.9179) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2174 data: 0.1262 max mem: 3536 +Test: [1230/2387] eta: 0:03:55 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0862 (0.3851) acc1: 100.0000 (86.0270) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2104 data: 0.1196 max mem: 3536 +Test: [1240/2387] eta: 0:03:53 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0921 (0.3828) acc1: 100.0000 (86.1345) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2102 data: 0.1188 max mem: 3536 +Test: [1250/2387] eta: 0:03:51 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0926 (0.3805) acc1: 99.4792 (86.2362) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2011 data: 0.1102 max mem: 3536 +Test: [1260/2387] eta: 0:03:49 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0804 (0.3781) acc1: 100.0000 (86.3452) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2043 data: 0.1140 max mem: 3536 +Test: [1270/2387] eta: 0:03:47 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0738 (0.3757) acc1: 100.0000 (86.4524) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2102 data: 0.1197 max mem: 3536 +Test: [1280/2387] eta: 0:03:45 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0730 (0.3734) acc1: 100.0000 (86.5567) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2049 data: 0.1118 max mem: 3536 +Test: [1290/2387] eta: 0:03:43 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0817 (0.3712) acc1: 99.7396 (86.6582) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2037 data: 0.1105 max mem: 3536 +Test: [1300/2387] eta: 0:03:41 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0837 (0.3690) acc1: 99.7396 (86.7570) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1982 data: 0.1057 max mem: 3536 +Test: [1310/2387] eta: 0:03:39 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0832 (0.3668) acc1: 99.7396 (86.8556) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2148 data: 0.1203 max mem: 3536 +Test: [1320/2387] eta: 0:03:37 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0788 (0.3646) acc1: 100.0000 (86.9547) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2384 data: 0.1459 max mem: 3536 +Test: [1330/2387] eta: 0:03:35 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0762 (0.3625) acc1: 100.0000 (87.0520) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1991 data: 0.1083 max mem: 3536 +Test: [1340/2387] eta: 0:03:33 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0825 (0.3613) acc1: 100.0000 (87.0934) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1731 data: 0.0809 max mem: 3536 +Test: [1350/2387] eta: 0:03:30 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0804 (0.3592) acc1: 100.0000 (87.1889) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1851 data: 0.0927 max mem: 3537 +Test: [1360/2387] eta: 0:03:28 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0784 (0.3572) acc1: 100.0000 (87.2830) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1599 data: 0.0655 max mem: 3537 +Test: [1370/2387] eta: 0:03:26 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0775 (0.3551) acc1: 100.0000 (87.3756) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1562 data: 0.0628 max mem: 3537 +Test: [1380/2387] eta: 0:03:23 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0768 (0.3531) acc1: 100.0000 (87.4670) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1748 data: 0.0826 max mem: 3537 +Test: [1390/2387] eta: 0:03:21 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0752 (0.3511) acc1: 100.0000 (87.5571) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1833 data: 0.0900 max mem: 3537 +Test: [1400/2387] eta: 0:03:19 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0731 (0.3491) acc1: 100.0000 (87.6459) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1770 data: 0.0849 max mem: 3537 +Test: [1410/2387] eta: 0:03:17 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0722 (0.3474) acc1: 100.0000 (87.7270) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1735 data: 0.0815 max mem: 3537 +Test: [1420/2387] eta: 0:03:15 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0982 (0.3457) acc1: 99.7396 (87.8119) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1685 data: 0.0763 max mem: 3537 +Test: [1430/2387] eta: 0:03:12 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1115 (0.3442) acc1: 99.4792 (87.8809) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1620 data: 0.0707 max mem: 3537 +Test: [1440/2387] eta: 0:03:10 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0913 (0.3424) acc1: 99.7396 (87.9643) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1725 data: 0.0812 max mem: 3537 +Test: [1450/2387] eta: 0:03:08 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0851 (0.3407) acc1: 100.0000 (88.0426) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1743 data: 0.0804 max mem: 3537 +Test: [1460/2387] eta: 0:03:06 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1047 (0.3393) acc1: 99.7396 (88.1125) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1716 data: 0.0783 max mem: 3537 +Test: [1470/2387] eta: 0:03:04 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1078 (0.3377) acc1: 99.7396 (88.1865) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1812 data: 0.0902 max mem: 3537 +Test: [1480/2387] eta: 0:03:01 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1162 (0.3364) acc1: 99.2188 (88.2510) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1769 data: 0.0854 max mem: 3537 +Test: [1490/2387] eta: 0:02:59 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1199 (0.3349) acc1: 98.6979 (88.3228) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1646 data: 0.0720 max mem: 3537 +Test: [1500/2387] eta: 0:02:57 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1085 (0.3333) acc1: 99.7396 (88.3952) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1688 data: 0.0764 max mem: 3537 +Test: [1510/2387] eta: 0:02:55 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0898 (0.3317) acc1: 99.7396 (88.4707) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1749 data: 0.0827 max mem: 3537 +Test: [1520/2387] eta: 0:02:53 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0908 (0.3304) acc1: 100.0000 (88.5333) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1907 data: 0.0988 max mem: 3537 +Test: [1530/2387] eta: 0:02:51 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1116 (0.3291) acc1: 98.9583 (88.5958) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1792 data: 0.0880 max mem: 3537 +Test: [1540/2387] eta: 0:02:49 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1251 (0.3280) acc1: 98.6979 (88.6393) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1698 data: 0.0785 max mem: 3537 +Test: [1550/2387] eta: 0:02:46 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1306 (0.3268) acc1: 98.4375 (88.6953) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1779 data: 0.0870 max mem: 3537 +Test: [1560/2387] eta: 0:02:44 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1273 (0.3255) acc1: 98.4375 (88.7552) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1784 data: 0.0868 max mem: 3537 +Test: [1570/2387] eta: 0:02:42 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0898 (0.3241) acc1: 100.0000 (88.8245) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1793 data: 0.0873 max mem: 3537 +Test: [1580/2387] eta: 0:02:40 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1024 (0.3228) acc1: 99.7396 (88.8859) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1765 data: 0.0847 max mem: 3538 +Test: [1590/2387] eta: 0:02:38 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1093 (0.3215) acc1: 99.4792 (88.9522) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1710 data: 0.0792 max mem: 3538 +Test: [1600/2387] eta: 0:02:36 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1253 (0.3203) acc1: 99.4792 (89.0150) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1692 data: 0.0777 max mem: 3538 +Test: [1610/2387] eta: 0:02:34 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1087 (0.3191) acc1: 99.2188 (89.0759) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1709 data: 0.0794 max mem: 3538 +Test: [1620/2387] eta: 0:02:32 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0990 (0.3178) acc1: 99.4792 (89.1317) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1839 data: 0.0930 max mem: 3538 +Test: [1630/2387] eta: 0:02:30 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0891 (0.3165) acc1: 100.0000 (89.1965) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1973 data: 0.1057 max mem: 3538 +Test: [1640/2387] eta: 0:02:28 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0939 (0.3151) acc1: 99.4792 (89.2601) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1925 data: 0.1001 max mem: 3538 +Test: [1650/2387] eta: 0:02:26 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0880 (0.3138) acc1: 99.7396 (89.3221) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1940 data: 0.1026 max mem: 3538 +Test: [1660/2387] eta: 0:02:24 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0840 (0.3124) acc1: 100.0000 (89.3852) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2022 data: 0.1093 max mem: 3538 +Test: [1670/2387] eta: 0:02:22 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0835 (0.3111) acc1: 100.0000 (89.4468) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2117 data: 0.1171 max mem: 3538 +Test: [1680/2387] eta: 0:02:20 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0918 (0.3099) acc1: 99.7396 (89.5039) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2058 data: 0.1111 max mem: 3538 +Test: [1690/2387] eta: 0:02:18 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0980 (0.3086) acc1: 99.4792 (89.5622) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1980 data: 0.1003 max mem: 3538 +Test: [1700/2387] eta: 0:02:16 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0907 (0.3073) acc1: 99.7396 (89.6207) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2050 data: 0.1059 max mem: 3538 +Test: [1710/2387] eta: 0:02:14 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0886 (0.3061) acc1: 99.7396 (89.6797) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2112 data: 0.1170 max mem: 3538 +Test: [1720/2387] eta: 0:02:12 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0903 (0.3049) acc1: 99.7396 (89.7360) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2160 data: 0.1236 max mem: 3538 +Test: [1730/2387] eta: 0:02:10 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0909 (0.3037) acc1: 99.7396 (89.7908) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2071 data: 0.1120 max mem: 3538 +Test: [1740/2387] eta: 0:02:08 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0937 (0.3026) acc1: 99.4792 (89.8433) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1978 data: 0.1028 max mem: 3538 +Test: [1750/2387] eta: 0:02:06 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0880 (0.3013) acc1: 100.0000 (89.9007) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1999 data: 0.1063 max mem: 3538 +Test: [1760/2387] eta: 0:02:04 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0752 (0.3000) acc1: 100.0000 (89.9579) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1971 data: 0.1044 max mem: 3538 +Test: [1770/2387] eta: 0:02:02 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0794 (0.2990) acc1: 99.7396 (90.0045) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2027 data: 0.1115 max mem: 3538 +Test: [1780/2387] eta: 0:02:00 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1184 (0.2981) acc1: 98.4375 (90.0488) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2028 data: 0.1126 max mem: 3538 +Test: [1790/2387] eta: 0:01:58 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1205 (0.2971) acc1: 98.6979 (90.0960) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2021 data: 0.1122 max mem: 3538 +Test: [1800/2387] eta: 0:01:56 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1214 (0.2962) acc1: 98.9583 (90.1396) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1910 data: 0.1006 max mem: 3538 +Test: [1810/2387] eta: 0:01:54 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1055 (0.2951) acc1: 99.2188 (90.1900) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1749 data: 0.0839 max mem: 3539 +Test: [1820/2387] eta: 0:01:52 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0857 (0.2940) acc1: 100.0000 (90.2429) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1751 data: 0.0836 max mem: 3539 +Test: [1830/2387] eta: 0:01:50 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0835 (0.2929) acc1: 100.0000 (90.2959) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1656 data: 0.0723 max mem: 3539 +Test: [1840/2387] eta: 0:01:48 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0891 (0.2918) acc1: 100.0000 (90.3485) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1872 data: 0.0931 max mem: 3539 +Test: [1850/2387] eta: 0:01:46 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0900 (0.2907) acc1: 100.0000 (90.4005) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2104 data: 0.1188 max mem: 3539 +Test: [1860/2387] eta: 0:01:44 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0897 (0.2896) acc1: 100.0000 (90.4520) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2043 data: 0.1133 max mem: 3539 +Test: [1870/2387] eta: 0:01:42 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0894 (0.2885) acc1: 100.0000 (90.5031) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1975 data: 0.1055 max mem: 3539 +Test: [1880/2387] eta: 0:01:40 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0896 (0.2875) acc1: 100.0000 (90.5534) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1969 data: 0.1055 max mem: 3539 +Test: [1890/2387] eta: 0:01:38 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0896 (0.2864) acc1: 100.0000 (90.6032) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2031 data: 0.1128 max mem: 3539 +Test: [1900/2387] eta: 0:01:36 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0898 (0.2854) acc1: 100.0000 (90.6527) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1868 data: 0.0966 max mem: 3539 +Test: [1910/2387] eta: 0:01:34 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0859 (0.2843) acc1: 100.0000 (90.7009) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1818 data: 0.0907 max mem: 3539 +Test: [1920/2387] eta: 0:01:32 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0859 (0.2836) acc1: 100.0000 (90.7414) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1907 data: 0.0997 max mem: 3539 +Test: [1930/2387] eta: 0:01:30 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0938 (0.2826) acc1: 100.0000 (90.7891) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1937 data: 0.1034 max mem: 3539 +Test: [1940/2387] eta: 0:01:28 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0917 (0.2819) acc1: 100.0000 (90.8052) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2075 data: 0.1169 max mem: 3539 +Test: [1950/2387] eta: 0:01:26 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1315 (0.2813) acc1: 98.4375 (90.8328) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2041 data: 0.1128 max mem: 3539 +Test: [1960/2387] eta: 0:01:24 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1048 (0.2803) acc1: 99.7396 (90.8792) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1875 data: 0.0961 max mem: 3539 +Test: [1970/2387] eta: 0:01:22 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0869 (0.2794) acc1: 100.0000 (90.9211) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1864 data: 0.0947 max mem: 3539 +Test: [1980/2387] eta: 0:01:20 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1044 (0.2785) acc1: 99.7396 (90.9613) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1819 data: 0.0905 max mem: 3539 +Test: [1990/2387] eta: 0:01:18 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1113 (0.2777) acc1: 99.4792 (91.0039) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1773 data: 0.0846 max mem: 3539 +Test: [2000/2387] eta: 0:01:16 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1115 (0.2769) acc1: 99.4792 (91.0461) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1819 data: 0.0893 max mem: 3539 +Test: [2010/2387] eta: 0:01:14 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1124 (0.2761) acc1: 99.4792 (91.0876) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1841 data: 0.0912 max mem: 3539 +Test: [2020/2387] eta: 0:01:12 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1126 (0.2753) acc1: 99.4792 (91.1291) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1659 data: 0.0726 max mem: 3539 +Test: [2030/2387] eta: 0:01:10 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1126 (0.2745) acc1: 99.4792 (91.1706) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0827 max mem: 3540 +Test: [2040/2387] eta: 0:01:08 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1126 (0.2737) acc1: 99.4792 (91.2107) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1919 data: 0.0990 max mem: 3540 +Test: [2050/2387] eta: 0:01:06 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1112 (0.2729) acc1: 99.4792 (91.2515) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0766 max mem: 3540 +Test: [2060/2387] eta: 0:01:04 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1087 (0.2721) acc1: 99.7396 (91.2930) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0775 max mem: 3540 +Test: [2070/2387] eta: 0:01:02 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1030 (0.2713) acc1: 99.7396 (91.3339) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1702 data: 0.0777 max mem: 3540 +Test: [2080/2387] eta: 0:01:00 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1036 (0.2705) acc1: 99.7396 (91.3747) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1808 data: 0.0881 max mem: 3540 +Test: [2090/2387] eta: 0:00:58 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1036 (0.2697) acc1: 100.0000 (91.4155) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2046 data: 0.1126 max mem: 3540 +Test: [2100/2387] eta: 0:00:56 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1042 (0.2689) acc1: 99.7396 (91.4546) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2073 data: 0.1152 max mem: 3540 +Test: [2110/2387] eta: 0:00:54 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1066 (0.2681) acc1: 99.7396 (91.4933) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2011 data: 0.1089 max mem: 3540 +Test: [2120/2387] eta: 0:00:52 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1073 (0.2674) acc1: 99.4792 (91.5280) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1949 data: 0.1027 max mem: 3540 +Test: [2130/2387] eta: 0:00:50 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1015 (0.2667) acc1: 100.0000 (91.5677) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2098 data: 0.1178 max mem: 3540 +Test: [2140/2387] eta: 0:00:48 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0987 (0.2660) acc1: 100.0000 (91.6004) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2037 data: 0.1121 max mem: 3540 +Test: [2150/2387] eta: 0:00:46 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1047 (0.2653) acc1: 99.4792 (91.6341) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1938 data: 0.1030 max mem: 3540 +Test: [2160/2387] eta: 0:00:44 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0893 (0.2645) acc1: 100.0000 (91.6708) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1979 data: 0.1060 max mem: 3540 +Test: [2170/2387] eta: 0:00:42 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0839 (0.2637) acc1: 100.0000 (91.7079) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1926 data: 0.1004 max mem: 3540 +Test: [2180/2387] eta: 0:00:40 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0963 (0.2636) acc1: 99.7396 (91.7030) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.1978 data: 0.1069 max mem: 3540 +Test: [2190/2387] eta: 0:00:38 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1098 (0.2631) acc1: 99.2188 (91.7248) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1996 data: 0.1090 max mem: 3540 +Test: [2200/2387] eta: 0:00:36 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0926 (0.2623) acc1: 99.7396 (91.7614) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1879 data: 0.0965 max mem: 3540 +Test: [2210/2387] eta: 0:00:34 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0927 (0.2616) acc1: 100.0000 (91.7975) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1708 data: 0.0797 max mem: 3540 +Test: [2220/2387] eta: 0:00:32 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1021 (0.2609) acc1: 99.7396 (91.8301) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1630 data: 0.0725 max mem: 3540 +Test: [2230/2387] eta: 0:00:30 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0958 (0.2602) acc1: 99.7396 (91.8656) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1543 data: 0.0641 max mem: 3540 +Test: [2240/2387] eta: 0:00:28 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0890 (0.2594) acc1: 100.0000 (91.9014) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1447 data: 0.0549 max mem: 3540 +Test: [2250/2387] eta: 0:00:26 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0814 (0.2587) acc1: 100.0000 (91.9367) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1428 data: 0.0530 max mem: 3540 +Test: [2260/2387] eta: 0:00:24 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0880 (0.2579) acc1: 100.0000 (91.9711) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1441 data: 0.0545 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0880 (0.2572) acc1: 100.0000 (92.0063) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1442 data: 0.0545 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0875 (0.2565) acc1: 100.0000 (92.0385) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1424 data: 0.0527 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0880 (0.2558) acc1: 100.0000 (92.0713) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1404 data: 0.0507 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0860 (0.2551) acc1: 100.0000 (92.1051) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1391 data: 0.0495 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0776 (0.2543) acc1: 100.0000 (92.1390) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1387 data: 0.0491 max mem: 3541 +Test: [2320/2387] eta: 0:00:13 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0844 (0.2537) acc1: 100.0000 (92.1689) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1383 data: 0.0486 max mem: 3541 +Test: [2330/2387] eta: 0:00:11 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1138 (0.2532) acc1: 99.7396 (92.1915) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1374 data: 0.0477 max mem: 3541 +Test: [2340/2387] eta: 0:00:09 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1138 (0.2526) acc1: 99.7396 (92.2238) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1357 data: 0.0461 max mem: 3541 +Test: [2350/2387] eta: 0:00:07 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1044 (0.2521) acc1: 100.0000 (92.2500) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1330 data: 0.0434 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1186 (0.2518) acc1: 99.7396 (92.2625) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1300 data: 0.0404 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1015 (0.2511) acc1: 99.7396 (92.2943) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1283 data: 0.0387 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0871 (0.2505) acc1: 99.7396 (92.3215) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1133 data: 0.0236 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0871 (0.2501) acc1: 99.7396 (92.3373) acc5: 100.0000 (100.0000) acc1_neg: 97.674 acc5_neg: 100.000 time: 0.0993 data: 0.0124 max mem: 3541 +Test: Total time: 0:07:39 (0.1927 s / it) +* Overall Acc@1 92.337 Acc@5 100.000 + target == real Acc@1 5.550 Acc@5 100.000 + target == fake Acc@1 98.432 Acc@5 100.000 + loss 0.250 + PR-AUC 0.9561 ROC-AUC 0.6422 EER 40.30% (thr=0.9141) +Accuracy of the network on the 916267 test images: 92.3% +[rank0]:[W726 03:09:58.998285355 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:10:20.941000 3133143 torch/distributed/run.py:766] +W0726 03:10:20.941000 3133143 torch/distributed/run.py:766] ***************************************** +W0726 03:10:20.941000 3133143 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:10:20.941000 3133143 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:10:54.499806498 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:10:54.726254349 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:10:54.892977372 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:10:54.926587846 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:10:54.019890669 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:10:54.077681313 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:10:54.162812017 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:10:55.188218254 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n5_seed5_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed5_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:09:05 acc1_pos: 13.021 acc5_pos: 100.000 loss: 1.7283 (1.7283) acc1: 13.0208 (13.0208) acc5: 100.0000 (100.0000) time: 8.0193 data: 6.0974 max mem: 3531 +Test: [ 10/517] eta: 0:06:55 acc1_pos: 9.635 acc5_pos: 100.000 loss: 1.6385 (1.5941) acc1: 14.0625 (17.9214) acc5: 100.0000 (100.0000) time: 0.8188 data: 0.5547 max mem: 3531 +Test: [ 20/517] eta: 0:03:56 acc1_pos: 4.948 acc5_pos: 100.000 loss: 1.6003 (1.6290) acc1: 14.5833 (16.8775) acc5: 100.0000 (100.0000) time: 0.0993 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:03:27 acc1_pos: 6.510 acc5_pos: 100.000 loss: 1.6944 (1.6754) acc1: 12.2396 (15.5914) acc5: 100.0000 (100.0000) time: 0.2099 data: 0.1133 max mem: 3531 +Test: [ 40/517] eta: 0:03:01 acc1_pos: 2.865 acc5_pos: 100.000 loss: 2.0061 (1.7780) acc1: 6.5104 (12.5635) acc5: 100.0000 (100.0000) time: 0.2799 data: 0.1837 max mem: 3531 +Test: [ 50/517] eta: 0:02:45 acc1_pos: 3.670 acc5_pos: 100.000 loss: 2.0172 (1.7316) acc1: 3.6458 (15.4565) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2449 data: 0.1486 max mem: 3531 +Test: [ 60/517] eta: 0:02:35 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1931 (1.4731) acc1: 95.3125 (28.9105) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2549 data: 0.1584 max mem: 3531 +Test: [ 70/517] eta: 0:02:25 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1560 (1.3035) acc1: 97.1354 (37.5954) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.2499 data: 0.1525 max mem: 3531 +Test: [ 80/517] eta: 0:02:19 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.3153 (1.1842) acc1: 86.7188 (43.6921) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.2599 data: 0.1630 max mem: 3531 +Test: [ 90/517] eta: 0:02:13 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.2233 (1.0767) acc1: 92.4479 (49.3933) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2699 data: 0.1718 max mem: 3531 +Test: [100/517] eta: 0:02:08 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.2049 (0.9904) acc1: 96.0938 (54.0635) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2599 data: 0.1634 max mem: 3531 +Test: [110/517] eta: 0:02:03 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1401 (0.9125) acc1: 98.6979 (58.1715) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2599 data: 0.1647 max mem: 3531 +Test: [120/517] eta: 0:01:59 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1382 (0.8520) acc1: 98.4375 (61.3378) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2649 data: 0.1690 max mem: 3531 +Test: [130/517] eta: 0:01:55 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1626 (0.7990) acc1: 96.8750 (64.0824) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2649 data: 0.1698 max mem: 3531 +Test: [140/517] eta: 0:01:50 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1297 (0.7509) acc1: 98.6979 (66.5854) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2549 data: 0.1600 max mem: 3531 +Test: [150/517] eta: 0:01:47 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1240 (0.7109) acc1: 99.4792 (68.6603) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2649 data: 0.1693 max mem: 3531 +Test: [160/517] eta: 0:01:43 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1312 (0.6749) acc1: 99.2188 (70.5697) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2649 data: 0.1676 max mem: 3531 +Test: [170/517] eta: 0:01:39 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1247 (0.6429) acc1: 99.4792 (72.2618) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2399 data: 0.1424 max mem: 3531 +Test: [180/517] eta: 0:01:35 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1306 (0.6151) acc1: 99.4792 (73.6490) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2299 data: 0.1343 max mem: 3531 +Test: [190/517] eta: 0:01:32 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.0942 (0.5875) acc1: 99.7396 (75.0273) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2349 data: 0.1375 max mem: 3531 +Test: [200/517] eta: 0:01:28 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.0900 (0.5640) acc1: 100.0000 (76.2580) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2299 data: 0.1327 max mem: 3531 +Test: [210/517] eta: 0:01:24 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1387 (0.5472) acc1: 98.1771 (77.0549) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2297 data: 0.1352 max mem: 3532 +Test: [220/517] eta: 0:01:21 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.2078 (0.5327) acc1: 92.9688 (77.7361) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2447 data: 0.1486 max mem: 3532 +Test: [230/517] eta: 0:01:18 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1995 (0.5177) acc1: 94.0104 (78.5647) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2599 data: 0.1613 max mem: 3532 +Test: [240/517] eta: 0:01:15 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1813 (0.5048) acc1: 97.3958 (79.3471) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2349 data: 0.1387 max mem: 3532 +Test: [250/517] eta: 0:01:12 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1608 (0.4901) acc1: 99.4792 (80.1430) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2199 data: 0.1262 max mem: 3532 +Test: [260/517] eta: 0:01:09 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1162 (0.4757) acc1: 100.0000 (80.8998) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2249 data: 0.1266 max mem: 3532 +Test: [270/517] eta: 0:01:06 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1202 (0.4630) acc1: 100.0000 (81.6036) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2399 data: 0.1395 max mem: 3532 +Test: [280/517] eta: 0:01:03 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1254 (0.4507) acc1: 100.0000 (82.2583) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2599 data: 0.1637 max mem: 3532 +Test: [290/517] eta: 0:01:00 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1207 (0.4394) acc1: 100.0000 (82.8546) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2399 data: 0.1457 max mem: 3532 +Test: [300/517] eta: 0:00:57 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1229 (0.4295) acc1: 99.4792 (83.3628) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.2299 data: 0.1331 max mem: 3532 +Test: [310/517] eta: 0:00:54 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1594 (0.4233) acc1: 97.6562 (83.7026) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2399 data: 0.1429 max mem: 3532 +Test: [320/517] eta: 0:00:52 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.2175 (0.4151) acc1: 94.7917 (84.1470) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2449 data: 0.1512 max mem: 3532 +Test: [330/517] eta: 0:00:49 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1176 (0.4062) acc1: 100.0000 (84.6087) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2399 data: 0.1455 max mem: 3532 +Test: [340/517] eta: 0:00:46 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1302 (0.3988) acc1: 99.2188 (85.0035) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2251 data: 0.1313 max mem: 3532 +Test: [350/517] eta: 0:00:43 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1934 (0.3961) acc1: 95.8333 (85.0576) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.2349 data: 0.1432 max mem: 3532 +Test: [360/517] eta: 0:00:40 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.2614 (0.3914) acc1: 88.2812 (85.2717) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2497 data: 0.1557 max mem: 3532 +Test: [370/517] eta: 0:00:38 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1210 (0.3838) acc1: 99.2188 (85.6553) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2351 data: 0.1416 max mem: 3532 +Test: [380/517] eta: 0:00:35 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1096 (0.3769) acc1: 99.4792 (86.0093) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2299 data: 0.1379 max mem: 3532 +Test: [390/517] eta: 0:00:32 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1305 (0.3710) acc1: 98.9583 (86.3285) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2397 data: 0.1463 max mem: 3532 +Test: [400/517] eta: 0:00:30 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1171 (0.3644) acc1: 99.7396 (86.6636) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2399 data: 0.1468 max mem: 3532 +Test: [410/517] eta: 0:00:27 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1099 (0.3584) acc1: 100.0000 (86.9823) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2399 data: 0.1461 max mem: 3532 +Test: [420/517] eta: 0:00:25 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1043 (0.3522) acc1: 100.0000 (87.2878) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2449 data: 0.1502 max mem: 3532 +Test: [430/517] eta: 0:00:22 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.0995 (0.3465) acc1: 99.7396 (87.5719) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2299 data: 0.1354 max mem: 3532 +Test: [440/517] eta: 0:00:19 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1038 (0.3410) acc1: 99.7396 (87.8514) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2399 data: 0.1454 max mem: 3533 +Test: [450/517] eta: 0:00:17 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.0964 (0.3355) acc1: 100.0000 (88.1202) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2449 data: 0.1484 max mem: 3533 +Test: [460/517] eta: 0:00:14 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1025 (0.3309) acc1: 100.0000 (88.3462) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2159 data: 0.1213 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.0996 (0.3260) acc1: 100.0000 (88.5909) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2364 data: 0.1431 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.0945 (0.3212) acc1: 100.0000 (88.8281) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2489 data: 0.1545 max mem: 3533 +Test: [490/517] eta: 0:00:06 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1031 (0.3169) acc1: 100.0000 (89.0508) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2234 data: 0.1301 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1160 (0.3135) acc1: 99.7396 (89.2512) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2363 data: 0.1441 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1155 (0.3095) acc1: 99.4792 (89.4534) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2178 data: 0.1273 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1122 (0.3072) acc1: 99.4792 (89.5594) acc5: 100.0000 (100.0000) acc1_neg: 99.293 acc5_neg: 100.000 time: 0.1722 data: 0.0761 max mem: 3533 +Test: Total time: 0:02:10 (0.2520 s / it) +* Overall Acc@1 89.559 Acc@5 100.000 + target == real Acc@1 11.378 Acc@5 100.000 + target == fake Acc@1 97.670 Acc@5 100.000 + loss 0.307 + PR-AUC 0.9391 ROC-AUC 0.6687 EER 37.67% (thr=0.8931) +Accuracy of the network on the 198427 test images: 89.6% +[rank0]:[W726 03:13:17.557665351 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 03:13:39.965000 3133715 torch/distributed/run.py:766] +W0726 03:13:39.965000 3133715 torch/distributed/run.py:766] ***************************************** +W0726 03:13:39.965000 3133715 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:13:39.965000 3133715 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:14:15.680305839 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:14:15.979317296 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:14:15.156487537 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:14:16.209789038 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank6]:[W726 03:14:16.209789547 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:14:16.629387602 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +[rank7]:[W726 03:14:16.650276394 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:14:16.660440363 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n5_seed5_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed5_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:39:17 acc1_pos: 14.062 acc5_pos: 100.000 loss: 1.2260 (1.2260) acc1: 14.0625 (14.0625) acc5: 100.0000 (100.0000) time: 6.0608 data: 4.6333 max mem: 3531 +Test: [ 10/389] eta: 0:04:00 acc1_pos: 87.760 acc5_pos: 100.000 loss: 0.9656 (0.8309) acc1: 39.0625 (52.3674) acc5: 100.0000 (100.0000) time: 0.6340 data: 0.4215 max mem: 3531 +Test: [ 20/389] eta: 0:02:20 acc1_pos: 59.635 acc5_pos: 100.000 loss: 0.5155 (0.7390) acc1: 59.6354 (59.9826) acc5: 100.0000 (100.0000) time: 0.0956 data: 0.0004 max mem: 3531 +Test: [ 30/389] eta: 0:01:54 acc1_pos: 38.802 acc5_pos: 100.000 loss: 0.8164 (0.8463) acc1: 40.6250 (49.6724) acc5: 100.0000 (100.0000) time: 0.1470 data: 0.0511 max mem: 3531 +Test: [ 40/389] eta: 0:01:44 acc1_pos: 70.312 acc5_pos: 100.000 loss: 0.9266 (0.8537) acc1: 37.7604 (48.0501) acc5: 100.0000 (100.0000) time: 0.2149 data: 0.1221 max mem: 3531 +Test: [ 50/389] eta: 0:01:36 acc1_pos: 0.781 acc5_pos: 100.000 loss: 0.9266 (0.9308) acc1: 40.3646 (45.9457) acc5: 100.0000 (100.0000) time: 0.2278 data: 0.1318 max mem: 3531 +Test: [ 60/389] eta: 0:01:26 acc1_pos: 31.510 acc5_pos: 100.000 loss: 0.9530 (1.0094) acc1: 31.5104 (42.2686) acc5: 100.0000 (100.0000) time: 0.1921 data: 0.0936 max mem: 3531 +Test: [ 70/389] eta: 0:01:20 acc1_pos: 23.438 acc5_pos: 100.000 loss: 0.9508 (1.0037) acc1: 28.9062 (41.0908) acc5: 100.0000 (100.0000) time: 0.1714 data: 0.0769 max mem: 3531 +Test: [ 80/389] eta: 0:01:15 acc1_pos: 94.792 acc5_pos: 100.000 loss: 0.9147 (0.9823) acc1: 36.4583 (41.9850) acc5: 100.0000 (100.0000) time: 0.1877 data: 0.0960 max mem: 3531 +Test: [ 90/389] eta: 0:01:11 acc1_pos: 64.323 acc5_pos: 100.000 loss: 0.6972 (0.9214) acc1: 63.8021 (46.2368) acc5: 100.0000 (100.0000) time: 0.1964 data: 0.1030 max mem: 3531 +Test: [100/389] eta: 0:01:06 acc1_pos: 4.167 acc5_pos: 100.000 loss: 0.8506 (0.9552) acc1: 35.4167 (43.7655) acc5: 100.0000 (100.0000) time: 0.1759 data: 0.0837 max mem: 3531 +Test: [110/389] eta: 0:01:03 acc1_pos: 17.448 acc5_pos: 100.000 loss: 1.3068 (0.9822) acc1: 17.4479 (41.9904) acc5: 100.0000 (100.0000) time: 0.1720 data: 0.0799 max mem: 3531 +Test: [120/389] eta: 0:00:59 acc1_pos: 63.281 acc5_pos: 100.000 loss: 1.0154 (0.9771) acc1: 19.2708 (42.2865) acc5: 100.0000 (100.0000) time: 0.1642 data: 0.0725 max mem: 3531 +Test: [130/389] eta: 0:00:55 acc1_pos: 53.988 acc5_pos: 100.000 loss: 0.6251 (0.9459) acc1: 63.2812 (44.5014) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.1399 data: 0.0493 max mem: 3531 +Test: [140/389] eta: 0:00:52 acc1_pos: 53.988 acc5_pos: 100.000 loss: 0.3786 (0.9054) acc1: 79.9479 (47.5639) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1397 data: 0.0483 max mem: 3531 +Test: [150/389] eta: 0:00:49 acc1_pos: 53.988 acc5_pos: 100.000 loss: 0.3786 (0.8789) acc1: 88.0208 (49.5654) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1456 data: 0.0538 max mem: 3531 +Test: [160/389] eta: 0:00:46 acc1_pos: 53.988 acc5_pos: 100.000 loss: 0.3378 (0.8443) acc1: 89.0625 (52.1222) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1509 data: 0.0580 max mem: 3531 +Test: [170/389] eta: 0:00:43 acc1_pos: 53.988 acc5_pos: 100.000 loss: 0.2132 (0.8074) acc1: 98.6979 (54.7012) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1442 data: 0.0519 max mem: 3531 +Test: [180/389] eta: 0:00:40 acc1_pos: 53.988 acc5_pos: 100.000 loss: 0.1760 (0.7750) acc1: 99.2188 (56.9737) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1361 data: 0.0450 max mem: 3531 +Test: [190/389] eta: 0:00:38 acc1_pos: 53.988 acc5_pos: 100.000 loss: 0.2407 (0.7563) acc1: 93.7500 (58.3592) acc5: 100.0000 (100.0000) acc1_neg: 59.115 acc5_neg: 100.000 time: 0.1461 data: 0.0535 max mem: 3531 +Test: [200/389] eta: 0:00:36 acc1_pos: 53.988 acc5_pos: 100.000 loss: 0.4447 (0.7471) acc1: 78.9062 (58.7855) acc5: 100.0000 (100.0000) acc1_neg: 53.906 acc5_neg: 100.000 time: 0.1627 data: 0.0690 max mem: 3531 +Test: [210/389] eta: 0:00:34 acc1_pos: 53.988 acc5_pos: 100.000 loss: 0.3358 (0.7294) acc1: 81.7708 (59.9082) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1655 data: 0.0730 max mem: 3532 +Test: [220/389] eta: 0:00:31 acc1_pos: 53.988 acc5_pos: 100.000 loss: 0.1726 (0.7090) acc1: 99.2188 (61.3063) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.1499 data: 0.0583 max mem: 3532 +Test: [230/389] eta: 0:00:29 acc1_pos: 53.988 acc5_pos: 100.000 loss: 0.2714 (0.7078) acc1: 81.7708 (61.3727) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.1482 data: 0.0570 max mem: 3532 +Test: [240/389] eta: 0:00:27 acc1_pos: 53.988 acc5_pos: 100.000 loss: 0.4322 (0.6929) acc1: 81.7708 (62.4384) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1599 data: 0.0680 max mem: 3532 +Test: [250/389] eta: 0:00:25 acc1_pos: 53.988 acc5_pos: 100.000 loss: 0.2027 (0.6734) acc1: 97.6562 (63.8955) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1529 data: 0.0608 max mem: 3532 +Test: [260/389] eta: 0:00:23 acc1_pos: 53.988 acc5_pos: 100.000 loss: 0.2397 (0.6622) acc1: 98.9583 (64.8517) acc5: 100.0000 (100.0000) acc1_neg: 34.375 acc5_neg: 100.000 time: 0.1428 data: 0.0505 max mem: 3532 +Test: [270/389] eta: 0:00:21 acc1_pos: 53.988 acc5_pos: 100.000 loss: 0.3792 (0.6664) acc1: 88.0208 (64.4421) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1527 data: 0.0608 max mem: 3532 +Test: [280/389] eta: 0:00:19 acc1_pos: 53.988 acc5_pos: 100.000 loss: 0.4271 (0.6598) acc1: 83.5938 (64.8956) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1640 data: 0.0728 max mem: 3532 +Test: [290/389] eta: 0:00:17 acc1_pos: 53.988 acc5_pos: 100.000 loss: 0.3007 (0.6470) acc1: 93.7500 (65.7932) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1692 data: 0.0745 max mem: 3532 +Test: [300/389] eta: 0:00:16 acc1_pos: 53.988 acc5_pos: 100.000 loss: 0.3795 (0.6385) acc1: 86.1979 (66.5274) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.1708 data: 0.0768 max mem: 3532 +Test: [310/389] eta: 0:00:14 acc1_pos: 53.988 acc5_pos: 100.000 loss: 0.4471 (0.6329) acc1: 86.1979 (66.9748) acc5: 100.0000 (100.0000) acc1_neg: 38.802 acc5_neg: 100.000 time: 0.1699 data: 0.0775 max mem: 3532 +Test: [320/389] eta: 0:00:12 acc1_pos: 53.988 acc5_pos: 100.000 loss: 0.4073 (0.6263) acc1: 84.3750 (67.4390) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1726 data: 0.0796 max mem: 3532 +Test: [330/389] eta: 0:00:10 acc1_pos: 53.988 acc5_pos: 100.000 loss: 0.5126 (0.6367) acc1: 47.6562 (66.3866) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.1750 data: 0.0832 max mem: 3532 +Test: [340/389] eta: 0:00:08 acc1_pos: 53.988 acc5_pos: 100.000 loss: 0.3307 (0.6250) acc1: 85.9375 (67.2998) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1863 data: 0.0928 max mem: 3532 +Test: [350/389] eta: 0:00:06 acc1_pos: 53.988 acc5_pos: 100.000 loss: 0.2662 (0.6249) acc1: 96.6146 (67.4316) acc5: 100.0000 (100.0000) acc1_neg: 74.740 acc5_neg: 100.000 time: 0.1754 data: 0.0813 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 53.988 acc5_pos: 100.000 loss: 0.4523 (0.6206) acc1: 80.9896 (67.8230) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.1685 data: 0.0727 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 53.988 acc5_pos: 100.000 loss: 0.3880 (0.6111) acc1: 87.2396 (68.5717) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1788 data: 0.0810 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 53.988 acc5_pos: 100.000 loss: 0.3416 (0.6061) acc1: 94.7917 (69.0056) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1561 data: 0.0620 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 53.988 acc5_pos: 100.000 loss: 0.3416 (0.5993) acc1: 94.7917 (69.5269) acc5: 100.0000 (100.0000) acc1_neg: 91.257 acc5_neg: 100.000 time: 0.1279 data: 0.0336 max mem: 3532 +Test: Total time: 0:01:09 (0.1776 s / it) +* Overall Acc@1 69.527 Acc@5 100.000 + target == real Acc@1 42.272 Acc@5 100.000 + target == fake Acc@1 82.196 Acc@5 100.000 + loss 0.599 + PR-AUC 0.7837 ROC-AUC 0.6810 EER 36.41% (thr=0.6929) +Accuracy of the network on the 149358 test images: 69.5% +[rank0]:[W726 03:15:41.339734567 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:16:02.563000 3134275 torch/distributed/run.py:766] +W0726 03:16:02.563000 3134275 torch/distributed/run.py:766] ***************************************** +W0726 03:16:02.563000 3134275 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:16:02.563000 3134275 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 03:16:36.088459686 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:16:37.382636280 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:16:37.535839993 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:16:37.586524935 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:16:37.857248060 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:16:37.941912323 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank6]:[W726 03:16:37.941948009 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:16:37.998298979 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n5_seed5_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed5_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:32:43 acc1_pos: 1.562 acc5_pos: 100.000 loss: 2.2060 (2.2060) acc1: 1.5625 (1.5625) acc5: 100.0000 (100.0000) time: 5.8971 data: 4.1964 max mem: 3531 +Test: [ 10/333] eta: 0:03:19 acc1_pos: 5.729 acc5_pos: 100.000 loss: 2.0619 (2.0423) acc1: 7.8125 (8.4754) acc5: 100.0000 (100.0000) time: 0.6178 data: 0.3819 max mem: 3531 +Test: [ 20/333] eta: 0:01:55 acc1_pos: 17.448 acc5_pos: 100.000 loss: 1.9966 (2.0039) acc1: 10.4167 (10.9871) acc5: 100.0000 (100.0000) time: 0.0920 data: 0.0004 max mem: 3531 +Test: [ 30/333] eta: 0:01:47 acc1_pos: 11.719 acc5_pos: 100.000 loss: 1.8535 (1.9438) acc1: 13.8021 (12.1388) acc5: 100.0000 (100.0000) time: 0.2120 data: 0.1184 max mem: 3531 +Test: [ 40/333] eta: 0:01:44 acc1_pos: 5.990 acc5_pos: 100.000 loss: 1.7808 (1.9049) acc1: 13.8021 (12.4746) acc5: 100.0000 (100.0000) time: 0.3444 data: 0.2501 max mem: 3531 +Test: [ 50/333] eta: 0:01:33 acc1_pos: 18.750 acc5_pos: 100.000 loss: 1.7719 (1.8877) acc1: 14.8438 (13.0055) acc5: 100.0000 (100.0000) time: 0.2947 data: 0.2009 max mem: 3531 +Test: [ 60/333] eta: 0:01:23 acc1_pos: 16.146 acc5_pos: 100.000 loss: 1.7871 (1.8729) acc1: 14.8438 (13.1190) acc5: 100.0000 (100.0000) time: 0.2017 data: 0.1106 max mem: 3531 +Test: [ 70/333] eta: 0:01:16 acc1_pos: 15.885 acc5_pos: 100.000 loss: 1.7871 (1.8595) acc1: 14.5833 (13.3546) acc5: 100.0000 (100.0000) time: 0.1801 data: 0.0872 max mem: 3531 +Test: [ 80/333] eta: 0:01:08 acc1_pos: 18.229 acc5_pos: 100.000 loss: 1.8406 (1.8514) acc1: 14.5833 (13.5320) acc5: 100.0000 (100.0000) time: 0.1639 data: 0.0670 max mem: 3531 +Test: [ 90/333] eta: 0:01:02 acc1_pos: 21.094 acc5_pos: 100.000 loss: 1.6169 (1.8085) acc1: 18.2292 (14.7064) acc5: 100.0000 (100.0000) time: 0.1514 data: 0.0542 max mem: 3531 +Test: [100/333] eta: 0:01:00 acc1_pos: 33.333 acc5_pos: 100.000 loss: 1.3685 (1.7585) acc1: 25.7812 (16.0814) acc5: 100.0000 (100.0000) time: 0.2064 data: 0.1105 max mem: 3531 +Test: [110/333] eta: 0:00:55 acc1_pos: 32.292 acc5_pos: 100.000 loss: 1.3546 (1.7226) acc1: 26.5625 (17.2180) acc5: 100.0000 (100.0000) time: 0.2134 data: 0.1188 max mem: 3531 +Test: [120/333] eta: 0:00:50 acc1_pos: 38.021 acc5_pos: 100.000 loss: 1.3728 (1.6883) acc1: 29.6875 (18.3518) acc5: 100.0000 (100.0000) time: 0.1381 data: 0.0413 max mem: 3531 +Test: [130/333] eta: 0:00:46 acc1_pos: 26.562 acc5_pos: 100.000 loss: 1.2750 (1.6571) acc1: 30.9896 (19.3504) acc5: 100.0000 (100.0000) time: 0.1203 data: 0.0207 max mem: 3531 +Test: [140/333] eta: 0:00:44 acc1_pos: 42.188 acc5_pos: 100.000 loss: 1.2256 (1.6301) acc1: 35.6771 (20.3531) acc5: 100.0000 (100.0000) time: 0.1723 data: 0.0761 max mem: 3531 +Test: [150/333] eta: 0:00:41 acc1_pos: 20.052 acc5_pos: 100.000 loss: 1.2528 (1.6129) acc1: 30.2083 (20.8902) acc5: 100.0000 (100.0000) time: 0.2060 data: 0.1141 max mem: 3531 +Test: [160/333] eta: 0:00:38 acc1_pos: 20.395 acc5_pos: 100.000 loss: 1.3038 (1.5725) acc1: 30.4688 (22.7695) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.2006 data: 0.1071 max mem: 3531 +Test: [170/333] eta: 0:00:36 acc1_pos: 20.395 acc5_pos: 100.000 loss: 0.3888 (1.4995) acc1: 81.2500 (26.5092) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1750 data: 0.0813 max mem: 3531 +Test: [180/333] eta: 0:00:33 acc1_pos: 20.395 acc5_pos: 100.000 loss: 0.3037 (1.4327) acc1: 87.7604 (29.9681) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1668 data: 0.0743 max mem: 3531 +Test: [190/333] eta: 0:00:31 acc1_pos: 20.395 acc5_pos: 100.000 loss: 0.2678 (1.3695) acc1: 90.6250 (33.2733) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1888 data: 0.0954 max mem: 3531 +Test: [200/333] eta: 0:00:28 acc1_pos: 20.395 acc5_pos: 100.000 loss: 0.2441 (1.3146) acc1: 90.8854 (36.1383) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1964 data: 0.1018 max mem: 3531 +Test: [210/333] eta: 0:00:26 acc1_pos: 20.395 acc5_pos: 100.000 loss: 0.2852 (1.2664) acc1: 89.8438 (38.6034) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.1888 data: 0.0919 max mem: 3532 +Test: [220/333] eta: 0:00:24 acc1_pos: 20.395 acc5_pos: 100.000 loss: 0.3026 (1.2255) acc1: 88.8021 (40.7240) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.2030 data: 0.1073 max mem: 3532 +Test: [230/333] eta: 0:00:21 acc1_pos: 20.395 acc5_pos: 100.000 loss: 0.2599 (1.1819) acc1: 92.1875 (43.0274) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1949 data: 0.1000 max mem: 3532 +Test: [240/333] eta: 0:00:19 acc1_pos: 20.395 acc5_pos: 100.000 loss: 0.2380 (1.1432) acc1: 92.7083 (45.0661) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1823 data: 0.0846 max mem: 3532 +Test: [250/333] eta: 0:00:17 acc1_pos: 20.395 acc5_pos: 100.000 loss: 0.2628 (1.1088) acc1: 90.1042 (46.8096) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.2005 data: 0.1058 max mem: 3532 +Test: [260/333] eta: 0:00:15 acc1_pos: 20.395 acc5_pos: 100.000 loss: 0.2844 (1.0784) acc1: 88.0208 (48.3367) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.2083 data: 0.1166 max mem: 3532 +Test: [270/333] eta: 0:00:13 acc1_pos: 20.395 acc5_pos: 100.000 loss: 0.2705 (1.0486) acc1: 88.0208 (49.8684) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.2111 data: 0.1191 max mem: 3532 +Test: [280/333] eta: 0:00:11 acc1_pos: 20.395 acc5_pos: 100.000 loss: 0.2746 (1.0253) acc1: 87.5000 (51.0092) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1940 data: 0.1030 max mem: 3532 +Test: [290/333] eta: 0:00:09 acc1_pos: 20.395 acc5_pos: 100.000 loss: 0.4365 (1.0054) acc1: 78.1250 (51.9249) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1797 data: 0.0884 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 20.395 acc5_pos: 100.000 loss: 0.3715 (0.9836) acc1: 83.3333 (53.0428) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.1870 data: 0.0944 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 20.395 acc5_pos: 100.000 loss: 0.3218 (0.9609) acc1: 87.2396 (54.2177) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1733 data: 0.0808 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 20.395 acc5_pos: 100.000 loss: 0.2636 (0.9400) acc1: 90.6250 (55.3154) acc5: 100.0000 (100.0000) acc1_neg: 78.646 acc5_neg: 100.000 time: 0.1466 data: 0.0557 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 20.395 acc5_pos: 100.000 loss: 0.3776 (0.9248) acc1: 83.8542 (56.0447) acc5: 100.0000 (100.0000) acc1_neg: 77.083 acc5_neg: 100.000 time: 0.1223 data: 0.0323 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 20.395 acc5_pos: 100.000 loss: 0.4095 (0.9219) acc1: 80.2083 (56.1744) acc5: 100.0000 (100.0000) acc1_neg: 78.028 acc5_neg: 100.000 time: 0.1208 data: 0.0274 max mem: 3532 +Test: Total time: 0:01:07 (0.2026 s / it) +* Overall Acc@1 56.174 Acc@5 100.000 + target == real Acc@1 21.197 Acc@5 100.000 + target == fake Acc@1 87.311 Acc@5 100.000 + loss 0.922 + PR-AUC 0.5608 ROC-AUC 0.5695 EER 44.35% (thr=0.8516) +Accuracy of the network on the 127843 test images: 56.2% +[rank0]:[W726 03:17:55.492934240 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:18:16.935000 3134833 torch/distributed/run.py:766] +W0726 03:18:16.935000 3134833 torch/distributed/run.py:766] ***************************************** +W0726 03:18:16.935000 3134833 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:18:16.935000 3134833 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:18:53.070630731 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:18:54.597861053 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:18:54.619978472 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:18:54.836209954 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:18:54.850559967 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:18:54.895571555 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:18:54.976837406 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:18:54.987216571 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n5_seed5_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed5_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:23:12 acc1_pos: 91.927 acc5_pos: 100.000 loss: 0.2832 (0.2832) acc1: 91.9271 (91.9271) acc5: 100.0000 (100.0000) time: 5.9238 data: 4.3723 max mem: 3531 +Test: [ 10/235] eta: 0:02:19 acc1_pos: 93.229 acc5_pos: 100.000 loss: 0.2832 (0.2830) acc1: 91.4062 (91.3826) acc5: 100.0000 (100.0000) time: 0.6221 data: 0.3978 max mem: 3531 +Test: [ 20/235] eta: 0:01:20 acc1_pos: 91.406 acc5_pos: 100.000 loss: 0.2750 (0.2781) acc1: 91.6667 (91.7907) acc5: 100.0000 (100.0000) time: 0.0946 data: 0.0004 max mem: 3531 +Test: [ 30/235] eta: 0:01:07 acc1_pos: 91.927 acc5_pos: 100.000 loss: 0.2761 (0.2789) acc1: 91.6667 (91.7171) acc5: 100.0000 (100.0000) time: 0.1645 data: 0.0692 max mem: 3531 +Test: [ 40/235] eta: 0:01:00 acc1_pos: 91.667 acc5_pos: 100.000 loss: 0.2798 (0.2784) acc1: 91.6667 (91.6730) acc5: 100.0000 (100.0000) time: 0.2421 data: 0.1487 max mem: 3531 +Test: [ 50/235] eta: 0:00:55 acc1_pos: 89.062 acc5_pos: 100.000 loss: 0.2831 (0.2803) acc1: 91.4062 (91.5799) acc5: 100.0000 (100.0000) time: 0.2650 data: 0.1711 max mem: 3531 +Test: [ 60/235] eta: 0:00:48 acc1_pos: 91.667 acc5_pos: 100.000 loss: 0.2872 (0.2815) acc1: 91.1458 (91.5429) acc5: 100.0000 (100.0000) time: 0.2131 data: 0.1190 max mem: 3531 +Test: [ 70/235] eta: 0:00:43 acc1_pos: 92.308 acc5_pos: 100.000 loss: 0.3076 (0.4128) acc1: 89.5833 (83.2086) acc5: 100.0000 (100.0000) acc1_neg: 20.052 acc5_neg: 100.000 time: 0.1652 data: 0.0702 max mem: 3531 +Test: [ 80/235] eta: 0:00:39 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.3353 (0.5287) acc1: 24.7396 (75.8906) acc5: 100.0000 (100.0000) acc1_neg: 21.875 acc5_neg: 100.000 time: 0.1939 data: 0.0991 max mem: 3531 +Test: [ 90/235] eta: 0:00:36 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.3457 (0.6186) acc1: 24.4792 (70.2181) acc5: 100.0000 (100.0000) acc1_neg: 21.094 acc5_neg: 100.000 time: 0.2131 data: 0.1193 max mem: 3531 +Test: [100/235] eta: 0:00:33 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.3299 (0.6904) acc1: 25.5208 (65.7204) acc5: 100.0000 (100.0000) acc1_neg: 20.312 acc5_neg: 100.000 time: 0.2199 data: 0.1228 max mem: 3531 +Test: [110/235] eta: 0:00:30 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.3389 (0.7485) acc1: 24.7396 (62.0613) acc5: 100.0000 (100.0000) acc1_neg: 25.521 acc5_neg: 100.000 time: 0.2004 data: 0.1030 max mem: 3531 +Test: [120/235] eta: 0:00:27 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.3254 (0.7956) acc1: 24.2188 (58.9230) acc5: 100.0000 (100.0000) acc1_neg: 23.958 acc5_neg: 100.000 time: 0.2099 data: 0.1164 max mem: 3531 +Test: [130/235] eta: 0:00:25 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.3109 (0.8347) acc1: 25.0000 (56.2917) acc5: 100.0000 (100.0000) acc1_neg: 25.521 acc5_neg: 100.000 time: 0.2147 data: 0.1222 max mem: 3531 +Test: [140/235] eta: 0:00:21 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.3109 (0.8696) acc1: 24.2188 (53.9709) acc5: 100.0000 (100.0000) acc1_neg: 21.094 acc5_neg: 100.000 time: 0.1467 data: 0.0512 max mem: 3531 +Test: [150/235] eta: 0:00:19 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.3053 (0.8985) acc1: 24.2188 (52.0609) acc5: 100.0000 (100.0000) acc1_neg: 25.260 acc5_neg: 100.000 time: 0.1819 data: 0.0872 max mem: 3531 +Test: [160/235] eta: 0:00:17 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.3018 (0.9240) acc1: 25.0000 (50.3348) acc5: 100.0000 (100.0000) acc1_neg: 22.656 acc5_neg: 100.000 time: 0.2089 data: 0.1183 max mem: 3531 +Test: [170/235] eta: 0:00:14 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.2948 (0.9451) acc1: 26.8229 (49.0908) acc5: 100.0000 (100.0000) acc1_neg: 27.083 acc5_neg: 100.000 time: 0.1677 data: 0.0768 max mem: 3531 +Test: [180/235] eta: 0:00:12 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.2895 (0.9632) acc1: 29.1667 (48.0936) acc5: 100.0000 (100.0000) acc1_neg: 28.125 acc5_neg: 100.000 time: 0.1919 data: 0.0996 max mem: 3531 +Test: [190/235] eta: 0:00:09 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.2834 (0.9800) acc1: 30.2083 (47.1722) acc5: 100.0000 (100.0000) acc1_neg: 28.646 acc5_neg: 100.000 time: 0.2012 data: 0.1082 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.2858 (0.9964) acc1: 29.1667 (46.2842) acc5: 100.0000 (100.0000) acc1_neg: 28.385 acc5_neg: 100.000 time: 0.1826 data: 0.0899 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.2977 (1.0103) acc1: 29.1667 (45.4915) acc5: 100.0000 (100.0000) acc1_neg: 26.302 acc5_neg: 100.000 time: 0.1512 data: 0.0594 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.2986 (1.0234) acc1: 29.4271 (44.7964) acc5: 100.0000 (100.0000) acc1_neg: 30.208 acc5_neg: 100.000 time: 0.1326 data: 0.0419 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.2986 (1.0347) acc1: 30.9896 (44.1908) acc5: 100.0000 (100.0000) acc1_neg: 30.990 acc5_neg: 100.000 time: 0.1268 data: 0.0369 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.2829 (1.0406) acc1: 30.4688 (44.0047) acc5: 100.0000 (100.0000) acc1_neg: 21.739 acc5_neg: 100.000 time: 0.1185 data: 0.0320 max mem: 3532 +Test: Total time: 0:00:48 (0.2058 s / it) +* Overall Acc@1 44.005 Acc@5 100.000 + target == real Acc@1 91.530 Acc@5 100.000 + target == fake Acc@1 26.816 Acc@5 100.000 + loss 1.041 + PR-AUC 0.8504 ROC-AUC 0.6843 EER 37.40% (thr=0.1885) +Accuracy of the network on the 89879 test images: 44.0% +[rank0]:[W726 03:19:50.453391061 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:20:11.691000 3135380 torch/distributed/run.py:766] +W0726 03:20:11.691000 3135380 torch/distributed/run.py:766] ***************************************** +W0726 03:20:11.691000 3135380 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:20:11.691000 3135380 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:20:47.913832800 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:20:47.015262502 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:20:47.015266502 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:20:48.596058008 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:20:48.686392828 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:20:48.823660552 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:20:48.966649274 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:20:48.097264050 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n5_seed5_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed5_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:30:31 acc1_pos: 5.208 acc5_pos: 100.000 loss: 1.8771 (1.8771) acc1: 5.2083 (5.2083) acc5: 100.0000 (100.0000) time: 6.8091 data: 5.1361 max mem: 3531 +Test: [ 10/269] eta: 0:03:01 acc1_pos: 17.448 acc5_pos: 100.000 loss: 1.7085 (1.6474) acc1: 12.5000 (12.3106) acc5: 100.0000 (100.0000) time: 0.7009 data: 0.4673 max mem: 3531 +Test: [ 20/269] eta: 0:01:43 acc1_pos: 16.406 acc5_pos: 100.000 loss: 1.5169 (1.5689) acc1: 13.0208 (15.4638) acc5: 100.0000 (100.0000) time: 0.0949 data: 0.0004 max mem: 3531 +Test: [ 30/269] eta: 0:01:22 acc1_pos: 14.323 acc5_pos: 100.000 loss: 1.4896 (1.5221) acc1: 15.1042 (17.2463) acc5: 100.0000 (100.0000) time: 0.1519 data: 0.0552 max mem: 3531 +Test: [ 40/269] eta: 0:01:16 acc1_pos: 9.483 acc5_pos: 100.000 loss: 0.8519 (1.2311) acc1: 46.8750 (34.2480) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.2528 data: 0.1604 max mem: 3531 +Test: [ 50/269] eta: 0:01:05 acc1_pos: 9.483 acc5_pos: 100.000 loss: 0.1753 (1.0285) acc1: 95.8333 (46.3848) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.2306 data: 0.1389 max mem: 3531 +Test: [ 60/269] eta: 0:00:58 acc1_pos: 9.483 acc5_pos: 100.000 loss: 0.2151 (0.9040) acc1: 94.2708 (53.6672) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1676 data: 0.0749 max mem: 3531 +Test: [ 70/269] eta: 0:00:53 acc1_pos: 9.483 acc5_pos: 100.000 loss: 0.2816 (0.8167) acc1: 88.5417 (58.9532) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1828 data: 0.0884 max mem: 3531 +Test: [ 80/269] eta: 0:00:48 acc1_pos: 9.483 acc5_pos: 100.000 loss: 0.2815 (0.7501) acc1: 89.0625 (62.9180) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1753 data: 0.0813 max mem: 3531 +Test: [ 90/269] eta: 0:00:43 acc1_pos: 9.483 acc5_pos: 100.000 loss: 0.2299 (0.6981) acc1: 94.5312 (66.1859) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1492 data: 0.0543 max mem: 3531 +Test: [100/269] eta: 0:00:39 acc1_pos: 9.483 acc5_pos: 100.000 loss: 0.2174 (0.6506) acc1: 95.3125 (69.0749) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1588 data: 0.0602 max mem: 3531 +Test: [110/269] eta: 0:00:37 acc1_pos: 9.483 acc5_pos: 100.000 loss: 0.2186 (0.6129) acc1: 94.7917 (71.2181) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1949 data: 0.0968 max mem: 3531 +Test: [120/269] eta: 0:00:34 acc1_pos: 9.483 acc5_pos: 100.000 loss: 0.2186 (0.5822) acc1: 93.4896 (73.0157) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.2149 data: 0.1205 max mem: 3531 +Test: [130/269] eta: 0:00:31 acc1_pos: 9.483 acc5_pos: 100.000 loss: 0.2808 (0.5616) acc1: 89.8438 (74.1253) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.1749 data: 0.0818 max mem: 3531 +Test: [140/269] eta: 0:00:28 acc1_pos: 9.483 acc5_pos: 100.000 loss: 0.3288 (0.5448) acc1: 87.2396 (75.1034) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.1560 data: 0.0629 max mem: 3531 +Test: [150/269] eta: 0:00:26 acc1_pos: 9.483 acc5_pos: 100.000 loss: 0.3090 (0.5295) acc1: 89.0625 (75.9899) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.2199 data: 0.1251 max mem: 3531 +Test: [160/269] eta: 0:00:24 acc1_pos: 9.483 acc5_pos: 100.000 loss: 0.2994 (0.5153) acc1: 89.5833 (76.7776) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.2243 data: 0.1280 max mem: 3531 +Test: [170/269] eta: 0:00:21 acc1_pos: 9.483 acc5_pos: 100.000 loss: 0.2660 (0.4999) acc1: 90.3646 (77.6849) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1949 data: 0.1003 max mem: 3531 +Test: [180/269] eta: 0:00:19 acc1_pos: 9.483 acc5_pos: 100.000 loss: 0.1686 (0.4799) acc1: 97.3958 (78.8358) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2094 data: 0.1159 max mem: 3531 +Test: [190/269] eta: 0:00:17 acc1_pos: 9.483 acc5_pos: 100.000 loss: 0.1541 (0.4673) acc1: 97.9167 (79.6411) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2058 data: 0.1121 max mem: 3531 +Test: [200/269] eta: 0:00:14 acc1_pos: 9.483 acc5_pos: 100.000 loss: 0.2721 (0.4585) acc1: 91.6667 (80.1863) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1699 data: 0.0750 max mem: 3531 +Test: [210/269] eta: 0:00:12 acc1_pos: 9.483 acc5_pos: 100.000 loss: 0.2745 (0.4492) acc1: 91.6667 (80.7921) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1689 data: 0.0725 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 9.483 acc5_pos: 100.000 loss: 0.2407 (0.4403) acc1: 94.0104 (81.3172) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1799 data: 0.0836 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 9.483 acc5_pos: 100.000 loss: 0.1991 (0.4296) acc1: 97.1354 (81.9771) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2066 data: 0.1138 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 9.483 acc5_pos: 100.000 loss: 0.2262 (0.4232) acc1: 94.7917 (82.3824) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.2050 data: 0.1124 max mem: 3532 +Test: [250/269] eta: 0:00:03 acc1_pos: 9.483 acc5_pos: 100.000 loss: 0.2899 (0.4203) acc1: 90.8854 (82.5542) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.1567 data: 0.0629 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 9.483 acc5_pos: 100.000 loss: 0.3025 (0.4170) acc1: 90.6250 (82.7417) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1377 data: 0.0463 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 9.483 acc5_pos: 100.000 loss: 0.2584 (0.4117) acc1: 92.4479 (83.0976) acc5: 100.0000 (100.0000) acc1_neg: 99.213 acc5_neg: 100.000 time: 0.1239 data: 0.0304 max mem: 3532 +Test: Total time: 0:00:54 (0.2038 s / it) +* Overall Acc@1 83.098 Acc@5 100.000 + target == real Acc@1 17.236 Acc@5 100.000 + target == fake Acc@1 92.086 Acc@5 100.000 + loss 0.412 + PR-AUC 0.9127 ROC-AUC 0.6263 EER 40.39% (thr=0.8311) +Accuracy of the network on the 103293 test images: 83.1% +[rank0]:[W726 03:21:54.284312873 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T03:22:01Z] eval_done tag=fixed500k_real_n5_seed5_resampled +[2026-07-26T03:22:01Z] eval_start tag=fixed500k_fake_n11_seed5_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed5_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T03:22:01Z +hostname=n116-040-168 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed5_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n11_seed5_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 03:22:16.719000 3135959 torch/distributed/run.py:766] +W0726 03:22:16.719000 3135959 torch/distributed/run.py:766] ***************************************** +W0726 03:22:16.719000 3135959 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:22:16.719000 3135959 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:22:48.878150617 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:22:49.411435974 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:22:49.675971250 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:22:49.834239939 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:22:49.893529243 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:22:49.893625463 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:22:49.121812111 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:22:50.256660039 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n11_seed5_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed5_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:22:21 acc1_pos: 91.667 acc5_pos: 100.000 loss: 0.2340 (0.2340) acc1: 91.6667 (91.6667) acc5: 100.0000 (100.0000) time: 5.7348 data: 3.3886 max mem: 3531 +Test: [ 10/234] eta: 0:02:15 acc1_pos: 86.458 acc5_pos: 100.000 loss: 0.4462 (0.4613) acc1: 78.6458 (77.9593) acc5: 100.0000 (100.0000) time: 0.6038 data: 0.3085 max mem: 3531 +Test: [ 20/234] eta: 0:01:34 acc1_pos: 93.750 acc5_pos: 100.000 loss: 0.4062 (0.4168) acc1: 80.4688 (81.3492) acc5: 100.0000 (100.0000) time: 0.1753 data: 0.0814 max mem: 3531 +Test: [ 30/234] eta: 0:01:22 acc1_pos: 59.635 acc5_pos: 100.000 loss: 0.2691 (0.3829) acc1: 90.3646 (83.1905) acc5: 100.0000 (100.0000) time: 0.2938 data: 0.1976 max mem: 3531 +Test: [ 40/234] eta: 0:01:12 acc1_pos: 99.740 acc5_pos: 100.000 loss: 0.2783 (0.4070) acc1: 90.1042 (80.7990) acc5: 100.0000 (100.0000) time: 0.3018 data: 0.2081 max mem: 3531 +Test: [ 50/234] eta: 0:01:04 acc1_pos: 98.958 acc5_pos: 100.000 loss: 0.1074 (0.3479) acc1: 99.7396 (84.5129) acc5: 100.0000 (100.0000) time: 0.2736 data: 0.1819 max mem: 3531 +Test: [ 60/234] eta: 0:00:56 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.1059 (0.3090) acc1: 99.7396 (87.0176) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2353 data: 0.1442 max mem: 3531 +Test: [ 70/234] eta: 0:00:52 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.1137 (0.3759) acc1: 99.7396 (83.2820) acc5: 100.0000 (100.0000) acc1_neg: 35.938 acc5_neg: 100.000 time: 0.2348 data: 0.1441 max mem: 3531 +Test: [ 80/234] eta: 0:00:47 acc1_pos: 98.621 acc5_pos: 100.000 loss: 1.0959 (0.4279) acc1: 42.1875 (80.3434) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2384 data: 0.1485 max mem: 3531 +Test: [ 90/234] eta: 0:00:42 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.1889 (0.3985) acc1: 94.0104 (82.0484) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.2083 data: 0.1179 max mem: 3531 +Test: [100/234] eta: 0:00:39 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.2153 (0.4073) acc1: 92.7083 (81.5156) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2442 data: 0.1489 max mem: 3531 +Test: [110/234] eta: 0:00:35 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.4105 (0.4066) acc1: 81.2500 (81.6324) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.2244 data: 0.1281 max mem: 3531 +Test: [120/234] eta: 0:00:31 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.3586 (0.4213) acc1: 84.3750 (81.0025) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.2049 data: 0.1121 max mem: 3531 +Test: [130/234] eta: 0:00:28 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.5251 (0.4366) acc1: 73.1771 (79.9817) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.2344 data: 0.1416 max mem: 3531 +Test: [140/234] eta: 0:00:25 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.7056 (0.4747) acc1: 63.8021 (78.0271) acc5: 100.0000 (100.0000) acc1_neg: 25.781 acc5_neg: 100.000 time: 0.2078 data: 0.1135 max mem: 3531 +Test: [150/234] eta: 0:00:22 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.6698 (0.4832) acc1: 64.8438 (77.5697) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.1949 data: 0.0986 max mem: 3531 +Test: [160/234] eta: 0:00:19 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.5565 (0.4865) acc1: 70.0521 (77.3567) acc5: 100.0000 (100.0000) acc1_neg: 64.844 acc5_neg: 100.000 time: 0.1931 data: 0.0989 max mem: 3531 +Test: [170/234] eta: 0:00:16 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.6604 (0.4972) acc1: 65.3646 (76.6676) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.1834 data: 0.0924 max mem: 3531 +Test: [180/234] eta: 0:00:13 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.6604 (0.5110) acc1: 65.3646 (75.8532) acc5: 100.0000 (100.0000) acc1_neg: 28.906 acc5_neg: 100.000 time: 0.1788 data: 0.0858 max mem: 3531 +Test: [190/234] eta: 0:00:10 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.8219 (0.5297) acc1: 58.3333 (74.8391) acc5: 100.0000 (100.0000) acc1_neg: 58.333 acc5_neg: 100.000 time: 0.1694 data: 0.0744 max mem: 3531 +Test: [200/234] eta: 0:00:08 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.8178 (0.5225) acc1: 58.3333 (75.2902) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1461 data: 0.0535 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.1203 (0.5051) acc1: 99.4792 (76.3786) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1201 data: 0.0300 max mem: 3532 +Test: [220/234] eta: 0:00:03 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.2186 (0.4937) acc1: 95.8333 (77.1517) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.1098 data: 0.0202 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.5026 (0.5188) acc1: 42.7083 (75.4633) acc5: 100.0000 (100.0000) acc1_neg: 38.281 acc5_neg: 100.000 time: 0.1148 data: 0.0254 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 98.621 acc5_pos: 100.000 loss: 1.0460 (0.5257) acc1: 39.3229 (75.0415) acc5: 100.0000 (100.0000) acc1_neg: 38.132 acc5_neg: 100.000 time: 0.1105 data: 0.0202 max mem: 3532 +Test: Total time: 0:00:52 (0.2228 s / it) +* Overall Acc@1 75.042 Acc@5 100.000 + target == real Acc@1 84.617 Acc@5 100.000 + target == fake Acc@1 72.343 Acc@5 100.000 + loss 0.526 + PR-AUC 0.9603 ROC-AUC 0.8710 EER 21.79% (thr=0.3772) +Accuracy of the network on the 89729 test images: 75.0% +[rank0]:[W726 03:23:49.044350232 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:24:07.437000 3136517 torch/distributed/run.py:766] +W0726 03:24:07.437000 3136517 torch/distributed/run.py:766] ***************************************** +W0726 03:24:07.437000 3136517 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:24:07.437000 3136517 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:24:33.340063888 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:24:33.694655548 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:24:33.880674248 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:24:33.048320456 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:24:33.108063699 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +| distributed init (rank 7): env:// +[rank0]:[W726 03:24:33.170550308 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:24:33.174770390 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:24:34.177351913 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n11_seed5_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed5_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:42:18 acc1_pos: 85.677 acc5_pos: 100.000 loss: 0.3872 (0.3872) acc1: 85.6771 (85.6771) acc5: 100.0000 (100.0000) time: 7.0963 data: 5.5719 max mem: 3531 +Test: [ 10/2387] eta: 0:28:47 acc1_pos: 46.615 acc5_pos: 100.000 loss: 0.7577 (0.7833) acc1: 53.1250 (51.7519) acc5: 100.0000 (100.0000) time: 0.7269 data: 0.5069 max mem: 3531 +Test: [ 20/2387] eta: 0:16:51 acc1_pos: 42.448 acc5_pos: 100.000 loss: 0.8092 (0.8411) acc1: 47.9167 (49.4172) acc5: 100.0000 (100.0000) time: 0.0939 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:14:01 acc1_pos: 29.948 acc5_pos: 100.000 loss: 0.8195 (0.8403) acc1: 44.7917 (48.6643) acc5: 100.0000 (100.0000) time: 0.1539 data: 0.0566 max mem: 3531 +Test: [ 40/2387] eta: 0:12:39 acc1_pos: 47.396 acc5_pos: 100.000 loss: 0.8195 (0.8234) acc1: 47.3958 (49.6443) acc5: 100.0000 (100.0000) time: 0.2149 data: 0.1200 max mem: 3531 +Test: [ 50/2387] eta: 0:11:58 acc1_pos: 57.292 acc5_pos: 100.000 loss: 0.7087 (0.8048) acc1: 57.2917 (51.1336) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1341 max mem: 3531 +Test: [ 60/2387] eta: 0:11:33 acc1_pos: 61.719 acc5_pos: 100.000 loss: 0.6595 (0.7799) acc1: 61.7188 (53.1506) acc5: 100.0000 (100.0000) time: 0.2451 data: 0.1505 max mem: 3531 +Test: [ 70/2387] eta: 0:11:02 acc1_pos: 56.510 acc5_pos: 100.000 loss: 0.6823 (0.7772) acc1: 57.2917 (53.3377) acc5: 100.0000 (100.0000) time: 0.2321 data: 0.1400 max mem: 3531 +Test: [ 80/2387] eta: 0:10:44 acc1_pos: 94.531 acc5_pos: 100.000 loss: 0.6191 (0.7361) acc1: 63.2812 (56.4686) acc5: 100.0000 (100.0000) time: 0.2221 data: 0.1257 max mem: 3531 +Test: [ 90/2387] eta: 0:10:30 acc1_pos: 97.917 acc5_pos: 100.000 loss: 0.3675 (0.6928) acc1: 84.3750 (59.6898) acc5: 100.0000 (100.0000) time: 0.2327 data: 0.1369 max mem: 3531 +Test: [ 100/2387] eta: 0:10:22 acc1_pos: 85.938 acc5_pos: 100.000 loss: 0.3335 (0.6665) acc1: 84.6354 (61.6981) acc5: 100.0000 (100.0000) time: 0.2427 data: 0.1504 max mem: 3531 +Test: [ 110/2387] eta: 0:10:16 acc1_pos: 67.708 acc5_pos: 100.000 loss: 0.3603 (0.6392) acc1: 83.0729 (63.7082) acc5: 100.0000 (100.0000) time: 0.2549 data: 0.1578 max mem: 3531 +Test: [ 120/2387] eta: 0:10:10 acc1_pos: 78.125 acc5_pos: 100.000 loss: 0.3767 (0.6214) acc1: 80.9896 (65.1494) acc5: 100.0000 (100.0000) time: 0.2550 data: 0.1569 max mem: 3531 +Test: [ 130/2387] eta: 0:10:05 acc1_pos: 82.812 acc5_pos: 100.000 loss: 0.3859 (0.6017) acc1: 82.5521 (66.8297) acc5: 100.0000 (100.0000) time: 0.2549 data: 0.1602 max mem: 3531 +Test: [ 140/2387] eta: 0:09:57 acc1_pos: 86.719 acc5_pos: 100.000 loss: 0.3806 (0.5885) acc1: 82.8125 (67.9355) acc5: 100.0000 (100.0000) time: 0.2446 data: 0.1476 max mem: 3531 +Test: [ 150/2387] eta: 0:10:05 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.3305 (0.5698) acc1: 88.2812 (69.4830) acc5: 100.0000 (100.0000) time: 0.2854 data: 0.1895 max mem: 3531 +Test: [ 160/2387] eta: 0:09:40 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3135 (0.5584) acc1: 88.8021 (70.3885) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.2249 data: 0.1302 max mem: 3531 +Test: [ 170/2387] eta: 0:09:32 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2681 (0.5407) acc1: 92.1875 (71.7425) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1649 data: 0.0714 max mem: 3531 +Test: [ 180/2387] eta: 0:09:26 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2594 (0.5249) acc1: 92.9688 (72.9426) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.2249 data: 0.1333 max mem: 3531 +Test: [ 190/2387] eta: 0:09:20 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2501 (0.5104) acc1: 94.2708 (74.0592) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2294 data: 0.1364 max mem: 3531 +Test: [ 200/2387] eta: 0:09:06 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2468 (0.4973) acc1: 94.5312 (75.0518) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1906 data: 0.0964 max mem: 3531 +Test: [ 210/2387] eta: 0:08:54 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2502 (0.4859) acc1: 93.2292 (75.8997) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1499 data: 0.0558 max mem: 3532 +Test: [ 220/2387] eta: 0:08:49 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2537 (0.4755) acc1: 93.2292 (76.6827) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1842 data: 0.0898 max mem: 3532 +Test: [ 230/2387] eta: 0:08:45 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2552 (0.4658) acc1: 93.4896 (77.4204) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2249 data: 0.1282 max mem: 3532 +Test: [ 240/2387] eta: 0:08:36 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2765 (0.4662) acc1: 92.4479 (77.5037) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.1999 data: 0.1028 max mem: 3532 +Test: [ 250/2387] eta: 0:08:22 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2397 (0.4535) acc1: 80.4688 (78.2920) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1359 data: 0.0404 max mem: 3532 +Test: [ 260/2387] eta: 0:08:19 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1023 (0.4399) acc1: 100.0000 (79.1238) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1649 data: 0.0697 max mem: 3532 +Test: [ 270/2387] eta: 0:08:12 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1001 (0.4273) acc1: 100.0000 (79.8941) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2038 data: 0.1069 max mem: 3532 +Test: [ 280/2387] eta: 0:08:06 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1001 (0.4158) acc1: 100.0000 (80.6096) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0831 max mem: 3532 +Test: [ 290/2387] eta: 0:07:58 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.0985 (0.4048) acc1: 100.0000 (81.2724) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1698 data: 0.0749 max mem: 3532 +Test: [ 300/2387] eta: 0:07:53 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.0936 (0.3947) acc1: 100.0000 (81.8920) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1698 data: 0.0750 max mem: 3532 +Test: [ 310/2387] eta: 0:07:47 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1033 (0.3860) acc1: 100.0000 (82.4600) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1814 data: 0.0843 max mem: 3532 +Test: [ 320/2387] eta: 0:07:43 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1247 (0.3778) acc1: 99.7396 (82.9983) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1861 data: 0.0913 max mem: 3532 +Test: [ 330/2387] eta: 0:07:39 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1236 (0.3701) acc1: 99.7396 (83.5064) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1940 data: 0.1012 max mem: 3532 +Test: [ 340/2387] eta: 0:07:35 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1236 (0.3629) acc1: 99.7396 (83.9825) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1934 data: 0.1002 max mem: 3532 +Test: [ 350/2387] eta: 0:07:30 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1207 (0.3559) acc1: 100.0000 (84.4381) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1801 data: 0.0872 max mem: 3532 +Test: [ 360/2387] eta: 0:07:27 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1219 (0.3514) acc1: 99.7396 (84.7920) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1949 data: 0.0981 max mem: 3532 +Test: [ 370/2387] eta: 0:07:23 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2013 (0.3475) acc1: 96.3542 (85.0994) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2039 data: 0.1068 max mem: 3532 +Test: [ 380/2387] eta: 0:07:23 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2053 (0.3437) acc1: 96.3542 (85.4050) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2209 data: 0.1275 max mem: 3532 +Test: [ 390/2387] eta: 0:07:22 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2049 (0.3402) acc1: 96.3542 (85.6837) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2535 data: 0.1622 max mem: 3532 +Test: [ 400/2387] eta: 0:07:20 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2040 (0.3368) acc1: 96.0938 (85.9518) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2362 data: 0.1451 max mem: 3532 +Test: [ 410/2387] eta: 0:07:18 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2001 (0.3335) acc1: 96.3542 (86.2169) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2187 data: 0.1264 max mem: 3532 +Test: [ 420/2387] eta: 0:07:13 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2001 (0.3303) acc1: 96.6146 (86.4620) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1914 data: 0.0984 max mem: 3532 +Test: [ 430/2387] eta: 0:07:06 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2056 (0.3300) acc1: 95.8333 (86.5472) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1424 data: 0.0462 max mem: 3532 +Test: [ 440/2387] eta: 0:07:02 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2149 (0.3280) acc1: 90.3646 (86.7235) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1461 data: 0.0486 max mem: 3533 +Test: [ 450/2387] eta: 0:06:57 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1550 (0.3242) acc1: 99.7396 (87.0098) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1674 data: 0.0747 max mem: 3533 +Test: [ 460/2387] eta: 0:06:53 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1529 (0.3205) acc1: 99.4792 (87.2808) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1698 data: 0.0760 max mem: 3533 +Test: [ 470/2387] eta: 0:06:51 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1554 (0.3170) acc1: 99.4792 (87.5382) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1926 data: 0.0991 max mem: 3533 +Test: [ 480/2387] eta: 0:06:46 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1579 (0.3137) acc1: 99.4792 (87.7853) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1801 data: 0.0889 max mem: 3533 +Test: [ 490/2387] eta: 0:06:44 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1574 (0.3105) acc1: 99.4792 (88.0256) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1765 data: 0.0854 max mem: 3533 +Test: [ 500/2387] eta: 0:06:42 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1574 (0.3074) acc1: 99.4792 (88.2532) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2057 data: 0.1148 max mem: 3533 +Test: [ 510/2387] eta: 0:06:39 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1580 (0.3049) acc1: 99.2188 (88.4586) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2089 data: 0.1164 max mem: 3533 +Test: [ 520/2387] eta: 0:06:37 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1630 (0.3027) acc1: 98.4375 (88.6501) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2092 data: 0.1177 max mem: 3533 +Test: [ 530/2387] eta: 0:06:34 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1432 (0.2997) acc1: 99.4792 (88.8609) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2017 data: 0.1111 max mem: 3533 +Test: [ 540/2387] eta: 0:06:30 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1406 (0.2968) acc1: 99.7396 (89.0620) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1746 data: 0.0817 max mem: 3533 +Test: [ 550/2387] eta: 0:06:26 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1414 (0.2939) acc1: 99.7396 (89.2563) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1589 data: 0.0643 max mem: 3533 +Test: [ 560/2387] eta: 0:06:23 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1400 (0.2912) acc1: 99.7396 (89.4450) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1608 data: 0.0665 max mem: 3533 +Test: [ 570/2387] eta: 0:06:19 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1399 (0.2886) acc1: 99.7396 (89.6253) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1595 data: 0.0657 max mem: 3533 +Test: [ 580/2387] eta: 0:06:16 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1412 (0.2860) acc1: 99.7396 (89.8007) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1690 data: 0.0724 max mem: 3533 +Test: [ 590/2387] eta: 0:06:13 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1432 (0.2836) acc1: 99.7396 (89.9702) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1840 data: 0.0879 max mem: 3533 +Test: [ 600/2387] eta: 0:06:10 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1486 (0.2826) acc1: 98.4375 (90.0994) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1826 data: 0.0897 max mem: 3533 +Test: [ 610/2387] eta: 0:06:08 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1991 (0.2810) acc1: 97.6562 (90.2231) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1825 data: 0.0900 max mem: 3533 +Test: [ 620/2387] eta: 0:06:05 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1859 (0.2795) acc1: 97.6562 (90.3403) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1835 data: 0.0916 max mem: 3533 +Test: [ 630/2387] eta: 0:06:03 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1859 (0.2780) acc1: 97.3958 (90.4529) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1992 data: 0.1061 max mem: 3533 +Test: [ 640/2387] eta: 0:06:00 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1876 (0.2766) acc1: 97.1354 (90.5563) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2061 data: 0.1124 max mem: 3533 +Test: [ 650/2387] eta: 0:05:58 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1880 (0.2753) acc1: 97.1354 (90.6618) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1884 data: 0.0966 max mem: 3533 +Test: [ 660/2387] eta: 0:05:56 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1868 (0.2739) acc1: 97.6562 (90.7664) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1932 data: 0.1015 max mem: 3533 +Test: [ 670/2387] eta: 0:05:53 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1847 (0.2726) acc1: 97.6562 (90.8714) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1942 data: 0.1011 max mem: 3534 +Test: [ 680/2387] eta: 0:05:50 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1909 (0.2727) acc1: 97.3958 (90.9363) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1799 data: 0.0877 max mem: 3534 +Test: [ 690/2387] eta: 0:05:48 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3238 (0.2758) acc1: 93.2292 (90.7384) acc5: 100.0000 (100.0000) acc1_neg: 68.229 acc5_neg: 100.000 time: 0.1868 data: 0.0953 max mem: 3534 +Test: [ 700/2387] eta: 0:05:45 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.5315 (0.2795) acc1: 72.3958 (90.4775) acc5: 100.0000 (100.0000) acc1_neg: 73.698 acc5_neg: 100.000 time: 0.1904 data: 0.0995 max mem: 3534 +Test: [ 710/2387] eta: 0:05:43 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.5360 (0.2832) acc1: 71.8750 (90.2009) acc5: 100.0000 (100.0000) acc1_neg: 73.698 acc5_neg: 100.000 time: 0.1783 data: 0.0881 max mem: 3534 +Test: [ 720/2387] eta: 0:05:40 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.5375 (0.2867) acc1: 71.6146 (89.9463) acc5: 100.0000 (100.0000) acc1_neg: 69.531 acc5_neg: 100.000 time: 0.1796 data: 0.0890 max mem: 3534 +Test: [ 730/2387] eta: 0:05:38 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.5404 (0.2904) acc1: 70.5729 (89.6774) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.1810 data: 0.0873 max mem: 3534 +Test: [ 740/2387] eta: 0:05:35 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.5410 (0.2938) acc1: 70.8333 (89.4407) acc5: 100.0000 (100.0000) acc1_neg: 71.354 acc5_neg: 100.000 time: 0.1864 data: 0.0926 max mem: 3534 +Test: [ 750/2387] eta: 0:05:33 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.5405 (0.2971) acc1: 71.3542 (89.1998) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.1854 data: 0.0941 max mem: 3534 +Test: [ 760/2387] eta: 0:05:30 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.5528 (0.3008) acc1: 70.5729 (88.9444) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.1794 data: 0.0868 max mem: 3534 +Test: [ 770/2387] eta: 0:05:27 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.5725 (0.3031) acc1: 71.0938 (88.8105) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1781 data: 0.0847 max mem: 3534 +Test: [ 780/2387] eta: 0:05:25 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.4478 (0.3046) acc1: 74.4792 (88.7157) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1753 data: 0.0833 max mem: 3534 +Test: [ 790/2387] eta: 0:05:22 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3113 (0.3050) acc1: 88.0208 (88.6921) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.1763 data: 0.0847 max mem: 3534 +Test: [ 800/2387] eta: 0:05:20 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2784 (0.3048) acc1: 91.4062 (88.7260) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.1822 data: 0.0881 max mem: 3534 +Test: [ 810/2387] eta: 0:05:18 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2849 (0.3049) acc1: 91.4062 (88.7337) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1885 data: 0.0947 max mem: 3534 +Test: [ 820/2387] eta: 0:05:15 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3112 (0.3057) acc1: 89.0625 (88.6888) acc5: 100.0000 (100.0000) acc1_neg: 60.938 acc5_neg: 100.000 time: 0.1875 data: 0.0968 max mem: 3534 +Test: [ 830/2387] eta: 0:05:13 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.4684 (0.3104) acc1: 75.0000 (88.3775) acc5: 100.0000 (100.0000) acc1_neg: 66.146 acc5_neg: 100.000 time: 0.1859 data: 0.0955 max mem: 3534 +Test: [ 840/2387] eta: 0:05:11 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2888 (0.3092) acc1: 78.6458 (88.4748) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1888 data: 0.0969 max mem: 3534 +Test: [ 850/2387] eta: 0:05:09 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1967 (0.3081) acc1: 97.9167 (88.5616) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1938 data: 0.0988 max mem: 3534 +Test: [ 860/2387] eta: 0:05:07 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1938 (0.3068) acc1: 97.6562 (88.6484) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1901 data: 0.0956 max mem: 3534 +Test: [ 870/2387] eta: 0:05:05 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1812 (0.3053) acc1: 98.1771 (88.7522) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1926 data: 0.0954 max mem: 3534 +Test: [ 880/2387] eta: 0:05:02 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1900 (0.3043) acc1: 96.8750 (88.8204) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1941 data: 0.0970 max mem: 3534 +Test: [ 890/2387] eta: 0:05:00 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2228 (0.3041) acc1: 94.2708 (88.8667) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1885 data: 0.0962 max mem: 3534 +Test: [ 900/2387] eta: 0:04:58 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3491 (0.3076) acc1: 88.2812 (88.6446) acc5: 100.0000 (100.0000) acc1_neg: 67.708 acc5_neg: 100.000 time: 0.1852 data: 0.0930 max mem: 3535 +Test: [ 910/2387] eta: 0:04:56 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.6892 (0.3119) acc1: 63.2812 (88.3716) acc5: 100.0000 (100.0000) acc1_neg: 60.938 acc5_neg: 100.000 time: 0.1847 data: 0.0934 max mem: 3535 +Test: [ 920/2387] eta: 0:04:53 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.7045 (0.3162) acc1: 63.2812 (88.1011) acc5: 100.0000 (100.0000) acc1_neg: 60.938 acc5_neg: 100.000 time: 0.1893 data: 0.0988 max mem: 3535 +Test: [ 930/2387] eta: 0:04:51 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.6947 (0.3203) acc1: 63.0208 (87.8298) acc5: 100.0000 (100.0000) acc1_neg: 56.510 acc5_neg: 100.000 time: 0.1864 data: 0.0949 max mem: 3535 +Test: [ 940/2387] eta: 0:04:49 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.7012 (0.3243) acc1: 62.5000 (87.5589) acc5: 100.0000 (100.0000) acc1_neg: 59.635 acc5_neg: 100.000 time: 0.1891 data: 0.0950 max mem: 3535 +Test: [ 950/2387] eta: 0:04:47 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.6999 (0.3282) acc1: 62.5000 (87.3017) acc5: 100.0000 (100.0000) acc1_neg: 63.281 acc5_neg: 100.000 time: 0.1897 data: 0.0939 max mem: 3535 +Test: [ 960/2387] eta: 0:04:45 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.6852 (0.3319) acc1: 63.2812 (87.0640) acc5: 100.0000 (100.0000) acc1_neg: 64.844 acc5_neg: 100.000 time: 0.1869 data: 0.0933 max mem: 3535 +Test: [ 970/2387] eta: 0:04:43 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.6922 (0.3357) acc1: 63.5417 (86.8137) acc5: 100.0000 (100.0000) acc1_neg: 63.542 acc5_neg: 100.000 time: 0.1986 data: 0.1064 max mem: 3535 +Test: [ 980/2387] eta: 0:04:41 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.6721 (0.3372) acc1: 64.3229 (86.7259) acc5: 100.0000 (100.0000) acc1_neg: 64.844 acc5_neg: 100.000 time: 0.1991 data: 0.1075 max mem: 3535 +Test: [ 990/2387] eta: 0:04:39 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.4340 (0.3387) acc1: 82.0312 (86.6391) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.1944 data: 0.1036 max mem: 3535 +Test: [1000/2387] eta: 0:04:36 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.4563 (0.3399) acc1: 79.9479 (86.5762) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.1907 data: 0.0991 max mem: 3535 +Test: [1010/2387] eta: 0:04:34 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.4929 (0.3424) acc1: 78.1250 (86.4210) acc5: 100.0000 (100.0000) acc1_neg: 62.240 acc5_neg: 100.000 time: 0.1817 data: 0.0905 max mem: 3535 +Test: [1020/2387] eta: 0:04:32 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.5210 (0.3444) acc1: 65.3646 (86.2918) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1788 data: 0.0879 max mem: 3535 +Test: [1030/2387] eta: 0:04:30 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3489 (0.3446) acc1: 84.3750 (86.2911) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1780 data: 0.0865 max mem: 3535 +Test: [1040/2387] eta: 0:04:27 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2879 (0.3438) acc1: 89.3229 (86.3450) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1752 data: 0.0824 max mem: 3535 +Test: [1050/2387] eta: 0:04:25 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2517 (0.3429) acc1: 93.4896 (86.4165) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1712 data: 0.0774 max mem: 3535 +Test: [1060/2387] eta: 0:04:23 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2561 (0.3425) acc1: 92.9688 (86.4610) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1757 data: 0.0838 max mem: 3535 +Test: [1070/2387] eta: 0:04:21 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2797 (0.3427) acc1: 90.3646 (86.4581) acc5: 100.0000 (100.0000) acc1_neg: 63.542 acc5_neg: 100.000 time: 0.1838 data: 0.0934 max mem: 3535 +Test: [1080/2387] eta: 0:04:18 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.5117 (0.3454) acc1: 75.7812 (86.2832) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.1858 data: 0.0950 max mem: 3535 +Test: [1090/2387] eta: 0:04:16 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.5147 (0.3461) acc1: 74.2188 (86.2383) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.1707 data: 0.0800 max mem: 3535 +Test: [1100/2387] eta: 0:04:14 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3502 (0.3460) acc1: 85.9375 (86.2403) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1722 data: 0.0808 max mem: 3535 +Test: [1110/2387] eta: 0:04:12 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3170 (0.3461) acc1: 88.0208 (86.2366) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1834 data: 0.0912 max mem: 3535 +Test: [1120/2387] eta: 0:04:09 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3298 (0.3462) acc1: 86.7188 (86.2311) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.1774 data: 0.0859 max mem: 3536 +Test: [1130/2387] eta: 0:04:07 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3301 (0.3462) acc1: 85.6771 (86.2308) acc5: 100.0000 (100.0000) acc1_neg: 78.646 acc5_neg: 100.000 time: 0.1855 data: 0.0946 max mem: 3536 +Test: [1140/2387] eta: 0:04:05 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.4161 (0.3490) acc1: 78.6458 (86.0420) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.1949 data: 0.1032 max mem: 3536 +Test: [1150/2387] eta: 0:04:03 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.6004 (0.3491) acc1: 67.4479 (86.0248) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1951 data: 0.1031 max mem: 3536 +Test: [1160/2387] eta: 0:04:02 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1170 (0.3471) acc1: 100.0000 (86.1427) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2021 data: 0.1095 max mem: 3536 +Test: [1170/2387] eta: 0:03:59 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1165 (0.3453) acc1: 100.0000 (86.2569) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1871 data: 0.0948 max mem: 3536 +Test: [1180/2387] eta: 0:03:57 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1156 (0.3433) acc1: 100.0000 (86.3710) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1863 data: 0.0957 max mem: 3536 +Test: [1190/2387] eta: 0:03:55 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1156 (0.3415) acc1: 100.0000 (86.4822) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1825 data: 0.0898 max mem: 3536 +Test: [1200/2387] eta: 0:03:52 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1216 (0.3398) acc1: 100.0000 (86.5882) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1447 data: 0.0503 max mem: 3536 +Test: [1210/2387] eta: 0:03:50 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1288 (0.3381) acc1: 100.0000 (86.6942) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1393 data: 0.0470 max mem: 3536 +Test: [1220/2387] eta: 0:03:48 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1035 (0.3362) acc1: 100.0000 (86.8024) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1506 data: 0.0589 max mem: 3536 +Test: [1230/2387] eta: 0:03:45 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.0956 (0.3342) acc1: 100.0000 (86.9096) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1510 data: 0.0584 max mem: 3536 +Test: [1240/2387] eta: 0:03:43 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.0940 (0.3323) acc1: 100.0000 (87.0148) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1504 data: 0.0582 max mem: 3536 +Test: [1250/2387] eta: 0:03:41 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.0940 (0.3304) acc1: 100.0000 (87.1182) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1564 data: 0.0648 max mem: 3536 +Test: [1260/2387] eta: 0:03:38 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.0926 (0.3285) acc1: 100.0000 (87.2189) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1601 data: 0.0675 max mem: 3536 +Test: [1270/2387] eta: 0:03:36 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.0989 (0.3268) acc1: 100.0000 (87.3176) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1614 data: 0.0677 max mem: 3536 +Test: [1280/2387] eta: 0:03:34 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1424 (0.3261) acc1: 97.6562 (87.3758) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1666 data: 0.0738 max mem: 3536 +Test: [1290/2387] eta: 0:03:32 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2385 (0.3257) acc1: 94.2708 (87.4104) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1600 data: 0.0672 max mem: 3536 +Test: [1300/2387] eta: 0:03:30 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2385 (0.3249) acc1: 93.2292 (87.4608) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1752 data: 0.0825 max mem: 3536 +Test: [1310/2387] eta: 0:03:28 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2379 (0.3245) acc1: 92.1875 (87.4970) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1898 data: 0.0951 max mem: 3536 +Test: [1320/2387] eta: 0:03:25 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2558 (0.3241) acc1: 91.1458 (87.5274) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1616 data: 0.0631 max mem: 3536 +Test: [1330/2387] eta: 0:03:23 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3225 (0.3247) acc1: 87.2396 (87.4992) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.1599 data: 0.0625 max mem: 3536 +Test: [1340/2387] eta: 0:03:21 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.4060 (0.3254) acc1: 83.0729 (87.4557) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1740 data: 0.0778 max mem: 3536 +Test: [1350/2387] eta: 0:03:19 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.0951 (0.3236) acc1: 100.0000 (87.5486) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0756 max mem: 3537 +Test: [1360/2387] eta: 0:03:17 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.0826 (0.3219) acc1: 100.0000 (87.6401) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1646 data: 0.0705 max mem: 3537 +Test: [1370/2387] eta: 0:03:15 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.0824 (0.3201) acc1: 100.0000 (87.7302) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1772 data: 0.0836 max mem: 3537 +Test: [1380/2387] eta: 0:03:13 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.0889 (0.3184) acc1: 100.0000 (87.8191) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1848 data: 0.0942 max mem: 3537 +Test: [1390/2387] eta: 0:03:11 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.0828 (0.3167) acc1: 100.0000 (87.9066) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1777 data: 0.0849 max mem: 3537 +Test: [1400/2387] eta: 0:03:09 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.0837 (0.3151) acc1: 100.0000 (87.9930) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1718 data: 0.0787 max mem: 3537 +Test: [1410/2387] eta: 0:03:07 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.0936 (0.3149) acc1: 100.0000 (88.0129) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.1670 data: 0.0747 max mem: 3537 +Test: [1420/2387] eta: 0:03:05 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3153 (0.3153) acc1: 88.2812 (87.9849) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1971 data: 0.1019 max mem: 3537 +Test: [1430/2387] eta: 0:03:03 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3192 (0.3154) acc1: 86.7188 (87.9855) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.2178 data: 0.1205 max mem: 3537 +Test: [1440/2387] eta: 0:03:01 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3213 (0.3156) acc1: 88.2812 (87.9813) acc5: 100.0000 (100.0000) acc1_neg: 69.010 acc5_neg: 100.000 time: 0.1749 data: 0.0799 max mem: 3537 +Test: [1450/2387] eta: 0:03:00 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3541 (0.3170) acc1: 80.9896 (87.8900) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1963 data: 0.1043 max mem: 3537 +Test: [1460/2387] eta: 0:02:58 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1812 (0.3161) acc1: 93.4896 (87.9519) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2556 data: 0.1648 max mem: 3537 +Test: [1470/2387] eta: 0:02:56 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1582 (0.3149) acc1: 97.9167 (88.0233) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2135 data: 0.1183 max mem: 3537 +Test: [1480/2387] eta: 0:02:54 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1437 (0.3138) acc1: 99.2188 (88.0933) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1707 data: 0.0741 max mem: 3537 +Test: [1490/2387] eta: 0:02:52 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1437 (0.3127) acc1: 99.2188 (88.1637) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1813 data: 0.0878 max mem: 3537 +Test: [1500/2387] eta: 0:02:50 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1554 (0.3117) acc1: 98.6979 (88.2245) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1940 data: 0.1015 max mem: 3537 +Test: [1510/2387] eta: 0:02:48 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1717 (0.3109) acc1: 98.9583 (88.2940) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2084 data: 0.1166 max mem: 3537 +Test: [1520/2387] eta: 0:02:46 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1786 (0.3100) acc1: 99.2188 (88.3602) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2070 data: 0.1145 max mem: 3537 +Test: [1530/2387] eta: 0:02:45 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1744 (0.3092) acc1: 98.4375 (88.4165) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2149 data: 0.1229 max mem: 3537 +Test: [1540/2387] eta: 0:02:43 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1579 (0.3083) acc1: 98.9583 (88.4754) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2148 data: 0.1218 max mem: 3537 +Test: [1550/2387] eta: 0:02:41 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1501 (0.3073) acc1: 99.4792 (88.5442) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1868 data: 0.0943 max mem: 3537 +Test: [1560/2387] eta: 0:02:39 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1500 (0.3064) acc1: 99.4792 (88.5984) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1753 data: 0.0842 max mem: 3537 +Test: [1570/2387] eta: 0:02:37 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1781 (0.3058) acc1: 97.9167 (88.6534) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1729 data: 0.0822 max mem: 3537 +Test: [1580/2387] eta: 0:02:35 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2178 (0.3054) acc1: 96.0938 (88.6853) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.1880 data: 0.0937 max mem: 3538 +Test: [1590/2387] eta: 0:02:33 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2414 (0.3052) acc1: 93.4896 (88.7072) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1803 data: 0.0855 max mem: 3538 +Test: [1600/2387] eta: 0:02:31 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1894 (0.3049) acc1: 96.0938 (88.7221) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1886 data: 0.0969 max mem: 3538 +Test: [1610/2387] eta: 0:02:29 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1869 (0.3043) acc1: 96.6146 (88.7605) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2066 data: 0.1136 max mem: 3538 +Test: [1620/2387] eta: 0:02:27 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2156 (0.3038) acc1: 94.5312 (88.7918) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2067 data: 0.1137 max mem: 3538 +Test: [1630/2387] eta: 0:02:25 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2361 (0.3042) acc1: 92.9688 (88.7706) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.1947 data: 0.1035 max mem: 3538 +Test: [1640/2387] eta: 0:02:23 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.4548 (0.3057) acc1: 78.1250 (88.6812) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.1855 data: 0.0951 max mem: 3538 +Test: [1650/2387] eta: 0:02:22 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.5377 (0.3071) acc1: 73.6979 (88.5926) acc5: 100.0000 (100.0000) acc1_neg: 76.302 acc5_neg: 100.000 time: 0.2160 data: 0.1226 max mem: 3538 +Test: [1660/2387] eta: 0:02:20 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.5164 (0.3083) acc1: 76.3021 (88.5263) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.2048 data: 0.1098 max mem: 3538 +Test: [1670/2387] eta: 0:02:18 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.4691 (0.3090) acc1: 77.8646 (88.4826) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.2276 data: 0.1350 max mem: 3538 +Test: [1680/2387] eta: 0:02:16 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.4192 (0.3098) acc1: 81.7708 (88.4269) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.2194 data: 0.1264 max mem: 3538 +Test: [1690/2387] eta: 0:02:14 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.4845 (0.3110) acc1: 75.7812 (88.3486) acc5: 100.0000 (100.0000) acc1_neg: 77.344 acc5_neg: 100.000 time: 0.1662 data: 0.0736 max mem: 3538 +Test: [1700/2387] eta: 0:02:12 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.5010 (0.3113) acc1: 75.7812 (88.3226) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1930 data: 0.1020 max mem: 3538 +Test: [1710/2387] eta: 0:02:10 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1140 (0.3101) acc1: 99.4792 (88.3887) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2141 data: 0.1209 max mem: 3538 +Test: [1720/2387] eta: 0:02:08 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1053 (0.3089) acc1: 99.7396 (88.4553) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1973 data: 0.1044 max mem: 3538 +Test: [1730/2387] eta: 0:02:06 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1004 (0.3078) acc1: 100.0000 (88.5203) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1862 data: 0.0956 max mem: 3538 +Test: [1740/2387] eta: 0:02:05 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.0995 (0.3066) acc1: 100.0000 (88.5855) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2044 data: 0.1137 max mem: 3538 +Test: [1750/2387] eta: 0:02:03 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.0990 (0.3054) acc1: 100.0000 (88.6480) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2133 data: 0.1217 max mem: 3538 +Test: [1760/2387] eta: 0:02:01 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1148 (0.3045) acc1: 99.4792 (88.7054) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1992 data: 0.1075 max mem: 3538 +Test: [1770/2387] eta: 0:01:59 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1299 (0.3035) acc1: 99.2188 (88.7627) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1897 data: 0.0985 max mem: 3538 +Test: [1780/2387] eta: 0:01:57 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1172 (0.3024) acc1: 99.4792 (88.8237) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1929 data: 0.1014 max mem: 3538 +Test: [1790/2387] eta: 0:01:55 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1026 (0.3013) acc1: 100.0000 (88.8848) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1863 data: 0.0949 max mem: 3538 +Test: [1800/2387] eta: 0:01:53 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1023 (0.3002) acc1: 100.0000 (88.9454) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1782 data: 0.0866 max mem: 3538 +Test: [1810/2387] eta: 0:01:51 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.0995 (0.2991) acc1: 100.0000 (89.0043) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1769 data: 0.0826 max mem: 3539 +Test: [1820/2387] eta: 0:01:49 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1040 (0.2982) acc1: 99.7396 (89.0559) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1770 data: 0.0834 max mem: 3539 +Test: [1830/2387] eta: 0:01:47 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1517 (0.2975) acc1: 98.9583 (89.1076) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1775 data: 0.0859 max mem: 3539 +Test: [1840/2387] eta: 0:01:45 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2037 (0.2971) acc1: 98.4375 (89.1580) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1714 data: 0.0776 max mem: 3539 +Test: [1850/2387] eta: 0:01:43 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2102 (0.2966) acc1: 98.4375 (89.2087) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1694 data: 0.0761 max mem: 3539 +Test: [1860/2387] eta: 0:01:41 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2098 (0.2962) acc1: 98.4375 (89.2566) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1736 data: 0.0823 max mem: 3539 +Test: [1870/2387] eta: 0:01:39 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2104 (0.2957) acc1: 98.1771 (89.3045) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1728 data: 0.0813 max mem: 3539 +Test: [1880/2387] eta: 0:01:37 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2103 (0.2952) acc1: 98.1771 (89.3521) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1743 data: 0.0822 max mem: 3539 +Test: [1890/2387] eta: 0:01:35 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2099 (0.2948) acc1: 98.4375 (89.4000) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1802 data: 0.0884 max mem: 3539 +Test: [1900/2387] eta: 0:01:33 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2099 (0.2944) acc1: 98.6979 (89.4458) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1767 data: 0.0855 max mem: 3539 +Test: [1910/2387] eta: 0:01:31 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3108 (0.2947) acc1: 94.5312 (89.4576) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1754 data: 0.0820 max mem: 3539 +Test: [1920/2387] eta: 0:01:29 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3367 (0.2953) acc1: 91.4062 (89.4173) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.1831 data: 0.0902 max mem: 3539 +Test: [1930/2387] eta: 0:01:27 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2656 (0.2950) acc1: 92.1875 (89.4421) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1798 data: 0.0877 max mem: 3539 +Test: [1940/2387] eta: 0:01:25 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2150 (0.2951) acc1: 95.8333 (89.4378) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1766 data: 0.0816 max mem: 3539 +Test: [1950/2387] eta: 0:01:23 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2409 (0.2952) acc1: 93.4896 (89.4360) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1824 data: 0.0876 max mem: 3539 +Test: [1960/2387] eta: 0:01:21 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2729 (0.2952) acc1: 90.6250 (89.4386) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.1882 data: 0.0942 max mem: 3539 +Test: [1970/2387] eta: 0:01:19 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3390 (0.2967) acc1: 83.0729 (89.3521) acc5: 100.0000 (100.0000) acc1_neg: 72.135 acc5_neg: 100.000 time: 0.1894 data: 0.0957 max mem: 3539 +Test: [1980/2387] eta: 0:01:18 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.4813 (0.2973) acc1: 73.4375 (89.3032) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1872 data: 0.0953 max mem: 3539 +Test: [1990/2387] eta: 0:01:16 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2329 (0.2970) acc1: 94.2708 (89.3319) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1803 data: 0.0889 max mem: 3539 +Test: [2000/2387] eta: 0:01:14 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2351 (0.2967) acc1: 94.5312 (89.3577) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1849 data: 0.0930 max mem: 3539 +Test: [2010/2387] eta: 0:01:12 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2387 (0.2965) acc1: 94.5312 (89.3848) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1939 data: 0.1018 max mem: 3539 +Test: [2020/2387] eta: 0:01:10 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2406 (0.2962) acc1: 94.2708 (89.4094) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1943 data: 0.1032 max mem: 3539 +Test: [2030/2387] eta: 0:01:08 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2406 (0.2959) acc1: 94.5312 (89.4350) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2005 data: 0.1091 max mem: 3540 +Test: [2040/2387] eta: 0:01:06 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2341 (0.2956) acc1: 94.7917 (89.4617) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1956 data: 0.1037 max mem: 3540 +Test: [2050/2387] eta: 0:01:04 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2398 (0.2955) acc1: 94.2708 (89.4840) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2037 data: 0.1127 max mem: 3540 +Test: [2060/2387] eta: 0:01:02 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2860 (0.2955) acc1: 92.7083 (89.4956) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1994 data: 0.1089 max mem: 3540 +Test: [2070/2387] eta: 0:01:00 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2943 (0.2955) acc1: 92.4479 (89.5112) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1885 data: 0.0984 max mem: 3540 +Test: [2080/2387] eta: 0:00:58 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2941 (0.2955) acc1: 92.4479 (89.5249) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.1897 data: 0.0998 max mem: 3540 +Test: [2090/2387] eta: 0:00:56 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2944 (0.2955) acc1: 92.4479 (89.5410) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1933 data: 0.1027 max mem: 3540 +Test: [2100/2387] eta: 0:00:55 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2937 (0.2955) acc1: 92.9688 (89.5573) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1937 data: 0.1028 max mem: 3540 +Test: [2110/2387] eta: 0:00:53 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2921 (0.2955) acc1: 92.9688 (89.5731) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1938 data: 0.1033 max mem: 3540 +Test: [2120/2387] eta: 0:00:51 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2979 (0.2963) acc1: 91.9271 (89.5249) acc5: 100.0000 (100.0000) acc1_neg: 61.979 acc5_neg: 100.000 time: 0.2048 data: 0.1148 max mem: 3540 +Test: [2130/2387] eta: 0:00:49 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.4034 (0.2968) acc1: 81.7708 (89.4880) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1845 data: 0.0926 max mem: 3540 +Test: [2140/2387] eta: 0:00:47 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3480 (0.2974) acc1: 82.8125 (89.4499) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1644 data: 0.0718 max mem: 3540 +Test: [2150/2387] eta: 0:00:45 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3556 (0.2978) acc1: 85.1562 (89.4325) acc5: 100.0000 (100.0000) acc1_neg: 78.646 acc5_neg: 100.000 time: 0.1637 data: 0.0724 max mem: 3540 +Test: [2160/2387] eta: 0:00:43 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3543 (0.2981) acc1: 86.4583 (89.4125) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.1630 data: 0.0712 max mem: 3540 +Test: [2170/2387] eta: 0:00:41 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.4391 (0.3000) acc1: 77.0833 (89.2964) acc5: 100.0000 (100.0000) acc1_neg: 57.812 acc5_neg: 100.000 time: 0.1800 data: 0.0882 max mem: 3540 +Test: [2180/2387] eta: 0:00:39 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.8026 (0.3033) acc1: 55.2083 (89.0712) acc5: 100.0000 (100.0000) acc1_neg: 19.010 acc5_neg: 100.000 time: 0.1727 data: 0.0802 max mem: 3540 +Test: [2190/2387] eta: 0:00:37 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.6183 (0.3038) acc1: 64.3229 (89.0381) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1576 data: 0.0649 max mem: 3540 +Test: [2200/2387] eta: 0:00:35 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2318 (0.3034) acc1: 95.8333 (89.0735) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1922 data: 0.0993 max mem: 3540 +Test: [2210/2387] eta: 0:00:33 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1942 (0.3029) acc1: 96.8750 (89.1036) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2154 data: 0.1233 max mem: 3540 +Test: [2220/2387] eta: 0:00:31 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1942 (0.3025) acc1: 96.3542 (89.1374) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2016 data: 0.1108 max mem: 3540 +Test: [2230/2387] eta: 0:00:30 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1920 (0.3019) acc1: 97.6562 (89.1762) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1924 data: 0.1019 max mem: 3540 +Test: [2240/2387] eta: 0:00:28 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1902 (0.3015) acc1: 97.6562 (89.2042) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1891 data: 0.0980 max mem: 3540 +Test: [2250/2387] eta: 0:00:26 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2113 (0.3012) acc1: 96.6146 (89.2371) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1926 data: 0.1006 max mem: 3540 +Test: [2260/2387] eta: 0:00:24 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2344 (0.3011) acc1: 96.6146 (89.2595) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1999 data: 0.1080 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2581 (0.3013) acc1: 93.2292 (89.2538) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1737 data: 0.0829 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2581 (0.3013) acc1: 90.3646 (89.2532) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1470 data: 0.0571 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2258 (0.3012) acc1: 96.0938 (89.2684) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.1489 data: 0.0589 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2617 (0.3011) acc1: 92.4479 (89.2787) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1541 data: 0.0637 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3003 (0.3017) acc1: 89.5833 (89.2426) acc5: 100.0000 (100.0000) acc1_neg: 59.375 acc5_neg: 100.000 time: 0.1537 data: 0.0631 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.4974 (0.3028) acc1: 75.2604 (89.1710) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.1476 data: 0.0572 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.4868 (0.3036) acc1: 77.3438 (89.1172) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.1369 data: 0.0466 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3410 (0.3040) acc1: 84.6354 (89.0906) acc5: 100.0000 (100.0000) acc1_neg: 68.490 acc5_neg: 100.000 time: 0.1211 data: 0.0314 max mem: 3541 +Test: [2350/2387] eta: 0:00:07 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3128 (0.3044) acc1: 88.2812 (89.0667) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1134 data: 0.0239 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3022 (0.3050) acc1: 86.9792 (89.0271) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.1156 data: 0.0261 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.4131 (0.3055) acc1: 82.0312 (88.9974) acc5: 100.0000 (100.0000) acc1_neg: 66.406 acc5_neg: 100.000 time: 0.1109 data: 0.0213 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.5160 (0.3073) acc1: 71.8750 (88.8759) acc5: 100.0000 (100.0000) acc1_neg: 55.729 acc5_neg: 100.000 time: 0.1033 data: 0.0137 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.6732 (0.3083) acc1: 63.0208 (88.8147) acc5: 100.0000 (100.0000) acc1_neg: 69.767 acc5_neg: 100.000 time: 0.0990 data: 0.0121 max mem: 3541 +Test: Total time: 0:07:29 (0.1884 s / it) +* Overall Acc@1 88.815 Acc@5 100.000 + target == real Acc@1 69.985 Acc@5 100.000 + target == fake Acc@1 90.137 Acc@5 100.000 + loss 0.308 + PR-AUC 0.9925 ROC-AUC 0.9086 EER 16.86% (thr=0.6538) +Accuracy of the network on the 916267 test images: 88.8% +[rank0]:[W726 03:32:18.658763272 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:32:40.637000 3137137 torch/distributed/run.py:766] +W0726 03:32:40.637000 3137137 torch/distributed/run.py:766] ***************************************** +W0726 03:32:40.637000 3137137 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:32:40.637000 3137137 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 03:33:16.503198896 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:33:16.603511972 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:33:16.746189038 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:33:17.320708948 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:33:17.363290865 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:33:17.392532382 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:33:17.444053552 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:33:17.493670085 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n11_seed5_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed5_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 0:57:37 acc1_pos: 85.677 acc5_pos: 100.000 loss: 0.3872 (0.3872) acc1: 85.6771 (85.6771) acc5: 100.0000 (100.0000) time: 6.6868 data: 4.9125 max mem: 3531 +Test: [ 10/517] eta: 0:05:50 acc1_pos: 46.615 acc5_pos: 100.000 loss: 0.7577 (0.7833) acc1: 53.1250 (51.7519) acc5: 100.0000 (100.0000) time: 0.6904 data: 0.4470 max mem: 3531 +Test: [ 20/517] eta: 0:03:23 acc1_pos: 42.448 acc5_pos: 100.000 loss: 0.8092 (0.8411) acc1: 47.9167 (49.4172) acc5: 100.0000 (100.0000) time: 0.0954 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:03:05 acc1_pos: 29.948 acc5_pos: 100.000 loss: 0.8195 (0.8403) acc1: 44.7917 (48.6643) acc5: 100.0000 (100.0000) time: 0.2099 data: 0.1139 max mem: 3531 +Test: [ 40/517] eta: 0:03:02 acc1_pos: 47.396 acc5_pos: 100.000 loss: 0.8195 (0.8234) acc1: 47.3958 (49.6443) acc5: 100.0000 (100.0000) time: 0.3549 data: 0.2606 max mem: 3531 +Test: [ 50/517] eta: 0:02:54 acc1_pos: 73.853 acc5_pos: 100.000 loss: 0.7225 (0.8163) acc1: 53.6458 (50.0562) acc5: 100.0000 (100.0000) acc1_neg: 34.375 acc5_neg: 100.000 time: 0.3619 data: 0.2686 max mem: 3531 +Test: [ 60/517] eta: 0:02:41 acc1_pos: 73.853 acc5_pos: 100.000 loss: 0.8053 (0.8205) acc1: 48.4375 (49.5517) acc5: 100.0000 (100.0000) acc1_neg: 58.333 acc5_neg: 100.000 time: 0.2917 data: 0.1987 max mem: 3531 +Test: [ 70/517] eta: 0:02:31 acc1_pos: 73.853 acc5_pos: 100.000 loss: 0.5679 (0.7728) acc1: 68.2292 (53.6422) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.2529 data: 0.1560 max mem: 3531 +Test: [ 80/517] eta: 0:02:25 acc1_pos: 73.853 acc5_pos: 100.000 loss: 0.5199 (0.7520) acc1: 72.3958 (55.1730) acc5: 100.0000 (100.0000) acc1_neg: 54.688 acc5_neg: 100.000 time: 0.2690 data: 0.1744 max mem: 3531 +Test: [ 90/517] eta: 0:02:21 acc1_pos: 73.853 acc5_pos: 100.000 loss: 0.6993 (0.7528) acc1: 64.0625 (56.1813) acc5: 100.0000 (100.0000) acc1_neg: 57.292 acc5_neg: 100.000 time: 0.2999 data: 0.2070 max mem: 3531 +Test: [100/517] eta: 0:02:13 acc1_pos: 73.853 acc5_pos: 100.000 loss: 0.7048 (0.7441) acc1: 64.5833 (57.1189) acc5: 100.0000 (100.0000) acc1_neg: 69.792 acc5_neg: 100.000 time: 0.2692 data: 0.1747 max mem: 3531 +Test: [110/517] eta: 0:02:07 acc1_pos: 73.853 acc5_pos: 100.000 loss: 0.6649 (0.7436) acc1: 67.1875 (57.7492) acc5: 100.0000 (100.0000) acc1_neg: 72.917 acc5_neg: 100.000 time: 0.2339 data: 0.1394 max mem: 3531 +Test: [120/517] eta: 0:02:01 acc1_pos: 73.853 acc5_pos: 100.000 loss: 0.6765 (0.7409) acc1: 58.0729 (57.6855) acc5: 100.0000 (100.0000) acc1_neg: 55.469 acc5_neg: 100.000 time: 0.2444 data: 0.1477 max mem: 3531 +Test: [130/517] eta: 0:01:57 acc1_pos: 73.853 acc5_pos: 100.000 loss: 0.6548 (0.7340) acc1: 57.5521 (58.1345) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.2563 data: 0.1601 max mem: 3531 +Test: [140/517] eta: 0:01:53 acc1_pos: 73.853 acc5_pos: 100.000 loss: 0.5893 (0.7203) acc1: 69.2708 (59.1127) acc5: 100.0000 (100.0000) acc1_neg: 75.521 acc5_neg: 100.000 time: 0.2641 data: 0.1694 max mem: 3531 +Test: [150/517] eta: 0:01:49 acc1_pos: 73.853 acc5_pos: 100.000 loss: 0.5158 (0.7027) acc1: 75.5208 (60.5805) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.2539 data: 0.1597 max mem: 3531 +Test: [160/517] eta: 0:01:45 acc1_pos: 73.853 acc5_pos: 100.000 loss: 0.4866 (0.6938) acc1: 78.3854 (61.3370) acc5: 100.0000 (100.0000) acc1_neg: 79.427 acc5_neg: 100.000 time: 0.2557 data: 0.1632 max mem: 3531 +Test: [170/517] eta: 0:01:41 acc1_pos: 73.853 acc5_pos: 100.000 loss: 0.5067 (0.6811) acc1: 77.3438 (62.3660) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.2545 data: 0.1594 max mem: 3531 +Test: [180/517] eta: 0:01:37 acc1_pos: 73.853 acc5_pos: 100.000 loss: 0.4650 (0.6815) acc1: 77.3438 (62.2525) acc5: 100.0000 (100.0000) acc1_neg: 52.604 acc5_neg: 100.000 time: 0.2391 data: 0.1422 max mem: 3531 +Test: [190/517] eta: 0:01:33 acc1_pos: 73.853 acc5_pos: 100.000 loss: 0.3638 (0.6638) acc1: 85.6771 (63.6044) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.2299 data: 0.1325 max mem: 3531 +Test: [200/517] eta: 0:01:29 acc1_pos: 73.853 acc5_pos: 100.000 loss: 0.3681 (0.6539) acc1: 85.1562 (64.4097) acc5: 100.0000 (100.0000) acc1_neg: 72.917 acc5_neg: 100.000 time: 0.2299 data: 0.1328 max mem: 3531 +Test: [210/517] eta: 0:01:26 acc1_pos: 73.853 acc5_pos: 100.000 loss: 0.5731 (0.6576) acc1: 72.9167 (64.6018) acc5: 100.0000 (100.0000) acc1_neg: 67.969 acc5_neg: 100.000 time: 0.2362 data: 0.1405 max mem: 3532 +Test: [220/517] eta: 0:01:22 acc1_pos: 73.853 acc5_pos: 100.000 loss: 0.7268 (0.6607) acc1: 69.2708 (64.7778) acc5: 100.0000 (100.0000) acc1_neg: 69.792 acc5_neg: 100.000 time: 0.2400 data: 0.1457 max mem: 3532 +Test: [230/517] eta: 0:01:19 acc1_pos: 73.853 acc5_pos: 100.000 loss: 0.7282 (0.6662) acc1: 63.5417 (64.2260) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.2386 data: 0.1436 max mem: 3532 +Test: [240/517] eta: 0:01:16 acc1_pos: 73.853 acc5_pos: 100.000 loss: 0.5730 (0.6604) acc1: 63.0208 (64.5217) acc5: 100.0000 (100.0000) acc1_neg: 50.260 acc5_neg: 100.000 time: 0.2397 data: 0.1430 max mem: 3532 +Test: [250/517] eta: 0:01:13 acc1_pos: 73.853 acc5_pos: 100.000 loss: 0.4755 (0.6535) acc1: 77.3438 (65.0803) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.2561 data: 0.1595 max mem: 3532 +Test: [260/517] eta: 0:01:10 acc1_pos: 73.853 acc5_pos: 100.000 loss: 0.4574 (0.6484) acc1: 79.9479 (65.4694) acc5: 100.0000 (100.0000) acc1_neg: 54.688 acc5_neg: 100.000 time: 0.2502 data: 0.1553 max mem: 3532 +Test: [270/517] eta: 0:01:07 acc1_pos: 73.853 acc5_pos: 100.000 loss: 0.7181 (0.6529) acc1: 57.8125 (65.0283) acc5: 100.0000 (100.0000) acc1_neg: 52.344 acc5_neg: 100.000 time: 0.2436 data: 0.1487 max mem: 3532 +Test: [280/517] eta: 0:01:04 acc1_pos: 73.853 acc5_pos: 100.000 loss: 0.7446 (0.6530) acc1: 54.6875 (65.0161) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.2545 data: 0.1579 max mem: 3532 +Test: [290/517] eta: 0:01:01 acc1_pos: 73.853 acc5_pos: 100.000 loss: 0.5258 (0.6474) acc1: 74.2188 (65.4335) acc5: 100.0000 (100.0000) acc1_neg: 74.740 acc5_neg: 100.000 time: 0.2467 data: 0.1506 max mem: 3532 +Test: [300/517] eta: 0:00:58 acc1_pos: 73.853 acc5_pos: 100.000 loss: 0.4566 (0.6416) acc1: 80.7292 (65.8811) acc5: 100.0000 (100.0000) acc1_neg: 57.812 acc5_neg: 100.000 time: 0.2499 data: 0.1555 max mem: 3532 +Test: [310/517] eta: 0:00:56 acc1_pos: 73.853 acc5_pos: 100.000 loss: 0.6150 (0.6438) acc1: 60.4167 (65.6526) acc5: 100.0000 (100.0000) acc1_neg: 59.896 acc5_neg: 100.000 time: 0.2543 data: 0.1594 max mem: 3532 +Test: [320/517] eta: 0:00:53 acc1_pos: 73.853 acc5_pos: 100.000 loss: 0.6992 (0.6443) acc1: 58.8542 (65.5869) acc5: 100.0000 (100.0000) acc1_neg: 65.365 acc5_neg: 100.000 time: 0.2552 data: 0.1601 max mem: 3532 +Test: [330/517] eta: 0:00:50 acc1_pos: 73.853 acc5_pos: 100.000 loss: 0.5929 (0.6406) acc1: 71.8750 (65.9216) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.2587 data: 0.1629 max mem: 3532 +Test: [340/517] eta: 0:00:47 acc1_pos: 73.853 acc5_pos: 100.000 loss: 0.5111 (0.6413) acc1: 72.3958 (65.8740) acc5: 100.0000 (100.0000) acc1_neg: 40.625 acc5_neg: 100.000 time: 0.2608 data: 0.1635 max mem: 3532 +Test: [350/517] eta: 0:00:44 acc1_pos: 73.853 acc5_pos: 100.000 loss: 0.6600 (0.6432) acc1: 64.3229 (65.6703) acc5: 100.0000 (100.0000) acc1_neg: 48.698 acc5_neg: 100.000 time: 0.2461 data: 0.1507 max mem: 3532 +Test: [360/517] eta: 0:00:42 acc1_pos: 73.853 acc5_pos: 100.000 loss: 0.6451 (0.6440) acc1: 63.5417 (65.5897) acc5: 100.0000 (100.0000) acc1_neg: 56.510 acc5_neg: 100.000 time: 0.2386 data: 0.1417 max mem: 3532 +Test: [370/517] eta: 0:00:39 acc1_pos: 73.853 acc5_pos: 100.000 loss: 0.5489 (0.6381) acc1: 70.3125 (66.0897) acc5: 100.0000 (100.0000) acc1_neg: 63.542 acc5_neg: 100.000 time: 0.2487 data: 0.1491 max mem: 3532 +Test: [380/517] eta: 0:00:36 acc1_pos: 73.853 acc5_pos: 100.000 loss: 0.4836 (0.6356) acc1: 78.1250 (66.2873) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.2499 data: 0.1511 max mem: 3532 +Test: [390/517] eta: 0:00:33 acc1_pos: 73.853 acc5_pos: 100.000 loss: 0.4860 (0.6309) acc1: 75.7812 (66.5774) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.2454 data: 0.1480 max mem: 3532 +Test: [400/517] eta: 0:00:31 acc1_pos: 73.853 acc5_pos: 100.000 loss: 0.4740 (0.6286) acc1: 75.5208 (66.7368) acc5: 100.0000 (100.0000) acc1_neg: 58.594 acc5_neg: 100.000 time: 0.2458 data: 0.1498 max mem: 3532 +Test: [410/517] eta: 0:00:28 acc1_pos: 73.853 acc5_pos: 100.000 loss: 0.5904 (0.6314) acc1: 67.4479 (66.4817) acc5: 100.0000 (100.0000) acc1_neg: 58.333 acc5_neg: 100.000 time: 0.2493 data: 0.1538 max mem: 3532 +Test: [420/517] eta: 0:00:25 acc1_pos: 73.853 acc5_pos: 100.000 loss: 0.6277 (0.6304) acc1: 63.5417 (66.5269) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.2440 data: 0.1480 max mem: 3532 +Test: [430/517] eta: 0:00:22 acc1_pos: 73.853 acc5_pos: 100.000 loss: 0.5795 (0.6295) acc1: 69.0104 (66.5688) acc5: 100.0000 (100.0000) acc1_neg: 35.677 acc5_neg: 100.000 time: 0.2399 data: 0.1424 max mem: 3532 +Test: [440/517] eta: 0:00:20 acc1_pos: 73.853 acc5_pos: 100.000 loss: 0.6662 (0.6333) acc1: 57.0312 (66.1978) acc5: 100.0000 (100.0000) acc1_neg: 47.917 acc5_neg: 100.000 time: 0.2399 data: 0.1417 max mem: 3533 +Test: [450/517] eta: 0:00:17 acc1_pos: 73.853 acc5_pos: 100.000 loss: 0.6545 (0.6319) acc1: 61.9792 (66.3150) acc5: 100.0000 (100.0000) acc1_neg: 77.344 acc5_neg: 100.000 time: 0.2449 data: 0.1473 max mem: 3533 +Test: [460/517] eta: 0:00:14 acc1_pos: 73.853 acc5_pos: 100.000 loss: 0.5189 (0.6297) acc1: 75.7812 (66.5650) acc5: 100.0000 (100.0000) acc1_neg: 63.021 acc5_neg: 100.000 time: 0.2399 data: 0.1429 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 73.853 acc5_pos: 100.000 loss: 0.5189 (0.6281) acc1: 73.1771 (66.6396) acc5: 100.0000 (100.0000) acc1_neg: 80.729 acc5_neg: 100.000 time: 0.2454 data: 0.1489 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 73.853 acc5_pos: 100.000 loss: 0.5084 (0.6233) acc1: 72.3958 (67.0359) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2399 data: 0.1427 max mem: 3533 +Test: [490/517] eta: 0:00:07 acc1_pos: 73.853 acc5_pos: 100.000 loss: 0.4140 (0.6219) acc1: 80.9896 (67.1154) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.2261 data: 0.1288 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 73.853 acc5_pos: 100.000 loss: 0.5682 (0.6218) acc1: 67.7083 (67.0518) acc5: 100.0000 (100.0000) acc1_neg: 59.896 acc5_neg: 100.000 time: 0.2283 data: 0.1334 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 73.853 acc5_pos: 100.000 loss: 0.6140 (0.6238) acc1: 61.1979 (66.9587) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1874 data: 0.0960 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 73.853 acc5_pos: 100.000 loss: 0.6347 (0.6234) acc1: 66.1458 (67.0151) acc5: 100.0000 (100.0000) acc1_neg: 64.311 acc5_neg: 100.000 time: 0.1876 data: 0.0960 max mem: 3533 +Test: Total time: 0:02:12 (0.2569 s / it) +* Overall Acc@1 67.015 Acc@5 100.000 + target == real Acc@1 50.547 Acc@5 100.000 + target == fake Acc@1 68.723 Acc@5 100.000 + loss 0.623 + PR-AUC 0.9345 ROC-AUC 0.6290 EER 40.84% (thr=0.5708) +Accuracy of the network on the 198427 test images: 67.0% +[rank0]:[W726 03:35:40.477933091 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 03:36:02.509000 3137709 torch/distributed/run.py:766] +W0726 03:36:02.509000 3137709 torch/distributed/run.py:766] ***************************************** +W0726 03:36:02.509000 3137709 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:36:02.509000 3137709 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 03:36:35.448172215 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:36:35.799053303 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:36:36.283197597 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:36:36.345488623 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:36:36.373369971 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:36:36.498551753 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:36:36.502226138 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:36:36.534532874 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n11_seed5_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed5_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:32:51 acc1_pos: 72.656 acc5_pos: 100.000 loss: 0.5677 (0.5677) acc1: 72.6562 (72.6562) acc5: 100.0000 (100.0000) time: 5.0684 data: 2.7204 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/389] eta: 0:03:25 acc1_pos: 90.885 acc5_pos: 100.000 loss: 0.5677 (0.5421) acc1: 72.6562 (70.0284) acc5: 100.0000 (100.0000) time: 0.5424 data: 0.2477 max mem: 3531 +Test: [ 20/389] eta: 0:02:02 acc1_pos: 85.677 acc5_pos: 100.000 loss: 0.4028 (0.4830) acc1: 85.6771 (75.9673) acc5: 100.0000 (100.0000) time: 0.0939 data: 0.0044 max mem: 3531 +Test: [ 30/389] eta: 0:01:46 acc1_pos: 77.344 acc5_pos: 100.000 loss: 0.4597 (0.4936) acc1: 85.4167 (76.8817) acc5: 100.0000 (100.0000) time: 0.1597 data: 0.0674 max mem: 3531 +Test: [ 40/389] eta: 0:01:38 acc1_pos: 40.365 acc5_pos: 100.000 loss: 0.5091 (0.5023) acc1: 77.3438 (76.4736) acc5: 100.0000 (100.0000) time: 0.2333 data: 0.1388 max mem: 3531 +Test: [ 50/389] eta: 0:01:33 acc1_pos: 69.792 acc5_pos: 100.000 loss: 0.4952 (0.5019) acc1: 76.0417 (76.7361) acc5: 100.0000 (100.0000) time: 0.2462 data: 0.1544 max mem: 3531 +Test: [ 60/389] eta: 0:01:23 acc1_pos: 19.271 acc5_pos: 100.000 loss: 0.6109 (0.4990) acc1: 70.0521 (76.4259) acc5: 100.0000 (100.0000) time: 0.1923 data: 0.1001 max mem: 3531 +Test: [ 70/389] eta: 0:01:17 acc1_pos: 90.365 acc5_pos: 100.000 loss: 0.6109 (0.5100) acc1: 66.1458 (74.8166) acc5: 100.0000 (100.0000) time: 0.1636 data: 0.0696 max mem: 3531 +Test: [ 80/389] eta: 0:01:13 acc1_pos: 98.698 acc5_pos: 100.000 loss: 0.4792 (0.5078) acc1: 75.5208 (75.6655) acc5: 100.0000 (100.0000) time: 0.1921 data: 0.1005 max mem: 3531 +Test: [ 90/389] eta: 0:01:09 acc1_pos: 75.781 acc5_pos: 100.000 loss: 0.3625 (0.4799) acc1: 91.9271 (77.6099) acc5: 100.0000 (100.0000) time: 0.1830 data: 0.0924 max mem: 3531 +Test: [100/389] eta: 0:01:05 acc1_pos: 76.042 acc5_pos: 100.000 loss: 0.5300 (0.4957) acc1: 76.0417 (76.1783) acc5: 100.0000 (100.0000) time: 0.1727 data: 0.0799 max mem: 3531 +Test: [110/389] eta: 0:01:01 acc1_pos: 99.740 acc5_pos: 100.000 loss: 0.6068 (0.5018) acc1: 68.4896 (75.3191) acc5: 100.0000 (100.0000) time: 0.1768 data: 0.0825 max mem: 3531 +Test: [120/389] eta: 0:00:58 acc1_pos: 92.708 acc5_pos: 100.000 loss: 0.4728 (0.5006) acc1: 79.4271 (75.2733) acc5: 100.0000 (100.0000) time: 0.1799 data: 0.0862 max mem: 3531 +Test: [130/389] eta: 0:00:55 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5689 (0.5153) acc1: 62.5000 (73.9941) acc5: 100.0000 (100.0000) acc1_neg: 77.344 acc5_neg: 100.000 time: 0.1699 data: 0.0751 max mem: 3531 +Test: [140/389] eta: 0:00:52 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.8980 (0.5538) acc1: 36.7188 (70.6302) acc5: 100.0000 (100.0000) acc1_neg: 0.781 acc5_neg: 100.000 time: 0.1758 data: 0.0813 max mem: 3531 +Test: [150/389] eta: 0:00:50 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.9380 (0.5895) acc1: 31.2500 (68.5430) acc5: 100.0000 (100.0000) acc1_neg: 17.448 acc5_neg: 100.000 time: 0.1924 data: 0.1002 max mem: 3531 +Test: [160/389] eta: 0:00:47 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.7313 (0.6008) acc1: 45.8333 (67.3670) acc5: 100.0000 (100.0000) acc1_neg: 63.802 acc5_neg: 100.000 time: 0.1790 data: 0.0853 max mem: 3531 +Test: [170/389] eta: 0:00:45 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6367 (0.6018) acc1: 63.8021 (67.4205) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.1725 data: 0.0781 max mem: 3531 +Test: [180/389] eta: 0:00:42 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5865 (0.6176) acc1: 70.0521 (65.9847) acc5: 100.0000 (100.0000) acc1_neg: 53.125 acc5_neg: 100.000 time: 0.1799 data: 0.0851 max mem: 3531 +Test: [190/389] eta: 0:00:40 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.8495 (0.6363) acc1: 33.0729 (64.5533) acc5: 100.0000 (100.0000) acc1_neg: 7.552 acc5_neg: 100.000 time: 0.1654 data: 0.0718 max mem: 3531 +Test: [200/389] eta: 0:00:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.8923 (0.6508) acc1: 28.6458 (63.0610) acc5: 100.0000 (100.0000) acc1_neg: 11.719 acc5_neg: 100.000 time: 0.1559 data: 0.0625 max mem: 3531 +Test: [210/389] eta: 0:00:35 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.8262 (0.6553) acc1: 28.6458 (62.5407) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1795 data: 0.0844 max mem: 3532 +Test: [220/389] eta: 0:00:33 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5258 (0.6484) acc1: 64.0625 (63.2447) acc5: 100.0000 (100.0000) acc1_neg: 28.646 acc5_neg: 100.000 time: 0.1894 data: 0.0967 max mem: 3532 +Test: [230/389] eta: 0:00:31 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6196 (0.6546) acc1: 49.7396 (62.4030) acc5: 100.0000 (100.0000) acc1_neg: 4.167 acc5_neg: 100.000 time: 0.1749 data: 0.0835 max mem: 3532 +Test: [240/389] eta: 0:00:29 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.8061 (0.6595) acc1: 43.4896 (61.8430) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.1753 data: 0.0837 max mem: 3532 +Test: [250/389] eta: 0:00:27 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6972 (0.6589) acc1: 53.9062 (61.8765) acc5: 100.0000 (100.0000) acc1_neg: 56.771 acc5_neg: 100.000 time: 0.1852 data: 0.0923 max mem: 3532 +Test: [260/389] eta: 0:00:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.7103 (0.6764) acc1: 52.8646 (60.3099) acc5: 100.0000 (100.0000) acc1_neg: 4.948 acc5_neg: 100.000 time: 0.1918 data: 0.0998 max mem: 3532 +Test: [270/389] eta: 0:00:23 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.1087 (0.6904) acc1: 12.5000 (58.8897) acc5: 100.0000 (100.0000) acc1_neg: 17.448 acc5_neg: 100.000 time: 0.1946 data: 0.1028 max mem: 3532 +Test: [280/389] eta: 0:00:21 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.1208 (0.7081) acc1: 9.1146 (57.2481) acc5: 100.0000 (100.0000) acc1_neg: 52.083 acc5_neg: 100.000 time: 0.1751 data: 0.0827 max mem: 3532 +Test: [290/389] eta: 0:00:19 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.1075 (0.7141) acc1: 10.4167 (56.7762) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1567 data: 0.0642 max mem: 3532 +Test: [300/389] eta: 0:00:17 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.0234 (0.7233) acc1: 9.8958 (55.6556) acc5: 100.0000 (100.0000) acc1_neg: 7.552 acc5_neg: 100.000 time: 0.1670 data: 0.0734 max mem: 3532 +Test: [310/389] eta: 0:00:15 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.0998 (0.7328) acc1: 8.5938 (54.9546) acc5: 100.0000 (100.0000) acc1_neg: 76.302 acc5_neg: 100.000 time: 0.1744 data: 0.0819 max mem: 3532 +Test: [320/389] eta: 0:00:13 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.7976 (0.7319) acc1: 45.8333 (55.0307) acc5: 100.0000 (100.0000) acc1_neg: 42.448 acc5_neg: 100.000 time: 0.1803 data: 0.0894 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.8487 (0.7366) acc1: 33.3333 (54.1793) acc5: 100.0000 (100.0000) acc1_neg: 42.188 acc5_neg: 100.000 time: 0.1823 data: 0.0914 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.8606 (0.7342) acc1: 34.8958 (54.4729) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.1667 data: 0.0758 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.8101 (0.7452) acc1: 41.1458 (53.6644) acc5: 100.0000 (100.0000) acc1_neg: 1.823 acc5_neg: 100.000 time: 0.1662 data: 0.0730 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.0836 (0.7554) acc1: 17.1875 (52.5998) acc5: 100.0000 (100.0000) acc1_neg: 18.750 acc5_neg: 100.000 time: 0.1659 data: 0.0730 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.0610 (0.7552) acc1: 19.7917 (52.5719) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1529 data: 0.0620 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.7521 (0.7583) acc1: 46.6146 (52.4750) acc5: 100.0000 (100.0000) acc1_neg: 14.323 acc5_neg: 100.000 time: 0.1393 data: 0.0483 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.8659 (0.7640) acc1: 38.2812 (52.0046) acc5: 100.0000 (100.0000) acc1_neg: 9.836 acc5_neg: 100.000 time: 0.1305 data: 0.0370 max mem: 3532 +Test: Total time: 0:01:12 (0.1861 s / it) +* Overall Acc@1 52.005 Acc@5 100.000 + target == real Acc@1 75.753 Acc@5 100.000 + target == fake Acc@1 40.966 Acc@5 100.000 + loss 0.764 + PR-AUC 0.7771 ROC-AUC 0.6513 EER 40.61% (thr=0.3906) +Accuracy of the network on the 149358 test images: 52.0% +[rank0]:[W726 03:37:58.618476575 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:38:20.045000 3138267 torch/distributed/run.py:766] +W0726 03:38:20.045000 3138267 torch/distributed/run.py:766] ***************************************** +W0726 03:38:20.045000 3138267 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:38:20.045000 3138267 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:38:55.737376704 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:38:56.376329386 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:38:56.733314167 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:38:56.757836703 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +| distributed init (rank 7): env:// +[rank0]:[W726 03:38:56.865774390 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:38:56.877155089 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W726 03:38:56.877207618 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W726 03:38:56.877238961 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n11_seed5_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed5_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:40:05 acc1_pos: 73.438 acc5_pos: 100.000 loss: 0.5367 (0.5367) acc1: 73.4375 (73.4375) acc5: 100.0000 (100.0000) time: 7.2234 data: 5.1430 max mem: 3531 +Test: [ 10/333] eta: 0:03:58 acc1_pos: 78.646 acc5_pos: 100.000 loss: 0.3756 (0.3682) acc1: 81.7708 (83.4754) acc5: 100.0000 (100.0000) time: 0.7386 data: 0.4680 max mem: 3531 +Test: [ 20/333] eta: 0:02:15 acc1_pos: 92.448 acc5_pos: 100.000 loss: 0.3752 (0.3641) acc1: 83.3333 (84.5114) acc5: 100.0000 (100.0000) time: 0.0936 data: 0.0005 max mem: 3531 +Test: [ 30/333] eta: 0:02:06 acc1_pos: 69.792 acc5_pos: 100.000 loss: 0.4349 (0.4229) acc1: 80.7292 (80.4267) acc5: 100.0000 (100.0000) time: 0.2419 data: 0.1476 max mem: 3531 +Test: [ 40/333] eta: 0:01:53 acc1_pos: 74.740 acc5_pos: 100.000 loss: 0.5663 (0.4594) acc1: 69.5312 (78.2076) acc5: 100.0000 (100.0000) time: 0.3401 data: 0.2483 max mem: 3531 +Test: [ 50/333] eta: 0:01:39 acc1_pos: 68.490 acc5_pos: 100.000 loss: 0.5750 (0.4782) acc1: 69.5312 (76.7004) acc5: 100.0000 (100.0000) time: 0.2465 data: 0.1518 max mem: 3531 +Test: [ 60/333] eta: 0:01:27 acc1_pos: 61.719 acc5_pos: 100.000 loss: 0.5594 (0.4974) acc1: 70.3125 (75.4397) acc5: 100.0000 (100.0000) time: 0.1847 data: 0.0900 max mem: 3531 +Test: [ 70/333] eta: 0:01:21 acc1_pos: 69.531 acc5_pos: 100.000 loss: 0.5594 (0.5061) acc1: 71.0938 (74.9817) acc5: 100.0000 (100.0000) time: 0.2114 data: 0.1204 max mem: 3531 +Test: [ 80/333] eta: 0:01:15 acc1_pos: 70.833 acc5_pos: 100.000 loss: 0.5495 (0.5138) acc1: 72.1354 (74.5017) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1381 max mem: 3531 +Test: [ 90/333] eta: 0:01:09 acc1_pos: 82.552 acc5_pos: 100.000 loss: 0.5031 (0.5091) acc1: 74.2188 (74.9514) acc5: 100.0000 (100.0000) time: 0.1933 data: 0.0992 max mem: 3531 +Test: [100/333] eta: 0:01:04 acc1_pos: 84.375 acc5_pos: 100.000 loss: 0.4532 (0.5054) acc1: 78.1250 (75.0155) acc5: 100.0000 (100.0000) time: 0.1856 data: 0.0919 max mem: 3531 +Test: [110/333] eta: 0:01:00 acc1_pos: 75.000 acc5_pos: 100.000 loss: 0.4692 (0.5048) acc1: 76.3021 (74.9460) acc5: 100.0000 (100.0000) time: 0.2099 data: 0.1173 max mem: 3531 +Test: [120/333] eta: 0:00:56 acc1_pos: 76.562 acc5_pos: 100.000 loss: 0.4692 (0.4992) acc1: 76.5625 (75.3078) acc5: 100.0000 (100.0000) time: 0.2142 data: 0.1210 max mem: 3531 +Test: [130/333] eta: 0:00:51 acc1_pos: 86.719 acc5_pos: 100.000 loss: 0.4378 (0.4960) acc1: 77.6042 (75.4672) acc5: 100.0000 (100.0000) time: 0.1627 data: 0.0664 max mem: 3531 +Test: [140/333] eta: 0:00:48 acc1_pos: 89.844 acc5_pos: 100.000 loss: 0.4226 (0.4897) acc1: 78.3854 (75.8699) acc5: 100.0000 (100.0000) time: 0.1570 data: 0.0597 max mem: 3531 +Test: [150/333] eta: 0:00:45 acc1_pos: 85.417 acc5_pos: 100.000 loss: 0.4365 (0.4880) acc1: 78.6458 (75.9710) acc5: 100.0000 (100.0000) time: 0.1881 data: 0.0954 max mem: 3531 +Test: [160/333] eta: 0:00:42 acc1_pos: 75.658 acc5_pos: 100.000 loss: 0.4534 (0.5072) acc1: 77.8646 (74.8156) acc5: 100.0000 (100.0000) acc1_neg: 33.073 acc5_neg: 100.000 time: 0.2042 data: 0.1138 max mem: 3531 +Test: [170/333] eta: 0:00:39 acc1_pos: 75.658 acc5_pos: 100.000 loss: 1.2085 (0.5510) acc1: 29.4271 (72.0532) acc5: 100.0000 (100.0000) acc1_neg: 32.031 acc5_neg: 100.000 time: 0.2191 data: 0.1289 max mem: 3531 +Test: [180/333] eta: 0:00:36 acc1_pos: 75.658 acc5_pos: 100.000 loss: 1.2203 (0.5871) acc1: 27.0833 (69.5859) acc5: 100.0000 (100.0000) acc1_neg: 41.146 acc5_neg: 100.000 time: 0.2085 data: 0.1175 max mem: 3531 +Test: [190/333] eta: 0:00:34 acc1_pos: 75.658 acc5_pos: 100.000 loss: 1.1155 (0.6092) acc1: 31.5104 (67.9265) acc5: 100.0000 (100.0000) acc1_neg: 37.760 acc5_neg: 100.000 time: 0.2128 data: 0.1213 max mem: 3531 +Test: [200/333] eta: 0:00:31 acc1_pos: 75.658 acc5_pos: 100.000 loss: 0.9799 (0.6276) acc1: 38.8021 (66.6991) acc5: 100.0000 (100.0000) acc1_neg: 38.802 acc5_neg: 100.000 time: 0.1925 data: 0.1013 max mem: 3531 +Test: [210/333] eta: 0:00:28 acc1_pos: 75.658 acc5_pos: 100.000 loss: 1.0585 (0.6502) acc1: 38.5417 (65.2905) acc5: 100.0000 (100.0000) acc1_neg: 44.010 acc5_neg: 100.000 time: 0.1768 data: 0.0828 max mem: 3532 +Test: [220/333] eta: 0:00:26 acc1_pos: 75.658 acc5_pos: 100.000 loss: 1.0747 (0.6767) acc1: 34.6354 (63.6265) acc5: 100.0000 (100.0000) acc1_neg: 34.635 acc5_neg: 100.000 time: 0.1982 data: 0.1039 max mem: 3532 +Test: [230/333] eta: 0:00:23 acc1_pos: 75.658 acc5_pos: 100.000 loss: 0.9283 (0.6845) acc1: 39.3229 (62.9442) acc5: 100.0000 (100.0000) acc1_neg: 38.542 acc5_neg: 100.000 time: 0.2031 data: 0.1111 max mem: 3532 +Test: [240/333] eta: 0:00:21 acc1_pos: 75.658 acc5_pos: 100.000 loss: 0.9092 (0.6963) acc1: 41.6667 (61.9727) acc5: 100.0000 (100.0000) acc1_neg: 42.708 acc5_neg: 100.000 time: 0.1999 data: 0.1069 max mem: 3532 +Test: [250/333] eta: 0:00:18 acc1_pos: 75.658 acc5_pos: 100.000 loss: 0.9889 (0.7090) acc1: 38.8021 (61.0786) acc5: 100.0000 (100.0000) acc1_neg: 39.062 acc5_neg: 100.000 time: 0.1917 data: 0.0991 max mem: 3532 +Test: [260/333] eta: 0:00:16 acc1_pos: 75.658 acc5_pos: 100.000 loss: 1.0254 (0.7219) acc1: 38.8021 (60.2091) acc5: 100.0000 (100.0000) acc1_neg: 45.052 acc5_neg: 100.000 time: 0.1968 data: 0.1010 max mem: 3532 +Test: [270/333] eta: 0:00:14 acc1_pos: 75.658 acc5_pos: 100.000 loss: 1.1079 (0.7378) acc1: 34.8958 (59.1425) acc5: 100.0000 (100.0000) acc1_neg: 30.208 acc5_neg: 100.000 time: 0.1967 data: 0.0995 max mem: 3532 +Test: [280/333] eta: 0:00:11 acc1_pos: 75.658 acc5_pos: 100.000 loss: 1.1574 (0.7529) acc1: 32.0312 (58.1573) acc5: 100.0000 (100.0000) acc1_neg: 44.792 acc5_neg: 100.000 time: 0.1874 data: 0.0941 max mem: 3532 +Test: [290/333] eta: 0:00:09 acc1_pos: 75.658 acc5_pos: 100.000 loss: 1.2398 (0.7712) acc1: 28.9062 (57.0706) acc5: 100.0000 (100.0000) acc1_neg: 22.396 acc5_neg: 100.000 time: 0.1763 data: 0.0842 max mem: 3532 +Test: [300/333] eta: 0:00:07 acc1_pos: 75.658 acc5_pos: 100.000 loss: 1.3075 (0.7924) acc1: 22.3958 (55.8737) acc5: 100.0000 (100.0000) acc1_neg: 26.042 acc5_neg: 100.000 time: 0.1426 data: 0.0518 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 75.658 acc5_pos: 100.000 loss: 1.2388 (0.7953) acc1: 26.0417 (55.6471) acc5: 100.0000 (100.0000) acc1_neg: 48.438 acc5_neg: 100.000 time: 0.1139 data: 0.0233 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 75.658 acc5_pos: 100.000 loss: 0.8342 (0.7994) acc1: 48.6979 (55.3300) acc5: 100.0000 (100.0000) acc1_neg: 32.031 acc5_neg: 100.000 time: 0.1045 data: 0.0142 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 75.658 acc5_pos: 100.000 loss: 1.2368 (0.8148) acc1: 32.0312 (54.5404) acc5: 100.0000 (100.0000) acc1_neg: 33.854 acc5_neg: 100.000 time: 0.1024 data: 0.0130 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 75.658 acc5_pos: 100.000 loss: 1.2740 (0.8177) acc1: 31.7708 (54.3941) acc5: 100.0000 (100.0000) acc1_neg: 26.197 acc5_neg: 100.000 time: 0.1044 data: 0.0116 max mem: 3532 +Test: Total time: 0:01:09 (0.2087 s / it) +* Overall Acc@1 54.394 Acc@5 100.000 + target == real Acc@1 76.086 Acc@5 100.000 + target == fake Acc@1 35.084 Acc@5 100.000 + loss 0.818 + PR-AUC 0.6070 ROC-AUC 0.6050 EER 42.92% (thr=0.3154) +Accuracy of the network on the 127843 test images: 54.4% +[rank0]:[W726 03:40:13.022417000 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:40:36.790000 3138828 torch/distributed/run.py:766] +W0726 03:40:36.790000 3138828 torch/distributed/run.py:766] ***************************************** +W0726 03:40:36.790000 3138828 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:40:36.790000 3138828 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:41:12.885965527 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:41:13.548750074 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:41:13.574209741 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:41:13.743797936 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:41:13.940797907 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:41:13.023084713 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:41:13.118034261 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:41:13.124050275 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n11_seed5_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed5_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:26:02 acc1_pos: 96.615 acc5_pos: 100.000 loss: 0.1540 (0.1540) acc1: 96.6146 (96.6146) acc5: 100.0000 (100.0000) time: 6.6475 data: 5.2608 max mem: 3531 +Test: [ 10/235] eta: 0:02:34 acc1_pos: 96.094 acc5_pos: 100.000 loss: 0.1675 (0.1670) acc1: 96.0938 (95.9991) acc5: 100.0000 (100.0000) time: 0.6886 data: 0.4786 max mem: 3531 +Test: [ 20/235] eta: 0:01:27 acc1_pos: 94.792 acc5_pos: 100.000 loss: 0.1642 (0.1618) acc1: 96.3542 (96.2674) acc5: 100.0000 (100.0000) time: 0.0964 data: 0.0004 max mem: 3531 +Test: [ 30/235] eta: 0:01:15 acc1_pos: 98.177 acc5_pos: 100.000 loss: 0.1590 (0.1609) acc1: 96.6146 (96.3626) acc5: 100.0000 (100.0000) time: 0.1949 data: 0.0992 max mem: 3531 +Test: [ 40/235] eta: 0:01:08 acc1_pos: 95.052 acc5_pos: 100.000 loss: 0.1628 (0.1616) acc1: 96.0938 (96.3351) acc5: 100.0000 (100.0000) time: 0.2899 data: 0.1959 max mem: 3531 +Test: [ 50/235] eta: 0:00:59 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.1628 (0.1620) acc1: 96.3542 (96.3491) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1454 max mem: 3531 +Test: [ 60/235] eta: 0:00:52 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.1660 (0.1638) acc1: 95.8333 (96.2645) acc5: 100.0000 (100.0000) time: 0.1910 data: 0.0981 max mem: 3531 +Test: [ 70/235] eta: 0:00:46 acc1_pos: 96.923 acc5_pos: 100.000 loss: 0.1866 (0.3711) acc1: 94.7917 (85.7908) acc5: 100.0000 (100.0000) acc1_neg: 11.458 acc5_neg: 100.000 time: 0.1964 data: 0.1047 max mem: 3531 +Test: [ 80/235] eta: 0:00:42 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.8271 (0.5518) acc1: 11.9792 (76.6686) acc5: 100.0000 (100.0000) acc1_neg: 10.938 acc5_neg: 100.000 time: 0.2107 data: 0.1202 max mem: 3531 +Test: [ 90/235] eta: 0:00:39 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.8392 (0.6934) acc1: 11.1979 (69.5255) acc5: 100.0000 (100.0000) acc1_neg: 10.156 acc5_neg: 100.000 time: 0.2183 data: 0.1269 max mem: 3531 +Test: [100/235] eta: 0:00:35 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.8363 (0.8055) acc1: 12.2396 (63.8562) acc5: 100.0000 (100.0000) acc1_neg: 11.719 acc5_neg: 100.000 time: 0.2144 data: 0.1227 max mem: 3531 +Test: [110/235] eta: 0:00:32 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.8268 (0.8961) acc1: 13.0208 (59.3093) acc5: 100.0000 (100.0000) acc1_neg: 13.021 acc5_neg: 100.000 time: 0.2066 data: 0.1137 max mem: 3531 +Test: [120/235] eta: 0:00:29 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.8620 (0.9868) acc1: 10.4167 (54.9651) acc5: 100.0000 (100.0000) acc1_neg: 5.469 acc5_neg: 100.000 time: 0.2050 data: 0.1121 max mem: 3531 +Test: [130/235] eta: 0:00:26 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.9975 (1.0647) acc1: 6.5104 (51.2643) acc5: 100.0000 (100.0000) acc1_neg: 6.510 acc5_neg: 100.000 time: 0.2009 data: 0.1082 max mem: 3531 +Test: [140/235] eta: 0:00:23 acc1_pos: 96.923 acc5_pos: 100.000 loss: 2.0225 (1.1334) acc1: 6.2500 (48.0404) acc5: 100.0000 (100.0000) acc1_neg: 4.427 acc5_neg: 100.000 time: 0.1986 data: 0.1061 max mem: 3531 +Test: [150/235] eta: 0:00:20 acc1_pos: 96.923 acc5_pos: 100.000 loss: 2.0263 (1.1914) acc1: 5.4688 (45.2539) acc5: 100.0000 (100.0000) acc1_neg: 5.469 acc5_neg: 100.000 time: 0.2223 data: 0.1309 max mem: 3531 +Test: [160/235] eta: 0:00:18 acc1_pos: 96.923 acc5_pos: 100.000 loss: 2.0119 (1.2426) acc1: 5.7292 (42.8491) acc5: 100.0000 (100.0000) acc1_neg: 8.594 acc5_neg: 100.000 time: 0.2311 data: 0.1355 max mem: 3531 +Test: [170/235] eta: 0:00:15 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.9585 (1.2691) acc1: 8.5938 (41.3347) acc5: 100.0000 (100.0000) acc1_neg: 16.406 acc5_neg: 100.000 time: 0.2251 data: 0.1275 max mem: 3531 +Test: [180/235] eta: 0:00:13 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.6789 (1.2914) acc1: 17.4479 (40.0567) acc5: 100.0000 (100.0000) acc1_neg: 17.448 acc5_neg: 100.000 time: 0.2262 data: 0.1325 max mem: 3531 +Test: [190/235] eta: 0:00:10 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.6534 (1.3101) acc1: 18.7500 (38.9643) acc5: 100.0000 (100.0000) acc1_neg: 20.052 acc5_neg: 100.000 time: 0.1957 data: 0.1039 max mem: 3531 +Test: [200/235] eta: 0:00:08 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.6554 (1.3281) acc1: 18.2292 (37.9042) acc5: 100.0000 (100.0000) acc1_neg: 18.229 acc5_neg: 100.000 time: 0.1736 data: 0.0819 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.6595 (1.3434) acc1: 17.7083 (36.9705) acc5: 100.0000 (100.0000) acc1_neg: 17.708 acc5_neg: 100.000 time: 0.1690 data: 0.0786 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.6526 (1.3570) acc1: 18.4896 (36.1496) acc5: 100.0000 (100.0000) acc1_neg: 17.448 acc5_neg: 100.000 time: 0.1275 data: 0.0379 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.6413 (1.3700) acc1: 19.2708 (35.3919) acc5: 100.0000 (100.0000) acc1_neg: 16.406 acc5_neg: 100.000 time: 0.1136 data: 0.0227 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.6526 (1.3749) acc1: 19.2708 (35.1628) acc5: 100.0000 (100.0000) acc1_neg: 17.391 acc5_neg: 100.000 time: 0.1104 data: 0.0227 max mem: 3532 +Test: Total time: 0:00:51 (0.2199 s / it) +* Overall Acc@1 35.163 Acc@5 100.000 + target == real Acc@1 96.259 Acc@5 100.000 + target == fake Acc@1 13.065 Acc@5 100.000 + loss 1.375 + PR-AUC 0.8467 ROC-AUC 0.6871 EER 36.63% (thr=0.1086) +Accuracy of the network on the 89879 test images: 35.2% +[rank0]:[W726 03:42:14.539046949 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:42:37.080000 3139387 torch/distributed/run.py:766] +W0726 03:42:37.080000 3139387 torch/distributed/run.py:766] ***************************************** +W0726 03:42:37.080000 3139387 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:42:37.080000 3139387 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:43:10.426970529 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:43:10.793590817 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:43:11.273777949 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:43:11.401368841 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:43:11.551367713 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:43:11.622850240 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:43:11.654330017 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:43:11.759377416 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n11_seed5_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed5_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:30:02 acc1_pos: 87.760 acc5_pos: 100.000 loss: 0.3520 (0.3520) acc1: 87.7604 (87.7604) acc5: 100.0000 (100.0000) time: 6.7009 data: 4.8987 max mem: 3531 +Test: [ 10/269] eta: 0:02:59 acc1_pos: 94.271 acc5_pos: 100.000 loss: 0.3222 (0.3259) acc1: 91.6667 (90.5540) acc5: 100.0000 (100.0000) time: 0.6914 data: 0.4458 max mem: 3531 +Test: [ 20/269] eta: 0:01:41 acc1_pos: 93.490 acc5_pos: 100.000 loss: 0.3006 (0.3110) acc1: 92.7083 (92.0387) acc5: 100.0000 (100.0000) time: 0.0949 data: 0.0005 max mem: 3531 +Test: [ 30/269] eta: 0:01:30 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.3006 (0.3063) acc1: 92.7083 (92.3639) acc5: 100.0000 (100.0000) time: 0.2035 data: 0.1075 max mem: 3531 +Test: [ 40/269] eta: 0:01:17 acc1_pos: 93.966 acc5_pos: 100.000 loss: 0.3526 (0.4884) acc1: 87.2396 (78.1758) acc5: 100.0000 (100.0000) acc1_neg: 12.240 acc5_neg: 100.000 time: 0.2599 data: 0.1666 max mem: 3531 +Test: [ 50/269] eta: 0:01:07 acc1_pos: 93.966 acc5_pos: 100.000 loss: 1.2502 (0.6635) acc1: 16.1458 (64.9510) acc5: 100.0000 (100.0000) acc1_neg: 1.823 acc5_neg: 100.000 time: 0.2011 data: 0.1056 max mem: 3531 +Test: [ 60/269] eta: 0:01:01 acc1_pos: 93.966 acc5_pos: 100.000 loss: 1.3786 (0.7883) acc1: 6.5104 (55.6395) acc5: 100.0000 (100.0000) acc1_neg: 9.635 acc5_neg: 100.000 time: 0.2099 data: 0.1129 max mem: 3531 +Test: [ 70/269] eta: 0:00:56 acc1_pos: 93.966 acc5_pos: 100.000 loss: 1.3115 (0.8464) acc1: 11.4583 (50.4842) acc5: 100.0000 (100.0000) acc1_neg: 16.667 acc5_neg: 100.000 time: 0.2149 data: 0.1172 max mem: 3531 +Test: [ 80/269] eta: 0:00:50 acc1_pos: 93.966 acc5_pos: 100.000 loss: 1.1675 (0.8812) acc1: 22.6562 (47.1226) acc5: 100.0000 (100.0000) acc1_neg: 25.521 acc5_neg: 100.000 time: 0.1849 data: 0.0870 max mem: 3531 +Test: [ 90/269] eta: 0:00:47 acc1_pos: 93.966 acc5_pos: 100.000 loss: 1.1675 (0.9214) acc1: 20.3125 (44.2022) acc5: 100.0000 (100.0000) acc1_neg: 11.198 acc5_neg: 100.000 time: 0.2104 data: 0.1147 max mem: 3531 +Test: [100/269] eta: 0:00:44 acc1_pos: 93.966 acc5_pos: 100.000 loss: 1.3174 (0.9649) acc1: 13.0208 (40.8880) acc5: 100.0000 (100.0000) acc1_neg: 5.469 acc5_neg: 100.000 time: 0.2403 data: 0.1464 max mem: 3531 +Test: [110/269] eta: 0:00:39 acc1_pos: 93.966 acc5_pos: 100.000 loss: 1.3174 (0.9832) acc1: 13.5417 (39.1470) acc5: 100.0000 (100.0000) acc1_neg: 15.104 acc5_neg: 100.000 time: 0.1644 data: 0.0702 max mem: 3531 +Test: [120/269] eta: 0:00:36 acc1_pos: 93.966 acc5_pos: 100.000 loss: 1.2611 (1.0192) acc1: 15.1042 (37.1363) acc5: 100.0000 (100.0000) acc1_neg: 11.458 acc5_neg: 100.000 time: 0.1495 data: 0.0537 max mem: 3531 +Test: [130/269] eta: 0:00:34 acc1_pos: 93.966 acc5_pos: 100.000 loss: 1.4034 (1.0452) acc1: 13.5417 (35.3809) acc5: 100.0000 (100.0000) acc1_neg: 16.667 acc5_neg: 100.000 time: 0.2499 data: 0.1545 max mem: 3531 +Test: [140/269] eta: 0:00:31 acc1_pos: 93.966 acc5_pos: 100.000 loss: 1.4699 (1.0781) acc1: 8.0729 (33.3315) acc5: 100.0000 (100.0000) acc1_neg: 4.688 acc5_neg: 100.000 time: 0.2599 data: 0.1622 max mem: 3531 +Test: [150/269] eta: 0:00:28 acc1_pos: 93.966 acc5_pos: 100.000 loss: 1.4401 (1.0904) acc1: 5.2083 (32.4452) acc5: 100.0000 (100.0000) acc1_neg: 34.375 acc5_neg: 100.000 time: 0.1599 data: 0.0594 max mem: 3531 +Test: [160/269] eta: 0:00:25 acc1_pos: 93.966 acc5_pos: 100.000 loss: 1.0838 (1.0917) acc1: 17.4479 (32.0927) acc5: 100.0000 (100.0000) acc1_neg: 16.927 acc5_neg: 100.000 time: 0.1356 data: 0.0406 max mem: 3531 +Test: [170/269] eta: 0:00:23 acc1_pos: 93.966 acc5_pos: 100.000 loss: 1.2405 (1.1077) acc1: 16.9271 (31.0490) acc5: 100.0000 (100.0000) acc1_neg: 13.542 acc5_neg: 100.000 time: 0.2138 data: 0.1207 max mem: 3531 +Test: [180/269] eta: 0:00:20 acc1_pos: 93.966 acc5_pos: 100.000 loss: 1.2903 (1.1173) acc1: 15.6250 (30.2673) acc5: 100.0000 (100.0000) acc1_neg: 15.365 acc5_neg: 100.000 time: 0.2437 data: 0.1504 max mem: 3531 +Test: [190/269] eta: 0:00:18 acc1_pos: 93.966 acc5_pos: 100.000 loss: 1.2798 (1.1264) acc1: 15.8854 (29.5062) acc5: 100.0000 (100.0000) acc1_neg: 17.969 acc5_neg: 100.000 time: 0.1803 data: 0.0881 max mem: 3531 +Test: [200/269] eta: 0:00:15 acc1_pos: 93.966 acc5_pos: 100.000 loss: 1.3962 (1.1446) acc1: 11.4583 (28.5707) acc5: 100.0000 (100.0000) acc1_neg: 13.021 acc5_neg: 100.000 time: 0.1493 data: 0.0564 max mem: 3531 +Test: [210/269] eta: 0:00:13 acc1_pos: 93.966 acc5_pos: 100.000 loss: 1.3529 (1.1463) acc1: 13.0208 (28.4385) acc5: 100.0000 (100.0000) acc1_neg: 29.948 acc5_neg: 100.000 time: 0.2044 data: 0.1128 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 93.966 acc5_pos: 100.000 loss: 1.1314 (1.1468) acc1: 24.4792 (28.2546) acc5: 100.0000 (100.0000) acc1_neg: 34.896 acc5_neg: 100.000 time: 0.2005 data: 0.1075 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 93.966 acc5_pos: 100.000 loss: 1.1326 (1.1460) acc1: 23.4375 (28.2197) acc5: 100.0000 (100.0000) acc1_neg: 13.802 acc5_neg: 100.000 time: 0.1546 data: 0.0606 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 93.966 acc5_pos: 100.000 loss: 1.2864 (1.1560) acc1: 15.1042 (27.5545) acc5: 100.0000 (100.0000) acc1_neg: 14.583 acc5_neg: 100.000 time: 0.1336 data: 0.0410 max mem: 3532 +Test: [250/269] eta: 0:00:04 acc1_pos: 93.966 acc5_pos: 100.000 loss: 1.3962 (1.1653) acc1: 12.2396 (26.9889) acc5: 100.0000 (100.0000) acc1_neg: 1.562 acc5_neg: 100.000 time: 0.1362 data: 0.0450 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 93.966 acc5_pos: 100.000 loss: 1.4661 (1.1765) acc1: 4.9479 (26.1694) acc5: 100.0000 (100.0000) acc1_neg: 13.802 acc5_neg: 100.000 time: 0.1570 data: 0.0672 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 93.966 acc5_pos: 100.000 loss: 1.3363 (1.1785) acc1: 7.0312 (25.8759) acc5: 100.0000 (100.0000) acc1_neg: 30.184 acc5_neg: 100.000 time: 0.1288 data: 0.0358 max mem: 3532 +Test: Total time: 0:00:56 (0.2083 s / it) +* Overall Acc@1 25.876 Acc@5 100.000 + target == real Acc@1 92.220 Acc@5 100.000 + target == fake Acc@1 16.822 Acc@5 100.000 + loss 1.179 + PR-AUC 0.9200 ROC-AUC 0.6298 EER 41.11% (thr=0.2368) +Accuracy of the network on the 103293 test images: 25.9% +[rank0]:[W726 03:44:18.200854424 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T03:44:24Z] eval_done tag=fixed500k_fake_n11_seed5_resampled +[2026-07-26T03:44:24Z] eval_start tag=fixed500k_real_n5_seed6_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed6_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T03:44:24Z +hostname=n116-040-168 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed6_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n5_seed6_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 03:44:39.751000 3139952 torch/distributed/run.py:766] +W0726 03:44:39.751000 3139952 torch/distributed/run.py:766] ***************************************** +W0726 03:44:39.751000 3139952 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:44:39.751000 3139952 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:45:15.360509409 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:45:15.472798897 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:45:15.819165984 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:45:16.235407950 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:45:16.285119581 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:45:16.433118700 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:45:16.529536761 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W726 03:45:16.536020540 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n5_seed6_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed6_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:23:19 acc1_pos: 75.000 acc5_pos: 100.000 loss: 0.4827 (0.4827) acc1: 75.0000 (75.0000) acc5: 100.0000 (100.0000) time: 5.9812 data: 4.4305 max mem: 3531 +Test: [ 10/234] eta: 0:02:21 acc1_pos: 60.938 acc5_pos: 100.000 loss: 0.4026 (0.4059) acc1: 90.3646 (83.5464) acc5: 100.0000 (100.0000) time: 0.6339 data: 0.4076 max mem: 3531 +Test: [ 20/234] eta: 0:01:20 acc1_pos: 26.823 acc5_pos: 100.000 loss: 0.6661 (0.6774) acc1: 60.6771 (64.1617) acc5: 100.0000 (100.0000) time: 0.0976 data: 0.0028 max mem: 3531 +Test: [ 30/234] eta: 0:01:08 acc1_pos: 16.146 acc5_pos: 100.000 loss: 1.3451 (0.8789) acc1: 26.3021 (53.0494) acc5: 100.0000 (100.0000) time: 0.1692 data: 0.0708 max mem: 3531 +Test: [ 40/234] eta: 0:00:59 acc1_pos: 40.885 acc5_pos: 100.000 loss: 1.4540 (0.9860) acc1: 20.0521 (46.2525) acc5: 100.0000 (100.0000) time: 0.2349 data: 0.1386 max mem: 3531 +Test: [ 50/234] eta: 0:00:55 acc1_pos: 36.198 acc5_pos: 100.000 loss: 0.9675 (0.9752) acc1: 40.8854 (46.0682) acc5: 100.0000 (100.0000) time: 0.2502 data: 0.1578 max mem: 3531 +Test: [ 60/234] eta: 0:00:50 acc1_pos: 40.000 acc5_pos: 100.000 loss: 0.4670 (0.8418) acc1: 51.8229 (54.5210) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2544 data: 0.1606 max mem: 3531 +Test: [ 70/234] eta: 0:00:44 acc1_pos: 40.000 acc5_pos: 100.000 loss: 0.1297 (0.7596) acc1: 99.4792 (59.8335) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.1893 data: 0.0973 max mem: 3531 +Test: [ 80/234] eta: 0:00:40 acc1_pos: 40.000 acc5_pos: 100.000 loss: 0.2994 (0.6991) acc1: 89.3229 (63.7635) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1804 data: 0.0890 max mem: 3531 +Test: [ 90/234] eta: 0:00:38 acc1_pos: 40.000 acc5_pos: 100.000 loss: 0.1673 (0.6361) acc1: 97.6562 (67.6110) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2590 data: 0.1657 max mem: 3531 +Test: [100/234] eta: 0:00:34 acc1_pos: 40.000 acc5_pos: 100.000 loss: 0.1712 (0.6006) acc1: 95.5729 (69.9129) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.2349 data: 0.1404 max mem: 3531 +Test: [110/234] eta: 0:00:30 acc1_pos: 40.000 acc5_pos: 100.000 loss: 0.2283 (0.5682) acc1: 93.2292 (72.0181) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1612 data: 0.0653 max mem: 3531 +Test: [120/234] eta: 0:00:27 acc1_pos: 40.000 acc5_pos: 100.000 loss: 0.2682 (0.5505) acc1: 91.1458 (73.1577) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1814 data: 0.0863 max mem: 3531 +Test: [130/234] eta: 0:00:25 acc1_pos: 40.000 acc5_pos: 100.000 loss: 0.3783 (0.5490) acc1: 74.7396 (73.1234) acc5: 100.0000 (100.0000) acc1_neg: 72.396 acc5_neg: 100.000 time: 0.2137 data: 0.1180 max mem: 3531 +Test: [140/234] eta: 0:00:22 acc1_pos: 40.000 acc5_pos: 100.000 loss: 0.3608 (0.5327) acc1: 85.1562 (74.1947) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.2068 data: 0.1122 max mem: 3531 +Test: [150/234] eta: 0:00:19 acc1_pos: 40.000 acc5_pos: 100.000 loss: 0.2677 (0.5146) acc1: 90.6250 (75.3760) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.1987 data: 0.1072 max mem: 3531 +Test: [160/234] eta: 0:00:17 acc1_pos: 40.000 acc5_pos: 100.000 loss: 0.3696 (0.5190) acc1: 84.1146 (75.0307) acc5: 100.0000 (100.0000) acc1_neg: 61.719 acc5_neg: 100.000 time: 0.2156 data: 0.1219 max mem: 3531 +Test: [170/234] eta: 0:00:15 acc1_pos: 40.000 acc5_pos: 100.000 loss: 0.7047 (0.5347) acc1: 60.6771 (73.9431) acc5: 100.0000 (100.0000) acc1_neg: 51.823 acc5_neg: 100.000 time: 0.2296 data: 0.1358 max mem: 3531 +Test: [180/234] eta: 0:00:12 acc1_pos: 40.000 acc5_pos: 100.000 loss: 0.5015 (0.5228) acc1: 65.1042 (74.8029) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.2016 data: 0.1095 max mem: 3531 +Test: [190/234] eta: 0:00:10 acc1_pos: 40.000 acc5_pos: 100.000 loss: 0.3580 (0.5158) acc1: 85.9375 (75.2741) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.1676 data: 0.0749 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 40.000 acc5_pos: 100.000 loss: 0.3801 (0.5015) acc1: 84.1146 (76.1803) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1415 data: 0.0506 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 40.000 acc5_pos: 100.000 loss: 0.1204 (0.4838) acc1: 99.4792 (77.2870) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1196 data: 0.0300 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 40.000 acc5_pos: 100.000 loss: 0.1417 (0.4688) acc1: 99.2188 (78.2699) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1164 data: 0.0269 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 40.000 acc5_pos: 100.000 loss: 0.1887 (0.4611) acc1: 91.9271 (78.8070) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1198 data: 0.0299 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 40.000 acc5_pos: 100.000 loss: 0.2780 (0.4589) acc1: 91.1458 (78.9377) acc5: 100.0000 (100.0000) acc1_neg: 91.440 acc5_neg: 100.000 time: 0.1192 data: 0.0284 max mem: 3532 +Test: Total time: 0:00:49 (0.2096 s / it) +* Overall Acc@1 78.938 Acc@5 100.000 + target == real Acc@1 46.024 Acc@5 100.000 + target == fake Acc@1 88.214 Acc@5 100.000 + loss 0.459 + PR-AUC 0.9160 ROC-AUC 0.7795 EER 28.33% (thr=0.7549) +Accuracy of the network on the 89729 test images: 78.9% +[rank0]:[W726 03:46:13.786949677 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:46:35.913000 3140511 torch/distributed/run.py:766] +W0726 03:46:35.913000 3140511 torch/distributed/run.py:766] ***************************************** +W0726 03:46:35.913000 3140511 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:46:35.913000 3140511 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +[rank3]:[W726 03:47:14.760939674 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:47:14.782737234 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank6]:[W726 03:47:14.795213665 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:47:14.089843134 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:47:14.095630846 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:47:15.215203642 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:47:15.317952032 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:47:15.461635352 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n5_seed6_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed6_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:22:42 acc1_pos: 55.990 acc5_pos: 100.000 loss: 0.7301 (0.7301) acc1: 55.9896 (55.9896) acc5: 100.0000 (100.0000) time: 6.6037 data: 5.1146 max mem: 3531 +Test: [ 10/2387] eta: 0:27:01 acc1_pos: 40.365 acc5_pos: 100.000 loss: 0.9474 (0.9238) acc1: 50.7812 (45.7386) acc5: 100.0000 (100.0000) time: 0.6820 data: 0.4653 max mem: 3531 +Test: [ 20/2387] eta: 0:15:48 acc1_pos: 71.875 acc5_pos: 100.000 loss: 0.8247 (0.8602) acc1: 48.6979 (50.7564) acc5: 100.0000 (100.0000) time: 0.0907 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:14:14 acc1_pos: 20.833 acc5_pos: 100.000 loss: 0.7865 (0.8901) acc1: 40.6250 (47.9503) acc5: 100.0000 (100.0000) time: 0.1870 data: 0.0947 max mem: 3531 +Test: [ 40/2387] eta: 0:13:10 acc1_pos: 17.708 acc5_pos: 100.000 loss: 1.0879 (0.9437) acc1: 28.3854 (42.9878) acc5: 100.0000 (100.0000) time: 0.2699 data: 0.1763 max mem: 3531 +Test: [ 50/2387] eta: 0:12:27 acc1_pos: 50.260 acc5_pos: 100.000 loss: 0.9625 (0.9126) acc1: 38.2812 (45.1848) acc5: 100.0000 (100.0000) time: 0.2536 data: 0.1588 max mem: 3531 +Test: [ 60/2387] eta: 0:12:01 acc1_pos: 38.542 acc5_pos: 100.000 loss: 0.9636 (0.9415) acc1: 42.1875 (43.9293) acc5: 100.0000 (100.0000) time: 0.2549 data: 0.1578 max mem: 3531 +Test: [ 70/2387] eta: 0:11:32 acc1_pos: 30.208 acc5_pos: 100.000 loss: 1.0438 (0.9648) acc1: 36.1979 (43.2108) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1483 max mem: 3531 +Test: [ 80/2387] eta: 0:11:01 acc1_pos: 40.625 acc5_pos: 100.000 loss: 0.9037 (0.9520) acc1: 43.7500 (44.1712) acc5: 100.0000 (100.0000) time: 0.2149 data: 0.1205 max mem: 3531 +Test: [ 90/2387] eta: 0:10:48 acc1_pos: 30.990 acc5_pos: 100.000 loss: 0.8789 (0.9470) acc1: 50.5208 (44.4082) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1302 max mem: 3531 +Test: [ 100/2387] eta: 0:10:36 acc1_pos: 47.396 acc5_pos: 100.000 loss: 0.8789 (0.9417) acc1: 47.3958 (44.7814) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1503 max mem: 3531 +Test: [ 110/2387] eta: 0:10:32 acc1_pos: 59.375 acc5_pos: 100.000 loss: 0.8000 (0.9245) acc1: 48.4375 (45.5330) acc5: 100.0000 (100.0000) time: 0.2549 data: 0.1616 max mem: 3531 +Test: [ 120/2387] eta: 0:10:25 acc1_pos: 63.281 acc5_pos: 100.000 loss: 0.7462 (0.9084) acc1: 55.9896 (46.6081) acc5: 100.0000 (100.0000) time: 0.2649 data: 0.1704 max mem: 3531 +Test: [ 130/2387] eta: 0:10:20 acc1_pos: 59.635 acc5_pos: 100.000 loss: 0.6913 (0.8975) acc1: 59.6354 (47.3263) acc5: 100.0000 (100.0000) time: 0.2599 data: 0.1637 max mem: 3531 +Test: [ 140/2387] eta: 0:10:12 acc1_pos: 79.948 acc5_pos: 100.000 loss: 0.6701 (0.8798) acc1: 59.6354 (48.3378) acc5: 100.0000 (100.0000) time: 0.2499 data: 0.1556 max mem: 3531 +Test: [ 150/2387] eta: 0:10:08 acc1_pos: 36.198 acc5_pos: 100.000 loss: 0.6674 (0.8715) acc1: 59.6354 (48.9773) acc5: 100.0000 (100.0000) time: 0.2549 data: 0.1600 max mem: 3531 +Test: [ 160/2387] eta: 0:09:59 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.6242 (0.8499) acc1: 64.8438 (50.4772) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2450 data: 0.1487 max mem: 3531 +Test: [ 170/2387] eta: 0:09:43 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2281 (0.8135) acc1: 93.4896 (53.0336) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1935 data: 0.0993 max mem: 3531 +Test: [ 180/2387] eta: 0:09:36 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2281 (0.7813) acc1: 94.0104 (55.2918) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1984 data: 0.1052 max mem: 3531 +Test: [ 190/2387] eta: 0:09:26 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2292 (0.7523) acc1: 94.0104 (57.3258) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2149 data: 0.1221 max mem: 3531 +Test: [ 200/2387] eta: 0:09:20 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2282 (0.7259) acc1: 94.0104 (59.1612) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2099 data: 0.1146 max mem: 3531 +Test: [ 210/2387] eta: 0:09:09 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2282 (0.7026) acc1: 94.2708 (60.8252) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1999 data: 0.1033 max mem: 3532 +Test: [ 220/2387] eta: 0:09:02 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2313 (0.6811) acc1: 94.2708 (62.3256) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1902 data: 0.0965 max mem: 3532 +Test: [ 230/2387] eta: 0:08:53 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2298 (0.6616) acc1: 94.0104 (63.6882) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1949 data: 0.1031 max mem: 3532 +Test: [ 240/2387] eta: 0:08:48 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2442 (0.6467) acc1: 91.9271 (64.7757) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.1996 data: 0.1043 max mem: 3532 +Test: [ 250/2387] eta: 0:08:44 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2175 (0.6269) acc1: 92.7083 (66.0774) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2199 data: 0.1237 max mem: 3532 +Test: [ 260/2387] eta: 0:08:34 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1239 (0.6074) acc1: 99.2188 (67.3501) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1910 data: 0.0947 max mem: 3532 +Test: [ 270/2387] eta: 0:08:30 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1158 (0.5893) acc1: 99.4792 (68.5232) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1874 data: 0.0920 max mem: 3532 +Test: [ 280/2387] eta: 0:08:23 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1180 (0.5728) acc1: 99.2188 (69.5924) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2045 data: 0.1138 max mem: 3532 +Test: [ 290/2387] eta: 0:08:16 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1177 (0.5573) acc1: 99.2188 (70.5926) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1825 data: 0.0907 max mem: 3532 +Test: [ 300/2387] eta: 0:08:14 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1060 (0.5423) acc1: 99.4792 (71.5583) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2105 data: 0.1170 max mem: 3532 +Test: [ 310/2387] eta: 0:08:10 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1060 (0.5294) acc1: 99.7396 (72.4637) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2255 data: 0.1339 max mem: 3532 +Test: [ 320/2387] eta: 0:08:03 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1886 (0.5189) acc1: 99.7396 (73.3069) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1887 data: 0.0979 max mem: 3532 +Test: [ 330/2387] eta: 0:07:59 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1897 (0.5089) acc1: 99.4792 (74.0984) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1867 data: 0.0958 max mem: 3532 +Test: [ 340/2387] eta: 0:07:53 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1869 (0.4995) acc1: 99.4792 (74.8450) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1900 data: 0.0995 max mem: 3532 +Test: [ 350/2387] eta: 0:07:47 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1773 (0.4883) acc1: 99.7396 (75.5594) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1675 data: 0.0749 max mem: 3532 +Test: [ 360/2387] eta: 0:07:41 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1265 (0.4806) acc1: 99.4792 (76.0886) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1651 data: 0.0717 max mem: 3532 +Test: [ 370/2387] eta: 0:07:39 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2195 (0.4736) acc1: 94.2708 (76.5885) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1949 data: 0.1024 max mem: 3532 +Test: [ 380/2387] eta: 0:07:34 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2220 (0.4670) acc1: 94.7917 (77.0724) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2048 data: 0.1119 max mem: 3532 +Test: [ 390/2387] eta: 0:07:29 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2238 (0.4608) acc1: 94.2708 (77.5076) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1750 data: 0.0822 max mem: 3532 +Test: [ 400/2387] eta: 0:07:23 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2237 (0.4548) acc1: 94.2708 (77.9334) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1608 data: 0.0661 max mem: 3532 +Test: [ 410/2387] eta: 0:07:20 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2162 (0.4490) acc1: 94.7917 (78.3506) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1802 data: 0.0859 max mem: 3532 +Test: [ 420/2387] eta: 0:07:17 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2168 (0.4437) acc1: 94.7917 (78.7293) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1988 data: 0.1074 max mem: 3532 +Test: [ 430/2387] eta: 0:07:12 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2338 (0.4393) acc1: 93.7500 (79.0573) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1895 data: 0.0976 max mem: 3532 +Test: [ 440/2387] eta: 0:07:08 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2386 (0.4336) acc1: 92.7083 (79.4495) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1801 data: 0.0890 max mem: 3533 +Test: [ 450/2387] eta: 0:07:04 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1184 (0.4266) acc1: 99.7396 (79.9017) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1724 data: 0.0791 max mem: 3533 +Test: [ 460/2387] eta: 0:07:01 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1174 (0.4199) acc1: 99.7396 (80.3309) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1785 data: 0.0835 max mem: 3533 +Test: [ 470/2387] eta: 0:06:57 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1171 (0.4135) acc1: 99.7396 (80.7458) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1870 data: 0.0947 max mem: 3533 +Test: [ 480/2387] eta: 0:06:53 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1165 (0.4073) acc1: 100.0000 (81.1428) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1750 data: 0.0834 max mem: 3533 +Test: [ 490/2387] eta: 0:06:49 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1157 (0.4014) acc1: 100.0000 (81.5247) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1707 data: 0.0773 max mem: 3533 +Test: [ 500/2387] eta: 0:06:45 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1169 (0.3957) acc1: 100.0000 (81.8883) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1763 data: 0.0838 max mem: 3533 +Test: [ 510/2387] eta: 0:06:42 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1181 (0.3905) acc1: 99.7396 (82.2387) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1802 data: 0.0895 max mem: 3533 +Test: [ 520/2387] eta: 0:06:38 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1536 (0.3864) acc1: 99.7396 (82.5596) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1737 data: 0.0827 max mem: 3533 +Test: [ 530/2387] eta: 0:06:35 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2188 (0.3834) acc1: 97.3958 (82.8253) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1731 data: 0.0810 max mem: 3533 +Test: [ 540/2387] eta: 0:06:31 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2204 (0.3803) acc1: 97.1354 (83.0854) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1755 data: 0.0839 max mem: 3533 +Test: [ 550/2387] eta: 0:06:28 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2203 (0.3776) acc1: 96.8750 (83.3305) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1774 data: 0.0851 max mem: 3533 +Test: [ 560/2387] eta: 0:06:25 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2247 (0.3749) acc1: 96.8750 (83.5682) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1767 data: 0.0839 max mem: 3533 +Test: [ 570/2387] eta: 0:06:22 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2244 (0.3722) acc1: 96.6146 (83.7994) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1745 data: 0.0831 max mem: 3533 +Test: [ 580/2387] eta: 0:06:18 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2203 (0.3697) acc1: 96.6146 (84.0187) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1782 data: 0.0853 max mem: 3533 +Test: [ 590/2387] eta: 0:06:15 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2227 (0.3672) acc1: 96.3542 (84.2269) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1754 data: 0.0830 max mem: 3533 +Test: [ 600/2387] eta: 0:06:12 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1753 (0.3637) acc1: 97.3958 (84.4825) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1738 data: 0.0807 max mem: 3533 +Test: [ 610/2387] eta: 0:06:09 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1532 (0.3602) acc1: 99.2188 (84.7113) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1751 data: 0.0819 max mem: 3533 +Test: [ 620/2387] eta: 0:06:06 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1474 (0.3567) acc1: 98.4375 (84.9315) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1705 data: 0.0752 max mem: 3533 +Test: [ 630/2387] eta: 0:06:02 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1437 (0.3534) acc1: 98.6979 (85.1463) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1621 data: 0.0656 max mem: 3533 +Test: [ 640/2387] eta: 0:05:59 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1453 (0.3502) acc1: 98.6979 (85.3561) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1680 data: 0.0753 max mem: 3533 +Test: [ 650/2387] eta: 0:05:57 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1451 (0.3470) acc1: 98.4375 (85.5587) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1849 data: 0.0901 max mem: 3533 +Test: [ 660/2387] eta: 0:05:54 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1449 (0.3439) acc1: 98.4375 (85.7543) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1863 data: 0.0895 max mem: 3533 +Test: [ 670/2387] eta: 0:05:51 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1454 (0.3410) acc1: 98.6979 (85.9460) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1767 data: 0.0806 max mem: 3534 +Test: [ 680/2387] eta: 0:05:49 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1533 (0.3386) acc1: 98.1771 (86.1230) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1754 data: 0.0822 max mem: 3534 +Test: [ 690/2387] eta: 0:05:46 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1506 (0.3358) acc1: 98.4375 (86.3065) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1849 data: 0.0910 max mem: 3534 +Test: [ 700/2387] eta: 0:05:44 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1386 (0.3331) acc1: 99.2188 (86.4895) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1892 data: 0.0922 max mem: 3534 +Test: [ 710/2387] eta: 0:05:41 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1386 (0.3303) acc1: 99.2188 (86.6708) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1813 data: 0.0843 max mem: 3534 +Test: [ 720/2387] eta: 0:05:39 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1363 (0.3277) acc1: 99.4792 (86.8441) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1856 data: 0.0914 max mem: 3534 +Test: [ 730/2387] eta: 0:05:37 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1379 (0.3251) acc1: 99.4792 (87.0173) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1974 data: 0.1061 max mem: 3534 +Test: [ 740/2387] eta: 0:05:34 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1398 (0.3226) acc1: 99.4792 (87.1812) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1895 data: 0.0985 max mem: 3534 +Test: [ 750/2387] eta: 0:05:32 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1420 (0.3201) acc1: 99.2188 (87.3426) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1856 data: 0.0945 max mem: 3534 +Test: [ 760/2387] eta: 0:05:29 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1354 (0.3177) acc1: 99.4792 (87.5000) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1872 data: 0.0957 max mem: 3534 +Test: [ 770/2387] eta: 0:05:27 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1350 (0.3155) acc1: 99.2188 (87.6436) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1843 data: 0.0931 max mem: 3534 +Test: [ 780/2387] eta: 0:05:25 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1414 (0.3139) acc1: 97.9167 (87.7354) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1886 data: 0.0972 max mem: 3534 +Test: [ 790/2387] eta: 0:05:22 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1761 (0.3125) acc1: 95.5729 (87.8319) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1867 data: 0.0950 max mem: 3534 +Test: [ 800/2387] eta: 0:05:21 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1779 (0.3110) acc1: 96.0938 (87.9428) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1964 data: 0.1051 max mem: 3534 +Test: [ 810/2387] eta: 0:05:18 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1752 (0.3095) acc1: 97.3958 (88.0526) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.2022 data: 0.1097 max mem: 3534 +Test: [ 820/2387] eta: 0:05:16 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1748 (0.3081) acc1: 96.8750 (88.1553) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1922 data: 0.0997 max mem: 3534 +Test: [ 830/2387] eta: 0:05:14 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2126 (0.3074) acc1: 94.5312 (88.1976) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1949 data: 0.1028 max mem: 3534 +Test: [ 840/2387] eta: 0:05:12 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2362 (0.3065) acc1: 92.1875 (88.2726) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1913 data: 0.0977 max mem: 3534 +Test: [ 850/2387] eta: 0:05:09 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1988 (0.3052) acc1: 96.0938 (88.3682) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.1735 data: 0.0801 max mem: 3534 +Test: [ 860/2387] eta: 0:05:07 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1960 (0.3038) acc1: 97.6562 (88.4673) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1843 data: 0.0926 max mem: 3534 +Test: [ 870/2387] eta: 0:05:05 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1811 (0.3024) acc1: 98.4375 (88.5734) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1944 data: 0.1017 max mem: 3534 +Test: [ 880/2387] eta: 0:05:02 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1767 (0.3012) acc1: 98.4375 (88.6640) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1807 data: 0.0877 max mem: 3534 +Test: [ 890/2387] eta: 0:05:00 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1589 (0.2996) acc1: 98.4375 (88.7732) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1746 data: 0.0834 max mem: 3534 +Test: [ 900/2387] eta: 0:04:58 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1616 (0.2990) acc1: 97.1354 (88.8324) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.2033 data: 0.1102 max mem: 3535 +Test: [ 910/2387] eta: 0:04:56 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2659 (0.2986) acc1: 93.2292 (88.8796) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.2056 data: 0.1111 max mem: 3535 +Test: [ 920/2387] eta: 0:04:52 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2667 (0.2983) acc1: 93.2292 (88.9265) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1349 data: 0.0425 max mem: 3535 +Test: [ 930/2387] eta: 0:04:50 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2743 (0.2981) acc1: 92.7083 (88.9635) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.1368 data: 0.0458 max mem: 3535 +Test: [ 940/2387] eta: 0:04:48 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2739 (0.2978) acc1: 92.4479 (89.0030) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1809 data: 0.0874 max mem: 3535 +Test: [ 950/2387] eta: 0:04:45 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2676 (0.2975) acc1: 92.9688 (89.0450) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1806 data: 0.0876 max mem: 3535 +Test: [ 960/2387] eta: 0:04:43 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2642 (0.2972) acc1: 92.9688 (89.0907) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1739 data: 0.0824 max mem: 3535 +Test: [ 970/2387] eta: 0:04:41 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2642 (0.2969) acc1: 93.2292 (89.1317) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1741 data: 0.0809 max mem: 3535 +Test: [ 980/2387] eta: 0:04:38 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2600 (0.2961) acc1: 93.7500 (89.1913) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1811 data: 0.0886 max mem: 3535 +Test: [ 990/2387] eta: 0:04:36 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2783 (0.2969) acc1: 91.1458 (89.1411) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.1747 data: 0.0836 max mem: 3535 +Test: [1000/2387] eta: 0:04:34 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.3415 (0.2977) acc1: 84.6354 (89.0906) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1790 data: 0.0875 max mem: 3535 +Test: [1010/2387] eta: 0:04:32 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.3415 (0.2984) acc1: 85.9375 (89.0530) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.1967 data: 0.1037 max mem: 3535 +Test: [1020/2387] eta: 0:04:30 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.3644 (0.2991) acc1: 85.1562 (89.0135) acc5: 100.0000 (100.0000) acc1_neg: 72.656 acc5_neg: 100.000 time: 0.1896 data: 0.0963 max mem: 3535 +Test: [1030/2387] eta: 0:04:27 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.3609 (0.3000) acc1: 84.8958 (88.9554) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1693 data: 0.0773 max mem: 3535 +Test: [1040/2387] eta: 0:04:25 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.3447 (0.3002) acc1: 82.8125 (88.9442) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.1688 data: 0.0764 max mem: 3535 +Test: [1050/2387] eta: 0:04:23 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.3381 (0.3015) acc1: 87.2396 (88.8677) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1742 data: 0.0825 max mem: 3535 +Test: [1060/2387] eta: 0:04:21 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.3490 (0.3020) acc1: 86.4583 (88.8330) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1763 data: 0.0846 max mem: 3535 +Test: [1070/2387] eta: 0:04:18 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.3490 (0.3026) acc1: 83.8542 (88.7907) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.1747 data: 0.0818 max mem: 3535 +Test: [1080/2387] eta: 0:04:16 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.3633 (0.3032) acc1: 82.8125 (88.7438) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1754 data: 0.0836 max mem: 3535 +Test: [1090/2387] eta: 0:04:14 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.3241 (0.3032) acc1: 86.1979 (88.7417) acc5: 100.0000 (100.0000) acc1_neg: 72.917 acc5_neg: 100.000 time: 0.1796 data: 0.0876 max mem: 3535 +Test: [1100/2387] eta: 0:04:12 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2854 (0.3033) acc1: 88.2812 (88.7210) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1715 data: 0.0757 max mem: 3535 +Test: [1110/2387] eta: 0:04:10 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.3215 (0.3039) acc1: 84.3750 (88.6675) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.1799 data: 0.0847 max mem: 3535 +Test: [1120/2387] eta: 0:04:07 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2795 (0.3035) acc1: 88.0208 (88.6964) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1835 data: 0.0923 max mem: 3536 +Test: [1130/2387] eta: 0:04:05 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2405 (0.3035) acc1: 92.7083 (88.6900) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.1727 data: 0.0819 max mem: 3536 +Test: [1140/2387] eta: 0:04:03 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.3315 (0.3040) acc1: 87.2396 (88.6562) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.1760 data: 0.0847 max mem: 3536 +Test: [1150/2387] eta: 0:04:01 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.3347 (0.3033) acc1: 88.5417 (88.6980) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1871 data: 0.0958 max mem: 3536 +Test: [1160/2387] eta: 0:03:59 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1272 (0.3017) acc1: 100.0000 (88.7927) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1862 data: 0.0896 max mem: 3536 +Test: [1170/2387] eta: 0:03:57 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1149 (0.3002) acc1: 100.0000 (88.8777) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1748 data: 0.0767 max mem: 3536 +Test: [1180/2387] eta: 0:03:55 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1170 (0.2988) acc1: 99.7396 (88.9591) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1786 data: 0.0827 max mem: 3536 +Test: [1190/2387] eta: 0:03:53 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1153 (0.2973) acc1: 99.7396 (89.0514) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0903 max mem: 3536 +Test: [1200/2387] eta: 0:03:51 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1096 (0.2957) acc1: 100.0000 (89.1395) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1888 data: 0.0980 max mem: 3536 +Test: [1210/2387] eta: 0:03:49 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1071 (0.2943) acc1: 100.0000 (89.2236) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1850 data: 0.0926 max mem: 3536 +Test: [1220/2387] eta: 0:03:47 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1245 (0.2931) acc1: 99.2188 (89.2943) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1804 data: 0.0877 max mem: 3536 +Test: [1230/2387] eta: 0:03:44 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1350 (0.2918) acc1: 98.4375 (89.3682) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1780 data: 0.0874 max mem: 3536 +Test: [1240/2387] eta: 0:03:42 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1390 (0.2905) acc1: 98.4375 (89.4419) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1748 data: 0.0822 max mem: 3536 +Test: [1250/2387] eta: 0:03:40 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1403 (0.2893) acc1: 98.4375 (89.5109) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1795 data: 0.0866 max mem: 3536 +Test: [1260/2387] eta: 0:03:38 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1403 (0.2881) acc1: 98.4375 (89.5823) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1850 data: 0.0923 max mem: 3536 +Test: [1270/2387] eta: 0:03:36 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1198 (0.2868) acc1: 99.2188 (89.6602) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1775 data: 0.0852 max mem: 3536 +Test: [1280/2387] eta: 0:03:34 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1293 (0.2857) acc1: 99.4792 (89.7309) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1765 data: 0.0856 max mem: 3536 +Test: [1290/2387] eta: 0:03:32 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1643 (0.2849) acc1: 97.6562 (89.7756) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1765 data: 0.0858 max mem: 3536 +Test: [1300/2387] eta: 0:03:30 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1801 (0.2842) acc1: 96.0938 (89.8211) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1745 data: 0.0842 max mem: 3536 +Test: [1310/2387] eta: 0:03:28 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1777 (0.2835) acc1: 96.6146 (89.8686) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1693 data: 0.0783 max mem: 3536 +Test: [1320/2387] eta: 0:03:26 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1982 (0.2829) acc1: 96.0938 (89.9120) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1678 data: 0.0762 max mem: 3536 +Test: [1330/2387] eta: 0:03:23 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1677 (0.2820) acc1: 97.6562 (89.9701) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1748 data: 0.0821 max mem: 3536 +Test: [1340/2387] eta: 0:03:21 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1677 (0.2823) acc1: 97.1354 (89.9585) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1692 data: 0.0765 max mem: 3536 +Test: [1350/2387] eta: 0:03:19 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.0937 (0.2808) acc1: 100.0000 (90.0328) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1692 data: 0.0774 max mem: 3537 +Test: [1360/2387] eta: 0:03:17 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.0868 (0.2794) acc1: 100.0000 (90.1061) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1748 data: 0.0831 max mem: 3537 +Test: [1370/2387] eta: 0:03:15 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.0883 (0.2781) acc1: 100.0000 (90.1782) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1750 data: 0.0837 max mem: 3537 +Test: [1380/2387] eta: 0:03:13 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.0887 (0.2767) acc1: 100.0000 (90.2492) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1810 data: 0.0900 max mem: 3537 +Test: [1390/2387] eta: 0:03:11 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.0865 (0.2753) acc1: 100.0000 (90.3193) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1819 data: 0.0898 max mem: 3537 +Test: [1400/2387] eta: 0:03:09 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.0842 (0.2740) acc1: 100.0000 (90.3884) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1738 data: 0.0786 max mem: 3537 +Test: [1410/2387] eta: 0:03:07 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.0920 (0.2737) acc1: 100.0000 (90.4033) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1771 data: 0.0815 max mem: 3537 +Test: [1420/2387] eta: 0:03:05 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2903 (0.2743) acc1: 88.8021 (90.3593) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.1800 data: 0.0871 max mem: 3537 +Test: [1430/2387] eta: 0:03:03 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.3257 (0.2747) acc1: 86.4583 (90.3397) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1799 data: 0.0881 max mem: 3537 +Test: [1440/2387] eta: 0:03:01 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.3238 (0.2751) acc1: 88.2812 (90.3106) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.1707 data: 0.0792 max mem: 3537 +Test: [1450/2387] eta: 0:02:59 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2954 (0.2750) acc1: 88.2812 (90.3136) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1700 data: 0.0786 max mem: 3537 +Test: [1460/2387] eta: 0:02:57 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1798 (0.2743) acc1: 96.0938 (90.3596) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1858 data: 0.0932 max mem: 3537 +Test: [1470/2387] eta: 0:02:55 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1502 (0.2734) acc1: 98.1771 (90.4133) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1912 data: 0.0982 max mem: 3537 +Test: [1480/2387] eta: 0:02:53 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1517 (0.2726) acc1: 98.1771 (90.4615) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1853 data: 0.0928 max mem: 3537 +Test: [1490/2387] eta: 0:02:51 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1511 (0.2717) acc1: 98.1771 (90.5155) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1772 data: 0.0842 max mem: 3537 +Test: [1500/2387] eta: 0:02:49 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1371 (0.2709) acc1: 98.9583 (90.5690) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1777 data: 0.0853 max mem: 3537 +Test: [1510/2387] eta: 0:02:47 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1281 (0.2699) acc1: 99.2188 (90.6272) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1826 data: 0.0916 max mem: 3537 +Test: [1520/2387] eta: 0:02:45 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1297 (0.2693) acc1: 98.9583 (90.6721) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1909 data: 0.1003 max mem: 3537 +Test: [1530/2387] eta: 0:02:44 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1643 (0.2685) acc1: 98.4375 (90.7235) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2041 data: 0.1126 max mem: 3537 +Test: [1540/2387] eta: 0:02:41 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1541 (0.2678) acc1: 99.2188 (90.7746) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1894 data: 0.0966 max mem: 3537 +Test: [1550/2387] eta: 0:02:40 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1541 (0.2671) acc1: 99.2188 (90.8272) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1871 data: 0.0953 max mem: 3537 +Test: [1560/2387] eta: 0:02:38 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1440 (0.2663) acc1: 99.4792 (90.8747) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1944 data: 0.1022 max mem: 3537 +Test: [1570/2387] eta: 0:02:36 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1310 (0.2654) acc1: 99.7396 (90.9302) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1846 data: 0.0911 max mem: 3537 +Test: [1580/2387] eta: 0:02:34 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1431 (0.2650) acc1: 99.2188 (90.9594) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1855 data: 0.0925 max mem: 3538 +Test: [1590/2387] eta: 0:02:32 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1876 (0.2649) acc1: 95.5729 (90.9653) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1873 data: 0.0931 max mem: 3538 +Test: [1600/2387] eta: 0:02:30 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2379 (0.2649) acc1: 93.2292 (90.9568) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1863 data: 0.0929 max mem: 3538 +Test: [1610/2387] eta: 0:02:28 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2200 (0.2646) acc1: 93.4896 (90.9816) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1814 data: 0.0900 max mem: 3538 +Test: [1620/2387] eta: 0:02:26 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2081 (0.2645) acc1: 94.7917 (90.9871) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1778 data: 0.0865 max mem: 3538 +Test: [1630/2387] eta: 0:02:24 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2106 (0.2645) acc1: 92.7083 (90.9830) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1764 data: 0.0851 max mem: 3538 +Test: [1640/2387] eta: 0:02:22 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2441 (0.2645) acc1: 91.6667 (90.9905) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.1757 data: 0.0847 max mem: 3538 +Test: [1650/2387] eta: 0:02:20 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2333 (0.2643) acc1: 93.7500 (91.0080) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1731 data: 0.0825 max mem: 3538 +Test: [1660/2387] eta: 0:02:18 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2298 (0.2643) acc1: 93.4896 (91.0107) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1808 data: 0.0902 max mem: 3538 +Test: [1670/2387] eta: 0:02:16 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2724 (0.2645) acc1: 92.1875 (91.0068) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1838 data: 0.0926 max mem: 3538 +Test: [1680/2387] eta: 0:02:14 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2960 (0.2647) acc1: 91.6667 (91.0007) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1807 data: 0.0866 max mem: 3538 +Test: [1690/2387] eta: 0:02:12 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2818 (0.2647) acc1: 91.9271 (91.0060) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1725 data: 0.0787 max mem: 3538 +Test: [1700/2387] eta: 0:02:10 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2600 (0.2645) acc1: 92.7083 (91.0186) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1623 data: 0.0711 max mem: 3538 +Test: [1710/2387] eta: 0:02:08 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1564 (0.2638) acc1: 97.3958 (91.0614) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1645 data: 0.0730 max mem: 3538 +Test: [1720/2387] eta: 0:02:06 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1318 (0.2630) acc1: 98.6979 (91.1039) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1703 data: 0.0791 max mem: 3538 +Test: [1730/2387] eta: 0:02:04 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1268 (0.2623) acc1: 98.6979 (91.1433) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1740 data: 0.0830 max mem: 3538 +Test: [1740/2387] eta: 0:02:02 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1310 (0.2615) acc1: 97.6562 (91.1817) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1700 data: 0.0791 max mem: 3538 +Test: [1750/2387] eta: 0:02:00 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1278 (0.2608) acc1: 98.6979 (91.2244) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1688 data: 0.0775 max mem: 3538 +Test: [1760/2387] eta: 0:01:58 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1245 (0.2601) acc1: 99.4792 (91.2680) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1675 data: 0.0760 max mem: 3538 +Test: [1770/2387] eta: 0:01:56 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1294 (0.2595) acc1: 98.9583 (91.3032) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1713 data: 0.0802 max mem: 3538 +Test: [1780/2387] eta: 0:01:54 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1439 (0.2588) acc1: 98.6979 (91.3438) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1784 data: 0.0871 max mem: 3538 +Test: [1790/2387] eta: 0:01:53 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1353 (0.2581) acc1: 98.6979 (91.3812) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1756 data: 0.0829 max mem: 3538 +Test: [1800/2387] eta: 0:01:51 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1326 (0.2575) acc1: 98.6979 (91.4175) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1735 data: 0.0805 max mem: 3538 +Test: [1810/2387] eta: 0:01:49 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1295 (0.2567) acc1: 98.9583 (91.4596) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1757 data: 0.0840 max mem: 3539 +Test: [1820/2387] eta: 0:01:47 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1191 (0.2560) acc1: 99.2188 (91.5009) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1797 data: 0.0878 max mem: 3539 +Test: [1830/2387] eta: 0:01:45 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1218 (0.2554) acc1: 99.7396 (91.5436) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1786 data: 0.0869 max mem: 3539 +Test: [1840/2387] eta: 0:01:43 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1664 (0.2549) acc1: 99.7396 (91.5876) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1850 data: 0.0945 max mem: 3539 +Test: [1850/2387] eta: 0:01:41 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1682 (0.2544) acc1: 99.4792 (91.6312) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1887 data: 0.0985 max mem: 3539 +Test: [1860/2387] eta: 0:01:39 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1678 (0.2540) acc1: 99.4792 (91.6739) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1756 data: 0.0850 max mem: 3539 +Test: [1870/2387] eta: 0:01:37 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1678 (0.2535) acc1: 99.4792 (91.7162) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1704 data: 0.0788 max mem: 3539 +Test: [1880/2387] eta: 0:01:35 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1681 (0.2531) acc1: 99.4792 (91.7579) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1743 data: 0.0828 max mem: 3539 +Test: [1890/2387] eta: 0:01:33 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1680 (0.2526) acc1: 99.7396 (91.7993) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1746 data: 0.0837 max mem: 3539 +Test: [1900/2387] eta: 0:01:31 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1707 (0.2522) acc1: 99.7396 (91.8405) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1747 data: 0.0840 max mem: 3539 +Test: [1910/2387] eta: 0:01:29 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1900 (0.2520) acc1: 99.2188 (91.8745) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1753 data: 0.0842 max mem: 3539 +Test: [1920/2387] eta: 0:01:28 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2128 (0.2527) acc1: 96.6146 (91.8388) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.1761 data: 0.0846 max mem: 3539 +Test: [1930/2387] eta: 0:01:26 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2433 (0.2525) acc1: 93.7500 (91.8599) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1764 data: 0.0838 max mem: 3539 +Test: [1940/2387] eta: 0:01:24 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2788 (0.2533) acc1: 92.7083 (91.8206) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.1734 data: 0.0802 max mem: 3539 +Test: [1950/2387] eta: 0:01:22 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.3017 (0.2535) acc1: 88.8021 (91.8186) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1828 data: 0.0869 max mem: 3539 +Test: [1960/2387] eta: 0:01:20 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2901 (0.2542) acc1: 88.5417 (91.7664) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1803 data: 0.0854 max mem: 3539 +Test: [1970/2387] eta: 0:01:18 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.3662 (0.2550) acc1: 83.0729 (91.6985) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.1758 data: 0.0830 max mem: 3539 +Test: [1980/2387] eta: 0:01:16 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.3195 (0.2549) acc1: 85.4167 (91.7002) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1798 data: 0.0872 max mem: 3539 +Test: [1990/2387] eta: 0:01:14 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1440 (0.2544) acc1: 98.6979 (91.7374) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1795 data: 0.0886 max mem: 3539 +Test: [2000/2387] eta: 0:01:12 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1432 (0.2538) acc1: 99.2188 (91.7743) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1821 data: 0.0906 max mem: 3539 +Test: [2010/2387] eta: 0:01:10 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1438 (0.2533) acc1: 98.9583 (91.8095) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1839 data: 0.0921 max mem: 3539 +Test: [2020/2387] eta: 0:01:09 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1473 (0.2527) acc1: 98.9583 (91.8454) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1824 data: 0.0902 max mem: 3539 +Test: [2030/2387] eta: 0:01:07 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1432 (0.2522) acc1: 99.2188 (91.8813) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1750 data: 0.0827 max mem: 3540 +Test: [2040/2387] eta: 0:01:05 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1416 (0.2517) acc1: 99.2188 (91.9162) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1749 data: 0.0828 max mem: 3540 +Test: [2050/2387] eta: 0:01:03 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1456 (0.2512) acc1: 99.2188 (91.9513) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1751 data: 0.0831 max mem: 3540 +Test: [2060/2387] eta: 0:01:01 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1581 (0.2508) acc1: 98.9583 (91.9846) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1713 data: 0.0792 max mem: 3540 +Test: [2070/2387] eta: 0:00:59 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1614 (0.2503) acc1: 98.6979 (92.0189) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1767 data: 0.0858 max mem: 3540 +Test: [2080/2387] eta: 0:00:57 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1608 (0.2499) acc1: 98.9583 (92.0515) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1985 data: 0.1056 max mem: 3540 +Test: [2090/2387] eta: 0:00:55 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1567 (0.2494) acc1: 98.9583 (92.0850) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1825 data: 0.0867 max mem: 3540 +Test: [2100/2387] eta: 0:00:53 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1596 (0.2490) acc1: 98.9583 (92.1177) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1581 data: 0.0649 max mem: 3540 +Test: [2110/2387] eta: 0:00:51 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1617 (0.2486) acc1: 98.9583 (92.1494) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1753 data: 0.0846 max mem: 3540 +Test: [2120/2387] eta: 0:00:50 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1676 (0.2487) acc1: 98.1771 (92.1519) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.2018 data: 0.1102 max mem: 3540 +Test: [2130/2387] eta: 0:00:48 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2696 (0.2490) acc1: 90.8854 (92.1337) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1895 data: 0.0956 max mem: 3540 +Test: [2140/2387] eta: 0:00:46 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.3121 (0.2497) acc1: 86.1979 (92.0904) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1816 data: 0.0878 max mem: 3540 +Test: [2150/2387] eta: 0:00:44 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.3105 (0.2500) acc1: 87.2396 (92.0812) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.2149 data: 0.1214 max mem: 3540 +Test: [2160/2387] eta: 0:00:42 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2664 (0.2501) acc1: 92.4479 (92.0737) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.2014 data: 0.1040 max mem: 3540 +Test: [2170/2387] eta: 0:00:40 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.3002 (0.2507) acc1: 88.2812 (92.0324) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1799 data: 0.0813 max mem: 3540 +Test: [2180/2387] eta: 0:00:38 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.3288 (0.2507) acc1: 86.7188 (92.0301) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1817 data: 0.0855 max mem: 3540 +Test: [2190/2387] eta: 0:00:36 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1381 (0.2503) acc1: 99.4792 (92.0598) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1699 data: 0.0745 max mem: 3540 +Test: [2200/2387] eta: 0:00:35 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1552 (0.2501) acc1: 96.8750 (92.0716) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1708 data: 0.0777 max mem: 3540 +Test: [2210/2387] eta: 0:00:33 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1790 (0.2498) acc1: 96.3542 (92.0934) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1801 data: 0.0886 max mem: 3540 +Test: [2220/2387] eta: 0:00:31 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1712 (0.2496) acc1: 97.6562 (92.1152) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1751 data: 0.0831 max mem: 3540 +Test: [2230/2387] eta: 0:00:29 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1810 (0.2492) acc1: 97.6562 (92.1391) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1764 data: 0.0830 max mem: 3540 +Test: [2240/2387] eta: 0:00:27 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1801 (0.2490) acc1: 97.6562 (92.1597) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1891 data: 0.0959 max mem: 3540 +Test: [2250/2387] eta: 0:00:25 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1565 (0.2485) acc1: 99.2188 (92.1905) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1859 data: 0.0946 max mem: 3540 +Test: [2260/2387] eta: 0:00:23 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1565 (0.2486) acc1: 98.6979 (92.1900) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.1820 data: 0.0903 max mem: 3541 +Test: [2270/2387] eta: 0:00:21 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2539 (0.2488) acc1: 90.6250 (92.1725) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1777 data: 0.0864 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.3096 (0.2492) acc1: 87.7604 (92.1451) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.1756 data: 0.0851 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2775 (0.2494) acc1: 91.4062 (92.1371) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1808 data: 0.0892 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2596 (0.2496) acc1: 92.7083 (92.1303) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1833 data: 0.0918 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2696 (0.2499) acc1: 89.8438 (92.1134) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1835 data: 0.0901 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2839 (0.2502) acc1: 89.5833 (92.0988) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1733 data: 0.0803 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.3402 (0.2511) acc1: 86.1979 (92.0422) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1783 data: 0.0872 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.3676 (0.2515) acc1: 82.8125 (92.0101) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.1999 data: 0.1061 max mem: 3541 +Test: [2350/2387] eta: 0:00:06 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.4280 (0.2527) acc1: 79.6875 (91.9418) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.2065 data: 0.1103 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.4245 (0.2531) acc1: 80.9896 (91.9152) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.2049 data: 0.1107 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.3483 (0.2540) acc1: 84.8958 (91.8516) acc5: 100.0000 (100.0000) acc1_neg: 60.417 acc5_neg: 100.000 time: 0.1771 data: 0.0859 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.3711 (0.2547) acc1: 83.0729 (91.8002) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1515 data: 0.0610 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.3711 (0.2547) acc1: 83.0729 (91.7961) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1343 data: 0.0469 max mem: 3541 +Test: Total time: 0:07:26 (0.1871 s / it) +* Overall Acc@1 91.796 Acc@5 100.000 + target == real Acc@1 49.156 Acc@5 100.000 + target == fake Acc@1 94.790 Acc@5 100.000 + loss 0.255 + PR-AUC 0.9894 ROC-AUC 0.8807 EER 20.05% (thr=0.7764) +Accuracy of the network on the 916267 test images: 91.8% +[rank0]:[W726 03:55:02.280939736 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:55:23.984000 3141141 torch/distributed/run.py:766] +W0726 03:55:23.984000 3141141 torch/distributed/run.py:766] ***************************************** +W0726 03:55:23.984000 3141141 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:55:23.984000 3141141 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:55:57.423878693 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:55:57.934794366 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:55:57.069102809 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:55:57.103841272 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W726 03:55:57.107776422 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:55:57.153079962 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:55:58.188558556 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W726 03:55:58.193830816 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n5_seed6_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed6_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:01:03 acc1_pos: 55.990 acc5_pos: 100.000 loss: 0.7301 (0.7301) acc1: 55.9896 (55.9896) acc5: 100.0000 (100.0000) time: 7.0853 data: 5.0902 max mem: 3531 +Test: [ 10/517] eta: 0:06:07 acc1_pos: 40.365 acc5_pos: 100.000 loss: 0.9474 (0.9238) acc1: 50.7812 (45.7386) acc5: 100.0000 (100.0000) time: 0.7258 data: 0.4631 max mem: 3531 +Test: [ 20/517] eta: 0:03:31 acc1_pos: 71.875 acc5_pos: 100.000 loss: 0.8247 (0.8602) acc1: 48.6979 (50.7564) acc5: 100.0000 (100.0000) time: 0.0927 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:02:44 acc1_pos: 20.833 acc5_pos: 100.000 loss: 0.7865 (0.8901) acc1: 40.6250 (47.9503) acc5: 100.0000 (100.0000) time: 0.1254 data: 0.0329 max mem: 3531 +Test: [ 40/517] eta: 0:02:30 acc1_pos: 17.708 acc5_pos: 100.000 loss: 1.0879 (0.9437) acc1: 28.3854 (42.9878) acc5: 100.0000 (100.0000) time: 0.2005 data: 0.1102 max mem: 3531 +Test: [ 50/517] eta: 0:02:20 acc1_pos: 78.440 acc5_pos: 100.000 loss: 0.9224 (0.9041) acc1: 38.2812 (45.5372) acc5: 100.0000 (100.0000) acc1_neg: 56.771 acc5_neg: 100.000 time: 0.2427 data: 0.1517 max mem: 3531 +Test: [ 60/517] eta: 0:02:15 acc1_pos: 78.440 acc5_pos: 100.000 loss: 0.6662 (0.8550) acc1: 61.1979 (49.1462) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.2594 data: 0.1664 max mem: 3531 +Test: [ 70/517] eta: 0:02:08 acc1_pos: 78.440 acc5_pos: 100.000 loss: 0.6862 (0.8337) acc1: 59.6354 (51.0527) acc5: 100.0000 (100.0000) acc1_neg: 66.927 acc5_neg: 100.000 time: 0.2545 data: 0.1581 max mem: 3531 +Test: [ 80/517] eta: 0:02:03 acc1_pos: 78.440 acc5_pos: 100.000 loss: 0.7889 (0.8439) acc1: 55.2083 (51.0995) acc5: 100.0000 (100.0000) acc1_neg: 46.094 acc5_neg: 100.000 time: 0.2349 data: 0.1378 max mem: 3531 +Test: [ 90/517] eta: 0:01:59 acc1_pos: 78.440 acc5_pos: 100.000 loss: 0.5503 (0.7953) acc1: 61.1979 (54.9622) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.2499 data: 0.1564 max mem: 3531 +Test: [100/517] eta: 0:01:55 acc1_pos: 78.440 acc5_pos: 100.000 loss: 0.4231 (0.7649) acc1: 82.5521 (57.1112) acc5: 100.0000 (100.0000) acc1_neg: 70.833 acc5_neg: 100.000 time: 0.2549 data: 0.1640 max mem: 3531 +Test: [110/517] eta: 0:01:51 acc1_pos: 78.440 acc5_pos: 100.000 loss: 0.5085 (0.7479) acc1: 73.1771 (58.2770) acc5: 100.0000 (100.0000) acc1_neg: 72.656 acc5_neg: 100.000 time: 0.2499 data: 0.1579 max mem: 3531 +Test: [120/517] eta: 0:01:48 acc1_pos: 78.440 acc5_pos: 100.000 loss: 0.4705 (0.7259) acc1: 74.4792 (59.8420) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.2559 data: 0.1644 max mem: 3531 +Test: [130/517] eta: 0:01:45 acc1_pos: 78.440 acc5_pos: 100.000 loss: 0.4591 (0.7093) acc1: 77.3438 (60.9375) acc5: 100.0000 (100.0000) acc1_neg: 65.365 acc5_neg: 100.000 time: 0.2617 data: 0.1673 max mem: 3531 +Test: [140/517] eta: 0:01:42 acc1_pos: 78.440 acc5_pos: 100.000 loss: 0.5449 (0.7015) acc1: 70.8333 (61.3291) acc5: 100.0000 (100.0000) acc1_neg: 62.240 acc5_neg: 100.000 time: 0.2602 data: 0.1658 max mem: 3531 +Test: [150/517] eta: 0:01:38 acc1_pos: 78.440 acc5_pos: 100.000 loss: 0.6047 (0.6937) acc1: 66.1458 (62.0344) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.2431 data: 0.1513 max mem: 3531 +Test: [160/517] eta: 0:01:35 acc1_pos: 78.440 acc5_pos: 100.000 loss: 0.6460 (0.6939) acc1: 66.1458 (62.1765) acc5: 100.0000 (100.0000) acc1_neg: 66.667 acc5_neg: 100.000 time: 0.2436 data: 0.1486 max mem: 3531 +Test: [170/517] eta: 0:01:32 acc1_pos: 78.440 acc5_pos: 100.000 loss: 0.6887 (0.6911) acc1: 65.3646 (62.5137) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.2550 data: 0.1605 max mem: 3531 +Test: [180/517] eta: 0:01:29 acc1_pos: 78.440 acc5_pos: 100.000 loss: 0.5105 (0.6746) acc1: 75.5208 (63.7042) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.2599 data: 0.1653 max mem: 3531 +Test: [190/517] eta: 0:01:27 acc1_pos: 78.440 acc5_pos: 100.000 loss: 0.3291 (0.6559) acc1: 86.1979 (65.0619) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.2749 data: 0.1787 max mem: 3531 +Test: [200/517] eta: 0:01:24 acc1_pos: 78.440 acc5_pos: 100.000 loss: 0.2887 (0.6388) acc1: 92.7083 (66.4024) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2549 data: 0.1610 max mem: 3531 +Test: [210/517] eta: 0:01:21 acc1_pos: 78.440 acc5_pos: 100.000 loss: 0.4049 (0.6436) acc1: 65.8854 (66.0471) acc5: 100.0000 (100.0000) acc1_neg: 60.156 acc5_neg: 100.000 time: 0.2549 data: 0.1606 max mem: 3532 +Test: [220/517] eta: 0:01:19 acc1_pos: 78.440 acc5_pos: 100.000 loss: 0.7575 (0.6518) acc1: 56.5104 (65.5225) acc5: 100.0000 (100.0000) acc1_neg: 47.396 acc5_neg: 100.000 time: 0.2749 data: 0.1756 max mem: 3532 +Test: [230/517] eta: 0:01:16 acc1_pos: 78.440 acc5_pos: 100.000 loss: 0.7575 (0.6572) acc1: 52.8646 (65.0106) acc5: 100.0000 (100.0000) acc1_neg: 32.031 acc5_neg: 100.000 time: 0.2599 data: 0.1610 max mem: 3532 +Test: [240/517] eta: 0:01:14 acc1_pos: 78.440 acc5_pos: 100.000 loss: 0.8484 (0.6788) acc1: 33.5938 (63.4477) acc5: 100.0000 (100.0000) acc1_neg: 9.375 acc5_neg: 100.000 time: 0.2999 data: 0.2027 max mem: 3532 +Test: [250/517] eta: 0:01:11 acc1_pos: 78.440 acc5_pos: 100.000 loss: 1.0487 (0.6805) acc1: 32.2917 (63.3943) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.2849 data: 0.1877 max mem: 3532 +Test: [260/517] eta: 0:01:08 acc1_pos: 78.440 acc5_pos: 100.000 loss: 0.5569 (0.6781) acc1: 67.1875 (63.4618) acc5: 100.0000 (100.0000) acc1_neg: 65.625 acc5_neg: 100.000 time: 0.2348 data: 0.1394 max mem: 3532 +Test: [270/517] eta: 0:01:05 acc1_pos: 78.440 acc5_pos: 100.000 loss: 0.6655 (0.6803) acc1: 60.1562 (63.2005) acc5: 100.0000 (100.0000) acc1_neg: 52.604 acc5_neg: 100.000 time: 0.2499 data: 0.1552 max mem: 3532 +Test: [280/517] eta: 0:01:02 acc1_pos: 78.440 acc5_pos: 100.000 loss: 0.7225 (0.6842) acc1: 58.0729 (62.9495) acc5: 100.0000 (100.0000) acc1_neg: 43.229 acc5_neg: 100.000 time: 0.2449 data: 0.1484 max mem: 3532 +Test: [290/517] eta: 0:00:59 acc1_pos: 78.440 acc5_pos: 100.000 loss: 0.7357 (0.6883) acc1: 55.9896 (62.5439) acc5: 100.0000 (100.0000) acc1_neg: 45.833 acc5_neg: 100.000 time: 0.2349 data: 0.1374 max mem: 3532 +Test: [300/517] eta: 0:00:57 acc1_pos: 78.440 acc5_pos: 100.000 loss: 0.7391 (0.6913) acc1: 55.9896 (62.3711) acc5: 100.0000 (100.0000) acc1_neg: 70.312 acc5_neg: 100.000 time: 0.2299 data: 0.1341 max mem: 3532 +Test: [310/517] eta: 0:00:54 acc1_pos: 78.440 acc5_pos: 100.000 loss: 0.6517 (0.6873) acc1: 62.5000 (62.6675) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.2299 data: 0.1353 max mem: 3532 +Test: [320/517] eta: 0:00:51 acc1_pos: 78.440 acc5_pos: 100.000 loss: 0.5211 (0.6824) acc1: 73.6979 (63.0849) acc5: 100.0000 (100.0000) acc1_neg: 73.698 acc5_neg: 100.000 time: 0.2349 data: 0.1416 max mem: 3532 +Test: [330/517] eta: 0:00:48 acc1_pos: 78.440 acc5_pos: 100.000 loss: 0.4766 (0.6767) acc1: 74.4792 (63.5739) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.2449 data: 0.1475 max mem: 3532 +Test: [340/517] eta: 0:00:46 acc1_pos: 78.440 acc5_pos: 100.000 loss: 0.4515 (0.6698) acc1: 82.5521 (64.1373) acc5: 100.0000 (100.0000) acc1_neg: 73.958 acc5_neg: 100.000 time: 0.2591 data: 0.1619 max mem: 3532 +Test: [350/517] eta: 0:00:43 acc1_pos: 78.440 acc5_pos: 100.000 loss: 0.4800 (0.6669) acc1: 76.8229 (64.2717) acc5: 100.0000 (100.0000) acc1_neg: 48.958 acc5_neg: 100.000 time: 0.2697 data: 0.1774 max mem: 3532 +Test: [360/517] eta: 0:00:40 acc1_pos: 78.440 acc5_pos: 100.000 loss: 0.5449 (0.6632) acc1: 68.2292 (64.5206) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2557 data: 0.1638 max mem: 3532 +Test: [370/517] eta: 0:00:38 acc1_pos: 78.440 acc5_pos: 100.000 loss: 0.3241 (0.6537) acc1: 93.2292 (65.3218) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.2595 data: 0.1682 max mem: 3532 +Test: [380/517] eta: 0:00:35 acc1_pos: 78.440 acc5_pos: 100.000 loss: 0.3012 (0.6448) acc1: 93.4896 (66.0317) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.2593 data: 0.1646 max mem: 3532 +Test: [390/517] eta: 0:00:33 acc1_pos: 78.440 acc5_pos: 100.000 loss: 0.3547 (0.6383) acc1: 84.1146 (66.4555) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.2468 data: 0.1515 max mem: 3532 +Test: [400/517] eta: 0:00:30 acc1_pos: 78.440 acc5_pos: 100.000 loss: 0.3869 (0.6321) acc1: 83.8542 (66.9186) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.2570 data: 0.1642 max mem: 3532 +Test: [410/517] eta: 0:00:27 acc1_pos: 78.440 acc5_pos: 100.000 loss: 0.3338 (0.6242) acc1: 90.8854 (67.5442) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2499 data: 0.1564 max mem: 3532 +Test: [420/517] eta: 0:00:25 acc1_pos: 78.440 acc5_pos: 100.000 loss: 0.3410 (0.6215) acc1: 90.8854 (67.7325) acc5: 100.0000 (100.0000) acc1_neg: 51.823 acc5_neg: 100.000 time: 0.2374 data: 0.1415 max mem: 3532 +Test: [430/517] eta: 0:00:22 acc1_pos: 78.440 acc5_pos: 100.000 loss: 0.7409 (0.6262) acc1: 50.5208 (67.2105) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.2435 data: 0.1463 max mem: 3532 +Test: [440/517] eta: 0:00:19 acc1_pos: 78.440 acc5_pos: 100.000 loss: 0.6047 (0.6228) acc1: 64.8438 (67.4863) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.2603 data: 0.1663 max mem: 3533 +Test: [450/517] eta: 0:00:17 acc1_pos: 78.440 acc5_pos: 100.000 loss: 0.4560 (0.6193) acc1: 80.7292 (67.7973) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.2561 data: 0.1622 max mem: 3533 +Test: [460/517] eta: 0:00:14 acc1_pos: 78.440 acc5_pos: 100.000 loss: 0.4560 (0.6161) acc1: 80.4688 (68.0354) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.2611 data: 0.1656 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 78.440 acc5_pos: 100.000 loss: 0.3272 (0.6097) acc1: 82.0312 (68.5963) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2678 data: 0.1739 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 78.440 acc5_pos: 100.000 loss: 0.3243 (0.6050) acc1: 94.7917 (69.0315) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2554 data: 0.1600 max mem: 3533 +Test: [490/517] eta: 0:00:07 acc1_pos: 78.440 acc5_pos: 100.000 loss: 0.4824 (0.6057) acc1: 77.6042 (68.9218) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.2649 data: 0.1693 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 78.440 acc5_pos: 100.000 loss: 0.6467 (0.6057) acc1: 62.2396 (68.8841) acc5: 100.0000 (100.0000) acc1_neg: 54.688 acc5_neg: 100.000 time: 0.2238 data: 0.1312 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 78.440 acc5_pos: 100.000 loss: 0.5369 (0.6041) acc1: 63.8021 (68.9681) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.1785 data: 0.0872 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 78.440 acc5_pos: 100.000 loss: 0.6126 (0.6060) acc1: 62.2396 (68.8203) acc5: 100.0000 (100.0000) acc1_neg: 50.530 acc5_neg: 100.000 time: 0.1787 data: 0.0867 max mem: 3533 +Test: Total time: 0:02:12 (0.2564 s / it) +* Overall Acc@1 68.820 Acc@5 100.000 + target == real Acc@1 45.024 Acc@5 100.000 + target == fake Acc@1 71.289 Acc@5 100.000 + loss 0.606 + PR-AUC 0.9276 ROC-AUC 0.6086 EER 42.52% (thr=0.6230) +Accuracy of the network on the 198427 test images: 68.8% +[rank0]:[W726 03:58:21.606061282 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 03:58:44.284000 3141701 torch/distributed/run.py:766] +W0726 03:58:44.284000 3141701 torch/distributed/run.py:766] ***************************************** +W0726 03:58:44.284000 3141701 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:58:44.284000 3141701 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 03:59:18.284133493 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:59:18.648542339 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:59:18.876947685 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:59:18.016672351 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:59:18.054109382 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:59:18.165034298 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:59:19.196361898 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank1]:[W726 03:59:19.196572629 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n5_seed6_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed6_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:33:49 acc1_pos: 22.917 acc5_pos: 100.000 loss: 1.5326 (1.5326) acc1: 22.9167 (22.9167) acc5: 100.0000 (100.0000) time: 5.2166 data: 3.1871 max mem: 3531 +Test: [ 10/389] eta: 0:03:30 acc1_pos: 90.885 acc5_pos: 100.000 loss: 1.1199 (1.1334) acc1: 24.2188 (36.6477) acc5: 100.0000 (100.0000) time: 0.5559 data: 0.2900 max mem: 3531 +Test: [ 20/389] eta: 0:02:10 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.8712 (1.2301) acc1: 34.6354 (38.0456) acc5: 100.0000 (100.0000) time: 0.1105 data: 0.0185 max mem: 3531 +Test: [ 30/389] eta: 0:01:52 acc1_pos: 19.792 acc5_pos: 100.000 loss: 2.2394 (1.5320) acc1: 0.2604 (26.8565) acc5: 100.0000 (100.0000) time: 0.1805 data: 0.0863 max mem: 3531 +Test: [ 40/389] eta: 0:01:47 acc1_pos: 38.542 acc5_pos: 100.000 loss: 1.7020 (1.4378) acc1: 2.6042 (27.5152) acc5: 100.0000 (100.0000) time: 0.2610 data: 0.1691 max mem: 3531 +Test: [ 50/389] eta: 0:01:35 acc1_pos: 33.594 acc5_pos: 100.000 loss: 1.1040 (1.3763) acc1: 19.2708 (26.9301) acc5: 100.0000 (100.0000) time: 0.2349 data: 0.1446 max mem: 3531 +Test: [ 60/389] eta: 0:01:29 acc1_pos: 3.125 acc5_pos: 100.000 loss: 0.9816 (1.4089) acc1: 19.2708 (25.1409) acc5: 100.0000 (100.0000) time: 0.1939 data: 0.1036 max mem: 3531 +Test: [ 70/389] eta: 0:01:24 acc1_pos: 30.469 acc5_pos: 100.000 loss: 1.0936 (1.3893) acc1: 3.1250 (24.7102) acc5: 100.0000 (100.0000) time: 0.2236 data: 0.1328 max mem: 3531 +Test: [ 80/389] eta: 0:01:20 acc1_pos: 63.281 acc5_pos: 100.000 loss: 1.0939 (1.3643) acc1: 19.7917 (25.2122) acc5: 100.0000 (100.0000) time: 0.2274 data: 0.1350 max mem: 3531 +Test: [ 90/389] eta: 0:01:14 acc1_pos: 61.719 acc5_pos: 100.000 loss: 0.8812 (1.2971) acc1: 48.9583 (29.2325) acc5: 100.0000 (100.0000) time: 0.1863 data: 0.0934 max mem: 3531 +Test: [100/389] eta: 0:01:11 acc1_pos: 1.042 acc5_pos: 100.000 loss: 1.0237 (1.3056) acc1: 39.0625 (28.4782) acc5: 100.0000 (100.0000) time: 0.1974 data: 0.1041 max mem: 3531 +Test: [110/389] eta: 0:01:07 acc1_pos: 77.865 acc5_pos: 100.000 loss: 1.4102 (1.3086) acc1: 15.3646 (28.6271) acc5: 100.0000 (100.0000) time: 0.2199 data: 0.1245 max mem: 3531 +Test: [120/389] eta: 0:01:03 acc1_pos: 66.406 acc5_pos: 100.000 loss: 1.2072 (1.2878) acc1: 35.6771 (30.4020) acc5: 100.0000 (100.0000) time: 0.1703 data: 0.0756 max mem: 3531 +Test: [130/389] eta: 0:00:59 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.5164 (1.2320) acc1: 66.4062 (33.9337) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1567 data: 0.0647 max mem: 3531 +Test: [140/389] eta: 0:00:56 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1449 (1.1690) acc1: 98.9583 (37.3670) acc5: 100.0000 (100.0000) acc1_neg: 19.010 acc5_neg: 100.000 time: 0.1848 data: 0.0938 max mem: 3531 +Test: [150/389] eta: 0:00:54 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.2417 (1.1195) acc1: 96.6146 (40.4232) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1985 data: 0.1076 max mem: 3531 +Test: [160/389] eta: 0:00:50 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.2532 (1.0771) acc1: 95.3125 (43.0723) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1726 data: 0.0811 max mem: 3531 +Test: [170/389] eta: 0:00:47 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1101 (1.0219) acc1: 100.0000 (46.3770) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1666 data: 0.0737 max mem: 3531 +Test: [180/389] eta: 0:00:45 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1031 (0.9773) acc1: 100.0000 (49.0950) acc5: 100.0000 (100.0000) acc1_neg: 66.146 acc5_neg: 100.000 time: 0.1793 data: 0.0860 max mem: 3531 +Test: [190/389] eta: 0:00:42 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1827 (0.9477) acc1: 98.6979 (50.9108) acc5: 100.0000 (100.0000) acc1_neg: 50.521 acc5_neg: 100.000 time: 0.1713 data: 0.0798 max mem: 3531 +Test: [200/389] eta: 0:00:40 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1866 (0.9120) acc1: 99.2188 (53.0628) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1716 data: 0.0815 max mem: 3531 +Test: [210/389] eta: 0:00:37 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1634 (0.8775) acc1: 99.4792 (55.1676) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1774 data: 0.0870 max mem: 3532 +Test: [220/389] eta: 0:00:35 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.0958 (0.8437) acc1: 100.0000 (57.0807) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1894 data: 0.0974 max mem: 3532 +Test: [230/389] eta: 0:00:33 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1093 (0.8219) acc1: 100.0000 (58.2274) acc5: 100.0000 (100.0000) acc1_neg: 40.885 acc5_neg: 100.000 time: 0.1951 data: 0.1033 max mem: 3532 +Test: [240/389] eta: 0:00:30 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.2303 (0.8007) acc1: 97.6562 (59.4993) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1868 data: 0.0957 max mem: 3532 +Test: [250/389] eta: 0:00:28 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.2077 (0.7756) acc1: 98.6979 (61.0682) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1816 data: 0.0906 max mem: 3532 +Test: [260/389] eta: 0:00:26 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1052 (0.7572) acc1: 99.7396 (61.9542) acc5: 100.0000 (100.0000) acc1_neg: 26.302 acc5_neg: 100.000 time: 0.1929 data: 0.1020 max mem: 3532 +Test: [270/389] eta: 0:00:24 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1652 (0.7491) acc1: 97.9167 (62.3059) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2023 data: 0.1108 max mem: 3532 +Test: [280/389] eta: 0:00:22 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.2458 (0.7373) acc1: 96.3542 (63.0079) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1984 data: 0.1069 max mem: 3532 +Test: [290/389] eta: 0:00:20 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1500 (0.7209) acc1: 99.2188 (63.9211) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1923 data: 0.0984 max mem: 3532 +Test: [300/389] eta: 0:00:18 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1854 (0.7027) acc1: 100.0000 (65.1033) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1905 data: 0.0970 max mem: 3532 +Test: [310/389] eta: 0:00:16 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1876 (0.6878) acc1: 98.9583 (66.0680) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1888 data: 0.0986 max mem: 3532 +Test: [320/389] eta: 0:00:14 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.2011 (0.6729) acc1: 98.4375 (67.0244) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1927 data: 0.1014 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1066 (0.6557) acc1: 100.0000 (68.0152) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1732 data: 0.0816 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1108 (0.6405) acc1: 100.0000 (68.9424) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1556 data: 0.0654 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1559 (0.6425) acc1: 99.2188 (68.4911) acc5: 100.0000 (100.0000) acc1_neg: 4.427 acc5_neg: 100.000 time: 0.1356 data: 0.0458 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 1.227 acc5_pos: 100.000 loss: 1.0028 (0.6538) acc1: 25.7812 (67.1687) acc5: 100.0000 (100.0000) acc1_neg: 29.167 acc5_neg: 100.000 time: 0.1161 data: 0.0265 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.5815 (0.6414) acc1: 29.1667 (67.8943) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1238 data: 0.0340 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1108 (0.6291) acc1: 100.0000 (68.6523) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1216 data: 0.0319 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1478 (0.6208) acc1: 100.0000 (69.2035) acc5: 100.0000 (100.0000) acc1_neg: 94.262 acc5_neg: 100.000 time: 0.1179 data: 0.0252 max mem: 3532 +Test: Total time: 0:01:14 (0.1917 s / it) +* Overall Acc@1 69.204 Acc@5 100.000 + target == real Acc@1 30.050 Acc@5 100.000 + target == fake Acc@1 87.403 Acc@5 100.000 + loss 0.621 + PR-AUC 0.8093 ROC-AUC 0.6956 EER 33.82% (thr=0.8232) +Accuracy of the network on the 149358 test images: 69.2% +[rank0]:[W726 04:00:40.080451075 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:01:03.310000 3142260 torch/distributed/run.py:766] +W0726 04:01:03.310000 3142260 torch/distributed/run.py:766] ***************************************** +W0726 04:01:03.310000 3142260 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:01:03.310000 3142260 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:01:38.877084746 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:01:38.960741169 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:01:39.565338644 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank4]:[W726 04:01:39.565337643 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:01:39.616961154 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:01:39.685338869 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +[rank0]:[W726 04:01:39.710825893 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:01:39.720057556 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n5_seed6_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed6_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:28:47 acc1_pos: 9.115 acc5_pos: 100.000 loss: 1.5681 (1.5681) acc1: 9.1146 (9.1146) acc5: 100.0000 (100.0000) time: 5.1887 data: 3.0747 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/333] eta: 0:03:10 acc1_pos: 11.979 acc5_pos: 100.000 loss: 1.5435 (1.5078) acc1: 11.4583 (10.9375) acc5: 100.0000 (100.0000) time: 0.5885 data: 0.3150 max mem: 3531 +Test: [ 20/333] eta: 0:02:04 acc1_pos: 6.250 acc5_pos: 100.000 loss: 1.5504 (1.5681) acc1: 8.3333 (10.0694) acc5: 100.0000 (100.0000) time: 0.1580 data: 0.0673 max mem: 3531 +Test: [ 30/333] eta: 0:01:52 acc1_pos: 65.625 acc5_pos: 100.000 loss: 1.1522 (1.3366) acc1: 17.7083 (22.5386) acc5: 100.0000 (100.0000) time: 0.2511 data: 0.1595 max mem: 3531 +Test: [ 40/333] eta: 0:01:34 acc1_pos: 59.896 acc5_pos: 100.000 loss: 0.7731 (1.1983) acc1: 53.6458 (30.8943) acc5: 100.0000 (100.0000) time: 0.2464 data: 0.1549 max mem: 3531 +Test: [ 50/333] eta: 0:01:26 acc1_pos: 41.667 acc5_pos: 100.000 loss: 0.7731 (1.1203) acc1: 53.6458 (35.2482) acc5: 100.0000 (100.0000) time: 0.2028 data: 0.1106 max mem: 3531 +Test: [ 60/333] eta: 0:01:19 acc1_pos: 59.635 acc5_pos: 100.000 loss: 0.7727 (1.0644) acc1: 53.1250 (38.2813) acc5: 100.0000 (100.0000) time: 0.2253 data: 0.1293 max mem: 3531 +Test: [ 70/333] eta: 0:01:12 acc1_pos: 57.292 acc5_pos: 100.000 loss: 0.8101 (1.0344) acc1: 55.2083 (40.2766) acc5: 100.0000 (100.0000) time: 0.1948 data: 0.0983 max mem: 3531 +Test: [ 80/333] eta: 0:01:05 acc1_pos: 46.615 acc5_pos: 100.000 loss: 0.8187 (1.0061) acc1: 52.0833 (42.1489) acc5: 100.0000 (100.0000) time: 0.1530 data: 0.0573 max mem: 3531 +Test: [ 90/333] eta: 0:01:00 acc1_pos: 57.292 acc5_pos: 100.000 loss: 0.7648 (0.9756) acc1: 52.8646 (43.6041) acc5: 100.0000 (100.0000) time: 0.1568 data: 0.0616 max mem: 3531 +Test: [100/333] eta: 0:00:56 acc1_pos: 53.385 acc5_pos: 100.000 loss: 0.7439 (0.9513) acc1: 55.9896 (45.0830) acc5: 100.0000 (100.0000) time: 0.1717 data: 0.0800 max mem: 3531 +Test: [110/333] eta: 0:00:52 acc1_pos: 52.865 acc5_pos: 100.000 loss: 0.7640 (0.9380) acc1: 53.3854 (45.7465) acc5: 100.0000 (100.0000) time: 0.1742 data: 0.0839 max mem: 3531 +Test: [120/333] eta: 0:00:49 acc1_pos: 59.115 acc5_pos: 100.000 loss: 0.7640 (0.9207) acc1: 54.9479 (46.9159) acc5: 100.0000 (100.0000) time: 0.1860 data: 0.0957 max mem: 3531 +Test: [130/333] eta: 0:00:46 acc1_pos: 76.823 acc5_pos: 100.000 loss: 0.7123 (0.9053) acc1: 60.4167 (47.8809) acc5: 100.0000 (100.0000) time: 0.1853 data: 0.0950 max mem: 3531 +Test: [140/333] eta: 0:00:43 acc1_pos: 67.708 acc5_pos: 100.000 loss: 0.7038 (0.8869) acc1: 60.6771 (48.9509) acc5: 100.0000 (100.0000) time: 0.1839 data: 0.0921 max mem: 3531 +Test: [150/333] eta: 0:00:40 acc1_pos: 62.760 acc5_pos: 100.000 loss: 0.6765 (0.8777) acc1: 58.8542 (49.4567) acc5: 100.0000 (100.0000) time: 0.1734 data: 0.0811 max mem: 3531 +Test: [160/333] eta: 0:00:37 acc1_pos: 57.566 acc5_pos: 100.000 loss: 0.7387 (0.8692) acc1: 55.9896 (50.0421) acc5: 100.0000 (100.0000) acc1_neg: 50.781 acc5_neg: 100.000 time: 0.1714 data: 0.0802 max mem: 3531 +Test: [170/333] eta: 0:00:35 acc1_pos: 57.566 acc5_pos: 100.000 loss: 0.6707 (0.8592) acc1: 59.3750 (50.7538) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.1764 data: 0.0842 max mem: 3531 +Test: [180/333] eta: 0:00:32 acc1_pos: 57.566 acc5_pos: 100.000 loss: 0.6323 (0.8438) acc1: 65.3646 (51.6877) acc5: 100.0000 (100.0000) acc1_neg: 76.042 acc5_neg: 100.000 time: 0.1666 data: 0.0736 max mem: 3531 +Test: [190/333] eta: 0:00:30 acc1_pos: 57.566 acc5_pos: 100.000 loss: 0.5770 (0.8284) acc1: 68.4896 (52.6832) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.1692 data: 0.0754 max mem: 3531 +Test: [200/333] eta: 0:00:27 acc1_pos: 57.566 acc5_pos: 100.000 loss: 0.5719 (0.8166) acc1: 68.7500 (53.5279) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.1862 data: 0.0941 max mem: 3531 +Test: [210/333] eta: 0:00:25 acc1_pos: 57.566 acc5_pos: 100.000 loss: 0.6011 (0.8048) acc1: 68.7500 (54.3827) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1839 data: 0.0937 max mem: 3532 +Test: [220/333] eta: 0:00:23 acc1_pos: 57.566 acc5_pos: 100.000 loss: 0.6011 (0.7956) acc1: 68.7500 (54.9903) acc5: 100.0000 (100.0000) acc1_neg: 69.271 acc5_neg: 100.000 time: 0.1837 data: 0.0918 max mem: 3532 +Test: [230/333] eta: 0:00:21 acc1_pos: 57.566 acc5_pos: 100.000 loss: 0.5471 (0.7844) acc1: 72.1354 (55.8351) acc5: 100.0000 (100.0000) acc1_neg: 72.135 acc5_neg: 100.000 time: 0.1868 data: 0.0926 max mem: 3532 +Test: [240/333] eta: 0:00:19 acc1_pos: 57.566 acc5_pos: 100.000 loss: 0.5582 (0.7823) acc1: 70.0521 (55.9172) acc5: 100.0000 (100.0000) acc1_neg: 56.510 acc5_neg: 100.000 time: 0.1810 data: 0.0854 max mem: 3532 +Test: [250/333] eta: 0:00:16 acc1_pos: 57.566 acc5_pos: 100.000 loss: 0.7210 (0.7797) acc1: 56.5104 (56.1068) acc5: 100.0000 (100.0000) acc1_neg: 61.458 acc5_neg: 100.000 time: 0.1782 data: 0.0855 max mem: 3532 +Test: [260/333] eta: 0:00:14 acc1_pos: 57.566 acc5_pos: 100.000 loss: 0.6398 (0.7730) acc1: 66.6667 (56.6112) acc5: 100.0000 (100.0000) acc1_neg: 66.667 acc5_neg: 100.000 time: 0.1739 data: 0.0816 max mem: 3532 +Test: [270/333] eta: 0:00:12 acc1_pos: 57.566 acc5_pos: 100.000 loss: 0.5638 (0.7638) acc1: 71.0938 (57.2571) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.1853 data: 0.0928 max mem: 3532 +Test: [280/333] eta: 0:00:10 acc1_pos: 57.566 acc5_pos: 100.000 loss: 0.5526 (0.7585) acc1: 70.3125 (57.6188) acc5: 100.0000 (100.0000) acc1_neg: 69.271 acc5_neg: 100.000 time: 0.1845 data: 0.0912 max mem: 3532 +Test: [290/333] eta: 0:00:08 acc1_pos: 57.566 acc5_pos: 100.000 loss: 0.7091 (0.7575) acc1: 61.4583 (57.6836) acc5: 100.0000 (100.0000) acc1_neg: 47.396 acc5_neg: 100.000 time: 0.1863 data: 0.0909 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 57.566 acc5_pos: 100.000 loss: 0.7373 (0.7580) acc1: 58.5938 (57.6066) acc5: 100.0000 (100.0000) acc1_neg: 53.125 acc5_neg: 100.000 time: 0.1949 data: 0.1000 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 57.566 acc5_pos: 100.000 loss: 0.7103 (0.7528) acc1: 58.8542 (57.9967) acc5: 100.0000 (100.0000) acc1_neg: 62.240 acc5_neg: 100.000 time: 0.1883 data: 0.0953 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 57.566 acc5_pos: 100.000 loss: 0.6577 (0.7501) acc1: 64.0625 (58.1857) acc5: 100.0000 (100.0000) acc1_neg: 70.573 acc5_neg: 100.000 time: 0.1635 data: 0.0726 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 57.566 acc5_pos: 100.000 loss: 0.6197 (0.7457) acc1: 66.4062 (58.4986) acc5: 100.0000 (100.0000) acc1_neg: 71.094 acc5_neg: 100.000 time: 0.1373 data: 0.0467 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 57.566 acc5_pos: 100.000 loss: 0.6119 (0.7449) acc1: 67.4479 (58.5570) acc5: 100.0000 (100.0000) acc1_neg: 69.859 acc5_neg: 100.000 time: 0.1353 data: 0.0415 max mem: 3532 +Test: Total time: 0:01:05 (0.1968 s / it) +* Overall Acc@1 58.557 Acc@5 100.000 + target == real Acc@1 49.975 Acc@5 100.000 + target == fake Acc@1 66.196 Acc@5 100.000 + loss 0.745 + PR-AUC 0.6040 ROC-AUC 0.6128 EER 42.21% (thr=0.5840) +Accuracy of the network on the 127843 test images: 58.6% +[rank0]:[W726 04:02:55.296599874 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:03:16.268000 3142820 torch/distributed/run.py:766] +W0726 04:03:16.268000 3142820 torch/distributed/run.py:766] ***************************************** +W0726 04:03:16.268000 3142820 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:03:16.268000 3142820 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 1): env:// +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:03:49.129616963 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:03:50.471993247 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:03:50.516805227 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W726 04:03:50.521177334 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:03:50.053767186 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:03:50.119907291 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:03:51.185786388 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:03:51.409642067 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n5_seed6_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed6_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:26:27 acc1_pos: 15.885 acc5_pos: 100.000 loss: 1.5725 (1.5725) acc1: 15.8854 (15.8854) acc5: 100.0000 (100.0000) time: 6.7557 data: 5.0463 max mem: 3531 +Test: [ 10/235] eta: 0:02:37 acc1_pos: 13.802 acc5_pos: 100.000 loss: 1.6118 (1.6082) acc1: 13.8021 (13.8021) acc5: 100.0000 (100.0000) time: 0.7022 data: 0.4592 max mem: 3531 +Test: [ 20/235] eta: 0:01:33 acc1_pos: 13.281 acc5_pos: 100.000 loss: 1.6118 (1.6092) acc1: 13.5417 (13.7773) acc5: 100.0000 (100.0000) time: 0.1184 data: 0.0221 max mem: 3531 +Test: [ 30/235] eta: 0:01:23 acc1_pos: 12.500 acc5_pos: 100.000 loss: 1.6066 (1.6091) acc1: 13.8021 (13.7937) acc5: 100.0000 (100.0000) time: 0.2464 data: 0.1520 max mem: 3531 +Test: [ 40/235] eta: 0:01:14 acc1_pos: 12.760 acc5_pos: 100.000 loss: 1.6068 (1.6110) acc1: 13.2812 (13.7640) acc5: 100.0000 (100.0000) time: 0.3249 data: 0.2333 max mem: 3531 +Test: [ 50/235] eta: 0:01:07 acc1_pos: 13.542 acc5_pos: 100.000 loss: 1.6068 (1.6113) acc1: 13.2812 (13.7102) acc5: 100.0000 (100.0000) time: 0.2934 data: 0.1977 max mem: 3531 +Test: [ 60/235] eta: 0:00:58 acc1_pos: 15.625 acc5_pos: 100.000 loss: 1.5998 (1.6080) acc1: 13.5417 (13.7807) acc5: 100.0000 (100.0000) time: 0.2315 data: 0.1328 max mem: 3531 +Test: [ 70/235] eta: 0:00:52 acc1_pos: 12.308 acc5_pos: 100.000 loss: 1.5428 (1.4316) acc1: 15.6250 (24.0207) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2023 data: 0.1069 max mem: 3531 +Test: [ 80/235] eta: 0:00:47 acc1_pos: 12.308 acc5_pos: 100.000 loss: 0.1870 (1.2777) acc1: 96.3542 (32.9861) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2170 data: 0.1246 max mem: 3531 +Test: [ 90/235] eta: 0:00:42 acc1_pos: 12.308 acc5_pos: 100.000 loss: 0.1838 (1.1581) acc1: 96.6146 (39.9525) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2149 data: 0.1235 max mem: 3531 +Test: [100/235] eta: 0:00:38 acc1_pos: 12.308 acc5_pos: 100.000 loss: 0.1838 (1.0617) acc1: 96.6146 (45.5703) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2080 data: 0.1141 max mem: 3531 +Test: [110/235] eta: 0:00:34 acc1_pos: 12.308 acc5_pos: 100.000 loss: 0.1831 (0.9831) acc1: 96.6146 (50.1431) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1858 data: 0.0927 max mem: 3531 +Test: [120/235] eta: 0:00:30 acc1_pos: 12.308 acc5_pos: 100.000 loss: 0.2140 (0.9208) acc1: 95.5729 (53.8524) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1907 data: 0.0974 max mem: 3531 +Test: [130/235] eta: 0:00:27 acc1_pos: 12.308 acc5_pos: 100.000 loss: 0.2373 (0.8690) acc1: 94.2708 (56.9199) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2036 data: 0.1098 max mem: 3531 +Test: [140/235] eta: 0:00:24 acc1_pos: 12.308 acc5_pos: 100.000 loss: 0.2377 (0.8241) acc1: 94.2708 (59.6206) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2093 data: 0.1176 max mem: 3531 +Test: [150/235] eta: 0:00:21 acc1_pos: 12.308 acc5_pos: 100.000 loss: 0.2343 (0.7850) acc1: 95.0521 (61.9861) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1968 data: 0.1050 max mem: 3531 +Test: [160/235] eta: 0:00:18 acc1_pos: 12.308 acc5_pos: 100.000 loss: 0.2368 (0.7515) acc1: 94.7917 (64.0010) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1888 data: 0.0957 max mem: 3531 +Test: [170/235] eta: 0:00:16 acc1_pos: 12.308 acc5_pos: 100.000 loss: 0.2328 (0.7197) acc1: 94.7917 (65.8215) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1966 data: 0.1012 max mem: 3531 +Test: [180/235] eta: 0:00:13 acc1_pos: 12.308 acc5_pos: 100.000 loss: 0.2040 (0.6909) acc1: 95.3125 (67.4724) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1890 data: 0.0950 max mem: 3531 +Test: [190/235] eta: 0:00:10 acc1_pos: 12.308 acc5_pos: 100.000 loss: 0.1984 (0.6651) acc1: 96.0938 (68.9600) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1581 data: 0.0673 max mem: 3531 +Test: [200/235] eta: 0:00:08 acc1_pos: 12.308 acc5_pos: 100.000 loss: 0.1995 (0.6422) acc1: 95.8333 (70.2801) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1324 data: 0.0424 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 12.308 acc5_pos: 100.000 loss: 0.2046 (0.6213) acc1: 95.5729 (71.4838) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1253 data: 0.0354 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 12.308 acc5_pos: 100.000 loss: 0.2050 (0.6022) acc1: 95.3125 (72.5785) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1222 data: 0.0327 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 12.308 acc5_pos: 100.000 loss: 0.1968 (0.5846) acc1: 95.3125 (73.5852) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1196 data: 0.0300 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 12.308 acc5_pos: 100.000 loss: 0.1947 (0.5779) acc1: 95.5729 (73.8693) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1184 data: 0.0319 max mem: 3532 +Test: Total time: 0:00:51 (0.2173 s / it) +* Overall Acc@1 73.869 Acc@5 100.000 + target == real Acc@1 13.769 Acc@5 100.000 + target == fake Acc@1 95.606 Acc@5 100.000 + loss 0.578 + PR-AUC 0.8113 ROC-AUC 0.6638 EER 37.29% (thr=0.8574) +Accuracy of the network on the 89879 test images: 73.9% +[rank0]:[W726 04:04:50.663863358 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:05:13.387000 3143379 torch/distributed/run.py:766] +W0726 04:05:13.387000 3143379 torch/distributed/run.py:766] ***************************************** +W0726 04:05:13.387000 3143379 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:05:13.387000 3143379 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:05:49.786304416 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:05:49.168908015 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:05:50.497584745 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:05:50.703666404 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:05:50.873318892 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:05:50.139975312 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:05:50.163587683 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:05:51.212828489 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n5_seed6_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed6_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:22:30 acc1_pos: 24.219 acc5_pos: 100.000 loss: 1.2462 (1.2462) acc1: 24.2188 (24.2188) acc5: 100.0000 (100.0000) time: 5.0202 data: 2.7597 max mem: 3531 +Test: [ 10/269] eta: 0:02:19 acc1_pos: 45.833 acc5_pos: 100.000 loss: 0.8996 (0.9328) acc1: 45.8333 (42.1402) acc5: 100.0000 (100.0000) time: 0.5381 data: 0.2512 max mem: 3531 +Test: [ 20/269] eta: 0:01:27 acc1_pos: 49.219 acc5_pos: 100.000 loss: 0.8103 (0.8885) acc1: 45.8333 (45.8333) acc5: 100.0000 (100.0000) time: 0.1194 data: 0.0291 max mem: 3531 +Test: [ 30/269] eta: 0:01:21 acc1_pos: 31.771 acc5_pos: 100.000 loss: 0.8797 (0.8744) acc1: 44.2708 (46.3038) acc5: 100.0000 (100.0000) time: 0.2310 data: 0.1380 max mem: 3531 +Test: [ 40/269] eta: 0:01:10 acc1_pos: 52.586 acc5_pos: 100.000 loss: 0.6422 (0.7923) acc1: 66.6667 (53.1123) acc5: 100.0000 (100.0000) acc1_neg: 66.667 acc5_neg: 100.000 time: 0.2586 data: 0.1633 max mem: 3531 +Test: [ 50/269] eta: 0:01:00 acc1_pos: 52.586 acc5_pos: 100.000 loss: 0.5861 (0.7663) acc1: 70.5729 (55.3360) acc5: 100.0000 (100.0000) acc1_neg: 45.312 acc5_neg: 100.000 time: 0.1796 data: 0.0862 max mem: 3531 +Test: [ 60/269] eta: 0:00:55 acc1_pos: 52.586 acc5_pos: 100.000 loss: 0.6909 (0.7598) acc1: 59.3750 (55.6481) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.1823 data: 0.0889 max mem: 3531 +Test: [ 70/269] eta: 0:00:51 acc1_pos: 52.586 acc5_pos: 100.000 loss: 0.5192 (0.7180) acc1: 74.4792 (59.0632) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.2060 data: 0.1132 max mem: 3531 +Test: [ 80/269] eta: 0:00:46 acc1_pos: 52.586 acc5_pos: 100.000 loss: 0.4753 (0.6877) acc1: 78.9062 (61.6770) acc5: 100.0000 (100.0000) acc1_neg: 78.646 acc5_neg: 100.000 time: 0.1896 data: 0.0994 max mem: 3531 +Test: [ 90/269] eta: 0:00:42 acc1_pos: 52.586 acc5_pos: 100.000 loss: 0.5924 (0.6904) acc1: 73.1771 (61.6472) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.1788 data: 0.0875 max mem: 3531 +Test: [100/269] eta: 0:00:39 acc1_pos: 52.586 acc5_pos: 100.000 loss: 0.6001 (0.6745) acc1: 69.5312 (63.2013) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1873 data: 0.0942 max mem: 3531 +Test: [110/269] eta: 0:00:36 acc1_pos: 52.586 acc5_pos: 100.000 loss: 0.4581 (0.6531) acc1: 78.9062 (64.9118) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1855 data: 0.0912 max mem: 3531 +Test: [120/269] eta: 0:00:33 acc1_pos: 52.586 acc5_pos: 100.000 loss: 0.4616 (0.6509) acc1: 78.6458 (65.0676) acc5: 100.0000 (100.0000) acc1_neg: 40.625 acc5_neg: 100.000 time: 0.1831 data: 0.0878 max mem: 3531 +Test: [130/269] eta: 0:00:31 acc1_pos: 52.586 acc5_pos: 100.000 loss: 0.6378 (0.6513) acc1: 67.4479 (64.9968) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.1926 data: 0.0964 max mem: 3531 +Test: [140/269] eta: 0:00:28 acc1_pos: 52.586 acc5_pos: 100.000 loss: 0.6768 (0.6565) acc1: 62.2396 (64.6683) acc5: 100.0000 (100.0000) acc1_neg: 44.010 acc5_neg: 100.000 time: 0.1938 data: 0.1003 max mem: 3531 +Test: [150/269] eta: 0:00:26 acc1_pos: 52.586 acc5_pos: 100.000 loss: 0.7577 (0.6652) acc1: 55.7292 (64.0073) acc5: 100.0000 (100.0000) acc1_neg: 61.719 acc5_neg: 100.000 time: 0.1949 data: 0.1035 max mem: 3531 +Test: [160/269] eta: 0:00:23 acc1_pos: 52.586 acc5_pos: 100.000 loss: 0.7230 (0.6644) acc1: 58.3333 (64.1968) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.1947 data: 0.1029 max mem: 3531 +Test: [170/269] eta: 0:00:21 acc1_pos: 52.586 acc5_pos: 100.000 loss: 0.5513 (0.6573) acc1: 73.4375 (64.8788) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.2012 data: 0.1101 max mem: 3531 +Test: [180/269] eta: 0:00:19 acc1_pos: 52.586 acc5_pos: 100.000 loss: 0.4720 (0.6439) acc1: 81.7708 (66.0235) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.2143 data: 0.1236 max mem: 3531 +Test: [190/269] eta: 0:00:17 acc1_pos: 52.586 acc5_pos: 100.000 loss: 0.4938 (0.6412) acc1: 81.2500 (66.2331) acc5: 100.0000 (100.0000) acc1_neg: 75.521 acc5_neg: 100.000 time: 0.2362 data: 0.1457 max mem: 3531 +Test: [200/269] eta: 0:00:15 acc1_pos: 52.586 acc5_pos: 100.000 loss: 0.5710 (0.6395) acc1: 69.2708 (66.3000) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.2112 data: 0.1205 max mem: 3531 +Test: [210/269] eta: 0:00:12 acc1_pos: 52.586 acc5_pos: 100.000 loss: 0.5811 (0.6364) acc1: 73.4375 (66.7099) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.2046 data: 0.1101 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 52.586 acc5_pos: 100.000 loss: 0.5811 (0.6373) acc1: 73.4375 (66.7539) acc5: 100.0000 (100.0000) acc1_neg: 69.531 acc5_neg: 100.000 time: 0.2208 data: 0.1264 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 52.586 acc5_pos: 100.000 loss: 0.5834 (0.6346) acc1: 70.5729 (67.0139) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.1781 data: 0.0855 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 52.586 acc5_pos: 100.000 loss: 0.5719 (0.6339) acc1: 71.8750 (67.0211) acc5: 100.0000 (100.0000) acc1_neg: 55.729 acc5_neg: 100.000 time: 0.1783 data: 0.0863 max mem: 3532 +Test: [250/269] eta: 0:00:04 acc1_pos: 52.586 acc5_pos: 100.000 loss: 0.6976 (0.6400) acc1: 58.5938 (66.4374) acc5: 100.0000 (100.0000) acc1_neg: 39.583 acc5_neg: 100.000 time: 0.1891 data: 0.0993 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 52.586 acc5_pos: 100.000 loss: 0.7502 (0.6477) acc1: 53.1250 (65.8016) acc5: 100.0000 (100.0000) acc1_neg: 72.917 acc5_neg: 100.000 time: 0.1518 data: 0.0625 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 52.586 acc5_pos: 100.000 loss: 0.6667 (0.6474) acc1: 58.0729 (65.7702) acc5: 100.0000 (100.0000) acc1_neg: 73.753 acc5_neg: 100.000 time: 0.1205 data: 0.0275 max mem: 3532 +Test: Total time: 0:00:55 (0.2075 s / it) +* Overall Acc@1 65.770 Acc@5 100.000 + target == real Acc@1 46.138 Acc@5 100.000 + target == fake Acc@1 68.449 Acc@5 100.000 + loss 0.647 + PR-AUC 0.9088 ROC-AUC 0.5989 EER 42.55% (thr=0.5898) +Accuracy of the network on the 103293 test images: 65.8% +[rank0]:[W726 04:06:55.917629123 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T04:07:02Z] eval_done tag=fixed500k_real_n5_seed6_resampled +[2026-07-26T04:07:02Z] eval_start tag=fixed500k_fake_n11_seed6_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed6_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T04:07:02Z +hostname=n116-040-168 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed6_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n11_seed6_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 04:07:17.271000 3143957 torch/distributed/run.py:766] +W0726 04:07:17.271000 3143957 torch/distributed/run.py:766] ***************************************** +W0726 04:07:17.271000 3143957 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:07:17.271000 3143957 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:07:54.840601796 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:07:54.859794963 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W726 04:07:54.861060515 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:07:54.968582935 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:07:54.995868193 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:07:54.037035934 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +[rank0]:[W726 04:07:54.061689504 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:07:54.069384508 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n11_seed6_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed6_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:25:59 acc1_pos: 32.552 acc5_pos: 100.000 loss: 0.8592 (0.8592) acc1: 32.5521 (32.5521) acc5: 100.0000 (100.0000) time: 6.6641 data: 5.1785 max mem: 3531 +Test: [ 10/234] eta: 0:02:34 acc1_pos: 49.219 acc5_pos: 100.000 loss: 0.7409 (0.7859) acc1: 49.2188 (48.5795) acc5: 100.0000 (100.0000) time: 0.6877 data: 0.4712 max mem: 3531 +Test: [ 20/234] eta: 0:01:28 acc1_pos: 95.052 acc5_pos: 100.000 loss: 0.7162 (0.7287) acc1: 49.2188 (53.0134) acc5: 100.0000 (100.0000) time: 0.1034 data: 0.0108 max mem: 3531 +Test: [ 30/234] eta: 0:01:14 acc1_pos: 44.010 acc5_pos: 100.000 loss: 0.2251 (0.5929) acc1: 94.2708 (64.1381) acc5: 100.0000 (100.0000) time: 0.1845 data: 0.0885 max mem: 3531 +Test: [ 40/234] eta: 0:01:05 acc1_pos: 98.958 acc5_pos: 100.000 loss: 0.2010 (0.5858) acc1: 95.8333 (64.5135) acc5: 100.0000 (100.0000) time: 0.2509 data: 0.1577 max mem: 3531 +Test: [ 50/234] eta: 0:00:55 acc1_pos: 98.698 acc5_pos: 100.000 loss: 0.1302 (0.4950) acc1: 98.1771 (71.2112) acc5: 100.0000 (100.0000) time: 0.2087 data: 0.1153 max mem: 3531 +Test: [ 60/234] eta: 0:00:47 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.1280 (0.4353) acc1: 99.2188 (75.9136) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1404 data: 0.0427 max mem: 3531 +Test: [ 70/234] eta: 0:00:42 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.1337 (0.4253) acc1: 100.0000 (77.1347) acc5: 100.0000 (100.0000) acc1_neg: 74.740 acc5_neg: 100.000 time: 0.1475 data: 0.0525 max mem: 3531 +Test: [ 80/234] eta: 0:00:39 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4431 (0.4191) acc1: 78.1250 (77.9964) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2059 data: 0.1141 max mem: 3531 +Test: [ 90/234] eta: 0:00:35 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.1843 (0.3898) acc1: 95.5729 (80.0738) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.2038 data: 0.1125 max mem: 3531 +Test: [100/234] eta: 0:00:33 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.1843 (0.3877) acc1: 94.7917 (80.4146) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2184 data: 0.1254 max mem: 3531 +Test: [110/234] eta: 0:00:29 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.3838 (0.3824) acc1: 82.2917 (81.0576) acc5: 100.0000 (100.0000) acc1_neg: 77.083 acc5_neg: 100.000 time: 0.2107 data: 0.1147 max mem: 3531 +Test: [120/234] eta: 0:00:26 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.2855 (0.3849) acc1: 90.6250 (81.3146) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1289 data: 0.0332 max mem: 3531 +Test: [130/234] eta: 0:00:22 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.3212 (0.3827) acc1: 90.3646 (81.8543) acc5: 100.0000 (100.0000) acc1_neg: 72.135 acc5_neg: 100.000 time: 0.0976 data: 0.0005 max mem: 3531 +Test: [140/234] eta: 0:00:20 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4339 (0.4229) acc1: 79.9479 (79.7595) acc5: 100.0000 (100.0000) acc1_neg: 27.344 acc5_neg: 100.000 time: 0.1314 data: 0.0309 max mem: 3531 +Test: [150/234] eta: 0:00:17 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4833 (0.4328) acc1: 73.4375 (79.3029) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.1808 data: 0.0793 max mem: 3531 +Test: [160/234] eta: 0:00:15 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.3843 (0.4275) acc1: 85.4167 (79.9026) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1937 data: 0.0977 max mem: 3531 +Test: [170/234] eta: 0:00:13 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.3754 (0.4247) acc1: 85.9375 (80.3302) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1864 data: 0.0942 max mem: 3531 +Test: [180/234] eta: 0:00:11 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4024 (0.4311) acc1: 84.8958 (79.9824) acc5: 100.0000 (100.0000) acc1_neg: 45.312 acc5_neg: 100.000 time: 0.1796 data: 0.0848 max mem: 3531 +Test: [190/234] eta: 0:00:09 acc1_pos: 97.931 acc5_pos: 100.000 loss: 1.0010 (0.4633) acc1: 45.3125 (77.9341) acc5: 100.0000 (100.0000) acc1_neg: 42.708 acc5_neg: 100.000 time: 0.1588 data: 0.0642 max mem: 3531 +Test: [200/234] eta: 0:00:06 acc1_pos: 97.931 acc5_pos: 100.000 loss: 1.0232 (0.4635) acc1: 42.1875 (77.9488) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1561 data: 0.0636 max mem: 3531 +Test: [210/234] eta: 0:00:04 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.1260 (0.4471) acc1: 99.7396 (78.9865) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1694 data: 0.0765 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.1067 (0.4330) acc1: 99.7396 (79.8301) acc5: 100.0000 (100.0000) acc1_neg: 78.646 acc5_neg: 100.000 time: 0.2037 data: 0.1095 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4253 (0.4589) acc1: 38.2812 (77.9491) acc5: 100.0000 (100.0000) acc1_neg: 35.938 acc5_neg: 100.000 time: 0.1980 data: 0.1048 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 97.931 acc5_pos: 100.000 loss: 1.0100 (0.4661) acc1: 37.7604 (77.4833) acc5: 100.0000 (100.0000) acc1_neg: 38.911 acc5_neg: 100.000 time: 0.1932 data: 0.0919 max mem: 3532 +Test: Total time: 0:00:47 (0.2014 s / it) +* Overall Acc@1 77.483 Acc@5 100.000 + target == real Acc@1 71.408 Acc@5 100.000 + target == fake Acc@1 79.196 Acc@5 100.000 + loss 0.466 + PR-AUC 0.9591 ROC-AUC 0.8625 EER 23.49% (thr=0.5620) +Accuracy of the network on the 89729 test images: 77.5% +[rank0]:[W726 04:08:53.135648037 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:09:15.195000 3144516 torch/distributed/run.py:766] +W0726 04:09:15.195000 3144516 torch/distributed/run.py:766] ***************************************** +W0726 04:09:15.195000 3144516 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:09:15.195000 3144516 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:09:48.765392374 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:09:49.361660763 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:09:49.532698046 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:09:49.857349976 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:09:49.945886015 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:09:50.187771273 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:09:50.372749051 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:09:50.628022107 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n11_seed6_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed6_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:08:23 acc1_pos: 13.542 acc5_pos: 100.000 loss: 1.3069 (1.3069) acc1: 13.5417 (13.5417) acc5: 100.0000 (100.0000) time: 6.2435 data: 4.7431 max mem: 3531 +Test: [ 10/2387] eta: 0:25:45 acc1_pos: 5.208 acc5_pos: 100.000 loss: 1.3841 (1.3325) acc1: 8.8542 (11.4347) acc5: 100.0000 (100.0000) time: 0.6502 data: 0.4315 max mem: 3531 +Test: [ 20/2387] eta: 0:15:14 acc1_pos: 3.385 acc5_pos: 100.000 loss: 1.3815 (1.3484) acc1: 8.5938 (9.9578) acc5: 100.0000 (100.0000) time: 0.0936 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:11:55 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.4295 (1.3738) acc1: 6.2500 (9.1398) acc5: 100.0000 (100.0000) time: 0.1131 data: 0.0165 max mem: 3531 +Test: [ 40/2387] eta: 0:11:12 acc1_pos: 11.458 acc5_pos: 100.000 loss: 1.4308 (1.3858) acc1: 5.7292 (8.6954) acc5: 100.0000 (100.0000) time: 0.1818 data: 0.0865 max mem: 3531 +Test: [ 50/2387] eta: 0:10:34 acc1_pos: 19.792 acc5_pos: 100.000 loss: 1.3787 (1.3715) acc1: 9.3750 (9.7631) acc5: 100.0000 (100.0000) time: 0.2216 data: 0.1271 max mem: 3531 +Test: [ 60/2387] eta: 0:10:11 acc1_pos: 22.396 acc5_pos: 100.000 loss: 1.3650 (1.3680) acc1: 12.7604 (10.3825) acc5: 100.0000 (100.0000) time: 0.2141 data: 0.1199 max mem: 3531 +Test: [ 70/2387] eta: 0:10:07 acc1_pos: 32.031 acc5_pos: 100.000 loss: 1.3700 (1.3734) acc1: 11.4583 (10.9008) acc5: 100.0000 (100.0000) time: 0.2382 data: 0.1408 max mem: 3531 +Test: [ 80/2387] eta: 0:09:49 acc1_pos: 32.031 acc5_pos: 100.000 loss: 1.3149 (1.3594) acc1: 11.4583 (11.5291) acc5: 100.0000 (100.0000) time: 0.2338 data: 0.1344 max mem: 3531 +Test: [ 90/2387] eta: 0:09:48 acc1_pos: 18.229 acc5_pos: 100.000 loss: 1.1321 (1.3281) acc1: 18.2292 (13.2526) acc5: 100.0000 (100.0000) time: 0.2349 data: 0.1397 max mem: 3531 +Test: [ 100/2387] eta: 0:09:44 acc1_pos: 19.792 acc5_pos: 100.000 loss: 1.0942 (1.3108) acc1: 21.3542 (13.8769) acc5: 100.0000 (100.0000) time: 0.2549 data: 0.1587 max mem: 3531 +Test: [ 110/2387] eta: 0:09:36 acc1_pos: 31.510 acc5_pos: 100.000 loss: 1.1169 (1.2913) acc1: 20.8333 (14.7945) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1429 max mem: 3531 +Test: [ 120/2387] eta: 0:09:33 acc1_pos: 29.167 acc5_pos: 100.000 loss: 1.1169 (1.2757) acc1: 24.7396 (15.6896) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1440 max mem: 3531 +Test: [ 130/2387] eta: 0:09:28 acc1_pos: 14.323 acc5_pos: 100.000 loss: 1.1214 (1.2664) acc1: 22.1354 (15.9769) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1483 max mem: 3531 +Test: [ 140/2387] eta: 0:09:22 acc1_pos: 21.094 acc5_pos: 100.000 loss: 1.1711 (1.2592) acc1: 20.8333 (16.3712) acc5: 100.0000 (100.0000) time: 0.2349 data: 0.1374 max mem: 3531 +Test: [ 150/2387] eta: 0:09:18 acc1_pos: 9.375 acc5_pos: 100.000 loss: 1.1711 (1.2497) acc1: 21.0938 (16.7098) acc5: 100.0000 (100.0000) time: 0.2349 data: 0.1372 max mem: 3531 +Test: [ 160/2387] eta: 0:09:20 acc1_pos: 27.397 acc5_pos: 100.000 loss: 1.0739 (1.2241) acc1: 23.9583 (18.4734) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2599 data: 0.1631 max mem: 3531 +Test: [ 170/2387] eta: 0:09:00 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.2886 (1.1684) acc1: 91.9271 (22.8938) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1999 data: 0.1004 max mem: 3531 +Test: [ 180/2387] eta: 0:08:56 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.2674 (1.1186) acc1: 93.7500 (26.8085) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1749 data: 0.0743 max mem: 3531 +Test: [ 190/2387] eta: 0:08:46 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.2667 (1.0737) acc1: 94.0104 (30.3529) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.2049 data: 0.1067 max mem: 3531 +Test: [ 200/2387] eta: 0:08:40 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.2632 (1.0337) acc1: 94.0104 (33.5277) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1899 data: 0.0921 max mem: 3531 +Test: [ 210/2387] eta: 0:08:32 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.2667 (0.9975) acc1: 93.4896 (36.3781) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1949 data: 0.0966 max mem: 3532 +Test: [ 220/2387] eta: 0:08:23 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.2701 (0.9647) acc1: 93.4896 (38.9777) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1799 data: 0.0822 max mem: 3532 +Test: [ 230/2387] eta: 0:08:15 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.2689 (0.9346) acc1: 93.7500 (41.3454) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1699 data: 0.0726 max mem: 3532 +Test: [ 240/2387] eta: 0:08:09 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.2602 (0.9063) acc1: 94.5312 (43.5782) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1749 data: 0.0775 max mem: 3532 +Test: [ 250/2387] eta: 0:08:01 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.2296 (0.8742) acc1: 96.0938 (45.8012) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0780 max mem: 3532 +Test: [ 260/2387] eta: 0:07:55 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.0840 (0.8437) acc1: 100.0000 (47.8778) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0730 max mem: 3532 +Test: [ 270/2387] eta: 0:07:50 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.0754 (0.8154) acc1: 100.0000 (49.7992) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0830 max mem: 3532 +Test: [ 280/2387] eta: 0:07:45 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.0769 (0.7893) acc1: 100.0000 (51.5847) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0930 max mem: 3532 +Test: [ 290/2387] eta: 0:07:43 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.0792 (0.7649) acc1: 100.0000 (53.2485) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2049 data: 0.1079 max mem: 3532 +Test: [ 300/2387] eta: 0:07:39 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.0788 (0.7420) acc1: 100.0000 (54.8017) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2099 data: 0.1133 max mem: 3532 +Test: [ 310/2387] eta: 0:07:34 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.0782 (0.7227) acc1: 100.0000 (56.2416) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0904 max mem: 3532 +Test: [ 320/2387] eta: 0:07:30 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.2020 (0.7065) acc1: 100.0000 (57.6032) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0896 max mem: 3532 +Test: [ 330/2387] eta: 0:07:27 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.2020 (0.6912) acc1: 100.0000 (58.8825) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2049 data: 0.1080 max mem: 3532 +Test: [ 340/2387] eta: 0:07:24 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.2016 (0.6769) acc1: 100.0000 (60.0868) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2049 data: 0.1100 max mem: 3532 +Test: [ 350/2387] eta: 0:07:19 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1957 (0.6609) acc1: 100.0000 (61.2239) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0897 max mem: 3532 +Test: [ 360/2387] eta: 0:07:17 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1344 (0.6481) acc1: 99.7396 (62.2374) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1899 data: 0.0915 max mem: 3532 +Test: [ 370/2387] eta: 0:07:14 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.2085 (0.6364) acc1: 97.6562 (63.1893) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2089 data: 0.1130 max mem: 3532 +Test: [ 380/2387] eta: 0:07:06 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.2098 (0.6251) acc1: 97.6562 (64.1021) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1565 data: 0.0632 max mem: 3532 +Test: [ 390/2387] eta: 0:07:00 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.2089 (0.6144) acc1: 97.6562 (64.9556) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1211 data: 0.0281 max mem: 3532 +Test: [ 400/2387] eta: 0:06:56 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.2067 (0.6043) acc1: 97.6562 (65.7659) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1564 data: 0.0644 max mem: 3532 +Test: [ 410/2387] eta: 0:06:54 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.2040 (0.5947) acc1: 97.6562 (66.5450) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1847 data: 0.0878 max mem: 3532 +Test: [ 420/2387] eta: 0:06:51 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.2034 (0.5853) acc1: 97.9167 (67.2871) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1923 data: 0.0927 max mem: 3532 +Test: [ 430/2387] eta: 0:06:47 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1994 (0.5764) acc1: 97.9167 (68.0044) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1813 data: 0.0868 max mem: 3532 +Test: [ 440/2387] eta: 0:06:44 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1916 (0.5663) acc1: 98.4375 (68.7169) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1765 data: 0.0802 max mem: 3533 +Test: [ 450/2387] eta: 0:06:40 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.0905 (0.5557) acc1: 100.0000 (69.4106) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1737 data: 0.0765 max mem: 3533 +Test: [ 460/2387] eta: 0:06:37 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.0899 (0.5456) acc1: 100.0000 (70.0741) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1794 data: 0.0865 max mem: 3533 +Test: [ 470/2387] eta: 0:06:34 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.0899 (0.5359) acc1: 100.0000 (70.7095) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1897 data: 0.0960 max mem: 3533 +Test: [ 480/2387] eta: 0:06:32 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.0898 (0.5266) acc1: 100.0000 (71.3184) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1983 data: 0.1045 max mem: 3533 +Test: [ 490/2387] eta: 0:06:28 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.0899 (0.5177) acc1: 100.0000 (71.9026) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0907 max mem: 3533 +Test: [ 500/2387] eta: 0:06:26 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.0902 (0.5092) acc1: 100.0000 (72.4634) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0869 max mem: 3533 +Test: [ 510/2387] eta: 0:06:24 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.0897 (0.5012) acc1: 100.0000 (73.0023) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.1023 max mem: 3533 +Test: [ 520/2387] eta: 0:06:21 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1240 (0.4940) acc1: 100.0000 (73.5200) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.1028 max mem: 3533 +Test: [ 530/2387] eta: 0:06:18 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1264 (0.4871) acc1: 100.0000 (74.0187) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1781 data: 0.0848 max mem: 3533 +Test: [ 540/2387] eta: 0:06:15 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1277 (0.4804) acc1: 100.0000 (74.4989) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1758 data: 0.0822 max mem: 3533 +Test: [ 550/2387] eta: 0:06:13 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1267 (0.4740) acc1: 100.0000 (74.9617) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1880 data: 0.0967 max mem: 3533 +Test: [ 560/2387] eta: 0:06:10 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1267 (0.4678) acc1: 100.0000 (75.4076) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1768 data: 0.0849 max mem: 3533 +Test: [ 570/2387] eta: 0:06:07 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1266 (0.4618) acc1: 100.0000 (75.8378) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1804 data: 0.0884 max mem: 3533 +Test: [ 580/2387] eta: 0:06:05 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1267 (0.4561) acc1: 100.0000 (76.2537) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1954 data: 0.1007 max mem: 3533 +Test: [ 590/2387] eta: 0:06:03 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1268 (0.4505) acc1: 100.0000 (76.6555) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2046 data: 0.1109 max mem: 3533 +Test: [ 600/2387] eta: 0:05:59 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1261 (0.4451) acc1: 100.0000 (77.0435) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1769 data: 0.0866 max mem: 3533 +Test: [ 610/2387] eta: 0:05:57 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1427 (0.4409) acc1: 99.7396 (77.3936) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1634 data: 0.0726 max mem: 3533 +Test: [ 620/2387] eta: 0:05:54 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1848 (0.4368) acc1: 98.6979 (77.7392) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1802 data: 0.0893 max mem: 3533 +Test: [ 630/2387] eta: 0:05:51 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1842 (0.4327) acc1: 98.9583 (78.0734) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1775 data: 0.0868 max mem: 3533 +Test: [ 640/2387] eta: 0:05:49 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1866 (0.4290) acc1: 98.6979 (78.3891) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1793 data: 0.0882 max mem: 3533 +Test: [ 650/2387] eta: 0:05:47 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1888 (0.4253) acc1: 98.4375 (78.6958) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1843 data: 0.0905 max mem: 3533 +Test: [ 660/2387] eta: 0:05:44 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1863 (0.4217) acc1: 98.4375 (78.9961) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1832 data: 0.0900 max mem: 3533 +Test: [ 670/2387] eta: 0:05:43 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1854 (0.4182) acc1: 98.4375 (79.2885) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2042 data: 0.1139 max mem: 3534 +Test: [ 680/2387] eta: 0:05:39 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1844 (0.4147) acc1: 98.6979 (79.5839) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1782 data: 0.0880 max mem: 3534 +Test: [ 690/2387] eta: 0:05:38 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1861 (0.4140) acc1: 98.6979 (79.6359) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.1765 data: 0.0859 max mem: 3534 +Test: [ 700/2387] eta: 0:05:35 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.4257 (0.4143) acc1: 77.3438 (79.6006) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.2052 data: 0.1141 max mem: 3534 +Test: [ 710/2387] eta: 0:05:33 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.4272 (0.4145) acc1: 77.0833 (79.5663) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.1836 data: 0.0924 max mem: 3534 +Test: [ 720/2387] eta: 0:05:30 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.4269 (0.4147) acc1: 77.3438 (79.5401) acc5: 100.0000 (100.0000) acc1_neg: 77.083 acc5_neg: 100.000 time: 0.1696 data: 0.0778 max mem: 3534 +Test: [ 730/2387] eta: 0:05:28 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.4324 (0.4150) acc1: 77.0833 (79.5065) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.1703 data: 0.0781 max mem: 3534 +Test: [ 740/2387] eta: 0:05:25 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.4359 (0.4152) acc1: 77.3438 (79.4851) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.1742 data: 0.0830 max mem: 3534 +Test: [ 750/2387] eta: 0:05:23 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.4219 (0.4153) acc1: 77.8646 (79.4607) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.1788 data: 0.0871 max mem: 3534 +Test: [ 760/2387] eta: 0:05:21 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.4143 (0.4147) acc1: 78.1250 (79.4976) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1889 data: 0.0951 max mem: 3534 +Test: [ 770/2387] eta: 0:05:18 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.2821 (0.4122) acc1: 88.8021 (79.6780) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1652 data: 0.0703 max mem: 3534 +Test: [ 780/2387] eta: 0:05:15 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1539 (0.4088) acc1: 98.6979 (79.9206) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1456 data: 0.0499 max mem: 3534 +Test: [ 790/2387] eta: 0:05:12 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1224 (0.4055) acc1: 99.2188 (80.1425) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1486 data: 0.0523 max mem: 3534 +Test: [ 800/2387] eta: 0:05:09 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1215 (0.4019) acc1: 99.7396 (80.3884) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1544 data: 0.0572 max mem: 3534 +Test: [ 810/2387] eta: 0:05:07 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1282 (0.3986) acc1: 99.7396 (80.6190) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1710 data: 0.0759 max mem: 3534 +Test: [ 820/2387] eta: 0:05:05 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1394 (0.3955) acc1: 99.2188 (80.8449) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1800 data: 0.0887 max mem: 3534 +Test: [ 830/2387] eta: 0:05:02 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1705 (0.3933) acc1: 98.4375 (81.0388) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1738 data: 0.0803 max mem: 3534 +Test: [ 840/2387] eta: 0:05:00 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.2357 (0.3917) acc1: 96.8750 (81.2054) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1795 data: 0.0822 max mem: 3534 +Test: [ 850/2387] eta: 0:04:58 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.2598 (0.3903) acc1: 94.7917 (81.3528) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1749 data: 0.0798 max mem: 3534 +Test: [ 860/2387] eta: 0:04:55 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.2685 (0.3887) acc1: 95.3125 (81.5131) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1707 data: 0.0762 max mem: 3534 +Test: [ 870/2387] eta: 0:04:53 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.2445 (0.3871) acc1: 95.3125 (81.6707) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1749 data: 0.0779 max mem: 3534 +Test: [ 880/2387] eta: 0:04:51 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.2424 (0.3857) acc1: 95.3125 (81.8178) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1691 data: 0.0731 max mem: 3534 +Test: [ 890/2387] eta: 0:04:48 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.2019 (0.3834) acc1: 98.6979 (82.0000) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1722 data: 0.0772 max mem: 3534 +Test: [ 900/2387] eta: 0:04:47 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.2122 (0.3854) acc1: 96.6146 (81.8766) acc5: 100.0000 (100.0000) acc1_neg: 65.365 acc5_neg: 100.000 time: 0.1960 data: 0.1019 max mem: 3535 +Test: [ 910/2387] eta: 0:04:45 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.6313 (0.3882) acc1: 65.3646 (81.6945) acc5: 100.0000 (100.0000) acc1_neg: 66.927 acc5_neg: 100.000 time: 0.2043 data: 0.1125 max mem: 3535 +Test: [ 920/2387] eta: 0:04:43 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.6396 (0.3910) acc1: 64.3229 (81.4991) acc5: 100.0000 (100.0000) acc1_neg: 63.021 acc5_neg: 100.000 time: 0.1956 data: 0.1051 max mem: 3535 +Test: [ 930/2387] eta: 0:04:41 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.6472 (0.3937) acc1: 64.3229 (81.3258) acc5: 100.0000 (100.0000) acc1_neg: 63.542 acc5_neg: 100.000 time: 0.2032 data: 0.1121 max mem: 3535 +Test: [ 940/2387] eta: 0:04:39 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.6494 (0.3964) acc1: 65.1042 (81.1451) acc5: 100.0000 (100.0000) acc1_neg: 64.583 acc5_neg: 100.000 time: 0.1980 data: 0.1054 max mem: 3535 +Test: [ 950/2387] eta: 0:04:38 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.6459 (0.3989) acc1: 64.8438 (80.9745) acc5: 100.0000 (100.0000) acc1_neg: 64.844 acc5_neg: 100.000 time: 0.1970 data: 0.1026 max mem: 3535 +Test: [ 960/2387] eta: 0:04:36 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.6291 (0.4013) acc1: 65.1042 (80.8248) acc5: 100.0000 (100.0000) acc1_neg: 71.094 acc5_neg: 100.000 time: 0.2057 data: 0.1135 max mem: 3535 +Test: [ 970/2387] eta: 0:04:34 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.6291 (0.4038) acc1: 64.8438 (80.6616) acc5: 100.0000 (100.0000) acc1_neg: 66.927 acc5_neg: 100.000 time: 0.2091 data: 0.1179 max mem: 3535 +Test: [ 980/2387] eta: 0:04:33 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.5936 (0.4033) acc1: 66.4062 (80.7045) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.2148 data: 0.1208 max mem: 3535 +Test: [ 990/2387] eta: 0:04:31 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1842 (0.4008) acc1: 97.9167 (80.8600) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2063 data: 0.1131 max mem: 3535 +Test: [1000/2387] eta: 0:04:29 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.0993 (0.3979) acc1: 99.7396 (81.0395) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1873 data: 0.0951 max mem: 3535 +Test: [1010/2387] eta: 0:04:27 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1049 (0.3952) acc1: 99.7396 (81.2142) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1873 data: 0.0956 max mem: 3535 +Test: [1020/2387] eta: 0:04:25 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1160 (0.3927) acc1: 99.7396 (81.3786) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1930 data: 0.1021 max mem: 3535 +Test: [1030/2387] eta: 0:04:23 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1117 (0.3900) acc1: 100.0000 (81.5549) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1863 data: 0.0941 max mem: 3535 +Test: [1040/2387] eta: 0:04:20 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.0961 (0.3871) acc1: 100.0000 (81.7313) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1802 data: 0.0865 max mem: 3535 +Test: [1050/2387] eta: 0:04:18 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.0920 (0.3843) acc1: 100.0000 (81.9039) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1861 data: 0.0919 max mem: 3535 +Test: [1060/2387] eta: 0:04:16 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.0959 (0.3816) acc1: 100.0000 (82.0740) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1906 data: 0.0986 max mem: 3535 +Test: [1070/2387] eta: 0:04:15 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1068 (0.3792) acc1: 100.0000 (82.2372) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1921 data: 0.1007 max mem: 3535 +Test: [1080/2387] eta: 0:04:13 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1402 (0.3772) acc1: 99.4792 (82.3820) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2076 data: 0.1159 max mem: 3535 +Test: [1090/2387] eta: 0:04:11 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1754 (0.3757) acc1: 96.8750 (82.4989) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2007 data: 0.1084 max mem: 3535 +Test: [1100/2387] eta: 0:04:09 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1952 (0.3741) acc1: 96.0938 (82.6240) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1804 data: 0.0882 max mem: 3535 +Test: [1110/2387] eta: 0:04:07 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.2017 (0.3728) acc1: 96.0938 (82.7307) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1891 data: 0.0960 max mem: 3535 +Test: [1120/2387] eta: 0:04:05 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.2240 (0.3716) acc1: 95.5729 (82.8346) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1954 data: 0.1022 max mem: 3536 +Test: [1130/2387] eta: 0:04:03 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.2022 (0.3700) acc1: 96.0938 (82.9571) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1950 data: 0.1037 max mem: 3536 +Test: [1140/2387] eta: 0:04:01 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1708 (0.3683) acc1: 97.6562 (83.0914) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1990 data: 0.1081 max mem: 3536 +Test: [1150/2387] eta: 0:03:59 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1842 (0.3672) acc1: 97.9167 (83.1989) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.2105 data: 0.1194 max mem: 3536 +Test: [1160/2387] eta: 0:03:57 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.2355 (0.3659) acc1: 95.3125 (83.3091) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1898 data: 0.0962 max mem: 3536 +Test: [1170/2387] eta: 0:03:56 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.2252 (0.3646) acc1: 96.6146 (83.4270) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2212 data: 0.1232 max mem: 3536 +Test: [1180/2387] eta: 0:03:54 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.2252 (0.3634) acc1: 97.3958 (83.5358) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2349 data: 0.1367 max mem: 3536 +Test: [1190/2387] eta: 0:03:51 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1871 (0.3621) acc1: 97.3958 (83.6541) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1435 data: 0.0465 max mem: 3536 +Test: [1200/2387] eta: 0:03:49 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1714 (0.3604) acc1: 99.2188 (83.7841) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1469 data: 0.0488 max mem: 3536 +Test: [1210/2387] eta: 0:03:48 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1421 (0.3586) acc1: 99.7396 (83.9157) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2004 data: 0.1054 max mem: 3536 +Test: [1220/2387] eta: 0:03:46 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.0995 (0.3564) acc1: 100.0000 (84.0463) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2045 data: 0.1111 max mem: 3536 +Test: [1230/2387] eta: 0:03:44 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.0850 (0.3542) acc1: 100.0000 (84.1755) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1987 data: 0.1047 max mem: 3536 +Test: [1240/2387] eta: 0:03:42 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.0884 (0.3521) acc1: 100.0000 (84.3018) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1994 data: 0.1052 max mem: 3536 +Test: [1250/2387] eta: 0:03:40 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.0891 (0.3500) acc1: 100.0000 (84.4252) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2006 data: 0.1075 max mem: 3536 +Test: [1260/2387] eta: 0:03:38 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.0873 (0.3479) acc1: 100.0000 (84.5479) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2062 data: 0.1131 max mem: 3536 +Test: [1270/2387] eta: 0:03:37 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.0799 (0.3458) acc1: 100.0000 (84.6692) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2199 data: 0.1255 max mem: 3536 +Test: [1280/2387] eta: 0:03:34 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.0796 (0.3438) acc1: 100.0000 (84.7881) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1873 data: 0.0895 max mem: 3536 +Test: [1290/2387] eta: 0:03:33 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.0881 (0.3419) acc1: 100.0000 (84.8972) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2053 data: 0.1078 max mem: 3536 +Test: [1300/2387] eta: 0:03:31 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.0915 (0.3400) acc1: 99.7396 (85.0083) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2549 data: 0.1620 max mem: 3536 +Test: [1310/2387] eta: 0:03:29 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.0940 (0.3381) acc1: 99.7396 (85.1181) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1802 data: 0.0846 max mem: 3536 +Test: [1320/2387] eta: 0:03:27 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1029 (0.3364) acc1: 99.7396 (85.2249) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1501 data: 0.0543 max mem: 3536 +Test: [1330/2387] eta: 0:03:25 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1066 (0.3347) acc1: 99.7396 (85.3276) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1958 data: 0.1018 max mem: 3536 +Test: [1340/2387] eta: 0:03:23 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1403 (0.3339) acc1: 98.4375 (85.3872) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2150 data: 0.1223 max mem: 3536 +Test: [1350/2387] eta: 0:03:22 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.0666 (0.3319) acc1: 100.0000 (85.4953) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2325 data: 0.1410 max mem: 3537 +Test: [1360/2387] eta: 0:03:20 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.0597 (0.3299) acc1: 100.0000 (85.6019) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2186 data: 0.1239 max mem: 3537 +Test: [1370/2387] eta: 0:03:18 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.0568 (0.3279) acc1: 100.0000 (85.7069) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1922 data: 0.0981 max mem: 3537 +Test: [1380/2387] eta: 0:03:15 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.0584 (0.3260) acc1: 100.0000 (85.8104) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1673 data: 0.0753 max mem: 3537 +Test: [1390/2387] eta: 0:03:14 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.0632 (0.3241) acc1: 100.0000 (85.9124) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1776 data: 0.0844 max mem: 3537 +Test: [1400/2387] eta: 0:03:11 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.0624 (0.3222) acc1: 100.0000 (86.0130) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1935 data: 0.1002 max mem: 3537 +Test: [1410/2387] eta: 0:03:09 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.0608 (0.3210) acc1: 100.0000 (86.0949) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1699 data: 0.0763 max mem: 3537 +Test: [1420/2387] eta: 0:03:07 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1684 (0.3202) acc1: 99.2188 (86.1631) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1698 data: 0.0762 max mem: 3537 +Test: [1430/2387] eta: 0:03:05 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1745 (0.3192) acc1: 97.6562 (86.2380) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1722 data: 0.0795 max mem: 3537 +Test: [1440/2387] eta: 0:03:03 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1696 (0.3182) acc1: 98.1771 (86.3235) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1662 data: 0.0742 max mem: 3537 +Test: [1450/2387] eta: 0:03:01 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1696 (0.3172) acc1: 98.9583 (86.4052) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1582 data: 0.0667 max mem: 3537 +Test: [1460/2387] eta: 0:02:59 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1329 (0.3159) acc1: 99.2188 (86.4927) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1655 data: 0.0735 max mem: 3537 +Test: [1470/2387] eta: 0:02:57 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1089 (0.3144) acc1: 100.0000 (86.5840) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1841 data: 0.0920 max mem: 3537 +Test: [1480/2387] eta: 0:02:55 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1016 (0.3131) acc1: 100.0000 (86.6723) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1806 data: 0.0890 max mem: 3537 +Test: [1490/2387] eta: 0:02:53 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1146 (0.3117) acc1: 99.7396 (86.7600) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2030 data: 0.1104 max mem: 3537 +Test: [1500/2387] eta: 0:02:51 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1190 (0.3104) acc1: 99.7396 (86.8456) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1997 data: 0.1071 max mem: 3537 +Test: [1510/2387] eta: 0:02:49 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1149 (0.3092) acc1: 100.0000 (86.9290) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1958 data: 0.1045 max mem: 3537 +Test: [1520/2387] eta: 0:02:47 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1272 (0.3084) acc1: 100.0000 (87.0004) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2009 data: 0.1098 max mem: 3537 +Test: [1530/2387] eta: 0:02:45 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.2043 (0.3078) acc1: 97.6562 (87.0710) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1930 data: 0.1012 max mem: 3537 +Test: [1540/2387] eta: 0:02:43 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.2147 (0.3071) acc1: 98.4375 (87.1431) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1986 data: 0.1070 max mem: 3537 +Test: [1550/2387] eta: 0:02:41 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1979 (0.3064) acc1: 98.4375 (87.2141) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1901 data: 0.0987 max mem: 3537 +Test: [1560/2387] eta: 0:02:39 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1886 (0.3058) acc1: 97.6562 (87.2745) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1770 data: 0.0856 max mem: 3537 +Test: [1570/2387] eta: 0:02:37 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1755 (0.3048) acc1: 98.9583 (87.3468) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1653 data: 0.0748 max mem: 3537 +Test: [1580/2387] eta: 0:02:35 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1346 (0.3038) acc1: 99.7396 (87.4206) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1640 data: 0.0739 max mem: 3538 +Test: [1590/2387] eta: 0:02:33 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1199 (0.3026) acc1: 99.7396 (87.4977) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1766 data: 0.0846 max mem: 3538 +Test: [1600/2387] eta: 0:02:31 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1156 (0.3016) acc1: 99.7396 (87.5615) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1853 data: 0.0920 max mem: 3538 +Test: [1610/2387] eta: 0:02:29 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1209 (0.3005) acc1: 99.7396 (87.6340) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1903 data: 0.0984 max mem: 3538 +Test: [1620/2387] eta: 0:02:27 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1236 (0.2995) acc1: 99.4792 (87.7050) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1785 data: 0.0873 max mem: 3538 +Test: [1630/2387] eta: 0:02:25 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1480 (0.2987) acc1: 98.9583 (87.7647) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1553 data: 0.0640 max mem: 3538 +Test: [1640/2387] eta: 0:02:23 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1946 (0.2982) acc1: 97.1354 (87.8150) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1618 data: 0.0694 max mem: 3538 +Test: [1650/2387] eta: 0:02:21 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1887 (0.2975) acc1: 97.9167 (87.8713) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1697 data: 0.0774 max mem: 3538 +Test: [1660/2387] eta: 0:02:19 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1823 (0.2969) acc1: 97.9167 (87.9285) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1825 data: 0.0902 max mem: 3538 +Test: [1670/2387] eta: 0:02:17 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1908 (0.2963) acc1: 97.1354 (87.9873) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1819 data: 0.0894 max mem: 3538 +Test: [1680/2387] eta: 0:02:15 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1911 (0.2957) acc1: 97.3958 (88.0393) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1707 data: 0.0765 max mem: 3538 +Test: [1690/2387] eta: 0:02:13 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1983 (0.2952) acc1: 96.8750 (88.0849) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1849 data: 0.0902 max mem: 3538 +Test: [1700/2387] eta: 0:02:11 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1930 (0.2944) acc1: 97.3958 (88.1439) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1891 data: 0.0963 max mem: 3538 +Test: [1710/2387] eta: 0:02:09 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1308 (0.2934) acc1: 99.2188 (88.2102) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1771 data: 0.0856 max mem: 3538 +Test: [1720/2387] eta: 0:02:07 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1057 (0.2923) acc1: 99.7396 (88.2764) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1780 data: 0.0846 max mem: 3538 +Test: [1730/2387] eta: 0:02:05 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1085 (0.2913) acc1: 99.7396 (88.3408) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1827 data: 0.0873 max mem: 3538 +Test: [1740/2387] eta: 0:02:03 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1072 (0.2903) acc1: 99.7396 (88.4047) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1868 data: 0.0906 max mem: 3538 +Test: [1750/2387] eta: 0:02:02 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1016 (0.2892) acc1: 99.7396 (88.4678) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1902 data: 0.0954 max mem: 3538 +Test: [1760/2387] eta: 0:02:00 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.0965 (0.2881) acc1: 100.0000 (88.5331) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2105 data: 0.1198 max mem: 3538 +Test: [1770/2387] eta: 0:01:58 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1016 (0.2872) acc1: 100.0000 (88.5933) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2146 data: 0.1235 max mem: 3538 +Test: [1780/2387] eta: 0:01:56 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1132 (0.2862) acc1: 99.4792 (88.6556) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1976 data: 0.1041 max mem: 3538 +Test: [1790/2387] eta: 0:01:54 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1015 (0.2852) acc1: 100.0000 (88.7172) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1659 data: 0.0731 max mem: 3538 +Test: [1800/2387] eta: 0:01:52 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.0957 (0.2842) acc1: 99.7396 (88.7782) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1430 data: 0.0528 max mem: 3538 +Test: [1810/2387] eta: 0:01:50 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.0996 (0.2831) acc1: 99.7396 (88.8388) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1789 data: 0.0864 max mem: 3539 +Test: [1820/2387] eta: 0:01:48 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.0928 (0.2821) acc1: 100.0000 (88.8983) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1835 data: 0.0895 max mem: 3539 +Test: [1830/2387] eta: 0:01:46 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.0974 (0.2812) acc1: 99.7396 (88.9580) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1655 data: 0.0733 max mem: 3539 +Test: [1840/2387] eta: 0:01:44 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1306 (0.2803) acc1: 99.7396 (89.0172) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1659 data: 0.0747 max mem: 3539 +Test: [1850/2387] eta: 0:01:42 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1311 (0.2795) acc1: 100.0000 (89.0756) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1674 data: 0.0750 max mem: 3539 +Test: [1860/2387] eta: 0:01:40 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1308 (0.2788) acc1: 100.0000 (89.1336) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1823 data: 0.0898 max mem: 3539 +Test: [1870/2387] eta: 0:01:38 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1316 (0.2780) acc1: 99.7396 (89.1907) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1869 data: 0.0949 max mem: 3539 +Test: [1880/2387] eta: 0:01:36 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1323 (0.2772) acc1: 99.7396 (89.2472) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1890 data: 0.0959 max mem: 3539 +Test: [1890/2387] eta: 0:01:34 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1323 (0.2764) acc1: 99.7396 (89.3027) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1813 data: 0.0889 max mem: 3539 +Test: [1900/2387] eta: 0:01:32 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1347 (0.2757) acc1: 99.7396 (89.3580) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1759 data: 0.0842 max mem: 3539 +Test: [1910/2387] eta: 0:01:30 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1424 (0.2750) acc1: 100.0000 (89.4123) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1743 data: 0.0810 max mem: 3539 +Test: [1920/2387] eta: 0:01:28 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1471 (0.2746) acc1: 99.7396 (89.4529) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1688 data: 0.0741 max mem: 3539 +Test: [1930/2387] eta: 0:01:26 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1378 (0.2739) acc1: 99.7396 (89.5058) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1698 data: 0.0765 max mem: 3539 +Test: [1940/2387] eta: 0:01:25 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1378 (0.2734) acc1: 99.7396 (89.5222) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1651 data: 0.0738 max mem: 3539 +Test: [1950/2387] eta: 0:01:23 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1616 (0.2729) acc1: 98.9583 (89.5624) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1646 data: 0.0745 max mem: 3539 +Test: [1960/2387] eta: 0:01:21 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1678 (0.2724) acc1: 98.6979 (89.6001) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1774 data: 0.0859 max mem: 3539 +Test: [1970/2387] eta: 0:01:19 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1938 (0.2722) acc1: 98.6979 (89.6306) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1794 data: 0.0880 max mem: 3539 +Test: [1980/2387] eta: 0:01:17 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1822 (0.2717) acc1: 98.4375 (89.6692) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1713 data: 0.0815 max mem: 3539 +Test: [1990/2387] eta: 0:01:15 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1582 (0.2712) acc1: 99.2188 (89.7170) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1697 data: 0.0794 max mem: 3539 +Test: [2000/2387] eta: 0:01:13 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1558 (0.2706) acc1: 99.4792 (89.7654) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1793 data: 0.0885 max mem: 3539 +Test: [2010/2387] eta: 0:01:11 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1558 (0.2700) acc1: 99.4792 (89.8123) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1804 data: 0.0895 max mem: 3539 +Test: [2020/2387] eta: 0:01:09 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1559 (0.2695) acc1: 99.4792 (89.8595) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1742 data: 0.0833 max mem: 3539 +Test: [2030/2387] eta: 0:01:07 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1549 (0.2689) acc1: 99.2188 (89.9059) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1800 data: 0.0884 max mem: 3540 +Test: [2040/2387] eta: 0:01:05 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1559 (0.2684) acc1: 99.2188 (89.9514) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1819 data: 0.0902 max mem: 3540 +Test: [2050/2387] eta: 0:01:03 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1553 (0.2678) acc1: 99.2188 (89.9981) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1852 data: 0.0945 max mem: 3540 +Test: [2060/2387] eta: 0:01:01 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1431 (0.2672) acc1: 99.4792 (90.0444) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1840 data: 0.0937 max mem: 3540 +Test: [2070/2387] eta: 0:01:00 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1407 (0.2666) acc1: 99.7396 (90.0915) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1839 data: 0.0894 max mem: 3540 +Test: [2080/2387] eta: 0:00:58 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1396 (0.2660) acc1: 99.7396 (90.1373) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2087 data: 0.1111 max mem: 3540 +Test: [2090/2387] eta: 0:00:56 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1392 (0.2653) acc1: 99.7396 (90.1833) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1950 data: 0.0977 max mem: 3540 +Test: [2100/2387] eta: 0:00:54 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1365 (0.2647) acc1: 99.7396 (90.2285) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1724 data: 0.0780 max mem: 3540 +Test: [2110/2387] eta: 0:00:52 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1366 (0.2641) acc1: 99.7396 (90.2731) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1815 data: 0.0894 max mem: 3540 +Test: [2120/2387] eta: 0:00:50 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1420 (0.2637) acc1: 99.4792 (90.3079) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1714 data: 0.0753 max mem: 3540 +Test: [2130/2387] eta: 0:00:48 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1492 (0.2631) acc1: 99.2188 (90.3498) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1732 data: 0.0759 max mem: 3540 +Test: [2140/2387] eta: 0:00:46 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1333 (0.2627) acc1: 99.2188 (90.3804) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1796 data: 0.0855 max mem: 3540 +Test: [2150/2387] eta: 0:00:44 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1561 (0.2622) acc1: 98.6979 (90.4168) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1709 data: 0.0793 max mem: 3540 +Test: [2160/2387] eta: 0:00:42 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1526 (0.2617) acc1: 98.6979 (90.4547) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1775 data: 0.0862 max mem: 3540 +Test: [2170/2387] eta: 0:00:41 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1625 (0.2614) acc1: 98.4375 (90.4857) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1854 data: 0.0936 max mem: 3540 +Test: [2180/2387] eta: 0:00:39 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1999 (0.2626) acc1: 95.8333 (90.4021) acc5: 100.0000 (100.0000) acc1_neg: 37.760 acc5_neg: 100.000 time: 0.1802 data: 0.0892 max mem: 3540 +Test: [2190/2387] eta: 0:00:37 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.3930 (0.2635) acc1: 82.2917 (90.3359) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1746 data: 0.0843 max mem: 3540 +Test: [2200/2387] eta: 0:00:35 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1510 (0.2629) acc1: 98.1771 (90.3750) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1782 data: 0.0857 max mem: 3540 +Test: [2210/2387] eta: 0:00:33 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1380 (0.2624) acc1: 98.9583 (90.4128) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1839 data: 0.0913 max mem: 3540 +Test: [2220/2387] eta: 0:00:31 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1455 (0.2619) acc1: 99.4792 (90.4496) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1762 data: 0.0850 max mem: 3540 +Test: [2230/2387] eta: 0:00:29 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1400 (0.2613) acc1: 99.4792 (90.4824) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1676 data: 0.0755 max mem: 3540 +Test: [2240/2387] eta: 0:00:27 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1181 (0.2608) acc1: 99.4792 (90.5161) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1667 data: 0.0756 max mem: 3540 +Test: [2250/2387] eta: 0:00:25 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1349 (0.2603) acc1: 99.4792 (90.5493) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1670 data: 0.0768 max mem: 3540 +Test: [2260/2387] eta: 0:00:23 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1252 (0.2596) acc1: 100.0000 (90.5901) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1696 data: 0.0790 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.0996 (0.2589) acc1: 100.0000 (90.6315) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1714 data: 0.0806 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.0920 (0.2583) acc1: 100.0000 (90.6610) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1673 data: 0.0765 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.0941 (0.2577) acc1: 100.0000 (90.7008) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1718 data: 0.0807 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1052 (0.2570) acc1: 100.0000 (90.7392) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1729 data: 0.0819 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1104 (0.2565) acc1: 100.0000 (90.7753) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1766 data: 0.0860 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1285 (0.2561) acc1: 99.2188 (90.8060) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1672 data: 0.0760 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1642 (0.2558) acc1: 98.6979 (90.8348) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1468 data: 0.0547 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1726 (0.2555) acc1: 98.6979 (90.8662) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1431 data: 0.0510 max mem: 3541 +Test: [2350/2387] eta: 0:00:06 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1656 (0.2553) acc1: 99.2188 (90.8859) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1429 data: 0.0520 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1656 (0.2550) acc1: 98.6979 (90.9111) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1454 data: 0.0554 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.1826 (0.2548) acc1: 98.1771 (90.9329) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1430 data: 0.0535 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.2135 (0.2547) acc1: 96.8750 (90.9529) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1374 data: 0.0476 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 27.397 acc5_pos: 100.000 loss: 0.2302 (0.2562) acc1: 94.7917 (90.8734) acc5: 100.0000 (100.0000) acc1_neg: 20.930 acc5_neg: 100.000 time: 0.1265 data: 0.0392 max mem: 3541 +Test: Total time: 0:07:26 (0.1869 s / it) +* Overall Acc@1 90.873 Acc@5 100.000 + target == real Acc@1 16.816 Acc@5 100.000 + target == fake Acc@1 96.074 Acc@5 100.000 + loss 0.256 + PR-AUC 0.9879 ROC-AUC 0.8478 EER 22.66% (thr=0.8115) +Accuracy of the network on the 916267 test images: 90.9% +[rank0]:[W726 04:17:36.678848623 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:17:58.227000 3145136 torch/distributed/run.py:766] +W0726 04:17:58.227000 3145136 torch/distributed/run.py:766] ***************************************** +W0726 04:17:58.227000 3145136 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:17:58.227000 3145136 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:18:31.210547246 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:18:31.330648477 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank1]:[W726 04:18:31.334320523 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank6]:[W726 04:18:31.336605037 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:18:31.359323117 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +[rank0]:[W726 04:18:31.547277394 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:18:31.552516182 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:18:31.558202898 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n11_seed6_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed6_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:12:05 acc1_pos: 13.542 acc5_pos: 100.000 loss: 1.3069 (1.3069) acc1: 13.5417 (13.5417) acc5: 100.0000 (100.0000) time: 8.3674 data: 6.1478 max mem: 3531 +Test: [ 10/517] eta: 0:07:11 acc1_pos: 5.208 acc5_pos: 100.000 loss: 1.3841 (1.3325) acc1: 8.8542 (11.4347) acc5: 100.0000 (100.0000) time: 0.8507 data: 0.5592 max mem: 3531 +Test: [ 20/517] eta: 0:04:05 acc1_pos: 3.385 acc5_pos: 100.000 loss: 1.3815 (1.3484) acc1: 8.5938 (9.9578) acc5: 100.0000 (100.0000) time: 0.0995 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:03:42 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.4295 (1.3738) acc1: 6.2500 (9.1398) acc5: 100.0000 (100.0000) time: 0.2406 data: 0.1459 max mem: 3531 +Test: [ 40/517] eta: 0:03:28 acc1_pos: 11.458 acc5_pos: 100.000 loss: 1.4308 (1.3858) acc1: 5.7292 (8.6954) acc5: 100.0000 (100.0000) time: 0.3799 data: 0.2868 max mem: 3531 +Test: [ 50/517] eta: 0:03:12 acc1_pos: 11.009 acc5_pos: 100.000 loss: 1.3671 (1.3261) acc1: 9.3750 (13.4906) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.3442 data: 0.2484 max mem: 3531 +Test: [ 60/517] eta: 0:03:04 acc1_pos: 11.009 acc5_pos: 100.000 loss: 0.2899 (1.1486) acc1: 96.0938 (27.3011) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.3359 data: 0.2433 max mem: 3531 +Test: [ 70/517] eta: 0:02:49 acc1_pos: 11.009 acc5_pos: 100.000 loss: 0.2484 (1.0185) acc1: 98.1771 (37.3606) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2925 data: 0.2025 max mem: 3531 +Test: [ 80/517] eta: 0:02:39 acc1_pos: 11.009 acc5_pos: 100.000 loss: 0.2360 (0.9218) acc1: 99.2188 (45.0167) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2449 data: 0.1523 max mem: 3531 +Test: [ 90/517] eta: 0:02:31 acc1_pos: 11.009 acc5_pos: 100.000 loss: 0.2324 (0.8455) acc1: 99.2188 (50.9501) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2691 data: 0.1759 max mem: 3531 +Test: [100/517] eta: 0:02:23 acc1_pos: 11.009 acc5_pos: 100.000 loss: 0.2050 (0.7816) acc1: 99.7396 (55.7782) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2639 data: 0.1731 max mem: 3531 +Test: [110/517] eta: 0:02:17 acc1_pos: 11.009 acc5_pos: 100.000 loss: 0.1862 (0.7267) acc1: 99.7396 (59.6988) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2644 data: 0.1729 max mem: 3531 +Test: [120/517] eta: 0:02:11 acc1_pos: 11.009 acc5_pos: 100.000 loss: 0.1902 (0.6843) acc1: 99.7396 (63.0036) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2569 data: 0.1621 max mem: 3531 +Test: [130/517] eta: 0:02:05 acc1_pos: 11.009 acc5_pos: 100.000 loss: 0.2092 (0.6471) acc1: 99.7396 (65.8059) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2486 data: 0.1533 max mem: 3531 +Test: [140/517] eta: 0:02:01 acc1_pos: 11.009 acc5_pos: 100.000 loss: 0.1975 (0.6161) acc1: 100.0000 (68.2181) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2637 data: 0.1704 max mem: 3531 +Test: [150/517] eta: 0:01:56 acc1_pos: 11.009 acc5_pos: 100.000 loss: 0.2023 (0.5884) acc1: 100.0000 (70.3194) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2676 data: 0.1755 max mem: 3531 +Test: [160/517] eta: 0:01:51 acc1_pos: 11.009 acc5_pos: 100.000 loss: 0.2039 (0.5648) acc1: 99.7396 (72.1241) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2556 data: 0.1613 max mem: 3531 +Test: [170/517] eta: 0:01:47 acc1_pos: 11.009 acc5_pos: 100.000 loss: 0.1886 (0.5421) acc1: 99.7396 (73.7390) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2549 data: 0.1585 max mem: 3531 +Test: [180/517] eta: 0:01:43 acc1_pos: 11.009 acc5_pos: 100.000 loss: 0.1762 (0.5226) acc1: 99.7396 (75.1108) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2703 data: 0.1722 max mem: 3531 +Test: [190/517] eta: 0:01:40 acc1_pos: 11.009 acc5_pos: 100.000 loss: 0.1615 (0.5033) acc1: 100.0000 (76.4139) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2785 data: 0.1779 max mem: 3531 +Test: [200/517] eta: 0:01:36 acc1_pos: 11.009 acc5_pos: 100.000 loss: 0.1564 (0.4870) acc1: 100.0000 (77.5873) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2644 data: 0.1661 max mem: 3531 +Test: [210/517] eta: 0:01:32 acc1_pos: 11.009 acc5_pos: 100.000 loss: 0.1975 (0.4771) acc1: 100.0000 (78.4360) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2336 data: 0.1395 max mem: 3532 +Test: [220/517] eta: 0:01:28 acc1_pos: 11.009 acc5_pos: 100.000 loss: 0.2565 (0.4664) acc1: 96.3542 (79.3057) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2535 data: 0.1605 max mem: 3532 +Test: [230/517] eta: 0:01:25 acc1_pos: 11.009 acc5_pos: 100.000 loss: 0.2551 (0.4578) acc1: 98.4375 (80.1227) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2699 data: 0.1744 max mem: 3532 +Test: [240/517] eta: 0:01:21 acc1_pos: 11.009 acc5_pos: 100.000 loss: 0.2655 (0.4493) acc1: 98.9583 (80.9226) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2398 data: 0.1433 max mem: 3532 +Test: [250/517] eta: 0:01:18 acc1_pos: 11.009 acc5_pos: 100.000 loss: 0.2250 (0.4401) acc1: 99.7396 (81.6671) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2448 data: 0.1469 max mem: 3532 +Test: [260/517] eta: 0:01:14 acc1_pos: 11.009 acc5_pos: 100.000 loss: 0.2033 (0.4307) acc1: 99.7396 (82.3555) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2449 data: 0.1457 max mem: 3532 +Test: [270/517] eta: 0:01:11 acc1_pos: 11.009 acc5_pos: 100.000 loss: 0.2081 (0.4233) acc1: 98.9583 (82.9461) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2349 data: 0.1360 max mem: 3532 +Test: [280/517] eta: 0:01:07 acc1_pos: 11.009 acc5_pos: 100.000 loss: 0.2180 (0.4155) acc1: 98.6979 (83.5326) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2299 data: 0.1338 max mem: 3532 +Test: [290/517] eta: 0:01:04 acc1_pos: 11.009 acc5_pos: 100.000 loss: 0.2180 (0.4089) acc1: 99.7396 (84.0842) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2299 data: 0.1359 max mem: 3532 +Test: [300/517] eta: 0:01:01 acc1_pos: 11.009 acc5_pos: 100.000 loss: 0.2112 (0.4023) acc1: 99.7396 (84.6043) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2349 data: 0.1389 max mem: 3532 +Test: [310/517] eta: 0:00:58 acc1_pos: 11.009 acc5_pos: 100.000 loss: 0.2165 (0.3968) acc1: 99.7396 (85.0759) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2299 data: 0.1330 max mem: 3532 +Test: [320/517] eta: 0:00:55 acc1_pos: 11.009 acc5_pos: 100.000 loss: 0.2165 (0.3909) acc1: 99.4792 (85.5035) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2349 data: 0.1368 max mem: 3532 +Test: [330/517] eta: 0:00:52 acc1_pos: 11.009 acc5_pos: 100.000 loss: 0.1943 (0.3847) acc1: 98.9583 (85.9092) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2398 data: 0.1407 max mem: 3532 +Test: [340/517] eta: 0:00:49 acc1_pos: 11.009 acc5_pos: 100.000 loss: 0.1793 (0.3791) acc1: 99.2188 (86.3064) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2349 data: 0.1358 max mem: 3532 +Test: [350/517] eta: 0:00:46 acc1_pos: 11.009 acc5_pos: 100.000 loss: 0.2228 (0.3752) acc1: 99.4792 (86.6587) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2499 data: 0.1508 max mem: 3532 +Test: [360/517] eta: 0:00:43 acc1_pos: 11.009 acc5_pos: 100.000 loss: 0.2211 (0.3703) acc1: 99.4792 (87.0073) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1563 max mem: 3532 +Test: [370/517] eta: 0:00:40 acc1_pos: 11.009 acc5_pos: 100.000 loss: 0.1635 (0.3646) acc1: 100.0000 (87.3575) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2499 data: 0.1496 max mem: 3532 +Test: [380/517] eta: 0:00:37 acc1_pos: 11.009 acc5_pos: 100.000 loss: 0.1635 (0.3599) acc1: 100.0000 (87.6852) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2449 data: 0.1424 max mem: 3532 +Test: [390/517] eta: 0:00:34 acc1_pos: 11.009 acc5_pos: 100.000 loss: 0.1786 (0.3552) acc1: 100.0000 (88.0002) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2308 data: 0.1332 max mem: 3532 +Test: [400/517] eta: 0:00:31 acc1_pos: 11.009 acc5_pos: 100.000 loss: 0.1756 (0.3506) acc1: 100.0000 (88.2988) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2353 data: 0.1405 max mem: 3532 +Test: [410/517] eta: 0:00:29 acc1_pos: 11.009 acc5_pos: 100.000 loss: 0.1842 (0.3469) acc1: 100.0000 (88.5810) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2462 data: 0.1508 max mem: 3532 +Test: [420/517] eta: 0:00:26 acc1_pos: 11.009 acc5_pos: 100.000 loss: 0.1973 (0.3438) acc1: 100.0000 (88.8386) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2511 data: 0.1554 max mem: 3532 +Test: [430/517] eta: 0:00:23 acc1_pos: 11.009 acc5_pos: 100.000 loss: 0.2085 (0.3405) acc1: 99.7396 (89.0752) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2327 data: 0.1353 max mem: 3532 +Test: [440/517] eta: 0:00:20 acc1_pos: 11.009 acc5_pos: 100.000 loss: 0.2068 (0.3376) acc1: 99.4792 (89.3117) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2233 data: 0.1278 max mem: 3533 +Test: [450/517] eta: 0:00:17 acc1_pos: 11.009 acc5_pos: 100.000 loss: 0.2068 (0.3348) acc1: 99.4792 (89.5377) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2362 data: 0.1429 max mem: 3533 +Test: [460/517] eta: 0:00:15 acc1_pos: 11.009 acc5_pos: 100.000 loss: 0.2022 (0.3321) acc1: 99.4792 (89.7472) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2399 data: 0.1453 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 11.009 acc5_pos: 100.000 loss: 0.1949 (0.3292) acc1: 99.7396 (89.9576) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2396 data: 0.1406 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 11.009 acc5_pos: 100.000 loss: 0.1909 (0.3268) acc1: 99.7396 (90.1442) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2399 data: 0.1403 max mem: 3533 +Test: [490/517] eta: 0:00:07 acc1_pos: 11.009 acc5_pos: 100.000 loss: 0.2022 (0.3242) acc1: 99.4792 (90.3322) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2227 data: 0.1286 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 11.009 acc5_pos: 100.000 loss: 0.2416 (0.3237) acc1: 98.4375 (90.4233) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1927 data: 0.1027 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 11.009 acc5_pos: 100.000 loss: 0.2566 (0.3219) acc1: 95.5729 (90.5557) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1698 data: 0.0804 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 11.009 acc5_pos: 100.000 loss: 0.2036 (0.3203) acc1: 97.9167 (90.6505) acc5: 100.0000 (100.0000) acc1_neg: 98.940 acc5_neg: 100.000 time: 0.1689 data: 0.0778 max mem: 3533 +Test: Total time: 0:02:15 (0.2619 s / it) +* Overall Acc@1 90.650 Acc@5 100.000 + target == real Acc@1 9.340 Acc@5 100.000 + target == fake Acc@1 99.086 Acc@5 100.000 + loss 0.320 + PR-AUC 0.9567 ROC-AUC 0.7352 EER 32.42% (thr=0.8042) +Accuracy of the network on the 198427 test images: 90.7% +[rank0]:[W726 04:20:56.153498992 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 04:21:19.680000 3145708 torch/distributed/run.py:766] +W0726 04:21:19.680000 3145708 torch/distributed/run.py:766] ***************************************** +W0726 04:21:19.680000 3145708 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:21:19.680000 3145708 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:21:54.436970380 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:21:54.511697228 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:21:54.896264269 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:21:54.976082541 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:21:54.018362879 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:21:54.019839353 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:21:54.022616984 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:21:54.051936391 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n11_seed6_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed6_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:30:09 acc1_pos: 83.594 acc5_pos: 100.000 loss: 0.4315 (0.4315) acc1: 83.5938 (83.5938) acc5: 100.0000 (100.0000) time: 4.6523 data: 3.0093 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/389] eta: 0:03:37 acc1_pos: 78.385 acc5_pos: 100.000 loss: 0.4673 (0.4437) acc1: 82.0312 (81.6761) acc5: 100.0000 (100.0000) time: 0.5748 data: 0.3440 max mem: 3531 +Test: [ 20/389] eta: 0:02:17 acc1_pos: 80.469 acc5_pos: 100.000 loss: 0.4673 (0.4606) acc1: 80.2083 (79.6751) acc5: 100.0000 (100.0000) time: 0.1598 data: 0.0677 max mem: 3531 +Test: [ 30/389] eta: 0:01:57 acc1_pos: 76.302 acc5_pos: 100.000 loss: 0.5852 (0.5299) acc1: 74.2188 (72.6311) acc5: 100.0000 (100.0000) time: 0.1931 data: 0.0988 max mem: 3531 +Test: [ 40/389] eta: 0:01:40 acc1_pos: 82.812 acc5_pos: 100.000 loss: 0.5501 (0.5253) acc1: 76.0417 (74.9301) acc5: 100.0000 (100.0000) time: 0.1971 data: 0.1052 max mem: 3531 +Test: [ 50/389] eta: 0:01:27 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.4725 (0.6064) acc1: 82.8125 (70.6138) acc5: 100.0000 (100.0000) time: 0.1501 data: 0.0581 max mem: 3531 +Test: [ 60/389] eta: 0:01:20 acc1_pos: 71.615 acc5_pos: 100.000 loss: 0.5807 (0.6654) acc1: 55.7292 (66.6325) acc5: 100.0000 (100.0000) time: 0.1577 data: 0.0650 max mem: 3531 +Test: [ 70/389] eta: 0:01:14 acc1_pos: 92.708 acc5_pos: 100.000 loss: 0.3696 (0.6205) acc1: 89.8438 (70.3895) acc5: 100.0000 (100.0000) time: 0.1679 data: 0.0762 max mem: 3531 +Test: [ 80/389] eta: 0:01:09 acc1_pos: 99.479 acc5_pos: 100.000 loss: 0.4024 (0.6017) acc1: 89.5833 (72.3187) acc5: 100.0000 (100.0000) time: 0.1603 data: 0.0679 max mem: 3531 +Test: [ 90/389] eta: 0:01:05 acc1_pos: 93.490 acc5_pos: 100.000 loss: 0.3777 (0.5674) acc1: 89.8438 (74.6766) acc5: 100.0000 (100.0000) time: 0.1699 data: 0.0779 max mem: 3531 +Test: [100/389] eta: 0:01:01 acc1_pos: 53.906 acc5_pos: 100.000 loss: 0.3818 (0.5755) acc1: 91.1458 (73.2983) acc5: 100.0000 (100.0000) time: 0.1695 data: 0.0764 max mem: 3531 +Test: [110/389] eta: 0:00:58 acc1_pos: 99.479 acc5_pos: 100.000 loss: 0.6333 (0.5731) acc1: 64.5833 (72.9636) acc5: 100.0000 (100.0000) time: 0.1604 data: 0.0677 max mem: 3531 +Test: [120/389] eta: 0:00:54 acc1_pos: 87.500 acc5_pos: 100.000 loss: 0.4938 (0.5694) acc1: 69.2708 (73.0716) acc5: 100.0000 (100.0000) time: 0.1563 data: 0.0656 max mem: 3531 +Test: [130/389] eta: 0:00:52 acc1_pos: 92.638 acc5_pos: 100.000 loss: 0.5558 (0.5791) acc1: 69.2708 (71.9704) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.1615 data: 0.0671 max mem: 3531 +Test: [140/389] eta: 0:00:49 acc1_pos: 92.638 acc5_pos: 100.000 loss: 0.7527 (0.5988) acc1: 38.0208 (69.4297) acc5: 100.0000 (100.0000) acc1_neg: 17.708 acc5_neg: 100.000 time: 0.1691 data: 0.0734 max mem: 3531 +Test: [150/389] eta: 0:00:46 acc1_pos: 92.638 acc5_pos: 100.000 loss: 0.8403 (0.6094) acc1: 38.0208 (68.3188) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.1651 data: 0.0730 max mem: 3531 +Test: [160/389] eta: 0:00:44 acc1_pos: 92.638 acc5_pos: 100.000 loss: 0.6741 (0.6126) acc1: 50.5208 (68.2631) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1643 data: 0.0736 max mem: 3531 +Test: [170/389] eta: 0:00:42 acc1_pos: 92.638 acc5_pos: 100.000 loss: 0.4303 (0.6048) acc1: 88.8021 (68.9267) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1604 data: 0.0694 max mem: 3531 +Test: [180/389] eta: 0:00:39 acc1_pos: 92.638 acc5_pos: 100.000 loss: 0.4723 (0.6030) acc1: 80.7292 (69.3442) acc5: 100.0000 (100.0000) acc1_neg: 51.823 acc5_neg: 100.000 time: 0.1528 data: 0.0589 max mem: 3531 +Test: [190/389] eta: 0:00:37 acc1_pos: 92.638 acc5_pos: 100.000 loss: 0.6384 (0.6207) acc1: 64.3229 (67.5993) acc5: 100.0000 (100.0000) acc1_neg: 25.000 acc5_neg: 100.000 time: 0.1642 data: 0.0705 max mem: 3531 +Test: [200/389] eta: 0:00:35 acc1_pos: 92.638 acc5_pos: 100.000 loss: 0.8875 (0.6358) acc1: 23.1771 (65.5991) acc5: 100.0000 (100.0000) acc1_neg: 11.198 acc5_neg: 100.000 time: 0.1873 data: 0.0953 max mem: 3531 +Test: [210/389] eta: 0:00:33 acc1_pos: 92.638 acc5_pos: 100.000 loss: 0.7842 (0.6334) acc1: 24.4792 (65.6731) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1943 data: 0.1023 max mem: 3532 +Test: [220/389] eta: 0:00:32 acc1_pos: 92.638 acc5_pos: 100.000 loss: 0.3783 (0.6306) acc1: 88.5417 (65.9290) acc5: 100.0000 (100.0000) acc1_neg: 5.469 acc5_neg: 100.000 time: 0.1922 data: 0.1018 max mem: 3532 +Test: [230/389] eta: 0:00:30 acc1_pos: 92.638 acc5_pos: 100.000 loss: 0.6143 (0.6357) acc1: 64.0625 (65.2631) acc5: 100.0000 (100.0000) acc1_neg: 64.062 acc5_neg: 100.000 time: 0.1888 data: 0.0985 max mem: 3532 +Test: [240/389] eta: 0:00:28 acc1_pos: 92.638 acc5_pos: 100.000 loss: 0.6143 (0.6341) acc1: 64.0625 (65.3192) acc5: 100.0000 (100.0000) acc1_neg: 41.667 acc5_neg: 100.000 time: 0.1906 data: 0.0973 max mem: 3532 +Test: [250/389] eta: 0:00:26 acc1_pos: 92.638 acc5_pos: 100.000 loss: 0.7241 (0.6384) acc1: 46.0938 (64.4495) acc5: 100.0000 (100.0000) acc1_neg: 46.094 acc5_neg: 100.000 time: 0.1984 data: 0.1028 max mem: 3532 +Test: [260/389] eta: 0:00:24 acc1_pos: 92.638 acc5_pos: 100.000 loss: 0.7400 (0.6450) acc1: 44.2708 (63.9737) acc5: 100.0000 (100.0000) acc1_neg: 3.646 acc5_neg: 100.000 time: 0.1920 data: 0.0992 max mem: 3532 +Test: [270/389] eta: 0:00:22 acc1_pos: 92.638 acc5_pos: 100.000 loss: 0.8078 (0.6605) acc1: 42.7083 (62.8094) acc5: 100.0000 (100.0000) acc1_neg: 76.042 acc5_neg: 100.000 time: 0.1885 data: 0.0965 max mem: 3532 +Test: [280/389] eta: 0:00:20 acc1_pos: 92.638 acc5_pos: 100.000 loss: 0.7250 (0.6644) acc1: 42.7083 (62.6196) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1945 data: 0.1020 max mem: 3532 +Test: [290/389] eta: 0:00:18 acc1_pos: 92.638 acc5_pos: 100.000 loss: 0.5612 (0.6580) acc1: 72.6562 (63.2428) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1891 data: 0.0979 max mem: 3532 +Test: [300/389] eta: 0:00:16 acc1_pos: 92.638 acc5_pos: 100.000 loss: 0.7926 (0.6662) acc1: 50.0000 (61.9584) acc5: 100.0000 (100.0000) acc1_neg: 24.740 acc5_neg: 100.000 time: 0.1808 data: 0.0891 max mem: 3532 +Test: [310/389] eta: 0:00:14 acc1_pos: 92.638 acc5_pos: 100.000 loss: 0.9401 (0.6778) acc1: 20.0521 (60.7801) acc5: 100.0000 (100.0000) acc1_neg: 11.458 acc5_neg: 100.000 time: 0.1742 data: 0.0826 max mem: 3532 +Test: [320/389] eta: 0:00:13 acc1_pos: 92.638 acc5_pos: 100.000 loss: 0.9165 (0.6807) acc1: 32.0312 (60.3680) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.1718 data: 0.0810 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 92.638 acc5_pos: 100.000 loss: 1.1330 (0.6976) acc1: 10.1562 (58.8227) acc5: 100.0000 (100.0000) acc1_neg: 16.406 acc5_neg: 100.000 time: 0.1700 data: 0.0791 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 92.638 acc5_pos: 100.000 loss: 0.7300 (0.6907) acc1: 47.3958 (59.5583) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1605 data: 0.0688 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 92.638 acc5_pos: 100.000 loss: 0.5493 (0.6944) acc1: 69.7917 (59.3312) acc5: 100.0000 (100.0000) acc1_neg: 47.656 acc5_neg: 100.000 time: 0.1548 data: 0.0636 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 92.638 acc5_pos: 100.000 loss: 0.5493 (0.6892) acc1: 73.9583 (59.9117) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1572 data: 0.0666 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 92.638 acc5_pos: 100.000 loss: 0.4337 (0.6873) acc1: 90.1042 (60.1394) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1463 data: 0.0548 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 92.638 acc5_pos: 100.000 loss: 0.5800 (0.6871) acc1: 69.0104 (60.1508) acc5: 100.0000 (100.0000) acc1_neg: 47.135 acc5_neg: 100.000 time: 0.1340 data: 0.0427 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 92.638 acc5_pos: 100.000 loss: 0.5943 (0.6853) acc1: 67.7083 (60.3316) acc5: 100.0000 (100.0000) acc1_neg: 87.978 acc5_neg: 100.000 time: 0.1314 data: 0.0367 max mem: 3532 +Test: Total time: 0:01:10 (0.1822 s / it) +* Overall Acc@1 60.332 Acc@5 100.000 + target == real Acc@1 72.738 Acc@5 100.000 + target == fake Acc@1 54.565 Acc@5 100.000 + loss 0.685 + PR-AUC 0.7825 ROC-AUC 0.6860 EER 35.76% (thr=0.4373) +Accuracy of the network on the 149358 test images: 60.3% +[rank0]:[W726 04:23:17.965481883 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:23:39.738000 3146266 torch/distributed/run.py:766] +W0726 04:23:39.738000 3146266 torch/distributed/run.py:766] ***************************************** +W0726 04:23:39.738000 3146266 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:23:39.738000 3146266 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:24:15.531163715 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:24:15.977052717 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:24:16.249162347 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:24:16.345835143 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +[rank4]:[W726 04:24:16.487073255 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:24:16.501013048 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:24:16.644855557 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:24:16.693522266 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n11_seed6_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed6_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:34:38 acc1_pos: 63.802 acc5_pos: 100.000 loss: 0.6377 (0.6377) acc1: 63.8021 (63.8021) acc5: 100.0000 (100.0000) time: 6.2426 data: 4.7173 max mem: 3531 +Test: [ 10/333] eta: 0:03:29 acc1_pos: 78.125 acc5_pos: 100.000 loss: 0.4578 (0.4390) acc1: 72.9167 (76.0417) acc5: 100.0000 (100.0000) time: 0.6494 data: 0.4292 max mem: 3531 +Test: [ 20/333] eta: 0:02:07 acc1_pos: 95.833 acc5_pos: 100.000 loss: 0.3330 (0.3752) acc1: 85.4167 (81.3244) acc5: 100.0000 (100.0000) time: 0.1143 data: 0.0222 max mem: 3531 +Test: [ 30/333] eta: 0:01:56 acc1_pos: 49.219 acc5_pos: 100.000 loss: 0.3876 (0.5266) acc1: 73.6979 (69.9597) acc5: 100.0000 (100.0000) time: 0.2400 data: 0.1478 max mem: 3531 +Test: [ 40/333] eta: 0:01:46 acc1_pos: 40.104 acc5_pos: 100.000 loss: 0.8638 (0.6131) acc1: 43.2292 (63.2050) acc5: 100.0000 (100.0000) time: 0.3192 data: 0.2289 max mem: 3531 +Test: [ 50/333] eta: 0:01:37 acc1_pos: 47.135 acc5_pos: 100.000 loss: 0.8653 (0.6688) acc1: 40.1042 (59.0176) acc5: 100.0000 (100.0000) time: 0.2791 data: 0.1878 max mem: 3531 +Test: [ 60/333] eta: 0:01:26 acc1_pos: 30.208 acc5_pos: 100.000 loss: 0.8950 (0.7090) acc1: 38.8021 (55.5669) acc5: 100.0000 (100.0000) time: 0.2156 data: 0.1204 max mem: 3531 +Test: [ 70/333] eta: 0:01:17 acc1_pos: 41.667 acc5_pos: 100.000 loss: 0.8981 (0.7379) acc1: 41.1458 (53.5211) acc5: 100.0000 (100.0000) time: 0.1701 data: 0.0733 max mem: 3531 +Test: [ 80/333] eta: 0:01:10 acc1_pos: 43.490 acc5_pos: 100.000 loss: 0.8981 (0.7562) acc1: 42.1875 (52.0287) acc5: 100.0000 (100.0000) time: 0.1699 data: 0.0747 max mem: 3531 +Test: [ 90/333] eta: 0:01:04 acc1_pos: 58.073 acc5_pos: 100.000 loss: 0.8293 (0.7623) acc1: 46.6146 (51.6026) acc5: 100.0000 (100.0000) time: 0.1647 data: 0.0714 max mem: 3531 +Test: [100/333] eta: 0:01:00 acc1_pos: 54.688 acc5_pos: 100.000 loss: 0.7859 (0.7633) acc1: 49.2188 (51.5522) acc5: 100.0000 (100.0000) time: 0.1799 data: 0.0881 max mem: 3531 +Test: [110/333] eta: 0:00:56 acc1_pos: 53.385 acc5_pos: 100.000 loss: 0.7667 (0.7650) acc1: 51.5625 (51.4217) acc5: 100.0000 (100.0000) time: 0.1899 data: 0.0987 max mem: 3531 +Test: [120/333] eta: 0:00:52 acc1_pos: 58.333 acc5_pos: 100.000 loss: 0.7516 (0.7644) acc1: 49.2188 (51.3430) acc5: 100.0000 (100.0000) time: 0.1809 data: 0.0861 max mem: 3531 +Test: [130/333] eta: 0:00:48 acc1_pos: 51.302 acc5_pos: 100.000 loss: 0.7516 (0.7647) acc1: 48.9583 (51.3558) acc5: 100.0000 (100.0000) time: 0.1706 data: 0.0755 max mem: 3531 +Test: [140/333] eta: 0:00:45 acc1_pos: 61.979 acc5_pos: 100.000 loss: 0.7329 (0.7621) acc1: 51.8229 (51.5145) acc5: 100.0000 (100.0000) time: 0.1660 data: 0.0741 max mem: 3531 +Test: [150/333] eta: 0:00:42 acc1_pos: 55.208 acc5_pos: 100.000 loss: 0.7435 (0.7649) acc1: 51.5625 (51.2555) acc5: 100.0000 (100.0000) time: 0.1893 data: 0.0952 max mem: 3531 +Test: [160/333] eta: 0:00:39 acc1_pos: 43.421 acc5_pos: 100.000 loss: 0.7754 (0.7683) acc1: 49.7396 (51.2147) acc5: 100.0000 (100.0000) acc1_neg: 52.344 acc5_neg: 100.000 time: 0.1799 data: 0.0830 max mem: 3531 +Test: [170/333] eta: 0:00:36 acc1_pos: 43.421 acc5_pos: 100.000 loss: 0.7693 (0.7688) acc1: 52.8646 (51.4635) acc5: 100.0000 (100.0000) acc1_neg: 58.594 acc5_neg: 100.000 time: 0.1404 data: 0.0459 max mem: 3531 +Test: [180/333] eta: 0:00:33 acc1_pos: 43.421 acc5_pos: 100.000 loss: 0.7580 (0.7669) acc1: 57.5521 (51.8359) acc5: 100.0000 (100.0000) acc1_neg: 61.979 acc5_neg: 100.000 time: 0.1579 data: 0.0631 max mem: 3531 +Test: [190/333] eta: 0:00:31 acc1_pos: 43.421 acc5_pos: 100.000 loss: 0.6387 (0.7540) acc1: 61.9792 (53.1100) acc5: 100.0000 (100.0000) acc1_neg: 71.354 acc5_neg: 100.000 time: 0.1844 data: 0.0894 max mem: 3531 +Test: [200/333] eta: 0:00:28 acc1_pos: 43.421 acc5_pos: 100.000 loss: 0.5375 (0.7473) acc1: 71.3542 (53.8674) acc5: 100.0000 (100.0000) acc1_neg: 64.062 acc5_neg: 100.000 time: 0.1799 data: 0.0855 max mem: 3531 +Test: [210/333] eta: 0:00:26 acc1_pos: 43.421 acc5_pos: 100.000 loss: 0.6953 (0.7470) acc1: 61.7188 (54.1173) acc5: 100.0000 (100.0000) acc1_neg: 58.854 acc5_neg: 100.000 time: 0.1606 data: 0.0644 max mem: 3532 +Test: [220/333] eta: 0:00:23 acc1_pos: 43.421 acc5_pos: 100.000 loss: 0.7296 (0.7487) acc1: 58.5938 (54.2114) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.1504 data: 0.0562 max mem: 3532 +Test: [230/333] eta: 0:00:21 acc1_pos: 43.421 acc5_pos: 100.000 loss: 0.5284 (0.7375) acc1: 72.1354 (55.3312) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1742 data: 0.0800 max mem: 3532 +Test: [240/333] eta: 0:00:19 acc1_pos: 43.421 acc5_pos: 100.000 loss: 0.5205 (0.7311) acc1: 75.0000 (55.9593) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.1793 data: 0.0835 max mem: 3532 +Test: [250/333] eta: 0:00:17 acc1_pos: 43.421 acc5_pos: 100.000 loss: 0.5793 (0.7243) acc1: 69.7917 (56.5986) acc5: 100.0000 (100.0000) acc1_neg: 72.396 acc5_neg: 100.000 time: 0.1513 data: 0.0551 max mem: 3532 +Test: [260/333] eta: 0:00:14 acc1_pos: 43.421 acc5_pos: 100.000 loss: 0.5271 (0.7172) acc1: 72.3958 (57.3136) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.1504 data: 0.0569 max mem: 3532 +Test: [270/333] eta: 0:00:12 acc1_pos: 43.421 acc5_pos: 100.000 loss: 0.5881 (0.7130) acc1: 71.0938 (57.7529) acc5: 100.0000 (100.0000) acc1_neg: 67.969 acc5_neg: 100.000 time: 0.1835 data: 0.0893 max mem: 3532 +Test: [280/333] eta: 0:00:10 acc1_pos: 43.421 acc5_pos: 100.000 loss: 0.6189 (0.7137) acc1: 65.6250 (57.7393) acc5: 100.0000 (100.0000) acc1_neg: 56.250 acc5_neg: 100.000 time: 0.1794 data: 0.0816 max mem: 3532 +Test: [290/333] eta: 0:00:08 acc1_pos: 43.421 acc5_pos: 100.000 loss: 0.8160 (0.7181) acc1: 51.0417 (57.4143) acc5: 100.0000 (100.0000) acc1_neg: 52.604 acc5_neg: 100.000 time: 0.1599 data: 0.0625 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 43.421 acc5_pos: 100.000 loss: 0.8277 (0.7230) acc1: 48.6979 (57.1896) acc5: 100.0000 (100.0000) acc1_neg: 53.906 acc5_neg: 100.000 time: 0.1550 data: 0.0598 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 43.421 acc5_pos: 100.000 loss: 0.7629 (0.7174) acc1: 55.9896 (57.7338) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.1771 data: 0.0818 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 43.421 acc5_pos: 100.000 loss: 0.4983 (0.7119) acc1: 77.8646 (58.2895) acc5: 100.0000 (100.0000) acc1_neg: 52.344 acc5_neg: 100.000 time: 0.1898 data: 0.0946 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 43.421 acc5_pos: 100.000 loss: 0.8480 (0.7180) acc1: 52.3438 (58.0257) acc5: 100.0000 (100.0000) acc1_neg: 54.688 acc5_neg: 100.000 time: 0.1496 data: 0.0549 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 43.421 acc5_pos: 100.000 loss: 0.8688 (0.7193) acc1: 51.8229 (57.9735) acc5: 100.0000 (100.0000) acc1_neg: 49.577 acc5_neg: 100.000 time: 0.1574 data: 0.0548 max mem: 3532 +Test: Total time: 0:01:05 (0.1963 s / it) +* Overall Acc@1 57.973 Acc@5 100.000 + target == real Acc@1 51.261 Acc@5 100.000 + target == fake Acc@1 63.949 Acc@5 100.000 + loss 0.719 + PR-AUC 0.6279 ROC-AUC 0.6122 EER 42.34% (thr=0.5483) +Accuracy of the network on the 127843 test images: 58.0% +[rank0]:[W726 04:25:34.919575854 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:25:56.788000 3146824 torch/distributed/run.py:766] +W0726 04:25:56.788000 3146824 torch/distributed/run.py:766] ***************************************** +W0726 04:25:56.788000 3146824 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:25:56.788000 3146824 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:26:30.475497734 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:26:31.210917108 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:26:31.338720701 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:26:31.421571530 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:26:31.608996040 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:26:31.675058052 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:26:31.737307296 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:26:31.757421264 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n11_seed6_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed6_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:25:48 acc1_pos: 97.656 acc5_pos: 100.000 loss: 0.1608 (0.1608) acc1: 97.6562 (97.6562) acc5: 100.0000 (100.0000) time: 6.5885 data: 5.1736 max mem: 3531 +Test: [ 10/235] eta: 0:02:34 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.1660 (0.1705) acc1: 97.6562 (96.8987) acc5: 100.0000 (100.0000) time: 0.6875 data: 0.4707 max mem: 3531 +Test: [ 20/235] eta: 0:01:43 acc1_pos: 97.917 acc5_pos: 100.000 loss: 0.1661 (0.1685) acc1: 97.1354 (97.0858) acc5: 100.0000 (100.0000) time: 0.1736 data: 0.0725 max mem: 3531 +Test: [ 30/235] eta: 0:01:23 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.1661 (0.1683) acc1: 96.8750 (97.1102) acc5: 100.0000 (100.0000) time: 0.2503 data: 0.1523 max mem: 3531 +Test: [ 40/235] eta: 0:01:09 acc1_pos: 96.615 acc5_pos: 100.000 loss: 0.1649 (0.1684) acc1: 97.1354 (97.1354) acc5: 100.0000 (100.0000) time: 0.2237 data: 0.1322 max mem: 3531 +Test: [ 50/235] eta: 0:00:59 acc1_pos: 95.833 acc5_pos: 100.000 loss: 0.1666 (0.1695) acc1: 97.1354 (97.0741) acc5: 100.0000 (100.0000) time: 0.1954 data: 0.1008 max mem: 3531 +Test: [ 60/235] eta: 0:00:51 acc1_pos: 95.833 acc5_pos: 100.000 loss: 0.1756 (0.1711) acc1: 96.3542 (96.9390) acc5: 100.0000 (100.0000) time: 0.1676 data: 0.0704 max mem: 3531 +Test: [ 70/235] eta: 0:00:45 acc1_pos: 93.846 acc5_pos: 100.000 loss: 0.1872 (0.3626) acc1: 95.3125 (86.3043) acc5: 100.0000 (100.0000) acc1_neg: 9.115 acc5_neg: 100.000 time: 0.1551 data: 0.0610 max mem: 3531 +Test: [ 80/235] eta: 0:00:40 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.7036 (0.5283) acc1: 11.1979 (77.0062) acc5: 100.0000 (100.0000) acc1_neg: 10.417 acc5_neg: 100.000 time: 0.1587 data: 0.0674 max mem: 3531 +Test: [ 90/235] eta: 0:00:36 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.7038 (0.6580) acc1: 10.6771 (69.7859) acc5: 100.0000 (100.0000) acc1_neg: 9.896 acc5_neg: 100.000 time: 0.1601 data: 0.0691 max mem: 3531 +Test: [100/235] eta: 0:00:32 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.6893 (0.7602) acc1: 10.9375 (64.0728) acc5: 100.0000 (100.0000) acc1_neg: 10.417 acc5_neg: 100.000 time: 0.1725 data: 0.0779 max mem: 3531 +Test: [110/235] eta: 0:00:29 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.6893 (0.8439) acc1: 11.7188 (59.3985) acc5: 100.0000 (100.0000) acc1_neg: 13.802 acc5_neg: 100.000 time: 0.1686 data: 0.0738 max mem: 3531 +Test: [120/235] eta: 0:00:26 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.7291 (0.9211) acc1: 10.1562 (55.1072) acc5: 100.0000 (100.0000) acc1_neg: 5.729 acc5_neg: 100.000 time: 0.1883 data: 0.0946 max mem: 3531 +Test: [130/235] eta: 0:00:24 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.7755 (0.9862) acc1: 7.2917 (51.4671) acc5: 100.0000 (100.0000) acc1_neg: 7.552 acc5_neg: 100.000 time: 0.1999 data: 0.1056 max mem: 3531 +Test: [140/235] eta: 0:00:21 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.7987 (1.0438) acc1: 6.7708 (48.3452) acc5: 100.0000 (100.0000) acc1_neg: 5.990 acc5_neg: 100.000 time: 0.1987 data: 0.1063 max mem: 3531 +Test: [150/235] eta: 0:00:19 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.7893 (1.0925) acc1: 7.8125 (45.6695) acc5: 100.0000 (100.0000) acc1_neg: 7.812 acc5_neg: 100.000 time: 0.2084 data: 0.1156 max mem: 3531 +Test: [160/235] eta: 0:00:17 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.7764 (1.1344) acc1: 7.5521 (43.3036) acc5: 100.0000 (100.0000) acc1_neg: 9.635 acc5_neg: 100.000 time: 0.2125 data: 0.1205 max mem: 3531 +Test: [170/235] eta: 0:00:14 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.7123 (1.1618) acc1: 9.6354 (41.6865) acc5: 100.0000 (100.0000) acc1_neg: 14.062 acc5_neg: 100.000 time: 0.1867 data: 0.0939 max mem: 3531 +Test: [180/235] eta: 0:00:12 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.5781 (1.1842) acc1: 16.4062 (40.3732) acc5: 100.0000 (100.0000) acc1_neg: 16.406 acc5_neg: 100.000 time: 0.1544 data: 0.0618 max mem: 3531 +Test: [190/235] eta: 0:00:09 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.5529 (1.2042) acc1: 17.9688 (39.1743) acc5: 100.0000 (100.0000) acc1_neg: 17.448 acc5_neg: 100.000 time: 0.1498 data: 0.0589 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.5763 (1.2236) acc1: 17.1875 (38.0377) acc5: 100.0000 (100.0000) acc1_neg: 16.927 acc5_neg: 100.000 time: 0.1478 data: 0.0569 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.5763 (1.2399) acc1: 16.4062 (37.0532) acc5: 100.0000 (100.0000) acc1_neg: 17.708 acc5_neg: 100.000 time: 0.1557 data: 0.0645 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.5569 (1.2551) acc1: 16.4062 (36.1484) acc5: 100.0000 (100.0000) acc1_neg: 16.146 acc5_neg: 100.000 time: 0.1554 data: 0.0641 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.5807 (1.2688) acc1: 16.9271 (35.3434) acc5: 100.0000 (100.0000) acc1_neg: 14.844 acc5_neg: 100.000 time: 0.1402 data: 0.0496 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.5828 (1.2751) acc1: 16.1458 (35.0861) acc5: 100.0000 (100.0000) acc1_neg: 13.043 acc5_neg: 100.000 time: 0.1353 data: 0.0471 max mem: 3532 +Test: Total time: 0:00:47 (0.2029 s / it) +* Overall Acc@1 35.086 Acc@5 100.000 + target == real Acc@1 96.951 Acc@5 100.000 + target == fake Acc@1 12.711 Acc@5 100.000 + loss 1.275 + PR-AUC 0.8570 ROC-AUC 0.7062 EER 35.27% (thr=0.1278) +Accuracy of the network on the 89879 test images: 35.1% +[rank0]:[W726 04:27:30.256545282 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:27:51.414000 3147383 torch/distributed/run.py:766] +W0726 04:27:51.414000 3147383 torch/distributed/run.py:766] ***************************************** +W0726 04:27:51.414000 3147383 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:27:51.414000 3147383 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:28:25.635702931 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:28:25.895995000 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:28:25.014500627 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:28:26.194793083 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:28:26.344501046 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:28:26.365040770 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:28:26.382650958 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank4]:[W726 04:28:26.382672870 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n11_seed6_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed6_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:22:11 acc1_pos: 40.625 acc5_pos: 100.000 loss: 0.8924 (0.8924) acc1: 40.6250 (40.6250) acc5: 100.0000 (100.0000) time: 4.9498 data: 2.8515 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/269] eta: 0:02:17 acc1_pos: 42.969 acc5_pos: 100.000 loss: 0.8860 (0.8647) acc1: 41.4062 (42.5663) acc5: 100.0000 (100.0000) time: 0.5328 data: 0.2609 max mem: 3531 +Test: [ 20/269] eta: 0:01:23 acc1_pos: 57.031 acc5_pos: 100.000 loss: 0.8384 (0.8453) acc1: 41.6667 (43.6012) acc5: 100.0000 (100.0000) time: 0.1050 data: 0.0126 max mem: 3531 +Test: [ 30/269] eta: 0:01:07 acc1_pos: 42.969 acc5_pos: 100.000 loss: 0.8280 (0.8408) acc1: 41.6667 (44.6153) acc5: 100.0000 (100.0000) time: 0.1444 data: 0.0487 max mem: 3531 +Test: [ 40/269] eta: 0:01:03 acc1_pos: 6.897 acc5_pos: 100.000 loss: 0.6114 (0.7672) acc1: 58.8542 (51.6578) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.2149 data: 0.1185 max mem: 3531 +Test: [ 50/269] eta: 0:00:57 acc1_pos: 6.897 acc5_pos: 100.000 loss: 0.5171 (0.7227) acc1: 76.5625 (55.8926) acc5: 100.0000 (100.0000) acc1_neg: 62.240 acc5_neg: 100.000 time: 0.2262 data: 0.1321 max mem: 3531 +Test: [ 60/269] eta: 0:00:51 acc1_pos: 6.897 acc5_pos: 100.000 loss: 0.5383 (0.6981) acc1: 71.0938 (58.2309) acc5: 100.0000 (100.0000) acc1_neg: 71.094 acc5_neg: 100.000 time: 0.1821 data: 0.0907 max mem: 3531 +Test: [ 70/269] eta: 0:00:48 acc1_pos: 6.897 acc5_pos: 100.000 loss: 0.4947 (0.6629) acc1: 75.7812 (61.6050) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.1936 data: 0.1012 max mem: 3531 +Test: [ 80/269] eta: 0:00:44 acc1_pos: 6.897 acc5_pos: 100.000 loss: 0.4328 (0.6354) acc1: 84.3750 (64.2940) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.2098 data: 0.1175 max mem: 3531 +Test: [ 90/269] eta: 0:00:41 acc1_pos: 6.897 acc5_pos: 100.000 loss: 0.4403 (0.6256) acc1: 82.8125 (65.2244) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.1949 data: 0.1024 max mem: 3531 +Test: [100/269] eta: 0:00:38 acc1_pos: 6.897 acc5_pos: 100.000 loss: 0.4696 (0.6092) acc1: 79.1667 (66.6203) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.1901 data: 0.0977 max mem: 3531 +Test: [110/269] eta: 0:00:35 acc1_pos: 6.897 acc5_pos: 100.000 loss: 0.4988 (0.6025) acc1: 73.6979 (67.0233) acc5: 100.0000 (100.0000) acc1_neg: 72.656 acc5_neg: 100.000 time: 0.1999 data: 0.1084 max mem: 3531 +Test: [120/269] eta: 0:00:33 acc1_pos: 6.897 acc5_pos: 100.000 loss: 0.4768 (0.5892) acc1: 79.4271 (68.3131) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.1976 data: 0.1043 max mem: 3531 +Test: [130/269] eta: 0:00:30 acc1_pos: 6.897 acc5_pos: 100.000 loss: 0.4563 (0.5848) acc1: 80.9896 (68.6685) acc5: 100.0000 (100.0000) acc1_neg: 61.979 acc5_neg: 100.000 time: 0.2049 data: 0.1116 max mem: 3531 +Test: [140/269] eta: 0:00:28 acc1_pos: 6.897 acc5_pos: 100.000 loss: 0.4998 (0.5802) acc1: 75.5208 (69.1286) acc5: 100.0000 (100.0000) acc1_neg: 76.302 acc5_neg: 100.000 time: 0.2149 data: 0.1201 max mem: 3531 +Test: [150/269] eta: 0:00:26 acc1_pos: 6.897 acc5_pos: 100.000 loss: 0.4810 (0.5741) acc1: 76.8229 (69.7244) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.1959 data: 0.1020 max mem: 3531 +Test: [160/269] eta: 0:00:23 acc1_pos: 6.897 acc5_pos: 100.000 loss: 0.4680 (0.5684) acc1: 80.2083 (70.2575) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.2019 data: 0.1109 max mem: 3531 +Test: [170/269] eta: 0:00:21 acc1_pos: 6.897 acc5_pos: 100.000 loss: 0.5351 (0.5674) acc1: 71.6146 (70.3780) acc5: 100.0000 (100.0000) acc1_neg: 67.708 acc5_neg: 100.000 time: 0.2260 data: 0.1332 max mem: 3531 +Test: [180/269] eta: 0:00:19 acc1_pos: 6.897 acc5_pos: 100.000 loss: 0.5021 (0.5620) acc1: 78.3854 (71.0304) acc5: 100.0000 (100.0000) acc1_neg: 74.219 acc5_neg: 100.000 time: 0.2081 data: 0.1159 max mem: 3531 +Test: [190/269] eta: 0:00:17 acc1_pos: 6.897 acc5_pos: 100.000 loss: 0.5072 (0.5623) acc1: 77.0833 (70.9779) acc5: 100.0000 (100.0000) acc1_neg: 63.281 acc5_neg: 100.000 time: 0.1989 data: 0.1069 max mem: 3531 +Test: [200/269] eta: 0:00:14 acc1_pos: 6.897 acc5_pos: 100.000 loss: 0.5934 (0.5652) acc1: 67.9688 (70.6869) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2104 data: 0.1188 max mem: 3531 +Test: [210/269] eta: 0:00:12 acc1_pos: 6.897 acc5_pos: 100.000 loss: 0.4930 (0.5602) acc1: 72.6562 (71.0950) acc5: 100.0000 (100.0000) acc1_neg: 72.656 acc5_neg: 100.000 time: 0.2057 data: 0.1151 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 6.897 acc5_pos: 100.000 loss: 0.4607 (0.5553) acc1: 78.9062 (71.5062) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1967 data: 0.1045 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 6.897 acc5_pos: 100.000 loss: 0.4423 (0.5499) acc1: 79.4271 (71.9336) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1781 data: 0.0859 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 6.897 acc5_pos: 100.000 loss: 0.4548 (0.5476) acc1: 79.4271 (72.1635) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1735 data: 0.0826 max mem: 3532 +Test: [250/269] eta: 0:00:03 acc1_pos: 6.897 acc5_pos: 100.000 loss: 0.5194 (0.5529) acc1: 70.5729 (71.5959) acc5: 100.0000 (100.0000) acc1_neg: 52.083 acc5_neg: 100.000 time: 0.1571 data: 0.0668 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 6.897 acc5_pos: 100.000 loss: 0.6727 (0.5572) acc1: 55.7292 (71.1187) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.1258 data: 0.0360 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 6.897 acc5_pos: 100.000 loss: 0.5569 (0.5551) acc1: 67.9688 (71.3369) acc5: 100.0000 (100.0000) acc1_neg: 95.013 acc5_neg: 100.000 time: 0.1201 data: 0.0268 max mem: 3532 +Test: Total time: 0:00:54 (0.2032 s / it) +* Overall Acc@1 71.337 Acc@5 100.000 + target == real Acc@1 43.808 Acc@5 100.000 + target == fake Acc@1 75.094 Acc@5 100.000 + loss 0.555 + PR-AUC 0.9269 ROC-AUC 0.6512 EER 39.15% (thr=0.5991) +Accuracy of the network on the 103293 test images: 71.3% +[rank0]:[W726 04:29:30.336706894 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T04:29:37Z] eval_done tag=fixed500k_fake_n11_seed6_resampled +[2026-07-26T04:29:38Z] eval_start tag=fixed500k_real_n5_seed7_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed7_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T04:29:38Z +hostname=n116-040-168 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed7_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n5_seed7_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 04:29:53.100000 3147947 torch/distributed/run.py:766] +W0726 04:29:53.100000 3147947 torch/distributed/run.py:766] ***************************************** +W0726 04:29:53.100000 3147947 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:29:53.100000 3147947 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:30:26.124395375 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:30:27.622392550 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:30:27.680698859 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:30:27.927356644 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +| distributed init (rank 6): env:// +| distributed init (rank 3): env:// +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:30:28.341025577 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W726 04:30:28.341076661 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank4]:[W726 04:30:28.341074718 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:30:28.346186352 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n5_seed7_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed7_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:19:12 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.6800 (2.6800) acc1: 0.0000 (0.0000) acc5: 100.0000 (100.0000) time: 4.9271 data: 2.5052 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/234] eta: 0:01:59 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.5993 (2.5685) acc1: 0.0000 (0.6866) acc5: 100.0000 (100.0000) time: 0.5345 data: 0.2311 max mem: 3531 +Test: [ 20/234] eta: 0:01:19 acc1_pos: 2.865 acc5_pos: 100.000 loss: 2.5993 (2.5905) acc1: 0.0000 (0.9053) acc5: 100.0000 (100.0000) time: 0.1440 data: 0.0521 max mem: 3531 +Test: [ 30/234] eta: 0:01:13 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.5299 (2.4816) acc1: 0.2604 (4.5027) acc5: 100.0000 (100.0000) time: 0.2642 data: 0.1729 max mem: 3531 +Test: [ 40/234] eta: 0:01:02 acc1_pos: 3.385 acc5_pos: 100.000 loss: 2.5016 (2.4787) acc1: 0.2604 (4.0206) acc5: 100.0000 (100.0000) time: 0.2712 data: 0.1813 max mem: 3531 +Test: [ 50/234] eta: 0:00:54 acc1_pos: 1.823 acc5_pos: 100.000 loss: 2.2982 (2.4331) acc1: 3.3854 (4.1718) acc5: 100.0000 (100.0000) time: 0.1980 data: 0.1074 max mem: 3531 +Test: [ 60/234] eta: 0:00:49 acc1_pos: 15.862 acc5_pos: 100.000 loss: 0.7856 (2.0594) acc1: 8.5938 (19.3605) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1984 data: 0.1073 max mem: 3531 +Test: [ 70/234] eta: 0:00:43 acc1_pos: 15.862 acc5_pos: 100.000 loss: 0.0843 (1.7833) acc1: 100.0000 (30.5531) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1937 data: 0.0997 max mem: 3531 +Test: [ 80/234] eta: 0:00:39 acc1_pos: 15.862 acc5_pos: 100.000 loss: 0.0878 (1.5743) acc1: 99.2188 (39.0239) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1813 data: 0.0872 max mem: 3531 +Test: [ 90/234] eta: 0:00:35 acc1_pos: 15.862 acc5_pos: 100.000 loss: 0.0755 (1.4090) acc1: 100.0000 (45.7160) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1747 data: 0.0839 max mem: 3531 +Test: [100/234] eta: 0:00:32 acc1_pos: 15.862 acc5_pos: 100.000 loss: 0.0756 (1.2796) acc1: 99.7396 (50.9205) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1733 data: 0.0813 max mem: 3531 +Test: [110/234] eta: 0:00:29 acc1_pos: 15.862 acc5_pos: 100.000 loss: 0.0769 (1.1710) acc1: 99.7396 (55.3116) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1762 data: 0.0847 max mem: 3531 +Test: [120/234] eta: 0:00:26 acc1_pos: 15.862 acc5_pos: 100.000 loss: 0.0755 (1.0812) acc1: 100.0000 (58.9553) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1713 data: 0.0811 max mem: 3531 +Test: [130/234] eta: 0:00:23 acc1_pos: 15.862 acc5_pos: 100.000 loss: 0.0744 (1.0048) acc1: 100.0000 (62.0547) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1836 data: 0.0925 max mem: 3531 +Test: [140/234] eta: 0:00:20 acc1_pos: 15.862 acc5_pos: 100.000 loss: 0.0744 (0.9428) acc1: 100.0000 (64.5445) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1861 data: 0.0926 max mem: 3531 +Test: [150/234] eta: 0:00:18 acc1_pos: 15.862 acc5_pos: 100.000 loss: 0.0768 (0.8868) acc1: 99.7396 (66.8064) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1751 data: 0.0813 max mem: 3531 +Test: [160/234] eta: 0:00:16 acc1_pos: 15.862 acc5_pos: 100.000 loss: 0.0739 (0.8367) acc1: 99.7396 (68.8390) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1776 data: 0.0857 max mem: 3531 +Test: [170/234] eta: 0:00:13 acc1_pos: 15.862 acc5_pos: 100.000 loss: 0.0732 (0.7923) acc1: 100.0000 (70.6430) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1853 data: 0.0941 max mem: 3531 +Test: [180/234] eta: 0:00:11 acc1_pos: 15.862 acc5_pos: 100.000 loss: 0.0781 (0.7544) acc1: 99.7396 (72.1656) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1674 data: 0.0746 max mem: 3531 +Test: [190/234] eta: 0:00:09 acc1_pos: 15.862 acc5_pos: 100.000 loss: 0.0774 (0.7189) acc1: 99.7396 (73.6107) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1617 data: 0.0673 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 15.862 acc5_pos: 100.000 loss: 0.0750 (0.6868) acc1: 100.0000 (74.9223) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1727 data: 0.0799 max mem: 3531 +Test: [210/234] eta: 0:00:04 acc1_pos: 15.862 acc5_pos: 100.000 loss: 0.0738 (0.6579) acc1: 100.0000 (76.1021) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1637 data: 0.0731 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 15.862 acc5_pos: 100.000 loss: 0.0812 (0.6320) acc1: 99.7396 (77.1682) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1609 data: 0.0709 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 15.862 acc5_pos: 100.000 loss: 0.0774 (0.6078) acc1: 99.7396 (78.1543) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1460 data: 0.0559 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 15.862 acc5_pos: 100.000 loss: 0.0763 (0.6010) acc1: 100.0000 (78.4016) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1377 data: 0.0463 max mem: 3532 +Test: Total time: 0:00:46 (0.2005 s / it) +* Overall Acc@1 78.402 Acc@5 100.000 + target == real Acc@1 4.258 Acc@5 100.000 + target == fake Acc@1 99.299 Acc@5 100.000 + loss 0.601 + PR-AUC 0.8408 ROC-AUC 0.6306 EER 41.42% (thr=0.9292) +Accuracy of the network on the 89729 test images: 78.4% +[rank0]:[W726 04:31:25.695988588 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:31:46.305000 3148505 torch/distributed/run.py:766] +W0726 04:31:46.305000 3148505 torch/distributed/run.py:766] ***************************************** +W0726 04:31:46.305000 3148505 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:31:46.305000 3148505 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:32:20.059514472 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:32:21.236131655 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:32:21.329408833 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:32:21.728812012 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:32:21.755323266 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:32:21.809255265 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:32:21.899221647 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:32:21.925639929 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n5_seed7_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed7_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:05:37 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.6643 (2.6643) acc1: 0.0000 (0.0000) acc5: 100.0000 (100.0000) time: 6.1740 data: 4.2133 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/2387] eta: 0:25:28 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.6643 (2.6664) acc1: 0.0000 (0.0000) acc5: 100.0000 (100.0000) time: 0.6431 data: 0.3834 max mem: 3531 +Test: [ 20/2387] eta: 0:15:18 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.6411 (2.6490) acc1: 0.0000 (0.0000) acc5: 100.0000 (100.0000) time: 0.0986 data: 0.0068 max mem: 3531 +Test: [ 30/2387] eta: 0:13:21 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.6411 (2.6596) acc1: 0.0000 (0.0000) acc5: 100.0000 (100.0000) time: 0.1736 data: 0.0796 max mem: 3531 +Test: [ 40/2387] eta: 0:12:44 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.6889 (2.6687) acc1: 0.0000 (0.0000) acc5: 100.0000 (100.0000) time: 0.2602 data: 0.1644 max mem: 3531 +Test: [ 50/2387] eta: 0:11:43 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.6978 (2.6756) acc1: 0.0000 (0.0000) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1445 max mem: 3531 +Test: [ 60/2387] eta: 0:12:25 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.6983 (2.6800) acc1: 0.0000 (0.0000) acc5: 100.0000 (100.0000) time: 0.3097 data: 0.2161 max mem: 3531 +Test: [ 70/2387] eta: 0:11:56 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.7123 (2.6877) acc1: 0.0000 (0.0000) acc5: 100.0000 (100.0000) time: 0.3299 data: 0.2343 max mem: 3531 +Test: [ 80/2387] eta: 0:11:58 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.6921 (2.6822) acc1: 0.0000 (0.0000) acc5: 100.0000 (100.0000) time: 0.2849 data: 0.1893 max mem: 3531 +Test: [ 90/2387] eta: 0:12:35 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.6266 (2.6761) acc1: 0.0000 (0.0000) acc5: 100.0000 (100.0000) time: 0.3999 data: 0.3039 max mem: 3531 +Test: [ 100/2387] eta: 0:11:40 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.6496 (2.6735) acc1: 0.0000 (0.0000) acc5: 100.0000 (100.0000) time: 0.2849 data: 0.1861 max mem: 3531 +Test: [ 110/2387] eta: 0:11:30 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.6496 (2.6706) acc1: 0.0000 (0.0047) acc5: 100.0000 (100.0000) time: 0.1849 data: 0.0863 max mem: 3531 +Test: [ 120/2387] eta: 0:11:20 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.6421 (2.6691) acc1: 0.0000 (0.0043) acc5: 100.0000 (100.0000) time: 0.2699 data: 0.1722 max mem: 3531 +Test: [ 130/2387] eta: 0:11:08 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.6311 (2.6656) acc1: 0.0000 (0.0040) acc5: 100.0000 (100.0000) time: 0.2571 data: 0.1601 max mem: 3531 +Test: [ 140/2387] eta: 0:10:52 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.6377 (2.6642) acc1: 0.0000 (0.0055) acc5: 100.0000 (100.0000) time: 0.2308 data: 0.1357 max mem: 3531 +Test: [ 150/2387] eta: 0:10:41 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.6444 (2.6626) acc1: 0.0000 (0.0052) acc5: 100.0000 (100.0000) time: 0.2277 data: 0.1307 max mem: 3531 +Test: [ 160/2387] eta: 0:10:28 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.6220 (2.5906) acc1: 0.0000 (2.7562) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2240 data: 0.1272 max mem: 3531 +Test: [ 170/2387] eta: 0:10:13 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0710 (2.4433) acc1: 100.0000 (8.4430) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1999 data: 0.1061 max mem: 3531 +Test: [ 180/2387] eta: 0:10:02 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0709 (2.3122) acc1: 100.0000 (13.5014) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2001 data: 0.1069 max mem: 3531 +Test: [ 190/2387] eta: 0:09:51 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0708 (2.1948) acc1: 100.0000 (18.0301) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2049 data: 0.1104 max mem: 3531 +Test: [ 200/2387] eta: 0:09:39 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0708 (2.0892) acc1: 100.0000 (22.1082) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1898 data: 0.0946 max mem: 3531 +Test: [ 210/2387] eta: 0:09:27 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0707 (1.9935) acc1: 100.0000 (25.7998) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0839 max mem: 3532 +Test: [ 220/2387] eta: 0:09:16 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0705 (1.9065) acc1: 100.0000 (29.1572) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1784 data: 0.0823 max mem: 3532 +Test: [ 230/2387] eta: 0:09:05 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0705 (1.8270) acc1: 100.0000 (32.2240) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1698 data: 0.0770 max mem: 3532 +Test: [ 240/2387] eta: 0:08:54 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0713 (1.7542) acc1: 100.0000 (35.0363) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1613 data: 0.0678 max mem: 3532 +Test: [ 250/2387] eta: 0:08:44 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0727 (1.6872) acc1: 100.0000 (37.6245) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1599 data: 0.0661 max mem: 3532 +Test: [ 260/2387] eta: 0:08:39 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0725 (1.6254) acc1: 100.0000 (40.0144) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0892 max mem: 3532 +Test: [ 270/2387] eta: 0:08:33 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0734 (1.5682) acc1: 100.0000 (42.2279) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2049 data: 0.1078 max mem: 3532 +Test: [ 280/2387] eta: 0:08:21 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0741 (1.5149) acc1: 100.0000 (44.2838) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1610 data: 0.0659 max mem: 3532 +Test: [ 290/2387] eta: 0:08:16 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0718 (1.4653) acc1: 100.0000 (46.1985) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1579 data: 0.0638 max mem: 3532 +Test: [ 300/2387] eta: 0:08:11 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0708 (1.4190) acc1: 100.0000 (47.9859) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1989 data: 0.1059 max mem: 3532 +Test: [ 310/2387] eta: 0:08:05 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0721 (1.3758) acc1: 100.0000 (49.6567) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1888 data: 0.0956 max mem: 3532 +Test: [ 320/2387] eta: 0:07:58 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0770 (1.3353) acc1: 100.0000 (51.2250) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1714 data: 0.0794 max mem: 3532 +Test: [ 330/2387] eta: 0:07:53 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0774 (1.2973) acc1: 100.0000 (52.6986) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1729 data: 0.0826 max mem: 3532 +Test: [ 340/2387] eta: 0:07:46 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0774 (1.2616) acc1: 100.0000 (54.0857) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0787 max mem: 3532 +Test: [ 350/2387] eta: 0:07:42 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0768 (1.2276) acc1: 100.0000 (55.3938) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1800 data: 0.0878 max mem: 3532 +Test: [ 360/2387] eta: 0:07:36 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0704 (1.1956) acc1: 100.0000 (56.6294) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1797 data: 0.0861 max mem: 3532 +Test: [ 370/2387] eta: 0:07:30 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0704 (1.1653) acc1: 100.0000 (57.7985) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1555 data: 0.0624 max mem: 3532 +Test: [ 380/2387] eta: 0:07:25 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0703 (1.1365) acc1: 100.0000 (58.9061) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1618 data: 0.0694 max mem: 3532 +Test: [ 390/2387] eta: 0:07:19 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0704 (1.1093) acc1: 100.0000 (59.9571) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1568 data: 0.0617 max mem: 3532 +Test: [ 400/2387] eta: 0:07:15 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0707 (1.0834) acc1: 100.0000 (60.9557) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1628 data: 0.0681 max mem: 3532 +Test: [ 410/2387] eta: 0:07:11 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0706 (1.0587) acc1: 100.0000 (61.9057) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1829 data: 0.0873 max mem: 3532 +Test: [ 420/2387] eta: 0:07:06 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0704 (1.0352) acc1: 100.0000 (62.8105) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1712 data: 0.0728 max mem: 3532 +Test: [ 430/2387] eta: 0:07:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0708 (1.0129) acc1: 100.0000 (63.6734) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1460 data: 0.0519 max mem: 3532 +Test: [ 440/2387] eta: 0:06:56 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0719 (0.9916) acc1: 100.0000 (64.4971) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1507 data: 0.0605 max mem: 3533 +Test: [ 450/2387] eta: 0:06:52 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0751 (0.9713) acc1: 100.0000 (65.2843) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1731 data: 0.0811 max mem: 3533 +Test: [ 460/2387] eta: 0:06:49 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0752 (0.9518) acc1: 100.0000 (66.0374) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1818 data: 0.0889 max mem: 3533 +Test: [ 470/2387] eta: 0:06:45 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0753 (0.9332) acc1: 100.0000 (66.7584) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0884 max mem: 3533 +Test: [ 480/2387] eta: 0:06:40 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0753 (0.9154) acc1: 100.0000 (67.4495) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1618 data: 0.0678 max mem: 3533 +Test: [ 490/2387] eta: 0:06:37 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0752 (0.8983) acc1: 100.0000 (68.1125) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1698 data: 0.0759 max mem: 3533 +Test: [ 500/2387] eta: 0:06:34 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0753 (0.8818) acc1: 100.0000 (68.7490) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1790 data: 0.0885 max mem: 3533 +Test: [ 510/2387] eta: 0:06:30 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0753 (0.8661) acc1: 100.0000 (69.3605) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1709 data: 0.0805 max mem: 3533 +Test: [ 520/2387] eta: 0:06:27 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0797 (0.8510) acc1: 100.0000 (69.9486) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1745 data: 0.0837 max mem: 3533 +Test: [ 530/2387] eta: 0:06:25 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0811 (0.8365) acc1: 100.0000 (70.5146) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1865 data: 0.0945 max mem: 3533 +Test: [ 540/2387] eta: 0:06:22 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0818 (0.8226) acc1: 100.0000 (71.0596) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1869 data: 0.0944 max mem: 3533 +Test: [ 550/2387] eta: 0:06:19 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0818 (0.8091) acc1: 100.0000 (71.5848) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1861 data: 0.0933 max mem: 3533 +Test: [ 560/2387] eta: 0:06:16 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0817 (0.7962) acc1: 100.0000 (72.0913) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1897 data: 0.0941 max mem: 3533 +Test: [ 570/2387] eta: 0:06:14 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0817 (0.7836) acc1: 100.0000 (72.5801) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1861 data: 0.0915 max mem: 3533 +Test: [ 580/2387] eta: 0:06:11 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0819 (0.7716) acc1: 100.0000 (73.0520) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1811 data: 0.0887 max mem: 3533 +Test: [ 590/2387] eta: 0:06:08 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0817 (0.7599) acc1: 100.0000 (73.5080) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1802 data: 0.0886 max mem: 3533 +Test: [ 600/2387] eta: 0:06:05 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0807 (0.7486) acc1: 100.0000 (73.9488) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1864 data: 0.0948 max mem: 3533 +Test: [ 610/2387] eta: 0:06:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0707 (0.7375) acc1: 100.0000 (74.3752) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1864 data: 0.0949 max mem: 3533 +Test: [ 620/2387] eta: 0:06:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0698 (0.7267) acc1: 100.0000 (74.7878) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1760 data: 0.0857 max mem: 3533 +Test: [ 630/2387] eta: 0:05:57 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0698 (0.7163) acc1: 100.0000 (75.1874) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1704 data: 0.0795 max mem: 3533 +Test: [ 640/2387] eta: 0:05:55 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0698 (0.7062) acc1: 100.0000 (75.5745) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1938 data: 0.0996 max mem: 3533 +Test: [ 650/2387] eta: 0:05:52 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0700 (0.6964) acc1: 100.0000 (75.9497) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1927 data: 0.0990 max mem: 3533 +Test: [ 660/2387] eta: 0:05:50 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0700 (0.6870) acc1: 100.0000 (76.3135) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1784 data: 0.0863 max mem: 3533 +Test: [ 670/2387] eta: 0:05:47 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0700 (0.6778) acc1: 100.0000 (76.6665) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1790 data: 0.0866 max mem: 3534 +Test: [ 680/2387] eta: 0:05:44 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0695 (0.6688) acc1: 100.0000 (77.0091) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1774 data: 0.0860 max mem: 3534 +Test: [ 690/2387] eta: 0:05:43 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0698 (0.6602) acc1: 100.0000 (77.3419) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1953 data: 0.1015 max mem: 3534 +Test: [ 700/2387] eta: 0:05:41 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0743 (0.6518) acc1: 100.0000 (77.6651) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2080 data: 0.1122 max mem: 3534 +Test: [ 710/2387] eta: 0:05:38 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0743 (0.6437) acc1: 100.0000 (77.9792) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1894 data: 0.0965 max mem: 3534 +Test: [ 720/2387] eta: 0:05:35 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0747 (0.6358) acc1: 100.0000 (78.2846) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1677 data: 0.0772 max mem: 3534 +Test: [ 730/2387] eta: 0:05:32 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0746 (0.6282) acc1: 100.0000 (78.5817) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1696 data: 0.0773 max mem: 3534 +Test: [ 740/2387] eta: 0:05:29 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0744 (0.6207) acc1: 100.0000 (78.8708) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1667 data: 0.0740 max mem: 3534 +Test: [ 750/2387] eta: 0:05:27 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0745 (0.6134) acc1: 100.0000 (79.1521) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0781 max mem: 3534 +Test: [ 760/2387] eta: 0:05:24 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0745 (0.6063) acc1: 100.0000 (79.4261) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1623 data: 0.0702 max mem: 3534 +Test: [ 770/2387] eta: 0:05:21 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0740 (0.5994) acc1: 100.0000 (79.6929) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1556 data: 0.0625 max mem: 3534 +Test: [ 780/2387] eta: 0:05:18 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0721 (0.5926) acc1: 100.0000 (79.9529) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1605 data: 0.0674 max mem: 3534 +Test: [ 790/2387] eta: 0:05:16 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0691 (0.5860) acc1: 100.0000 (80.2064) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1642 data: 0.0716 max mem: 3534 +Test: [ 800/2387] eta: 0:05:14 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0712 (0.5796) acc1: 100.0000 (80.4535) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1796 data: 0.0874 max mem: 3534 +Test: [ 810/2387] eta: 0:05:11 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0703 (0.5733) acc1: 100.0000 (80.6945) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1901 data: 0.0977 max mem: 3534 +Test: [ 820/2387] eta: 0:05:10 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0680 (0.5672) acc1: 100.0000 (80.9296) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2069 data: 0.1125 max mem: 3534 +Test: [ 830/2387] eta: 0:05:08 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0700 (0.5612) acc1: 100.0000 (81.1588) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2097 data: 0.1142 max mem: 3534 +Test: [ 840/2387] eta: 0:05:06 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0708 (0.5554) acc1: 100.0000 (81.3828) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1973 data: 0.1024 max mem: 3534 +Test: [ 850/2387] eta: 0:05:04 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0727 (0.5497) acc1: 100.0000 (81.6016) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1961 data: 0.1034 max mem: 3534 +Test: [ 860/2387] eta: 0:05:02 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0740 (0.5442) acc1: 100.0000 (81.8153) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1890 data: 0.0966 max mem: 3534 +Test: [ 870/2387] eta: 0:04:59 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0744 (0.5388) acc1: 100.0000 (82.0241) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1787 data: 0.0866 max mem: 3534 +Test: [ 880/2387] eta: 0:04:57 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0704 (0.5335) acc1: 100.0000 (82.2281) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1784 data: 0.0879 max mem: 3534 +Test: [ 890/2387] eta: 0:04:55 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0686 (0.5282) acc1: 100.0000 (82.4276) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0932 max mem: 3534 +Test: [ 900/2387] eta: 0:04:52 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0697 (0.5232) acc1: 100.0000 (82.6226) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0803 max mem: 3535 +Test: [ 910/2387] eta: 0:04:50 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0796 (0.5184) acc1: 100.0000 (82.8134) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1653 data: 0.0701 max mem: 3535 +Test: [ 920/2387] eta: 0:04:48 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0794 (0.5136) acc1: 100.0000 (83.0000) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1706 data: 0.0778 max mem: 3535 +Test: [ 930/2387] eta: 0:04:45 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0794 (0.5089) acc1: 100.0000 (83.1826) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1800 data: 0.0878 max mem: 3535 +Test: [ 940/2387] eta: 0:04:43 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0795 (0.5044) acc1: 100.0000 (83.3613) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1855 data: 0.0915 max mem: 3535 +Test: [ 950/2387] eta: 0:04:41 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0791 (0.4999) acc1: 100.0000 (83.5362) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1862 data: 0.0920 max mem: 3535 +Test: [ 960/2387] eta: 0:04:39 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0791 (0.4955) acc1: 100.0000 (83.7076) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1891 data: 0.0966 max mem: 3535 +Test: [ 970/2387] eta: 0:04:37 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0793 (0.4912) acc1: 100.0000 (83.8754) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1882 data: 0.0955 max mem: 3535 +Test: [ 980/2387] eta: 0:04:35 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0789 (0.4870) acc1: 100.0000 (84.0397) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1805 data: 0.0880 max mem: 3535 +Test: [ 990/2387] eta: 0:04:33 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0767 (0.4829) acc1: 100.0000 (84.2008) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0871 max mem: 3535 +Test: [1000/2387] eta: 0:04:30 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0729 (0.4788) acc1: 100.0000 (84.3586) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1805 data: 0.0871 max mem: 3535 +Test: [1010/2387] eta: 0:04:28 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0725 (0.4747) acc1: 100.0000 (84.5133) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0833 max mem: 3535 +Test: [1020/2387] eta: 0:04:26 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0719 (0.4708) acc1: 100.0000 (84.6650) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1756 data: 0.0832 max mem: 3535 +Test: [1030/2387] eta: 0:04:24 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0719 (0.4669) acc1: 100.0000 (84.8137) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1883 data: 0.0939 max mem: 3535 +Test: [1040/2387] eta: 0:04:22 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0715 (0.4631) acc1: 100.0000 (84.9596) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1863 data: 0.0936 max mem: 3535 +Test: [1050/2387] eta: 0:04:20 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0722 (0.4594) acc1: 100.0000 (85.1027) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1732 data: 0.0820 max mem: 3535 +Test: [1060/2387] eta: 0:04:17 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0724 (0.4558) acc1: 100.0000 (85.2431) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1720 data: 0.0810 max mem: 3535 +Test: [1070/2387] eta: 0:04:15 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0708 (0.4522) acc1: 100.0000 (85.3809) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1731 data: 0.0818 max mem: 3535 +Test: [1080/2387] eta: 0:04:13 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0741 (0.4487) acc1: 100.0000 (85.5162) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0830 max mem: 3535 +Test: [1090/2387] eta: 0:04:11 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0753 (0.4453) acc1: 100.0000 (85.6489) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1765 data: 0.0854 max mem: 3535 +Test: [1100/2387] eta: 0:04:09 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0748 (0.4419) acc1: 100.0000 (85.7793) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1841 data: 0.0922 max mem: 3535 +Test: [1110/2387] eta: 0:04:07 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0737 (0.4386) acc1: 100.0000 (85.9073) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1712 data: 0.0787 max mem: 3535 +Test: [1120/2387] eta: 0:04:05 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0734 (0.4353) acc1: 100.0000 (86.0330) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0824 max mem: 3536 +Test: [1130/2387] eta: 0:04:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0729 (0.4321) acc1: 100.0000 (86.1565) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1861 data: 0.0918 max mem: 3536 +Test: [1140/2387] eta: 0:04:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0733 (0.4290) acc1: 100.0000 (86.2778) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1599 data: 0.0669 max mem: 3536 +Test: [1150/2387] eta: 0:03:58 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0749 (0.4259) acc1: 100.0000 (86.3970) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1569 data: 0.0653 max mem: 3536 +Test: [1160/2387] eta: 0:03:56 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0732 (0.4229) acc1: 100.0000 (86.5142) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1644 data: 0.0726 max mem: 3536 +Test: [1170/2387] eta: 0:03:54 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0736 (0.4199) acc1: 100.0000 (86.6293) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1712 data: 0.0768 max mem: 3536 +Test: [1180/2387] eta: 0:03:51 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0750 (0.4170) acc1: 100.0000 (86.7426) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0846 max mem: 3536 +Test: [1190/2387] eta: 0:03:49 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0727 (0.4141) acc1: 100.0000 (86.8539) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1721 data: 0.0799 max mem: 3536 +Test: [1200/2387] eta: 0:03:47 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0685 (0.4112) acc1: 100.0000 (86.9633) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1729 data: 0.0778 max mem: 3536 +Test: [1210/2387] eta: 0:03:45 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0688 (0.4084) acc1: 100.0000 (87.0710) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1848 data: 0.0877 max mem: 3536 +Test: [1220/2387] eta: 0:03:43 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0721 (0.4056) acc1: 100.0000 (87.1769) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1921 data: 0.0984 max mem: 3536 +Test: [1230/2387] eta: 0:03:41 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0744 (0.4029) acc1: 100.0000 (87.2810) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0978 max mem: 3536 +Test: [1240/2387] eta: 0:03:39 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0756 (0.4003) acc1: 100.0000 (87.3835) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1797 data: 0.0868 max mem: 3536 +Test: [1250/2387] eta: 0:03:37 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0756 (0.3977) acc1: 100.0000 (87.4844) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1775 data: 0.0850 max mem: 3536 +Test: [1260/2387] eta: 0:03:36 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0727 (0.3951) acc1: 100.0000 (87.5836) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2071 data: 0.1125 max mem: 3536 +Test: [1270/2387] eta: 0:03:34 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0724 (0.3926) acc1: 100.0000 (87.6813) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1923 data: 0.0986 max mem: 3536 +Test: [1280/2387] eta: 0:03:31 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0717 (0.3901) acc1: 100.0000 (87.7775) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1637 data: 0.0733 max mem: 3536 +Test: [1290/2387] eta: 0:03:30 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0708 (0.3876) acc1: 100.0000 (87.8722) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2049 data: 0.1135 max mem: 3536 +Test: [1300/2387] eta: 0:03:28 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0711 (0.3852) acc1: 100.0000 (87.9654) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2105 data: 0.1159 max mem: 3536 +Test: [1310/2387] eta: 0:03:26 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0719 (0.3828) acc1: 100.0000 (88.0572) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1865 data: 0.0903 max mem: 3536 +Test: [1320/2387] eta: 0:03:24 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0714 (0.3804) acc1: 100.0000 (88.1476) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2136 data: 0.1208 max mem: 3536 +Test: [1330/2387] eta: 0:03:22 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0713 (0.3781) acc1: 100.0000 (88.2366) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2033 data: 0.1095 max mem: 3536 +Test: [1340/2387] eta: 0:03:20 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0719 (0.3758) acc1: 100.0000 (88.3244) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0805 max mem: 3536 +Test: [1350/2387] eta: 0:03:18 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0693 (0.3735) acc1: 100.0000 (88.4108) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1804 data: 0.0896 max mem: 3537 +Test: [1360/2387] eta: 0:03:16 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0664 (0.3713) acc1: 100.0000 (88.4959) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1797 data: 0.0863 max mem: 3537 +Test: [1370/2387] eta: 0:03:15 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0670 (0.3691) acc1: 100.0000 (88.5798) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1894 data: 0.0935 max mem: 3537 +Test: [1380/2387] eta: 0:03:13 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0680 (0.3669) acc1: 100.0000 (88.6625) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2006 data: 0.1081 max mem: 3537 +Test: [1390/2387] eta: 0:03:11 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0656 (0.3647) acc1: 100.0000 (88.7440) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1960 data: 0.1052 max mem: 3537 +Test: [1400/2387] eta: 0:03:09 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0656 (0.3626) acc1: 100.0000 (88.8244) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1922 data: 0.0983 max mem: 3537 +Test: [1410/2387] eta: 0:03:07 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0681 (0.3605) acc1: 100.0000 (88.9036) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1950 data: 0.1014 max mem: 3537 +Test: [1420/2387] eta: 0:03:05 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0719 (0.3585) acc1: 100.0000 (88.9817) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2018 data: 0.1091 max mem: 3537 +Test: [1430/2387] eta: 0:03:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0733 (0.3565) acc1: 100.0000 (89.0587) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2101 data: 0.1147 max mem: 3537 +Test: [1440/2387] eta: 0:03:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0727 (0.3545) acc1: 100.0000 (89.1346) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2013 data: 0.1072 max mem: 3537 +Test: [1450/2387] eta: 0:03:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0727 (0.3526) acc1: 100.0000 (89.2093) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1985 data: 0.1045 max mem: 3537 +Test: [1460/2387] eta: 0:02:58 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0736 (0.3507) acc1: 100.0000 (89.2832) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2086 data: 0.1135 max mem: 3537 +Test: [1470/2387] eta: 0:02:56 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0728 (0.3488) acc1: 100.0000 (89.3560) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2200 data: 0.1277 max mem: 3537 +Test: [1480/2387] eta: 0:02:54 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0742 (0.3470) acc1: 100.0000 (89.4279) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2107 data: 0.1186 max mem: 3537 +Test: [1490/2387] eta: 0:02:52 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0741 (0.3451) acc1: 100.0000 (89.4988) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1866 data: 0.0905 max mem: 3537 +Test: [1500/2387] eta: 0:02:50 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0703 (0.3433) acc1: 100.0000 (89.5686) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1908 data: 0.0964 max mem: 3537 +Test: [1510/2387] eta: 0:02:49 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0701 (0.3415) acc1: 100.0000 (89.6376) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2165 data: 0.1232 max mem: 3537 +Test: [1520/2387] eta: 0:02:46 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0720 (0.3397) acc1: 100.0000 (89.7058) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1931 data: 0.0964 max mem: 3537 +Test: [1530/2387] eta: 0:02:44 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0720 (0.3380) acc1: 100.0000 (89.7730) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1670 data: 0.0718 max mem: 3537 +Test: [1540/2387] eta: 0:02:43 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0730 (0.3363) acc1: 100.0000 (89.8394) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0971 max mem: 3537 +Test: [1550/2387] eta: 0:02:41 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0730 (0.3345) acc1: 100.0000 (89.9049) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1877 data: 0.0944 max mem: 3537 +Test: [1560/2387] eta: 0:02:39 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0706 (0.3328) acc1: 100.0000 (89.9695) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1804 data: 0.0875 max mem: 3537 +Test: [1570/2387] eta: 0:02:37 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0681 (0.3312) acc1: 100.0000 (90.0334) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1816 data: 0.0867 max mem: 3537 +Test: [1580/2387] eta: 0:02:35 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0686 (0.3295) acc1: 100.0000 (90.0964) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1756 data: 0.0793 max mem: 3538 +Test: [1590/2387] eta: 0:02:33 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0771 (0.3279) acc1: 100.0000 (90.1587) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1768 data: 0.0831 max mem: 3538 +Test: [1600/2387] eta: 0:02:31 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0766 (0.3264) acc1: 100.0000 (90.2201) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1805 data: 0.0883 max mem: 3538 +Test: [1610/2387] eta: 0:02:29 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0751 (0.3248) acc1: 100.0000 (90.2808) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1862 data: 0.0933 max mem: 3538 +Test: [1620/2387] eta: 0:02:27 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0751 (0.3233) acc1: 100.0000 (90.3408) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1903 data: 0.0967 max mem: 3538 +Test: [1630/2387] eta: 0:02:25 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0770 (0.3218) acc1: 100.0000 (90.3997) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1887 data: 0.0961 max mem: 3538 +Test: [1640/2387] eta: 0:02:23 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0774 (0.3203) acc1: 100.0000 (90.4582) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1929 data: 0.0997 max mem: 3538 +Test: [1650/2387] eta: 0:02:21 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0772 (0.3188) acc1: 100.0000 (90.5160) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1879 data: 0.0937 max mem: 3538 +Test: [1660/2387] eta: 0:02:19 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0767 (0.3174) acc1: 100.0000 (90.5729) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1823 data: 0.0887 max mem: 3538 +Test: [1670/2387] eta: 0:02:17 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0764 (0.3159) acc1: 100.0000 (90.6292) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0937 max mem: 3538 +Test: [1680/2387] eta: 0:02:15 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0769 (0.3145) acc1: 100.0000 (90.6850) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1925 data: 0.0978 max mem: 3538 +Test: [1690/2387] eta: 0:02:13 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0769 (0.3131) acc1: 100.0000 (90.7400) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1856 data: 0.0925 max mem: 3538 +Test: [1700/2387] eta: 0:02:11 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0759 (0.3117) acc1: 100.0000 (90.7945) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1763 data: 0.0826 max mem: 3538 +Test: [1710/2387] eta: 0:02:09 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0738 (0.3103) acc1: 100.0000 (90.8483) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1703 data: 0.0772 max mem: 3538 +Test: [1720/2387] eta: 0:02:07 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0738 (0.3089) acc1: 100.0000 (90.9015) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1700 data: 0.0744 max mem: 3538 +Test: [1730/2387] eta: 0:02:05 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0748 (0.3076) acc1: 100.0000 (90.9540) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1688 data: 0.0752 max mem: 3538 +Test: [1740/2387] eta: 0:02:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0748 (0.3062) acc1: 100.0000 (91.0060) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1634 data: 0.0722 max mem: 3538 +Test: [1750/2387] eta: 0:02:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0717 (0.3049) acc1: 100.0000 (91.0573) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1634 data: 0.0693 max mem: 3538 +Test: [1760/2387] eta: 0:01:59 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0712 (0.3035) acc1: 100.0000 (91.1081) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1659 data: 0.0724 max mem: 3538 +Test: [1770/2387] eta: 0:01:57 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0722 (0.3022) acc1: 100.0000 (91.1583) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1619 data: 0.0716 max mem: 3538 +Test: [1780/2387] eta: 0:01:55 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0728 (0.3010) acc1: 100.0000 (91.2080) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1572 data: 0.0648 max mem: 3538 +Test: [1790/2387] eta: 0:01:53 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0738 (0.2997) acc1: 100.0000 (91.2571) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1597 data: 0.0658 max mem: 3538 +Test: [1800/2387] eta: 0:01:51 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0755 (0.2985) acc1: 100.0000 (91.3056) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1666 data: 0.0722 max mem: 3538 +Test: [1810/2387] eta: 0:01:49 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0737 (0.2972) acc1: 100.0000 (91.3536) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1665 data: 0.0737 max mem: 3539 +Test: [1820/2387] eta: 0:01:47 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0703 (0.2960) acc1: 100.0000 (91.4011) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1603 data: 0.0688 max mem: 3539 +Test: [1830/2387] eta: 0:01:45 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0720 (0.2947) acc1: 100.0000 (91.4481) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1579 data: 0.0638 max mem: 3539 +Test: [1840/2387] eta: 0:01:43 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0712 (0.2935) acc1: 100.0000 (91.4945) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1575 data: 0.0633 max mem: 3539 +Test: [1850/2387] eta: 0:01:41 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0710 (0.2923) acc1: 100.0000 (91.5403) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1648 data: 0.0715 max mem: 3539 +Test: [1860/2387] eta: 0:01:39 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0710 (0.2911) acc1: 100.0000 (91.5858) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1684 data: 0.0735 max mem: 3539 +Test: [1870/2387] eta: 0:01:37 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0709 (0.2900) acc1: 100.0000 (91.6303) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1680 data: 0.0745 max mem: 3539 +Test: [1880/2387] eta: 0:01:35 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0711 (0.2888) acc1: 100.0000 (91.6748) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1687 data: 0.0775 max mem: 3539 +Test: [1890/2387] eta: 0:01:33 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0710 (0.2876) acc1: 100.0000 (91.7186) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1743 data: 0.0833 max mem: 3539 +Test: [1900/2387] eta: 0:01:32 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0708 (0.2865) acc1: 100.0000 (91.7620) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1796 data: 0.0887 max mem: 3539 +Test: [1910/2387] eta: 0:01:30 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0702 (0.2854) acc1: 100.0000 (91.8048) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1724 data: 0.0795 max mem: 3539 +Test: [1920/2387] eta: 0:01:28 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0716 (0.2843) acc1: 100.0000 (91.8475) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1719 data: 0.0784 max mem: 3539 +Test: [1930/2387] eta: 0:01:26 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0743 (0.2832) acc1: 100.0000 (91.8897) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1774 data: 0.0849 max mem: 3539 +Test: [1940/2387] eta: 0:01:24 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0715 (0.2821) acc1: 100.0000 (91.9315) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1804 data: 0.0884 max mem: 3539 +Test: [1950/2387] eta: 0:01:22 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0715 (0.2810) acc1: 100.0000 (91.9729) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0837 max mem: 3539 +Test: [1960/2387] eta: 0:01:20 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0718 (0.2800) acc1: 100.0000 (92.0138) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1789 data: 0.0862 max mem: 3539 +Test: [1970/2387] eta: 0:01:18 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0718 (0.2789) acc1: 100.0000 (92.0543) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1944 data: 0.1010 max mem: 3539 +Test: [1980/2387] eta: 0:01:16 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0737 (0.2779) acc1: 100.0000 (92.0939) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1846 data: 0.0929 max mem: 3539 +Test: [1990/2387] eta: 0:01:14 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0844 (0.2769) acc1: 99.7396 (92.1322) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1667 data: 0.0744 max mem: 3539 +Test: [2000/2387] eta: 0:01:12 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0874 (0.2760) acc1: 99.7396 (92.1690) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1746 data: 0.0829 max mem: 3539 +Test: [2010/2387] eta: 0:01:11 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0880 (0.2751) acc1: 99.4792 (92.2051) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1776 data: 0.0864 max mem: 3539 +Test: [2020/2387] eta: 0:01:09 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0858 (0.2741) acc1: 99.7396 (92.2430) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1704 data: 0.0784 max mem: 3539 +Test: [2030/2387] eta: 0:01:07 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0844 (0.2732) acc1: 99.7396 (92.2799) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1840 data: 0.0917 max mem: 3540 +Test: [2040/2387] eta: 0:01:05 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0863 (0.2723) acc1: 99.7396 (92.3159) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1912 data: 0.0984 max mem: 3540 +Test: [2050/2387] eta: 0:01:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0861 (0.2714) acc1: 99.4792 (92.3514) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1861 data: 0.0946 max mem: 3540 +Test: [2060/2387] eta: 0:01:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0844 (0.2705) acc1: 99.4792 (92.3866) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1881 data: 0.0976 max mem: 3540 +Test: [2070/2387] eta: 0:00:59 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0814 (0.2696) acc1: 99.7396 (92.4226) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1839 data: 0.0923 max mem: 3540 +Test: [2080/2387] eta: 0:00:57 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0814 (0.2687) acc1: 100.0000 (92.4583) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1720 data: 0.0797 max mem: 3540 +Test: [2090/2387] eta: 0:00:55 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0812 (0.2678) acc1: 100.0000 (92.4934) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1720 data: 0.0798 max mem: 3540 +Test: [2100/2387] eta: 0:00:53 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0812 (0.2669) acc1: 99.7396 (92.5280) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1742 data: 0.0826 max mem: 3540 +Test: [2110/2387] eta: 0:00:52 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0826 (0.2660) acc1: 99.7396 (92.5620) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1772 data: 0.0864 max mem: 3540 +Test: [2120/2387] eta: 0:00:50 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0785 (0.2651) acc1: 100.0000 (92.5968) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1775 data: 0.0865 max mem: 3540 +Test: [2130/2387] eta: 0:00:48 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0737 (0.2642) acc1: 100.0000 (92.6316) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1710 data: 0.0796 max mem: 3540 +Test: [2140/2387] eta: 0:00:46 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0716 (0.2633) acc1: 100.0000 (92.6660) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1723 data: 0.0802 max mem: 3540 +Test: [2150/2387] eta: 0:00:44 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0719 (0.2624) acc1: 100.0000 (92.7001) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1794 data: 0.0876 max mem: 3540 +Test: [2160/2387] eta: 0:00:42 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0715 (0.2616) acc1: 100.0000 (92.7339) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1730 data: 0.0813 max mem: 3540 +Test: [2170/2387] eta: 0:00:40 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0703 (0.2607) acc1: 100.0000 (92.7674) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1911 data: 0.0998 max mem: 3540 +Test: [2180/2387] eta: 0:00:38 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0728 (0.2598) acc1: 100.0000 (92.8003) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2076 data: 0.1174 max mem: 3540 +Test: [2190/2387] eta: 0:00:37 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0719 (0.2590) acc1: 100.0000 (92.8331) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1957 data: 0.1044 max mem: 3540 +Test: [2200/2387] eta: 0:00:35 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0709 (0.2581) acc1: 100.0000 (92.8657) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1905 data: 0.0988 max mem: 3540 +Test: [2210/2387] eta: 0:00:33 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0715 (0.2573) acc1: 100.0000 (92.8980) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1873 data: 0.0952 max mem: 3540 +Test: [2220/2387] eta: 0:00:31 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0737 (0.2564) acc1: 100.0000 (92.9299) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1885 data: 0.0966 max mem: 3540 +Test: [2230/2387] eta: 0:00:29 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0737 (0.2556) acc1: 100.0000 (92.9616) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1901 data: 0.0988 max mem: 3540 +Test: [2240/2387] eta: 0:00:27 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0707 (0.2548) acc1: 100.0000 (92.9930) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1948 data: 0.1036 max mem: 3540 +Test: [2250/2387] eta: 0:00:25 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0687 (0.2540) acc1: 100.0000 (93.0242) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1947 data: 0.1014 max mem: 3540 +Test: [2260/2387] eta: 0:00:23 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0674 (0.2531) acc1: 100.0000 (93.0550) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1950 data: 0.1015 max mem: 3541 +Test: [2270/2387] eta: 0:00:21 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0679 (0.2523) acc1: 100.0000 (93.0856) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0981 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0672 (0.2515) acc1: 100.0000 (93.1159) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1825 data: 0.0902 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0673 (0.2507) acc1: 100.0000 (93.1460) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1815 data: 0.0902 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0666 (0.2499) acc1: 100.0000 (93.1757) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1767 data: 0.0865 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0666 (0.2491) acc1: 100.0000 (93.2053) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1601 data: 0.0697 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0686 (0.2483) acc1: 100.0000 (93.2343) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1518 data: 0.0610 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0692 (0.2476) acc1: 100.0000 (93.2634) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1494 data: 0.0589 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0685 (0.2468) acc1: 100.0000 (93.2921) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1453 data: 0.0551 max mem: 3541 +Test: [2350/2387] eta: 0:00:06 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0692 (0.2461) acc1: 100.0000 (93.3207) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1495 data: 0.0589 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0688 (0.2453) acc1: 100.0000 (93.3490) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1462 data: 0.0551 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0679 (0.2446) acc1: 100.0000 (93.3770) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1389 data: 0.0479 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0684 (0.2438) acc1: 100.0000 (93.4048) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1359 data: 0.0445 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0687 (0.2434) acc1: 100.0000 (93.4189) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1291 data: 0.0406 max mem: 3541 +Test: Total time: 0:07:24 (0.1864 s / it) +* Overall Acc@1 93.419 Acc@5 100.000 + target == real Acc@1 0.005 Acc@5 100.000 + target == fake Acc@1 99.979 Acc@5 100.000 + loss 0.243 + PR-AUC 0.9315 ROC-AUC 0.5000 EER 49.95% (thr=0.9307) +Accuracy of the network on the 916267 test images: 93.4% +[rank0]:[W726 04:40:09.462081439 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:40:31.258000 3149123 torch/distributed/run.py:766] +W0726 04:40:31.258000 3149123 torch/distributed/run.py:766] ***************************************** +W0726 04:40:31.258000 3149123 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:40:31.258000 3149123 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:41:05.413957320 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:41:05.734777012 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:41:05.820013686 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 6): env:// +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:41:06.178765751 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:41:06.186536211 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:41:06.264507022 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:41:06.291916489 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:41:06.327340162 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n5_seed7_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed7_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 0:57:34 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.6643 (2.6643) acc1: 0.0000 (0.0000) acc5: 100.0000 (100.0000) time: 6.6809 data: 4.9194 max mem: 3531 +Test: [ 10/517] eta: 0:05:49 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.6643 (2.6664) acc1: 0.0000 (0.0000) acc5: 100.0000 (100.0000) time: 0.6893 data: 0.4477 max mem: 3531 +Test: [ 20/517] eta: 0:03:22 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.6411 (2.6490) acc1: 0.0000 (0.0000) acc5: 100.0000 (100.0000) time: 0.0945 data: 0.0005 max mem: 3531 +Test: [ 30/517] eta: 0:02:47 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.6411 (2.6596) acc1: 0.0000 (0.0000) acc5: 100.0000 (100.0000) time: 0.1544 data: 0.0549 max mem: 3531 +Test: [ 40/517] eta: 0:02:35 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.6889 (2.6687) acc1: 0.0000 (0.0000) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1416 max mem: 3531 +Test: [ 50/517] eta: 0:02:25 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.6889 (2.5480) acc1: 0.0000 (4.7692) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2599 data: 0.1606 max mem: 3531 +Test: [ 60/517] eta: 0:02:19 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0744 (2.1422) acc1: 100.0000 (20.3808) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2649 data: 0.1664 max mem: 3531 +Test: [ 70/517] eta: 0:02:13 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0738 (1.8508) acc1: 100.0000 (31.5948) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2649 data: 0.1678 max mem: 3531 +Test: [ 80/517] eta: 0:02:08 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0757 (1.6317) acc1: 100.0000 (40.0399) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1575 max mem: 3531 +Test: [ 90/517] eta: 0:02:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0778 (1.4611) acc1: 100.0000 (46.6289) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1582 max mem: 3531 +Test: [100/517] eta: 0:01:59 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0777 (1.3237) acc1: 100.0000 (51.9132) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2599 data: 0.1592 max mem: 3531 +Test: [110/517] eta: 0:01:54 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0736 (1.2111) acc1: 100.0000 (56.2453) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2419 data: 0.1432 max mem: 3531 +Test: [120/517] eta: 0:01:50 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0722 (1.1169) acc1: 100.0000 (59.8614) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2399 data: 0.1452 max mem: 3531 +Test: [130/517] eta: 0:01:47 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0709 (1.0372) acc1: 100.0000 (62.9254) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2679 data: 0.1723 max mem: 3531 +Test: [140/517] eta: 0:01:43 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0756 (0.9691) acc1: 100.0000 (65.5548) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2550 data: 0.1616 max mem: 3531 +Test: [150/517] eta: 0:01:40 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0761 (0.9099) acc1: 100.0000 (67.8360) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2490 data: 0.1583 max mem: 3531 +Test: [160/517] eta: 0:01:38 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0747 (0.8580) acc1: 100.0000 (69.8337) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2703 data: 0.1798 max mem: 3531 +Test: [170/517] eta: 0:01:34 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0747 (0.8122) acc1: 100.0000 (71.5978) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2662 data: 0.1749 max mem: 3531 +Test: [180/517] eta: 0:01:32 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0708 (0.7711) acc1: 100.0000 (73.1670) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2585 data: 0.1646 max mem: 3531 +Test: [190/517] eta: 0:01:28 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0693 (0.7344) acc1: 100.0000 (74.5719) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2486 data: 0.1547 max mem: 3531 +Test: [200/517] eta: 0:01:25 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0695 (0.7015) acc1: 100.0000 (75.8370) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2518 data: 0.1551 max mem: 3531 +Test: [210/517] eta: 0:01:22 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0738 (0.6717) acc1: 100.0000 (76.9821) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2599 data: 0.1636 max mem: 3532 +Test: [220/517] eta: 0:01:20 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0737 (0.6446) acc1: 100.0000 (78.0237) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2538 data: 0.1595 max mem: 3532 +Test: [230/517] eta: 0:01:17 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0737 (0.6199) acc1: 100.0000 (78.9750) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2657 data: 0.1718 max mem: 3532 +Test: [240/517] eta: 0:01:14 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0811 (0.5977) acc1: 100.0000 (79.8474) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2546 data: 0.1629 max mem: 3532 +Test: [250/517] eta: 0:01:11 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0834 (0.5770) acc1: 100.0000 (80.6503) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2497 data: 0.1578 max mem: 3532 +Test: [260/517] eta: 0:01:08 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0772 (0.5579) acc1: 100.0000 (81.3917) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2624 data: 0.1679 max mem: 3532 +Test: [270/517] eta: 0:01:06 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0779 (0.5402) acc1: 100.0000 (82.0783) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2705 data: 0.1762 max mem: 3532 +Test: [280/517] eta: 0:01:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0786 (0.5239) acc1: 100.0000 (82.7161) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2570 data: 0.1640 max mem: 3532 +Test: [290/517] eta: 0:01:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0782 (0.5086) acc1: 100.0000 (83.3101) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2486 data: 0.1523 max mem: 3532 +Test: [300/517] eta: 0:00:57 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0765 (0.4943) acc1: 100.0000 (83.8645) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2701 data: 0.1716 max mem: 3532 +Test: [310/517] eta: 0:00:55 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0820 (0.4810) acc1: 100.0000 (84.3834) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2649 data: 0.1667 max mem: 3532 +Test: [320/517] eta: 0:00:52 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0822 (0.4684) acc1: 100.0000 (84.8699) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2595 data: 0.1618 max mem: 3532 +Test: [330/517] eta: 0:00:49 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0723 (0.4565) acc1: 100.0000 (85.3270) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1562 max mem: 3532 +Test: [340/517] eta: 0:00:47 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0722 (0.4452) acc1: 100.0000 (85.7573) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2738 data: 0.1763 max mem: 3532 +Test: [350/517] eta: 0:00:44 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0724 (0.4346) acc1: 100.0000 (86.1630) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2651 data: 0.1680 max mem: 3532 +Test: [360/517] eta: 0:00:41 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0722 (0.4245) acc1: 100.0000 (86.5463) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2449 data: 0.1473 max mem: 3532 +Test: [370/517] eta: 0:00:39 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0657 (0.4149) acc1: 100.0000 (86.9090) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2586 data: 0.1613 max mem: 3532 +Test: [380/517] eta: 0:00:36 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0660 (0.4058) acc1: 100.0000 (87.2526) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2503 data: 0.1526 max mem: 3532 +Test: [390/517] eta: 0:00:33 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0701 (0.3972) acc1: 100.0000 (87.5786) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2498 data: 0.1528 max mem: 3532 +Test: [400/517] eta: 0:00:30 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0691 (0.3889) acc1: 100.0000 (87.8884) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2495 data: 0.1524 max mem: 3532 +Test: [410/517] eta: 0:00:28 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0686 (0.3812) acc1: 100.0000 (88.1830) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1563 max mem: 3532 +Test: [420/517] eta: 0:00:25 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0705 (0.3739) acc1: 100.0000 (88.4637) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2499 data: 0.1514 max mem: 3532 +Test: [430/517] eta: 0:00:22 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0745 (0.3669) acc1: 100.0000 (88.7314) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2462 data: 0.1488 max mem: 3532 +Test: [440/517] eta: 0:00:20 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0746 (0.3603) acc1: 100.0000 (88.9869) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2554 data: 0.1593 max mem: 3533 +Test: [450/517] eta: 0:00:17 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0736 (0.3539) acc1: 100.0000 (89.2311) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2492 data: 0.1535 max mem: 3533 +Test: [460/517] eta: 0:00:14 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0703 (0.3477) acc1: 100.0000 (89.4647) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2495 data: 0.1531 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0702 (0.3418) acc1: 100.0000 (89.6884) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2561 data: 0.1593 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0713 (0.3362) acc1: 100.0000 (89.9028) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2435 data: 0.1469 max mem: 3533 +Test: [490/517] eta: 0:00:07 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0731 (0.3309) acc1: 100.0000 (90.1084) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2134 data: 0.1177 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0761 (0.3258) acc1: 100.0000 (90.3058) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1857 data: 0.0931 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0744 (0.3208) acc1: 100.0000 (90.4956) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1734 data: 0.0839 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0726 (0.3179) acc1: 100.0000 (90.6011) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0838 max mem: 3533 +Test: Total time: 0:02:12 (0.2557 s / it) +* Overall Acc@1 90.601 Acc@5 100.000 + target == real Acc@1 0.000 Acc@5 100.000 + target == fake Acc@1 100.000 Acc@5 100.000 + loss 0.318 + PR-AUC 0.8909 ROC-AUC 0.4533 EER 53.25% (thr=0.9307) +Accuracy of the network on the 198427 test images: 90.6% +[rank0]:[W726 04:43:29.430042095 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 04:43:48.495000 3149697 torch/distributed/run.py:766] +W0726 04:43:48.495000 3149697 torch/distributed/run.py:766] ***************************************** +W0726 04:43:48.495000 3149697 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:43:48.495000 3149697 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:44:15.280590990 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:44:15.761440927 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:44:15.923861522 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +NCCL version 2.26.2+cuda12.2 +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:44:15.106680367 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:44:16.242431188 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:44:16.284104324 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:44:16.396981758 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:44:16.492973732 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n5_seed7_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed7_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:28:58 acc1_pos: 3.125 acc5_pos: 100.000 loss: 2.1170 (2.1170) acc1: 3.1250 (3.1250) acc5: 100.0000 (100.0000) time: 4.4684 data: 2.5064 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/389] eta: 0:03:15 acc1_pos: 1.302 acc5_pos: 100.000 loss: 2.5049 (2.4793) acc1: 1.0417 (1.4678) acc5: 100.0000 (100.0000) time: 0.5148 data: 0.2546 max mem: 3531 +Test: [ 20/389] eta: 0:02:04 acc1_pos: 0.260 acc5_pos: 100.000 loss: 2.5949 (2.5467) acc1: 0.0000 (1.5377) acc5: 100.0000 (100.0000) time: 0.1298 data: 0.0383 max mem: 3531 +Test: [ 30/389] eta: 0:01:47 acc1_pos: 0.260 acc5_pos: 100.000 loss: 2.6596 (2.5783) acc1: 0.0000 (1.0501) acc5: 100.0000 (100.0000) time: 0.1795 data: 0.0856 max mem: 3531 +Test: [ 40/389] eta: 0:01:40 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.6193 (2.5677) acc1: 0.0000 (1.3847) acc5: 100.0000 (100.0000) time: 0.2363 data: 0.1439 max mem: 3531 +Test: [ 50/389] eta: 0:01:29 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.5380 (2.5690) acc1: 0.0000 (1.1132) acc5: 100.0000 (100.0000) time: 0.2111 data: 0.1209 max mem: 3531 +Test: [ 60/389] eta: 0:01:20 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.5370 (2.5586) acc1: 0.0000 (0.9477) acc5: 100.0000 (100.0000) time: 0.1586 data: 0.0676 max mem: 3531 +Test: [ 70/389] eta: 0:01:14 acc1_pos: 4.427 acc5_pos: 100.000 loss: 2.4912 (2.5350) acc1: 0.0000 (0.9426) acc5: 100.0000 (100.0000) time: 0.1514 data: 0.0575 max mem: 3531 +Test: [ 80/389] eta: 0:01:10 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.4809 (2.5330) acc1: 0.0000 (0.9066) acc5: 100.0000 (100.0000) time: 0.1716 data: 0.0772 max mem: 3531 +Test: [ 90/389] eta: 0:01:05 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.4992 (2.5305) acc1: 0.0000 (0.8156) acc5: 100.0000 (100.0000) time: 0.1749 data: 0.0833 max mem: 3531 +Test: [100/389] eta: 0:01:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.4461 (2.4577) acc1: 0.0000 (3.4396) acc5: 100.0000 (100.0000) time: 0.1601 data: 0.0686 max mem: 3531 +Test: [110/389] eta: 0:00:58 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.4603 (2.4621) acc1: 0.0000 (3.1414) acc5: 100.0000 (100.0000) time: 0.1641 data: 0.0721 max mem: 3531 +Test: [120/389] eta: 0:00:55 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.4603 (2.4356) acc1: 0.0000 (3.7599) acc5: 100.0000 (100.0000) time: 0.1726 data: 0.0799 max mem: 3531 +Test: [130/389] eta: 0:00:52 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.6492 (2.3048) acc1: 19.2708 (9.2557) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1686 data: 0.0758 max mem: 3531 +Test: [140/389] eta: 0:00:49 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0851 (2.1468) acc1: 100.0000 (15.6896) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1608 data: 0.0695 max mem: 3531 +Test: [150/389] eta: 0:00:47 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0737 (2.0101) acc1: 100.0000 (21.2714) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1642 data: 0.0715 max mem: 3531 +Test: [160/389] eta: 0:00:44 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0737 (1.9142) acc1: 100.0000 (24.7638) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1661 data: 0.0710 max mem: 3531 +Test: [170/389] eta: 0:00:42 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0668 (1.8064) acc1: 100.0000 (29.1621) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1513 data: 0.0564 max mem: 3531 +Test: [180/389] eta: 0:00:39 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0745 (1.7136) acc1: 100.0000 (32.8931) acc5: 100.0000 (100.0000) acc1_neg: 66.927 acc5_neg: 100.000 time: 0.1480 data: 0.0554 max mem: 3531 +Test: [190/389] eta: 0:00:37 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0845 (1.6408) acc1: 100.0000 (35.6021) acc5: 100.0000 (100.0000) acc1_neg: 69.792 acc5_neg: 100.000 time: 0.1563 data: 0.0650 max mem: 3531 +Test: [200/389] eta: 0:00:35 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0853 (1.5648) acc1: 100.0000 (38.7075) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1708 data: 0.0800 max mem: 3531 +Test: [210/389] eta: 0:00:33 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0740 (1.4941) acc1: 100.0000 (41.6124) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1579 data: 0.0675 max mem: 3532 +Test: [220/389] eta: 0:00:31 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0710 (1.4314) acc1: 100.0000 (44.1683) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1385 data: 0.0440 max mem: 3532 +Test: [230/389] eta: 0:00:29 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0843 (1.3786) acc1: 100.0000 (46.1355) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1641 data: 0.0680 max mem: 3532 +Test: [240/389] eta: 0:00:27 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0910 (1.3253) acc1: 100.0000 (48.3694) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0749 max mem: 3532 +Test: [250/389] eta: 0:00:25 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0783 (1.2766) acc1: 100.0000 (50.4264) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1553 data: 0.0611 max mem: 3532 +Test: [260/389] eta: 0:00:23 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0918 (1.2387) acc1: 100.0000 (51.9407) acc5: 100.0000 (100.0000) acc1_neg: 15.104 acc5_neg: 100.000 time: 0.1499 data: 0.0566 max mem: 3532 +Test: [270/389] eta: 0:00:21 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0973 (1.2178) acc1: 100.0000 (52.6061) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1846 data: 0.0902 max mem: 3532 +Test: [280/389] eta: 0:00:20 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0923 (1.1900) acc1: 100.0000 (53.5884) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2360 data: 0.1422 max mem: 3532 +Test: [290/389] eta: 0:00:17 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0857 (1.1532) acc1: 100.0000 (55.1296) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1750 data: 0.0806 max mem: 3532 +Test: [300/389] eta: 0:00:16 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1241 (1.1230) acc1: 98.4375 (56.3166) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1566 data: 0.0625 max mem: 3532 +Test: [310/389] eta: 0:00:14 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2196 (1.0966) acc1: 92.4479 (57.3009) acc5: 100.0000 (100.0000) acc1_neg: 52.083 acc5_neg: 100.000 time: 0.2057 data: 0.1143 max mem: 3532 +Test: [320/389] eta: 0:00:12 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1882 (1.0681) acc1: 92.4479 (58.4331) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2021 data: 0.1085 max mem: 3532 +Test: [330/389] eta: 0:00:10 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0736 (1.0383) acc1: 100.0000 (59.6669) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1960 data: 0.0999 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0708 (1.0108) acc1: 100.0000 (60.7977) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1986 data: 0.1040 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0803 (0.9856) acc1: 100.0000 (61.8360) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2028 data: 0.1077 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0641 (0.9600) acc1: 100.0000 (62.8932) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1981 data: 0.1007 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0650 (0.9362) acc1: 100.0000 (63.8933) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1993 data: 0.1041 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0736 (0.9137) acc1: 100.0000 (64.8410) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1625 data: 0.0718 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0755 (0.8972) acc1: 100.0000 (65.5231) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1240 data: 0.0308 max mem: 3532 +Test: Total time: 0:01:11 (0.1835 s / it) +* Overall Acc@1 65.523 Acc@5 100.000 + target == real Acc@1 3.686 Acc@5 100.000 + target == fake Acc@1 94.267 Acc@5 100.000 + loss 0.897 + PR-AUC 0.6873 ROC-AUC 0.5205 EER 47.54% (thr=0.9258) +Accuracy of the network on the 149358 test images: 65.5% +[rank0]:[W726 04:45:40.774490310 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:46:02.255000 3150256 torch/distributed/run.py:766] +W0726 04:46:02.255000 3150256 torch/distributed/run.py:766] ***************************************** +W0726 04:46:02.255000 3150256 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:46:02.255000 3150256 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:46:35.332500666 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:46:35.669119571 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:46:35.787525234 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:46:35.915885907 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:46:36.233665275 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:46:36.269577500 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:46:36.278603044 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:46:36.305265732 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n5_seed7_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed7_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:30:24 acc1_pos: 0.260 acc5_pos: 100.000 loss: 2.6244 (2.6244) acc1: 0.2604 (0.2604) acc5: 100.0000 (100.0000) time: 5.4788 data: 3.2780 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/333] eta: 0:03:07 acc1_pos: 2.083 acc5_pos: 100.000 loss: 2.4984 (2.4923) acc1: 0.2604 (1.5152) acc5: 100.0000 (100.0000) time: 0.5804 data: 0.2983 max mem: 3531 +Test: [ 20/333] eta: 0:01:52 acc1_pos: 1.823 acc5_pos: 100.000 loss: 2.4527 (2.4519) acc1: 1.0417 (1.6121) acc5: 100.0000 (100.0000) time: 0.1048 data: 0.0147 max mem: 3531 +Test: [ 30/333] eta: 0:01:31 acc1_pos: 0.521 acc5_pos: 100.000 loss: 2.4966 (2.4809) acc1: 0.5208 (1.4365) acc5: 100.0000 (100.0000) time: 0.1490 data: 0.0556 max mem: 3531 +Test: [ 40/333] eta: 0:01:31 acc1_pos: 0.260 acc5_pos: 100.000 loss: 2.5548 (2.4965) acc1: 0.2604 (1.3211) acc5: 100.0000 (100.0000) time: 0.2643 data: 0.1684 max mem: 3531 +Test: [ 50/333] eta: 0:01:23 acc1_pos: 8.073 acc5_pos: 100.000 loss: 2.5620 (2.5092) acc1: 0.5208 (1.4246) acc5: 100.0000 (100.0000) time: 0.2807 data: 0.1878 max mem: 3531 +Test: [ 60/333] eta: 0:01:14 acc1_pos: 0.781 acc5_pos: 100.000 loss: 2.5341 (2.5047) acc1: 1.0417 (1.5796) acc5: 100.0000 (100.0000) time: 0.1852 data: 0.0920 max mem: 3531 +Test: [ 70/333] eta: 0:01:08 acc1_pos: 0.260 acc5_pos: 100.000 loss: 2.5254 (2.5073) acc1: 0.5208 (1.4781) acc5: 100.0000 (100.0000) time: 0.1794 data: 0.0840 max mem: 3531 +Test: [ 80/333] eta: 0:01:05 acc1_pos: 0.781 acc5_pos: 100.000 loss: 2.5277 (2.5113) acc1: 0.2604 (1.3760) acc5: 100.0000 (100.0000) time: 0.2207 data: 0.1281 max mem: 3531 +Test: [ 90/333] eta: 0:01:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.5402 (2.5147) acc1: 0.0000 (1.2592) acc5: 100.0000 (100.0000) time: 0.2222 data: 0.1296 max mem: 3531 +Test: [100/333] eta: 0:00:58 acc1_pos: 0.260 acc5_pos: 100.000 loss: 2.5727 (2.5222) acc1: 0.0000 (1.1654) acc5: 100.0000 (100.0000) time: 0.2097 data: 0.1167 max mem: 3531 +Test: [110/333] eta: 0:00:54 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.5794 (2.5274) acc1: 0.0000 (1.0839) acc5: 100.0000 (100.0000) time: 0.2172 data: 0.1257 max mem: 3531 +Test: [120/333] eta: 0:00:52 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.5771 (2.5316) acc1: 0.0000 (1.0008) acc5: 100.0000 (100.0000) time: 0.2197 data: 0.1284 max mem: 3531 +Test: [130/333] eta: 0:00:49 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.5843 (2.5358) acc1: 0.0000 (0.9264) acc5: 100.0000 (100.0000) time: 0.2208 data: 0.1297 max mem: 3531 +Test: [140/333] eta: 0:00:46 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.6077 (2.5407) acc1: 0.0000 (0.8662) acc5: 100.0000 (100.0000) time: 0.2149 data: 0.1224 max mem: 3531 +Test: [150/333] eta: 0:00:43 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.5918 (2.5437) acc1: 0.0000 (0.8123) acc5: 100.0000 (100.0000) time: 0.2087 data: 0.1160 max mem: 3531 +Test: [160/333] eta: 0:00:40 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.5767 (2.4815) acc1: 0.0000 (3.3547) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2089 data: 0.1158 max mem: 3531 +Test: [170/333] eta: 0:00:38 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0975 (2.3420) acc1: 99.2188 (8.9699) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2008 data: 0.1038 max mem: 3531 +Test: [180/333] eta: 0:00:35 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0971 (2.2184) acc1: 99.2188 (13.9474) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2002 data: 0.1055 max mem: 3531 +Test: [190/333] eta: 0:00:33 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0967 (2.1069) acc1: 99.4792 (18.4323) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2100 data: 0.1176 max mem: 3531 +Test: [200/333] eta: 0:00:30 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0856 (2.0065) acc1: 99.7396 (22.4736) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2096 data: 0.1141 max mem: 3531 +Test: [210/333] eta: 0:00:28 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0908 (1.9159) acc1: 99.4792 (26.1095) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2054 data: 0.1113 max mem: 3532 +Test: [220/333] eta: 0:00:25 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0907 (1.8334) acc1: 99.7396 (29.4400) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1999 data: 0.1076 max mem: 3532 +Test: [230/333] eta: 0:00:23 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0899 (1.7582) acc1: 100.0000 (32.4630) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1906 data: 0.0983 max mem: 3532 +Test: [240/333] eta: 0:00:20 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0905 (1.6893) acc1: 99.7396 (35.2319) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1899 data: 0.0972 max mem: 3532 +Test: [250/333] eta: 0:00:18 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0907 (1.6256) acc1: 99.4792 (37.7936) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1977 data: 0.1044 max mem: 3532 +Test: [260/333] eta: 0:00:16 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0895 (1.5667) acc1: 99.4792 (40.1640) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0941 max mem: 3532 +Test: [270/333] eta: 0:00:13 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0895 (1.5123) acc1: 99.7396 (42.3566) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1799 data: 0.0821 max mem: 3532 +Test: [280/333] eta: 0:00:11 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0870 (1.4616) acc1: 99.7396 (44.3950) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1723 data: 0.0746 max mem: 3532 +Test: [290/333] eta: 0:00:09 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0896 (1.4146) acc1: 99.7396 (46.2870) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1535 data: 0.0596 max mem: 3532 +Test: [300/333] eta: 0:00:07 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0923 (1.3709) acc1: 99.4792 (48.0447) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1370 data: 0.0469 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0959 (1.3298) acc1: 99.2188 (49.6952) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1226 data: 0.0327 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0916 (1.2912) acc1: 99.4792 (51.2494) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1205 data: 0.0309 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0915 (1.2550) acc1: 99.4792 (52.7033) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1200 data: 0.0306 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0923 (1.2481) acc1: 99.4792 (52.9720) acc5: 100.0000 (100.0000) acc1_neg: 99.437 acc5_neg: 100.000 time: 0.1204 data: 0.0277 max mem: 3532 +Test: Total time: 0:01:07 (0.2042 s / it) +* Overall Acc@1 52.972 Acc@5 100.000 + target == real Acc@1 0.784 Acc@5 100.000 + target == fake Acc@1 99.429 Acc@5 100.000 + loss 1.248 + PR-AUC 0.5077 ROC-AUC 0.4740 EER 52.28% (thr=0.9263) +Accuracy of the network on the 127843 test images: 53.0% +[rank0]:[W726 04:47:51.457818878 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:48:12.660000 3150815 torch/distributed/run.py:766] +W0726 04:48:12.660000 3150815 torch/distributed/run.py:766] ***************************************** +W0726 04:48:12.660000 3150815 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:48:12.660000 3150815 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:48:47.205635857 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:48:47.262841161 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:48:47.705336699 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:48:47.744301548 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:48:47.813108931 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:48:47.137977127 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:48:47.169506711 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:48:48.305985123 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n5_seed7_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed7_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:23:08 acc1_pos: 0.521 acc5_pos: 100.000 loss: 2.5588 (2.5588) acc1: 0.5208 (0.5208) acc5: 100.0000 (100.0000) time: 5.9080 data: 4.4132 max mem: 3531 +Test: [ 10/235] eta: 0:02:19 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.5857 (2.5804) acc1: 0.2604 (0.1894) acc5: 100.0000 (100.0000) time: 0.6186 data: 0.4015 max mem: 3531 +Test: [ 20/235] eta: 0:01:19 acc1_pos: 0.260 acc5_pos: 100.000 loss: 2.5841 (2.5781) acc1: 0.0000 (0.1736) acc5: 100.0000 (100.0000) time: 0.0926 data: 0.0003 max mem: 3531 +Test: [ 30/235] eta: 0:01:16 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.5802 (2.5793) acc1: 0.0000 (0.1512) acc5: 100.0000 (100.0000) time: 0.2386 data: 0.1442 max mem: 3531 +Test: [ 40/235] eta: 0:01:09 acc1_pos: 0.260 acc5_pos: 100.000 loss: 2.5820 (2.5796) acc1: 0.0000 (0.1461) acc5: 100.0000 (100.0000) time: 0.3473 data: 0.2552 max mem: 3531 +Test: [ 50/235] eta: 0:01:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.5820 (2.5801) acc1: 0.0000 (0.1328) acc5: 100.0000 (100.0000) time: 0.2524 data: 0.1620 max mem: 3531 +Test: [ 60/235] eta: 0:00:52 acc1_pos: 0.260 acc5_pos: 100.000 loss: 2.5832 (2.5803) acc1: 0.0000 (0.1281) acc5: 100.0000 (100.0000) time: 0.1867 data: 0.0956 max mem: 3531 +Test: [ 70/235] eta: 0:00:48 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.5648 (2.2695) acc1: 0.2604 (12.5330) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2199 data: 0.1285 max mem: 3531 +Test: [ 80/235] eta: 0:00:44 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0800 (1.9993) acc1: 100.0000 (23.3218) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2341 data: 0.1410 max mem: 3531 +Test: [ 90/235] eta: 0:00:39 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0802 (1.7885) acc1: 100.0000 (31.7365) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1998 data: 0.1049 max mem: 3531 +Test: [100/235] eta: 0:00:36 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0802 (1.6194) acc1: 100.0000 (38.4824) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1910 data: 0.0974 max mem: 3531 +Test: [110/235] eta: 0:00:32 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0801 (1.4808) acc1: 100.0000 (44.0151) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1969 data: 0.1045 max mem: 3531 +Test: [120/235] eta: 0:00:29 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0821 (1.3653) acc1: 100.0000 (48.6377) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2038 data: 0.1094 max mem: 3531 +Test: [130/235] eta: 0:00:26 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0845 (1.2676) acc1: 100.0000 (52.5545) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2086 data: 0.1142 max mem: 3531 +Test: [140/235] eta: 0:00:23 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0849 (1.1837) acc1: 100.0000 (55.9176) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2128 data: 0.1194 max mem: 3531 +Test: [150/235] eta: 0:00:21 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0849 (1.1110) acc1: 100.0000 (58.8335) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2142 data: 0.1155 max mem: 3531 +Test: [160/235] eta: 0:00:18 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0848 (1.0473) acc1: 100.0000 (61.3839) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2179 data: 0.1214 max mem: 3531 +Test: [170/235] eta: 0:00:15 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0839 (0.9906) acc1: 100.0000 (63.6361) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2156 data: 0.1243 max mem: 3531 +Test: [180/235] eta: 0:00:13 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0758 (0.9401) acc1: 100.0000 (65.6365) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2062 data: 0.1144 max mem: 3531 +Test: [190/235] eta: 0:00:10 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0752 (0.8948) acc1: 100.0000 (67.4343) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2062 data: 0.1149 max mem: 3531 +Test: [200/235] eta: 0:00:08 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0752 (0.8540) acc1: 100.0000 (69.0493) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2015 data: 0.1108 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0752 (0.8171) acc1: 100.0000 (70.5087) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1767 data: 0.0864 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0743 (0.7835) acc1: 100.0000 (71.8420) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1470 data: 0.0570 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0751 (0.7529) acc1: 100.0000 (73.0531) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1289 data: 0.0394 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0751 (0.7414) acc1: 100.0000 (73.3998) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1230 data: 0.0367 max mem: 3532 +Test: Total time: 0:00:52 (0.2238 s / it) +* Overall Acc@1 73.400 Acc@5 100.000 + target == real Acc@1 0.126 Acc@5 100.000 + target == fake Acc@1 99.902 Acc@5 100.000 + loss 0.741 + PR-AUC 0.7561 ROC-AUC 0.5384 EER 47.34% (thr=0.9268) +Accuracy of the network on the 89879 test images: 73.4% +[rank0]:[W726 04:49:48.701650954 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:50:10.247000 3151374 torch/distributed/run.py:766] +W0726 04:50:10.247000 3151374 torch/distributed/run.py:766] ***************************************** +W0726 04:50:10.247000 3151374 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:50:10.247000 3151374 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:50:43.501311931 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:50:43.001011811 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:50:43.159806482 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +[rank0]:[W726 04:50:44.509338427 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +| distributed init (rank 5): env:// +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:50:44.519752103 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:50:44.525839502 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W726 04:50:44.525893152 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:50:44.531984286 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n5_seed7_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed7_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:24:08 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.7542 (2.7542) acc1: 0.0000 (0.0000) acc5: 100.0000 (100.0000) time: 5.3850 data: 3.9448 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/269] eta: 0:02:34 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.6990 (2.6934) acc1: 0.0000 (0.0000) acc5: 100.0000 (100.0000) time: 0.5951 data: 0.3828 max mem: 3531 +Test: [ 20/269] eta: 0:01:29 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.6767 (2.6776) acc1: 0.0000 (0.0000) acc5: 100.0000 (100.0000) time: 0.1062 data: 0.0135 max mem: 3531 +Test: [ 30/269] eta: 0:01:11 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.6554 (2.6651) acc1: 0.0000 (0.0084) acc5: 100.0000 (100.0000) time: 0.1332 data: 0.0365 max mem: 3531 +Test: [ 40/269] eta: 0:01:07 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.4846 (2.1168) acc1: 0.0000 (21.2208) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2258 data: 0.1305 max mem: 3531 +Test: [ 50/269] eta: 0:00:59 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0706 (1.7160) acc1: 100.0000 (36.6677) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2359 data: 0.1440 max mem: 3531 +Test: [ 60/269] eta: 0:00:53 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0716 (1.4464) acc1: 100.0000 (47.0500) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1803 data: 0.0861 max mem: 3531 +Test: [ 70/269] eta: 0:00:48 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0707 (1.2525) acc1: 100.0000 (54.5078) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1757 data: 0.0812 max mem: 3531 +Test: [ 80/269] eta: 0:00:45 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0687 (1.1064) acc1: 100.0000 (60.1241) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1954 data: 0.1004 max mem: 3531 +Test: [ 90/269] eta: 0:00:42 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0700 (0.9927) acc1: 100.0000 (64.5061) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2096 data: 0.1129 max mem: 3531 +Test: [100/269] eta: 0:00:39 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0703 (0.9014) acc1: 100.0000 (68.0203) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2048 data: 0.1125 max mem: 3531 +Test: [110/269] eta: 0:00:36 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0726 (0.8270) acc1: 100.0000 (70.9014) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1783 data: 0.0863 max mem: 3531 +Test: [120/269] eta: 0:00:33 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0724 (0.7645) acc1: 100.0000 (73.3062) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1805 data: 0.0874 max mem: 3531 +Test: [130/269] eta: 0:00:31 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0716 (0.7118) acc1: 100.0000 (75.3439) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2055 data: 0.1139 max mem: 3531 +Test: [140/269] eta: 0:00:28 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0745 (0.6667) acc1: 100.0000 (77.0926) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1982 data: 0.1070 max mem: 3531 +Test: [150/269] eta: 0:00:26 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0753 (0.6275) acc1: 100.0000 (78.6096) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1902 data: 0.0983 max mem: 3531 +Test: [160/269] eta: 0:00:23 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0756 (0.5932) acc1: 100.0000 (79.9382) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1998 data: 0.1077 max mem: 3531 +Test: [170/269] eta: 0:00:21 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0765 (0.5630) acc1: 100.0000 (81.1114) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2020 data: 0.1108 max mem: 3531 +Test: [180/269] eta: 0:00:19 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0740 (0.5358) acc1: 100.0000 (82.1550) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1844 data: 0.0933 max mem: 3531 +Test: [190/269] eta: 0:00:16 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0708 (0.5116) acc1: 100.0000 (83.0893) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1892 data: 0.0967 max mem: 3531 +Test: [200/269] eta: 0:00:14 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0715 (0.4897) acc1: 100.0000 (83.9306) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2105 data: 0.1165 max mem: 3531 +Test: [210/269] eta: 0:00:12 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0701 (0.4698) acc1: 100.0000 (84.6922) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1995 data: 0.1074 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0696 (0.4518) acc1: 100.0000 (85.3849) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1903 data: 0.0977 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0729 (0.4355) acc1: 100.0000 (86.0164) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1931 data: 0.0987 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0739 (0.4206) acc1: 100.0000 (86.5966) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1877 data: 0.0949 max mem: 3532 +Test: [250/269] eta: 0:00:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0737 (0.4070) acc1: 100.0000 (87.1296) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1663 data: 0.0753 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0751 (0.3942) acc1: 100.0000 (87.6227) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1368 data: 0.0461 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0725 (0.3846) acc1: 100.0000 (87.9905) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1392 data: 0.0355 max mem: 3532 +Test: Total time: 0:00:54 (0.2039 s / it) +* Overall Acc@1 87.990 Acc@5 100.000 + target == real Acc@1 0.008 Acc@5 100.000 + target == fake Acc@1 99.998 Acc@5 100.000 + loss 0.385 + PR-AUC 0.8747 ROC-AUC 0.4950 EER 50.27% (thr=0.9316) +Accuracy of the network on the 103293 test images: 88.0% +[rank0]:[W726 04:51:48.351498957 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T04:51:55Z] eval_done tag=fixed500k_real_n5_seed7_resampled +[2026-07-26T04:51:55Z] eval_start tag=fixed500k_fake_n11_seed7_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed7_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T04:51:55Z +hostname=n116-040-168 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed7_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n11_seed7_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 04:52:07.712000 3151939 torch/distributed/run.py:766] +W0726 04:52:07.712000 3151939 torch/distributed/run.py:766] ***************************************** +W0726 04:52:07.712000 3151939 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:52:07.712000 3151939 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:52:34.605111061 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:52:34.702213601 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:52:35.184654470 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:52:35.234432526 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:52:35.458250845 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +| distributed init (rank 4): env:// +[rank1]:[W726 04:52:35.486403493 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:52:35.493135158 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W726 04:52:35.493162713 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n11_seed7_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed7_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:19:20 acc1_pos: 36.198 acc5_pos: 100.000 loss: 0.8930 (0.8930) acc1: 36.1979 (36.1979) acc5: 100.0000 (100.0000) time: 4.9600 data: 3.3176 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/234] eta: 0:01:59 acc1_pos: 25.781 acc5_pos: 100.000 loss: 0.8930 (0.8152) acc1: 43.7500 (46.0227) acc5: 100.0000 (100.0000) time: 0.5325 data: 0.3019 max mem: 3531 +Test: [ 20/234] eta: 0:01:08 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.8683 (0.7532) acc1: 43.7500 (49.5660) acc5: 100.0000 (100.0000) time: 0.0897 data: 0.0003 max mem: 3531 +Test: [ 30/234] eta: 0:00:50 acc1_pos: 61.198 acc5_pos: 100.000 loss: 0.2035 (0.5953) acc1: 92.9688 (62.6260) acc5: 100.0000 (100.0000) time: 0.0916 data: 0.0003 max mem: 3531 +Test: [ 40/234] eta: 0:00:43 acc1_pos: 97.917 acc5_pos: 100.000 loss: 0.1985 (0.5622) acc1: 94.7917 (65.0851) acc5: 100.0000 (100.0000) time: 0.1267 data: 0.0328 max mem: 3531 +Test: [ 50/234] eta: 0:00:39 acc1_pos: 95.833 acc5_pos: 100.000 loss: 0.1460 (0.4788) acc1: 96.6146 (71.4154) acc5: 100.0000 (100.0000) time: 0.1652 data: 0.0725 max mem: 3531 +Test: [ 60/234] eta: 0:00:36 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.1179 (0.4160) acc1: 98.6979 (76.0587) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1699 data: 0.0769 max mem: 3531 +Test: [ 70/234] eta: 0:00:32 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.0951 (0.4894) acc1: 99.7396 (72.8433) acc5: 100.0000 (100.0000) acc1_neg: 24.479 acc5_neg: 100.000 time: 0.1558 data: 0.0625 max mem: 3531 +Test: [ 80/234] eta: 0:00:29 acc1_pos: 95.862 acc5_pos: 100.000 loss: 1.3139 (0.5510) acc1: 32.0312 (70.2321) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1400 data: 0.0486 max mem: 3531 +Test: [ 90/234] eta: 0:00:27 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.3086 (0.5145) acc1: 89.0625 (72.8280) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1601 data: 0.0689 max mem: 3531 +Test: [100/234] eta: 0:00:25 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.2932 (0.5083) acc1: 88.0208 (73.4891) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1752 data: 0.0828 max mem: 3531 +Test: [110/234] eta: 0:00:23 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.3772 (0.4929) acc1: 83.0729 (74.7279) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1686 data: 0.0759 max mem: 3531 +Test: [120/234] eta: 0:00:21 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.3017 (0.4947) acc1: 86.9792 (75.0129) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.1796 data: 0.0845 max mem: 3531 +Test: [130/234] eta: 0:00:19 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.4329 (0.4943) acc1: 82.8125 (75.4532) acc5: 100.0000 (100.0000) acc1_neg: 69.792 acc5_neg: 100.000 time: 0.1899 data: 0.0933 max mem: 3531 +Test: [140/234] eta: 0:00:17 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.5962 (0.5289) acc1: 71.8750 (73.7736) acc5: 100.0000 (100.0000) acc1_neg: 27.083 acc5_neg: 100.000 time: 0.1799 data: 0.0853 max mem: 3531 +Test: [150/234] eta: 0:00:15 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.6282 (0.5391) acc1: 68.7500 (73.2340) acc5: 100.0000 (100.0000) acc1_neg: 76.042 acc5_neg: 100.000 time: 0.1618 data: 0.0687 max mem: 3531 +Test: [160/234] eta: 0:00:13 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.5413 (0.5375) acc1: 72.1354 (73.4116) acc5: 100.0000 (100.0000) acc1_neg: 72.917 acc5_neg: 100.000 time: 0.1712 data: 0.0795 max mem: 3531 +Test: [170/234] eta: 0:00:11 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.5795 (0.5378) acc1: 72.9167 (73.5060) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1780 data: 0.0845 max mem: 3531 +Test: [180/234] eta: 0:00:09 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.6040 (0.5580) acc1: 69.7917 (72.6260) acc5: 100.0000 (100.0000) acc1_neg: 14.062 acc5_neg: 100.000 time: 0.1889 data: 0.0953 max mem: 3531 +Test: [190/234] eta: 0:00:08 acc1_pos: 95.862 acc5_pos: 100.000 loss: 1.3940 (0.6037) acc1: 30.7292 (70.3589) acc5: 100.0000 (100.0000) acc1_neg: 30.729 acc5_neg: 100.000 time: 0.2060 data: 0.1146 max mem: 3531 +Test: [200/234] eta: 0:00:06 acc1_pos: 95.862 acc5_pos: 100.000 loss: 1.3698 (0.6036) acc1: 30.7292 (70.5600) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2095 data: 0.1172 max mem: 3531 +Test: [210/234] eta: 0:00:04 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.1359 (0.5870) acc1: 98.9583 (71.5961) acc5: 100.0000 (100.0000) acc1_neg: 78.646 acc5_neg: 100.000 time: 0.2045 data: 0.1096 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.4675 (0.5841) acc1: 79.4271 (71.8243) acc5: 100.0000 (100.0000) acc1_neg: 60.938 acc5_neg: 100.000 time: 0.2149 data: 0.1180 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.8700 (0.6391) acc1: 13.5417 (69.1930) acc5: 100.0000 (100.0000) acc1_neg: 13.281 acc5_neg: 100.000 time: 0.2064 data: 0.1108 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 95.862 acc5_pos: 100.000 loss: 1.7967 (0.6542) acc1: 12.7604 (68.5286) acc5: 100.0000 (100.0000) acc1_neg: 12.840 acc5_neg: 100.000 time: 0.2045 data: 0.1079 max mem: 3532 +Test: Total time: 0:00:44 (0.1885 s / it) +* Overall Acc@1 68.529 Acc@5 100.000 + target == real Acc@1 71.595 Acc@5 100.000 + target == fake Acc@1 67.664 Acc@5 100.000 + loss 0.654 + PR-AUC 0.9332 ROC-AUC 0.7871 EER 30.76% (thr=0.4692) +Accuracy of the network on the 89729 test images: 68.5% +[rank0]:[W726 04:53:31.419917056 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:53:53.099000 3152497 torch/distributed/run.py:766] +W0726 04:53:53.099000 3152497 torch/distributed/run.py:766] ***************************************** +W0726 04:53:53.099000 3152497 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:53:53.099000 3152497 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:54:27.049782642 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:54:28.224748326 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:54:28.402766604 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:54:28.684715325 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:54:28.715598046 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:54:28.776777336 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:54:28.813687760 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:54:28.818701473 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n11_seed7_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed7_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 3:51:34 acc1_pos: 35.677 acc5_pos: 100.000 loss: 0.9114 (0.9114) acc1: 35.6771 (35.6771) acc5: 100.0000 (100.0000) time: 5.8211 data: 3.8365 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/2387] eta: 0:24:14 acc1_pos: 16.146 acc5_pos: 100.000 loss: 1.2961 (1.2912) acc1: 14.0625 (15.1515) acc5: 100.0000 (100.0000) time: 0.6117 data: 0.3500 max mem: 3531 +Test: [ 20/2387] eta: 0:14:46 acc1_pos: 8.854 acc5_pos: 100.000 loss: 1.2965 (1.2808) acc1: 13.8021 (15.7490) acc5: 100.0000 (100.0000) time: 0.1022 data: 0.0127 max mem: 3531 +Test: [ 30/2387] eta: 0:11:26 acc1_pos: 1.042 acc5_pos: 100.000 loss: 1.3299 (1.3242) acc1: 10.1562 (13.1972) acc5: 100.0000 (100.0000) time: 0.1150 data: 0.0247 max mem: 3531 +Test: [ 40/2387] eta: 0:11:46 acc1_pos: 13.021 acc5_pos: 100.000 loss: 1.3299 (1.3230) acc1: 7.0312 (12.0554) acc5: 100.0000 (100.0000) time: 0.2240 data: 0.1303 max mem: 3531 +Test: [ 50/2387] eta: 0:11:23 acc1_pos: 35.938 acc5_pos: 100.000 loss: 1.2432 (1.2828) acc1: 11.4583 (13.9706) acc5: 100.0000 (100.0000) time: 0.2944 data: 0.1991 max mem: 3531 +Test: [ 60/2387] eta: 0:11:04 acc1_pos: 28.125 acc5_pos: 100.000 loss: 1.1270 (1.2601) acc1: 19.2708 (14.8822) acc5: 100.0000 (100.0000) time: 0.2542 data: 0.1610 max mem: 3531 +Test: [ 70/2387] eta: 0:10:41 acc1_pos: 30.990 acc5_pos: 100.000 loss: 1.1611 (1.2616) acc1: 18.2292 (14.9391) acc5: 100.0000 (100.0000) time: 0.2372 data: 0.1450 max mem: 3531 +Test: [ 80/2387] eta: 0:10:21 acc1_pos: 13.542 acc5_pos: 100.000 loss: 1.2422 (1.2590) acc1: 13.5417 (14.9113) acc5: 100.0000 (100.0000) time: 0.2203 data: 0.1292 max mem: 3531 +Test: [ 90/2387] eta: 0:10:16 acc1_pos: 18.750 acc5_pos: 100.000 loss: 1.2058 (1.2510) acc1: 14.8438 (15.2988) acc5: 100.0000 (100.0000) time: 0.2376 data: 0.1448 max mem: 3531 +Test: [ 100/2387] eta: 0:10:07 acc1_pos: 13.802 acc5_pos: 100.000 loss: 1.2359 (1.2547) acc1: 14.8438 (14.9985) acc5: 100.0000 (100.0000) time: 0.2507 data: 0.1555 max mem: 3531 +Test: [ 110/2387] eta: 0:10:02 acc1_pos: 24.479 acc5_pos: 100.000 loss: 1.2359 (1.2526) acc1: 13.5417 (15.0690) acc5: 100.0000 (100.0000) time: 0.2470 data: 0.1520 max mem: 3531 +Test: [ 120/2387] eta: 0:09:59 acc1_pos: 21.875 acc5_pos: 100.000 loss: 1.2197 (1.2513) acc1: 14.5833 (15.0654) acc5: 100.0000 (100.0000) time: 0.2581 data: 0.1633 max mem: 3531 +Test: [ 130/2387] eta: 0:09:49 acc1_pos: 17.969 acc5_pos: 100.000 loss: 1.2485 (1.2558) acc1: 14.5833 (14.8716) acc5: 100.0000 (100.0000) time: 0.2446 data: 0.1487 max mem: 3531 +Test: [ 140/2387] eta: 0:09:43 acc1_pos: 17.188 acc5_pos: 100.000 loss: 1.2086 (1.2545) acc1: 14.0625 (14.9490) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1339 max mem: 3531 +Test: [ 150/2387] eta: 0:09:49 acc1_pos: 5.990 acc5_pos: 100.000 loss: 1.1973 (1.2509) acc1: 13.2812 (14.9248) acc5: 100.0000 (100.0000) time: 0.2782 data: 0.1818 max mem: 3531 +Test: [ 160/2387] eta: 0:09:28 acc1_pos: 25.114 acc5_pos: 100.000 loss: 1.1583 (1.2273) acc1: 14.5833 (16.8397) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2252 data: 0.1273 max mem: 3531 +Test: [ 170/2387] eta: 0:09:17 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.2611 (1.1704) acc1: 94.7917 (21.4943) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1605 data: 0.0634 max mem: 3531 +Test: [ 180/2387] eta: 0:09:10 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.2505 (1.1196) acc1: 96.3542 (25.6287) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2020 data: 0.1055 max mem: 3531 +Test: [ 190/2387] eta: 0:08:59 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.2507 (1.0742) acc1: 96.3542 (29.3426) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1947 data: 0.1003 max mem: 3531 +Test: [ 200/2387] eta: 0:08:50 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.2513 (1.0332) acc1: 96.6146 (32.6881) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1788 data: 0.0842 max mem: 3531 +Test: [ 210/2387] eta: 0:08:39 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.2513 (0.9962) acc1: 96.3542 (35.7042) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1722 data: 0.0764 max mem: 3532 +Test: [ 220/2387] eta: 0:08:32 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.2546 (0.9626) acc1: 96.3542 (38.4545) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1749 data: 0.0815 max mem: 3532 +Test: [ 230/2387] eta: 0:08:29 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.2514 (0.9318) acc1: 96.3542 (40.9869) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2085 data: 0.1152 max mem: 3532 +Test: [ 240/2387] eta: 0:08:20 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.2421 (0.9028) acc1: 97.9167 (43.3524) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1993 data: 0.1046 max mem: 3532 +Test: [ 250/2387] eta: 0:08:06 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.2247 (0.8735) acc1: 97.9167 (45.5293) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1335 data: 0.0371 max mem: 3532 +Test: [ 260/2387] eta: 0:07:59 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1507 (0.8454) acc1: 98.6979 (47.5615) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1342 data: 0.0369 max mem: 3532 +Test: [ 270/2387] eta: 0:07:54 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1359 (0.8194) acc1: 99.2188 (49.4648) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1765 data: 0.0813 max mem: 3532 +Test: [ 280/2387] eta: 0:07:50 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1359 (0.7955) acc1: 99.2188 (51.2103) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1901 data: 0.0973 max mem: 3532 +Test: [ 290/2387] eta: 0:07:44 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1357 (0.7728) acc1: 99.2188 (52.8637) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1890 data: 0.0951 max mem: 3532 +Test: [ 300/2387] eta: 0:07:38 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1249 (0.7513) acc1: 99.7396 (54.4098) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1716 data: 0.0779 max mem: 3532 +Test: [ 310/2387] eta: 0:07:32 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1206 (0.7350) acc1: 100.0000 (55.7334) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1651 data: 0.0728 max mem: 3532 +Test: [ 320/2387] eta: 0:07:30 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.3889 (0.7245) acc1: 91.4062 (56.8195) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.1930 data: 0.1004 max mem: 3532 +Test: [ 330/2387] eta: 0:07:28 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.3946 (0.7144) acc1: 91.1458 (57.8676) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.2141 data: 0.1201 max mem: 3532 +Test: [ 340/2387] eta: 0:07:26 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.3933 (0.7051) acc1: 91.4062 (58.8450) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.2203 data: 0.1234 max mem: 3532 +Test: [ 350/2387] eta: 0:07:22 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.3805 (0.6916) acc1: 92.4479 (59.9841) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2049 data: 0.1075 max mem: 3532 +Test: [ 360/2387] eta: 0:07:13 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.2150 (0.6785) acc1: 98.6979 (61.0371) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1396 data: 0.0412 max mem: 3532 +Test: [ 370/2387] eta: 0:07:08 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.2215 (0.6663) acc1: 97.9167 (62.0213) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1349 data: 0.0385 max mem: 3532 +Test: [ 380/2387] eta: 0:07:06 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.2224 (0.6546) acc1: 97.9167 (62.9696) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1862 data: 0.0939 max mem: 3532 +Test: [ 390/2387] eta: 0:07:04 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.2213 (0.6435) acc1: 97.9167 (63.8607) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2117 data: 0.1211 max mem: 3532 +Test: [ 400/2387] eta: 0:06:58 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.2186 (0.6329) acc1: 97.9167 (64.7165) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1808 data: 0.0903 max mem: 3532 +Test: [ 410/2387] eta: 0:06:55 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.2173 (0.6229) acc1: 97.9167 (65.5300) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1581 data: 0.0667 max mem: 3532 +Test: [ 420/2387] eta: 0:06:54 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.2173 (0.6132) acc1: 98.1771 (66.3067) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2087 data: 0.1171 max mem: 3532 +Test: [ 430/2387] eta: 0:06:49 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.2161 (0.6041) acc1: 98.1771 (67.0479) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1958 data: 0.1039 max mem: 3532 +Test: [ 440/2387] eta: 0:06:45 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.2104 (0.5942) acc1: 98.6979 (67.7798) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1590 data: 0.0674 max mem: 3533 +Test: [ 450/2387] eta: 0:06:42 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1399 (0.5841) acc1: 100.0000 (68.4913) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1797 data: 0.0892 max mem: 3533 +Test: [ 460/2387] eta: 0:06:39 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1411 (0.5745) acc1: 100.0000 (69.1720) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1869 data: 0.0944 max mem: 3533 +Test: [ 470/2387] eta: 0:06:36 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1417 (0.5654) acc1: 100.0000 (69.8237) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1802 data: 0.0876 max mem: 3533 +Test: [ 480/2387] eta: 0:06:33 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1433 (0.5566) acc1: 100.0000 (70.4489) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1848 data: 0.0930 max mem: 3533 +Test: [ 490/2387] eta: 0:06:31 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1426 (0.5482) acc1: 100.0000 (71.0497) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1906 data: 0.0977 max mem: 3533 +Test: [ 500/2387] eta: 0:06:30 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1410 (0.5401) acc1: 100.0000 (71.6234) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2192 data: 0.1256 max mem: 3533 +Test: [ 510/2387] eta: 0:06:27 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1435 (0.5324) acc1: 100.0000 (72.1787) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2140 data: 0.1208 max mem: 3533 +Test: [ 520/2387] eta: 0:06:23 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1555 (0.5253) acc1: 100.0000 (72.7117) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1637 data: 0.0668 max mem: 3533 +Test: [ 530/2387] eta: 0:06:20 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1496 (0.5182) acc1: 100.0000 (73.2256) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1712 data: 0.0731 max mem: 3533 +Test: [ 540/2387] eta: 0:06:19 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1479 (0.5113) acc1: 100.0000 (73.7205) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2062 data: 0.1117 max mem: 3533 +Test: [ 550/2387] eta: 0:06:16 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1479 (0.5047) acc1: 100.0000 (74.1975) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1937 data: 0.1007 max mem: 3533 +Test: [ 560/2387] eta: 0:06:13 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1482 (0.4984) acc1: 100.0000 (74.6560) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1804 data: 0.0888 max mem: 3533 +Test: [ 570/2387] eta: 0:06:10 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1475 (0.4922) acc1: 100.0000 (75.0999) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1790 data: 0.0876 max mem: 3533 +Test: [ 580/2387] eta: 0:06:07 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1473 (0.4863) acc1: 100.0000 (75.5285) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1713 data: 0.0799 max mem: 3533 +Test: [ 590/2387] eta: 0:06:04 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1473 (0.4806) acc1: 100.0000 (75.9425) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0888 max mem: 3533 +Test: [ 600/2387] eta: 0:06:01 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1532 (0.4754) acc1: 100.0000 (76.3415) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1734 data: 0.0830 max mem: 3533 +Test: [ 610/2387] eta: 0:05:59 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1714 (0.4707) acc1: 99.7396 (76.7147) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1788 data: 0.0881 max mem: 3533 +Test: [ 620/2387] eta: 0:05:56 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1877 (0.4661) acc1: 99.4792 (77.0817) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1838 data: 0.0926 max mem: 3533 +Test: [ 630/2387] eta: 0:05:54 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1877 (0.4617) acc1: 99.4792 (77.4358) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1800 data: 0.0891 max mem: 3533 +Test: [ 640/2387] eta: 0:05:51 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1904 (0.4575) acc1: 99.4792 (77.7781) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1814 data: 0.0906 max mem: 3533 +Test: [ 650/2387] eta: 0:05:49 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1909 (0.4534) acc1: 99.4792 (78.1090) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1880 data: 0.0963 max mem: 3533 +Test: [ 660/2387] eta: 0:05:46 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1900 (0.4494) acc1: 99.4792 (78.4327) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1955 data: 0.1018 max mem: 3533 +Test: [ 670/2387] eta: 0:05:45 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1890 (0.4455) acc1: 99.4792 (78.7444) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2139 data: 0.1204 max mem: 3534 +Test: [ 680/2387] eta: 0:05:42 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1925 (0.4418) acc1: 99.4792 (79.0504) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1961 data: 0.1049 max mem: 3534 +Test: [ 690/2387] eta: 0:05:40 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.2044 (0.4395) acc1: 98.9583 (79.2183) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1802 data: 0.0876 max mem: 3534 +Test: [ 700/2387] eta: 0:05:39 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.3016 (0.4377) acc1: 86.9792 (79.3298) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.2098 data: 0.1141 max mem: 3534 +Test: [ 710/2387] eta: 0:05:36 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.3073 (0.4358) acc1: 86.9792 (79.4469) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1981 data: 0.1018 max mem: 3534 +Test: [ 720/2387] eta: 0:05:34 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.2955 (0.4339) acc1: 87.7604 (79.5618) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.1949 data: 0.0995 max mem: 3534 +Test: [ 730/2387] eta: 0:05:32 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.3050 (0.4323) acc1: 87.2396 (79.6576) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.1952 data: 0.1010 max mem: 3534 +Test: [ 740/2387] eta: 0:05:29 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.3061 (0.4306) acc1: 87.2396 (79.7694) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1792 data: 0.0866 max mem: 3534 +Test: [ 750/2387] eta: 0:05:27 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.2965 (0.4288) acc1: 88.2812 (79.8800) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1754 data: 0.0834 max mem: 3534 +Test: [ 760/2387] eta: 0:05:24 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.2953 (0.4271) acc1: 88.2812 (79.9904) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1707 data: 0.0788 max mem: 3534 +Test: [ 770/2387] eta: 0:05:22 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.2657 (0.4245) acc1: 90.6250 (80.1708) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1733 data: 0.0826 max mem: 3534 +Test: [ 780/2387] eta: 0:05:19 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1896 (0.4212) acc1: 97.6562 (80.4031) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1762 data: 0.0858 max mem: 3534 +Test: [ 790/2387] eta: 0:05:16 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1699 (0.4186) acc1: 98.1771 (80.5787) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1597 data: 0.0684 max mem: 3534 +Test: [ 800/2387] eta: 0:05:14 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1560 (0.4153) acc1: 98.1771 (80.7961) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1578 data: 0.0650 max mem: 3534 +Test: [ 810/2387] eta: 0:05:12 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1479 (0.4121) acc1: 98.9583 (81.0166) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1847 data: 0.0925 max mem: 3534 +Test: [ 820/2387] eta: 0:05:09 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1631 (0.4094) acc1: 98.6979 (81.2205) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1864 data: 0.0952 max mem: 3534 +Test: [ 830/2387] eta: 0:05:06 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1905 (0.4071) acc1: 97.6562 (81.3973) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1656 data: 0.0749 max mem: 3534 +Test: [ 840/2387] eta: 0:05:04 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.3441 (0.4075) acc1: 85.6771 (81.3708) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1686 data: 0.0775 max mem: 3534 +Test: [ 850/2387] eta: 0:05:02 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.4238 (0.4080) acc1: 80.2083 (81.3290) acc5: 100.0000 (100.0000) acc1_neg: 74.219 acc5_neg: 100.000 time: 0.1794 data: 0.0884 max mem: 3534 +Test: [ 860/2387] eta: 0:04:59 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.4196 (0.4082) acc1: 78.9062 (81.2981) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.1752 data: 0.0836 max mem: 3534 +Test: [ 870/2387] eta: 0:04:57 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.4059 (0.4081) acc1: 81.5104 (81.3253) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1763 data: 0.0833 max mem: 3534 +Test: [ 880/2387] eta: 0:04:55 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.4027 (0.4083) acc1: 83.8542 (81.3325) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1858 data: 0.0942 max mem: 3534 +Test: [ 890/2387] eta: 0:04:53 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.3396 (0.4068) acc1: 87.5000 (81.4835) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1882 data: 0.0956 max mem: 3534 +Test: [ 900/2387] eta: 0:04:51 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.3396 (0.4114) acc1: 87.5000 (81.1890) acc5: 100.0000 (100.0000) acc1_neg: 50.521 acc5_neg: 100.000 time: 0.1799 data: 0.0856 max mem: 3535 +Test: [ 910/2387] eta: 0:04:48 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.9316 (0.4173) acc1: 45.3125 (80.7966) acc5: 100.0000 (100.0000) acc1_neg: 44.531 acc5_neg: 100.000 time: 0.1758 data: 0.0828 max mem: 3535 +Test: [ 920/2387] eta: 0:04:46 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.9514 (0.4231) acc1: 44.5312 (80.4032) acc5: 100.0000 (100.0000) acc1_neg: 41.146 acc5_neg: 100.000 time: 0.1837 data: 0.0917 max mem: 3535 +Test: [ 930/2387] eta: 0:04:44 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.9536 (0.4288) acc1: 45.0521 (80.0178) acc5: 100.0000 (100.0000) acc1_neg: 40.625 acc5_neg: 100.000 time: 0.1890 data: 0.0958 max mem: 3535 +Test: [ 940/2387] eta: 0:04:43 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.9547 (0.4344) acc1: 45.0521 (79.6518) acc5: 100.0000 (100.0000) acc1_neg: 45.573 acc5_neg: 100.000 time: 0.1985 data: 0.1054 max mem: 3535 +Test: [ 950/2387] eta: 0:04:41 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.9541 (0.4398) acc1: 44.5312 (79.2800) acc5: 100.0000 (100.0000) acc1_neg: 46.875 acc5_neg: 100.000 time: 0.2071 data: 0.1164 max mem: 3535 +Test: [ 960/2387] eta: 0:04:39 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.9374 (0.4449) acc1: 45.3125 (78.9315) acc5: 100.0000 (100.0000) acc1_neg: 48.438 acc5_neg: 100.000 time: 0.2065 data: 0.1139 max mem: 3535 +Test: [ 970/2387] eta: 0:04:37 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.9376 (0.4501) acc1: 46.0938 (78.5852) acc5: 100.0000 (100.0000) acc1_neg: 46.354 acc5_neg: 100.000 time: 0.2090 data: 0.1163 max mem: 3535 +Test: [ 980/2387] eta: 0:04:35 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.9177 (0.4498) acc1: 46.6146 (78.6015) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.2099 data: 0.1194 max mem: 3535 +Test: [ 990/2387] eta: 0:04:33 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.2595 (0.4482) acc1: 92.9688 (78.7278) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.1985 data: 0.1074 max mem: 3535 +Test: [1000/2387] eta: 0:04:31 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.2882 (0.4473) acc1: 89.5833 (78.7939) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1946 data: 0.1038 max mem: 3535 +Test: [1010/2387] eta: 0:04:29 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.3195 (0.4462) acc1: 87.5000 (78.8831) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1968 data: 0.1066 max mem: 3535 +Test: [1020/2387] eta: 0:04:28 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.3148 (0.4444) acc1: 89.3229 (79.0139) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2009 data: 0.1105 max mem: 3535 +Test: [1030/2387] eta: 0:04:26 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1839 (0.4419) acc1: 96.6146 (79.1785) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2008 data: 0.1097 max mem: 3535 +Test: [1040/2387] eta: 0:04:24 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1611 (0.4394) acc1: 98.9583 (79.3603) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1991 data: 0.1080 max mem: 3535 +Test: [1050/2387] eta: 0:04:22 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1578 (0.4367) acc1: 98.9583 (79.5438) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1952 data: 0.1045 max mem: 3535 +Test: [1060/2387] eta: 0:04:20 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1587 (0.4340) acc1: 98.9583 (79.7282) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2010 data: 0.1104 max mem: 3535 +Test: [1070/2387] eta: 0:04:18 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1592 (0.4315) acc1: 99.2188 (79.9046) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2042 data: 0.1122 max mem: 3535 +Test: [1080/2387] eta: 0:04:16 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1767 (0.4294) acc1: 97.9167 (80.0539) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1998 data: 0.1077 max mem: 3535 +Test: [1090/2387] eta: 0:04:14 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1842 (0.4273) acc1: 97.3958 (80.2052) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1999 data: 0.1068 max mem: 3535 +Test: [1100/2387] eta: 0:04:12 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.2082 (0.4254) acc1: 96.0938 (80.3408) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1756 data: 0.0824 max mem: 3535 +Test: [1110/2387] eta: 0:04:09 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.2125 (0.4237) acc1: 95.8333 (80.4488) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1625 data: 0.0709 max mem: 3535 +Test: [1120/2387] eta: 0:04:07 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.2163 (0.4218) acc1: 95.8333 (80.5837) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1589 data: 0.0657 max mem: 3536 +Test: [1130/2387] eta: 0:04:05 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.2163 (0.4200) acc1: 96.0938 (80.7218) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1567 data: 0.0638 max mem: 3536 +Test: [1140/2387] eta: 0:04:02 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.2105 (0.4182) acc1: 96.6146 (80.8645) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1653 data: 0.0736 max mem: 3536 +Test: [1150/2387] eta: 0:04:00 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.2309 (0.4171) acc1: 95.5729 (80.9699) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.1706 data: 0.0792 max mem: 3536 +Test: [1160/2387] eta: 0:03:58 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.2568 (0.4157) acc1: 94.2708 (81.0923) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1784 data: 0.0878 max mem: 3536 +Test: [1170/2387] eta: 0:03:56 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.2676 (0.4143) acc1: 94.2708 (81.2104) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1724 data: 0.0805 max mem: 3536 +Test: [1180/2387] eta: 0:03:54 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.2795 (0.4132) acc1: 95.3125 (81.3201) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1604 data: 0.0681 max mem: 3536 +Test: [1190/2387] eta: 0:03:51 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.2503 (0.4118) acc1: 95.3125 (81.4429) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1646 data: 0.0731 max mem: 3536 +Test: [1200/2387] eta: 0:03:49 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.2049 (0.4101) acc1: 97.6562 (81.5772) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1660 data: 0.0747 max mem: 3536 +Test: [1210/2387] eta: 0:03:47 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1907 (0.4084) acc1: 99.2188 (81.7059) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1707 data: 0.0783 max mem: 3536 +Test: [1220/2387] eta: 0:03:45 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1772 (0.4064) acc1: 98.1771 (81.8308) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1730 data: 0.0801 max mem: 3536 +Test: [1230/2387] eta: 0:03:43 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1635 (0.4045) acc1: 96.8750 (81.9509) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1639 data: 0.0726 max mem: 3536 +Test: [1240/2387] eta: 0:03:41 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1635 (0.4026) acc1: 96.8750 (82.0699) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1641 data: 0.0731 max mem: 3536 +Test: [1250/2387] eta: 0:03:38 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1523 (0.4006) acc1: 97.1354 (82.1949) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1667 data: 0.0755 max mem: 3536 +Test: [1260/2387] eta: 0:03:36 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1478 (0.3987) acc1: 97.9167 (82.3169) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1698 data: 0.0757 max mem: 3536 +Test: [1270/2387] eta: 0:03:34 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1228 (0.3965) acc1: 99.7396 (82.4548) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1684 data: 0.0732 max mem: 3536 +Test: [1280/2387] eta: 0:03:32 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1177 (0.3943) acc1: 100.0000 (82.5903) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1859 data: 0.0912 max mem: 3536 +Test: [1290/2387] eta: 0:03:30 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1280 (0.3924) acc1: 99.7396 (82.7078) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2012 data: 0.1075 max mem: 3536 +Test: [1300/2387] eta: 0:03:28 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1282 (0.3904) acc1: 99.4792 (82.8309) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1860 data: 0.0944 max mem: 3536 +Test: [1310/2387] eta: 0:03:27 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1326 (0.3885) acc1: 99.4792 (82.9502) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2006 data: 0.1091 max mem: 3536 +Test: [1320/2387] eta: 0:03:24 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1413 (0.3867) acc1: 99.2188 (83.0656) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1885 data: 0.0981 max mem: 3536 +Test: [1330/2387] eta: 0:03:22 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1416 (0.3849) acc1: 99.4792 (83.1864) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1674 data: 0.0769 max mem: 3536 +Test: [1340/2387] eta: 0:03:20 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1485 (0.3843) acc1: 98.9583 (83.2298) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1709 data: 0.0801 max mem: 3536 +Test: [1350/2387] eta: 0:03:18 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1064 (0.3821) acc1: 100.0000 (83.3540) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1712 data: 0.0787 max mem: 3537 +Test: [1360/2387] eta: 0:03:16 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.0883 (0.3800) acc1: 100.0000 (83.4763) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1742 data: 0.0818 max mem: 3537 +Test: [1370/2387] eta: 0:03:14 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.0914 (0.3779) acc1: 100.0000 (83.5964) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1802 data: 0.0894 max mem: 3537 +Test: [1380/2387] eta: 0:03:12 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.0949 (0.3759) acc1: 100.0000 (83.7152) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1938 data: 0.1030 max mem: 3537 +Test: [1390/2387] eta: 0:03:10 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.0960 (0.3739) acc1: 100.0000 (83.8323) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1831 data: 0.0882 max mem: 3537 +Test: [1400/2387] eta: 0:03:08 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.0907 (0.3718) acc1: 100.0000 (83.9477) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1560 data: 0.0602 max mem: 3537 +Test: [1410/2387] eta: 0:03:06 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.0887 (0.3700) acc1: 100.0000 (84.0581) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1556 data: 0.0645 max mem: 3537 +Test: [1420/2387] eta: 0:03:04 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1277 (0.3686) acc1: 99.7396 (84.1485) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1838 data: 0.0929 max mem: 3537 +Test: [1430/2387] eta: 0:03:02 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1302 (0.3669) acc1: 99.7396 (84.2551) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1918 data: 0.0996 max mem: 3537 +Test: [1440/2387] eta: 0:03:00 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1334 (0.3653) acc1: 99.7396 (84.3624) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1749 data: 0.0825 max mem: 3537 +Test: [1450/2387] eta: 0:02:58 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1421 (0.3639) acc1: 99.7396 (84.4633) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1686 data: 0.0767 max mem: 3537 +Test: [1460/2387] eta: 0:02:56 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1637 (0.3627) acc1: 98.6979 (84.5618) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1649 data: 0.0727 max mem: 3537 +Test: [1470/2387] eta: 0:02:54 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1600 (0.3613) acc1: 98.9583 (84.6606) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1758 data: 0.0824 max mem: 3537 +Test: [1480/2387] eta: 0:02:52 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1514 (0.3600) acc1: 99.2188 (84.7502) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1976 data: 0.1051 max mem: 3537 +Test: [1490/2387] eta: 0:02:50 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1633 (0.3587) acc1: 98.4375 (84.8448) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1785 data: 0.0871 max mem: 3537 +Test: [1500/2387] eta: 0:02:48 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1736 (0.3574) acc1: 98.4375 (84.9376) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1728 data: 0.0811 max mem: 3537 +Test: [1510/2387] eta: 0:02:46 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1774 (0.3563) acc1: 99.2188 (85.0316) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1840 data: 0.0901 max mem: 3537 +Test: [1520/2387] eta: 0:02:44 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.2039 (0.3565) acc1: 96.3542 (85.0138) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.1716 data: 0.0781 max mem: 3537 +Test: [1530/2387] eta: 0:02:42 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.3985 (0.3571) acc1: 82.8125 (84.9780) acc5: 100.0000 (100.0000) acc1_neg: 72.656 acc5_neg: 100.000 time: 0.1770 data: 0.0861 max mem: 3537 +Test: [1540/2387] eta: 0:02:40 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.4249 (0.3574) acc1: 78.9062 (84.9533) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.1946 data: 0.0993 max mem: 3537 +Test: [1550/2387] eta: 0:02:39 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.3952 (0.3576) acc1: 82.2917 (84.9440) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1928 data: 0.0976 max mem: 3537 +Test: [1560/2387] eta: 0:02:37 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.3744 (0.3579) acc1: 82.8125 (84.9370) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.1798 data: 0.0889 max mem: 3537 +Test: [1570/2387] eta: 0:02:35 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.3349 (0.3573) acc1: 89.5833 (85.0024) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1904 data: 0.0997 max mem: 3537 +Test: [1580/2387] eta: 0:02:33 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.2168 (0.3562) acc1: 97.9167 (85.0798) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1930 data: 0.1017 max mem: 3538 +Test: [1590/2387] eta: 0:02:31 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1443 (0.3548) acc1: 99.4792 (85.1700) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1832 data: 0.0915 max mem: 3538 +Test: [1600/2387] eta: 0:02:29 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1288 (0.3537) acc1: 99.2188 (85.2262) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1951 data: 0.1027 max mem: 3538 +Test: [1610/2387] eta: 0:02:27 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1360 (0.3524) acc1: 98.9583 (85.3121) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2020 data: 0.1093 max mem: 3538 +Test: [1620/2387] eta: 0:02:25 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1400 (0.3512) acc1: 98.9583 (85.3948) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1970 data: 0.1046 max mem: 3538 +Test: [1630/2387] eta: 0:02:23 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1673 (0.3501) acc1: 98.4375 (85.4729) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1914 data: 0.1001 max mem: 3538 +Test: [1640/2387] eta: 0:02:21 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1675 (0.3491) acc1: 98.6979 (85.5452) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1848 data: 0.0947 max mem: 3538 +Test: [1650/2387] eta: 0:02:20 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1670 (0.3480) acc1: 98.1771 (85.6195) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1991 data: 0.1086 max mem: 3538 +Test: [1660/2387] eta: 0:02:18 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1723 (0.3470) acc1: 98.1771 (85.6926) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1883 data: 0.0966 max mem: 3538 +Test: [1670/2387] eta: 0:02:16 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1615 (0.3459) acc1: 98.4375 (85.7695) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1884 data: 0.0948 max mem: 3538 +Test: [1680/2387] eta: 0:02:14 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1632 (0.3450) acc1: 98.6979 (85.8363) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1947 data: 0.0987 max mem: 3538 +Test: [1690/2387] eta: 0:02:12 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1708 (0.3440) acc1: 98.4375 (85.9075) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1731 data: 0.0783 max mem: 3538 +Test: [1700/2387] eta: 0:02:10 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1730 (0.3431) acc1: 98.4375 (85.9649) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1823 data: 0.0912 max mem: 3538 +Test: [1710/2387] eta: 0:02:08 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1632 (0.3423) acc1: 97.6562 (86.0139) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1876 data: 0.0972 max mem: 3538 +Test: [1720/2387] eta: 0:02:06 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1632 (0.3414) acc1: 95.8333 (86.0687) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1922 data: 0.0997 max mem: 3538 +Test: [1730/2387] eta: 0:02:04 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1922 (0.3406) acc1: 95.0521 (86.1177) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1870 data: 0.0933 max mem: 3538 +Test: [1740/2387] eta: 0:02:02 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1764 (0.3397) acc1: 95.3125 (86.1759) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1774 data: 0.0836 max mem: 3538 +Test: [1750/2387] eta: 0:02:00 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1578 (0.3387) acc1: 97.9167 (86.2385) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1869 data: 0.0940 max mem: 3538 +Test: [1760/2387] eta: 0:01:59 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1417 (0.3375) acc1: 99.2188 (86.3144) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1886 data: 0.0953 max mem: 3538 +Test: [1770/2387] eta: 0:01:57 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1505 (0.3369) acc1: 98.4375 (86.3520) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1868 data: 0.0939 max mem: 3538 +Test: [1780/2387] eta: 0:01:55 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1865 (0.3360) acc1: 95.3125 (86.4032) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1851 data: 0.0930 max mem: 3538 +Test: [1790/2387] eta: 0:01:53 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1829 (0.3352) acc1: 96.0938 (86.4579) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1657 data: 0.0731 max mem: 3538 +Test: [1800/2387] eta: 0:01:51 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1741 (0.3344) acc1: 96.0938 (86.5072) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1751 data: 0.0837 max mem: 3538 +Test: [1810/2387] eta: 0:01:49 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1726 (0.3335) acc1: 96.8750 (86.5670) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1872 data: 0.0961 max mem: 3539 +Test: [1820/2387] eta: 0:01:47 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1442 (0.3325) acc1: 98.6979 (86.6314) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1851 data: 0.0932 max mem: 3539 +Test: [1830/2387] eta: 0:01:45 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1212 (0.3313) acc1: 99.4792 (86.7030) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1756 data: 0.0835 max mem: 3539 +Test: [1840/2387] eta: 0:01:43 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1182 (0.3302) acc1: 100.0000 (86.7751) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1660 data: 0.0748 max mem: 3539 +Test: [1850/2387] eta: 0:01:41 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1175 (0.3290) acc1: 100.0000 (86.8465) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1793 data: 0.0881 max mem: 3539 +Test: [1860/2387] eta: 0:01:39 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1172 (0.3279) acc1: 100.0000 (86.9172) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0888 max mem: 3539 +Test: [1870/2387] eta: 0:01:37 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1165 (0.3267) acc1: 100.0000 (86.9870) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1848 data: 0.0943 max mem: 3539 +Test: [1880/2387] eta: 0:01:35 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1160 (0.3256) acc1: 100.0000 (87.0560) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1841 data: 0.0933 max mem: 3539 +Test: [1890/2387] eta: 0:01:34 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1168 (0.3245) acc1: 100.0000 (87.1245) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0885 max mem: 3539 +Test: [1900/2387] eta: 0:01:32 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1175 (0.3234) acc1: 100.0000 (87.1922) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1828 data: 0.0903 max mem: 3539 +Test: [1910/2387] eta: 0:01:30 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1189 (0.3224) acc1: 100.0000 (87.2563) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1807 data: 0.0888 max mem: 3539 +Test: [1920/2387] eta: 0:01:28 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1201 (0.3215) acc1: 100.0000 (87.3112) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1764 data: 0.0846 max mem: 3539 +Test: [1930/2387] eta: 0:01:26 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1197 (0.3205) acc1: 100.0000 (87.3744) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1755 data: 0.0830 max mem: 3539 +Test: [1940/2387] eta: 0:01:24 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1130 (0.3198) acc1: 100.0000 (87.4041) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1795 data: 0.0881 max mem: 3539 +Test: [1950/2387] eta: 0:01:22 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1165 (0.3188) acc1: 100.0000 (87.4669) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1854 data: 0.0942 max mem: 3539 +Test: [1960/2387] eta: 0:01:20 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1395 (0.3180) acc1: 100.0000 (87.5256) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1913 data: 0.1000 max mem: 3539 +Test: [1970/2387] eta: 0:01:18 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1732 (0.3174) acc1: 99.4792 (87.5807) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1955 data: 0.1042 max mem: 3539 +Test: [1980/2387] eta: 0:01:16 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1740 (0.3167) acc1: 98.9583 (87.6280) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1974 data: 0.1055 max mem: 3539 +Test: [1990/2387] eta: 0:01:14 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1766 (0.3160) acc1: 97.6562 (87.6802) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1842 data: 0.0921 max mem: 3539 +Test: [2000/2387] eta: 0:01:13 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1814 (0.3154) acc1: 97.9167 (87.7306) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1799 data: 0.0885 max mem: 3539 +Test: [2010/2387] eta: 0:01:11 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1814 (0.3147) acc1: 97.9167 (87.7819) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1930 data: 0.1017 max mem: 3539 +Test: [2020/2387] eta: 0:01:09 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1809 (0.3140) acc1: 97.9167 (87.8319) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1925 data: 0.0992 max mem: 3539 +Test: [2030/2387] eta: 0:01:07 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1800 (0.3134) acc1: 97.6562 (87.8820) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1973 data: 0.1037 max mem: 3540 +Test: [2040/2387] eta: 0:01:05 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1765 (0.3127) acc1: 97.6562 (87.9315) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1993 data: 0.1079 max mem: 3540 +Test: [2050/2387] eta: 0:01:03 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1753 (0.3120) acc1: 98.1771 (87.9833) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1989 data: 0.1082 max mem: 3540 +Test: [2060/2387] eta: 0:01:01 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1601 (0.3113) acc1: 98.6979 (88.0355) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1985 data: 0.1081 max mem: 3540 +Test: [2070/2387] eta: 0:00:59 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1551 (0.3105) acc1: 98.9583 (88.0896) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1935 data: 0.1029 max mem: 3540 +Test: [2080/2387] eta: 0:00:58 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1528 (0.3097) acc1: 99.2188 (88.1421) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1916 data: 0.1007 max mem: 3540 +Test: [2090/2387] eta: 0:00:56 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1508 (0.3090) acc1: 98.9583 (88.1943) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1914 data: 0.1005 max mem: 3540 +Test: [2100/2387] eta: 0:00:54 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1524 (0.3083) acc1: 98.9583 (88.2453) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1858 data: 0.0939 max mem: 3540 +Test: [2110/2387] eta: 0:00:52 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1524 (0.3075) acc1: 98.9583 (88.2969) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1849 data: 0.0921 max mem: 3540 +Test: [2120/2387] eta: 0:00:50 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1517 (0.3069) acc1: 98.9583 (88.3450) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1838 data: 0.0923 max mem: 3540 +Test: [2130/2387] eta: 0:00:48 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1485 (0.3061) acc1: 98.6979 (88.3942) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1745 data: 0.0845 max mem: 3540 +Test: [2140/2387] eta: 0:00:46 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1482 (0.3057) acc1: 98.6979 (88.4211) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1898 data: 0.0997 max mem: 3540 +Test: [2150/2387] eta: 0:00:44 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1609 (0.3050) acc1: 98.1771 (88.4661) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1789 data: 0.0872 max mem: 3540 +Test: [2160/2387] eta: 0:00:42 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1538 (0.3044) acc1: 98.9583 (88.5138) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1788 data: 0.0858 max mem: 3540 +Test: [2170/2387] eta: 0:00:41 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1722 (0.3038) acc1: 98.4375 (88.5562) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2053 data: 0.1135 max mem: 3540 +Test: [2180/2387] eta: 0:00:39 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.2107 (0.3052) acc1: 96.0938 (88.4643) acc5: 100.0000 (100.0000) acc1_neg: 28.385 acc5_neg: 100.000 time: 0.1941 data: 0.1019 max mem: 3540 +Test: [2190/2387] eta: 0:00:37 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.2663 (0.3055) acc1: 90.6250 (88.4421) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1917 data: 0.0987 max mem: 3540 +Test: [2200/2387] eta: 0:00:35 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1483 (0.3047) acc1: 98.9583 (88.4927) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1949 data: 0.1022 max mem: 3540 +Test: [2210/2387] eta: 0:00:33 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1202 (0.3039) acc1: 99.7396 (88.5381) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1747 data: 0.0815 max mem: 3540 +Test: [2220/2387] eta: 0:00:31 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1356 (0.3032) acc1: 99.4792 (88.5824) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1976 data: 0.1054 max mem: 3540 +Test: [2230/2387] eta: 0:00:29 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1242 (0.3024) acc1: 99.7396 (88.6258) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2031 data: 0.1123 max mem: 3540 +Test: [2240/2387] eta: 0:00:27 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1197 (0.3017) acc1: 99.4792 (88.6702) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1845 data: 0.0929 max mem: 3540 +Test: [2250/2387] eta: 0:00:25 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1437 (0.3012) acc1: 99.4792 (88.7105) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1950 data: 0.1021 max mem: 3540 +Test: [2260/2387] eta: 0:00:24 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1879 (0.3007) acc1: 98.9583 (88.7553) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1852 data: 0.0933 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1400 (0.2999) acc1: 100.0000 (88.8043) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1795 data: 0.0885 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1309 (0.2995) acc1: 99.7396 (88.8314) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1596 data: 0.0680 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1460 (0.2987) acc1: 99.7396 (88.8780) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1412 data: 0.0500 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1462 (0.2981) acc1: 99.7396 (88.9228) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1461 data: 0.0560 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1540 (0.2975) acc1: 99.7396 (88.9685) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1486 data: 0.0579 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1570 (0.2970) acc1: 99.4792 (89.0057) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1469 data: 0.0550 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1570 (0.2963) acc1: 98.9583 (89.0473) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1456 data: 0.0541 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1513 (0.2957) acc1: 98.9583 (89.0893) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1445 data: 0.0537 max mem: 3541 +Test: [2350/2387] eta: 0:00:06 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1520 (0.2953) acc1: 99.4792 (89.1086) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1382 data: 0.0476 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1540 (0.2948) acc1: 99.4792 (89.1515) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1386 data: 0.0484 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1603 (0.2942) acc1: 99.2188 (89.1935) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1319 data: 0.0415 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1733 (0.2938) acc1: 98.6979 (89.2289) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1230 data: 0.0323 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 25.114 acc5_pos: 100.000 loss: 0.1826 (0.2942) acc1: 98.1771 (89.2123) acc5: 100.0000 (100.0000) acc1_neg: 62.791 acc5_neg: 100.000 time: 0.1174 data: 0.0297 max mem: 3541 +Test: Total time: 0:07:25 (0.1867 s / it) +* Overall Acc@1 89.212 Acc@5 100.000 + target == real Acc@1 14.891 Acc@5 100.000 + target == fake Acc@1 94.432 Acc@5 100.000 + loss 0.294 + PR-AUC 0.9843 ROC-AUC 0.8082 EER 26.00% (thr=0.7974) +Accuracy of the network on the 916267 test images: 89.2% +[rank0]:[W726 05:02:15.843709362 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:02:38.110000 3153115 torch/distributed/run.py:766] +W0726 05:02:38.110000 3153115 torch/distributed/run.py:766] ***************************************** +W0726 05:02:38.110000 3153115 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:02:38.110000 3153115 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 05:03:12.315284137 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:03:12.686869979 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:03:12.727869959 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:03:12.147210127 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:03:12.160924315 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:03:13.319710455 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:03:13.336917004 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank1]:[W726 05:03:13.337823201 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n11_seed7_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed7_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:01:08 acc1_pos: 35.677 acc5_pos: 100.000 loss: 0.9114 (0.9114) acc1: 35.6771 (35.6771) acc5: 100.0000 (100.0000) time: 7.0961 data: 5.4964 max mem: 3531 +Test: [ 10/517] eta: 0:06:08 acc1_pos: 16.146 acc5_pos: 100.000 loss: 1.2961 (1.2912) acc1: 14.0625 (15.1515) acc5: 100.0000 (100.0000) time: 0.7269 data: 0.5000 max mem: 3531 +Test: [ 20/517] eta: 0:03:31 acc1_pos: 8.854 acc5_pos: 100.000 loss: 1.2965 (1.2808) acc1: 13.8021 (15.7490) acc5: 100.0000 (100.0000) time: 0.0911 data: 0.0003 max mem: 3531 +Test: [ 30/517] eta: 0:02:53 acc1_pos: 1.042 acc5_pos: 100.000 loss: 1.3299 (1.3242) acc1: 10.1562 (13.1972) acc5: 100.0000 (100.0000) time: 0.1517 data: 0.0608 max mem: 3531 +Test: [ 40/517] eta: 0:02:36 acc1_pos: 13.021 acc5_pos: 100.000 loss: 1.3299 (1.3230) acc1: 7.0312 (12.0554) acc5: 100.0000 (100.0000) time: 0.2270 data: 0.1334 max mem: 3531 +Test: [ 50/517] eta: 0:02:29 acc1_pos: 27.982 acc5_pos: 100.000 loss: 1.2432 (1.2557) acc1: 11.4583 (16.3246) acc5: 100.0000 (100.0000) acc1_neg: 70.573 acc5_neg: 100.000 time: 0.2642 data: 0.1651 max mem: 3531 +Test: [ 60/517] eta: 0:02:22 acc1_pos: 27.982 acc5_pos: 100.000 loss: 0.5209 (1.1204) acc1: 75.0000 (27.5700) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.2778 data: 0.1802 max mem: 3531 +Test: [ 70/517] eta: 0:02:14 acc1_pos: 27.982 acc5_pos: 100.000 loss: 0.2984 (1.0007) acc1: 93.2292 (37.3936) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2549 data: 0.1589 max mem: 3531 +Test: [ 80/517] eta: 0:02:07 acc1_pos: 27.982 acc5_pos: 100.000 loss: 0.2904 (0.9183) acc1: 96.0938 (44.4573) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.2349 data: 0.1373 max mem: 3531 +Test: [ 90/517] eta: 0:02:02 acc1_pos: 27.982 acc5_pos: 100.000 loss: 0.3126 (0.8494) acc1: 96.0938 (50.2690) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2399 data: 0.1422 max mem: 3531 +Test: [100/517] eta: 0:01:56 acc1_pos: 27.982 acc5_pos: 100.000 loss: 0.2656 (0.7913) acc1: 97.6562 (54.9892) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2299 data: 0.1321 max mem: 3531 +Test: [110/517] eta: 0:01:52 acc1_pos: 27.982 acc5_pos: 100.000 loss: 0.2348 (0.7403) acc1: 97.9167 (58.8049) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2249 data: 0.1277 max mem: 3531 +Test: [120/517] eta: 0:01:48 acc1_pos: 27.982 acc5_pos: 100.000 loss: 0.2453 (0.7017) acc1: 98.1771 (62.0308) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2399 data: 0.1441 max mem: 3531 +Test: [130/517] eta: 0:01:44 acc1_pos: 27.982 acc5_pos: 100.000 loss: 0.2782 (0.6701) acc1: 98.1771 (64.7145) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2349 data: 0.1404 max mem: 3531 +Test: [140/517] eta: 0:01:41 acc1_pos: 27.982 acc5_pos: 100.000 loss: 0.2672 (0.6408) acc1: 98.1771 (67.1044) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2349 data: 0.1424 max mem: 3531 +Test: [150/517] eta: 0:01:38 acc1_pos: 27.982 acc5_pos: 100.000 loss: 0.2573 (0.6148) acc1: 98.9583 (69.2122) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2502 data: 0.1595 max mem: 3531 +Test: [160/517] eta: 0:01:34 acc1_pos: 27.982 acc5_pos: 100.000 loss: 0.2758 (0.5969) acc1: 97.6562 (70.8091) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2449 data: 0.1545 max mem: 3531 +Test: [170/517] eta: 0:01:32 acc1_pos: 27.982 acc5_pos: 100.000 loss: 0.3191 (0.5813) acc1: 94.5312 (72.1781) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2496 data: 0.1555 max mem: 3531 +Test: [180/517] eta: 0:01:29 acc1_pos: 27.982 acc5_pos: 100.000 loss: 0.3202 (0.5686) acc1: 93.7500 (73.2577) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.2599 data: 0.1613 max mem: 3531 +Test: [190/517] eta: 0:01:25 acc1_pos: 27.982 acc5_pos: 100.000 loss: 0.2842 (0.5524) acc1: 96.3542 (74.5473) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2249 data: 0.1273 max mem: 3531 +Test: [200/517] eta: 0:01:22 acc1_pos: 27.982 acc5_pos: 100.000 loss: 0.2382 (0.5361) acc1: 98.6979 (75.7605) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2299 data: 0.1327 max mem: 3531 +Test: [210/517] eta: 0:01:19 acc1_pos: 27.982 acc5_pos: 100.000 loss: 0.3022 (0.5324) acc1: 91.6667 (76.0503) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.2449 data: 0.1459 max mem: 3532 +Test: [220/517] eta: 0:01:17 acc1_pos: 27.982 acc5_pos: 100.000 loss: 0.4813 (0.5306) acc1: 79.4271 (76.1736) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.2399 data: 0.1422 max mem: 3532 +Test: [230/517] eta: 0:01:14 acc1_pos: 27.982 acc5_pos: 100.000 loss: 0.4620 (0.5252) acc1: 80.7292 (76.7091) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2449 data: 0.1499 max mem: 3532 +Test: [240/517] eta: 0:01:11 acc1_pos: 27.982 acc5_pos: 100.000 loss: 0.2731 (0.5128) acc1: 97.6562 (77.6279) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2299 data: 0.1341 max mem: 3532 +Test: [250/517] eta: 0:01:08 acc1_pos: 27.982 acc5_pos: 100.000 loss: 0.2485 (0.5033) acc1: 98.6979 (78.4342) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2299 data: 0.1316 max mem: 3532 +Test: [260/517] eta: 0:01:05 acc1_pos: 27.982 acc5_pos: 100.000 loss: 0.2872 (0.4956) acc1: 96.8750 (79.1327) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2299 data: 0.1310 max mem: 3532 +Test: [270/517] eta: 0:01:02 acc1_pos: 27.982 acc5_pos: 100.000 loss: 0.2872 (0.4875) acc1: 95.3125 (79.7048) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2349 data: 0.1371 max mem: 3532 +Test: [280/517] eta: 0:01:00 acc1_pos: 27.982 acc5_pos: 100.000 loss: 0.2651 (0.4795) acc1: 94.7917 (80.2575) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2399 data: 0.1445 max mem: 3532 +Test: [290/517] eta: 0:00:57 acc1_pos: 27.982 acc5_pos: 100.000 loss: 0.2713 (0.4732) acc1: 95.5729 (80.8151) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2387 data: 0.1439 max mem: 3532 +Test: [300/517] eta: 0:00:54 acc1_pos: 27.982 acc5_pos: 100.000 loss: 0.2859 (0.4672) acc1: 95.5729 (81.3019) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2438 data: 0.1492 max mem: 3532 +Test: [310/517] eta: 0:00:52 acc1_pos: 27.982 acc5_pos: 100.000 loss: 0.3068 (0.4631) acc1: 95.0521 (81.7147) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2449 data: 0.1498 max mem: 3532 +Test: [320/517] eta: 0:00:49 acc1_pos: 27.982 acc5_pos: 100.000 loss: 0.3245 (0.4573) acc1: 94.5312 (82.1440) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2499 data: 0.1517 max mem: 3532 +Test: [330/517] eta: 0:00:47 acc1_pos: 27.982 acc5_pos: 100.000 loss: 0.2349 (0.4506) acc1: 96.0938 (82.5647) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2499 data: 0.1492 max mem: 3532 +Test: [340/517] eta: 0:00:44 acc1_pos: 27.982 acc5_pos: 100.000 loss: 0.2293 (0.4443) acc1: 96.8750 (83.0011) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2499 data: 0.1513 max mem: 3532 +Test: [350/517] eta: 0:00:42 acc1_pos: 27.982 acc5_pos: 100.000 loss: 0.2548 (0.4396) acc1: 96.3542 (83.3689) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.2499 data: 0.1542 max mem: 3532 +Test: [360/517] eta: 0:00:39 acc1_pos: 27.982 acc5_pos: 100.000 loss: 0.2767 (0.4358) acc1: 95.5729 (83.6875) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2499 data: 0.1527 max mem: 3532 +Test: [370/517] eta: 0:00:37 acc1_pos: 27.982 acc5_pos: 100.000 loss: 0.2396 (0.4295) acc1: 97.6562 (84.1027) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2399 data: 0.1417 max mem: 3532 +Test: [380/517] eta: 0:00:34 acc1_pos: 27.982 acc5_pos: 100.000 loss: 0.1891 (0.4234) acc1: 100.0000 (84.5076) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2399 data: 0.1422 max mem: 3532 +Test: [390/517] eta: 0:00:31 acc1_pos: 27.982 acc5_pos: 100.000 loss: 0.1993 (0.4179) acc1: 99.4792 (84.8885) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2449 data: 0.1483 max mem: 3532 +Test: [400/517] eta: 0:00:29 acc1_pos: 27.982 acc5_pos: 100.000 loss: 0.2048 (0.4127) acc1: 99.4792 (85.2595) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1592 max mem: 3532 +Test: [410/517] eta: 0:00:26 acc1_pos: 27.982 acc5_pos: 100.000 loss: 0.2098 (0.4085) acc1: 99.4792 (85.6017) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2499 data: 0.1532 max mem: 3532 +Test: [420/517] eta: 0:00:24 acc1_pos: 27.982 acc5_pos: 100.000 loss: 0.2588 (0.4057) acc1: 99.2188 (85.8787) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2449 data: 0.1464 max mem: 3532 +Test: [430/517] eta: 0:00:21 acc1_pos: 27.982 acc5_pos: 100.000 loss: 0.2942 (0.4042) acc1: 95.5729 (86.0638) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.2499 data: 0.1535 max mem: 3532 +Test: [440/517] eta: 0:00:19 acc1_pos: 27.982 acc5_pos: 100.000 loss: 0.3173 (0.4023) acc1: 95.0521 (86.2493) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.2399 data: 0.1435 max mem: 3533 +Test: [450/517] eta: 0:00:16 acc1_pos: 27.982 acc5_pos: 100.000 loss: 0.2864 (0.3993) acc1: 96.3542 (86.4889) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2449 data: 0.1489 max mem: 3533 +Test: [460/517] eta: 0:00:14 acc1_pos: 27.982 acc5_pos: 100.000 loss: 0.2741 (0.3970) acc1: 96.6146 (86.6786) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.2499 data: 0.1556 max mem: 3533 +Test: [470/517] eta: 0:00:11 acc1_pos: 27.982 acc5_pos: 100.000 loss: 0.2702 (0.3941) acc1: 96.8750 (86.9206) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2449 data: 0.1501 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 27.982 acc5_pos: 100.000 loss: 0.2528 (0.3910) acc1: 99.2188 (87.1860) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2399 data: 0.1440 max mem: 3533 +Test: [490/517] eta: 0:00:06 acc1_pos: 27.982 acc5_pos: 100.000 loss: 0.2528 (0.3905) acc1: 99.2188 (87.2322) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2449 data: 0.1481 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 27.982 acc5_pos: 100.000 loss: 0.2845 (0.3882) acc1: 97.3958 (87.4397) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2549 data: 0.1590 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 27.982 acc5_pos: 100.000 loss: 0.2842 (0.3872) acc1: 97.6562 (87.5357) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2274 data: 0.1338 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 27.982 acc5_pos: 100.000 loss: 0.2842 (0.3860) acc1: 97.6562 (87.6337) acc5: 100.0000 (100.0000) acc1_neg: 91.166 acc5_neg: 100.000 time: 0.1904 data: 0.0957 max mem: 3533 +Test: Total time: 0:02:09 (0.2501 s / it) +* Overall Acc@1 87.634 Acc@5 100.000 + target == real Acc@1 13.190 Acc@5 100.000 + target == fake Acc@1 95.356 Acc@5 100.000 + loss 0.386 + PR-AUC 0.9366 ROC-AUC 0.6408 EER 39.57% (thr=0.7656) +Accuracy of the network on the 198427 test images: 87.6% +[rank0]:[W726 05:05:37.450169591 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 05:05:59.724000 3153687 torch/distributed/run.py:766] +W0726 05:05:59.724000 3153687 torch/distributed/run.py:766] ***************************************** +W0726 05:05:59.724000 3153687 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:05:59.724000 3153687 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 6): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:06:32.523310747 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W726 05:06:32.523309561 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:06:32.865136071 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:06:32.051636359 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:06:32.112116136 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:06:32.164988719 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +[rank0]:[W726 05:06:33.230639946 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:06:33.241899123 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n11_seed7_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed7_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:36:39 acc1_pos: 91.406 acc5_pos: 100.000 loss: 0.4037 (0.4037) acc1: 91.4062 (91.4062) acc5: 100.0000 (100.0000) time: 5.6554 data: 4.0043 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/389] eta: 0:03:45 acc1_pos: 68.750 acc5_pos: 100.000 loss: 0.4568 (0.5182) acc1: 80.2083 (76.4678) acc5: 100.0000 (100.0000) time: 0.5958 data: 0.3643 max mem: 3531 +Test: [ 20/389] eta: 0:02:10 acc1_pos: 71.094 acc5_pos: 100.000 loss: 0.5398 (0.5401) acc1: 71.0938 (75.2108) acc5: 100.0000 (100.0000) time: 0.0898 data: 0.0003 max mem: 3531 +Test: [ 30/389] eta: 0:01:37 acc1_pos: 18.490 acc5_pos: 100.000 loss: 0.5694 (0.5900) acc1: 68.7500 (67.8091) acc5: 100.0000 (100.0000) time: 0.0924 data: 0.0003 max mem: 3531 +Test: [ 40/389] eta: 0:01:23 acc1_pos: 61.719 acc5_pos: 100.000 loss: 0.6867 (0.6013) acc1: 58.8542 (67.1430) acc5: 100.0000 (100.0000) time: 0.1176 data: 0.0249 max mem: 3531 +Test: [ 50/389] eta: 0:01:15 acc1_pos: 99.219 acc5_pos: 100.000 loss: 0.4829 (0.5588) acc1: 84.1146 (71.6248) acc5: 100.0000 (100.0000) time: 0.1449 data: 0.0532 max mem: 3531 +Test: [ 60/389] eta: 0:01:09 acc1_pos: 81.510 acc5_pos: 100.000 loss: 0.3339 (0.5225) acc1: 95.0521 (75.4781) acc5: 100.0000 (100.0000) time: 0.1548 data: 0.0628 max mem: 3531 +Test: [ 70/389] eta: 0:01:05 acc1_pos: 71.354 acc5_pos: 100.000 loss: 0.3767 (0.5190) acc1: 93.4896 (75.9206) acc5: 100.0000 (100.0000) time: 0.1603 data: 0.0693 max mem: 3531 +Test: [ 80/389] eta: 0:01:00 acc1_pos: 97.917 acc5_pos: 100.000 loss: 0.5072 (0.5269) acc1: 82.0312 (75.5208) acc5: 100.0000 (100.0000) time: 0.1499 data: 0.0584 max mem: 3531 +Test: [ 90/389] eta: 0:00:57 acc1_pos: 86.198 acc5_pos: 100.000 loss: 0.4186 (0.5129) acc1: 85.1562 (76.8086) acc5: 100.0000 (100.0000) time: 0.1495 data: 0.0537 max mem: 3531 +Test: [100/389] eta: 0:00:54 acc1_pos: 38.281 acc5_pos: 100.000 loss: 0.4406 (0.5239) acc1: 82.0312 (75.7838) acc5: 100.0000 (100.0000) time: 0.1499 data: 0.0545 max mem: 3531 +Test: [110/389] eta: 0:00:51 acc1_pos: 99.219 acc5_pos: 100.000 loss: 0.5302 (0.5232) acc1: 71.3542 (75.3191) acc5: 100.0000 (100.0000) time: 0.1502 data: 0.0583 max mem: 3531 +Test: [120/389] eta: 0:00:48 acc1_pos: 92.448 acc5_pos: 100.000 loss: 0.4810 (0.5175) acc1: 76.0417 (75.8092) acc5: 100.0000 (100.0000) time: 0.1549 data: 0.0610 max mem: 3531 +Test: [130/389] eta: 0:00:46 acc1_pos: 90.184 acc5_pos: 100.000 loss: 0.5604 (0.5371) acc1: 68.4896 (73.7218) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.1546 data: 0.0611 max mem: 3531 +Test: [140/389] eta: 0:00:44 acc1_pos: 90.184 acc5_pos: 100.000 loss: 0.7764 (0.5695) acc1: 41.4062 (70.6468) acc5: 100.0000 (100.0000) acc1_neg: 0.000 acc5_neg: 100.000 time: 0.1599 data: 0.0660 max mem: 3531 +Test: [150/389] eta: 0:00:42 acc1_pos: 90.184 acc5_pos: 100.000 loss: 0.8006 (0.5903) acc1: 41.4062 (68.8724) acc5: 100.0000 (100.0000) acc1_neg: 60.417 acc5_neg: 100.000 time: 0.1599 data: 0.0625 max mem: 3531 +Test: [160/389] eta: 0:00:40 acc1_pos: 90.184 acc5_pos: 100.000 loss: 0.6709 (0.5933) acc1: 55.7292 (68.3311) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.1699 data: 0.0719 max mem: 3531 +Test: [170/389] eta: 0:00:38 acc1_pos: 90.184 acc5_pos: 100.000 loss: 0.5306 (0.5950) acc1: 60.1562 (68.4896) acc5: 100.0000 (100.0000) acc1_neg: 27.865 acc5_neg: 100.000 time: 0.1750 data: 0.0792 max mem: 3531 +Test: [180/389] eta: 0:00:36 acc1_pos: 90.184 acc5_pos: 100.000 loss: 0.6114 (0.6110) acc1: 55.9896 (67.0897) acc5: 100.0000 (100.0000) acc1_neg: 47.656 acc5_neg: 100.000 time: 0.1649 data: 0.0711 max mem: 3531 +Test: [190/389] eta: 0:00:34 acc1_pos: 90.184 acc5_pos: 100.000 loss: 0.9272 (0.6237) acc1: 34.1146 (65.7354) acc5: 100.0000 (100.0000) acc1_neg: 0.521 acc5_neg: 100.000 time: 0.1649 data: 0.0710 max mem: 3531 +Test: [200/389] eta: 0:00:32 acc1_pos: 90.184 acc5_pos: 100.000 loss: 0.9692 (0.6573) acc1: 17.1875 (63.2165) acc5: 100.0000 (100.0000) acc1_neg: 1.823 acc5_neg: 100.000 time: 0.1599 data: 0.0659 max mem: 3531 +Test: [210/389] eta: 0:00:31 acc1_pos: 90.184 acc5_pos: 100.000 loss: 1.1886 (0.6708) acc1: 4.9479 (61.9952) acc5: 100.0000 (100.0000) acc1_neg: 38.021 acc5_neg: 100.000 time: 0.1520 data: 0.0577 max mem: 3532 +Test: [220/389] eta: 0:00:29 acc1_pos: 90.184 acc5_pos: 100.000 loss: 0.8188 (0.6797) acc1: 36.4583 (60.6252) acc5: 100.0000 (100.0000) acc1_neg: 21.354 acc5_neg: 100.000 time: 0.1449 data: 0.0515 max mem: 3532 +Test: [230/389] eta: 0:00:27 acc1_pos: 90.184 acc5_pos: 100.000 loss: 0.8196 (0.6817) acc1: 35.4167 (60.1157) acc5: 100.0000 (100.0000) acc1_neg: 48.438 acc5_neg: 100.000 time: 0.1431 data: 0.0496 max mem: 3532 +Test: [240/389] eta: 0:00:25 acc1_pos: 90.184 acc5_pos: 100.000 loss: 0.7182 (0.6835) acc1: 50.0000 (59.8256) acc5: 100.0000 (100.0000) acc1_neg: 30.990 acc5_neg: 100.000 time: 0.1549 data: 0.0599 max mem: 3532 +Test: [250/389] eta: 0:00:23 acc1_pos: 90.184 acc5_pos: 100.000 loss: 0.7182 (0.6871) acc1: 46.6146 (59.0689) acc5: 100.0000 (100.0000) acc1_neg: 50.260 acc5_neg: 100.000 time: 0.1596 data: 0.0608 max mem: 3532 +Test: [260/389] eta: 0:00:21 acc1_pos: 90.184 acc5_pos: 100.000 loss: 0.7375 (0.6934) acc1: 34.6354 (58.0809) acc5: 100.0000 (100.0000) acc1_neg: 69.010 acc5_neg: 100.000 time: 0.1549 data: 0.0557 max mem: 3532 +Test: [270/389] eta: 0:00:20 acc1_pos: 90.184 acc5_pos: 100.000 loss: 0.7374 (0.6951) acc1: 41.4062 (57.9509) acc5: 100.0000 (100.0000) acc1_neg: 49.219 acc5_neg: 100.000 time: 0.1549 data: 0.0576 max mem: 3532 +Test: [280/389] eta: 0:00:18 acc1_pos: 90.184 acc5_pos: 100.000 loss: 0.8061 (0.7031) acc1: 41.6667 (57.2342) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1558 data: 0.0596 max mem: 3532 +Test: [290/389] eta: 0:00:16 acc1_pos: 90.184 acc5_pos: 100.000 loss: 0.8144 (0.7032) acc1: 43.4896 (57.3346) acc5: 100.0000 (100.0000) acc1_neg: 79.427 acc5_neg: 100.000 time: 0.1553 data: 0.0621 max mem: 3532 +Test: [300/389] eta: 0:00:14 acc1_pos: 90.184 acc5_pos: 100.000 loss: 0.9221 (0.7187) acc1: 25.0000 (55.8806) acc5: 100.0000 (100.0000) acc1_neg: 4.948 acc5_neg: 100.000 time: 0.1639 data: 0.0694 max mem: 3532 +Test: [310/389] eta: 0:00:13 acc1_pos: 90.184 acc5_pos: 100.000 loss: 1.2046 (0.7333) acc1: 8.0729 (54.8181) acc5: 100.0000 (100.0000) acc1_neg: 4.427 acc5_neg: 100.000 time: 0.1595 data: 0.0622 max mem: 3532 +Test: [320/389] eta: 0:00:11 acc1_pos: 90.184 acc5_pos: 100.000 loss: 1.0036 (0.7388) acc1: 14.5833 (54.2308) acc5: 100.0000 (100.0000) acc1_neg: 56.250 acc5_neg: 100.000 time: 0.1499 data: 0.0541 max mem: 3532 +Test: [330/389] eta: 0:00:09 acc1_pos: 90.184 acc5_pos: 100.000 loss: 0.9447 (0.7458) acc1: 22.1354 (53.3846) acc5: 100.0000 (100.0000) acc1_neg: 44.531 acc5_neg: 100.000 time: 0.1599 data: 0.0653 max mem: 3532 +Test: [340/389] eta: 0:00:08 acc1_pos: 90.184 acc5_pos: 100.000 loss: 0.9039 (0.7452) acc1: 31.7708 (53.4221) acc5: 100.0000 (100.0000) acc1_neg: 54.948 acc5_neg: 100.000 time: 0.1803 data: 0.0858 max mem: 3532 +Test: [350/389] eta: 0:00:06 acc1_pos: 90.184 acc5_pos: 100.000 loss: 0.7726 (0.7544) acc1: 45.0521 (52.9781) acc5: 100.0000 (100.0000) acc1_neg: 6.510 acc5_neg: 100.000 time: 0.1899 data: 0.0940 max mem: 3532 +Test: [360/389] eta: 0:00:04 acc1_pos: 90.184 acc5_pos: 100.000 loss: 1.0816 (0.7653) acc1: 22.3958 (52.0603) acc5: 100.0000 (100.0000) acc1_neg: 20.052 acc5_neg: 100.000 time: 0.1495 data: 0.0533 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 90.184 acc5_pos: 100.000 loss: 1.0455 (0.7629) acc1: 24.2188 (52.2637) acc5: 100.0000 (100.0000) acc1_neg: 44.531 acc5_neg: 100.000 time: 0.1349 data: 0.0393 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 90.184 acc5_pos: 100.000 loss: 0.8333 (0.7661) acc1: 38.8021 (51.7894) acc5: 100.0000 (100.0000) acc1_neg: 23.698 acc5_neg: 100.000 time: 0.1668 data: 0.0734 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 90.184 acc5_pos: 100.000 loss: 0.8680 (0.7671) acc1: 36.9792 (51.6484) acc5: 100.0000 (100.0000) acc1_neg: 62.842 acc5_neg: 100.000 time: 0.1616 data: 0.0655 max mem: 3532 +Test: Total time: 0:01:05 (0.1674 s / it) +* Overall Acc@1 51.648 Acc@5 100.000 + target == real Acc@1 75.734 Acc@5 100.000 + target == fake Acc@1 40.453 Acc@5 100.000 + loss 0.767 + PR-AUC 0.7740 ROC-AUC 0.6255 EER 41.28% (thr=0.3948) +Accuracy of the network on the 149358 test images: 51.6% +[rank0]:[W726 05:07:55.102692421 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:08:15.420000 3154250 torch/distributed/run.py:766] +W0726 05:08:15.420000 3154250 torch/distributed/run.py:766] ***************************************** +W0726 05:08:15.420000 3154250 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:08:15.420000 3154250 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:08:42.897269815 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:08:42.048493737 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:08:43.240979969 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:08:43.388371034 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:08:43.492602797 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:08:43.590957520 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +[rank0]:[W726 05:08:43.616154888 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:08:43.620743366 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n11_seed7_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed7_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:23:40 acc1_pos: 76.562 acc5_pos: 100.000 loss: 0.4875 (0.4875) acc1: 76.5625 (76.5625) acc5: 100.0000 (100.0000) time: 4.2648 data: 2.3765 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/333] eta: 0:02:43 acc1_pos: 75.000 acc5_pos: 100.000 loss: 0.4095 (0.4043) acc1: 77.3438 (78.8116) acc5: 100.0000 (100.0000) time: 0.5051 data: 0.2522 max mem: 3531 +Test: [ 20/333] eta: 0:01:36 acc1_pos: 90.365 acc5_pos: 100.000 loss: 0.3946 (0.3851) acc1: 78.9062 (80.1587) acc5: 100.0000 (100.0000) time: 0.1095 data: 0.0201 max mem: 3531 +Test: [ 30/333] eta: 0:01:12 acc1_pos: 51.562 acc5_pos: 100.000 loss: 0.4524 (0.5115) acc1: 73.1771 (70.6233) acc5: 100.0000 (100.0000) time: 0.0933 data: 0.0004 max mem: 3531 +Test: [ 40/333] eta: 0:01:02 acc1_pos: 48.438 acc5_pos: 100.000 loss: 0.8157 (0.5893) acc1: 48.4375 (65.1804) acc5: 100.0000 (100.0000) time: 0.1133 data: 0.0178 max mem: 3531 +Test: [ 50/333] eta: 0:00:57 acc1_pos: 48.958 acc5_pos: 100.000 loss: 0.8157 (0.6294) acc1: 48.4375 (62.2447) acc5: 100.0000 (100.0000) time: 0.1468 data: 0.0525 max mem: 3531 +Test: [ 60/333] eta: 0:00:52 acc1_pos: 45.052 acc5_pos: 100.000 loss: 0.7892 (0.6546) acc1: 48.9583 (60.3441) acc5: 100.0000 (100.0000) time: 0.1549 data: 0.0629 max mem: 3531 +Test: [ 70/333] eta: 0:00:49 acc1_pos: 50.781 acc5_pos: 100.000 loss: 0.7833 (0.6798) acc1: 49.2188 (58.3810) acc5: 100.0000 (100.0000) time: 0.1440 data: 0.0532 max mem: 3531 +Test: [ 80/333] eta: 0:00:45 acc1_pos: 51.042 acc5_pos: 100.000 loss: 0.7833 (0.6943) acc1: 48.9583 (57.0891) acc5: 100.0000 (100.0000) time: 0.1403 data: 0.0482 max mem: 3531 +Test: [ 90/333] eta: 0:00:43 acc1_pos: 61.198 acc5_pos: 100.000 loss: 0.7682 (0.6983) acc1: 51.0417 (56.9683) acc5: 100.0000 (100.0000) time: 0.1504 data: 0.0571 max mem: 3531 +Test: [100/333] eta: 0:00:41 acc1_pos: 64.583 acc5_pos: 100.000 loss: 0.7121 (0.7018) acc1: 54.1667 (56.5955) acc5: 100.0000 (100.0000) time: 0.1657 data: 0.0737 max mem: 3531 +Test: [110/333] eta: 0:00:40 acc1_pos: 57.292 acc5_pos: 100.000 loss: 0.7344 (0.7059) acc1: 54.1667 (56.3743) acc5: 100.0000 (100.0000) time: 0.1988 data: 0.1085 max mem: 3531 +Test: [120/333] eta: 0:00:39 acc1_pos: 63.542 acc5_pos: 100.000 loss: 0.7192 (0.7065) acc1: 55.4688 (56.3210) acc5: 100.0000 (100.0000) time: 0.2136 data: 0.1199 max mem: 3531 +Test: [130/333] eta: 0:00:37 acc1_pos: 72.396 acc5_pos: 100.000 loss: 0.6995 (0.7082) acc1: 54.9479 (56.3454) acc5: 100.0000 (100.0000) time: 0.1961 data: 0.0992 max mem: 3531 +Test: [140/333] eta: 0:00:35 acc1_pos: 59.896 acc5_pos: 100.000 loss: 0.6995 (0.7078) acc1: 54.9479 (56.2574) acc5: 100.0000 (100.0000) time: 0.1965 data: 0.1014 max mem: 3531 +Test: [150/333] eta: 0:00:33 acc1_pos: 52.344 acc5_pos: 100.000 loss: 0.7146 (0.7096) acc1: 53.6458 (56.0655) acc5: 100.0000 (100.0000) time: 0.1737 data: 0.0803 max mem: 3531 +Test: [160/333] eta: 0:00:31 acc1_pos: 39.803 acc5_pos: 100.000 loss: 0.7422 (0.7159) acc1: 50.5208 (55.7033) acc5: 100.0000 (100.0000) acc1_neg: 54.427 acc5_neg: 100.000 time: 0.1445 data: 0.0498 max mem: 3531 +Test: [170/333] eta: 0:00:30 acc1_pos: 39.803 acc5_pos: 100.000 loss: 0.8481 (0.7254) acc1: 49.7396 (55.3484) acc5: 100.0000 (100.0000) acc1_neg: 48.958 acc5_neg: 100.000 time: 0.1948 data: 0.1014 max mem: 3531 +Test: [180/333] eta: 0:00:28 acc1_pos: 39.803 acc5_pos: 100.000 loss: 0.8506 (0.7314) acc1: 50.2604 (55.1350) acc5: 100.0000 (100.0000) acc1_neg: 57.812 acc5_neg: 100.000 time: 0.2200 data: 0.1272 max mem: 3531 +Test: [190/333] eta: 0:00:25 acc1_pos: 39.803 acc5_pos: 100.000 loss: 0.7891 (0.7313) acc1: 53.1250 (55.1756) acc5: 100.0000 (100.0000) acc1_neg: 50.521 acc5_neg: 100.000 time: 0.1448 data: 0.0512 max mem: 3531 +Test: [200/333] eta: 0:00:24 acc1_pos: 39.803 acc5_pos: 100.000 loss: 0.7102 (0.7301) acc1: 59.8958 (55.5426) acc5: 100.0000 (100.0000) acc1_neg: 64.583 acc5_neg: 100.000 time: 0.1487 data: 0.0572 max mem: 3531 +Test: [210/333] eta: 0:00:22 acc1_pos: 39.803 acc5_pos: 100.000 loss: 0.7383 (0.7328) acc1: 59.8958 (55.6477) acc5: 100.0000 (100.0000) acc1_neg: 55.990 acc5_neg: 100.000 time: 0.2106 data: 0.1196 max mem: 3532 +Test: [220/333] eta: 0:00:20 acc1_pos: 39.803 acc5_pos: 100.000 loss: 0.7659 (0.7417) acc1: 55.9896 (55.2767) acc5: 100.0000 (100.0000) acc1_neg: 64.583 acc5_neg: 100.000 time: 0.2249 data: 0.1328 max mem: 3532 +Test: [230/333] eta: 0:00:19 acc1_pos: 39.803 acc5_pos: 100.000 loss: 0.5864 (0.7329) acc1: 65.3646 (56.1057) acc5: 100.0000 (100.0000) acc1_neg: 78.646 acc5_neg: 100.000 time: 0.2047 data: 0.1123 max mem: 3532 +Test: [240/333] eta: 0:00:17 acc1_pos: 39.803 acc5_pos: 100.000 loss: 0.5642 (0.7282) acc1: 72.1354 (56.5709) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.1798 data: 0.0887 max mem: 3532 +Test: [250/333] eta: 0:00:15 acc1_pos: 39.803 acc5_pos: 100.000 loss: 0.6510 (0.7284) acc1: 63.8021 (56.7252) acc5: 100.0000 (100.0000) acc1_neg: 49.479 acc5_neg: 100.000 time: 0.1836 data: 0.0926 max mem: 3532 +Test: [260/333] eta: 0:00:13 acc1_pos: 39.803 acc5_pos: 100.000 loss: 0.7886 (0.7333) acc1: 56.2500 (56.5413) acc5: 100.0000 (100.0000) acc1_neg: 57.812 acc5_neg: 100.000 time: 0.1846 data: 0.0934 max mem: 3532 +Test: [270/333] eta: 0:00:11 acc1_pos: 39.803 acc5_pos: 100.000 loss: 0.7709 (0.7331) acc1: 56.7708 (56.7074) acc5: 100.0000 (100.0000) acc1_neg: 56.771 acc5_neg: 100.000 time: 0.1848 data: 0.0939 max mem: 3532 +Test: [280/333] eta: 0:00:09 acc1_pos: 39.803 acc5_pos: 100.000 loss: 0.7419 (0.7347) acc1: 59.6354 (56.6615) acc5: 100.0000 (100.0000) acc1_neg: 61.979 acc5_neg: 100.000 time: 0.1858 data: 0.0950 max mem: 3532 +Test: [290/333] eta: 0:00:07 acc1_pos: 39.803 acc5_pos: 100.000 loss: 0.8448 (0.7391) acc1: 51.0417 (56.4209) acc5: 100.0000 (100.0000) acc1_neg: 46.615 acc5_neg: 100.000 time: 0.1769 data: 0.0856 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 39.803 acc5_pos: 100.000 loss: 0.8886 (0.7467) acc1: 46.6146 (56.0285) acc5: 100.0000 (100.0000) acc1_neg: 48.438 acc5_neg: 100.000 time: 0.1736 data: 0.0802 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 39.803 acc5_pos: 100.000 loss: 0.8611 (0.7443) acc1: 48.6979 (56.3028) acc5: 100.0000 (100.0000) acc1_neg: 66.927 acc5_neg: 100.000 time: 0.1778 data: 0.0831 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 39.803 acc5_pos: 100.000 loss: 0.6269 (0.7422) acc1: 66.9271 (56.5567) acc5: 100.0000 (100.0000) acc1_neg: 39.844 acc5_neg: 100.000 time: 0.1899 data: 0.0952 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 39.803 acc5_pos: 100.000 loss: 1.0034 (0.7525) acc1: 43.4896 (56.0698) acc5: 100.0000 (100.0000) acc1_neg: 40.104 acc5_neg: 100.000 time: 0.1826 data: 0.0902 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 39.803 acc5_pos: 100.000 loss: 1.0597 (0.7545) acc1: 41.6667 (55.9655) acc5: 100.0000 (100.0000) acc1_neg: 41.127 acc5_neg: 100.000 time: 0.1867 data: 0.0904 max mem: 3532 +Test: Total time: 0:01:01 (0.1851 s / it) +* Overall Acc@1 55.966 Acc@5 100.000 + target == real Acc@1 55.903 Acc@5 100.000 + target == fake Acc@1 56.021 Acc@5 100.000 + loss 0.755 + PR-AUC 0.5999 ROC-AUC 0.5912 EER 44.05% (thr=0.5005) +Accuracy of the network on the 127843 test images: 56.0% +[rank0]:[W726 05:10:00.884532882 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:10:22.583000 3154808 torch/distributed/run.py:766] +W0726 05:10:22.583000 3154808 torch/distributed/run.py:766] ***************************************** +W0726 05:10:22.583000 3154808 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:10:22.583000 3154808 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:10:56.080272814 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:10:57.358116199 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:10:57.588189663 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:10:57.899531907 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:10:57.017374309 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 05:10:57.116161053 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank1]:[W726 05:10:57.116196361 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W726 05:10:57.120755351 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n11_seed7_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed7_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:23:15 acc1_pos: 96.615 acc5_pos: 100.000 loss: 0.1458 (0.1458) acc1: 96.6146 (96.6146) acc5: 100.0000 (100.0000) time: 5.9395 data: 3.8013 max mem: 3531 +Test: [ 10/235] eta: 0:02:21 acc1_pos: 97.656 acc5_pos: 100.000 loss: 0.1532 (0.1536) acc1: 96.6146 (96.6619) acc5: 100.0000 (100.0000) time: 0.6292 data: 0.3460 max mem: 3531 +Test: [ 20/235] eta: 0:01:23 acc1_pos: 95.573 acc5_pos: 100.000 loss: 0.1498 (0.1504) acc1: 96.6146 (96.7510) acc5: 100.0000 (100.0000) time: 0.1088 data: 0.0147 max mem: 3531 +Test: [ 30/235] eta: 0:01:15 acc1_pos: 97.396 acc5_pos: 100.000 loss: 0.1454 (0.1503) acc1: 96.8750 (96.8330) acc5: 100.0000 (100.0000) time: 0.2250 data: 0.1329 max mem: 3531 +Test: [ 40/235] eta: 0:01:10 acc1_pos: 97.135 acc5_pos: 100.000 loss: 0.1478 (0.1500) acc1: 97.1354 (96.9131) acc5: 100.0000 (100.0000) time: 0.3333 data: 0.2389 max mem: 3531 +Test: [ 50/235] eta: 0:01:01 acc1_pos: 96.094 acc5_pos: 100.000 loss: 0.1521 (0.1510) acc1: 97.1354 (96.8597) acc5: 100.0000 (100.0000) time: 0.2729 data: 0.1783 max mem: 3531 +Test: [ 60/235] eta: 0:00:54 acc1_pos: 98.177 acc5_pos: 100.000 loss: 0.1535 (0.1520) acc1: 96.6146 (96.8408) acc5: 100.0000 (100.0000) time: 0.2160 data: 0.1235 max mem: 3531 +Test: [ 70/235] eta: 0:00:49 acc1_pos: 96.923 acc5_pos: 100.000 loss: 0.1678 (0.3839) acc1: 95.5729 (85.7835) acc5: 100.0000 (100.0000) acc1_neg: 5.729 acc5_neg: 100.000 time: 0.2199 data: 0.1257 max mem: 3531 +Test: [ 80/235] eta: 0:00:45 acc1_pos: 96.923 acc5_pos: 100.000 loss: 2.0199 (0.5863) acc1: 7.2917 (76.0385) acc5: 100.0000 (100.0000) acc1_neg: 5.729 acc5_neg: 100.000 time: 0.2288 data: 0.1329 max mem: 3531 +Test: [ 90/235] eta: 0:00:41 acc1_pos: 96.923 acc5_pos: 100.000 loss: 2.0210 (0.7435) acc1: 6.7708 (68.5125) acc5: 100.0000 (100.0000) acc1_neg: 6.771 acc5_neg: 100.000 time: 0.2249 data: 0.1313 max mem: 3531 +Test: [100/235] eta: 0:00:37 acc1_pos: 96.923 acc5_pos: 100.000 loss: 2.0095 (0.8691) acc1: 7.2917 (62.4948) acc5: 100.0000 (100.0000) acc1_neg: 7.552 acc5_neg: 100.000 time: 0.2066 data: 0.1137 max mem: 3531 +Test: [110/235] eta: 0:00:33 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.9778 (0.9686) acc1: 7.5521 (57.6248) acc5: 100.0000 (100.0000) acc1_neg: 7.292 acc5_neg: 100.000 time: 0.2152 data: 0.1234 max mem: 3531 +Test: [120/235] eta: 0:00:30 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.9317 (1.0365) acc1: 9.6354 (53.7857) acc5: 100.0000 (100.0000) acc1_neg: 11.719 acc5_neg: 100.000 time: 0.2106 data: 0.1183 max mem: 3531 +Test: [130/235] eta: 0:00:27 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.7451 (1.0907) acc1: 11.7188 (50.6560) acc5: 100.0000 (100.0000) acc1_neg: 13.021 acc5_neg: 100.000 time: 0.2011 data: 0.1076 max mem: 3531 +Test: [140/235] eta: 0:00:24 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.7557 (1.1400) acc1: 11.9792 (47.9093) acc5: 100.0000 (100.0000) acc1_neg: 10.417 acc5_neg: 100.000 time: 0.2036 data: 0.1104 max mem: 3531 +Test: [150/235] eta: 0:00:21 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.7589 (1.1810) acc1: 10.6771 (45.4988) acc5: 100.0000 (100.0000) acc1_neg: 12.240 acc5_neg: 100.000 time: 0.2034 data: 0.1069 max mem: 3531 +Test: [160/235] eta: 0:00:18 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.7482 (1.2158) acc1: 11.1979 (43.3942) acc5: 100.0000 (100.0000) acc1_neg: 12.500 acc5_neg: 100.000 time: 0.2103 data: 0.1120 max mem: 3531 +Test: [170/235] eta: 0:00:16 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.7682 (1.2578) acc1: 10.4167 (41.4794) acc5: 100.0000 (100.0000) acc1_neg: 9.896 acc5_neg: 100.000 time: 0.2081 data: 0.1132 max mem: 3531 +Test: [180/235] eta: 0:00:13 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.9455 (1.2957) acc1: 10.4167 (39.8322) acc5: 100.0000 (100.0000) acc1_neg: 10.156 acc5_neg: 100.000 time: 0.2340 data: 0.1409 max mem: 3531 +Test: [190/235] eta: 0:00:11 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.9390 (1.3296) acc1: 10.6771 (38.2908) acc5: 100.0000 (100.0000) acc1_neg: 11.458 acc5_neg: 100.000 time: 0.2327 data: 0.1405 max mem: 3531 +Test: [200/235] eta: 0:00:08 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.9458 (1.3607) acc1: 10.1562 (36.9001) acc5: 100.0000 (100.0000) acc1_neg: 11.198 acc5_neg: 100.000 time: 0.1610 data: 0.0671 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.9494 (1.3886) acc1: 10.1562 (35.6487) acc5: 100.0000 (100.0000) acc1_neg: 13.021 acc5_neg: 100.000 time: 0.1233 data: 0.0292 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.9436 (1.4137) acc1: 10.4167 (34.5093) acc5: 100.0000 (100.0000) acc1_neg: 8.594 acc5_neg: 100.000 time: 0.1494 data: 0.0583 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.9436 (1.4367) acc1: 10.4167 (33.4686) acc5: 100.0000 (100.0000) acc1_neg: 8.594 acc5_neg: 100.000 time: 0.1462 data: 0.0566 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.9442 (1.4460) acc1: 9.8958 (33.1702) acc5: 100.0000 (100.0000) acc1_neg: 4.348 acc5_neg: 100.000 time: 0.1322 data: 0.0462 max mem: 3532 +Test: Total time: 0:00:52 (0.2249 s / it) +* Overall Acc@1 33.170 Acc@5 100.000 + target == real Acc@1 96.858 Acc@5 100.000 + target == fake Acc@1 10.135 Acc@5 100.000 + loss 1.446 + PR-AUC 0.8377 ROC-AUC 0.6713 EER 37.47% (thr=0.1002) +Accuracy of the network on the 89879 test images: 33.2% +[rank0]:[W726 05:11:58.308969035 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:12:19.742000 3155366 torch/distributed/run.py:766] +W0726 05:12:19.742000 3155366 torch/distributed/run.py:766] ***************************************** +W0726 05:12:19.742000 3155366 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:12:19.742000 3155366 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:12:54.373967081 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:12:54.548628732 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:12:54.664370721 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:12:54.732454284 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 3): env:// +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:12:54.039043335 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:12:54.041920059 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:12:54.062938770 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W726 05:12:54.065652496 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n11_seed7_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed7_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:24:09 acc1_pos: 88.021 acc5_pos: 100.000 loss: 0.3281 (0.3281) acc1: 88.0208 (88.0208) acc5: 100.0000 (100.0000) time: 5.3893 data: 3.7975 max mem: 3531 +Test: [ 10/269] eta: 0:02:28 acc1_pos: 94.531 acc5_pos: 100.000 loss: 0.3363 (0.3337) acc1: 90.8854 (89.6307) acc5: 100.0000 (100.0000) time: 0.5715 data: 0.3456 max mem: 3531 +Test: [ 20/269] eta: 0:01:25 acc1_pos: 95.052 acc5_pos: 100.000 loss: 0.3252 (0.3463) acc1: 91.1458 (89.0129) acc5: 100.0000 (100.0000) time: 0.0924 data: 0.0025 max mem: 3531 +Test: [ 30/269] eta: 0:01:11 acc1_pos: 87.500 acc5_pos: 100.000 loss: 0.3252 (0.3477) acc1: 91.1458 (89.4237) acc5: 100.0000 (100.0000) time: 0.1525 data: 0.0594 max mem: 3531 +Test: [ 40/269] eta: 0:01:03 acc1_pos: 54.310 acc5_pos: 100.000 loss: 0.4379 (0.5022) acc1: 80.4688 (77.2358) acc5: 100.0000 (100.0000) acc1_neg: 17.969 acc5_neg: 100.000 time: 0.2077 data: 0.1128 max mem: 3531 +Test: [ 50/269] eta: 0:00:58 acc1_pos: 54.310 acc5_pos: 100.000 loss: 1.0891 (0.6267) acc1: 28.3854 (66.9373) acc5: 100.0000 (100.0000) acc1_neg: 11.719 acc5_neg: 100.000 time: 0.2099 data: 0.1170 max mem: 3531 +Test: [ 60/269] eta: 0:00:51 acc1_pos: 54.310 acc5_pos: 100.000 loss: 1.1348 (0.7170) acc1: 23.6979 (59.3878) acc5: 100.0000 (100.0000) acc1_neg: 17.969 acc5_neg: 100.000 time: 0.1822 data: 0.0859 max mem: 3531 +Test: [ 70/269] eta: 0:00:46 acc1_pos: 54.310 acc5_pos: 100.000 loss: 1.1598 (0.7755) acc1: 21.6146 (54.4968) acc5: 100.0000 (100.0000) acc1_neg: 23.698 acc5_neg: 100.000 time: 0.1569 data: 0.0610 max mem: 3531 +Test: [ 80/269] eta: 0:00:43 acc1_pos: 54.310 acc5_pos: 100.000 loss: 1.1555 (0.8156) acc1: 23.1771 (51.1638) acc5: 100.0000 (100.0000) acc1_neg: 20.833 acc5_neg: 100.000 time: 0.1849 data: 0.0946 max mem: 3531 +Test: [ 90/269] eta: 0:00:40 acc1_pos: 54.310 acc5_pos: 100.000 loss: 1.2828 (0.8703) acc1: 20.0521 (47.5733) acc5: 100.0000 (100.0000) acc1_neg: 17.708 acc5_neg: 100.000 time: 0.2029 data: 0.1069 max mem: 3531 +Test: [100/269] eta: 0:00:37 acc1_pos: 54.310 acc5_pos: 100.000 loss: 1.2400 (0.9034) acc1: 20.0521 (45.1887) acc5: 100.0000 (100.0000) acc1_neg: 16.146 acc5_neg: 100.000 time: 0.1751 data: 0.0756 max mem: 3531 +Test: [110/269] eta: 0:00:33 acc1_pos: 54.310 acc5_pos: 100.000 loss: 1.1968 (0.9228) acc1: 22.9167 (43.5600) acc5: 100.0000 (100.0000) acc1_neg: 27.865 acc5_neg: 100.000 time: 0.1365 data: 0.0401 max mem: 3531 +Test: [120/269] eta: 0:00:30 acc1_pos: 54.310 acc5_pos: 100.000 loss: 1.0750 (0.9340) acc1: 27.8646 (42.7858) acc5: 100.0000 (100.0000) acc1_neg: 35.156 acc5_neg: 100.000 time: 0.1451 data: 0.0507 max mem: 3531 +Test: [130/269] eta: 0:00:28 acc1_pos: 54.310 acc5_pos: 100.000 loss: 1.0725 (0.9532) acc1: 33.5938 (41.4301) acc5: 100.0000 (100.0000) acc1_neg: 21.354 acc5_neg: 100.000 time: 0.1808 data: 0.0878 max mem: 3531 +Test: [140/269] eta: 0:00:27 acc1_pos: 54.310 acc5_pos: 100.000 loss: 1.1791 (0.9728) acc1: 21.3542 (39.9509) acc5: 100.0000 (100.0000) acc1_neg: 13.281 acc5_neg: 100.000 time: 0.2246 data: 0.1326 max mem: 3531 +Test: [150/269] eta: 0:00:24 acc1_pos: 54.310 acc5_pos: 100.000 loss: 1.1791 (0.9789) acc1: 21.0938 (39.2902) acc5: 100.0000 (100.0000) acc1_neg: 36.719 acc5_neg: 100.000 time: 0.2253 data: 0.1315 max mem: 3531 +Test: [160/269] eta: 0:00:22 acc1_pos: 54.310 acc5_pos: 100.000 loss: 1.0792 (0.9822) acc1: 24.7396 (38.7616) acc5: 100.0000 (100.0000) acc1_neg: 20.312 acc5_neg: 100.000 time: 0.1499 data: 0.0552 max mem: 3531 +Test: [170/269] eta: 0:00:20 acc1_pos: 54.310 acc5_pos: 100.000 loss: 1.1025 (0.9974) acc1: 22.6562 (37.6020) acc5: 100.0000 (100.0000) acc1_neg: 14.323 acc5_neg: 100.000 time: 0.1520 data: 0.0575 max mem: 3531 +Test: [180/269] eta: 0:00:17 acc1_pos: 54.310 acc5_pos: 100.000 loss: 1.1765 (1.0073) acc1: 17.4479 (36.6986) acc5: 100.0000 (100.0000) acc1_neg: 16.146 acc5_neg: 100.000 time: 0.1952 data: 0.1000 max mem: 3531 +Test: [190/269] eta: 0:00:16 acc1_pos: 54.310 acc5_pos: 100.000 loss: 1.1332 (1.0107) acc1: 23.1771 (36.2743) acc5: 100.0000 (100.0000) acc1_neg: 22.656 acc5_neg: 100.000 time: 0.2049 data: 0.1118 max mem: 3531 +Test: [200/269] eta: 0:00:13 acc1_pos: 54.310 acc5_pos: 100.000 loss: 1.1767 (1.0234) acc1: 18.7500 (35.2754) acc5: 100.0000 (100.0000) acc1_neg: 30.208 acc5_neg: 100.000 time: 0.1803 data: 0.0863 max mem: 3531 +Test: [210/269] eta: 0:00:11 acc1_pos: 54.310 acc5_pos: 100.000 loss: 1.1463 (1.0248) acc1: 20.0521 (34.9637) acc5: 100.0000 (100.0000) acc1_neg: 21.615 acc5_neg: 100.000 time: 0.1766 data: 0.0822 max mem: 3532 +Test: [220/269] eta: 0:00:09 acc1_pos: 54.310 acc5_pos: 100.000 loss: 1.0002 (1.0256) acc1: 27.8646 (34.6554) acc5: 100.0000 (100.0000) acc1_neg: 34.896 acc5_neg: 100.000 time: 0.2092 data: 0.1160 max mem: 3532 +Test: [230/269] eta: 0:00:07 acc1_pos: 54.310 acc5_pos: 100.000 loss: 0.9769 (1.0204) acc1: 34.8958 (34.9477) acc5: 100.0000 (100.0000) acc1_neg: 46.875 acc5_neg: 100.000 time: 0.1844 data: 0.0903 max mem: 3532 +Test: [240/269] eta: 0:00:05 acc1_pos: 54.310 acc5_pos: 100.000 loss: 1.0029 (1.0243) acc1: 29.6875 (34.5220) acc5: 100.0000 (100.0000) acc1_neg: 20.833 acc5_neg: 100.000 time: 0.1761 data: 0.0773 max mem: 3532 +Test: [250/269] eta: 0:00:03 acc1_pos: 54.310 acc5_pos: 100.000 loss: 1.1188 (1.0288) acc1: 20.8333 (34.3605) acc5: 100.0000 (100.0000) acc1_neg: 9.896 acc5_neg: 100.000 time: 0.1887 data: 0.0917 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 54.310 acc5_pos: 100.000 loss: 1.2572 (1.0390) acc1: 16.9271 (33.7923) acc5: 100.0000 (100.0000) acc1_neg: 32.031 acc5_neg: 100.000 time: 0.1732 data: 0.0821 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 54.310 acc5_pos: 100.000 loss: 1.0302 (1.0360) acc1: 27.0833 (33.9190) acc5: 100.0000 (100.0000) acc1_neg: 62.467 acc5_neg: 100.000 time: 0.1483 data: 0.0543 max mem: 3532 +Test: Total time: 0:00:52 (0.1951 s / it) +* Overall Acc@1 33.919 Acc@5 100.000 + target == real Acc@1 88.818 Acc@5 100.000 + target == fake Acc@1 26.427 Acc@5 100.000 + loss 1.036 + PR-AUC 0.9283 ROC-AUC 0.6560 EER 39.24% (thr=0.2866) +Accuracy of the network on the 103293 test images: 33.9% +[rank0]:[W726 05:13:56.914393857 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T05:14:03Z] eval_done tag=fixed500k_fake_n11_seed7_resampled +[2026-07-26T05:14:04Z] eval_start tag=fixed500k_real_n5_seed8_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed8_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T05:14:04Z +hostname=n116-040-168 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed8_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n5_seed8_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 05:14:16.904000 3155931 torch/distributed/run.py:766] +W0726 05:14:16.904000 3155931 torch/distributed/run.py:766] ***************************************** +W0726 05:14:16.904000 3155931 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:14:16.904000 3155931 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:14:44.143029058 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:14:45.609350091 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:14:45.665367851 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:14:45.854924728 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:14:45.980977885 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:14:45.986709054 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:14:45.986922850 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:14:45.032485965 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n5_seed8_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed8_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:25:42 acc1_pos: 2.604 acc5_pos: 100.000 loss: 1.7370 (1.7370) acc1: 2.6042 (2.6042) acc5: 100.0000 (100.0000) time: 6.5922 data: 5.0149 max mem: 3531 +Test: [ 10/234] eta: 0:02:37 acc1_pos: 3.906 acc5_pos: 100.000 loss: 1.6915 (1.6121) acc1: 7.0312 (9.6354) acc5: 100.0000 (100.0000) time: 0.7049 data: 0.4779 max mem: 3531 +Test: [ 20/234] eta: 0:01:28 acc1_pos: 25.781 acc5_pos: 100.000 loss: 1.5187 (1.4848) acc1: 25.5208 (20.0893) acc5: 100.0000 (100.0000) time: 0.1066 data: 0.0123 max mem: 3531 +Test: [ 30/234] eta: 0:01:17 acc1_pos: 10.677 acc5_pos: 100.000 loss: 1.5439 (1.5095) acc1: 27.6042 (23.3871) acc5: 100.0000 (100.0000) time: 0.2047 data: 0.1110 max mem: 3531 +Test: [ 40/234] eta: 0:01:08 acc1_pos: 96.354 acc5_pos: 100.000 loss: 1.7115 (1.4180) acc1: 25.2604 (28.2647) acc5: 100.0000 (100.0000) time: 0.2844 data: 0.1941 max mem: 3531 +Test: [ 50/234] eta: 0:00:58 acc1_pos: 95.833 acc5_pos: 100.000 loss: 0.1555 (1.1686) acc1: 95.5729 (41.7177) acc5: 100.0000 (100.0000) time: 0.2237 data: 0.1336 max mem: 3531 +Test: [ 60/234] eta: 0:00:51 acc1_pos: 91.034 acc5_pos: 100.000 loss: 0.1220 (0.9923) acc1: 98.1771 (51.2082) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1905 data: 0.0953 max mem: 3531 +Test: [ 70/234] eta: 0:00:47 acc1_pos: 91.034 acc5_pos: 100.000 loss: 0.0892 (0.9127) acc1: 99.7396 (55.4761) acc5: 100.0000 (100.0000) acc1_neg: 69.531 acc5_neg: 100.000 time: 0.2199 data: 0.1218 max mem: 3531 +Test: [ 80/234] eta: 0:00:43 acc1_pos: 91.034 acc5_pos: 100.000 loss: 0.5481 (0.8501) acc1: 73.4375 (58.8317) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2206 data: 0.1242 max mem: 3531 +Test: [ 90/234] eta: 0:00:38 acc1_pos: 91.034 acc5_pos: 100.000 loss: 0.1156 (0.7721) acc1: 97.6562 (63.0008) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.1699 data: 0.0710 max mem: 3531 +Test: [100/234] eta: 0:00:35 acc1_pos: 91.034 acc5_pos: 100.000 loss: 0.1586 (0.7283) acc1: 96.6146 (65.3672) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1914 data: 0.0938 max mem: 3531 +Test: [110/234] eta: 0:00:31 acc1_pos: 91.034 acc5_pos: 100.000 loss: 0.2020 (0.6972) acc1: 90.8854 (67.0115) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.2116 data: 0.1197 max mem: 3531 +Test: [120/234] eta: 0:00:28 acc1_pos: 91.034 acc5_pos: 100.000 loss: 0.1818 (0.6741) acc1: 94.0104 (68.2808) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2037 data: 0.1133 max mem: 3531 +Test: [130/234] eta: 0:00:25 acc1_pos: 91.034 acc5_pos: 100.000 loss: 0.1966 (0.6420) acc1: 94.5312 (70.0561) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1997 data: 0.1082 max mem: 3531 +Test: [140/234] eta: 0:00:22 acc1_pos: 91.034 acc5_pos: 100.000 loss: 0.2387 (0.6319) acc1: 91.9271 (70.6837) acc5: 100.0000 (100.0000) acc1_neg: 35.156 acc5_neg: 100.000 time: 0.1889 data: 0.0937 max mem: 3531 +Test: [150/234] eta: 0:00:20 acc1_pos: 91.034 acc5_pos: 100.000 loss: 0.3145 (0.6232) acc1: 82.5521 (71.1524) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1986 data: 0.1015 max mem: 3531 +Test: [160/234] eta: 0:00:17 acc1_pos: 91.034 acc5_pos: 100.000 loss: 0.3243 (0.6028) acc1: 88.2812 (72.2875) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.2073 data: 0.1143 max mem: 3531 +Test: [170/234] eta: 0:00:15 acc1_pos: 91.034 acc5_pos: 100.000 loss: 0.3311 (0.5853) acc1: 86.9792 (73.2608) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1966 data: 0.1065 max mem: 3531 +Test: [180/234] eta: 0:00:12 acc1_pos: 91.034 acc5_pos: 100.000 loss: 0.3253 (0.5805) acc1: 86.9792 (73.5181) acc5: 100.0000 (100.0000) acc1_neg: 38.542 acc5_neg: 100.000 time: 0.1792 data: 0.0855 max mem: 3531 +Test: [190/234] eta: 0:00:10 acc1_pos: 91.034 acc5_pos: 100.000 loss: 0.2399 (0.5650) acc1: 92.1875 (74.3865) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1695 data: 0.0720 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 91.034 acc5_pos: 100.000 loss: 0.2325 (0.5450) acc1: 92.7083 (75.5066) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1474 data: 0.0535 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 91.034 acc5_pos: 100.000 loss: 0.1177 (0.5268) acc1: 99.4792 (76.5131) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1364 data: 0.0467 max mem: 3532 +Test: [220/234] eta: 0:00:03 acc1_pos: 91.034 acc5_pos: 100.000 loss: 0.2413 (0.5145) acc1: 91.6667 (77.1694) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1286 data: 0.0386 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 91.034 acc5_pos: 100.000 loss: 0.3070 (0.5110) acc1: 81.7708 (77.2581) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.1208 data: 0.0309 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 91.034 acc5_pos: 100.000 loss: 0.4095 (0.5099) acc1: 80.2083 (77.2816) acc5: 100.0000 (100.0000) acc1_neg: 79.767 acc5_neg: 100.000 time: 0.1217 data: 0.0308 max mem: 3532 +Test: Total time: 0:00:49 (0.2102 s / it) +* Overall Acc@1 77.282 Acc@5 100.000 + target == real Acc@1 42.080 Acc@5 100.000 + target == fake Acc@1 87.203 Acc@5 100.000 + loss 0.510 + PR-AUC 0.8943 ROC-AUC 0.7472 EER 31.47% (thr=0.8398) +Accuracy of the network on the 89729 test images: 77.3% +[rank0]:[W726 05:15:44.050811555 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:16:04.169000 3156489 torch/distributed/run.py:766] +W0726 05:16:04.169000 3156489 torch/distributed/run.py:766] ***************************************** +W0726 05:16:04.169000 3156489 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:16:04.169000 3156489 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:16:31.640885911 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:16:31.675605869 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:16:31.891321964 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:16:31.981234082 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 1): env:// +| distributed init (rank 2): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:16:32.417465742 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank1]:[W726 05:16:32.417465938 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:16:32.420781661 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:16:32.450265217 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n5_seed8_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed8_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:34:32 acc1_pos: 25.521 acc5_pos: 100.000 loss: 1.3603 (1.3603) acc1: 25.5208 (25.5208) acc5: 100.0000 (100.0000) time: 6.9008 data: 5.1347 max mem: 3531 +Test: [ 10/2387] eta: 0:28:05 acc1_pos: 9.896 acc5_pos: 100.000 loss: 1.6665 (1.6688) acc1: 10.6771 (11.6477) acc5: 100.0000 (100.0000) time: 0.7093 data: 0.4672 max mem: 3531 +Test: [ 20/2387] eta: 0:16:32 acc1_pos: 1.562 acc5_pos: 100.000 loss: 1.7167 (1.7196) acc1: 8.0729 (9.4122) acc5: 100.0000 (100.0000) time: 0.0950 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:13:36 acc1_pos: 14.323 acc5_pos: 100.000 loss: 1.7144 (1.7191) acc1: 7.8125 (9.6270) acc5: 100.0000 (100.0000) time: 0.1466 data: 0.0497 max mem: 3531 +Test: [ 40/2387] eta: 0:12:41 acc1_pos: 6.771 acc5_pos: 100.000 loss: 1.7144 (1.6909) acc1: 9.3750 (10.8041) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1302 max mem: 3531 +Test: [ 50/2387] eta: 0:11:38 acc1_pos: 4.427 acc5_pos: 100.000 loss: 1.5857 (1.6402) acc1: 12.2396 (12.7451) acc5: 100.0000 (100.0000) time: 0.2251 data: 0.1325 max mem: 3531 +Test: [ 60/2387] eta: 0:11:11 acc1_pos: 1.823 acc5_pos: 100.000 loss: 1.7315 (1.6828) acc1: 9.3750 (11.7614) acc5: 100.0000 (100.0000) time: 0.2149 data: 0.1213 max mem: 3531 +Test: [ 70/2387] eta: 0:10:49 acc1_pos: 8.333 acc5_pos: 100.000 loss: 1.9099 (1.7347) acc1: 7.5521 (11.1062) acc5: 100.0000 (100.0000) time: 0.2331 data: 0.1361 max mem: 3531 +Test: [ 80/2387] eta: 0:10:38 acc1_pos: 9.896 acc5_pos: 100.000 loss: 1.8024 (1.7380) acc1: 8.3333 (10.8539) acc5: 100.0000 (100.0000) time: 0.2408 data: 0.1457 max mem: 3531 +Test: [ 90/2387] eta: 0:10:25 acc1_pos: 1.302 acc5_pos: 100.000 loss: 1.7727 (1.7407) acc1: 9.1146 (10.3852) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1458 max mem: 3531 +Test: [ 100/2387] eta: 0:10:22 acc1_pos: 5.990 acc5_pos: 100.000 loss: 1.7274 (1.7355) acc1: 9.1146 (10.3857) acc5: 100.0000 (100.0000) time: 0.2540 data: 0.1544 max mem: 3531 +Test: [ 110/2387] eta: 0:10:13 acc1_pos: 7.552 acc5_pos: 100.000 loss: 1.6948 (1.7286) acc1: 9.6354 (10.3721) acc5: 100.0000 (100.0000) time: 0.2549 data: 0.1586 max mem: 3531 +Test: [ 120/2387] eta: 0:10:03 acc1_pos: 7.031 acc5_pos: 100.000 loss: 1.6604 (1.7232) acc1: 9.6354 (10.3930) acc5: 100.0000 (100.0000) time: 0.2349 data: 0.1371 max mem: 3531 +Test: [ 130/2387] eta: 0:09:52 acc1_pos: 4.167 acc5_pos: 100.000 loss: 1.7686 (1.7240) acc1: 8.5938 (10.3570) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1280 max mem: 3531 +Test: [ 140/2387] eta: 0:09:43 acc1_pos: 9.635 acc5_pos: 100.000 loss: 1.6764 (1.7177) acc1: 9.1146 (10.5072) acc5: 100.0000 (100.0000) time: 0.2199 data: 0.1224 max mem: 3531 +Test: [ 150/2387] eta: 0:09:33 acc1_pos: 13.542 acc5_pos: 100.000 loss: 1.6558 (1.7138) acc1: 9.1146 (10.5115) acc5: 100.0000 (100.0000) time: 0.2161 data: 0.1193 max mem: 3531 +Test: [ 160/2387] eta: 0:09:25 acc1_pos: 37.900 acc5_pos: 100.000 loss: 1.5509 (1.6659) acc1: 9.1146 (12.9772) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2149 data: 0.1214 max mem: 3531 +Test: [ 170/2387] eta: 0:09:10 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1682 (1.5782) acc1: 96.6146 (17.8713) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1893 data: 0.0958 max mem: 3531 +Test: [ 180/2387] eta: 0:08:52 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1658 (1.5005) acc1: 96.6146 (22.2217) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1403 data: 0.0463 max mem: 3531 +Test: [ 190/2387] eta: 0:08:45 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1648 (1.4305) acc1: 96.6146 (26.1330) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1593 data: 0.0658 max mem: 3531 +Test: [ 200/2387] eta: 0:08:34 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1648 (1.3677) acc1: 96.6146 (29.6409) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1796 data: 0.0863 max mem: 3531 +Test: [ 210/2387] eta: 0:08:30 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1675 (1.3108) acc1: 96.8750 (32.8347) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1899 data: 0.0976 max mem: 3532 +Test: [ 220/2387] eta: 0:08:25 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1668 (1.2590) acc1: 97.1354 (35.7572) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2148 data: 0.1221 max mem: 3532 +Test: [ 230/2387] eta: 0:08:15 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1628 (1.2115) acc1: 97.3958 (38.4233) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1799 data: 0.0862 max mem: 3532 +Test: [ 240/2387] eta: 0:08:10 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1603 (1.1679) acc1: 97.3958 (40.8789) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1752 data: 0.0818 max mem: 3532 +Test: [ 250/2387] eta: 0:08:03 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1582 (1.1273) acc1: 97.6562 (43.1524) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1867 data: 0.0945 max mem: 3532 +Test: [ 260/2387] eta: 0:08:00 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1328 (1.0885) acc1: 99.2188 (45.3055) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1947 data: 0.1019 max mem: 3532 +Test: [ 270/2387] eta: 0:07:55 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1149 (1.0527) acc1: 99.7396 (47.2921) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2051 data: 0.1113 max mem: 3532 +Test: [ 280/2387] eta: 0:07:53 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1135 (1.0195) acc1: 99.7396 (49.1446) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2097 data: 0.1165 max mem: 3532 +Test: [ 290/2387] eta: 0:07:49 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1247 (0.9892) acc1: 99.2188 (50.8412) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2083 data: 0.1159 max mem: 3532 +Test: [ 300/2387] eta: 0:07:44 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1244 (0.9603) acc1: 99.2188 (52.4510) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1905 data: 0.0980 max mem: 3532 +Test: [ 310/2387] eta: 0:07:39 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1035 (0.9329) acc1: 100.0000 (53.9674) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1867 data: 0.0944 max mem: 3532 +Test: [ 320/2387] eta: 0:07:35 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0894 (0.9066) acc1: 99.7396 (55.3941) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1870 data: 0.0951 max mem: 3532 +Test: [ 330/2387] eta: 0:07:32 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0852 (0.8818) acc1: 99.7396 (56.7362) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2026 data: 0.1076 max mem: 3532 +Test: [ 340/2387] eta: 0:07:28 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0852 (0.8584) acc1: 99.7396 (57.9996) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2022 data: 0.1075 max mem: 3532 +Test: [ 350/2387] eta: 0:07:25 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0827 (0.8359) acc1: 100.0000 (59.1940) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1965 data: 0.1043 max mem: 3532 +Test: [ 360/2387] eta: 0:07:22 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0717 (0.8167) acc1: 99.2188 (60.2652) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2021 data: 0.1103 max mem: 3532 +Test: [ 370/2387] eta: 0:07:17 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1542 (0.7989) acc1: 97.9167 (61.2716) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1860 data: 0.0928 max mem: 3532 +Test: [ 380/2387] eta: 0:07:13 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1529 (0.7820) acc1: 97.9167 (62.2293) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1727 data: 0.0760 max mem: 3532 +Test: [ 390/2387] eta: 0:07:09 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1529 (0.7659) acc1: 97.6562 (63.1334) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1823 data: 0.0878 max mem: 3532 +Test: [ 400/2387] eta: 0:07:05 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1537 (0.7507) acc1: 97.6562 (63.9898) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1799 data: 0.0872 max mem: 3532 +Test: [ 410/2387] eta: 0:07:01 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1477 (0.7360) acc1: 97.6562 (64.8140) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1711 data: 0.0759 max mem: 3532 +Test: [ 420/2387] eta: 0:06:57 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1463 (0.7220) acc1: 97.6562 (65.6021) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1726 data: 0.0780 max mem: 3532 +Test: [ 430/2387] eta: 0:06:53 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1497 (0.7087) acc1: 97.9167 (66.3531) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1792 data: 0.0866 max mem: 3532 +Test: [ 440/2387] eta: 0:06:50 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1439 (0.6953) acc1: 98.4375 (67.1019) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1788 data: 0.0853 max mem: 3533 +Test: [ 450/2387] eta: 0:06:46 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0934 (0.6819) acc1: 100.0000 (67.8313) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1745 data: 0.0813 max mem: 3533 +Test: [ 460/2387] eta: 0:06:42 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0934 (0.6692) acc1: 100.0000 (68.5291) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1756 data: 0.0839 max mem: 3533 +Test: [ 470/2387] eta: 0:06:40 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0935 (0.6570) acc1: 100.0000 (69.1967) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1810 data: 0.0892 max mem: 3533 +Test: [ 480/2387] eta: 0:06:36 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0930 (0.6453) acc1: 100.0000 (69.8371) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1844 data: 0.0926 max mem: 3533 +Test: [ 490/2387] eta: 0:06:33 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0933 (0.6340) acc1: 100.0000 (70.4515) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1794 data: 0.0878 max mem: 3533 +Test: [ 500/2387] eta: 0:06:30 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0942 (0.6233) acc1: 100.0000 (71.0413) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1746 data: 0.0834 max mem: 3533 +Test: [ 510/2387] eta: 0:06:30 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0934 (0.6129) acc1: 100.0000 (71.6080) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2156 data: 0.1242 max mem: 3533 +Test: [ 520/2387] eta: 0:06:29 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0924 (0.6030) acc1: 100.0000 (72.1529) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2491 data: 0.1556 max mem: 3533 +Test: [ 530/2387] eta: 0:06:24 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1057 (0.5936) acc1: 100.0000 (72.6768) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1836 data: 0.0886 max mem: 3533 +Test: [ 540/2387] eta: 0:06:21 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1070 (0.5847) acc1: 100.0000 (73.1819) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1572 data: 0.0628 max mem: 3533 +Test: [ 550/2387] eta: 0:06:18 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1074 (0.5760) acc1: 100.0000 (73.6686) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1906 data: 0.0968 max mem: 3533 +Test: [ 560/2387] eta: 0:06:17 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1064 (0.5676) acc1: 100.0000 (74.1380) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2124 data: 0.1210 max mem: 3533 +Test: [ 570/2387] eta: 0:06:15 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1066 (0.5596) acc1: 100.0000 (74.5909) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2117 data: 0.1206 max mem: 3533 +Test: [ 580/2387] eta: 0:06:11 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1066 (0.5518) acc1: 100.0000 (75.0282) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1819 data: 0.0908 max mem: 3533 +Test: [ 590/2387] eta: 0:06:08 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1063 (0.5442) acc1: 100.0000 (75.4508) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1723 data: 0.0819 max mem: 3533 +Test: [ 600/2387] eta: 0:06:06 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0954 (0.5367) acc1: 100.0000 (75.8592) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1784 data: 0.0863 max mem: 3533 +Test: [ 610/2387] eta: 0:06:03 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0923 (0.5294) acc1: 100.0000 (76.2543) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1751 data: 0.0833 max mem: 3533 +Test: [ 620/2387] eta: 0:05:59 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0921 (0.5224) acc1: 100.0000 (76.6367) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1620 data: 0.0702 max mem: 3533 +Test: [ 630/2387] eta: 0:05:56 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0914 (0.5156) acc1: 100.0000 (77.0066) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1597 data: 0.0654 max mem: 3533 +Test: [ 640/2387] eta: 0:05:53 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0917 (0.5089) acc1: 100.0000 (77.3653) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1728 data: 0.0787 max mem: 3533 +Test: [ 650/2387] eta: 0:05:50 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0917 (0.5025) acc1: 100.0000 (77.7130) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1605 data: 0.0682 max mem: 3533 +Test: [ 660/2387] eta: 0:05:46 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0919 (0.4963) acc1: 100.0000 (78.0501) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1444 data: 0.0521 max mem: 3533 +Test: [ 670/2387] eta: 0:05:43 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0911 (0.4903) acc1: 100.0000 (78.3773) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1507 data: 0.0575 max mem: 3534 +Test: [ 680/2387] eta: 0:05:41 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0898 (0.4844) acc1: 100.0000 (78.6944) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1666 data: 0.0742 max mem: 3534 +Test: [ 690/2387] eta: 0:05:38 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0892 (0.4787) acc1: 100.0000 (78.9975) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1692 data: 0.0770 max mem: 3534 +Test: [ 700/2387] eta: 0:05:35 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0893 (0.4732) acc1: 99.4792 (79.2893) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1601 data: 0.0687 max mem: 3534 +Test: [ 710/2387] eta: 0:05:33 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0911 (0.4678) acc1: 99.4792 (79.5747) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1808 data: 0.0904 max mem: 3534 +Test: [ 720/2387] eta: 0:05:30 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0895 (0.4626) acc1: 99.4792 (79.8497) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1876 data: 0.0960 max mem: 3534 +Test: [ 730/2387] eta: 0:05:28 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0894 (0.4575) acc1: 99.7396 (80.1207) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1813 data: 0.0901 max mem: 3534 +Test: [ 740/2387] eta: 0:05:26 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0899 (0.4525) acc1: 99.4792 (80.3809) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1949 data: 0.1015 max mem: 3534 +Test: [ 750/2387] eta: 0:05:24 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0914 (0.4477) acc1: 99.4792 (80.6373) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2034 data: 0.1084 max mem: 3534 +Test: [ 760/2387] eta: 0:05:22 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0903 (0.4430) acc1: 99.7396 (80.8876) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2072 data: 0.1146 max mem: 3534 +Test: [ 770/2387] eta: 0:05:20 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0895 (0.4388) acc1: 99.7396 (81.1257) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1998 data: 0.1079 max mem: 3534 +Test: [ 780/2387] eta: 0:05:18 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1202 (0.4360) acc1: 99.4792 (81.2620) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1876 data: 0.0958 max mem: 3534 +Test: [ 790/2387] eta: 0:05:16 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1538 (0.4326) acc1: 98.4375 (81.4620) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1902 data: 0.0989 max mem: 3534 +Test: [ 800/2387] eta: 0:05:15 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1354 (0.4288) acc1: 98.9583 (81.6879) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2099 data: 0.1182 max mem: 3534 +Test: [ 810/2387] eta: 0:05:14 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1354 (0.4255) acc1: 99.2188 (81.8954) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2432 data: 0.1475 max mem: 3534 +Test: [ 820/2387] eta: 0:05:12 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1509 (0.4222) acc1: 98.4375 (82.1004) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2299 data: 0.1318 max mem: 3534 +Test: [ 830/2387] eta: 0:05:08 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1736 (0.4190) acc1: 98.4375 (82.2851) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1567 data: 0.0600 max mem: 3534 +Test: [ 840/2387] eta: 0:05:06 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1589 (0.4159) acc1: 97.1354 (82.4610) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1554 data: 0.0615 max mem: 3534 +Test: [ 850/2387] eta: 0:05:05 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1362 (0.4125) acc1: 99.2188 (82.6531) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2181 data: 0.1255 max mem: 3534 +Test: [ 860/2387] eta: 0:05:03 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1172 (0.4091) acc1: 99.4792 (82.8412) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2244 data: 0.1297 max mem: 3534 +Test: [ 870/2387] eta: 0:05:00 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1166 (0.4059) acc1: 99.4792 (83.0182) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1700 data: 0.0746 max mem: 3534 +Test: [ 880/2387] eta: 0:04:58 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1361 (0.4031) acc1: 98.4375 (83.1781) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1499 data: 0.0565 max mem: 3534 +Test: [ 890/2387] eta: 0:04:55 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1305 (0.3999) acc1: 98.6979 (83.3593) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1807 data: 0.0887 max mem: 3534 +Test: [ 900/2387] eta: 0:04:53 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1297 (0.3979) acc1: 98.9583 (83.4805) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1949 data: 0.1017 max mem: 3535 +Test: [ 910/2387] eta: 0:04:51 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.2316 (0.3961) acc1: 93.2292 (83.5840) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1941 data: 0.1007 max mem: 3535 +Test: [ 920/2387] eta: 0:04:49 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.2492 (0.3945) acc1: 92.7083 (83.6825) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1899 data: 0.0967 max mem: 3535 +Test: [ 930/2387] eta: 0:04:47 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.2441 (0.3928) acc1: 92.7083 (83.7823) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1801 data: 0.0870 max mem: 3535 +Test: [ 940/2387] eta: 0:04:45 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.2360 (0.3912) acc1: 92.7083 (83.8777) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1805 data: 0.0868 max mem: 3535 +Test: [ 950/2387] eta: 0:04:42 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.2358 (0.3896) acc1: 92.9688 (83.9774) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1813 data: 0.0878 max mem: 3535 +Test: [ 960/2387] eta: 0:04:40 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.2331 (0.3879) acc1: 93.4896 (84.0796) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1699 data: 0.0786 max mem: 3535 +Test: [ 970/2387] eta: 0:04:38 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.2349 (0.3864) acc1: 93.2292 (84.1728) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1700 data: 0.0772 max mem: 3535 +Test: [ 980/2387] eta: 0:04:36 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.2313 (0.3844) acc1: 93.4896 (84.2906) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1842 data: 0.0902 max mem: 3535 +Test: [ 990/2387] eta: 0:04:34 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1919 (0.3824) acc1: 96.0938 (84.4165) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2082 data: 0.1148 max mem: 3535 +Test: [1000/2387] eta: 0:04:31 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1492 (0.3800) acc1: 98.4375 (84.5597) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1799 data: 0.0829 max mem: 3535 +Test: [1010/2387] eta: 0:04:29 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1490 (0.3780) acc1: 98.9583 (84.6913) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1649 data: 0.0661 max mem: 3535 +Test: [1020/2387] eta: 0:04:27 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1854 (0.3760) acc1: 97.3958 (84.8076) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1899 data: 0.0943 max mem: 3535 +Test: [1030/2387] eta: 0:04:25 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1841 (0.3743) acc1: 96.8750 (84.9148) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1799 data: 0.0839 max mem: 3535 +Test: [1040/2387] eta: 0:04:23 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1730 (0.3723) acc1: 98.4375 (85.0384) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1730 data: 0.0754 max mem: 3535 +Test: [1050/2387] eta: 0:04:21 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1592 (0.3703) acc1: 98.9583 (85.1699) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1766 data: 0.0789 max mem: 3535 +Test: [1060/2387] eta: 0:04:19 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1621 (0.3687) acc1: 98.6979 (85.2726) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1868 data: 0.0901 max mem: 3535 +Test: [1070/2387] eta: 0:04:16 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1746 (0.3668) acc1: 98.6979 (85.3965) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1881 data: 0.0937 max mem: 3535 +Test: [1080/2387] eta: 0:04:14 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1746 (0.3653) acc1: 97.6562 (85.4930) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1849 data: 0.0919 max mem: 3535 +Test: [1090/2387] eta: 0:04:12 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.2148 (0.3650) acc1: 94.5312 (85.5222) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.1767 data: 0.0857 max mem: 3535 +Test: [1100/2387] eta: 0:04:10 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.2591 (0.3647) acc1: 91.6667 (85.5281) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1699 data: 0.0788 max mem: 3535 +Test: [1110/2387] eta: 0:04:08 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.2974 (0.3644) acc1: 90.3646 (85.5512) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1705 data: 0.0792 max mem: 3535 +Test: [1120/2387] eta: 0:04:06 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.2854 (0.3636) acc1: 91.6667 (85.6132) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1714 data: 0.0787 max mem: 3536 +Test: [1130/2387] eta: 0:04:04 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.2689 (0.3632) acc1: 92.1875 (85.6297) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1875 data: 0.0924 max mem: 3536 +Test: [1140/2387] eta: 0:04:02 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.2303 (0.3620) acc1: 93.2292 (85.7095) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1947 data: 0.1016 max mem: 3536 +Test: [1150/2387] eta: 0:04:00 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1652 (0.3601) acc1: 97.3958 (85.8194) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1860 data: 0.0933 max mem: 3536 +Test: [1160/2387] eta: 0:03:58 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0987 (0.3578) acc1: 100.0000 (85.9409) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1815 data: 0.0875 max mem: 3536 +Test: [1170/2387] eta: 0:03:55 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0910 (0.3555) acc1: 100.0000 (86.0578) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1811 data: 0.0887 max mem: 3536 +Test: [1180/2387] eta: 0:03:53 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0840 (0.3533) acc1: 100.0000 (86.1701) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1819 data: 0.0908 max mem: 3536 +Test: [1190/2387] eta: 0:03:51 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0881 (0.3512) acc1: 100.0000 (86.2823) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1839 data: 0.0934 max mem: 3536 +Test: [1200/2387] eta: 0:03:49 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0915 (0.3490) acc1: 100.0000 (86.3926) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1867 data: 0.0959 max mem: 3536 +Test: [1210/2387] eta: 0:03:47 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0854 (0.3470) acc1: 100.0000 (86.4955) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1890 data: 0.0977 max mem: 3536 +Test: [1220/2387] eta: 0:03:45 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1375 (0.3456) acc1: 99.2188 (86.5842) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1769 data: 0.0850 max mem: 3536 +Test: [1230/2387] eta: 0:03:43 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1399 (0.3438) acc1: 99.2188 (86.6870) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1591 data: 0.0672 max mem: 3536 +Test: [1240/2387] eta: 0:03:41 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1212 (0.3421) acc1: 99.4792 (86.7846) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1520 data: 0.0601 max mem: 3536 +Test: [1250/2387] eta: 0:03:39 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1356 (0.3405) acc1: 99.4792 (86.8770) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1695 data: 0.0775 max mem: 3536 +Test: [1260/2387] eta: 0:03:36 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1500 (0.3390) acc1: 98.6979 (86.9631) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1744 data: 0.0821 max mem: 3536 +Test: [1270/2387] eta: 0:03:34 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1275 (0.3374) acc1: 99.4792 (87.0574) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1651 data: 0.0730 max mem: 3536 +Test: [1280/2387] eta: 0:03:32 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1306 (0.3360) acc1: 99.2188 (87.1412) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1608 data: 0.0690 max mem: 3536 +Test: [1290/2387] eta: 0:03:30 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1592 (0.3347) acc1: 98.1771 (87.2158) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1543 data: 0.0617 max mem: 3536 +Test: [1300/2387] eta: 0:03:28 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1585 (0.3333) acc1: 98.9583 (87.3012) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1544 data: 0.0624 max mem: 3536 +Test: [1310/2387] eta: 0:03:25 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1585 (0.3320) acc1: 98.4375 (87.3772) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1527 data: 0.0613 max mem: 3536 +Test: [1320/2387] eta: 0:03:23 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1769 (0.3311) acc1: 97.1354 (87.4336) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1538 data: 0.0626 max mem: 3536 +Test: [1330/2387] eta: 0:03:21 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1788 (0.3298) acc1: 97.6562 (87.5112) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1521 data: 0.0612 max mem: 3536 +Test: [1340/2387] eta: 0:03:19 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1607 (0.3287) acc1: 98.6979 (87.5783) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1578 data: 0.0665 max mem: 3536 +Test: [1350/2387] eta: 0:03:17 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0767 (0.3268) acc1: 100.0000 (87.6702) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1764 data: 0.0843 max mem: 3537 +Test: [1360/2387] eta: 0:03:15 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0693 (0.3249) acc1: 100.0000 (87.7608) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1766 data: 0.0854 max mem: 3537 +Test: [1370/2387] eta: 0:03:12 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0685 (0.3230) acc1: 100.0000 (87.8501) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1552 data: 0.0643 max mem: 3537 +Test: [1380/2387] eta: 0:03:10 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0685 (0.3212) acc1: 100.0000 (87.9381) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1473 data: 0.0554 max mem: 3537 +Test: [1390/2387] eta: 0:03:08 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0669 (0.3193) acc1: 100.0000 (88.0248) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1452 data: 0.0538 max mem: 3537 +Test: [1400/2387] eta: 0:03:06 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0685 (0.3176) acc1: 100.0000 (88.1102) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1391 data: 0.0478 max mem: 3537 +Test: [1410/2387] eta: 0:03:04 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0718 (0.3164) acc1: 100.0000 (88.1773) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1419 data: 0.0477 max mem: 3537 +Test: [1420/2387] eta: 0:03:01 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1953 (0.3163) acc1: 95.5729 (88.1799) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1451 data: 0.0491 max mem: 3537 +Test: [1430/2387] eta: 0:02:59 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.2076 (0.3154) acc1: 95.0521 (88.2359) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1573 data: 0.0616 max mem: 3537 +Test: [1440/2387] eta: 0:02:57 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1760 (0.3147) acc1: 96.0938 (88.2870) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1676 data: 0.0719 max mem: 3537 +Test: [1450/2387] eta: 0:02:55 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1663 (0.3138) acc1: 96.6146 (88.3378) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1698 data: 0.0750 max mem: 3537 +Test: [1460/2387] eta: 0:02:53 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1655 (0.3127) acc1: 96.6146 (88.3957) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1707 data: 0.0779 max mem: 3537 +Test: [1470/2387] eta: 0:02:51 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1164 (0.3113) acc1: 100.0000 (88.4733) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1503 data: 0.0594 max mem: 3537 +Test: [1480/2387] eta: 0:02:49 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1053 (0.3100) acc1: 100.0000 (88.5411) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1446 data: 0.0542 max mem: 3537 +Test: [1490/2387] eta: 0:02:47 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1110 (0.3088) acc1: 99.2188 (88.6040) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1529 data: 0.0609 max mem: 3537 +Test: [1500/2387] eta: 0:02:45 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1299 (0.3076) acc1: 98.9583 (88.6704) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1510 data: 0.0584 max mem: 3537 +Test: [1510/2387] eta: 0:02:43 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1128 (0.3063) acc1: 99.4792 (88.7443) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1653 data: 0.0742 max mem: 3537 +Test: [1520/2387] eta: 0:02:41 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1102 (0.3052) acc1: 99.4792 (88.8052) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1636 data: 0.0730 max mem: 3537 +Test: [1530/2387] eta: 0:02:39 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1179 (0.3039) acc1: 99.4792 (88.8718) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1642 data: 0.0718 max mem: 3537 +Test: [1540/2387] eta: 0:02:37 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1067 (0.3027) acc1: 99.7396 (88.9395) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1791 data: 0.0869 max mem: 3537 +Test: [1550/2387] eta: 0:02:35 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1045 (0.3015) acc1: 99.7396 (89.0049) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1808 data: 0.0880 max mem: 3537 +Test: [1560/2387] eta: 0:02:33 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1253 (0.3004) acc1: 98.9583 (89.0640) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1853 data: 0.0885 max mem: 3537 +Test: [1570/2387] eta: 0:02:32 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1139 (0.2992) acc1: 99.4792 (89.1311) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1843 data: 0.0888 max mem: 3537 +Test: [1580/2387] eta: 0:02:30 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1118 (0.2982) acc1: 99.4792 (89.1836) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1861 data: 0.0934 max mem: 3538 +Test: [1590/2387] eta: 0:02:28 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1630 (0.2980) acc1: 97.1354 (89.1931) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1881 data: 0.0949 max mem: 3538 +Test: [1600/2387] eta: 0:02:26 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1754 (0.2973) acc1: 97.3958 (89.2248) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1749 data: 0.0829 max mem: 3538 +Test: [1610/2387] eta: 0:02:24 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1612 (0.2965) acc1: 98.4375 (89.2752) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1771 data: 0.0848 max mem: 3538 +Test: [1620/2387] eta: 0:02:22 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1809 (0.2960) acc1: 97.1354 (89.3160) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1783 data: 0.0859 max mem: 3538 +Test: [1630/2387] eta: 0:02:20 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1837 (0.2952) acc1: 96.8750 (89.3659) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1849 data: 0.0943 max mem: 3538 +Test: [1640/2387] eta: 0:02:18 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1617 (0.2944) acc1: 98.6979 (89.4177) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1734 data: 0.0805 max mem: 3538 +Test: [1650/2387] eta: 0:02:16 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1605 (0.2936) acc1: 98.9583 (89.4685) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1630 data: 0.0699 max mem: 3538 +Test: [1660/2387] eta: 0:02:15 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1632 (0.2928) acc1: 98.4375 (89.5134) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1868 data: 0.0958 max mem: 3538 +Test: [1670/2387] eta: 0:02:13 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1653 (0.2921) acc1: 97.9167 (89.5570) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1922 data: 0.1002 max mem: 3538 +Test: [1680/2387] eta: 0:02:11 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1663 (0.2914) acc1: 97.6562 (89.6030) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1901 data: 0.0986 max mem: 3538 +Test: [1690/2387] eta: 0:02:09 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1664 (0.2908) acc1: 97.9167 (89.6460) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1924 data: 0.1022 max mem: 3538 +Test: [1700/2387] eta: 0:02:07 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1688 (0.2901) acc1: 98.4375 (89.6827) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1897 data: 0.0983 max mem: 3538 +Test: [1710/2387] eta: 0:02:05 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1572 (0.2893) acc1: 98.6979 (89.7294) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1809 data: 0.0887 max mem: 3538 +Test: [1720/2387] eta: 0:02:03 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1365 (0.2884) acc1: 98.6979 (89.7811) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1759 data: 0.0843 max mem: 3538 +Test: [1730/2387] eta: 0:02:02 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1158 (0.2874) acc1: 98.9583 (89.8313) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1798 data: 0.0890 max mem: 3538 +Test: [1740/2387] eta: 0:02:00 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1299 (0.2866) acc1: 98.6979 (89.8753) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1732 data: 0.0833 max mem: 3538 +Test: [1750/2387] eta: 0:01:58 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1507 (0.2859) acc1: 98.4375 (89.9220) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1647 data: 0.0746 max mem: 3538 +Test: [1760/2387] eta: 0:01:56 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1275 (0.2850) acc1: 99.7396 (89.9700) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1703 data: 0.0789 max mem: 3538 +Test: [1770/2387] eta: 0:01:54 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1331 (0.2843) acc1: 98.9583 (90.0098) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1693 data: 0.0774 max mem: 3538 +Test: [1780/2387] eta: 0:01:52 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1293 (0.2833) acc1: 98.9583 (90.0613) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1696 data: 0.0781 max mem: 3538 +Test: [1790/2387] eta: 0:01:50 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1055 (0.2823) acc1: 99.7396 (90.1109) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1773 data: 0.0824 max mem: 3538 +Test: [1800/2387] eta: 0:01:48 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1011 (0.2814) acc1: 99.7396 (90.1620) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1866 data: 0.0924 max mem: 3538 +Test: [1810/2387] eta: 0:01:46 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1123 (0.2805) acc1: 99.7396 (90.2083) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1844 data: 0.0927 max mem: 3539 +Test: [1820/2387] eta: 0:01:44 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1123 (0.2796) acc1: 99.7396 (90.2590) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1742 data: 0.0824 max mem: 3539 +Test: [1830/2387] eta: 0:01:43 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1085 (0.2786) acc1: 100.0000 (90.3100) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1730 data: 0.0818 max mem: 3539 +Test: [1840/2387] eta: 0:01:41 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0930 (0.2776) acc1: 100.0000 (90.3626) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1714 data: 0.0797 max mem: 3539 +Test: [1850/2387] eta: 0:01:39 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0930 (0.2766) acc1: 100.0000 (90.4145) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1751 data: 0.0833 max mem: 3539 +Test: [1860/2387] eta: 0:01:37 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0938 (0.2756) acc1: 100.0000 (90.4659) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2000 data: 0.1064 max mem: 3539 +Test: [1870/2387] eta: 0:01:35 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0933 (0.2746) acc1: 100.0000 (90.5169) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2036 data: 0.1078 max mem: 3539 +Test: [1880/2387] eta: 0:01:34 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0929 (0.2737) acc1: 100.0000 (90.5673) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2079 data: 0.1121 max mem: 3539 +Test: [1890/2387] eta: 0:01:32 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0932 (0.2727) acc1: 100.0000 (90.6172) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1876 data: 0.0942 max mem: 3539 +Test: [1900/2387] eta: 0:01:30 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0940 (0.2718) acc1: 100.0000 (90.6660) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1848 data: 0.0901 max mem: 3539 +Test: [1910/2387] eta: 0:01:28 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1375 (0.2713) acc1: 99.4792 (90.6976) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.2322 data: 0.1358 max mem: 3539 +Test: [1920/2387] eta: 0:01:26 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1810 (0.2715) acc1: 94.5312 (90.6741) acc5: 100.0000 (100.0000) acc1_neg: 50.260 acc5_neg: 100.000 time: 0.2134 data: 0.1199 max mem: 3539 +Test: [1930/2387] eta: 0:01:24 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.2185 (0.2712) acc1: 93.7500 (90.6978) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1773 data: 0.0855 max mem: 3539 +Test: [1940/2387] eta: 0:01:23 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.2105 (0.2712) acc1: 97.9167 (90.7232) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1889 data: 0.0974 max mem: 3539 +Test: [1950/2387] eta: 0:01:21 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.2105 (0.2710) acc1: 97.9167 (90.7279) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.1840 data: 0.0931 max mem: 3539 +Test: [1960/2387] eta: 0:01:19 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.2488 (0.2713) acc1: 90.8854 (90.6964) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1757 data: 0.0845 max mem: 3539 +Test: [1970/2387] eta: 0:01:17 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.2703 (0.2714) acc1: 90.8854 (90.6858) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1749 data: 0.0838 max mem: 3539 +Test: [1980/2387] eta: 0:01:15 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.2376 (0.2709) acc1: 93.2292 (90.7110) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1695 data: 0.0781 max mem: 3539 +Test: [1990/2387] eta: 0:01:13 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1072 (0.2701) acc1: 99.4792 (90.7551) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1782 data: 0.0860 max mem: 3539 +Test: [2000/2387] eta: 0:01:11 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1060 (0.2693) acc1: 99.4792 (90.7989) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1697 data: 0.0781 max mem: 3539 +Test: [2010/2387] eta: 0:01:09 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1061 (0.2685) acc1: 99.4792 (90.8419) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1709 data: 0.0762 max mem: 3539 +Test: [2020/2387] eta: 0:01:08 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1064 (0.2677) acc1: 99.4792 (90.8841) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1896 data: 0.0954 max mem: 3539 +Test: [2030/2387] eta: 0:01:06 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1052 (0.2669) acc1: 99.4792 (90.9268) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1944 data: 0.1030 max mem: 3540 +Test: [2040/2387] eta: 0:01:04 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1052 (0.2661) acc1: 99.4792 (90.9689) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1861 data: 0.0942 max mem: 3540 +Test: [2050/2387] eta: 0:01:02 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1048 (0.2653) acc1: 99.7396 (91.0118) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2096 data: 0.1185 max mem: 3540 +Test: [2060/2387] eta: 0:01:00 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1069 (0.2645) acc1: 99.7396 (91.0541) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1981 data: 0.1055 max mem: 3540 +Test: [2070/2387] eta: 0:00:58 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1090 (0.2638) acc1: 99.7396 (91.0963) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1648 data: 0.0723 max mem: 3540 +Test: [2080/2387] eta: 0:00:56 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1085 (0.2630) acc1: 99.7396 (91.1381) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1700 data: 0.0776 max mem: 3540 +Test: [2090/2387] eta: 0:00:55 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1080 (0.2623) acc1: 99.7396 (91.1793) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1764 data: 0.0840 max mem: 3540 +Test: [2100/2387] eta: 0:00:53 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1082 (0.2616) acc1: 99.7396 (91.2200) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1874 data: 0.0959 max mem: 3540 +Test: [2110/2387] eta: 0:00:51 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1097 (0.2608) acc1: 99.7396 (91.2602) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1953 data: 0.1006 max mem: 3540 +Test: [2120/2387] eta: 0:00:49 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1131 (0.2603) acc1: 99.4792 (91.2935) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1936 data: 0.0996 max mem: 3540 +Test: [2130/2387] eta: 0:00:47 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1718 (0.2604) acc1: 97.6562 (91.2894) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1894 data: 0.0989 max mem: 3540 +Test: [2140/2387] eta: 0:00:45 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1978 (0.2601) acc1: 96.8750 (91.3091) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1810 data: 0.0904 max mem: 3540 +Test: [2150/2387] eta: 0:00:43 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1623 (0.2597) acc1: 97.9167 (91.3397) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1771 data: 0.0866 max mem: 3540 +Test: [2160/2387] eta: 0:00:42 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1717 (0.2595) acc1: 98.4375 (91.3643) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1844 data: 0.0928 max mem: 3540 +Test: [2170/2387] eta: 0:00:40 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1874 (0.2591) acc1: 97.6562 (91.3935) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1911 data: 0.0984 max mem: 3540 +Test: [2180/2387] eta: 0:00:38 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1606 (0.2586) acc1: 99.2188 (91.4223) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1885 data: 0.0966 max mem: 3540 +Test: [2190/2387] eta: 0:00:36 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1192 (0.2580) acc1: 99.2188 (91.4522) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1906 data: 0.0993 max mem: 3540 +Test: [2200/2387] eta: 0:00:34 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1885 (0.2579) acc1: 97.6562 (91.4616) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1943 data: 0.1022 max mem: 3540 +Test: [2210/2387] eta: 0:00:32 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1891 (0.2575) acc1: 95.8333 (91.4875) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1903 data: 0.0960 max mem: 3540 +Test: [2220/2387] eta: 0:00:30 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1611 (0.2572) acc1: 98.9583 (91.5107) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1859 data: 0.0929 max mem: 3540 +Test: [2230/2387] eta: 0:00:29 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1801 (0.2568) acc1: 98.6979 (91.5332) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1888 data: 0.0950 max mem: 3540 +Test: [2240/2387] eta: 0:00:27 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1901 (0.2569) acc1: 95.3125 (91.5304) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1989 data: 0.1038 max mem: 3540 +Test: [2250/2387] eta: 0:00:25 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1847 (0.2563) acc1: 95.5729 (91.5608) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1899 data: 0.0967 max mem: 3540 +Test: [2260/2387] eta: 0:00:23 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1184 (0.2558) acc1: 99.4792 (91.5917) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1787 data: 0.0850 max mem: 3541 +Test: [2270/2387] eta: 0:00:21 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1601 (0.2557) acc1: 98.6979 (91.5912) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1912 data: 0.0966 max mem: 3541 +Test: [2280/2387] eta: 0:00:19 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1602 (0.2553) acc1: 98.6979 (91.6176) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1963 data: 0.1032 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1434 (0.2548) acc1: 99.4792 (91.6498) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2036 data: 0.1085 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1470 (0.2545) acc1: 98.6979 (91.6765) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2246 data: 0.1295 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1573 (0.2540) acc1: 98.4375 (91.7061) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2061 data: 0.1143 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1588 (0.2538) acc1: 97.6562 (91.7202) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1776 data: 0.0838 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1635 (0.2539) acc1: 96.8750 (91.7129) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1855 data: 0.0931 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1532 (0.2539) acc1: 98.1771 (91.7123) acc5: 100.0000 (100.0000) acc1_neg: 72.135 acc5_neg: 100.000 time: 0.1839 data: 0.0915 max mem: 3541 +Test: [2350/2387] eta: 0:00:06 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1735 (0.2536) acc1: 98.4375 (91.7373) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1665 data: 0.0733 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1898 (0.2535) acc1: 97.1354 (91.7508) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1562 data: 0.0653 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.2306 (0.2535) acc1: 95.0521 (91.7567) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1484 data: 0.0578 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.2061 (0.2533) acc1: 95.5729 (91.7759) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1469 data: 0.0560 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1804 (0.2531) acc1: 97.3958 (91.7854) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1358 data: 0.0477 max mem: 3541 +Test: Total time: 0:07:22 (0.1854 s / it) +* Overall Acc@1 91.785 Acc@5 100.000 + target == real Acc@1 10.567 Acc@5 100.000 + target == fake Acc@1 97.489 Acc@5 100.000 + loss 0.253 + PR-AUC 0.9733 ROC-AUC 0.7479 EER 30.48% (thr=0.8828) +Accuracy of the network on the 916267 test images: 91.8% +[rank0]:[W726 05:24:18.415808714 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:24:39.975000 3157108 torch/distributed/run.py:766] +W0726 05:24:39.975000 3157108 torch/distributed/run.py:766] ***************************************** +W0726 05:24:39.975000 3157108 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:24:39.975000 3157108 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 05:25:17.871313701 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:25:17.980343385 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:25:17.047281929 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:25:18.376775902 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:25:18.617605238 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:25:18.660809306 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:25:18.693755835 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:25:18.739463778 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n5_seed8_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed8_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 0:57:40 acc1_pos: 25.521 acc5_pos: 100.000 loss: 1.3603 (1.3603) acc1: 25.5208 (25.5208) acc5: 100.0000 (100.0000) time: 6.6941 data: 5.0126 max mem: 3531 +Test: [ 10/517] eta: 0:05:49 acc1_pos: 9.896 acc5_pos: 100.000 loss: 1.6665 (1.6688) acc1: 10.6771 (11.6477) acc5: 100.0000 (100.0000) time: 0.6903 data: 0.4561 max mem: 3531 +Test: [ 20/517] eta: 0:03:21 acc1_pos: 1.562 acc5_pos: 100.000 loss: 1.7167 (1.7196) acc1: 8.0729 (9.4122) acc5: 100.0000 (100.0000) time: 0.0917 data: 0.0005 max mem: 3531 +Test: [ 30/517] eta: 0:02:47 acc1_pos: 14.323 acc5_pos: 100.000 loss: 1.7144 (1.7191) acc1: 7.8125 (9.6270) acc5: 100.0000 (100.0000) time: 0.1546 data: 0.0560 max mem: 3531 +Test: [ 40/517] eta: 0:02:32 acc1_pos: 6.771 acc5_pos: 100.000 loss: 1.7144 (1.6909) acc1: 9.3750 (10.8041) acc5: 100.0000 (100.0000) time: 0.2278 data: 0.1279 max mem: 3531 +Test: [ 50/517] eta: 0:02:20 acc1_pos: 12.385 acc5_pos: 100.000 loss: 1.3940 (1.5754) acc1: 20.3125 (16.0488) acc5: 100.0000 (100.0000) acc1_neg: 78.646 acc5_neg: 100.000 time: 0.2311 data: 0.1381 max mem: 3531 +Test: [ 60/517] eta: 0:02:12 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.4093 (1.3727) acc1: 78.6458 (27.8432) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2321 data: 0.1417 max mem: 3531 +Test: [ 70/517] eta: 0:02:06 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.2461 (1.2112) acc1: 93.7500 (37.5624) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2389 data: 0.1456 max mem: 3531 +Test: [ 80/517] eta: 0:02:01 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.2364 (1.0897) acc1: 97.6562 (44.9878) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2396 data: 0.1459 max mem: 3531 +Test: [ 90/517] eta: 0:01:57 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.1728 (0.9867) acc1: 98.4375 (50.9243) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2500 data: 0.1588 max mem: 3531 +Test: [100/517] eta: 0:01:53 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.1585 (0.9090) acc1: 98.6979 (55.5177) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.2481 data: 0.1566 max mem: 3531 +Test: [110/517] eta: 0:01:49 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.2560 (0.8561) acc1: 90.3646 (58.4154) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.2361 data: 0.1455 max mem: 3531 +Test: [120/517] eta: 0:01:47 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.2643 (0.8052) acc1: 90.3646 (61.3550) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2599 data: 0.1665 max mem: 3531 +Test: [130/517] eta: 0:01:43 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.2180 (0.7606) acc1: 95.5729 (63.9830) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2634 data: 0.1662 max mem: 3531 +Test: [140/517] eta: 0:01:40 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.2555 (0.7279) acc1: 93.2292 (65.9002) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.2399 data: 0.1429 max mem: 3531 +Test: [150/517] eta: 0:01:37 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.2770 (0.6959) acc1: 92.1875 (67.7721) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2449 data: 0.1495 max mem: 3531 +Test: [160/517] eta: 0:01:34 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.2646 (0.6692) acc1: 94.0104 (69.4083) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2551 data: 0.1620 max mem: 3531 +Test: [170/517] eta: 0:01:31 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.2646 (0.6443) acc1: 94.5312 (70.9278) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2449 data: 0.1520 max mem: 3531 +Test: [180/517] eta: 0:01:28 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.2561 (0.6201) acc1: 94.7917 (72.2275) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2447 data: 0.1497 max mem: 3531 +Test: [190/517] eta: 0:01:25 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.1191 (0.5939) acc1: 100.0000 (73.6816) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2499 data: 0.1510 max mem: 3531 +Test: [200/517] eta: 0:01:22 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.1208 (0.5707) acc1: 100.0000 (74.9909) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2299 data: 0.1307 max mem: 3531 +Test: [210/517] eta: 0:01:19 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.1424 (0.5612) acc1: 94.5312 (75.5949) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.2449 data: 0.1459 max mem: 3532 +Test: [220/517] eta: 0:01:16 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.3719 (0.5527) acc1: 88.8021 (76.1571) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.2550 data: 0.1552 max mem: 3532 +Test: [230/517] eta: 0:01:13 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.3180 (0.5407) acc1: 90.1042 (76.9875) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2348 data: 0.1383 max mem: 3532 +Test: [240/517] eta: 0:01:10 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.2487 (0.5280) acc1: 96.8750 (77.8419) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2248 data: 0.1284 max mem: 3532 +Test: [250/517] eta: 0:01:08 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.2154 (0.5158) acc1: 97.6562 (78.6313) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2249 data: 0.1248 max mem: 3532 +Test: [260/517] eta: 0:01:05 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.2154 (0.5046) acc1: 97.9167 (79.3582) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2299 data: 0.1310 max mem: 3532 +Test: [270/517] eta: 0:01:02 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.2071 (0.4930) acc1: 98.9583 (80.1036) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2299 data: 0.1357 max mem: 3532 +Test: [280/517] eta: 0:00:59 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.1806 (0.4818) acc1: 99.4792 (80.7922) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2357 data: 0.1414 max mem: 3532 +Test: [290/517] eta: 0:00:57 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.1736 (0.4711) acc1: 99.4792 (81.4361) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2359 data: 0.1421 max mem: 3532 +Test: [300/517] eta: 0:00:54 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.1689 (0.4609) acc1: 99.4792 (82.0390) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2341 data: 0.1425 max mem: 3532 +Test: [310/517] eta: 0:00:51 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.1693 (0.4520) acc1: 99.7396 (82.6015) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2389 data: 0.1464 max mem: 3532 +Test: [320/517] eta: 0:00:49 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.1959 (0.4444) acc1: 98.9583 (83.0437) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.2458 data: 0.1531 max mem: 3532 +Test: [330/517] eta: 0:00:46 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.2125 (0.4373) acc1: 96.6146 (83.4537) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2451 data: 0.1539 max mem: 3532 +Test: [340/517] eta: 0:00:44 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.1907 (0.4300) acc1: 97.6562 (83.8794) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2446 data: 0.1521 max mem: 3532 +Test: [350/517] eta: 0:00:41 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.1907 (0.4239) acc1: 97.9167 (84.2259) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2413 data: 0.1489 max mem: 3532 +Test: [360/517] eta: 0:00:39 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.2163 (0.4184) acc1: 96.3542 (84.5799) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2343 data: 0.1438 max mem: 3532 +Test: [370/517] eta: 0:00:36 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.2261 (0.4130) acc1: 97.9167 (84.9337) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2344 data: 0.1424 max mem: 3532 +Test: [380/517] eta: 0:00:34 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.2391 (0.4091) acc1: 95.0521 (85.1590) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2349 data: 0.1382 max mem: 3532 +Test: [390/517] eta: 0:00:31 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.2514 (0.4050) acc1: 94.0104 (85.3068) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.2487 data: 0.1511 max mem: 3532 +Test: [400/517] eta: 0:00:29 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.2967 (0.4041) acc1: 85.6771 (85.2264) acc5: 100.0000 (100.0000) acc1_neg: 79.427 acc5_neg: 100.000 time: 0.2499 data: 0.1541 max mem: 3532 +Test: [410/517] eta: 0:00:26 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.3061 (0.4007) acc1: 87.5000 (85.4160) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.2301 data: 0.1353 max mem: 3532 +Test: [420/517] eta: 0:00:24 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.2282 (0.3968) acc1: 95.3125 (85.6827) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.2263 data: 0.1328 max mem: 3532 +Test: [430/517] eta: 0:00:21 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.3423 (0.3968) acc1: 92.1875 (85.7121) acc5: 100.0000 (100.0000) acc1_neg: 71.615 acc5_neg: 100.000 time: 0.2347 data: 0.1420 max mem: 3532 +Test: [440/517] eta: 0:00:19 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.3844 (0.3947) acc1: 90.6250 (85.7397) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2444 data: 0.1511 max mem: 3533 +Test: [450/517] eta: 0:00:16 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.2978 (0.3939) acc1: 91.4062 (85.7481) acc5: 100.0000 (100.0000) acc1_neg: 74.219 acc5_neg: 100.000 time: 0.2399 data: 0.1453 max mem: 3533 +Test: [460/517] eta: 0:00:14 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.3041 (0.3928) acc1: 89.0625 (85.7195) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2308 data: 0.1358 max mem: 3533 +Test: [470/517] eta: 0:00:11 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.1785 (0.3881) acc1: 95.5729 (85.9911) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2263 data: 0.1337 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.1729 (0.3851) acc1: 98.9583 (86.1801) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2343 data: 0.1443 max mem: 3533 +Test: [490/517] eta: 0:00:06 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.2892 (0.3878) acc1: 91.6667 (85.9333) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.2585 data: 0.1652 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.1682 (0.3828) acc1: 98.9583 (86.2140) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2438 data: 0.1468 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.1544 (0.3790) acc1: 99.7396 (86.4532) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2010 data: 0.1041 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.1744 (0.3767) acc1: 99.4792 (86.5946) acc5: 100.0000 (100.0000) acc1_neg: 98.587 acc5_neg: 100.000 time: 0.2257 data: 0.1290 max mem: 3533 +Test: Total time: 0:02:06 (0.2449 s / it) +* Overall Acc@1 86.595 Acc@5 100.000 + target == real Acc@1 12.879 Acc@5 100.000 + target == fake Acc@1 94.242 Acc@5 100.000 + loss 0.377 + PR-AUC 0.9178 ROC-AUC 0.5753 EER 44.39% (thr=0.8418) +Accuracy of the network on the 198427 test images: 86.6% +[rank0]:[W726 05:27:39.315811804 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 05:28:00.989000 3157681 torch/distributed/run.py:766] +W0726 05:28:00.989000 3157681 torch/distributed/run.py:766] ***************************************** +W0726 05:28:00.989000 3157681 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:28:00.989000 3157681 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:28:36.766490265 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank1]:[W726 05:28:36.766490413 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:28:36.089210074 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:28:36.137579579 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 3): env:// +| distributed init (rank 7): env:// +| distributed init (rank 5): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:28:37.534060145 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W726 05:28:37.534112647 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W726 05:28:37.534128786 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W726 05:28:37.534147796 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n5_seed8_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed8_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:36:24 acc1_pos: 13.802 acc5_pos: 100.000 loss: 1.8261 (1.8261) acc1: 13.8021 (13.8021) acc5: 100.0000 (100.0000) time: 5.6145 data: 3.4736 max mem: 3531 +Test: [ 10/389] eta: 0:03:44 acc1_pos: 78.125 acc5_pos: 100.000 loss: 1.7054 (1.5554) acc1: 10.1562 (17.3532) acc5: 100.0000 (100.0000) time: 0.5921 data: 0.3161 max mem: 3531 +Test: [ 20/389] eta: 0:02:12 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.4773 (1.5487) acc1: 10.1562 (23.8715) acc5: 100.0000 (100.0000) time: 0.0972 data: 0.0049 max mem: 3531 +Test: [ 30/389] eta: 0:02:02 acc1_pos: 7.552 acc5_pos: 100.000 loss: 1.8383 (1.6661) acc1: 1.3021 (16.8683) acc5: 100.0000 (100.0000) time: 0.2043 data: 0.1112 max mem: 3531 +Test: [ 40/389] eta: 0:01:48 acc1_pos: 26.562 acc5_pos: 100.000 loss: 1.6977 (1.5937) acc1: 1.8229 (17.6575) acc5: 100.0000 (100.0000) time: 0.2572 data: 0.1659 max mem: 3531 +Test: [ 50/389] eta: 0:01:35 acc1_pos: 10.417 acc5_pos: 100.000 loss: 1.3934 (1.5952) acc1: 10.1562 (15.7016) acc5: 100.0000 (100.0000) time: 0.1878 data: 0.0957 max mem: 3531 +Test: [ 60/389] eta: 0:01:25 acc1_pos: 43.229 acc5_pos: 100.000 loss: 1.4147 (1.6343) acc1: 7.8125 (14.5449) acc5: 100.0000 (100.0000) time: 0.1579 data: 0.0643 max mem: 3531 +Test: [ 70/389] eta: 0:01:20 acc1_pos: 35.938 acc5_pos: 100.000 loss: 1.3772 (1.5611) acc1: 9.3750 (17.0408) acc5: 100.0000 (100.0000) time: 0.1754 data: 0.0802 max mem: 3531 +Test: [ 80/389] eta: 0:01:16 acc1_pos: 80.469 acc5_pos: 100.000 loss: 1.0963 (1.5151) acc1: 28.9062 (19.5956) acc5: 100.0000 (100.0000) time: 0.2096 data: 0.1160 max mem: 3531 +Test: [ 90/389] eta: 0:01:11 acc1_pos: 53.646 acc5_pos: 100.000 loss: 0.7120 (1.4268) acc1: 53.6458 (24.5650) acc5: 100.0000 (100.0000) time: 0.1923 data: 0.1000 max mem: 3531 +Test: [100/389] eta: 0:01:07 acc1_pos: 10.156 acc5_pos: 100.000 loss: 1.1079 (1.4490) acc1: 40.3646 (23.4865) acc5: 100.0000 (100.0000) time: 0.1821 data: 0.0904 max mem: 3531 +Test: [110/389] eta: 0:01:04 acc1_pos: 97.396 acc5_pos: 100.000 loss: 1.5034 (1.4385) acc1: 9.1146 (24.6974) acc5: 100.0000 (100.0000) time: 0.1930 data: 0.1026 max mem: 3531 +Test: [120/389] eta: 0:01:01 acc1_pos: 29.688 acc5_pos: 100.000 loss: 1.3765 (1.4317) acc1: 29.6875 (25.5919) acc5: 100.0000 (100.0000) time: 0.2033 data: 0.1116 max mem: 3531 +Test: [130/389] eta: 0:00:58 acc1_pos: 27.607 acc5_pos: 100.000 loss: 0.6783 (1.3556) acc1: 55.2083 (29.5305) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2063 data: 0.1113 max mem: 3531 +Test: [140/389] eta: 0:00:55 acc1_pos: 27.607 acc5_pos: 100.000 loss: 0.1456 (1.2786) acc1: 99.2188 (33.8043) acc5: 100.0000 (100.0000) acc1_neg: 48.698 acc5_neg: 100.000 time: 0.1893 data: 0.0932 max mem: 3531 +Test: [150/389] eta: 0:00:53 acc1_pos: 27.607 acc5_pos: 100.000 loss: 0.1852 (1.2230) acc1: 98.4375 (36.7946) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1903 data: 0.0962 max mem: 3531 +Test: [160/389] eta: 0:00:50 acc1_pos: 27.607 acc5_pos: 100.000 loss: 0.1953 (1.1590) acc1: 98.4375 (40.3824) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1830 data: 0.0904 max mem: 3531 +Test: [170/389] eta: 0:00:47 acc1_pos: 27.607 acc5_pos: 100.000 loss: 0.1542 (1.1008) acc1: 99.4792 (43.8353) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1776 data: 0.0857 max mem: 3531 +Test: [180/389] eta: 0:00:44 acc1_pos: 27.607 acc5_pos: 100.000 loss: 0.1542 (1.0506) acc1: 100.0000 (46.7470) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.1775 data: 0.0861 max mem: 3531 +Test: [190/389] eta: 0:00:42 acc1_pos: 27.607 acc5_pos: 100.000 loss: 0.1785 (1.0131) acc1: 99.4792 (48.9311) acc5: 100.0000 (100.0000) acc1_neg: 69.010 acc5_neg: 100.000 time: 0.1699 data: 0.0759 max mem: 3531 +Test: [200/389] eta: 0:00:39 acc1_pos: 27.607 acc5_pos: 100.000 loss: 0.1785 (0.9757) acc1: 98.6979 (51.0274) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1748 data: 0.0763 max mem: 3531 +Test: [210/389] eta: 0:00:37 acc1_pos: 27.607 acc5_pos: 100.000 loss: 0.1470 (0.9380) acc1: 99.4792 (53.2978) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1847 data: 0.0874 max mem: 3532 +Test: [220/389] eta: 0:00:35 acc1_pos: 27.607 acc5_pos: 100.000 loss: 0.1246 (0.9009) acc1: 100.0000 (55.4110) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1768 data: 0.0831 max mem: 3532 +Test: [230/389] eta: 0:00:32 acc1_pos: 27.607 acc5_pos: 100.000 loss: 0.1260 (0.8759) acc1: 100.0000 (56.8667) acc5: 100.0000 (100.0000) acc1_neg: 72.917 acc5_neg: 100.000 time: 0.1754 data: 0.0820 max mem: 3532 +Test: [240/389] eta: 0:00:30 acc1_pos: 27.607 acc5_pos: 100.000 loss: 0.3373 (0.8581) acc1: 86.1979 (57.9357) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1925 data: 0.1004 max mem: 3532 +Test: [250/389] eta: 0:00:28 acc1_pos: 27.607 acc5_pos: 100.000 loss: 0.2044 (0.8309) acc1: 95.3125 (59.5649) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1775 data: 0.0865 max mem: 3532 +Test: [260/389] eta: 0:00:26 acc1_pos: 27.607 acc5_pos: 100.000 loss: 0.1214 (0.8094) acc1: 99.7396 (60.6112) acc5: 100.0000 (100.0000) acc1_neg: 57.292 acc5_neg: 100.000 time: 0.1709 data: 0.0785 max mem: 3532 +Test: [270/389] eta: 0:00:24 acc1_pos: 27.607 acc5_pos: 100.000 loss: 0.1380 (0.7953) acc1: 97.9167 (61.3344) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1807 data: 0.0873 max mem: 3532 +Test: [280/389] eta: 0:00:21 acc1_pos: 27.607 acc5_pos: 100.000 loss: 0.1662 (0.7739) acc1: 98.4375 (62.6149) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1742 data: 0.0814 max mem: 3532 +Test: [290/389] eta: 0:00:19 acc1_pos: 27.607 acc5_pos: 100.000 loss: 0.2085 (0.7578) acc1: 98.4375 (63.5819) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1668 data: 0.0755 max mem: 3532 +Test: [300/389] eta: 0:00:17 acc1_pos: 27.607 acc5_pos: 100.000 loss: 0.1488 (0.7372) acc1: 100.0000 (64.7910) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1689 data: 0.0777 max mem: 3532 +Test: [310/389] eta: 0:00:15 acc1_pos: 27.607 acc5_pos: 100.000 loss: 0.1474 (0.7198) acc1: 100.0000 (65.8276) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1853 data: 0.0936 max mem: 3532 +Test: [320/389] eta: 0:00:13 acc1_pos: 27.607 acc5_pos: 100.000 loss: 0.1493 (0.7027) acc1: 98.9583 (66.8330) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2045 data: 0.1139 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 27.607 acc5_pos: 100.000 loss: 0.1187 (0.6850) acc1: 100.0000 (67.8350) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2177 data: 0.1278 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 27.607 acc5_pos: 100.000 loss: 0.1163 (0.6686) acc1: 100.0000 (68.7660) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1848 data: 0.0919 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 27.607 acc5_pos: 100.000 loss: 0.1322 (0.6708) acc1: 99.4792 (68.4577) acc5: 100.0000 (100.0000) acc1_neg: 14.583 acc5_neg: 100.000 time: 0.1523 data: 0.0590 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 27.607 acc5_pos: 100.000 loss: 0.8386 (0.6785) acc1: 47.6562 (67.6824) acc5: 100.0000 (100.0000) acc1_neg: 49.219 acc5_neg: 100.000 time: 0.1393 data: 0.0488 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 27.607 acc5_pos: 100.000 loss: 0.6776 (0.6673) acc1: 50.5208 (68.3387) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1281 data: 0.0381 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 27.607 acc5_pos: 100.000 loss: 0.1228 (0.6537) acc1: 100.0000 (69.1054) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1239 data: 0.0343 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 27.607 acc5_pos: 100.000 loss: 0.1239 (0.6450) acc1: 100.0000 (69.6086) acc5: 100.0000 (100.0000) acc1_neg: 71.585 acc5_neg: 100.000 time: 0.1164 data: 0.0235 max mem: 3532 +Test: Total time: 0:01:14 (0.1906 s / it) +* Overall Acc@1 69.609 Acc@5 100.000 + target == real Acc@1 25.933 Acc@5 100.000 + target == fake Acc@1 89.910 Acc@5 100.000 + loss 0.645 + PR-AUC 0.7616 ROC-AUC 0.6521 EER 38.18% (thr=0.8462) +Accuracy of the network on the 149358 test images: 69.6% +[rank0]:[W726 05:30:00.084469734 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:30:22.035000 3158243 torch/distributed/run.py:766] +W0726 05:30:22.035000 3158243 torch/distributed/run.py:766] ***************************************** +W0726 05:30:22.035000 3158243 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:30:22.035000 3158243 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:30:55.259790928 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:30:55.495068608 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W726 05:30:55.495118855 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:30:55.517091843 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:30:55.585817830 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +| distributed init (rank 7): env:// +[rank0]:[W726 05:30:55.636611789 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:30:55.647779344 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W726 05:30:55.647813358 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n5_seed8_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed8_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:27:45 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.5991 (2.5991) acc1: 0.0000 (0.0000) acc5: 100.0000 (100.0000) time: 5.0013 data: 3.1063 max mem: 3531 +Test: [ 10/333] eta: 0:03:00 acc1_pos: 1.042 acc5_pos: 100.000 loss: 2.5452 (2.4654) acc1: 0.2604 (1.3494) acc5: 100.0000 (100.0000) time: 0.5578 data: 0.3042 max mem: 3531 +Test: [ 20/333] eta: 0:01:55 acc1_pos: 0.781 acc5_pos: 100.000 loss: 2.4038 (2.4344) acc1: 0.7812 (1.6121) acc5: 100.0000 (100.0000) time: 0.1376 data: 0.0447 max mem: 3531 +Test: [ 30/333] eta: 0:01:37 acc1_pos: 17.969 acc5_pos: 100.000 loss: 1.9451 (2.1985) acc1: 3.9062 (5.7880) acc5: 100.0000 (100.0000) time: 0.1919 data: 0.0966 max mem: 3531 +Test: [ 40/333] eta: 0:01:29 acc1_pos: 31.250 acc5_pos: 100.000 loss: 1.6728 (2.0617) acc1: 11.4583 (8.1745) acc5: 100.0000 (100.0000) time: 0.2366 data: 0.1437 max mem: 3531 +Test: [ 50/333] eta: 0:01:20 acc1_pos: 13.021 acc5_pos: 100.000 loss: 1.6940 (1.9886) acc1: 13.0208 (9.1963) acc5: 100.0000 (100.0000) time: 0.2288 data: 0.1376 max mem: 3531 +Test: [ 60/333] eta: 0:01:14 acc1_pos: 8.594 acc5_pos: 100.000 loss: 1.7114 (1.9434) acc1: 13.0208 (9.9428) acc5: 100.0000 (100.0000) time: 0.2107 data: 0.1165 max mem: 3531 +Test: [ 70/333] eta: 0:01:11 acc1_pos: 15.104 acc5_pos: 100.000 loss: 1.7091 (1.9023) acc1: 11.1979 (10.4350) acc5: 100.0000 (100.0000) time: 0.2349 data: 0.1386 max mem: 3531 +Test: [ 80/333] eta: 0:01:03 acc1_pos: 13.802 acc5_pos: 100.000 loss: 1.6641 (1.8748) acc1: 13.8021 (10.9697) acc5: 100.0000 (100.0000) time: 0.1874 data: 0.0914 max mem: 3531 +Test: [ 90/333] eta: 0:00:59 acc1_pos: 20.312 acc5_pos: 100.000 loss: 1.5890 (1.8417) acc1: 13.8021 (11.2924) acc5: 100.0000 (100.0000) time: 0.1449 data: 0.0505 max mem: 3531 +Test: [100/333] eta: 0:00:56 acc1_pos: 8.333 acc5_pos: 100.000 loss: 1.5180 (1.8130) acc1: 13.2812 (11.7523) acc5: 100.0000 (100.0000) time: 0.1949 data: 0.1033 max mem: 3531 +Test: [110/333] eta: 0:00:52 acc1_pos: 13.542 acc5_pos: 100.000 loss: 1.5501 (1.7962) acc1: 13.2812 (11.9627) acc5: 100.0000 (100.0000) time: 0.2062 data: 0.1148 max mem: 3531 +Test: [120/333] eta: 0:00:49 acc1_pos: 15.885 acc5_pos: 100.000 loss: 1.6064 (1.7785) acc1: 13.5417 (12.2095) acc5: 100.0000 (100.0000) time: 0.2000 data: 0.1084 max mem: 3531 +Test: [130/333] eta: 0:00:46 acc1_pos: 17.188 acc5_pos: 100.000 loss: 1.5956 (1.7600) acc1: 14.3229 (12.6312) acc5: 100.0000 (100.0000) time: 0.1815 data: 0.0896 max mem: 3531 +Test: [140/333] eta: 0:00:43 acc1_pos: 15.365 acc5_pos: 100.000 loss: 1.4817 (1.7365) acc1: 15.8854 (13.2831) acc5: 100.0000 (100.0000) time: 0.1699 data: 0.0769 max mem: 3531 +Test: [150/333] eta: 0:00:40 acc1_pos: 13.021 acc5_pos: 100.000 loss: 1.4948 (1.7231) acc1: 17.1875 (13.6055) acc5: 100.0000 (100.0000) time: 0.1871 data: 0.0928 max mem: 3531 +Test: [160/333] eta: 0:00:38 acc1_pos: 14.145 acc5_pos: 100.000 loss: 1.4771 (1.6779) acc1: 20.3125 (15.7123) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.1815 data: 0.0864 max mem: 3531 +Test: [170/333] eta: 0:00:35 acc1_pos: 14.145 acc5_pos: 100.000 loss: 0.3152 (1.5972) acc1: 87.2396 (20.0064) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1904 data: 0.0972 max mem: 3531 +Test: [180/333] eta: 0:00:33 acc1_pos: 14.145 acc5_pos: 100.000 loss: 0.2884 (1.5244) acc1: 89.8438 (23.9540) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1948 data: 0.1031 max mem: 3531 +Test: [190/333] eta: 0:00:30 acc1_pos: 14.145 acc5_pos: 100.000 loss: 0.2757 (1.4576) acc1: 91.6667 (27.5946) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1796 data: 0.0885 max mem: 3531 +Test: [200/333] eta: 0:00:28 acc1_pos: 14.145 acc5_pos: 100.000 loss: 0.2646 (1.3990) acc1: 91.1458 (30.7395) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1835 data: 0.0930 max mem: 3531 +Test: [210/333] eta: 0:00:26 acc1_pos: 14.145 acc5_pos: 100.000 loss: 0.2617 (1.3450) acc1: 91.1458 (33.6283) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1878 data: 0.0949 max mem: 3532 +Test: [220/333] eta: 0:00:23 acc1_pos: 14.145 acc5_pos: 100.000 loss: 0.2617 (1.2968) acc1: 91.4062 (36.2191) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1849 data: 0.0909 max mem: 3532 +Test: [230/333] eta: 0:00:21 acc1_pos: 14.145 acc5_pos: 100.000 loss: 0.2695 (1.2519) acc1: 92.9688 (38.6702) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1819 data: 0.0844 max mem: 3532 +Test: [240/333] eta: 0:00:19 acc1_pos: 14.145 acc5_pos: 100.000 loss: 0.3041 (1.2136) acc1: 88.5417 (40.6823) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.1857 data: 0.0896 max mem: 3532 +Test: [250/333] eta: 0:00:17 acc1_pos: 14.145 acc5_pos: 100.000 loss: 0.3190 (1.1772) acc1: 87.5000 (42.5963) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1941 data: 0.1033 max mem: 3532 +Test: [260/333] eta: 0:00:15 acc1_pos: 14.145 acc5_pos: 100.000 loss: 0.2630 (1.1419) acc1: 90.6250 (44.4993) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1979 data: 0.1051 max mem: 3532 +Test: [270/333] eta: 0:00:13 acc1_pos: 14.145 acc5_pos: 100.000 loss: 0.2479 (1.1094) acc1: 92.1875 (46.2110) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1900 data: 0.0965 max mem: 3532 +Test: [280/333] eta: 0:00:10 acc1_pos: 14.145 acc5_pos: 100.000 loss: 0.2727 (1.0806) acc1: 89.3229 (47.7322) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.1701 data: 0.0790 max mem: 3532 +Test: [290/333] eta: 0:00:08 acc1_pos: 14.145 acc5_pos: 100.000 loss: 0.3210 (1.0549) acc1: 87.7604 (49.0926) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.1554 data: 0.0652 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 14.145 acc5_pos: 100.000 loss: 0.3197 (1.0301) acc1: 88.0208 (50.4335) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1489 data: 0.0588 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 14.145 acc5_pos: 100.000 loss: 0.2973 (1.0063) acc1: 89.5833 (51.6965) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1447 data: 0.0549 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 14.145 acc5_pos: 100.000 loss: 0.3291 (0.9855) acc1: 88.0208 (52.7899) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.1366 data: 0.0468 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 14.145 acc5_pos: 100.000 loss: 0.2868 (0.9640) acc1: 88.8021 (53.9409) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1307 data: 0.0412 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 14.145 acc5_pos: 100.000 loss: 0.2818 (0.9598) acc1: 90.3646 (54.1586) acc5: 100.0000 (100.0000) acc1_neg: 90.423 acc5_neg: 100.000 time: 0.1329 data: 0.0399 max mem: 3532 +Test: Total time: 0:01:05 (0.1958 s / it) +* Overall Acc@1 54.159 Acc@5 100.000 + target == real Acc@1 13.819 Acc@5 100.000 + target == fake Acc@1 90.069 Acc@5 100.000 + loss 0.960 + PR-AUC 0.5355 ROC-AUC 0.5294 EER 47.46% (thr=0.8447) +Accuracy of the network on the 127843 test images: 54.2% +[rank0]:[W726 05:32:10.068866864 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:32:32.211000 3158803 torch/distributed/run.py:766] +W0726 05:32:32.211000 3158803 torch/distributed/run.py:766] ***************************************** +W0726 05:32:32.211000 3158803 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:32:32.211000 3158803 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 05:33:06.510680656 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +NCCL version 2.26.2+cuda12.2 +[rank1]:[W726 05:33:06.645890502 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:33:06.853923853 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:33:06.895065816 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:33:06.068013604 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:33:07.182519577 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +[rank4]:[W726 05:33:07.311112361 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:33:07.316127411 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n5_seed8_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed8_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:26:11 acc1_pos: 10.156 acc5_pos: 100.000 loss: 2.0302 (2.0302) acc1: 10.1562 (10.1562) acc5: 100.0000 (100.0000) time: 6.6893 data: 5.0707 max mem: 3531 +Test: [ 10/235] eta: 0:02:35 acc1_pos: 8.073 acc5_pos: 100.000 loss: 2.0797 (2.0711) acc1: 6.5104 (7.1259) acc5: 100.0000 (100.0000) time: 0.6911 data: 0.4614 max mem: 3531 +Test: [ 20/235] eta: 0:01:28 acc1_pos: 4.427 acc5_pos: 100.000 loss: 2.0797 (2.0737) acc1: 6.2500 (6.3368) acc5: 100.0000 (100.0000) time: 0.0956 data: 0.0017 max mem: 3531 +Test: [ 30/235] eta: 0:01:29 acc1_pos: 8.333 acc5_pos: 100.000 loss: 2.0730 (2.0749) acc1: 5.9896 (6.4768) acc5: 100.0000 (100.0000) time: 0.2949 data: 0.2001 max mem: 3531 +Test: [ 40/235] eta: 0:01:24 acc1_pos: 7.031 acc5_pos: 100.000 loss: 2.0628 (2.0714) acc1: 7.0312 (6.5612) acc5: 100.0000 (100.0000) time: 0.4541 data: 0.3615 max mem: 3531 +Test: [ 50/235] eta: 0:01:13 acc1_pos: 7.031 acc5_pos: 100.000 loss: 2.0579 (2.0696) acc1: 6.5104 (6.4645) acc5: 100.0000 (100.0000) time: 0.3326 data: 0.2413 max mem: 3531 +Test: [ 60/235] eta: 0:01:05 acc1_pos: 6.510 acc5_pos: 100.000 loss: 2.0741 (2.0697) acc1: 6.2500 (6.4891) acc5: 100.0000 (100.0000) time: 0.2490 data: 0.1577 max mem: 3531 +Test: [ 70/235] eta: 0:00:57 acc1_pos: 7.692 acc5_pos: 100.000 loss: 2.0161 (1.8301) acc1: 7.5521 (17.8257) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2262 data: 0.1345 max mem: 3531 +Test: [ 80/235] eta: 0:00:50 acc1_pos: 7.692 acc5_pos: 100.000 loss: 0.1371 (1.6210) acc1: 97.6562 (27.6878) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1969 data: 0.1064 max mem: 3531 +Test: [ 90/235] eta: 0:00:45 acc1_pos: 7.692 acc5_pos: 100.000 loss: 0.1364 (1.4584) acc1: 97.6562 (35.3451) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1885 data: 0.0949 max mem: 3531 +Test: [100/235] eta: 0:00:41 acc1_pos: 7.692 acc5_pos: 100.000 loss: 0.1364 (1.3272) acc1: 97.9167 (41.5610) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2216 data: 0.1273 max mem: 3531 +Test: [110/235] eta: 0:00:37 acc1_pos: 7.692 acc5_pos: 100.000 loss: 0.1368 (1.2203) acc1: 97.9167 (46.5700) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2473 data: 0.1565 max mem: 3531 +Test: [120/235] eta: 0:00:33 acc1_pos: 7.692 acc5_pos: 100.000 loss: 0.1498 (1.1339) acc1: 96.8750 (50.7231) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1932 data: 0.0996 max mem: 3531 +Test: [130/235] eta: 0:00:29 acc1_pos: 7.692 acc5_pos: 100.000 loss: 0.1837 (1.0618) acc1: 96.6146 (54.1726) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1608 data: 0.0636 max mem: 3531 +Test: [140/235] eta: 0:00:26 acc1_pos: 7.692 acc5_pos: 100.000 loss: 0.1923 (1.0001) acc1: 96.0938 (57.1328) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1947 data: 0.0932 max mem: 3531 +Test: [150/235] eta: 0:00:23 acc1_pos: 7.692 acc5_pos: 100.000 loss: 0.1923 (0.9464) acc1: 96.0938 (59.6941) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2260 data: 0.1284 max mem: 3531 +Test: [160/235] eta: 0:00:19 acc1_pos: 7.692 acc5_pos: 100.000 loss: 0.1869 (0.8995) acc1: 96.0938 (61.9533) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1982 data: 0.1079 max mem: 3531 +Test: [170/235] eta: 0:00:17 acc1_pos: 7.692 acc5_pos: 100.000 loss: 0.2111 (0.8606) acc1: 94.5312 (63.6757) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1887 data: 0.0977 max mem: 3531 +Test: [180/235] eta: 0:00:14 acc1_pos: 7.692 acc5_pos: 100.000 loss: 0.2366 (0.8267) acc1: 90.3646 (65.1675) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.2238 data: 0.1318 max mem: 3531 +Test: [190/235] eta: 0:00:11 acc1_pos: 7.692 acc5_pos: 100.000 loss: 0.2453 (0.7960) acc1: 90.6250 (66.5290) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.2211 data: 0.1298 max mem: 3531 +Test: [200/235] eta: 0:00:08 acc1_pos: 7.692 acc5_pos: 100.000 loss: 0.2480 (0.7689) acc1: 90.3646 (67.7006) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1865 data: 0.0963 max mem: 3531 +Test: [210/235] eta: 0:00:06 acc1_pos: 7.692 acc5_pos: 100.000 loss: 0.2501 (0.7439) acc1: 89.8438 (68.7944) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1474 data: 0.0570 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 7.692 acc5_pos: 100.000 loss: 0.2322 (0.7207) acc1: 90.8854 (69.8294) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1312 data: 0.0415 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 7.692 acc5_pos: 100.000 loss: 0.2305 (0.6996) acc1: 91.4062 (70.7589) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1260 data: 0.0366 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 7.692 acc5_pos: 100.000 loss: 0.2326 (0.6918) acc1: 91.3044 (71.0288) acc5: 100.0000 (100.0000) acc1_neg: 91.304 acc5_neg: 100.000 time: 0.1205 data: 0.0296 max mem: 3532 +Test: Total time: 0:00:55 (0.2357 s / it) +* Overall Acc@1 71.029 Acc@5 100.000 + target == real Acc@1 6.505 Acc@5 100.000 + target == fake Acc@1 94.366 Acc@5 100.000 + loss 0.692 + PR-AUC 0.7647 ROC-AUC 0.5841 EER 41.92% (thr=0.9028) +Accuracy of the network on the 89879 test images: 71.0% +[rank0]:[W726 05:34:10.061981890 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:34:32.945000 3159363 torch/distributed/run.py:766] +W0726 05:34:32.945000 3159363 torch/distributed/run.py:766] ***************************************** +W0726 05:34:32.945000 3159363 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:34:32.945000 3159363 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:35:05.563643941 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:35:05.881578473 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:35:06.368625970 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:35:06.414628234 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:35:06.465247128 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +[rank0]:[W726 05:35:06.495420615 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:35:06.502715918 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:35:06.506627160 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n5_seed8_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed8_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:20:56 acc1_pos: 4.167 acc5_pos: 100.000 loss: 2.0830 (2.0830) acc1: 4.1667 (4.1667) acc5: 100.0000 (100.0000) time: 4.6725 data: 2.6543 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/269] eta: 0:02:11 acc1_pos: 16.927 acc5_pos: 100.000 loss: 1.6419 (1.6964) acc1: 9.1146 (10.1326) acc5: 100.0000 (100.0000) time: 0.5095 data: 0.2447 max mem: 3531 +Test: [ 20/269] eta: 0:01:21 acc1_pos: 8.594 acc5_pos: 100.000 loss: 1.6047 (1.6440) acc1: 8.5938 (10.3919) acc5: 100.0000 (100.0000) time: 0.1082 data: 0.0178 max mem: 3531 +Test: [ 30/269] eta: 0:01:15 acc1_pos: 2.083 acc5_pos: 100.000 loss: 1.5906 (1.6263) acc1: 8.0729 (12.9452) acc5: 100.0000 (100.0000) time: 0.2072 data: 0.1138 max mem: 3531 +Test: [ 40/269] eta: 0:01:08 acc1_pos: 11.207 acc5_pos: 100.000 loss: 1.0424 (1.3280) acc1: 36.9792 (30.3989) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.2706 data: 0.1781 max mem: 3531 +Test: [ 50/269] eta: 0:01:01 acc1_pos: 11.207 acc5_pos: 100.000 loss: 0.2481 (1.1200) acc1: 96.0938 (43.0862) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2332 data: 0.1417 max mem: 3531 +Test: [ 60/269] eta: 0:00:52 acc1_pos: 11.207 acc5_pos: 100.000 loss: 0.2716 (0.9833) acc1: 94.0104 (51.1868) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1564 data: 0.0648 max mem: 3531 +Test: [ 70/269] eta: 0:00:46 acc1_pos: 11.207 acc5_pos: 100.000 loss: 0.2283 (0.8754) acc1: 93.4896 (57.3687) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1134 data: 0.0232 max mem: 3531 +Test: [ 80/269] eta: 0:00:42 acc1_pos: 11.207 acc5_pos: 100.000 loss: 0.2132 (0.7934) acc1: 96.0938 (62.1689) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1454 data: 0.0530 max mem: 3531 +Test: [ 90/269] eta: 0:00:38 acc1_pos: 11.207 acc5_pos: 100.000 loss: 0.2834 (0.7461) acc1: 93.2292 (64.8495) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.1553 data: 0.0629 max mem: 3531 +Test: [100/269] eta: 0:00:35 acc1_pos: 11.207 acc5_pos: 100.000 loss: 0.2990 (0.6994) acc1: 90.8854 (67.5588) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1520 data: 0.0593 max mem: 3531 +Test: [110/269] eta: 0:00:33 acc1_pos: 11.207 acc5_pos: 100.000 loss: 0.2192 (0.6540) acc1: 97.3958 (70.3266) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1670 data: 0.0728 max mem: 3531 +Test: [120/269] eta: 0:00:30 acc1_pos: 11.207 acc5_pos: 100.000 loss: 0.2035 (0.6168) acc1: 98.1771 (72.6003) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1808 data: 0.0883 max mem: 3531 +Test: [130/269] eta: 0:00:28 acc1_pos: 11.207 acc5_pos: 100.000 loss: 0.2042 (0.5901) acc1: 97.6562 (74.1869) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1871 data: 0.0961 max mem: 3531 +Test: [140/269] eta: 0:00:26 acc1_pos: 11.207 acc5_pos: 100.000 loss: 0.2967 (0.5718) acc1: 92.7083 (75.1792) acc5: 100.0000 (100.0000) acc1_neg: 59.635 acc5_neg: 100.000 time: 0.1909 data: 0.0979 max mem: 3531 +Test: [150/269] eta: 0:00:24 acc1_pos: 11.207 acc5_pos: 100.000 loss: 0.3290 (0.5577) acc1: 91.1458 (75.9054) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1897 data: 0.0959 max mem: 3531 +Test: [160/269] eta: 0:00:21 acc1_pos: 11.207 acc5_pos: 100.000 loss: 0.2817 (0.5394) acc1: 94.5312 (77.0478) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1851 data: 0.0934 max mem: 3531 +Test: [170/269] eta: 0:00:19 acc1_pos: 11.207 acc5_pos: 100.000 loss: 0.2481 (0.5233) acc1: 96.3542 (78.1539) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1879 data: 0.0956 max mem: 3531 +Test: [180/269] eta: 0:00:17 acc1_pos: 11.207 acc5_pos: 100.000 loss: 0.2417 (0.5060) acc1: 96.6146 (79.1739) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1835 data: 0.0910 max mem: 3531 +Test: [190/269] eta: 0:00:15 acc1_pos: 11.207 acc5_pos: 100.000 loss: 0.2415 (0.4932) acc1: 95.5729 (79.9766) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1749 data: 0.0831 max mem: 3531 +Test: [200/269] eta: 0:00:13 acc1_pos: 11.207 acc5_pos: 100.000 loss: 0.2415 (0.4813) acc1: 95.3125 (80.7058) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1859 data: 0.0945 max mem: 3531 +Test: [210/269] eta: 0:00:11 acc1_pos: 11.207 acc5_pos: 100.000 loss: 0.2687 (0.4721) acc1: 94.5312 (81.2426) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1854 data: 0.0928 max mem: 3532 +Test: [220/269] eta: 0:00:09 acc1_pos: 11.207 acc5_pos: 100.000 loss: 0.2937 (0.4638) acc1: 91.9271 (81.6094) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1832 data: 0.0894 max mem: 3532 +Test: [230/269] eta: 0:00:07 acc1_pos: 11.207 acc5_pos: 100.000 loss: 0.1720 (0.4503) acc1: 98.1771 (82.3537) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1811 data: 0.0880 max mem: 3532 +Test: [240/269] eta: 0:00:05 acc1_pos: 11.207 acc5_pos: 100.000 loss: 0.2063 (0.4496) acc1: 96.6146 (82.2517) acc5: 100.0000 (100.0000) acc1_neg: 56.510 acc5_neg: 100.000 time: 0.1775 data: 0.0858 max mem: 3532 +Test: [250/269] eta: 0:00:03 acc1_pos: 11.207 acc5_pos: 100.000 loss: 0.3688 (0.4481) acc1: 83.3333 (82.2107) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.1759 data: 0.0840 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 11.207 acc5_pos: 100.000 loss: 0.3746 (0.4487) acc1: 83.3333 (81.9933) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1681 data: 0.0756 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 11.207 acc5_pos: 100.000 loss: 0.2560 (0.4430) acc1: 90.6250 (82.3676) acc5: 100.0000 (100.0000) acc1_neg: 96.588 acc5_neg: 100.000 time: 0.1651 data: 0.0700 max mem: 3532 +Test: Total time: 0:00:51 (0.1931 s / it) +* Overall Acc@1 82.368 Acc@5 100.000 + target == real Acc@1 12.698 Acc@5 100.000 + target == fake Acc@1 91.876 Acc@5 100.000 + loss 0.443 + PR-AUC 0.8840 ROC-AUC 0.5469 EER 46.05% (thr=0.8213) +Accuracy of the network on the 103293 test images: 82.4% +[rank0]:[W726 05:36:11.462182242 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T05:36:17Z] eval_done tag=fixed500k_real_n5_seed8_resampled +[2026-07-26T05:36:18Z] eval_start tag=fixed500k_fake_n11_seed8_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed8_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T05:36:18Z +hostname=n116-040-168 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed8_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n11_seed8_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 05:36:33.503000 3159930 torch/distributed/run.py:766] +W0726 05:36:33.503000 3159930 torch/distributed/run.py:766] ***************************************** +W0726 05:36:33.503000 3159930 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:36:33.503000 3159930 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:37:07.533821381 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W726 05:37:07.533825932 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:37:07.681747418 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:37:07.784297075 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:37:07.952666015 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:37:07.980347596 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:37:07.092451572 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:37:07.123608544 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n11_seed8_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed8_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:22:12 acc1_pos: 37.500 acc5_pos: 100.000 loss: 0.8600 (0.8600) acc1: 37.5000 (37.5000) acc5: 100.0000 (100.0000) time: 5.6932 data: 4.2014 max mem: 3531 +Test: [ 10/234] eta: 0:02:14 acc1_pos: 12.500 acc5_pos: 100.000 loss: 1.0874 (1.1030) acc1: 25.7812 (25.7813) acc5: 100.0000 (100.0000) time: 0.5993 data: 0.3823 max mem: 3531 +Test: [ 20/234] eta: 0:01:19 acc1_pos: 88.281 acc5_pos: 100.000 loss: 1.0874 (1.0557) acc1: 25.7812 (32.8621) acc5: 100.0000 (100.0000) time: 0.1071 data: 0.0140 max mem: 3531 +Test: [ 30/234] eta: 0:01:11 acc1_pos: 29.427 acc5_pos: 100.000 loss: 0.5096 (0.8874) acc1: 76.8229 (45.9761) acc5: 100.0000 (100.0000) time: 0.2171 data: 0.1207 max mem: 3531 +Test: [ 40/234] eta: 0:01:05 acc1_pos: 97.396 acc5_pos: 100.000 loss: 0.5028 (0.8770) acc1: 77.3438 (47.7134) acc5: 100.0000 (100.0000) time: 0.2974 data: 0.2026 max mem: 3531 +Test: [ 50/234] eta: 0:00:56 acc1_pos: 97.396 acc5_pos: 100.000 loss: 0.2002 (0.7421) acc1: 96.8750 (57.4602) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1382 max mem: 3531 +Test: [ 60/234] eta: 0:00:49 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.2064 (0.6593) acc1: 97.3958 (64.0283) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1841 data: 0.0933 max mem: 3531 +Test: [ 70/234] eta: 0:00:44 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.2483 (0.6671) acc1: 97.1354 (63.2372) acc5: 100.0000 (100.0000) acc1_neg: 35.938 acc5_neg: 100.000 time: 0.1966 data: 0.1057 max mem: 3531 +Test: [ 80/234] eta: 0:00:41 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.8872 (0.6640) acc1: 40.8854 (63.0176) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2052 data: 0.1152 max mem: 3531 +Test: [ 90/234] eta: 0:00:37 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.2395 (0.6196) acc1: 92.9688 (66.4635) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.1988 data: 0.1089 max mem: 3531 +Test: [100/234] eta: 0:00:33 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.2399 (0.5996) acc1: 92.9688 (67.9455) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1930 data: 0.1026 max mem: 3531 +Test: [110/234] eta: 0:00:30 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4883 (0.5887) acc1: 76.5625 (68.8861) acc5: 100.0000 (100.0000) acc1_neg: 72.656 acc5_neg: 100.000 time: 0.1963 data: 0.1057 max mem: 3531 +Test: [120/234] eta: 0:00:27 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4171 (0.5768) acc1: 76.5625 (69.8562) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1749 data: 0.0825 max mem: 3531 +Test: [130/234] eta: 0:00:24 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.3130 (0.5597) acc1: 91.9271 (71.2289) acc5: 100.0000 (100.0000) acc1_neg: 66.146 acc5_neg: 100.000 time: 0.1679 data: 0.0750 max mem: 3531 +Test: [140/234] eta: 0:00:21 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.5240 (0.5877) acc1: 73.9583 (69.5405) acc5: 100.0000 (100.0000) acc1_neg: 18.490 acc5_neg: 100.000 time: 0.1910 data: 0.0989 max mem: 3531 +Test: [150/234] eta: 0:00:19 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.6074 (0.5912) acc1: 65.6250 (69.4795) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.1769 data: 0.0848 max mem: 3531 +Test: [160/234] eta: 0:00:16 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4605 (0.5778) acc1: 80.4688 (70.4807) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.1493 data: 0.0572 max mem: 3531 +Test: [170/234] eta: 0:00:14 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4597 (0.5707) acc1: 81.2500 (71.1105) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1805 data: 0.0878 max mem: 3531 +Test: [180/234] eta: 0:00:11 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4614 (0.5811) acc1: 77.3438 (70.4406) acc5: 100.0000 (100.0000) acc1_neg: 15.625 acc5_neg: 100.000 time: 0.1964 data: 0.1039 max mem: 3531 +Test: [190/234] eta: 0:00:09 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.7108 (0.5899) acc1: 55.2083 (69.5353) acc5: 100.0000 (100.0000) acc1_neg: 51.042 acc5_neg: 100.000 time: 0.1723 data: 0.0796 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.7038 (0.5771) acc1: 55.7292 (70.2153) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1737 data: 0.0793 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.1115 (0.5584) acc1: 99.7396 (71.4887) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1885 data: 0.0953 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.3144 (0.5479) acc1: 92.4479 (72.4053) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1611 data: 0.0706 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.3941 (0.5502) acc1: 70.8333 (72.1039) acc5: 100.0000 (100.0000) acc1_neg: 64.583 acc5_neg: 100.000 time: 0.1340 data: 0.0436 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.5786 (0.5506) acc1: 67.4479 (72.0536) acc5: 100.0000 (100.0000) acc1_neg: 70.428 acc5_neg: 100.000 time: 0.1348 data: 0.0435 max mem: 3532 +Test: Total time: 0:00:48 (0.2060 s / it) +* Overall Acc@1 72.054 Acc@5 100.000 + target == real Acc@1 57.758 Acc@5 100.000 + target == fake Acc@1 76.083 Acc@5 100.000 + loss 0.551 + PR-AUC 0.9018 ROC-AUC 0.7430 EER 33.28% (thr=0.6055) +Accuracy of the network on the 89729 test images: 72.1% +[rank0]:[W726 05:38:07.506864483 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:38:26.339000 3160488 torch/distributed/run.py:766] +W0726 05:38:26.339000 3160488 torch/distributed/run.py:766] ***************************************** +W0726 05:38:26.339000 3160488 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:38:26.339000 3160488 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:38:54.768739574 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:38:54.832308798 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:38:54.069525377 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:38:54.160489570 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:38:55.579482196 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +| distributed init (rank 1): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:38:55.598151366 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank4]:[W726 05:38:55.600660627 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:38:55.604705055 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n11_seed8_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed8_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 3:58:07 acc1_pos: 5.469 acc5_pos: 100.000 loss: 1.5227 (1.5227) acc1: 5.4688 (5.4688) acc5: 100.0000 (100.0000) time: 5.9857 data: 3.3895 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/2387] eta: 0:24:47 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.4520 (1.4896) acc1: 5.4688 (5.7528) acc5: 100.0000 (100.0000) time: 0.6260 data: 0.3085 max mem: 3531 +Test: [ 20/2387] eta: 0:15:15 acc1_pos: 0.260 acc5_pos: 100.000 loss: 1.5670 (1.5778) acc1: 2.0833 (3.9063) acc5: 100.0000 (100.0000) time: 0.1068 data: 0.0148 max mem: 3531 +Test: [ 30/2387] eta: 0:11:40 acc1_pos: 3.646 acc5_pos: 100.000 loss: 1.5832 (1.5576) acc1: 2.0833 (4.0575) acc5: 100.0000 (100.0000) time: 0.1166 data: 0.0209 max mem: 3531 +Test: [ 40/2387] eta: 0:11:16 acc1_pos: 2.865 acc5_pos: 100.000 loss: 1.5139 (1.5536) acc1: 2.6042 (3.7729) acc5: 100.0000 (100.0000) time: 0.1848 data: 0.0881 max mem: 3531 +Test: [ 50/2387] eta: 0:11:09 acc1_pos: 12.760 acc5_pos: 100.000 loss: 1.4848 (1.5471) acc1: 2.8646 (4.6569) acc5: 100.0000 (100.0000) time: 0.2699 data: 0.1733 max mem: 3531 +Test: [ 60/2387] eta: 0:10:41 acc1_pos: 25.781 acc5_pos: 100.000 loss: 1.4848 (1.5296) acc1: 7.5521 (5.7292) acc5: 100.0000 (100.0000) time: 0.2499 data: 0.1517 max mem: 3531 +Test: [ 70/2387] eta: 0:10:14 acc1_pos: 5.208 acc5_pos: 100.000 loss: 1.4589 (1.5294) acc1: 6.5104 (5.7365) acc5: 100.0000 (100.0000) time: 0.2099 data: 0.1127 max mem: 3531 +Test: [ 80/2387] eta: 0:09:55 acc1_pos: 58.073 acc5_pos: 100.000 loss: 1.3852 (1.4727) acc1: 8.8542 (8.3751) acc5: 100.0000 (100.0000) time: 0.2049 data: 0.1105 max mem: 3531 +Test: [ 90/2387] eta: 0:09:43 acc1_pos: 39.062 acc5_pos: 100.000 loss: 0.9808 (1.4074) acc1: 35.9375 (11.9906) acc5: 100.0000 (100.0000) time: 0.2149 data: 0.1207 max mem: 3531 +Test: [ 100/2387] eta: 0:09:35 acc1_pos: 40.104 acc5_pos: 100.000 loss: 0.8881 (1.3580) acc1: 39.0625 (14.7664) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1302 max mem: 3531 +Test: [ 110/2387] eta: 0:09:26 acc1_pos: 44.531 acc5_pos: 100.000 loss: 0.8732 (1.3115) acc1: 40.1042 (17.4761) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1309 max mem: 3531 +Test: [ 120/2387] eta: 0:09:18 acc1_pos: 26.302 acc5_pos: 100.000 loss: 0.8937 (1.2845) acc1: 38.2812 (18.8748) acc5: 100.0000 (100.0000) time: 0.2199 data: 0.1262 max mem: 3531 +Test: [ 130/2387] eta: 0:09:14 acc1_pos: 30.729 acc5_pos: 100.000 loss: 0.8949 (1.2539) acc1: 34.8958 (20.4775) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1363 max mem: 3531 +Test: [ 140/2387] eta: 0:09:08 acc1_pos: 31.510 acc5_pos: 100.000 loss: 0.8927 (1.2293) acc1: 41.6667 (21.9618) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1358 max mem: 3531 +Test: [ 150/2387] eta: 0:09:03 acc1_pos: 17.188 acc5_pos: 100.000 loss: 0.8737 (1.2070) acc1: 41.6667 (23.1529) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1297 max mem: 3531 +Test: [ 160/2387] eta: 0:08:55 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.8279 (1.1764) acc1: 42.9688 (25.1100) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2149 data: 0.1178 max mem: 3531 +Test: [ 170/2387] eta: 0:08:43 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2100 (1.1195) acc1: 98.6979 (29.4530) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1849 data: 0.0880 max mem: 3531 +Test: [ 180/2387] eta: 0:08:32 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2053 (1.0691) acc1: 99.2188 (33.3103) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1649 data: 0.0685 max mem: 3531 +Test: [ 190/2387] eta: 0:08:23 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2068 (1.0239) acc1: 99.2188 (36.7624) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1699 data: 0.0716 max mem: 3531 +Test: [ 200/2387] eta: 0:08:15 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2044 (0.9833) acc1: 99.2188 (39.8735) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0765 max mem: 3531 +Test: [ 210/2387] eta: 0:08:07 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2068 (0.9466) acc1: 99.2188 (42.6861) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1699 data: 0.0716 max mem: 3532 +Test: [ 220/2387] eta: 0:08:00 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2059 (0.9130) acc1: 99.2188 (45.2442) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1749 data: 0.0771 max mem: 3532 +Test: [ 230/2387] eta: 0:07:53 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2047 (0.8823) acc1: 99.2188 (47.5909) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1767 data: 0.0829 max mem: 3532 +Test: [ 240/2387] eta: 0:07:49 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2095 (0.8593) acc1: 98.4375 (49.4424) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.1849 data: 0.0922 max mem: 3532 +Test: [ 250/2387] eta: 0:07:44 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2021 (0.8299) acc1: 94.5312 (51.4287) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1932 data: 0.0994 max mem: 3532 +Test: [ 260/2387] eta: 0:07:40 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.0884 (0.8013) acc1: 100.0000 (53.2896) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0979 max mem: 3532 +Test: [ 270/2387] eta: 0:07:36 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.0822 (0.7749) acc1: 100.0000 (55.0113) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1904 data: 0.0974 max mem: 3532 +Test: [ 280/2387] eta: 0:07:31 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.0877 (0.7506) acc1: 100.0000 (56.6049) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1825 data: 0.0898 max mem: 3532 +Test: [ 290/2387] eta: 0:07:26 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.0927 (0.7281) acc1: 100.0000 (58.0935) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1750 data: 0.0841 max mem: 3532 +Test: [ 300/2387] eta: 0:07:23 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1000 (0.7072) acc1: 100.0000 (59.4840) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1889 data: 0.0962 max mem: 3532 +Test: [ 310/2387] eta: 0:07:20 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1067 (0.6931) acc1: 100.0000 (60.6963) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2017 data: 0.1090 max mem: 3532 +Test: [ 320/2387] eta: 0:07:17 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.4515 (0.6856) acc1: 96.0938 (61.7723) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1983 data: 0.1068 max mem: 3532 +Test: [ 330/2387] eta: 0:07:13 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.4532 (0.6786) acc1: 95.5729 (62.7950) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1942 data: 0.1026 max mem: 3532 +Test: [ 340/2387] eta: 0:07:10 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.4512 (0.6719) acc1: 95.5729 (63.7547) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1949 data: 0.1037 max mem: 3532 +Test: [ 350/2387] eta: 0:07:07 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.4418 (0.6605) acc1: 97.1354 (64.7651) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1882 data: 0.0962 max mem: 3532 +Test: [ 360/2387] eta: 0:07:03 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2370 (0.6473) acc1: 99.7396 (65.7310) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1804 data: 0.0890 max mem: 3532 +Test: [ 370/2387] eta: 0:06:59 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1764 (0.6347) acc1: 99.7396 (66.6421) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1768 data: 0.0863 max mem: 3532 +Test: [ 380/2387] eta: 0:06:55 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1764 (0.6226) acc1: 99.7396 (67.5108) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1769 data: 0.0842 max mem: 3532 +Test: [ 390/2387] eta: 0:06:52 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1787 (0.6113) acc1: 99.7396 (68.3311) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1847 data: 0.0903 max mem: 3532 +Test: [ 400/2387] eta: 0:06:49 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1795 (0.6005) acc1: 99.4792 (69.1111) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1836 data: 0.0886 max mem: 3532 +Test: [ 410/2387] eta: 0:06:47 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1770 (0.5902) acc1: 99.7396 (69.8525) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1899 data: 0.0952 max mem: 3532 +Test: [ 420/2387] eta: 0:06:44 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1770 (0.5804) acc1: 99.7396 (70.5599) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1942 data: 0.1016 max mem: 3532 +Test: [ 430/2387] eta: 0:06:41 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1799 (0.5728) acc1: 99.4792 (71.1608) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1938 data: 0.1005 max mem: 3532 +Test: [ 440/2387] eta: 0:06:40 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1829 (0.5642) acc1: 96.0938 (71.7675) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2074 data: 0.1138 max mem: 3533 +Test: [ 450/2387] eta: 0:06:37 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1245 (0.5545) acc1: 100.0000 (72.3935) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2061 data: 0.1152 max mem: 3533 +Test: [ 460/2387] eta: 0:06:35 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1218 (0.5451) acc1: 100.0000 (72.9924) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1987 data: 0.1079 max mem: 3533 +Test: [ 470/2387] eta: 0:06:33 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1214 (0.5361) acc1: 100.0000 (73.5658) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2002 data: 0.1083 max mem: 3533 +Test: [ 480/2387] eta: 0:06:30 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1230 (0.5276) acc1: 100.0000 (74.1153) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1933 data: 0.1016 max mem: 3533 +Test: [ 490/2387] eta: 0:06:28 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1244 (0.5194) acc1: 100.0000 (74.6425) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1958 data: 0.1021 max mem: 3533 +Test: [ 500/2387] eta: 0:06:26 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1233 (0.5115) acc1: 100.0000 (75.1476) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2082 data: 0.1147 max mem: 3533 +Test: [ 510/2387] eta: 0:06:23 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1227 (0.5039) acc1: 100.0000 (75.6340) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1998 data: 0.1084 max mem: 3533 +Test: [ 520/2387] eta: 0:06:21 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1336 (0.4968) acc1: 100.0000 (76.1016) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1915 data: 0.0976 max mem: 3533 +Test: [ 530/2387] eta: 0:06:18 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1338 (0.4900) acc1: 100.0000 (76.5517) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1921 data: 0.0962 max mem: 3533 +Test: [ 540/2387] eta: 0:06:17 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1337 (0.4834) acc1: 100.0000 (76.9851) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1999 data: 0.1061 max mem: 3533 +Test: [ 550/2387] eta: 0:06:14 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1338 (0.4771) acc1: 100.0000 (77.4028) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.1014 max mem: 3533 +Test: [ 560/2387] eta: 0:06:13 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1330 (0.4709) acc1: 100.0000 (77.8056) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2086 data: 0.1137 max mem: 3533 +Test: [ 570/2387] eta: 0:06:12 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1331 (0.4650) acc1: 100.0000 (78.1943) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2455 data: 0.1493 max mem: 3533 +Test: [ 580/2387] eta: 0:06:08 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1334 (0.4593) acc1: 100.0000 (78.5696) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1916 data: 0.0941 max mem: 3533 +Test: [ 590/2387] eta: 0:06:04 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1332 (0.4538) acc1: 100.0000 (78.9322) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1393 data: 0.0448 max mem: 3533 +Test: [ 600/2387] eta: 0:06:02 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1343 (0.4485) acc1: 100.0000 (79.2828) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1660 data: 0.0731 max mem: 3533 +Test: [ 610/2387] eta: 0:05:59 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1420 (0.4437) acc1: 100.0000 (79.6219) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1867 data: 0.0946 max mem: 3533 +Test: [ 620/2387] eta: 0:05:59 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1502 (0.4390) acc1: 100.0000 (79.9500) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2284 data: 0.1352 max mem: 3533 +Test: [ 630/2387] eta: 0:05:56 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1502 (0.4344) acc1: 100.0000 (80.2678) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2181 data: 0.1221 max mem: 3533 +Test: [ 640/2387] eta: 0:05:53 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1526 (0.4300) acc1: 100.0000 (80.5756) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1685 data: 0.0743 max mem: 3533 +Test: [ 650/2387] eta: 0:05:51 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1527 (0.4258) acc1: 100.0000 (80.8740) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1816 data: 0.0880 max mem: 3533 +Test: [ 660/2387] eta: 0:05:49 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1531 (0.4216) acc1: 100.0000 (81.1629) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2063 data: 0.1129 max mem: 3533 +Test: [ 670/2387] eta: 0:05:46 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1529 (0.4176) acc1: 100.0000 (81.4437) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1854 data: 0.0904 max mem: 3534 +Test: [ 680/2387] eta: 0:05:44 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1555 (0.4146) acc1: 100.0000 (81.7162) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1760 data: 0.0786 max mem: 3534 +Test: [ 690/2387] eta: 0:05:42 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2451 (0.4168) acc1: 100.0000 (81.5289) acc5: 100.0000 (100.0000) acc1_neg: 54.167 acc5_neg: 100.000 time: 0.2082 data: 0.1147 max mem: 3534 +Test: [ 700/2387] eta: 0:05:39 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.6791 (0.4206) acc1: 58.0729 (81.1961) acc5: 100.0000 (100.0000) acc1_neg: 59.115 acc5_neg: 100.000 time: 0.1900 data: 0.0980 max mem: 3534 +Test: [ 710/2387] eta: 0:05:38 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.6850 (0.4243) acc1: 58.0729 (80.8632) acc5: 100.0000 (100.0000) acc1_neg: 55.208 acc5_neg: 100.000 time: 0.1944 data: 0.1021 max mem: 3534 +Test: [ 720/2387] eta: 0:05:35 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.6884 (0.4280) acc1: 56.7708 (80.5222) acc5: 100.0000 (100.0000) acc1_neg: 55.208 acc5_neg: 100.000 time: 0.1938 data: 0.1012 max mem: 3534 +Test: [ 730/2387] eta: 0:05:33 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.6907 (0.4316) acc1: 56.7708 (80.1891) acc5: 100.0000 (100.0000) acc1_neg: 56.771 acc5_neg: 100.000 time: 0.1725 data: 0.0798 max mem: 3534 +Test: [ 740/2387] eta: 0:05:30 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.6869 (0.4350) acc1: 56.7708 (79.8833) acc5: 100.0000 (100.0000) acc1_neg: 57.292 acc5_neg: 100.000 time: 0.1900 data: 0.0986 max mem: 3534 +Test: [ 750/2387] eta: 0:05:28 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.6841 (0.4383) acc1: 58.0729 (79.5914) acc5: 100.0000 (100.0000) acc1_neg: 59.635 acc5_neg: 100.000 time: 0.1929 data: 0.1015 max mem: 3534 +Test: [ 760/2387] eta: 0:05:26 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.6850 (0.4416) acc1: 58.0729 (79.3039) acc5: 100.0000 (100.0000) acc1_neg: 62.760 acc5_neg: 100.000 time: 0.1818 data: 0.0906 max mem: 3534 +Test: [ 770/2387] eta: 0:05:23 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.6802 (0.4419) acc1: 58.8542 (79.2663) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1813 data: 0.0899 max mem: 3534 +Test: [ 780/2387] eta: 0:05:21 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2159 (0.4388) acc1: 98.9583 (79.5191) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1879 data: 0.0958 max mem: 3534 +Test: [ 790/2387] eta: 0:05:18 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1758 (0.4356) acc1: 99.7396 (79.7112) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.1789 data: 0.0862 max mem: 3534 +Test: [ 800/2387] eta: 0:05:16 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1466 (0.4321) acc1: 100.0000 (79.9583) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1777 data: 0.0825 max mem: 3534 +Test: [ 810/2387] eta: 0:05:14 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1650 (0.4290) acc1: 100.0000 (80.1981) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1759 data: 0.0798 max mem: 3534 +Test: [ 820/2387] eta: 0:05:11 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1735 (0.4263) acc1: 99.7396 (80.4088) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1730 data: 0.0803 max mem: 3534 +Test: [ 830/2387] eta: 0:05:09 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.3062 (0.4261) acc1: 91.1458 (80.4568) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1778 data: 0.0855 max mem: 3534 +Test: [ 840/2387] eta: 0:05:06 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.3038 (0.4239) acc1: 92.4479 (80.6626) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1758 data: 0.0834 max mem: 3534 +Test: [ 850/2387] eta: 0:05:04 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2174 (0.4213) acc1: 98.9583 (80.8754) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1732 data: 0.0818 max mem: 3534 +Test: [ 860/2387] eta: 0:05:01 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1992 (0.4187) acc1: 99.2188 (81.0824) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1706 data: 0.0793 max mem: 3534 +Test: [ 870/2387] eta: 0:04:59 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1920 (0.4163) acc1: 99.2188 (81.2892) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1832 data: 0.0925 max mem: 3534 +Test: [ 880/2387] eta: 0:04:57 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2016 (0.4140) acc1: 99.2188 (81.4882) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1792 data: 0.0888 max mem: 3534 +Test: [ 890/2387] eta: 0:04:55 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2023 (0.4118) acc1: 98.6979 (81.6758) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1784 data: 0.0866 max mem: 3534 +Test: [ 900/2387] eta: 0:04:53 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2619 (0.4155) acc1: 94.0104 (81.3734) acc5: 100.0000 (100.0000) acc1_neg: 46.354 acc5_neg: 100.000 time: 0.1932 data: 0.1013 max mem: 3535 +Test: [ 910/2387] eta: 0:04:51 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.8574 (0.4204) acc1: 44.5312 (80.9736) acc5: 100.0000 (100.0000) acc1_neg: 44.531 acc5_neg: 100.000 time: 0.1937 data: 0.1019 max mem: 3535 +Test: [ 920/2387] eta: 0:04:49 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.8572 (0.4252) acc1: 44.7917 (80.5700) acc5: 100.0000 (100.0000) acc1_neg: 41.146 acc5_neg: 100.000 time: 0.2110 data: 0.1182 max mem: 3535 +Test: [ 930/2387] eta: 0:04:47 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.8547 (0.4299) acc1: 43.4896 (80.1686) acc5: 100.0000 (100.0000) acc1_neg: 43.490 acc5_neg: 100.000 time: 0.1961 data: 0.1039 max mem: 3535 +Test: [ 940/2387] eta: 0:04:45 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.8566 (0.4344) acc1: 43.4896 (79.7893) acc5: 100.0000 (100.0000) acc1_neg: 43.750 acc5_neg: 100.000 time: 0.1853 data: 0.0935 max mem: 3535 +Test: [ 950/2387] eta: 0:04:43 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.8619 (0.4390) acc1: 42.9688 (79.3986) acc5: 100.0000 (100.0000) acc1_neg: 46.615 acc5_neg: 100.000 time: 0.2149 data: 0.1235 max mem: 3535 +Test: [ 960/2387] eta: 0:04:41 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.8601 (0.4433) acc1: 42.1875 (79.0331) acc5: 100.0000 (100.0000) acc1_neg: 49.219 acc5_neg: 100.000 time: 0.1889 data: 0.0962 max mem: 3535 +Test: [ 970/2387] eta: 0:04:38 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.8596 (0.4476) acc1: 43.4896 (78.6737) acc5: 100.0000 (100.0000) acc1_neg: 43.229 acc5_neg: 100.000 time: 0.1674 data: 0.0747 max mem: 3535 +Test: [ 980/2387] eta: 0:04:36 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.8494 (0.4474) acc1: 45.0521 (78.6995) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1843 data: 0.0911 max mem: 3535 +Test: [ 990/2387] eta: 0:04:34 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.4647 (0.4485) acc1: 74.2188 (78.6340) acc5: 100.0000 (100.0000) acc1_neg: 68.490 acc5_neg: 100.000 time: 0.1775 data: 0.0835 max mem: 3535 +Test: [1000/2387] eta: 0:04:32 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.5760 (0.4498) acc1: 68.7500 (78.5457) acc5: 100.0000 (100.0000) acc1_neg: 65.885 acc5_neg: 100.000 time: 0.1823 data: 0.0902 max mem: 3535 +Test: [1010/2387] eta: 0:04:30 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.5802 (0.4522) acc1: 66.6667 (78.3391) acc5: 100.0000 (100.0000) acc1_neg: 42.708 acc5_neg: 100.000 time: 0.1951 data: 0.1040 max mem: 3535 +Test: [1020/2387] eta: 0:04:28 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.6692 (0.4524) acc1: 58.0729 (78.3035) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1890 data: 0.0964 max mem: 3535 +Test: [1030/2387] eta: 0:04:26 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1723 (0.4496) acc1: 99.4792 (78.5099) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1857 data: 0.0931 max mem: 3535 +Test: [1040/2387] eta: 0:04:24 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1454 (0.4466) acc1: 100.0000 (78.7121) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1806 data: 0.0888 max mem: 3535 +Test: [1050/2387] eta: 0:04:21 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1332 (0.4437) acc1: 100.0000 (78.9139) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1720 data: 0.0803 max mem: 3535 +Test: [1060/2387] eta: 0:04:19 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1406 (0.4410) acc1: 100.0000 (79.1102) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1653 data: 0.0744 max mem: 3535 +Test: [1070/2387] eta: 0:04:17 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1684 (0.4388) acc1: 100.0000 (79.2873) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1732 data: 0.0819 max mem: 3535 +Test: [1080/2387] eta: 0:04:15 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2588 (0.4377) acc1: 96.6146 (79.4293) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1836 data: 0.0918 max mem: 3535 +Test: [1090/2387] eta: 0:04:12 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2669 (0.4359) acc1: 97.1354 (79.6056) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1586 data: 0.0670 max mem: 3535 +Test: [1100/2387] eta: 0:04:10 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2209 (0.4339) acc1: 99.4792 (79.7833) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1469 data: 0.0561 max mem: 3535 +Test: [1110/2387] eta: 0:04:07 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1877 (0.4320) acc1: 99.4792 (79.9160) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1597 data: 0.0684 max mem: 3535 +Test: [1120/2387] eta: 0:04:05 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1877 (0.4300) acc1: 99.7396 (80.0757) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1717 data: 0.0800 max mem: 3536 +Test: [1130/2387] eta: 0:04:03 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2203 (0.4283) acc1: 99.7396 (80.2392) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1700 data: 0.0776 max mem: 3536 +Test: [1140/2387] eta: 0:04:01 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2862 (0.4285) acc1: 95.8333 (80.2611) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.1584 data: 0.0663 max mem: 3536 +Test: [1150/2387] eta: 0:03:58 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.4088 (0.4273) acc1: 86.9792 (80.3590) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1650 data: 0.0737 max mem: 3536 +Test: [1160/2387] eta: 0:03:56 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1707 (0.4249) acc1: 100.0000 (80.5275) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1652 data: 0.0738 max mem: 3536 +Test: [1170/2387] eta: 0:03:54 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1329 (0.4224) acc1: 100.0000 (80.6907) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1587 data: 0.0669 max mem: 3536 +Test: [1180/2387] eta: 0:03:52 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1407 (0.4201) acc1: 99.7396 (80.8471) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1685 data: 0.0763 max mem: 3536 +Test: [1190/2387] eta: 0:03:50 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1594 (0.4180) acc1: 99.7396 (81.0038) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1717 data: 0.0801 max mem: 3536 +Test: [1200/2387] eta: 0:03:47 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1549 (0.4158) acc1: 100.0000 (81.1596) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1635 data: 0.0726 max mem: 3536 +Test: [1210/2387] eta: 0:03:45 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1584 (0.4138) acc1: 99.7396 (81.3121) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1609 data: 0.0695 max mem: 3536 +Test: [1220/2387] eta: 0:03:43 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1436 (0.4115) acc1: 100.0000 (81.4635) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1578 data: 0.0667 max mem: 3536 +Test: [1230/2387] eta: 0:03:41 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1124 (0.4090) acc1: 100.0000 (81.6132) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1655 data: 0.0742 max mem: 3536 +Test: [1240/2387] eta: 0:03:39 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1146 (0.4067) acc1: 100.0000 (81.7603) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1701 data: 0.0785 max mem: 3536 +Test: [1250/2387] eta: 0:03:36 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1219 (0.4044) acc1: 100.0000 (81.9039) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1597 data: 0.0686 max mem: 3536 +Test: [1260/2387] eta: 0:03:34 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1233 (0.4022) acc1: 100.0000 (82.0455) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1659 data: 0.0748 max mem: 3536 +Test: [1270/2387] eta: 0:03:32 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1269 (0.4001) acc1: 100.0000 (82.1861) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1739 data: 0.0829 max mem: 3536 +Test: [1280/2387] eta: 0:03:30 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1269 (0.3979) acc1: 100.0000 (82.3230) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1621 data: 0.0709 max mem: 3536 +Test: [1290/2387] eta: 0:03:28 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1234 (0.3959) acc1: 100.0000 (82.4419) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1534 data: 0.0621 max mem: 3536 +Test: [1300/2387] eta: 0:03:26 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1182 (0.3938) acc1: 100.0000 (82.5709) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1617 data: 0.0702 max mem: 3536 +Test: [1310/2387] eta: 0:03:24 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1210 (0.3918) acc1: 100.0000 (82.6963) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1593 data: 0.0680 max mem: 3536 +Test: [1320/2387] eta: 0:03:21 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1323 (0.3899) acc1: 100.0000 (82.8147) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1514 data: 0.0603 max mem: 3536 +Test: [1330/2387] eta: 0:03:19 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1548 (0.3887) acc1: 98.9583 (82.9223) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1563 data: 0.0657 max mem: 3536 +Test: [1340/2387] eta: 0:03:17 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2584 (0.3896) acc1: 94.2708 (82.8397) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1569 data: 0.0668 max mem: 3536 +Test: [1350/2387] eta: 0:03:15 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1199 (0.3875) acc1: 99.7396 (82.9654) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1500 data: 0.0590 max mem: 3537 +Test: [1360/2387] eta: 0:03:13 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.0813 (0.3852) acc1: 100.0000 (83.0905) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1565 data: 0.0628 max mem: 3537 +Test: [1370/2387] eta: 0:03:11 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.0756 (0.3830) acc1: 100.0000 (83.2139) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1594 data: 0.0652 max mem: 3537 +Test: [1380/2387] eta: 0:03:09 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.0809 (0.3809) acc1: 100.0000 (83.3354) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1530 data: 0.0607 max mem: 3537 +Test: [1390/2387] eta: 0:03:06 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.0862 (0.3788) acc1: 100.0000 (83.4552) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1566 data: 0.0656 max mem: 3537 +Test: [1400/2387] eta: 0:03:04 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1030 (0.3770) acc1: 100.0000 (83.5709) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1579 data: 0.0673 max mem: 3537 +Test: [1410/2387] eta: 0:03:02 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1400 (0.3755) acc1: 100.0000 (83.6870) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1522 data: 0.0615 max mem: 3537 +Test: [1420/2387] eta: 0:03:00 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1553 (0.3741) acc1: 100.0000 (83.7799) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1495 data: 0.0593 max mem: 3537 +Test: [1430/2387] eta: 0:02:58 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1512 (0.3725) acc1: 100.0000 (83.8904) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1632 data: 0.0728 max mem: 3537 +Test: [1440/2387] eta: 0:02:56 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1537 (0.3712) acc1: 100.0000 (83.9949) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1642 data: 0.0732 max mem: 3537 +Test: [1450/2387] eta: 0:02:54 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1994 (0.3704) acc1: 98.4375 (84.0728) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1576 data: 0.0664 max mem: 3537 +Test: [1460/2387] eta: 0:02:52 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1441 (0.3688) acc1: 98.6979 (84.1788) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1651 data: 0.0739 max mem: 3537 +Test: [1470/2387] eta: 0:02:50 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1117 (0.3670) acc1: 100.0000 (84.2863) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1639 data: 0.0720 max mem: 3537 +Test: [1480/2387] eta: 0:02:48 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1007 (0.3652) acc1: 100.0000 (84.3917) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1575 data: 0.0659 max mem: 3537 +Test: [1490/2387] eta: 0:02:46 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1106 (0.3636) acc1: 100.0000 (84.4960) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1617 data: 0.0712 max mem: 3537 +Test: [1500/2387] eta: 0:02:44 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1149 (0.3619) acc1: 100.0000 (84.5978) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1654 data: 0.0751 max mem: 3537 +Test: [1510/2387] eta: 0:02:42 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1203 (0.3605) acc1: 100.0000 (84.6987) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1632 data: 0.0726 max mem: 3537 +Test: [1520/2387] eta: 0:02:40 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1596 (0.3595) acc1: 99.2188 (84.7887) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1616 data: 0.0708 max mem: 3537 +Test: [1530/2387] eta: 0:02:38 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1955 (0.3584) acc1: 98.9583 (84.8790) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1649 data: 0.0737 max mem: 3537 +Test: [1540/2387] eta: 0:02:36 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1735 (0.3572) acc1: 99.2188 (84.9736) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1670 data: 0.0756 max mem: 3537 +Test: [1550/2387] eta: 0:02:34 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1684 (0.3561) acc1: 99.4792 (85.0651) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1625 data: 0.0708 max mem: 3537 +Test: [1560/2387] eta: 0:02:32 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1881 (0.3550) acc1: 99.2188 (85.1553) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1579 data: 0.0656 max mem: 3537 +Test: [1570/2387] eta: 0:02:30 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1859 (0.3539) acc1: 99.2188 (85.2431) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1586 data: 0.0669 max mem: 3537 +Test: [1580/2387] eta: 0:02:28 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1846 (0.3529) acc1: 99.7396 (85.3351) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1621 data: 0.0705 max mem: 3538 +Test: [1590/2387] eta: 0:02:26 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1819 (0.3518) acc1: 100.0000 (85.4267) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1649 data: 0.0732 max mem: 3538 +Test: [1600/2387] eta: 0:02:24 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1596 (0.3508) acc1: 100.0000 (85.4858) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1677 data: 0.0762 max mem: 3538 +Test: [1610/2387] eta: 0:02:23 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1473 (0.3495) acc1: 100.0000 (85.5746) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1736 data: 0.0823 max mem: 3538 +Test: [1620/2387] eta: 0:02:21 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1624 (0.3485) acc1: 99.7396 (85.6584) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1756 data: 0.0833 max mem: 3538 +Test: [1630/2387] eta: 0:02:19 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1989 (0.3479) acc1: 99.2188 (85.7279) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1717 data: 0.0794 max mem: 3538 +Test: [1640/2387] eta: 0:02:17 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2733 (0.3478) acc1: 95.5729 (85.7633) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1745 data: 0.0833 max mem: 3538 +Test: [1650/2387] eta: 0:02:15 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.3062 (0.3474) acc1: 94.7917 (85.8276) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1667 data: 0.0750 max mem: 3538 +Test: [1660/2387] eta: 0:02:13 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2846 (0.3472) acc1: 95.8333 (85.8817) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1610 data: 0.0692 max mem: 3538 +Test: [1670/2387] eta: 0:02:11 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2890 (0.3468) acc1: 95.5729 (85.9380) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1643 data: 0.0737 max mem: 3538 +Test: [1680/2387] eta: 0:02:09 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2890 (0.3465) acc1: 96.3542 (85.9939) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1654 data: 0.0747 max mem: 3538 +Test: [1690/2387] eta: 0:02:07 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.3011 (0.3463) acc1: 97.6562 (86.0473) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1683 data: 0.0767 max mem: 3538 +Test: [1700/2387] eta: 0:02:05 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.3011 (0.3458) acc1: 97.6562 (86.1125) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1647 data: 0.0723 max mem: 3538 +Test: [1710/2387] eta: 0:02:03 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1959 (0.3448) acc1: 99.2188 (86.1895) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1588 data: 0.0666 max mem: 3538 +Test: [1720/2387] eta: 0:02:02 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1635 (0.3437) acc1: 99.7396 (86.2689) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1607 data: 0.0692 max mem: 3538 +Test: [1730/2387] eta: 0:02:00 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1603 (0.3427) acc1: 99.7396 (86.3444) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1721 data: 0.0813 max mem: 3538 +Test: [1740/2387] eta: 0:01:58 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1616 (0.3417) acc1: 99.7396 (86.4206) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1742 data: 0.0831 max mem: 3538 +Test: [1750/2387] eta: 0:01:56 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1603 (0.3407) acc1: 100.0000 (86.4955) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1597 data: 0.0666 max mem: 3538 +Test: [1760/2387] eta: 0:01:54 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1772 (0.3399) acc1: 99.7396 (86.5661) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1602 data: 0.0674 max mem: 3538 +Test: [1770/2387] eta: 0:01:52 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2076 (0.3393) acc1: 98.6979 (86.6304) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1642 data: 0.0719 max mem: 3538 +Test: [1780/2387] eta: 0:01:50 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1950 (0.3384) acc1: 98.9583 (86.7046) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1628 data: 0.0700 max mem: 3538 +Test: [1790/2387] eta: 0:01:48 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1721 (0.3374) acc1: 100.0000 (86.7769) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1691 data: 0.0776 max mem: 3538 +Test: [1800/2387] eta: 0:01:46 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1713 (0.3365) acc1: 99.7396 (86.8474) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1672 data: 0.0767 max mem: 3538 +Test: [1810/2387] eta: 0:01:45 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1638 (0.3357) acc1: 99.4792 (86.9168) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1696 data: 0.0790 max mem: 3539 +Test: [1820/2387] eta: 0:01:43 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1728 (0.3349) acc1: 99.4792 (86.9817) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1832 data: 0.0885 max mem: 3539 +Test: [1830/2387] eta: 0:01:41 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2024 (0.3340) acc1: 99.4792 (87.0476) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1957 data: 0.1007 max mem: 3539 +Test: [1840/2387] eta: 0:01:39 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.0847 (0.3326) acc1: 100.0000 (87.1177) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1918 data: 0.1008 max mem: 3539 +Test: [1850/2387] eta: 0:01:37 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.0835 (0.3313) acc1: 100.0000 (87.1872) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1912 data: 0.0989 max mem: 3539 +Test: [1860/2387] eta: 0:01:36 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.0834 (0.3300) acc1: 100.0000 (87.2561) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1925 data: 0.0984 max mem: 3539 +Test: [1870/2387] eta: 0:01:34 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.0849 (0.3286) acc1: 100.0000 (87.3242) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1824 data: 0.0899 max mem: 3539 +Test: [1880/2387] eta: 0:01:32 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.0838 (0.3273) acc1: 100.0000 (87.3916) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1943 data: 0.1030 max mem: 3539 +Test: [1890/2387] eta: 0:01:30 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.0833 (0.3261) acc1: 100.0000 (87.4583) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1921 data: 0.0986 max mem: 3539 +Test: [1900/2387] eta: 0:01:28 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.0845 (0.3248) acc1: 100.0000 (87.5242) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1700 data: 0.0774 max mem: 3539 +Test: [1910/2387] eta: 0:01:26 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1150 (0.3238) acc1: 100.0000 (87.5889) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1700 data: 0.0792 max mem: 3539 +Test: [1920/2387] eta: 0:01:25 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1398 (0.3231) acc1: 100.0000 (87.6531) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1772 data: 0.0847 max mem: 3539 +Test: [1930/2387] eta: 0:01:23 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1617 (0.3222) acc1: 100.0000 (87.7167) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1852 data: 0.0930 max mem: 3539 +Test: [1940/2387] eta: 0:01:21 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1385 (0.3216) acc1: 100.0000 (87.7457) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1876 data: 0.0957 max mem: 3539 +Test: [1950/2387] eta: 0:01:19 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1465 (0.3207) acc1: 100.0000 (87.8085) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1943 data: 0.1018 max mem: 3539 +Test: [1960/2387] eta: 0:01:17 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1481 (0.3200) acc1: 100.0000 (87.8655) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2078 data: 0.1157 max mem: 3539 +Test: [1970/2387] eta: 0:01:16 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2519 (0.3198) acc1: 99.7396 (87.8907) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.1925 data: 0.1008 max mem: 3539 +Test: [1980/2387] eta: 0:01:14 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2736 (0.3193) acc1: 92.4479 (87.9295) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1939 data: 0.1009 max mem: 3539 +Test: [1990/2387] eta: 0:01:12 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1128 (0.3183) acc1: 99.2188 (87.9862) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1872 data: 0.0944 max mem: 3539 +Test: [2000/2387] eta: 0:01:10 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1128 (0.3172) acc1: 99.2188 (88.0426) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1742 data: 0.0824 max mem: 3539 +Test: [2010/2387] eta: 0:01:08 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1132 (0.3162) acc1: 99.4792 (88.0997) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1854 data: 0.0938 max mem: 3539 +Test: [2020/2387] eta: 0:01:07 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1122 (0.3152) acc1: 99.4792 (88.1557) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1795 data: 0.0894 max mem: 3539 +Test: [2030/2387] eta: 0:01:05 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1125 (0.3142) acc1: 99.4792 (88.2105) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1830 data: 0.0924 max mem: 3540 +Test: [2040/2387] eta: 0:01:03 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1125 (0.3132) acc1: 99.4792 (88.2659) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1877 data: 0.0959 max mem: 3540 +Test: [2050/2387] eta: 0:01:01 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1118 (0.3123) acc1: 99.4792 (88.3207) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1797 data: 0.0886 max mem: 3540 +Test: [2060/2387] eta: 0:00:59 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1085 (0.3113) acc1: 99.4792 (88.3744) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1778 data: 0.0875 max mem: 3540 +Test: [2070/2387] eta: 0:00:57 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1084 (0.3103) acc1: 99.4792 (88.4286) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1729 data: 0.0819 max mem: 3540 +Test: [2080/2387] eta: 0:00:56 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1084 (0.3093) acc1: 99.4792 (88.4816) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1658 data: 0.0748 max mem: 3540 +Test: [2090/2387] eta: 0:00:54 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1093 (0.3084) acc1: 99.2188 (88.5336) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1674 data: 0.0768 max mem: 3540 +Test: [2100/2387] eta: 0:00:52 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1085 (0.3074) acc1: 99.4792 (88.5863) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1691 data: 0.0777 max mem: 3540 +Test: [2110/2387] eta: 0:00:50 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1081 (0.3065) acc1: 99.4792 (88.6378) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1745 data: 0.0832 max mem: 3540 +Test: [2120/2387] eta: 0:00:48 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1215 (0.3059) acc1: 99.7396 (88.6902) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1756 data: 0.0833 max mem: 3540 +Test: [2130/2387] eta: 0:00:46 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1535 (0.3052) acc1: 100.0000 (88.7416) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1790 data: 0.0861 max mem: 3540 +Test: [2140/2387] eta: 0:00:45 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1526 (0.3047) acc1: 100.0000 (88.7678) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1832 data: 0.0918 max mem: 3540 +Test: [2150/2387] eta: 0:00:43 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1554 (0.3040) acc1: 100.0000 (88.8195) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0884 max mem: 3540 +Test: [2160/2387] eta: 0:00:41 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1540 (0.3033) acc1: 100.0000 (88.8674) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1843 data: 0.0913 max mem: 3540 +Test: [2170/2387] eta: 0:00:39 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1737 (0.3032) acc1: 99.4792 (88.8986) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1849 data: 0.0928 max mem: 3540 +Test: [2180/2387] eta: 0:00:37 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2982 (0.3047) acc1: 90.3646 (88.7832) acc5: 100.0000 (100.0000) acc1_neg: 24.740 acc5_neg: 100.000 time: 0.1799 data: 0.0870 max mem: 3540 +Test: [2190/2387] eta: 0:00:35 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.4753 (0.3056) acc1: 77.8646 (88.7162) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1863 data: 0.0929 max mem: 3540 +Test: [2200/2387] eta: 0:00:34 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1785 (0.3050) acc1: 98.9583 (88.7668) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1857 data: 0.0950 max mem: 3540 +Test: [2210/2387] eta: 0:00:32 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1531 (0.3043) acc1: 100.0000 (88.8175) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1784 data: 0.0829 max mem: 3540 +Test: [2220/2387] eta: 0:00:30 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1423 (0.3035) acc1: 100.0000 (88.8659) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1753 data: 0.0801 max mem: 3540 +Test: [2230/2387] eta: 0:00:28 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1365 (0.3028) acc1: 100.0000 (88.9158) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1727 data: 0.0808 max mem: 3540 +Test: [2240/2387] eta: 0:00:26 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1419 (0.3022) acc1: 100.0000 (88.9624) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1686 data: 0.0758 max mem: 3540 +Test: [2250/2387] eta: 0:00:24 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1721 (0.3016) acc1: 99.7396 (89.0066) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1671 data: 0.0762 max mem: 3540 +Test: [2260/2387] eta: 0:00:23 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1547 (0.3010) acc1: 99.7396 (89.0543) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1649 data: 0.0713 max mem: 3541 +Test: [2270/2387] eta: 0:00:21 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1319 (0.3002) acc1: 100.0000 (89.1025) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1599 data: 0.0645 max mem: 3541 +Test: [2280/2387] eta: 0:00:19 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1170 (0.2996) acc1: 100.0000 (89.1264) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1608 data: 0.0682 max mem: 3541 +Test: [2290/2387] eta: 0:00:17 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1157 (0.2988) acc1: 100.0000 (89.1738) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1615 data: 0.0702 max mem: 3541 +Test: [2300/2387] eta: 0:00:15 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1224 (0.2981) acc1: 100.0000 (89.2181) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1689 data: 0.0740 max mem: 3541 +Test: [2310/2387] eta: 0:00:13 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1455 (0.2976) acc1: 99.7396 (89.2576) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1638 data: 0.0690 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2003 (0.2974) acc1: 98.6979 (89.2845) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1512 data: 0.0594 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1941 (0.2969) acc1: 100.0000 (89.3302) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1544 data: 0.0608 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1598 (0.2962) acc1: 100.0000 (89.3754) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1835 data: 0.0894 max mem: 3541 +Test: [2350/2387] eta: 0:00:06 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1303 (0.2959) acc1: 100.0000 (89.3913) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2125 data: 0.1199 max mem: 3541 +Test: [2360/2387] eta: 0:00:04 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1604 (0.2954) acc1: 100.0000 (89.4353) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1906 data: 0.0980 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1646 (0.2949) acc1: 100.0000 (89.4758) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1679 data: 0.0727 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2225 (0.2950) acc1: 98.1771 (89.4850) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1610 data: 0.0666 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2409 (0.2947) acc1: 96.0938 (89.4996) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1463 data: 0.0568 max mem: 3541 +Test: Total time: 0:07:13 (0.1814 s / it) +* Overall Acc@1 89.500 Acc@5 100.000 + target == real Acc@1 23.721 Acc@5 100.000 + target == fake Acc@1 94.119 Acc@5 100.000 + loss 0.295 + PR-AUC 0.9811 ROC-AUC 0.7950 EER 27.13% (thr=0.7949) +Accuracy of the network on the 916267 test images: 89.5% +[rank0]:[W726 05:46:44.745820994 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:47:05.320000 3161108 torch/distributed/run.py:766] +W0726 05:47:05.320000 3161108 torch/distributed/run.py:766] ***************************************** +W0726 05:47:05.320000 3161108 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:47:05.320000 3161108 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:47:33.090509828 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:47:34.530499023 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:47:34.610950341 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:47:34.821432172 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:47:34.931181385 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:47:34.085355347 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:47:34.123403183 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W726 05:47:34.123401889 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n11_seed8_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed8_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:07:33 acc1_pos: 5.469 acc5_pos: 100.000 loss: 1.5227 (1.5227) acc1: 5.4688 (5.4688) acc5: 100.0000 (100.0000) time: 7.8398 data: 4.9318 max mem: 3531 +Test: [ 10/517] eta: 0:06:42 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.4520 (1.4896) acc1: 5.4688 (5.7528) acc5: 100.0000 (100.0000) time: 0.7941 data: 0.4486 max mem: 3531 +Test: [ 20/517] eta: 0:03:48 acc1_pos: 0.260 acc5_pos: 100.000 loss: 1.5670 (1.5778) acc1: 2.0833 (3.9063) acc5: 100.0000 (100.0000) time: 0.0916 data: 0.0003 max mem: 3531 +Test: [ 30/517] eta: 0:02:55 acc1_pos: 3.646 acc5_pos: 100.000 loss: 1.5832 (1.5576) acc1: 2.0833 (4.0575) acc5: 100.0000 (100.0000) time: 0.1233 data: 0.0315 max mem: 3531 +Test: [ 40/517] eta: 0:02:40 acc1_pos: 2.865 acc5_pos: 100.000 loss: 1.5139 (1.5536) acc1: 2.6042 (3.7729) acc5: 100.0000 (100.0000) time: 0.2049 data: 0.1097 max mem: 3531 +Test: [ 50/517] eta: 0:02:28 acc1_pos: 12.844 acc5_pos: 100.000 loss: 1.4553 (1.4837) acc1: 2.8646 (8.9001) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2485 data: 0.1535 max mem: 3531 +Test: [ 60/517] eta: 0:02:19 acc1_pos: 12.844 acc5_pos: 100.000 loss: 0.2309 (1.2728) acc1: 98.9583 (23.7577) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2449 data: 0.1514 max mem: 3531 +Test: [ 70/517] eta: 0:02:12 acc1_pos: 12.844 acc5_pos: 100.000 loss: 0.2070 (1.1230) acc1: 99.7396 (34.3493) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2398 data: 0.1424 max mem: 3531 +Test: [ 80/517] eta: 0:02:05 acc1_pos: 12.844 acc5_pos: 100.000 loss: 0.1855 (1.0082) acc1: 100.0000 (42.4029) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2299 data: 0.1353 max mem: 3531 +Test: [ 90/517] eta: 0:02:00 acc1_pos: 12.844 acc5_pos: 100.000 loss: 0.1831 (0.9189) acc1: 100.0000 (48.7237) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2399 data: 0.1443 max mem: 3531 +Test: [100/517] eta: 0:01:55 acc1_pos: 12.844 acc5_pos: 100.000 loss: 0.1806 (0.8469) acc1: 100.0000 (53.7825) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2399 data: 0.1443 max mem: 3531 +Test: [110/517] eta: 0:01:51 acc1_pos: 12.844 acc5_pos: 100.000 loss: 0.1662 (0.7847) acc1: 100.0000 (57.9275) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2349 data: 0.1429 max mem: 3531 +Test: [120/517] eta: 0:01:48 acc1_pos: 12.844 acc5_pos: 100.000 loss: 0.1863 (0.7362) acc1: 100.0000 (61.4045) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2499 data: 0.1569 max mem: 3531 +Test: [130/517] eta: 0:01:44 acc1_pos: 12.844 acc5_pos: 100.000 loss: 0.1949 (0.6947) acc1: 100.0000 (64.3468) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2499 data: 0.1554 max mem: 3531 +Test: [140/517] eta: 0:01:41 acc1_pos: 12.844 acc5_pos: 100.000 loss: 0.1358 (0.6545) acc1: 100.0000 (66.8754) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2499 data: 0.1555 max mem: 3531 +Test: [150/517] eta: 0:01:38 acc1_pos: 12.844 acc5_pos: 100.000 loss: 0.1255 (0.6197) acc1: 100.0000 (69.0673) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2451 data: 0.1520 max mem: 3531 +Test: [160/517] eta: 0:01:35 acc1_pos: 12.844 acc5_pos: 100.000 loss: 0.1396 (0.5905) acc1: 100.0000 (70.9886) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2499 data: 0.1569 max mem: 3531 +Test: [170/517] eta: 0:01:32 acc1_pos: 12.844 acc5_pos: 100.000 loss: 0.1477 (0.5649) acc1: 100.0000 (72.6806) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2647 data: 0.1702 max mem: 3531 +Test: [180/517] eta: 0:01:29 acc1_pos: 12.844 acc5_pos: 100.000 loss: 0.1467 (0.5430) acc1: 100.0000 (74.1871) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2449 data: 0.1516 max mem: 3531 +Test: [190/517] eta: 0:01:26 acc1_pos: 12.844 acc5_pos: 100.000 loss: 0.1342 (0.5211) acc1: 100.0000 (75.5372) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1607 max mem: 3531 +Test: [200/517] eta: 0:01:23 acc1_pos: 12.844 acc5_pos: 100.000 loss: 0.1286 (0.5022) acc1: 100.0000 (76.7335) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2649 data: 0.1681 max mem: 3531 +Test: [210/517] eta: 0:01:20 acc1_pos: 12.844 acc5_pos: 100.000 loss: 0.1514 (0.4859) acc1: 100.0000 (77.8288) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2449 data: 0.1473 max mem: 3532 +Test: [220/517] eta: 0:01:18 acc1_pos: 12.844 acc5_pos: 100.000 loss: 0.1592 (0.4717) acc1: 100.0000 (78.8026) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2549 data: 0.1569 max mem: 3532 +Test: [230/517] eta: 0:01:15 acc1_pos: 12.844 acc5_pos: 100.000 loss: 0.1669 (0.4586) acc1: 99.7396 (79.7010) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2499 data: 0.1529 max mem: 3532 +Test: [240/517] eta: 0:01:13 acc1_pos: 12.844 acc5_pos: 100.000 loss: 0.1271 (0.4444) acc1: 100.0000 (80.5433) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2599 data: 0.1631 max mem: 3532 +Test: [250/517] eta: 0:01:10 acc1_pos: 12.844 acc5_pos: 100.000 loss: 0.1271 (0.4323) acc1: 100.0000 (81.3164) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2749 data: 0.1786 max mem: 3532 +Test: [260/517] eta: 0:01:07 acc1_pos: 12.844 acc5_pos: 100.000 loss: 0.1422 (0.4216) acc1: 100.0000 (82.0293) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2499 data: 0.1524 max mem: 3532 +Test: [270/517] eta: 0:01:04 acc1_pos: 12.844 acc5_pos: 100.000 loss: 0.1452 (0.4117) acc1: 100.0000 (82.6924) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2449 data: 0.1484 max mem: 3532 +Test: [280/517] eta: 0:01:02 acc1_pos: 12.844 acc5_pos: 100.000 loss: 0.1408 (0.4023) acc1: 100.0000 (83.2953) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2499 data: 0.1546 max mem: 3532 +Test: [290/517] eta: 0:00:59 acc1_pos: 12.844 acc5_pos: 100.000 loss: 0.1327 (0.3928) acc1: 100.0000 (83.8631) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1591 max mem: 3532 +Test: [300/517] eta: 0:00:56 acc1_pos: 12.844 acc5_pos: 100.000 loss: 0.1353 (0.3849) acc1: 100.0000 (84.3992) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2499 data: 0.1544 max mem: 3532 +Test: [310/517] eta: 0:00:54 acc1_pos: 12.844 acc5_pos: 100.000 loss: 0.1725 (0.3785) acc1: 100.0000 (84.8967) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2499 data: 0.1561 max mem: 3532 +Test: [320/517] eta: 0:00:51 acc1_pos: 12.844 acc5_pos: 100.000 loss: 0.1785 (0.3721) acc1: 100.0000 (85.3574) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2749 data: 0.1796 max mem: 3532 +Test: [330/517] eta: 0:00:48 acc1_pos: 12.844 acc5_pos: 100.000 loss: 0.1421 (0.3650) acc1: 100.0000 (85.7982) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2699 data: 0.1731 max mem: 3532 +Test: [340/517] eta: 0:00:46 acc1_pos: 12.844 acc5_pos: 100.000 loss: 0.1338 (0.3595) acc1: 100.0000 (86.2033) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2449 data: 0.1509 max mem: 3532 +Test: [350/517] eta: 0:00:43 acc1_pos: 12.844 acc5_pos: 100.000 loss: 0.1510 (0.3543) acc1: 100.0000 (86.5934) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2449 data: 0.1508 max mem: 3532 +Test: [360/517] eta: 0:00:40 acc1_pos: 12.844 acc5_pos: 100.000 loss: 0.1569 (0.3492) acc1: 100.0000 (86.9575) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2451 data: 0.1505 max mem: 3532 +Test: [370/517] eta: 0:00:38 acc1_pos: 12.844 acc5_pos: 100.000 loss: 0.1514 (0.3441) acc1: 100.0000 (87.3070) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2449 data: 0.1510 max mem: 3532 +Test: [380/517] eta: 0:00:35 acc1_pos: 12.844 acc5_pos: 100.000 loss: 0.1607 (0.3403) acc1: 100.0000 (87.6401) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2555 data: 0.1580 max mem: 3532 +Test: [390/517] eta: 0:00:32 acc1_pos: 12.844 acc5_pos: 100.000 loss: 0.2129 (0.3370) acc1: 100.0000 (87.9516) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1568 max mem: 3532 +Test: [400/517] eta: 0:00:30 acc1_pos: 12.844 acc5_pos: 100.000 loss: 0.1876 (0.3331) acc1: 100.0000 (88.2494) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2541 data: 0.1570 max mem: 3532 +Test: [410/517] eta: 0:00:27 acc1_pos: 12.844 acc5_pos: 100.000 loss: 0.1908 (0.3300) acc1: 100.0000 (88.5315) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2511 data: 0.1535 max mem: 3532 +Test: [420/517] eta: 0:00:25 acc1_pos: 12.844 acc5_pos: 100.000 loss: 0.1735 (0.3260) acc1: 100.0000 (88.8015) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2449 data: 0.1490 max mem: 3532 +Test: [430/517] eta: 0:00:22 acc1_pos: 12.844 acc5_pos: 100.000 loss: 0.1618 (0.3221) acc1: 100.0000 (89.0468) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2437 data: 0.1475 max mem: 3532 +Test: [440/517] eta: 0:00:19 acc1_pos: 12.844 acc5_pos: 100.000 loss: 0.1770 (0.3191) acc1: 100.0000 (89.2952) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2349 data: 0.1373 max mem: 3533 +Test: [450/517] eta: 0:00:17 acc1_pos: 12.844 acc5_pos: 100.000 loss: 0.1751 (0.3157) acc1: 100.0000 (89.5279) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2399 data: 0.1423 max mem: 3533 +Test: [460/517] eta: 0:00:14 acc1_pos: 12.844 acc5_pos: 100.000 loss: 0.1728 (0.3130) acc1: 100.0000 (89.7500) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2507 data: 0.1549 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 12.844 acc5_pos: 100.000 loss: 0.1903 (0.3104) acc1: 100.0000 (89.9610) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1596 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 12.844 acc5_pos: 100.000 loss: 0.1714 (0.3070) acc1: 100.0000 (90.1697) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2691 data: 0.1728 max mem: 3533 +Test: [490/517] eta: 0:00:06 acc1_pos: 12.844 acc5_pos: 100.000 loss: 0.1463 (0.3042) acc1: 100.0000 (90.3651) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2456 data: 0.1506 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 12.844 acc5_pos: 100.000 loss: 0.1916 (0.3027) acc1: 99.7396 (90.5320) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2207 data: 0.1278 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 12.844 acc5_pos: 100.000 loss: 0.2387 (0.3012) acc1: 99.4792 (90.6540) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2069 data: 0.1161 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 12.844 acc5_pos: 100.000 loss: 0.1700 (0.2994) acc1: 100.0000 (90.7578) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1860 data: 0.0942 max mem: 3533 +Test: Total time: 0:02:11 (0.2553 s / it) +* Overall Acc@1 90.758 Acc@5 100.000 + target == real Acc@1 4.338 Acc@5 100.000 + target == fake Acc@1 99.723 Acc@5 100.000 + loss 0.299 + PR-AUC 0.9591 ROC-AUC 0.7411 EER 32.10% (thr=0.8335) +Accuracy of the network on the 198427 test images: 90.8% +[rank0]:[W726 05:49:59.558220964 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 05:50:23.137000 3161681 torch/distributed/run.py:766] +W0726 05:50:23.137000 3161681 torch/distributed/run.py:766] ***************************************** +W0726 05:50:23.137000 3161681 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:50:23.137000 3161681 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:50:57.060013033 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:50:57.163391557 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:50:58.335735401 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:50:58.626641098 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:50:58.659769141 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:50:58.697265541 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:50:58.711447439 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:50:58.722184385 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n11_seed8_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed8_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:35:17 acc1_pos: 54.427 acc5_pos: 100.000 loss: 0.6875 (0.6875) acc1: 54.4271 (54.4271) acc5: 100.0000 (100.0000) time: 5.4427 data: 3.3719 max mem: 3531 +Test: [ 10/389] eta: 0:03:40 acc1_pos: 12.240 acc5_pos: 100.000 loss: 0.9615 (0.9587) acc1: 33.3333 (30.2794) acc5: 100.0000 (100.0000) time: 0.5818 data: 0.3084 max mem: 3531 +Test: [ 20/389] eta: 0:02:14 acc1_pos: 6.250 acc5_pos: 100.000 loss: 1.0766 (1.0741) acc1: 10.9375 (19.2460) acc5: 100.0000 (100.0000) time: 0.1100 data: 0.0182 max mem: 3531 +Test: [ 30/389] eta: 0:01:55 acc1_pos: 65.104 acc5_pos: 100.000 loss: 1.0719 (1.0434) acc1: 9.8958 (20.3293) acc5: 100.0000 (100.0000) time: 0.1797 data: 0.0865 max mem: 3531 +Test: [ 40/389] eta: 0:01:52 acc1_pos: 20.312 acc5_pos: 100.000 loss: 0.9522 (0.9892) acc1: 25.2604 (24.8666) acc5: 100.0000 (100.0000) time: 0.2816 data: 0.1883 max mem: 3531 +Test: [ 50/389] eta: 0:01:40 acc1_pos: 76.562 acc5_pos: 100.000 loss: 0.8349 (0.9674) acc1: 32.5521 (27.0118) acc5: 100.0000 (100.0000) time: 0.2585 data: 0.1687 max mem: 3531 +Test: [ 60/389] eta: 0:01:28 acc1_pos: 27.865 acc5_pos: 100.000 loss: 0.5704 (0.9095) acc1: 73.4375 (34.0591) acc5: 100.0000 (100.0000) time: 0.1577 data: 0.0656 max mem: 3531 +Test: [ 70/389] eta: 0:01:20 acc1_pos: 59.375 acc5_pos: 100.000 loss: 0.5671 (0.9054) acc1: 73.6979 (35.7138) acc5: 100.0000 (100.0000) time: 0.1432 data: 0.0500 max mem: 3531 +Test: [ 80/389] eta: 0:01:17 acc1_pos: 73.958 acc5_pos: 100.000 loss: 0.9742 (0.9156) acc1: 36.7188 (35.7768) acc5: 100.0000 (100.0000) time: 0.1955 data: 0.1042 max mem: 3531 +Test: [ 90/389] eta: 0:01:11 acc1_pos: 36.979 acc5_pos: 100.000 loss: 0.7374 (0.8917) acc1: 46.6146 (37.7747) acc5: 100.0000 (100.0000) time: 0.1799 data: 0.0887 max mem: 3531 +Test: [100/389] eta: 0:01:07 acc1_pos: 34.115 acc5_pos: 100.000 loss: 0.7048 (0.8838) acc1: 48.9583 (38.7583) acc5: 100.0000 (100.0000) time: 0.1543 data: 0.0601 max mem: 3531 +Test: [110/389] eta: 0:01:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.7956 (0.8774) acc1: 43.4896 (39.6631) acc5: 100.0000 (100.0000) time: 0.1938 data: 0.0994 max mem: 3531 +Test: [120/389] eta: 0:01:00 acc1_pos: 29.167 acc5_pos: 100.000 loss: 0.8087 (0.8660) acc1: 39.3229 (40.0525) acc5: 100.0000 (100.0000) time: 0.1921 data: 0.0980 max mem: 3531 +Test: [130/389] eta: 0:00:57 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.5930 (0.8382) acc1: 62.7604 (42.7759) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1760 data: 0.0824 max mem: 3531 +Test: [140/389] eta: 0:00:53 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.4342 (0.8173) acc1: 87.2396 (45.1370) acc5: 100.0000 (100.0000) acc1_neg: 43.750 acc5_neg: 100.000 time: 0.1577 data: 0.0620 max mem: 3531 +Test: [150/389] eta: 0:00:51 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.4134 (0.7914) acc1: 90.3646 (47.6494) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1622 data: 0.0663 max mem: 3531 +Test: [160/389] eta: 0:00:48 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.3077 (0.7635) acc1: 95.5729 (50.3364) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1775 data: 0.0868 max mem: 3531 +Test: [170/389] eta: 0:00:45 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.2914 (0.7421) acc1: 97.9167 (52.5874) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1676 data: 0.0761 max mem: 3531 +Test: [180/389] eta: 0:00:43 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.3241 (0.7192) acc1: 97.3958 (54.9609) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1644 data: 0.0709 max mem: 3531 +Test: [190/389] eta: 0:00:40 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.3458 (0.7056) acc1: 95.8333 (56.3945) acc5: 100.0000 (100.0000) acc1_neg: 71.354 acc5_neg: 100.000 time: 0.1621 data: 0.0691 max mem: 3531 +Test: [200/389] eta: 0:00:38 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.3567 (0.6885) acc1: 97.1354 (58.2828) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1695 data: 0.0778 max mem: 3531 +Test: [210/389] eta: 0:00:36 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.3054 (0.6680) acc1: 99.7396 (60.2377) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1832 data: 0.0922 max mem: 3532 +Test: [220/389] eta: 0:00:33 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.2119 (0.6495) acc1: 100.0000 (61.9038) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1857 data: 0.0936 max mem: 3532 +Test: [230/389] eta: 0:00:31 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.3436 (0.6388) acc1: 97.9167 (63.1505) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1564 data: 0.0644 max mem: 3532 +Test: [240/389] eta: 0:00:29 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.3472 (0.6263) acc1: 97.1354 (64.3391) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1480 data: 0.0559 max mem: 3532 +Test: [250/389] eta: 0:00:27 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.2896 (0.6134) acc1: 99.4792 (65.7339) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1583 data: 0.0663 max mem: 3532 +Test: [260/389] eta: 0:00:24 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.2840 (0.6081) acc1: 99.2188 (66.1498) acc5: 100.0000 (100.0000) acc1_neg: 17.969 acc5_neg: 100.000 time: 0.1559 data: 0.0655 max mem: 3532 +Test: [270/389] eta: 0:00:22 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.4512 (0.6098) acc1: 84.8958 (66.0392) acc5: 100.0000 (100.0000) acc1_neg: 72.917 acc5_neg: 100.000 time: 0.1682 data: 0.0773 max mem: 3532 +Test: [280/389] eta: 0:00:20 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.5809 (0.6132) acc1: 75.5208 (65.9123) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1794 data: 0.0870 max mem: 3532 +Test: [290/389] eta: 0:00:18 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.4994 (0.6052) acc1: 77.0833 (66.7329) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1811 data: 0.0880 max mem: 3532 +Test: [300/389] eta: 0:00:17 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.2901 (0.5953) acc1: 99.7396 (67.7715) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1801 data: 0.0881 max mem: 3532 +Test: [310/389] eta: 0:00:15 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.3097 (0.5905) acc1: 95.8333 (68.3665) acc5: 100.0000 (100.0000) acc1_neg: 50.260 acc5_neg: 100.000 time: 0.1873 data: 0.0960 max mem: 3532 +Test: [320/389] eta: 0:00:13 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.3948 (0.5859) acc1: 94.0104 (68.7605) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1905 data: 0.0985 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.4144 (0.5813) acc1: 94.2708 (69.4313) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.1829 data: 0.0912 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.3271 (0.5723) acc1: 96.6146 (70.2400) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1845 data: 0.0935 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.3241 (0.5788) acc1: 98.9583 (69.5075) acc5: 100.0000 (100.0000) acc1_neg: 0.260 acc5_neg: 100.000 time: 0.1855 data: 0.0935 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.6310 (0.5837) acc1: 58.0729 (68.8950) acc5: 100.0000 (100.0000) acc1_neg: 63.802 acc5_neg: 100.000 time: 0.1806 data: 0.0880 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.6078 (0.5789) acc1: 63.2812 (69.2666) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1795 data: 0.0879 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.3591 (0.5728) acc1: 98.1771 (69.8928) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.1645 data: 0.0745 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.3653 (0.5697) acc1: 95.3125 (70.1871) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.1602 data: 0.0653 max mem: 3532 +Test: Total time: 0:01:13 (0.1891 s / it) +* Overall Acc@1 70.187 Acc@5 100.000 + target == real Acc@1 39.859 Acc@5 100.000 + target == fake Acc@1 84.284 Acc@5 100.000 + loss 0.570 + PR-AUC 0.8437 ROC-AUC 0.7199 EER 33.35% (thr=0.6484) +Accuracy of the network on the 149358 test images: 70.2% +[rank0]:[W726 05:52:23.030709798 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:52:43.702000 3162239 torch/distributed/run.py:766] +W0726 05:52:43.702000 3162239 torch/distributed/run.py:766] ***************************************** +W0726 05:52:43.702000 3162239 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:52:43.702000 3162239 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:53:15.375497224 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:53:15.108882099 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:53:16.237592364 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:53:16.254167334 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:53:16.423945116 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:53:16.530833852 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:53:16.595903908 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:53:16.733514011 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n11_seed8_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed8_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:29:59 acc1_pos: 75.260 acc5_pos: 100.000 loss: 0.5519 (0.5519) acc1: 75.2604 (75.2604) acc5: 100.0000 (100.0000) time: 5.4045 data: 3.4816 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/333] eta: 0:03:05 acc1_pos: 78.125 acc5_pos: 100.000 loss: 0.4620 (0.4737) acc1: 79.4271 (78.8589) acc5: 100.0000 (100.0000) time: 0.5744 data: 0.3169 max mem: 3531 +Test: [ 20/333] eta: 0:01:47 acc1_pos: 90.625 acc5_pos: 100.000 loss: 0.4473 (0.4663) acc1: 78.1250 (77.3810) acc5: 100.0000 (100.0000) time: 0.0907 data: 0.0005 max mem: 3531 +Test: [ 30/333] eta: 0:01:30 acc1_pos: 33.333 acc5_pos: 100.000 loss: 0.6101 (0.6476) acc1: 59.3750 (64.2305) acc5: 100.0000 (100.0000) time: 0.1481 data: 0.0578 max mem: 3531 +Test: [ 40/333] eta: 0:01:27 acc1_pos: 37.500 acc5_pos: 100.000 loss: 1.0011 (0.7434) acc1: 34.8958 (56.7073) acc5: 100.0000 (100.0000) time: 0.2479 data: 0.1571 max mem: 3531 +Test: [ 50/333] eta: 0:01:22 acc1_pos: 36.458 acc5_pos: 100.000 loss: 1.0011 (0.8004) acc1: 36.4583 (52.2212) acc5: 100.0000 (100.0000) time: 0.2763 data: 0.1850 max mem: 3531 +Test: [ 60/333] eta: 0:01:15 acc1_pos: 23.177 acc5_pos: 100.000 loss: 1.0200 (0.8356) acc1: 35.4167 (49.3554) acc5: 100.0000 (100.0000) time: 0.2349 data: 0.1407 max mem: 3531 +Test: [ 70/333] eta: 0:01:09 acc1_pos: 28.125 acc5_pos: 100.000 loss: 1.0203 (0.8607) acc1: 33.8542 (47.2088) acc5: 100.0000 (100.0000) time: 0.1985 data: 0.1027 max mem: 3531 +Test: [ 80/333] eta: 0:01:05 acc1_pos: 28.906 acc5_pos: 100.000 loss: 1.0151 (0.8820) acc1: 33.8542 (45.4861) acc5: 100.0000 (100.0000) time: 0.2049 data: 0.1104 max mem: 3531 +Test: [ 90/333] eta: 0:01:01 acc1_pos: 41.927 acc5_pos: 100.000 loss: 0.9686 (0.8890) acc1: 35.1562 (44.5084) acc5: 100.0000 (100.0000) time: 0.2099 data: 0.1132 max mem: 3531 +Test: [100/333] eta: 0:00:56 acc1_pos: 41.146 acc5_pos: 100.000 loss: 0.9339 (0.8938) acc1: 38.5417 (43.8944) acc5: 100.0000 (100.0000) time: 0.1799 data: 0.0821 max mem: 3531 +Test: [110/333] eta: 0:00:53 acc1_pos: 46.354 acc5_pos: 100.000 loss: 0.9088 (0.8962) acc1: 37.7604 (43.3441) acc5: 100.0000 (100.0000) time: 0.1949 data: 0.1013 max mem: 3531 +Test: [120/333] eta: 0:00:52 acc1_pos: 42.188 acc5_pos: 100.000 loss: 0.9088 (0.8981) acc1: 36.7188 (43.1711) acc5: 100.0000 (100.0000) time: 0.2599 data: 0.1656 max mem: 3531 +Test: [130/333] eta: 0:00:49 acc1_pos: 52.344 acc5_pos: 100.000 loss: 0.9009 (0.8986) acc1: 40.6250 (43.0781) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1444 max mem: 3531 +Test: [140/333] eta: 0:00:45 acc1_pos: 51.823 acc5_pos: 100.000 loss: 0.8903 (0.8970) acc1: 41.6667 (43.0574) acc5: 100.0000 (100.0000) time: 0.1756 data: 0.0814 max mem: 3531 +Test: [150/333] eta: 0:00:41 acc1_pos: 40.885 acc5_pos: 100.000 loss: 0.8735 (0.8984) acc1: 40.8854 (42.8618) acc5: 100.0000 (100.0000) time: 0.1406 data: 0.0466 max mem: 3531 +Test: [160/333] eta: 0:00:39 acc1_pos: 28.289 acc5_pos: 100.000 loss: 0.8564 (0.8937) acc1: 42.4479 (43.2955) acc5: 100.0000 (100.0000) acc1_neg: 59.896 acc5_neg: 100.000 time: 0.1642 data: 0.0712 max mem: 3531 +Test: [170/333] eta: 0:00:36 acc1_pos: 28.289 acc5_pos: 100.000 loss: 0.6673 (0.8743) acc1: 60.4167 (44.9013) acc5: 100.0000 (100.0000) acc1_neg: 74.219 acc5_neg: 100.000 time: 0.1902 data: 0.0929 max mem: 3531 +Test: [180/333] eta: 0:00:34 acc1_pos: 28.289 acc5_pos: 100.000 loss: 0.5370 (0.8551) acc1: 72.9167 (46.5153) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.1871 data: 0.0884 max mem: 3531 +Test: [190/333] eta: 0:00:31 acc1_pos: 28.289 acc5_pos: 100.000 loss: 0.4976 (0.8353) acc1: 74.7396 (48.1225) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.1954 data: 0.0973 max mem: 3531 +Test: [200/333] eta: 0:00:28 acc1_pos: 28.289 acc5_pos: 100.000 loss: 0.4760 (0.8181) acc1: 76.5625 (49.5932) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.1732 data: 0.0771 max mem: 3531 +Test: [210/333] eta: 0:00:26 acc1_pos: 28.289 acc5_pos: 100.000 loss: 0.5150 (0.8068) acc1: 73.6979 (50.5418) acc5: 100.0000 (100.0000) acc1_neg: 61.198 acc5_neg: 100.000 time: 0.1545 data: 0.0599 max mem: 3532 +Test: [220/333] eta: 0:00:24 acc1_pos: 28.289 acc5_pos: 100.000 loss: 0.6067 (0.8040) acc1: 61.1979 (50.7813) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.1763 data: 0.0810 max mem: 3532 +Test: [230/333] eta: 0:00:22 acc1_pos: 28.289 acc5_pos: 100.000 loss: 0.3702 (0.7841) acc1: 85.9375 (52.4351) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.2138 data: 0.1195 max mem: 3532 +Test: [240/333] eta: 0:00:19 acc1_pos: 28.289 acc5_pos: 100.000 loss: 0.3702 (0.7682) acc1: 85.9375 (53.8101) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.1979 data: 0.1023 max mem: 3532 +Test: [250/333] eta: 0:00:17 acc1_pos: 28.289 acc5_pos: 100.000 loss: 0.4225 (0.7584) acc1: 82.5521 (54.6605) acc5: 100.0000 (100.0000) acc1_neg: 63.021 acc5_neg: 100.000 time: 0.1668 data: 0.0721 max mem: 3532 +Test: [260/333] eta: 0:00:15 acc1_pos: 28.289 acc5_pos: 100.000 loss: 0.5954 (0.7530) acc1: 69.0104 (55.1325) acc5: 100.0000 (100.0000) acc1_neg: 67.708 acc5_neg: 100.000 time: 0.2149 data: 0.1227 max mem: 3532 +Test: [270/333] eta: 0:00:13 acc1_pos: 28.289 acc5_pos: 100.000 loss: 0.5597 (0.7447) acc1: 69.7917 (55.8349) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.2179 data: 0.1232 max mem: 3532 +Test: [280/333] eta: 0:00:11 acc1_pos: 28.289 acc5_pos: 100.000 loss: 0.5437 (0.7394) acc1: 72.6562 (56.2444) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.1709 data: 0.0749 max mem: 3532 +Test: [290/333] eta: 0:00:08 acc1_pos: 28.289 acc5_pos: 100.000 loss: 0.6330 (0.7365) acc1: 63.2812 (56.4415) acc5: 100.0000 (100.0000) acc1_neg: 66.406 acc5_neg: 100.000 time: 0.1799 data: 0.0875 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 28.289 acc5_pos: 100.000 loss: 0.6450 (0.7347) acc1: 60.9375 (56.6168) acc5: 100.0000 (100.0000) acc1_neg: 67.708 acc5_neg: 100.000 time: 0.1684 data: 0.0772 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 28.289 acc5_pos: 100.000 loss: 0.6119 (0.7266) acc1: 67.7083 (57.3360) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.1459 data: 0.0556 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 28.289 acc5_pos: 100.000 loss: 0.4447 (0.7188) acc1: 81.2500 (58.0267) acc5: 100.0000 (100.0000) acc1_neg: 51.562 acc5_neg: 100.000 time: 0.1346 data: 0.0444 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 28.289 acc5_pos: 100.000 loss: 0.7677 (0.7221) acc1: 54.6875 (57.8432) acc5: 100.0000 (100.0000) acc1_neg: 54.688 acc5_neg: 100.000 time: 0.1257 data: 0.0351 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 28.289 acc5_pos: 100.000 loss: 0.7944 (0.7229) acc1: 54.1667 (57.7873) acc5: 100.0000 (100.0000) acc1_neg: 49.577 acc5_neg: 100.000 time: 0.1224 data: 0.0286 max mem: 3532 +Test: Total time: 0:01:06 (0.2004 s / it) +* Overall Acc@1 57.787 Acc@5 100.000 + target == real Acc@1 42.853 Acc@5 100.000 + target == fake Acc@1 71.082 Acc@5 100.000 + loss 0.723 + PR-AUC 0.6138 ROC-AUC 0.6049 EER 42.64% (thr=0.5947) +Accuracy of the network on the 127843 test images: 57.8% +[rank0]:[W726 05:54:34.644928475 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:54:56.356000 3162798 torch/distributed/run.py:766] +W0726 05:54:56.356000 3162798 torch/distributed/run.py:766] ***************************************** +W0726 05:54:56.356000 3162798 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:54:56.356000 3162798 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:55:29.403841252 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:55:29.906504178 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:55:29.940925533 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:55:29.014991598 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:55:29.044085690 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:55:29.072193035 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:55:30.183690413 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank4]:[W726 05:55:30.183729255 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n11_seed8_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed8_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:20:57 acc1_pos: 87.240 acc5_pos: 100.000 loss: 0.3773 (0.3773) acc1: 87.2396 (87.2396) acc5: 100.0000 (100.0000) time: 5.3530 data: 3.6944 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/235] eta: 0:02:07 acc1_pos: 85.677 acc5_pos: 100.000 loss: 0.3737 (0.3679) acc1: 87.2396 (87.3816) acc5: 100.0000 (100.0000) time: 0.5684 data: 0.3363 max mem: 3531 +Test: [ 20/235] eta: 0:01:14 acc1_pos: 85.677 acc5_pos: 100.000 loss: 0.3640 (0.3600) acc1: 87.2396 (87.9712) acc5: 100.0000 (100.0000) time: 0.0972 data: 0.0053 max mem: 3531 +Test: [ 30/235] eta: 0:00:59 acc1_pos: 89.844 acc5_pos: 100.000 loss: 0.3498 (0.3586) acc1: 88.2812 (88.0208) acc5: 100.0000 (100.0000) time: 0.1368 data: 0.0440 max mem: 3531 +Test: [ 40/235] eta: 0:00:52 acc1_pos: 89.062 acc5_pos: 100.000 loss: 0.3562 (0.3587) acc1: 88.2812 (88.1098) acc5: 100.0000 (100.0000) time: 0.1902 data: 0.0993 max mem: 3531 +Test: [ 50/235] eta: 0:00:47 acc1_pos: 85.677 acc5_pos: 100.000 loss: 0.3610 (0.3599) acc1: 88.0208 (88.0106) acc5: 100.0000 (100.0000) time: 0.2096 data: 0.1195 max mem: 3531 +Test: [ 60/235] eta: 0:00:43 acc1_pos: 88.281 acc5_pos: 100.000 loss: 0.3683 (0.3614) acc1: 88.2812 (87.9611) acc5: 100.0000 (100.0000) time: 0.2061 data: 0.1142 max mem: 3531 +Test: [ 70/235] eta: 0:00:40 acc1_pos: 83.077 acc5_pos: 100.000 loss: 0.3854 (0.4857) acc1: 85.1562 (78.9943) acc5: 100.0000 (100.0000) acc1_neg: 14.323 acc5_neg: 100.000 time: 0.2148 data: 0.1194 max mem: 3531 +Test: [ 80/235] eta: 0:00:37 acc1_pos: 83.077 acc5_pos: 100.000 loss: 1.3605 (0.5941) acc1: 16.1458 (71.1484) acc5: 100.0000 (100.0000) acc1_neg: 16.927 acc5_neg: 100.000 time: 0.2176 data: 0.1225 max mem: 3531 +Test: [ 90/235] eta: 0:00:34 acc1_pos: 83.077 acc5_pos: 100.000 loss: 1.3622 (0.6795) acc1: 15.1042 (64.9210) acc5: 100.0000 (100.0000) acc1_neg: 14.062 acc5_neg: 100.000 time: 0.2049 data: 0.1108 max mem: 3531 +Test: [100/235] eta: 0:00:31 acc1_pos: 83.077 acc5_pos: 100.000 loss: 1.3525 (0.7460) acc1: 15.3646 (60.1124) acc5: 100.0000 (100.0000) acc1_neg: 15.625 acc5_neg: 100.000 time: 0.1965 data: 0.1033 max mem: 3531 +Test: [110/235] eta: 0:00:28 acc1_pos: 83.077 acc5_pos: 100.000 loss: 1.3437 (0.8008) acc1: 15.3646 (56.0600) acc5: 100.0000 (100.0000) acc1_neg: 15.365 acc5_neg: 100.000 time: 0.2049 data: 0.1127 max mem: 3531 +Test: [120/235] eta: 0:00:26 acc1_pos: 83.077 acc5_pos: 100.000 loss: 1.2902 (0.8340) acc1: 17.9688 (53.2326) acc5: 100.0000 (100.0000) acc1_neg: 25.260 acc5_neg: 100.000 time: 0.2198 data: 0.1285 max mem: 3531 +Test: [130/235] eta: 0:00:23 acc1_pos: 83.077 acc5_pos: 100.000 loss: 1.1808 (0.8604) acc1: 22.9167 (50.9940) acc5: 100.0000 (100.0000) acc1_neg: 25.260 acc5_neg: 100.000 time: 0.2060 data: 0.1132 max mem: 3531 +Test: [140/235] eta: 0:00:21 acc1_pos: 83.077 acc5_pos: 100.000 loss: 1.1779 (0.8847) acc1: 23.6979 (48.9787) acc5: 100.0000 (100.0000) acc1_neg: 19.792 acc5_neg: 100.000 time: 0.1896 data: 0.0964 max mem: 3531 +Test: [150/235] eta: 0:00:19 acc1_pos: 83.077 acc5_pos: 100.000 loss: 1.1875 (0.9043) acc1: 23.6979 (47.2975) acc5: 100.0000 (100.0000) acc1_neg: 22.396 acc5_neg: 100.000 time: 0.2011 data: 0.1078 max mem: 3531 +Test: [160/235] eta: 0:00:16 acc1_pos: 83.077 acc5_pos: 100.000 loss: 1.1843 (0.9222) acc1: 23.6979 (45.8139) acc5: 100.0000 (100.0000) acc1_neg: 25.260 acc5_neg: 100.000 time: 0.2012 data: 0.1042 max mem: 3531 +Test: [170/235] eta: 0:00:14 acc1_pos: 83.077 acc5_pos: 100.000 loss: 1.1318 (0.9281) acc1: 25.5208 (45.0186) acc5: 100.0000 (100.0000) acc1_neg: 34.115 acc5_neg: 100.000 time: 0.1897 data: 0.0930 max mem: 3531 +Test: [180/235] eta: 0:00:12 acc1_pos: 83.077 acc5_pos: 100.000 loss: 1.0114 (0.9320) acc1: 33.8542 (44.4032) acc5: 100.0000 (100.0000) acc1_neg: 32.552 acc5_neg: 100.000 time: 0.2575 data: 0.1636 max mem: 3531 +Test: [190/235] eta: 0:00:10 acc1_pos: 83.077 acc5_pos: 100.000 loss: 1.0073 (0.9357) acc1: 32.8125 (43.7936) acc5: 100.0000 (100.0000) acc1_neg: 32.292 acc5_neg: 100.000 time: 0.2578 data: 0.1641 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 83.077 acc5_pos: 100.000 loss: 1.0065 (0.9393) acc1: 32.2917 (43.2706) acc5: 100.0000 (100.0000) acc1_neg: 31.771 acc5_neg: 100.000 time: 0.2120 data: 0.1172 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 83.077 acc5_pos: 100.000 loss: 0.9930 (0.9416) acc1: 33.8542 (42.8688) acc5: 100.0000 (100.0000) acc1_neg: 33.594 acc5_neg: 100.000 time: 0.1941 data: 0.1019 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 83.077 acc5_pos: 100.000 loss: 0.9898 (0.9440) acc1: 33.8542 (42.4727) acc5: 100.0000 (100.0000) acc1_neg: 33.594 acc5_neg: 100.000 time: 0.1604 data: 0.0704 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 83.077 acc5_pos: 100.000 loss: 0.9921 (0.9464) acc1: 33.5938 (42.1199) acc5: 100.0000 (100.0000) acc1_neg: 35.938 acc5_neg: 100.000 time: 0.1347 data: 0.0448 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 83.077 acc5_pos: 100.000 loss: 0.9978 (0.9478) acc1: 33.5938 (42.0120) acc5: 100.0000 (100.0000) acc1_neg: 26.087 acc5_neg: 100.000 time: 0.1209 data: 0.0345 max mem: 3532 +Test: Total time: 0:00:49 (0.2127 s / it) +* Overall Acc@1 42.012 Acc@5 100.000 + target == real Acc@1 87.919 Acc@5 100.000 + target == fake Acc@1 25.408 Acc@5 100.000 + loss 0.948 + PR-AUC 0.8189 ROC-AUC 0.6563 EER 38.35% (thr=0.2734) +Accuracy of the network on the 89879 test images: 42.0% +[rank0]:[W726 05:56:28.509403695 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:56:51.000000 3163356 torch/distributed/run.py:766] +W0726 05:56:51.000000 3163356 torch/distributed/run.py:766] ***************************************** +W0726 05:56:51.000000 3163356 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:56:51.000000 3163356 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 05:57:25.499758043 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:57:25.785838736 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:57:25.835722946 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:57:25.859883928 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:57:26.320164506 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:57:26.380353437 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +[rank6]:[W726 05:57:26.398690384 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:57:26.401304172 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n11_seed8_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed8_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:29:53 acc1_pos: 39.583 acc5_pos: 100.000 loss: 0.8744 (0.8744) acc1: 39.5833 (39.5833) acc5: 100.0000 (100.0000) time: 6.6675 data: 5.1582 max mem: 3531 +Test: [ 10/269] eta: 0:03:17 acc1_pos: 55.208 acc5_pos: 100.000 loss: 0.8177 (0.7864) acc1: 46.0938 (46.4725) acc5: 100.0000 (100.0000) time: 0.7622 data: 0.5391 max mem: 3531 +Test: [ 20/269] eta: 0:01:51 acc1_pos: 45.312 acc5_pos: 100.000 loss: 0.7937 (0.7630) acc1: 46.0938 (49.3800) acc5: 100.0000 (100.0000) time: 0.1358 data: 0.0389 max mem: 3531 +Test: [ 30/269] eta: 0:01:31 acc1_pos: 56.771 acc5_pos: 100.000 loss: 0.7630 (0.7512) acc1: 49.7396 (50.5292) acc5: 100.0000 (100.0000) time: 0.1749 data: 0.0795 max mem: 3531 +Test: [ 40/269] eta: 0:01:17 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.6381 (0.6828) acc1: 64.8438 (57.6855) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.2272 data: 0.1358 max mem: 3531 +Test: [ 50/269] eta: 0:01:08 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.4598 (0.6446) acc1: 83.0729 (62.1477) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.2026 data: 0.1102 max mem: 3531 +Test: [ 60/269] eta: 0:01:02 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.5102 (0.6262) acc1: 78.6458 (64.5108) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.2127 data: 0.1204 max mem: 3531 +Test: [ 70/269] eta: 0:00:58 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.5452 (0.6147) acc1: 74.7396 (65.9698) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.2481 data: 0.1558 max mem: 3531 +Test: [ 80/269] eta: 0:00:54 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.4872 (0.5940) acc1: 77.8646 (68.5957) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2549 data: 0.1610 max mem: 3531 +Test: [ 90/269] eta: 0:00:48 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.4427 (0.5800) acc1: 87.7604 (70.5701) acc5: 100.0000 (100.0000) acc1_neg: 72.917 acc5_neg: 100.000 time: 0.1759 data: 0.0801 max mem: 3531 +Test: [100/269] eta: 0:00:45 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.4998 (0.5778) acc1: 83.3333 (70.8514) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.1898 data: 0.0943 max mem: 3531 +Test: [110/269] eta: 0:00:42 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.5424 (0.5769) acc1: 74.2188 (71.0140) acc5: 100.0000 (100.0000) acc1_neg: 68.490 acc5_neg: 100.000 time: 0.2558 data: 0.1616 max mem: 3531 +Test: [120/269] eta: 0:00:38 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.5111 (0.5665) acc1: 78.6458 (72.0192) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1949 data: 0.1008 max mem: 3531 +Test: [130/269] eta: 0:00:34 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.4797 (0.5638) acc1: 78.9062 (72.1275) acc5: 100.0000 (100.0000) acc1_neg: 58.854 acc5_neg: 100.000 time: 0.1421 data: 0.0470 max mem: 3531 +Test: [140/269] eta: 0:00:31 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.5802 (0.5673) acc1: 64.0625 (71.5149) acc5: 100.0000 (100.0000) acc1_neg: 64.062 acc5_neg: 100.000 time: 0.1849 data: 0.0874 max mem: 3531 +Test: [150/269] eta: 0:00:28 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.5610 (0.5628) acc1: 64.0625 (71.8336) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.2049 data: 0.1083 max mem: 3531 +Test: [160/269] eta: 0:00:25 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.4056 (0.5529) acc1: 87.7604 (72.9830) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1615 data: 0.0673 max mem: 3531 +Test: [170/269] eta: 0:00:23 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.4170 (0.5469) acc1: 89.3229 (73.6766) acc5: 100.0000 (100.0000) acc1_neg: 63.021 acc5_neg: 100.000 time: 0.2413 data: 0.1467 max mem: 3531 +Test: [180/269] eta: 0:00:21 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.4459 (0.5433) acc1: 78.9062 (73.9857) acc5: 100.0000 (100.0000) acc1_neg: 70.833 acc5_neg: 100.000 time: 0.2883 data: 0.1915 max mem: 3531 +Test: [190/269] eta: 0:00:18 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.4862 (0.5424) acc1: 76.5625 (74.0170) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1701 data: 0.0725 max mem: 3531 +Test: [200/269] eta: 0:00:16 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.4861 (0.5388) acc1: 76.5625 (74.3561) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.1517 data: 0.0563 max mem: 3531 +Test: [210/269] eta: 0:00:13 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.5058 (0.5383) acc1: 75.0000 (74.3027) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1911 data: 0.0992 max mem: 3532 +Test: [220/269] eta: 0:00:11 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.5058 (0.5348) acc1: 75.0000 (74.6088) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1685 data: 0.0778 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.4404 (0.5323) acc1: 82.0312 (74.8410) acc5: 100.0000 (100.0000) acc1_neg: 55.469 acc5_neg: 100.000 time: 0.1520 data: 0.0618 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.5463 (0.5348) acc1: 74.2188 (74.5959) acc5: 100.0000 (100.0000) acc1_neg: 72.917 acc5_neg: 100.000 time: 0.1367 data: 0.0469 max mem: 3532 +Test: [250/269] eta: 0:00:04 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.5744 (0.5348) acc1: 69.7917 (74.6026) acc5: 100.0000 (100.0000) acc1_neg: 64.323 acc5_neg: 100.000 time: 0.1208 data: 0.0311 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.5947 (0.5370) acc1: 65.6250 (74.3175) acc5: 100.0000 (100.0000) acc1_neg: 66.667 acc5_neg: 100.000 time: 0.1124 data: 0.0230 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.5578 (0.5360) acc1: 70.8333 (74.4726) acc5: 100.0000 (100.0000) acc1_neg: 92.913 acc5_neg: 100.000 time: 0.1132 data: 0.0202 max mem: 3532 +Test: Total time: 0:00:56 (0.2088 s / it) +* Overall Acc@1 74.473 Acc@5 100.000 + target == real Acc@1 50.008 Acc@5 100.000 + target == fake Acc@1 77.811 Acc@5 100.000 + loss 0.536 + PR-AUC 0.9423 ROC-AUC 0.7109 EER 34.77% (thr=0.5659) +Accuracy of the network on the 103293 test images: 74.5% +[rank0]:[W726 05:58:30.487180192 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T05:58:37Z] eval_done tag=fixed500k_fake_n11_seed8_resampled +[2026-07-26T05:58:38Z] eval_start tag=fixed500k_real_n5_seed9_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed9_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T05:58:38Z +hostname=n116-040-168 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed9_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n5_seed9_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 05:58:53.436000 3163920 torch/distributed/run.py:766] +W0726 05:58:53.436000 3163920 torch/distributed/run.py:766] ***************************************** +W0726 05:58:53.436000 3163920 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:58:53.436000 3163920 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:59:27.224110989 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:59:27.926418776 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:59:27.010147901 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:59:27.112732462 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:59:27.118961552 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:59:28.391037572 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:59:28.444571955 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:59:28.448926432 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n5_seed9_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed9_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:25:31 acc1_pos: 4.427 acc5_pos: 100.000 loss: 1.9410 (1.9410) acc1: 4.4271 (4.4271) acc5: 100.0000 (100.0000) time: 6.5435 data: 5.0186 max mem: 3531 +Test: [ 10/234] eta: 0:02:33 acc1_pos: 11.979 acc5_pos: 100.000 loss: 1.8117 (1.8652) acc1: 8.3333 (8.7358) acc5: 100.0000 (100.0000) time: 0.6836 data: 0.4566 max mem: 3531 +Test: [ 20/234] eta: 0:01:26 acc1_pos: 82.812 acc5_pos: 100.000 loss: 1.6274 (1.4702) acc1: 15.6250 (28.2614) acc5: 100.0000 (100.0000) time: 0.0988 data: 0.0004 max mem: 3531 +Test: [ 30/234] eta: 0:01:05 acc1_pos: 40.365 acc5_pos: 100.000 loss: 0.4067 (1.1565) acc1: 82.2917 (44.5985) acc5: 100.0000 (100.0000) time: 0.1219 data: 0.0229 max mem: 3531 +Test: [ 40/234] eta: 0:00:57 acc1_pos: 44.010 acc5_pos: 100.000 loss: 0.9558 (1.1387) acc1: 47.9167 (43.2609) acc5: 100.0000 (100.0000) time: 0.1816 data: 0.0853 max mem: 3531 +Test: [ 50/234] eta: 0:00:52 acc1_pos: 43.490 acc5_pos: 100.000 loss: 0.9806 (1.0999) acc1: 45.3125 (44.3730) acc5: 100.0000 (100.0000) time: 0.2229 data: 0.1263 max mem: 3531 +Test: [ 60/234] eta: 0:00:46 acc1_pos: 49.655 acc5_pos: 100.000 loss: 0.4382 (0.9400) acc1: 58.0729 (53.1720) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2132 data: 0.1176 max mem: 3531 +Test: [ 70/234] eta: 0:00:43 acc1_pos: 49.655 acc5_pos: 100.000 loss: 0.0906 (0.8427) acc1: 99.7396 (58.4067) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.2127 data: 0.1211 max mem: 3531 +Test: [ 80/234] eta: 0:00:39 acc1_pos: 49.655 acc5_pos: 100.000 loss: 0.3096 (0.7715) acc1: 86.1979 (62.3007) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2206 data: 0.1303 max mem: 3531 +Test: [ 90/234] eta: 0:00:36 acc1_pos: 49.655 acc5_pos: 100.000 loss: 0.2725 (0.7128) acc1: 88.8021 (65.5191) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2236 data: 0.1335 max mem: 3531 +Test: [100/234] eta: 0:00:34 acc1_pos: 49.655 acc5_pos: 100.000 loss: 0.2819 (0.6801) acc1: 88.5417 (67.1927) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2463 data: 0.1551 max mem: 3531 +Test: [110/234] eta: 0:00:31 acc1_pos: 49.655 acc5_pos: 100.000 loss: 0.2247 (0.6393) acc1: 90.6250 (69.4773) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.2308 data: 0.1375 max mem: 3531 +Test: [120/234] eta: 0:00:28 acc1_pos: 49.655 acc5_pos: 100.000 loss: 0.2034 (0.6101) acc1: 94.0104 (71.0550) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.2232 data: 0.1280 max mem: 3531 +Test: [130/234] eta: 0:00:25 acc1_pos: 49.655 acc5_pos: 100.000 loss: 0.1894 (0.5830) acc1: 94.7917 (72.5668) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2449 data: 0.1516 max mem: 3531 +Test: [140/234] eta: 0:00:23 acc1_pos: 49.655 acc5_pos: 100.000 loss: 0.3373 (0.5740) acc1: 85.4167 (72.9591) acc5: 100.0000 (100.0000) acc1_neg: 69.531 acc5_neg: 100.000 time: 0.2306 data: 0.1401 max mem: 3531 +Test: [150/234] eta: 0:00:20 acc1_pos: 49.655 acc5_pos: 100.000 loss: 0.3762 (0.5613) acc1: 80.2083 (73.5479) acc5: 100.0000 (100.0000) acc1_neg: 69.010 acc5_neg: 100.000 time: 0.2344 data: 0.1444 max mem: 3531 +Test: [160/234] eta: 0:00:18 acc1_pos: 49.655 acc5_pos: 100.000 loss: 0.2394 (0.5382) acc1: 89.5833 (74.8657) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.2299 data: 0.1384 max mem: 3531 +Test: [170/234] eta: 0:00:15 acc1_pos: 49.655 acc5_pos: 100.000 loss: 0.2154 (0.5195) acc1: 92.9688 (75.9396) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2279 data: 0.1335 max mem: 3531 +Test: [180/234] eta: 0:00:13 acc1_pos: 49.655 acc5_pos: 100.000 loss: 0.2387 (0.5076) acc1: 90.8854 (76.5524) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.2567 data: 0.1637 max mem: 3531 +Test: [190/234] eta: 0:00:10 acc1_pos: 49.655 acc5_pos: 100.000 loss: 0.2131 (0.4918) acc1: 93.7500 (77.5128) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2106 data: 0.1205 max mem: 3531 +Test: [200/234] eta: 0:00:08 acc1_pos: 49.655 acc5_pos: 100.000 loss: 0.1912 (0.4741) acc1: 96.0938 (78.5383) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1415 data: 0.0514 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 49.655 acc5_pos: 100.000 loss: 0.0982 (0.4577) acc1: 99.7396 (79.5234) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1332 data: 0.0435 max mem: 3532 +Test: [220/234] eta: 0:00:03 acc1_pos: 49.655 acc5_pos: 100.000 loss: 0.1886 (0.4470) acc1: 97.9167 (80.2743) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.1238 data: 0.0342 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 49.655 acc5_pos: 100.000 loss: 0.4240 (0.4653) acc1: 52.0833 (78.8318) acc5: 100.0000 (100.0000) acc1_neg: 45.573 acc5_neg: 100.000 time: 0.1225 data: 0.0331 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 49.655 acc5_pos: 100.000 loss: 0.8382 (0.4699) acc1: 48.4375 (78.4908) acc5: 100.0000 (100.0000) acc1_neg: 51.362 acc5_neg: 100.000 time: 0.1233 data: 0.0330 max mem: 3532 +Test: Total time: 0:00:51 (0.2209 s / it) +* Overall Acc@1 78.491 Acc@5 100.000 + target == real Acc@1 44.412 Acc@5 100.000 + target == fake Acc@1 88.096 Acc@5 100.000 + loss 0.470 + PR-AUC 0.9100 ROC-AUC 0.7659 EER 30.07% (thr=0.8120) +Accuracy of the network on the 89729 test images: 78.5% +[rank0]:[W726 06:00:27.437517059 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 06:00:49.041000 3164479 torch/distributed/run.py:766] +W0726 06:00:49.041000 3164479 torch/distributed/run.py:766] ***************************************** +W0726 06:00:49.041000 3164479 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:00:49.041000 3164479 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:01:21.467283605 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:01:21.743319452 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 06:01:21.926800739 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:01:22.290785578 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:01:22.438697063 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:01:22.469723446 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:01:22.506488919 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank1]:[W726 06:01:22.506674472 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n5_seed9_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed9_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 5:01:34 acc1_pos: 18.750 acc5_pos: 100.000 loss: 1.4829 (1.4829) acc1: 18.7500 (18.7500) acc5: 100.0000 (100.0000) time: 7.5803 data: 6.0786 max mem: 3531 +Test: [ 10/2387] eta: 0:30:34 acc1_pos: 13.021 acc5_pos: 100.000 loss: 1.4829 (1.4937) acc1: 15.8854 (18.8447) acc5: 100.0000 (100.0000) time: 0.7718 data: 0.5530 max mem: 3531 +Test: [ 20/2387] eta: 0:17:49 acc1_pos: 4.948 acc5_pos: 100.000 loss: 1.4528 (1.5081) acc1: 15.8854 (17.9563) acc5: 100.0000 (100.0000) time: 0.0954 data: 0.0005 max mem: 3531 +Test: [ 30/2387] eta: 0:15:57 acc1_pos: 3.906 acc5_pos: 100.000 loss: 1.5183 (1.5488) acc1: 15.3646 (16.3138) acc5: 100.0000 (100.0000) time: 0.2055 data: 0.1108 max mem: 3531 +Test: [ 40/2387] eta: 0:14:35 acc1_pos: 10.156 acc5_pos: 100.000 loss: 1.5519 (1.5459) acc1: 14.0625 (16.3237) acc5: 100.0000 (100.0000) time: 0.2899 data: 0.1943 max mem: 3531 +Test: [ 50/2387] eta: 0:13:25 acc1_pos: 12.240 acc5_pos: 100.000 loss: 1.6309 (1.5667) acc1: 13.8021 (16.2684) acc5: 100.0000 (100.0000) time: 0.2493 data: 0.1510 max mem: 3531 +Test: [ 60/2387] eta: 0:12:57 acc1_pos: 24.740 acc5_pos: 100.000 loss: 1.6051 (1.5677) acc1: 19.0104 (16.8673) acc5: 100.0000 (100.0000) time: 0.2549 data: 0.1583 max mem: 3531 +Test: [ 70/2387] eta: 0:12:07 acc1_pos: 25.781 acc5_pos: 100.000 loss: 1.5835 (1.5879) acc1: 19.5312 (16.9748) acc5: 100.0000 (100.0000) time: 0.2349 data: 0.1391 max mem: 3531 +Test: [ 80/2387] eta: 0:11:57 acc1_pos: 1.562 acc5_pos: 100.000 loss: 1.7921 (1.6293) acc1: 11.4583 (15.7890) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1448 max mem: 3531 +Test: [ 90/2387] eta: 0:11:41 acc1_pos: 5.208 acc5_pos: 100.000 loss: 1.9115 (1.6624) acc1: 8.0729 (15.0755) acc5: 100.0000 (100.0000) time: 0.2759 data: 0.1804 max mem: 3531 +Test: [ 100/2387] eta: 0:11:22 acc1_pos: 1.042 acc5_pos: 100.000 loss: 1.9409 (1.6958) acc1: 8.3333 (14.2224) acc5: 100.0000 (100.0000) time: 0.2487 data: 0.1556 max mem: 3531 +Test: [ 110/2387] eta: 0:11:09 acc1_pos: 8.594 acc5_pos: 100.000 loss: 1.9616 (1.7187) acc1: 8.5938 (13.8959) acc5: 100.0000 (100.0000) time: 0.2427 data: 0.1471 max mem: 3531 +Test: [ 120/2387] eta: 0:10:58 acc1_pos: 7.812 acc5_pos: 100.000 loss: 1.9779 (1.7424) acc1: 7.8125 (13.3523) acc5: 100.0000 (100.0000) time: 0.2499 data: 0.1510 max mem: 3531 +Test: [ 130/2387] eta: 0:10:47 acc1_pos: 10.677 acc5_pos: 100.000 loss: 2.1001 (1.7696) acc1: 2.6042 (12.6491) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1471 max mem: 3531 +Test: [ 140/2387] eta: 0:10:38 acc1_pos: 2.344 acc5_pos: 100.000 loss: 2.1001 (1.7925) acc1: 2.3438 (12.0364) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1485 max mem: 3531 +Test: [ 150/2387] eta: 0:10:33 acc1_pos: 7.292 acc5_pos: 100.000 loss: 2.0293 (1.8050) acc1: 4.9479 (11.7515) acc5: 100.0000 (100.0000) time: 0.2599 data: 0.1639 max mem: 3531 +Test: [ 160/2387] eta: 0:10:19 acc1_pos: 11.872 acc5_pos: 100.000 loss: 1.9661 (1.7690) acc1: 5.9896 (13.7956) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2359 data: 0.1405 max mem: 3531 +Test: [ 170/2387] eta: 0:10:03 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.2199 (1.6782) acc1: 93.2292 (18.4896) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1899 data: 0.0958 max mem: 3531 +Test: [ 180/2387] eta: 0:09:51 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.2140 (1.5973) acc1: 94.7917 (22.7181) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1889 data: 0.0928 max mem: 3531 +Test: [ 190/2387] eta: 0:09:40 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.2119 (1.5246) acc1: 94.7917 (26.5093) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1949 data: 0.0988 max mem: 3531 +Test: [ 200/2387] eta: 0:09:33 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.2044 (1.4592) acc1: 94.7917 (29.8935) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2099 data: 0.1148 max mem: 3531 +Test: [ 210/2387] eta: 0:09:24 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.2122 (1.4005) acc1: 94.5312 (32.9322) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2149 data: 0.1189 max mem: 3532 +Test: [ 220/2387] eta: 0:09:15 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.2149 (1.3468) acc1: 93.7500 (35.7018) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1949 data: 0.0988 max mem: 3532 +Test: [ 230/2387] eta: 0:09:09 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.2149 (1.2976) acc1: 94.2708 (38.2587) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2049 data: 0.1066 max mem: 3532 +Test: [ 240/2387] eta: 0:09:02 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1840 (1.2494) acc1: 95.8333 (40.7503) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2099 data: 0.1114 max mem: 3532 +Test: [ 250/2387] eta: 0:08:56 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1123 (1.2031) acc1: 99.4792 (43.1016) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2099 data: 0.1133 max mem: 3532 +Test: [ 260/2387] eta: 0:08:49 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0808 (1.1600) acc1: 100.0000 (45.2816) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2053 data: 0.1103 max mem: 3532 +Test: [ 270/2387] eta: 0:08:36 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0767 (1.1202) acc1: 100.0000 (47.2949) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1549 data: 0.0591 max mem: 3532 +Test: [ 280/2387] eta: 0:08:32 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0770 (1.0831) acc1: 100.0000 (49.1706) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1694 data: 0.0729 max mem: 3532 +Test: [ 290/2387] eta: 0:08:29 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0778 (1.0486) acc1: 100.0000 (50.9137) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2249 data: 0.1291 max mem: 3532 +Test: [ 300/2387] eta: 0:08:25 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0765 (1.0163) acc1: 100.0000 (52.5384) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2249 data: 0.1276 max mem: 3532 +Test: [ 310/2387] eta: 0:08:18 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0795 (0.9895) acc1: 100.0000 (53.9573) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2006 data: 0.1025 max mem: 3532 +Test: [ 320/2387] eta: 0:08:10 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.2782 (0.9678) acc1: 94.5312 (55.1945) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1668 data: 0.0693 max mem: 3532 +Test: [ 330/2387] eta: 0:08:06 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.2951 (0.9474) acc1: 93.7500 (56.3719) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1842 data: 0.0823 max mem: 3532 +Test: [ 340/2387] eta: 0:08:02 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.2910 (0.9281) acc1: 94.2708 (57.5040) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2134 data: 0.1150 max mem: 3532 +Test: [ 350/2387] eta: 0:07:58 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.2711 (0.9054) acc1: 96.6146 (58.6917) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2099 data: 0.1177 max mem: 3532 +Test: [ 360/2387] eta: 0:07:53 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1535 (0.8847) acc1: 98.1771 (59.7653) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1944 data: 0.1003 max mem: 3532 +Test: [ 370/2387] eta: 0:07:48 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1593 (0.8651) acc1: 97.3958 (60.7789) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1802 data: 0.0880 max mem: 3532 +Test: [ 380/2387] eta: 0:07:43 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1582 (0.8466) acc1: 97.3958 (61.7461) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1870 data: 0.0928 max mem: 3532 +Test: [ 390/2387] eta: 0:07:40 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1590 (0.8290) acc1: 97.6562 (62.6625) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2011 data: 0.1070 max mem: 3532 +Test: [ 400/2387] eta: 0:07:35 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1591 (0.8123) acc1: 97.3958 (63.5287) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1949 data: 0.1024 max mem: 3532 +Test: [ 410/2387] eta: 0:07:32 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1613 (0.7965) acc1: 97.1354 (64.3514) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1935 data: 0.0955 max mem: 3532 +Test: [ 420/2387] eta: 0:07:29 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1603 (0.7814) acc1: 97.1354 (65.1320) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2128 data: 0.1137 max mem: 3532 +Test: [ 430/2387] eta: 0:07:24 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1544 (0.7663) acc1: 97.6562 (65.9108) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1919 data: 0.0957 max mem: 3532 +Test: [ 440/2387] eta: 0:07:20 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1091 (0.7512) acc1: 99.4792 (66.6743) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1821 data: 0.0875 max mem: 3533 +Test: [ 450/2387] eta: 0:07:16 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0976 (0.7367) acc1: 100.0000 (67.4133) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1879 data: 0.0904 max mem: 3533 +Test: [ 460/2387] eta: 0:07:12 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0973 (0.7229) acc1: 100.0000 (68.1201) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1854 data: 0.0881 max mem: 3533 +Test: [ 470/2387] eta: 0:07:09 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0972 (0.7096) acc1: 100.0000 (68.7970) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0972 max mem: 3533 +Test: [ 480/2387] eta: 0:07:05 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0975 (0.6969) acc1: 100.0000 (69.4457) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1879 data: 0.0960 max mem: 3533 +Test: [ 490/2387] eta: 0:07:01 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0980 (0.6847) acc1: 100.0000 (70.0680) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1827 data: 0.0910 max mem: 3533 +Test: [ 500/2387] eta: 0:06:57 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0982 (0.6730) acc1: 100.0000 (70.6649) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1771 data: 0.0862 max mem: 3533 +Test: [ 510/2387] eta: 0:06:54 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0973 (0.6617) acc1: 100.0000 (71.2390) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1855 data: 0.0941 max mem: 3533 +Test: [ 520/2387] eta: 0:06:50 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0955 (0.6508) acc1: 100.0000 (71.7910) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1836 data: 0.0924 max mem: 3533 +Test: [ 530/2387] eta: 0:06:46 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1013 (0.6405) acc1: 100.0000 (72.3218) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1791 data: 0.0880 max mem: 3533 +Test: [ 540/2387] eta: 0:06:43 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1050 (0.6306) acc1: 100.0000 (72.8334) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1764 data: 0.0834 max mem: 3533 +Test: [ 550/2387] eta: 0:06:39 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1040 (0.6211) acc1: 100.0000 (73.3264) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1754 data: 0.0829 max mem: 3533 +Test: [ 560/2387] eta: 0:06:36 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1036 (0.6119) acc1: 100.0000 (73.8014) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1771 data: 0.0866 max mem: 3533 +Test: [ 570/2387] eta: 0:06:32 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1032 (0.6030) acc1: 100.0000 (74.2603) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1723 data: 0.0805 max mem: 3533 +Test: [ 580/2387] eta: 0:06:29 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1032 (0.5944) acc1: 100.0000 (74.7033) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1763 data: 0.0835 max mem: 3533 +Test: [ 590/2387] eta: 0:06:25 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1046 (0.5861) acc1: 100.0000 (75.1304) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1744 data: 0.0823 max mem: 3533 +Test: [ 600/2387] eta: 0:06:22 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0999 (0.5779) acc1: 100.0000 (75.5434) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1682 data: 0.0762 max mem: 3533 +Test: [ 610/2387] eta: 0:06:18 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1191 (0.5714) acc1: 98.6979 (75.8968) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1712 data: 0.0769 max mem: 3533 +Test: [ 620/2387] eta: 0:06:15 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1749 (0.5650) acc1: 97.3958 (76.2434) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1760 data: 0.0804 max mem: 3533 +Test: [ 630/2387] eta: 0:06:12 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1761 (0.5588) acc1: 97.3958 (76.5802) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1807 data: 0.0871 max mem: 3533 +Test: [ 640/2387] eta: 0:06:09 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1767 (0.5529) acc1: 97.3958 (76.9005) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1838 data: 0.0902 max mem: 3533 +Test: [ 650/2387] eta: 0:06:07 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1753 (0.5471) acc1: 97.6562 (77.2201) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1946 data: 0.1013 max mem: 3533 +Test: [ 660/2387] eta: 0:06:04 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1778 (0.5416) acc1: 97.3958 (77.5179) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1899 data: 0.0966 max mem: 3533 +Test: [ 670/2387] eta: 0:06:01 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1782 (0.5362) acc1: 96.8750 (77.8106) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1744 data: 0.0821 max mem: 3534 +Test: [ 680/2387] eta: 0:05:59 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1714 (0.5301) acc1: 97.6562 (78.1239) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0993 max mem: 3534 +Test: [ 690/2387] eta: 0:05:56 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1789 (0.5302) acc1: 97.1354 (77.9972) acc5: 100.0000 (100.0000) acc1_neg: 55.729 acc5_neg: 100.000 time: 0.1964 data: 0.1041 max mem: 3534 +Test: [ 700/2387] eta: 0:05:54 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.6980 (0.5326) acc1: 58.8542 (77.7197) acc5: 100.0000 (100.0000) acc1_neg: 60.677 acc5_neg: 100.000 time: 0.1899 data: 0.0967 max mem: 3534 +Test: [ 710/2387] eta: 0:05:51 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.7119 (0.5352) acc1: 58.0729 (77.4393) acc5: 100.0000 (100.0000) acc1_neg: 56.771 acc5_neg: 100.000 time: 0.1934 data: 0.1008 max mem: 3534 +Test: [ 720/2387] eta: 0:05:48 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.7119 (0.5375) acc1: 57.8125 (77.1769) acc5: 100.0000 (100.0000) acc1_neg: 55.990 acc5_neg: 100.000 time: 0.1853 data: 0.0929 max mem: 3534 +Test: [ 730/2387] eta: 0:05:45 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.7041 (0.5400) acc1: 58.0729 (76.9052) acc5: 100.0000 (100.0000) acc1_neg: 57.812 acc5_neg: 100.000 time: 0.1786 data: 0.0857 max mem: 3534 +Test: [ 740/2387] eta: 0:05:43 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.7107 (0.5424) acc1: 57.8125 (76.6370) acc5: 100.0000 (100.0000) acc1_neg: 56.771 acc5_neg: 100.000 time: 0.1946 data: 0.0980 max mem: 3534 +Test: [ 750/2387] eta: 0:05:41 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.7100 (0.5445) acc1: 58.0729 (76.3981) acc5: 100.0000 (100.0000) acc1_neg: 61.979 acc5_neg: 100.000 time: 0.2065 data: 0.1098 max mem: 3534 +Test: [ 760/2387] eta: 0:05:39 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.6973 (0.5446) acc1: 58.0729 (76.2911) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.2055 data: 0.1122 max mem: 3534 +Test: [ 770/2387] eta: 0:05:36 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.3112 (0.5409) acc1: 86.1979 (76.4716) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1936 data: 0.1006 max mem: 3534 +Test: [ 780/2387] eta: 0:05:34 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.2818 (0.5378) acc1: 88.2812 (76.6265) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1853 data: 0.0924 max mem: 3534 +Test: [ 790/2387] eta: 0:05:31 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.2680 (0.5343) acc1: 88.5417 (76.7834) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.1899 data: 0.0972 max mem: 3534 +Test: [ 800/2387] eta: 0:05:29 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.2415 (0.5305) acc1: 91.9271 (76.9799) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.1938 data: 0.1001 max mem: 3534 +Test: [ 810/2387] eta: 0:05:27 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1808 (0.5262) acc1: 95.8333 (77.2249) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1970 data: 0.1026 max mem: 3534 +Test: [ 820/2387] eta: 0:05:25 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1699 (0.5221) acc1: 96.6146 (77.4367) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.1962 data: 0.1032 max mem: 3534 +Test: [ 830/2387] eta: 0:05:22 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1994 (0.5188) acc1: 92.9688 (77.5882) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1977 data: 0.1062 max mem: 3534 +Test: [ 840/2387] eta: 0:05:20 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.2293 (0.5153) acc1: 91.1458 (77.7770) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1974 data: 0.1065 max mem: 3534 +Test: [ 850/2387] eta: 0:05:18 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.2207 (0.5121) acc1: 94.0104 (77.9622) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.1975 data: 0.1066 max mem: 3534 +Test: [ 860/2387] eta: 0:05:16 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.2070 (0.5086) acc1: 94.0104 (78.1453) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2061 data: 0.1156 max mem: 3534 +Test: [ 870/2387] eta: 0:05:14 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1974 (0.5054) acc1: 94.7917 (78.3199) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2033 data: 0.1122 max mem: 3534 +Test: [ 880/2387] eta: 0:05:12 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.2230 (0.5024) acc1: 92.9688 (78.4862) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.2033 data: 0.1115 max mem: 3534 +Test: [ 890/2387] eta: 0:05:09 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1716 (0.4984) acc1: 95.5729 (78.6964) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2036 data: 0.1122 max mem: 3534 +Test: [ 900/2387] eta: 0:05:07 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1563 (0.4983) acc1: 92.9688 (78.6614) acc5: 100.0000 (100.0000) acc1_neg: 70.052 acc5_neg: 100.000 time: 0.1977 data: 0.1073 max mem: 3535 +Test: [ 910/2387] eta: 0:05:05 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.5591 (0.4989) acc1: 70.0521 (78.5827) acc5: 100.0000 (100.0000) acc1_neg: 68.750 acc5_neg: 100.000 time: 0.2079 data: 0.1163 max mem: 3535 +Test: [ 920/2387] eta: 0:05:03 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.5598 (0.4996) acc1: 70.0521 (78.4951) acc5: 100.0000 (100.0000) acc1_neg: 69.531 acc5_neg: 100.000 time: 0.1986 data: 0.1060 max mem: 3535 +Test: [ 930/2387] eta: 0:05:00 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.5637 (0.5003) acc1: 70.0521 (78.4061) acc5: 100.0000 (100.0000) acc1_neg: 69.271 acc5_neg: 100.000 time: 0.1893 data: 0.0980 max mem: 3535 +Test: [ 940/2387] eta: 0:04:58 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.5756 (0.5010) acc1: 70.5729 (78.3273) acc5: 100.0000 (100.0000) acc1_neg: 66.406 acc5_neg: 100.000 time: 0.1814 data: 0.0909 max mem: 3535 +Test: [ 950/2387] eta: 0:04:56 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.5575 (0.5017) acc1: 70.3125 (78.2414) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.1832 data: 0.0929 max mem: 3535 +Test: [ 960/2387] eta: 0:04:53 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.5481 (0.5022) acc1: 70.5729 (78.1711) acc5: 100.0000 (100.0000) acc1_neg: 72.396 acc5_neg: 100.000 time: 0.1951 data: 0.1044 max mem: 3535 +Test: [ 970/2387] eta: 0:04:51 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.5518 (0.5028) acc1: 69.5312 (78.0880) acc5: 100.0000 (100.0000) acc1_neg: 69.531 acc5_neg: 100.000 time: 0.1947 data: 0.1029 max mem: 3535 +Test: [ 980/2387] eta: 0:04:49 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.5462 (0.5018) acc1: 71.8750 (78.1266) acc5: 100.0000 (100.0000) acc1_neg: 59.635 acc5_neg: 100.000 time: 0.1997 data: 0.1058 max mem: 3535 +Test: [ 990/2387] eta: 0:04:47 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.2023 (0.4995) acc1: 91.6667 (78.2330) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1993 data: 0.1063 max mem: 3535 +Test: [1000/2387] eta: 0:04:45 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1102 (0.4959) acc1: 99.4792 (78.4273) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1988 data: 0.1078 max mem: 3535 +Test: [1010/2387] eta: 0:04:43 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1102 (0.4922) acc1: 99.4792 (78.6257) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2015 data: 0.1104 max mem: 3535 +Test: [1020/2387] eta: 0:04:41 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1079 (0.4888) acc1: 99.4792 (78.8078) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2096 data: 0.1177 max mem: 3535 +Test: [1030/2387] eta: 0:04:39 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1079 (0.4852) acc1: 99.4792 (79.0037) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2162 data: 0.1240 max mem: 3535 +Test: [1040/2387] eta: 0:04:37 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1008 (0.4814) acc1: 99.7396 (79.2012) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2132 data: 0.1195 max mem: 3535 +Test: [1050/2387] eta: 0:04:35 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0863 (0.4777) acc1: 100.0000 (79.3981) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2145 data: 0.1190 max mem: 3535 +Test: [1060/2387] eta: 0:04:33 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0863 (0.4741) acc1: 100.0000 (79.5883) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2064 data: 0.1133 max mem: 3535 +Test: [1070/2387] eta: 0:04:31 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0911 (0.4706) acc1: 99.7396 (79.7711) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1986 data: 0.1073 max mem: 3535 +Test: [1080/2387] eta: 0:04:29 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1075 (0.4677) acc1: 98.1771 (79.9221) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2060 data: 0.1148 max mem: 3535 +Test: [1090/2387] eta: 0:04:27 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1693 (0.4652) acc1: 96.3542 (80.0670) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2161 data: 0.1231 max mem: 3535 +Test: [1100/2387] eta: 0:04:25 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.2050 (0.4632) acc1: 95.8333 (80.1835) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2074 data: 0.1145 max mem: 3535 +Test: [1110/2387] eta: 0:04:22 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.2095 (0.4614) acc1: 93.7500 (80.2644) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1923 data: 0.1003 max mem: 3535 +Test: [1120/2387] eta: 0:04:20 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.2168 (0.4592) acc1: 93.7500 (80.3898) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2014 data: 0.1076 max mem: 3536 +Test: [1130/2387] eta: 0:04:18 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1671 (0.4567) acc1: 96.8750 (80.5355) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2099 data: 0.1165 max mem: 3536 +Test: [1140/2387] eta: 0:04:16 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1444 (0.4539) acc1: 98.4375 (80.6881) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1809 data: 0.0877 max mem: 3536 +Test: [1150/2387] eta: 0:04:13 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1558 (0.4516) acc1: 97.9167 (80.8172) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1699 data: 0.0774 max mem: 3536 +Test: [1160/2387] eta: 0:04:11 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1809 (0.4495) acc1: 97.1354 (80.9335) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1858 data: 0.0933 max mem: 3536 +Test: [1170/2387] eta: 0:04:09 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.2004 (0.4473) acc1: 94.7917 (81.0539) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1868 data: 0.0939 max mem: 3536 +Test: [1180/2387] eta: 0:04:07 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1855 (0.4453) acc1: 94.7917 (81.1603) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1837 data: 0.0923 max mem: 3536 +Test: [1190/2387] eta: 0:04:04 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1884 (0.4435) acc1: 94.7917 (81.2609) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1835 data: 0.0906 max mem: 3536 +Test: [1200/2387] eta: 0:04:02 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1699 (0.4410) acc1: 96.8750 (81.3990) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1891 data: 0.0935 max mem: 3536 +Test: [1210/2387] eta: 0:04:00 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1275 (0.4385) acc1: 99.2188 (81.5453) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1813 data: 0.0872 max mem: 3536 +Test: [1220/2387] eta: 0:03:57 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1160 (0.4358) acc1: 99.4792 (81.6834) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1717 data: 0.0808 max mem: 3536 +Test: [1230/2387] eta: 0:03:55 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1064 (0.4331) acc1: 99.4792 (81.8263) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0847 max mem: 3536 +Test: [1240/2387] eta: 0:03:53 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0980 (0.4304) acc1: 99.4792 (81.9683) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1849 data: 0.0944 max mem: 3536 +Test: [1250/2387] eta: 0:03:51 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0947 (0.4277) acc1: 99.4792 (82.1079) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1900 data: 0.0976 max mem: 3536 +Test: [1260/2387] eta: 0:03:49 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0976 (0.4251) acc1: 99.4792 (82.2407) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1880 data: 0.0951 max mem: 3536 +Test: [1270/2387] eta: 0:03:46 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0980 (0.4225) acc1: 99.2188 (82.3734) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1840 data: 0.0928 max mem: 3536 +Test: [1280/2387] eta: 0:03:44 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0887 (0.4200) acc1: 100.0000 (82.5082) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1755 data: 0.0847 max mem: 3536 +Test: [1290/2387] eta: 0:03:42 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0986 (0.4177) acc1: 99.7396 (82.6235) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1750 data: 0.0825 max mem: 3536 +Test: [1300/2387] eta: 0:03:40 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1089 (0.4153) acc1: 98.6979 (82.7460) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1712 data: 0.0781 max mem: 3536 +Test: [1310/2387] eta: 0:03:37 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1073 (0.4131) acc1: 98.9583 (82.8653) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1715 data: 0.0803 max mem: 3536 +Test: [1320/2387] eta: 0:03:35 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1073 (0.4109) acc1: 98.9583 (82.9797) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1756 data: 0.0842 max mem: 3536 +Test: [1330/2387] eta: 0:03:33 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1166 (0.4089) acc1: 98.4375 (83.0829) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1686 data: 0.0763 max mem: 3536 +Test: [1340/2387] eta: 0:03:31 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1688 (0.4074) acc1: 94.7917 (83.1558) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1703 data: 0.0787 max mem: 3536 +Test: [1350/2387] eta: 0:03:28 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0821 (0.4050) acc1: 100.0000 (83.2805) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1716 data: 0.0794 max mem: 3537 +Test: [1360/2387] eta: 0:03:26 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0761 (0.4025) acc1: 100.0000 (83.4034) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1735 data: 0.0811 max mem: 3537 +Test: [1370/2387] eta: 0:03:24 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0759 (0.4002) acc1: 100.0000 (83.5239) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1780 data: 0.0868 max mem: 3537 +Test: [1380/2387] eta: 0:03:22 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0764 (0.3979) acc1: 100.0000 (83.6428) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1853 data: 0.0929 max mem: 3537 +Test: [1390/2387] eta: 0:03:19 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0757 (0.3956) acc1: 100.0000 (83.7604) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1692 data: 0.0770 max mem: 3537 +Test: [1400/2387] eta: 0:03:17 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0710 (0.3932) acc1: 100.0000 (83.8763) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1557 data: 0.0652 max mem: 3537 +Test: [1410/2387] eta: 0:03:15 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0725 (0.3918) acc1: 100.0000 (83.9555) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.1813 data: 0.0903 max mem: 3537 +Test: [1420/2387] eta: 0:03:13 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.2294 (0.3910) acc1: 92.7083 (83.9883) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1821 data: 0.0906 max mem: 3537 +Test: [1430/2387] eta: 0:03:11 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.2493 (0.3899) acc1: 91.9271 (84.0487) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1686 data: 0.0770 max mem: 3537 +Test: [1440/2387] eta: 0:03:09 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1617 (0.3882) acc1: 96.8750 (84.1401) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1699 data: 0.0770 max mem: 3537 +Test: [1450/2387] eta: 0:03:06 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1454 (0.3868) acc1: 97.6562 (84.2156) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1620 data: 0.0690 max mem: 3537 +Test: [1460/2387] eta: 0:03:04 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1353 (0.3853) acc1: 97.6562 (84.3023) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1600 data: 0.0692 max mem: 3537 +Test: [1470/2387] eta: 0:03:02 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1393 (0.3837) acc1: 96.8750 (84.3877) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1679 data: 0.0773 max mem: 3537 +Test: [1480/2387] eta: 0:03:00 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1393 (0.3822) acc1: 96.8750 (84.4714) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1673 data: 0.0762 max mem: 3537 +Test: [1490/2387] eta: 0:02:57 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1392 (0.3806) acc1: 98.1771 (84.5584) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1654 data: 0.0745 max mem: 3537 +Test: [1500/2387] eta: 0:02:55 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1386 (0.3790) acc1: 97.6562 (84.6401) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1679 data: 0.0767 max mem: 3537 +Test: [1510/2387] eta: 0:02:53 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1196 (0.3773) acc1: 98.6979 (84.7362) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1650 data: 0.0736 max mem: 3537 +Test: [1520/2387] eta: 0:02:51 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1464 (0.3762) acc1: 97.9167 (84.7946) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1745 data: 0.0808 max mem: 3537 +Test: [1530/2387] eta: 0:02:49 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.2236 (0.3754) acc1: 92.7083 (84.8450) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1795 data: 0.0849 max mem: 3537 +Test: [1540/2387] eta: 0:02:47 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.2237 (0.3744) acc1: 92.7083 (84.8975) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1716 data: 0.0795 max mem: 3537 +Test: [1550/2387] eta: 0:02:45 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.2169 (0.3734) acc1: 92.7083 (84.9506) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1707 data: 0.0797 max mem: 3537 +Test: [1560/2387] eta: 0:02:43 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.2142 (0.3725) acc1: 93.2292 (85.0039) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1706 data: 0.0783 max mem: 3537 +Test: [1570/2387] eta: 0:02:41 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.2005 (0.3712) acc1: 95.5729 (85.0803) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1738 data: 0.0804 max mem: 3537 +Test: [1580/2387] eta: 0:02:38 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1709 (0.3701) acc1: 95.8333 (85.1416) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.1725 data: 0.0791 max mem: 3538 +Test: [1590/2387] eta: 0:02:36 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.2105 (0.3696) acc1: 94.0104 (85.1697) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1671 data: 0.0749 max mem: 3538 +Test: [1600/2387] eta: 0:02:34 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.2620 (0.3688) acc1: 90.8854 (85.2052) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1706 data: 0.0801 max mem: 3538 +Test: [1610/2387] eta: 0:02:32 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1861 (0.3676) acc1: 96.0938 (85.2692) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1798 data: 0.0893 max mem: 3538 +Test: [1620/2387] eta: 0:02:30 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1861 (0.3666) acc1: 96.0938 (85.3291) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1859 data: 0.0947 max mem: 3538 +Test: [1630/2387] eta: 0:02:28 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.2108 (0.3657) acc1: 94.7917 (85.3779) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.1814 data: 0.0887 max mem: 3538 +Test: [1640/2387] eta: 0:02:26 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.2309 (0.3648) acc1: 91.6667 (85.4195) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1788 data: 0.0857 max mem: 3538 +Test: [1650/2387] eta: 0:02:24 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1798 (0.3638) acc1: 95.5729 (85.4772) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1839 data: 0.0911 max mem: 3538 +Test: [1660/2387] eta: 0:02:22 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1812 (0.3630) acc1: 94.2708 (85.5172) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1651 data: 0.0726 max mem: 3538 +Test: [1670/2387] eta: 0:02:20 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.2140 (0.3620) acc1: 92.4479 (85.5658) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1694 data: 0.0776 max mem: 3538 +Test: [1680/2387] eta: 0:02:18 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1784 (0.3611) acc1: 94.0104 (85.6140) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1996 data: 0.1054 max mem: 3538 +Test: [1690/2387] eta: 0:02:16 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1651 (0.3601) acc1: 95.5729 (85.6654) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2052 data: 0.1117 max mem: 3538 +Test: [1700/2387] eta: 0:02:14 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1631 (0.3590) acc1: 95.8333 (85.7235) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2048 data: 0.1133 max mem: 3538 +Test: [1710/2387] eta: 0:02:12 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1566 (0.3578) acc1: 96.8750 (85.7893) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1937 data: 0.1001 max mem: 3538 +Test: [1720/2387] eta: 0:02:10 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1371 (0.3565) acc1: 97.6562 (85.8587) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1833 data: 0.0881 max mem: 3538 +Test: [1730/2387] eta: 0:02:08 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1264 (0.3552) acc1: 97.6562 (85.9262) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1810 data: 0.0850 max mem: 3538 +Test: [1740/2387] eta: 0:02:06 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1190 (0.3539) acc1: 98.4375 (85.9954) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1860 data: 0.0920 max mem: 3538 +Test: [1750/2387] eta: 0:02:04 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1265 (0.3527) acc1: 98.4375 (86.0603) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1882 data: 0.0961 max mem: 3538 +Test: [1760/2387] eta: 0:02:02 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1265 (0.3514) acc1: 97.9167 (86.1256) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1757 data: 0.0841 max mem: 3538 +Test: [1770/2387] eta: 0:02:00 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1312 (0.3503) acc1: 97.9167 (86.1860) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1789 data: 0.0862 max mem: 3538 +Test: [1780/2387] eta: 0:01:58 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1613 (0.3493) acc1: 96.3542 (86.2365) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1720 data: 0.0783 max mem: 3538 +Test: [1790/2387] eta: 0:01:56 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1572 (0.3482) acc1: 96.6146 (86.3009) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1767 data: 0.0834 max mem: 3538 +Test: [1800/2387] eta: 0:01:54 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1482 (0.3471) acc1: 97.3958 (86.3597) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1796 data: 0.0870 max mem: 3538 +Test: [1810/2387] eta: 0:01:52 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1429 (0.3459) acc1: 97.9167 (86.4245) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1655 data: 0.0738 max mem: 3539 +Test: [1820/2387] eta: 0:01:50 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1138 (0.3446) acc1: 98.9583 (86.4905) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1799 data: 0.0885 max mem: 3539 +Test: [1830/2387] eta: 0:01:48 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1017 (0.3434) acc1: 99.2188 (86.5562) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1882 data: 0.0965 max mem: 3539 +Test: [1840/2387] eta: 0:01:46 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0910 (0.3420) acc1: 100.0000 (86.6292) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1861 data: 0.0939 max mem: 3539 +Test: [1850/2387] eta: 0:01:44 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0904 (0.3407) acc1: 100.0000 (86.7014) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1881 data: 0.0960 max mem: 3539 +Test: [1860/2387] eta: 0:01:42 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0904 (0.3393) acc1: 100.0000 (86.7728) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2050 data: 0.1138 max mem: 3539 +Test: [1870/2387] eta: 0:01:40 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0905 (0.3380) acc1: 100.0000 (86.8435) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1858 data: 0.0945 max mem: 3539 +Test: [1880/2387] eta: 0:01:38 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0894 (0.3367) acc1: 100.0000 (86.9134) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1848 data: 0.0921 max mem: 3539 +Test: [1890/2387] eta: 0:01:36 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0898 (0.3354) acc1: 100.0000 (86.9826) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1902 data: 0.0945 max mem: 3539 +Test: [1900/2387] eta: 0:01:34 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0910 (0.3341) acc1: 100.0000 (87.0511) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1823 data: 0.0875 max mem: 3539 +Test: [1910/2387] eta: 0:01:32 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0938 (0.3329) acc1: 100.0000 (87.1148) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1947 data: 0.1031 max mem: 3539 +Test: [1920/2387] eta: 0:01:30 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1030 (0.3323) acc1: 100.0000 (87.1545) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1893 data: 0.0977 max mem: 3539 +Test: [1930/2387] eta: 0:01:28 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1786 (0.3320) acc1: 97.1354 (87.1689) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1911 data: 0.0983 max mem: 3539 +Test: [1940/2387] eta: 0:01:26 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.2052 (0.3315) acc1: 96.6146 (87.1803) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1910 data: 0.0964 max mem: 3539 +Test: [1950/2387] eta: 0:01:25 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1939 (0.3310) acc1: 95.0521 (87.2070) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1879 data: 0.0914 max mem: 3539 +Test: [1960/2387] eta: 0:01:23 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1939 (0.3303) acc1: 95.0521 (87.2494) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1923 data: 0.0974 max mem: 3539 +Test: [1970/2387] eta: 0:01:21 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1988 (0.3302) acc1: 94.7917 (87.2503) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.1949 data: 0.1024 max mem: 3539 +Test: [1980/2387] eta: 0:01:19 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1941 (0.3293) acc1: 93.7500 (87.2916) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1911 data: 0.0983 max mem: 3539 +Test: [1990/2387] eta: 0:01:17 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0894 (0.3281) acc1: 99.7396 (87.3543) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1927 data: 0.1013 max mem: 3539 +Test: [2000/2387] eta: 0:01:15 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0902 (0.3270) acc1: 99.7396 (87.4166) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1890 data: 0.0984 max mem: 3539 +Test: [2010/2387] eta: 0:01:13 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0928 (0.3258) acc1: 99.7396 (87.4785) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1845 data: 0.0935 max mem: 3539 +Test: [2020/2387] eta: 0:01:11 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0914 (0.3246) acc1: 99.7396 (87.5394) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1815 data: 0.0887 max mem: 3539 +Test: [2030/2387] eta: 0:01:09 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0914 (0.3235) acc1: 99.7396 (87.6001) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1751 data: 0.0823 max mem: 3540 +Test: [2040/2387] eta: 0:01:07 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0911 (0.3224) acc1: 99.7396 (87.6601) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1816 data: 0.0907 max mem: 3540 +Test: [2050/2387] eta: 0:01:05 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0879 (0.3212) acc1: 99.7396 (87.7195) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1821 data: 0.0909 max mem: 3540 +Test: [2060/2387] eta: 0:01:03 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0851 (0.3201) acc1: 100.0000 (87.7789) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1685 data: 0.0772 max mem: 3540 +Test: [2070/2387] eta: 0:01:01 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0839 (0.3189) acc1: 100.0000 (87.8379) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1711 data: 0.0791 max mem: 3540 +Test: [2080/2387] eta: 0:00:59 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0837 (0.3178) acc1: 100.0000 (87.8962) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1701 data: 0.0778 max mem: 3540 +Test: [2090/2387] eta: 0:00:57 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0831 (0.3167) acc1: 100.0000 (87.9540) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1689 data: 0.0771 max mem: 3540 +Test: [2100/2387] eta: 0:00:55 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0826 (0.3156) acc1: 100.0000 (88.0113) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1734 data: 0.0812 max mem: 3540 +Test: [2110/2387] eta: 0:00:53 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0840 (0.3145) acc1: 100.0000 (88.0681) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1660 data: 0.0732 max mem: 3540 +Test: [2120/2387] eta: 0:00:51 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0855 (0.3142) acc1: 100.0000 (88.0838) acc5: 100.0000 (100.0000) acc1_neg: 70.573 acc5_neg: 100.000 time: 0.1638 data: 0.0701 max mem: 3540 +Test: [2130/2387] eta: 0:00:49 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.2248 (0.3139) acc1: 93.7500 (88.1081) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1648 data: 0.0720 max mem: 3540 +Test: [2140/2387] eta: 0:00:47 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1969 (0.3135) acc1: 95.0521 (88.1271) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1702 data: 0.0780 max mem: 3540 +Test: [2150/2387] eta: 0:00:45 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1913 (0.3130) acc1: 93.7500 (88.1596) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1749 data: 0.0831 max mem: 3540 +Test: [2160/2387] eta: 0:00:43 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1671 (0.3123) acc1: 98.1771 (88.2012) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1723 data: 0.0804 max mem: 3540 +Test: [2170/2387] eta: 0:00:41 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1485 (0.3118) acc1: 98.1771 (88.2315) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.1662 data: 0.0745 max mem: 3540 +Test: [2180/2387] eta: 0:00:39 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.2322 (0.3125) acc1: 90.8854 (88.1685) acc5: 100.0000 (100.0000) acc1_neg: 52.865 acc5_neg: 100.000 time: 0.1650 data: 0.0734 max mem: 3540 +Test: [2190/2387] eta: 0:00:37 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1682 (0.3120) acc1: 90.8854 (88.1934) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1741 data: 0.0818 max mem: 3540 +Test: [2200/2387] eta: 0:00:35 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1323 (0.3114) acc1: 98.4375 (88.2330) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1707 data: 0.0786 max mem: 3540 +Test: [2210/2387] eta: 0:00:34 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1767 (0.3108) acc1: 96.6146 (88.2617) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1781 data: 0.0857 max mem: 3540 +Test: [2220/2387] eta: 0:00:32 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1778 (0.3103) acc1: 95.3125 (88.2932) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1779 data: 0.0858 max mem: 3540 +Test: [2230/2387] eta: 0:00:30 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1523 (0.3096) acc1: 97.6562 (88.3327) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1664 data: 0.0742 max mem: 3540 +Test: [2240/2387] eta: 0:00:28 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1458 (0.3091) acc1: 97.6562 (88.3649) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1783 data: 0.0856 max mem: 3540 +Test: [2250/2387] eta: 0:00:26 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1363 (0.3084) acc1: 97.3958 (88.3986) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1904 data: 0.0990 max mem: 3540 +Test: [2260/2387] eta: 0:00:24 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1170 (0.3076) acc1: 99.4792 (88.4472) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1949 data: 0.1019 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1165 (0.3068) acc1: 99.4792 (88.4924) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2033 data: 0.1094 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1122 (0.3062) acc1: 98.9583 (88.5188) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2113 data: 0.1203 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1001 (0.3053) acc1: 99.7396 (88.5671) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1895 data: 0.0972 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1022 (0.3044) acc1: 100.0000 (88.6131) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1609 data: 0.0687 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1034 (0.3036) acc1: 99.7396 (88.6547) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1437 data: 0.0531 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1273 (0.3031) acc1: 97.6562 (88.6834) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1321 data: 0.0412 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1771 (0.3027) acc1: 97.1354 (88.7091) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1273 data: 0.0370 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1870 (0.3026) acc1: 96.8750 (88.7104) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1239 data: 0.0337 max mem: 3541 +Test: [2350/2387] eta: 0:00:07 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1870 (0.3024) acc1: 94.7917 (88.7152) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1209 data: 0.0310 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1697 (0.3020) acc1: 97.9167 (88.7426) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1229 data: 0.0332 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1862 (0.3016) acc1: 97.3958 (88.7653) acc5: 100.0000 (100.0000) acc1_neg: 72.917 acc5_neg: 100.000 time: 0.1293 data: 0.0397 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1647 (0.3012) acc1: 95.3125 (88.7836) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.1257 data: 0.0364 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1627 (0.3007) acc1: 97.3958 (88.8011) acc5: 100.0000 (100.0000) acc1_neg: 97.674 acc5_neg: 100.000 time: 0.1161 data: 0.0295 max mem: 3541 +Test: Total time: 0:07:33 (0.1898 s / it) +* Overall Acc@1 88.801 Acc@5 100.000 + target == real Acc@1 11.427 Acc@5 100.000 + target == fake Acc@1 94.235 Acc@5 100.000 + loss 0.301 + PR-AUC 0.9686 ROC-AUC 0.6817 EER 37.18% (thr=0.8916) +Accuracy of the network on the 916267 test images: 88.8% +[rank0]:[W726 06:09:08.121945648 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 06:09:28.454000 3165097 torch/distributed/run.py:766] +W0726 06:09:28.454000 3165097 torch/distributed/run.py:766] ***************************************** +W0726 06:09:28.454000 3165097 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:09:28.454000 3165097 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:09:56.335896235 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:09:56.593998486 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:09:56.727099689 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 06:09:56.912564512 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:09:56.026267744 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 6): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:09:56.052919184 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W726 06:09:56.052930947 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:09:56.141817016 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n5_seed9_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed9_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 0:50:08 acc1_pos: 18.750 acc5_pos: 100.000 loss: 1.4829 (1.4829) acc1: 18.7500 (18.7500) acc5: 100.0000 (100.0000) time: 5.8199 data: 3.6297 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/517] eta: 0:05:20 acc1_pos: 13.021 acc5_pos: 100.000 loss: 1.4829 (1.4937) acc1: 15.8854 (18.8447) acc5: 100.0000 (100.0000) time: 0.6324 data: 0.3522 max mem: 3531 +Test: [ 20/517] eta: 0:03:05 acc1_pos: 4.948 acc5_pos: 100.000 loss: 1.4528 (1.5081) acc1: 15.8854 (17.9563) acc5: 100.0000 (100.0000) time: 0.1017 data: 0.0124 max mem: 3531 +Test: [ 30/517] eta: 0:02:27 acc1_pos: 3.906 acc5_pos: 100.000 loss: 1.5183 (1.5488) acc1: 15.3646 (16.3138) acc5: 100.0000 (100.0000) time: 0.1223 data: 0.0317 max mem: 3531 +Test: [ 40/517] eta: 0:02:20 acc1_pos: 10.156 acc5_pos: 100.000 loss: 1.5519 (1.5459) acc1: 14.0625 (16.3237) acc5: 100.0000 (100.0000) time: 0.2120 data: 0.1198 max mem: 3531 +Test: [ 50/517] eta: 0:02:12 acc1_pos: 25.229 acc5_pos: 100.000 loss: 1.6051 (1.4963) acc1: 16.1458 (19.9959) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.2513 data: 0.1601 max mem: 3531 +Test: [ 60/517] eta: 0:02:07 acc1_pos: 25.229 acc5_pos: 100.000 loss: 0.2837 (1.2899) acc1: 90.1042 (32.1551) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2446 data: 0.1494 max mem: 3531 +Test: [ 70/517] eta: 0:02:03 acc1_pos: 25.229 acc5_pos: 100.000 loss: 0.1920 (1.1331) acc1: 98.6979 (41.5053) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2627 data: 0.1661 max mem: 3531 +Test: [ 80/517] eta: 0:02:00 acc1_pos: 25.229 acc5_pos: 100.000 loss: 0.1751 (1.0149) acc1: 99.2188 (48.6304) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2649 data: 0.1722 max mem: 3531 +Test: [ 90/517] eta: 0:01:56 acc1_pos: 25.229 acc5_pos: 100.000 loss: 0.2371 (0.9386) acc1: 94.7917 (52.8503) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.2549 data: 0.1614 max mem: 3531 +Test: [100/517] eta: 0:01:53 acc1_pos: 25.229 acc5_pos: 100.000 loss: 0.3145 (0.8749) acc1: 88.5417 (56.5027) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2549 data: 0.1608 max mem: 3531 +Test: [110/517] eta: 0:01:49 acc1_pos: 25.229 acc5_pos: 100.000 loss: 0.2190 (0.8133) acc1: 96.8750 (60.2829) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2549 data: 0.1604 max mem: 3531 +Test: [120/517] eta: 0:01:46 acc1_pos: 25.229 acc5_pos: 100.000 loss: 0.2205 (0.7760) acc1: 90.6250 (62.1901) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.2607 data: 0.1669 max mem: 3531 +Test: [130/517] eta: 0:01:44 acc1_pos: 25.229 acc5_pos: 100.000 loss: 0.3338 (0.7405) acc1: 85.4167 (64.2056) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2699 data: 0.1751 max mem: 3531 +Test: [140/517] eta: 0:01:40 acc1_pos: 25.229 acc5_pos: 100.000 loss: 0.3206 (0.7104) acc1: 89.0625 (65.9501) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.2541 data: 0.1591 max mem: 3531 +Test: [150/517] eta: 0:01:37 acc1_pos: 25.229 acc5_pos: 100.000 loss: 0.2701 (0.6780) acc1: 91.4062 (67.8377) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2399 data: 0.1467 max mem: 3531 +Test: [160/517] eta: 0:01:33 acc1_pos: 25.229 acc5_pos: 100.000 loss: 0.1748 (0.6459) acc1: 97.1354 (69.7237) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2351 data: 0.1410 max mem: 3531 +Test: [170/517] eta: 0:01:30 acc1_pos: 25.229 acc5_pos: 100.000 loss: 0.1504 (0.6171) acc1: 99.2188 (71.4105) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2305 data: 0.1379 max mem: 3531 +Test: [180/517] eta: 0:01:27 acc1_pos: 25.229 acc5_pos: 100.000 loss: 0.1732 (0.5965) acc1: 97.3958 (72.5397) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.2352 data: 0.1451 max mem: 3531 +Test: [190/517] eta: 0:01:24 acc1_pos: 25.229 acc5_pos: 100.000 loss: 0.1184 (0.5709) acc1: 98.4375 (73.9774) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2443 data: 0.1535 max mem: 3531 +Test: [200/517] eta: 0:01:22 acc1_pos: 25.229 acc5_pos: 100.000 loss: 0.1117 (0.5552) acc1: 100.0000 (74.7875) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.2543 data: 0.1613 max mem: 3531 +Test: [210/517] eta: 0:01:19 acc1_pos: 25.229 acc5_pos: 100.000 loss: 0.3439 (0.5467) acc1: 83.5938 (75.1999) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.2499 data: 0.1553 max mem: 3532 +Test: [220/517] eta: 0:01:16 acc1_pos: 25.229 acc5_pos: 100.000 loss: 0.3409 (0.5365) acc1: 84.6354 (75.7176) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.2299 data: 0.1357 max mem: 3532 +Test: [230/517] eta: 0:01:13 acc1_pos: 25.229 acc5_pos: 100.000 loss: 0.2784 (0.5237) acc1: 89.5833 (76.5095) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2152 data: 0.1223 max mem: 3532 +Test: [240/517] eta: 0:01:10 acc1_pos: 25.229 acc5_pos: 100.000 loss: 0.1980 (0.5096) acc1: 96.6146 (77.3762) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2199 data: 0.1274 max mem: 3532 +Test: [250/517] eta: 0:01:07 acc1_pos: 25.229 acc5_pos: 100.000 loss: 0.1953 (0.4979) acc1: 97.6562 (78.1426) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2346 data: 0.1422 max mem: 3532 +Test: [260/517] eta: 0:01:05 acc1_pos: 25.229 acc5_pos: 100.000 loss: 0.2010 (0.4864) acc1: 97.6562 (78.9013) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2454 data: 0.1532 max mem: 3532 +Test: [270/517] eta: 0:01:02 acc1_pos: 25.229 acc5_pos: 100.000 loss: 0.2169 (0.4776) acc1: 96.8750 (79.5145) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2500 data: 0.1588 max mem: 3532 +Test: [280/517] eta: 0:00:59 acc1_pos: 25.229 acc5_pos: 100.000 loss: 0.2511 (0.4694) acc1: 94.2708 (80.0406) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2459 data: 0.1556 max mem: 3532 +Test: [290/517] eta: 0:00:57 acc1_pos: 25.229 acc5_pos: 100.000 loss: 0.2103 (0.4595) acc1: 94.2708 (80.6334) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2418 data: 0.1501 max mem: 3532 +Test: [300/517] eta: 0:00:54 acc1_pos: 25.229 acc5_pos: 100.000 loss: 0.1468 (0.4486) acc1: 98.4375 (81.2448) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2444 data: 0.1520 max mem: 3532 +Test: [310/517] eta: 0:00:52 acc1_pos: 25.229 acc5_pos: 100.000 loss: 0.1514 (0.4404) acc1: 98.4375 (81.7265) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2455 data: 0.1509 max mem: 3532 +Test: [320/517] eta: 0:00:49 acc1_pos: 25.229 acc5_pos: 100.000 loss: 0.1877 (0.4331) acc1: 95.8333 (82.1594) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2463 data: 0.1503 max mem: 3532 +Test: [330/517] eta: 0:00:47 acc1_pos: 25.229 acc5_pos: 100.000 loss: 0.1844 (0.4253) acc1: 96.6146 (82.6284) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2529 data: 0.1604 max mem: 3532 +Test: [340/517] eta: 0:00:44 acc1_pos: 25.229 acc5_pos: 100.000 loss: 0.1720 (0.4185) acc1: 97.9167 (83.0118) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.2475 data: 0.1564 max mem: 3532 +Test: [350/517] eta: 0:00:41 acc1_pos: 25.229 acc5_pos: 100.000 loss: 0.2174 (0.4145) acc1: 94.0104 (83.2376) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2416 data: 0.1484 max mem: 3532 +Test: [360/517] eta: 0:00:39 acc1_pos: 25.229 acc5_pos: 100.000 loss: 0.2366 (0.4096) acc1: 93.4896 (83.5375) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2462 data: 0.1525 max mem: 3532 +Test: [370/517] eta: 0:00:37 acc1_pos: 25.229 acc5_pos: 100.000 loss: 0.1833 (0.4030) acc1: 98.1771 (83.9538) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2688 data: 0.1752 max mem: 3532 +Test: [380/517] eta: 0:00:34 acc1_pos: 25.229 acc5_pos: 100.000 loss: 0.1695 (0.3976) acc1: 98.4375 (84.2923) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2645 data: 0.1705 max mem: 3532 +Test: [390/517] eta: 0:00:31 acc1_pos: 25.229 acc5_pos: 100.000 loss: 0.2009 (0.3925) acc1: 97.9167 (84.6427) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2337 data: 0.1395 max mem: 3532 +Test: [400/517] eta: 0:00:29 acc1_pos: 25.229 acc5_pos: 100.000 loss: 0.1746 (0.3868) acc1: 98.1771 (84.9627) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2401 data: 0.1447 max mem: 3532 +Test: [410/517] eta: 0:00:26 acc1_pos: 25.229 acc5_pos: 100.000 loss: 0.1711 (0.3844) acc1: 96.3542 (85.1277) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.2499 data: 0.1549 max mem: 3532 +Test: [420/517] eta: 0:00:24 acc1_pos: 25.229 acc5_pos: 100.000 loss: 0.1680 (0.3792) acc1: 99.2188 (85.4692) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2409 data: 0.1473 max mem: 3532 +Test: [430/517] eta: 0:00:21 acc1_pos: 25.229 acc5_pos: 100.000 loss: 0.1636 (0.3748) acc1: 99.4792 (85.7460) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.2502 data: 0.1583 max mem: 3532 +Test: [440/517] eta: 0:00:19 acc1_pos: 25.229 acc5_pos: 100.000 loss: 0.2700 (0.3741) acc1: 88.8021 (85.7332) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.2588 data: 0.1647 max mem: 3533 +Test: [450/517] eta: 0:00:16 acc1_pos: 25.229 acc5_pos: 100.000 loss: 0.2353 (0.3701) acc1: 89.5833 (85.9900) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2496 data: 0.1527 max mem: 3533 +Test: [460/517] eta: 0:00:14 acc1_pos: 25.229 acc5_pos: 100.000 loss: 0.2160 (0.3670) acc1: 96.6146 (86.1985) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.2443 data: 0.1470 max mem: 3533 +Test: [470/517] eta: 0:00:11 acc1_pos: 25.229 acc5_pos: 100.000 loss: 0.1744 (0.3628) acc1: 98.1771 (86.4600) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2405 data: 0.1423 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 25.229 acc5_pos: 100.000 loss: 0.1583 (0.3583) acc1: 98.9583 (86.7258) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2352 data: 0.1376 max mem: 3533 +Test: [490/517] eta: 0:00:06 acc1_pos: 25.229 acc5_pos: 100.000 loss: 0.1407 (0.3539) acc1: 99.2188 (86.9765) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2318 data: 0.1346 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 25.229 acc5_pos: 100.000 loss: 0.1949 (0.3515) acc1: 96.3542 (87.1294) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2410 data: 0.1473 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 25.229 acc5_pos: 100.000 loss: 0.2348 (0.3491) acc1: 94.7917 (87.2411) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2326 data: 0.1403 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 25.229 acc5_pos: 100.000 loss: 0.1013 (0.3461) acc1: 97.3958 (87.3792) acc5: 100.0000 (100.0000) acc1_neg: 99.647 acc5_neg: 100.000 time: 0.2411 data: 0.1380 max mem: 3533 +Test: Total time: 0:02:08 (0.2482 s / it) +* Overall Acc@1 87.379 Acc@5 100.000 + target == real Acc@1 16.306 Acc@5 100.000 + target == fake Acc@1 94.752 Acc@5 100.000 + loss 0.346 + PR-AUC 0.9420 ROC-AUC 0.6602 EER 38.62% (thr=0.8540) +Accuracy of the network on the 198427 test images: 87.4% +[rank0]:[W726 06:12:18.938052570 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 06:12:41.406000 3165674 torch/distributed/run.py:766] +W0726 06:12:41.406000 3165674 torch/distributed/run.py:766] ***************************************** +W0726 06:12:41.406000 3165674 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:12:41.406000 3165674 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 06:13:18.578367563 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:13:18.623885954 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:13:18.736923904 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:13:18.766185906 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:13:19.294919327 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:13:19.299542996 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:13:19.325509787 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:13:19.387501725 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n5_seed9_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed9_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:37:39 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.3861 (2.3861) acc1: 0.0000 (0.0000) acc5: 100.0000 (100.0000) time: 5.8095 data: 4.3042 max mem: 3531 +Test: [ 10/389] eta: 0:03:51 acc1_pos: 3.906 acc5_pos: 100.000 loss: 2.1070 (2.0959) acc1: 3.9062 (3.5985) acc5: 100.0000 (100.0000) time: 0.6099 data: 0.3916 max mem: 3531 +Test: [ 20/389] eta: 0:02:14 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.1070 (2.2003) acc1: 0.2604 (1.9717) acc5: 100.0000 (100.0000) time: 0.0931 data: 0.0004 max mem: 3531 +Test: [ 30/389] eta: 0:01:51 acc1_pos: 0.260 acc5_pos: 100.000 loss: 2.2587 (2.2591) acc1: 0.0000 (1.3441) acc5: 100.0000 (100.0000) time: 0.1440 data: 0.0494 max mem: 3531 +Test: [ 40/389] eta: 0:01:42 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.3826 (2.2992) acc1: 0.0000 (1.0480) acc5: 100.0000 (100.0000) time: 0.2213 data: 0.1293 max mem: 3531 +Test: [ 50/389] eta: 0:01:35 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.5164 (2.3545) acc1: 0.0000 (0.8425) acc5: 100.0000 (100.0000) time: 0.2359 data: 0.1452 max mem: 3531 +Test: [ 60/389] eta: 0:01:27 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.5747 (2.3515) acc1: 0.0000 (1.0032) acc5: 100.0000 (100.0000) time: 0.2085 data: 0.1173 max mem: 3531 +Test: [ 70/389] eta: 0:01:20 acc1_pos: 1.823 acc5_pos: 100.000 loss: 2.0556 (2.3020) acc1: 1.8229 (1.2837) acc5: 100.0000 (100.0000) time: 0.1846 data: 0.0933 max mem: 3531 +Test: [ 80/389] eta: 0:01:15 acc1_pos: 1.042 acc5_pos: 100.000 loss: 1.9103 (2.2483) acc1: 1.0417 (1.5078) acc5: 100.0000 (100.0000) time: 0.1699 data: 0.0784 max mem: 3531 +Test: [ 90/389] eta: 0:01:09 acc1_pos: 1.302 acc5_pos: 100.000 loss: 1.7868 (2.1914) acc1: 1.0417 (2.0175) acc5: 100.0000 (100.0000) time: 0.1642 data: 0.0726 max mem: 3531 +Test: [100/389] eta: 0:01:06 acc1_pos: 21.094 acc5_pos: 100.000 loss: 1.8604 (2.1726) acc1: 1.8229 (2.3128) acc5: 100.0000 (100.0000) time: 0.1771 data: 0.0845 max mem: 3531 +Test: [110/389] eta: 0:01:03 acc1_pos: 2.604 acc5_pos: 100.000 loss: 1.9528 (2.1458) acc1: 1.8229 (2.9514) acc5: 100.0000 (100.0000) time: 0.1915 data: 0.0970 max mem: 3531 +Test: [120/389] eta: 0:00:59 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.9528 (2.1399) acc1: 0.0000 (2.9507) acc5: 100.0000 (100.0000) time: 0.1748 data: 0.0790 max mem: 3531 +Test: [130/389] eta: 0:00:55 acc1_pos: 3.067 acc5_pos: 100.000 loss: 1.8072 (2.0206) acc1: 1.3021 (8.5301) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1452 data: 0.0517 max mem: 3531 +Test: [140/389] eta: 0:00:53 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.0999 (1.8855) acc1: 100.0000 (15.0118) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1765 data: 0.0844 max mem: 3531 +Test: [150/389] eta: 0:00:50 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.0946 (1.7678) acc1: 100.0000 (20.6367) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1871 data: 0.0953 max mem: 3531 +Test: [160/389] eta: 0:00:47 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.1002 (1.6687) acc1: 100.0000 (25.2216) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1579 data: 0.0661 max mem: 3531 +Test: [170/389] eta: 0:00:45 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.0850 (1.5768) acc1: 100.0000 (29.5839) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1748 data: 0.0824 max mem: 3531 +Test: [180/389] eta: 0:00:42 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.0832 (1.4950) acc1: 100.0000 (33.4412) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1593 data: 0.0668 max mem: 3531 +Test: [190/389] eta: 0:00:39 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.1074 (1.4275) acc1: 100.0000 (36.6329) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1449 data: 0.0518 max mem: 3531 +Test: [200/389] eta: 0:00:37 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.1074 (1.3609) acc1: 100.0000 (39.7842) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1638 data: 0.0720 max mem: 3531 +Test: [210/389] eta: 0:00:35 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.0706 (1.3002) acc1: 100.0000 (42.6380) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1767 data: 0.0857 max mem: 3532 +Test: [220/389] eta: 0:00:33 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.0680 (1.2445) acc1: 100.0000 (45.2336) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0873 max mem: 3532 +Test: [230/389] eta: 0:00:31 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.0801 (1.1963) acc1: 100.0000 (47.5368) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1731 data: 0.0793 max mem: 3532 +Test: [240/389] eta: 0:00:29 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.1076 (1.1515) acc1: 100.0000 (49.7126) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1816 data: 0.0892 max mem: 3532 +Test: [250/389] eta: 0:00:26 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.0954 (1.1092) acc1: 100.0000 (51.7161) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1772 data: 0.0849 max mem: 3532 +Test: [260/389] eta: 0:00:25 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.1007 (1.0735) acc1: 100.0000 (53.3894) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1749 data: 0.0831 max mem: 3532 +Test: [270/389] eta: 0:00:22 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.1110 (1.0407) acc1: 100.0000 (54.9739) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1809 data: 0.0902 max mem: 3532 +Test: [280/389] eta: 0:00:20 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.0998 (1.0085) acc1: 100.0000 (56.5280) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1532 data: 0.0618 max mem: 3532 +Test: [290/389] eta: 0:00:18 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.0762 (0.9775) acc1: 100.0000 (58.0085) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1585 data: 0.0650 max mem: 3532 +Test: [300/389] eta: 0:00:17 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.1159 (0.9490) acc1: 100.0000 (59.4036) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2056 data: 0.1126 max mem: 3532 +Test: [310/389] eta: 0:00:15 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.1169 (0.9222) acc1: 100.0000 (60.6989) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2131 data: 0.1212 max mem: 3532 +Test: [320/389] eta: 0:00:13 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.1067 (0.8968) acc1: 100.0000 (61.9078) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1993 data: 0.1070 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.1179 (0.8732) acc1: 100.0000 (63.0586) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1983 data: 0.1049 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.1080 (0.8501) acc1: 100.0000 (64.1419) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1948 data: 0.1009 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.0922 (0.8296) acc1: 100.0000 (65.1205) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2051 data: 0.1118 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.1026 (0.8095) acc1: 100.0000 (66.0860) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1902 data: 0.0983 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.1065 (0.7916) acc1: 100.0000 (66.9889) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1644 data: 0.0745 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.1122 (0.7736) acc1: 100.0000 (67.8553) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1512 data: 0.0617 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.0965 (0.7597) acc1: 100.0000 (68.5126) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1293 data: 0.0365 max mem: 3532 +Test: Total time: 0:01:13 (0.1895 s / it) +* Overall Acc@1 68.513 Acc@5 100.000 + target == real Acc@1 2.920 Acc@5 100.000 + target == fake Acc@1 99.002 Acc@5 100.000 + loss 0.760 + PR-AUC 0.7627 ROC-AUC 0.6234 EER 41.67% (thr=0.9082) +Accuracy of the network on the 149358 test images: 68.5% +[rank0]:[W726 06:14:40.784960860 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 06:15:01.567000 3166234 torch/distributed/run.py:766] +W0726 06:15:01.567000 3166234 torch/distributed/run.py:766] ***************************************** +W0726 06:15:01.567000 3166234 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:15:01.567000 3166234 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:15:35.526822629 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:15:35.868387302 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 06:15:35.868406395 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:15:36.329564195 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank4]:[W726 06:15:36.332476864 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +[rank1]:[W726 06:15:36.478603879 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:15:36.488795706 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W726 06:15:36.488975774 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n5_seed9_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed9_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:37:53 acc1_pos: 88.021 acc5_pos: 100.000 loss: 0.2616 (0.2616) acc1: 88.0208 (88.0208) acc5: 100.0000 (100.0000) time: 6.8263 data: 5.0637 max mem: 3531 +Test: [ 10/333] eta: 0:03:47 acc1_pos: 3.125 acc5_pos: 100.000 loss: 0.2627 (0.8281) acc1: 88.0208 (61.5057) acc5: 100.0000 (100.0000) time: 0.7057 data: 0.4607 max mem: 3531 +Test: [ 20/333] eta: 0:02:10 acc1_pos: 10.417 acc5_pos: 100.000 loss: 1.8066 (1.3256) acc1: 9.1146 (35.3299) acc5: 100.0000 (100.0000) time: 0.0968 data: 0.0004 max mem: 3531 +Test: [ 30/333] eta: 0:01:45 acc1_pos: 16.146 acc5_pos: 100.000 loss: 1.7450 (1.3868) acc1: 10.4167 (31.8464) acc5: 100.0000 (100.0000) time: 0.1509 data: 0.0540 max mem: 3531 +Test: [ 40/333] eta: 0:01:36 acc1_pos: 20.052 acc5_pos: 100.000 loss: 1.4854 (1.4126) acc1: 24.7396 (29.9035) acc5: 100.0000 (100.0000) time: 0.2349 data: 0.1419 max mem: 3531 +Test: [ 50/333] eta: 0:01:27 acc1_pos: 32.292 acc5_pos: 100.000 loss: 1.4803 (1.4261) acc1: 23.9583 (28.5692) acc5: 100.0000 (100.0000) time: 0.2457 data: 0.1509 max mem: 3531 +Test: [ 60/333] eta: 0:01:21 acc1_pos: 17.448 acc5_pos: 100.000 loss: 1.4552 (1.4318) acc1: 24.4792 (28.1335) acc5: 100.0000 (100.0000) time: 0.2325 data: 0.1358 max mem: 3531 +Test: [ 70/333] eta: 0:01:15 acc1_pos: 27.083 acc5_pos: 100.000 loss: 1.4314 (1.4290) acc1: 25.0000 (27.9343) acc5: 100.0000 (100.0000) time: 0.2294 data: 0.1366 max mem: 3531 +Test: [ 80/333] eta: 0:01:11 acc1_pos: 25.521 acc5_pos: 100.000 loss: 1.3920 (1.4273) acc1: 27.0833 (28.0221) acc5: 100.0000 (100.0000) time: 0.2373 data: 0.1457 max mem: 3531 +Test: [ 90/333] eta: 0:01:04 acc1_pos: 43.750 acc5_pos: 100.000 loss: 1.3264 (1.4010) acc1: 33.0729 (28.8433) acc5: 100.0000 (100.0000) time: 0.1887 data: 0.0929 max mem: 3531 +Test: [100/333] eta: 0:00:59 acc1_pos: 37.760 acc5_pos: 100.000 loss: 1.1547 (1.3796) acc1: 35.6771 (29.2285) acc5: 100.0000 (100.0000) time: 0.1535 data: 0.0588 max mem: 3531 +Test: [110/333] eta: 0:00:55 acc1_pos: 42.188 acc5_pos: 100.000 loss: 1.1306 (1.3544) acc1: 35.4167 (30.0816) acc5: 100.0000 (100.0000) time: 0.1867 data: 0.0963 max mem: 3531 +Test: [120/333] eta: 0:00:52 acc1_pos: 44.531 acc5_pos: 100.000 loss: 1.1250 (1.3386) acc1: 35.4167 (30.4967) acc5: 100.0000 (100.0000) time: 0.1914 data: 0.1008 max mem: 3531 +Test: [130/333] eta: 0:00:48 acc1_pos: 40.625 acc5_pos: 100.000 loss: 1.1182 (1.3217) acc1: 35.9375 (31.0870) acc5: 100.0000 (100.0000) time: 0.1849 data: 0.0936 max mem: 3531 +Test: [140/333] eta: 0:00:45 acc1_pos: 33.333 acc5_pos: 100.000 loss: 1.1083 (1.3120) acc1: 38.2812 (31.4107) acc5: 100.0000 (100.0000) time: 0.1748 data: 0.0817 max mem: 3531 +Test: [150/333] eta: 0:00:42 acc1_pos: 32.552 acc5_pos: 100.000 loss: 1.0901 (1.2995) acc1: 36.1979 (31.7691) acc5: 100.0000 (100.0000) time: 0.1780 data: 0.0845 max mem: 3531 +Test: [160/333] eta: 0:00:39 acc1_pos: 31.579 acc5_pos: 100.000 loss: 1.0617 (1.2751) acc1: 38.2812 (32.9468) acc5: 100.0000 (100.0000) acc1_neg: 72.396 acc5_neg: 100.000 time: 0.1883 data: 0.0953 max mem: 3531 +Test: [170/333] eta: 0:00:37 acc1_pos: 31.579 acc5_pos: 100.000 loss: 0.6071 (1.2344) acc1: 68.2292 (35.1334) acc5: 100.0000 (100.0000) acc1_neg: 71.094 acc5_neg: 100.000 time: 0.1910 data: 0.0986 max mem: 3531 +Test: [180/333] eta: 0:00:34 acc1_pos: 31.579 acc5_pos: 100.000 loss: 0.5445 (1.1944) acc1: 72.9167 (37.3345) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.1779 data: 0.0863 max mem: 3531 +Test: [190/333] eta: 0:00:31 acc1_pos: 31.579 acc5_pos: 100.000 loss: 0.5207 (1.1577) acc1: 75.0000 (39.4224) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.1881 data: 0.0945 max mem: 3531 +Test: [200/333] eta: 0:00:29 acc1_pos: 31.579 acc5_pos: 100.000 loss: 0.4875 (1.1246) acc1: 76.0417 (41.3000) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.1939 data: 0.0997 max mem: 3531 +Test: [210/333] eta: 0:00:26 acc1_pos: 31.579 acc5_pos: 100.000 loss: 0.5321 (1.0984) acc1: 73.4375 (42.7355) acc5: 100.0000 (100.0000) acc1_neg: 65.365 acc5_neg: 100.000 time: 0.1816 data: 0.0885 max mem: 3532 +Test: [220/333] eta: 0:00:24 acc1_pos: 31.579 acc5_pos: 100.000 loss: 0.5488 (1.0763) acc1: 71.6146 (43.8136) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.1798 data: 0.0872 max mem: 3532 +Test: [230/333] eta: 0:00:22 acc1_pos: 31.579 acc5_pos: 100.000 loss: 0.4163 (1.0457) acc1: 78.9062 (45.5222) acc5: 100.0000 (100.0000) acc1_neg: 78.646 acc5_neg: 100.000 time: 0.1827 data: 0.0910 max mem: 3532 +Test: [240/333] eta: 0:00:19 acc1_pos: 31.579 acc5_pos: 100.000 loss: 0.4163 (1.0211) acc1: 79.4271 (46.9020) acc5: 100.0000 (100.0000) acc1_neg: 79.427 acc5_neg: 100.000 time: 0.1780 data: 0.0868 max mem: 3532 +Test: [250/333] eta: 0:00:17 acc1_pos: 31.579 acc5_pos: 100.000 loss: 0.4545 (0.9979) acc1: 79.4271 (48.1968) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.1897 data: 0.0965 max mem: 3532 +Test: [260/333] eta: 0:00:15 acc1_pos: 31.579 acc5_pos: 100.000 loss: 0.4355 (0.9778) acc1: 77.3438 (49.3086) acc5: 100.0000 (100.0000) acc1_neg: 76.302 acc5_neg: 100.000 time: 0.1769 data: 0.0826 max mem: 3532 +Test: [270/333] eta: 0:00:13 acc1_pos: 31.579 acc5_pos: 100.000 loss: 0.4736 (0.9591) acc1: 77.3438 (50.3796) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.1573 data: 0.0651 max mem: 3532 +Test: [280/333] eta: 0:00:11 acc1_pos: 31.579 acc5_pos: 100.000 loss: 0.4767 (0.9475) acc1: 75.7812 (50.9786) acc5: 100.0000 (100.0000) acc1_neg: 68.490 acc5_neg: 100.000 time: 0.1785 data: 0.0877 max mem: 3532 +Test: [290/333] eta: 0:00:08 acc1_pos: 31.579 acc5_pos: 100.000 loss: 0.7008 (0.9394) acc1: 61.7188 (51.3620) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.1790 data: 0.0876 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 31.579 acc5_pos: 100.000 loss: 0.5740 (0.9261) acc1: 70.0521 (52.0747) acc5: 100.0000 (100.0000) acc1_neg: 70.052 acc5_neg: 100.000 time: 0.1745 data: 0.0823 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 31.579 acc5_pos: 100.000 loss: 0.5181 (0.9116) acc1: 72.9167 (52.8587) acc5: 100.0000 (100.0000) acc1_neg: 72.917 acc5_neg: 100.000 time: 0.1538 data: 0.0627 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 31.579 acc5_pos: 100.000 loss: 0.5152 (0.8996) acc1: 73.9583 (53.4519) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.1262 data: 0.0365 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 31.579 acc5_pos: 100.000 loss: 0.5619 (0.8905) acc1: 69.7917 (53.9629) acc5: 100.0000 (100.0000) acc1_neg: 72.135 acc5_neg: 100.000 time: 0.1197 data: 0.0302 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 31.579 acc5_pos: 100.000 loss: 0.5667 (0.8886) acc1: 70.3125 (54.0765) acc5: 100.0000 (100.0000) acc1_neg: 74.366 acc5_neg: 100.000 time: 0.1260 data: 0.0333 max mem: 3532 +Test: Total time: 0:01:06 (0.1996 s / it) +* Overall Acc@1 54.076 Acc@5 100.000 + target == real Acc@1 31.959 Acc@5 100.000 + target == fake Acc@1 73.765 Acc@5 100.000 + loss 0.889 + PR-AUC 0.5539 ROC-AUC 0.5419 EER 47.29% (thr=0.7441) +Accuracy of the network on the 127843 test images: 54.1% +[rank0]:[W726 06:16:53.945921878 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 06:17:15.843000 3166792 torch/distributed/run.py:766] +W0726 06:17:15.843000 3166792 torch/distributed/run.py:766] ***************************************** +W0726 06:17:15.843000 3166792 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:17:15.843000 3166792 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:17:53.652643286 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:17:53.732151132 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:17:53.113586649 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:17:54.262863676 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:17:54.282080325 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:17:54.328060812 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +[rank0]:[W726 06:17:54.339301853 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:17:54.345131678 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n5_seed9_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed9_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:20:07 acc1_pos: 52.604 acc5_pos: 100.000 loss: 0.8222 (0.8222) acc1: 52.6042 (52.6042) acc5: 100.0000 (100.0000) time: 5.1387 data: 3.1829 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/235] eta: 0:02:03 acc1_pos: 60.938 acc5_pos: 100.000 loss: 0.7765 (0.7747) acc1: 59.1146 (59.0199) acc5: 100.0000 (100.0000) time: 0.5506 data: 0.2915 max mem: 3531 +Test: [ 20/235] eta: 0:01:20 acc1_pos: 56.771 acc5_pos: 100.000 loss: 0.7648 (0.7755) acc1: 59.3750 (59.0650) acc5: 100.0000 (100.0000) time: 0.1352 data: 0.0436 max mem: 3531 +Test: [ 30/235] eta: 0:00:58 acc1_pos: 59.375 acc5_pos: 100.000 loss: 0.7648 (0.7729) acc1: 59.3750 (59.1818) acc5: 100.0000 (100.0000) time: 0.1392 data: 0.0427 max mem: 3531 +Test: [ 40/235] eta: 0:00:50 acc1_pos: 60.156 acc5_pos: 100.000 loss: 0.7715 (0.7717) acc1: 59.3750 (59.2924) acc5: 100.0000 (100.0000) time: 0.1399 data: 0.0451 max mem: 3531 +Test: [ 50/235] eta: 0:00:47 acc1_pos: 57.812 acc5_pos: 100.000 loss: 0.7729 (0.7737) acc1: 59.1146 (59.2116) acc5: 100.0000 (100.0000) time: 0.2099 data: 0.1182 max mem: 3531 +Test: [ 60/235] eta: 0:00:44 acc1_pos: 63.281 acc5_pos: 100.000 loss: 0.7687 (0.7706) acc1: 59.8958 (59.4518) acc5: 100.0000 (100.0000) time: 0.2411 data: 0.1476 max mem: 3531 +Test: [ 70/235] eta: 0:00:41 acc1_pos: 53.846 acc5_pos: 100.000 loss: 0.7507 (0.7701) acc1: 59.8958 (59.3273) acc5: 100.0000 (100.0000) acc1_neg: 55.729 acc5_neg: 100.000 time: 0.2361 data: 0.1440 max mem: 3531 +Test: [ 80/235] eta: 0:00:38 acc1_pos: 53.846 acc5_pos: 100.000 loss: 0.7678 (0.7703) acc1: 58.5938 (59.2078) acc5: 100.0000 (100.0000) acc1_neg: 58.073 acc5_neg: 100.000 time: 0.2254 data: 0.1352 max mem: 3531 +Test: [ 90/235] eta: 0:00:35 acc1_pos: 53.846 acc5_pos: 100.000 loss: 0.7712 (0.7709) acc1: 58.3333 (59.1146) acc5: 100.0000 (100.0000) acc1_neg: 58.333 acc5_neg: 100.000 time: 0.2241 data: 0.1328 max mem: 3531 +Test: [100/235] eta: 0:00:32 acc1_pos: 53.846 acc5_pos: 100.000 loss: 0.7737 (0.7696) acc1: 58.0729 (59.0501) acc5: 100.0000 (100.0000) acc1_neg: 55.208 acc5_neg: 100.000 time: 0.2199 data: 0.1266 max mem: 3531 +Test: [110/235] eta: 0:00:29 acc1_pos: 53.846 acc5_pos: 100.000 loss: 0.7663 (0.7698) acc1: 58.0729 (58.9363) acc5: 100.0000 (100.0000) acc1_neg: 58.333 acc5_neg: 100.000 time: 0.1909 data: 0.0991 max mem: 3531 +Test: [120/235] eta: 0:00:26 acc1_pos: 53.846 acc5_pos: 100.000 loss: 0.7933 (0.7912) acc1: 52.8646 (57.4251) acc5: 100.0000 (100.0000) acc1_neg: 35.417 acc5_neg: 100.000 time: 0.1807 data: 0.0904 max mem: 3531 +Test: [130/235] eta: 0:00:23 acc1_pos: 53.846 acc5_pos: 100.000 loss: 1.0342 (0.8095) acc1: 40.3646 (56.1268) acc5: 100.0000 (100.0000) acc1_neg: 42.708 acc5_neg: 100.000 time: 0.1846 data: 0.0935 max mem: 3531 +Test: [140/235] eta: 0:00:21 acc1_pos: 53.846 acc5_pos: 100.000 loss: 1.0388 (0.8275) acc1: 38.5417 (54.8242) acc5: 100.0000 (100.0000) acc1_neg: 38.021 acc5_neg: 100.000 time: 0.1774 data: 0.0853 max mem: 3531 +Test: [150/235] eta: 0:00:18 acc1_pos: 53.846 acc5_pos: 100.000 loss: 1.0579 (0.8418) acc1: 38.0208 (53.7890) acc5: 100.0000 (100.0000) acc1_neg: 36.458 acc5_neg: 100.000 time: 0.1792 data: 0.0873 max mem: 3531 +Test: [160/235] eta: 0:00:16 acc1_pos: 53.846 acc5_pos: 100.000 loss: 1.0474 (0.8544) acc1: 38.2812 (52.8290) acc5: 100.0000 (100.0000) acc1_neg: 38.281 acc5_neg: 100.000 time: 0.1754 data: 0.0840 max mem: 3531 +Test: [170/235] eta: 0:00:14 acc1_pos: 53.846 acc5_pos: 100.000 loss: 1.0095 (0.8343) acc1: 39.8438 (54.1195) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.1765 data: 0.0844 max mem: 3531 +Test: [180/235] eta: 0:00:11 acc1_pos: 53.846 acc5_pos: 100.000 loss: 0.4560 (0.8131) acc1: 79.1667 (55.5335) acc5: 100.0000 (100.0000) acc1_neg: 77.344 acc5_neg: 100.000 time: 0.1853 data: 0.0938 max mem: 3531 +Test: [190/235] eta: 0:00:09 acc1_pos: 53.846 acc5_pos: 100.000 loss: 0.4464 (0.7937) acc1: 79.4271 (56.8076) acc5: 100.0000 (100.0000) acc1_neg: 80.729 acc5_neg: 100.000 time: 0.1930 data: 0.1021 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 53.846 acc5_pos: 100.000 loss: 0.4429 (0.7766) acc1: 79.6875 (57.9317) acc5: 100.0000 (100.0000) acc1_neg: 77.083 acc5_neg: 100.000 time: 0.1853 data: 0.0949 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 53.846 acc5_pos: 100.000 loss: 0.4426 (0.7608) acc1: 79.6875 (58.9764) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.1606 data: 0.0689 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 53.846 acc5_pos: 100.000 loss: 0.4456 (0.7465) acc1: 79.9479 (59.9441) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.1379 data: 0.0457 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 53.846 acc5_pos: 100.000 loss: 0.4197 (0.7320) acc1: 80.7292 (60.8800) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.1261 data: 0.0361 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 53.846 acc5_pos: 100.000 loss: 0.4197 (0.7266) acc1: 80.7292 (61.1289) acc5: 100.0000 (100.0000) acc1_neg: 82.609 acc5_neg: 100.000 time: 0.1304 data: 0.0382 max mem: 3532 +Test: Total time: 0:00:47 (0.2005 s / it) +* Overall Acc@1 61.129 Acc@5 100.000 + target == real Acc@1 59.414 Acc@5 100.000 + target == fake Acc@1 61.749 Acc@5 100.000 + loss 0.727 + PR-AUC 0.8327 ROC-AUC 0.6524 EER 39.22% (thr=0.5156) +Accuracy of the network on the 89879 test images: 61.1% +[rank0]:[W726 06:18:51.858301767 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 06:19:13.796000 3167353 torch/distributed/run.py:766] +W0726 06:19:13.796000 3167353 torch/distributed/run.py:766] ***************************************** +W0726 06:19:13.796000 3167353 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:19:13.796000 3167353 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:19:46.053897570 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 06:19:47.911642851 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +[rank6]:[W726 06:19:47.062531125 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:19:47.069823670 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:19:48.490420867 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:19:48.541927716 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:19:48.625202510 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:19:48.671174734 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n5_seed9_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed9_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:22:14 acc1_pos: 5.208 acc5_pos: 100.000 loss: 1.9857 (1.9857) acc1: 5.2083 (5.2083) acc5: 100.0000 (100.0000) time: 4.9599 data: 2.8849 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/269] eta: 0:02:17 acc1_pos: 11.719 acc5_pos: 100.000 loss: 1.7904 (1.7546) acc1: 11.7188 (12.1449) acc5: 100.0000 (100.0000) time: 0.5324 data: 0.2626 max mem: 3531 +Test: [ 20/269] eta: 0:01:20 acc1_pos: 18.490 acc5_pos: 100.000 loss: 1.7904 (1.8076) acc1: 8.0729 (9.7470) acc5: 100.0000 (100.0000) time: 0.0921 data: 0.0013 max mem: 3531 +Test: [ 30/269] eta: 0:01:00 acc1_pos: 7.292 acc5_pos: 100.000 loss: 1.7818 (1.8112) acc1: 7.0312 (10.0050) acc5: 100.0000 (100.0000) time: 0.0963 data: 0.0013 max mem: 3531 +Test: [ 40/269] eta: 0:00:51 acc1_pos: 14.655 acc5_pos: 100.000 loss: 1.5782 (1.4564) acc1: 19.5312 (28.8681) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1213 data: 0.0265 max mem: 3531 +Test: [ 50/269] eta: 0:00:47 acc1_pos: 14.655 acc5_pos: 100.000 loss: 0.1707 (1.2118) acc1: 95.0521 (41.5799) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.1615 data: 0.0662 max mem: 3531 +Test: [ 60/269] eta: 0:00:43 acc1_pos: 14.655 acc5_pos: 100.000 loss: 0.2066 (1.0549) acc1: 93.4896 (49.5261) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1777 data: 0.0794 max mem: 3531 +Test: [ 70/269] eta: 0:00:41 acc1_pos: 14.655 acc5_pos: 100.000 loss: 0.2253 (0.9375) acc1: 93.4896 (55.7035) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1834 data: 0.0873 max mem: 3531 +Test: [ 80/269] eta: 0:00:38 acc1_pos: 14.655 acc5_pos: 100.000 loss: 0.1933 (0.8467) acc1: 95.5729 (60.4842) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0903 max mem: 3531 +Test: [ 90/269] eta: 0:00:36 acc1_pos: 14.655 acc5_pos: 100.000 loss: 0.1788 (0.7751) acc1: 96.0938 (64.3802) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2049 data: 0.1066 max mem: 3531 +Test: [100/269] eta: 0:00:34 acc1_pos: 14.655 acc5_pos: 100.000 loss: 0.1946 (0.7246) acc1: 94.0104 (66.8291) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1999 data: 0.1022 max mem: 3531 +Test: [110/269] eta: 0:00:31 acc1_pos: 14.655 acc5_pos: 100.000 loss: 0.3060 (0.6965) acc1: 85.1562 (68.2245) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.1701 data: 0.0772 max mem: 3531 +Test: [120/269] eta: 0:00:30 acc1_pos: 14.655 acc5_pos: 100.000 loss: 0.2772 (0.6602) acc1: 88.0208 (70.2501) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1999 data: 0.1065 max mem: 3531 +Test: [130/269] eta: 0:00:28 acc1_pos: 14.655 acc5_pos: 100.000 loss: 0.2247 (0.6278) acc1: 95.3125 (72.0917) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2197 data: 0.1236 max mem: 3531 +Test: [140/269] eta: 0:00:26 acc1_pos: 14.655 acc5_pos: 100.000 loss: 0.1908 (0.5956) acc1: 98.9583 (73.9990) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1990 data: 0.1037 max mem: 3531 +Test: [150/269] eta: 0:00:23 acc1_pos: 14.655 acc5_pos: 100.000 loss: 0.1608 (0.5664) acc1: 98.6979 (75.6174) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1854 data: 0.0916 max mem: 3531 +Test: [160/269] eta: 0:00:21 acc1_pos: 14.655 acc5_pos: 100.000 loss: 0.1466 (0.5417) acc1: 98.4375 (76.9911) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1749 data: 0.0816 max mem: 3531 +Test: [170/269] eta: 0:00:19 acc1_pos: 14.655 acc5_pos: 100.000 loss: 0.1879 (0.5222) acc1: 96.8750 (78.1159) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1585 data: 0.0659 max mem: 3531 +Test: [180/269] eta: 0:00:17 acc1_pos: 14.655 acc5_pos: 100.000 loss: 0.1988 (0.5049) acc1: 95.8333 (79.0343) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1949 data: 0.1002 max mem: 3531 +Test: [190/269] eta: 0:00:15 acc1_pos: 14.655 acc5_pos: 100.000 loss: 0.2032 (0.4903) acc1: 94.0104 (79.8116) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2149 data: 0.1182 max mem: 3531 +Test: [200/269] eta: 0:00:13 acc1_pos: 14.655 acc5_pos: 100.000 loss: 0.2450 (0.4788) acc1: 92.9688 (80.3638) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1666 data: 0.0718 max mem: 3531 +Test: [210/269] eta: 0:00:11 acc1_pos: 14.655 acc5_pos: 100.000 loss: 0.2409 (0.4664) acc1: 89.8438 (80.9007) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1799 data: 0.0875 max mem: 3532 +Test: [220/269] eta: 0:00:09 acc1_pos: 14.655 acc5_pos: 100.000 loss: 0.2017 (0.4541) acc1: 92.1875 (81.4303) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2195 data: 0.1276 max mem: 3532 +Test: [230/269] eta: 0:00:07 acc1_pos: 14.655 acc5_pos: 100.000 loss: 0.1475 (0.4423) acc1: 97.6562 (82.0549) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1999 data: 0.1059 max mem: 3532 +Test: [240/269] eta: 0:00:05 acc1_pos: 14.655 acc5_pos: 100.000 loss: 0.2236 (0.4353) acc1: 93.4896 (82.3922) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1743 data: 0.0769 max mem: 3532 +Test: [250/269] eta: 0:00:03 acc1_pos: 14.655 acc5_pos: 100.000 loss: 0.1908 (0.4249) acc1: 97.6562 (83.0283) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1849 data: 0.0894 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 14.655 acc5_pos: 100.000 loss: 0.1853 (0.4167) acc1: 97.6562 (83.4850) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1874 data: 0.0956 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 14.655 acc5_pos: 100.000 loss: 0.2164 (0.4124) acc1: 94.2708 (83.6949) acc5: 100.0000 (100.0000) acc1_neg: 95.276 acc5_neg: 100.000 time: 0.1740 data: 0.0678 max mem: 3532 +Test: Total time: 0:00:52 (0.1952 s / it) +* Overall Acc@1 83.695 Acc@5 100.000 + target == real Acc@1 10.343 Acc@5 100.000 + target == fake Acc@1 93.706 Acc@5 100.000 + loss 0.412 + PR-AUC 0.8997 ROC-AUC 0.5762 EER 43.97% (thr=0.8765) +Accuracy of the network on the 103293 test images: 83.7% +[rank0]:[W726 06:20:52.761406750 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T06:21:00Z] eval_done tag=fixed500k_real_n5_seed9_resampled +[2026-07-26T06:21:00Z] eval_start tag=fixed500k_fake_n11_seed9_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed9_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T06:21:00Z +hostname=n116-040-168 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed9_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n11_seed9_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 06:21:15.735000 3167917 torch/distributed/run.py:766] +W0726 06:21:15.735000 3167917 torch/distributed/run.py:766] ***************************************** +W0726 06:21:15.735000 3167917 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:21:15.735000 3167917 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 06:21:49.445901823 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:21:49.869917167 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:21:49.043356511 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:21:50.233370235 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:21:50.342111724 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:21:50.415257048 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:21:50.500249673 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:21:50.690152286 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n11_seed9_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed9_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:18:42 acc1_pos: 0.260 acc5_pos: 100.000 loss: 1.4066 (1.4066) acc1: 0.2604 (0.2604) acc5: 100.0000 (100.0000) time: 4.7954 data: 2.6575 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/234] eta: 0:01:57 acc1_pos: 1.042 acc5_pos: 100.000 loss: 1.3195 (1.2075) acc1: 13.5417 (17.2112) acc5: 100.0000 (100.0000) time: 0.5248 data: 0.2492 max mem: 3531 +Test: [ 20/234] eta: 0:01:13 acc1_pos: 93.229 acc5_pos: 100.000 loss: 1.2614 (1.1170) acc1: 11.1979 (25.3100) acc5: 100.0000 (100.0000) time: 0.1224 data: 0.0326 max mem: 3531 +Test: [ 30/234] eta: 0:00:54 acc1_pos: 33.854 acc5_pos: 100.000 loss: 0.2576 (0.8739) acc1: 92.7083 (44.1532) acc5: 100.0000 (100.0000) time: 0.1234 data: 0.0287 max mem: 3531 +Test: [ 40/234] eta: 0:00:50 acc1_pos: 99.479 acc5_pos: 100.000 loss: 0.2494 (0.8187) acc1: 93.4896 (48.0246) acc5: 100.0000 (100.0000) time: 0.1749 data: 0.0778 max mem: 3531 +Test: [ 50/234] eta: 0:00:50 acc1_pos: 98.958 acc5_pos: 100.000 loss: 0.1172 (0.6804) acc1: 99.2188 (58.1087) acc5: 100.0000 (100.0000) time: 0.2813 data: 0.1879 max mem: 3531 +Test: [ 60/234] eta: 0:00:43 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1079 (0.5858) acc1: 99.7396 (64.9590) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2249 data: 0.1326 max mem: 3531 +Test: [ 70/234] eta: 0:00:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1074 (0.6076) acc1: 99.7396 (64.8511) acc5: 100.0000 (100.0000) acc1_neg: 43.490 acc5_neg: 100.000 time: 0.1435 data: 0.0502 max mem: 3531 +Test: [ 80/234] eta: 0:00:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.0309 (0.6246) acc1: 48.1771 (64.8052) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1915 data: 0.0987 max mem: 3531 +Test: [ 90/234] eta: 0:00:33 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1678 (0.5736) acc1: 96.0938 (68.2721) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.2299 data: 0.1385 max mem: 3531 +Test: [100/234] eta: 0:00:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1842 (0.5545) acc1: 96.0938 (69.7530) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2033 data: 0.1093 max mem: 3531 +Test: [110/234] eta: 0:00:27 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3243 (0.5339) acc1: 85.6771 (71.3800) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.1449 data: 0.0466 max mem: 3531 +Test: [120/234] eta: 0:00:24 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2204 (0.5296) acc1: 89.0625 (72.1483) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1220 data: 0.0248 max mem: 3531 +Test: [130/234] eta: 0:00:21 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2295 (0.5132) acc1: 95.5729 (73.6045) acc5: 100.0000 (100.0000) acc1_neg: 72.396 acc5_neg: 100.000 time: 0.1549 data: 0.0602 max mem: 3531 +Test: [140/234] eta: 0:00:19 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4148 (0.5595) acc1: 80.7292 (71.3634) acc5: 100.0000 (100.0000) acc1_neg: 5.208 acc5_neg: 100.000 time: 0.1778 data: 0.0807 max mem: 3531 +Test: [150/234] eta: 0:00:17 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5197 (0.5608) acc1: 72.1354 (71.5249) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.1953 data: 0.0992 max mem: 3531 +Test: [160/234] eta: 0:00:15 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3217 (0.5457) acc1: 86.9792 (72.7145) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.2106 data: 0.1179 max mem: 3531 +Test: [170/234] eta: 0:00:13 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3912 (0.5358) acc1: 86.1979 (73.6081) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1845 data: 0.0902 max mem: 3531 +Test: [180/234] eta: 0:00:10 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4319 (0.5485) acc1: 83.3333 (72.9641) acc5: 100.0000 (100.0000) acc1_neg: 19.010 acc5_neg: 100.000 time: 0.1592 data: 0.0628 max mem: 3531 +Test: [190/234] eta: 0:00:08 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.9637 (0.5727) acc1: 47.3958 (71.5655) acc5: 100.0000 (100.0000) acc1_neg: 47.135 acc5_neg: 100.000 time: 0.1956 data: 0.0995 max mem: 3531 +Test: [200/234] eta: 0:00:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.9634 (0.5662) acc1: 49.2188 (72.0382) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2349 data: 0.1404 max mem: 3531 +Test: [210/234] eta: 0:00:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1465 (0.5513) acc1: 99.4792 (72.9808) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.1850 data: 0.0908 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4506 (0.5490) acc1: 78.3854 (73.0875) acc5: 100.0000 (100.0000) acc1_neg: 56.510 acc5_neg: 100.000 time: 0.1419 data: 0.0490 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.8436 (0.5962) acc1: 13.2812 (70.4095) acc5: 100.0000 (100.0000) acc1_neg: 13.281 acc5_neg: 100.000 time: 0.1545 data: 0.0639 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.6069 (0.6094) acc1: 12.5000 (69.7333) acc5: 100.0000 (100.0000) acc1_neg: 11.284 acc5_neg: 100.000 time: 0.1585 data: 0.0673 max mem: 3532 +Test: Total time: 0:00:46 (0.1976 s / it) +* Overall Acc@1 69.733 Acc@5 100.000 + target == real Acc@1 58.417 Acc@5 100.000 + target == fake Acc@1 72.923 Acc@5 100.000 + loss 0.609 + PR-AUC 0.9299 ROC-AUC 0.7796 EER 32.95% (thr=0.6470) +Accuracy of the network on the 89729 test images: 69.7% +[rank0]:[W726 06:22:48.629694968 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 06:23:10.058000 3168475 torch/distributed/run.py:766] +W0726 06:23:10.058000 3168475 torch/distributed/run.py:766] ***************************************** +W0726 06:23:10.058000 3168475 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:23:10.058000 3168475 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:23:42.577454841 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:23:42.942960300 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:23:43.182200415 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:23:43.219240198 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:23:43.515607116 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 06:23:43.537279250 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:23:43.551508016 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W726 06:23:43.551520254 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n11_seed9_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed9_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 3:59:17 acc1_pos: 1.823 acc5_pos: 100.000 loss: 1.6874 (1.6874) acc1: 1.8229 (1.8229) acc5: 100.0000 (100.0000) time: 6.0150 data: 4.2813 max mem: 3531 +Test: [ 10/2387] eta: 0:24:58 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.6435 (1.6195) acc1: 0.7812 (2.4858) acc5: 100.0000 (100.0000) time: 0.6302 data: 0.3895 max mem: 3531 +Test: [ 20/2387] eta: 0:14:50 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.6435 (1.6408) acc1: 0.7812 (2.7034) acc5: 100.0000 (100.0000) time: 0.0942 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:12:38 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.6746 (1.6578) acc1: 0.7812 (2.0077) acc5: 100.0000 (100.0000) time: 0.1520 data: 0.0590 max mem: 3531 +Test: [ 40/2387] eta: 0:11:53 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.6560 (1.6458) acc1: 0.0000 (1.6514) acc5: 100.0000 (100.0000) time: 0.2279 data: 0.1350 max mem: 3531 +Test: [ 50/2387] eta: 0:11:13 acc1_pos: 10.156 acc5_pos: 100.000 loss: 1.5682 (1.6336) acc1: 0.5208 (1.8842) acc5: 100.0000 (100.0000) time: 0.2358 data: 0.1402 max mem: 3531 +Test: [ 60/2387] eta: 0:10:57 acc1_pos: 4.688 acc5_pos: 100.000 loss: 1.5856 (1.6276) acc1: 1.5625 (2.1773) acc5: 100.0000 (100.0000) time: 0.2392 data: 0.1458 max mem: 3531 +Test: [ 70/2387] eta: 0:10:48 acc1_pos: 8.333 acc5_pos: 100.000 loss: 1.6318 (1.6372) acc1: 0.7812 (2.1420) acc5: 100.0000 (100.0000) time: 0.2587 data: 0.1658 max mem: 3531 +Test: [ 80/2387] eta: 0:10:36 acc1_pos: 0.521 acc5_pos: 100.000 loss: 1.6909 (1.6496) acc1: 0.2604 (1.9162) acc5: 100.0000 (100.0000) time: 0.2549 data: 0.1602 max mem: 3531 +Test: [ 90/2387] eta: 0:10:27 acc1_pos: 5.469 acc5_pos: 100.000 loss: 1.7414 (1.6553) acc1: 0.2604 (1.9946) acc5: 100.0000 (100.0000) time: 0.2489 data: 0.1557 max mem: 3531 +Test: [ 100/2387] eta: 0:10:19 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.7242 (1.6623) acc1: 0.5208 (1.9441) acc5: 100.0000 (100.0000) time: 0.2501 data: 0.1580 max mem: 3531 +Test: [ 110/2387] eta: 0:10:14 acc1_pos: 7.812 acc5_pos: 100.000 loss: 1.7242 (1.6657) acc1: 0.0000 (1.9285) acc5: 100.0000 (100.0000) time: 0.2559 data: 0.1632 max mem: 3531 +Test: [ 120/2387] eta: 0:10:07 acc1_pos: 1.823 acc5_pos: 100.000 loss: 1.6863 (1.6686) acc1: 0.5208 (1.9843) acc5: 100.0000 (100.0000) time: 0.2547 data: 0.1621 max mem: 3531 +Test: [ 130/2387] eta: 0:09:58 acc1_pos: 5.208 acc5_pos: 100.000 loss: 1.7068 (1.6731) acc1: 0.2604 (1.9541) acc5: 100.0000 (100.0000) time: 0.2395 data: 0.1466 max mem: 3531 +Test: [ 140/2387] eta: 0:09:55 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.7304 (1.6770) acc1: 0.0000 (1.9171) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1506 max mem: 3531 +Test: [ 150/2387] eta: 0:09:45 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.6775 (1.6770) acc1: 0.2604 (1.9212) acc5: 100.0000 (100.0000) time: 0.2396 data: 0.1457 max mem: 3531 +Test: [ 160/2387] eta: 0:09:33 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.6702 (1.6432) acc1: 1.0417 (4.0534) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2049 data: 0.1115 max mem: 3531 +Test: [ 170/2387] eta: 0:09:15 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1628 (1.5564) acc1: 99.2188 (9.6506) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1662 data: 0.0739 max mem: 3531 +Test: [ 180/2387] eta: 0:09:22 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1570 (1.4792) acc1: 99.7396 (14.6193) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2351 data: 0.1451 max mem: 3531 +Test: [ 190/2387] eta: 0:09:10 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1565 (1.4099) acc1: 99.7396 (19.0772) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2482 data: 0.1536 max mem: 3531 +Test: [ 200/2387] eta: 0:08:57 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1562 (1.3475) acc1: 99.7396 (23.0929) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1647 data: 0.0665 max mem: 3531 +Test: [ 210/2387] eta: 0:08:44 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1582 (1.2913) acc1: 99.7396 (26.7192) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1513 data: 0.0549 max mem: 3532 +Test: [ 220/2387] eta: 0:08:39 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1599 (1.2400) acc1: 99.7396 (30.0292) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1751 data: 0.0819 max mem: 3532 +Test: [ 230/2387] eta: 0:08:31 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1580 (1.1931) acc1: 100.0000 (33.0492) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1935 data: 0.1010 max mem: 3532 +Test: [ 240/2387] eta: 0:08:24 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1518 (1.1496) acc1: 100.0000 (35.8240) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1847 data: 0.0888 max mem: 3532 +Test: [ 250/2387] eta: 0:08:16 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1386 (1.1082) acc1: 100.0000 (38.3788) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1822 data: 0.0848 max mem: 3532 +Test: [ 260/2387] eta: 0:08:08 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1003 (1.0695) acc1: 100.0000 (40.7397) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1649 data: 0.0678 max mem: 3532 +Test: [ 270/2387] eta: 0:08:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0942 (1.0335) acc1: 100.0000 (42.9255) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1625 data: 0.0638 max mem: 3532 +Test: [ 280/2387] eta: 0:07:57 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0970 (1.0003) acc1: 100.0000 (44.9566) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0873 max mem: 3532 +Test: [ 290/2387] eta: 0:07:53 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1002 (0.9695) acc1: 100.0000 (46.8482) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2016 data: 0.1072 max mem: 3532 +Test: [ 300/2387] eta: 0:07:47 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1033 (0.9407) acc1: 100.0000 (48.6140) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1911 data: 0.0957 max mem: 3532 +Test: [ 310/2387] eta: 0:07:43 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1040 (0.9170) acc1: 100.0000 (50.2119) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1901 data: 0.0964 max mem: 3532 +Test: [ 320/2387] eta: 0:07:39 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2929 (0.8977) acc1: 99.2188 (51.7264) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1987 data: 0.1063 max mem: 3532 +Test: [ 330/2387] eta: 0:07:36 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2944 (0.8795) acc1: 98.9583 (53.1588) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1981 data: 0.1018 max mem: 3532 +Test: [ 340/2387] eta: 0:07:32 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2935 (0.8624) acc1: 98.9583 (54.5042) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1952 data: 0.0998 max mem: 3532 +Test: [ 350/2387] eta: 0:07:27 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2882 (0.8431) acc1: 99.4792 (55.7952) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1824 data: 0.0883 max mem: 3532 +Test: [ 360/2387] eta: 0:07:23 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1654 (0.8237) acc1: 100.0000 (57.0190) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1804 data: 0.0876 max mem: 3532 +Test: [ 370/2387] eta: 0:07:19 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1400 (0.8053) acc1: 100.0000 (58.1761) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1883 data: 0.0975 max mem: 3532 +Test: [ 380/2387] eta: 0:07:15 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1403 (0.7878) acc1: 100.0000 (59.2711) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1900 data: 0.0985 max mem: 3532 +Test: [ 390/2387] eta: 0:07:11 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1404 (0.7712) acc1: 100.0000 (60.3114) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1854 data: 0.0898 max mem: 3532 +Test: [ 400/2387] eta: 0:07:07 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1384 (0.7555) acc1: 100.0000 (61.3005) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1818 data: 0.0864 max mem: 3532 +Test: [ 410/2387] eta: 0:07:02 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1373 (0.7405) acc1: 100.0000 (62.2396) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1697 data: 0.0771 max mem: 3532 +Test: [ 420/2387] eta: 0:07:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1369 (0.7261) acc1: 100.0000 (63.1340) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1823 data: 0.0898 max mem: 3532 +Test: [ 430/2387] eta: 0:06:56 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1347 (0.7124) acc1: 100.0000 (63.9882) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1848 data: 0.0920 max mem: 3532 +Test: [ 440/2387] eta: 0:06:52 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1309 (0.6988) acc1: 100.0000 (64.8048) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1747 data: 0.0795 max mem: 3533 +Test: [ 450/2387] eta: 0:06:47 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1020 (0.6856) acc1: 100.0000 (65.5852) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1664 data: 0.0722 max mem: 3533 +Test: [ 460/2387] eta: 0:06:43 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1010 (0.6729) acc1: 100.0000 (66.3311) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1548 data: 0.0629 max mem: 3533 +Test: [ 470/2387] eta: 0:06:40 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1010 (0.6608) acc1: 100.0000 (67.0460) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1711 data: 0.0796 max mem: 3533 +Test: [ 480/2387] eta: 0:06:36 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1020 (0.6492) acc1: 100.0000 (67.7311) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1679 data: 0.0761 max mem: 3533 +Test: [ 490/2387] eta: 0:06:33 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1019 (0.6380) acc1: 100.0000 (68.3883) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1697 data: 0.0786 max mem: 3533 +Test: [ 500/2387] eta: 0:06:29 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1019 (0.6273) acc1: 100.0000 (69.0187) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1708 data: 0.0771 max mem: 3533 +Test: [ 510/2387] eta: 0:06:26 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1023 (0.6171) acc1: 100.0000 (69.6250) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1676 data: 0.0733 max mem: 3533 +Test: [ 520/2387] eta: 0:06:23 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1142 (0.6076) acc1: 100.0000 (70.2080) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1811 data: 0.0879 max mem: 3533 +Test: [ 530/2387] eta: 0:06:20 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1139 (0.5982) acc1: 100.0000 (70.7691) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1894 data: 0.0942 max mem: 3533 +Test: [ 540/2387] eta: 0:06:18 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1121 (0.5892) acc1: 100.0000 (71.3094) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1913 data: 0.0971 max mem: 3533 +Test: [ 550/2387] eta: 0:06:15 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1116 (0.5806) acc1: 100.0000 (71.8301) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1837 data: 0.0908 max mem: 3533 +Test: [ 560/2387] eta: 0:06:13 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1116 (0.5722) acc1: 100.0000 (72.3322) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1889 data: 0.0959 max mem: 3533 +Test: [ 570/2387] eta: 0:06:10 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1113 (0.5641) acc1: 100.0000 (72.8168) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1889 data: 0.0964 max mem: 3533 +Test: [ 580/2387] eta: 0:06:07 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1112 (0.5564) acc1: 100.0000 (73.2847) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1791 data: 0.0872 max mem: 3533 +Test: [ 590/2387] eta: 0:06:05 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1121 (0.5488) acc1: 100.0000 (73.7367) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1951 data: 0.1040 max mem: 3533 +Test: [ 600/2387] eta: 0:06:02 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1151 (0.5417) acc1: 100.0000 (74.1737) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1943 data: 0.1032 max mem: 3533 +Test: [ 610/2387] eta: 0:06:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1194 (0.5348) acc1: 100.0000 (74.5955) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1797 data: 0.0884 max mem: 3533 +Test: [ 620/2387] eta: 0:05:57 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1164 (0.5281) acc1: 100.0000 (75.0038) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1812 data: 0.0894 max mem: 3533 +Test: [ 630/2387] eta: 0:05:55 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1162 (0.5215) acc1: 100.0000 (75.3987) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1868 data: 0.0927 max mem: 3533 +Test: [ 640/2387] eta: 0:05:52 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1163 (0.5152) acc1: 100.0000 (75.7825) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1832 data: 0.0893 max mem: 3533 +Test: [ 650/2387] eta: 0:05:49 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1166 (0.5091) acc1: 100.0000 (76.1541) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1778 data: 0.0837 max mem: 3533 +Test: [ 660/2387] eta: 0:05:47 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1169 (0.5032) acc1: 100.0000 (76.5144) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1808 data: 0.0866 max mem: 3533 +Test: [ 670/2387] eta: 0:05:44 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1168 (0.4974) acc1: 100.0000 (76.8637) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1830 data: 0.0914 max mem: 3534 +Test: [ 680/2387] eta: 0:05:42 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1166 (0.4918) acc1: 100.0000 (77.2034) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1856 data: 0.0944 max mem: 3534 +Test: [ 690/2387] eta: 0:05:40 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1180 (0.4883) acc1: 100.0000 (77.4093) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1818 data: 0.0870 max mem: 3534 +Test: [ 700/2387] eta: 0:05:37 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2938 (0.4856) acc1: 88.0208 (77.5574) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1710 data: 0.0762 max mem: 3534 +Test: [ 710/2387] eta: 0:05:34 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2935 (0.4829) acc1: 88.0208 (77.7082) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1702 data: 0.0757 max mem: 3534 +Test: [ 720/2387] eta: 0:05:31 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2910 (0.4802) acc1: 88.2812 (77.8606) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1712 data: 0.0752 max mem: 3534 +Test: [ 730/2387] eta: 0:05:29 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2946 (0.4778) acc1: 88.0208 (77.9925) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1646 data: 0.0721 max mem: 3534 +Test: [ 740/2387] eta: 0:05:27 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.3020 (0.4754) acc1: 87.5000 (78.1257) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.1799 data: 0.0877 max mem: 3534 +Test: [ 750/2387] eta: 0:05:23 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2968 (0.4729) acc1: 87.7604 (78.2595) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1741 data: 0.0811 max mem: 3534 +Test: [ 760/2387] eta: 0:05:21 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2854 (0.4699) acc1: 88.0208 (78.4320) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1599 data: 0.0654 max mem: 3534 +Test: [ 770/2387] eta: 0:05:19 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1548 (0.4656) acc1: 96.3542 (78.6847) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1746 data: 0.0813 max mem: 3534 +Test: [ 780/2387] eta: 0:05:16 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1182 (0.4611) acc1: 100.0000 (78.9533) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1766 data: 0.0852 max mem: 3534 +Test: [ 790/2387] eta: 0:05:14 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1083 (0.4572) acc1: 100.0000 (79.1696) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.1849 data: 0.0915 max mem: 3534 +Test: [ 800/2387] eta: 0:05:12 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1066 (0.4529) acc1: 100.0000 (79.4274) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1836 data: 0.0901 max mem: 3534 +Test: [ 810/2387] eta: 0:05:09 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1109 (0.4487) acc1: 100.0000 (79.6798) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1600 data: 0.0670 max mem: 3534 +Test: [ 820/2387] eta: 0:05:06 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1145 (0.4447) acc1: 100.0000 (79.9232) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1545 data: 0.0625 max mem: 3534 +Test: [ 830/2387] eta: 0:05:04 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1228 (0.4410) acc1: 100.0000 (80.1529) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1625 data: 0.0714 max mem: 3534 +Test: [ 840/2387] eta: 0:05:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1499 (0.4377) acc1: 99.7396 (80.3830) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1664 data: 0.0743 max mem: 3534 +Test: [ 850/2387] eta: 0:04:59 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1632 (0.4344) acc1: 100.0000 (80.6095) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1680 data: 0.0735 max mem: 3534 +Test: [ 860/2387] eta: 0:04:56 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1551 (0.4312) acc1: 100.0000 (80.8335) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1623 data: 0.0667 max mem: 3534 +Test: [ 870/2387] eta: 0:04:53 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1549 (0.4280) acc1: 100.0000 (81.0524) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1516 data: 0.0589 max mem: 3534 +Test: [ 880/2387] eta: 0:04:51 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1611 (0.4251) acc1: 100.0000 (81.2663) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1660 data: 0.0725 max mem: 3534 +Test: [ 890/2387] eta: 0:04:49 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1511 (0.4219) acc1: 100.0000 (81.4762) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1777 data: 0.0846 max mem: 3534 +Test: [ 900/2387] eta: 0:04:47 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1474 (0.4247) acc1: 99.7396 (81.2619) acc5: 100.0000 (100.0000) acc1_neg: 57.031 acc5_neg: 100.000 time: 0.1716 data: 0.0812 max mem: 3535 +Test: [ 910/2387] eta: 0:04:44 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.7961 (0.4289) acc1: 53.6458 (80.9556) acc5: 100.0000 (100.0000) acc1_neg: 53.385 acc5_neg: 100.000 time: 0.1727 data: 0.0800 max mem: 3535 +Test: [ 920/2387] eta: 0:04:42 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.8087 (0.4331) acc1: 52.8646 (80.6415) acc5: 100.0000 (100.0000) acc1_neg: 46.615 acc5_neg: 100.000 time: 0.1732 data: 0.0800 max mem: 3535 +Test: [ 930/2387] eta: 0:04:40 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.8087 (0.4371) acc1: 52.8646 (80.3527) acc5: 100.0000 (100.0000) acc1_neg: 48.177 acc5_neg: 100.000 time: 0.1720 data: 0.0798 max mem: 3535 +Test: [ 940/2387] eta: 0:04:37 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.8103 (0.4412) acc1: 53.9062 (80.0558) acc5: 100.0000 (100.0000) acc1_neg: 51.302 acc5_neg: 100.000 time: 0.1677 data: 0.0739 max mem: 3535 +Test: [ 950/2387] eta: 0:04:35 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.8106 (0.4450) acc1: 53.6458 (79.7672) acc5: 100.0000 (100.0000) acc1_neg: 54.688 acc5_neg: 100.000 time: 0.1648 data: 0.0702 max mem: 3535 +Test: [ 960/2387] eta: 0:04:33 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.7928 (0.4486) acc1: 53.9062 (79.5019) acc5: 100.0000 (100.0000) acc1_neg: 55.208 acc5_neg: 100.000 time: 0.1726 data: 0.0761 max mem: 3535 +Test: [ 970/2387] eta: 0:04:31 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.7996 (0.4524) acc1: 53.6458 (79.2227) acc5: 100.0000 (100.0000) acc1_neg: 51.302 acc5_neg: 100.000 time: 0.1832 data: 0.0860 max mem: 3535 +Test: [ 980/2387] eta: 0:04:29 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.7851 (0.4514) acc1: 53.6458 (79.2750) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1729 data: 0.0798 max mem: 3535 +Test: [ 990/2387] eta: 0:04:27 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1330 (0.4483) acc1: 99.4792 (79.4736) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1799 data: 0.0867 max mem: 3535 +Test: [1000/2387] eta: 0:04:25 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1281 (0.4453) acc1: 99.4792 (79.6620) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1819 data: 0.0873 max mem: 3535 +Test: [1010/2387] eta: 0:04:23 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1281 (0.4423) acc1: 99.4792 (79.8544) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1791 data: 0.0860 max mem: 3535 +Test: [1020/2387] eta: 0:04:20 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1351 (0.4393) acc1: 99.7396 (80.0428) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1806 data: 0.0891 max mem: 3535 +Test: [1030/2387] eta: 0:04:19 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1274 (0.4363) acc1: 100.0000 (80.2338) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1942 data: 0.1034 max mem: 3535 +Test: [1040/2387] eta: 0:04:17 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1136 (0.4332) acc1: 100.0000 (80.4235) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1969 data: 0.1058 max mem: 3535 +Test: [1050/2387] eta: 0:04:15 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1102 (0.4301) acc1: 100.0000 (80.6092) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1888 data: 0.0984 max mem: 3535 +Test: [1060/2387] eta: 0:04:13 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1170 (0.4272) acc1: 100.0000 (80.7910) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1871 data: 0.0949 max mem: 3535 +Test: [1070/2387] eta: 0:04:11 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1190 (0.4244) acc1: 100.0000 (80.9696) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1999 data: 0.1039 max mem: 3535 +Test: [1080/2387] eta: 0:04:09 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1226 (0.4217) acc1: 100.0000 (81.1411) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2043 data: 0.1098 max mem: 3535 +Test: [1090/2387] eta: 0:04:08 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1246 (0.4190) acc1: 100.0000 (81.3128) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1857 data: 0.0917 max mem: 3535 +Test: [1100/2387] eta: 0:04:06 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1246 (0.4163) acc1: 100.0000 (81.4820) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.0991 max mem: 3535 +Test: [1110/2387] eta: 0:04:04 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1257 (0.4139) acc1: 100.0000 (81.6164) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1950 data: 0.0996 max mem: 3535 +Test: [1120/2387] eta: 0:04:02 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1265 (0.4113) acc1: 100.0000 (81.7790) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1851 data: 0.0905 max mem: 3536 +Test: [1130/2387] eta: 0:04:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1248 (0.4088) acc1: 100.0000 (81.9391) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1744 data: 0.0805 max mem: 3536 +Test: [1140/2387] eta: 0:03:58 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1213 (0.4063) acc1: 100.0000 (82.0958) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1861 data: 0.0932 max mem: 3536 +Test: [1150/2387] eta: 0:03:56 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1246 (0.4041) acc1: 100.0000 (82.2489) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2008 data: 0.1094 max mem: 3536 +Test: [1160/2387] eta: 0:03:54 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1412 (0.4018) acc1: 100.0000 (82.4005) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1852 data: 0.0939 max mem: 3536 +Test: [1170/2387] eta: 0:03:52 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1283 (0.3994) acc1: 100.0000 (82.5505) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1807 data: 0.0873 max mem: 3536 +Test: [1180/2387] eta: 0:03:50 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1364 (0.3973) acc1: 100.0000 (82.6974) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1832 data: 0.0884 max mem: 3536 +Test: [1190/2387] eta: 0:03:48 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1420 (0.3952) acc1: 100.0000 (82.8427) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1821 data: 0.0891 max mem: 3536 +Test: [1200/2387] eta: 0:03:46 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1406 (0.3929) acc1: 100.0000 (82.9855) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0970 max mem: 3536 +Test: [1210/2387] eta: 0:03:44 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1241 (0.3907) acc1: 100.0000 (83.1260) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1895 data: 0.0945 max mem: 3536 +Test: [1220/2387] eta: 0:03:42 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1121 (0.3883) acc1: 100.0000 (83.2638) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1923 data: 0.0991 max mem: 3536 +Test: [1230/2387] eta: 0:03:40 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1012 (0.3860) acc1: 100.0000 (83.3998) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2005 data: 0.1069 max mem: 3536 +Test: [1240/2387] eta: 0:03:39 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1010 (0.3837) acc1: 100.0000 (83.5335) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2036 data: 0.1099 max mem: 3536 +Test: [1250/2387] eta: 0:03:37 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1022 (0.3815) acc1: 100.0000 (83.6652) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1894 data: 0.0973 max mem: 3536 +Test: [1260/2387] eta: 0:03:35 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1067 (0.3793) acc1: 100.0000 (83.7947) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1901 data: 0.0975 max mem: 3536 +Test: [1270/2387] eta: 0:03:33 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1008 (0.3771) acc1: 100.0000 (83.9222) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1950 data: 0.1039 max mem: 3536 +Test: [1280/2387] eta: 0:03:31 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1012 (0.3750) acc1: 100.0000 (84.0473) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1736 data: 0.0813 max mem: 3536 +Test: [1290/2387] eta: 0:03:29 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1063 (0.3731) acc1: 100.0000 (84.1588) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1778 data: 0.0854 max mem: 3536 +Test: [1300/2387] eta: 0:03:27 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1038 (0.3710) acc1: 100.0000 (84.2765) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1820 data: 0.0907 max mem: 3536 +Test: [1310/2387] eta: 0:03:25 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1043 (0.3691) acc1: 100.0000 (84.3919) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1786 data: 0.0872 max mem: 3536 +Test: [1320/2387] eta: 0:03:23 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1179 (0.3673) acc1: 100.0000 (84.5020) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1883 data: 0.0952 max mem: 3536 +Test: [1330/2387] eta: 0:03:21 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1199 (0.3654) acc1: 100.0000 (84.6164) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1940 data: 0.1003 max mem: 3536 +Test: [1340/2387] eta: 0:03:19 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1291 (0.3641) acc1: 99.7396 (84.6919) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2020 data: 0.1109 max mem: 3536 +Test: [1350/2387] eta: 0:03:17 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0825 (0.3620) acc1: 100.0000 (84.8052) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1901 data: 0.0997 max mem: 3537 +Test: [1360/2387] eta: 0:03:15 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0749 (0.3599) acc1: 100.0000 (84.9169) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1942 data: 0.1011 max mem: 3537 +Test: [1370/2387] eta: 0:03:14 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0735 (0.3578) acc1: 100.0000 (85.0269) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2016 data: 0.1081 max mem: 3537 +Test: [1380/2387] eta: 0:03:12 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0749 (0.3558) acc1: 100.0000 (85.1353) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1860 data: 0.0943 max mem: 3537 +Test: [1390/2387] eta: 0:03:10 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0769 (0.3538) acc1: 100.0000 (85.2422) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1847 data: 0.0933 max mem: 3537 +Test: [1400/2387] eta: 0:03:08 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0777 (0.3518) acc1: 100.0000 (85.3475) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1832 data: 0.0918 max mem: 3537 +Test: [1410/2387] eta: 0:03:06 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0778 (0.3501) acc1: 100.0000 (85.4501) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1746 data: 0.0836 max mem: 3537 +Test: [1420/2387] eta: 0:03:04 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1175 (0.3487) acc1: 100.0000 (85.5318) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1707 data: 0.0806 max mem: 3537 +Test: [1430/2387] eta: 0:03:02 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1192 (0.3471) acc1: 100.0000 (85.6325) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1789 data: 0.0874 max mem: 3537 +Test: [1440/2387] eta: 0:03:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1192 (0.3455) acc1: 100.0000 (85.7317) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1881 data: 0.0962 max mem: 3537 +Test: [1450/2387] eta: 0:02:58 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1202 (0.3440) acc1: 100.0000 (85.8273) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1896 data: 0.0991 max mem: 3537 +Test: [1460/2387] eta: 0:02:56 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1110 (0.3424) acc1: 100.0000 (85.9234) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1894 data: 0.0985 max mem: 3537 +Test: [1470/2387] eta: 0:02:54 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0998 (0.3407) acc1: 100.0000 (86.0191) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1888 data: 0.0962 max mem: 3537 +Test: [1480/2387] eta: 0:02:52 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0947 (0.3391) acc1: 100.0000 (86.1133) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1872 data: 0.0949 max mem: 3537 +Test: [1490/2387] eta: 0:02:50 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0976 (0.3375) acc1: 100.0000 (86.2065) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2018 data: 0.1089 max mem: 3537 +Test: [1500/2387] eta: 0:02:49 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1027 (0.3359) acc1: 100.0000 (86.2977) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2101 data: 0.1166 max mem: 3537 +Test: [1510/2387] eta: 0:02:47 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1085 (0.3345) acc1: 100.0000 (86.3884) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2057 data: 0.1145 max mem: 3537 +Test: [1520/2387] eta: 0:02:45 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1205 (0.3333) acc1: 100.0000 (86.4751) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2038 data: 0.1115 max mem: 3537 +Test: [1530/2387] eta: 0:02:43 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1532 (0.3321) acc1: 100.0000 (86.5612) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2072 data: 0.1143 max mem: 3537 +Test: [1540/2387] eta: 0:02:41 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1447 (0.3309) acc1: 100.0000 (86.6478) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2090 data: 0.1165 max mem: 3537 +Test: [1550/2387] eta: 0:02:39 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1447 (0.3297) acc1: 100.0000 (86.7334) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1911 data: 0.0990 max mem: 3537 +Test: [1560/2387] eta: 0:02:37 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1525 (0.3286) acc1: 100.0000 (86.8173) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1863 data: 0.0953 max mem: 3537 +Test: [1570/2387] eta: 0:02:35 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1413 (0.3273) acc1: 100.0000 (86.9013) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1891 data: 0.0978 max mem: 3537 +Test: [1580/2387] eta: 0:02:33 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1248 (0.3260) acc1: 100.0000 (86.9836) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1830 data: 0.0914 max mem: 3538 +Test: [1590/2387] eta: 0:02:31 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1073 (0.3246) acc1: 100.0000 (87.0648) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1748 data: 0.0837 max mem: 3538 +Test: [1600/2387] eta: 0:02:30 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1017 (0.3234) acc1: 100.0000 (87.1243) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1773 data: 0.0867 max mem: 3538 +Test: [1610/2387] eta: 0:02:27 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0994 (0.3221) acc1: 100.0000 (87.2042) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1713 data: 0.0789 max mem: 3538 +Test: [1620/2387] eta: 0:02:25 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1099 (0.3208) acc1: 100.0000 (87.2820) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1628 data: 0.0698 max mem: 3538 +Test: [1630/2387] eta: 0:02:23 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1157 (0.3196) acc1: 99.7396 (87.3557) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1644 data: 0.0721 max mem: 3538 +Test: [1640/2387] eta: 0:02:21 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1170 (0.3184) acc1: 99.7396 (87.4300) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1640 data: 0.0726 max mem: 3538 +Test: [1650/2387] eta: 0:02:19 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1170 (0.3172) acc1: 99.7396 (87.5016) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1605 data: 0.0693 max mem: 3538 +Test: [1660/2387] eta: 0:02:17 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1170 (0.3160) acc1: 99.7396 (87.5756) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1572 data: 0.0663 max mem: 3538 +Test: [1670/2387] eta: 0:02:15 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1134 (0.3148) acc1: 99.7396 (87.6465) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1602 data: 0.0700 max mem: 3538 +Test: [1680/2387] eta: 0:02:13 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1129 (0.3137) acc1: 99.7396 (87.7135) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1646 data: 0.0725 max mem: 3538 +Test: [1690/2387] eta: 0:02:11 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1158 (0.3126) acc1: 100.0000 (87.7784) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1700 data: 0.0763 max mem: 3538 +Test: [1700/2387] eta: 0:02:09 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1167 (0.3115) acc1: 99.7396 (87.8480) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1775 data: 0.0841 max mem: 3538 +Test: [1710/2387] eta: 0:02:08 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1149 (0.3103) acc1: 100.0000 (87.9186) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1818 data: 0.0885 max mem: 3538 +Test: [1720/2387] eta: 0:02:06 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1115 (0.3091) acc1: 100.0000 (87.9883) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1705 data: 0.0788 max mem: 3538 +Test: [1730/2387] eta: 0:02:04 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1115 (0.3080) acc1: 100.0000 (88.0572) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1736 data: 0.0828 max mem: 3538 +Test: [1740/2387] eta: 0:02:02 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1117 (0.3069) acc1: 100.0000 (88.1254) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1748 data: 0.0834 max mem: 3538 +Test: [1750/2387] eta: 0:02:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1141 (0.3058) acc1: 100.0000 (88.1932) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1662 data: 0.0753 max mem: 3538 +Test: [1760/2387] eta: 0:01:58 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1117 (0.3047) acc1: 100.0000 (88.2603) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1695 data: 0.0779 max mem: 3538 +Test: [1770/2387] eta: 0:01:56 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1092 (0.3036) acc1: 100.0000 (88.3260) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1767 data: 0.0845 max mem: 3538 +Test: [1780/2387] eta: 0:01:54 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1086 (0.3025) acc1: 100.0000 (88.3915) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1835 data: 0.0922 max mem: 3538 +Test: [1790/2387] eta: 0:01:52 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1018 (0.3014) acc1: 100.0000 (88.4560) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1771 data: 0.0868 max mem: 3538 +Test: [1800/2387] eta: 0:01:50 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1005 (0.3003) acc1: 100.0000 (88.5198) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1780 data: 0.0863 max mem: 3538 +Test: [1810/2387] eta: 0:01:48 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1058 (0.2992) acc1: 100.0000 (88.5829) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1816 data: 0.0890 max mem: 3539 +Test: [1820/2387] eta: 0:01:46 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1034 (0.2982) acc1: 100.0000 (88.6456) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1633 data: 0.0719 max mem: 3539 +Test: [1830/2387] eta: 0:01:44 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1034 (0.2972) acc1: 100.0000 (88.7074) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1678 data: 0.0748 max mem: 3539 +Test: [1840/2387] eta: 0:01:42 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1493 (0.2964) acc1: 100.0000 (88.7681) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1805 data: 0.0842 max mem: 3539 +Test: [1850/2387] eta: 0:01:41 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1521 (0.2957) acc1: 100.0000 (88.8288) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1906 data: 0.0962 max mem: 3539 +Test: [1860/2387] eta: 0:01:39 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1516 (0.2949) acc1: 100.0000 (88.8887) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1987 data: 0.1065 max mem: 3539 +Test: [1870/2387] eta: 0:01:37 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1509 (0.2941) acc1: 100.0000 (88.9475) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1773 data: 0.0854 max mem: 3539 +Test: [1880/2387] eta: 0:01:35 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1508 (0.2933) acc1: 100.0000 (89.0062) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1900 data: 0.0980 max mem: 3539 +Test: [1890/2387] eta: 0:01:33 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1512 (0.2926) acc1: 100.0000 (89.0643) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1768 data: 0.0841 max mem: 3539 +Test: [1900/2387] eta: 0:01:31 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1519 (0.2919) acc1: 100.0000 (89.1218) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1668 data: 0.0748 max mem: 3539 +Test: [1910/2387] eta: 0:01:29 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1558 (0.2911) acc1: 100.0000 (89.1786) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0987 max mem: 3539 +Test: [1920/2387] eta: 0:01:27 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1308 (0.2903) acc1: 100.0000 (89.2348) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1842 data: 0.0936 max mem: 3539 +Test: [1930/2387] eta: 0:01:25 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1146 (0.2893) acc1: 100.0000 (89.2906) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1864 data: 0.0957 max mem: 3539 +Test: [1940/2387] eta: 0:01:23 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1081 (0.2887) acc1: 100.0000 (89.3114) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1926 data: 0.1015 max mem: 3539 +Test: [1950/2387] eta: 0:01:22 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1123 (0.2878) acc1: 100.0000 (89.3662) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1983 data: 0.1072 max mem: 3539 +Test: [1960/2387] eta: 0:01:20 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1199 (0.2870) acc1: 100.0000 (89.4204) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0873 max mem: 3539 +Test: [1970/2387] eta: 0:01:18 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1399 (0.2863) acc1: 100.0000 (89.4692) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1722 data: 0.0790 max mem: 3539 +Test: [1980/2387] eta: 0:01:16 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1643 (0.2859) acc1: 99.2188 (89.5017) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1770 data: 0.0864 max mem: 3539 +Test: [1990/2387] eta: 0:01:14 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2531 (0.2858) acc1: 93.7500 (89.5217) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1730 data: 0.0815 max mem: 3539 +Test: [2000/2387] eta: 0:01:12 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2712 (0.2857) acc1: 93.2292 (89.5383) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1811 data: 0.0879 max mem: 3539 +Test: [2010/2387] eta: 0:01:10 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2725 (0.2857) acc1: 92.7083 (89.5519) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1796 data: 0.0876 max mem: 3539 +Test: [2020/2387] eta: 0:01:08 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2771 (0.2856) acc1: 91.9271 (89.5665) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1921 data: 0.1013 max mem: 3539 +Test: [2030/2387] eta: 0:01:06 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2733 (0.2856) acc1: 92.1875 (89.5801) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1922 data: 0.1003 max mem: 3540 +Test: [2040/2387] eta: 0:01:05 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2731 (0.2855) acc1: 92.7083 (89.5978) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1794 data: 0.0858 max mem: 3540 +Test: [2050/2387] eta: 0:01:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2651 (0.2854) acc1: 93.7500 (89.6193) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1775 data: 0.0833 max mem: 3540 +Test: [2060/2387] eta: 0:01:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2411 (0.2851) acc1: 94.5312 (89.6449) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1710 data: 0.0784 max mem: 3540 +Test: [2070/2387] eta: 0:00:59 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2358 (0.2849) acc1: 95.0521 (89.6710) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1749 data: 0.0834 max mem: 3540 +Test: [2080/2387] eta: 0:00:57 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2257 (0.2846) acc1: 95.3125 (89.6987) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1788 data: 0.0871 max mem: 3540 +Test: [2090/2387] eta: 0:00:55 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2272 (0.2843) acc1: 95.3125 (89.7263) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1758 data: 0.0842 max mem: 3540 +Test: [2100/2387] eta: 0:00:53 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2331 (0.2841) acc1: 95.0521 (89.7512) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1805 data: 0.0854 max mem: 3540 +Test: [2110/2387] eta: 0:00:51 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2280 (0.2838) acc1: 95.0521 (89.7791) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1865 data: 0.0888 max mem: 3540 +Test: [2120/2387] eta: 0:00:50 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1991 (0.2832) acc1: 96.6146 (89.8213) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1842 data: 0.0902 max mem: 3540 +Test: [2130/2387] eta: 0:00:48 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1362 (0.2825) acc1: 100.0000 (89.8671) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1853 data: 0.0942 max mem: 3540 +Test: [2140/2387] eta: 0:00:46 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1301 (0.2820) acc1: 99.7396 (89.8912) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1853 data: 0.0937 max mem: 3540 +Test: [2150/2387] eta: 0:00:44 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1281 (0.2813) acc1: 99.7396 (89.9373) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1886 data: 0.0949 max mem: 3540 +Test: [2160/2387] eta: 0:00:42 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1296 (0.2806) acc1: 100.0000 (89.9833) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1958 data: 0.1015 max mem: 3540 +Test: [2170/2387] eta: 0:00:40 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1374 (0.2800) acc1: 100.0000 (90.0266) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1933 data: 0.1011 max mem: 3540 +Test: [2180/2387] eta: 0:00:38 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1463 (0.2807) acc1: 98.9583 (89.9783) acc5: 100.0000 (100.0000) acc1_neg: 48.438 acc5_neg: 100.000 time: 0.1856 data: 0.0943 max mem: 3540 +Test: [2190/2387] eta: 0:00:36 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.3398 (0.2814) acc1: 81.2500 (89.9235) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1857 data: 0.0945 max mem: 3540 +Test: [2200/2387] eta: 0:00:35 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1142 (0.2806) acc1: 100.0000 (89.9693) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1878 data: 0.0942 max mem: 3540 +Test: [2210/2387] eta: 0:00:33 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1032 (0.2798) acc1: 100.0000 (90.0147) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1889 data: 0.0912 max mem: 3540 +Test: [2220/2387] eta: 0:00:31 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1008 (0.2790) acc1: 100.0000 (90.0589) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1884 data: 0.0928 max mem: 3540 +Test: [2230/2387] eta: 0:00:29 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1014 (0.2783) acc1: 100.0000 (90.1035) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1853 data: 0.0929 max mem: 3540 +Test: [2240/2387] eta: 0:00:27 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1035 (0.2775) acc1: 100.0000 (90.1476) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1817 data: 0.0894 max mem: 3540 +Test: [2250/2387] eta: 0:00:25 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1140 (0.2768) acc1: 100.0000 (90.1912) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1865 data: 0.0941 max mem: 3540 +Test: [2260/2387] eta: 0:00:23 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1204 (0.2761) acc1: 100.0000 (90.2344) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2259 data: 0.1336 max mem: 3541 +Test: [2270/2387] eta: 0:00:21 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1123 (0.2754) acc1: 100.0000 (90.2774) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2034 data: 0.1123 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1029 (0.2749) acc1: 100.0000 (90.3007) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1539 data: 0.0621 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1049 (0.2741) acc1: 100.0000 (90.3430) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1828 data: 0.0911 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1122 (0.2735) acc1: 100.0000 (90.3845) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1808 data: 0.0869 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1173 (0.2728) acc1: 100.0000 (90.4251) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1750 data: 0.0810 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1161 (0.2722) acc1: 100.0000 (90.4635) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1744 data: 0.0837 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1237 (0.2716) acc1: 100.0000 (90.5041) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1530 data: 0.0624 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1219 (0.2709) acc1: 100.0000 (90.5446) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1531 data: 0.0628 max mem: 3541 +Test: [2350/2387] eta: 0:00:06 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1194 (0.2706) acc1: 100.0000 (90.5636) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1531 data: 0.0633 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1250 (0.2700) acc1: 100.0000 (90.6036) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1431 data: 0.0535 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1307 (0.2694) acc1: 100.0000 (90.6425) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1380 data: 0.0484 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1307 (0.2689) acc1: 100.0000 (90.6812) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1411 data: 0.0512 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1377 (0.2698) acc1: 99.7396 (90.6413) acc5: 100.0000 (100.0000) acc1_neg: 46.512 acc5_neg: 100.000 time: 0.1344 data: 0.0471 max mem: 3541 +Test: Total time: 0:07:24 (0.1862 s / it) +* Overall Acc@1 90.641 Acc@5 100.000 + target == real Acc@1 1.856 Acc@5 100.000 + target == fake Acc@1 96.876 Acc@5 100.000 + loss 0.270 + PR-AUC 0.9841 ROC-AUC 0.8027 EER 25.55% (thr=0.8501) +Accuracy of the network on the 916267 test images: 90.6% +[rank0]:[W726 06:31:34.205286893 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 06:31:55.377000 3169094 torch/distributed/run.py:766] +W0726 06:31:55.377000 3169094 torch/distributed/run.py:766] ***************************************** +W0726 06:31:55.377000 3169094 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:31:55.377000 3169094 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:32:31.598703919 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:32:31.081806235 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 06:32:32.240779779 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:32:32.454050884 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:32:32.586800130 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:32:32.619669824 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:32:32.649440811 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:32:32.686658578 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n11_seed9_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed9_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:08:07 acc1_pos: 1.823 acc5_pos: 100.000 loss: 1.6874 (1.6874) acc1: 1.8229 (1.8229) acc5: 100.0000 (100.0000) time: 7.9059 data: 5.5839 max mem: 3531 +Test: [ 10/517] eta: 0:06:45 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.6435 (1.6195) acc1: 0.7812 (2.4858) acc5: 100.0000 (100.0000) time: 0.8004 data: 0.5080 max mem: 3531 +Test: [ 20/517] eta: 0:03:51 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.6435 (1.6408) acc1: 0.7812 (2.7034) acc5: 100.0000 (100.0000) time: 0.0944 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:03:24 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.6746 (1.6578) acc1: 0.7812 (2.0077) acc5: 100.0000 (100.0000) time: 0.2105 data: 0.1146 max mem: 3531 +Test: [ 40/517] eta: 0:03:13 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.6560 (1.6458) acc1: 0.0000 (1.6514) acc5: 100.0000 (100.0000) time: 0.3399 data: 0.2466 max mem: 3531 +Test: [ 50/517] eta: 0:02:56 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.5467 (1.5670) acc1: 0.5208 (6.4134) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.3139 data: 0.2161 max mem: 3531 +Test: [ 60/517] eta: 0:02:43 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1949 (1.3407) acc1: 99.7396 (21.7341) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2599 data: 0.1605 max mem: 3531 +Test: [ 70/517] eta: 0:02:33 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1912 (1.1808) acc1: 100.0000 (32.7538) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1577 max mem: 3531 +Test: [ 80/517] eta: 0:02:25 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2085 (1.0624) acc1: 100.0000 (40.9915) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2600 data: 0.1634 max mem: 3531 +Test: [ 90/517] eta: 0:02:18 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1942 (0.9657) acc1: 100.0000 (47.4588) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2555 data: 0.1601 max mem: 3531 +Test: [100/517] eta: 0:02:12 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1821 (0.8884) acc1: 100.0000 (52.6506) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2548 data: 0.1589 max mem: 3531 +Test: [110/517] eta: 0:02:06 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1857 (0.8258) acc1: 100.0000 (56.8741) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2546 data: 0.1585 max mem: 3531 +Test: [120/517] eta: 0:02:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1975 (0.7744) acc1: 100.0000 (60.3908) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2548 data: 0.1586 max mem: 3531 +Test: [130/517] eta: 0:01:57 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1921 (0.7299) acc1: 100.0000 (63.4144) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2546 data: 0.1584 max mem: 3531 +Test: [140/517] eta: 0:01:52 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1852 (0.6909) acc1: 100.0000 (66.0092) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2410 data: 0.1455 max mem: 3531 +Test: [150/517] eta: 0:01:47 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1824 (0.6574) acc1: 100.0000 (68.2568) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2298 data: 0.1335 max mem: 3531 +Test: [160/517] eta: 0:01:43 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1841 (0.6280) acc1: 100.0000 (70.2284) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2288 data: 0.1321 max mem: 3531 +Test: [170/517] eta: 0:01:39 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1816 (0.6018) acc1: 100.0000 (71.9679) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2299 data: 0.1324 max mem: 3531 +Test: [180/517] eta: 0:01:35 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1743 (0.5779) acc1: 100.0000 (73.5166) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2299 data: 0.1318 max mem: 3531 +Test: [190/517] eta: 0:01:31 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1768 (0.5571) acc1: 100.0000 (74.9032) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2256 data: 0.1292 max mem: 3531 +Test: [200/517] eta: 0:01:28 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1710 (0.5375) acc1: 100.0000 (76.1518) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2311 data: 0.1361 max mem: 3531 +Test: [210/517] eta: 0:01:24 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1855 (0.5219) acc1: 100.0000 (77.2697) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2442 data: 0.1503 max mem: 3532 +Test: [220/517] eta: 0:01:21 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2072 (0.5076) acc1: 99.7396 (78.2758) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2401 data: 0.1456 max mem: 3532 +Test: [230/517] eta: 0:01:18 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2001 (0.4944) acc1: 100.0000 (79.2084) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2550 data: 0.1608 max mem: 3532 +Test: [240/517] eta: 0:01:15 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1963 (0.4822) acc1: 100.0000 (80.0646) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2636 data: 0.1685 max mem: 3532 +Test: [250/517] eta: 0:01:12 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2022 (0.4714) acc1: 100.0000 (80.8454) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2398 data: 0.1429 max mem: 3532 +Test: [260/517] eta: 0:01:10 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2024 (0.4612) acc1: 100.0000 (81.5793) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2648 data: 0.1682 max mem: 3532 +Test: [270/517] eta: 0:01:08 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2014 (0.4515) acc1: 100.0000 (82.2552) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.3499 data: 0.2521 max mem: 3532 +Test: [280/517] eta: 0:01:06 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1959 (0.4425) acc1: 100.0000 (82.8848) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.3600 data: 0.2613 max mem: 3532 +Test: [290/517] eta: 0:01:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1971 (0.4342) acc1: 100.0000 (83.4711) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.3349 data: 0.2364 max mem: 3532 +Test: [300/517] eta: 0:01:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1891 (0.4257) acc1: 100.0000 (84.0185) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.3399 data: 0.2412 max mem: 3532 +Test: [310/517] eta: 0:00:58 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1795 (0.4180) acc1: 100.0000 (84.5324) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2749 data: 0.1767 max mem: 3532 +Test: [320/517] eta: 0:00:55 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1892 (0.4110) acc1: 100.0000 (85.0143) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2399 data: 0.1413 max mem: 3532 +Test: [330/517] eta: 0:00:52 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1897 (0.4041) acc1: 100.0000 (85.4670) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2399 data: 0.1410 max mem: 3532 +Test: [340/517] eta: 0:00:49 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1783 (0.3973) acc1: 100.0000 (85.8917) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2244 data: 0.1287 max mem: 3532 +Test: [350/517] eta: 0:00:46 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1735 (0.3911) acc1: 100.0000 (86.2936) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2287 data: 0.1340 max mem: 3532 +Test: [360/517] eta: 0:00:43 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1782 (0.3853) acc1: 100.0000 (86.6733) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2296 data: 0.1336 max mem: 3532 +Test: [370/517] eta: 0:00:40 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1782 (0.3797) acc1: 100.0000 (87.0325) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2449 data: 0.1479 max mem: 3532 +Test: [380/517] eta: 0:00:37 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1809 (0.3747) acc1: 100.0000 (87.3729) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2445 data: 0.1470 max mem: 3532 +Test: [390/517] eta: 0:00:34 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1942 (0.3702) acc1: 100.0000 (87.6958) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2448 data: 0.1482 max mem: 3532 +Test: [400/517] eta: 0:00:31 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1975 (0.3658) acc1: 100.0000 (88.0026) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2363 data: 0.1400 max mem: 3532 +Test: [410/517] eta: 0:00:28 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2016 (0.3620) acc1: 100.0000 (88.2946) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2349 data: 0.1399 max mem: 3532 +Test: [420/517] eta: 0:00:26 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2050 (0.3582) acc1: 100.0000 (88.5726) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2436 data: 0.1485 max mem: 3532 +Test: [430/517] eta: 0:00:23 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2021 (0.3549) acc1: 100.0000 (88.8347) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2499 data: 0.1540 max mem: 3532 +Test: [440/517] eta: 0:00:20 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2083 (0.3519) acc1: 100.0000 (89.0826) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2499 data: 0.1524 max mem: 3533 +Test: [450/517] eta: 0:00:17 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2198 (0.3489) acc1: 100.0000 (89.3246) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2310 data: 0.1337 max mem: 3533 +Test: [460/517] eta: 0:00:15 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2198 (0.3463) acc1: 100.0000 (89.5562) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2399 data: 0.1452 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2057 (0.3432) acc1: 100.0000 (89.7780) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2400 data: 0.1433 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1982 (0.3402) acc1: 100.0000 (89.9905) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2364 data: 0.1380 max mem: 3533 +Test: [490/517] eta: 0:00:07 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1982 (0.3376) acc1: 100.0000 (90.1890) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2432 data: 0.1480 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1786 (0.3342) acc1: 100.0000 (90.3849) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1975 data: 0.1057 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1725 (0.3316) acc1: 100.0000 (90.5689) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1520 data: 0.0621 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1698 (0.3296) acc1: 100.0000 (90.6736) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1535 data: 0.0621 max mem: 3533 +Test: Total time: 0:02:14 (0.2599 s / it) +* Overall Acc@1 90.674 Acc@5 100.000 + target == real Acc@1 1.727 Acc@5 100.000 + target == fake Acc@1 99.901 Acc@5 100.000 + loss 0.330 + PR-AUC 0.9336 ROC-AUC 0.6316 EER 40.75% (thr=0.8242) +Accuracy of the network on the 198427 test images: 90.7% +[rank0]:[W726 06:34:55.219886332 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 06:35:14.081000 3169667 torch/distributed/run.py:766] +W0726 06:35:14.081000 3169667 torch/distributed/run.py:766] ***************************************** +W0726 06:35:14.081000 3169667 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:35:14.081000 3169667 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:35:40.568906405 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:35:40.874180862 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W726 06:35:40.878665298 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:35:41.222361685 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:35:41.449393519 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +[rank0]:[W726 06:35:41.479386658 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:35:41.484862525 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:35:41.512259673 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n11_seed9_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed9_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:36:41 acc1_pos: 42.969 acc5_pos: 100.000 loss: 0.7466 (0.7466) acc1: 42.9688 (42.9688) acc5: 100.0000 (100.0000) time: 5.6605 data: 3.7906 max mem: 3531 +Test: [ 10/389] eta: 0:03:47 acc1_pos: 79.427 acc5_pos: 100.000 loss: 0.6889 (0.7083) acc1: 49.4792 (49.8816) acc5: 100.0000 (100.0000) time: 0.6015 data: 0.3451 max mem: 3531 +Test: [ 20/389] eta: 0:02:13 acc1_pos: 55.990 acc5_pos: 100.000 loss: 0.6692 (0.6783) acc1: 53.1250 (53.5466) acc5: 100.0000 (100.0000) time: 0.0974 data: 0.0005 max mem: 3531 +Test: [ 30/389] eta: 0:01:39 acc1_pos: 69.792 acc5_pos: 100.000 loss: 0.6477 (0.6706) acc1: 60.4167 (56.5776) acc5: 100.0000 (100.0000) time: 0.0996 data: 0.0005 max mem: 3531 +Test: [ 40/389] eta: 0:01:25 acc1_pos: 10.677 acc5_pos: 100.000 loss: 0.5943 (0.6714) acc1: 69.7917 (57.2917) acc5: 100.0000 (100.0000) time: 0.1206 data: 0.0254 max mem: 3531 +Test: [ 50/389] eta: 0:01:17 acc1_pos: 97.656 acc5_pos: 100.000 loss: 0.5005 (0.6326) acc1: 72.1354 (61.0498) acc5: 100.0000 (100.0000) time: 0.1499 data: 0.0570 max mem: 3531 +Test: [ 60/389] eta: 0:01:09 acc1_pos: 55.469 acc5_pos: 100.000 loss: 0.4263 (0.5947) acc1: 84.1146 (65.2066) acc5: 100.0000 (100.0000) time: 0.1444 data: 0.0512 max mem: 3531 +Test: [ 70/389] eta: 0:01:05 acc1_pos: 80.990 acc5_pos: 100.000 loss: 0.5025 (0.5896) acc1: 79.9479 (65.9845) acc5: 100.0000 (100.0000) time: 0.1456 data: 0.0550 max mem: 3531 +Test: [ 80/389] eta: 0:01:01 acc1_pos: 87.760 acc5_pos: 100.000 loss: 0.5762 (0.5831) acc1: 71.0938 (67.1843) acc5: 100.0000 (100.0000) time: 0.1617 data: 0.0698 max mem: 3531 +Test: [ 90/389] eta: 0:00:58 acc1_pos: 27.865 acc5_pos: 100.000 loss: 0.4498 (0.5700) acc1: 85.4167 (68.2006) acc5: 100.0000 (100.0000) time: 0.1717 data: 0.0797 max mem: 3531 +Test: [100/389] eta: 0:00:55 acc1_pos: 72.656 acc5_pos: 100.000 loss: 0.6506 (0.5845) acc1: 55.9896 (66.3289) acc5: 100.0000 (100.0000) time: 0.1681 data: 0.0778 max mem: 3531 +Test: [110/389] eta: 0:00:53 acc1_pos: 99.740 acc5_pos: 100.000 loss: 0.7109 (0.5907) acc1: 53.1250 (65.5054) acc5: 100.0000 (100.0000) time: 0.1622 data: 0.0689 max mem: 3531 +Test: [120/389] eta: 0:00:50 acc1_pos: 69.792 acc5_pos: 100.000 loss: 0.6160 (0.5876) acc1: 64.3229 (65.6013) acc5: 100.0000 (100.0000) time: 0.1691 data: 0.0705 max mem: 3531 +Test: [130/389] eta: 0:00:48 acc1_pos: 84.049 acc5_pos: 100.000 loss: 0.6322 (0.5964) acc1: 60.9375 (65.4123) acc5: 100.0000 (100.0000) acc1_neg: 59.375 acc5_neg: 100.000 time: 0.1695 data: 0.0693 max mem: 3531 +Test: [140/389] eta: 0:00:46 acc1_pos: 84.049 acc5_pos: 100.000 loss: 0.6322 (0.6104) acc1: 61.4583 (65.0691) acc5: 100.0000 (100.0000) acc1_neg: 1.042 acc5_neg: 100.000 time: 0.1660 data: 0.0659 max mem: 3531 +Test: [150/389] eta: 0:00:44 acc1_pos: 84.049 acc5_pos: 100.000 loss: 0.5856 (0.6230) acc1: 72.3958 (64.3471) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1749 data: 0.0770 max mem: 3531 +Test: [160/389] eta: 0:00:42 acc1_pos: 84.049 acc5_pos: 100.000 loss: 0.5856 (0.6220) acc1: 72.3958 (64.8421) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1787 data: 0.0820 max mem: 3531 +Test: [170/389] eta: 0:00:39 acc1_pos: 84.049 acc5_pos: 100.000 loss: 0.4813 (0.6179) acc1: 90.8854 (65.6113) acc5: 100.0000 (100.0000) acc1_neg: 30.208 acc5_neg: 100.000 time: 0.1567 data: 0.0585 max mem: 3531 +Test: [180/389] eta: 0:00:38 acc1_pos: 84.049 acc5_pos: 100.000 loss: 0.4905 (0.6210) acc1: 69.7917 (65.3315) acc5: 100.0000 (100.0000) acc1_neg: 65.104 acc5_neg: 100.000 time: 0.1786 data: 0.0842 max mem: 3531 +Test: [190/389] eta: 0:00:36 acc1_pos: 84.049 acc5_pos: 100.000 loss: 0.6609 (0.6251) acc1: 61.7188 (64.7415) acc5: 100.0000 (100.0000) acc1_neg: 11.979 acc5_neg: 100.000 time: 0.1942 data: 0.1031 max mem: 3531 +Test: [200/389] eta: 0:00:33 acc1_pos: 84.049 acc5_pos: 100.000 loss: 0.6132 (0.6269) acc1: 61.9792 (64.5678) acc5: 100.0000 (100.0000) acc1_neg: 47.917 acc5_neg: 100.000 time: 0.1361 data: 0.0421 max mem: 3531 +Test: [210/389] eta: 0:00:31 acc1_pos: 84.049 acc5_pos: 100.000 loss: 0.6171 (0.6260) acc1: 67.1875 (64.6463) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1306 data: 0.0352 max mem: 3532 +Test: [220/389] eta: 0:00:30 acc1_pos: 84.049 acc5_pos: 100.000 loss: 0.5036 (0.6162) acc1: 88.2812 (65.9750) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1709 data: 0.0782 max mem: 3532 +Test: [230/389] eta: 0:00:28 acc1_pos: 84.049 acc5_pos: 100.000 loss: 0.5209 (0.6194) acc1: 88.2812 (65.7028) acc5: 100.0000 (100.0000) acc1_neg: 25.781 acc5_neg: 100.000 time: 0.1632 data: 0.0712 max mem: 3532 +Test: [240/389] eta: 0:00:26 acc1_pos: 84.049 acc5_pos: 100.000 loss: 0.6315 (0.6220) acc1: 62.2396 (65.4435) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.1439 data: 0.0499 max mem: 3532 +Test: [250/389] eta: 0:00:24 acc1_pos: 84.049 acc5_pos: 100.000 loss: 0.5700 (0.6160) acc1: 78.3854 (66.5525) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1548 data: 0.0586 max mem: 3532 +Test: [260/389] eta: 0:00:22 acc1_pos: 84.049 acc5_pos: 100.000 loss: 0.5238 (0.6189) acc1: 87.2396 (66.3244) acc5: 100.0000 (100.0000) acc1_neg: 57.552 acc5_neg: 100.000 time: 0.1599 data: 0.0633 max mem: 3532 +Test: [270/389] eta: 0:00:20 acc1_pos: 84.049 acc5_pos: 100.000 loss: 0.6801 (0.6227) acc1: 57.5521 (65.8566) acc5: 100.0000 (100.0000) acc1_neg: 73.698 acc5_neg: 100.000 time: 0.1406 data: 0.0415 max mem: 3532 +Test: [280/389] eta: 0:00:18 acc1_pos: 84.049 acc5_pos: 100.000 loss: 0.6606 (0.6271) acc1: 59.1146 (65.2413) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.1349 data: 0.0366 max mem: 3532 +Test: [290/389] eta: 0:00:16 acc1_pos: 84.049 acc5_pos: 100.000 loss: 0.7786 (0.6355) acc1: 43.2292 (64.2665) acc5: 100.0000 (100.0000) acc1_neg: 20.833 acc5_neg: 100.000 time: 0.1397 data: 0.0449 max mem: 3532 +Test: [300/389] eta: 0:00:15 acc1_pos: 84.049 acc5_pos: 100.000 loss: 0.7627 (0.6381) acc1: 45.3125 (63.9656) acc5: 100.0000 (100.0000) acc1_neg: 36.719 acc5_neg: 100.000 time: 0.1374 data: 0.0433 max mem: 3532 +Test: [310/389] eta: 0:00:13 acc1_pos: 84.049 acc5_pos: 100.000 loss: 0.6764 (0.6415) acc1: 53.1250 (63.5551) acc5: 100.0000 (100.0000) acc1_neg: 53.125 acc5_neg: 100.000 time: 0.1355 data: 0.0429 max mem: 3532 +Test: [320/389] eta: 0:00:11 acc1_pos: 84.049 acc5_pos: 100.000 loss: 0.6269 (0.6399) acc1: 67.1875 (63.8848) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.1442 data: 0.0519 max mem: 3532 +Test: [330/389] eta: 0:00:09 acc1_pos: 84.049 acc5_pos: 100.000 loss: 0.6324 (0.6404) acc1: 67.4479 (63.8768) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.1568 data: 0.0626 max mem: 3532 +Test: [340/389] eta: 0:00:08 acc1_pos: 84.049 acc5_pos: 100.000 loss: 0.6227 (0.6363) acc1: 67.4479 (64.3290) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.1681 data: 0.0741 max mem: 3532 +Test: [350/389] eta: 0:00:06 acc1_pos: 84.049 acc5_pos: 100.000 loss: 0.5832 (0.6518) acc1: 67.9688 (63.5869) acc5: 100.0000 (100.0000) acc1_neg: 0.260 acc5_neg: 100.000 time: 0.1498 data: 0.0577 max mem: 3532 +Test: [360/389] eta: 0:00:04 acc1_pos: 84.049 acc5_pos: 100.000 loss: 1.2384 (0.6718) acc1: 21.6146 (62.3233) acc5: 100.0000 (100.0000) acc1_neg: 21.615 acc5_neg: 100.000 time: 0.1399 data: 0.0476 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 84.049 acc5_pos: 100.000 loss: 0.8721 (0.6711) acc1: 26.0417 (62.3189) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1484 data: 0.0563 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 84.049 acc5_pos: 100.000 loss: 0.6854 (0.6737) acc1: 53.1250 (62.1084) acc5: 100.0000 (100.0000) acc1_neg: 15.104 acc5_neg: 100.000 time: 0.1338 data: 0.0420 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 84.049 acc5_pos: 100.000 loss: 0.6857 (0.6742) acc1: 57.8125 (61.9880) acc5: 100.0000 (100.0000) acc1_neg: 25.683 acc5_neg: 100.000 time: 0.1343 data: 0.0381 max mem: 3532 +Test: Total time: 0:01:04 (0.1649 s / it) +* Overall Acc@1 61.988 Acc@5 100.000 + target == real Acc@1 66.045 Acc@5 100.000 + target == fake Acc@1 60.102 Acc@5 100.000 + loss 0.674 + PR-AUC 0.8019 ROC-AUC 0.6792 EER 36.89% (thr=0.4856) +Accuracy of the network on the 149358 test images: 62.0% +[rank0]:[W726 06:37:04.962452701 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 06:37:23.079000 3170225 torch/distributed/run.py:766] +W0726 06:37:23.079000 3170225 torch/distributed/run.py:766] ***************************************** +W0726 06:37:23.079000 3170225 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:37:23.079000 3170225 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:37:52.705046970 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:37:52.842933818 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:37:52.984114661 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:37:53.182121545 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:37:53.285995242 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:37:53.363261755 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 06:37:53.433695879 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:37:53.447578498 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n11_seed9_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed9_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:34:25 acc1_pos: 54.948 acc5_pos: 100.000 loss: 0.8235 (0.8235) acc1: 54.9479 (54.9479) acc5: 100.0000 (100.0000) time: 6.2017 data: 4.6500 max mem: 3531 +Test: [ 10/333] eta: 0:03:33 acc1_pos: 80.208 acc5_pos: 100.000 loss: 0.6280 (0.5968) acc1: 66.1458 (68.8447) acc5: 100.0000 (100.0000) time: 0.6598 data: 0.4348 max mem: 3531 +Test: [ 20/333] eta: 0:02:04 acc1_pos: 89.062 acc5_pos: 100.000 loss: 0.5048 (0.5225) acc1: 73.6979 (72.6190) acc5: 100.0000 (100.0000) time: 0.1068 data: 0.0134 max mem: 3531 +Test: [ 30/333] eta: 0:01:56 acc1_pos: 36.979 acc5_pos: 100.000 loss: 0.5585 (0.6946) acc1: 65.6250 (59.1314) acc5: 100.0000 (100.0000) time: 0.2352 data: 0.1425 max mem: 3531 +Test: [ 40/333] eta: 0:01:43 acc1_pos: 25.781 acc5_pos: 100.000 loss: 1.1246 (0.7980) acc1: 25.7812 (51.0480) acc5: 100.0000 (100.0000) time: 0.3099 data: 0.2185 max mem: 3531 +Test: [ 50/333] eta: 0:01:32 acc1_pos: 27.865 acc5_pos: 100.000 loss: 1.1242 (0.8621) acc1: 23.6979 (45.9712) acc5: 100.0000 (100.0000) time: 0.2344 data: 0.1419 max mem: 3531 +Test: [ 60/333] eta: 0:01:21 acc1_pos: 19.792 acc5_pos: 100.000 loss: 1.1405 (0.9078) acc1: 23.6979 (42.5162) acc5: 100.0000 (100.0000) time: 0.1799 data: 0.0863 max mem: 3531 +Test: [ 70/333] eta: 0:01:16 acc1_pos: 15.104 acc5_pos: 100.000 loss: 1.1497 (0.9396) acc1: 24.4792 (40.2619) acc5: 100.0000 (100.0000) time: 0.2049 data: 0.1115 max mem: 3531 +Test: [ 80/333] eta: 0:01:10 acc1_pos: 28.906 acc5_pos: 100.000 loss: 1.0709 (0.9564) acc1: 27.0833 (38.5545) acc5: 100.0000 (100.0000) time: 0.2271 data: 0.1340 max mem: 3531 +Test: [ 90/333] eta: 0:01:04 acc1_pos: 37.240 acc5_pos: 100.000 loss: 1.0371 (0.9659) acc1: 26.5625 (37.3741) acc5: 100.0000 (100.0000) time: 0.1765 data: 0.0821 max mem: 3531 +Test: [100/333] eta: 0:01:01 acc1_pos: 34.115 acc5_pos: 100.000 loss: 1.0202 (0.9741) acc1: 26.5625 (36.4145) acc5: 100.0000 (100.0000) time: 0.2026 data: 0.1082 max mem: 3531 +Test: [110/333] eta: 0:00:57 acc1_pos: 32.812 acc5_pos: 100.000 loss: 1.0220 (0.9834) acc1: 25.2604 (35.5222) acc5: 100.0000 (100.0000) time: 0.2226 data: 0.1291 max mem: 3531 +Test: [120/333] eta: 0:00:53 acc1_pos: 23.698 acc5_pos: 100.000 loss: 1.0517 (0.9874) acc1: 25.5208 (34.9884) acc5: 100.0000 (100.0000) time: 0.1929 data: 0.0983 max mem: 3531 +Test: [130/333] eta: 0:00:51 acc1_pos: 46.354 acc5_pos: 100.000 loss: 0.9929 (0.9886) acc1: 28.3854 (34.7229) acc5: 100.0000 (100.0000) time: 0.2199 data: 0.1249 max mem: 3531 +Test: [140/333] eta: 0:00:47 acc1_pos: 56.771 acc5_pos: 100.000 loss: 0.9929 (0.9905) acc1: 29.9479 (34.3602) acc5: 100.0000 (100.0000) time: 0.2219 data: 0.1269 max mem: 3531 +Test: [150/333] eta: 0:00:44 acc1_pos: 33.073 acc5_pos: 100.000 loss: 1.0330 (0.9940) acc1: 29.6875 (33.9197) acc5: 100.0000 (100.0000) time: 0.1728 data: 0.0774 max mem: 3531 +Test: [160/333] eta: 0:00:41 acc1_pos: 18.750 acc5_pos: 100.000 loss: 0.9870 (0.9878) acc1: 30.7292 (34.5529) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.1905 data: 0.0954 max mem: 3531 +Test: [170/333] eta: 0:00:38 acc1_pos: 18.750 acc5_pos: 100.000 loss: 0.6705 (0.9639) acc1: 67.1875 (36.8360) acc5: 100.0000 (100.0000) acc1_neg: 75.521 acc5_neg: 100.000 time: 0.1942 data: 0.1009 max mem: 3531 +Test: [180/333] eta: 0:00:35 acc1_pos: 18.750 acc5_pos: 100.000 loss: 0.5705 (0.9420) acc1: 74.2188 (38.9114) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1693 data: 0.0768 max mem: 3531 +Test: [190/333] eta: 0:00:33 acc1_pos: 18.750 acc5_pos: 100.000 loss: 0.5140 (0.9171) acc1: 75.5208 (41.0981) acc5: 100.0000 (100.0000) acc1_neg: 75.521 acc5_neg: 100.000 time: 0.2093 data: 0.1159 max mem: 3531 +Test: [200/333] eta: 0:00:31 acc1_pos: 18.750 acc5_pos: 100.000 loss: 0.4665 (0.8957) acc1: 79.1667 (43.0491) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.2358 data: 0.1410 max mem: 3531 +Test: [210/333] eta: 0:00:28 acc1_pos: 18.750 acc5_pos: 100.000 loss: 0.5423 (0.8799) acc1: 75.7812 (44.5325) acc5: 100.0000 (100.0000) acc1_neg: 66.146 acc5_neg: 100.000 time: 0.1741 data: 0.0772 max mem: 3532 +Test: [220/333] eta: 0:00:25 acc1_pos: 18.750 acc5_pos: 100.000 loss: 0.5640 (0.8678) acc1: 74.2188 (45.6436) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.1467 data: 0.0519 max mem: 3532 +Test: [230/333] eta: 0:00:23 acc1_pos: 18.750 acc5_pos: 100.000 loss: 0.4236 (0.8473) acc1: 83.3333 (47.4838) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1859 data: 0.0940 max mem: 3532 +Test: [240/333] eta: 0:00:20 acc1_pos: 18.750 acc5_pos: 100.000 loss: 0.4013 (0.8287) acc1: 86.7188 (49.1420) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.2073 data: 0.1162 max mem: 3532 +Test: [250/333] eta: 0:00:18 acc1_pos: 18.750 acc5_pos: 100.000 loss: 0.4076 (0.8137) acc1: 85.1562 (50.4658) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.1982 data: 0.1065 max mem: 3532 +Test: [260/333] eta: 0:00:16 acc1_pos: 18.750 acc5_pos: 100.000 loss: 0.4839 (0.8018) acc1: 80.2083 (51.5396) acc5: 100.0000 (100.0000) acc1_neg: 77.344 acc5_neg: 100.000 time: 0.1809 data: 0.0897 max mem: 3532 +Test: [270/333] eta: 0:00:13 acc1_pos: 18.750 acc5_pos: 100.000 loss: 0.4916 (0.7905) acc1: 77.8646 (52.5350) acc5: 100.0000 (100.0000) acc1_neg: 77.344 acc5_neg: 100.000 time: 0.1818 data: 0.0914 max mem: 3532 +Test: [280/333] eta: 0:00:11 acc1_pos: 18.750 acc5_pos: 100.000 loss: 0.5030 (0.7816) acc1: 77.3438 (53.3307) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.1989 data: 0.1075 max mem: 3532 +Test: [290/333] eta: 0:00:09 acc1_pos: 18.750 acc5_pos: 100.000 loss: 0.5894 (0.7760) acc1: 71.0938 (53.8687) acc5: 100.0000 (100.0000) acc1_neg: 69.531 acc5_neg: 100.000 time: 0.1980 data: 0.1057 max mem: 3532 +Test: [300/333] eta: 0:00:07 acc1_pos: 18.750 acc5_pos: 100.000 loss: 0.6351 (0.7727) acc1: 66.9271 (54.2921) acc5: 100.0000 (100.0000) acc1_neg: 72.135 acc5_neg: 100.000 time: 0.1728 data: 0.0811 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 18.750 acc5_pos: 100.000 loss: 0.6051 (0.7625) acc1: 70.5729 (55.2092) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1549 data: 0.0646 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 18.750 acc5_pos: 100.000 loss: 0.4070 (0.7522) acc1: 86.1979 (56.1591) acc5: 100.0000 (100.0000) acc1_neg: 64.062 acc5_neg: 100.000 time: 0.1274 data: 0.0376 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 18.750 acc5_pos: 100.000 loss: 0.6607 (0.7512) acc1: 66.4062 (56.3704) acc5: 100.0000 (100.0000) acc1_neg: 66.406 acc5_neg: 100.000 time: 0.1114 data: 0.0213 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 18.750 acc5_pos: 100.000 loss: 0.6864 (0.7511) acc1: 64.0625 (56.4098) acc5: 100.0000 (100.0000) acc1_neg: 63.944 acc5_neg: 100.000 time: 0.1224 data: 0.0270 max mem: 3532 +Test: Total time: 0:01:09 (0.2081 s / it) +* Overall Acc@1 56.410 Acc@5 100.000 + target == real Acc@1 33.720 Acc@5 100.000 + target == fake Acc@1 76.608 Acc@5 100.000 + loss 0.751 + PR-AUC 0.5924 ROC-AUC 0.5781 EER 45.07% (thr=0.6665) +Accuracy of the network on the 127843 test images: 56.4% +[rank0]:[W726 06:39:10.406170989 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 06:39:33.187000 3170784 torch/distributed/run.py:766] +W0726 06:39:33.187000 3170784 torch/distributed/run.py:766] ***************************************** +W0726 06:39:33.187000 3170784 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:39:33.187000 3170784 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 2): env:// +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:40:06.983107369 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:40:06.130296376 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:40:07.397069890 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:40:07.701725128 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:40:07.719217381 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 06:40:07.759921124 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:40:07.787521531 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:40:07.872436664 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n11_seed9_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed9_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:18:26 acc1_pos: 93.750 acc5_pos: 100.000 loss: 0.2057 (0.2057) acc1: 93.7500 (93.7500) acc5: 100.0000 (100.0000) time: 4.7065 data: 2.8796 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/235] eta: 0:01:54 acc1_pos: 92.969 acc5_pos: 100.000 loss: 0.2042 (0.2066) acc1: 94.2708 (94.5076) acc5: 100.0000 (100.0000) time: 0.5096 data: 0.2621 max mem: 3531 +Test: [ 20/235] eta: 0:01:06 acc1_pos: 94.792 acc5_pos: 100.000 loss: 0.2025 (0.2021) acc1: 94.7917 (94.8289) acc5: 100.0000 (100.0000) time: 0.0900 data: 0.0004 max mem: 3531 +Test: [ 30/235] eta: 0:00:49 acc1_pos: 95.573 acc5_pos: 100.000 loss: 0.1965 (0.2010) acc1: 95.3125 (94.9765) acc5: 100.0000 (100.0000) time: 0.0923 data: 0.0005 max mem: 3531 +Test: [ 40/235] eta: 0:00:42 acc1_pos: 93.490 acc5_pos: 100.000 loss: 0.1961 (0.2005) acc1: 95.3125 (94.9632) acc5: 100.0000 (100.0000) time: 0.1172 data: 0.0230 max mem: 3531 +Test: [ 50/235] eta: 0:00:38 acc1_pos: 92.969 acc5_pos: 100.000 loss: 0.2041 (0.2022) acc1: 94.5312 (94.8223) acc5: 100.0000 (100.0000) time: 0.1599 data: 0.0631 max mem: 3531 +Test: [ 60/235] eta: 0:00:34 acc1_pos: 94.531 acc5_pos: 100.000 loss: 0.2106 (0.2045) acc1: 94.5312 (94.7106) acc5: 100.0000 (100.0000) time: 0.1649 data: 0.0704 max mem: 3531 +Test: [ 70/235] eta: 0:00:31 acc1_pos: 92.308 acc5_pos: 100.000 loss: 0.2332 (0.3939) acc1: 92.7083 (84.2210) acc5: 100.0000 (100.0000) acc1_neg: 10.156 acc5_neg: 100.000 time: 0.1504 data: 0.0596 max mem: 3531 +Test: [ 80/235] eta: 0:00:29 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.7195 (0.5589) acc1: 10.4167 (75.1415) acc5: 100.0000 (100.0000) acc1_neg: 8.594 acc5_neg: 100.000 time: 0.1549 data: 0.0610 max mem: 3531 +Test: [ 90/235] eta: 0:00:27 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.7248 (0.6876) acc1: 11.1979 (68.1691) acc5: 100.0000 (100.0000) acc1_neg: 8.594 acc5_neg: 100.000 time: 0.1747 data: 0.0807 max mem: 3531 +Test: [100/235] eta: 0:00:25 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.7264 (0.7898) acc1: 11.4583 (62.5645) acc5: 100.0000 (100.0000) acc1_neg: 9.896 acc5_neg: 100.000 time: 0.1954 data: 0.1046 max mem: 3531 +Test: [110/235] eta: 0:00:23 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.6991 (0.8722) acc1: 11.7188 (58.0330) acc5: 100.0000 (100.0000) acc1_neg: 11.979 acc5_neg: 100.000 time: 0.2010 data: 0.1102 max mem: 3531 +Test: [120/235] eta: 0:00:21 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.6991 (0.9414) acc1: 11.9792 (54.2226) acc5: 100.0000 (100.0000) acc1_neg: 12.500 acc5_neg: 100.000 time: 0.1893 data: 0.0977 max mem: 3531 +Test: [130/235] eta: 0:00:19 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.7040 (0.9999) acc1: 12.2396 (51.0377) acc5: 100.0000 (100.0000) acc1_neg: 12.240 acc5_neg: 100.000 time: 0.1835 data: 0.0904 max mem: 3531 +Test: [140/235] eta: 0:00:18 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.7335 (1.0523) acc1: 11.1979 (48.1882) acc5: 100.0000 (100.0000) acc1_neg: 10.938 acc5_neg: 100.000 time: 0.2199 data: 0.1264 max mem: 3531 +Test: [150/235] eta: 0:00:16 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.7268 (1.0953) acc1: 10.9375 (45.7850) acc5: 100.0000 (100.0000) acc1_neg: 14.062 acc5_neg: 100.000 time: 0.2349 data: 0.1391 max mem: 3531 +Test: [160/235] eta: 0:00:14 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.6968 (1.1332) acc1: 11.7188 (43.7112) acc5: 100.0000 (100.0000) acc1_neg: 15.104 acc5_neg: 100.000 time: 0.1599 data: 0.0605 max mem: 3531 +Test: [170/235] eta: 0:00:12 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.7142 (1.1680) acc1: 10.1562 (41.6910) acc5: 100.0000 (100.0000) acc1_neg: 9.635 acc5_neg: 100.000 time: 0.1154 data: 0.0191 max mem: 3531 +Test: [180/235] eta: 0:00:10 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.7283 (1.1995) acc1: 8.8542 (39.9171) acc5: 100.0000 (100.0000) acc1_neg: 8.073 acc5_neg: 100.000 time: 0.1549 data: 0.0611 max mem: 3531 +Test: [190/235] eta: 0:00:08 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.7177 (1.2265) acc1: 9.1146 (38.3222) acc5: 100.0000 (100.0000) acc1_neg: 8.854 acc5_neg: 100.000 time: 0.1894 data: 0.0947 max mem: 3531 +Test: [200/235] eta: 0:00:06 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.7060 (1.2513) acc1: 9.1146 (36.8639) acc5: 100.0000 (100.0000) acc1_neg: 9.115 acc5_neg: 100.000 time: 0.1899 data: 0.0939 max mem: 3531 +Test: [210/235] eta: 0:00:04 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.7276 (1.2735) acc1: 9.1146 (35.5672) acc5: 100.0000 (100.0000) acc1_neg: 9.635 acc5_neg: 100.000 time: 0.1849 data: 0.0877 max mem: 3532 +Test: [220/235] eta: 0:00:02 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.7203 (1.2937) acc1: 9.3750 (34.3832) acc5: 100.0000 (100.0000) acc1_neg: 8.594 acc5_neg: 100.000 time: 0.1699 data: 0.0734 max mem: 3532 +Test: [230/235] eta: 0:00:00 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.7154 (1.3122) acc1: 8.8542 (33.3097) acc5: 100.0000 (100.0000) acc1_neg: 7.812 acc5_neg: 100.000 time: 0.1449 data: 0.0486 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.7272 (1.3202) acc1: 8.5938 (32.9966) acc5: 100.0000 (100.0000) acc1_neg: 0.000 acc5_neg: 100.000 time: 0.1210 data: 0.0276 max mem: 3532 +Test: Total time: 0:00:42 (0.1823 s / it) +* Overall Acc@1 32.997 Acc@5 100.000 + target == real Acc@1 94.718 Acc@5 100.000 + target == fake Acc@1 10.673 Acc@5 100.000 + loss 1.320 + PR-AUC 0.8183 ROC-AUC 0.6596 EER 38.29% (thr=0.1284) +Accuracy of the network on the 89879 test images: 33.0% +[rank0]:[W726 06:41:05.108687630 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 06:41:27.095000 3171342 torch/distributed/run.py:766] +W0726 06:41:27.095000 3171342 torch/distributed/run.py:766] ***************************************** +W0726 06:41:27.095000 3171342 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:41:27.095000 3171342 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:41:58.961966556 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:41:59.367466132 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 06:41:59.491990240 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:41:59.927327360 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +[rank6]:[W726 06:41:59.045976586 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:41:59.054604309 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:41:59.087774045 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W726 06:41:59.087826559 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n11_seed9_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed9_resampled_n116-040-168_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:32:19 acc1_pos: 17.188 acc5_pos: 100.000 loss: 1.1526 (1.1526) acc1: 17.1875 (17.1875) acc5: 100.0000 (100.0000) time: 7.2118 data: 4.9008 max mem: 3531 +Test: [ 10/269] eta: 0:03:13 acc1_pos: 33.854 acc5_pos: 100.000 loss: 0.8764 (0.9097) acc1: 30.7292 (31.9129) acc5: 100.0000 (100.0000) time: 0.7458 data: 0.4459 max mem: 3531 +Test: [ 20/269] eta: 0:01:49 acc1_pos: 47.917 acc5_pos: 100.000 loss: 0.8707 (0.8880) acc1: 33.0729 (33.5441) acc5: 100.0000 (100.0000) time: 0.0995 data: 0.0004 max mem: 3531 +Test: [ 30/269] eta: 0:01:32 acc1_pos: 18.490 acc5_pos: 100.000 loss: 0.8921 (0.8987) acc1: 28.3854 (32.7285) acc5: 100.0000 (100.0000) time: 0.1899 data: 0.0922 max mem: 3531 +Test: [ 40/269] eta: 0:01:20 acc1_pos: 6.034 acc5_pos: 100.000 loss: 0.6397 (0.7923) acc1: 59.1146 (44.8552) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2649 data: 0.1710 max mem: 3531 +Test: [ 50/269] eta: 0:01:08 acc1_pos: 6.034 acc5_pos: 100.000 loss: 0.3832 (0.7190) acc1: 91.1458 (53.6101) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.1999 data: 0.1032 max mem: 3531 +Test: [ 60/269] eta: 0:01:00 acc1_pos: 6.034 acc5_pos: 100.000 loss: 0.4440 (0.6763) acc1: 87.7604 (58.9694) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1649 data: 0.0661 max mem: 3531 +Test: [ 70/269] eta: 0:00:55 acc1_pos: 6.034 acc5_pos: 100.000 loss: 0.4440 (0.6437) acc1: 87.7604 (63.2702) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1899 data: 0.0948 max mem: 3531 +Test: [ 80/269] eta: 0:00:50 acc1_pos: 6.034 acc5_pos: 100.000 loss: 0.4329 (0.6180) acc1: 91.4062 (66.6827) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1999 data: 0.1059 max mem: 3531 +Test: [ 90/269] eta: 0:00:46 acc1_pos: 6.034 acc5_pos: 100.000 loss: 0.4811 (0.6168) acc1: 83.5938 (66.9729) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.1849 data: 0.0911 max mem: 3531 +Test: [100/269] eta: 0:00:42 acc1_pos: 6.034 acc5_pos: 100.000 loss: 0.5597 (0.6086) acc1: 73.6979 (67.9997) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.1773 data: 0.0834 max mem: 3531 +Test: [110/269] eta: 0:00:38 acc1_pos: 6.034 acc5_pos: 100.000 loss: 0.5134 (0.5989) acc1: 81.2500 (69.3342) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1827 data: 0.0890 max mem: 3531 +Test: [120/269] eta: 0:00:35 acc1_pos: 6.034 acc5_pos: 100.000 loss: 0.5134 (0.5914) acc1: 81.2500 (70.1575) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.1924 data: 0.0996 max mem: 3531 +Test: [130/269] eta: 0:00:32 acc1_pos: 6.034 acc5_pos: 100.000 loss: 0.5488 (0.5912) acc1: 74.4792 (70.1654) acc5: 100.0000 (100.0000) acc1_neg: 58.594 acc5_neg: 100.000 time: 0.1921 data: 0.0964 max mem: 3531 +Test: [140/269] eta: 0:00:30 acc1_pos: 6.034 acc5_pos: 100.000 loss: 0.6168 (0.5931) acc1: 67.4479 (69.9837) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.1801 data: 0.0842 max mem: 3531 +Test: [150/269] eta: 0:00:27 acc1_pos: 6.034 acc5_pos: 100.000 loss: 0.6226 (0.5899) acc1: 64.0625 (70.2573) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1724 data: 0.0792 max mem: 3531 +Test: [160/269] eta: 0:00:24 acc1_pos: 6.034 acc5_pos: 100.000 loss: 0.4664 (0.5808) acc1: 83.0729 (71.1649) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.1625 data: 0.0705 max mem: 3531 +Test: [170/269] eta: 0:00:22 acc1_pos: 6.034 acc5_pos: 100.000 loss: 0.4743 (0.5773) acc1: 83.0729 (71.7181) acc5: 100.0000 (100.0000) acc1_neg: 76.302 acc5_neg: 100.000 time: 0.1793 data: 0.0890 max mem: 3531 +Test: [180/269] eta: 0:00:19 acc1_pos: 6.034 acc5_pos: 100.000 loss: 0.4725 (0.5691) acc1: 86.9792 (72.6577) acc5: 100.0000 (100.0000) acc1_neg: 76.042 acc5_neg: 100.000 time: 0.2143 data: 0.1240 max mem: 3531 +Test: [190/269] eta: 0:00:17 acc1_pos: 6.034 acc5_pos: 100.000 loss: 0.4397 (0.5663) acc1: 87.5000 (72.9576) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.2182 data: 0.1265 max mem: 3531 +Test: [200/269] eta: 0:00:14 acc1_pos: 6.034 acc5_pos: 100.000 loss: 0.5287 (0.5662) acc1: 75.7812 (72.9750) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.1677 data: 0.0716 max mem: 3531 +Test: [210/269] eta: 0:00:12 acc1_pos: 6.034 acc5_pos: 100.000 loss: 0.4971 (0.5613) acc1: 78.1250 (73.6152) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1897 data: 0.0930 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 6.034 acc5_pos: 100.000 loss: 0.4304 (0.5552) acc1: 89.8438 (74.2565) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2179 data: 0.1215 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 6.034 acc5_pos: 100.000 loss: 0.3660 (0.5468) acc1: 94.0104 (75.0981) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1503 data: 0.0556 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 6.034 acc5_pos: 100.000 loss: 0.4114 (0.5454) acc1: 88.8021 (75.2842) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.1418 data: 0.0504 max mem: 3532 +Test: [250/269] eta: 0:00:03 acc1_pos: 6.034 acc5_pos: 100.000 loss: 0.5460 (0.5491) acc1: 72.9167 (74.8298) acc5: 100.0000 (100.0000) acc1_neg: 46.354 acc5_neg: 100.000 time: 0.1723 data: 0.0803 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 6.034 acc5_pos: 100.000 loss: 0.6554 (0.5509) acc1: 60.4167 (74.6338) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1573 data: 0.0666 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 6.034 acc5_pos: 100.000 loss: 0.4672 (0.5468) acc1: 82.8125 (75.1435) acc5: 100.0000 (100.0000) acc1_neg: 98.950 acc5_neg: 100.000 time: 0.1398 data: 0.0466 max mem: 3532 +Test: Total time: 0:00:55 (0.2045 s / it) +* Overall Acc@1 75.144 Acc@5 100.000 + target == real Acc@1 31.990 Acc@5 100.000 + target == fake Acc@1 81.033 Acc@5 100.000 + loss 0.547 + PR-AUC 0.9139 ROC-AUC 0.6095 EER 42.58% (thr=0.6221) +Accuracy of the network on the 103293 test images: 75.1% +[rank0]:[W726 06:43:04.293758225 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T06:43:10Z] eval_done tag=fixed500k_fake_n11_seed9_resampled +eval_queue_done_utc=2026-07-26T06:43:10Z diff --git a/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_1958_seed5_9/eval_queue_n116-040-168.pid b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_1958_seed5_9/eval_queue_n116-040-168.pid new file mode 100644 index 0000000000000000000000000000000000000000..731112648e1baeb2de0745d3111388c66e002553 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_1958_seed5_9/eval_queue_n116-040-168.pid @@ -0,0 +1 @@ +3131945 diff --git a/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_1958_seed5_9/eval_queue_n116-040-213.log b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_1958_seed5_9/eval_queue_n116-040-213.log new file mode 100644 index 0000000000000000000000000000000000000000..837336bcaed05a13c063745632a195816d9deb25 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_1958_seed5_9/eval_queue_n116-040-213.log @@ -0,0 +1,10270 @@ +eval_queue_started_utc=2026-07-26T02:59:39Z +hostname=n116-040-213 +train_tag=20260725_1958_seed5_9 +eval_tag=20260725_1958_seed5_9 +real_level=32 +fake_level=64 +eval_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9 +batch_per_gpu=256 +seeds=5 6 7 8 9 +[2026-07-26T02:59:39Z] eval_start tag=fixed500k_real_n32_seed5_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n32_classseed5_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T02:59:39Z +hostname=n116-040-213 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n32_classseed5_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n32_seed5_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 02:59:52.665000 1171834 torch/distributed/run.py:766] +W0726 02:59:52.665000 1171834 torch/distributed/run.py:766] ***************************************** +W0726 02:59:52.665000 1171834 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 02:59:52.665000 1171834 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:00:23.788250653 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:00:23.022759422 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W726 03:00:23.022761979 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:00:23.029780884 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:00:24.229019975 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 1): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:00:24.378564660 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:00:24.384260080 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:00:24.420545060 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n32_seed5_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n32_classseed5_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:18:55 acc1_pos: 65.104 acc5_pos: 100.000 loss: 0.5595 (0.5595) acc1: 65.1042 (65.1042) acc5: 100.0000 (100.0000) time: 4.8521 data: 3.1528 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/234] eta: 0:01:57 acc1_pos: 37.760 acc5_pos: 100.000 loss: 0.7688 (0.7608) acc1: 48.4375 (48.5085) acc5: 100.0000 (100.0000) time: 0.5232 data: 0.2870 max mem: 3531 +Test: [ 20/234] eta: 0:01:09 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.6712 (0.6343) acc1: 61.9792 (60.0942) acc5: 100.0000 (100.0000) time: 0.0987 data: 0.0080 max mem: 3531 +Test: [ 30/234] eta: 0:01:03 acc1_pos: 70.573 acc5_pos: 100.000 loss: 0.2103 (0.5112) acc1: 95.3125 (70.2957) acc5: 100.0000 (100.0000) time: 0.1933 data: 0.1020 max mem: 3531 +Test: [ 40/234] eta: 0:01:00 acc1_pos: 97.396 acc5_pos: 100.000 loss: 0.2096 (0.5016) acc1: 95.5729 (70.7190) acc5: 100.0000 (100.0000) time: 0.2938 data: 0.2022 max mem: 3531 +Test: [ 50/234] eta: 0:00:49 acc1_pos: 97.135 acc5_pos: 100.000 loss: 0.1659 (0.4343) acc1: 96.8750 (75.8681) acc5: 100.0000 (100.0000) time: 0.2108 data: 0.1195 max mem: 3531 +Test: [ 60/234] eta: 0:00:44 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.1104 (0.3732) acc1: 97.9167 (79.7814) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1374 data: 0.0444 max mem: 3531 +Test: [ 70/234] eta: 0:00:39 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.0589 (0.3858) acc1: 100.0000 (79.2510) acc5: 100.0000 (100.0000) acc1_neg: 60.938 acc5_neg: 100.000 time: 0.1656 data: 0.0711 max mem: 3531 +Test: [ 80/234] eta: 0:00:36 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.6232 (0.3976) acc1: 65.6250 (78.9609) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1802 data: 0.0871 max mem: 3531 +Test: [ 90/234] eta: 0:00:32 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.2139 (0.3716) acc1: 92.7083 (80.7979) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1762 data: 0.0828 max mem: 3531 +Test: [100/234] eta: 0:00:29 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.2246 (0.3719) acc1: 92.7083 (81.0051) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1652 data: 0.0725 max mem: 3531 +Test: [110/234] eta: 0:00:26 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.3923 (0.3717) acc1: 80.2083 (81.3227) acc5: 100.0000 (100.0000) acc1_neg: 73.958 acc5_neg: 100.000 time: 0.1635 data: 0.0724 max mem: 3531 +Test: [120/234] eta: 0:00:24 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.3887 (0.3795) acc1: 85.1562 (81.2694) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1685 data: 0.0770 max mem: 3531 +Test: [130/234] eta: 0:00:21 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.4285 (0.3852) acc1: 83.0729 (81.3057) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.1683 data: 0.0745 max mem: 3531 +Test: [140/234] eta: 0:00:19 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.5016 (0.4104) acc1: 77.6042 (80.1049) acc5: 100.0000 (100.0000) acc1_neg: 37.500 acc5_neg: 100.000 time: 0.1711 data: 0.0759 max mem: 3531 +Test: [150/234] eta: 0:00:17 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.4178 (0.4166) acc1: 77.6042 (79.8065) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.1788 data: 0.0852 max mem: 3531 +Test: [160/234] eta: 0:00:14 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.3971 (0.4170) acc1: 81.5104 (80.0320) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.1728 data: 0.0805 max mem: 3531 +Test: [170/234] eta: 0:00:12 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.4825 (0.4208) acc1: 78.9062 (79.9997) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1660 data: 0.0698 max mem: 3531 +Test: [180/234] eta: 0:00:10 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.4997 (0.4288) acc1: 78.3854 (79.5292) acc5: 100.0000 (100.0000) acc1_neg: 45.833 acc5_neg: 100.000 time: 0.1499 data: 0.0514 max mem: 3531 +Test: [190/234] eta: 0:00:08 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.6414 (0.4406) acc1: 64.5833 (78.7726) acc5: 100.0000 (100.0000) acc1_neg: 69.271 acc5_neg: 100.000 time: 0.1521 data: 0.0555 max mem: 3531 +Test: [200/234] eta: 0:00:06 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.6292 (0.4340) acc1: 65.6250 (79.1978) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1723 data: 0.0778 max mem: 3531 +Test: [210/234] eta: 0:00:04 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.1177 (0.4193) acc1: 100.0000 (80.1590) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1777 data: 0.0838 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.1338 (0.4077) acc1: 98.9583 (80.9318) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.1782 data: 0.0821 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.3629 (0.4274) acc1: 48.4375 (79.4214) acc5: 100.0000 (100.0000) acc1_neg: 45.833 acc5_neg: 100.000 time: 0.1634 data: 0.0688 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.8385 (0.4326) acc1: 48.1771 (79.0636) acc5: 100.0000 (100.0000) acc1_neg: 48.249 acc5_neg: 100.000 time: 0.1582 data: 0.0551 max mem: 3532 +Test: Total time: 0:00:44 (0.1916 s / it) +* Overall Acc@1 79.064 Acc@5 100.000 + target == real Acc@1 76.000 Acc@5 100.000 + target == fake Acc@1 79.927 Acc@5 100.000 + loss 0.433 + PR-AUC 0.9650 ROC-AUC 0.8822 EER 21.53% (thr=0.5215) +Accuracy of the network on the 89729 test images: 79.1% +[rank0]:[W726 03:01:21.896569788 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:01:40.615000 1172383 torch/distributed/run.py:766] +W0726 03:01:40.615000 1172383 torch/distributed/run.py:766] ***************************************** +W0726 03:01:40.615000 1172383 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:01:40.615000 1172383 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:02:09.734159914 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:02:09.903475277 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:02:09.963626980 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:02:10.226277630 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:02:10.261502668 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:02:10.318886659 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:02:10.357436560 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:02:10.387491623 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n32_seed5_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n32_classseed5_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:27:02 acc1_pos: 48.698 acc5_pos: 100.000 loss: 0.7613 (0.7613) acc1: 48.6979 (48.6979) acc5: 100.0000 (100.0000) time: 6.7126 data: 4.8457 max mem: 3531 +Test: [ 10/2387] eta: 0:27:25 acc1_pos: 22.396 acc5_pos: 100.000 loss: 0.9578 (0.9759) acc1: 26.0417 (26.9886) acc5: 100.0000 (100.0000) time: 0.6924 data: 0.4410 max mem: 3531 +Test: [ 20/2387] eta: 0:16:00 acc1_pos: 17.448 acc5_pos: 100.000 loss: 0.9464 (0.9979) acc1: 22.3958 (24.9752) acc5: 100.0000 (100.0000) time: 0.0904 data: 0.0005 max mem: 3531 +Test: [ 30/2387] eta: 0:13:33 acc1_pos: 6.510 acc5_pos: 100.000 loss: 0.9967 (1.0092) acc1: 19.5312 (23.8155) acc5: 100.0000 (100.0000) time: 0.1540 data: 0.0593 max mem: 3531 +Test: [ 40/2387] eta: 0:12:31 acc1_pos: 12.500 acc5_pos: 100.000 loss: 1.0393 (1.0300) acc1: 14.5833 (21.1890) acc5: 100.0000 (100.0000) time: 0.2304 data: 0.1347 max mem: 3531 +Test: [ 50/2387] eta: 0:11:40 acc1_pos: 37.500 acc5_pos: 100.000 loss: 1.0105 (1.0065) acc1: 17.9688 (24.0247) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1365 max mem: 3531 +Test: [ 60/2387] eta: 0:11:37 acc1_pos: 44.792 acc5_pos: 100.000 loss: 0.8759 (0.9893) acc1: 36.1979 (25.7215) acc5: 100.0000 (100.0000) time: 0.2583 data: 0.1635 max mem: 3531 +Test: [ 70/2387] eta: 0:11:05 acc1_pos: 29.688 acc5_pos: 100.000 loss: 0.9752 (0.9924) acc1: 29.1667 (25.6272) acc5: 100.0000 (100.0000) time: 0.2549 data: 0.1545 max mem: 3531 +Test: [ 80/2387] eta: 0:10:49 acc1_pos: 73.438 acc5_pos: 100.000 loss: 0.9492 (0.9658) acc1: 29.6875 (28.3179) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1232 max mem: 3531 +Test: [ 90/2387] eta: 0:10:35 acc1_pos: 59.635 acc5_pos: 100.000 loss: 0.6686 (0.9286) acc1: 53.3854 (32.0541) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1454 max mem: 3531 +Test: [ 100/2387] eta: 0:10:40 acc1_pos: 58.073 acc5_pos: 100.000 loss: 0.6505 (0.9106) acc1: 58.0729 (34.1223) acc5: 100.0000 (100.0000) time: 0.2749 data: 0.1807 max mem: 3531 +Test: [ 110/2387] eta: 0:10:27 acc1_pos: 49.479 acc5_pos: 100.000 loss: 0.6861 (0.8912) acc1: 52.8646 (36.1627) acc5: 100.0000 (100.0000) time: 0.2699 data: 0.1756 max mem: 3531 +Test: [ 120/2387] eta: 0:10:21 acc1_pos: 41.146 acc5_pos: 100.000 loss: 0.7045 (0.8768) acc1: 55.7292 (37.7346) acc5: 100.0000 (100.0000) time: 0.2448 data: 0.1496 max mem: 3531 +Test: [ 130/2387] eta: 0:10:14 acc1_pos: 52.344 acc5_pos: 100.000 loss: 0.7027 (0.8642) acc1: 54.6875 (38.9313) acc5: 100.0000 (100.0000) time: 0.2549 data: 0.1581 max mem: 3531 +Test: [ 140/2387] eta: 0:10:06 acc1_pos: 58.333 acc5_pos: 100.000 loss: 0.6988 (0.8535) acc1: 53.9062 (39.9029) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1489 max mem: 3531 +Test: [ 150/2387] eta: 0:10:04 acc1_pos: 58.594 acc5_pos: 100.000 loss: 0.6603 (0.8394) acc1: 58.0729 (41.3614) acc5: 100.0000 (100.0000) time: 0.2549 data: 0.1592 max mem: 3531 +Test: [ 160/2387] eta: 0:09:53 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.6246 (0.8172) acc1: 64.0625 (43.5478) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2399 data: 0.1440 max mem: 3531 +Test: [ 170/2387] eta: 0:09:39 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2153 (0.7816) acc1: 97.1354 (46.7547) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1950 data: 0.0968 max mem: 3531 +Test: [ 180/2387] eta: 0:09:31 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2060 (0.7500) acc1: 98.4375 (49.5900) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1999 data: 0.1000 max mem: 3531 +Test: [ 190/2387] eta: 0:09:21 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2040 (0.7215) acc1: 98.1771 (52.1365) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2049 data: 0.1052 max mem: 3531 +Test: [ 200/2387] eta: 0:09:10 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2028 (0.6958) acc1: 98.4375 (54.4258) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1849 data: 0.0886 max mem: 3531 +Test: [ 210/2387] eta: 0:09:06 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2093 (0.6729) acc1: 98.1771 (56.4882) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2101 data: 0.1162 max mem: 3532 +Test: [ 220/2387] eta: 0:08:58 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2103 (0.6518) acc1: 98.1771 (58.3675) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2149 data: 0.1216 max mem: 3532 +Test: [ 230/2387] eta: 0:08:52 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2083 (0.6326) acc1: 98.1771 (60.1055) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2004 data: 0.1058 max mem: 3532 +Test: [ 240/2387] eta: 0:08:45 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2109 (0.6179) acc1: 97.6562 (61.5696) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2049 data: 0.1099 max mem: 3532 +Test: [ 250/2387] eta: 0:08:36 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2004 (0.5978) acc1: 96.3542 (63.0789) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1842 data: 0.0895 max mem: 3532 +Test: [ 260/2387] eta: 0:08:28 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0932 (0.5783) acc1: 100.0000 (64.4935) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1713 data: 0.0766 max mem: 3532 +Test: [ 270/2387] eta: 0:08:22 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0830 (0.5600) acc1: 100.0000 (65.8018) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1809 data: 0.0885 max mem: 3532 +Test: [ 280/2387] eta: 0:08:16 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0824 (0.5433) acc1: 100.0000 (67.0188) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1895 data: 0.0978 max mem: 3532 +Test: [ 290/2387] eta: 0:08:12 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0845 (0.5275) acc1: 100.0000 (68.1522) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2007 data: 0.1077 max mem: 3532 +Test: [ 300/2387] eta: 0:08:06 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0846 (0.5131) acc1: 100.0000 (69.2103) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1961 data: 0.1021 max mem: 3532 +Test: [ 310/2387] eta: 0:08:01 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1006 (0.5038) acc1: 100.0000 (70.1459) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1908 data: 0.0978 max mem: 3532 +Test: [ 320/2387] eta: 0:07:57 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.3612 (0.4994) acc1: 96.6146 (70.9566) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1999 data: 0.1055 max mem: 3532 +Test: [ 330/2387] eta: 0:07:52 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.3644 (0.4952) acc1: 96.3542 (71.7310) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1932 data: 0.0993 max mem: 3532 +Test: [ 340/2387] eta: 0:07:45 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.3568 (0.4912) acc1: 96.6146 (72.4516) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1692 data: 0.0769 max mem: 3532 +Test: [ 350/2387] eta: 0:07:42 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.3505 (0.4833) acc1: 97.3958 (73.2216) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1826 data: 0.0893 max mem: 3532 +Test: [ 360/2387] eta: 0:07:35 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1836 (0.4748) acc1: 99.7396 (73.9468) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1857 data: 0.0903 max mem: 3532 +Test: [ 370/2387] eta: 0:07:31 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1778 (0.4668) acc1: 99.2188 (74.6196) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1676 data: 0.0693 max mem: 3532 +Test: [ 380/2387] eta: 0:07:26 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1774 (0.4591) acc1: 99.2188 (75.2645) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1776 data: 0.0808 max mem: 3532 +Test: [ 390/2387] eta: 0:07:23 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1739 (0.4519) acc1: 99.2188 (75.8725) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1889 data: 0.0954 max mem: 3532 +Test: [ 400/2387] eta: 0:07:19 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1766 (0.4450) acc1: 98.9583 (76.4534) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1971 data: 0.1033 max mem: 3532 +Test: [ 410/2387] eta: 0:07:16 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1762 (0.4384) acc1: 99.2188 (77.0073) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1997 data: 0.1053 max mem: 3532 +Test: [ 420/2387] eta: 0:07:11 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1725 (0.4321) acc1: 99.2188 (77.5386) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1858 data: 0.0939 max mem: 3532 +Test: [ 430/2387] eta: 0:07:08 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1769 (0.4271) acc1: 98.9583 (78.0090) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1796 data: 0.0871 max mem: 3532 +Test: [ 440/2387] eta: 0:07:04 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1795 (0.4207) acc1: 98.1771 (78.4864) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1917 data: 0.0987 max mem: 3533 +Test: [ 450/2387] eta: 0:07:01 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0719 (0.4129) acc1: 100.0000 (78.9634) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1875 data: 0.0951 max mem: 3533 +Test: [ 460/2387] eta: 0:06:57 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0715 (0.4055) acc1: 100.0000 (79.4197) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1817 data: 0.0896 max mem: 3533 +Test: [ 470/2387] eta: 0:06:53 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0718 (0.3984) acc1: 100.0000 (79.8567) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1781 data: 0.0864 max mem: 3533 +Test: [ 480/2387] eta: 0:06:50 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0721 (0.3917) acc1: 100.0000 (80.2755) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1884 data: 0.0964 max mem: 3533 +Test: [ 490/2387] eta: 0:06:47 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0725 (0.3852) acc1: 100.0000 (80.6772) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1940 data: 0.1023 max mem: 3533 +Test: [ 500/2387] eta: 0:06:44 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0725 (0.3789) acc1: 100.0000 (81.0629) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1902 data: 0.0979 max mem: 3533 +Test: [ 510/2387] eta: 0:06:41 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0726 (0.3733) acc1: 100.0000 (81.4335) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1859 data: 0.0929 max mem: 3533 +Test: [ 520/2387] eta: 0:06:37 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1064 (0.3685) acc1: 100.0000 (81.7898) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1755 data: 0.0826 max mem: 3533 +Test: [ 530/2387] eta: 0:06:34 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1062 (0.3635) acc1: 100.0000 (82.1328) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1727 data: 0.0777 max mem: 3533 +Test: [ 540/2387] eta: 0:06:31 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1053 (0.3587) acc1: 100.0000 (82.4630) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1813 data: 0.0875 max mem: 3533 +Test: [ 550/2387] eta: 0:06:27 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1051 (0.3541) acc1: 100.0000 (82.7813) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1829 data: 0.0915 max mem: 3533 +Test: [ 560/2387] eta: 0:06:24 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1048 (0.3497) acc1: 100.0000 (83.0882) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1796 data: 0.0880 max mem: 3533 +Test: [ 570/2387] eta: 0:06:22 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1039 (0.3454) acc1: 100.0000 (83.3844) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1892 data: 0.0973 max mem: 3533 +Test: [ 580/2387] eta: 0:06:19 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1045 (0.3413) acc1: 100.0000 (83.6704) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1881 data: 0.0956 max mem: 3533 +Test: [ 590/2387] eta: 0:06:16 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1054 (0.3373) acc1: 100.0000 (83.9467) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1857 data: 0.0920 max mem: 3533 +Test: [ 600/2387] eta: 0:06:13 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1088 (0.3339) acc1: 100.0000 (84.2138) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1903 data: 0.0974 max mem: 3533 +Test: [ 610/2387] eta: 0:06:12 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1335 (0.3306) acc1: 100.0000 (84.4688) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2092 data: 0.1171 max mem: 3533 +Test: [ 620/2387] eta: 0:06:08 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1352 (0.3275) acc1: 100.0000 (84.7172) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1943 data: 0.1021 max mem: 3533 +Test: [ 630/2387] eta: 0:06:07 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1358 (0.3245) acc1: 100.0000 (84.9569) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1992 data: 0.1058 max mem: 3533 +Test: [ 640/2387] eta: 0:06:05 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1368 (0.3216) acc1: 99.7396 (85.1879) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2205 data: 0.1255 max mem: 3533 +Test: [ 650/2387] eta: 0:06:02 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1373 (0.3188) acc1: 99.7396 (85.4135) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1952 data: 0.0996 max mem: 3533 +Test: [ 660/2387] eta: 0:05:58 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1373 (0.3160) acc1: 100.0000 (85.6326) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1513 data: 0.0547 max mem: 3533 +Test: [ 670/2387] eta: 0:05:53 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1366 (0.3134) acc1: 100.0000 (85.8451) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1117 data: 0.0174 max mem: 3534 +Test: [ 680/2387] eta: 0:05:51 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1397 (0.3112) acc1: 100.0000 (86.0526) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1524 data: 0.0590 max mem: 3534 +Test: [ 690/2387] eta: 0:05:48 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1817 (0.3114) acc1: 100.0000 (86.0800) acc5: 100.0000 (100.0000) acc1_neg: 80.729 acc5_neg: 100.000 time: 0.1894 data: 0.0955 max mem: 3534 +Test: [ 700/2387] eta: 0:05:45 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.3666 (0.3122) acc1: 84.1146 (86.0497) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1775 data: 0.0839 max mem: 3534 +Test: [ 710/2387] eta: 0:05:43 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.3717 (0.3130) acc1: 83.3333 (86.0159) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.1877 data: 0.0937 max mem: 3534 +Test: [ 720/2387] eta: 0:05:41 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.3638 (0.3138) acc1: 83.8542 (85.9978) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.2073 data: 0.1113 max mem: 3534 +Test: [ 730/2387] eta: 0:05:38 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.3704 (0.3147) acc1: 83.5938 (85.9589) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.1753 data: 0.0787 max mem: 3534 +Test: [ 740/2387] eta: 0:05:33 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.3703 (0.3154) acc1: 83.5938 (85.9305) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.1199 data: 0.0231 max mem: 3534 +Test: [ 750/2387] eta: 0:05:30 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.3588 (0.3160) acc1: 84.3750 (85.9101) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1177 data: 0.0210 max mem: 3534 +Test: [ 760/2387] eta: 0:05:27 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.3478 (0.3163) acc1: 84.6354 (85.9033) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1552 data: 0.0621 max mem: 3534 +Test: [ 770/2387] eta: 0:05:25 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.3133 (0.3155) acc1: 89.3229 (86.0000) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1723 data: 0.0805 max mem: 3534 +Test: [ 780/2387] eta: 0:05:23 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1881 (0.3136) acc1: 97.3958 (86.1629) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1855 data: 0.0936 max mem: 3534 +Test: [ 790/2387] eta: 0:05:21 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1522 (0.3118) acc1: 99.2188 (86.3020) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.2014 data: 0.1102 max mem: 3534 +Test: [ 800/2387] eta: 0:05:19 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1431 (0.3098) acc1: 99.2188 (86.4622) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2055 data: 0.1148 max mem: 3534 +Test: [ 810/2387] eta: 0:05:17 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1556 (0.3080) acc1: 99.7396 (86.6137) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1989 data: 0.1074 max mem: 3534 +Test: [ 820/2387] eta: 0:05:15 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1654 (0.3066) acc1: 99.7396 (86.7536) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.2004 data: 0.1082 max mem: 3534 +Test: [ 830/2387] eta: 0:05:13 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2252 (0.3068) acc1: 93.4896 (86.7871) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.2058 data: 0.1129 max mem: 3534 +Test: [ 840/2387] eta: 0:05:10 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2691 (0.3058) acc1: 94.7917 (86.9191) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1834 data: 0.0909 max mem: 3534 +Test: [ 850/2387] eta: 0:05:08 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2143 (0.3047) acc1: 97.6562 (87.0355) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1746 data: 0.0831 max mem: 3534 +Test: [ 860/2387] eta: 0:05:05 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2006 (0.3035) acc1: 98.6979 (87.1712) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1666 data: 0.0750 max mem: 3534 +Test: [ 870/2387] eta: 0:05:02 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1996 (0.3022) acc1: 98.9583 (87.2976) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1635 data: 0.0720 max mem: 3534 +Test: [ 880/2387] eta: 0:05:00 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2003 (0.3011) acc1: 98.1771 (87.4190) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1644 data: 0.0725 max mem: 3534 +Test: [ 890/2387] eta: 0:04:58 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2054 (0.3002) acc1: 98.1771 (87.5386) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1811 data: 0.0860 max mem: 3534 +Test: [ 900/2387] eta: 0:04:56 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2302 (0.3027) acc1: 96.3542 (87.3997) acc5: 100.0000 (100.0000) acc1_neg: 69.010 acc5_neg: 100.000 time: 0.2046 data: 0.1077 max mem: 3535 +Test: [ 910/2387] eta: 0:04:52 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.6024 (0.3061) acc1: 69.5312 (87.1990) acc5: 100.0000 (100.0000) acc1_neg: 69.792 acc5_neg: 100.000 time: 0.1585 data: 0.0624 max mem: 3535 +Test: [ 920/2387] eta: 0:04:50 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.6123 (0.3094) acc1: 68.7500 (86.9984) acc5: 100.0000 (100.0000) acc1_neg: 67.969 acc5_neg: 100.000 time: 0.1536 data: 0.0588 max mem: 3535 +Test: [ 930/2387] eta: 0:04:48 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.6123 (0.3126) acc1: 68.2292 (86.7974) acc5: 100.0000 (100.0000) acc1_neg: 65.104 acc5_neg: 100.000 time: 0.1854 data: 0.0921 max mem: 3535 +Test: [ 940/2387] eta: 0:04:46 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.6181 (0.3158) acc1: 68.4896 (86.6056) acc5: 100.0000 (100.0000) acc1_neg: 64.583 acc5_neg: 100.000 time: 0.1753 data: 0.0812 max mem: 3535 +Test: [ 950/2387] eta: 0:04:44 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.6111 (0.3189) acc1: 69.5312 (86.4145) acc5: 100.0000 (100.0000) acc1_neg: 69.531 acc5_neg: 100.000 time: 0.1795 data: 0.0841 max mem: 3535 +Test: [ 960/2387] eta: 0:04:41 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.5936 (0.3217) acc1: 70.3125 (86.2453) acc5: 100.0000 (100.0000) acc1_neg: 70.573 acc5_neg: 100.000 time: 0.1695 data: 0.0741 max mem: 3535 +Test: [ 970/2387] eta: 0:04:39 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.5878 (0.3247) acc1: 70.5729 (86.0700) acc5: 100.0000 (100.0000) acc1_neg: 71.094 acc5_neg: 100.000 time: 0.1658 data: 0.0726 max mem: 3535 +Test: [ 980/2387] eta: 0:04:37 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.5861 (0.3252) acc1: 71.0938 (86.0636) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.1851 data: 0.0940 max mem: 3535 +Test: [ 990/2387] eta: 0:04:35 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2108 (0.3242) acc1: 95.3125 (86.1448) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1976 data: 0.1038 max mem: 3535 +Test: [1000/2387] eta: 0:04:33 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1710 (0.3227) acc1: 98.9583 (86.2619) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2037 data: 0.1082 max mem: 3535 +Test: [1010/2387] eta: 0:04:31 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1757 (0.3220) acc1: 98.4375 (86.3388) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.1964 data: 0.1015 max mem: 3535 +Test: [1020/2387] eta: 0:04:29 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2276 (0.3212) acc1: 96.6146 (86.4117) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2023 data: 0.1087 max mem: 3535 +Test: [1030/2387] eta: 0:04:27 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1551 (0.3196) acc1: 99.2188 (86.5387) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2069 data: 0.1143 max mem: 3535 +Test: [1040/2387] eta: 0:04:25 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1256 (0.3177) acc1: 100.0000 (86.6660) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2001 data: 0.1040 max mem: 3535 +Test: [1050/2387] eta: 0:04:23 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1253 (0.3159) acc1: 100.0000 (86.7918) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2084 data: 0.1108 max mem: 3535 +Test: [1060/2387] eta: 0:04:21 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1321 (0.3143) acc1: 99.7396 (86.9109) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2045 data: 0.1093 max mem: 3535 +Test: [1070/2387] eta: 0:04:20 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1617 (0.3132) acc1: 99.2188 (87.0125) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.2115 data: 0.1180 max mem: 3535 +Test: [1080/2387] eta: 0:04:18 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2469 (0.3130) acc1: 93.4896 (87.0628) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2141 data: 0.1226 max mem: 3535 +Test: [1090/2387] eta: 0:04:16 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2327 (0.3121) acc1: 94.0104 (87.1465) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2028 data: 0.1115 max mem: 3535 +Test: [1100/2387] eta: 0:04:14 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1972 (0.3110) acc1: 96.6146 (87.2339) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1907 data: 0.0985 max mem: 3535 +Test: [1110/2387] eta: 0:04:12 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1972 (0.3102) acc1: 97.1354 (87.3015) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2063 data: 0.1148 max mem: 3535 +Test: [1120/2387] eta: 0:04:10 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2076 (0.3094) acc1: 96.6146 (87.3748) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2000 data: 0.1083 max mem: 3536 +Test: [1130/2387] eta: 0:04:09 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1984 (0.3085) acc1: 96.8750 (87.4588) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2234 data: 0.1300 max mem: 3536 +Test: [1140/2387] eta: 0:04:06 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2569 (0.3084) acc1: 95.5729 (87.5071) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2236 data: 0.1278 max mem: 3536 +Test: [1150/2387] eta: 0:04:04 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2684 (0.3078) acc1: 94.5312 (87.5760) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1575 data: 0.0617 max mem: 3536 +Test: [1160/2387] eta: 0:04:02 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1646 (0.3064) acc1: 98.1771 (87.6694) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1725 data: 0.0800 max mem: 3536 +Test: [1170/2387] eta: 0:04:00 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1437 (0.3052) acc1: 98.9583 (87.7649) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1987 data: 0.1067 max mem: 3536 +Test: [1180/2387] eta: 0:03:58 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1583 (0.3040) acc1: 98.9583 (87.8530) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2042 data: 0.1116 max mem: 3536 +Test: [1190/2387] eta: 0:03:56 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1583 (0.3028) acc1: 99.2188 (87.9471) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1907 data: 0.0993 max mem: 3536 +Test: [1200/2387] eta: 0:03:54 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1349 (0.3015) acc1: 99.7396 (88.0462) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1794 data: 0.0884 max mem: 3536 +Test: [1210/2387] eta: 0:03:52 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1380 (0.3002) acc1: 100.0000 (88.1417) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1793 data: 0.0880 max mem: 3536 +Test: [1220/2387] eta: 0:03:50 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0967 (0.2985) acc1: 99.7396 (88.2373) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1841 data: 0.0929 max mem: 3536 +Test: [1230/2387] eta: 0:03:48 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0825 (0.2967) acc1: 100.0000 (88.3324) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1933 data: 0.1007 max mem: 3536 +Test: [1240/2387] eta: 0:03:46 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0822 (0.2950) acc1: 100.0000 (88.4258) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1959 data: 0.1033 max mem: 3536 +Test: [1250/2387] eta: 0:03:44 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0832 (0.2933) acc1: 100.0000 (88.5177) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2030 data: 0.1117 max mem: 3536 +Test: [1260/2387] eta: 0:03:42 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0788 (0.2916) acc1: 100.0000 (88.6088) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2038 data: 0.1127 max mem: 3536 +Test: [1270/2387] eta: 0:03:40 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0864 (0.2902) acc1: 100.0000 (88.6978) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1828 data: 0.0914 max mem: 3536 +Test: [1280/2387] eta: 0:03:37 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1088 (0.2889) acc1: 100.0000 (88.7834) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1711 data: 0.0789 max mem: 3536 +Test: [1290/2387] eta: 0:03:35 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1159 (0.2876) acc1: 99.7396 (88.8588) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1702 data: 0.0778 max mem: 3536 +Test: [1300/2387] eta: 0:03:33 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1213 (0.2863) acc1: 99.7396 (88.9374) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1735 data: 0.0816 max mem: 3536 +Test: [1310/2387] eta: 0:03:31 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1287 (0.2852) acc1: 99.4792 (89.0118) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1853 data: 0.0943 max mem: 3536 +Test: [1320/2387] eta: 0:03:29 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1355 (0.2841) acc1: 99.2188 (89.0869) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1794 data: 0.0871 max mem: 3536 +Test: [1330/2387] eta: 0:03:27 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1500 (0.2836) acc1: 98.9583 (89.1415) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1777 data: 0.0851 max mem: 3536 +Test: [1340/2387] eta: 0:03:25 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2612 (0.2837) acc1: 93.7500 (89.1516) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1759 data: 0.0830 max mem: 3536 +Test: [1350/2387] eta: 0:03:23 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0533 (0.2819) acc1: 100.0000 (89.2319) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1668 data: 0.0731 max mem: 3537 +Test: [1360/2387] eta: 0:03:20 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0394 (0.2801) acc1: 100.0000 (89.3111) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1778 data: 0.0856 max mem: 3537 +Test: [1370/2387] eta: 0:03:18 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0371 (0.2784) acc1: 100.0000 (89.3890) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1757 data: 0.0824 max mem: 3537 +Test: [1380/2387] eta: 0:03:16 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0399 (0.2767) acc1: 100.0000 (89.4659) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1768 data: 0.0834 max mem: 3537 +Test: [1390/2387] eta: 0:03:14 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0436 (0.2750) acc1: 100.0000 (89.5416) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1897 data: 0.0952 max mem: 3537 +Test: [1400/2387] eta: 0:03:12 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0476 (0.2734) acc1: 100.0000 (89.6162) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1847 data: 0.0908 max mem: 3537 +Test: [1410/2387] eta: 0:03:10 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0622 (0.2727) acc1: 100.0000 (89.6717) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1722 data: 0.0800 max mem: 3537 +Test: [1420/2387] eta: 0:03:08 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1949 (0.2723) acc1: 97.1354 (89.7056) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1801 data: 0.0870 max mem: 3537 +Test: [1430/2387] eta: 0:03:06 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2060 (0.2719) acc1: 96.6146 (89.7511) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1950 data: 0.1013 max mem: 3537 +Test: [1440/2387] eta: 0:03:04 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2105 (0.2715) acc1: 97.3958 (89.8058) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1780 data: 0.0849 max mem: 3537 +Test: [1450/2387] eta: 0:03:02 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2328 (0.2715) acc1: 97.1354 (89.8271) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1744 data: 0.0820 max mem: 3537 +Test: [1460/2387] eta: 0:03:00 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1616 (0.2707) acc1: 97.1354 (89.8881) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1803 data: 0.0880 max mem: 3537 +Test: [1470/2387] eta: 0:02:58 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1283 (0.2697) acc1: 99.4792 (89.9546) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1870 data: 0.0959 max mem: 3537 +Test: [1480/2387] eta: 0:02:56 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1250 (0.2687) acc1: 99.7396 (90.0170) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1960 data: 0.1047 max mem: 3537 +Test: [1490/2387] eta: 0:02:54 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1223 (0.2678) acc1: 99.7396 (90.0815) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1995 data: 0.1080 max mem: 3537 +Test: [1500/2387] eta: 0:02:52 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1271 (0.2668) acc1: 99.7396 (90.1432) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2033 data: 0.1097 max mem: 3537 +Test: [1510/2387] eta: 0:02:50 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1339 (0.2661) acc1: 99.7396 (90.2077) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1750 data: 0.0791 max mem: 3537 +Test: [1520/2387] eta: 0:02:48 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1653 (0.2656) acc1: 99.7396 (90.2617) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1709 data: 0.0778 max mem: 3537 +Test: [1530/2387] eta: 0:02:46 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1824 (0.2651) acc1: 98.6979 (90.3113) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1846 data: 0.0936 max mem: 3537 +Test: [1540/2387] eta: 0:02:44 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1668 (0.2644) acc1: 98.9583 (90.3673) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1893 data: 0.0971 max mem: 3537 +Test: [1550/2387] eta: 0:02:42 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1636 (0.2638) acc1: 99.4792 (90.4232) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2034 data: 0.1100 max mem: 3537 +Test: [1560/2387] eta: 0:02:40 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1562 (0.2633) acc1: 98.9583 (90.4764) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1853 data: 0.0900 max mem: 3537 +Test: [1570/2387] eta: 0:02:38 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1722 (0.2628) acc1: 98.9583 (90.5290) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1606 data: 0.0653 max mem: 3537 +Test: [1580/2387] eta: 0:02:36 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1722 (0.2622) acc1: 98.6979 (90.5805) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1652 data: 0.0708 max mem: 3538 +Test: [1590/2387] eta: 0:02:34 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1673 (0.2616) acc1: 98.9583 (90.6348) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1700 data: 0.0766 max mem: 3538 +Test: [1600/2387] eta: 0:02:32 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1426 (0.2609) acc1: 99.2188 (90.6722) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1743 data: 0.0811 max mem: 3538 +Test: [1610/2387] eta: 0:02:30 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1465 (0.2603) acc1: 99.2188 (90.7226) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1725 data: 0.0795 max mem: 3538 +Test: [1620/2387] eta: 0:02:28 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1511 (0.2596) acc1: 98.9583 (90.7715) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1707 data: 0.0787 max mem: 3538 +Test: [1630/2387] eta: 0:02:26 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1715 (0.2595) acc1: 98.1771 (90.8065) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1768 data: 0.0840 max mem: 3538 +Test: [1640/2387] eta: 0:02:24 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2662 (0.2597) acc1: 94.7917 (90.8100) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1833 data: 0.0910 max mem: 3538 +Test: [1650/2387] eta: 0:02:22 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2736 (0.2598) acc1: 92.4479 (90.8241) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1832 data: 0.0884 max mem: 3538 +Test: [1660/2387] eta: 0:02:20 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2756 (0.2600) acc1: 92.4479 (90.8260) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1940 data: 0.0988 max mem: 3538 +Test: [1670/2387] eta: 0:02:18 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2762 (0.2601) acc1: 92.7083 (90.8399) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2008 data: 0.1087 max mem: 3538 +Test: [1680/2387] eta: 0:02:16 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2751 (0.2603) acc1: 93.4896 (90.8526) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1756 data: 0.0818 max mem: 3538 +Test: [1690/2387] eta: 0:02:14 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2841 (0.2605) acc1: 92.4479 (90.8540) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1973 data: 0.1035 max mem: 3538 +Test: [1700/2387] eta: 0:02:12 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2800 (0.2603) acc1: 94.0104 (90.8767) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2044 data: 0.1125 max mem: 3538 +Test: [1710/2387] eta: 0:02:10 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1149 (0.2594) acc1: 99.4792 (90.9285) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1840 data: 0.0901 max mem: 3538 +Test: [1720/2387] eta: 0:02:09 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0987 (0.2585) acc1: 100.0000 (90.9801) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2411 data: 0.1473 max mem: 3538 +Test: [1730/2387] eta: 0:02:07 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0973 (0.2576) acc1: 100.0000 (91.0289) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2349 data: 0.1410 max mem: 3538 +Test: [1740/2387] eta: 0:02:05 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0937 (0.2566) acc1: 100.0000 (91.0793) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1752 data: 0.0816 max mem: 3538 +Test: [1750/2387] eta: 0:02:03 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.0925 (0.2557) acc1: 100.0000 (91.1289) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1891 data: 0.0969 max mem: 3538 +Test: [1760/2387] eta: 0:02:01 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1034 (0.2550) acc1: 99.7396 (91.1756) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2038 data: 0.1111 max mem: 3538 +Test: [1770/2387] eta: 0:01:59 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1322 (0.2543) acc1: 99.2188 (91.2192) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1982 data: 0.1046 max mem: 3538 +Test: [1780/2387] eta: 0:01:57 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1177 (0.2535) acc1: 99.7396 (91.2678) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2266 data: 0.1306 max mem: 3538 +Test: [1790/2387] eta: 0:01:55 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1092 (0.2527) acc1: 99.7396 (91.3142) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2160 data: 0.1168 max mem: 3538 +Test: [1800/2387] eta: 0:01:53 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1040 (0.2519) acc1: 99.7396 (91.3595) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1822 data: 0.0849 max mem: 3538 +Test: [1810/2387] eta: 0:01:51 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1019 (0.2510) acc1: 99.7396 (91.4061) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2088 data: 0.1152 max mem: 3539 +Test: [1820/2387] eta: 0:01:49 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1010 (0.2503) acc1: 100.0000 (91.4507) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2061 data: 0.1119 max mem: 3539 +Test: [1830/2387] eta: 0:01:47 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1203 (0.2497) acc1: 100.0000 (91.4944) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1731 data: 0.0794 max mem: 3539 +Test: [1840/2387] eta: 0:01:46 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1204 (0.2489) acc1: 100.0000 (91.5406) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2187 data: 0.1268 max mem: 3539 +Test: [1850/2387] eta: 0:01:44 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1180 (0.2482) acc1: 100.0000 (91.5862) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2217 data: 0.1255 max mem: 3539 +Test: [1860/2387] eta: 0:01:42 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1180 (0.2475) acc1: 100.0000 (91.6314) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1777 data: 0.0802 max mem: 3539 +Test: [1870/2387] eta: 0:01:40 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1180 (0.2469) acc1: 100.0000 (91.6761) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0833 max mem: 3539 +Test: [1880/2387] eta: 0:01:38 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1176 (0.2462) acc1: 100.0000 (91.7202) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2226 data: 0.1258 max mem: 3539 +Test: [1890/2387] eta: 0:01:36 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1178 (0.2455) acc1: 100.0000 (91.7640) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2220 data: 0.1271 max mem: 3539 +Test: [1900/2387] eta: 0:01:34 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1200 (0.2449) acc1: 100.0000 (91.8071) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1566 data: 0.0636 max mem: 3539 +Test: [1910/2387] eta: 0:01:32 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1773 (0.2446) acc1: 100.0000 (91.8464) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1455 data: 0.0546 max mem: 3539 +Test: [1920/2387] eta: 0:01:30 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1869 (0.2444) acc1: 99.4792 (91.8624) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1752 data: 0.0842 max mem: 3539 +Test: [1930/2387] eta: 0:01:28 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1549 (0.2439) acc1: 98.9583 (91.8994) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1994 data: 0.1085 max mem: 3539 +Test: [1940/2387] eta: 0:01:26 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1566 (0.2437) acc1: 99.4792 (91.8982) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1841 data: 0.0913 max mem: 3539 +Test: [1950/2387] eta: 0:01:24 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1694 (0.2434) acc1: 98.4375 (91.9215) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1726 data: 0.0774 max mem: 3539 +Test: [1960/2387] eta: 0:01:22 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1868 (0.2432) acc1: 98.4375 (91.9494) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1802 data: 0.0856 max mem: 3539 +Test: [1970/2387] eta: 0:01:20 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2504 (0.2438) acc1: 93.2292 (91.9078) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.1799 data: 0.0854 max mem: 3539 +Test: [1980/2387] eta: 0:01:18 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2977 (0.2438) acc1: 88.8021 (91.9122) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1911 data: 0.0974 max mem: 3539 +Test: [1990/2387] eta: 0:01:16 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1331 (0.2433) acc1: 98.4375 (91.9454) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1925 data: 0.1012 max mem: 3539 +Test: [2000/2387] eta: 0:01:14 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1348 (0.2428) acc1: 98.4375 (91.9775) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1804 data: 0.0873 max mem: 3539 +Test: [2010/2387] eta: 0:01:12 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1386 (0.2423) acc1: 98.4375 (92.0098) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1828 data: 0.0893 max mem: 3539 +Test: [2020/2387] eta: 0:01:10 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1352 (0.2417) acc1: 98.4375 (92.0427) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1812 data: 0.0880 max mem: 3539 +Test: [2030/2387] eta: 0:01:09 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1365 (0.2412) acc1: 98.4375 (92.0739) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1924 data: 0.0968 max mem: 3540 +Test: [2040/2387] eta: 0:01:07 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1354 (0.2407) acc1: 98.1771 (92.1058) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1960 data: 0.1017 max mem: 3540 +Test: [2050/2387] eta: 0:01:05 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1342 (0.2402) acc1: 98.9583 (92.1385) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1819 data: 0.0897 max mem: 3540 +Test: [2060/2387] eta: 0:01:03 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1399 (0.2397) acc1: 98.9583 (92.1709) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1765 data: 0.0836 max mem: 3540 +Test: [2070/2387] eta: 0:01:01 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1429 (0.2393) acc1: 98.9583 (92.2044) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1805 data: 0.0875 max mem: 3540 +Test: [2080/2387] eta: 0:00:59 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1404 (0.2388) acc1: 98.9583 (92.2369) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1827 data: 0.0868 max mem: 3540 +Test: [2090/2387] eta: 0:00:57 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1408 (0.2383) acc1: 99.2188 (92.2703) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1847 data: 0.0878 max mem: 3540 +Test: [2100/2387] eta: 0:00:55 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1408 (0.2379) acc1: 99.2188 (92.3027) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1993 data: 0.1042 max mem: 3540 +Test: [2110/2387] eta: 0:00:53 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1408 (0.2374) acc1: 98.9583 (92.3349) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1955 data: 0.1023 max mem: 3540 +Test: [2120/2387] eta: 0:00:51 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1515 (0.2372) acc1: 98.9583 (92.3614) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1649 data: 0.0720 max mem: 3540 +Test: [2130/2387] eta: 0:00:49 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1767 (0.2368) acc1: 98.6979 (92.3951) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1690 data: 0.0742 max mem: 3540 +Test: [2140/2387] eta: 0:00:47 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1497 (0.2366) acc1: 99.7396 (92.4169) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1899 data: 0.0921 max mem: 3540 +Test: [2150/2387] eta: 0:00:45 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1761 (0.2363) acc1: 98.9583 (92.4454) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1665 data: 0.0685 max mem: 3540 +Test: [2160/2387] eta: 0:00:43 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1802 (0.2361) acc1: 98.9583 (92.4724) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1516 data: 0.0587 max mem: 3540 +Test: [2170/2387] eta: 0:00:41 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2061 (0.2363) acc1: 97.6562 (92.4805) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1765 data: 0.0841 max mem: 3540 +Test: [2180/2387] eta: 0:00:39 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.3184 (0.2376) acc1: 90.1042 (92.3996) acc5: 100.0000 (100.0000) acc1_neg: 51.042 acc5_neg: 100.000 time: 0.1844 data: 0.0914 max mem: 3540 +Test: [2190/2387] eta: 0:00:37 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.3184 (0.2377) acc1: 90.6250 (92.3907) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1921 data: 0.0977 max mem: 3540 +Test: [2200/2387] eta: 0:00:35 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1445 (0.2373) acc1: 98.6979 (92.4214) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1792 data: 0.0835 max mem: 3540 +Test: [2210/2387] eta: 0:00:34 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1284 (0.2368) acc1: 99.7396 (92.4539) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1612 data: 0.0683 max mem: 3540 +Test: [2220/2387] eta: 0:00:32 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1333 (0.2363) acc1: 99.7396 (92.4833) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1947 data: 0.1025 max mem: 3540 +Test: [2230/2387] eta: 0:00:30 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1368 (0.2359) acc1: 99.4792 (92.5153) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1959 data: 0.1029 max mem: 3540 +Test: [2240/2387] eta: 0:00:28 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1235 (0.2354) acc1: 99.7396 (92.5453) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1814 data: 0.0887 max mem: 3540 +Test: [2250/2387] eta: 0:00:26 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1443 (0.2351) acc1: 98.9583 (92.5703) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1871 data: 0.0927 max mem: 3540 +Test: [2260/2387] eta: 0:00:24 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1587 (0.2348) acc1: 99.4792 (92.6016) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1844 data: 0.0895 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1501 (0.2344) acc1: 100.0000 (92.6339) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1814 data: 0.0888 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1275 (0.2340) acc1: 100.0000 (92.6548) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1904 data: 0.0993 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1272 (0.2336) acc1: 100.0000 (92.6864) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2055 data: 0.1148 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1455 (0.2332) acc1: 100.0000 (92.7147) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1759 data: 0.0851 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1633 (0.2331) acc1: 99.2188 (92.7371) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1528 data: 0.0597 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2421 (0.2333) acc1: 96.6146 (92.7449) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1627 data: 0.0688 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2370 (0.2333) acc1: 96.8750 (92.7634) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1533 data: 0.0614 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2001 (0.2331) acc1: 97.6562 (92.7824) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1421 data: 0.0513 max mem: 3541 +Test: [2350/2387] eta: 0:00:07 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.1945 (0.2332) acc1: 97.1354 (92.7859) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1344 data: 0.0440 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2178 (0.2332) acc1: 96.6146 (92.7979) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1276 data: 0.0370 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.2329 (0.2333) acc1: 96.6146 (92.8036) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.1246 data: 0.0338 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.3016 (0.2338) acc1: 92.7083 (92.7830) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.1213 data: 0.0307 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 69.863 acc5_pos: 100.000 loss: 0.3232 (0.2340) acc1: 89.3229 (92.7739) acc5: 100.0000 (100.0000) acc1_neg: 93.023 acc5_neg: 100.000 time: 0.1146 data: 0.0269 max mem: 3541 +Test: Total time: 0:07:33 (0.1901 s / it) +* Overall Acc@1 92.774 Acc@5 100.000 + target == real Acc@1 42.089 Acc@5 100.000 + target == fake Acc@1 96.333 Acc@5 100.000 + loss 0.234 + PR-AUC 0.9944 ROC-AUC 0.9260 EER 14.09% (thr=0.7173) +Accuracy of the network on the 916267 test images: 92.8% +[rank0]:[W726 03:10:00.208744579 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:10:19.914000 1173015 torch/distributed/run.py:766] +W0726 03:10:19.914000 1173015 torch/distributed/run.py:766] ***************************************** +W0726 03:10:19.914000 1173015 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:10:19.914000 1173015 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:10:49.553308937 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:10:49.154842346 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:10:50.380404043 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:10:50.585398944 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:10:50.591525087 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:10:50.614906004 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:10:50.689954719 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:10:50.707358596 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n32_seed5_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n32_classseed5_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:00:29 acc1_pos: 48.698 acc5_pos: 100.000 loss: 0.7613 (0.7613) acc1: 48.6979 (48.6979) acc5: 100.0000 (100.0000) time: 7.0196 data: 5.0139 max mem: 3531 +Test: [ 10/517] eta: 0:06:05 acc1_pos: 22.396 acc5_pos: 100.000 loss: 0.9578 (0.9759) acc1: 26.0417 (26.9886) acc5: 100.0000 (100.0000) time: 0.7207 data: 0.4562 max mem: 3531 +Test: [ 20/517] eta: 0:03:30 acc1_pos: 17.448 acc5_pos: 100.000 loss: 0.9464 (0.9979) acc1: 22.3958 (24.9752) acc5: 100.0000 (100.0000) time: 0.0945 data: 0.0024 max mem: 3531 +Test: [ 30/517] eta: 0:02:56 acc1_pos: 6.510 acc5_pos: 100.000 loss: 0.9967 (1.0092) acc1: 19.5312 (23.8155) acc5: 100.0000 (100.0000) time: 0.1640 data: 0.0678 max mem: 3531 +Test: [ 40/517] eta: 0:02:39 acc1_pos: 12.500 acc5_pos: 100.000 loss: 1.0393 (1.0300) acc1: 14.5833 (21.1890) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1414 max mem: 3531 +Test: [ 50/517] eta: 0:02:28 acc1_pos: 65.596 acc5_pos: 100.000 loss: 1.0105 (0.9868) acc1: 17.9688 (26.2766) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.2500 data: 0.1546 max mem: 3531 +Test: [ 60/517] eta: 0:02:21 acc1_pos: 65.596 acc5_pos: 100.000 loss: 0.5048 (0.8979) acc1: 77.8646 (36.4071) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2599 data: 0.1667 max mem: 3531 +Test: [ 70/517] eta: 0:02:14 acc1_pos: 65.596 acc5_pos: 100.000 loss: 0.3606 (0.8183) acc1: 95.5729 (44.8797) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2549 data: 0.1609 max mem: 3531 +Test: [ 80/517] eta: 0:02:09 acc1_pos: 65.596 acc5_pos: 100.000 loss: 0.3290 (0.7597) acc1: 96.6146 (51.2635) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.2549 data: 0.1573 max mem: 3531 +Test: [ 90/517] eta: 0:02:04 acc1_pos: 65.596 acc5_pos: 100.000 loss: 0.3527 (0.7145) acc1: 97.6562 (56.3673) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2659 data: 0.1664 max mem: 3531 +Test: [100/517] eta: 0:02:01 acc1_pos: 65.596 acc5_pos: 100.000 loss: 0.3305 (0.6771) acc1: 98.4375 (60.5482) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2664 data: 0.1710 max mem: 3531 +Test: [110/517] eta: 0:01:56 acc1_pos: 65.596 acc5_pos: 100.000 loss: 0.3220 (0.6440) acc1: 98.6979 (63.8819) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2639 data: 0.1670 max mem: 3531 +Test: [120/517] eta: 0:01:52 acc1_pos: 65.596 acc5_pos: 100.000 loss: 0.3512 (0.6222) acc1: 95.0521 (66.2771) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2534 data: 0.1540 max mem: 3531 +Test: [130/517] eta: 0:01:49 acc1_pos: 65.596 acc5_pos: 100.000 loss: 0.3726 (0.6022) acc1: 93.7500 (68.4637) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2599 data: 0.1610 max mem: 3531 +Test: [140/517] eta: 0:01:45 acc1_pos: 65.596 acc5_pos: 100.000 loss: 0.3517 (0.5855) acc1: 95.8333 (70.4048) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2549 data: 0.1578 max mem: 3531 +Test: [150/517] eta: 0:01:42 acc1_pos: 65.596 acc5_pos: 100.000 loss: 0.3477 (0.5678) acc1: 97.6562 (72.0785) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2449 data: 0.1498 max mem: 3531 +Test: [160/517] eta: 0:01:38 acc1_pos: 65.596 acc5_pos: 100.000 loss: 0.2978 (0.5504) acc1: 98.9583 (73.7659) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2499 data: 0.1555 max mem: 3531 +Test: [170/517] eta: 0:01:34 acc1_pos: 65.596 acc5_pos: 100.000 loss: 0.2839 (0.5338) acc1: 99.2188 (75.2513) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2399 data: 0.1413 max mem: 3531 +Test: [180/517] eta: 0:01:31 acc1_pos: 65.596 acc5_pos: 100.000 loss: 0.2674 (0.5219) acc1: 98.9583 (76.3640) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.2349 data: 0.1349 max mem: 3531 +Test: [190/517] eta: 0:01:28 acc1_pos: 65.596 acc5_pos: 100.000 loss: 0.2153 (0.5046) acc1: 98.9583 (77.5619) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2549 data: 0.1563 max mem: 3531 +Test: [200/517] eta: 0:01:25 acc1_pos: 65.596 acc5_pos: 100.000 loss: 0.2072 (0.4930) acc1: 99.4792 (78.6458) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2399 data: 0.1422 max mem: 3531 +Test: [210/517] eta: 0:01:22 acc1_pos: 65.596 acc5_pos: 100.000 loss: 0.3204 (0.4875) acc1: 95.8333 (79.1124) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.2299 data: 0.1350 max mem: 3532 +Test: [220/517] eta: 0:01:19 acc1_pos: 65.596 acc5_pos: 100.000 loss: 0.3801 (0.4824) acc1: 87.5000 (79.5614) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.2451 data: 0.1516 max mem: 3532 +Test: [230/517] eta: 0:01:16 acc1_pos: 65.596 acc5_pos: 100.000 loss: 0.3973 (0.4789) acc1: 92.7083 (80.2072) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2449 data: 0.1505 max mem: 3532 +Test: [240/517] eta: 0:01:13 acc1_pos: 65.596 acc5_pos: 100.000 loss: 0.3635 (0.4712) acc1: 95.8333 (80.9096) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2497 data: 0.1525 max mem: 3532 +Test: [250/517] eta: 0:01:10 acc1_pos: 65.596 acc5_pos: 100.000 loss: 0.3035 (0.4653) acc1: 98.4375 (81.6069) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2499 data: 0.1531 max mem: 3532 +Test: [260/517] eta: 0:01:07 acc1_pos: 65.596 acc5_pos: 100.000 loss: 0.3169 (0.4602) acc1: 99.2188 (82.2637) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2449 data: 0.1499 max mem: 3532 +Test: [270/517] eta: 0:01:04 acc1_pos: 65.596 acc5_pos: 100.000 loss: 0.3518 (0.4570) acc1: 97.1354 (82.7500) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2449 data: 0.1487 max mem: 3532 +Test: [280/517] eta: 0:01:02 acc1_pos: 65.596 acc5_pos: 100.000 loss: 0.3579 (0.4531) acc1: 96.6146 (83.2416) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2449 data: 0.1479 max mem: 3532 +Test: [290/517] eta: 0:00:59 acc1_pos: 65.596 acc5_pos: 100.000 loss: 0.3269 (0.4490) acc1: 97.6562 (83.7504) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2399 data: 0.1429 max mem: 3532 +Test: [300/517] eta: 0:00:56 acc1_pos: 65.596 acc5_pos: 100.000 loss: 0.3228 (0.4445) acc1: 98.4375 (84.2348) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2349 data: 0.1379 max mem: 3532 +Test: [310/517] eta: 0:00:53 acc1_pos: 65.596 acc5_pos: 100.000 loss: 0.3536 (0.4427) acc1: 98.1771 (84.6672) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2399 data: 0.1445 max mem: 3532 +Test: [320/517] eta: 0:00:51 acc1_pos: 65.596 acc5_pos: 100.000 loss: 0.3574 (0.4394) acc1: 97.6562 (85.0411) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2549 data: 0.1589 max mem: 3532 +Test: [330/517] eta: 0:00:48 acc1_pos: 65.596 acc5_pos: 100.000 loss: 0.3055 (0.4349) acc1: 98.1771 (85.4324) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2499 data: 0.1534 max mem: 3532 +Test: [340/517] eta: 0:00:45 acc1_pos: 65.596 acc5_pos: 100.000 loss: 0.2951 (0.4318) acc1: 98.4375 (85.7809) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.2399 data: 0.1426 max mem: 3532 +Test: [350/517] eta: 0:00:43 acc1_pos: 65.596 acc5_pos: 100.000 loss: 0.3631 (0.4305) acc1: 97.1354 (86.0488) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2500 data: 0.1540 max mem: 3532 +Test: [360/517] eta: 0:00:40 acc1_pos: 65.596 acc5_pos: 100.000 loss: 0.3752 (0.4284) acc1: 96.0938 (86.3177) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2549 data: 0.1613 max mem: 3532 +Test: [370/517] eta: 0:00:37 acc1_pos: 65.596 acc5_pos: 100.000 loss: 0.2880 (0.4241) acc1: 99.4792 (86.6808) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2498 data: 0.1557 max mem: 3532 +Test: [380/517] eta: 0:00:35 acc1_pos: 65.596 acc5_pos: 100.000 loss: 0.2763 (0.4208) acc1: 99.7396 (86.9737) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2448 data: 0.1491 max mem: 3532 +Test: [390/517] eta: 0:00:32 acc1_pos: 65.596 acc5_pos: 100.000 loss: 0.2962 (0.4172) acc1: 98.6979 (87.2742) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2399 data: 0.1430 max mem: 3532 +Test: [400/517] eta: 0:00:30 acc1_pos: 65.596 acc5_pos: 100.000 loss: 0.2881 (0.4143) acc1: 99.4792 (87.5747) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2399 data: 0.1417 max mem: 3532 +Test: [410/517] eta: 0:00:27 acc1_pos: 65.596 acc5_pos: 100.000 loss: 0.3224 (0.4125) acc1: 99.2188 (87.8257) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2399 data: 0.1411 max mem: 3532 +Test: [420/517] eta: 0:00:24 acc1_pos: 65.596 acc5_pos: 100.000 loss: 0.3386 (0.4106) acc1: 99.2188 (88.0951) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2399 data: 0.1419 max mem: 3532 +Test: [430/517] eta: 0:00:22 acc1_pos: 65.596 acc5_pos: 100.000 loss: 0.3199 (0.4080) acc1: 98.9583 (88.3223) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2349 data: 0.1388 max mem: 3532 +Test: [440/517] eta: 0:00:19 acc1_pos: 65.596 acc5_pos: 100.000 loss: 0.3036 (0.4059) acc1: 98.4375 (88.5369) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2449 data: 0.1481 max mem: 3533 +Test: [450/517] eta: 0:00:17 acc1_pos: 65.596 acc5_pos: 100.000 loss: 0.2934 (0.4032) acc1: 99.4792 (88.7819) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1578 max mem: 3533 +Test: [460/517] eta: 0:00:14 acc1_pos: 65.596 acc5_pos: 100.000 loss: 0.3022 (0.4015) acc1: 99.7396 (88.9806) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2499 data: 0.1541 max mem: 3533 +Test: [470/517] eta: 0:00:11 acc1_pos: 65.596 acc5_pos: 100.000 loss: 0.3104 (0.3993) acc1: 99.2188 (89.1891) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2449 data: 0.1474 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 65.596 acc5_pos: 100.000 loss: 0.2970 (0.3969) acc1: 99.7396 (89.4025) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2349 data: 0.1372 max mem: 3533 +Test: [490/517] eta: 0:00:06 acc1_pos: 65.596 acc5_pos: 100.000 loss: 0.3064 (0.3958) acc1: 98.1771 (89.4900) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2349 data: 0.1378 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 65.596 acc5_pos: 100.000 loss: 0.3656 (0.3957) acc1: 91.9271 (89.5345) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.2399 data: 0.1431 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 65.596 acc5_pos: 100.000 loss: 0.3863 (0.3956) acc1: 89.8438 (89.5747) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2113 data: 0.1182 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 65.596 acc5_pos: 100.000 loss: 0.3526 (0.3949) acc1: 96.6146 (89.6758) acc5: 100.0000 (100.0000) acc1_neg: 99.293 acc5_neg: 100.000 time: 0.1774 data: 0.0840 max mem: 3533 +Test: Total time: 0:02:10 (0.2526 s / it) +* Overall Acc@1 89.676 Acc@5 100.000 + target == real Acc@1 23.303 Acc@5 100.000 + target == fake Acc@1 96.561 Acc@5 100.000 + loss 0.395 + PR-AUC 0.9580 ROC-AUC 0.7418 EER 32.25% (thr=0.6875) +Accuracy of the network on the 198427 test images: 89.7% +[rank0]:[W726 03:13:12.078437182 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 03:13:34.346000 1173577 torch/distributed/run.py:766] +W0726 03:13:34.346000 1173577 torch/distributed/run.py:766] ***************************************** +W0726 03:13:34.346000 1173577 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:13:34.346000 1173577 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:14:04.548730067 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 03:14:04.088207644 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank6]:[W726 03:14:04.090041854 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:14:05.212651739 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:14:05.318161398 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +[rank5]:[W726 03:14:05.336913440 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:14:05.347440905 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:14:05.419074282 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n32_seed5_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n32_classseed5_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:44:42 acc1_pos: 38.021 acc5_pos: 100.000 loss: 0.7931 (0.7931) acc1: 38.0208 (38.0208) acc5: 100.0000 (100.0000) time: 6.8968 data: 5.0990 max mem: 3531 +Test: [ 10/389] eta: 0:04:28 acc1_pos: 89.844 acc5_pos: 100.000 loss: 0.5884 (0.6138) acc1: 65.8854 (64.8201) acc5: 100.0000 (100.0000) time: 0.7094 data: 0.4640 max mem: 3531 +Test: [ 20/389] eta: 0:02:34 acc1_pos: 10.938 acc5_pos: 100.000 loss: 0.5884 (0.7608) acc1: 62.5000 (54.5387) acc5: 100.0000 (100.0000) time: 0.0945 data: 0.0014 max mem: 3531 +Test: [ 30/389] eta: 0:02:11 acc1_pos: 34.375 acc5_pos: 100.000 loss: 0.8023 (0.7938) acc1: 38.0208 (48.3871) acc5: 100.0000 (100.0000) time: 0.1791 data: 0.0786 max mem: 3531 +Test: [ 40/389] eta: 0:01:59 acc1_pos: 22.135 acc5_pos: 100.000 loss: 0.7980 (0.7959) acc1: 38.0208 (46.5320) acc5: 100.0000 (100.0000) time: 0.2655 data: 0.1663 max mem: 3531 +Test: [ 50/389] eta: 0:01:45 acc1_pos: 64.062 acc5_pos: 100.000 loss: 0.7907 (0.8035) acc1: 37.2396 (45.1848) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1309 max mem: 3531 +Test: [ 60/389] eta: 0:01:34 acc1_pos: 77.604 acc5_pos: 100.000 loss: 0.6753 (0.7770) acc1: 60.6771 (48.1557) acc5: 100.0000 (100.0000) time: 0.1743 data: 0.0786 max mem: 3531 +Test: [ 70/389] eta: 0:01:28 acc1_pos: 63.021 acc5_pos: 100.000 loss: 0.6247 (0.7551) acc1: 63.0208 (50.6309) acc5: 100.0000 (100.0000) time: 0.1912 data: 0.0964 max mem: 3531 +Test: [ 80/389] eta: 0:01:21 acc1_pos: 88.802 acc5_pos: 100.000 loss: 0.6554 (0.7591) acc1: 63.0208 (50.8648) acc5: 100.0000 (100.0000) time: 0.1850 data: 0.0928 max mem: 3531 +Test: [ 90/389] eta: 0:01:13 acc1_pos: 60.417 acc5_pos: 100.000 loss: 0.4614 (0.7281) acc1: 74.7396 (53.6544) acc5: 100.0000 (100.0000) time: 0.1405 data: 0.0479 max mem: 3531 +Test: [100/389] eta: 0:01:09 acc1_pos: 78.385 acc5_pos: 100.000 loss: 0.4872 (0.7276) acc1: 77.3438 (54.2492) acc5: 100.0000 (100.0000) time: 0.1448 data: 0.0505 max mem: 3531 +Test: [110/389] eta: 0:01:04 acc1_pos: 86.979 acc5_pos: 100.000 loss: 0.6248 (0.7374) acc1: 67.7083 (54.4107) acc5: 100.0000 (100.0000) time: 0.1614 data: 0.0672 max mem: 3531 +Test: [120/389] eta: 0:01:01 acc1_pos: 68.229 acc5_pos: 100.000 loss: 0.5973 (0.7198) acc1: 71.3542 (56.1897) acc5: 100.0000 (100.0000) time: 0.1649 data: 0.0683 max mem: 3531 +Test: [130/389] eta: 0:00:57 acc1_pos: 45.399 acc5_pos: 100.000 loss: 0.5768 (0.7097) acc1: 71.3542 (57.1644) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1666 data: 0.0683 max mem: 3531 +Test: [140/389] eta: 0:00:54 acc1_pos: 45.399 acc5_pos: 100.000 loss: 0.5461 (0.7011) acc1: 75.0000 (58.3149) acc5: 100.0000 (100.0000) acc1_neg: 35.938 acc5_neg: 100.000 time: 0.1656 data: 0.0697 max mem: 3531 +Test: [150/389] eta: 0:00:51 acc1_pos: 45.399 acc5_pos: 100.000 loss: 0.5241 (0.6907) acc1: 75.0000 (59.0784) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1749 data: 0.0803 max mem: 3531 +Test: [160/389] eta: 0:00:48 acc1_pos: 45.399 acc5_pos: 100.000 loss: 0.5241 (0.6755) acc1: 72.9167 (60.0738) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1708 data: 0.0764 max mem: 3531 +Test: [170/389] eta: 0:00:46 acc1_pos: 45.399 acc5_pos: 100.000 loss: 0.2540 (0.6636) acc1: 98.6979 (61.0852) acc5: 100.0000 (100.0000) acc1_neg: 57.812 acc5_neg: 100.000 time: 0.1711 data: 0.0782 max mem: 3531 +Test: [180/389] eta: 0:00:43 acc1_pos: 45.399 acc5_pos: 100.000 loss: 0.3016 (0.6523) acc1: 88.2812 (62.2827) acc5: 100.0000 (100.0000) acc1_neg: 65.885 acc5_neg: 100.000 time: 0.1783 data: 0.0866 max mem: 3531 +Test: [190/389] eta: 0:00:41 acc1_pos: 45.399 acc5_pos: 100.000 loss: 0.4401 (0.6543) acc1: 78.6458 (62.0501) acc5: 100.0000 (100.0000) acc1_neg: 5.990 acc5_neg: 100.000 time: 0.1796 data: 0.0870 max mem: 3531 +Test: [200/389] eta: 0:00:38 acc1_pos: 45.399 acc5_pos: 100.000 loss: 0.5979 (0.6541) acc1: 71.3542 (62.0141) acc5: 100.0000 (100.0000) acc1_neg: 67.969 acc5_neg: 100.000 time: 0.1751 data: 0.0810 max mem: 3531 +Test: [210/389] eta: 0:00:36 acc1_pos: 45.399 acc5_pos: 100.000 loss: 0.5314 (0.6391) acc1: 74.4792 (63.2381) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1603 data: 0.0652 max mem: 3532 +Test: [220/389] eta: 0:00:34 acc1_pos: 45.399 acc5_pos: 100.000 loss: 0.2124 (0.6243) acc1: 99.2188 (64.5869) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.1598 data: 0.0648 max mem: 3532 +Test: [230/389] eta: 0:00:32 acc1_pos: 45.399 acc5_pos: 100.000 loss: 0.4879 (0.6223) acc1: 80.2083 (64.7299) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.1931 data: 0.0985 max mem: 3532 +Test: [240/389] eta: 0:00:29 acc1_pos: 45.399 acc5_pos: 100.000 loss: 0.5256 (0.6177) acc1: 72.6562 (65.1441) acc5: 100.0000 (100.0000) acc1_neg: 39.844 acc5_neg: 100.000 time: 0.1995 data: 0.1023 max mem: 3532 +Test: [250/389] eta: 0:00:27 acc1_pos: 45.399 acc5_pos: 100.000 loss: 0.3982 (0.6104) acc1: 82.8125 (65.8055) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1550 data: 0.0560 max mem: 3532 +Test: [260/389] eta: 0:00:25 acc1_pos: 45.399 acc5_pos: 100.000 loss: 0.2848 (0.6039) acc1: 98.4375 (66.4571) acc5: 100.0000 (100.0000) acc1_neg: 16.406 acc5_neg: 100.000 time: 0.1538 data: 0.0557 max mem: 3532 +Test: [270/389] eta: 0:00:23 acc1_pos: 45.399 acc5_pos: 100.000 loss: 0.4730 (0.6127) acc1: 68.4896 (65.6654) acc5: 100.0000 (100.0000) acc1_neg: 65.625 acc5_neg: 100.000 time: 0.1911 data: 0.0970 max mem: 3532 +Test: [280/389] eta: 0:00:21 acc1_pos: 45.399 acc5_pos: 100.000 loss: 0.6213 (0.6169) acc1: 60.1562 (65.3396) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1887 data: 0.0955 max mem: 3532 +Test: [290/389] eta: 0:00:19 acc1_pos: 45.399 acc5_pos: 100.000 loss: 0.5892 (0.6109) acc1: 66.9271 (65.8568) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1328 data: 0.0387 max mem: 3532 +Test: [300/389] eta: 0:00:17 acc1_pos: 45.399 acc5_pos: 100.000 loss: 0.5669 (0.6112) acc1: 72.3958 (65.8612) acc5: 100.0000 (100.0000) acc1_neg: 69.792 acc5_neg: 100.000 time: 0.1322 data: 0.0404 max mem: 3532 +Test: [310/389] eta: 0:00:15 acc1_pos: 45.399 acc5_pos: 100.000 loss: 0.6677 (0.6187) acc1: 58.3333 (65.2206) acc5: 100.0000 (100.0000) acc1_neg: 3.385 acc5_neg: 100.000 time: 0.1824 data: 0.0903 max mem: 3532 +Test: [320/389] eta: 0:00:13 acc1_pos: 45.399 acc5_pos: 100.000 loss: 0.7030 (0.6179) acc1: 51.8229 (65.2956) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1983 data: 0.1054 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 45.399 acc5_pos: 100.000 loss: 0.6838 (0.6201) acc1: 57.2917 (65.0161) acc5: 100.0000 (100.0000) acc1_neg: 68.229 acc5_neg: 100.000 time: 0.1742 data: 0.0825 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 45.399 acc5_pos: 100.000 loss: 0.4389 (0.6092) acc1: 71.0938 (65.9526) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1698 data: 0.0779 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 45.399 acc5_pos: 100.000 loss: 0.2485 (0.6124) acc1: 99.2188 (65.7726) acc5: 100.0000 (100.0000) acc1_neg: 47.656 acc5_neg: 100.000 time: 0.1977 data: 0.1058 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 45.399 acc5_pos: 100.000 loss: 0.5980 (0.6129) acc1: 64.8438 (65.6957) acc5: 100.0000 (100.0000) acc1_neg: 71.094 acc5_neg: 100.000 time: 0.1701 data: 0.0793 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 45.399 acc5_pos: 100.000 loss: 0.5840 (0.6111) acc1: 66.9271 (65.8601) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1487 data: 0.0575 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 45.399 acc5_pos: 100.000 loss: 0.4719 (0.6084) acc1: 79.9479 (66.2101) acc5: 100.0000 (100.0000) acc1_neg: 57.031 acc5_neg: 100.000 time: 0.1434 data: 0.0526 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 45.399 acc5_pos: 100.000 loss: 0.4463 (0.6046) acc1: 82.8125 (66.5903) acc5: 100.0000 (100.0000) acc1_neg: 93.443 acc5_neg: 100.000 time: 0.1411 data: 0.0383 max mem: 3532 +Test: Total time: 0:01:12 (0.1873 s / it) +* Overall Acc@1 66.590 Acc@5 100.000 + target == real Acc@1 55.786 Acc@5 100.000 + target == fake Acc@1 71.612 Acc@5 100.000 + loss 0.605 + PR-AUC 0.8328 ROC-AUC 0.7030 EER 35.48% (thr=0.5542) +Accuracy of the network on the 149358 test images: 66.6% +[rank0]:[W726 03:15:27.791275318 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:15:48.164000 1174137 torch/distributed/run.py:766] +W0726 03:15:48.164000 1174137 torch/distributed/run.py:766] ***************************************** +W0726 03:15:48.164000 1174137 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:15:48.164000 1174137 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:16:16.458759776 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:16:17.291039859 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:16:17.729078114 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:16:17.879428222 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank4]:[W726 03:16:17.880048928 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:16:17.974294703 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank1]:[W726 03:16:17.974294878 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:16:17.124083072 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n32_seed5_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n32_classseed5_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:37:48 acc1_pos: 16.406 acc5_pos: 100.000 loss: 1.3769 (1.3769) acc1: 16.4062 (16.4062) acc5: 100.0000 (100.0000) time: 6.8137 data: 5.1454 max mem: 3531 +Test: [ 10/333] eta: 0:03:48 acc1_pos: 86.719 acc5_pos: 100.000 loss: 1.0780 (0.8414) acc1: 39.8438 (50.5682) acc5: 100.0000 (100.0000) time: 0.7083 data: 0.4682 max mem: 3531 +Test: [ 20/333] eta: 0:02:11 acc1_pos: 97.135 acc5_pos: 100.000 loss: 0.3193 (0.5799) acc1: 86.7188 (69.4940) acc5: 100.0000 (100.0000) time: 0.0988 data: 0.0005 max mem: 3531 +Test: [ 30/333] eta: 0:01:38 acc1_pos: 63.542 acc5_pos: 100.000 loss: 0.3562 (0.6139) acc1: 81.2500 (66.4819) acc5: 100.0000 (100.0000) time: 0.1149 data: 0.0182 max mem: 3531 +Test: [ 40/333] eta: 0:01:24 acc1_pos: 60.938 acc5_pos: 100.000 loss: 0.6896 (0.6425) acc1: 55.7292 (63.3575) acc5: 100.0000 (100.0000) time: 0.1517 data: 0.0589 max mem: 3531 +Test: [ 50/333] eta: 0:01:14 acc1_pos: 54.948 acc5_pos: 100.000 loss: 0.7224 (0.6587) acc1: 54.9479 (61.3971) acc5: 100.0000 (100.0000) time: 0.1671 data: 0.0760 max mem: 3531 +Test: [ 60/333] eta: 0:01:06 acc1_pos: 41.927 acc5_pos: 100.000 loss: 0.7168 (0.6720) acc1: 54.9479 (59.9129) acc5: 100.0000 (100.0000) time: 0.1542 data: 0.0638 max mem: 3531 +Test: [ 70/333] eta: 0:01:01 acc1_pos: 46.354 acc5_pos: 100.000 loss: 0.6984 (0.6752) acc1: 55.4688 (59.5694) acc5: 100.0000 (100.0000) time: 0.1587 data: 0.0683 max mem: 3531 +Test: [ 80/333] eta: 0:00:56 acc1_pos: 61.198 acc5_pos: 100.000 loss: 0.7055 (0.6818) acc1: 55.9896 (58.8606) acc5: 100.0000 (100.0000) time: 0.1645 data: 0.0732 max mem: 3531 +Test: [ 90/333] eta: 0:00:52 acc1_pos: 68.229 acc5_pos: 100.000 loss: 0.6719 (0.6748) acc1: 58.5938 (59.4780) acc5: 100.0000 (100.0000) time: 0.1542 data: 0.0627 max mem: 3531 +Test: [100/333] eta: 0:00:49 acc1_pos: 69.792 acc5_pos: 100.000 loss: 0.5964 (0.6709) acc1: 65.6250 (59.7901) acc5: 100.0000 (100.0000) time: 0.1675 data: 0.0749 max mem: 3531 +Test: [110/333] eta: 0:00:47 acc1_pos: 74.740 acc5_pos: 100.000 loss: 0.6284 (0.6654) acc1: 61.7188 (60.2243) acc5: 100.0000 (100.0000) time: 0.2005 data: 0.1057 max mem: 3531 +Test: [120/333] eta: 0:00:44 acc1_pos: 69.010 acc5_pos: 100.000 loss: 0.6284 (0.6624) acc1: 61.7188 (60.4081) acc5: 100.0000 (100.0000) time: 0.1966 data: 0.1039 max mem: 3531 +Test: [130/333] eta: 0:00:41 acc1_pos: 83.333 acc5_pos: 100.000 loss: 0.5835 (0.6568) acc1: 66.4062 (60.9693) acc5: 100.0000 (100.0000) time: 0.1641 data: 0.0723 max mem: 3531 +Test: [140/333] eta: 0:00:39 acc1_pos: 79.167 acc5_pos: 100.000 loss: 0.5782 (0.6508) acc1: 69.7917 (61.5304) acc5: 100.0000 (100.0000) time: 0.1699 data: 0.0763 max mem: 3531 +Test: [150/333] eta: 0:00:37 acc1_pos: 70.573 acc5_pos: 100.000 loss: 0.5855 (0.6486) acc1: 67.1875 (61.7032) acc5: 100.0000 (100.0000) time: 0.1899 data: 0.0941 max mem: 3531 +Test: [160/333] eta: 0:00:35 acc1_pos: 49.671 acc5_pos: 100.000 loss: 0.6124 (0.6539) acc1: 64.8438 (61.3128) acc5: 100.0000 (100.0000) acc1_neg: 45.052 acc5_neg: 100.000 time: 0.1999 data: 0.1017 max mem: 3531 +Test: [170/333] eta: 0:00:33 acc1_pos: 49.671 acc5_pos: 100.000 loss: 0.7634 (0.6624) acc1: 52.6042 (60.6786) acc5: 100.0000 (100.0000) acc1_neg: 54.688 acc5_neg: 100.000 time: 0.2000 data: 0.1043 max mem: 3531 +Test: [180/333] eta: 0:00:31 acc1_pos: 49.671 acc5_pos: 100.000 loss: 0.7467 (0.6660) acc1: 53.6458 (60.3922) acc5: 100.0000 (100.0000) acc1_neg: 51.042 acc5_neg: 100.000 time: 0.1949 data: 0.1001 max mem: 3531 +Test: [190/333] eta: 0:00:28 acc1_pos: 49.671 acc5_pos: 100.000 loss: 0.6916 (0.6649) acc1: 57.2917 (60.6389) acc5: 100.0000 (100.0000) acc1_neg: 63.281 acc5_neg: 100.000 time: 0.1960 data: 0.1001 max mem: 3531 +Test: [200/333] eta: 0:00:26 acc1_pos: 49.671 acc5_pos: 100.000 loss: 0.6779 (0.6656) acc1: 62.5000 (60.7574) acc5: 100.0000 (100.0000) acc1_neg: 60.938 acc5_neg: 100.000 time: 0.1961 data: 0.1025 max mem: 3531 +Test: [210/333] eta: 0:00:24 acc1_pos: 49.671 acc5_pos: 100.000 loss: 0.7051 (0.6690) acc1: 60.1562 (60.5648) acc5: 100.0000 (100.0000) acc1_neg: 54.427 acc5_neg: 100.000 time: 0.1887 data: 0.0956 max mem: 3532 +Test: [220/333] eta: 0:00:22 acc1_pos: 49.671 acc5_pos: 100.000 loss: 0.7194 (0.6748) acc1: 56.5104 (60.1374) acc5: 100.0000 (100.0000) acc1_neg: 63.281 acc5_neg: 100.000 time: 0.1737 data: 0.0779 max mem: 3532 +Test: [230/333] eta: 0:00:20 acc1_pos: 49.671 acc5_pos: 100.000 loss: 0.5994 (0.6694) acc1: 65.8854 (60.8338) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.1848 data: 0.0892 max mem: 3532 +Test: [240/333] eta: 0:00:18 acc1_pos: 49.671 acc5_pos: 100.000 loss: 0.5838 (0.6691) acc1: 68.4896 (60.8705) acc5: 100.0000 (100.0000) acc1_neg: 60.156 acc5_neg: 100.000 time: 0.1849 data: 0.0872 max mem: 3532 +Test: [250/333] eta: 0:00:16 acc1_pos: 49.671 acc5_pos: 100.000 loss: 0.6762 (0.6698) acc1: 61.7188 (60.8794) acc5: 100.0000 (100.0000) acc1_neg: 54.948 acc5_neg: 100.000 time: 0.1699 data: 0.0706 max mem: 3532 +Test: [260/333] eta: 0:00:14 acc1_pos: 49.671 acc5_pos: 100.000 loss: 0.6825 (0.6707) acc1: 60.9375 (60.8906) acc5: 100.0000 (100.0000) acc1_neg: 59.635 acc5_neg: 100.000 time: 0.1999 data: 0.1032 max mem: 3532 +Test: [270/333] eta: 0:00:12 acc1_pos: 49.671 acc5_pos: 100.000 loss: 0.6669 (0.6700) acc1: 62.2396 (61.0086) acc5: 100.0000 (100.0000) acc1_neg: 61.979 acc5_neg: 100.000 time: 0.1909 data: 0.0943 max mem: 3532 +Test: [280/333] eta: 0:00:10 acc1_pos: 49.671 acc5_pos: 100.000 loss: 0.6669 (0.6730) acc1: 61.9792 (60.7142) acc5: 100.0000 (100.0000) acc1_neg: 54.167 acc5_neg: 100.000 time: 0.1461 data: 0.0508 max mem: 3532 +Test: [290/333] eta: 0:00:08 acc1_pos: 49.671 acc5_pos: 100.000 loss: 0.8240 (0.6795) acc1: 45.3125 (60.1482) acc5: 100.0000 (100.0000) acc1_neg: 46.615 acc5_neg: 100.000 time: 0.2039 data: 0.1094 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 49.671 acc5_pos: 100.000 loss: 0.8597 (0.6858) acc1: 44.7917 (59.6882) acc5: 100.0000 (100.0000) acc1_neg: 45.312 acc5_neg: 100.000 time: 0.2391 data: 0.1440 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 49.671 acc5_pos: 100.000 loss: 0.7862 (0.6837) acc1: 51.0417 (59.8992) acc5: 100.0000 (100.0000) acc1_neg: 64.062 acc5_neg: 100.000 time: 0.2022 data: 0.1055 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 49.671 acc5_pos: 100.000 loss: 0.6430 (0.6832) acc1: 64.8438 (59.9851) acc5: 100.0000 (100.0000) acc1_neg: 51.823 acc5_neg: 100.000 time: 0.1540 data: 0.0593 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 49.671 acc5_pos: 100.000 loss: 0.8165 (0.6887) acc1: 50.5208 (59.5859) acc5: 100.0000 (100.0000) acc1_neg: 47.656 acc5_neg: 100.000 time: 0.1268 data: 0.0351 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 49.671 acc5_pos: 100.000 loss: 0.8436 (0.6897) acc1: 48.4507 (59.5160) acc5: 100.0000 (100.0000) acc1_neg: 48.451 acc5_neg: 100.000 time: 0.1296 data: 0.0351 max mem: 3532 +Test: Total time: 0:01:04 (0.1938 s / it) +* Overall Acc@1 59.516 Acc@5 100.000 + target == real Acc@1 61.889 Acc@5 100.000 + target == fake Acc@1 57.404 Acc@5 100.000 + loss 0.690 + PR-AUC 0.6468 ROC-AUC 0.6360 EER 40.34% (thr=0.4858) +Accuracy of the network on the 127843 test images: 59.5% +[rank0]:[W726 03:17:35.607975305 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:17:55.793000 1174696 torch/distributed/run.py:766] +W0726 03:17:55.793000 1174696 torch/distributed/run.py:766] ***************************************** +W0726 03:17:55.793000 1174696 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:17:55.793000 1174696 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:18:25.723584210 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:18:25.109299214 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:18:26.490382354 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:18:26.618032553 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:18:26.857643788 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:18:26.062436652 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:18:26.079854807 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:18:26.114870515 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n32_seed5_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n32_classseed5_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:23:59 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.2097 (0.2097) acc1: 95.3125 (95.3125) acc5: 100.0000 (100.0000) time: 6.1243 data: 4.6755 max mem: 3531 +Test: [ 10/235] eta: 0:02:26 acc1_pos: 94.271 acc5_pos: 100.000 loss: 0.2153 (0.2186) acc1: 95.3125 (94.9337) acc5: 100.0000 (100.0000) time: 0.6496 data: 0.4358 max mem: 3531 +Test: [ 20/235] eta: 0:01:34 acc1_pos: 95.573 acc5_pos: 100.000 loss: 0.2104 (0.2142) acc1: 95.5729 (95.3125) acc5: 100.0000 (100.0000) time: 0.1577 data: 0.0664 max mem: 3531 +Test: [ 30/235] eta: 0:01:26 acc1_pos: 96.354 acc5_pos: 100.000 loss: 0.2062 (0.2126) acc1: 95.5729 (95.4049) acc5: 100.0000 (100.0000) time: 0.2935 data: 0.2018 max mem: 3531 +Test: [ 40/235] eta: 0:01:12 acc1_pos: 94.792 acc5_pos: 100.000 loss: 0.2099 (0.2128) acc1: 95.5729 (95.4141) acc5: 100.0000 (100.0000) time: 0.3011 data: 0.2087 max mem: 3531 +Test: [ 50/235] eta: 0:00:59 acc1_pos: 94.010 acc5_pos: 100.000 loss: 0.2142 (0.2140) acc1: 95.3125 (95.3534) acc5: 100.0000 (100.0000) time: 0.1655 data: 0.0730 max mem: 3531 +Test: [ 60/235] eta: 0:00:54 acc1_pos: 94.271 acc5_pos: 100.000 loss: 0.2185 (0.2148) acc1: 95.3125 (95.3253) acc5: 100.0000 (100.0000) time: 0.1847 data: 0.0924 max mem: 3531 +Test: [ 70/235] eta: 0:00:48 acc1_pos: 98.462 acc5_pos: 100.000 loss: 0.2321 (0.3690) acc1: 94.0104 (86.0072) acc5: 100.0000 (100.0000) acc1_neg: 19.271 acc5_neg: 100.000 time: 0.2299 data: 0.1363 max mem: 3531 +Test: [ 80/235] eta: 0:00:43 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4426 (0.5028) acc1: 20.8333 (77.9450) acc5: 100.0000 (100.0000) acc1_neg: 20.833 acc5_neg: 100.000 time: 0.1764 data: 0.0825 max mem: 3531 +Test: [ 90/235] eta: 0:00:39 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4566 (0.6072) acc1: 20.5729 (71.6432) acc5: 100.0000 (100.0000) acc1_neg: 21.875 acc5_neg: 100.000 time: 0.1849 data: 0.0899 max mem: 3531 +Test: [100/235] eta: 0:00:35 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4567 (0.6885) acc1: 20.8333 (66.7053) acc5: 100.0000 (100.0000) acc1_neg: 18.229 acc5_neg: 100.000 time: 0.2024 data: 0.1080 max mem: 3531 +Test: [110/235] eta: 0:00:31 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4483 (0.7563) acc1: 20.8333 (62.5493) acc5: 100.0000 (100.0000) acc1_neg: 21.094 acc5_neg: 100.000 time: 0.1870 data: 0.0954 max mem: 3531 +Test: [120/235] eta: 0:00:29 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4962 (0.8272) acc1: 17.9688 (58.5141) acc5: 100.0000 (100.0000) acc1_neg: 14.583 acc5_neg: 100.000 time: 0.2073 data: 0.1149 max mem: 3531 +Test: [130/235] eta: 0:00:26 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.6156 (0.8872) acc1: 13.5417 (55.0334) acc5: 100.0000 (100.0000) acc1_neg: 13.802 acc5_neg: 100.000 time: 0.2193 data: 0.1269 max mem: 3531 +Test: [140/235] eta: 0:00:23 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.6261 (0.9420) acc1: 12.2396 (51.9891) acc5: 100.0000 (100.0000) acc1_neg: 10.417 acc5_neg: 100.000 time: 0.2099 data: 0.1136 max mem: 3531 +Test: [150/235] eta: 0:00:20 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.6407 (0.9868) acc1: 12.2396 (49.4240) acc5: 100.0000 (100.0000) acc1_neg: 13.802 acc5_neg: 100.000 time: 0.2035 data: 0.1043 max mem: 3531 +Test: [160/235] eta: 0:00:18 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.6086 (1.0265) acc1: 13.2812 (47.1678) acc5: 100.0000 (100.0000) acc1_neg: 14.062 acc5_neg: 100.000 time: 0.1899 data: 0.0953 max mem: 3531 +Test: [170/235] eta: 0:00:15 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.5675 (1.0336) acc1: 14.3229 (46.3435) acc5: 100.0000 (100.0000) acc1_neg: 30.729 acc5_neg: 100.000 time: 0.1810 data: 0.0882 max mem: 3531 +Test: [180/235] eta: 0:00:12 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.1073 (1.0367) acc1: 35.4167 (45.8276) acc5: 100.0000 (100.0000) acc1_neg: 37.240 acc5_neg: 100.000 time: 0.1817 data: 0.0893 max mem: 3531 +Test: [190/235] eta: 0:00:10 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.0838 (1.0395) acc1: 36.9792 (45.3275) acc5: 100.0000 (100.0000) acc1_neg: 38.281 acc5_neg: 100.000 time: 0.1991 data: 0.1066 max mem: 3531 +Test: [200/235] eta: 0:00:08 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.0838 (1.0428) acc1: 34.6354 (44.7826) acc5: 100.0000 (100.0000) acc1_neg: 33.594 acc5_neg: 100.000 time: 0.1911 data: 0.0993 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.0823 (1.0443) acc1: 34.6354 (44.4103) acc5: 100.0000 (100.0000) acc1_neg: 38.542 acc5_neg: 100.000 time: 0.1516 data: 0.0609 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.0759 (1.0462) acc1: 37.2396 (44.0894) acc5: 100.0000 (100.0000) acc1_neg: 34.115 acc5_neg: 100.000 time: 0.1332 data: 0.0426 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.0759 (1.0472) acc1: 37.2396 (43.8210) acc5: 100.0000 (100.0000) acc1_neg: 39.583 acc5_neg: 100.000 time: 0.1304 data: 0.0402 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.0816 (1.0491) acc1: 36.4583 (43.7221) acc5: 100.0000 (100.0000) acc1_neg: 34.783 acc5_neg: 100.000 time: 0.1243 data: 0.0324 max mem: 3532 +Test: Total time: 0:00:50 (0.2156 s / it) +* Overall Acc@1 43.722 Acc@5 100.000 + target == real Acc@1 95.317 Acc@5 100.000 + target == fake Acc@1 25.061 Acc@5 100.000 + loss 1.049 + PR-AUC 0.8803 ROC-AUC 0.7301 EER 33.12% (thr=0.1802) +Accuracy of the network on the 89879 test images: 43.7% +[rank0]:[W726 03:19:26.854469021 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:19:46.052000 1175254 torch/distributed/run.py:766] +W0726 03:19:46.052000 1175254 torch/distributed/run.py:766] ***************************************** +W0726 03:19:46.052000 1175254 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:19:46.052000 1175254 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:20:16.546362237 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:20:16.702423666 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:20:16.070341381 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:20:17.235113870 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:20:17.520238683 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:20:17.557221876 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:20:17.584569364 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:20:17.677558676 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n32_seed5_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n32_classseed5_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:25:06 acc1_pos: 48.438 acc5_pos: 100.000 loss: 0.7539 (0.7539) acc1: 48.4375 (48.4375) acc5: 100.0000 (100.0000) time: 5.5994 data: 3.8569 max mem: 3531 +Test: [ 10/269] eta: 0:02:33 acc1_pos: 40.625 acc5_pos: 100.000 loss: 0.7421 (0.6982) acc1: 48.9583 (53.7169) acc5: 100.0000 (100.0000) time: 0.5914 data: 0.3510 max mem: 3531 +Test: [ 20/269] eta: 0:01:32 acc1_pos: 71.354 acc5_pos: 100.000 loss: 0.6175 (0.6604) acc1: 63.0208 (59.0154) acc5: 100.0000 (100.0000) time: 0.1117 data: 0.0196 max mem: 3531 +Test: [ 30/269] eta: 0:01:22 acc1_pos: 40.885 acc5_pos: 100.000 loss: 0.5515 (0.6338) acc1: 72.6562 (62.2984) acc5: 100.0000 (100.0000) time: 0.2114 data: 0.1173 max mem: 3531 +Test: [ 40/269] eta: 0:01:18 acc1_pos: 47.414 acc5_pos: 100.000 loss: 0.5515 (0.6216) acc1: 73.9583 (64.3039) acc5: 100.0000 (100.0000) acc1_neg: 66.406 acc5_neg: 100.000 time: 0.3123 data: 0.2191 max mem: 3531 +Test: [ 50/269] eta: 0:01:06 acc1_pos: 47.414 acc5_pos: 100.000 loss: 0.5237 (0.6009) acc1: 79.9479 (67.6573) acc5: 100.0000 (100.0000) acc1_neg: 54.688 acc5_neg: 100.000 time: 0.2350 data: 0.1432 max mem: 3531 +Test: [ 60/269] eta: 0:01:01 acc1_pos: 47.414 acc5_pos: 100.000 loss: 0.5255 (0.5964) acc1: 77.6042 (68.5920) acc5: 100.0000 (100.0000) acc1_neg: 61.458 acc5_neg: 100.000 time: 0.1893 data: 0.0944 max mem: 3531 +Test: [ 70/269] eta: 0:00:56 acc1_pos: 47.414 acc5_pos: 100.000 loss: 0.5560 (0.5900) acc1: 73.6979 (69.6009) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.2448 data: 0.1443 max mem: 3531 +Test: [ 80/269] eta: 0:00:50 acc1_pos: 47.414 acc5_pos: 100.000 loss: 0.5076 (0.5797) acc1: 82.0312 (71.1452) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.1994 data: 0.1016 max mem: 3531 +Test: [ 90/269] eta: 0:00:47 acc1_pos: 47.414 acc5_pos: 100.000 loss: 0.5784 (0.5867) acc1: 71.3542 (70.0693) acc5: 100.0000 (100.0000) acc1_neg: 71.354 acc5_neg: 100.000 time: 0.1893 data: 0.0918 max mem: 3531 +Test: [100/269] eta: 0:00:43 acc1_pos: 47.414 acc5_pos: 100.000 loss: 0.6384 (0.5904) acc1: 64.5833 (69.7092) acc5: 100.0000 (100.0000) acc1_neg: 67.969 acc5_neg: 100.000 time: 0.2130 data: 0.1146 max mem: 3531 +Test: [110/269] eta: 0:00:39 acc1_pos: 47.414 acc5_pos: 100.000 loss: 0.6211 (0.5936) acc1: 64.0625 (69.3131) acc5: 100.0000 (100.0000) acc1_neg: 60.938 acc5_neg: 100.000 time: 0.1900 data: 0.0956 max mem: 3531 +Test: [120/269] eta: 0:00:36 acc1_pos: 47.414 acc5_pos: 100.000 loss: 0.6330 (0.5962) acc1: 61.1979 (68.8770) acc5: 100.0000 (100.0000) acc1_neg: 48.438 acc5_neg: 100.000 time: 0.1863 data: 0.0915 max mem: 3531 +Test: [130/269] eta: 0:00:33 acc1_pos: 47.414 acc5_pos: 100.000 loss: 0.6453 (0.6021) acc1: 61.4583 (68.1556) acc5: 100.0000 (100.0000) acc1_neg: 45.833 acc5_neg: 100.000 time: 0.1767 data: 0.0803 max mem: 3531 +Test: [140/269] eta: 0:00:30 acc1_pos: 47.414 acc5_pos: 100.000 loss: 0.6955 (0.6147) acc1: 53.9062 (66.6537) acc5: 100.0000 (100.0000) acc1_neg: 44.271 acc5_neg: 100.000 time: 0.1867 data: 0.0919 max mem: 3531 +Test: [150/269] eta: 0:00:27 acc1_pos: 47.414 acc5_pos: 100.000 loss: 0.7406 (0.6227) acc1: 47.6562 (65.6353) acc5: 100.0000 (100.0000) acc1_neg: 58.333 acc5_neg: 100.000 time: 0.1890 data: 0.0967 max mem: 3531 +Test: [160/269] eta: 0:00:25 acc1_pos: 47.414 acc5_pos: 100.000 loss: 0.6586 (0.6208) acc1: 58.3333 (65.9695) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1728 data: 0.0791 max mem: 3531 +Test: [170/269] eta: 0:00:22 acc1_pos: 47.414 acc5_pos: 100.000 loss: 0.5228 (0.6146) acc1: 80.9896 (66.7870) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1771 data: 0.0832 max mem: 3531 +Test: [180/269] eta: 0:00:19 acc1_pos: 47.414 acc5_pos: 100.000 loss: 0.4755 (0.6073) acc1: 83.0729 (67.7472) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.1768 data: 0.0853 max mem: 3531 +Test: [190/269] eta: 0:00:17 acc1_pos: 47.414 acc5_pos: 100.000 loss: 0.4967 (0.6071) acc1: 78.6458 (67.7247) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.1746 data: 0.0841 max mem: 3531 +Test: [200/269] eta: 0:00:15 acc1_pos: 47.414 acc5_pos: 100.000 loss: 0.6233 (0.6092) acc1: 64.8438 (67.4661) acc5: 100.0000 (100.0000) acc1_neg: 66.406 acc5_neg: 100.000 time: 0.1927 data: 0.1023 max mem: 3531 +Test: [210/269] eta: 0:00:12 acc1_pos: 47.414 acc5_pos: 100.000 loss: 0.6598 (0.6153) acc1: 53.6458 (66.5815) acc5: 100.0000 (100.0000) acc1_neg: 51.823 acc5_neg: 100.000 time: 0.1980 data: 0.1069 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 47.414 acc5_pos: 100.000 loss: 0.6947 (0.6199) acc1: 49.2188 (65.9809) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.1840 data: 0.0926 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 47.414 acc5_pos: 100.000 loss: 0.5708 (0.6176) acc1: 69.0104 (66.2721) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.1694 data: 0.0786 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 47.414 acc5_pos: 100.000 loss: 0.5660 (0.6188) acc1: 73.1771 (66.2063) acc5: 100.0000 (100.0000) acc1_neg: 59.896 acc5_neg: 100.000 time: 0.1480 data: 0.0578 max mem: 3532 +Test: [250/269] eta: 0:00:03 acc1_pos: 47.414 acc5_pos: 100.000 loss: 0.5733 (0.6235) acc1: 59.8958 (65.4683) acc5: 100.0000 (100.0000) acc1_neg: 17.188 acc5_neg: 100.000 time: 0.1377 data: 0.0475 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 47.414 acc5_pos: 100.000 loss: 0.7432 (0.6287) acc1: 41.9271 (64.9186) acc5: 100.0000 (100.0000) acc1_neg: 77.344 acc5_neg: 100.000 time: 0.1237 data: 0.0331 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 47.414 acc5_pos: 100.000 loss: 0.6136 (0.6268) acc1: 65.8854 (65.2077) acc5: 100.0000 (100.0000) acc1_neg: 84.252 acc5_neg: 100.000 time: 0.1244 data: 0.0303 max mem: 3532 +Test: Total time: 0:00:54 (0.2035 s / it) +* Overall Acc@1 65.208 Acc@5 100.000 + target == real Acc@1 61.391 Acc@5 100.000 + target == fake Acc@1 65.729 Acc@5 100.000 + loss 0.627 + PR-AUC 0.9383 ROC-AUC 0.6935 EER 36.45% (thr=0.5093) +Accuracy of the network on the 103293 test images: 65.2% +[rank0]:[W726 03:21:22.033622482 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T03:21:30Z] eval_done tag=fixed500k_real_n32_seed5_resampled +[2026-07-26T03:21:31Z] eval_start tag=fixed500k_fake_n64_seed5_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed5_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T03:21:31Z +hostname=n116-040-213 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed5_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n64_seed5_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 03:21:44.241000 1175831 torch/distributed/run.py:766] +W0726 03:21:44.241000 1175831 torch/distributed/run.py:766] ***************************************** +W0726 03:21:44.241000 1175831 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:21:44.241000 1175831 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:22:14.537647840 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 0): env:// +[rank5]:[W726 03:22:14.558973243 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:22:14.858958481 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:22:14.050922169 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:22:14.148544192 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:22:15.324956139 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:22:15.375290900 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W726 03:22:15.375328658 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n64_seed5_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed5_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:21:44 acc1_pos: 60.417 acc5_pos: 100.000 loss: 0.6075 (0.6075) acc1: 60.4167 (60.4167) acc5: 100.0000 (100.0000) time: 5.5769 data: 4.0985 max mem: 3531 +Test: [ 10/234] eta: 0:02:12 acc1_pos: 55.729 acc5_pos: 100.000 loss: 0.6075 (0.6313) acc1: 63.0208 (61.9555) acc5: 100.0000 (100.0000) time: 0.5919 data: 0.3730 max mem: 3531 +Test: [ 20/234] eta: 0:01:20 acc1_pos: 96.354 acc5_pos: 100.000 loss: 0.7024 (0.6344) acc1: 55.7292 (58.5690) acc5: 100.0000 (100.0000) time: 0.1165 data: 0.0229 max mem: 3531 +Test: [ 30/234] eta: 0:01:06 acc1_pos: 88.802 acc5_pos: 100.000 loss: 0.2239 (0.5020) acc1: 96.0938 (70.6065) acc5: 100.0000 (100.0000) time: 0.1798 data: 0.0829 max mem: 3531 +Test: [ 40/234] eta: 0:00:57 acc1_pos: 98.958 acc5_pos: 100.000 loss: 0.2155 (0.4556) acc1: 96.3542 (75.0762) acc5: 100.0000 (100.0000) time: 0.2151 data: 0.1181 max mem: 3531 +Test: [ 50/234] eta: 0:00:50 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1476 (0.3942) acc1: 97.6562 (79.6058) acc5: 100.0000 (100.0000) time: 0.1950 data: 0.1019 max mem: 3531 +Test: [ 60/234] eta: 0:00:45 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.1173 (0.3451) acc1: 99.2188 (82.9278) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1847 data: 0.0927 max mem: 3531 +Test: [ 70/234] eta: 0:00:42 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.0987 (0.3725) acc1: 99.7396 (81.3234) acc5: 100.0000 (100.0000) acc1_neg: 56.771 acc5_neg: 100.000 time: 0.2114 data: 0.1201 max mem: 3531 +Test: [ 80/234] eta: 0:00:38 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.7149 (0.3954) acc1: 59.3750 (80.2437) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2208 data: 0.1305 max mem: 3531 +Test: [ 90/234] eta: 0:00:35 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.2592 (0.3738) acc1: 89.5833 (81.7565) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.2033 data: 0.1126 max mem: 3531 +Test: [100/234] eta: 0:00:32 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.2779 (0.3771) acc1: 87.5000 (81.7373) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1964 data: 0.1047 max mem: 3531 +Test: [110/234] eta: 0:00:29 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.4519 (0.3811) acc1: 77.6042 (81.6629) acc5: 100.0000 (100.0000) acc1_neg: 65.625 acc5_neg: 100.000 time: 0.1899 data: 0.0963 max mem: 3531 +Test: [120/234] eta: 0:00:26 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.4542 (0.3957) acc1: 76.5625 (81.0003) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.1844 data: 0.0896 max mem: 3531 +Test: [130/234] eta: 0:00:23 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.4542 (0.3997) acc1: 82.8125 (81.2917) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.1948 data: 0.0978 max mem: 3531 +Test: [140/234] eta: 0:00:21 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.4610 (0.4212) acc1: 81.5104 (80.2231) acc5: 100.0000 (100.0000) acc1_neg: 40.625 acc5_neg: 100.000 time: 0.2030 data: 0.1074 max mem: 3531 +Test: [150/234] eta: 0:00:18 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.4266 (0.4260) acc1: 78.9062 (79.9634) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.1849 data: 0.0919 max mem: 3531 +Test: [160/234] eta: 0:00:16 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.4153 (0.4281) acc1: 80.2083 (80.0789) acc5: 100.0000 (100.0000) acc1_neg: 72.135 acc5_neg: 100.000 time: 0.1761 data: 0.0817 max mem: 3531 +Test: [170/234] eta: 0:00:14 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.5538 (0.4362) acc1: 72.1354 (79.7347) acc5: 100.0000 (100.0000) acc1_neg: 77.344 acc5_neg: 100.000 time: 0.2001 data: 0.1062 max mem: 3531 +Test: [180/234] eta: 0:00:11 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.5780 (0.4500) acc1: 70.8333 (78.9293) acc5: 100.0000 (100.0000) acc1_neg: 36.979 acc5_neg: 100.000 time: 0.2091 data: 0.1138 max mem: 3531 +Test: [190/234] eta: 0:00:09 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.7893 (0.4684) acc1: 55.9896 (77.6587) acc5: 100.0000 (100.0000) acc1_neg: 57.292 acc5_neg: 100.000 time: 0.2097 data: 0.1137 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.7669 (0.4629) acc1: 56.7708 (77.9475) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1955 data: 0.1030 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.1171 (0.4494) acc1: 99.4792 (78.8544) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1714 data: 0.0809 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.2816 (0.4435) acc1: 91.6667 (79.3175) acc5: 100.0000 (100.0000) acc1_neg: 72.656 acc5_neg: 100.000 time: 0.1763 data: 0.0858 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.5387 (0.4725) acc1: 31.7708 (77.0664) acc5: 100.0000 (100.0000) acc1_neg: 26.302 acc5_neg: 100.000 time: 0.1559 data: 0.0658 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 98.621 acc5_pos: 100.000 loss: 1.0790 (0.4803) acc1: 28.7938 (76.5137) acc5: 100.0000 (100.0000) acc1_neg: 28.794 acc5_neg: 100.000 time: 0.1371 data: 0.0459 max mem: 3532 +Test: Total time: 0:00:48 (0.2089 s / it) +* Overall Acc@1 76.514 Acc@5 100.000 + target == real Acc@1 79.746 Acc@5 100.000 + target == fake Acc@1 75.603 Acc@5 100.000 + loss 0.480 + PR-AUC 0.9620 ROC-AUC 0.8727 EER 22.65% (thr=0.4751) +Accuracy of the network on the 89729 test images: 76.5% +[rank0]:[W726 03:23:13.152450012 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:23:33.122000 1176378 torch/distributed/run.py:766] +W0726 03:23:33.122000 1176378 torch/distributed/run.py:766] ***************************************** +W0726 03:23:33.122000 1176378 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:23:33.122000 1176378 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 03:24:02.737659410 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:24:02.814869844 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:24:02.156239192 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:24:03.197213438 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:24:03.494394243 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:24:03.547322758 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:24:03.601136993 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:24:03.639742281 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n64_seed5_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed5_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 3:59:23 acc1_pos: 37.240 acc5_pos: 100.000 loss: 0.8449 (0.8449) acc1: 37.2396 (37.2396) acc5: 100.0000 (100.0000) time: 6.0173 data: 3.7945 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/2387] eta: 0:24:56 acc1_pos: 13.021 acc5_pos: 100.000 loss: 0.9412 (0.9537) acc1: 28.9062 (27.6278) acc5: 100.0000 (100.0000) time: 0.6297 data: 0.3452 max mem: 3531 +Test: [ 20/2387] eta: 0:14:42 acc1_pos: 24.479 acc5_pos: 100.000 loss: 0.9269 (0.9599) acc1: 24.7396 (27.0461) acc5: 100.0000 (100.0000) time: 0.0906 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:12:38 acc1_pos: 5.208 acc5_pos: 100.000 loss: 0.9862 (0.9714) acc1: 20.3125 (25.0672) acc5: 100.0000 (100.0000) time: 0.1524 data: 0.0619 max mem: 3531 +Test: [ 40/2387] eta: 0:13:08 acc1_pos: 17.708 acc5_pos: 100.000 loss: 0.9944 (0.9806) acc1: 17.4479 (22.5673) acc5: 100.0000 (100.0000) time: 0.2970 data: 0.2055 max mem: 3531 +Test: [ 50/2387] eta: 0:13:09 acc1_pos: 39.583 acc5_pos: 100.000 loss: 0.9506 (0.9691) acc1: 21.8750 (24.1677) acc5: 100.0000 (100.0000) time: 0.3627 data: 0.2710 max mem: 3531 +Test: [ 60/2387] eta: 0:13:03 acc1_pos: 41.927 acc5_pos: 100.000 loss: 0.9413 (0.9675) acc1: 27.0833 (24.6030) acc5: 100.0000 (100.0000) time: 0.3379 data: 0.2454 max mem: 3531 +Test: [ 70/2387] eta: 0:12:58 acc1_pos: 28.125 acc5_pos: 100.000 loss: 1.0069 (0.9796) acc1: 21.6146 (23.7383) acc5: 100.0000 (100.0000) time: 0.3318 data: 0.2373 max mem: 3531 +Test: [ 80/2387] eta: 0:13:05 acc1_pos: 45.573 acc5_pos: 100.000 loss: 0.9703 (0.9683) acc1: 25.7812 (24.5981) acc5: 100.0000 (100.0000) time: 0.3534 data: 0.2592 max mem: 3531 +Test: [ 90/2387] eta: 0:12:15 acc1_pos: 40.625 acc5_pos: 100.000 loss: 0.8250 (0.9490) acc1: 38.8021 (26.3765) acc5: 100.0000 (100.0000) time: 0.2636 data: 0.1720 max mem: 3531 +Test: [ 100/2387] eta: 0:12:07 acc1_pos: 44.271 acc5_pos: 100.000 loss: 0.7993 (0.9416) acc1: 39.5833 (27.0421) acc5: 100.0000 (100.0000) time: 0.2262 data: 0.1319 max mem: 3531 +Test: [ 110/2387] eta: 0:11:41 acc1_pos: 38.802 acc5_pos: 100.000 loss: 0.7935 (0.9269) acc1: 38.8021 (28.3549) acc5: 100.0000 (100.0000) time: 0.2541 data: 0.1536 max mem: 3531 +Test: [ 120/2387] eta: 0:11:26 acc1_pos: 28.646 acc5_pos: 100.000 loss: 0.7981 (0.9180) acc1: 40.8854 (29.2592) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1254 max mem: 3531 +Test: [ 130/2387] eta: 0:11:14 acc1_pos: 47.396 acc5_pos: 100.000 loss: 0.8064 (0.9115) acc1: 38.2812 (29.7591) acc5: 100.0000 (100.0000) time: 0.2467 data: 0.1493 max mem: 3531 +Test: [ 140/2387] eta: 0:11:01 acc1_pos: 43.229 acc5_pos: 100.000 loss: 0.7970 (0.9053) acc1: 36.9792 (30.3044) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1491 max mem: 3531 +Test: [ 150/2387] eta: 0:10:50 acc1_pos: 16.927 acc5_pos: 100.000 loss: 0.7917 (0.8974) acc1: 40.8854 (30.9344) acc5: 100.0000 (100.0000) time: 0.2364 data: 0.1419 max mem: 3531 +Test: [ 160/2387] eta: 0:10:30 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.7528 (0.8815) acc1: 41.4062 (32.5537) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2024 data: 0.1086 max mem: 3531 +Test: [ 170/2387] eta: 0:10:17 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2651 (0.8449) acc1: 96.0938 (36.3472) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1886 data: 0.0970 max mem: 3531 +Test: [ 180/2387] eta: 0:10:03 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2552 (0.8122) acc1: 97.3958 (39.7258) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1974 data: 0.1054 max mem: 3531 +Test: [ 190/2387] eta: 0:09:51 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2512 (0.7828) acc1: 97.3958 (42.7533) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1906 data: 0.0964 max mem: 3531 +Test: [ 200/2387] eta: 0:09:46 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2490 (0.7564) acc1: 97.6562 (45.4835) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2198 data: 0.1268 max mem: 3531 +Test: [ 210/2387] eta: 0:09:34 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2516 (0.7328) acc1: 97.3958 (47.9401) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2126 data: 0.1181 max mem: 3532 +Test: [ 220/2387] eta: 0:09:22 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2581 (0.7113) acc1: 97.3958 (50.1791) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1749 data: 0.0750 max mem: 3532 +Test: [ 230/2387] eta: 0:09:17 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2579 (0.6915) acc1: 97.3958 (52.2254) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2010 data: 0.1019 max mem: 3532 +Test: [ 240/2387] eta: 0:09:13 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2612 (0.6742) acc1: 97.1354 (54.0845) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2355 data: 0.1406 max mem: 3532 +Test: [ 250/2387] eta: 0:09:00 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2448 (0.6527) acc1: 97.9167 (55.8983) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1856 data: 0.0903 max mem: 3532 +Test: [ 260/2387] eta: 0:08:46 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1098 (0.6318) acc1: 100.0000 (57.5870) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1243 data: 0.0275 max mem: 3532 +Test: [ 270/2387] eta: 0:08:40 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1016 (0.6123) acc1: 100.0000 (59.1492) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1576 data: 0.0628 max mem: 3532 +Test: [ 280/2387] eta: 0:08:32 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1053 (0.5945) acc1: 100.0000 (60.5983) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0938 max mem: 3532 +Test: [ 290/2387] eta: 0:08:25 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1109 (0.5780) acc1: 100.0000 (61.9523) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1770 data: 0.0797 max mem: 3532 +Test: [ 300/2387] eta: 0:08:19 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1140 (0.5626) acc1: 100.0000 (63.2155) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1810 data: 0.0872 max mem: 3532 +Test: [ 310/2387] eta: 0:08:14 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1192 (0.5544) acc1: 100.0000 (64.1161) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.1956 data: 0.1016 max mem: 3532 +Test: [ 320/2387] eta: 0:08:10 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.5336 (0.5540) acc1: 80.9896 (64.6215) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.2042 data: 0.1089 max mem: 3532 +Test: [ 330/2387] eta: 0:08:05 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.5422 (0.5535) acc1: 80.9896 (65.1207) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.2030 data: 0.1094 max mem: 3532 +Test: [ 340/2387] eta: 0:08:01 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.5370 (0.5531) acc1: 81.2500 (65.5914) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.2017 data: 0.1087 max mem: 3532 +Test: [ 350/2387] eta: 0:07:55 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.5299 (0.5454) acc1: 83.0729 (66.5034) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1945 data: 0.1014 max mem: 3532 +Test: [ 360/2387] eta: 0:07:51 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2381 (0.5364) acc1: 99.4792 (67.4003) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1927 data: 0.0988 max mem: 3532 +Test: [ 370/2387] eta: 0:07:46 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2185 (0.5279) acc1: 98.6979 (68.2411) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1893 data: 0.0938 max mem: 3532 +Test: [ 380/2387] eta: 0:07:39 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2185 (0.5198) acc1: 98.6979 (69.0419) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1648 data: 0.0688 max mem: 3532 +Test: [ 390/2387] eta: 0:07:33 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2183 (0.5120) acc1: 98.9583 (69.8063) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1549 data: 0.0587 max mem: 3532 +Test: [ 400/2387] eta: 0:07:27 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2183 (0.5047) acc1: 98.9583 (70.5294) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1549 data: 0.0590 max mem: 3532 +Test: [ 410/2387] eta: 0:07:23 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2155 (0.4977) acc1: 98.9583 (71.2230) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1649 data: 0.0691 max mem: 3532 +Test: [ 420/2387] eta: 0:07:20 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2143 (0.4910) acc1: 98.9583 (71.8812) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1899 data: 0.0926 max mem: 3532 +Test: [ 430/2387] eta: 0:07:16 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2210 (0.4852) acc1: 98.4375 (72.4895) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1949 data: 0.0964 max mem: 3532 +Test: [ 440/2387] eta: 0:07:12 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2269 (0.4779) acc1: 98.6979 (73.0991) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0865 max mem: 3533 +Test: [ 450/2387] eta: 0:07:08 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1096 (0.4698) acc1: 100.0000 (73.6956) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1795 data: 0.0835 max mem: 3533 +Test: [ 460/2387] eta: 0:07:06 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1094 (0.4619) acc1: 100.0000 (74.2662) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2044 data: 0.1075 max mem: 3533 +Test: [ 470/2387] eta: 0:07:02 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1094 (0.4545) acc1: 100.0000 (74.8126) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2049 data: 0.1071 max mem: 3533 +Test: [ 480/2387] eta: 0:06:57 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1099 (0.4473) acc1: 100.0000 (75.3362) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0733 max mem: 3533 +Test: [ 490/2387] eta: 0:06:53 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1108 (0.4404) acc1: 100.0000 (75.8385) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1649 data: 0.0666 max mem: 3533 +Test: [ 500/2387] eta: 0:06:48 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1103 (0.4339) acc1: 100.0000 (76.3208) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1599 data: 0.0636 max mem: 3533 +Test: [ 510/2387] eta: 0:06:45 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1108 (0.4279) acc1: 100.0000 (76.7842) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1599 data: 0.0660 max mem: 3533 +Test: [ 520/2387] eta: 0:06:41 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1587 (0.4229) acc1: 100.0000 (77.2298) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0753 max mem: 3533 +Test: [ 530/2387] eta: 0:06:39 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1571 (0.4179) acc1: 100.0000 (77.6586) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2014 data: 0.1055 max mem: 3533 +Test: [ 540/2387] eta: 0:06:36 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1548 (0.4130) acc1: 100.0000 (78.0716) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2032 data: 0.1074 max mem: 3533 +Test: [ 550/2387] eta: 0:06:32 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1548 (0.4083) acc1: 100.0000 (78.4695) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1746 data: 0.0816 max mem: 3533 +Test: [ 560/2387] eta: 0:06:29 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1552 (0.4038) acc1: 100.0000 (78.8533) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1739 data: 0.0798 max mem: 3533 +Test: [ 570/2387] eta: 0:06:26 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1541 (0.3995) acc1: 100.0000 (79.2237) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1862 data: 0.0910 max mem: 3533 +Test: [ 580/2387] eta: 0:06:23 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1544 (0.3953) acc1: 100.0000 (79.5813) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1834 data: 0.0898 max mem: 3533 +Test: [ 590/2387] eta: 0:06:19 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1561 (0.3912) acc1: 100.0000 (79.9268) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1724 data: 0.0768 max mem: 3533 +Test: [ 600/2387] eta: 0:06:16 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1608 (0.3876) acc1: 100.0000 (80.2603) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1740 data: 0.0758 max mem: 3533 +Test: [ 610/2387] eta: 0:06:13 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1837 (0.3846) acc1: 99.7396 (80.5753) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1799 data: 0.0810 max mem: 3533 +Test: [ 620/2387] eta: 0:06:10 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2031 (0.3818) acc1: 99.4792 (80.8780) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1812 data: 0.0823 max mem: 3533 +Test: [ 630/2387] eta: 0:06:07 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2039 (0.3790) acc1: 99.4792 (81.1716) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1703 data: 0.0731 max mem: 3533 +Test: [ 640/2387] eta: 0:06:04 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2066 (0.3763) acc1: 99.4792 (81.4544) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1642 data: 0.0674 max mem: 3533 +Test: [ 650/2387] eta: 0:06:01 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2073 (0.3738) acc1: 99.4792 (81.7296) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1696 data: 0.0720 max mem: 3533 +Test: [ 660/2387] eta: 0:05:57 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2073 (0.3712) acc1: 99.4792 (81.9989) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1692 data: 0.0711 max mem: 3533 +Test: [ 670/2387] eta: 0:05:54 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2053 (0.3688) acc1: 99.4792 (82.2598) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1618 data: 0.0632 max mem: 3534 +Test: [ 680/2387] eta: 0:05:51 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2112 (0.3665) acc1: 99.7396 (82.5177) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1649 data: 0.0681 max mem: 3534 +Test: [ 690/2387] eta: 0:05:48 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2169 (0.3679) acc1: 99.2188 (82.4654) acc5: 100.0000 (100.0000) acc1_neg: 68.229 acc5_neg: 100.000 time: 0.1599 data: 0.0631 max mem: 3534 +Test: [ 700/2387] eta: 0:05:45 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.5402 (0.3704) acc1: 71.8750 (82.3132) acc5: 100.0000 (100.0000) acc1_neg: 74.740 acc5_neg: 100.000 time: 0.1661 data: 0.0710 max mem: 3534 +Test: [ 710/2387] eta: 0:05:42 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.5432 (0.3729) acc1: 71.3542 (82.1551) acc5: 100.0000 (100.0000) acc1_neg: 70.573 acc5_neg: 100.000 time: 0.1740 data: 0.0816 max mem: 3534 +Test: [ 720/2387] eta: 0:05:39 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.5439 (0.3751) acc1: 71.3542 (82.0211) acc5: 100.0000 (100.0000) acc1_neg: 72.656 acc5_neg: 100.000 time: 0.1623 data: 0.0690 max mem: 3534 +Test: [ 730/2387] eta: 0:05:36 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.5529 (0.3777) acc1: 70.5729 (81.8627) acc5: 100.0000 (100.0000) acc1_neg: 67.969 acc5_neg: 100.000 time: 0.1653 data: 0.0694 max mem: 3534 +Test: [ 740/2387] eta: 0:05:33 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.5499 (0.3800) acc1: 70.5729 (81.7280) acc5: 100.0000 (100.0000) acc1_neg: 68.229 acc5_neg: 100.000 time: 0.1656 data: 0.0711 max mem: 3534 +Test: [ 750/2387] eta: 0:05:30 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.5394 (0.3821) acc1: 70.8333 (81.5968) acc5: 100.0000 (100.0000) acc1_neg: 72.396 acc5_neg: 100.000 time: 0.1589 data: 0.0666 max mem: 3534 +Test: [ 760/2387] eta: 0:05:28 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.5295 (0.3833) acc1: 71.3542 (81.5354) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.1701 data: 0.0762 max mem: 3534 +Test: [ 770/2387] eta: 0:05:25 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.3614 (0.3820) acc1: 85.1562 (81.6624) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1734 data: 0.0792 max mem: 3534 +Test: [ 780/2387] eta: 0:05:22 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1928 (0.3795) acc1: 96.3542 (81.8622) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1662 data: 0.0717 max mem: 3534 +Test: [ 790/2387] eta: 0:05:19 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1713 (0.3771) acc1: 97.6562 (82.0316) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1637 data: 0.0714 max mem: 3534 +Test: [ 800/2387] eta: 0:05:17 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1582 (0.3746) acc1: 98.9583 (82.2449) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1669 data: 0.0758 max mem: 3534 +Test: [ 810/2387] eta: 0:05:14 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1793 (0.3723) acc1: 98.9583 (82.4455) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1677 data: 0.0736 max mem: 3534 +Test: [ 820/2387] eta: 0:05:12 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1941 (0.3704) acc1: 98.6979 (82.6295) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1680 data: 0.0746 max mem: 3534 +Test: [ 830/2387] eta: 0:05:09 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2367 (0.3694) acc1: 95.3125 (82.7586) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1784 data: 0.0867 max mem: 3534 +Test: [ 840/2387] eta: 0:05:07 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2936 (0.3685) acc1: 95.3125 (82.9066) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1729 data: 0.0787 max mem: 3534 +Test: [ 850/2387] eta: 0:05:04 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2917 (0.3677) acc1: 95.8333 (83.0331) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1635 data: 0.0673 max mem: 3534 +Test: [ 860/2387] eta: 0:05:02 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2824 (0.3666) acc1: 95.8333 (83.1987) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1638 data: 0.0683 max mem: 3534 +Test: [ 870/2387] eta: 0:04:59 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2680 (0.3656) acc1: 97.1354 (83.3480) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1649 data: 0.0715 max mem: 3534 +Test: [ 880/2387] eta: 0:04:57 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2837 (0.3648) acc1: 96.8750 (83.4912) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1699 data: 0.0749 max mem: 3534 +Test: [ 890/2387] eta: 0:04:54 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2571 (0.3633) acc1: 98.6979 (83.6636) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1649 data: 0.0670 max mem: 3534 +Test: [ 900/2387] eta: 0:04:51 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2698 (0.3683) acc1: 95.3125 (83.3377) acc5: 100.0000 (100.0000) acc1_neg: 45.573 acc5_neg: 100.000 time: 0.1601 data: 0.0655 max mem: 3535 +Test: [ 910/2387] eta: 0:04:50 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.9442 (0.3747) acc1: 43.4896 (82.9071) acc5: 100.0000 (100.0000) acc1_neg: 42.448 acc5_neg: 100.000 time: 0.1799 data: 0.0850 max mem: 3535 +Test: [ 920/2387] eta: 0:04:47 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.9537 (0.3810) acc1: 43.2292 (82.4842) acc5: 100.0000 (100.0000) acc1_neg: 42.188 acc5_neg: 100.000 time: 0.1947 data: 0.0958 max mem: 3535 +Test: [ 930/2387] eta: 0:04:45 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.9496 (0.3871) acc1: 44.0104 (82.0822) acc5: 100.0000 (100.0000) acc1_neg: 42.448 acc5_neg: 100.000 time: 0.1899 data: 0.0947 max mem: 3535 +Test: [ 940/2387] eta: 0:04:43 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.9496 (0.3931) acc1: 44.0104 (81.6803) acc5: 100.0000 (100.0000) acc1_neg: 40.885 acc5_neg: 100.000 time: 0.1749 data: 0.0803 max mem: 3535 +Test: [ 950/2387] eta: 0:04:41 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.9484 (0.3989) acc1: 43.4896 (81.2749) acc5: 100.0000 (100.0000) acc1_neg: 45.052 acc5_neg: 100.000 time: 0.1749 data: 0.0795 max mem: 3535 +Test: [ 960/2387] eta: 0:04:38 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.9346 (0.4045) acc1: 44.2708 (80.8953) acc5: 100.0000 (100.0000) acc1_neg: 47.396 acc5_neg: 100.000 time: 0.1799 data: 0.0872 max mem: 3535 +Test: [ 970/2387] eta: 0:04:36 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.9361 (0.4101) acc1: 44.0104 (80.5135) acc5: 100.0000 (100.0000) acc1_neg: 47.656 acc5_neg: 100.000 time: 0.1799 data: 0.0843 max mem: 3535 +Test: [ 980/2387] eta: 0:04:34 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.9313 (0.4107) acc1: 44.2708 (80.4879) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.1798 data: 0.0841 max mem: 3535 +Test: [ 990/2387] eta: 0:04:32 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2014 (0.4086) acc1: 95.8333 (80.6456) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1799 data: 0.0857 max mem: 3535 +Test: [1000/2387] eta: 0:04:30 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1733 (0.4064) acc1: 99.2188 (80.8122) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1899 data: 0.0951 max mem: 3535 +Test: [1010/2387] eta: 0:04:28 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1908 (0.4047) acc1: 98.1771 (80.9641) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1949 data: 0.0990 max mem: 3535 +Test: [1020/2387] eta: 0:04:26 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2271 (0.4031) acc1: 97.3958 (81.1013) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1899 data: 0.0917 max mem: 3535 +Test: [1030/2387] eta: 0:04:24 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1939 (0.4010) acc1: 98.4375 (81.2695) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0767 max mem: 3535 +Test: [1040/2387] eta: 0:04:22 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1641 (0.3986) acc1: 99.7396 (81.4444) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1799 data: 0.0829 max mem: 3535 +Test: [1050/2387] eta: 0:04:19 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1569 (0.3964) acc1: 99.7396 (81.6150) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1700 data: 0.0755 max mem: 3535 +Test: [1060/2387] eta: 0:04:17 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1646 (0.3944) acc1: 99.7396 (81.7826) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0749 max mem: 3535 +Test: [1070/2387] eta: 0:04:15 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1809 (0.3926) acc1: 99.2188 (81.9313) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1748 data: 0.0798 max mem: 3535 +Test: [1080/2387] eta: 0:04:13 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2232 (0.3914) acc1: 97.1354 (82.0626) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1699 data: 0.0764 max mem: 3535 +Test: [1090/2387] eta: 0:04:12 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2007 (0.3896) acc1: 97.1354 (82.2062) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2576 data: 0.1648 max mem: 3535 +Test: [1100/2387] eta: 0:04:09 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1865 (0.3878) acc1: 97.9167 (82.3465) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2149 data: 0.1223 max mem: 3535 +Test: [1110/2387] eta: 0:04:07 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1840 (0.3863) acc1: 97.9167 (82.4480) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1472 data: 0.0544 max mem: 3535 +Test: [1120/2387] eta: 0:04:05 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1899 (0.3846) acc1: 98.4375 (82.5888) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1849 data: 0.0907 max mem: 3536 +Test: [1130/2387] eta: 0:04:03 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1924 (0.3831) acc1: 98.4375 (82.7186) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1612 data: 0.0662 max mem: 3536 +Test: [1140/2387] eta: 0:04:01 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2320 (0.3819) acc1: 97.3958 (82.8461) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1599 data: 0.0656 max mem: 3536 +Test: [1150/2387] eta: 0:03:58 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2672 (0.3811) acc1: 96.6146 (82.9587) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1636 data: 0.0693 max mem: 3536 +Test: [1160/2387] eta: 0:03:56 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2740 (0.3799) acc1: 94.2708 (83.0648) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1749 data: 0.0776 max mem: 3536 +Test: [1170/2387] eta: 0:03:54 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2399 (0.3788) acc1: 95.8333 (83.1734) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1749 data: 0.0762 max mem: 3536 +Test: [1180/2387] eta: 0:03:52 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2399 (0.3778) acc1: 96.0938 (83.2762) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1549 data: 0.0570 max mem: 3536 +Test: [1190/2387] eta: 0:03:50 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2334 (0.3768) acc1: 97.1354 (83.3895) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1599 data: 0.0614 max mem: 3536 +Test: [1200/2387] eta: 0:03:48 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2151 (0.3754) acc1: 98.6979 (83.5209) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1901 data: 0.0956 max mem: 3536 +Test: [1210/2387] eta: 0:03:46 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1832 (0.3739) acc1: 99.4792 (83.6505) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.1009 max mem: 3536 +Test: [1220/2387] eta: 0:03:43 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1359 (0.3718) acc1: 99.7396 (83.7823) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1447 data: 0.0487 max mem: 3536 +Test: [1230/2387] eta: 0:03:41 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1072 (0.3696) acc1: 99.7396 (83.9121) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1301 data: 0.0333 max mem: 3536 +Test: [1240/2387] eta: 0:03:39 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1076 (0.3675) acc1: 100.0000 (84.0399) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0731 max mem: 3536 +Test: [1250/2387] eta: 0:03:37 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1126 (0.3655) acc1: 100.0000 (84.1664) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1767 data: 0.0824 max mem: 3536 +Test: [1260/2387] eta: 0:03:35 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1120 (0.3635) acc1: 100.0000 (84.2899) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0870 max mem: 3536 +Test: [1270/2387] eta: 0:03:33 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1118 (0.3615) acc1: 100.0000 (84.4133) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1779 data: 0.0845 max mem: 3536 +Test: [1280/2387] eta: 0:03:30 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1152 (0.3597) acc1: 100.0000 (84.5328) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1425 data: 0.0482 max mem: 3536 +Test: [1290/2387] eta: 0:03:28 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1279 (0.3580) acc1: 99.4792 (84.6334) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1649 data: 0.0679 max mem: 3536 +Test: [1300/2387] eta: 0:03:27 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1235 (0.3562) acc1: 99.4792 (84.7423) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2227 data: 0.1271 max mem: 3536 +Test: [1310/2387] eta: 0:03:25 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1282 (0.3546) acc1: 99.4792 (84.8484) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2156 data: 0.1207 max mem: 3536 +Test: [1320/2387] eta: 0:03:23 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1527 (0.3532) acc1: 99.4792 (84.9495) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1815 data: 0.0841 max mem: 3536 +Test: [1330/2387] eta: 0:03:21 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1671 (0.3519) acc1: 99.4792 (85.0543) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1709 data: 0.0750 max mem: 3536 +Test: [1340/2387] eta: 0:03:19 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2186 (0.3516) acc1: 97.9167 (85.0860) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1653 data: 0.0705 max mem: 3536 +Test: [1350/2387] eta: 0:03:17 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.0882 (0.3496) acc1: 100.0000 (85.1963) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1632 data: 0.0685 max mem: 3537 +Test: [1360/2387] eta: 0:03:15 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.0643 (0.3474) acc1: 100.0000 (85.3051) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1654 data: 0.0691 max mem: 3537 +Test: [1370/2387] eta: 0:03:13 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.0592 (0.3454) acc1: 100.0000 (85.4121) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0744 max mem: 3537 +Test: [1380/2387] eta: 0:03:11 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.0639 (0.3434) acc1: 100.0000 (85.5177) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1769 data: 0.0809 max mem: 3537 +Test: [1390/2387] eta: 0:03:09 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.0767 (0.3415) acc1: 100.0000 (85.6219) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1770 data: 0.0781 max mem: 3537 +Test: [1400/2387] eta: 0:03:06 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.0776 (0.3396) acc1: 100.0000 (85.7245) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1700 data: 0.0736 max mem: 3537 +Test: [1410/2387] eta: 0:03:05 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.0835 (0.3384) acc1: 100.0000 (85.8089) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1830 data: 0.0867 max mem: 3537 +Test: [1420/2387] eta: 0:03:03 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1955 (0.3376) acc1: 97.9167 (85.8739) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1898 data: 0.0912 max mem: 3537 +Test: [1430/2387] eta: 0:03:01 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1975 (0.3366) acc1: 97.3958 (85.9550) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1811 data: 0.0824 max mem: 3537 +Test: [1440/2387] eta: 0:02:59 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2057 (0.3358) acc1: 97.9167 (86.0411) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1828 data: 0.0875 max mem: 3537 +Test: [1450/2387] eta: 0:02:57 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2182 (0.3351) acc1: 98.6979 (86.1207) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1804 data: 0.0880 max mem: 3537 +Test: [1460/2387] eta: 0:02:55 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1870 (0.3341) acc1: 98.6979 (86.2065) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1831 data: 0.0894 max mem: 3537 +Test: [1470/2387] eta: 0:02:53 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1480 (0.3327) acc1: 99.4792 (86.2988) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1830 data: 0.0856 max mem: 3537 +Test: [1480/2387] eta: 0:02:51 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1442 (0.3315) acc1: 99.7396 (86.3866) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1798 data: 0.0804 max mem: 3537 +Test: [1490/2387] eta: 0:02:49 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1505 (0.3304) acc1: 99.7396 (86.4760) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1763 data: 0.0787 max mem: 3537 +Test: [1500/2387] eta: 0:02:47 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1621 (0.3293) acc1: 99.7396 (86.5628) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1539 data: 0.0578 max mem: 3537 +Test: [1510/2387] eta: 0:02:45 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1744 (0.3283) acc1: 100.0000 (86.6505) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1635 data: 0.0672 max mem: 3537 +Test: [1520/2387] eta: 0:02:43 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2052 (0.3279) acc1: 99.2188 (86.7181) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1761 data: 0.0802 max mem: 3537 +Test: [1530/2387] eta: 0:02:41 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2602 (0.3275) acc1: 97.1354 (86.7706) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1517 data: 0.0586 max mem: 3537 +Test: [1540/2387] eta: 0:02:39 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2403 (0.3270) acc1: 97.6562 (86.8423) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1686 data: 0.0767 max mem: 3537 +Test: [1550/2387] eta: 0:02:37 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2403 (0.3265) acc1: 98.1771 (86.9095) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1990 data: 0.1061 max mem: 3537 +Test: [1560/2387] eta: 0:02:35 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2514 (0.3261) acc1: 97.3958 (86.9748) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1900 data: 0.0936 max mem: 3537 +Test: [1570/2387] eta: 0:02:33 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2280 (0.3254) acc1: 99.2188 (87.0518) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1795 data: 0.0835 max mem: 3537 +Test: [1580/2387] eta: 0:02:31 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1972 (0.3245) acc1: 99.2188 (87.1266) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1703 data: 0.0789 max mem: 3538 +Test: [1590/2387] eta: 0:02:29 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1695 (0.3235) acc1: 98.9583 (87.2016) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1727 data: 0.0817 max mem: 3538 +Test: [1600/2387] eta: 0:02:28 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1558 (0.3227) acc1: 98.9583 (87.2469) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1944 data: 0.1020 max mem: 3538 +Test: [1610/2387] eta: 0:02:26 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1558 (0.3217) acc1: 99.2188 (87.3209) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2086 data: 0.1160 max mem: 3538 +Test: [1620/2387] eta: 0:02:24 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1746 (0.3209) acc1: 99.2188 (87.3882) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2148 data: 0.1218 max mem: 3538 +Test: [1630/2387] eta: 0:02:22 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1945 (0.3202) acc1: 97.9167 (87.4451) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1928 data: 0.0983 max mem: 3538 +Test: [1640/2387] eta: 0:02:20 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2171 (0.3198) acc1: 97.3958 (87.5013) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1904 data: 0.0968 max mem: 3538 +Test: [1650/2387] eta: 0:02:19 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2248 (0.3192) acc1: 97.3958 (87.5604) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2104 data: 0.1172 max mem: 3538 +Test: [1660/2387] eta: 0:02:17 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2297 (0.3187) acc1: 97.1354 (87.6141) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2154 data: 0.1228 max mem: 3538 +Test: [1670/2387] eta: 0:02:15 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2354 (0.3183) acc1: 97.1354 (87.6688) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2244 data: 0.1299 max mem: 3538 +Test: [1680/2387] eta: 0:02:13 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2356 (0.3179) acc1: 97.1354 (87.7214) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2155 data: 0.1181 max mem: 3538 +Test: [1690/2387] eta: 0:02:11 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2314 (0.3174) acc1: 97.9167 (87.7760) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1921 data: 0.0980 max mem: 3538 +Test: [1700/2387] eta: 0:02:10 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2305 (0.3168) acc1: 97.9167 (87.8359) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1983 data: 0.1055 max mem: 3538 +Test: [1710/2387] eta: 0:02:08 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1687 (0.3157) acc1: 98.6979 (87.9024) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1942 data: 0.1002 max mem: 3538 +Test: [1720/2387] eta: 0:02:06 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1313 (0.3147) acc1: 99.4792 (87.9694) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2202 data: 0.1280 max mem: 3538 +Test: [1730/2387] eta: 0:02:04 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1335 (0.3137) acc1: 99.4792 (88.0321) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2083 data: 0.1126 max mem: 3538 +Test: [1740/2387] eta: 0:02:02 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1315 (0.3126) acc1: 99.7396 (88.0977) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1658 data: 0.0652 max mem: 3538 +Test: [1750/2387] eta: 0:02:00 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1264 (0.3116) acc1: 100.0000 (88.1635) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1923 data: 0.0969 max mem: 3538 +Test: [1760/2387] eta: 0:01:58 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1320 (0.3106) acc1: 100.0000 (88.2307) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1955 data: 0.1049 max mem: 3538 +Test: [1770/2387] eta: 0:01:56 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1412 (0.3098) acc1: 99.7396 (88.2877) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1840 data: 0.0911 max mem: 3538 +Test: [1780/2387] eta: 0:01:54 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1504 (0.3089) acc1: 98.9583 (88.3503) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1882 data: 0.0937 max mem: 3538 +Test: [1790/2387] eta: 0:01:53 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1422 (0.3079) acc1: 99.4792 (88.4112) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1884 data: 0.0940 max mem: 3538 +Test: [1800/2387] eta: 0:01:51 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1409 (0.3070) acc1: 99.2188 (88.4705) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1799 data: 0.0827 max mem: 3538 +Test: [1810/2387] eta: 0:01:49 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1409 (0.3060) acc1: 99.2188 (88.5322) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1756 data: 0.0795 max mem: 3539 +Test: [1820/2387] eta: 0:01:47 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1317 (0.3051) acc1: 100.0000 (88.5947) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1712 data: 0.0790 max mem: 3539 +Test: [1830/2387] eta: 0:01:45 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1434 (0.3042) acc1: 100.0000 (88.6566) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1743 data: 0.0824 max mem: 3539 +Test: [1840/2387] eta: 0:01:43 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1557 (0.3034) acc1: 100.0000 (88.7182) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1772 data: 0.0855 max mem: 3539 +Test: [1850/2387] eta: 0:01:41 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1571 (0.3026) acc1: 100.0000 (88.7792) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1750 data: 0.0831 max mem: 3539 +Test: [1860/2387] eta: 0:01:39 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1565 (0.3019) acc1: 100.0000 (88.8394) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1771 data: 0.0840 max mem: 3539 +Test: [1870/2387] eta: 0:01:37 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1561 (0.3011) acc1: 100.0000 (88.8990) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1766 data: 0.0823 max mem: 3539 +Test: [1880/2387] eta: 0:01:35 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1553 (0.3003) acc1: 100.0000 (88.9580) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1700 data: 0.0755 max mem: 3539 +Test: [1890/2387] eta: 0:01:33 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1558 (0.2995) acc1: 100.0000 (89.0164) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1681 data: 0.0757 max mem: 3539 +Test: [1900/2387] eta: 0:01:31 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1574 (0.2988) acc1: 100.0000 (89.0741) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1740 data: 0.0820 max mem: 3539 +Test: [1910/2387] eta: 0:01:29 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1702 (0.2982) acc1: 100.0000 (89.1295) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1774 data: 0.0847 max mem: 3539 +Test: [1920/2387] eta: 0:01:27 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1948 (0.2978) acc1: 99.7396 (89.1535) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1714 data: 0.0799 max mem: 3539 +Test: [1930/2387] eta: 0:01:26 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1594 (0.2971) acc1: 99.4792 (89.2068) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1774 data: 0.0848 max mem: 3539 +Test: [1940/2387] eta: 0:01:24 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1528 (0.2967) acc1: 99.7396 (89.2243) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1985 data: 0.1055 max mem: 3539 +Test: [1950/2387] eta: 0:01:22 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1748 (0.2961) acc1: 98.9583 (89.2695) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1999 data: 0.1073 max mem: 3539 +Test: [1960/2387] eta: 0:01:20 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1883 (0.2957) acc1: 98.6979 (89.3132) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1813 data: 0.0886 max mem: 3539 +Test: [1970/2387] eta: 0:01:18 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2288 (0.2956) acc1: 97.1354 (89.3187) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.1816 data: 0.0908 max mem: 3539 +Test: [1980/2387] eta: 0:01:16 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2453 (0.2953) acc1: 95.3125 (89.3521) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1984 data: 0.1050 max mem: 3539 +Test: [1990/2387] eta: 0:01:14 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1933 (0.2948) acc1: 97.9167 (89.3947) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1988 data: 0.1050 max mem: 3539 +Test: [2000/2387] eta: 0:01:12 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1943 (0.2943) acc1: 97.9167 (89.4368) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1878 data: 0.0961 max mem: 3539 +Test: [2010/2387] eta: 0:01:11 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1964 (0.2939) acc1: 97.9167 (89.4777) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1962 data: 0.1028 max mem: 3539 +Test: [2020/2387] eta: 0:01:09 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1988 (0.2934) acc1: 97.6562 (89.5183) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2020 data: 0.1084 max mem: 3539 +Test: [2030/2387] eta: 0:01:07 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1966 (0.2929) acc1: 97.3958 (89.5568) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1898 data: 0.0955 max mem: 3540 +Test: [2040/2387] eta: 0:01:05 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1947 (0.2924) acc1: 97.6562 (89.5975) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1920 data: 0.0962 max mem: 3540 +Test: [2050/2387] eta: 0:01:03 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1944 (0.2920) acc1: 97.9167 (89.6388) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1945 data: 0.1001 max mem: 3540 +Test: [2060/2387] eta: 0:01:01 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2071 (0.2916) acc1: 97.9167 (89.6785) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1980 data: 0.1054 max mem: 3540 +Test: [2070/2387] eta: 0:00:59 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2131 (0.2912) acc1: 97.9167 (89.7209) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2060 data: 0.1143 max mem: 3540 +Test: [2080/2387] eta: 0:00:57 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2069 (0.2908) acc1: 98.1771 (89.7608) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1850 data: 0.0931 max mem: 3540 +Test: [2090/2387] eta: 0:00:56 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2073 (0.2904) acc1: 97.9167 (89.8008) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1924 data: 0.0996 max mem: 3540 +Test: [2100/2387] eta: 0:00:54 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2092 (0.2900) acc1: 98.1771 (89.8415) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1882 data: 0.0935 max mem: 3540 +Test: [2110/2387] eta: 0:00:52 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2071 (0.2896) acc1: 98.1771 (89.8817) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1549 data: 0.0606 max mem: 3540 +Test: [2120/2387] eta: 0:00:50 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2070 (0.2893) acc1: 98.1771 (89.9167) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1733 data: 0.0795 max mem: 3540 +Test: [2130/2387] eta: 0:00:48 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1746 (0.2887) acc1: 99.2188 (89.9601) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2337 data: 0.1412 max mem: 3540 +Test: [2140/2387] eta: 0:00:46 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1632 (0.2883) acc1: 99.4792 (89.9831) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1997 data: 0.1091 max mem: 3540 +Test: [2150/2387] eta: 0:00:44 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1854 (0.2878) acc1: 98.4375 (90.0232) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1423 data: 0.0504 max mem: 3540 +Test: [2160/2387] eta: 0:00:42 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1899 (0.2874) acc1: 98.6979 (90.0615) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1984 data: 0.1054 max mem: 3540 +Test: [2170/2387] eta: 0:00:40 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2105 (0.2873) acc1: 97.9167 (90.0897) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2044 data: 0.1110 max mem: 3540 +Test: [2180/2387] eta: 0:00:39 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2773 (0.2882) acc1: 93.7500 (90.0353) acc5: 100.0000 (100.0000) acc1_neg: 54.427 acc5_neg: 100.000 time: 0.1750 data: 0.0822 max mem: 3540 +Test: [2190/2387] eta: 0:00:37 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.3744 (0.2888) acc1: 86.4583 (89.9993) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1942 data: 0.1003 max mem: 3540 +Test: [2200/2387] eta: 0:00:35 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1652 (0.2882) acc1: 98.4375 (90.0413) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2000 data: 0.1028 max mem: 3540 +Test: [2210/2387] eta: 0:00:33 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1428 (0.2875) acc1: 99.7396 (90.0844) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1812 data: 0.0851 max mem: 3540 +Test: [2220/2387] eta: 0:00:31 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1471 (0.2869) acc1: 99.7396 (90.1239) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1818 data: 0.0887 max mem: 3540 +Test: [2230/2387] eta: 0:00:29 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1496 (0.2863) acc1: 99.4792 (90.1665) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1856 data: 0.0928 max mem: 3540 +Test: [2240/2387] eta: 0:00:27 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1379 (0.2856) acc1: 99.4792 (90.2061) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1887 data: 0.0957 max mem: 3540 +Test: [2250/2387] eta: 0:00:25 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1605 (0.2852) acc1: 99.2188 (90.2445) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1987 data: 0.1068 max mem: 3540 +Test: [2260/2387] eta: 0:00:23 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1843 (0.2848) acc1: 99.7396 (90.2868) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1940 data: 0.1019 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1655 (0.2842) acc1: 100.0000 (90.3294) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1869 data: 0.0950 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1408 (0.2837) acc1: 100.0000 (90.3521) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1798 data: 0.0889 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1408 (0.2831) acc1: 99.7396 (90.3937) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1721 data: 0.0807 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1595 (0.2827) acc1: 99.7396 (90.4320) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1781 data: 0.0856 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1850 (0.2824) acc1: 99.7396 (90.4659) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1816 data: 0.0883 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2148 (0.2822) acc1: 98.6979 (90.4954) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1783 data: 0.0850 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2277 (0.2820) acc1: 97.3958 (90.5193) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.1935 data: 0.0998 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2129 (0.2816) acc1: 97.6562 (90.5492) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1908 data: 0.0981 max mem: 3541 +Test: [2350/2387] eta: 0:00:06 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.1937 (0.2816) acc1: 97.6562 (90.5551) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1697 data: 0.0778 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2213 (0.2815) acc1: 95.8333 (90.5759) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1695 data: 0.0775 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2508 (0.2814) acc1: 95.8333 (90.5917) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1640 data: 0.0724 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.2596 (0.2814) acc1: 94.0104 (90.6049) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1496 data: 0.0578 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 41.553 acc5_pos: 100.000 loss: 0.3271 (0.2816) acc1: 90.6977 (90.6015) acc5: 100.0000 (100.0000) acc1_neg: 90.698 acc5_neg: 100.000 time: 0.1394 data: 0.0508 max mem: 3541 +Test: Total time: 0:07:28 (0.1880 s / it) +* Overall Acc@1 90.602 Acc@5 100.000 + target == real Acc@1 31.176 Acc@5 100.000 + target == fake Acc@1 94.775 Acc@5 100.000 + loss 0.282 + PR-AUC 0.9919 ROC-AUC 0.8924 EER 17.49% (thr=0.7100) +Accuracy of the network on the 916267 test images: 90.6% +[rank0]:[W726 03:32:00.397442693 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:32:21.865000 1177009 torch/distributed/run.py:766] +W0726 03:32:21.865000 1177009 torch/distributed/run.py:766] ***************************************** +W0726 03:32:21.865000 1177009 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:32:21.865000 1177009 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:32:51.826233784 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:32:51.114169033 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:32:52.208568249 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:32:52.534518908 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:32:52.574593601 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:32:52.774102451 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:32:52.790551065 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank4]:[W726 03:32:52.790670684 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n64_seed5_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed5_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:02:13 acc1_pos: 37.240 acc5_pos: 100.000 loss: 0.8449 (0.8449) acc1: 37.2396 (37.2396) acc5: 100.0000 (100.0000) time: 7.2212 data: 5.7075 max mem: 3531 +Test: [ 10/517] eta: 0:06:15 acc1_pos: 13.021 acc5_pos: 100.000 loss: 0.9412 (0.9537) acc1: 28.9062 (27.6278) acc5: 100.0000 (100.0000) time: 0.7402 data: 0.5192 max mem: 3531 +Test: [ 20/517] eta: 0:03:35 acc1_pos: 24.479 acc5_pos: 100.000 loss: 0.9269 (0.9599) acc1: 24.7396 (27.0461) acc5: 100.0000 (100.0000) time: 0.0949 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:03:01 acc1_pos: 5.208 acc5_pos: 100.000 loss: 0.9862 (0.9714) acc1: 20.3125 (25.0672) acc5: 100.0000 (100.0000) time: 0.1692 data: 0.0753 max mem: 3531 +Test: [ 40/517] eta: 0:02:44 acc1_pos: 17.708 acc5_pos: 100.000 loss: 0.9944 (0.9806) acc1: 17.4479 (22.5673) acc5: 100.0000 (100.0000) time: 0.2511 data: 0.1559 max mem: 3531 +Test: [ 50/517] eta: 0:02:33 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.9305 (0.9428) acc1: 21.8750 (27.0629) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2645 data: 0.1666 max mem: 3531 +Test: [ 60/517] eta: 0:02:28 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.4434 (0.8545) acc1: 83.8542 (37.7732) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.2838 data: 0.1873 max mem: 3531 +Test: [ 70/517] eta: 0:02:21 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.3776 (0.7853) acc1: 94.2708 (45.8847) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2799 data: 0.1826 max mem: 3531 +Test: [ 80/517] eta: 0:02:13 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.3813 (0.7369) acc1: 94.2708 (51.8647) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.2449 data: 0.1488 max mem: 3531 +Test: [ 90/517] eta: 0:02:09 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.3931 (0.6975) acc1: 93.4896 (56.4961) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.2599 data: 0.1664 max mem: 3531 +Test: [100/517] eta: 0:02:04 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.3545 (0.6622) acc1: 94.5312 (60.5146) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2749 data: 0.1798 max mem: 3531 +Test: [110/517] eta: 0:01:59 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.3096 (0.6293) acc1: 98.9583 (63.9194) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2549 data: 0.1572 max mem: 3531 +Test: [120/517] eta: 0:01:56 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.3466 (0.6091) acc1: 96.3542 (66.4794) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2599 data: 0.1611 max mem: 3531 +Test: [130/517] eta: 0:01:51 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.3825 (0.5916) acc1: 95.0521 (68.6108) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2599 data: 0.1593 max mem: 3531 +Test: [140/517] eta: 0:01:48 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.3455 (0.5727) acc1: 97.6562 (70.7502) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2549 data: 0.1548 max mem: 3531 +Test: [150/517] eta: 0:01:45 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.3193 (0.5554) acc1: 98.4375 (72.5442) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2649 data: 0.1665 max mem: 3531 +Test: [160/517] eta: 0:01:42 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.3066 (0.5402) acc1: 98.1771 (74.1233) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2849 data: 0.1864 max mem: 3531 +Test: [170/517] eta: 0:01:38 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.3066 (0.5264) acc1: 97.9167 (75.4569) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2649 data: 0.1663 max mem: 3531 +Test: [180/517] eta: 0:01:35 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.3099 (0.5141) acc1: 97.1354 (76.6388) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2549 data: 0.1568 max mem: 3531 +Test: [190/517] eta: 0:01:31 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.2297 (0.4983) acc1: 99.4792 (77.8523) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1583 max mem: 3531 +Test: [200/517] eta: 0:01:28 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.2295 (0.4871) acc1: 99.7396 (78.9322) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2399 data: 0.1425 max mem: 3531 +Test: [210/517] eta: 0:01:25 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.3193 (0.4824) acc1: 95.3125 (79.4481) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2549 data: 0.1584 max mem: 3532 +Test: [220/517] eta: 0:01:23 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.4004 (0.4790) acc1: 88.0208 (79.8112) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.3199 data: 0.2269 max mem: 3532 +Test: [230/517] eta: 0:01:19 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.4032 (0.4751) acc1: 89.0625 (80.5071) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2749 data: 0.1816 max mem: 3532 +Test: [240/517] eta: 0:01:17 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.3627 (0.4684) acc1: 98.1771 (81.2338) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.2449 data: 0.1482 max mem: 3532 +Test: [250/517] eta: 0:01:14 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.2999 (0.4627) acc1: 99.2188 (81.9119) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2700 data: 0.1739 max mem: 3532 +Test: [260/517] eta: 0:01:11 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.3152 (0.4577) acc1: 99.4792 (82.5451) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2449 data: 0.1492 max mem: 3532 +Test: [270/517] eta: 0:01:07 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.3548 (0.4547) acc1: 95.8333 (82.9932) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2449 data: 0.1478 max mem: 3532 +Test: [280/517] eta: 0:01:05 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.3548 (0.4507) acc1: 95.3125 (83.4622) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2449 data: 0.1500 max mem: 3532 +Test: [290/517] eta: 0:01:02 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.3366 (0.4475) acc1: 97.1354 (83.9589) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2549 data: 0.1597 max mem: 3532 +Test: [300/517] eta: 0:00:59 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.3366 (0.4438) acc1: 98.4375 (84.4589) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2649 data: 0.1701 max mem: 3532 +Test: [310/517] eta: 0:00:56 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.3582 (0.4418) acc1: 99.2188 (84.9151) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2749 data: 0.1814 max mem: 3532 +Test: [320/517] eta: 0:00:53 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.3664 (0.4387) acc1: 98.9583 (85.3339) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2649 data: 0.1709 max mem: 3532 +Test: [330/517] eta: 0:00:51 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.3155 (0.4345) acc1: 98.9583 (85.7487) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2699 data: 0.1760 max mem: 3532 +Test: [340/517] eta: 0:00:48 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.2960 (0.4314) acc1: 99.2188 (86.0612) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.2599 data: 0.1658 max mem: 3532 +Test: [350/517] eta: 0:00:45 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.3539 (0.4300) acc1: 96.8750 (86.3359) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.2549 data: 0.1600 max mem: 3532 +Test: [360/517] eta: 0:00:42 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.3618 (0.4282) acc1: 95.8333 (86.5802) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2549 data: 0.1590 max mem: 3532 +Test: [370/517] eta: 0:00:39 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.3176 (0.4244) acc1: 99.4792 (86.9392) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2499 data: 0.1544 max mem: 3532 +Test: [380/517] eta: 0:00:37 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.2937 (0.4216) acc1: 99.7396 (87.2560) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2549 data: 0.1600 max mem: 3532 +Test: [390/517] eta: 0:00:34 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.3166 (0.4193) acc1: 99.4792 (87.5699) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2599 data: 0.1627 max mem: 3532 +Test: [400/517] eta: 0:00:31 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.3286 (0.4167) acc1: 99.7396 (87.8630) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1558 max mem: 3532 +Test: [410/517] eta: 0:00:28 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.3335 (0.4153) acc1: 99.2188 (88.1324) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2499 data: 0.1512 max mem: 3532 +Test: [420/517] eta: 0:00:26 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.3557 (0.4138) acc1: 99.2188 (88.3808) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2749 data: 0.1783 max mem: 3532 +Test: [430/517] eta: 0:00:23 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.3519 (0.4124) acc1: 98.6979 (88.5447) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.2649 data: 0.1677 max mem: 3532 +Test: [440/517] eta: 0:00:20 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.3826 (0.4117) acc1: 95.0521 (88.7011) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2549 data: 0.1588 max mem: 3533 +Test: [450/517] eta: 0:00:18 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.3594 (0.4103) acc1: 97.9167 (88.9464) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.3449 data: 0.2490 max mem: 3533 +Test: [460/517] eta: 0:00:15 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.3519 (0.4094) acc1: 99.4792 (89.1207) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.3549 data: 0.2591 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.3460 (0.4078) acc1: 98.9583 (89.3318) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2449 data: 0.1522 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.3444 (0.4065) acc1: 99.7396 (89.5460) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1939 data: 0.1022 max mem: 3533 +Test: [490/517] eta: 0:00:07 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.3504 (0.4052) acc1: 99.4792 (89.6820) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1844 data: 0.0940 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.3889 (0.4059) acc1: 95.0521 (89.6764) acc5: 100.0000 (100.0000) acc1_neg: 79.427 acc5_neg: 100.000 time: 0.1747 data: 0.0847 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.3932 (0.4050) acc1: 95.0521 (89.7831) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1737 data: 0.0839 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.3258 (0.4038) acc1: 97.9167 (89.8885) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1812 data: 0.0816 max mem: 3533 +Test: Total time: 0:02:16 (0.2646 s / it) +* Overall Acc@1 89.888 Acc@5 100.000 + target == real Acc@1 23.668 Acc@5 100.000 + target == fake Acc@1 96.758 Acc@5 100.000 + loss 0.404 + PR-AUC 0.9607 ROC-AUC 0.7544 EER 31.28% (thr=0.6797) +Accuracy of the network on the 198427 test images: 89.9% +[rank0]:[W726 03:35:17.635356108 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 03:35:38.774000 1177581 torch/distributed/run.py:766] +W0726 03:35:38.774000 1177581 torch/distributed/run.py:766] ***************************************** +W0726 03:35:38.774000 1177581 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:35:38.774000 1177581 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:36:07.701037253 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank1]:[W726 03:36:07.716594571 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:36:07.065078682 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +[rank4]:[W726 03:36:08.264976118 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:36:08.269752564 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:36:08.405895070 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:36:08.448211984 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:36:08.449785344 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n64_seed5_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed5_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:39:35 acc1_pos: 61.719 acc5_pos: 100.000 loss: 0.6477 (0.6477) acc1: 61.7188 (61.7188) acc5: 100.0000 (100.0000) time: 6.1069 data: 4.7597 max mem: 3531 +Test: [ 10/389] eta: 0:04:01 acc1_pos: 86.198 acc5_pos: 100.000 loss: 0.5396 (0.5599) acc1: 69.7917 (70.7150) acc5: 100.0000 (100.0000) time: 0.6374 data: 0.4331 max mem: 3531 +Test: [ 20/389] eta: 0:02:22 acc1_pos: 38.021 acc5_pos: 100.000 loss: 0.5396 (0.6048) acc1: 63.5417 (64.2857) acc5: 100.0000 (100.0000) time: 0.1005 data: 0.0080 max mem: 3531 +Test: [ 30/389] eta: 0:01:55 acc1_pos: 80.469 acc5_pos: 100.000 loss: 0.5676 (0.6035) acc1: 62.5000 (66.7759) acc5: 100.0000 (100.0000) time: 0.1499 data: 0.0550 max mem: 3531 +Test: [ 40/389] eta: 0:01:45 acc1_pos: 39.583 acc5_pos: 100.000 loss: 0.5554 (0.6027) acc1: 76.0417 (67.4606) acc5: 100.0000 (100.0000) time: 0.2160 data: 0.1223 max mem: 3531 +Test: [ 50/389] eta: 0:01:33 acc1_pos: 77.344 acc5_pos: 100.000 loss: 0.5409 (0.5945) acc1: 76.0417 (69.0155) acc5: 100.0000 (100.0000) time: 0.2045 data: 0.1088 max mem: 3531 +Test: [ 60/389] eta: 0:01:22 acc1_pos: 90.885 acc5_pos: 100.000 loss: 0.5409 (0.5791) acc1: 76.0417 (70.9699) acc5: 100.0000 (100.0000) time: 0.1449 data: 0.0471 max mem: 3531 +Test: [ 70/389] eta: 0:01:17 acc1_pos: 75.781 acc5_pos: 100.000 loss: 0.4930 (0.5707) acc1: 82.8125 (72.3261) acc5: 100.0000 (100.0000) time: 0.1569 data: 0.0597 max mem: 3531 +Test: [ 80/389] eta: 0:01:12 acc1_pos: 91.406 acc5_pos: 100.000 loss: 0.5288 (0.5751) acc1: 78.1250 (71.8525) acc5: 100.0000 (100.0000) time: 0.1883 data: 0.0937 max mem: 3531 +Test: [ 90/389] eta: 0:01:09 acc1_pos: 72.396 acc5_pos: 100.000 loss: 0.4285 (0.5576) acc1: 85.6771 (73.4375) acc5: 100.0000 (100.0000) time: 0.1934 data: 0.1012 max mem: 3531 +Test: [100/389] eta: 0:01:05 acc1_pos: 73.177 acc5_pos: 100.000 loss: 0.4350 (0.5630) acc1: 89.3229 (72.8831) acc5: 100.0000 (100.0000) time: 0.1953 data: 0.1028 max mem: 3531 +Test: [110/389] eta: 0:01:02 acc1_pos: 94.271 acc5_pos: 100.000 loss: 0.6008 (0.5719) acc1: 68.2292 (71.6873) acc5: 100.0000 (100.0000) time: 0.1844 data: 0.0905 max mem: 3531 +Test: [120/389] eta: 0:00:58 acc1_pos: 75.781 acc5_pos: 100.000 loss: 0.5464 (0.5641) acc1: 75.7812 (72.5228) acc5: 100.0000 (100.0000) time: 0.1743 data: 0.0785 max mem: 3531 +Test: [130/389] eta: 0:00:55 acc1_pos: 49.080 acc5_pos: 100.000 loss: 0.5677 (0.5743) acc1: 75.5208 (71.4456) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.1725 data: 0.0774 max mem: 3531 +Test: [140/389] eta: 0:00:52 acc1_pos: 49.080 acc5_pos: 100.000 loss: 0.6517 (0.5862) acc1: 64.0625 (70.6542) acc5: 100.0000 (100.0000) acc1_neg: 9.635 acc5_neg: 100.000 time: 0.1699 data: 0.0768 max mem: 3531 +Test: [150/389] eta: 0:00:49 acc1_pos: 49.080 acc5_pos: 100.000 loss: 0.6639 (0.5951) acc1: 45.5729 (69.3726) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1662 data: 0.0735 max mem: 3531 +Test: [160/389] eta: 0:00:47 acc1_pos: 49.080 acc5_pos: 100.000 loss: 0.6639 (0.5939) acc1: 53.9062 (69.3630) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1660 data: 0.0732 max mem: 3531 +Test: [170/389] eta: 0:00:44 acc1_pos: 49.080 acc5_pos: 100.000 loss: 0.4531 (0.5969) acc1: 94.2708 (69.3515) acc5: 100.0000 (100.0000) acc1_neg: 23.958 acc5_neg: 100.000 time: 0.1665 data: 0.0747 max mem: 3531 +Test: [180/389] eta: 0:00:42 acc1_pos: 49.080 acc5_pos: 100.000 loss: 0.5080 (0.5999) acc1: 77.3438 (69.0608) acc5: 100.0000 (100.0000) acc1_neg: 63.021 acc5_neg: 100.000 time: 0.1760 data: 0.0844 max mem: 3531 +Test: [190/389] eta: 0:00:39 acc1_pos: 49.080 acc5_pos: 100.000 loss: 0.6816 (0.6077) acc1: 60.6771 (68.0178) acc5: 100.0000 (100.0000) acc1_neg: 2.083 acc5_neg: 100.000 time: 0.1670 data: 0.0749 max mem: 3531 +Test: [200/389] eta: 0:00:37 acc1_pos: 49.080 acc5_pos: 100.000 loss: 0.6758 (0.6138) acc1: 50.5208 (67.2069) acc5: 100.0000 (100.0000) acc1_neg: 48.177 acc5_neg: 100.000 time: 0.1519 data: 0.0595 max mem: 3531 +Test: [210/389] eta: 0:00:35 acc1_pos: 49.080 acc5_pos: 100.000 loss: 0.6085 (0.6074) acc1: 67.4479 (67.8700) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1615 data: 0.0690 max mem: 3532 +Test: [220/389] eta: 0:00:33 acc1_pos: 49.080 acc5_pos: 100.000 loss: 0.3813 (0.5992) acc1: 95.8333 (68.8914) acc5: 100.0000 (100.0000) acc1_neg: 77.344 acc5_neg: 100.000 time: 0.1768 data: 0.0847 max mem: 3532 +Test: [230/389] eta: 0:00:30 acc1_pos: 49.080 acc5_pos: 100.000 loss: 0.5200 (0.6040) acc1: 77.3438 (68.3295) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.1734 data: 0.0820 max mem: 3532 +Test: [240/389] eta: 0:00:28 acc1_pos: 49.080 acc5_pos: 100.000 loss: 0.6303 (0.6053) acc1: 60.9375 (68.1762) acc5: 100.0000 (100.0000) acc1_neg: 55.208 acc5_neg: 100.000 time: 0.1629 data: 0.0716 max mem: 3532 +Test: [250/389] eta: 0:00:26 acc1_pos: 49.080 acc5_pos: 100.000 loss: 0.5122 (0.5996) acc1: 77.8646 (68.8776) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1630 data: 0.0717 max mem: 3532 +Test: [260/389] eta: 0:00:24 acc1_pos: 49.080 acc5_pos: 100.000 loss: 0.4618 (0.5981) acc1: 96.3542 (69.2419) acc5: 100.0000 (100.0000) acc1_neg: 14.844 acc5_neg: 100.000 time: 0.1626 data: 0.0706 max mem: 3532 +Test: [270/389] eta: 0:00:22 acc1_pos: 49.080 acc5_pos: 100.000 loss: 0.6763 (0.6083) acc1: 52.0833 (67.9438) acc5: 100.0000 (100.0000) acc1_neg: 42.448 acc5_neg: 100.000 time: 0.1732 data: 0.0812 max mem: 3532 +Test: [280/389] eta: 0:00:20 acc1_pos: 49.080 acc5_pos: 100.000 loss: 0.8334 (0.6185) acc1: 35.4167 (66.7862) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1800 data: 0.0886 max mem: 3532 +Test: [290/389] eta: 0:00:18 acc1_pos: 49.080 acc5_pos: 100.000 loss: 0.6842 (0.6178) acc1: 43.7500 (66.8090) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.1728 data: 0.0808 max mem: 3532 +Test: [300/389] eta: 0:00:16 acc1_pos: 49.080 acc5_pos: 100.000 loss: 0.6960 (0.6222) acc1: 54.4271 (66.1346) acc5: 100.0000 (100.0000) acc1_neg: 43.750 acc5_neg: 100.000 time: 0.1822 data: 0.0897 max mem: 3532 +Test: [310/389] eta: 0:00:14 acc1_pos: 49.080 acc5_pos: 100.000 loss: 0.7868 (0.6309) acc1: 39.3229 (65.2616) acc5: 100.0000 (100.0000) acc1_neg: 1.823 acc5_neg: 100.000 time: 0.1850 data: 0.0898 max mem: 3532 +Test: [320/389] eta: 0:00:13 acc1_pos: 49.080 acc5_pos: 100.000 loss: 0.7828 (0.6337) acc1: 39.3229 (65.0190) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.1821 data: 0.0879 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 49.080 acc5_pos: 100.000 loss: 0.7893 (0.6383) acc1: 35.9375 (64.1726) acc5: 100.0000 (100.0000) acc1_neg: 45.833 acc5_neg: 100.000 time: 0.1872 data: 0.0957 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 49.080 acc5_pos: 100.000 loss: 0.5419 (0.6312) acc1: 66.9271 (65.0912) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1886 data: 0.0967 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 49.080 acc5_pos: 100.000 loss: 0.4013 (0.6408) acc1: 97.1354 (64.4847) acc5: 100.0000 (100.0000) acc1_neg: 2.604 acc5_neg: 100.000 time: 0.1918 data: 0.1008 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 49.080 acc5_pos: 100.000 loss: 1.0927 (0.6545) acc1: 15.1042 (63.0418) acc5: 100.0000 (100.0000) acc1_neg: 17.188 acc5_neg: 100.000 time: 0.1809 data: 0.0897 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 49.080 acc5_pos: 100.000 loss: 1.0613 (0.6542) acc1: 16.9271 (63.0903) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.1771 data: 0.0860 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 49.080 acc5_pos: 100.000 loss: 0.6410 (0.6549) acc1: 60.4167 (63.0092) acc5: 100.0000 (100.0000) acc1_neg: 40.885 acc5_neg: 100.000 time: 0.1404 data: 0.0503 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 49.080 acc5_pos: 100.000 loss: 0.5906 (0.6529) acc1: 74.2188 (63.1764) acc5: 100.0000 (100.0000) acc1_neg: 82.514 acc5_neg: 100.000 time: 0.1144 data: 0.0209 max mem: 3532 +Test: Total time: 0:01:12 (0.1858 s / it) +* Overall Acc@1 63.176 Acc@5 100.000 + target == real Acc@1 72.014 Acc@5 100.000 + target == fake Acc@1 59.068 Acc@5 100.000 + loss 0.653 + PR-AUC 0.8326 ROC-AUC 0.7039 EER 34.93% (thr=0.4666) +Accuracy of the network on the 149358 test images: 63.2% +[rank0]:[W726 03:37:31.458016990 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:37:51.045000 1178131 torch/distributed/run.py:766] +W0726 03:37:51.045000 1178131 torch/distributed/run.py:766] ***************************************** +W0726 03:37:51.045000 1178131 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:37:51.045000 1178131 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:38:23.360727598 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:38:23.523154009 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:38:23.674522264 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:38:23.714954177 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:38:23.834958206 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:38:23.884648715 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:38:23.914546421 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:38:23.924857518 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n64_seed5_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed5_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:30:25 acc1_pos: 84.375 acc5_pos: 100.000 loss: 0.3623 (0.3623) acc1: 84.3750 (84.3750) acc5: 100.0000 (100.0000) time: 5.4825 data: 3.6995 max mem: 3531 +Test: [ 10/333] eta: 0:03:07 acc1_pos: 85.677 acc5_pos: 100.000 loss: 0.2777 (0.2807) acc1: 87.7604 (88.6364) acc5: 100.0000 (100.0000) time: 0.5808 data: 0.3368 max mem: 3531 +Test: [ 20/333] eta: 0:01:49 acc1_pos: 94.531 acc5_pos: 100.000 loss: 0.2872 (0.2897) acc1: 87.7604 (88.7773) acc5: 100.0000 (100.0000) time: 0.0948 data: 0.0019 max mem: 3531 +Test: [ 30/333] eta: 0:01:33 acc1_pos: 63.281 acc5_pos: 100.000 loss: 0.3698 (0.4204) acc1: 77.8646 (78.3938) acc5: 100.0000 (100.0000) time: 0.1602 data: 0.0635 max mem: 3531 +Test: [ 40/333] eta: 0:01:28 acc1_pos: 56.510 acc5_pos: 100.000 loss: 0.7284 (0.4980) acc1: 55.4688 (71.9004) acc5: 100.0000 (100.0000) time: 0.2499 data: 0.1518 max mem: 3531 +Test: [ 50/333] eta: 0:01:21 acc1_pos: 53.646 acc5_pos: 100.000 loss: 0.7303 (0.5429) acc1: 53.1250 (68.0453) acc5: 100.0000 (100.0000) time: 0.2542 data: 0.1583 max mem: 3531 +Test: [ 60/333] eta: 0:01:14 acc1_pos: 42.969 acc5_pos: 100.000 loss: 0.7144 (0.5723) acc1: 53.1250 (65.6848) acc5: 100.0000 (100.0000) time: 0.2106 data: 0.1139 max mem: 3531 +Test: [ 70/333] eta: 0:01:07 acc1_pos: 49.219 acc5_pos: 100.000 loss: 0.7144 (0.5924) acc1: 53.6458 (64.2569) acc5: 100.0000 (100.0000) time: 0.1755 data: 0.0786 max mem: 3531 +Test: [ 80/333] eta: 0:01:00 acc1_pos: 62.760 acc5_pos: 100.000 loss: 0.7080 (0.6081) acc1: 54.9479 (62.9244) acc5: 100.0000 (100.0000) time: 0.1406 data: 0.0460 max mem: 3531 +Test: [ 90/333] eta: 0:00:56 acc1_pos: 64.323 acc5_pos: 100.000 loss: 0.6570 (0.6114) acc1: 59.1146 (62.7175) acc5: 100.0000 (100.0000) time: 0.1445 data: 0.0508 max mem: 3531 +Test: [100/333] eta: 0:00:53 acc1_pos: 70.312 acc5_pos: 100.000 loss: 0.6388 (0.6168) acc1: 59.8958 (62.4665) acc5: 100.0000 (100.0000) time: 0.1789 data: 0.0870 max mem: 3531 +Test: [110/333] eta: 0:00:49 acc1_pos: 67.448 acc5_pos: 100.000 loss: 0.6638 (0.6202) acc1: 57.0312 (62.1246) acc5: 100.0000 (100.0000) time: 0.1847 data: 0.0929 max mem: 3531 +Test: [120/333] eta: 0:00:46 acc1_pos: 66.667 acc5_pos: 100.000 loss: 0.6299 (0.6224) acc1: 59.3750 (61.9490) acc5: 100.0000 (100.0000) time: 0.1845 data: 0.0899 max mem: 3531 +Test: [130/333] eta: 0:00:44 acc1_pos: 76.562 acc5_pos: 100.000 loss: 0.6094 (0.6226) acc1: 62.5000 (62.0249) acc5: 100.0000 (100.0000) time: 0.1898 data: 0.0918 max mem: 3531 +Test: [140/333] eta: 0:00:41 acc1_pos: 74.479 acc5_pos: 100.000 loss: 0.6142 (0.6225) acc1: 62.2396 (62.0641) acc5: 100.0000 (100.0000) time: 0.1724 data: 0.0760 max mem: 3531 +Test: [150/333] eta: 0:00:39 acc1_pos: 65.365 acc5_pos: 100.000 loss: 0.6388 (0.6252) acc1: 60.4167 (61.9240) acc5: 100.0000 (100.0000) time: 0.1816 data: 0.0893 max mem: 3531 +Test: [160/333] eta: 0:00:36 acc1_pos: 54.276 acc5_pos: 100.000 loss: 0.6759 (0.6349) acc1: 57.0312 (61.3386) acc5: 100.0000 (100.0000) acc1_neg: 47.135 acc5_neg: 100.000 time: 0.2024 data: 0.1099 max mem: 3531 +Test: [170/333] eta: 0:00:34 acc1_pos: 54.276 acc5_pos: 100.000 loss: 0.7679 (0.6461) acc1: 54.6875 (60.7913) acc5: 100.0000 (100.0000) acc1_neg: 54.948 acc5_neg: 100.000 time: 0.2038 data: 0.1103 max mem: 3531 +Test: [180/333] eta: 0:00:32 acc1_pos: 54.276 acc5_pos: 100.000 loss: 0.7546 (0.6495) acc1: 56.5104 (60.7044) acc5: 100.0000 (100.0000) acc1_neg: 55.990 acc5_neg: 100.000 time: 0.1849 data: 0.0887 max mem: 3531 +Test: [190/333] eta: 0:00:30 acc1_pos: 54.276 acc5_pos: 100.000 loss: 0.7104 (0.6521) acc1: 58.3333 (60.7262) acc5: 100.0000 (100.0000) acc1_neg: 61.719 acc5_neg: 100.000 time: 0.2044 data: 0.1054 max mem: 3531 +Test: [200/333] eta: 0:00:28 acc1_pos: 54.276 acc5_pos: 100.000 loss: 0.7115 (0.6540) acc1: 60.1562 (60.8753) acc5: 100.0000 (100.0000) acc1_neg: 61.198 acc5_neg: 100.000 time: 0.2399 data: 0.1416 max mem: 3531 +Test: [210/333] eta: 0:00:25 acc1_pos: 54.276 acc5_pos: 100.000 loss: 0.7270 (0.6597) acc1: 59.8958 (60.6919) acc5: 100.0000 (100.0000) acc1_neg: 50.781 acc5_neg: 100.000 time: 0.1699 data: 0.0713 max mem: 3532 +Test: [220/333] eta: 0:00:23 acc1_pos: 54.276 acc5_pos: 100.000 loss: 0.7763 (0.6673) acc1: 56.7708 (60.2682) acc5: 100.0000 (100.0000) acc1_neg: 70.312 acc5_neg: 100.000 time: 0.1388 data: 0.0429 max mem: 3532 +Test: [230/333] eta: 0:00:21 acc1_pos: 54.276 acc5_pos: 100.000 loss: 0.5402 (0.6604) acc1: 71.6146 (61.0852) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.1910 data: 0.0989 max mem: 3532 +Test: [240/333] eta: 0:00:19 acc1_pos: 54.276 acc5_pos: 100.000 loss: 0.5390 (0.6584) acc1: 76.0417 (61.3449) acc5: 100.0000 (100.0000) acc1_neg: 68.229 acc5_neg: 100.000 time: 0.2110 data: 0.1177 max mem: 3532 +Test: [250/333] eta: 0:00:17 acc1_pos: 54.276 acc5_pos: 100.000 loss: 0.6335 (0.6601) acc1: 64.5833 (61.3650) acc5: 100.0000 (100.0000) acc1_neg: 47.917 acc5_neg: 100.000 time: 0.1988 data: 0.1015 max mem: 3532 +Test: [260/333] eta: 0:00:14 acc1_pos: 54.276 acc5_pos: 100.000 loss: 0.7013 (0.6637) acc1: 60.1562 (61.2159) acc5: 100.0000 (100.0000) acc1_neg: 60.156 acc5_neg: 100.000 time: 0.1449 data: 0.0459 max mem: 3532 +Test: [270/333] eta: 0:00:12 acc1_pos: 54.276 acc5_pos: 100.000 loss: 0.6718 (0.6627) acc1: 62.2396 (61.4237) acc5: 100.0000 (100.0000) acc1_neg: 66.406 acc5_neg: 100.000 time: 0.1271 data: 0.0305 max mem: 3532 +Test: [280/333] eta: 0:00:10 acc1_pos: 54.276 acc5_pos: 100.000 loss: 0.6407 (0.6658) acc1: 64.0625 (61.2081) acc5: 100.0000 (100.0000) acc1_neg: 60.938 acc5_neg: 100.000 time: 0.1600 data: 0.0659 max mem: 3532 +Test: [290/333] eta: 0:00:08 acc1_pos: 54.276 acc5_pos: 100.000 loss: 0.8318 (0.6725) acc1: 47.1354 (60.6995) acc5: 100.0000 (100.0000) acc1_neg: 50.781 acc5_neg: 100.000 time: 0.1877 data: 0.0930 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 54.276 acc5_pos: 100.000 loss: 0.8643 (0.6785) acc1: 47.1354 (60.3085) acc5: 100.0000 (100.0000) acc1_neg: 54.427 acc5_neg: 100.000 time: 0.2048 data: 0.1075 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 54.276 acc5_pos: 100.000 loss: 0.7661 (0.6771) acc1: 54.4271 (60.5071) acc5: 100.0000 (100.0000) acc1_neg: 67.708 acc5_neg: 100.000 time: 0.1849 data: 0.0878 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 54.276 acc5_pos: 100.000 loss: 0.6092 (0.6763) acc1: 67.7083 (60.6828) acc5: 100.0000 (100.0000) acc1_neg: 44.531 acc5_neg: 100.000 time: 0.1689 data: 0.0717 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 54.276 acc5_pos: 100.000 loss: 0.8715 (0.6845) acc1: 46.8750 (60.1877) acc5: 100.0000 (100.0000) acc1_neg: 45.312 acc5_neg: 100.000 time: 0.1520 data: 0.0571 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 54.276 acc5_pos: 100.000 loss: 0.9123 (0.6860) acc1: 45.3125 (60.1050) acc5: 100.0000 (100.0000) acc1_neg: 44.789 acc5_neg: 100.000 time: 0.1551 data: 0.0570 max mem: 3532 +Test: Total time: 0:01:05 (0.1953 s / it) +* Overall Acc@1 60.105 Acc@5 100.000 + target == real Acc@1 61.851 Acc@5 100.000 + target == fake Acc@1 58.551 Acc@5 100.000 + loss 0.686 + PR-AUC 0.6567 ROC-AUC 0.6442 EER 39.86% (thr=0.4897) +Accuracy of the network on the 127843 test images: 60.1% +[rank0]:[W726 03:39:42.699083377 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:40:02.665000 1178694 torch/distributed/run.py:766] +W0726 03:40:02.665000 1178694 torch/distributed/run.py:766] ***************************************** +W0726 03:40:02.665000 1178694 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:40:02.665000 1178694 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:40:32.260934348 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:40:32.375852556 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:40:32.537547175 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:40:32.847463096 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:40:32.890138594 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:40:32.926130753 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:40:32.965547723 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:40:32.996147613 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n64_seed5_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed5_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:20:10 acc1_pos: 97.135 acc5_pos: 100.000 loss: 0.1986 (0.1986) acc1: 97.1354 (97.1354) acc5: 100.0000 (100.0000) time: 5.1514 data: 3.4174 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/235] eta: 0:02:03 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.2071 (0.2069) acc1: 96.3542 (96.1885) acc5: 100.0000 (100.0000) time: 0.5505 data: 0.3111 max mem: 3531 +Test: [ 20/235] eta: 0:01:11 acc1_pos: 96.615 acc5_pos: 100.000 loss: 0.2002 (0.2031) acc1: 96.0938 (96.3666) acc5: 100.0000 (100.0000) time: 0.0907 data: 0.0005 max mem: 3531 +Test: [ 30/235] eta: 0:01:01 acc1_pos: 97.135 acc5_pos: 100.000 loss: 0.1971 (0.2016) acc1: 96.0938 (96.3374) acc5: 100.0000 (100.0000) time: 0.1652 data: 0.0715 max mem: 3531 +Test: [ 40/235] eta: 0:01:00 acc1_pos: 95.052 acc5_pos: 100.000 loss: 0.1970 (0.2011) acc1: 96.6146 (96.3986) acc5: 100.0000 (100.0000) time: 0.2880 data: 0.1922 max mem: 3531 +Test: [ 50/235] eta: 0:00:53 acc1_pos: 94.531 acc5_pos: 100.000 loss: 0.2018 (0.2022) acc1: 96.6146 (96.3286) acc5: 100.0000 (100.0000) time: 0.2699 data: 0.1755 max mem: 3531 +Test: [ 60/235] eta: 0:00:49 acc1_pos: 96.094 acc5_pos: 100.000 loss: 0.2099 (0.2035) acc1: 96.0938 (96.2645) acc5: 100.0000 (100.0000) time: 0.2189 data: 0.1231 max mem: 3531 +Test: [ 70/235] eta: 0:00:45 acc1_pos: 98.462 acc5_pos: 100.000 loss: 0.2209 (0.3520) acc1: 94.0104 (86.7628) acc5: 100.0000 (100.0000) acc1_neg: 17.969 acc5_neg: 100.000 time: 0.2402 data: 0.1447 max mem: 3531 +Test: [ 80/235] eta: 0:00:42 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4031 (0.4820) acc1: 20.3125 (78.4883) acc5: 100.0000 (100.0000) acc1_neg: 16.927 acc5_neg: 100.000 time: 0.2576 data: 0.1631 max mem: 3531 +Test: [ 90/235] eta: 0:00:37 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4044 (0.5827) acc1: 19.7917 (72.0667) acc5: 100.0000 (100.0000) acc1_neg: 19.531 acc5_neg: 100.000 time: 0.2021 data: 0.1050 max mem: 3531 +Test: [100/235] eta: 0:00:34 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.3995 (0.6621) acc1: 20.3125 (67.0173) acc5: 100.0000 (100.0000) acc1_neg: 19.531 acc5_neg: 100.000 time: 0.1761 data: 0.0808 max mem: 3531 +Test: [110/235] eta: 0:00:31 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.3752 (0.7275) acc1: 20.3125 (62.8543) acc5: 100.0000 (100.0000) acc1_neg: 22.656 acc5_neg: 100.000 time: 0.2224 data: 0.1303 max mem: 3531 +Test: [120/235] eta: 0:00:28 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4507 (0.7963) acc1: 17.7083 (58.7078) acc5: 100.0000 (100.0000) acc1_neg: 11.719 acc5_neg: 100.000 time: 0.2213 data: 0.1295 max mem: 3531 +Test: [130/235] eta: 0:00:26 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.5648 (0.8553) acc1: 12.5000 (55.1805) acc5: 100.0000 (100.0000) acc1_neg: 12.760 acc5_neg: 100.000 time: 0.2251 data: 0.1339 max mem: 3531 +Test: [140/235] eta: 0:00:23 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.5943 (0.9086) acc1: 11.7188 (52.0852) acc5: 100.0000 (100.0000) acc1_neg: 9.896 acc5_neg: 100.000 time: 0.2123 data: 0.1185 max mem: 3531 +Test: [150/235] eta: 0:00:20 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.5925 (0.9528) acc1: 11.7188 (49.4430) acc5: 100.0000 (100.0000) acc1_neg: 13.281 acc5_neg: 100.000 time: 0.2163 data: 0.1197 max mem: 3531 +Test: [160/235] eta: 0:00:17 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.5614 (0.9914) acc1: 12.2396 (47.1306) acc5: 100.0000 (100.0000) acc1_neg: 12.240 acc5_neg: 100.000 time: 0.1975 data: 0.1035 max mem: 3531 +Test: [170/235] eta: 0:00:15 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.5112 (1.0024) acc1: 13.8021 (46.0785) acc5: 100.0000 (100.0000) acc1_neg: 28.125 acc5_neg: 100.000 time: 0.2000 data: 0.1080 max mem: 3531 +Test: [180/235] eta: 0:00:12 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.1403 (1.0093) acc1: 30.9896 (45.2578) acc5: 100.0000 (100.0000) acc1_neg: 30.990 acc5_neg: 100.000 time: 0.2189 data: 0.1271 max mem: 3531 +Test: [190/235] eta: 0:00:10 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.1073 (1.0152) acc1: 32.0312 (44.5735) acc5: 100.0000 (100.0000) acc1_neg: 31.771 acc5_neg: 100.000 time: 0.2009 data: 0.1080 max mem: 3531 +Test: [200/235] eta: 0:00:08 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.1250 (1.0213) acc1: 31.7708 (43.8536) acc5: 100.0000 (100.0000) acc1_neg: 29.688 acc5_neg: 100.000 time: 0.1861 data: 0.0923 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.1360 (1.0259) acc1: 29.6875 (43.2909) acc5: 100.0000 (100.0000) acc1_neg: 30.469 acc5_neg: 100.000 time: 0.1478 data: 0.0563 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.1199 (1.0302) acc1: 31.5104 (42.7873) acc5: 100.0000 (100.0000) acc1_neg: 29.427 acc5_neg: 100.000 time: 0.1300 data: 0.0397 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.1146 (1.0337) acc1: 31.7708 (42.3510) acc5: 100.0000 (100.0000) acc1_neg: 30.729 acc5_neg: 100.000 time: 0.1251 data: 0.0350 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.1146 (1.0361) acc1: 31.5104 (42.2079) acc5: 100.0000 (100.0000) acc1_neg: 30.435 acc5_neg: 100.000 time: 0.1199 data: 0.0331 max mem: 3532 +Test: Total time: 0:00:50 (0.2166 s / it) +* Overall Acc@1 42.208 Acc@5 100.000 + target == real Acc@1 96.264 Acc@5 100.000 + target == fake Acc@1 22.657 Acc@5 100.000 + loss 1.036 + PR-AUC 0.8885 ROC-AUC 0.7522 EER 31.34% (thr=0.1796) +Accuracy of the network on the 89879 test images: 42.2% +[rank0]:[W726 03:41:30.042401294 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:41:51.397000 1179253 torch/distributed/run.py:766] +W0726 03:41:51.397000 1179253 torch/distributed/run.py:766] ***************************************** +W0726 03:41:51.397000 1179253 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:41:51.397000 1179253 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:42:19.157231222 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:42:20.795269069 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:42:20.051391963 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:42:20.120991736 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:42:21.422294002 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank1]:[W726 03:42:21.423865655 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:42:21.545812417 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:42:21.627761687 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n64_seed5_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed5_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:25:19 acc1_pos: 63.021 acc5_pos: 100.000 loss: 0.6400 (0.6400) acc1: 63.0208 (63.0208) acc5: 100.0000 (100.0000) time: 5.6482 data: 4.0371 max mem: 3531 +Test: [ 10/269] eta: 0:02:34 acc1_pos: 86.719 acc5_pos: 100.000 loss: 0.5314 (0.5315) acc1: 77.0833 (76.9176) acc5: 100.0000 (100.0000) time: 0.5956 data: 0.3674 max mem: 3531 +Test: [ 20/269] eta: 0:01:29 acc1_pos: 85.417 acc5_pos: 100.000 loss: 0.4842 (0.5108) acc1: 79.6875 (79.7371) acc5: 100.0000 (100.0000) time: 0.0932 data: 0.0004 max mem: 3531 +Test: [ 30/269] eta: 0:01:11 acc1_pos: 77.083 acc5_pos: 100.000 loss: 0.4795 (0.5048) acc1: 82.8125 (80.7628) acc5: 100.0000 (100.0000) time: 0.1336 data: 0.0388 max mem: 3531 +Test: [ 40/269] eta: 0:01:04 acc1_pos: 86.207 acc5_pos: 100.000 loss: 0.5495 (0.5319) acc1: 76.8229 (77.6740) acc5: 100.0000 (100.0000) acc1_neg: 55.469 acc5_neg: 100.000 time: 0.2049 data: 0.1125 max mem: 3531 +Test: [ 50/269] eta: 0:00:59 acc1_pos: 86.207 acc5_pos: 100.000 loss: 0.6206 (0.5604) acc1: 64.5833 (74.0605) acc5: 100.0000 (100.0000) acc1_neg: 42.188 acc5_neg: 100.000 time: 0.2267 data: 0.1347 max mem: 3531 +Test: [ 60/269] eta: 0:00:55 acc1_pos: 86.207 acc5_pos: 100.000 loss: 0.6700 (0.5870) acc1: 58.8542 (70.5473) acc5: 100.0000 (100.0000) acc1_neg: 40.104 acc5_neg: 100.000 time: 0.2253 data: 0.1331 max mem: 3531 +Test: [ 70/269] eta: 0:00:51 acc1_pos: 86.207 acc5_pos: 100.000 loss: 0.6815 (0.6013) acc1: 54.6875 (68.5263) acc5: 100.0000 (100.0000) acc1_neg: 63.542 acc5_neg: 100.000 time: 0.2345 data: 0.1438 max mem: 3531 +Test: [ 80/269] eta: 0:00:47 acc1_pos: 86.207 acc5_pos: 100.000 loss: 0.6551 (0.6090) acc1: 57.5521 (67.4865) acc5: 100.0000 (100.0000) acc1_neg: 52.604 acc5_neg: 100.000 time: 0.2054 data: 0.1135 max mem: 3531 +Test: [ 90/269] eta: 0:00:42 acc1_pos: 86.207 acc5_pos: 100.000 loss: 0.7548 (0.6348) acc1: 44.2708 (64.0196) acc5: 100.0000 (100.0000) acc1_neg: 48.177 acc5_neg: 100.000 time: 0.1691 data: 0.0770 max mem: 3531 +Test: [100/269] eta: 0:00:39 acc1_pos: 86.207 acc5_pos: 100.000 loss: 0.8301 (0.6498) acc1: 36.4583 (62.1468) acc5: 100.0000 (100.0000) acc1_neg: 36.198 acc5_neg: 100.000 time: 0.1739 data: 0.0826 max mem: 3531 +Test: [110/269] eta: 0:00:37 acc1_pos: 86.207 acc5_pos: 100.000 loss: 0.8015 (0.6614) acc1: 39.3229 (60.3369) acc5: 100.0000 (100.0000) acc1_neg: 32.031 acc5_neg: 100.000 time: 0.2048 data: 0.1121 max mem: 3531 +Test: [120/269] eta: 0:00:35 acc1_pos: 86.207 acc5_pos: 100.000 loss: 0.7221 (0.6639) acc1: 51.0417 (59.9733) acc5: 100.0000 (100.0000) acc1_neg: 51.562 acc5_neg: 100.000 time: 0.2452 data: 0.1525 max mem: 3531 +Test: [130/269] eta: 0:00:32 acc1_pos: 86.207 acc5_pos: 100.000 loss: 0.7453 (0.6778) acc1: 44.5312 (58.3393) acc5: 100.0000 (100.0000) acc1_neg: 33.854 acc5_neg: 100.000 time: 0.2508 data: 0.1585 max mem: 3531 +Test: [140/269] eta: 0:00:30 acc1_pos: 86.207 acc5_pos: 100.000 loss: 0.8310 (0.6884) acc1: 37.7604 (56.8964) acc5: 100.0000 (100.0000) acc1_neg: 34.115 acc5_neg: 100.000 time: 0.2196 data: 0.1262 max mem: 3531 +Test: [150/269] eta: 0:00:27 acc1_pos: 86.207 acc5_pos: 100.000 loss: 0.7980 (0.6911) acc1: 44.2708 (56.5794) acc5: 100.0000 (100.0000) acc1_neg: 61.198 acc5_neg: 100.000 time: 0.2278 data: 0.1325 max mem: 3531 +Test: [160/269] eta: 0:00:25 acc1_pos: 86.207 acc5_pos: 100.000 loss: 0.7372 (0.6924) acc1: 47.3958 (56.3584) acc5: 100.0000 (100.0000) acc1_neg: 45.833 acc5_neg: 100.000 time: 0.2400 data: 0.1437 max mem: 3531 +Test: [170/269] eta: 0:00:22 acc1_pos: 86.207 acc5_pos: 100.000 loss: 0.7544 (0.6986) acc1: 45.8333 (55.4901) acc5: 100.0000 (100.0000) acc1_neg: 38.542 acc5_neg: 100.000 time: 0.1965 data: 0.1002 max mem: 3531 +Test: [180/269] eta: 0:00:20 acc1_pos: 86.207 acc5_pos: 100.000 loss: 0.7316 (0.6992) acc1: 47.1354 (55.4342) acc5: 100.0000 (100.0000) acc1_neg: 43.229 acc5_neg: 100.000 time: 0.1703 data: 0.0744 max mem: 3531 +Test: [190/269] eta: 0:00:17 acc1_pos: 86.207 acc5_pos: 100.000 loss: 0.7159 (0.7031) acc1: 51.0417 (54.8348) acc5: 100.0000 (100.0000) acc1_neg: 57.031 acc5_neg: 100.000 time: 0.1789 data: 0.0850 max mem: 3531 +Test: [200/269] eta: 0:00:15 acc1_pos: 86.207 acc5_pos: 100.000 loss: 0.7731 (0.7082) acc1: 45.5729 (54.3804) acc5: 100.0000 (100.0000) acc1_neg: 52.865 acc5_neg: 100.000 time: 0.1943 data: 0.1019 max mem: 3531 +Test: [210/269] eta: 0:00:13 acc1_pos: 86.207 acc5_pos: 100.000 loss: 0.7713 (0.7091) acc1: 45.8333 (54.1827) acc5: 100.0000 (100.0000) acc1_neg: 50.521 acc5_neg: 100.000 time: 0.2043 data: 0.1113 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 86.207 acc5_pos: 100.000 loss: 0.7103 (0.7106) acc1: 47.3958 (54.0088) acc5: 100.0000 (100.0000) acc1_neg: 72.917 acc5_neg: 100.000 time: 0.1803 data: 0.0879 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 86.207 acc5_pos: 100.000 loss: 0.6645 (0.7086) acc1: 57.5521 (54.2636) acc5: 100.0000 (100.0000) acc1_neg: 58.073 acc5_neg: 100.000 time: 0.1799 data: 0.0884 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 86.207 acc5_pos: 100.000 loss: 0.7549 (0.7130) acc1: 48.9583 (53.7517) acc5: 100.0000 (100.0000) acc1_neg: 48.698 acc5_neg: 100.000 time: 0.1998 data: 0.1070 max mem: 3532 +Test: [250/269] eta: 0:00:04 acc1_pos: 86.207 acc5_pos: 100.000 loss: 0.7925 (0.7146) acc1: 40.8854 (53.5006) acc5: 100.0000 (100.0000) acc1_neg: 19.531 acc5_neg: 100.000 time: 0.1779 data: 0.0848 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 86.207 acc5_pos: 100.000 loss: 0.8166 (0.7193) acc1: 35.9375 (52.8536) acc5: 100.0000 (100.0000) acc1_neg: 46.615 acc5_neg: 100.000 time: 0.1467 data: 0.0558 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 86.207 acc5_pos: 100.000 loss: 0.7729 (0.7198) acc1: 39.3229 (52.7015) acc5: 100.0000 (100.0000) acc1_neg: 54.331 acc5_neg: 100.000 time: 0.1395 data: 0.0458 max mem: 3532 +Test: Total time: 0:00:56 (0.2117 s / it) +* Overall Acc@1 52.702 Acc@5 100.000 + target == real Acc@1 80.498 Acc@5 100.000 + target == fake Acc@1 48.908 Acc@5 100.000 + loss 0.720 + PR-AUC 0.9443 ROC-AUC 0.7158 EER 34.73% (thr=0.4333) +Accuracy of the network on the 103293 test images: 52.7% +[rank0]:[W726 03:43:27.048470192 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T03:43:34Z] eval_done tag=fixed500k_fake_n64_seed5_resampled +[2026-07-26T03:43:34Z] eval_start tag=fixed500k_real_n32_seed6_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n32_classseed6_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T03:43:34Z +hostname=n116-040-213 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n32_classseed6_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n32_seed6_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 03:43:47.483000 1179829 torch/distributed/run.py:766] +W0726 03:43:47.483000 1179829 torch/distributed/run.py:766] ***************************************** +W0726 03:43:47.483000 1179829 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:43:47.483000 1179829 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:44:18.182871509 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:44:18.963463077 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:44:18.071993348 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:44:19.445673550 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:44:19.521521991 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:44:19.568720080 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W726 03:44:19.570674115 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +NCCL version 2.26.2+cuda12.2 +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:44:19.678086195 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n32_seed6_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n32_classseed6_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:24:46 acc1_pos: 72.135 acc5_pos: 100.000 loss: 0.5488 (0.5488) acc1: 72.1354 (72.1354) acc5: 100.0000 (100.0000) time: 6.3527 data: 4.8115 max mem: 3531 +Test: [ 10/234] eta: 0:02:31 acc1_pos: 48.698 acc5_pos: 100.000 loss: 0.6014 (0.5908) acc1: 69.5312 (69.0814) acc5: 100.0000 (100.0000) time: 0.6785 data: 0.4537 max mem: 3531 +Test: [ 20/234] eta: 0:01:25 acc1_pos: 92.708 acc5_pos: 100.000 loss: 0.5765 (0.5700) acc1: 69.5312 (71.0441) acc5: 100.0000 (100.0000) time: 0.1022 data: 0.0092 max mem: 3531 +Test: [ 30/234] eta: 0:01:16 acc1_pos: 73.438 acc5_pos: 100.000 loss: 0.3167 (0.4893) acc1: 90.3646 (77.0497) acc5: 100.0000 (100.0000) time: 0.2117 data: 0.1184 max mem: 3531 +Test: [ 40/234] eta: 0:01:09 acc1_pos: 99.219 acc5_pos: 100.000 loss: 0.3073 (0.4752) acc1: 91.4062 (77.3374) acc5: 100.0000 (100.0000) time: 0.3169 data: 0.2236 max mem: 3531 +Test: [ 50/234] eta: 0:00:58 acc1_pos: 98.698 acc5_pos: 100.000 loss: 0.1403 (0.4085) acc1: 98.1771 (81.4951) acc5: 100.0000 (100.0000) time: 0.2313 data: 0.1378 max mem: 3531 +Test: [ 60/234] eta: 0:00:51 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.1217 (0.3589) acc1: 98.9583 (84.5116) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1682 data: 0.0761 max mem: 3531 +Test: [ 70/234] eta: 0:00:47 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.1062 (0.3678) acc1: 100.0000 (83.8982) acc5: 100.0000 (100.0000) acc1_neg: 70.312 acc5_neg: 100.000 time: 0.2142 data: 0.1220 max mem: 3531 +Test: [ 80/234] eta: 0:00:43 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.5343 (0.3775) acc1: 72.3958 (83.5487) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2267 data: 0.1342 max mem: 3531 +Test: [ 90/234] eta: 0:00:39 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.2273 (0.3552) acc1: 95.0521 (84.9731) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.2065 data: 0.1154 max mem: 3531 +Test: [100/234] eta: 0:00:35 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.2273 (0.3581) acc1: 94.0104 (84.7721) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2079 data: 0.1157 max mem: 3531 +Test: [110/234] eta: 0:00:32 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.3810 (0.3600) acc1: 83.0729 (84.7011) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.2177 data: 0.1233 max mem: 3531 +Test: [120/234] eta: 0:00:29 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.3810 (0.3759) acc1: 83.5938 (83.8477) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.2018 data: 0.1076 max mem: 3531 +Test: [130/234] eta: 0:00:26 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4677 (0.3842) acc1: 79.6875 (83.4168) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.1914 data: 0.0962 max mem: 3531 +Test: [140/234] eta: 0:00:23 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4920 (0.4052) acc1: 77.0833 (82.3951) acc5: 100.0000 (100.0000) acc1_neg: 37.240 acc5_neg: 100.000 time: 0.2057 data: 0.1107 max mem: 3531 +Test: [150/234] eta: 0:00:20 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4597 (0.4135) acc1: 77.3438 (81.9398) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.2192 data: 0.1267 max mem: 3531 +Test: [160/234] eta: 0:00:18 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4597 (0.4187) acc1: 77.3438 (81.5978) acc5: 100.0000 (100.0000) acc1_neg: 62.760 acc5_neg: 100.000 time: 0.2192 data: 0.1261 max mem: 3531 +Test: [170/234] eta: 0:00:15 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.6083 (0.4294) acc1: 66.4062 (80.8388) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.1950 data: 0.1029 max mem: 3531 +Test: [180/234] eta: 0:00:12 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.5843 (0.4414) acc1: 66.9271 (80.0544) acc5: 100.0000 (100.0000) acc1_neg: 31.771 acc5_neg: 100.000 time: 0.1637 data: 0.0733 max mem: 3531 +Test: [190/234] eta: 0:00:10 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.6890 (0.4565) acc1: 60.9375 (78.9622) acc5: 100.0000 (100.0000) acc1_neg: 61.458 acc5_neg: 100.000 time: 0.1440 data: 0.0513 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.6890 (0.4500) acc1: 60.9375 (79.2846) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1500 data: 0.0564 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.1109 (0.4346) acc1: 99.7396 (80.2355) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1474 data: 0.0561 max mem: 3532 +Test: [220/234] eta: 0:00:03 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.1472 (0.4226) acc1: 98.6979 (80.9978) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.1316 data: 0.0413 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.3228 (0.4344) acc1: 60.9375 (80.0032) acc5: 100.0000 (100.0000) acc1_neg: 58.854 acc5_neg: 100.000 time: 0.1200 data: 0.0299 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.6835 (0.4376) acc1: 59.1146 (79.7501) acc5: 100.0000 (100.0000) acc1_neg: 57.198 acc5_neg: 100.000 time: 0.1178 data: 0.0267 max mem: 3532 +Test: Total time: 0:00:49 (0.2131 s / it) +* Overall Acc@1 79.750 Acc@5 100.000 + target == real Acc@1 81.616 Acc@5 100.000 + target == fake Acc@1 79.224 Acc@5 100.000 + loss 0.438 + PR-AUC 0.9671 ROC-AUC 0.8883 EER 19.82% (thr=0.4858) +Accuracy of the network on the 89729 test images: 79.8% +[rank0]:[W726 03:45:21.237410866 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:45:41.062000 1180375 torch/distributed/run.py:766] +W0726 03:45:41.062000 1180375 torch/distributed/run.py:766] ***************************************** +W0726 03:45:41.062000 1180375 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:45:41.062000 1180375 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:46:10.239532088 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:46:10.753735670 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:46:10.813734438 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:46:10.838110234 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:46:10.876594885 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:46:10.988553149 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:46:10.021359586 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:46:10.099738339 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n32_seed6_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n32_classseed6_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:23:57 acc1_pos: 60.677 acc5_pos: 100.000 loss: 0.6390 (0.6390) acc1: 60.6771 (60.6771) acc5: 100.0000 (100.0000) time: 6.6347 data: 5.1905 max mem: 3531 +Test: [ 10/2387] eta: 0:27:09 acc1_pos: 33.854 acc5_pos: 100.000 loss: 0.7790 (0.7502) acc1: 51.0417 (51.6809) acc5: 100.0000 (100.0000) time: 0.6854 data: 0.4723 max mem: 3531 +Test: [ 20/2387] eta: 0:15:53 acc1_pos: 63.542 acc5_pos: 100.000 loss: 0.7770 (0.7568) acc1: 49.7396 (51.6121) acc5: 100.0000 (100.0000) time: 0.0914 data: 0.0005 max mem: 3531 +Test: [ 30/2387] eta: 0:13:41 acc1_pos: 38.542 acc5_pos: 100.000 loss: 0.7581 (0.7578) acc1: 52.3438 (52.2345) acc5: 100.0000 (100.0000) time: 0.1635 data: 0.0714 max mem: 3531 +Test: [ 40/2387] eta: 0:13:26 acc1_pos: 53.385 acc5_pos: 100.000 loss: 0.7581 (0.7630) acc1: 50.2604 (51.2830) acc5: 100.0000 (100.0000) time: 0.2812 data: 0.1852 max mem: 3531 +Test: [ 50/2387] eta: 0:13:02 acc1_pos: 49.219 acc5_pos: 100.000 loss: 0.7526 (0.7634) acc1: 49.2188 (51.0774) acc5: 100.0000 (100.0000) time: 0.3137 data: 0.2153 max mem: 3531 +Test: [ 60/2387] eta: 0:12:14 acc1_pos: 65.365 acc5_pos: 100.000 loss: 0.7997 (0.7728) acc1: 47.1354 (50.2348) acc5: 100.0000 (100.0000) time: 0.2586 data: 0.1623 max mem: 3531 +Test: [ 70/2387] eta: 0:11:37 acc1_pos: 35.677 acc5_pos: 100.000 loss: 0.8611 (0.7905) acc1: 39.0625 (48.2138) acc5: 100.0000 (100.0000) time: 0.2143 data: 0.1190 max mem: 3531 +Test: [ 80/2387] eta: 0:11:13 acc1_pos: 76.562 acc5_pos: 100.000 loss: 0.8041 (0.7753) acc1: 43.7500 (49.8553) acc5: 100.0000 (100.0000) time: 0.2199 data: 0.1244 max mem: 3531 +Test: [ 90/2387] eta: 0:10:55 acc1_pos: 65.365 acc5_pos: 100.000 loss: 0.6040 (0.7525) acc1: 65.3646 (52.1606) acc5: 100.0000 (100.0000) time: 0.2295 data: 0.1328 max mem: 3531 +Test: [ 100/2387] eta: 0:10:42 acc1_pos: 70.833 acc5_pos: 100.000 loss: 0.5971 (0.7434) acc1: 65.3646 (53.3132) acc5: 100.0000 (100.0000) time: 0.2350 data: 0.1388 max mem: 3531 +Test: [ 110/2387] eta: 0:10:30 acc1_pos: 59.375 acc5_pos: 100.000 loss: 0.6192 (0.7308) acc1: 64.3229 (54.6218) acc5: 100.0000 (100.0000) time: 0.2401 data: 0.1453 max mem: 3531 +Test: [ 120/2387] eta: 0:10:23 acc1_pos: 56.510 acc5_pos: 100.000 loss: 0.6308 (0.7244) acc1: 62.2396 (55.1761) acc5: 100.0000 (100.0000) time: 0.2456 data: 0.1504 max mem: 3531 +Test: [ 130/2387] eta: 0:10:24 acc1_pos: 65.625 acc5_pos: 100.000 loss: 0.6039 (0.7155) acc1: 63.5417 (56.1009) acc5: 100.0000 (100.0000) time: 0.2747 data: 0.1789 max mem: 3531 +Test: [ 140/2387] eta: 0:10:20 acc1_pos: 76.042 acc5_pos: 100.000 loss: 0.5928 (0.7097) acc1: 65.6250 (56.5603) acc5: 100.0000 (100.0000) time: 0.2841 data: 0.1882 max mem: 3531 +Test: [ 150/2387] eta: 0:10:06 acc1_pos: 59.375 acc5_pos: 100.000 loss: 0.5885 (0.7010) acc1: 66.1458 (57.4797) acc5: 100.0000 (100.0000) time: 0.2355 data: 0.1390 max mem: 3531 +Test: [ 160/2387] eta: 0:09:54 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.5540 (0.6871) acc1: 72.9167 (58.8040) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2006 data: 0.1045 max mem: 3531 +Test: [ 170/2387] eta: 0:09:40 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2833 (0.6628) acc1: 94.0104 (60.9284) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1925 data: 0.0960 max mem: 3531 +Test: [ 180/2387] eta: 0:09:30 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2674 (0.6411) acc1: 94.5312 (62.7993) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1907 data: 0.0948 max mem: 3531 +Test: [ 190/2387] eta: 0:09:18 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2641 (0.6215) acc1: 94.7917 (64.4688) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1879 data: 0.0937 max mem: 3531 +Test: [ 200/2387] eta: 0:09:12 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2625 (0.6039) acc1: 94.7917 (65.9761) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2006 data: 0.1068 max mem: 3531 +Test: [ 210/2387] eta: 0:09:01 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2684 (0.5882) acc1: 95.3125 (67.3677) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1964 data: 0.1019 max mem: 3532 +Test: [ 220/2387] eta: 0:08:49 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2741 (0.5740) acc1: 94.7917 (68.5992) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1588 data: 0.0649 max mem: 3532 +Test: [ 230/2387] eta: 0:08:46 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2741 (0.5610) acc1: 94.5312 (69.7173) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1921 data: 0.0990 max mem: 3532 +Test: [ 240/2387] eta: 0:08:38 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2872 (0.5513) acc1: 93.4896 (70.6669) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.2139 data: 0.1173 max mem: 3532 +Test: [ 250/2387] eta: 0:08:33 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2562 (0.5346) acc1: 93.7500 (71.7910) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1999 data: 0.1032 max mem: 3532 +Test: [ 260/2387] eta: 0:08:23 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1056 (0.5179) acc1: 100.0000 (72.8718) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1765 data: 0.0805 max mem: 3532 +Test: [ 270/2387] eta: 0:08:17 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.0983 (0.5024) acc1: 100.0000 (73.8718) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1660 data: 0.0710 max mem: 3532 +Test: [ 280/2387] eta: 0:08:11 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.0987 (0.4881) acc1: 100.0000 (74.8017) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1902 data: 0.0982 max mem: 3532 +Test: [ 290/2387] eta: 0:08:06 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.0923 (0.4745) acc1: 100.0000 (75.6676) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1938 data: 0.0977 max mem: 3532 +Test: [ 300/2387] eta: 0:08:00 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.0914 (0.4618) acc1: 100.0000 (76.4760) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1850 data: 0.0889 max mem: 3532 +Test: [ 310/2387] eta: 0:07:58 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.0962 (0.4524) acc1: 100.0000 (77.2140) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2020 data: 0.1105 max mem: 3532 +Test: [ 320/2387] eta: 0:07:57 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2494 (0.4462) acc1: 99.7396 (77.9076) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2445 data: 0.1521 max mem: 3532 +Test: [ 330/2387] eta: 0:07:55 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2507 (0.4403) acc1: 99.7396 (78.5601) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2480 data: 0.1541 max mem: 3532 +Test: [ 340/2387] eta: 0:07:52 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2501 (0.4347) acc1: 99.4792 (79.1713) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2302 data: 0.1339 max mem: 3532 +Test: [ 350/2387] eta: 0:07:47 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2426 (0.4267) acc1: 99.7396 (79.7639) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2047 data: 0.1061 max mem: 3532 +Test: [ 360/2387] eta: 0:07:39 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1715 (0.4209) acc1: 99.7396 (80.2552) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1537 data: 0.0548 max mem: 3532 +Test: [ 370/2387] eta: 0:07:34 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2225 (0.4156) acc1: 97.1354 (80.7116) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1516 data: 0.0543 max mem: 3532 +Test: [ 380/2387] eta: 0:07:31 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2237 (0.4105) acc1: 97.1354 (81.1468) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1899 data: 0.0961 max mem: 3532 +Test: [ 390/2387] eta: 0:07:25 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2231 (0.4057) acc1: 97.1354 (81.5517) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1749 data: 0.0827 max mem: 3532 +Test: [ 400/2387] eta: 0:07:21 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2231 (0.4012) acc1: 96.8750 (81.9423) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1730 data: 0.0801 max mem: 3532 +Test: [ 410/2387] eta: 0:07:17 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2191 (0.3968) acc1: 97.3958 (82.3189) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1845 data: 0.0920 max mem: 3532 +Test: [ 420/2387] eta: 0:07:14 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2191 (0.3926) acc1: 97.3958 (82.6746) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1999 data: 0.1072 max mem: 3532 +Test: [ 430/2387] eta: 0:07:13 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2276 (0.3894) acc1: 96.6146 (82.9787) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2324 data: 0.1387 max mem: 3532 +Test: [ 440/2387] eta: 0:07:09 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2315 (0.3846) acc1: 95.8333 (83.3227) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2117 data: 0.1159 max mem: 3533 +Test: [ 450/2387] eta: 0:07:03 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1128 (0.3786) acc1: 100.0000 (83.6925) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1549 data: 0.0595 max mem: 3533 +Test: [ 460/2387] eta: 0:07:00 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1126 (0.3728) acc1: 100.0000 (84.0462) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1600 data: 0.0679 max mem: 3533 +Test: [ 470/2387] eta: 0:06:56 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1127 (0.3673) acc1: 100.0000 (84.3850) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1864 data: 0.0942 max mem: 3533 +Test: [ 480/2387] eta: 0:06:54 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1130 (0.3620) acc1: 100.0000 (84.7096) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2039 data: 0.1109 max mem: 3533 +Test: [ 490/2387] eta: 0:06:50 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1130 (0.3569) acc1: 100.0000 (85.0210) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1984 data: 0.1033 max mem: 3533 +Test: [ 500/2387] eta: 0:06:46 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1126 (0.3520) acc1: 100.0000 (85.3200) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1646 data: 0.0689 max mem: 3533 +Test: [ 510/2387] eta: 0:06:43 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1126 (0.3476) acc1: 100.0000 (85.6073) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1719 data: 0.0763 max mem: 3533 +Test: [ 520/2387] eta: 0:06:40 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1493 (0.3439) acc1: 100.0000 (85.8835) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1973 data: 0.1008 max mem: 3533 +Test: [ 530/2387] eta: 0:06:38 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1473 (0.3401) acc1: 100.0000 (86.1494) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2092 data: 0.1149 max mem: 3533 +Test: [ 540/2387] eta: 0:06:35 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1453 (0.3365) acc1: 100.0000 (86.4054) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2024 data: 0.1097 max mem: 3533 +Test: [ 550/2387] eta: 0:06:32 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1453 (0.3331) acc1: 100.0000 (86.6516) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1836 data: 0.0896 max mem: 3533 +Test: [ 560/2387] eta: 0:06:28 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1453 (0.3298) acc1: 100.0000 (86.8882) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1704 data: 0.0771 max mem: 3533 +Test: [ 570/2387] eta: 0:06:24 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1452 (0.3265) acc1: 100.0000 (87.1178) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1658 data: 0.0729 max mem: 3533 +Test: [ 580/2387] eta: 0:06:21 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1454 (0.3234) acc1: 100.0000 (87.3395) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1655 data: 0.0712 max mem: 3533 +Test: [ 590/2387] eta: 0:06:17 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1470 (0.3204) acc1: 100.0000 (87.5533) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1702 data: 0.0772 max mem: 3533 +Test: [ 600/2387] eta: 0:06:14 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1503 (0.3177) acc1: 100.0000 (87.7591) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1683 data: 0.0763 max mem: 3533 +Test: [ 610/2387] eta: 0:06:10 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1690 (0.3157) acc1: 100.0000 (87.9437) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1589 data: 0.0670 max mem: 3533 +Test: [ 620/2387] eta: 0:06:07 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1958 (0.3138) acc1: 98.9583 (88.1248) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1554 data: 0.0638 max mem: 3533 +Test: [ 630/2387] eta: 0:06:04 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1970 (0.3120) acc1: 99.2188 (88.2965) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1665 data: 0.0756 max mem: 3533 +Test: [ 640/2387] eta: 0:06:01 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1997 (0.3102) acc1: 98.6979 (88.4596) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1694 data: 0.0773 max mem: 3533 +Test: [ 650/2387] eta: 0:05:57 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1997 (0.3085) acc1: 98.6979 (88.6201) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1555 data: 0.0636 max mem: 3533 +Test: [ 660/2387] eta: 0:05:54 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1954 (0.3068) acc1: 98.9583 (88.7781) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1512 data: 0.0579 max mem: 3533 +Test: [ 670/2387] eta: 0:05:51 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1977 (0.3052) acc1: 98.9583 (88.9286) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1674 data: 0.0722 max mem: 3534 +Test: [ 680/2387] eta: 0:05:48 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2019 (0.3043) acc1: 99.4792 (89.0851) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1751 data: 0.0794 max mem: 3534 +Test: [ 690/2387] eta: 0:05:45 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2692 (0.3048) acc1: 98.9583 (89.0783) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.1649 data: 0.0702 max mem: 3534 +Test: [ 700/2387] eta: 0:05:41 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.3633 (0.3056) acc1: 84.8958 (89.0150) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.1545 data: 0.0630 max mem: 3534 +Test: [ 710/2387] eta: 0:05:39 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.3642 (0.3065) acc1: 84.3750 (88.9566) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1599 data: 0.0680 max mem: 3534 +Test: [ 720/2387] eta: 0:05:36 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.3624 (0.3072) acc1: 85.1562 (88.9039) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.1639 data: 0.0702 max mem: 3534 +Test: [ 730/2387] eta: 0:05:33 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.3628 (0.3081) acc1: 84.3750 (88.8324) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.1599 data: 0.0649 max mem: 3534 +Test: [ 740/2387] eta: 0:05:30 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.3631 (0.3088) acc1: 84.3750 (88.7799) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.1553 data: 0.0592 max mem: 3534 +Test: [ 750/2387] eta: 0:05:27 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.3544 (0.3094) acc1: 85.1562 (88.7265) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1550 data: 0.0592 max mem: 3534 +Test: [ 760/2387] eta: 0:05:24 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.3492 (0.3099) acc1: 85.1562 (88.6827) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1547 data: 0.0614 max mem: 3534 +Test: [ 770/2387] eta: 0:05:21 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.3293 (0.3094) acc1: 87.2396 (88.7176) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1548 data: 0.0596 max mem: 3534 +Test: [ 780/2387] eta: 0:05:18 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2155 (0.3080) acc1: 96.0938 (88.8264) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1607 data: 0.0644 max mem: 3534 +Test: [ 790/2387] eta: 0:05:15 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1831 (0.3064) acc1: 97.3958 (88.9328) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1649 data: 0.0693 max mem: 3534 +Test: [ 800/2387] eta: 0:05:13 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1710 (0.3050) acc1: 97.6562 (89.0404) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1739 data: 0.0769 max mem: 3534 +Test: [ 810/2387] eta: 0:05:12 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1908 (0.3038) acc1: 97.6562 (89.1354) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2013 data: 0.1064 max mem: 3534 +Test: [ 820/2387] eta: 0:05:09 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1990 (0.3028) acc1: 95.8333 (89.2109) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.2056 data: 0.1132 max mem: 3534 +Test: [ 830/2387] eta: 0:05:07 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2717 (0.3033) acc1: 91.6667 (89.1976) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1836 data: 0.0911 max mem: 3534 +Test: [ 840/2387] eta: 0:05:06 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.3085 (0.3030) acc1: 91.6667 (89.2681) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1991 data: 0.1052 max mem: 3534 +Test: [ 850/2387] eta: 0:05:04 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2742 (0.3027) acc1: 95.3125 (89.3275) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.2204 data: 0.1275 max mem: 3534 +Test: [ 860/2387] eta: 0:05:02 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2622 (0.3021) acc1: 96.6146 (89.4209) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2099 data: 0.1168 max mem: 3534 +Test: [ 870/2387] eta: 0:04:59 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2603 (0.3017) acc1: 97.1354 (89.4987) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1692 data: 0.0710 max mem: 3534 +Test: [ 880/2387] eta: 0:04:57 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2588 (0.3014) acc1: 95.8333 (89.5550) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1651 data: 0.0668 max mem: 3534 +Test: [ 890/2387] eta: 0:04:55 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2474 (0.3007) acc1: 97.3958 (89.6547) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1999 data: 0.1047 max mem: 3534 +Test: [ 900/2387] eta: 0:04:53 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2703 (0.3037) acc1: 95.3125 (89.4333) acc5: 100.0000 (100.0000) acc1_neg: 65.365 acc5_neg: 100.000 time: 0.1951 data: 0.0993 max mem: 3535 +Test: [ 910/2387] eta: 0:04:51 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.6504 (0.3075) acc1: 63.0208 (89.1423) acc5: 100.0000 (100.0000) acc1_neg: 62.760 acc5_neg: 100.000 time: 0.1821 data: 0.0884 max mem: 3535 +Test: [ 920/2387] eta: 0:04:49 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.6592 (0.3113) acc1: 62.7604 (88.8584) acc5: 100.0000 (100.0000) acc1_neg: 62.760 acc5_neg: 100.000 time: 0.1897 data: 0.0967 max mem: 3535 +Test: [ 930/2387] eta: 0:04:47 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.6604 (0.3149) acc1: 63.2812 (88.5898) acc5: 100.0000 (100.0000) acc1_neg: 59.896 acc5_neg: 100.000 time: 0.1928 data: 0.0996 max mem: 3535 +Test: [ 940/2387] eta: 0:04:45 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.6604 (0.3185) acc1: 63.0208 (88.3122) acc5: 100.0000 (100.0000) acc1_neg: 59.375 acc5_neg: 100.000 time: 0.1947 data: 0.0974 max mem: 3535 +Test: [ 950/2387] eta: 0:04:43 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.6409 (0.3220) acc1: 62.5000 (88.0416) acc5: 100.0000 (100.0000) acc1_neg: 63.542 acc5_neg: 100.000 time: 0.1999 data: 0.1019 max mem: 3535 +Test: [ 960/2387] eta: 0:04:42 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.6355 (0.3253) acc1: 63.5417 (87.7878) acc5: 100.0000 (100.0000) acc1_neg: 63.542 acc5_neg: 100.000 time: 0.2257 data: 0.1311 max mem: 3535 +Test: [ 970/2387] eta: 0:04:39 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.6509 (0.3287) acc1: 62.2396 (87.5244) acc5: 100.0000 (100.0000) acc1_neg: 64.844 acc5_neg: 100.000 time: 0.2049 data: 0.1089 max mem: 3535 +Test: [ 980/2387] eta: 0:04:36 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.6361 (0.3295) acc1: 63.2812 (87.4681) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.1291 data: 0.0311 max mem: 3535 +Test: [ 990/2387] eta: 0:04:33 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2575 (0.3288) acc1: 94.0104 (87.5229) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1299 data: 0.0342 max mem: 3535 +Test: [1000/2387] eta: 0:04:31 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2242 (0.3277) acc1: 96.6146 (87.6046) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1699 data: 0.0741 max mem: 3535 +Test: [1010/2387] eta: 0:04:29 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2386 (0.3273) acc1: 95.5729 (87.6448) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.2004 data: 0.1044 max mem: 3535 +Test: [1020/2387] eta: 0:04:27 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2411 (0.3268) acc1: 92.9688 (87.6819) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1856 data: 0.0904 max mem: 3535 +Test: [1030/2387] eta: 0:04:25 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2080 (0.3256) acc1: 97.6562 (87.7763) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1694 data: 0.0737 max mem: 3535 +Test: [1040/2387] eta: 0:04:23 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1815 (0.3241) acc1: 98.9583 (87.8830) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1849 data: 0.0916 max mem: 3535 +Test: [1050/2387] eta: 0:04:20 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1789 (0.3229) acc1: 98.6979 (87.9797) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1799 data: 0.0880 max mem: 3535 +Test: [1060/2387] eta: 0:04:18 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1898 (0.3217) acc1: 97.9167 (88.0658) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1713 data: 0.0787 max mem: 3535 +Test: [1070/2387] eta: 0:04:16 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1940 (0.3209) acc1: 96.8750 (88.1281) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.1799 data: 0.0812 max mem: 3535 +Test: [1080/2387] eta: 0:04:14 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2850 (0.3209) acc1: 92.1875 (88.1488) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1917 data: 0.0938 max mem: 3535 +Test: [1090/2387] eta: 0:04:12 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2645 (0.3201) acc1: 92.1875 (88.2070) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1884 data: 0.0948 max mem: 3535 +Test: [1100/2387] eta: 0:04:10 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2342 (0.3194) acc1: 94.5312 (88.2571) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1838 data: 0.0897 max mem: 3535 +Test: [1110/2387] eta: 0:04:08 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2393 (0.3188) acc1: 94.7917 (88.3134) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1878 data: 0.0957 max mem: 3535 +Test: [1120/2387] eta: 0:04:06 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2469 (0.3184) acc1: 94.5312 (88.3456) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.1863 data: 0.0953 max mem: 3536 +Test: [1130/2387] eta: 0:04:04 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2675 (0.3178) acc1: 93.2292 (88.3929) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1918 data: 0.1002 max mem: 3536 +Test: [1140/2387] eta: 0:04:02 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2857 (0.3177) acc1: 92.4479 (88.4173) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.2004 data: 0.1064 max mem: 3536 +Test: [1150/2387] eta: 0:04:00 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2621 (0.3172) acc1: 92.4479 (88.4756) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1873 data: 0.0916 max mem: 3536 +Test: [1160/2387] eta: 0:03:58 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1932 (0.3160) acc1: 98.9583 (88.5643) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1838 data: 0.0898 max mem: 3536 +Test: [1170/2387] eta: 0:03:56 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1705 (0.3147) acc1: 99.7396 (88.6555) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1918 data: 0.0996 max mem: 3536 +Test: [1180/2387] eta: 0:03:54 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1715 (0.3135) acc1: 99.4792 (88.7399) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1888 data: 0.0971 max mem: 3536 +Test: [1190/2387] eta: 0:03:52 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1715 (0.3123) acc1: 99.4792 (88.8288) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1708 data: 0.0779 max mem: 3536 +Test: [1200/2387] eta: 0:03:50 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1603 (0.3111) acc1: 99.7396 (88.9205) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1864 data: 0.0930 max mem: 3536 +Test: [1210/2387] eta: 0:03:48 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1553 (0.3098) acc1: 99.7396 (89.0098) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1893 data: 0.0968 max mem: 3536 +Test: [1220/2387] eta: 0:03:46 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1262 (0.3083) acc1: 99.7396 (89.0964) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1800 data: 0.0885 max mem: 3536 +Test: [1230/2387] eta: 0:03:44 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1099 (0.3066) acc1: 100.0000 (89.1844) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1951 data: 0.1034 max mem: 3536 +Test: [1240/2387] eta: 0:03:42 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1068 (0.3050) acc1: 100.0000 (89.2707) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1913 data: 0.0996 max mem: 3536 +Test: [1250/2387] eta: 0:03:40 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1068 (0.3035) acc1: 100.0000 (89.3541) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1894 data: 0.0936 max mem: 3536 +Test: [1260/2387] eta: 0:03:38 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1024 (0.3019) acc1: 100.0000 (89.4369) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1969 data: 0.0979 max mem: 3536 +Test: [1270/2387] eta: 0:03:36 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1021 (0.3003) acc1: 100.0000 (89.5196) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1855 data: 0.0908 max mem: 3536 +Test: [1280/2387] eta: 0:03:34 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1095 (0.2990) acc1: 99.7396 (89.5959) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1868 data: 0.0915 max mem: 3536 +Test: [1290/2387] eta: 0:03:32 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1357 (0.2978) acc1: 99.2188 (89.6687) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2099 data: 0.1123 max mem: 3536 +Test: [1300/2387] eta: 0:03:30 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1448 (0.2967) acc1: 98.6979 (89.7373) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1964 data: 0.1016 max mem: 3536 +Test: [1310/2387] eta: 0:03:28 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1456 (0.2956) acc1: 99.2188 (89.8042) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1803 data: 0.0882 max mem: 3536 +Test: [1320/2387] eta: 0:03:27 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1532 (0.2945) acc1: 99.2188 (89.8735) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2068 data: 0.1156 max mem: 3536 +Test: [1330/2387] eta: 0:03:24 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1625 (0.2939) acc1: 98.1771 (89.9197) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1889 data: 0.0974 max mem: 3536 +Test: [1340/2387] eta: 0:03:22 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2362 (0.2943) acc1: 94.2708 (89.8966) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1546 data: 0.0610 max mem: 3536 +Test: [1350/2387] eta: 0:03:20 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.0862 (0.2927) acc1: 100.0000 (89.9714) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1732 data: 0.0792 max mem: 3537 +Test: [1360/2387] eta: 0:03:18 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.0741 (0.2911) acc1: 100.0000 (90.0450) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0873 max mem: 3537 +Test: [1370/2387] eta: 0:03:16 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.0741 (0.2895) acc1: 100.0000 (90.1177) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1795 data: 0.0874 max mem: 3537 +Test: [1380/2387] eta: 0:03:14 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.0720 (0.2879) acc1: 100.0000 (90.1892) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2001 data: 0.1089 max mem: 3537 +Test: [1390/2387] eta: 0:03:12 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.0666 (0.2864) acc1: 100.0000 (90.2597) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1821 data: 0.0890 max mem: 3537 +Test: [1400/2387] eta: 0:03:10 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.0648 (0.2848) acc1: 100.0000 (90.3293) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1951 data: 0.1016 max mem: 3537 +Test: [1410/2387] eta: 0:03:08 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.0723 (0.2843) acc1: 100.0000 (90.3605) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1999 data: 0.1078 max mem: 3537 +Test: [1420/2387] eta: 0:03:06 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2536 (0.2841) acc1: 93.4896 (90.3789) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.1329 data: 0.0388 max mem: 3537 +Test: [1430/2387] eta: 0:03:04 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2580 (0.2840) acc1: 92.4479 (90.3861) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1407 data: 0.0459 max mem: 3537 +Test: [1440/2387] eta: 0:03:02 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2661 (0.2839) acc1: 92.9688 (90.4016) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.1786 data: 0.0859 max mem: 3537 +Test: [1450/2387] eta: 0:03:00 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2661 (0.2841) acc1: 91.6667 (90.4057) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1747 data: 0.0836 max mem: 3537 +Test: [1460/2387] eta: 0:02:58 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1985 (0.2834) acc1: 97.1354 (90.4517) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1679 data: 0.0735 max mem: 3537 +Test: [1470/2387] eta: 0:02:55 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1585 (0.2826) acc1: 98.6979 (90.5121) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1703 data: 0.0738 max mem: 3537 +Test: [1480/2387] eta: 0:02:53 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1611 (0.2818) acc1: 99.2188 (90.5640) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1756 data: 0.0830 max mem: 3537 +Test: [1490/2387] eta: 0:02:51 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1670 (0.2810) acc1: 98.6979 (90.6182) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1788 data: 0.0851 max mem: 3537 +Test: [1500/2387] eta: 0:02:49 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1635 (0.2802) acc1: 98.6979 (90.6725) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1791 data: 0.0805 max mem: 3537 +Test: [1510/2387] eta: 0:02:47 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1679 (0.2795) acc1: 99.2188 (90.7305) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1774 data: 0.0816 max mem: 3537 +Test: [1520/2387] eta: 0:02:45 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1860 (0.2792) acc1: 98.6979 (90.7710) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1753 data: 0.0839 max mem: 3537 +Test: [1530/2387] eta: 0:02:43 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2358 (0.2789) acc1: 96.6146 (90.8043) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1780 data: 0.0850 max mem: 3537 +Test: [1540/2387] eta: 0:02:42 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2384 (0.2786) acc1: 97.9167 (90.8501) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1800 data: 0.0855 max mem: 3537 +Test: [1550/2387] eta: 0:02:40 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2172 (0.2782) acc1: 98.4375 (90.8948) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1795 data: 0.0860 max mem: 3537 +Test: [1560/2387] eta: 0:02:38 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2172 (0.2779) acc1: 98.1771 (90.9318) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1862 data: 0.0924 max mem: 3537 +Test: [1570/2387] eta: 0:02:36 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2076 (0.2775) acc1: 98.4375 (90.9804) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1756 data: 0.0827 max mem: 3537 +Test: [1580/2387] eta: 0:02:34 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2066 (0.2770) acc1: 98.4375 (91.0152) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1781 data: 0.0871 max mem: 3538 +Test: [1590/2387] eta: 0:02:32 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1946 (0.2765) acc1: 96.8750 (91.0566) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1939 data: 0.1019 max mem: 3538 +Test: [1600/2387] eta: 0:02:30 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1703 (0.2759) acc1: 97.9167 (91.0959) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2058 data: 0.1130 max mem: 3538 +Test: [1610/2387] eta: 0:02:28 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1703 (0.2754) acc1: 98.4375 (91.1342) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1956 data: 0.1039 max mem: 3538 +Test: [1620/2387] eta: 0:02:26 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2071 (0.2751) acc1: 95.5729 (91.1547) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1759 data: 0.0836 max mem: 3538 +Test: [1630/2387] eta: 0:02:24 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2205 (0.2749) acc1: 94.7917 (91.1701) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1802 data: 0.0878 max mem: 3538 +Test: [1640/2387] eta: 0:02:22 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2818 (0.2752) acc1: 91.9271 (91.1655) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2180 data: 0.1269 max mem: 3538 +Test: [1650/2387] eta: 0:02:20 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.3002 (0.2753) acc1: 90.8854 (91.1703) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1942 data: 0.1001 max mem: 3538 +Test: [1660/2387] eta: 0:02:18 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2928 (0.2754) acc1: 90.8854 (91.1651) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1503 data: 0.0560 max mem: 3538 +Test: [1670/2387] eta: 0:02:16 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2928 (0.2755) acc1: 90.8854 (91.1700) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1908 data: 0.0993 max mem: 3538 +Test: [1680/2387] eta: 0:02:14 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2825 (0.2756) acc1: 91.4062 (91.1685) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1961 data: 0.1053 max mem: 3538 +Test: [1690/2387] eta: 0:02:13 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.3028 (0.2758) acc1: 91.1458 (91.1674) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1859 data: 0.0949 max mem: 3538 +Test: [1700/2387] eta: 0:02:11 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2974 (0.2757) acc1: 91.6667 (91.1798) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1853 data: 0.0936 max mem: 3538 +Test: [1710/2387] eta: 0:02:09 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1523 (0.2749) acc1: 98.9583 (91.2267) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1880 data: 0.0956 max mem: 3538 +Test: [1720/2387] eta: 0:02:07 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1327 (0.2740) acc1: 99.7396 (91.2751) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1881 data: 0.0962 max mem: 3538 +Test: [1730/2387] eta: 0:02:05 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1245 (0.2732) acc1: 99.7396 (91.3188) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1762 data: 0.0853 max mem: 3538 +Test: [1740/2387] eta: 0:02:03 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1227 (0.2724) acc1: 99.4792 (91.3656) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1776 data: 0.0869 max mem: 3538 +Test: [1750/2387] eta: 0:02:01 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1273 (0.2716) acc1: 99.7396 (91.4125) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1802 data: 0.0890 max mem: 3538 +Test: [1760/2387] eta: 0:01:59 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1313 (0.2709) acc1: 99.4792 (91.4539) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1690 data: 0.0774 max mem: 3538 +Test: [1770/2387] eta: 0:01:57 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1639 (0.2704) acc1: 98.6979 (91.4890) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1657 data: 0.0745 max mem: 3538 +Test: [1780/2387] eta: 0:01:55 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1668 (0.2698) acc1: 98.6979 (91.5302) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1641 data: 0.0728 max mem: 3538 +Test: [1790/2387] eta: 0:01:53 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1582 (0.2690) acc1: 98.9583 (91.5729) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1776 data: 0.0859 max mem: 3538 +Test: [1800/2387] eta: 0:01:51 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1435 (0.2684) acc1: 99.2188 (91.6142) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1693 data: 0.0760 max mem: 3538 +Test: [1810/2387] eta: 0:01:49 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1435 (0.2676) acc1: 99.2188 (91.6570) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1615 data: 0.0652 max mem: 3539 +Test: [1820/2387] eta: 0:01:47 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1323 (0.2669) acc1: 99.7396 (91.6980) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1870 data: 0.0921 max mem: 3539 +Test: [1830/2387] eta: 0:01:45 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1534 (0.2664) acc1: 99.4792 (91.7392) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1912 data: 0.1003 max mem: 3539 +Test: [1840/2387] eta: 0:01:43 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1688 (0.2659) acc1: 100.0000 (91.7839) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1877 data: 0.0961 max mem: 3539 +Test: [1850/2387] eta: 0:01:41 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1691 (0.2653) acc1: 100.0000 (91.8280) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1831 data: 0.0900 max mem: 3539 +Test: [1860/2387] eta: 0:01:40 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1682 (0.2648) acc1: 100.0000 (91.8717) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2023 data: 0.1102 max mem: 3539 +Test: [1870/2387] eta: 0:01:38 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1671 (0.2643) acc1: 100.0000 (91.9143) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1836 data: 0.0930 max mem: 3539 +Test: [1880/2387] eta: 0:01:36 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1684 (0.2638) acc1: 100.0000 (91.9569) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1669 data: 0.0756 max mem: 3539 +Test: [1890/2387] eta: 0:01:34 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1684 (0.2633) acc1: 100.0000 (91.9994) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1815 data: 0.0888 max mem: 3539 +Test: [1900/2387] eta: 0:01:32 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1714 (0.2629) acc1: 100.0000 (92.0406) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1681 data: 0.0761 max mem: 3539 +Test: [1910/2387] eta: 0:01:30 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2107 (0.2626) acc1: 99.7396 (92.0762) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1630 data: 0.0720 max mem: 3539 +Test: [1920/2387] eta: 0:01:28 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2171 (0.2628) acc1: 97.3958 (92.0692) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1676 data: 0.0746 max mem: 3539 +Test: [1930/2387] eta: 0:01:26 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2070 (0.2624) acc1: 96.8750 (92.1017) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1669 data: 0.0730 max mem: 3539 +Test: [1940/2387] eta: 0:01:24 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1994 (0.2623) acc1: 97.3958 (92.1216) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1713 data: 0.0771 max mem: 3539 +Test: [1950/2387] eta: 0:01:22 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2403 (0.2623) acc1: 95.5729 (92.1230) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1729 data: 0.0793 max mem: 3539 +Test: [1960/2387] eta: 0:01:20 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2591 (0.2624) acc1: 93.2292 (92.1234) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1750 data: 0.0833 max mem: 3539 +Test: [1970/2387] eta: 0:01:18 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.3231 (0.2630) acc1: 89.5833 (92.0736) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.1849 data: 0.0923 max mem: 3539 +Test: [1980/2387] eta: 0:01:16 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.3395 (0.2631) acc1: 87.5000 (92.0722) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1938 data: 0.1004 max mem: 3539 +Test: [1990/2387] eta: 0:01:15 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1677 (0.2626) acc1: 98.6979 (92.1055) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1903 data: 0.0956 max mem: 3539 +Test: [2000/2387] eta: 0:01:13 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1672 (0.2621) acc1: 98.6979 (92.1383) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1756 data: 0.0792 max mem: 3539 +Test: [2010/2387] eta: 0:01:11 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1688 (0.2617) acc1: 98.9583 (92.1720) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1811 data: 0.0877 max mem: 3539 +Test: [2020/2387] eta: 0:01:09 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1681 (0.2612) acc1: 98.9583 (92.2044) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1920 data: 0.0978 max mem: 3539 +Test: [2030/2387] eta: 0:01:07 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1681 (0.2608) acc1: 98.6979 (92.2366) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1956 data: 0.0993 max mem: 3540 +Test: [2040/2387] eta: 0:01:05 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1681 (0.2603) acc1: 98.6979 (92.2688) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1839 data: 0.0905 max mem: 3540 +Test: [2050/2387] eta: 0:01:03 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1680 (0.2599) acc1: 98.9583 (92.3015) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1791 data: 0.0876 max mem: 3540 +Test: [2060/2387] eta: 0:01:01 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1733 (0.2595) acc1: 98.6979 (92.3324) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1693 data: 0.0777 max mem: 3540 +Test: [2070/2387] eta: 0:00:59 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1759 (0.2591) acc1: 98.9583 (92.3657) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1600 data: 0.0673 max mem: 3540 +Test: [2080/2387] eta: 0:00:57 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1734 (0.2586) acc1: 98.9583 (92.3962) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1761 data: 0.0813 max mem: 3540 +Test: [2090/2387] eta: 0:00:55 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1722 (0.2582) acc1: 98.6979 (92.4266) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1839 data: 0.0885 max mem: 3540 +Test: [2100/2387] eta: 0:00:54 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1721 (0.2578) acc1: 98.6979 (92.4576) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1718 data: 0.0788 max mem: 3540 +Test: [2110/2387] eta: 0:00:52 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1735 (0.2574) acc1: 98.9583 (92.4876) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1729 data: 0.0791 max mem: 3540 +Test: [2120/2387] eta: 0:00:50 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1874 (0.2573) acc1: 98.4375 (92.5024) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2051 data: 0.1112 max mem: 3540 +Test: [2130/2387] eta: 0:00:48 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2009 (0.2571) acc1: 97.6562 (92.5253) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1898 data: 0.0968 max mem: 3540 +Test: [2140/2387] eta: 0:00:46 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2110 (0.2569) acc1: 97.9167 (92.5457) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1650 data: 0.0717 max mem: 3540 +Test: [2150/2387] eta: 0:00:44 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2286 (0.2568) acc1: 96.3542 (92.5566) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1823 data: 0.0910 max mem: 3540 +Test: [2160/2387] eta: 0:00:42 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2355 (0.2567) acc1: 95.3125 (92.5726) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1744 data: 0.0840 max mem: 3540 +Test: [2170/2387] eta: 0:00:40 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2396 (0.2570) acc1: 93.2292 (92.5626) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1625 data: 0.0712 max mem: 3540 +Test: [2180/2387] eta: 0:00:38 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.3397 (0.2579) acc1: 86.9792 (92.4971) acc5: 100.0000 (100.0000) acc1_neg: 63.542 acc5_neg: 100.000 time: 0.1685 data: 0.0756 max mem: 3540 +Test: [2190/2387] eta: 0:00:37 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.3088 (0.2581) acc1: 89.5833 (92.4851) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1860 data: 0.0927 max mem: 3540 +Test: [2200/2387] eta: 0:00:35 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1914 (0.2577) acc1: 97.3958 (92.5112) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1869 data: 0.0950 max mem: 3540 +Test: [2210/2387] eta: 0:00:33 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1674 (0.2573) acc1: 98.9583 (92.5400) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1806 data: 0.0877 max mem: 3540 +Test: [2220/2387] eta: 0:00:31 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1698 (0.2570) acc1: 98.6979 (92.5659) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1995 data: 0.1062 max mem: 3540 +Test: [2230/2387] eta: 0:00:29 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1770 (0.2566) acc1: 98.4375 (92.5907) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1796 data: 0.0886 max mem: 3540 +Test: [2240/2387] eta: 0:00:27 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1697 (0.2563) acc1: 98.4375 (92.6129) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1706 data: 0.0796 max mem: 3540 +Test: [2250/2387] eta: 0:00:25 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1869 (0.2560) acc1: 98.6979 (92.6402) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1894 data: 0.0946 max mem: 3540 +Test: [2260/2387] eta: 0:00:23 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1911 (0.2557) acc1: 99.4792 (92.6639) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1792 data: 0.0834 max mem: 3541 +Test: [2270/2387] eta: 0:00:21 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1824 (0.2554) acc1: 99.4792 (92.6933) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1739 data: 0.0796 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1800 (0.2551) acc1: 99.7396 (92.7182) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1742 data: 0.0813 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1793 (0.2547) acc1: 98.4375 (92.7429) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1709 data: 0.0793 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.1881 (0.2545) acc1: 97.6562 (92.7598) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1642 data: 0.0689 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2015 (0.2545) acc1: 96.8750 (92.7668) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1539 data: 0.0595 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2652 (0.2547) acc1: 92.7083 (92.7566) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1529 data: 0.0607 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.3001 (0.2549) acc1: 90.8854 (92.7493) acc5: 100.0000 (100.0000) acc1_neg: 79.427 acc5_neg: 100.000 time: 0.1599 data: 0.0679 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2793 (0.2549) acc1: 92.9688 (92.7513) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1654 data: 0.0734 max mem: 3541 +Test: [2350/2387] eta: 0:00:06 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.2663 (0.2552) acc1: 91.9271 (92.7430) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1750 data: 0.0828 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.3179 (0.2555) acc1: 89.3229 (92.7204) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2074 data: 0.1164 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.3241 (0.2559) acc1: 88.5417 (92.6922) acc5: 100.0000 (100.0000) acc1_neg: 79.427 acc5_neg: 100.000 time: 0.1849 data: 0.0940 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.3521 (0.2565) acc1: 84.3750 (92.6550) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.1623 data: 0.0704 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 71.689 acc5_pos: 100.000 loss: 0.3698 (0.2566) acc1: 83.3333 (92.6496) acc5: 100.0000 (100.0000) acc1_neg: 95.349 acc5_neg: 100.000 time: 0.1625 data: 0.0732 max mem: 3541 +Test: Total time: 0:07:26 (0.1871 s / it) +* Overall Acc@1 92.650 Acc@5 100.000 + target == real Acc@1 57.958 Acc@5 100.000 + target == fake Acc@1 95.086 Acc@5 100.000 + loss 0.257 + PR-AUC 0.9948 ROC-AUC 0.9320 EER 13.93% (thr=0.6768) +Accuracy of the network on the 916267 test images: 92.6% +[rank0]:[W726 03:54:03.946703889 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:54:24.033000 1181007 torch/distributed/run.py:766] +W0726 03:54:24.033000 1181007 torch/distributed/run.py:766] ***************************************** +W0726 03:54:24.033000 1181007 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:54:24.033000 1181007 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 03:54:52.310958794 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:54:52.791870314 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:54:52.819451581 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:54:52.827080504 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:54:52.879763876 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:54:52.932570488 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:54:52.038434813 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:54:52.070484313 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n32_seed6_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n32_classseed6_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:06:19 acc1_pos: 60.677 acc5_pos: 100.000 loss: 0.6390 (0.6390) acc1: 60.6771 (60.6771) acc5: 100.0000 (100.0000) time: 7.6964 data: 6.0782 max mem: 3531 +Test: [ 10/517] eta: 0:06:36 acc1_pos: 33.854 acc5_pos: 100.000 loss: 0.7790 (0.7502) acc1: 51.0417 (51.6809) acc5: 100.0000 (100.0000) time: 0.7830 data: 0.5529 max mem: 3531 +Test: [ 20/517] eta: 0:03:47 acc1_pos: 63.542 acc5_pos: 100.000 loss: 0.7770 (0.7568) acc1: 49.7396 (51.6121) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:03:13 acc1_pos: 38.542 acc5_pos: 100.000 loss: 0.7581 (0.7578) acc1: 52.3438 (52.2345) acc5: 100.0000 (100.0000) time: 0.1859 data: 0.0865 max mem: 3531 +Test: [ 40/517] eta: 0:02:53 acc1_pos: 53.385 acc5_pos: 100.000 loss: 0.7581 (0.7630) acc1: 50.2604 (51.2830) acc5: 100.0000 (100.0000) time: 0.2649 data: 0.1684 max mem: 3531 +Test: [ 50/517] eta: 0:02:43 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.7239 (0.7506) acc1: 51.8229 (52.5582) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.2740 data: 0.1788 max mem: 3531 +Test: [ 60/517] eta: 0:02:34 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.5494 (0.7089) acc1: 70.5729 (57.2575) acc5: 100.0000 (100.0000) acc1_neg: 75.521 acc5_neg: 100.000 time: 0.2849 data: 0.1892 max mem: 3531 +Test: [ 70/517] eta: 0:02:27 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.4851 (0.6763) acc1: 81.7708 (60.7431) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.2809 data: 0.1877 max mem: 3531 +Test: [ 80/517] eta: 0:02:21 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.4900 (0.6553) acc1: 78.3854 (63.0369) acc5: 100.0000 (100.0000) acc1_neg: 74.740 acc5_neg: 100.000 time: 0.2811 data: 0.1889 max mem: 3531 +Test: [ 90/517] eta: 0:02:16 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.4724 (0.6326) acc1: 85.1562 (65.9112) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.2839 data: 0.1908 max mem: 3531 +Test: [100/517] eta: 0:02:10 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.4241 (0.6102) acc1: 90.1042 (68.4251) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2636 data: 0.1690 max mem: 3531 +Test: [110/517] eta: 0:02:04 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.3678 (0.5858) acc1: 94.0104 (70.8310) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2406 data: 0.1449 max mem: 3531 +Test: [120/517] eta: 0:01:59 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.3897 (0.5773) acc1: 88.5417 (71.6124) acc5: 100.0000 (100.0000) acc1_neg: 77.344 acc5_neg: 100.000 time: 0.2513 data: 0.1556 max mem: 3531 +Test: [130/517] eta: 0:01:54 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.4930 (0.5685) acc1: 78.6458 (72.5549) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2401 data: 0.1468 max mem: 3531 +Test: [140/517] eta: 0:01:49 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.4064 (0.5558) acc1: 89.5833 (74.0285) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2236 data: 0.1295 max mem: 3531 +Test: [150/517] eta: 0:01:46 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.4005 (0.5450) acc1: 94.0104 (75.2794) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.2497 data: 0.1545 max mem: 3531 +Test: [160/517] eta: 0:01:42 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.3950 (0.5359) acc1: 90.1042 (76.1420) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.2549 data: 0.1587 max mem: 3531 +Test: [170/517] eta: 0:01:38 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.3887 (0.5261) acc1: 90.1042 (77.0361) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2492 data: 0.1523 max mem: 3531 +Test: [180/517] eta: 0:01:35 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.3885 (0.5179) acc1: 90.3646 (77.8027) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.2502 data: 0.1567 max mem: 3531 +Test: [190/517] eta: 0:01:31 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.2826 (0.5051) acc1: 97.3958 (78.8722) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2399 data: 0.1467 max mem: 3531 +Test: [200/517] eta: 0:01:28 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.2954 (0.4957) acc1: 97.9167 (79.7873) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2446 data: 0.1461 max mem: 3531 +Test: [210/517] eta: 0:01:25 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.3431 (0.4953) acc1: 91.9271 (79.8442) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.2499 data: 0.1506 max mem: 3532 +Test: [220/517] eta: 0:01:21 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.5100 (0.4961) acc1: 77.6042 (79.7806) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.2499 data: 0.1557 max mem: 3532 +Test: [230/517] eta: 0:01:19 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.5193 (0.4981) acc1: 77.0833 (79.6300) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.2599 data: 0.1609 max mem: 3532 +Test: [240/517] eta: 0:01:16 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.5089 (0.4962) acc1: 83.8542 (80.0495) acc5: 100.0000 (100.0000) acc1_neg: 72.135 acc5_neg: 100.000 time: 0.2602 data: 0.1601 max mem: 3532 +Test: [250/517] eta: 0:01:13 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.4156 (0.4930) acc1: 93.7500 (80.6109) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2599 data: 0.1639 max mem: 3532 +Test: [260/517] eta: 0:01:10 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.4089 (0.4901) acc1: 94.0104 (81.1133) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.2545 data: 0.1587 max mem: 3532 +Test: [270/517] eta: 0:01:07 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.4481 (0.4898) acc1: 88.8021 (81.2298) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.2549 data: 0.1582 max mem: 3532 +Test: [280/517] eta: 0:01:04 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.4701 (0.4881) acc1: 84.8958 (81.4604) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.2499 data: 0.1526 max mem: 3532 +Test: [290/517] eta: 0:01:01 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.4365 (0.4875) acc1: 87.2396 (81.6330) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.2349 data: 0.1398 max mem: 3532 +Test: [300/517] eta: 0:00:58 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.4365 (0.4862) acc1: 87.5000 (81.8003) acc5: 100.0000 (100.0000) acc1_neg: 75.781 acc5_neg: 100.000 time: 0.2399 data: 0.1462 max mem: 3532 +Test: [310/517] eta: 0:00:55 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.4872 (0.4879) acc1: 81.5104 (81.5782) acc5: 100.0000 (100.0000) acc1_neg: 69.271 acc5_neg: 100.000 time: 0.2449 data: 0.1488 max mem: 3532 +Test: [320/517] eta: 0:00:52 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.5054 (0.4866) acc1: 79.6875 (81.6873) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.2499 data: 0.1533 max mem: 3532 +Test: [330/517] eta: 0:00:50 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.4073 (0.4834) acc1: 90.1042 (82.0076) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2499 data: 0.1547 max mem: 3532 +Test: [340/517] eta: 0:00:47 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.3694 (0.4822) acc1: 93.2292 (82.0885) acc5: 100.0000 (100.0000) acc1_neg: 54.948 acc5_neg: 100.000 time: 0.2449 data: 0.1493 max mem: 3532 +Test: [350/517] eta: 0:00:44 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.4555 (0.4830) acc1: 89.0625 (82.0253) acc5: 100.0000 (100.0000) acc1_neg: 67.708 acc5_neg: 100.000 time: 0.2449 data: 0.1462 max mem: 3532 +Test: [360/517] eta: 0:00:41 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.4723 (0.4828) acc1: 83.0729 (82.0500) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.2449 data: 0.1439 max mem: 3532 +Test: [370/517] eta: 0:00:38 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.4303 (0.4791) acc1: 88.8021 (82.4328) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2449 data: 0.1475 max mem: 3532 +Test: [380/517] eta: 0:00:36 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.3410 (0.4763) acc1: 95.0521 (82.6553) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2499 data: 0.1550 max mem: 3532 +Test: [390/517] eta: 0:00:33 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.3606 (0.4735) acc1: 91.9271 (82.8325) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.2649 data: 0.1702 max mem: 3532 +Test: [400/517] eta: 0:00:30 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.3739 (0.4716) acc1: 91.1458 (82.9625) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2499 data: 0.1560 max mem: 3532 +Test: [410/517] eta: 0:00:28 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.4243 (0.4711) acc1: 87.2396 (83.0501) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.2399 data: 0.1445 max mem: 3532 +Test: [420/517] eta: 0:00:25 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.4835 (0.4713) acc1: 85.4167 (83.0921) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2549 data: 0.1565 max mem: 3532 +Test: [430/517] eta: 0:00:22 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.4676 (0.4707) acc1: 85.4167 (83.1774) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.2425 data: 0.1473 max mem: 3532 +Test: [440/517] eta: 0:00:20 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.4713 (0.4711) acc1: 85.1562 (83.1202) acc5: 100.0000 (100.0000) acc1_neg: 77.083 acc5_neg: 100.000 time: 0.2249 data: 0.1297 max mem: 3533 +Test: [450/517] eta: 0:00:17 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.4602 (0.4706) acc1: 84.1146 (83.1543) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.2323 data: 0.1335 max mem: 3533 +Test: [460/517] eta: 0:00:14 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.4582 (0.4708) acc1: 84.6354 (83.1368) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.2499 data: 0.1519 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.4411 (0.4693) acc1: 86.1979 (83.3073) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.2499 data: 0.1520 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.4131 (0.4685) acc1: 91.9271 (83.4411) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2399 data: 0.1434 max mem: 3533 +Test: [490/517] eta: 0:00:07 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.4107 (0.4669) acc1: 91.9271 (83.6091) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.2299 data: 0.1366 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.4518 (0.4682) acc1: 87.2396 (83.4555) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.2265 data: 0.1348 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.4633 (0.4671) acc1: 83.0729 (83.5688) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2061 data: 0.1151 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.3711 (0.4658) acc1: 92.7083 (83.7008) acc5: 100.0000 (100.0000) acc1_neg: 96.820 acc5_neg: 100.000 time: 0.1854 data: 0.0921 max mem: 3533 +Test: Total time: 0:02:13 (0.2579 s / it) +* Overall Acc@1 83.701 Acc@5 100.000 + target == real Acc@1 51.228 Acc@5 100.000 + target == fake Acc@1 87.070 Acc@5 100.000 + loss 0.466 + PR-AUC 0.9608 ROC-AUC 0.7679 EER 29.89% (thr=0.6025) +Accuracy of the network on the 198427 test images: 83.7% +[rank0]:[W726 03:57:16.644211860 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 03:57:36.945000 1181569 torch/distributed/run.py:766] +W0726 03:57:36.945000 1181569 torch/distributed/run.py:766] ***************************************** +W0726 03:57:36.945000 1181569 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:57:36.945000 1181569 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:58:04.903817822 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:58:05.352646877 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:58:05.551364906 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:58:05.827296530 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:58:05.996968521 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +[rank5]:[W726 03:58:05.034296739 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:58:05.059456488 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank6]:[W726 03:58:05.059468430 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n32_seed6_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n32_classseed6_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:33:21 acc1_pos: 53.906 acc5_pos: 100.000 loss: 0.6883 (0.6883) acc1: 53.9062 (53.9062) acc5: 100.0000 (100.0000) time: 5.1448 data: 2.7046 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/389] eta: 0:03:46 acc1_pos: 90.104 acc5_pos: 100.000 loss: 0.6643 (0.6177) acc1: 57.0312 (62.5710) acc5: 100.0000 (100.0000) time: 0.5980 data: 0.2942 max mem: 3531 +Test: [ 20/389] eta: 0:02:23 acc1_pos: 49.740 acc5_pos: 100.000 loss: 0.5706 (0.6201) acc1: 57.0312 (61.7436) acc5: 100.0000 (100.0000) time: 0.1499 data: 0.0584 max mem: 3531 +Test: [ 30/389] eta: 0:02:05 acc1_pos: 69.010 acc5_pos: 100.000 loss: 0.5706 (0.6048) acc1: 68.4896 (65.0958) acc5: 100.0000 (100.0000) time: 0.2148 data: 0.1194 max mem: 3531 +Test: [ 40/389] eta: 0:01:51 acc1_pos: 44.531 acc5_pos: 100.000 loss: 0.5639 (0.6080) acc1: 69.0104 (64.8819) acc5: 100.0000 (100.0000) time: 0.2461 data: 0.1519 max mem: 3531 +Test: [ 50/389] eta: 0:01:42 acc1_pos: 34.115 acc5_pos: 100.000 loss: 0.5820 (0.6224) acc1: 66.1458 (63.9451) acc5: 100.0000 (100.0000) time: 0.2287 data: 0.1378 max mem: 3531 +Test: [ 60/389] eta: 0:01:34 acc1_pos: 67.969 acc5_pos: 100.000 loss: 0.6566 (0.6397) acc1: 46.3542 (62.4957) acc5: 100.0000 (100.0000) time: 0.2199 data: 0.1273 max mem: 3531 +Test: [ 70/389] eta: 0:01:30 acc1_pos: 83.854 acc5_pos: 100.000 loss: 0.5243 (0.6252) acc1: 68.4896 (63.8168) acc5: 100.0000 (100.0000) time: 0.2332 data: 0.1410 max mem: 3531 +Test: [ 80/389] eta: 0:01:24 acc1_pos: 95.052 acc5_pos: 100.000 loss: 0.5816 (0.6186) acc1: 68.4896 (64.6219) acc5: 100.0000 (100.0000) time: 0.2389 data: 0.1484 max mem: 3531 +Test: [ 90/389] eta: 0:01:19 acc1_pos: 87.500 acc5_pos: 100.000 loss: 0.3639 (0.5900) acc1: 86.1979 (67.2333) acc5: 100.0000 (100.0000) time: 0.2082 data: 0.1167 max mem: 3531 +Test: [100/389] eta: 0:01:17 acc1_pos: 32.812 acc5_pos: 100.000 loss: 0.5896 (0.6131) acc1: 51.5625 (64.5034) acc5: 100.0000 (100.0000) time: 0.2455 data: 0.1531 max mem: 3531 +Test: [110/389] eta: 0:01:12 acc1_pos: 99.740 acc5_pos: 100.000 loss: 0.8299 (0.6246) acc1: 38.5417 (63.0302) acc5: 100.0000 (100.0000) time: 0.2252 data: 0.1316 max mem: 3531 +Test: [120/389] eta: 0:01:06 acc1_pos: 80.990 acc5_pos: 100.000 loss: 0.6867 (0.6238) acc1: 58.5938 (62.9692) acc5: 100.0000 (100.0000) time: 0.1491 data: 0.0543 max mem: 3531 +Test: [130/389] eta: 0:01:02 acc1_pos: 35.583 acc5_pos: 100.000 loss: 0.6447 (0.6247) acc1: 56.5104 (62.9910) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1493 data: 0.0548 max mem: 3531 +Test: [140/389] eta: 0:00:59 acc1_pos: 35.583 acc5_pos: 100.000 loss: 0.6240 (0.6298) acc1: 65.8854 (62.8398) acc5: 100.0000 (100.0000) acc1_neg: 6.510 acc5_neg: 100.000 time: 0.1673 data: 0.0743 max mem: 3531 +Test: [150/389] eta: 0:00:55 acc1_pos: 35.583 acc5_pos: 100.000 loss: 0.6240 (0.6390) acc1: 56.7708 (61.9671) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1632 data: 0.0716 max mem: 3531 +Test: [160/389] eta: 0:00:52 acc1_pos: 35.583 acc5_pos: 100.000 loss: 0.5740 (0.6279) acc1: 56.7708 (63.1988) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1584 data: 0.0670 max mem: 3531 +Test: [170/389] eta: 0:00:49 acc1_pos: 35.583 acc5_pos: 100.000 loss: 0.3078 (0.6133) acc1: 98.4375 (64.6778) acc5: 100.0000 (100.0000) acc1_neg: 73.698 acc5_neg: 100.000 time: 0.1651 data: 0.0718 max mem: 3531 +Test: [180/389] eta: 0:00:46 acc1_pos: 35.583 acc5_pos: 100.000 loss: 0.3918 (0.6069) acc1: 86.7188 (65.5775) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.1859 data: 0.0926 max mem: 3531 +Test: [190/389] eta: 0:00:43 acc1_pos: 35.583 acc5_pos: 100.000 loss: 0.5242 (0.6100) acc1: 75.5208 (65.2146) acc5: 100.0000 (100.0000) acc1_neg: 20.573 acc5_neg: 100.000 time: 0.1923 data: 0.1004 max mem: 3531 +Test: [200/389] eta: 0:00:41 acc1_pos: 35.583 acc5_pos: 100.000 loss: 0.5812 (0.6125) acc1: 68.7500 (65.1171) acc5: 100.0000 (100.0000) acc1_neg: 55.469 acc5_neg: 100.000 time: 0.1790 data: 0.0869 max mem: 3531 +Test: [210/389] eta: 0:00:38 acc1_pos: 35.583 acc5_pos: 100.000 loss: 0.4429 (0.6044) acc1: 77.3438 (65.8817) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1838 data: 0.0908 max mem: 3532 +Test: [220/389] eta: 0:00:36 acc1_pos: 35.583 acc5_pos: 100.000 loss: 0.3040 (0.5909) acc1: 96.6146 (67.1345) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.1589 data: 0.0642 max mem: 3532 +Test: [230/389] eta: 0:00:33 acc1_pos: 35.583 acc5_pos: 100.000 loss: 0.5099 (0.5941) acc1: 79.6875 (66.6565) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.1384 data: 0.0412 max mem: 3532 +Test: [240/389] eta: 0:00:31 acc1_pos: 35.583 acc5_pos: 100.000 loss: 0.6079 (0.5936) acc1: 56.2500 (66.7596) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.1524 data: 0.0539 max mem: 3532 +Test: [250/389] eta: 0:00:28 acc1_pos: 35.583 acc5_pos: 100.000 loss: 0.4839 (0.5849) acc1: 82.8125 (67.8422) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1599 data: 0.0649 max mem: 3532 +Test: [260/389] eta: 0:00:26 acc1_pos: 35.583 acc5_pos: 100.000 loss: 0.4429 (0.5823) acc1: 88.2812 (68.3399) acc5: 100.0000 (100.0000) acc1_neg: 51.562 acc5_neg: 100.000 time: 0.1769 data: 0.0823 max mem: 3532 +Test: [270/389] eta: 0:00:24 acc1_pos: 35.583 acc5_pos: 100.000 loss: 0.5821 (0.5864) acc1: 66.4062 (67.8429) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.1745 data: 0.0777 max mem: 3532 +Test: [280/389] eta: 0:00:22 acc1_pos: 35.583 acc5_pos: 100.000 loss: 0.6625 (0.5886) acc1: 61.1979 (67.5860) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1766 data: 0.0798 max mem: 3532 +Test: [290/389] eta: 0:00:19 acc1_pos: 35.583 acc5_pos: 100.000 loss: 0.5568 (0.5863) acc1: 67.4479 (67.9598) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1705 data: 0.0755 max mem: 3532 +Test: [300/389] eta: 0:00:17 acc1_pos: 35.583 acc5_pos: 100.000 loss: 0.5504 (0.5847) acc1: 79.9479 (68.3624) acc5: 100.0000 (100.0000) acc1_neg: 68.490 acc5_neg: 100.000 time: 0.1504 data: 0.0561 max mem: 3532 +Test: [310/389] eta: 0:00:15 acc1_pos: 35.583 acc5_pos: 100.000 loss: 0.5654 (0.5880) acc1: 75.5208 (68.0458) acc5: 100.0000 (100.0000) acc1_neg: 20.573 acc5_neg: 100.000 time: 0.1493 data: 0.0555 max mem: 3532 +Test: [320/389] eta: 0:00:13 acc1_pos: 35.583 acc5_pos: 100.000 loss: 0.5239 (0.5844) acc1: 71.6146 (68.5366) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1628 data: 0.0671 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 35.583 acc5_pos: 100.000 loss: 0.6781 (0.5918) acc1: 61.7188 (67.4519) acc5: 100.0000 (100.0000) acc1_neg: 68.750 acc5_neg: 100.000 time: 0.1599 data: 0.0635 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 35.583 acc5_pos: 100.000 loss: 0.4159 (0.5849) acc1: 68.7500 (68.2658) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1448 data: 0.0494 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 35.583 acc5_pos: 100.000 loss: 0.3690 (0.5980) acc1: 92.4479 (67.5970) acc5: 100.0000 (100.0000) acc1_neg: 1.042 acc5_neg: 100.000 time: 0.1480 data: 0.0509 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 35.583 acc5_pos: 100.000 loss: 1.0546 (0.6142) acc1: 25.7812 (66.2807) acc5: 100.0000 (100.0000) acc1_neg: 26.042 acc5_neg: 100.000 time: 0.1410 data: 0.0474 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 35.583 acc5_pos: 100.000 loss: 0.8967 (0.6129) acc1: 28.6458 (66.3340) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1337 data: 0.0432 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 35.583 acc5_pos: 100.000 loss: 0.5260 (0.6111) acc1: 70.0521 (66.4685) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.1302 data: 0.0398 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 35.583 acc5_pos: 100.000 loss: 0.4555 (0.6083) acc1: 84.1146 (66.7765) acc5: 100.0000 (100.0000) acc1_neg: 75.137 acc5_neg: 100.000 time: 0.1342 data: 0.0330 max mem: 3532 +Test: Total time: 0:01:13 (0.1883 s / it) +* Overall Acc@1 66.776 Acc@5 100.000 + target == real Acc@1 62.722 Acc@5 100.000 + target == fake Acc@1 68.661 Acc@5 100.000 + loss 0.608 + PR-AUC 0.8426 ROC-AUC 0.7243 EER 34.09% (thr=0.5190) +Accuracy of the network on the 149358 test images: 66.8% +[rank0]:[W726 03:59:35.503124065 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:59:54.968000 1182127 torch/distributed/run.py:766] +W0726 03:59:54.968000 1182127 torch/distributed/run.py:766] ***************************************** +W0726 03:59:54.968000 1182127 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:59:54.968000 1182127 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:00:25.702094598 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:00:25.099587627 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:00:26.452701015 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:00:26.609137412 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:00:26.744377418 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:00:26.799978452 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:00:26.159903020 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:00:27.380815231 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n32_seed6_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n32_classseed6_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:30:48 acc1_pos: 90.625 acc5_pos: 100.000 loss: 0.2902 (0.2902) acc1: 90.6250 (90.6250) acc5: 100.0000 (100.0000) time: 5.5505 data: 3.8486 max mem: 3531 +Test: [ 10/333] eta: 0:03:09 acc1_pos: 86.719 acc5_pos: 100.000 loss: 0.2902 (0.2701) acc1: 90.6250 (91.2642) acc5: 100.0000 (100.0000) time: 0.5867 data: 0.3503 max mem: 3531 +Test: [ 20/333] eta: 0:01:49 acc1_pos: 96.615 acc5_pos: 100.000 loss: 0.2967 (0.2905) acc1: 89.8438 (90.5754) acc5: 100.0000 (100.0000) time: 0.0904 data: 0.0005 max mem: 3531 +Test: [ 30/333] eta: 0:01:40 acc1_pos: 66.667 acc5_pos: 100.000 loss: 0.3856 (0.4058) acc1: 76.0417 (81.4348) acc5: 100.0000 (100.0000) time: 0.1928 data: 0.1021 max mem: 3531 +Test: [ 40/333] eta: 0:01:35 acc1_pos: 63.542 acc5_pos: 100.000 loss: 0.6601 (0.4700) acc1: 59.3750 (75.9146) acc5: 100.0000 (100.0000) time: 0.2974 data: 0.2020 max mem: 3531 +Test: [ 50/333] eta: 0:01:28 acc1_pos: 63.281 acc5_pos: 100.000 loss: 0.6717 (0.5072) acc1: 60.9375 (72.9984) acc5: 100.0000 (100.0000) time: 0.2849 data: 0.1883 max mem: 3531 +Test: [ 60/333] eta: 0:01:21 acc1_pos: 60.156 acc5_pos: 100.000 loss: 0.6792 (0.5379) acc1: 60.9375 (70.6796) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1484 max mem: 3531 +Test: [ 70/333] eta: 0:01:14 acc1_pos: 53.646 acc5_pos: 100.000 loss: 0.6792 (0.5576) acc1: 58.0729 (69.2452) acc5: 100.0000 (100.0000) time: 0.2014 data: 0.1023 max mem: 3531 +Test: [ 80/333] eta: 0:01:08 acc1_pos: 66.927 acc5_pos: 100.000 loss: 0.6876 (0.5741) acc1: 58.0729 (68.0105) acc5: 100.0000 (100.0000) time: 0.1810 data: 0.0856 max mem: 3531 +Test: [ 90/333] eta: 0:01:04 acc1_pos: 74.740 acc5_pos: 100.000 loss: 0.6357 (0.5728) acc1: 65.8854 (68.2978) acc5: 100.0000 (100.0000) time: 0.2088 data: 0.1148 max mem: 3531 +Test: [100/333] eta: 0:00:59 acc1_pos: 76.302 acc5_pos: 100.000 loss: 0.5480 (0.5709) acc1: 72.1354 (68.6082) acc5: 100.0000 (100.0000) time: 0.1984 data: 0.1017 max mem: 3531 +Test: [110/333] eta: 0:00:54 acc1_pos: 78.125 acc5_pos: 100.000 loss: 0.5641 (0.5700) acc1: 71.0938 (68.7312) acc5: 100.0000 (100.0000) time: 0.1401 data: 0.0439 max mem: 3531 +Test: [120/333] eta: 0:00:51 acc1_pos: 77.865 acc5_pos: 100.000 loss: 0.5153 (0.5675) acc1: 71.6146 (69.1804) acc5: 100.0000 (100.0000) time: 0.1806 data: 0.0869 max mem: 3531 +Test: [130/333] eta: 0:00:48 acc1_pos: 85.677 acc5_pos: 100.000 loss: 0.5152 (0.5647) acc1: 74.7396 (69.6008) acc5: 100.0000 (100.0000) time: 0.2140 data: 0.1194 max mem: 3531 +Test: [140/333] eta: 0:00:44 acc1_pos: 88.281 acc5_pos: 100.000 loss: 0.5037 (0.5601) acc1: 77.3438 (70.2516) acc5: 100.0000 (100.0000) time: 0.1512 data: 0.0552 max mem: 3531 +Test: [150/333] eta: 0:00:41 acc1_pos: 81.250 acc5_pos: 100.000 loss: 0.5094 (0.5604) acc1: 77.3438 (70.4005) acc5: 100.0000 (100.0000) time: 0.1449 data: 0.0494 max mem: 3531 +Test: [160/333] eta: 0:00:38 acc1_pos: 61.513 acc5_pos: 100.000 loss: 0.5562 (0.5703) acc1: 70.8333 (69.7512) acc5: 100.0000 (100.0000) acc1_neg: 38.021 acc5_neg: 100.000 time: 0.1879 data: 0.0926 max mem: 3531 +Test: [170/333] eta: 0:00:36 acc1_pos: 61.513 acc5_pos: 100.000 loss: 0.8199 (0.5881) acc1: 45.8333 (68.1941) acc5: 100.0000 (100.0000) acc1_neg: 49.740 acc5_neg: 100.000 time: 0.1799 data: 0.0852 max mem: 3531 +Test: [180/333] eta: 0:00:33 acc1_pos: 61.513 acc5_pos: 100.000 loss: 0.7845 (0.5976) acc1: 47.9167 (67.2335) acc5: 100.0000 (100.0000) acc1_neg: 47.396 acc5_neg: 100.000 time: 0.1649 data: 0.0699 max mem: 3531 +Test: [190/333] eta: 0:00:30 acc1_pos: 61.513 acc5_pos: 100.000 loss: 0.7444 (0.6047) acc1: 51.3021 (66.6476) acc5: 100.0000 (100.0000) acc1_neg: 52.604 acc5_neg: 100.000 time: 0.1749 data: 0.0777 max mem: 3531 +Test: [200/333] eta: 0:00:28 acc1_pos: 61.513 acc5_pos: 100.000 loss: 0.7409 (0.6113) acc1: 54.6875 (66.1497) acc5: 100.0000 (100.0000) acc1_neg: 50.260 acc5_neg: 100.000 time: 0.1849 data: 0.0887 max mem: 3531 +Test: [210/333] eta: 0:00:26 acc1_pos: 61.513 acc5_pos: 100.000 loss: 0.7661 (0.6200) acc1: 53.6458 (65.4201) acc5: 100.0000 (100.0000) acc1_neg: 45.833 acc5_neg: 100.000 time: 0.2105 data: 0.1160 max mem: 3532 +Test: [220/333] eta: 0:00:24 acc1_pos: 61.513 acc5_pos: 100.000 loss: 0.8010 (0.6306) acc1: 50.5208 (64.5987) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.2162 data: 0.1216 max mem: 3532 +Test: [230/333] eta: 0:00:21 acc1_pos: 61.513 acc5_pos: 100.000 loss: 0.5868 (0.6275) acc1: 67.9688 (64.9948) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.1793 data: 0.0846 max mem: 3532 +Test: [240/333] eta: 0:00:20 acc1_pos: 61.513 acc5_pos: 100.000 loss: 0.5779 (0.6305) acc1: 68.2292 (64.6309) acc5: 100.0000 (100.0000) acc1_neg: 54.688 acc5_neg: 100.000 time: 0.2385 data: 0.1454 max mem: 3532 +Test: [250/333] eta: 0:00:18 acc1_pos: 61.513 acc5_pos: 100.000 loss: 0.7144 (0.6348) acc1: 55.7292 (64.2742) acc5: 100.0000 (100.0000) acc1_neg: 50.000 acc5_neg: 100.000 time: 0.2799 data: 0.1851 max mem: 3532 +Test: [260/333] eta: 0:00:15 acc1_pos: 61.513 acc5_pos: 100.000 loss: 0.7498 (0.6403) acc1: 53.9062 (63.8460) acc5: 100.0000 (100.0000) acc1_neg: 57.031 acc5_neg: 100.000 time: 0.1768 data: 0.0821 max mem: 3532 +Test: [270/333] eta: 0:00:13 acc1_pos: 61.513 acc5_pos: 100.000 loss: 0.7044 (0.6417) acc1: 59.1146 (63.7761) acc5: 100.0000 (100.0000) acc1_neg: 59.896 acc5_neg: 100.000 time: 0.1763 data: 0.0841 max mem: 3532 +Test: [280/333] eta: 0:00:11 acc1_pos: 61.513 acc5_pos: 100.000 loss: 0.6912 (0.6484) acc1: 59.1146 (63.1219) acc5: 100.0000 (100.0000) acc1_neg: 48.438 acc5_neg: 100.000 time: 0.2297 data: 0.1358 max mem: 3532 +Test: [290/333] eta: 0:00:09 acc1_pos: 61.513 acc5_pos: 100.000 loss: 0.9249 (0.6587) acc1: 36.1979 (62.1760) acc5: 100.0000 (100.0000) acc1_neg: 41.667 acc5_neg: 100.000 time: 0.2102 data: 0.1177 max mem: 3532 +Test: [300/333] eta: 0:00:07 acc1_pos: 61.513 acc5_pos: 100.000 loss: 0.9345 (0.6675) acc1: 36.4583 (61.4341) acc5: 100.0000 (100.0000) acc1_neg: 41.667 acc5_neg: 100.000 time: 0.2044 data: 0.1136 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 61.513 acc5_pos: 100.000 loss: 0.8296 (0.6674) acc1: 46.0938 (61.4458) acc5: 100.0000 (100.0000) acc1_neg: 56.250 acc5_neg: 100.000 time: 0.1965 data: 0.1048 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 61.513 acc5_pos: 100.000 loss: 0.6853 (0.6688) acc1: 58.8542 (61.2993) acc5: 100.0000 (100.0000) acc1_neg: 41.406 acc5_neg: 100.000 time: 0.1536 data: 0.0616 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 61.513 acc5_pos: 100.000 loss: 0.8695 (0.6767) acc1: 42.4479 (60.6787) acc5: 100.0000 (100.0000) acc1_neg: 42.188 acc5_neg: 100.000 time: 0.1188 data: 0.0276 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 61.513 acc5_pos: 100.000 loss: 0.9117 (0.6782) acc1: 42.1875 (60.5688) acc5: 100.0000 (100.0000) acc1_neg: 42.817 acc5_neg: 100.000 time: 0.1188 data: 0.0245 max mem: 3532 +Test: Total time: 0:01:09 (0.2077 s / it) +* Overall Acc@1 60.569 Acc@5 100.000 + target == real Acc@1 70.688 Acc@5 100.000 + target == fake Acc@1 51.561 Acc@5 100.000 + loss 0.678 + PR-AUC 0.6661 ROC-AUC 0.6605 EER 38.49% (thr=0.4380) +Accuracy of the network on the 127843 test images: 60.6% +[rank0]:[W726 04:01:47.590397218 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:02:08.607000 1182690 torch/distributed/run.py:766] +W0726 04:02:08.607000 1182690 torch/distributed/run.py:766] ***************************************** +W0726 04:02:08.607000 1182690 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:02:08.607000 1182690 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:02:37.070736781 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:02:37.141587263 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:02:38.232436616 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:02:38.440956195 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:02:38.667836357 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:02:38.758557910 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:02:38.789732752 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:02:38.848869610 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n32_seed6_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n32_classseed6_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:25:00 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.2512 (0.2512) acc1: 95.3125 (95.3125) acc5: 100.0000 (100.0000) time: 6.3836 data: 4.9112 max mem: 3531 +Test: [ 10/235] eta: 0:02:29 acc1_pos: 93.490 acc5_pos: 100.000 loss: 0.2579 (0.2605) acc1: 94.7917 (94.4366) acc5: 100.0000 (100.0000) time: 0.6628 data: 0.4471 max mem: 3531 +Test: [ 20/235] eta: 0:01:24 acc1_pos: 95.573 acc5_pos: 100.000 loss: 0.2536 (0.2553) acc1: 95.0521 (94.8041) acc5: 100.0000 (100.0000) time: 0.0932 data: 0.0006 max mem: 3531 +Test: [ 30/235] eta: 0:01:14 acc1_pos: 95.833 acc5_pos: 100.000 loss: 0.2465 (0.2530) acc1: 95.3125 (94.9009) acc5: 100.0000 (100.0000) time: 0.1978 data: 0.1012 max mem: 3531 +Test: [ 40/235] eta: 0:01:04 acc1_pos: 94.792 acc5_pos: 100.000 loss: 0.2485 (0.2525) acc1: 95.3125 (94.9505) acc5: 100.0000 (100.0000) time: 0.2653 data: 0.1697 max mem: 3531 +Test: [ 50/235] eta: 0:00:55 acc1_pos: 93.229 acc5_pos: 100.000 loss: 0.2524 (0.2534) acc1: 94.7917 (94.9397) acc5: 100.0000 (100.0000) time: 0.2049 data: 0.1124 max mem: 3531 +Test: [ 60/235] eta: 0:00:50 acc1_pos: 94.010 acc5_pos: 100.000 loss: 0.2565 (0.2541) acc1: 94.7917 (94.9880) acc5: 100.0000 (100.0000) time: 0.2046 data: 0.1122 max mem: 3531 +Test: [ 70/235] eta: 0:00:44 acc1_pos: 93.846 acc5_pos: 100.000 loss: 0.2753 (0.3805) acc1: 93.2292 (85.8935) acc5: 100.0000 (100.0000) acc1_neg: 19.531 acc5_neg: 100.000 time: 0.1961 data: 0.1035 max mem: 3531 +Test: [ 80/235] eta: 0:00:39 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.2613 (0.4906) acc1: 22.9167 (78.0189) acc5: 100.0000 (100.0000) acc1_neg: 19.531 acc5_neg: 100.000 time: 0.1549 data: 0.0623 max mem: 3531 +Test: [ 90/235] eta: 0:00:35 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.2764 (0.5767) acc1: 22.1354 (71.8693) acc5: 100.0000 (100.0000) acc1_neg: 21.354 acc5_neg: 100.000 time: 0.1405 data: 0.0472 max mem: 3531 +Test: [100/235] eta: 0:00:31 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.2776 (0.6447) acc1: 21.8750 (66.9812) acc5: 100.0000 (100.0000) acc1_neg: 21.615 acc5_neg: 100.000 time: 0.1555 data: 0.0626 max mem: 3531 +Test: [110/235] eta: 0:00:29 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.2617 (0.6998) acc1: 22.1354 (62.9434) acc5: 100.0000 (100.0000) acc1_neg: 22.135 acc5_neg: 100.000 time: 0.1844 data: 0.0916 max mem: 3531 +Test: [120/235] eta: 0:00:26 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.3107 (0.7620) acc1: 19.5312 (58.8348) acc5: 100.0000 (100.0000) acc1_neg: 14.323 acc5_neg: 100.000 time: 0.1943 data: 0.1007 max mem: 3531 +Test: [130/235] eta: 0:00:23 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.4548 (0.8151) acc1: 12.7604 (55.3197) acc5: 100.0000 (100.0000) acc1_neg: 13.281 acc5_neg: 100.000 time: 0.1843 data: 0.0905 max mem: 3531 +Test: [140/235] eta: 0:00:21 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.4839 (0.8632) acc1: 12.2396 (52.2255) acc5: 100.0000 (100.0000) acc1_neg: 7.552 acc5_neg: 100.000 time: 0.1799 data: 0.0868 max mem: 3531 +Test: [150/235] eta: 0:00:18 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.4739 (0.9024) acc1: 12.2396 (49.5947) acc5: 100.0000 (100.0000) acc1_neg: 13.802 acc5_neg: 100.000 time: 0.1800 data: 0.0860 max mem: 3531 +Test: [160/235] eta: 0:00:16 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.4636 (0.9375) acc1: 12.5000 (47.3166) acc5: 100.0000 (100.0000) acc1_neg: 12.760 acc5_neg: 100.000 time: 0.1749 data: 0.0812 max mem: 3531 +Test: [170/235] eta: 0:00:13 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.4439 (0.9503) acc1: 13.8021 (46.2080) acc5: 100.0000 (100.0000) acc1_neg: 27.865 acc5_neg: 100.000 time: 0.1800 data: 0.0846 max mem: 3531 +Test: [180/235] eta: 0:00:11 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.1250 (0.9592) acc1: 30.2083 (45.4046) acc5: 100.0000 (100.0000) acc1_neg: 30.208 acc5_neg: 100.000 time: 0.1755 data: 0.0792 max mem: 3531 +Test: [190/235] eta: 0:00:09 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.1135 (0.9672) acc1: 30.9896 (44.6703) acc5: 100.0000 (100.0000) acc1_neg: 30.990 acc5_neg: 100.000 time: 0.1646 data: 0.0725 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.1163 (0.9753) acc1: 29.4271 (43.9055) acc5: 100.0000 (100.0000) acc1_neg: 27.865 acc5_neg: 100.000 time: 0.1743 data: 0.0828 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.1106 (0.9817) acc1: 28.9062 (43.2946) acc5: 100.0000 (100.0000) acc1_neg: 31.771 acc5_neg: 100.000 time: 0.1794 data: 0.0829 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.1078 (0.9873) acc1: 30.9896 (42.7413) acc5: 100.0000 (100.0000) acc1_neg: 31.510 acc5_neg: 100.000 time: 0.1750 data: 0.0779 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.0956 (0.9919) acc1: 31.7708 (42.3171) acc5: 100.0000 (100.0000) acc1_neg: 31.771 acc5_neg: 100.000 time: 0.1635 data: 0.0713 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.1109 (0.9953) acc1: 31.7708 (42.1667) acc5: 100.0000 (100.0000) acc1_neg: 34.783 acc5_neg: 100.000 time: 0.1446 data: 0.0562 max mem: 3532 +Test: Total time: 0:00:47 (0.2021 s / it) +* Overall Acc@1 42.167 Acc@5 100.000 + target == real Acc@1 94.994 Acc@5 100.000 + target == fake Acc@1 23.060 Acc@5 100.000 + loss 0.995 + PR-AUC 0.8691 ROC-AUC 0.7117 EER 34.61% (thr=0.2164) +Accuracy of the network on the 89879 test images: 42.2% +[rank0]:[W726 04:03:36.424781278 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:03:57.030000 1183249 torch/distributed/run.py:766] +W0726 04:03:57.030000 1183249 torch/distributed/run.py:766] ***************************************** +W0726 04:03:57.030000 1183249 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:03:57.030000 1183249 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 0): env:// +[rank4]:[W726 04:04:25.679137930 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:04:25.782906429 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:04:26.348558518 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:04:26.496233290 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +[rank6]:[W726 04:04:26.538950998 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:04:26.553705890 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:04:26.654563174 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:04:26.712138648 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n32_seed6_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n32_classseed6_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:31:19 acc1_pos: 36.979 acc5_pos: 100.000 loss: 0.8329 (0.8329) acc1: 36.9792 (36.9792) acc5: 100.0000 (100.0000) time: 6.9866 data: 5.3017 max mem: 3531 +Test: [ 10/269] eta: 0:03:07 acc1_pos: 52.604 acc5_pos: 100.000 loss: 0.6851 (0.6892) acc1: 56.7708 (58.3097) acc5: 100.0000 (100.0000) time: 0.7240 data: 0.4823 max mem: 3531 +Test: [ 20/269] eta: 0:01:46 acc1_pos: 59.896 acc5_pos: 100.000 loss: 0.6621 (0.6874) acc1: 56.7708 (58.3209) acc5: 100.0000 (100.0000) time: 0.0988 data: 0.0004 max mem: 3531 +Test: [ 30/269] eta: 0:01:29 acc1_pos: 72.135 acc5_pos: 100.000 loss: 0.6621 (0.6786) acc1: 55.2083 (59.1566) acc5: 100.0000 (100.0000) time: 0.1801 data: 0.0827 max mem: 3531 +Test: [ 40/269] eta: 0:01:19 acc1_pos: 76.724 acc5_pos: 100.000 loss: 0.5643 (0.6349) acc1: 72.1354 (64.4563) acc5: 100.0000 (100.0000) acc1_neg: 71.615 acc5_neg: 100.000 time: 0.2653 data: 0.1694 max mem: 3531 +Test: [ 50/269] eta: 0:01:11 acc1_pos: 76.724 acc5_pos: 100.000 loss: 0.5643 (0.6367) acc1: 71.6146 (64.3229) acc5: 100.0000 (100.0000) acc1_neg: 37.760 acc5_neg: 100.000 time: 0.2498 data: 0.1542 max mem: 3531 +Test: [ 60/269] eta: 0:01:01 acc1_pos: 76.724 acc5_pos: 100.000 loss: 0.6665 (0.6579) acc1: 59.6354 (61.7657) acc5: 100.0000 (100.0000) acc1_neg: 31.771 acc5_neg: 100.000 time: 0.1805 data: 0.0845 max mem: 3531 +Test: [ 70/269] eta: 0:00:56 acc1_pos: 76.724 acc5_pos: 100.000 loss: 0.7684 (0.6676) acc1: 47.3958 (60.2333) acc5: 100.0000 (100.0000) acc1_neg: 54.948 acc5_neg: 100.000 time: 0.1765 data: 0.0813 max mem: 3531 +Test: [ 80/269] eta: 0:00:51 acc1_pos: 76.724 acc5_pos: 100.000 loss: 0.6862 (0.6701) acc1: 51.0417 (59.6193) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.2196 data: 0.1266 max mem: 3531 +Test: [ 90/269] eta: 0:00:48 acc1_pos: 76.724 acc5_pos: 100.000 loss: 0.7065 (0.6756) acc1: 53.9062 (58.8227) acc5: 100.0000 (100.0000) acc1_neg: 60.417 acc5_neg: 100.000 time: 0.2241 data: 0.1311 max mem: 3531 +Test: [100/269] eta: 0:00:44 acc1_pos: 76.724 acc5_pos: 100.000 loss: 0.7052 (0.6760) acc1: 55.2083 (58.7923) acc5: 100.0000 (100.0000) acc1_neg: 54.948 acc5_neg: 100.000 time: 0.2093 data: 0.1173 max mem: 3531 +Test: [110/269] eta: 0:00:40 acc1_pos: 76.724 acc5_pos: 100.000 loss: 0.6748 (0.6758) acc1: 54.9479 (58.3005) acc5: 100.0000 (100.0000) acc1_neg: 45.052 acc5_neg: 100.000 time: 0.1904 data: 0.0986 max mem: 3531 +Test: [120/269] eta: 0:00:38 acc1_pos: 76.724 acc5_pos: 100.000 loss: 0.6273 (0.6635) acc1: 59.1146 (59.6074) acc5: 100.0000 (100.0000) acc1_neg: 66.667 acc5_neg: 100.000 time: 0.2205 data: 0.1259 max mem: 3531 +Test: [130/269] eta: 0:00:35 acc1_pos: 76.724 acc5_pos: 100.000 loss: 0.5788 (0.6576) acc1: 67.7083 (60.2497) acc5: 100.0000 (100.0000) acc1_neg: 61.979 acc5_neg: 100.000 time: 0.2408 data: 0.1464 max mem: 3531 +Test: [140/269] eta: 0:00:32 acc1_pos: 76.724 acc5_pos: 100.000 loss: 0.6226 (0.6611) acc1: 60.6771 (59.7684) acc5: 100.0000 (100.0000) acc1_neg: 45.573 acc5_neg: 100.000 time: 0.2103 data: 0.1181 max mem: 3531 +Test: [150/269] eta: 0:00:29 acc1_pos: 76.724 acc5_pos: 100.000 loss: 0.6770 (0.6590) acc1: 52.6042 (59.9786) acc5: 100.0000 (100.0000) acc1_neg: 75.781 acc5_neg: 100.000 time: 0.2027 data: 0.1093 max mem: 3531 +Test: [160/269] eta: 0:00:26 acc1_pos: 76.724 acc5_pos: 100.000 loss: 0.5657 (0.6529) acc1: 71.3542 (60.6771) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.2208 data: 0.1266 max mem: 3531 +Test: [170/269] eta: 0:00:24 acc1_pos: 76.724 acc5_pos: 100.000 loss: 0.5411 (0.6493) acc1: 71.8750 (61.1279) acc5: 100.0000 (100.0000) acc1_neg: 73.958 acc5_neg: 100.000 time: 0.2216 data: 0.1285 max mem: 3531 +Test: [180/269] eta: 0:00:21 acc1_pos: 76.724 acc5_pos: 100.000 loss: 0.5183 (0.6405) acc1: 77.8646 (62.1878) acc5: 100.0000 (100.0000) acc1_neg: 62.760 acc5_neg: 100.000 time: 0.2280 data: 0.1334 max mem: 3531 +Test: [190/269] eta: 0:00:19 acc1_pos: 76.724 acc5_pos: 100.000 loss: 0.5358 (0.6390) acc1: 72.9167 (62.2559) acc5: 100.0000 (100.0000) acc1_neg: 69.531 acc5_neg: 100.000 time: 0.2144 data: 0.1202 max mem: 3531 +Test: [200/269] eta: 0:00:16 acc1_pos: 76.724 acc5_pos: 100.000 loss: 0.6044 (0.6379) acc1: 62.5000 (62.3381) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.2214 data: 0.1264 max mem: 3531 +Test: [210/269] eta: 0:00:14 acc1_pos: 76.724 acc5_pos: 100.000 loss: 0.6121 (0.6373) acc1: 63.5417 (62.3963) acc5: 100.0000 (100.0000) acc1_neg: 71.094 acc5_neg: 100.000 time: 0.2244 data: 0.1290 max mem: 3532 +Test: [220/269] eta: 0:00:11 acc1_pos: 76.724 acc5_pos: 100.000 loss: 0.5898 (0.6345) acc1: 64.5833 (62.7675) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.2010 data: 0.1090 max mem: 3532 +Test: [230/269] eta: 0:00:09 acc1_pos: 76.724 acc5_pos: 100.000 loss: 0.4296 (0.6261) acc1: 81.2500 (63.7468) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.1708 data: 0.0797 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 76.724 acc5_pos: 100.000 loss: 0.5461 (0.6286) acc1: 73.1771 (63.4433) acc5: 100.0000 (100.0000) acc1_neg: 58.333 acc5_neg: 100.000 time: 0.1326 data: 0.0421 max mem: 3532 +Test: [250/269] eta: 0:00:04 acc1_pos: 76.724 acc5_pos: 100.000 loss: 0.6867 (0.6330) acc1: 55.9896 (62.9887) acc5: 100.0000 (100.0000) acc1_neg: 28.646 acc5_neg: 100.000 time: 0.1291 data: 0.0390 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 76.724 acc5_pos: 100.000 loss: 0.7716 (0.6411) acc1: 46.6146 (62.1638) acc5: 100.0000 (100.0000) acc1_neg: 55.469 acc5_neg: 100.000 time: 0.1271 data: 0.0371 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 76.724 acc5_pos: 100.000 loss: 0.7716 (0.6429) acc1: 42.7083 (61.9510) acc5: 100.0000 (100.0000) acc1_neg: 75.591 acc5_neg: 100.000 time: 0.1260 data: 0.0324 max mem: 3532 +Test: Total time: 0:00:58 (0.2187 s / it) +* Overall Acc@1 61.951 Acc@5 100.000 + target == real Acc@1 59.312 Acc@5 100.000 + target == fake Acc@1 62.311 Acc@5 100.000 + loss 0.643 + PR-AUC 0.9275 ROC-AUC 0.6542 EER 39.13% (thr=0.5083) +Accuracy of the network on the 103293 test images: 62.0% +[rank0]:[W726 04:05:33.501451748 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T04:05:40Z] eval_done tag=fixed500k_real_n32_seed6_resampled +[2026-07-26T04:05:40Z] eval_start tag=fixed500k_fake_n64_seed6_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed6_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T04:05:40Z +hostname=n116-040-213 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed6_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n64_seed6_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 04:05:53.392000 1183825 torch/distributed/run.py:766] +W0726 04:05:53.392000 1183825 torch/distributed/run.py:766] ***************************************** +W0726 04:05:53.392000 1183825 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:05:53.392000 1183825 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:06:23.945186422 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:06:25.204307931 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:06:25.346737011 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:06:25.382427537 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:06:25.552830483 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:06:25.579411812 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:06:25.674193993 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:06:25.773966797 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n64_seed6_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed6_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:21:25 acc1_pos: 52.344 acc5_pos: 100.000 loss: 0.6944 (0.6944) acc1: 52.3438 (52.3438) acc5: 100.0000 (100.0000) time: 5.4950 data: 3.3616 max mem: 3531 +Test: [ 10/234] eta: 0:02:10 acc1_pos: 39.583 acc5_pos: 100.000 loss: 0.6944 (0.7097) acc1: 52.0833 (54.2140) acc5: 100.0000 (100.0000) time: 0.5819 data: 0.3060 max mem: 3531 +Test: [ 20/234] eta: 0:01:15 acc1_pos: 94.792 acc5_pos: 100.000 loss: 0.7944 (0.7097) acc1: 46.0938 (53.4350) acc5: 100.0000 (100.0000) time: 0.0951 data: 0.0017 max mem: 3531 +Test: [ 30/234] eta: 0:01:03 acc1_pos: 80.208 acc5_pos: 100.000 loss: 0.2350 (0.5588) acc1: 94.7917 (66.3474) acc5: 100.0000 (100.0000) time: 0.1617 data: 0.0673 max mem: 3531 +Test: [ 40/234] eta: 0:00:57 acc1_pos: 99.219 acc5_pos: 100.000 loss: 0.2186 (0.5115) acc1: 95.3125 (70.5158) acc5: 100.0000 (100.0000) time: 0.2362 data: 0.1445 max mem: 3531 +Test: [ 50/234] eta: 0:00:51 acc1_pos: 98.958 acc5_pos: 100.000 loss: 0.1395 (0.4376) acc1: 97.9167 (75.9600) acc5: 100.0000 (100.0000) time: 0.2330 data: 0.1410 max mem: 3531 +Test: [ 60/234] eta: 0:00:46 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.1169 (0.3849) acc1: 98.9583 (79.8711) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2045 data: 0.1112 max mem: 3531 +Test: [ 70/234] eta: 0:00:42 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.1157 (0.3924) acc1: 100.0000 (79.6545) acc5: 100.0000 (100.0000) acc1_neg: 66.667 acc5_neg: 100.000 time: 0.2018 data: 0.1065 max mem: 3531 +Test: [ 80/234] eta: 0:00:39 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.5587 (0.3988) acc1: 69.5312 (79.7164) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2216 data: 0.1253 max mem: 3531 +Test: [ 90/234] eta: 0:00:35 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.2047 (0.3751) acc1: 94.0104 (81.4818) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.2190 data: 0.1241 max mem: 3531 +Test: [100/234] eta: 0:00:32 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.2111 (0.3792) acc1: 91.4062 (81.3531) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1984 data: 0.1046 max mem: 3531 +Test: [110/234] eta: 0:00:29 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.4229 (0.3800) acc1: 79.1667 (81.5175) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.1749 data: 0.0816 max mem: 3531 +Test: [120/234] eta: 0:00:26 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.4116 (0.3902) acc1: 80.7292 (81.1575) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.1637 data: 0.0676 max mem: 3531 +Test: [130/234] eta: 0:00:23 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.4022 (0.3907) acc1: 84.1146 (81.6317) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.1791 data: 0.0808 max mem: 3531 +Test: [140/234] eta: 0:00:21 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.4337 (0.4188) acc1: 81.7708 (80.2878) acc5: 100.0000 (100.0000) acc1_neg: 34.375 acc5_neg: 100.000 time: 0.1961 data: 0.1017 max mem: 3531 +Test: [150/234] eta: 0:00:18 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.4674 (0.4269) acc1: 75.2604 (79.8738) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.1626 data: 0.0705 max mem: 3531 +Test: [160/234] eta: 0:00:16 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.4219 (0.4243) acc1: 79.6875 (80.2892) acc5: 100.0000 (100.0000) acc1_neg: 79.427 acc5_neg: 100.000 time: 0.1546 data: 0.0628 max mem: 3531 +Test: [170/234] eta: 0:00:13 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.4454 (0.4263) acc1: 80.4688 (80.3774) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1947 data: 0.0997 max mem: 3531 +Test: [180/234] eta: 0:00:11 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.4812 (0.4374) acc1: 78.1250 (79.6947) acc5: 100.0000 (100.0000) acc1_neg: 33.854 acc5_neg: 100.000 time: 0.2065 data: 0.1116 max mem: 3531 +Test: [190/234] eta: 0:00:09 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.6990 (0.4525) acc1: 60.9375 (78.6254) acc5: 100.0000 (100.0000) acc1_neg: 60.417 acc5_neg: 100.000 time: 0.2047 data: 0.1114 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.6832 (0.4462) acc1: 60.9375 (78.9555) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1910 data: 0.0958 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.1112 (0.4311) acc1: 99.7396 (79.9319) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1683 data: 0.0746 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.1588 (0.4201) acc1: 99.2188 (80.7068) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.1491 data: 0.0549 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.4020 (0.4410) acc1: 46.0938 (78.9942) acc5: 100.0000 (100.0000) acc1_neg: 43.490 acc5_neg: 100.000 time: 0.1374 data: 0.0438 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.8764 (0.4468) acc1: 43.4896 (78.5599) acc5: 100.0000 (100.0000) acc1_neg: 43.580 acc5_neg: 100.000 time: 0.1357 data: 0.0346 max mem: 3532 +Test: Total time: 0:00:47 (0.2042 s / it) +* Overall Acc@1 78.560 Acc@5 100.000 + target == real Acc@1 76.111 Acc@5 100.000 + target == fake Acc@1 79.250 Acc@5 100.000 + loss 0.447 + PR-AUC 0.9635 ROC-AUC 0.8783 EER 22.02% (thr=0.5205) +Accuracy of the network on the 89729 test images: 78.6% +[rank0]:[W726 04:07:24.576078902 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:07:45.144000 1184373 torch/distributed/run.py:766] +W0726 04:07:45.144000 1184373 torch/distributed/run.py:766] ***************************************** +W0726 04:07:45.144000 1184373 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:07:45.144000 1184373 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:08:14.828590510 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:08:15.318080140 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:08:15.374277528 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:08:15.425309847 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:08:15.766994629 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:08:15.837989822 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:08:15.868526427 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:08:15.896599793 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n64_seed6_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed6_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:44:10 acc1_pos: 30.990 acc5_pos: 100.000 loss: 1.0068 (1.0068) acc1: 30.9896 (30.9896) acc5: 100.0000 (100.0000) time: 7.1430 data: 5.7461 max mem: 3531 +Test: [ 10/2387] eta: 0:29:05 acc1_pos: 9.896 acc5_pos: 100.000 loss: 1.0760 (1.0773) acc1: 20.8333 (21.4252) acc5: 100.0000 (100.0000) time: 0.7341 data: 0.5228 max mem: 3531 +Test: [ 20/2387] eta: 0:17:02 acc1_pos: 9.896 acc5_pos: 100.000 loss: 1.0760 (1.0865) acc1: 17.4479 (20.3869) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:14:49 acc1_pos: 6.771 acc5_pos: 100.000 loss: 1.0979 (1.0933) acc1: 17.4479 (20.2369) acc5: 100.0000 (100.0000) time: 0.1811 data: 0.0854 max mem: 3531 +Test: [ 40/2387] eta: 0:13:25 acc1_pos: 20.312 acc5_pos: 100.000 loss: 1.1093 (1.1009) acc1: 16.1458 (18.9977) acc5: 100.0000 (100.0000) time: 0.2499 data: 0.1569 max mem: 3531 +Test: [ 50/2387] eta: 0:12:21 acc1_pos: 32.812 acc5_pos: 100.000 loss: 1.0633 (1.0962) acc1: 16.4062 (19.5976) acc5: 100.0000 (100.0000) time: 0.2245 data: 0.1314 max mem: 3531 +Test: [ 60/2387] eta: 0:11:40 acc1_pos: 45.312 acc5_pos: 100.000 loss: 1.0706 (1.0953) acc1: 20.0521 (20.0521) acc5: 100.0000 (100.0000) time: 0.2149 data: 0.1219 max mem: 3531 +Test: [ 70/2387] eta: 0:11:08 acc1_pos: 24.479 acc5_pos: 100.000 loss: 1.1456 (1.1097) acc1: 17.4479 (19.5423) acc5: 100.0000 (100.0000) time: 0.2153 data: 0.1206 max mem: 3531 +Test: [ 80/2387] eta: 0:10:54 acc1_pos: 41.667 acc5_pos: 100.000 loss: 1.0967 (1.0958) acc1: 17.1875 (20.1807) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1321 max mem: 3531 +Test: [ 90/2387] eta: 0:10:37 acc1_pos: 46.094 acc5_pos: 100.000 loss: 0.9405 (1.0717) acc1: 28.1250 (21.9408) acc5: 100.0000 (100.0000) time: 0.2386 data: 0.1422 max mem: 3531 +Test: [ 100/2387] eta: 0:10:24 acc1_pos: 33.594 acc5_pos: 100.000 loss: 0.8984 (1.0636) acc1: 33.5938 (22.3907) acc5: 100.0000 (100.0000) time: 0.2301 data: 0.1379 max mem: 3531 +Test: [ 110/2387] eta: 0:10:10 acc1_pos: 27.344 acc5_pos: 100.000 loss: 0.8986 (1.0491) acc1: 27.3438 (23.3366) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1334 max mem: 3531 +Test: [ 120/2387] eta: 0:10:25 acc1_pos: 32.292 acc5_pos: 100.000 loss: 0.9214 (1.0415) acc1: 33.3333 (24.0121) acc5: 100.0000 (100.0000) time: 0.2897 data: 0.1937 max mem: 3531 +Test: [ 130/2387] eta: 0:10:19 acc1_pos: 30.990 acc5_pos: 100.000 loss: 0.9214 (1.0331) acc1: 32.8125 (24.3957) acc5: 100.0000 (100.0000) time: 0.3099 data: 0.2106 max mem: 3531 +Test: [ 140/2387] eta: 0:10:08 acc1_pos: 28.385 acc5_pos: 100.000 loss: 0.9222 (1.0273) acc1: 27.6042 (24.6546) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1418 max mem: 3531 +Test: [ 150/2387] eta: 0:09:54 acc1_pos: 9.635 acc5_pos: 100.000 loss: 0.9178 (1.0183) acc1: 28.3854 (25.1207) acc5: 100.0000 (100.0000) time: 0.2099 data: 0.1129 max mem: 3531 +Test: [ 160/2387] eta: 0:09:39 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.8541 (0.9956) acc1: 35.6771 (27.0995) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1861 data: 0.0887 max mem: 3531 +Test: [ 170/2387] eta: 0:09:28 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.2130 (0.9492) acc1: 97.1354 (31.2622) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1849 data: 0.0881 max mem: 3531 +Test: [ 180/2387] eta: 0:09:20 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.2003 (0.9079) acc1: 98.4375 (34.9865) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2037 data: 0.1067 max mem: 3531 +Test: [ 190/2387] eta: 0:09:08 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1984 (0.8707) acc1: 98.6979 (38.3235) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1899 data: 0.0911 max mem: 3531 +Test: [ 200/2387] eta: 0:08:53 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1960 (0.8373) acc1: 98.6979 (41.3169) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1549 data: 0.0561 max mem: 3531 +Test: [ 210/2387] eta: 0:08:45 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.2011 (0.8072) acc1: 98.6979 (44.0351) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1649 data: 0.0659 max mem: 3532 +Test: [ 220/2387] eta: 0:08:39 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.2023 (0.7799) acc1: 98.6979 (46.4991) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1949 data: 0.0969 max mem: 3532 +Test: [ 230/2387] eta: 0:08:32 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.2036 (0.7549) acc1: 98.4375 (48.7486) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1973 data: 0.1030 max mem: 3532 +Test: [ 240/2387] eta: 0:08:30 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.2087 (0.7327) acc1: 98.4375 (50.8039) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2149 data: 0.1238 max mem: 3532 +Test: [ 250/2387] eta: 0:08:19 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1900 (0.7077) acc1: 98.9583 (52.7556) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1875 data: 0.0941 max mem: 3532 +Test: [ 260/2387] eta: 0:08:11 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.0900 (0.6840) acc1: 100.0000 (54.5658) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1525 data: 0.0574 max mem: 3532 +Test: [ 270/2387] eta: 0:08:04 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.0873 (0.6620) acc1: 100.0000 (56.2414) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1662 data: 0.0739 max mem: 3532 +Test: [ 280/2387] eta: 0:07:59 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.0869 (0.6417) acc1: 100.0000 (57.7986) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1793 data: 0.0872 max mem: 3532 +Test: [ 290/2387] eta: 0:07:53 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.0816 (0.6224) acc1: 100.0000 (59.2488) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1835 data: 0.0903 max mem: 3532 +Test: [ 300/2387] eta: 0:07:46 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.0788 (0.6044) acc1: 100.0000 (60.6027) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1679 data: 0.0727 max mem: 3532 +Test: [ 310/2387] eta: 0:07:42 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.0830 (0.5908) acc1: 100.0000 (61.8393) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1769 data: 0.0812 max mem: 3532 +Test: [ 320/2387] eta: 0:07:37 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.3018 (0.5820) acc1: 98.6979 (62.9795) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1868 data: 0.0937 max mem: 3532 +Test: [ 330/2387] eta: 0:07:35 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.3058 (0.5736) acc1: 98.6979 (64.0609) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2052 data: 0.1129 max mem: 3532 +Test: [ 340/2387] eta: 0:07:32 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.3013 (0.5656) acc1: 98.6979 (65.0751) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2232 data: 0.1318 max mem: 3532 +Test: [ 350/2387] eta: 0:07:28 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.2966 (0.5540) acc1: 98.9583 (66.0642) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2035 data: 0.1131 max mem: 3532 +Test: [ 360/2387] eta: 0:07:26 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1612 (0.5432) acc1: 99.7396 (66.9870) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1998 data: 0.1063 max mem: 3532 +Test: [ 370/2387] eta: 0:07:22 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1667 (0.5331) acc1: 99.4792 (67.8628) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2040 data: 0.1070 max mem: 3532 +Test: [ 380/2387] eta: 0:07:22 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1673 (0.5235) acc1: 99.4792 (68.6899) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2252 data: 0.1292 max mem: 3532 +Test: [ 390/2387] eta: 0:07:17 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1699 (0.5145) acc1: 99.4792 (69.4726) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2099 data: 0.1165 max mem: 3532 +Test: [ 400/2387] eta: 0:07:14 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1684 (0.5059) acc1: 99.4792 (70.2203) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1921 data: 0.1007 max mem: 3532 +Test: [ 410/2387] eta: 0:07:10 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1660 (0.4976) acc1: 99.4792 (70.9334) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1949 data: 0.1016 max mem: 3532 +Test: [ 420/2387] eta: 0:07:07 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1656 (0.4897) acc1: 99.7396 (71.6115) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1803 data: 0.0845 max mem: 3532 +Test: [ 430/2387] eta: 0:07:03 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1705 (0.4826) acc1: 99.2188 (72.2496) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1832 data: 0.0886 max mem: 3532 +Test: [ 440/2387] eta: 0:06:58 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1733 (0.4748) acc1: 99.4792 (72.8736) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1737 data: 0.0815 max mem: 3533 +Test: [ 450/2387] eta: 0:06:55 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1024 (0.4665) acc1: 100.0000 (73.4750) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1765 data: 0.0833 max mem: 3533 +Test: [ 460/2387] eta: 0:06:51 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1024 (0.4586) acc1: 100.0000 (74.0504) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1832 data: 0.0889 max mem: 3533 +Test: [ 470/2387] eta: 0:06:47 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1025 (0.4511) acc1: 100.0000 (74.6014) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1733 data: 0.0800 max mem: 3533 +Test: [ 480/2387] eta: 0:06:44 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1030 (0.4438) acc1: 100.0000 (75.1294) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0819 max mem: 3533 +Test: [ 490/2387] eta: 0:06:39 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1028 (0.4369) acc1: 100.0000 (75.6359) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1618 data: 0.0693 max mem: 3533 +Test: [ 500/2387] eta: 0:06:36 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1027 (0.4302) acc1: 100.0000 (76.1222) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1608 data: 0.0685 max mem: 3533 +Test: [ 510/2387] eta: 0:06:33 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1025 (0.4239) acc1: 100.0000 (76.5895) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1797 data: 0.0879 max mem: 3533 +Test: [ 520/2387] eta: 0:06:29 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1244 (0.4182) acc1: 100.0000 (77.0388) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1702 data: 0.0781 max mem: 3533 +Test: [ 530/2387] eta: 0:06:26 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1269 (0.4127) acc1: 100.0000 (77.4713) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1753 data: 0.0833 max mem: 3533 +Test: [ 540/2387] eta: 0:06:23 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1273 (0.4075) acc1: 100.0000 (77.8877) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1896 data: 0.0980 max mem: 3533 +Test: [ 550/2387] eta: 0:06:21 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1273 (0.4024) acc1: 100.0000 (78.2890) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2024 data: 0.1090 max mem: 3533 +Test: [ 560/2387] eta: 0:06:20 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1275 (0.3975) acc1: 100.0000 (78.6760) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2140 data: 0.1203 max mem: 3533 +Test: [ 570/2387] eta: 0:06:15 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1269 (0.3928) acc1: 100.0000 (79.0495) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1682 data: 0.0755 max mem: 3533 +Test: [ 580/2387] eta: 0:06:12 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1264 (0.3882) acc1: 100.0000 (79.4101) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1548 data: 0.0623 max mem: 3533 +Test: [ 590/2387] eta: 0:06:09 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1276 (0.3838) acc1: 100.0000 (79.7584) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1859 data: 0.0931 max mem: 3533 +Test: [ 600/2387] eta: 0:06:07 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1267 (0.3795) acc1: 100.0000 (80.0948) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1817 data: 0.0892 max mem: 3533 +Test: [ 610/2387] eta: 0:06:04 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1386 (0.3760) acc1: 100.0000 (80.4168) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1850 data: 0.0930 max mem: 3533 +Test: [ 620/2387] eta: 0:06:01 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1612 (0.3726) acc1: 99.7396 (80.7271) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1830 data: 0.0885 max mem: 3533 +Test: [ 630/2387] eta: 0:05:58 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1612 (0.3692) acc1: 99.7396 (81.0280) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1694 data: 0.0747 max mem: 3533 +Test: [ 640/2387] eta: 0:05:55 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1650 (0.3661) acc1: 99.7396 (81.3166) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1721 data: 0.0803 max mem: 3533 +Test: [ 650/2387] eta: 0:05:53 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1650 (0.3630) acc1: 99.4792 (81.5968) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1936 data: 0.0960 max mem: 3533 +Test: [ 660/2387] eta: 0:05:51 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1639 (0.3600) acc1: 99.7396 (81.8721) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1942 data: 0.0962 max mem: 3533 +Test: [ 670/2387] eta: 0:05:48 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1629 (0.3571) acc1: 99.7396 (82.1384) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1843 data: 0.0925 max mem: 3534 +Test: [ 680/2387] eta: 0:05:46 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1629 (0.3542) acc1: 99.7396 (82.3995) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1839 data: 0.0913 max mem: 3534 +Test: [ 690/2387] eta: 0:05:43 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1676 (0.3541) acc1: 99.7396 (82.4454) acc5: 100.0000 (100.0000) acc1_neg: 77.083 acc5_neg: 100.000 time: 0.1861 data: 0.0939 max mem: 3534 +Test: [ 700/2387] eta: 0:05:41 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.4039 (0.3549) acc1: 80.9896 (82.4165) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.1843 data: 0.0910 max mem: 3534 +Test: [ 710/2387] eta: 0:05:39 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.4071 (0.3557) acc1: 80.4688 (82.3865) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.1886 data: 0.0948 max mem: 3534 +Test: [ 720/2387] eta: 0:05:36 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.4043 (0.3563) acc1: 80.4688 (82.3686) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.1806 data: 0.0879 max mem: 3534 +Test: [ 730/2387] eta: 0:05:33 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.4129 (0.3572) acc1: 79.6875 (82.3366) acc5: 100.0000 (100.0000) acc1_neg: 79.427 acc5_neg: 100.000 time: 0.1755 data: 0.0833 max mem: 3534 +Test: [ 740/2387] eta: 0:05:31 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.4129 (0.3579) acc1: 79.9479 (82.3117) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.1795 data: 0.0882 max mem: 3534 +Test: [ 750/2387] eta: 0:05:28 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.4018 (0.3585) acc1: 80.7292 (82.2924) acc5: 100.0000 (100.0000) acc1_neg: 80.729 acc5_neg: 100.000 time: 0.1732 data: 0.0802 max mem: 3534 +Test: [ 760/2387] eta: 0:05:26 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.3966 (0.3585) acc1: 80.7292 (82.3132) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1875 data: 0.0933 max mem: 3534 +Test: [ 770/2387] eta: 0:05:24 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.2892 (0.3568) acc1: 89.3229 (82.4619) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1961 data: 0.1038 max mem: 3534 +Test: [ 780/2387] eta: 0:05:22 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1686 (0.3541) acc1: 97.6562 (82.6741) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1938 data: 0.1018 max mem: 3534 +Test: [ 790/2387] eta: 0:05:19 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1368 (0.3517) acc1: 99.4792 (82.8513) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1866 data: 0.0942 max mem: 3534 +Test: [ 800/2387] eta: 0:05:17 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1328 (0.3490) acc1: 100.0000 (83.0638) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1707 data: 0.0782 max mem: 3534 +Test: [ 810/2387] eta: 0:05:14 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1380 (0.3464) acc1: 99.7396 (83.2672) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1647 data: 0.0727 max mem: 3534 +Test: [ 820/2387] eta: 0:05:12 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1437 (0.3441) acc1: 99.4792 (83.4545) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1789 data: 0.0866 max mem: 3534 +Test: [ 830/2387] eta: 0:05:09 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1940 (0.3428) acc1: 97.9167 (83.5972) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1849 data: 0.0906 max mem: 3534 +Test: [ 840/2387] eta: 0:05:07 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.2220 (0.3417) acc1: 97.1354 (83.7538) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1797 data: 0.0857 max mem: 3534 +Test: [ 850/2387] eta: 0:05:05 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.2490 (0.3407) acc1: 95.8333 (83.8903) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1746 data: 0.0800 max mem: 3534 +Test: [ 860/2387] eta: 0:05:02 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.2348 (0.3394) acc1: 98.1771 (84.0550) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1713 data: 0.0765 max mem: 3534 +Test: [ 870/2387] eta: 0:05:00 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.2195 (0.3381) acc1: 98.9583 (84.2150) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1950 data: 0.1020 max mem: 3534 +Test: [ 880/2387] eta: 0:04:58 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.2118 (0.3370) acc1: 98.4375 (84.3570) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1834 data: 0.0896 max mem: 3534 +Test: [ 890/2387] eta: 0:04:56 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1968 (0.3353) acc1: 98.6979 (84.5197) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1785 data: 0.0837 max mem: 3534 +Test: [ 900/2387] eta: 0:04:54 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.2203 (0.3381) acc1: 97.3958 (84.3501) acc5: 100.0000 (100.0000) acc1_neg: 65.625 acc5_neg: 100.000 time: 0.1937 data: 0.0998 max mem: 3535 +Test: [ 910/2387] eta: 0:04:51 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.6704 (0.3418) acc1: 61.1979 (84.1009) acc5: 100.0000 (100.0000) acc1_neg: 61.979 acc5_neg: 100.000 time: 0.1822 data: 0.0903 max mem: 3535 +Test: [ 920/2387] eta: 0:04:50 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.6818 (0.3454) acc1: 61.1979 (83.8553) acc5: 100.0000 (100.0000) acc1_neg: 60.677 acc5_neg: 100.000 time: 0.1961 data: 0.1033 max mem: 3535 +Test: [ 930/2387] eta: 0:04:47 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.6818 (0.3490) acc1: 61.4583 (83.6172) acc5: 100.0000 (100.0000) acc1_neg: 58.854 acc5_neg: 100.000 time: 0.1926 data: 0.0995 max mem: 3535 +Test: [ 940/2387] eta: 0:04:45 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.6886 (0.3525) acc1: 61.4583 (83.3762) acc5: 100.0000 (100.0000) acc1_neg: 57.292 acc5_neg: 100.000 time: 0.1685 data: 0.0745 max mem: 3535 +Test: [ 950/2387] eta: 0:04:42 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.6624 (0.3558) acc1: 61.4583 (83.1460) acc5: 100.0000 (100.0000) acc1_neg: 61.458 acc5_neg: 100.000 time: 0.1699 data: 0.0746 max mem: 3535 +Test: [ 960/2387] eta: 0:04:40 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.6606 (0.3590) acc1: 61.9792 (82.9385) acc5: 100.0000 (100.0000) acc1_neg: 65.104 acc5_neg: 100.000 time: 0.1799 data: 0.0850 max mem: 3535 +Test: [ 970/2387] eta: 0:04:38 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.6681 (0.3623) acc1: 62.5000 (82.7186) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.1899 data: 0.0966 max mem: 3535 +Test: [ 980/2387] eta: 0:04:36 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.6565 (0.3624) acc1: 62.5000 (82.7286) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.1949 data: 0.1021 max mem: 3535 +Test: [ 990/2387] eta: 0:04:34 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1601 (0.3605) acc1: 98.6979 (82.8614) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2002 data: 0.1048 max mem: 3535 +Test: [1000/2387] eta: 0:04:32 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1278 (0.3583) acc1: 99.7396 (83.0175) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2000 data: 0.1048 max mem: 3535 +Test: [1010/2387] eta: 0:04:30 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1373 (0.3565) acc1: 99.4792 (83.1592) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1898 data: 0.0973 max mem: 3535 +Test: [1020/2387] eta: 0:04:28 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1580 (0.3549) acc1: 98.4375 (83.2951) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1898 data: 0.0977 max mem: 3535 +Test: [1030/2387] eta: 0:04:26 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1475 (0.3529) acc1: 99.4792 (83.4495) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1800 data: 0.0878 max mem: 3535 +Test: [1040/2387] eta: 0:04:23 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1277 (0.3507) acc1: 100.0000 (83.6080) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1576 data: 0.0656 max mem: 3535 +Test: [1050/2387] eta: 0:04:22 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1206 (0.3485) acc1: 100.0000 (83.7620) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0820 max mem: 3535 +Test: [1060/2387] eta: 0:04:19 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1255 (0.3465) acc1: 100.0000 (83.9126) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1832 data: 0.0898 max mem: 3535 +Test: [1070/2387] eta: 0:04:16 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1321 (0.3447) acc1: 99.7396 (84.0511) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1505 data: 0.0579 max mem: 3535 +Test: [1080/2387] eta: 0:04:14 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1818 (0.3435) acc1: 97.9167 (84.1726) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1568 data: 0.0655 max mem: 3535 +Test: [1090/2387] eta: 0:04:12 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1832 (0.3419) acc1: 97.9167 (84.3036) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0886 max mem: 3535 +Test: [1100/2387] eta: 0:04:10 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1697 (0.3403) acc1: 99.2188 (84.4339) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1698 data: 0.0778 max mem: 3535 +Test: [1110/2387] eta: 0:04:08 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1627 (0.3390) acc1: 99.2188 (84.5426) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1659 data: 0.0726 max mem: 3535 +Test: [1120/2387] eta: 0:04:05 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1754 (0.3376) acc1: 99.2188 (84.6677) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1730 data: 0.0802 max mem: 3536 +Test: [1130/2387] eta: 0:04:03 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1759 (0.3362) acc1: 98.4375 (84.7929) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1662 data: 0.0710 max mem: 3536 +Test: [1140/2387] eta: 0:04:01 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1885 (0.3350) acc1: 98.4375 (84.9088) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1850 data: 0.0877 max mem: 3536 +Test: [1150/2387] eta: 0:03:59 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.2022 (0.3340) acc1: 98.4375 (85.0176) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1947 data: 0.1015 max mem: 3536 +Test: [1160/2387] eta: 0:03:57 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1954 (0.3327) acc1: 98.4375 (85.1244) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1847 data: 0.0925 max mem: 3536 +Test: [1170/2387] eta: 0:03:55 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1805 (0.3315) acc1: 98.9583 (85.2428) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1932 data: 0.0987 max mem: 3536 +Test: [1180/2387] eta: 0:03:53 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1819 (0.3303) acc1: 99.2188 (85.3540) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1953 data: 0.0992 max mem: 3536 +Test: [1190/2387] eta: 0:03:52 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1675 (0.3290) acc1: 99.2188 (85.4691) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1940 data: 0.0971 max mem: 3536 +Test: [1200/2387] eta: 0:03:49 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1452 (0.3275) acc1: 99.7396 (85.5888) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1886 data: 0.0937 max mem: 3536 +Test: [1210/2387] eta: 0:03:47 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1370 (0.3259) acc1: 100.0000 (85.7070) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1855 data: 0.0932 max mem: 3536 +Test: [1220/2387] eta: 0:03:46 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1065 (0.3241) acc1: 100.0000 (85.8230) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1918 data: 0.1004 max mem: 3536 +Test: [1230/2387] eta: 0:03:44 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.0907 (0.3222) acc1: 100.0000 (85.9375) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2043 data: 0.1131 max mem: 3536 +Test: [1240/2387] eta: 0:03:42 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.0923 (0.3203) acc1: 100.0000 (86.0506) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1988 data: 0.1064 max mem: 3536 +Test: [1250/2387] eta: 0:03:39 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.0923 (0.3185) acc1: 100.0000 (86.1617) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1698 data: 0.0767 max mem: 3536 +Test: [1260/2387] eta: 0:03:37 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.0816 (0.3166) acc1: 100.0000 (86.2708) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1697 data: 0.0777 max mem: 3536 +Test: [1270/2387] eta: 0:03:35 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.0808 (0.3148) acc1: 100.0000 (86.3786) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1805 data: 0.0894 max mem: 3536 +Test: [1280/2387] eta: 0:03:33 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.0941 (0.3132) acc1: 100.0000 (86.4831) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1716 data: 0.0802 max mem: 3536 +Test: [1290/2387] eta: 0:03:31 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1155 (0.3117) acc1: 99.7396 (86.5769) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1763 data: 0.0845 max mem: 3536 +Test: [1300/2387] eta: 0:03:29 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1184 (0.3103) acc1: 99.7396 (86.6751) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1927 data: 0.1005 max mem: 3536 +Test: [1310/2387] eta: 0:03:27 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1192 (0.3089) acc1: 99.7396 (86.7700) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1975 data: 0.1038 max mem: 3536 +Test: [1320/2387] eta: 0:03:25 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1223 (0.3075) acc1: 99.4792 (86.8619) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1848 data: 0.0907 max mem: 3536 +Test: [1330/2387] eta: 0:03:23 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1378 (0.3064) acc1: 99.7396 (86.9514) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1827 data: 0.0906 max mem: 3536 +Test: [1340/2387] eta: 0:03:21 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1753 (0.3061) acc1: 97.6562 (86.9799) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0907 max mem: 3536 +Test: [1350/2387] eta: 0:03:19 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.0769 (0.3044) acc1: 100.0000 (87.0763) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1825 data: 0.0880 max mem: 3537 +Test: [1360/2387] eta: 0:03:18 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.0692 (0.3027) acc1: 100.0000 (87.1713) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1964 data: 0.1050 max mem: 3537 +Test: [1370/2387] eta: 0:03:16 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.0684 (0.3010) acc1: 100.0000 (87.2648) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1911 data: 0.0982 max mem: 3537 +Test: [1380/2387] eta: 0:03:14 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.0641 (0.2993) acc1: 100.0000 (87.3571) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0869 max mem: 3537 +Test: [1390/2387] eta: 0:03:12 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.0626 (0.2976) acc1: 100.0000 (87.4480) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1833 data: 0.0888 max mem: 3537 +Test: [1400/2387] eta: 0:03:10 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.0606 (0.2959) acc1: 100.0000 (87.5375) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2084 data: 0.1124 max mem: 3537 +Test: [1410/2387] eta: 0:03:08 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.0666 (0.2948) acc1: 100.0000 (87.6117) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1949 data: 0.0985 max mem: 3537 +Test: [1420/2387] eta: 0:03:06 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1745 (0.2941) acc1: 98.6979 (87.6737) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1765 data: 0.0794 max mem: 3537 +Test: [1430/2387] eta: 0:03:04 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1762 (0.2933) acc1: 98.1771 (87.7459) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2020 data: 0.1083 max mem: 3537 +Test: [1440/2387] eta: 0:03:02 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1762 (0.2925) acc1: 98.1771 (87.8206) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1933 data: 0.1012 max mem: 3537 +Test: [1450/2387] eta: 0:03:00 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1810 (0.2919) acc1: 98.6979 (87.8886) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1856 data: 0.0929 max mem: 3537 +Test: [1460/2387] eta: 0:02:58 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1525 (0.2910) acc1: 98.6979 (87.9615) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2149 data: 0.1215 max mem: 3537 +Test: [1470/2387] eta: 0:02:56 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1345 (0.2899) acc1: 99.7396 (88.0424) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1987 data: 0.1046 max mem: 3537 +Test: [1480/2387] eta: 0:02:54 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1231 (0.2888) acc1: 99.7396 (88.1193) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1721 data: 0.0788 max mem: 3537 +Test: [1490/2387] eta: 0:02:52 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1315 (0.2878) acc1: 99.7396 (88.1969) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1849 data: 0.0923 max mem: 3537 +Test: [1500/2387] eta: 0:02:50 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1329 (0.2867) acc1: 99.7396 (88.2724) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1951 data: 0.1030 max mem: 3537 +Test: [1510/2387] eta: 0:02:49 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1390 (0.2858) acc1: 99.7396 (88.3488) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1981 data: 0.1045 max mem: 3537 +Test: [1520/2387] eta: 0:02:47 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1659 (0.2854) acc1: 99.4792 (88.4105) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1893 data: 0.0954 max mem: 3537 +Test: [1530/2387] eta: 0:02:45 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.2222 (0.2851) acc1: 97.6562 (88.4619) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1905 data: 0.0986 max mem: 3537 +Test: [1540/2387] eta: 0:02:43 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.2204 (0.2846) acc1: 98.6979 (88.5261) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1929 data: 0.1010 max mem: 3537 +Test: [1550/2387] eta: 0:02:41 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.2055 (0.2841) acc1: 99.2188 (88.5899) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1863 data: 0.0923 max mem: 3537 +Test: [1560/2387] eta: 0:02:39 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1961 (0.2836) acc1: 99.2188 (88.6491) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1724 data: 0.0784 max mem: 3537 +Test: [1570/2387] eta: 0:02:37 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1669 (0.2828) acc1: 99.4792 (88.7152) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1789 data: 0.0870 max mem: 3537 +Test: [1580/2387] eta: 0:02:35 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1564 (0.2821) acc1: 99.4792 (88.7826) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1891 data: 0.0977 max mem: 3538 +Test: [1590/2387] eta: 0:02:33 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1438 (0.2812) acc1: 99.4792 (88.8509) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1727 data: 0.0795 max mem: 3538 +Test: [1600/2387] eta: 0:02:31 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1274 (0.2804) acc1: 99.7396 (88.9028) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1637 data: 0.0701 max mem: 3538 +Test: [1610/2387] eta: 0:02:29 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1247 (0.2795) acc1: 99.7396 (88.9692) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1644 data: 0.0728 max mem: 3538 +Test: [1620/2387] eta: 0:02:27 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1396 (0.2786) acc1: 99.7396 (89.0304) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1753 data: 0.0841 max mem: 3538 +Test: [1630/2387] eta: 0:02:25 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1594 (0.2781) acc1: 99.2188 (89.0833) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1837 data: 0.0914 max mem: 3538 +Test: [1640/2387] eta: 0:02:23 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.2066 (0.2778) acc1: 97.1354 (89.1238) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1790 data: 0.0868 max mem: 3538 +Test: [1650/2387] eta: 0:02:21 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.2108 (0.2774) acc1: 96.8750 (89.1726) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1635 data: 0.0724 max mem: 3538 +Test: [1660/2387] eta: 0:02:19 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.2066 (0.2770) acc1: 97.1354 (89.2201) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1558 data: 0.0647 max mem: 3538 +Test: [1670/2387] eta: 0:02:17 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.2115 (0.2766) acc1: 97.1354 (89.2663) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1641 data: 0.0732 max mem: 3538 +Test: [1680/2387] eta: 0:02:14 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.2115 (0.2763) acc1: 97.3958 (89.3084) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1639 data: 0.0697 max mem: 3538 +Test: [1690/2387] eta: 0:02:12 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.2160 (0.2760) acc1: 97.3958 (89.3459) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1625 data: 0.0673 max mem: 3538 +Test: [1700/2387] eta: 0:02:10 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.2109 (0.2754) acc1: 97.1354 (89.3946) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1549 data: 0.0601 max mem: 3538 +Test: [1710/2387] eta: 0:02:08 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1155 (0.2744) acc1: 99.4792 (89.4552) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1691 data: 0.0752 max mem: 3538 +Test: [1720/2387] eta: 0:02:06 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1030 (0.2735) acc1: 100.0000 (89.5158) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1801 data: 0.0888 max mem: 3538 +Test: [1730/2387] eta: 0:02:04 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1029 (0.2725) acc1: 100.0000 (89.5749) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1546 data: 0.0604 max mem: 3538 +Test: [1740/2387] eta: 0:02:02 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.0981 (0.2715) acc1: 100.0000 (89.6337) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1547 data: 0.0586 max mem: 3538 +Test: [1750/2387] eta: 0:02:00 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.0898 (0.2705) acc1: 100.0000 (89.6925) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1602 data: 0.0665 max mem: 3538 +Test: [1760/2387] eta: 0:01:58 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.0900 (0.2695) acc1: 100.0000 (89.7509) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1752 data: 0.0820 max mem: 3538 +Test: [1770/2387] eta: 0:01:57 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1077 (0.2687) acc1: 100.0000 (89.8048) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1845 data: 0.0911 max mem: 3538 +Test: [1780/2387] eta: 0:01:55 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1262 (0.2679) acc1: 99.7396 (89.8610) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1621 data: 0.0705 max mem: 3538 +Test: [1790/2387] eta: 0:01:53 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1160 (0.2670) acc1: 100.0000 (89.9165) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1694 data: 0.0770 max mem: 3538 +Test: [1800/2387] eta: 0:01:51 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1123 (0.2661) acc1: 100.0000 (89.9706) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1788 data: 0.0844 max mem: 3538 +Test: [1810/2387] eta: 0:01:49 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1039 (0.2652) acc1: 100.0000 (90.0248) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1740 data: 0.0792 max mem: 3539 +Test: [1820/2387] eta: 0:01:47 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.0933 (0.2642) acc1: 100.0000 (90.0794) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1873 data: 0.0932 max mem: 3539 +Test: [1830/2387] eta: 0:01:45 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1049 (0.2635) acc1: 100.0000 (90.1333) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1921 data: 0.0979 max mem: 3539 +Test: [1840/2387] eta: 0:01:43 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1452 (0.2628) acc1: 100.0000 (90.1869) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1762 data: 0.0829 max mem: 3539 +Test: [1850/2387] eta: 0:01:41 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1460 (0.2622) acc1: 100.0000 (90.2398) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1659 data: 0.0744 max mem: 3539 +Test: [1860/2387] eta: 0:01:39 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1457 (0.2616) acc1: 100.0000 (90.2922) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1777 data: 0.0852 max mem: 3539 +Test: [1870/2387] eta: 0:01:37 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1451 (0.2609) acc1: 100.0000 (90.3441) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1784 data: 0.0858 max mem: 3539 +Test: [1880/2387] eta: 0:01:35 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1449 (0.2603) acc1: 100.0000 (90.3955) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1751 data: 0.0815 max mem: 3539 +Test: [1890/2387] eta: 0:01:33 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1449 (0.2597) acc1: 100.0000 (90.4462) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1982 data: 0.1038 max mem: 3539 +Test: [1900/2387] eta: 0:01:32 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1468 (0.2591) acc1: 100.0000 (90.4965) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1988 data: 0.1058 max mem: 3539 +Test: [1910/2387] eta: 0:01:30 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1613 (0.2587) acc1: 100.0000 (90.5453) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1872 data: 0.0951 max mem: 3539 +Test: [1920/2387] eta: 0:01:28 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1665 (0.2584) acc1: 100.0000 (90.5735) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1801 data: 0.0862 max mem: 3539 +Test: [1930/2387] eta: 0:01:26 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1457 (0.2577) acc1: 100.0000 (90.6219) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1857 data: 0.0916 max mem: 3539 +Test: [1940/2387] eta: 0:01:24 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1441 (0.2574) acc1: 100.0000 (90.6339) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1901 data: 0.0987 max mem: 3539 +Test: [1950/2387] eta: 0:01:22 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1593 (0.2570) acc1: 99.2188 (90.6736) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2087 data: 0.1178 max mem: 3539 +Test: [1960/2387] eta: 0:01:20 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1626 (0.2566) acc1: 99.2188 (90.7109) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2090 data: 0.1169 max mem: 3539 +Test: [1970/2387] eta: 0:01:18 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.2021 (0.2566) acc1: 97.1354 (90.7196) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1825 data: 0.0907 max mem: 3539 +Test: [1980/2387] eta: 0:01:16 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.2231 (0.2564) acc1: 95.3125 (90.7491) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1655 data: 0.0740 max mem: 3539 +Test: [1990/2387] eta: 0:01:14 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1759 (0.2561) acc1: 99.2188 (90.7910) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1553 data: 0.0623 max mem: 3539 +Test: [2000/2387] eta: 0:01:13 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1759 (0.2557) acc1: 98.9583 (90.8324) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1729 data: 0.0798 max mem: 3539 +Test: [2010/2387] eta: 0:01:11 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1794 (0.2553) acc1: 98.9583 (90.8726) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1706 data: 0.0792 max mem: 3539 +Test: [2020/2387] eta: 0:01:09 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1797 (0.2549) acc1: 98.9583 (90.9129) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1675 data: 0.0764 max mem: 3539 +Test: [2030/2387] eta: 0:01:07 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1768 (0.2545) acc1: 99.2188 (90.9531) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1773 data: 0.0854 max mem: 3540 +Test: [2040/2387] eta: 0:01:05 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1768 (0.2542) acc1: 99.2188 (90.9939) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1872 data: 0.0940 max mem: 3540 +Test: [2050/2387] eta: 0:01:03 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1750 (0.2538) acc1: 99.2188 (91.0345) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1928 data: 0.1004 max mem: 3540 +Test: [2060/2387] eta: 0:01:01 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1761 (0.2534) acc1: 99.2188 (91.0737) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1928 data: 0.1013 max mem: 3540 +Test: [2070/2387] eta: 0:00:59 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1723 (0.2530) acc1: 99.2188 (91.1143) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1940 data: 0.1005 max mem: 3540 +Test: [2080/2387] eta: 0:00:57 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1720 (0.2526) acc1: 99.4792 (91.1542) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1954 data: 0.1006 max mem: 3540 +Test: [2090/2387] eta: 0:00:55 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1697 (0.2522) acc1: 99.4792 (91.1933) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1828 data: 0.0902 max mem: 3540 +Test: [2100/2387] eta: 0:00:54 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1698 (0.2518) acc1: 99.4792 (91.2327) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1816 data: 0.0898 max mem: 3540 +Test: [2110/2387] eta: 0:00:52 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1718 (0.2514) acc1: 99.4792 (91.2710) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1946 data: 0.1026 max mem: 3540 +Test: [2120/2387] eta: 0:00:50 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1705 (0.2511) acc1: 99.2188 (91.3031) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1881 data: 0.0964 max mem: 3540 +Test: [2130/2387] eta: 0:00:48 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1570 (0.2506) acc1: 99.7396 (91.3421) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1901 data: 0.0985 max mem: 3540 +Test: [2140/2387] eta: 0:00:46 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1474 (0.2503) acc1: 99.7396 (91.3640) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1791 data: 0.0879 max mem: 3540 +Test: [2150/2387] eta: 0:00:44 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1557 (0.2499) acc1: 99.4792 (91.4018) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1570 data: 0.0639 max mem: 3540 +Test: [2160/2387] eta: 0:00:42 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1591 (0.2495) acc1: 99.4792 (91.4363) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1587 data: 0.0651 max mem: 3540 +Test: [2170/2387] eta: 0:00:40 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1727 (0.2493) acc1: 98.6979 (91.4596) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1799 data: 0.0866 max mem: 3540 +Test: [2180/2387] eta: 0:00:38 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.2601 (0.2502) acc1: 94.0104 (91.4053) acc5: 100.0000 (100.0000) acc1_neg: 55.729 acc5_neg: 100.000 time: 0.1887 data: 0.0928 max mem: 3540 +Test: [2190/2387] eta: 0:00:37 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.2900 (0.2506) acc1: 91.4062 (91.3862) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1449 data: 0.0467 max mem: 3540 +Test: [2200/2387] eta: 0:00:35 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1510 (0.2501) acc1: 99.2188 (91.4235) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1507 data: 0.0535 max mem: 3540 +Test: [2210/2387] eta: 0:00:33 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1452 (0.2496) acc1: 99.4792 (91.4602) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1873 data: 0.0932 max mem: 3540 +Test: [2220/2387] eta: 0:00:31 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1460 (0.2491) acc1: 99.4792 (91.4949) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1941 data: 0.1008 max mem: 3540 +Test: [2230/2387] eta: 0:00:29 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1330 (0.2486) acc1: 100.0000 (91.5317) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1937 data: 0.0988 max mem: 3540 +Test: [2240/2387] eta: 0:00:27 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1250 (0.2481) acc1: 99.7396 (91.5646) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1914 data: 0.0978 max mem: 3540 +Test: [2250/2387] eta: 0:00:25 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1492 (0.2477) acc1: 99.2188 (91.5973) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1899 data: 0.0981 max mem: 3540 +Test: [2260/2387] eta: 0:00:23 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1521 (0.2473) acc1: 100.0000 (91.6338) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1907 data: 0.0992 max mem: 3541 +Test: [2270/2387] eta: 0:00:21 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1321 (0.2467) acc1: 100.0000 (91.6705) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1890 data: 0.0973 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1198 (0.2463) acc1: 100.0000 (91.6921) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1798 data: 0.0878 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1174 (0.2458) acc1: 100.0000 (91.7280) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1797 data: 0.0883 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1275 (0.2453) acc1: 100.0000 (91.7612) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1880 data: 0.0965 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1387 (0.2450) acc1: 99.7396 (91.7916) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1920 data: 0.0999 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1716 (0.2448) acc1: 98.6979 (91.8162) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1767 data: 0.0842 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1975 (0.2447) acc1: 98.4375 (91.8401) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1805 data: 0.0885 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1961 (0.2444) acc1: 98.4375 (91.8678) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1827 data: 0.0898 max mem: 3541 +Test: [2350/2387] eta: 0:00:06 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.1816 (0.2444) acc1: 98.4375 (91.8767) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1643 data: 0.0712 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.2160 (0.2443) acc1: 98.1771 (91.8940) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1502 data: 0.0579 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.2247 (0.2443) acc1: 97.3958 (91.9064) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.1396 data: 0.0482 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.2292 (0.2444) acc1: 95.0521 (91.9140) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1359 data: 0.0452 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 36.073 acc5_pos: 100.000 loss: 0.2919 (0.2446) acc1: 91.6667 (91.9112) acc5: 100.0000 (100.0000) acc1_neg: 88.372 acc5_neg: 100.000 time: 0.1301 data: 0.0418 max mem: 3541 +Test: Total time: 0:07:26 (0.1870 s / it) +* Overall Acc@1 91.911 Acc@5 100.000 + target == real Acc@1 25.365 Acc@5 100.000 + target == fake Acc@1 96.584 Acc@5 100.000 + loss 0.245 + PR-AUC 0.9925 ROC-AUC 0.9015 EER 16.78% (thr=0.7656) +Accuracy of the network on the 916267 test images: 91.9% +[rank0]:[W726 04:16:07.346215042 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:16:27.070000 1185005 torch/distributed/run.py:766] +W0726 04:16:27.070000 1185005 torch/distributed/run.py:766] ***************************************** +W0726 04:16:27.070000 1185005 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:16:27.070000 1185005 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 04:16:57.174504061 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:16:57.386333125 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +[rank4]:[W726 04:16:57.869191594 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:16:57.874130321 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:16:57.923392979 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:16:57.086223955 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:16:57.110171888 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:16:57.167245039 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n64_seed6_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed6_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 0:58:59 acc1_pos: 30.990 acc5_pos: 100.000 loss: 1.0068 (1.0068) acc1: 30.9896 (30.9896) acc5: 100.0000 (100.0000) time: 6.8458 data: 5.0772 max mem: 3531 +Test: [ 10/517] eta: 0:05:58 acc1_pos: 9.896 acc5_pos: 100.000 loss: 1.0760 (1.0773) acc1: 20.8333 (21.4252) acc5: 100.0000 (100.0000) time: 0.7067 data: 0.4619 max mem: 3531 +Test: [ 20/517] eta: 0:03:26 acc1_pos: 9.896 acc5_pos: 100.000 loss: 1.0760 (1.0865) acc1: 17.4479 (20.3869) acc5: 100.0000 (100.0000) time: 0.0936 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:02:53 acc1_pos: 6.771 acc5_pos: 100.000 loss: 1.0979 (1.0933) acc1: 17.4479 (20.2369) acc5: 100.0000 (100.0000) time: 0.1642 data: 0.0692 max mem: 3531 +Test: [ 40/517] eta: 0:02:38 acc1_pos: 20.312 acc5_pos: 100.000 loss: 1.1093 (1.1009) acc1: 16.1458 (18.9977) acc5: 100.0000 (100.0000) time: 0.2457 data: 0.1520 max mem: 3531 +Test: [ 50/517] eta: 0:02:27 acc1_pos: 24.312 acc5_pos: 100.000 loss: 1.0633 (1.0628) acc1: 16.4062 (22.8452) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2541 data: 0.1617 max mem: 3531 +Test: [ 60/517] eta: 0:02:21 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.3529 (0.9413) acc1: 91.1458 (34.9129) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2640 data: 0.1706 max mem: 3531 +Test: [ 70/517] eta: 0:02:14 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.3035 (0.8514) acc1: 98.4375 (43.8674) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2642 data: 0.1718 max mem: 3531 +Test: [ 80/517] eta: 0:02:10 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.3035 (0.7849) acc1: 98.6979 (50.5723) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2652 data: 0.1743 max mem: 3531 +Test: [ 90/517] eta: 0:02:04 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.3256 (0.7346) acc1: 98.1771 (55.8207) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2603 data: 0.1692 max mem: 3531 +Test: [100/517] eta: 0:02:00 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.2851 (0.6888) acc1: 98.9583 (60.1408) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2496 data: 0.1560 max mem: 3531 +Test: [110/517] eta: 0:01:57 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.2558 (0.6490) acc1: 99.7396 (63.6965) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2651 data: 0.1709 max mem: 3531 +Test: [120/517] eta: 0:01:52 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.2725 (0.6217) acc1: 98.9583 (66.4622) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2499 data: 0.1556 max mem: 3531 +Test: [130/517] eta: 0:01:49 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.3151 (0.5972) acc1: 97.1354 (68.8395) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2597 data: 0.1655 max mem: 3531 +Test: [140/517] eta: 0:01:47 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.2780 (0.5739) acc1: 99.4792 (71.0310) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2916 data: 0.1999 max mem: 3531 +Test: [150/517] eta: 0:01:43 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.2623 (0.5534) acc1: 100.0000 (72.9305) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2655 data: 0.1736 max mem: 3531 +Test: [160/517] eta: 0:01:40 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.2668 (0.5359) acc1: 99.7396 (74.5325) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2532 data: 0.1601 max mem: 3531 +Test: [170/517] eta: 0:01:36 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.2593 (0.5191) acc1: 99.4792 (75.9960) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2634 data: 0.1668 max mem: 3531 +Test: [180/517] eta: 0:01:33 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.2507 (0.5044) acc1: 99.4792 (77.2114) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2599 data: 0.1612 max mem: 3531 +Test: [190/517] eta: 0:01:30 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.2117 (0.4887) acc1: 99.7396 (78.4031) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1596 max mem: 3531 +Test: [200/517] eta: 0:01:27 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.2080 (0.4764) acc1: 100.0000 (79.4672) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2499 data: 0.1556 max mem: 3531 +Test: [210/517] eta: 0:01:23 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.2685 (0.4700) acc1: 98.4375 (80.1639) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2456 data: 0.1509 max mem: 3532 +Test: [220/517] eta: 0:01:20 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.3346 (0.4639) acc1: 93.7500 (80.7904) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2449 data: 0.1491 max mem: 3532 +Test: [230/517] eta: 0:01:17 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.3239 (0.4574) acc1: 96.6146 (81.5668) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2356 data: 0.1398 max mem: 3532 +Test: [240/517] eta: 0:01:14 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.3058 (0.4509) acc1: 99.7396 (82.3187) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2354 data: 0.1428 max mem: 3532 +Test: [250/517] eta: 0:01:12 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.2839 (0.4437) acc1: 99.7396 (83.0055) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2636 data: 0.1701 max mem: 3532 +Test: [260/517] eta: 0:01:09 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.2580 (0.4367) acc1: 99.7396 (83.6406) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2593 data: 0.1618 max mem: 3532 +Test: [270/517] eta: 0:01:06 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.2911 (0.4321) acc1: 99.4792 (84.2203) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2462 data: 0.1503 max mem: 3532 +Test: [280/517] eta: 0:01:03 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.3005 (0.4273) acc1: 99.2188 (84.7438) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2699 data: 0.1769 max mem: 3532 +Test: [290/517] eta: 0:01:00 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.2899 (0.4225) acc1: 99.4792 (85.2466) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2645 data: 0.1690 max mem: 3532 +Test: [300/517] eta: 0:00:58 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.2798 (0.4178) acc1: 99.7396 (85.7281) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2554 data: 0.1606 max mem: 3532 +Test: [310/517] eta: 0:00:55 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.2912 (0.4145) acc1: 99.7396 (86.1552) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2590 data: 0.1647 max mem: 3532 +Test: [320/517] eta: 0:00:52 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.2954 (0.4102) acc1: 99.4792 (86.5735) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2556 data: 0.1607 max mem: 3532 +Test: [330/517] eta: 0:00:49 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.2475 (0.4050) acc1: 100.0000 (86.9760) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2548 data: 0.1611 max mem: 3532 +Test: [340/517] eta: 0:00:47 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.2374 (0.4017) acc1: 100.0000 (87.2442) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.2487 data: 0.1543 max mem: 3532 +Test: [350/517] eta: 0:00:44 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.2972 (0.4000) acc1: 97.1354 (87.5141) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2449 data: 0.1506 max mem: 3532 +Test: [360/517] eta: 0:00:41 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.3043 (0.3973) acc1: 97.6562 (87.7979) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2511 data: 0.1576 max mem: 3532 +Test: [370/517] eta: 0:00:39 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.2706 (0.3932) acc1: 99.2188 (88.1198) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2899 data: 0.1967 max mem: 3532 +Test: [380/517] eta: 0:00:36 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.2505 (0.3901) acc1: 99.4792 (88.3783) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2737 data: 0.1780 max mem: 3532 +Test: [390/517] eta: 0:00:33 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.2757 (0.3874) acc1: 99.2188 (88.6556) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2299 data: 0.1307 max mem: 3532 +Test: [400/517] eta: 0:00:30 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.2768 (0.3842) acc1: 99.7396 (88.9248) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2360 data: 0.1387 max mem: 3532 +Test: [410/517] eta: 0:00:28 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.2794 (0.3830) acc1: 99.2188 (89.1208) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2449 data: 0.1517 max mem: 3532 +Test: [420/517] eta: 0:00:25 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.3113 (0.3811) acc1: 99.2188 (89.3662) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2500 data: 0.1573 max mem: 3532 +Test: [430/517] eta: 0:00:22 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.2900 (0.3787) acc1: 99.7396 (89.5870) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2454 data: 0.1510 max mem: 3532 +Test: [440/517] eta: 0:00:20 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.2939 (0.3772) acc1: 98.6979 (89.7859) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2986 data: 0.2021 max mem: 3533 +Test: [450/517] eta: 0:00:17 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.2912 (0.3752) acc1: 98.9583 (90.0014) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.3044 data: 0.2050 max mem: 3533 +Test: [460/517] eta: 0:00:15 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.2860 (0.3734) acc1: 99.4792 (90.1725) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.3349 data: 0.2357 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.2768 (0.3714) acc1: 99.4792 (90.3640) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2899 data: 0.1937 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.2768 (0.3692) acc1: 99.7396 (90.5600) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2049 data: 0.1079 max mem: 3533 +Test: [490/517] eta: 0:00:07 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.2673 (0.3669) acc1: 99.7396 (90.7332) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2257 data: 0.1294 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.3174 (0.3670) acc1: 99.2188 (90.8657) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1982 data: 0.1055 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.3397 (0.3655) acc1: 98.4375 (90.9812) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1731 data: 0.0825 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.2382 (0.3640) acc1: 99.7396 (91.0809) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1598 data: 0.0681 max mem: 3533 +Test: Total time: 0:02:14 (0.2605 s / it) +* Overall Acc@1 91.081 Acc@5 100.000 + target == real Acc@1 19.088 Acc@5 100.000 + target == fake Acc@1 98.549 Acc@5 100.000 + loss 0.364 + PR-AUC 0.9559 ROC-AUC 0.7470 EER 31.58% (thr=0.7324) +Accuracy of the network on the 198427 test images: 91.1% +[rank0]:[W726 04:19:23.651951577 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 04:19:45.754000 1185565 torch/distributed/run.py:766] +W0726 04:19:45.754000 1185565 torch/distributed/run.py:766] ***************************************** +W0726 04:19:45.754000 1185565 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:19:45.754000 1185565 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:20:16.145594265 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:20:17.318954994 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:20:17.576893624 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:20:17.595339211 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:20:17.660623497 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:20:17.809974320 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:20:17.821452368 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:20:17.847832795 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n64_seed6_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed6_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:30:56 acc1_pos: 56.771 acc5_pos: 100.000 loss: 0.6743 (0.6743) acc1: 56.7708 (56.7708) acc5: 100.0000 (100.0000) time: 4.7731 data: 2.3693 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/389] eta: 0:03:15 acc1_pos: 90.625 acc5_pos: 100.000 loss: 0.6148 (0.6041) acc1: 61.1979 (63.9441) acc5: 100.0000 (100.0000) time: 0.5159 data: 0.2157 max mem: 3531 +Test: [ 20/389] eta: 0:01:55 acc1_pos: 40.625 acc5_pos: 100.000 loss: 0.6014 (0.6185) acc1: 61.1979 (62.0784) acc5: 100.0000 (100.0000) time: 0.0902 data: 0.0004 max mem: 3531 +Test: [ 30/389] eta: 0:01:43 acc1_pos: 66.927 acc5_pos: 100.000 loss: 0.7041 (0.6559) acc1: 54.1667 (57.6025) acc5: 100.0000 (100.0000) time: 0.1652 data: 0.0705 max mem: 3531 +Test: [ 40/389] eta: 0:01:42 acc1_pos: 15.625 acc5_pos: 100.000 loss: 0.7015 (0.6721) acc1: 54.1667 (55.6529) acc5: 100.0000 (100.0000) time: 0.2749 data: 0.1751 max mem: 3531 +Test: [ 50/389] eta: 0:01:34 acc1_pos: 57.031 acc5_pos: 100.000 loss: 0.6528 (0.6753) acc1: 56.5104 (55.8772) acc5: 100.0000 (100.0000) time: 0.2584 data: 0.1630 max mem: 3531 +Test: [ 60/389] eta: 0:01:29 acc1_pos: 62.760 acc5_pos: 100.000 loss: 0.6936 (0.6618) acc1: 54.6875 (58.0003) acc5: 100.0000 (100.0000) time: 0.2245 data: 0.1339 max mem: 3531 +Test: [ 70/389] eta: 0:01:23 acc1_pos: 45.052 acc5_pos: 100.000 loss: 0.5883 (0.6523) acc1: 66.1458 (59.2503) acc5: 100.0000 (100.0000) time: 0.2205 data: 0.1293 max mem: 3531 +Test: [ 80/389] eta: 0:01:18 acc1_pos: 93.750 acc5_pos: 100.000 loss: 0.6182 (0.6549) acc1: 66.1458 (59.5358) acc5: 100.0000 (100.0000) time: 0.1944 data: 0.1014 max mem: 3531 +Test: [ 90/389] eta: 0:01:13 acc1_pos: 76.823 acc5_pos: 100.000 loss: 0.3695 (0.6266) acc1: 80.4688 (62.6030) acc5: 100.0000 (100.0000) time: 0.1967 data: 0.1033 max mem: 3531 +Test: [100/389] eta: 0:01:09 acc1_pos: 60.417 acc5_pos: 100.000 loss: 0.5872 (0.6364) acc1: 67.7083 (61.4145) acc5: 100.0000 (100.0000) time: 0.1854 data: 0.0920 max mem: 3531 +Test: [110/389] eta: 0:01:05 acc1_pos: 97.396 acc5_pos: 100.000 loss: 0.7103 (0.6445) acc1: 51.0417 (60.5363) acc5: 100.0000 (100.0000) time: 0.1801 data: 0.0875 max mem: 3531 +Test: [120/389] eta: 0:01:01 acc1_pos: 73.438 acc5_pos: 100.000 loss: 0.5500 (0.6371) acc1: 67.7083 (61.5078) acc5: 100.0000 (100.0000) time: 0.1743 data: 0.0818 max mem: 3531 +Test: [130/389] eta: 0:00:58 acc1_pos: 34.356 acc5_pos: 100.000 loss: 0.5500 (0.6354) acc1: 67.4479 (61.8957) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1696 data: 0.0760 max mem: 3531 +Test: [140/389] eta: 0:00:55 acc1_pos: 34.356 acc5_pos: 100.000 loss: 0.5819 (0.6336) acc1: 67.4479 (62.4834) acc5: 100.0000 (100.0000) acc1_neg: 25.781 acc5_neg: 100.000 time: 0.1899 data: 0.0977 max mem: 3531 +Test: [150/389] eta: 0:00:52 acc1_pos: 34.356 acc5_pos: 100.000 loss: 0.5819 (0.6351) acc1: 72.6562 (62.3172) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1743 data: 0.0829 max mem: 3531 +Test: [160/389] eta: 0:00:49 acc1_pos: 34.356 acc5_pos: 100.000 loss: 0.6052 (0.6272) acc1: 62.2396 (63.3670) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1672 data: 0.0759 max mem: 3531 +Test: [170/389] eta: 0:00:46 acc1_pos: 34.356 acc5_pos: 100.000 loss: 0.3562 (0.6178) acc1: 98.4375 (64.6336) acc5: 100.0000 (100.0000) acc1_neg: 74.740 acc5_neg: 100.000 time: 0.1786 data: 0.0852 max mem: 3531 +Test: [180/389] eta: 0:00:44 acc1_pos: 34.356 acc5_pos: 100.000 loss: 0.4178 (0.6117) acc1: 87.2396 (65.7041) acc5: 100.0000 (100.0000) acc1_neg: 68.229 acc5_neg: 100.000 time: 0.1812 data: 0.0860 max mem: 3531 +Test: [190/389] eta: 0:00:42 acc1_pos: 34.356 acc5_pos: 100.000 loss: 0.5174 (0.6155) acc1: 80.4688 (65.2610) acc5: 100.0000 (100.0000) acc1_neg: 25.521 acc5_neg: 100.000 time: 0.1899 data: 0.0964 max mem: 3531 +Test: [200/389] eta: 0:00:39 acc1_pos: 34.356 acc5_pos: 100.000 loss: 0.6136 (0.6168) acc1: 66.4062 (65.2156) acc5: 100.0000 (100.0000) acc1_neg: 65.885 acc5_neg: 100.000 time: 0.1666 data: 0.0727 max mem: 3531 +Test: [210/389] eta: 0:00:36 acc1_pos: 34.356 acc5_pos: 100.000 loss: 0.4783 (0.6068) acc1: 75.2604 (66.2470) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1449 data: 0.0507 max mem: 3532 +Test: [220/389] eta: 0:00:34 acc1_pos: 34.356 acc5_pos: 100.000 loss: 0.3114 (0.5957) acc1: 99.2188 (67.3784) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.1576 data: 0.0629 max mem: 3532 +Test: [230/389] eta: 0:00:31 acc1_pos: 34.356 acc5_pos: 100.000 loss: 0.4773 (0.5960) acc1: 73.1771 (67.3228) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1563 data: 0.0628 max mem: 3532 +Test: [240/389] eta: 0:00:29 acc1_pos: 34.356 acc5_pos: 100.000 loss: 0.5174 (0.5920) acc1: 80.7292 (67.9947) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.1620 data: 0.0705 max mem: 3532 +Test: [250/389] eta: 0:00:27 acc1_pos: 34.356 acc5_pos: 100.000 loss: 0.4560 (0.5846) acc1: 86.1979 (69.0498) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1685 data: 0.0742 max mem: 3532 +Test: [260/389] eta: 0:00:25 acc1_pos: 34.356 acc5_pos: 100.000 loss: 0.4880 (0.5833) acc1: 86.1979 (69.3856) acc5: 100.0000 (100.0000) acc1_neg: 26.042 acc5_neg: 100.000 time: 0.1670 data: 0.0734 max mem: 3532 +Test: [270/389] eta: 0:00:23 acc1_pos: 34.356 acc5_pos: 100.000 loss: 0.5754 (0.5882) acc1: 75.5208 (68.7298) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.1858 data: 0.0945 max mem: 3532 +Test: [280/389] eta: 0:00:21 acc1_pos: 34.356 acc5_pos: 100.000 loss: 0.5981 (0.5905) acc1: 66.6667 (68.5128) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1808 data: 0.0881 max mem: 3532 +Test: [290/389] eta: 0:00:19 acc1_pos: 34.356 acc5_pos: 100.000 loss: 0.5410 (0.5869) acc1: 77.8646 (69.0919) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1614 data: 0.0684 max mem: 3532 +Test: [300/389] eta: 0:00:17 acc1_pos: 34.356 acc5_pos: 100.000 loss: 0.5260 (0.5857) acc1: 86.1979 (69.5287) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.1723 data: 0.0789 max mem: 3532 +Test: [310/389] eta: 0:00:15 acc1_pos: 34.356 acc5_pos: 100.000 loss: 0.5765 (0.5902) acc1: 72.1354 (69.0456) acc5: 100.0000 (100.0000) acc1_neg: 23.698 acc5_neg: 100.000 time: 0.1832 data: 0.0898 max mem: 3532 +Test: [320/389] eta: 0:00:13 acc1_pos: 34.356 acc5_pos: 100.000 loss: 0.6597 (0.5892) acc1: 64.5833 (69.2149) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1948 data: 0.1026 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 34.356 acc5_pos: 100.000 loss: 0.7453 (0.5936) acc1: 42.7083 (68.4683) acc5: 100.0000 (100.0000) acc1_neg: 63.802 acc5_neg: 100.000 time: 0.1792 data: 0.0857 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 34.356 acc5_pos: 100.000 loss: 0.4608 (0.5876) acc1: 68.4896 (69.2747) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1577 data: 0.0650 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 34.356 acc5_pos: 100.000 loss: 0.3803 (0.5954) acc1: 98.1771 (68.5949) acc5: 100.0000 (100.0000) acc1_neg: 1.562 acc5_neg: 100.000 time: 0.1630 data: 0.0704 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 34.356 acc5_pos: 100.000 loss: 0.7796 (0.6028) acc1: 43.2292 (67.6499) acc5: 100.0000 (100.0000) acc1_neg: 43.490 acc5_neg: 100.000 time: 0.1725 data: 0.0801 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 34.356 acc5_pos: 100.000 loss: 0.7590 (0.6018) acc1: 44.2708 (67.6873) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1676 data: 0.0771 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 34.356 acc5_pos: 100.000 loss: 0.5396 (0.6008) acc1: 67.7083 (67.7979) acc5: 100.0000 (100.0000) acc1_neg: 60.677 acc5_neg: 100.000 time: 0.1507 data: 0.0600 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 34.356 acc5_pos: 100.000 loss: 0.4838 (0.5982) acc1: 81.7708 (68.1785) acc5: 100.0000 (100.0000) acc1_neg: 79.508 acc5_neg: 100.000 time: 0.1309 data: 0.0368 max mem: 3532 +Test: Total time: 0:01:13 (0.1879 s / it) +* Overall Acc@1 68.178 Acc@5 100.000 + target == real Acc@1 61.129 Acc@5 100.000 + target == fake Acc@1 71.455 Acc@5 100.000 + loss 0.598 + PR-AUC 0.8446 ROC-AUC 0.7325 EER 33.21% (thr=0.5259) +Accuracy of the network on the 149358 test images: 68.2% +[rank0]:[W726 04:21:41.038963260 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:22:02.736000 1186123 torch/distributed/run.py:766] +W0726 04:22:02.736000 1186123 torch/distributed/run.py:766] ***************************************** +W0726 04:22:02.736000 1186123 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:22:02.736000 1186123 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:22:31.987319316 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:22:32.668395650 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:22:32.693876227 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:22:32.993933326 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:22:32.031691187 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:22:32.043395636 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +NCCL version 2.26.2+cuda12.2 +[rank1]:[W726 04:22:32.134341393 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:22:32.149284914 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n64_seed6_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed6_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:31:29 acc1_pos: 82.031 acc5_pos: 100.000 loss: 0.4102 (0.4102) acc1: 82.0312 (82.0312) acc5: 100.0000 (100.0000) time: 5.6746 data: 3.9775 max mem: 3531 +Test: [ 10/333] eta: 0:03:13 acc1_pos: 83.854 acc5_pos: 100.000 loss: 0.3335 (0.3268) acc1: 85.6771 (86.8608) acc5: 100.0000 (100.0000) time: 0.5997 data: 0.3620 max mem: 3531 +Test: [ 20/333] eta: 0:01:52 acc1_pos: 94.531 acc5_pos: 100.000 loss: 0.3321 (0.3271) acc1: 86.9792 (86.4831) acc5: 100.0000 (100.0000) time: 0.0943 data: 0.0005 max mem: 3531 +Test: [ 30/333] eta: 0:01:33 acc1_pos: 55.469 acc5_pos: 100.000 loss: 0.4267 (0.4717) acc1: 71.0938 (74.7312) acc5: 100.0000 (100.0000) time: 0.1477 data: 0.0535 max mem: 3531 +Test: [ 40/333] eta: 0:01:23 acc1_pos: 50.000 acc5_pos: 100.000 loss: 0.7984 (0.5553) acc1: 46.8750 (67.2193) acc5: 100.0000 (100.0000) time: 0.2058 data: 0.1120 max mem: 3531 +Test: [ 50/333] eta: 0:01:18 acc1_pos: 48.438 acc5_pos: 100.000 loss: 0.7984 (0.6065) acc1: 44.5312 (63.0004) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1324 max mem: 3531 +Test: [ 60/333] eta: 0:01:12 acc1_pos: 35.677 acc5_pos: 100.000 loss: 0.7837 (0.6417) acc1: 45.8333 (60.2118) acc5: 100.0000 (100.0000) time: 0.2286 data: 0.1317 max mem: 3531 +Test: [ 70/333] eta: 0:01:07 acc1_pos: 41.146 acc5_pos: 100.000 loss: 0.7837 (0.6665) acc1: 46.3542 (58.3994) acc5: 100.0000 (100.0000) time: 0.2049 data: 0.1110 max mem: 3531 +Test: [ 80/333] eta: 0:01:02 acc1_pos: 53.906 acc5_pos: 100.000 loss: 0.7953 (0.6841) acc1: 46.6146 (56.8930) acc5: 100.0000 (100.0000) time: 0.1823 data: 0.0873 max mem: 3531 +Test: [ 90/333] eta: 0:00:59 acc1_pos: 59.896 acc5_pos: 100.000 loss: 0.7279 (0.6859) acc1: 52.8646 (56.8052) acc5: 100.0000 (100.0000) time: 0.1999 data: 0.1000 max mem: 3531 +Test: [100/333] eta: 0:00:56 acc1_pos: 60.417 acc5_pos: 100.000 loss: 0.6963 (0.6890) acc1: 57.8125 (56.7322) acc5: 100.0000 (100.0000) time: 0.2353 data: 0.1382 max mem: 3531 +Test: [110/333] eta: 0:00:53 acc1_pos: 65.365 acc5_pos: 100.000 loss: 0.7112 (0.6908) acc1: 56.2500 (56.7380) acc5: 100.0000 (100.0000) time: 0.2071 data: 0.1165 max mem: 3531 +Test: [120/333] eta: 0:00:50 acc1_pos: 63.021 acc5_pos: 100.000 loss: 0.6633 (0.6907) acc1: 58.3333 (56.7536) acc5: 100.0000 (100.0000) time: 0.2020 data: 0.1083 max mem: 3531 +Test: [130/333] eta: 0:00:47 acc1_pos: 69.010 acc5_pos: 100.000 loss: 0.6444 (0.6891) acc1: 60.9375 (57.0213) acc5: 100.0000 (100.0000) time: 0.2178 data: 0.1203 max mem: 3531 +Test: [140/333] eta: 0:00:44 acc1_pos: 77.604 acc5_pos: 100.000 loss: 0.6444 (0.6866) acc1: 59.6354 (57.2400) acc5: 100.0000 (100.0000) time: 0.1967 data: 0.1017 max mem: 3531 +Test: [150/333] eta: 0:00:41 acc1_pos: 63.021 acc5_pos: 100.000 loss: 0.6754 (0.6887) acc1: 57.2917 (57.1520) acc5: 100.0000 (100.0000) time: 0.1806 data: 0.0880 max mem: 3531 +Test: [160/333] eta: 0:00:38 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.7174 (0.6932) acc1: 53.9062 (56.9019) acc5: 100.0000 (100.0000) acc1_neg: 53.385 acc5_neg: 100.000 time: 0.1832 data: 0.0897 max mem: 3531 +Test: [170/333] eta: 0:00:36 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.7140 (0.6972) acc1: 55.7292 (56.7906) acc5: 100.0000 (100.0000) acc1_neg: 60.677 acc5_neg: 100.000 time: 0.1959 data: 0.1030 max mem: 3531 +Test: [180/333] eta: 0:00:33 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.6908 (0.6948) acc1: 60.4167 (57.1579) acc5: 100.0000 (100.0000) acc1_neg: 65.625 acc5_neg: 100.000 time: 0.1905 data: 0.0966 max mem: 3531 +Test: [190/333] eta: 0:00:31 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.6300 (0.6898) acc1: 65.6250 (57.7702) acc5: 100.0000 (100.0000) acc1_neg: 67.708 acc5_neg: 100.000 time: 0.2046 data: 0.1118 max mem: 3531 +Test: [200/333] eta: 0:00:29 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.5936 (0.6863) acc1: 67.7083 (58.2996) acc5: 100.0000 (100.0000) acc1_neg: 64.062 acc5_neg: 100.000 time: 0.2107 data: 0.1196 max mem: 3531 +Test: [210/333] eta: 0:00:26 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.6684 (0.6871) acc1: 62.7604 (58.4123) acc5: 100.0000 (100.0000) acc1_neg: 55.208 acc5_neg: 100.000 time: 0.1652 data: 0.0724 max mem: 3532 +Test: [220/333] eta: 0:00:24 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.7281 (0.6896) acc1: 57.8125 (58.2626) acc5: 100.0000 (100.0000) acc1_neg: 72.656 acc5_neg: 100.000 time: 0.1643 data: 0.0678 max mem: 3532 +Test: [230/333] eta: 0:00:22 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.5011 (0.6803) acc1: 77.8646 (59.3085) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.2131 data: 0.1162 max mem: 3532 +Test: [240/333] eta: 0:00:20 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.4965 (0.6757) acc1: 78.3854 (59.8310) acc5: 100.0000 (100.0000) acc1_neg: 70.312 acc5_neg: 100.000 time: 0.2168 data: 0.1199 max mem: 3532 +Test: [250/333] eta: 0:00:17 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.5766 (0.6727) acc1: 70.3125 (60.2019) acc5: 100.0000 (100.0000) acc1_neg: 61.979 acc5_neg: 100.000 time: 0.1757 data: 0.0810 max mem: 3532 +Test: [260/333] eta: 0:00:15 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.5991 (0.6708) acc1: 67.4479 (60.4935) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.1742 data: 0.0828 max mem: 3532 +Test: [270/333] eta: 0:00:13 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.5929 (0.6676) acc1: 69.0104 (60.8674) acc5: 100.0000 (100.0000) acc1_neg: 67.969 acc5_neg: 100.000 time: 0.1744 data: 0.0829 max mem: 3532 +Test: [280/333] eta: 0:00:11 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.5965 (0.6685) acc1: 68.2292 (60.8254) acc5: 100.0000 (100.0000) acc1_neg: 63.802 acc5_neg: 100.000 time: 0.1694 data: 0.0762 max mem: 3532 +Test: [290/333] eta: 0:00:08 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.7665 (0.6726) acc1: 50.7812 (60.4695) acc5: 100.0000 (100.0000) acc1_neg: 57.292 acc5_neg: 100.000 time: 0.1825 data: 0.0886 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.7969 (0.6765) acc1: 51.3021 (60.2177) acc5: 100.0000 (100.0000) acc1_neg: 53.125 acc5_neg: 100.000 time: 0.1565 data: 0.0651 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.6981 (0.6725) acc1: 59.3750 (60.6712) acc5: 100.0000 (100.0000) acc1_neg: 75.521 acc5_neg: 100.000 time: 0.1327 data: 0.0417 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.5302 (0.6693) acc1: 75.5208 (61.0381) acc5: 100.0000 (100.0000) acc1_neg: 53.125 acc5_neg: 100.000 time: 0.1326 data: 0.0414 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.7651 (0.6740) acc1: 54.4271 (60.7495) acc5: 100.0000 (100.0000) acc1_neg: 54.167 acc5_neg: 100.000 time: 0.1101 data: 0.0197 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.7982 (0.6749) acc1: 54.1667 (60.7057) acc5: 100.0000 (100.0000) acc1_neg: 51.831 acc5_neg: 100.000 time: 0.1136 data: 0.0196 max mem: 3532 +Test: Total time: 0:01:05 (0.1979 s / it) +* Overall Acc@1 60.706 Acc@5 100.000 + target == real Acc@1 57.145 Acc@5 100.000 + target == fake Acc@1 63.875 Acc@5 100.000 + loss 0.675 + PR-AUC 0.6552 ROC-AUC 0.6487 EER 39.50% (thr=0.5225) +Accuracy of the network on the 127843 test images: 60.7% +[rank0]:[W726 04:23:48.914322804 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:24:09.061000 1186686 torch/distributed/run.py:766] +W0726 04:24:09.061000 1186686 torch/distributed/run.py:766] ***************************************** +W0726 04:24:09.061000 1186686 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:24:09.061000 1186686 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:24:36.122054705 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:24:36.128363951 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:24:37.189624626 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:24:37.402778239 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 7): env:// +| distributed init (rank 5): env:// +| distributed init (rank 3): env:// +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:24:37.684152179 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W726 04:24:37.684179219 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W726 04:24:37.684192726 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W726 04:24:37.684210385 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n64_seed6_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed6_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:18:15 acc1_pos: 95.833 acc5_pos: 100.000 loss: 0.2298 (0.2298) acc1: 95.8333 (95.8333) acc5: 100.0000 (100.0000) time: 4.6631 data: 2.6227 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/235] eta: 0:01:53 acc1_pos: 94.271 acc5_pos: 100.000 loss: 0.2298 (0.2331) acc1: 95.8333 (95.4545) acc5: 100.0000 (100.0000) time: 0.5064 data: 0.2392 max mem: 3531 +Test: [ 20/235] eta: 0:01:08 acc1_pos: 96.615 acc5_pos: 100.000 loss: 0.2286 (0.2290) acc1: 95.8333 (95.7341) acc5: 100.0000 (100.0000) time: 0.1005 data: 0.0081 max mem: 3531 +Test: [ 30/235] eta: 0:00:53 acc1_pos: 96.354 acc5_pos: 100.000 loss: 0.2231 (0.2284) acc1: 96.0938 (95.7661) acc5: 100.0000 (100.0000) time: 0.1251 data: 0.0293 max mem: 3531 +Test: [ 40/235] eta: 0:00:50 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.2249 (0.2277) acc1: 96.3542 (95.7317) acc5: 100.0000 (100.0000) time: 0.1949 data: 0.0993 max mem: 3531 +Test: [ 50/235] eta: 0:00:46 acc1_pos: 94.271 acc5_pos: 100.000 loss: 0.2281 (0.2292) acc1: 95.0521 (95.5729) acc5: 100.0000 (100.0000) time: 0.2356 data: 0.1420 max mem: 3531 +Test: [ 60/235] eta: 0:00:42 acc1_pos: 94.010 acc5_pos: 100.000 loss: 0.2343 (0.2301) acc1: 95.3125 (95.5729) acc5: 100.0000 (100.0000) time: 0.2049 data: 0.1117 max mem: 3531 +Test: [ 70/235] eta: 0:00:40 acc1_pos: 98.462 acc5_pos: 100.000 loss: 0.2492 (0.3679) acc1: 94.0104 (86.4693) acc5: 100.0000 (100.0000) acc1_neg: 21.615 acc5_neg: 100.000 time: 0.2292 data: 0.1368 max mem: 3531 +Test: [ 80/235] eta: 0:00:36 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.3329 (0.4887) acc1: 22.9167 (78.4979) acc5: 100.0000 (100.0000) acc1_neg: 20.573 acc5_neg: 100.000 time: 0.2249 data: 0.1312 max mem: 3531 +Test: [ 90/235] eta: 0:00:32 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.3535 (0.5831) acc1: 21.8750 (72.2900) acc5: 100.0000 (100.0000) acc1_neg: 19.010 acc5_neg: 100.000 time: 0.1659 data: 0.0686 max mem: 3531 +Test: [100/235] eta: 0:00:30 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.3535 (0.6570) acc1: 21.8750 (67.3964) acc5: 100.0000 (100.0000) acc1_neg: 22.135 acc5_neg: 100.000 time: 0.1899 data: 0.0927 max mem: 3531 +Test: [110/235] eta: 0:00:27 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.3302 (0.7183) acc1: 22.1354 (63.3774) acc5: 100.0000 (100.0000) acc1_neg: 25.521 acc5_neg: 100.000 time: 0.2044 data: 0.1077 max mem: 3531 +Test: [120/235] eta: 0:00:25 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.3973 (0.7820) acc1: 19.7917 (59.3320) acc5: 100.0000 (100.0000) acc1_neg: 16.667 acc5_neg: 100.000 time: 0.2050 data: 0.1074 max mem: 3531 +Test: [130/235] eta: 0:00:23 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4875 (0.8361) acc1: 13.8021 (55.8862) acc5: 100.0000 (100.0000) acc1_neg: 13.542 acc5_neg: 100.000 time: 0.2163 data: 0.1184 max mem: 3531 +Test: [140/235] eta: 0:00:21 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.5301 (0.8854) acc1: 13.8021 (52.8664) acc5: 100.0000 (100.0000) acc1_neg: 9.375 acc5_neg: 100.000 time: 0.2198 data: 0.1233 max mem: 3531 +Test: [150/235] eta: 0:00:18 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.5194 (0.9257) acc1: 13.5417 (50.2742) acc5: 100.0000 (100.0000) acc1_neg: 14.323 acc5_neg: 100.000 time: 0.1828 data: 0.0844 max mem: 3531 +Test: [160/235] eta: 0:00:16 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4874 (0.9614) acc1: 14.3229 (48.0639) acc5: 100.0000 (100.0000) acc1_neg: 14.844 acc5_neg: 100.000 time: 0.1499 data: 0.0517 max mem: 3531 +Test: [170/235] eta: 0:00:13 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4576 (0.9721) acc1: 16.1458 (47.0319) acc5: 100.0000 (100.0000) acc1_neg: 31.250 acc5_neg: 100.000 time: 0.1949 data: 0.1017 max mem: 3531 +Test: [180/235] eta: 0:00:11 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.1057 (0.9789) acc1: 33.3333 (46.3556) acc5: 100.0000 (100.0000) acc1_neg: 34.375 acc5_neg: 100.000 time: 0.2149 data: 0.1176 max mem: 3531 +Test: [190/235] eta: 0:00:09 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.0906 (0.9851) acc1: 34.3750 (45.7120) acc5: 100.0000 (100.0000) acc1_neg: 34.115 acc5_neg: 100.000 time: 0.2348 data: 0.1331 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.0906 (0.9914) acc1: 33.0729 (45.0378) acc5: 100.0000 (100.0000) acc1_neg: 33.073 acc5_neg: 100.000 time: 0.2404 data: 0.1433 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.0953 (0.9962) acc1: 33.0729 (44.5288) acc5: 100.0000 (100.0000) acc1_neg: 33.854 acc5_neg: 100.000 time: 0.2172 data: 0.1250 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.0942 (1.0004) acc1: 34.6354 (44.0634) acc5: 100.0000 (100.0000) acc1_neg: 32.031 acc5_neg: 100.000 time: 0.1716 data: 0.0809 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.0868 (1.0040) acc1: 34.3750 (43.6609) acc5: 100.0000 (100.0000) acc1_neg: 33.594 acc5_neg: 100.000 time: 0.1314 data: 0.0413 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.0895 (1.0070) acc1: 34.1146 (43.5207) acc5: 100.0000 (100.0000) acc1_neg: 34.783 acc5_neg: 100.000 time: 0.1303 data: 0.0434 max mem: 3532 +Test: Total time: 0:00:48 (0.2082 s / it) +* Overall Acc@1 43.521 Acc@5 100.000 + target == real Acc@1 95.606 Acc@5 100.000 + target == fake Acc@1 24.683 Acc@5 100.000 + loss 1.007 + PR-AUC 0.8808 ROC-AUC 0.7291 EER 33.05% (thr=0.2001) +Accuracy of the network on the 89879 test images: 43.5% +[rank0]:[W726 04:25:35.361556466 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:25:54.814000 1187245 torch/distributed/run.py:766] +W0726 04:25:54.814000 1187245 torch/distributed/run.py:766] ***************************************** +W0726 04:25:54.814000 1187245 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:25:54.814000 1187245 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 04:26:24.627423718 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:26:24.132049865 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:26:25.303433213 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:26:25.336786378 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +[rank5]:[W726 04:26:25.559809562 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:26:25.566645119 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:26:25.614224590 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:26:25.917747304 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n64_seed6_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed6_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:28:50 acc1_pos: 55.208 acc5_pos: 100.000 loss: 0.6886 (0.6886) acc1: 55.2083 (55.2083) acc5: 100.0000 (100.0000) time: 6.4347 data: 4.9702 max mem: 3531 +Test: [ 10/269] eta: 0:02:53 acc1_pos: 62.500 acc5_pos: 100.000 loss: 0.6898 (0.6635) acc1: 55.9896 (60.1563) acc5: 100.0000 (100.0000) time: 0.6686 data: 0.4522 max mem: 3531 +Test: [ 20/269] eta: 0:01:47 acc1_pos: 67.969 acc5_pos: 100.000 loss: 0.6870 (0.6507) acc1: 55.9896 (61.4459) acc5: 100.0000 (100.0000) time: 0.1310 data: 0.0371 max mem: 3531 +Test: [ 30/269] eta: 0:01:29 acc1_pos: 65.104 acc5_pos: 100.000 loss: 0.6242 (0.6448) acc1: 65.1042 (62.7436) acc5: 100.0000 (100.0000) time: 0.2112 data: 0.1160 max mem: 3531 +Test: [ 40/269] eta: 0:01:20 acc1_pos: 56.034 acc5_pos: 100.000 loss: 0.5802 (0.6183) acc1: 68.2292 (65.8219) acc5: 100.0000 (100.0000) acc1_neg: 78.646 acc5_neg: 100.000 time: 0.2699 data: 0.1748 max mem: 3531 +Test: [ 50/269] eta: 0:01:10 acc1_pos: 56.034 acc5_pos: 100.000 loss: 0.5447 (0.6092) acc1: 75.7812 (66.9424) acc5: 100.0000 (100.0000) acc1_neg: 53.646 acc5_neg: 100.000 time: 0.2437 data: 0.1487 max mem: 3531 +Test: [ 60/269] eta: 0:01:03 acc1_pos: 56.034 acc5_pos: 100.000 loss: 0.5747 (0.6167) acc1: 68.2292 (66.0178) acc5: 100.0000 (100.0000) acc1_neg: 50.260 acc5_neg: 100.000 time: 0.1999 data: 0.1054 max mem: 3531 +Test: [ 70/269] eta: 0:00:57 acc1_pos: 56.034 acc5_pos: 100.000 loss: 0.6357 (0.6160) acc1: 59.1146 (65.9221) acc5: 100.0000 (100.0000) acc1_neg: 69.271 acc5_neg: 100.000 time: 0.1949 data: 0.0997 max mem: 3531 +Test: [ 80/269] eta: 0:00:52 acc1_pos: 56.034 acc5_pos: 100.000 loss: 0.5899 (0.6099) acc1: 69.2708 (66.7213) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.2053 data: 0.1076 max mem: 3531 +Test: [ 90/269] eta: 0:00:49 acc1_pos: 56.034 acc5_pos: 100.000 loss: 0.5949 (0.6174) acc1: 68.2292 (65.6565) acc5: 100.0000 (100.0000) acc1_neg: 55.469 acc5_neg: 100.000 time: 0.2299 data: 0.1342 max mem: 3531 +Test: [100/269] eta: 0:00:44 acc1_pos: 56.034 acc5_pos: 100.000 loss: 0.6812 (0.6244) acc1: 55.7292 (64.8644) acc5: 100.0000 (100.0000) acc1_neg: 48.438 acc5_neg: 100.000 time: 0.2108 data: 0.1191 max mem: 3531 +Test: [110/269] eta: 0:00:41 acc1_pos: 56.034 acc5_pos: 100.000 loss: 0.6812 (0.6283) acc1: 56.7708 (64.2643) acc5: 100.0000 (100.0000) acc1_neg: 53.125 acc5_neg: 100.000 time: 0.1962 data: 0.1050 max mem: 3531 +Test: [120/269] eta: 0:00:38 acc1_pos: 56.034 acc5_pos: 100.000 loss: 0.6370 (0.6208) acc1: 61.9792 (65.1214) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.2086 data: 0.1162 max mem: 3531 +Test: [130/269] eta: 0:00:34 acc1_pos: 56.034 acc5_pos: 100.000 loss: 0.5647 (0.6222) acc1: 70.0521 (64.8895) acc5: 100.0000 (100.0000) acc1_neg: 54.688 acc5_neg: 100.000 time: 0.1986 data: 0.1042 max mem: 3531 +Test: [140/269] eta: 0:00:32 acc1_pos: 56.034 acc5_pos: 100.000 loss: 0.6809 (0.6271) acc1: 54.9479 (64.1862) acc5: 100.0000 (100.0000) acc1_neg: 50.521 acc5_neg: 100.000 time: 0.2120 data: 0.1176 max mem: 3531 +Test: [150/269] eta: 0:00:29 acc1_pos: 56.034 acc5_pos: 100.000 loss: 0.6515 (0.6264) acc1: 54.9479 (64.1660) acc5: 100.0000 (100.0000) acc1_neg: 69.010 acc5_neg: 100.000 time: 0.2104 data: 0.1179 max mem: 3531 +Test: [160/269] eta: 0:00:26 acc1_pos: 56.034 acc5_pos: 100.000 loss: 0.5750 (0.6237) acc1: 69.0104 (64.5542) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.1891 data: 0.0986 max mem: 3531 +Test: [170/269] eta: 0:00:23 acc1_pos: 56.034 acc5_pos: 100.000 loss: 0.5952 (0.6234) acc1: 69.5312 (64.6732) acc5: 100.0000 (100.0000) acc1_neg: 68.490 acc5_neg: 100.000 time: 0.2050 data: 0.1140 max mem: 3531 +Test: [180/269] eta: 0:00:21 acc1_pos: 56.034 acc5_pos: 100.000 loss: 0.5810 (0.6188) acc1: 69.5312 (65.3401) acc5: 100.0000 (100.0000) acc1_neg: 64.583 acc5_neg: 100.000 time: 0.1961 data: 0.1017 max mem: 3531 +Test: [190/269] eta: 0:00:18 acc1_pos: 56.034 acc5_pos: 100.000 loss: 0.5673 (0.6185) acc1: 70.3125 (65.3932) acc5: 100.0000 (100.0000) acc1_neg: 67.708 acc5_neg: 100.000 time: 0.1844 data: 0.0901 max mem: 3531 +Test: [200/269] eta: 0:00:16 acc1_pos: 56.034 acc5_pos: 100.000 loss: 0.6468 (0.6193) acc1: 62.7604 (65.3646) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1924 data: 0.0990 max mem: 3531 +Test: [210/269] eta: 0:00:13 acc1_pos: 56.034 acc5_pos: 100.000 loss: 0.6016 (0.6173) acc1: 63.2812 (65.5645) acc5: 100.0000 (100.0000) acc1_neg: 72.917 acc5_neg: 100.000 time: 0.1854 data: 0.0900 max mem: 3532 +Test: [220/269] eta: 0:00:11 acc1_pos: 56.034 acc5_pos: 100.000 loss: 0.5630 (0.6148) acc1: 68.7500 (65.8359) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1923 data: 0.0984 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 56.034 acc5_pos: 100.000 loss: 0.5038 (0.6100) acc1: 75.7812 (66.3589) acc5: 100.0000 (100.0000) acc1_neg: 80.729 acc5_neg: 100.000 time: 0.1874 data: 0.0959 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 56.034 acc5_pos: 100.000 loss: 0.5774 (0.6117) acc1: 70.8333 (66.2204) acc5: 100.0000 (100.0000) acc1_neg: 70.833 acc5_neg: 100.000 time: 0.1501 data: 0.0596 max mem: 3532 +Test: [250/269] eta: 0:00:04 acc1_pos: 56.034 acc5_pos: 100.000 loss: 0.6647 (0.6151) acc1: 59.3750 (65.7412) acc5: 100.0000 (100.0000) acc1_neg: 36.198 acc5_neg: 100.000 time: 0.1261 data: 0.0359 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 56.034 acc5_pos: 100.000 loss: 0.7073 (0.6205) acc1: 46.0938 (65.0214) acc5: 100.0000 (100.0000) acc1_neg: 60.417 acc5_neg: 100.000 time: 0.1239 data: 0.0339 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 56.034 acc5_pos: 100.000 loss: 0.6771 (0.6199) acc1: 54.9479 (65.1399) acc5: 100.0000 (100.0000) acc1_neg: 86.877 acc5_neg: 100.000 time: 0.1261 data: 0.0326 max mem: 3532 +Test: Total time: 0:00:57 (0.2128 s / it) +* Overall Acc@1 65.140 Acc@5 100.000 + target == real Acc@1 62.262 Acc@5 100.000 + target == fake Acc@1 65.533 Acc@5 100.000 + loss 0.620 + PR-AUC 0.9374 ROC-AUC 0.6943 EER 36.03% (thr=0.5078) +Accuracy of the network on the 103293 test images: 65.1% +[rank0]:[W726 04:27:31.186805767 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T04:27:37Z] eval_done tag=fixed500k_fake_n64_seed6_resampled +[2026-07-26T04:27:38Z] eval_start tag=fixed500k_real_n32_seed7_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n32_classseed7_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T04:27:38Z +hostname=n116-040-213 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n32_classseed7_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n32_seed7_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 04:27:51.238000 1187810 torch/distributed/run.py:766] +W0726 04:27:51.238000 1187810 torch/distributed/run.py:766] ***************************************** +W0726 04:27:51.238000 1187810 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:27:51.238000 1187810 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:28:21.748613673 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:28:22.177987083 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:28:22.633496241 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:28:22.643620389 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:28:22.791617533 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:28:22.849271129 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W726 04:28:22.849344787 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:28:22.057944188 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n32_seed7_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n32_classseed7_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:26:17 acc1_pos: 13.802 acc5_pos: 100.000 loss: 1.2723 (1.2723) acc1: 13.8021 (13.8021) acc5: 100.0000 (100.0000) time: 6.7411 data: 5.1036 max mem: 3531 +Test: [ 10/234] eta: 0:02:47 acc1_pos: 29.948 acc5_pos: 100.000 loss: 0.9872 (1.0003) acc1: 29.9479 (38.2102) acc5: 100.0000 (100.0000) time: 0.7459 data: 0.5116 max mem: 3531 +Test: [ 20/234] eta: 0:01:33 acc1_pos: 95.833 acc5_pos: 100.000 loss: 0.9872 (1.0033) acc1: 34.6354 (41.4559) acc5: 100.0000 (100.0000) time: 0.1226 data: 0.0264 max mem: 3531 +Test: [ 30/234] eta: 0:01:22 acc1_pos: 75.521 acc5_pos: 100.000 loss: 0.2780 (0.7733) acc1: 92.7083 (57.5773) acc5: 100.0000 (100.0000) time: 0.2174 data: 0.1151 max mem: 3531 +Test: [ 40/234] eta: 0:01:12 acc1_pos: 98.958 acc5_pos: 100.000 loss: 0.2715 (0.6745) acc1: 93.4896 (63.7767) acc5: 100.0000 (100.0000) time: 0.3065 data: 0.2081 max mem: 3531 +Test: [ 50/234] eta: 0:01:02 acc1_pos: 97.917 acc5_pos: 100.000 loss: 0.1534 (0.5707) acc1: 97.6562 (70.4759) acc5: 100.0000 (100.0000) time: 0.2418 data: 0.1481 max mem: 3531 +Test: [ 60/234] eta: 0:00:57 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.1142 (0.4911) acc1: 98.9583 (75.2946) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2333 data: 0.1375 max mem: 3531 +Test: [ 70/234] eta: 0:00:50 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.0848 (0.4861) acc1: 100.0000 (75.3374) acc5: 100.0000 (100.0000) acc1_neg: 61.198 acc5_neg: 100.000 time: 0.2199 data: 0.1264 max mem: 3531 +Test: [ 80/234] eta: 0:00:45 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.6161 (0.4834) acc1: 65.3646 (75.5401) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1899 data: 0.0954 max mem: 3531 +Test: [ 90/234] eta: 0:00:40 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.2307 (0.4492) acc1: 91.9271 (77.7301) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.1903 data: 0.0948 max mem: 3531 +Test: [100/234] eta: 0:00:36 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.2359 (0.4388) acc1: 91.1458 (78.4318) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1859 data: 0.0905 max mem: 3531 +Test: [110/234] eta: 0:00:32 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.3650 (0.4291) acc1: 82.2917 (79.1385) acc5: 100.0000 (100.0000) acc1_neg: 77.083 acc5_neg: 100.000 time: 0.1948 data: 0.1008 max mem: 3531 +Test: [120/234] eta: 0:00:29 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.3423 (0.4300) acc1: 85.9375 (79.2205) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1989 data: 0.1072 max mem: 3531 +Test: [130/234] eta: 0:00:26 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.2840 (0.4216) acc1: 90.6250 (80.0135) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.2117 data: 0.1199 max mem: 3531 +Test: [140/234] eta: 0:00:23 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.3552 (0.4367) acc1: 85.1562 (79.2405) acc5: 100.0000 (100.0000) acc1_neg: 41.406 acc5_neg: 100.000 time: 0.2213 data: 0.1299 max mem: 3531 +Test: [150/234] eta: 0:00:21 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.3633 (0.4365) acc1: 82.5521 (79.2598) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.2177 data: 0.1260 max mem: 3531 +Test: [160/234] eta: 0:00:18 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.3266 (0.4286) acc1: 85.4167 (79.9463) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.2098 data: 0.1182 max mem: 3531 +Test: [170/234] eta: 0:00:15 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.3818 (0.4257) acc1: 85.6771 (80.3363) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1962 data: 0.1055 max mem: 3531 +Test: [180/234] eta: 0:00:13 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.3958 (0.4317) acc1: 83.3333 (79.9335) acc5: 100.0000 (100.0000) acc1_neg: 45.052 acc5_neg: 100.000 time: 0.1987 data: 0.1073 max mem: 3531 +Test: [190/234] eta: 0:00:10 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.6592 (0.4450) acc1: 63.8021 (79.0317) acc5: 100.0000 (100.0000) acc1_neg: 63.281 acc5_neg: 100.000 time: 0.1887 data: 0.0960 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.6546 (0.4382) acc1: 64.0625 (79.3999) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1490 data: 0.0572 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.0959 (0.4235) acc1: 99.7396 (80.3318) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1314 data: 0.0413 max mem: 3532 +Test: [220/234] eta: 0:00:03 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.1927 (0.4143) acc1: 97.6562 (81.0167) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.1302 data: 0.0401 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.4071 (0.4338) acc1: 50.5208 (79.4124) acc5: 100.0000 (100.0000) acc1_neg: 44.792 acc5_neg: 100.000 time: 0.1229 data: 0.0330 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.8482 (0.4391) acc1: 44.7917 (79.0369) acc5: 100.0000 (100.0000) acc1_neg: 43.969 acc5_neg: 100.000 time: 0.1201 data: 0.0291 max mem: 3532 +Test: Total time: 0:00:51 (0.2195 s / it) +* Overall Acc@1 79.037 Acc@5 100.000 + target == real Acc@1 70.673 Acc@5 100.000 + target == fake Acc@1 81.394 Acc@5 100.000 + loss 0.439 + PR-AUC 0.9523 ROC-AUC 0.8539 EER 23.75% (thr=0.5869) +Accuracy of the network on the 89729 test images: 79.0% +[rank0]:[W726 04:29:23.534881992 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:29:43.694000 1188368 torch/distributed/run.py:766] +W0726 04:29:43.694000 1188368 torch/distributed/run.py:766] ***************************************** +W0726 04:29:43.694000 1188368 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:29:43.694000 1188368 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:30:12.151980009 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:30:13.267502595 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:30:13.512232911 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:30:13.536928974 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:30:13.922688342 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 6): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:30:13.966571554 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W726 04:30:13.966780856 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:30:13.970367601 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n32_seed7_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n32_classseed7_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:44:12 acc1_pos: 5.990 acc5_pos: 100.000 loss: 1.6408 (1.6408) acc1: 5.9896 (5.9896) acc5: 100.0000 (100.0000) time: 7.1439 data: 5.7497 max mem: 3531 +Test: [ 10/2387] eta: 0:28:59 acc1_pos: 4.427 acc5_pos: 100.000 loss: 1.6408 (1.5809) acc1: 4.4271 (7.9309) acc5: 100.0000 (100.0000) time: 0.7317 data: 0.5231 max mem: 3531 +Test: [ 20/2387] eta: 0:16:56 acc1_pos: 2.604 acc5_pos: 100.000 loss: 1.5813 (1.5730) acc1: 4.4271 (7.8621) acc5: 100.0000 (100.0000) time: 0.0936 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:13:59 acc1_pos: 0.521 acc5_pos: 100.000 loss: 1.6525 (1.5938) acc1: 4.6875 (7.0565) acc5: 100.0000 (100.0000) time: 0.1499 data: 0.0543 max mem: 3531 +Test: [ 40/2387] eta: 0:12:55 acc1_pos: 1.823 acc5_pos: 100.000 loss: 1.6525 (1.6099) acc1: 1.8229 (6.1928) acc5: 100.0000 (100.0000) time: 0.2263 data: 0.1304 max mem: 3531 +Test: [ 50/2387] eta: 0:12:27 acc1_pos: 24.740 acc5_pos: 100.000 loss: 1.5709 (1.6062) acc1: 2.6042 (6.3164) acc5: 100.0000 (100.0000) time: 0.2634 data: 0.1689 max mem: 3531 +Test: [ 60/2387] eta: 0:11:35 acc1_pos: 20.833 acc5_pos: 100.000 loss: 1.5528 (1.5910) acc1: 6.2500 (6.8477) acc5: 100.0000 (100.0000) time: 0.2341 data: 0.1407 max mem: 3531 +Test: [ 70/2387] eta: 0:11:06 acc1_pos: 19.271 acc5_pos: 100.000 loss: 1.5946 (1.6000) acc1: 6.2500 (6.8002) acc5: 100.0000 (100.0000) time: 0.2049 data: 0.1082 max mem: 3531 +Test: [ 80/2387] eta: 0:10:49 acc1_pos: 4.688 acc5_pos: 100.000 loss: 1.6260 (1.6040) acc1: 4.6875 (6.8126) acc5: 100.0000 (100.0000) time: 0.2293 data: 0.1323 max mem: 3531 +Test: [ 90/2387] eta: 0:10:29 acc1_pos: 9.115 acc5_pos: 100.000 loss: 1.6033 (1.5963) acc1: 7.2917 (7.1858) acc5: 100.0000 (100.0000) time: 0.2264 data: 0.1299 max mem: 3531 +Test: [ 100/2387] eta: 0:10:16 acc1_pos: 1.562 acc5_pos: 100.000 loss: 1.5955 (1.6068) acc1: 6.2500 (6.9539) acc5: 100.0000 (100.0000) time: 0.2199 data: 0.1250 max mem: 3531 +Test: [ 110/2387] eta: 0:10:07 acc1_pos: 17.448 acc5_pos: 100.000 loss: 1.6082 (1.6013) acc1: 5.2083 (7.1486) acc5: 100.0000 (100.0000) time: 0.2334 data: 0.1388 max mem: 3531 +Test: [ 120/2387] eta: 0:09:54 acc1_pos: 11.719 acc5_pos: 100.000 loss: 1.5549 (1.5976) acc1: 9.1146 (7.4186) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1295 max mem: 3531 +Test: [ 130/2387] eta: 0:09:45 acc1_pos: 12.240 acc5_pos: 100.000 loss: 1.5501 (1.6003) acc1: 8.3333 (7.3871) acc5: 100.0000 (100.0000) time: 0.2199 data: 0.1268 max mem: 3531 +Test: [ 140/2387] eta: 0:09:43 acc1_pos: 5.469 acc5_pos: 100.000 loss: 1.6081 (1.6020) acc1: 5.9896 (7.4357) acc5: 100.0000 (100.0000) time: 0.2454 data: 0.1533 max mem: 3531 +Test: [ 150/2387] eta: 0:09:54 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.6081 (1.6030) acc1: 5.9896 (7.4331) acc5: 100.0000 (100.0000) time: 0.3049 data: 0.2137 max mem: 3531 +Test: [ 160/2387] eta: 0:09:36 acc1_pos: 4.110 acc5_pos: 100.000 loss: 1.5593 (1.5698) acc1: 8.3333 (9.6095) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2545 data: 0.1578 max mem: 3531 +Test: [ 170/2387] eta: 0:09:26 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1705 (1.4875) acc1: 98.6979 (14.8529) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1802 data: 0.0811 max mem: 3531 +Test: [ 180/2387] eta: 0:09:18 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1609 (1.4141) acc1: 99.2188 (19.5125) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2049 data: 0.1095 max mem: 3531 +Test: [ 190/2387] eta: 0:09:06 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1590 (1.3483) acc1: 99.4792 (23.6952) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1896 data: 0.0937 max mem: 3531 +Test: [ 200/2387] eta: 0:08:50 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1574 (1.2891) acc1: 99.2188 (27.4461) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1501 data: 0.0547 max mem: 3531 +Test: [ 210/2387] eta: 0:08:44 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1592 (1.2357) acc1: 99.2188 (30.8452) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1675 data: 0.0750 max mem: 3532 +Test: [ 220/2387] eta: 0:08:37 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1592 (1.1870) acc1: 99.2188 (33.9390) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1965 data: 0.1046 max mem: 3532 +Test: [ 230/2387] eta: 0:08:30 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1596 (1.1426) acc1: 99.4792 (36.7695) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1925 data: 0.0995 max mem: 3532 +Test: [ 240/2387] eta: 0:08:22 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1514 (1.1010) acc1: 99.4792 (39.3791) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1847 data: 0.0919 max mem: 3532 +Test: [ 250/2387] eta: 0:08:13 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1310 (1.0604) acc1: 99.7396 (41.7922) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1614 data: 0.0687 max mem: 3532 +Test: [ 260/2387] eta: 0:08:06 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0731 (1.0225) acc1: 100.0000 (44.0224) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1654 data: 0.0733 max mem: 3532 +Test: [ 270/2387] eta: 0:08:00 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0699 (0.9874) acc1: 100.0000 (46.0870) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1787 data: 0.0859 max mem: 3532 +Test: [ 280/2387] eta: 0:07:55 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0688 (0.9549) acc1: 100.0000 (48.0056) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1827 data: 0.0892 max mem: 3532 +Test: [ 290/2387] eta: 0:07:52 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0690 (0.9245) acc1: 100.0000 (49.7924) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2009 data: 0.1088 max mem: 3532 +Test: [ 300/2387] eta: 0:07:48 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0690 (0.8960) acc1: 100.0000 (51.4604) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2113 data: 0.1179 max mem: 3532 +Test: [ 310/2387] eta: 0:07:45 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0679 (0.8732) acc1: 100.0000 (52.9667) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2089 data: 0.1150 max mem: 3532 +Test: [ 320/2387] eta: 0:07:41 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.3266 (0.8563) acc1: 97.1354 (54.3476) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2004 data: 0.1081 max mem: 3532 +Test: [ 330/2387] eta: 0:07:39 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.3311 (0.8404) acc1: 97.1354 (55.6497) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2093 data: 0.1170 max mem: 3532 +Test: [ 340/2387] eta: 0:07:34 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.3261 (0.8254) acc1: 97.1354 (56.8609) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2078 data: 0.1126 max mem: 3532 +Test: [ 350/2387] eta: 0:07:30 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.3200 (0.8068) acc1: 97.6562 (58.0818) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1884 data: 0.0918 max mem: 3532 +Test: [ 360/2387] eta: 0:07:26 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1464 (0.7883) acc1: 100.0000 (59.2329) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1870 data: 0.0921 max mem: 3532 +Test: [ 370/2387] eta: 0:07:24 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1382 (0.7707) acc1: 99.7396 (60.3170) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2064 data: 0.1133 max mem: 3532 +Test: [ 380/2387] eta: 0:07:20 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1369 (0.7541) acc1: 99.7396 (61.3476) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2034 data: 0.1103 max mem: 3532 +Test: [ 390/2387] eta: 0:07:14 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1361 (0.7383) acc1: 99.7396 (62.3268) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1676 data: 0.0730 max mem: 3532 +Test: [ 400/2387] eta: 0:07:11 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1348 (0.7233) acc1: 99.7396 (63.2559) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1718 data: 0.0794 max mem: 3532 +Test: [ 410/2387] eta: 0:07:08 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1345 (0.7090) acc1: 99.7396 (64.1385) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1958 data: 0.1050 max mem: 3532 +Test: [ 420/2387] eta: 0:07:04 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1348 (0.6954) acc1: 99.4792 (64.9780) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1925 data: 0.1014 max mem: 3532 +Test: [ 430/2387] eta: 0:07:01 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1319 (0.6823) acc1: 99.4792 (65.7833) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1869 data: 0.0960 max mem: 3532 +Test: [ 440/2387] eta: 0:06:58 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1255 (0.6690) acc1: 99.7396 (66.5551) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1947 data: 0.1025 max mem: 3533 +Test: [ 450/2387] eta: 0:06:54 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0777 (0.6559) acc1: 100.0000 (67.2966) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1869 data: 0.0937 max mem: 3533 +Test: [ 460/2387] eta: 0:06:51 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0775 (0.6433) acc1: 100.0000 (68.0060) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1851 data: 0.0925 max mem: 3533 +Test: [ 470/2387] eta: 0:06:48 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0774 (0.6313) acc1: 100.0000 (68.6853) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1945 data: 0.1028 max mem: 3533 +Test: [ 480/2387] eta: 0:06:46 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0778 (0.6198) acc1: 100.0000 (69.3363) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2048 data: 0.1127 max mem: 3533 +Test: [ 490/2387] eta: 0:06:42 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0778 (0.6088) acc1: 100.0000 (69.9609) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1944 data: 0.1018 max mem: 3533 +Test: [ 500/2387] eta: 0:06:40 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0779 (0.5982) acc1: 100.0000 (70.5604) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1825 data: 0.0908 max mem: 3533 +Test: [ 510/2387] eta: 0:06:36 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0780 (0.5881) acc1: 100.0000 (71.1366) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1838 data: 0.0925 max mem: 3533 +Test: [ 520/2387] eta: 0:06:34 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0935 (0.5787) acc1: 100.0000 (71.6906) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1934 data: 0.1021 max mem: 3533 +Test: [ 530/2387] eta: 0:06:31 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0964 (0.5696) acc1: 100.0000 (72.2237) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2004 data: 0.1079 max mem: 3533 +Test: [ 540/2387] eta: 0:06:28 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0955 (0.5608) acc1: 100.0000 (72.7371) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1887 data: 0.0958 max mem: 3533 +Test: [ 550/2387] eta: 0:06:25 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0952 (0.5524) acc1: 100.0000 (73.2319) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1838 data: 0.0903 max mem: 3533 +Test: [ 560/2387] eta: 0:06:22 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0953 (0.5442) acc1: 100.0000 (73.7091) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1833 data: 0.0890 max mem: 3533 +Test: [ 570/2387] eta: 0:06:20 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0951 (0.5364) acc1: 100.0000 (74.1695) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2012 data: 0.1071 max mem: 3533 +Test: [ 580/2387] eta: 0:06:19 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0951 (0.5288) acc1: 100.0000 (74.6141) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2222 data: 0.1291 max mem: 3533 +Test: [ 590/2387] eta: 0:06:18 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0956 (0.5214) acc1: 100.0000 (75.0436) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2345 data: 0.1427 max mem: 3533 +Test: [ 600/2387] eta: 0:06:14 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0970 (0.5144) acc1: 100.0000 (75.4584) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2069 data: 0.1141 max mem: 3533 +Test: [ 610/2387] eta: 0:06:12 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1096 (0.5082) acc1: 100.0000 (75.8563) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1889 data: 0.0945 max mem: 3533 +Test: [ 620/2387] eta: 0:06:10 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1324 (0.5021) acc1: 99.7396 (76.2409) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1978 data: 0.1042 max mem: 3533 +Test: [ 630/2387] eta: 0:06:08 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1334 (0.4963) acc1: 99.7396 (76.6116) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2021 data: 0.1086 max mem: 3533 +Test: [ 640/2387] eta: 0:06:05 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1334 (0.4906) acc1: 99.7396 (76.9720) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2104 data: 0.1171 max mem: 3533 +Test: [ 650/2387] eta: 0:06:03 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1343 (0.4852) acc1: 99.7396 (77.3209) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2106 data: 0.1192 max mem: 3533 +Test: [ 660/2387] eta: 0:06:02 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1337 (0.4799) acc1: 99.7396 (77.6605) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2160 data: 0.1198 max mem: 3533 +Test: [ 670/2387] eta: 0:06:01 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1326 (0.4747) acc1: 99.7396 (77.9888) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2433 data: 0.1474 max mem: 3534 +Test: [ 680/2387] eta: 0:05:57 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1303 (0.4694) acc1: 99.7396 (78.3105) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2084 data: 0.1174 max mem: 3534 +Test: [ 690/2387] eta: 0:05:55 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1308 (0.4675) acc1: 99.7396 (78.4175) acc5: 100.0000 (100.0000) acc1_neg: 77.344 acc5_neg: 100.000 time: 0.1686 data: 0.0746 max mem: 3534 +Test: [ 700/2387] eta: 0:05:52 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.4006 (0.4666) acc1: 80.9896 (78.4567) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.1892 data: 0.0924 max mem: 3534 +Test: [ 710/2387] eta: 0:05:50 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.4076 (0.4658) acc1: 80.9896 (78.4942) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1952 data: 0.1014 max mem: 3534 +Test: [ 720/2387] eta: 0:05:48 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.4048 (0.4649) acc1: 81.2500 (78.5357) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.2070 data: 0.1159 max mem: 3534 +Test: [ 730/2387] eta: 0:05:45 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.4107 (0.4643) acc1: 80.9896 (78.5571) acc5: 100.0000 (100.0000) acc1_neg: 79.427 acc5_neg: 100.000 time: 0.1902 data: 0.0957 max mem: 3534 +Test: [ 740/2387] eta: 0:05:42 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.4177 (0.4637) acc1: 80.7292 (78.5917) acc5: 100.0000 (100.0000) acc1_neg: 78.646 acc5_neg: 100.000 time: 0.1754 data: 0.0778 max mem: 3534 +Test: [ 750/2387] eta: 0:05:39 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.4090 (0.4629) acc1: 81.2500 (78.6285) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.1799 data: 0.0841 max mem: 3534 +Test: [ 760/2387] eta: 0:05:37 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.3991 (0.4613) acc1: 81.7708 (78.7180) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1838 data: 0.0903 max mem: 3534 +Test: [ 770/2387] eta: 0:05:35 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.2358 (0.4576) acc1: 92.4479 (78.9400) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1942 data: 0.0977 max mem: 3534 +Test: [ 780/2387] eta: 0:05:33 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1059 (0.4530) acc1: 100.0000 (79.2073) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1987 data: 0.1023 max mem: 3534 +Test: [ 790/2387] eta: 0:05:30 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0951 (0.4486) acc1: 100.0000 (79.4620) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1953 data: 0.1001 max mem: 3534 +Test: [ 800/2387] eta: 0:05:27 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0951 (0.4442) acc1: 100.0000 (79.7132) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1846 data: 0.0890 max mem: 3534 +Test: [ 810/2387] eta: 0:05:25 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1025 (0.4400) acc1: 100.0000 (79.9611) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1795 data: 0.0843 max mem: 3534 +Test: [ 820/2387] eta: 0:05:22 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1016 (0.4359) acc1: 100.0000 (80.2017) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1761 data: 0.0796 max mem: 3534 +Test: [ 830/2387] eta: 0:05:19 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1188 (0.4324) acc1: 99.7396 (80.4199) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1718 data: 0.0785 max mem: 3534 +Test: [ 840/2387] eta: 0:05:17 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1547 (0.4292) acc1: 99.2188 (80.6434) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1716 data: 0.0807 max mem: 3534 +Test: [ 850/2387] eta: 0:05:14 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1661 (0.4262) acc1: 99.2188 (80.8458) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1730 data: 0.0800 max mem: 3534 +Test: [ 860/2387] eta: 0:05:11 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1569 (0.4231) acc1: 99.4792 (81.0601) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1702 data: 0.0771 max mem: 3534 +Test: [ 870/2387] eta: 0:05:09 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1521 (0.4201) acc1: 99.7396 (81.2638) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1720 data: 0.0808 max mem: 3534 +Test: [ 880/2387] eta: 0:05:07 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1533 (0.4173) acc1: 99.4792 (81.4611) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1920 data: 0.0996 max mem: 3534 +Test: [ 890/2387] eta: 0:05:04 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1415 (0.4141) acc1: 99.7396 (81.6636) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1928 data: 0.1003 max mem: 3534 +Test: [ 900/2387] eta: 0:05:02 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1442 (0.4161) acc1: 97.6562 (81.4885) acc5: 100.0000 (100.0000) acc1_neg: 61.719 acc5_neg: 100.000 time: 0.1801 data: 0.0866 max mem: 3535 +Test: [ 910/2387] eta: 0:05:00 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.7038 (0.4193) acc1: 58.0729 (81.2371) acc5: 100.0000 (100.0000) acc1_neg: 57.812 acc5_neg: 100.000 time: 0.1822 data: 0.0887 max mem: 3535 +Test: [ 920/2387] eta: 0:04:57 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.7175 (0.4225) acc1: 57.8125 (80.9780) acc5: 100.0000 (100.0000) acc1_neg: 54.688 acc5_neg: 100.000 time: 0.1844 data: 0.0922 max mem: 3535 +Test: [ 930/2387] eta: 0:04:55 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.7147 (0.4256) acc1: 57.5521 (80.7370) acc5: 100.0000 (100.0000) acc1_neg: 54.167 acc5_neg: 100.000 time: 0.1753 data: 0.0823 max mem: 3535 +Test: [ 940/2387] eta: 0:04:52 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.7070 (0.4286) acc1: 58.3333 (80.5017) acc5: 100.0000 (100.0000) acc1_neg: 56.510 acc5_neg: 100.000 time: 0.1676 data: 0.0746 max mem: 3535 +Test: [ 950/2387] eta: 0:04:50 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.7064 (0.4315) acc1: 58.3333 (80.2615) acc5: 100.0000 (100.0000) acc1_neg: 58.073 acc5_neg: 100.000 time: 0.1750 data: 0.0828 max mem: 3535 +Test: [ 960/2387] eta: 0:04:47 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.6915 (0.4342) acc1: 58.5938 (80.0422) acc5: 100.0000 (100.0000) acc1_neg: 59.896 acc5_neg: 100.000 time: 0.1682 data: 0.0755 max mem: 3535 +Test: [ 970/2387] eta: 0:04:45 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.6996 (0.4371) acc1: 58.8542 (79.8052) acc5: 100.0000 (100.0000) acc1_neg: 59.115 acc5_neg: 100.000 time: 0.1644 data: 0.0701 max mem: 3535 +Test: [ 980/2387] eta: 0:04:42 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.6878 (0.4363) acc1: 59.1146 (79.8293) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.1696 data: 0.0765 max mem: 3535 +Test: [ 990/2387] eta: 0:04:40 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1174 (0.4333) acc1: 98.6979 (79.9978) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1632 data: 0.0714 max mem: 3535 +Test: [1000/2387] eta: 0:04:37 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0952 (0.4300) acc1: 100.0000 (80.1891) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1669 data: 0.0745 max mem: 3535 +Test: [1010/2387] eta: 0:04:35 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1019 (0.4269) acc1: 99.7396 (80.3799) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1804 data: 0.0879 max mem: 3535 +Test: [1020/2387] eta: 0:04:33 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1039 (0.4240) acc1: 99.7396 (80.5573) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1783 data: 0.0853 max mem: 3535 +Test: [1030/2387] eta: 0:04:30 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1007 (0.4208) acc1: 100.0000 (80.7420) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1687 data: 0.0736 max mem: 3535 +Test: [1040/2387] eta: 0:04:28 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0933 (0.4177) acc1: 100.0000 (80.9268) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1736 data: 0.0768 max mem: 3535 +Test: [1050/2387] eta: 0:04:26 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0895 (0.4146) acc1: 100.0000 (81.1078) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1722 data: 0.0775 max mem: 3535 +Test: [1060/2387] eta: 0:04:23 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0927 (0.4116) acc1: 100.0000 (81.2849) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1638 data: 0.0715 max mem: 3535 +Test: [1070/2387] eta: 0:04:21 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0962 (0.4087) acc1: 100.0000 (81.4572) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1739 data: 0.0820 max mem: 3535 +Test: [1080/2387] eta: 0:04:19 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1089 (0.4061) acc1: 100.0000 (81.6210) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1777 data: 0.0850 max mem: 3535 +Test: [1090/2387] eta: 0:04:16 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1112 (0.4034) acc1: 100.0000 (81.7875) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1680 data: 0.0736 max mem: 3535 +Test: [1100/2387] eta: 0:04:14 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1109 (0.4009) acc1: 100.0000 (81.9473) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1633 data: 0.0706 max mem: 3535 +Test: [1110/2387] eta: 0:04:11 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1130 (0.3985) acc1: 99.7396 (82.0988) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1610 data: 0.0704 max mem: 3535 +Test: [1120/2387] eta: 0:04:09 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1149 (0.3960) acc1: 99.7396 (82.2547) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1616 data: 0.0701 max mem: 3536 +Test: [1130/2387] eta: 0:04:07 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1149 (0.3935) acc1: 99.7396 (82.4100) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1735 data: 0.0797 max mem: 3536 +Test: [1140/2387] eta: 0:04:05 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1180 (0.3911) acc1: 99.7396 (82.5576) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1785 data: 0.0856 max mem: 3536 +Test: [1150/2387] eta: 0:04:02 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1271 (0.3891) acc1: 100.0000 (82.7007) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1707 data: 0.0798 max mem: 3536 +Test: [1160/2387] eta: 0:04:00 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1443 (0.3869) acc1: 100.0000 (82.8437) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1735 data: 0.0815 max mem: 3536 +Test: [1170/2387] eta: 0:03:58 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1285 (0.3847) acc1: 100.0000 (82.9862) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1896 data: 0.0940 max mem: 3536 +Test: [1180/2387] eta: 0:03:56 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1401 (0.3827) acc1: 99.7396 (83.1236) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1854 data: 0.0906 max mem: 3536 +Test: [1190/2387] eta: 0:03:54 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1422 (0.3807) acc1: 99.7396 (83.2623) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1859 data: 0.0944 max mem: 3536 +Test: [1200/2387] eta: 0:03:52 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1208 (0.3785) acc1: 100.0000 (83.4016) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1874 data: 0.0948 max mem: 3536 +Test: [1210/2387] eta: 0:03:50 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1069 (0.3763) acc1: 100.0000 (83.5385) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1997 data: 0.1073 max mem: 3536 +Test: [1220/2387] eta: 0:03:48 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0878 (0.3738) acc1: 100.0000 (83.6727) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1970 data: 0.1050 max mem: 3536 +Test: [1230/2387] eta: 0:03:46 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0696 (0.3714) acc1: 100.0000 (83.8051) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1822 data: 0.0890 max mem: 3536 +Test: [1240/2387] eta: 0:03:44 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0730 (0.3690) acc1: 100.0000 (83.9352) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1848 data: 0.0928 max mem: 3536 +Test: [1250/2387] eta: 0:03:42 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0736 (0.3666) acc1: 100.0000 (84.0632) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1831 data: 0.0918 max mem: 3536 +Test: [1260/2387] eta: 0:03:40 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0704 (0.3643) acc1: 100.0000 (84.1895) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1880 data: 0.0963 max mem: 3536 +Test: [1270/2387] eta: 0:03:38 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0667 (0.3619) acc1: 100.0000 (84.3137) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1710 data: 0.0796 max mem: 3536 +Test: [1280/2387] eta: 0:03:36 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0668 (0.3597) acc1: 100.0000 (84.4360) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1681 data: 0.0769 max mem: 3536 +Test: [1290/2387] eta: 0:03:34 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0772 (0.3575) acc1: 100.0000 (84.5531) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1772 data: 0.0839 max mem: 3536 +Test: [1300/2387] eta: 0:03:32 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0794 (0.3554) acc1: 100.0000 (84.6704) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1857 data: 0.0921 max mem: 3536 +Test: [1310/2387] eta: 0:03:30 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0811 (0.3534) acc1: 100.0000 (84.7848) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1947 data: 0.1001 max mem: 3536 +Test: [1320/2387] eta: 0:03:28 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0850 (0.3514) acc1: 100.0000 (84.8972) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1885 data: 0.0928 max mem: 3536 +Test: [1330/2387] eta: 0:03:26 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0832 (0.3494) acc1: 100.0000 (85.0068) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1874 data: 0.0943 max mem: 3536 +Test: [1340/2387] eta: 0:03:24 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1044 (0.3484) acc1: 99.7396 (85.0691) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2101 data: 0.1171 max mem: 3536 +Test: [1350/2387] eta: 0:03:21 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0596 (0.3462) acc1: 100.0000 (85.1796) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1824 data: 0.0892 max mem: 3537 +Test: [1360/2387] eta: 0:03:19 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0542 (0.3440) acc1: 100.0000 (85.2885) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1543 data: 0.0619 max mem: 3537 +Test: [1370/2387] eta: 0:03:17 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0518 (0.3419) acc1: 100.0000 (85.3958) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0851 max mem: 3537 +Test: [1380/2387] eta: 0:03:15 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0512 (0.3398) acc1: 100.0000 (85.5015) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1704 data: 0.0755 max mem: 3537 +Test: [1390/2387] eta: 0:03:13 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0518 (0.3378) acc1: 100.0000 (85.6058) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1751 data: 0.0833 max mem: 3537 +Test: [1400/2387] eta: 0:03:11 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0492 (0.3357) acc1: 100.0000 (85.7085) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1893 data: 0.0980 max mem: 3537 +Test: [1410/2387] eta: 0:03:09 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0464 (0.3340) acc1: 100.0000 (85.8061) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1859 data: 0.0948 max mem: 3537 +Test: [1420/2387] eta: 0:03:07 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1092 (0.3325) acc1: 100.0000 (85.8986) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1915 data: 0.1004 max mem: 3537 +Test: [1430/2387] eta: 0:03:05 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1133 (0.3311) acc1: 99.4792 (85.9906) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1912 data: 0.0999 max mem: 3537 +Test: [1440/2387] eta: 0:03:03 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1122 (0.3296) acc1: 99.7396 (86.0870) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1806 data: 0.0884 max mem: 3537 +Test: [1450/2387] eta: 0:03:01 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1090 (0.3281) acc1: 100.0000 (86.1782) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1753 data: 0.0831 max mem: 3537 +Test: [1460/2387] eta: 0:02:59 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1049 (0.3266) acc1: 99.7396 (86.2705) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1836 data: 0.0926 max mem: 3537 +Test: [1470/2387] eta: 0:02:57 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0964 (0.3250) acc1: 100.0000 (86.3631) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1869 data: 0.0952 max mem: 3537 +Test: [1480/2387] eta: 0:02:55 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0891 (0.3235) acc1: 100.0000 (86.4527) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1854 data: 0.0935 max mem: 3537 +Test: [1490/2387] eta: 0:02:53 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0923 (0.3220) acc1: 100.0000 (86.5409) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1878 data: 0.0961 max mem: 3537 +Test: [1500/2387] eta: 0:02:51 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1034 (0.3206) acc1: 99.7396 (86.6285) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1980 data: 0.1029 max mem: 3537 +Test: [1510/2387] eta: 0:02:49 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1046 (0.3191) acc1: 100.0000 (86.7170) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1824 data: 0.0870 max mem: 3537 +Test: [1520/2387] eta: 0:02:47 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1173 (0.3181) acc1: 100.0000 (86.8001) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1750 data: 0.0825 max mem: 3537 +Test: [1530/2387] eta: 0:02:45 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1543 (0.3171) acc1: 99.4792 (86.8773) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1710 data: 0.0781 max mem: 3537 +Test: [1540/2387] eta: 0:02:43 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1543 (0.3160) acc1: 99.7396 (86.9577) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1598 data: 0.0669 max mem: 3537 +Test: [1550/2387] eta: 0:02:41 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1479 (0.3150) acc1: 99.4792 (87.0361) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1652 data: 0.0734 max mem: 3537 +Test: [1560/2387] eta: 0:02:39 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1481 (0.3140) acc1: 99.4792 (87.1108) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1654 data: 0.0739 max mem: 3537 +Test: [1570/2387] eta: 0:02:37 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1443 (0.3128) acc1: 100.0000 (87.1902) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1549 data: 0.0632 max mem: 3537 +Test: [1580/2387] eta: 0:02:35 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1090 (0.3115) acc1: 100.0000 (87.2706) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1594 data: 0.0645 max mem: 3538 +Test: [1590/2387] eta: 0:02:33 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0995 (0.3102) acc1: 99.7396 (87.3483) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1733 data: 0.0774 max mem: 3538 +Test: [1600/2387] eta: 0:02:31 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0956 (0.3090) acc1: 99.7396 (87.4208) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1698 data: 0.0762 max mem: 3538 +Test: [1610/2387] eta: 0:02:28 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0913 (0.3076) acc1: 100.0000 (87.4977) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1579 data: 0.0647 max mem: 3538 +Test: [1620/2387] eta: 0:02:26 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0981 (0.3064) acc1: 100.0000 (87.5725) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1575 data: 0.0651 max mem: 3538 +Test: [1630/2387] eta: 0:02:24 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1056 (0.3053) acc1: 99.7396 (87.6445) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1725 data: 0.0811 max mem: 3538 +Test: [1640/2387] eta: 0:02:22 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1188 (0.3042) acc1: 99.4792 (87.7131) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1676 data: 0.0757 max mem: 3538 +Test: [1650/2387] eta: 0:02:20 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1173 (0.3030) acc1: 99.4792 (87.7819) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1739 data: 0.0790 max mem: 3538 +Test: [1660/2387] eta: 0:02:19 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1173 (0.3020) acc1: 99.4792 (87.8434) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1853 data: 0.0899 max mem: 3538 +Test: [1670/2387] eta: 0:02:17 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1219 (0.3010) acc1: 99.4792 (87.9110) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1851 data: 0.0924 max mem: 3538 +Test: [1680/2387] eta: 0:02:15 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1221 (0.3000) acc1: 99.4792 (87.9714) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1910 data: 0.0990 max mem: 3538 +Test: [1690/2387] eta: 0:02:13 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1221 (0.2990) acc1: 99.4792 (88.0367) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1960 data: 0.1023 max mem: 3538 +Test: [1700/2387] eta: 0:02:11 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1163 (0.2979) acc1: 99.4792 (88.1052) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2082 data: 0.1147 max mem: 3538 +Test: [1710/2387] eta: 0:02:09 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0914 (0.2966) acc1: 100.0000 (88.1744) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1944 data: 0.1011 max mem: 3538 +Test: [1720/2387] eta: 0:02:07 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0811 (0.2954) acc1: 100.0000 (88.2428) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1903 data: 0.0960 max mem: 3538 +Test: [1730/2387] eta: 0:02:05 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0811 (0.2942) acc1: 100.0000 (88.3094) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2054 data: 0.1111 max mem: 3538 +Test: [1740/2387] eta: 0:02:03 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0773 (0.2929) acc1: 100.0000 (88.3759) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1971 data: 0.1026 max mem: 3538 +Test: [1750/2387] eta: 0:02:01 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0745 (0.2917) acc1: 100.0000 (88.4422) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1936 data: 0.1009 max mem: 3538 +Test: [1760/2387] eta: 0:02:00 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0755 (0.2905) acc1: 100.0000 (88.5075) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1912 data: 0.0996 max mem: 3538 +Test: [1770/2387] eta: 0:01:58 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0773 (0.2894) acc1: 100.0000 (88.5695) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2022 data: 0.1106 max mem: 3538 +Test: [1780/2387] eta: 0:01:56 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0872 (0.2882) acc1: 100.0000 (88.6332) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2123 data: 0.1209 max mem: 3538 +Test: [1790/2387] eta: 0:01:54 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0798 (0.2871) acc1: 100.0000 (88.6959) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2050 data: 0.1118 max mem: 3538 +Test: [1800/2387] eta: 0:01:52 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0822 (0.2860) acc1: 99.7396 (88.7574) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2046 data: 0.1112 max mem: 3538 +Test: [1810/2387] eta: 0:01:50 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0856 (0.2848) acc1: 100.0000 (88.8192) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2138 data: 0.1196 max mem: 3539 +Test: [1820/2387] eta: 0:01:48 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0743 (0.2837) acc1: 100.0000 (88.8805) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2095 data: 0.1147 max mem: 3539 +Test: [1830/2387] eta: 0:01:46 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0743 (0.2826) acc1: 100.0000 (88.9412) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1897 data: 0.0967 max mem: 3539 +Test: [1840/2387] eta: 0:01:44 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1023 (0.2816) acc1: 100.0000 (89.0013) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1815 data: 0.0872 max mem: 3539 +Test: [1850/2387] eta: 0:01:43 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1035 (0.2806) acc1: 100.0000 (89.0607) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1798 data: 0.0858 max mem: 3539 +Test: [1860/2387] eta: 0:01:41 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1035 (0.2797) acc1: 100.0000 (89.1195) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1825 data: 0.0901 max mem: 3539 +Test: [1870/2387] eta: 0:01:39 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1032 (0.2787) acc1: 100.0000 (89.1776) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1855 data: 0.0905 max mem: 3539 +Test: [1880/2387] eta: 0:01:37 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1021 (0.2778) acc1: 100.0000 (89.2351) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1811 data: 0.0865 max mem: 3539 +Test: [1890/2387] eta: 0:01:35 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1027 (0.2769) acc1: 100.0000 (89.2921) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2148 data: 0.1204 max mem: 3539 +Test: [1900/2387] eta: 0:01:33 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1031 (0.2760) acc1: 100.0000 (89.3484) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2053 data: 0.1111 max mem: 3539 +Test: [1910/2387] eta: 0:01:31 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1056 (0.2751) acc1: 100.0000 (89.4041) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1784 data: 0.0821 max mem: 3539 +Test: [1920/2387] eta: 0:01:29 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1098 (0.2743) acc1: 100.0000 (89.4567) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2044 data: 0.1083 max mem: 3539 +Test: [1930/2387] eta: 0:01:27 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0966 (0.2734) acc1: 100.0000 (89.5057) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.1020 max mem: 3539 +Test: [1940/2387] eta: 0:01:25 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0899 (0.2726) acc1: 100.0000 (89.5560) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1850 data: 0.0923 max mem: 3539 +Test: [1950/2387] eta: 0:01:23 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0909 (0.2717) acc1: 100.0000 (89.6092) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1852 data: 0.0938 max mem: 3539 +Test: [1960/2387] eta: 0:01:21 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1049 (0.2709) acc1: 100.0000 (89.6558) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1770 data: 0.0858 max mem: 3539 +Test: [1970/2387] eta: 0:01:19 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1157 (0.2703) acc1: 100.0000 (89.7025) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1804 data: 0.0897 max mem: 3539 +Test: [1980/2387] eta: 0:01:18 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1470 (0.2697) acc1: 98.9583 (89.7390) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1820 data: 0.0912 max mem: 3539 +Test: [1990/2387] eta: 0:01:16 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1559 (0.2691) acc1: 98.6979 (89.7833) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1801 data: 0.0890 max mem: 3539 +Test: [2000/2387] eta: 0:01:14 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1576 (0.2686) acc1: 98.6979 (89.8268) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1794 data: 0.0884 max mem: 3539 +Test: [2010/2387] eta: 0:01:12 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1591 (0.2681) acc1: 98.4375 (89.8699) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1669 data: 0.0748 max mem: 3539 +Test: [2020/2387] eta: 0:01:10 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1606 (0.2675) acc1: 98.4375 (89.9129) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1670 data: 0.0729 max mem: 3539 +Test: [2030/2387] eta: 0:01:08 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1597 (0.2670) acc1: 98.4375 (89.9544) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1702 data: 0.0766 max mem: 3540 +Test: [2040/2387] eta: 0:01:06 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1577 (0.2665) acc1: 98.4375 (89.9981) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1629 data: 0.0693 max mem: 3540 +Test: [2050/2387] eta: 0:01:04 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1548 (0.2659) acc1: 98.9583 (90.0417) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1624 data: 0.0695 max mem: 3540 +Test: [2060/2387] eta: 0:01:02 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1365 (0.2653) acc1: 98.9583 (90.0861) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1573 data: 0.0661 max mem: 3540 +Test: [2070/2387] eta: 0:01:00 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1310 (0.2646) acc1: 99.4792 (90.1318) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1640 data: 0.0717 max mem: 3540 +Test: [2080/2387] eta: 0:00:58 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1302 (0.2640) acc1: 99.4792 (90.1760) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1689 data: 0.0762 max mem: 3540 +Test: [2090/2387] eta: 0:00:56 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1292 (0.2633) acc1: 99.4792 (90.2202) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1625 data: 0.0706 max mem: 3540 +Test: [2100/2387] eta: 0:00:54 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1313 (0.2627) acc1: 99.4792 (90.2643) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1663 data: 0.0743 max mem: 3540 +Test: [2110/2387] eta: 0:00:52 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1305 (0.2620) acc1: 99.4792 (90.3078) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1647 data: 0.0732 max mem: 3540 +Test: [2120/2387] eta: 0:00:50 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1261 (0.2613) acc1: 99.4792 (90.3523) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1670 data: 0.0754 max mem: 3540 +Test: [2130/2387] eta: 0:00:48 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0995 (0.2606) acc1: 100.0000 (90.3972) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1888 data: 0.0961 max mem: 3540 +Test: [2140/2387] eta: 0:00:46 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0995 (0.2599) acc1: 100.0000 (90.4388) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1800 data: 0.0880 max mem: 3540 +Test: [2150/2387] eta: 0:00:44 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1058 (0.2592) acc1: 100.0000 (90.4821) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1613 data: 0.0697 max mem: 3540 +Test: [2160/2387] eta: 0:00:43 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1048 (0.2585) acc1: 100.0000 (90.5240) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1824 data: 0.0911 max mem: 3540 +Test: [2170/2387] eta: 0:00:41 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1044 (0.2579) acc1: 100.0000 (90.5631) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1921 data: 0.1000 max mem: 3540 +Test: [2180/2387] eta: 0:00:39 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1388 (0.2583) acc1: 97.6562 (90.5328) acc5: 100.0000 (100.0000) acc1_neg: 60.417 acc5_neg: 100.000 time: 0.1823 data: 0.0893 max mem: 3540 +Test: [2190/2387] eta: 0:00:37 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.2508 (0.2586) acc1: 90.3646 (90.5196) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1753 data: 0.0837 max mem: 3540 +Test: [2200/2387] eta: 0:00:35 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1033 (0.2578) acc1: 99.7396 (90.5624) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1781 data: 0.0871 max mem: 3540 +Test: [2210/2387] eta: 0:00:33 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0878 (0.2571) acc1: 100.0000 (90.6044) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1770 data: 0.0861 max mem: 3540 +Test: [2220/2387] eta: 0:00:31 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0974 (0.2563) acc1: 100.0000 (90.6429) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1732 data: 0.0806 max mem: 3540 +Test: [2230/2387] eta: 0:00:29 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0905 (0.2556) acc1: 100.0000 (90.6838) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1844 data: 0.0912 max mem: 3540 +Test: [2240/2387] eta: 0:00:27 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0883 (0.2549) acc1: 100.0000 (90.7224) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1960 data: 0.1042 max mem: 3540 +Test: [2250/2387] eta: 0:00:25 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1021 (0.2543) acc1: 100.0000 (90.7600) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2093 data: 0.1174 max mem: 3540 +Test: [2260/2387] eta: 0:00:24 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0984 (0.2536) acc1: 100.0000 (90.8009) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2125 data: 0.1195 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0846 (0.2529) acc1: 100.0000 (90.8413) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2036 data: 0.1084 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0845 (0.2522) acc1: 100.0000 (90.8765) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2016 data: 0.1054 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0847 (0.2515) acc1: 100.0000 (90.9160) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2113 data: 0.1159 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0881 (0.2508) acc1: 100.0000 (90.9537) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2129 data: 0.1167 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0901 (0.2501) acc1: 100.0000 (90.9909) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1961 data: 0.1018 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.0972 (0.2496) acc1: 99.7396 (91.0248) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1853 data: 0.0931 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1093 (0.2490) acc1: 99.7396 (91.0593) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2024 data: 0.1096 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1119 (0.2485) acc1: 100.0000 (91.0936) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1841 data: 0.0911 max mem: 3541 +Test: [2350/2387] eta: 0:00:07 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1086 (0.2481) acc1: 100.0000 (91.1230) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1333 data: 0.0411 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1088 (0.2475) acc1: 100.0000 (91.1600) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1288 data: 0.0375 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1244 (0.2470) acc1: 100.0000 (91.1939) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1368 data: 0.0461 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1280 (0.2466) acc1: 99.4792 (91.2218) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1313 data: 0.0412 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 4.110 acc5_pos: 100.000 loss: 0.1613 (0.2465) acc1: 97.1354 (91.2327) acc5: 100.0000 (100.0000) acc1_neg: 93.023 acc5_neg: 100.000 time: 0.1283 data: 0.0405 max mem: 3541 +Test: Total time: 0:07:31 (0.1890 s / it) +* Overall Acc@1 91.233 Acc@5 100.000 + target == real Acc@1 7.347 Acc@5 100.000 + target == fake Acc@1 97.124 Acc@5 100.000 + loss 0.247 + PR-AUC 0.9849 ROC-AUC 0.8142 EER 25.74% (thr=0.8647) +Accuracy of the network on the 916267 test images: 91.2% +[rank0]:[W726 04:38:05.203167415 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:38:25.306000 1188988 torch/distributed/run.py:766] +W0726 04:38:25.306000 1188988 torch/distributed/run.py:766] ***************************************** +W0726 04:38:25.306000 1188988 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:38:25.306000 1188988 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 3): env:// +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:38:54.446825658 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:38:54.642099003 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:38:54.902391166 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:38:55.174648974 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:38:55.214880216 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:38:55.223528835 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:38:55.250363714 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W726 04:38:55.254411546 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n32_seed7_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n32_classseed7_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:04:37 acc1_pos: 5.990 acc5_pos: 100.000 loss: 1.6408 (1.6408) acc1: 5.9896 (5.9896) acc5: 100.0000 (100.0000) time: 7.4999 data: 5.7352 max mem: 3531 +Test: [ 10/517] eta: 0:06:27 acc1_pos: 4.427 acc5_pos: 100.000 loss: 1.6408 (1.5809) acc1: 4.4271 (7.9309) acc5: 100.0000 (100.0000) time: 0.7641 data: 0.5218 max mem: 3531 +Test: [ 20/517] eta: 0:03:41 acc1_pos: 2.604 acc5_pos: 100.000 loss: 1.5813 (1.5730) acc1: 4.4271 (7.8621) acc5: 100.0000 (100.0000) time: 0.0932 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:03:14 acc1_pos: 0.521 acc5_pos: 100.000 loss: 1.6525 (1.5938) acc1: 4.6875 (7.0565) acc5: 100.0000 (100.0000) time: 0.1980 data: 0.1033 max mem: 3531 +Test: [ 40/517] eta: 0:03:01 acc1_pos: 1.823 acc5_pos: 100.000 loss: 1.6525 (1.6099) acc1: 1.8229 (6.1928) acc5: 100.0000 (100.0000) time: 0.3099 data: 0.2159 max mem: 3531 +Test: [ 50/517] eta: 0:02:51 acc1_pos: 1.376 acc5_pos: 100.000 loss: 1.5709 (1.5446) acc1: 2.6042 (10.3860) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.3198 data: 0.2244 max mem: 3531 +Test: [ 60/517] eta: 0:02:37 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1835 (1.3192) acc1: 98.6979 (25.0000) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2749 data: 0.1796 max mem: 3531 +Test: [ 70/517] eta: 0:02:28 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1677 (1.1577) acc1: 100.0000 (35.5157) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2449 data: 0.1488 max mem: 3531 +Test: [ 80/517] eta: 0:02:21 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1715 (1.0367) acc1: 100.0000 (43.3963) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1561 max mem: 3531 +Test: [ 90/517] eta: 0:02:13 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1786 (0.9423) acc1: 99.7396 (49.5822) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2349 data: 0.1341 max mem: 3531 +Test: [100/517] eta: 0:02:06 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1627 (0.8642) acc1: 100.0000 (54.5741) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2299 data: 0.1308 max mem: 3531 +Test: [110/517] eta: 0:02:01 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1433 (0.7989) acc1: 100.0000 (58.6665) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2407 data: 0.1449 max mem: 3531 +Test: [120/517] eta: 0:01:57 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1579 (0.7483) acc1: 100.0000 (61.9533) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2504 data: 0.1533 max mem: 3531 +Test: [130/517] eta: 0:01:55 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1812 (0.7047) acc1: 99.2188 (64.7642) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2904 data: 0.1954 max mem: 3531 +Test: [140/517] eta: 0:01:49 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1567 (0.6654) acc1: 100.0000 (67.2632) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2644 data: 0.1733 max mem: 3531 +Test: [150/517] eta: 0:01:45 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1483 (0.6313) acc1: 100.0000 (69.4295) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2196 data: 0.1289 max mem: 3531 +Test: [160/517] eta: 0:01:41 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1550 (0.6022) acc1: 100.0000 (71.3267) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2399 data: 0.1484 max mem: 3531 +Test: [170/517] eta: 0:01:37 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1564 (0.5763) acc1: 100.0000 (72.9943) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2298 data: 0.1368 max mem: 3531 +Test: [180/517] eta: 0:01:33 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1563 (0.5527) acc1: 100.0000 (74.4792) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2254 data: 0.1318 max mem: 3531 +Test: [190/517] eta: 0:01:30 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1398 (0.5309) acc1: 100.0000 (75.8153) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2345 data: 0.1391 max mem: 3531 +Test: [200/517] eta: 0:01:26 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1374 (0.5113) acc1: 100.0000 (77.0186) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2294 data: 0.1324 max mem: 3531 +Test: [210/517] eta: 0:01:23 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1418 (0.4969) acc1: 99.2188 (77.9966) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2345 data: 0.1390 max mem: 3532 +Test: [220/517] eta: 0:01:20 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.2149 (0.4842) acc1: 97.1354 (78.8556) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2399 data: 0.1437 max mem: 3532 +Test: [230/517] eta: 0:01:17 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1886 (0.4706) acc1: 98.1771 (79.7608) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2366 data: 0.1412 max mem: 3532 +Test: [240/517] eta: 0:01:14 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1673 (0.4584) acc1: 100.0000 (80.5995) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2349 data: 0.1427 max mem: 3532 +Test: [250/517] eta: 0:01:10 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1616 (0.4463) acc1: 100.0000 (81.3683) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2333 data: 0.1402 max mem: 3532 +Test: [260/517] eta: 0:01:08 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1474 (0.4349) acc1: 100.0000 (82.0791) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2399 data: 0.1441 max mem: 3532 +Test: [270/517] eta: 0:01:05 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1531 (0.4247) acc1: 100.0000 (82.7385) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2499 data: 0.1540 max mem: 3532 +Test: [280/517] eta: 0:01:02 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1615 (0.4158) acc1: 100.0000 (83.3352) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2449 data: 0.1499 max mem: 3532 +Test: [290/517] eta: 0:00:59 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1743 (0.4076) acc1: 99.7396 (83.8926) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2349 data: 0.1397 max mem: 3532 +Test: [300/517] eta: 0:00:56 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1664 (0.3993) acc1: 99.7396 (84.4252) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2249 data: 0.1292 max mem: 3532 +Test: [310/517] eta: 0:00:53 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1527 (0.3913) acc1: 100.0000 (84.9260) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2299 data: 0.1341 max mem: 3532 +Test: [320/517] eta: 0:00:51 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1458 (0.3836) acc1: 100.0000 (85.3948) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2762 data: 0.1813 max mem: 3532 +Test: [330/517] eta: 0:00:48 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1410 (0.3761) acc1: 100.0000 (85.8352) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2709 data: 0.1770 max mem: 3532 +Test: [340/517] eta: 0:00:46 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1286 (0.3697) acc1: 100.0000 (86.2369) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2388 data: 0.1450 max mem: 3532 +Test: [350/517] eta: 0:00:43 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1579 (0.3643) acc1: 100.0000 (86.6238) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2347 data: 0.1417 max mem: 3532 +Test: [360/517] eta: 0:00:40 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1664 (0.3587) acc1: 100.0000 (86.9900) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2404 data: 0.1463 max mem: 3532 +Test: [370/517] eta: 0:00:38 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1552 (0.3529) acc1: 100.0000 (87.3407) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2954 data: 0.1997 max mem: 3532 +Test: [380/517] eta: 0:00:35 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1440 (0.3477) acc1: 100.0000 (87.6716) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2901 data: 0.1968 max mem: 3532 +Test: [390/517] eta: 0:00:33 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1593 (0.3431) acc1: 100.0000 (87.9869) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2287 data: 0.1346 max mem: 3532 +Test: [400/517] eta: 0:00:30 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1543 (0.3382) acc1: 100.0000 (88.2858) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2143 data: 0.1184 max mem: 3532 +Test: [410/517] eta: 0:00:27 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1600 (0.3343) acc1: 100.0000 (88.5708) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2198 data: 0.1258 max mem: 3532 +Test: [420/517] eta: 0:00:24 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1525 (0.3297) acc1: 100.0000 (88.8411) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2257 data: 0.1335 max mem: 3532 +Test: [430/517] eta: 0:00:22 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1497 (0.3257) acc1: 100.0000 (89.0957) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2377 data: 0.1434 max mem: 3532 +Test: [440/517] eta: 0:00:19 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1662 (0.3225) acc1: 99.7396 (89.3341) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2407 data: 0.1452 max mem: 3533 +Test: [450/517] eta: 0:00:17 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1732 (0.3190) acc1: 100.0000 (89.5701) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2637 data: 0.1701 max mem: 3533 +Test: [460/517] eta: 0:00:14 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1732 (0.3163) acc1: 100.0000 (89.7890) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2678 data: 0.1743 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1648 (0.3128) acc1: 100.0000 (90.0035) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2433 data: 0.1485 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1548 (0.3096) acc1: 100.0000 (90.2114) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1568 max mem: 3533 +Test: [490/517] eta: 0:00:06 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1521 (0.3063) acc1: 100.0000 (90.4107) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2749 data: 0.1762 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1540 (0.3035) acc1: 100.0000 (90.6016) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2449 data: 0.1491 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1709 (0.3010) acc1: 100.0000 (90.7809) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2230 data: 0.1279 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1435 (0.2991) acc1: 100.0000 (90.8833) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2212 data: 0.1159 max mem: 3533 +Test: Total time: 0:02:12 (0.2568 s / it) +* Overall Acc@1 90.883 Acc@5 100.000 + target == real Acc@1 5.898 Acc@5 100.000 + target == fake Acc@1 99.700 Acc@5 100.000 + loss 0.299 + PR-AUC 0.9500 ROC-AUC 0.7171 EER 33.97% (thr=0.8481) +Accuracy of the network on the 198427 test images: 90.9% +[rank0]:[W726 04:41:19.016543340 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 04:41:40.804000 1189558 torch/distributed/run.py:766] +W0726 04:41:40.804000 1189558 torch/distributed/run.py:766] ***************************************** +W0726 04:41:40.804000 1189558 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:41:40.804000 1189558 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:42:09.042887730 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:42:10.214392895 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:42:10.348337965 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +| distributed init (rank 5): env:// +| distributed init (rank 6): env:// +| distributed init (rank 4): env:// +[rank0]:[W726 04:42:10.625174711 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:42:10.626639913 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W726 04:42:10.626716867 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:42:10.630774233 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank4]:[W726 04:42:10.630870929 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n32_seed7_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n32_classseed7_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:31:16 acc1_pos: 30.990 acc5_pos: 100.000 loss: 0.9156 (0.9156) acc1: 30.9896 (30.9896) acc5: 100.0000 (100.0000) time: 4.8230 data: 2.8574 max mem: 3531 +Test: [ 10/389] eta: 0:03:17 acc1_pos: 33.594 acc5_pos: 100.000 loss: 0.9156 (1.0007) acc1: 30.9896 (27.2491) acc5: 100.0000 (100.0000) time: 0.5207 data: 0.2602 max mem: 3531 +Test: [ 20/389] eta: 0:02:10 acc1_pos: 0.260 acc5_pos: 100.000 loss: 1.0042 (1.1190) acc1: 17.9688 (21.1930) acc5: 100.0000 (100.0000) time: 0.1297 data: 0.0396 max mem: 3531 +Test: [ 30/389] eta: 0:01:44 acc1_pos: 7.292 acc5_pos: 100.000 loss: 1.1182 (1.1314) acc1: 2.3438 (15.7174) acc5: 100.0000 (100.0000) time: 0.1633 data: 0.0729 max mem: 3531 +Test: [ 40/389] eta: 0:01:36 acc1_pos: 3.646 acc5_pos: 100.000 loss: 1.0977 (1.1051) acc1: 5.9896 (17.3463) acc5: 100.0000 (100.0000) time: 0.1945 data: 0.1036 max mem: 3531 +Test: [ 50/389] eta: 0:01:28 acc1_pos: 31.250 acc5_pos: 100.000 loss: 1.0425 (1.1058) acc1: 12.5000 (16.5901) acc5: 100.0000 (100.0000) time: 0.2171 data: 0.1252 max mem: 3531 +Test: [ 60/389] eta: 0:01:23 acc1_pos: 14.583 acc5_pos: 100.000 loss: 0.9660 (1.0658) acc1: 26.0417 (19.7532) acc5: 100.0000 (100.0000) time: 0.2042 data: 0.1110 max mem: 3531 +Test: [ 70/389] eta: 0:01:17 acc1_pos: 60.156 acc5_pos: 100.000 loss: 0.8134 (1.0318) acc1: 34.1146 (22.5719) acc5: 100.0000 (100.0000) time: 0.2001 data: 0.1049 max mem: 3531 +Test: [ 80/389] eta: 0:01:14 acc1_pos: 82.031 acc5_pos: 100.000 loss: 0.8134 (1.0156) acc1: 34.6354 (24.6817) acc5: 100.0000 (100.0000) time: 0.2067 data: 0.1131 max mem: 3531 +Test: [ 90/389] eta: 0:01:10 acc1_pos: 25.781 acc5_pos: 100.000 loss: 0.7230 (0.9770) acc1: 53.3854 (28.1908) acc5: 100.0000 (100.0000) time: 0.2088 data: 0.1168 max mem: 3531 +Test: [100/389] eta: 0:01:04 acc1_pos: 47.917 acc5_pos: 100.000 loss: 0.8422 (0.9813) acc1: 43.2292 (28.3003) acc5: 100.0000 (100.0000) time: 0.1542 data: 0.0599 max mem: 3531 +Test: [110/389] eta: 0:01:01 acc1_pos: 95.833 acc5_pos: 100.000 loss: 0.9368 (0.9788) acc1: 29.4271 (29.6406) acc5: 100.0000 (100.0000) time: 0.1451 data: 0.0497 max mem: 3531 +Test: [120/389] eta: 0:00:58 acc1_pos: 36.979 acc5_pos: 100.000 loss: 0.8776 (0.9652) acc1: 36.9792 (30.8217) acc5: 100.0000 (100.0000) time: 0.1888 data: 0.0923 max mem: 3531 +Test: [130/389] eta: 0:00:55 acc1_pos: 17.791 acc5_pos: 100.000 loss: 0.5596 (0.9332) acc1: 59.1146 (33.8939) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1904 data: 0.0953 max mem: 3531 +Test: [140/389] eta: 0:00:53 acc1_pos: 17.791 acc5_pos: 100.000 loss: 0.3476 (0.8951) acc1: 92.1875 (37.7567) acc5: 100.0000 (100.0000) acc1_neg: 58.333 acc5_neg: 100.000 time: 0.1865 data: 0.0946 max mem: 3531 +Test: [150/389] eta: 0:00:50 acc1_pos: 17.791 acc5_pos: 100.000 loss: 0.3476 (0.8619) acc1: 96.6146 (41.2528) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1838 data: 0.0928 max mem: 3531 +Test: [160/389] eta: 0:00:47 acc1_pos: 17.791 acc5_pos: 100.000 loss: 0.3702 (0.8256) acc1: 96.6146 (44.5393) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1787 data: 0.0878 max mem: 3531 +Test: [170/389] eta: 0:00:45 acc1_pos: 17.791 acc5_pos: 100.000 loss: 0.1814 (0.7912) acc1: 100.0000 (47.6060) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1807 data: 0.0901 max mem: 3531 +Test: [180/389] eta: 0:00:43 acc1_pos: 17.791 acc5_pos: 100.000 loss: 0.2205 (0.7639) acc1: 99.4792 (50.3050) acc5: 100.0000 (100.0000) acc1_neg: 79.427 acc5_neg: 100.000 time: 0.1795 data: 0.0888 max mem: 3531 +Test: [190/389] eta: 0:00:41 acc1_pos: 17.791 acc5_pos: 100.000 loss: 0.3180 (0.7474) acc1: 97.6562 (52.0261) acc5: 100.0000 (100.0000) acc1_neg: 60.417 acc5_neg: 100.000 time: 0.1932 data: 0.1023 max mem: 3531 +Test: [200/389] eta: 0:00:38 acc1_pos: 17.791 acc5_pos: 100.000 loss: 0.3135 (0.7303) acc1: 98.9583 (53.9399) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1824 data: 0.0886 max mem: 3531 +Test: [210/389] eta: 0:00:36 acc1_pos: 17.791 acc5_pos: 100.000 loss: 0.3053 (0.7070) acc1: 99.7396 (56.1056) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1650 data: 0.0712 max mem: 3532 +Test: [220/389] eta: 0:00:33 acc1_pos: 17.791 acc5_pos: 100.000 loss: 0.1769 (0.6847) acc1: 100.0000 (58.0717) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1748 data: 0.0822 max mem: 3532 +Test: [230/389] eta: 0:00:31 acc1_pos: 17.791 acc5_pos: 100.000 loss: 0.3312 (0.6723) acc1: 98.4375 (59.4539) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1699 data: 0.0756 max mem: 3532 +Test: [240/389] eta: 0:00:29 acc1_pos: 17.791 acc5_pos: 100.000 loss: 0.3574 (0.6589) acc1: 95.5729 (60.8792) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1776 data: 0.0845 max mem: 3532 +Test: [250/389] eta: 0:00:27 acc1_pos: 17.791 acc5_pos: 100.000 loss: 0.2935 (0.6427) acc1: 99.2188 (62.4315) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1817 data: 0.0874 max mem: 3532 +Test: [260/389] eta: 0:00:25 acc1_pos: 17.791 acc5_pos: 100.000 loss: 0.2517 (0.6321) acc1: 100.0000 (63.4030) acc5: 100.0000 (100.0000) acc1_neg: 33.333 acc5_neg: 100.000 time: 0.1728 data: 0.0787 max mem: 3532 +Test: [270/389] eta: 0:00:23 acc1_pos: 17.791 acc5_pos: 100.000 loss: 0.3680 (0.6295) acc1: 96.0938 (63.6743) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1831 data: 0.0898 max mem: 3532 +Test: [280/389] eta: 0:00:21 acc1_pos: 17.791 acc5_pos: 100.000 loss: 0.3978 (0.6240) acc1: 83.5938 (64.1987) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1888 data: 0.0925 max mem: 3532 +Test: [290/389] eta: 0:00:19 acc1_pos: 17.791 acc5_pos: 100.000 loss: 0.3664 (0.6146) acc1: 93.7500 (65.1480) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1801 data: 0.0861 max mem: 3532 +Test: [300/389] eta: 0:00:17 acc1_pos: 17.791 acc5_pos: 100.000 loss: 0.3210 (0.6054) acc1: 97.9167 (66.1874) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1709 data: 0.0782 max mem: 3532 +Test: [310/389] eta: 0:00:15 acc1_pos: 17.791 acc5_pos: 100.000 loss: 0.3349 (0.6015) acc1: 95.3125 (66.6709) acc5: 100.0000 (100.0000) acc1_neg: 47.135 acc5_neg: 100.000 time: 0.1747 data: 0.0807 max mem: 3532 +Test: [320/389] eta: 0:00:13 acc1_pos: 17.791 acc5_pos: 100.000 loss: 0.4042 (0.5940) acc1: 88.5417 (67.4641) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1798 data: 0.0869 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 17.791 acc5_pos: 100.000 loss: 0.4383 (0.5894) acc1: 95.5729 (68.2347) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1706 data: 0.0776 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 17.791 acc5_pos: 100.000 loss: 0.2701 (0.5788) acc1: 97.1354 (69.0876) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1859 data: 0.0928 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 17.791 acc5_pos: 100.000 loss: 0.2378 (0.5807) acc1: 99.2188 (68.8272) acc5: 100.0000 (100.0000) acc1_neg: 40.625 acc5_neg: 100.000 time: 0.1873 data: 0.0951 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 17.791 acc5_pos: 100.000 loss: 0.5662 (0.5817) acc1: 68.4896 (68.6894) acc5: 100.0000 (100.0000) acc1_neg: 72.656 acc5_neg: 100.000 time: 0.1580 data: 0.0653 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 17.791 acc5_pos: 100.000 loss: 0.5447 (0.5760) acc1: 71.8750 (69.2477) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1285 data: 0.0360 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 17.791 acc5_pos: 100.000 loss: 0.3651 (0.5703) acc1: 92.4479 (69.8422) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1199 data: 0.0290 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 17.791 acc5_pos: 100.000 loss: 0.3370 (0.5656) acc1: 97.1354 (70.3504) acc5: 100.0000 (100.0000) acc1_neg: 98.361 acc5_neg: 100.000 time: 0.1161 data: 0.0218 max mem: 3532 +Test: Total time: 0:01:12 (0.1865 s / it) +* Overall Acc@1 70.350 Acc@5 100.000 + target == real Acc@1 30.567 Acc@5 100.000 + target == fake Acc@1 88.843 Acc@5 100.000 + loss 0.566 + PR-AUC 0.8340 ROC-AUC 0.7112 EER 34.65% (thr=0.6865) +Accuracy of the network on the 149358 test images: 70.4% +[rank0]:[W726 04:43:31.139240650 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:43:51.183000 1190117 torch/distributed/run.py:766] +W0726 04:43:51.183000 1190117 torch/distributed/run.py:766] ***************************************** +W0726 04:43:51.183000 1190117 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:43:51.183000 1190117 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:44:20.777792199 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:44:21.266520729 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W726 04:44:21.271100286 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:44:21.505127627 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:44:21.670084809 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +[rank0]:[W726 04:44:21.880810596 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:44:21.888359395 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:44:21.894186998 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n32_seed7_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n32_classseed7_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:40:29 acc1_pos: 85.938 acc5_pos: 100.000 loss: 0.3590 (0.3590) acc1: 85.9375 (85.9375) acc5: 100.0000 (100.0000) time: 7.2948 data: 5.4415 max mem: 3531 +Test: [ 10/333] eta: 0:04:02 acc1_pos: 85.677 acc5_pos: 100.000 loss: 0.2806 (0.2891) acc1: 87.7604 (88.5890) acc5: 100.0000 (100.0000) time: 0.7515 data: 0.4951 max mem: 3531 +Test: [ 20/333] eta: 0:02:18 acc1_pos: 94.010 acc5_pos: 100.000 loss: 0.2810 (0.2991) acc1: 88.0208 (88.5417) acc5: 100.0000 (100.0000) time: 0.0985 data: 0.0004 max mem: 3531 +Test: [ 30/333] eta: 0:02:00 acc1_pos: 49.219 acc5_pos: 100.000 loss: 0.3940 (0.4899) acc1: 73.9583 (74.4960) acc5: 100.0000 (100.0000) time: 0.2049 data: 0.1072 max mem: 3531 +Test: [ 40/333] eta: 0:01:41 acc1_pos: 40.365 acc5_pos: 100.000 loss: 0.9043 (0.5987) acc1: 42.7083 (66.2665) acc5: 100.0000 (100.0000) time: 0.2499 data: 0.1541 max mem: 3531 +Test: [ 50/333] eta: 0:01:34 acc1_pos: 43.229 acc5_pos: 100.000 loss: 0.9043 (0.6625) acc1: 41.4062 (61.4328) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1339 max mem: 3531 +Test: [ 60/333] eta: 0:01:23 acc1_pos: 29.948 acc5_pos: 100.000 loss: 0.8756 (0.7006) acc1: 42.9688 (58.6919) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1287 max mem: 3531 +Test: [ 70/333] eta: 0:01:16 acc1_pos: 44.531 acc5_pos: 100.000 loss: 0.8756 (0.7292) acc1: 42.9688 (56.4921) acc5: 100.0000 (100.0000) time: 0.1799 data: 0.0834 max mem: 3531 +Test: [ 80/333] eta: 0:01:09 acc1_pos: 48.177 acc5_pos: 100.000 loss: 0.8863 (0.7513) acc1: 44.5312 (54.9190) acc5: 100.0000 (100.0000) time: 0.1799 data: 0.0844 max mem: 3531 +Test: [ 90/333] eta: 0:01:06 acc1_pos: 50.521 acc5_pos: 100.000 loss: 0.8444 (0.7584) acc1: 48.1771 (54.3555) acc5: 100.0000 (100.0000) time: 0.2152 data: 0.1212 max mem: 3531 +Test: [100/333] eta: 0:01:01 acc1_pos: 56.771 acc5_pos: 100.000 loss: 0.8407 (0.7699) acc1: 48.9583 (53.5814) acc5: 100.0000 (100.0000) time: 0.2149 data: 0.1187 max mem: 3531 +Test: [110/333] eta: 0:00:55 acc1_pos: 52.865 acc5_pos: 100.000 loss: 0.8600 (0.7785) acc1: 47.3958 (53.0781) acc5: 100.0000 (100.0000) time: 0.1446 data: 0.0467 max mem: 3531 +Test: [120/333] eta: 0:00:52 acc1_pos: 59.896 acc5_pos: 100.000 loss: 0.8600 (0.7848) acc1: 47.3958 (52.5052) acc5: 100.0000 (100.0000) time: 0.1549 data: 0.0587 max mem: 3531 +Test: [130/333] eta: 0:00:49 acc1_pos: 62.500 acc5_pos: 100.000 loss: 0.7634 (0.7882) acc1: 48.9583 (52.2404) acc5: 100.0000 (100.0000) time: 0.2199 data: 0.1243 max mem: 3531 +Test: [140/333] eta: 0:00:46 acc1_pos: 69.792 acc5_pos: 100.000 loss: 0.8122 (0.7921) acc1: 48.9583 (52.0261) acc5: 100.0000 (100.0000) time: 0.2199 data: 0.1246 max mem: 3531 +Test: [150/333] eta: 0:00:43 acc1_pos: 50.521 acc5_pos: 100.000 loss: 0.8559 (0.7974) acc1: 45.8333 (51.6401) acc5: 100.0000 (100.0000) time: 0.1699 data: 0.0744 max mem: 3531 +Test: [160/333] eta: 0:00:40 acc1_pos: 39.145 acc5_pos: 100.000 loss: 0.8607 (0.8018) acc1: 47.9167 (51.4380) acc5: 100.0000 (100.0000) acc1_neg: 55.729 acc5_neg: 100.000 time: 0.1599 data: 0.0621 max mem: 3531 +Test: [170/333] eta: 0:00:37 acc1_pos: 39.145 acc5_pos: 100.000 loss: 0.8231 (0.7996) acc1: 53.3854 (51.8305) acc5: 100.0000 (100.0000) acc1_neg: 61.198 acc5_neg: 100.000 time: 0.2099 data: 0.1105 max mem: 3531 +Test: [180/333] eta: 0:00:35 acc1_pos: 39.145 acc5_pos: 100.000 loss: 0.7057 (0.7929) acc1: 61.1979 (52.5035) acc5: 100.0000 (100.0000) acc1_neg: 64.062 acc5_neg: 100.000 time: 0.2149 data: 0.1166 max mem: 3531 +Test: [190/333] eta: 0:00:32 acc1_pos: 39.145 acc5_pos: 100.000 loss: 0.6480 (0.7822) acc1: 64.0625 (53.4222) acc5: 100.0000 (100.0000) acc1_neg: 64.062 acc5_neg: 100.000 time: 0.1599 data: 0.0636 max mem: 3531 +Test: [200/333] eta: 0:00:29 acc1_pos: 39.145 acc5_pos: 100.000 loss: 0.6191 (0.7737) acc1: 66.6667 (54.1939) acc5: 100.0000 (100.0000) acc1_neg: 63.802 acc5_neg: 100.000 time: 0.1649 data: 0.0683 max mem: 3531 +Test: [210/333] eta: 0:00:27 acc1_pos: 39.145 acc5_pos: 100.000 loss: 0.6724 (0.7713) acc1: 64.8438 (54.5814) acc5: 100.0000 (100.0000) acc1_neg: 52.865 acc5_neg: 100.000 time: 0.2012 data: 0.1065 max mem: 3532 +Test: [220/333] eta: 0:00:25 acc1_pos: 39.145 acc5_pos: 100.000 loss: 0.7251 (0.7716) acc1: 61.1979 (54.6922) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.2007 data: 0.1091 max mem: 3532 +Test: [230/333] eta: 0:00:22 acc1_pos: 39.145 acc5_pos: 100.000 loss: 0.5106 (0.7587) acc1: 75.0000 (55.7923) acc5: 100.0000 (100.0000) acc1_neg: 74.740 acc5_neg: 100.000 time: 0.1957 data: 0.1047 max mem: 3532 +Test: [240/333] eta: 0:00:20 acc1_pos: 39.145 acc5_pos: 100.000 loss: 0.5085 (0.7503) acc1: 77.0833 (56.4845) acc5: 100.0000 (100.0000) acc1_neg: 71.615 acc5_neg: 100.000 time: 0.1857 data: 0.0929 max mem: 3532 +Test: [250/333] eta: 0:00:17 acc1_pos: 39.145 acc5_pos: 100.000 loss: 0.5838 (0.7443) acc1: 70.3125 (57.0053) acc5: 100.0000 (100.0000) acc1_neg: 64.062 acc5_neg: 100.000 time: 0.1639 data: 0.0698 max mem: 3532 +Test: [260/333] eta: 0:00:15 acc1_pos: 39.145 acc5_pos: 100.000 loss: 0.6143 (0.7405) acc1: 66.9271 (57.3924) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.1802 data: 0.0858 max mem: 3532 +Test: [270/333] eta: 0:00:13 acc1_pos: 39.145 acc5_pos: 100.000 loss: 0.5828 (0.7339) acc1: 69.2708 (57.9413) acc5: 100.0000 (100.0000) acc1_neg: 71.615 acc5_neg: 100.000 time: 0.1888 data: 0.0923 max mem: 3532 +Test: [280/333] eta: 0:00:11 acc1_pos: 39.145 acc5_pos: 100.000 loss: 0.5752 (0.7322) acc1: 69.5312 (58.0933) acc5: 100.0000 (100.0000) acc1_neg: 68.229 acc5_neg: 100.000 time: 0.1729 data: 0.0774 max mem: 3532 +Test: [290/333] eta: 0:00:09 acc1_pos: 39.145 acc5_pos: 100.000 loss: 0.7779 (0.7344) acc1: 54.6875 (57.9181) acc5: 100.0000 (100.0000) acc1_neg: 57.031 acc5_neg: 100.000 time: 0.1776 data: 0.0850 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 39.145 acc5_pos: 100.000 loss: 0.8030 (0.7368) acc1: 53.9062 (57.8532) acc5: 100.0000 (100.0000) acc1_neg: 56.510 acc5_neg: 100.000 time: 0.1725 data: 0.0811 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 39.145 acc5_pos: 100.000 loss: 0.7077 (0.7311) acc1: 60.1562 (58.3266) acc5: 100.0000 (100.0000) acc1_neg: 72.396 acc5_neg: 100.000 time: 0.1654 data: 0.0741 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 39.145 acc5_pos: 100.000 loss: 0.5237 (0.7260) acc1: 73.1771 (58.7682) acc5: 100.0000 (100.0000) acc1_neg: 53.906 acc5_neg: 100.000 time: 0.1468 data: 0.0553 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 39.145 acc5_pos: 100.000 loss: 0.7661 (0.7298) acc1: 57.0312 (58.6016) acc5: 100.0000 (100.0000) acc1_neg: 54.688 acc5_neg: 100.000 time: 0.1163 data: 0.0251 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 39.145 acc5_pos: 100.000 loss: 0.8262 (0.7306) acc1: 53.9062 (58.5640) acc5: 100.0000 (100.0000) acc1_neg: 51.549 acc5_neg: 100.000 time: 0.1195 data: 0.0234 max mem: 3532 +Test: Total time: 0:01:07 (0.2029 s / it) +* Overall Acc@1 58.564 Acc@5 100.000 + target == real Acc@1 51.443 Acc@5 100.000 + target == fake Acc@1 64.903 Acc@5 100.000 + loss 0.731 + PR-AUC 0.6198 ROC-AUC 0.6129 EER 41.95% (thr=0.5586) +Accuracy of the network on the 127843 test images: 58.6% +[rank0]:[W726 04:45:40.158624791 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:46:01.127000 1190675 torch/distributed/run.py:766] +W0726 04:46:01.127000 1190675 torch/distributed/run.py:766] ***************************************** +W0726 04:46:01.127000 1190675 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:46:01.127000 1190675 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:46:32.662991360 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:46:32.875454946 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:46:32.001062243 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:46:32.058346643 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:46:33.177539260 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:46:33.207651595 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:46:33.219743115 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:46:33.251143953 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n32_seed7_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n32_classseed7_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:19:47 acc1_pos: 94.271 acc5_pos: 100.000 loss: 0.2486 (0.2486) acc1: 94.2708 (94.2708) acc5: 100.0000 (100.0000) time: 5.0511 data: 3.2383 max mem: 3531 +Test: [ 10/235] eta: 0:02:03 acc1_pos: 93.750 acc5_pos: 100.000 loss: 0.2533 (0.2541) acc1: 94.2708 (94.2708) acc5: 100.0000 (100.0000) time: 0.5476 data: 0.3011 max mem: 3531 +Test: [ 20/235] eta: 0:01:12 acc1_pos: 94.792 acc5_pos: 100.000 loss: 0.2511 (0.2508) acc1: 94.2708 (94.3452) acc5: 100.0000 (100.0000) time: 0.1009 data: 0.0079 max mem: 3531 +Test: [ 30/235] eta: 0:00:54 acc1_pos: 96.354 acc5_pos: 100.000 loss: 0.2456 (0.2487) acc1: 94.5312 (94.5397) acc5: 100.0000 (100.0000) time: 0.1072 data: 0.0077 max mem: 3531 +Test: [ 40/235] eta: 0:00:49 acc1_pos: 93.750 acc5_pos: 100.000 loss: 0.2445 (0.2486) acc1: 95.0521 (94.5313) acc5: 100.0000 (100.0000) time: 0.1699 data: 0.0701 max mem: 3531 +Test: [ 50/235] eta: 0:00:45 acc1_pos: 93.750 acc5_pos: 100.000 loss: 0.2467 (0.2495) acc1: 94.2708 (94.5364) acc5: 100.0000 (100.0000) time: 0.2207 data: 0.1248 max mem: 3531 +Test: [ 60/235] eta: 0:00:42 acc1_pos: 93.229 acc5_pos: 100.000 loss: 0.2524 (0.2508) acc1: 94.2708 (94.4672) acc5: 100.0000 (100.0000) time: 0.2169 data: 0.1237 max mem: 3531 +Test: [ 70/235] eta: 0:00:39 acc1_pos: 98.462 acc5_pos: 100.000 loss: 0.2672 (0.3694) acc1: 91.4062 (86.0145) acc5: 100.0000 (100.0000) acc1_neg: 23.177 acc5_neg: 100.000 time: 0.2191 data: 0.1267 max mem: 3531 +Test: [ 80/235] eta: 0:00:36 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.2034 (0.4738) acc1: 26.5625 (78.5848) acc5: 100.0000 (100.0000) acc1_neg: 22.917 acc5_neg: 100.000 time: 0.2179 data: 0.1234 max mem: 3531 +Test: [ 90/235] eta: 0:00:33 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.2173 (0.5556) acc1: 25.7812 (72.7965) acc5: 100.0000 (100.0000) acc1_neg: 26.302 acc5_neg: 100.000 time: 0.1956 data: 0.1012 max mem: 3531 +Test: [100/235] eta: 0:00:30 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.2173 (0.6201) acc1: 25.0000 (68.2292) acc5: 100.0000 (100.0000) acc1_neg: 25.000 acc5_neg: 100.000 time: 0.1899 data: 0.0977 max mem: 3531 +Test: [110/235] eta: 0:00:28 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.1906 (0.6724) acc1: 26.8229 (64.5247) acc5: 100.0000 (100.0000) acc1_neg: 29.427 acc5_neg: 100.000 time: 0.2193 data: 0.1255 max mem: 3531 +Test: [120/235] eta: 0:00:26 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.2470 (0.7389) acc1: 23.6979 (60.4102) acc5: 100.0000 (100.0000) acc1_neg: 15.625 acc5_neg: 100.000 time: 0.2199 data: 0.1227 max mem: 3531 +Test: [130/235] eta: 0:00:23 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4982 (0.7974) acc1: 14.3229 (56.8881) acc5: 100.0000 (100.0000) acc1_neg: 13.542 acc5_neg: 100.000 time: 0.1856 data: 0.0900 max mem: 3531 +Test: [140/235] eta: 0:00:20 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.5179 (0.8498) acc1: 13.8021 (53.8065) acc5: 100.0000 (100.0000) acc1_neg: 10.156 acc5_neg: 100.000 time: 0.1899 data: 0.0955 max mem: 3531 +Test: [150/235] eta: 0:00:18 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.5109 (0.8928) acc1: 13.8021 (51.1400) acc5: 100.0000 (100.0000) acc1_neg: 15.885 acc5_neg: 100.000 time: 0.2092 data: 0.1142 max mem: 3531 +Test: [160/235] eta: 0:00:16 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.5013 (0.9307) acc1: 14.3229 (48.8532) acc5: 100.0000 (100.0000) acc1_neg: 14.844 acc5_neg: 100.000 time: 0.2011 data: 0.1080 max mem: 3531 +Test: [170/235] eta: 0:00:14 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4740 (0.9462) acc1: 15.3646 (47.6563) acc5: 100.0000 (100.0000) acc1_neg: 28.125 acc5_neg: 100.000 time: 0.1999 data: 0.1067 max mem: 3531 +Test: [180/235] eta: 0:00:11 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.1526 (0.9572) acc1: 30.4688 (46.7498) acc5: 100.0000 (100.0000) acc1_neg: 32.031 acc5_neg: 100.000 time: 0.2002 data: 0.1076 max mem: 3531 +Test: [190/235] eta: 0:00:09 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.1419 (0.9668) acc1: 30.9896 (45.9424) acc5: 100.0000 (100.0000) acc1_neg: 31.771 acc5_neg: 100.000 time: 0.2018 data: 0.1103 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.1534 (0.9768) acc1: 30.2083 (45.1143) acc5: 100.0000 (100.0000) acc1_neg: 28.385 acc5_neg: 100.000 time: 0.1900 data: 0.0956 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.1746 (0.9845) acc1: 30.2083 (44.4967) acc5: 100.0000 (100.0000) acc1_neg: 32.031 acc5_neg: 100.000 time: 0.1958 data: 0.1020 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.1271 (0.9910) acc1: 31.7708 (43.9032) acc5: 100.0000 (100.0000) acc1_neg: 28.906 acc5_neg: 100.000 time: 0.1946 data: 0.1025 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.1306 (0.9972) acc1: 31.2500 (43.3780) acc5: 100.0000 (100.0000) acc1_neg: 31.771 acc5_neg: 100.000 time: 0.1542 data: 0.0619 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.1396 (1.0011) acc1: 30.7292 (43.1981) acc5: 100.0000 (100.0000) acc1_neg: 21.739 acc5_neg: 100.000 time: 0.1239 data: 0.0349 max mem: 3532 +Test: Total time: 0:00:48 (0.2069 s / it) +* Overall Acc@1 43.198 Acc@5 100.000 + target == real Acc@1 94.471 Acc@5 100.000 + target == fake Acc@1 24.654 Acc@5 100.000 + loss 1.001 + PR-AUC 0.8708 ROC-AUC 0.7167 EER 34.13% (thr=0.2091) +Accuracy of the network on the 89879 test images: 43.2% +[rank0]:[W726 04:47:31.643077986 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:47:50.825000 1191236 torch/distributed/run.py:766] +W0726 04:47:50.825000 1191236 torch/distributed/run.py:766] ***************************************** +W0726 04:47:50.825000 1191236 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:47:50.825000 1191236 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:48:22.385978800 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:48:22.128995095 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:48:23.181204572 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank4]:[W726 04:48:23.181350214 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:48:23.304352774 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:48:23.408370731 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:48:23.629572304 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:48:23.777418204 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n32_seed7_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n32_classseed7_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:22:14 acc1_pos: 46.875 acc5_pos: 100.000 loss: 0.7545 (0.7545) acc1: 46.8750 (46.8750) acc5: 100.0000 (100.0000) time: 4.9622 data: 2.9178 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/269] eta: 0:02:18 acc1_pos: 78.646 acc5_pos: 100.000 loss: 0.6470 (0.6431) acc1: 64.3229 (63.7784) acc5: 100.0000 (100.0000) time: 0.5332 data: 0.2656 max mem: 3531 +Test: [ 20/269] eta: 0:01:20 acc1_pos: 66.406 acc5_pos: 100.000 loss: 0.6297 (0.6271) acc1: 65.6250 (64.9182) acc5: 100.0000 (100.0000) time: 0.0903 data: 0.0004 max mem: 3531 +Test: [ 30/269] eta: 0:00:59 acc1_pos: 66.927 acc5_pos: 100.000 loss: 0.6079 (0.6188) acc1: 66.9271 (65.8266) acc5: 100.0000 (100.0000) time: 0.0903 data: 0.0004 max mem: 3531 +Test: [ 40/269] eta: 0:00:50 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.5076 (0.5858) acc1: 79.1667 (69.9060) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1111 data: 0.0190 max mem: 3531 +Test: [ 50/269] eta: 0:00:47 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.5300 (0.5895) acc1: 75.7812 (69.5721) acc5: 100.0000 (100.0000) acc1_neg: 49.479 acc5_neg: 100.000 time: 0.1652 data: 0.0725 max mem: 3531 +Test: [ 60/269] eta: 0:00:43 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.6075 (0.6045) acc1: 66.9271 (68.0243) acc5: 100.0000 (100.0000) acc1_neg: 51.562 acc5_neg: 100.000 time: 0.1888 data: 0.0960 max mem: 3531 +Test: [ 70/269] eta: 0:00:40 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.6309 (0.6058) acc1: 63.0208 (67.8404) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.1745 data: 0.0790 max mem: 3531 +Test: [ 80/269] eta: 0:00:38 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.6103 (0.6053) acc1: 64.5833 (67.9495) acc5: 100.0000 (100.0000) acc1_neg: 62.240 acc5_neg: 100.000 time: 0.1818 data: 0.0882 max mem: 3531 +Test: [ 90/269] eta: 0:00:36 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.6532 (0.6188) acc1: 60.1562 (66.3948) acc5: 100.0000 (100.0000) acc1_neg: 66.146 acc5_neg: 100.000 time: 0.2057 data: 0.1151 max mem: 3531 +Test: [100/269] eta: 0:00:34 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.7016 (0.6239) acc1: 55.4688 (65.7797) acc5: 100.0000 (100.0000) acc1_neg: 48.438 acc5_neg: 100.000 time: 0.2233 data: 0.1324 max mem: 3531 +Test: [110/269] eta: 0:00:32 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.7012 (0.6327) acc1: 56.2500 (64.6068) acc5: 100.0000 (100.0000) acc1_neg: 40.625 acc5_neg: 100.000 time: 0.2191 data: 0.1272 max mem: 3531 +Test: [120/269] eta: 0:00:30 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.6582 (0.6293) acc1: 58.5938 (64.9148) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.2048 data: 0.1132 max mem: 3531 +Test: [130/269] eta: 0:00:28 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.6386 (0.6362) acc1: 60.9375 (64.1619) acc5: 100.0000 (100.0000) acc1_neg: 54.948 acc5_neg: 100.000 time: 0.1912 data: 0.0995 max mem: 3531 +Test: [140/269] eta: 0:00:26 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.7328 (0.6447) acc1: 52.8646 (63.2406) acc5: 100.0000 (100.0000) acc1_neg: 45.573 acc5_neg: 100.000 time: 0.1956 data: 0.1015 max mem: 3531 +Test: [150/269] eta: 0:00:24 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.7306 (0.6463) acc1: 53.1250 (63.0001) acc5: 100.0000 (100.0000) acc1_neg: 65.104 acc5_neg: 100.000 time: 0.2036 data: 0.1100 max mem: 3531 +Test: [160/269] eta: 0:00:22 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.6746 (0.6475) acc1: 57.0312 (62.8397) acc5: 100.0000 (100.0000) acc1_neg: 58.854 acc5_neg: 100.000 time: 0.2141 data: 0.1207 max mem: 3531 +Test: [170/269] eta: 0:00:20 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.7321 (0.6529) acc1: 56.5104 (62.3051) acc5: 100.0000 (100.0000) acc1_neg: 59.635 acc5_neg: 100.000 time: 0.2127 data: 0.1162 max mem: 3531 +Test: [180/269] eta: 0:00:18 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.6458 (0.6499) acc1: 59.1146 (62.5201) acc5: 100.0000 (100.0000) acc1_neg: 59.115 acc5_neg: 100.000 time: 0.2100 data: 0.1123 max mem: 3531 +Test: [190/269] eta: 0:00:16 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.5824 (0.6487) acc1: 66.9271 (62.5736) acc5: 100.0000 (100.0000) acc1_neg: 64.844 acc5_neg: 100.000 time: 0.2121 data: 0.1168 max mem: 3531 +Test: [200/269] eta: 0:00:14 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.6129 (0.6475) acc1: 64.8438 (62.7241) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.1948 data: 0.1009 max mem: 3531 +Test: [210/269] eta: 0:00:12 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.5931 (0.6444) acc1: 66.9271 (62.9653) acc5: 100.0000 (100.0000) acc1_neg: 68.490 acc5_neg: 100.000 time: 0.2121 data: 0.1183 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.5566 (0.6421) acc1: 68.4896 (63.1764) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.2281 data: 0.1362 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.5069 (0.6367) acc1: 72.9167 (63.6476) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.1943 data: 0.1004 max mem: 3532 +Test: [240/269] eta: 0:00:05 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.6365 (0.6397) acc1: 63.8021 (63.3623) acc5: 100.0000 (100.0000) acc1_neg: 63.802 acc5_neg: 100.000 time: 0.1919 data: 0.0977 max mem: 3532 +Test: [250/269] eta: 0:00:03 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.6849 (0.6431) acc1: 57.5521 (63.0385) acc5: 100.0000 (100.0000) acc1_neg: 38.802 acc5_neg: 100.000 time: 0.1803 data: 0.0892 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.7465 (0.6495) acc1: 45.0521 (62.4351) acc5: 100.0000 (100.0000) acc1_neg: 62.240 acc5_neg: 100.000 time: 0.1386 data: 0.0481 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.6895 (0.6489) acc1: 52.0833 (62.5154) acc5: 100.0000 (100.0000) acc1_neg: 77.690 acc5_neg: 100.000 time: 0.1395 data: 0.0370 max mem: 3532 +Test: Total time: 0:00:53 (0.2001 s / it) +* Overall Acc@1 62.515 Acc@5 100.000 + target == real Acc@1 65.656 Acc@5 100.000 + target == fake Acc@1 62.087 Acc@5 100.000 + loss 0.649 + PR-AUC 0.9362 ROC-AUC 0.6909 EER 36.27% (thr=0.4902) +Accuracy of the network on the 103293 test images: 62.5% +[rank0]:[W726 04:49:28.823043273 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T04:49:35Z] eval_done tag=fixed500k_real_n32_seed7_resampled +[2026-07-26T04:49:36Z] eval_start tag=fixed500k_fake_n64_seed7_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed7_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T04:49:36Z +hostname=n116-040-213 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed7_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n64_seed7_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 04:49:49.375000 1191800 torch/distributed/run.py:766] +W0726 04:49:49.375000 1191800 torch/distributed/run.py:766] ***************************************** +W0726 04:49:49.375000 1191800 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:49:49.375000 1191800 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:50:17.548662994 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:50:17.963369433 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:50:18.588672405 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:50:18.671761367 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:50:18.685218513 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:50:18.745584080 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W726 04:50:18.746328658 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:50:18.802065428 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n64_seed7_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed7_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:25:02 acc1_pos: 35.156 acc5_pos: 100.000 loss: 0.8585 (0.8585) acc1: 35.1562 (35.1562) acc5: 100.0000 (100.0000) time: 6.4228 data: 4.7371 max mem: 3531 +Test: [ 10/234] eta: 0:02:30 acc1_pos: 16.667 acc5_pos: 100.000 loss: 0.8864 (0.8891) acc1: 36.9792 (36.7661) acc5: 100.0000 (100.0000) time: 0.6709 data: 0.4310 max mem: 3531 +Test: [ 20/234] eta: 0:01:25 acc1_pos: 95.833 acc5_pos: 100.000 loss: 0.9328 (0.8607) acc1: 33.3333 (39.6701) acc5: 100.0000 (100.0000) time: 0.0963 data: 0.0005 max mem: 3531 +Test: [ 30/234] eta: 0:01:09 acc1_pos: 56.250 acc5_pos: 100.000 loss: 0.2314 (0.6775) acc1: 94.5312 (55.4856) acc5: 100.0000 (100.0000) time: 0.1583 data: 0.0607 max mem: 3531 +Test: [ 40/234] eta: 0:01:02 acc1_pos: 99.219 acc5_pos: 100.000 loss: 0.2299 (0.6415) acc1: 95.0521 (58.5112) acc5: 100.0000 (100.0000) time: 0.2447 data: 0.1468 max mem: 3531 +Test: [ 50/234] eta: 0:00:56 acc1_pos: 97.917 acc5_pos: 100.000 loss: 0.1432 (0.5428) acc1: 97.9167 (66.3246) acc5: 100.0000 (100.0000) time: 0.2599 data: 0.1628 max mem: 3531 +Test: [ 60/234] eta: 0:00:48 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.1136 (0.4692) acc1: 98.9583 (71.8280) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1863 data: 0.0893 max mem: 3531 +Test: [ 70/234] eta: 0:00:44 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.0949 (0.4887) acc1: 100.0000 (71.1708) acc5: 100.0000 (100.0000) acc1_neg: 48.958 acc5_neg: 100.000 time: 0.1771 data: 0.0814 max mem: 3531 +Test: [ 80/234] eta: 0:00:40 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.8288 (0.5035) acc1: 52.6042 (70.7658) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2256 data: 0.1322 max mem: 3531 +Test: [ 90/234] eta: 0:00:37 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.1825 (0.4673) acc1: 92.9688 (73.5005) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.2150 data: 0.1219 max mem: 3531 +Test: [100/234] eta: 0:00:33 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.2107 (0.4615) acc1: 92.9688 (74.2316) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2034 data: 0.1113 max mem: 3531 +Test: [110/234] eta: 0:00:30 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.3897 (0.4504) acc1: 80.4688 (75.3214) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.2029 data: 0.1127 max mem: 3531 +Test: [120/234] eta: 0:00:27 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.2936 (0.4500) acc1: 88.2812 (75.8308) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1951 data: 0.1049 max mem: 3531 +Test: [130/234] eta: 0:00:25 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.2936 (0.4418) acc1: 93.4896 (76.9402) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.1923 data: 0.1020 max mem: 3531 +Test: [140/234] eta: 0:00:22 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.3559 (0.4653) acc1: 85.1562 (76.0047) acc5: 100.0000 (100.0000) acc1_neg: 32.031 acc5_neg: 100.000 time: 0.1994 data: 0.1083 max mem: 3531 +Test: [150/234] eta: 0:00:19 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4506 (0.4702) acc1: 74.2188 (75.8916) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.1878 data: 0.0955 max mem: 3531 +Test: [160/234] eta: 0:00:17 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.3737 (0.4606) acc1: 87.7604 (76.8585) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.1847 data: 0.0904 max mem: 3531 +Test: [170/234] eta: 0:00:14 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.3737 (0.4556) acc1: 89.5833 (77.5356) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1804 data: 0.0866 max mem: 3531 +Test: [180/234] eta: 0:00:12 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.3924 (0.4686) acc1: 84.8958 (76.8186) acc5: 100.0000 (100.0000) acc1_neg: 25.521 acc5_neg: 100.000 time: 0.1949 data: 0.1028 max mem: 3531 +Test: [190/234] eta: 0:00:09 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4840 (0.4713) acc1: 76.3021 (76.7316) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.2138 data: 0.1182 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4653 (0.4599) acc1: 77.8646 (77.4681) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1864 data: 0.0898 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.1004 (0.4451) acc1: 99.7396 (78.4274) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1544 data: 0.0618 max mem: 3532 +Test: [220/234] eta: 0:00:03 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.2260 (0.4362) acc1: 94.2708 (79.0618) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.1367 data: 0.0464 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.3628 (0.4468) acc1: 60.4167 (78.1385) acc5: 100.0000 (100.0000) acc1_neg: 59.115 acc5_neg: 100.000 time: 0.1296 data: 0.0396 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.6569 (0.4495) acc1: 59.1146 (77.9090) acc5: 100.0000 (100.0000) acc1_neg: 61.868 acc5_neg: 100.000 time: 0.1322 data: 0.0395 max mem: 3532 +Test: Total time: 0:00:49 (0.2105 s / it) +* Overall Acc@1 77.909 Acc@5 100.000 + target == real Acc@1 66.557 Acc@5 100.000 + target == fake Acc@1 81.109 Acc@5 100.000 + loss 0.449 + PR-AUC 0.9569 ROC-AUC 0.8582 EER 24.61% (thr=0.6021) +Accuracy of the network on the 89729 test images: 77.9% +[rank0]:[W726 04:51:18.441104014 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:51:39.245000 1192359 torch/distributed/run.py:766] +W0726 04:51:39.245000 1192359 torch/distributed/run.py:766] ***************************************** +W0726 04:51:39.245000 1192359 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:51:39.245000 1192359 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 2): env:// +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:52:09.903960158 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:52:09.063681889 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:52:10.179051238 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:52:10.201295302 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:52:10.329652798 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +[rank0]:[W726 04:52:10.479219114 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:52:10.483867320 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:52:10.501496795 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n64_seed7_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed7_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 3:50:45 acc1_pos: 6.771 acc5_pos: 100.000 loss: 1.3141 (1.3141) acc1: 6.7708 (6.7708) acc5: 100.0000 (100.0000) time: 5.8002 data: 3.0284 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/2387] eta: 0:24:13 acc1_pos: 0.521 acc5_pos: 100.000 loss: 1.3795 (1.3538) acc1: 6.5104 (7.2443) acc5: 100.0000 (100.0000) time: 0.6117 data: 0.2757 max mem: 3531 +Test: [ 20/2387] eta: 0:14:21 acc1_pos: 1.823 acc5_pos: 100.000 loss: 1.3795 (1.3700) acc1: 6.5104 (6.9072) acc5: 100.0000 (100.0000) time: 0.0919 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:11:41 acc1_pos: 0.521 acc5_pos: 100.000 loss: 1.4534 (1.3844) acc1: 6.5104 (6.7792) acc5: 100.0000 (100.0000) time: 0.1252 data: 0.0343 max mem: 3531 +Test: [ 40/2387] eta: 0:11:43 acc1_pos: 1.042 acc5_pos: 100.000 loss: 1.4239 (1.3972) acc1: 1.3021 (5.7990) acc5: 100.0000 (100.0000) time: 0.2329 data: 0.1402 max mem: 3531 +Test: [ 50/2387] eta: 0:11:36 acc1_pos: 24.740 acc5_pos: 100.000 loss: 1.3928 (1.3788) acc1: 4.4271 (6.9291) acc5: 100.0000 (100.0000) time: 0.2983 data: 0.2052 max mem: 3531 +Test: [ 60/2387] eta: 0:12:08 acc1_pos: 16.146 acc5_pos: 100.000 loss: 1.3111 (1.3681) acc1: 7.0312 (7.4539) acc5: 100.0000 (100.0000) time: 0.3399 data: 0.2456 max mem: 3531 +Test: [ 70/2387] eta: 0:10:59 acc1_pos: 18.229 acc5_pos: 100.000 loss: 1.3405 (1.3773) acc1: 7.0312 (7.4090) acc5: 100.0000 (100.0000) time: 0.2498 data: 0.1546 max mem: 3531 +Test: [ 80/2387] eta: 0:10:43 acc1_pos: 35.938 acc5_pos: 100.000 loss: 1.3263 (1.3499) acc1: 8.8542 (8.4491) acc5: 100.0000 (100.0000) time: 0.1749 data: 0.0811 max mem: 3531 +Test: [ 90/2387] eta: 0:10:20 acc1_pos: 30.208 acc5_pos: 100.000 loss: 1.0690 (1.3163) acc1: 19.7917 (10.1591) acc5: 100.0000 (100.0000) time: 0.2198 data: 0.1256 max mem: 3531 +Test: [ 100/2387] eta: 0:10:05 acc1_pos: 11.979 acc5_pos: 100.000 loss: 1.0491 (1.3004) acc1: 17.7083 (10.5920) acc5: 100.0000 (100.0000) time: 0.2064 data: 0.1098 max mem: 3531 +Test: [ 110/2387] eta: 0:09:50 acc1_pos: 22.656 acc5_pos: 100.000 loss: 1.0826 (1.2791) acc1: 13.8021 (11.2871) acc5: 100.0000 (100.0000) time: 0.2099 data: 0.1137 max mem: 3531 +Test: [ 120/2387] eta: 0:09:40 acc1_pos: 19.792 acc5_pos: 100.000 loss: 1.0879 (1.2659) acc1: 17.7083 (11.9792) acc5: 100.0000 (100.0000) time: 0.2134 data: 0.1181 max mem: 3531 +Test: [ 130/2387] eta: 0:09:35 acc1_pos: 19.010 acc5_pos: 100.000 loss: 1.1047 (1.2542) acc1: 17.7083 (12.2694) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1324 max mem: 3531 +Test: [ 140/2387] eta: 0:09:28 acc1_pos: 15.365 acc5_pos: 100.000 loss: 1.0966 (1.2439) acc1: 16.1458 (12.6237) acc5: 100.0000 (100.0000) time: 0.2349 data: 0.1362 max mem: 3531 +Test: [ 150/2387] eta: 0:09:22 acc1_pos: 4.427 acc5_pos: 100.000 loss: 1.0612 (1.2307) acc1: 17.7083 (13.0864) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1307 max mem: 3531 +Test: [ 160/2387] eta: 0:09:14 acc1_pos: 20.091 acc5_pos: 100.000 loss: 1.0029 (1.2000) acc1: 22.6562 (15.3258) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2199 data: 0.1229 max mem: 3531 +Test: [ 170/2387] eta: 0:09:00 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1864 (1.1403) acc1: 98.9583 (20.2470) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1852 data: 0.0891 max mem: 3531 +Test: [ 180/2387] eta: 0:08:50 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1784 (1.0871) acc1: 99.4792 (24.6259) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1718 data: 0.0766 max mem: 3531 +Test: [ 190/2387] eta: 0:08:40 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1781 (1.0395) acc1: 99.4792 (28.5422) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1795 data: 0.0864 max mem: 3531 +Test: [ 200/2387] eta: 0:08:28 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1737 (0.9965) acc1: 99.4792 (32.0727) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1629 data: 0.0694 max mem: 3531 +Test: [ 210/2387] eta: 0:08:23 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1772 (0.9578) acc1: 99.4792 (35.2624) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1749 data: 0.0785 max mem: 3532 +Test: [ 220/2387] eta: 0:08:15 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1779 (0.9225) acc1: 99.4792 (38.1693) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1899 data: 0.0914 max mem: 3532 +Test: [ 230/2387] eta: 0:08:10 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1784 (0.8904) acc1: 99.4792 (40.8234) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1849 data: 0.0864 max mem: 3532 +Test: [ 240/2387] eta: 0:08:05 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1843 (0.8630) acc1: 99.2188 (43.2065) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1967 data: 0.0989 max mem: 3532 +Test: [ 250/2387] eta: 0:07:59 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1741 (0.8327) acc1: 98.9583 (45.4598) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0947 max mem: 3532 +Test: [ 260/2387] eta: 0:07:55 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.0811 (0.8039) acc1: 100.0000 (47.5495) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1931 data: 0.0983 max mem: 3532 +Test: [ 270/2387] eta: 0:07:49 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.0787 (0.7772) acc1: 100.0000 (49.4840) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1904 data: 0.0935 max mem: 3532 +Test: [ 280/2387] eta: 0:07:43 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.0793 (0.7525) acc1: 100.0000 (51.2817) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1715 data: 0.0743 max mem: 3532 +Test: [ 290/2387] eta: 0:07:39 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.0803 (0.7294) acc1: 100.0000 (52.9559) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1857 data: 0.0902 max mem: 3532 +Test: [ 300/2387] eta: 0:07:33 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.0812 (0.7080) acc1: 100.0000 (54.5188) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1833 data: 0.0886 max mem: 3532 +Test: [ 310/2387] eta: 0:07:30 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.0827 (0.6914) acc1: 100.0000 (55.9377) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1853 data: 0.0890 max mem: 3532 +Test: [ 320/2387] eta: 0:07:25 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.3163 (0.6799) acc1: 97.6562 (57.2227) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1849 data: 0.0890 max mem: 3532 +Test: [ 330/2387] eta: 0:07:20 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.3237 (0.6690) acc1: 97.6562 (58.4569) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1704 data: 0.0745 max mem: 3532 +Test: [ 340/2387] eta: 0:07:17 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.3190 (0.6588) acc1: 97.6562 (59.6011) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1819 data: 0.0877 max mem: 3532 +Test: [ 350/2387] eta: 0:07:15 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.3140 (0.6449) acc1: 98.1771 (60.7431) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2026 data: 0.1114 max mem: 3532 +Test: [ 360/2387] eta: 0:07:09 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1505 (0.6313) acc1: 100.0000 (61.8277) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1878 data: 0.0926 max mem: 3532 +Test: [ 370/2387] eta: 0:07:07 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1550 (0.6185) acc1: 99.7396 (62.8496) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0795 max mem: 3532 +Test: [ 380/2387] eta: 0:07:03 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1561 (0.6063) acc1: 99.7396 (63.8178) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1931 data: 0.0948 max mem: 3532 +Test: [ 390/2387] eta: 0:07:00 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1546 (0.5948) acc1: 99.7396 (64.7379) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1849 data: 0.0903 max mem: 3532 +Test: [ 400/2387] eta: 0:06:57 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1546 (0.5838) acc1: 99.7396 (65.6133) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1966 data: 0.1023 max mem: 3532 +Test: [ 410/2387] eta: 0:06:56 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1531 (0.5734) acc1: 99.7396 (66.4424) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2141 data: 0.1200 max mem: 3532 +Test: [ 420/2387] eta: 0:06:52 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1519 (0.5634) acc1: 99.7396 (67.2370) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1999 data: 0.1069 max mem: 3532 +Test: [ 430/2387] eta: 0:06:47 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1548 (0.5545) acc1: 99.7396 (67.9845) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1657 data: 0.0703 max mem: 3532 +Test: [ 440/2387] eta: 0:06:46 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1573 (0.5452) acc1: 99.4792 (68.7016) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1816 data: 0.0849 max mem: 3533 +Test: [ 450/2387] eta: 0:06:43 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1039 (0.5354) acc1: 100.0000 (69.3956) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2067 data: 0.1091 max mem: 3533 +Test: [ 460/2387] eta: 0:06:41 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1032 (0.5260) acc1: 100.0000 (70.0594) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2049 data: 0.1089 max mem: 3533 +Test: [ 470/2387] eta: 0:06:37 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1032 (0.5170) acc1: 100.0000 (70.6951) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1882 data: 0.0969 max mem: 3533 +Test: [ 480/2387] eta: 0:06:33 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1038 (0.5084) acc1: 100.0000 (71.3044) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1645 data: 0.0727 max mem: 3533 +Test: [ 490/2387] eta: 0:06:31 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1038 (0.5002) acc1: 100.0000 (71.8888) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1803 data: 0.0865 max mem: 3533 +Test: [ 500/2387] eta: 0:06:28 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1045 (0.4923) acc1: 100.0000 (72.4494) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1836 data: 0.0876 max mem: 3533 +Test: [ 510/2387] eta: 0:06:24 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1046 (0.4850) acc1: 100.0000 (72.9885) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1646 data: 0.0694 max mem: 3533 +Test: [ 520/2387] eta: 0:06:20 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1275 (0.4784) acc1: 100.0000 (73.5070) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1635 data: 0.0689 max mem: 3533 +Test: [ 530/2387] eta: 0:06:17 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1262 (0.4717) acc1: 100.0000 (74.0059) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1697 data: 0.0752 max mem: 3533 +Test: [ 540/2387] eta: 0:06:14 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1249 (0.4653) acc1: 100.0000 (74.4864) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1642 data: 0.0695 max mem: 3533 +Test: [ 550/2387] eta: 0:06:11 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1250 (0.4592) acc1: 100.0000 (74.9494) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1673 data: 0.0718 max mem: 3533 +Test: [ 560/2387] eta: 0:06:08 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1254 (0.4532) acc1: 100.0000 (75.3960) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1728 data: 0.0795 max mem: 3533 +Test: [ 570/2387] eta: 0:06:05 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1254 (0.4475) acc1: 100.0000 (75.8269) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1675 data: 0.0736 max mem: 3533 +Test: [ 580/2387] eta: 0:06:02 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1245 (0.4419) acc1: 100.0000 (76.2429) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1741 data: 0.0754 max mem: 3533 +Test: [ 590/2387] eta: 0:06:00 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1249 (0.4366) acc1: 100.0000 (76.6449) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1860 data: 0.0891 max mem: 3533 +Test: [ 600/2387] eta: 0:05:58 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1285 (0.4318) acc1: 100.0000 (77.0335) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1970 data: 0.1053 max mem: 3533 +Test: [ 610/2387] eta: 0:05:57 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1422 (0.4270) acc1: 100.0000 (77.4090) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2159 data: 0.1248 max mem: 3533 +Test: [ 620/2387] eta: 0:05:55 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1359 (0.4223) acc1: 100.0000 (77.7723) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2239 data: 0.1331 max mem: 3533 +Test: [ 630/2387] eta: 0:05:54 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1362 (0.4178) acc1: 100.0000 (78.1242) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2235 data: 0.1312 max mem: 3533 +Test: [ 640/2387] eta: 0:05:51 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1375 (0.4134) acc1: 100.0000 (78.4650) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1999 data: 0.1070 max mem: 3533 +Test: [ 650/2387] eta: 0:05:49 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1372 (0.4092) acc1: 100.0000 (78.7958) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1844 data: 0.0930 max mem: 3533 +Test: [ 660/2387] eta: 0:05:47 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1373 (0.4051) acc1: 100.0000 (79.1166) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2054 data: 0.1147 max mem: 3533 +Test: [ 670/2387] eta: 0:05:45 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1371 (0.4011) acc1: 100.0000 (79.4279) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2010 data: 0.1105 max mem: 3534 +Test: [ 680/2387] eta: 0:05:43 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1388 (0.3975) acc1: 100.0000 (79.7299) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2055 data: 0.1147 max mem: 3534 +Test: [ 690/2387] eta: 0:05:41 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1720 (0.3959) acc1: 100.0000 (79.8869) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.2085 data: 0.1160 max mem: 3534 +Test: [ 700/2387] eta: 0:05:39 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.3212 (0.3948) acc1: 86.9792 (79.9932) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1988 data: 0.1061 max mem: 3534 +Test: [ 710/2387] eta: 0:05:37 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.3181 (0.3938) acc1: 87.2396 (80.0966) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.2067 data: 0.1149 max mem: 3534 +Test: [ 720/2387] eta: 0:05:35 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.3146 (0.3927) acc1: 87.2396 (80.2007) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1936 data: 0.1005 max mem: 3534 +Test: [ 730/2387] eta: 0:05:33 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.3203 (0.3918) acc1: 87.2396 (80.2899) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.1944 data: 0.1011 max mem: 3534 +Test: [ 740/2387] eta: 0:05:31 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.3228 (0.3908) acc1: 86.9792 (80.3844) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.1959 data: 0.1045 max mem: 3534 +Test: [ 750/2387] eta: 0:05:29 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.3162 (0.3899) acc1: 86.9792 (80.4816) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.2001 data: 0.1089 max mem: 3534 +Test: [ 760/2387] eta: 0:05:26 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.3149 (0.3888) acc1: 87.7604 (80.5933) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1794 data: 0.0881 max mem: 3534 +Test: [ 770/2387] eta: 0:05:23 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.2791 (0.3867) acc1: 91.6667 (80.7812) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1547 data: 0.0615 max mem: 3534 +Test: [ 780/2387] eta: 0:05:21 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1859 (0.3839) acc1: 98.9583 (81.0156) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1744 data: 0.0806 max mem: 3534 +Test: [ 790/2387] eta: 0:05:18 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1521 (0.3812) acc1: 99.4792 (81.2144) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1815 data: 0.0873 max mem: 3534 +Test: [ 800/2387] eta: 0:05:16 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1310 (0.3782) acc1: 99.7396 (81.4457) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1908 data: 0.0970 max mem: 3534 +Test: [ 810/2387] eta: 0:05:14 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1522 (0.3755) acc1: 100.0000 (81.6655) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1971 data: 0.1047 max mem: 3534 +Test: [ 820/2387] eta: 0:05:12 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1586 (0.3730) acc1: 99.7396 (81.8736) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1918 data: 0.0997 max mem: 3534 +Test: [ 830/2387] eta: 0:05:10 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.2065 (0.3717) acc1: 97.9167 (82.0278) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1823 data: 0.0901 max mem: 3534 +Test: [ 840/2387] eta: 0:05:07 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.2142 (0.3696) acc1: 97.9167 (82.2276) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1776 data: 0.0833 max mem: 3534 +Test: [ 850/2387] eta: 0:05:05 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1847 (0.3673) acc1: 99.4792 (82.4251) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1739 data: 0.0806 max mem: 3534 +Test: [ 860/2387] eta: 0:05:03 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1700 (0.3650) acc1: 99.7396 (82.6244) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1859 data: 0.0946 max mem: 3534 +Test: [ 870/2387] eta: 0:05:00 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1645 (0.3628) acc1: 99.7396 (82.8206) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1836 data: 0.0925 max mem: 3534 +Test: [ 880/2387] eta: 0:04:58 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1656 (0.3608) acc1: 99.7396 (83.0079) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1805 data: 0.0868 max mem: 3534 +Test: [ 890/2387] eta: 0:04:56 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1760 (0.3587) acc1: 99.7396 (83.1948) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1999 data: 0.1040 max mem: 3534 +Test: [ 900/2387] eta: 0:04:54 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.2061 (0.3628) acc1: 97.1354 (82.9076) acc5: 100.0000 (100.0000) acc1_neg: 51.042 acc5_neg: 100.000 time: 0.2067 data: 0.1106 max mem: 3535 +Test: [ 910/2387] eta: 0:04:53 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.8525 (0.3682) acc1: 46.8750 (82.5132) acc5: 100.0000 (100.0000) acc1_neg: 44.531 acc5_neg: 100.000 time: 0.2145 data: 0.1175 max mem: 3535 +Test: [ 920/2387] eta: 0:04:50 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.8582 (0.3735) acc1: 46.8750 (82.1257) acc5: 100.0000 (100.0000) acc1_neg: 42.708 acc5_neg: 100.000 time: 0.1986 data: 0.1026 max mem: 3535 +Test: [ 930/2387] eta: 0:04:49 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.8556 (0.3786) acc1: 47.6562 (81.7627) acc5: 100.0000 (100.0000) acc1_neg: 43.490 acc5_neg: 100.000 time: 0.2000 data: 0.1072 max mem: 3535 +Test: [ 940/2387] eta: 0:04:47 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.8556 (0.3837) acc1: 47.6562 (81.4022) acc5: 100.0000 (100.0000) acc1_neg: 47.396 acc5_neg: 100.000 time: 0.2130 data: 0.1215 max mem: 3535 +Test: [ 950/2387] eta: 0:04:45 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.8511 (0.3886) acc1: 46.3542 (81.0252) acc5: 100.0000 (100.0000) acc1_neg: 48.177 acc5_neg: 100.000 time: 0.2030 data: 0.1108 max mem: 3535 +Test: [ 960/2387] eta: 0:04:43 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.8391 (0.3933) acc1: 46.6146 (80.6755) acc5: 100.0000 (100.0000) acc1_neg: 50.521 acc5_neg: 100.000 time: 0.2095 data: 0.1149 max mem: 3535 +Test: [ 970/2387] eta: 0:04:41 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.8426 (0.3980) acc1: 46.3542 (80.3186) acc5: 100.0000 (100.0000) acc1_neg: 47.917 acc5_neg: 100.000 time: 0.2080 data: 0.1134 max mem: 3535 +Test: [ 980/2387] eta: 0:04:39 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.8194 (0.3981) acc1: 47.9167 (80.3180) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.1968 data: 0.1022 max mem: 3535 +Test: [ 990/2387] eta: 0:04:37 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1895 (0.3962) acc1: 98.1771 (80.4892) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1969 data: 0.1013 max mem: 3535 +Test: [1000/2387] eta: 0:04:35 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1926 (0.3942) acc1: 98.9583 (80.6605) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1939 data: 0.1012 max mem: 3535 +Test: [1010/2387] eta: 0:04:33 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1995 (0.3928) acc1: 98.1771 (80.8106) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1919 data: 0.1012 max mem: 3535 +Test: [1020/2387] eta: 0:04:31 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.2301 (0.3912) acc1: 98.6979 (80.9605) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1865 data: 0.0945 max mem: 3535 +Test: [1030/2387] eta: 0:04:29 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1700 (0.3890) acc1: 99.2188 (81.1386) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1913 data: 0.0985 max mem: 3535 +Test: [1040/2387] eta: 0:04:27 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1440 (0.3866) acc1: 100.0000 (81.3188) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1879 data: 0.0954 max mem: 3535 +Test: [1050/2387] eta: 0:04:25 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1370 (0.3842) acc1: 100.0000 (81.4941) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1832 data: 0.0911 max mem: 3535 +Test: [1060/2387] eta: 0:04:23 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1395 (0.3820) acc1: 100.0000 (81.6655) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2041 data: 0.1117 max mem: 3535 +Test: [1070/2387] eta: 0:04:20 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1506 (0.3801) acc1: 99.7396 (81.8236) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1903 data: 0.0971 max mem: 3535 +Test: [1080/2387] eta: 0:04:19 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.2023 (0.3787) acc1: 98.6979 (81.9717) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2147 data: 0.1221 max mem: 3535 +Test: [1090/2387] eta: 0:04:17 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1981 (0.3769) acc1: 98.6979 (82.1286) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2065 data: 0.1146 max mem: 3535 +Test: [1100/2387] eta: 0:04:15 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1730 (0.3750) acc1: 99.7396 (82.2862) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1706 data: 0.0752 max mem: 3535 +Test: [1110/2387] eta: 0:04:13 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1647 (0.3734) acc1: 99.7396 (82.4131) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1977 data: 0.1017 max mem: 3535 +Test: [1120/2387] eta: 0:04:11 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1692 (0.3716) acc1: 99.2188 (82.5595) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2006 data: 0.1088 max mem: 3536 +Test: [1130/2387] eta: 0:04:09 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1641 (0.3698) acc1: 99.4792 (82.7114) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2052 data: 0.1130 max mem: 3536 +Test: [1140/2387] eta: 0:04:07 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1970 (0.3685) acc1: 99.7396 (82.8490) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1989 data: 0.1068 max mem: 3536 +Test: [1150/2387] eta: 0:04:05 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1982 (0.3670) acc1: 99.4792 (82.9856) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2102 data: 0.1187 max mem: 3536 +Test: [1160/2387] eta: 0:04:03 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1448 (0.3650) acc1: 100.0000 (83.1285) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1919 data: 0.0988 max mem: 3536 +Test: [1170/2387] eta: 0:04:01 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1343 (0.3630) acc1: 100.0000 (83.2722) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1706 data: 0.0764 max mem: 3536 +Test: [1180/2387] eta: 0:03:59 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1309 (0.3611) acc1: 100.0000 (83.4114) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1992 data: 0.1056 max mem: 3536 +Test: [1190/2387] eta: 0:03:57 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1313 (0.3593) acc1: 99.7396 (83.5498) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2000 data: 0.1076 max mem: 3536 +Test: [1200/2387] eta: 0:03:54 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1283 (0.3574) acc1: 100.0000 (83.6868) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1794 data: 0.0876 max mem: 3536 +Test: [1210/2387] eta: 0:03:52 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1263 (0.3555) acc1: 100.0000 (83.8215) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1834 data: 0.0902 max mem: 3536 +Test: [1220/2387] eta: 0:03:50 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.0882 (0.3532) acc1: 100.0000 (83.9533) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1967 data: 0.0992 max mem: 3536 +Test: [1230/2387] eta: 0:03:48 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.0774 (0.3510) acc1: 100.0000 (84.0837) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.0991 max mem: 3536 +Test: [1240/2387] eta: 0:03:46 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.0757 (0.3487) acc1: 100.0000 (84.2117) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2032 data: 0.1115 max mem: 3536 +Test: [1250/2387] eta: 0:03:45 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.0762 (0.3466) acc1: 100.0000 (84.3377) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2051 data: 0.1141 max mem: 3536 +Test: [1260/2387] eta: 0:03:43 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.0762 (0.3444) acc1: 100.0000 (84.4615) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2061 data: 0.1154 max mem: 3536 +Test: [1270/2387] eta: 0:03:41 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.0779 (0.3424) acc1: 100.0000 (84.5838) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2053 data: 0.1135 max mem: 3536 +Test: [1280/2387] eta: 0:03:39 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.0925 (0.3405) acc1: 100.0000 (84.7033) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2031 data: 0.1111 max mem: 3536 +Test: [1290/2387] eta: 0:03:37 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1061 (0.3388) acc1: 100.0000 (84.8115) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2074 data: 0.1160 max mem: 3536 +Test: [1300/2387] eta: 0:03:35 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1047 (0.3370) acc1: 100.0000 (84.9243) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2241 data: 0.1327 max mem: 3536 +Test: [1310/2387] eta: 0:03:33 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1074 (0.3354) acc1: 100.0000 (85.0343) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2079 data: 0.1134 max mem: 3536 +Test: [1320/2387] eta: 0:03:32 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1156 (0.3338) acc1: 100.0000 (85.1401) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2113 data: 0.1168 max mem: 3536 +Test: [1330/2387] eta: 0:03:29 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1325 (0.3324) acc1: 100.0000 (85.2445) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2100 data: 0.1165 max mem: 3536 +Test: [1340/2387] eta: 0:03:27 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1675 (0.3319) acc1: 98.9583 (85.2897) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1885 data: 0.0950 max mem: 3536 +Test: [1350/2387] eta: 0:03:25 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.0666 (0.3299) acc1: 100.0000 (85.3985) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1862 data: 0.0941 max mem: 3537 +Test: [1360/2387] eta: 0:03:23 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.0610 (0.3279) acc1: 100.0000 (85.5058) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1716 data: 0.0795 max mem: 3537 +Test: [1370/2387] eta: 0:03:21 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.0586 (0.3260) acc1: 100.0000 (85.6116) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1652 data: 0.0735 max mem: 3537 +Test: [1380/2387] eta: 0:03:19 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.0586 (0.3241) acc1: 100.0000 (85.7157) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1668 data: 0.0747 max mem: 3537 +Test: [1390/2387] eta: 0:03:17 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.0603 (0.3222) acc1: 100.0000 (85.8184) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1789 data: 0.0862 max mem: 3537 +Test: [1400/2387] eta: 0:03:15 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.0583 (0.3203) acc1: 100.0000 (85.9197) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1918 data: 0.0990 max mem: 3537 +Test: [1410/2387] eta: 0:03:13 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.0606 (0.3189) acc1: 100.0000 (86.0152) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2064 data: 0.1148 max mem: 3537 +Test: [1420/2387] eta: 0:03:11 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1412 (0.3179) acc1: 100.0000 (86.0951) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2115 data: 0.1202 max mem: 3537 +Test: [1430/2387] eta: 0:03:09 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1475 (0.3167) acc1: 100.0000 (86.1894) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2017 data: 0.1097 max mem: 3537 +Test: [1440/2387] eta: 0:03:07 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1449 (0.3156) acc1: 100.0000 (86.2825) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1962 data: 0.1034 max mem: 3537 +Test: [1450/2387] eta: 0:03:05 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1569 (0.3147) acc1: 99.7396 (86.3699) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1976 data: 0.1056 max mem: 3537 +Test: [1460/2387] eta: 0:03:03 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1322 (0.3135) acc1: 99.2188 (86.4605) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1950 data: 0.1037 max mem: 3537 +Test: [1470/2387] eta: 0:03:01 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1107 (0.3121) acc1: 100.0000 (86.5522) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2001 data: 0.1085 max mem: 3537 +Test: [1480/2387] eta: 0:02:59 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1066 (0.3107) acc1: 100.0000 (86.6428) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.1015 max mem: 3537 +Test: [1490/2387] eta: 0:02:57 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1118 (0.3094) acc1: 100.0000 (86.7322) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1753 data: 0.0820 max mem: 3537 +Test: [1500/2387] eta: 0:02:55 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1169 (0.3081) acc1: 100.0000 (86.8196) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1686 data: 0.0768 max mem: 3537 +Test: [1510/2387] eta: 0:02:53 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1211 (0.3070) acc1: 100.0000 (86.9066) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1901 data: 0.0983 max mem: 3537 +Test: [1520/2387] eta: 0:02:51 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1548 (0.3061) acc1: 100.0000 (86.9891) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2047 data: 0.1124 max mem: 3537 +Test: [1530/2387] eta: 0:02:49 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1571 (0.3052) acc1: 99.7396 (87.0708) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1979 data: 0.1060 max mem: 3537 +Test: [1540/2387] eta: 0:02:47 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1482 (0.3042) acc1: 99.7396 (87.1532) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1950 data: 0.1034 max mem: 3537 +Test: [1550/2387] eta: 0:02:45 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1459 (0.3032) acc1: 99.7396 (87.2344) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1967 data: 0.1046 max mem: 3537 +Test: [1560/2387] eta: 0:02:43 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1488 (0.3023) acc1: 99.7396 (87.3150) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1993 data: 0.1055 max mem: 3537 +Test: [1570/2387] eta: 0:02:41 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1481 (0.3013) acc1: 100.0000 (87.3949) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1778 data: 0.0830 max mem: 3537 +Test: [1580/2387] eta: 0:02:38 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1507 (0.3004) acc1: 100.0000 (87.4733) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1579 data: 0.0650 max mem: 3538 +Test: [1590/2387] eta: 0:02:36 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1437 (0.2994) acc1: 100.0000 (87.5507) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1611 data: 0.0697 max mem: 3538 +Test: [1600/2387] eta: 0:02:34 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1275 (0.2985) acc1: 100.0000 (87.6096) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1619 data: 0.0704 max mem: 3538 +Test: [1610/2387] eta: 0:02:32 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1240 (0.2975) acc1: 100.0000 (87.6853) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1587 data: 0.0668 max mem: 3538 +Test: [1620/2387] eta: 0:02:30 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1344 (0.2965) acc1: 100.0000 (87.7572) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1573 data: 0.0649 max mem: 3538 +Test: [1630/2387] eta: 0:02:28 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1539 (0.2958) acc1: 99.7396 (87.8235) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1649 data: 0.0706 max mem: 3538 +Test: [1640/2387] eta: 0:02:26 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1918 (0.2954) acc1: 98.6979 (87.8798) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1725 data: 0.0787 max mem: 3538 +Test: [1650/2387] eta: 0:02:24 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.2001 (0.2947) acc1: 98.6979 (87.9448) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1655 data: 0.0731 max mem: 3538 +Test: [1660/2387] eta: 0:02:22 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1916 (0.2941) acc1: 99.2188 (88.0103) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1697 data: 0.0745 max mem: 3538 +Test: [1670/2387] eta: 0:02:19 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1922 (0.2935) acc1: 99.2188 (88.0754) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1664 data: 0.0695 max mem: 3538 +Test: [1680/2387] eta: 0:02:17 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1940 (0.2930) acc1: 98.9583 (88.1328) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1583 data: 0.0626 max mem: 3538 +Test: [1690/2387] eta: 0:02:15 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1976 (0.2925) acc1: 98.4375 (88.1904) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1696 data: 0.0758 max mem: 3538 +Test: [1700/2387] eta: 0:02:13 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1995 (0.2918) acc1: 99.4792 (88.2538) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1896 data: 0.0949 max mem: 3538 +Test: [1710/2387] eta: 0:02:11 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.0978 (0.2906) acc1: 100.0000 (88.3220) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1892 data: 0.0952 max mem: 3538 +Test: [1720/2387] eta: 0:02:09 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.0895 (0.2894) acc1: 100.0000 (88.3896) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1818 data: 0.0895 max mem: 3538 +Test: [1730/2387] eta: 0:02:07 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.0849 (0.2883) acc1: 100.0000 (88.4561) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1703 data: 0.0757 max mem: 3538 +Test: [1740/2387] eta: 0:02:05 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.0848 (0.2871) acc1: 100.0000 (88.5215) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1653 data: 0.0699 max mem: 3538 +Test: [1750/2387] eta: 0:02:03 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.0851 (0.2860) acc1: 100.0000 (88.5869) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1607 data: 0.0654 max mem: 3538 +Test: [1760/2387] eta: 0:02:01 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.0888 (0.2849) acc1: 100.0000 (88.6517) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1756 data: 0.0819 max mem: 3538 +Test: [1770/2387] eta: 0:01:59 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.0997 (0.2839) acc1: 100.0000 (88.7144) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1885 data: 0.0962 max mem: 3538 +Test: [1780/2387] eta: 0:01:57 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.0999 (0.2829) acc1: 100.0000 (88.7778) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1688 data: 0.0743 max mem: 3538 +Test: [1790/2387] eta: 0:01:55 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.0952 (0.2818) acc1: 100.0000 (88.8395) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1437 data: 0.0455 max mem: 3538 +Test: [1800/2387] eta: 0:01:53 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.0923 (0.2807) acc1: 100.0000 (88.9008) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1316 data: 0.0357 max mem: 3538 +Test: [1810/2387] eta: 0:01:51 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.0908 (0.2797) acc1: 100.0000 (88.9617) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1671 data: 0.0757 max mem: 3539 +Test: [1820/2387] eta: 0:01:49 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.0949 (0.2787) acc1: 100.0000 (89.0223) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1848 data: 0.0937 max mem: 3539 +Test: [1830/2387] eta: 0:01:47 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1019 (0.2778) acc1: 100.0000 (89.0823) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1694 data: 0.0778 max mem: 3539 +Test: [1840/2387] eta: 0:01:45 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1161 (0.2769) acc1: 100.0000 (89.1416) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1599 data: 0.0682 max mem: 3539 +Test: [1850/2387] eta: 0:01:43 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1175 (0.2760) acc1: 100.0000 (89.2002) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1661 data: 0.0748 max mem: 3539 +Test: [1860/2387] eta: 0:01:41 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1179 (0.2752) acc1: 100.0000 (89.2583) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1751 data: 0.0833 max mem: 3539 +Test: [1870/2387] eta: 0:01:39 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1177 (0.2743) acc1: 100.0000 (89.3157) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1719 data: 0.0797 max mem: 3539 +Test: [1880/2387] eta: 0:01:37 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1174 (0.2735) acc1: 100.0000 (89.3725) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1656 data: 0.0741 max mem: 3539 +Test: [1890/2387] eta: 0:01:35 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1171 (0.2727) acc1: 100.0000 (89.4287) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1707 data: 0.0791 max mem: 3539 +Test: [1900/2387] eta: 0:01:33 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1175 (0.2719) acc1: 100.0000 (89.4843) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1796 data: 0.0875 max mem: 3539 +Test: [1910/2387] eta: 0:01:31 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1379 (0.2712) acc1: 100.0000 (89.5392) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1841 data: 0.0909 max mem: 3539 +Test: [1920/2387] eta: 0:01:29 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1446 (0.2707) acc1: 100.0000 (89.5927) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1877 data: 0.0950 max mem: 3539 +Test: [1930/2387] eta: 0:01:27 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1294 (0.2699) acc1: 100.0000 (89.6466) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1863 data: 0.0934 max mem: 3539 +Test: [1940/2387] eta: 0:01:25 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1294 (0.2694) acc1: 100.0000 (89.6653) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0857 max mem: 3539 +Test: [1950/2387] eta: 0:01:23 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1335 (0.2688) acc1: 100.0000 (89.7181) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1795 data: 0.0872 max mem: 3539 +Test: [1960/2387] eta: 0:01:21 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1407 (0.2682) acc1: 100.0000 (89.7630) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1799 data: 0.0891 max mem: 3539 +Test: [1970/2387] eta: 0:01:19 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1713 (0.2680) acc1: 100.0000 (89.8037) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1678 data: 0.0765 max mem: 3539 +Test: [1980/2387] eta: 0:01:17 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.2065 (0.2676) acc1: 98.9583 (89.8414) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1569 data: 0.0639 max mem: 3539 +Test: [1990/2387] eta: 0:01:15 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1499 (0.2670) acc1: 98.6979 (89.8864) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1641 data: 0.0692 max mem: 3539 +Test: [2000/2387] eta: 0:01:14 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1479 (0.2664) acc1: 98.6979 (89.9290) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1759 data: 0.0820 max mem: 3539 +Test: [2010/2387] eta: 0:01:12 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1479 (0.2658) acc1: 98.6979 (89.9722) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1719 data: 0.0799 max mem: 3539 +Test: [2020/2387] eta: 0:01:10 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1457 (0.2653) acc1: 98.9583 (90.0159) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1781 data: 0.0862 max mem: 3539 +Test: [2030/2387] eta: 0:01:08 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1460 (0.2647) acc1: 98.9583 (90.0578) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1973 data: 0.1057 max mem: 3540 +Test: [2040/2387] eta: 0:01:06 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1465 (0.2641) acc1: 98.6979 (90.1006) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1858 data: 0.0948 max mem: 3540 +Test: [2050/2387] eta: 0:01:04 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1450 (0.2635) acc1: 98.6979 (90.1434) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1674 data: 0.0758 max mem: 3540 +Test: [2060/2387] eta: 0:01:02 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1486 (0.2630) acc1: 98.6979 (90.1839) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1677 data: 0.0753 max mem: 3540 +Test: [2070/2387] eta: 0:01:00 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1486 (0.2624) acc1: 98.9583 (90.2273) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1770 data: 0.0846 max mem: 3540 +Test: [2080/2387] eta: 0:00:58 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1456 (0.2619) acc1: 98.9583 (90.2684) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1868 data: 0.0943 max mem: 3540 +Test: [2090/2387] eta: 0:00:56 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1456 (0.2613) acc1: 98.9583 (90.3102) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2006 data: 0.1067 max mem: 3540 +Test: [2100/2387] eta: 0:00:54 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1485 (0.2608) acc1: 98.9583 (90.3521) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2145 data: 0.1218 max mem: 3540 +Test: [2110/2387] eta: 0:00:52 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1490 (0.2603) acc1: 99.2188 (90.3931) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1931 data: 0.1018 max mem: 3540 +Test: [2120/2387] eta: 0:00:51 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1522 (0.2599) acc1: 99.2188 (90.4336) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1812 data: 0.0879 max mem: 3540 +Test: [2130/2387] eta: 0:00:49 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1562 (0.2594) acc1: 99.7396 (90.4759) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1906 data: 0.0974 max mem: 3540 +Test: [2140/2387] eta: 0:00:47 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1540 (0.2591) acc1: 99.7396 (90.5025) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1791 data: 0.0874 max mem: 3540 +Test: [2150/2387] eta: 0:00:45 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1559 (0.2586) acc1: 99.7396 (90.5456) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1820 data: 0.0900 max mem: 3540 +Test: [2160/2387] eta: 0:00:43 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1540 (0.2581) acc1: 99.7396 (90.5852) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1820 data: 0.0903 max mem: 3540 +Test: [2170/2387] eta: 0:00:41 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1705 (0.2580) acc1: 99.4792 (90.6179) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1771 data: 0.0860 max mem: 3540 +Test: [2180/2387] eta: 0:00:39 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.2488 (0.2593) acc1: 96.3542 (90.5372) acc5: 100.0000 (100.0000) acc1_neg: 36.198 acc5_neg: 100.000 time: 0.2097 data: 0.1139 max mem: 3540 +Test: [2190/2387] eta: 0:00:37 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.2235 (0.2594) acc1: 95.5729 (90.5326) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2490 data: 0.1520 max mem: 3540 +Test: [2200/2387] eta: 0:00:35 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1312 (0.2588) acc1: 99.7396 (90.5754) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.1001 max mem: 3540 +Test: [2210/2387] eta: 0:00:33 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1128 (0.2581) acc1: 100.0000 (90.6181) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1667 data: 0.0726 max mem: 3540 +Test: [2220/2387] eta: 0:00:31 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1128 (0.2575) acc1: 100.0000 (90.6587) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2006 data: 0.1077 max mem: 3540 +Test: [2230/2387] eta: 0:00:29 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1128 (0.2568) acc1: 100.0000 (90.7005) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1704 data: 0.0765 max mem: 3540 +Test: [2240/2387] eta: 0:00:28 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1083 (0.2562) acc1: 100.0000 (90.7418) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1700 data: 0.0735 max mem: 3540 +Test: [2250/2387] eta: 0:00:26 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1356 (0.2558) acc1: 100.0000 (90.7813) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1926 data: 0.0965 max mem: 3540 +Test: [2260/2387] eta: 0:00:24 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1410 (0.2553) acc1: 100.0000 (90.8220) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1892 data: 0.0951 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1410 (0.2548) acc1: 100.0000 (90.8624) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1805 data: 0.0861 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1304 (0.2544) acc1: 100.0000 (90.8868) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1784 data: 0.0838 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1267 (0.2538) acc1: 100.0000 (90.9266) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1796 data: 0.0867 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1326 (0.2533) acc1: 100.0000 (90.9641) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0934 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1498 (0.2530) acc1: 99.7396 (90.9995) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1836 data: 0.0904 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1805 (0.2527) acc1: 99.4792 (91.0305) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1724 data: 0.0800 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1811 (0.2524) acc1: 99.4792 (91.0653) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1656 data: 0.0742 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1498 (0.2520) acc1: 100.0000 (91.1018) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1492 data: 0.0583 max mem: 3541 +Test: [2350/2387] eta: 0:00:07 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1436 (0.2518) acc1: 100.0000 (91.1204) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1364 data: 0.0463 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1492 (0.2515) acc1: 100.0000 (91.1531) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1412 data: 0.0511 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1700 (0.2512) acc1: 100.0000 (91.1866) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1336 data: 0.0436 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.1987 (0.2512) acc1: 99.2188 (91.2089) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1269 data: 0.0339 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 20.091 acc5_pos: 100.000 loss: 0.2271 (0.2510) acc1: 98.1771 (91.2243) acc5: 100.0000 (100.0000) acc1_neg: 97.674 acc5_neg: 100.000 time: 0.1256 data: 0.0353 max mem: 3541 +Test: Total time: 0:07:31 (0.1893 s / it) +* Overall Acc@1 91.224 Acc@5 100.000 + target == real Acc@1 13.166 Acc@5 100.000 + target == fake Acc@1 96.706 Acc@5 100.000 + loss 0.251 + PR-AUC 0.9901 ROC-AUC 0.8735 EER 19.60% (thr=0.8057) +Accuracy of the network on the 916267 test images: 91.2% +[rank0]:[W726 04:59:58.130085601 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:00:20.171000 1192977 torch/distributed/run.py:766] +W0726 05:00:20.171000 1192977 torch/distributed/run.py:766] ***************************************** +W0726 05:00:20.171000 1192977 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:00:20.171000 1192977 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:00:47.956781300 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:00:48.176515537 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W726 05:00:48.189765618 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:00:48.263369436 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:00:48.595031844 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:00:48.705828685 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:00:48.708834704 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +NCCL version 2.26.2+cuda12.2 +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:00:48.741873494 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n64_seed7_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed7_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:00:23 acc1_pos: 6.771 acc5_pos: 100.000 loss: 1.3141 (1.3141) acc1: 6.7708 (6.7708) acc5: 100.0000 (100.0000) time: 7.0086 data: 5.4950 max mem: 3531 +Test: [ 10/517] eta: 0:06:04 acc1_pos: 0.521 acc5_pos: 100.000 loss: 1.3795 (1.3538) acc1: 6.5104 (7.2443) acc5: 100.0000 (100.0000) time: 0.7193 data: 0.5000 max mem: 3531 +Test: [ 20/517] eta: 0:03:29 acc1_pos: 1.823 acc5_pos: 100.000 loss: 1.3795 (1.3700) acc1: 6.5104 (6.9072) acc5: 100.0000 (100.0000) time: 0.0932 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:02:59 acc1_pos: 0.521 acc5_pos: 100.000 loss: 1.4534 (1.3844) acc1: 6.5104 (6.7792) acc5: 100.0000 (100.0000) time: 0.1741 data: 0.0787 max mem: 3531 +Test: [ 40/517] eta: 0:02:40 acc1_pos: 1.042 acc5_pos: 100.000 loss: 1.4239 (1.3972) acc1: 1.3021 (5.7990) acc5: 100.0000 (100.0000) time: 0.2452 data: 0.1522 max mem: 3531 +Test: [ 50/517] eta: 0:02:29 acc1_pos: 14.679 acc5_pos: 100.000 loss: 1.3928 (1.3304) acc1: 4.4271 (10.7690) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2455 data: 0.1537 max mem: 3531 +Test: [ 60/517] eta: 0:02:22 acc1_pos: 14.679 acc5_pos: 100.000 loss: 0.2515 (1.1506) acc1: 99.4792 (25.3543) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2619 data: 0.1673 max mem: 3531 +Test: [ 70/517] eta: 0:02:15 acc1_pos: 14.679 acc5_pos: 100.000 loss: 0.2138 (1.0193) acc1: 100.0000 (35.8678) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2590 data: 0.1654 max mem: 3531 +Test: [ 80/517] eta: 0:02:14 acc1_pos: 14.679 acc5_pos: 100.000 loss: 0.2199 (0.9215) acc1: 100.0000 (43.7757) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2977 data: 0.2038 max mem: 3531 +Test: [ 90/517] eta: 0:02:08 acc1_pos: 14.679 acc5_pos: 100.000 loss: 0.2290 (0.8463) acc1: 100.0000 (49.9485) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2991 data: 0.2032 max mem: 3531 +Test: [100/517] eta: 0:02:04 acc1_pos: 14.679 acc5_pos: 100.000 loss: 0.2116 (0.7834) acc1: 100.0000 (54.8989) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2556 data: 0.1622 max mem: 3531 +Test: [110/517] eta: 0:01:59 acc1_pos: 14.679 acc5_pos: 100.000 loss: 0.1960 (0.7309) acc1: 100.0000 (58.9550) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2602 data: 0.1672 max mem: 3531 +Test: [120/517] eta: 0:01:55 acc1_pos: 14.679 acc5_pos: 100.000 loss: 0.2159 (0.6894) acc1: 100.0000 (62.3386) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2611 data: 0.1641 max mem: 3531 +Test: [130/517] eta: 0:01:51 acc1_pos: 14.679 acc5_pos: 100.000 loss: 0.2235 (0.6538) acc1: 100.0000 (65.2095) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2546 data: 0.1580 max mem: 3531 +Test: [140/517] eta: 0:01:47 acc1_pos: 14.679 acc5_pos: 100.000 loss: 0.2017 (0.6212) acc1: 100.0000 (67.6769) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2372 data: 0.1428 max mem: 3531 +Test: [150/517] eta: 0:01:42 acc1_pos: 14.679 acc5_pos: 100.000 loss: 0.1912 (0.5924) acc1: 100.0000 (69.8175) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2298 data: 0.1314 max mem: 3531 +Test: [160/517] eta: 0:01:39 acc1_pos: 14.679 acc5_pos: 100.000 loss: 0.1940 (0.5679) acc1: 100.0000 (71.6922) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2355 data: 0.1356 max mem: 3531 +Test: [170/517] eta: 0:01:36 acc1_pos: 14.679 acc5_pos: 100.000 loss: 0.1940 (0.5461) acc1: 100.0000 (73.3446) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2492 data: 0.1520 max mem: 3531 +Test: [180/517] eta: 0:01:32 acc1_pos: 14.679 acc5_pos: 100.000 loss: 0.1892 (0.5269) acc1: 100.0000 (74.8072) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2411 data: 0.1454 max mem: 3531 +Test: [190/517] eta: 0:01:29 acc1_pos: 14.679 acc5_pos: 100.000 loss: 0.1650 (0.5078) acc1: 100.0000 (76.1262) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2348 data: 0.1400 max mem: 3531 +Test: [200/517] eta: 0:01:25 acc1_pos: 14.679 acc5_pos: 100.000 loss: 0.1646 (0.4909) acc1: 100.0000 (77.3140) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2393 data: 0.1441 max mem: 3531 +Test: [210/517] eta: 0:01:22 acc1_pos: 14.679 acc5_pos: 100.000 loss: 0.1860 (0.4794) acc1: 99.7396 (78.3077) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2355 data: 0.1421 max mem: 3532 +Test: [220/517] eta: 0:01:19 acc1_pos: 14.679 acc5_pos: 100.000 loss: 0.2514 (0.4692) acc1: 97.9167 (79.1879) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2408 data: 0.1488 max mem: 3532 +Test: [230/517] eta: 0:01:16 acc1_pos: 14.679 acc5_pos: 100.000 loss: 0.2416 (0.4590) acc1: 98.6979 (80.0640) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2543 data: 0.1610 max mem: 3532 +Test: [240/517] eta: 0:01:13 acc1_pos: 14.679 acc5_pos: 100.000 loss: 0.2172 (0.4486) acc1: 100.0000 (80.8858) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2428 data: 0.1494 max mem: 3532 +Test: [250/517] eta: 0:01:10 acc1_pos: 14.679 acc5_pos: 100.000 loss: 0.1961 (0.4389) acc1: 100.0000 (81.6391) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2399 data: 0.1442 max mem: 3532 +Test: [260/517] eta: 0:01:08 acc1_pos: 14.679 acc5_pos: 100.000 loss: 0.1919 (0.4297) acc1: 99.7396 (82.3346) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2519 data: 0.1539 max mem: 3532 +Test: [270/517] eta: 0:01:05 acc1_pos: 14.679 acc5_pos: 100.000 loss: 0.2025 (0.4215) acc1: 100.0000 (82.9864) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2468 data: 0.1516 max mem: 3532 +Test: [280/517] eta: 0:01:02 acc1_pos: 14.679 acc5_pos: 100.000 loss: 0.2084 (0.4140) acc1: 100.0000 (83.5836) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2479 data: 0.1546 max mem: 3532 +Test: [290/517] eta: 0:00:59 acc1_pos: 14.679 acc5_pos: 100.000 loss: 0.2087 (0.4073) acc1: 99.7396 (84.1423) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2379 data: 0.1420 max mem: 3532 +Test: [300/517] eta: 0:00:56 acc1_pos: 14.679 acc5_pos: 100.000 loss: 0.2050 (0.4003) acc1: 100.0000 (84.6674) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2501 data: 0.1535 max mem: 3532 +Test: [310/517] eta: 0:00:54 acc1_pos: 14.679 acc5_pos: 100.000 loss: 0.2120 (0.3945) acc1: 100.0000 (85.1604) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2605 data: 0.1662 max mem: 3532 +Test: [320/517] eta: 0:00:51 acc1_pos: 14.679 acc5_pos: 100.000 loss: 0.2120 (0.3886) acc1: 100.0000 (85.6227) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2477 data: 0.1538 max mem: 3532 +Test: [330/517] eta: 0:00:48 acc1_pos: 14.679 acc5_pos: 100.000 loss: 0.1918 (0.3825) acc1: 100.0000 (86.0563) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2593 data: 0.1620 max mem: 3532 +Test: [340/517] eta: 0:00:46 acc1_pos: 14.679 acc5_pos: 100.000 loss: 0.1780 (0.3769) acc1: 100.0000 (86.4583) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2568 data: 0.1562 max mem: 3532 +Test: [350/517] eta: 0:00:43 acc1_pos: 14.679 acc5_pos: 100.000 loss: 0.2042 (0.3726) acc1: 100.0000 (86.8352) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2399 data: 0.1414 max mem: 3532 +Test: [360/517] eta: 0:00:40 acc1_pos: 14.679 acc5_pos: 100.000 loss: 0.2132 (0.3683) acc1: 99.7396 (87.1927) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2499 data: 0.1530 max mem: 3532 +Test: [370/517] eta: 0:00:38 acc1_pos: 14.679 acc5_pos: 100.000 loss: 0.1932 (0.3630) acc1: 100.0000 (87.5379) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2399 data: 0.1437 max mem: 3532 +Test: [380/517] eta: 0:00:35 acc1_pos: 14.679 acc5_pos: 100.000 loss: 0.1758 (0.3584) acc1: 100.0000 (87.8629) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2300 data: 0.1335 max mem: 3532 +Test: [390/517] eta: 0:00:32 acc1_pos: 14.679 acc5_pos: 100.000 loss: 0.1943 (0.3545) acc1: 100.0000 (88.1734) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2560 data: 0.1572 max mem: 3532 +Test: [400/517] eta: 0:00:30 acc1_pos: 14.679 acc5_pos: 100.000 loss: 0.1931 (0.3502) acc1: 100.0000 (88.4683) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2461 data: 0.1516 max mem: 3532 +Test: [410/517] eta: 0:00:27 acc1_pos: 14.679 acc5_pos: 100.000 loss: 0.1925 (0.3472) acc1: 100.0000 (88.7489) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2390 data: 0.1474 max mem: 3532 +Test: [420/517] eta: 0:00:25 acc1_pos: 14.679 acc5_pos: 100.000 loss: 0.2132 (0.3440) acc1: 100.0000 (89.0161) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2637 data: 0.1699 max mem: 3532 +Test: [430/517] eta: 0:00:22 acc1_pos: 14.679 acc5_pos: 100.000 loss: 0.2092 (0.3408) acc1: 100.0000 (89.2704) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2597 data: 0.1657 max mem: 3532 +Test: [440/517] eta: 0:00:19 acc1_pos: 14.679 acc5_pos: 100.000 loss: 0.2154 (0.3381) acc1: 100.0000 (89.5125) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2551 data: 0.1588 max mem: 3533 +Test: [450/517] eta: 0:00:17 acc1_pos: 14.679 acc5_pos: 100.000 loss: 0.2147 (0.3351) acc1: 100.0000 (89.7450) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2749 data: 0.1777 max mem: 3533 +Test: [460/517] eta: 0:00:14 acc1_pos: 14.679 acc5_pos: 100.000 loss: 0.2031 (0.3323) acc1: 100.0000 (89.9663) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2713 data: 0.1756 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 14.679 acc5_pos: 100.000 loss: 0.1951 (0.3292) acc1: 100.0000 (90.1794) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2624 data: 0.1691 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 14.679 acc5_pos: 100.000 loss: 0.1816 (0.3262) acc1: 100.0000 (90.3835) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2733 data: 0.1802 max mem: 3533 +Test: [490/517] eta: 0:00:07 acc1_pos: 14.679 acc5_pos: 100.000 loss: 0.1940 (0.3237) acc1: 100.0000 (90.5783) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2677 data: 0.1696 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 14.679 acc5_pos: 100.000 loss: 0.2140 (0.3219) acc1: 100.0000 (90.7622) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2313 data: 0.1329 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 14.679 acc5_pos: 100.000 loss: 0.2222 (0.3200) acc1: 100.0000 (90.9389) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1827 data: 0.0894 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 14.679 acc5_pos: 100.000 loss: 0.2032 (0.3186) acc1: 100.0000 (91.0395) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1817 data: 0.0893 max mem: 3533 +Test: Total time: 0:02:11 (0.2552 s / it) +* Overall Acc@1 91.040 Acc@5 100.000 + target == real Acc@1 6.306 Acc@5 100.000 + target == fake Acc@1 99.830 Acc@5 100.000 + loss 0.319 + PR-AUC 0.9552 ROC-AUC 0.7405 EER 32.27% (thr=0.8008) +Accuracy of the network on the 198427 test images: 91.0% +[rank0]:[W726 05:03:12.020218061 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 05:03:32.832000 1193550 torch/distributed/run.py:766] +W0726 05:03:32.832000 1193550 torch/distributed/run.py:766] ***************************************** +W0726 05:03:32.832000 1193550 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:03:32.832000 1193550 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:04:01.457185562 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:04:01.076527445 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:04:01.109443329 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:04:01.141291211 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:04:02.280731599 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:04:02.372594895 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:04:02.376211935 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:04:02.387167801 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n64_seed7_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed7_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:35:14 acc1_pos: 35.677 acc5_pos: 100.000 loss: 0.8567 (0.8567) acc1: 35.6771 (35.6771) acc5: 100.0000 (100.0000) time: 5.4345 data: 3.3451 max mem: 3531 +Test: [ 10/389] eta: 0:03:38 acc1_pos: 71.875 acc5_pos: 100.000 loss: 0.6976 (0.7383) acc1: 48.6979 (50.5208) acc5: 100.0000 (100.0000) time: 0.5762 data: 0.3045 max mem: 3531 +Test: [ 20/389] eta: 0:02:07 acc1_pos: 19.271 acc5_pos: 100.000 loss: 0.7598 (0.7790) acc1: 38.2812 (43.3780) acc5: 100.0000 (100.0000) time: 0.0904 data: 0.0004 max mem: 3531 +Test: [ 30/389] eta: 0:01:43 acc1_pos: 61.198 acc5_pos: 100.000 loss: 0.7598 (0.7611) acc1: 41.1458 (45.5393) acc5: 100.0000 (100.0000) time: 0.1318 data: 0.0396 max mem: 3531 +Test: [ 40/389] eta: 0:01:39 acc1_pos: 24.740 acc5_pos: 100.000 loss: 0.6949 (0.7496) acc1: 53.6458 (47.9294) acc5: 100.0000 (100.0000) time: 0.2215 data: 0.1262 max mem: 3531 +Test: [ 50/389] eta: 0:01:31 acc1_pos: 88.802 acc5_pos: 100.000 loss: 0.6167 (0.7138) acc1: 67.1875 (52.7522) acc5: 100.0000 (100.0000) time: 0.2417 data: 0.1462 max mem: 3531 +Test: [ 60/389] eta: 0:01:22 acc1_pos: 49.479 acc5_pos: 100.000 loss: 0.5105 (0.6836) acc1: 83.5938 (56.7666) acc5: 100.0000 (100.0000) time: 0.1799 data: 0.0858 max mem: 3531 +Test: [ 70/389] eta: 0:01:16 acc1_pos: 57.812 acc5_pos: 100.000 loss: 0.5806 (0.6877) acc1: 69.5312 (56.0336) acc5: 100.0000 (100.0000) time: 0.1599 data: 0.0649 max mem: 3531 +Test: [ 80/389] eta: 0:01:12 acc1_pos: 91.667 acc5_pos: 100.000 loss: 0.7066 (0.6904) acc1: 46.8750 (55.3466) acc5: 100.0000 (100.0000) time: 0.1899 data: 0.0948 max mem: 3531 +Test: [ 90/389] eta: 0:01:06 acc1_pos: 61.198 acc5_pos: 100.000 loss: 0.5350 (0.6712) acc1: 70.3125 (57.3603) acc5: 100.0000 (100.0000) time: 0.1634 data: 0.0687 max mem: 3531 +Test: [100/389] eta: 0:01:01 acc1_pos: 47.917 acc5_pos: 100.000 loss: 0.6572 (0.6818) acc1: 61.1979 (56.4150) acc5: 100.0000 (100.0000) time: 0.1171 data: 0.0217 max mem: 3531 +Test: [110/389] eta: 0:00:58 acc1_pos: 99.479 acc5_pos: 100.000 loss: 0.7768 (0.6935) acc1: 47.9167 (55.6259) acc5: 100.0000 (100.0000) time: 0.1472 data: 0.0535 max mem: 3531 +Test: [120/389] eta: 0:00:55 acc1_pos: 64.062 acc5_pos: 100.000 loss: 0.6170 (0.6868) acc1: 54.6875 (56.3533) acc5: 100.0000 (100.0000) time: 0.1735 data: 0.0820 max mem: 3531 +Test: [130/389] eta: 0:00:52 acc1_pos: 60.123 acc5_pos: 100.000 loss: 0.5821 (0.6770) acc1: 69.5312 (57.5958) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1782 data: 0.0873 max mem: 3531 +Test: [140/389] eta: 0:00:50 acc1_pos: 60.123 acc5_pos: 100.000 loss: 0.5070 (0.6664) acc1: 78.6458 (58.7323) acc5: 100.0000 (100.0000) acc1_neg: 21.615 acc5_neg: 100.000 time: 0.1840 data: 0.0912 max mem: 3531 +Test: [150/389] eta: 0:00:47 acc1_pos: 60.123 acc5_pos: 100.000 loss: 0.5246 (0.6587) acc1: 77.3438 (59.4974) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1564 data: 0.0619 max mem: 3531 +Test: [160/389] eta: 0:00:44 acc1_pos: 60.123 acc5_pos: 100.000 loss: 0.5450 (0.6441) acc1: 75.5208 (60.8631) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1487 data: 0.0555 max mem: 3531 +Test: [170/389] eta: 0:00:42 acc1_pos: 60.123 acc5_pos: 100.000 loss: 0.2388 (0.6286) acc1: 99.7396 (62.1954) acc5: 100.0000 (100.0000) acc1_neg: 68.490 acc5_neg: 100.000 time: 0.1684 data: 0.0769 max mem: 3531 +Test: [180/389] eta: 0:00:40 acc1_pos: 60.123 acc5_pos: 100.000 loss: 0.3321 (0.6177) acc1: 90.1042 (63.6568) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.1634 data: 0.0717 max mem: 3531 +Test: [190/389] eta: 0:00:38 acc1_pos: 60.123 acc5_pos: 100.000 loss: 0.4448 (0.6129) acc1: 87.5000 (64.3052) acc5: 100.0000 (100.0000) acc1_neg: 52.083 acc5_neg: 100.000 time: 0.1689 data: 0.0761 max mem: 3531 +Test: [200/389] eta: 0:00:36 acc1_pos: 60.123 acc5_pos: 100.000 loss: 0.5280 (0.6083) acc1: 80.9896 (65.1430) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.1828 data: 0.0903 max mem: 3531 +Test: [210/389] eta: 0:00:34 acc1_pos: 60.123 acc5_pos: 100.000 loss: 0.4150 (0.5977) acc1: 88.5417 (66.3199) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0881 max mem: 3532 +Test: [220/389] eta: 0:00:32 acc1_pos: 60.123 acc5_pos: 100.000 loss: 0.3018 (0.5842) acc1: 98.9583 (67.6966) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1848 data: 0.0871 max mem: 3532 +Test: [230/389] eta: 0:00:30 acc1_pos: 60.123 acc5_pos: 100.000 loss: 0.4222 (0.5821) acc1: 94.2708 (67.9361) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.2185 data: 0.1206 max mem: 3532 +Test: [240/389] eta: 0:00:28 acc1_pos: 60.123 acc5_pos: 100.000 loss: 0.4391 (0.5764) acc1: 85.9375 (68.7630) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1999 data: 0.1072 max mem: 3532 +Test: [250/389] eta: 0:00:26 acc1_pos: 60.123 acc5_pos: 100.000 loss: 0.3898 (0.5668) acc1: 96.0938 (69.9255) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1417 data: 0.0478 max mem: 3532 +Test: [260/389] eta: 0:00:24 acc1_pos: 60.123 acc5_pos: 100.000 loss: 0.3733 (0.5641) acc1: 98.9583 (70.4133) acc5: 100.0000 (100.0000) acc1_neg: 35.417 acc5_neg: 100.000 time: 0.1649 data: 0.0691 max mem: 3532 +Test: [270/389] eta: 0:00:22 acc1_pos: 60.123 acc5_pos: 100.000 loss: 0.5165 (0.5677) acc1: 77.6042 (69.9704) acc5: 100.0000 (100.0000) acc1_neg: 71.615 acc5_neg: 100.000 time: 0.1846 data: 0.0885 max mem: 3532 +Test: [280/389] eta: 0:00:20 acc1_pos: 60.123 acc5_pos: 100.000 loss: 0.6039 (0.5716) acc1: 60.9375 (69.5396) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1707 data: 0.0777 max mem: 3532 +Test: [290/389] eta: 0:00:18 acc1_pos: 60.123 acc5_pos: 100.000 loss: 0.5618 (0.5672) acc1: 70.0521 (70.1174) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1749 data: 0.0828 max mem: 3532 +Test: [300/389] eta: 0:00:16 acc1_pos: 60.123 acc5_pos: 100.000 loss: 0.5355 (0.5667) acc1: 85.1562 (70.4829) acc5: 100.0000 (100.0000) acc1_neg: 76.042 acc5_neg: 100.000 time: 0.1749 data: 0.0801 max mem: 3532 +Test: [310/389] eta: 0:00:14 acc1_pos: 60.123 acc5_pos: 100.000 loss: 0.5663 (0.5693) acc1: 76.0417 (70.2078) acc5: 100.0000 (100.0000) acc1_neg: 12.500 acc5_neg: 100.000 time: 0.1799 data: 0.0835 max mem: 3532 +Test: [320/389] eta: 0:00:12 acc1_pos: 60.123 acc5_pos: 100.000 loss: 0.5181 (0.5671) acc1: 68.7500 (70.4464) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1799 data: 0.0852 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 60.123 acc5_pos: 100.000 loss: 0.5058 (0.5655) acc1: 88.2812 (70.9797) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.1800 data: 0.0876 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 60.123 acc5_pos: 100.000 loss: 0.3438 (0.5574) acc1: 94.2708 (71.7429) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1741 data: 0.0800 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 60.123 acc5_pos: 100.000 loss: 0.2828 (0.5680) acc1: 99.4792 (70.9877) acc5: 100.0000 (100.0000) acc1_neg: 0.781 acc5_neg: 100.000 time: 0.1711 data: 0.0767 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 60.123 acc5_pos: 100.000 loss: 0.9134 (0.5802) acc1: 30.4688 (69.6820) acc5: 100.0000 (100.0000) acc1_neg: 30.469 acc5_neg: 100.000 time: 0.1799 data: 0.0865 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 60.123 acc5_pos: 100.000 loss: 0.8816 (0.5783) acc1: 32.8125 (69.8275) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1714 data: 0.0788 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 60.123 acc5_pos: 100.000 loss: 0.5134 (0.5766) acc1: 74.4792 (70.0610) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.1537 data: 0.0630 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 60.123 acc5_pos: 100.000 loss: 0.4527 (0.5741) acc1: 82.0312 (70.3417) acc5: 100.0000 (100.0000) acc1_neg: 83.880 acc5_neg: 100.000 time: 0.1437 data: 0.0466 max mem: 3532 +Test: Total time: 0:01:11 (0.1838 s / it) +* Overall Acc@1 70.342 Acc@5 100.000 + target == real Acc@1 56.249 Acc@5 100.000 + target == fake Acc@1 76.893 Acc@5 100.000 + loss 0.574 + PR-AUC 0.8592 ROC-AUC 0.7457 EER 31.53% (thr=0.5562) +Accuracy of the network on the 149358 test images: 70.3% +[rank0]:[W726 05:05:25.603894062 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:05:45.396000 1194108 torch/distributed/run.py:766] +W0726 05:05:45.396000 1194108 torch/distributed/run.py:766] ***************************************** +W0726 05:05:45.396000 1194108 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:05:45.396000 1194108 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:06:15.185382364 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:06:15.693530304 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:06:15.806719508 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:06:15.843542671 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:06:15.021129603 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:06:15.090777622 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:06:16.278489916 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank6]:[W726 05:06:16.285209385 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n64_seed7_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed7_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:38:54 acc1_pos: 78.646 acc5_pos: 100.000 loss: 0.4738 (0.4738) acc1: 78.6458 (78.6458) acc5: 100.0000 (100.0000) time: 7.0092 data: 5.4090 max mem: 3531 +Test: [ 10/333] eta: 0:03:53 acc1_pos: 79.688 acc5_pos: 100.000 loss: 0.4134 (0.3965) acc1: 79.6875 (79.7349) acc5: 100.0000 (100.0000) time: 0.7239 data: 0.4921 max mem: 3531 +Test: [ 20/333] eta: 0:02:13 acc1_pos: 89.323 acc5_pos: 100.000 loss: 0.4002 (0.3991) acc1: 78.9062 (79.1915) acc5: 100.0000 (100.0000) time: 0.0976 data: 0.0004 max mem: 3531 +Test: [ 30/333] eta: 0:01:52 acc1_pos: 47.917 acc5_pos: 100.000 loss: 0.5246 (0.5673) acc1: 69.7917 (66.9271) acc5: 100.0000 (100.0000) time: 0.1749 data: 0.0791 max mem: 3531 +Test: [ 40/333] eta: 0:01:36 acc1_pos: 33.594 acc5_pos: 100.000 loss: 0.9481 (0.6599) acc1: 39.3229 (59.5592) acc5: 100.0000 (100.0000) time: 0.2269 data: 0.1349 max mem: 3531 +Test: [ 50/333] eta: 0:01:26 acc1_pos: 41.667 acc5_pos: 100.000 loss: 0.9338 (0.7159) acc1: 36.1979 (55.2288) acc5: 100.0000 (100.0000) time: 0.2049 data: 0.1128 max mem: 3531 +Test: [ 60/333] eta: 0:01:18 acc1_pos: 24.219 acc5_pos: 100.000 loss: 0.9085 (0.7516) acc1: 39.0625 (52.2925) acc5: 100.0000 (100.0000) time: 0.1986 data: 0.1068 max mem: 3531 +Test: [ 70/333] eta: 0:01:13 acc1_pos: 32.552 acc5_pos: 100.000 loss: 0.9109 (0.7792) acc1: 37.2396 (50.0734) acc5: 100.0000 (100.0000) time: 0.2134 data: 0.1217 max mem: 3531 +Test: [ 80/333] eta: 0:01:08 acc1_pos: 40.365 acc5_pos: 100.000 loss: 0.9098 (0.7981) acc1: 39.0625 (48.6722) acc5: 100.0000 (100.0000) time: 0.2162 data: 0.1246 max mem: 3531 +Test: [ 90/333] eta: 0:01:04 acc1_pos: 50.260 acc5_pos: 100.000 loss: 0.8755 (0.8014) acc1: 41.4062 (48.2572) acc5: 100.0000 (100.0000) time: 0.2132 data: 0.1216 max mem: 3531 +Test: [100/333] eta: 0:01:00 acc1_pos: 53.125 acc5_pos: 100.000 loss: 0.8136 (0.8051) acc1: 44.2708 (48.0172) acc5: 100.0000 (100.0000) time: 0.2230 data: 0.1293 max mem: 3531 +Test: [110/333] eta: 0:00:56 acc1_pos: 46.354 acc5_pos: 100.000 loss: 0.8172 (0.8087) acc1: 44.7917 (47.7055) acc5: 100.0000 (100.0000) time: 0.2135 data: 0.1127 max mem: 3531 +Test: [120/333] eta: 0:00:53 acc1_pos: 48.438 acc5_pos: 100.000 loss: 0.8158 (0.8085) acc1: 44.7917 (47.5723) acc5: 100.0000 (100.0000) time: 0.1999 data: 0.0978 max mem: 3531 +Test: [130/333] eta: 0:00:49 acc1_pos: 60.417 acc5_pos: 100.000 loss: 0.7619 (0.8080) acc1: 48.4375 (47.5747) acc5: 100.0000 (100.0000) time: 0.1921 data: 0.0958 max mem: 3531 +Test: [140/333] eta: 0:00:46 acc1_pos: 69.010 acc5_pos: 100.000 loss: 0.7619 (0.8062) acc1: 50.7812 (47.6489) acc5: 100.0000 (100.0000) time: 0.1966 data: 0.1037 max mem: 3531 +Test: [150/333] eta: 0:00:43 acc1_pos: 51.302 acc5_pos: 100.000 loss: 0.8036 (0.8081) acc1: 48.4375 (47.4959) acc5: 100.0000 (100.0000) time: 0.2024 data: 0.1090 max mem: 3531 +Test: [160/333] eta: 0:00:40 acc1_pos: 35.526 acc5_pos: 100.000 loss: 0.8207 (0.8084) acc1: 48.9583 (47.6611) acc5: 100.0000 (100.0000) acc1_neg: 58.594 acc5_neg: 100.000 time: 0.1905 data: 0.0957 max mem: 3531 +Test: [170/333] eta: 0:00:37 acc1_pos: 35.526 acc5_pos: 100.000 loss: 0.7395 (0.8007) acc1: 55.9896 (48.6096) acc5: 100.0000 (100.0000) acc1_neg: 68.229 acc5_neg: 100.000 time: 0.1706 data: 0.0771 max mem: 3531 +Test: [180/333] eta: 0:00:35 acc1_pos: 35.526 acc5_pos: 100.000 loss: 0.6275 (0.7901) acc1: 66.9271 (49.7497) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.1877 data: 0.0916 max mem: 3531 +Test: [190/333] eta: 0:00:32 acc1_pos: 35.526 acc5_pos: 100.000 loss: 0.5645 (0.7771) acc1: 70.0521 (51.0144) acc5: 100.0000 (100.0000) acc1_neg: 70.052 acc5_neg: 100.000 time: 0.2201 data: 0.1245 max mem: 3531 +Test: [200/333] eta: 0:00:30 acc1_pos: 35.526 acc5_pos: 100.000 loss: 0.5484 (0.7653) acc1: 73.6979 (52.1961) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.2099 data: 0.1169 max mem: 3531 +Test: [210/333] eta: 0:00:27 acc1_pos: 35.526 acc5_pos: 100.000 loss: 0.5798 (0.7591) acc1: 71.6146 (52.9399) acc5: 100.0000 (100.0000) acc1_neg: 56.510 acc5_neg: 100.000 time: 0.1559 data: 0.0610 max mem: 3532 +Test: [220/333] eta: 0:00:25 acc1_pos: 35.526 acc5_pos: 100.000 loss: 0.6221 (0.7583) acc1: 64.8438 (53.1922) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.1649 data: 0.0699 max mem: 3532 +Test: [230/333] eta: 0:00:22 acc1_pos: 35.526 acc5_pos: 100.000 loss: 0.4374 (0.7428) acc1: 79.9479 (54.6683) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.2080 data: 0.1143 max mem: 3532 +Test: [240/333] eta: 0:00:20 acc1_pos: 35.526 acc5_pos: 100.000 loss: 0.4236 (0.7315) acc1: 83.3333 (55.7800) acc5: 100.0000 (100.0000) acc1_neg: 80.729 acc5_neg: 100.000 time: 0.1902 data: 0.0968 max mem: 3532 +Test: [250/333] eta: 0:00:18 acc1_pos: 35.526 acc5_pos: 100.000 loss: 0.4809 (0.7240) acc1: 78.1250 (56.5249) acc5: 100.0000 (100.0000) acc1_neg: 65.365 acc5_neg: 100.000 time: 0.1757 data: 0.0834 max mem: 3532 +Test: [260/333] eta: 0:00:15 acc1_pos: 35.526 acc5_pos: 100.000 loss: 0.5620 (0.7194) acc1: 72.9167 (57.0672) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.1946 data: 0.1029 max mem: 3532 +Test: [270/333] eta: 0:00:13 acc1_pos: 35.526 acc5_pos: 100.000 loss: 0.5362 (0.7122) acc1: 74.2188 (57.7366) acc5: 100.0000 (100.0000) acc1_neg: 74.219 acc5_neg: 100.000 time: 0.1914 data: 0.0976 max mem: 3532 +Test: [280/333] eta: 0:00:11 acc1_pos: 35.526 acc5_pos: 100.000 loss: 0.5324 (0.7095) acc1: 73.4375 (58.0396) acc5: 100.0000 (100.0000) acc1_neg: 71.094 acc5_neg: 100.000 time: 0.1782 data: 0.0815 max mem: 3532 +Test: [290/333] eta: 0:00:09 acc1_pos: 35.526 acc5_pos: 100.000 loss: 0.7136 (0.7105) acc1: 57.8125 (58.0121) acc5: 100.0000 (100.0000) acc1_neg: 61.458 acc5_neg: 100.000 time: 0.1688 data: 0.0728 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 35.526 acc5_pos: 100.000 loss: 0.7497 (0.7120) acc1: 56.7708 (58.0201) acc5: 100.0000 (100.0000) acc1_neg: 61.979 acc5_neg: 100.000 time: 0.1460 data: 0.0541 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 35.526 acc5_pos: 100.000 loss: 0.6738 (0.7053) acc1: 63.2812 (58.6758) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.1419 data: 0.0517 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 35.526 acc5_pos: 100.000 loss: 0.4699 (0.6994) acc1: 80.7292 (59.2566) acc5: 100.0000 (100.0000) acc1_neg: 51.562 acc5_neg: 100.000 time: 0.1331 data: 0.0431 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 35.526 acc5_pos: 100.000 loss: 0.7526 (0.7036) acc1: 55.9896 (59.0548) acc5: 100.0000 (100.0000) acc1_neg: 55.990 acc5_neg: 100.000 time: 0.1138 data: 0.0239 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 35.526 acc5_pos: 100.000 loss: 0.8010 (0.7044) acc1: 53.6458 (59.0161) acc5: 100.0000 (100.0000) acc1_neg: 52.394 acc5_neg: 100.000 time: 0.1231 data: 0.0301 max mem: 3532 +Test: Total time: 0:01:07 (0.2032 s / it) +* Overall Acc@1 59.016 Acc@5 100.000 + target == real Acc@1 47.469 Acc@5 100.000 + target == fake Acc@1 69.295 Acc@5 100.000 + loss 0.704 + PR-AUC 0.6316 ROC-AUC 0.6215 EER 41.42% (thr=0.5776) +Accuracy of the network on the 127843 test images: 59.0% +[rank0]:[W726 05:07:33.661454676 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:07:53.228000 1194667 torch/distributed/run.py:766] +W0726 05:07:53.228000 1194667 torch/distributed/run.py:766] ***************************************** +W0726 05:07:53.228000 1194667 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:07:53.228000 1194667 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:08:23.527858600 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:08:23.761480723 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:08:23.891247137 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:08:24.272806043 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:08:24.385429965 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:08:24.510911400 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:08:24.587533593 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:08:24.612729638 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n64_seed7_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed7_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:26:09 acc1_pos: 92.188 acc5_pos: 100.000 loss: 0.2428 (0.2428) acc1: 92.1875 (92.1875) acc5: 100.0000 (100.0000) time: 6.6771 data: 5.1501 max mem: 3531 +Test: [ 10/235] eta: 0:02:35 acc1_pos: 92.969 acc5_pos: 100.000 loss: 0.2492 (0.2489) acc1: 92.9688 (93.2055) acc5: 100.0000 (100.0000) time: 0.6922 data: 0.4686 max mem: 3531 +Test: [ 20/235] eta: 0:01:31 acc1_pos: 94.010 acc5_pos: 100.000 loss: 0.2443 (0.2433) acc1: 93.4896 (93.6012) acc5: 100.0000 (100.0000) time: 0.1118 data: 0.0184 max mem: 3531 +Test: [ 30/235] eta: 0:01:17 acc1_pos: 95.052 acc5_pos: 100.000 loss: 0.2347 (0.2412) acc1: 93.7500 (93.6072) acc5: 100.0000 (100.0000) time: 0.2059 data: 0.1124 max mem: 3531 +Test: [ 40/235] eta: 0:01:09 acc1_pos: 93.490 acc5_pos: 100.000 loss: 0.2347 (0.2408) acc1: 93.4896 (93.6039) acc5: 100.0000 (100.0000) time: 0.2899 data: 0.1958 max mem: 3531 +Test: [ 50/235] eta: 0:01:00 acc1_pos: 92.708 acc5_pos: 100.000 loss: 0.2383 (0.2417) acc1: 93.4896 (93.5406) acc5: 100.0000 (100.0000) time: 0.2539 data: 0.1579 max mem: 3531 +Test: [ 60/235] eta: 0:00:51 acc1_pos: 92.708 acc5_pos: 100.000 loss: 0.2450 (0.2434) acc1: 93.7500 (93.5067) acc5: 100.0000 (100.0000) time: 0.1649 data: 0.0672 max mem: 3531 +Test: [ 70/235] eta: 0:00:45 acc1_pos: 93.846 acc5_pos: 100.000 loss: 0.2652 (0.3871) acc1: 91.4062 (84.5987) acc5: 100.0000 (100.0000) acc1_neg: 21.354 acc5_neg: 100.000 time: 0.1461 data: 0.0505 max mem: 3531 +Test: [ 80/235] eta: 0:00:42 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.4095 (0.5148) acc1: 21.8750 (76.8261) acc5: 100.0000 (100.0000) acc1_neg: 21.354 acc5_neg: 100.000 time: 0.2127 data: 0.1207 max mem: 3531 +Test: [ 90/235] eta: 0:00:37 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.4190 (0.6136) acc1: 21.3542 (70.7189) acc5: 100.0000 (100.0000) acc1_neg: 22.917 acc5_neg: 100.000 time: 0.2037 data: 0.1115 max mem: 3531 +Test: [100/235] eta: 0:00:33 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.4177 (0.6914) acc1: 21.3542 (65.9447) acc5: 100.0000 (100.0000) acc1_neg: 19.010 acc5_neg: 100.000 time: 0.1571 data: 0.0625 max mem: 3531 +Test: [110/235] eta: 0:00:30 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.3799 (0.7547) acc1: 22.3958 (62.0425) acc5: 100.0000 (100.0000) acc1_neg: 23.958 acc5_neg: 100.000 time: 0.1566 data: 0.0627 max mem: 3531 +Test: [120/235] eta: 0:00:27 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.4490 (0.8227) acc1: 18.7500 (58.0406) acc5: 100.0000 (100.0000) acc1_neg: 15.104 acc5_neg: 100.000 time: 0.1704 data: 0.0787 max mem: 3531 +Test: [130/235] eta: 0:00:24 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.5841 (0.8814) acc1: 12.7604 (54.6000) acc5: 100.0000 (100.0000) acc1_neg: 14.062 acc5_neg: 100.000 time: 0.1932 data: 0.1010 max mem: 3531 +Test: [140/235] eta: 0:00:22 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.6064 (0.9344) acc1: 12.7604 (51.6142) acc5: 100.0000 (100.0000) acc1_neg: 8.594 acc5_neg: 100.000 time: 0.2244 data: 0.1281 max mem: 3531 +Test: [150/235] eta: 0:00:19 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.6093 (0.9778) acc1: 11.9792 (49.0290) acc5: 100.0000 (100.0000) acc1_neg: 15.885 acc5_neg: 100.000 time: 0.1949 data: 0.0952 max mem: 3531 +Test: [160/235] eta: 0:00:16 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.5840 (1.0158) acc1: 13.0208 (46.8054) acc5: 100.0000 (100.0000) acc1_neg: 14.583 acc5_neg: 100.000 time: 0.1199 data: 0.0202 max mem: 3531 +Test: [170/235] eta: 0:00:14 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.5478 (1.0244) acc1: 16.1458 (46.0100) acc5: 100.0000 (100.0000) acc1_neg: 30.729 acc5_neg: 100.000 time: 0.1399 data: 0.0395 max mem: 3531 +Test: [180/235] eta: 0:00:12 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.1167 (1.0285) acc1: 35.9375 (45.4406) acc5: 100.0000 (100.0000) acc1_neg: 36.198 acc5_neg: 100.000 time: 0.1907 data: 0.0938 max mem: 3531 +Test: [190/235] eta: 0:00:09 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.0834 (1.0316) acc1: 35.9375 (44.9103) acc5: 100.0000 (100.0000) acc1_neg: 37.500 acc5_neg: 100.000 time: 0.2149 data: 0.1202 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.0834 (1.0354) acc1: 34.6354 (44.3848) acc5: 100.0000 (100.0000) acc1_neg: 32.031 acc5_neg: 100.000 time: 0.2086 data: 0.1114 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.0980 (1.0379) acc1: 34.6354 (44.0228) acc5: 100.0000 (100.0000) acc1_neg: 35.677 acc5_neg: 100.000 time: 0.1544 data: 0.0588 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.0924 (1.0401) acc1: 36.1979 (43.6840) acc5: 100.0000 (100.0000) acc1_neg: 35.938 acc5_neg: 100.000 time: 0.1296 data: 0.0377 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.0786 (1.0418) acc1: 36.1979 (43.3994) acc5: 100.0000 (100.0000) acc1_neg: 40.104 acc5_neg: 100.000 time: 0.1250 data: 0.0350 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.0799 (1.0436) acc1: 35.9375 (43.2860) acc5: 100.0000 (100.0000) acc1_neg: 30.435 acc5_neg: 100.000 time: 0.1234 data: 0.0349 max mem: 3532 +Test: Total time: 0:00:47 (0.2039 s / it) +* Overall Acc@1 43.286 Acc@5 100.000 + target == real Acc@1 93.512 Acc@5 100.000 + target == fake Acc@1 25.120 Acc@5 100.000 + loss 1.044 + PR-AUC 0.8663 ROC-AUC 0.7125 EER 34.58% (thr=0.1893) +Accuracy of the network on the 89879 test images: 43.3% +[rank0]:[W726 05:09:23.109528553 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:09:43.856000 1195215 torch/distributed/run.py:766] +W0726 05:09:43.856000 1195215 torch/distributed/run.py:766] ***************************************** +W0726 05:09:43.856000 1195215 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:09:43.856000 1195215 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:10:12.149072018 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:10:13.420429875 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +[rank6]:[W726 05:10:13.987766816 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:10:13.994841588 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:10:14.244655165 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank4]:[W726 05:10:14.247163845 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:10:14.260957735 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:10:14.277883086 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n64_seed7_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed7_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:26:08 acc1_pos: 38.542 acc5_pos: 100.000 loss: 0.8294 (0.8294) acc1: 38.5417 (38.5417) acc5: 100.0000 (100.0000) time: 5.8305 data: 4.4566 max mem: 3531 +Test: [ 10/269] eta: 0:02:38 acc1_pos: 50.521 acc5_pos: 100.000 loss: 0.7973 (0.7832) acc1: 44.2708 (45.9044) acc5: 100.0000 (100.0000) time: 0.6122 data: 0.4055 max mem: 3531 +Test: [ 20/269] eta: 0:01:30 acc1_pos: 47.656 acc5_pos: 100.000 loss: 0.7410 (0.7322) acc1: 47.6562 (51.1657) acc5: 100.0000 (100.0000) time: 0.0920 data: 0.0019 max mem: 3531 +Test: [ 30/269] eta: 0:01:12 acc1_pos: 50.781 acc5_pos: 100.000 loss: 0.6925 (0.7154) acc1: 50.7812 (53.5198) acc5: 100.0000 (100.0000) time: 0.1312 data: 0.0399 max mem: 3531 +Test: [ 40/269] eta: 0:01:05 acc1_pos: 34.483 acc5_pos: 100.000 loss: 0.5316 (0.6550) acc1: 75.7812 (60.8867) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.2012 data: 0.1089 max mem: 3531 +Test: [ 50/269] eta: 0:00:58 acc1_pos: 34.483 acc5_pos: 100.000 loss: 0.4346 (0.6161) acc1: 85.9375 (65.3237) acc5: 100.0000 (100.0000) acc1_neg: 66.146 acc5_neg: 100.000 time: 0.2170 data: 0.1249 max mem: 3531 +Test: [ 60/269] eta: 0:00:53 acc1_pos: 34.483 acc5_pos: 100.000 loss: 0.4545 (0.5957) acc1: 82.8125 (67.7425) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.1938 data: 0.1007 max mem: 3531 +Test: [ 70/269] eta: 0:00:48 acc1_pos: 34.483 acc5_pos: 100.000 loss: 0.4430 (0.5721) acc1: 85.1562 (70.6793) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1872 data: 0.0934 max mem: 3531 +Test: [ 80/269] eta: 0:00:44 acc1_pos: 34.483 acc5_pos: 100.000 loss: 0.3998 (0.5503) acc1: 89.8438 (73.2060) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1788 data: 0.0867 max mem: 3531 +Test: [ 90/269] eta: 0:00:41 acc1_pos: 34.483 acc5_pos: 100.000 loss: 0.4530 (0.5470) acc1: 86.7188 (73.7237) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1820 data: 0.0887 max mem: 3531 +Test: [100/269] eta: 0:00:38 acc1_pos: 34.483 acc5_pos: 100.000 loss: 0.5140 (0.5405) acc1: 79.6875 (74.5797) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.1996 data: 0.1035 max mem: 3531 +Test: [110/269] eta: 0:00:35 acc1_pos: 34.483 acc5_pos: 100.000 loss: 0.5229 (0.5419) acc1: 77.6042 (74.4698) acc5: 100.0000 (100.0000) acc1_neg: 64.844 acc5_neg: 100.000 time: 0.2010 data: 0.1056 max mem: 3531 +Test: [120/269] eta: 0:00:33 acc1_pos: 34.483 acc5_pos: 100.000 loss: 0.5286 (0.5404) acc1: 76.8229 (74.7138) acc5: 100.0000 (100.0000) acc1_neg: 74.740 acc5_neg: 100.000 time: 0.1976 data: 0.1002 max mem: 3531 +Test: [130/269] eta: 0:00:30 acc1_pos: 34.483 acc5_pos: 100.000 loss: 0.5286 (0.5441) acc1: 74.7396 (74.2486) acc5: 100.0000 (100.0000) acc1_neg: 58.594 acc5_neg: 100.000 time: 0.2022 data: 0.1057 max mem: 3531 +Test: [140/269] eta: 0:00:28 acc1_pos: 34.483 acc5_pos: 100.000 loss: 0.6387 (0.5540) acc1: 58.5938 (72.8594) acc5: 100.0000 (100.0000) acc1_neg: 50.000 acc5_neg: 100.000 time: 0.2111 data: 0.1184 max mem: 3531 +Test: [150/269] eta: 0:00:26 acc1_pos: 34.483 acc5_pos: 100.000 loss: 0.6751 (0.5589) acc1: 54.9479 (72.1803) acc5: 100.0000 (100.0000) acc1_neg: 71.094 acc5_neg: 100.000 time: 0.2049 data: 0.1100 max mem: 3531 +Test: [160/269] eta: 0:00:23 acc1_pos: 34.483 acc5_pos: 100.000 loss: 0.5145 (0.5558) acc1: 74.7396 (72.6288) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.2062 data: 0.1112 max mem: 3531 +Test: [170/269] eta: 0:00:21 acc1_pos: 34.483 acc5_pos: 100.000 loss: 0.5051 (0.5525) acc1: 81.2500 (73.1101) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.2150 data: 0.1224 max mem: 3531 +Test: [180/269] eta: 0:00:19 acc1_pos: 34.483 acc5_pos: 100.000 loss: 0.4464 (0.5452) acc1: 88.8021 (74.0605) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.2006 data: 0.1092 max mem: 3531 +Test: [190/269] eta: 0:00:17 acc1_pos: 34.483 acc5_pos: 100.000 loss: 0.4353 (0.5436) acc1: 86.1979 (74.2924) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1890 data: 0.0982 max mem: 3531 +Test: [200/269] eta: 0:00:14 acc1_pos: 34.483 acc5_pos: 100.000 loss: 0.5242 (0.5421) acc1: 76.8229 (74.5116) acc5: 100.0000 (100.0000) acc1_neg: 71.354 acc5_neg: 100.000 time: 0.1930 data: 0.0988 max mem: 3531 +Test: [210/269] eta: 0:00:12 acc1_pos: 34.483 acc5_pos: 100.000 loss: 0.5295 (0.5428) acc1: 74.4792 (74.4199) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.1958 data: 0.0980 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 34.483 acc5_pos: 100.000 loss: 0.5238 (0.5434) acc1: 74.7396 (74.3484) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1964 data: 0.1016 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 34.483 acc5_pos: 100.000 loss: 0.4553 (0.5393) acc1: 84.8958 (74.7734) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.2032 data: 0.1106 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 34.483 acc5_pos: 100.000 loss: 0.4963 (0.5397) acc1: 82.0312 (74.8541) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.2147 data: 0.1218 max mem: 3532 +Test: [250/269] eta: 0:00:04 acc1_pos: 34.483 acc5_pos: 100.000 loss: 0.5545 (0.5425) acc1: 71.8750 (74.4781) acc5: 100.0000 (100.0000) acc1_neg: 49.740 acc5_neg: 100.000 time: 0.1887 data: 0.0971 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 34.483 acc5_pos: 100.000 loss: 0.6244 (0.5454) acc1: 64.5833 (74.1290) acc5: 100.0000 (100.0000) acc1_neg: 77.083 acc5_neg: 100.000 time: 0.1430 data: 0.0523 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 34.483 acc5_pos: 100.000 loss: 0.5314 (0.5434) acc1: 76.8229 (74.3884) acc5: 100.0000 (100.0000) acc1_neg: 88.189 acc5_neg: 100.000 time: 0.1343 data: 0.0405 max mem: 3532 +Test: Total time: 0:00:55 (0.2070 s / it) +* Overall Acc@1 74.388 Acc@5 100.000 + target == real Acc@1 52.854 Acc@5 100.000 + target == fake Acc@1 77.327 Acc@5 100.000 + loss 0.543 + PR-AUC 0.9427 ROC-AUC 0.7221 EER 34.03% (thr=0.5630) +Accuracy of the network on the 103293 test images: 74.4% +[rank0]:[W726 05:11:20.176650792 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T05:11:26Z] eval_done tag=fixed500k_fake_n64_seed7_resampled +[2026-07-26T05:11:27Z] eval_start tag=fixed500k_real_n32_seed8_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n32_classseed8_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T05:11:27Z +hostname=n116-040-213 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n32_classseed8_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n32_seed8_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 05:11:40.418000 1195791 torch/distributed/run.py:766] +W0726 05:11:40.418000 1195791 torch/distributed/run.py:766] ***************************************** +W0726 05:11:40.418000 1195791 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:11:40.418000 1195791 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:12:08.170849357 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:12:09.651712472 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:12:09.747408047 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:12:09.994810723 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:12:09.028445585 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:12:09.166938641 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:12:10.194717354 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:12:10.197153960 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n32_seed8_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n32_classseed8_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:23:16 acc1_pos: 46.875 acc5_pos: 100.000 loss: 0.7345 (0.7345) acc1: 46.8750 (46.8750) acc5: 100.0000 (100.0000) time: 5.9690 data: 4.1264 max mem: 3531 +Test: [ 10/234] eta: 0:02:20 acc1_pos: 53.385 acc5_pos: 100.000 loss: 0.6320 (0.6768) acc1: 63.8021 (61.4583) acc5: 100.0000 (100.0000) time: 0.6284 data: 0.3755 max mem: 3531 +Test: [ 20/234] eta: 0:01:24 acc1_pos: 96.615 acc5_pos: 100.000 loss: 0.6320 (0.6986) acc1: 53.3854 (57.5769) acc5: 100.0000 (100.0000) time: 0.1154 data: 0.0214 max mem: 3531 +Test: [ 30/234] eta: 0:01:10 acc1_pos: 77.344 acc5_pos: 100.000 loss: 0.1898 (0.5439) acc1: 96.3542 (69.4640) acc5: 100.0000 (100.0000) time: 0.1901 data: 0.0966 max mem: 3531 +Test: [ 40/234] eta: 0:01:01 acc1_pos: 94.531 acc5_pos: 100.000 loss: 0.2202 (0.5073) acc1: 94.0104 (73.0755) acc5: 100.0000 (100.0000) time: 0.2349 data: 0.1413 max mem: 3531 +Test: [ 50/234] eta: 0:00:54 acc1_pos: 94.792 acc5_pos: 100.000 loss: 0.2353 (0.4524) acc1: 93.2292 (77.2569) acc5: 100.0000 (100.0000) time: 0.2129 data: 0.1202 max mem: 3531 +Test: [ 60/234] eta: 0:00:48 acc1_pos: 90.345 acc5_pos: 100.000 loss: 0.1642 (0.3947) acc1: 96.3542 (80.9255) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2021 data: 0.1109 max mem: 3531 +Test: [ 70/234] eta: 0:00:44 acc1_pos: 90.345 acc5_pos: 100.000 loss: 0.0959 (0.3940) acc1: 100.0000 (80.9236) acc5: 100.0000 (100.0000) acc1_neg: 70.052 acc5_neg: 100.000 time: 0.2057 data: 0.1148 max mem: 3531 +Test: [ 80/234] eta: 0:00:39 acc1_pos: 90.345 acc5_pos: 100.000 loss: 0.5114 (0.3966) acc1: 72.6562 (80.9896) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1979 data: 0.1050 max mem: 3531 +Test: [ 90/234] eta: 0:00:36 acc1_pos: 90.345 acc5_pos: 100.000 loss: 0.2438 (0.3729) acc1: 92.1875 (82.5864) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1904 data: 0.0969 max mem: 3531 +Test: [100/234] eta: 0:00:32 acc1_pos: 90.345 acc5_pos: 100.000 loss: 0.2452 (0.3755) acc1: 89.8438 (82.4876) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1948 data: 0.1005 max mem: 3531 +Test: [110/234] eta: 0:00:29 acc1_pos: 90.345 acc5_pos: 100.000 loss: 0.3615 (0.3710) acc1: 82.2917 (82.8524) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.1960 data: 0.1022 max mem: 3531 +Test: [120/234] eta: 0:00:27 acc1_pos: 90.345 acc5_pos: 100.000 loss: 0.3322 (0.3765) acc1: 82.2917 (82.5564) acc5: 100.0000 (100.0000) acc1_neg: 77.344 acc5_neg: 100.000 time: 0.1899 data: 0.0959 max mem: 3531 +Test: [130/234] eta: 0:00:24 acc1_pos: 90.345 acc5_pos: 100.000 loss: 0.3905 (0.3777) acc1: 80.7292 (82.7389) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.1888 data: 0.0947 max mem: 3531 +Test: [140/234] eta: 0:00:21 acc1_pos: 90.345 acc5_pos: 100.000 loss: 0.4492 (0.3985) acc1: 78.1250 (81.7284) acc5: 100.0000 (100.0000) acc1_neg: 51.562 acc5_neg: 100.000 time: 0.1852 data: 0.0941 max mem: 3531 +Test: [150/234] eta: 0:00:18 acc1_pos: 90.345 acc5_pos: 100.000 loss: 0.3875 (0.4022) acc1: 79.4271 (81.5466) acc5: 100.0000 (100.0000) acc1_neg: 79.427 acc5_neg: 100.000 time: 0.1701 data: 0.0787 max mem: 3531 +Test: [160/234] eta: 0:00:16 acc1_pos: 90.345 acc5_pos: 100.000 loss: 0.3680 (0.4013) acc1: 82.0312 (81.7902) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.1749 data: 0.0805 max mem: 3531 +Test: [170/234] eta: 0:00:14 acc1_pos: 90.345 acc5_pos: 100.000 loss: 0.4374 (0.4037) acc1: 81.7708 (81.7922) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.1787 data: 0.0831 max mem: 3531 +Test: [180/234] eta: 0:00:11 acc1_pos: 90.345 acc5_pos: 100.000 loss: 0.4437 (0.4093) acc1: 81.2500 (81.3378) acc5: 100.0000 (100.0000) acc1_neg: 51.042 acc5_neg: 100.000 time: 0.1758 data: 0.0825 max mem: 3531 +Test: [190/234] eta: 0:00:09 acc1_pos: 90.345 acc5_pos: 100.000 loss: 0.6415 (0.4223) acc1: 65.1042 (80.4374) acc5: 100.0000 (100.0000) acc1_neg: 66.146 acc5_neg: 100.000 time: 0.1759 data: 0.0821 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 90.345 acc5_pos: 100.000 loss: 0.6317 (0.4166) acc1: 65.8854 (80.7616) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0869 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 90.345 acc5_pos: 100.000 loss: 0.1066 (0.4027) acc1: 99.7396 (81.6585) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1761 data: 0.0833 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 90.345 acc5_pos: 100.000 loss: 0.1503 (0.3923) acc1: 99.4792 (82.4060) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1490 data: 0.0552 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 90.345 acc5_pos: 100.000 loss: 0.3189 (0.4031) acc1: 67.1875 (81.5566) acc5: 100.0000 (100.0000) acc1_neg: 62.760 acc5_neg: 100.000 time: 0.1317 data: 0.0403 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 90.345 acc5_pos: 100.000 loss: 0.6242 (0.4062) acc1: 63.0350 (81.3204) acc5: 100.0000 (100.0000) acc1_neg: 63.035 acc5_neg: 100.000 time: 0.1316 data: 0.0403 max mem: 3532 +Test: Total time: 0:00:47 (0.2037 s / it) +* Overall Acc@1 81.320 Acc@5 100.000 + target == real Acc@1 77.353 Acc@5 100.000 + target == fake Acc@1 82.439 Acc@5 100.000 + loss 0.406 + PR-AUC 0.9670 ROC-AUC 0.8931 EER 19.87% (thr=0.5366) +Accuracy of the network on the 89729 test images: 81.3% +[rank0]:[W726 05:13:07.556393295 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:13:26.736000 1196349 torch/distributed/run.py:766] +W0726 05:13:26.736000 1196349 torch/distributed/run.py:766] ***************************************** +W0726 05:13:26.736000 1196349 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:13:26.736000 1196349 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:13:56.202794761 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:13:56.512030896 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:13:56.838443593 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:13:56.975771753 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:13:56.058874223 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:13:57.184537029 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:13:57.194281704 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:13:57.298171203 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n32_seed8_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n32_classseed8_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:23:26 acc1_pos: 34.635 acc5_pos: 100.000 loss: 1.0795 (1.0795) acc1: 34.6354 (34.6354) acc5: 100.0000 (100.0000) time: 6.6220 data: 4.7843 max mem: 3531 +Test: [ 10/2387] eta: 0:27:08 acc1_pos: 21.615 acc5_pos: 100.000 loss: 1.0795 (1.1239) acc1: 29.1667 (29.3797) acc5: 100.0000 (100.0000) time: 0.6850 data: 0.4352 max mem: 3531 +Test: [ 20/2387] eta: 0:15:54 acc1_pos: 26.042 acc5_pos: 100.000 loss: 1.0737 (1.0904) acc1: 26.0417 (30.9772) acc5: 100.0000 (100.0000) time: 0.0924 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:13:18 acc1_pos: 10.417 acc5_pos: 100.000 loss: 1.1429 (1.1329) acc1: 22.3958 (28.4442) acc5: 100.0000 (100.0000) time: 0.1482 data: 0.0541 max mem: 3531 +Test: [ 40/2387] eta: 0:13:08 acc1_pos: 11.458 acc5_pos: 100.000 loss: 1.2650 (1.1854) acc1: 11.4583 (24.2505) acc5: 100.0000 (100.0000) time: 0.2649 data: 0.1686 max mem: 3531 +Test: [ 50/2387] eta: 0:12:03 acc1_pos: 37.240 acc5_pos: 100.000 loss: 1.2495 (1.1721) acc1: 15.3646 (24.4587) acc5: 100.0000 (100.0000) time: 0.2649 data: 0.1686 max mem: 3531 +Test: [ 60/2387] eta: 0:11:21 acc1_pos: 41.146 acc5_pos: 100.000 loss: 1.1086 (1.1679) acc1: 24.7396 (24.7609) acc5: 100.0000 (100.0000) time: 0.2049 data: 0.1094 max mem: 3531 +Test: [ 70/2387] eta: 0:10:46 acc1_pos: 35.156 acc5_pos: 100.000 loss: 1.1930 (1.1768) acc1: 24.2188 (24.1894) acc5: 100.0000 (100.0000) time: 0.2006 data: 0.1052 max mem: 3531 +Test: [ 80/2387] eta: 0:10:23 acc1_pos: 19.531 acc5_pos: 100.000 loss: 1.2537 (1.1893) acc1: 16.6667 (23.2510) acc5: 100.0000 (100.0000) time: 0.2017 data: 0.1087 max mem: 3531 +Test: [ 90/2387] eta: 0:10:14 acc1_pos: 27.604 acc5_pos: 100.000 loss: 1.2537 (1.1925) acc1: 16.6667 (22.8251) acc5: 100.0000 (100.0000) time: 0.2277 data: 0.1361 max mem: 3531 +Test: [ 100/2387] eta: 0:09:57 acc1_pos: 11.719 acc5_pos: 100.000 loss: 1.2607 (1.2094) acc1: 11.9792 (21.6378) acc5: 100.0000 (100.0000) time: 0.2253 data: 0.1300 max mem: 3531 +Test: [ 110/2387] eta: 0:09:46 acc1_pos: 24.219 acc5_pos: 100.000 loss: 1.2441 (1.2112) acc1: 14.3229 (21.3987) acc5: 100.0000 (100.0000) time: 0.2119 data: 0.1173 max mem: 3531 +Test: [ 120/2387] eta: 0:09:37 acc1_pos: 22.135 acc5_pos: 100.000 loss: 1.2229 (1.2133) acc1: 20.0521 (21.3456) acc5: 100.0000 (100.0000) time: 0.2195 data: 0.1288 max mem: 3531 +Test: [ 130/2387] eta: 0:09:27 acc1_pos: 21.354 acc5_pos: 100.000 loss: 1.2192 (1.2196) acc1: 17.4479 (20.9288) acc5: 100.0000 (100.0000) time: 0.2180 data: 0.1251 max mem: 3531 +Test: [ 140/2387] eta: 0:09:20 acc1_pos: 26.042 acc5_pos: 100.000 loss: 1.2475 (1.2232) acc1: 17.1875 (20.7761) acc5: 100.0000 (100.0000) time: 0.2188 data: 0.1219 max mem: 3531 +Test: [ 150/2387] eta: 0:09:15 acc1_pos: 5.208 acc5_pos: 100.000 loss: 1.2428 (1.2226) acc1: 19.0104 (20.7471) acc5: 100.0000 (100.0000) time: 0.2258 data: 0.1288 max mem: 3531 +Test: [ 160/2387] eta: 0:09:02 acc1_pos: 4.566 acc5_pos: 100.000 loss: 1.1633 (1.2004) acc1: 19.2708 (22.4217) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2009 data: 0.1067 max mem: 3531 +Test: [ 170/2387] eta: 0:08:48 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.2248 (1.1430) acc1: 96.3542 (26.7940) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1654 data: 0.0713 max mem: 3531 +Test: [ 180/2387] eta: 0:08:34 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.2163 (1.0917) acc1: 97.3958 (30.6990) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1503 data: 0.0576 max mem: 3531 +Test: [ 190/2387] eta: 0:08:23 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.2121 (1.0456) acc1: 97.6562 (34.1950) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1489 data: 0.0578 max mem: 3531 +Test: [ 200/2387] eta: 0:08:16 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.2114 (1.0042) acc1: 97.6562 (37.3432) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1729 data: 0.0800 max mem: 3531 +Test: [ 210/2387] eta: 0:08:11 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.2148 (0.9670) acc1: 97.3958 (40.1634) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1944 data: 0.0990 max mem: 3532 +Test: [ 220/2387] eta: 0:08:07 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.2156 (0.9330) acc1: 97.1354 (42.7472) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2007 data: 0.1055 max mem: 3532 +Test: [ 230/2387] eta: 0:08:01 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.2184 (0.9021) acc1: 97.1354 (45.1085) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1949 data: 0.1012 max mem: 3532 +Test: [ 240/2387] eta: 0:07:53 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.2045 (0.8724) acc1: 97.9167 (47.3386) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1758 data: 0.0827 max mem: 3532 +Test: [ 250/2387] eta: 0:07:48 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1758 (0.8418) acc1: 99.4792 (49.4325) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1720 data: 0.0798 max mem: 3532 +Test: [ 260/2387] eta: 0:07:44 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.0917 (0.8130) acc1: 100.0000 (51.3699) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1881 data: 0.0960 max mem: 3532 +Test: [ 270/2387] eta: 0:07:37 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.0910 (0.7864) acc1: 100.0000 (53.1634) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1811 data: 0.0887 max mem: 3532 +Test: [ 280/2387] eta: 0:07:34 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.0910 (0.7617) acc1: 100.0000 (54.8302) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1813 data: 0.0889 max mem: 3532 +Test: [ 290/2387] eta: 0:07:29 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.0853 (0.7385) acc1: 100.0000 (56.3824) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1916 data: 0.0996 max mem: 3532 +Test: [ 300/2387] eta: 0:07:25 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.0841 (0.7168) acc1: 100.0000 (57.8315) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1845 data: 0.0928 max mem: 3532 +Test: [ 310/2387] eta: 0:07:23 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.0880 (0.7010) acc1: 100.0000 (59.1112) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1957 data: 0.1038 max mem: 3532 +Test: [ 320/2387] eta: 0:07:22 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.3986 (0.6918) acc1: 95.3125 (60.2098) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2197 data: 0.1276 max mem: 3532 +Test: [ 330/2387] eta: 0:07:20 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.3993 (0.6829) acc1: 94.7917 (61.2805) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2296 data: 0.1356 max mem: 3532 +Test: [ 340/2387] eta: 0:07:19 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.3955 (0.6745) acc1: 95.5729 (62.2801) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2266 data: 0.1297 max mem: 3532 +Test: [ 350/2387] eta: 0:07:16 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.3893 (0.6604) acc1: 96.3542 (63.3391) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2158 data: 0.1213 max mem: 3532 +Test: [ 360/2387] eta: 0:07:13 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1681 (0.6469) acc1: 99.4792 (64.3236) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2037 data: 0.1120 max mem: 3532 +Test: [ 370/2387] eta: 0:07:11 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1785 (0.6343) acc1: 98.9583 (65.2495) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2034 data: 0.1112 max mem: 3532 +Test: [ 380/2387] eta: 0:07:07 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1787 (0.6223) acc1: 98.4375 (66.1205) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1961 data: 0.1037 max mem: 3532 +Test: [ 390/2387] eta: 0:07:05 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1768 (0.6109) acc1: 98.4375 (66.9524) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1949 data: 0.1017 max mem: 3532 +Test: [ 400/2387] eta: 0:07:01 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1768 (0.6002) acc1: 98.6979 (67.7421) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1979 data: 0.1053 max mem: 3532 +Test: [ 410/2387] eta: 0:06:59 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1726 (0.5898) acc1: 98.6979 (68.4953) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1955 data: 0.1028 max mem: 3532 +Test: [ 420/2387] eta: 0:06:55 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1737 (0.5800) acc1: 98.6979 (69.2152) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1923 data: 0.1000 max mem: 3532 +Test: [ 430/2387] eta: 0:06:53 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1722 (0.5704) acc1: 98.9583 (69.9053) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1910 data: 0.1000 max mem: 3532 +Test: [ 440/2387] eta: 0:06:49 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1585 (0.5602) acc1: 99.2188 (70.5800) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1883 data: 0.0964 max mem: 3533 +Test: [ 450/2387] eta: 0:06:46 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.0951 (0.5499) acc1: 100.0000 (71.2323) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1805 data: 0.0892 max mem: 3533 +Test: [ 460/2387] eta: 0:06:43 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.0952 (0.5400) acc1: 100.0000 (71.8564) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1854 data: 0.0945 max mem: 3533 +Test: [ 470/2387] eta: 0:06:40 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.0956 (0.5306) acc1: 100.0000 (72.4539) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1885 data: 0.0969 max mem: 3533 +Test: [ 480/2387] eta: 0:06:37 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.0957 (0.5215) acc1: 100.0000 (73.0266) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1839 data: 0.0923 max mem: 3533 +Test: [ 490/2387] eta: 0:06:33 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.0955 (0.5129) acc1: 100.0000 (73.5759) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1812 data: 0.0895 max mem: 3533 +Test: [ 500/2387] eta: 0:06:31 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.0953 (0.5045) acc1: 100.0000 (74.1034) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1871 data: 0.0957 max mem: 3533 +Test: [ 510/2387] eta: 0:06:28 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.0950 (0.4966) acc1: 100.0000 (74.6101) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1910 data: 0.1001 max mem: 3533 +Test: [ 520/2387] eta: 0:06:25 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1123 (0.4893) acc1: 100.0000 (75.0975) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1822 data: 0.0910 max mem: 3533 +Test: [ 530/2387] eta: 0:06:22 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1167 (0.4823) acc1: 100.0000 (75.5664) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1766 data: 0.0846 max mem: 3533 +Test: [ 540/2387] eta: 0:06:19 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1181 (0.4756) acc1: 100.0000 (76.0181) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1797 data: 0.0871 max mem: 3533 +Test: [ 550/2387] eta: 0:06:15 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1181 (0.4691) acc1: 100.0000 (76.4533) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1702 data: 0.0786 max mem: 3533 +Test: [ 560/2387] eta: 0:06:12 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1182 (0.4629) acc1: 100.0000 (76.8731) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1630 data: 0.0721 max mem: 3533 +Test: [ 570/2387] eta: 0:06:09 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1186 (0.4568) acc1: 100.0000 (77.2781) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1748 data: 0.0837 max mem: 3533 +Test: [ 580/2387] eta: 0:06:07 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1186 (0.4510) acc1: 100.0000 (77.6692) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1868 data: 0.0955 max mem: 3533 +Test: [ 590/2387] eta: 0:06:04 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1189 (0.4454) acc1: 100.0000 (78.0470) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1763 data: 0.0840 max mem: 3533 +Test: [ 600/2387] eta: 0:06:01 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1155 (0.4399) acc1: 100.0000 (78.4123) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1758 data: 0.0826 max mem: 3533 +Test: [ 610/2387] eta: 0:05:59 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1262 (0.4352) acc1: 100.0000 (78.7622) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1889 data: 0.0960 max mem: 3533 +Test: [ 620/2387] eta: 0:05:56 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1535 (0.4307) acc1: 99.7396 (79.0987) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1749 data: 0.0820 max mem: 3533 +Test: [ 630/2387] eta: 0:05:53 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1546 (0.4263) acc1: 99.7396 (79.4254) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1653 data: 0.0716 max mem: 3533 +Test: [ 640/2387] eta: 0:05:50 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1554 (0.4221) acc1: 99.7396 (79.7407) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1761 data: 0.0831 max mem: 3533 +Test: [ 650/2387] eta: 0:05:48 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1561 (0.4180) acc1: 99.7396 (80.0451) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1853 data: 0.0937 max mem: 3533 +Test: [ 660/2387] eta: 0:05:45 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1553 (0.4140) acc1: 99.7396 (80.3419) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1782 data: 0.0873 max mem: 3533 +Test: [ 670/2387] eta: 0:05:42 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1519 (0.4101) acc1: 99.7396 (80.6294) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1658 data: 0.0739 max mem: 3534 +Test: [ 680/2387] eta: 0:05:40 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1500 (0.4060) acc1: 99.7396 (80.9131) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1721 data: 0.0798 max mem: 3534 +Test: [ 690/2387] eta: 0:05:37 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1481 (0.4045) acc1: 99.7396 (81.0137) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.1850 data: 0.0934 max mem: 3534 +Test: [ 700/2387] eta: 0:05:35 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.3574 (0.4040) acc1: 83.0729 (81.0390) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.1892 data: 0.0975 max mem: 3534 +Test: [ 710/2387] eta: 0:05:33 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.3624 (0.4034) acc1: 83.0729 (81.0669) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.1925 data: 0.0979 max mem: 3534 +Test: [ 720/2387] eta: 0:05:31 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.3605 (0.4028) acc1: 83.5938 (81.1023) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.1910 data: 0.0955 max mem: 3534 +Test: [ 730/2387] eta: 0:05:28 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.3647 (0.4024) acc1: 83.5938 (81.1257) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.1859 data: 0.0911 max mem: 3534 +Test: [ 740/2387] eta: 0:05:26 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.3658 (0.4019) acc1: 82.8125 (81.1558) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.1878 data: 0.0924 max mem: 3534 +Test: [ 750/2387] eta: 0:05:24 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.3608 (0.4014) acc1: 82.8125 (81.1865) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.1890 data: 0.0921 max mem: 3534 +Test: [ 760/2387] eta: 0:05:22 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.3565 (0.4000) acc1: 83.5938 (81.2740) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1866 data: 0.0903 max mem: 3534 +Test: [ 770/2387] eta: 0:05:19 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1961 (0.3970) acc1: 93.7500 (81.4635) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1817 data: 0.0896 max mem: 3534 +Test: [ 780/2387] eta: 0:05:17 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1644 (0.3937) acc1: 96.8750 (81.6821) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1896 data: 0.0973 max mem: 3534 +Test: [ 790/2387] eta: 0:05:15 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1336 (0.3909) acc1: 98.9583 (81.8492) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1953 data: 0.1030 max mem: 3534 +Test: [ 800/2387] eta: 0:05:13 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1287 (0.3876) acc1: 99.4792 (82.0660) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1874 data: 0.0952 max mem: 3534 +Test: [ 810/2387] eta: 0:05:10 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1228 (0.3844) acc1: 99.4792 (82.2779) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1732 data: 0.0798 max mem: 3534 +Test: [ 820/2387] eta: 0:05:08 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1228 (0.3813) acc1: 99.7396 (82.4820) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1670 data: 0.0750 max mem: 3534 +Test: [ 830/2387] eta: 0:05:06 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1396 (0.3788) acc1: 98.9583 (82.6589) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1867 data: 0.0938 max mem: 3534 +Test: [ 840/2387] eta: 0:05:04 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.2193 (0.3774) acc1: 96.6146 (82.8063) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.1847 data: 0.0893 max mem: 3534 +Test: [ 850/2387] eta: 0:05:01 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.2606 (0.3761) acc1: 95.5729 (82.9438) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1631 data: 0.0689 max mem: 3534 +Test: [ 860/2387] eta: 0:04:59 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.2531 (0.3746) acc1: 96.3542 (83.1083) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1693 data: 0.0773 max mem: 3534 +Test: [ 870/2387] eta: 0:04:56 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.2425 (0.3730) acc1: 98.1771 (83.2744) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1740 data: 0.0827 max mem: 3534 +Test: [ 880/2387] eta: 0:04:53 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.2178 (0.3716) acc1: 98.1771 (83.4170) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1504 data: 0.0571 max mem: 3534 +Test: [ 890/2387] eta: 0:04:51 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1919 (0.3694) acc1: 99.2188 (83.5926) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1581 data: 0.0631 max mem: 3534 +Test: [ 900/2387] eta: 0:04:49 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1996 (0.3708) acc1: 97.9167 (83.4981) acc5: 100.0000 (100.0000) acc1_neg: 67.969 acc5_neg: 100.000 time: 0.1736 data: 0.0804 max mem: 3535 +Test: [ 910/2387] eta: 0:04:46 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.5712 (0.3730) acc1: 69.0104 (83.3433) acc5: 100.0000 (100.0000) acc1_neg: 69.010 acc5_neg: 100.000 time: 0.1630 data: 0.0691 max mem: 3535 +Test: [ 920/2387] eta: 0:04:44 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.5626 (0.3752) acc1: 69.7917 (83.1897) acc5: 100.0000 (100.0000) acc1_neg: 69.792 acc5_neg: 100.000 time: 0.1677 data: 0.0727 max mem: 3535 +Test: [ 930/2387] eta: 0:04:42 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.5704 (0.3772) acc1: 69.7917 (83.0455) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.1713 data: 0.0790 max mem: 3535 +Test: [ 940/2387] eta: 0:04:39 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.5783 (0.3793) acc1: 68.7500 (82.8930) acc5: 100.0000 (100.0000) acc1_neg: 67.708 acc5_neg: 100.000 time: 0.1697 data: 0.0778 max mem: 3535 +Test: [ 950/2387] eta: 0:04:37 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.5791 (0.3813) acc1: 69.0104 (82.7476) acc5: 100.0000 (100.0000) acc1_neg: 69.010 acc5_neg: 100.000 time: 0.1599 data: 0.0679 max mem: 3535 +Test: [ 960/2387] eta: 0:04:34 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.5585 (0.3831) acc1: 69.5312 (82.6158) acc5: 100.0000 (100.0000) acc1_neg: 72.396 acc5_neg: 100.000 time: 0.1587 data: 0.0671 max mem: 3535 +Test: [ 970/2387] eta: 0:04:32 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.5656 (0.3851) acc1: 69.2708 (82.4722) acc5: 100.0000 (100.0000) acc1_neg: 68.750 acc5_neg: 100.000 time: 0.1641 data: 0.0703 max mem: 3535 +Test: [ 980/2387] eta: 0:04:30 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.5479 (0.3845) acc1: 70.0521 (82.5120) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.1684 data: 0.0752 max mem: 3535 +Test: [ 990/2387] eta: 0:04:27 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1729 (0.3825) acc1: 98.1771 (82.6325) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1557 data: 0.0642 max mem: 3535 +Test: [1000/2387] eta: 0:04:25 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1236 (0.3800) acc1: 99.2188 (82.7849) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1507 data: 0.0587 max mem: 3535 +Test: [1010/2387] eta: 0:04:23 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1289 (0.3777) acc1: 99.2188 (82.9423) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1771 data: 0.0825 max mem: 3535 +Test: [1020/2387] eta: 0:04:20 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1338 (0.3755) acc1: 98.9583 (83.0834) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1704 data: 0.0752 max mem: 3535 +Test: [1030/2387] eta: 0:04:18 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1317 (0.3732) acc1: 99.4792 (83.2363) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1449 data: 0.0519 max mem: 3535 +Test: [1040/2387] eta: 0:04:15 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1303 (0.3708) acc1: 99.7396 (83.3929) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1446 data: 0.0521 max mem: 3535 +Test: [1050/2387] eta: 0:04:13 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1154 (0.3685) acc1: 100.0000 (83.5469) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1549 data: 0.0608 max mem: 3535 +Test: [1060/2387] eta: 0:04:11 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1140 (0.3661) acc1: 99.7396 (83.6978) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1592 data: 0.0659 max mem: 3535 +Test: [1070/2387] eta: 0:04:09 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1166 (0.3639) acc1: 99.7396 (83.8459) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1569 data: 0.0655 max mem: 3535 +Test: [1080/2387] eta: 0:04:06 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1475 (0.3622) acc1: 99.2188 (83.9688) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1554 data: 0.0635 max mem: 3535 +Test: [1090/2387] eta: 0:04:04 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1706 (0.3605) acc1: 97.6562 (84.0843) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1679 data: 0.0760 max mem: 3535 +Test: [1100/2387] eta: 0:04:02 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1710 (0.3589) acc1: 97.1354 (84.2061) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1686 data: 0.0750 max mem: 3535 +Test: [1110/2387] eta: 0:03:59 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1801 (0.3577) acc1: 97.1354 (84.2803) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1400 data: 0.0444 max mem: 3535 +Test: [1120/2387] eta: 0:03:57 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.2014 (0.3563) acc1: 96.3542 (84.3901) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1469 data: 0.0520 max mem: 3536 +Test: [1130/2387] eta: 0:03:56 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1641 (0.3545) acc1: 97.9167 (84.5141) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1871 data: 0.0940 max mem: 3536 +Test: [1140/2387] eta: 0:03:54 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1590 (0.3529) acc1: 98.4375 (84.6354) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1977 data: 0.1052 max mem: 3536 +Test: [1150/2387] eta: 0:03:52 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1754 (0.3515) acc1: 98.6979 (84.7461) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1787 data: 0.0854 max mem: 3536 +Test: [1160/2387] eta: 0:03:50 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1569 (0.3500) acc1: 98.9583 (84.8590) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1722 data: 0.0796 max mem: 3536 +Test: [1170/2387] eta: 0:03:48 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1645 (0.3484) acc1: 99.2188 (84.9819) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1908 data: 0.0978 max mem: 3536 +Test: [1180/2387] eta: 0:03:46 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1741 (0.3470) acc1: 99.2188 (85.0947) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1804 data: 0.0868 max mem: 3536 +Test: [1190/2387] eta: 0:03:44 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1525 (0.3455) acc1: 99.2188 (85.2153) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1954 data: 0.1029 max mem: 3536 +Test: [1200/2387] eta: 0:03:43 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1352 (0.3437) acc1: 99.7396 (85.3377) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2221 data: 0.1300 max mem: 3536 +Test: [1210/2387] eta: 0:03:41 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1293 (0.3419) acc1: 100.0000 (85.4573) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2197 data: 0.1255 max mem: 3536 +Test: [1220/2387] eta: 0:03:39 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1102 (0.3399) acc1: 100.0000 (85.5751) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1649 data: 0.0706 max mem: 3536 +Test: [1230/2387] eta: 0:03:37 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.0962 (0.3379) acc1: 100.0000 (85.6919) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1603 data: 0.0685 max mem: 3536 +Test: [1240/2387] eta: 0:03:35 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.0973 (0.3360) acc1: 100.0000 (85.8068) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2114 data: 0.1200 max mem: 3536 +Test: [1250/2387] eta: 0:03:33 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.0944 (0.3341) acc1: 100.0000 (85.9196) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1943 data: 0.1024 max mem: 3536 +Test: [1260/2387] eta: 0:03:31 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.0860 (0.3321) acc1: 100.0000 (86.0311) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1760 data: 0.0820 max mem: 3536 +Test: [1270/2387] eta: 0:03:29 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.0835 (0.3302) acc1: 100.0000 (86.1408) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1727 data: 0.0796 max mem: 3536 +Test: [1280/2387] eta: 0:03:27 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.0856 (0.3283) acc1: 100.0000 (86.2473) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1681 data: 0.0744 max mem: 3536 +Test: [1290/2387] eta: 0:03:25 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.0982 (0.3267) acc1: 99.4792 (86.3339) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1754 data: 0.0817 max mem: 3536 +Test: [1300/2387] eta: 0:03:23 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1053 (0.3250) acc1: 99.2188 (86.4295) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1896 data: 0.0982 max mem: 3536 +Test: [1310/2387] eta: 0:03:21 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1053 (0.3235) acc1: 99.4792 (86.5219) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1746 data: 0.0834 max mem: 3536 +Test: [1320/2387] eta: 0:03:19 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1033 (0.3219) acc1: 99.4792 (86.6113) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1646 data: 0.0735 max mem: 3536 +Test: [1330/2387] eta: 0:03:18 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1012 (0.3204) acc1: 99.7396 (86.7041) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1908 data: 0.0978 max mem: 3536 +Test: [1340/2387] eta: 0:03:16 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1358 (0.3196) acc1: 98.4375 (86.7539) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1948 data: 0.1001 max mem: 3536 +Test: [1350/2387] eta: 0:03:13 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.0684 (0.3177) acc1: 100.0000 (86.8519) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1606 data: 0.0655 max mem: 3537 +Test: [1360/2387] eta: 0:03:11 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.0634 (0.3158) acc1: 100.0000 (86.9486) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1597 data: 0.0645 max mem: 3537 +Test: [1370/2387] eta: 0:03:10 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.0635 (0.3140) acc1: 100.0000 (87.0436) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1905 data: 0.0950 max mem: 3537 +Test: [1380/2387] eta: 0:03:08 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.0635 (0.3122) acc1: 100.0000 (87.1374) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2078 data: 0.1144 max mem: 3537 +Test: [1390/2387] eta: 0:03:06 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.0621 (0.3104) acc1: 100.0000 (87.2298) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2091 data: 0.1166 max mem: 3537 +Test: [1400/2387] eta: 0:03:05 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.0588 (0.3086) acc1: 100.0000 (87.3210) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2155 data: 0.1230 max mem: 3537 +Test: [1410/2387] eta: 0:03:03 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.0590 (0.3074) acc1: 100.0000 (87.3954) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1957 data: 0.1048 max mem: 3537 +Test: [1420/2387] eta: 0:03:01 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1568 (0.3066) acc1: 97.9167 (87.4491) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1908 data: 0.0992 max mem: 3537 +Test: [1430/2387] eta: 0:02:59 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1768 (0.3058) acc1: 97.6562 (87.5111) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2101 data: 0.1186 max mem: 3537 +Test: [1440/2387] eta: 0:02:57 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1630 (0.3047) acc1: 97.9167 (87.5929) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2040 data: 0.1128 max mem: 3537 +Test: [1450/2387] eta: 0:02:56 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1553 (0.3038) acc1: 99.4792 (87.6648) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1967 data: 0.1041 max mem: 3537 +Test: [1460/2387] eta: 0:02:54 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1424 (0.3027) acc1: 98.6979 (87.7422) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1840 data: 0.0918 max mem: 3537 +Test: [1470/2387] eta: 0:02:52 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1322 (0.3015) acc1: 99.4792 (87.8226) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1822 data: 0.0894 max mem: 3537 +Test: [1480/2387] eta: 0:02:50 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1235 (0.3003) acc1: 99.4792 (87.8983) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1917 data: 0.0985 max mem: 3537 +Test: [1490/2387] eta: 0:02:48 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1251 (0.2992) acc1: 99.4792 (87.9763) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1905 data: 0.0994 max mem: 3537 +Test: [1500/2387] eta: 0:02:46 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1251 (0.2980) acc1: 99.4792 (88.0522) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1853 data: 0.0897 max mem: 3537 +Test: [1510/2387] eta: 0:02:44 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1246 (0.2969) acc1: 100.0000 (88.1299) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1893 data: 0.0902 max mem: 3537 +Test: [1520/2387] eta: 0:02:43 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1488 (0.2963) acc1: 99.4792 (88.1932) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1979 data: 0.1021 max mem: 3537 +Test: [1530/2387] eta: 0:02:41 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.2245 (0.2959) acc1: 97.6562 (88.2457) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1956 data: 0.1023 max mem: 3537 +Test: [1540/2387] eta: 0:02:39 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.2187 (0.2954) acc1: 98.4375 (88.3096) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1792 data: 0.0873 max mem: 3537 +Test: [1550/2387] eta: 0:02:37 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.2121 (0.2948) acc1: 98.6979 (88.3770) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1805 data: 0.0895 max mem: 3537 +Test: [1560/2387] eta: 0:02:35 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1905 (0.2942) acc1: 99.4792 (88.4354) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1795 data: 0.0880 max mem: 3537 +Test: [1570/2387] eta: 0:02:33 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1689 (0.2934) acc1: 99.7396 (88.5027) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1754 data: 0.0826 max mem: 3537 +Test: [1580/2387] eta: 0:02:31 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1458 (0.2925) acc1: 99.2188 (88.5685) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1899 data: 0.0965 max mem: 3538 +Test: [1590/2387] eta: 0:02:29 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1454 (0.2915) acc1: 98.9583 (88.6333) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2005 data: 0.1056 max mem: 3538 +Test: [1600/2387] eta: 0:02:27 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1454 (0.2908) acc1: 98.6979 (88.6685) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1904 data: 0.0946 max mem: 3538 +Test: [1610/2387] eta: 0:02:25 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1313 (0.2898) acc1: 98.9583 (88.7337) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1727 data: 0.0793 max mem: 3538 +Test: [1620/2387] eta: 0:02:23 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1313 (0.2889) acc1: 98.9583 (88.7932) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1544 data: 0.0622 max mem: 3538 +Test: [1630/2387] eta: 0:02:21 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1444 (0.2881) acc1: 98.4375 (88.8524) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1505 data: 0.0574 max mem: 3538 +Test: [1640/2387] eta: 0:02:19 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1581 (0.2874) acc1: 98.6979 (88.9063) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1666 data: 0.0739 max mem: 3538 +Test: [1650/2387] eta: 0:02:18 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1542 (0.2866) acc1: 98.4375 (88.9616) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1893 data: 0.0958 max mem: 3538 +Test: [1660/2387] eta: 0:02:16 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1563 (0.2859) acc1: 98.1771 (89.0130) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2136 data: 0.1204 max mem: 3538 +Test: [1670/2387] eta: 0:02:14 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1600 (0.2852) acc1: 98.4375 (89.0680) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2149 data: 0.1231 max mem: 3538 +Test: [1680/2387] eta: 0:02:12 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1565 (0.2845) acc1: 98.4375 (89.1167) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2102 data: 0.1193 max mem: 3538 +Test: [1690/2387] eta: 0:02:10 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1725 (0.2840) acc1: 97.3958 (89.1595) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1974 data: 0.1068 max mem: 3538 +Test: [1700/2387] eta: 0:02:08 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1684 (0.2831) acc1: 98.6979 (89.2182) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1843 data: 0.0919 max mem: 3538 +Test: [1710/2387] eta: 0:02:07 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1219 (0.2821) acc1: 99.4792 (89.2792) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0805 max mem: 3538 +Test: [1720/2387] eta: 0:02:05 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1036 (0.2811) acc1: 100.0000 (89.3400) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1656 data: 0.0723 max mem: 3538 +Test: [1730/2387] eta: 0:02:03 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1071 (0.2801) acc1: 100.0000 (89.3999) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1675 data: 0.0756 max mem: 3538 +Test: [1740/2387] eta: 0:02:01 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1044 (0.2791) acc1: 100.0000 (89.4598) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1709 data: 0.0775 max mem: 3538 +Test: [1750/2387] eta: 0:01:59 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.0951 (0.2781) acc1: 100.0000 (89.5195) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1727 data: 0.0793 max mem: 3538 +Test: [1760/2387] eta: 0:01:57 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.0918 (0.2770) acc1: 100.0000 (89.5790) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1782 data: 0.0861 max mem: 3538 +Test: [1770/2387] eta: 0:01:55 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.0979 (0.2762) acc1: 100.0000 (89.6294) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1769 data: 0.0848 max mem: 3538 +Test: [1780/2387] eta: 0:01:53 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1234 (0.2753) acc1: 99.2188 (89.6842) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1749 data: 0.0822 max mem: 3538 +Test: [1790/2387] eta: 0:01:51 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1188 (0.2744) acc1: 99.7396 (89.7383) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1734 data: 0.0795 max mem: 3538 +Test: [1800/2387] eta: 0:01:49 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1186 (0.2736) acc1: 99.4792 (89.7914) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1729 data: 0.0782 max mem: 3538 +Test: [1810/2387] eta: 0:01:47 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1046 (0.2727) acc1: 99.4792 (89.8462) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1766 data: 0.0835 max mem: 3539 +Test: [1820/2387] eta: 0:01:45 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.0956 (0.2717) acc1: 100.0000 (89.9017) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1724 data: 0.0808 max mem: 3539 +Test: [1830/2387] eta: 0:01:43 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.0918 (0.2707) acc1: 100.0000 (89.9567) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1678 data: 0.0766 max mem: 3539 +Test: [1840/2387] eta: 0:01:42 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1151 (0.2699) acc1: 100.0000 (90.0112) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1674 data: 0.0761 max mem: 3539 +Test: [1850/2387] eta: 0:01:40 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1158 (0.2691) acc1: 100.0000 (90.0651) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1677 data: 0.0764 max mem: 3539 +Test: [1860/2387] eta: 0:01:38 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1157 (0.2682) acc1: 100.0000 (90.1184) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1709 data: 0.0788 max mem: 3539 +Test: [1870/2387] eta: 0:01:36 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1147 (0.2674) acc1: 100.0000 (90.1713) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1774 data: 0.0852 max mem: 3539 +Test: [1880/2387] eta: 0:01:34 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1156 (0.2666) acc1: 100.0000 (90.2235) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1740 data: 0.0828 max mem: 3539 +Test: [1890/2387] eta: 0:01:32 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1157 (0.2658) acc1: 100.0000 (90.2752) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1715 data: 0.0797 max mem: 3539 +Test: [1900/2387] eta: 0:01:30 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1167 (0.2650) acc1: 100.0000 (90.3264) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1748 data: 0.0831 max mem: 3539 +Test: [1910/2387] eta: 0:01:28 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1189 (0.2643) acc1: 100.0000 (90.3752) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1800 data: 0.0885 max mem: 3539 +Test: [1920/2387] eta: 0:01:26 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1222 (0.2639) acc1: 100.0000 (90.4002) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1853 data: 0.0927 max mem: 3539 +Test: [1930/2387] eta: 0:01:25 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1383 (0.2633) acc1: 99.4792 (90.4477) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1888 data: 0.0957 max mem: 3539 +Test: [1940/2387] eta: 0:01:23 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1437 (0.2630) acc1: 99.4792 (90.4582) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1919 data: 0.0998 max mem: 3539 +Test: [1950/2387] eta: 0:01:21 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1675 (0.2626) acc1: 98.4375 (90.4926) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1991 data: 0.1071 max mem: 3539 +Test: [1960/2387] eta: 0:01:19 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1546 (0.2620) acc1: 98.9583 (90.5326) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2050 data: 0.1103 max mem: 3539 +Test: [1970/2387] eta: 0:01:17 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1593 (0.2618) acc1: 98.9583 (90.5587) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.2148 data: 0.1188 max mem: 3539 +Test: [1980/2387] eta: 0:01:16 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1670 (0.2613) acc1: 98.6979 (90.5924) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2300 data: 0.1349 max mem: 3539 +Test: [1990/2387] eta: 0:01:14 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1511 (0.2608) acc1: 99.2188 (90.6359) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2180 data: 0.1250 max mem: 3539 +Test: [2000/2387] eta: 0:01:12 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1519 (0.2603) acc1: 99.2188 (90.6790) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1847 data: 0.0929 max mem: 3539 +Test: [2010/2387] eta: 0:01:10 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1520 (0.2598) acc1: 99.2188 (90.7216) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1727 data: 0.0792 max mem: 3539 +Test: [2020/2387] eta: 0:01:08 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1549 (0.2592) acc1: 99.2188 (90.7635) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1729 data: 0.0786 max mem: 3539 +Test: [2030/2387] eta: 0:01:06 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1537 (0.2587) acc1: 99.4792 (90.8057) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1739 data: 0.0800 max mem: 3540 +Test: [2040/2387] eta: 0:01:04 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1535 (0.2582) acc1: 99.2188 (90.8470) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1715 data: 0.0783 max mem: 3540 +Test: [2050/2387] eta: 0:01:02 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1510 (0.2576) acc1: 99.2188 (90.8885) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1813 data: 0.0890 max mem: 3540 +Test: [2060/2387] eta: 0:01:00 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1421 (0.2571) acc1: 99.4792 (90.9300) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1884 data: 0.0964 max mem: 3540 +Test: [2070/2387] eta: 0:00:59 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1354 (0.2565) acc1: 99.7396 (90.9724) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1599 data: 0.0673 max mem: 3540 +Test: [2080/2387] eta: 0:00:57 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1354 (0.2559) acc1: 99.4792 (91.0137) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1542 data: 0.0610 max mem: 3540 +Test: [2090/2387] eta: 0:00:55 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1347 (0.2553) acc1: 99.4792 (91.0552) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1648 data: 0.0727 max mem: 3540 +Test: [2100/2387] eta: 0:00:53 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1347 (0.2548) acc1: 99.4792 (91.0955) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1670 data: 0.0759 max mem: 3540 +Test: [2110/2387] eta: 0:00:51 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1370 (0.2542) acc1: 99.4792 (91.1355) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1687 data: 0.0738 max mem: 3540 +Test: [2120/2387] eta: 0:00:49 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1370 (0.2538) acc1: 99.4792 (91.1628) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1683 data: 0.0739 max mem: 3540 +Test: [2130/2387] eta: 0:00:47 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1429 (0.2533) acc1: 99.4792 (91.2007) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1719 data: 0.0798 max mem: 3540 +Test: [2140/2387] eta: 0:00:45 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1454 (0.2530) acc1: 98.6979 (91.2130) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1663 data: 0.0727 max mem: 3540 +Test: [2150/2387] eta: 0:00:44 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1654 (0.2526) acc1: 98.4375 (91.2486) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1729 data: 0.0797 max mem: 3540 +Test: [2160/2387] eta: 0:00:42 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1362 (0.2520) acc1: 99.4792 (91.2849) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1835 data: 0.0909 max mem: 3540 +Test: [2170/2387] eta: 0:00:40 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1426 (0.2516) acc1: 99.2188 (91.3174) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1968 data: 0.1035 max mem: 3540 +Test: [2180/2387] eta: 0:00:38 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1847 (0.2526) acc1: 96.8750 (91.2491) acc5: 100.0000 (100.0000) acc1_neg: 48.958 acc5_neg: 100.000 time: 0.1866 data: 0.0909 max mem: 3540 +Test: [2190/2387] eta: 0:00:36 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1819 (0.2526) acc1: 95.8333 (91.2463) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1753 data: 0.0812 max mem: 3540 +Test: [2200/2387] eta: 0:00:34 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1388 (0.2521) acc1: 99.4792 (91.2847) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1806 data: 0.0871 max mem: 3540 +Test: [2210/2387] eta: 0:00:32 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1388 (0.2516) acc1: 99.7396 (91.3199) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1751 data: 0.0813 max mem: 3540 +Test: [2220/2387] eta: 0:00:30 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1417 (0.2511) acc1: 99.4792 (91.3535) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1751 data: 0.0820 max mem: 3540 +Test: [2230/2387] eta: 0:00:29 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1371 (0.2506) acc1: 99.7396 (91.3898) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1775 data: 0.0841 max mem: 3540 +Test: [2240/2387] eta: 0:00:27 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1251 (0.2500) acc1: 99.7396 (91.4231) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1828 data: 0.0914 max mem: 3540 +Test: [2250/2387] eta: 0:00:25 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1367 (0.2496) acc1: 99.7396 (91.4568) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1830 data: 0.0917 max mem: 3540 +Test: [2260/2387] eta: 0:00:23 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1425 (0.2491) acc1: 99.7396 (91.4928) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1790 data: 0.0866 max mem: 3541 +Test: [2270/2387] eta: 0:00:21 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1223 (0.2485) acc1: 100.0000 (91.5296) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1739 data: 0.0806 max mem: 3541 +Test: [2280/2387] eta: 0:00:19 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1223 (0.2482) acc1: 100.0000 (91.5454) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1766 data: 0.0846 max mem: 3541 +Test: [2290/2387] eta: 0:00:17 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1275 (0.2476) acc1: 100.0000 (91.5818) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1686 data: 0.0772 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1167 (0.2470) acc1: 100.0000 (91.6153) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1651 data: 0.0726 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1160 (0.2465) acc1: 99.4792 (91.6484) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1715 data: 0.0790 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1404 (0.2462) acc1: 99.4792 (91.6718) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1780 data: 0.0855 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1684 (0.2460) acc1: 98.1771 (91.6895) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1652 data: 0.0713 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1775 (0.2458) acc1: 96.8750 (91.7106) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1629 data: 0.0692 max mem: 3541 +Test: [2350/2387] eta: 0:00:06 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1866 (0.2459) acc1: 96.3542 (91.7067) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1734 data: 0.0817 max mem: 3541 +Test: [2360/2387] eta: 0:00:04 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1983 (0.2457) acc1: 96.3542 (91.7220) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1718 data: 0.0794 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1952 (0.2455) acc1: 97.1354 (91.7462) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1846 data: 0.0904 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1952 (0.2453) acc1: 97.6562 (91.7669) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1620 data: 0.0691 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 4.566 acc5_pos: 100.000 loss: 0.1943 (0.2451) acc1: 97.1354 (91.7767) acc5: 100.0000 (100.0000) acc1_neg: 95.349 acc5_neg: 100.000 time: 0.1394 data: 0.0505 max mem: 3541 +Test: Total time: 0:07:20 (0.1846 s / it) +* Overall Acc@1 91.777 Acc@5 100.000 + target == real Acc@1 20.387 Acc@5 100.000 + target == fake Acc@1 96.790 Acc@5 100.000 + loss 0.245 + PR-AUC 0.9889 ROC-AUC 0.8612 EER 21.76% (thr=0.8198) +Accuracy of the network on the 916267 test images: 91.8% +[rank0]:[W726 05:21:40.807864851 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:22:00.908000 1196969 torch/distributed/run.py:766] +W0726 05:22:00.908000 1196969 torch/distributed/run.py:766] ***************************************** +W0726 05:22:00.908000 1196969 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:22:00.908000 1196969 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:22:31.123944761 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:22:32.417904478 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:22:32.715269842 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:22:32.762505238 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:22:32.763192905 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:22:32.806282282 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:22:32.876155754 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:22:32.939045346 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n32_seed8_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n32_classseed8_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:01:05 acc1_pos: 34.635 acc5_pos: 100.000 loss: 1.0795 (1.0795) acc1: 34.6354 (34.6354) acc5: 100.0000 (100.0000) time: 7.0897 data: 5.4711 max mem: 3531 +Test: [ 10/517] eta: 0:06:08 acc1_pos: 21.615 acc5_pos: 100.000 loss: 1.0795 (1.1239) acc1: 29.1667 (29.3797) acc5: 100.0000 (100.0000) time: 0.7266 data: 0.4978 max mem: 3531 +Test: [ 20/517] eta: 0:03:30 acc1_pos: 26.042 acc5_pos: 100.000 loss: 1.0737 (1.0904) acc1: 26.0417 (30.9772) acc5: 100.0000 (100.0000) time: 0.0904 data: 0.0005 max mem: 3531 +Test: [ 30/517] eta: 0:02:45 acc1_pos: 10.417 acc5_pos: 100.000 loss: 1.1429 (1.1329) acc1: 22.3958 (28.4442) acc5: 100.0000 (100.0000) time: 0.1276 data: 0.0369 max mem: 3531 +Test: [ 40/517] eta: 0:02:34 acc1_pos: 11.458 acc5_pos: 100.000 loss: 1.2650 (1.1854) acc1: 11.4583 (24.2505) acc5: 100.0000 (100.0000) time: 0.2198 data: 0.1287 max mem: 3531 +Test: [ 50/517] eta: 0:02:23 acc1_pos: 34.862 acc5_pos: 100.000 loss: 1.2495 (1.1387) acc1: 15.3646 (27.2110) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2547 data: 0.1619 max mem: 3531 +Test: [ 60/517] eta: 0:02:14 acc1_pos: 34.862 acc5_pos: 100.000 loss: 0.3710 (1.0023) acc1: 84.6354 (37.7348) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2347 data: 0.1390 max mem: 3531 +Test: [ 70/517] eta: 0:02:08 acc1_pos: 34.862 acc5_pos: 100.000 loss: 0.2822 (0.8989) acc1: 94.2708 (45.9250) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.2376 data: 0.1430 max mem: 3531 +Test: [ 80/517] eta: 0:02:03 acc1_pos: 34.862 acc5_pos: 100.000 loss: 0.2845 (0.8265) acc1: 94.5312 (51.8293) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2462 data: 0.1543 max mem: 3531 +Test: [ 90/517] eta: 0:01:59 acc1_pos: 34.862 acc5_pos: 100.000 loss: 0.3094 (0.7685) acc1: 96.3542 (56.7794) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2570 data: 0.1663 max mem: 3531 +Test: [100/517] eta: 0:01:54 acc1_pos: 34.862 acc5_pos: 100.000 loss: 0.2596 (0.7170) acc1: 97.9167 (60.9014) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2389 data: 0.1446 max mem: 3531 +Test: [110/517] eta: 0:01:50 acc1_pos: 34.862 acc5_pos: 100.000 loss: 0.2368 (0.6737) acc1: 98.9583 (64.2267) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2293 data: 0.1350 max mem: 3531 +Test: [120/517] eta: 0:01:46 acc1_pos: 34.862 acc5_pos: 100.000 loss: 0.2534 (0.6399) acc1: 96.6146 (66.8862) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2447 data: 0.1536 max mem: 3531 +Test: [130/517] eta: 0:01:43 acc1_pos: 34.862 acc5_pos: 100.000 loss: 0.2606 (0.6106) acc1: 96.8750 (69.1675) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2438 data: 0.1512 max mem: 3531 +Test: [140/517] eta: 0:01:40 acc1_pos: 34.862 acc5_pos: 100.000 loss: 0.2476 (0.5844) acc1: 98.4375 (71.2821) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2513 data: 0.1594 max mem: 3531 +Test: [150/517] eta: 0:01:36 acc1_pos: 34.862 acc5_pos: 100.000 loss: 0.2421 (0.5619) acc1: 99.2188 (73.1012) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2449 data: 0.1540 max mem: 3531 +Test: [160/517] eta: 0:01:34 acc1_pos: 34.862 acc5_pos: 100.000 loss: 0.2744 (0.5457) acc1: 97.9167 (74.5196) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2405 data: 0.1454 max mem: 3531 +Test: [170/517] eta: 0:01:31 acc1_pos: 34.862 acc5_pos: 100.000 loss: 0.2925 (0.5306) acc1: 95.8333 (75.7691) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2506 data: 0.1535 max mem: 3531 +Test: [180/517] eta: 0:01:28 acc1_pos: 34.862 acc5_pos: 100.000 loss: 0.2522 (0.5143) acc1: 96.0938 (76.9222) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2494 data: 0.1562 max mem: 3531 +Test: [190/517] eta: 0:01:25 acc1_pos: 34.862 acc5_pos: 100.000 loss: 0.2172 (0.4973) acc1: 98.1771 (78.1141) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2538 data: 0.1619 max mem: 3531 +Test: [200/517] eta: 0:01:22 acc1_pos: 34.862 acc5_pos: 100.000 loss: 0.1904 (0.4823) acc1: 99.7396 (79.1770) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2445 data: 0.1527 max mem: 3531 +Test: [210/517] eta: 0:01:19 acc1_pos: 34.862 acc5_pos: 100.000 loss: 0.2069 (0.4768) acc1: 96.8750 (79.6320) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.2410 data: 0.1485 max mem: 3532 +Test: [220/517] eta: 0:01:17 acc1_pos: 34.862 acc5_pos: 100.000 loss: 0.3713 (0.4721) acc1: 86.9792 (80.0033) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.2539 data: 0.1592 max mem: 3532 +Test: [230/517] eta: 0:01:14 acc1_pos: 34.862 acc5_pos: 100.000 loss: 0.3365 (0.4655) acc1: 91.6667 (80.6784) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2507 data: 0.1566 max mem: 3532 +Test: [240/517] eta: 0:01:11 acc1_pos: 34.862 acc5_pos: 100.000 loss: 0.3230 (0.4601) acc1: 95.5729 (81.2975) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.2549 data: 0.1607 max mem: 3532 +Test: [250/517] eta: 0:01:08 acc1_pos: 34.862 acc5_pos: 100.000 loss: 0.2842 (0.4507) acc1: 98.1771 (81.9773) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2480 data: 0.1515 max mem: 3532 +Test: [260/517] eta: 0:01:05 acc1_pos: 34.862 acc5_pos: 100.000 loss: 0.1909 (0.4411) acc1: 99.4792 (82.6449) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2219 data: 0.1267 max mem: 3532 +Test: [270/517] eta: 0:01:03 acc1_pos: 34.862 acc5_pos: 100.000 loss: 0.2440 (0.4349) acc1: 99.2188 (83.2276) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2399 data: 0.1465 max mem: 3532 +Test: [280/517] eta: 0:01:00 acc1_pos: 34.862 acc5_pos: 100.000 loss: 0.2730 (0.4294) acc1: 98.1771 (83.7077) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2480 data: 0.1556 max mem: 3532 +Test: [290/517] eta: 0:00:58 acc1_pos: 34.862 acc5_pos: 100.000 loss: 0.2848 (0.4247) acc1: 97.3958 (84.1745) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2399 data: 0.1460 max mem: 3532 +Test: [300/517] eta: 0:00:55 acc1_pos: 34.862 acc5_pos: 100.000 loss: 0.2848 (0.4201) acc1: 97.3958 (84.5887) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2425 data: 0.1485 max mem: 3532 +Test: [310/517] eta: 0:00:52 acc1_pos: 34.862 acc5_pos: 100.000 loss: 0.2683 (0.4153) acc1: 98.1771 (85.0231) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2367 data: 0.1458 max mem: 3532 +Test: [320/517] eta: 0:00:50 acc1_pos: 34.862 acc5_pos: 100.000 loss: 0.2463 (0.4095) acc1: 98.6979 (85.4499) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2473 data: 0.1548 max mem: 3532 +Test: [330/517] eta: 0:00:47 acc1_pos: 34.862 acc5_pos: 100.000 loss: 0.2092 (0.4033) acc1: 99.2188 (85.8643) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2581 data: 0.1646 max mem: 3532 +Test: [340/517] eta: 0:00:44 acc1_pos: 34.862 acc5_pos: 100.000 loss: 0.1931 (0.3984) acc1: 99.2188 (86.1941) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.2467 data: 0.1533 max mem: 3532 +Test: [350/517] eta: 0:00:42 acc1_pos: 34.862 acc5_pos: 100.000 loss: 0.2949 (0.3967) acc1: 96.3542 (86.3990) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.2468 data: 0.1544 max mem: 3532 +Test: [360/517] eta: 0:00:39 acc1_pos: 34.862 acc5_pos: 100.000 loss: 0.3193 (0.3937) acc1: 94.5312 (86.6394) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2539 data: 0.1626 max mem: 3532 +Test: [370/517] eta: 0:00:37 acc1_pos: 34.862 acc5_pos: 100.000 loss: 0.1963 (0.3876) acc1: 99.7396 (86.9974) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2480 data: 0.1544 max mem: 3532 +Test: [380/517] eta: 0:00:34 acc1_pos: 34.862 acc5_pos: 100.000 loss: 0.1695 (0.3824) acc1: 100.0000 (87.3230) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2446 data: 0.1495 max mem: 3532 +Test: [390/517] eta: 0:00:32 acc1_pos: 34.862 acc5_pos: 100.000 loss: 0.1932 (0.3786) acc1: 99.2188 (87.5633) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2449 data: 0.1497 max mem: 3532 +Test: [400/517] eta: 0:00:29 acc1_pos: 34.862 acc5_pos: 100.000 loss: 0.1879 (0.3738) acc1: 99.7396 (87.8682) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2444 data: 0.1492 max mem: 3532 +Test: [410/517] eta: 0:00:27 acc1_pos: 34.862 acc5_pos: 100.000 loss: 0.2003 (0.3706) acc1: 99.7396 (88.1298) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2399 data: 0.1445 max mem: 3532 +Test: [420/517] eta: 0:00:24 acc1_pos: 34.862 acc5_pos: 100.000 loss: 0.2262 (0.3670) acc1: 99.7396 (88.3969) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2504 data: 0.1550 max mem: 3532 +Test: [430/517] eta: 0:00:21 acc1_pos: 34.862 acc5_pos: 100.000 loss: 0.2201 (0.3640) acc1: 99.7396 (88.6335) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2463 data: 0.1522 max mem: 3532 +Test: [440/517] eta: 0:00:19 acc1_pos: 34.862 acc5_pos: 100.000 loss: 0.2220 (0.3607) acc1: 98.9583 (88.8712) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2344 data: 0.1421 max mem: 3533 +Test: [450/517] eta: 0:00:16 acc1_pos: 34.862 acc5_pos: 100.000 loss: 0.2095 (0.3571) acc1: 100.0000 (89.1179) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2428 data: 0.1507 max mem: 3533 +Test: [460/517] eta: 0:00:14 acc1_pos: 34.862 acc5_pos: 100.000 loss: 0.2035 (0.3541) acc1: 100.0000 (89.3365) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2392 data: 0.1441 max mem: 3533 +Test: [470/517] eta: 0:00:11 acc1_pos: 34.862 acc5_pos: 100.000 loss: 0.1939 (0.3506) acc1: 99.7396 (89.5496) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2264 data: 0.1300 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 34.862 acc5_pos: 100.000 loss: 0.1891 (0.3473) acc1: 99.7396 (89.7539) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2359 data: 0.1415 max mem: 3533 +Test: [490/517] eta: 0:00:06 acc1_pos: 34.862 acc5_pos: 100.000 loss: 0.2165 (0.3449) acc1: 99.2188 (89.9318) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2861 data: 0.1928 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 34.862 acc5_pos: 100.000 loss: 0.2486 (0.3439) acc1: 98.4375 (90.0376) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2939 data: 0.2016 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 34.862 acc5_pos: 100.000 loss: 0.2488 (0.3418) acc1: 98.4375 (90.2066) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2089 data: 0.1153 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 34.862 acc5_pos: 100.000 loss: 0.2297 (0.3404) acc1: 98.9583 (90.3007) acc5: 100.0000 (100.0000) acc1_neg: 99.293 acc5_neg: 100.000 time: 0.2105 data: 0.1152 max mem: 3533 +Test: Total time: 0:02:08 (0.2493 s / it) +* Overall Acc@1 90.301 Acc@5 100.000 + target == real Acc@1 23.989 Acc@5 100.000 + target == fake Acc@1 97.180 Acc@5 100.000 + loss 0.340 + PR-AUC 0.9549 ROC-AUC 0.7363 EER 32.73% (thr=0.7705) +Accuracy of the network on the 198427 test images: 90.3% +[rank0]:[W726 05:24:55.972276074 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 05:25:16.348000 1197541 torch/distributed/run.py:766] +W0726 05:25:16.348000 1197541 torch/distributed/run.py:766] ***************************************** +W0726 05:25:16.348000 1197541 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:25:16.348000 1197541 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:25:45.743241471 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:25:45.838731098 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:25:45.890012269 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:25:46.217635061 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:25:46.438870786 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:25:46.510036381 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:25:46.588146570 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W726 05:25:46.588312203 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n32_seed8_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n32_classseed8_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:40:26 acc1_pos: 57.031 acc5_pos: 100.000 loss: 0.6720 (0.6720) acc1: 57.0312 (57.0312) acc5: 100.0000 (100.0000) time: 6.2370 data: 4.7058 max mem: 3531 +Test: [ 10/389] eta: 0:04:06 acc1_pos: 97.135 acc5_pos: 100.000 loss: 0.4756 (0.4980) acc1: 77.0833 (74.5739) acc5: 100.0000 (100.0000) time: 0.6491 data: 0.4282 max mem: 3531 +Test: [ 20/389] eta: 0:02:22 acc1_pos: 35.156 acc5_pos: 100.000 loss: 0.4763 (0.5477) acc1: 77.0833 (69.9033) acc5: 100.0000 (100.0000) time: 0.0935 data: 0.0004 max mem: 3531 +Test: [ 30/389] eta: 0:02:00 acc1_pos: 73.958 acc5_pos: 100.000 loss: 0.7125 (0.5964) acc1: 50.5208 (64.4657) acc5: 100.0000 (100.0000) time: 0.1633 data: 0.0684 max mem: 3531 +Test: [ 40/389] eta: 0:01:50 acc1_pos: 28.125 acc5_pos: 100.000 loss: 0.6861 (0.6120) acc1: 55.4688 (63.0907) acc5: 100.0000 (100.0000) time: 0.2408 data: 0.1465 max mem: 3531 +Test: [ 50/389] eta: 0:01:38 acc1_pos: 84.635 acc5_pos: 100.000 loss: 0.6087 (0.6109) acc1: 64.0625 (63.8940) acc5: 100.0000 (100.0000) time: 0.2225 data: 0.1280 max mem: 3531 +Test: [ 60/389] eta: 0:01:31 acc1_pos: 80.990 acc5_pos: 100.000 loss: 0.5131 (0.5899) acc1: 81.7708 (67.3198) acc5: 100.0000 (100.0000) time: 0.1978 data: 0.1035 max mem: 3531 +Test: [ 70/389] eta: 0:01:24 acc1_pos: 67.969 acc5_pos: 100.000 loss: 0.5100 (0.5952) acc1: 79.6875 (66.5126) acc5: 100.0000 (100.0000) time: 0.1926 data: 0.0969 max mem: 3531 +Test: [ 80/389] eta: 0:01:18 acc1_pos: 90.625 acc5_pos: 100.000 loss: 0.6218 (0.5971) acc1: 61.4583 (66.7567) acc5: 100.0000 (100.0000) time: 0.1864 data: 0.0925 max mem: 3531 +Test: [ 90/389] eta: 0:01:14 acc1_pos: 65.104 acc5_pos: 100.000 loss: 0.4024 (0.5801) acc1: 83.5938 (68.6298) acc5: 100.0000 (100.0000) time: 0.1946 data: 0.1032 max mem: 3531 +Test: [100/389] eta: 0:01:10 acc1_pos: 77.344 acc5_pos: 100.000 loss: 0.4166 (0.5822) acc1: 80.2083 (68.4097) acc5: 100.0000 (100.0000) time: 0.2001 data: 0.1081 max mem: 3531 +Test: [110/389] eta: 0:01:06 acc1_pos: 71.615 acc5_pos: 100.000 loss: 0.5783 (0.5920) acc1: 67.7083 (67.3705) acc5: 100.0000 (100.0000) time: 0.1957 data: 0.1029 max mem: 3531 +Test: [120/389] eta: 0:01:03 acc1_pos: 89.323 acc5_pos: 100.000 loss: 0.4604 (0.5806) acc1: 73.4375 (68.7651) acc5: 100.0000 (100.0000) time: 0.1908 data: 0.0979 max mem: 3531 +Test: [130/389] eta: 0:00:59 acc1_pos: 62.577 acc5_pos: 100.000 loss: 0.5014 (0.5837) acc1: 78.3854 (68.6208) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.1776 data: 0.0860 max mem: 3531 +Test: [140/389] eta: 0:00:56 acc1_pos: 62.577 acc5_pos: 100.000 loss: 0.5866 (0.5919) acc1: 64.5833 (67.9207) acc5: 100.0000 (100.0000) acc1_neg: 26.042 acc5_neg: 100.000 time: 0.1684 data: 0.0767 max mem: 3531 +Test: [150/389] eta: 0:00:53 acc1_pos: 62.577 acc5_pos: 100.000 loss: 0.6219 (0.5944) acc1: 69.0104 (67.7894) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1717 data: 0.0790 max mem: 3531 +Test: [160/389] eta: 0:00:49 acc1_pos: 62.577 acc5_pos: 100.000 loss: 0.6219 (0.5966) acc1: 69.0104 (67.9413) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1661 data: 0.0744 max mem: 3531 +Test: [170/389] eta: 0:00:47 acc1_pos: 62.577 acc5_pos: 100.000 loss: 0.3753 (0.5910) acc1: 95.8333 (68.3936) acc5: 100.0000 (100.0000) acc1_neg: 46.615 acc5_neg: 100.000 time: 0.1706 data: 0.0794 max mem: 3531 +Test: [180/389] eta: 0:00:45 acc1_pos: 62.577 acc5_pos: 100.000 loss: 0.4250 (0.5889) acc1: 88.0208 (68.8766) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.1969 data: 0.1041 max mem: 3531 +Test: [190/389] eta: 0:00:42 acc1_pos: 62.577 acc5_pos: 100.000 loss: 0.6263 (0.6003) acc1: 65.3646 (67.8529) acc5: 100.0000 (100.0000) acc1_neg: 3.646 acc5_neg: 100.000 time: 0.1820 data: 0.0888 max mem: 3531 +Test: [200/389] eta: 0:00:39 acc1_pos: 62.577 acc5_pos: 100.000 loss: 0.8630 (0.6114) acc1: 30.7292 (66.4568) acc5: 100.0000 (100.0000) acc1_neg: 19.010 acc5_neg: 100.000 time: 0.1732 data: 0.0800 max mem: 3531 +Test: [210/389] eta: 0:00:37 acc1_pos: 62.577 acc5_pos: 100.000 loss: 0.5545 (0.6078) acc1: 33.0729 (66.7802) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1901 data: 0.0964 max mem: 3532 +Test: [220/389] eta: 0:00:35 acc1_pos: 62.577 acc5_pos: 100.000 loss: 0.3143 (0.6003) acc1: 96.8750 (67.5834) acc5: 100.0000 (100.0000) acc1_neg: 60.938 acc5_neg: 100.000 time: 0.1759 data: 0.0821 max mem: 3532 +Test: [230/389] eta: 0:00:32 acc1_pos: 62.577 acc5_pos: 100.000 loss: 0.5497 (0.6034) acc1: 67.1875 (67.0691) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1680 data: 0.0754 max mem: 3532 +Test: [240/389] eta: 0:00:30 acc1_pos: 62.577 acc5_pos: 100.000 loss: 0.5328 (0.6013) acc1: 69.7917 (67.4652) acc5: 100.0000 (100.0000) acc1_neg: 54.948 acc5_neg: 100.000 time: 0.1741 data: 0.0824 max mem: 3532 +Test: [250/389] eta: 0:00:28 acc1_pos: 62.577 acc5_pos: 100.000 loss: 0.4926 (0.5971) acc1: 82.5521 (68.1244) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1735 data: 0.0813 max mem: 3532 +Test: [260/389] eta: 0:00:26 acc1_pos: 62.577 acc5_pos: 100.000 loss: 0.5993 (0.5993) acc1: 71.3542 (68.0875) acc5: 100.0000 (100.0000) acc1_neg: 12.760 acc5_neg: 100.000 time: 0.1739 data: 0.0796 max mem: 3532 +Test: [270/389] eta: 0:00:24 acc1_pos: 62.577 acc5_pos: 100.000 loss: 0.6358 (0.6120) acc1: 58.8542 (66.9367) acc5: 100.0000 (100.0000) acc1_neg: 44.010 acc5_neg: 100.000 time: 0.1816 data: 0.0876 max mem: 3532 +Test: [280/389] eta: 0:00:21 acc1_pos: 62.577 acc5_pos: 100.000 loss: 0.7459 (0.6217) acc1: 41.4062 (65.9809) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1794 data: 0.0874 max mem: 3532 +Test: [290/389] eta: 0:00:19 acc1_pos: 62.577 acc5_pos: 100.000 loss: 0.6719 (0.6169) acc1: 50.0000 (66.4483) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1689 data: 0.0768 max mem: 3532 +Test: [300/389] eta: 0:00:17 acc1_pos: 62.577 acc5_pos: 100.000 loss: 0.6504 (0.6202) acc1: 64.0625 (66.0403) acc5: 100.0000 (100.0000) acc1_neg: 55.469 acc5_neg: 100.000 time: 0.1623 data: 0.0693 max mem: 3532 +Test: [310/389] eta: 0:00:15 acc1_pos: 62.577 acc5_pos: 100.000 loss: 0.7385 (0.6295) acc1: 49.7396 (65.1578) acc5: 100.0000 (100.0000) acc1_neg: 3.125 acc5_neg: 100.000 time: 0.1649 data: 0.0718 max mem: 3532 +Test: [320/389] eta: 0:00:13 acc1_pos: 62.577 acc5_pos: 100.000 loss: 0.7705 (0.6314) acc1: 48.9583 (65.0174) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1586 data: 0.0657 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 62.577 acc5_pos: 100.000 loss: 0.9151 (0.6433) acc1: 13.8021 (63.5755) acc5: 100.0000 (100.0000) acc1_neg: 45.312 acc5_neg: 100.000 time: 0.1602 data: 0.0670 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 62.577 acc5_pos: 100.000 loss: 0.5689 (0.6353) acc1: 66.1458 (64.5222) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1684 data: 0.0757 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 62.577 acc5_pos: 100.000 loss: 0.3601 (0.6418) acc1: 97.1354 (64.0143) acc5: 100.0000 (100.0000) acc1_neg: 9.635 acc5_neg: 100.000 time: 0.1598 data: 0.0672 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 62.577 acc5_pos: 100.000 loss: 0.8227 (0.6479) acc1: 35.9375 (63.1175) acc5: 100.0000 (100.0000) acc1_neg: 36.979 acc5_neg: 100.000 time: 0.1528 data: 0.0608 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 62.577 acc5_pos: 100.000 loss: 0.8018 (0.6465) acc1: 39.5833 (63.2483) acc5: 100.0000 (100.0000) acc1_neg: 74.740 acc5_neg: 100.000 time: 0.1496 data: 0.0585 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 62.577 acc5_pos: 100.000 loss: 0.6050 (0.6482) acc1: 64.0625 (63.0003) acc5: 100.0000 (100.0000) acc1_neg: 42.188 acc5_neg: 100.000 time: 0.1352 data: 0.0451 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 62.577 acc5_pos: 100.000 loss: 0.6065 (0.6468) acc1: 65.1042 (63.1657) acc5: 100.0000 (100.0000) acc1_neg: 83.060 acc5_neg: 100.000 time: 0.1278 data: 0.0343 max mem: 3532 +Test: Total time: 0:01:13 (0.1884 s / it) +* Overall Acc@1 63.166 Acc@5 100.000 + target == real Acc@1 68.478 Acc@5 100.000 + target == fake Acc@1 60.697 Acc@5 100.000 + loss 0.647 + PR-AUC 0.8378 ROC-AUC 0.7017 EER 35.80% (thr=0.4785) +Accuracy of the network on the 149358 test images: 63.2% +[rank0]:[W726 05:27:10.592014349 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:27:30.653000 1198100 torch/distributed/run.py:766] +W0726 05:27:30.653000 1198100 torch/distributed/run.py:766] ***************************************** +W0726 05:27:30.653000 1198100 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:27:30.653000 1198100 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:27:59.013498165 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +| distributed init (rank 5): env:// +[rank2]:[W726 05:28:00.324270257 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:28:00.331583880 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W726 05:28:00.337533934 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:28:00.651110011 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:28:00.654016360 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:28:00.678391085 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:28:00.691795108 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n32_seed8_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n32_classseed8_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:34:18 acc1_pos: 20.312 acc5_pos: 100.000 loss: 1.2313 (1.2313) acc1: 20.3125 (20.3125) acc5: 100.0000 (100.0000) time: 6.1802 data: 4.7352 max mem: 3531 +Test: [ 10/333] eta: 0:03:28 acc1_pos: 86.979 acc5_pos: 100.000 loss: 0.9681 (0.7855) acc1: 39.5833 (52.9830) acc5: 100.0000 (100.0000) time: 0.6441 data: 0.4309 max mem: 3531 +Test: [ 20/333] eta: 0:02:00 acc1_pos: 94.531 acc5_pos: 100.000 loss: 0.3130 (0.5514) acc1: 87.5000 (70.4613) acc5: 100.0000 (100.0000) time: 0.0952 data: 0.0004 max mem: 3531 +Test: [ 30/333] eta: 0:01:32 acc1_pos: 57.812 acc5_pos: 100.000 loss: 0.3571 (0.6133) acc1: 78.3854 (65.6082) acc5: 100.0000 (100.0000) time: 0.1199 data: 0.0212 max mem: 3531 +Test: [ 40/333] eta: 0:01:24 acc1_pos: 58.073 acc5_pos: 100.000 loss: 0.7504 (0.6529) acc1: 52.6042 (62.2840) acc5: 100.0000 (100.0000) time: 0.1899 data: 0.0920 max mem: 3531 +Test: [ 50/333] eta: 0:01:17 acc1_pos: 52.344 acc5_pos: 100.000 loss: 0.7493 (0.6805) acc1: 52.3438 (60.1358) acc5: 100.0000 (100.0000) time: 0.2218 data: 0.1268 max mem: 3531 +Test: [ 60/333] eta: 0:01:11 acc1_pos: 47.656 acc5_pos: 100.000 loss: 0.7676 (0.6979) acc1: 52.3438 (58.9780) acc5: 100.0000 (100.0000) time: 0.2028 data: 0.1109 max mem: 3531 +Test: [ 70/333] eta: 0:01:06 acc1_pos: 45.833 acc5_pos: 100.000 loss: 0.7516 (0.7091) acc1: 53.1250 (58.0252) acc5: 100.0000 (100.0000) time: 0.1984 data: 0.1044 max mem: 3531 +Test: [ 80/333] eta: 0:01:02 acc1_pos: 55.990 acc5_pos: 100.000 loss: 0.7652 (0.7167) acc1: 49.7396 (57.1341) acc5: 100.0000 (100.0000) time: 0.2032 data: 0.1088 max mem: 3531 +Test: [ 90/333] eta: 0:01:00 acc1_pos: 67.969 acc5_pos: 100.000 loss: 0.7160 (0.7093) acc1: 55.9896 (57.9670) acc5: 100.0000 (100.0000) time: 0.2322 data: 0.1382 max mem: 3531 +Test: [100/333] eta: 0:00:57 acc1_pos: 67.448 acc5_pos: 100.000 loss: 0.6420 (0.7068) acc1: 64.5833 (58.2818) acc5: 100.0000 (100.0000) time: 0.2359 data: 0.1423 max mem: 3531 +Test: [110/333] eta: 0:00:54 acc1_pos: 67.969 acc5_pos: 100.000 loss: 0.6731 (0.7059) acc1: 59.1146 (58.3615) acc5: 100.0000 (100.0000) time: 0.2371 data: 0.1456 max mem: 3531 +Test: [120/333] eta: 0:00:50 acc1_pos: 68.490 acc5_pos: 100.000 loss: 0.6537 (0.7014) acc1: 59.1146 (58.7035) acc5: 100.0000 (100.0000) time: 0.1949 data: 0.0999 max mem: 3531 +Test: [130/333] eta: 0:00:47 acc1_pos: 78.646 acc5_pos: 100.000 loss: 0.5905 (0.6959) acc1: 66.4062 (59.3134) acc5: 100.0000 (100.0000) time: 0.1667 data: 0.0710 max mem: 3531 +Test: [140/333] eta: 0:00:44 acc1_pos: 80.208 acc5_pos: 100.000 loss: 0.6186 (0.6915) acc1: 65.8854 (59.6317) acc5: 100.0000 (100.0000) time: 0.1928 data: 0.1007 max mem: 3531 +Test: [150/333] eta: 0:00:43 acc1_pos: 69.531 acc5_pos: 100.000 loss: 0.6472 (0.6913) acc1: 62.2396 (59.6785) acc5: 100.0000 (100.0000) time: 0.2481 data: 0.1568 max mem: 3531 +Test: [160/333] eta: 0:00:40 acc1_pos: 54.276 acc5_pos: 100.000 loss: 0.6772 (0.6953) acc1: 60.6771 (59.4284) acc5: 100.0000 (100.0000) acc1_neg: 49.479 acc5_neg: 100.000 time: 0.2498 data: 0.1549 max mem: 3531 +Test: [170/333] eta: 0:00:37 acc1_pos: 54.276 acc5_pos: 100.000 loss: 0.7584 (0.7012) acc1: 52.8646 (58.9395) acc5: 100.0000 (100.0000) acc1_neg: 53.906 acc5_neg: 100.000 time: 0.2118 data: 0.1145 max mem: 3531 +Test: [180/333] eta: 0:00:35 acc1_pos: 54.276 acc5_pos: 100.000 loss: 0.7512 (0.7024) acc1: 54.4271 (58.8570) acc5: 100.0000 (100.0000) acc1_neg: 50.000 acc5_neg: 100.000 time: 0.2330 data: 0.1365 max mem: 3531 +Test: [190/333] eta: 0:00:33 acc1_pos: 54.276 acc5_pos: 100.000 loss: 0.6759 (0.6970) acc1: 60.9375 (59.3273) acc5: 100.0000 (100.0000) acc1_neg: 60.417 acc5_neg: 100.000 time: 0.2138 data: 0.1201 max mem: 3531 +Test: [200/333] eta: 0:00:30 acc1_pos: 54.276 acc5_pos: 100.000 loss: 0.6229 (0.6941) acc1: 63.0208 (59.6030) acc5: 100.0000 (100.0000) acc1_neg: 60.677 acc5_neg: 100.000 time: 0.2017 data: 0.1110 max mem: 3531 +Test: [210/333] eta: 0:00:28 acc1_pos: 54.276 acc5_pos: 100.000 loss: 0.6942 (0.6963) acc1: 59.8958 (59.4811) acc5: 100.0000 (100.0000) acc1_neg: 50.781 acc5_neg: 100.000 time: 0.2091 data: 0.1169 max mem: 3532 +Test: [220/333] eta: 0:00:25 acc1_pos: 54.276 acc5_pos: 100.000 loss: 0.7291 (0.7017) acc1: 58.0729 (59.1641) acc5: 100.0000 (100.0000) acc1_neg: 65.625 acc5_neg: 100.000 time: 0.2151 data: 0.1223 max mem: 3532 +Test: [230/333] eta: 0:00:23 acc1_pos: 54.276 acc5_pos: 100.000 loss: 0.5708 (0.6943) acc1: 66.4062 (59.8214) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.2223 data: 0.1308 max mem: 3532 +Test: [240/333] eta: 0:00:21 acc1_pos: 54.276 acc5_pos: 100.000 loss: 0.5667 (0.6923) acc1: 70.8333 (59.9509) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.2201 data: 0.1257 max mem: 3532 +Test: [250/333] eta: 0:00:18 acc1_pos: 54.276 acc5_pos: 100.000 loss: 0.6474 (0.6907) acc1: 61.1979 (60.0577) acc5: 100.0000 (100.0000) acc1_neg: 54.427 acc5_neg: 100.000 time: 0.2127 data: 0.1183 max mem: 3532 +Test: [260/333] eta: 0:00:16 acc1_pos: 54.276 acc5_pos: 100.000 loss: 0.6260 (0.6880) acc1: 62.5000 (60.2820) acc5: 100.0000 (100.0000) acc1_neg: 69.010 acc5_neg: 100.000 time: 0.2101 data: 0.1193 max mem: 3532 +Test: [270/333] eta: 0:00:14 acc1_pos: 54.276 acc5_pos: 100.000 loss: 0.6004 (0.6850) acc1: 67.4479 (60.5531) acc5: 100.0000 (100.0000) acc1_neg: 61.719 acc5_neg: 100.000 time: 0.2068 data: 0.1152 max mem: 3532 +Test: [280/333] eta: 0:00:11 acc1_pos: 54.276 acc5_pos: 100.000 loss: 0.6232 (0.6870) acc1: 64.8438 (60.3574) acc5: 100.0000 (100.0000) acc1_neg: 58.854 acc5_neg: 100.000 time: 0.1970 data: 0.1040 max mem: 3532 +Test: [290/333] eta: 0:00:09 acc1_pos: 54.276 acc5_pos: 100.000 loss: 0.8244 (0.6926) acc1: 45.8333 (59.8538) acc5: 100.0000 (100.0000) acc1_neg: 45.573 acc5_neg: 100.000 time: 0.1952 data: 0.1027 max mem: 3532 +Test: [300/333] eta: 0:00:07 acc1_pos: 54.276 acc5_pos: 100.000 loss: 0.8563 (0.6980) acc1: 45.8333 (59.4321) acc5: 100.0000 (100.0000) acc1_neg: 46.615 acc5_neg: 100.000 time: 0.1756 data: 0.0838 max mem: 3532 +Test: [310/333] eta: 0:00:05 acc1_pos: 54.276 acc5_pos: 100.000 loss: 0.7608 (0.6946) acc1: 52.0833 (59.7326) acc5: 100.0000 (100.0000) acc1_neg: 69.010 acc5_neg: 100.000 time: 0.1482 data: 0.0569 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 54.276 acc5_pos: 100.000 loss: 0.5670 (0.6918) acc1: 68.7500 (59.9664) acc5: 100.0000 (100.0000) acc1_neg: 53.385 acc5_neg: 100.000 time: 0.1369 data: 0.0467 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 54.276 acc5_pos: 100.000 loss: 0.7856 (0.6961) acc1: 52.3438 (59.6401) acc5: 100.0000 (100.0000) acc1_neg: 52.083 acc5_neg: 100.000 time: 0.1259 data: 0.0360 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 54.276 acc5_pos: 100.000 loss: 0.8059 (0.6969) acc1: 52.0833 (59.5887) acc5: 100.0000 (100.0000) acc1_neg: 52.394 acc5_neg: 100.000 time: 0.1275 data: 0.0344 max mem: 3532 +Test: Total time: 0:01:10 (0.2131 s / it) +* Overall Acc@1 59.589 Acc@5 100.000 + target == real Acc@1 59.819 Acc@5 100.000 + target == fake Acc@1 59.383 Acc@5 100.000 + loss 0.697 + PR-AUC 0.6421 ROC-AUC 0.6352 EER 40.41% (thr=0.4983) +Accuracy of the network on the 127843 test images: 59.6% +[rank0]:[W726 05:29:19.201493602 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:29:39.703000 1198647 torch/distributed/run.py:766] +W0726 05:29:39.703000 1198647 torch/distributed/run.py:766] ***************************************** +W0726 05:29:39.703000 1198647 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:29:39.703000 1198647 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:30:09.655297007 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:30:09.681946733 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:30:09.722932410 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:30:09.914207484 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:30:09.008641910 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +| distributed init (rank 7): env:// +[rank0]:[W726 05:30:09.036341246 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:30:09.046847267 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W726 05:30:09.046863667 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n32_seed8_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n32_classseed8_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:19:16 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.1803 (0.1803) acc1: 96.8750 (96.8750) acc5: 100.0000 (100.0000) time: 4.9196 data: 3.1256 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/235] eta: 0:02:03 acc1_pos: 96.354 acc5_pos: 100.000 loss: 0.1808 (0.1831) acc1: 97.1354 (96.8987) acc5: 100.0000 (100.0000) time: 0.5503 data: 0.3054 max mem: 3531 +Test: [ 20/235] eta: 0:01:12 acc1_pos: 97.917 acc5_pos: 100.000 loss: 0.1788 (0.1797) acc1: 97.3958 (97.1478) acc5: 100.0000 (100.0000) time: 0.1101 data: 0.0197 max mem: 3531 +Test: [ 30/235] eta: 0:00:55 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.1775 (0.1795) acc1: 97.1354 (97.0766) acc5: 100.0000 (100.0000) time: 0.1163 data: 0.0229 max mem: 3531 +Test: [ 40/235] eta: 0:00:52 acc1_pos: 96.615 acc5_pos: 100.000 loss: 0.1757 (0.1789) acc1: 97.1354 (97.0973) acc5: 100.0000 (100.0000) time: 0.1966 data: 0.1010 max mem: 3531 +Test: [ 50/235] eta: 0:00:51 acc1_pos: 96.094 acc5_pos: 100.000 loss: 0.1796 (0.1801) acc1: 96.8750 (97.0282) acc5: 100.0000 (100.0000) time: 0.2937 data: 0.1997 max mem: 3531 +Test: [ 60/235] eta: 0:00:47 acc1_pos: 97.135 acc5_pos: 100.000 loss: 0.1834 (0.1809) acc1: 96.6146 (96.9689) acc5: 100.0000 (100.0000) time: 0.2736 data: 0.1809 max mem: 3531 +Test: [ 70/235] eta: 0:00:42 acc1_pos: 96.923 acc5_pos: 100.000 loss: 0.1957 (0.3567) acc1: 95.5729 (86.8728) acc5: 100.0000 (100.0000) acc1_neg: 14.323 acc5_neg: 100.000 time: 0.1948 data: 0.1028 max mem: 3531 +Test: [ 80/235] eta: 0:00:36 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.5916 (0.5104) acc1: 15.8854 (78.0125) acc5: 100.0000 (100.0000) acc1_neg: 13.542 acc5_neg: 100.000 time: 0.1361 data: 0.0421 max mem: 3531 +Test: [ 90/235] eta: 0:00:33 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.5953 (0.6310) acc1: 15.1042 (71.0995) acc5: 100.0000 (100.0000) acc1_neg: 15.104 acc5_neg: 100.000 time: 0.1399 data: 0.0404 max mem: 3531 +Test: [100/235] eta: 0:00:30 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.5953 (0.7251) acc1: 15.1042 (65.6869) acc5: 100.0000 (100.0000) acc1_neg: 12.240 acc5_neg: 100.000 time: 0.1599 data: 0.0604 max mem: 3531 +Test: [110/235] eta: 0:00:27 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.5883 (0.8031) acc1: 15.6250 (61.1862) acc5: 100.0000 (100.0000) acc1_neg: 18.490 acc5_neg: 100.000 time: 0.1652 data: 0.0699 max mem: 3531 +Test: [120/235] eta: 0:00:24 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.6480 (0.8809) acc1: 13.2812 (56.9430) acc5: 100.0000 (100.0000) acc1_neg: 10.417 acc5_neg: 100.000 time: 0.1778 data: 0.0833 max mem: 3531 +Test: [130/235] eta: 0:00:22 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.7473 (0.9467) acc1: 9.6354 (53.3258) acc5: 100.0000 (100.0000) acc1_neg: 8.073 acc5_neg: 100.000 time: 0.1846 data: 0.0891 max mem: 3531 +Test: [140/235] eta: 0:00:19 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.7704 (1.0064) acc1: 9.1146 (50.1644) acc5: 100.0000 (100.0000) acc1_neg: 4.948 acc5_neg: 100.000 time: 0.1793 data: 0.0861 max mem: 3531 +Test: [150/235] eta: 0:00:17 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.7761 (1.0557) acc1: 8.3333 (47.4200) acc5: 100.0000 (100.0000) acc1_neg: 10.677 acc5_neg: 100.000 time: 0.1849 data: 0.0932 max mem: 3531 +Test: [160/235] eta: 0:00:15 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.7434 (1.0989) acc1: 8.8542 (45.0456) acc5: 100.0000 (100.0000) acc1_neg: 8.594 acc5_neg: 100.000 time: 0.1983 data: 0.1042 max mem: 3531 +Test: [170/235] eta: 0:00:13 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.7105 (1.1135) acc1: 10.9375 (43.8261) acc5: 100.0000 (100.0000) acc1_neg: 20.312 acc5_neg: 100.000 time: 0.1949 data: 0.1006 max mem: 3531 +Test: [180/235] eta: 0:00:11 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.3111 (1.1234) acc1: 26.0417 (42.8781) acc5: 100.0000 (100.0000) acc1_neg: 28.906 acc5_neg: 100.000 time: 0.1753 data: 0.0814 max mem: 3531 +Test: [190/235] eta: 0:00:09 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.2824 (1.1318) acc1: 27.0833 (42.0880) acc5: 100.0000 (100.0000) acc1_neg: 28.125 acc5_neg: 100.000 time: 0.1823 data: 0.0871 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.2878 (1.1409) acc1: 26.5625 (41.2456) acc5: 100.0000 (100.0000) acc1_neg: 24.219 acc5_neg: 100.000 time: 0.2039 data: 0.1101 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.3063 (1.1480) acc1: 24.7396 (40.5189) acc5: 100.0000 (100.0000) acc1_neg: 26.302 acc5_neg: 100.000 time: 0.1748 data: 0.0808 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.3054 (1.1549) acc1: 25.5208 (39.8697) acc5: 100.0000 (100.0000) acc1_neg: 25.521 acc5_neg: 100.000 time: 0.1719 data: 0.0777 max mem: 3532 +Test: [230/235] eta: 0:00:00 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.2934 (1.1603) acc1: 26.5625 (39.3263) acc5: 100.0000 (100.0000) acc1_neg: 27.344 acc5_neg: 100.000 time: 0.1642 data: 0.0711 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.2990 (1.1639) acc1: 26.3021 (39.1393) acc5: 100.0000 (100.0000) acc1_neg: 26.087 acc5_neg: 100.000 time: 0.1605 data: 0.0709 max mem: 3532 +Test: Total time: 0:00:46 (0.1978 s / it) +* Overall Acc@1 39.139 Acc@5 100.000 + target == real Acc@1 96.980 Acc@5 100.000 + target == fake Acc@1 18.220 Acc@5 100.000 + loss 1.164 + PR-AUC 0.8749 ROC-AUC 0.7231 EER 33.65% (thr=0.1486) +Accuracy of the network on the 89879 test images: 39.1% +[rank0]:[W726 05:31:08.915684517 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:31:28.845000 1199207 torch/distributed/run.py:766] +W0726 05:31:28.845000 1199207 torch/distributed/run.py:766] ***************************************** +W0726 05:31:28.845000 1199207 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:31:28.845000 1199207 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:31:58.846812140 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:31:59.193632886 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:31:59.374530523 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:31:59.494291456 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +NCCL version 2.26.2+cuda12.2 +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:31:59.670712311 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:31:59.830117499 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:31:59.837754801 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:31:59.859631656 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n32_seed8_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n32_classseed8_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:21:37 acc1_pos: 52.344 acc5_pos: 100.000 loss: 0.7595 (0.7595) acc1: 52.3438 (52.3438) acc5: 100.0000 (100.0000) time: 4.8233 data: 2.5143 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/269] eta: 0:02:17 acc1_pos: 77.083 acc5_pos: 100.000 loss: 0.6799 (0.6518) acc1: 60.6771 (63.5180) acc5: 100.0000 (100.0000) time: 0.5304 data: 0.2388 max mem: 3531 +Test: [ 20/269] eta: 0:01:21 acc1_pos: 72.917 acc5_pos: 100.000 loss: 0.5887 (0.6056) acc1: 62.7604 (68.1796) acc5: 100.0000 (100.0000) time: 0.1028 data: 0.0112 max mem: 3531 +Test: [ 30/269] eta: 0:01:01 acc1_pos: 71.094 acc5_pos: 100.000 loss: 0.5744 (0.5830) acc1: 72.6562 (70.2537) acc5: 100.0000 (100.0000) time: 0.1085 data: 0.0139 max mem: 3531 +Test: [ 40/269] eta: 0:00:55 acc1_pos: 25.000 acc5_pos: 100.000 loss: 0.5744 (0.5817) acc1: 71.0938 (70.1283) acc5: 100.0000 (100.0000) acc1_neg: 69.271 acc5_neg: 100.000 time: 0.1558 data: 0.0619 max mem: 3531 +Test: [ 50/269] eta: 0:00:54 acc1_pos: 25.000 acc5_pos: 100.000 loss: 0.5977 (0.5937) acc1: 68.4896 (69.3117) acc5: 100.0000 (100.0000) acc1_neg: 47.396 acc5_neg: 100.000 time: 0.2339 data: 0.1422 max mem: 3531 +Test: [ 60/269] eta: 0:00:50 acc1_pos: 25.000 acc5_pos: 100.000 loss: 0.6765 (0.6095) acc1: 58.0729 (67.4821) acc5: 100.0000 (100.0000) acc1_neg: 47.656 acc5_neg: 100.000 time: 0.2440 data: 0.1490 max mem: 3531 +Test: [ 70/269] eta: 0:00:48 acc1_pos: 25.000 acc5_pos: 100.000 loss: 0.6771 (0.6147) acc1: 57.5521 (66.5786) acc5: 100.0000 (100.0000) acc1_neg: 64.844 acc5_neg: 100.000 time: 0.2224 data: 0.1263 max mem: 3531 +Test: [ 80/269] eta: 0:00:44 acc1_pos: 25.000 acc5_pos: 100.000 loss: 0.6009 (0.6112) acc1: 64.8438 (66.8789) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.2188 data: 0.1249 max mem: 3531 +Test: [ 90/269] eta: 0:00:42 acc1_pos: 25.000 acc5_pos: 100.000 loss: 0.6225 (0.6251) acc1: 62.7604 (65.1271) acc5: 100.0000 (100.0000) acc1_neg: 49.219 acc5_neg: 100.000 time: 0.2199 data: 0.1273 max mem: 3531 +Test: [100/269] eta: 0:00:39 acc1_pos: 25.000 acc5_pos: 100.000 loss: 0.7399 (0.6355) acc1: 49.4792 (63.8356) acc5: 100.0000 (100.0000) acc1_neg: 43.750 acc5_neg: 100.000 time: 0.2108 data: 0.1176 max mem: 3531 +Test: [110/269] eta: 0:00:36 acc1_pos: 25.000 acc5_pos: 100.000 loss: 0.7413 (0.6431) acc1: 51.5625 (63.0865) acc5: 100.0000 (100.0000) acc1_neg: 57.031 acc5_neg: 100.000 time: 0.1974 data: 0.1035 max mem: 3531 +Test: [120/269] eta: 0:00:33 acc1_pos: 25.000 acc5_pos: 100.000 loss: 0.7110 (0.6447) acc1: 54.6875 (62.9025) acc5: 100.0000 (100.0000) acc1_neg: 41.927 acc5_neg: 100.000 time: 0.2025 data: 0.1092 max mem: 3531 +Test: [130/269] eta: 0:00:31 acc1_pos: 25.000 acc5_pos: 100.000 loss: 0.7336 (0.6582) acc1: 53.6458 (61.3947) acc5: 100.0000 (100.0000) acc1_neg: 30.208 acc5_neg: 100.000 time: 0.1927 data: 0.0987 max mem: 3531 +Test: [140/269] eta: 0:00:28 acc1_pos: 25.000 acc5_pos: 100.000 loss: 0.8893 (0.6794) acc1: 33.5938 (59.1183) acc5: 100.0000 (100.0000) acc1_neg: 23.177 acc5_neg: 100.000 time: 0.1804 data: 0.0880 max mem: 3531 +Test: [150/269] eta: 0:00:26 acc1_pos: 25.000 acc5_pos: 100.000 loss: 0.8989 (0.6914) acc1: 29.4271 (57.7349) acc5: 100.0000 (100.0000) acc1_neg: 41.667 acc5_neg: 100.000 time: 0.1904 data: 0.0961 max mem: 3531 +Test: [160/269] eta: 0:00:23 acc1_pos: 25.000 acc5_pos: 100.000 loss: 0.7541 (0.6923) acc1: 47.9167 (57.6427) acc5: 100.0000 (100.0000) acc1_neg: 66.146 acc5_neg: 100.000 time: 0.1962 data: 0.1021 max mem: 3531 +Test: [170/269] eta: 0:00:21 acc1_pos: 25.000 acc5_pos: 100.000 loss: 0.7083 (0.6926) acc1: 57.2917 (57.6998) acc5: 100.0000 (100.0000) acc1_neg: 68.229 acc5_neg: 100.000 time: 0.1884 data: 0.0975 max mem: 3531 +Test: [180/269] eta: 0:00:19 acc1_pos: 25.000 acc5_pos: 100.000 loss: 0.6152 (0.6841) acc1: 67.1875 (58.6973) acc5: 100.0000 (100.0000) acc1_neg: 66.927 acc5_neg: 100.000 time: 0.1932 data: 0.1016 max mem: 3531 +Test: [190/269] eta: 0:00:17 acc1_pos: 25.000 acc5_pos: 100.000 loss: 0.6152 (0.6859) acc1: 67.1875 (58.4547) acc5: 100.0000 (100.0000) acc1_neg: 54.688 acc5_neg: 100.000 time: 0.2187 data: 0.1266 max mem: 3531 +Test: [200/269] eta: 0:00:14 acc1_pos: 25.000 acc5_pos: 100.000 loss: 0.7252 (0.6896) acc1: 51.8229 (58.1040) acc5: 100.0000 (100.0000) acc1_neg: 74.740 acc5_neg: 100.000 time: 0.1932 data: 0.1000 max mem: 3531 +Test: [210/269] eta: 0:00:12 acc1_pos: 25.000 acc5_pos: 100.000 loss: 0.7441 (0.6909) acc1: 50.2604 (57.8421) acc5: 100.0000 (100.0000) acc1_neg: 56.771 acc5_neg: 100.000 time: 0.1670 data: 0.0722 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 25.000 acc5_pos: 100.000 loss: 0.7152 (0.6933) acc1: 48.6979 (57.4178) acc5: 100.0000 (100.0000) acc1_neg: 60.938 acc5_neg: 100.000 time: 0.1953 data: 0.1017 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 25.000 acc5_pos: 100.000 loss: 0.7098 (0.6934) acc1: 47.6562 (57.2691) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.2026 data: 0.1106 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 25.000 acc5_pos: 100.000 loss: 0.6783 (0.6925) acc1: 58.3333 (57.4710) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.1724 data: 0.0801 max mem: 3532 +Test: [250/269] eta: 0:00:03 acc1_pos: 25.000 acc5_pos: 100.000 loss: 0.7354 (0.7011) acc1: 52.6042 (56.6349) acc5: 100.0000 (100.0000) acc1_neg: 15.365 acc5_neg: 100.000 time: 0.1384 data: 0.0471 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 25.000 acc5_pos: 100.000 loss: 0.9197 (0.7095) acc1: 29.9479 (55.7880) acc5: 100.0000 (100.0000) acc1_neg: 56.510 acc5_neg: 100.000 time: 0.1296 data: 0.0393 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 25.000 acc5_pos: 100.000 loss: 0.7704 (0.7080) acc1: 45.0521 (55.9409) acc5: 100.0000 (100.0000) acc1_neg: 82.415 acc5_neg: 100.000 time: 0.1307 data: 0.0370 max mem: 3532 +Test: Total time: 0:00:53 (0.2005 s / it) +* Overall Acc@1 55.941 Acc@5 100.000 + target == real Acc@1 69.566 Acc@5 100.000 + target == fake Acc@1 54.081 Acc@5 100.000 + loss 0.708 + PR-AUC 0.9288 ROC-AUC 0.6655 EER 38.03% (thr=0.4536) +Accuracy of the network on the 103293 test images: 55.9% +[rank0]:[W726 05:33:02.784754650 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T05:33:08Z] eval_done tag=fixed500k_real_n32_seed8_resampled +[2026-07-26T05:33:09Z] eval_start tag=fixed500k_fake_n64_seed8_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed8_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T05:33:09Z +hostname=n116-040-213 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed8_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n64_seed8_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 05:33:22.880000 1199783 torch/distributed/run.py:766] +W0726 05:33:22.880000 1199783 torch/distributed/run.py:766] ***************************************** +W0726 05:33:22.880000 1199783 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:33:22.880000 1199783 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:33:53.860452720 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:33:53.945562974 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:33:53.084081564 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:33:53.168060088 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 05:33:54.320012100 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W726 05:33:54.320130133 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:33:54.355568732 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:33:54.555295732 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n64_seed8_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed8_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:21:52 acc1_pos: 52.865 acc5_pos: 100.000 loss: 0.6835 (0.6835) acc1: 52.8646 (52.8646) acc5: 100.0000 (100.0000) time: 5.6098 data: 4.0642 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/234] eta: 0:02:13 acc1_pos: 52.604 acc5_pos: 100.000 loss: 0.6835 (0.7016) acc1: 52.8646 (57.2680) acc5: 100.0000 (100.0000) time: 0.5946 data: 0.3698 max mem: 3531 +Test: [ 20/234] eta: 0:01:15 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.6974 (0.7035) acc1: 52.6042 (55.6548) acc5: 100.0000 (100.0000) time: 0.0917 data: 0.0004 max mem: 3531 +Test: [ 30/234] eta: 0:00:55 acc1_pos: 80.469 acc5_pos: 100.000 loss: 0.2057 (0.5511) acc1: 95.8333 (68.0696) acc5: 100.0000 (100.0000) time: 0.0931 data: 0.0004 max mem: 3531 +Test: [ 40/234] eta: 0:00:49 acc1_pos: 98.958 acc5_pos: 100.000 loss: 0.2046 (0.5026) acc1: 96.0938 (72.1164) acc5: 100.0000 (100.0000) time: 0.1490 data: 0.0535 max mem: 3531 +Test: [ 50/234] eta: 0:00:44 acc1_pos: 98.958 acc5_pos: 100.000 loss: 0.1287 (0.4286) acc1: 98.4375 (77.3284) acc5: 100.0000 (100.0000) time: 0.1910 data: 0.0981 max mem: 3531 +Test: [ 60/234] eta: 0:00:39 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.1057 (0.3721) acc1: 98.9583 (81.0280) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1710 data: 0.0808 max mem: 3531 +Test: [ 70/234] eta: 0:00:36 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.0832 (0.3927) acc1: 100.0000 (79.7902) acc5: 100.0000 (100.0000) acc1_neg: 58.854 acc5_neg: 100.000 time: 0.1688 data: 0.0766 max mem: 3531 +Test: [ 80/234] eta: 0:00:34 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.6953 (0.4091) acc1: 59.3750 (78.9255) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2029 data: 0.1097 max mem: 3531 +Test: [ 90/234] eta: 0:00:31 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.2380 (0.3854) acc1: 91.1458 (80.6204) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.2249 data: 0.1329 max mem: 3531 +Test: [100/234] eta: 0:00:29 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.2429 (0.3917) acc1: 87.7604 (80.3811) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2096 data: 0.1154 max mem: 3531 +Test: [110/234] eta: 0:00:27 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4453 (0.3939) acc1: 76.8229 (80.4359) acc5: 100.0000 (100.0000) acc1_neg: 74.219 acc5_neg: 100.000 time: 0.2257 data: 0.1312 max mem: 3531 +Test: [120/234] eta: 0:00:25 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4109 (0.4029) acc1: 78.6458 (80.0749) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.2312 data: 0.1390 max mem: 3531 +Test: [130/234] eta: 0:00:22 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.3824 (0.4017) acc1: 86.1979 (80.5344) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.1755 data: 0.0837 max mem: 3531 +Test: [140/234] eta: 0:00:19 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4331 (0.4291) acc1: 79.1667 (79.1944) acc5: 100.0000 (100.0000) acc1_neg: 33.333 acc5_neg: 100.000 time: 0.1614 data: 0.0700 max mem: 3531 +Test: [150/234] eta: 0:00:18 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4718 (0.4389) acc1: 70.8333 (78.6441) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.2290 data: 0.1365 max mem: 3531 +Test: [160/234] eta: 0:00:16 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4374 (0.4363) acc1: 77.8646 (79.0502) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.2460 data: 0.1520 max mem: 3531 +Test: [170/234] eta: 0:00:13 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4528 (0.4381) acc1: 79.1667 (79.1286) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.1651 data: 0.0722 max mem: 3531 +Test: [180/234] eta: 0:00:11 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4927 (0.4503) acc1: 77.6042 (78.3869) acc5: 100.0000 (100.0000) acc1_neg: 30.729 acc5_neg: 100.000 time: 0.1510 data: 0.0585 max mem: 3531 +Test: [190/234] eta: 0:00:09 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.6649 (0.4629) acc1: 63.2812 (77.5265) acc5: 100.0000 (100.0000) acc1_neg: 63.802 acc5_neg: 100.000 time: 0.1859 data: 0.0926 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.6496 (0.4551) acc1: 63.8021 (77.9877) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1863 data: 0.0930 max mem: 3531 +Test: [210/234] eta: 0:00:04 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.0972 (0.4394) acc1: 99.7396 (78.9828) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1808 data: 0.0885 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.1731 (0.4286) acc1: 97.9167 (79.7287) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.1719 data: 0.0796 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.3990 (0.4473) acc1: 46.3542 (78.1318) acc5: 100.0000 (100.0000) acc1_neg: 42.708 acc5_neg: 100.000 time: 0.1544 data: 0.0621 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.8325 (0.4525) acc1: 43.4896 (77.7318) acc5: 100.0000 (100.0000) acc1_neg: 44.747 acc5_neg: 100.000 time: 0.1400 data: 0.0472 max mem: 3532 +Test: Total time: 0:00:46 (0.2003 s / it) +* Overall Acc@1 77.732 Acc@5 100.000 + target == real Acc@1 77.480 Acc@5 100.000 + target == fake Acc@1 77.803 Acc@5 100.000 + loss 0.452 + PR-AUC 0.9623 ROC-AUC 0.8756 EER 22.34% (thr=0.5024) +Accuracy of the network on the 89729 test images: 77.7% +[rank0]:[W726 05:34:52.706498346 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:35:12.063000 1200329 torch/distributed/run.py:766] +W0726 05:35:12.063000 1200329 torch/distributed/run.py:766] ***************************************** +W0726 05:35:12.063000 1200329 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:35:12.063000 1200329 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:35:40.688280827 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:35:40.069374457 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:35:41.556278906 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:35:41.643990379 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:35:41.666191602 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:35:41.702733547 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:35:41.716402769 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:35:41.731776840 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n64_seed8_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed8_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:42:30 acc1_pos: 32.292 acc5_pos: 100.000 loss: 1.0394 (1.0394) acc1: 32.2917 (32.2917) acc5: 100.0000 (100.0000) time: 7.1013 data: 5.6345 max mem: 3531 +Test: [ 10/2387] eta: 0:28:50 acc1_pos: 6.510 acc5_pos: 100.000 loss: 1.1738 (1.1414) acc1: 21.6146 (22.2538) acc5: 100.0000 (100.0000) time: 0.7279 data: 0.5127 max mem: 3531 +Test: [ 20/2387] eta: 0:16:52 acc1_pos: 9.115 acc5_pos: 100.000 loss: 1.1738 (1.1428) acc1: 20.0521 (21.5402) acc5: 100.0000 (100.0000) time: 0.0939 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:13:27 acc1_pos: 11.719 acc5_pos: 100.000 loss: 1.1697 (1.1542) acc1: 17.7083 (20.6317) acc5: 100.0000 (100.0000) time: 0.1305 data: 0.0327 max mem: 3531 +Test: [ 40/2387] eta: 0:12:17 acc1_pos: 20.573 acc5_pos: 100.000 loss: 1.1594 (1.1629) acc1: 14.5833 (19.0676) acc5: 100.0000 (100.0000) time: 0.1949 data: 0.0992 max mem: 3531 +Test: [ 50/2387] eta: 0:11:26 acc1_pos: 39.062 acc5_pos: 100.000 loss: 1.1161 (1.1473) acc1: 16.1458 (19.9602) acc5: 100.0000 (100.0000) time: 0.2181 data: 0.1261 max mem: 3531 +Test: [ 60/2387] eta: 0:10:55 acc1_pos: 41.406 acc5_pos: 100.000 loss: 1.0491 (1.1382) acc1: 25.5208 (20.8333) acc5: 100.0000 (100.0000) time: 0.2149 data: 0.1222 max mem: 3531 +Test: [ 70/2387] eta: 0:10:25 acc1_pos: 29.948 acc5_pos: 100.000 loss: 1.1210 (1.1433) acc1: 22.6562 (20.3859) acc5: 100.0000 (100.0000) time: 0.2098 data: 0.1173 max mem: 3531 +Test: [ 80/2387] eta: 0:10:04 acc1_pos: 36.979 acc5_pos: 100.000 loss: 1.0923 (1.1302) acc1: 20.3125 (20.8237) acc5: 100.0000 (100.0000) time: 0.2027 data: 0.1088 max mem: 3531 +Test: [ 90/2387] eta: 0:09:55 acc1_pos: 44.010 acc5_pos: 100.000 loss: 0.9876 (1.1065) acc1: 28.9062 (22.3186) acc5: 100.0000 (100.0000) time: 0.2202 data: 0.1266 max mem: 3531 +Test: [ 100/2387] eta: 0:09:43 acc1_pos: 31.250 acc5_pos: 100.000 loss: 0.9246 (1.0980) acc1: 31.2500 (22.7181) acc5: 100.0000 (100.0000) time: 0.2272 data: 0.1332 max mem: 3531 +Test: [ 110/2387] eta: 0:09:34 acc1_pos: 29.167 acc5_pos: 100.000 loss: 0.9246 (1.0825) acc1: 29.1667 (23.6838) acc5: 100.0000 (100.0000) time: 0.2221 data: 0.1233 max mem: 3531 +Test: [ 120/2387] eta: 0:09:26 acc1_pos: 29.427 acc5_pos: 100.000 loss: 0.9573 (1.0735) acc1: 31.2500 (24.3005) acc5: 100.0000 (100.0000) time: 0.2225 data: 0.1240 max mem: 3531 +Test: [ 130/2387] eta: 0:09:22 acc1_pos: 26.823 acc5_pos: 100.000 loss: 0.9685 (1.0662) acc1: 29.4271 (24.5706) acc5: 100.0000 (100.0000) time: 0.2325 data: 0.1350 max mem: 3531 +Test: [ 140/2387] eta: 0:09:17 acc1_pos: 27.344 acc5_pos: 100.000 loss: 0.9685 (1.0595) acc1: 27.3438 (24.9224) acc5: 100.0000 (100.0000) time: 0.2383 data: 0.1393 max mem: 3531 +Test: [ 150/2387] eta: 0:09:28 acc1_pos: 8.073 acc5_pos: 100.000 loss: 0.9288 (1.0516) acc1: 30.4688 (25.2259) acc5: 100.0000 (100.0000) time: 0.2858 data: 0.1899 max mem: 3531 +Test: [ 160/2387] eta: 0:09:26 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.8948 (1.0286) acc1: 30.7292 (27.0494) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2970 data: 0.2050 max mem: 3531 +Test: [ 170/2387] eta: 0:09:12 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.2230 (0.9808) acc1: 96.8750 (31.2058) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2106 data: 0.1183 max mem: 3531 +Test: [ 180/2387] eta: 0:08:51 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.2093 (0.9382) acc1: 98.1771 (34.9002) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1318 data: 0.0375 max mem: 3531 +Test: [ 190/2387] eta: 0:08:37 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.2083 (0.9000) acc1: 98.1771 (38.2063) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1202 data: 0.0254 max mem: 3531 +Test: [ 200/2387] eta: 0:08:30 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.2056 (0.8655) acc1: 97.9167 (41.1873) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1671 data: 0.0753 max mem: 3531 +Test: [ 210/2387] eta: 0:08:27 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.2102 (0.8347) acc1: 97.9167 (43.8598) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2085 data: 0.1167 max mem: 3532 +Test: [ 220/2387] eta: 0:08:21 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.2123 (0.8065) acc1: 97.9167 (46.3082) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2101 data: 0.1170 max mem: 3532 +Test: [ 230/2387] eta: 0:08:14 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.2136 (0.7808) acc1: 97.9167 (48.5435) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1903 data: 0.0983 max mem: 3532 +Test: [ 240/2387] eta: 0:08:06 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.2190 (0.7577) acc1: 97.9167 (50.5954) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1740 data: 0.0834 max mem: 3532 +Test: [ 250/2387] eta: 0:08:02 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1974 (0.7316) acc1: 98.4375 (52.5544) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1821 data: 0.0909 max mem: 3532 +Test: [ 260/2387] eta: 0:07:57 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.0829 (0.7067) acc1: 100.0000 (54.3722) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1983 data: 0.1055 max mem: 3532 +Test: [ 270/2387] eta: 0:07:53 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.0787 (0.6836) acc1: 100.0000 (56.0540) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1969 data: 0.0995 max mem: 3532 +Test: [ 280/2387] eta: 0:07:49 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.0802 (0.6623) acc1: 100.0000 (57.6160) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2049 data: 0.1065 max mem: 3532 +Test: [ 290/2387] eta: 0:07:45 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.0831 (0.6424) acc1: 100.0000 (59.0725) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2005 data: 0.1068 max mem: 3532 +Test: [ 300/2387] eta: 0:07:40 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.0818 (0.6238) acc1: 100.0000 (60.4322) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1873 data: 0.0962 max mem: 3532 +Test: [ 310/2387] eta: 0:07:34 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.0867 (0.6109) acc1: 100.0000 (61.6158) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1762 data: 0.0855 max mem: 3532 +Test: [ 320/2387] eta: 0:07:31 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.3907 (0.6042) acc1: 95.0521 (62.6404) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1885 data: 0.0974 max mem: 3532 +Test: [ 330/2387] eta: 0:07:28 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.3927 (0.5977) acc1: 95.0521 (63.6369) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2005 data: 0.1090 max mem: 3532 +Test: [ 340/2387] eta: 0:07:24 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.3889 (0.5916) acc1: 95.0521 (64.5566) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1938 data: 0.0962 max mem: 3532 +Test: [ 350/2387] eta: 0:07:19 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.3819 (0.5798) acc1: 95.5729 (65.5471) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1785 data: 0.0804 max mem: 3532 +Test: [ 360/2387] eta: 0:07:15 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1602 (0.5684) acc1: 99.4792 (66.4762) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1725 data: 0.0793 max mem: 3532 +Test: [ 370/2387] eta: 0:07:11 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1683 (0.5576) acc1: 99.2188 (67.3567) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1838 data: 0.0894 max mem: 3532 +Test: [ 380/2387] eta: 0:07:07 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1687 (0.5474) acc1: 98.9583 (68.1806) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1823 data: 0.0886 max mem: 3532 +Test: [ 390/2387] eta: 0:07:03 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1699 (0.5377) acc1: 98.9583 (68.9718) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1800 data: 0.0878 max mem: 3532 +Test: [ 400/2387] eta: 0:07:00 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1685 (0.5286) acc1: 98.9583 (69.7254) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1829 data: 0.0910 max mem: 3532 +Test: [ 410/2387] eta: 0:06:56 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1666 (0.5197) acc1: 99.2188 (70.4411) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1851 data: 0.0936 max mem: 3532 +Test: [ 420/2387] eta: 0:06:54 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1665 (0.5114) acc1: 98.9583 (71.1210) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1930 data: 0.1019 max mem: 3532 +Test: [ 430/2387] eta: 0:06:49 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1710 (0.5037) acc1: 98.9583 (71.7614) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1817 data: 0.0906 max mem: 3532 +Test: [ 440/2387] eta: 0:06:46 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1757 (0.4951) acc1: 98.9583 (72.3941) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1716 data: 0.0802 max mem: 3533 +Test: [ 450/2387] eta: 0:06:43 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.0805 (0.4859) acc1: 100.0000 (73.0062) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1878 data: 0.0955 max mem: 3533 +Test: [ 460/2387] eta: 0:06:39 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.0801 (0.4771) acc1: 100.0000 (73.5917) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1776 data: 0.0847 max mem: 3533 +Test: [ 470/2387] eta: 0:06:36 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.0800 (0.4687) acc1: 100.0000 (74.1524) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1705 data: 0.0781 max mem: 3533 +Test: [ 480/2387] eta: 0:06:34 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.0802 (0.4606) acc1: 100.0000 (74.6898) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1916 data: 0.0999 max mem: 3533 +Test: [ 490/2387] eta: 0:06:30 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.0807 (0.4529) acc1: 100.0000 (75.2053) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1793 data: 0.0881 max mem: 3533 +Test: [ 500/2387] eta: 0:06:26 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.0809 (0.4455) acc1: 100.0000 (75.7002) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1594 data: 0.0684 max mem: 3533 +Test: [ 510/2387] eta: 0:06:23 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.0807 (0.4386) acc1: 100.0000 (76.1757) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1687 data: 0.0754 max mem: 3533 +Test: [ 520/2387] eta: 0:06:20 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1125 (0.4325) acc1: 100.0000 (76.6330) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1705 data: 0.0768 max mem: 3533 +Test: [ 530/2387] eta: 0:06:17 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1120 (0.4264) acc1: 100.0000 (77.0730) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1711 data: 0.0795 max mem: 3533 +Test: [ 540/2387] eta: 0:06:14 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1111 (0.4206) acc1: 100.0000 (77.4968) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1800 data: 0.0878 max mem: 3533 +Test: [ 550/2387] eta: 0:06:12 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1107 (0.4149) acc1: 100.0000 (77.9052) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1999 data: 0.1052 max mem: 3533 +Test: [ 560/2387] eta: 0:06:10 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1105 (0.4095) acc1: 100.0000 (78.2991) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2053 data: 0.1103 max mem: 3533 +Test: [ 570/2387] eta: 0:06:08 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1105 (0.4043) acc1: 100.0000 (78.6791) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2011 data: 0.1088 max mem: 3533 +Test: [ 580/2387] eta: 0:06:06 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1106 (0.3992) acc1: 100.0000 (79.0461) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2031 data: 0.1124 max mem: 3533 +Test: [ 590/2387] eta: 0:06:04 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1110 (0.3944) acc1: 100.0000 (79.4006) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2006 data: 0.1099 max mem: 3533 +Test: [ 600/2387] eta: 0:06:01 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1139 (0.3899) acc1: 100.0000 (79.7434) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1923 data: 0.0997 max mem: 3533 +Test: [ 610/2387] eta: 0:05:59 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1367 (0.3861) acc1: 100.0000 (80.0724) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.1022 max mem: 3533 +Test: [ 620/2387] eta: 0:05:58 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1554 (0.3824) acc1: 99.7396 (80.3895) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2097 data: 0.1179 max mem: 3533 +Test: [ 630/2387] eta: 0:05:56 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1581 (0.3788) acc1: 99.7396 (80.6962) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2112 data: 0.1174 max mem: 3533 +Test: [ 640/2387] eta: 0:05:54 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1581 (0.3755) acc1: 99.7396 (80.9920) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2088 data: 0.1150 max mem: 3533 +Test: [ 650/2387] eta: 0:05:52 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1587 (0.3721) acc1: 99.4792 (81.2760) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2084 data: 0.1140 max mem: 3533 +Test: [ 660/2387] eta: 0:05:50 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1587 (0.3689) acc1: 99.7396 (81.5561) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2101 data: 0.1164 max mem: 3533 +Test: [ 670/2387] eta: 0:05:48 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1563 (0.3658) acc1: 99.7396 (81.8271) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2059 data: 0.1146 max mem: 3534 +Test: [ 680/2387] eta: 0:05:46 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1564 (0.3627) acc1: 100.0000 (82.0936) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2048 data: 0.1112 max mem: 3534 +Test: [ 690/2387] eta: 0:05:45 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1603 (0.3630) acc1: 100.0000 (82.1044) acc5: 100.0000 (100.0000) acc1_neg: 72.396 acc5_neg: 100.000 time: 0.2138 data: 0.1200 max mem: 3534 +Test: [ 700/2387] eta: 0:05:42 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.4528 (0.3643) acc1: 77.0833 (82.0257) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.2015 data: 0.1098 max mem: 3534 +Test: [ 710/2387] eta: 0:05:39 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.4557 (0.3656) acc1: 76.5625 (81.9463) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.1700 data: 0.0771 max mem: 3534 +Test: [ 720/2387] eta: 0:05:37 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.4557 (0.3668) acc1: 77.0833 (81.8810) acc5: 100.0000 (100.0000) acc1_neg: 77.344 acc5_neg: 100.000 time: 0.1687 data: 0.0751 max mem: 3534 +Test: [ 730/2387] eta: 0:05:34 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.4563 (0.3682) acc1: 76.5625 (81.8050) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.1778 data: 0.0852 max mem: 3534 +Test: [ 740/2387] eta: 0:05:31 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.4599 (0.3694) acc1: 76.5625 (81.7399) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.1714 data: 0.0772 max mem: 3534 +Test: [ 750/2387] eta: 0:05:29 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.4538 (0.3705) acc1: 77.6042 (81.6869) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.1743 data: 0.0802 max mem: 3534 +Test: [ 760/2387] eta: 0:05:26 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.4425 (0.3710) acc1: 77.6042 (81.6764) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.1796 data: 0.0883 max mem: 3534 +Test: [ 770/2387] eta: 0:05:24 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.3190 (0.3695) acc1: 86.4583 (81.8124) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1758 data: 0.0841 max mem: 3534 +Test: [ 780/2387] eta: 0:05:21 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1768 (0.3667) acc1: 97.9167 (82.0293) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1805 data: 0.0877 max mem: 3534 +Test: [ 790/2387] eta: 0:05:20 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1413 (0.3640) acc1: 98.6979 (82.2100) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.2026 data: 0.1107 max mem: 3534 +Test: [ 800/2387] eta: 0:05:16 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1307 (0.3611) acc1: 99.7396 (82.4279) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1740 data: 0.0810 max mem: 3534 +Test: [ 810/2387] eta: 0:05:14 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1402 (0.3585) acc1: 99.7396 (82.6330) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1571 data: 0.0619 max mem: 3534 +Test: [ 820/2387] eta: 0:05:12 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1527 (0.3561) acc1: 99.4792 (82.8227) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1868 data: 0.0931 max mem: 3534 +Test: [ 830/2387] eta: 0:05:09 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1988 (0.3549) acc1: 96.8750 (82.9592) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1784 data: 0.0867 max mem: 3534 +Test: [ 840/2387] eta: 0:05:07 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.2567 (0.3541) acc1: 93.7500 (83.0865) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1708 data: 0.0786 max mem: 3534 +Test: [ 850/2387] eta: 0:05:04 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.2883 (0.3533) acc1: 93.7500 (83.1962) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1780 data: 0.0844 max mem: 3534 +Test: [ 860/2387] eta: 0:05:02 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.2697 (0.3521) acc1: 94.5312 (83.3406) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1781 data: 0.0852 max mem: 3534 +Test: [ 870/2387] eta: 0:05:00 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.2446 (0.3510) acc1: 96.3542 (83.4849) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1783 data: 0.0848 max mem: 3534 +Test: [ 880/2387] eta: 0:04:58 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.2641 (0.3502) acc1: 96.0938 (83.6041) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1930 data: 0.0980 max mem: 3534 +Test: [ 890/2387] eta: 0:04:56 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.2416 (0.3486) acc1: 97.3958 (83.7691) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1926 data: 0.0986 max mem: 3534 +Test: [ 900/2387] eta: 0:04:53 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.2426 (0.3509) acc1: 97.3958 (83.6244) acc5: 100.0000 (100.0000) acc1_neg: 66.667 acc5_neg: 100.000 time: 0.1760 data: 0.0836 max mem: 3535 +Test: [ 910/2387] eta: 0:04:51 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.6274 (0.3539) acc1: 63.8021 (83.4057) acc5: 100.0000 (100.0000) acc1_neg: 62.760 acc5_neg: 100.000 time: 0.1706 data: 0.0791 max mem: 3535 +Test: [ 920/2387] eta: 0:04:48 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.6368 (0.3570) acc1: 63.0208 (83.1840) acc5: 100.0000 (100.0000) acc1_neg: 60.417 acc5_neg: 100.000 time: 0.1693 data: 0.0781 max mem: 3535 +Test: [ 930/2387] eta: 0:04:46 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.6373 (0.3600) acc1: 63.2812 (82.9756) acc5: 100.0000 (100.0000) acc1_neg: 60.156 acc5_neg: 100.000 time: 0.1670 data: 0.0758 max mem: 3535 +Test: [ 940/2387] eta: 0:04:43 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.6402 (0.3629) acc1: 63.2812 (82.7616) acc5: 100.0000 (100.0000) acc1_neg: 60.677 acc5_neg: 100.000 time: 0.1609 data: 0.0698 max mem: 3535 +Test: [ 950/2387] eta: 0:04:41 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.6375 (0.3657) acc1: 63.2812 (82.5587) acc5: 100.0000 (100.0000) acc1_neg: 64.323 acc5_neg: 100.000 time: 0.1620 data: 0.0706 max mem: 3535 +Test: [ 960/2387] eta: 0:04:39 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.6144 (0.3683) acc1: 64.3229 (82.3705) acc5: 100.0000 (100.0000) acc1_neg: 64.583 acc5_neg: 100.000 time: 0.1691 data: 0.0767 max mem: 3535 +Test: [ 970/2387] eta: 0:04:36 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.6255 (0.3711) acc1: 63.8021 (82.1721) acc5: 100.0000 (100.0000) acc1_neg: 65.885 acc5_neg: 100.000 time: 0.1660 data: 0.0734 max mem: 3535 +Test: [ 980/2387] eta: 0:04:34 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.6136 (0.3709) acc1: 64.8438 (82.1956) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.1712 data: 0.0789 max mem: 3535 +Test: [ 990/2387] eta: 0:04:32 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1804 (0.3690) acc1: 96.3542 (82.3295) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1769 data: 0.0847 max mem: 3535 +Test: [1000/2387] eta: 0:04:30 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1339 (0.3669) acc1: 99.4792 (82.4883) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1815 data: 0.0896 max mem: 3535 +Test: [1010/2387] eta: 0:04:27 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1547 (0.3652) acc1: 99.2188 (82.6312) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1802 data: 0.0888 max mem: 3535 +Test: [1020/2387] eta: 0:04:25 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1637 (0.3635) acc1: 98.9583 (82.7709) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1681 data: 0.0768 max mem: 3535 +Test: [1030/2387] eta: 0:04:23 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1435 (0.3614) acc1: 99.4792 (82.9277) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1704 data: 0.0780 max mem: 3535 +Test: [1040/2387] eta: 0:04:21 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1264 (0.3591) acc1: 99.7396 (83.0889) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1765 data: 0.0832 max mem: 3535 +Test: [1050/2387] eta: 0:04:18 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1156 (0.3568) acc1: 100.0000 (83.2474) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1722 data: 0.0793 max mem: 3535 +Test: [1060/2387] eta: 0:04:16 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1169 (0.3547) acc1: 100.0000 (83.4018) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1690 data: 0.0760 max mem: 3535 +Test: [1070/2387] eta: 0:04:14 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1266 (0.3528) acc1: 99.7396 (83.5446) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1691 data: 0.0765 max mem: 3535 +Test: [1080/2387] eta: 0:04:12 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1876 (0.3516) acc1: 98.1771 (83.6687) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1696 data: 0.0782 max mem: 3535 +Test: [1090/2387] eta: 0:04:10 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.2061 (0.3501) acc1: 96.3542 (83.7923) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1745 data: 0.0835 max mem: 3535 +Test: [1100/2387] eta: 0:04:08 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1813 (0.3486) acc1: 97.3958 (83.9197) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1818 data: 0.0900 max mem: 3535 +Test: [1110/2387] eta: 0:04:05 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1813 (0.3474) acc1: 97.9167 (84.0227) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1779 data: 0.0854 max mem: 3535 +Test: [1120/2387] eta: 0:04:03 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1935 (0.3461) acc1: 97.1354 (84.1378) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1737 data: 0.0812 max mem: 3536 +Test: [1130/2387] eta: 0:04:01 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1833 (0.3448) acc1: 97.1354 (84.2580) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1786 data: 0.0865 max mem: 3536 +Test: [1140/2387] eta: 0:03:59 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1973 (0.3436) acc1: 97.3958 (84.3757) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1794 data: 0.0884 max mem: 3536 +Test: [1150/2387] eta: 0:03:57 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.2108 (0.3427) acc1: 97.9167 (84.4838) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1809 data: 0.0895 max mem: 3536 +Test: [1160/2387] eta: 0:03:55 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.2072 (0.3414) acc1: 97.9167 (84.5811) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1797 data: 0.0869 max mem: 3536 +Test: [1170/2387] eta: 0:03:53 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1905 (0.3402) acc1: 96.6146 (84.6906) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1772 data: 0.0837 max mem: 3536 +Test: [1180/2387] eta: 0:03:51 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1905 (0.3390) acc1: 96.8750 (84.7959) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1727 data: 0.0799 max mem: 3536 +Test: [1190/2387] eta: 0:03:49 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1773 (0.3378) acc1: 97.9167 (84.9063) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1722 data: 0.0804 max mem: 3536 +Test: [1200/2387] eta: 0:03:47 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1646 (0.3363) acc1: 99.4792 (85.0288) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1867 data: 0.0940 max mem: 3536 +Test: [1210/2387] eta: 0:03:45 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1501 (0.3348) acc1: 99.7396 (85.1500) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1726 data: 0.0800 max mem: 3536 +Test: [1220/2387] eta: 0:03:43 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1124 (0.3329) acc1: 99.7396 (85.2704) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1612 data: 0.0673 max mem: 3536 +Test: [1230/2387] eta: 0:03:41 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.0912 (0.3309) acc1: 99.7396 (85.3887) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.0997 max mem: 3536 +Test: [1240/2387] eta: 0:03:38 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.0925 (0.3290) acc1: 100.0000 (85.5054) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1779 data: 0.0847 max mem: 3536 +Test: [1250/2387] eta: 0:03:36 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.0926 (0.3271) acc1: 100.0000 (85.6203) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1472 data: 0.0547 max mem: 3536 +Test: [1260/2387] eta: 0:03:34 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.0910 (0.3252) acc1: 100.0000 (85.7324) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1612 data: 0.0693 max mem: 3536 +Test: [1270/2387] eta: 0:03:32 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.0875 (0.3234) acc1: 100.0000 (85.8445) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1696 data: 0.0782 max mem: 3536 +Test: [1280/2387] eta: 0:03:30 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.0895 (0.3216) acc1: 100.0000 (85.9521) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1650 data: 0.0737 max mem: 3536 +Test: [1290/2387] eta: 0:03:28 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1049 (0.3201) acc1: 99.2188 (86.0480) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1647 data: 0.0701 max mem: 3536 +Test: [1300/2387] eta: 0:03:26 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1027 (0.3184) acc1: 99.2188 (86.1499) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1680 data: 0.0737 max mem: 3536 +Test: [1310/2387] eta: 0:03:24 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1050 (0.3169) acc1: 99.4792 (86.2482) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1613 data: 0.0704 max mem: 3536 +Test: [1320/2387] eta: 0:03:22 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1176 (0.3154) acc1: 99.4792 (86.3442) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1643 data: 0.0720 max mem: 3536 +Test: [1330/2387] eta: 0:03:19 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1233 (0.3142) acc1: 99.7396 (86.4396) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1667 data: 0.0719 max mem: 3536 +Test: [1340/2387] eta: 0:03:17 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1779 (0.3140) acc1: 98.9583 (86.4605) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1608 data: 0.0669 max mem: 3536 +Test: [1350/2387] eta: 0:03:15 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.0646 (0.3121) acc1: 100.0000 (86.5607) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1761 data: 0.0845 max mem: 3537 +Test: [1360/2387] eta: 0:03:13 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.0557 (0.3102) acc1: 100.0000 (86.6594) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1540 data: 0.0617 max mem: 3537 +Test: [1370/2387] eta: 0:03:11 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.0515 (0.3083) acc1: 100.0000 (86.7567) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1445 data: 0.0526 max mem: 3537 +Test: [1380/2387] eta: 0:03:09 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.0517 (0.3065) acc1: 100.0000 (86.8526) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1716 data: 0.0772 max mem: 3537 +Test: [1390/2387] eta: 0:03:07 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.0561 (0.3047) acc1: 100.0000 (86.9472) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1707 data: 0.0760 max mem: 3537 +Test: [1400/2387] eta: 0:03:05 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.0561 (0.3029) acc1: 100.0000 (87.0403) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1835 data: 0.0904 max mem: 3537 +Test: [1410/2387] eta: 0:03:04 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.0590 (0.3018) acc1: 100.0000 (87.1163) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2110 data: 0.1175 max mem: 3537 +Test: [1420/2387] eta: 0:03:02 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1721 (0.3011) acc1: 97.9167 (87.1787) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2090 data: 0.1175 max mem: 3537 +Test: [1430/2387] eta: 0:03:00 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1732 (0.3003) acc1: 97.1354 (87.2465) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1816 data: 0.0875 max mem: 3537 +Test: [1440/2387] eta: 0:02:58 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1842 (0.2995) acc1: 97.6562 (87.3238) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1799 data: 0.0816 max mem: 3537 +Test: [1450/2387] eta: 0:02:56 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1842 (0.2989) acc1: 98.6979 (87.3920) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1838 data: 0.0865 max mem: 3537 +Test: [1460/2387] eta: 0:02:55 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1554 (0.2979) acc1: 98.6979 (87.4681) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2338 data: 0.1374 max mem: 3537 +Test: [1470/2387] eta: 0:02:53 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1240 (0.2966) acc1: 99.4792 (87.5510) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2453 data: 0.1496 max mem: 3537 +Test: [1480/2387] eta: 0:02:50 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1155 (0.2955) acc1: 99.7396 (87.6278) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1610 data: 0.0657 max mem: 3537 +Test: [1490/2387] eta: 0:02:49 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1207 (0.2943) acc1: 99.4792 (87.7082) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1710 data: 0.0766 max mem: 3537 +Test: [1500/2387] eta: 0:02:47 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1237 (0.2932) acc1: 99.4792 (87.7859) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2061 data: 0.1142 max mem: 3537 +Test: [1510/2387] eta: 0:02:45 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1341 (0.2922) acc1: 99.7396 (87.8661) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1743 data: 0.0822 max mem: 3537 +Test: [1520/2387] eta: 0:02:43 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1510 (0.2918) acc1: 99.2188 (87.9190) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1684 data: 0.0766 max mem: 3537 +Test: [1530/2387] eta: 0:02:41 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.2309 (0.2916) acc1: 95.8333 (87.9591) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1797 data: 0.0874 max mem: 3537 +Test: [1540/2387] eta: 0:02:39 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.2233 (0.2911) acc1: 96.0938 (88.0186) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1753 data: 0.0824 max mem: 3537 +Test: [1550/2387] eta: 0:02:37 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.2153 (0.2906) acc1: 97.3958 (88.0801) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1892 data: 0.0969 max mem: 3537 +Test: [1560/2387] eta: 0:02:35 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.2141 (0.2903) acc1: 97.1354 (88.1318) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1845 data: 0.0926 max mem: 3537 +Test: [1570/2387] eta: 0:02:33 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.2010 (0.2896) acc1: 97.6562 (88.1992) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1766 data: 0.0854 max mem: 3537 +Test: [1580/2387] eta: 0:02:31 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1710 (0.2888) acc1: 99.4792 (88.2691) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1939 data: 0.1024 max mem: 3538 +Test: [1590/2387] eta: 0:02:30 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1542 (0.2879) acc1: 99.4792 (88.3389) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1931 data: 0.1016 max mem: 3538 +Test: [1600/2387] eta: 0:02:28 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1381 (0.2871) acc1: 99.4792 (88.3906) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1861 data: 0.0944 max mem: 3538 +Test: [1610/2387] eta: 0:02:26 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1330 (0.2862) acc1: 99.4792 (88.4578) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1829 data: 0.0902 max mem: 3538 +Test: [1620/2387] eta: 0:02:24 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1534 (0.2854) acc1: 98.9583 (88.5185) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2137 data: 0.1207 max mem: 3538 +Test: [1630/2387] eta: 0:02:22 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1677 (0.2848) acc1: 97.9167 (88.5725) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2034 data: 0.1104 max mem: 3538 +Test: [1640/2387] eta: 0:02:20 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.2169 (0.2846) acc1: 97.3958 (88.6144) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1853 data: 0.0933 max mem: 3538 +Test: [1650/2387] eta: 0:02:19 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.2184 (0.2842) acc1: 96.0938 (88.6634) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2050 data: 0.1130 max mem: 3538 +Test: [1660/2387] eta: 0:02:17 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.2255 (0.2840) acc1: 96.6146 (88.7041) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2128 data: 0.1184 max mem: 3538 +Test: [1670/2387] eta: 0:02:15 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.2289 (0.2836) acc1: 96.6146 (88.7475) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2182 data: 0.1242 max mem: 3538 +Test: [1680/2387] eta: 0:02:13 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.2263 (0.2833) acc1: 96.6146 (88.7883) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.2273 data: 0.1353 max mem: 3538 +Test: [1690/2387] eta: 0:02:12 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.2256 (0.2831) acc1: 96.0938 (88.8266) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2417 data: 0.1495 max mem: 3538 +Test: [1700/2387] eta: 0:02:10 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.2247 (0.2825) acc1: 97.3958 (88.8802) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1995 data: 0.1071 max mem: 3538 +Test: [1710/2387] eta: 0:02:08 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1362 (0.2816) acc1: 99.2188 (88.9427) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1952 data: 0.1035 max mem: 3538 +Test: [1720/2387] eta: 0:02:06 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1088 (0.2806) acc1: 99.7396 (89.0058) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2161 data: 0.1245 max mem: 3538 +Test: [1730/2387] eta: 0:02:04 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1088 (0.2796) acc1: 99.7396 (89.0661) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1901 data: 0.0988 max mem: 3538 +Test: [1740/2387] eta: 0:02:02 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1033 (0.2786) acc1: 99.7396 (89.1267) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1832 data: 0.0917 max mem: 3538 +Test: [1750/2387] eta: 0:02:00 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1045 (0.2777) acc1: 100.0000 (89.1877) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1815 data: 0.0871 max mem: 3538 +Test: [1760/2387] eta: 0:01:58 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1045 (0.2767) acc1: 100.0000 (89.2488) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1674 data: 0.0734 max mem: 3538 +Test: [1770/2387] eta: 0:01:56 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1182 (0.2760) acc1: 99.7396 (89.2994) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1711 data: 0.0798 max mem: 3538 +Test: [1780/2387] eta: 0:01:54 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1314 (0.2751) acc1: 98.9583 (89.3555) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1852 data: 0.0936 max mem: 3538 +Test: [1790/2387] eta: 0:01:53 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1163 (0.2743) acc1: 99.4792 (89.4113) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1799 data: 0.0871 max mem: 3538 +Test: [1800/2387] eta: 0:01:51 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1163 (0.2734) acc1: 99.4792 (89.4661) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1705 data: 0.0784 max mem: 3538 +Test: [1810/2387] eta: 0:01:49 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1165 (0.2725) acc1: 99.4792 (89.5216) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1717 data: 0.0794 max mem: 3539 +Test: [1820/2387] eta: 0:01:47 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1120 (0.2716) acc1: 100.0000 (89.5785) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1723 data: 0.0795 max mem: 3539 +Test: [1830/2387] eta: 0:01:45 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1131 (0.2708) acc1: 100.0000 (89.6350) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1683 data: 0.0768 max mem: 3539 +Test: [1840/2387] eta: 0:01:43 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1129 (0.2699) acc1: 100.0000 (89.6913) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1820 data: 0.0910 max mem: 3539 +Test: [1850/2387] eta: 0:01:41 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1124 (0.2691) acc1: 100.0000 (89.7468) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1925 data: 0.1001 max mem: 3539 +Test: [1860/2387] eta: 0:01:39 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1127 (0.2682) acc1: 100.0000 (89.8019) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2285 data: 0.1358 max mem: 3539 +Test: [1870/2387] eta: 0:01:37 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1127 (0.2674) acc1: 100.0000 (89.8564) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2040 data: 0.1124 max mem: 3539 +Test: [1880/2387] eta: 0:01:35 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1122 (0.2666) acc1: 100.0000 (89.9103) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1751 data: 0.0834 max mem: 3539 +Test: [1890/2387] eta: 0:01:34 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1122 (0.2658) acc1: 100.0000 (89.9637) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1980 data: 0.1053 max mem: 3539 +Test: [1900/2387] eta: 0:01:32 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1138 (0.2650) acc1: 100.0000 (90.0165) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1955 data: 0.1016 max mem: 3539 +Test: [1910/2387] eta: 0:01:30 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1408 (0.2644) acc1: 100.0000 (90.0678) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2013 data: 0.1078 max mem: 3539 +Test: [1920/2387] eta: 0:01:28 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1433 (0.2640) acc1: 100.0000 (90.1031) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1976 data: 0.1017 max mem: 3539 +Test: [1930/2387] eta: 0:01:26 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1390 (0.2633) acc1: 100.0000 (90.1542) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1946 data: 0.1000 max mem: 3539 +Test: [1940/2387] eta: 0:01:24 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1424 (0.2630) acc1: 100.0000 (90.1678) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1949 data: 0.1026 max mem: 3539 +Test: [1950/2387] eta: 0:01:22 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1516 (0.2625) acc1: 99.4792 (90.2115) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1919 data: 0.0983 max mem: 3539 +Test: [1960/2387] eta: 0:01:20 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1602 (0.2620) acc1: 99.4792 (90.2522) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2021 data: 0.1082 max mem: 3539 +Test: [1970/2387] eta: 0:01:18 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.2070 (0.2621) acc1: 98.4375 (90.2607) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.1803 data: 0.0850 max mem: 3539 +Test: [1980/2387] eta: 0:01:16 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.2255 (0.2617) acc1: 97.3958 (90.2943) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1530 data: 0.0596 max mem: 3539 +Test: [1990/2387] eta: 0:01:15 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1352 (0.2611) acc1: 99.2188 (90.3386) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1639 data: 0.0728 max mem: 3539 +Test: [2000/2387] eta: 0:01:13 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1346 (0.2605) acc1: 99.2188 (90.3836) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1896 data: 0.0990 max mem: 3539 +Test: [2010/2387] eta: 0:01:11 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1352 (0.2599) acc1: 99.2188 (90.4267) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1883 data: 0.0973 max mem: 3539 +Test: [2020/2387] eta: 0:01:09 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1359 (0.2592) acc1: 99.2188 (90.4701) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1748 data: 0.0825 max mem: 3539 +Test: [2030/2387] eta: 0:01:07 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1338 (0.2586) acc1: 99.2188 (90.5131) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1803 data: 0.0871 max mem: 3540 +Test: [2040/2387] eta: 0:01:05 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1338 (0.2580) acc1: 99.2188 (90.5560) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1710 data: 0.0788 max mem: 3540 +Test: [2050/2387] eta: 0:01:03 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1338 (0.2574) acc1: 99.2188 (90.5995) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1643 data: 0.0721 max mem: 3540 +Test: [2060/2387] eta: 0:01:01 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1374 (0.2569) acc1: 99.4792 (90.6419) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1606 data: 0.0688 max mem: 3540 +Test: [2070/2387] eta: 0:00:59 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1377 (0.2563) acc1: 99.4792 (90.6851) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1642 data: 0.0719 max mem: 3540 +Test: [2080/2387] eta: 0:00:57 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1369 (0.2557) acc1: 99.4792 (90.7271) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1584 data: 0.0661 max mem: 3540 +Test: [2090/2387] eta: 0:00:55 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1375 (0.2551) acc1: 99.2188 (90.7682) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1617 data: 0.0702 max mem: 3540 +Test: [2100/2387] eta: 0:00:53 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1377 (0.2546) acc1: 99.4792 (90.8091) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1695 data: 0.0774 max mem: 3540 +Test: [2110/2387] eta: 0:00:52 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1377 (0.2540) acc1: 99.4792 (90.8494) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1704 data: 0.0789 max mem: 3540 +Test: [2120/2387] eta: 0:00:50 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1401 (0.2537) acc1: 99.2188 (90.8841) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1812 data: 0.0878 max mem: 3540 +Test: [2130/2387] eta: 0:00:48 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1444 (0.2531) acc1: 99.4792 (90.9238) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1790 data: 0.0852 max mem: 3540 +Test: [2140/2387] eta: 0:00:46 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1417 (0.2528) acc1: 99.4792 (90.9454) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1992 data: 0.1073 max mem: 3540 +Test: [2150/2387] eta: 0:00:44 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1546 (0.2524) acc1: 98.6979 (90.9834) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2385 data: 0.1460 max mem: 3540 +Test: [2160/2387] eta: 0:00:42 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1546 (0.2519) acc1: 99.2188 (91.0188) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2158 data: 0.1208 max mem: 3540 +Test: [2170/2387] eta: 0:00:40 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1629 (0.2518) acc1: 98.1771 (91.0444) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1657 data: 0.0690 max mem: 3540 +Test: [2180/2387] eta: 0:00:38 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.2546 (0.2527) acc1: 95.0521 (90.9923) acc5: 100.0000 (100.0000) acc1_neg: 56.250 acc5_neg: 100.000 time: 0.1633 data: 0.0663 max mem: 3540 +Test: [2190/2387] eta: 0:00:37 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.2736 (0.2529) acc1: 92.4479 (90.9797) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1893 data: 0.0936 max mem: 3540 +Test: [2200/2387] eta: 0:00:35 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1594 (0.2524) acc1: 98.4375 (91.0165) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1910 data: 0.0984 max mem: 3540 +Test: [2210/2387] eta: 0:00:33 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1405 (0.2519) acc1: 99.2188 (91.0522) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1860 data: 0.0921 max mem: 3540 +Test: [2220/2387] eta: 0:00:31 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1405 (0.2515) acc1: 99.2188 (91.0870) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1866 data: 0.0909 max mem: 3540 +Test: [2230/2387] eta: 0:00:29 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1358 (0.2509) acc1: 99.4792 (91.1249) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1936 data: 0.0993 max mem: 3540 +Test: [2240/2387] eta: 0:00:27 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1337 (0.2505) acc1: 99.2188 (91.1584) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1821 data: 0.0872 max mem: 3540 +Test: [2250/2387] eta: 0:00:25 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1507 (0.2501) acc1: 98.9583 (91.1921) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1868 data: 0.0924 max mem: 3540 +Test: [2260/2387] eta: 0:00:23 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1473 (0.2496) acc1: 100.0000 (91.2305) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1902 data: 0.0977 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1249 (0.2490) acc1: 100.0000 (91.2688) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1804 data: 0.0887 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1165 (0.2486) acc1: 100.0000 (91.2911) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1861 data: 0.0943 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1125 (0.2480) acc1: 100.0000 (91.3287) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1696 data: 0.0767 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1233 (0.2475) acc1: 100.0000 (91.3628) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1602 data: 0.0670 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1367 (0.2471) acc1: 100.0000 (91.3952) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1633 data: 0.0711 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1671 (0.2470) acc1: 99.2188 (91.4205) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1653 data: 0.0737 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.2133 (0.2468) acc1: 97.6562 (91.4424) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.1681 data: 0.0751 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1835 (0.2466) acc1: 98.6979 (91.4708) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1442 data: 0.0520 max mem: 3541 +Test: [2350/2387] eta: 0:00:06 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.1782 (0.2466) acc1: 98.6979 (91.4789) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1325 data: 0.0422 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.2076 (0.2466) acc1: 97.3958 (91.4946) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1464 data: 0.0562 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.2346 (0.2465) acc1: 97.3958 (91.5115) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.1351 data: 0.0448 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.2359 (0.2466) acc1: 96.0938 (91.5222) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1344 data: 0.0437 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 26.027 acc5_pos: 100.000 loss: 0.2758 (0.2467) acc1: 92.7083 (91.5253) acc5: 100.0000 (100.0000) acc1_neg: 90.698 acc5_neg: 100.000 time: 0.1338 data: 0.0458 max mem: 3541 +Test: Total time: 0:07:25 (0.1866 s / it) +* Overall Acc@1 91.525 Acc@5 100.000 + target == real Acc@1 25.290 Acc@5 100.000 + target == fake Acc@1 96.177 Acc@5 100.000 + loss 0.247 + PR-AUC 0.9913 ROC-AUC 0.8873 EER 18.76% (thr=0.7754) +Accuracy of the network on the 916267 test images: 91.5% +[rank0]:[W726 05:43:32.435258498 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:43:52.115000 1200951 torch/distributed/run.py:766] +W0726 05:43:52.115000 1200951 torch/distributed/run.py:766] ***************************************** +W0726 05:43:52.115000 1200951 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:43:52.115000 1200951 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 7): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:44:23.676895385 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:44:23.680199226 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:44:23.119610920 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:44:24.216901065 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:44:24.229508589 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +| distributed init (rank 2): env:// +[rank0]:[W726 05:44:24.462262118 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:44:24.468318904 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank1]:[W726 05:44:24.468338584 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n64_seed8_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed8_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 0:57:07 acc1_pos: 32.292 acc5_pos: 100.000 loss: 1.0394 (1.0394) acc1: 32.2917 (32.2917) acc5: 100.0000 (100.0000) time: 6.6292 data: 4.6804 max mem: 3531 +Test: [ 10/517] eta: 0:05:48 acc1_pos: 6.510 acc5_pos: 100.000 loss: 1.1738 (1.1414) acc1: 21.6146 (22.2538) acc5: 100.0000 (100.0000) time: 0.6870 data: 0.4258 max mem: 3531 +Test: [ 20/517] eta: 0:03:20 acc1_pos: 9.115 acc5_pos: 100.000 loss: 1.1738 (1.1428) acc1: 20.0521 (21.5402) acc5: 100.0000 (100.0000) time: 0.0924 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:02:52 acc1_pos: 11.719 acc5_pos: 100.000 loss: 1.1697 (1.1542) acc1: 17.7083 (20.6317) acc5: 100.0000 (100.0000) time: 0.1713 data: 0.0763 max mem: 3531 +Test: [ 40/517] eta: 0:02:48 acc1_pos: 20.573 acc5_pos: 100.000 loss: 1.1594 (1.1629) acc1: 14.5833 (19.0676) acc5: 100.0000 (100.0000) time: 0.3001 data: 0.2049 max mem: 3531 +Test: [ 50/517] eta: 0:02:36 acc1_pos: 30.275 acc5_pos: 100.000 loss: 1.1161 (1.1132) acc1: 16.1458 (23.0290) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.3067 data: 0.2155 max mem: 3531 +Test: [ 60/517] eta: 0:02:25 acc1_pos: 30.275 acc5_pos: 100.000 loss: 0.3410 (0.9819) acc1: 88.8021 (34.8190) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2461 data: 0.1555 max mem: 3531 +Test: [ 70/517] eta: 0:02:19 acc1_pos: 30.275 acc5_pos: 100.000 loss: 0.2751 (0.8793) acc1: 97.6562 (43.8050) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2528 data: 0.1605 max mem: 3531 +Test: [ 80/517] eta: 0:02:13 acc1_pos: 30.275 acc5_pos: 100.000 loss: 0.2556 (0.8032) acc1: 99.2188 (50.5369) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2644 data: 0.1682 max mem: 3531 +Test: [ 90/517] eta: 0:02:08 acc1_pos: 30.275 acc5_pos: 100.000 loss: 0.2788 (0.7457) acc1: 98.4375 (55.7692) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2599 data: 0.1630 max mem: 3531 +Test: [100/517] eta: 0:02:02 acc1_pos: 30.275 acc5_pos: 100.000 loss: 0.2487 (0.6959) acc1: 98.6979 (60.0789) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2491 data: 0.1545 max mem: 3531 +Test: [110/517] eta: 0:01:58 acc1_pos: 30.275 acc5_pos: 100.000 loss: 0.2217 (0.6522) acc1: 99.7396 (63.6425) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2449 data: 0.1509 max mem: 3531 +Test: [120/517] eta: 0:01:54 acc1_pos: 30.275 acc5_pos: 100.000 loss: 0.2501 (0.6208) acc1: 99.4792 (66.5311) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2649 data: 0.1692 max mem: 3531 +Test: [130/517] eta: 0:01:50 acc1_pos: 30.275 acc5_pos: 100.000 loss: 0.2675 (0.5936) acc1: 98.6979 (68.9289) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2649 data: 0.1704 max mem: 3531 +Test: [140/517] eta: 0:01:47 acc1_pos: 30.275 acc5_pos: 100.000 loss: 0.2150 (0.5657) acc1: 99.7396 (71.1270) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2649 data: 0.1734 max mem: 3531 +Test: [150/517] eta: 0:01:45 acc1_pos: 30.275 acc5_pos: 100.000 loss: 0.2008 (0.5419) acc1: 100.0000 (73.0288) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2899 data: 0.1975 max mem: 3531 +Test: [160/517] eta: 0:01:41 acc1_pos: 30.275 acc5_pos: 100.000 loss: 0.2172 (0.5224) acc1: 100.0000 (74.6862) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2849 data: 0.1917 max mem: 3531 +Test: [170/517] eta: 0:01:38 acc1_pos: 30.275 acc5_pos: 100.000 loss: 0.2257 (0.5046) acc1: 99.7396 (76.1452) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2499 data: 0.1563 max mem: 3531 +Test: [180/517] eta: 0:01:34 acc1_pos: 30.275 acc5_pos: 100.000 loss: 0.2244 (0.4894) acc1: 99.4792 (77.3553) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2499 data: 0.1555 max mem: 3531 +Test: [190/517] eta: 0:01:31 acc1_pos: 30.275 acc5_pos: 100.000 loss: 0.1884 (0.4736) acc1: 99.7396 (78.5340) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2649 data: 0.1665 max mem: 3531 +Test: [200/517] eta: 0:01:28 acc1_pos: 30.275 acc5_pos: 100.000 loss: 0.1898 (0.4599) acc1: 100.0000 (79.5942) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2699 data: 0.1690 max mem: 3531 +Test: [210/517] eta: 0:01:25 acc1_pos: 30.275 acc5_pos: 100.000 loss: 0.2145 (0.4533) acc1: 97.3958 (80.2589) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2649 data: 0.1682 max mem: 3532 +Test: [220/517] eta: 0:01:22 acc1_pos: 30.275 acc5_pos: 100.000 loss: 0.3249 (0.4477) acc1: 92.1875 (80.8034) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2599 data: 0.1646 max mem: 3532 +Test: [230/517] eta: 0:01:19 acc1_pos: 30.275 acc5_pos: 100.000 loss: 0.3105 (0.4407) acc1: 94.5312 (81.5386) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2599 data: 0.1630 max mem: 3532 +Test: [240/517] eta: 0:01:17 acc1_pos: 30.275 acc5_pos: 100.000 loss: 0.2682 (0.4334) acc1: 99.2188 (82.2776) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2891 data: 0.1923 max mem: 3532 +Test: [250/517] eta: 0:01:14 acc1_pos: 30.275 acc5_pos: 100.000 loss: 0.2478 (0.4254) acc1: 99.4792 (82.9557) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2842 data: 0.1867 max mem: 3532 +Test: [260/517] eta: 0:01:11 acc1_pos: 30.275 acc5_pos: 100.000 loss: 0.2194 (0.4178) acc1: 99.4792 (83.5918) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2599 data: 0.1594 max mem: 3532 +Test: [270/517] eta: 0:01:08 acc1_pos: 30.275 acc5_pos: 100.000 loss: 0.2470 (0.4120) acc1: 99.4792 (84.1770) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2649 data: 0.1670 max mem: 3532 +Test: [280/517] eta: 0:01:05 acc1_pos: 30.275 acc5_pos: 100.000 loss: 0.2486 (0.4060) acc1: 99.2188 (84.7105) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2649 data: 0.1713 max mem: 3532 +Test: [290/517] eta: 0:01:02 acc1_pos: 30.275 acc5_pos: 100.000 loss: 0.2400 (0.4003) acc1: 99.2188 (85.2126) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2650 data: 0.1707 max mem: 3532 +Test: [300/517] eta: 0:00:59 acc1_pos: 30.275 acc5_pos: 100.000 loss: 0.2386 (0.3952) acc1: 99.4792 (85.6909) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2599 data: 0.1639 max mem: 3532 +Test: [310/517] eta: 0:00:57 acc1_pos: 30.275 acc5_pos: 100.000 loss: 0.2721 (0.3916) acc1: 99.4792 (86.1142) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2604 data: 0.1645 max mem: 3532 +Test: [320/517] eta: 0:00:54 acc1_pos: 30.275 acc5_pos: 100.000 loss: 0.2721 (0.3867) acc1: 99.4792 (86.5240) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2649 data: 0.1713 max mem: 3532 +Test: [330/517] eta: 0:00:51 acc1_pos: 30.275 acc5_pos: 100.000 loss: 0.2124 (0.3809) acc1: 99.4792 (86.9217) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2693 data: 0.1752 max mem: 3532 +Test: [340/517] eta: 0:00:48 acc1_pos: 30.275 acc5_pos: 100.000 loss: 0.1901 (0.3768) acc1: 99.7396 (87.2289) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2649 data: 0.1681 max mem: 3532 +Test: [350/517] eta: 0:00:45 acc1_pos: 30.275 acc5_pos: 100.000 loss: 0.2640 (0.3748) acc1: 97.6562 (87.5007) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2599 data: 0.1624 max mem: 3532 +Test: [360/517] eta: 0:00:43 acc1_pos: 30.275 acc5_pos: 100.000 loss: 0.2817 (0.3722) acc1: 96.6146 (87.7546) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2699 data: 0.1735 max mem: 3532 +Test: [370/517] eta: 0:00:40 acc1_pos: 30.275 acc5_pos: 100.000 loss: 0.2406 (0.3677) acc1: 99.4792 (88.0805) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2599 data: 0.1635 max mem: 3532 +Test: [380/517] eta: 0:00:37 acc1_pos: 30.275 acc5_pos: 100.000 loss: 0.2114 (0.3643) acc1: 99.7396 (88.3763) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1570 max mem: 3532 +Test: [390/517] eta: 0:00:34 acc1_pos: 30.275 acc5_pos: 100.000 loss: 0.2400 (0.3616) acc1: 99.4792 (88.6322) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2699 data: 0.1709 max mem: 3532 +Test: [400/517] eta: 0:00:31 acc1_pos: 30.275 acc5_pos: 100.000 loss: 0.2436 (0.3583) acc1: 99.7396 (88.9086) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2649 data: 0.1656 max mem: 3532 +Test: [410/517] eta: 0:00:29 acc1_pos: 30.275 acc5_pos: 100.000 loss: 0.2441 (0.3563) acc1: 99.7396 (89.1442) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2499 data: 0.1487 max mem: 3532 +Test: [420/517] eta: 0:00:26 acc1_pos: 30.275 acc5_pos: 100.000 loss: 0.2533 (0.3537) acc1: 99.4792 (89.3953) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2449 data: 0.1438 max mem: 3532 +Test: [430/517] eta: 0:00:23 acc1_pos: 30.275 acc5_pos: 100.000 loss: 0.2465 (0.3511) acc1: 99.4792 (89.6093) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2499 data: 0.1509 max mem: 3532 +Test: [440/517] eta: 0:00:20 acc1_pos: 30.275 acc5_pos: 100.000 loss: 0.2492 (0.3493) acc1: 98.6979 (89.8071) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2399 data: 0.1424 max mem: 3533 +Test: [450/517] eta: 0:00:18 acc1_pos: 30.275 acc5_pos: 100.000 loss: 0.2513 (0.3470) acc1: 98.9583 (90.0199) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2299 data: 0.1335 max mem: 3533 +Test: [460/517] eta: 0:00:15 acc1_pos: 30.275 acc5_pos: 100.000 loss: 0.2513 (0.3452) acc1: 99.4792 (90.2115) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2399 data: 0.1446 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 30.275 acc5_pos: 100.000 loss: 0.2461 (0.3428) acc1: 99.4792 (90.4072) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2399 data: 0.1457 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 30.275 acc5_pos: 100.000 loss: 0.2405 (0.3407) acc1: 99.4792 (90.5985) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2073 data: 0.1138 max mem: 3533 +Test: [490/517] eta: 0:00:07 acc1_pos: 30.275 acc5_pos: 100.000 loss: 0.2414 (0.3386) acc1: 99.7396 (90.7751) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1706 data: 0.0792 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 30.275 acc5_pos: 100.000 loss: 0.2580 (0.3379) acc1: 99.4792 (90.8995) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.1613 data: 0.0715 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 30.275 acc5_pos: 100.000 loss: 0.2637 (0.3367) acc1: 97.9167 (91.0123) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1619 data: 0.0721 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 30.275 acc5_pos: 100.000 loss: 0.2344 (0.3352) acc1: 99.4792 (91.1111) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1592 data: 0.0672 max mem: 3533 +Test: Total time: 0:02:14 (0.2598 s / it) +* Overall Acc@1 91.111 Acc@5 100.000 + target == real Acc@1 19.340 Acc@5 100.000 + target == fake Acc@1 98.557 Acc@5 100.000 + loss 0.335 + PR-AUC 0.9617 ROC-AUC 0.7686 EER 30.06% (thr=0.7593) +Accuracy of the network on the 198427 test images: 91.1% +[rank0]:[W726 05:46:48.383288536 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 05:47:08.811000 1201523 torch/distributed/run.py:766] +W0726 05:47:08.811000 1201523 torch/distributed/run.py:766] ***************************************** +W0726 05:47:08.811000 1201523 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:47:08.811000 1201523 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:47:37.971524254 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:47:37.022464083 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +NCCL version 2.26.2+cuda12.2 +[rank1]:[W726 05:47:38.338524986 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:47:38.482890415 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:47:38.747454334 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:47:38.756338094 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +[rank4]:[W726 05:47:38.781748485 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:47:38.787074427 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n64_seed8_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed8_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:41:59 acc1_pos: 47.656 acc5_pos: 100.000 loss: 0.7639 (0.7639) acc1: 47.6562 (47.6562) acc5: 100.0000 (100.0000) time: 6.4768 data: 5.1379 max mem: 3531 +Test: [ 10/389] eta: 0:04:19 acc1_pos: 95.573 acc5_pos: 100.000 loss: 0.5831 (0.5686) acc1: 77.0833 (68.5369) acc5: 100.0000 (100.0000) time: 0.6843 data: 0.4806 max mem: 3531 +Test: [ 20/389] eta: 0:02:28 acc1_pos: 38.281 acc5_pos: 100.000 loss: 0.5625 (0.5997) acc1: 74.2188 (65.1538) acc5: 100.0000 (100.0000) time: 0.0999 data: 0.0083 max mem: 3531 +Test: [ 30/389] eta: 0:02:16 acc1_pos: 78.906 acc5_pos: 100.000 loss: 0.6469 (0.6212) acc1: 62.7604 (63.1972) acc5: 100.0000 (100.0000) time: 0.2126 data: 0.1172 max mem: 3531 +Test: [ 40/389] eta: 0:02:01 acc1_pos: 25.521 acc5_pos: 100.000 loss: 0.6381 (0.6255) acc1: 63.8021 (63.2368) acc5: 100.0000 (100.0000) time: 0.2882 data: 0.1935 max mem: 3531 +Test: [ 50/389] eta: 0:01:45 acc1_pos: 82.552 acc5_pos: 100.000 loss: 0.5160 (0.6048) acc1: 81.2500 (66.3705) acc5: 100.0000 (100.0000) time: 0.2053 data: 0.1110 max mem: 3531 +Test: [ 60/389] eta: 0:01:37 acc1_pos: 73.438 acc5_pos: 100.000 loss: 0.5147 (0.5915) acc1: 80.2083 (68.3316) acc5: 100.0000 (100.0000) time: 0.1922 data: 0.0967 max mem: 3531 +Test: [ 70/389] eta: 0:01:32 acc1_pos: 69.792 acc5_pos: 100.000 loss: 0.5255 (0.5908) acc1: 76.8229 (68.4822) acc5: 100.0000 (100.0000) time: 0.2392 data: 0.1455 max mem: 3531 +Test: [ 80/389] eta: 0:01:26 acc1_pos: 93.229 acc5_pos: 100.000 loss: 0.5801 (0.5987) acc1: 68.4896 (67.3804) acc5: 100.0000 (100.0000) time: 0.2244 data: 0.1291 max mem: 3531 +Test: [ 90/389] eta: 0:01:21 acc1_pos: 81.250 acc5_pos: 100.000 loss: 0.3836 (0.5752) acc1: 84.6354 (69.7058) acc5: 100.0000 (100.0000) time: 0.2112 data: 0.1165 max mem: 3531 +Test: [100/389] eta: 0:01:16 acc1_pos: 40.885 acc5_pos: 100.000 loss: 0.5102 (0.5882) acc1: 74.2188 (68.2421) acc5: 100.0000 (100.0000) time: 0.2075 data: 0.1142 max mem: 3531 +Test: [110/389] eta: 0:01:12 acc1_pos: 97.135 acc5_pos: 100.000 loss: 0.8051 (0.6059) acc1: 51.0417 (66.5118) acc5: 100.0000 (100.0000) time: 0.1904 data: 0.0979 max mem: 3531 +Test: [120/389] eta: 0:01:09 acc1_pos: 79.688 acc5_pos: 100.000 loss: 0.5392 (0.5986) acc1: 67.7083 (67.2887) acc5: 100.0000 (100.0000) time: 0.2191 data: 0.1285 max mem: 3531 +Test: [130/389] eta: 0:01:04 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.5392 (0.5977) acc1: 72.1354 (67.4559) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1993 data: 0.1081 max mem: 3531 +Test: [140/389] eta: 0:01:01 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.5773 (0.6034) acc1: 66.9271 (67.0767) acc5: 100.0000 (100.0000) acc1_neg: 10.417 acc5_neg: 100.000 time: 0.1840 data: 0.0926 max mem: 3531 +Test: [150/389] eta: 0:00:57 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.6504 (0.6081) acc1: 60.1562 (66.3683) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1951 data: 0.1034 max mem: 3531 +Test: [160/389] eta: 0:00:54 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.6625 (0.6011) acc1: 56.5104 (67.0096) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1814 data: 0.0897 max mem: 3531 +Test: [170/389] eta: 0:00:51 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.3155 (0.5934) acc1: 97.6562 (67.6200) acc5: 100.0000 (100.0000) acc1_neg: 41.927 acc5_neg: 100.000 time: 0.1855 data: 0.0947 max mem: 3531 +Test: [180/389] eta: 0:00:48 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.4292 (0.5914) acc1: 84.8958 (68.0824) acc5: 100.0000 (100.0000) acc1_neg: 66.406 acc5_neg: 100.000 time: 0.1726 data: 0.0809 max mem: 3531 +Test: [190/389] eta: 0:00:45 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.6055 (0.5981) acc1: 66.4062 (67.3388) acc5: 100.0000 (100.0000) acc1_neg: 7.031 acc5_neg: 100.000 time: 0.1630 data: 0.0696 max mem: 3531 +Test: [200/389] eta: 0:00:42 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.8570 (0.6080) acc1: 32.5521 (66.0564) acc5: 100.0000 (100.0000) acc1_neg: 16.667 acc5_neg: 100.000 time: 0.1765 data: 0.0833 max mem: 3531 +Test: [210/389] eta: 0:00:39 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.4802 (0.6039) acc1: 34.8958 (66.3853) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1776 data: 0.0848 max mem: 3532 +Test: [220/389] eta: 0:00:37 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.3186 (0.5924) acc1: 98.4375 (67.5092) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.1738 data: 0.0816 max mem: 3532 +Test: [230/389] eta: 0:00:34 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.4794 (0.5941) acc1: 75.0000 (67.2044) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1739 data: 0.0807 max mem: 3532 +Test: [240/389] eta: 0:00:32 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.5562 (0.5920) acc1: 69.7917 (67.4620) acc5: 100.0000 (100.0000) acc1_neg: 56.510 acc5_neg: 100.000 time: 0.1865 data: 0.0928 max mem: 3532 +Test: [250/389] eta: 0:00:30 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.4669 (0.5853) acc1: 77.0833 (68.2499) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2132 data: 0.1201 max mem: 3532 +Test: [260/389] eta: 0:00:28 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.4363 (0.5836) acc1: 92.7083 (68.6243) acc5: 100.0000 (100.0000) acc1_neg: 16.406 acc5_neg: 100.000 time: 0.2028 data: 0.1089 max mem: 3532 +Test: [270/389] eta: 0:00:25 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.5884 (0.5948) acc1: 59.8958 (67.5440) acc5: 100.0000 (100.0000) acc1_neg: 59.896 acc5_neg: 100.000 time: 0.1799 data: 0.0867 max mem: 3532 +Test: [280/389] eta: 0:00:23 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.6547 (0.6007) acc1: 52.6042 (66.9234) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1907 data: 0.0978 max mem: 3532 +Test: [290/389] eta: 0:00:21 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.6264 (0.5966) acc1: 64.3229 (67.3951) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1965 data: 0.1046 max mem: 3532 +Test: [300/389] eta: 0:00:19 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.5897 (0.6010) acc1: 54.4271 (66.8103) acc5: 100.0000 (100.0000) acc1_neg: 40.885 acc5_neg: 100.000 time: 0.1962 data: 0.1045 max mem: 3532 +Test: [310/389] eta: 0:00:16 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.7664 (0.6095) acc1: 45.0521 (66.0144) acc5: 100.0000 (100.0000) acc1_neg: 6.771 acc5_neg: 100.000 time: 0.1851 data: 0.0931 max mem: 3532 +Test: [320/389] eta: 0:00:14 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.7034 (0.6100) acc1: 50.0000 (65.9552) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1743 data: 0.0830 max mem: 3532 +Test: [330/389] eta: 0:00:12 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.7840 (0.6156) acc1: 40.6250 (65.1671) acc5: 100.0000 (100.0000) acc1_neg: 40.365 acc5_neg: 100.000 time: 0.1674 data: 0.0766 max mem: 3532 +Test: [340/389] eta: 0:00:10 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.4228 (0.6070) acc1: 65.8854 (66.0725) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1446 data: 0.0545 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.3281 (0.6171) acc1: 97.9167 (65.4647) acc5: 100.0000 (100.0000) acc1_neg: 0.521 acc5_neg: 100.000 time: 0.1270 data: 0.0369 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.8812 (0.6269) acc1: 34.1146 (64.4030) acc5: 100.0000 (100.0000) acc1_neg: 35.417 acc5_neg: 100.000 time: 0.1140 data: 0.0237 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.8496 (0.6251) acc1: 35.4167 (64.5833) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1049 data: 0.0146 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.5797 (0.6257) acc1: 71.8750 (64.5170) acc5: 100.0000 (100.0000) acc1_neg: 60.938 acc5_neg: 100.000 time: 0.1007 data: 0.0107 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 67.485 acc5_pos: 100.000 loss: 0.5565 (0.6236) acc1: 71.8750 (64.7357) acc5: 100.0000 (100.0000) acc1_neg: 74.590 acc5_neg: 100.000 time: 0.1024 data: 0.0090 max mem: 3532 +Test: Total time: 0:01:16 (0.1955 s / it) +* Overall Acc@1 64.736 Acc@5 100.000 + target == real Acc@1 67.134 Acc@5 100.000 + target == fake Acc@1 63.621 Acc@5 100.000 + loss 0.624 + PR-AUC 0.8431 ROC-AUC 0.7158 EER 34.78% (thr=0.4890) +Accuracy of the network on the 149358 test images: 64.7% +[rank0]:[W726 05:49:02.325351588 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:49:22.678000 1202082 torch/distributed/run.py:766] +W0726 05:49:22.678000 1202082 torch/distributed/run.py:766] ***************************************** +W0726 05:49:22.678000 1202082 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:49:22.678000 1202082 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:49:53.877590625 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:49:53.992208963 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:49:53.130508186 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W726 05:49:53.135273125 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:49:54.399090128 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:49:54.432439711 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank1]:[W726 05:49:54.437519275 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W726 05:49:54.439574874 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n64_seed8_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed8_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:32:18 acc1_pos: 84.375 acc5_pos: 100.000 loss: 0.3636 (0.3636) acc1: 84.3750 (84.3750) acc5: 100.0000 (100.0000) time: 5.8228 data: 4.4696 max mem: 3531 +Test: [ 10/333] eta: 0:03:17 acc1_pos: 85.677 acc5_pos: 100.000 loss: 0.3048 (0.2953) acc1: 87.7604 (88.2576) acc5: 100.0000 (100.0000) time: 0.6113 data: 0.4066 max mem: 3531 +Test: [ 20/333] eta: 0:01:54 acc1_pos: 93.229 acc5_pos: 100.000 loss: 0.2982 (0.3057) acc1: 87.7604 (87.5496) acc5: 100.0000 (100.0000) time: 0.0929 data: 0.0004 max mem: 3531 +Test: [ 30/333] eta: 0:01:37 acc1_pos: 59.115 acc5_pos: 100.000 loss: 0.4051 (0.4561) acc1: 73.6979 (76.2601) acc5: 100.0000 (100.0000) time: 0.1628 data: 0.0665 max mem: 3531 +Test: [ 40/333] eta: 0:01:31 acc1_pos: 53.646 acc5_pos: 100.000 loss: 0.7940 (0.5432) acc1: 49.7396 (69.1756) acc5: 100.0000 (100.0000) time: 0.2596 data: 0.1645 max mem: 3531 +Test: [ 50/333] eta: 0:01:22 acc1_pos: 51.823 acc5_pos: 100.000 loss: 0.7948 (0.5927) acc1: 51.8229 (65.4463) acc5: 100.0000 (100.0000) time: 0.2451 data: 0.1519 max mem: 3531 +Test: [ 60/333] eta: 0:01:12 acc1_pos: 43.490 acc5_pos: 100.000 loss: 0.7888 (0.6268) acc1: 49.7396 (62.8031) acc5: 100.0000 (100.0000) time: 0.1702 data: 0.0754 max mem: 3531 +Test: [ 70/333] eta: 0:01:07 acc1_pos: 52.344 acc5_pos: 100.000 loss: 0.7888 (0.6485) acc1: 49.7396 (61.2713) acc5: 100.0000 (100.0000) time: 0.1705 data: 0.0748 max mem: 3531 +Test: [ 80/333] eta: 0:01:03 acc1_pos: 49.219 acc5_pos: 100.000 loss: 0.7665 (0.6655) acc1: 49.7396 (59.6772) acc5: 100.0000 (100.0000) time: 0.2049 data: 0.1099 max mem: 3531 +Test: [ 90/333] eta: 0:00:59 acc1_pos: 59.635 acc5_pos: 100.000 loss: 0.7085 (0.6695) acc1: 53.9062 (59.4351) acc5: 100.0000 (100.0000) time: 0.1905 data: 0.0972 max mem: 3531 +Test: [100/333] eta: 0:00:55 acc1_pos: 62.500 acc5_pos: 100.000 loss: 0.6965 (0.6748) acc1: 59.6354 (59.1069) acc5: 100.0000 (100.0000) time: 0.1814 data: 0.0888 max mem: 3531 +Test: [110/333] eta: 0:00:52 acc1_pos: 66.146 acc5_pos: 100.000 loss: 0.7059 (0.6804) acc1: 55.2083 (58.7416) acc5: 100.0000 (100.0000) time: 0.1937 data: 0.1014 max mem: 3531 +Test: [120/333] eta: 0:00:48 acc1_pos: 61.458 acc5_pos: 100.000 loss: 0.7034 (0.6831) acc1: 55.2083 (58.5077) acc5: 100.0000 (100.0000) time: 0.1895 data: 0.0962 max mem: 3531 +Test: [130/333] eta: 0:00:45 acc1_pos: 74.740 acc5_pos: 100.000 loss: 0.6559 (0.6834) acc1: 58.3333 (58.5460) acc5: 100.0000 (100.0000) time: 0.1817 data: 0.0883 max mem: 3531 +Test: [140/333] eta: 0:00:43 acc1_pos: 76.823 acc5_pos: 100.000 loss: 0.6560 (0.6835) acc1: 59.8958 (58.6381) acc5: 100.0000 (100.0000) time: 0.1910 data: 0.0988 max mem: 3531 +Test: [150/333] eta: 0:00:40 acc1_pos: 63.802 acc5_pos: 100.000 loss: 0.7139 (0.6869) acc1: 58.8542 (58.3833) acc5: 100.0000 (100.0000) time: 0.2004 data: 0.1074 max mem: 3531 +Test: [160/333] eta: 0:00:38 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.7174 (0.6929) acc1: 50.7812 (58.0438) acc5: 100.0000 (100.0000) acc1_neg: 50.781 acc5_neg: 100.000 time: 0.1887 data: 0.0949 max mem: 3531 +Test: [170/333] eta: 0:00:35 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.7302 (0.6979) acc1: 56.5104 (57.8551) acc5: 100.0000 (100.0000) acc1_neg: 58.333 acc5_neg: 100.000 time: 0.1803 data: 0.0865 max mem: 3531 +Test: [180/333] eta: 0:00:33 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.7043 (0.6964) acc1: 59.1146 (58.0758) acc5: 100.0000 (100.0000) acc1_neg: 54.167 acc5_neg: 100.000 time: 0.1875 data: 0.0952 max mem: 3531 +Test: [190/333] eta: 0:00:30 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.6489 (0.6923) acc1: 63.2812 (58.5460) acc5: 100.0000 (100.0000) acc1_neg: 63.802 acc5_neg: 100.000 time: 0.1832 data: 0.0907 max mem: 3531 +Test: [200/333] eta: 0:00:28 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.6311 (0.6896) acc1: 64.5833 (58.9734) acc5: 100.0000 (100.0000) acc1_neg: 64.583 acc5_neg: 100.000 time: 0.1813 data: 0.0854 max mem: 3531 +Test: [210/333] eta: 0:00:26 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.6782 (0.6915) acc1: 63.5417 (59.0492) acc5: 100.0000 (100.0000) acc1_neg: 52.865 acc5_neg: 100.000 time: 0.1887 data: 0.0920 max mem: 3532 +Test: [220/333] eta: 0:00:23 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.7457 (0.6961) acc1: 58.0729 (58.8165) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.1914 data: 0.0970 max mem: 3532 +Test: [230/333] eta: 0:00:21 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.5101 (0.6865) acc1: 75.2604 (59.7200) acc5: 100.0000 (100.0000) acc1_neg: 80.729 acc5_neg: 100.000 time: 0.1988 data: 0.1061 max mem: 3532 +Test: [240/333] eta: 0:00:19 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.4950 (0.6826) acc1: 75.2604 (60.0720) acc5: 100.0000 (100.0000) acc1_neg: 67.969 acc5_neg: 100.000 time: 0.2010 data: 0.1083 max mem: 3532 +Test: [250/333] eta: 0:00:17 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.6093 (0.6806) acc1: 67.9688 (60.3264) acc5: 100.0000 (100.0000) acc1_neg: 58.333 acc5_neg: 100.000 time: 0.1941 data: 0.1002 max mem: 3532 +Test: [260/333] eta: 0:00:15 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.6204 (0.6794) acc1: 67.4479 (60.5234) acc5: 100.0000 (100.0000) acc1_neg: 68.750 acc5_neg: 100.000 time: 0.1925 data: 0.0997 max mem: 3532 +Test: [270/333] eta: 0:00:13 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.5927 (0.6753) acc1: 68.7500 (60.9068) acc5: 100.0000 (100.0000) acc1_neg: 68.750 acc5_neg: 100.000 time: 0.1967 data: 0.1031 max mem: 3532 +Test: [280/333] eta: 0:00:11 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.5718 (0.6767) acc1: 68.7500 (60.8133) acc5: 100.0000 (100.0000) acc1_neg: 64.844 acc5_neg: 100.000 time: 0.1946 data: 0.1010 max mem: 3532 +Test: [290/333] eta: 0:00:08 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.8218 (0.6818) acc1: 49.4792 (60.3764) acc5: 100.0000 (100.0000) acc1_neg: 52.083 acc5_neg: 100.000 time: 0.1909 data: 0.0981 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.8218 (0.6860) acc1: 48.9583 (60.1303) acc5: 100.0000 (100.0000) acc1_neg: 54.948 acc5_neg: 100.000 time: 0.1880 data: 0.0953 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.7173 (0.6827) acc1: 58.5938 (60.4795) acc5: 100.0000 (100.0000) acc1_neg: 72.135 acc5_neg: 100.000 time: 0.1810 data: 0.0896 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.5674 (0.6804) acc1: 70.5729 (60.7103) acc5: 100.0000 (100.0000) acc1_neg: 47.656 acc5_neg: 100.000 time: 0.1824 data: 0.0908 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.8078 (0.6865) acc1: 53.1250 (60.3553) acc5: 100.0000 (100.0000) acc1_neg: 51.823 acc5_neg: 100.000 time: 0.1543 data: 0.0635 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 51.316 acc5_pos: 100.000 loss: 0.8500 (0.6877) acc1: 50.7812 (60.2974) acc5: 100.0000 (100.0000) acc1_neg: 49.859 acc5_neg: 100.000 time: 0.1507 data: 0.0566 max mem: 3532 +Test: Total time: 0:01:07 (0.2024 s / it) +* Overall Acc@1 60.297 Acc@5 100.000 + target == real Acc@1 58.378 Acc@5 100.000 + target == fake Acc@1 62.006 Acc@5 100.000 + loss 0.688 + PR-AUC 0.6543 ROC-AUC 0.6435 EER 39.85% (thr=0.5132) +Accuracy of the network on the 127843 test images: 60.3% +[rank0]:[W726 05:51:11.527482859 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:51:31.332000 1202643 torch/distributed/run.py:766] +W0726 05:51:31.332000 1202643 torch/distributed/run.py:766] ***************************************** +W0726 05:51:31.332000 1202643 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:51:31.332000 1202643 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:52:00.896691794 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:52:00.993966486 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:52:01.231641260 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W726 05:52:01.234541858 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:52:01.269991120 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +[rank0]:[W726 05:52:01.420530824 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:52:01.428035690 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:52:01.466653466 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n64_seed8_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed8_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:19:12 acc1_pos: 96.354 acc5_pos: 100.000 loss: 0.2088 (0.2088) acc1: 96.3542 (96.3542) acc5: 100.0000 (100.0000) time: 4.9045 data: 2.9372 max mem: 3531 +Test: [ 10/235] eta: 0:01:58 acc1_pos: 95.573 acc5_pos: 100.000 loss: 0.2103 (0.2113) acc1: 96.3542 (96.0701) acc5: 100.0000 (100.0000) time: 0.5279 data: 0.2674 max mem: 3531 +Test: [ 20/235] eta: 0:01:08 acc1_pos: 97.135 acc5_pos: 100.000 loss: 0.2058 (0.2080) acc1: 96.3542 (96.4286) acc5: 100.0000 (100.0000) time: 0.0904 data: 0.0004 max mem: 3531 +Test: [ 30/235] eta: 0:00:52 acc1_pos: 97.135 acc5_pos: 100.000 loss: 0.2039 (0.2068) acc1: 96.6146 (96.4130) acc5: 100.0000 (100.0000) time: 0.1098 data: 0.0189 max mem: 3531 +Test: [ 40/235] eta: 0:00:47 acc1_pos: 95.833 acc5_pos: 100.000 loss: 0.2051 (0.2063) acc1: 96.6146 (96.4240) acc5: 100.0000 (100.0000) time: 0.1688 data: 0.0771 max mem: 3531 +Test: [ 50/235] eta: 0:00:47 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.2059 (0.2077) acc1: 96.6146 (96.3491) acc5: 100.0000 (100.0000) time: 0.2553 data: 0.1634 max mem: 3531 +Test: [ 60/235] eta: 0:00:44 acc1_pos: 95.052 acc5_pos: 100.000 loss: 0.2140 (0.2089) acc1: 95.8333 (96.3029) acc5: 100.0000 (100.0000) time: 0.2759 data: 0.1833 max mem: 3531 +Test: [ 70/235] eta: 0:00:40 acc1_pos: 98.462 acc5_pos: 100.000 loss: 0.2279 (0.3499) acc1: 95.0521 (87.0489) acc5: 100.0000 (100.0000) acc1_neg: 19.792 acc5_neg: 100.000 time: 0.2157 data: 0.1209 max mem: 3531 +Test: [ 80/235] eta: 0:00:37 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.3350 (0.4733) acc1: 22.1354 (78.9545) acc5: 100.0000 (100.0000) acc1_neg: 19.271 acc5_neg: 100.000 time: 0.1920 data: 0.0963 max mem: 3531 +Test: [ 90/235] eta: 0:00:33 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.3536 (0.5705) acc1: 21.6146 (72.6305) acc5: 100.0000 (100.0000) acc1_neg: 21.094 acc5_neg: 100.000 time: 0.1932 data: 0.1003 max mem: 3531 +Test: [100/235] eta: 0:00:31 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.3510 (0.6464) acc1: 21.0938 (67.6697) acc5: 100.0000 (100.0000) acc1_neg: 18.750 acc5_neg: 100.000 time: 0.1898 data: 0.0951 max mem: 3531 +Test: [110/235] eta: 0:00:28 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.3320 (0.7090) acc1: 21.6146 (63.5417) acc5: 100.0000 (100.0000) acc1_neg: 24.479 acc5_neg: 100.000 time: 0.2022 data: 0.1068 max mem: 3531 +Test: [120/235] eta: 0:00:25 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.3943 (0.7809) acc1: 18.4896 (59.3126) acc5: 100.0000 (100.0000) acc1_neg: 13.021 acc5_neg: 100.000 time: 0.1993 data: 0.1065 max mem: 3531 +Test: [130/235] eta: 0:00:23 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.5898 (0.8428) acc1: 11.9792 (55.6596) acc5: 100.0000 (100.0000) acc1_neg: 10.677 acc5_neg: 100.000 time: 0.1888 data: 0.0954 max mem: 3531 +Test: [140/235] eta: 0:00:20 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.6139 (0.8987) acc1: 10.9375 (52.5044) acc5: 100.0000 (100.0000) acc1_neg: 8.073 acc5_neg: 100.000 time: 0.1725 data: 0.0779 max mem: 3531 +Test: [150/235] eta: 0:00:18 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.6183 (0.9450) acc1: 10.9375 (49.7603) acc5: 100.0000 (100.0000) acc1_neg: 12.240 acc5_neg: 100.000 time: 0.1657 data: 0.0687 max mem: 3531 +Test: [160/235] eta: 0:00:16 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.5952 (0.9855) acc1: 11.1979 (47.3926) acc5: 100.0000 (100.0000) acc1_neg: 12.500 acc5_neg: 100.000 time: 0.1919 data: 0.0963 max mem: 3531 +Test: [170/235] eta: 0:00:13 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.5531 (1.0005) acc1: 13.0208 (46.1806) acc5: 100.0000 (100.0000) acc1_neg: 23.958 acc5_neg: 100.000 time: 0.1938 data: 0.1019 max mem: 3531 +Test: [180/235] eta: 0:00:11 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.1935 (1.0110) acc1: 29.1667 (45.2550) acc5: 100.0000 (100.0000) acc1_neg: 29.427 acc5_neg: 100.000 time: 0.1899 data: 0.0952 max mem: 3531 +Test: [190/235] eta: 0:00:09 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.1814 (1.0207) acc1: 29.1667 (44.3731) acc5: 100.0000 (100.0000) acc1_neg: 29.167 acc5_neg: 100.000 time: 0.2112 data: 0.1167 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.1916 (1.0304) acc1: 27.3438 (43.5025) acc5: 100.0000 (100.0000) acc1_neg: 26.302 acc5_neg: 100.000 time: 0.1979 data: 0.1068 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.2152 (1.0380) acc1: 27.3438 (42.8169) acc5: 100.0000 (100.0000) acc1_neg: 28.646 acc5_neg: 100.000 time: 0.1574 data: 0.0640 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.1870 (1.0449) acc1: 29.4271 (42.1887) acc5: 100.0000 (100.0000) acc1_neg: 29.427 acc5_neg: 100.000 time: 0.1452 data: 0.0521 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.1870 (1.0511) acc1: 29.6875 (41.6554) acc5: 100.0000 (100.0000) acc1_neg: 29.688 acc5_neg: 100.000 time: 0.1409 data: 0.0501 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.1893 (1.0547) acc1: 29.4271 (41.4902) acc5: 100.0000 (100.0000) acc1_neg: 34.783 acc5_neg: 100.000 time: 0.1259 data: 0.0387 max mem: 3532 +Test: Total time: 0:00:47 (0.2001 s / it) +* Overall Acc@1 41.490 Acc@5 100.000 + target == real Acc@1 96.314 Acc@5 100.000 + target == fake Acc@1 21.662 Acc@5 100.000 + loss 1.055 + PR-AUC 0.8815 ROC-AUC 0.7359 EER 32.51% (thr=0.1804) +Accuracy of the network on the 89879 test images: 41.5% +[rank0]:[W726 05:52:59.694683392 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:53:19.045000 1203204 torch/distributed/run.py:766] +W0726 05:53:19.045000 1203204 torch/distributed/run.py:766] ***************************************** +W0726 05:53:19.045000 1203204 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:53:19.045000 1203204 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:53:48.892930677 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:53:49.301293652 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:53:49.420158334 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:53:49.545808302 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:53:49.659639985 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:53:49.676654826 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:53:49.692838879 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:53:49.723066454 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n64_seed8_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed8_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:27:57 acc1_pos: 59.115 acc5_pos: 100.000 loss: 0.6898 (0.6898) acc1: 59.1146 (59.1146) acc5: 100.0000 (100.0000) time: 6.2366 data: 4.9407 max mem: 3531 +Test: [ 10/269] eta: 0:02:48 acc1_pos: 80.990 acc5_pos: 100.000 loss: 0.6119 (0.6199) acc1: 65.8854 (65.8381) acc5: 100.0000 (100.0000) time: 0.6493 data: 0.4495 max mem: 3531 +Test: [ 20/269] eta: 0:01:36 acc1_pos: 80.469 acc5_pos: 100.000 loss: 0.6070 (0.6131) acc1: 67.1875 (66.4187) acc5: 100.0000 (100.0000) time: 0.0942 data: 0.0004 max mem: 3531 +Test: [ 30/269] eta: 0:01:12 acc1_pos: 67.448 acc5_pos: 100.000 loss: 0.5652 (0.5994) acc1: 69.7917 (68.5316) acc5: 100.0000 (100.0000) time: 0.1150 data: 0.0196 max mem: 3531 +Test: [ 40/269] eta: 0:01:02 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.5564 (0.5928) acc1: 71.3542 (69.3280) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.1568 data: 0.0645 max mem: 3531 +Test: [ 50/269] eta: 0:00:55 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.5604 (0.5995) acc1: 69.7917 (68.7653) acc5: 100.0000 (100.0000) acc1_neg: 48.698 acc5_neg: 100.000 time: 0.1746 data: 0.0811 max mem: 3531 +Test: [ 60/269] eta: 0:00:50 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.6747 (0.6134) acc1: 58.8542 (67.3028) acc5: 100.0000 (100.0000) acc1_neg: 53.385 acc5_neg: 100.000 time: 0.1717 data: 0.0735 max mem: 3531 +Test: [ 70/269] eta: 0:00:45 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.6777 (0.6150) acc1: 57.5521 (66.8060) acc5: 100.0000 (100.0000) acc1_neg: 69.792 acc5_neg: 100.000 time: 0.1712 data: 0.0745 max mem: 3531 +Test: [ 80/269] eta: 0:00:42 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.6090 (0.6117) acc1: 66.6667 (66.9303) acc5: 100.0000 (100.0000) acc1_neg: 65.104 acc5_neg: 100.000 time: 0.1668 data: 0.0749 max mem: 3531 +Test: [ 90/269] eta: 0:00:39 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.6179 (0.6226) acc1: 62.7604 (65.5134) acc5: 100.0000 (100.0000) acc1_neg: 58.594 acc5_neg: 100.000 time: 0.1795 data: 0.0839 max mem: 3531 +Test: [100/269] eta: 0:00:36 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.6880 (0.6296) acc1: 55.2083 (64.7870) acc5: 100.0000 (100.0000) acc1_neg: 51.562 acc5_neg: 100.000 time: 0.1862 data: 0.0895 max mem: 3531 +Test: [110/269] eta: 0:00:34 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.7191 (0.6422) acc1: 52.0833 (63.4666) acc5: 100.0000 (100.0000) acc1_neg: 48.698 acc5_neg: 100.000 time: 0.2000 data: 0.1059 max mem: 3531 +Test: [120/269] eta: 0:00:31 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.6933 (0.6416) acc1: 54.9479 (63.4857) acc5: 100.0000 (100.0000) acc1_neg: 59.635 acc5_neg: 100.000 time: 0.2110 data: 0.1159 max mem: 3531 +Test: [130/269] eta: 0:00:28 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.6573 (0.6484) acc1: 55.7292 (62.6352) acc5: 100.0000 (100.0000) acc1_neg: 46.354 acc5_neg: 100.000 time: 0.1612 data: 0.0668 max mem: 3531 +Test: [140/269] eta: 0:00:26 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.7164 (0.6541) acc1: 51.8229 (61.8111) acc5: 100.0000 (100.0000) acc1_neg: 52.604 acc5_neg: 100.000 time: 0.1637 data: 0.0708 max mem: 3531 +Test: [150/269] eta: 0:00:24 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.6643 (0.6512) acc1: 54.6875 (61.9999) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.2118 data: 0.1183 max mem: 3531 +Test: [160/269] eta: 0:00:22 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.6011 (0.6489) acc1: 64.0625 (62.1264) acc5: 100.0000 (100.0000) acc1_neg: 67.969 acc5_neg: 100.000 time: 0.1942 data: 0.0983 max mem: 3531 +Test: [170/269] eta: 0:00:20 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.6549 (0.6522) acc1: 56.5104 (61.7644) acc5: 100.0000 (100.0000) acc1_neg: 66.146 acc5_neg: 100.000 time: 0.1923 data: 0.0982 max mem: 3531 +Test: [180/269] eta: 0:00:18 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.6566 (0.6517) acc1: 57.5521 (61.7662) acc5: 100.0000 (100.0000) acc1_neg: 57.552 acc5_neg: 100.000 time: 0.1948 data: 0.1033 max mem: 3531 +Test: [190/269] eta: 0:00:16 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.6480 (0.6535) acc1: 59.8958 (61.5947) acc5: 100.0000 (100.0000) acc1_neg: 51.562 acc5_neg: 100.000 time: 0.1894 data: 0.0981 max mem: 3531 +Test: [200/269] eta: 0:00:14 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.7402 (0.6594) acc1: 50.2604 (61.0463) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.2054 data: 0.1127 max mem: 3531 +Test: [210/269] eta: 0:00:12 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.6723 (0.6580) acc1: 54.9479 (61.1165) acc5: 100.0000 (100.0000) acc1_neg: 68.490 acc5_neg: 100.000 time: 0.2030 data: 0.1075 max mem: 3532 +Test: [220/269] eta: 0:00:09 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.5962 (0.6559) acc1: 64.0625 (61.3040) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.1985 data: 0.1036 max mem: 3532 +Test: [230/269] eta: 0:00:07 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.6027 (0.6523) acc1: 65.8854 (61.6173) acc5: 100.0000 (100.0000) acc1_neg: 70.573 acc5_neg: 100.000 time: 0.2049 data: 0.1112 max mem: 3532 +Test: [240/269] eta: 0:00:05 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.6376 (0.6553) acc1: 63.5417 (61.3524) acc5: 100.0000 (100.0000) acc1_neg: 58.333 acc5_neg: 100.000 time: 0.1999 data: 0.1041 max mem: 3532 +Test: [250/269] eta: 0:00:03 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.7145 (0.6588) acc1: 53.9062 (60.9998) acc5: 100.0000 (100.0000) acc1_neg: 37.240 acc5_neg: 100.000 time: 0.2038 data: 0.1094 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.7929 (0.6652) acc1: 44.7917 (60.2989) acc5: 100.0000 (100.0000) acc1_neg: 55.729 acc5_neg: 100.000 time: 0.1764 data: 0.0854 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.7405 (0.6647) acc1: 44.7917 (60.3768) acc5: 100.0000 (100.0000) acc1_neg: 81.365 acc5_neg: 100.000 time: 0.1396 data: 0.0450 max mem: 3532 +Test: Total time: 0:00:53 (0.2005 s / it) +* Overall Acc@1 60.377 Acc@5 100.000 + target == real Acc@1 67.825 Acc@5 100.000 + target == fake Acc@1 59.360 Acc@5 100.000 + loss 0.665 + PR-AUC 0.9361 ROC-AUC 0.6853 EER 36.71% (thr=0.4773) +Accuracy of the network on the 103293 test images: 60.4% +[rank0]:[W726 05:54:52.519081593 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T05:54:59Z] eval_done tag=fixed500k_fake_n64_seed8_resampled +[2026-07-26T05:54:59Z] eval_start tag=fixed500k_real_n32_seed9_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n32_classseed9_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T05:54:59Z +hostname=n116-040-213 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n32_classseed9_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n32_seed9_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 05:55:12.970000 1203771 torch/distributed/run.py:766] +W0726 05:55:12.970000 1203771 torch/distributed/run.py:766] ***************************************** +W0726 05:55:12.970000 1203771 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:55:12.970000 1203771 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:55:42.083672278 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:55:43.815250546 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:55:43.939828815 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:55:43.978662768 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:55:43.070614585 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:55:43.109336562 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +[rank7]:[W726 05:55:44.243929406 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:55:44.250766878 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n32_seed9_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n32_classseed9_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:20:18 acc1_pos: 94.010 acc5_pos: 100.000 loss: 0.3972 (0.3972) acc1: 94.0104 (94.0104) acc5: 100.0000 (100.0000) time: 5.2058 data: 3.3828 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/234] eta: 0:02:16 acc1_pos: 86.979 acc5_pos: 100.000 loss: 0.4686 (0.4745) acc1: 86.9792 (83.9489) acc5: 100.0000 (100.0000) time: 0.6097 data: 0.3622 max mem: 3531 +Test: [ 20/234] eta: 0:01:27 acc1_pos: 95.052 acc5_pos: 100.000 loss: 0.4703 (0.4695) acc1: 86.4583 (82.2917) acc5: 100.0000 (100.0000) time: 0.1687 data: 0.0748 max mem: 3531 +Test: [ 30/234] eta: 0:01:12 acc1_pos: 77.604 acc5_pos: 100.000 loss: 0.2171 (0.3991) acc1: 95.0521 (85.7527) acc5: 100.0000 (100.0000) time: 0.2169 data: 0.1209 max mem: 3531 +Test: [ 40/234] eta: 0:01:00 acc1_pos: 96.354 acc5_pos: 100.000 loss: 0.2157 (0.4080) acc1: 95.0521 (84.0003) acc5: 100.0000 (100.0000) time: 0.2103 data: 0.1178 max mem: 3531 +Test: [ 50/234] eta: 0:00:51 acc1_pos: 96.615 acc5_pos: 100.000 loss: 0.1767 (0.3613) acc1: 95.3125 (86.3613) acc5: 100.0000 (100.0000) time: 0.1552 data: 0.0646 max mem: 3531 +Test: [ 60/234] eta: 0:00:46 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.1406 (0.3190) acc1: 97.6562 (88.5673) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1733 data: 0.0831 max mem: 3531 +Test: [ 70/234] eta: 0:00:43 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.1014 (0.3392) acc1: 100.0000 (86.9535) acc5: 100.0000 (100.0000) acc1_neg: 66.406 acc5_neg: 100.000 time: 0.2386 data: 0.1461 max mem: 3531 +Test: [ 80/234] eta: 0:00:38 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.6163 (0.3614) acc1: 66.6667 (85.6031) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.2051 data: 0.1118 max mem: 3531 +Test: [ 90/234] eta: 0:00:34 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.3062 (0.3460) acc1: 88.8021 (86.4469) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.1523 data: 0.0602 max mem: 3531 +Test: [100/234] eta: 0:00:32 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.3062 (0.3586) acc1: 84.3750 (85.5018) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.2056 data: 0.1123 max mem: 3531 +Test: [110/234] eta: 0:00:29 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.4429 (0.3672) acc1: 75.5208 (84.8888) acc5: 100.0000 (100.0000) acc1_neg: 65.625 acc5_neg: 100.000 time: 0.2241 data: 0.1311 max mem: 3531 +Test: [120/234] eta: 0:00:26 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.4569 (0.3827) acc1: 79.1667 (83.8606) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.1885 data: 0.0951 max mem: 3531 +Test: [130/234] eta: 0:00:23 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.5277 (0.3952) acc1: 74.4792 (83.0431) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.1643 data: 0.0707 max mem: 3531 +Test: [140/234] eta: 0:00:21 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.5366 (0.4229) acc1: 72.3958 (81.6452) acc5: 100.0000 (100.0000) acc1_neg: 39.062 acc5_neg: 100.000 time: 0.1699 data: 0.0769 max mem: 3531 +Test: [150/234] eta: 0:00:18 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.4984 (0.4340) acc1: 74.4792 (80.9189) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.1970 data: 0.1035 max mem: 3531 +Test: [160/234] eta: 0:00:16 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.4916 (0.4386) acc1: 74.4792 (80.6305) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.1904 data: 0.0984 max mem: 3531 +Test: [170/234] eta: 0:00:14 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.5596 (0.4459) acc1: 70.3125 (80.0332) acc5: 100.0000 (100.0000) acc1_neg: 68.750 acc5_neg: 100.000 time: 0.1935 data: 0.1005 max mem: 3531 +Test: [180/234] eta: 0:00:11 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.5596 (0.4583) acc1: 69.7917 (79.2400) acc5: 100.0000 (100.0000) acc1_neg: 38.021 acc5_neg: 100.000 time: 0.2022 data: 0.1084 max mem: 3531 +Test: [190/234] eta: 0:00:09 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.5800 (0.4660) acc1: 69.0104 (78.6486) acc5: 100.0000 (100.0000) acc1_neg: 68.490 acc5_neg: 100.000 time: 0.1842 data: 0.0919 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.5728 (0.4569) acc1: 70.5729 (79.1563) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1750 data: 0.0835 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.1126 (0.4422) acc1: 99.7396 (80.0899) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1816 data: 0.0890 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.2113 (0.4332) acc1: 97.3958 (80.7563) acc5: 100.0000 (100.0000) acc1_neg: 76.302 acc5_neg: 100.000 time: 0.1689 data: 0.0764 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.4572 (0.4550) acc1: 43.7500 (78.9187) acc5: 100.0000 (100.0000) acc1_neg: 40.104 acc5_neg: 100.000 time: 0.1391 data: 0.0488 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.9191 (0.4609) acc1: 39.8438 (78.4741) acc5: 100.0000 (100.0000) acc1_neg: 41.245 acc5_neg: 100.000 time: 0.1297 data: 0.0384 max mem: 3532 +Test: Total time: 0:00:48 (0.2064 s / it) +* Overall Acc@1 78.474 Acc@5 100.000 + target == real Acc@1 86.426 Acc@5 100.000 + target == fake Acc@1 76.233 Acc@5 100.000 + loss 0.461 + PR-AUC 0.9679 ROC-AUC 0.8910 EER 19.41% (thr=0.4412) +Accuracy of the network on the 89729 test images: 78.5% +[rank0]:[W726 05:56:41.745752114 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:57:01.628000 1204330 torch/distributed/run.py:766] +W0726 05:57:01.628000 1204330 torch/distributed/run.py:766] ***************************************** +W0726 05:57:01.628000 1204330 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:57:01.628000 1204330 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:57:31.017802211 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:57:32.426502351 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:57:32.539411957 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:57:32.845259247 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:57:32.900055229 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:57:32.905389873 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:57:32.930940835 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:57:32.001073949 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n32_seed9_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n32_classseed9_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:51:08 acc1_pos: 55.990 acc5_pos: 100.000 loss: 0.6930 (0.6930) acc1: 55.9896 (55.9896) acc5: 100.0000 (100.0000) time: 7.3181 data: 5.6905 max mem: 3531 +Test: [ 10/2387] eta: 0:29:46 acc1_pos: 37.240 acc5_pos: 100.000 loss: 0.7581 (0.7741) acc1: 46.3542 (48.1771) acc5: 100.0000 (100.0000) time: 0.7515 data: 0.5177 max mem: 3531 +Test: [ 20/2387] eta: 0:17:21 acc1_pos: 42.969 acc5_pos: 100.000 loss: 0.7581 (0.7812) acc1: 45.8333 (47.5322) acc5: 100.0000 (100.0000) time: 0.0960 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:15:15 acc1_pos: 32.812 acc5_pos: 100.000 loss: 0.7961 (0.7920) acc1: 42.9688 (46.7910) acc5: 100.0000 (100.0000) time: 0.1886 data: 0.0913 max mem: 3531 +Test: [ 40/2387] eta: 0:14:26 acc1_pos: 63.542 acc5_pos: 100.000 loss: 0.7878 (0.7917) acc1: 44.7917 (46.5320) acc5: 100.0000 (100.0000) time: 0.2949 data: 0.1981 max mem: 3531 +Test: [ 50/2387] eta: 0:13:18 acc1_pos: 48.698 acc5_pos: 100.000 loss: 0.7611 (0.7783) acc1: 47.3958 (47.8299) acc5: 100.0000 (100.0000) time: 0.2699 data: 0.1745 max mem: 3531 +Test: [ 60/2387] eta: 0:12:40 acc1_pos: 69.531 acc5_pos: 100.000 loss: 0.7090 (0.7750) acc1: 52.8646 (48.6253) acc5: 100.0000 (100.0000) time: 0.2402 data: 0.1470 max mem: 3531 +Test: [ 70/2387] eta: 0:12:05 acc1_pos: 47.396 acc5_pos: 100.000 loss: 0.7742 (0.7779) acc1: 49.2188 (48.3751) acc5: 100.0000 (100.0000) time: 0.2406 data: 0.1479 max mem: 3531 +Test: [ 80/2387] eta: 0:11:38 acc1_pos: 77.083 acc5_pos: 100.000 loss: 0.7404 (0.7598) acc1: 54.9479 (50.5498) acc5: 100.0000 (100.0000) time: 0.2296 data: 0.1322 max mem: 3531 +Test: [ 90/2387] eta: 0:11:15 acc1_pos: 80.208 acc5_pos: 100.000 loss: 0.5884 (0.7347) acc1: 73.9583 (53.5113) acc5: 100.0000 (100.0000) time: 0.2256 data: 0.1283 max mem: 3531 +Test: [ 100/2387] eta: 0:10:57 acc1_pos: 78.646 acc5_pos: 100.000 loss: 0.5228 (0.7202) acc1: 78.1250 (55.4791) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1268 max mem: 3531 +Test: [ 110/2387] eta: 0:10:42 acc1_pos: 69.792 acc5_pos: 100.000 loss: 0.5360 (0.7035) acc1: 78.1250 (57.5169) acc5: 100.0000 (100.0000) time: 0.2285 data: 0.1307 max mem: 3531 +Test: [ 120/2387] eta: 0:10:28 acc1_pos: 67.188 acc5_pos: 100.000 loss: 0.5541 (0.6928) acc1: 70.8333 (58.6583) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1317 max mem: 3531 +Test: [ 130/2387] eta: 0:10:17 acc1_pos: 72.656 acc5_pos: 100.000 loss: 0.5771 (0.6834) acc1: 71.3542 (59.8859) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1297 max mem: 3531 +Test: [ 140/2387] eta: 0:10:07 acc1_pos: 85.417 acc5_pos: 100.000 loss: 0.5759 (0.6753) acc1: 72.9167 (60.7861) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1347 max mem: 3531 +Test: [ 150/2387] eta: 0:10:02 acc1_pos: 72.656 acc5_pos: 100.000 loss: 0.5396 (0.6663) acc1: 73.6979 (61.8533) acc5: 100.0000 (100.0000) time: 0.2412 data: 0.1490 max mem: 3531 +Test: [ 160/2387] eta: 0:09:48 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.5201 (0.6541) acc1: 77.8646 (63.0872) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.2221 data: 0.1309 max mem: 3531 +Test: [ 170/2387] eta: 0:09:37 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3579 (0.6362) acc1: 89.3229 (64.7250) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1936 data: 0.1005 max mem: 3531 +Test: [ 180/2387] eta: 0:09:28 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3442 (0.6202) acc1: 90.8854 (66.1775) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.2027 data: 0.1079 max mem: 3531 +Test: [ 190/2387] eta: 0:09:18 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3409 (0.6055) acc1: 90.8854 (67.4888) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.2009 data: 0.1067 max mem: 3531 +Test: [ 200/2387] eta: 0:09:10 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3395 (0.5924) acc1: 91.4062 (68.6800) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1999 data: 0.1056 max mem: 3531 +Test: [ 210/2387] eta: 0:09:02 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3436 (0.5810) acc1: 91.6667 (69.7312) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.2038 data: 0.1087 max mem: 3532 +Test: [ 220/2387] eta: 0:08:55 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3497 (0.5704) acc1: 90.8854 (70.7049) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.1999 data: 0.1042 max mem: 3532 +Test: [ 230/2387] eta: 0:08:45 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3474 (0.5607) acc1: 90.8854 (71.5954) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1855 data: 0.0915 max mem: 3532 +Test: [ 240/2387] eta: 0:08:40 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3541 (0.5550) acc1: 89.8438 (72.1700) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.1899 data: 0.0957 max mem: 3532 +Test: [ 250/2387] eta: 0:08:33 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3291 (0.5392) acc1: 87.5000 (73.1989) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1993 data: 0.1016 max mem: 3532 +Test: [ 260/2387] eta: 0:08:25 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1245 (0.5229) acc1: 100.0000 (74.2257) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0824 max mem: 3532 +Test: [ 270/2387] eta: 0:08:20 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1160 (0.5079) acc1: 100.0000 (75.1749) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0952 max mem: 3532 +Test: [ 280/2387] eta: 0:08:17 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1180 (0.4941) acc1: 100.0000 (76.0584) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2149 data: 0.1205 max mem: 3532 +Test: [ 290/2387] eta: 0:08:07 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1149 (0.4812) acc1: 100.0000 (76.8793) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1751 data: 0.0801 max mem: 3532 +Test: [ 300/2387] eta: 0:08:04 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1141 (0.4692) acc1: 100.0000 (77.6448) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0786 max mem: 3532 +Test: [ 310/2387] eta: 0:08:02 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1307 (0.4624) acc1: 100.0000 (78.2807) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2296 data: 0.1347 max mem: 3532 +Test: [ 320/2387] eta: 0:07:56 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.4170 (0.4612) acc1: 94.5312 (78.7545) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2099 data: 0.1155 max mem: 3532 +Test: [ 330/2387] eta: 0:07:51 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.4195 (0.4598) acc1: 94.0104 (79.2202) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1799 data: 0.0830 max mem: 3532 +Test: [ 340/2387] eta: 0:07:48 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.4132 (0.4585) acc1: 94.0104 (79.6562) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1999 data: 0.1027 max mem: 3532 +Test: [ 350/2387] eta: 0:07:44 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.4011 (0.4525) acc1: 95.8333 (80.2135) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2149 data: 0.1206 max mem: 3532 +Test: [ 360/2387] eta: 0:07:38 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2750 (0.4478) acc1: 96.3542 (80.6440) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1849 data: 0.0897 max mem: 3532 +Test: [ 370/2387] eta: 0:07:36 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2902 (0.4436) acc1: 95.3125 (81.0331) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1905 data: 0.0952 max mem: 3532 +Test: [ 380/2387] eta: 0:07:32 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2902 (0.4395) acc1: 95.3125 (81.4106) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2052 data: 0.1140 max mem: 3532 +Test: [ 390/2387] eta: 0:07:24 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2866 (0.4356) acc1: 95.0521 (81.7628) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1559 data: 0.0625 max mem: 3532 +Test: [ 400/2387] eta: 0:07:21 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2870 (0.4319) acc1: 94.7917 (82.0929) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1661 data: 0.0703 max mem: 3532 +Test: [ 410/2387] eta: 0:07:19 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2870 (0.4284) acc1: 94.7917 (82.4146) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2183 data: 0.1207 max mem: 3532 +Test: [ 420/2387] eta: 0:07:14 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2860 (0.4249) acc1: 95.3125 (82.7296) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1985 data: 0.1030 max mem: 3532 +Test: [ 430/2387] eta: 0:07:12 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2897 (0.4229) acc1: 94.7917 (82.9376) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.1915 data: 0.0998 max mem: 3532 +Test: [ 440/2387] eta: 0:07:09 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2947 (0.4179) acc1: 91.9271 (83.2400) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2078 data: 0.1157 max mem: 3533 +Test: [ 450/2387] eta: 0:07:04 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1026 (0.4110) acc1: 100.0000 (83.6117) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1783 data: 0.0819 max mem: 3533 +Test: [ 460/2387] eta: 0:07:01 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1018 (0.4043) acc1: 100.0000 (83.9671) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1739 data: 0.0773 max mem: 3533 +Test: [ 470/2387] eta: 0:06:57 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1021 (0.3978) acc1: 100.0000 (84.3075) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0966 max mem: 3533 +Test: [ 480/2387] eta: 0:06:53 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1026 (0.3917) acc1: 100.0000 (84.6338) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1835 data: 0.0880 max mem: 3533 +Test: [ 490/2387] eta: 0:06:50 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1026 (0.3858) acc1: 100.0000 (84.9468) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1820 data: 0.0866 max mem: 3533 +Test: [ 500/2387] eta: 0:06:48 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1027 (0.3802) acc1: 100.0000 (85.2472) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2052 data: 0.1123 max mem: 3533 +Test: [ 510/2387] eta: 0:06:45 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1029 (0.3752) acc1: 100.0000 (85.5359) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2078 data: 0.1148 max mem: 3533 +Test: [ 520/2387] eta: 0:06:41 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1848 (0.3717) acc1: 100.0000 (85.8135) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1793 data: 0.0847 max mem: 3533 +Test: [ 530/2387] eta: 0:06:38 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1858 (0.3682) acc1: 100.0000 (86.0802) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1819 data: 0.0883 max mem: 3533 +Test: [ 540/2387] eta: 0:06:35 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1849 (0.3648) acc1: 100.0000 (86.3375) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1945 data: 0.1004 max mem: 3533 +Test: [ 550/2387] eta: 0:06:32 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1849 (0.3615) acc1: 100.0000 (86.5855) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1795 data: 0.0844 max mem: 3533 +Test: [ 560/2387] eta: 0:06:28 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1854 (0.3584) acc1: 100.0000 (86.8237) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1625 data: 0.0674 max mem: 3533 +Test: [ 570/2387] eta: 0:06:26 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1845 (0.3553) acc1: 100.0000 (87.0531) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1934 data: 0.0989 max mem: 3533 +Test: [ 580/2387] eta: 0:06:23 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1846 (0.3524) acc1: 100.0000 (87.2759) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2073 data: 0.1132 max mem: 3533 +Test: [ 590/2387] eta: 0:06:20 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1852 (0.3496) acc1: 100.0000 (87.4912) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1775 data: 0.0838 max mem: 3533 +Test: [ 600/2387] eta: 0:06:16 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1900 (0.3470) acc1: 100.0000 (87.6976) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1699 data: 0.0785 max mem: 3533 +Test: [ 610/2387] eta: 0:06:14 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2167 (0.3456) acc1: 98.9583 (87.8648) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1824 data: 0.0895 max mem: 3533 +Test: [ 620/2387] eta: 0:06:11 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2587 (0.3443) acc1: 97.6562 (88.0221) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1866 data: 0.0930 max mem: 3533 +Test: [ 630/2387] eta: 0:06:08 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2617 (0.3429) acc1: 97.6562 (88.1772) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1849 data: 0.0937 max mem: 3533 +Test: [ 640/2387] eta: 0:06:05 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2617 (0.3417) acc1: 97.6562 (88.3243) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1862 data: 0.0947 max mem: 3533 +Test: [ 650/2387] eta: 0:06:03 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2613 (0.3405) acc1: 97.9167 (88.4725) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1870 data: 0.0933 max mem: 3533 +Test: [ 660/2387] eta: 0:06:01 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2627 (0.3394) acc1: 97.9167 (88.6149) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1996 data: 0.1059 max mem: 3533 +Test: [ 670/2387] eta: 0:05:58 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2609 (0.3382) acc1: 97.6562 (88.7497) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2086 data: 0.1167 max mem: 3534 +Test: [ 680/2387] eta: 0:05:56 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2664 (0.3380) acc1: 97.6562 (88.8805) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1950 data: 0.1031 max mem: 3534 +Test: [ 690/2387] eta: 0:05:54 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3538 (0.3391) acc1: 96.8750 (88.8096) acc5: 100.0000 (100.0000) acc1_neg: 76.302 acc5_neg: 100.000 time: 0.1954 data: 0.1031 max mem: 3534 +Test: [ 700/2387] eta: 0:05:51 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.4273 (0.3404) acc1: 79.9479 (88.6747) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.1973 data: 0.1055 max mem: 3534 +Test: [ 710/2387] eta: 0:05:49 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.4287 (0.3417) acc1: 79.9479 (88.5501) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.1960 data: 0.1015 max mem: 3534 +Test: [ 720/2387] eta: 0:05:46 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.4287 (0.3429) acc1: 79.9479 (88.4344) acc5: 100.0000 (100.0000) acc1_neg: 79.427 acc5_neg: 100.000 time: 0.1947 data: 0.1001 max mem: 3534 +Test: [ 730/2387] eta: 0:05:44 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.4323 (0.3442) acc1: 79.1667 (88.3012) acc5: 100.0000 (100.0000) acc1_neg: 77.344 acc5_neg: 100.000 time: 0.1945 data: 0.1034 max mem: 3534 +Test: [ 740/2387] eta: 0:05:42 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.4323 (0.3454) acc1: 79.4271 (88.1857) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.1999 data: 0.1091 max mem: 3534 +Test: [ 750/2387] eta: 0:05:40 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.4253 (0.3465) acc1: 80.2083 (88.0767) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.2154 data: 0.1223 max mem: 3534 +Test: [ 760/2387] eta: 0:05:38 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.4196 (0.3473) acc1: 80.2083 (87.9924) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.2151 data: 0.1217 max mem: 3534 +Test: [ 770/2387] eta: 0:05:36 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3604 (0.3468) acc1: 84.8958 (88.0212) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2039 data: 0.1118 max mem: 3534 +Test: [ 780/2387] eta: 0:05:34 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2534 (0.3451) acc1: 94.0104 (88.1295) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2146 data: 0.1207 max mem: 3534 +Test: [ 790/2387] eta: 0:05:32 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2123 (0.3435) acc1: 97.6562 (88.2075) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.2030 data: 0.1060 max mem: 3534 +Test: [ 800/2387] eta: 0:05:29 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1931 (0.3418) acc1: 97.9167 (88.3281) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1964 data: 0.1014 max mem: 3534 +Test: [ 810/2387] eta: 0:05:28 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2022 (0.3404) acc1: 97.9167 (88.4299) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2199 data: 0.1285 max mem: 3534 +Test: [ 820/2387] eta: 0:05:25 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2390 (0.3393) acc1: 96.0938 (88.5096) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.2083 data: 0.1118 max mem: 3534 +Test: [ 830/2387] eta: 0:05:23 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2923 (0.3402) acc1: 88.2812 (88.4336) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1952 data: 0.0983 max mem: 3534 +Test: [ 840/2387] eta: 0:05:21 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3868 (0.3406) acc1: 84.6354 (88.4522) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.2023 data: 0.1105 max mem: 3534 +Test: [ 850/2387] eta: 0:05:18 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3602 (0.3409) acc1: 90.3646 (88.4777) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1862 data: 0.0952 max mem: 3534 +Test: [ 860/2387] eta: 0:05:15 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3453 (0.3409) acc1: 92.1875 (88.5344) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1675 data: 0.0744 max mem: 3534 +Test: [ 870/2387] eta: 0:05:13 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3420 (0.3408) acc1: 94.0104 (88.5859) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1750 data: 0.0793 max mem: 3534 +Test: [ 880/2387] eta: 0:05:11 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3345 (0.3410) acc1: 93.7500 (88.6206) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2114 data: 0.1170 max mem: 3534 +Test: [ 890/2387] eta: 0:05:09 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3357 (0.3411) acc1: 93.7500 (88.6773) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2014 data: 0.1093 max mem: 3534 +Test: [ 900/2387] eta: 0:05:06 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3812 (0.3452) acc1: 87.2396 (88.3581) acc5: 100.0000 (100.0000) acc1_neg: 55.469 acc5_neg: 100.000 time: 0.1606 data: 0.0686 max mem: 3535 +Test: [ 910/2387] eta: 0:05:03 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.8003 (0.3503) acc1: 50.7812 (87.9488) acc5: 100.0000 (100.0000) acc1_neg: 50.260 acc5_neg: 100.000 time: 0.1516 data: 0.0596 max mem: 3535 +Test: [ 920/2387] eta: 0:05:00 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.8129 (0.3552) acc1: 50.2604 (87.5396) acc5: 100.0000 (100.0000) acc1_neg: 47.656 acc5_neg: 100.000 time: 0.1690 data: 0.0772 max mem: 3535 +Test: [ 930/2387] eta: 0:04:58 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.8123 (0.3601) acc1: 51.0417 (87.1587) acc5: 100.0000 (100.0000) acc1_neg: 47.656 acc5_neg: 100.000 time: 0.1821 data: 0.0895 max mem: 3535 +Test: [ 940/2387] eta: 0:04:55 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.8126 (0.3648) acc1: 51.8229 (86.7783) acc5: 100.0000 (100.0000) acc1_neg: 49.219 acc5_neg: 100.000 time: 0.1757 data: 0.0838 max mem: 3535 +Test: [ 950/2387] eta: 0:04:53 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.7971 (0.3694) acc1: 50.5208 (86.3956) acc5: 100.0000 (100.0000) acc1_neg: 52.865 acc5_neg: 100.000 time: 0.1827 data: 0.0913 max mem: 3535 +Test: [ 960/2387] eta: 0:04:51 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.7906 (0.3737) acc1: 51.8229 (86.0426) acc5: 100.0000 (100.0000) acc1_neg: 52.083 acc5_neg: 100.000 time: 0.1797 data: 0.0865 max mem: 3535 +Test: [ 970/2387] eta: 0:04:48 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.7943 (0.3782) acc1: 51.8229 (85.6800) acc5: 100.0000 (100.0000) acc1_neg: 51.562 acc5_neg: 100.000 time: 0.1737 data: 0.0791 max mem: 3535 +Test: [ 980/2387] eta: 0:04:46 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.7791 (0.3793) acc1: 52.0833 (85.5927) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.1723 data: 0.0763 max mem: 3535 +Test: [ 990/2387] eta: 0:04:44 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2787 (0.3783) acc1: 91.9271 (85.6592) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1849 data: 0.0888 max mem: 3535 +Test: [1000/2387] eta: 0:04:41 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2075 (0.3767) acc1: 98.6979 (85.7694) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1807 data: 0.0855 max mem: 3535 +Test: [1010/2387] eta: 0:04:38 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2276 (0.3760) acc1: 96.8750 (85.8218) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.1658 data: 0.0724 max mem: 3535 +Test: [1020/2387] eta: 0:04:36 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2851 (0.3754) acc1: 93.2292 (85.8574) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1788 data: 0.0869 max mem: 3535 +Test: [1030/2387] eta: 0:04:34 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2380 (0.3740) acc1: 96.8750 (85.9691) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1941 data: 0.0992 max mem: 3535 +Test: [1040/2387] eta: 0:04:32 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2052 (0.3723) acc1: 98.6979 (86.0929) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1805 data: 0.0858 max mem: 3535 +Test: [1050/2387] eta: 0:04:29 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1969 (0.3707) acc1: 99.2188 (86.2135) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1649 data: 0.0727 max mem: 3535 +Test: [1060/2387] eta: 0:04:27 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1985 (0.3693) acc1: 98.4375 (86.3184) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1700 data: 0.0755 max mem: 3535 +Test: [1070/2387] eta: 0:04:24 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2306 (0.3684) acc1: 96.8750 (86.3883) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.1649 data: 0.0701 max mem: 3535 +Test: [1080/2387] eta: 0:04:22 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3146 (0.3684) acc1: 89.5833 (86.3832) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.1694 data: 0.0760 max mem: 3535 +Test: [1090/2387] eta: 0:04:20 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3074 (0.3675) acc1: 89.5833 (86.4435) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1799 data: 0.0868 max mem: 3535 +Test: [1100/2387] eta: 0:04:17 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2549 (0.3665) acc1: 93.7500 (86.5071) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1795 data: 0.0863 max mem: 3535 +Test: [1110/2387] eta: 0:04:15 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2549 (0.3659) acc1: 94.2708 (86.5329) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1849 data: 0.0886 max mem: 3535 +Test: [1120/2387] eta: 0:04:13 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2714 (0.3651) acc1: 90.8854 (86.5810) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1813 data: 0.0849 max mem: 3536 +Test: [1130/2387] eta: 0:04:11 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2688 (0.3642) acc1: 92.9688 (86.6451) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1768 data: 0.0831 max mem: 3536 +Test: [1140/2387] eta: 0:04:09 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3176 (0.3642) acc1: 89.5833 (86.6535) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.1945 data: 0.0995 max mem: 3536 +Test: [1150/2387] eta: 0:04:07 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3552 (0.3640) acc1: 89.3229 (86.6939) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1901 data: 0.0952 max mem: 3536 +Test: [1160/2387] eta: 0:04:04 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2648 (0.3631) acc1: 94.2708 (86.7715) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1817 data: 0.0893 max mem: 3536 +Test: [1170/2387] eta: 0:04:02 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2551 (0.3621) acc1: 97.3958 (86.8644) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1887 data: 0.0975 max mem: 3536 +Test: [1180/2387] eta: 0:04:00 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2551 (0.3612) acc1: 97.3958 (86.9490) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1950 data: 0.1040 max mem: 3536 +Test: [1190/2387] eta: 0:03:58 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2350 (0.3602) acc1: 97.3958 (87.0380) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1943 data: 0.1017 max mem: 3536 +Test: [1200/2387] eta: 0:03:56 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2383 (0.3593) acc1: 97.3958 (87.1277) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1982 data: 0.1053 max mem: 3536 +Test: [1210/2387] eta: 0:03:54 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2373 (0.3583) acc1: 99.4792 (87.2151) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1946 data: 0.1028 max mem: 3536 +Test: [1220/2387] eta: 0:03:52 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1504 (0.3565) acc1: 99.4792 (87.3153) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1840 data: 0.0919 max mem: 3536 +Test: [1230/2387] eta: 0:03:50 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1357 (0.3546) acc1: 100.0000 (87.4169) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1778 data: 0.0862 max mem: 3536 +Test: [1240/2387] eta: 0:03:48 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1271 (0.3528) acc1: 100.0000 (87.5162) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1797 data: 0.0881 max mem: 3536 +Test: [1250/2387] eta: 0:03:45 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1233 (0.3510) acc1: 100.0000 (87.6145) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1770 data: 0.0857 max mem: 3536 +Test: [1260/2387] eta: 0:03:43 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1236 (0.3492) acc1: 99.7396 (87.7096) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1733 data: 0.0819 max mem: 3536 +Test: [1270/2387] eta: 0:03:41 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1326 (0.3476) acc1: 99.7396 (87.8045) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1827 data: 0.0905 max mem: 3536 +Test: [1280/2387] eta: 0:03:39 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1368 (0.3460) acc1: 99.7396 (87.8950) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1866 data: 0.0924 max mem: 3536 +Test: [1290/2387] eta: 0:03:37 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1453 (0.3445) acc1: 99.2188 (87.9680) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1840 data: 0.0895 max mem: 3536 +Test: [1300/2387] eta: 0:03:35 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1481 (0.3430) acc1: 98.9583 (88.0465) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1851 data: 0.0932 max mem: 3536 +Test: [1310/2387] eta: 0:03:33 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1511 (0.3417) acc1: 98.6979 (88.1204) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1930 data: 0.1009 max mem: 3536 +Test: [1320/2387] eta: 0:03:31 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1709 (0.3404) acc1: 98.6979 (88.1931) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1914 data: 0.0972 max mem: 3536 +Test: [1330/2387] eta: 0:03:29 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1862 (0.3398) acc1: 96.8750 (88.2337) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1825 data: 0.0874 max mem: 3536 +Test: [1340/2387] eta: 0:03:27 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2873 (0.3398) acc1: 89.0625 (88.2263) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1841 data: 0.0913 max mem: 3536 +Test: [1350/2387] eta: 0:03:25 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.0859 (0.3379) acc1: 100.0000 (88.3134) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1888 data: 0.0983 max mem: 3537 +Test: [1360/2387] eta: 0:03:23 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.0735 (0.3359) acc1: 100.0000 (88.3993) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1941 data: 0.1023 max mem: 3537 +Test: [1370/2387] eta: 0:03:21 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.0719 (0.3340) acc1: 100.0000 (88.4839) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1934 data: 0.1006 max mem: 3537 +Test: [1380/2387] eta: 0:03:18 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.0724 (0.3322) acc1: 100.0000 (88.5673) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1671 data: 0.0724 max mem: 3537 +Test: [1390/2387] eta: 0:03:16 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.0724 (0.3303) acc1: 100.0000 (88.6495) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1610 data: 0.0640 max mem: 3537 +Test: [1400/2387] eta: 0:03:14 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.0769 (0.3285) acc1: 100.0000 (88.7305) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1659 data: 0.0721 max mem: 3537 +Test: [1410/2387] eta: 0:03:12 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.0834 (0.3279) acc1: 100.0000 (88.7661) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1541 data: 0.0604 max mem: 3537 +Test: [1420/2387] eta: 0:03:09 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2819 (0.3277) acc1: 92.1875 (88.7728) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1496 data: 0.0552 max mem: 3537 +Test: [1430/2387] eta: 0:03:07 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2854 (0.3275) acc1: 91.9271 (88.7897) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1564 data: 0.0622 max mem: 3537 +Test: [1440/2387] eta: 0:03:05 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2950 (0.3273) acc1: 91.9271 (88.8052) acc5: 100.0000 (100.0000) acc1_neg: 74.740 acc5_neg: 100.000 time: 0.1668 data: 0.0731 max mem: 3537 +Test: [1450/2387] eta: 0:03:03 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3162 (0.3275) acc1: 89.0625 (88.7931) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1587 data: 0.0677 max mem: 3537 +Test: [1460/2387] eta: 0:03:00 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2408 (0.3268) acc1: 93.7500 (88.8449) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1528 data: 0.0613 max mem: 3537 +Test: [1470/2387] eta: 0:02:58 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1830 (0.3257) acc1: 97.9167 (88.9126) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1620 data: 0.0666 max mem: 3537 +Test: [1480/2387] eta: 0:02:56 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1784 (0.3248) acc1: 98.6979 (88.9695) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1698 data: 0.0717 max mem: 3537 +Test: [1490/2387] eta: 0:02:54 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1847 (0.3239) acc1: 98.4375 (89.0316) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1599 data: 0.0615 max mem: 3537 +Test: [1500/2387] eta: 0:02:52 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1846 (0.3230) acc1: 98.6979 (89.0920) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1549 data: 0.0562 max mem: 3537 +Test: [1510/2387] eta: 0:02:50 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2161 (0.3224) acc1: 98.9583 (89.1544) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1649 data: 0.0667 max mem: 3537 +Test: [1520/2387] eta: 0:02:48 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2550 (0.3222) acc1: 97.1354 (89.1834) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1549 data: 0.0594 max mem: 3537 +Test: [1530/2387] eta: 0:02:45 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3068 (0.3222) acc1: 93.2292 (89.2125) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1349 data: 0.0396 max mem: 3537 +Test: [1540/2387] eta: 0:02:43 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2981 (0.3221) acc1: 95.5729 (89.2526) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1399 data: 0.0417 max mem: 3537 +Test: [1550/2387] eta: 0:02:41 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2805 (0.3219) acc1: 96.3542 (89.2986) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1799 data: 0.0817 max mem: 3537 +Test: [1560/2387] eta: 0:02:39 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2774 (0.3217) acc1: 95.8333 (89.3376) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.2049 data: 0.1063 max mem: 3537 +Test: [1570/2387] eta: 0:02:37 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2861 (0.3215) acc1: 95.8333 (89.3828) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1574 data: 0.0597 max mem: 3537 +Test: [1580/2387] eta: 0:02:35 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2484 (0.3210) acc1: 96.8750 (89.4290) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1529 data: 0.0579 max mem: 3538 +Test: [1590/2387] eta: 0:02:33 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2271 (0.3203) acc1: 96.8750 (89.4737) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1925 data: 0.1000 max mem: 3538 +Test: [1600/2387] eta: 0:02:31 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1944 (0.3197) acc1: 97.1354 (89.4999) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2019 data: 0.1095 max mem: 3538 +Test: [1610/2387] eta: 0:02:29 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1887 (0.3189) acc1: 97.6562 (89.5486) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1998 data: 0.1044 max mem: 3538 +Test: [1620/2387] eta: 0:02:28 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2098 (0.3184) acc1: 94.2708 (89.5769) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2015 data: 0.1067 max mem: 3538 +Test: [1630/2387] eta: 0:02:26 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2493 (0.3183) acc1: 93.7500 (89.5913) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.2049 data: 0.1119 max mem: 3538 +Test: [1640/2387] eta: 0:02:24 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3558 (0.3187) acc1: 87.2396 (89.5564) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.1983 data: 0.0997 max mem: 3538 +Test: [1650/2387] eta: 0:02:22 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3837 (0.3190) acc1: 83.5938 (89.5234) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1929 data: 0.0942 max mem: 3538 +Test: [1660/2387] eta: 0:02:20 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3888 (0.3195) acc1: 83.8542 (89.4861) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.1803 data: 0.0861 max mem: 3538 +Test: [1670/2387] eta: 0:02:18 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3877 (0.3197) acc1: 84.3750 (89.4675) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1739 data: 0.0811 max mem: 3538 +Test: [1680/2387] eta: 0:02:16 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3700 (0.3200) acc1: 84.8958 (89.4438) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.1709 data: 0.0800 max mem: 3538 +Test: [1690/2387] eta: 0:02:14 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3650 (0.3205) acc1: 84.3750 (89.4102) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.1708 data: 0.0799 max mem: 3538 +Test: [1700/2387] eta: 0:02:12 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3607 (0.3204) acc1: 84.3750 (89.4091) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1685 data: 0.0750 max mem: 3538 +Test: [1710/2387] eta: 0:02:10 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1782 (0.3195) acc1: 97.9167 (89.4652) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1669 data: 0.0715 max mem: 3538 +Test: [1720/2387] eta: 0:02:08 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1630 (0.3186) acc1: 99.4792 (89.5222) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1709 data: 0.0773 max mem: 3538 +Test: [1730/2387] eta: 0:02:06 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1532 (0.3177) acc1: 99.4792 (89.5764) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1686 data: 0.0762 max mem: 3538 +Test: [1740/2387] eta: 0:02:04 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1518 (0.3167) acc1: 99.4792 (89.6312) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1884 data: 0.0963 max mem: 3538 +Test: [1750/2387] eta: 0:02:02 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1537 (0.3158) acc1: 99.4792 (89.6867) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2023 data: 0.1092 max mem: 3538 +Test: [1760/2387] eta: 0:02:00 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1672 (0.3151) acc1: 98.4375 (89.7342) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1860 data: 0.0925 max mem: 3538 +Test: [1770/2387] eta: 0:01:58 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1966 (0.3145) acc1: 97.6562 (89.7771) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1852 data: 0.0932 max mem: 3538 +Test: [1780/2387] eta: 0:01:56 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1989 (0.3138) acc1: 97.6562 (89.8247) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1934 data: 0.1021 max mem: 3538 +Test: [1790/2387] eta: 0:01:54 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1876 (0.3131) acc1: 98.4375 (89.8744) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1976 data: 0.1052 max mem: 3538 +Test: [1800/2387] eta: 0:01:52 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1720 (0.3123) acc1: 98.9583 (89.9247) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1974 data: 0.1029 max mem: 3538 +Test: [1810/2387] eta: 0:01:50 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1693 (0.3115) acc1: 99.2188 (89.9758) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1944 data: 0.0992 max mem: 3539 +Test: [1820/2387] eta: 0:01:48 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1693 (0.3108) acc1: 98.9583 (90.0211) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1820 data: 0.0867 max mem: 3539 +Test: [1830/2387] eta: 0:01:46 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1796 (0.3101) acc1: 98.6979 (90.0692) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0747 max mem: 3539 +Test: [1840/2387] eta: 0:01:45 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1456 (0.3092) acc1: 100.0000 (90.1231) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1920 data: 0.0990 max mem: 3539 +Test: [1850/2387] eta: 0:01:43 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1446 (0.3083) acc1: 100.0000 (90.1762) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2161 data: 0.1246 max mem: 3539 +Test: [1860/2387] eta: 0:01:41 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1441 (0.3074) acc1: 100.0000 (90.2286) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2198 data: 0.1278 max mem: 3539 +Test: [1870/2387] eta: 0:01:39 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1443 (0.3065) acc1: 100.0000 (90.2805) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2095 data: 0.1174 max mem: 3539 +Test: [1880/2387] eta: 0:01:37 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1443 (0.3057) acc1: 100.0000 (90.3322) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2029 data: 0.1082 max mem: 3539 +Test: [1890/2387] eta: 0:01:35 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1441 (0.3048) acc1: 100.0000 (90.3832) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1991 data: 0.1032 max mem: 3539 +Test: [1900/2387] eta: 0:01:33 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1462 (0.3040) acc1: 100.0000 (90.4331) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1899 data: 0.0961 max mem: 3539 +Test: [1910/2387] eta: 0:01:31 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1996 (0.3035) acc1: 99.4792 (90.4774) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1861 data: 0.0923 max mem: 3539 +Test: [1920/2387] eta: 0:01:29 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2258 (0.3036) acc1: 97.6562 (90.4745) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1809 data: 0.0885 max mem: 3539 +Test: [1930/2387] eta: 0:01:27 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2340 (0.3032) acc1: 95.3125 (90.5061) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1842 data: 0.0935 max mem: 3539 +Test: [1940/2387] eta: 0:01:25 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2340 (0.3031) acc1: 96.8750 (90.5045) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1870 data: 0.0955 max mem: 3539 +Test: [1950/2387] eta: 0:01:23 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2645 (0.3030) acc1: 93.4896 (90.5138) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1872 data: 0.0940 max mem: 3539 +Test: [1960/2387] eta: 0:01:22 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2901 (0.3031) acc1: 92.7083 (90.5176) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2018 data: 0.1073 max mem: 3539 +Test: [1970/2387] eta: 0:01:20 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3536 (0.3038) acc1: 86.1979 (90.4494) acc5: 100.0000 (100.0000) acc1_neg: 70.573 acc5_neg: 100.000 time: 0.1888 data: 0.0957 max mem: 3539 +Test: [1980/2387] eta: 0:01:18 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3713 (0.3037) acc1: 84.3750 (90.4537) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1549 data: 0.0617 max mem: 3539 +Test: [1990/2387] eta: 0:01:16 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1455 (0.3029) acc1: 99.4792 (90.4992) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1797 data: 0.0860 max mem: 3539 +Test: [2000/2387] eta: 0:01:14 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1446 (0.3021) acc1: 99.4792 (90.5439) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2144 data: 0.1225 max mem: 3539 +Test: [2010/2387] eta: 0:01:12 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1446 (0.3013) acc1: 99.4792 (90.5884) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1936 data: 0.1025 max mem: 3539 +Test: [2020/2387] eta: 0:01:10 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1452 (0.3006) acc1: 99.4792 (90.6322) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1765 data: 0.0851 max mem: 3539 +Test: [2030/2387] eta: 0:01:08 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1439 (0.2998) acc1: 99.4792 (90.6760) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1910 data: 0.0988 max mem: 3540 +Test: [2040/2387] eta: 0:01:06 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1441 (0.2990) acc1: 99.4792 (90.7192) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1852 data: 0.0925 max mem: 3540 +Test: [2050/2387] eta: 0:01:04 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1449 (0.2983) acc1: 99.4792 (90.7620) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1785 data: 0.0867 max mem: 3540 +Test: [2060/2387] eta: 0:01:02 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1542 (0.2976) acc1: 99.4792 (90.8032) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1835 data: 0.0910 max mem: 3540 +Test: [2070/2387] eta: 0:01:00 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1570 (0.2969) acc1: 99.2188 (90.8446) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1804 data: 0.0863 max mem: 3540 +Test: [2080/2387] eta: 0:00:58 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1542 (0.2962) acc1: 99.4792 (90.8854) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1721 data: 0.0790 max mem: 3540 +Test: [2090/2387] eta: 0:00:56 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1542 (0.2956) acc1: 99.2188 (90.9258) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1794 data: 0.0874 max mem: 3540 +Test: [2100/2387] eta: 0:00:54 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1546 (0.2949) acc1: 99.2188 (90.9661) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1926 data: 0.0994 max mem: 3540 +Test: [2110/2387] eta: 0:00:53 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1559 (0.2942) acc1: 99.4792 (91.0062) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1849 data: 0.0880 max mem: 3540 +Test: [2120/2387] eta: 0:00:51 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1666 (0.2940) acc1: 98.6979 (91.0262) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1699 data: 0.0705 max mem: 3540 +Test: [2130/2387] eta: 0:00:49 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2213 (0.2937) acc1: 96.8750 (91.0550) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1649 data: 0.0686 max mem: 3540 +Test: [2140/2387] eta: 0:00:47 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2093 (0.2935) acc1: 97.9167 (91.0688) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1899 data: 0.0976 max mem: 3540 +Test: [2150/2387] eta: 0:00:45 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2263 (0.2932) acc1: 96.6146 (91.0943) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2001 data: 0.1070 max mem: 3540 +Test: [2160/2387] eta: 0:00:43 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2263 (0.2930) acc1: 96.6146 (91.1159) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1775 data: 0.0829 max mem: 3540 +Test: [2170/2387] eta: 0:00:41 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2715 (0.2933) acc1: 92.9688 (91.1016) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.1769 data: 0.0838 max mem: 3540 +Test: [2180/2387] eta: 0:00:39 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.4095 (0.2944) acc1: 81.7708 (91.0139) acc5: 100.0000 (100.0000) acc1_neg: 52.083 acc5_neg: 100.000 time: 0.1789 data: 0.0882 max mem: 3540 +Test: [2190/2387] eta: 0:00:37 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3639 (0.2946) acc1: 86.7188 (91.0007) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1505 data: 0.0591 max mem: 3540 +Test: [2200/2387] eta: 0:00:35 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2503 (0.2943) acc1: 95.3125 (91.0282) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1545 data: 0.0602 max mem: 3540 +Test: [2210/2387] eta: 0:00:33 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2173 (0.2940) acc1: 98.1771 (91.0619) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1901 data: 0.0953 max mem: 3540 +Test: [2220/2387] eta: 0:00:31 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2183 (0.2937) acc1: 98.1771 (91.0908) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1737 data: 0.0799 max mem: 3540 +Test: [2230/2387] eta: 0:00:29 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2217 (0.2934) acc1: 97.3958 (91.1203) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1624 data: 0.0691 max mem: 3540 +Test: [2240/2387] eta: 0:00:28 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2163 (0.2930) acc1: 97.6562 (91.1491) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1849 data: 0.0925 max mem: 3540 +Test: [2250/2387] eta: 0:00:26 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2317 (0.2930) acc1: 97.9167 (91.1707) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1846 data: 0.0860 max mem: 3540 +Test: [2260/2387] eta: 0:00:24 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2317 (0.2927) acc1: 98.1771 (91.2014) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1754 data: 0.0776 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1963 (0.2922) acc1: 98.9583 (91.2377) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0771 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1694 (0.2918) acc1: 99.7396 (91.2569) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1694 data: 0.0760 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1685 (0.2913) acc1: 99.4792 (91.2911) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1574 data: 0.0652 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.1776 (0.2909) acc1: 98.9583 (91.3185) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1482 data: 0.0562 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2194 (0.2908) acc1: 97.3958 (91.3372) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1410 data: 0.0492 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2930 (0.2909) acc1: 92.4479 (91.3318) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1378 data: 0.0467 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3189 (0.2911) acc1: 90.8854 (91.3276) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.1362 data: 0.0458 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2955 (0.2911) acc1: 92.9688 (91.3325) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1288 data: 0.0382 max mem: 3541 +Test: [2350/2387] eta: 0:00:06 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.2901 (0.2913) acc1: 92.9688 (91.3176) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1278 data: 0.0373 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3286 (0.2916) acc1: 89.3229 (91.3046) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1232 data: 0.0330 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3440 (0.2918) acc1: 88.2812 (91.2872) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.1267 data: 0.0366 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.3695 (0.2925) acc1: 84.3750 (91.2474) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.1260 data: 0.0357 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 75.799 acc5_pos: 100.000 loss: 0.4103 (0.2926) acc1: 84.1146 (91.2366) acc5: 100.0000 (100.0000) acc1_neg: 93.023 acc5_neg: 100.000 time: 0.1148 data: 0.0273 max mem: 3541 +Test: Total time: 0:07:28 (0.1879 s / it) +* Overall Acc@1 91.237 Acc@5 100.000 + target == real Acc@1 62.475 Acc@5 100.000 + target == fake Acc@1 93.256 Acc@5 100.000 + loss 0.293 + PR-AUC 0.9939 ROC-AUC 0.9209 EER 14.96% (thr=0.6348) +Accuracy of the network on the 916267 test images: 91.2% +[rank0]:[W726 06:05:21.181180147 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 06:05:40.887000 1204949 torch/distributed/run.py:766] +W0726 06:05:40.887000 1204949 torch/distributed/run.py:766] ***************************************** +W0726 06:05:40.887000 1204949 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:05:40.887000 1204949 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:06:08.855826153 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:06:09.196325439 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:06:09.430893716 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:06:09.580086688 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 06:06:09.600514434 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 3): env:// +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:06:09.647550312 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank6]:[W726 06:06:09.647573283 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W726 06:06:09.647603663 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n32_seed9_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n32_classseed9_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:02:35 acc1_pos: 55.990 acc5_pos: 100.000 loss: 0.6930 (0.6930) acc1: 55.9896 (55.9896) acc5: 100.0000 (100.0000) time: 7.2631 data: 5.3899 max mem: 3531 +Test: [ 10/517] eta: 0:06:16 acc1_pos: 37.240 acc5_pos: 100.000 loss: 0.7581 (0.7741) acc1: 46.3542 (48.1771) acc5: 100.0000 (100.0000) time: 0.7423 data: 0.4903 max mem: 3531 +Test: [ 20/517] eta: 0:03:35 acc1_pos: 42.969 acc5_pos: 100.000 loss: 0.7581 (0.7812) acc1: 45.8333 (47.5322) acc5: 100.0000 (100.0000) time: 0.0912 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:02:54 acc1_pos: 32.812 acc5_pos: 100.000 loss: 0.7961 (0.7920) acc1: 42.9688 (46.7910) acc5: 100.0000 (100.0000) time: 0.1475 data: 0.0552 max mem: 3531 +Test: [ 40/517] eta: 0:02:46 acc1_pos: 63.542 acc5_pos: 100.000 loss: 0.7878 (0.7917) acc1: 44.7917 (46.5320) acc5: 100.0000 (100.0000) time: 0.2599 data: 0.1632 max mem: 3531 +Test: [ 50/517] eta: 0:02:33 acc1_pos: 79.358 acc5_pos: 100.000 loss: 0.7301 (0.7693) acc1: 47.3958 (48.9124) acc5: 100.0000 (100.0000) acc1_neg: 77.344 acc5_neg: 100.000 time: 0.2835 data: 0.1860 max mem: 3531 +Test: [ 60/517] eta: 0:02:23 acc1_pos: 79.358 acc5_pos: 100.000 loss: 0.5809 (0.7294) acc1: 70.3125 (53.4580) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.2449 data: 0.1515 max mem: 3531 +Test: [ 70/517] eta: 0:02:15 acc1_pos: 79.358 acc5_pos: 100.000 loss: 0.4648 (0.6881) acc1: 85.1562 (58.4324) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.2349 data: 0.1387 max mem: 3531 +Test: [ 80/517] eta: 0:02:09 acc1_pos: 79.358 acc5_pos: 100.000 loss: 0.4380 (0.6564) acc1: 87.7604 (62.2653) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.2449 data: 0.1476 max mem: 3531 +Test: [ 90/517] eta: 0:02:05 acc1_pos: 79.358 acc5_pos: 100.000 loss: 0.4380 (0.6321) acc1: 88.2812 (65.0298) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.2660 data: 0.1723 max mem: 3531 +Test: [100/517] eta: 0:02:01 acc1_pos: 79.358 acc5_pos: 100.000 loss: 0.4219 (0.6105) acc1: 89.3229 (67.4273) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.2649 data: 0.1734 max mem: 3531 +Test: [110/517] eta: 0:01:56 acc1_pos: 79.358 acc5_pos: 100.000 loss: 0.3900 (0.5918) acc1: 90.3646 (69.4116) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2544 data: 0.1639 max mem: 3531 +Test: [120/517] eta: 0:01:53 acc1_pos: 79.358 acc5_pos: 100.000 loss: 0.4252 (0.5789) acc1: 90.1042 (71.0981) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.2662 data: 0.1747 max mem: 3531 +Test: [130/517] eta: 0:01:51 acc1_pos: 79.358 acc5_pos: 100.000 loss: 0.4283 (0.5666) acc1: 90.1042 (72.6085) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.2843 data: 0.1881 max mem: 3531 +Test: [140/517] eta: 0:01:46 acc1_pos: 79.358 acc5_pos: 100.000 loss: 0.4255 (0.5580) acc1: 89.5833 (73.6628) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.2648 data: 0.1694 max mem: 3531 +Test: [150/517] eta: 0:01:42 acc1_pos: 79.358 acc5_pos: 100.000 loss: 0.4118 (0.5485) acc1: 88.5417 (74.7258) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2349 data: 0.1383 max mem: 3531 +Test: [160/517] eta: 0:01:39 acc1_pos: 79.358 acc5_pos: 100.000 loss: 0.4099 (0.5400) acc1: 90.1042 (75.6874) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2486 data: 0.1503 max mem: 3531 +Test: [170/517] eta: 0:01:36 acc1_pos: 79.358 acc5_pos: 100.000 loss: 0.3979 (0.5305) acc1: 90.3646 (76.6447) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.2599 data: 0.1663 max mem: 3531 +Test: [180/517] eta: 0:01:33 acc1_pos: 79.358 acc5_pos: 100.000 loss: 0.3845 (0.5232) acc1: 90.3646 (77.2301) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.2499 data: 0.1569 max mem: 3531 +Test: [190/517] eta: 0:01:30 acc1_pos: 79.358 acc5_pos: 100.000 loss: 0.2948 (0.5109) acc1: 97.3958 (78.3527) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2561 data: 0.1633 max mem: 3531 +Test: [200/517] eta: 0:01:26 acc1_pos: 79.358 acc5_pos: 100.000 loss: 0.2931 (0.5039) acc1: 97.3958 (79.1809) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2549 data: 0.1625 max mem: 3531 +Test: [210/517] eta: 0:01:24 acc1_pos: 79.358 acc5_pos: 100.000 loss: 0.4121 (0.5030) acc1: 89.3229 (79.2531) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.2637 data: 0.1698 max mem: 3532 +Test: [220/517] eta: 0:01:21 acc1_pos: 79.358 acc5_pos: 100.000 loss: 0.4726 (0.5016) acc1: 80.7292 (79.3670) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.2604 data: 0.1672 max mem: 3532 +Test: [230/517] eta: 0:01:18 acc1_pos: 79.358 acc5_pos: 100.000 loss: 0.5008 (0.5029) acc1: 80.2083 (79.3673) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.2457 data: 0.1540 max mem: 3532 +Test: [240/517] eta: 0:01:15 acc1_pos: 79.358 acc5_pos: 100.000 loss: 0.5028 (0.5003) acc1: 83.0729 (79.6292) acc5: 100.0000 (100.0000) acc1_neg: 68.229 acc5_neg: 100.000 time: 0.2845 data: 0.1914 max mem: 3532 +Test: [250/517] eta: 0:01:12 acc1_pos: 79.358 acc5_pos: 100.000 loss: 0.4279 (0.4981) acc1: 88.0208 (80.0952) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2896 data: 0.1947 max mem: 3532 +Test: [260/517] eta: 0:01:10 acc1_pos: 79.358 acc5_pos: 100.000 loss: 0.4236 (0.4957) acc1: 94.5312 (80.5905) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.2600 data: 0.1632 max mem: 3532 +Test: [270/517] eta: 0:01:07 acc1_pos: 79.358 acc5_pos: 100.000 loss: 0.4418 (0.4937) acc1: 90.1042 (80.9031) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.2545 data: 0.1569 max mem: 3532 +Test: [280/517] eta: 0:01:04 acc1_pos: 79.358 acc5_pos: 100.000 loss: 0.4309 (0.4908) acc1: 89.3229 (81.2315) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.2448 data: 0.1472 max mem: 3532 +Test: [290/517] eta: 0:01:01 acc1_pos: 79.358 acc5_pos: 100.000 loss: 0.4145 (0.4891) acc1: 91.1458 (81.5104) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.2452 data: 0.1485 max mem: 3532 +Test: [300/517] eta: 0:00:58 acc1_pos: 79.358 acc5_pos: 100.000 loss: 0.4234 (0.4869) acc1: 90.1042 (81.7907) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.2659 data: 0.1705 max mem: 3532 +Test: [310/517] eta: 0:00:55 acc1_pos: 79.358 acc5_pos: 100.000 loss: 0.4819 (0.4882) acc1: 82.8125 (81.6193) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.2495 data: 0.1536 max mem: 3532 +Test: [320/517] eta: 0:00:52 acc1_pos: 79.358 acc5_pos: 100.000 loss: 0.4954 (0.4861) acc1: 78.9062 (81.7984) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2254 data: 0.1290 max mem: 3532 +Test: [330/517] eta: 0:00:49 acc1_pos: 79.358 acc5_pos: 100.000 loss: 0.3510 (0.4818) acc1: 94.0104 (82.1988) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2399 data: 0.1446 max mem: 3532 +Test: [340/517] eta: 0:00:47 acc1_pos: 79.358 acc5_pos: 100.000 loss: 0.3336 (0.4802) acc1: 95.5729 (82.3131) acc5: 100.0000 (100.0000) acc1_neg: 67.708 acc5_neg: 100.000 time: 0.2395 data: 0.1444 max mem: 3532 +Test: [350/517] eta: 0:00:44 acc1_pos: 79.358 acc5_pos: 100.000 loss: 0.5089 (0.4825) acc1: 78.3854 (82.0505) acc5: 100.0000 (100.0000) acc1_neg: 65.885 acc5_neg: 100.000 time: 0.2299 data: 0.1346 max mem: 3532 +Test: [360/517] eta: 0:00:41 acc1_pos: 79.358 acc5_pos: 100.000 loss: 0.5155 (0.4826) acc1: 77.0833 (82.0233) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.2338 data: 0.1351 max mem: 3532 +Test: [370/517] eta: 0:00:38 acc1_pos: 79.358 acc5_pos: 100.000 loss: 0.3960 (0.4793) acc1: 91.6667 (82.3984) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2449 data: 0.1438 max mem: 3532 +Test: [380/517] eta: 0:00:36 acc1_pos: 79.358 acc5_pos: 100.000 loss: 0.3649 (0.4766) acc1: 95.0521 (82.7059) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2449 data: 0.1454 max mem: 3532 +Test: [390/517] eta: 0:00:33 acc1_pos: 79.358 acc5_pos: 100.000 loss: 0.3711 (0.4737) acc1: 94.2708 (82.9917) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.2199 data: 0.1209 max mem: 3532 +Test: [400/517] eta: 0:00:30 acc1_pos: 79.358 acc5_pos: 100.000 loss: 0.3627 (0.4716) acc1: 93.7500 (83.1606) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2302 data: 0.1328 max mem: 3532 +Test: [410/517] eta: 0:00:27 acc1_pos: 79.358 acc5_pos: 100.000 loss: 0.4193 (0.4708) acc1: 89.5833 (83.3181) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.2349 data: 0.1381 max mem: 3532 +Test: [420/517] eta: 0:00:25 acc1_pos: 79.358 acc5_pos: 100.000 loss: 0.4577 (0.4703) acc1: 90.1042 (83.4991) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.2266 data: 0.1304 max mem: 3532 +Test: [430/517] eta: 0:00:22 acc1_pos: 79.358 acc5_pos: 100.000 loss: 0.4560 (0.4694) acc1: 90.1042 (83.5998) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.2303 data: 0.1370 max mem: 3532 +Test: [440/517] eta: 0:00:19 acc1_pos: 79.358 acc5_pos: 100.000 loss: 0.4692 (0.4696) acc1: 83.0729 (83.5985) acc5: 100.0000 (100.0000) acc1_neg: 76.042 acc5_neg: 100.000 time: 0.2293 data: 0.1361 max mem: 3533 +Test: [450/517] eta: 0:00:17 acc1_pos: 79.358 acc5_pos: 100.000 loss: 0.4535 (0.4689) acc1: 84.3750 (83.7364) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.2544 data: 0.1608 max mem: 3533 +Test: [460/517] eta: 0:00:14 acc1_pos: 79.358 acc5_pos: 100.000 loss: 0.4493 (0.4688) acc1: 88.5417 (83.7898) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.2584 data: 0.1664 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 79.358 acc5_pos: 100.000 loss: 0.4412 (0.4677) acc1: 88.0208 (83.8990) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.2449 data: 0.1497 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 79.358 acc5_pos: 100.000 loss: 0.4362 (0.4673) acc1: 90.8854 (84.0610) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2598 data: 0.1637 max mem: 3533 +Test: [490/517] eta: 0:00:06 acc1_pos: 79.358 acc5_pos: 100.000 loss: 0.4379 (0.4667) acc1: 90.8854 (84.1602) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.2749 data: 0.1814 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 79.358 acc5_pos: 100.000 loss: 0.4427 (0.4672) acc1: 88.8021 (84.1577) acc5: 100.0000 (100.0000) acc1_neg: 71.354 acc5_neg: 100.000 time: 0.2334 data: 0.1416 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 79.358 acc5_pos: 100.000 loss: 0.4355 (0.4667) acc1: 88.8021 (84.2058) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1796 data: 0.0895 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 79.358 acc5_pos: 100.000 loss: 0.4201 (0.4658) acc1: 90.3646 (84.2990) acc5: 100.0000 (100.0000) acc1_neg: 88.693 acc5_neg: 100.000 time: 0.1739 data: 0.0819 max mem: 3533 +Test: Total time: 0:02:11 (0.2544 s / it) +* Overall Acc@1 84.299 Acc@5 100.000 + target == real Acc@1 47.539 Acc@5 100.000 + target == fake Acc@1 88.112 Acc@5 100.000 + loss 0.466 + PR-AUC 0.9632 ROC-AUC 0.7688 EER 30.21% (thr=0.5996) +Accuracy of the network on the 198427 test images: 84.3% +[rank0]:[W726 06:08:33.485461956 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 06:08:54.763000 1205522 torch/distributed/run.py:766] +W0726 06:08:54.763000 1205522 torch/distributed/run.py:766] ***************************************** +W0726 06:08:54.763000 1205522 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:08:54.763000 1205522 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:09:23.314551790 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:09:23.687021445 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:09:23.993267908 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:09:23.120562788 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 06:09:23.120565925 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:09:23.128182644 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:09:23.132820437 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:09:23.158438222 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n32_seed9_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n32_classseed9_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:44:13 acc1_pos: 36.979 acc5_pos: 100.000 loss: 0.8020 (0.8020) acc1: 36.9792 (36.9792) acc5: 100.0000 (100.0000) time: 6.8212 data: 5.0957 max mem: 3531 +Test: [ 10/389] eta: 0:04:27 acc1_pos: 95.052 acc5_pos: 100.000 loss: 0.6223 (0.6644) acc1: 69.0104 (59.1619) acc5: 100.0000 (100.0000) time: 0.7055 data: 0.4636 max mem: 3531 +Test: [ 20/389] eta: 0:02:33 acc1_pos: 39.583 acc5_pos: 100.000 loss: 0.6404 (0.6492) acc1: 62.2396 (60.7639) acc5: 100.0000 (100.0000) time: 0.0969 data: 0.0004 max mem: 3531 +Test: [ 30/389] eta: 0:02:05 acc1_pos: 65.625 acc5_pos: 100.000 loss: 0.7452 (0.7060) acc1: 39.8438 (53.7382) acc5: 100.0000 (100.0000) time: 0.1549 data: 0.0569 max mem: 3531 +Test: [ 40/389] eta: 0:01:49 acc1_pos: 33.594 acc5_pos: 100.000 loss: 0.8210 (0.7240) acc1: 33.5938 (51.3402) acc5: 100.0000 (100.0000) time: 0.2067 data: 0.1113 max mem: 3531 +Test: [ 50/389] eta: 0:01:38 acc1_pos: 51.302 acc5_pos: 100.000 loss: 0.7733 (0.7228) acc1: 48.1771 (51.6953) acc5: 100.0000 (100.0000) time: 0.1953 data: 0.1030 max mem: 3531 +Test: [ 60/389] eta: 0:01:31 acc1_pos: 10.156 acc5_pos: 100.000 loss: 0.7560 (0.7188) acc1: 48.4375 (52.3139) acc5: 100.0000 (100.0000) time: 0.2032 data: 0.1095 max mem: 3531 +Test: [ 70/389] eta: 0:01:22 acc1_pos: 64.323 acc5_pos: 100.000 loss: 0.7772 (0.7304) acc1: 48.1771 (50.2824) acc5: 100.0000 (100.0000) time: 0.1847 data: 0.0896 max mem: 3531 +Test: [ 80/389] eta: 0:01:17 acc1_pos: 90.625 acc5_pos: 100.000 loss: 0.6551 (0.7142) acc1: 55.9896 (52.9032) acc5: 100.0000 (100.0000) time: 0.1670 data: 0.0745 max mem: 3531 +Test: [ 90/389] eta: 0:01:12 acc1_pos: 46.875 acc5_pos: 100.000 loss: 0.4474 (0.6901) acc1: 75.7812 (55.3886) acc5: 100.0000 (100.0000) time: 0.1857 data: 0.0924 max mem: 3531 +Test: [100/389] eta: 0:01:08 acc1_pos: 42.448 acc5_pos: 100.000 loss: 0.7505 (0.7045) acc1: 47.9167 (53.6072) acc5: 100.0000 (100.0000) time: 0.1898 data: 0.0966 max mem: 3531 +Test: [110/389] eta: 0:01:04 acc1_pos: 90.104 acc5_pos: 100.000 loss: 0.7878 (0.7035) acc1: 41.4062 (53.4980) acc5: 100.0000 (100.0000) time: 0.1767 data: 0.0848 max mem: 3531 +Test: [120/389] eta: 0:01:01 acc1_pos: 83.333 acc5_pos: 100.000 loss: 0.6465 (0.7016) acc1: 54.9479 (53.8654) acc5: 100.0000 (100.0000) time: 0.1787 data: 0.0861 max mem: 3531 +Test: [130/389] eta: 0:00:58 acc1_pos: 44.785 acc5_pos: 100.000 loss: 0.6064 (0.6909) acc1: 66.4062 (55.3038) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1945 data: 0.1014 max mem: 3531 +Test: [140/389] eta: 0:00:55 acc1_pos: 44.785 acc5_pos: 100.000 loss: 0.5134 (0.6816) acc1: 78.3854 (56.4864) acc5: 100.0000 (100.0000) acc1_neg: 36.979 acc5_neg: 100.000 time: 0.1898 data: 0.0972 max mem: 3531 +Test: [150/389] eta: 0:00:52 acc1_pos: 44.785 acc5_pos: 100.000 loss: 0.5173 (0.6718) acc1: 81.7708 (57.8401) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1860 data: 0.0944 max mem: 3531 +Test: [160/389] eta: 0:00:49 acc1_pos: 44.785 acc5_pos: 100.000 loss: 0.5065 (0.6589) acc1: 83.0729 (59.3054) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1824 data: 0.0910 max mem: 3531 +Test: [170/389] eta: 0:00:47 acc1_pos: 44.785 acc5_pos: 100.000 loss: 0.3144 (0.6409) acc1: 97.3958 (61.2512) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.1776 data: 0.0862 max mem: 3531 +Test: [180/389] eta: 0:00:44 acc1_pos: 44.785 acc5_pos: 100.000 loss: 0.3002 (0.6287) acc1: 97.3958 (62.6885) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.1790 data: 0.0878 max mem: 3531 +Test: [190/389] eta: 0:00:42 acc1_pos: 44.785 acc5_pos: 100.000 loss: 0.4385 (0.6267) acc1: 85.9375 (62.9363) acc5: 100.0000 (100.0000) acc1_neg: 23.438 acc5_neg: 100.000 time: 0.1747 data: 0.0835 max mem: 3531 +Test: [200/389] eta: 0:00:39 acc1_pos: 44.785 acc5_pos: 100.000 loss: 0.5087 (0.6245) acc1: 67.1875 (63.1893) acc5: 100.0000 (100.0000) acc1_neg: 59.115 acc5_neg: 100.000 time: 0.1675 data: 0.0756 max mem: 3531 +Test: [210/389] eta: 0:00:37 acc1_pos: 44.785 acc5_pos: 100.000 loss: 0.4727 (0.6142) acc1: 81.2500 (64.2291) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1700 data: 0.0774 max mem: 3532 +Test: [220/389] eta: 0:00:34 acc1_pos: 44.785 acc5_pos: 100.000 loss: 0.3003 (0.6005) acc1: 99.2188 (65.6580) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1849 data: 0.0899 max mem: 3532 +Test: [230/389] eta: 0:00:32 acc1_pos: 44.785 acc5_pos: 100.000 loss: 0.4119 (0.5963) acc1: 91.9271 (66.0692) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1762 data: 0.0822 max mem: 3532 +Test: [240/389] eta: 0:00:30 acc1_pos: 44.785 acc5_pos: 100.000 loss: 0.4555 (0.5902) acc1: 84.6354 (66.8990) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.1572 data: 0.0659 max mem: 3532 +Test: [250/389] eta: 0:00:28 acc1_pos: 44.785 acc5_pos: 100.000 loss: 0.4215 (0.5827) acc1: 89.3229 (68.0154) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1659 data: 0.0743 max mem: 3532 +Test: [260/389] eta: 0:00:25 acc1_pos: 44.785 acc5_pos: 100.000 loss: 0.4941 (0.5840) acc1: 85.1562 (67.8281) acc5: 100.0000 (100.0000) acc1_neg: 21.094 acc5_neg: 100.000 time: 0.1776 data: 0.0862 max mem: 3532 +Test: [270/389] eta: 0:00:23 acc1_pos: 44.785 acc5_pos: 100.000 loss: 0.5558 (0.5862) acc1: 61.4583 (67.5776) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1799 data: 0.0874 max mem: 3532 +Test: [280/389] eta: 0:00:21 acc1_pos: 44.785 acc5_pos: 100.000 loss: 0.5384 (0.5861) acc1: 76.0417 (67.6694) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0927 max mem: 3532 +Test: [290/389] eta: 0:00:19 acc1_pos: 44.785 acc5_pos: 100.000 loss: 0.4250 (0.5807) acc1: 86.7188 (68.3222) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1833 data: 0.0915 max mem: 3532 +Test: [300/389] eta: 0:00:17 acc1_pos: 44.785 acc5_pos: 100.000 loss: 0.5531 (0.5800) acc1: 79.6875 (68.6453) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.1848 data: 0.0905 max mem: 3532 +Test: [310/389] eta: 0:00:15 acc1_pos: 44.785 acc5_pos: 100.000 loss: 0.5612 (0.5809) acc1: 78.1250 (68.6638) acc5: 100.0000 (100.0000) acc1_neg: 33.073 acc5_neg: 100.000 time: 0.1882 data: 0.0941 max mem: 3532 +Test: [320/389] eta: 0:00:13 acc1_pos: 44.785 acc5_pos: 100.000 loss: 0.4943 (0.5783) acc1: 80.9896 (69.0575) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1850 data: 0.0934 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 44.785 acc5_pos: 100.000 loss: 0.6683 (0.5821) acc1: 57.5521 (68.4613) acc5: 100.0000 (100.0000) acc1_neg: 70.312 acc5_neg: 100.000 time: 0.1849 data: 0.0926 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 44.785 acc5_pos: 100.000 loss: 0.4352 (0.5747) acc1: 70.3125 (69.2533) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1704 data: 0.0777 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 44.785 acc5_pos: 100.000 loss: 0.3409 (0.5771) acc1: 96.0938 (69.2130) acc5: 100.0000 (100.0000) acc1_neg: 67.969 acc5_neg: 100.000 time: 0.1663 data: 0.0750 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 44.785 acc5_pos: 100.000 loss: 0.4921 (0.5755) acc1: 80.7292 (69.4505) acc5: 100.0000 (100.0000) acc1_neg: 80.729 acc5_neg: 100.000 time: 0.1459 data: 0.0557 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 44.785 acc5_pos: 100.000 loss: 0.4873 (0.5707) acc1: 81.2500 (69.9489) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1211 data: 0.0311 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 44.785 acc5_pos: 100.000 loss: 0.3968 (0.5709) acc1: 86.4583 (69.9065) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1159 data: 0.0260 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 44.785 acc5_pos: 100.000 loss: 0.3920 (0.5678) acc1: 91.1458 (70.3042) acc5: 100.0000 (100.0000) acc1_neg: 75.137 acc5_neg: 100.000 time: 0.1184 data: 0.0251 max mem: 3532 +Test: Total time: 0:01:13 (0.1887 s / it) +* Overall Acc@1 70.304 Acc@5 100.000 + target == real Acc@1 53.814 Acc@5 100.000 + target == fake Acc@1 77.969 Acc@5 100.000 + loss 0.568 + PR-AUC 0.8554 ROC-AUC 0.7465 EER 32.08% (thr=0.5649) +Accuracy of the network on the 149358 test images: 70.3% +[rank0]:[W726 06:10:46.881067912 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 06:11:06.013000 1206081 torch/distributed/run.py:766] +W0726 06:11:06.013000 1206081 torch/distributed/run.py:766] ***************************************** +W0726 06:11:06.013000 1206081 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:11:06.013000 1206081 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:11:36.243779179 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 06:11:36.245770310 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:11:36.324857675 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:11:36.565886915 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:11:36.709481927 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +| distributed init (rank 2): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:11:36.852396307 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W726 06:11:36.852396527 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank6]:[W726 06:11:36.852399580 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n32_seed9_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n32_classseed9_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:40:16 acc1_pos: 86.979 acc5_pos: 100.000 loss: 0.3312 (0.3312) acc1: 86.9792 (86.9792) acc5: 100.0000 (100.0000) time: 7.2568 data: 5.4851 max mem: 3531 +Test: [ 10/333] eta: 0:04:01 acc1_pos: 38.281 acc5_pos: 100.000 loss: 0.3312 (0.5350) acc1: 86.9792 (72.3722) acc5: 100.0000 (100.0000) time: 0.7481 data: 0.4990 max mem: 3531 +Test: [ 20/333] eta: 0:02:17 acc1_pos: 51.823 acc5_pos: 100.000 loss: 0.9369 (0.7755) acc1: 38.2812 (53.4102) acc5: 100.0000 (100.0000) time: 0.0979 data: 0.0004 max mem: 3531 +Test: [ 30/333] eta: 0:01:49 acc1_pos: 72.917 acc5_pos: 100.000 loss: 0.8552 (0.7397) acc1: 48.1771 (56.7456) acc5: 100.0000 (100.0000) time: 0.1493 data: 0.0480 max mem: 3531 +Test: [ 40/333] eta: 0:01:35 acc1_pos: 66.146 acc5_pos: 100.000 loss: 0.6278 (0.7148) acc1: 64.8438 (58.6954) acc5: 100.0000 (100.0000) time: 0.2099 data: 0.1115 max mem: 3531 +Test: [ 50/333] eta: 0:01:24 acc1_pos: 63.281 acc5_pos: 100.000 loss: 0.6205 (0.6975) acc1: 65.6250 (60.2482) acc5: 100.0000 (100.0000) time: 0.2049 data: 0.1110 max mem: 3531 +Test: [ 60/333] eta: 0:01:17 acc1_pos: 61.198 acc5_pos: 100.000 loss: 0.6197 (0.6880) acc1: 63.8021 (61.0442) acc5: 100.0000 (100.0000) time: 0.1899 data: 0.0957 max mem: 3531 +Test: [ 70/333] eta: 0:01:10 acc1_pos: 60.677 acc5_pos: 100.000 loss: 0.6191 (0.6782) acc1: 66.1458 (62.1626) acc5: 100.0000 (100.0000) time: 0.1855 data: 0.0926 max mem: 3531 +Test: [ 80/333] eta: 0:01:04 acc1_pos: 69.792 acc5_pos: 100.000 loss: 0.6035 (0.6702) acc1: 66.4062 (62.7379) acc5: 100.0000 (100.0000) time: 0.1799 data: 0.0863 max mem: 3531 +Test: [ 90/333] eta: 0:00:59 acc1_pos: 76.823 acc5_pos: 100.000 loss: 0.5750 (0.6571) acc1: 71.8750 (64.0539) acc5: 100.0000 (100.0000) time: 0.1711 data: 0.0773 max mem: 3531 +Test: [100/333] eta: 0:00:55 acc1_pos: 76.562 acc5_pos: 100.000 loss: 0.5560 (0.6494) acc1: 73.1771 (64.7896) acc5: 100.0000 (100.0000) time: 0.1656 data: 0.0739 max mem: 3531 +Test: [110/333] eta: 0:00:52 acc1_pos: 80.469 acc5_pos: 100.000 loss: 0.5584 (0.6418) acc1: 72.9167 (65.3740) acc5: 100.0000 (100.0000) time: 0.1739 data: 0.0819 max mem: 3531 +Test: [120/333] eta: 0:00:49 acc1_pos: 75.260 acc5_pos: 100.000 loss: 0.5281 (0.6341) acc1: 71.3542 (66.0468) acc5: 100.0000 (100.0000) time: 0.1958 data: 0.1034 max mem: 3531 +Test: [130/333] eta: 0:00:46 acc1_pos: 83.854 acc5_pos: 100.000 loss: 0.5163 (0.6278) acc1: 75.2604 (66.6229) acc5: 100.0000 (100.0000) time: 0.1961 data: 0.1045 max mem: 3531 +Test: [140/333] eta: 0:00:43 acc1_pos: 80.729 acc5_pos: 100.000 loss: 0.5226 (0.6211) acc1: 75.7812 (67.2706) acc5: 100.0000 (100.0000) time: 0.1893 data: 0.0979 max mem: 3531 +Test: [150/333] eta: 0:00:40 acc1_pos: 74.740 acc5_pos: 100.000 loss: 0.5332 (0.6175) acc1: 74.7396 (67.5548) acc5: 100.0000 (100.0000) time: 0.1850 data: 0.0942 max mem: 3531 +Test: [160/333] eta: 0:00:37 acc1_pos: 66.118 acc5_pos: 100.000 loss: 0.5746 (0.6233) acc1: 68.4896 (66.9724) acc5: 100.0000 (100.0000) acc1_neg: 35.417 acc5_neg: 100.000 time: 0.1706 data: 0.0797 max mem: 3531 +Test: [170/333] eta: 0:00:35 acc1_pos: 66.118 acc5_pos: 100.000 loss: 0.8112 (0.6361) acc1: 46.0938 (65.6189) acc5: 100.0000 (100.0000) acc1_neg: 49.479 acc5_neg: 100.000 time: 0.2027 data: 0.1103 max mem: 3531 +Test: [180/333] eta: 0:00:33 acc1_pos: 66.118 acc5_pos: 100.000 loss: 0.7821 (0.6413) acc1: 46.6146 (64.9042) acc5: 100.0000 (100.0000) acc1_neg: 65.625 acc5_neg: 100.000 time: 0.2181 data: 0.1252 max mem: 3531 +Test: [190/333] eta: 0:00:30 acc1_pos: 66.118 acc5_pos: 100.000 loss: 0.7276 (0.6431) acc1: 52.3438 (64.5165) acc5: 100.0000 (100.0000) acc1_neg: 51.302 acc5_neg: 100.000 time: 0.1649 data: 0.0704 max mem: 3531 +Test: [200/333] eta: 0:00:28 acc1_pos: 66.118 acc5_pos: 100.000 loss: 0.6813 (0.6460) acc1: 54.4271 (64.1778) acc5: 100.0000 (100.0000) acc1_neg: 54.427 acc5_neg: 100.000 time: 0.1853 data: 0.0896 max mem: 3531 +Test: [210/333] eta: 0:00:27 acc1_pos: 66.118 acc5_pos: 100.000 loss: 0.7432 (0.6511) acc1: 52.8646 (63.6058) acc5: 100.0000 (100.0000) acc1_neg: 52.083 acc5_neg: 100.000 time: 0.2899 data: 0.1956 max mem: 3532 +Test: [220/333] eta: 0:00:24 acc1_pos: 66.118 acc5_pos: 100.000 loss: 0.7465 (0.6584) acc1: 51.3021 (62.8417) acc5: 100.0000 (100.0000) acc1_neg: 61.198 acc5_neg: 100.000 time: 0.2609 data: 0.1653 max mem: 3532 +Test: [230/333] eta: 0:00:22 acc1_pos: 66.118 acc5_pos: 100.000 loss: 0.6532 (0.6567) acc1: 61.1979 (62.9769) acc5: 100.0000 (100.0000) acc1_neg: 66.146 acc5_neg: 100.000 time: 0.1399 data: 0.0427 max mem: 3532 +Test: [240/333] eta: 0:00:19 acc1_pos: 66.118 acc5_pos: 100.000 loss: 0.6318 (0.6615) acc1: 61.1979 (62.3185) acc5: 100.0000 (100.0000) acc1_neg: 49.219 acc5_neg: 100.000 time: 0.1559 data: 0.0614 max mem: 3532 +Test: [250/333] eta: 0:00:17 acc1_pos: 66.118 acc5_pos: 100.000 loss: 0.7624 (0.6637) acc1: 45.8333 (62.0113) acc5: 100.0000 (100.0000) acc1_neg: 61.719 acc5_neg: 100.000 time: 0.2021 data: 0.1094 max mem: 3532 +Test: [260/333] eta: 0:00:15 acc1_pos: 66.118 acc5_pos: 100.000 loss: 0.6783 (0.6640) acc1: 59.8958 (61.9852) acc5: 100.0000 (100.0000) acc1_neg: 60.938 acc5_neg: 100.000 time: 0.1793 data: 0.0874 max mem: 3532 +Test: [270/333] eta: 0:00:13 acc1_pos: 66.118 acc5_pos: 100.000 loss: 0.6388 (0.6633) acc1: 62.2396 (62.0186) acc5: 100.0000 (100.0000) acc1_neg: 63.802 acc5_neg: 100.000 time: 0.1801 data: 0.0874 max mem: 3532 +Test: [280/333] eta: 0:00:11 acc1_pos: 66.118 acc5_pos: 100.000 loss: 0.6827 (0.6690) acc1: 58.3333 (61.4370) acc5: 100.0000 (100.0000) acc1_neg: 43.750 acc5_neg: 100.000 time: 0.1879 data: 0.0948 max mem: 3532 +Test: [290/333] eta: 0:00:08 acc1_pos: 66.118 acc5_pos: 100.000 loss: 0.9242 (0.6781) acc1: 36.4583 (60.5616) acc5: 100.0000 (100.0000) acc1_neg: 39.062 acc5_neg: 100.000 time: 0.1815 data: 0.0903 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 66.118 acc5_pos: 100.000 loss: 0.8703 (0.6839) acc1: 39.0625 (59.9849) acc5: 100.0000 (100.0000) acc1_neg: 46.094 acc5_neg: 100.000 time: 0.1829 data: 0.0902 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 66.118 acc5_pos: 100.000 loss: 0.8027 (0.6832) acc1: 48.4375 (60.0164) acc5: 100.0000 (100.0000) acc1_neg: 53.125 acc5_neg: 100.000 time: 0.1584 data: 0.0662 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 66.118 acc5_pos: 100.000 loss: 0.7228 (0.6852) acc1: 53.1250 (59.7677) acc5: 100.0000 (100.0000) acc1_neg: 47.917 acc5_neg: 100.000 time: 0.1270 data: 0.0369 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 66.118 acc5_pos: 100.000 loss: 0.8079 (0.6908) acc1: 47.6562 (59.2641) acc5: 100.0000 (100.0000) acc1_neg: 42.969 acc5_neg: 100.000 time: 0.1206 data: 0.0307 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 66.118 acc5_pos: 100.000 loss: 0.8391 (0.6917) acc1: 46.4789 (59.1867) acc5: 100.0000 (100.0000) acc1_neg: 46.479 acc5_neg: 100.000 time: 0.1225 data: 0.0295 max mem: 3532 +Test: Total time: 0:01:06 (0.2009 s / it) +* Overall Acc@1 59.187 Acc@5 100.000 + target == real Acc@1 67.823 Acc@5 100.000 + target == fake Acc@1 51.498 Acc@5 100.000 + loss 0.692 + PR-AUC 0.6375 ROC-AUC 0.6358 EER 40.08% (thr=0.4521) +Accuracy of the network on the 127843 test images: 59.2% +[rank0]:[W726 06:12:53.504873045 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 06:13:13.157000 1206639 torch/distributed/run.py:766] +W0726 06:13:13.157000 1206639 torch/distributed/run.py:766] ***************************************** +W0726 06:13:13.157000 1206639 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:13:13.157000 1206639 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:13:40.460989887 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:13:40.749661361 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:13:40.970247516 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:13:40.012885069 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +[rank0]:[W726 06:13:41.275532995 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:13:41.281613886 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:13:41.295041442 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:13:41.337142857 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n32_seed9_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n32_classseed9_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:27:36 acc1_pos: 95.833 acc5_pos: 100.000 loss: 0.2224 (0.2224) acc1: 95.8333 (95.8333) acc5: 100.0000 (100.0000) time: 7.0486 data: 5.5021 max mem: 3531 +Test: [ 10/235] eta: 0:02:49 acc1_pos: 95.052 acc5_pos: 100.000 loss: 0.2285 (0.2322) acc1: 95.0521 (95.0521) acc5: 100.0000 (100.0000) time: 0.7531 data: 0.5269 max mem: 3531 +Test: [ 20/235] eta: 0:01:35 acc1_pos: 95.573 acc5_pos: 100.000 loss: 0.2261 (0.2285) acc1: 95.3125 (95.2505) acc5: 100.0000 (100.0000) time: 0.1117 data: 0.0149 max mem: 3531 +Test: [ 30/235] eta: 0:01:24 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.2229 (0.2275) acc1: 95.5729 (95.3545) acc5: 100.0000 (100.0000) time: 0.2262 data: 0.1303 max mem: 3531 +Test: [ 40/235] eta: 0:01:16 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.2229 (0.2267) acc1: 95.5729 (95.4014) acc5: 100.0000 (100.0000) time: 0.3406 data: 0.2487 max mem: 3531 +Test: [ 50/235] eta: 0:01:05 acc1_pos: 94.010 acc5_pos: 100.000 loss: 0.2223 (0.2273) acc1: 95.3125 (95.2819) acc5: 100.0000 (100.0000) time: 0.2635 data: 0.1710 max mem: 3531 +Test: [ 60/235] eta: 0:00:55 acc1_pos: 95.833 acc5_pos: 100.000 loss: 0.2288 (0.2281) acc1: 95.3125 (95.2741) acc5: 100.0000 (100.0000) time: 0.1643 data: 0.0678 max mem: 3531 +Test: [ 70/235] eta: 0:00:48 acc1_pos: 98.462 acc5_pos: 100.000 loss: 0.2479 (0.3690) acc1: 93.7500 (86.1172) acc5: 100.0000 (100.0000) acc1_neg: 22.396 acc5_neg: 100.000 time: 0.1449 data: 0.0479 max mem: 3531 +Test: [ 80/235] eta: 0:00:43 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.3601 (0.4934) acc1: 22.3958 (78.1346) acc5: 100.0000 (100.0000) acc1_neg: 17.969 acc5_neg: 100.000 time: 0.1801 data: 0.0858 max mem: 3531 +Test: [ 90/235] eta: 0:00:39 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.3750 (0.5900) acc1: 21.0938 (71.8922) acc5: 100.0000 (100.0000) acc1_neg: 20.573 acc5_neg: 100.000 time: 0.1949 data: 0.1006 max mem: 3531 +Test: [100/235] eta: 0:00:35 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.3647 (0.6659) acc1: 21.0938 (66.9477) acc5: 100.0000 (100.0000) acc1_neg: 17.969 acc5_neg: 100.000 time: 0.1847 data: 0.0878 max mem: 3531 +Test: [110/235] eta: 0:00:32 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.3443 (0.7284) acc1: 21.3542 (62.9106) acc5: 100.0000 (100.0000) acc1_neg: 24.740 acc5_neg: 100.000 time: 0.1949 data: 0.0976 max mem: 3531 +Test: [120/235] eta: 0:00:29 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4159 (0.7964) acc1: 17.9688 (58.7724) acc5: 100.0000 (100.0000) acc1_neg: 11.979 acc5_neg: 100.000 time: 0.1949 data: 0.0995 max mem: 3531 +Test: [130/235] eta: 0:00:26 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.5585 (0.8547) acc1: 12.2396 (55.1805) acc5: 100.0000 (100.0000) acc1_neg: 11.719 acc5_neg: 100.000 time: 0.1949 data: 0.0974 max mem: 3531 +Test: [140/235] eta: 0:00:23 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.5832 (0.9076) acc1: 10.6771 (52.0298) acc5: 100.0000 (100.0000) acc1_neg: 8.073 acc5_neg: 100.000 time: 0.1999 data: 0.0988 max mem: 3531 +Test: [150/235] eta: 0:00:20 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.5849 (0.9513) acc1: 11.1979 (49.3722) acc5: 100.0000 (100.0000) acc1_neg: 12.240 acc5_neg: 100.000 time: 0.2154 data: 0.1168 max mem: 3531 +Test: [160/235] eta: 0:00:18 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.5723 (0.9900) acc1: 12.2396 (47.0303) acc5: 100.0000 (100.0000) acc1_neg: 11.979 acc5_neg: 100.000 time: 0.2099 data: 0.1154 max mem: 3531 +Test: [170/235] eta: 0:00:15 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.5297 (0.9948) acc1: 13.0208 (46.3328) acc5: 100.0000 (100.0000) acc1_neg: 33.333 acc5_neg: 100.000 time: 0.1894 data: 0.0938 max mem: 3531 +Test: [180/235] eta: 0:00:12 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.0106 (0.9956) acc1: 37.5000 (45.8938) acc5: 100.0000 (100.0000) acc1_neg: 36.458 acc5_neg: 100.000 time: 0.1799 data: 0.0836 max mem: 3531 +Test: [190/235] eta: 0:00:10 acc1_pos: 98.462 acc5_pos: 100.000 loss: 0.9922 (0.9958) acc1: 38.5417 (45.5088) acc5: 100.0000 (100.0000) acc1_neg: 40.104 acc5_neg: 100.000 time: 0.1599 data: 0.0645 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.0041 (0.9968) acc1: 37.7604 (45.1259) acc5: 100.0000 (100.0000) acc1_neg: 37.240 acc5_neg: 100.000 time: 0.1659 data: 0.0711 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.0120 (0.9971) acc1: 37.2396 (44.8398) acc5: 100.0000 (100.0000) acc1_neg: 37.240 acc5_neg: 100.000 time: 0.1675 data: 0.0742 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.0092 (0.9975) acc1: 39.0625 (44.5619) acc5: 100.0000 (100.0000) acc1_neg: 36.719 acc5_neg: 100.000 time: 0.1373 data: 0.0462 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.0069 (0.9973) acc1: 39.5833 (44.3712) acc5: 100.0000 (100.0000) acc1_neg: 43.490 acc5_neg: 100.000 time: 0.1167 data: 0.0262 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.0025 (0.9982) acc1: 39.3229 (44.2895) acc5: 100.0000 (100.0000) acc1_neg: 34.783 acc5_neg: 100.000 time: 0.1135 data: 0.0262 max mem: 3532 +Test: Total time: 0:00:50 (0.2131 s / it) +* Overall Acc@1 44.290 Acc@5 100.000 + target == real Acc@1 95.279 Acc@5 100.000 + target == fake Acc@1 25.848 Acc@5 100.000 + loss 0.998 + PR-AUC 0.8834 ROC-AUC 0.7396 EER 32.46% (thr=0.1953) +Accuracy of the network on the 89879 test images: 44.3% +[rank0]:[W726 06:14:40.194521815 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 06:14:59.690000 1207185 torch/distributed/run.py:766] +W0726 06:14:59.690000 1207185 torch/distributed/run.py:766] ***************************************** +W0726 06:14:59.690000 1207185 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:14:59.690000 1207185 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:15:29.645908016 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:15:29.943651318 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:15:30.012869594 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 06:15:30.126779145 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:15:31.303180486 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W726 06:15:31.304231081 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:15:31.361607871 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:15:31.449885967 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n32_seed9_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n32_classseed9_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:23:21 acc1_pos: 57.292 acc5_pos: 100.000 loss: 0.6724 (0.6724) acc1: 57.2917 (57.2917) acc5: 100.0000 (100.0000) time: 5.2086 data: 3.1344 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/269] eta: 0:02:23 acc1_pos: 72.917 acc5_pos: 100.000 loss: 0.5596 (0.5557) acc1: 73.6979 (73.9820) acc5: 100.0000 (100.0000) time: 0.5558 data: 0.2854 max mem: 3531 +Test: [ 20/269] eta: 0:01:24 acc1_pos: 85.677 acc5_pos: 100.000 loss: 0.5564 (0.5670) acc1: 73.6979 (72.7183) acc5: 100.0000 (100.0000) time: 0.0969 data: 0.0039 max mem: 3531 +Test: [ 30/269] eta: 0:01:13 acc1_pos: 76.562 acc5_pos: 100.000 loss: 0.5445 (0.5603) acc1: 76.0417 (73.9835) acc5: 100.0000 (100.0000) time: 0.1695 data: 0.0757 max mem: 3531 +Test: [ 40/269] eta: 0:01:08 acc1_pos: 81.897 acc5_pos: 100.000 loss: 0.5600 (0.5645) acc1: 73.6979 (73.1707) acc5: 100.0000 (100.0000) acc1_neg: 69.531 acc5_neg: 100.000 time: 0.2536 data: 0.1623 max mem: 3531 +Test: [ 50/269] eta: 0:01:02 acc1_pos: 81.897 acc5_pos: 100.000 loss: 0.6351 (0.5925) acc1: 61.9792 (69.3679) acc5: 100.0000 (100.0000) acc1_neg: 39.323 acc5_neg: 100.000 time: 0.2499 data: 0.1567 max mem: 3531 +Test: [ 60/269] eta: 0:00:57 acc1_pos: 81.897 acc5_pos: 100.000 loss: 0.7318 (0.6257) acc1: 46.0938 (64.7712) acc5: 100.0000 (100.0000) acc1_neg: 29.167 acc5_neg: 100.000 time: 0.2353 data: 0.1414 max mem: 3531 +Test: [ 70/269] eta: 0:00:54 acc1_pos: 81.897 acc5_pos: 100.000 loss: 0.7963 (0.6479) acc1: 41.9271 (61.8655) acc5: 100.0000 (100.0000) acc1_neg: 47.656 acc5_neg: 100.000 time: 0.2399 data: 0.1469 max mem: 3531 +Test: [ 80/269] eta: 0:00:48 acc1_pos: 81.897 acc5_pos: 100.000 loss: 0.7690 (0.6581) acc1: 46.0938 (60.3138) acc5: 100.0000 (100.0000) acc1_neg: 53.646 acc5_neg: 100.000 time: 0.1938 data: 0.0989 max mem: 3531 +Test: [ 90/269] eta: 0:00:45 acc1_pos: 81.897 acc5_pos: 100.000 loss: 0.7720 (0.6754) acc1: 43.7500 (57.8468) acc5: 100.0000 (100.0000) acc1_neg: 52.344 acc5_neg: 100.000 time: 0.1850 data: 0.0901 max mem: 3531 +Test: [100/269] eta: 0:00:42 acc1_pos: 81.897 acc5_pos: 100.000 loss: 0.7857 (0.6838) acc1: 41.1458 (56.6548) acc5: 100.0000 (100.0000) acc1_neg: 37.500 acc5_neg: 100.000 time: 0.2357 data: 0.1404 max mem: 3531 +Test: [110/269] eta: 0:00:39 acc1_pos: 81.897 acc5_pos: 100.000 loss: 0.7869 (0.6980) acc1: 40.1042 (54.8470) acc5: 100.0000 (100.0000) acc1_neg: 29.948 acc5_neg: 100.000 time: 0.2262 data: 0.1322 max mem: 3531 +Test: [120/269] eta: 0:00:36 acc1_pos: 81.897 acc5_pos: 100.000 loss: 0.7166 (0.6966) acc1: 48.6979 (55.0405) acc5: 100.0000 (100.0000) acc1_neg: 51.823 acc5_neg: 100.000 time: 0.1912 data: 0.1005 max mem: 3531 +Test: [130/269] eta: 0:00:33 acc1_pos: 81.897 acc5_pos: 100.000 loss: 0.7122 (0.7033) acc1: 53.9062 (54.4152) acc5: 100.0000 (100.0000) acc1_neg: 39.844 acc5_neg: 100.000 time: 0.2104 data: 0.1171 max mem: 3531 +Test: [140/269] eta: 0:00:30 acc1_pos: 81.897 acc5_pos: 100.000 loss: 0.7929 (0.7105) acc1: 39.8438 (53.4316) acc5: 100.0000 (100.0000) acc1_neg: 39.844 acc5_neg: 100.000 time: 0.1986 data: 0.1045 max mem: 3531 +Test: [150/269] eta: 0:00:27 acc1_pos: 81.897 acc5_pos: 100.000 loss: 0.7510 (0.7097) acc1: 42.1875 (53.5027) acc5: 100.0000 (100.0000) acc1_neg: 64.323 acc5_neg: 100.000 time: 0.1833 data: 0.0909 max mem: 3531 +Test: [160/269] eta: 0:00:25 acc1_pos: 81.897 acc5_pos: 100.000 loss: 0.6283 (0.7049) acc1: 59.3750 (54.0664) acc5: 100.0000 (100.0000) acc1_neg: 63.542 acc5_neg: 100.000 time: 0.2220 data: 0.1286 max mem: 3531 +Test: [170/269] eta: 0:00:22 acc1_pos: 81.897 acc5_pos: 100.000 loss: 0.6451 (0.7031) acc1: 59.3750 (54.2169) acc5: 100.0000 (100.0000) acc1_neg: 56.250 acc5_neg: 100.000 time: 0.2162 data: 0.1227 max mem: 3531 +Test: [180/269] eta: 0:00:20 acc1_pos: 81.897 acc5_pos: 100.000 loss: 0.6451 (0.6995) acc1: 56.2500 (54.7220) acc5: 100.0000 (100.0000) acc1_neg: 54.688 acc5_neg: 100.000 time: 0.2018 data: 0.1096 max mem: 3531 +Test: [190/269] eta: 0:00:18 acc1_pos: 81.897 acc5_pos: 100.000 loss: 0.6679 (0.7018) acc1: 54.6875 (54.4284) acc5: 100.0000 (100.0000) acc1_neg: 38.802 acc5_neg: 100.000 time: 0.1984 data: 0.1037 max mem: 3531 +Test: [200/269] eta: 0:00:15 acc1_pos: 81.897 acc5_pos: 100.000 loss: 0.8016 (0.7105) acc1: 40.3646 (53.7961) acc5: 100.0000 (100.0000) acc1_neg: 66.406 acc5_neg: 100.000 time: 0.1899 data: 0.0926 max mem: 3531 +Test: [210/269] eta: 0:00:13 acc1_pos: 81.897 acc5_pos: 100.000 loss: 0.7433 (0.7096) acc1: 49.7396 (53.8482) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.1999 data: 0.1027 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 81.897 acc5_pos: 100.000 loss: 0.6753 (0.7074) acc1: 53.9062 (54.1125) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.1959 data: 0.1013 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 81.897 acc5_pos: 100.000 loss: 0.6444 (0.7037) acc1: 57.2917 (54.4541) acc5: 100.0000 (100.0000) acc1_neg: 64.062 acc5_neg: 100.000 time: 0.1784 data: 0.0868 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 81.897 acc5_pos: 100.000 loss: 0.6895 (0.7068) acc1: 53.1250 (54.1483) acc5: 100.0000 (100.0000) acc1_neg: 53.646 acc5_neg: 100.000 time: 0.1437 data: 0.0526 max mem: 3532 +Test: [250/269] eta: 0:00:04 acc1_pos: 81.897 acc5_pos: 100.000 loss: 0.7470 (0.7094) acc1: 48.6979 (53.8129) acc5: 100.0000 (100.0000) acc1_neg: 32.812 acc5_neg: 100.000 time: 0.1153 data: 0.0249 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 81.897 acc5_pos: 100.000 loss: 0.8148 (0.7155) acc1: 36.7188 (53.0761) acc5: 100.0000 (100.0000) acc1_neg: 50.781 acc5_neg: 100.000 time: 0.1313 data: 0.0411 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 81.897 acc5_pos: 100.000 loss: 0.7650 (0.7160) acc1: 38.2812 (52.9329) acc5: 100.0000 (100.0000) acc1_neg: 46.194 acc5_neg: 100.000 time: 0.1326 data: 0.0386 max mem: 3532 +Test: Total time: 0:00:56 (0.2083 s / it) +* Overall Acc@1 52.933 Acc@5 100.000 + target == real Acc@1 73.686 Acc@5 100.000 + target == fake Acc@1 50.101 Acc@5 100.000 + loss 0.716 + PR-AUC 0.9320 ROC-AUC 0.6715 EER 38.05% (thr=0.4436) +Accuracy of the network on the 103293 test images: 52.9% +[rank0]:[W726 06:16:36.433287657 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T06:16:42Z] eval_done tag=fixed500k_real_n32_seed9_resampled +[2026-07-26T06:16:42Z] eval_start tag=fixed500k_fake_n64_seed9_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed9_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T06:16:42Z +hostname=n116-040-213 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed9_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n64_seed9_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 06:16:55.485000 1207763 torch/distributed/run.py:766] +W0726 06:16:55.485000 1207763 torch/distributed/run.py:766] ***************************************** +W0726 06:16:55.485000 1207763 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:16:55.485000 1207763 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:17:23.565170123 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:17:23.630120524 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:17:23.071371772 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 06:17:23.098072640 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:17:24.212137898 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:17:24.308813626 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:17:24.419171925 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:17:24.439931676 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n64_seed9_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed9_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:20:54 acc1_pos: 88.021 acc5_pos: 100.000 loss: 0.4306 (0.4306) acc1: 88.0208 (88.0208) acc5: 100.0000 (100.0000) time: 5.3595 data: 3.4391 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/234] eta: 0:02:07 acc1_pos: 68.750 acc5_pos: 100.000 loss: 0.5996 (0.5906) acc1: 69.0104 (70.6913) acc5: 100.0000 (100.0000) time: 0.5708 data: 0.3146 max mem: 3531 +Test: [ 20/234] eta: 0:01:20 acc1_pos: 96.615 acc5_pos: 100.000 loss: 0.6155 (0.5767) acc1: 68.7500 (69.1964) acc5: 100.0000 (100.0000) time: 0.1277 data: 0.0349 max mem: 3531 +Test: [ 30/234] eta: 0:01:13 acc1_pos: 76.562 acc5_pos: 100.000 loss: 0.1913 (0.4647) acc1: 96.3542 (77.0665) acc5: 100.0000 (100.0000) time: 0.2439 data: 0.1483 max mem: 3531 +Test: [ 40/234] eta: 0:01:07 acc1_pos: 99.219 acc5_pos: 100.000 loss: 0.1879 (0.4450) acc1: 96.3542 (78.0170) acc5: 100.0000 (100.0000) time: 0.3144 data: 0.2201 max mem: 3531 +Test: [ 50/234] eta: 0:00:58 acc1_pos: 99.740 acc5_pos: 100.000 loss: 0.1121 (0.3787) acc1: 98.6979 (82.1283) acc5: 100.0000 (100.0000) time: 0.2547 data: 0.1618 max mem: 3531 +Test: [ 60/234] eta: 0:00:55 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.1151 (0.3379) acc1: 99.4792 (85.0324) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2579 data: 0.1659 max mem: 3531 +Test: [ 70/234] eta: 0:00:49 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.1328 (0.3748) acc1: 99.7396 (82.9849) acc5: 100.0000 (100.0000) acc1_neg: 54.948 acc5_neg: 100.000 time: 0.2643 data: 0.1731 max mem: 3531 +Test: [ 80/234] eta: 0:00:45 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.8012 (0.4051) acc1: 57.0312 (81.4333) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2282 data: 0.1366 max mem: 3531 +Test: [ 90/234] eta: 0:00:41 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.3186 (0.3885) acc1: 85.6771 (82.5321) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.2264 data: 0.1337 max mem: 3531 +Test: [100/234] eta: 0:00:37 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.3254 (0.3951) acc1: 86.7188 (82.1473) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2114 data: 0.1183 max mem: 3531 +Test: [110/234] eta: 0:00:33 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.5063 (0.4054) acc1: 72.9167 (81.6019) acc5: 100.0000 (100.0000) acc1_neg: 57.292 acc5_neg: 100.000 time: 0.2076 data: 0.1162 max mem: 3531 +Test: [120/234] eta: 0:00:30 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.4947 (0.4205) acc1: 71.8750 (80.7722) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.1927 data: 0.1004 max mem: 3531 +Test: [130/234] eta: 0:00:26 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.4741 (0.4246) acc1: 78.1250 (80.7987) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.1798 data: 0.0857 max mem: 3531 +Test: [140/234] eta: 0:00:24 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.4741 (0.4484) acc1: 74.2188 (79.6930) acc5: 100.0000 (100.0000) acc1_neg: 39.062 acc5_neg: 100.000 time: 0.2105 data: 0.1174 max mem: 3531 +Test: [150/234] eta: 0:00:21 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.5244 (0.4595) acc1: 73.4375 (79.0597) acc5: 100.0000 (100.0000) acc1_neg: 74.219 acc5_neg: 100.000 time: 0.2292 data: 0.1343 max mem: 3531 +Test: [160/234] eta: 0:00:18 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.5189 (0.4605) acc1: 73.4375 (79.0534) acc5: 100.0000 (100.0000) acc1_neg: 66.667 acc5_neg: 100.000 time: 0.2117 data: 0.1163 max mem: 3531 +Test: [170/234] eta: 0:00:15 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.5795 (0.4667) acc1: 70.5729 (78.6367) acc5: 100.0000 (100.0000) acc1_neg: 76.302 acc5_neg: 100.000 time: 0.2058 data: 0.1121 max mem: 3531 +Test: [180/234] eta: 0:00:13 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.5795 (0.4815) acc1: 69.7917 (77.7207) acc5: 100.0000 (100.0000) acc1_neg: 33.594 acc5_neg: 100.000 time: 0.2097 data: 0.1162 max mem: 3531 +Test: [190/234] eta: 0:00:10 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.7346 (0.4961) acc1: 57.0312 (76.6320) acc5: 100.0000 (100.0000) acc1_neg: 58.854 acc5_neg: 100.000 time: 0.2016 data: 0.1095 max mem: 3531 +Test: [200/234] eta: 0:00:08 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.7230 (0.4882) acc1: 59.8958 (77.0522) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1621 data: 0.0712 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.1171 (0.4723) acc1: 99.7396 (78.0547) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1237 data: 0.0335 max mem: 3532 +Test: [220/234] eta: 0:00:03 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.2166 (0.4619) acc1: 95.8333 (78.7590) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.1040 data: 0.0138 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.4148 (0.4778) acc1: 51.8229 (77.3776) acc5: 100.0000 (100.0000) acc1_neg: 47.656 acc5_neg: 100.000 time: 0.1076 data: 0.0168 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.8096 (0.4821) acc1: 48.9583 (77.0554) acc5: 100.0000 (100.0000) acc1_neg: 52.529 acc5_neg: 100.000 time: 0.1087 data: 0.0168 max mem: 3532 +Test: Total time: 0:00:51 (0.2206 s / it) +* Overall Acc@1 77.055 Acc@5 100.000 + target == real Acc@1 82.255 Acc@5 100.000 + target == fake Acc@1 75.590 Acc@5 100.000 + loss 0.482 + PR-AUC 0.9633 ROC-AUC 0.8770 EER 21.70% (thr=0.4587) +Accuracy of the network on the 89729 test images: 77.1% +[rank0]:[W726 06:18:25.003623511 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 06:18:45.067000 1208310 torch/distributed/run.py:766] +W0726 06:18:45.067000 1208310 torch/distributed/run.py:766] ***************************************** +W0726 06:18:45.067000 1208310 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:18:45.067000 1208310 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:19:14.908034683 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:19:14.994236833 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:19:15.313059659 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:19:15.600692893 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 06:19:15.609860876 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:19:15.623285179 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:19:15.723143471 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W726 06:19:15.723157061 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n64_seed9_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed9_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 5:10:10 acc1_pos: 40.365 acc5_pos: 100.000 loss: 0.8450 (0.8450) acc1: 40.3646 (40.3646) acc5: 100.0000 (100.0000) time: 7.7967 data: 5.8674 max mem: 3531 +Test: [ 10/2387] eta: 0:31:27 acc1_pos: 19.271 acc5_pos: 100.000 loss: 0.8780 (0.8764) acc1: 39.0625 (38.0445) acc5: 100.0000 (100.0000) time: 0.7942 data: 0.5338 max mem: 3531 +Test: [ 20/2387] eta: 0:18:17 acc1_pos: 32.552 acc5_pos: 100.000 loss: 0.8991 (0.9081) acc1: 33.8542 (35.5159) acc5: 100.0000 (100.0000) time: 0.0969 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:14:44 acc1_pos: 23.958 acc5_pos: 100.000 loss: 0.9033 (0.9097) acc1: 32.5521 (35.3411) acc5: 100.0000 (100.0000) time: 0.1449 data: 0.0484 max mem: 3531 +Test: [ 40/2387] eta: 0:13:02 acc1_pos: 44.792 acc5_pos: 100.000 loss: 0.8545 (0.8929) acc1: 37.7604 (36.5219) acc5: 100.0000 (100.0000) time: 0.1968 data: 0.1029 max mem: 3531 +Test: [ 50/2387] eta: 0:12:02 acc1_pos: 45.312 acc5_pos: 100.000 loss: 0.8037 (0.8809) acc1: 44.0104 (37.7094) acc5: 100.0000 (100.0000) time: 0.2069 data: 0.1131 max mem: 3531 +Test: [ 60/2387] eta: 0:11:21 acc1_pos: 55.469 acc5_pos: 100.000 loss: 0.8577 (0.8842) acc1: 40.3646 (37.8159) acc5: 100.0000 (100.0000) time: 0.2091 data: 0.1171 max mem: 3531 +Test: [ 70/2387] eta: 0:10:54 acc1_pos: 39.583 acc5_pos: 100.000 loss: 0.9252 (0.8960) acc1: 35.4167 (36.8765) acc5: 100.0000 (100.0000) time: 0.2143 data: 0.1216 max mem: 3531 +Test: [ 80/2387] eta: 0:10:47 acc1_pos: 67.969 acc5_pos: 100.000 loss: 0.8933 (0.8753) acc1: 39.5833 (38.9564) acc5: 100.0000 (100.0000) time: 0.2438 data: 0.1484 max mem: 3531 +Test: [ 90/2387] eta: 0:10:32 acc1_pos: 57.031 acc5_pos: 100.000 loss: 0.6786 (0.8482) acc1: 55.7292 (41.3690) acc5: 100.0000 (100.0000) time: 0.2505 data: 0.1564 max mem: 3531 +Test: [ 100/2387] eta: 0:10:19 acc1_pos: 61.198 acc5_pos: 100.000 loss: 0.6346 (0.8347) acc1: 58.3333 (42.9739) acc5: 100.0000 (100.0000) time: 0.2314 data: 0.1403 max mem: 3531 +Test: [ 110/2387] eta: 0:10:07 acc1_pos: 60.156 acc5_pos: 100.000 loss: 0.6420 (0.8174) acc1: 63.0208 (44.8714) acc5: 100.0000 (100.0000) time: 0.2279 data: 0.1332 max mem: 3531 +Test: [ 120/2387] eta: 0:09:58 acc1_pos: 58.073 acc5_pos: 100.000 loss: 0.6533 (0.8080) acc1: 60.1562 (45.7042) acc5: 100.0000 (100.0000) time: 0.2286 data: 0.1328 max mem: 3531 +Test: [ 130/2387] eta: 0:09:50 acc1_pos: 51.823 acc5_pos: 100.000 loss: 0.7032 (0.7994) acc1: 57.5521 (46.5848) acc5: 100.0000 (100.0000) time: 0.2332 data: 0.1367 max mem: 3531 +Test: [ 140/2387] eta: 0:09:48 acc1_pos: 58.073 acc5_pos: 100.000 loss: 0.6789 (0.7912) acc1: 56.2500 (47.2444) acc5: 100.0000 (100.0000) time: 0.2511 data: 0.1526 max mem: 3531 +Test: [ 150/2387] eta: 0:09:40 acc1_pos: 39.062 acc5_pos: 100.000 loss: 0.6623 (0.7819) acc1: 57.5521 (48.0443) acc5: 100.0000 (100.0000) time: 0.2444 data: 0.1490 max mem: 3531 +Test: [ 160/2387] eta: 0:09:31 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.6356 (0.7672) acc1: 60.4167 (49.4258) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2185 data: 0.1225 max mem: 3531 +Test: [ 170/2387] eta: 0:09:16 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2700 (0.7376) acc1: 95.0521 (52.1701) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1874 data: 0.0913 max mem: 3531 +Test: [ 180/2387] eta: 0:09:09 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2593 (0.7112) acc1: 96.0938 (54.6069) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1853 data: 0.0916 max mem: 3531 +Test: [ 190/2387] eta: 0:08:57 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2577 (0.6874) acc1: 96.0938 (56.7899) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1896 data: 0.0924 max mem: 3531 +Test: [ 200/2387] eta: 0:08:48 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2538 (0.6660) acc1: 96.3542 (58.7453) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1773 data: 0.0785 max mem: 3531 +Test: [ 210/2387] eta: 0:08:45 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2608 (0.6470) acc1: 96.6146 (60.5339) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2101 data: 0.1162 max mem: 3532 +Test: [ 220/2387] eta: 0:08:39 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2622 (0.6295) acc1: 96.3542 (62.1500) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2185 data: 0.1251 max mem: 3532 +Test: [ 230/2387] eta: 0:08:37 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2607 (0.6135) acc1: 95.8333 (63.6364) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2253 data: 0.1311 max mem: 3532 +Test: [ 240/2387] eta: 0:08:29 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2657 (0.6010) acc1: 94.7917 (64.8816) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.2096 data: 0.1163 max mem: 3532 +Test: [ 250/2387] eta: 0:08:23 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2507 (0.5818) acc1: 94.7917 (66.2434) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1842 data: 0.0908 max mem: 3532 +Test: [ 260/2387] eta: 0:08:19 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.0919 (0.5628) acc1: 100.0000 (67.5367) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2090 data: 0.1167 max mem: 3532 +Test: [ 270/2387] eta: 0:08:14 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.0843 (0.5452) acc1: 100.0000 (68.7337) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2111 data: 0.1200 max mem: 3532 +Test: [ 280/2387] eta: 0:08:09 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.0872 (0.5290) acc1: 100.0000 (69.8463) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1998 data: 0.1090 max mem: 3532 +Test: [ 290/2387] eta: 0:08:03 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.0898 (0.5140) acc1: 100.0000 (70.8826) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1887 data: 0.0937 max mem: 3532 +Test: [ 300/2387] eta: 0:07:56 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.0887 (0.5000) acc1: 100.0000 (71.8499) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1698 data: 0.0732 max mem: 3532 +Test: [ 310/2387] eta: 0:07:49 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.0996 (0.4920) acc1: 100.0000 (72.6345) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1599 data: 0.0645 max mem: 3532 +Test: [ 320/2387] eta: 0:07:43 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.4318 (0.4902) acc1: 92.4479 (73.2331) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1654 data: 0.0692 max mem: 3532 +Test: [ 330/2387] eta: 0:07:39 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.4334 (0.4884) acc1: 92.4479 (73.8301) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1799 data: 0.0851 max mem: 3532 +Test: [ 340/2387] eta: 0:07:34 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.4301 (0.4868) acc1: 92.4479 (74.3730) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1843 data: 0.0897 max mem: 3532 +Test: [ 350/2387] eta: 0:07:30 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.4222 (0.4789) acc1: 94.0104 (75.0794) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0898 max mem: 3532 +Test: [ 360/2387] eta: 0:07:25 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2030 (0.4715) acc1: 98.6979 (75.7243) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1799 data: 0.0849 max mem: 3532 +Test: [ 370/2387] eta: 0:07:20 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2129 (0.4646) acc1: 98.1771 (76.3288) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1699 data: 0.0734 max mem: 3532 +Test: [ 380/2387] eta: 0:07:15 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2129 (0.4579) acc1: 98.1771 (76.9008) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1699 data: 0.0716 max mem: 3532 +Test: [ 390/2387] eta: 0:07:10 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2125 (0.4517) acc1: 98.1771 (77.4470) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1599 data: 0.0628 max mem: 3532 +Test: [ 400/2387] eta: 0:07:06 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2125 (0.4457) acc1: 98.1771 (77.9646) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1658 data: 0.0724 max mem: 3532 +Test: [ 410/2387] eta: 0:07:01 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2121 (0.4401) acc1: 98.1771 (78.4621) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1699 data: 0.0752 max mem: 3532 +Test: [ 420/2387] eta: 0:06:56 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2116 (0.4346) acc1: 98.1771 (78.9353) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1557 data: 0.0575 max mem: 3532 +Test: [ 430/2387] eta: 0:06:54 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2171 (0.4303) acc1: 97.6562 (79.3371) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1899 data: 0.0943 max mem: 3532 +Test: [ 440/2387] eta: 0:06:51 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2186 (0.4243) acc1: 96.3542 (79.7649) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2040 data: 0.1106 max mem: 3533 +Test: [ 450/2387] eta: 0:06:46 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.0938 (0.4170) acc1: 100.0000 (80.2135) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1609 data: 0.0656 max mem: 3533 +Test: [ 460/2387] eta: 0:06:42 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.0929 (0.4099) acc1: 100.0000 (80.6427) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1591 data: 0.0647 max mem: 3533 +Test: [ 470/2387] eta: 0:06:40 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.0929 (0.4032) acc1: 100.0000 (81.0537) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1903 data: 0.0987 max mem: 3533 +Test: [ 480/2387] eta: 0:06:37 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.0943 (0.3968) acc1: 100.0000 (81.4476) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1898 data: 0.0986 max mem: 3533 +Test: [ 490/2387] eta: 0:06:33 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.0940 (0.3906) acc1: 100.0000 (81.8255) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1736 data: 0.0819 max mem: 3533 +Test: [ 500/2387] eta: 0:06:29 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.0943 (0.3847) acc1: 100.0000 (82.1882) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1625 data: 0.0695 max mem: 3533 +Test: [ 510/2387] eta: 0:06:25 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.0944 (0.3793) acc1: 100.0000 (82.5368) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1582 data: 0.0651 max mem: 3533 +Test: [ 520/2387] eta: 0:06:22 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1380 (0.3747) acc1: 100.0000 (82.8720) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1636 data: 0.0705 max mem: 3533 +Test: [ 530/2387] eta: 0:06:18 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1408 (0.3703) acc1: 100.0000 (83.1945) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1627 data: 0.0699 max mem: 3533 +Test: [ 540/2387] eta: 0:06:16 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1401 (0.3661) acc1: 100.0000 (83.5052) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1756 data: 0.0824 max mem: 3533 +Test: [ 550/2387] eta: 0:06:13 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1402 (0.3620) acc1: 100.0000 (83.8045) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1800 data: 0.0862 max mem: 3533 +Test: [ 560/2387] eta: 0:06:09 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1408 (0.3580) acc1: 100.0000 (84.0932) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1629 data: 0.0705 max mem: 3533 +Test: [ 570/2387] eta: 0:06:06 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1397 (0.3542) acc1: 100.0000 (84.3718) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1563 data: 0.0646 max mem: 3533 +Test: [ 580/2387] eta: 0:06:02 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1397 (0.3505) acc1: 100.0000 (84.6408) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1568 data: 0.0655 max mem: 3533 +Test: [ 590/2387] eta: 0:05:59 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1408 (0.3470) acc1: 100.0000 (84.9007) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1610 data: 0.0679 max mem: 3533 +Test: [ 600/2387] eta: 0:05:56 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1410 (0.3436) acc1: 100.0000 (85.1519) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1667 data: 0.0691 max mem: 3533 +Test: [ 610/2387] eta: 0:05:52 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1581 (0.3410) acc1: 100.0000 (85.3856) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1552 data: 0.0584 max mem: 3533 +Test: [ 620/2387] eta: 0:05:49 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1876 (0.3386) acc1: 99.4792 (85.6100) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1464 data: 0.0533 max mem: 3533 +Test: [ 630/2387] eta: 0:05:46 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1879 (0.3362) acc1: 99.2188 (85.8252) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1604 data: 0.0675 max mem: 3533 +Test: [ 640/2387] eta: 0:05:43 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1923 (0.3340) acc1: 99.2188 (86.0313) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1643 data: 0.0725 max mem: 3533 +Test: [ 650/2387] eta: 0:05:41 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1928 (0.3318) acc1: 99.2188 (86.2355) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1647 data: 0.0719 max mem: 3533 +Test: [ 660/2387] eta: 0:05:38 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1898 (0.3297) acc1: 99.4792 (86.4343) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1690 data: 0.0756 max mem: 3533 +Test: [ 670/2387] eta: 0:05:35 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1891 (0.3276) acc1: 99.4792 (86.6275) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1698 data: 0.0785 max mem: 3534 +Test: [ 680/2387] eta: 0:05:33 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1920 (0.3263) acc1: 99.2188 (86.8147) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1722 data: 0.0807 max mem: 3534 +Test: [ 690/2387] eta: 0:05:30 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2668 (0.3283) acc1: 98.9583 (86.7059) acc5: 100.0000 (100.0000) acc1_neg: 68.490 acc5_neg: 100.000 time: 0.1671 data: 0.0751 max mem: 3534 +Test: [ 700/2387] eta: 0:05:27 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.5250 (0.3311) acc1: 73.4375 (86.5148) acc5: 100.0000 (100.0000) acc1_neg: 76.042 acc5_neg: 100.000 time: 0.1651 data: 0.0740 max mem: 3534 +Test: [ 710/2387] eta: 0:05:25 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.5306 (0.3339) acc1: 72.6562 (86.3129) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.1798 data: 0.0878 max mem: 3534 +Test: [ 720/2387] eta: 0:05:23 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.5288 (0.3366) acc1: 72.3958 (86.1304) acc5: 100.0000 (100.0000) acc1_neg: 74.740 acc5_neg: 100.000 time: 0.1896 data: 0.0930 max mem: 3534 +Test: [ 730/2387] eta: 0:05:22 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.5353 (0.3394) acc1: 71.6146 (85.9272) acc5: 100.0000 (100.0000) acc1_neg: 68.229 acc5_neg: 100.000 time: 0.1994 data: 0.1028 max mem: 3534 +Test: [ 740/2387] eta: 0:05:19 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.5353 (0.3420) acc1: 71.6146 (85.7488) acc5: 100.0000 (100.0000) acc1_neg: 72.135 acc5_neg: 100.000 time: 0.1778 data: 0.0847 max mem: 3534 +Test: [ 750/2387] eta: 0:05:17 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.5222 (0.3444) acc1: 72.1354 (85.5793) acc5: 100.0000 (100.0000) acc1_neg: 73.698 acc5_neg: 100.000 time: 0.1640 data: 0.0715 max mem: 3534 +Test: [ 760/2387] eta: 0:05:14 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.5178 (0.3463) acc1: 73.4375 (85.4502) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.1796 data: 0.0875 max mem: 3534 +Test: [ 770/2387] eta: 0:05:12 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.4169 (0.3460) acc1: 80.4688 (85.4819) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1819 data: 0.0896 max mem: 3534 +Test: [ 780/2387] eta: 0:05:09 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2271 (0.3443) acc1: 95.3125 (85.6214) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1700 data: 0.0777 max mem: 3534 +Test: [ 790/2387] eta: 0:05:08 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1830 (0.3425) acc1: 97.6562 (85.7367) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.1790 data: 0.0874 max mem: 3534 +Test: [ 800/2387] eta: 0:05:06 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1744 (0.3405) acc1: 98.1771 (85.8907) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1999 data: 0.1071 max mem: 3534 +Test: [ 810/2387] eta: 0:05:04 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1901 (0.3388) acc1: 98.6979 (86.0329) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2032 data: 0.1106 max mem: 3534 +Test: [ 820/2387] eta: 0:05:03 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2173 (0.3376) acc1: 96.6146 (86.1402) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.2164 data: 0.1235 max mem: 3534 +Test: [ 830/2387] eta: 0:05:01 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2681 (0.3379) acc1: 91.6667 (86.1374) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.2009 data: 0.1063 max mem: 3534 +Test: [ 840/2387] eta: 0:04:59 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2930 (0.3371) acc1: 91.6667 (86.2574) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1833 data: 0.0908 max mem: 3534 +Test: [ 850/2387] eta: 0:04:57 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2642 (0.3362) acc1: 96.3542 (86.3671) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1901 data: 0.0990 max mem: 3534 +Test: [ 860/2387] eta: 0:04:55 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2414 (0.3351) acc1: 97.1354 (86.5046) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1968 data: 0.1043 max mem: 3534 +Test: [ 870/2387] eta: 0:04:53 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2324 (0.3340) acc1: 98.4375 (86.6300) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1994 data: 0.1069 max mem: 3534 +Test: [ 880/2387] eta: 0:04:51 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2537 (0.3332) acc1: 96.8750 (86.7412) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2006 data: 0.1093 max mem: 3534 +Test: [ 890/2387] eta: 0:04:49 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2497 (0.3322) acc1: 96.8750 (86.8687) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1937 data: 0.1024 max mem: 3534 +Test: [ 900/2387] eta: 0:04:47 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2665 (0.3369) acc1: 95.5729 (86.5471) acc5: 100.0000 (100.0000) acc1_neg: 51.562 acc5_neg: 100.000 time: 0.1956 data: 0.1039 max mem: 3535 +Test: [ 910/2387] eta: 0:04:46 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.8761 (0.3429) acc1: 47.1354 (86.1153) acc5: 100.0000 (100.0000) acc1_neg: 44.792 acc5_neg: 100.000 time: 0.2149 data: 0.1231 max mem: 3535 +Test: [ 920/2387] eta: 0:04:44 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.8907 (0.3488) acc1: 47.3958 (85.6963) acc5: 100.0000 (100.0000) acc1_neg: 44.271 acc5_neg: 100.000 time: 0.2176 data: 0.1252 max mem: 3535 +Test: [ 930/2387] eta: 0:04:42 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.8759 (0.3545) acc1: 47.9167 (85.2933) acc5: 100.0000 (100.0000) acc1_neg: 42.188 acc5_neg: 100.000 time: 0.2026 data: 0.1099 max mem: 3535 +Test: [ 940/2387] eta: 0:04:40 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.8900 (0.3602) acc1: 47.3958 (84.8867) acc5: 100.0000 (100.0000) acc1_neg: 44.010 acc5_neg: 100.000 time: 0.1911 data: 0.0989 max mem: 3535 +Test: [ 950/2387] eta: 0:04:38 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.8798 (0.3657) acc1: 46.3542 (84.4851) acc5: 100.0000 (100.0000) acc1_neg: 48.698 acc5_neg: 100.000 time: 0.1850 data: 0.0934 max mem: 3535 +Test: [ 960/2387] eta: 0:04:36 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.8686 (0.3709) acc1: 47.6562 (84.1043) acc5: 100.0000 (100.0000) acc1_neg: 50.260 acc5_neg: 100.000 time: 0.1807 data: 0.0892 max mem: 3535 +Test: [ 970/2387] eta: 0:04:34 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.8775 (0.3763) acc1: 46.8750 (83.7171) acc5: 100.0000 (100.0000) acc1_neg: 46.875 acc5_neg: 100.000 time: 0.1829 data: 0.0913 max mem: 3535 +Test: [ 980/2387] eta: 0:04:32 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.8644 (0.3773) acc1: 46.8750 (83.6567) acc5: 100.0000 (100.0000) acc1_neg: 79.427 acc5_neg: 100.000 time: 0.1819 data: 0.0907 max mem: 3535 +Test: [ 990/2387] eta: 0:04:30 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2745 (0.3763) acc1: 92.7083 (83.7367) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1770 data: 0.0851 max mem: 3535 +Test: [1000/2387] eta: 0:04:28 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2451 (0.3751) acc1: 95.0521 (83.8438) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1766 data: 0.0843 max mem: 3535 +Test: [1010/2387] eta: 0:04:26 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2594 (0.3746) acc1: 93.4896 (83.8936) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1820 data: 0.0903 max mem: 3535 +Test: [1020/2387] eta: 0:04:23 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2849 (0.3738) acc1: 91.6667 (83.9559) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1833 data: 0.0910 max mem: 3535 +Test: [1030/2387] eta: 0:04:21 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1934 (0.3720) acc1: 97.3958 (84.0974) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0866 max mem: 3535 +Test: [1040/2387] eta: 0:04:19 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1730 (0.3700) acc1: 99.4792 (84.2429) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1752 data: 0.0807 max mem: 3535 +Test: [1050/2387] eta: 0:04:17 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1605 (0.3680) acc1: 99.7396 (84.3876) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1729 data: 0.0782 max mem: 3535 +Test: [1060/2387] eta: 0:04:15 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1650 (0.3662) acc1: 99.4792 (84.5233) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1696 data: 0.0770 max mem: 3535 +Test: [1070/2387] eta: 0:04:13 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1857 (0.3649) acc1: 98.1771 (84.6361) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1755 data: 0.0840 max mem: 3535 +Test: [1080/2387] eta: 0:04:11 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2491 (0.3642) acc1: 94.2708 (84.7077) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1869 data: 0.0959 max mem: 3535 +Test: [1090/2387] eta: 0:04:09 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2682 (0.3630) acc1: 92.9688 (84.7984) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1883 data: 0.0973 max mem: 3535 +Test: [1100/2387] eta: 0:04:07 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2208 (0.3617) acc1: 96.0938 (84.9027) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1981 data: 0.1053 max mem: 3535 +Test: [1110/2387] eta: 0:04:05 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2184 (0.3607) acc1: 96.0938 (84.9753) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1932 data: 0.0997 max mem: 3535 +Test: [1120/2387] eta: 0:04:03 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2349 (0.3597) acc1: 95.3125 (85.0605) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.1885 data: 0.0948 max mem: 3536 +Test: [1130/2387] eta: 0:04:01 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2358 (0.3586) acc1: 95.0521 (85.1519) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1850 data: 0.0919 max mem: 3536 +Test: [1140/2387] eta: 0:03:59 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2719 (0.3582) acc1: 93.4896 (85.2005) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1861 data: 0.0946 max mem: 3536 +Test: [1150/2387] eta: 0:03:57 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2867 (0.3573) acc1: 91.9271 (85.2859) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1984 data: 0.1069 max mem: 3536 +Test: [1160/2387] eta: 0:03:55 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1892 (0.3557) acc1: 98.4375 (85.3965) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1811 data: 0.0885 max mem: 3536 +Test: [1170/2387] eta: 0:03:53 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1621 (0.3542) acc1: 100.0000 (85.5183) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1813 data: 0.0873 max mem: 3536 +Test: [1180/2387] eta: 0:03:51 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1621 (0.3526) acc1: 99.7396 (85.6334) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1881 data: 0.0943 max mem: 3536 +Test: [1190/2387] eta: 0:03:49 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1717 (0.3512) acc1: 99.4792 (85.7477) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1805 data: 0.0875 max mem: 3536 +Test: [1200/2387] eta: 0:03:47 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1658 (0.3497) acc1: 99.7396 (85.8638) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1898 data: 0.0979 max mem: 3536 +Test: [1210/2387] eta: 0:03:45 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1674 (0.3482) acc1: 100.0000 (85.9786) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1951 data: 0.1042 max mem: 3536 +Test: [1220/2387] eta: 0:03:43 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1132 (0.3462) acc1: 100.0000 (86.0913) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1948 data: 0.1036 max mem: 3536 +Test: [1230/2387] eta: 0:03:42 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.0952 (0.3442) acc1: 100.0000 (86.2043) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1948 data: 0.1038 max mem: 3536 +Test: [1240/2387] eta: 0:03:40 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.0929 (0.3422) acc1: 100.0000 (86.3152) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1921 data: 0.0991 max mem: 3536 +Test: [1250/2387] eta: 0:03:37 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.0942 (0.3402) acc1: 100.0000 (86.4240) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1806 data: 0.0869 max mem: 3536 +Test: [1260/2387] eta: 0:03:35 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.0961 (0.3383) acc1: 100.0000 (86.5304) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0779 max mem: 3536 +Test: [1270/2387] eta: 0:03:33 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.0967 (0.3364) acc1: 100.0000 (86.6362) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1770 data: 0.0821 max mem: 3536 +Test: [1280/2387] eta: 0:03:31 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1102 (0.3347) acc1: 100.0000 (86.7373) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1646 data: 0.0704 max mem: 3536 +Test: [1290/2387] eta: 0:03:29 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1223 (0.3332) acc1: 99.7396 (86.8249) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1564 data: 0.0651 max mem: 3536 +Test: [1300/2387] eta: 0:03:27 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1253 (0.3316) acc1: 99.4792 (86.9177) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1704 data: 0.0765 max mem: 3536 +Test: [1310/2387] eta: 0:03:25 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1277 (0.3302) acc1: 99.2188 (87.0064) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1736 data: 0.0796 max mem: 3536 +Test: [1320/2387] eta: 0:03:23 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1450 (0.3288) acc1: 99.2188 (87.0917) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1723 data: 0.0804 max mem: 3536 +Test: [1330/2387] eta: 0:03:21 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1594 (0.3279) acc1: 98.1771 (87.1629) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1704 data: 0.0787 max mem: 3536 +Test: [1340/2387] eta: 0:03:19 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2328 (0.3280) acc1: 93.2292 (87.1565) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1816 data: 0.0876 max mem: 3536 +Test: [1350/2387] eta: 0:03:17 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.0769 (0.3261) acc1: 100.0000 (87.2515) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1854 data: 0.0919 max mem: 3537 +Test: [1360/2387] eta: 0:03:15 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.0665 (0.3242) acc1: 100.0000 (87.3452) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1731 data: 0.0819 max mem: 3537 +Test: [1370/2387] eta: 0:03:13 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.0656 (0.3223) acc1: 100.0000 (87.4375) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1751 data: 0.0842 max mem: 3537 +Test: [1380/2387] eta: 0:03:11 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.0651 (0.3205) acc1: 100.0000 (87.5285) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1781 data: 0.0866 max mem: 3537 +Test: [1390/2387] eta: 0:03:09 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.0624 (0.3186) acc1: 100.0000 (87.6181) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1741 data: 0.0824 max mem: 3537 +Test: [1400/2387] eta: 0:03:07 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.0644 (0.3168) acc1: 100.0000 (87.7065) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1742 data: 0.0834 max mem: 3537 +Test: [1410/2387] eta: 0:03:05 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.0762 (0.3159) acc1: 100.0000 (87.7689) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1769 data: 0.0856 max mem: 3537 +Test: [1420/2387] eta: 0:03:03 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2133 (0.3153) acc1: 96.6146 (87.8181) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1789 data: 0.0858 max mem: 3537 +Test: [1430/2387] eta: 0:03:01 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2133 (0.3146) acc1: 96.0938 (87.8758) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1803 data: 0.0870 max mem: 3537 +Test: [1440/2387] eta: 0:02:59 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2276 (0.3141) acc1: 95.5729 (87.9287) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1811 data: 0.0896 max mem: 3537 +Test: [1450/2387] eta: 0:02:57 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2388 (0.3138) acc1: 94.5312 (87.9623) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1782 data: 0.0875 max mem: 3537 +Test: [1460/2387] eta: 0:02:55 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1816 (0.3128) acc1: 96.8750 (88.0299) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1781 data: 0.0865 max mem: 3537 +Test: [1470/2387] eta: 0:02:53 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1452 (0.3116) acc1: 99.2188 (88.1063) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1850 data: 0.0921 max mem: 3537 +Test: [1480/2387] eta: 0:02:51 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1352 (0.3105) acc1: 99.4792 (88.1773) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1906 data: 0.0960 max mem: 3537 +Test: [1490/2387] eta: 0:02:49 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1372 (0.3094) acc1: 99.4792 (88.2502) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1984 data: 0.1047 max mem: 3537 +Test: [1500/2387] eta: 0:02:48 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1417 (0.3083) acc1: 99.4792 (88.3208) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2048 data: 0.1114 max mem: 3537 +Test: [1510/2387] eta: 0:02:46 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1580 (0.3074) acc1: 99.4792 (88.3957) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2028 data: 0.1098 max mem: 3537 +Test: [1520/2387] eta: 0:02:44 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1811 (0.3068) acc1: 99.2188 (88.4579) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2033 data: 0.1118 max mem: 3537 +Test: [1530/2387] eta: 0:02:42 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2258 (0.3063) acc1: 98.1771 (88.5145) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2000 data: 0.1076 max mem: 3537 +Test: [1540/2387] eta: 0:02:40 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2140 (0.3057) acc1: 98.9583 (88.5792) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1949 data: 0.1032 max mem: 3537 +Test: [1550/2387] eta: 0:02:38 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2025 (0.3051) acc1: 99.2188 (88.6469) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1933 data: 0.1023 max mem: 3537 +Test: [1560/2387] eta: 0:02:37 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2025 (0.3045) acc1: 98.9583 (88.7067) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2024 data: 0.1109 max mem: 3537 +Test: [1570/2387] eta: 0:02:35 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2005 (0.3039) acc1: 98.9583 (88.7686) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2100 data: 0.1185 max mem: 3537 +Test: [1580/2387] eta: 0:02:33 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2041 (0.3033) acc1: 98.9583 (88.8174) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1974 data: 0.1054 max mem: 3538 +Test: [1590/2387] eta: 0:02:31 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1849 (0.3025) acc1: 97.1354 (88.8748) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1887 data: 0.0962 max mem: 3538 +Test: [1600/2387] eta: 0:02:29 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1738 (0.3018) acc1: 97.9167 (88.9106) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1902 data: 0.0985 max mem: 3538 +Test: [1610/2387] eta: 0:02:27 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1579 (0.3010) acc1: 98.6979 (88.9707) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1825 data: 0.0911 max mem: 3538 +Test: [1620/2387] eta: 0:02:25 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1929 (0.3005) acc1: 97.3958 (89.0087) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1620 data: 0.0706 max mem: 3538 +Test: [1630/2387] eta: 0:02:23 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2364 (0.3003) acc1: 94.5312 (89.0349) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1578 data: 0.0644 max mem: 3538 +Test: [1640/2387] eta: 0:02:21 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.3113 (0.3006) acc1: 89.5833 (89.0243) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1546 data: 0.0608 max mem: 3538 +Test: [1650/2387] eta: 0:02:19 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.3178 (0.3006) acc1: 89.5833 (89.0295) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1464 data: 0.0548 max mem: 3538 +Test: [1660/2387] eta: 0:02:17 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.3119 (0.3008) acc1: 88.5417 (89.0230) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1491 data: 0.0581 max mem: 3538 +Test: [1670/2387] eta: 0:02:15 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.3130 (0.3009) acc1: 88.2812 (89.0240) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1494 data: 0.0562 max mem: 3538 +Test: [1680/2387] eta: 0:02:13 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.3171 (0.3010) acc1: 90.1042 (89.0233) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1457 data: 0.0516 max mem: 3538 +Test: [1690/2387] eta: 0:02:11 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.3171 (0.3012) acc1: 89.8438 (89.0158) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1481 data: 0.0560 max mem: 3538 +Test: [1700/2387] eta: 0:02:09 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.3126 (0.3010) acc1: 89.8438 (89.0359) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1555 data: 0.0645 max mem: 3538 +Test: [1710/2387] eta: 0:02:07 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1377 (0.3000) acc1: 99.2188 (89.0971) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1549 data: 0.0633 max mem: 3538 +Test: [1720/2387] eta: 0:02:05 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1273 (0.2989) acc1: 99.7396 (89.1586) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1701 data: 0.0785 max mem: 3538 +Test: [1730/2387] eta: 0:02:03 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1194 (0.2979) acc1: 99.7396 (89.2185) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1608 data: 0.0690 max mem: 3538 +Test: [1740/2387] eta: 0:02:01 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1187 (0.2970) acc1: 99.7396 (89.2771) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1512 data: 0.0588 max mem: 3538 +Test: [1750/2387] eta: 0:01:59 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1174 (0.2960) acc1: 99.7396 (89.3357) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1719 data: 0.0781 max mem: 3538 +Test: [1760/2387] eta: 0:01:57 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1289 (0.2951) acc1: 99.4792 (89.3918) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1599 data: 0.0654 max mem: 3538 +Test: [1770/2387] eta: 0:01:55 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1633 (0.2945) acc1: 98.4375 (89.4391) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1518 data: 0.0583 max mem: 3538 +Test: [1780/2387] eta: 0:01:53 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1671 (0.2937) acc1: 98.4375 (89.4928) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1786 data: 0.0843 max mem: 3538 +Test: [1790/2387] eta: 0:01:51 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1580 (0.2929) acc1: 99.2188 (89.5480) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1742 data: 0.0800 max mem: 3538 +Test: [1800/2387] eta: 0:01:49 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1462 (0.2921) acc1: 99.7396 (89.6016) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1573 data: 0.0624 max mem: 3538 +Test: [1810/2387] eta: 0:01:47 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1462 (0.2913) acc1: 99.4792 (89.6557) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1638 data: 0.0695 max mem: 3539 +Test: [1820/2387] eta: 0:01:45 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1422 (0.2905) acc1: 99.4792 (89.7069) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1626 data: 0.0701 max mem: 3539 +Test: [1830/2387] eta: 0:01:43 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1465 (0.2897) acc1: 99.2188 (89.7598) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1597 data: 0.0673 max mem: 3539 +Test: [1840/2387] eta: 0:01:41 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1345 (0.2889) acc1: 100.0000 (89.8155) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1647 data: 0.0704 max mem: 3539 +Test: [1850/2387] eta: 0:01:39 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1334 (0.2881) acc1: 100.0000 (89.8703) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1808 data: 0.0858 max mem: 3539 +Test: [1860/2387] eta: 0:01:38 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1333 (0.2872) acc1: 100.0000 (89.9248) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0958 max mem: 3539 +Test: [1870/2387] eta: 0:01:36 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1334 (0.2864) acc1: 100.0000 (89.9785) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1664 data: 0.0729 max mem: 3539 +Test: [1880/2387] eta: 0:01:34 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1330 (0.2856) acc1: 100.0000 (90.0316) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1454 data: 0.0536 max mem: 3539 +Test: [1890/2387] eta: 0:01:32 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1326 (0.2848) acc1: 100.0000 (90.0843) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1500 data: 0.0582 max mem: 3539 +Test: [1900/2387] eta: 0:01:30 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1341 (0.2840) acc1: 100.0000 (90.1364) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1512 data: 0.0593 max mem: 3539 +Test: [1910/2387] eta: 0:01:28 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1632 (0.2834) acc1: 100.0000 (90.1850) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1497 data: 0.0574 max mem: 3539 +Test: [1920/2387] eta: 0:01:26 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1719 (0.2833) acc1: 99.7396 (90.1966) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1581 data: 0.0654 max mem: 3539 +Test: [1930/2387] eta: 0:01:24 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1747 (0.2827) acc1: 98.9583 (90.2440) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1851 data: 0.0898 max mem: 3539 +Test: [1940/2387] eta: 0:01:22 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1684 (0.2823) acc1: 99.2188 (90.2563) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1848 data: 0.0898 max mem: 3539 +Test: [1950/2387] eta: 0:01:20 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1960 (0.2819) acc1: 98.9583 (90.2908) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1425 data: 0.0481 max mem: 3539 +Test: [1960/2387] eta: 0:01:18 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2013 (0.2817) acc1: 98.4375 (90.3152) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1558 data: 0.0592 max mem: 3539 +Test: [1970/2387] eta: 0:01:17 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2660 (0.2820) acc1: 93.4896 (90.2872) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.2097 data: 0.1162 max mem: 3539 +Test: [1980/2387] eta: 0:01:15 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2742 (0.2818) acc1: 91.4062 (90.3039) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1962 data: 0.1043 max mem: 3539 +Test: [1990/2387] eta: 0:01:13 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1438 (0.2811) acc1: 98.6979 (90.3461) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1633 data: 0.0710 max mem: 3539 +Test: [2000/2387] eta: 0:01:11 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1429 (0.2804) acc1: 98.6979 (90.3887) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1726 data: 0.0810 max mem: 3539 +Test: [2010/2387] eta: 0:01:09 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1420 (0.2797) acc1: 98.6979 (90.4305) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1665 data: 0.0730 max mem: 3539 +Test: [2020/2387] eta: 0:01:07 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1438 (0.2791) acc1: 98.6979 (90.4723) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1449 data: 0.0510 max mem: 3539 +Test: [2030/2387] eta: 0:01:05 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1438 (0.2784) acc1: 98.6979 (90.5137) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1510 data: 0.0574 max mem: 3540 +Test: [2040/2387] eta: 0:01:03 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1428 (0.2778) acc1: 98.9583 (90.5553) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1634 data: 0.0680 max mem: 3540 +Test: [2050/2387] eta: 0:01:01 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1426 (0.2771) acc1: 99.2188 (90.5977) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1640 data: 0.0703 max mem: 3540 +Test: [2060/2387] eta: 0:01:00 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1484 (0.2765) acc1: 98.9583 (90.6374) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1575 data: 0.0656 max mem: 3540 +Test: [2070/2387] eta: 0:00:58 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1533 (0.2759) acc1: 98.9583 (90.6786) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1547 data: 0.0614 max mem: 3540 +Test: [2080/2387] eta: 0:00:56 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1485 (0.2753) acc1: 98.9583 (90.7184) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1566 data: 0.0634 max mem: 3540 +Test: [2090/2387] eta: 0:00:54 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1484 (0.2747) acc1: 99.2188 (90.7585) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1603 data: 0.0684 max mem: 3540 +Test: [2100/2387] eta: 0:00:52 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1486 (0.2741) acc1: 99.2188 (90.7988) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1599 data: 0.0686 max mem: 3540 +Test: [2110/2387] eta: 0:00:50 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1487 (0.2735) acc1: 99.2188 (90.8379) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1509 data: 0.0585 max mem: 3540 +Test: [2120/2387] eta: 0:00:48 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1630 (0.2732) acc1: 98.6979 (90.8652) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1498 data: 0.0569 max mem: 3540 +Test: [2130/2387] eta: 0:00:47 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1766 (0.2728) acc1: 98.4375 (90.9018) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1548 data: 0.0630 max mem: 3540 +Test: [2140/2387] eta: 0:00:45 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1713 (0.2724) acc1: 98.4375 (90.9202) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1594 data: 0.0681 max mem: 3540 +Test: [2150/2387] eta: 0:00:43 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1848 (0.2720) acc1: 98.4375 (90.9570) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1655 data: 0.0735 max mem: 3540 +Test: [2160/2387] eta: 0:00:41 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1875 (0.2717) acc1: 98.6979 (90.9875) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1680 data: 0.0761 max mem: 3540 +Test: [2170/2387] eta: 0:00:39 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2256 (0.2718) acc1: 95.5729 (90.9994) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1657 data: 0.0744 max mem: 3540 +Test: [2180/2387] eta: 0:00:37 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.3224 (0.2731) acc1: 89.5833 (90.9091) acc5: 100.0000 (100.0000) acc1_neg: 40.625 acc5_neg: 100.000 time: 0.1649 data: 0.0714 max mem: 3540 +Test: [2190/2387] eta: 0:00:35 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.3642 (0.2737) acc1: 85.4167 (90.8748) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1634 data: 0.0702 max mem: 3540 +Test: [2200/2387] eta: 0:00:34 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1819 (0.2732) acc1: 97.6562 (90.9111) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1660 data: 0.0745 max mem: 3540 +Test: [2210/2387] eta: 0:00:32 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1643 (0.2727) acc1: 99.7396 (90.9503) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1786 data: 0.0867 max mem: 3540 +Test: [2220/2387] eta: 0:00:30 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1643 (0.2722) acc1: 99.4792 (90.9858) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1598 data: 0.0674 max mem: 3540 +Test: [2230/2387] eta: 0:00:28 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1617 (0.2717) acc1: 99.4792 (91.0239) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1507 data: 0.0581 max mem: 3540 +Test: [2240/2387] eta: 0:00:26 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1572 (0.2712) acc1: 99.2188 (91.0596) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1665 data: 0.0726 max mem: 3540 +Test: [2250/2387] eta: 0:00:24 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1719 (0.2709) acc1: 99.2188 (91.0919) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1712 data: 0.0754 max mem: 3540 +Test: [2260/2387] eta: 0:00:23 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1892 (0.2705) acc1: 99.4792 (91.1279) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1770 data: 0.0814 max mem: 3541 +Test: [2270/2387] eta: 0:00:21 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1586 (0.2699) acc1: 99.7396 (91.1656) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1729 data: 0.0793 max mem: 3541 +Test: [2280/2387] eta: 0:00:19 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1368 (0.2695) acc1: 100.0000 (91.1857) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1678 data: 0.0748 max mem: 3541 +Test: [2290/2387] eta: 0:00:17 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1337 (0.2689) acc1: 100.0000 (91.2227) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1664 data: 0.0740 max mem: 3541 +Test: [2300/2387] eta: 0:00:15 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1473 (0.2685) acc1: 99.7396 (91.2547) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1792 data: 0.0862 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.1788 (0.2682) acc1: 98.6979 (91.2829) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1887 data: 0.0963 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2145 (0.2681) acc1: 96.3542 (91.2961) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1788 data: 0.0879 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2475 (0.2681) acc1: 94.7917 (91.3089) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1753 data: 0.0843 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2212 (0.2679) acc1: 96.3542 (91.3326) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1809 data: 0.0901 max mem: 3541 +Test: [2350/2387] eta: 0:00:06 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2064 (0.2680) acc1: 97.1354 (91.3366) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1764 data: 0.0844 max mem: 3541 +Test: [2360/2387] eta: 0:00:04 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2371 (0.2679) acc1: 97.1354 (91.3468) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1684 data: 0.0762 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2589 (0.2680) acc1: 93.2292 (91.3451) acc5: 100.0000 (100.0000) acc1_neg: 76.042 acc5_neg: 100.000 time: 0.1683 data: 0.0769 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.2945 (0.2684) acc1: 89.8438 (91.3332) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.1694 data: 0.0770 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 70.320 acc5_pos: 100.000 loss: 0.3127 (0.2683) acc1: 89.8438 (91.3356) acc5: 100.0000 (100.0000) acc1_neg: 95.349 acc5_neg: 100.000 time: 0.1550 data: 0.0658 max mem: 3541 +Test: Total time: 0:07:13 (0.1817 s / it) +* Overall Acc@1 91.336 Acc@5 100.000 + target == real Acc@1 48.461 Acc@5 100.000 + target == fake Acc@1 94.347 Acc@5 100.000 + loss 0.268 + PR-AUC 0.9933 ROC-AUC 0.9113 EER 15.95% (thr=0.6997) +Accuracy of the network on the 916267 test images: 91.3% +[rank0]:[W726 06:26:58.978686645 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 06:27:19.522000 1208942 torch/distributed/run.py:766] +W0726 06:27:19.522000 1208942 torch/distributed/run.py:766] ***************************************** +W0726 06:27:19.522000 1208942 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:27:19.522000 1208942 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:27:47.843598315 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 06:27:48.303310574 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:27:48.489348677 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:27:48.922388774 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:27:48.994425851 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +[rank1]:[W726 06:27:48.023823125 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:27:48.037696965 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:27:48.123821115 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n64_seed9_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed9_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:09:09 acc1_pos: 40.365 acc5_pos: 100.000 loss: 0.8450 (0.8450) acc1: 40.3646 (40.3646) acc5: 100.0000 (100.0000) time: 8.0253 data: 6.3589 max mem: 3531 +Test: [ 10/517] eta: 0:06:54 acc1_pos: 19.271 acc5_pos: 100.000 loss: 0.8780 (0.8764) acc1: 39.0625 (38.0445) acc5: 100.0000 (100.0000) time: 0.8184 data: 0.5784 max mem: 3531 +Test: [ 20/517] eta: 0:03:56 acc1_pos: 32.552 acc5_pos: 100.000 loss: 0.8991 (0.9081) acc1: 33.8542 (35.5159) acc5: 100.0000 (100.0000) time: 0.0988 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:03:35 acc1_pos: 23.958 acc5_pos: 100.000 loss: 0.9033 (0.9097) acc1: 32.5521 (35.3411) acc5: 100.0000 (100.0000) time: 0.2349 data: 0.1394 max mem: 3531 +Test: [ 40/517] eta: 0:03:17 acc1_pos: 44.792 acc5_pos: 100.000 loss: 0.8545 (0.8929) acc1: 37.7604 (36.5219) acc5: 100.0000 (100.0000) time: 0.3499 data: 0.2560 max mem: 3531 +Test: [ 50/517] eta: 0:03:04 acc1_pos: 69.725 acc5_pos: 100.000 loss: 0.7902 (0.8618) acc1: 44.0104 (39.6957) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.3222 data: 0.2250 max mem: 3531 +Test: [ 60/517] eta: 0:02:52 acc1_pos: 69.725 acc5_pos: 100.000 loss: 0.4446 (0.7889) acc1: 81.5104 (47.7544) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.2999 data: 0.2049 max mem: 3531 +Test: [ 70/517] eta: 0:02:40 acc1_pos: 69.725 acc5_pos: 100.000 loss: 0.4037 (0.7351) acc1: 91.1458 (53.8622) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.2675 data: 0.1748 max mem: 3531 +Test: [ 80/517] eta: 0:02:30 acc1_pos: 69.725 acc5_pos: 100.000 loss: 0.4030 (0.6953) acc1: 91.1458 (58.1854) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.2467 data: 0.1535 max mem: 3531 +Test: [ 90/517] eta: 0:02:22 acc1_pos: 69.725 acc5_pos: 100.000 loss: 0.4140 (0.6639) acc1: 89.3229 (61.7731) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.2449 data: 0.1508 max mem: 3531 +Test: [100/517] eta: 0:02:15 acc1_pos: 69.725 acc5_pos: 100.000 loss: 0.3876 (0.6354) acc1: 92.7083 (64.9237) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2431 data: 0.1440 max mem: 3531 +Test: [110/517] eta: 0:02:09 acc1_pos: 69.725 acc5_pos: 100.000 loss: 0.3326 (0.6059) acc1: 96.6146 (67.8163) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2451 data: 0.1448 max mem: 3531 +Test: [120/517] eta: 0:02:04 acc1_pos: 69.725 acc5_pos: 100.000 loss: 0.3535 (0.5863) acc1: 95.3125 (69.8971) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2549 data: 0.1570 max mem: 3531 +Test: [130/517] eta: 0:01:59 acc1_pos: 69.725 acc5_pos: 100.000 loss: 0.3624 (0.5684) acc1: 93.4896 (71.7995) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2563 data: 0.1584 max mem: 3531 +Test: [140/517] eta: 0:01:54 acc1_pos: 69.725 acc5_pos: 100.000 loss: 0.3351 (0.5508) acc1: 97.1354 (73.6813) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2424 data: 0.1464 max mem: 3531 +Test: [150/517] eta: 0:01:50 acc1_pos: 69.725 acc5_pos: 100.000 loss: 0.3200 (0.5359) acc1: 98.1771 (75.2794) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2433 data: 0.1496 max mem: 3531 +Test: [160/517] eta: 0:01:45 acc1_pos: 69.725 acc5_pos: 100.000 loss: 0.3119 (0.5215) acc1: 97.9167 (76.6773) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2478 data: 0.1526 max mem: 3531 +Test: [170/517] eta: 0:01:41 acc1_pos: 69.725 acc5_pos: 100.000 loss: 0.3083 (0.5077) acc1: 98.1771 (77.9255) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2419 data: 0.1447 max mem: 3531 +Test: [180/517] eta: 0:01:38 acc1_pos: 69.725 acc5_pos: 100.000 loss: 0.2832 (0.4963) acc1: 97.3958 (78.8473) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.2504 data: 0.1552 max mem: 3531 +Test: [190/517] eta: 0:01:34 acc1_pos: 69.725 acc5_pos: 100.000 loss: 0.2484 (0.4834) acc1: 99.4792 (79.9343) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2486 data: 0.1539 max mem: 3531 +Test: [200/517] eta: 0:01:30 acc1_pos: 69.725 acc5_pos: 100.000 loss: 0.2602 (0.4729) acc1: 99.7396 (80.9157) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2440 data: 0.1469 max mem: 3531 +Test: [210/517] eta: 0:01:27 acc1_pos: 69.725 acc5_pos: 100.000 loss: 0.2885 (0.4693) acc1: 97.6562 (81.2401) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.2502 data: 0.1522 max mem: 3532 +Test: [220/517] eta: 0:01:23 acc1_pos: 69.725 acc5_pos: 100.000 loss: 0.4015 (0.4659) acc1: 88.0208 (81.5552) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.2449 data: 0.1474 max mem: 3532 +Test: [230/517] eta: 0:01:20 acc1_pos: 69.725 acc5_pos: 100.000 loss: 0.4050 (0.4636) acc1: 89.5833 (82.0087) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2340 data: 0.1358 max mem: 3532 +Test: [240/517] eta: 0:01:17 acc1_pos: 69.725 acc5_pos: 100.000 loss: 0.3958 (0.4595) acc1: 93.4896 (82.5359) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.2313 data: 0.1331 max mem: 3532 +Test: [250/517] eta: 0:01:14 acc1_pos: 69.725 acc5_pos: 100.000 loss: 0.3436 (0.4554) acc1: 97.9167 (83.0874) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2408 data: 0.1448 max mem: 3532 +Test: [260/517] eta: 0:01:11 acc1_pos: 69.725 acc5_pos: 100.000 loss: 0.3346 (0.4509) acc1: 99.2188 (83.6855) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2501 data: 0.1549 max mem: 3532 +Test: [270/517] eta: 0:01:08 acc1_pos: 69.725 acc5_pos: 100.000 loss: 0.3696 (0.4484) acc1: 97.3958 (84.1415) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2556 data: 0.1606 max mem: 3532 +Test: [280/517] eta: 0:01:05 acc1_pos: 69.725 acc5_pos: 100.000 loss: 0.3697 (0.4450) acc1: 96.8750 (84.6021) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2483 data: 0.1532 max mem: 3532 +Test: [290/517] eta: 0:01:02 acc1_pos: 69.725 acc5_pos: 100.000 loss: 0.3462 (0.4418) acc1: 97.6562 (85.0498) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2485 data: 0.1517 max mem: 3532 +Test: [300/517] eta: 0:00:59 acc1_pos: 69.725 acc5_pos: 100.000 loss: 0.3462 (0.4385) acc1: 97.1354 (85.4435) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2549 data: 0.1573 max mem: 3532 +Test: [310/517] eta: 0:00:56 acc1_pos: 69.725 acc5_pos: 100.000 loss: 0.3565 (0.4367) acc1: 96.3542 (85.7407) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2420 data: 0.1444 max mem: 3532 +Test: [320/517] eta: 0:00:53 acc1_pos: 69.725 acc5_pos: 100.000 loss: 0.3561 (0.4336) acc1: 95.8333 (86.0511) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2458 data: 0.1506 max mem: 3532 +Test: [330/517] eta: 0:00:50 acc1_pos: 69.725 acc5_pos: 100.000 loss: 0.3035 (0.4295) acc1: 96.3542 (86.3875) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2545 data: 0.1602 max mem: 3532 +Test: [340/517] eta: 0:00:47 acc1_pos: 69.725 acc5_pos: 100.000 loss: 0.2914 (0.4279) acc1: 98.1771 (86.4652) acc5: 100.0000 (100.0000) acc1_neg: 67.708 acc5_neg: 100.000 time: 0.2402 data: 0.1454 max mem: 3532 +Test: [350/517] eta: 0:00:44 acc1_pos: 69.725 acc5_pos: 100.000 loss: 0.3695 (0.4281) acc1: 91.4062 (86.4784) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.2395 data: 0.1450 max mem: 3532 +Test: [360/517] eta: 0:00:42 acc1_pos: 69.725 acc5_pos: 100.000 loss: 0.3790 (0.4267) acc1: 92.1875 (86.6531) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.2588 data: 0.1638 max mem: 3532 +Test: [370/517] eta: 0:00:39 acc1_pos: 69.725 acc5_pos: 100.000 loss: 0.3269 (0.4234) acc1: 97.1354 (86.9750) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2443 data: 0.1483 max mem: 3532 +Test: [380/517] eta: 0:00:36 acc1_pos: 69.725 acc5_pos: 100.000 loss: 0.3077 (0.4220) acc1: 96.6146 (87.1029) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.2360 data: 0.1393 max mem: 3532 +Test: [390/517] eta: 0:00:33 acc1_pos: 69.725 acc5_pos: 100.000 loss: 0.3812 (0.4209) acc1: 91.6667 (87.2309) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2440 data: 0.1478 max mem: 3532 +Test: [400/517] eta: 0:00:31 acc1_pos: 69.725 acc5_pos: 100.000 loss: 0.3307 (0.4183) acc1: 95.0521 (87.4786) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2356 data: 0.1386 max mem: 3532 +Test: [410/517] eta: 0:00:28 acc1_pos: 69.725 acc5_pos: 100.000 loss: 0.3256 (0.4180) acc1: 95.3125 (87.5849) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.2202 data: 0.1244 max mem: 3532 +Test: [420/517] eta: 0:00:25 acc1_pos: 69.725 acc5_pos: 100.000 loss: 0.4007 (0.4173) acc1: 94.0104 (87.7579) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2195 data: 0.1240 max mem: 3532 +Test: [430/517] eta: 0:00:22 acc1_pos: 69.725 acc5_pos: 100.000 loss: 0.3718 (0.4161) acc1: 94.2708 (87.8716) acc5: 100.0000 (100.0000) acc1_neg: 74.219 acc5_neg: 100.000 time: 0.2446 data: 0.1484 max mem: 3532 +Test: [440/517] eta: 0:00:20 acc1_pos: 69.725 acc5_pos: 100.000 loss: 0.4311 (0.4172) acc1: 86.9792 (87.7699) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.2385 data: 0.1423 max mem: 3533 +Test: [450/517] eta: 0:00:17 acc1_pos: 69.725 acc5_pos: 100.000 loss: 0.4311 (0.4166) acc1: 88.0208 (87.9152) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.2357 data: 0.1392 max mem: 3533 +Test: [460/517] eta: 0:00:14 acc1_pos: 69.725 acc5_pos: 100.000 loss: 0.3958 (0.4168) acc1: 91.1458 (87.9547) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.2718 data: 0.1754 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 69.725 acc5_pos: 100.000 loss: 0.4045 (0.4162) acc1: 91.1458 (88.0573) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2740 data: 0.1783 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 69.725 acc5_pos: 100.000 loss: 0.3672 (0.4151) acc1: 94.5312 (88.2168) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2405 data: 0.1446 max mem: 3533 +Test: [490/517] eta: 0:00:07 acc1_pos: 69.725 acc5_pos: 100.000 loss: 0.3594 (0.4137) acc1: 96.3542 (88.3757) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2123 data: 0.1174 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 69.725 acc5_pos: 100.000 loss: 0.3802 (0.4149) acc1: 95.0521 (88.3197) acc5: 100.0000 (100.0000) acc1_neg: 72.656 acc5_neg: 100.000 time: 0.1937 data: 0.1027 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 69.725 acc5_pos: 100.000 loss: 0.4034 (0.4148) acc1: 90.1042 (88.3495) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1679 data: 0.0779 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 69.725 acc5_pos: 100.000 loss: 0.3836 (0.4141) acc1: 95.5729 (88.4466) acc5: 100.0000 (100.0000) acc1_neg: 96.820 acc5_neg: 100.000 time: 0.1695 data: 0.0778 max mem: 3533 +Test: Total time: 0:02:12 (0.2555 s / it) +* Overall Acc@1 88.447 Acc@5 100.000 + target == real Acc@1 37.212 Acc@5 100.000 + target == fake Acc@1 93.762 Acc@5 100.000 + loss 0.414 + PR-AUC 0.9613 ROC-AUC 0.7678 EER 30.07% (thr=0.6606) +Accuracy of the network on the 198427 test images: 88.4% +[rank0]:[W726 06:30:10.125012387 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 06:30:31.518000 1209502 torch/distributed/run.py:766] +W0726 06:30:31.518000 1209502 torch/distributed/run.py:766] ***************************************** +W0726 06:30:31.518000 1209502 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:30:31.518000 1209502 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:30:59.800754067 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:31:00.609286801 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W726 06:31:00.609328185 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:31:00.796113811 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 1): env:// +| distributed init (rank 3): env:// +[rank0]:[W726 06:31:00.910636215 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:31:00.924324427 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W726 06:31:00.924351646 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W726 06:31:00.924361516 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n64_seed9_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed9_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:31:49 acc1_pos: 48.177 acc5_pos: 100.000 loss: 0.7005 (0.7005) acc1: 48.1771 (48.1771) acc5: 100.0000 (100.0000) time: 4.9094 data: 3.3326 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/389] eta: 0:03:20 acc1_pos: 92.188 acc5_pos: 100.000 loss: 0.5987 (0.5926) acc1: 70.3125 (66.7614) acc5: 100.0000 (100.0000) time: 0.5284 data: 0.3034 max mem: 3531 +Test: [ 20/389] eta: 0:02:05 acc1_pos: 51.823 acc5_pos: 100.000 loss: 0.5754 (0.5990) acc1: 70.3125 (67.1751) acc5: 100.0000 (100.0000) time: 0.1122 data: 0.0202 max mem: 3531 +Test: [ 30/389] eta: 0:01:48 acc1_pos: 80.990 acc5_pos: 100.000 loss: 0.6388 (0.6102) acc1: 63.8021 (66.5491) acc5: 100.0000 (100.0000) time: 0.1769 data: 0.0826 max mem: 3531 +Test: [ 40/389] eta: 0:01:40 acc1_pos: 39.323 acc5_pos: 100.000 loss: 0.6388 (0.6150) acc1: 63.8021 (66.0252) acc5: 100.0000 (100.0000) time: 0.2333 data: 0.1401 max mem: 3531 +Test: [ 50/389] eta: 0:01:28 acc1_pos: 84.896 acc5_pos: 100.000 loss: 0.5449 (0.5989) acc1: 74.7396 (68.0198) acc5: 100.0000 (100.0000) time: 0.2013 data: 0.1088 max mem: 3531 +Test: [ 60/389] eta: 0:01:21 acc1_pos: 77.604 acc5_pos: 100.000 loss: 0.4663 (0.5743) acc1: 84.3750 (70.9614) acc5: 100.0000 (100.0000) time: 0.1607 data: 0.0685 max mem: 3531 +Test: [ 70/389] eta: 0:01:16 acc1_pos: 71.875 acc5_pos: 100.000 loss: 0.4669 (0.5768) acc1: 80.4688 (70.4079) acc5: 100.0000 (100.0000) time: 0.1791 data: 0.0876 max mem: 3531 +Test: [ 80/389] eta: 0:01:12 acc1_pos: 95.052 acc5_pos: 100.000 loss: 0.5865 (0.5837) acc1: 64.0625 (69.5891) acc5: 100.0000 (100.0000) time: 0.2020 data: 0.1095 max mem: 3531 +Test: [ 90/389] eta: 0:01:08 acc1_pos: 74.479 acc5_pos: 100.000 loss: 0.4035 (0.5629) acc1: 82.5521 (71.3971) acc5: 100.0000 (100.0000) time: 0.1908 data: 0.0979 max mem: 3531 +Test: [100/389] eta: 0:01:04 acc1_pos: 59.375 acc5_pos: 100.000 loss: 0.5416 (0.5730) acc1: 75.7812 (70.2609) acc5: 100.0000 (100.0000) time: 0.1686 data: 0.0773 max mem: 3531 +Test: [110/389] eta: 0:01:01 acc1_pos: 99.740 acc5_pos: 100.000 loss: 0.6933 (0.5804) acc1: 59.8958 (69.0831) acc5: 100.0000 (100.0000) time: 0.1889 data: 0.0983 max mem: 3531 +Test: [120/389] eta: 0:00:58 acc1_pos: 82.552 acc5_pos: 100.000 loss: 0.5262 (0.5759) acc1: 74.4792 (69.4645) acc5: 100.0000 (100.0000) time: 0.2033 data: 0.1111 max mem: 3531 +Test: [130/389] eta: 0:00:55 acc1_pos: 50.920 acc5_pos: 100.000 loss: 0.5511 (0.5809) acc1: 72.3958 (68.9587) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1835 data: 0.0876 max mem: 3531 +Test: [140/389] eta: 0:00:53 acc1_pos: 50.920 acc5_pos: 100.000 loss: 0.6170 (0.5853) acc1: 66.1458 (68.6392) acc5: 100.0000 (100.0000) acc1_neg: 15.885 acc5_neg: 100.000 time: 0.1770 data: 0.0825 max mem: 3531 +Test: [150/389] eta: 0:00:50 acc1_pos: 50.920 acc5_pos: 100.000 loss: 0.6273 (0.5963) acc1: 56.7708 (67.3203) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1817 data: 0.0904 max mem: 3531 +Test: [160/389] eta: 0:00:48 acc1_pos: 50.920 acc5_pos: 100.000 loss: 0.6776 (0.5903) acc1: 56.7708 (67.9930) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1848 data: 0.0934 max mem: 3531 +Test: [170/389] eta: 0:00:45 acc1_pos: 50.920 acc5_pos: 100.000 loss: 0.3389 (0.5833) acc1: 97.9167 (68.6129) acc5: 100.0000 (100.0000) acc1_neg: 47.656 acc5_neg: 100.000 time: 0.1886 data: 0.0960 max mem: 3531 +Test: [180/389] eta: 0:00:43 acc1_pos: 50.920 acc5_pos: 100.000 loss: 0.3935 (0.5812) acc1: 84.3750 (68.9054) acc5: 100.0000 (100.0000) acc1_neg: 72.135 acc5_neg: 100.000 time: 0.1846 data: 0.0921 max mem: 3531 +Test: [190/389] eta: 0:00:40 acc1_pos: 50.920 acc5_pos: 100.000 loss: 0.5704 (0.5871) acc1: 67.1875 (68.1828) acc5: 100.0000 (100.0000) acc1_neg: 6.771 acc5_neg: 100.000 time: 0.1790 data: 0.0876 max mem: 3531 +Test: [200/389] eta: 0:00:38 acc1_pos: 50.920 acc5_pos: 100.000 loss: 0.6702 (0.5935) acc1: 45.3125 (67.3793) acc5: 100.0000 (100.0000) acc1_neg: 37.760 acc5_neg: 100.000 time: 0.1832 data: 0.0908 max mem: 3531 +Test: [210/389] eta: 0:00:36 acc1_pos: 50.920 acc5_pos: 100.000 loss: 0.4383 (0.5881) acc1: 63.0208 (67.8861) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0951 max mem: 3532 +Test: [220/389] eta: 0:00:34 acc1_pos: 50.920 acc5_pos: 100.000 loss: 0.3323 (0.5788) acc1: 97.1354 (68.8843) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.1949 data: 0.1005 max mem: 3532 +Test: [230/389] eta: 0:00:31 acc1_pos: 50.920 acc5_pos: 100.000 loss: 0.4720 (0.5816) acc1: 75.2604 (68.4208) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.1619 data: 0.0679 max mem: 3532 +Test: [240/389] eta: 0:00:29 acc1_pos: 50.920 acc5_pos: 100.000 loss: 0.5256 (0.5799) acc1: 73.1771 (68.7446) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1449 data: 0.0491 max mem: 3532 +Test: [250/389] eta: 0:00:27 acc1_pos: 50.920 acc5_pos: 100.000 loss: 0.4666 (0.5742) acc1: 86.7188 (69.7450) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1729 data: 0.0793 max mem: 3532 +Test: [260/389] eta: 0:00:25 acc1_pos: 50.920 acc5_pos: 100.000 loss: 0.5009 (0.5771) acc1: 86.7188 (69.7837) acc5: 100.0000 (100.0000) acc1_neg: 16.667 acc5_neg: 100.000 time: 0.1865 data: 0.0947 max mem: 3532 +Test: [270/389] eta: 0:00:23 acc1_pos: 50.920 acc5_pos: 100.000 loss: 0.6051 (0.5867) acc1: 69.0104 (68.7587) acc5: 100.0000 (100.0000) acc1_neg: 69.531 acc5_neg: 100.000 time: 0.1864 data: 0.0944 max mem: 3532 +Test: [280/389] eta: 0:00:21 acc1_pos: 50.920 acc5_pos: 100.000 loss: 0.6463 (0.5942) acc1: 53.1250 (68.1004) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1733 data: 0.0814 max mem: 3532 +Test: [290/389] eta: 0:00:19 acc1_pos: 50.920 acc5_pos: 100.000 loss: 0.6152 (0.5912) acc1: 63.8021 (68.4252) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.1683 data: 0.0739 max mem: 3532 +Test: [300/389] eta: 0:00:17 acc1_pos: 50.920 acc5_pos: 100.000 loss: 0.5704 (0.5913) acc1: 74.4792 (68.5692) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.1679 data: 0.0735 max mem: 3532 +Test: [310/389] eta: 0:00:15 acc1_pos: 50.920 acc5_pos: 100.000 loss: 0.6137 (0.5982) acc1: 66.4062 (67.9210) acc5: 100.0000 (100.0000) acc1_neg: 6.510 acc5_neg: 100.000 time: 0.1550 data: 0.0631 max mem: 3532 +Test: [320/389] eta: 0:00:13 acc1_pos: 50.920 acc5_pos: 100.000 loss: 0.6342 (0.5987) acc1: 61.4583 (67.8828) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1580 data: 0.0662 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 50.920 acc5_pos: 100.000 loss: 0.7895 (0.6052) acc1: 38.2812 (66.9892) acc5: 100.0000 (100.0000) acc1_neg: 46.354 acc5_neg: 100.000 time: 0.1571 data: 0.0657 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 50.920 acc5_pos: 100.000 loss: 0.5004 (0.5976) acc1: 65.3646 (67.8382) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1539 data: 0.0631 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 50.920 acc5_pos: 100.000 loss: 0.3388 (0.6067) acc1: 97.9167 (67.2469) acc5: 100.0000 (100.0000) acc1_neg: 9.635 acc5_neg: 100.000 time: 0.1544 data: 0.0625 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 50.920 acc5_pos: 100.000 loss: 0.7073 (0.6120) acc1: 49.4792 (66.6393) acc5: 100.0000 (100.0000) acc1_neg: 55.990 acc5_neg: 100.000 time: 0.1502 data: 0.0578 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 50.920 acc5_pos: 100.000 loss: 0.6867 (0.6120) acc1: 56.2500 (66.6976) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1495 data: 0.0589 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 50.920 acc5_pos: 100.000 loss: 0.5425 (0.6113) acc1: 64.5833 (66.7255) acc5: 100.0000 (100.0000) acc1_neg: 60.417 acc5_neg: 100.000 time: 0.1338 data: 0.0438 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 50.920 acc5_pos: 100.000 loss: 0.5026 (0.6088) acc1: 79.6875 (67.0590) acc5: 100.0000 (100.0000) acc1_neg: 76.503 acc5_neg: 100.000 time: 0.1247 data: 0.0312 max mem: 3532 +Test: Total time: 0:01:11 (0.1841 s / it) +* Overall Acc@1 67.059 Acc@5 100.000 + target == real Acc@1 69.138 Acc@5 100.000 + target == fake Acc@1 66.093 Acc@5 100.000 + loss 0.609 + PR-AUC 0.8560 ROC-AUC 0.7407 EER 32.58% (thr=0.4907) +Accuracy of the network on the 149358 test images: 67.1% +[rank0]:[W726 06:32:22.867478894 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 06:32:42.997000 1210061 torch/distributed/run.py:766] +W0726 06:32:42.997000 1210061 torch/distributed/run.py:766] ***************************************** +W0726 06:32:42.997000 1210061 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:32:42.997000 1210061 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:33:13.435201945 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:33:13.561286091 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 06:33:13.637350645 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:33:13.065635341 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +[rank1]:[W726 06:33:13.092725731 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:33:13.101644021 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:33:14.207377942 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:33:14.393738885 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n64_seed9_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed9_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:32:42 acc1_pos: 83.594 acc5_pos: 100.000 loss: 0.3849 (0.3849) acc1: 83.5938 (83.5938) acc5: 100.0000 (100.0000) time: 5.8932 data: 4.5049 max mem: 3531 +Test: [ 10/333] eta: 0:03:19 acc1_pos: 88.281 acc5_pos: 100.000 loss: 0.2865 (0.2932) acc1: 88.2812 (89.3703) acc5: 100.0000 (100.0000) time: 0.6181 data: 0.4100 max mem: 3531 +Test: [ 20/333] eta: 0:02:03 acc1_pos: 95.573 acc5_pos: 100.000 loss: 0.2808 (0.2807) acc1: 89.8438 (89.6949) acc5: 100.0000 (100.0000) time: 0.1200 data: 0.0270 max mem: 3531 +Test: [ 30/333] eta: 0:01:40 acc1_pos: 66.146 acc5_pos: 100.000 loss: 0.3583 (0.3998) acc1: 80.2083 (81.1324) acc5: 100.0000 (100.0000) time: 0.1751 data: 0.0800 max mem: 3531 +Test: [ 40/333] eta: 0:01:36 acc1_pos: 66.146 acc5_pos: 100.000 loss: 0.6509 (0.4693) acc1: 61.7188 (75.7495) acc5: 100.0000 (100.0000) time: 0.2599 data: 0.1661 max mem: 3531 +Test: [ 50/333] eta: 0:01:25 acc1_pos: 63.802 acc5_pos: 100.000 loss: 0.6509 (0.5075) acc1: 60.4167 (72.6205) acc5: 100.0000 (100.0000) time: 0.2544 data: 0.1593 max mem: 3531 +Test: [ 60/333] eta: 0:01:13 acc1_pos: 59.115 acc5_pos: 100.000 loss: 0.6447 (0.5317) acc1: 60.6771 (71.0425) acc5: 100.0000 (100.0000) time: 0.1449 data: 0.0467 max mem: 3531 +Test: [ 70/333] eta: 0:01:05 acc1_pos: 58.073 acc5_pos: 100.000 loss: 0.6511 (0.5495) acc1: 60.6771 (69.8980) acc5: 100.0000 (100.0000) time: 0.1100 data: 0.0118 max mem: 3531 +Test: [ 80/333] eta: 0:01:01 acc1_pos: 64.844 acc5_pos: 100.000 loss: 0.6606 (0.5620) acc1: 60.4167 (68.7275) acc5: 100.0000 (100.0000) time: 0.1624 data: 0.0676 max mem: 3531 +Test: [ 90/333] eta: 0:00:57 acc1_pos: 70.312 acc5_pos: 100.000 loss: 0.5965 (0.5623) acc1: 66.9271 (69.0419) acc5: 100.0000 (100.0000) time: 0.1925 data: 0.1004 max mem: 3531 +Test: [100/333] eta: 0:00:54 acc1_pos: 76.042 acc5_pos: 100.000 loss: 0.5710 (0.5646) acc1: 70.3125 (69.0001) acc5: 100.0000 (100.0000) time: 0.1947 data: 0.1026 max mem: 3531 +Test: [110/333] eta: 0:00:50 acc1_pos: 77.604 acc5_pos: 100.000 loss: 0.5809 (0.5663) acc1: 67.9688 (68.8227) acc5: 100.0000 (100.0000) time: 0.1936 data: 0.0966 max mem: 3531 +Test: [120/333] eta: 0:00:47 acc1_pos: 75.521 acc5_pos: 100.000 loss: 0.5403 (0.5663) acc1: 70.3125 (68.9179) acc5: 100.0000 (100.0000) time: 0.1799 data: 0.0831 max mem: 3531 +Test: [130/333] eta: 0:00:44 acc1_pos: 81.771 acc5_pos: 100.000 loss: 0.5201 (0.5653) acc1: 75.5208 (69.1615) acc5: 100.0000 (100.0000) time: 0.1653 data: 0.0730 max mem: 3531 +Test: [140/333] eta: 0:00:42 acc1_pos: 79.948 acc5_pos: 100.000 loss: 0.5367 (0.5635) acc1: 74.7396 (69.4112) acc5: 100.0000 (100.0000) time: 0.1795 data: 0.0862 max mem: 3531 +Test: [150/333] eta: 0:00:39 acc1_pos: 72.917 acc5_pos: 100.000 loss: 0.5553 (0.5648) acc1: 71.0938 (69.3484) acc5: 100.0000 (100.0000) time: 0.2033 data: 0.1088 max mem: 3531 +Test: [160/333] eta: 0:00:37 acc1_pos: 64.145 acc5_pos: 100.000 loss: 0.5826 (0.5786) acc1: 68.4896 (68.4734) acc5: 100.0000 (100.0000) acc1_neg: 34.115 acc5_neg: 100.000 time: 0.2005 data: 0.1076 max mem: 3531 +Test: [170/333] eta: 0:00:35 acc1_pos: 64.145 acc5_pos: 100.000 loss: 0.9090 (0.6007) acc1: 42.1875 (66.7885) acc5: 100.0000 (100.0000) acc1_neg: 46.094 acc5_neg: 100.000 time: 0.2019 data: 0.1096 max mem: 3531 +Test: [180/333] eta: 0:00:32 acc1_pos: 64.145 acc5_pos: 100.000 loss: 0.8762 (0.6133) acc1: 44.0104 (65.7214) acc5: 100.0000 (100.0000) acc1_neg: 44.531 acc5_neg: 100.000 time: 0.1889 data: 0.0933 max mem: 3531 +Test: [190/333] eta: 0:00:30 acc1_pos: 64.145 acc5_pos: 100.000 loss: 0.8130 (0.6212) acc1: 47.3958 (65.0974) acc5: 100.0000 (100.0000) acc1_neg: 51.562 acc5_neg: 100.000 time: 0.1856 data: 0.0912 max mem: 3531 +Test: [200/333] eta: 0:00:27 acc1_pos: 64.145 acc5_pos: 100.000 loss: 0.7787 (0.6294) acc1: 51.3021 (64.5600) acc5: 100.0000 (100.0000) acc1_neg: 51.302 acc5_neg: 100.000 time: 0.1885 data: 0.0950 max mem: 3531 +Test: [210/333] eta: 0:00:25 acc1_pos: 64.145 acc5_pos: 100.000 loss: 0.8466 (0.6418) acc1: 49.4792 (63.7046) acc5: 100.0000 (100.0000) acc1_neg: 44.010 acc5_neg: 100.000 time: 0.1789 data: 0.0840 max mem: 3532 +Test: [220/333] eta: 0:00:23 acc1_pos: 64.145 acc5_pos: 100.000 loss: 0.8803 (0.6555) acc1: 45.0521 (62.7993) acc5: 100.0000 (100.0000) acc1_neg: 66.146 acc5_neg: 100.000 time: 0.2099 data: 0.1160 max mem: 3532 +Test: [230/333] eta: 0:00:21 acc1_pos: 64.145 acc5_pos: 100.000 loss: 0.6170 (0.6524) acc1: 65.8854 (63.1324) acc5: 100.0000 (100.0000) acc1_neg: 69.271 acc5_neg: 100.000 time: 0.2082 data: 0.1153 max mem: 3532 +Test: [240/333] eta: 0:00:19 acc1_pos: 64.145 acc5_pos: 100.000 loss: 0.5973 (0.6556) acc1: 65.8854 (62.7150) acc5: 100.0000 (100.0000) acc1_neg: 53.125 acc5_neg: 100.000 time: 0.1659 data: 0.0724 max mem: 3532 +Test: [250/333] eta: 0:00:17 acc1_pos: 64.145 acc5_pos: 100.000 loss: 0.7580 (0.6596) acc1: 53.1250 (62.4035) acc5: 100.0000 (100.0000) acc1_neg: 48.177 acc5_neg: 100.000 time: 0.1853 data: 0.0900 max mem: 3532 +Test: [260/333] eta: 0:00:15 acc1_pos: 64.145 acc5_pos: 100.000 loss: 0.7623 (0.6647) acc1: 54.1667 (62.0460) acc5: 100.0000 (100.0000) acc1_neg: 59.375 acc5_neg: 100.000 time: 0.1967 data: 0.1030 max mem: 3532 +Test: [270/333] eta: 0:00:12 acc1_pos: 64.145 acc5_pos: 100.000 loss: 0.7389 (0.6666) acc1: 56.7708 (61.9196) acc5: 100.0000 (100.0000) acc1_neg: 59.375 acc5_neg: 100.000 time: 0.1845 data: 0.0918 max mem: 3532 +Test: [280/333] eta: 0:00:10 acc1_pos: 64.145 acc5_pos: 100.000 loss: 0.7154 (0.6744) acc1: 55.7292 (61.2600) acc5: 100.0000 (100.0000) acc1_neg: 47.135 acc5_neg: 100.000 time: 0.1777 data: 0.0852 max mem: 3532 +Test: [290/333] eta: 0:00:08 acc1_pos: 64.145 acc5_pos: 100.000 loss: 0.9987 (0.6856) acc1: 34.1146 (60.3415) acc5: 100.0000 (100.0000) acc1_neg: 39.583 acc5_neg: 100.000 time: 0.1728 data: 0.0814 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 64.145 acc5_pos: 100.000 loss: 0.9675 (0.6944) acc1: 37.2396 (59.6882) acc5: 100.0000 (100.0000) acc1_neg: 45.573 acc5_neg: 100.000 time: 0.1793 data: 0.0872 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 64.145 acc5_pos: 100.000 loss: 0.8637 (0.6951) acc1: 45.5729 (59.6044) acc5: 100.0000 (100.0000) acc1_neg: 57.031 acc5_neg: 100.000 time: 0.1673 data: 0.0754 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 64.145 acc5_pos: 100.000 loss: 0.6762 (0.6963) acc1: 58.8542 (59.4707) acc5: 100.0000 (100.0000) acc1_neg: 34.635 acc5_neg: 100.000 time: 0.1461 data: 0.0550 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 64.145 acc5_pos: 100.000 loss: 1.0412 (0.7090) acc1: 36.4583 (58.6669) acc5: 100.0000 (100.0000) acc1_neg: 33.854 acc5_neg: 100.000 time: 0.1305 data: 0.0392 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 64.145 acc5_pos: 100.000 loss: 1.0770 (0.7114) acc1: 34.6354 (58.5359) acc5: 100.0000 (100.0000) acc1_neg: 37.746 acc5_neg: 100.000 time: 0.1238 data: 0.0294 max mem: 3532 +Test: Total time: 0:01:05 (0.1963 s / it) +* Overall Acc@1 58.536 Acc@5 100.000 + target == real Acc@1 69.436 Acc@5 100.000 + target == fake Acc@1 48.833 Acc@5 100.000 + loss 0.711 + PR-AUC 0.6429 ROC-AUC 0.6338 EER 40.67% (thr=0.4312) +Accuracy of the network on the 127843 test images: 58.5% +[rank0]:[W726 06:34:32.524928751 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 06:34:52.470000 1210619 torch/distributed/run.py:766] +W0726 06:34:52.470000 1210619 torch/distributed/run.py:766] ***************************************** +W0726 06:34:52.470000 1210619 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:34:52.470000 1210619 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:35:23.629670246 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:35:23.644874294 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +[rank6]:[W726 06:35:23.076688910 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:35:23.106209881 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 06:35:23.117440931 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 1): env:// +| distributed init (rank 7): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:35:24.546773292 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:35:24.554081260 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W726 06:35:24.554122146 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n64_seed9_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed9_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:26:52 acc1_pos: 97.135 acc5_pos: 100.000 loss: 0.1790 (0.1790) acc1: 97.1354 (97.1354) acc5: 100.0000 (100.0000) time: 6.8635 data: 5.1024 max mem: 3531 +Test: [ 10/235] eta: 0:02:38 acc1_pos: 96.094 acc5_pos: 100.000 loss: 0.1916 (0.1922) acc1: 96.3542 (96.3542) acc5: 100.0000 (100.0000) time: 0.7063 data: 0.4643 max mem: 3531 +Test: [ 20/235] eta: 0:01:29 acc1_pos: 96.615 acc5_pos: 100.000 loss: 0.1863 (0.1893) acc1: 96.6146 (96.6022) acc5: 100.0000 (100.0000) time: 0.0949 data: 0.0004 max mem: 3531 +Test: [ 30/235] eta: 0:01:23 acc1_pos: 97.396 acc5_pos: 100.000 loss: 0.1845 (0.1883) acc1: 96.8750 (96.6482) acc5: 100.0000 (100.0000) time: 0.2446 data: 0.1472 max mem: 3531 +Test: [ 40/235] eta: 0:01:15 acc1_pos: 95.833 acc5_pos: 100.000 loss: 0.1861 (0.1877) acc1: 96.3542 (96.5701) acc5: 100.0000 (100.0000) time: 0.3602 data: 0.2668 max mem: 3531 +Test: [ 50/235] eta: 0:01:06 acc1_pos: 97.135 acc5_pos: 100.000 loss: 0.1845 (0.1882) acc1: 96.3542 (96.5278) acc5: 100.0000 (100.0000) time: 0.2849 data: 0.1917 max mem: 3531 +Test: [ 60/235] eta: 0:00:59 acc1_pos: 96.354 acc5_pos: 100.000 loss: 0.1910 (0.1894) acc1: 96.0938 (96.4481) acc5: 100.0000 (100.0000) time: 0.2396 data: 0.1449 max mem: 3531 +Test: [ 70/235] eta: 0:00:53 acc1_pos: 98.462 acc5_pos: 100.000 loss: 0.2097 (0.3515) acc1: 94.0104 (86.6197) acc5: 100.0000 (100.0000) acc1_neg: 16.667 acc5_neg: 100.000 time: 0.2349 data: 0.1427 max mem: 3531 +Test: [ 80/235] eta: 0:00:47 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4892 (0.4934) acc1: 17.4479 (78.0543) acc5: 100.0000 (100.0000) acc1_neg: 14.844 acc5_neg: 100.000 time: 0.2010 data: 0.1082 max mem: 3531 +Test: [ 90/235] eta: 0:00:42 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4927 (0.6038) acc1: 16.9271 (71.4286) acc5: 100.0000 (100.0000) acc1_neg: 16.146 acc5_neg: 100.000 time: 0.1725 data: 0.0793 max mem: 3531 +Test: [100/235] eta: 0:00:38 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4913 (0.6916) acc1: 17.7083 (66.1561) acc5: 100.0000 (100.0000) acc1_neg: 14.844 acc5_neg: 100.000 time: 0.1826 data: 0.0899 max mem: 3531 +Test: [110/235] eta: 0:00:33 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4753 (0.7627) acc1: 17.7083 (61.8056) acc5: 100.0000 (100.0000) acc1_neg: 19.271 acc5_neg: 100.000 time: 0.1783 data: 0.0839 max mem: 3531 +Test: [120/235] eta: 0:00:30 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.5362 (0.8465) acc1: 14.3229 (57.3175) acc5: 100.0000 (100.0000) acc1_neg: 6.771 acc5_neg: 100.000 time: 0.1688 data: 0.0759 max mem: 3531 +Test: [130/235] eta: 0:00:27 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.8018 (0.9195) acc1: 6.7708 (53.4729) acc5: 100.0000 (100.0000) acc1_neg: 7.552 acc5_neg: 100.000 time: 0.1945 data: 0.1028 max mem: 3531 +Test: [140/235] eta: 0:00:23 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.8243 (0.9842) acc1: 6.2500 (50.1330) acc5: 100.0000 (100.0000) acc1_neg: 4.688 acc5_neg: 100.000 time: 0.1871 data: 0.0940 max mem: 3531 +Test: [150/235] eta: 0:00:21 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.8139 (1.0383) acc1: 6.2500 (47.2234) acc5: 100.0000 (100.0000) acc1_neg: 6.510 acc5_neg: 100.000 time: 0.1705 data: 0.0780 max mem: 3531 +Test: [160/235] eta: 0:00:18 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.8030 (1.0857) acc1: 6.5104 (44.7415) acc5: 100.0000 (100.0000) acc1_neg: 7.552 acc5_neg: 100.000 time: 0.1958 data: 0.1044 max mem: 3531 +Test: [170/235] eta: 0:00:15 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.7655 (1.1005) acc1: 8.5938 (43.5612) acc5: 100.0000 (100.0000) acc1_neg: 21.615 acc5_neg: 100.000 time: 0.1901 data: 0.0981 max mem: 3531 +Test: [180/235] eta: 0:00:13 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.2900 (1.1103) acc1: 26.5625 (42.6364) acc5: 100.0000 (100.0000) acc1_neg: 26.823 acc5_neg: 100.000 time: 0.2030 data: 0.1116 max mem: 3531 +Test: [190/235] eta: 0:00:10 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.2607 (1.1185) acc1: 27.0833 (41.8235) acc5: 100.0000 (100.0000) acc1_neg: 25.521 acc5_neg: 100.000 time: 0.2080 data: 0.1166 max mem: 3531 +Test: [200/235] eta: 0:00:08 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.2786 (1.1271) acc1: 25.5208 (40.9930) acc5: 100.0000 (100.0000) acc1_neg: 23.958 acc5_neg: 100.000 time: 0.1577 data: 0.0663 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.2892 (1.1336) acc1: 24.7396 (40.3251) acc5: 100.0000 (100.0000) acc1_neg: 28.125 acc5_neg: 100.000 time: 0.1467 data: 0.0561 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.2782 (1.1400) acc1: 26.8229 (39.7365) acc5: 100.0000 (100.0000) acc1_neg: 26.302 acc5_neg: 100.000 time: 0.1472 data: 0.0569 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.2620 (1.1449) acc1: 28.1250 (39.2395) acc5: 100.0000 (100.0000) acc1_neg: 29.167 acc5_neg: 100.000 time: 0.1192 data: 0.0290 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.2687 (1.1482) acc1: 27.6042 (39.0814) acc5: 100.0000 (100.0000) acc1_neg: 30.435 acc5_neg: 100.000 time: 0.1144 data: 0.0243 max mem: 3532 +Test: Total time: 0:00:51 (0.2180 s / it) +* Overall Acc@1 39.081 Acc@5 100.000 + target == real Acc@1 96.427 Acc@5 100.000 + target == fake Acc@1 18.341 Acc@5 100.000 + loss 1.148 + PR-AUC 0.8720 ROC-AUC 0.7241 EER 33.89% (thr=0.1530) +Accuracy of the network on the 89879 test images: 39.1% +[rank0]:[W726 06:36:24.844135754 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 06:36:44.885000 1211178 torch/distributed/run.py:766] +W0726 06:36:44.885000 1211178 torch/distributed/run.py:766] ***************************************** +W0726 06:36:44.885000 1211178 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:36:44.885000 1211178 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:37:12.619579314 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 06:37:12.735321739 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:37:12.808582284 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:37:12.160812934 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:37:13.288292731 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:37:13.341212779 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:37:13.409046627 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W726 06:37:13.413198415 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n64_seed9_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n64_classseed9_resampled_n116-040-213_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:23:12 acc1_pos: 79.167 acc5_pos: 100.000 loss: 0.5255 (0.5255) acc1: 79.1667 (79.1667) acc5: 100.0000 (100.0000) time: 5.1781 data: 3.2588 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/269] eta: 0:02:28 acc1_pos: 93.229 acc5_pos: 100.000 loss: 0.4251 (0.4325) acc1: 86.9792 (87.2633) acc5: 100.0000 (100.0000) time: 0.5751 data: 0.3190 max mem: 3531 +Test: [ 20/269] eta: 0:01:32 acc1_pos: 92.188 acc5_pos: 100.000 loss: 0.4225 (0.4218) acc1: 89.8438 (88.2316) acc5: 100.0000 (100.0000) time: 0.1301 data: 0.0402 max mem: 3531 +Test: [ 30/269] eta: 0:01:09 acc1_pos: 90.625 acc5_pos: 100.000 loss: 0.3901 (0.4200) acc1: 90.6250 (88.2813) acc5: 100.0000 (100.0000) time: 0.1366 data: 0.0425 max mem: 3531 +Test: [ 40/269] eta: 0:01:08 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.5060 (0.4816) acc1: 75.5208 (80.9451) acc5: 100.0000 (100.0000) acc1_neg: 45.573 acc5_neg: 100.000 time: 0.2204 data: 0.1249 max mem: 3531 +Test: [ 50/269] eta: 0:01:01 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.7623 (0.5554) acc1: 45.5729 (72.8554) acc5: 100.0000 (100.0000) acc1_neg: 21.354 acc5_neg: 100.000 time: 0.2644 data: 0.1708 max mem: 3531 +Test: [ 60/269] eta: 0:00:55 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.9082 (0.6180) acc1: 34.6354 (65.5951) acc5: 100.0000 (100.0000) acc1_neg: 18.490 acc5_neg: 100.000 time: 0.1982 data: 0.1049 max mem: 3531 +Test: [ 70/269] eta: 0:00:52 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.9232 (0.6579) acc1: 28.9062 (60.8935) acc5: 100.0000 (100.0000) acc1_neg: 38.802 acc5_neg: 100.000 time: 0.2200 data: 0.1280 max mem: 3531 +Test: [ 80/269] eta: 0:00:49 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.8309 (0.6779) acc1: 35.9375 (58.1276) acc5: 100.0000 (100.0000) acc1_neg: 34.896 acc5_neg: 100.000 time: 0.2495 data: 0.1565 max mem: 3531 +Test: [ 90/269] eta: 0:00:45 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.8685 (0.7100) acc1: 30.9896 (54.3641) acc5: 100.0000 (100.0000) acc1_neg: 27.604 acc5_neg: 100.000 time: 0.2197 data: 0.1239 max mem: 3531 +Test: [100/269] eta: 0:00:42 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.9479 (0.7318) acc1: 27.0833 (51.8178) acc5: 100.0000 (100.0000) acc1_neg: 18.750 acc5_neg: 100.000 time: 0.2149 data: 0.1194 max mem: 3531 +Test: [110/269] eta: 0:00:39 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.9461 (0.7540) acc1: 26.3021 (49.2188) acc5: 100.0000 (100.0000) acc1_neg: 23.438 acc5_neg: 100.000 time: 0.2078 data: 0.1154 max mem: 3531 +Test: [120/269] eta: 0:00:35 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.9311 (0.7670) acc1: 26.3021 (48.0157) acc5: 100.0000 (100.0000) acc1_neg: 30.469 acc5_neg: 100.000 time: 0.1799 data: 0.0865 max mem: 3531 +Test: [130/269] eta: 0:00:32 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.9597 (0.7861) acc1: 26.3021 (46.1852) acc5: 100.0000 (100.0000) acc1_neg: 20.052 acc5_neg: 100.000 time: 0.1820 data: 0.0885 max mem: 3531 +Test: [140/269] eta: 0:00:30 acc1_pos: 87.931 acc5_pos: 100.000 loss: 1.0101 (0.8004) acc1: 24.2188 (44.8194) acc5: 100.0000 (100.0000) acc1_neg: 25.781 acc5_neg: 100.000 time: 0.2008 data: 0.1085 max mem: 3531 +Test: [150/269] eta: 0:00:27 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.9142 (0.8029) acc1: 29.9479 (44.6209) acc5: 100.0000 (100.0000) acc1_neg: 44.531 acc5_neg: 100.000 time: 0.2073 data: 0.1153 max mem: 3531 +Test: [160/269] eta: 0:00:25 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.8292 (0.8040) acc1: 38.5417 (44.3921) acc5: 100.0000 (100.0000) acc1_neg: 38.542 acc5_neg: 100.000 time: 0.2008 data: 0.1087 max mem: 3531 +Test: [170/269] eta: 0:00:22 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.8621 (0.8104) acc1: 30.7292 (43.5292) acc5: 100.0000 (100.0000) acc1_neg: 30.729 acc5_neg: 100.000 time: 0.2081 data: 0.1120 max mem: 3531 +Test: [180/269] eta: 0:00:20 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.8461 (0.8097) acc1: 33.5938 (43.4320) acc5: 100.0000 (100.0000) acc1_neg: 35.156 acc5_neg: 100.000 time: 0.2058 data: 0.1099 max mem: 3531 +Test: [190/269] eta: 0:00:17 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.8417 (0.8152) acc1: 35.4167 (42.8447) acc5: 100.0000 (100.0000) acc1_neg: 33.594 acc5_neg: 100.000 time: 0.1897 data: 0.0977 max mem: 3531 +Test: [200/269] eta: 0:00:15 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.9718 (0.8273) acc1: 27.0833 (41.9491) acc5: 100.0000 (100.0000) acc1_neg: 37.500 acc5_neg: 100.000 time: 0.1854 data: 0.0938 max mem: 3531 +Test: [210/269] eta: 0:00:13 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.9166 (0.8279) acc1: 28.9062 (41.8111) acc5: 100.0000 (100.0000) acc1_neg: 38.542 acc5_neg: 100.000 time: 0.1958 data: 0.1040 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.8096 (0.8269) acc1: 39.8438 (41.8635) acc5: 100.0000 (100.0000) acc1_neg: 66.927 acc5_neg: 100.000 time: 0.1995 data: 0.1058 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.7648 (0.8228) acc1: 46.3542 (42.2461) acc5: 100.0000 (100.0000) acc1_neg: 47.917 acc5_neg: 100.000 time: 0.1918 data: 0.0988 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.8423 (0.8285) acc1: 36.7188 (41.5878) acc5: 100.0000 (100.0000) acc1_neg: 30.208 acc5_neg: 100.000 time: 0.1686 data: 0.0781 max mem: 3532 +Test: [250/269] eta: 0:00:04 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.9510 (0.8367) acc1: 24.2188 (41.0348) acc5: 100.0000 (100.0000) acc1_neg: 13.021 acc5_neg: 100.000 time: 0.1490 data: 0.0581 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 87.931 acc5_pos: 100.000 loss: 1.1072 (0.8470) acc1: 14.3229 (40.1670) acc5: 100.0000 (100.0000) acc1_neg: 30.208 acc5_neg: 100.000 time: 0.1397 data: 0.0488 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.9659 (0.8488) acc1: 21.3542 (39.8565) acc5: 100.0000 (100.0000) acc1_neg: 26.247 acc5_neg: 100.000 time: 0.1307 data: 0.0362 max mem: 3532 +Test: Total time: 0:00:56 (0.2093 s / it) +* Overall Acc@1 39.857 Acc@5 100.000 + target == real Acc@1 88.093 Acc@5 100.000 + target == fake Acc@1 33.274 Acc@5 100.000 + loss 0.849 + PR-AUC 0.9382 ROC-AUC 0.6939 EER 35.98% (thr=0.3582) +Accuracy of the network on the 103293 test images: 39.9% +[rank0]:[W726 06:38:20.355356444 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T06:38:27Z] eval_done tag=fixed500k_fake_n64_seed9_resampled +eval_queue_done_utc=2026-07-26T06:38:27Z diff --git a/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_1958_seed5_9/eval_queue_n116-040-213.pid b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_1958_seed5_9/eval_queue_n116-040-213.pid new file mode 100644 index 0000000000000000000000000000000000000000..31ace642d35624f76b5040dbf39421a0bdde8b36 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_1958_seed5_9/eval_queue_n116-040-213.pid @@ -0,0 +1 @@ +1171813 diff --git a/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_1958_seed5_9/eval_queue_n116-041-023.log b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_1958_seed5_9/eval_queue_n116-041-023.log new file mode 100644 index 0000000000000000000000000000000000000000..d72b49bca8aa092a179d8fcfb2c4287fe5043ad3 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_1958_seed5_9/eval_queue_n116-041-023.log @@ -0,0 +1,10270 @@ +eval_queue_started_utc=2026-07-26T02:59:38Z +hostname=n116-041-023 +train_tag=20260725_1958_seed5_9 +eval_tag=20260725_1958_seed5_9 +real_level=23 +fake_level=45 +eval_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9 +batch_per_gpu=256 +seeds=5 6 7 8 9 +[2026-07-26T02:59:38Z] eval_start tag=fixed500k_real_n23_seed5_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed5_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T02:59:39Z +hostname=n116-041-023 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed5_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n23_seed5_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 02:59:52.300000 1169687 torch/distributed/run.py:766] +W0726 02:59:52.300000 1169687 torch/distributed/run.py:766] ***************************************** +W0726 02:59:52.300000 1169687 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 02:59:52.300000 1169687 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:00:23.747209932 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:00:23.012201799 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:00:23.226012690 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:00:24.504181974 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:00:24.758595483 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:00:24.988160937 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:00:24.020413401 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:00:24.055072323 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n23_seed5_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed5_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:22:22 acc1_pos: 95.573 acc5_pos: 100.000 loss: 0.3712 (0.3712) acc1: 95.5729 (95.5729) acc5: 100.0000 (100.0000) time: 5.7370 data: 3.6620 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/234] eta: 0:02:15 acc1_pos: 86.979 acc5_pos: 100.000 loss: 0.4436 (0.4536) acc1: 86.9792 (83.3333) acc5: 100.0000 (100.0000) time: 0.6037 data: 0.3333 max mem: 3531 +Test: [ 20/234] eta: 0:01:16 acc1_pos: 91.406 acc5_pos: 100.000 loss: 0.4435 (0.4600) acc1: 83.0729 (81.5972) acc5: 100.0000 (100.0000) time: 0.0902 data: 0.0004 max mem: 3531 +Test: [ 30/234] eta: 0:00:58 acc1_pos: 84.896 acc5_pos: 100.000 loss: 0.3042 (0.4047) acc1: 89.5833 (84.4842) acc5: 100.0000 (100.0000) time: 0.1113 data: 0.0209 max mem: 3531 +Test: [ 40/234] eta: 0:00:53 acc1_pos: 97.396 acc5_pos: 100.000 loss: 0.3041 (0.3884) acc1: 89.5833 (85.3341) acc5: 100.0000 (100.0000) time: 0.1865 data: 0.0945 max mem: 3531 +Test: [ 50/234] eta: 0:00:49 acc1_pos: 97.917 acc5_pos: 100.000 loss: 0.1898 (0.3482) acc1: 96.6146 (87.6685) acc5: 100.0000 (100.0000) time: 0.2366 data: 0.1442 max mem: 3531 +Test: [ 60/234] eta: 0:00:44 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.1323 (0.3071) acc1: 97.9167 (89.6644) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2092 data: 0.1154 max mem: 3531 +Test: [ 70/234] eta: 0:00:39 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.0945 (0.3150) acc1: 100.0000 (88.8131) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.1839 data: 0.0905 max mem: 3531 +Test: [ 80/234] eta: 0:00:36 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.4675 (0.3202) acc1: 77.0833 (88.2684) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0989 max mem: 3531 +Test: [ 90/234] eta: 0:00:33 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.1282 (0.2978) acc1: 98.4375 (89.3601) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.2087 data: 0.1167 max mem: 3531 +Test: [100/234] eta: 0:00:31 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.1729 (0.2992) acc1: 91.6667 (89.0573) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2112 data: 0.1179 max mem: 3531 +Test: [110/234] eta: 0:00:28 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.3080 (0.2998) acc1: 85.9375 (88.9335) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.2170 data: 0.1245 max mem: 3531 +Test: [120/234] eta: 0:00:25 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.3294 (0.3106) acc1: 85.9375 (88.2425) acc5: 100.0000 (100.0000) acc1_neg: 72.396 acc5_neg: 100.000 time: 0.2051 data: 0.1133 max mem: 3531 +Test: [130/234] eta: 0:00:23 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.4943 (0.3275) acc1: 75.5208 (87.0925) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.1978 data: 0.1053 max mem: 3531 +Test: [140/234] eta: 0:00:21 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.4994 (0.3528) acc1: 72.6562 (85.6438) acc5: 100.0000 (100.0000) acc1_neg: 39.583 acc5_neg: 100.000 time: 0.1987 data: 0.1036 max mem: 3531 +Test: [150/234] eta: 0:00:18 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.3659 (0.3570) acc1: 82.8125 (85.2649) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.1773 data: 0.0833 max mem: 3531 +Test: [160/234] eta: 0:00:16 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.3367 (0.3661) acc1: 82.8125 (84.5287) acc5: 100.0000 (100.0000) acc1_neg: 57.292 acc5_neg: 100.000 time: 0.1812 data: 0.0903 max mem: 3531 +Test: [170/234] eta: 0:00:13 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.6586 (0.3830) acc1: 60.6771 (83.2222) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.1975 data: 0.1048 max mem: 3531 +Test: [180/234] eta: 0:00:11 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.6111 (0.3895) acc1: 62.2396 (82.6988) acc5: 100.0000 (100.0000) acc1_neg: 51.562 acc5_neg: 100.000 time: 0.2183 data: 0.1230 max mem: 3531 +Test: [190/234] eta: 0:00:09 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.7186 (0.4077) acc1: 62.2396 (81.5404) acc5: 100.0000 (100.0000) acc1_neg: 62.240 acc5_neg: 100.000 time: 0.1958 data: 0.1011 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.7220 (0.4048) acc1: 62.2396 (81.7462) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1899 data: 0.0949 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.1260 (0.3915) acc1: 99.7396 (82.5940) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2193 data: 0.1249 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.1201 (0.3797) acc1: 99.4792 (83.3098) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1933 data: 0.1016 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.2800 (0.3888) acc1: 75.0000 (82.6829) acc5: 100.0000 (100.0000) acc1_neg: 68.229 acc5_neg: 100.000 time: 0.1510 data: 0.0605 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.5691 (0.3913) acc1: 69.7917 (82.5229) acc5: 100.0000 (100.0000) acc1_neg: 67.315 acc5_neg: 100.000 time: 0.1189 data: 0.0282 max mem: 3532 +Test: Total time: 0:00:49 (0.2095 s / it) +* Overall Acc@1 82.523 Acc@5 100.000 + target == real Acc@1 87.729 Acc@5 100.000 + target == fake Acc@1 81.056 Acc@5 100.000 + loss 0.391 + PR-AUC 0.9767 ROC-AUC 0.9180 EER 16.06% (thr=0.4495) +Accuracy of the network on the 89729 test images: 82.5% +[rank0]:[W726 03:01:24.821946905 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:01:44.570000 1170233 torch/distributed/run.py:766] +W0726 03:01:44.570000 1170233 torch/distributed/run.py:766] ***************************************** +W0726 03:01:44.570000 1170233 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:01:44.570000 1170233 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:02:12.299243768 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:02:12.356269292 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:02:13.892248558 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:02:13.966796028 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:02:13.999985680 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:02:13.007864422 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:02:13.046227845 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:02:13.135238661 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n23_seed5_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed5_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:51:30 acc1_pos: 73.698 acc5_pos: 100.000 loss: 0.5223 (0.5223) acc1: 73.6979 (73.6979) acc5: 100.0000 (100.0000) time: 7.3275 data: 5.9180 max mem: 3531 +Test: [ 10/2387] eta: 0:29:38 acc1_pos: 42.708 acc5_pos: 100.000 loss: 0.7132 (0.7385) acc1: 54.9479 (51.4441) acc5: 100.0000 (100.0000) time: 0.7483 data: 0.5384 max mem: 3531 +Test: [ 20/2387] eta: 0:17:17 acc1_pos: 55.469 acc5_pos: 100.000 loss: 0.7177 (0.7433) acc1: 53.9062 (52.9018) acc5: 100.0000 (100.0000) time: 0.0938 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:15:08 acc1_pos: 22.135 acc5_pos: 100.000 loss: 0.7438 (0.7613) acc1: 49.2188 (50.6048) acc5: 100.0000 (100.0000) time: 0.1857 data: 0.0921 max mem: 3531 +Test: [ 40/2387] eta: 0:13:38 acc1_pos: 51.823 acc5_pos: 100.000 loss: 0.8185 (0.7767) acc1: 38.5417 (47.8277) acc5: 100.0000 (100.0000) time: 0.2549 data: 0.1591 max mem: 3531 +Test: [ 50/2387] eta: 0:12:42 acc1_pos: 51.302 acc5_pos: 100.000 loss: 0.7712 (0.7594) acc1: 46.3542 (49.7141) acc5: 100.0000 (100.0000) time: 0.2346 data: 0.1378 max mem: 3531 +Test: [ 60/2387] eta: 0:12:00 acc1_pos: 59.635 acc5_pos: 100.000 loss: 0.7084 (0.7577) acc1: 51.3021 (50.1750) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1369 max mem: 3531 +Test: [ 70/2387] eta: 0:11:33 acc1_pos: 38.542 acc5_pos: 100.000 loss: 0.7737 (0.7682) acc1: 47.1354 (49.3104) acc5: 100.0000 (100.0000) time: 0.2300 data: 0.1359 max mem: 3531 +Test: [ 80/2387] eta: 0:11:12 acc1_pos: 89.062 acc5_pos: 100.000 loss: 0.7193 (0.7443) acc1: 54.4271 (52.1573) acc5: 100.0000 (100.0000) time: 0.2354 data: 0.1415 max mem: 3531 +Test: [ 90/2387] eta: 0:10:58 acc1_pos: 77.604 acc5_pos: 100.000 loss: 0.5217 (0.7162) acc1: 77.8646 (55.5002) acc5: 100.0000 (100.0000) time: 0.2430 data: 0.1501 max mem: 3531 +Test: [ 100/2387] eta: 0:10:49 acc1_pos: 88.542 acc5_pos: 100.000 loss: 0.5055 (0.6990) acc1: 82.2917 (57.7584) acc5: 100.0000 (100.0000) time: 0.2543 data: 0.1587 max mem: 3531 +Test: [ 110/2387] eta: 0:10:37 acc1_pos: 71.615 acc5_pos: 100.000 loss: 0.4879 (0.6793) acc1: 82.2917 (60.0601) acc5: 100.0000 (100.0000) time: 0.2499 data: 0.1530 max mem: 3531 +Test: [ 120/2387] eta: 0:10:29 acc1_pos: 88.802 acc5_pos: 100.000 loss: 0.4793 (0.6650) acc1: 79.6875 (61.6843) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1456 max mem: 3531 +Test: [ 130/2387] eta: 0:10:18 acc1_pos: 72.917 acc5_pos: 100.000 loss: 0.4957 (0.6528) acc1: 78.9062 (63.1421) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1403 max mem: 3531 +Test: [ 140/2387] eta: 0:10:19 acc1_pos: 86.719 acc5_pos: 100.000 loss: 0.4957 (0.6424) acc1: 79.4271 (64.3414) acc5: 100.0000 (100.0000) time: 0.2649 data: 0.1689 max mem: 3531 +Test: [ 150/2387] eta: 0:10:11 acc1_pos: 86.979 acc5_pos: 100.000 loss: 0.4602 (0.6294) acc1: 82.0312 (65.8595) acc5: 100.0000 (100.0000) time: 0.2700 data: 0.1746 max mem: 3531 +Test: [ 160/2387] eta: 0:09:56 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.4165 (0.6129) acc1: 91.1458 (67.4172) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2112 data: 0.1154 max mem: 3531 +Test: [ 170/2387] eta: 0:09:45 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.2012 (0.5883) acc1: 95.5729 (69.1033) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1948 data: 0.0963 max mem: 3531 +Test: [ 180/2387] eta: 0:09:32 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1991 (0.5667) acc1: 95.8333 (70.6074) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1935 data: 0.0968 max mem: 3531 +Test: [ 190/2387] eta: 0:09:25 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1929 (0.5472) acc1: 95.8333 (71.9391) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1998 data: 0.1048 max mem: 3531 +Test: [ 200/2387] eta: 0:09:15 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1919 (0.5295) acc1: 96.0938 (73.1460) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2049 data: 0.1081 max mem: 3531 +Test: [ 210/2387] eta: 0:09:01 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1914 (0.5137) acc1: 96.8750 (74.2619) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1649 data: 0.0690 max mem: 3532 +Test: [ 220/2387] eta: 0:08:50 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1914 (0.4993) acc1: 96.8750 (75.2498) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1520 data: 0.0546 max mem: 3532 +Test: [ 230/2387] eta: 0:08:42 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1910 (0.4860) acc1: 96.6146 (76.1702) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1749 data: 0.0771 max mem: 3532 +Test: [ 240/2387] eta: 0:08:33 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1986 (0.4777) acc1: 94.7917 (76.8402) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1778 data: 0.0829 max mem: 3532 +Test: [ 250/2387] eta: 0:08:26 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1949 (0.4640) acc1: 94.0104 (77.7224) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1752 data: 0.0824 max mem: 3532 +Test: [ 260/2387] eta: 0:08:19 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1067 (0.4502) acc1: 100.0000 (78.5680) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1798 data: 0.0877 max mem: 3532 +Test: [ 270/2387] eta: 0:08:12 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.0964 (0.4371) acc1: 100.0000 (79.3541) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1761 data: 0.0824 max mem: 3532 +Test: [ 280/2387] eta: 0:08:09 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.0969 (0.4252) acc1: 100.0000 (80.0879) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1999 data: 0.1070 max mem: 3532 +Test: [ 290/2387] eta: 0:08:05 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1052 (0.4144) acc1: 100.0000 (80.7659) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2159 data: 0.1246 max mem: 3532 +Test: [ 300/2387] eta: 0:08:01 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1144 (0.4047) acc1: 100.0000 (81.4040) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2099 data: 0.1174 max mem: 3532 +Test: [ 310/2387] eta: 0:07:57 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1292 (0.3968) acc1: 100.0000 (81.9911) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2096 data: 0.1159 max mem: 3532 +Test: [ 320/2387] eta: 0:07:57 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1818 (0.3902) acc1: 99.7396 (82.5391) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2312 data: 0.1368 max mem: 3532 +Test: [ 330/2387] eta: 0:07:52 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1822 (0.3839) acc1: 99.4792 (83.0564) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2282 data: 0.1347 max mem: 3532 +Test: [ 340/2387] eta: 0:07:51 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1787 (0.3779) acc1: 100.0000 (83.5495) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2186 data: 0.1250 max mem: 3532 +Test: [ 350/2387] eta: 0:07:47 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1761 (0.3706) acc1: 100.0000 (84.0174) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2199 data: 0.1260 max mem: 3532 +Test: [ 360/2387] eta: 0:07:43 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1304 (0.3650) acc1: 99.7396 (84.3923) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2049 data: 0.1132 max mem: 3532 +Test: [ 370/2387] eta: 0:07:40 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1751 (0.3600) acc1: 97.3958 (84.7442) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2102 data: 0.1178 max mem: 3532 +Test: [ 380/2387] eta: 0:07:37 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1760 (0.3551) acc1: 97.3958 (85.0872) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2123 data: 0.1199 max mem: 3532 +Test: [ 390/2387] eta: 0:07:33 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1758 (0.3506) acc1: 98.1771 (85.4180) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2019 data: 0.1101 max mem: 3532 +Test: [ 400/2387] eta: 0:07:29 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1758 (0.3462) acc1: 97.6562 (85.7193) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1986 data: 0.1070 max mem: 3532 +Test: [ 410/2387] eta: 0:07:26 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1739 (0.3420) acc1: 97.9167 (86.0211) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2088 data: 0.1175 max mem: 3532 +Test: [ 420/2387] eta: 0:07:25 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1710 (0.3380) acc1: 97.9167 (86.2963) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2252 data: 0.1337 max mem: 3532 +Test: [ 430/2387] eta: 0:07:21 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1762 (0.3355) acc1: 97.3958 (86.5091) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2132 data: 0.1197 max mem: 3532 +Test: [ 440/2387] eta: 0:07:18 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1855 (0.3317) acc1: 95.3125 (86.7601) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2042 data: 0.1107 max mem: 3533 +Test: [ 450/2387] eta: 0:07:15 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1089 (0.3268) acc1: 100.0000 (87.0537) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2126 data: 0.1214 max mem: 3533 +Test: [ 460/2387] eta: 0:07:12 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1085 (0.3221) acc1: 100.0000 (87.3339) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2092 data: 0.1144 max mem: 3533 +Test: [ 470/2387] eta: 0:07:10 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1085 (0.3175) acc1: 100.0000 (87.6028) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2181 data: 0.1202 max mem: 3533 +Test: [ 480/2387] eta: 0:07:08 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1086 (0.3132) acc1: 100.0000 (87.8606) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2249 data: 0.1280 max mem: 3533 +Test: [ 490/2387] eta: 0:07:04 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1092 (0.3090) acc1: 100.0000 (88.1078) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2047 data: 0.1085 max mem: 3533 +Test: [ 500/2387] eta: 0:07:00 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1093 (0.3051) acc1: 100.0000 (88.3452) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1858 data: 0.0917 max mem: 3533 +Test: [ 510/2387] eta: 0:06:58 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1094 (0.3016) acc1: 100.0000 (88.5733) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1990 data: 0.1075 max mem: 3533 +Test: [ 520/2387] eta: 0:06:54 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1415 (0.2988) acc1: 100.0000 (88.7926) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1989 data: 0.1072 max mem: 3533 +Test: [ 530/2387] eta: 0:06:51 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1410 (0.2958) acc1: 100.0000 (89.0037) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1951 data: 0.1013 max mem: 3533 +Test: [ 540/2387] eta: 0:06:48 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1394 (0.2929) acc1: 100.0000 (89.2069) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1962 data: 0.1030 max mem: 3533 +Test: [ 550/2387] eta: 0:06:45 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1392 (0.2901) acc1: 100.0000 (89.4023) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2003 data: 0.1094 max mem: 3533 +Test: [ 560/2387] eta: 0:06:43 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1383 (0.2875) acc1: 100.0000 (89.5912) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2065 data: 0.1158 max mem: 3533 +Test: [ 570/2387] eta: 0:06:40 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1371 (0.2848) acc1: 100.0000 (89.7735) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2046 data: 0.1122 max mem: 3533 +Test: [ 580/2387] eta: 0:06:37 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1374 (0.2824) acc1: 100.0000 (89.9495) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1939 data: 0.1010 max mem: 3533 +Test: [ 590/2387] eta: 0:06:33 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1406 (0.2800) acc1: 100.0000 (90.1196) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1797 data: 0.0878 max mem: 3533 +Test: [ 600/2387] eta: 0:06:29 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1441 (0.2781) acc1: 100.0000 (90.2836) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1712 data: 0.0794 max mem: 3533 +Test: [ 610/2387] eta: 0:06:25 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1356 (0.2756) acc1: 100.0000 (90.4358) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1626 data: 0.0711 max mem: 3533 +Test: [ 620/2387] eta: 0:06:22 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1287 (0.2733) acc1: 99.7396 (90.5835) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1617 data: 0.0701 max mem: 3533 +Test: [ 630/2387] eta: 0:06:19 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1306 (0.2711) acc1: 99.4792 (90.7241) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1720 data: 0.0790 max mem: 3533 +Test: [ 640/2387] eta: 0:06:16 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1306 (0.2689) acc1: 99.4792 (90.8578) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1816 data: 0.0892 max mem: 3533 +Test: [ 650/2387] eta: 0:06:12 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1295 (0.2668) acc1: 99.4792 (90.9910) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1820 data: 0.0911 max mem: 3533 +Test: [ 660/2387] eta: 0:06:10 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1311 (0.2647) acc1: 99.7396 (91.1222) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1847 data: 0.0927 max mem: 3533 +Test: [ 670/2387] eta: 0:06:07 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1311 (0.2627) acc1: 99.7396 (91.2483) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1831 data: 0.0916 max mem: 3534 +Test: [ 680/2387] eta: 0:06:03 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1324 (0.2612) acc1: 100.0000 (91.3749) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1751 data: 0.0834 max mem: 3534 +Test: [ 690/2387] eta: 0:06:00 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1737 (0.2621) acc1: 99.4792 (91.3177) acc5: 100.0000 (100.0000) acc1_neg: 80.729 acc5_neg: 100.000 time: 0.1746 data: 0.0807 max mem: 3534 +Test: [ 700/2387] eta: 0:05:57 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.3805 (0.2638) acc1: 83.3333 (91.2019) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.1764 data: 0.0820 max mem: 3534 +Test: [ 710/2387] eta: 0:05:54 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.3805 (0.2655) acc1: 83.5938 (91.0913) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.1706 data: 0.0781 max mem: 3534 +Test: [ 720/2387] eta: 0:05:51 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.3723 (0.2670) acc1: 83.3333 (90.9887) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.1631 data: 0.0692 max mem: 3534 +Test: [ 730/2387] eta: 0:05:48 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.3839 (0.2687) acc1: 82.2917 (90.8694) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.1652 data: 0.0718 max mem: 3534 +Test: [ 740/2387] eta: 0:05:45 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.3893 (0.2702) acc1: 83.5938 (90.7642) acc5: 100.0000 (100.0000) acc1_neg: 80.729 acc5_neg: 100.000 time: 0.1838 data: 0.0906 max mem: 3534 +Test: [ 750/2387] eta: 0:05:43 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.3758 (0.2716) acc1: 83.5938 (90.6645) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.2004 data: 0.1072 max mem: 3534 +Test: [ 760/2387] eta: 0:05:41 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.3684 (0.2727) acc1: 83.0729 (90.5884) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.2097 data: 0.1174 max mem: 3534 +Test: [ 770/2387] eta: 0:05:40 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.3093 (0.2727) acc1: 88.8021 (90.6115) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2422 data: 0.1493 max mem: 3534 +Test: [ 780/2387] eta: 0:05:37 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.2408 (0.2719) acc1: 93.2292 (90.6700) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2177 data: 0.1262 max mem: 3534 +Test: [ 790/2387] eta: 0:05:34 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1980 (0.2713) acc1: 95.8333 (90.6928) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.1715 data: 0.0796 max mem: 3534 +Test: [ 800/2387] eta: 0:05:33 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1796 (0.2701) acc1: 98.1771 (90.7928) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1957 data: 0.1024 max mem: 3534 +Test: [ 810/2387] eta: 0:05:30 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1815 (0.2693) acc1: 98.1771 (90.8617) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2145 data: 0.1194 max mem: 3534 +Test: [ 820/2387] eta: 0:05:29 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.2261 (0.2692) acc1: 96.0938 (90.9032) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.2150 data: 0.1212 max mem: 3534 +Test: [ 830/2387] eta: 0:05:26 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.3074 (0.2705) acc1: 90.3646 (90.8381) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.2114 data: 0.1197 max mem: 3534 +Test: [ 840/2387] eta: 0:05:24 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1659 (0.2688) acc1: 94.2708 (90.9384) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2099 data: 0.1167 max mem: 3534 +Test: [ 850/2387] eta: 0:05:22 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1143 (0.2669) acc1: 99.7396 (91.0427) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2150 data: 0.1217 max mem: 3534 +Test: [ 860/2387] eta: 0:05:20 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1066 (0.2651) acc1: 99.7396 (91.1446) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2178 data: 0.1221 max mem: 3534 +Test: [ 870/2387] eta: 0:05:19 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1066 (0.2633) acc1: 99.7396 (91.2433) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2255 data: 0.1278 max mem: 3534 +Test: [ 880/2387] eta: 0:05:18 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1146 (0.2617) acc1: 99.7396 (91.3386) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2556 data: 0.1613 max mem: 3534 +Test: [ 890/2387] eta: 0:05:16 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1308 (0.2606) acc1: 99.4792 (91.4250) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2486 data: 0.1570 max mem: 3534 +Test: [ 900/2387] eta: 0:05:13 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1834 (0.2616) acc1: 98.4375 (91.3802) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.1947 data: 0.1019 max mem: 3535 +Test: [ 910/2387] eta: 0:05:11 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.3932 (0.2630) acc1: 84.3750 (91.3116) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1903 data: 0.0962 max mem: 3535 +Test: [ 920/2387] eta: 0:05:08 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.3924 (0.2644) acc1: 85.4167 (91.2530) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.1981 data: 0.1059 max mem: 3535 +Test: [ 930/2387] eta: 0:05:06 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.3898 (0.2658) acc1: 85.4167 (91.1802) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.1833 data: 0.0918 max mem: 3535 +Test: [ 940/2387] eta: 0:05:03 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.3986 (0.2672) acc1: 84.6354 (91.1157) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.1824 data: 0.0876 max mem: 3535 +Test: [ 950/2387] eta: 0:05:01 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.3921 (0.2684) acc1: 85.1562 (91.0552) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1848 data: 0.0884 max mem: 3535 +Test: [ 960/2387] eta: 0:04:58 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.3831 (0.2696) acc1: 86.1979 (91.0079) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1639 data: 0.0702 max mem: 3535 +Test: [ 970/2387] eta: 0:04:55 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.3837 (0.2709) acc1: 86.4583 (90.9525) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1555 data: 0.0635 max mem: 3535 +Test: [ 980/2387] eta: 0:04:53 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.3713 (0.2708) acc1: 87.7604 (90.9656) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1790 data: 0.0850 max mem: 3535 +Test: [ 990/2387] eta: 0:04:50 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.2110 (0.2704) acc1: 92.9688 (90.9871) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1856 data: 0.0924 max mem: 3535 +Test: [1000/2387] eta: 0:04:48 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.2068 (0.2700) acc1: 92.7083 (91.0054) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1892 data: 0.0990 max mem: 3535 +Test: [1010/2387] eta: 0:04:46 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.2503 (0.2704) acc1: 91.9271 (90.9882) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.2047 data: 0.1145 max mem: 3535 +Test: [1020/2387] eta: 0:04:44 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.2960 (0.2707) acc1: 90.3646 (90.9670) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2044 data: 0.1125 max mem: 3535 +Test: [1030/2387] eta: 0:04:41 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.2107 (0.2699) acc1: 93.7500 (91.0158) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1950 data: 0.1018 max mem: 3535 +Test: [1040/2387] eta: 0:04:39 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1535 (0.2689) acc1: 97.3958 (91.0733) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1849 data: 0.0919 max mem: 3535 +Test: [1050/2387] eta: 0:04:37 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1535 (0.2679) acc1: 97.9167 (91.1359) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1790 data: 0.0865 max mem: 3535 +Test: [1060/2387] eta: 0:04:34 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1830 (0.2672) acc1: 96.6146 (91.1807) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1759 data: 0.0844 max mem: 3535 +Test: [1070/2387] eta: 0:04:32 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.2118 (0.2672) acc1: 94.7917 (91.1903) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.1761 data: 0.0854 max mem: 3535 +Test: [1080/2387] eta: 0:04:29 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.3180 (0.2682) acc1: 86.9792 (91.1340) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.1698 data: 0.0796 max mem: 3535 +Test: [1090/2387] eta: 0:04:27 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.3141 (0.2682) acc1: 86.4583 (91.1301) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.1648 data: 0.0735 max mem: 3535 +Test: [1100/2387] eta: 0:04:24 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.2524 (0.2680) acc1: 91.1458 (91.1250) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1740 data: 0.0818 max mem: 3535 +Test: [1110/2387] eta: 0:04:22 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.2497 (0.2681) acc1: 89.8438 (91.0919) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1789 data: 0.0858 max mem: 3535 +Test: [1120/2387] eta: 0:04:19 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.2497 (0.2679) acc1: 89.8438 (91.0922) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1824 data: 0.0880 max mem: 3536 +Test: [1130/2387] eta: 0:04:17 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.2411 (0.2679) acc1: 91.1458 (91.0952) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1762 data: 0.0809 max mem: 3536 +Test: [1140/2387] eta: 0:04:15 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.3109 (0.2688) acc1: 86.7188 (91.0322) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1686 data: 0.0747 max mem: 3536 +Test: [1150/2387] eta: 0:04:12 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.3635 (0.2683) acc1: 85.4167 (91.0513) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1652 data: 0.0729 max mem: 3536 +Test: [1160/2387] eta: 0:04:10 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.0864 (0.2667) acc1: 100.0000 (91.1283) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1617 data: 0.0704 max mem: 3536 +Test: [1170/2387] eta: 0:04:07 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.0806 (0.2651) acc1: 100.0000 (91.2041) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1619 data: 0.0707 max mem: 3536 +Test: [1180/2387] eta: 0:04:05 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.0806 (0.2635) acc1: 100.0000 (91.2781) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1624 data: 0.0704 max mem: 3536 +Test: [1190/2387] eta: 0:04:02 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.0869 (0.2621) acc1: 100.0000 (91.3514) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1636 data: 0.0711 max mem: 3536 +Test: [1200/2387] eta: 0:04:00 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.0953 (0.2608) acc1: 100.0000 (91.4229) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1659 data: 0.0726 max mem: 3536 +Test: [1210/2387] eta: 0:03:58 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1008 (0.2596) acc1: 100.0000 (91.4929) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1704 data: 0.0769 max mem: 3536 +Test: [1220/2387] eta: 0:03:55 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1074 (0.2583) acc1: 100.0000 (91.5575) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1742 data: 0.0820 max mem: 3536 +Test: [1230/2387] eta: 0:03:53 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.0921 (0.2568) acc1: 100.0000 (91.6248) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1776 data: 0.0847 max mem: 3536 +Test: [1240/2387] eta: 0:03:51 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.0809 (0.2554) acc1: 100.0000 (91.6912) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1760 data: 0.0825 max mem: 3536 +Test: [1250/2387] eta: 0:03:48 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.0832 (0.2541) acc1: 100.0000 (91.7570) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1672 data: 0.0755 max mem: 3536 +Test: [1260/2387] eta: 0:03:46 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.0955 (0.2529) acc1: 100.0000 (91.8203) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1702 data: 0.0766 max mem: 3536 +Test: [1270/2387] eta: 0:03:44 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1068 (0.2519) acc1: 99.7396 (91.8822) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1779 data: 0.0843 max mem: 3536 +Test: [1280/2387] eta: 0:03:42 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1400 (0.2512) acc1: 99.7396 (91.9372) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1789 data: 0.0866 max mem: 3536 +Test: [1290/2387] eta: 0:03:40 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1692 (0.2507) acc1: 98.4375 (91.9763) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1822 data: 0.0878 max mem: 3536 +Test: [1300/2387] eta: 0:03:37 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1742 (0.2502) acc1: 98.1771 (92.0184) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1849 data: 0.0915 max mem: 3536 +Test: [1310/2387] eta: 0:03:35 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1686 (0.2497) acc1: 97.9167 (92.0582) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1915 data: 0.0978 max mem: 3536 +Test: [1320/2387] eta: 0:03:33 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1970 (0.2493) acc1: 98.1771 (92.1017) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1836 data: 0.0886 max mem: 3536 +Test: [1330/2387] eta: 0:03:31 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.2201 (0.2494) acc1: 96.0938 (92.1114) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1864 data: 0.0923 max mem: 3536 +Test: [1340/2387] eta: 0:03:29 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.2890 (0.2503) acc1: 91.1458 (92.0521) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1964 data: 0.1031 max mem: 3536 +Test: [1350/2387] eta: 0:03:27 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.0831 (0.2490) acc1: 100.0000 (92.1110) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1851 data: 0.0937 max mem: 3537 +Test: [1360/2387] eta: 0:03:25 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.0665 (0.2476) acc1: 100.0000 (92.1689) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1884 data: 0.0973 max mem: 3537 +Test: [1370/2387] eta: 0:03:23 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.0628 (0.2463) acc1: 100.0000 (92.2261) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1911 data: 0.0998 max mem: 3537 +Test: [1380/2387] eta: 0:03:21 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.0629 (0.2450) acc1: 100.0000 (92.2824) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1779 data: 0.0870 max mem: 3537 +Test: [1390/2387] eta: 0:03:18 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.0665 (0.2438) acc1: 100.0000 (92.3378) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1725 data: 0.0786 max mem: 3537 +Test: [1400/2387] eta: 0:03:16 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.0775 (0.2426) acc1: 100.0000 (92.3925) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1771 data: 0.0822 max mem: 3537 +Test: [1410/2387] eta: 0:03:14 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.0884 (0.2424) acc1: 100.0000 (92.4106) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1894 data: 0.0966 max mem: 3537 +Test: [1420/2387] eta: 0:03:12 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.2267 (0.2425) acc1: 93.4896 (92.3946) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1952 data: 0.1029 max mem: 3537 +Test: [1430/2387] eta: 0:03:10 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.2267 (0.2425) acc1: 93.4896 (92.3988) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1856 data: 0.0936 max mem: 3537 +Test: [1440/2387] eta: 0:03:08 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.2390 (0.2428) acc1: 92.9688 (92.3874) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.1744 data: 0.0818 max mem: 3537 +Test: [1450/2387] eta: 0:03:06 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.3093 (0.2433) acc1: 89.3229 (92.3636) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1810 data: 0.0884 max mem: 3537 +Test: [1460/2387] eta: 0:03:04 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1667 (0.2427) acc1: 96.3542 (92.3982) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1923 data: 0.1003 max mem: 3537 +Test: [1470/2387] eta: 0:03:02 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1345 (0.2419) acc1: 98.6979 (92.4424) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1999 data: 0.1055 max mem: 3537 +Test: [1480/2387] eta: 0:03:00 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1253 (0.2412) acc1: 98.9583 (92.4834) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2104 data: 0.1164 max mem: 3537 +Test: [1490/2387] eta: 0:02:58 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1244 (0.2405) acc1: 98.9583 (92.5258) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2082 data: 0.1161 max mem: 3537 +Test: [1500/2387] eta: 0:02:56 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1345 (0.2398) acc1: 98.9583 (92.5662) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1718 data: 0.0773 max mem: 3537 +Test: [1510/2387] eta: 0:02:54 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1429 (0.2392) acc1: 99.7396 (92.6148) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1666 data: 0.0707 max mem: 3537 +Test: [1520/2387] eta: 0:02:52 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1298 (0.2384) acc1: 100.0000 (92.6606) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1971 data: 0.1036 max mem: 3537 +Test: [1530/2387] eta: 0:02:50 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1028 (0.2374) acc1: 99.7396 (92.7077) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1804 data: 0.0895 max mem: 3537 +Test: [1540/2387] eta: 0:02:47 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.0926 (0.2365) acc1: 99.7396 (92.7535) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1706 data: 0.0782 max mem: 3537 +Test: [1550/2387] eta: 0:02:45 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.0924 (0.2357) acc1: 99.7396 (92.7992) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1742 data: 0.0814 max mem: 3537 +Test: [1560/2387] eta: 0:02:43 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1017 (0.2348) acc1: 99.7396 (92.8440) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1644 data: 0.0733 max mem: 3537 +Test: [1570/2387] eta: 0:02:41 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1121 (0.2342) acc1: 100.0000 (92.8875) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1649 data: 0.0718 max mem: 3537 +Test: [1580/2387] eta: 0:02:39 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1459 (0.2340) acc1: 98.9583 (92.9071) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1803 data: 0.0861 max mem: 3538 +Test: [1590/2387] eta: 0:02:37 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1925 (0.2337) acc1: 97.1354 (92.9259) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1844 data: 0.0918 max mem: 3538 +Test: [1600/2387] eta: 0:02:35 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1800 (0.2336) acc1: 96.6146 (92.9253) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1773 data: 0.0833 max mem: 3538 +Test: [1610/2387] eta: 0:02:33 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1733 (0.2332) acc1: 97.1354 (92.9529) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1757 data: 0.0818 max mem: 3538 +Test: [1620/2387] eta: 0:02:31 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1934 (0.2332) acc1: 95.8333 (92.9575) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2037 data: 0.1126 max mem: 3538 +Test: [1630/2387] eta: 0:02:29 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.2512 (0.2334) acc1: 92.1875 (92.9483) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.2347 data: 0.1442 max mem: 3538 +Test: [1640/2387] eta: 0:02:27 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.2895 (0.2340) acc1: 90.8854 (92.9273) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1759 data: 0.0841 max mem: 3538 +Test: [1650/2387] eta: 0:02:25 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.3248 (0.2345) acc1: 89.3229 (92.9076) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1383 data: 0.0433 max mem: 3538 +Test: [1660/2387] eta: 0:02:23 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.3176 (0.2351) acc1: 89.8438 (92.8824) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.1748 data: 0.0788 max mem: 3538 +Test: [1670/2387] eta: 0:02:21 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.3297 (0.2356) acc1: 90.1042 (92.8572) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1821 data: 0.0884 max mem: 3538 +Test: [1680/2387] eta: 0:02:19 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.3297 (0.2362) acc1: 90.1042 (92.8411) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1749 data: 0.0833 max mem: 3538 +Test: [1690/2387] eta: 0:02:16 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.3332 (0.2367) acc1: 89.5833 (92.8164) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1754 data: 0.0832 max mem: 3538 +Test: [1700/2387] eta: 0:02:14 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.3332 (0.2369) acc1: 88.0208 (92.8091) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1699 data: 0.0734 max mem: 3538 +Test: [1710/2387] eta: 0:02:12 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1109 (0.2361) acc1: 98.6979 (92.8470) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1624 data: 0.0645 max mem: 3538 +Test: [1720/2387] eta: 0:02:10 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.0939 (0.2352) acc1: 99.4792 (92.8854) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1710 data: 0.0773 max mem: 3538 +Test: [1730/2387] eta: 0:02:08 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.0870 (0.2344) acc1: 99.4792 (92.9208) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1747 data: 0.0836 max mem: 3538 +Test: [1740/2387] eta: 0:02:06 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.0888 (0.2336) acc1: 99.7396 (92.9586) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1547 data: 0.0613 max mem: 3538 +Test: [1750/2387] eta: 0:02:04 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1032 (0.2329) acc1: 99.7396 (92.9949) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1472 data: 0.0537 max mem: 3538 +Test: [1760/2387] eta: 0:02:02 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1227 (0.2324) acc1: 99.2188 (93.0257) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1491 data: 0.0578 max mem: 3538 +Test: [1770/2387] eta: 0:02:00 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1403 (0.2319) acc1: 98.6979 (93.0558) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1462 data: 0.0546 max mem: 3538 +Test: [1780/2387] eta: 0:01:58 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1120 (0.2312) acc1: 98.9583 (93.0898) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1503 data: 0.0589 max mem: 3538 +Test: [1790/2387] eta: 0:01:56 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.0970 (0.2304) acc1: 99.4792 (93.1251) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1566 data: 0.0650 max mem: 3538 +Test: [1800/2387] eta: 0:01:53 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.0916 (0.2297) acc1: 99.7396 (93.1606) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1611 data: 0.0677 max mem: 3538 +Test: [1810/2387] eta: 0:01:51 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1014 (0.2290) acc1: 99.7396 (93.1960) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1679 data: 0.0744 max mem: 3539 +Test: [1820/2387] eta: 0:01:49 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1124 (0.2285) acc1: 99.4792 (93.2282) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1760 data: 0.0846 max mem: 3539 +Test: [1830/2387] eta: 0:01:47 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1522 (0.2281) acc1: 99.4792 (93.2609) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1716 data: 0.0775 max mem: 3539 +Test: [1840/2387] eta: 0:01:45 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1599 (0.2277) acc1: 100.0000 (93.2975) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1696 data: 0.0748 max mem: 3539 +Test: [1850/2387] eta: 0:01:43 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1593 (0.2274) acc1: 100.0000 (93.3334) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1772 data: 0.0854 max mem: 3539 +Test: [1860/2387] eta: 0:01:42 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1597 (0.2270) acc1: 100.0000 (93.3688) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1784 data: 0.0872 max mem: 3539 +Test: [1870/2387] eta: 0:01:40 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1576 (0.2266) acc1: 100.0000 (93.4040) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1732 data: 0.0813 max mem: 3539 +Test: [1880/2387] eta: 0:01:38 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1583 (0.2263) acc1: 100.0000 (93.4388) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1695 data: 0.0779 max mem: 3539 +Test: [1890/2387] eta: 0:01:36 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1598 (0.2259) acc1: 100.0000 (93.4733) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0795 max mem: 3539 +Test: [1900/2387] eta: 0:01:34 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1608 (0.2256) acc1: 100.0000 (93.5073) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1694 data: 0.0780 max mem: 3539 +Test: [1910/2387] eta: 0:01:32 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.2308 (0.2258) acc1: 99.7396 (93.5335) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1712 data: 0.0791 max mem: 3539 +Test: [1920/2387] eta: 0:01:30 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.2500 (0.2262) acc1: 98.1771 (93.5056) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1748 data: 0.0823 max mem: 3539 +Test: [1930/2387] eta: 0:01:28 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.2038 (0.2259) acc1: 94.7917 (93.5213) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1781 data: 0.0845 max mem: 3539 +Test: [1940/2387] eta: 0:01:26 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1789 (0.2261) acc1: 96.3542 (93.4990) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1762 data: 0.0836 max mem: 3539 +Test: [1950/2387] eta: 0:01:24 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.2040 (0.2261) acc1: 94.7917 (93.5033) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1776 data: 0.0851 max mem: 3539 +Test: [1960/2387] eta: 0:01:22 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.2091 (0.2264) acc1: 94.2708 (93.4731) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1811 data: 0.0881 max mem: 3539 +Test: [1970/2387] eta: 0:01:20 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.3379 (0.2275) acc1: 82.8125 (93.3943) acc5: 100.0000 (100.0000) acc1_neg: 72.917 acc5_neg: 100.000 time: 0.1967 data: 0.1049 max mem: 3539 +Test: [1980/2387] eta: 0:01:18 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.3591 (0.2277) acc1: 86.1979 (93.3839) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1970 data: 0.1039 max mem: 3539 +Test: [1990/2387] eta: 0:01:16 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1387 (0.2272) acc1: 99.2188 (93.4142) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1799 data: 0.0856 max mem: 3539 +Test: [2000/2387] eta: 0:01:14 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1375 (0.2268) acc1: 99.4792 (93.4439) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2098 data: 0.1163 max mem: 3539 +Test: [2010/2387] eta: 0:01:12 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1387 (0.2264) acc1: 99.4792 (93.4722) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2076 data: 0.1147 max mem: 3539 +Test: [2020/2387] eta: 0:01:10 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1397 (0.2259) acc1: 99.2188 (93.5013) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1699 data: 0.0762 max mem: 3539 +Test: [2030/2387] eta: 0:01:08 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1416 (0.2255) acc1: 99.2188 (93.5298) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1874 data: 0.0943 max mem: 3540 +Test: [2040/2387] eta: 0:01:06 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1386 (0.2251) acc1: 99.4792 (93.5599) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2021 data: 0.1094 max mem: 3540 +Test: [2050/2387] eta: 0:01:04 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1386 (0.2247) acc1: 99.4792 (93.5870) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1847 data: 0.0920 max mem: 3540 +Test: [2060/2387] eta: 0:01:02 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1674 (0.2245) acc1: 98.6979 (93.6109) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1677 data: 0.0753 max mem: 3540 +Test: [2070/2387] eta: 0:01:00 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1721 (0.2242) acc1: 98.6979 (93.6370) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1568 data: 0.0631 max mem: 3540 +Test: [2080/2387] eta: 0:00:58 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1695 (0.2240) acc1: 98.9583 (93.6622) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1599 data: 0.0654 max mem: 3540 +Test: [2090/2387] eta: 0:00:57 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1694 (0.2237) acc1: 98.6979 (93.6867) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1629 data: 0.0693 max mem: 3540 +Test: [2100/2387] eta: 0:00:55 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1696 (0.2234) acc1: 98.9583 (93.7121) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1799 data: 0.0847 max mem: 3540 +Test: [2110/2387] eta: 0:00:53 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1695 (0.2232) acc1: 98.6979 (93.7364) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1849 data: 0.0885 max mem: 3540 +Test: [2120/2387] eta: 0:00:51 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1797 (0.2233) acc1: 98.4375 (93.7405) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1559 data: 0.0615 max mem: 3540 +Test: [2130/2387] eta: 0:00:49 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.2207 (0.2232) acc1: 95.3125 (93.7518) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1511 data: 0.0570 max mem: 3540 +Test: [2140/2387] eta: 0:00:47 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.2119 (0.2234) acc1: 95.5729 (93.7360) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1645 data: 0.0697 max mem: 3540 +Test: [2150/2387] eta: 0:00:45 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.2041 (0.2233) acc1: 95.5729 (93.7479) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1713 data: 0.0747 max mem: 3540 +Test: [2160/2387] eta: 0:00:43 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.2310 (0.2234) acc1: 96.8750 (93.7580) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1758 data: 0.0804 max mem: 3540 +Test: [2170/2387] eta: 0:00:41 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.2635 (0.2240) acc1: 91.4062 (93.7228) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.1675 data: 0.0756 max mem: 3540 +Test: [2180/2387] eta: 0:00:39 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.3780 (0.2250) acc1: 83.3333 (93.6695) acc5: 100.0000 (100.0000) acc1_neg: 75.781 acc5_neg: 100.000 time: 0.1633 data: 0.0704 max mem: 3540 +Test: [2190/2387] eta: 0:00:37 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.3118 (0.2249) acc1: 89.3229 (93.6731) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1746 data: 0.0793 max mem: 3540 +Test: [2200/2387] eta: 0:00:35 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1655 (0.2246) acc1: 97.9167 (93.6924) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1753 data: 0.0799 max mem: 3540 +Test: [2210/2387] eta: 0:00:33 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1213 (0.2241) acc1: 99.4792 (93.7164) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1822 data: 0.0874 max mem: 3540 +Test: [2220/2387] eta: 0:00:31 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1198 (0.2237) acc1: 99.4792 (93.7353) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1997 data: 0.1071 max mem: 3540 +Test: [2230/2387] eta: 0:00:29 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1250 (0.2233) acc1: 98.9583 (93.7557) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1776 data: 0.0837 max mem: 3540 +Test: [2240/2387] eta: 0:00:28 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1336 (0.2229) acc1: 99.2188 (93.7780) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1763 data: 0.0803 max mem: 3540 +Test: [2250/2387] eta: 0:00:26 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1400 (0.2226) acc1: 99.7396 (93.8028) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2125 data: 0.1184 max mem: 3540 +Test: [2260/2387] eta: 0:00:24 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1567 (0.2224) acc1: 99.4792 (93.8215) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1832 data: 0.0901 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1800 (0.2222) acc1: 98.4375 (93.8409) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2093 data: 0.1161 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1770 (0.2222) acc1: 97.9167 (93.8403) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2473 data: 0.1554 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1770 (0.2220) acc1: 97.9167 (93.8617) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2003 data: 0.1097 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.1891 (0.2220) acc1: 98.4375 (93.8789) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1774 data: 0.0862 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.2185 (0.2222) acc1: 96.3542 (93.8806) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.2235 data: 0.1323 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.2991 (0.2226) acc1: 92.4479 (93.8648) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.2299 data: 0.1374 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.3072 (0.2230) acc1: 90.6250 (93.8509) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.2158 data: 0.1227 max mem: 3541 +Test: [2340/2387] eta: 0:00:09 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.2568 (0.2231) acc1: 90.8854 (93.8371) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.2070 data: 0.1139 max mem: 3541 +Test: [2350/2387] eta: 0:00:07 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.2516 (0.2235) acc1: 91.9271 (93.8039) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2006 data: 0.1067 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.2705 (0.2238) acc1: 90.8854 (93.7846) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.2157 data: 0.1225 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.2932 (0.2243) acc1: 89.3229 (93.7512) acc5: 100.0000 (100.0000) acc1_neg: 68.229 acc5_neg: 100.000 time: 0.1881 data: 0.0962 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.3841 (0.2252) acc1: 83.5938 (93.6984) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1710 data: 0.0803 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 96.347 acc5_pos: 100.000 loss: 0.3866 (0.2255) acc1: 81.7708 (93.6762) acc5: 100.0000 (100.0000) acc1_neg: 95.349 acc5_neg: 100.000 time: 0.1599 data: 0.0714 max mem: 3541 +Test: Total time: 0:07:37 (0.1917 s / it) +* Overall Acc@1 93.676 Acc@5 100.000 + target == real Acc@1 66.660 Acc@5 100.000 + target == fake Acc@1 95.573 Acc@5 100.000 + loss 0.226 + PR-AUC 0.9967 ROC-AUC 0.9544 EER 10.67% (thr=0.6733) +Accuracy of the network on the 916267 test images: 93.7% +[rank0]:[W726 03:10:13.667361653 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:10:34.298000 1170863 torch/distributed/run.py:766] +W0726 03:10:34.298000 1170863 torch/distributed/run.py:766] ***************************************** +W0726 03:10:34.298000 1170863 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:10:34.298000 1170863 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:11:02.693537639 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:11:02.953651297 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W726 03:11:02.954220304 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:11:02.203254672 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:11:02.273804421 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:11:03.444163186 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:11:03.509353308 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank1]:[W726 03:11:03.509437541 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n23_seed5_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed5_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 0:57:36 acc1_pos: 73.698 acc5_pos: 100.000 loss: 0.5223 (0.5223) acc1: 73.6979 (73.6979) acc5: 100.0000 (100.0000) time: 6.6865 data: 4.9005 max mem: 3531 +Test: [ 10/517] eta: 0:05:49 acc1_pos: 42.708 acc5_pos: 100.000 loss: 0.7132 (0.7385) acc1: 54.9479 (51.4441) acc5: 100.0000 (100.0000) time: 0.6902 data: 0.4459 max mem: 3531 +Test: [ 20/517] eta: 0:03:25 acc1_pos: 55.469 acc5_pos: 100.000 loss: 0.7177 (0.7433) acc1: 53.9062 (52.9018) acc5: 100.0000 (100.0000) time: 0.0997 data: 0.0051 max mem: 3531 +Test: [ 30/517] eta: 0:02:51 acc1_pos: 22.135 acc5_pos: 100.000 loss: 0.7438 (0.7613) acc1: 49.2188 (50.6048) acc5: 100.0000 (100.0000) time: 0.1671 data: 0.0726 max mem: 3531 +Test: [ 40/517] eta: 0:02:36 acc1_pos: 51.823 acc5_pos: 100.000 loss: 0.8185 (0.7767) acc1: 38.5417 (47.8277) acc5: 100.0000 (100.0000) time: 0.2377 data: 0.1453 max mem: 3531 +Test: [ 50/517] eta: 0:02:27 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.7674 (0.7595) acc1: 46.8750 (49.7141) acc5: 100.0000 (100.0000) acc1_neg: 47.135 acc5_neg: 100.000 time: 0.2578 data: 0.1639 max mem: 3531 +Test: [ 60/517] eta: 0:02:19 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.6652 (0.7474) acc1: 59.3750 (51.1996) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.2606 data: 0.1672 max mem: 3531 +Test: [ 70/517] eta: 0:02:14 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.5077 (0.7105) acc1: 75.0000 (55.6778) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.2643 data: 0.1691 max mem: 3531 +Test: [ 80/517] eta: 0:02:09 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.4961 (0.6903) acc1: 79.1667 (58.1822) acc5: 100.0000 (100.0000) acc1_neg: 54.948 acc5_neg: 100.000 time: 0.2665 data: 0.1706 max mem: 3531 +Test: [ 90/517] eta: 0:02:12 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.4968 (0.6689) acc1: 79.9479 (60.9289) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.3399 data: 0.2438 max mem: 3531 +Test: [100/517] eta: 0:02:06 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.4503 (0.6453) acc1: 86.4583 (63.4643) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.3349 data: 0.2367 max mem: 3531 +Test: [110/517] eta: 0:02:01 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.4040 (0.6245) acc1: 86.9792 (65.3740) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.2498 data: 0.1526 max mem: 3531 +Test: [120/517] eta: 0:01:57 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.4730 (0.6166) acc1: 77.0833 (66.0490) acc5: 100.0000 (100.0000) acc1_neg: 72.917 acc5_neg: 100.000 time: 0.2499 data: 0.1550 max mem: 3531 +Test: [130/517] eta: 0:01:53 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.5161 (0.6083) acc1: 75.0000 (66.9927) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.2549 data: 0.1587 max mem: 3531 +Test: [140/517] eta: 0:01:49 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.5036 (0.6001) acc1: 81.5104 (68.3178) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.2599 data: 0.1615 max mem: 3531 +Test: [150/517] eta: 0:01:45 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.4801 (0.5904) acc1: 84.3750 (69.5071) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.2549 data: 0.1578 max mem: 3531 +Test: [160/517] eta: 0:01:41 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.4689 (0.5825) acc1: 82.8125 (70.2429) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.2499 data: 0.1529 max mem: 3531 +Test: [170/517] eta: 0:01:38 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.4553 (0.5732) acc1: 82.2917 (71.0115) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2667 data: 0.1704 max mem: 3531 +Test: [180/517] eta: 0:01:37 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.4330 (0.5658) acc1: 85.4167 (71.8189) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.3227 data: 0.2265 max mem: 3531 +Test: [190/517] eta: 0:01:35 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.3442 (0.5541) acc1: 90.8854 (72.9508) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.3536 data: 0.2591 max mem: 3531 +Test: [200/517] eta: 0:01:33 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.3809 (0.5466) acc1: 92.4479 (73.8560) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.3400 data: 0.2473 max mem: 3531 +Test: [210/517] eta: 0:01:31 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.4455 (0.5471) acc1: 83.8542 (73.8818) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.3444 data: 0.2507 max mem: 3532 +Test: [220/517] eta: 0:01:29 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.5782 (0.5482) acc1: 72.6562 (73.8440) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.3798 data: 0.2841 max mem: 3532 +Test: [230/517] eta: 0:01:25 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.5699 (0.5489) acc1: 75.5208 (73.9989) acc5: 100.0000 (100.0000) acc1_neg: 75.781 acc5_neg: 100.000 time: 0.3049 data: 0.2071 max mem: 3532 +Test: [240/517] eta: 0:01:23 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.5473 (0.5474) acc1: 78.9062 (74.0696) acc5: 100.0000 (100.0000) acc1_neg: 57.292 acc5_neg: 100.000 time: 0.3022 data: 0.2047 max mem: 3532 +Test: [250/517] eta: 0:01:19 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.4700 (0.5452) acc1: 81.5104 (74.4750) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.3020 data: 0.2038 max mem: 3532 +Test: [260/517] eta: 0:01:16 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.4628 (0.5429) acc1: 87.5000 (74.9202) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.2297 data: 0.1336 max mem: 3532 +Test: [270/517] eta: 0:01:12 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.4764 (0.5408) acc1: 85.9375 (75.2739) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.2477 data: 0.1514 max mem: 3532 +Test: [280/517] eta: 0:01:09 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.4632 (0.5376) acc1: 85.6771 (75.6867) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.2403 data: 0.1418 max mem: 3532 +Test: [290/517] eta: 0:01:05 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.4419 (0.5352) acc1: 86.7188 (76.0569) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.2416 data: 0.1434 max mem: 3532 +Test: [300/517] eta: 0:01:02 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.4664 (0.5331) acc1: 85.4167 (76.3125) acc5: 100.0000 (100.0000) acc1_neg: 73.698 acc5_neg: 100.000 time: 0.2517 data: 0.1564 max mem: 3532 +Test: [310/517] eta: 0:01:00 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.4952 (0.5332) acc1: 79.4271 (76.3163) acc5: 100.0000 (100.0000) acc1_neg: 73.698 acc5_neg: 100.000 time: 0.2895 data: 0.1968 max mem: 3532 +Test: [320/517] eta: 0:00:56 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.5009 (0.5309) acc1: 82.2917 (76.5414) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.2776 data: 0.1827 max mem: 3532 +Test: [330/517] eta: 0:00:54 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.3992 (0.5267) acc1: 91.6667 (77.0172) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2686 data: 0.1704 max mem: 3532 +Test: [340/517] eta: 0:00:50 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.3959 (0.5244) acc1: 93.4896 (77.3216) acc5: 100.0000 (100.0000) acc1_neg: 66.667 acc5_neg: 100.000 time: 0.2649 data: 0.1657 max mem: 3532 +Test: [350/517] eta: 0:00:47 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.4998 (0.5245) acc1: 77.3438 (77.2822) acc5: 100.0000 (100.0000) acc1_neg: 66.406 acc5_neg: 100.000 time: 0.2303 data: 0.1328 max mem: 3532 +Test: [360/517] eta: 0:00:44 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.5075 (0.5240) acc1: 77.3438 (77.3784) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.2449 data: 0.1463 max mem: 3532 +Test: [370/517] eta: 0:00:41 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.4086 (0.5200) acc1: 95.8333 (77.9264) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2595 data: 0.1605 max mem: 3532 +Test: [380/517] eta: 0:00:38 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.3731 (0.5164) acc1: 96.6146 (78.3362) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2449 data: 0.1484 max mem: 3532 +Test: [390/517] eta: 0:00:35 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.3725 (0.5126) acc1: 94.7917 (78.7717) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2299 data: 0.1326 max mem: 3532 +Test: [400/517] eta: 0:00:32 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.3758 (0.5095) acc1: 94.5312 (79.1335) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2302 data: 0.1323 max mem: 3532 +Test: [410/517] eta: 0:00:29 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.3901 (0.5072) acc1: 96.6146 (79.5836) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2450 data: 0.1502 max mem: 3532 +Test: [420/517] eta: 0:00:27 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.4663 (0.5074) acc1: 94.2708 (79.7259) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.2567 data: 0.1602 max mem: 3532 +Test: [430/517] eta: 0:00:24 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.5156 (0.5088) acc1: 78.9062 (79.5769) acc5: 100.0000 (100.0000) acc1_neg: 68.490 acc5_neg: 100.000 time: 0.2501 data: 0.1536 max mem: 3532 +Test: [440/517] eta: 0:00:21 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.5145 (0.5086) acc1: 78.9062 (79.6792) acc5: 100.0000 (100.0000) acc1_neg: 80.729 acc5_neg: 100.000 time: 0.2257 data: 0.1323 max mem: 3533 +Test: [450/517] eta: 0:00:18 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.4591 (0.5071) acc1: 90.6250 (79.9866) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2171 data: 0.1237 max mem: 3533 +Test: [460/517] eta: 0:00:15 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.4578 (0.5062) acc1: 91.9271 (80.1767) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.2421 data: 0.1502 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.4387 (0.5045) acc1: 89.8438 (80.4157) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2442 data: 0.1504 max mem: 3533 +Test: [480/517] eta: 0:00:10 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.4229 (0.5030) acc1: 92.9688 (80.7026) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2517 data: 0.1572 max mem: 3533 +Test: [490/517] eta: 0:00:07 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.4453 (0.5028) acc1: 91.1458 (80.6873) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.2895 data: 0.1968 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.5136 (0.5033) acc1: 71.6146 (80.6304) acc5: 100.0000 (100.0000) acc1_neg: 64.323 acc5_neg: 100.000 time: 0.2414 data: 0.1501 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.5569 (0.5045) acc1: 71.6146 (80.5187) acc5: 100.0000 (100.0000) acc1_neg: 76.302 acc5_neg: 100.000 time: 0.1676 data: 0.0774 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.5569 (0.5059) acc1: 71.0938 (80.3636) acc5: 100.0000 (100.0000) acc1_neg: 68.551 acc5_neg: 100.000 time: 0.1685 data: 0.0768 max mem: 3533 +Test: Total time: 0:02:19 (0.2692 s / it) +* Overall Acc@1 80.364 Acc@5 100.000 + target == real Acc@1 49.571 Acc@5 100.000 + target == fake Acc@1 83.558 Acc@5 100.000 + loss 0.506 + PR-AUC 0.9540 ROC-AUC 0.7288 EER 33.02% (thr=0.5859) +Accuracy of the network on the 198427 test images: 80.4% +[rank0]:[W726 03:13:35.188386939 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 03:13:56.418000 1171436 torch/distributed/run.py:766] +W0726 03:13:56.418000 1171436 torch/distributed/run.py:766] ***************************************** +W0726 03:13:56.418000 1171436 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:13:56.418000 1171436 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:14:28.138219772 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:14:28.317427188 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:14:28.386642755 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:14:29.489992313 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:14:29.602124488 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:14:29.635474875 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:14:29.771633038 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:14:29.797368421 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n23_seed5_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed5_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:43:27 acc1_pos: 3.646 acc5_pos: 100.000 loss: 1.8555 (1.8555) acc1: 3.6458 (3.6458) acc5: 100.0000 (100.0000) time: 6.7036 data: 5.1612 max mem: 3531 +Test: [ 10/389] eta: 0:04:25 acc1_pos: 41.927 acc5_pos: 100.000 loss: 1.1841 (1.3106) acc1: 14.8438 (16.5720) acc5: 100.0000 (100.0000) time: 0.6999 data: 0.4696 max mem: 3531 +Test: [ 20/389] eta: 0:02:34 acc1_pos: 0.260 acc5_pos: 100.000 loss: 1.1841 (1.3518) acc1: 18.4896 (18.3656) acc5: 100.0000 (100.0000) time: 0.1047 data: 0.0036 max mem: 3531 +Test: [ 30/389] eta: 0:02:18 acc1_pos: 1.823 acc5_pos: 100.000 loss: 1.6364 (1.4344) acc1: 0.5208 (13.1888) acc5: 100.0000 (100.0000) time: 0.2115 data: 0.1075 max mem: 3531 +Test: [ 40/389] eta: 0:01:59 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.5465 (1.4362) acc1: 1.5625 (12.9319) acc5: 100.0000 (100.0000) time: 0.2609 data: 0.1630 max mem: 3531 +Test: [ 50/389] eta: 0:01:46 acc1_pos: 88.281 acc5_pos: 100.000 loss: 1.2069 (1.3578) acc1: 7.0312 (18.2802) acc5: 100.0000 (100.0000) time: 0.2033 data: 0.1084 max mem: 3531 +Test: [ 60/389] eta: 0:01:31 acc1_pos: 12.760 acc5_pos: 100.000 loss: 1.0124 (1.3281) acc1: 12.7604 (22.8569) acc5: 100.0000 (100.0000) time: 0.1489 data: 0.0498 max mem: 3531 +Test: [ 70/389] eta: 0:01:22 acc1_pos: 6.510 acc5_pos: 100.000 loss: 1.5525 (1.3705) acc1: 2.0833 (20.0337) acc5: 100.0000 (100.0000) time: 0.1179 data: 0.0214 max mem: 3531 +Test: [ 80/389] eta: 0:01:16 acc1_pos: 53.385 acc5_pos: 100.000 loss: 1.6082 (1.3792) acc1: 1.3021 (20.7112) acc5: 100.0000 (100.0000) time: 0.1554 data: 0.0603 max mem: 3531 +Test: [ 90/389] eta: 0:01:11 acc1_pos: 11.458 acc5_pos: 100.000 loss: 1.0214 (1.3344) acc1: 30.7292 (23.0769) acc5: 100.0000 (100.0000) time: 0.1681 data: 0.0744 max mem: 3531 +Test: [100/389] eta: 0:01:07 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.1256 (1.3343) acc1: 21.3542 (22.1380) acc5: 100.0000 (100.0000) time: 0.1744 data: 0.0783 max mem: 3531 +Test: [110/389] eta: 0:01:04 acc1_pos: 97.396 acc5_pos: 100.000 loss: 1.3491 (1.3425) acc1: 11.9792 (23.1349) acc5: 100.0000 (100.0000) time: 0.1947 data: 0.0980 max mem: 3531 +Test: [120/389] eta: 0:01:01 acc1_pos: 27.083 acc5_pos: 100.000 loss: 1.2312 (1.3189) acc1: 31.2500 (24.5739) acc5: 100.0000 (100.0000) time: 0.1971 data: 0.1058 max mem: 3531 +Test: [130/389] eta: 0:00:58 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.6334 (1.2661) acc1: 49.4792 (27.9401) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2096 data: 0.1170 max mem: 3531 +Test: [140/389] eta: 0:00:56 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2015 (1.1993) acc1: 96.3542 (31.9167) acc5: 100.0000 (100.0000) acc1_neg: 28.385 acc5_neg: 100.000 time: 0.2242 data: 0.1313 max mem: 3531 +Test: [150/389] eta: 0:00:53 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2113 (1.1406) acc1: 99.2188 (35.7719) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2001 data: 0.1083 max mem: 3531 +Test: [160/389] eta: 0:00:51 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2235 (1.0843) acc1: 98.6979 (39.4038) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1995 data: 0.1078 max mem: 3531 +Test: [170/389] eta: 0:00:48 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2327 (1.0361) acc1: 100.0000 (42.8682) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1991 data: 0.1084 max mem: 3531 +Test: [180/389] eta: 0:00:45 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2342 (0.9933) acc1: 99.7396 (45.8377) acc5: 100.0000 (100.0000) acc1_neg: 71.615 acc5_neg: 100.000 time: 0.1896 data: 0.0976 max mem: 3531 +Test: [190/389] eta: 0:00:43 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1965 (0.9544) acc1: 99.7396 (48.4116) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1880 data: 0.0953 max mem: 3531 +Test: [200/389] eta: 0:00:40 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1750 (0.9196) acc1: 100.0000 (50.8344) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1766 data: 0.0848 max mem: 3531 +Test: [210/389] eta: 0:00:38 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1745 (0.8848) acc1: 100.0000 (53.1571) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1746 data: 0.0831 max mem: 3532 +Test: [220/389] eta: 0:00:35 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1691 (0.8527) acc1: 100.0000 (55.2284) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1857 data: 0.0935 max mem: 3532 +Test: [230/389] eta: 0:00:33 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1697 (0.8258) acc1: 100.0000 (57.1282) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1993 data: 0.1062 max mem: 3532 +Test: [240/389] eta: 0:00:31 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2077 (0.8012) acc1: 99.4792 (58.7645) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1973 data: 0.1051 max mem: 3532 +Test: [250/389] eta: 0:00:29 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1610 (0.7753) acc1: 100.0000 (60.4073) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2008 data: 0.1095 max mem: 3532 +Test: [260/389] eta: 0:00:27 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1479 (0.7549) acc1: 100.0000 (61.6709) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2049 data: 0.1132 max mem: 3532 +Test: [270/389] eta: 0:00:25 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1863 (0.7371) acc1: 99.2188 (62.9939) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2027 data: 0.1105 max mem: 3532 +Test: [280/389] eta: 0:00:22 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2555 (0.7214) acc1: 99.2188 (63.9606) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2008 data: 0.1097 max mem: 3532 +Test: [290/389] eta: 0:00:20 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2168 (0.7050) acc1: 98.4375 (65.1122) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2073 data: 0.1150 max mem: 3532 +Test: [300/389] eta: 0:00:18 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2591 (0.6905) acc1: 99.7396 (66.2566) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2002 data: 0.1068 max mem: 3532 +Test: [310/389] eta: 0:00:16 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2831 (0.6791) acc1: 99.2188 (67.1339) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.1787 data: 0.0870 max mem: 3532 +Test: [320/389] eta: 0:00:14 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2709 (0.6655) acc1: 93.4896 (68.0199) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1717 data: 0.0801 max mem: 3532 +Test: [330/389] eta: 0:00:12 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1862 (0.6512) acc1: 100.0000 (68.9577) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1698 data: 0.0780 max mem: 3532 +Test: [340/389] eta: 0:00:10 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1666 (0.6368) acc1: 100.0000 (69.8031) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1643 data: 0.0733 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1444 (0.6358) acc1: 100.0000 (69.4719) acc5: 100.0000 (100.0000) acc1_neg: 18.490 acc5_neg: 100.000 time: 0.1691 data: 0.0786 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.4926 (0.6347) acc1: 71.6146 (69.2377) acc5: 100.0000 (100.0000) acc1_neg: 77.344 acc5_neg: 100.000 time: 0.1665 data: 0.0753 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.3958 (0.6248) acc1: 77.3438 (69.9643) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1467 data: 0.0557 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2081 (0.6135) acc1: 99.7396 (70.7452) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1276 data: 0.0379 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1932 (0.6052) acc1: 99.7396 (71.2603) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1198 data: 0.0266 max mem: 3532 +Test: Total time: 0:01:16 (0.1978 s / it) +* Overall Acc@1 71.260 Acc@5 100.000 + target == real Acc@1 24.110 Acc@5 100.000 + target == fake Acc@1 93.177 Acc@5 100.000 + loss 0.605 + PR-AUC 0.7673 ROC-AUC 0.6583 EER 38.41% (thr=0.8037) +Accuracy of the network on the 149358 test images: 71.3% +[rank0]:[W726 03:15:53.290968221 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:16:15.356000 1171994 torch/distributed/run.py:766] +W0726 03:16:15.356000 1171994 torch/distributed/run.py:766] ***************************************** +W0726 03:16:15.356000 1171994 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:16:15.356000 1171994 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:16:47.673351798 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:16:47.831255370 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:16:47.082631014 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:16:47.153801413 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank1]:[W726 03:16:47.155699010 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:16:47.291255968 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:16:47.322322085 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:16:47.324716435 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n23_seed5_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed5_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:32:38 acc1_pos: 10.938 acc5_pos: 100.000 loss: 1.4095 (1.4095) acc1: 10.9375 (10.9375) acc5: 100.0000 (100.0000) time: 5.8805 data: 4.3625 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/333] eta: 0:03:19 acc1_pos: 50.000 acc5_pos: 100.000 loss: 1.1864 (1.0657) acc1: 35.1562 (33.3807) acc5: 100.0000 (100.0000) time: 0.6167 data: 0.3971 max mem: 3531 +Test: [ 20/333] eta: 0:02:00 acc1_pos: 56.250 acc5_pos: 100.000 loss: 0.8747 (0.9695) acc1: 43.2292 (39.3229) acc5: 100.0000 (100.0000) time: 0.1105 data: 0.0171 max mem: 3531 +Test: [ 30/333] eta: 0:01:40 acc1_pos: 77.865 acc5_pos: 100.000 loss: 0.7388 (0.8518) acc1: 53.1250 (48.0343) acc5: 100.0000 (100.0000) time: 0.1765 data: 0.0817 max mem: 3531 +Test: [ 40/333] eta: 0:01:40 acc1_pos: 71.615 acc5_pos: 100.000 loss: 0.5665 (0.7860) acc1: 66.4062 (52.9853) acc5: 100.0000 (100.0000) time: 0.2956 data: 0.2026 max mem: 3531 +Test: [ 50/333] eta: 0:01:27 acc1_pos: 66.146 acc5_pos: 100.000 loss: 0.5892 (0.7466) acc1: 66.9271 (55.8058) acc5: 100.0000 (100.0000) time: 0.2738 data: 0.1804 max mem: 3531 +Test: [ 60/333] eta: 0:01:17 acc1_pos: 66.927 acc5_pos: 100.000 loss: 0.5853 (0.7211) acc1: 66.9271 (57.7869) acc5: 100.0000 (100.0000) time: 0.1693 data: 0.0746 max mem: 3531 +Test: [ 70/333] eta: 0:01:13 acc1_pos: 66.927 acc5_pos: 100.000 loss: 0.5703 (0.7002) acc1: 68.4896 (59.4080) acc5: 100.0000 (100.0000) time: 0.2049 data: 0.1095 max mem: 3531 +Test: [ 80/333] eta: 0:01:08 acc1_pos: 70.573 acc5_pos: 100.000 loss: 0.5759 (0.6864) acc1: 67.9688 (60.6353) acc5: 100.0000 (100.0000) time: 0.2324 data: 0.1383 max mem: 3531 +Test: [ 90/333] eta: 0:01:02 acc1_pos: 80.469 acc5_pos: 100.000 loss: 0.5634 (0.6687) acc1: 69.7917 (62.1423) acc5: 100.0000 (100.0000) time: 0.1815 data: 0.0894 max mem: 3531 +Test: [100/333] eta: 0:00:58 acc1_pos: 79.427 acc5_pos: 100.000 loss: 0.5176 (0.6547) acc1: 72.3958 (63.1755) acc5: 100.0000 (100.0000) time: 0.1745 data: 0.0824 max mem: 3531 +Test: [110/333] eta: 0:00:55 acc1_pos: 72.656 acc5_pos: 100.000 loss: 0.5223 (0.6424) acc1: 72.6562 (64.2080) acc5: 100.0000 (100.0000) time: 0.1956 data: 0.1028 max mem: 3531 +Test: [120/333] eta: 0:00:51 acc1_pos: 76.302 acc5_pos: 100.000 loss: 0.5008 (0.6312) acc1: 73.9583 (65.0999) acc5: 100.0000 (100.0000) time: 0.1979 data: 0.1042 max mem: 3531 +Test: [130/333] eta: 0:00:48 acc1_pos: 89.323 acc5_pos: 100.000 loss: 0.4850 (0.6205) acc1: 75.7812 (65.9888) acc5: 100.0000 (100.0000) time: 0.2077 data: 0.1139 max mem: 3531 +Test: [140/333] eta: 0:00:45 acc1_pos: 80.208 acc5_pos: 100.000 loss: 0.4768 (0.6110) acc1: 76.5625 (66.8218) acc5: 100.0000 (100.0000) time: 0.2053 data: 0.1125 max mem: 3531 +Test: [150/333] eta: 0:00:43 acc1_pos: 76.562 acc5_pos: 100.000 loss: 0.4825 (0.6047) acc1: 76.5625 (67.2875) acc5: 100.0000 (100.0000) time: 0.2076 data: 0.1122 max mem: 3531 +Test: [160/333] eta: 0:00:40 acc1_pos: 66.776 acc5_pos: 100.000 loss: 0.4982 (0.6095) acc1: 73.6979 (66.8737) acc5: 100.0000 (100.0000) acc1_neg: 40.365 acc5_neg: 100.000 time: 0.2061 data: 0.1117 max mem: 3531 +Test: [170/333] eta: 0:00:38 acc1_pos: 66.776 acc5_pos: 100.000 loss: 0.8109 (0.6237) acc1: 48.1771 (65.6524) acc5: 100.0000 (100.0000) acc1_neg: 54.948 acc5_neg: 100.000 time: 0.2223 data: 0.1309 max mem: 3531 +Test: [180/333] eta: 0:00:35 acc1_pos: 66.776 acc5_pos: 100.000 loss: 0.7897 (0.6305) acc1: 48.6979 (64.9948) acc5: 100.0000 (100.0000) acc1_neg: 61.198 acc5_neg: 100.000 time: 0.2217 data: 0.1302 max mem: 3531 +Test: [190/333] eta: 0:00:33 acc1_pos: 66.776 acc5_pos: 100.000 loss: 0.7268 (0.6337) acc1: 54.9479 (64.6665) acc5: 100.0000 (100.0000) acc1_neg: 54.948 acc5_neg: 100.000 time: 0.1998 data: 0.1064 max mem: 3531 +Test: [200/333] eta: 0:00:30 acc1_pos: 66.776 acc5_pos: 100.000 loss: 0.7075 (0.6370) acc1: 57.5521 (64.3773) acc5: 100.0000 (100.0000) acc1_neg: 57.552 acc5_neg: 100.000 time: 0.2165 data: 0.1202 max mem: 3531 +Test: [210/333] eta: 0:00:28 acc1_pos: 66.776 acc5_pos: 100.000 loss: 0.7152 (0.6417) acc1: 57.0312 (63.9699) acc5: 100.0000 (100.0000) acc1_neg: 64.844 acc5_neg: 100.000 time: 0.2199 data: 0.1239 max mem: 3532 +Test: [220/333] eta: 0:00:26 acc1_pos: 66.776 acc5_pos: 100.000 loss: 0.7196 (0.6502) acc1: 56.5104 (63.2329) acc5: 100.0000 (100.0000) acc1_neg: 58.333 acc5_neg: 100.000 time: 0.2299 data: 0.1368 max mem: 3532 +Test: [230/333] eta: 0:00:24 acc1_pos: 66.776 acc5_pos: 100.000 loss: 0.6718 (0.6500) acc1: 62.2396 (63.4177) acc5: 100.0000 (100.0000) acc1_neg: 65.885 acc5_neg: 100.000 time: 0.2799 data: 0.1875 max mem: 3532 +Test: [240/333] eta: 0:00:21 acc1_pos: 66.776 acc5_pos: 100.000 loss: 0.6718 (0.6557) acc1: 64.5833 (62.8209) acc5: 100.0000 (100.0000) acc1_neg: 43.750 acc5_neg: 100.000 time: 0.2345 data: 0.1385 max mem: 3532 +Test: [250/333] eta: 0:00:19 acc1_pos: 66.776 acc5_pos: 100.000 loss: 0.7996 (0.6607) acc1: 46.0938 (62.3164) acc5: 100.0000 (100.0000) acc1_neg: 54.427 acc5_neg: 100.000 time: 0.1695 data: 0.0721 max mem: 3532 +Test: [260/333] eta: 0:00:16 acc1_pos: 66.776 acc5_pos: 100.000 loss: 0.7580 (0.6637) acc1: 53.6458 (62.0989) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.2110 data: 0.1154 max mem: 3532 +Test: [270/333] eta: 0:00:14 acc1_pos: 66.776 acc5_pos: 100.000 loss: 0.7186 (0.6646) acc1: 58.3333 (62.0215) acc5: 100.0000 (100.0000) acc1_neg: 58.333 acc5_neg: 100.000 time: 0.2199 data: 0.1267 max mem: 3532 +Test: [280/333] eta: 0:00:11 acc1_pos: 66.776 acc5_pos: 100.000 loss: 0.7186 (0.6708) acc1: 57.0312 (61.4769) acc5: 100.0000 (100.0000) acc1_neg: 46.094 acc5_neg: 100.000 time: 0.1813 data: 0.0862 max mem: 3532 +Test: [290/333] eta: 0:00:09 acc1_pos: 66.776 acc5_pos: 100.000 loss: 0.9378 (0.6807) acc1: 38.5417 (60.6601) acc5: 100.0000 (100.0000) acc1_neg: 35.417 acc5_neg: 100.000 time: 0.1756 data: 0.0812 max mem: 3532 +Test: [300/333] eta: 0:00:07 acc1_pos: 66.776 acc5_pos: 100.000 loss: 0.9378 (0.6883) acc1: 38.8021 (60.0074) acc5: 100.0000 (100.0000) acc1_neg: 42.708 acc5_neg: 100.000 time: 0.1938 data: 0.1037 max mem: 3532 +Test: [310/333] eta: 0:00:05 acc1_pos: 66.776 acc5_pos: 100.000 loss: 0.8560 (0.6882) acc1: 45.3125 (59.9963) acc5: 100.0000 (100.0000) acc1_neg: 50.521 acc5_neg: 100.000 time: 0.1622 data: 0.0723 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 66.776 acc5_pos: 100.000 loss: 0.7504 (0.6909) acc1: 53.1250 (59.7052) acc5: 100.0000 (100.0000) acc1_neg: 52.083 acc5_neg: 100.000 time: 0.1200 data: 0.0302 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 66.776 acc5_pos: 100.000 loss: 0.8325 (0.6967) acc1: 46.6146 (59.2531) acc5: 100.0000 (100.0000) acc1_neg: 40.885 acc5_neg: 100.000 time: 0.1178 data: 0.0280 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 66.776 acc5_pos: 100.000 loss: 0.8536 (0.6977) acc1: 47.1354 (59.1859) acc5: 100.0000 (100.0000) acc1_neg: 47.324 acc5_neg: 100.000 time: 0.1209 data: 0.0280 max mem: 3532 +Test: Total time: 0:01:11 (0.2147 s / it) +* Overall Acc@1 59.186 Acc@5 100.000 + target == real Acc@1 67.650 Acc@5 100.000 + target == fake Acc@1 51.651 Acc@5 100.000 + loss 0.698 + PR-AUC 0.6365 ROC-AUC 0.6389 EER 39.76% (thr=0.4460) +Accuracy of the network on the 127843 test images: 59.2% +[rank0]:[W726 03:18:07.988373764 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:18:27.659000 1172555 torch/distributed/run.py:766] +W0726 03:18:27.659000 1172555 torch/distributed/run.py:766] ***************************************** +W0726 03:18:27.659000 1172555 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:18:27.659000 1172555 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:18:55.164029082 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:18:56.718034935 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:18:56.938157762 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:18:56.022885219 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:18:56.045917022 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:18:56.076345067 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:18:56.161605431 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:18:56.162091138 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n23_seed5_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed5_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:25:56 acc1_pos: 94.792 acc5_pos: 100.000 loss: 0.2214 (0.2214) acc1: 94.7917 (94.7917) acc5: 100.0000 (100.0000) time: 6.6239 data: 5.0430 max mem: 3531 +Test: [ 10/235] eta: 0:02:34 acc1_pos: 94.271 acc5_pos: 100.000 loss: 0.2349 (0.2351) acc1: 94.2708 (93.6790) acc5: 100.0000 (100.0000) time: 0.6869 data: 0.4588 max mem: 3531 +Test: [ 20/235] eta: 0:01:28 acc1_pos: 92.969 acc5_pos: 100.000 loss: 0.2332 (0.2299) acc1: 94.0104 (93.8368) acc5: 100.0000 (100.0000) time: 0.1015 data: 0.0056 max mem: 3531 +Test: [ 30/235] eta: 0:01:27 acc1_pos: 96.094 acc5_pos: 100.000 loss: 0.2193 (0.2287) acc1: 93.4896 (93.8340) acc5: 100.0000 (100.0000) time: 0.2817 data: 0.1861 max mem: 3531 +Test: [ 40/235] eta: 0:01:18 acc1_pos: 94.271 acc5_pos: 100.000 loss: 0.2268 (0.2287) acc1: 94.2708 (93.9215) acc5: 100.0000 (100.0000) time: 0.3957 data: 0.3026 max mem: 3531 +Test: [ 50/235] eta: 0:01:09 acc1_pos: 90.625 acc5_pos: 100.000 loss: 0.2268 (0.2294) acc1: 94.2708 (93.8317) acc5: 100.0000 (100.0000) time: 0.3001 data: 0.2081 max mem: 3531 +Test: [ 60/235] eta: 0:00:58 acc1_pos: 94.010 acc5_pos: 100.000 loss: 0.2368 (0.2306) acc1: 93.4896 (93.7927) acc5: 100.0000 (100.0000) time: 0.1931 data: 0.0994 max mem: 3531 +Test: [ 70/235] eta: 0:00:53 acc1_pos: 96.923 acc5_pos: 100.000 loss: 0.2503 (0.3623) acc1: 92.4479 (85.3727) acc5: 100.0000 (100.0000) acc1_neg: 24.219 acc5_neg: 100.000 time: 0.1857 data: 0.0919 max mem: 3531 +Test: [ 80/235] eta: 0:00:47 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.2821 (0.4778) acc1: 26.5625 (78.1025) acc5: 100.0000 (100.0000) acc1_neg: 26.042 acc5_neg: 100.000 time: 0.2326 data: 0.1419 max mem: 3531 +Test: [ 90/235] eta: 0:00:43 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.3020 (0.5685) acc1: 26.0417 (72.3443) acc5: 100.0000 (100.0000) acc1_neg: 28.646 acc5_neg: 100.000 time: 0.2182 data: 0.1269 max mem: 3531 +Test: [100/235] eta: 0:00:39 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.3020 (0.6390) acc1: 25.5208 (67.8527) acc5: 100.0000 (100.0000) acc1_neg: 22.396 acc5_neg: 100.000 time: 0.2137 data: 0.1191 max mem: 3531 +Test: [110/235] eta: 0:00:35 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.2823 (0.6972) acc1: 27.0833 (64.1587) acc5: 100.0000 (100.0000) acc1_neg: 30.469 acc5_neg: 100.000 time: 0.2105 data: 0.1157 max mem: 3531 +Test: [120/235] eta: 0:00:31 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.3408 (0.7628) acc1: 22.3958 (60.3715) acc5: 100.0000 (100.0000) acc1_neg: 19.792 acc5_neg: 100.000 time: 0.2067 data: 0.1144 max mem: 3531 +Test: [130/235] eta: 0:00:28 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.5028 (0.8195) acc1: 17.1875 (57.1068) acc5: 100.0000 (100.0000) acc1_neg: 17.188 acc5_neg: 100.000 time: 0.2156 data: 0.1239 max mem: 3531 +Test: [140/235] eta: 0:00:25 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.5400 (0.8713) acc1: 16.4062 (54.1777) acc5: 100.0000 (100.0000) acc1_neg: 13.802 acc5_neg: 100.000 time: 0.2336 data: 0.1413 max mem: 3531 +Test: [150/235] eta: 0:00:22 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.5323 (0.9133) acc1: 15.8854 (51.7246) acc5: 100.0000 (100.0000) acc1_neg: 16.927 acc5_neg: 100.000 time: 0.2183 data: 0.1239 max mem: 3531 +Test: [160/235] eta: 0:00:19 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.5136 (0.9508) acc1: 16.9271 (49.5665) acc5: 100.0000 (100.0000) acc1_neg: 17.448 acc5_neg: 100.000 time: 0.2005 data: 0.1076 max mem: 3531 +Test: [170/235] eta: 0:00:16 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.4513 (0.9527) acc1: 17.7083 (49.0710) acc5: 100.0000 (100.0000) acc1_neg: 40.104 acc5_neg: 100.000 time: 0.2041 data: 0.1131 max mem: 3531 +Test: [180/235] eta: 0:00:14 acc1_pos: 96.923 acc5_pos: 100.000 loss: 0.9261 (0.9510) acc1: 43.2292 (48.8260) acc5: 100.0000 (100.0000) acc1_neg: 42.708 acc5_neg: 100.000 time: 0.2016 data: 0.1112 max mem: 3531 +Test: [190/235] eta: 0:00:11 acc1_pos: 96.923 acc5_pos: 100.000 loss: 0.9063 (0.9490) acc1: 44.5312 (48.6107) acc5: 100.0000 (100.0000) acc1_neg: 46.615 acc5_neg: 100.000 time: 0.1870 data: 0.0968 max mem: 3531 +Test: [200/235] eta: 0:00:08 acc1_pos: 96.923 acc5_pos: 100.000 loss: 0.9063 (0.9477) acc1: 44.5312 (48.3637) acc5: 100.0000 (100.0000) acc1_neg: 45.573 acc5_neg: 100.000 time: 0.1611 data: 0.0708 max mem: 3531 +Test: [210/235] eta: 0:00:06 acc1_pos: 96.923 acc5_pos: 100.000 loss: 0.9234 (0.9461) acc1: 43.4896 (48.2339) acc5: 100.0000 (100.0000) acc1_neg: 45.312 acc5_neg: 100.000 time: 0.1403 data: 0.0500 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 96.923 acc5_pos: 100.000 loss: 0.9212 (0.9448) acc1: 44.7917 (48.0981) acc5: 100.0000 (100.0000) acc1_neg: 39.583 acc5_neg: 100.000 time: 0.1298 data: 0.0396 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 96.923 acc5_pos: 100.000 loss: 0.9078 (0.9430) acc1: 45.5729 (48.0012) acc5: 100.0000 (100.0000) acc1_neg: 45.052 acc5_neg: 100.000 time: 0.1198 data: 0.0302 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 96.923 acc5_pos: 100.000 loss: 0.9095 (0.9430) acc1: 45.5729 (47.9311) acc5: 100.0000 (100.0000) acc1_neg: 43.478 acc5_neg: 100.000 time: 0.1097 data: 0.0233 max mem: 3532 +Test: Total time: 0:00:53 (0.2284 s / it) +* Overall Acc@1 47.931 Acc@5 100.000 + target == real Acc@1 93.792 Acc@5 100.000 + target == fake Acc@1 31.344 Acc@5 100.000 + loss 0.943 + PR-AUC 0.8953 ROC-AUC 0.7712 EER 29.80% (thr=0.2001) +Accuracy of the network on the 89879 test images: 47.9% +[rank0]:[W726 03:19:57.919796012 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:20:17.746000 1173102 torch/distributed/run.py:766] +W0726 03:20:17.746000 1173102 torch/distributed/run.py:766] ***************************************** +W0726 03:20:17.746000 1173102 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:20:17.746000 1173102 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:20:47.731879693 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:20:47.818516090 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:20:47.081552333 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:20:47.268915873 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:20:47.344119855 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:20:47.356547135 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +[rank0]:[W726 03:20:47.396750630 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:20:47.407417626 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n23_seed5_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed5_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:30:04 acc1_pos: 90.885 acc5_pos: 100.000 loss: 0.4156 (0.4156) acc1: 90.8854 (90.8854) acc5: 100.0000 (100.0000) time: 6.7068 data: 4.9848 max mem: 3531 +Test: [ 10/269] eta: 0:03:02 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.3376 (0.3471) acc1: 93.2292 (93.6553) acc5: 100.0000 (100.0000) time: 0.7062 data: 0.4614 max mem: 3531 +Test: [ 20/269] eta: 0:01:43 acc1_pos: 97.135 acc5_pos: 100.000 loss: 0.3140 (0.3299) acc1: 95.3125 (94.7297) acc5: 100.0000 (100.0000) time: 0.1030 data: 0.0048 max mem: 3531 +Test: [ 30/269] eta: 0:01:25 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.3174 (0.3259) acc1: 96.8750 (95.1781) acc5: 100.0000 (100.0000) time: 0.1660 data: 0.0700 max mem: 3531 +Test: [ 40/269] eta: 0:01:14 acc1_pos: 89.655 acc5_pos: 100.000 loss: 0.3695 (0.4760) acc1: 90.8854 (78.7030) acc5: 100.0000 (100.0000) acc1_neg: 9.115 acc5_neg: 100.000 time: 0.2299 data: 0.1355 max mem: 3531 +Test: [ 50/269] eta: 0:01:05 acc1_pos: 89.655 acc5_pos: 100.000 loss: 1.0850 (0.6102) acc1: 11.4583 (65.3084) acc5: 100.0000 (100.0000) acc1_neg: 4.167 acc5_neg: 100.000 time: 0.2118 data: 0.1183 max mem: 3531 +Test: [ 60/269] eta: 0:01:00 acc1_pos: 89.655 acc5_pos: 100.000 loss: 1.1923 (0.7098) acc1: 8.3333 (55.9768) acc5: 100.0000 (100.0000) acc1_neg: 7.552 acc5_neg: 100.000 time: 0.2099 data: 0.1186 max mem: 3531 +Test: [ 70/269] eta: 0:00:56 acc1_pos: 89.655 acc5_pos: 100.000 loss: 1.1093 (0.7548) acc1: 11.4583 (50.5869) acc5: 100.0000 (100.0000) acc1_neg: 20.052 acc5_neg: 100.000 time: 0.2349 data: 0.1421 max mem: 3531 +Test: [ 80/269] eta: 0:00:51 acc1_pos: 89.655 acc5_pos: 100.000 loss: 0.9909 (0.7858) acc1: 18.2292 (46.8300) acc5: 100.0000 (100.0000) acc1_neg: 22.917 acc5_neg: 100.000 time: 0.2306 data: 0.1380 max mem: 3531 +Test: [ 90/269] eta: 0:00:48 acc1_pos: 89.655 acc5_pos: 100.000 loss: 1.1647 (0.8319) acc1: 15.3646 (43.1090) acc5: 100.0000 (100.0000) acc1_neg: 9.896 acc5_neg: 100.000 time: 0.2269 data: 0.1342 max mem: 3531 +Test: [100/269] eta: 0:00:44 acc1_pos: 89.655 acc5_pos: 100.000 loss: 1.1673 (0.8615) acc1: 13.0208 (40.1867) acc5: 100.0000 (100.0000) acc1_neg: 8.854 acc5_neg: 100.000 time: 0.2152 data: 0.1227 max mem: 3531 +Test: [110/269] eta: 0:00:41 acc1_pos: 89.655 acc5_pos: 100.000 loss: 1.1115 (0.8820) acc1: 13.0208 (37.9012) acc5: 100.0000 (100.0000) acc1_neg: 12.760 acc5_neg: 100.000 time: 0.2149 data: 0.1241 max mem: 3531 +Test: [120/269] eta: 0:00:38 acc1_pos: 89.655 acc5_pos: 100.000 loss: 1.0562 (0.8951) acc1: 19.2708 (36.8565) acc5: 100.0000 (100.0000) acc1_neg: 22.656 acc5_neg: 100.000 time: 0.2455 data: 0.1533 max mem: 3531 +Test: [130/269] eta: 0:00:35 acc1_pos: 89.655 acc5_pos: 100.000 loss: 1.0894 (0.9161) acc1: 19.7917 (35.4405) acc5: 100.0000 (100.0000) acc1_neg: 16.146 acc5_neg: 100.000 time: 0.2454 data: 0.1532 max mem: 3531 +Test: [140/269] eta: 0:00:32 acc1_pos: 89.655 acc5_pos: 100.000 loss: 1.2471 (0.9418) acc1: 10.4167 (33.6251) acc5: 100.0000 (100.0000) acc1_neg: 9.375 acc5_neg: 100.000 time: 0.2319 data: 0.1398 max mem: 3531 +Test: [150/269] eta: 0:00:30 acc1_pos: 89.655 acc5_pos: 100.000 loss: 1.2984 (0.9626) acc1: 8.3333 (32.2261) acc5: 100.0000 (100.0000) acc1_neg: 22.656 acc5_neg: 100.000 time: 0.2151 data: 0.1208 max mem: 3531 +Test: [160/269] eta: 0:00:27 acc1_pos: 89.655 acc5_pos: 100.000 loss: 1.1375 (0.9730) acc1: 12.7604 (31.2338) acc5: 100.0000 (100.0000) acc1_neg: 11.719 acc5_neg: 100.000 time: 0.2021 data: 0.1080 max mem: 3531 +Test: [170/269] eta: 0:00:24 acc1_pos: 89.655 acc5_pos: 100.000 loss: 1.0699 (0.9819) acc1: 16.1458 (30.2281) acc5: 100.0000 (100.0000) acc1_neg: 17.708 acc5_neg: 100.000 time: 0.2052 data: 0.1125 max mem: 3531 +Test: [180/269] eta: 0:00:21 acc1_pos: 89.655 acc5_pos: 100.000 loss: 1.0339 (0.9779) acc1: 20.5729 (30.6184) acc5: 100.0000 (100.0000) acc1_neg: 18.229 acc5_neg: 100.000 time: 0.2105 data: 0.1187 max mem: 3531 +Test: [190/269] eta: 0:00:19 acc1_pos: 89.655 acc5_pos: 100.000 loss: 0.9539 (0.9829) acc1: 26.3021 (30.1006) acc5: 100.0000 (100.0000) acc1_neg: 26.302 acc5_neg: 100.000 time: 0.2146 data: 0.1215 max mem: 3531 +Test: [200/269] eta: 0:00:16 acc1_pos: 89.655 acc5_pos: 100.000 loss: 1.1324 (0.9935) acc1: 14.3229 (29.2612) acc5: 100.0000 (100.0000) acc1_neg: 17.969 acc5_neg: 100.000 time: 0.2039 data: 0.1112 max mem: 3531 +Test: [210/269] eta: 0:00:14 acc1_pos: 89.655 acc5_pos: 100.000 loss: 1.1923 (1.0036) acc1: 12.7604 (28.4755) acc5: 100.0000 (100.0000) acc1_neg: 18.229 acc5_neg: 100.000 time: 0.1942 data: 0.1019 max mem: 3532 +Test: [220/269] eta: 0:00:11 acc1_pos: 89.655 acc5_pos: 100.000 loss: 1.1638 (1.0122) acc1: 8.3333 (27.9895) acc5: 100.0000 (100.0000) acc1_neg: 36.979 acc5_neg: 100.000 time: 0.1834 data: 0.0896 max mem: 3532 +Test: [230/269] eta: 0:00:09 acc1_pos: 89.655 acc5_pos: 100.000 loss: 1.0028 (1.0117) acc1: 21.3542 (27.9198) acc5: 100.0000 (100.0000) acc1_neg: 32.292 acc5_neg: 100.000 time: 0.1722 data: 0.0799 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 89.655 acc5_pos: 100.000 loss: 1.0539 (1.0163) acc1: 20.8333 (27.4507) acc5: 100.0000 (100.0000) acc1_neg: 20.573 acc5_neg: 100.000 time: 0.1541 data: 0.0634 max mem: 3532 +Test: [250/269] eta: 0:00:04 acc1_pos: 89.655 acc5_pos: 100.000 loss: 1.1164 (1.0218) acc1: 14.3229 (27.0263) acc5: 100.0000 (100.0000) acc1_neg: 2.604 acc5_neg: 100.000 time: 0.1296 data: 0.0392 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 89.655 acc5_pos: 100.000 loss: 1.2157 (1.0300) acc1: 8.5938 (26.3161) acc5: 100.0000 (100.0000) acc1_neg: 17.188 acc5_neg: 100.000 time: 0.1223 data: 0.0326 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 89.655 acc5_pos: 100.000 loss: 1.1958 (1.0329) acc1: 9.1146 (25.9485) acc5: 100.0000 (100.0000) acc1_neg: 20.735 acc5_neg: 100.000 time: 0.1165 data: 0.0234 max mem: 3532 +Test: Total time: 0:00:58 (0.2180 s / it) +* Overall Acc@1 25.949 Acc@5 100.000 + target == real Acc@1 95.026 Acc@5 100.000 + target == fake Acc@1 16.521 Acc@5 100.000 + loss 1.033 + PR-AUC 0.9331 ROC-AUC 0.6758 EER 37.73% (thr=0.2849) +Accuracy of the network on the 103293 test images: 25.9% +[rank0]:[W726 03:21:53.303341787 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T03:22:01Z] eval_done tag=fixed500k_real_n23_seed5_resampled +[2026-07-26T03:22:02Z] eval_start tag=fixed500k_fake_n45_seed5_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed5_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T03:22:02Z +hostname=n116-041-023 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed5_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n45_seed5_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 03:22:15.080000 1173679 torch/distributed/run.py:766] +W0726 03:22:15.080000 1173679 torch/distributed/run.py:766] ***************************************** +W0726 03:22:15.080000 1173679 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:22:15.080000 1173679 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:22:45.838773357 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:22:45.315340464 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:22:45.400275574 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:22:46.457343851 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:22:46.518653932 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:22:46.562658180 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:22:46.595500565 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W726 03:22:46.603585595 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n45_seed5_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed5_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:24:34 acc1_pos: 77.865 acc5_pos: 100.000 loss: 0.5140 (0.5140) acc1: 77.8646 (77.8646) acc5: 100.0000 (100.0000) time: 6.2993 data: 4.8770 max mem: 3531 +Test: [ 10/234] eta: 0:02:26 acc1_pos: 47.917 acc5_pos: 100.000 loss: 0.6080 (0.6660) acc1: 68.2292 (59.8011) acc5: 100.0000 (100.0000) time: 0.6546 data: 0.4437 max mem: 3531 +Test: [ 20/234] eta: 0:01:22 acc1_pos: 96.354 acc5_pos: 100.000 loss: 0.7258 (0.6675) acc1: 60.1562 (57.5149) acc5: 100.0000 (100.0000) time: 0.0918 data: 0.0004 max mem: 3531 +Test: [ 30/234] eta: 0:01:14 acc1_pos: 71.615 acc5_pos: 100.000 loss: 0.2079 (0.5324) acc1: 95.0521 (68.5904) acc5: 100.0000 (100.0000) time: 0.2069 data: 0.1148 max mem: 3531 +Test: [ 40/234] eta: 0:01:08 acc1_pos: 99.219 acc5_pos: 100.000 loss: 0.2061 (0.5106) acc1: 95.3125 (69.8361) acc5: 100.0000 (100.0000) time: 0.3128 data: 0.2217 max mem: 3531 +Test: [ 50/234] eta: 0:01:02 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1380 (0.4367) acc1: 98.4375 (75.4749) acc5: 100.0000 (100.0000) time: 0.3051 data: 0.2145 max mem: 3531 +Test: [ 60/234] eta: 0:00:55 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.1125 (0.3795) acc1: 98.9583 (79.4740) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2482 data: 0.1556 max mem: 3531 +Test: [ 70/234] eta: 0:00:50 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.0897 (0.4043) acc1: 99.7396 (78.2094) acc5: 100.0000 (100.0000) acc1_neg: 54.427 acc5_neg: 100.000 time: 0.2286 data: 0.1342 max mem: 3531 +Test: [ 80/234] eta: 0:00:45 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.7529 (0.4237) acc1: 57.2917 (77.3438) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2412 data: 0.1472 max mem: 3531 +Test: [ 90/234] eta: 0:00:41 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.2012 (0.3950) acc1: 92.4479 (79.3727) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.2084 data: 0.1158 max mem: 3531 +Test: [100/234] eta: 0:00:38 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.2280 (0.3921) acc1: 92.4479 (79.8706) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2356 data: 0.1436 max mem: 3531 +Test: [110/234] eta: 0:00:34 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.3818 (0.3893) acc1: 81.2500 (80.3655) acc5: 100.0000 (100.0000) acc1_neg: 74.219 acc5_neg: 100.000 time: 0.2498 data: 0.1574 max mem: 3531 +Test: [120/234] eta: 0:00:31 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.3595 (0.3944) acc1: 88.0208 (80.4838) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2135 data: 0.1214 max mem: 3531 +Test: [130/234] eta: 0:00:28 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.3784 (0.3947) acc1: 89.0625 (80.9896) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.2590 data: 0.1674 max mem: 3531 +Test: [140/234] eta: 0:00:25 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4103 (0.4176) acc1: 83.0729 (79.9479) acc5: 100.0000 (100.0000) acc1_neg: 41.406 acc5_neg: 100.000 time: 0.2455 data: 0.1547 max mem: 3531 +Test: [150/234] eta: 0:00:22 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.3809 (0.4208) acc1: 79.1667 (79.8772) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.1846 data: 0.0941 max mem: 3531 +Test: [160/234] eta: 0:00:19 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.3776 (0.4212) acc1: 82.5521 (80.0773) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.1968 data: 0.1067 max mem: 3531 +Test: [170/234] eta: 0:00:16 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.5291 (0.4271) acc1: 75.7812 (79.9144) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.2012 data: 0.1113 max mem: 3531 +Test: [180/234] eta: 0:00:13 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.5430 (0.4332) acc1: 74.7396 (79.5738) acc5: 100.0000 (100.0000) acc1_neg: 51.302 acc5_neg: 100.000 time: 0.1991 data: 0.1093 max mem: 3531 +Test: [190/234] eta: 0:00:11 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.7227 (0.4497) acc1: 59.3750 (78.4400) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.1933 data: 0.1035 max mem: 3531 +Test: [200/234] eta: 0:00:08 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.7227 (0.4440) acc1: 59.1146 (78.7443) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1880 data: 0.0982 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.1042 (0.4292) acc1: 99.7396 (79.6961) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1606 data: 0.0708 max mem: 3532 +Test: [220/234] eta: 0:00:03 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.1849 (0.4190) acc1: 97.6562 (80.4181) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.1350 data: 0.0453 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.3401 (0.4315) acc1: 61.4583 (79.4113) acc5: 100.0000 (100.0000) acc1_neg: 55.990 acc5_neg: 100.000 time: 0.1247 data: 0.0351 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.6894 (0.4349) acc1: 58.7549 (79.1784) acc5: 100.0000 (100.0000) acc1_neg: 58.755 acc5_neg: 100.000 time: 0.1217 data: 0.0311 max mem: 3532 +Test: Total time: 0:00:54 (0.2312 s / it) +* Overall Acc@1 79.178 Acc@5 100.000 + target == real Acc@1 75.640 Acc@5 100.000 + target == fake Acc@1 80.176 Acc@5 100.000 + loss 0.435 + PR-AUC 0.9643 ROC-AUC 0.8818 EER 21.60% (thr=0.5269) +Accuracy of the network on the 89729 test images: 79.2% +[rank0]:[W726 03:23:47.401824430 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:24:07.761000 1174240 torch/distributed/run.py:766] +W0726 03:24:07.761000 1174240 torch/distributed/run.py:766] ***************************************** +W0726 03:24:07.761000 1174240 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:24:07.761000 1174240 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:24:36.884562665 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:24:36.284916444 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:24:36.335366734 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:24:37.444746000 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:24:37.462773104 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 03:24:37.601975684 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank4]:[W726 03:24:37.609584372 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:24:37.654911822 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n45_seed5_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed5_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:55:51 acc1_pos: 28.906 acc5_pos: 100.000 loss: 0.9809 (0.9809) acc1: 28.9062 (28.9062) acc5: 100.0000 (100.0000) time: 7.4368 data: 5.9961 max mem: 3531 +Test: [ 10/2387] eta: 0:31:20 acc1_pos: 14.323 acc5_pos: 100.000 loss: 1.0883 (1.0778) acc1: 26.8229 (24.4555) acc5: 100.0000 (100.0000) time: 0.7910 data: 0.5782 max mem: 3531 +Test: [ 20/2387] eta: 0:18:07 acc1_pos: 29.948 acc5_pos: 100.000 loss: 1.0110 (1.0715) acc1: 24.7396 (24.8388) acc5: 100.0000 (100.0000) time: 0.1105 data: 0.0184 max mem: 3531 +Test: [ 30/2387] eta: 0:15:25 acc1_pos: 1.823 acc5_pos: 100.000 loss: 1.0825 (1.0951) acc1: 16.9271 (22.1858) acc5: 100.0000 (100.0000) time: 0.1739 data: 0.0815 max mem: 3531 +Test: [ 40/2387] eta: 0:15:02 acc1_pos: 10.156 acc5_pos: 100.000 loss: 1.1920 (1.1377) acc1: 10.9375 (18.6039) acc5: 100.0000 (100.0000) time: 0.3063 data: 0.2142 max mem: 3531 +Test: [ 50/2387] eta: 0:14:25 acc1_pos: 36.979 acc5_pos: 100.000 loss: 1.1799 (1.1241) acc1: 11.7188 (19.8938) acc5: 100.0000 (100.0000) time: 0.3359 data: 0.2424 max mem: 3531 +Test: [ 60/2387] eta: 0:13:20 acc1_pos: 41.406 acc5_pos: 100.000 loss: 1.0805 (1.1208) acc1: 20.0521 (20.1332) acc5: 100.0000 (100.0000) time: 0.2612 data: 0.1679 max mem: 3531 +Test: [ 70/2387] eta: 0:12:36 acc1_pos: 28.906 acc5_pos: 100.000 loss: 1.1403 (1.1347) acc1: 14.5833 (19.2268) acc5: 100.0000 (100.0000) time: 0.2149 data: 0.1222 max mem: 3531 +Test: [ 80/2387] eta: 0:12:05 acc1_pos: 62.760 acc5_pos: 100.000 loss: 1.0844 (1.1021) acc1: 24.2188 (21.6821) acc5: 100.0000 (100.0000) time: 0.2240 data: 0.1310 max mem: 3531 +Test: [ 90/2387] eta: 0:11:46 acc1_pos: 40.885 acc5_pos: 100.000 loss: 0.7949 (1.0627) acc1: 42.4479 (24.7911) acc5: 100.0000 (100.0000) time: 0.2391 data: 0.1470 max mem: 3531 +Test: [ 100/2387] eta: 0:11:25 acc1_pos: 47.656 acc5_pos: 100.000 loss: 0.7728 (1.0394) acc1: 42.4479 (26.5986) acc5: 100.0000 (100.0000) time: 0.2395 data: 0.1475 max mem: 3531 +Test: [ 110/2387] eta: 0:11:07 acc1_pos: 53.906 acc5_pos: 100.000 loss: 0.7316 (1.0131) acc1: 47.6562 (28.7279) acc5: 100.0000 (100.0000) time: 0.2281 data: 0.1347 max mem: 3531 +Test: [ 120/2387] eta: 0:10:52 acc1_pos: 37.760 acc5_pos: 100.000 loss: 0.7652 (0.9961) acc1: 46.8750 (30.0405) acc5: 100.0000 (100.0000) time: 0.2285 data: 0.1324 max mem: 3531 +Test: [ 130/2387] eta: 0:10:38 acc1_pos: 43.229 acc5_pos: 100.000 loss: 0.7933 (0.9808) acc1: 43.2292 (30.9955) acc5: 100.0000 (100.0000) time: 0.2260 data: 0.1286 max mem: 3531 +Test: [ 140/2387] eta: 0:10:25 acc1_pos: 42.969 acc5_pos: 100.000 loss: 0.7730 (0.9667) acc1: 43.2292 (32.1402) acc5: 100.0000 (100.0000) time: 0.2221 data: 0.1289 max mem: 3531 +Test: [ 150/2387] eta: 0:10:13 acc1_pos: 25.781 acc5_pos: 100.000 loss: 0.7502 (0.9526) acc1: 47.3958 (33.1333) acc5: 100.0000 (100.0000) time: 0.2169 data: 0.1252 max mem: 3531 +Test: [ 160/2387] eta: 0:09:58 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.7233 (0.9297) acc1: 49.7396 (35.0835) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2008 data: 0.1068 max mem: 3531 +Test: [ 170/2387] eta: 0:09:45 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2388 (0.8887) acc1: 97.1354 (38.7747) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1875 data: 0.0933 max mem: 3531 +Test: [ 180/2387] eta: 0:09:36 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2291 (0.8521) acc1: 97.9167 (42.0422) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1996 data: 0.1065 max mem: 3531 +Test: [ 190/2387] eta: 0:09:20 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2259 (0.8193) acc1: 97.9167 (44.9839) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1792 data: 0.0853 max mem: 3531 +Test: [ 200/2387] eta: 0:09:08 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2227 (0.7897) acc1: 98.4375 (47.6303) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1571 data: 0.0625 max mem: 3531 +Test: [ 210/2387] eta: 0:08:59 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2257 (0.7633) acc1: 98.1771 (50.0074) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1799 data: 0.0865 max mem: 3532 +Test: [ 220/2387] eta: 0:08:53 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2296 (0.7390) acc1: 98.1771 (52.1976) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2006 data: 0.1091 max mem: 3532 +Test: [ 230/2387] eta: 0:08:47 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2294 (0.7169) acc1: 98.4375 (54.1892) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2099 data: 0.1163 max mem: 3532 +Test: [ 240/2387] eta: 0:08:37 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2335 (0.6994) acc1: 96.6146 (55.8999) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1850 data: 0.0898 max mem: 3532 +Test: [ 250/2387] eta: 0:08:30 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2227 (0.6759) acc1: 96.3542 (57.6372) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1715 data: 0.0792 max mem: 3532 +Test: [ 260/2387] eta: 0:08:22 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.0829 (0.6530) acc1: 100.0000 (59.2603) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1744 data: 0.0835 max mem: 3532 +Test: [ 270/2387] eta: 0:08:17 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.0762 (0.6317) acc1: 100.0000 (60.7626) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1853 data: 0.0934 max mem: 3532 +Test: [ 280/2387] eta: 0:08:15 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.0800 (0.6122) acc1: 100.0000 (62.1590) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2247 data: 0.1318 max mem: 3532 +Test: [ 290/2387] eta: 0:08:05 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.0814 (0.5940) acc1: 100.0000 (63.4593) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1890 data: 0.0958 max mem: 3532 +Test: [ 300/2387] eta: 0:08:01 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.0866 (0.5772) acc1: 100.0000 (64.6733) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1657 data: 0.0713 max mem: 3532 +Test: [ 310/2387] eta: 0:08:01 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.0924 (0.5650) acc1: 100.0000 (65.7682) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2357 data: 0.1417 max mem: 3532 +Test: [ 320/2387] eta: 0:07:56 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.3131 (0.5573) acc1: 98.4375 (66.7738) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2247 data: 0.1317 max mem: 3532 +Test: [ 330/2387] eta: 0:07:52 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.3163 (0.5499) acc1: 98.4375 (67.7335) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1915 data: 0.0976 max mem: 3532 +Test: [ 340/2387] eta: 0:07:48 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.3131 (0.5430) acc1: 98.4375 (68.6286) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2093 data: 0.1144 max mem: 3532 +Test: [ 350/2387] eta: 0:07:43 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.3069 (0.5323) acc1: 98.6979 (69.5164) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1984 data: 0.1053 max mem: 3532 +Test: [ 360/2387] eta: 0:07:39 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1836 (0.5227) acc1: 99.7396 (70.3421) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1828 data: 0.0909 max mem: 3532 +Test: [ 370/2387] eta: 0:07:34 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1916 (0.5139) acc1: 99.2188 (71.1183) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1833 data: 0.0916 max mem: 3532 +Test: [ 380/2387] eta: 0:07:30 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1930 (0.5055) acc1: 98.9583 (71.8497) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1850 data: 0.0937 max mem: 3532 +Test: [ 390/2387] eta: 0:07:25 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1930 (0.4975) acc1: 98.9583 (72.5450) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1763 data: 0.0854 max mem: 3532 +Test: [ 400/2387] eta: 0:07:20 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1942 (0.4900) acc1: 98.9583 (73.2024) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1681 data: 0.0769 max mem: 3532 +Test: [ 410/2387] eta: 0:07:15 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1915 (0.4827) acc1: 99.2188 (73.8386) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1739 data: 0.0822 max mem: 3532 +Test: [ 420/2387] eta: 0:07:11 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1906 (0.4758) acc1: 99.2188 (74.4427) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1762 data: 0.0829 max mem: 3532 +Test: [ 430/2387] eta: 0:07:07 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1954 (0.4703) acc1: 98.9583 (74.9873) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1760 data: 0.0815 max mem: 3532 +Test: [ 440/2387] eta: 0:07:03 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1970 (0.4631) acc1: 97.6562 (75.5285) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1781 data: 0.0859 max mem: 3533 +Test: [ 450/2387] eta: 0:07:01 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.0823 (0.4546) acc1: 100.0000 (76.0711) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1954 data: 0.1027 max mem: 3533 +Test: [ 460/2387] eta: 0:06:58 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.0816 (0.4465) acc1: 100.0000 (76.5902) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2069 data: 0.1131 max mem: 3533 +Test: [ 470/2387] eta: 0:06:55 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.0818 (0.4388) acc1: 100.0000 (77.0872) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2059 data: 0.1119 max mem: 3533 +Test: [ 480/2387] eta: 0:06:52 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.0830 (0.4314) acc1: 100.0000 (77.5636) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2030 data: 0.1080 max mem: 3533 +Test: [ 490/2387] eta: 0:06:50 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.0830 (0.4243) acc1: 100.0000 (78.0205) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2016 data: 0.1082 max mem: 3533 +Test: [ 500/2387] eta: 0:06:47 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.0829 (0.4175) acc1: 100.0000 (78.4592) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2061 data: 0.1129 max mem: 3533 +Test: [ 510/2387] eta: 0:06:45 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.0829 (0.4112) acc1: 100.0000 (78.8808) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2049 data: 0.1107 max mem: 3533 +Test: [ 520/2387] eta: 0:06:42 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1206 (0.4059) acc1: 100.0000 (79.2861) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2028 data: 0.1103 max mem: 3533 +Test: [ 530/2387] eta: 0:06:39 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1211 (0.4005) acc1: 100.0000 (79.6762) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2009 data: 0.1097 max mem: 3533 +Test: [ 540/2387] eta: 0:06:36 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1201 (0.3954) acc1: 100.0000 (80.0519) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1979 data: 0.1069 max mem: 3533 +Test: [ 550/2387] eta: 0:06:33 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1202 (0.3904) acc1: 100.0000 (80.4139) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1893 data: 0.0984 max mem: 3533 +Test: [ 560/2387] eta: 0:06:30 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1196 (0.3855) acc1: 100.0000 (80.7631) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1776 data: 0.0864 max mem: 3533 +Test: [ 570/2387] eta: 0:06:27 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1198 (0.3809) acc1: 100.0000 (81.1000) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1766 data: 0.0840 max mem: 3533 +Test: [ 580/2387] eta: 0:06:23 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1205 (0.3764) acc1: 100.0000 (81.4253) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1807 data: 0.0886 max mem: 3533 +Test: [ 590/2387] eta: 0:06:20 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1210 (0.3721) acc1: 100.0000 (81.7395) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1731 data: 0.0819 max mem: 3533 +Test: [ 600/2387] eta: 0:06:17 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1241 (0.3682) acc1: 100.0000 (82.0434) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1757 data: 0.0841 max mem: 3533 +Test: [ 610/2387] eta: 0:06:13 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1439 (0.3647) acc1: 100.0000 (82.3347) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1679 data: 0.0758 max mem: 3533 +Test: [ 620/2387] eta: 0:06:10 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1508 (0.3613) acc1: 100.0000 (82.6171) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1635 data: 0.0714 max mem: 3533 +Test: [ 630/2387] eta: 0:06:07 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1509 (0.3579) acc1: 100.0000 (82.8909) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1745 data: 0.0834 max mem: 3533 +Test: [ 640/2387] eta: 0:06:05 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1529 (0.3548) acc1: 100.0000 (83.1554) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1960 data: 0.1016 max mem: 3533 +Test: [ 650/2387] eta: 0:06:02 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1540 (0.3517) acc1: 100.0000 (83.4125) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1945 data: 0.0996 max mem: 3533 +Test: [ 660/2387] eta: 0:06:00 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1520 (0.3487) acc1: 100.0000 (83.6615) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1982 data: 0.1062 max mem: 3533 +Test: [ 670/2387] eta: 0:05:58 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1519 (0.3457) acc1: 100.0000 (83.9035) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2186 data: 0.1260 max mem: 3534 +Test: [ 680/2387] eta: 0:05:56 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1563 (0.3433) acc1: 100.0000 (84.1398) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2117 data: 0.1192 max mem: 3534 +Test: [ 690/2387] eta: 0:05:55 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1966 (0.3459) acc1: 100.0000 (83.9947) acc5: 100.0000 (100.0000) acc1_neg: 61.719 acc5_neg: 100.000 time: 0.2186 data: 0.1266 max mem: 3534 +Test: [ 700/2387] eta: 0:05:52 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.6317 (0.3501) acc1: 65.1042 (83.7275) acc5: 100.0000 (100.0000) acc1_neg: 68.490 acc5_neg: 100.000 time: 0.2115 data: 0.1191 max mem: 3534 +Test: [ 710/2387] eta: 0:05:51 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.6330 (0.3542) acc1: 65.1042 (83.4645) acc5: 100.0000 (100.0000) acc1_neg: 64.323 acc5_neg: 100.000 time: 0.2122 data: 0.1199 max mem: 3534 +Test: [ 720/2387] eta: 0:05:48 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.6324 (0.3580) acc1: 65.1042 (83.2156) acc5: 100.0000 (100.0000) acc1_neg: 66.146 acc5_neg: 100.000 time: 0.2080 data: 0.1148 max mem: 3534 +Test: [ 730/2387] eta: 0:05:46 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.6407 (0.3620) acc1: 64.5833 (82.9529) acc5: 100.0000 (100.0000) acc1_neg: 59.115 acc5_neg: 100.000 time: 0.1975 data: 0.1044 max mem: 3534 +Test: [ 740/2387] eta: 0:05:44 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.6451 (0.3657) acc1: 64.0625 (82.7085) acc5: 100.0000 (100.0000) acc1_neg: 61.719 acc5_neg: 100.000 time: 0.2104 data: 0.1183 max mem: 3534 +Test: [ 750/2387] eta: 0:05:42 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.6281 (0.3693) acc1: 64.8438 (82.4758) acc5: 100.0000 (100.0000) acc1_neg: 66.927 acc5_neg: 100.000 time: 0.2178 data: 0.1243 max mem: 3534 +Test: [ 760/2387] eta: 0:05:39 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.6236 (0.3718) acc1: 66.9271 (82.3252) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.2000 data: 0.1073 max mem: 3534 +Test: [ 770/2387] eta: 0:05:37 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.4400 (0.3711) acc1: 79.1667 (82.4099) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1872 data: 0.0965 max mem: 3534 +Test: [ 780/2387] eta: 0:05:34 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1998 (0.3688) acc1: 97.1354 (82.6078) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1887 data: 0.0974 max mem: 3534 +Test: [ 790/2387] eta: 0:05:32 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1750 (0.3666) acc1: 98.1771 (82.7769) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1854 data: 0.0936 max mem: 3534 +Test: [ 800/2387] eta: 0:05:29 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1549 (0.3642) acc1: 98.4375 (82.9760) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1863 data: 0.0948 max mem: 3534 +Test: [ 810/2387] eta: 0:05:27 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1658 (0.3619) acc1: 98.9583 (83.1689) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1870 data: 0.0955 max mem: 3534 +Test: [ 820/2387] eta: 0:05:24 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1714 (0.3600) acc1: 98.9583 (83.3495) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1839 data: 0.0924 max mem: 3534 +Test: [ 830/2387] eta: 0:05:22 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2429 (0.3594) acc1: 96.0938 (83.4490) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1786 data: 0.0879 max mem: 3534 +Test: [ 840/2387] eta: 0:05:19 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2643 (0.3580) acc1: 95.8333 (83.6139) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1812 data: 0.0894 max mem: 3534 +Test: [ 850/2387] eta: 0:05:17 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2296 (0.3566) acc1: 97.3958 (83.7743) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1844 data: 0.0913 max mem: 3534 +Test: [ 860/2387] eta: 0:05:14 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2219 (0.3549) acc1: 98.1771 (83.9488) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1918 data: 0.0988 max mem: 3534 +Test: [ 870/2387] eta: 0:05:12 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2153 (0.3534) acc1: 98.9583 (84.1128) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1895 data: 0.0954 max mem: 3534 +Test: [ 880/2387] eta: 0:05:10 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2250 (0.3519) acc1: 98.6979 (84.2745) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1914 data: 0.0982 max mem: 3534 +Test: [ 890/2387] eta: 0:05:08 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2201 (0.3503) acc1: 99.2188 (84.4375) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1994 data: 0.1080 max mem: 3534 +Test: [ 900/2387] eta: 0:05:05 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2504 (0.3537) acc1: 95.3125 (84.2319) acc5: 100.0000 (100.0000) acc1_neg: 60.938 acc5_neg: 100.000 time: 0.1936 data: 0.1026 max mem: 3535 +Test: [ 910/2387] eta: 0:05:03 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.7586 (0.3582) acc1: 58.0729 (83.9428) acc5: 100.0000 (100.0000) acc1_neg: 57.812 acc5_neg: 100.000 time: 0.1949 data: 0.0996 max mem: 3535 +Test: [ 920/2387] eta: 0:05:01 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.7652 (0.3627) acc1: 57.5521 (83.6582) acc5: 100.0000 (100.0000) acc1_neg: 56.510 acc5_neg: 100.000 time: 0.1947 data: 0.0981 max mem: 3535 +Test: [ 930/2387] eta: 0:04:59 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.7652 (0.3669) acc1: 58.3333 (83.3879) acc5: 100.0000 (100.0000) acc1_neg: 51.823 acc5_neg: 100.000 time: 0.2022 data: 0.1079 max mem: 3535 +Test: [ 940/2387] eta: 0:04:57 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.7624 (0.3711) acc1: 58.0729 (83.1103) acc5: 100.0000 (100.0000) acc1_neg: 53.385 acc5_neg: 100.000 time: 0.2199 data: 0.1245 max mem: 3535 +Test: [ 950/2387] eta: 0:04:54 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.7587 (0.3753) acc1: 57.5521 (82.8336) acc5: 100.0000 (100.0000) acc1_neg: 57.552 acc5_neg: 100.000 time: 0.1888 data: 0.0934 max mem: 3535 +Test: [ 960/2387] eta: 0:04:52 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.7498 (0.3792) acc1: 58.5938 (82.5830) acc5: 100.0000 (100.0000) acc1_neg: 60.156 acc5_neg: 100.000 time: 0.1849 data: 0.0872 max mem: 3535 +Test: [ 970/2387] eta: 0:04:50 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.7498 (0.3831) acc1: 58.5938 (82.3236) acc5: 100.0000 (100.0000) acc1_neg: 61.979 acc5_neg: 100.000 time: 0.2037 data: 0.1072 max mem: 3535 +Test: [ 980/2387] eta: 0:04:48 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.7343 (0.3832) acc1: 58.8542 (82.3371) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.1909 data: 0.0956 max mem: 3535 +Test: [ 990/2387] eta: 0:04:46 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2498 (0.3820) acc1: 93.2292 (82.4514) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2020 data: 0.1070 max mem: 3535 +Test: [1000/2387] eta: 0:04:44 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2521 (0.3811) acc1: 93.7500 (82.5536) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1964 data: 0.1031 max mem: 3535 +Test: [1010/2387] eta: 0:04:41 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2897 (0.3804) acc1: 92.7083 (82.6319) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.1891 data: 0.0967 max mem: 3535 +Test: [1020/2387] eta: 0:04:39 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.3127 (0.3794) acc1: 90.1042 (82.7230) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1912 data: 0.1004 max mem: 3535 +Test: [1030/2387] eta: 0:04:37 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1913 (0.3776) acc1: 97.6562 (82.8767) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1819 data: 0.0892 max mem: 3535 +Test: [1040/2387] eta: 0:04:34 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1765 (0.3755) acc1: 99.2188 (83.0339) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1775 data: 0.0851 max mem: 3535 +Test: [1050/2387] eta: 0:04:32 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1679 (0.3735) acc1: 99.7396 (83.1899) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1819 data: 0.0913 max mem: 3535 +Test: [1060/2387] eta: 0:04:30 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1621 (0.3717) acc1: 99.7396 (83.3412) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1894 data: 0.0984 max mem: 3535 +Test: [1070/2387] eta: 0:04:28 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1785 (0.3701) acc1: 99.2188 (83.4783) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1896 data: 0.0976 max mem: 3535 +Test: [1080/2387] eta: 0:04:25 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2353 (0.3691) acc1: 97.1354 (83.5909) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1889 data: 0.0970 max mem: 3535 +Test: [1090/2387] eta: 0:04:23 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2370 (0.3678) acc1: 95.8333 (83.7016) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1877 data: 0.0952 max mem: 3535 +Test: [1100/2387] eta: 0:04:21 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2156 (0.3664) acc1: 96.6146 (83.8222) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1899 data: 0.0959 max mem: 3535 +Test: [1110/2387] eta: 0:04:19 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2117 (0.3653) acc1: 96.6146 (83.9160) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1889 data: 0.0959 max mem: 3535 +Test: [1120/2387] eta: 0:04:17 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2146 (0.3640) acc1: 96.6146 (84.0251) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1850 data: 0.0931 max mem: 3536 +Test: [1130/2387] eta: 0:04:15 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2085 (0.3628) acc1: 96.8750 (84.1335) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2031 data: 0.1111 max mem: 3536 +Test: [1140/2387] eta: 0:04:13 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2663 (0.3623) acc1: 95.8333 (84.2193) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2048 data: 0.1129 max mem: 3536 +Test: [1150/2387] eta: 0:04:10 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2753 (0.3612) acc1: 95.0521 (84.3202) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1956 data: 0.1038 max mem: 3536 +Test: [1160/2387] eta: 0:04:09 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1848 (0.3595) acc1: 98.9583 (84.4439) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2106 data: 0.1188 max mem: 3536 +Test: [1170/2387] eta: 0:04:06 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1577 (0.3578) acc1: 98.9583 (84.5703) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2042 data: 0.1084 max mem: 3536 +Test: [1180/2387] eta: 0:04:05 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1577 (0.3562) acc1: 99.4792 (84.6925) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2217 data: 0.1256 max mem: 3536 +Test: [1190/2387] eta: 0:04:02 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1574 (0.3546) acc1: 99.4792 (84.8171) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1901 data: 0.0960 max mem: 3536 +Test: [1200/2387] eta: 0:04:00 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1472 (0.3528) acc1: 99.7396 (84.9425) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1755 data: 0.0801 max mem: 3536 +Test: [1210/2387] eta: 0:03:58 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1430 (0.3511) acc1: 100.0000 (85.0659) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2083 data: 0.1146 max mem: 3536 +Test: [1220/2387] eta: 0:03:56 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1006 (0.3490) acc1: 100.0000 (85.1872) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1968 data: 0.1025 max mem: 3536 +Test: [1230/2387] eta: 0:03:54 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.0850 (0.3468) acc1: 100.0000 (85.3071) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1964 data: 0.1024 max mem: 3536 +Test: [1240/2387] eta: 0:03:52 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.0825 (0.3447) acc1: 100.0000 (85.4255) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1931 data: 0.0995 max mem: 3536 +Test: [1250/2387] eta: 0:03:50 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.0826 (0.3427) acc1: 100.0000 (85.5418) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1864 data: 0.0908 max mem: 3536 +Test: [1260/2387] eta: 0:03:48 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.0826 (0.3406) acc1: 100.0000 (85.6562) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1825 data: 0.0882 max mem: 3536 +Test: [1270/2387] eta: 0:03:45 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.0901 (0.3387) acc1: 100.0000 (85.7691) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1851 data: 0.0917 max mem: 3536 +Test: [1280/2387] eta: 0:03:43 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1034 (0.3369) acc1: 100.0000 (85.8785) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1953 data: 0.1007 max mem: 3536 +Test: [1290/2387] eta: 0:03:41 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1034 (0.3351) acc1: 99.7396 (85.9752) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2036 data: 0.1101 max mem: 3536 +Test: [1300/2387] eta: 0:03:39 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1004 (0.3334) acc1: 99.7396 (86.0766) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1851 data: 0.0938 max mem: 3536 +Test: [1310/2387] eta: 0:03:37 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1089 (0.3318) acc1: 99.7396 (86.1731) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1793 data: 0.0871 max mem: 3536 +Test: [1320/2387] eta: 0:03:35 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1221 (0.3302) acc1: 99.2188 (86.2697) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1958 data: 0.1028 max mem: 3536 +Test: [1330/2387] eta: 0:03:33 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1281 (0.3290) acc1: 99.4792 (86.3632) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2069 data: 0.1113 max mem: 3536 +Test: [1340/2387] eta: 0:03:31 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1815 (0.3287) acc1: 98.4375 (86.3909) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2048 data: 0.1092 max mem: 3536 +Test: [1350/2387] eta: 0:03:29 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.0644 (0.3267) acc1: 100.0000 (86.4917) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2081 data: 0.1147 max mem: 3537 +Test: [1360/2387] eta: 0:03:27 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.0455 (0.3246) acc1: 100.0000 (86.5909) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2129 data: 0.1196 max mem: 3537 +Test: [1370/2387] eta: 0:03:25 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.0445 (0.3226) acc1: 100.0000 (86.6887) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0937 max mem: 3537 +Test: [1380/2387] eta: 0:03:22 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.0483 (0.3206) acc1: 100.0000 (86.7851) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1617 data: 0.0695 max mem: 3537 +Test: [1390/2387] eta: 0:03:20 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.0551 (0.3187) acc1: 100.0000 (86.8801) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1647 data: 0.0716 max mem: 3537 +Test: [1400/2387] eta: 0:03:18 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.0553 (0.3169) acc1: 100.0000 (86.9738) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1600 data: 0.0679 max mem: 3537 +Test: [1410/2387] eta: 0:03:16 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.0622 (0.3156) acc1: 100.0000 (87.0543) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1638 data: 0.0711 max mem: 3537 +Test: [1420/2387] eta: 0:03:13 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1554 (0.3147) acc1: 99.4792 (87.1243) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1687 data: 0.0767 max mem: 3537 +Test: [1430/2387] eta: 0:03:11 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1697 (0.3137) acc1: 98.6979 (87.2037) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1624 data: 0.0710 max mem: 3537 +Test: [1440/2387] eta: 0:03:09 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1787 (0.3128) acc1: 99.2188 (87.2875) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1696 data: 0.0775 max mem: 3537 +Test: [1450/2387] eta: 0:03:07 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1852 (0.3121) acc1: 99.4792 (87.3577) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1688 data: 0.0770 max mem: 3537 +Test: [1460/2387] eta: 0:03:04 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1694 (0.3110) acc1: 98.4375 (87.4362) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1581 data: 0.0658 max mem: 3537 +Test: [1470/2387] eta: 0:03:02 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1288 (0.3097) acc1: 99.7396 (87.5204) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1635 data: 0.0713 max mem: 3537 +Test: [1480/2387] eta: 0:03:00 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1223 (0.3085) acc1: 99.7396 (87.6016) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1581 data: 0.0670 max mem: 3537 +Test: [1490/2387] eta: 0:02:58 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1285 (0.3073) acc1: 99.7396 (87.6825) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1539 data: 0.0634 max mem: 3537 +Test: [1500/2387] eta: 0:02:55 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1307 (0.3061) acc1: 99.7396 (87.7632) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1648 data: 0.0725 max mem: 3537 +Test: [1510/2387] eta: 0:02:53 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1337 (0.3050) acc1: 100.0000 (87.8442) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1642 data: 0.0701 max mem: 3537 +Test: [1520/2387] eta: 0:02:51 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1644 (0.3043) acc1: 100.0000 (87.9154) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1535 data: 0.0611 max mem: 3537 +Test: [1530/2387] eta: 0:02:49 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1932 (0.3037) acc1: 98.6979 (87.9802) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1854 data: 0.0941 max mem: 3537 +Test: [1540/2387] eta: 0:02:47 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1857 (0.3029) acc1: 98.6979 (88.0529) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2008 data: 0.1092 max mem: 3537 +Test: [1550/2387] eta: 0:02:45 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1834 (0.3021) acc1: 99.4792 (88.1234) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1896 data: 0.0979 max mem: 3537 +Test: [1560/2387] eta: 0:02:43 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1812 (0.3014) acc1: 99.4792 (88.1940) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2036 data: 0.1105 max mem: 3537 +Test: [1570/2387] eta: 0:02:41 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1749 (0.3006) acc1: 99.7396 (88.2660) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2056 data: 0.1124 max mem: 3537 +Test: [1580/2387] eta: 0:02:39 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1749 (0.2998) acc1: 99.7396 (88.3345) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2081 data: 0.1168 max mem: 3538 +Test: [1590/2387] eta: 0:02:37 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1563 (0.2989) acc1: 99.4792 (88.4047) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2036 data: 0.1120 max mem: 3538 +Test: [1600/2387] eta: 0:02:36 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1504 (0.2981) acc1: 99.2188 (88.4530) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2117 data: 0.1170 max mem: 3538 +Test: [1610/2387] eta: 0:02:34 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1512 (0.2973) acc1: 99.4792 (88.5208) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2110 data: 0.1152 max mem: 3538 +Test: [1620/2387] eta: 0:02:32 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1594 (0.2964) acc1: 99.4792 (88.5852) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2020 data: 0.1094 max mem: 3538 +Test: [1630/2387] eta: 0:02:30 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1709 (0.2959) acc1: 98.4375 (88.6336) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1970 data: 0.1049 max mem: 3538 +Test: [1640/2387] eta: 0:02:27 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2183 (0.2956) acc1: 97.6562 (88.6778) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1757 data: 0.0826 max mem: 3538 +Test: [1650/2387] eta: 0:02:25 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2188 (0.2952) acc1: 97.6562 (88.7308) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1738 data: 0.0789 max mem: 3538 +Test: [1660/2387] eta: 0:02:23 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2188 (0.2948) acc1: 97.6562 (88.7794) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1789 data: 0.0848 max mem: 3538 +Test: [1670/2387] eta: 0:02:21 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2231 (0.2944) acc1: 97.1354 (88.8270) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1773 data: 0.0860 max mem: 3538 +Test: [1680/2387] eta: 0:02:19 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2195 (0.2941) acc1: 96.8750 (88.8727) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1818 data: 0.0902 max mem: 3538 +Test: [1690/2387] eta: 0:02:17 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2315 (0.2938) acc1: 97.1354 (88.9202) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1953 data: 0.1039 max mem: 3538 +Test: [1700/2387] eta: 0:02:15 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2315 (0.2932) acc1: 97.9167 (88.9743) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1910 data: 0.0998 max mem: 3538 +Test: [1710/2387] eta: 0:02:13 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1212 (0.2921) acc1: 99.2188 (89.0368) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1746 data: 0.0833 max mem: 3538 +Test: [1720/2387] eta: 0:02:11 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1062 (0.2910) acc1: 100.0000 (89.0996) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1876 data: 0.0965 max mem: 3538 +Test: [1730/2387] eta: 0:02:09 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1049 (0.2900) acc1: 100.0000 (89.1613) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1892 data: 0.0983 max mem: 3538 +Test: [1740/2387] eta: 0:02:07 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1049 (0.2890) acc1: 100.0000 (89.2227) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1820 data: 0.0906 max mem: 3538 +Test: [1750/2387] eta: 0:02:05 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1037 (0.2879) acc1: 100.0000 (89.2837) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1931 data: 0.1002 max mem: 3538 +Test: [1760/2387] eta: 0:02:03 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1118 (0.2870) acc1: 100.0000 (89.3442) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2040 data: 0.1108 max mem: 3538 +Test: [1770/2387] eta: 0:02:01 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1358 (0.2863) acc1: 99.7396 (89.3956) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2117 data: 0.1200 max mem: 3538 +Test: [1780/2387] eta: 0:01:59 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1452 (0.2855) acc1: 99.2188 (89.4526) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2072 data: 0.1163 max mem: 3538 +Test: [1790/2387] eta: 0:01:57 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1387 (0.2846) acc1: 99.7396 (89.5087) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2003 data: 0.1091 max mem: 3538 +Test: [1800/2387] eta: 0:01:55 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1365 (0.2838) acc1: 99.4792 (89.5635) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1891 data: 0.0952 max mem: 3538 +Test: [1810/2387] eta: 0:01:53 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1365 (0.2829) acc1: 99.7396 (89.6199) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1763 data: 0.0816 max mem: 3539 +Test: [1820/2387] eta: 0:01:51 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1184 (0.2821) acc1: 100.0000 (89.6756) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1652 data: 0.0732 max mem: 3539 +Test: [1830/2387] eta: 0:01:49 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1332 (0.2813) acc1: 100.0000 (89.7308) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1561 data: 0.0635 max mem: 3539 +Test: [1840/2387] eta: 0:01:47 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1152 (0.2804) acc1: 100.0000 (89.7866) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1612 data: 0.0661 max mem: 3539 +Test: [1850/2387] eta: 0:01:45 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1141 (0.2795) acc1: 100.0000 (89.8418) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1709 data: 0.0770 max mem: 3539 +Test: [1860/2387] eta: 0:01:43 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1137 (0.2786) acc1: 100.0000 (89.8964) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1967 data: 0.1042 max mem: 3539 +Test: [1870/2387] eta: 0:01:41 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1152 (0.2777) acc1: 100.0000 (89.9504) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2138 data: 0.1209 max mem: 3539 +Test: [1880/2387] eta: 0:01:39 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1147 (0.2769) acc1: 100.0000 (90.0038) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2205 data: 0.1266 max mem: 3539 +Test: [1890/2387] eta: 0:01:37 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1143 (0.2760) acc1: 100.0000 (90.0567) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2075 data: 0.1149 max mem: 3539 +Test: [1900/2387] eta: 0:01:35 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1153 (0.2752) acc1: 100.0000 (90.1087) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1922 data: 0.1009 max mem: 3539 +Test: [1910/2387] eta: 0:01:33 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1539 (0.2746) acc1: 100.0000 (90.1591) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2073 data: 0.1163 max mem: 3539 +Test: [1920/2387] eta: 0:01:31 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1726 (0.2742) acc1: 99.7396 (90.1879) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2121 data: 0.1210 max mem: 3539 +Test: [1930/2387] eta: 0:01:29 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1382 (0.2735) acc1: 100.0000 (90.2382) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2003 data: 0.1091 max mem: 3539 +Test: [1940/2387] eta: 0:01:28 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1358 (0.2731) acc1: 100.0000 (90.2527) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1923 data: 0.1019 max mem: 3539 +Test: [1950/2387] eta: 0:01:26 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1619 (0.2726) acc1: 99.4792 (90.2942) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1966 data: 0.1057 max mem: 3539 +Test: [1960/2387] eta: 0:01:24 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1612 (0.2721) acc1: 99.2188 (90.3392) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2020 data: 0.1106 max mem: 3539 +Test: [1970/2387] eta: 0:01:22 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1989 (0.2721) acc1: 97.9167 (90.3548) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.2051 data: 0.1137 max mem: 3539 +Test: [1980/2387] eta: 0:01:20 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2431 (0.2717) acc1: 97.1354 (90.3854) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1997 data: 0.1086 max mem: 3539 +Test: [1990/2387] eta: 0:01:18 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1346 (0.2711) acc1: 98.6979 (90.4265) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1973 data: 0.1027 max mem: 3539 +Test: [2000/2387] eta: 0:01:16 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1342 (0.2704) acc1: 98.6979 (90.4683) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2068 data: 0.1105 max mem: 3539 +Test: [2010/2387] eta: 0:01:14 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1327 (0.2697) acc1: 98.9583 (90.5088) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1951 data: 0.1024 max mem: 3539 +Test: [2020/2387] eta: 0:01:12 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1331 (0.2690) acc1: 98.6979 (90.5491) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1800 data: 0.0883 max mem: 3539 +Test: [2030/2387] eta: 0:01:10 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1355 (0.2684) acc1: 98.6979 (90.5882) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1900 data: 0.0968 max mem: 3540 +Test: [2040/2387] eta: 0:01:08 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1336 (0.2677) acc1: 98.6979 (90.6286) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1936 data: 0.0969 max mem: 3540 +Test: [2050/2387] eta: 0:01:06 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1319 (0.2671) acc1: 98.9583 (90.6697) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1905 data: 0.0946 max mem: 3540 +Test: [2060/2387] eta: 0:01:04 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1383 (0.2665) acc1: 98.9583 (90.7090) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1886 data: 0.0960 max mem: 3540 +Test: [2070/2387] eta: 0:01:02 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1399 (0.2658) acc1: 98.6979 (90.7490) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1807 data: 0.0873 max mem: 3540 +Test: [2080/2387] eta: 0:01:00 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1374 (0.2652) acc1: 98.9583 (90.7874) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1849 data: 0.0917 max mem: 3540 +Test: [2090/2387] eta: 0:00:58 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1387 (0.2646) acc1: 98.9583 (90.8269) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2014 data: 0.1077 max mem: 3540 +Test: [2100/2387] eta: 0:00:56 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1387 (0.2640) acc1: 98.9583 (90.8663) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1949 data: 0.1013 max mem: 3540 +Test: [2110/2387] eta: 0:00:54 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1373 (0.2634) acc1: 99.2188 (90.9050) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1918 data: 0.0977 max mem: 3540 +Test: [2120/2387] eta: 0:00:52 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1454 (0.2631) acc1: 98.6979 (90.9369) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1825 data: 0.0886 max mem: 3540 +Test: [2130/2387] eta: 0:00:50 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1638 (0.2626) acc1: 98.9583 (90.9735) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1616 data: 0.0704 max mem: 3540 +Test: [2140/2387] eta: 0:00:48 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1640 (0.2624) acc1: 98.9583 (90.9956) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1673 data: 0.0761 max mem: 3540 +Test: [2150/2387] eta: 0:00:46 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1819 (0.2620) acc1: 98.9583 (91.0338) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1774 data: 0.0865 max mem: 3540 +Test: [2160/2387] eta: 0:00:44 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1626 (0.2616) acc1: 99.4792 (91.0698) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0891 max mem: 3540 +Test: [2170/2387] eta: 0:00:42 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1972 (0.2615) acc1: 98.1771 (91.0959) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1751 data: 0.0838 max mem: 3540 +Test: [2180/2387] eta: 0:00:40 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2740 (0.2628) acc1: 93.7500 (91.0135) acc5: 100.0000 (100.0000) acc1_neg: 42.708 acc5_neg: 100.000 time: 0.1711 data: 0.0790 max mem: 3540 +Test: [2190/2387] eta: 0:00:38 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2780 (0.2630) acc1: 91.9271 (91.0004) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1710 data: 0.0781 max mem: 3540 +Test: [2200/2387] eta: 0:00:36 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1536 (0.2625) acc1: 98.9583 (91.0391) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1796 data: 0.0865 max mem: 3540 +Test: [2210/2387] eta: 0:00:34 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1308 (0.2618) acc1: 99.7396 (91.0779) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1884 data: 0.0948 max mem: 3540 +Test: [2220/2387] eta: 0:00:32 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1278 (0.2613) acc1: 100.0000 (91.1146) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1759 data: 0.0828 max mem: 3540 +Test: [2230/2387] eta: 0:00:30 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1314 (0.2607) acc1: 100.0000 (91.1533) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1704 data: 0.0790 max mem: 3540 +Test: [2240/2387] eta: 0:00:28 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1275 (0.2601) acc1: 100.0000 (91.1905) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1612 data: 0.0693 max mem: 3540 +Test: [2250/2387] eta: 0:00:26 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1455 (0.2597) acc1: 100.0000 (91.2265) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1468 data: 0.0547 max mem: 3540 +Test: [2260/2387] eta: 0:00:24 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1554 (0.2593) acc1: 100.0000 (91.2640) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1473 data: 0.0550 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1452 (0.2587) acc1: 100.0000 (91.3013) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1386 data: 0.0463 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1352 (0.2583) acc1: 100.0000 (91.3229) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1276 data: 0.0367 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1379 (0.2578) acc1: 100.0000 (91.3604) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1241 data: 0.0337 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1483 (0.2574) acc1: 100.0000 (91.3956) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1227 data: 0.0327 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1602 (0.2571) acc1: 99.2188 (91.4261) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1225 data: 0.0326 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1955 (0.2569) acc1: 98.9583 (91.4515) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1227 data: 0.0328 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1955 (0.2567) acc1: 98.9583 (91.4782) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1254 data: 0.0355 max mem: 3541 +Test: [2340/2387] eta: 0:00:09 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1790 (0.2564) acc1: 99.2188 (91.5081) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1248 data: 0.0350 max mem: 3541 +Test: [2350/2387] eta: 0:00:07 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1732 (0.2563) acc1: 99.4792 (91.5205) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1210 data: 0.0312 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2098 (0.2562) acc1: 98.9583 (91.5425) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1216 data: 0.0317 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2098 (0.2560) acc1: 98.9583 (91.5655) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1209 data: 0.0311 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2509 (0.2561) acc1: 97.3958 (91.5803) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1194 data: 0.0296 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2667 (0.2561) acc1: 94.2708 (91.5876) acc5: 100.0000 (100.0000) acc1_neg: 93.023 acc5_neg: 100.000 time: 0.1141 data: 0.0269 max mem: 3541 +Test: Total time: 0:07:36 (0.1912 s / it) +* Overall Acc@1 91.588 Acc@5 100.000 + target == real Acc@1 33.551 Acc@5 100.000 + target == fake Acc@1 95.663 Acc@5 100.000 + loss 0.256 + PR-AUC 0.9919 ROC-AUC 0.8959 EER 17.71% (thr=0.7524) +Accuracy of the network on the 916267 test images: 91.6% +[rank0]:[W726 03:32:35.195626562 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:32:57.216000 1174859 torch/distributed/run.py:766] +W0726 03:32:57.216000 1174859 torch/distributed/run.py:766] ***************************************** +W0726 03:32:57.216000 1174859 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:32:57.216000 1174859 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:33:29.604153460 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:33:29.947386367 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:33:29.244755904 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:33:29.356950084 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:33:30.599586352 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:33:30.672690072 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:33:30.703277102 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank6]:[W726 03:33:30.703320607 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n45_seed5_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed5_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:04:33 acc1_pos: 28.906 acc5_pos: 100.000 loss: 0.9809 (0.9809) acc1: 28.9062 (28.9062) acc5: 100.0000 (100.0000) time: 7.4930 data: 5.8741 max mem: 3531 +Test: [ 10/517] eta: 0:06:30 acc1_pos: 14.323 acc5_pos: 100.000 loss: 1.0883 (1.0778) acc1: 26.8229 (24.4555) acc5: 100.0000 (100.0000) time: 0.7701 data: 0.5344 max mem: 3531 +Test: [ 20/517] eta: 0:03:43 acc1_pos: 29.948 acc5_pos: 100.000 loss: 1.0110 (1.0715) acc1: 24.7396 (24.8388) acc5: 100.0000 (100.0000) time: 0.0968 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:03:02 acc1_pos: 1.823 acc5_pos: 100.000 loss: 1.0825 (1.0951) acc1: 16.9271 (22.1858) acc5: 100.0000 (100.0000) time: 0.1588 data: 0.0629 max mem: 3531 +Test: [ 40/517] eta: 0:02:45 acc1_pos: 10.156 acc5_pos: 100.000 loss: 1.1920 (1.1377) acc1: 10.9375 (18.6039) acc5: 100.0000 (100.0000) time: 0.2398 data: 0.1433 max mem: 3531 +Test: [ 50/517] eta: 0:02:34 acc1_pos: 52.294 acc5_pos: 100.000 loss: 1.1799 (1.0872) acc1: 11.7188 (23.1158) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2590 data: 0.1649 max mem: 3531 +Test: [ 60/517] eta: 0:02:24 acc1_pos: 52.294 acc5_pos: 100.000 loss: 0.3549 (0.9606) acc1: 87.7604 (34.9940) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2514 data: 0.1600 max mem: 3531 +Test: [ 70/517] eta: 0:02:16 acc1_pos: 52.294 acc5_pos: 100.000 loss: 0.2956 (0.8655) acc1: 97.6562 (43.9004) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2450 data: 0.1521 max mem: 3531 +Test: [ 80/517] eta: 0:02:10 acc1_pos: 52.294 acc5_pos: 100.000 loss: 0.2956 (0.8007) acc1: 96.8750 (50.1961) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.2449 data: 0.1502 max mem: 3531 +Test: [ 90/517] eta: 0:02:05 acc1_pos: 52.294 acc5_pos: 100.000 loss: 0.3300 (0.7482) acc1: 96.8750 (55.4487) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2505 data: 0.1580 max mem: 3531 +Test: [100/517] eta: 0:02:00 acc1_pos: 52.294 acc5_pos: 100.000 loss: 0.2955 (0.7026) acc1: 98.4375 (59.6896) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2484 data: 0.1573 max mem: 3531 +Test: [110/517] eta: 0:01:55 acc1_pos: 52.294 acc5_pos: 100.000 loss: 0.2300 (0.6585) acc1: 99.4792 (63.2742) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2440 data: 0.1519 max mem: 3531 +Test: [120/517] eta: 0:01:51 acc1_pos: 52.294 acc5_pos: 100.000 loss: 0.2581 (0.6270) acc1: 99.4792 (66.2212) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2500 data: 0.1582 max mem: 3531 +Test: [130/517] eta: 0:01:48 acc1_pos: 52.294 acc5_pos: 100.000 loss: 0.2791 (0.6015) acc1: 98.6979 (68.6188) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2599 data: 0.1666 max mem: 3531 +Test: [140/517] eta: 0:01:44 acc1_pos: 52.294 acc5_pos: 100.000 loss: 0.2695 (0.5761) acc1: 99.4792 (70.8278) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2547 data: 0.1573 max mem: 3531 +Test: [150/517] eta: 0:01:41 acc1_pos: 52.294 acc5_pos: 100.000 loss: 0.2388 (0.5538) acc1: 99.7396 (72.7339) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2499 data: 0.1525 max mem: 3531 +Test: [160/517] eta: 0:01:37 acc1_pos: 52.294 acc5_pos: 100.000 loss: 0.2546 (0.5349) acc1: 99.4792 (74.3498) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2470 data: 0.1539 max mem: 3531 +Test: [170/517] eta: 0:01:34 acc1_pos: 52.294 acc5_pos: 100.000 loss: 0.2543 (0.5182) acc1: 98.6979 (75.7219) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2470 data: 0.1558 max mem: 3531 +Test: [180/517] eta: 0:01:31 acc1_pos: 52.294 acc5_pos: 100.000 loss: 0.2318 (0.5024) acc1: 98.6979 (76.9495) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2628 data: 0.1695 max mem: 3531 +Test: [190/517] eta: 0:01:28 acc1_pos: 52.294 acc5_pos: 100.000 loss: 0.1768 (0.4842) acc1: 99.7396 (78.1536) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2479 data: 0.1532 max mem: 3531 +Test: [200/517] eta: 0:01:25 acc1_pos: 52.294 acc5_pos: 100.000 loss: 0.1708 (0.4705) acc1: 100.0000 (79.2276) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2363 data: 0.1426 max mem: 3531 +Test: [210/517] eta: 0:01:22 acc1_pos: 52.294 acc5_pos: 100.000 loss: 0.2333 (0.4622) acc1: 98.9583 (80.0355) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2523 data: 0.1586 max mem: 3532 +Test: [220/517] eta: 0:01:19 acc1_pos: 52.294 acc5_pos: 100.000 loss: 0.3119 (0.4560) acc1: 94.7917 (80.6738) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2593 data: 0.1661 max mem: 3532 +Test: [230/517] eta: 0:01:16 acc1_pos: 52.294 acc5_pos: 100.000 loss: 0.3164 (0.4498) acc1: 96.0938 (81.4270) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2444 data: 0.1514 max mem: 3532 +Test: [240/517] eta: 0:01:13 acc1_pos: 52.294 acc5_pos: 100.000 loss: 0.2933 (0.4422) acc1: 98.6979 (82.1231) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.2490 data: 0.1539 max mem: 3532 +Test: [250/517] eta: 0:01:11 acc1_pos: 52.294 acc5_pos: 100.000 loss: 0.2501 (0.4350) acc1: 99.2188 (82.7720) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2630 data: 0.1639 max mem: 3532 +Test: [260/517] eta: 0:01:08 acc1_pos: 52.294 acc5_pos: 100.000 loss: 0.2598 (0.4286) acc1: 99.2188 (83.3952) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2599 data: 0.1614 max mem: 3532 +Test: [270/517] eta: 0:01:05 acc1_pos: 52.294 acc5_pos: 100.000 loss: 0.2751 (0.4234) acc1: 98.9583 (83.9512) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2599 data: 0.1633 max mem: 3532 +Test: [280/517] eta: 0:01:02 acc1_pos: 52.294 acc5_pos: 100.000 loss: 0.2703 (0.4178) acc1: 98.9583 (84.4668) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2549 data: 0.1576 max mem: 3532 +Test: [290/517] eta: 0:01:00 acc1_pos: 52.294 acc5_pos: 100.000 loss: 0.2578 (0.4130) acc1: 99.2188 (84.9549) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2616 data: 0.1659 max mem: 3532 +Test: [300/517] eta: 0:00:57 acc1_pos: 52.294 acc5_pos: 100.000 loss: 0.2642 (0.4086) acc1: 99.2188 (85.4193) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2649 data: 0.1715 max mem: 3532 +Test: [310/517] eta: 0:00:54 acc1_pos: 52.294 acc5_pos: 100.000 loss: 0.3205 (0.4062) acc1: 99.2188 (85.8454) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2532 data: 0.1602 max mem: 3532 +Test: [320/517] eta: 0:00:52 acc1_pos: 52.294 acc5_pos: 100.000 loss: 0.3252 (0.4019) acc1: 98.9583 (86.2360) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2549 data: 0.1610 max mem: 3532 +Test: [330/517] eta: 0:00:49 acc1_pos: 52.294 acc5_pos: 100.000 loss: 0.2330 (0.3968) acc1: 99.2188 (86.6283) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2798 data: 0.1852 max mem: 3532 +Test: [340/517] eta: 0:00:46 acc1_pos: 52.294 acc5_pos: 100.000 loss: 0.2295 (0.3926) acc1: 99.7396 (86.9998) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2638 data: 0.1700 max mem: 3532 +Test: [350/517] eta: 0:00:44 acc1_pos: 52.294 acc5_pos: 100.000 loss: 0.2831 (0.3903) acc1: 99.2188 (87.3012) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2338 data: 0.1401 max mem: 3532 +Test: [360/517] eta: 0:00:41 acc1_pos: 52.294 acc5_pos: 100.000 loss: 0.2928 (0.3875) acc1: 98.1771 (87.5736) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2549 data: 0.1580 max mem: 3532 +Test: [370/517] eta: 0:00:38 acc1_pos: 52.294 acc5_pos: 100.000 loss: 0.2146 (0.3819) acc1: 99.7396 (87.9078) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1583 max mem: 3532 +Test: [380/517] eta: 0:00:36 acc1_pos: 52.294 acc5_pos: 100.000 loss: 0.1836 (0.3773) acc1: 100.0000 (88.2184) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2449 data: 0.1511 max mem: 3532 +Test: [390/517] eta: 0:00:33 acc1_pos: 52.294 acc5_pos: 100.000 loss: 0.2016 (0.3730) acc1: 100.0000 (88.5190) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1617 max mem: 3532 +Test: [400/517] eta: 0:00:30 acc1_pos: 52.294 acc5_pos: 100.000 loss: 0.2016 (0.3686) acc1: 100.0000 (88.7982) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1606 max mem: 3532 +Test: [410/517] eta: 0:00:28 acc1_pos: 52.294 acc5_pos: 100.000 loss: 0.2071 (0.3650) acc1: 100.0000 (89.0707) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1591 max mem: 3532 +Test: [420/517] eta: 0:00:25 acc1_pos: 52.294 acc5_pos: 100.000 loss: 0.2276 (0.3618) acc1: 100.0000 (89.3229) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2498 data: 0.1550 max mem: 3532 +Test: [430/517] eta: 0:00:22 acc1_pos: 52.294 acc5_pos: 100.000 loss: 0.2276 (0.3591) acc1: 100.0000 (89.5453) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2398 data: 0.1434 max mem: 3532 +Test: [440/517] eta: 0:00:20 acc1_pos: 52.294 acc5_pos: 100.000 loss: 0.2283 (0.3563) acc1: 99.4792 (89.7729) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2399 data: 0.1419 max mem: 3533 +Test: [450/517] eta: 0:00:17 acc1_pos: 52.294 acc5_pos: 100.000 loss: 0.2228 (0.3532) acc1: 100.0000 (89.9985) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.3099 data: 0.2131 max mem: 3533 +Test: [460/517] eta: 0:00:15 acc1_pos: 52.294 acc5_pos: 100.000 loss: 0.2232 (0.3507) acc1: 100.0000 (90.1991) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.3499 data: 0.2527 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 52.294 acc5_pos: 100.000 loss: 0.2200 (0.3480) acc1: 100.0000 (90.4044) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.3449 data: 0.2488 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 52.294 acc5_pos: 100.000 loss: 0.2147 (0.3452) acc1: 100.0000 (90.6033) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.3449 data: 0.2503 max mem: 3533 +Test: [490/517] eta: 0:00:07 acc1_pos: 52.294 acc5_pos: 100.000 loss: 0.2325 (0.3433) acc1: 100.0000 (90.7735) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2618 data: 0.1658 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 52.294 acc5_pos: 100.000 loss: 0.2974 (0.3440) acc1: 98.9583 (90.7409) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.2045 data: 0.1097 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 52.294 acc5_pos: 100.000 loss: 0.3120 (0.3429) acc1: 96.0938 (90.8650) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1825 data: 0.0916 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 52.294 acc5_pos: 100.000 loss: 0.2966 (0.3419) acc1: 98.9583 (90.9533) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1656 data: 0.0736 max mem: 3533 +Test: Total time: 0:02:16 (0.2636 s / it) +* Overall Acc@1 90.953 Acc@5 100.000 + target == real Acc@1 19.458 Acc@5 100.000 + target == fake Acc@1 98.370 Acc@5 100.000 + loss 0.342 + PR-AUC 0.9642 ROC-AUC 0.7695 EER 30.28% (thr=0.7466) +Accuracy of the network on the 198427 test images: 91.0% +[rank0]:[W726 03:35:56.694067217 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 03:36:16.081000 1175431 torch/distributed/run.py:766] +W0726 03:36:16.081000 1175431 torch/distributed/run.py:766] ***************************************** +W0726 03:36:16.081000 1175431 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:36:16.081000 1175431 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 03:36:45.705537784 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:36:45.967953347 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:36:46.540662472 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:36:46.647256185 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:36:46.803389667 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:36:46.956880237 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:36:46.048648738 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W726 03:36:46.055718634 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n45_seed5_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed5_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:33:15 acc1_pos: 22.135 acc5_pos: 100.000 loss: 0.9205 (0.9205) acc1: 22.1354 (22.1354) acc5: 100.0000 (100.0000) time: 5.1296 data: 2.7427 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/389] eta: 0:03:46 acc1_pos: 75.521 acc5_pos: 100.000 loss: 0.6692 (0.6589) acc1: 51.3021 (56.2737) acc5: 100.0000 (100.0000) time: 0.5972 data: 0.2987 max mem: 3531 +Test: [ 20/389] eta: 0:02:20 acc1_pos: 20.312 acc5_pos: 100.000 loss: 0.6692 (0.7150) acc1: 48.4375 (48.8591) acc5: 100.0000 (100.0000) time: 0.1434 data: 0.0523 max mem: 3531 +Test: [ 30/389] eta: 0:01:58 acc1_pos: 51.823 acc5_pos: 100.000 loss: 0.8264 (0.7474) acc1: 33.5938 (44.7665) acc5: 100.0000 (100.0000) time: 0.1816 data: 0.0878 max mem: 3531 +Test: [ 40/389] eta: 0:01:53 acc1_pos: 17.188 acc5_pos: 100.000 loss: 0.8164 (0.7562) acc1: 33.5938 (44.5821) acc5: 100.0000 (100.0000) time: 0.2671 data: 0.1698 max mem: 3531 +Test: [ 50/389] eta: 0:01:41 acc1_pos: 68.750 acc5_pos: 100.000 loss: 0.6891 (0.7466) acc1: 53.3854 (46.4869) acc5: 100.0000 (100.0000) time: 0.2498 data: 0.1530 max mem: 3531 +Test: [ 60/389] eta: 0:01:28 acc1_pos: 64.323 acc5_pos: 100.000 loss: 0.6264 (0.7221) acc1: 67.1875 (50.1195) acc5: 100.0000 (100.0000) time: 0.1577 data: 0.0616 max mem: 3531 +Test: [ 70/389] eta: 0:01:24 acc1_pos: 55.208 acc5_pos: 100.000 loss: 0.6321 (0.7172) acc1: 64.3229 (50.4438) acc5: 100.0000 (100.0000) time: 0.1754 data: 0.0808 max mem: 3531 +Test: [ 80/389] eta: 0:01:19 acc1_pos: 84.896 acc5_pos: 100.000 loss: 0.6844 (0.7235) acc1: 51.3021 (50.6430) acc5: 100.0000 (100.0000) time: 0.2108 data: 0.1177 max mem: 3531 +Test: [ 90/389] eta: 0:01:15 acc1_pos: 66.406 acc5_pos: 100.000 loss: 0.5350 (0.6976) acc1: 71.3542 (53.4655) acc5: 100.0000 (100.0000) time: 0.2193 data: 0.1256 max mem: 3531 +Test: [100/389] eta: 0:01:11 acc1_pos: 36.458 acc5_pos: 100.000 loss: 0.5850 (0.7070) acc1: 70.8333 (52.7331) acc5: 100.0000 (100.0000) time: 0.2190 data: 0.1247 max mem: 3531 +Test: [110/389] eta: 0:01:09 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.7811 (0.7146) acc1: 41.4062 (52.3414) acc5: 100.0000 (100.0000) time: 0.2202 data: 0.1259 max mem: 3531 +Test: [120/389] eta: 0:01:05 acc1_pos: 65.625 acc5_pos: 100.000 loss: 0.6353 (0.7060) acc1: 63.0208 (53.4349) acc5: 100.0000 (100.0000) time: 0.2271 data: 0.1342 max mem: 3531 +Test: [130/389] eta: 0:01:01 acc1_pos: 50.307 acc5_pos: 100.000 loss: 0.5690 (0.6981) acc1: 71.8750 (54.6835) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1766 data: 0.0814 max mem: 3531 +Test: [140/389] eta: 0:00:59 acc1_pos: 50.307 acc5_pos: 100.000 loss: 0.4585 (0.6878) acc1: 81.5104 (56.1632) acc5: 100.0000 (100.0000) acc1_neg: 22.135 acc5_neg: 100.000 time: 0.2050 data: 0.1106 max mem: 3531 +Test: [150/389] eta: 0:00:55 acc1_pos: 50.307 acc5_pos: 100.000 loss: 0.4585 (0.6782) acc1: 81.5104 (57.0588) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2163 data: 0.1217 max mem: 3531 +Test: [160/389] eta: 0:00:52 acc1_pos: 50.307 acc5_pos: 100.000 loss: 0.4337 (0.6584) acc1: 82.2917 (58.9561) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1703 data: 0.0762 max mem: 3531 +Test: [170/389] eta: 0:00:50 acc1_pos: 50.307 acc5_pos: 100.000 loss: 0.2715 (0.6461) acc1: 98.9583 (60.1075) acc5: 100.0000 (100.0000) acc1_neg: 52.865 acc5_neg: 100.000 time: 0.2115 data: 0.1202 max mem: 3531 +Test: [180/389] eta: 0:00:48 acc1_pos: 50.307 acc5_pos: 100.000 loss: 0.3536 (0.6337) acc1: 93.4896 (61.7188) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2222 data: 0.1273 max mem: 3531 +Test: [190/389] eta: 0:00:45 acc1_pos: 50.307 acc5_pos: 100.000 loss: 0.4257 (0.6281) acc1: 90.6250 (62.2696) acc5: 100.0000 (100.0000) acc1_neg: 36.198 acc5_neg: 100.000 time: 0.2058 data: 0.1088 max mem: 3531 +Test: [200/389] eta: 0:00:43 acc1_pos: 50.307 acc5_pos: 100.000 loss: 0.5438 (0.6244) acc1: 72.6562 (62.7319) acc5: 100.0000 (100.0000) acc1_neg: 66.667 acc5_neg: 100.000 time: 0.2284 data: 0.1350 max mem: 3531 +Test: [210/389] eta: 0:00:40 acc1_pos: 50.307 acc5_pos: 100.000 loss: 0.3779 (0.6115) acc1: 78.1250 (63.9366) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1758 data: 0.0848 max mem: 3532 +Test: [220/389] eta: 0:00:37 acc1_pos: 50.307 acc5_pos: 100.000 loss: 0.2487 (0.5961) acc1: 100.0000 (65.4188) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1614 data: 0.0691 max mem: 3532 +Test: [230/389] eta: 0:00:35 acc1_pos: 50.307 acc5_pos: 100.000 loss: 0.3237 (0.5914) acc1: 90.6250 (65.8877) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.2039 data: 0.1108 max mem: 3532 +Test: [240/389] eta: 0:00:32 acc1_pos: 50.307 acc5_pos: 100.000 loss: 0.4331 (0.5848) acc1: 85.1562 (66.7066) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1874 data: 0.0933 max mem: 3532 +Test: [250/389] eta: 0:00:30 acc1_pos: 50.307 acc5_pos: 100.000 loss: 0.3374 (0.5744) acc1: 93.4896 (67.8297) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1950 data: 0.1016 max mem: 3532 +Test: [260/389] eta: 0:00:28 acc1_pos: 50.307 acc5_pos: 100.000 loss: 0.3312 (0.5689) acc1: 99.4792 (68.5006) acc5: 100.0000 (100.0000) acc1_neg: 26.562 acc5_neg: 100.000 time: 0.2023 data: 0.1094 max mem: 3532 +Test: [270/389] eta: 0:00:25 acc1_pos: 50.307 acc5_pos: 100.000 loss: 0.4655 (0.5719) acc1: 84.3750 (68.2388) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.1900 data: 0.0976 max mem: 3532 +Test: [280/389] eta: 0:00:23 acc1_pos: 50.307 acc5_pos: 100.000 loss: 0.5266 (0.5758) acc1: 73.6979 (67.9891) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1775 data: 0.0869 max mem: 3532 +Test: [290/389] eta: 0:00:21 acc1_pos: 50.307 acc5_pos: 100.000 loss: 0.5129 (0.5688) acc1: 81.2500 (68.7097) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1789 data: 0.0875 max mem: 3532 +Test: [300/389] eta: 0:00:19 acc1_pos: 50.307 acc5_pos: 100.000 loss: 0.5077 (0.5684) acc1: 81.2500 (69.0217) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.1795 data: 0.0877 max mem: 3532 +Test: [310/389] eta: 0:00:16 acc1_pos: 50.307 acc5_pos: 100.000 loss: 0.5603 (0.5723) acc1: 77.0833 (68.7182) acc5: 100.0000 (100.0000) acc1_neg: 6.250 acc5_neg: 100.000 time: 0.1572 data: 0.0662 max mem: 3532 +Test: [320/389] eta: 0:00:14 acc1_pos: 50.307 acc5_pos: 100.000 loss: 0.5528 (0.5713) acc1: 67.9688 (68.8084) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1432 data: 0.0525 max mem: 3532 +Test: [330/389] eta: 0:00:12 acc1_pos: 50.307 acc5_pos: 100.000 loss: 0.4779 (0.5691) acc1: 89.0625 (69.3833) acc5: 100.0000 (100.0000) acc1_neg: 74.740 acc5_neg: 100.000 time: 0.1435 data: 0.0527 max mem: 3532 +Test: [340/389] eta: 0:00:10 acc1_pos: 50.307 acc5_pos: 100.000 loss: 0.3363 (0.5600) acc1: 92.4479 (70.1880) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1372 data: 0.0462 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 50.307 acc5_pos: 100.000 loss: 0.2612 (0.5639) acc1: 99.2188 (70.0669) acc5: 100.0000 (100.0000) acc1_neg: 64.844 acc5_neg: 100.000 time: 0.1447 data: 0.0533 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 50.307 acc5_pos: 100.000 loss: 0.4594 (0.5621) acc1: 85.4167 (70.4034) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1510 data: 0.0596 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 50.307 acc5_pos: 100.000 loss: 0.4484 (0.5592) acc1: 87.5000 (70.6579) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1677 data: 0.0767 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 50.307 acc5_pos: 100.000 loss: 0.3926 (0.5576) acc1: 93.2292 (70.8402) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.1523 data: 0.0621 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 50.307 acc5_pos: 100.000 loss: 0.3926 (0.5538) acc1: 90.3646 (71.2704) acc5: 100.0000 (100.0000) acc1_neg: 96.175 acc5_neg: 100.000 time: 0.1061 data: 0.0130 max mem: 3532 +Test: Total time: 0:01:16 (0.1979 s / it) +* Overall Acc@1 71.270 Acc@5 100.000 + target == real Acc@1 52.972 Acc@5 100.000 + target == fake Acc@1 79.776 Acc@5 100.000 + loss 0.554 + PR-AUC 0.8720 ROC-AUC 0.7611 EER 30.80% (thr=0.5752) +Accuracy of the network on the 149358 test images: 71.3% +[rank0]:[W726 03:38:12.946223879 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:38:32.201000 1175989 torch/distributed/run.py:766] +W0726 03:38:32.201000 1175989 torch/distributed/run.py:766] ***************************************** +W0726 03:38:32.201000 1175989 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:38:32.201000 1175989 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:39:00.475218263 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:39:00.534043901 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:39:00.799828369 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +NCCL version 2.26.2+cuda12.2 +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:39:00.894888506 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:39:00.030448751 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:39:00.168578971 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:39:00.182608384 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:39:00.205988288 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n45_seed5_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed5_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:37:47 acc1_pos: 81.250 acc5_pos: 100.000 loss: 0.4210 (0.4210) acc1: 81.2500 (81.2500) acc5: 100.0000 (100.0000) time: 6.8089 data: 4.9983 max mem: 3531 +Test: [ 10/333] eta: 0:03:48 acc1_pos: 86.198 acc5_pos: 100.000 loss: 0.3395 (0.3365) acc1: 83.3333 (84.5644) acc5: 100.0000 (100.0000) time: 0.7066 data: 0.4548 max mem: 3531 +Test: [ 20/333] eta: 0:02:10 acc1_pos: 94.792 acc5_pos: 100.000 loss: 0.3299 (0.3346) acc1: 84.8958 (85.3299) acc5: 100.0000 (100.0000) time: 0.0981 data: 0.0004 max mem: 3531 +Test: [ 30/333] eta: 0:01:57 acc1_pos: 55.990 acc5_pos: 100.000 loss: 0.4072 (0.4817) acc1: 75.2604 (74.1347) acc5: 100.0000 (100.0000) time: 0.2100 data: 0.1136 max mem: 3531 +Test: [ 40/333] eta: 0:01:45 acc1_pos: 50.521 acc5_pos: 100.000 loss: 0.7936 (0.5625) acc1: 48.1771 (67.5559) acc5: 100.0000 (100.0000) time: 0.2970 data: 0.2012 max mem: 3531 +Test: [ 50/333] eta: 0:01:31 acc1_pos: 49.219 acc5_pos: 100.000 loss: 0.7936 (0.6108) acc1: 47.1354 (63.2761) acc5: 100.0000 (100.0000) time: 0.2260 data: 0.1294 max mem: 3531 +Test: [ 60/333] eta: 0:01:23 acc1_pos: 35.156 acc5_pos: 100.000 loss: 0.8327 (0.6455) acc1: 44.7917 (60.4124) acc5: 100.0000 (100.0000) time: 0.1978 data: 0.1013 max mem: 3531 +Test: [ 70/333] eta: 0:01:15 acc1_pos: 50.521 acc5_pos: 100.000 loss: 0.8144 (0.6680) acc1: 44.5312 (58.8285) acc5: 100.0000 (100.0000) time: 0.2002 data: 0.1047 max mem: 3531 +Test: [ 80/333] eta: 0:01:10 acc1_pos: 55.729 acc5_pos: 100.000 loss: 0.7854 (0.6832) acc1: 48.4375 (57.5521) acc5: 100.0000 (100.0000) time: 0.1906 data: 0.0963 max mem: 3531 +Test: [ 90/333] eta: 0:01:05 acc1_pos: 62.760 acc5_pos: 100.000 loss: 0.7349 (0.6864) acc1: 54.6875 (57.4090) acc5: 100.0000 (100.0000) time: 0.2001 data: 0.1056 max mem: 3531 +Test: [100/333] eta: 0:01:02 acc1_pos: 62.240 acc5_pos: 100.000 loss: 0.6895 (0.6885) acc1: 58.3333 (57.4928) acc5: 100.0000 (100.0000) time: 0.2204 data: 0.1270 max mem: 3531 +Test: [110/333] eta: 0:00:57 acc1_pos: 53.906 acc5_pos: 100.000 loss: 0.7132 (0.6905) acc1: 55.2083 (57.2846) acc5: 100.0000 (100.0000) time: 0.2096 data: 0.1170 max mem: 3531 +Test: [120/333] eta: 0:00:54 acc1_pos: 57.812 acc5_pos: 100.000 loss: 0.7103 (0.6902) acc1: 53.9062 (57.1604) acc5: 100.0000 (100.0000) time: 0.1898 data: 0.0980 max mem: 3531 +Test: [130/333] eta: 0:00:50 acc1_pos: 76.823 acc5_pos: 100.000 loss: 0.6597 (0.6888) acc1: 56.7708 (57.3771) acc5: 100.0000 (100.0000) time: 0.1989 data: 0.1061 max mem: 3531 +Test: [140/333] eta: 0:00:48 acc1_pos: 73.958 acc5_pos: 100.000 loss: 0.6597 (0.6873) acc1: 57.2917 (57.4136) acc5: 100.0000 (100.0000) time: 0.2237 data: 0.1290 max mem: 3531 +Test: [150/333] eta: 0:00:44 acc1_pos: 58.594 acc5_pos: 100.000 loss: 0.6805 (0.6890) acc1: 58.0729 (57.3417) acc5: 100.0000 (100.0000) time: 0.2012 data: 0.1050 max mem: 3531 +Test: [160/333] eta: 0:00:41 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.7098 (0.6951) acc1: 51.8229 (57.0135) acc5: 100.0000 (100.0000) acc1_neg: 46.094 acc5_neg: 100.000 time: 0.1809 data: 0.0850 max mem: 3531 +Test: [170/333] eta: 0:00:39 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.7361 (0.6996) acc1: 53.6458 (56.9155) acc5: 100.0000 (100.0000) acc1_neg: 59.115 acc5_neg: 100.000 time: 0.2141 data: 0.1181 max mem: 3531 +Test: [180/333] eta: 0:00:36 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.7207 (0.6997) acc1: 58.3333 (57.0629) acc5: 100.0000 (100.0000) acc1_neg: 54.688 acc5_neg: 100.000 time: 0.1958 data: 0.1006 max mem: 3531 +Test: [190/333] eta: 0:00:33 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.6703 (0.6962) acc1: 60.1562 (57.4730) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.1895 data: 0.0965 max mem: 3531 +Test: [200/333] eta: 0:00:31 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.6473 (0.6931) acc1: 65.6250 (57.9550) acc5: 100.0000 (100.0000) acc1_neg: 65.885 acc5_neg: 100.000 time: 0.2107 data: 0.1187 max mem: 3531 +Test: [210/333] eta: 0:00:28 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.6654 (0.6945) acc1: 64.8438 (58.1124) acc5: 100.0000 (100.0000) acc1_neg: 57.292 acc5_neg: 100.000 time: 0.1966 data: 0.1052 max mem: 3532 +Test: [220/333] eta: 0:00:26 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.7875 (0.7025) acc1: 57.2917 (57.7147) acc5: 100.0000 (100.0000) acc1_neg: 76.042 acc5_neg: 100.000 time: 0.1989 data: 0.1061 max mem: 3532 +Test: [230/333] eta: 0:00:23 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.4643 (0.6905) acc1: 80.9896 (58.9105) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.2140 data: 0.1216 max mem: 3532 +Test: [240/333] eta: 0:00:21 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.4551 (0.6843) acc1: 80.9896 (59.5382) acc5: 100.0000 (100.0000) acc1_neg: 72.917 acc5_neg: 100.000 time: 0.2072 data: 0.1158 max mem: 3532 +Test: [250/333] eta: 0:00:18 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.5699 (0.6828) acc1: 70.0521 (59.8004) acc5: 100.0000 (100.0000) acc1_neg: 52.344 acc5_neg: 100.000 time: 0.2143 data: 0.1204 max mem: 3532 +Test: [260/333] eta: 0:00:16 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.6841 (0.6846) acc1: 62.2396 (59.7761) acc5: 100.0000 (100.0000) acc1_neg: 65.365 acc5_neg: 100.000 time: 0.2268 data: 0.1332 max mem: 3532 +Test: [270/333] eta: 0:00:14 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.6327 (0.6817) acc1: 65.3646 (60.1063) acc5: 100.0000 (100.0000) acc1_neg: 68.229 acc5_neg: 100.000 time: 0.2188 data: 0.1263 max mem: 3532 +Test: [280/333] eta: 0:00:11 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.6099 (0.6827) acc1: 67.1875 (60.0867) acc5: 100.0000 (100.0000) acc1_neg: 65.365 acc5_neg: 100.000 time: 0.1858 data: 0.0936 max mem: 3532 +Test: [290/333] eta: 0:00:09 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.7853 (0.6878) acc1: 51.8229 (59.7016) acc5: 100.0000 (100.0000) acc1_neg: 49.219 acc5_neg: 100.000 time: 0.1503 data: 0.0593 max mem: 3532 +Test: [300/333] eta: 0:00:07 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.8567 (0.6946) acc1: 47.6562 (59.2521) acc5: 100.0000 (100.0000) acc1_neg: 52.604 acc5_neg: 100.000 time: 0.1346 data: 0.0438 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.8064 (0.6911) acc1: 52.6042 (59.6069) acc5: 100.0000 (100.0000) acc1_neg: 73.958 acc5_neg: 100.000 time: 0.1437 data: 0.0534 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.5423 (0.6879) acc1: 73.9583 (59.9818) acc5: 100.0000 (100.0000) acc1_neg: 46.615 acc5_neg: 100.000 time: 0.1506 data: 0.0607 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.8497 (0.6952) acc1: 48.6979 (59.5449) acc5: 100.0000 (100.0000) acc1_neg: 48.698 acc5_neg: 100.000 time: 0.1386 data: 0.0489 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.8938 (0.6966) acc1: 46.7606 (59.4636) acc5: 100.0000 (100.0000) acc1_neg: 46.761 acc5_neg: 100.000 time: 0.1386 data: 0.0453 max mem: 3532 +Test: Total time: 0:01:10 (0.2126 s / it) +* Overall Acc@1 59.464 Acc@5 100.000 + target == real Acc@1 57.363 Acc@5 100.000 + target == fake Acc@1 61.334 Acc@5 100.000 + loss 0.697 + PR-AUC 0.6468 ROC-AUC 0.6332 EER 40.65% (thr=0.5132) +Accuracy of the network on the 127843 test images: 59.5% +[rank0]:[W726 03:40:19.551293770 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:40:39.628000 1176548 torch/distributed/run.py:766] +W0726 03:40:39.628000 1176548 torch/distributed/run.py:766] ***************************************** +W0726 03:40:39.628000 1176548 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:40:39.628000 1176548 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:41:08.191178054 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 03:41:08.235419838 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W726 03:41:08.235423873 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:41:09.472373873 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:41:09.508621589 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:41:09.657103620 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:41:09.710871740 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:41:09.727110134 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n45_seed5_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed5_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:19:59 acc1_pos: 95.573 acc5_pos: 100.000 loss: 0.1885 (0.1885) acc1: 95.5729 (95.5729) acc5: 100.0000 (100.0000) time: 5.1046 data: 2.8420 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/235] eta: 0:02:05 acc1_pos: 95.573 acc5_pos: 100.000 loss: 0.1997 (0.1997) acc1: 95.5729 (95.5492) acc5: 100.0000 (100.0000) time: 0.5579 data: 0.2707 max mem: 3531 +Test: [ 20/235] eta: 0:01:18 acc1_pos: 96.615 acc5_pos: 100.000 loss: 0.1933 (0.1952) acc1: 95.5729 (95.9201) acc5: 100.0000 (100.0000) time: 0.1263 data: 0.0358 max mem: 3531 +Test: [ 30/235] eta: 0:01:09 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.1897 (0.1936) acc1: 96.3542 (96.0265) acc5: 100.0000 (100.0000) time: 0.2175 data: 0.1252 max mem: 3531 +Test: [ 40/235] eta: 0:01:03 acc1_pos: 96.354 acc5_pos: 100.000 loss: 0.1910 (0.1933) acc1: 96.3542 (96.0747) acc5: 100.0000 (100.0000) time: 0.2902 data: 0.1986 max mem: 3531 +Test: [ 50/235] eta: 0:00:54 acc1_pos: 94.531 acc5_pos: 100.000 loss: 0.1910 (0.1940) acc1: 96.0938 (96.0835) acc5: 100.0000 (100.0000) time: 0.2221 data: 0.1296 max mem: 3531 +Test: [ 60/235] eta: 0:00:46 acc1_pos: 96.094 acc5_pos: 100.000 loss: 0.1977 (0.1952) acc1: 95.8333 (96.0340) acc5: 100.0000 (100.0000) time: 0.1366 data: 0.0434 max mem: 3531 +Test: [ 70/235] eta: 0:00:41 acc1_pos: 96.923 acc5_pos: 100.000 loss: 0.2144 (0.3787) acc1: 94.2708 (85.7064) acc5: 100.0000 (100.0000) acc1_neg: 11.979 acc5_neg: 100.000 time: 0.1399 data: 0.0478 max mem: 3531 +Test: [ 80/235] eta: 0:00:39 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.6647 (0.5389) acc1: 12.7604 (76.6750) acc5: 100.0000 (100.0000) acc1_neg: 11.198 acc5_neg: 100.000 time: 0.2251 data: 0.1330 max mem: 3531 +Test: [ 90/235] eta: 0:00:36 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.6842 (0.6644) acc1: 12.5000 (69.6944) acc5: 100.0000 (100.0000) acc1_neg: 11.979 acc5_neg: 100.000 time: 0.2549 data: 0.1589 max mem: 3531 +Test: [100/235] eta: 0:00:32 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.6881 (0.7634) acc1: 13.0208 (64.1527) acc5: 100.0000 (100.0000) acc1_neg: 13.021 acc5_neg: 100.000 time: 0.1877 data: 0.0893 max mem: 3531 +Test: [110/235] eta: 0:00:29 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.6579 (0.8444) acc1: 13.0208 (59.5416) acc5: 100.0000 (100.0000) acc1_neg: 14.583 acc5_neg: 100.000 time: 0.1726 data: 0.0752 max mem: 3531 +Test: [120/235] eta: 0:00:26 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.6569 (0.9115) acc1: 11.7188 (55.5613) acc5: 100.0000 (100.0000) acc1_neg: 12.240 acc5_neg: 100.000 time: 0.1816 data: 0.0824 max mem: 3531 +Test: [130/235] eta: 0:00:24 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.6411 (0.9669) acc1: 11.7188 (52.2225) acc5: 100.0000 (100.0000) acc1_neg: 11.979 acc5_neg: 100.000 time: 0.1929 data: 0.0942 max mem: 3531 +Test: [140/235] eta: 0:00:21 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.6634 (1.0172) acc1: 11.4583 (49.2742) acc5: 100.0000 (100.0000) acc1_neg: 7.292 acc5_neg: 100.000 time: 0.1983 data: 0.1002 max mem: 3531 +Test: [150/235] eta: 0:00:19 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.6584 (1.0586) acc1: 10.6771 (46.7284) acc5: 100.0000 (100.0000) acc1_neg: 12.760 acc5_neg: 100.000 time: 0.1899 data: 0.0955 max mem: 3531 +Test: [160/235] eta: 0:00:16 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.6348 (1.0947) acc1: 11.4583 (44.5685) acc5: 100.0000 (100.0000) acc1_neg: 12.240 acc5_neg: 100.000 time: 0.1926 data: 0.1012 max mem: 3531 +Test: [170/235] eta: 0:00:14 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.5825 (1.1143) acc1: 13.5417 (43.1165) acc5: 100.0000 (100.0000) acc1_neg: 17.969 acc5_neg: 100.000 time: 0.1971 data: 0.1056 max mem: 3531 +Test: [180/235] eta: 0:00:12 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.4117 (1.1302) acc1: 20.3125 (41.8739) acc5: 100.0000 (100.0000) acc1_neg: 22.917 acc5_neg: 100.000 time: 0.2045 data: 0.1133 max mem: 3531 +Test: [190/235] eta: 0:00:09 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.3828 (1.1439) acc1: 20.8333 (40.7327) acc5: 100.0000 (100.0000) acc1_neg: 19.271 acc5_neg: 100.000 time: 0.1992 data: 0.1077 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.3961 (1.1570) acc1: 19.7917 (39.6883) acc5: 100.0000 (100.0000) acc1_neg: 19.792 acc5_neg: 100.000 time: 0.2042 data: 0.1118 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.4073 (1.1682) acc1: 19.2708 (38.7638) acc5: 100.0000 (100.0000) acc1_neg: 19.271 acc5_neg: 100.000 time: 0.2078 data: 0.1134 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.3898 (1.1783) acc1: 20.3125 (37.9619) acc5: 100.0000 (100.0000) acc1_neg: 20.052 acc5_neg: 100.000 time: 0.1838 data: 0.0883 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.3806 (1.1872) acc1: 21.6146 (37.2486) acc5: 100.0000 (100.0000) acc1_neg: 20.573 acc5_neg: 100.000 time: 0.1681 data: 0.0753 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.3913 (1.1913) acc1: 20.3125 (37.0064) acc5: 100.0000 (100.0000) acc1_neg: 21.739 acc5_neg: 100.000 time: 0.1507 data: 0.0625 max mem: 3532 +Test: Total time: 0:00:49 (0.2127 s / it) +* Overall Acc@1 37.006 Acc@5 100.000 + target == real Acc@1 96.046 Acc@5 100.000 + target == fake Acc@1 15.653 Acc@5 100.000 + loss 1.191 + PR-AUC 0.8566 ROC-AUC 0.6982 EER 35.33% (thr=0.1509) +Accuracy of the network on the 89879 test images: 37.0% +[rank0]:[W726 03:42:09.637048065 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:42:28.562000 1177094 torch/distributed/run.py:766] +W0726 03:42:28.562000 1177094 torch/distributed/run.py:766] ***************************************** +W0726 03:42:28.562000 1177094 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:42:28.562000 1177094 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:42:55.285162096 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:42:55.372476443 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:42:56.684147887 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:42:56.698057834 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:42:56.985862528 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:42:56.054144502 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:42:56.074898829 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:42:56.086651576 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n45_seed5_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed5_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:27:31 acc1_pos: 51.823 acc5_pos: 100.000 loss: 0.7186 (0.7186) acc1: 51.8229 (51.8229) acc5: 100.0000 (100.0000) time: 6.1386 data: 4.6921 max mem: 3531 +Test: [ 10/269] eta: 0:02:47 acc1_pos: 70.312 acc5_pos: 100.000 loss: 0.6134 (0.6467) acc1: 66.1458 (62.2159) acc5: 100.0000 (100.0000) time: 0.6465 data: 0.4270 max mem: 3531 +Test: [ 20/269] eta: 0:01:36 acc1_pos: 79.167 acc5_pos: 100.000 loss: 0.6121 (0.6250) acc1: 65.1042 (64.8438) acc5: 100.0000 (100.0000) time: 0.0986 data: 0.0011 max mem: 3531 +Test: [ 30/269] eta: 0:01:26 acc1_pos: 58.594 acc5_pos: 100.000 loss: 0.5970 (0.6073) acc1: 67.4479 (66.9019) acc5: 100.0000 (100.0000) time: 0.2053 data: 0.1105 max mem: 3531 +Test: [ 40/269] eta: 0:01:18 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.5640 (0.5937) acc1: 73.6979 (68.8961) acc5: 100.0000 (100.0000) acc1_neg: 64.323 acc5_neg: 100.000 time: 0.2972 data: 0.2063 max mem: 3531 +Test: [ 50/269] eta: 0:01:11 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.5410 (0.5913) acc1: 75.7812 (69.2964) acc5: 100.0000 (100.0000) acc1_neg: 44.792 acc5_neg: 100.000 time: 0.2696 data: 0.1777 max mem: 3531 +Test: [ 60/269] eta: 0:01:04 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.5564 (0.5973) acc1: 71.6146 (68.6774) acc5: 100.0000 (100.0000) acc1_neg: 59.896 acc5_neg: 100.000 time: 0.2440 data: 0.1504 max mem: 3531 +Test: [ 70/269] eta: 0:00:56 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.5610 (0.5922) acc1: 67.7083 (69.3699) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.1752 data: 0.0809 max mem: 3531 +Test: [ 80/269] eta: 0:00:51 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.5358 (0.5896) acc1: 74.2188 (69.7338) acc5: 100.0000 (100.0000) acc1_neg: 60.417 acc5_neg: 100.000 time: 0.1586 data: 0.0643 max mem: 3531 +Test: [ 90/269] eta: 0:00:48 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.6369 (0.5995) acc1: 63.5417 (68.3723) acc5: 100.0000 (100.0000) acc1_neg: 72.135 acc5_neg: 100.000 time: 0.2347 data: 0.1429 max mem: 3531 +Test: [100/269] eta: 0:00:44 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.6474 (0.6005) acc1: 58.8542 (68.1209) acc5: 100.0000 (100.0000) acc1_neg: 59.896 acc5_neg: 100.000 time: 0.2199 data: 0.1262 max mem: 3531 +Test: [110/269] eta: 0:00:40 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.6394 (0.6054) acc1: 61.1979 (67.2978) acc5: 100.0000 (100.0000) acc1_neg: 48.177 acc5_neg: 100.000 time: 0.1798 data: 0.0806 max mem: 3531 +Test: [120/269] eta: 0:00:37 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.6016 (0.6056) acc1: 64.0625 (67.1402) acc5: 100.0000 (100.0000) acc1_neg: 55.469 acc5_neg: 100.000 time: 0.1949 data: 0.0991 max mem: 3531 +Test: [130/269] eta: 0:00:34 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.6642 (0.6126) acc1: 57.8125 (66.1955) acc5: 100.0000 (100.0000) acc1_neg: 52.083 acc5_neg: 100.000 time: 0.2149 data: 0.1242 max mem: 3531 +Test: [140/269] eta: 0:00:31 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.7154 (0.6191) acc1: 52.6042 (65.2999) acc5: 100.0000 (100.0000) acc1_neg: 55.990 acc5_neg: 100.000 time: 0.2249 data: 0.1326 max mem: 3531 +Test: [150/269] eta: 0:00:28 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.6610 (0.6169) acc1: 56.2500 (65.4094) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.1907 data: 0.0957 max mem: 3531 +Test: [160/269] eta: 0:00:26 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.5436 (0.6143) acc1: 66.6667 (65.6751) acc5: 100.0000 (100.0000) acc1_neg: 60.417 acc5_neg: 100.000 time: 0.2003 data: 0.1061 max mem: 3531 +Test: [170/269] eta: 0:00:23 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.6288 (0.6181) acc1: 62.5000 (65.2747) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.2291 data: 0.1352 max mem: 3531 +Test: [180/269] eta: 0:00:21 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.5681 (0.6135) acc1: 70.5729 (65.7905) acc5: 100.0000 (100.0000) acc1_neg: 58.073 acc5_neg: 100.000 time: 0.1915 data: 0.0963 max mem: 3531 +Test: [190/269] eta: 0:00:18 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.5457 (0.6139) acc1: 71.3542 (65.6836) acc5: 100.0000 (100.0000) acc1_neg: 68.750 acc5_neg: 100.000 time: 0.1873 data: 0.0929 max mem: 3531 +Test: [200/269] eta: 0:00:16 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.6588 (0.6185) acc1: 60.4167 (65.2104) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.2129 data: 0.1186 max mem: 3531 +Test: [210/269] eta: 0:00:13 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.6588 (0.6190) acc1: 60.9375 (65.1301) acc5: 100.0000 (100.0000) acc1_neg: 65.365 acc5_neg: 100.000 time: 0.2064 data: 0.1121 max mem: 3532 +Test: [220/269] eta: 0:00:11 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.5998 (0.6188) acc1: 65.3646 (65.1525) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.1622 data: 0.0696 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.5603 (0.6165) acc1: 69.5312 (65.3736) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.1262 data: 0.0354 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.6182 (0.6185) acc1: 63.5417 (65.1398) acc5: 100.0000 (100.0000) acc1_neg: 59.375 acc5_neg: 100.000 time: 0.1282 data: 0.0371 max mem: 3532 +Test: [250/269] eta: 0:00:04 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.6602 (0.6207) acc1: 58.0729 (64.8479) acc5: 100.0000 (100.0000) acc1_neg: 32.552 acc5_neg: 100.000 time: 0.1208 data: 0.0299 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.7555 (0.6266) acc1: 47.9167 (64.2810) acc5: 100.0000 (100.0000) acc1_neg: 67.708 acc5_neg: 100.000 time: 0.1127 data: 0.0225 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.6591 (0.6259) acc1: 56.7708 (64.3645) acc5: 100.0000 (100.0000) acc1_neg: 78.215 acc5_neg: 100.000 time: 0.1077 data: 0.0144 max mem: 3532 +Test: Total time: 0:00:56 (0.2084 s / it) +* Overall Acc@1 64.364 Acc@5 100.000 + target == real Acc@1 66.624 Acc@5 100.000 + target == fake Acc@1 64.056 Acc@5 100.000 + loss 0.626 + PR-AUC 0.9440 ROC-AUC 0.7130 EER 34.62% (thr=0.4932) +Accuracy of the network on the 103293 test images: 64.4% +[rank0]:[W726 03:44:01.759480735 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T03:44:08Z] eval_done tag=fixed500k_fake_n45_seed5_resampled +[2026-07-26T03:44:08Z] eval_start tag=fixed500k_real_n23_seed6_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed6_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T03:44:08Z +hostname=n116-041-023 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed6_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n23_seed6_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 03:44:21.280000 1177670 torch/distributed/run.py:766] +W0726 03:44:21.280000 1177670 torch/distributed/run.py:766] ***************************************** +W0726 03:44:21.280000 1177670 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:44:21.280000 1177670 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:44:49.403201975 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:44:50.589642425 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:44:50.853371304 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:44:50.966303336 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:44:50.042665113 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:44:50.129621295 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:44:50.152961729 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:44:50.190394949 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n23_seed6_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed6_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:18:57 acc1_pos: 7.812 acc5_pos: 100.000 loss: 1.3275 (1.3275) acc1: 7.8125 (7.8125) acc5: 100.0000 (100.0000) time: 4.8631 data: 3.1789 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/234] eta: 0:01:57 acc1_pos: 34.635 acc5_pos: 100.000 loss: 1.0368 (1.0770) acc1: 28.9062 (30.1610) acc5: 100.0000 (100.0000) time: 0.5241 data: 0.2893 max mem: 3531 +Test: [ 20/234] eta: 0:01:07 acc1_pos: 95.052 acc5_pos: 100.000 loss: 1.0709 (1.1057) acc1: 28.3854 (33.1721) acc5: 100.0000 (100.0000) time: 0.0902 data: 0.0004 max mem: 3531 +Test: [ 30/234] eta: 0:00:50 acc1_pos: 78.125 acc5_pos: 100.000 loss: 0.2109 (0.8237) acc1: 95.0521 (52.7554) acc5: 100.0000 (100.0000) time: 0.0932 data: 0.0004 max mem: 3531 +Test: [ 40/234] eta: 0:00:43 acc1_pos: 98.177 acc5_pos: 100.000 loss: 0.1838 (0.7200) acc1: 96.6146 (59.5592) acc5: 100.0000 (100.0000) time: 0.1280 data: 0.0339 max mem: 3531 +Test: [ 50/234] eta: 0:00:41 acc1_pos: 97.396 acc5_pos: 100.000 loss: 0.1503 (0.6070) acc1: 96.8750 (67.0037) acc5: 100.0000 (100.0000) time: 0.1999 data: 0.1070 max mem: 3531 +Test: [ 60/234] eta: 0:00:38 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.1231 (0.5252) acc1: 98.6979 (72.3873) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2222 data: 0.1298 max mem: 3531 +Test: [ 70/234] eta: 0:00:36 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.1083 (0.5214) acc1: 100.0000 (72.7516) acc5: 100.0000 (100.0000) acc1_neg: 59.115 acc5_neg: 100.000 time: 0.2153 data: 0.1226 max mem: 3531 +Test: [ 80/234] eta: 0:00:34 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.6648 (0.5195) acc1: 64.0625 (73.1739) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2176 data: 0.1222 max mem: 3531 +Test: [ 90/234] eta: 0:00:31 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.2936 (0.4902) acc1: 89.3229 (75.2060) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.1865 data: 0.0903 max mem: 3531 +Test: [100/234] eta: 0:00:29 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.3579 (0.4860) acc1: 86.1979 (75.6059) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1957 data: 0.0993 max mem: 3531 +Test: [110/234] eta: 0:00:26 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.4534 (0.4791) acc1: 78.6458 (76.1449) acc5: 100.0000 (100.0000) acc1_neg: 70.833 acc5_neg: 100.000 time: 0.2211 data: 0.1263 max mem: 3531 +Test: [120/234] eta: 0:00:24 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.4391 (0.4785) acc1: 79.6875 (76.4097) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.2257 data: 0.1327 max mem: 3531 +Test: [130/234] eta: 0:00:22 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.2697 (0.4647) acc1: 89.3229 (77.5604) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.2304 data: 0.1379 max mem: 3531 +Test: [140/234] eta: 0:00:20 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.3225 (0.4842) acc1: 88.0208 (76.6752) acc5: 100.0000 (100.0000) acc1_neg: 35.938 acc5_neg: 100.000 time: 0.2183 data: 0.1258 max mem: 3531 +Test: [150/234] eta: 0:00:18 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.4725 (0.4899) acc1: 76.3021 (76.4263) acc5: 100.0000 (100.0000) acc1_neg: 80.729 acc5_neg: 100.000 time: 0.1999 data: 0.1065 max mem: 3531 +Test: [160/234] eta: 0:00:15 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.3498 (0.4766) acc1: 85.4167 (77.4457) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.1961 data: 0.1043 max mem: 3531 +Test: [170/234] eta: 0:00:13 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.3105 (0.4667) acc1: 90.6250 (78.2697) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.2071 data: 0.1161 max mem: 3531 +Test: [180/234] eta: 0:00:11 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.3422 (0.4748) acc1: 87.5000 (77.7265) acc5: 100.0000 (100.0000) acc1_neg: 37.500 acc5_neg: 100.000 time: 0.2087 data: 0.1167 max mem: 3531 +Test: [190/234] eta: 0:00:09 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.4695 (0.4762) acc1: 75.7812 (77.5769) acc5: 100.0000 (100.0000) acc1_neg: 75.781 acc5_neg: 100.000 time: 0.1993 data: 0.1069 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.4544 (0.4642) acc1: 77.8646 (78.2507) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.1024 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.0935 (0.4484) acc1: 100.0000 (79.2383) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1917 data: 0.0996 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.2030 (0.4388) acc1: 97.3958 (79.9456) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.1658 data: 0.0744 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.4438 (0.4597) acc1: 45.0521 (78.2231) acc5: 100.0000 (100.0000) acc1_neg: 42.969 acc5_neg: 100.000 time: 0.1381 data: 0.0474 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.8995 (0.4655) acc1: 42.1875 (77.8143) acc5: 100.0000 (100.0000) acc1_neg: 43.969 acc5_neg: 100.000 time: 0.1279 data: 0.0363 max mem: 3532 +Test: Total time: 0:00:47 (0.2050 s / it) +* Overall Acc@1 77.814 Acc@5 100.000 + target == real Acc@1 67.226 Acc@5 100.000 + target == fake Acc@1 80.799 Acc@5 100.000 + loss 0.465 + PR-AUC 0.9473 ROC-AUC 0.8390 EER 25.70% (thr=0.6196) +Accuracy of the network on the 89729 test images: 77.8% +[rank0]:[W726 03:45:47.144129097 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:46:07.660000 1178231 torch/distributed/run.py:766] +W0726 03:46:07.660000 1178231 torch/distributed/run.py:766] ***************************************** +W0726 03:46:07.660000 1178231 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:46:07.660000 1178231 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 03:46:37.261778398 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:46:38.817303956 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:46:38.161810591 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:46:38.201877539 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:46:38.247733868 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:46:38.387983574 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:46:39.562455378 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:46:39.638499720 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n23_seed6_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed6_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:51:00 acc1_pos: 2.604 acc5_pos: 100.000 loss: 1.6428 (1.6428) acc1: 2.6042 (2.6042) acc5: 100.0000 (100.0000) time: 7.3150 data: 5.8200 max mem: 3531 +Test: [ 10/2387] eta: 0:29:35 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.7974 (1.7769) acc1: 2.6042 (2.9830) acc5: 100.0000 (100.0000) time: 0.7469 data: 0.5295 max mem: 3531 +Test: [ 20/2387] eta: 0:17:12 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.7974 (1.7675) acc1: 1.3021 (2.8150) acc5: 100.0000 (100.0000) time: 0.0924 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:14:25 acc1_pos: 1.823 acc5_pos: 100.000 loss: 1.8467 (1.7861) acc1: 1.3021 (2.6042) acc5: 100.0000 (100.0000) time: 0.1581 data: 0.0620 max mem: 3531 +Test: [ 40/2387] eta: 0:13:25 acc1_pos: 1.042 acc5_pos: 100.000 loss: 1.8558 (1.7981) acc1: 1.0417 (2.3819) acc5: 100.0000 (100.0000) time: 0.2453 data: 0.1507 max mem: 3531 +Test: [ 50/2387] eta: 0:12:27 acc1_pos: 25.781 acc5_pos: 100.000 loss: 1.7589 (1.7672) acc1: 1.3021 (3.3241) acc5: 100.0000 (100.0000) time: 0.2462 data: 0.1537 max mem: 3531 +Test: [ 60/2387] eta: 0:12:14 acc1_pos: 19.531 acc5_pos: 100.000 loss: 1.5808 (1.7350) acc1: 5.4688 (4.0130) acc5: 100.0000 (100.0000) time: 0.2595 data: 0.1667 max mem: 3531 +Test: [ 70/2387] eta: 0:12:13 acc1_pos: 22.135 acc5_pos: 100.000 loss: 1.6544 (1.7243) acc1: 3.1250 (4.2180) acc5: 100.0000 (100.0000) time: 0.3084 data: 0.2133 max mem: 3531 +Test: [ 80/2387] eta: 0:11:42 acc1_pos: 10.938 acc5_pos: 100.000 loss: 1.6389 (1.7115) acc1: 3.1250 (4.4399) acc5: 100.0000 (100.0000) time: 0.2699 data: 0.1746 max mem: 3531 +Test: [ 90/2387] eta: 0:11:33 acc1_pos: 21.094 acc5_pos: 100.000 loss: 1.5226 (1.6893) acc1: 7.8125 (5.2885) acc5: 100.0000 (100.0000) time: 0.2493 data: 0.1541 max mem: 3531 +Test: [ 100/2387] eta: 0:11:29 acc1_pos: 0.521 acc5_pos: 100.000 loss: 1.5397 (1.6876) acc1: 7.8125 (5.3347) acc5: 100.0000 (100.0000) time: 0.2899 data: 0.1920 max mem: 3531 +Test: [ 110/2387] eta: 0:11:13 acc1_pos: 14.062 acc5_pos: 100.000 loss: 1.6013 (1.6757) acc1: 5.9896 (5.6494) acc5: 100.0000 (100.0000) time: 0.2699 data: 0.1732 max mem: 3531 +Test: [ 120/2387] eta: 0:11:07 acc1_pos: 8.854 acc5_pos: 100.000 loss: 1.5324 (1.6624) acc1: 8.8542 (5.9724) acc5: 100.0000 (100.0000) time: 0.2599 data: 0.1645 max mem: 3531 +Test: [ 130/2387] eta: 0:11:30 acc1_pos: 9.635 acc5_pos: 100.000 loss: 1.5470 (1.6596) acc1: 8.0729 (5.9518) acc5: 100.0000 (100.0000) time: 0.3599 data: 0.2650 max mem: 3531 +Test: [ 140/2387] eta: 0:10:55 acc1_pos: 10.677 acc5_pos: 100.000 loss: 1.5817 (1.6538) acc1: 4.9479 (6.1281) acc5: 100.0000 (100.0000) time: 0.2749 data: 0.1763 max mem: 3531 +Test: [ 150/2387] eta: 0:10:51 acc1_pos: 4.167 acc5_pos: 100.000 loss: 1.5345 (1.6458) acc1: 8.5938 (6.3155) acc5: 100.0000 (100.0000) time: 0.1949 data: 0.0974 max mem: 3531 +Test: [ 160/2387] eta: 0:10:41 acc1_pos: 0.913 acc5_pos: 100.000 loss: 1.4562 (1.6086) acc1: 11.1979 (8.5226) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2599 data: 0.1626 max mem: 3531 +Test: [ 170/2387] eta: 0:10:25 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1594 (1.5235) acc1: 97.9167 (13.8371) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2149 data: 0.1174 max mem: 3531 +Test: [ 180/2387] eta: 0:10:06 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1507 (1.4477) acc1: 99.4792 (18.5630) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1699 data: 0.0744 max mem: 3531 +Test: [ 190/2387] eta: 0:09:52 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1497 (1.3797) acc1: 99.4792 (22.8076) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1649 data: 0.0666 max mem: 3531 +Test: [ 200/2387] eta: 0:09:40 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1505 (1.3187) acc1: 99.4792 (26.6104) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1804 data: 0.0856 max mem: 3531 +Test: [ 210/2387] eta: 0:09:31 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1539 (1.2635) acc1: 99.2188 (30.0479) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1899 data: 0.0955 max mem: 3532 +Test: [ 220/2387] eta: 0:09:21 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1525 (1.2132) acc1: 99.2188 (33.1813) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1943 data: 0.0994 max mem: 3532 +Test: [ 230/2387] eta: 0:09:09 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1525 (1.1674) acc1: 99.2188 (36.0469) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1749 data: 0.0794 max mem: 3532 +Test: [ 240/2387] eta: 0:09:01 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1544 (1.1254) acc1: 99.4792 (38.6746) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1749 data: 0.0781 max mem: 3532 +Test: [ 250/2387] eta: 0:08:51 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1445 (1.0842) acc1: 99.7396 (41.1147) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0834 max mem: 3532 +Test: [ 260/2387] eta: 0:08:42 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0827 (1.0459) acc1: 100.0000 (43.3709) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0725 max mem: 3532 +Test: [ 270/2387] eta: 0:08:34 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0803 (1.0103) acc1: 100.0000 (45.4605) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0749 max mem: 3532 +Test: [ 280/2387] eta: 0:08:26 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0803 (0.9773) acc1: 100.0000 (47.4014) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1703 data: 0.0782 max mem: 3532 +Test: [ 290/2387] eta: 0:08:18 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0795 (0.9465) acc1: 100.0000 (49.2089) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0774 max mem: 3532 +Test: [ 300/2387] eta: 0:08:11 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0798 (0.9178) acc1: 100.0000 (50.8963) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1665 data: 0.0733 max mem: 3532 +Test: [ 310/2387] eta: 0:08:05 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0839 (0.8945) acc1: 100.0000 (52.4266) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1699 data: 0.0779 max mem: 3532 +Test: [ 320/2387] eta: 0:07:59 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.3297 (0.8771) acc1: 96.8750 (53.8081) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1794 data: 0.0855 max mem: 3532 +Test: [ 330/2387] eta: 0:07:53 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.3303 (0.8606) acc1: 96.8750 (55.1147) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1820 data: 0.0882 max mem: 3532 +Test: [ 340/2387] eta: 0:07:48 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.3286 (0.8450) acc1: 97.1354 (56.3516) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1754 data: 0.0826 max mem: 3532 +Test: [ 350/2387] eta: 0:07:44 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.3230 (0.8255) acc1: 98.6979 (57.5818) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1889 data: 0.0961 max mem: 3532 +Test: [ 360/2387] eta: 0:07:41 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1349 (0.8063) acc1: 100.0000 (58.7503) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2131 data: 0.1221 max mem: 3532 +Test: [ 370/2387] eta: 0:07:38 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1320 (0.7881) acc1: 100.0000 (59.8586) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2139 data: 0.1231 max mem: 3532 +Test: [ 380/2387] eta: 0:07:35 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1311 (0.7709) acc1: 100.0000 (60.9047) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2076 data: 0.1148 max mem: 3532 +Test: [ 390/2387] eta: 0:07:32 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1327 (0.7546) acc1: 99.7396 (61.8986) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2097 data: 0.1166 max mem: 3532 +Test: [ 400/2387] eta: 0:07:29 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1329 (0.7391) acc1: 99.7396 (62.8422) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2119 data: 0.1201 max mem: 3532 +Test: [ 410/2387] eta: 0:07:26 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1322 (0.7244) acc1: 99.7396 (63.7375) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2173 data: 0.1257 max mem: 3532 +Test: [ 420/2387] eta: 0:07:22 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1319 (0.7103) acc1: 99.7396 (64.5920) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2072 data: 0.1157 max mem: 3532 +Test: [ 430/2387] eta: 0:07:19 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1338 (0.6970) acc1: 99.7396 (65.4057) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1993 data: 0.1066 max mem: 3532 +Test: [ 440/2387] eta: 0:07:15 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1347 (0.6838) acc1: 99.7396 (66.1866) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1976 data: 0.1008 max mem: 3533 +Test: [ 450/2387] eta: 0:07:12 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0977 (0.6708) acc1: 100.0000 (66.9363) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1945 data: 0.0983 max mem: 3533 +Test: [ 460/2387] eta: 0:07:09 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0977 (0.6584) acc1: 100.0000 (67.6535) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1998 data: 0.1077 max mem: 3533 +Test: [ 470/2387] eta: 0:07:06 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0978 (0.6465) acc1: 100.0000 (68.3403) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2035 data: 0.1100 max mem: 3533 +Test: [ 480/2387] eta: 0:07:02 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0980 (0.6351) acc1: 100.0000 (68.9985) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1973 data: 0.1029 max mem: 3533 +Test: [ 490/2387] eta: 0:06:59 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0982 (0.6242) acc1: 100.0000 (69.6299) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1909 data: 0.0989 max mem: 3533 +Test: [ 500/2387] eta: 0:06:56 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0982 (0.6137) acc1: 100.0000 (70.2361) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2001 data: 0.1096 max mem: 3533 +Test: [ 510/2387] eta: 0:06:52 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0981 (0.6037) acc1: 100.0000 (70.8186) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1889 data: 0.0982 max mem: 3533 +Test: [ 520/2387] eta: 0:06:49 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1126 (0.5943) acc1: 100.0000 (71.3787) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1773 data: 0.0847 max mem: 3533 +Test: [ 530/2387] eta: 0:06:45 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1123 (0.5852) acc1: 100.0000 (71.9177) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0801 max mem: 3533 +Test: [ 540/2387] eta: 0:06:42 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1110 (0.5764) acc1: 100.0000 (72.4367) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0956 max mem: 3533 +Test: [ 550/2387] eta: 0:06:39 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1107 (0.5679) acc1: 100.0000 (72.9370) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1999 data: 0.1040 max mem: 3533 +Test: [ 560/2387] eta: 0:06:36 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1107 (0.5598) acc1: 100.0000 (73.4194) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1857 data: 0.0893 max mem: 3533 +Test: [ 570/2387] eta: 0:06:32 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1109 (0.5519) acc1: 100.0000 (73.8849) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1769 data: 0.0829 max mem: 3533 +Test: [ 580/2387] eta: 0:06:30 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1102 (0.5443) acc1: 100.0000 (74.3344) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1995 data: 0.1065 max mem: 3533 +Test: [ 590/2387] eta: 0:06:28 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1108 (0.5370) acc1: 100.0000 (74.7687) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2187 data: 0.1257 max mem: 3533 +Test: [ 600/2387] eta: 0:06:25 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1121 (0.5300) acc1: 100.0000 (75.1881) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1895 data: 0.0953 max mem: 3533 +Test: [ 610/2387] eta: 0:06:26 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1233 (0.5235) acc1: 100.0000 (75.5903) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2485 data: 0.1558 max mem: 3533 +Test: [ 620/2387] eta: 0:06:23 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1328 (0.5173) acc1: 99.7396 (75.9762) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2626 data: 0.1716 max mem: 3533 +Test: [ 630/2387] eta: 0:06:18 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1342 (0.5112) acc1: 99.7396 (76.3512) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1509 data: 0.0592 max mem: 3533 +Test: [ 640/2387] eta: 0:06:18 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1359 (0.5054) acc1: 99.4792 (76.7104) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2097 data: 0.1183 max mem: 3533 +Test: [ 650/2387] eta: 0:06:17 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1368 (0.4998) acc1: 99.4792 (77.0585) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2795 data: 0.1879 max mem: 3533 +Test: [ 660/2387] eta: 0:06:14 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1356 (0.4942) acc1: 99.4792 (77.4013) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2103 data: 0.1169 max mem: 3533 +Test: [ 670/2387] eta: 0:06:11 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1350 (0.4889) acc1: 99.7396 (77.7322) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1949 data: 0.1024 max mem: 3534 +Test: [ 680/2387] eta: 0:06:08 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1312 (0.4836) acc1: 99.7396 (78.0577) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1929 data: 0.1007 max mem: 3534 +Test: [ 690/2387] eta: 0:06:06 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1306 (0.4811) acc1: 99.7396 (78.1740) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.1899 data: 0.0972 max mem: 3534 +Test: [ 700/2387] eta: 0:06:03 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.3703 (0.4797) acc1: 80.9896 (78.2108) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1991 data: 0.1052 max mem: 3534 +Test: [ 710/2387] eta: 0:06:00 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.3760 (0.4783) acc1: 80.7292 (78.2506) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1911 data: 0.0965 max mem: 3534 +Test: [ 720/2387] eta: 0:05:58 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.3719 (0.4769) acc1: 81.5104 (78.2919) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.1948 data: 0.1027 max mem: 3534 +Test: [ 730/2387] eta: 0:05:55 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.3719 (0.4755) acc1: 81.2500 (78.3359) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.1852 data: 0.0933 max mem: 3534 +Test: [ 740/2387] eta: 0:05:52 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.3744 (0.4742) acc1: 81.2500 (78.3714) acc5: 100.0000 (100.0000) acc1_neg: 77.344 acc5_neg: 100.000 time: 0.1921 data: 0.1003 max mem: 3534 +Test: [ 750/2387] eta: 0:05:50 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.3676 (0.4728) acc1: 81.5104 (78.4170) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.2016 data: 0.1109 max mem: 3534 +Test: [ 760/2387] eta: 0:05:46 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.3561 (0.4707) acc1: 82.2917 (78.5144) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1693 data: 0.0781 max mem: 3534 +Test: [ 770/2387] eta: 0:05:44 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.2135 (0.4667) acc1: 92.1875 (78.7424) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1772 data: 0.0843 max mem: 3534 +Test: [ 780/2387] eta: 0:05:42 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1021 (0.4619) acc1: 100.0000 (79.0146) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2133 data: 0.1197 max mem: 3534 +Test: [ 790/2387] eta: 0:05:39 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0904 (0.4577) acc1: 100.0000 (79.2394) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.2044 data: 0.1105 max mem: 3534 +Test: [ 800/2387] eta: 0:05:37 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0904 (0.4532) acc1: 100.0000 (79.4970) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.1007 max mem: 3534 +Test: [ 810/2387] eta: 0:05:35 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0908 (0.4487) acc1: 100.0000 (79.7479) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2017 data: 0.1066 max mem: 3534 +Test: [ 820/2387] eta: 0:05:33 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1003 (0.4446) acc1: 100.0000 (79.9907) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2099 data: 0.1146 max mem: 3534 +Test: [ 830/2387] eta: 0:05:30 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1173 (0.4412) acc1: 99.7396 (80.1992) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2081 data: 0.1122 max mem: 3534 +Test: [ 840/2387] eta: 0:05:28 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1705 (0.4383) acc1: 98.6979 (80.4112) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2049 data: 0.1087 max mem: 3534 +Test: [ 850/2387] eta: 0:05:26 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.2002 (0.4357) acc1: 97.9167 (80.6034) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2023 data: 0.1087 max mem: 3534 +Test: [ 860/2387] eta: 0:05:23 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1893 (0.4328) acc1: 98.4375 (80.8172) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1872 data: 0.0954 max mem: 3534 +Test: [ 870/2387] eta: 0:05:21 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1845 (0.4300) acc1: 99.2188 (81.0234) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1997 data: 0.1077 max mem: 3534 +Test: [ 880/2387] eta: 0:05:18 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1807 (0.4274) acc1: 98.6979 (81.2169) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2076 data: 0.1109 max mem: 3534 +Test: [ 890/2387] eta: 0:05:16 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1558 (0.4243) acc1: 99.4792 (81.4251) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1944 data: 0.0953 max mem: 3534 +Test: [ 900/2387] eta: 0:05:15 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1762 (0.4263) acc1: 98.6979 (81.2564) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.2299 data: 0.1335 max mem: 3535 +Test: [ 910/2387] eta: 0:05:12 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.7059 (0.4294) acc1: 58.3333 (81.0019) acc5: 100.0000 (100.0000) acc1_neg: 55.990 acc5_neg: 100.000 time: 0.2132 data: 0.1180 max mem: 3535 +Test: [ 920/2387] eta: 0:05:09 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.7219 (0.4326) acc1: 56.5104 (80.7408) acc5: 100.0000 (100.0000) acc1_neg: 58.073 acc5_neg: 100.000 time: 0.1499 data: 0.0526 max mem: 3535 +Test: [ 930/2387] eta: 0:05:06 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.7241 (0.4356) acc1: 56.7708 (80.4934) acc5: 100.0000 (100.0000) acc1_neg: 53.646 acc5_neg: 100.000 time: 0.1612 data: 0.0652 max mem: 3535 +Test: [ 940/2387] eta: 0:05:04 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.7279 (0.4386) acc1: 56.5104 (80.2404) acc5: 100.0000 (100.0000) acc1_neg: 55.469 acc5_neg: 100.000 time: 0.1885 data: 0.0966 max mem: 3535 +Test: [ 950/2387] eta: 0:05:00 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.7063 (0.4414) acc1: 55.9896 (79.9887) acc5: 100.0000 (100.0000) acc1_neg: 58.073 acc5_neg: 100.000 time: 0.1617 data: 0.0711 max mem: 3535 +Test: [ 960/2387] eta: 0:04:57 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.7019 (0.4441) acc1: 57.0312 (79.7653) acc5: 100.0000 (100.0000) acc1_neg: 62.240 acc5_neg: 100.000 time: 0.1367 data: 0.0465 max mem: 3535 +Test: [ 970/2387] eta: 0:04:55 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.7080 (0.4470) acc1: 57.0312 (79.5263) acc5: 100.0000 (100.0000) acc1_neg: 57.292 acc5_neg: 100.000 time: 0.1596 data: 0.0649 max mem: 3535 +Test: [ 980/2387] eta: 0:04:52 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.6872 (0.4460) acc1: 57.2917 (79.5667) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.1745 data: 0.0761 max mem: 3535 +Test: [ 990/2387] eta: 0:04:49 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1218 (0.4427) acc1: 99.7396 (79.7600) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1618 data: 0.0666 max mem: 3535 +Test: [1000/2387] eta: 0:04:47 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0938 (0.4394) acc1: 99.7396 (79.9471) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1595 data: 0.0665 max mem: 3535 +Test: [1010/2387] eta: 0:04:44 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1053 (0.4363) acc1: 99.7396 (80.1329) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1749 data: 0.0802 max mem: 3535 +Test: [1020/2387] eta: 0:04:42 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1191 (0.4334) acc1: 99.7396 (80.3101) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1849 data: 0.0883 max mem: 3535 +Test: [1030/2387] eta: 0:04:39 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1072 (0.4301) acc1: 99.7396 (80.4991) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0791 max mem: 3535 +Test: [1040/2387] eta: 0:04:37 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0976 (0.4270) acc1: 100.0000 (80.6839) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1599 data: 0.0648 max mem: 3535 +Test: [1050/2387] eta: 0:04:34 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0954 (0.4238) acc1: 100.0000 (80.8674) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0738 max mem: 3535 +Test: [1060/2387] eta: 0:04:32 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0931 (0.4208) acc1: 100.0000 (81.0475) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0896 max mem: 3535 +Test: [1070/2387] eta: 0:04:29 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0991 (0.4179) acc1: 100.0000 (81.2220) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1754 data: 0.0812 max mem: 3535 +Test: [1080/2387] eta: 0:04:27 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1209 (0.4154) acc1: 100.0000 (81.3830) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1649 data: 0.0688 max mem: 3535 +Test: [1090/2387] eta: 0:04:25 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1155 (0.4125) acc1: 100.0000 (81.5527) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1794 data: 0.0814 max mem: 3535 +Test: [1100/2387] eta: 0:04:22 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1025 (0.4098) acc1: 100.0000 (81.7197) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0742 max mem: 3535 +Test: [1110/2387] eta: 0:04:19 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1051 (0.4073) acc1: 100.0000 (81.8632) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1549 data: 0.0595 max mem: 3535 +Test: [1120/2387] eta: 0:04:17 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1154 (0.4047) acc1: 99.7396 (82.0208) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1668 data: 0.0709 max mem: 3536 +Test: [1130/2387] eta: 0:04:15 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1133 (0.4020) acc1: 99.7396 (82.1772) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1949 data: 0.0995 max mem: 3536 +Test: [1140/2387] eta: 0:04:13 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1142 (0.3997) acc1: 100.0000 (82.3280) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1930 data: 0.0969 max mem: 3536 +Test: [1150/2387] eta: 0:04:11 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1387 (0.3977) acc1: 100.0000 (82.4648) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1799 data: 0.0853 max mem: 3536 +Test: [1160/2387] eta: 0:04:09 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1610 (0.3956) acc1: 99.4792 (82.6091) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2066 data: 0.1138 max mem: 3536 +Test: [1170/2387] eta: 0:04:07 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1582 (0.3937) acc1: 99.2188 (82.7513) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2018 data: 0.1097 max mem: 3536 +Test: [1180/2387] eta: 0:04:05 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1631 (0.3918) acc1: 99.2188 (82.8884) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1935 data: 0.1018 max mem: 3536 +Test: [1190/2387] eta: 0:04:03 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1531 (0.3898) acc1: 99.7396 (83.0287) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2080 data: 0.1164 max mem: 3536 +Test: [1200/2387] eta: 0:04:01 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1259 (0.3876) acc1: 99.7396 (83.1692) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2054 data: 0.1142 max mem: 3536 +Test: [1210/2387] eta: 0:03:58 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1233 (0.3854) acc1: 100.0000 (83.3077) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2003 data: 0.1094 max mem: 3536 +Test: [1220/2387] eta: 0:03:56 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1002 (0.3830) acc1: 100.0000 (83.4436) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1954 data: 0.1040 max mem: 3536 +Test: [1230/2387] eta: 0:03:54 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0867 (0.3806) acc1: 100.0000 (83.5777) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1960 data: 0.1031 max mem: 3536 +Test: [1240/2387] eta: 0:03:52 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0879 (0.3782) acc1: 100.0000 (83.7098) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1986 data: 0.1053 max mem: 3536 +Test: [1250/2387] eta: 0:03:50 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0834 (0.3759) acc1: 100.0000 (83.8396) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1885 data: 0.0971 max mem: 3536 +Test: [1260/2387] eta: 0:03:48 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0812 (0.3736) acc1: 100.0000 (83.9673) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1807 data: 0.0880 max mem: 3536 +Test: [1270/2387] eta: 0:03:45 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0794 (0.3713) acc1: 100.0000 (84.0935) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1803 data: 0.0869 max mem: 3536 +Test: [1280/2387] eta: 0:03:43 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0797 (0.3690) acc1: 100.0000 (84.2170) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1854 data: 0.0938 max mem: 3536 +Test: [1290/2387] eta: 0:03:41 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0825 (0.3669) acc1: 100.0000 (84.3312) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1920 data: 0.0961 max mem: 3536 +Test: [1300/2387] eta: 0:03:39 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0854 (0.3648) acc1: 100.0000 (84.4487) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1936 data: 0.0970 max mem: 3536 +Test: [1310/2387] eta: 0:03:37 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0859 (0.3627) acc1: 100.0000 (84.5631) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1950 data: 0.1013 max mem: 3536 +Test: [1320/2387] eta: 0:03:35 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0883 (0.3607) acc1: 100.0000 (84.6731) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1928 data: 0.0992 max mem: 3536 +Test: [1330/2387] eta: 0:03:33 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0937 (0.3588) acc1: 100.0000 (84.7845) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2003 data: 0.1065 max mem: 3536 +Test: [1340/2387] eta: 0:03:31 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1135 (0.3576) acc1: 99.4792 (84.8620) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2087 data: 0.1157 max mem: 3536 +Test: [1350/2387] eta: 0:03:29 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0806 (0.3554) acc1: 100.0000 (84.9741) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1984 data: 0.1069 max mem: 3537 +Test: [1360/2387] eta: 0:03:27 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0731 (0.3534) acc1: 100.0000 (85.0845) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2033 data: 0.1114 max mem: 3537 +Test: [1370/2387] eta: 0:03:25 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0727 (0.3513) acc1: 100.0000 (85.1933) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2205 data: 0.1288 max mem: 3537 +Test: [1380/2387] eta: 0:03:23 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0715 (0.3493) acc1: 100.0000 (85.3005) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1864 data: 0.0940 max mem: 3537 +Test: [1390/2387] eta: 0:03:21 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0708 (0.3473) acc1: 100.0000 (85.4062) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1628 data: 0.0700 max mem: 3537 +Test: [1400/2387] eta: 0:03:19 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0685 (0.3453) acc1: 100.0000 (85.5104) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2035 data: 0.1084 max mem: 3537 +Test: [1410/2387] eta: 0:03:17 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0682 (0.3435) acc1: 100.0000 (85.6110) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2149 data: 0.1191 max mem: 3537 +Test: [1420/2387] eta: 0:03:15 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0985 (0.3420) acc1: 100.0000 (85.6963) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1905 data: 0.0957 max mem: 3537 +Test: [1430/2387] eta: 0:03:12 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1080 (0.3404) acc1: 99.7396 (85.7925) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1671 data: 0.0737 max mem: 3537 +Test: [1440/2387] eta: 0:03:10 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1080 (0.3388) acc1: 99.7396 (85.8891) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1641 data: 0.0727 max mem: 3537 +Test: [1450/2387] eta: 0:03:08 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1199 (0.3375) acc1: 100.0000 (85.9790) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1827 data: 0.0913 max mem: 3537 +Test: [1460/2387] eta: 0:03:06 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1207 (0.3359) acc1: 100.0000 (86.0740) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1817 data: 0.0900 max mem: 3537 +Test: [1470/2387] eta: 0:03:04 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1028 (0.3343) acc1: 100.0000 (86.1684) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1838 data: 0.0892 max mem: 3537 +Test: [1480/2387] eta: 0:03:01 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1036 (0.3328) acc1: 100.0000 (86.2596) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1804 data: 0.0865 max mem: 3537 +Test: [1490/2387] eta: 0:02:59 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1065 (0.3313) acc1: 100.0000 (86.3506) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1624 data: 0.0694 max mem: 3537 +Test: [1500/2387] eta: 0:02:57 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1074 (0.3298) acc1: 100.0000 (86.4401) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1683 data: 0.0736 max mem: 3537 +Test: [1510/2387] eta: 0:02:55 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1153 (0.3285) acc1: 100.0000 (86.5293) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1711 data: 0.0784 max mem: 3537 +Test: [1520/2387] eta: 0:02:53 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1408 (0.3275) acc1: 99.7396 (86.6073) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1608 data: 0.0703 max mem: 3537 +Test: [1530/2387] eta: 0:02:50 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1791 (0.3266) acc1: 98.1771 (86.6793) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1597 data: 0.0695 max mem: 3537 +Test: [1540/2387] eta: 0:02:48 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1783 (0.3256) acc1: 98.1771 (86.7612) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1610 data: 0.0703 max mem: 3537 +Test: [1550/2387] eta: 0:02:46 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1757 (0.3247) acc1: 99.7396 (86.8403) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1594 data: 0.0680 max mem: 3537 +Test: [1560/2387] eta: 0:02:44 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1757 (0.3237) acc1: 99.2188 (86.9138) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1622 data: 0.0697 max mem: 3537 +Test: [1570/2387] eta: 0:02:42 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1530 (0.3226) acc1: 99.7396 (86.9957) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1730 data: 0.0785 max mem: 3537 +Test: [1580/2387] eta: 0:02:40 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1259 (0.3212) acc1: 100.0000 (87.0770) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1654 data: 0.0711 max mem: 3538 +Test: [1590/2387] eta: 0:02:37 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0956 (0.3198) acc1: 100.0000 (87.1581) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1549 data: 0.0621 max mem: 3538 +Test: [1600/2387] eta: 0:02:35 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0955 (0.3186) acc1: 100.0000 (87.2245) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1765 data: 0.0841 max mem: 3538 +Test: [1610/2387] eta: 0:02:33 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0926 (0.3172) acc1: 100.0000 (87.3031) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1959 data: 0.1043 max mem: 3538 +Test: [1620/2387] eta: 0:02:31 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0947 (0.3158) acc1: 100.0000 (87.3787) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1978 data: 0.1052 max mem: 3538 +Test: [1630/2387] eta: 0:02:29 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1059 (0.3147) acc1: 99.7396 (87.4489) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1903 data: 0.0965 max mem: 3538 +Test: [1640/2387] eta: 0:02:27 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1518 (0.3138) acc1: 99.2188 (87.5138) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1849 data: 0.0924 max mem: 3538 +Test: [1650/2387] eta: 0:02:25 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1415 (0.3127) acc1: 99.2188 (87.5819) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1797 data: 0.0869 max mem: 3538 +Test: [1660/2387] eta: 0:02:23 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1408 (0.3118) acc1: 99.2188 (87.6489) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1709 data: 0.0779 max mem: 3538 +Test: [1670/2387] eta: 0:02:21 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1408 (0.3107) acc1: 99.4792 (87.7182) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1712 data: 0.0792 max mem: 3538 +Test: [1680/2387] eta: 0:02:19 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1403 (0.3099) acc1: 99.7396 (87.7781) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1689 data: 0.0770 max mem: 3538 +Test: [1690/2387] eta: 0:02:17 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1529 (0.3090) acc1: 98.9583 (87.8383) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1693 data: 0.0758 max mem: 3538 +Test: [1700/2387] eta: 0:02:15 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1421 (0.3079) acc1: 99.2188 (87.9065) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1727 data: 0.0780 max mem: 3538 +Test: [1710/2387] eta: 0:02:13 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1115 (0.3067) acc1: 100.0000 (87.9759) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1803 data: 0.0876 max mem: 3538 +Test: [1720/2387] eta: 0:02:11 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0929 (0.3055) acc1: 100.0000 (88.0450) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1782 data: 0.0865 max mem: 3538 +Test: [1730/2387] eta: 0:02:09 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0956 (0.3043) acc1: 100.0000 (88.1126) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1725 data: 0.0794 max mem: 3538 +Test: [1740/2387] eta: 0:02:07 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0965 (0.3030) acc1: 100.0000 (88.1803) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1744 data: 0.0808 max mem: 3538 +Test: [1750/2387] eta: 0:02:05 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0850 (0.3018) acc1: 100.0000 (88.2472) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1719 data: 0.0797 max mem: 3538 +Test: [1760/2387] eta: 0:02:02 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0850 (0.3006) acc1: 100.0000 (88.3138) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1644 data: 0.0728 max mem: 3538 +Test: [1770/2387] eta: 0:02:00 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0861 (0.2995) acc1: 100.0000 (88.3752) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1692 data: 0.0770 max mem: 3538 +Test: [1780/2387] eta: 0:01:58 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0983 (0.2984) acc1: 99.4792 (88.4387) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1886 data: 0.0947 max mem: 3538 +Test: [1790/2387] eta: 0:01:56 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0955 (0.2973) acc1: 99.7396 (88.5014) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1875 data: 0.0905 max mem: 3538 +Test: [1800/2387] eta: 0:01:54 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0969 (0.2962) acc1: 99.7396 (88.5631) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1836 data: 0.0860 max mem: 3538 +Test: [1810/2387] eta: 0:01:53 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0897 (0.2950) acc1: 100.0000 (88.6259) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1952 data: 0.1012 max mem: 3539 +Test: [1820/2387] eta: 0:01:51 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0890 (0.2939) acc1: 100.0000 (88.6867) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2046 data: 0.1131 max mem: 3539 +Test: [1830/2387] eta: 0:01:49 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0911 (0.2929) acc1: 100.0000 (88.7485) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2045 data: 0.1122 max mem: 3539 +Test: [1840/2387] eta: 0:01:47 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1230 (0.2919) acc1: 100.0000 (88.8096) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2117 data: 0.1184 max mem: 3539 +Test: [1850/2387] eta: 0:01:45 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1248 (0.2910) acc1: 100.0000 (88.8700) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2003 data: 0.1082 max mem: 3539 +Test: [1860/2387] eta: 0:01:43 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1246 (0.2901) acc1: 100.0000 (88.9298) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1707 data: 0.0795 max mem: 3539 +Test: [1870/2387] eta: 0:01:41 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1237 (0.2893) acc1: 100.0000 (88.9890) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1713 data: 0.0796 max mem: 3539 +Test: [1880/2387] eta: 0:01:39 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1236 (0.2884) acc1: 100.0000 (89.0475) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1759 data: 0.0846 max mem: 3539 +Test: [1890/2387] eta: 0:01:37 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1238 (0.2875) acc1: 100.0000 (89.1055) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1808 data: 0.0888 max mem: 3539 +Test: [1900/2387] eta: 0:01:35 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1246 (0.2867) acc1: 100.0000 (89.1628) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1948 data: 0.1013 max mem: 3539 +Test: [1910/2387] eta: 0:01:33 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1273 (0.2860) acc1: 100.0000 (89.2195) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1802 data: 0.0878 max mem: 3539 +Test: [1920/2387] eta: 0:01:31 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1161 (0.2850) acc1: 100.0000 (89.2755) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1780 data: 0.0839 max mem: 3539 +Test: [1930/2387] eta: 0:01:29 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0936 (0.2840) acc1: 100.0000 (89.3310) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1783 data: 0.0828 max mem: 3539 +Test: [1940/2387] eta: 0:01:27 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0936 (0.2833) acc1: 100.0000 (89.3511) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1750 data: 0.0812 max mem: 3539 +Test: [1950/2387] eta: 0:01:25 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0952 (0.2823) acc1: 100.0000 (89.4057) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1735 data: 0.0806 max mem: 3539 +Test: [1960/2387] eta: 0:01:23 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0950 (0.2815) acc1: 100.0000 (89.4544) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1660 data: 0.0736 max mem: 3539 +Test: [1970/2387] eta: 0:01:21 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1152 (0.2809) acc1: 100.0000 (89.4998) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1747 data: 0.0818 max mem: 3539 +Test: [1980/2387] eta: 0:01:19 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1576 (0.2803) acc1: 99.7396 (89.5411) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1764 data: 0.0832 max mem: 3539 +Test: [1990/2387] eta: 0:01:17 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1576 (0.2797) acc1: 99.4792 (89.5903) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1716 data: 0.0796 max mem: 3539 +Test: [2000/2387] eta: 0:01:15 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1590 (0.2791) acc1: 99.4792 (89.6388) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1855 data: 0.0945 max mem: 3539 +Test: [2010/2387] eta: 0:01:13 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1602 (0.2785) acc1: 99.4792 (89.6876) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1835 data: 0.0899 max mem: 3539 +Test: [2020/2387] eta: 0:01:11 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1604 (0.2779) acc1: 99.2188 (89.7350) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1767 data: 0.0812 max mem: 3539 +Test: [2030/2387] eta: 0:01:09 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1597 (0.2773) acc1: 99.2188 (89.7814) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1824 data: 0.0890 max mem: 3540 +Test: [2040/2387] eta: 0:01:07 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1577 (0.2767) acc1: 99.2188 (89.8290) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1796 data: 0.0874 max mem: 3540 +Test: [2050/2387] eta: 0:01:05 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1556 (0.2761) acc1: 99.7396 (89.8773) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1871 data: 0.0948 max mem: 3540 +Test: [2060/2387] eta: 0:01:03 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1524 (0.2755) acc1: 99.7396 (89.9244) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2029 data: 0.1097 max mem: 3540 +Test: [2070/2387] eta: 0:01:01 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1504 (0.2749) acc1: 99.7396 (89.9716) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1821 data: 0.0888 max mem: 3540 +Test: [2080/2387] eta: 0:00:59 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1479 (0.2743) acc1: 99.7396 (90.0181) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1773 data: 0.0855 max mem: 3540 +Test: [2090/2387] eta: 0:00:57 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1466 (0.2737) acc1: 99.7396 (90.0641) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1887 data: 0.0967 max mem: 3540 +Test: [2100/2387] eta: 0:00:55 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1476 (0.2731) acc1: 99.7396 (90.1100) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1700 data: 0.0740 max mem: 3540 +Test: [2110/2387] eta: 0:00:53 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1476 (0.2725) acc1: 99.7396 (90.1555) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1689 data: 0.0734 max mem: 3540 +Test: [2120/2387] eta: 0:00:51 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1403 (0.2718) acc1: 100.0000 (90.2012) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1670 data: 0.0723 max mem: 3540 +Test: [2130/2387] eta: 0:00:49 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1004 (0.2709) acc1: 100.0000 (90.2471) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1854 data: 0.0906 max mem: 3540 +Test: [2140/2387] eta: 0:00:47 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1016 (0.2703) acc1: 100.0000 (90.2749) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2060 data: 0.1142 max mem: 3540 +Test: [2150/2387] eta: 0:00:45 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1049 (0.2696) acc1: 100.0000 (90.3197) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2109 data: 0.1194 max mem: 3540 +Test: [2160/2387] eta: 0:00:44 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0999 (0.2688) acc1: 100.0000 (90.3624) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2124 data: 0.1214 max mem: 3540 +Test: [2170/2387] eta: 0:00:42 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1127 (0.2682) acc1: 100.0000 (90.4018) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2003 data: 0.1090 max mem: 3540 +Test: [2180/2387] eta: 0:00:40 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1487 (0.2695) acc1: 97.1354 (90.3119) acc5: 100.0000 (100.0000) acc1_neg: 31.771 acc5_neg: 100.000 time: 0.2017 data: 0.1083 max mem: 3540 +Test: [2190/2387] eta: 0:00:38 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.2222 (0.2697) acc1: 92.7083 (90.2936) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2090 data: 0.1149 max mem: 3540 +Test: [2200/2387] eta: 0:00:36 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1042 (0.2689) acc1: 99.7396 (90.3375) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2118 data: 0.1179 max mem: 3540 +Test: [2210/2387] eta: 0:00:34 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1017 (0.2682) acc1: 100.0000 (90.3807) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2040 data: 0.1105 max mem: 3540 +Test: [2220/2387] eta: 0:00:32 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1052 (0.2675) acc1: 100.0000 (90.4222) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2130 data: 0.1209 max mem: 3540 +Test: [2230/2387] eta: 0:00:30 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1031 (0.2667) acc1: 100.0000 (90.4642) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2206 data: 0.1283 max mem: 3540 +Test: [2240/2387] eta: 0:00:28 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0893 (0.2660) acc1: 100.0000 (90.5032) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2118 data: 0.1196 max mem: 3540 +Test: [2250/2387] eta: 0:00:26 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1099 (0.2654) acc1: 100.0000 (90.5426) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2067 data: 0.1149 max mem: 3540 +Test: [2260/2387] eta: 0:00:24 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1100 (0.2647) acc1: 100.0000 (90.5845) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1903 data: 0.0977 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0922 (0.2639) acc1: 100.0000 (90.6259) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.1029 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0913 (0.2633) acc1: 100.0000 (90.6509) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1925 data: 0.1011 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0901 (0.2625) acc1: 100.0000 (90.6917) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1605 data: 0.0687 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0901 (0.2618) acc1: 100.0000 (90.7300) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1593 data: 0.0673 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.0993 (0.2612) acc1: 100.0000 (90.7678) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1807 data: 0.0895 max mem: 3541 +Test: [2320/2387] eta: 0:00:13 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1151 (0.2606) acc1: 100.0000 (90.8005) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1719 data: 0.0809 max mem: 3541 +Test: [2330/2387] eta: 0:00:11 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1175 (0.2600) acc1: 100.0000 (90.8397) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1600 data: 0.0693 max mem: 3541 +Test: [2340/2387] eta: 0:00:09 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1104 (0.2594) acc1: 100.0000 (90.8789) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1454 data: 0.0556 max mem: 3541 +Test: [2350/2387] eta: 0:00:07 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1111 (0.2590) acc1: 100.0000 (90.8973) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1313 data: 0.0415 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1172 (0.2584) acc1: 100.0000 (90.9356) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1382 data: 0.0483 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1123 (0.2579) acc1: 100.0000 (90.9719) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1366 data: 0.0468 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1463 (0.2575) acc1: 99.7396 (91.0015) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1133 data: 0.0235 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 0.913 acc5_pos: 100.000 loss: 0.1755 (0.2574) acc1: 98.6979 (91.0093) acc5: 100.0000 (100.0000) acc1_neg: 95.349 acc5_neg: 100.000 time: 0.1167 data: 0.0291 max mem: 3541 +Test: Total time: 0:07:39 (0.1924 s / it) +* Overall Acc@1 91.009 Acc@5 100.000 + target == real Acc@1 6.232 Acc@5 100.000 + target == fake Acc@1 96.963 Acc@5 100.000 + loss 0.257 + PR-AUC 0.9835 ROC-AUC 0.8011 EER 27.03% (thr=0.8672) +Accuracy of the network on the 916267 test images: 91.0% +[rank0]:[W726 03:54:41.607761607 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:55:01.798000 1178851 torch/distributed/run.py:766] +W0726 03:55:01.798000 1178851 torch/distributed/run.py:766] ***************************************** +W0726 03:55:01.798000 1178851 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:55:01.798000 1178851 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:55:29.063019161 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:55:30.496310955 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:55:30.726346454 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:55:30.758264063 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:55:30.866535825 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W726 03:55:30.866536665 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:55:30.882296351 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:55:30.916489903 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n23_seed6_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed6_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 0:59:24 acc1_pos: 2.604 acc5_pos: 100.000 loss: 1.6428 (1.6428) acc1: 2.6042 (2.6042) acc5: 100.0000 (100.0000) time: 6.8952 data: 5.2089 max mem: 3531 +Test: [ 10/517] eta: 0:05:59 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.7974 (1.7769) acc1: 2.6042 (2.9830) acc5: 100.0000 (100.0000) time: 0.7088 data: 0.4739 max mem: 3531 +Test: [ 20/517] eta: 0:03:26 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.7974 (1.7675) acc1: 1.3021 (2.8150) acc5: 100.0000 (100.0000) time: 0.0906 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:02:45 acc1_pos: 1.823 acc5_pos: 100.000 loss: 1.8467 (1.7861) acc1: 1.3021 (2.6042) acc5: 100.0000 (100.0000) time: 0.1364 data: 0.0451 max mem: 3531 +Test: [ 40/517] eta: 0:02:41 acc1_pos: 1.042 acc5_pos: 100.000 loss: 1.8558 (1.7981) acc1: 1.0417 (2.3819) acc5: 100.0000 (100.0000) time: 0.2599 data: 0.1627 max mem: 3531 +Test: [ 50/517] eta: 0:02:31 acc1_pos: 15.138 acc5_pos: 100.000 loss: 1.7589 (1.7048) acc1: 1.3021 (7.3683) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2990 data: 0.2010 max mem: 3531 +Test: [ 60/517] eta: 0:02:24 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.1765 (1.4527) acc1: 98.6979 (22.4940) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2699 data: 0.1714 max mem: 3531 +Test: [ 70/517] eta: 0:02:19 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.1465 (1.2679) acc1: 100.0000 (33.4104) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2850 data: 0.1878 max mem: 3531 +Test: [ 80/517] eta: 0:02:13 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.1392 (1.1291) acc1: 100.0000 (41.6313) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2749 data: 0.1804 max mem: 3531 +Test: [ 90/517] eta: 0:02:09 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.1432 (1.0206) acc1: 100.0000 (48.0454) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2698 data: 0.1713 max mem: 3531 +Test: [100/517] eta: 0:02:06 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.1383 (0.9322) acc1: 100.0000 (53.1895) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2900 data: 0.1886 max mem: 3531 +Test: [110/517] eta: 0:02:01 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.1334 (0.8607) acc1: 100.0000 (57.3996) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2799 data: 0.1824 max mem: 3531 +Test: [120/517] eta: 0:01:57 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.1378 (0.8011) acc1: 100.0000 (60.9203) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2551 data: 0.1639 max mem: 3531 +Test: [130/517] eta: 0:01:54 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.1387 (0.7502) acc1: 100.0000 (63.9035) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2699 data: 0.1771 max mem: 3531 +Test: [140/517] eta: 0:01:51 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.1297 (0.7059) acc1: 100.0000 (66.4635) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2946 data: 0.1998 max mem: 3531 +Test: [150/517] eta: 0:01:47 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.1276 (0.6679) acc1: 100.0000 (68.6827) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2749 data: 0.1809 max mem: 3531 +Test: [160/517] eta: 0:01:43 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.1371 (0.6349) acc1: 100.0000 (70.6279) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2649 data: 0.1689 max mem: 3531 +Test: [170/517] eta: 0:01:40 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.1354 (0.6055) acc1: 100.0000 (72.3456) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2699 data: 0.1724 max mem: 3531 +Test: [180/517] eta: 0:01:37 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.1265 (0.5790) acc1: 100.0000 (73.8734) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2799 data: 0.1820 max mem: 3531 +Test: [190/517] eta: 0:01:34 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.1380 (0.5566) acc1: 100.0000 (75.2413) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2706 data: 0.1740 max mem: 3531 +Test: [200/517] eta: 0:01:30 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.1470 (0.5357) acc1: 100.0000 (76.4731) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1613 max mem: 3531 +Test: [210/517] eta: 0:01:27 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.1393 (0.5170) acc1: 100.0000 (77.5832) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2691 data: 0.1728 max mem: 3532 +Test: [220/517] eta: 0:01:24 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.1393 (0.4998) acc1: 100.0000 (78.5963) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2799 data: 0.1852 max mem: 3532 +Test: [230/517] eta: 0:01:21 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.1436 (0.4847) acc1: 100.0000 (79.5229) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2655 data: 0.1727 max mem: 3532 +Test: [240/517] eta: 0:01:18 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.1562 (0.4714) acc1: 100.0000 (80.3726) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2498 data: 0.1560 max mem: 3532 +Test: [250/517] eta: 0:01:15 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.1573 (0.4582) acc1: 100.0000 (81.1545) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2693 data: 0.1743 max mem: 3532 +Test: [260/517] eta: 0:01:12 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.1272 (0.4456) acc1: 100.0000 (81.8766) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1561 max mem: 3532 +Test: [270/517] eta: 0:01:09 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.1466 (0.4348) acc1: 100.0000 (82.5454) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2500 data: 0.1510 max mem: 3532 +Test: [280/517] eta: 0:01:06 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.1509 (0.4248) acc1: 100.0000 (83.1665) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2749 data: 0.1804 max mem: 3532 +Test: [290/517] eta: 0:01:03 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.1496 (0.4154) acc1: 100.0000 (83.7450) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2799 data: 0.1871 max mem: 3532 +Test: [300/517] eta: 0:01:00 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.1454 (0.4063) acc1: 100.0000 (84.2850) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2649 data: 0.1683 max mem: 3532 +Test: [310/517] eta: 0:00:57 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.1428 (0.3979) acc1: 100.0000 (84.7903) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2599 data: 0.1641 max mem: 3532 +Test: [320/517] eta: 0:00:54 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.1416 (0.3895) acc1: 100.0000 (85.2641) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2749 data: 0.1791 max mem: 3532 +Test: [330/517] eta: 0:00:52 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.1171 (0.3812) acc1: 100.0000 (85.7093) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2749 data: 0.1796 max mem: 3532 +Test: [340/517] eta: 0:00:49 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.1152 (0.3738) acc1: 100.0000 (86.1261) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2749 data: 0.1837 max mem: 3532 +Test: [350/517] eta: 0:00:46 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.1485 (0.3676) acc1: 100.0000 (86.5207) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2649 data: 0.1704 max mem: 3532 +Test: [360/517] eta: 0:00:43 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.1495 (0.3615) acc1: 100.0000 (86.8933) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2399 data: 0.1448 max mem: 3532 +Test: [370/517] eta: 0:00:40 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.1321 (0.3550) acc1: 100.0000 (87.2459) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2499 data: 0.1564 max mem: 3532 +Test: [380/517] eta: 0:00:37 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.1210 (0.3489) acc1: 100.0000 (87.5800) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2849 data: 0.1882 max mem: 3532 +Test: [390/517] eta: 0:00:35 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.1247 (0.3435) acc1: 100.0000 (87.8970) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2899 data: 0.1928 max mem: 3532 +Test: [400/517] eta: 0:00:32 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.1243 (0.3381) acc1: 100.0000 (88.1988) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2699 data: 0.1749 max mem: 3532 +Test: [410/517] eta: 0:00:29 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.1304 (0.3332) acc1: 100.0000 (88.4859) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1598 max mem: 3532 +Test: [420/517] eta: 0:00:26 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.1422 (0.3287) acc1: 100.0000 (88.7594) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2649 data: 0.1686 max mem: 3532 +Test: [430/517] eta: 0:00:24 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.1415 (0.3243) acc1: 100.0000 (89.0196) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2799 data: 0.1832 max mem: 3532 +Test: [440/517] eta: 0:00:21 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.1415 (0.3202) acc1: 100.0000 (89.2674) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2799 data: 0.1834 max mem: 3533 +Test: [450/517] eta: 0:00:18 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.1427 (0.3162) acc1: 100.0000 (89.5054) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1570 max mem: 3533 +Test: [460/517] eta: 0:00:15 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.1419 (0.3125) acc1: 100.0000 (89.7325) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2699 data: 0.1716 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.1366 (0.3087) acc1: 100.0000 (89.9505) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2799 data: 0.1837 max mem: 3533 +Test: [480/517] eta: 0:00:10 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.1331 (0.3053) acc1: 100.0000 (90.1594) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2764 data: 0.1830 max mem: 3533 +Test: [490/517] eta: 0:00:07 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.1464 (0.3019) acc1: 100.0000 (90.3598) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2384 data: 0.1475 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.1487 (0.2991) acc1: 100.0000 (90.5522) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1808 data: 0.0910 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.1614 (0.2965) acc1: 100.0000 (90.7356) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1805 data: 0.0908 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 15.138 acc5_pos: 100.000 loss: 0.1526 (0.2947) acc1: 100.0000 (90.8384) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1737 data: 0.0824 max mem: 3533 +Test: Total time: 0:02:19 (0.2702 s / it) +* Overall Acc@1 90.838 Acc@5 100.000 + target == real Acc@1 2.729 Acc@5 100.000 + target == fake Acc@1 99.979 Acc@5 100.000 + loss 0.295 + PR-AUC 0.9469 ROC-AUC 0.6980 EER 36.48% (thr=0.8667) +Accuracy of the network on the 198427 test images: 90.8% +[rank0]:[W726 03:57:59.732765978 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 03:58:20.528000 1179428 torch/distributed/run.py:766] +W0726 03:58:20.528000 1179428 torch/distributed/run.py:766] ***************************************** +W0726 03:58:20.528000 1179428 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:58:20.528000 1179428 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:58:49.573634438 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:58:49.045215236 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:58:49.086757872 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:58:49.312654645 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:58:50.435316284 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 7): env:// +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:58:50.646951885 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W726 03:58:50.647220727 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:58:50.676923921 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n23_seed6_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed6_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:48:23 acc1_pos: 1.823 acc5_pos: 100.000 loss: 1.2574 (1.2574) acc1: 1.8229 (1.8229) acc5: 100.0000 (100.0000) time: 7.4652 data: 5.5398 max mem: 3531 +Test: [ 10/389] eta: 0:04:50 acc1_pos: 80.990 acc5_pos: 100.000 loss: 0.9162 (0.9413) acc1: 37.5000 (36.5294) acc5: 100.0000 (100.0000) time: 0.7671 data: 0.5040 max mem: 3531 +Test: [ 20/389] eta: 0:02:45 acc1_pos: 1.302 acc5_pos: 100.000 loss: 0.9162 (1.0141) acc1: 30.2083 (34.1518) acc5: 100.0000 (100.0000) time: 0.0986 data: 0.0004 max mem: 3531 +Test: [ 30/389] eta: 0:02:16 acc1_pos: 12.500 acc5_pos: 100.000 loss: 1.1610 (1.0974) acc1: 7.5521 (25.5376) acc5: 100.0000 (100.0000) time: 0.1655 data: 0.0666 max mem: 3531 +Test: [ 40/389] eta: 0:01:57 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.1558 (1.1299) acc1: 9.6354 (21.9322) acc5: 100.0000 (100.0000) time: 0.2199 data: 0.1240 max mem: 3531 +Test: [ 50/389] eta: 0:01:44 acc1_pos: 41.927 acc5_pos: 100.000 loss: 1.1209 (1.1363) acc1: 10.9375 (21.3184) acc5: 100.0000 (100.0000) time: 0.1993 data: 0.1030 max mem: 3531 +Test: [ 60/389] eta: 0:01:35 acc1_pos: 15.104 acc5_pos: 100.000 loss: 0.9947 (1.1118) acc1: 15.6250 (21.6530) acc5: 100.0000 (100.0000) time: 0.1899 data: 0.0926 max mem: 3531 +Test: [ 70/389] eta: 0:01:27 acc1_pos: 21.615 acc5_pos: 100.000 loss: 1.0232 (1.0998) acc1: 17.7083 (21.6256) acc5: 100.0000 (100.0000) time: 0.1878 data: 0.0943 max mem: 3531 +Test: [ 80/389] eta: 0:01:21 acc1_pos: 72.917 acc5_pos: 100.000 loss: 1.0347 (1.0821) acc1: 17.7083 (23.1289) acc5: 100.0000 (100.0000) time: 0.1816 data: 0.0909 max mem: 3531 +Test: [ 90/389] eta: 0:01:15 acc1_pos: 33.333 acc5_pos: 100.000 loss: 0.6850 (1.0346) acc1: 53.1250 (27.1120) acc5: 100.0000 (100.0000) time: 0.1748 data: 0.0839 max mem: 3531 +Test: [100/389] eta: 0:01:12 acc1_pos: 32.552 acc5_pos: 100.000 loss: 0.9203 (1.0374) acc1: 33.0729 (26.6579) acc5: 100.0000 (100.0000) time: 0.1959 data: 0.1045 max mem: 3531 +Test: [110/389] eta: 0:01:07 acc1_pos: 84.375 acc5_pos: 100.000 loss: 0.9940 (1.0348) acc1: 23.4375 (27.8411) acc5: 100.0000 (100.0000) time: 0.1879 data: 0.0967 max mem: 3531 +Test: [120/389] eta: 0:01:04 acc1_pos: 23.177 acc5_pos: 100.000 loss: 0.9176 (1.0217) acc1: 35.4167 (28.6222) acc5: 100.0000 (100.0000) time: 0.1885 data: 0.0974 max mem: 3531 +Test: [130/389] eta: 0:01:00 acc1_pos: 2.454 acc5_pos: 100.000 loss: 0.7492 (0.9857) acc1: 48.1771 (31.8603) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1841 data: 0.0921 max mem: 3531 +Test: [140/389] eta: 0:00:56 acc1_pos: 2.454 acc5_pos: 100.000 loss: 0.3254 (0.9431) acc1: 94.5312 (35.8230) acc5: 100.0000 (100.0000) acc1_neg: 57.552 acc5_neg: 100.000 time: 0.1575 data: 0.0645 max mem: 3531 +Test: [150/389] eta: 0:00:53 acc1_pos: 2.454 acc5_pos: 100.000 loss: 0.3034 (0.9052) acc1: 97.6562 (39.4764) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1669 data: 0.0746 max mem: 3531 +Test: [160/389] eta: 0:00:50 acc1_pos: 2.454 acc5_pos: 100.000 loss: 0.3034 (0.8664) acc1: 97.6562 (42.9493) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1726 data: 0.0805 max mem: 3531 +Test: [170/389] eta: 0:00:47 acc1_pos: 2.454 acc5_pos: 100.000 loss: 0.2032 (0.8315) acc1: 100.0000 (46.1531) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1803 data: 0.0889 max mem: 3531 +Test: [180/389] eta: 0:00:45 acc1_pos: 2.454 acc5_pos: 100.000 loss: 0.2300 (0.8026) acc1: 99.4792 (48.8303) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.1791 data: 0.0887 max mem: 3531 +Test: [190/389] eta: 0:00:42 acc1_pos: 2.454 acc5_pos: 100.000 loss: 0.2984 (0.7831) acc1: 96.3542 (50.6599) acc5: 100.0000 (100.0000) acc1_neg: 67.969 acc5_neg: 100.000 time: 0.1687 data: 0.0778 max mem: 3531 +Test: [200/389] eta: 0:00:39 acc1_pos: 2.454 acc5_pos: 100.000 loss: 0.3600 (0.7645) acc1: 94.0104 (52.5588) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1595 data: 0.0683 max mem: 3531 +Test: [210/389] eta: 0:00:37 acc1_pos: 2.454 acc5_pos: 100.000 loss: 0.2773 (0.7413) acc1: 96.3542 (54.6418) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1600 data: 0.0692 max mem: 3532 +Test: [220/389] eta: 0:00:34 acc1_pos: 2.454 acc5_pos: 100.000 loss: 0.1961 (0.7188) acc1: 100.0000 (56.6742) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1684 data: 0.0773 max mem: 3532 +Test: [230/389] eta: 0:00:32 acc1_pos: 2.454 acc5_pos: 100.000 loss: 0.2780 (0.7040) acc1: 98.6979 (58.1056) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1672 data: 0.0754 max mem: 3532 +Test: [240/389] eta: 0:00:30 acc1_pos: 2.454 acc5_pos: 100.000 loss: 0.3191 (0.6882) acc1: 95.8333 (59.6646) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1629 data: 0.0714 max mem: 3532 +Test: [250/389] eta: 0:00:28 acc1_pos: 2.454 acc5_pos: 100.000 loss: 0.2880 (0.6707) acc1: 98.6979 (61.2508) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1685 data: 0.0766 max mem: 3532 +Test: [260/389] eta: 0:00:25 acc1_pos: 2.454 acc5_pos: 100.000 loss: 0.2552 (0.6591) acc1: 99.7396 (62.2935) acc5: 100.0000 (100.0000) acc1_neg: 33.073 acc5_neg: 100.000 time: 0.1635 data: 0.0709 max mem: 3532 +Test: [270/389] eta: 0:00:23 acc1_pos: 2.454 acc5_pos: 100.000 loss: 0.3096 (0.6577) acc1: 99.2188 (62.2742) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1696 data: 0.0764 max mem: 3532 +Test: [280/389] eta: 0:00:21 acc1_pos: 2.454 acc5_pos: 100.000 loss: 0.3522 (0.6490) acc1: 86.4583 (63.0913) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1766 data: 0.0830 max mem: 3532 +Test: [290/389] eta: 0:00:19 acc1_pos: 2.454 acc5_pos: 100.000 loss: 0.3335 (0.6366) acc1: 98.9583 (64.2460) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1704 data: 0.0777 max mem: 3532 +Test: [300/389] eta: 0:00:17 acc1_pos: 2.454 acc5_pos: 100.000 loss: 0.3187 (0.6259) acc1: 99.2188 (65.4070) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1737 data: 0.0815 max mem: 3532 +Test: [310/389] eta: 0:00:15 acc1_pos: 2.454 acc5_pos: 100.000 loss: 0.3209 (0.6177) acc1: 98.6979 (66.2815) acc5: 100.0000 (100.0000) acc1_neg: 70.312 acc5_neg: 100.000 time: 0.1736 data: 0.0819 max mem: 3532 +Test: [320/389] eta: 0:00:13 acc1_pos: 2.454 acc5_pos: 100.000 loss: 0.3132 (0.6088) acc1: 96.0938 (67.0869) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1806 data: 0.0888 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 2.454 acc5_pos: 100.000 loss: 0.4403 (0.6050) acc1: 89.5833 (67.6989) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.1738 data: 0.0808 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 2.454 acc5_pos: 100.000 loss: 0.2585 (0.5935) acc1: 91.1458 (68.5789) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1568 data: 0.0634 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 2.454 acc5_pos: 100.000 loss: 0.2117 (0.5927) acc1: 99.7396 (68.4814) acc5: 100.0000 (100.0000) acc1_neg: 50.781 acc5_neg: 100.000 time: 0.1634 data: 0.0699 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 2.454 acc5_pos: 100.000 loss: 0.4899 (0.5912) acc1: 80.2083 (68.6505) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.1626 data: 0.0697 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 2.454 acc5_pos: 100.000 loss: 0.4687 (0.5849) acc1: 82.8125 (69.2280) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1534 data: 0.0625 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 2.454 acc5_pos: 100.000 loss: 0.3226 (0.5789) acc1: 91.4062 (69.8299) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1407 data: 0.0506 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 2.454 acc5_pos: 100.000 loss: 0.3008 (0.5729) acc1: 98.6979 (70.3538) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1353 data: 0.0387 max mem: 3532 +Test: Total time: 0:01:13 (0.1887 s / it) +* Overall Acc@1 70.354 Acc@5 100.000 + target == real Acc@1 28.258 Acc@5 100.000 + target == fake Acc@1 89.921 Acc@5 100.000 + loss 0.573 + PR-AUC 0.8144 ROC-AUC 0.6939 EER 35.88% (thr=0.7056) +Accuracy of the network on the 149358 test images: 70.4% +[rank0]:[W726 04:00:14.673029214 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:00:33.481000 1179987 torch/distributed/run.py:766] +W0726 04:00:33.481000 1179987 torch/distributed/run.py:766] ***************************************** +W0726 04:00:33.481000 1179987 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:00:33.481000 1179987 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:01:02.853250980 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:01:02.366049008 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:01:03.496610181 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:01:03.973100933 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:01:03.016851611 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:01:03.044226231 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W726 04:01:03.044254167 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:01:03.150294860 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n23_seed6_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed6_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:30:13 acc1_pos: 10.677 acc5_pos: 100.000 loss: 1.3863 (1.3863) acc1: 10.6771 (10.6771) acc5: 100.0000 (100.0000) time: 5.4450 data: 4.1042 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/333] eta: 0:03:08 acc1_pos: 18.750 acc5_pos: 100.000 loss: 1.2270 (1.2247) acc1: 18.7500 (19.8864) acc5: 100.0000 (100.0000) time: 0.5821 data: 0.3787 max mem: 3531 +Test: [ 20/333] eta: 0:01:48 acc1_pos: 25.781 acc5_pos: 100.000 loss: 1.2270 (1.2387) acc1: 18.7500 (19.9405) acc5: 100.0000 (100.0000) time: 0.0930 data: 0.0032 max mem: 3531 +Test: [ 30/333] eta: 0:01:20 acc1_pos: 45.052 acc5_pos: 100.000 loss: 1.1545 (1.1900) acc1: 21.3542 (23.1015) acc5: 100.0000 (100.0000) time: 0.0919 data: 0.0009 max mem: 3531 +Test: [ 40/333] eta: 0:01:16 acc1_pos: 29.427 acc5_pos: 100.000 loss: 1.0507 (1.1613) acc1: 29.4271 (25.0699) acc5: 100.0000 (100.0000) time: 0.1673 data: 0.0759 max mem: 3531 +Test: [ 50/333] eta: 0:01:13 acc1_pos: 39.323 acc5_pos: 100.000 loss: 1.0201 (1.1381) acc1: 33.3333 (26.7923) acc5: 100.0000 (100.0000) time: 0.2521 data: 0.1568 max mem: 3531 +Test: [ 60/333] eta: 0:01:09 acc1_pos: 20.833 acc5_pos: 100.000 loss: 1.0137 (1.1229) acc1: 33.3333 (27.6682) acc5: 100.0000 (100.0000) time: 0.2401 data: 0.1431 max mem: 3531 +Test: [ 70/333] eta: 0:01:05 acc1_pos: 26.823 acc5_pos: 100.000 loss: 1.0618 (1.1130) acc1: 31.2500 (28.3524) acc5: 100.0000 (100.0000) time: 0.2184 data: 0.1249 max mem: 3531 +Test: [ 80/333] eta: 0:01:02 acc1_pos: 37.240 acc5_pos: 100.000 loss: 1.0237 (1.1019) acc1: 34.6354 (29.0156) acc5: 100.0000 (100.0000) time: 0.2241 data: 0.1297 max mem: 3531 +Test: [ 90/333] eta: 0:00:58 acc1_pos: 33.854 acc5_pos: 100.000 loss: 1.0237 (1.0935) acc1: 34.6354 (29.5215) acc5: 100.0000 (100.0000) time: 0.2195 data: 0.1261 max mem: 3531 +Test: [100/333] eta: 0:00:55 acc1_pos: 32.812 acc5_pos: 100.000 loss: 1.0608 (1.0927) acc1: 32.2917 (29.7571) acc5: 100.0000 (100.0000) time: 0.2068 data: 0.1151 max mem: 3531 +Test: [110/333] eta: 0:00:52 acc1_pos: 39.323 acc5_pos: 100.000 loss: 1.0430 (1.0900) acc1: 32.8125 (30.2060) acc5: 100.0000 (100.0000) time: 0.2072 data: 0.1153 max mem: 3531 +Test: [120/333] eta: 0:00:49 acc1_pos: 28.906 acc5_pos: 100.000 loss: 1.1268 (1.0903) acc1: 30.9896 (30.3504) acc5: 100.0000 (100.0000) time: 0.2165 data: 0.1249 max mem: 3531 +Test: [130/333] eta: 0:00:47 acc1_pos: 39.062 acc5_pos: 100.000 loss: 1.0925 (1.0879) acc1: 30.9896 (30.6636) acc5: 100.0000 (100.0000) time: 0.2173 data: 0.1256 max mem: 3531 +Test: [140/333] eta: 0:00:44 acc1_pos: 40.104 acc5_pos: 100.000 loss: 1.0627 (1.0874) acc1: 33.8542 (30.8917) acc5: 100.0000 (100.0000) time: 0.2164 data: 0.1231 max mem: 3531 +Test: [150/333] eta: 0:00:42 acc1_pos: 29.688 acc5_pos: 100.000 loss: 1.0642 (1.0895) acc1: 31.7708 (30.8654) acc5: 100.0000 (100.0000) time: 0.2149 data: 0.1200 max mem: 3531 +Test: [160/333] eta: 0:00:39 acc1_pos: 30.592 acc5_pos: 100.000 loss: 1.0329 (1.0757) acc1: 34.3750 (32.0587) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.2056 data: 0.1118 max mem: 3531 +Test: [170/333] eta: 0:00:37 acc1_pos: 30.592 acc5_pos: 100.000 loss: 0.5296 (1.0406) acc1: 73.4375 (34.8258) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.2099 data: 0.1139 max mem: 3531 +Test: [180/333] eta: 0:00:34 acc1_pos: 30.592 acc5_pos: 100.000 loss: 0.4591 (1.0070) acc1: 80.7292 (37.5173) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.2170 data: 0.1210 max mem: 3531 +Test: [190/333] eta: 0:00:32 acc1_pos: 30.592 acc5_pos: 100.000 loss: 0.3894 (0.9720) acc1: 85.9375 (40.3619) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.2200 data: 0.1261 max mem: 3531 +Test: [200/333] eta: 0:00:30 acc1_pos: 30.592 acc5_pos: 100.000 loss: 0.3510 (0.9422) acc1: 89.5833 (42.7304) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.2146 data: 0.1216 max mem: 3531 +Test: [210/333] eta: 0:00:27 acc1_pos: 30.592 acc5_pos: 100.000 loss: 0.4066 (0.9173) acc1: 84.6354 (44.6720) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.1964 data: 0.1057 max mem: 3532 +Test: [220/333] eta: 0:00:25 acc1_pos: 30.592 acc5_pos: 100.000 loss: 0.4186 (0.8961) acc1: 83.8542 (46.3306) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.1973 data: 0.1047 max mem: 3532 +Test: [230/333] eta: 0:00:22 acc1_pos: 30.592 acc5_pos: 100.000 loss: 0.3908 (0.8737) acc1: 87.5000 (48.1489) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1913 data: 0.0975 max mem: 3532 +Test: [240/333] eta: 0:00:20 acc1_pos: 30.592 acc5_pos: 100.000 loss: 0.3955 (0.8557) acc1: 86.7188 (49.6315) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.2049 data: 0.1129 max mem: 3532 +Test: [250/333] eta: 0:00:18 acc1_pos: 30.592 acc5_pos: 100.000 loss: 0.4358 (0.8380) acc1: 83.0729 (50.9836) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.2043 data: 0.1128 max mem: 3532 +Test: [260/333] eta: 0:00:15 acc1_pos: 30.592 acc5_pos: 100.000 loss: 0.3658 (0.8191) acc1: 87.5000 (52.4515) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1817 data: 0.0891 max mem: 3532 +Test: [270/333] eta: 0:00:13 acc1_pos: 30.592 acc5_pos: 100.000 loss: 0.3508 (0.8018) acc1: 89.3229 (53.8121) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.1991 data: 0.1060 max mem: 3532 +Test: [280/333] eta: 0:00:11 acc1_pos: 30.592 acc5_pos: 100.000 loss: 0.3612 (0.7897) acc1: 88.5417 (54.7292) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.2068 data: 0.1150 max mem: 3532 +Test: [290/333] eta: 0:00:09 acc1_pos: 30.592 acc5_pos: 100.000 loss: 0.5286 (0.7818) acc1: 71.3542 (55.2871) acc5: 100.0000 (100.0000) acc1_neg: 71.354 acc5_neg: 100.000 time: 0.1918 data: 0.1007 max mem: 3532 +Test: [300/333] eta: 0:00:07 acc1_pos: 30.592 acc5_pos: 100.000 loss: 0.5240 (0.7723) acc1: 73.6979 (56.0501) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.1717 data: 0.0809 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 30.592 acc5_pos: 100.000 loss: 0.4539 (0.7597) acc1: 80.4688 (57.0681) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.1558 data: 0.0656 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 30.592 acc5_pos: 100.000 loss: 0.3854 (0.7487) acc1: 88.8021 (57.9512) acc5: 100.0000 (100.0000) acc1_neg: 75.521 acc5_neg: 100.000 time: 0.1373 data: 0.0472 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 30.592 acc5_pos: 100.000 loss: 0.5070 (0.7420) acc1: 75.5208 (58.4356) acc5: 100.0000 (100.0000) acc1_neg: 75.521 acc5_neg: 100.000 time: 0.1179 data: 0.0283 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 30.592 acc5_pos: 100.000 loss: 0.5111 (0.7406) acc1: 74.9296 (58.5288) acc5: 100.0000 (100.0000) acc1_neg: 74.930 acc5_neg: 100.000 time: 0.1285 data: 0.0357 max mem: 3532 +Test: Total time: 0:01:09 (0.2076 s / it) +* Overall Acc@1 58.529 Acc@5 100.000 + target == real Acc@1 30.958 Acc@5 100.000 + target == fake Acc@1 83.072 Acc@5 100.000 + loss 0.741 + PR-AUC 0.6050 ROC-AUC 0.6061 EER 42.36% (thr=0.6997) +Accuracy of the network on the 127843 test images: 58.5% +[rank0]:[W726 04:02:20.140423597 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:02:40.723000 1180546 torch/distributed/run.py:766] +W0726 04:02:40.723000 1180546 torch/distributed/run.py:766] ***************************************** +W0726 04:02:40.723000 1180546 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:02:40.723000 1180546 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:03:07.040124140 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:03:07.308210112 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:03:07.391106021 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:03:08.956946116 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:03:08.027929015 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:03:08.033790938 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:03:08.053246725 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:03:08.059122389 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n23_seed6_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed6_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:23:55 acc1_pos: 89.583 acc5_pos: 100.000 loss: 0.2977 (0.2977) acc1: 89.5833 (89.5833) acc5: 100.0000 (100.0000) time: 6.1065 data: 4.6300 max mem: 3531 +Test: [ 10/235] eta: 0:02:24 acc1_pos: 91.927 acc5_pos: 100.000 loss: 0.2838 (0.2896) acc1: 91.9271 (91.5009) acc5: 100.0000 (100.0000) time: 0.6423 data: 0.4230 max mem: 3531 +Test: [ 20/235] eta: 0:01:29 acc1_pos: 91.406 acc5_pos: 100.000 loss: 0.2792 (0.2864) acc1: 91.9271 (91.6791) acc5: 100.0000 (100.0000) time: 0.1324 data: 0.0372 max mem: 3531 +Test: [ 30/235] eta: 0:01:14 acc1_pos: 94.271 acc5_pos: 100.000 loss: 0.2812 (0.2854) acc1: 92.1875 (91.9103) acc5: 100.0000 (100.0000) time: 0.2094 data: 0.1145 max mem: 3531 +Test: [ 40/235] eta: 0:01:08 acc1_pos: 92.448 acc5_pos: 100.000 loss: 0.2812 (0.2845) acc1: 92.4479 (91.9652) acc5: 100.0000 (100.0000) time: 0.2849 data: 0.1933 max mem: 3531 +Test: [ 50/235] eta: 0:01:00 acc1_pos: 90.365 acc5_pos: 100.000 loss: 0.2826 (0.2856) acc1: 92.4479 (91.8913) acc5: 100.0000 (100.0000) time: 0.2710 data: 0.1808 max mem: 3531 +Test: [ 60/235] eta: 0:00:51 acc1_pos: 92.448 acc5_pos: 100.000 loss: 0.2885 (0.2861) acc1: 91.1458 (91.8332) acc5: 100.0000 (100.0000) time: 0.1676 data: 0.0757 max mem: 3531 +Test: [ 70/235] eta: 0:00:48 acc1_pos: 93.846 acc5_pos: 100.000 loss: 0.3013 (0.4145) acc1: 89.5833 (83.4947) acc5: 100.0000 (100.0000) acc1_neg: 23.438 acc5_neg: 100.000 time: 0.2039 data: 0.1087 max mem: 3531 +Test: [ 80/235] eta: 0:00:42 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.3141 (0.5259) acc1: 24.4792 (76.1638) acc5: 100.0000 (100.0000) acc1_neg: 23.438 acc5_neg: 100.000 time: 0.2133 data: 0.1181 max mem: 3531 +Test: [ 90/235] eta: 0:00:37 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.3224 (0.6136) acc1: 24.4792 (70.4642) acc5: 100.0000 (100.0000) acc1_neg: 23.698 acc5_neg: 100.000 time: 0.1449 data: 0.0521 max mem: 3531 +Test: [100/235] eta: 0:00:33 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.3113 (0.6820) acc1: 24.2188 (66.0143) acc5: 100.0000 (100.0000) acc1_neg: 23.698 acc5_neg: 100.000 time: 0.1645 data: 0.0702 max mem: 3531 +Test: [110/235] eta: 0:00:30 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.2989 (0.7379) acc1: 25.5208 (62.3616) acc5: 100.0000 (100.0000) acc1_neg: 26.042 acc5_neg: 100.000 time: 0.1768 data: 0.0809 max mem: 3531 +Test: [120/235] eta: 0:00:27 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.3536 (0.7956) acc1: 22.9167 (58.7961) acc5: 100.0000 (100.0000) acc1_neg: 19.531 acc5_neg: 100.000 time: 0.1999 data: 0.1037 max mem: 3531 +Test: [130/235] eta: 0:00:24 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.4332 (0.8443) acc1: 19.5312 (55.8306) acc5: 100.0000 (100.0000) acc1_neg: 20.833 acc5_neg: 100.000 time: 0.1929 data: 0.0972 max mem: 3531 +Test: [140/235] eta: 0:00:21 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.4572 (0.8889) acc1: 18.2292 (53.1398) acc5: 100.0000 (100.0000) acc1_neg: 15.104 acc5_neg: 100.000 time: 0.1625 data: 0.0667 max mem: 3531 +Test: [150/235] eta: 0:00:19 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.4601 (0.9257) acc1: 18.2292 (50.8451) acc5: 100.0000 (100.0000) acc1_neg: 21.354 acc5_neg: 100.000 time: 0.1666 data: 0.0723 max mem: 3531 +Test: [160/235] eta: 0:00:16 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.4423 (0.9579) acc1: 19.0104 (48.8953) acc5: 100.0000 (100.0000) acc1_neg: 21.094 acc5_neg: 100.000 time: 0.1738 data: 0.0821 max mem: 3531 +Test: [170/235] eta: 0:00:14 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.3927 (0.9703) acc1: 21.8750 (47.8771) acc5: 100.0000 (100.0000) acc1_neg: 27.865 acc5_neg: 100.000 time: 0.1701 data: 0.0776 max mem: 3531 +Test: [180/235] eta: 0:00:12 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.1512 (0.9799) acc1: 32.8125 (47.0692) acc5: 100.0000 (100.0000) acc1_neg: 32.812 acc5_neg: 100.000 time: 0.1752 data: 0.0823 max mem: 3531 +Test: [190/235] eta: 0:00:09 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.1380 (0.9884) acc1: 32.8125 (46.3364) acc5: 100.0000 (100.0000) acc1_neg: 31.250 acc5_neg: 100.000 time: 0.1786 data: 0.0858 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.1403 (0.9967) acc1: 31.7708 (45.6234) acc5: 100.0000 (100.0000) acc1_neg: 30.208 acc5_neg: 100.000 time: 0.1797 data: 0.0883 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.1355 (1.0030) acc1: 33.0729 (45.0422) acc5: 100.0000 (100.0000) acc1_neg: 34.115 acc5_neg: 100.000 time: 0.1950 data: 0.1017 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.1337 (1.0092) acc1: 33.5938 (44.5136) acc5: 100.0000 (100.0000) acc1_neg: 33.333 acc5_neg: 100.000 time: 0.1817 data: 0.0877 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.1180 (1.0140) acc1: 33.8542 (44.0769) acc5: 100.0000 (100.0000) acc1_neg: 35.156 acc5_neg: 100.000 time: 0.1247 data: 0.0334 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.1313 (1.0177) acc1: 33.5938 (43.9235) acc5: 100.0000 (100.0000) acc1_neg: 21.739 acc5_neg: 100.000 time: 0.1155 data: 0.0275 max mem: 3532 +Test: Total time: 0:00:48 (0.2055 s / it) +* Overall Acc@1 43.923 Acc@5 100.000 + target == real Acc@1 91.878 Acc@5 100.000 + target == fake Acc@1 26.579 Acc@5 100.000 + loss 1.018 + PR-AUC 0.8506 ROC-AUC 0.6809 EER 36.96% (thr=0.2164) +Accuracy of the network on the 89879 test images: 43.9% +[rank0]:[W726 04:04:07.090449288 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:04:26.956000 1181094 torch/distributed/run.py:766] +W0726 04:04:26.956000 1181094 torch/distributed/run.py:766] ***************************************** +W0726 04:04:26.956000 1181094 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:04:26.956000 1181094 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:04:54.237603897 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:04:55.545324078 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:04:55.728454297 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:04:55.820268546 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:04:55.036090052 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:04:55.045063770 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:04:55.052506398 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:04:55.083583083 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n23_seed6_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed6_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:30:18 acc1_pos: 36.458 acc5_pos: 100.000 loss: 0.8754 (0.8754) acc1: 36.4583 (36.4583) acc5: 100.0000 (100.0000) time: 6.7599 data: 5.2739 max mem: 3531 +Test: [ 10/269] eta: 0:03:00 acc1_pos: 60.156 acc5_pos: 100.000 loss: 0.8423 (0.8142) acc1: 47.3958 (46.8277) acc5: 100.0000 (100.0000) time: 0.6968 data: 0.4799 max mem: 3531 +Test: [ 20/269] eta: 0:01:42 acc1_pos: 51.042 acc5_pos: 100.000 loss: 0.8143 (0.7895) acc1: 48.1771 (48.4127) acc5: 100.0000 (100.0000) time: 0.0945 data: 0.0005 max mem: 3531 +Test: [ 30/269] eta: 0:01:30 acc1_pos: 56.510 acc5_pos: 100.000 loss: 0.7051 (0.7701) acc1: 55.2083 (50.8149) acc5: 100.0000 (100.0000) time: 0.2042 data: 0.1050 max mem: 3531 +Test: [ 40/269] eta: 0:01:17 acc1_pos: 23.276 acc5_pos: 100.000 loss: 0.5841 (0.6983) acc1: 67.1875 (58.3460) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.2606 data: 0.1641 max mem: 3531 +Test: [ 50/269] eta: 0:01:10 acc1_pos: 23.276 acc5_pos: 100.000 loss: 0.4281 (0.6553) acc1: 84.6354 (62.7196) acc5: 100.0000 (100.0000) acc1_neg: 68.229 acc5_neg: 100.000 time: 0.2349 data: 0.1391 max mem: 3531 +Test: [ 60/269] eta: 0:01:03 acc1_pos: 23.276 acc5_pos: 100.000 loss: 0.4997 (0.6384) acc1: 75.2604 (64.4040) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.2392 data: 0.1443 max mem: 3531 +Test: [ 70/269] eta: 0:00:58 acc1_pos: 23.276 acc5_pos: 100.000 loss: 0.5730 (0.6262) acc1: 70.3125 (65.5700) acc5: 100.0000 (100.0000) acc1_neg: 80.729 acc5_neg: 100.000 time: 0.2249 data: 0.1274 max mem: 3531 +Test: [ 80/269] eta: 0:00:53 acc1_pos: 23.276 acc5_pos: 100.000 loss: 0.5214 (0.6091) acc1: 78.3854 (67.3290) acc5: 100.0000 (100.0000) acc1_neg: 78.646 acc5_neg: 100.000 time: 0.2199 data: 0.1216 max mem: 3531 +Test: [ 90/269] eta: 0:00:49 acc1_pos: 23.276 acc5_pos: 100.000 loss: 0.5214 (0.6079) acc1: 74.7396 (67.2390) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.2154 data: 0.1229 max mem: 3531 +Test: [100/269] eta: 0:00:46 acc1_pos: 23.276 acc5_pos: 100.000 loss: 0.5251 (0.6022) acc1: 66.9271 (67.6465) acc5: 100.0000 (100.0000) acc1_neg: 66.667 acc5_neg: 100.000 time: 0.2306 data: 0.1388 max mem: 3531 +Test: [110/269] eta: 0:00:43 acc1_pos: 23.276 acc5_pos: 100.000 loss: 0.5224 (0.5974) acc1: 72.9167 (68.2292) acc5: 100.0000 (100.0000) acc1_neg: 66.146 acc5_neg: 100.000 time: 0.2394 data: 0.1477 max mem: 3531 +Test: [120/269] eta: 0:00:40 acc1_pos: 23.276 acc5_pos: 100.000 loss: 0.5394 (0.5911) acc1: 73.4375 (69.0233) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.2470 data: 0.1556 max mem: 3531 +Test: [130/269] eta: 0:00:36 acc1_pos: 23.276 acc5_pos: 100.000 loss: 0.5738 (0.5983) acc1: 69.7917 (68.1576) acc5: 100.0000 (100.0000) acc1_neg: 46.354 acc5_neg: 100.000 time: 0.2214 data: 0.1309 max mem: 3531 +Test: [140/269] eta: 0:00:33 acc1_pos: 23.276 acc5_pos: 100.000 loss: 0.6805 (0.6035) acc1: 58.0729 (67.5033) acc5: 100.0000 (100.0000) acc1_neg: 64.062 acc5_neg: 100.000 time: 0.1950 data: 0.1040 max mem: 3531 +Test: [150/269] eta: 0:00:30 acc1_pos: 23.276 acc5_pos: 100.000 loss: 0.5922 (0.6012) acc1: 64.0625 (67.6652) acc5: 100.0000 (100.0000) acc1_neg: 61.458 acc5_neg: 100.000 time: 0.2060 data: 0.1131 max mem: 3531 +Test: [160/269] eta: 0:00:27 acc1_pos: 23.276 acc5_pos: 100.000 loss: 0.4995 (0.5940) acc1: 75.5208 (68.4605) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.2319 data: 0.1381 max mem: 3531 +Test: [170/269] eta: 0:00:25 acc1_pos: 23.276 acc5_pos: 100.000 loss: 0.4699 (0.5876) acc1: 83.5938 (69.2312) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.2423 data: 0.1467 max mem: 3531 +Test: [180/269] eta: 0:00:22 acc1_pos: 23.276 acc5_pos: 100.000 loss: 0.4532 (0.5792) acc1: 86.4583 (70.2924) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.2116 data: 0.1153 max mem: 3531 +Test: [190/269] eta: 0:00:19 acc1_pos: 23.276 acc5_pos: 100.000 loss: 0.4532 (0.5778) acc1: 87.5000 (70.4516) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.2064 data: 0.1137 max mem: 3531 +Test: [200/269] eta: 0:00:17 acc1_pos: 23.276 acc5_pos: 100.000 loss: 0.6002 (0.5799) acc1: 66.1458 (70.2127) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.2131 data: 0.1191 max mem: 3531 +Test: [210/269] eta: 0:00:14 acc1_pos: 23.276 acc5_pos: 100.000 loss: 0.5598 (0.5765) acc1: 69.7917 (70.5519) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.1846 data: 0.0896 max mem: 3532 +Test: [220/269] eta: 0:00:11 acc1_pos: 23.276 acc5_pos: 100.000 loss: 0.4856 (0.5737) acc1: 79.4271 (70.9064) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1650 data: 0.0739 max mem: 3532 +Test: [230/269] eta: 0:00:09 acc1_pos: 23.276 acc5_pos: 100.000 loss: 0.4856 (0.5730) acc1: 77.8646 (70.9427) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1585 data: 0.0686 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 23.276 acc5_pos: 100.000 loss: 0.5439 (0.5732) acc1: 70.8333 (70.9500) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.1360 data: 0.0461 max mem: 3532 +Test: [250/269] eta: 0:00:04 acc1_pos: 23.276 acc5_pos: 100.000 loss: 0.6010 (0.5744) acc1: 65.6250 (70.6850) acc5: 100.0000 (100.0000) acc1_neg: 41.927 acc5_neg: 100.000 time: 0.1214 data: 0.0315 max mem: 3532 +Test: [260/269] eta: 0:00:02 acc1_pos: 23.276 acc5_pos: 100.000 loss: 0.6495 (0.5771) acc1: 59.1146 (70.2806) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.1208 data: 0.0312 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 23.276 acc5_pos: 100.000 loss: 0.5689 (0.5744) acc1: 68.2292 (70.5517) acc5: 100.0000 (100.0000) acc1_neg: 88.976 acc5_neg: 100.000 time: 0.1201 data: 0.0269 max mem: 3532 +Test: Total time: 0:00:59 (0.2194 s / it) +* Overall Acc@1 70.552 Acc@5 100.000 + target == real Acc@1 49.984 Acc@5 100.000 + target == fake Acc@1 73.359 Acc@5 100.000 + loss 0.574 + PR-AUC 0.9253 ROC-AUC 0.6631 EER 38.12% (thr=0.5669) +Accuracy of the network on the 103293 test images: 70.6% +[rank0]:[W726 04:06:02.958958267 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T04:06:09Z] eval_done tag=fixed500k_real_n23_seed6_resampled +[2026-07-26T04:06:09Z] eval_start tag=fixed500k_fake_n45_seed6_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed6_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T04:06:09Z +hostname=n116-041-023 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed6_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n45_seed6_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 04:06:22.258000 1181675 torch/distributed/run.py:766] +W0726 04:06:22.258000 1181675 torch/distributed/run.py:766] ***************************************** +W0726 04:06:22.258000 1181675 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:06:22.258000 1181675 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:06:49.119451253 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:06:49.172583741 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 1): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:06:50.702321993 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W726 04:06:50.704426617 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:06:50.932141094 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:06:50.960940577 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:06:50.984277331 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:06:50.173384574 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n45_seed6_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed6_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:25:30 acc1_pos: 62.500 acc5_pos: 100.000 loss: 0.6500 (0.6500) acc1: 62.5000 (62.5000) acc5: 100.0000 (100.0000) time: 6.5387 data: 4.9325 max mem: 3531 +Test: [ 10/234] eta: 0:02:31 acc1_pos: 59.896 acc5_pos: 100.000 loss: 0.6241 (0.6252) acc1: 72.6562 (67.5900) acc5: 100.0000 (100.0000) time: 0.6785 data: 0.4487 max mem: 3531 +Test: [ 20/234] eta: 0:01:26 acc1_pos: 96.354 acc5_pos: 100.000 loss: 0.6241 (0.6528) acc1: 59.8958 (62.0908) acc5: 100.0000 (100.0000) time: 0.0972 data: 0.0017 max mem: 3531 +Test: [ 30/234] eta: 0:01:11 acc1_pos: 91.406 acc5_pos: 100.000 loss: 0.1942 (0.5061) acc1: 96.3542 (73.1603) acc5: 100.0000 (100.0000) time: 0.1672 data: 0.0726 max mem: 3531 +Test: [ 40/234] eta: 0:01:05 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1804 (0.4596) acc1: 96.8750 (77.1977) acc5: 100.0000 (100.0000) time: 0.2714 data: 0.1804 max mem: 3531 +Test: [ 50/234] eta: 0:00:59 acc1_pos: 97.396 acc5_pos: 100.000 loss: 0.1665 (0.4006) acc1: 96.3542 (81.0611) acc5: 100.0000 (100.0000) time: 0.2789 data: 0.1857 max mem: 3531 +Test: [ 60/234] eta: 0:00:51 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.1340 (0.3524) acc1: 98.1771 (84.1359) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2135 data: 0.1169 max mem: 3531 +Test: [ 70/234] eta: 0:00:47 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.1045 (0.3652) acc1: 100.0000 (83.2710) acc5: 100.0000 (100.0000) acc1_neg: 65.885 acc5_neg: 100.000 time: 0.2054 data: 0.1079 max mem: 3531 +Test: [ 80/234] eta: 0:00:43 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.5663 (0.3789) acc1: 68.7500 (82.6550) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2268 data: 0.1330 max mem: 3531 +Test: [ 90/234] eta: 0:00:38 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.3120 (0.3658) acc1: 89.0625 (83.6882) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.2050 data: 0.1127 max mem: 3531 +Test: [100/234] eta: 0:00:35 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.3133 (0.3779) acc1: 84.3750 (82.9311) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2055 data: 0.1125 max mem: 3531 +Test: [110/234] eta: 0:00:32 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.5100 (0.3858) acc1: 72.6562 (82.4934) acc5: 100.0000 (100.0000) acc1_neg: 65.885 acc5_neg: 100.000 time: 0.2145 data: 0.1225 max mem: 3531 +Test: [120/234] eta: 0:00:29 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.5100 (0.4022) acc1: 74.4792 (81.4717) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.2050 data: 0.1132 max mem: 3531 +Test: [130/234] eta: 0:00:26 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.4581 (0.4071) acc1: 79.1667 (81.4230) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.2094 data: 0.1152 max mem: 3531 +Test: [140/234] eta: 0:00:23 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.4581 (0.4284) acc1: 79.4271 (80.5149) acc5: 100.0000 (100.0000) acc1_neg: 44.531 acc5_neg: 100.000 time: 0.2051 data: 0.1104 max mem: 3531 +Test: [150/234] eta: 0:00:20 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.5206 (0.4400) acc1: 73.4375 (79.8220) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.1956 data: 0.1034 max mem: 3531 +Test: [160/234] eta: 0:00:17 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.4841 (0.4412) acc1: 73.9583 (79.8072) acc5: 100.0000 (100.0000) acc1_neg: 69.010 acc5_neg: 100.000 time: 0.1955 data: 0.1016 max mem: 3531 +Test: [170/234] eta: 0:00:15 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.5408 (0.4468) acc1: 72.6562 (79.4377) acc5: 100.0000 (100.0000) acc1_neg: 72.656 acc5_neg: 100.000 time: 0.1856 data: 0.0914 max mem: 3531 +Test: [180/234] eta: 0:00:12 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.5431 (0.4588) acc1: 71.8750 (78.6948) acc5: 100.0000 (100.0000) acc1_neg: 35.156 acc5_neg: 100.000 time: 0.1773 data: 0.0851 max mem: 3531 +Test: [190/234] eta: 0:00:10 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.7530 (0.4760) acc1: 54.6875 (77.3860) acc5: 100.0000 (100.0000) acc1_neg: 53.646 acc5_neg: 100.000 time: 0.1785 data: 0.0861 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.7530 (0.4697) acc1: 54.6875 (77.6625) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1781 data: 0.0859 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.1098 (0.4539) acc1: 99.7396 (78.6890) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1620 data: 0.0710 max mem: 3532 +Test: [220/234] eta: 0:00:03 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.1860 (0.4431) acc1: 98.6979 (79.4695) acc5: 100.0000 (100.0000) acc1_neg: 79.427 acc5_neg: 100.000 time: 0.1375 data: 0.0472 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.3970 (0.4611) acc1: 50.0000 (77.9536) acc5: 100.0000 (100.0000) acc1_neg: 46.354 acc5_neg: 100.000 time: 0.1253 data: 0.0352 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.8456 (0.4660) acc1: 46.3542 (77.5881) acc5: 100.0000 (100.0000) acc1_neg: 47.860 acc5_neg: 100.000 time: 0.1172 data: 0.0261 max mem: 3532 +Test: Total time: 0:00:50 (0.2163 s / it) +* Overall Acc@1 77.588 Acc@5 100.000 + target == real Acc@1 81.170 Acc@5 100.000 + target == fake Acc@1 76.579 Acc@5 100.000 + loss 0.466 + PR-AUC 0.9623 ROC-AUC 0.8777 EER 21.35% (thr=0.4707) +Accuracy of the network on the 89729 test images: 77.6% +[rank0]:[W726 04:07:48.947329956 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:08:08.568000 1182234 torch/distributed/run.py:766] +W0726 04:08:08.568000 1182234 torch/distributed/run.py:766] ***************************************** +W0726 04:08:08.568000 1182234 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:08:08.568000 1182234 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 04:08:37.478381102 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:08:37.936076197 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:08:37.281090528 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:08:37.391556337 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:08:38.546937458 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:08:38.656819424 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:08:38.679882795 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W726 04:08:38.679923295 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n45_seed6_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed6_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 5:13:10 acc1_pos: 43.229 acc5_pos: 100.000 loss: 0.8561 (0.8561) acc1: 43.2292 (43.2292) acc5: 100.0000 (100.0000) time: 7.8719 data: 6.2321 max mem: 3531 +Test: [ 10/2387] eta: 0:31:39 acc1_pos: 21.615 acc5_pos: 100.000 loss: 0.9761 (0.9600) acc1: 33.8542 (33.3807) acc5: 100.0000 (100.0000) time: 0.7993 data: 0.5669 max mem: 3531 +Test: [ 20/2387] eta: 0:18:23 acc1_pos: 33.073 acc5_pos: 100.000 loss: 0.9761 (0.9658) acc1: 32.8125 (33.8790) acc5: 100.0000 (100.0000) time: 0.0959 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:15:42 acc1_pos: 20.573 acc5_pos: 100.000 loss: 0.9290 (0.9625) acc1: 32.8125 (34.1146) acc5: 100.0000 (100.0000) time: 0.1799 data: 0.0831 max mem: 3531 +Test: [ 40/2387] eta: 0:14:40 acc1_pos: 30.208 acc5_pos: 100.000 loss: 0.9545 (0.9708) acc1: 28.3854 (32.3234) acc5: 100.0000 (100.0000) time: 0.2799 data: 0.1869 max mem: 3531 +Test: [ 50/2387] eta: 0:13:25 acc1_pos: 41.927 acc5_pos: 100.000 loss: 0.9592 (0.9693) acc1: 29.9479 (32.5419) acc5: 100.0000 (100.0000) time: 0.2599 data: 0.1651 max mem: 3531 +Test: [ 60/2387] eta: 0:12:42 acc1_pos: 53.385 acc5_pos: 100.000 loss: 0.9610 (0.9711) acc1: 33.3333 (32.6545) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1331 max mem: 3531 +Test: [ 70/2387] eta: 0:12:07 acc1_pos: 39.323 acc5_pos: 100.000 loss: 1.0214 (0.9795) acc1: 29.4271 (32.1413) acc5: 100.0000 (100.0000) time: 0.2349 data: 0.1409 max mem: 3531 +Test: [ 80/2387] eta: 0:11:43 acc1_pos: 50.781 acc5_pos: 100.000 loss: 0.9613 (0.9664) acc1: 32.5521 (32.9925) acc5: 100.0000 (100.0000) time: 0.2349 data: 0.1428 max mem: 3531 +Test: [ 90/2387] eta: 0:11:23 acc1_pos: 51.302 acc5_pos: 100.000 loss: 0.7992 (0.9446) acc1: 41.4062 (34.7556) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1424 max mem: 3531 +Test: [ 100/2387] eta: 0:11:09 acc1_pos: 49.740 acc5_pos: 100.000 loss: 0.7970 (0.9377) acc1: 46.0938 (35.2774) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1439 max mem: 3531 +Test: [ 110/2387] eta: 0:10:58 acc1_pos: 41.406 acc5_pos: 100.000 loss: 0.7970 (0.9224) acc1: 46.0938 (36.6554) acc5: 100.0000 (100.0000) time: 0.2499 data: 0.1509 max mem: 3531 +Test: [ 120/2387] eta: 0:10:59 acc1_pos: 44.531 acc5_pos: 100.000 loss: 0.7608 (0.9135) acc1: 46.8750 (37.4742) acc5: 100.0000 (100.0000) time: 0.2799 data: 0.1823 max mem: 3531 +Test: [ 130/2387] eta: 0:10:47 acc1_pos: 52.083 acc5_pos: 100.000 loss: 0.7701 (0.9033) acc1: 47.9167 (38.1620) acc5: 100.0000 (100.0000) time: 0.2749 data: 0.1792 max mem: 3531 +Test: [ 140/2387] eta: 0:10:38 acc1_pos: 52.604 acc5_pos: 100.000 loss: 0.7701 (0.8960) acc1: 46.3542 (38.6229) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1498 max mem: 3531 +Test: [ 150/2387] eta: 0:10:27 acc1_pos: 19.010 acc5_pos: 100.000 loss: 0.7668 (0.8882) acc1: 46.3542 (39.2626) acc5: 100.0000 (100.0000) time: 0.2396 data: 0.1451 max mem: 3531 +Test: [ 160/2387] eta: 0:10:12 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.7453 (0.8706) acc1: 50.5208 (40.8903) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2099 data: 0.1173 max mem: 3531 +Test: [ 170/2387] eta: 0:10:02 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2566 (0.8341) acc1: 95.8333 (44.1505) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2052 data: 0.1110 max mem: 3531 +Test: [ 180/2387] eta: 0:09:52 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2437 (0.8015) acc1: 96.3542 (47.0275) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2149 data: 0.1167 max mem: 3531 +Test: [ 190/2387] eta: 0:09:41 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2411 (0.7722) acc1: 96.3542 (49.6155) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2049 data: 0.1065 max mem: 3531 +Test: [ 200/2387] eta: 0:09:22 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2410 (0.7457) acc1: 96.3542 (51.9369) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1549 data: 0.0587 max mem: 3531 +Test: [ 210/2387] eta: 0:09:13 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2408 (0.7221) acc1: 95.8333 (54.0186) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1559 data: 0.0621 max mem: 3532 +Test: [ 220/2387] eta: 0:09:13 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2462 (0.7006) acc1: 96.0938 (55.9260) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2399 data: 0.1477 max mem: 3532 +Test: [ 230/2387] eta: 0:09:05 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2466 (0.6809) acc1: 96.0938 (57.6671) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2346 data: 0.1415 max mem: 3532 +Test: [ 240/2387] eta: 0:08:49 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2552 (0.6642) acc1: 95.5729 (59.2259) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1499 data: 0.0546 max mem: 3532 +Test: [ 250/2387] eta: 0:08:44 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2304 (0.6423) acc1: 96.3542 (60.8286) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1592 data: 0.0619 max mem: 3532 +Test: [ 260/2387] eta: 0:08:40 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.0919 (0.6212) acc1: 100.0000 (62.3294) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2213 data: 0.1266 max mem: 3532 +Test: [ 270/2387] eta: 0:08:35 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.0918 (0.6017) acc1: 100.0000 (63.7166) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2204 data: 0.1298 max mem: 3532 +Test: [ 280/2387] eta: 0:08:25 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.0938 (0.5837) acc1: 100.0000 (65.0078) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1786 data: 0.0856 max mem: 3532 +Test: [ 290/2387] eta: 0:08:19 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.0852 (0.5665) acc1: 100.0000 (66.2103) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1694 data: 0.0742 max mem: 3532 +Test: [ 300/2387] eta: 0:08:14 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.0828 (0.5505) acc1: 100.0000 (67.3329) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1918 data: 0.0989 max mem: 3532 +Test: [ 310/2387] eta: 0:08:08 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.0945 (0.5391) acc1: 100.0000 (68.3280) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1899 data: 0.0966 max mem: 3532 +Test: [ 320/2387] eta: 0:08:07 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.3128 (0.5323) acc1: 97.6562 (69.2319) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2240 data: 0.1293 max mem: 3532 +Test: [ 330/2387] eta: 0:08:02 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.3184 (0.5257) acc1: 97.3958 (70.0820) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2215 data: 0.1284 max mem: 3532 +Test: [ 340/2387] eta: 0:07:55 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.3137 (0.5195) acc1: 97.6562 (70.8952) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1689 data: 0.0745 max mem: 3532 +Test: [ 350/2387] eta: 0:07:50 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.3027 (0.5097) acc1: 98.6979 (71.7125) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1733 data: 0.0773 max mem: 3532 +Test: [ 360/2387] eta: 0:07:44 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1969 (0.5012) acc1: 98.9583 (72.4485) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1801 data: 0.0863 max mem: 3532 +Test: [ 370/2387] eta: 0:07:39 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2045 (0.4932) acc1: 98.1771 (73.1420) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1768 data: 0.0848 max mem: 3532 +Test: [ 380/2387] eta: 0:07:35 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2055 (0.4857) acc1: 97.9167 (73.7875) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1861 data: 0.0931 max mem: 3532 +Test: [ 390/2387] eta: 0:07:31 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2074 (0.4786) acc1: 97.6562 (74.4006) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1887 data: 0.0947 max mem: 3532 +Test: [ 400/2387] eta: 0:07:28 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2074 (0.4719) acc1: 97.6562 (74.9812) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2035 data: 0.1109 max mem: 3532 +Test: [ 410/2387] eta: 0:07:24 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2053 (0.4654) acc1: 97.9167 (75.5474) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2042 data: 0.1105 max mem: 3532 +Test: [ 420/2387] eta: 0:07:19 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2033 (0.4591) acc1: 98.4375 (76.0843) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1775 data: 0.0826 max mem: 3532 +Test: [ 430/2387] eta: 0:07:16 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2107 (0.4538) acc1: 97.6562 (76.5728) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1826 data: 0.0905 max mem: 3532 +Test: [ 440/2387] eta: 0:07:13 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2127 (0.4470) acc1: 97.3958 (77.0774) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2094 data: 0.1172 max mem: 3533 +Test: [ 450/2387] eta: 0:07:10 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.0964 (0.4393) acc1: 100.0000 (77.5857) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2100 data: 0.1175 max mem: 3533 +Test: [ 460/2387] eta: 0:07:08 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.0964 (0.4318) acc1: 100.0000 (78.0719) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2147 data: 0.1235 max mem: 3533 +Test: [ 470/2387] eta: 0:07:06 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.0970 (0.4247) acc1: 100.0000 (78.5375) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2205 data: 0.1289 max mem: 3533 +Test: [ 480/2387] eta: 0:07:03 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.0973 (0.4179) acc1: 100.0000 (78.9837) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2108 data: 0.1197 max mem: 3533 +Test: [ 490/2387] eta: 0:07:00 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.0972 (0.4114) acc1: 100.0000 (79.4117) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2098 data: 0.1187 max mem: 3533 +Test: [ 500/2387] eta: 0:06:57 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.0969 (0.4051) acc1: 100.0000 (79.8226) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2054 data: 0.1139 max mem: 3533 +Test: [ 510/2387] eta: 0:06:55 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.0969 (0.3993) acc1: 100.0000 (80.2175) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2104 data: 0.1186 max mem: 3533 +Test: [ 520/2387] eta: 0:06:53 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1325 (0.3942) acc1: 100.0000 (80.5972) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2218 data: 0.1295 max mem: 3533 +Test: [ 530/2387] eta: 0:06:50 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1414 (0.3895) acc1: 100.0000 (80.9626) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2194 data: 0.1277 max mem: 3533 +Test: [ 540/2387] eta: 0:06:48 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1442 (0.3850) acc1: 100.0000 (81.3145) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2216 data: 0.1283 max mem: 3533 +Test: [ 550/2387] eta: 0:06:46 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1444 (0.3806) acc1: 100.0000 (81.6536) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2154 data: 0.1220 max mem: 3533 +Test: [ 560/2387] eta: 0:06:44 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1453 (0.3764) acc1: 100.0000 (81.9807) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2202 data: 0.1228 max mem: 3533 +Test: [ 570/2387] eta: 0:06:42 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1437 (0.3724) acc1: 100.0000 (82.2962) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2392 data: 0.1423 max mem: 3533 +Test: [ 580/2387] eta: 0:06:40 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1434 (0.3685) acc1: 100.0000 (82.6009) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2264 data: 0.1346 max mem: 3533 +Test: [ 590/2387] eta: 0:06:38 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1448 (0.3647) acc1: 100.0000 (82.8953) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2301 data: 0.1379 max mem: 3533 +Test: [ 600/2387] eta: 0:06:36 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1374 (0.3609) acc1: 100.0000 (83.1791) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2356 data: 0.1410 max mem: 3533 +Test: [ 610/2387] eta: 0:06:31 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1571 (0.3583) acc1: 99.4792 (83.4348) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1756 data: 0.0809 max mem: 3533 +Test: [ 620/2387] eta: 0:06:28 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2032 (0.3558) acc1: 98.9583 (83.6864) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1575 data: 0.0666 max mem: 3533 +Test: [ 630/2387] eta: 0:06:25 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2032 (0.3534) acc1: 98.9583 (83.9256) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1905 data: 0.0979 max mem: 3533 +Test: [ 640/2387] eta: 0:06:23 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2046 (0.3511) acc1: 98.6979 (84.1601) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1997 data: 0.1057 max mem: 3533 +Test: [ 650/2387] eta: 0:06:19 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2046 (0.3489) acc1: 98.9583 (84.3882) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1875 data: 0.0940 max mem: 3533 +Test: [ 660/2387] eta: 0:06:16 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2044 (0.3467) acc1: 98.9583 (84.6078) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1827 data: 0.0900 max mem: 3533 +Test: [ 670/2387] eta: 0:06:13 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2041 (0.3446) acc1: 98.9583 (84.8221) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1921 data: 0.0995 max mem: 3534 +Test: [ 680/2387] eta: 0:06:11 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2076 (0.3427) acc1: 99.4792 (85.0415) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1920 data: 0.0964 max mem: 3534 +Test: [ 690/2387] eta: 0:06:07 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2239 (0.3438) acc1: 99.2188 (85.0040) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.1827 data: 0.0883 max mem: 3534 +Test: [ 700/2387] eta: 0:06:05 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.4861 (0.3458) acc1: 76.3021 (84.8706) acc5: 100.0000 (100.0000) acc1_neg: 79.427 acc5_neg: 100.000 time: 0.1820 data: 0.0904 max mem: 3534 +Test: [ 710/2387] eta: 0:06:01 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.4933 (0.3479) acc1: 75.5208 (84.7314) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.1720 data: 0.0806 max mem: 3534 +Test: [ 720/2387] eta: 0:05:58 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.4853 (0.3497) acc1: 75.5208 (84.6087) acc5: 100.0000 (100.0000) acc1_neg: 75.781 acc5_neg: 100.000 time: 0.1648 data: 0.0729 max mem: 3534 +Test: [ 730/2387] eta: 0:05:55 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.4848 (0.3517) acc1: 75.7812 (84.4762) acc5: 100.0000 (100.0000) acc1_neg: 73.958 acc5_neg: 100.000 time: 0.1724 data: 0.0780 max mem: 3534 +Test: [ 740/2387] eta: 0:05:52 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.4848 (0.3535) acc1: 75.0000 (84.3571) acc5: 100.0000 (100.0000) acc1_neg: 73.698 acc5_neg: 100.000 time: 0.1692 data: 0.0740 max mem: 3534 +Test: [ 750/2387] eta: 0:05:49 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.4756 (0.3552) acc1: 76.0417 (84.2474) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.1798 data: 0.0865 max mem: 3534 +Test: [ 760/2387] eta: 0:05:46 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.4711 (0.3561) acc1: 76.8229 (84.1861) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1886 data: 0.0969 max mem: 3534 +Test: [ 770/2387] eta: 0:05:44 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.3337 (0.3548) acc1: 86.1979 (84.2882) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1899 data: 0.0983 max mem: 3534 +Test: [ 780/2387] eta: 0:05:41 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1552 (0.3520) acc1: 98.6979 (84.4830) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1971 data: 0.1055 max mem: 3534 +Test: [ 790/2387] eta: 0:05:39 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1366 (0.3495) acc1: 99.7396 (84.6427) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1949 data: 0.1014 max mem: 3534 +Test: [ 800/2387] eta: 0:05:36 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1324 (0.3469) acc1: 99.4792 (84.8276) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1897 data: 0.0953 max mem: 3534 +Test: [ 810/2387] eta: 0:05:33 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1432 (0.3445) acc1: 99.4792 (85.0031) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1750 data: 0.0820 max mem: 3534 +Test: [ 820/2387] eta: 0:05:30 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1546 (0.3423) acc1: 99.2188 (85.1613) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1690 data: 0.0759 max mem: 3534 +Test: [ 830/2387] eta: 0:05:28 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2004 (0.3415) acc1: 95.5729 (85.2528) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2014 data: 0.1083 max mem: 3534 +Test: [ 840/2387] eta: 0:05:25 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2617 (0.3409) acc1: 94.7917 (85.3557) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1888 data: 0.0965 max mem: 3534 +Test: [ 850/2387] eta: 0:05:23 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2912 (0.3405) acc1: 94.7917 (85.4473) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1733 data: 0.0800 max mem: 3534 +Test: [ 860/2387] eta: 0:05:20 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2836 (0.3398) acc1: 95.0521 (85.5706) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1928 data: 0.0989 max mem: 3534 +Test: [ 870/2387] eta: 0:05:17 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2687 (0.3391) acc1: 96.6146 (85.6804) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1775 data: 0.0846 max mem: 3534 +Test: [ 880/2387] eta: 0:05:15 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2793 (0.3385) acc1: 95.5729 (85.7785) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1736 data: 0.0820 max mem: 3534 +Test: [ 890/2387] eta: 0:05:12 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2569 (0.3374) acc1: 96.6146 (85.9127) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1874 data: 0.0967 max mem: 3534 +Test: [ 900/2387] eta: 0:05:10 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2767 (0.3415) acc1: 95.0521 (85.6138) acc5: 100.0000 (100.0000) acc1_neg: 50.521 acc5_neg: 100.000 time: 0.1737 data: 0.0817 max mem: 3535 +Test: [ 910/2387] eta: 0:05:07 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.8070 (0.3466) acc1: 49.7396 (85.2100) acc5: 100.0000 (100.0000) acc1_neg: 50.260 acc5_neg: 100.000 time: 0.1774 data: 0.0833 max mem: 3535 +Test: [ 920/2387] eta: 0:05:05 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.8179 (0.3517) acc1: 48.9583 (84.8237) acc5: 100.0000 (100.0000) acc1_neg: 46.615 acc5_neg: 100.000 time: 0.1974 data: 0.1027 max mem: 3535 +Test: [ 930/2387] eta: 0:05:02 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.8194 (0.3566) acc1: 49.7396 (84.4561) acc5: 100.0000 (100.0000) acc1_neg: 48.438 acc5_neg: 100.000 time: 0.1875 data: 0.0946 max mem: 3535 +Test: [ 940/2387] eta: 0:05:00 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.8259 (0.3615) acc1: 49.7396 (84.0836) acc5: 100.0000 (100.0000) acc1_neg: 47.917 acc5_neg: 100.000 time: 0.1785 data: 0.0856 max mem: 3535 +Test: [ 950/2387] eta: 0:04:58 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.7965 (0.3662) acc1: 48.9583 (83.7137) acc5: 100.0000 (100.0000) acc1_neg: 51.042 acc5_neg: 100.000 time: 0.1846 data: 0.0920 max mem: 3535 +Test: [ 960/2387] eta: 0:04:55 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.7912 (0.3707) acc1: 49.2188 (83.3640) acc5: 100.0000 (100.0000) acc1_neg: 52.865 acc5_neg: 100.000 time: 0.1954 data: 0.1027 max mem: 3535 +Test: [ 970/2387] eta: 0:04:53 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.8029 (0.3752) acc1: 48.9583 (83.0000) acc5: 100.0000 (100.0000) acc1_neg: 49.479 acc5_neg: 100.000 time: 0.2061 data: 0.1106 max mem: 3535 +Test: [ 980/2387] eta: 0:04:51 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.7992 (0.3758) acc1: 49.4792 (82.9551) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.2085 data: 0.1133 max mem: 3535 +Test: [ 990/2387] eta: 0:04:49 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1706 (0.3738) acc1: 98.9583 (83.0863) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2069 data: 0.1146 max mem: 3535 +Test: [1000/2387] eta: 0:04:47 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1259 (0.3714) acc1: 99.2188 (83.2376) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2093 data: 0.1143 max mem: 3535 +Test: [1010/2387] eta: 0:04:45 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1385 (0.3696) acc1: 98.4375 (83.3653) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2137 data: 0.1185 max mem: 3535 +Test: [1020/2387] eta: 0:04:43 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1686 (0.3680) acc1: 97.1354 (83.4838) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2049 data: 0.1140 max mem: 3535 +Test: [1030/2387] eta: 0:04:41 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1457 (0.3658) acc1: 99.2188 (83.6309) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1987 data: 0.1063 max mem: 3535 +Test: [1040/2387] eta: 0:04:39 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1206 (0.3635) acc1: 99.7396 (83.7856) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2072 data: 0.1147 max mem: 3535 +Test: [1050/2387] eta: 0:04:37 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1213 (0.3613) acc1: 99.7396 (83.9354) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2005 data: 0.1073 max mem: 3535 +Test: [1060/2387] eta: 0:04:35 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1230 (0.3591) acc1: 99.7396 (84.0812) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2083 data: 0.1146 max mem: 3535 +Test: [1070/2387] eta: 0:04:32 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1452 (0.3574) acc1: 99.4792 (84.2058) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.2110 data: 0.1201 max mem: 3535 +Test: [1080/2387] eta: 0:04:30 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1925 (0.3563) acc1: 96.6146 (84.3049) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1956 data: 0.1052 max mem: 3535 +Test: [1090/2387] eta: 0:04:28 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1867 (0.3546) acc1: 96.6146 (84.4289) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1880 data: 0.0962 max mem: 3535 +Test: [1100/2387] eta: 0:04:25 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1658 (0.3530) acc1: 98.4375 (84.5496) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1766 data: 0.0849 max mem: 3535 +Test: [1110/2387] eta: 0:04:23 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1730 (0.3516) acc1: 98.4375 (84.6589) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1893 data: 0.0978 max mem: 3535 +Test: [1120/2387] eta: 0:04:21 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1852 (0.3501) acc1: 98.1771 (84.7762) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1727 data: 0.0792 max mem: 3536 +Test: [1130/2387] eta: 0:04:18 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1852 (0.3486) acc1: 98.1771 (84.8912) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1652 data: 0.0721 max mem: 3536 +Test: [1140/2387] eta: 0:04:16 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1964 (0.3476) acc1: 96.8750 (84.9892) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1890 data: 0.0972 max mem: 3536 +Test: [1150/2387] eta: 0:04:14 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2333 (0.3468) acc1: 96.0938 (85.0850) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1948 data: 0.1030 max mem: 3536 +Test: [1160/2387] eta: 0:04:11 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2260 (0.3457) acc1: 97.6562 (85.1764) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1778 data: 0.0836 max mem: 3536 +Test: [1170/2387] eta: 0:04:09 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2096 (0.3447) acc1: 97.6562 (85.2848) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1710 data: 0.0769 max mem: 3536 +Test: [1180/2387] eta: 0:04:07 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2272 (0.3437) acc1: 97.6562 (85.3754) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1898 data: 0.0989 max mem: 3536 +Test: [1190/2387] eta: 0:04:05 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2105 (0.3427) acc1: 97.6562 (85.4764) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1909 data: 0.0995 max mem: 3536 +Test: [1200/2387] eta: 0:04:02 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2016 (0.3414) acc1: 98.4375 (85.5888) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1828 data: 0.0910 max mem: 3536 +Test: [1210/2387] eta: 0:04:00 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1768 (0.3401) acc1: 99.4792 (85.7025) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1817 data: 0.0910 max mem: 3536 +Test: [1220/2387] eta: 0:03:58 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1157 (0.3382) acc1: 99.7396 (85.8179) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1745 data: 0.0837 max mem: 3536 +Test: [1230/2387] eta: 0:03:56 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1011 (0.3363) acc1: 99.7396 (85.9320) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1758 data: 0.0838 max mem: 3536 +Test: [1240/2387] eta: 0:03:53 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1022 (0.3344) acc1: 100.0000 (86.0439) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0932 max mem: 3536 +Test: [1250/2387] eta: 0:03:51 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1013 (0.3325) acc1: 100.0000 (86.1542) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1730 data: 0.0826 max mem: 3536 +Test: [1260/2387] eta: 0:03:48 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.0912 (0.3306) acc1: 100.0000 (86.2634) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1624 data: 0.0715 max mem: 3536 +Test: [1270/2387] eta: 0:03:46 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.0894 (0.3287) acc1: 100.0000 (86.3713) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1649 data: 0.0717 max mem: 3536 +Test: [1280/2387] eta: 0:03:44 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.0934 (0.3269) acc1: 100.0000 (86.4750) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1681 data: 0.0746 max mem: 3536 +Test: [1290/2387] eta: 0:03:41 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1046 (0.3252) acc1: 99.2188 (86.5683) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1680 data: 0.0763 max mem: 3536 +Test: [1300/2387] eta: 0:03:39 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1056 (0.3236) acc1: 99.2188 (86.6645) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1705 data: 0.0787 max mem: 3536 +Test: [1310/2387] eta: 0:03:37 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1069 (0.3220) acc1: 99.4792 (86.7577) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1651 data: 0.0730 max mem: 3536 +Test: [1320/2387] eta: 0:03:35 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1109 (0.3205) acc1: 99.4792 (86.8498) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1657 data: 0.0745 max mem: 3536 +Test: [1330/2387] eta: 0:03:32 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1187 (0.3192) acc1: 99.2188 (86.9371) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1753 data: 0.0831 max mem: 3536 +Test: [1340/2387] eta: 0:03:30 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1749 (0.3189) acc1: 98.1771 (86.9580) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1785 data: 0.0860 max mem: 3536 +Test: [1350/2387] eta: 0:03:28 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.0733 (0.3171) acc1: 100.0000 (87.0545) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1782 data: 0.0862 max mem: 3537 +Test: [1360/2387] eta: 0:03:26 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.0656 (0.3152) acc1: 100.0000 (87.1497) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1792 data: 0.0863 max mem: 3537 +Test: [1370/2387] eta: 0:03:24 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.0636 (0.3134) acc1: 100.0000 (87.2434) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1777 data: 0.0839 max mem: 3537 +Test: [1380/2387] eta: 0:03:21 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.0606 (0.3116) acc1: 100.0000 (87.3358) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1754 data: 0.0816 max mem: 3537 +Test: [1390/2387] eta: 0:03:19 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.0581 (0.3098) acc1: 100.0000 (87.4268) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1721 data: 0.0797 max mem: 3537 +Test: [1400/2387] eta: 0:03:17 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.0581 (0.3080) acc1: 100.0000 (87.5165) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1770 data: 0.0835 max mem: 3537 +Test: [1410/2387] eta: 0:03:15 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.0644 (0.3070) acc1: 100.0000 (87.5836) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1802 data: 0.0857 max mem: 3537 +Test: [1420/2387] eta: 0:03:13 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1819 (0.3063) acc1: 97.9167 (87.6439) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1729 data: 0.0809 max mem: 3537 +Test: [1430/2387] eta: 0:03:11 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2090 (0.3057) acc1: 95.5729 (87.6971) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1862 data: 0.0948 max mem: 3537 +Test: [1440/2387] eta: 0:03:09 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2107 (0.3050) acc1: 95.3125 (87.7561) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.1873 data: 0.0953 max mem: 3537 +Test: [1450/2387] eta: 0:03:07 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2062 (0.3046) acc1: 95.3125 (87.8071) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1836 data: 0.0907 max mem: 3537 +Test: [1460/2387] eta: 0:03:04 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1797 (0.3038) acc1: 96.8750 (87.8715) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1766 data: 0.0841 max mem: 3537 +Test: [1470/2387] eta: 0:03:02 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1558 (0.3027) acc1: 98.6979 (87.9488) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1662 data: 0.0741 max mem: 3537 +Test: [1480/2387] eta: 0:03:00 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1494 (0.3017) acc1: 99.2188 (88.0177) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1888 data: 0.0957 max mem: 3537 +Test: [1490/2387] eta: 0:02:58 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1529 (0.3008) acc1: 98.9583 (88.0910) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1811 data: 0.0879 max mem: 3537 +Test: [1500/2387] eta: 0:02:56 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1559 (0.2998) acc1: 98.9583 (88.1638) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1813 data: 0.0888 max mem: 3537 +Test: [1510/2387] eta: 0:02:54 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1674 (0.2991) acc1: 99.4792 (88.2358) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1906 data: 0.0980 max mem: 3537 +Test: [1520/2387] eta: 0:02:52 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2145 (0.2988) acc1: 97.6562 (88.2802) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1848 data: 0.0925 max mem: 3537 +Test: [1530/2387] eta: 0:02:50 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2629 (0.2987) acc1: 94.0104 (88.3211) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1865 data: 0.0947 max mem: 3537 +Test: [1540/2387] eta: 0:02:48 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2555 (0.2984) acc1: 95.8333 (88.3759) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1897 data: 0.0973 max mem: 3537 +Test: [1550/2387] eta: 0:02:46 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2400 (0.2981) acc1: 97.3958 (88.4290) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2035 data: 0.1112 max mem: 3537 +Test: [1560/2387] eta: 0:02:44 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2400 (0.2978) acc1: 97.1354 (88.4793) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1823 data: 0.0900 max mem: 3537 +Test: [1570/2387] eta: 0:02:42 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2258 (0.2973) acc1: 97.1354 (88.5392) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1987 data: 0.1070 max mem: 3537 +Test: [1580/2387] eta: 0:02:40 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1835 (0.2965) acc1: 98.9583 (88.6051) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2174 data: 0.1253 max mem: 3538 +Test: [1590/2387] eta: 0:02:38 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1463 (0.2955) acc1: 99.2188 (88.6724) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1777 data: 0.0829 max mem: 3538 +Test: [1600/2387] eta: 0:02:36 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1283 (0.2946) acc1: 99.7396 (88.7269) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1676 data: 0.0730 max mem: 3538 +Test: [1610/2387] eta: 0:02:34 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1283 (0.2936) acc1: 99.7396 (88.7929) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1976 data: 0.1042 max mem: 3538 +Test: [1620/2387] eta: 0:02:32 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1522 (0.2928) acc1: 99.4792 (88.8495) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2194 data: 0.1264 max mem: 3538 +Test: [1630/2387] eta: 0:02:30 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1632 (0.2922) acc1: 98.1771 (88.8984) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1970 data: 0.1055 max mem: 3538 +Test: [1640/2387] eta: 0:02:28 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2187 (0.2920) acc1: 96.0938 (88.9359) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1876 data: 0.0950 max mem: 3538 +Test: [1650/2387] eta: 0:02:26 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2288 (0.2916) acc1: 95.8333 (88.9781) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1805 data: 0.0866 max mem: 3538 +Test: [1660/2387] eta: 0:02:24 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2243 (0.2913) acc1: 95.5729 (89.0106) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1849 data: 0.0922 max mem: 3538 +Test: [1670/2387] eta: 0:02:22 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2363 (0.2909) acc1: 95.5729 (89.0513) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2178 data: 0.1234 max mem: 3538 +Test: [1680/2387] eta: 0:02:20 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2363 (0.2906) acc1: 95.3125 (89.0837) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.2123 data: 0.1184 max mem: 3538 +Test: [1690/2387] eta: 0:02:18 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2303 (0.2904) acc1: 95.3125 (89.1144) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1883 data: 0.0967 max mem: 3538 +Test: [1700/2387] eta: 0:02:16 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2285 (0.2898) acc1: 96.0938 (89.1582) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1748 data: 0.0825 max mem: 3538 +Test: [1710/2387] eta: 0:02:14 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1257 (0.2888) acc1: 99.2188 (89.2185) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1737 data: 0.0822 max mem: 3538 +Test: [1720/2387] eta: 0:02:12 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1150 (0.2878) acc1: 99.7396 (89.2801) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1940 data: 0.1027 max mem: 3538 +Test: [1730/2387] eta: 0:02:10 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1150 (0.2869) acc1: 99.7396 (89.3387) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2067 data: 0.1154 max mem: 3538 +Test: [1740/2387] eta: 0:02:08 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1138 (0.2859) acc1: 99.7396 (89.3985) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2085 data: 0.1172 max mem: 3538 +Test: [1750/2387] eta: 0:02:06 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1053 (0.2849) acc1: 100.0000 (89.4569) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2024 data: 0.1109 max mem: 3538 +Test: [1760/2387] eta: 0:02:04 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1145 (0.2839) acc1: 100.0000 (89.5152) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1956 data: 0.1028 max mem: 3538 +Test: [1770/2387] eta: 0:02:02 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1333 (0.2832) acc1: 99.4792 (89.5654) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2031 data: 0.1095 max mem: 3538 +Test: [1780/2387] eta: 0:02:00 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1460 (0.2824) acc1: 98.9583 (89.6186) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2014 data: 0.1094 max mem: 3538 +Test: [1790/2387] eta: 0:01:58 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1453 (0.2816) acc1: 98.9583 (89.6723) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2027 data: 0.1119 max mem: 3538 +Test: [1800/2387] eta: 0:01:56 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1433 (0.2809) acc1: 99.2188 (89.7239) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1875 data: 0.0955 max mem: 3538 +Test: [1810/2387] eta: 0:01:54 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1280 (0.2800) acc1: 99.2188 (89.7783) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1725 data: 0.0795 max mem: 3539 +Test: [1820/2387] eta: 0:01:52 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1194 (0.2791) acc1: 99.7396 (89.8320) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1876 data: 0.0952 max mem: 3539 +Test: [1830/2387] eta: 0:01:50 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1296 (0.2783) acc1: 100.0000 (89.8870) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1889 data: 0.0964 max mem: 3539 +Test: [1840/2387] eta: 0:01:48 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1363 (0.2775) acc1: 100.0000 (89.9419) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1822 data: 0.0904 max mem: 3539 +Test: [1850/2387] eta: 0:01:46 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1363 (0.2768) acc1: 100.0000 (89.9960) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1885 data: 0.0970 max mem: 3539 +Test: [1860/2387] eta: 0:01:44 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1359 (0.2760) acc1: 100.0000 (90.0496) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1803 data: 0.0888 max mem: 3539 +Test: [1870/2387] eta: 0:01:42 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1355 (0.2753) acc1: 100.0000 (90.1026) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1764 data: 0.0836 max mem: 3539 +Test: [1880/2387] eta: 0:01:40 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1342 (0.2745) acc1: 100.0000 (90.1551) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1761 data: 0.0837 max mem: 3539 +Test: [1890/2387] eta: 0:01:38 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1345 (0.2738) acc1: 100.0000 (90.2072) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1701 data: 0.0798 max mem: 3539 +Test: [1900/2387] eta: 0:01:36 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1365 (0.2731) acc1: 100.0000 (90.2586) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1803 data: 0.0901 max mem: 3539 +Test: [1910/2387] eta: 0:01:33 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1640 (0.2726) acc1: 100.0000 (90.3080) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1780 data: 0.0871 max mem: 3539 +Test: [1920/2387] eta: 0:01:32 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1716 (0.2723) acc1: 99.7396 (90.3308) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1880 data: 0.0957 max mem: 3539 +Test: [1930/2387] eta: 0:01:29 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1512 (0.2716) acc1: 99.7396 (90.3800) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1818 data: 0.0894 max mem: 3539 +Test: [1940/2387] eta: 0:01:27 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1546 (0.2712) acc1: 99.7396 (90.4221) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1699 data: 0.0784 max mem: 3539 +Test: [1950/2387] eta: 0:01:25 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1855 (0.2708) acc1: 98.6979 (90.4576) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1877 data: 0.0953 max mem: 3539 +Test: [1960/2387] eta: 0:01:23 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1772 (0.2704) acc1: 98.6979 (90.4890) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1863 data: 0.0943 max mem: 3539 +Test: [1970/2387] eta: 0:01:21 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2185 (0.2706) acc1: 92.1875 (90.4770) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.1696 data: 0.0786 max mem: 3539 +Test: [1980/2387] eta: 0:01:19 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2571 (0.2704) acc1: 91.9271 (90.4997) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1740 data: 0.0803 max mem: 3539 +Test: [1990/2387] eta: 0:01:18 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1703 (0.2699) acc1: 98.6979 (90.5421) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1947 data: 0.1006 max mem: 3539 +Test: [2000/2387] eta: 0:01:15 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1702 (0.2694) acc1: 98.9583 (90.5836) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1743 data: 0.0817 max mem: 3539 +Test: [2010/2387] eta: 0:01:14 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1727 (0.2689) acc1: 98.6979 (90.6240) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1794 data: 0.0853 max mem: 3539 +Test: [2020/2387] eta: 0:01:12 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1727 (0.2685) acc1: 98.6979 (90.6638) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2071 data: 0.1114 max mem: 3539 +Test: [2030/2387] eta: 0:01:10 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1711 (0.2680) acc1: 98.6979 (90.7037) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1899 data: 0.0948 max mem: 3540 +Test: [2040/2387] eta: 0:01:08 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1703 (0.2675) acc1: 98.9583 (90.7447) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1967 data: 0.1038 max mem: 3540 +Test: [2050/2387] eta: 0:01:06 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1687 (0.2670) acc1: 98.9583 (90.7851) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1899 data: 0.0978 max mem: 3540 +Test: [2060/2387] eta: 0:01:04 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1725 (0.2666) acc1: 98.9583 (90.8222) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1772 data: 0.0849 max mem: 3540 +Test: [2070/2387] eta: 0:01:02 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1763 (0.2661) acc1: 98.6979 (90.8613) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2184 data: 0.1267 max mem: 3540 +Test: [2080/2387] eta: 0:01:00 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1704 (0.2657) acc1: 98.6979 (90.8991) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2289 data: 0.1376 max mem: 3540 +Test: [2090/2387] eta: 0:00:58 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1709 (0.2652) acc1: 98.6979 (90.9369) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1868 data: 0.0948 max mem: 3540 +Test: [2100/2387] eta: 0:00:56 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1723 (0.2648) acc1: 98.6979 (90.9742) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1917 data: 0.0997 max mem: 3540 +Test: [2110/2387] eta: 0:00:54 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1712 (0.2643) acc1: 98.6979 (91.0120) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2077 data: 0.1164 max mem: 3540 +Test: [2120/2387] eta: 0:00:52 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1710 (0.2639) acc1: 98.6979 (91.0439) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1946 data: 0.1041 max mem: 3540 +Test: [2130/2387] eta: 0:00:50 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1471 (0.2634) acc1: 99.2188 (91.0831) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1935 data: 0.1024 max mem: 3540 +Test: [2140/2387] eta: 0:00:48 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1499 (0.2629) acc1: 99.4792 (91.1091) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1736 data: 0.0823 max mem: 3540 +Test: [2150/2387] eta: 0:00:46 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1691 (0.2625) acc1: 98.9583 (91.1457) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1881 data: 0.0958 max mem: 3540 +Test: [2160/2387] eta: 0:00:44 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1691 (0.2620) acc1: 98.9583 (91.1790) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2043 data: 0.1097 max mem: 3540 +Test: [2170/2387] eta: 0:00:42 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1815 (0.2619) acc1: 97.6562 (91.1999) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1703 data: 0.0766 max mem: 3540 +Test: [2180/2387] eta: 0:00:40 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2528 (0.2628) acc1: 90.8854 (91.1395) acc5: 100.0000 (100.0000) acc1_neg: 53.385 acc5_neg: 100.000 time: 0.1766 data: 0.0842 max mem: 3540 +Test: [2190/2387] eta: 0:00:38 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.3001 (0.2631) acc1: 90.6250 (91.1246) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2064 data: 0.1143 max mem: 3540 +Test: [2200/2387] eta: 0:00:36 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1721 (0.2626) acc1: 98.4375 (91.1609) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1939 data: 0.1020 max mem: 3540 +Test: [2210/2387] eta: 0:00:34 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1532 (0.2621) acc1: 99.2188 (91.1962) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1895 data: 0.0981 max mem: 3540 +Test: [2220/2387] eta: 0:00:32 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1532 (0.2617) acc1: 99.2188 (91.2299) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1945 data: 0.1021 max mem: 3540 +Test: [2230/2387] eta: 0:00:30 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1517 (0.2612) acc1: 99.4792 (91.2647) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1961 data: 0.1030 max mem: 3540 +Test: [2240/2387] eta: 0:00:28 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1441 (0.2607) acc1: 99.2188 (91.2955) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1744 data: 0.0818 max mem: 3540 +Test: [2250/2387] eta: 0:00:26 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1674 (0.2604) acc1: 99.2188 (91.3258) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1588 data: 0.0666 max mem: 3540 +Test: [2260/2387] eta: 0:00:24 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1740 (0.2600) acc1: 98.9583 (91.3593) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1674 data: 0.0762 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1384 (0.2594) acc1: 99.4792 (91.3958) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1558 data: 0.0641 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1281 (0.2590) acc1: 100.0000 (91.4204) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1612 data: 0.0698 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1311 (0.2584) acc1: 100.0000 (91.4572) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1545 data: 0.0641 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1350 (0.2579) acc1: 99.7396 (91.4885) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1363 data: 0.0460 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1455 (0.2576) acc1: 98.6979 (91.5144) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1373 data: 0.0470 max mem: 3541 +Test: [2320/2387] eta: 0:00:13 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.1898 (0.2575) acc1: 96.6146 (91.5281) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1376 data: 0.0475 max mem: 3541 +Test: [2330/2387] eta: 0:00:11 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2337 (0.2575) acc1: 96.3542 (91.5442) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.1341 data: 0.0438 max mem: 3541 +Test: [2340/2387] eta: 0:00:09 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2187 (0.2573) acc1: 96.3542 (91.5661) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1327 data: 0.0415 max mem: 3541 +Test: [2350/2387] eta: 0:00:07 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2174 (0.2573) acc1: 95.0521 (91.5660) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1330 data: 0.0421 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2606 (0.2574) acc1: 93.4896 (91.5721) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1290 data: 0.0387 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2419 (0.2574) acc1: 94.0104 (91.5786) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.1276 data: 0.0368 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2732 (0.2576) acc1: 90.8854 (91.5728) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1249 data: 0.0336 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 61.187 acc5_pos: 100.000 loss: 0.2740 (0.2576) acc1: 90.8854 (91.5773) acc5: 100.0000 (100.0000) acc1_neg: 95.349 acc5_neg: 100.000 time: 0.1157 data: 0.0272 max mem: 3541 +Test: Total time: 0:07:39 (0.1927 s / it) +* Overall Acc@1 91.577 Acc@5 100.000 + target == real Acc@1 39.629 Acc@5 100.000 + target == fake Acc@1 95.225 Acc@5 100.000 + loss 0.258 + PR-AUC 0.9923 ROC-AUC 0.9001 EER 17.43% (thr=0.7354) +Accuracy of the network on the 916267 test images: 91.6% +[rank0]:[W726 04:16:39.806825605 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:16:59.873000 1182855 torch/distributed/run.py:766] +W0726 04:16:59.873000 1182855 torch/distributed/run.py:766] ***************************************** +W0726 04:16:59.873000 1182855 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:16:59.873000 1182855 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:17:29.068942662 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:17:30.536220528 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:17:30.656175850 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:17:30.925139659 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:17:30.031838900 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +| distributed init (rank 3): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:17:30.060914743 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:17:30.065432647 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W726 04:17:30.065448863 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n45_seed6_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed6_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:01:15 acc1_pos: 43.229 acc5_pos: 100.000 loss: 0.8561 (0.8561) acc1: 43.2292 (43.2292) acc5: 100.0000 (100.0000) time: 7.1087 data: 5.3108 max mem: 3531 +Test: [ 10/517] eta: 0:06:09 acc1_pos: 21.615 acc5_pos: 100.000 loss: 0.9761 (0.9600) acc1: 33.8542 (33.3807) acc5: 100.0000 (100.0000) time: 0.7288 data: 0.4831 max mem: 3531 +Test: [ 20/517] eta: 0:03:32 acc1_pos: 33.073 acc5_pos: 100.000 loss: 0.9761 (0.9658) acc1: 32.8125 (33.8790) acc5: 100.0000 (100.0000) time: 0.0945 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:02:58 acc1_pos: 20.573 acc5_pos: 100.000 loss: 0.9290 (0.9625) acc1: 32.8125 (34.1146) acc5: 100.0000 (100.0000) time: 0.1676 data: 0.0705 max mem: 3531 +Test: [ 40/517] eta: 0:02:37 acc1_pos: 30.208 acc5_pos: 100.000 loss: 0.9545 (0.9708) acc1: 28.3854 (32.3234) acc5: 100.0000 (100.0000) time: 0.2290 data: 0.1326 max mem: 3531 +Test: [ 50/517] eta: 0:02:27 acc1_pos: 50.917 acc5_pos: 100.000 loss: 0.9158 (0.9433) acc1: 29.9479 (35.1920) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2355 data: 0.1396 max mem: 3531 +Test: [ 60/517] eta: 0:02:20 acc1_pos: 50.917 acc5_pos: 100.000 loss: 0.3934 (0.8483) acc1: 83.5938 (44.5441) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.2593 data: 0.1645 max mem: 3531 +Test: [ 70/517] eta: 0:02:13 acc1_pos: 50.917 acc5_pos: 100.000 loss: 0.3371 (0.7747) acc1: 95.0521 (51.8303) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.2594 data: 0.1645 max mem: 3531 +Test: [ 80/517] eta: 0:02:08 acc1_pos: 50.917 acc5_pos: 100.000 loss: 0.3338 (0.7224) acc1: 96.0938 (57.1599) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2522 data: 0.1588 max mem: 3531 +Test: [ 90/517] eta: 0:02:03 acc1_pos: 50.917 acc5_pos: 100.000 loss: 0.3365 (0.6786) acc1: 96.3542 (61.5528) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2512 data: 0.1586 max mem: 3531 +Test: [100/517] eta: 0:01:59 acc1_pos: 50.917 acc5_pos: 100.000 loss: 0.2864 (0.6383) acc1: 98.1771 (65.2254) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2527 data: 0.1551 max mem: 3531 +Test: [110/517] eta: 0:01:55 acc1_pos: 50.917 acc5_pos: 100.000 loss: 0.2454 (0.6027) acc1: 99.4792 (68.2831) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2546 data: 0.1561 max mem: 3531 +Test: [120/517] eta: 0:01:50 acc1_pos: 50.917 acc5_pos: 100.000 loss: 0.2657 (0.5784) acc1: 97.6562 (70.5557) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2457 data: 0.1526 max mem: 3531 +Test: [130/517] eta: 0:01:47 acc1_pos: 50.917 acc5_pos: 100.000 loss: 0.3102 (0.5576) acc1: 95.3125 (72.4018) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2488 data: 0.1575 max mem: 3531 +Test: [140/517] eta: 0:01:44 acc1_pos: 50.917 acc5_pos: 100.000 loss: 0.3023 (0.5383) acc1: 97.9167 (74.2871) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2601 data: 0.1666 max mem: 3531 +Test: [150/517] eta: 0:01:40 acc1_pos: 50.917 acc5_pos: 100.000 loss: 0.2813 (0.5215) acc1: 98.9583 (75.8882) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2469 data: 0.1539 max mem: 3531 +Test: [160/517] eta: 0:01:37 acc1_pos: 50.917 acc5_pos: 100.000 loss: 0.2981 (0.5086) acc1: 97.3958 (77.1383) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2456 data: 0.1540 max mem: 3531 +Test: [170/517] eta: 0:01:34 acc1_pos: 50.917 acc5_pos: 100.000 loss: 0.3093 (0.4958) acc1: 95.8333 (78.2651) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2648 data: 0.1731 max mem: 3531 +Test: [180/517] eta: 0:01:30 acc1_pos: 50.917 acc5_pos: 100.000 loss: 0.2508 (0.4813) acc1: 96.6146 (79.3667) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2478 data: 0.1545 max mem: 3531 +Test: [190/517] eta: 0:01:27 acc1_pos: 50.917 acc5_pos: 100.000 loss: 0.2209 (0.4675) acc1: 99.4792 (80.4169) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2378 data: 0.1401 max mem: 3531 +Test: [200/517] eta: 0:01:25 acc1_pos: 50.917 acc5_pos: 100.000 loss: 0.2285 (0.4562) acc1: 99.2188 (81.3524) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2566 data: 0.1598 max mem: 3531 +Test: [210/517] eta: 0:01:22 acc1_pos: 50.917 acc5_pos: 100.000 loss: 0.2656 (0.4507) acc1: 98.1771 (81.9634) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2514 data: 0.1579 max mem: 3532 +Test: [220/517] eta: 0:01:19 acc1_pos: 50.917 acc5_pos: 100.000 loss: 0.3493 (0.4460) acc1: 93.7500 (82.5038) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2403 data: 0.1473 max mem: 3532 +Test: [230/517] eta: 0:01:16 acc1_pos: 50.917 acc5_pos: 100.000 loss: 0.3530 (0.4415) acc1: 94.5312 (83.1259) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2446 data: 0.1515 max mem: 3532 +Test: [240/517] eta: 0:01:13 acc1_pos: 50.917 acc5_pos: 100.000 loss: 0.3232 (0.4361) acc1: 98.1771 (83.7710) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2577 data: 0.1646 max mem: 3532 +Test: [250/517] eta: 0:01:10 acc1_pos: 50.917 acc5_pos: 100.000 loss: 0.3010 (0.4294) acc1: 99.2188 (84.3812) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2598 data: 0.1644 max mem: 3532 +Test: [260/517] eta: 0:01:07 acc1_pos: 50.917 acc5_pos: 100.000 loss: 0.2579 (0.4233) acc1: 99.4792 (84.9507) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2549 data: 0.1588 max mem: 3532 +Test: [270/517] eta: 0:01:05 acc1_pos: 50.917 acc5_pos: 100.000 loss: 0.3121 (0.4205) acc1: 98.9583 (85.4023) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2687 data: 0.1721 max mem: 3532 +Test: [280/517] eta: 0:01:02 acc1_pos: 50.917 acc5_pos: 100.000 loss: 0.3418 (0.4173) acc1: 96.8750 (85.7522) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2649 data: 0.1674 max mem: 3532 +Test: [290/517] eta: 0:00:59 acc1_pos: 50.917 acc5_pos: 100.000 loss: 0.3285 (0.4140) acc1: 96.6146 (86.1406) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2452 data: 0.1493 max mem: 3532 +Test: [300/517] eta: 0:00:57 acc1_pos: 50.917 acc5_pos: 100.000 loss: 0.3116 (0.4102) acc1: 98.1771 (86.5648) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2449 data: 0.1485 max mem: 3532 +Test: [310/517] eta: 0:00:54 acc1_pos: 50.917 acc5_pos: 100.000 loss: 0.3119 (0.4075) acc1: 98.4375 (86.9080) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2445 data: 0.1459 max mem: 3532 +Test: [320/517] eta: 0:00:51 acc1_pos: 50.917 acc5_pos: 100.000 loss: 0.3087 (0.4035) acc1: 98.9583 (87.2720) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2348 data: 0.1370 max mem: 3532 +Test: [330/517] eta: 0:00:48 acc1_pos: 50.917 acc5_pos: 100.000 loss: 0.2649 (0.3989) acc1: 99.4792 (87.6345) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2313 data: 0.1357 max mem: 3532 +Test: [340/517] eta: 0:00:46 acc1_pos: 50.917 acc5_pos: 100.000 loss: 0.2491 (0.3965) acc1: 99.4792 (87.7795) acc5: 100.0000 (100.0000) acc1_neg: 77.083 acc5_neg: 100.000 time: 0.2599 data: 0.1637 max mem: 3532 +Test: [350/517] eta: 0:00:43 acc1_pos: 50.917 acc5_pos: 100.000 loss: 0.3151 (0.3960) acc1: 94.5312 (87.8940) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.2801 data: 0.1839 max mem: 3532 +Test: [360/517] eta: 0:00:41 acc1_pos: 50.917 acc5_pos: 100.000 loss: 0.3416 (0.3946) acc1: 93.2292 (88.0482) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2606 data: 0.1659 max mem: 3532 +Test: [370/517] eta: 0:00:38 acc1_pos: 50.917 acc5_pos: 100.000 loss: 0.2963 (0.3909) acc1: 96.8750 (88.3444) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2537 data: 0.1592 max mem: 3532 +Test: [380/517] eta: 0:00:35 acc1_pos: 50.917 acc5_pos: 100.000 loss: 0.2635 (0.3889) acc1: 98.4375 (88.5048) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2555 data: 0.1606 max mem: 3532 +Test: [390/517] eta: 0:00:33 acc1_pos: 50.917 acc5_pos: 100.000 loss: 0.3068 (0.3865) acc1: 95.8333 (88.7142) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2545 data: 0.1574 max mem: 3532 +Test: [400/517] eta: 0:00:30 acc1_pos: 50.917 acc5_pos: 100.000 loss: 0.2814 (0.3837) acc1: 96.3542 (88.9190) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2555 data: 0.1567 max mem: 3532 +Test: [410/517] eta: 0:00:27 acc1_pos: 50.917 acc5_pos: 100.000 loss: 0.2920 (0.3828) acc1: 95.0521 (88.9972) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.2517 data: 0.1566 max mem: 3532 +Test: [420/517] eta: 0:00:25 acc1_pos: 50.917 acc5_pos: 100.000 loss: 0.3036 (0.3808) acc1: 96.3542 (89.2023) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2548 data: 0.1622 max mem: 3532 +Test: [430/517] eta: 0:00:22 acc1_pos: 50.917 acc5_pos: 100.000 loss: 0.3001 (0.3789) acc1: 97.1354 (89.3803) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2590 data: 0.1658 max mem: 3532 +Test: [440/517] eta: 0:00:20 acc1_pos: 50.917 acc5_pos: 100.000 loss: 0.3035 (0.3776) acc1: 97.3958 (89.5650) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2546 data: 0.1611 max mem: 3533 +Test: [450/517] eta: 0:00:17 acc1_pos: 50.917 acc5_pos: 100.000 loss: 0.3105 (0.3759) acc1: 97.3958 (89.7439) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2513 data: 0.1572 max mem: 3533 +Test: [460/517] eta: 0:00:14 acc1_pos: 50.917 acc5_pos: 100.000 loss: 0.3196 (0.3750) acc1: 96.6146 (89.8534) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.2604 data: 0.1673 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 50.917 acc5_pos: 100.000 loss: 0.3196 (0.3734) acc1: 96.6146 (90.0124) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2636 data: 0.1714 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 50.917 acc5_pos: 100.000 loss: 0.2979 (0.3717) acc1: 98.6979 (90.2038) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2545 data: 0.1614 max mem: 3533 +Test: [490/517] eta: 0:00:07 acc1_pos: 50.917 acc5_pos: 100.000 loss: 0.2617 (0.3693) acc1: 99.2188 (90.3683) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2594 data: 0.1666 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 50.917 acc5_pos: 100.000 loss: 0.3088 (0.3698) acc1: 96.8750 (90.4082) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.2404 data: 0.1471 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 50.917 acc5_pos: 100.000 loss: 0.3578 (0.3686) acc1: 95.0521 (90.5154) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2013 data: 0.1085 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 50.917 acc5_pos: 100.000 loss: 0.2664 (0.3673) acc1: 98.1771 (90.6157) acc5: 100.0000 (100.0000) acc1_neg: 99.647 acc5_neg: 100.000 time: 0.2045 data: 0.1084 max mem: 3533 +Test: Total time: 0:02:12 (0.2565 s / it) +* Overall Acc@1 90.616 Acc@5 100.000 + target == real Acc@1 32.182 Acc@5 100.000 + target == fake Acc@1 96.678 Acc@5 100.000 + loss 0.367 + PR-AUC 0.9642 ROC-AUC 0.7836 EER 28.46% (thr=0.7061) +Accuracy of the network on the 198427 test images: 90.6% +[rank0]:[W726 04:19:58.406737352 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 04:20:18.789000 1183428 torch/distributed/run.py:766] +W0726 04:20:18.789000 1183428 torch/distributed/run.py:766] ***************************************** +W0726 04:20:18.789000 1183428 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:20:18.789000 1183428 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:20:47.591750329 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:20:47.919446874 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:20:47.098867643 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:20:48.456488007 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:20:48.481481283 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:20:48.610194140 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:20:48.630851893 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W726 04:20:48.630968902 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n45_seed6_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed6_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:45:11 acc1_pos: 43.750 acc5_pos: 100.000 loss: 0.7904 (0.7904) acc1: 43.7500 (43.7500) acc5: 100.0000 (100.0000) time: 6.9710 data: 5.3140 max mem: 3531 +Test: [ 10/389] eta: 0:04:33 acc1_pos: 95.833 acc5_pos: 100.000 loss: 0.5769 (0.5908) acc1: 77.6042 (66.8324) acc5: 100.0000 (100.0000) time: 0.7206 data: 0.4836 max mem: 3531 +Test: [ 20/389] eta: 0:02:36 acc1_pos: 28.125 acc5_pos: 100.000 loss: 0.5396 (0.6203) acc1: 74.7396 (62.3388) acc5: 100.0000 (100.0000) time: 0.0978 data: 0.0005 max mem: 3531 +Test: [ 30/389] eta: 0:02:24 acc1_pos: 74.740 acc5_pos: 100.000 loss: 0.7516 (0.6617) acc1: 46.8750 (57.6949) acc5: 100.0000 (100.0000) time: 0.2255 data: 0.1304 max mem: 3531 +Test: [ 40/389] eta: 0:02:06 acc1_pos: 8.854 acc5_pos: 100.000 loss: 0.7489 (0.6850) acc1: 47.3958 (55.6148) acc5: 100.0000 (100.0000) time: 0.2951 data: 0.2041 max mem: 3531 +Test: [ 50/389] eta: 0:01:58 acc1_pos: 69.792 acc5_pos: 100.000 loss: 0.6312 (0.6834) acc1: 64.3229 (56.3777) acc5: 100.0000 (100.0000) time: 0.2707 data: 0.1768 max mem: 3531 +Test: [ 60/389] eta: 0:01:47 acc1_pos: 56.250 acc5_pos: 100.000 loss: 0.6088 (0.6729) acc1: 66.4062 (57.9278) acc5: 100.0000 (100.0000) time: 0.2596 data: 0.1640 max mem: 3531 +Test: [ 70/389] eta: 0:01:39 acc1_pos: 54.948 acc5_pos: 100.000 loss: 0.6108 (0.6724) acc1: 65.6250 (57.8895) acc5: 100.0000 (100.0000) time: 0.2104 data: 0.1137 max mem: 3531 +Test: [ 80/389] eta: 0:01:31 acc1_pos: 89.583 acc5_pos: 100.000 loss: 0.7119 (0.6784) acc1: 53.9062 (57.3495) acc5: 100.0000 (100.0000) time: 0.1949 data: 0.0988 max mem: 3531 +Test: [ 90/389] eta: 0:01:24 acc1_pos: 82.292 acc5_pos: 100.000 loss: 0.4140 (0.6477) acc1: 79.6875 (60.4396) acc5: 100.0000 (100.0000) time: 0.1880 data: 0.0933 max mem: 3531 +Test: [100/389] eta: 0:01:19 acc1_pos: 44.271 acc5_pos: 100.000 loss: 0.5493 (0.6577) acc1: 70.5729 (59.4627) acc5: 100.0000 (100.0000) time: 0.1911 data: 0.0969 max mem: 3531 +Test: [110/389] eta: 0:01:14 acc1_pos: 90.625 acc5_pos: 100.000 loss: 0.7351 (0.6648) acc1: 50.2604 (58.8260) acc5: 100.0000 (100.0000) time: 0.1904 data: 0.0979 max mem: 3531 +Test: [120/389] eta: 0:01:09 acc1_pos: 66.146 acc5_pos: 100.000 loss: 0.6048 (0.6581) acc1: 64.8438 (59.5988) acc5: 100.0000 (100.0000) time: 0.1843 data: 0.0924 max mem: 3531 +Test: [130/389] eta: 0:01:05 acc1_pos: 15.337 acc5_pos: 100.000 loss: 0.6027 (0.6593) acc1: 66.1458 (59.4346) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.1873 data: 0.0962 max mem: 3531 +Test: [140/389] eta: 0:01:02 acc1_pos: 15.337 acc5_pos: 100.000 loss: 0.5195 (0.6501) acc1: 68.4896 (60.6715) acc5: 100.0000 (100.0000) acc1_neg: 49.479 acc5_neg: 100.000 time: 0.1960 data: 0.1047 max mem: 3531 +Test: [150/389] eta: 0:00:59 acc1_pos: 15.337 acc5_pos: 100.000 loss: 0.4931 (0.6466) acc1: 76.5625 (61.0858) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2020 data: 0.1106 max mem: 3531 +Test: [160/389] eta: 0:00:56 acc1_pos: 15.337 acc5_pos: 100.000 loss: 0.4931 (0.6352) acc1: 72.1354 (62.4224) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2083 data: 0.1168 max mem: 3531 +Test: [170/389] eta: 0:00:52 acc1_pos: 15.337 acc5_pos: 100.000 loss: 0.4349 (0.6277) acc1: 87.7604 (63.4214) acc5: 100.0000 (100.0000) acc1_neg: 57.812 acc5_neg: 100.000 time: 0.1849 data: 0.0913 max mem: 3531 +Test: [180/389] eta: 0:00:49 acc1_pos: 15.337 acc5_pos: 100.000 loss: 0.4745 (0.6245) acc1: 81.5104 (64.1042) acc5: 100.0000 (100.0000) acc1_neg: 64.062 acc5_neg: 100.000 time: 0.1799 data: 0.0864 max mem: 3531 +Test: [190/389] eta: 0:00:46 acc1_pos: 15.337 acc5_pos: 100.000 loss: 0.5648 (0.6260) acc1: 71.3542 (63.8743) acc5: 100.0000 (100.0000) acc1_neg: 23.177 acc5_neg: 100.000 time: 0.1899 data: 0.0975 max mem: 3531 +Test: [200/389] eta: 0:00:43 acc1_pos: 15.337 acc5_pos: 100.000 loss: 0.6070 (0.6274) acc1: 56.2500 (63.7179) acc5: 100.0000 (100.0000) acc1_neg: 48.177 acc5_neg: 100.000 time: 0.1801 data: 0.0882 max mem: 3531 +Test: [210/389] eta: 0:00:41 acc1_pos: 15.337 acc5_pos: 100.000 loss: 0.4446 (0.6173) acc1: 71.6146 (64.5611) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1822 data: 0.0913 max mem: 3532 +Test: [220/389] eta: 0:00:38 acc1_pos: 15.337 acc5_pos: 100.000 loss: 0.2825 (0.6032) acc1: 98.1771 (66.0209) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1750 data: 0.0847 max mem: 3532 +Test: [230/389] eta: 0:00:35 acc1_pos: 15.337 acc5_pos: 100.000 loss: 0.4033 (0.6027) acc1: 90.3646 (66.0252) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1801 data: 0.0890 max mem: 3532 +Test: [240/389] eta: 0:00:33 acc1_pos: 15.337 acc5_pos: 100.000 loss: 0.4937 (0.5982) acc1: 78.3854 (66.6742) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.1774 data: 0.0853 max mem: 3532 +Test: [250/389] eta: 0:00:30 acc1_pos: 15.337 acc5_pos: 100.000 loss: 0.4768 (0.5886) acc1: 87.5000 (67.8266) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1774 data: 0.0848 max mem: 3532 +Test: [260/389] eta: 0:00:28 acc1_pos: 15.337 acc5_pos: 100.000 loss: 0.4358 (0.5872) acc1: 93.4896 (68.1454) acc5: 100.0000 (100.0000) acc1_neg: 16.927 acc5_neg: 100.000 time: 0.1951 data: 0.1029 max mem: 3532 +Test: [270/389] eta: 0:00:26 acc1_pos: 15.337 acc5_pos: 100.000 loss: 0.5377 (0.5958) acc1: 80.2083 (67.3768) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.2005 data: 0.1083 max mem: 3532 +Test: [280/389] eta: 0:00:23 acc1_pos: 15.337 acc5_pos: 100.000 loss: 0.5612 (0.5972) acc1: 64.8438 (67.2857) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2021 data: 0.1092 max mem: 3532 +Test: [290/389] eta: 0:00:21 acc1_pos: 15.337 acc5_pos: 100.000 loss: 0.5070 (0.5925) acc1: 78.3854 (67.9303) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.2027 data: 0.1103 max mem: 3532 +Test: [300/389] eta: 0:00:19 acc1_pos: 15.337 acc5_pos: 100.000 loss: 0.5281 (0.5915) acc1: 80.2083 (68.2370) acc5: 100.0000 (100.0000) acc1_neg: 75.781 acc5_neg: 100.000 time: 0.1897 data: 0.0982 max mem: 3532 +Test: [310/389] eta: 0:00:17 acc1_pos: 15.337 acc5_pos: 100.000 loss: 0.5977 (0.5954) acc1: 70.5729 (67.9085) acc5: 100.0000 (100.0000) acc1_neg: 31.250 acc5_neg: 100.000 time: 0.1805 data: 0.0895 max mem: 3532 +Test: [320/389] eta: 0:00:14 acc1_pos: 15.337 acc5_pos: 100.000 loss: 0.6083 (0.5928) acc1: 69.0104 (68.2389) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1776 data: 0.0867 max mem: 3532 +Test: [330/389] eta: 0:00:12 acc1_pos: 15.337 acc5_pos: 100.000 loss: 0.7284 (0.5973) acc1: 50.2604 (67.6084) acc5: 100.0000 (100.0000) acc1_neg: 64.583 acc5_neg: 100.000 time: 0.1647 data: 0.0736 max mem: 3532 +Test: [340/389] eta: 0:00:10 acc1_pos: 15.337 acc5_pos: 100.000 loss: 0.4572 (0.5904) acc1: 67.4479 (68.4377) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1669 data: 0.0760 max mem: 3532 +Test: [350/389] eta: 0:00:08 acc1_pos: 15.337 acc5_pos: 100.000 loss: 0.3809 (0.5991) acc1: 96.8750 (67.7833) acc5: 100.0000 (100.0000) acc1_neg: 3.385 acc5_neg: 100.000 time: 0.1635 data: 0.0734 max mem: 3532 +Test: [360/389] eta: 0:00:06 acc1_pos: 15.337 acc5_pos: 100.000 loss: 0.7968 (0.6069) acc1: 41.6667 (66.8578) acc5: 100.0000 (100.0000) acc1_neg: 42.448 acc5_neg: 100.000 time: 0.1514 data: 0.0617 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 15.337 acc5_pos: 100.000 loss: 0.7599 (0.6044) acc1: 45.3125 (67.0668) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1401 data: 0.0504 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 15.337 acc5_pos: 100.000 loss: 0.4733 (0.6021) acc1: 75.7812 (67.3051) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.1340 data: 0.0444 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 15.337 acc5_pos: 100.000 loss: 0.4594 (0.5991) acc1: 81.5104 (67.6589) acc5: 100.0000 (100.0000) acc1_neg: 95.355 acc5_neg: 100.000 time: 0.1321 data: 0.0392 max mem: 3532 +Test: Total time: 0:01:19 (0.2038 s / it) +* Overall Acc@1 67.659 Acc@5 100.000 + target == real Acc@1 58.846 Acc@5 100.000 + target == fake Acc@1 71.755 Acc@5 100.000 + loss 0.599 + PR-AUC 0.8363 ROC-AUC 0.7211 EER 34.07% (thr=0.5381) +Accuracy of the network on the 149358 test images: 67.7% +[rank0]:[W726 04:22:14.175195995 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:22:34.834000 1183989 torch/distributed/run.py:766] +W0726 04:22:34.834000 1183989 torch/distributed/run.py:766] ***************************************** +W0726 04:22:34.834000 1183989 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:22:34.834000 1183989 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:23:03.079496319 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:23:03.325488321 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:23:04.492866309 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:23:04.796242153 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:23:04.969747216 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 7): env:// +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:23:04.047951360 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:23:04.053352045 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:23:04.078210341 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n45_seed6_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed6_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:29:01 acc1_pos: 89.062 acc5_pos: 100.000 loss: 0.3300 (0.3300) acc1: 89.0625 (89.0625) acc5: 100.0000 (100.0000) time: 5.2296 data: 3.5815 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/333] eta: 0:03:00 acc1_pos: 89.583 acc5_pos: 100.000 loss: 0.2945 (0.2772) acc1: 90.3646 (91.0038) acc5: 100.0000 (100.0000) time: 0.5573 data: 0.3259 max mem: 3531 +Test: [ 20/333] eta: 0:01:44 acc1_pos: 97.135 acc5_pos: 100.000 loss: 0.2936 (0.2904) acc1: 90.3646 (90.1662) acc5: 100.0000 (100.0000) time: 0.0902 data: 0.0004 max mem: 3531 +Test: [ 30/333] eta: 0:01:20 acc1_pos: 61.719 acc5_pos: 100.000 loss: 0.3942 (0.4196) acc1: 76.5625 (79.8471) acc5: 100.0000 (100.0000) time: 0.1043 data: 0.0109 max mem: 3531 +Test: [ 40/333] eta: 0:01:19 acc1_pos: 59.635 acc5_pos: 100.000 loss: 0.7125 (0.4928) acc1: 56.5104 (73.7297) acc5: 100.0000 (100.0000) time: 0.2069 data: 0.1117 max mem: 3531 +Test: [ 50/333] eta: 0:01:14 acc1_pos: 62.760 acc5_pos: 100.000 loss: 0.7125 (0.5349) acc1: 54.9479 (70.3278) acc5: 100.0000 (100.0000) time: 0.2568 data: 0.1649 max mem: 3531 +Test: [ 60/333] eta: 0:01:06 acc1_pos: 52.865 acc5_pos: 100.000 loss: 0.7104 (0.5644) acc1: 56.2500 (67.9645) acc5: 100.0000 (100.0000) time: 0.1882 data: 0.0980 max mem: 3531 +Test: [ 70/333] eta: 0:01:00 acc1_pos: 53.125 acc5_pos: 100.000 loss: 0.7104 (0.5849) acc1: 57.0312 (66.5860) acc5: 100.0000 (100.0000) time: 0.1484 data: 0.0579 max mem: 3531 +Test: [ 80/333] eta: 0:00:57 acc1_pos: 62.500 acc5_pos: 100.000 loss: 0.6942 (0.6012) acc1: 56.2500 (65.1942) acc5: 100.0000 (100.0000) time: 0.1788 data: 0.0851 max mem: 3531 +Test: [ 90/333] eta: 0:00:54 acc1_pos: 65.365 acc5_pos: 100.000 loss: 0.6625 (0.6050) acc1: 62.5000 (65.1414) acc5: 100.0000 (100.0000) time: 0.1996 data: 0.1053 max mem: 3531 +Test: [100/333] eta: 0:00:50 acc1_pos: 66.406 acc5_pos: 100.000 loss: 0.6528 (0.6106) acc1: 63.2812 (64.7793) acc5: 100.0000 (100.0000) time: 0.1700 data: 0.0783 max mem: 3531 +Test: [110/333] eta: 0:00:48 acc1_pos: 68.750 acc5_pos: 100.000 loss: 0.6516 (0.6136) acc1: 62.2396 (64.6842) acc5: 100.0000 (100.0000) time: 0.1849 data: 0.0908 max mem: 3531 +Test: [120/333] eta: 0:00:45 acc1_pos: 69.271 acc5_pos: 100.000 loss: 0.6227 (0.6162) acc1: 63.2812 (64.6049) acc5: 100.0000 (100.0000) time: 0.2099 data: 0.1163 max mem: 3531 +Test: [130/333] eta: 0:00:42 acc1_pos: 75.260 acc5_pos: 100.000 loss: 0.6055 (0.6167) acc1: 67.7083 (64.7642) acc5: 100.0000 (100.0000) time: 0.1854 data: 0.0941 max mem: 3531 +Test: [140/333] eta: 0:00:40 acc1_pos: 72.917 acc5_pos: 100.000 loss: 0.6055 (0.6159) acc1: 67.7083 (64.9435) acc5: 100.0000 (100.0000) time: 0.1898 data: 0.0978 max mem: 3531 +Test: [150/333] eta: 0:00:38 acc1_pos: 66.146 acc5_pos: 100.000 loss: 0.6103 (0.6185) acc1: 66.4062 (64.8420) acc5: 100.0000 (100.0000) time: 0.2166 data: 0.1235 max mem: 3531 +Test: [160/333] eta: 0:00:36 acc1_pos: 52.961 acc5_pos: 100.000 loss: 0.6383 (0.6273) acc1: 62.7604 (64.1968) acc5: 100.0000 (100.0000) acc1_neg: 46.094 acc5_neg: 100.000 time: 0.1916 data: 0.0994 max mem: 3531 +Test: [170/333] eta: 0:00:34 acc1_pos: 52.961 acc5_pos: 100.000 loss: 0.7397 (0.6369) acc1: 53.9062 (63.4579) acc5: 100.0000 (100.0000) acc1_neg: 55.990 acc5_neg: 100.000 time: 0.1889 data: 0.0983 max mem: 3531 +Test: [180/333] eta: 0:00:32 acc1_pos: 52.961 acc5_pos: 100.000 loss: 0.7135 (0.6387) acc1: 56.5104 (63.2913) acc5: 100.0000 (100.0000) acc1_neg: 48.438 acc5_neg: 100.000 time: 0.2186 data: 0.1266 max mem: 3531 +Test: [190/333] eta: 0:00:29 acc1_pos: 52.961 acc5_pos: 100.000 loss: 0.6661 (0.6402) acc1: 61.7188 (63.2104) acc5: 100.0000 (100.0000) acc1_neg: 62.240 acc5_neg: 100.000 time: 0.2073 data: 0.1121 max mem: 3531 +Test: [200/333] eta: 0:00:27 acc1_pos: 52.961 acc5_pos: 100.000 loss: 0.6700 (0.6408) acc1: 61.7188 (63.2761) acc5: 100.0000 (100.0000) acc1_neg: 60.938 acc5_neg: 100.000 time: 0.1995 data: 0.1054 max mem: 3531 +Test: [210/333] eta: 0:00:25 acc1_pos: 52.961 acc5_pos: 100.000 loss: 0.6870 (0.6449) acc1: 59.8958 (63.0640) acc5: 100.0000 (100.0000) acc1_neg: 51.562 acc5_neg: 100.000 time: 0.2073 data: 0.1155 max mem: 3532 +Test: [220/333] eta: 0:00:23 acc1_pos: 52.961 acc5_pos: 100.000 loss: 0.7443 (0.6509) acc1: 55.9896 (62.6473) acc5: 100.0000 (100.0000) acc1_neg: 71.094 acc5_neg: 100.000 time: 0.2117 data: 0.1199 max mem: 3532 +Test: [230/333] eta: 0:00:21 acc1_pos: 52.961 acc5_pos: 100.000 loss: 0.5330 (0.6447) acc1: 73.1771 (63.3275) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.2051 data: 0.1120 max mem: 3532 +Test: [240/333] eta: 0:00:19 acc1_pos: 52.961 acc5_pos: 100.000 loss: 0.5330 (0.6442) acc1: 73.9583 (63.3580) acc5: 100.0000 (100.0000) acc1_neg: 63.542 acc5_neg: 100.000 time: 0.2047 data: 0.1116 max mem: 3532 +Test: [250/333] eta: 0:00:17 acc1_pos: 52.961 acc5_pos: 100.000 loss: 0.6577 (0.6454) acc1: 63.5417 (63.3030) acc5: 100.0000 (100.0000) acc1_neg: 52.604 acc5_neg: 100.000 time: 0.2040 data: 0.1130 max mem: 3532 +Test: [260/333] eta: 0:00:15 acc1_pos: 52.961 acc5_pos: 100.000 loss: 0.6761 (0.6479) acc1: 61.9792 (63.1825) acc5: 100.0000 (100.0000) acc1_neg: 65.104 acc5_neg: 100.000 time: 0.2091 data: 0.1181 max mem: 3532 +Test: [270/333] eta: 0:00:13 acc1_pos: 52.961 acc5_pos: 100.000 loss: 0.6165 (0.6450) acc1: 66.1458 (63.4927) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.2170 data: 0.1266 max mem: 3532 +Test: [280/333] eta: 0:00:11 acc1_pos: 52.961 acc5_pos: 100.000 loss: 0.5693 (0.6475) acc1: 69.5312 (63.2655) acc5: 100.0000 (100.0000) acc1_neg: 58.854 acc5_neg: 100.000 time: 0.2042 data: 0.1122 max mem: 3532 +Test: [290/333] eta: 0:00:08 acc1_pos: 52.961 acc5_pos: 100.000 loss: 0.8198 (0.6537) acc1: 47.9167 (62.6790) acc5: 100.0000 (100.0000) acc1_neg: 52.865 acc5_neg: 100.000 time: 0.2083 data: 0.1158 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 52.961 acc5_pos: 100.000 loss: 0.8293 (0.6594) acc1: 45.8333 (62.2482) acc5: 100.0000 (100.0000) acc1_neg: 50.000 acc5_neg: 100.000 time: 0.2057 data: 0.1139 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 52.961 acc5_pos: 100.000 loss: 0.7201 (0.6569) acc1: 57.8125 (62.4941) acc5: 100.0000 (100.0000) acc1_neg: 64.323 acc5_neg: 100.000 time: 0.1791 data: 0.0881 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 52.961 acc5_pos: 100.000 loss: 0.6115 (0.6564) acc1: 66.9271 (62.5625) acc5: 100.0000 (100.0000) acc1_neg: 51.823 acc5_neg: 100.000 time: 0.1400 data: 0.0499 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 52.961 acc5_pos: 100.000 loss: 0.7808 (0.6618) acc1: 51.8229 (62.1467) acc5: 100.0000 (100.0000) acc1_neg: 49.740 acc5_neg: 100.000 time: 0.1221 data: 0.0323 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 52.961 acc5_pos: 100.000 loss: 0.8124 (0.6628) acc1: 50.0000 (62.0832) acc5: 100.0000 (100.0000) acc1_neg: 49.577 acc5_neg: 100.000 time: 0.1164 data: 0.0234 max mem: 3532 +Test: Total time: 0:01:07 (0.2021 s / it) +* Overall Acc@1 62.083 Acc@5 100.000 + target == real Acc@1 64.790 Acc@5 100.000 + target == fake Acc@1 59.673 Acc@5 100.000 + loss 0.663 + PR-AUC 0.6813 ROC-AUC 0.6705 EER 37.76% (thr=0.4824) +Accuracy of the network on the 127843 test images: 62.1% +[rank0]:[W726 04:24:21.913354125 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:24:41.923000 1184553 torch/distributed/run.py:766] +W0726 04:24:41.923000 1184553 torch/distributed/run.py:766] ***************************************** +W0726 04:24:41.923000 1184553 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:24:41.923000 1184553 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:25:11.694681254 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:25:11.201627627 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:25:11.233475682 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:25:12.425651792 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +[rank0]:[W726 04:25:12.679565920 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:25:12.681951646 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +[rank6]:[W726 04:25:12.712234812 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:25:12.725349377 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n45_seed6_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed6_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:25:27 acc1_pos: 94.010 acc5_pos: 100.000 loss: 0.2382 (0.2382) acc1: 94.0104 (94.0104) acc5: 100.0000 (100.0000) time: 6.4987 data: 4.6710 max mem: 3531 +Test: [ 10/235] eta: 0:02:31 acc1_pos: 94.792 acc5_pos: 100.000 loss: 0.2380 (0.2400) acc1: 94.7917 (94.6023) acc5: 100.0000 (100.0000) time: 0.6740 data: 0.4250 max mem: 3531 +Test: [ 20/235] eta: 0:01:26 acc1_pos: 96.094 acc5_pos: 100.000 loss: 0.2355 (0.2354) acc1: 95.3125 (94.8537) acc5: 100.0000 (100.0000) time: 0.0957 data: 0.0005 max mem: 3531 +Test: [ 30/235] eta: 0:01:04 acc1_pos: 95.833 acc5_pos: 100.000 loss: 0.2316 (0.2341) acc1: 95.3125 (94.9513) acc5: 100.0000 (100.0000) time: 0.1149 data: 0.0156 max mem: 3531 +Test: [ 40/235] eta: 0:00:53 acc1_pos: 94.010 acc5_pos: 100.000 loss: 0.2270 (0.2332) acc1: 94.7917 (94.9886) acc5: 100.0000 (100.0000) time: 0.1412 data: 0.0461 max mem: 3531 +Test: [ 50/235] eta: 0:00:46 acc1_pos: 93.750 acc5_pos: 100.000 loss: 0.2293 (0.2342) acc1: 94.2708 (94.9397) acc5: 100.0000 (100.0000) time: 0.1614 data: 0.0697 max mem: 3531 +Test: [ 60/235] eta: 0:00:41 acc1_pos: 94.792 acc5_pos: 100.000 loss: 0.2384 (0.2349) acc1: 94.7917 (94.9240) acc5: 100.0000 (100.0000) time: 0.1652 data: 0.0730 max mem: 3531 +Test: [ 70/235] eta: 0:00:37 acc1_pos: 92.308 acc5_pos: 100.000 loss: 0.2530 (0.3727) acc1: 92.7083 (85.9632) acc5: 100.0000 (100.0000) acc1_neg: 20.833 acc5_neg: 100.000 time: 0.1642 data: 0.0724 max mem: 3531 +Test: [ 80/235] eta: 0:00:34 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.3363 (0.4935) acc1: 23.4375 (78.2086) acc5: 100.0000 (100.0000) acc1_neg: 21.094 acc5_neg: 100.000 time: 0.1633 data: 0.0704 max mem: 3531 +Test: [ 90/235] eta: 0:00:30 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.3566 (0.5879) acc1: 22.9167 (72.1669) acc5: 100.0000 (100.0000) acc1_neg: 24.219 acc5_neg: 100.000 time: 0.1540 data: 0.0577 max mem: 3531 +Test: [100/235] eta: 0:00:28 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.3458 (0.6617) acc1: 23.4375 (67.4144) acc5: 100.0000 (100.0000) acc1_neg: 23.438 acc5_neg: 100.000 time: 0.1649 data: 0.0690 max mem: 3531 +Test: [110/235] eta: 0:00:26 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.3353 (0.7222) acc1: 23.9583 (63.5135) acc5: 100.0000 (100.0000) acc1_neg: 25.521 acc5_neg: 100.000 time: 0.2023 data: 0.1078 max mem: 3531 +Test: [120/235] eta: 0:00:24 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.3895 (0.7840) acc1: 19.5312 (59.5795) acc5: 100.0000 (100.0000) acc1_neg: 18.229 acc5_neg: 100.000 time: 0.2277 data: 0.1299 max mem: 3531 +Test: [130/235] eta: 0:00:22 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.4702 (0.8357) acc1: 16.4062 (56.3673) acc5: 100.0000 (100.0000) acc1_neg: 17.708 acc5_neg: 100.000 time: 0.2350 data: 0.1396 max mem: 3531 +Test: [140/235] eta: 0:00:19 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.4819 (0.8832) acc1: 16.4062 (53.4611) acc5: 100.0000 (100.0000) acc1_neg: 11.979 acc5_neg: 100.000 time: 0.1976 data: 0.1064 max mem: 3531 +Test: [150/235] eta: 0:00:17 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.4870 (0.9219) acc1: 16.4062 (51.0192) acc5: 100.0000 (100.0000) acc1_neg: 16.406 acc5_neg: 100.000 time: 0.1829 data: 0.0889 max mem: 3531 +Test: [160/235] eta: 0:00:15 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.4638 (0.9560) acc1: 16.6667 (48.9066) acc5: 100.0000 (100.0000) acc1_neg: 17.188 acc5_neg: 100.000 time: 0.2062 data: 0.1127 max mem: 3531 +Test: [170/235] eta: 0:00:13 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.4243 (0.9633) acc1: 17.9688 (48.0537) acc5: 100.0000 (100.0000) acc1_neg: 33.073 acc5_neg: 100.000 time: 0.2058 data: 0.1145 max mem: 3531 +Test: [180/235] eta: 0:00:11 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.0465 (0.9675) acc1: 36.4583 (47.4390) acc5: 100.0000 (100.0000) acc1_neg: 33.594 acc5_neg: 100.000 time: 0.2089 data: 0.1166 max mem: 3531 +Test: [190/235] eta: 0:00:09 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.0307 (0.9713) acc1: 36.4583 (46.8559) acc5: 100.0000 (100.0000) acc1_neg: 36.198 acc5_neg: 100.000 time: 0.2036 data: 0.1114 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.0307 (0.9754) acc1: 35.4167 (46.2855) acc5: 100.0000 (100.0000) acc1_neg: 33.854 acc5_neg: 100.000 time: 0.1916 data: 0.0981 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.0398 (0.9783) acc1: 35.6771 (45.8654) acc5: 100.0000 (100.0000) acc1_neg: 38.542 acc5_neg: 100.000 time: 0.2176 data: 0.1245 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.0369 (0.9808) acc1: 36.7188 (45.4751) acc5: 100.0000 (100.0000) acc1_neg: 33.333 acc5_neg: 100.000 time: 0.2235 data: 0.1323 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.0266 (0.9826) acc1: 38.0208 (45.1457) acc5: 100.0000 (100.0000) acc1_neg: 35.677 acc5_neg: 100.000 time: 0.1785 data: 0.0865 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.0363 (0.9847) acc1: 37.5000 (45.0339) acc5: 100.0000 (100.0000) acc1_neg: 34.783 acc5_neg: 100.000 time: 0.1290 data: 0.0401 max mem: 3532 +Test: Total time: 0:00:48 (0.2063 s / it) +* Overall Acc@1 45.034 Acc@5 100.000 + target == real Acc@1 94.898 Acc@5 100.000 + target == fake Acc@1 26.999 Acc@5 100.000 + loss 0.985 + PR-AUC 0.8840 ROC-AUC 0.7387 EER 32.33% (thr=0.2064) +Accuracy of the network on the 89879 test images: 45.0% +[rank0]:[W726 04:26:12.712520220 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:26:32.747000 1185104 torch/distributed/run.py:766] +W0726 04:26:32.747000 1185104 torch/distributed/run.py:766] ***************************************** +W0726 04:26:32.747000 1185104 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:26:32.747000 1185104 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:27:00.304047384 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:27:01.671926421 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:27:01.970642471 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:27:01.160529649 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:27:01.387630957 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:27:01.412411873 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:27:02.456845410 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:27:02.466550137 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n45_seed6_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed6_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:24:43 acc1_pos: 54.427 acc5_pos: 100.000 loss: 0.7150 (0.7150) acc1: 54.4271 (54.4271) acc5: 100.0000 (100.0000) time: 5.5157 data: 3.7061 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/269] eta: 0:02:40 acc1_pos: 56.510 acc5_pos: 100.000 loss: 0.6688 (0.6460) acc1: 58.3333 (61.9318) acc5: 100.0000 (100.0000) time: 0.6190 data: 0.3728 max mem: 3531 +Test: [ 20/269] eta: 0:01:33 acc1_pos: 62.500 acc5_pos: 100.000 loss: 0.6673 (0.6331) acc1: 58.3333 (63.4921) acc5: 100.0000 (100.0000) time: 0.1183 data: 0.0249 max mem: 3531 +Test: [ 30/269] eta: 0:01:22 acc1_pos: 68.750 acc5_pos: 100.000 loss: 0.6102 (0.6223) acc1: 62.5000 (65.3394) acc5: 100.0000 (100.0000) time: 0.1944 data: 0.0983 max mem: 3531 +Test: [ 40/269] eta: 0:01:14 acc1_pos: 65.517 acc5_pos: 100.000 loss: 0.5810 (0.6007) acc1: 68.7500 (67.8862) acc5: 100.0000 (100.0000) acc1_neg: 78.646 acc5_neg: 100.000 time: 0.2713 data: 0.1783 max mem: 3531 +Test: [ 50/269] eta: 0:01:06 acc1_pos: 65.517 acc5_pos: 100.000 loss: 0.5428 (0.5993) acc1: 75.2604 (68.0453) acc5: 100.0000 (100.0000) acc1_neg: 45.052 acc5_neg: 100.000 time: 0.2359 data: 0.1450 max mem: 3531 +Test: [ 60/269] eta: 0:00:59 acc1_pos: 65.517 acc5_pos: 100.000 loss: 0.5927 (0.6125) acc1: 67.1875 (66.3209) acc5: 100.0000 (100.0000) acc1_neg: 42.448 acc5_neg: 100.000 time: 0.2092 data: 0.1175 max mem: 3531 +Test: [ 70/269] eta: 0:00:55 acc1_pos: 65.517 acc5_pos: 100.000 loss: 0.6698 (0.6140) acc1: 56.7708 (65.8341) acc5: 100.0000 (100.0000) acc1_neg: 64.323 acc5_neg: 100.000 time: 0.2232 data: 0.1305 max mem: 3531 +Test: [ 80/269] eta: 0:00:51 acc1_pos: 65.517 acc5_pos: 100.000 loss: 0.5892 (0.6094) acc1: 69.7917 (66.3645) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.2244 data: 0.1318 max mem: 3531 +Test: [ 90/269] eta: 0:00:47 acc1_pos: 65.517 acc5_pos: 100.000 loss: 0.6219 (0.6218) acc1: 58.0729 (64.8151) acc5: 100.0000 (100.0000) acc1_neg: 48.438 acc5_neg: 100.000 time: 0.2009 data: 0.1095 max mem: 3531 +Test: [100/269] eta: 0:00:43 acc1_pos: 65.517 acc5_pos: 100.000 loss: 0.7212 (0.6322) acc1: 50.5208 (63.4643) acc5: 100.0000 (100.0000) acc1_neg: 48.698 acc5_neg: 100.000 time: 0.2020 data: 0.1083 max mem: 3531 +Test: [110/269] eta: 0:00:40 acc1_pos: 65.517 acc5_pos: 100.000 loss: 0.7225 (0.6397) acc1: 51.0417 (62.3850) acc5: 100.0000 (100.0000) acc1_neg: 44.271 acc5_neg: 100.000 time: 0.2045 data: 0.1098 max mem: 3531 +Test: [120/269] eta: 0:00:36 acc1_pos: 65.517 acc5_pos: 100.000 loss: 0.6822 (0.6348) acc1: 57.0312 (63.0316) acc5: 100.0000 (100.0000) acc1_neg: 67.708 acc5_neg: 100.000 time: 0.1983 data: 0.1059 max mem: 3531 +Test: [130/269] eta: 0:00:33 acc1_pos: 65.517 acc5_pos: 100.000 loss: 0.6142 (0.6356) acc1: 65.6250 (63.0526) acc5: 100.0000 (100.0000) acc1_neg: 50.260 acc5_neg: 100.000 time: 0.1993 data: 0.1076 max mem: 3531 +Test: [140/269] eta: 0:00:31 acc1_pos: 65.517 acc5_pos: 100.000 loss: 0.7105 (0.6423) acc1: 54.4271 (62.2562) acc5: 100.0000 (100.0000) acc1_neg: 49.219 acc5_neg: 100.000 time: 0.1997 data: 0.1090 max mem: 3531 +Test: [150/269] eta: 0:00:28 acc1_pos: 65.517 acc5_pos: 100.000 loss: 0.7105 (0.6424) acc1: 54.4271 (62.1516) acc5: 100.0000 (100.0000) acc1_neg: 67.708 acc5_neg: 100.000 time: 0.1973 data: 0.1059 max mem: 3531 +Test: [160/269] eta: 0:00:25 acc1_pos: 65.517 acc5_pos: 100.000 loss: 0.5925 (0.6393) acc1: 67.1875 (62.5291) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.1998 data: 0.1082 max mem: 3531 +Test: [170/269] eta: 0:00:23 acc1_pos: 65.517 acc5_pos: 100.000 loss: 0.5949 (0.6392) acc1: 68.4896 (62.6142) acc5: 100.0000 (100.0000) acc1_neg: 72.135 acc5_neg: 100.000 time: 0.1990 data: 0.1079 max mem: 3531 +Test: [180/269] eta: 0:00:20 acc1_pos: 65.517 acc5_pos: 100.000 loss: 0.5676 (0.6329) acc1: 72.1354 (63.3949) acc5: 100.0000 (100.0000) acc1_neg: 65.885 acc5_neg: 100.000 time: 0.2009 data: 0.1079 max mem: 3531 +Test: [190/269] eta: 0:00:18 acc1_pos: 65.517 acc5_pos: 100.000 loss: 0.5671 (0.6332) acc1: 72.1354 (63.2990) acc5: 100.0000 (100.0000) acc1_neg: 70.312 acc5_neg: 100.000 time: 0.2141 data: 0.1210 max mem: 3531 +Test: [200/269] eta: 0:00:15 acc1_pos: 65.517 acc5_pos: 100.000 loss: 0.6400 (0.6337) acc1: 62.5000 (63.3305) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.2185 data: 0.1248 max mem: 3531 +Test: [210/269] eta: 0:00:13 acc1_pos: 65.517 acc5_pos: 100.000 loss: 0.6277 (0.6321) acc1: 63.2812 (63.4614) acc5: 100.0000 (100.0000) acc1_neg: 73.958 acc5_neg: 100.000 time: 0.2045 data: 0.1115 max mem: 3532 +Test: [220/269] eta: 0:00:11 acc1_pos: 65.517 acc5_pos: 100.000 loss: 0.5651 (0.6303) acc1: 65.8854 (63.7349) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1779 data: 0.0877 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 65.517 acc5_pos: 100.000 loss: 0.5091 (0.6259) acc1: 76.0417 (64.2293) acc5: 100.0000 (100.0000) acc1_neg: 76.042 acc5_neg: 100.000 time: 0.1558 data: 0.0658 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 65.517 acc5_pos: 100.000 loss: 0.6024 (0.6280) acc1: 66.9271 (64.0593) acc5: 100.0000 (100.0000) acc1_neg: 70.573 acc5_neg: 100.000 time: 0.1362 data: 0.0462 max mem: 3532 +Test: [250/269] eta: 0:00:04 acc1_pos: 65.517 acc5_pos: 100.000 loss: 0.6739 (0.6315) acc1: 55.7292 (63.6133) acc5: 100.0000 (100.0000) acc1_neg: 27.344 acc5_neg: 100.000 time: 0.1238 data: 0.0340 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 65.517 acc5_pos: 100.000 loss: 0.7644 (0.6380) acc1: 44.5312 (62.8063) acc5: 100.0000 (100.0000) acc1_neg: 56.250 acc5_neg: 100.000 time: 0.1194 data: 0.0295 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 65.517 acc5_pos: 100.000 loss: 0.7068 (0.6384) acc1: 44.5312 (62.7661) acc5: 100.0000 (100.0000) acc1_neg: 81.890 acc5_neg: 100.000 time: 0.1234 data: 0.0301 max mem: 3532 +Test: Total time: 0:00:56 (0.2091 s / it) +* Overall Acc@1 62.766 Acc@5 100.000 + target == real Acc@1 65.269 Acc@5 100.000 + target == fake Acc@1 62.424 Acc@5 100.000 + loss 0.638 + PR-AUC 0.9387 ROC-AUC 0.6933 EER 36.22% (thr=0.4932) +Accuracy of the network on the 103293 test images: 62.8% +[rank0]:[W726 04:28:07.524246698 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T04:28:13Z] eval_done tag=fixed500k_fake_n45_seed6_resampled +[2026-07-26T04:28:14Z] eval_start tag=fixed500k_real_n23_seed7_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed7_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T04:28:14Z +hostname=n116-041-023 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed7_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n23_seed7_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 04:28:27.682000 1185681 torch/distributed/run.py:766] +W0726 04:28:27.682000 1185681 torch/distributed/run.py:766] ***************************************** +W0726 04:28:27.682000 1185681 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:28:27.682000 1185681 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:28:55.102133907 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:28:55.345156088 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:28:56.898857908 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:28:56.994657201 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:28:56.098336421 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:28:56.103794897 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:28:56.108694685 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:28:56.136266436 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n23_seed7_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed7_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:21:19 acc1_pos: 66.667 acc5_pos: 100.000 loss: 0.6065 (0.6065) acc1: 66.6667 (66.6667) acc5: 100.0000 (100.0000) time: 5.4688 data: 3.8654 max mem: 3531 +Test: [ 10/234] eta: 0:02:09 acc1_pos: 34.115 acc5_pos: 100.000 loss: 0.8095 (0.8347) acc1: 49.4792 (43.8684) acc5: 100.0000 (100.0000) time: 0.5791 data: 0.3518 max mem: 3531 +Test: [ 20/234] eta: 0:01:14 acc1_pos: 94.531 acc5_pos: 100.000 loss: 0.7563 (0.7565) acc1: 49.7396 (53.0258) acc5: 100.0000 (100.0000) time: 0.0942 data: 0.0004 max mem: 3531 +Test: [ 30/234] eta: 0:00:55 acc1_pos: 50.000 acc5_pos: 100.000 loss: 0.2363 (0.6160) acc1: 94.2708 (64.1717) acc5: 100.0000 (100.0000) time: 0.1013 data: 0.0035 max mem: 3531 +Test: [ 40/234] eta: 0:00:48 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.2234 (0.6084) acc1: 94.7917 (64.5579) acc5: 100.0000 (100.0000) time: 0.1471 data: 0.0506 max mem: 3531 +Test: [ 50/234] eta: 0:00:43 acc1_pos: 97.656 acc5_pos: 100.000 loss: 0.1758 (0.5221) acc1: 96.0938 (70.8742) acc5: 100.0000 (100.0000) time: 0.1793 data: 0.0865 max mem: 3531 +Test: [ 60/234] eta: 0:00:38 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.1341 (0.4517) acc1: 98.1771 (75.6190) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1682 data: 0.0762 max mem: 3531 +Test: [ 70/234] eta: 0:00:36 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.0902 (0.4444) acc1: 100.0000 (76.2691) acc5: 100.0000 (100.0000) acc1_neg: 70.052 acc5_neg: 100.000 time: 0.1883 data: 0.0922 max mem: 3531 +Test: [ 80/234] eta: 0:00:32 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.4946 (0.4403) acc1: 74.7396 (76.9612) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1807 data: 0.0854 max mem: 3531 +Test: [ 90/234] eta: 0:00:29 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.2163 (0.4092) acc1: 92.7083 (79.0951) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1555 data: 0.0645 max mem: 3531 +Test: [100/234] eta: 0:00:27 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.2304 (0.4051) acc1: 92.7083 (79.5070) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1593 data: 0.0683 max mem: 3531 +Test: [110/234] eta: 0:00:24 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.3746 (0.3976) acc1: 81.7708 (80.1591) acc5: 100.0000 (100.0000) acc1_neg: 76.042 acc5_neg: 100.000 time: 0.1666 data: 0.0754 max mem: 3531 +Test: [120/234] eta: 0:00:22 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.2740 (0.3994) acc1: 87.5000 (80.3741) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1768 data: 0.0848 max mem: 3531 +Test: [130/234] eta: 0:00:20 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.3428 (0.3976) acc1: 86.9792 (80.8484) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.2030 data: 0.1097 max mem: 3531 +Test: [140/234] eta: 0:00:18 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.4174 (0.4157) acc1: 81.5104 (80.0440) acc5: 100.0000 (100.0000) acc1_neg: 46.094 acc5_neg: 100.000 time: 0.2184 data: 0.1265 max mem: 3531 +Test: [150/234] eta: 0:00:16 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.3889 (0.4179) acc1: 80.2083 (80.0031) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.1857 data: 0.0932 max mem: 3531 +Test: [160/234] eta: 0:00:14 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.3805 (0.4144) acc1: 83.3333 (80.4590) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.1798 data: 0.0858 max mem: 3531 +Test: [170/234] eta: 0:00:12 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.4077 (0.4146) acc1: 84.1146 (80.6698) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2138 data: 0.1207 max mem: 3531 +Test: [180/234] eta: 0:00:10 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.4238 (0.4212) acc1: 82.5521 (80.2515) acc5: 100.0000 (100.0000) acc1_neg: 47.656 acc5_neg: 100.000 time: 0.2167 data: 0.1254 max mem: 3531 +Test: [190/234] eta: 0:00:08 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.6443 (0.4343) acc1: 66.6667 (79.4475) acc5: 100.0000 (100.0000) acc1_neg: 63.802 acc5_neg: 100.000 time: 0.2124 data: 0.1218 max mem: 3531 +Test: [200/234] eta: 0:00:06 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.6342 (0.4281) acc1: 67.4479 (79.8184) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2081 data: 0.1173 max mem: 3531 +Test: [210/234] eta: 0:00:04 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.1096 (0.4136) acc1: 99.7396 (80.7625) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2097 data: 0.1187 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.1440 (0.4022) acc1: 99.2188 (81.5269) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.1984 data: 0.1072 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.2998 (0.4120) acc1: 69.7917 (80.7934) acc5: 100.0000 (100.0000) acc1_neg: 65.104 acc5_neg: 100.000 time: 0.1454 data: 0.0547 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.6041 (0.4145) acc1: 65.8854 (80.5871) acc5: 100.0000 (100.0000) acc1_neg: 62.646 acc5_neg: 100.000 time: 0.1342 data: 0.0429 max mem: 3532 +Test: Total time: 0:00:46 (0.1966 s / it) +* Overall Acc@1 80.587 Acc@5 100.000 + target == real Acc@1 71.058 Acc@5 100.000 + target == fake Acc@1 83.273 Acc@5 100.000 + loss 0.415 + PR-AUC 0.9611 ROC-AUC 0.8749 EER 21.90% (thr=0.5854) +Accuracy of the network on the 89729 test images: 80.6% +[rank0]:[W726 04:29:55.782275594 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:30:15.320000 1186243 torch/distributed/run.py:766] +W0726 04:30:15.320000 1186243 torch/distributed/run.py:766] ***************************************** +W0726 04:30:15.320000 1186243 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:30:15.320000 1186243 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:30:44.238391295 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:30:45.810753495 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:30:45.875223826 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:30:45.089908045 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:30:45.122707084 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +[rank4]:[W726 04:30:45.210357280 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:30:45.232658636 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +NCCL version 2.26.2+cuda12.2 +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:30:45.277477788 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n23_seed7_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed7_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:46:16 acc1_pos: 38.802 acc5_pos: 100.000 loss: 0.8960 (0.8960) acc1: 38.8021 (38.8021) acc5: 100.0000 (100.0000) time: 7.1960 data: 5.7375 max mem: 3531 +Test: [ 10/2387] eta: 0:29:11 acc1_pos: 36.198 acc5_pos: 100.000 loss: 0.8503 (0.7707) acc1: 43.7500 (50.9706) acc5: 100.0000 (100.0000) time: 0.7368 data: 0.5220 max mem: 3531 +Test: [ 20/2387] eta: 0:17:06 acc1_pos: 46.094 acc5_pos: 100.000 loss: 0.7885 (0.8039) acc1: 46.0938 (49.5164) acc5: 100.0000 (100.0000) time: 0.0954 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:15:05 acc1_pos: 36.979 acc5_pos: 100.000 loss: 0.8267 (0.8054) acc1: 46.0938 (49.0591) acc5: 100.0000 (100.0000) time: 0.1899 data: 0.0941 max mem: 3531 +Test: [ 40/2387] eta: 0:13:33 acc1_pos: 44.271 acc5_pos: 100.000 loss: 0.8329 (0.8193) acc1: 42.4479 (47.0910) acc5: 100.0000 (100.0000) time: 0.2553 data: 0.1640 max mem: 3531 +Test: [ 50/2387] eta: 0:12:45 acc1_pos: 25.000 acc5_pos: 100.000 loss: 0.8668 (0.8515) acc1: 36.9792 (44.4904) acc5: 100.0000 (100.0000) time: 0.2402 data: 0.1481 max mem: 3531 +Test: [ 60/2387] eta: 0:12:09 acc1_pos: 40.365 acc5_pos: 100.000 loss: 0.9815 (0.8780) acc1: 29.4271 (42.4778) acc5: 100.0000 (100.0000) time: 0.2460 data: 0.1529 max mem: 3531 +Test: [ 70/2387] eta: 0:11:42 acc1_pos: 32.031 acc5_pos: 100.000 loss: 0.9927 (0.9095) acc1: 27.6042 (40.0968) acc5: 100.0000 (100.0000) time: 0.2410 data: 0.1492 max mem: 3531 +Test: [ 80/2387] eta: 0:11:32 acc1_pos: 76.302 acc5_pos: 100.000 loss: 0.9470 (0.8866) acc1: 34.1146 (42.0010) acc5: 100.0000 (100.0000) time: 0.2585 data: 0.1655 max mem: 3531 +Test: [ 90/2387] eta: 0:11:16 acc1_pos: 58.854 acc5_pos: 100.000 loss: 0.6412 (0.8564) acc1: 59.1146 (44.5084) acc5: 100.0000 (100.0000) time: 0.2636 data: 0.1684 max mem: 3531 +Test: [ 100/2387] eta: 0:11:08 acc1_pos: 61.979 acc5_pos: 100.000 loss: 0.6399 (0.8426) acc1: 59.8958 (45.8282) acc5: 100.0000 (100.0000) time: 0.2618 data: 0.1653 max mem: 3531 +Test: [ 110/2387] eta: 0:10:54 acc1_pos: 56.250 acc5_pos: 100.000 loss: 0.6636 (0.8251) acc1: 61.1979 (47.6117) acc5: 100.0000 (100.0000) time: 0.2556 data: 0.1600 max mem: 3531 +Test: [ 120/2387] eta: 0:10:48 acc1_pos: 55.729 acc5_pos: 100.000 loss: 0.6873 (0.8190) acc1: 60.6771 (47.9705) acc5: 100.0000 (100.0000) time: 0.2531 data: 0.1572 max mem: 3531 +Test: [ 130/2387] eta: 0:10:39 acc1_pos: 59.635 acc5_pos: 100.000 loss: 0.6873 (0.8062) acc1: 59.3750 (49.0597) acc5: 100.0000 (100.0000) time: 0.2604 data: 0.1662 max mem: 3531 +Test: [ 140/2387] eta: 0:10:30 acc1_pos: 54.427 acc5_pos: 100.000 loss: 0.6948 (0.7991) acc1: 55.9896 (49.5807) acc5: 100.0000 (100.0000) time: 0.2498 data: 0.1566 max mem: 3531 +Test: [ 150/2387] eta: 0:10:18 acc1_pos: 51.302 acc5_pos: 100.000 loss: 0.6711 (0.7878) acc1: 55.7292 (50.4639) acc5: 100.0000 (100.0000) time: 0.2304 data: 0.1361 max mem: 3531 +Test: [ 160/2387] eta: 0:10:06 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.6113 (0.7710) acc1: 67.1875 (51.9733) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.2115 data: 0.1179 max mem: 3531 +Test: [ 170/2387] eta: 0:09:49 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.3084 (0.7436) acc1: 90.6250 (54.3357) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1882 data: 0.0948 max mem: 3531 +Test: [ 180/2387] eta: 0:09:39 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2963 (0.7190) acc1: 92.4479 (56.4572) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1834 data: 0.0886 max mem: 3531 +Test: [ 190/2387] eta: 0:09:29 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2961 (0.6969) acc1: 92.7083 (58.3579) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1998 data: 0.1036 max mem: 3531 +Test: [ 200/2387] eta: 0:09:16 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2957 (0.6770) acc1: 92.7083 (60.0591) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1848 data: 0.0911 max mem: 3531 +Test: [ 210/2387] eta: 0:09:07 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.3001 (0.6594) acc1: 92.7083 (61.5966) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1799 data: 0.0860 max mem: 3532 +Test: [ 220/2387] eta: 0:08:59 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.3050 (0.6433) acc1: 91.9271 (62.9690) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1899 data: 0.0960 max mem: 3532 +Test: [ 230/2387] eta: 0:08:52 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.3050 (0.6285) acc1: 91.6667 (64.2440) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1949 data: 0.1008 max mem: 3532 +Test: [ 240/2387] eta: 0:08:42 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.3143 (0.6181) acc1: 90.3646 (65.1744) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.1849 data: 0.0900 max mem: 3532 +Test: [ 250/2387] eta: 0:08:34 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2864 (0.5989) acc1: 88.8021 (66.4986) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0795 max mem: 3532 +Test: [ 260/2387] eta: 0:08:28 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1001 (0.5797) acc1: 100.0000 (67.7822) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0907 max mem: 3532 +Test: [ 270/2387] eta: 0:08:24 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.0943 (0.5618) acc1: 100.0000 (68.9681) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2049 data: 0.1120 max mem: 3532 +Test: [ 280/2387] eta: 0:08:21 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.0943 (0.5454) acc1: 100.0000 (70.0706) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2249 data: 0.1299 max mem: 3532 +Test: [ 290/2387] eta: 0:08:16 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.0965 (0.5301) acc1: 100.0000 (71.0982) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2149 data: 0.1207 max mem: 3532 +Test: [ 300/2387] eta: 0:08:10 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.0970 (0.5159) acc1: 100.0000 (72.0584) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.1007 max mem: 3532 +Test: [ 310/2387] eta: 0:08:08 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1143 (0.5062) acc1: 100.0000 (72.8631) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2149 data: 0.1166 max mem: 3532 +Test: [ 320/2387] eta: 0:08:06 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.3255 (0.5009) acc1: 94.7917 (73.5397) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2349 data: 0.1361 max mem: 3532 +Test: [ 330/2387] eta: 0:08:02 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.3296 (0.4958) acc1: 94.7917 (74.1936) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2249 data: 0.1272 max mem: 3532 +Test: [ 340/2387] eta: 0:08:00 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.3296 (0.4910) acc1: 94.7917 (74.7969) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2299 data: 0.1326 max mem: 3532 +Test: [ 350/2387] eta: 0:07:56 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.3194 (0.4823) acc1: 95.5729 (75.4956) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2222 data: 0.1285 max mem: 3532 +Test: [ 360/2387] eta: 0:07:52 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2158 (0.4752) acc1: 98.1771 (76.0972) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2074 data: 0.1157 max mem: 3532 +Test: [ 370/2387] eta: 0:07:49 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2286 (0.4686) acc1: 96.8750 (76.6552) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2108 data: 0.1192 max mem: 3532 +Test: [ 380/2387] eta: 0:07:47 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2283 (0.4624) acc1: 96.8750 (77.1838) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2234 data: 0.1315 max mem: 3532 +Test: [ 390/2387] eta: 0:07:43 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2294 (0.4565) acc1: 96.6146 (77.6788) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2245 data: 0.1324 max mem: 3532 +Test: [ 400/2387] eta: 0:07:42 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2316 (0.4509) acc1: 96.6146 (78.1458) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2315 data: 0.1405 max mem: 3532 +Test: [ 410/2387] eta: 0:07:38 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2281 (0.4455) acc1: 96.6146 (78.5964) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2272 data: 0.1362 max mem: 3532 +Test: [ 420/2387] eta: 0:07:35 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2309 (0.4405) acc1: 96.6146 (79.0219) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2057 data: 0.1142 max mem: 3532 +Test: [ 430/2387] eta: 0:07:31 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2361 (0.4368) acc1: 96.0938 (79.3636) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.2021 data: 0.1109 max mem: 3532 +Test: [ 440/2387] eta: 0:07:28 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2467 (0.4311) acc1: 94.2708 (79.7719) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2044 data: 0.1137 max mem: 3533 +Test: [ 450/2387] eta: 0:07:24 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1048 (0.4239) acc1: 100.0000 (80.2205) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1989 data: 0.1069 max mem: 3533 +Test: [ 460/2387] eta: 0:07:18 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1044 (0.4169) acc1: 100.0000 (80.6495) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1733 data: 0.0807 max mem: 3533 +Test: [ 470/2387] eta: 0:07:13 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1042 (0.4103) acc1: 100.0000 (81.0604) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1614 data: 0.0699 max mem: 3533 +Test: [ 480/2387] eta: 0:07:09 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1042 (0.4040) acc1: 100.0000 (81.4541) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1647 data: 0.0738 max mem: 3533 +Test: [ 490/2387] eta: 0:07:05 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1054 (0.3979) acc1: 100.0000 (81.8318) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1752 data: 0.0820 max mem: 3533 +Test: [ 500/2387] eta: 0:07:01 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1046 (0.3920) acc1: 100.0000 (82.1939) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1789 data: 0.0831 max mem: 3533 +Test: [ 510/2387] eta: 0:06:57 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1062 (0.3868) acc1: 100.0000 (82.5419) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1788 data: 0.0856 max mem: 3533 +Test: [ 520/2387] eta: 0:06:53 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1442 (0.3824) acc1: 100.0000 (82.8755) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1808 data: 0.0894 max mem: 3533 +Test: [ 530/2387] eta: 0:06:50 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1442 (0.3779) acc1: 100.0000 (83.1980) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1885 data: 0.0937 max mem: 3533 +Test: [ 540/2387] eta: 0:06:47 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1433 (0.3736) acc1: 100.0000 (83.5081) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1940 data: 0.0981 max mem: 3533 +Test: [ 550/2387] eta: 0:06:44 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1421 (0.3694) acc1: 100.0000 (83.8074) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1898 data: 0.0962 max mem: 3533 +Test: [ 560/2387] eta: 0:06:41 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1421 (0.3654) acc1: 100.0000 (84.0960) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1920 data: 0.0977 max mem: 3533 +Test: [ 570/2387] eta: 0:06:37 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1413 (0.3615) acc1: 100.0000 (84.3745) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1875 data: 0.0946 max mem: 3533 +Test: [ 580/2387] eta: 0:06:35 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1422 (0.3577) acc1: 100.0000 (84.6435) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1996 data: 0.1092 max mem: 3533 +Test: [ 590/2387] eta: 0:06:32 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1431 (0.3541) acc1: 100.0000 (84.9033) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2033 data: 0.1118 max mem: 3533 +Test: [ 600/2387] eta: 0:06:28 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1498 (0.3510) acc1: 100.0000 (85.1493) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1752 data: 0.0837 max mem: 3533 +Test: [ 610/2387] eta: 0:06:25 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2100 (0.3500) acc1: 96.8750 (85.3182) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1684 data: 0.0774 max mem: 3533 +Test: [ 620/2387] eta: 0:06:22 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2873 (0.3491) acc1: 95.3125 (85.4825) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1810 data: 0.0886 max mem: 3533 +Test: [ 630/2387] eta: 0:06:18 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2881 (0.3482) acc1: 95.3125 (85.6375) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1777 data: 0.0843 max mem: 3533 +Test: [ 640/2387] eta: 0:06:15 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2940 (0.3474) acc1: 95.3125 (85.7860) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1744 data: 0.0793 max mem: 3533 +Test: [ 650/2387] eta: 0:06:12 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2939 (0.3466) acc1: 95.5729 (85.9399) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1752 data: 0.0814 max mem: 3533 +Test: [ 660/2387] eta: 0:06:09 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2940 (0.3458) acc1: 95.5729 (86.0872) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1797 data: 0.0888 max mem: 3533 +Test: [ 670/2387] eta: 0:06:06 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2930 (0.3450) acc1: 95.3125 (86.2266) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1882 data: 0.0971 max mem: 3534 +Test: [ 680/2387] eta: 0:06:03 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.3003 (0.3447) acc1: 95.0521 (86.3513) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1774 data: 0.0851 max mem: 3534 +Test: [ 690/2387] eta: 0:06:00 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.3455 (0.3471) acc1: 93.2292 (86.2070) acc5: 100.0000 (100.0000) acc1_neg: 68.490 acc5_neg: 100.000 time: 0.1736 data: 0.0804 max mem: 3534 +Test: [ 700/2387] eta: 0:05:57 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.5614 (0.3503) acc1: 70.8333 (85.9810) acc5: 100.0000 (100.0000) acc1_neg: 72.656 acc5_neg: 100.000 time: 0.1772 data: 0.0841 max mem: 3534 +Test: [ 710/2387] eta: 0:05:54 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.5682 (0.3535) acc1: 69.5312 (85.7511) acc5: 100.0000 (100.0000) acc1_neg: 69.010 acc5_neg: 100.000 time: 0.1768 data: 0.0848 max mem: 3534 +Test: [ 720/2387] eta: 0:05:52 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.5686 (0.3565) acc1: 70.3125 (85.5481) acc5: 100.0000 (100.0000) acc1_neg: 71.094 acc5_neg: 100.000 time: 0.1889 data: 0.0980 max mem: 3534 +Test: [ 730/2387] eta: 0:05:49 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.5751 (0.3596) acc1: 70.3125 (85.3394) acc5: 100.0000 (100.0000) acc1_neg: 69.792 acc5_neg: 100.000 time: 0.1946 data: 0.1037 max mem: 3534 +Test: [ 740/2387] eta: 0:05:47 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.5751 (0.3624) acc1: 70.0521 (85.1373) acc5: 100.0000 (100.0000) acc1_neg: 67.969 acc5_neg: 100.000 time: 0.1962 data: 0.1028 max mem: 3534 +Test: [ 750/2387] eta: 0:05:44 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.5611 (0.3651) acc1: 71.0938 (84.9607) acc5: 100.0000 (100.0000) acc1_neg: 74.219 acc5_neg: 100.000 time: 0.2032 data: 0.1066 max mem: 3534 +Test: [ 760/2387] eta: 0:05:42 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.5515 (0.3672) acc1: 71.8750 (84.8127) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.1992 data: 0.1046 max mem: 3534 +Test: [ 770/2387] eta: 0:05:40 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.4242 (0.3667) acc1: 79.9479 (84.8475) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1960 data: 0.1039 max mem: 3534 +Test: [ 780/2387] eta: 0:05:37 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2605 (0.3652) acc1: 92.4479 (84.9675) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2043 data: 0.1123 max mem: 3534 +Test: [ 790/2387] eta: 0:05:35 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2256 (0.3633) acc1: 95.3125 (85.0874) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.2109 data: 0.1179 max mem: 3534 +Test: [ 800/2387] eta: 0:05:33 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1991 (0.3615) acc1: 95.8333 (85.2203) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.2090 data: 0.1162 max mem: 3534 +Test: [ 810/2387] eta: 0:05:31 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2230 (0.3599) acc1: 95.3125 (85.3412) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2028 data: 0.1110 max mem: 3534 +Test: [ 820/2387] eta: 0:05:28 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2337 (0.3584) acc1: 94.7917 (85.4490) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1845 data: 0.0917 max mem: 3534 +Test: [ 830/2387] eta: 0:05:26 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.3110 (0.3587) acc1: 89.0625 (85.4320) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1911 data: 0.0988 max mem: 3534 +Test: [ 840/2387] eta: 0:05:23 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.3934 (0.3588) acc1: 85.1562 (85.4696) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.1899 data: 0.0971 max mem: 3534 +Test: [ 850/2387] eta: 0:05:20 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.3535 (0.3588) acc1: 86.7188 (85.5094) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1713 data: 0.0772 max mem: 3534 +Test: [ 860/2387] eta: 0:05:18 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.3217 (0.3584) acc1: 90.6250 (85.5906) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1806 data: 0.0878 max mem: 3534 +Test: [ 870/2387] eta: 0:05:15 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.3167 (0.3581) acc1: 91.1458 (85.6496) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1846 data: 0.0901 max mem: 3534 +Test: [ 880/2387] eta: 0:05:13 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.3517 (0.3579) acc1: 90.8854 (85.6954) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1892 data: 0.0944 max mem: 3534 +Test: [ 890/2387] eta: 0:05:10 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.3461 (0.3577) acc1: 89.8438 (85.7560) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1825 data: 0.0900 max mem: 3534 +Test: [ 900/2387] eta: 0:05:08 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.3899 (0.3617) acc1: 82.2917 (85.4502) acc5: 100.0000 (100.0000) acc1_neg: 53.646 acc5_neg: 100.000 time: 0.1849 data: 0.0926 max mem: 3535 +Test: [ 910/2387] eta: 0:05:06 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.8081 (0.3668) acc1: 48.6979 (85.0408) acc5: 100.0000 (100.0000) acc1_neg: 47.917 acc5_neg: 100.000 time: 0.1868 data: 0.0946 max mem: 3535 +Test: [ 920/2387] eta: 0:05:03 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.8232 (0.3717) acc1: 48.4375 (84.6456) acc5: 100.0000 (100.0000) acc1_neg: 45.833 acc5_neg: 100.000 time: 0.1902 data: 0.0974 max mem: 3535 +Test: [ 930/2387] eta: 0:05:01 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.8199 (0.3764) acc1: 49.4792 (84.2768) acc5: 100.0000 (100.0000) acc1_neg: 44.531 acc5_neg: 100.000 time: 0.1830 data: 0.0897 max mem: 3535 +Test: [ 940/2387] eta: 0:04:58 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.8171 (0.3811) acc1: 49.4792 (83.9112) acc5: 100.0000 (100.0000) acc1_neg: 47.135 acc5_neg: 100.000 time: 0.1710 data: 0.0785 max mem: 3535 +Test: [ 950/2387] eta: 0:04:56 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.8171 (0.3856) acc1: 49.2188 (83.5436) acc5: 100.0000 (100.0000) acc1_neg: 49.219 acc5_neg: 100.000 time: 0.1870 data: 0.0954 max mem: 3535 +Test: [ 960/2387] eta: 0:04:54 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.8047 (0.3900) acc1: 49.2188 (83.1875) acc5: 100.0000 (100.0000) acc1_neg: 52.083 acc5_neg: 100.000 time: 0.1884 data: 0.0948 max mem: 3535 +Test: [ 970/2387] eta: 0:04:51 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.8084 (0.3944) acc1: 48.9583 (82.8264) acc5: 100.0000 (100.0000) acc1_neg: 46.615 acc5_neg: 100.000 time: 0.1892 data: 0.0956 max mem: 3535 +Test: [ 980/2387] eta: 0:04:49 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.7983 (0.3956) acc1: 50.7812 (82.7416) acc5: 100.0000 (100.0000) acc1_neg: 68.750 acc5_neg: 100.000 time: 0.1874 data: 0.0960 max mem: 3535 +Test: [ 990/2387] eta: 0:04:47 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.3008 (0.3943) acc1: 90.1042 (82.8298) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1885 data: 0.0978 max mem: 3535 +Test: [1000/2387] eta: 0:04:44 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2346 (0.3926) acc1: 95.0521 (82.9478) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1929 data: 0.1012 max mem: 3535 +Test: [1010/2387] eta: 0:04:42 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2393 (0.3916) acc1: 92.9688 (83.0222) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1898 data: 0.0979 max mem: 3535 +Test: [1020/2387] eta: 0:04:40 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2748 (0.3908) acc1: 90.1042 (83.0913) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1994 data: 0.1055 max mem: 3535 +Test: [1030/2387] eta: 0:04:39 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2316 (0.3893) acc1: 95.0521 (83.2123) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2249 data: 0.1287 max mem: 3535 +Test: [1040/2387] eta: 0:04:36 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1940 (0.3873) acc1: 97.3958 (83.3541) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1978 data: 0.1018 max mem: 3535 +Test: [1050/2387] eta: 0:04:34 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1832 (0.3854) acc1: 98.4375 (83.4936) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1799 data: 0.0863 max mem: 3535 +Test: [1060/2387] eta: 0:04:31 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1893 (0.3838) acc1: 97.9167 (83.6168) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1899 data: 0.0982 max mem: 3535 +Test: [1070/2387] eta: 0:04:29 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2151 (0.3825) acc1: 96.0938 (83.7178) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.1949 data: 0.1029 max mem: 3535 +Test: [1080/2387] eta: 0:04:27 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2747 (0.3820) acc1: 91.4062 (83.7677) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1970 data: 0.1025 max mem: 3535 +Test: [1090/2387] eta: 0:04:25 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.3024 (0.3810) acc1: 90.1042 (83.8458) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.1749 data: 0.0802 max mem: 3535 +Test: [1100/2387] eta: 0:04:23 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2599 (0.3801) acc1: 91.6667 (83.9050) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1999 data: 0.1074 max mem: 3535 +Test: [1110/2387] eta: 0:04:21 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2599 (0.3793) acc1: 91.6667 (83.9657) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2249 data: 0.1302 max mem: 3535 +Test: [1120/2387] eta: 0:04:19 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2760 (0.3786) acc1: 90.3646 (84.0182) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.2064 data: 0.1074 max mem: 3536 +Test: [1130/2387] eta: 0:04:16 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2421 (0.3775) acc1: 92.7083 (84.0971) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1522 data: 0.0543 max mem: 3536 +Test: [1140/2387] eta: 0:04:13 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2708 (0.3773) acc1: 91.4062 (84.1468) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1451 data: 0.0516 max mem: 3536 +Test: [1150/2387] eta: 0:04:12 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.3532 (0.3769) acc1: 86.9792 (84.1997) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1976 data: 0.1041 max mem: 3536 +Test: [1160/2387] eta: 0:04:09 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2768 (0.3758) acc1: 91.9271 (84.2777) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1938 data: 0.1005 max mem: 3536 +Test: [1170/2387] eta: 0:04:07 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2394 (0.3748) acc1: 95.0521 (84.3663) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1662 data: 0.0736 max mem: 3536 +Test: [1180/2387] eta: 0:04:05 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2399 (0.3737) acc1: 94.7917 (84.4489) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1862 data: 0.0934 max mem: 3536 +Test: [1190/2387] eta: 0:04:02 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2313 (0.3728) acc1: 94.7917 (84.5329) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1812 data: 0.0895 max mem: 3536 +Test: [1200/2387] eta: 0:04:00 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2295 (0.3716) acc1: 95.8333 (84.6346) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1807 data: 0.0897 max mem: 3536 +Test: [1210/2387] eta: 0:03:59 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2295 (0.3705) acc1: 96.8750 (84.7238) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2290 data: 0.1370 max mem: 3536 +Test: [1220/2387] eta: 0:03:57 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1305 (0.3685) acc1: 98.6979 (84.8432) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2291 data: 0.1365 max mem: 3536 +Test: [1230/2387] eta: 0:03:55 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1139 (0.3664) acc1: 99.7396 (84.9637) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2181 data: 0.1251 max mem: 3536 +Test: [1240/2387] eta: 0:03:53 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1002 (0.3643) acc1: 99.7396 (85.0836) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2062 data: 0.1132 max mem: 3536 +Test: [1250/2387] eta: 0:03:51 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1025 (0.3622) acc1: 99.7396 (85.2008) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1924 data: 0.1001 max mem: 3536 +Test: [1260/2387] eta: 0:03:48 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1107 (0.3602) acc1: 99.7396 (85.3144) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1953 data: 0.1025 max mem: 3536 +Test: [1270/2387] eta: 0:03:47 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1113 (0.3584) acc1: 99.7396 (85.4275) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2089 data: 0.1161 max mem: 3536 +Test: [1280/2387] eta: 0:03:45 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1247 (0.3566) acc1: 99.7396 (85.5376) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2134 data: 0.1216 max mem: 3536 +Test: [1290/2387] eta: 0:03:42 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1427 (0.3550) acc1: 99.2188 (85.6345) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1999 data: 0.1086 max mem: 3536 +Test: [1300/2387] eta: 0:03:40 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1427 (0.3534) acc1: 98.4375 (85.7313) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1984 data: 0.1063 max mem: 3536 +Test: [1310/2387] eta: 0:03:38 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1462 (0.3519) acc1: 98.6979 (85.8249) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1892 data: 0.0961 max mem: 3536 +Test: [1320/2387] eta: 0:03:36 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1524 (0.3505) acc1: 98.6979 (85.9170) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1903 data: 0.0971 max mem: 3536 +Test: [1330/2387] eta: 0:03:34 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1708 (0.3496) acc1: 96.8750 (85.9864) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1989 data: 0.1065 max mem: 3536 +Test: [1340/2387] eta: 0:03:32 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2491 (0.3490) acc1: 93.2292 (86.0296) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1951 data: 0.1028 max mem: 3536 +Test: [1350/2387] eta: 0:03:30 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.0804 (0.3470) acc1: 100.0000 (86.1330) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1922 data: 0.1004 max mem: 3537 +Test: [1360/2387] eta: 0:03:28 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.0697 (0.3449) acc1: 100.0000 (86.2348) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1959 data: 0.1044 max mem: 3537 +Test: [1370/2387] eta: 0:03:26 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.0655 (0.3429) acc1: 100.0000 (86.3352) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2053 data: 0.1130 max mem: 3537 +Test: [1380/2387] eta: 0:03:24 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.0666 (0.3409) acc1: 100.0000 (86.4342) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1984 data: 0.1064 max mem: 3537 +Test: [1390/2387] eta: 0:03:22 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.0682 (0.3389) acc1: 100.0000 (86.5317) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1896 data: 0.0967 max mem: 3537 +Test: [1400/2387] eta: 0:03:19 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.0682 (0.3370) acc1: 100.0000 (86.6279) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1893 data: 0.0962 max mem: 3537 +Test: [1410/2387] eta: 0:03:17 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.0752 (0.3362) acc1: 100.0000 (86.6813) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1884 data: 0.0972 max mem: 3537 +Test: [1420/2387] eta: 0:03:15 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2336 (0.3356) acc1: 95.5729 (86.7292) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1858 data: 0.0943 max mem: 3537 +Test: [1430/2387] eta: 0:03:13 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2628 (0.3352) acc1: 93.2292 (86.7659) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1906 data: 0.0973 max mem: 3537 +Test: [1440/2387] eta: 0:03:11 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2767 (0.3347) acc1: 93.2292 (86.8178) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.2006 data: 0.1081 max mem: 3537 +Test: [1450/2387] eta: 0:03:09 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2811 (0.3345) acc1: 93.4896 (86.8480) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1982 data: 0.1054 max mem: 3537 +Test: [1460/2387] eta: 0:03:07 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2021 (0.3336) acc1: 97.3958 (86.9180) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2032 data: 0.1100 max mem: 3537 +Test: [1470/2387] eta: 0:03:05 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1776 (0.3325) acc1: 97.9167 (86.9928) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1999 data: 0.1072 max mem: 3537 +Test: [1480/2387] eta: 0:03:03 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1662 (0.3315) acc1: 98.1771 (87.0634) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1812 data: 0.0882 max mem: 3537 +Test: [1490/2387] eta: 0:03:01 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1705 (0.3304) acc1: 98.1771 (87.1367) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1803 data: 0.0892 max mem: 3537 +Test: [1500/2387] eta: 0:02:58 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1769 (0.3294) acc1: 98.1771 (87.2077) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1842 data: 0.0926 max mem: 3537 +Test: [1510/2387] eta: 0:02:56 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1967 (0.3288) acc1: 97.9167 (87.2716) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1779 data: 0.0858 max mem: 3537 +Test: [1520/2387] eta: 0:02:54 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2566 (0.3287) acc1: 94.7917 (87.3046) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1803 data: 0.0892 max mem: 3537 +Test: [1530/2387] eta: 0:02:52 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.3253 (0.3287) acc1: 90.1042 (87.3192) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1843 data: 0.0927 max mem: 3537 +Test: [1540/2387] eta: 0:02:50 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2965 (0.3285) acc1: 92.9688 (87.3636) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1875 data: 0.0952 max mem: 3537 +Test: [1550/2387] eta: 0:02:48 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2824 (0.3282) acc1: 94.2708 (87.4070) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1930 data: 0.1010 max mem: 3537 +Test: [1560/2387] eta: 0:02:46 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2824 (0.3280) acc1: 94.2708 (87.4436) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.1810 data: 0.0880 max mem: 3537 +Test: [1570/2387] eta: 0:02:44 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2643 (0.3277) acc1: 94.7917 (87.4829) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1709 data: 0.0783 max mem: 3537 +Test: [1580/2387] eta: 0:02:42 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2678 (0.3274) acc1: 94.7917 (87.5137) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1864 data: 0.0954 max mem: 3538 +Test: [1590/2387] eta: 0:02:39 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2361 (0.3269) acc1: 94.5312 (87.5552) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1843 data: 0.0932 max mem: 3538 +Test: [1600/2387] eta: 0:02:37 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2170 (0.3262) acc1: 94.7917 (87.5958) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1813 data: 0.0897 max mem: 3538 +Test: [1610/2387] eta: 0:02:35 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2097 (0.3256) acc1: 95.3125 (87.6471) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1836 data: 0.0919 max mem: 3538 +Test: [1620/2387] eta: 0:02:33 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2347 (0.3251) acc1: 94.0104 (87.6798) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1740 data: 0.0824 max mem: 3538 +Test: [1630/2387] eta: 0:02:31 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2643 (0.3249) acc1: 91.9271 (87.6922) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1823 data: 0.0904 max mem: 3538 +Test: [1640/2387] eta: 0:02:29 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.3429 (0.3251) acc1: 86.9792 (87.6858) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1898 data: 0.0978 max mem: 3538 +Test: [1650/2387] eta: 0:02:27 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.3584 (0.3252) acc1: 85.9375 (87.6893) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1875 data: 0.0950 max mem: 3538 +Test: [1660/2387] eta: 0:02:25 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.3151 (0.3252) acc1: 89.5833 (87.6896) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1852 data: 0.0919 max mem: 3538 +Test: [1670/2387] eta: 0:02:23 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.3082 (0.3251) acc1: 89.5833 (87.7063) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1862 data: 0.0938 max mem: 3538 +Test: [1680/2387] eta: 0:02:21 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.3008 (0.3251) acc1: 91.1458 (87.7181) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.2079 data: 0.1169 max mem: 3538 +Test: [1690/2387] eta: 0:02:19 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.3318 (0.3251) acc1: 88.2812 (87.7179) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.1900 data: 0.0978 max mem: 3538 +Test: [1700/2387] eta: 0:02:17 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.3267 (0.3249) acc1: 88.0208 (87.7405) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1758 data: 0.0837 max mem: 3538 +Test: [1710/2387] eta: 0:02:15 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1589 (0.3239) acc1: 97.9167 (87.8043) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1887 data: 0.0973 max mem: 3538 +Test: [1720/2387] eta: 0:02:13 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1401 (0.3228) acc1: 98.6979 (87.8662) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1841 data: 0.0916 max mem: 3538 +Test: [1730/2387] eta: 0:02:11 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1310 (0.3217) acc1: 99.2188 (87.9292) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2012 data: 0.1086 max mem: 3538 +Test: [1740/2387] eta: 0:02:09 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1280 (0.3207) acc1: 99.2188 (87.9917) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1994 data: 0.1066 max mem: 3538 +Test: [1750/2387] eta: 0:02:07 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1367 (0.3196) acc1: 98.9583 (88.0549) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1952 data: 0.1035 max mem: 3538 +Test: [1760/2387] eta: 0:02:05 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1367 (0.3188) acc1: 98.4375 (88.1059) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2056 data: 0.1138 max mem: 3538 +Test: [1770/2387] eta: 0:02:03 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2045 (0.3183) acc1: 95.8333 (88.1451) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.2049 data: 0.1126 max mem: 3538 +Test: [1780/2387] eta: 0:02:01 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2202 (0.3177) acc1: 95.5729 (88.1843) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2042 data: 0.1129 max mem: 3538 +Test: [1790/2387] eta: 0:01:59 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2030 (0.3170) acc1: 96.6146 (88.2370) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2040 data: 0.1098 max mem: 3538 +Test: [1800/2387] eta: 0:01:57 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1997 (0.3163) acc1: 97.3958 (88.2846) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2049 data: 0.1088 max mem: 3538 +Test: [1810/2387] eta: 0:01:55 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1969 (0.3156) acc1: 97.1354 (88.3362) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2057 data: 0.1121 max mem: 3539 +Test: [1820/2387] eta: 0:01:53 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1695 (0.3148) acc1: 98.1771 (88.3881) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1964 data: 0.1036 max mem: 3539 +Test: [1830/2387] eta: 0:01:51 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1695 (0.3141) acc1: 97.9167 (88.4377) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1992 data: 0.1063 max mem: 3539 +Test: [1840/2387] eta: 0:01:49 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1377 (0.3131) acc1: 100.0000 (88.5005) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2110 data: 0.1191 max mem: 3539 +Test: [1850/2387] eta: 0:01:47 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1369 (0.3122) acc1: 100.0000 (88.5622) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1964 data: 0.1048 max mem: 3539 +Test: [1860/2387] eta: 0:01:45 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1369 (0.3112) acc1: 100.0000 (88.6237) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1895 data: 0.0980 max mem: 3539 +Test: [1870/2387] eta: 0:01:43 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1369 (0.3103) acc1: 100.0000 (88.6843) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2009 data: 0.1090 max mem: 3539 +Test: [1880/2387] eta: 0:01:41 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1373 (0.3094) acc1: 100.0000 (88.7444) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1976 data: 0.1022 max mem: 3539 +Test: [1890/2387] eta: 0:01:39 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1373 (0.3085) acc1: 100.0000 (88.8037) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1979 data: 0.1032 max mem: 3539 +Test: [1900/2387] eta: 0:01:37 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1378 (0.3076) acc1: 100.0000 (88.8625) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2012 data: 0.1089 max mem: 3539 +Test: [1910/2387] eta: 0:01:35 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1570 (0.3069) acc1: 100.0000 (88.9171) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1979 data: 0.1049 max mem: 3539 +Test: [1920/2387] eta: 0:01:33 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1980 (0.3072) acc1: 96.6146 (88.9020) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2010 data: 0.1090 max mem: 3539 +Test: [1930/2387] eta: 0:01:31 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2409 (0.3068) acc1: 95.3125 (88.9310) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2162 data: 0.1221 max mem: 3539 +Test: [1940/2387] eta: 0:01:29 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2226 (0.3066) acc1: 96.3542 (88.9354) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2190 data: 0.1225 max mem: 3539 +Test: [1950/2387] eta: 0:01:27 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2719 (0.3066) acc1: 92.9688 (88.9441) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.1986 data: 0.1041 max mem: 3539 +Test: [1960/2387] eta: 0:01:25 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2942 (0.3066) acc1: 91.4062 (88.9515) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.1908 data: 0.0965 max mem: 3539 +Test: [1970/2387] eta: 0:01:23 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.3168 (0.3069) acc1: 89.8438 (88.9480) acc5: 100.0000 (100.0000) acc1_neg: 78.646 acc5_neg: 100.000 time: 0.1954 data: 0.1013 max mem: 3539 +Test: [1980/2387] eta: 0:01:21 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.3255 (0.3068) acc1: 89.0625 (88.9501) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1924 data: 0.0994 max mem: 3539 +Test: [1990/2387] eta: 0:01:19 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1455 (0.3060) acc1: 99.2188 (89.0019) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1967 data: 0.1046 max mem: 3539 +Test: [2000/2387] eta: 0:01:17 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1450 (0.3052) acc1: 99.2188 (89.0535) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2002 data: 0.1094 max mem: 3539 +Test: [2010/2387] eta: 0:01:15 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1450 (0.3044) acc1: 99.2188 (89.1032) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1855 data: 0.0925 max mem: 3539 +Test: [2020/2387] eta: 0:01:13 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1475 (0.3036) acc1: 99.2188 (89.1530) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1796 data: 0.0857 max mem: 3539 +Test: [2030/2387] eta: 0:01:11 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1469 (0.3029) acc1: 99.2188 (89.2017) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1770 data: 0.0841 max mem: 3540 +Test: [2040/2387] eta: 0:01:09 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1455 (0.3021) acc1: 99.2188 (89.2515) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1735 data: 0.0808 max mem: 3540 +Test: [2050/2387] eta: 0:01:07 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1455 (0.3014) acc1: 99.2188 (89.3008) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1723 data: 0.0792 max mem: 3540 +Test: [2060/2387] eta: 0:01:05 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1672 (0.3007) acc1: 99.2188 (89.3462) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1635 data: 0.0695 max mem: 3540 +Test: [2070/2387] eta: 0:01:03 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1693 (0.3001) acc1: 98.9583 (89.3933) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1814 data: 0.0880 max mem: 3540 +Test: [2080/2387] eta: 0:01:01 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1688 (0.2995) acc1: 98.9583 (89.4390) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1979 data: 0.1053 max mem: 3540 +Test: [2090/2387] eta: 0:00:59 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1690 (0.2988) acc1: 98.9583 (89.4846) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1849 data: 0.0914 max mem: 3540 +Test: [2100/2387] eta: 0:00:57 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1694 (0.2982) acc1: 98.9583 (89.5300) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1638 data: 0.0710 max mem: 3540 +Test: [2110/2387] eta: 0:00:54 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1700 (0.2976) acc1: 98.9583 (89.5748) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1624 data: 0.0717 max mem: 3540 +Test: [2120/2387] eta: 0:00:52 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1748 (0.2974) acc1: 98.6979 (89.5950) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1626 data: 0.0711 max mem: 3540 +Test: [2130/2387] eta: 0:00:50 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1956 (0.2970) acc1: 97.1354 (89.6265) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1543 data: 0.0626 max mem: 3540 +Test: [2140/2387] eta: 0:00:48 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1993 (0.2966) acc1: 96.8750 (89.6525) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1594 data: 0.0670 max mem: 3540 +Test: [2150/2387] eta: 0:00:46 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2039 (0.2962) acc1: 96.3542 (89.6837) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1608 data: 0.0676 max mem: 3540 +Test: [2160/2387] eta: 0:00:44 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2001 (0.2959) acc1: 96.6146 (89.7137) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1598 data: 0.0677 max mem: 3540 +Test: [2170/2387] eta: 0:00:42 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2549 (0.2959) acc1: 94.5312 (89.7201) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.1584 data: 0.0657 max mem: 3540 +Test: [2180/2387] eta: 0:00:40 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.3295 (0.2973) acc1: 87.7604 (89.6187) acc5: 100.0000 (100.0000) acc1_neg: 39.323 acc5_neg: 100.000 time: 0.1589 data: 0.0651 max mem: 3540 +Test: [2190/2387] eta: 0:00:38 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.3151 (0.2974) acc1: 89.5833 (89.6150) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1562 data: 0.0623 max mem: 3540 +Test: [2200/2387] eta: 0:00:36 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2028 (0.2970) acc1: 97.6562 (89.6539) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1658 data: 0.0733 max mem: 3540 +Test: [2210/2387] eta: 0:00:34 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1992 (0.2966) acc1: 98.1771 (89.6899) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1724 data: 0.0793 max mem: 3540 +Test: [2220/2387] eta: 0:00:32 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1992 (0.2962) acc1: 98.1771 (89.7212) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1595 data: 0.0646 max mem: 3540 +Test: [2230/2387] eta: 0:00:30 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1817 (0.2957) acc1: 97.6562 (89.7546) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1553 data: 0.0623 max mem: 3540 +Test: [2240/2387] eta: 0:00:28 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1959 (0.2953) acc1: 97.6562 (89.7830) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1626 data: 0.0692 max mem: 3540 +Test: [2250/2387] eta: 0:00:26 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2389 (0.2952) acc1: 94.5312 (89.8005) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1689 data: 0.0713 max mem: 3540 +Test: [2260/2387] eta: 0:00:24 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2389 (0.2949) acc1: 95.8333 (89.8319) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1817 data: 0.0851 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1745 (0.2943) acc1: 98.9583 (89.8710) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1865 data: 0.0930 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1624 (0.2938) acc1: 99.4792 (89.9004) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1652 data: 0.0712 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1476 (0.2932) acc1: 99.7396 (89.9420) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1502 data: 0.0577 max mem: 3541 +Test: [2300/2387] eta: 0:00:17 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1668 (0.2927) acc1: 99.2188 (89.9776) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1592 data: 0.0691 max mem: 3541 +Test: [2310/2387] eta: 0:00:15 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.1942 (0.2924) acc1: 97.9167 (90.0047) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1610 data: 0.0706 max mem: 3541 +Test: [2320/2387] eta: 0:00:13 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2449 (0.2924) acc1: 95.0521 (90.0163) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1505 data: 0.0600 max mem: 3541 +Test: [2330/2387] eta: 0:00:11 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2900 (0.2926) acc1: 92.4479 (90.0084) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.1374 data: 0.0474 max mem: 3541 +Test: [2340/2387] eta: 0:00:09 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2789 (0.2926) acc1: 92.4479 (90.0176) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1325 data: 0.0423 max mem: 3541 +Test: [2350/2387] eta: 0:00:07 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2647 (0.2926) acc1: 92.4479 (90.0192) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1351 data: 0.0442 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2827 (0.2929) acc1: 92.1875 (90.0141) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1295 data: 0.0389 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.2899 (0.2930) acc1: 89.8438 (90.0075) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1251 data: 0.0352 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.3552 (0.2935) acc1: 83.5938 (89.9748) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.1210 data: 0.0313 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 63.927 acc5_pos: 100.000 loss: 0.4333 (0.2935) acc1: 80.4688 (89.9731) acc5: 100.0000 (100.0000) acc1_neg: 95.349 acc5_neg: 100.000 time: 0.1169 data: 0.0271 max mem: 3541 +Test: Total time: 0:07:41 (0.1935 s / it) +* Overall Acc@1 89.973 Acc@5 100.000 + target == real Acc@1 51.020 Acc@5 100.000 + target == fake Acc@1 92.709 Acc@5 100.000 + loss 0.293 + PR-AUC 0.9905 ROC-AUC 0.8837 EER 19.42% (thr=0.6914) +Accuracy of the network on the 916267 test images: 90.0% +[rank0]:[W726 04:38:42.309967266 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:39:03.361000 1186865 torch/distributed/run.py:766] +W0726 04:39:03.361000 1186865 torch/distributed/run.py:766] ***************************************** +W0726 04:39:03.361000 1186865 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:39:03.361000 1186865 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:39:31.341299812 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:39:31.358407609 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:39:32.783169871 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:39:32.806751652 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W726 04:39:32.812602641 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:39:32.887315014 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:39:32.933448035 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:39:32.935396267 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n23_seed7_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed7_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:00:50 acc1_pos: 38.802 acc5_pos: 100.000 loss: 0.8960 (0.8960) acc1: 38.8021 (38.8021) acc5: 100.0000 (100.0000) time: 7.0611 data: 5.5868 max mem: 3531 +Test: [ 10/517] eta: 0:06:08 acc1_pos: 36.198 acc5_pos: 100.000 loss: 0.8503 (0.7707) acc1: 43.7500 (50.9706) acc5: 100.0000 (100.0000) time: 0.7264 data: 0.5083 max mem: 3531 +Test: [ 20/517] eta: 0:03:32 acc1_pos: 46.094 acc5_pos: 100.000 loss: 0.7885 (0.8039) acc1: 46.0938 (49.5164) acc5: 100.0000 (100.0000) time: 0.0955 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:02:48 acc1_pos: 36.979 acc5_pos: 100.000 loss: 0.8267 (0.8054) acc1: 46.0938 (49.0591) acc5: 100.0000 (100.0000) time: 0.1359 data: 0.0408 max mem: 3531 +Test: [ 40/517] eta: 0:02:32 acc1_pos: 44.271 acc5_pos: 100.000 loss: 0.8329 (0.8193) acc1: 42.4479 (47.0910) acc5: 100.0000 (100.0000) time: 0.2053 data: 0.1138 max mem: 3531 +Test: [ 50/517] eta: 0:02:21 acc1_pos: 22.936 acc5_pos: 100.000 loss: 0.8321 (0.8228) acc1: 43.2292 (47.1201) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.2396 data: 0.1458 max mem: 3531 +Test: [ 60/517] eta: 0:02:15 acc1_pos: 22.936 acc5_pos: 100.000 loss: 0.4825 (0.7571) acc1: 75.7812 (53.4366) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.2496 data: 0.1545 max mem: 3531 +Test: [ 70/517] eta: 0:02:10 acc1_pos: 22.936 acc5_pos: 100.000 loss: 0.4505 (0.7255) acc1: 81.2500 (56.3270) acc5: 100.0000 (100.0000) acc1_neg: 64.062 acc5_neg: 100.000 time: 0.2634 data: 0.1685 max mem: 3531 +Test: [ 80/517] eta: 0:02:06 acc1_pos: 22.936 acc5_pos: 100.000 loss: 0.5516 (0.7085) acc1: 66.6667 (57.6743) acc5: 100.0000 (100.0000) acc1_neg: 66.406 acc5_neg: 100.000 time: 0.2699 data: 0.1711 max mem: 3531 +Test: [ 90/517] eta: 0:01:59 acc1_pos: 22.936 acc5_pos: 100.000 loss: 0.5585 (0.6909) acc1: 69.2708 (59.4351) acc5: 100.0000 (100.0000) acc1_neg: 71.354 acc5_neg: 100.000 time: 0.2399 data: 0.1437 max mem: 3531 +Test: [100/517] eta: 0:01:56 acc1_pos: 22.936 acc5_pos: 100.000 loss: 0.4870 (0.6685) acc1: 80.4688 (61.6801) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.2399 data: 0.1486 max mem: 3531 +Test: [110/517] eta: 0:01:52 acc1_pos: 22.936 acc5_pos: 100.000 loss: 0.3664 (0.6393) acc1: 88.5417 (64.3675) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.2499 data: 0.1576 max mem: 3531 +Test: [120/517] eta: 0:01:48 acc1_pos: 22.936 acc5_pos: 100.000 loss: 0.4002 (0.6265) acc1: 83.5938 (65.6272) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.2449 data: 0.1503 max mem: 3531 +Test: [130/517] eta: 0:01:44 acc1_pos: 22.936 acc5_pos: 100.000 loss: 0.4834 (0.6138) acc1: 79.6875 (66.9887) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2457 data: 0.1513 max mem: 3531 +Test: [140/517] eta: 0:01:41 acc1_pos: 22.936 acc5_pos: 100.000 loss: 0.4290 (0.5994) acc1: 85.9375 (68.5764) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.2405 data: 0.1476 max mem: 3531 +Test: [150/517] eta: 0:01:38 acc1_pos: 22.936 acc5_pos: 100.000 loss: 0.3935 (0.5841) acc1: 90.3646 (70.1297) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.2540 data: 0.1618 max mem: 3531 +Test: [160/517] eta: 0:01:35 acc1_pos: 22.936 acc5_pos: 100.000 loss: 0.3831 (0.5718) acc1: 90.3646 (71.3946) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.2544 data: 0.1615 max mem: 3531 +Test: [170/517] eta: 0:01:32 acc1_pos: 22.936 acc5_pos: 100.000 loss: 0.3741 (0.5590) acc1: 90.6250 (72.6425) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.2407 data: 0.1445 max mem: 3531 +Test: [180/517] eta: 0:01:29 acc1_pos: 22.936 acc5_pos: 100.000 loss: 0.3641 (0.5481) acc1: 91.1458 (73.6289) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.2610 data: 0.1640 max mem: 3531 +Test: [190/517] eta: 0:01:26 acc1_pos: 22.936 acc5_pos: 100.000 loss: 0.2333 (0.5301) acc1: 98.4375 (74.9850) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2691 data: 0.1745 max mem: 3531 +Test: [200/517] eta: 0:01:23 acc1_pos: 22.936 acc5_pos: 100.000 loss: 0.2141 (0.5183) acc1: 99.2188 (76.0754) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2304 data: 0.1378 max mem: 3531 +Test: [210/517] eta: 0:01:20 acc1_pos: 22.936 acc5_pos: 100.000 loss: 0.3424 (0.5173) acc1: 93.4896 (76.3292) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.2253 data: 0.1345 max mem: 3532 +Test: [220/517] eta: 0:01:17 acc1_pos: 22.936 acc5_pos: 100.000 loss: 0.5038 (0.5167) acc1: 80.9896 (76.5401) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.2432 data: 0.1488 max mem: 3532 +Test: [230/517] eta: 0:01:14 acc1_pos: 22.936 acc5_pos: 100.000 loss: 0.4708 (0.5135) acc1: 82.5521 (76.9526) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.2445 data: 0.1479 max mem: 3532 +Test: [240/517] eta: 0:01:12 acc1_pos: 22.936 acc5_pos: 100.000 loss: 0.4403 (0.5104) acc1: 85.1562 (77.3232) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.2449 data: 0.1482 max mem: 3532 +Test: [250/517] eta: 0:01:09 acc1_pos: 22.936 acc5_pos: 100.000 loss: 0.3877 (0.5055) acc1: 91.4062 (77.8967) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2649 data: 0.1657 max mem: 3532 +Test: [260/517] eta: 0:01:06 acc1_pos: 22.936 acc5_pos: 100.000 loss: 0.3642 (0.4999) acc1: 94.2708 (78.5351) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.2505 data: 0.1524 max mem: 3532 +Test: [270/517] eta: 0:01:03 acc1_pos: 22.936 acc5_pos: 100.000 loss: 0.3712 (0.4960) acc1: 91.4062 (78.9629) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.2349 data: 0.1395 max mem: 3532 +Test: [280/517] eta: 0:01:01 acc1_pos: 22.936 acc5_pos: 100.000 loss: 0.3807 (0.4922) acc1: 90.3646 (79.3743) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.2593 data: 0.1659 max mem: 3532 +Test: [290/517] eta: 0:00:59 acc1_pos: 22.936 acc5_pos: 100.000 loss: 0.3749 (0.4879) acc1: 92.9688 (79.8853) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2749 data: 0.1797 max mem: 3532 +Test: [300/517] eta: 0:00:56 acc1_pos: 22.936 acc5_pos: 100.000 loss: 0.3586 (0.4837) acc1: 94.0104 (80.2966) acc5: 100.0000 (100.0000) acc1_neg: 80.729 acc5_neg: 100.000 time: 0.2649 data: 0.1693 max mem: 3532 +Test: [310/517] eta: 0:00:53 acc1_pos: 22.936 acc5_pos: 100.000 loss: 0.4235 (0.4836) acc1: 89.3229 (80.3440) acc5: 100.0000 (100.0000) acc1_neg: 77.083 acc5_neg: 100.000 time: 0.2499 data: 0.1561 max mem: 3532 +Test: [320/517] eta: 0:00:51 acc1_pos: 22.936 acc5_pos: 100.000 loss: 0.4411 (0.4803) acc1: 84.6354 (80.6489) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2549 data: 0.1602 max mem: 3532 +Test: [330/517] eta: 0:00:48 acc1_pos: 22.936 acc5_pos: 100.000 loss: 0.3205 (0.4752) acc1: 93.7500 (81.0714) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2549 data: 0.1589 max mem: 3532 +Test: [340/517] eta: 0:00:45 acc1_pos: 22.936 acc5_pos: 100.000 loss: 0.3013 (0.4737) acc1: 94.0104 (81.1736) acc5: 100.0000 (100.0000) acc1_neg: 52.865 acc5_neg: 100.000 time: 0.2499 data: 0.1530 max mem: 3532 +Test: [350/517] eta: 0:00:43 acc1_pos: 22.936 acc5_pos: 100.000 loss: 0.4775 (0.4759) acc1: 79.9479 (80.9317) acc5: 100.0000 (100.0000) acc1_neg: 58.854 acc5_neg: 100.000 time: 0.2449 data: 0.1484 max mem: 3532 +Test: [360/517] eta: 0:00:40 acc1_pos: 22.936 acc5_pos: 100.000 loss: 0.4916 (0.4762) acc1: 79.1667 (80.9167) acc5: 100.0000 (100.0000) acc1_neg: 72.656 acc5_neg: 100.000 time: 0.2399 data: 0.1444 max mem: 3532 +Test: [370/517] eta: 0:00:37 acc1_pos: 22.936 acc5_pos: 100.000 loss: 0.4396 (0.4725) acc1: 85.9375 (81.2774) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2399 data: 0.1457 max mem: 3532 +Test: [380/517] eta: 0:00:35 acc1_pos: 22.936 acc5_pos: 100.000 loss: 0.3476 (0.4702) acc1: 91.4062 (81.4701) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.2451 data: 0.1526 max mem: 3532 +Test: [390/517] eta: 0:00:32 acc1_pos: 22.936 acc5_pos: 100.000 loss: 0.3897 (0.4686) acc1: 85.6771 (81.5317) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.2400 data: 0.1482 max mem: 3532 +Test: [400/517] eta: 0:00:30 acc1_pos: 22.936 acc5_pos: 100.000 loss: 0.3919 (0.4660) acc1: 85.6771 (81.7663) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.3156 data: 0.2235 max mem: 3532 +Test: [410/517] eta: 0:00:27 acc1_pos: 22.936 acc5_pos: 100.000 loss: 0.4068 (0.4660) acc1: 84.8958 (81.7892) acc5: 100.0000 (100.0000) acc1_neg: 72.917 acc5_neg: 100.000 time: 0.3148 data: 0.2199 max mem: 3532 +Test: [420/517] eta: 0:00:25 acc1_pos: 22.936 acc5_pos: 100.000 loss: 0.4287 (0.4648) acc1: 88.5417 (82.0214) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2339 data: 0.1372 max mem: 3532 +Test: [430/517] eta: 0:00:22 acc1_pos: 22.936 acc5_pos: 100.000 loss: 0.4034 (0.4630) acc1: 91.9271 (82.1738) acc5: 100.0000 (100.0000) acc1_neg: 73.698 acc5_neg: 100.000 time: 0.2207 data: 0.1247 max mem: 3532 +Test: [440/517] eta: 0:00:19 acc1_pos: 22.936 acc5_pos: 100.000 loss: 0.4444 (0.4637) acc1: 83.8542 (82.0962) acc5: 100.0000 (100.0000) acc1_neg: 69.271 acc5_neg: 100.000 time: 0.2164 data: 0.1180 max mem: 3533 +Test: [450/517] eta: 0:00:17 acc1_pos: 22.936 acc5_pos: 100.000 loss: 0.4150 (0.4614) acc1: 84.3750 (82.3598) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2380 data: 0.1397 max mem: 3533 +Test: [460/517] eta: 0:00:14 acc1_pos: 22.936 acc5_pos: 100.000 loss: 0.3716 (0.4607) acc1: 92.1875 (82.4645) acc5: 100.0000 (100.0000) acc1_neg: 74.740 acc5_neg: 100.000 time: 0.2423 data: 0.1458 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 22.936 acc5_pos: 100.000 loss: 0.3707 (0.4586) acc1: 91.4062 (82.6909) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.2506 data: 0.1567 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 22.936 acc5_pos: 100.000 loss: 0.3349 (0.4559) acc1: 95.3125 (82.9798) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2599 data: 0.1664 max mem: 3533 +Test: [490/517] eta: 0:00:06 acc1_pos: 22.936 acc5_pos: 100.000 loss: 0.2948 (0.4526) acc1: 97.1354 (83.2649) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.2502 data: 0.1555 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 22.936 acc5_pos: 100.000 loss: 0.3863 (0.4552) acc1: 85.1562 (82.9362) acc5: 100.0000 (100.0000) acc1_neg: 68.490 acc5_neg: 100.000 time: 0.2349 data: 0.1422 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 22.936 acc5_pos: 100.000 loss: 0.4792 (0.4535) acc1: 79.4271 (83.0948) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2079 data: 0.1165 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 22.936 acc5_pos: 100.000 loss: 0.3425 (0.4519) acc1: 94.2708 (83.2467) acc5: 100.0000 (100.0000) acc1_neg: 97.527 acc5_neg: 100.000 time: 0.1953 data: 0.1028 max mem: 3533 +Test: Total time: 0:02:11 (0.2544 s / it) +* Overall Acc@1 83.247 Acc@5 100.000 + target == real Acc@1 45.164 Acc@5 100.000 + target == fake Acc@1 87.197 Acc@5 100.000 + loss 0.452 + PR-AUC 0.9562 ROC-AUC 0.7402 EER 31.95% (thr=0.6357) +Accuracy of the network on the 198427 test images: 83.2% +[rank0]:[W726 04:42:01.547969517 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 04:42:20.950000 1187438 torch/distributed/run.py:766] +W0726 04:42:20.950000 1187438 torch/distributed/run.py:766] ***************************************** +W0726 04:42:20.950000 1187438 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:42:20.950000 1187438 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:42:49.952017535 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:42:50.484933778 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:42:50.800740135 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:42:50.065671029 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:42:50.093315227 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:42:50.098478592 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:42:50.116609437 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank6]:[W726 04:42:50.116617343 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n23_seed7_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed7_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:45:18 acc1_pos: 40.625 acc5_pos: 100.000 loss: 0.8069 (0.8069) acc1: 40.6250 (40.6250) acc5: 100.0000 (100.0000) time: 6.9897 data: 5.4960 max mem: 3531 +Test: [ 10/389] eta: 0:04:33 acc1_pos: 49.219 acc5_pos: 100.000 loss: 0.7344 (0.6761) acc1: 49.2188 (58.5938) acc5: 100.0000 (100.0000) time: 0.7212 data: 0.5001 max mem: 3531 +Test: [ 20/389] eta: 0:02:36 acc1_pos: 60.417 acc5_pos: 100.000 loss: 0.6931 (0.6983) acc1: 54.9479 (57.2545) acc5: 100.0000 (100.0000) time: 0.0972 data: 0.0004 max mem: 3531 +Test: [ 30/389] eta: 0:02:17 acc1_pos: 79.688 acc5_pos: 100.000 loss: 0.5808 (0.6304) acc1: 72.3958 (65.7342) acc5: 100.0000 (100.0000) time: 0.1949 data: 0.0954 max mem: 3531 +Test: [ 40/389] eta: 0:01:57 acc1_pos: 18.490 acc5_pos: 100.000 loss: 0.5577 (0.6492) acc1: 77.8646 (64.3356) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1481 max mem: 3531 +Test: [ 50/389] eta: 0:01:45 acc1_pos: 89.062 acc5_pos: 100.000 loss: 0.5424 (0.6151) acc1: 81.5104 (68.4896) acc5: 100.0000 (100.0000) time: 0.1999 data: 0.1061 max mem: 3531 +Test: [ 60/389] eta: 0:01:35 acc1_pos: 81.250 acc5_pos: 100.000 loss: 0.4331 (0.5804) acc1: 90.1042 (72.3105) acc5: 100.0000 (100.0000) time: 0.1899 data: 0.0941 max mem: 3531 +Test: [ 70/389] eta: 0:01:25 acc1_pos: 58.073 acc5_pos: 100.000 loss: 0.4447 (0.5929) acc1: 81.2500 (70.7526) acc5: 100.0000 (100.0000) time: 0.1649 data: 0.0684 max mem: 3531 +Test: [ 80/389] eta: 0:01:19 acc1_pos: 98.177 acc5_pos: 100.000 loss: 0.6359 (0.6099) acc1: 58.0729 (68.7179) acc5: 100.0000 (100.0000) time: 0.1650 data: 0.0721 max mem: 3531 +Test: [ 90/389] eta: 0:01:14 acc1_pos: 88.021 acc5_pos: 100.000 loss: 0.4067 (0.5868) acc1: 85.1562 (70.9106) acc5: 100.0000 (100.0000) time: 0.1799 data: 0.0880 max mem: 3531 +Test: [100/389] eta: 0:01:11 acc1_pos: 49.219 acc5_pos: 100.000 loss: 0.5099 (0.5948) acc1: 76.8229 (70.0676) acc5: 100.0000 (100.0000) time: 0.2102 data: 0.1184 max mem: 3531 +Test: [110/389] eta: 0:01:08 acc1_pos: 97.135 acc5_pos: 100.000 loss: 0.6568 (0.6157) acc1: 60.6771 (68.3371) acc5: 100.0000 (100.0000) time: 0.2199 data: 0.1253 max mem: 3531 +Test: [120/389] eta: 0:01:04 acc1_pos: 73.958 acc5_pos: 100.000 loss: 0.5294 (0.6100) acc1: 73.9583 (68.7823) acc5: 100.0000 (100.0000) time: 0.1898 data: 0.0937 max mem: 3531 +Test: [130/389] eta: 0:01:00 acc1_pos: 71.779 acc5_pos: 100.000 loss: 0.5377 (0.6153) acc1: 73.9583 (68.0025) acc5: 100.0000 (100.0000) acc1_neg: 75.521 acc5_neg: 100.000 time: 0.1699 data: 0.0760 max mem: 3531 +Test: [140/389] eta: 0:00:56 acc1_pos: 71.779 acc5_pos: 100.000 loss: 0.6362 (0.6157) acc1: 64.5833 (67.8081) acc5: 100.0000 (100.0000) acc1_neg: 31.510 acc5_neg: 100.000 time: 0.1546 data: 0.0594 max mem: 3531 +Test: [150/389] eta: 0:00:53 acc1_pos: 71.779 acc5_pos: 100.000 loss: 0.6425 (0.6211) acc1: 54.1667 (66.8133) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1626 data: 0.0675 max mem: 3531 +Test: [160/389] eta: 0:00:51 acc1_pos: 71.779 acc5_pos: 100.000 loss: 0.6101 (0.6115) acc1: 62.5000 (67.7439) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1972 data: 0.1050 max mem: 3531 +Test: [170/389] eta: 0:00:48 acc1_pos: 71.779 acc5_pos: 100.000 loss: 0.3551 (0.6048) acc1: 97.9167 (68.4942) acc5: 100.0000 (100.0000) acc1_neg: 64.062 acc5_neg: 100.000 time: 0.2038 data: 0.1132 max mem: 3531 +Test: [180/389] eta: 0:00:46 acc1_pos: 71.779 acc5_pos: 100.000 loss: 0.4888 (0.6021) acc1: 79.9479 (68.7802) acc5: 100.0000 (100.0000) acc1_neg: 65.625 acc5_neg: 100.000 time: 0.2085 data: 0.1155 max mem: 3531 +Test: [190/389] eta: 0:00:43 acc1_pos: 71.779 acc5_pos: 100.000 loss: 0.5648 (0.6039) acc1: 70.8333 (68.3028) acc5: 100.0000 (100.0000) acc1_neg: 29.688 acc5_neg: 100.000 time: 0.1883 data: 0.0904 max mem: 3531 +Test: [200/389] eta: 0:00:40 acc1_pos: 71.779 acc5_pos: 100.000 loss: 0.7305 (0.6085) acc1: 45.3125 (67.5308) acc5: 100.0000 (100.0000) acc1_neg: 40.625 acc5_neg: 100.000 time: 0.1590 data: 0.0605 max mem: 3531 +Test: [210/389] eta: 0:00:38 acc1_pos: 71.779 acc5_pos: 100.000 loss: 0.4577 (0.6036) acc1: 45.3125 (67.9367) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1816 data: 0.0864 max mem: 3532 +Test: [220/389] eta: 0:00:35 acc1_pos: 71.779 acc5_pos: 100.000 loss: 0.3796 (0.5984) acc1: 95.3125 (68.4212) acc5: 100.0000 (100.0000) acc1_neg: 40.625 acc5_neg: 100.000 time: 0.1649 data: 0.0683 max mem: 3532 +Test: [230/389] eta: 0:00:33 acc1_pos: 71.779 acc5_pos: 100.000 loss: 0.5698 (0.6041) acc1: 65.8854 (67.6599) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.1581 data: 0.0627 max mem: 3532 +Test: [240/389] eta: 0:00:31 acc1_pos: 71.779 acc5_pos: 100.000 loss: 0.5763 (0.6014) acc1: 65.8854 (68.0455) acc5: 100.0000 (100.0000) acc1_neg: 71.354 acc5_neg: 100.000 time: 0.1866 data: 0.0944 max mem: 3532 +Test: [250/389] eta: 0:00:28 acc1_pos: 71.779 acc5_pos: 100.000 loss: 0.4704 (0.5958) acc1: 86.4583 (68.8538) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1910 data: 0.0989 max mem: 3532 +Test: [260/389] eta: 0:00:26 acc1_pos: 71.779 acc5_pos: 100.000 loss: 0.5007 (0.5971) acc1: 82.0312 (68.5405) acc5: 100.0000 (100.0000) acc1_neg: 24.479 acc5_neg: 100.000 time: 0.1800 data: 0.0886 max mem: 3532 +Test: [270/389] eta: 0:00:24 acc1_pos: 71.779 acc5_pos: 100.000 loss: 0.6820 (0.6038) acc1: 45.0521 (67.5046) acc5: 100.0000 (100.0000) acc1_neg: 57.552 acc5_neg: 100.000 time: 0.1796 data: 0.0885 max mem: 3532 +Test: [280/389] eta: 0:00:22 acc1_pos: 71.779 acc5_pos: 100.000 loss: 0.6892 (0.6094) acc1: 50.2604 (66.8159) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1880 data: 0.0970 max mem: 3532 +Test: [290/389] eta: 0:00:20 acc1_pos: 71.779 acc5_pos: 100.000 loss: 0.5597 (0.6051) acc1: 58.8542 (67.3101) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1946 data: 0.1025 max mem: 3532 +Test: [300/389] eta: 0:00:18 acc1_pos: 71.779 acc5_pos: 100.000 loss: 0.4024 (0.5998) acc1: 89.8438 (67.9757) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1886 data: 0.0961 max mem: 3532 +Test: [310/389] eta: 0:00:16 acc1_pos: 71.779 acc5_pos: 100.000 loss: 0.5199 (0.6043) acc1: 71.8750 (67.5970) acc5: 100.0000 (100.0000) acc1_neg: 2.083 acc5_neg: 100.000 time: 0.1709 data: 0.0794 max mem: 3532 +Test: [320/389] eta: 0:00:13 acc1_pos: 71.779 acc5_pos: 100.000 loss: 0.6215 (0.6035) acc1: 61.9792 (67.6832) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1514 data: 0.0606 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 71.779 acc5_pos: 100.000 loss: 0.8448 (0.6157) acc1: 33.5938 (66.1915) acc5: 100.0000 (100.0000) acc1_neg: 33.594 acc5_neg: 100.000 time: 0.1634 data: 0.0710 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 71.779 acc5_pos: 100.000 loss: 0.4796 (0.6080) acc1: 66.1458 (67.0500) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1608 data: 0.0686 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 71.779 acc5_pos: 100.000 loss: 0.3737 (0.6127) acc1: 97.1354 (66.6667) acc5: 100.0000 (100.0000) acc1_neg: 29.948 acc5_neg: 100.000 time: 0.1281 data: 0.0376 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 71.779 acc5_pos: 100.000 loss: 0.6585 (0.6151) acc1: 58.3333 (66.2822) acc5: 100.0000 (100.0000) acc1_neg: 61.458 acc5_neg: 100.000 time: 0.1274 data: 0.0371 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 71.779 acc5_pos: 100.000 loss: 0.6560 (0.6177) acc1: 59.6354 (66.2448) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1321 data: 0.0414 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 71.779 acc5_pos: 100.000 loss: 0.5452 (0.6173) acc1: 71.6146 (66.3242) acc5: 100.0000 (100.0000) acc1_neg: 71.615 acc5_neg: 100.000 time: 0.1246 data: 0.0337 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 71.779 acc5_pos: 100.000 loss: 0.4878 (0.6155) acc1: 79.4271 (66.4906) acc5: 100.0000 (100.0000) acc1_neg: 62.842 acc5_neg: 100.000 time: 0.1187 data: 0.0247 max mem: 3532 +Test: Total time: 0:01:13 (0.1901 s / it) +* Overall Acc@1 66.491 Acc@5 100.000 + target == real Acc@1 68.585 Acc@5 100.000 + target == fake Acc@1 65.517 Acc@5 100.000 + loss 0.615 + PR-AUC 0.8228 ROC-AUC 0.7268 EER 33.00% (thr=0.4902) +Accuracy of the network on the 149358 test images: 66.5% +[rank0]:[W726 04:44:15.727423819 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:44:34.958000 1187998 torch/distributed/run.py:766] +W0726 04:44:34.958000 1187998 torch/distributed/run.py:766] ***************************************** +W0726 04:44:34.958000 1187998 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:44:34.958000 1187998 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:45:03.783682628 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:45:03.829847559 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:45:03.145425150 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:45:03.145463170 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 7): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:45:03.299259627 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W726 04:45:03.299301404 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:45:03.405743169 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:45:04.519890068 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n23_seed7_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed7_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:32:28 acc1_pos: 12.500 acc5_pos: 100.000 loss: 1.4611 (1.4611) acc1: 12.5000 (12.5000) acc5: 100.0000 (100.0000) time: 5.8501 data: 4.1674 max mem: 3531 +Test: [ 10/333] eta: 0:03:18 acc1_pos: 88.021 acc5_pos: 100.000 loss: 1.2026 (0.9300) acc1: 37.2396 (48.5085) acc5: 100.0000 (100.0000) time: 0.6136 data: 0.3792 max mem: 3531 +Test: [ 20/333] eta: 0:01:54 acc1_pos: 95.052 acc5_pos: 100.000 loss: 0.3546 (0.6410) acc1: 85.4167 (66.7783) acc5: 100.0000 (100.0000) time: 0.0914 data: 0.0004 max mem: 3531 +Test: [ 30/333] eta: 0:01:31 acc1_pos: 46.094 acc5_pos: 100.000 loss: 0.3971 (0.7051) acc1: 76.0417 (60.7275) acc5: 100.0000 (100.0000) time: 0.1317 data: 0.0366 max mem: 3531 +Test: [ 40/333] eta: 0:01:26 acc1_pos: 41.927 acc5_pos: 100.000 loss: 0.8522 (0.7469) acc1: 44.2708 (55.9197) acc5: 100.0000 (100.0000) time: 0.2199 data: 0.1225 max mem: 3531 +Test: [ 50/333] eta: 0:01:17 acc1_pos: 39.062 acc5_pos: 100.000 loss: 0.8906 (0.7782) acc1: 40.6250 (52.5991) acc5: 100.0000 (100.0000) time: 0.2303 data: 0.1359 max mem: 3531 +Test: [ 60/333] eta: 0:01:11 acc1_pos: 38.281 acc5_pos: 100.000 loss: 0.8797 (0.7949) acc1: 40.6250 (51.2252) acc5: 100.0000 (100.0000) time: 0.1999 data: 0.1071 max mem: 3531 +Test: [ 70/333] eta: 0:01:05 acc1_pos: 41.667 acc5_pos: 100.000 loss: 0.8797 (0.8068) acc1: 46.0938 (50.4695) acc5: 100.0000 (100.0000) time: 0.1842 data: 0.0894 max mem: 3531 +Test: [ 80/333] eta: 0:01:03 acc1_pos: 39.323 acc5_pos: 100.000 loss: 0.8843 (0.8201) acc1: 42.1875 (49.0998) acc5: 100.0000 (100.0000) time: 0.2085 data: 0.1110 max mem: 3531 +Test: [ 90/333] eta: 0:00:59 acc1_pos: 65.625 acc5_pos: 100.000 loss: 0.7997 (0.8048) acc1: 43.4896 (50.4350) acc5: 100.0000 (100.0000) time: 0.2361 data: 0.1370 max mem: 3531 +Test: [100/333] eta: 0:00:56 acc1_pos: 70.312 acc5_pos: 100.000 loss: 0.6229 (0.7857) acc1: 66.4062 (51.9699) acc5: 100.0000 (100.0000) time: 0.2023 data: 0.1049 max mem: 3531 +Test: [110/333] eta: 0:00:52 acc1_pos: 64.583 acc5_pos: 100.000 loss: 0.6307 (0.7727) acc1: 65.3646 (53.0523) acc5: 100.0000 (100.0000) time: 0.1837 data: 0.0882 max mem: 3531 +Test: [120/333] eta: 0:00:48 acc1_pos: 78.646 acc5_pos: 100.000 loss: 0.6032 (0.7598) acc1: 64.5833 (54.0892) acc5: 100.0000 (100.0000) time: 0.1739 data: 0.0753 max mem: 3531 +Test: [130/333] eta: 0:00:46 acc1_pos: 71.354 acc5_pos: 100.000 loss: 0.5876 (0.7472) acc1: 65.8854 (55.0970) acc5: 100.0000 (100.0000) time: 0.1936 data: 0.0962 max mem: 3531 +Test: [140/333] eta: 0:00:44 acc1_pos: 74.740 acc5_pos: 100.000 loss: 0.5544 (0.7342) acc1: 68.4896 (56.1392) acc5: 100.0000 (100.0000) time: 0.2225 data: 0.1264 max mem: 3531 +Test: [150/333] eta: 0:00:41 acc1_pos: 70.312 acc5_pos: 100.000 loss: 0.5714 (0.7270) acc1: 67.1875 (56.7208) acc5: 100.0000 (100.0000) time: 0.2212 data: 0.1245 max mem: 3531 +Test: [160/333] eta: 0:00:38 acc1_pos: 55.263 acc5_pos: 100.000 loss: 0.5860 (0.7252) acc1: 61.9792 (57.0086) acc5: 100.0000 (100.0000) acc1_neg: 51.042 acc5_neg: 100.000 time: 0.2030 data: 0.1084 max mem: 3531 +Test: [170/333] eta: 0:00:36 acc1_pos: 55.263 acc5_pos: 100.000 loss: 0.7421 (0.7292) acc1: 55.7292 (56.8074) acc5: 100.0000 (100.0000) acc1_neg: 59.635 acc5_neg: 100.000 time: 0.1862 data: 0.0904 max mem: 3531 +Test: [180/333] eta: 0:00:33 acc1_pos: 55.263 acc5_pos: 100.000 loss: 0.7196 (0.7269) acc1: 59.6354 (57.0514) acc5: 100.0000 (100.0000) acc1_neg: 48.438 acc5_neg: 100.000 time: 0.1791 data: 0.0828 max mem: 3531 +Test: [190/333] eta: 0:00:31 acc1_pos: 55.263 acc5_pos: 100.000 loss: 0.6515 (0.7198) acc1: 62.2396 (57.6775) acc5: 100.0000 (100.0000) acc1_neg: 69.010 acc5_neg: 100.000 time: 0.1936 data: 0.0994 max mem: 3531 +Test: [200/333] eta: 0:00:28 acc1_pos: 55.263 acc5_pos: 100.000 loss: 0.6197 (0.7153) acc1: 65.1042 (58.1079) acc5: 100.0000 (100.0000) acc1_neg: 60.938 acc5_neg: 100.000 time: 0.1957 data: 0.1024 max mem: 3531 +Test: [210/333] eta: 0:00:26 acc1_pos: 55.263 acc5_pos: 100.000 loss: 0.6836 (0.7156) acc1: 63.0208 (58.1840) acc5: 100.0000 (100.0000) acc1_neg: 51.302 acc5_neg: 100.000 time: 0.1666 data: 0.0726 max mem: 3532 +Test: [220/333] eta: 0:00:24 acc1_pos: 55.263 acc5_pos: 100.000 loss: 0.7065 (0.7198) acc1: 56.5104 (57.9351) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.1772 data: 0.0839 max mem: 3532 +Test: [230/333] eta: 0:00:21 acc1_pos: 55.263 acc5_pos: 100.000 loss: 0.5164 (0.7096) acc1: 74.7396 (58.8767) acc5: 100.0000 (100.0000) acc1_neg: 72.396 acc5_neg: 100.000 time: 0.1900 data: 0.0958 max mem: 3532 +Test: [240/333] eta: 0:00:19 acc1_pos: 55.263 acc5_pos: 100.000 loss: 0.5164 (0.7040) acc1: 74.7396 (59.3653) acc5: 100.0000 (100.0000) acc1_neg: 66.406 acc5_neg: 100.000 time: 0.1694 data: 0.0750 max mem: 3532 +Test: [250/333] eta: 0:00:17 acc1_pos: 55.263 acc5_pos: 100.000 loss: 0.6000 (0.7005) acc1: 69.0104 (59.7153) acc5: 100.0000 (100.0000) acc1_neg: 59.896 acc5_neg: 100.000 time: 0.1759 data: 0.0813 max mem: 3532 +Test: [260/333] eta: 0:00:15 acc1_pos: 55.263 acc5_pos: 100.000 loss: 0.6287 (0.6986) acc1: 66.9271 (59.9527) acc5: 100.0000 (100.0000) acc1_neg: 66.667 acc5_neg: 100.000 time: 0.1851 data: 0.0910 max mem: 3532 +Test: [270/333] eta: 0:00:13 acc1_pos: 55.263 acc5_pos: 100.000 loss: 0.6532 (0.6965) acc1: 64.3229 (60.1514) acc5: 100.0000 (100.0000) acc1_neg: 63.802 acc5_neg: 100.000 time: 0.1889 data: 0.0974 max mem: 3532 +Test: [280/333] eta: 0:00:10 acc1_pos: 55.263 acc5_pos: 100.000 loss: 0.6635 (0.6997) acc1: 62.7604 (59.8560) acc5: 100.0000 (100.0000) acc1_neg: 52.083 acc5_neg: 100.000 time: 0.1744 data: 0.0837 max mem: 3532 +Test: [290/333] eta: 0:00:08 acc1_pos: 55.263 acc5_pos: 100.000 loss: 0.8771 (0.7059) acc1: 46.0938 (59.4197) acc5: 100.0000 (100.0000) acc1_neg: 52.604 acc5_neg: 100.000 time: 0.1621 data: 0.0701 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 55.263 acc5_pos: 100.000 loss: 0.8373 (0.7096) acc1: 49.7396 (59.1648) acc5: 100.0000 (100.0000) acc1_neg: 54.688 acc5_neg: 100.000 time: 0.1768 data: 0.0843 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 55.263 acc5_pos: 100.000 loss: 0.7321 (0.7048) acc1: 55.2083 (59.5768) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.1725 data: 0.0817 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 55.263 acc5_pos: 100.000 loss: 0.5339 (0.7005) acc1: 73.9583 (59.9437) acc5: 100.0000 (100.0000) acc1_neg: 53.906 acc5_neg: 100.000 time: 0.1618 data: 0.0692 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 55.263 acc5_pos: 100.000 loss: 0.7639 (0.7036) acc1: 55.4688 (59.7054) acc5: 100.0000 (100.0000) acc1_neg: 55.729 acc5_neg: 100.000 time: 0.1507 data: 0.0583 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 55.263 acc5_pos: 100.000 loss: 0.7789 (0.7041) acc1: 53.9062 (59.6677) acc5: 100.0000 (100.0000) acc1_neg: 55.775 acc5_neg: 100.000 time: 0.1500 data: 0.0548 max mem: 3532 +Test: Total time: 0:01:06 (0.2001 s / it) +* Overall Acc@1 59.668 Acc@5 100.000 + target == real Acc@1 57.306 Acc@5 100.000 + target == fake Acc@1 61.770 Acc@5 100.000 + loss 0.704 + PR-AUC 0.6339 ROC-AUC 0.6307 EER 40.42% (thr=0.5171) +Accuracy of the network on the 127843 test images: 59.7% +[rank0]:[W726 04:46:22.835580360 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:46:41.904000 1188557 torch/distributed/run.py:766] +W0726 04:46:41.904000 1188557 torch/distributed/run.py:766] ***************************************** +W0726 04:46:41.904000 1188557 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:46:41.904000 1188557 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:47:10.626178458 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:47:10.220666329 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:47:10.245591815 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:47:10.370488260 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:47:11.416967501 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:47:11.421773976 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:47:11.486355378 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:47:11.496835983 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n23_seed7_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed7_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:24:19 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.2308 (0.2308) acc1: 95.3125 (95.3125) acc5: 100.0000 (100.0000) time: 6.2090 data: 4.0469 max mem: 3531 +Test: [ 10/235] eta: 0:02:26 acc1_pos: 94.531 acc5_pos: 100.000 loss: 0.2377 (0.2391) acc1: 94.7917 (94.6496) acc5: 100.0000 (100.0000) time: 0.6517 data: 0.3683 max mem: 3531 +Test: [ 20/235] eta: 0:01:28 acc1_pos: 95.052 acc5_pos: 100.000 loss: 0.2337 (0.2344) acc1: 94.7917 (94.8909) acc5: 100.0000 (100.0000) time: 0.1204 data: 0.0254 max mem: 3531 +Test: [ 30/235] eta: 0:01:10 acc1_pos: 96.354 acc5_pos: 100.000 loss: 0.2277 (0.2332) acc1: 95.0521 (94.9429) acc5: 100.0000 (100.0000) time: 0.1781 data: 0.0848 max mem: 3531 +Test: [ 40/235] eta: 0:01:03 acc1_pos: 94.792 acc5_pos: 100.000 loss: 0.2277 (0.2332) acc1: 95.0521 (94.9060) acc5: 100.0000 (100.0000) time: 0.2374 data: 0.1460 max mem: 3531 +Test: [ 50/235] eta: 0:00:56 acc1_pos: 92.708 acc5_pos: 100.000 loss: 0.2353 (0.2348) acc1: 94.7917 (94.7559) acc5: 100.0000 (100.0000) time: 0.2393 data: 0.1476 max mem: 3531 +Test: [ 60/235] eta: 0:00:51 acc1_pos: 94.010 acc5_pos: 100.000 loss: 0.2406 (0.2363) acc1: 94.5312 (94.7106) acc5: 100.0000 (100.0000) time: 0.2228 data: 0.1302 max mem: 3531 +Test: [ 70/235] eta: 0:00:46 acc1_pos: 96.923 acc5_pos: 100.000 loss: 0.2567 (0.3767) acc1: 92.1875 (85.5707) acc5: 100.0000 (100.0000) acc1_neg: 20.312 acc5_neg: 100.000 time: 0.2306 data: 0.1361 max mem: 3531 +Test: [ 80/235] eta: 0:00:42 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.3647 (0.5009) acc1: 21.3542 (77.5334) acc5: 100.0000 (100.0000) acc1_neg: 18.229 acc5_neg: 100.000 time: 0.2309 data: 0.1369 max mem: 3531 +Test: [ 90/235] eta: 0:00:39 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.3848 (0.5978) acc1: 20.8333 (71.2769) acc5: 100.0000 (100.0000) acc1_neg: 19.792 acc5_neg: 100.000 time: 0.2391 data: 0.1468 max mem: 3531 +Test: [100/235] eta: 0:00:35 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.3787 (0.6743) acc1: 20.8333 (66.4398) acc5: 100.0000 (100.0000) acc1_neg: 20.833 acc5_neg: 100.000 time: 0.1946 data: 0.0994 max mem: 3531 +Test: [110/235] eta: 0:00:31 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.3663 (0.7366) acc1: 21.3542 (62.3921) acc5: 100.0000 (100.0000) acc1_neg: 21.875 acc5_neg: 100.000 time: 0.1661 data: 0.0707 max mem: 3531 +Test: [120/235] eta: 0:00:28 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.4320 (0.8065) acc1: 18.4896 (58.2559) acc5: 100.0000 (100.0000) acc1_neg: 12.500 acc5_neg: 100.000 time: 0.2063 data: 0.1125 max mem: 3531 +Test: [130/235] eta: 0:00:26 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.5971 (0.8667) acc1: 12.2396 (54.7332) acc5: 100.0000 (100.0000) acc1_neg: 12.240 acc5_neg: 100.000 time: 0.2141 data: 0.1215 max mem: 3531 +Test: [140/235] eta: 0:00:22 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.6137 (0.9206) acc1: 11.9792 (51.6826) acc5: 100.0000 (100.0000) acc1_neg: 9.375 acc5_neg: 100.000 time: 0.1798 data: 0.0870 max mem: 3531 +Test: [150/235] eta: 0:00:20 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.6016 (0.9644) acc1: 11.9792 (49.0428) acc5: 100.0000 (100.0000) acc1_neg: 12.500 acc5_neg: 100.000 time: 0.1613 data: 0.0647 max mem: 3531 +Test: [160/235] eta: 0:00:17 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.5878 (1.0034) acc1: 11.7188 (46.7682) acc5: 100.0000 (100.0000) acc1_neg: 13.281 acc5_neg: 100.000 time: 0.1775 data: 0.0812 max mem: 3531 +Test: [170/235] eta: 0:00:15 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.5594 (1.0196) acc1: 14.3229 (45.4937) acc5: 100.0000 (100.0000) acc1_neg: 22.396 acc5_neg: 100.000 time: 0.1888 data: 0.0964 max mem: 3531 +Test: [180/235] eta: 0:00:12 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.2402 (1.0315) acc1: 27.6042 (44.5802) acc5: 100.0000 (100.0000) acc1_neg: 29.948 acc5_neg: 100.000 time: 0.1899 data: 0.0989 max mem: 3531 +Test: [190/235] eta: 0:00:10 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.2256 (1.0422) acc1: 28.6458 (43.6996) acc5: 100.0000 (100.0000) acc1_neg: 25.781 acc5_neg: 100.000 time: 0.1609 data: 0.0696 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.2295 (1.0526) acc1: 25.7812 (42.8236) acc5: 100.0000 (100.0000) acc1_neg: 25.260 acc5_neg: 100.000 time: 0.1524 data: 0.0612 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.2335 (1.0613) acc1: 26.3021 (42.0962) acc5: 100.0000 (100.0000) acc1_neg: 27.865 acc5_neg: 100.000 time: 0.1659 data: 0.0751 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.2284 (1.0687) acc1: 27.8646 (41.4628) acc5: 100.0000 (100.0000) acc1_neg: 24.479 acc5_neg: 100.000 time: 0.1541 data: 0.0639 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.2230 (1.0756) acc1: 28.1250 (40.8832) acc5: 100.0000 (100.0000) acc1_neg: 28.906 acc5_neg: 100.000 time: 0.1432 data: 0.0534 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.2335 (1.0798) acc1: 27.8646 (40.7003) acc5: 100.0000 (100.0000) acc1_neg: 30.435 acc5_neg: 100.000 time: 0.1353 data: 0.0487 max mem: 3532 +Test: Total time: 0:00:49 (0.2107 s / it) +* Overall Acc@1 40.700 Acc@5 100.000 + target == real Acc@1 94.722 Acc@5 100.000 + target == fake Acc@1 21.162 Acc@5 100.000 + loss 1.080 + PR-AUC 0.8590 ROC-AUC 0.6953 EER 35.65% (thr=0.1849) +Accuracy of the network on the 89879 test images: 40.7% +[rank0]:[W726 04:48:10.115514983 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:48:30.053000 1189106 torch/distributed/run.py:766] +W0726 04:48:30.053000 1189106 torch/distributed/run.py:766] ***************************************** +W0726 04:48:30.053000 1189106 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:48:30.053000 1189106 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:48:58.232100262 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:48:59.555127322 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:48:59.826484650 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:48:59.135731799 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:48:59.218230461 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:48:59.246543763 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +[rank0]:[W726 04:48:59.266255658 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:48:59.273086880 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n23_seed7_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed7_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:23:11 acc1_pos: 68.490 acc5_pos: 100.000 loss: 0.6044 (0.6044) acc1: 68.4896 (68.4896) acc5: 100.0000 (100.0000) time: 5.1739 data: 3.0313 max mem: 3531 +Test: [ 10/269] eta: 0:02:27 acc1_pos: 66.406 acc5_pos: 100.000 loss: 0.5942 (0.6214) acc1: 67.4479 (65.4119) acc5: 100.0000 (100.0000) time: 0.5692 data: 0.2923 max mem: 3531 +Test: [ 20/269] eta: 0:01:33 acc1_pos: 56.250 acc5_pos: 100.000 loss: 0.5886 (0.5826) acc1: 66.4062 (68.6632) acc5: 100.0000 (100.0000) time: 0.1362 data: 0.0430 max mem: 3531 +Test: [ 30/269] eta: 0:01:12 acc1_pos: 89.323 acc5_pos: 100.000 loss: 0.5710 (0.5709) acc1: 72.6562 (69.9009) acc5: 100.0000 (100.0000) time: 0.1574 data: 0.0620 max mem: 3531 +Test: [ 40/269] eta: 0:01:05 acc1_pos: 72.414 acc5_pos: 100.000 loss: 0.5442 (0.5647) acc1: 72.1354 (71.2398) acc5: 100.0000 (100.0000) acc1_neg: 66.146 acc5_neg: 100.000 time: 0.1903 data: 0.0976 max mem: 3531 +Test: [ 50/269] eta: 0:00:58 acc1_pos: 72.414 acc5_pos: 100.000 loss: 0.6037 (0.6038) acc1: 66.1458 (67.5347) acc5: 100.0000 (100.0000) acc1_neg: 31.510 acc5_neg: 100.000 time: 0.2119 data: 0.1163 max mem: 3531 +Test: [ 60/269] eta: 0:00:53 acc1_pos: 72.414 acc5_pos: 100.000 loss: 0.8072 (0.6535) acc1: 43.2292 (62.6494) acc5: 100.0000 (100.0000) acc1_neg: 25.781 acc5_neg: 100.000 time: 0.1995 data: 0.1000 max mem: 3531 +Test: [ 70/269] eta: 0:00:48 acc1_pos: 72.414 acc5_pos: 100.000 loss: 0.8237 (0.6702) acc1: 41.6667 (60.4827) acc5: 100.0000 (100.0000) acc1_neg: 51.562 acc5_neg: 100.000 time: 0.1872 data: 0.0912 max mem: 3531 +Test: [ 80/269] eta: 0:00:44 acc1_pos: 72.414 acc5_pos: 100.000 loss: 0.7462 (0.6775) acc1: 47.9167 (59.5293) acc5: 100.0000 (100.0000) acc1_neg: 74.219 acc5_neg: 100.000 time: 0.1754 data: 0.0816 max mem: 3531 +Test: [ 90/269] eta: 0:00:41 acc1_pos: 72.414 acc5_pos: 100.000 loss: 0.6774 (0.6753) acc1: 53.1250 (59.7585) acc5: 100.0000 (100.0000) acc1_neg: 53.125 acc5_neg: 100.000 time: 0.1776 data: 0.0828 max mem: 3531 +Test: [100/269] eta: 0:00:37 acc1_pos: 72.414 acc5_pos: 100.000 loss: 0.6475 (0.6760) acc1: 59.3750 (59.5710) acc5: 100.0000 (100.0000) acc1_neg: 46.875 acc5_neg: 100.000 time: 0.1722 data: 0.0788 max mem: 3531 +Test: [110/269] eta: 0:00:35 acc1_pos: 72.414 acc5_pos: 100.000 loss: 0.7179 (0.6836) acc1: 53.9062 (58.6243) acc5: 100.0000 (100.0000) acc1_neg: 43.229 acc5_neg: 100.000 time: 0.1839 data: 0.0924 max mem: 3531 +Test: [120/269] eta: 0:00:32 acc1_pos: 72.414 acc5_pos: 100.000 loss: 0.7501 (0.6865) acc1: 49.2188 (58.3204) acc5: 100.0000 (100.0000) acc1_neg: 49.219 acc5_neg: 100.000 time: 0.1979 data: 0.1044 max mem: 3531 +Test: [130/269] eta: 0:00:29 acc1_pos: 72.414 acc5_pos: 100.000 loss: 0.7504 (0.6925) acc1: 51.5625 (57.7091) acc5: 100.0000 (100.0000) acc1_neg: 44.792 acc5_neg: 100.000 time: 0.1481 data: 0.0518 max mem: 3531 +Test: [140/269] eta: 0:00:26 acc1_pos: 72.414 acc5_pos: 100.000 loss: 0.8562 (0.7087) acc1: 42.7083 (56.1983) acc5: 100.0000 (100.0000) acc1_neg: 29.688 acc5_neg: 100.000 time: 0.1404 data: 0.0460 max mem: 3531 +Test: [150/269] eta: 0:00:24 acc1_pos: 72.414 acc5_pos: 100.000 loss: 0.8697 (0.7174) acc1: 40.6250 (55.5964) acc5: 100.0000 (100.0000) acc1_neg: 44.271 acc5_neg: 100.000 time: 0.1866 data: 0.0957 max mem: 3531 +Test: [160/269] eta: 0:00:22 acc1_pos: 72.414 acc5_pos: 100.000 loss: 0.7781 (0.7211) acc1: 44.5312 (55.2374) acc5: 100.0000 (100.0000) acc1_neg: 48.177 acc5_neg: 100.000 time: 0.1985 data: 0.1049 max mem: 3531 +Test: [170/269] eta: 0:00:20 acc1_pos: 72.414 acc5_pos: 100.000 loss: 0.7728 (0.7240) acc1: 48.1771 (54.9007) acc5: 100.0000 (100.0000) acc1_neg: 58.333 acc5_neg: 100.000 time: 0.1961 data: 0.0998 max mem: 3531 +Test: [180/269] eta: 0:00:18 acc1_pos: 72.414 acc5_pos: 100.000 loss: 0.6583 (0.7170) acc1: 58.3333 (55.5968) acc5: 100.0000 (100.0000) acc1_neg: 56.771 acc5_neg: 100.000 time: 0.1772 data: 0.0839 max mem: 3531 +Test: [190/269] eta: 0:00:16 acc1_pos: 72.414 acc5_pos: 100.000 loss: 0.6163 (0.7139) acc1: 63.5417 (55.8941) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.1722 data: 0.0793 max mem: 3531 +Test: [200/269] eta: 0:00:13 acc1_pos: 72.414 acc5_pos: 100.000 loss: 0.6054 (0.7078) acc1: 65.6250 (56.5519) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.1876 data: 0.0910 max mem: 3531 +Test: [210/269] eta: 0:00:11 acc1_pos: 72.414 acc5_pos: 100.000 loss: 0.6400 (0.7091) acc1: 62.2396 (56.4265) acc5: 100.0000 (100.0000) acc1_neg: 56.510 acc5_neg: 100.000 time: 0.1866 data: 0.0886 max mem: 3532 +Test: [220/269] eta: 0:00:09 acc1_pos: 72.414 acc5_pos: 100.000 loss: 0.6671 (0.7079) acc1: 58.3333 (56.6636) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.1800 data: 0.0849 max mem: 3532 +Test: [230/269] eta: 0:00:07 acc1_pos: 72.414 acc5_pos: 100.000 loss: 0.5892 (0.7021) acc1: 67.7083 (57.1880) acc5: 100.0000 (100.0000) acc1_neg: 67.708 acc5_neg: 100.000 time: 0.1830 data: 0.0915 max mem: 3532 +Test: [240/269] eta: 0:00:05 acc1_pos: 72.414 acc5_pos: 100.000 loss: 0.6513 (0.7042) acc1: 61.4583 (56.8908) acc5: 100.0000 (100.0000) acc1_neg: 61.458 acc5_neg: 100.000 time: 0.1822 data: 0.0897 max mem: 3532 +Test: [250/269] eta: 0:00:03 acc1_pos: 72.414 acc5_pos: 100.000 loss: 0.7948 (0.7136) acc1: 43.4896 (56.1597) acc5: 100.0000 (100.0000) acc1_neg: 15.885 acc5_neg: 100.000 time: 0.1873 data: 0.0948 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 72.414 acc5_pos: 100.000 loss: 0.9438 (0.7262) acc1: 32.5521 (55.0946) acc5: 100.0000 (100.0000) acc1_neg: 39.583 acc5_neg: 100.000 time: 0.1692 data: 0.0783 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 72.414 acc5_pos: 100.000 loss: 0.8923 (0.7288) acc1: 32.5521 (54.8421) acc5: 100.0000 (100.0000) acc1_neg: 65.354 acc5_neg: 100.000 time: 0.1449 data: 0.0509 max mem: 3532 +Test: Total time: 0:00:52 (0.1955 s / it) +* Overall Acc@1 54.842 Acc@5 100.000 + target == real Acc@1 69.985 Acc@5 100.000 + target == fake Acc@1 52.775 Acc@5 100.000 + loss 0.729 + PR-AUC 0.9291 ROC-AUC 0.6640 EER 38.64% (thr=0.4517) +Accuracy of the network on the 103293 test images: 54.8% +[rank0]:[W726 04:50:04.560343566 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T04:50:10Z] eval_done tag=fixed500k_real_n23_seed7_resampled +[2026-07-26T04:50:10Z] eval_start tag=fixed500k_fake_n45_seed7_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed7_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T04:50:10Z +hostname=n116-041-023 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed7_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n45_seed7_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 04:50:23.737000 1189681 torch/distributed/run.py:766] +W0726 04:50:23.737000 1189681 torch/distributed/run.py:766] ***************************************** +W0726 04:50:23.737000 1189681 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:50:23.737000 1189681 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 1): env:// +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:50:52.523628722 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank1]:[W726 04:50:52.531685593 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:50:52.906018819 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:50:52.941718575 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:50:52.023835118 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:50:52.219046330 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:50:52.235475145 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:50:52.236072730 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n45_seed7_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed7_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:22:16 acc1_pos: 60.156 acc5_pos: 100.000 loss: 0.6768 (0.6768) acc1: 60.1562 (60.1562) acc5: 100.0000 (100.0000) time: 5.7128 data: 4.1641 max mem: 3531 +Test: [ 10/234] eta: 0:02:14 acc1_pos: 35.938 acc5_pos: 100.000 loss: 0.7224 (0.7081) acc1: 49.7396 (53.2907) acc5: 100.0000 (100.0000) time: 0.6012 data: 0.3788 max mem: 3531 +Test: [ 20/234] eta: 0:01:19 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.7431 (0.7007) acc1: 48.6979 (54.2039) acc5: 100.0000 (100.0000) time: 0.1030 data: 0.0116 max mem: 3531 +Test: [ 30/234] eta: 0:01:03 acc1_pos: 72.656 acc5_pos: 100.000 loss: 0.2520 (0.5612) acc1: 94.0104 (66.2550) acc5: 100.0000 (100.0000) time: 0.1501 data: 0.0562 max mem: 3531 +Test: [ 40/234] eta: 0:00:58 acc1_pos: 99.219 acc5_pos: 100.000 loss: 0.2395 (0.5198) acc1: 94.5312 (69.6265) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1322 max mem: 3531 +Test: [ 50/234] eta: 0:00:52 acc1_pos: 98.177 acc5_pos: 100.000 loss: 0.1399 (0.4441) acc1: 97.9167 (75.2196) acc5: 100.0000 (100.0000) time: 0.2477 data: 0.1565 max mem: 3531 +Test: [ 60/234] eta: 0:00:47 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.1148 (0.3879) acc1: 98.9583 (79.2649) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2249 data: 0.1300 max mem: 3531 +Test: [ 70/234] eta: 0:00:43 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.1013 (0.4003) acc1: 100.0000 (78.5321) acc5: 100.0000 (100.0000) acc1_neg: 60.417 acc5_neg: 100.000 time: 0.2170 data: 0.1221 max mem: 3531 +Test: [ 80/234] eta: 0:00:39 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.6224 (0.4122) acc1: 63.8021 (78.1443) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2052 data: 0.1142 max mem: 3531 +Test: [ 90/234] eta: 0:00:35 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.2461 (0.3875) acc1: 94.0104 (80.0481) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.1781 data: 0.0875 max mem: 3531 +Test: [100/234] eta: 0:00:32 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.2441 (0.3925) acc1: 88.5417 (80.0124) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1925 data: 0.1011 max mem: 3531 +Test: [110/234] eta: 0:00:29 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.3937 (0.3909) acc1: 80.2083 (80.4148) acc5: 100.0000 (100.0000) acc1_neg: 77.344 acc5_neg: 100.000 time: 0.1946 data: 0.1020 max mem: 3531 +Test: [120/234] eta: 0:00:26 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.3676 (0.4008) acc1: 85.9375 (80.1954) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1828 data: 0.0902 max mem: 3531 +Test: [130/234] eta: 0:00:23 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.3827 (0.4009) acc1: 86.1979 (80.6815) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.1899 data: 0.0978 max mem: 3531 +Test: [140/234] eta: 0:00:21 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4214 (0.4258) acc1: 80.9896 (79.4991) acc5: 100.0000 (100.0000) acc1_neg: 34.896 acc5_neg: 100.000 time: 0.1842 data: 0.0921 max mem: 3531 +Test: [150/234] eta: 0:00:18 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4788 (0.4345) acc1: 73.9583 (79.0270) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.1815 data: 0.0904 max mem: 3531 +Test: [160/234] eta: 0:00:16 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4293 (0.4340) acc1: 77.6042 (79.2783) acc5: 100.0000 (100.0000) acc1_neg: 77.083 acc5_neg: 100.000 time: 0.1762 data: 0.0855 max mem: 3531 +Test: [170/234] eta: 0:00:13 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4898 (0.4372) acc1: 78.1250 (79.2474) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.1782 data: 0.0864 max mem: 3531 +Test: [180/234] eta: 0:00:11 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.5063 (0.4459) acc1: 76.3021 (78.7753) acc5: 100.0000 (100.0000) acc1_neg: 39.062 acc5_neg: 100.000 time: 0.1797 data: 0.0881 max mem: 3531 +Test: [190/234] eta: 0:00:09 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.6803 (0.4600) acc1: 62.7604 (77.8141) acc5: 100.0000 (100.0000) acc1_neg: 59.896 acc5_neg: 100.000 time: 0.1849 data: 0.0927 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.6803 (0.4530) acc1: 62.7604 (78.2170) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1938 data: 0.1011 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.0995 (0.4377) acc1: 99.7396 (79.2148) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1766 data: 0.0841 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.1855 (0.4275) acc1: 98.6979 (79.9927) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.1349 data: 0.0431 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.3974 (0.4459) acc1: 53.3854 (78.5748) acc5: 100.0000 (100.0000) acc1_neg: 48.958 acc5_neg: 100.000 time: 0.1241 data: 0.0341 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.8377 (0.4509) acc1: 48.1771 (78.2088) acc5: 100.0000 (100.0000) acc1_neg: 45.136 acc5_neg: 100.000 time: 0.1276 data: 0.0366 max mem: 3532 +Test: Total time: 0:00:47 (0.2033 s / it) +* Overall Acc@1 78.209 Acc@5 100.000 + target == real Acc@1 75.386 Acc@5 100.000 + target == fake Acc@1 79.004 Acc@5 100.000 + loss 0.451 + PR-AUC 0.9616 ROC-AUC 0.8737 EER 22.42% (thr=0.5220) +Accuracy of the network on the 89729 test images: 78.2% +[rank0]:[W726 04:51:50.425260070 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:52:08.879000 1190228 torch/distributed/run.py:766] +W0726 04:52:08.879000 1190228 torch/distributed/run.py:766] ***************************************** +W0726 04:52:08.879000 1190228 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:52:08.879000 1190228 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:52:38.805890073 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:52:38.911757918 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:52:38.077287108 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:52:38.226161890 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:52:38.284696229 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:52:38.313623389 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +NCCL version 2.26.2+cuda12.2 +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:52:39.456724303 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:52:39.491722843 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n45_seed7_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed7_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:07:31 acc1_pos: 26.562 acc5_pos: 100.000 loss: 1.0232 (1.0232) acc1: 26.5625 (26.5625) acc5: 100.0000 (100.0000) time: 6.2217 data: 4.8461 max mem: 3531 +Test: [ 10/2387] eta: 0:25:39 acc1_pos: 10.938 acc5_pos: 100.000 loss: 1.1113 (1.0809) acc1: 16.4062 (20.5256) acc5: 100.0000 (100.0000) time: 0.6476 data: 0.4409 max mem: 3531 +Test: [ 20/2387] eta: 0:15:17 acc1_pos: 15.104 acc5_pos: 100.000 loss: 1.1287 (1.1067) acc1: 15.6250 (19.7297) acc5: 100.0000 (100.0000) time: 0.0957 data: 0.0056 max mem: 3531 +Test: [ 30/2387] eta: 0:12:13 acc1_pos: 4.688 acc5_pos: 100.000 loss: 1.1324 (1.1204) acc1: 15.6250 (19.3884) acc5: 100.0000 (100.0000) time: 0.1262 data: 0.0360 max mem: 3531 +Test: [ 40/2387] eta: 0:11:13 acc1_pos: 10.938 acc5_pos: 100.000 loss: 1.1524 (1.1374) acc1: 14.5833 (17.6639) acc5: 100.0000 (100.0000) time: 0.1814 data: 0.0911 max mem: 3531 +Test: [ 50/2387] eta: 0:10:47 acc1_pos: 31.250 acc5_pos: 100.000 loss: 1.1135 (1.1242) acc1: 14.5833 (18.8113) acc5: 100.0000 (100.0000) time: 0.2243 data: 0.1316 max mem: 3531 +Test: [ 60/2387] eta: 0:10:30 acc1_pos: 42.969 acc5_pos: 100.000 loss: 1.0895 (1.1232) acc1: 22.6562 (19.5654) acc5: 100.0000 (100.0000) time: 0.2384 data: 0.1444 max mem: 3531 +Test: [ 70/2387] eta: 0:10:22 acc1_pos: 27.083 acc5_pos: 100.000 loss: 1.1631 (1.1314) acc1: 19.7917 (19.0581) acc5: 100.0000 (100.0000) time: 0.2469 data: 0.1540 max mem: 3531 +Test: [ 80/2387] eta: 0:10:11 acc1_pos: 55.990 acc5_pos: 100.000 loss: 1.0679 (1.1084) acc1: 21.3542 (20.7948) acc5: 100.0000 (100.0000) time: 0.2474 data: 0.1540 max mem: 3531 +Test: [ 90/2387] eta: 0:10:02 acc1_pos: 46.615 acc5_pos: 100.000 loss: 0.8452 (1.0762) acc1: 36.1979 (23.2515) acc5: 100.0000 (100.0000) time: 0.2394 data: 0.1464 max mem: 3531 +Test: [ 100/2387] eta: 0:09:55 acc1_pos: 39.844 acc5_pos: 100.000 loss: 0.8073 (1.0607) acc1: 39.8438 (24.4044) acc5: 100.0000 (100.0000) time: 0.2400 data: 0.1445 max mem: 3531 +Test: [ 110/2387] eta: 0:09:55 acc1_pos: 34.375 acc5_pos: 100.000 loss: 0.8394 (1.0405) acc1: 39.8438 (26.0370) acc5: 100.0000 (100.0000) time: 0.2589 data: 0.1641 max mem: 3531 +Test: [ 120/2387] eta: 0:09:46 acc1_pos: 30.208 acc5_pos: 100.000 loss: 0.8738 (1.0278) acc1: 40.1042 (27.1393) acc5: 100.0000 (100.0000) time: 0.2523 data: 0.1597 max mem: 3531 +Test: [ 130/2387] eta: 0:09:44 acc1_pos: 35.156 acc5_pos: 100.000 loss: 0.8730 (1.0155) acc1: 40.1042 (28.0137) acc5: 100.0000 (100.0000) time: 0.2443 data: 0.1493 max mem: 3531 +Test: [ 140/2387] eta: 0:09:40 acc1_pos: 39.323 acc5_pos: 100.000 loss: 0.8603 (1.0050) acc1: 39.3229 (28.6643) acc5: 100.0000 (100.0000) time: 0.2556 data: 0.1604 max mem: 3531 +Test: [ 150/2387] eta: 0:09:36 acc1_pos: 12.500 acc5_pos: 100.000 loss: 0.8331 (0.9936) acc1: 38.0208 (29.2857) acc5: 100.0000 (100.0000) time: 0.2499 data: 0.1562 max mem: 3531 +Test: [ 160/2387] eta: 0:09:28 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.7906 (0.9730) acc1: 44.7917 (31.1756) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2341 data: 0.1415 max mem: 3531 +Test: [ 170/2387] eta: 0:09:16 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2821 (0.9315) acc1: 94.5312 (34.9781) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2011 data: 0.1073 max mem: 3531 +Test: [ 180/2387] eta: 0:09:04 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2627 (0.8944) acc1: 96.3542 (38.3705) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1748 data: 0.0802 max mem: 3531 +Test: [ 190/2387] eta: 0:08:53 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2604 (0.8610) acc1: 96.3542 (41.4035) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1736 data: 0.0752 max mem: 3531 +Test: [ 200/2387] eta: 0:08:48 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2588 (0.8310) acc1: 96.3542 (44.1322) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1999 data: 0.1001 max mem: 3531 +Test: [ 210/2387] eta: 0:08:41 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2593 (0.8042) acc1: 96.0938 (46.5800) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2052 data: 0.1091 max mem: 3532 +Test: [ 220/2387] eta: 0:08:36 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2604 (0.7797) acc1: 96.0938 (48.8228) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2009 data: 0.1050 max mem: 3532 +Test: [ 230/2387] eta: 0:08:34 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2609 (0.7574) acc1: 96.3542 (50.8827) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2296 data: 0.1334 max mem: 3532 +Test: [ 240/2387] eta: 0:08:36 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2707 (0.7386) acc1: 95.3125 (52.6625) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2654 data: 0.1698 max mem: 3532 +Test: [ 250/2387] eta: 0:08:24 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2454 (0.7139) acc1: 95.8333 (54.5256) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2049 data: 0.1089 max mem: 3532 +Test: [ 260/2387] eta: 0:08:17 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.0910 (0.6898) acc1: 100.0000 (56.2680) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1543 data: 0.0597 max mem: 3532 +Test: [ 270/2387] eta: 0:08:14 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.0853 (0.6676) acc1: 100.0000 (57.8778) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2057 data: 0.1123 max mem: 3532 +Test: [ 280/2387] eta: 0:08:09 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.0864 (0.6472) acc1: 100.0000 (59.3731) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2147 data: 0.1221 max mem: 3532 +Test: [ 290/2387] eta: 0:08:08 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.0861 (0.6278) acc1: 100.0000 (60.7693) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2201 data: 0.1269 max mem: 3532 +Test: [ 300/2387] eta: 0:08:06 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.0828 (0.6098) acc1: 100.0000 (62.0726) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2392 data: 0.1440 max mem: 3532 +Test: [ 310/2387] eta: 0:08:03 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.0914 (0.5967) acc1: 100.0000 (63.2327) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2298 data: 0.1346 max mem: 3532 +Test: [ 320/2387] eta: 0:08:01 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.3260 (0.5885) acc1: 97.6562 (64.3051) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2367 data: 0.1435 max mem: 3532 +Test: [ 330/2387] eta: 0:07:58 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.3298 (0.5806) acc1: 97.6562 (65.3142) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2317 data: 0.1394 max mem: 3532 +Test: [ 340/2387] eta: 0:07:54 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.3289 (0.5732) acc1: 97.6562 (66.2596) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2101 data: 0.1155 max mem: 3532 +Test: [ 350/2387] eta: 0:07:51 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.3174 (0.5630) acc1: 97.9167 (67.2112) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2138 data: 0.1188 max mem: 3532 +Test: [ 360/2387] eta: 0:07:49 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2070 (0.5534) acc1: 98.9583 (68.0726) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2279 data: 0.1337 max mem: 3532 +Test: [ 370/2387] eta: 0:07:45 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2178 (0.5443) acc1: 98.1771 (68.8890) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2182 data: 0.1241 max mem: 3532 +Test: [ 380/2387] eta: 0:07:38 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2182 (0.5357) acc1: 98.1771 (69.6598) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1766 data: 0.0817 max mem: 3532 +Test: [ 390/2387] eta: 0:07:35 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2165 (0.5276) acc1: 98.4375 (70.3971) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1802 data: 0.0868 max mem: 3532 +Test: [ 400/2387] eta: 0:07:31 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2158 (0.5198) acc1: 98.4375 (71.0905) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2034 data: 0.1107 max mem: 3532 +Test: [ 410/2387] eta: 0:07:28 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2153 (0.5124) acc1: 98.4375 (71.7508) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2026 data: 0.1091 max mem: 3532 +Test: [ 420/2387] eta: 0:07:25 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2139 (0.5053) acc1: 98.4375 (72.3804) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2055 data: 0.1117 max mem: 3532 +Test: [ 430/2387] eta: 0:07:20 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2187 (0.4992) acc1: 97.6562 (72.9584) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1919 data: 0.0990 max mem: 3532 +Test: [ 440/2387] eta: 0:07:16 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2257 (0.4916) acc1: 97.1354 (73.5408) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1798 data: 0.0885 max mem: 3533 +Test: [ 450/2387] eta: 0:07:12 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1072 (0.4831) acc1: 100.0000 (74.1275) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1788 data: 0.0852 max mem: 3533 +Test: [ 460/2387] eta: 0:07:08 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1062 (0.4749) acc1: 100.0000 (74.6887) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1795 data: 0.0866 max mem: 3533 +Test: [ 470/2387] eta: 0:07:04 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1060 (0.4671) acc1: 100.0000 (75.2261) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1828 data: 0.0903 max mem: 3533 +Test: [ 480/2387] eta: 0:07:01 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1071 (0.4596) acc1: 100.0000 (75.7412) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0970 max mem: 3533 +Test: [ 490/2387] eta: 0:06:57 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1071 (0.4524) acc1: 100.0000 (76.2353) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1832 data: 0.0920 max mem: 3533 +Test: [ 500/2387] eta: 0:06:53 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1066 (0.4455) acc1: 100.0000 (76.7091) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1763 data: 0.0835 max mem: 3533 +Test: [ 510/2387] eta: 0:06:50 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1073 (0.4392) acc1: 100.0000 (77.1649) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1845 data: 0.0912 max mem: 3533 +Test: [ 520/2387] eta: 0:06:46 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1531 (0.4339) acc1: 100.0000 (77.6032) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1872 data: 0.0950 max mem: 3533 +Test: [ 530/2387] eta: 0:06:43 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1506 (0.4285) acc1: 100.0000 (78.0250) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1817 data: 0.0897 max mem: 3533 +Test: [ 540/2387] eta: 0:06:39 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1479 (0.4233) acc1: 100.0000 (78.4311) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1734 data: 0.0816 max mem: 3533 +Test: [ 550/2387] eta: 0:06:36 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1479 (0.4183) acc1: 100.0000 (78.8226) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1812 data: 0.0884 max mem: 3533 +Test: [ 560/2387] eta: 0:06:33 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1484 (0.4135) acc1: 100.0000 (79.2001) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1878 data: 0.0948 max mem: 3533 +Test: [ 570/2387] eta: 0:06:30 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1476 (0.4088) acc1: 100.0000 (79.5644) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1836 data: 0.0916 max mem: 3533 +Test: [ 580/2387] eta: 0:06:27 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1463 (0.4043) acc1: 100.0000 (79.9161) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1836 data: 0.0924 max mem: 3533 +Test: [ 590/2387] eta: 0:06:23 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1468 (0.4000) acc1: 100.0000 (80.2555) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1827 data: 0.0922 max mem: 3533 +Test: [ 600/2387] eta: 0:06:20 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1550 (0.3961) acc1: 100.0000 (80.5827) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1805 data: 0.0900 max mem: 3533 +Test: [ 610/2387] eta: 0:06:17 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1943 (0.3938) acc1: 98.6979 (80.8528) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1801 data: 0.0877 max mem: 3533 +Test: [ 620/2387] eta: 0:06:15 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2551 (0.3916) acc1: 97.1354 (81.1125) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1879 data: 0.0943 max mem: 3533 +Test: [ 630/2387] eta: 0:06:12 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2552 (0.3894) acc1: 97.3958 (81.3647) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1945 data: 0.1023 max mem: 3533 +Test: [ 640/2387] eta: 0:06:09 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2544 (0.3874) acc1: 97.1354 (81.6108) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1916 data: 0.1000 max mem: 3533 +Test: [ 650/2387] eta: 0:06:06 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2544 (0.3854) acc1: 97.1354 (81.8472) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1856 data: 0.0942 max mem: 3533 +Test: [ 660/2387] eta: 0:06:03 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2569 (0.3835) acc1: 97.1354 (82.0829) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1813 data: 0.0905 max mem: 3533 +Test: [ 670/2387] eta: 0:06:01 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2556 (0.3816) acc1: 97.1354 (82.3099) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1844 data: 0.0937 max mem: 3534 +Test: [ 680/2387] eta: 0:05:58 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2615 (0.3798) acc1: 97.6562 (82.5551) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1843 data: 0.0933 max mem: 3534 +Test: [ 690/2387] eta: 0:05:55 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2704 (0.3797) acc1: 97.3958 (82.6030) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.1719 data: 0.0805 max mem: 3534 +Test: [ 700/2387] eta: 0:05:52 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.4012 (0.3801) acc1: 81.2500 (82.5837) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.1700 data: 0.0783 max mem: 3534 +Test: [ 710/2387] eta: 0:05:49 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.4012 (0.3804) acc1: 81.2500 (82.5550) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.1766 data: 0.0852 max mem: 3534 +Test: [ 720/2387] eta: 0:05:46 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.4026 (0.3807) acc1: 80.9896 (82.5362) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.1857 data: 0.0946 max mem: 3534 +Test: [ 730/2387] eta: 0:05:44 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.4063 (0.3812) acc1: 80.4688 (82.5086) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.1915 data: 0.1003 max mem: 3534 +Test: [ 740/2387] eta: 0:05:42 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.4038 (0.3815) acc1: 80.4688 (82.4902) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.1900 data: 0.0991 max mem: 3534 +Test: [ 750/2387] eta: 0:05:39 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.3957 (0.3818) acc1: 81.5104 (82.4779) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.1873 data: 0.0958 max mem: 3534 +Test: [ 760/2387] eta: 0:05:36 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.3910 (0.3816) acc1: 81.7708 (82.4908) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1849 data: 0.0927 max mem: 3534 +Test: [ 770/2387] eta: 0:05:34 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.3019 (0.3796) acc1: 88.5417 (82.6328) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1861 data: 0.0933 max mem: 3534 +Test: [ 780/2387] eta: 0:05:31 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1418 (0.3764) acc1: 99.2188 (82.8488) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1819 data: 0.0885 max mem: 3534 +Test: [ 790/2387] eta: 0:05:29 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1235 (0.3738) acc1: 99.4792 (83.0133) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1770 data: 0.0846 max mem: 3534 +Test: [ 800/2387] eta: 0:05:26 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1197 (0.3706) acc1: 99.4792 (83.2231) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0824 max mem: 3534 +Test: [ 810/2387] eta: 0:05:23 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1233 (0.3677) acc1: 99.7396 (83.4232) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1748 data: 0.0828 max mem: 3534 +Test: [ 820/2387] eta: 0:05:21 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1342 (0.3650) acc1: 99.7396 (83.6099) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1736 data: 0.0832 max mem: 3534 +Test: [ 830/2387] eta: 0:05:18 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1869 (0.3636) acc1: 96.6146 (83.7260) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1684 data: 0.0772 max mem: 3534 +Test: [ 840/2387] eta: 0:05:15 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2855 (0.3631) acc1: 94.0104 (83.8322) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1668 data: 0.0742 max mem: 3534 +Test: [ 850/2387] eta: 0:05:12 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.3092 (0.3626) acc1: 91.6667 (83.9276) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1659 data: 0.0737 max mem: 3534 +Test: [ 860/2387] eta: 0:05:09 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.3017 (0.3618) acc1: 93.4896 (84.0447) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1609 data: 0.0694 max mem: 3534 +Test: [ 870/2387] eta: 0:05:07 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2815 (0.3612) acc1: 94.0104 (84.1600) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1588 data: 0.0665 max mem: 3534 +Test: [ 880/2387] eta: 0:05:04 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.3056 (0.3607) acc1: 94.2708 (84.2618) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1617 data: 0.0697 max mem: 3534 +Test: [ 890/2387] eta: 0:05:01 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2858 (0.3598) acc1: 96.0938 (84.3963) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1634 data: 0.0720 max mem: 3534 +Test: [ 900/2387] eta: 0:04:59 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.3316 (0.3636) acc1: 91.6667 (84.1287) acc5: 100.0000 (100.0000) acc1_neg: 53.906 acc5_neg: 100.000 time: 0.1670 data: 0.0754 max mem: 3535 +Test: [ 910/2387] eta: 0:04:56 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.7931 (0.3684) acc1: 51.0417 (83.7698) acc5: 100.0000 (100.0000) acc1_neg: 49.479 acc5_neg: 100.000 time: 0.1738 data: 0.0821 max mem: 3535 +Test: [ 920/2387] eta: 0:04:54 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.8071 (0.3731) acc1: 50.7812 (83.4128) acc5: 100.0000 (100.0000) acc1_neg: 48.698 acc5_neg: 100.000 time: 0.1720 data: 0.0809 max mem: 3535 +Test: [ 930/2387] eta: 0:04:51 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.8055 (0.3776) acc1: 51.5625 (83.0726) acc5: 100.0000 (100.0000) acc1_neg: 46.094 acc5_neg: 100.000 time: 0.1725 data: 0.0819 max mem: 3535 +Test: [ 940/2387] eta: 0:04:49 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.8069 (0.3821) acc1: 51.3021 (82.7378) acc5: 100.0000 (100.0000) acc1_neg: 50.521 acc5_neg: 100.000 time: 0.1910 data: 0.0999 max mem: 3535 +Test: [ 950/2387] eta: 0:04:47 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.7967 (0.3864) acc1: 50.5208 (82.3913) acc5: 100.0000 (100.0000) acc1_neg: 52.865 acc5_neg: 100.000 time: 0.1897 data: 0.0972 max mem: 3535 +Test: [ 960/2387] eta: 0:04:45 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.7848 (0.3906) acc1: 50.2604 (82.0681) acc5: 100.0000 (100.0000) acc1_neg: 54.427 acc5_neg: 100.000 time: 0.1981 data: 0.1061 max mem: 3535 +Test: [ 970/2387] eta: 0:04:43 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.7877 (0.3949) acc1: 50.2604 (81.7301) acc5: 100.0000 (100.0000) acc1_neg: 52.083 acc5_neg: 100.000 time: 0.2050 data: 0.1146 max mem: 3535 +Test: [ 980/2387] eta: 0:04:41 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.7774 (0.3952) acc1: 51.8229 (81.7074) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.1702 data: 0.0789 max mem: 3535 +Test: [ 990/2387] eta: 0:04:39 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2021 (0.3932) acc1: 95.5729 (81.8463) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1765 data: 0.0848 max mem: 3535 +Test: [1000/2387] eta: 0:04:36 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1697 (0.3912) acc1: 98.9583 (81.9948) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1901 data: 0.0994 max mem: 3535 +Test: [1010/2387] eta: 0:04:34 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1925 (0.3895) acc1: 97.6562 (82.1217) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1794 data: 0.0883 max mem: 3535 +Test: [1020/2387] eta: 0:04:32 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1982 (0.3880) acc1: 97.3958 (82.2478) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1857 data: 0.0938 max mem: 3535 +Test: [1030/2387] eta: 0:04:30 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1679 (0.3858) acc1: 98.9583 (82.4079) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1941 data: 0.1031 max mem: 3535 +Test: [1040/2387] eta: 0:04:28 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1414 (0.3834) acc1: 99.4792 (82.5726) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1851 data: 0.0948 max mem: 3535 +Test: [1050/2387] eta: 0:04:25 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1275 (0.3811) acc1: 99.7396 (82.7352) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1752 data: 0.0848 max mem: 3535 +Test: [1060/2387] eta: 0:04:23 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1382 (0.3790) acc1: 99.7396 (82.8918) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1818 data: 0.0910 max mem: 3535 +Test: [1070/2387] eta: 0:04:21 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1493 (0.3771) acc1: 98.9583 (83.0274) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1843 data: 0.0933 max mem: 3535 +Test: [1080/2387] eta: 0:04:19 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2107 (0.3758) acc1: 97.3958 (83.1421) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1815 data: 0.0888 max mem: 3535 +Test: [1090/2387] eta: 0:04:17 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1778 (0.3738) acc1: 97.9167 (83.2894) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1849 data: 0.0906 max mem: 3535 +Test: [1100/2387] eta: 0:04:14 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1563 (0.3718) acc1: 99.4792 (83.4320) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1832 data: 0.0883 max mem: 3535 +Test: [1110/2387] eta: 0:04:12 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1499 (0.3704) acc1: 99.2188 (83.5305) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1851 data: 0.0920 max mem: 3535 +Test: [1120/2387] eta: 0:04:10 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1510 (0.3685) acc1: 98.9583 (83.6674) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1813 data: 0.0902 max mem: 3536 +Test: [1130/2387] eta: 0:04:08 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1568 (0.3667) acc1: 98.9583 (83.8049) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1763 data: 0.0854 max mem: 3536 +Test: [1140/2387] eta: 0:04:06 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1713 (0.3652) acc1: 98.9583 (83.9274) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1792 data: 0.0881 max mem: 3536 +Test: [1150/2387] eta: 0:04:03 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2192 (0.3647) acc1: 97.3958 (84.0191) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1755 data: 0.0830 max mem: 3536 +Test: [1160/2387] eta: 0:04:01 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2885 (0.3639) acc1: 93.2292 (84.1002) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1714 data: 0.0799 max mem: 3536 +Test: [1170/2387] eta: 0:03:59 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2715 (0.3631) acc1: 94.2708 (84.1935) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1827 data: 0.0918 max mem: 3536 +Test: [1180/2387] eta: 0:03:57 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2802 (0.3625) acc1: 94.2708 (84.2700) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1931 data: 0.1018 max mem: 3536 +Test: [1190/2387] eta: 0:03:55 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2581 (0.3617) acc1: 95.0521 (84.3558) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1870 data: 0.0943 max mem: 3536 +Test: [1200/2387] eta: 0:03:53 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2284 (0.3606) acc1: 97.3958 (84.4734) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1864 data: 0.0935 max mem: 3536 +Test: [1210/2387] eta: 0:03:51 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2225 (0.3594) acc1: 99.2188 (84.5907) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1800 data: 0.0885 max mem: 3536 +Test: [1220/2387] eta: 0:03:49 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1180 (0.3573) acc1: 99.7396 (84.7141) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1739 data: 0.0813 max mem: 3536 +Test: [1230/2387] eta: 0:03:47 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.0976 (0.3551) acc1: 99.7396 (84.8370) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1766 data: 0.0842 max mem: 3536 +Test: [1240/2387] eta: 0:03:44 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.0978 (0.3531) acc1: 99.7396 (84.9569) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1791 data: 0.0880 max mem: 3536 +Test: [1250/2387] eta: 0:03:42 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.0984 (0.3510) acc1: 99.7396 (85.0755) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1884 data: 0.0962 max mem: 3536 +Test: [1260/2387] eta: 0:03:40 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.0939 (0.3490) acc1: 99.7396 (85.1926) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0966 max mem: 3536 +Test: [1270/2387] eta: 0:03:38 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.0886 (0.3470) acc1: 100.0000 (85.3085) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1855 data: 0.0921 max mem: 3536 +Test: [1280/2387] eta: 0:03:36 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.0931 (0.3451) acc1: 100.0000 (85.4209) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1773 data: 0.0851 max mem: 3536 +Test: [1290/2387] eta: 0:03:34 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1038 (0.3434) acc1: 99.7396 (85.5153) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1827 data: 0.0919 max mem: 3536 +Test: [1300/2387] eta: 0:03:32 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1058 (0.3416) acc1: 99.7396 (85.6188) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1897 data: 0.0993 max mem: 3536 +Test: [1310/2387] eta: 0:03:30 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1062 (0.3399) acc1: 99.7396 (85.7186) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1769 data: 0.0853 max mem: 3536 +Test: [1320/2387] eta: 0:03:28 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1215 (0.3383) acc1: 99.4792 (85.8131) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1743 data: 0.0817 max mem: 3536 +Test: [1330/2387] eta: 0:03:26 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1279 (0.3369) acc1: 99.4792 (85.9113) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1875 data: 0.0960 max mem: 3536 +Test: [1340/2387] eta: 0:03:24 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1669 (0.3364) acc1: 98.1771 (85.9457) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1902 data: 0.0993 max mem: 3536 +Test: [1350/2387] eta: 0:03:22 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.0749 (0.3344) acc1: 100.0000 (86.0497) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1807 data: 0.0877 max mem: 3537 +Test: [1360/2387] eta: 0:03:20 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.0641 (0.3324) acc1: 100.0000 (86.1522) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1831 data: 0.0892 max mem: 3537 +Test: [1370/2387] eta: 0:03:18 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.0612 (0.3305) acc1: 100.0000 (86.2532) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1910 data: 0.0991 max mem: 3537 +Test: [1380/2387] eta: 0:03:16 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.0636 (0.3285) acc1: 100.0000 (86.3527) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1924 data: 0.1013 max mem: 3537 +Test: [1390/2387] eta: 0:03:14 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.0639 (0.3267) acc1: 100.0000 (86.4508) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1954 data: 0.1042 max mem: 3537 +Test: [1400/2387] eta: 0:03:12 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.0628 (0.3248) acc1: 100.0000 (86.5476) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1881 data: 0.0972 max mem: 3537 +Test: [1410/2387] eta: 0:03:10 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.0663 (0.3235) acc1: 100.0000 (86.6278) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1781 data: 0.0862 max mem: 3537 +Test: [1420/2387] eta: 0:03:08 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1654 (0.3227) acc1: 98.4375 (86.6905) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1816 data: 0.0895 max mem: 3537 +Test: [1430/2387] eta: 0:03:06 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1934 (0.3218) acc1: 97.9167 (86.7655) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1806 data: 0.0896 max mem: 3537 +Test: [1440/2387] eta: 0:03:04 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1838 (0.3209) acc1: 98.4375 (86.8498) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1876 data: 0.0965 max mem: 3537 +Test: [1450/2387] eta: 0:03:02 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1819 (0.3202) acc1: 98.9583 (86.9149) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1901 data: 0.0990 max mem: 3537 +Test: [1460/2387] eta: 0:03:00 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1691 (0.3191) acc1: 98.4375 (86.9938) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1831 data: 0.0911 max mem: 3537 +Test: [1470/2387] eta: 0:02:58 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1385 (0.3178) acc1: 99.4792 (87.0795) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1795 data: 0.0877 max mem: 3537 +Test: [1480/2387] eta: 0:02:56 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1299 (0.3166) acc1: 99.4792 (87.1598) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1766 data: 0.0834 max mem: 3537 +Test: [1490/2387] eta: 0:02:54 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1400 (0.3155) acc1: 99.4792 (87.2429) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1820 data: 0.0888 max mem: 3537 +Test: [1500/2387] eta: 0:02:52 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1463 (0.3143) acc1: 99.4792 (87.3234) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1900 data: 0.0975 max mem: 3537 +Test: [1510/2387] eta: 0:02:50 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1646 (0.3136) acc1: 99.4792 (87.3968) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1991 data: 0.1069 max mem: 3537 +Test: [1520/2387] eta: 0:02:48 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2265 (0.3134) acc1: 97.6562 (87.4456) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.2004 data: 0.1093 max mem: 3537 +Test: [1530/2387] eta: 0:02:46 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2898 (0.3134) acc1: 94.7917 (87.4784) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2027 data: 0.1113 max mem: 3537 +Test: [1540/2387] eta: 0:02:44 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2738 (0.3131) acc1: 95.3125 (87.5303) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2015 data: 0.1106 max mem: 3537 +Test: [1550/2387] eta: 0:02:42 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2706 (0.3129) acc1: 95.5729 (87.5824) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2001 data: 0.1066 max mem: 3537 +Test: [1560/2387] eta: 0:02:40 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2706 (0.3128) acc1: 96.0938 (87.6290) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.2043 data: 0.1107 max mem: 3537 +Test: [1570/2387] eta: 0:02:38 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2520 (0.3123) acc1: 96.8750 (87.6873) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2027 data: 0.1098 max mem: 3537 +Test: [1580/2387] eta: 0:02:36 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2133 (0.3114) acc1: 98.9583 (87.7583) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1936 data: 0.1001 max mem: 3538 +Test: [1590/2387] eta: 0:02:34 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1234 (0.3103) acc1: 99.4792 (87.8324) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1832 data: 0.0917 max mem: 3538 +Test: [1600/2387] eta: 0:02:32 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1200 (0.3094) acc1: 99.7396 (87.8764) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1767 data: 0.0853 max mem: 3538 +Test: [1610/2387] eta: 0:02:30 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1197 (0.3082) acc1: 99.7396 (87.9479) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1789 data: 0.0874 max mem: 3538 +Test: [1620/2387] eta: 0:02:29 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1261 (0.3071) acc1: 99.7396 (88.0176) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1977 data: 0.1066 max mem: 3538 +Test: [1630/2387] eta: 0:02:27 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1406 (0.3063) acc1: 99.2188 (88.0804) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2040 data: 0.1124 max mem: 3538 +Test: [1640/2387] eta: 0:02:24 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1816 (0.3057) acc1: 98.4375 (88.1322) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1699 data: 0.0772 max mem: 3538 +Test: [1650/2387] eta: 0:02:22 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1904 (0.3049) acc1: 97.9167 (88.1904) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1700 data: 0.0778 max mem: 3538 +Test: [1660/2387] eta: 0:02:21 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1773 (0.3042) acc1: 98.1771 (88.2453) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1913 data: 0.0993 max mem: 3538 +Test: [1670/2387] eta: 0:02:18 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1773 (0.3035) acc1: 98.4375 (88.3046) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1546 data: 0.0602 max mem: 3538 +Test: [1680/2387] eta: 0:02:16 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1832 (0.3029) acc1: 98.4375 (88.3558) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1586 data: 0.0649 max mem: 3538 +Test: [1690/2387] eta: 0:02:14 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1855 (0.3023) acc1: 97.6562 (88.4071) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1853 data: 0.0931 max mem: 3538 +Test: [1700/2387] eta: 0:02:12 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1822 (0.3015) acc1: 97.9167 (88.4654) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1670 data: 0.0745 max mem: 3538 +Test: [1710/2387] eta: 0:02:10 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1330 (0.3004) acc1: 98.9583 (88.5292) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1764 data: 0.0826 max mem: 3538 +Test: [1720/2387] eta: 0:02:08 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1066 (0.2993) acc1: 99.4792 (88.5934) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1749 data: 0.0801 max mem: 3538 +Test: [1730/2387] eta: 0:02:06 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1085 (0.2983) acc1: 99.4792 (88.6553) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1708 data: 0.0782 max mem: 3538 +Test: [1740/2387] eta: 0:02:04 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1061 (0.2972) acc1: 99.7396 (88.7171) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1760 data: 0.0845 max mem: 3538 +Test: [1750/2387] eta: 0:02:02 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1010 (0.2961) acc1: 100.0000 (88.7805) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1640 data: 0.0727 max mem: 3538 +Test: [1760/2387] eta: 0:02:00 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1058 (0.2951) acc1: 100.0000 (88.8436) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1730 data: 0.0826 max mem: 3538 +Test: [1770/2387] eta: 0:01:58 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1266 (0.2942) acc1: 99.7396 (88.8981) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1729 data: 0.0807 max mem: 3538 +Test: [1780/2387] eta: 0:01:56 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1301 (0.2933) acc1: 98.9583 (88.9581) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1684 data: 0.0744 max mem: 3538 +Test: [1790/2387] eta: 0:01:54 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1253 (0.2923) acc1: 99.7396 (89.0167) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1634 data: 0.0710 max mem: 3538 +Test: [1800/2387] eta: 0:01:52 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1194 (0.2914) acc1: 99.4792 (89.0729) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1584 data: 0.0674 max mem: 3538 +Test: [1810/2387] eta: 0:01:50 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1187 (0.2904) acc1: 99.4792 (89.1317) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1622 data: 0.0713 max mem: 3539 +Test: [1820/2387] eta: 0:01:48 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1158 (0.2894) acc1: 100.0000 (89.1901) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1676 data: 0.0748 max mem: 3539 +Test: [1830/2387] eta: 0:01:46 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1282 (0.2886) acc1: 100.0000 (89.2480) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1679 data: 0.0740 max mem: 3539 +Test: [1840/2387] eta: 0:01:44 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1542 (0.2879) acc1: 100.0000 (89.3064) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1592 data: 0.0675 max mem: 3539 +Test: [1850/2387] eta: 0:01:42 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1569 (0.2872) acc1: 100.0000 (89.3639) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1492 data: 0.0586 max mem: 3539 +Test: [1860/2387] eta: 0:01:40 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1573 (0.2865) acc1: 100.0000 (89.4207) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1524 data: 0.0607 max mem: 3539 +Test: [1870/2387] eta: 0:01:38 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1571 (0.2858) acc1: 100.0000 (89.4764) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1624 data: 0.0709 max mem: 3539 +Test: [1880/2387] eta: 0:01:36 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1570 (0.2851) acc1: 100.0000 (89.5320) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1581 data: 0.0671 max mem: 3539 +Test: [1890/2387] eta: 0:01:34 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1563 (0.2845) acc1: 100.0000 (89.5871) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1493 data: 0.0566 max mem: 3539 +Test: [1900/2387] eta: 0:01:32 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1587 (0.2838) acc1: 100.0000 (89.6417) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1540 data: 0.0612 max mem: 3539 +Test: [1910/2387] eta: 0:01:30 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1777 (0.2833) acc1: 100.0000 (89.6933) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1627 data: 0.0692 max mem: 3539 +Test: [1920/2387] eta: 0:01:28 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1824 (0.2829) acc1: 99.4792 (89.7182) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1667 data: 0.0725 max mem: 3539 +Test: [1930/2387] eta: 0:01:26 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1665 (0.2823) acc1: 99.2188 (89.7580) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1671 data: 0.0722 max mem: 3539 +Test: [1940/2387] eta: 0:01:24 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1608 (0.2820) acc1: 99.2188 (89.7734) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1730 data: 0.0753 max mem: 3539 +Test: [1950/2387] eta: 0:01:22 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1820 (0.2816) acc1: 97.9167 (89.8065) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1699 data: 0.0735 max mem: 3539 +Test: [1960/2387] eta: 0:01:20 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1820 (0.2812) acc1: 97.3958 (89.8461) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1599 data: 0.0656 max mem: 3539 +Test: [1970/2387] eta: 0:01:18 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2029 (0.2812) acc1: 97.3958 (89.8579) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1623 data: 0.0690 max mem: 3539 +Test: [1980/2387] eta: 0:01:16 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2390 (0.2811) acc1: 93.7500 (89.8787) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1581 data: 0.0659 max mem: 3539 +Test: [1990/2387] eta: 0:01:15 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2241 (0.2809) acc1: 97.3958 (89.9154) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1675 data: 0.0745 max mem: 3539 +Test: [2000/2387] eta: 0:01:13 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2271 (0.2806) acc1: 97.3958 (89.9523) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1720 data: 0.0794 max mem: 3539 +Test: [2010/2387] eta: 0:01:11 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2285 (0.2804) acc1: 97.3958 (89.9878) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1568 data: 0.0651 max mem: 3539 +Test: [2020/2387] eta: 0:01:09 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2306 (0.2801) acc1: 97.1354 (90.0231) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1616 data: 0.0697 max mem: 3539 +Test: [2030/2387] eta: 0:01:07 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2300 (0.2799) acc1: 97.1354 (90.0578) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1687 data: 0.0775 max mem: 3540 +Test: [2040/2387] eta: 0:01:05 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2279 (0.2796) acc1: 97.1354 (90.0941) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1646 data: 0.0743 max mem: 3540 +Test: [2050/2387] eta: 0:01:03 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2259 (0.2793) acc1: 97.1354 (90.1296) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1671 data: 0.0744 max mem: 3540 +Test: [2060/2387] eta: 0:01:01 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2220 (0.2791) acc1: 97.3958 (90.1652) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1732 data: 0.0804 max mem: 3540 +Test: [2070/2387] eta: 0:00:59 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2194 (0.2788) acc1: 97.6562 (90.2030) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1671 data: 0.0764 max mem: 3540 +Test: [2080/2387] eta: 0:00:57 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2133 (0.2785) acc1: 97.9167 (90.2397) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1650 data: 0.0743 max mem: 3540 +Test: [2090/2387] eta: 0:00:55 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2145 (0.2782) acc1: 97.6562 (90.2760) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1623 data: 0.0714 max mem: 3540 +Test: [2100/2387] eta: 0:00:53 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2154 (0.2779) acc1: 97.9167 (90.3123) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1556 data: 0.0641 max mem: 3540 +Test: [2110/2387] eta: 0:00:51 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2157 (0.2776) acc1: 97.9167 (90.3489) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1633 data: 0.0719 max mem: 3540 +Test: [2120/2387] eta: 0:00:50 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2017 (0.2771) acc1: 97.9167 (90.3866) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1709 data: 0.0788 max mem: 3540 +Test: [2130/2387] eta: 0:00:48 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1454 (0.2764) acc1: 99.7396 (90.4292) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1642 data: 0.0724 max mem: 3540 +Test: [2140/2387] eta: 0:00:46 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1333 (0.2760) acc1: 99.7396 (90.4530) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1611 data: 0.0701 max mem: 3540 +Test: [2150/2387] eta: 0:00:44 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1395 (0.2754) acc1: 99.2188 (90.4945) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1672 data: 0.0754 max mem: 3540 +Test: [2160/2387] eta: 0:00:42 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1445 (0.2748) acc1: 99.2188 (90.5335) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1670 data: 0.0751 max mem: 3540 +Test: [2170/2387] eta: 0:00:40 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1606 (0.2745) acc1: 98.9583 (90.5629) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1673 data: 0.0731 max mem: 3540 +Test: [2180/2387] eta: 0:00:38 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2483 (0.2753) acc1: 95.0521 (90.5185) acc5: 100.0000 (100.0000) acc1_neg: 60.677 acc5_neg: 100.000 time: 0.1680 data: 0.0740 max mem: 3540 +Test: [2190/2387] eta: 0:00:36 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.3055 (0.2756) acc1: 87.7604 (90.4969) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0780 max mem: 3540 +Test: [2200/2387] eta: 0:00:34 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1872 (0.2750) acc1: 98.1771 (90.5366) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1719 data: 0.0790 max mem: 3540 +Test: [2210/2387] eta: 0:00:33 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1531 (0.2745) acc1: 99.2188 (90.5733) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1714 data: 0.0789 max mem: 3540 +Test: [2220/2387] eta: 0:00:31 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1587 (0.2740) acc1: 99.2188 (90.6079) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1701 data: 0.0762 max mem: 3540 +Test: [2230/2387] eta: 0:00:29 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1540 (0.2735) acc1: 99.4792 (90.6429) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1759 data: 0.0806 max mem: 3540 +Test: [2240/2387] eta: 0:00:27 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1357 (0.2730) acc1: 99.2188 (90.6750) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1913 data: 0.0986 max mem: 3540 +Test: [2250/2387] eta: 0:00:25 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1882 (0.2728) acc1: 97.9167 (90.7023) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1773 data: 0.0852 max mem: 3540 +Test: [2260/2387] eta: 0:00:23 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1978 (0.2723) acc1: 99.2188 (90.7397) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1780 data: 0.0852 max mem: 3541 +Test: [2270/2387] eta: 0:00:21 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1260 (0.2716) acc1: 100.0000 (90.7800) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0963 max mem: 3541 +Test: [2280/2387] eta: 0:00:19 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1177 (0.2712) acc1: 100.0000 (90.8014) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1725 data: 0.0757 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1162 (0.2705) acc1: 100.0000 (90.8413) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1899 data: 0.0943 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1205 (0.2700) acc1: 100.0000 (90.8783) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2129 data: 0.1206 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1385 (0.2695) acc1: 100.0000 (90.9120) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1999 data: 0.1066 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1747 (0.2692) acc1: 98.9583 (90.9361) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1792 data: 0.0851 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2064 (0.2690) acc1: 97.6562 (90.9603) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1753 data: 0.0835 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1856 (0.2687) acc1: 97.9167 (90.9894) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1706 data: 0.0797 max mem: 3541 +Test: [2350/2387] eta: 0:00:06 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1753 (0.2686) acc1: 98.4375 (90.9981) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1563 data: 0.0652 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.1918 (0.2685) acc1: 98.4375 (91.0168) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1610 data: 0.0698 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2013 (0.2683) acc1: 98.4375 (91.0355) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1754 data: 0.0841 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2422 (0.2684) acc1: 93.4896 (91.0370) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.1738 data: 0.0832 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 46.575 acc5_pos: 100.000 loss: 0.2642 (0.2684) acc1: 91.9271 (91.0401) acc5: 100.0000 (100.0000) acc1_neg: 90.698 acc5_neg: 100.000 time: 0.1608 data: 0.0724 max mem: 3541 +Test: Total time: 0:07:24 (0.1863 s / it) +* Overall Acc@1 91.040 Acc@5 100.000 + target == real Acc@1 29.762 Acc@5 100.000 + target == fake Acc@1 95.343 Acc@5 100.000 + loss 0.268 + PR-AUC 0.9899 ROC-AUC 0.8718 EER 20.64% (thr=0.7490) +Accuracy of the network on the 916267 test images: 91.0% +[rank0]:[W726 05:00:39.230620673 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:01:00.629000 1190859 torch/distributed/run.py:766] +W0726 05:01:00.629000 1190859 torch/distributed/run.py:766] ***************************************** +W0726 05:01:00.629000 1190859 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:01:00.629000 1190859 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:01:30.246743316 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:01:31.508578271 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:01:31.755146156 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:01:31.818747495 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:01:31.914423222 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +NCCL version 2.26.2+cuda12.2 +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:01:31.936886519 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +[rank6]:[W726 05:01:31.963025116 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:01:31.971389761 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n45_seed7_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed7_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:07:03 acc1_pos: 26.562 acc5_pos: 100.000 loss: 1.0232 (1.0232) acc1: 26.5625 (26.5625) acc5: 100.0000 (100.0000) time: 7.7814 data: 5.7602 max mem: 3531 +Test: [ 10/517] eta: 0:06:40 acc1_pos: 10.938 acc5_pos: 100.000 loss: 1.1113 (1.0809) acc1: 16.4062 (20.5256) acc5: 100.0000 (100.0000) time: 0.7893 data: 0.5240 max mem: 3531 +Test: [ 20/517] eta: 0:03:47 acc1_pos: 15.104 acc5_pos: 100.000 loss: 1.1287 (1.1067) acc1: 15.6250 (19.7297) acc5: 100.0000 (100.0000) time: 0.0922 data: 0.0003 max mem: 3531 +Test: [ 30/517] eta: 0:03:10 acc1_pos: 4.688 acc5_pos: 100.000 loss: 1.1324 (1.1204) acc1: 15.6250 (19.3884) acc5: 100.0000 (100.0000) time: 0.1731 data: 0.0806 max mem: 3531 +Test: [ 40/517] eta: 0:02:51 acc1_pos: 10.938 acc5_pos: 100.000 loss: 1.1524 (1.1374) acc1: 14.5833 (17.6639) acc5: 100.0000 (100.0000) time: 0.2550 data: 0.1611 max mem: 3531 +Test: [ 50/517] eta: 0:02:42 acc1_pos: 36.697 acc5_pos: 100.000 loss: 1.1135 (1.0888) acc1: 14.5833 (22.1150) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2790 data: 0.1838 max mem: 3531 +Test: [ 60/517] eta: 0:02:30 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.3105 (0.9595) acc1: 93.4896 (34.4390) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2709 data: 0.1780 max mem: 3531 +Test: [ 70/517] eta: 0:02:23 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2794 (0.8636) acc1: 98.1771 (43.3759) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2499 data: 0.1582 max mem: 3531 +Test: [ 80/517] eta: 0:02:18 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2697 (0.7927) acc1: 98.4375 (50.0193) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2742 data: 0.1829 max mem: 3531 +Test: [ 90/517] eta: 0:02:13 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2934 (0.7372) acc1: 97.9167 (55.3257) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2849 data: 0.1919 max mem: 3531 +Test: [100/517] eta: 0:02:08 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2404 (0.6874) acc1: 99.7396 (59.6870) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2696 data: 0.1753 max mem: 3531 +Test: [110/517] eta: 0:02:03 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2123 (0.6439) acc1: 100.0000 (63.3071) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2609 data: 0.1674 max mem: 3531 +Test: [120/517] eta: 0:01:59 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2376 (0.6128) acc1: 99.7396 (66.1781) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2649 data: 0.1716 max mem: 3531 +Test: [130/517] eta: 0:01:54 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2616 (0.5855) acc1: 98.6979 (68.6287) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2592 data: 0.1668 max mem: 3531 +Test: [140/517] eta: 0:01:51 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2440 (0.5602) acc1: 100.0000 (70.8537) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2606 data: 0.1695 max mem: 3531 +Test: [150/517] eta: 0:01:47 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2254 (0.5375) acc1: 100.0000 (72.7770) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2714 data: 0.1797 max mem: 3531 +Test: [160/517] eta: 0:01:44 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2460 (0.5202) acc1: 100.0000 (74.4452) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2792 data: 0.1845 max mem: 3531 +Test: [170/517] eta: 0:01:40 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2537 (0.5037) acc1: 99.7396 (75.9183) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2680 data: 0.1719 max mem: 3531 +Test: [180/517] eta: 0:01:36 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2372 (0.4885) acc1: 99.4792 (77.1812) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2449 data: 0.1491 max mem: 3531 +Test: [190/517] eta: 0:01:33 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.1808 (0.4724) acc1: 100.0000 (78.3745) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2402 data: 0.1466 max mem: 3531 +Test: [200/517] eta: 0:01:30 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.1870 (0.4587) acc1: 100.0000 (79.4504) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2551 data: 0.1644 max mem: 3531 +Test: [210/517] eta: 0:01:27 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2114 (0.4511) acc1: 99.2188 (80.2157) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2746 data: 0.1818 max mem: 3532 +Test: [220/517] eta: 0:01:24 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2913 (0.4439) acc1: 96.0938 (80.9425) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2796 data: 0.1848 max mem: 3532 +Test: [230/517] eta: 0:01:21 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2911 (0.4370) acc1: 97.1354 (81.7066) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2649 data: 0.1698 max mem: 3532 +Test: [240/517] eta: 0:01:17 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2568 (0.4283) acc1: 99.7396 (82.4592) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2453 data: 0.1501 max mem: 3532 +Test: [250/517] eta: 0:01:14 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2255 (0.4204) acc1: 99.7396 (83.1476) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2549 data: 0.1566 max mem: 3532 +Test: [260/517] eta: 0:01:11 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2261 (0.4130) acc1: 99.7396 (83.7803) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2549 data: 0.1582 max mem: 3532 +Test: [270/517] eta: 0:01:08 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2364 (0.4070) acc1: 99.4792 (84.3529) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2549 data: 0.1624 max mem: 3532 +Test: [280/517] eta: 0:01:05 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2366 (0.4010) acc1: 99.4792 (84.8727) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2646 data: 0.1711 max mem: 3532 +Test: [290/517] eta: 0:01:02 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2324 (0.3953) acc1: 99.4792 (85.3674) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2649 data: 0.1703 max mem: 3532 +Test: [300/517] eta: 0:01:00 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2293 (0.3898) acc1: 99.7396 (85.8475) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2649 data: 0.1692 max mem: 3532 +Test: [310/517] eta: 0:00:57 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2446 (0.3857) acc1: 100.0000 (86.2976) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2652 data: 0.1702 max mem: 3532 +Test: [320/517] eta: 0:00:54 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2446 (0.3809) acc1: 100.0000 (86.7188) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2699 data: 0.1758 max mem: 3532 +Test: [330/517] eta: 0:00:51 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2175 (0.3755) acc1: 100.0000 (87.1169) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2746 data: 0.1798 max mem: 3532 +Test: [340/517] eta: 0:00:48 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.1989 (0.3717) acc1: 100.0000 (87.4191) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.2699 data: 0.1732 max mem: 3532 +Test: [350/517] eta: 0:00:46 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2591 (0.3695) acc1: 98.1771 (87.7115) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2701 data: 0.1744 max mem: 3532 +Test: [360/517] eta: 0:00:43 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2722 (0.3667) acc1: 98.1771 (88.0006) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2749 data: 0.1791 max mem: 3532 +Test: [370/517] eta: 0:00:40 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2374 (0.3625) acc1: 99.2188 (88.3192) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2600 data: 0.1646 max mem: 3532 +Test: [380/517] eta: 0:00:37 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2076 (0.3591) acc1: 100.0000 (88.6073) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2599 data: 0.1640 max mem: 3532 +Test: [390/517] eta: 0:00:34 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2351 (0.3564) acc1: 99.4792 (88.8780) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2695 data: 0.1691 max mem: 3532 +Test: [400/517] eta: 0:00:32 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2438 (0.3536) acc1: 99.7396 (89.1463) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2649 data: 0.1647 max mem: 3532 +Test: [410/517] eta: 0:00:29 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2719 (0.3522) acc1: 100.0000 (89.3774) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2599 data: 0.1615 max mem: 3532 +Test: [420/517] eta: 0:00:26 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2580 (0.3497) acc1: 100.0000 (89.6229) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2600 data: 0.1638 max mem: 3532 +Test: [430/517] eta: 0:00:23 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2448 (0.3476) acc1: 100.0000 (89.8425) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2599 data: 0.1655 max mem: 3532 +Test: [440/517] eta: 0:00:21 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2609 (0.3457) acc1: 99.2188 (90.0557) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2548 data: 0.1600 max mem: 3533 +Test: [450/517] eta: 0:00:18 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2516 (0.3435) acc1: 99.7396 (90.2716) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2549 data: 0.1592 max mem: 3533 +Test: [460/517] eta: 0:00:15 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2450 (0.3415) acc1: 99.7396 (90.4736) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2599 data: 0.1631 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2411 (0.3393) acc1: 100.0000 (90.6714) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1581 max mem: 3533 +Test: [480/517] eta: 0:00:10 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2225 (0.3369) acc1: 100.0000 (90.8627) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2289 data: 0.1336 max mem: 3533 +Test: [490/517] eta: 0:00:07 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2391 (0.3349) acc1: 100.0000 (91.0360) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1935 data: 0.1018 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2591 (0.3343) acc1: 99.4792 (91.1781) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1820 data: 0.0923 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2674 (0.3331) acc1: 97.9167 (91.3120) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1787 data: 0.0891 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2378 (0.3318) acc1: 99.2188 (91.4039) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1768 data: 0.0853 max mem: 3533 +Test: Total time: 0:02:17 (0.2653 s / it) +* Overall Acc@1 91.404 Acc@5 100.000 + target == real Acc@1 18.295 Acc@5 100.000 + target == fake Acc@1 98.988 Acc@5 100.000 + loss 0.332 + PR-AUC 0.9663 ROC-AUC 0.7925 EER 27.84% (thr=0.7544) +Accuracy of the network on the 198427 test images: 91.4% +[rank0]:[W726 05:03:58.818364620 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 05:04:19.439000 1191432 torch/distributed/run.py:766] +W0726 05:04:19.439000 1191432 torch/distributed/run.py:766] ***************************************** +W0726 05:04:19.439000 1191432 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:04:19.439000 1191432 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 05:04:46.249826868 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:04:47.427462453 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:04:47.935397621 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:04:47.961902092 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:04:47.144757064 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W726 05:04:47.144797068 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:04:47.376975584 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:04:48.464763884 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n45_seed7_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed7_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:45:30 acc1_pos: 89.583 acc5_pos: 100.000 loss: 0.4589 (0.4589) acc1: 89.5833 (89.5833) acc5: 100.0000 (100.0000) time: 7.0203 data: 5.4167 max mem: 3531 +Test: [ 10/389] eta: 0:04:35 acc1_pos: 94.271 acc5_pos: 100.000 loss: 0.3806 (0.4321) acc1: 94.0104 (85.7244) acc5: 100.0000 (100.0000) time: 0.7262 data: 0.4928 max mem: 3531 +Test: [ 20/389] eta: 0:02:37 acc1_pos: 66.406 acc5_pos: 100.000 loss: 0.5129 (0.4893) acc1: 77.8646 (79.8611) acc5: 100.0000 (100.0000) time: 0.0984 data: 0.0004 max mem: 3531 +Test: [ 30/389] eta: 0:02:10 acc1_pos: 92.448 acc5_pos: 100.000 loss: 0.4391 (0.4672) acc1: 92.4479 (84.3834) acc5: 100.0000 (100.0000) time: 0.1628 data: 0.0651 max mem: 3531 +Test: [ 40/389] eta: 0:01:52 acc1_pos: 76.042 acc5_pos: 100.000 loss: 0.4322 (0.4661) acc1: 93.4896 (84.5020) acc5: 100.0000 (100.0000) time: 0.2121 data: 0.1185 max mem: 3531 +Test: [ 50/389] eta: 0:01:38 acc1_pos: 73.438 acc5_pos: 100.000 loss: 0.4494 (0.4739) acc1: 86.7188 (84.3035) acc5: 100.0000 (100.0000) time: 0.1771 data: 0.0855 max mem: 3531 +Test: [ 60/389] eta: 0:01:30 acc1_pos: 94.792 acc5_pos: 100.000 loss: 0.5058 (0.4725) acc1: 82.5521 (84.1573) acc5: 100.0000 (100.0000) time: 0.1736 data: 0.0817 max mem: 3531 +Test: [ 70/389] eta: 0:01:23 acc1_pos: 91.667 acc5_pos: 100.000 loss: 0.3616 (0.4575) acc1: 95.8333 (85.6037) acc5: 100.0000 (100.0000) time: 0.1906 data: 0.0966 max mem: 3531 +Test: [ 80/389] eta: 0:01:17 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.3639 (0.4581) acc1: 91.6667 (85.3942) acc5: 100.0000 (100.0000) time: 0.1844 data: 0.0900 max mem: 3531 +Test: [ 90/389] eta: 0:01:12 acc1_pos: 97.135 acc5_pos: 100.000 loss: 0.3100 (0.4410) acc1: 94.0104 (86.6358) acc5: 100.0000 (100.0000) time: 0.1749 data: 0.0823 max mem: 3531 +Test: [100/389] eta: 0:01:09 acc1_pos: 97.396 acc5_pos: 100.000 loss: 0.3752 (0.4461) acc1: 95.5729 (86.0948) acc5: 100.0000 (100.0000) time: 0.1949 data: 0.1026 max mem: 3531 +Test: [110/389] eta: 0:01:05 acc1_pos: 96.094 acc5_pos: 100.000 loss: 0.4327 (0.4562) acc1: 88.8021 (84.4477) acc5: 100.0000 (100.0000) time: 0.2010 data: 0.1088 max mem: 3531 +Test: [120/389] eta: 0:01:01 acc1_pos: 82.292 acc5_pos: 100.000 loss: 0.3787 (0.4515) acc1: 91.6667 (84.7990) acc5: 100.0000 (100.0000) time: 0.1773 data: 0.0833 max mem: 3531 +Test: [130/389] eta: 0:00:59 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.5138 (0.4738) acc1: 69.7917 (82.0472) acc5: 100.0000 (100.0000) acc1_neg: 47.396 acc5_neg: 100.000 time: 0.1892 data: 0.0954 max mem: 3531 +Test: [140/389] eta: 0:00:56 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.7159 (0.4955) acc1: 48.6979 (79.2590) acc5: 100.0000 (100.0000) acc1_neg: 10.938 acc5_neg: 100.000 time: 0.1986 data: 0.1058 max mem: 3531 +Test: [150/389] eta: 0:00:52 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.7262 (0.5117) acc1: 44.2708 (77.1075) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1711 data: 0.0787 max mem: 3531 +Test: [160/389] eta: 0:00:49 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.7262 (0.5201) acc1: 46.6146 (76.5253) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1503 data: 0.0586 max mem: 3531 +Test: [170/389] eta: 0:00:46 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.4382 (0.5262) acc1: 95.8333 (76.1467) acc5: 100.0000 (100.0000) acc1_neg: 19.531 acc5_neg: 100.000 time: 0.1650 data: 0.0721 max mem: 3531 +Test: [180/389] eta: 0:00:44 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.6496 (0.5376) acc1: 62.7604 (74.8489) acc5: 100.0000 (100.0000) acc1_neg: 42.188 acc5_neg: 100.000 time: 0.1682 data: 0.0747 max mem: 3531 +Test: [190/389] eta: 0:00:41 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.7691 (0.5564) acc1: 42.1875 (72.7735) acc5: 100.0000 (100.0000) acc1_neg: 0.260 acc5_neg: 100.000 time: 0.1589 data: 0.0647 max mem: 3531 +Test: [200/389] eta: 0:00:38 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.7578 (0.5699) acc1: 44.0104 (71.4125) acc5: 100.0000 (100.0000) acc1_neg: 48.958 acc5_neg: 100.000 time: 0.1549 data: 0.0617 max mem: 3531 +Test: [210/389] eta: 0:00:36 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.6392 (0.5675) acc1: 52.3438 (71.7676) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1619 data: 0.0687 max mem: 3532 +Test: [220/389] eta: 0:00:34 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.4333 (0.5640) acc1: 95.5729 (71.9563) acc5: 100.0000 (100.0000) acc1_neg: 29.948 acc5_neg: 100.000 time: 0.1623 data: 0.0695 max mem: 3532 +Test: [230/389] eta: 0:00:31 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.6872 (0.5763) acc1: 51.8229 (70.4591) acc5: 100.0000 (100.0000) acc1_neg: 64.844 acc5_neg: 100.000 time: 0.1568 data: 0.0661 max mem: 3532 +Test: [240/389] eta: 0:00:29 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.6874 (0.5805) acc1: 51.5625 (70.0715) acc5: 100.0000 (100.0000) acc1_neg: 48.177 acc5_neg: 100.000 time: 0.1552 data: 0.0640 max mem: 3532 +Test: [250/389] eta: 0:00:27 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.5835 (0.5814) acc1: 69.7917 (70.1672) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1509 data: 0.0594 max mem: 3532 +Test: [260/389] eta: 0:00:25 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.6176 (0.5877) acc1: 70.8333 (69.6570) acc5: 100.0000 (100.0000) acc1_neg: 6.771 acc5_neg: 100.000 time: 0.1625 data: 0.0711 max mem: 3532 +Test: [270/389] eta: 0:00:23 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.8050 (0.6064) acc1: 31.5104 (67.8630) acc5: 100.0000 (100.0000) acc1_neg: 32.812 acc5_neg: 100.000 time: 0.1594 data: 0.0685 max mem: 3532 +Test: [280/389] eta: 0:00:20 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.8801 (0.6183) acc1: 28.6458 (66.6018) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1548 data: 0.0639 max mem: 3532 +Test: [290/389] eta: 0:00:18 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.7427 (0.6178) acc1: 46.3542 (66.6166) acc5: 100.0000 (100.0000) acc1_neg: 68.750 acc5_neg: 100.000 time: 0.1644 data: 0.0731 max mem: 3532 +Test: [300/389] eta: 0:00:16 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.7387 (0.6231) acc1: 43.2292 (65.6259) acc5: 100.0000 (100.0000) acc1_neg: 43.229 acc5_neg: 100.000 time: 0.1651 data: 0.0742 max mem: 3532 +Test: [310/389] eta: 0:00:14 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.7967 (0.6329) acc1: 32.8125 (64.5390) acc5: 100.0000 (100.0000) acc1_neg: 2.604 acc5_neg: 100.000 time: 0.1648 data: 0.0719 max mem: 3532 +Test: [320/389] eta: 0:00:13 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.8324 (0.6362) acc1: 35.6771 (64.2264) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1750 data: 0.0815 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 72.393 acc5_pos: 100.000 loss: 1.0672 (0.6535) acc1: 8.0729 (62.5519) acc5: 100.0000 (100.0000) acc1_neg: 8.073 acc5_neg: 100.000 time: 0.1770 data: 0.0841 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.6119 (0.6490) acc1: 61.7188 (63.3927) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.1855 data: 0.0931 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.5080 (0.6569) acc1: 82.0312 (62.8517) acc5: 100.0000 (100.0000) acc1_neg: 16.927 acc5_neg: 100.000 time: 0.1842 data: 0.0935 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.8407 (0.6627) acc1: 32.5521 (61.9496) acc5: 100.0000 (100.0000) acc1_neg: 37.240 acc5_neg: 100.000 time: 0.1764 data: 0.0853 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.8279 (0.6651) acc1: 33.8542 (61.6830) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.1569 data: 0.0656 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.7128 (0.6667) acc1: 54.1667 (61.5479) acc5: 100.0000 (100.0000) acc1_neg: 41.927 acc5_neg: 100.000 time: 0.1388 data: 0.0471 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.6254 (0.6665) acc1: 69.5312 (61.6104) acc5: 100.0000 (100.0000) acc1_neg: 79.508 acc5_neg: 100.000 time: 0.1434 data: 0.0483 max mem: 3532 +Test: Total time: 0:01:12 (0.1853 s / it) +* Overall Acc@1 61.610 Acc@5 100.000 + target == real Acc@1 84.216 Acc@5 100.000 + target == fake Acc@1 51.103 Acc@5 100.000 + loss 0.667 + PR-AUC 0.8549 ROC-AUC 0.7526 EER 30.66% (thr=0.4062) +Accuracy of the network on the 149358 test images: 61.6% +[rank0]:[W726 05:06:11.348375177 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:06:32.430000 1191990 torch/distributed/run.py:766] +W0726 05:06:32.430000 1191990 torch/distributed/run.py:766] ***************************************** +W0726 05:06:32.430000 1191990 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:06:32.430000 1191990 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:07:02.108453721 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:07:02.309615566 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:07:03.442437748 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:07:03.549909031 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:07:03.640820166 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:07:03.752571959 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:07:03.805068541 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:07:03.835373596 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n45_seed7_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed7_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:39:00 acc1_pos: 83.073 acc5_pos: 100.000 loss: 0.4309 (0.4309) acc1: 83.0729 (83.0729) acc5: 100.0000 (100.0000) time: 7.0285 data: 5.5057 max mem: 3531 +Test: [ 10/333] eta: 0:04:07 acc1_pos: 87.760 acc5_pos: 100.000 loss: 0.3289 (0.3366) acc1: 83.3333 (85.7955) acc5: 100.0000 (100.0000) time: 0.7659 data: 0.5345 max mem: 3531 +Test: [ 20/333] eta: 0:02:20 acc1_pos: 95.573 acc5_pos: 100.000 loss: 0.3233 (0.3263) acc1: 87.7604 (86.5451) acc5: 100.0000 (100.0000) time: 0.1198 data: 0.0191 max mem: 3531 +Test: [ 30/333] eta: 0:01:54 acc1_pos: 54.167 acc5_pos: 100.000 loss: 0.3906 (0.4693) acc1: 74.4792 (75.0756) acc5: 100.0000 (100.0000) time: 0.1660 data: 0.0702 max mem: 3531 +Test: [ 40/333] eta: 0:01:40 acc1_pos: 51.823 acc5_pos: 100.000 loss: 0.7831 (0.5483) acc1: 47.9167 (68.1402) acc5: 100.0000 (100.0000) time: 0.2316 data: 0.1399 max mem: 3531 +Test: [ 50/333] eta: 0:01:28 acc1_pos: 46.354 acc5_pos: 100.000 loss: 0.7882 (0.5966) acc1: 46.3542 (64.1646) acc5: 100.0000 (100.0000) time: 0.2089 data: 0.1142 max mem: 3531 +Test: [ 60/333] eta: 0:01:20 acc1_pos: 41.927 acc5_pos: 100.000 loss: 0.7920 (0.6294) acc1: 46.8750 (61.5779) acc5: 100.0000 (100.0000) time: 0.1932 data: 0.0931 max mem: 3531 +Test: [ 70/333] eta: 0:01:14 acc1_pos: 39.323 acc5_pos: 100.000 loss: 0.7920 (0.6530) acc1: 46.8750 (59.7675) acc5: 100.0000 (100.0000) time: 0.2054 data: 0.1055 max mem: 3531 +Test: [ 80/333] eta: 0:01:10 acc1_pos: 52.865 acc5_pos: 100.000 loss: 0.7667 (0.6702) acc1: 51.0417 (58.2272) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1280 max mem: 3531 +Test: [ 90/333] eta: 0:01:04 acc1_pos: 61.719 acc5_pos: 100.000 loss: 0.7269 (0.6713) acc1: 52.8646 (58.2418) acc5: 100.0000 (100.0000) time: 0.2144 data: 0.1172 max mem: 3531 +Test: [100/333] eta: 0:00:58 acc1_pos: 65.625 acc5_pos: 100.000 loss: 0.6662 (0.6739) acc1: 58.5938 (58.1657) acc5: 100.0000 (100.0000) time: 0.1449 data: 0.0458 max mem: 3531 +Test: [110/333] eta: 0:00:54 acc1_pos: 60.938 acc5_pos: 100.000 loss: 0.6963 (0.6770) acc1: 54.9479 (57.8430) acc5: 100.0000 (100.0000) time: 0.1371 data: 0.0413 max mem: 3531 +Test: [120/333] eta: 0:00:51 acc1_pos: 63.281 acc5_pos: 100.000 loss: 0.6780 (0.6770) acc1: 55.4688 (57.8060) acc5: 100.0000 (100.0000) time: 0.1849 data: 0.0926 max mem: 3531 +Test: [130/333] eta: 0:00:48 acc1_pos: 69.010 acc5_pos: 100.000 loss: 0.6363 (0.6760) acc1: 58.8542 (57.9139) acc5: 100.0000 (100.0000) time: 0.1997 data: 0.1040 max mem: 3531 +Test: [140/333] eta: 0:00:44 acc1_pos: 77.604 acc5_pos: 100.000 loss: 0.6294 (0.6732) acc1: 59.3750 (58.1486) acc5: 100.0000 (100.0000) time: 0.1921 data: 0.0977 max mem: 3531 +Test: [150/333] eta: 0:00:42 acc1_pos: 62.240 acc5_pos: 100.000 loss: 0.6556 (0.6752) acc1: 58.5938 (58.0126) acc5: 100.0000 (100.0000) time: 0.1829 data: 0.0903 max mem: 3531 +Test: [160/333] eta: 0:00:39 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.7125 (0.6814) acc1: 53.9062 (57.6588) acc5: 100.0000 (100.0000) acc1_neg: 48.958 acc5_neg: 100.000 time: 0.1779 data: 0.0834 max mem: 3531 +Test: [170/333] eta: 0:00:36 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.6780 (0.6843) acc1: 60.1562 (57.7744) acc5: 100.0000 (100.0000) acc1_neg: 65.365 acc5_neg: 100.000 time: 0.1815 data: 0.0870 max mem: 3531 +Test: [180/333] eta: 0:00:34 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.6466 (0.6806) acc1: 65.3646 (58.3520) acc5: 100.0000 (100.0000) acc1_neg: 63.802 acc5_neg: 100.000 time: 0.1948 data: 0.0997 max mem: 3531 +Test: [190/333] eta: 0:00:31 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.5964 (0.6742) acc1: 68.7500 (59.1269) acc5: 100.0000 (100.0000) acc1_neg: 75.781 acc5_neg: 100.000 time: 0.1940 data: 0.0994 max mem: 3531 +Test: [200/333] eta: 0:00:29 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.5585 (0.6701) acc1: 69.2708 (59.6834) acc5: 100.0000 (100.0000) acc1_neg: 69.271 acc5_neg: 100.000 time: 0.2104 data: 0.1179 max mem: 3531 +Test: [210/333] eta: 0:00:27 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.6372 (0.6705) acc1: 66.9271 (59.9255) acc5: 100.0000 (100.0000) acc1_neg: 59.635 acc5_neg: 100.000 time: 0.2161 data: 0.1254 max mem: 3532 +Test: [220/333] eta: 0:00:25 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.6832 (0.6731) acc1: 61.7188 (59.8322) acc5: 100.0000 (100.0000) acc1_neg: 70.833 acc5_neg: 100.000 time: 0.2402 data: 0.1487 max mem: 3532 +Test: [230/333] eta: 0:00:22 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.5064 (0.6648) acc1: 76.5625 (60.7864) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.2094 data: 0.1164 max mem: 3532 +Test: [240/333] eta: 0:00:20 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.5041 (0.6607) acc1: 76.8229 (61.2519) acc5: 100.0000 (100.0000) acc1_neg: 71.354 acc5_neg: 100.000 time: 0.1741 data: 0.0795 max mem: 3532 +Test: [250/333] eta: 0:00:18 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.5847 (0.6591) acc1: 69.2708 (61.5341) acc5: 100.0000 (100.0000) acc1_neg: 60.677 acc5_neg: 100.000 time: 0.2135 data: 0.1178 max mem: 3532 +Test: [260/333] eta: 0:00:15 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.6251 (0.6591) acc1: 67.9688 (61.7247) acc5: 100.0000 (100.0000) acc1_neg: 70.573 acc5_neg: 100.000 time: 0.1965 data: 0.1007 max mem: 3532 +Test: [270/333] eta: 0:00:13 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.5924 (0.6560) acc1: 69.2708 (62.0666) acc5: 100.0000 (100.0000) acc1_neg: 70.833 acc5_neg: 100.000 time: 0.1804 data: 0.0849 max mem: 3532 +Test: [280/333] eta: 0:00:11 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.5851 (0.6573) acc1: 68.4896 (62.0135) acc5: 100.0000 (100.0000) acc1_neg: 65.625 acc5_neg: 100.000 time: 0.1991 data: 0.1056 max mem: 3532 +Test: [290/333] eta: 0:00:09 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.7866 (0.6622) acc1: 52.3438 (61.6319) acc5: 100.0000 (100.0000) acc1_neg: 55.208 acc5_neg: 100.000 time: 0.1961 data: 0.1045 max mem: 3532 +Test: [300/333] eta: 0:00:07 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.8086 (0.6667) acc1: 50.7812 (61.4012) acc5: 100.0000 (100.0000) acc1_neg: 57.812 acc5_neg: 100.000 time: 0.1702 data: 0.0792 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.7091 (0.6626) acc1: 60.6771 (61.8527) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.1411 data: 0.0509 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.5703 (0.6607) acc1: 72.6562 (62.1195) acc5: 100.0000 (100.0000) acc1_neg: 54.427 acc5_neg: 100.000 time: 0.1201 data: 0.0303 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.7620 (0.6661) acc1: 55.9896 (61.8265) acc5: 100.0000 (100.0000) acc1_neg: 55.990 acc5_neg: 100.000 time: 0.1142 data: 0.0246 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 47.039 acc5_pos: 100.000 loss: 0.8090 (0.6671) acc1: 54.4271 (61.7773) acc5: 100.0000 (100.0000) acc1_neg: 51.549 acc5_neg: 100.000 time: 0.1148 data: 0.0220 max mem: 3532 +Test: Total time: 0:01:08 (0.2045 s / it) +* Overall Acc@1 61.777 Acc@5 100.000 + target == real Acc@1 57.987 Acc@5 100.000 + target == fake Acc@1 65.151 Acc@5 100.000 + loss 0.667 + PR-AUC 0.6745 ROC-AUC 0.6636 EER 38.40% (thr=0.5249) +Accuracy of the network on the 127843 test images: 61.8% +[rank0]:[W726 05:08:19.334248940 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:08:39.269000 1192537 torch/distributed/run.py:766] +W0726 05:08:39.269000 1192537 torch/distributed/run.py:766] ***************************************** +W0726 05:08:39.269000 1192537 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:08:39.269000 1192537 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:09:09.064632689 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:09:09.336283688 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:09:09.356173908 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:09:09.357518327 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:09:09.399423609 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 7): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:09:10.666249805 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W726 05:09:10.666396820 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:09:10.681481667 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n45_seed7_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed7_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:22:55 acc1_pos: 94.010 acc5_pos: 100.000 loss: 0.2139 (0.2139) acc1: 94.0104 (94.0104) acc5: 100.0000 (100.0000) time: 5.8551 data: 4.2406 max mem: 3531 +Test: [ 10/235] eta: 0:02:18 acc1_pos: 94.531 acc5_pos: 100.000 loss: 0.2208 (0.2236) acc1: 94.7917 (94.6496) acc5: 100.0000 (100.0000) time: 0.6141 data: 0.3859 max mem: 3531 +Test: [ 20/235] eta: 0:01:18 acc1_pos: 96.354 acc5_pos: 100.000 loss: 0.2165 (0.2174) acc1: 95.0521 (95.0521) acc5: 100.0000 (100.0000) time: 0.0899 data: 0.0004 max mem: 3531 +Test: [ 30/235] eta: 0:01:01 acc1_pos: 96.094 acc5_pos: 100.000 loss: 0.2111 (0.2159) acc1: 95.0521 (95.1193) acc5: 100.0000 (100.0000) time: 0.1251 data: 0.0349 max mem: 3531 +Test: [ 40/235] eta: 0:00:58 acc1_pos: 94.531 acc5_pos: 100.000 loss: 0.2115 (0.2155) acc1: 95.3125 (95.1156) acc5: 100.0000 (100.0000) time: 0.2301 data: 0.1370 max mem: 3531 +Test: [ 50/235] eta: 0:00:54 acc1_pos: 94.792 acc5_pos: 100.000 loss: 0.2115 (0.2165) acc1: 95.3125 (95.0878) acc5: 100.0000 (100.0000) time: 0.2826 data: 0.1878 max mem: 3531 +Test: [ 60/235] eta: 0:00:49 acc1_pos: 94.531 acc5_pos: 100.000 loss: 0.2228 (0.2181) acc1: 94.7917 (95.0435) acc5: 100.0000 (100.0000) time: 0.2562 data: 0.1637 max mem: 3531 +Test: [ 70/235] eta: 0:00:46 acc1_pos: 98.462 acc5_pos: 100.000 loss: 0.2413 (0.3761) acc1: 93.2292 (85.5487) acc5: 100.0000 (100.0000) acc1_neg: 17.708 acc5_neg: 100.000 time: 0.2577 data: 0.1640 max mem: 3531 +Test: [ 80/235] eta: 0:00:43 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4865 (0.5154) acc1: 19.5312 (77.3566) acc5: 100.0000 (100.0000) acc1_neg: 17.448 acc5_neg: 100.000 time: 0.2559 data: 0.1617 max mem: 3531 +Test: [ 90/235] eta: 0:00:40 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.5017 (0.6241) acc1: 19.2708 (70.9077) acc5: 100.0000 (100.0000) acc1_neg: 18.750 acc5_neg: 100.000 time: 0.2548 data: 0.1633 max mem: 3531 +Test: [100/235] eta: 0:00:36 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.5017 (0.7092) acc1: 19.0104 (65.8235) acc5: 100.0000 (100.0000) acc1_neg: 17.188 acc5_neg: 100.000 time: 0.2340 data: 0.1405 max mem: 3531 +Test: [110/235] eta: 0:00:33 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4790 (0.7789) acc1: 19.0104 (61.6578) acc5: 100.0000 (100.0000) acc1_neg: 22.656 acc5_neg: 100.000 time: 0.2226 data: 0.1275 max mem: 3531 +Test: [120/235] eta: 0:00:30 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4897 (0.8382) acc1: 18.4896 (58.0579) acc5: 100.0000 (100.0000) acc1_neg: 20.833 acc5_neg: 100.000 time: 0.2346 data: 0.1414 max mem: 3531 +Test: [130/235] eta: 0:00:27 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4897 (0.8863) acc1: 18.7500 (55.1229) acc5: 100.0000 (100.0000) acc1_neg: 19.010 acc5_neg: 100.000 time: 0.2338 data: 0.1405 max mem: 3531 +Test: [140/235] eta: 0:00:24 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4902 (0.9309) acc1: 18.7500 (52.4823) acc5: 100.0000 (100.0000) acc1_neg: 12.500 acc5_neg: 100.000 time: 0.2363 data: 0.1424 max mem: 3531 +Test: [150/235] eta: 0:00:22 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4903 (0.9671) acc1: 18.2292 (50.2570) acc5: 100.0000 (100.0000) acc1_neg: 21.354 acc5_neg: 100.000 time: 0.2494 data: 0.1548 max mem: 3531 +Test: [160/235] eta: 0:00:19 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4741 (0.9989) acc1: 19.0104 (48.3032) acc5: 100.0000 (100.0000) acc1_neg: 19.010 acc5_neg: 100.000 time: 0.2423 data: 0.1433 max mem: 3531 +Test: [170/235] eta: 0:00:16 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4315 (1.0141) acc1: 21.3542 (47.1232) acc5: 100.0000 (100.0000) acc1_neg: 26.302 acc5_neg: 100.000 time: 0.1995 data: 0.1021 max mem: 3531 +Test: [180/235] eta: 0:00:13 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.2295 (1.0257) acc1: 29.1667 (46.1686) acc5: 100.0000 (100.0000) acc1_neg: 31.250 acc5_neg: 100.000 time: 0.1741 data: 0.0823 max mem: 3531 +Test: [190/235] eta: 0:00:10 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.2088 (1.0360) acc1: 29.4271 (45.2648) acc5: 100.0000 (100.0000) acc1_neg: 29.427 acc5_neg: 100.000 time: 0.1589 data: 0.0684 max mem: 3531 +Test: [200/235] eta: 0:00:08 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.2258 (1.0463) acc1: 27.6042 (44.3823) acc5: 100.0000 (100.0000) acc1_neg: 27.865 acc5_neg: 100.000 time: 0.1513 data: 0.0611 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.2258 (1.0543) acc1: 28.3854 (43.7043) acc5: 100.0000 (100.0000) acc1_neg: 29.688 acc5_neg: 100.000 time: 0.1356 data: 0.0454 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.2135 (1.0614) acc1: 29.6875 (43.0819) acc5: 100.0000 (100.0000) acc1_neg: 26.823 acc5_neg: 100.000 time: 0.1190 data: 0.0290 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.2135 (1.0675) acc1: 29.6875 (42.5505) acc5: 100.0000 (100.0000) acc1_neg: 31.250 acc5_neg: 100.000 time: 0.1163 data: 0.0261 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.2164 (1.0711) acc1: 29.6875 (42.3647) acc5: 100.0000 (100.0000) acc1_neg: 30.435 acc5_neg: 100.000 time: 0.1124 data: 0.0252 max mem: 3532 +Test: Total time: 0:00:51 (0.2211 s / it) +* Overall Acc@1 42.365 Acc@5 100.000 + target == real Acc@1 95.053 Acc@5 100.000 + target == fake Acc@1 23.308 Acc@5 100.000 + loss 1.071 + PR-AUC 0.8732 ROC-AUC 0.7223 EER 33.84% (thr=0.1772) +Accuracy of the network on the 89879 test images: 42.4% +[rank0]:[W726 05:10:11.919716642 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:10:31.643000 1193095 torch/distributed/run.py:766] +W0726 05:10:31.643000 1193095 torch/distributed/run.py:766] ***************************************** +W0726 05:10:31.643000 1193095 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:10:31.643000 1193095 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:10:58.385088263 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:10:59.657886544 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:10:59.919282951 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:10:59.020685768 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:10:59.061633198 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +[rank0]:[W726 05:10:59.080311731 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:10:59.087550654 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:10:59.109992714 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n45_seed7_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed7_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:31:30 acc1_pos: 45.052 acc5_pos: 100.000 loss: 0.7619 (0.7619) acc1: 45.0521 (45.0521) acc5: 100.0000 (100.0000) time: 7.0294 data: 5.3638 max mem: 3531 +Test: [ 10/269] eta: 0:03:07 acc1_pos: 67.969 acc5_pos: 100.000 loss: 0.6912 (0.6967) acc1: 55.9896 (56.1316) acc5: 100.0000 (100.0000) time: 0.7252 data: 0.4881 max mem: 3531 +Test: [ 20/269] eta: 0:01:46 acc1_pos: 59.896 acc5_pos: 100.000 loss: 0.6776 (0.6739) acc1: 57.0312 (58.6186) acc5: 100.0000 (100.0000) time: 0.0973 data: 0.0005 max mem: 3531 +Test: [ 30/269] eta: 0:01:28 acc1_pos: 59.115 acc5_pos: 100.000 loss: 0.6464 (0.6553) acc1: 59.8958 (60.8787) acc5: 100.0000 (100.0000) time: 0.1749 data: 0.0785 max mem: 3531 +Test: [ 40/269] eta: 0:01:16 acc1_pos: 70.690 acc5_pos: 100.000 loss: 0.5022 (0.6118) acc1: 78.9062 (66.3237) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.2349 data: 0.1408 max mem: 3531 +Test: [ 50/269] eta: 0:01:10 acc1_pos: 70.690 acc5_pos: 100.000 loss: 0.4834 (0.5943) acc1: 81.5104 (68.3160) acc5: 100.0000 (100.0000) acc1_neg: 55.208 acc5_neg: 100.000 time: 0.2517 data: 0.1578 max mem: 3531 +Test: [ 60/269] eta: 0:01:02 acc1_pos: 70.690 acc5_pos: 100.000 loss: 0.5189 (0.5942) acc1: 75.0000 (68.3530) acc5: 100.0000 (100.0000) acc1_neg: 60.677 acc5_neg: 100.000 time: 0.2349 data: 0.1422 max mem: 3531 +Test: [ 70/269] eta: 0:00:54 acc1_pos: 70.690 acc5_pos: 100.000 loss: 0.5189 (0.5828) acc1: 75.0000 (69.6670) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.1431 data: 0.0471 max mem: 3531 +Test: [ 80/269] eta: 0:00:49 acc1_pos: 70.690 acc5_pos: 100.000 loss: 0.5045 (0.5701) acc1: 79.4271 (71.3027) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1499 data: 0.0528 max mem: 3531 +Test: [ 90/269] eta: 0:00:45 acc1_pos: 70.690 acc5_pos: 100.000 loss: 0.5224 (0.5719) acc1: 75.2604 (71.0737) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.1999 data: 0.1046 max mem: 3531 +Test: [100/269] eta: 0:00:42 acc1_pos: 70.690 acc5_pos: 100.000 loss: 0.5739 (0.5698) acc1: 69.7917 (71.2717) acc5: 100.0000 (100.0000) acc1_neg: 66.927 acc5_neg: 100.000 time: 0.1899 data: 0.0930 max mem: 3531 +Test: [110/269] eta: 0:00:39 acc1_pos: 70.690 acc5_pos: 100.000 loss: 0.5904 (0.5734) acc1: 68.4896 (70.6668) acc5: 100.0000 (100.0000) acc1_neg: 55.990 acc5_neg: 100.000 time: 0.1962 data: 0.0996 max mem: 3531 +Test: [120/269] eta: 0:00:36 acc1_pos: 70.690 acc5_pos: 100.000 loss: 0.5617 (0.5677) acc1: 69.2708 (71.3004) acc5: 100.0000 (100.0000) acc1_neg: 77.083 acc5_neg: 100.000 time: 0.2052 data: 0.1110 max mem: 3531 +Test: [130/269] eta: 0:00:32 acc1_pos: 70.690 acc5_pos: 100.000 loss: 0.5215 (0.5696) acc1: 75.7812 (71.0500) acc5: 100.0000 (100.0000) acc1_neg: 61.198 acc5_neg: 100.000 time: 0.1849 data: 0.0896 max mem: 3531 +Test: [140/269] eta: 0:00:29 acc1_pos: 70.690 acc5_pos: 100.000 loss: 0.6253 (0.5752) acc1: 61.7188 (70.3698) acc5: 100.0000 (100.0000) acc1_neg: 52.604 acc5_neg: 100.000 time: 0.1696 data: 0.0746 max mem: 3531 +Test: [150/269] eta: 0:00:27 acc1_pos: 70.690 acc5_pos: 100.000 loss: 0.6139 (0.5765) acc1: 61.7188 (70.1763) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.1998 data: 0.1062 max mem: 3531 +Test: [160/269] eta: 0:00:25 acc1_pos: 70.690 acc5_pos: 100.000 loss: 0.5165 (0.5748) acc1: 75.2604 (70.4484) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.2216 data: 0.1273 max mem: 3531 +Test: [170/269] eta: 0:00:22 acc1_pos: 70.690 acc5_pos: 100.000 loss: 0.5439 (0.5736) acc1: 75.2604 (70.6247) acc5: 100.0000 (100.0000) acc1_neg: 74.740 acc5_neg: 100.000 time: 0.1816 data: 0.0875 max mem: 3531 +Test: [180/269] eta: 0:00:19 acc1_pos: 70.690 acc5_pos: 100.000 loss: 0.5162 (0.5685) acc1: 76.0417 (71.1945) acc5: 100.0000 (100.0000) acc1_neg: 65.885 acc5_neg: 100.000 time: 0.1696 data: 0.0772 max mem: 3531 +Test: [190/269] eta: 0:00:17 acc1_pos: 70.690 acc5_pos: 100.000 loss: 0.5162 (0.5682) acc1: 75.7812 (71.1551) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.2093 data: 0.1171 max mem: 3531 +Test: [200/269] eta: 0:00:15 acc1_pos: 70.690 acc5_pos: 100.000 loss: 0.5460 (0.5684) acc1: 70.0521 (71.1171) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.1986 data: 0.1026 max mem: 3531 +Test: [210/269] eta: 0:00:12 acc1_pos: 70.690 acc5_pos: 100.000 loss: 0.5572 (0.5696) acc1: 68.4896 (70.8790) acc5: 100.0000 (100.0000) acc1_neg: 74.219 acc5_neg: 100.000 time: 0.1505 data: 0.0527 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 70.690 acc5_pos: 100.000 loss: 0.5572 (0.5702) acc1: 68.4896 (70.7992) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1704 data: 0.0735 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 70.690 acc5_pos: 100.000 loss: 0.4707 (0.5654) acc1: 79.9479 (71.2989) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.2165 data: 0.1218 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 70.690 acc5_pos: 100.000 loss: 0.5428 (0.5674) acc1: 74.2188 (71.1651) acc5: 100.0000 (100.0000) acc1_neg: 70.052 acc5_neg: 100.000 time: 0.1890 data: 0.0957 max mem: 3532 +Test: [250/269] eta: 0:00:04 acc1_pos: 70.690 acc5_pos: 100.000 loss: 0.5779 (0.5716) acc1: 66.9271 (70.6590) acc5: 100.0000 (100.0000) acc1_neg: 37.760 acc5_neg: 100.000 time: 0.1508 data: 0.0578 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 70.690 acc5_pos: 100.000 loss: 0.6995 (0.5780) acc1: 54.4271 (69.9343) acc5: 100.0000 (100.0000) acc1_neg: 70.052 acc5_neg: 100.000 time: 0.1549 data: 0.0629 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 70.690 acc5_pos: 100.000 loss: 0.5919 (0.5779) acc1: 55.4688 (70.0241) acc5: 100.0000 (100.0000) acc1_neg: 80.052 acc5_neg: 100.000 time: 0.1338 data: 0.0398 max mem: 3532 +Test: Total time: 0:00:55 (0.2074 s / it) +* Overall Acc@1 70.024 Acc@5 100.000 + target == real Acc@1 60.988 Acc@5 100.000 + target == fake Acc@1 71.257 Acc@5 100.000 + loss 0.578 + PR-AUC 0.9464 ROC-AUC 0.7277 EER 33.40% (thr=0.5278) +Accuracy of the network on the 103293 test images: 70.0% +[rank0]:[W726 05:12:04.559966585 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T05:12:11Z] eval_done tag=fixed500k_fake_n45_seed7_resampled +[2026-07-26T05:12:11Z] eval_start tag=fixed500k_real_n23_seed8_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed8_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T05:12:11Z +hostname=n116-041-023 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed8_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n23_seed8_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 05:12:24.515000 1193670 torch/distributed/run.py:766] +W0726 05:12:24.515000 1193670 torch/distributed/run.py:766] ***************************************** +W0726 05:12:24.515000 1193670 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:12:24.515000 1193670 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:12:53.430542503 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:12:53.460627362 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:12:53.778618701 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:12:53.926974299 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:12:53.133027690 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:12:53.224687469 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:12:53.346192078 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:12:53.361456314 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n23_seed8_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed8_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:26:28 acc1_pos: 20.573 acc5_pos: 100.000 loss: 1.2420 (1.2420) acc1: 20.5729 (20.5729) acc5: 100.0000 (100.0000) time: 6.7892 data: 5.2223 max mem: 3531 +Test: [ 10/234] eta: 0:02:38 acc1_pos: 29.167 acc5_pos: 100.000 loss: 1.0308 (1.0166) acc1: 34.1146 (37.1922) acc5: 100.0000 (100.0000) time: 0.7060 data: 0.4751 max mem: 3531 +Test: [ 20/234] eta: 0:01:28 acc1_pos: 92.188 acc5_pos: 100.000 loss: 1.0308 (1.0296) acc1: 33.3333 (39.5833) acc5: 100.0000 (100.0000) time: 0.0968 data: 0.0004 max mem: 3531 +Test: [ 30/234] eta: 0:01:15 acc1_pos: 78.125 acc5_pos: 100.000 loss: 0.2779 (0.7885) acc1: 91.6667 (56.3676) acc5: 100.0000 (100.0000) time: 0.1879 data: 0.0905 max mem: 3531 +Test: [ 40/234] eta: 0:01:04 acc1_pos: 98.177 acc5_pos: 100.000 loss: 0.2757 (0.6879) acc1: 92.7083 (63.4337) acc5: 100.0000 (100.0000) time: 0.2470 data: 0.1490 max mem: 3531 +Test: [ 50/234] eta: 0:00:57 acc1_pos: 97.656 acc5_pos: 100.000 loss: 0.1837 (0.5876) acc1: 96.8750 (70.1134) acc5: 100.0000 (100.0000) time: 0.2165 data: 0.1229 max mem: 3531 +Test: [ 60/234] eta: 0:00:51 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.1337 (0.5058) acc1: 98.1771 (74.9829) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2178 data: 0.1245 max mem: 3531 +Test: [ 70/234] eta: 0:00:46 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.0871 (0.4982) acc1: 100.0000 (75.1357) acc5: 100.0000 (100.0000) acc1_neg: 61.979 acc5_neg: 100.000 time: 0.2186 data: 0.1248 max mem: 3531 +Test: [ 80/234] eta: 0:00:41 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.6053 (0.4905) acc1: 67.4479 (75.4597) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1956 data: 0.1030 max mem: 3531 +Test: [ 90/234] eta: 0:00:38 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.1413 (0.4518) acc1: 94.5312 (77.7558) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1896 data: 0.0944 max mem: 3531 +Test: [100/234] eta: 0:00:34 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.1910 (0.4431) acc1: 90.1042 (78.2591) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1999 data: 0.1059 max mem: 3531 +Test: [110/234] eta: 0:00:31 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.3686 (0.4329) acc1: 82.0312 (78.9555) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.1898 data: 0.0983 max mem: 3531 +Test: [120/234] eta: 0:00:27 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.3360 (0.4297) acc1: 85.4167 (79.1990) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1855 data: 0.0939 max mem: 3531 +Test: [130/234] eta: 0:00:25 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.2527 (0.4195) acc1: 93.7500 (80.0553) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1958 data: 0.1047 max mem: 3531 +Test: [140/234] eta: 0:00:22 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.3441 (0.4338) acc1: 85.6771 (79.3938) acc5: 100.0000 (100.0000) acc1_neg: 51.562 acc5_neg: 100.000 time: 0.2051 data: 0.1135 max mem: 3531 +Test: [150/234] eta: 0:00:19 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.3914 (0.4332) acc1: 80.2083 (79.3771) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.2017 data: 0.1083 max mem: 3531 +Test: [160/234] eta: 0:00:17 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.3281 (0.4248) acc1: 83.5938 (79.9932) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.2064 data: 0.1107 max mem: 3531 +Test: [170/234] eta: 0:00:14 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.3831 (0.4223) acc1: 86.7188 (80.3088) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.2095 data: 0.1143 max mem: 3531 +Test: [180/234] eta: 0:00:12 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.4066 (0.4289) acc1: 82.8125 (79.7839) acc5: 100.0000 (100.0000) acc1_neg: 40.104 acc5_neg: 100.000 time: 0.1992 data: 0.1062 max mem: 3531 +Test: [190/234] eta: 0:00:10 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.5694 (0.4373) acc1: 71.0938 (79.2703) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.1764 data: 0.0848 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.5658 (0.4294) acc1: 72.6562 (79.7562) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1519 data: 0.0617 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.1096 (0.4155) acc1: 99.7396 (80.6489) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1389 data: 0.0487 max mem: 3532 +Test: [220/234] eta: 0:00:03 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.1802 (0.4062) acc1: 96.8750 (81.2960) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.1342 data: 0.0440 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.4014 (0.4260) acc1: 50.7812 (79.7743) acc5: 100.0000 (100.0000) acc1_neg: 44.271 acc5_neg: 100.000 time: 0.1286 data: 0.0384 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.8406 (0.4315) acc1: 46.8750 (79.3790) acc5: 100.0000 (100.0000) acc1_neg: 46.304 acc5_neg: 100.000 time: 0.1302 data: 0.0392 max mem: 3532 +Test: Total time: 0:00:49 (0.2113 s / it) +* Overall Acc@1 79.379 Acc@5 100.000 + target == real Acc@1 70.298 Acc@5 100.000 + target == fake Acc@1 81.939 Acc@5 100.000 + loss 0.431 + PR-AUC 0.9541 ROC-AUC 0.8573 EER 23.79% (thr=0.6055) +Accuracy of the network on the 89729 test images: 79.4% +[rank0]:[W726 05:13:52.907663726 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:14:11.475000 1194217 torch/distributed/run.py:766] +W0726 05:14:11.475000 1194217 torch/distributed/run.py:766] ***************************************** +W0726 05:14:11.475000 1194217 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:14:11.475000 1194217 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:14:41.510886360 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:14:42.553441203 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:14:42.577818419 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +| distributed init (rank 3): env:// +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:14:42.938334298 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:14:42.950705751 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W726 05:14:42.952030714 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:14:42.967918324 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:14:42.213840637 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n23_seed8_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed8_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:46:37 acc1_pos: 9.115 acc5_pos: 100.000 loss: 1.4840 (1.4840) acc1: 9.1146 (9.1146) acc5: 100.0000 (100.0000) time: 7.2046 data: 5.6413 max mem: 3531 +Test: [ 10/2387] eta: 0:29:11 acc1_pos: 1.302 acc5_pos: 100.000 loss: 1.4640 (1.4671) acc1: 6.7708 (8.6648) acc5: 100.0000 (100.0000) time: 0.7371 data: 0.5132 max mem: 3531 +Test: [ 20/2387] eta: 0:17:06 acc1_pos: 5.208 acc5_pos: 100.000 loss: 1.4483 (1.4717) acc1: 7.5521 (10.0074) acc5: 100.0000 (100.0000) time: 0.0950 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:14:21 acc1_pos: 1.302 acc5_pos: 100.000 loss: 1.5426 (1.4999) acc1: 7.8125 (8.8878) acc5: 100.0000 (100.0000) time: 0.1610 data: 0.0651 max mem: 3531 +Test: [ 40/2387] eta: 0:13:15 acc1_pos: 3.125 acc5_pos: 100.000 loss: 1.5514 (1.5220) acc1: 3.6458 (7.3742) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1441 max mem: 3531 +Test: [ 50/2387] eta: 0:12:32 acc1_pos: 26.823 acc5_pos: 100.000 loss: 1.5016 (1.5102) acc1: 3.6458 (7.7921) acc5: 100.0000 (100.0000) time: 0.2544 data: 0.1567 max mem: 3531 +Test: [ 60/2387] eta: 0:12:08 acc1_pos: 19.792 acc5_pos: 100.000 loss: 1.4980 (1.5153) acc1: 7.8125 (8.1498) acc5: 100.0000 (100.0000) time: 0.2598 data: 0.1635 max mem: 3531 +Test: [ 70/2387] eta: 0:11:39 acc1_pos: 16.406 acc5_pos: 100.000 loss: 1.6148 (1.5391) acc1: 6.2500 (7.8932) acc5: 100.0000 (100.0000) time: 0.2503 data: 0.1539 max mem: 3531 +Test: [ 80/2387] eta: 0:11:29 acc1_pos: 5.729 acc5_pos: 100.000 loss: 1.6886 (1.5558) acc1: 5.7292 (7.7675) acc5: 100.0000 (100.0000) time: 0.2559 data: 0.1612 max mem: 3531 +Test: [ 90/2387] eta: 0:11:16 acc1_pos: 10.417 acc5_pos: 100.000 loss: 1.5955 (1.5594) acc1: 8.0729 (8.2418) acc5: 100.0000 (100.0000) time: 0.2688 data: 0.1726 max mem: 3531 +Test: [ 100/2387] eta: 0:11:14 acc1_pos: 8.594 acc5_pos: 100.000 loss: 1.6502 (1.5762) acc1: 8.0729 (8.1477) acc5: 100.0000 (100.0000) time: 0.2789 data: 0.1808 max mem: 3531 +Test: [ 110/2387] eta: 0:10:58 acc1_pos: 19.531 acc5_pos: 100.000 loss: 1.5828 (1.5710) acc1: 8.0729 (8.3873) acc5: 100.0000 (100.0000) time: 0.2649 data: 0.1679 max mem: 3531 +Test: [ 120/2387] eta: 0:10:38 acc1_pos: 12.240 acc5_pos: 100.000 loss: 1.5426 (1.5669) acc1: 9.1146 (8.8047) acc5: 100.0000 (100.0000) time: 0.2152 data: 0.1181 max mem: 3531 +Test: [ 130/2387] eta: 0:10:30 acc1_pos: 19.010 acc5_pos: 100.000 loss: 1.5426 (1.5684) acc1: 9.8958 (8.9516) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1276 max mem: 3531 +Test: [ 140/2387] eta: 0:10:21 acc1_pos: 16.927 acc5_pos: 100.000 loss: 1.5189 (1.5653) acc1: 9.8958 (9.2402) acc5: 100.0000 (100.0000) time: 0.2447 data: 0.1443 max mem: 3531 +Test: [ 150/2387] eta: 0:10:10 acc1_pos: 0.260 acc5_pos: 100.000 loss: 1.5416 (1.5682) acc1: 10.1562 (9.2232) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1274 max mem: 3531 +Test: [ 160/2387] eta: 0:09:53 acc1_pos: 1.370 acc5_pos: 100.000 loss: 1.5609 (1.5362) acc1: 10.1562 (11.3758) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1949 data: 0.0957 max mem: 3531 +Test: [ 170/2387] eta: 0:09:43 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1386 (1.4541) acc1: 98.6979 (16.5174) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1899 data: 0.0929 max mem: 3531 +Test: [ 180/2387] eta: 0:09:30 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1329 (1.3813) acc1: 99.2188 (21.0866) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1949 data: 0.0976 max mem: 3531 +Test: [ 190/2387] eta: 0:09:12 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1329 (1.3159) acc1: 99.2188 (25.1772) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1499 data: 0.0529 max mem: 3531 +Test: [ 200/2387] eta: 0:09:08 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1309 (1.2569) acc1: 99.2188 (28.8661) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1821 data: 0.0856 max mem: 3531 +Test: [ 210/2387] eta: 0:09:02 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1325 (1.2037) acc1: 99.4792 (32.2040) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2317 data: 0.1379 max mem: 3532 +Test: [ 220/2387] eta: 0:08:56 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1338 (1.1553) acc1: 99.4792 (35.2364) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2116 data: 0.1198 max mem: 3532 +Test: [ 230/2387] eta: 0:08:49 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1327 (1.1110) acc1: 99.4792 (38.0118) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2029 data: 0.1108 max mem: 3532 +Test: [ 240/2387] eta: 0:08:43 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1260 (1.0700) acc1: 99.4792 (40.5667) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2058 data: 0.1139 max mem: 3532 +Test: [ 250/2387] eta: 0:08:38 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1169 (1.0306) acc1: 99.7396 (42.9304) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2118 data: 0.1208 max mem: 3532 +Test: [ 260/2387] eta: 0:08:33 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0706 (0.9937) acc1: 100.0000 (45.1169) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2145 data: 0.1232 max mem: 3532 +Test: [ 270/2387] eta: 0:08:29 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0689 (0.9596) acc1: 100.0000 (47.1412) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2122 data: 0.1183 max mem: 3532 +Test: [ 280/2387] eta: 0:08:25 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0680 (0.9279) acc1: 100.0000 (49.0223) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2152 data: 0.1207 max mem: 3532 +Test: [ 290/2387] eta: 0:08:21 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0686 (0.8985) acc1: 100.0000 (50.7741) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2218 data: 0.1299 max mem: 3532 +Test: [ 300/2387] eta: 0:08:18 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0686 (0.8709) acc1: 100.0000 (52.4095) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2296 data: 0.1364 max mem: 3532 +Test: [ 310/2387] eta: 0:08:16 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0696 (0.8475) acc1: 100.0000 (53.8996) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2337 data: 0.1397 max mem: 3532 +Test: [ 320/2387] eta: 0:08:12 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.2231 (0.8282) acc1: 97.9167 (55.2481) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2244 data: 0.1320 max mem: 3532 +Test: [ 330/2387] eta: 0:08:09 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.2256 (0.8099) acc1: 97.6562 (56.5356) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2267 data: 0.1324 max mem: 3532 +Test: [ 340/2387] eta: 0:08:06 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.2251 (0.7927) acc1: 97.9167 (57.7461) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2307 data: 0.1366 max mem: 3532 +Test: [ 350/2387] eta: 0:08:01 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.2142 (0.7728) acc1: 98.4375 (58.9417) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2104 data: 0.1169 max mem: 3532 +Test: [ 360/2387] eta: 0:07:59 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1095 (0.7544) acc1: 100.0000 (60.0726) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2141 data: 0.1177 max mem: 3532 +Test: [ 370/2387] eta: 0:07:55 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1141 (0.7372) acc1: 99.7396 (61.1404) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2269 data: 0.1311 max mem: 3532 +Test: [ 380/2387] eta: 0:07:52 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1148 (0.7209) acc1: 99.7396 (62.1535) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2218 data: 0.1293 max mem: 3532 +Test: [ 390/2387] eta: 0:07:50 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1151 (0.7054) acc1: 99.7396 (63.1147) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2290 data: 0.1371 max mem: 3532 +Test: [ 400/2387] eta: 0:07:45 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1151 (0.6907) acc1: 99.7396 (64.0307) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2101 data: 0.1186 max mem: 3532 +Test: [ 410/2387] eta: 0:07:43 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1149 (0.6766) acc1: 100.0000 (64.9020) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2065 data: 0.1140 max mem: 3532 +Test: [ 420/2387] eta: 0:07:38 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1149 (0.6633) acc1: 100.0000 (65.7308) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2028 data: 0.1069 max mem: 3532 +Test: [ 430/2387] eta: 0:07:34 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1143 (0.6506) acc1: 99.7396 (66.5168) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1871 data: 0.0921 max mem: 3532 +Test: [ 440/2387] eta: 0:07:29 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1140 (0.6380) acc1: 99.7396 (67.2696) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1928 data: 0.1013 max mem: 3533 +Test: [ 450/2387] eta: 0:07:25 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0804 (0.6256) acc1: 100.0000 (67.9953) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1878 data: 0.0943 max mem: 3533 +Test: [ 460/2387] eta: 0:07:21 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0799 (0.6138) acc1: 100.0000 (68.6896) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1893 data: 0.0959 max mem: 3533 +Test: [ 470/2387] eta: 0:07:17 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0802 (0.6025) acc1: 100.0000 (69.3543) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1877 data: 0.0949 max mem: 3533 +Test: [ 480/2387] eta: 0:07:13 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0801 (0.5916) acc1: 100.0000 (69.9914) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1885 data: 0.0950 max mem: 3533 +Test: [ 490/2387] eta: 0:07:09 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0803 (0.5812) acc1: 100.0000 (70.6026) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1887 data: 0.0967 max mem: 3533 +Test: [ 500/2387] eta: 0:07:06 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0807 (0.5712) acc1: 100.0000 (71.1894) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1855 data: 0.0913 max mem: 3533 +Test: [ 510/2387] eta: 0:07:02 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0810 (0.5617) acc1: 100.0000 (71.7532) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1886 data: 0.0940 max mem: 3533 +Test: [ 520/2387] eta: 0:06:58 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1001 (0.5529) acc1: 100.0000 (72.2954) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1845 data: 0.0898 max mem: 3533 +Test: [ 530/2387] eta: 0:06:55 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1064 (0.5446) acc1: 100.0000 (72.8171) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1910 data: 0.0946 max mem: 3533 +Test: [ 540/2387] eta: 0:06:52 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1079 (0.5365) acc1: 100.0000 (73.3196) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1927 data: 0.0967 max mem: 3533 +Test: [ 550/2387] eta: 0:06:48 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1082 (0.5287) acc1: 100.0000 (73.8038) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1853 data: 0.0892 max mem: 3533 +Test: [ 560/2387] eta: 0:06:45 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1082 (0.5212) acc1: 100.0000 (74.2707) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1876 data: 0.0934 max mem: 3533 +Test: [ 570/2387] eta: 0:06:42 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1081 (0.5140) acc1: 100.0000 (74.7213) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1884 data: 0.0965 max mem: 3533 +Test: [ 580/2387] eta: 0:06:38 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1081 (0.5070) acc1: 100.0000 (75.1564) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1906 data: 0.0976 max mem: 3533 +Test: [ 590/2387] eta: 0:06:35 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1081 (0.5003) acc1: 100.0000 (75.5768) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1870 data: 0.0937 max mem: 3533 +Test: [ 600/2387] eta: 0:06:32 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1021 (0.4936) acc1: 100.0000 (75.9832) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1814 data: 0.0898 max mem: 3533 +Test: [ 610/2387] eta: 0:06:28 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0993 (0.4871) acc1: 100.0000 (76.3762) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1828 data: 0.0914 max mem: 3533 +Test: [ 620/2387] eta: 0:06:25 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0971 (0.4809) acc1: 100.0000 (76.7558) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1860 data: 0.0938 max mem: 3533 +Test: [ 630/2387] eta: 0:06:23 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0977 (0.4748) acc1: 100.0000 (77.1238) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1951 data: 0.1028 max mem: 3533 +Test: [ 640/2387] eta: 0:06:20 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0985 (0.4689) acc1: 100.0000 (77.4799) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2065 data: 0.1151 max mem: 3533 +Test: [ 650/2387] eta: 0:06:17 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0982 (0.4632) acc1: 100.0000 (77.8246) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1957 data: 0.1026 max mem: 3533 +Test: [ 660/2387] eta: 0:06:15 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0975 (0.4577) acc1: 100.0000 (78.1593) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0968 max mem: 3533 +Test: [ 670/2387] eta: 0:06:11 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0977 (0.4524) acc1: 100.0000 (78.4848) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1846 data: 0.0939 max mem: 3534 +Test: [ 680/2387] eta: 0:06:09 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0957 (0.4471) acc1: 100.0000 (78.8007) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1833 data: 0.0921 max mem: 3534 +Test: [ 690/2387] eta: 0:06:06 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0977 (0.4461) acc1: 100.0000 (78.8444) acc5: 100.0000 (100.0000) acc1_neg: 72.396 acc5_neg: 100.000 time: 0.1965 data: 0.1041 max mem: 3534 +Test: [ 700/2387] eta: 0:06:03 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.4817 (0.4467) acc1: 75.0000 (78.7959) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.1951 data: 0.1034 max mem: 3534 +Test: [ 710/2387] eta: 0:06:00 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.4875 (0.4473) acc1: 75.0000 (78.7381) acc5: 100.0000 (100.0000) acc1_neg: 74.219 acc5_neg: 100.000 time: 0.1865 data: 0.0947 max mem: 3534 +Test: [ 720/2387] eta: 0:05:57 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.4803 (0.4478) acc1: 75.5208 (78.7036) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.1717 data: 0.0790 max mem: 3534 +Test: [ 730/2387] eta: 0:05:54 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.4852 (0.4485) acc1: 75.0000 (78.6501) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.1799 data: 0.0873 max mem: 3534 +Test: [ 740/2387] eta: 0:05:51 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.4925 (0.4491) acc1: 75.0000 (78.5994) acc5: 100.0000 (100.0000) acc1_neg: 72.656 acc5_neg: 100.000 time: 0.1793 data: 0.0853 max mem: 3534 +Test: [ 750/2387] eta: 0:05:48 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.4909 (0.4496) acc1: 74.7396 (78.5602) acc5: 100.0000 (100.0000) acc1_neg: 76.042 acc5_neg: 100.000 time: 0.1699 data: 0.0766 max mem: 3534 +Test: [ 760/2387] eta: 0:05:45 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.4765 (0.4489) acc1: 76.0417 (78.5996) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1814 data: 0.0891 max mem: 3534 +Test: [ 770/2387] eta: 0:05:43 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.2518 (0.4456) acc1: 91.9271 (78.8184) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0911 max mem: 3534 +Test: [ 780/2387] eta: 0:05:40 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1325 (0.4414) acc1: 99.7396 (79.0830) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1827 data: 0.0883 max mem: 3534 +Test: [ 790/2387] eta: 0:05:37 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1119 (0.4376) acc1: 99.7396 (79.3125) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1798 data: 0.0852 max mem: 3534 +Test: [ 800/2387] eta: 0:05:34 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1147 (0.4335) acc1: 99.7396 (79.5675) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1753 data: 0.0820 max mem: 3534 +Test: [ 810/2387] eta: 0:05:32 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1078 (0.4296) acc1: 100.0000 (79.8150) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1789 data: 0.0856 max mem: 3534 +Test: [ 820/2387] eta: 0:05:29 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1152 (0.4259) acc1: 100.0000 (80.0561) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1788 data: 0.0851 max mem: 3534 +Test: [ 830/2387] eta: 0:05:26 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1406 (0.4228) acc1: 99.4792 (80.2770) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1764 data: 0.0839 max mem: 3534 +Test: [ 840/2387] eta: 0:05:23 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1442 (0.4194) acc1: 98.9583 (80.4945) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1775 data: 0.0856 max mem: 3534 +Test: [ 850/2387] eta: 0:05:21 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1410 (0.4162) acc1: 98.4375 (80.7025) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1799 data: 0.0868 max mem: 3534 +Test: [ 860/2387] eta: 0:05:18 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1256 (0.4127) acc1: 99.4792 (80.9188) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1823 data: 0.0878 max mem: 3534 +Test: [ 870/2387] eta: 0:05:16 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1239 (0.4094) acc1: 99.4792 (81.1238) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1860 data: 0.0934 max mem: 3534 +Test: [ 880/2387] eta: 0:05:13 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1258 (0.4063) acc1: 99.2188 (81.3260) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1820 data: 0.0899 max mem: 3534 +Test: [ 890/2387] eta: 0:05:11 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1101 (0.4029) acc1: 99.7396 (81.5335) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1750 data: 0.0826 max mem: 3534 +Test: [ 900/2387] eta: 0:05:08 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1268 (0.4042) acc1: 98.9583 (81.4223) acc5: 100.0000 (100.0000) acc1_neg: 66.927 acc5_neg: 100.000 time: 0.1853 data: 0.0936 max mem: 3535 +Test: [ 910/2387] eta: 0:05:06 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.6171 (0.4065) acc1: 65.1042 (81.2554) acc5: 100.0000 (100.0000) acc1_neg: 65.365 acc5_neg: 100.000 time: 0.1839 data: 0.0925 max mem: 3535 +Test: [ 920/2387] eta: 0:05:03 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.6189 (0.4089) acc1: 65.3646 (81.0826) acc5: 100.0000 (100.0000) acc1_neg: 63.281 acc5_neg: 100.000 time: 0.1801 data: 0.0871 max mem: 3535 +Test: [ 930/2387] eta: 0:05:01 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.6189 (0.4111) acc1: 64.5833 (80.9071) acc5: 100.0000 (100.0000) acc1_neg: 61.458 acc5_neg: 100.000 time: 0.1792 data: 0.0859 max mem: 3535 +Test: [ 940/2387] eta: 0:04:58 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.6225 (0.4133) acc1: 64.3229 (80.7353) acc5: 100.0000 (100.0000) acc1_neg: 59.896 acc5_neg: 100.000 time: 0.1734 data: 0.0819 max mem: 3535 +Test: [ 950/2387] eta: 0:04:55 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.6080 (0.4154) acc1: 64.3229 (80.5679) acc5: 100.0000 (100.0000) acc1_neg: 68.229 acc5_neg: 100.000 time: 0.1713 data: 0.0797 max mem: 3535 +Test: [ 960/2387] eta: 0:04:53 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.6041 (0.4174) acc1: 65.3646 (80.4132) acc5: 100.0000 (100.0000) acc1_neg: 65.104 acc5_neg: 100.000 time: 0.1756 data: 0.0839 max mem: 3535 +Test: [ 970/2387] eta: 0:04:51 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.6062 (0.4194) acc1: 65.3646 (80.2558) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.1803 data: 0.0884 max mem: 3535 +Test: [ 980/2387] eta: 0:04:48 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.6032 (0.4184) acc1: 67.4479 (80.3092) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.1786 data: 0.0853 max mem: 3535 +Test: [ 990/2387] eta: 0:04:46 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1393 (0.4157) acc1: 98.6979 (80.4782) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1780 data: 0.0844 max mem: 3535 +Test: [1000/2387] eta: 0:04:43 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1109 (0.4128) acc1: 100.0000 (80.6623) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1810 data: 0.0886 max mem: 3535 +Test: [1010/2387] eta: 0:04:41 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1164 (0.4100) acc1: 99.7396 (80.8430) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1800 data: 0.0885 max mem: 3535 +Test: [1020/2387] eta: 0:04:38 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1261 (0.4074) acc1: 99.4792 (81.0120) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1759 data: 0.0839 max mem: 3535 +Test: [1030/2387] eta: 0:04:36 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1125 (0.4046) acc1: 100.0000 (81.1917) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1856 data: 0.0939 max mem: 3535 +Test: [1040/2387] eta: 0:04:34 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1034 (0.4016) acc1: 100.0000 (81.3721) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1890 data: 0.0978 max mem: 3535 +Test: [1050/2387] eta: 0:04:32 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0966 (0.3988) acc1: 100.0000 (81.5488) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1842 data: 0.0909 max mem: 3535 +Test: [1060/2387] eta: 0:04:29 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0977 (0.3960) acc1: 100.0000 (81.7217) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1732 data: 0.0792 max mem: 3535 +Test: [1070/2387] eta: 0:04:27 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1041 (0.3934) acc1: 100.0000 (81.8905) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1661 data: 0.0742 max mem: 3535 +Test: [1080/2387] eta: 0:04:25 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1229 (0.3910) acc1: 100.0000 (82.0452) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1854 data: 0.0941 max mem: 3535 +Test: [1090/2387] eta: 0:04:22 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1312 (0.3887) acc1: 99.4792 (82.2036) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1954 data: 0.1030 max mem: 3535 +Test: [1100/2387] eta: 0:04:20 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1268 (0.3864) acc1: 99.4792 (82.3588) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1929 data: 0.1012 max mem: 3535 +Test: [1110/2387] eta: 0:04:18 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1328 (0.3845) acc1: 99.4792 (82.4904) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1958 data: 0.1049 max mem: 3535 +Test: [1120/2387] eta: 0:04:16 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1328 (0.3822) acc1: 99.7396 (82.6427) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2044 data: 0.1124 max mem: 3536 +Test: [1130/2387] eta: 0:04:14 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1334 (0.3800) acc1: 100.0000 (82.7941) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2088 data: 0.1152 max mem: 3536 +Test: [1140/2387] eta: 0:04:12 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1364 (0.3778) acc1: 99.7396 (82.9396) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2054 data: 0.1128 max mem: 3536 +Test: [1150/2387] eta: 0:04:10 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1328 (0.3757) acc1: 99.7396 (83.0822) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2044 data: 0.1140 max mem: 3536 +Test: [1160/2387] eta: 0:04:08 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1150 (0.3734) acc1: 99.7396 (83.2198) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2082 data: 0.1170 max mem: 3536 +Test: [1170/2387] eta: 0:04:06 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1090 (0.3711) acc1: 99.7396 (83.3589) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2037 data: 0.1101 max mem: 3536 +Test: [1180/2387] eta: 0:04:04 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1010 (0.3690) acc1: 99.7396 (83.4914) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2041 data: 0.1105 max mem: 3536 +Test: [1190/2387] eta: 0:04:02 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1036 (0.3669) acc1: 99.7396 (83.6261) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2057 data: 0.1141 max mem: 3536 +Test: [1200/2387] eta: 0:04:00 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0938 (0.3645) acc1: 100.0000 (83.7624) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2040 data: 0.1120 max mem: 3536 +Test: [1210/2387] eta: 0:03:58 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0808 (0.3622) acc1: 100.0000 (83.8959) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2016 data: 0.1089 max mem: 3536 +Test: [1220/2387] eta: 0:03:56 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0809 (0.3599) acc1: 100.0000 (84.0265) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2005 data: 0.1077 max mem: 3536 +Test: [1230/2387] eta: 0:03:54 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0748 (0.3576) acc1: 100.0000 (84.1563) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1896 data: 0.0959 max mem: 3536 +Test: [1240/2387] eta: 0:03:52 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0726 (0.3553) acc1: 100.0000 (84.2835) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1720 data: 0.0780 max mem: 3536 +Test: [1250/2387] eta: 0:03:49 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0736 (0.3530) acc1: 100.0000 (84.4087) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1778 data: 0.0850 max mem: 3536 +Test: [1260/2387] eta: 0:03:47 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0756 (0.3508) acc1: 100.0000 (84.5320) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1790 data: 0.0878 max mem: 3536 +Test: [1270/2387] eta: 0:03:45 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0717 (0.3486) acc1: 100.0000 (84.6532) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1817 data: 0.0903 max mem: 3536 +Test: [1280/2387] eta: 0:03:43 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0730 (0.3466) acc1: 100.0000 (84.7728) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1844 data: 0.0921 max mem: 3536 +Test: [1290/2387] eta: 0:03:41 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0913 (0.3447) acc1: 100.0000 (84.8817) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1781 data: 0.0858 max mem: 3536 +Test: [1300/2387] eta: 0:03:38 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0938 (0.3428) acc1: 99.7396 (84.9943) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1764 data: 0.0823 max mem: 3536 +Test: [1310/2387] eta: 0:03:36 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0938 (0.3410) acc1: 100.0000 (85.1044) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1756 data: 0.0826 max mem: 3536 +Test: [1320/2387] eta: 0:03:34 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1000 (0.3393) acc1: 100.0000 (85.2111) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1820 data: 0.0908 max mem: 3536 +Test: [1330/2387] eta: 0:03:32 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1008 (0.3375) acc1: 100.0000 (85.3183) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1779 data: 0.0858 max mem: 3536 +Test: [1340/2387] eta: 0:03:30 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1078 (0.3365) acc1: 99.7396 (85.3924) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1838 data: 0.0914 max mem: 3536 +Test: [1350/2387] eta: 0:03:28 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0621 (0.3344) acc1: 100.0000 (85.5005) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1879 data: 0.0958 max mem: 3537 +Test: [1360/2387] eta: 0:03:25 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0526 (0.3323) acc1: 100.0000 (85.6071) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1875 data: 0.0958 max mem: 3537 +Test: [1370/2387] eta: 0:03:23 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0519 (0.3303) acc1: 100.0000 (85.7120) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1975 data: 0.1058 max mem: 3537 +Test: [1380/2387] eta: 0:03:21 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0531 (0.3283) acc1: 100.0000 (85.8155) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1851 data: 0.0935 max mem: 3537 +Test: [1390/2387] eta: 0:03:19 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0537 (0.3263) acc1: 100.0000 (85.9175) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1944 data: 0.1024 max mem: 3537 +Test: [1400/2387] eta: 0:03:17 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0508 (0.3243) acc1: 100.0000 (86.0180) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2026 data: 0.1076 max mem: 3537 +Test: [1410/2387] eta: 0:03:15 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0520 (0.3229) acc1: 100.0000 (86.1082) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1961 data: 0.1015 max mem: 3537 +Test: [1420/2387] eta: 0:03:13 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1419 (0.3218) acc1: 99.4792 (86.1864) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1998 data: 0.1082 max mem: 3537 +Test: [1430/2387] eta: 0:03:11 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1464 (0.3206) acc1: 98.4375 (86.2696) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2087 data: 0.1164 max mem: 3537 +Test: [1440/2387] eta: 0:03:09 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1404 (0.3194) acc1: 99.4792 (86.3633) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2063 data: 0.1140 max mem: 3537 +Test: [1450/2387] eta: 0:03:08 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1318 (0.3181) acc1: 99.7396 (86.4524) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2172 data: 0.1260 max mem: 3537 +Test: [1460/2387] eta: 0:03:05 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1100 (0.3166) acc1: 99.2188 (86.5380) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2031 data: 0.1111 max mem: 3537 +Test: [1470/2387] eta: 0:03:03 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0903 (0.3151) acc1: 100.0000 (86.6290) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1776 data: 0.0829 max mem: 3537 +Test: [1480/2387] eta: 0:03:01 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0876 (0.3135) acc1: 100.0000 (86.7173) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1815 data: 0.0850 max mem: 3537 +Test: [1490/2387] eta: 0:02:59 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0888 (0.3121) acc1: 100.0000 (86.8040) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1901 data: 0.0956 max mem: 3537 +Test: [1500/2387] eta: 0:02:57 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0954 (0.3106) acc1: 99.7396 (86.8903) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2100 data: 0.1175 max mem: 3537 +Test: [1510/2387] eta: 0:02:55 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0929 (0.3092) acc1: 100.0000 (86.9771) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2072 data: 0.1149 max mem: 3537 +Test: [1520/2387] eta: 0:02:53 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0980 (0.3080) acc1: 100.0000 (87.0542) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1909 data: 0.0989 max mem: 3537 +Test: [1530/2387] eta: 0:02:51 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1243 (0.3068) acc1: 99.2188 (87.1358) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1874 data: 0.0952 max mem: 3537 +Test: [1540/2387] eta: 0:02:49 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1179 (0.3056) acc1: 99.7396 (87.2156) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1880 data: 0.0961 max mem: 3537 +Test: [1550/2387] eta: 0:02:47 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1168 (0.3044) acc1: 99.7396 (87.2930) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1849 data: 0.0936 max mem: 3537 +Test: [1560/2387] eta: 0:02:45 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1202 (0.3032) acc1: 99.4792 (87.3679) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1891 data: 0.0951 max mem: 3537 +Test: [1570/2387] eta: 0:02:43 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0988 (0.3019) acc1: 99.7396 (87.4475) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1901 data: 0.0957 max mem: 3537 +Test: [1580/2387] eta: 0:02:41 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0988 (0.3007) acc1: 100.0000 (87.5257) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1902 data: 0.0983 max mem: 3538 +Test: [1590/2387] eta: 0:02:39 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1189 (0.2996) acc1: 99.7396 (87.6007) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2010 data: 0.1084 max mem: 3538 +Test: [1600/2387] eta: 0:02:37 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1189 (0.2986) acc1: 99.7396 (87.6615) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2026 data: 0.1096 max mem: 3538 +Test: [1610/2387] eta: 0:02:35 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1144 (0.2975) acc1: 99.7396 (87.7355) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1911 data: 0.0993 max mem: 3538 +Test: [1620/2387] eta: 0:02:33 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1193 (0.2964) acc1: 99.7396 (87.8065) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1940 data: 0.1020 max mem: 3538 +Test: [1630/2387] eta: 0:02:31 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1276 (0.2955) acc1: 99.4792 (87.8751) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1985 data: 0.1059 max mem: 3538 +Test: [1640/2387] eta: 0:02:29 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1447 (0.2946) acc1: 98.6979 (87.9375) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2087 data: 0.1168 max mem: 3538 +Test: [1650/2387] eta: 0:02:27 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1346 (0.2937) acc1: 99.2188 (88.0043) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2061 data: 0.1130 max mem: 3538 +Test: [1660/2387] eta: 0:02:25 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1344 (0.2928) acc1: 99.2188 (88.0682) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1961 data: 0.1031 max mem: 3538 +Test: [1670/2387] eta: 0:02:23 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1471 (0.2920) acc1: 98.9583 (88.1313) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1983 data: 0.1069 max mem: 3538 +Test: [1680/2387] eta: 0:02:21 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1587 (0.2913) acc1: 98.9583 (88.1891) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2004 data: 0.1085 max mem: 3538 +Test: [1690/2387] eta: 0:02:19 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1484 (0.2905) acc1: 98.6979 (88.2469) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1956 data: 0.1040 max mem: 3538 +Test: [1700/2387] eta: 0:02:17 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1337 (0.2895) acc1: 99.4792 (88.3125) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2046 data: 0.1136 max mem: 3538 +Test: [1710/2387] eta: 0:02:15 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0890 (0.2883) acc1: 100.0000 (88.3800) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2106 data: 0.1195 max mem: 3538 +Test: [1720/2387] eta: 0:02:13 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0802 (0.2871) acc1: 100.0000 (88.4471) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2153 data: 0.1227 max mem: 3538 +Test: [1730/2387] eta: 0:02:11 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0780 (0.2859) acc1: 100.0000 (88.5122) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2128 data: 0.1195 max mem: 3538 +Test: [1740/2387] eta: 0:02:09 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0800 (0.2847) acc1: 100.0000 (88.5771) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1947 data: 0.1028 max mem: 3538 +Test: [1750/2387] eta: 0:02:07 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0836 (0.2836) acc1: 100.0000 (88.6422) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2049 data: 0.1135 max mem: 3538 +Test: [1760/2387] eta: 0:02:05 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0782 (0.2824) acc1: 100.0000 (88.7067) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2131 data: 0.1217 max mem: 3538 +Test: [1770/2387] eta: 0:02:03 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0794 (0.2813) acc1: 100.0000 (88.7678) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2166 data: 0.1250 max mem: 3538 +Test: [1780/2387] eta: 0:02:01 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0815 (0.2802) acc1: 100.0000 (88.8300) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2202 data: 0.1269 max mem: 3538 +Test: [1790/2387] eta: 0:01:59 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0749 (0.2790) acc1: 100.0000 (88.8914) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2182 data: 0.1240 max mem: 3538 +Test: [1800/2387] eta: 0:01:57 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0730 (0.2779) acc1: 100.0000 (88.9523) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2255 data: 0.1326 max mem: 3538 +Test: [1810/2387] eta: 0:01:55 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0744 (0.2768) acc1: 100.0000 (89.0132) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2203 data: 0.1265 max mem: 3539 +Test: [1820/2387] eta: 0:01:53 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0724 (0.2757) acc1: 100.0000 (89.0735) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2039 data: 0.1095 max mem: 3539 +Test: [1830/2387] eta: 0:01:51 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0731 (0.2747) acc1: 100.0000 (89.1330) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1999 data: 0.1071 max mem: 3539 +Test: [1840/2387] eta: 0:01:49 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1031 (0.2737) acc1: 100.0000 (89.1921) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1904 data: 0.0982 max mem: 3539 +Test: [1850/2387] eta: 0:01:47 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1039 (0.2728) acc1: 100.0000 (89.2505) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1947 data: 0.1033 max mem: 3539 +Test: [1860/2387] eta: 0:01:45 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1042 (0.2719) acc1: 100.0000 (89.3082) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2018 data: 0.1104 max mem: 3539 +Test: [1870/2387] eta: 0:01:43 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1045 (0.2710) acc1: 100.0000 (89.3654) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1926 data: 0.1013 max mem: 3539 +Test: [1880/2387] eta: 0:01:41 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1035 (0.2701) acc1: 100.0000 (89.4219) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1979 data: 0.1067 max mem: 3539 +Test: [1890/2387] eta: 0:01:39 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1033 (0.2693) acc1: 100.0000 (89.4778) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2025 data: 0.1113 max mem: 3539 +Test: [1900/2387] eta: 0:01:37 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1042 (0.2684) acc1: 100.0000 (89.5332) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2012 data: 0.1099 max mem: 3539 +Test: [1910/2387] eta: 0:01:35 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1066 (0.2676) acc1: 100.0000 (89.5878) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2009 data: 0.1091 max mem: 3539 +Test: [1920/2387] eta: 0:01:33 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1218 (0.2670) acc1: 100.0000 (89.6332) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2014 data: 0.1089 max mem: 3539 +Test: [1930/2387] eta: 0:01:31 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1112 (0.2661) acc1: 100.0000 (89.6857) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2043 data: 0.1122 max mem: 3539 +Test: [1940/2387] eta: 0:01:29 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1112 (0.2656) acc1: 100.0000 (89.7348) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2047 data: 0.1131 max mem: 3539 +Test: [1950/2387] eta: 0:01:27 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1152 (0.2648) acc1: 100.0000 (89.7864) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1992 data: 0.1072 max mem: 3539 +Test: [1960/2387] eta: 0:01:25 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1204 (0.2642) acc1: 100.0000 (89.8350) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2036 data: 0.1119 max mem: 3539 +Test: [1970/2387] eta: 0:01:23 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1411 (0.2637) acc1: 99.7396 (89.8755) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.2218 data: 0.1306 max mem: 3539 +Test: [1980/2387] eta: 0:01:21 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1326 (0.2631) acc1: 99.4792 (89.9174) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2051 data: 0.1143 max mem: 3539 +Test: [1990/2387] eta: 0:01:19 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1337 (0.2625) acc1: 98.9583 (89.9634) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1954 data: 0.1040 max mem: 3539 +Test: [2000/2387] eta: 0:01:17 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1394 (0.2619) acc1: 98.9583 (90.0089) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2082 data: 0.1165 max mem: 3539 +Test: [2010/2387] eta: 0:01:15 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1400 (0.2613) acc1: 99.2188 (90.0548) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2086 data: 0.1178 max mem: 3539 +Test: [2020/2387] eta: 0:01:13 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1402 (0.2607) acc1: 99.2188 (90.0995) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2241 data: 0.1328 max mem: 3539 +Test: [2030/2387] eta: 0:01:11 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1399 (0.2601) acc1: 98.9583 (90.1431) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2092 data: 0.1175 max mem: 3540 +Test: [2040/2387] eta: 0:01:09 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1380 (0.2595) acc1: 99.2188 (90.1879) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1930 data: 0.1022 max mem: 3540 +Test: [2050/2387] eta: 0:01:07 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1385 (0.2589) acc1: 99.2188 (90.2323) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2116 data: 0.1209 max mem: 3540 +Test: [2060/2387] eta: 0:01:05 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1484 (0.2584) acc1: 98.9583 (90.2729) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2148 data: 0.1237 max mem: 3540 +Test: [2070/2387] eta: 0:01:03 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1500 (0.2578) acc1: 98.9583 (90.3154) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2101 data: 0.1179 max mem: 3540 +Test: [2080/2387] eta: 0:01:01 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1473 (0.2573) acc1: 98.9583 (90.3551) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2092 data: 0.1169 max mem: 3540 +Test: [2090/2387] eta: 0:00:59 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1473 (0.2568) acc1: 98.6979 (90.3957) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2173 data: 0.1256 max mem: 3540 +Test: [2100/2387] eta: 0:00:57 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1466 (0.2563) acc1: 98.9583 (90.4366) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2228 data: 0.1314 max mem: 3540 +Test: [2110/2387] eta: 0:00:55 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1455 (0.2557) acc1: 98.9583 (90.4773) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2056 data: 0.1145 max mem: 3540 +Test: [2120/2387] eta: 0:00:53 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1442 (0.2552) acc1: 99.2188 (90.5172) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2146 data: 0.1219 max mem: 3540 +Test: [2130/2387] eta: 0:00:51 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1231 (0.2546) acc1: 100.0000 (90.5610) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2248 data: 0.1322 max mem: 3540 +Test: [2140/2387] eta: 0:00:49 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1248 (0.2542) acc1: 99.7396 (90.5911) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2140 data: 0.1230 max mem: 3540 +Test: [2150/2387] eta: 0:00:47 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1330 (0.2536) acc1: 99.7396 (90.6334) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1991 data: 0.1054 max mem: 3540 +Test: [2160/2387] eta: 0:00:45 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1312 (0.2530) acc1: 100.0000 (90.6755) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1829 data: 0.0886 max mem: 3540 +Test: [2170/2387] eta: 0:00:43 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1362 (0.2526) acc1: 100.0000 (90.7141) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1833 data: 0.0914 max mem: 3540 +Test: [2180/2387] eta: 0:00:41 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1606 (0.2529) acc1: 97.6562 (90.6947) acc5: 100.0000 (100.0000) acc1_neg: 67.708 acc5_neg: 100.000 time: 0.1698 data: 0.0784 max mem: 3540 +Test: [2190/2387] eta: 0:00:39 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.2460 (0.2530) acc1: 92.9688 (90.6931) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1568 data: 0.0660 max mem: 3540 +Test: [2200/2387] eta: 0:00:37 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1117 (0.2524) acc1: 99.4792 (90.7347) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1550 data: 0.0646 max mem: 3540 +Test: [2210/2387] eta: 0:00:35 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1029 (0.2517) acc1: 100.0000 (90.7764) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1516 data: 0.0615 max mem: 3540 +Test: [2220/2387] eta: 0:00:33 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1040 (0.2511) acc1: 100.0000 (90.8146) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1513 data: 0.0609 max mem: 3540 +Test: [2230/2387] eta: 0:00:31 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1044 (0.2504) acc1: 100.0000 (90.8550) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1485 data: 0.0580 max mem: 3540 +Test: [2240/2387] eta: 0:00:29 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1029 (0.2498) acc1: 100.0000 (90.8946) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1440 data: 0.0534 max mem: 3540 +Test: [2250/2387] eta: 0:00:27 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0981 (0.2491) acc1: 100.0000 (90.9339) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1440 data: 0.0537 max mem: 3540 +Test: [2260/2387] eta: 0:00:25 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.0968 (0.2484) acc1: 100.0000 (90.9740) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1442 data: 0.0538 max mem: 3541 +Test: [2270/2387] eta: 0:00:23 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1057 (0.2478) acc1: 100.0000 (91.0136) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1465 data: 0.0557 max mem: 3541 +Test: [2280/2387] eta: 0:00:21 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1063 (0.2473) acc1: 100.0000 (91.0405) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1390 data: 0.0485 max mem: 3541 +Test: [2290/2387] eta: 0:00:19 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1037 (0.2467) acc1: 100.0000 (91.0795) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1329 data: 0.0431 max mem: 3541 +Test: [2300/2387] eta: 0:00:17 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1037 (0.2461) acc1: 100.0000 (91.1173) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1361 data: 0.0463 max mem: 3541 +Test: [2310/2387] eta: 0:00:15 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1138 (0.2456) acc1: 100.0000 (91.1539) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1357 data: 0.0459 max mem: 3541 +Test: [2320/2387] eta: 0:00:13 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1163 (0.2451) acc1: 100.0000 (91.1885) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1366 data: 0.0468 max mem: 3541 +Test: [2330/2387] eta: 0:00:11 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1366 (0.2448) acc1: 99.7396 (91.2183) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1406 data: 0.0508 max mem: 3541 +Test: [2340/2387] eta: 0:00:09 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1371 (0.2443) acc1: 99.7396 (91.2531) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1323 data: 0.0425 max mem: 3541 +Test: [2350/2387] eta: 0:00:07 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1371 (0.2441) acc1: 99.7396 (91.2723) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1282 data: 0.0384 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1418 (0.2437) acc1: 99.7396 (91.3071) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1337 data: 0.0439 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1504 (0.2434) acc1: 100.0000 (91.3396) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1310 data: 0.0412 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1585 (0.2431) acc1: 99.4792 (91.3697) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1295 data: 0.0397 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 1.370 acc5_pos: 100.000 loss: 0.1813 (0.2430) acc1: 97.6562 (91.3769) acc5: 100.0000 (100.0000) acc1_neg: 93.023 acc5_neg: 100.000 time: 0.1239 data: 0.0368 max mem: 3541 +Test: Total time: 0:07:47 (0.1960 s / it) +* Overall Acc@1 91.377 Acc@5 100.000 + target == real Acc@1 9.108 Acc@5 100.000 + target == fake Acc@1 97.154 Acc@5 100.000 + loss 0.243 + PR-AUC 0.9859 ROC-AUC 0.8302 EER 23.95% (thr=0.8696) +Accuracy of the network on the 916267 test images: 91.4% +[rank0]:[W726 05:22:47.701238875 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:23:07.843000 1194847 torch/distributed/run.py:766] +W0726 05:23:07.843000 1194847 torch/distributed/run.py:766] ***************************************** +W0726 05:23:07.843000 1194847 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:23:07.843000 1194847 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:23:35.276268143 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:23:36.016862508 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:23:36.098013021 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:23:37.486869529 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:23:37.678608487 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank1]:[W726 05:23:37.678618702 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:23:37.700374348 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:23:37.700386819 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n23_seed8_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed8_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:07:14 acc1_pos: 9.115 acc5_pos: 100.000 loss: 1.4840 (1.4840) acc1: 9.1146 (9.1146) acc5: 100.0000 (100.0000) time: 7.8043 data: 6.2179 max mem: 3531 +Test: [ 10/517] eta: 0:06:43 acc1_pos: 1.302 acc5_pos: 100.000 loss: 1.4640 (1.4671) acc1: 6.7708 (8.6648) acc5: 100.0000 (100.0000) time: 0.7964 data: 0.5656 max mem: 3531 +Test: [ 20/517] eta: 0:03:50 acc1_pos: 5.208 acc5_pos: 100.000 loss: 1.4483 (1.4717) acc1: 7.5521 (10.0074) acc5: 100.0000 (100.0000) time: 0.0977 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:03:21 acc1_pos: 1.302 acc5_pos: 100.000 loss: 1.5426 (1.4999) acc1: 7.8125 (8.8878) acc5: 100.0000 (100.0000) time: 0.2049 data: 0.1033 max mem: 3531 +Test: [ 40/517] eta: 0:03:01 acc1_pos: 3.125 acc5_pos: 100.000 loss: 1.5514 (1.5220) acc1: 3.6458 (7.3742) acc5: 100.0000 (100.0000) time: 0.2914 data: 0.1925 max mem: 3531 +Test: [ 50/517] eta: 0:02:45 acc1_pos: 8.257 acc5_pos: 100.000 loss: 1.4972 (1.4490) acc1: 3.6458 (11.7545) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2602 data: 0.1657 max mem: 3531 +Test: [ 60/517] eta: 0:02:33 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.2249 (1.2450) acc1: 98.9583 (26.1399) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2451 data: 0.1527 max mem: 3531 +Test: [ 70/517] eta: 0:02:25 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.2026 (1.0982) acc1: 99.7396 (36.4950) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2507 data: 0.1586 max mem: 3531 +Test: [ 80/517] eta: 0:02:18 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.2058 (0.9897) acc1: 100.0000 (44.2162) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2631 data: 0.1702 max mem: 3531 +Test: [ 90/517] eta: 0:02:14 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1876 (0.9009) acc1: 99.7396 (50.3119) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2787 data: 0.1842 max mem: 3531 +Test: [100/517] eta: 0:02:08 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1786 (0.8299) acc1: 99.7396 (55.2186) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2699 data: 0.1701 max mem: 3531 +Test: [110/517] eta: 0:02:04 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1764 (0.7710) acc1: 99.7396 (59.2084) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2599 data: 0.1612 max mem: 3531 +Test: [120/517] eta: 0:01:59 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1901 (0.7239) acc1: 99.7396 (62.5366) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2649 data: 0.1704 max mem: 3531 +Test: [130/517] eta: 0:01:55 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1953 (0.6835) acc1: 99.7396 (65.3785) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2649 data: 0.1692 max mem: 3531 +Test: [140/517] eta: 0:01:51 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1906 (0.6489) acc1: 100.0000 (67.8302) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2649 data: 0.1690 max mem: 3531 +Test: [150/517] eta: 0:01:47 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1864 (0.6183) acc1: 100.0000 (69.9469) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2599 data: 0.1661 max mem: 3531 +Test: [160/517] eta: 0:01:44 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1809 (0.5909) acc1: 100.0000 (71.8119) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2599 data: 0.1643 max mem: 3531 +Test: [170/517] eta: 0:01:39 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1675 (0.5660) acc1: 100.0000 (73.4466) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2399 data: 0.1439 max mem: 3531 +Test: [180/517] eta: 0:01:35 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1648 (0.5434) acc1: 99.7396 (74.8935) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2299 data: 0.1359 max mem: 3531 +Test: [190/517] eta: 0:01:32 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1371 (0.5221) acc1: 99.7396 (76.2012) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1600 max mem: 3531 +Test: [200/517] eta: 0:01:29 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1458 (0.5035) acc1: 100.0000 (77.3761) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1591 max mem: 3531 +Test: [210/517] eta: 0:01:26 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1566 (0.4931) acc1: 98.9583 (78.1818) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2499 data: 0.1553 max mem: 3532 +Test: [220/517] eta: 0:01:23 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.2881 (0.4838) acc1: 93.4896 (78.8980) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2649 data: 0.1689 max mem: 3532 +Test: [230/517] eta: 0:01:20 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.2280 (0.4717) acc1: 96.8750 (79.7709) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2799 data: 0.1820 max mem: 3532 +Test: [240/517] eta: 0:01:17 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.2095 (0.4612) acc1: 99.7396 (80.6038) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2599 data: 0.1629 max mem: 3532 +Test: [250/517] eta: 0:01:13 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1954 (0.4500) acc1: 99.7396 (81.3631) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2349 data: 0.1391 max mem: 3532 +Test: [260/517] eta: 0:01:11 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1639 (0.4393) acc1: 99.7396 (82.0672) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2499 data: 0.1552 max mem: 3532 +Test: [270/517] eta: 0:01:08 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.2043 (0.4317) acc1: 99.7396 (82.7135) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1605 max mem: 3532 +Test: [280/517] eta: 0:01:05 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.2275 (0.4242) acc1: 99.4792 (83.2703) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2499 data: 0.1560 max mem: 3532 +Test: [290/517] eta: 0:01:02 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1997 (0.4160) acc1: 99.7396 (83.8148) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2449 data: 0.1501 max mem: 3532 +Test: [300/517] eta: 0:00:59 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1842 (0.4088) acc1: 99.7396 (84.3352) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2399 data: 0.1438 max mem: 3532 +Test: [310/517] eta: 0:00:56 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1719 (0.4009) acc1: 99.7396 (84.8364) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2449 data: 0.1477 max mem: 3532 +Test: [320/517] eta: 0:00:53 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1700 (0.3938) acc1: 100.0000 (85.3063) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2499 data: 0.1496 max mem: 3532 +Test: [330/517] eta: 0:00:50 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1674 (0.3867) acc1: 100.0000 (85.7479) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2401 data: 0.1428 max mem: 3532 +Test: [340/517] eta: 0:00:47 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1533 (0.3807) acc1: 100.0000 (86.1269) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2349 data: 0.1405 max mem: 3532 +Test: [350/517] eta: 0:00:44 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1598 (0.3754) acc1: 100.0000 (86.5103) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2359 data: 0.1396 max mem: 3532 +Test: [360/517] eta: 0:00:42 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1738 (0.3699) acc1: 100.0000 (86.8782) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2499 data: 0.1541 max mem: 3532 +Test: [370/517] eta: 0:00:39 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1738 (0.3645) acc1: 100.0000 (87.2312) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2588 data: 0.1612 max mem: 3532 +Test: [380/517] eta: 0:00:36 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1820 (0.3601) acc1: 100.0000 (87.5554) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2449 data: 0.1468 max mem: 3532 +Test: [390/517] eta: 0:00:33 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1907 (0.3558) acc1: 99.7396 (87.8690) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2499 data: 0.1536 max mem: 3532 +Test: [400/517] eta: 0:00:31 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1762 (0.3511) acc1: 100.0000 (88.1709) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1575 max mem: 3532 +Test: [410/517] eta: 0:00:28 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1750 (0.3470) acc1: 100.0000 (88.4587) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2349 data: 0.1380 max mem: 3532 +Test: [420/517] eta: 0:00:25 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1804 (0.3430) acc1: 100.0000 (88.7291) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2453 data: 0.1514 max mem: 3532 +Test: [430/517] eta: 0:00:22 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1852 (0.3397) acc1: 99.7396 (88.9779) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2349 data: 0.1415 max mem: 3532 +Test: [440/517] eta: 0:00:20 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.2052 (0.3366) acc1: 99.7396 (89.2219) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2195 data: 0.1270 max mem: 3533 +Test: [450/517] eta: 0:00:17 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1960 (0.3334) acc1: 100.0000 (89.4603) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2499 data: 0.1545 max mem: 3533 +Test: [460/517] eta: 0:00:14 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1960 (0.3306) acc1: 100.0000 (89.6873) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2549 data: 0.1571 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1721 (0.3271) acc1: 100.0000 (89.9062) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2603 data: 0.1658 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1675 (0.3238) acc1: 100.0000 (90.1155) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2814 data: 0.1878 max mem: 3533 +Test: [490/517] eta: 0:00:07 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1844 (0.3213) acc1: 100.0000 (90.3078) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.3096 data: 0.2148 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.2113 (0.3193) acc1: 99.7396 (90.4971) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2724 data: 0.1769 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1972 (0.3167) acc1: 100.0000 (90.6821) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1954 data: 0.1024 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1678 (0.3149) acc1: 100.0000 (90.7850) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1951 data: 0.1024 max mem: 3533 +Test: Total time: 0:02:14 (0.2605 s / it) +* Overall Acc@1 90.785 Acc@5 100.000 + target == real Acc@1 7.346 Acc@5 100.000 + target == fake Acc@1 99.441 Acc@5 100.000 + loss 0.315 + PR-AUC 0.9458 ROC-AUC 0.6989 EER 35.39% (thr=0.8257) +Accuracy of the network on the 198427 test images: 90.8% +[rank0]:[W726 05:26:05.315291625 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 05:26:26.056000 1195422 torch/distributed/run.py:766] +W0726 05:26:26.056000 1195422 torch/distributed/run.py:766] ***************************************** +W0726 05:26:26.056000 1195422 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:26:26.056000 1195422 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:26:55.727689748 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:26:55.000320436 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:26:55.177433014 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:26:55.287828765 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:26:55.311847975 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:26:55.359310507 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 7): env:// +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:26:56.484588181 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:26:56.490523206 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n23_seed8_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed8_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:30:27 acc1_pos: 18.229 acc5_pos: 100.000 loss: 1.4768 (1.4768) acc1: 18.2292 (18.2292) acc5: 100.0000 (100.0000) time: 4.6988 data: 2.6066 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/389] eta: 0:03:22 acc1_pos: 78.385 acc5_pos: 100.000 loss: 1.0684 (1.0984) acc1: 23.1771 (29.0246) acc5: 100.0000 (100.0000) time: 0.5338 data: 0.2620 max mem: 3531 +Test: [ 20/389] eta: 0:02:13 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.9607 (1.2262) acc1: 23.1771 (28.9683) acc5: 100.0000 (100.0000) time: 0.1439 data: 0.0531 max mem: 3531 +Test: [ 30/389] eta: 0:01:51 acc1_pos: 7.031 acc5_pos: 100.000 loss: 1.3460 (1.2987) acc1: 1.5625 (20.4301) acc5: 100.0000 (100.0000) time: 0.1890 data: 0.0956 max mem: 3531 +Test: [ 40/389] eta: 0:01:38 acc1_pos: 0.260 acc5_pos: 100.000 loss: 1.3389 (1.2882) acc1: 2.0833 (19.0803) acc5: 100.0000 (100.0000) time: 0.1990 data: 0.1055 max mem: 3531 +Test: [ 50/389] eta: 0:01:28 acc1_pos: 76.042 acc5_pos: 100.000 loss: 0.9337 (1.2233) acc1: 14.8438 (23.1056) acc5: 100.0000 (100.0000) time: 0.1849 data: 0.0927 max mem: 3531 +Test: [ 60/389] eta: 0:01:21 acc1_pos: 11.458 acc5_pos: 100.000 loss: 0.7925 (1.2031) acc1: 38.2812 (25.8197) acc5: 100.0000 (100.0000) time: 0.1749 data: 0.0831 max mem: 3531 +Test: [ 70/389] eta: 0:01:15 acc1_pos: 17.708 acc5_pos: 100.000 loss: 1.1832 (1.1950) acc1: 11.4583 (25.0954) acc5: 100.0000 (100.0000) time: 0.1749 data: 0.0833 max mem: 3531 +Test: [ 80/389] eta: 0:01:10 acc1_pos: 64.323 acc5_pos: 100.000 loss: 1.3406 (1.2059) acc1: 4.9479 (25.3537) acc5: 100.0000 (100.0000) time: 0.1720 data: 0.0779 max mem: 3531 +Test: [ 90/389] eta: 0:01:06 acc1_pos: 43.750 acc5_pos: 100.000 loss: 0.8011 (1.1514) acc1: 46.6146 (28.7317) acc5: 100.0000 (100.0000) time: 0.1669 data: 0.0727 max mem: 3531 +Test: [100/389] eta: 0:01:02 acc1_pos: 45.573 acc5_pos: 100.000 loss: 0.9183 (1.1536) acc1: 44.2708 (28.0167) acc5: 100.0000 (100.0000) time: 0.1725 data: 0.0810 max mem: 3531 +Test: [110/389] eta: 0:00:58 acc1_pos: 97.135 acc5_pos: 100.000 loss: 1.1516 (1.1502) acc1: 25.5208 (29.5983) acc5: 100.0000 (100.0000) time: 0.1584 data: 0.0643 max mem: 3531 +Test: [120/389] eta: 0:00:55 acc1_pos: 46.615 acc5_pos: 100.000 loss: 0.9054 (1.1369) acc1: 41.1458 (30.6711) acc5: 100.0000 (100.0000) time: 0.1469 data: 0.0517 max mem: 3531 +Test: [130/389] eta: 0:00:52 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.6168 (1.0927) acc1: 59.6354 (33.9695) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1594 data: 0.0636 max mem: 3531 +Test: [140/389] eta: 0:00:50 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.2218 (1.0350) acc1: 96.8750 (37.9063) acc5: 100.0000 (100.0000) acc1_neg: 53.646 acc5_neg: 100.000 time: 0.1780 data: 0.0809 max mem: 3531 +Test: [150/389] eta: 0:00:47 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.2554 (0.9903) acc1: 98.4375 (41.3718) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1660 data: 0.0721 max mem: 3531 +Test: [160/389] eta: 0:00:44 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.2944 (0.9445) acc1: 98.4375 (44.6817) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1399 data: 0.0452 max mem: 3531 +Test: [170/389] eta: 0:00:41 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.2135 (0.9034) acc1: 99.7396 (47.8542) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1457 data: 0.0507 max mem: 3531 +Test: [180/389] eta: 0:00:39 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.2240 (0.8704) acc1: 99.4792 (50.3180) acc5: 100.0000 (100.0000) acc1_neg: 76.302 acc5_neg: 100.000 time: 0.1667 data: 0.0743 max mem: 3531 +Test: [190/389] eta: 0:00:37 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.2613 (0.8411) acc1: 98.1771 (52.5633) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1530 data: 0.0619 max mem: 3531 +Test: [200/389] eta: 0:00:35 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1980 (0.8107) acc1: 99.7396 (54.7030) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1403 data: 0.0485 max mem: 3531 +Test: [210/389] eta: 0:00:33 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1674 (0.7816) acc1: 100.0000 (56.7770) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1600 data: 0.0679 max mem: 3532 +Test: [220/389] eta: 0:00:31 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1429 (0.7531) acc1: 100.0000 (58.7245) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1639 data: 0.0729 max mem: 3532 +Test: [230/389] eta: 0:00:29 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1515 (0.7318) acc1: 100.0000 (60.4144) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1612 data: 0.0696 max mem: 3532 +Test: [240/389] eta: 0:00:27 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.2722 (0.7142) acc1: 98.4375 (61.7544) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1650 data: 0.0737 max mem: 3532 +Test: [250/389] eta: 0:00:25 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.2252 (0.6920) acc1: 100.0000 (63.2771) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1648 data: 0.0725 max mem: 3532 +Test: [260/389] eta: 0:00:23 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1607 (0.6768) acc1: 100.0000 (64.2152) acc5: 100.0000 (100.0000) acc1_neg: 25.521 acc5_neg: 100.000 time: 0.1560 data: 0.0617 max mem: 3532 +Test: [270/389] eta: 0:00:21 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.2154 (0.6720) acc1: 98.4375 (64.4402) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1536 data: 0.0606 max mem: 3532 +Test: [280/389] eta: 0:00:19 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.3030 (0.6600) acc1: 94.7917 (65.3090) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1481 data: 0.0571 max mem: 3532 +Test: [290/389] eta: 0:00:17 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.2405 (0.6482) acc1: 97.9167 (66.2980) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1521 data: 0.0600 max mem: 3532 +Test: [300/389] eta: 0:00:15 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.2292 (0.6343) acc1: 99.4792 (67.4081) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1524 data: 0.0597 max mem: 3532 +Test: [310/389] eta: 0:00:13 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.2493 (0.6244) acc1: 98.1771 (68.2560) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.1527 data: 0.0609 max mem: 3532 +Test: [320/389] eta: 0:00:12 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.2843 (0.6131) acc1: 97.1354 (69.0875) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1646 data: 0.0725 max mem: 3532 +Test: [330/389] eta: 0:00:10 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1783 (0.5997) acc1: 100.0000 (69.9962) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1699 data: 0.0770 max mem: 3532 +Test: [340/389] eta: 0:00:08 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1512 (0.5871) acc1: 100.0000 (70.8249) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1870 data: 0.0943 max mem: 3532 +Test: [350/389] eta: 0:00:06 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1842 (0.5890) acc1: 99.4792 (70.5173) acc5: 100.0000 (100.0000) acc1_neg: 41.406 acc5_neg: 100.000 time: 0.1959 data: 0.1042 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.5911 (0.5902) acc1: 62.2396 (70.2512) acc5: 100.0000 (100.0000) acc1_neg: 67.708 acc5_neg: 100.000 time: 0.1861 data: 0.0947 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.5517 (0.5831) acc1: 70.8333 (70.7638) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1714 data: 0.0794 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.2379 (0.5738) acc1: 99.4792 (71.4533) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1565 data: 0.0655 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.2128 (0.5671) acc1: 99.7396 (71.9506) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1529 data: 0.0596 max mem: 3532 +Test: Total time: 0:01:08 (0.1761 s / it) +* Overall Acc@1 71.951 Acc@5 100.000 + target == real Acc@1 30.214 Acc@5 100.000 + target == fake Acc@1 91.351 Acc@5 100.000 + loss 0.567 + PR-AUC 0.8157 ROC-AUC 0.7151 EER 33.06% (thr=0.7676) +Accuracy of the network on the 149358 test images: 72.0% +[rank0]:[W726 05:28:18.271150789 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:28:38.019000 1195983 torch/distributed/run.py:766] +W0726 05:28:38.019000 1195983 torch/distributed/run.py:766] ***************************************** +W0726 05:28:38.019000 1195983 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:28:38.019000 1195983 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:29:06.513055796 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:29:06.244730284 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:29:07.418635851 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:29:07.431071219 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank4]:[W726 05:29:07.434634679 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 3): env:// +[rank0]:[W726 05:29:07.517440178 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:29:07.518070095 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:29:07.522915064 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n23_seed8_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed8_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:35:00 acc1_pos: 85.156 acc5_pos: 100.000 loss: 0.3194 (0.3194) acc1: 85.1562 (85.1562) acc5: 100.0000 (100.0000) time: 6.3090 data: 4.8510 max mem: 3531 +Test: [ 10/333] eta: 0:03:32 acc1_pos: 89.323 acc5_pos: 100.000 loss: 0.2337 (0.2523) acc1: 90.8854 (90.8854) acc5: 100.0000 (100.0000) time: 0.6586 data: 0.4414 max mem: 3531 +Test: [ 20/333] eta: 0:02:04 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.2818 (0.2668) acc1: 90.8854 (90.8358) acc5: 100.0000 (100.0000) time: 0.1018 data: 0.0087 max mem: 3531 +Test: [ 30/333] eta: 0:01:42 acc1_pos: 51.823 acc5_pos: 100.000 loss: 0.3735 (0.4811) acc1: 75.2604 (75.8233) acc5: 100.0000 (100.0000) time: 0.1599 data: 0.0654 max mem: 3531 +Test: [ 40/333] eta: 0:01:29 acc1_pos: 38.021 acc5_pos: 100.000 loss: 0.9958 (0.6143) acc1: 40.8854 (66.1077) acc5: 100.0000 (100.0000) time: 0.2099 data: 0.1145 max mem: 3531 +Test: [ 50/333] eta: 0:01:20 acc1_pos: 42.448 acc5_pos: 100.000 loss: 1.0012 (0.6960) acc1: 38.0208 (60.5443) acc5: 100.0000 (100.0000) time: 0.2026 data: 0.1096 max mem: 3531 +Test: [ 60/333] eta: 0:01:13 acc1_pos: 33.594 acc5_pos: 100.000 loss: 0.9926 (0.7430) acc1: 38.2812 (57.0014) acc5: 100.0000 (100.0000) time: 0.1977 data: 0.1032 max mem: 3531 +Test: [ 70/333] eta: 0:01:08 acc1_pos: 41.667 acc5_pos: 100.000 loss: 0.9891 (0.7767) acc1: 38.8021 (54.9809) acc5: 100.0000 (100.0000) time: 0.1998 data: 0.1054 max mem: 3531 +Test: [ 80/333] eta: 0:01:05 acc1_pos: 42.708 acc5_pos: 100.000 loss: 0.9622 (0.8012) acc1: 40.6250 (52.9225) acc5: 100.0000 (100.0000) time: 0.2191 data: 0.1286 max mem: 3531 +Test: [ 90/333] eta: 0:00:59 acc1_pos: 52.865 acc5_pos: 100.000 loss: 0.8815 (0.8095) acc1: 40.6250 (52.1606) acc5: 100.0000 (100.0000) time: 0.1938 data: 0.1033 max mem: 3531 +Test: [100/333] eta: 0:00:55 acc1_pos: 50.000 acc5_pos: 100.000 loss: 0.8693 (0.8210) acc1: 46.6146 (51.3717) acc5: 100.0000 (100.0000) time: 0.1536 data: 0.0620 max mem: 3531 +Test: [110/333] eta: 0:00:52 acc1_pos: 47.917 acc5_pos: 100.000 loss: 0.9009 (0.8282) acc1: 42.7083 (50.7695) acc5: 100.0000 (100.0000) time: 0.1848 data: 0.0934 max mem: 3531 +Test: [120/333] eta: 0:00:48 acc1_pos: 53.906 acc5_pos: 100.000 loss: 0.8810 (0.8340) acc1: 44.7917 (50.3121) acc5: 100.0000 (100.0000) time: 0.1747 data: 0.0837 max mem: 3531 +Test: [130/333] eta: 0:00:45 acc1_pos: 58.333 acc5_pos: 100.000 loss: 0.8511 (0.8354) acc1: 48.6979 (50.1153) acc5: 100.0000 (100.0000) time: 0.1701 data: 0.0789 max mem: 3531 +Test: [140/333] eta: 0:00:42 acc1_pos: 66.406 acc5_pos: 100.000 loss: 0.8657 (0.8379) acc1: 43.7500 (49.8005) acc5: 100.0000 (100.0000) time: 0.1851 data: 0.0940 max mem: 3531 +Test: [150/333] eta: 0:00:40 acc1_pos: 43.229 acc5_pos: 100.000 loss: 0.8921 (0.8431) acc1: 42.1875 (49.3722) acc5: 100.0000 (100.0000) time: 0.1784 data: 0.0861 max mem: 3531 +Test: [160/333] eta: 0:00:37 acc1_pos: 32.566 acc5_pos: 100.000 loss: 0.8832 (0.8457) acc1: 43.2292 (49.2155) acc5: 100.0000 (100.0000) acc1_neg: 54.427 acc5_neg: 100.000 time: 0.1843 data: 0.0921 max mem: 3531 +Test: [170/333] eta: 0:00:34 acc1_pos: 32.566 acc5_pos: 100.000 loss: 0.7965 (0.8379) acc1: 53.9062 (49.9543) acc5: 100.0000 (100.0000) acc1_neg: 65.885 acc5_neg: 100.000 time: 0.1776 data: 0.0857 max mem: 3531 +Test: [180/333] eta: 0:00:32 acc1_pos: 32.566 acc5_pos: 100.000 loss: 0.6372 (0.8250) acc1: 66.4062 (50.9740) acc5: 100.0000 (100.0000) acc1_neg: 69.010 acc5_neg: 100.000 time: 0.1751 data: 0.0806 max mem: 3531 +Test: [190/333] eta: 0:00:29 acc1_pos: 32.566 acc5_pos: 100.000 loss: 0.5892 (0.8107) acc1: 70.0521 (52.1270) acc5: 100.0000 (100.0000) acc1_neg: 70.052 acc5_neg: 100.000 time: 0.1730 data: 0.0774 max mem: 3531 +Test: [200/333] eta: 0:00:27 acc1_pos: 32.566 acc5_pos: 100.000 loss: 0.5651 (0.8003) acc1: 70.0521 (53.0110) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.1714 data: 0.0776 max mem: 3531 +Test: [210/333] eta: 0:00:25 acc1_pos: 32.566 acc5_pos: 100.000 loss: 0.6468 (0.7952) acc1: 65.1042 (53.4767) acc5: 100.0000 (100.0000) acc1_neg: 50.781 acc5_neg: 100.000 time: 0.1669 data: 0.0747 max mem: 3532 +Test: [220/333] eta: 0:00:23 acc1_pos: 32.566 acc5_pos: 100.000 loss: 0.6810 (0.7916) acc1: 61.9792 (53.8391) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.1752 data: 0.0812 max mem: 3532 +Test: [230/333] eta: 0:00:21 acc1_pos: 32.566 acc5_pos: 100.000 loss: 0.4565 (0.7757) acc1: 78.6458 (55.0449) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.1870 data: 0.0930 max mem: 3532 +Test: [240/333] eta: 0:00:18 acc1_pos: 32.566 acc5_pos: 100.000 loss: 0.4441 (0.7640) acc1: 80.7292 (55.9496) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.1828 data: 0.0896 max mem: 3532 +Test: [250/333] eta: 0:00:16 acc1_pos: 32.566 acc5_pos: 100.000 loss: 0.5321 (0.7560) acc1: 73.1771 (56.5789) acc5: 100.0000 (100.0000) acc1_neg: 62.760 acc5_neg: 100.000 time: 0.1699 data: 0.0761 max mem: 3532 +Test: [260/333] eta: 0:00:14 acc1_pos: 32.566 acc5_pos: 100.000 loss: 0.5767 (0.7504) acc1: 71.0938 (57.0273) acc5: 100.0000 (100.0000) acc1_neg: 65.885 acc5_neg: 100.000 time: 0.1716 data: 0.0773 max mem: 3532 +Test: [270/333] eta: 0:00:12 acc1_pos: 32.566 acc5_pos: 100.000 loss: 0.5482 (0.7422) acc1: 72.6562 (57.6972) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.1776 data: 0.0813 max mem: 3532 +Test: [280/333] eta: 0:00:10 acc1_pos: 32.566 acc5_pos: 100.000 loss: 0.5399 (0.7389) acc1: 72.9167 (57.9561) acc5: 100.0000 (100.0000) acc1_neg: 68.229 acc5_neg: 100.000 time: 0.1832 data: 0.0879 max mem: 3532 +Test: [290/333] eta: 0:00:08 acc1_pos: 32.566 acc5_pos: 100.000 loss: 0.7294 (0.7394) acc1: 56.5104 (57.9270) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.1823 data: 0.0876 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 32.566 acc5_pos: 100.000 loss: 0.7292 (0.7390) acc1: 58.5938 (58.0141) acc5: 100.0000 (100.0000) acc1_neg: 61.979 acc5_neg: 100.000 time: 0.1657 data: 0.0724 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 32.566 acc5_pos: 100.000 loss: 0.6538 (0.7327) acc1: 63.0208 (58.5058) acc5: 100.0000 (100.0000) acc1_neg: 73.958 acc5_neg: 100.000 time: 0.1658 data: 0.0746 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 32.566 acc5_pos: 100.000 loss: 0.5115 (0.7267) acc1: 74.4792 (58.9832) acc5: 100.0000 (100.0000) acc1_neg: 57.812 acc5_neg: 100.000 time: 0.1646 data: 0.0716 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 32.566 acc5_pos: 100.000 loss: 0.6847 (0.7283) acc1: 58.3333 (58.9171) acc5: 100.0000 (100.0000) acc1_neg: 58.073 acc5_neg: 100.000 time: 0.1534 data: 0.0611 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 32.566 acc5_pos: 100.000 loss: 0.7599 (0.7286) acc1: 57.8125 (58.9043) acc5: 100.0000 (100.0000) acc1_neg: 56.901 acc5_neg: 100.000 time: 0.1516 data: 0.0562 max mem: 3532 +Test: Total time: 0:01:04 (0.1938 s / it) +* Overall Acc@1 58.904 Acc@5 100.000 + target == real Acc@1 49.130 Acc@5 100.000 + target == fake Acc@1 67.606 Acc@5 100.000 + loss 0.729 + PR-AUC 0.6276 ROC-AUC 0.6189 EER 41.64% (thr=0.5850) +Accuracy of the network on the 127843 test images: 58.9% +[rank0]:[W726 05:30:23.090876845 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:30:43.409000 1196529 torch/distributed/run.py:766] +W0726 05:30:43.409000 1196529 torch/distributed/run.py:766] ***************************************** +W0726 05:30:43.409000 1196529 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:30:43.409000 1196529 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:31:10.068485785 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:31:11.751878192 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +[rank5]:[W726 05:31:11.903499422 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank4]:[W726 05:31:11.904608508 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:31:11.912093997 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +NCCL version 2.26.2+cuda12.2 +[rank1]:[W726 05:31:11.070182251 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:31:11.145639979 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W726 05:31:11.145711648 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n23_seed8_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed8_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:22:39 acc1_pos: 87.500 acc5_pos: 100.000 loss: 0.3364 (0.3364) acc1: 87.5000 (87.5000) acc5: 100.0000 (100.0000) time: 5.7855 data: 3.1437 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/235] eta: 0:02:16 acc1_pos: 90.104 acc5_pos: 100.000 loss: 0.3328 (0.3335) acc1: 87.7604 (88.0682) acc5: 100.0000 (100.0000) time: 0.6077 data: 0.2861 max mem: 3531 +Test: [ 20/235] eta: 0:01:18 acc1_pos: 88.281 acc5_pos: 100.000 loss: 0.3246 (0.3259) acc1: 88.2812 (88.6533) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0068 max mem: 3531 +Test: [ 30/235] eta: 0:00:57 acc1_pos: 90.625 acc5_pos: 100.000 loss: 0.3160 (0.3236) acc1: 89.5833 (88.7685) acc5: 100.0000 (100.0000) time: 0.0975 data: 0.0069 max mem: 3531 +Test: [ 40/235] eta: 0:00:48 acc1_pos: 91.146 acc5_pos: 100.000 loss: 0.3152 (0.3224) acc1: 89.3229 (88.8529) acc5: 100.0000 (100.0000) time: 0.1274 data: 0.0360 max mem: 3531 +Test: [ 50/235] eta: 0:00:44 acc1_pos: 84.896 acc5_pos: 100.000 loss: 0.3182 (0.3237) acc1: 88.2812 (88.6897) acc5: 100.0000 (100.0000) time: 0.1749 data: 0.0782 max mem: 3531 +Test: [ 60/235] eta: 0:00:39 acc1_pos: 89.844 acc5_pos: 100.000 loss: 0.3265 (0.3238) acc1: 87.7604 (88.7380) acc5: 100.0000 (100.0000) time: 0.1734 data: 0.0760 max mem: 3531 +Test: [ 70/235] eta: 0:00:36 acc1_pos: 89.231 acc5_pos: 100.000 loss: 0.3420 (0.4173) acc1: 86.9792 (82.3540) acc5: 100.0000 (100.0000) acc1_neg: 35.156 acc5_neg: 100.000 time: 0.1849 data: 0.0904 max mem: 3531 +Test: [ 80/235] eta: 0:00:33 acc1_pos: 89.231 acc5_pos: 100.000 loss: 1.0782 (0.5015) acc1: 37.7604 (76.7265) acc5: 100.0000 (100.0000) acc1_neg: 34.896 acc5_neg: 100.000 time: 0.1999 data: 0.1030 max mem: 3531 +Test: [ 90/235] eta: 0:00:30 acc1_pos: 89.231 acc5_pos: 100.000 loss: 1.1026 (0.5665) acc1: 35.6771 (72.3644) acc5: 100.0000 (100.0000) acc1_neg: 39.323 acc5_neg: 100.000 time: 0.1702 data: 0.0740 max mem: 3531 +Test: [100/235] eta: 0:00:27 acc1_pos: 89.231 acc5_pos: 100.000 loss: 1.0877 (0.6165) acc1: 36.9792 (68.9666) acc5: 100.0000 (100.0000) acc1_neg: 38.542 acc5_neg: 100.000 time: 0.1549 data: 0.0611 max mem: 3531 +Test: [110/235] eta: 0:00:25 acc1_pos: 89.231 acc5_pos: 100.000 loss: 1.0738 (0.6592) acc1: 37.7604 (66.1505) acc5: 100.0000 (100.0000) acc1_neg: 38.542 acc5_neg: 100.000 time: 0.1546 data: 0.0617 max mem: 3531 +Test: [120/235] eta: 0:00:22 acc1_pos: 89.231 acc5_pos: 100.000 loss: 1.1307 (0.7176) acc1: 32.8125 (62.5732) acc5: 100.0000 (100.0000) acc1_neg: 23.958 acc5_neg: 100.000 time: 0.1499 data: 0.0555 max mem: 3531 +Test: [130/235] eta: 0:00:20 acc1_pos: 89.231 acc5_pos: 100.000 loss: 1.3814 (0.7684) acc1: 21.3542 (59.4645) acc5: 100.0000 (100.0000) acc1_neg: 22.135 acc5_neg: 100.000 time: 0.1499 data: 0.0552 max mem: 3531 +Test: [140/235] eta: 0:00:18 acc1_pos: 89.231 acc5_pos: 100.000 loss: 1.4198 (0.8155) acc1: 20.5729 (56.6434) acc5: 100.0000 (100.0000) acc1_neg: 16.927 acc5_neg: 100.000 time: 0.1549 data: 0.0606 max mem: 3531 +Test: [150/235] eta: 0:00:16 acc1_pos: 89.231 acc5_pos: 100.000 loss: 1.4187 (0.8538) acc1: 20.5729 (54.3150) acc5: 100.0000 (100.0000) acc1_neg: 19.010 acc5_neg: 100.000 time: 0.1599 data: 0.0658 max mem: 3531 +Test: [160/235] eta: 0:00:14 acc1_pos: 89.231 acc5_pos: 100.000 loss: 1.3991 (0.8878) acc1: 21.3542 (52.2402) acc5: 100.0000 (100.0000) acc1_neg: 23.698 acc5_neg: 100.000 time: 0.1649 data: 0.0700 max mem: 3531 +Test: [170/235] eta: 0:00:12 acc1_pos: 89.231 acc5_pos: 100.000 loss: 1.3318 (0.8851) acc1: 22.6562 (52.2067) acc5: 100.0000 (100.0000) acc1_neg: 52.865 acc5_neg: 100.000 time: 0.1599 data: 0.0633 max mem: 3531 +Test: [180/235] eta: 0:00:10 acc1_pos: 89.231 acc5_pos: 100.000 loss: 0.7806 (0.8792) acc1: 55.7292 (52.4315) acc5: 100.0000 (100.0000) acc1_neg: 53.646 acc5_neg: 100.000 time: 0.1499 data: 0.0516 max mem: 3531 +Test: [190/235] eta: 0:00:08 acc1_pos: 89.231 acc5_pos: 100.000 loss: 0.7817 (0.8741) acc1: 55.7292 (52.5892) acc5: 100.0000 (100.0000) acc1_neg: 55.469 acc5_neg: 100.000 time: 0.1549 data: 0.0553 max mem: 3531 +Test: [200/235] eta: 0:00:06 acc1_pos: 89.231 acc5_pos: 100.000 loss: 0.7825 (0.8700) acc1: 53.6458 (52.6534) acc5: 100.0000 (100.0000) acc1_neg: 52.604 acc5_neg: 100.000 time: 0.1649 data: 0.0676 max mem: 3531 +Test: [210/235] eta: 0:00:04 acc1_pos: 89.231 acc5_pos: 100.000 loss: 0.7863 (0.8654) acc1: 53.6458 (52.8387) acc5: 100.0000 (100.0000) acc1_neg: 59.115 acc5_neg: 100.000 time: 0.1649 data: 0.0676 max mem: 3532 +Test: [220/235] eta: 0:00:02 acc1_pos: 89.231 acc5_pos: 100.000 loss: 0.7680 (0.8608) acc1: 56.7708 (53.0437) acc5: 100.0000 (100.0000) acc1_neg: 56.771 acc5_neg: 100.000 time: 0.1585 data: 0.0630 max mem: 3532 +Test: [230/235] eta: 0:00:00 acc1_pos: 89.231 acc5_pos: 100.000 loss: 0.7684 (0.8571) acc1: 56.2500 (53.1645) acc5: 100.0000 (100.0000) acc1_neg: 54.688 acc5_neg: 100.000 time: 0.1544 data: 0.0598 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 89.231 acc5_pos: 100.000 loss: 0.7707 (0.8564) acc1: 56.2500 (53.1871) acc5: 100.0000 (100.0000) acc1_neg: 43.478 acc5_neg: 100.000 time: 0.1218 data: 0.0302 max mem: 3532 +Test: Total time: 0:00:41 (0.1781 s / it) +* Overall Acc@1 53.187 Acc@5 100.000 + target == real Acc@1 88.732 Acc@5 100.000 + target == fake Acc@1 40.331 Acc@5 100.000 + loss 0.856 + PR-AUC 0.8797 ROC-AUC 0.7298 EER 32.89% (thr=0.2600) +Accuracy of the network on the 89879 test images: 53.2% +[rank0]:[W726 05:32:11.071953986 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:32:32.197000 1197088 torch/distributed/run.py:766] +W0726 05:32:32.197000 1197088 torch/distributed/run.py:766] ***************************************** +W0726 05:32:32.197000 1197088 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:32:32.197000 1197088 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:33:01.378837521 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:33:02.781781872 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:33:02.960598859 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:33:02.972250015 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:33:02.037158296 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:33:02.089923575 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:33:02.192540069 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:33:02.224489626 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n23_seed8_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed8_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:35:08 acc1_pos: 25.521 acc5_pos: 100.000 loss: 1.0682 (1.0682) acc1: 25.5208 (25.5208) acc5: 100.0000 (100.0000) time: 7.8389 data: 5.6591 max mem: 3531 +Test: [ 10/269] eta: 0:03:27 acc1_pos: 60.156 acc5_pos: 100.000 loss: 0.7896 (0.7865) acc1: 50.2604 (50.2367) acc5: 100.0000 (100.0000) time: 0.7999 data: 0.5149 max mem: 3531 +Test: [ 20/269] eta: 0:01:56 acc1_pos: 51.823 acc5_pos: 100.000 loss: 0.7506 (0.7626) acc1: 50.5208 (51.6617) acc5: 100.0000 (100.0000) time: 0.0980 data: 0.0004 max mem: 3531 +Test: [ 30/269] eta: 0:01:34 acc1_pos: 44.271 acc5_pos: 100.000 loss: 0.7155 (0.7550) acc1: 51.8229 (52.4866) acc5: 100.0000 (100.0000) time: 0.1722 data: 0.0732 max mem: 3531 +Test: [ 40/269] eta: 0:01:20 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.5275 (0.6754) acc1: 69.5312 (59.8831) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.2326 data: 0.1379 max mem: 3531 +Test: [ 50/269] eta: 0:01:12 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.4439 (0.6412) acc1: 80.9896 (63.2915) acc5: 100.0000 (100.0000) acc1_neg: 63.021 acc5_neg: 100.000 time: 0.2326 data: 0.1399 max mem: 3531 +Test: [ 60/269] eta: 0:01:05 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.5255 (0.6266) acc1: 74.7396 (64.7413) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.2290 data: 0.1343 max mem: 3531 +Test: [ 70/269] eta: 0:00:59 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.4723 (0.6003) acc1: 75.5208 (67.3269) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.2149 data: 0.1218 max mem: 3531 +Test: [ 80/269] eta: 0:00:54 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.4271 (0.5811) acc1: 84.1146 (69.2098) acc5: 100.0000 (100.0000) acc1_neg: 72.917 acc5_neg: 100.000 time: 0.2138 data: 0.1185 max mem: 3531 +Test: [ 90/269] eta: 0:00:49 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.5358 (0.5864) acc1: 71.8750 (68.4982) acc5: 100.0000 (100.0000) acc1_neg: 69.010 acc5_neg: 100.000 time: 0.2099 data: 0.1144 max mem: 3531 +Test: [100/269] eta: 0:00:45 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.5880 (0.5833) acc1: 68.4896 (68.9485) acc5: 100.0000 (100.0000) acc1_neg: 73.958 acc5_neg: 100.000 time: 0.2093 data: 0.1153 max mem: 3531 +Test: [110/269] eta: 0:00:42 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.5880 (0.5834) acc1: 68.2292 (68.9072) acc5: 100.0000 (100.0000) acc1_neg: 64.844 acc5_neg: 100.000 time: 0.2207 data: 0.1271 max mem: 3531 +Test: [120/269] eta: 0:00:39 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.5326 (0.5734) acc1: 72.1354 (69.9531) acc5: 100.0000 (100.0000) acc1_neg: 77.344 acc5_neg: 100.000 time: 0.2098 data: 0.1176 max mem: 3531 +Test: [130/269] eta: 0:00:35 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.4944 (0.5729) acc1: 77.3438 (69.9209) acc5: 100.0000 (100.0000) acc1_neg: 65.625 acc5_neg: 100.000 time: 0.2056 data: 0.1134 max mem: 3531 +Test: [140/269] eta: 0:00:33 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.5570 (0.5736) acc1: 67.9688 (69.7769) acc5: 100.0000 (100.0000) acc1_neg: 55.469 acc5_neg: 100.000 time: 0.2345 data: 0.1425 max mem: 3531 +Test: [150/269] eta: 0:00:30 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.5437 (0.5722) acc1: 66.6667 (69.8538) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.2180 data: 0.1241 max mem: 3531 +Test: [160/269] eta: 0:00:27 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.5242 (0.5686) acc1: 72.6562 (70.1605) acc5: 100.0000 (100.0000) acc1_neg: 74.740 acc5_neg: 100.000 time: 0.2005 data: 0.1073 max mem: 3531 +Test: [170/269] eta: 0:00:24 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.5628 (0.5690) acc1: 72.6562 (70.1084) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.2204 data: 0.1271 max mem: 3531 +Test: [180/269] eta: 0:00:22 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.5594 (0.5674) acc1: 72.9167 (70.2780) acc5: 100.0000 (100.0000) acc1_neg: 64.062 acc5_neg: 100.000 time: 0.2243 data: 0.1314 max mem: 3531 +Test: [190/269] eta: 0:00:19 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.5172 (0.5667) acc1: 73.4375 (70.3139) acc5: 100.0000 (100.0000) acc1_neg: 73.698 acc5_neg: 100.000 time: 0.2059 data: 0.1142 max mem: 3531 +Test: [200/269] eta: 0:00:16 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.5264 (0.5654) acc1: 72.3958 (70.4343) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.2124 data: 0.1183 max mem: 3531 +Test: [210/269] eta: 0:00:14 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.5273 (0.5624) acc1: 72.3958 (70.6457) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.2084 data: 0.1144 max mem: 3532 +Test: [220/269] eta: 0:00:11 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.4985 (0.5596) acc1: 72.6562 (70.8451) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1711 data: 0.0796 max mem: 3532 +Test: [230/269] eta: 0:00:09 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.4289 (0.5545) acc1: 79.9479 (71.2798) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.1480 data: 0.0577 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.5078 (0.5572) acc1: 75.0000 (71.0754) acc5: 100.0000 (100.0000) acc1_neg: 68.229 acc5_neg: 100.000 time: 0.1325 data: 0.0427 max mem: 3532 +Test: [250/269] eta: 0:00:04 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.5581 (0.5582) acc1: 68.7500 (70.9983) acc5: 100.0000 (100.0000) acc1_neg: 61.198 acc5_neg: 100.000 time: 0.1205 data: 0.0308 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.5624 (0.5619) acc1: 68.7500 (70.6727) acc5: 100.0000 (100.0000) acc1_neg: 72.135 acc5_neg: 100.000 time: 0.1110 data: 0.0215 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 37.069 acc5_pos: 100.000 loss: 0.5410 (0.5599) acc1: 71.8750 (70.9341) acc5: 100.0000 (100.0000) acc1_neg: 94.226 acc5_neg: 100.000 time: 0.1210 data: 0.0279 max mem: 3532 +Test: Total time: 0:00:58 (0.2177 s / it) +* Overall Acc@1 70.934 Acc@5 100.000 + target == real Acc@1 51.911 Acc@5 100.000 + target == fake Acc@1 73.530 Acc@5 100.000 + loss 0.560 + PR-AUC 0.9350 ROC-AUC 0.6894 EER 35.94% (thr=0.5659) +Accuracy of the network on the 103293 test images: 70.9% +[rank0]:[W726 05:34:08.098299960 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T05:34:16Z] eval_done tag=fixed500k_real_n23_seed8_resampled +[2026-07-26T05:34:16Z] eval_start tag=fixed500k_fake_n45_seed8_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed8_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T05:34:16Z +hostname=n116-041-023 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed8_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n45_seed8_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 05:34:29.457000 1197665 torch/distributed/run.py:766] +W0726 05:34:29.457000 1197665 torch/distributed/run.py:766] ***************************************** +W0726 05:34:29.457000 1197665 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:34:29.457000 1197665 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:34:57.265110041 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:34:58.567156690 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:34:58.851223369 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:34:58.891604113 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:34:58.125672560 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:34:58.159226790 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W726 05:34:58.161602136 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:34:58.198985291 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n45_seed8_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed8_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:20:45 acc1_pos: 45.833 acc5_pos: 100.000 loss: 0.7841 (0.7841) acc1: 45.8333 (45.8333) acc5: 100.0000 (100.0000) time: 5.3222 data: 3.5747 max mem: 3531 +Test: [ 10/234] eta: 0:02:06 acc1_pos: 25.781 acc5_pos: 100.000 loss: 0.8602 (0.9072) acc1: 37.7604 (36.6241) acc5: 100.0000 (100.0000) time: 0.5658 data: 0.3253 max mem: 3531 +Test: [ 20/234] eta: 0:01:15 acc1_pos: 95.052 acc5_pos: 100.000 loss: 0.8602 (0.8545) acc1: 37.7604 (42.2371) acc5: 100.0000 (100.0000) time: 0.1066 data: 0.0126 max mem: 3531 +Test: [ 30/234] eta: 0:01:00 acc1_pos: 67.188 acc5_pos: 100.000 loss: 0.3190 (0.6845) acc1: 91.1458 (57.5101) acc5: 100.0000 (100.0000) time: 0.1465 data: 0.0505 max mem: 3531 +Test: [ 40/234] eta: 0:00:53 acc1_pos: 98.958 acc5_pos: 100.000 loss: 0.3131 (0.6380) acc1: 92.4479 (60.8994) acc5: 100.0000 (100.0000) time: 0.1899 data: 0.0939 max mem: 3531 +Test: [ 50/234] eta: 0:00:47 acc1_pos: 99.219 acc5_pos: 100.000 loss: 0.1683 (0.5449) acc1: 97.6562 (68.2649) acc5: 100.0000 (100.0000) time: 0.2050 data: 0.1103 max mem: 3531 +Test: [ 60/234] eta: 0:00:44 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.1616 (0.4818) acc1: 99.2188 (73.4418) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2250 data: 0.1315 max mem: 3531 +Test: [ 70/234] eta: 0:00:40 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.1646 (0.4928) acc1: 99.7396 (73.0670) acc5: 100.0000 (100.0000) acc1_neg: 53.906 acc5_neg: 100.000 time: 0.2150 data: 0.1201 max mem: 3531 +Test: [ 80/234] eta: 0:00:36 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.7351 (0.5039) acc1: 57.2917 (72.7013) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1703 data: 0.0768 max mem: 3531 +Test: [ 90/234] eta: 0:00:33 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.3390 (0.4825) acc1: 85.1562 (74.6938) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.1726 data: 0.0793 max mem: 3531 +Test: [100/234] eta: 0:00:30 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.3662 (0.4848) acc1: 84.3750 (74.7757) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1815 data: 0.0885 max mem: 3531 +Test: [110/234] eta: 0:00:27 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4720 (0.4846) acc1: 73.9583 (75.0493) acc5: 100.0000 (100.0000) acc1_neg: 66.146 acc5_neg: 100.000 time: 0.1722 data: 0.0802 max mem: 3531 +Test: [120/234] eta: 0:00:24 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4302 (0.4879) acc1: 81.2500 (75.0495) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1628 data: 0.0701 max mem: 3531 +Test: [130/234] eta: 0:00:22 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.3663 (0.4803) acc1: 88.2812 (75.8926) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.1697 data: 0.0734 max mem: 3531 +Test: [140/234] eta: 0:00:19 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4122 (0.4982) acc1: 80.4688 (75.0831) acc5: 100.0000 (100.0000) acc1_neg: 35.938 acc5_neg: 100.000 time: 0.1932 data: 0.0968 max mem: 3531 +Test: [150/234] eta: 0:00:17 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.6003 (0.5117) acc1: 67.1875 (74.2774) acc5: 100.0000 (100.0000) acc1_neg: 70.573 acc5_neg: 100.000 time: 0.1969 data: 0.1012 max mem: 3531 +Test: [160/234] eta: 0:00:15 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4936 (0.5061) acc1: 76.3021 (74.8787) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.2115 data: 0.1140 max mem: 3531 +Test: [170/234] eta: 0:00:13 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4644 (0.5030) acc1: 80.4688 (75.2482) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.2049 data: 0.1096 max mem: 3531 +Test: [180/234] eta: 0:00:11 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4798 (0.5128) acc1: 77.8646 (74.4979) acc5: 100.0000 (100.0000) acc1_neg: 26.042 acc5_neg: 100.000 time: 0.1828 data: 0.0906 max mem: 3531 +Test: [190/234] eta: 0:00:09 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.5315 (0.5153) acc1: 71.6146 (74.3374) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.1779 data: 0.0840 max mem: 3531 +Test: [200/234] eta: 0:00:06 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.5136 (0.5020) acc1: 75.7812 (75.1503) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1671 data: 0.0709 max mem: 3531 +Test: [210/234] eta: 0:00:04 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.0817 (0.4847) acc1: 99.7396 (76.2897) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1739 data: 0.0778 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.2476 (0.4751) acc1: 97.9167 (77.1646) acc5: 100.0000 (100.0000) acc1_neg: 76.302 acc5_neg: 100.000 time: 0.1505 data: 0.0575 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4655 (0.4932) acc1: 39.0625 (75.3190) acc5: 100.0000 (100.0000) acc1_neg: 34.375 acc5_neg: 100.000 time: 0.1211 data: 0.0308 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.8787 (0.4982) acc1: 35.9375 (74.8766) acc5: 100.0000 (100.0000) acc1_neg: 39.689 acc5_neg: 100.000 time: 0.1058 data: 0.0143 max mem: 3532 +Test: Total time: 0:00:45 (0.1958 s / it) +* Overall Acc@1 74.877 Acc@5 100.000 + target == real Acc@1 68.483 Acc@5 100.000 + target == fake Acc@1 76.679 Acc@5 100.000 + loss 0.498 + PR-AUC 0.9415 ROC-AUC 0.8241 EER 26.84% (thr=0.5430) +Accuracy of the network on the 89729 test images: 74.9% +[rank0]:[W726 05:35:56.684338914 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:36:15.627000 1198211 torch/distributed/run.py:766] +W0726 05:36:15.627000 1198211 torch/distributed/run.py:766] ***************************************** +W0726 05:36:15.627000 1198211 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:36:15.627000 1198211 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:36:46.897957340 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:36:46.975368820 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:36:46.239963831 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:36:46.311653944 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:36:47.490758015 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:36:47.637604019 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:36:47.685881760 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:36:47.698292919 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n45_seed8_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed8_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:29:33 acc1_pos: 10.677 acc5_pos: 100.000 loss: 1.3080 (1.3080) acc1: 10.6771 (10.6771) acc5: 100.0000 (100.0000) time: 6.7755 data: 4.8175 max mem: 3531 +Test: [ 10/2387] eta: 0:27:39 acc1_pos: 2.344 acc5_pos: 100.000 loss: 1.2910 (1.2873) acc1: 7.5521 (11.5294) acc5: 100.0000 (100.0000) time: 0.6980 data: 0.4384 max mem: 3531 +Test: [ 20/2387] eta: 0:16:08 acc1_pos: 2.865 acc5_pos: 100.000 loss: 1.2430 (1.2947) acc1: 7.5521 (10.0694) acc5: 100.0000 (100.0000) time: 0.0909 data: 0.0005 max mem: 3531 +Test: [ 30/2387] eta: 0:13:11 acc1_pos: 8.854 acc5_pos: 100.000 loss: 1.2736 (1.2935) acc1: 8.3333 (10.6267) acc5: 100.0000 (100.0000) time: 0.1368 data: 0.0453 max mem: 3531 +Test: [ 40/2387] eta: 0:11:58 acc1_pos: 10.417 acc5_pos: 100.000 loss: 1.2850 (1.2918) acc1: 7.2917 (9.6989) acc5: 100.0000 (100.0000) time: 0.1975 data: 0.1023 max mem: 3531 +Test: [ 50/2387] eta: 0:11:18 acc1_pos: 32.031 acc5_pos: 100.000 loss: 1.1670 (1.2751) acc1: 9.1146 (11.2183) acc5: 100.0000 (100.0000) time: 0.2198 data: 0.1233 max mem: 3531 +Test: [ 60/2387] eta: 0:11:03 acc1_pos: 35.677 acc5_pos: 100.000 loss: 1.1474 (1.2620) acc1: 17.1875 (12.3506) acc5: 100.0000 (100.0000) time: 0.2423 data: 0.1481 max mem: 3531 +Test: [ 70/2387] eta: 0:10:52 acc1_pos: 19.531 acc5_pos: 100.000 loss: 1.2176 (1.2644) acc1: 16.1458 (12.2249) acc5: 100.0000 (100.0000) time: 0.2590 data: 0.1635 max mem: 3531 +Test: [ 80/2387] eta: 0:10:29 acc1_pos: 32.812 acc5_pos: 100.000 loss: 1.1803 (1.2409) acc1: 16.1458 (13.3102) acc5: 100.0000 (100.0000) time: 0.2349 data: 0.1380 max mem: 3531 +Test: [ 90/2387] eta: 0:10:28 acc1_pos: 33.073 acc5_pos: 100.000 loss: 1.0164 (1.2059) acc1: 26.3021 (15.5649) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1493 max mem: 3531 +Test: [ 100/2387] eta: 0:10:22 acc1_pos: 20.833 acc5_pos: 100.000 loss: 0.9429 (1.1889) acc1: 25.0000 (16.3083) acc5: 100.0000 (100.0000) time: 0.2699 data: 0.1747 max mem: 3531 +Test: [ 110/2387] eta: 0:10:08 acc1_pos: 28.125 acc5_pos: 100.000 loss: 0.9429 (1.1668) acc1: 25.0000 (17.6004) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1431 max mem: 3531 +Test: [ 120/2387] eta: 0:10:01 acc1_pos: 26.823 acc5_pos: 100.000 loss: 1.0006 (1.1565) acc1: 28.1250 (18.5154) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1318 max mem: 3531 +Test: [ 130/2387] eta: 0:09:50 acc1_pos: 19.792 acc5_pos: 100.000 loss: 1.0433 (1.1463) acc1: 26.8229 (18.9786) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1330 max mem: 3531 +Test: [ 140/2387] eta: 0:09:44 acc1_pos: 24.219 acc5_pos: 100.000 loss: 1.0144 (1.1351) acc1: 27.0833 (19.7159) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1352 max mem: 3531 +Test: [ 150/2387] eta: 0:09:38 acc1_pos: 11.458 acc5_pos: 100.000 loss: 1.0078 (1.1244) acc1: 29.1667 (20.2590) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1440 max mem: 3531 +Test: [ 160/2387] eta: 0:09:46 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.9439 (1.0996) acc1: 30.2083 (22.3295) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2849 data: 0.1900 max mem: 3531 +Test: [ 170/2387] eta: 0:09:23 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.2334 (1.0483) acc1: 97.9167 (26.7788) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2199 data: 0.1249 max mem: 3531 +Test: [ 180/2387] eta: 0:09:13 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.2193 (1.0026) acc1: 98.1771 (30.7364) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1500 data: 0.0546 max mem: 3531 +Test: [ 190/2387] eta: 0:09:06 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.2184 (0.9616) acc1: 98.1771 (34.2850) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1999 data: 0.1060 max mem: 3531 +Test: [ 200/2387] eta: 0:08:58 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.2189 (0.9247) acc1: 98.4375 (37.4883) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2048 data: 0.1090 max mem: 3531 +Test: [ 210/2387] eta: 0:08:49 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.2231 (0.8916) acc1: 98.4375 (40.3671) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1899 data: 0.0928 max mem: 3532 +Test: [ 220/2387] eta: 0:08:36 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.2236 (0.8612) acc1: 98.6979 (43.0100) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1616 data: 0.0658 max mem: 3532 +Test: [ 230/2387] eta: 0:08:28 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.2216 (0.8336) acc1: 98.9583 (45.4207) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1599 data: 0.0648 max mem: 3532 +Test: [ 240/2387] eta: 0:08:19 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.2282 (0.8102) acc1: 98.1771 (47.5579) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1708 data: 0.0735 max mem: 3532 +Test: [ 250/2387] eta: 0:08:14 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.2121 (0.7830) acc1: 98.1771 (49.6120) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0848 max mem: 3532 +Test: [ 260/2387] eta: 0:08:08 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1019 (0.7568) acc1: 99.7396 (51.5366) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1883 data: 0.0977 max mem: 3532 +Test: [ 270/2387] eta: 0:08:00 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.0944 (0.7324) acc1: 99.7396 (53.3172) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1704 data: 0.0802 max mem: 3532 +Test: [ 280/2387] eta: 0:07:56 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.0932 (0.7103) acc1: 99.7396 (54.9618) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1839 data: 0.0915 max mem: 3532 +Test: [ 290/2387] eta: 0:07:49 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.0897 (0.6891) acc1: 100.0000 (56.5006) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1806 data: 0.0863 max mem: 3532 +Test: [ 300/2387] eta: 0:07:43 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.0876 (0.6692) acc1: 100.0000 (57.9388) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1673 data: 0.0742 max mem: 3532 +Test: [ 310/2387] eta: 0:07:38 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.0934 (0.6566) acc1: 100.0000 (59.1070) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.1808 data: 0.0892 max mem: 3532 +Test: [ 320/2387] eta: 0:07:34 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.5090 (0.6521) acc1: 86.7188 (59.9445) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1832 data: 0.0911 max mem: 3532 +Test: [ 330/2387] eta: 0:07:30 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.5132 (0.6478) acc1: 85.9375 (60.7691) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.1927 data: 0.0968 max mem: 3532 +Test: [ 340/2387] eta: 0:07:27 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.5071 (0.6437) acc1: 86.7188 (61.5263) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1960 data: 0.1014 max mem: 3532 +Test: [ 350/2387] eta: 0:07:22 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.4997 (0.6330) acc1: 87.7604 (62.5705) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1806 data: 0.0895 max mem: 3532 +Test: [ 360/2387] eta: 0:07:16 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.2188 (0.6208) acc1: 99.4792 (63.5842) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1608 data: 0.0696 max mem: 3532 +Test: [ 370/2387] eta: 0:07:11 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1924 (0.6093) acc1: 99.2188 (64.5440) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1643 data: 0.0723 max mem: 3532 +Test: [ 380/2387] eta: 0:07:07 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1920 (0.5983) acc1: 99.2188 (65.4596) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1738 data: 0.0806 max mem: 3532 +Test: [ 390/2387] eta: 0:07:03 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1923 (0.5879) acc1: 99.2188 (66.3243) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1698 data: 0.0728 max mem: 3532 +Test: [ 400/2387] eta: 0:06:57 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1928 (0.5781) acc1: 99.2188 (67.1472) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1534 data: 0.0556 max mem: 3532 +Test: [ 410/2387] eta: 0:06:53 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1913 (0.5687) acc1: 99.2188 (67.9282) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1549 data: 0.0609 max mem: 3532 +Test: [ 420/2387] eta: 0:06:49 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1900 (0.5597) acc1: 99.2188 (68.6758) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1703 data: 0.0783 max mem: 3532 +Test: [ 430/2387] eta: 0:06:46 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1945 (0.5518) acc1: 99.2188 (69.3645) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1799 data: 0.0882 max mem: 3532 +Test: [ 440/2387] eta: 0:06:43 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1969 (0.5426) acc1: 98.6979 (70.0356) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1844 data: 0.0910 max mem: 3533 +Test: [ 450/2387] eta: 0:06:40 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.0866 (0.5325) acc1: 100.0000 (70.7000) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1852 data: 0.0895 max mem: 3533 +Test: [ 460/2387] eta: 0:06:38 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.0866 (0.5229) acc1: 100.0000 (71.3355) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1999 data: 0.1043 max mem: 3533 +Test: [ 470/2387] eta: 0:06:37 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.0869 (0.5136) acc1: 100.0000 (71.9441) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2149 data: 0.1218 max mem: 3533 +Test: [ 480/2387] eta: 0:06:33 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.0872 (0.5048) acc1: 100.0000 (72.5274) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1999 data: 0.1074 max mem: 3533 +Test: [ 490/2387] eta: 0:06:30 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.0875 (0.4963) acc1: 100.0000 (73.0869) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1796 data: 0.0872 max mem: 3533 +Test: [ 500/2387] eta: 0:06:27 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.0878 (0.4881) acc1: 100.0000 (73.6241) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1815 data: 0.0899 max mem: 3533 +Test: [ 510/2387] eta: 0:06:23 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.0877 (0.4803) acc1: 100.0000 (74.1403) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1668 data: 0.0746 max mem: 3533 +Test: [ 520/2387] eta: 0:06:21 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.0925 (0.4731) acc1: 100.0000 (74.6366) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1691 data: 0.0758 max mem: 3533 +Test: [ 530/2387] eta: 0:06:17 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1231 (0.4665) acc1: 100.0000 (75.1143) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1780 data: 0.0838 max mem: 3533 +Test: [ 540/2387] eta: 0:06:14 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1243 (0.4602) acc1: 100.0000 (75.5743) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1608 data: 0.0665 max mem: 3533 +Test: [ 550/2387] eta: 0:06:11 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1246 (0.4541) acc1: 100.0000 (76.0176) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1677 data: 0.0738 max mem: 3533 +Test: [ 560/2387] eta: 0:06:08 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1251 (0.4482) acc1: 100.0000 (76.4446) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1742 data: 0.0803 max mem: 3533 +Test: [ 570/2387] eta: 0:06:05 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1251 (0.4425) acc1: 100.0000 (76.8571) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1745 data: 0.0787 max mem: 3533 +Test: [ 580/2387] eta: 0:06:02 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1236 (0.4370) acc1: 100.0000 (77.2555) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1789 data: 0.0844 max mem: 3533 +Test: [ 590/2387] eta: 0:05:59 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1236 (0.4317) acc1: 100.0000 (77.6403) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1729 data: 0.0816 max mem: 3533 +Test: [ 600/2387] eta: 0:05:56 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1203 (0.4262) acc1: 100.0000 (78.0123) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1710 data: 0.0799 max mem: 3533 +Test: [ 610/2387] eta: 0:05:55 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1212 (0.4221) acc1: 100.0000 (78.3688) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1896 data: 0.0978 max mem: 3533 +Test: [ 620/2387] eta: 0:05:52 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1762 (0.4181) acc1: 99.7396 (78.7125) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1903 data: 0.0975 max mem: 3533 +Test: [ 630/2387] eta: 0:05:50 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1765 (0.4143) acc1: 99.7396 (79.0478) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0963 max mem: 3533 +Test: [ 640/2387] eta: 0:05:49 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1792 (0.4106) acc1: 99.7396 (79.3706) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2135 data: 0.1193 max mem: 3533 +Test: [ 650/2387] eta: 0:05:47 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1805 (0.4071) acc1: 99.7396 (79.6827) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2087 data: 0.1152 max mem: 3533 +Test: [ 660/2387] eta: 0:05:45 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1785 (0.4036) acc1: 99.7396 (79.9881) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2007 data: 0.1088 max mem: 3533 +Test: [ 670/2387] eta: 0:05:43 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1785 (0.4003) acc1: 99.7396 (80.2813) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2023 data: 0.1115 max mem: 3534 +Test: [ 680/2387] eta: 0:05:41 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1821 (0.3980) acc1: 99.4792 (80.5613) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2018 data: 0.1107 max mem: 3534 +Test: [ 690/2387] eta: 0:05:39 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.2760 (0.3980) acc1: 98.6979 (80.6191) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.2086 data: 0.1137 max mem: 3534 +Test: [ 700/2387] eta: 0:05:38 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.4379 (0.3986) acc1: 79.6875 (80.6144) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.2188 data: 0.1239 max mem: 3534 +Test: [ 710/2387] eta: 0:05:36 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.4360 (0.3991) acc1: 79.9479 (80.6101) acc5: 100.0000 (100.0000) acc1_neg: 80.729 acc5_neg: 100.000 time: 0.2232 data: 0.1314 max mem: 3534 +Test: [ 720/2387] eta: 0:05:35 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.4357 (0.3996) acc1: 80.7292 (80.6031) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.2332 data: 0.1412 max mem: 3534 +Test: [ 730/2387] eta: 0:05:33 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.4418 (0.4002) acc1: 80.4688 (80.5956) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.2116 data: 0.1141 max mem: 3534 +Test: [ 740/2387] eta: 0:05:31 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.4397 (0.4007) acc1: 79.4271 (80.5847) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.1964 data: 0.0996 max mem: 3534 +Test: [ 750/2387] eta: 0:05:29 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.4323 (0.4011) acc1: 79.9479 (80.5853) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.2049 data: 0.1138 max mem: 3534 +Test: [ 760/2387] eta: 0:05:27 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.4270 (0.4014) acc1: 80.4688 (80.5957) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.1933 data: 0.1023 max mem: 3534 +Test: [ 770/2387] eta: 0:05:24 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.3828 (0.4001) acc1: 83.5938 (80.7285) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1903 data: 0.0986 max mem: 3534 +Test: [ 780/2387] eta: 0:05:23 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.2408 (0.3977) acc1: 96.6146 (80.9522) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1999 data: 0.1039 max mem: 3534 +Test: [ 790/2387] eta: 0:05:21 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1983 (0.3952) acc1: 98.6979 (81.1450) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.2045 data: 0.1073 max mem: 3534 +Test: [ 800/2387] eta: 0:05:19 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1801 (0.3925) acc1: 99.4792 (81.3729) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2007 data: 0.1073 max mem: 3534 +Test: [ 810/2387] eta: 0:05:17 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1767 (0.3902) acc1: 99.4792 (81.5875) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2157 data: 0.1241 max mem: 3534 +Test: [ 820/2387] eta: 0:05:15 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1925 (0.3881) acc1: 98.6979 (81.7737) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.2165 data: 0.1242 max mem: 3534 +Test: [ 830/2387] eta: 0:05:13 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.2765 (0.3878) acc1: 93.7500 (81.8608) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1868 data: 0.0921 max mem: 3534 +Test: [ 840/2387] eta: 0:05:10 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.3338 (0.3873) acc1: 90.1042 (81.9520) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1846 data: 0.0893 max mem: 3534 +Test: [ 850/2387] eta: 0:05:08 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.3219 (0.3866) acc1: 90.8854 (82.0466) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1935 data: 0.0989 max mem: 3534 +Test: [ 860/2387] eta: 0:05:06 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.3151 (0.3856) acc1: 92.4479 (82.1831) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1877 data: 0.0929 max mem: 3534 +Test: [ 870/2387] eta: 0:05:04 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.3151 (0.3848) acc1: 92.7083 (82.2959) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1856 data: 0.0901 max mem: 3534 +Test: [ 880/2387] eta: 0:05:02 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.3258 (0.3843) acc1: 91.4062 (82.3886) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1921 data: 0.0965 max mem: 3534 +Test: [ 890/2387] eta: 0:05:00 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.2863 (0.3830) acc1: 94.5312 (82.5427) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2003 data: 0.1071 max mem: 3534 +Test: [ 900/2387] eta: 0:04:58 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.3087 (0.3850) acc1: 93.4896 (82.3989) acc5: 100.0000 (100.0000) acc1_neg: 65.104 acc5_neg: 100.000 time: 0.1986 data: 0.1072 max mem: 3535 +Test: [ 910/2387] eta: 0:04:55 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.6193 (0.3876) acc1: 63.5417 (82.1868) acc5: 100.0000 (100.0000) acc1_neg: 60.677 acc5_neg: 100.000 time: 0.1887 data: 0.0977 max mem: 3535 +Test: [ 920/2387] eta: 0:04:53 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.6273 (0.3902) acc1: 62.5000 (81.9770) acc5: 100.0000 (100.0000) acc1_neg: 59.896 acc5_neg: 100.000 time: 0.1876 data: 0.0963 max mem: 3535 +Test: [ 930/2387] eta: 0:04:51 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.6247 (0.3927) acc1: 63.0208 (81.7798) acc5: 100.0000 (100.0000) acc1_neg: 61.458 acc5_neg: 100.000 time: 0.1946 data: 0.1026 max mem: 3535 +Test: [ 940/2387] eta: 0:04:49 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.6286 (0.3952) acc1: 61.9792 (81.5741) acc5: 100.0000 (100.0000) acc1_neg: 60.677 acc5_neg: 100.000 time: 0.1977 data: 0.1064 max mem: 3535 +Test: [ 950/2387] eta: 0:04:47 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.6332 (0.3976) acc1: 61.4583 (81.3771) acc5: 100.0000 (100.0000) acc1_neg: 64.844 acc5_neg: 100.000 time: 0.1978 data: 0.1067 max mem: 3535 +Test: [ 960/2387] eta: 0:04:45 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.6199 (0.3999) acc1: 63.5417 (81.1907) acc5: 100.0000 (100.0000) acc1_neg: 65.625 acc5_neg: 100.000 time: 0.1939 data: 0.1028 max mem: 3535 +Test: [ 970/2387] eta: 0:04:43 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.6239 (0.4023) acc1: 62.7604 (80.9888) acc5: 100.0000 (100.0000) acc1_neg: 63.021 acc5_neg: 100.000 time: 0.1854 data: 0.0944 max mem: 3535 +Test: [ 980/2387] eta: 0:04:40 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.6125 (0.4020) acc1: 64.0625 (81.0299) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.1793 data: 0.0878 max mem: 3535 +Test: [ 990/2387] eta: 0:04:38 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.2385 (0.4003) acc1: 97.1354 (81.1843) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1750 data: 0.0828 max mem: 3535 +Test: [1000/2387] eta: 0:04:35 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.2130 (0.3985) acc1: 98.9583 (81.3491) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1666 data: 0.0743 max mem: 3535 +Test: [1010/2387] eta: 0:04:33 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.2237 (0.3974) acc1: 98.4375 (81.4677) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1747 data: 0.0833 max mem: 3535 +Test: [1020/2387] eta: 0:04:31 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.2736 (0.3962) acc1: 96.6146 (81.5869) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1701 data: 0.0791 max mem: 3535 +Test: [1030/2387] eta: 0:04:28 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.2031 (0.3941) acc1: 97.6562 (81.7423) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1658 data: 0.0746 max mem: 3535 +Test: [1040/2387] eta: 0:04:26 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1606 (0.3917) acc1: 99.7396 (81.9152) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1776 data: 0.0866 max mem: 3535 +Test: [1050/2387] eta: 0:04:25 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1365 (0.3893) acc1: 100.0000 (82.0845) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2039 data: 0.1128 max mem: 3535 +Test: [1060/2387] eta: 0:04:22 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1393 (0.3872) acc1: 99.7396 (82.2472) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1857 data: 0.0943 max mem: 3535 +Test: [1070/2387] eta: 0:04:20 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1631 (0.3854) acc1: 99.4792 (82.3897) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1742 data: 0.0833 max mem: 3535 +Test: [1080/2387] eta: 0:04:18 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.2183 (0.3844) acc1: 96.0938 (82.4964) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1933 data: 0.1021 max mem: 3535 +Test: [1090/2387] eta: 0:04:15 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.2103 (0.3826) acc1: 98.4375 (82.6488) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1523 data: 0.0593 max mem: 3535 +Test: [1100/2387] eta: 0:04:13 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1639 (0.3807) acc1: 99.4792 (82.7993) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1599 data: 0.0629 max mem: 3535 +Test: [1110/2387] eta: 0:04:11 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1613 (0.3791) acc1: 99.4792 (82.9016) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1975 data: 0.1005 max mem: 3535 +Test: [1120/2387] eta: 0:04:09 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1840 (0.3775) acc1: 99.4792 (83.0416) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2149 data: 0.1185 max mem: 3536 +Test: [1130/2387] eta: 0:04:07 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1733 (0.3757) acc1: 99.4792 (83.1841) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2050 data: 0.1080 max mem: 3536 +Test: [1140/2387] eta: 0:04:04 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.2143 (0.3748) acc1: 98.6979 (83.2877) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1499 data: 0.0512 max mem: 3536 +Test: [1150/2387] eta: 0:04:03 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.2651 (0.3739) acc1: 95.0521 (83.3933) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1661 data: 0.0683 max mem: 3536 +Test: [1160/2387] eta: 0:04:00 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.2176 (0.3724) acc1: 98.9583 (83.5240) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1859 data: 0.0930 max mem: 3536 +Test: [1170/2387] eta: 0:03:59 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1747 (0.3708) acc1: 99.7396 (83.6587) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1946 data: 0.1028 max mem: 3536 +Test: [1180/2387] eta: 0:03:57 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1863 (0.3694) acc1: 98.6979 (83.7779) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2328 data: 0.1401 max mem: 3536 +Test: [1190/2387] eta: 0:03:54 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1888 (0.3678) acc1: 97.6562 (83.8996) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1742 data: 0.0819 max mem: 3536 +Test: [1200/2387] eta: 0:03:51 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1881 (0.3663) acc1: 98.9583 (84.0285) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1071 data: 0.0131 max mem: 3536 +Test: [1210/2387] eta: 0:03:50 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1940 (0.3648) acc1: 99.7396 (84.1546) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1499 data: 0.0554 max mem: 3536 +Test: [1220/2387] eta: 0:03:47 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1412 (0.3628) acc1: 99.7396 (84.2773) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1860 data: 0.0929 max mem: 3536 +Test: [1230/2387] eta: 0:03:45 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1058 (0.3607) acc1: 99.4792 (84.4004) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1805 data: 0.0877 max mem: 3536 +Test: [1240/2387] eta: 0:03:43 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1112 (0.3587) acc1: 99.7396 (84.5208) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1836 data: 0.0916 max mem: 3536 +Test: [1250/2387] eta: 0:03:41 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1105 (0.3568) acc1: 99.7396 (84.6392) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1813 data: 0.0892 max mem: 3536 +Test: [1260/2387] eta: 0:03:39 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1010 (0.3548) acc1: 99.4792 (84.7571) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1906 data: 0.0993 max mem: 3536 +Test: [1270/2387] eta: 0:03:37 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.0975 (0.3527) acc1: 100.0000 (84.8764) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2013 data: 0.1097 max mem: 3536 +Test: [1280/2387] eta: 0:03:35 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.0975 (0.3508) acc1: 100.0000 (84.9924) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2036 data: 0.1104 max mem: 3536 +Test: [1290/2387] eta: 0:03:34 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1189 (0.3491) acc1: 99.7396 (85.0911) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2060 data: 0.1130 max mem: 3536 +Test: [1300/2387] eta: 0:03:32 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1115 (0.3473) acc1: 99.7396 (85.1981) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2163 data: 0.1251 max mem: 3536 +Test: [1310/2387] eta: 0:03:30 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1115 (0.3457) acc1: 99.7396 (85.3021) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2178 data: 0.1266 max mem: 3536 +Test: [1320/2387] eta: 0:03:28 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1307 (0.3441) acc1: 99.4792 (85.3999) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1950 data: 0.1035 max mem: 3536 +Test: [1330/2387] eta: 0:03:26 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1392 (0.3429) acc1: 98.4375 (85.4965) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1819 data: 0.0900 max mem: 3536 +Test: [1340/2387] eta: 0:03:24 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.2191 (0.3428) acc1: 96.6146 (85.4988) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1883 data: 0.0965 max mem: 3536 +Test: [1350/2387] eta: 0:03:22 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.0791 (0.3408) acc1: 100.0000 (85.6061) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1853 data: 0.0937 max mem: 3537 +Test: [1360/2387] eta: 0:03:20 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.0616 (0.3387) acc1: 100.0000 (85.7119) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1857 data: 0.0916 max mem: 3537 +Test: [1370/2387] eta: 0:03:18 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.0589 (0.3367) acc1: 100.0000 (85.8161) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1859 data: 0.0922 max mem: 3537 +Test: [1380/2387] eta: 0:03:16 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.0570 (0.3347) acc1: 100.0000 (85.9188) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2062 data: 0.1147 max mem: 3537 +Test: [1390/2387] eta: 0:03:14 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.0613 (0.3327) acc1: 100.0000 (86.0201) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1934 data: 0.1013 max mem: 3537 +Test: [1400/2387] eta: 0:03:12 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.0601 (0.3308) acc1: 100.0000 (86.1198) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1605 data: 0.0670 max mem: 3537 +Test: [1410/2387] eta: 0:03:10 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.0652 (0.3295) acc1: 100.0000 (86.2086) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1899 data: 0.0959 max mem: 3537 +Test: [1420/2387] eta: 0:03:08 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1793 (0.3287) acc1: 99.2188 (86.2822) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2017 data: 0.1073 max mem: 3537 +Test: [1430/2387] eta: 0:03:06 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1892 (0.3277) acc1: 98.9583 (86.3692) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2003 data: 0.1066 max mem: 3537 +Test: [1440/2387] eta: 0:03:04 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1892 (0.3268) acc1: 99.2188 (86.4502) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.2049 data: 0.1128 max mem: 3537 +Test: [1450/2387] eta: 0:03:02 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.2088 (0.3263) acc1: 98.1771 (86.5161) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2004 data: 0.1078 max mem: 3537 +Test: [1460/2387] eta: 0:03:00 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1915 (0.3252) acc1: 98.1771 (86.5983) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1912 data: 0.0980 max mem: 3537 +Test: [1470/2387] eta: 0:02:58 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1589 (0.3240) acc1: 99.7396 (86.6871) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1841 data: 0.0904 max mem: 3537 +Test: [1480/2387] eta: 0:02:56 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1459 (0.3228) acc1: 99.7396 (86.7731) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1772 data: 0.0822 max mem: 3537 +Test: [1490/2387] eta: 0:02:54 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1526 (0.3217) acc1: 99.7396 (86.8576) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1755 data: 0.0818 max mem: 3537 +Test: [1500/2387] eta: 0:02:52 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1526 (0.3206) acc1: 99.4792 (86.9391) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1828 data: 0.0913 max mem: 3537 +Test: [1510/2387] eta: 0:02:50 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1595 (0.3196) acc1: 99.7396 (87.0233) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1843 data: 0.0927 max mem: 3537 +Test: [1520/2387] eta: 0:02:48 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1931 (0.3195) acc1: 98.1771 (87.0475) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1784 data: 0.0859 max mem: 3537 +Test: [1530/2387] eta: 0:02:46 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.3071 (0.3194) acc1: 91.6667 (87.0800) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1806 data: 0.0877 max mem: 3537 +Test: [1540/2387] eta: 0:02:44 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.2846 (0.3191) acc1: 93.4896 (87.1331) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1904 data: 0.0983 max mem: 3537 +Test: [1550/2387] eta: 0:02:42 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.2821 (0.3190) acc1: 94.5312 (87.1775) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1918 data: 0.1002 max mem: 3537 +Test: [1560/2387] eta: 0:02:40 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.2954 (0.3189) acc1: 93.2292 (87.2076) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.1771 data: 0.0850 max mem: 3537 +Test: [1570/2387] eta: 0:02:38 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.2822 (0.3184) acc1: 96.0938 (87.2727) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1827 data: 0.0899 max mem: 3537 +Test: [1580/2387] eta: 0:02:36 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.2192 (0.3178) acc1: 98.1771 (87.3427) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1990 data: 0.1069 max mem: 3538 +Test: [1590/2387] eta: 0:02:35 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1992 (0.3169) acc1: 99.2188 (87.4123) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2156 data: 0.1245 max mem: 3538 +Test: [1600/2387] eta: 0:02:32 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1694 (0.3163) acc1: 99.4792 (87.4561) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1906 data: 0.1001 max mem: 3538 +Test: [1610/2387] eta: 0:02:30 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1694 (0.3154) acc1: 99.2188 (87.5278) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1609 data: 0.0703 max mem: 3538 +Test: [1620/2387] eta: 0:02:28 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1764 (0.3146) acc1: 99.2188 (87.5948) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1933 data: 0.1016 max mem: 3538 +Test: [1630/2387] eta: 0:02:26 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.2078 (0.3143) acc1: 97.3958 (87.6450) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1998 data: 0.1073 max mem: 3538 +Test: [1640/2387] eta: 0:02:24 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.3193 (0.3147) acc1: 91.9271 (87.6408) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.1889 data: 0.0975 max mem: 3538 +Test: [1650/2387] eta: 0:02:22 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.3338 (0.3147) acc1: 90.8854 (87.6733) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1801 data: 0.0883 max mem: 3538 +Test: [1660/2387] eta: 0:02:21 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.3290 (0.3148) acc1: 90.6250 (87.6869) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.2005 data: 0.1070 max mem: 3538 +Test: [1670/2387] eta: 0:02:19 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.3368 (0.3150) acc1: 90.3646 (87.7073) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1997 data: 0.1068 max mem: 3538 +Test: [1680/2387] eta: 0:02:17 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.3399 (0.3152) acc1: 90.8854 (87.7167) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.1842 data: 0.0916 max mem: 3538 +Test: [1690/2387] eta: 0:02:15 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.3568 (0.3154) acc1: 90.6250 (87.7244) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.2005 data: 0.1038 max mem: 3538 +Test: [1700/2387] eta: 0:02:13 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.3515 (0.3152) acc1: 91.6667 (87.7642) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1856 data: 0.0897 max mem: 3538 +Test: [1710/2387] eta: 0:02:11 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1754 (0.3143) acc1: 97.3958 (87.8271) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1727 data: 0.0815 max mem: 3538 +Test: [1720/2387] eta: 0:02:09 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1400 (0.3132) acc1: 99.4792 (87.8945) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1797 data: 0.0887 max mem: 3538 +Test: [1730/2387] eta: 0:02:07 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1330 (0.3123) acc1: 99.4792 (87.9572) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1837 data: 0.0923 max mem: 3538 +Test: [1740/2387] eta: 0:02:05 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1457 (0.3113) acc1: 99.2188 (88.0196) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1923 data: 0.1005 max mem: 3538 +Test: [1750/2387] eta: 0:02:03 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1206 (0.3103) acc1: 99.7396 (88.0851) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1874 data: 0.0960 max mem: 3538 +Test: [1760/2387] eta: 0:02:01 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1289 (0.3094) acc1: 99.7396 (88.1511) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1872 data: 0.0964 max mem: 3538 +Test: [1770/2387] eta: 0:01:59 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1706 (0.3088) acc1: 98.6979 (88.1911) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1954 data: 0.1014 max mem: 3538 +Test: [1780/2387] eta: 0:01:57 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1995 (0.3082) acc1: 96.8750 (88.2446) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1986 data: 0.1045 max mem: 3538 +Test: [1790/2387] eta: 0:01:55 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1895 (0.3074) acc1: 98.4375 (88.3029) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2082 data: 0.1167 max mem: 3538 +Test: [1800/2387] eta: 0:01:53 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1681 (0.3067) acc1: 98.4375 (88.3543) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2101 data: 0.1174 max mem: 3538 +Test: [1810/2387] eta: 0:01:52 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1679 (0.3059) acc1: 97.9167 (88.4104) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2252 data: 0.1312 max mem: 3539 +Test: [1820/2387] eta: 0:01:50 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1540 (0.3051) acc1: 98.9583 (88.4687) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2417 data: 0.1485 max mem: 3539 +Test: [1830/2387] eta: 0:01:48 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1522 (0.3042) acc1: 99.7396 (88.5292) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2091 data: 0.1173 max mem: 3539 +Test: [1840/2387] eta: 0:01:46 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.0899 (0.3030) acc1: 100.0000 (88.5913) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2117 data: 0.1175 max mem: 3539 +Test: [1850/2387] eta: 0:01:44 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.0893 (0.3018) acc1: 100.0000 (88.6530) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2186 data: 0.1229 max mem: 3539 +Test: [1860/2387] eta: 0:01:42 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.0895 (0.3007) acc1: 100.0000 (88.7139) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2089 data: 0.1160 max mem: 3539 +Test: [1870/2387] eta: 0:01:40 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.0898 (0.2996) acc1: 100.0000 (88.7742) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2253 data: 0.1326 max mem: 3539 +Test: [1880/2387] eta: 0:01:38 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.0893 (0.2985) acc1: 100.0000 (88.8339) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2193 data: 0.1269 max mem: 3539 +Test: [1890/2387] eta: 0:01:36 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.0887 (0.2973) acc1: 100.0000 (88.8930) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2156 data: 0.1243 max mem: 3539 +Test: [1900/2387] eta: 0:01:35 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.0904 (0.2963) acc1: 100.0000 (88.9514) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2030 data: 0.1116 max mem: 3539 +Test: [1910/2387] eta: 0:01:33 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1063 (0.2953) acc1: 100.0000 (89.0089) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1853 data: 0.0944 max mem: 3539 +Test: [1920/2387] eta: 0:01:31 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1268 (0.2948) acc1: 100.0000 (89.0626) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1847 data: 0.0911 max mem: 3539 +Test: [1930/2387] eta: 0:01:29 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1511 (0.2940) acc1: 100.0000 (89.1193) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1818 data: 0.0877 max mem: 3539 +Test: [1940/2387] eta: 0:01:27 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1511 (0.2936) acc1: 100.0000 (89.1400) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1898 data: 0.0973 max mem: 3539 +Test: [1950/2387] eta: 0:01:25 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1593 (0.2930) acc1: 100.0000 (89.1946) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1789 data: 0.0862 max mem: 3539 +Test: [1960/2387] eta: 0:01:23 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1607 (0.2924) acc1: 100.0000 (89.2420) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1858 data: 0.0938 max mem: 3539 +Test: [1970/2387] eta: 0:01:21 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.2229 (0.2924) acc1: 99.2188 (89.2681) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.1925 data: 0.0995 max mem: 3539 +Test: [1980/2387] eta: 0:01:19 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.2567 (0.2920) acc1: 98.1771 (89.3028) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1603 data: 0.0636 max mem: 3539 +Test: [1990/2387] eta: 0:01:17 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1481 (0.2913) acc1: 98.9583 (89.3514) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1706 data: 0.0736 max mem: 3539 +Test: [2000/2387] eta: 0:01:15 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1473 (0.2906) acc1: 98.9583 (89.3984) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1806 data: 0.0868 max mem: 3539 +Test: [2010/2387] eta: 0:01:13 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1510 (0.2899) acc1: 98.6979 (89.4444) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1642 data: 0.0687 max mem: 3539 +Test: [2020/2387] eta: 0:01:11 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1520 (0.2892) acc1: 98.6979 (89.4916) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1686 data: 0.0741 max mem: 3539 +Test: [2030/2387] eta: 0:01:09 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1495 (0.2885) acc1: 98.9583 (89.5381) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1861 data: 0.0954 max mem: 3540 +Test: [2040/2387] eta: 0:01:07 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1492 (0.2878) acc1: 98.9583 (89.5852) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1942 data: 0.1024 max mem: 3540 +Test: [2050/2387] eta: 0:01:05 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1451 (0.2871) acc1: 99.2188 (89.6327) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1845 data: 0.0915 max mem: 3540 +Test: [2060/2387] eta: 0:01:03 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1363 (0.2864) acc1: 99.2188 (89.6784) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1855 data: 0.0929 max mem: 3540 +Test: [2070/2387] eta: 0:01:01 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1322 (0.2856) acc1: 99.2188 (89.7258) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1802 data: 0.0890 max mem: 3540 +Test: [2080/2387] eta: 0:00:59 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1286 (0.2848) acc1: 99.2188 (89.7710) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1722 data: 0.0811 max mem: 3540 +Test: [2090/2387] eta: 0:00:57 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1298 (0.2841) acc1: 99.2188 (89.8175) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1838 data: 0.0918 max mem: 3540 +Test: [2100/2387] eta: 0:00:55 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1295 (0.2834) acc1: 99.4792 (89.8622) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1778 data: 0.0862 max mem: 3540 +Test: [2110/2387] eta: 0:00:53 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1295 (0.2826) acc1: 99.2188 (89.9064) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1849 data: 0.0926 max mem: 3540 +Test: [2120/2387] eta: 0:00:51 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1354 (0.2822) acc1: 98.9583 (89.9490) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2091 data: 0.1145 max mem: 3540 +Test: [2130/2387] eta: 0:00:49 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1418 (0.2815) acc1: 99.7396 (89.9903) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1949 data: 0.0984 max mem: 3540 +Test: [2140/2387] eta: 0:00:47 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1486 (0.2811) acc1: 99.7396 (90.0176) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1964 data: 0.1017 max mem: 3540 +Test: [2150/2387] eta: 0:00:45 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1573 (0.2805) acc1: 99.4792 (90.0613) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2099 data: 0.1180 max mem: 3540 +Test: [2160/2387] eta: 0:00:43 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1462 (0.2800) acc1: 99.7396 (90.1021) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1841 data: 0.0915 max mem: 3540 +Test: [2170/2387] eta: 0:00:42 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1766 (0.2798) acc1: 99.2188 (90.1278) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1805 data: 0.0887 max mem: 3540 +Test: [2180/2387] eta: 0:00:40 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.2912 (0.2809) acc1: 91.9271 (90.0459) acc5: 100.0000 (100.0000) acc1_neg: 41.667 acc5_neg: 100.000 time: 0.1896 data: 0.0969 max mem: 3540 +Test: [2190/2387] eta: 0:00:38 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.3198 (0.2812) acc1: 91.6667 (90.0345) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1961 data: 0.1024 max mem: 3540 +Test: [2200/2387] eta: 0:00:36 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1879 (0.2807) acc1: 97.6562 (90.0758) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1996 data: 0.1066 max mem: 3540 +Test: [2210/2387] eta: 0:00:34 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1571 (0.2801) acc1: 99.7396 (90.1157) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2009 data: 0.1078 max mem: 3540 +Test: [2220/2387] eta: 0:00:32 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1448 (0.2796) acc1: 99.7396 (90.1574) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1998 data: 0.1076 max mem: 3540 +Test: [2230/2387] eta: 0:00:30 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1396 (0.2789) acc1: 99.7396 (90.1971) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1987 data: 0.1068 max mem: 3540 +Test: [2240/2387] eta: 0:00:28 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1392 (0.2784) acc1: 99.7396 (90.2329) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1970 data: 0.1031 max mem: 3540 +Test: [2250/2387] eta: 0:00:26 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1688 (0.2781) acc1: 99.2188 (90.2687) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1906 data: 0.0969 max mem: 3540 +Test: [2260/2387] eta: 0:00:24 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1874 (0.2776) acc1: 99.4792 (90.3080) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2027 data: 0.1108 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1245 (0.2768) acc1: 100.0000 (90.3506) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2038 data: 0.1115 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1017 (0.2763) acc1: 100.0000 (90.3761) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1954 data: 0.1035 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1006 (0.2755) acc1: 100.0000 (90.4180) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1967 data: 0.1054 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1049 (0.2748) acc1: 100.0000 (90.4566) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1651 data: 0.0743 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1271 (0.2743) acc1: 100.0000 (90.4910) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1481 data: 0.0578 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1766 (0.2741) acc1: 99.4792 (90.5193) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1472 data: 0.0570 max mem: 3541 +Test: [2330/2387] eta: 0:00:11 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.2174 (0.2739) acc1: 98.9583 (90.5483) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.1311 data: 0.0402 max mem: 3541 +Test: [2340/2387] eta: 0:00:09 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.2174 (0.2736) acc1: 99.4792 (90.5809) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1280 data: 0.0374 max mem: 3541 +Test: [2350/2387] eta: 0:00:07 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1717 (0.2735) acc1: 99.7396 (90.5973) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1262 data: 0.0356 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1985 (0.2734) acc1: 99.4792 (90.6264) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1213 data: 0.0306 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.1994 (0.2732) acc1: 98.6979 (90.6509) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1174 data: 0.0272 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.2702 (0.2735) acc1: 95.3125 (90.6534) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.1181 data: 0.0280 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 23.288 acc5_pos: 100.000 loss: 0.3362 (0.2736) acc1: 91.6667 (90.6586) acc5: 100.0000 (100.0000) acc1_neg: 90.698 acc5_neg: 100.000 time: 0.1129 data: 0.0256 max mem: 3541 +Test: Total time: 0:07:37 (0.1915 s / it) +* Overall Acc@1 90.659 Acc@5 100.000 + target == real Acc@1 20.382 Acc@5 100.000 + target == fake Acc@1 95.594 Acc@5 100.000 + loss 0.274 + PR-AUC 0.9861 ROC-AUC 0.8322 EER 24.41% (thr=0.7666) +Accuracy of the network on the 916267 test images: 90.7% +[rank0]:[W726 05:44:40.748085232 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:45:00.269000 1198832 torch/distributed/run.py:766] +W0726 05:45:00.269000 1198832 torch/distributed/run.py:766] ***************************************** +W0726 05:45:00.269000 1198832 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:45:00.269000 1198832 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:45:29.276012916 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:45:30.575681327 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:45:30.819730037 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:45:30.852794368 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:45:30.888600865 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:45:30.075905389 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:45:30.108338887 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W726 05:45:30.108519790 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n45_seed8_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed8_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 0:58:38 acc1_pos: 10.677 acc5_pos: 100.000 loss: 1.3080 (1.3080) acc1: 10.6771 (10.6771) acc5: 100.0000 (100.0000) time: 6.8063 data: 4.8074 max mem: 3531 +Test: [ 10/517] eta: 0:05:57 acc1_pos: 2.344 acc5_pos: 100.000 loss: 1.2910 (1.2873) acc1: 7.5521 (11.5294) acc5: 100.0000 (100.0000) time: 0.7049 data: 0.4375 max mem: 3531 +Test: [ 20/517] eta: 0:03:27 acc1_pos: 2.865 acc5_pos: 100.000 loss: 1.2430 (1.2947) acc1: 7.5521 (10.0694) acc5: 100.0000 (100.0000) time: 0.0973 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:03:06 acc1_pos: 8.854 acc5_pos: 100.000 loss: 1.2736 (1.2935) acc1: 8.3333 (10.6267) acc5: 100.0000 (100.0000) time: 0.2059 data: 0.1103 max mem: 3531 +Test: [ 40/517] eta: 0:03:03 acc1_pos: 10.417 acc5_pos: 100.000 loss: 1.2850 (1.2918) acc1: 7.2917 (9.6989) acc5: 100.0000 (100.0000) time: 0.3503 data: 0.2589 max mem: 3531 +Test: [ 50/517] eta: 0:02:51 acc1_pos: 24.771 acc5_pos: 100.000 loss: 1.1670 (1.2353) acc1: 9.1146 (14.5118) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.3403 data: 0.2464 max mem: 3531 +Test: [ 60/517] eta: 0:02:38 acc1_pos: 24.771 acc5_pos: 100.000 loss: 0.3461 (1.0844) acc1: 91.1458 (27.6810) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2670 data: 0.1719 max mem: 3531 +Test: [ 70/517] eta: 0:02:27 acc1_pos: 24.771 acc5_pos: 100.000 loss: 0.2933 (0.9706) acc1: 96.8750 (37.7311) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2411 data: 0.1491 max mem: 3531 +Test: [ 80/517] eta: 0:02:19 acc1_pos: 24.771 acc5_pos: 100.000 loss: 0.2793 (0.8846) acc1: 99.4792 (45.3093) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2393 data: 0.1490 max mem: 3531 +Test: [ 90/517] eta: 0:02:13 acc1_pos: 24.771 acc5_pos: 100.000 loss: 0.2826 (0.8194) acc1: 98.6979 (51.1905) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2448 data: 0.1545 max mem: 3531 +Test: [100/517] eta: 0:02:07 acc1_pos: 24.771 acc5_pos: 100.000 loss: 0.3045 (0.7691) acc1: 98.6979 (55.8504) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2448 data: 0.1535 max mem: 3531 +Test: [110/517] eta: 0:02:01 acc1_pos: 24.771 acc5_pos: 100.000 loss: 0.2857 (0.7230) acc1: 98.4375 (59.6683) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2423 data: 0.1503 max mem: 3531 +Test: [120/517] eta: 0:01:57 acc1_pos: 24.771 acc5_pos: 100.000 loss: 0.2916 (0.6887) acc1: 98.4375 (62.8465) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2533 data: 0.1588 max mem: 3531 +Test: [130/517] eta: 0:01:53 acc1_pos: 24.771 acc5_pos: 100.000 loss: 0.3081 (0.6592) acc1: 97.6562 (65.4183) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2549 data: 0.1590 max mem: 3531 +Test: [140/517] eta: 0:01:48 acc1_pos: 24.771 acc5_pos: 100.000 loss: 0.2120 (0.6267) acc1: 99.4792 (67.8616) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2417 data: 0.1461 max mem: 3531 +Test: [150/517] eta: 0:01:44 acc1_pos: 24.771 acc5_pos: 100.000 loss: 0.2032 (0.5990) acc1: 100.0000 (69.9624) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2322 data: 0.1387 max mem: 3531 +Test: [160/517] eta: 0:01:40 acc1_pos: 24.771 acc5_pos: 100.000 loss: 0.2032 (0.5740) acc1: 100.0000 (71.8281) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2428 data: 0.1521 max mem: 3531 +Test: [170/517] eta: 0:01:37 acc1_pos: 24.771 acc5_pos: 100.000 loss: 0.1978 (0.5519) acc1: 100.0000 (73.4725) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2552 data: 0.1617 max mem: 3531 +Test: [180/517] eta: 0:01:34 acc1_pos: 24.771 acc5_pos: 100.000 loss: 0.2078 (0.5351) acc1: 99.7396 (74.7770) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2509 data: 0.1577 max mem: 3531 +Test: [190/517] eta: 0:01:30 acc1_pos: 24.771 acc5_pos: 100.000 loss: 0.2099 (0.5181) acc1: 99.7396 (76.0880) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2442 data: 0.1528 max mem: 3531 +Test: [200/517] eta: 0:01:26 acc1_pos: 24.771 acc5_pos: 100.000 loss: 0.2118 (0.5035) acc1: 99.7396 (77.2647) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2257 data: 0.1330 max mem: 3531 +Test: [210/517] eta: 0:01:23 acc1_pos: 24.771 acc5_pos: 100.000 loss: 0.2402 (0.4943) acc1: 99.2188 (78.1127) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2338 data: 0.1412 max mem: 3532 +Test: [220/517] eta: 0:01:20 acc1_pos: 24.771 acc5_pos: 100.000 loss: 0.3026 (0.4857) acc1: 94.2708 (78.8685) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2532 data: 0.1604 max mem: 3532 +Test: [230/517] eta: 0:01:17 acc1_pos: 24.771 acc5_pos: 100.000 loss: 0.3044 (0.4783) acc1: 96.3542 (79.6830) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2467 data: 0.1516 max mem: 3532 +Test: [240/517] eta: 0:01:14 acc1_pos: 24.771 acc5_pos: 100.000 loss: 0.3013 (0.4696) acc1: 98.9583 (80.5044) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2460 data: 0.1511 max mem: 3532 +Test: [250/517] eta: 0:01:11 acc1_pos: 24.771 acc5_pos: 100.000 loss: 0.2426 (0.4598) acc1: 99.7396 (81.2542) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2549 data: 0.1626 max mem: 3532 +Test: [260/517] eta: 0:01:08 acc1_pos: 24.771 acc5_pos: 100.000 loss: 0.2006 (0.4501) acc1: 99.7396 (81.9624) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2507 data: 0.1584 max mem: 3532 +Test: [270/517] eta: 0:01:06 acc1_pos: 24.771 acc5_pos: 100.000 loss: 0.2381 (0.4430) acc1: 99.7396 (82.6049) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2449 data: 0.1528 max mem: 3532 +Test: [280/517] eta: 0:01:03 acc1_pos: 24.771 acc5_pos: 100.000 loss: 0.2558 (0.4365) acc1: 99.4792 (83.1674) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2529 data: 0.1558 max mem: 3532 +Test: [290/517] eta: 0:01:00 acc1_pos: 24.771 acc5_pos: 100.000 loss: 0.2645 (0.4306) acc1: 99.4792 (83.7011) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2485 data: 0.1471 max mem: 3532 +Test: [300/517] eta: 0:00:57 acc1_pos: 24.771 acc5_pos: 100.000 loss: 0.2649 (0.4253) acc1: 99.4792 (84.2167) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2450 data: 0.1479 max mem: 3532 +Test: [310/517] eta: 0:00:54 acc1_pos: 24.771 acc5_pos: 100.000 loss: 0.2747 (0.4211) acc1: 98.9583 (84.6672) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2517 data: 0.1557 max mem: 3532 +Test: [320/517] eta: 0:00:52 acc1_pos: 24.771 acc5_pos: 100.000 loss: 0.2747 (0.4151) acc1: 98.9583 (85.1287) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2511 data: 0.1522 max mem: 3532 +Test: [330/517] eta: 0:00:49 acc1_pos: 24.771 acc5_pos: 100.000 loss: 0.1996 (0.4085) acc1: 100.0000 (85.5756) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2440 data: 0.1483 max mem: 3532 +Test: [340/517] eta: 0:00:46 acc1_pos: 24.771 acc5_pos: 100.000 loss: 0.1921 (0.4051) acc1: 100.0000 (85.7901) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.2499 data: 0.1589 max mem: 3532 +Test: [350/517] eta: 0:00:44 acc1_pos: 24.771 acc5_pos: 100.000 loss: 0.3692 (0.4052) acc1: 95.5729 (85.9620) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2525 data: 0.1595 max mem: 3532 +Test: [360/517] eta: 0:00:41 acc1_pos: 24.771 acc5_pos: 100.000 loss: 0.3670 (0.4024) acc1: 95.5729 (86.2881) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2484 data: 0.1532 max mem: 3532 +Test: [370/517] eta: 0:00:38 acc1_pos: 24.771 acc5_pos: 100.000 loss: 0.2224 (0.3970) acc1: 99.7396 (86.6556) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2649 data: 0.1708 max mem: 3532 +Test: [380/517] eta: 0:00:36 acc1_pos: 24.771 acc5_pos: 100.000 loss: 0.2177 (0.3933) acc1: 100.0000 (87.0024) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2698 data: 0.1761 max mem: 3532 +Test: [390/517] eta: 0:00:33 acc1_pos: 24.771 acc5_pos: 100.000 loss: 0.2749 (0.3912) acc1: 99.7396 (87.2283) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2548 data: 0.1589 max mem: 3532 +Test: [400/517] eta: 0:00:30 acc1_pos: 24.771 acc5_pos: 100.000 loss: 0.2647 (0.3877) acc1: 98.9583 (87.5292) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2649 data: 0.1665 max mem: 3532 +Test: [410/517] eta: 0:00:28 acc1_pos: 24.771 acc5_pos: 100.000 loss: 0.2647 (0.3863) acc1: 99.2188 (87.7826) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2712 data: 0.1736 max mem: 3532 +Test: [420/517] eta: 0:00:25 acc1_pos: 24.771 acc5_pos: 100.000 loss: 0.2626 (0.3829) acc1: 99.7396 (88.0716) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2461 data: 0.1517 max mem: 3532 +Test: [430/517] eta: 0:00:22 acc1_pos: 24.771 acc5_pos: 100.000 loss: 0.2212 (0.3789) acc1: 100.0000 (88.3405) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2399 data: 0.1467 max mem: 3532 +Test: [440/517] eta: 0:00:20 acc1_pos: 24.771 acc5_pos: 100.000 loss: 0.2622 (0.3773) acc1: 99.2188 (88.5612) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2737 data: 0.1790 max mem: 3533 +Test: [450/517] eta: 0:00:17 acc1_pos: 24.771 acc5_pos: 100.000 loss: 0.2796 (0.3744) acc1: 98.4375 (88.7871) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2885 data: 0.1894 max mem: 3533 +Test: [460/517] eta: 0:00:15 acc1_pos: 24.771 acc5_pos: 100.000 loss: 0.2538 (0.3724) acc1: 99.2188 (88.9840) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.3099 data: 0.2105 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 24.771 acc5_pos: 100.000 loss: 0.2538 (0.3695) acc1: 99.2188 (89.2013) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2799 data: 0.1835 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 24.771 acc5_pos: 100.000 loss: 0.2366 (0.3666) acc1: 99.4792 (89.4106) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2349 data: 0.1380 max mem: 3533 +Test: [490/517] eta: 0:00:07 acc1_pos: 24.771 acc5_pos: 100.000 loss: 0.2449 (0.3642) acc1: 99.2188 (89.5844) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2299 data: 0.1329 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 24.771 acc5_pos: 100.000 loss: 0.3062 (0.3642) acc1: 98.6979 (89.7419) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1980 data: 0.1017 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 24.771 acc5_pos: 100.000 loss: 0.3532 (0.3641) acc1: 94.7917 (89.7408) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1755 data: 0.0829 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 24.771 acc5_pos: 100.000 loss: 0.2835 (0.3627) acc1: 99.7396 (89.8547) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1756 data: 0.0829 max mem: 3533 +Test: Total time: 0:02:13 (0.2581 s / it) +* Overall Acc@1 89.855 Acc@5 100.000 + target == real Acc@1 10.456 Acc@5 100.000 + target == fake Acc@1 98.092 Acc@5 100.000 + loss 0.363 + PR-AUC 0.9460 ROC-AUC 0.6793 EER 37.02% (thr=0.7583) +Accuracy of the network on the 198427 test images: 89.9% +[rank0]:[W726 05:47:55.168100759 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 05:48:15.443000 1199402 torch/distributed/run.py:766] +W0726 05:48:15.443000 1199402 torch/distributed/run.py:766] ***************************************** +W0726 05:48:15.443000 1199402 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:48:15.443000 1199402 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:48:46.866280504 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:48:46.250227935 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:48:46.333315196 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +[rank0]:[W726 05:48:47.510024872 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:48:47.518822486 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:48:47.521615223 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 6): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:48:47.731741297 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:48:47.735535000 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n45_seed8_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed8_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:37:53 acc1_pos: 94.531 acc5_pos: 100.000 loss: 0.3819 (0.3819) acc1: 94.5312 (94.5312) acc5: 100.0000 (100.0000) time: 5.8434 data: 4.4557 max mem: 3531 +Test: [ 10/389] eta: 0:03:52 acc1_pos: 84.635 acc5_pos: 100.000 loss: 0.4786 (0.5185) acc1: 77.6042 (73.6269) acc5: 100.0000 (100.0000) time: 0.6131 data: 0.4054 max mem: 3531 +Test: [ 20/389] eta: 0:02:15 acc1_pos: 33.854 acc5_pos: 100.000 loss: 0.5590 (0.6060) acc1: 75.5208 (62.8224) acc5: 100.0000 (100.0000) time: 0.0946 data: 0.0036 max mem: 3531 +Test: [ 30/389] eta: 0:01:53 acc1_pos: 94.271 acc5_pos: 100.000 loss: 0.5081 (0.5677) acc1: 80.4688 (70.0689) acc5: 100.0000 (100.0000) time: 0.1535 data: 0.0577 max mem: 3531 +Test: [ 40/389] eta: 0:01:48 acc1_pos: 63.802 acc5_pos: 100.000 loss: 0.4968 (0.5739) acc1: 83.5938 (69.2010) acc5: 100.0000 (100.0000) time: 0.2531 data: 0.1569 max mem: 3531 +Test: [ 50/389] eta: 0:01:42 acc1_pos: 96.094 acc5_pos: 100.000 loss: 0.5604 (0.5629) acc1: 72.9167 (70.7261) acc5: 100.0000 (100.0000) time: 0.2804 data: 0.1884 max mem: 3531 +Test: [ 60/389] eta: 0:01:32 acc1_pos: 38.281 acc5_pos: 100.000 loss: 0.4272 (0.5444) acc1: 90.3646 (73.4759) acc5: 100.0000 (100.0000) time: 0.2200 data: 0.1280 max mem: 3531 +Test: [ 70/389] eta: 0:01:28 acc1_pos: 77.865 acc5_pos: 100.000 loss: 0.4384 (0.5685) acc1: 82.8125 (70.9214) acc5: 100.0000 (100.0000) time: 0.2087 data: 0.1159 max mem: 3531 +Test: [ 80/389] eta: 0:01:21 acc1_pos: 97.396 acc5_pos: 100.000 loss: 0.7427 (0.5799) acc1: 44.7917 (69.5666) acc5: 100.0000 (100.0000) time: 0.2107 data: 0.1193 max mem: 3531 +Test: [ 90/389] eta: 0:01:17 acc1_pos: 73.958 acc5_pos: 100.000 loss: 0.4616 (0.5666) acc1: 75.0000 (71.3542) acc5: 100.0000 (100.0000) time: 0.1947 data: 0.1038 max mem: 3531 +Test: [100/389] eta: 0:01:12 acc1_pos: 77.604 acc5_pos: 100.000 loss: 0.4506 (0.5662) acc1: 83.8542 (71.4728) acc5: 100.0000 (100.0000) time: 0.2039 data: 0.1122 max mem: 3531 +Test: [110/389] eta: 0:01:08 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5296 (0.5650) acc1: 77.6042 (71.2251) acc5: 100.0000 (100.0000) time: 0.1887 data: 0.0963 max mem: 3531 +Test: [120/389] eta: 0:01:05 acc1_pos: 92.448 acc5_pos: 100.000 loss: 0.4193 (0.5549) acc1: 85.9375 (72.3205) acc5: 100.0000 (100.0000) time: 0.1915 data: 0.0993 max mem: 3531 +Test: [130/389] eta: 0:01:01 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.5442 (0.5707) acc1: 71.8750 (70.4954) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.1817 data: 0.0906 max mem: 3531 +Test: [140/389] eta: 0:00:57 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.8330 (0.5932) acc1: 36.7188 (67.7305) acc5: 100.0000 (100.0000) acc1_neg: 10.156 acc5_neg: 100.000 time: 0.1686 data: 0.0778 max mem: 3531 +Test: [150/389] eta: 0:00:54 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.8573 (0.6089) acc1: 29.4271 (65.9251) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1862 data: 0.0953 max mem: 3531 +Test: [160/389] eta: 0:00:51 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.8279 (0.6095) acc1: 35.6771 (66.0828) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.1896 data: 0.0974 max mem: 3531 +Test: [170/389] eta: 0:00:49 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.6308 (0.6171) acc1: 63.5417 (65.3433) acc5: 100.0000 (100.0000) acc1_neg: 52.344 acc5_neg: 100.000 time: 0.1852 data: 0.0934 max mem: 3531 +Test: [180/389] eta: 0:00:46 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.6392 (0.6208) acc1: 63.5417 (64.9589) acc5: 100.0000 (100.0000) acc1_neg: 64.583 acc5_neg: 100.000 time: 0.1684 data: 0.0777 max mem: 3531 +Test: [190/389] eta: 0:00:43 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.6459 (0.6309) acc1: 56.7708 (63.9889) acc5: 100.0000 (100.0000) acc1_neg: 38.542 acc5_neg: 100.000 time: 0.1727 data: 0.0808 max mem: 3531 +Test: [200/389] eta: 0:00:40 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.6616 (0.6332) acc1: 59.3750 (63.9511) acc5: 100.0000 (100.0000) acc1_neg: 64.844 acc5_neg: 100.000 time: 0.1850 data: 0.0927 max mem: 3531 +Test: [210/389] eta: 0:00:38 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.6247 (0.6279) acc1: 67.7083 (64.7598) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1811 data: 0.0900 max mem: 3532 +Test: [220/389] eta: 0:00:36 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.4650 (0.6254) acc1: 87.7604 (65.0877) acc5: 100.0000 (100.0000) acc1_neg: 42.448 acc5_neg: 100.000 time: 0.1915 data: 0.1003 max mem: 3532 +Test: [230/389] eta: 0:00:33 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.6729 (0.6293) acc1: 53.9062 (64.4672) acc5: 100.0000 (100.0000) acc1_neg: 40.104 acc5_neg: 100.000 time: 0.1975 data: 0.1058 max mem: 3532 +Test: [240/389] eta: 0:00:31 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.6968 (0.6305) acc1: 53.6458 (64.3564) acc5: 100.0000 (100.0000) acc1_neg: 70.833 acc5_neg: 100.000 time: 0.2047 data: 0.1127 max mem: 3532 +Test: [250/389] eta: 0:00:29 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.5705 (0.6279) acc1: 73.9583 (64.9278) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.2081 data: 0.1164 max mem: 3532 +Test: [260/389] eta: 0:00:27 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.5788 (0.6340) acc1: 73.9583 (64.3748) acc5: 100.0000 (100.0000) acc1_neg: 15.104 acc5_neg: 100.000 time: 0.2105 data: 0.1192 max mem: 3532 +Test: [270/389] eta: 0:00:25 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.7645 (0.6476) acc1: 39.8438 (63.2591) acc5: 100.0000 (100.0000) acc1_neg: 71.615 acc5_neg: 100.000 time: 0.2103 data: 0.1181 max mem: 3532 +Test: [280/389] eta: 0:00:23 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.7571 (0.6539) acc1: 39.8438 (62.8874) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1994 data: 0.1068 max mem: 3532 +Test: [290/389] eta: 0:00:20 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.6094 (0.6553) acc1: 63.5417 (62.8249) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.1974 data: 0.1056 max mem: 3532 +Test: [300/389] eta: 0:00:18 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.5863 (0.6534) acc1: 71.6146 (63.1575) acc5: 100.0000 (100.0000) acc1_neg: 71.615 acc5_neg: 100.000 time: 0.2005 data: 0.1094 max mem: 3532 +Test: [310/389] eta: 0:00:16 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.6054 (0.6606) acc1: 57.2917 (62.4925) acc5: 100.0000 (100.0000) acc1_neg: 1.823 acc5_neg: 100.000 time: 0.2039 data: 0.1126 max mem: 3532 +Test: [320/389] eta: 0:00:14 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.8194 (0.6662) acc1: 40.3646 (62.0384) acc5: 100.0000 (100.0000) acc1_neg: 70.573 acc5_neg: 100.000 time: 0.1917 data: 0.1008 max mem: 3532 +Test: [330/389] eta: 0:00:12 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.9750 (0.6753) acc1: 21.8750 (60.8203) acc5: 100.0000 (100.0000) acc1_neg: 39.062 acc5_neg: 100.000 time: 0.1729 data: 0.0828 max mem: 3532 +Test: [340/389] eta: 0:00:10 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.6895 (0.6703) acc1: 55.9896 (61.5943) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1632 data: 0.0725 max mem: 3532 +Test: [350/389] eta: 0:00:08 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.5521 (0.6763) acc1: 77.8646 (61.0903) acc5: 100.0000 (100.0000) acc1_neg: 7.812 acc5_neg: 100.000 time: 0.1553 data: 0.0642 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.7927 (0.6808) acc1: 42.4479 (60.3517) acc5: 100.0000 (100.0000) acc1_neg: 42.448 acc5_neg: 100.000 time: 0.1530 data: 0.0623 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.7800 (0.6796) acc1: 42.4479 (60.3402) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1469 data: 0.0560 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.6058 (0.6781) acc1: 55.2083 (60.5199) acc5: 100.0000 (100.0000) acc1_neg: 53.646 acc5_neg: 100.000 time: 0.1357 data: 0.0454 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.6573 (0.6779) acc1: 64.2076 (60.5538) acc5: 100.0000 (100.0000) acc1_neg: 64.208 acc5_neg: 100.000 time: 0.1321 data: 0.0392 max mem: 3532 +Test: Total time: 0:01:17 (0.2000 s / it) +* Overall Acc@1 60.554 Acc@5 100.000 + target == real Acc@1 72.111 Acc@5 100.000 + target == fake Acc@1 55.182 Acc@5 100.000 + loss 0.678 + PR-AUC 0.8123 ROC-AUC 0.6861 EER 36.21% (thr=0.4570) +Accuracy of the network on the 149358 test images: 60.6% +[rank0]:[W726 05:50:12.405863518 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:50:32.615000 1199960 torch/distributed/run.py:766] +W0726 05:50:32.615000 1199960 torch/distributed/run.py:766] ***************************************** +W0726 05:50:32.615000 1199960 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:50:32.615000 1199960 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:51:00.799165316 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:51:00.133690227 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:51:00.225631223 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:51:00.242542861 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:51:00.280906960 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:51:01.629893996 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:51:01.651670031 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank6]:[W726 05:51:01.654212586 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n45_seed8_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed8_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:30:19 acc1_pos: 78.385 acc5_pos: 100.000 loss: 0.5010 (0.5010) acc1: 78.3854 (78.3854) acc5: 100.0000 (100.0000) time: 5.4631 data: 3.5625 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/333] eta: 0:03:07 acc1_pos: 80.469 acc5_pos: 100.000 loss: 0.4429 (0.4306) acc1: 82.8125 (82.1970) acc5: 100.0000 (100.0000) time: 0.5808 data: 0.3265 max mem: 3531 +Test: [ 20/333] eta: 0:01:52 acc1_pos: 92.448 acc5_pos: 100.000 loss: 0.4213 (0.4198) acc1: 83.0729 (81.7584) acc5: 100.0000 (100.0000) time: 0.1049 data: 0.0148 max mem: 3531 +Test: [ 30/333] eta: 0:01:30 acc1_pos: 48.698 acc5_pos: 100.000 loss: 0.5245 (0.5530) acc1: 67.4479 (71.6146) acc5: 100.0000 (100.0000) time: 0.1447 data: 0.0524 max mem: 3531 +Test: [ 40/333] eta: 0:01:23 acc1_pos: 51.823 acc5_pos: 100.000 loss: 0.8436 (0.6296) acc1: 46.6146 (64.8819) acc5: 100.0000 (100.0000) time: 0.2069 data: 0.1114 max mem: 3531 +Test: [ 50/333] eta: 0:01:18 acc1_pos: 48.958 acc5_pos: 100.000 loss: 0.8385 (0.6690) acc1: 46.6146 (61.2132) acc5: 100.0000 (100.0000) time: 0.2387 data: 0.1425 max mem: 3531 +Test: [ 60/333] eta: 0:01:10 acc1_pos: 41.667 acc5_pos: 100.000 loss: 0.8015 (0.6944) acc1: 48.9583 (59.0591) acc5: 100.0000 (100.0000) time: 0.2013 data: 0.1067 max mem: 3531 +Test: [ 70/333] eta: 0:01:05 acc1_pos: 49.740 acc5_pos: 100.000 loss: 0.8255 (0.7142) acc1: 49.2188 (57.6401) acc5: 100.0000 (100.0000) time: 0.1834 data: 0.0911 max mem: 3531 +Test: [ 80/333] eta: 0:01:01 acc1_pos: 47.656 acc5_pos: 100.000 loss: 0.8284 (0.7305) acc1: 47.3958 (56.1761) acc5: 100.0000 (100.0000) time: 0.1975 data: 0.1039 max mem: 3531 +Test: [ 90/333] eta: 0:00:57 acc1_pos: 56.250 acc5_pos: 100.000 loss: 0.7663 (0.7334) acc1: 48.1771 (55.9095) acc5: 100.0000 (100.0000) time: 0.1815 data: 0.0873 max mem: 3531 +Test: [100/333] eta: 0:00:53 acc1_pos: 61.979 acc5_pos: 100.000 loss: 0.7516 (0.7379) acc1: 55.9896 (55.6338) acc5: 100.0000 (100.0000) time: 0.1780 data: 0.0863 max mem: 3531 +Test: [110/333] eta: 0:00:50 acc1_pos: 61.719 acc5_pos: 100.000 loss: 0.7638 (0.7412) acc1: 50.5208 (55.3022) acc5: 100.0000 (100.0000) time: 0.1808 data: 0.0883 max mem: 3531 +Test: [120/333] eta: 0:00:46 acc1_pos: 55.469 acc5_pos: 100.000 loss: 0.7638 (0.7427) acc1: 53.3854 (55.0512) acc5: 100.0000 (100.0000) time: 0.1669 data: 0.0736 max mem: 3531 +Test: [130/333] eta: 0:00:43 acc1_pos: 65.365 acc5_pos: 100.000 loss: 0.7553 (0.7435) acc1: 54.4271 (55.1050) acc5: 100.0000 (100.0000) time: 0.1639 data: 0.0716 max mem: 3531 +Test: [140/333] eta: 0:00:40 acc1_pos: 67.969 acc5_pos: 100.000 loss: 0.7224 (0.7414) acc1: 55.2083 (55.2287) acc5: 100.0000 (100.0000) time: 0.1655 data: 0.0723 max mem: 3531 +Test: [150/333] eta: 0:00:38 acc1_pos: 61.198 acc5_pos: 100.000 loss: 0.7224 (0.7434) acc1: 57.8125 (55.0617) acc5: 100.0000 (100.0000) time: 0.1728 data: 0.0802 max mem: 3531 +Test: [160/333] eta: 0:00:35 acc1_pos: 43.750 acc5_pos: 100.000 loss: 0.7589 (0.7468) acc1: 49.7396 (54.7376) acc5: 100.0000 (100.0000) acc1_neg: 50.521 acc5_neg: 100.000 time: 0.1731 data: 0.0804 max mem: 3531 +Test: [170/333] eta: 0:00:33 acc1_pos: 43.750 acc5_pos: 100.000 loss: 0.7628 (0.7490) acc1: 49.7396 (54.5504) acc5: 100.0000 (100.0000) acc1_neg: 57.031 acc5_neg: 100.000 time: 0.1885 data: 0.0936 max mem: 3531 +Test: [180/333] eta: 0:00:31 acc1_pos: 43.750 acc5_pos: 100.000 loss: 0.7196 (0.7456) acc1: 56.7708 (54.8299) acc5: 100.0000 (100.0000) acc1_neg: 58.854 acc5_neg: 100.000 time: 0.1968 data: 0.1035 max mem: 3531 +Test: [190/333] eta: 0:00:29 acc1_pos: 43.750 acc5_pos: 100.000 loss: 0.6618 (0.7374) acc1: 59.8958 (55.4647) acc5: 100.0000 (100.0000) acc1_neg: 65.885 acc5_neg: 100.000 time: 0.1905 data: 0.0988 max mem: 3531 +Test: [200/333] eta: 0:00:27 acc1_pos: 43.750 acc5_pos: 100.000 loss: 0.5960 (0.7337) acc1: 63.2812 (55.7875) acc5: 100.0000 (100.0000) acc1_neg: 56.510 acc5_neg: 100.000 time: 0.1775 data: 0.0867 max mem: 3531 +Test: [210/333] eta: 0:00:24 acc1_pos: 43.750 acc5_pos: 100.000 loss: 0.7229 (0.7349) acc1: 57.0312 (55.7674) acc5: 100.0000 (100.0000) acc1_neg: 57.031 acc5_neg: 100.000 time: 0.1671 data: 0.0754 max mem: 3532 +Test: [220/333] eta: 0:00:22 acc1_pos: 43.750 acc5_pos: 100.000 loss: 0.7562 (0.7371) acc1: 55.4688 (55.5371) acc5: 100.0000 (100.0000) acc1_neg: 66.927 acc5_neg: 100.000 time: 0.1780 data: 0.0858 max mem: 3532 +Test: [230/333] eta: 0:00:20 acc1_pos: 43.750 acc5_pos: 100.000 loss: 0.5639 (0.7284) acc1: 66.1458 (56.3289) acc5: 100.0000 (100.0000) acc1_neg: 76.302 acc5_neg: 100.000 time: 0.1877 data: 0.0964 max mem: 3532 +Test: [240/333] eta: 0:00:18 acc1_pos: 43.750 acc5_pos: 100.000 loss: 0.5606 (0.7239) acc1: 67.7083 (56.6714) acc5: 100.0000 (100.0000) acc1_neg: 66.667 acc5_neg: 100.000 time: 0.1966 data: 0.1052 max mem: 3532 +Test: [250/333] eta: 0:00:16 acc1_pos: 43.750 acc5_pos: 100.000 loss: 0.6186 (0.7198) acc1: 65.8854 (57.0064) acc5: 100.0000 (100.0000) acc1_neg: 54.948 acc5_neg: 100.000 time: 0.1838 data: 0.0925 max mem: 3532 +Test: [260/333] eta: 0:00:14 acc1_pos: 43.750 acc5_pos: 100.000 loss: 0.6186 (0.7162) acc1: 65.1042 (57.2967) acc5: 100.0000 (100.0000) acc1_neg: 65.625 acc5_neg: 100.000 time: 0.1751 data: 0.0816 max mem: 3532 +Test: [270/333] eta: 0:00:12 acc1_pos: 43.750 acc5_pos: 100.000 loss: 0.6302 (0.7135) acc1: 63.8021 (57.5540) acc5: 100.0000 (100.0000) acc1_neg: 63.802 acc5_neg: 100.000 time: 0.1938 data: 0.1006 max mem: 3532 +Test: [280/333] eta: 0:00:10 acc1_pos: 43.750 acc5_pos: 100.000 loss: 0.6529 (0.7138) acc1: 62.5000 (57.4576) acc5: 100.0000 (100.0000) acc1_neg: 65.625 acc5_neg: 100.000 time: 0.1897 data: 0.0962 max mem: 3532 +Test: [290/333] eta: 0:00:08 acc1_pos: 43.750 acc5_pos: 100.000 loss: 0.7722 (0.7162) acc1: 48.1771 (57.1619) acc5: 100.0000 (100.0000) acc1_neg: 56.510 acc5_neg: 100.000 time: 0.1798 data: 0.0862 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 43.750 acc5_pos: 100.000 loss: 0.7724 (0.7179) acc1: 50.0000 (57.0434) acc5: 100.0000 (100.0000) acc1_neg: 56.510 acc5_neg: 100.000 time: 0.1876 data: 0.0930 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 43.750 acc5_pos: 100.000 loss: 0.7000 (0.7142) acc1: 57.0312 (57.3955) acc5: 100.0000 (100.0000) acc1_neg: 69.010 acc5_neg: 100.000 time: 0.1793 data: 0.0839 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 43.750 acc5_pos: 100.000 loss: 0.5797 (0.7106) acc1: 69.0104 (57.7151) acc5: 100.0000 (100.0000) acc1_neg: 47.656 acc5_neg: 100.000 time: 0.1570 data: 0.0644 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 43.750 acc5_pos: 100.000 loss: 0.8216 (0.7153) acc1: 49.4792 (57.3719) acc5: 100.0000 (100.0000) acc1_neg: 49.479 acc5_neg: 100.000 time: 0.1381 data: 0.0467 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 43.750 acc5_pos: 100.000 loss: 0.8484 (0.7163) acc1: 48.9583 (57.2968) acc5: 100.0000 (100.0000) acc1_neg: 44.225 acc5_neg: 100.000 time: 0.1413 data: 0.0477 max mem: 3532 +Test: Total time: 0:01:04 (0.1932 s / it) +* Overall Acc@1 57.297 Acc@5 100.000 + target == real Acc@1 54.953 Acc@5 100.000 + target == fake Acc@1 59.383 Acc@5 100.000 + loss 0.716 + PR-AUC 0.6081 ROC-AUC 0.5995 EER 42.88% (thr=0.5142) +Accuracy of the network on the 127843 test images: 57.3% +[rank0]:[W726 05:52:20.447549193 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:52:39.431000 1200520 torch/distributed/run.py:766] +W0726 05:52:39.431000 1200520 torch/distributed/run.py:766] ***************************************** +W0726 05:52:39.431000 1200520 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:52:39.431000 1200520 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 05:53:09.824859558 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:53:09.874043166 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:53:09.079459029 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:53:09.112948666 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:53:09.333245852 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:53:10.416402457 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:53:10.493556077 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:53:10.499270780 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n45_seed8_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed8_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:18:51 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.2805 (0.2805) acc1: 95.3125 (95.3125) acc5: 100.0000 (100.0000) time: 4.8140 data: 2.8349 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/235] eta: 0:02:03 acc1_pos: 94.531 acc5_pos: 100.000 loss: 0.2797 (0.2838) acc1: 94.5312 (94.0578) acc5: 100.0000 (100.0000) time: 0.5470 data: 0.2854 max mem: 3531 +Test: [ 20/235] eta: 0:01:11 acc1_pos: 95.052 acc5_pos: 100.000 loss: 0.2775 (0.2804) acc1: 94.5312 (94.2584) acc5: 100.0000 (100.0000) time: 0.1104 data: 0.0206 max mem: 3531 +Test: [ 30/235] eta: 0:00:52 acc1_pos: 95.052 acc5_pos: 100.000 loss: 0.2748 (0.2794) acc1: 94.5312 (94.2876) acc5: 100.0000 (100.0000) time: 0.0954 data: 0.0056 max mem: 3531 +Test: [ 40/235] eta: 0:00:44 acc1_pos: 94.010 acc5_pos: 100.000 loss: 0.2764 (0.2792) acc1: 94.5312 (94.2899) acc5: 100.0000 (100.0000) time: 0.1184 data: 0.0244 max mem: 3531 +Test: [ 50/235] eta: 0:00:41 acc1_pos: 95.052 acc5_pos: 100.000 loss: 0.2796 (0.2802) acc1: 94.5312 (94.2402) acc5: 100.0000 (100.0000) time: 0.1731 data: 0.0773 max mem: 3531 +Test: [ 60/235] eta: 0:00:38 acc1_pos: 95.052 acc5_pos: 100.000 loss: 0.2837 (0.2814) acc1: 94.0104 (94.2068) acc5: 100.0000 (100.0000) time: 0.2099 data: 0.1167 max mem: 3531 +Test: [ 70/235] eta: 0:00:36 acc1_pos: 93.846 acc5_pos: 100.000 loss: 0.2983 (0.3918) acc1: 92.7083 (85.7138) acc5: 100.0000 (100.0000) acc1_neg: 21.615 acc5_neg: 100.000 time: 0.2249 data: 0.1300 max mem: 3531 +Test: [ 80/235] eta: 0:00:35 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.1629 (0.4880) acc1: 26.5625 (78.2890) acc5: 100.0000 (100.0000) acc1_neg: 21.875 acc5_neg: 100.000 time: 0.2410 data: 0.1440 max mem: 3531 +Test: [ 90/235] eta: 0:00:32 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.1732 (0.5640) acc1: 25.5208 (72.4416) acc5: 100.0000 (100.0000) acc1_neg: 24.740 acc5_neg: 100.000 time: 0.2099 data: 0.1146 max mem: 3531 +Test: [100/235] eta: 0:00:28 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.1732 (0.6222) acc1: 25.2604 (67.8759) acc5: 100.0000 (100.0000) acc1_neg: 25.260 acc5_neg: 100.000 time: 0.1338 data: 0.0374 max mem: 3531 +Test: [110/235] eta: 0:00:25 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.1615 (0.6716) acc1: 26.0417 (64.1329) acc5: 100.0000 (100.0000) acc1_neg: 30.208 acc5_neg: 100.000 time: 0.1154 data: 0.0188 max mem: 3531 +Test: [120/235] eta: 0:00:22 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.2045 (0.7298) acc1: 22.6562 (60.0486) acc5: 100.0000 (100.0000) acc1_neg: 15.885 acc5_neg: 100.000 time: 0.1372 data: 0.0441 max mem: 3531 +Test: [130/235] eta: 0:00:20 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.3857 (0.7803) acc1: 13.8021 (56.5084) acc5: 100.0000 (100.0000) acc1_neg: 13.802 acc5_neg: 100.000 time: 0.1459 data: 0.0545 max mem: 3531 +Test: [140/235] eta: 0:00:18 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.4175 (0.8258) acc1: 13.0208 (53.4427) acc5: 100.0000 (100.0000) acc1_neg: 11.719 acc5_neg: 100.000 time: 0.1483 data: 0.0572 max mem: 3531 +Test: [150/235] eta: 0:00:16 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.4103 (0.8635) acc1: 13.5417 (50.8433) acc5: 100.0000 (100.0000) acc1_neg: 14.062 acc5_neg: 100.000 time: 0.1584 data: 0.0646 max mem: 3531 +Test: [160/235] eta: 0:00:14 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.3949 (0.8967) acc1: 14.3229 (48.5879) acc5: 100.0000 (100.0000) acc1_neg: 15.625 acc5_neg: 100.000 time: 0.1692 data: 0.0722 max mem: 3531 +Test: [170/235] eta: 0:00:12 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.3498 (0.9016) acc1: 16.4062 (47.9334) acc5: 100.0000 (100.0000) acc1_neg: 37.240 acc5_neg: 100.000 time: 0.1599 data: 0.0630 max mem: 3531 +Test: [180/235] eta: 0:00:10 acc1_pos: 93.846 acc5_pos: 100.000 loss: 0.9256 (0.9031) acc1: 40.1042 (47.5656) acc5: 100.0000 (100.0000) acc1_neg: 40.885 acc5_neg: 100.000 time: 0.1549 data: 0.0594 max mem: 3531 +Test: [190/235] eta: 0:00:08 acc1_pos: 93.846 acc5_pos: 100.000 loss: 0.9220 (0.9042) acc1: 41.1458 (47.2281) acc5: 100.0000 (100.0000) acc1_neg: 45.312 acc5_neg: 100.000 time: 0.1649 data: 0.0692 max mem: 3531 +Test: [200/235] eta: 0:00:06 acc1_pos: 93.846 acc5_pos: 100.000 loss: 0.9241 (0.9061) acc1: 40.3646 (46.8465) acc5: 100.0000 (100.0000) acc1_neg: 36.979 acc5_neg: 100.000 time: 0.1649 data: 0.0689 max mem: 3531 +Test: [210/235] eta: 0:00:04 acc1_pos: 93.846 acc5_pos: 100.000 loss: 0.9381 (0.9073) acc1: 39.5833 (46.5492) acc5: 100.0000 (100.0000) acc1_neg: 40.885 acc5_neg: 100.000 time: 0.1599 data: 0.0609 max mem: 3532 +Test: [220/235] eta: 0:00:02 acc1_pos: 93.846 acc5_pos: 100.000 loss: 0.9325 (0.9082) acc1: 40.8854 (46.3035) acc5: 100.0000 (100.0000) acc1_neg: 38.281 acc5_neg: 100.000 time: 0.1899 data: 0.0921 max mem: 3532 +Test: [230/235] eta: 0:00:00 acc1_pos: 93.846 acc5_pos: 100.000 loss: 0.9193 (0.9084) acc1: 41.1458 (46.1095) acc5: 100.0000 (100.0000) acc1_neg: 43.229 acc5_neg: 100.000 time: 0.2049 data: 0.1103 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 93.846 acc5_pos: 100.000 loss: 0.9193 (0.9094) acc1: 40.8854 (46.0386) acc5: 100.0000 (100.0000) acc1_neg: 34.783 acc5_neg: 100.000 time: 0.1452 data: 0.0530 max mem: 3532 +Test: Total time: 0:00:42 (0.1828 s / it) +* Overall Acc@1 46.039 Acc@5 100.000 + target == real Acc@1 94.211 Acc@5 100.000 + target == fake Acc@1 28.616 Acc@5 100.000 + loss 0.909 + PR-AUC 0.8795 ROC-AUC 0.7273 EER 33.30% (thr=0.2546) +Accuracy of the network on the 89879 test images: 46.0% +[rank0]:[W726 05:54:08.430138008 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:54:28.347000 1201078 torch/distributed/run.py:766] +W0726 05:54:28.347000 1201078 torch/distributed/run.py:766] ***************************************** +W0726 05:54:28.347000 1201078 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:54:28.347000 1201078 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 4): env:// +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:54:56.246507105 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:54:56.330642542 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:54:57.559101044 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:54:57.803238045 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:54:57.875684086 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:54:57.893560102 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +[rank0]:[W726 05:54:57.913637015 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:54:57.919630597 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n45_seed8_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed8_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:30:26 acc1_pos: 47.396 acc5_pos: 100.000 loss: 0.7486 (0.7486) acc1: 47.3958 (47.3958) acc5: 100.0000 (100.0000) time: 6.7888 data: 5.3268 max mem: 3531 +Test: [ 10/269] eta: 0:03:01 acc1_pos: 78.125 acc5_pos: 100.000 loss: 0.6647 (0.6648) acc1: 60.1562 (61.8845) acc5: 100.0000 (100.0000) time: 0.6996 data: 0.4847 max mem: 3531 +Test: [ 20/269] eta: 0:01:42 acc1_pos: 78.385 acc5_pos: 100.000 loss: 0.6291 (0.6279) acc1: 64.3229 (66.3194) acc5: 100.0000 (100.0000) time: 0.0933 data: 0.0005 max mem: 3531 +Test: [ 30/269] eta: 0:01:16 acc1_pos: 67.188 acc5_pos: 100.000 loss: 0.5878 (0.6111) acc1: 73.1771 (68.6324) acc5: 100.0000 (100.0000) time: 0.1096 data: 0.0145 max mem: 3531 +Test: [ 40/269] eta: 0:01:04 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.5442 (0.6011) acc1: 73.1771 (69.8552) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.1409 data: 0.0483 max mem: 3531 +Test: [ 50/269] eta: 0:00:55 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.5476 (0.6006) acc1: 73.1771 (70.4197) acc5: 100.0000 (100.0000) acc1_neg: 65.365 acc5_neg: 100.000 time: 0.1532 data: 0.0617 max mem: 3531 +Test: [ 60/269] eta: 0:00:49 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.6202 (0.6067) acc1: 66.4062 (69.7490) acc5: 100.0000 (100.0000) acc1_neg: 55.208 acc5_neg: 100.000 time: 0.1449 data: 0.0529 max mem: 3531 +Test: [ 70/269] eta: 0:00:44 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.6452 (0.6136) acc1: 59.6354 (68.2438) acc5: 100.0000 (100.0000) acc1_neg: 64.844 acc5_neg: 100.000 time: 0.1507 data: 0.0589 max mem: 3531 +Test: [ 80/269] eta: 0:00:41 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.6242 (0.6125) acc1: 61.1979 (68.0877) acc5: 100.0000 (100.0000) acc1_neg: 59.375 acc5_neg: 100.000 time: 0.1739 data: 0.0823 max mem: 3531 +Test: [ 90/269] eta: 0:00:38 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.6489 (0.6241) acc1: 59.3750 (65.8253) acc5: 100.0000 (100.0000) acc1_neg: 54.688 acc5_neg: 100.000 time: 0.1747 data: 0.0817 max mem: 3531 +Test: [100/269] eta: 0:00:35 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.7059 (0.6314) acc1: 49.7396 (64.5911) acc5: 100.0000 (100.0000) acc1_neg: 44.531 acc5_neg: 100.000 time: 0.1605 data: 0.0665 max mem: 3531 +Test: [110/269] eta: 0:00:32 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.6940 (0.6375) acc1: 52.8646 (63.6003) acc5: 100.0000 (100.0000) acc1_neg: 58.073 acc5_neg: 100.000 time: 0.1593 data: 0.0652 max mem: 3531 +Test: [120/269] eta: 0:00:30 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.6725 (0.6382) acc1: 58.0729 (63.4965) acc5: 100.0000 (100.0000) acc1_neg: 62.240 acc5_neg: 100.000 time: 0.1843 data: 0.0896 max mem: 3531 +Test: [130/269] eta: 0:00:28 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.6710 (0.6451) acc1: 57.0312 (62.2336) acc5: 100.0000 (100.0000) acc1_neg: 37.760 acc5_neg: 100.000 time: 0.2249 data: 0.1301 max mem: 3531 +Test: [140/269] eta: 0:00:27 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.7888 (0.6569) acc1: 40.3646 (60.5570) acc5: 100.0000 (100.0000) acc1_neg: 44.271 acc5_neg: 100.000 time: 0.2499 data: 0.1539 max mem: 3531 +Test: [150/269] eta: 0:00:24 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.7572 (0.6587) acc1: 41.6667 (60.1614) acc5: 100.0000 (100.0000) acc1_neg: 59.115 acc5_neg: 100.000 time: 0.1799 data: 0.0818 max mem: 3531 +Test: [160/269] eta: 0:00:22 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.6094 (0.6557) acc1: 60.1562 (60.5703) acc5: 100.0000 (100.0000) acc1_neg: 66.667 acc5_neg: 100.000 time: 0.1472 data: 0.0508 max mem: 3531 +Test: [170/269] eta: 0:00:20 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.6412 (0.6559) acc1: 60.1562 (60.5111) acc5: 100.0000 (100.0000) acc1_neg: 48.698 acc5_neg: 100.000 time: 0.2114 data: 0.1171 max mem: 3531 +Test: [180/269] eta: 0:00:18 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.6768 (0.6576) acc1: 57.8125 (60.2109) acc5: 100.0000 (100.0000) acc1_neg: 52.865 acc5_neg: 100.000 time: 0.2010 data: 0.1078 max mem: 3531 +Test: [190/269] eta: 0:00:15 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.6768 (0.6598) acc1: 55.4688 (59.8822) acc5: 100.0000 (100.0000) acc1_neg: 43.229 acc5_neg: 100.000 time: 0.1696 data: 0.0789 max mem: 3531 +Test: [200/269] eta: 0:00:13 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.7795 (0.6677) acc1: 43.2292 (58.8801) acc5: 100.0000 (100.0000) acc1_neg: 52.604 acc5_neg: 100.000 time: 0.1735 data: 0.0832 max mem: 3531 +Test: [210/269] eta: 0:00:11 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.7219 (0.6679) acc1: 46.3542 (58.8344) acc5: 100.0000 (100.0000) acc1_neg: 59.896 acc5_neg: 100.000 time: 0.1756 data: 0.0839 max mem: 3532 +Test: [220/269] eta: 0:00:09 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.6521 (0.6667) acc1: 57.5521 (58.9143) acc5: 100.0000 (100.0000) acc1_neg: 68.750 acc5_neg: 100.000 time: 0.1748 data: 0.0815 max mem: 3532 +Test: [230/269] eta: 0:00:07 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.6572 (0.6653) acc1: 54.9479 (58.9613) acc5: 100.0000 (100.0000) acc1_neg: 57.552 acc5_neg: 100.000 time: 0.1809 data: 0.0883 max mem: 3532 +Test: [240/269] eta: 0:00:05 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.6943 (0.6695) acc1: 54.1667 (58.5278) acc5: 100.0000 (100.0000) acc1_neg: 54.167 acc5_neg: 100.000 time: 0.1741 data: 0.0825 max mem: 3532 +Test: [250/269] eta: 0:00:03 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.7772 (0.6753) acc1: 48.1771 (57.7586) acc5: 100.0000 (100.0000) acc1_neg: 22.396 acc5_neg: 100.000 time: 0.1688 data: 0.0754 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.8566 (0.6815) acc1: 28.6458 (56.9554) acc5: 100.0000 (100.0000) acc1_neg: 52.604 acc5_neg: 100.000 time: 0.1516 data: 0.0582 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 68.103 acc5_pos: 100.000 loss: 0.7221 (0.6809) acc1: 44.5312 (56.9709) acc5: 100.0000 (100.0000) acc1_neg: 67.717 acc5_neg: 100.000 time: 0.1341 data: 0.0398 max mem: 3532 +Test: Total time: 0:00:51 (0.1912 s / it) +* Overall Acc@1 56.971 Acc@5 100.000 + target == real Acc@1 68.236 Acc@5 100.000 + target == fake Acc@1 55.434 Acc@5 100.000 + loss 0.681 + PR-AUC 0.9292 ROC-AUC 0.6620 EER 37.94% (thr=0.4734) +Accuracy of the network on the 103293 test images: 57.0% +[rank0]:[W726 05:56:02.777570146 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T05:56:09Z] eval_done tag=fixed500k_fake_n45_seed8_resampled +[2026-07-26T05:56:10Z] eval_start tag=fixed500k_real_n23_seed9_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed9_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T05:56:10Z +hostname=n116-041-023 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed9_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n23_seed9_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 05:56:22.855000 1201641 torch/distributed/run.py:766] +W0726 05:56:22.855000 1201641 torch/distributed/run.py:766] ***************************************** +W0726 05:56:22.855000 1201641 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:56:22.855000 1201641 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:56:51.993510488 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:56:51.306928787 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:56:51.377046099 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:56:52.473987330 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +[rank0]:[W726 05:56:52.679837941 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:56:52.682120149 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:56:52.693436533 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W726 05:56:52.693695859 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n23_seed9_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed9_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:22:41 acc1_pos: 89.062 acc5_pos: 100.000 loss: 0.3464 (0.3464) acc1: 89.0625 (89.0625) acc5: 100.0000 (100.0000) time: 5.8203 data: 3.6365 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/234] eta: 0:02:17 acc1_pos: 96.354 acc5_pos: 100.000 loss: 0.4276 (0.4995) acc1: 85.1562 (81.3920) acc5: 100.0000 (100.0000) time: 0.6160 data: 0.3356 max mem: 3531 +Test: [ 20/234] eta: 0:01:25 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.4234 (0.4409) acc1: 87.2396 (85.1935) acc5: 100.0000 (100.0000) time: 0.1277 data: 0.0315 max mem: 3531 +Test: [ 30/234] eta: 0:01:12 acc1_pos: 64.062 acc5_pos: 100.000 loss: 0.2414 (0.3934) acc1: 94.7917 (86.6515) acc5: 100.0000 (100.0000) time: 0.2099 data: 0.1124 max mem: 3531 +Test: [ 40/234] eta: 0:01:06 acc1_pos: 94.531 acc5_pos: 100.000 loss: 0.2398 (0.4246) acc1: 93.4896 (82.7807) acc5: 100.0000 (100.0000) time: 0.2849 data: 0.1915 max mem: 3531 +Test: [ 50/234] eta: 0:00:59 acc1_pos: 96.354 acc5_pos: 100.000 loss: 0.2308 (0.3846) acc1: 93.4896 (85.0388) acc5: 100.0000 (100.0000) time: 0.2808 data: 0.1849 max mem: 3531 +Test: [ 60/234] eta: 0:00:50 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.1547 (0.3368) acc1: 96.3542 (87.4488) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0772 max mem: 3531 +Test: [ 70/234] eta: 0:00:45 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.0875 (0.3441) acc1: 100.0000 (86.6014) acc5: 100.0000 (100.0000) acc1_neg: 72.917 acc5_neg: 100.000 time: 0.1490 data: 0.0528 max mem: 3531 +Test: [ 80/234] eta: 0:00:42 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.5074 (0.3485) acc1: 73.1771 (86.1015) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2449 data: 0.1513 max mem: 3531 +Test: [ 90/234] eta: 0:00:38 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.1540 (0.3255) acc1: 96.0938 (87.3312) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.2353 data: 0.1440 max mem: 3531 +Test: [100/234] eta: 0:00:34 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.2136 (0.3287) acc1: 92.9688 (86.9663) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1749 data: 0.0817 max mem: 3531 +Test: [110/234] eta: 0:00:31 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.3635 (0.3294) acc1: 82.2917 (86.8337) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.1868 data: 0.0935 max mem: 3531 +Test: [120/234] eta: 0:00:28 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.3635 (0.3345) acc1: 83.5938 (86.5767) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1927 data: 0.1011 max mem: 3531 +Test: [130/234] eta: 0:00:25 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.4702 (0.3532) acc1: 77.3438 (85.0529) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.1835 data: 0.0890 max mem: 3531 +Test: [140/234] eta: 0:00:22 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.5051 (0.3813) acc1: 71.0938 (83.5901) acc5: 100.0000 (100.0000) acc1_neg: 41.406 acc5_neg: 100.000 time: 0.1890 data: 0.0959 max mem: 3531 +Test: [150/234] eta: 0:00:19 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.3966 (0.3871) acc1: 80.9896 (83.2557) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1962 data: 0.1059 max mem: 3531 +Test: [160/234] eta: 0:00:17 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.3917 (0.3948) acc1: 81.7708 (82.5068) acc5: 100.0000 (100.0000) acc1_neg: 54.427 acc5_neg: 100.000 time: 0.1880 data: 0.0933 max mem: 3531 +Test: [170/234] eta: 0:00:14 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.6570 (0.4095) acc1: 59.3750 (81.3033) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.1848 data: 0.0877 max mem: 3531 +Test: [180/234] eta: 0:00:12 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.6284 (0.4177) acc1: 63.0208 (80.6975) acc5: 100.0000 (100.0000) acc1_neg: 45.052 acc5_neg: 100.000 time: 0.1972 data: 0.1035 max mem: 3531 +Test: [190/234] eta: 0:00:09 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.7386 (0.4357) acc1: 59.1146 (79.4284) acc5: 100.0000 (100.0000) acc1_neg: 53.906 acc5_neg: 100.000 time: 0.1952 data: 0.1036 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.7347 (0.4311) acc1: 59.1146 (79.6720) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1771 data: 0.0855 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.1152 (0.4160) acc1: 99.7396 (80.6292) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1572 data: 0.0659 max mem: 3532 +Test: [220/234] eta: 0:00:03 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.1061 (0.4032) acc1: 100.0000 (81.3949) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.1378 data: 0.0480 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.3939 (0.4266) acc1: 43.2292 (79.4395) acc5: 100.0000 (100.0000) acc1_neg: 35.417 acc5_neg: 100.000 time: 0.1251 data: 0.0355 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.9170 (0.4330) acc1: 38.1323 (78.9644) acc5: 100.0000 (100.0000) acc1_neg: 38.132 acc5_neg: 100.000 time: 0.1235 data: 0.0328 max mem: 3532 +Test: Total time: 0:00:49 (0.2111 s / it) +* Overall Acc@1 78.964 Acc@5 100.000 + target == real Acc@1 85.103 Acc@5 100.000 + target == fake Acc@1 77.234 Acc@5 100.000 + loss 0.433 + PR-AUC 0.9686 ROC-AUC 0.8914 EER 19.56% (thr=0.4480) +Accuracy of the network on the 89729 test images: 79.0% +[rank0]:[W726 05:57:48.274230458 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:58:08.053000 1202199 torch/distributed/run.py:766] +W0726 05:58:08.053000 1202199 torch/distributed/run.py:766] ***************************************** +W0726 05:58:08.053000 1202199 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:58:08.053000 1202199 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:58:36.356104934 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:58:37.622865735 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:58:37.779888240 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:58:37.081790566 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:58:37.131903761 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:58:37.194772343 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:58:37.273583493 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:58:37.374247450 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n23_seed9_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed9_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:35:13 acc1_pos: 78.125 acc5_pos: 100.000 loss: 0.4971 (0.4971) acc1: 78.1250 (78.1250) acc5: 100.0000 (100.0000) time: 6.9182 data: 5.2538 max mem: 3531 +Test: [ 10/2387] eta: 0:28:09 acc1_pos: 52.083 acc5_pos: 100.000 loss: 0.6286 (0.6279) acc1: 67.1875 (68.4422) acc5: 100.0000 (100.0000) time: 0.7107 data: 0.4780 max mem: 3531 +Test: [ 20/2387] eta: 0:16:22 acc1_pos: 64.583 acc5_pos: 100.000 loss: 0.6469 (0.6702) acc1: 64.5833 (65.4762) acc5: 100.0000 (100.0000) time: 0.0900 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:14:48 acc1_pos: 44.531 acc5_pos: 100.000 loss: 0.6855 (0.6785) acc1: 60.9375 (63.2897) acc5: 100.0000 (100.0000) time: 0.1936 data: 0.0981 max mem: 3531 +Test: [ 40/2387] eta: 0:14:24 acc1_pos: 70.312 acc5_pos: 100.000 loss: 0.6612 (0.6730) acc1: 61.9792 (63.6433) acc5: 100.0000 (100.0000) time: 0.3192 data: 0.2235 max mem: 3531 +Test: [ 50/2387] eta: 0:13:42 acc1_pos: 52.083 acc5_pos: 100.000 loss: 0.6203 (0.6669) acc1: 65.3646 (64.1289) acc5: 100.0000 (100.0000) time: 0.3125 data: 0.2202 max mem: 3531 +Test: [ 60/2387] eta: 0:12:54 acc1_pos: 66.146 acc5_pos: 100.000 loss: 0.6280 (0.6715) acc1: 66.1458 (63.8362) acc5: 100.0000 (100.0000) time: 0.2605 data: 0.1666 max mem: 3531 +Test: [ 70/2387] eta: 0:12:20 acc1_pos: 44.792 acc5_pos: 100.000 loss: 0.7143 (0.6848) acc1: 57.8125 (62.7384) acc5: 100.0000 (100.0000) time: 0.2373 data: 0.1406 max mem: 3531 +Test: [ 80/2387] eta: 0:11:43 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.6773 (0.6568) acc1: 68.7500 (65.4803) acc5: 100.0000 (100.0000) time: 0.2199 data: 0.1225 max mem: 3531 +Test: [ 90/2387] eta: 0:11:11 acc1_pos: 97.135 acc5_pos: 100.000 loss: 0.3820 (0.6251) acc1: 91.6667 (68.6298) acc5: 100.0000 (100.0000) time: 0.1949 data: 0.0983 max mem: 3531 +Test: [ 100/2387] eta: 0:10:55 acc1_pos: 95.573 acc5_pos: 100.000 loss: 0.3820 (0.6052) acc1: 91.6667 (70.5239) acc5: 100.0000 (100.0000) time: 0.2111 data: 0.1114 max mem: 3531 +Test: [ 110/2387] eta: 0:10:29 acc1_pos: 77.344 acc5_pos: 100.000 loss: 0.4098 (0.5862) acc1: 89.0625 (72.1988) acc5: 100.0000 (100.0000) time: 0.2053 data: 0.1098 max mem: 3531 +Test: [ 120/2387] eta: 0:10:29 acc1_pos: 97.135 acc5_pos: 100.000 loss: 0.3862 (0.5719) acc1: 89.0625 (73.5236) acc5: 100.0000 (100.0000) time: 0.2326 data: 0.1369 max mem: 3531 +Test: [ 130/2387] eta: 0:10:20 acc1_pos: 86.979 acc5_pos: 100.000 loss: 0.3893 (0.5595) acc1: 89.5833 (74.7694) acc5: 100.0000 (100.0000) time: 0.2655 data: 0.1676 max mem: 3531 +Test: [ 140/2387] eta: 0:10:13 acc1_pos: 96.615 acc5_pos: 100.000 loss: 0.3893 (0.5492) acc1: 90.3646 (75.9013) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1497 max mem: 3531 +Test: [ 150/2387] eta: 0:10:08 acc1_pos: 99.740 acc5_pos: 100.000 loss: 0.3654 (0.5354) acc1: 93.4896 (77.1920) acc5: 100.0000 (100.0000) time: 0.2540 data: 0.1559 max mem: 3531 +Test: [ 160/2387] eta: 0:09:55 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.3331 (0.5264) acc1: 93.4896 (77.8403) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2299 data: 0.1317 max mem: 3531 +Test: [ 170/2387] eta: 0:09:43 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.2169 (0.5076) acc1: 93.4896 (78.8057) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1982 data: 0.1047 max mem: 3531 +Test: [ 180/2387] eta: 0:09:34 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.2099 (0.4912) acc1: 93.7500 (79.6544) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1999 data: 0.1056 max mem: 3531 +Test: [ 190/2387] eta: 0:09:24 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.2033 (0.4762) acc1: 93.7500 (80.4197) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2016 data: 0.1041 max mem: 3531 +Test: [ 200/2387] eta: 0:09:12 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.2024 (0.4626) acc1: 94.0104 (81.1230) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1849 data: 0.0892 max mem: 3531 +Test: [ 210/2387] eta: 0:09:08 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.2047 (0.4506) acc1: 94.5312 (81.7511) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2021 data: 0.1085 max mem: 3532 +Test: [ 220/2387] eta: 0:09:02 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.2082 (0.4395) acc1: 94.2708 (82.3117) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2267 data: 0.1337 max mem: 3532 +Test: [ 230/2387] eta: 0:08:53 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.2045 (0.4294) acc1: 94.0104 (82.8362) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1992 data: 0.1067 max mem: 3532 +Test: [ 240/2387] eta: 0:08:46 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.2096 (0.4251) acc1: 91.4062 (83.0502) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1904 data: 0.0974 max mem: 3532 +Test: [ 250/2387] eta: 0:08:46 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.2016 (0.4133) acc1: 89.3229 (83.6404) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2355 data: 0.1424 max mem: 3532 +Test: [ 260/2387] eta: 0:08:36 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.0963 (0.4011) acc1: 99.4792 (84.2433) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2126 data: 0.1190 max mem: 3532 +Test: [ 270/2387] eta: 0:08:32 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.0882 (0.3895) acc1: 99.4792 (84.7969) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1929 data: 0.0981 max mem: 3532 +Test: [ 280/2387] eta: 0:08:28 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.0861 (0.3789) acc1: 99.7396 (85.3305) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2205 data: 0.1274 max mem: 3532 +Test: [ 290/2387] eta: 0:08:20 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.0955 (0.3693) acc1: 99.7396 (85.8068) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1917 data: 0.0988 max mem: 3532 +Test: [ 300/2387] eta: 0:08:14 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.0996 (0.3606) acc1: 99.4792 (86.2559) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1795 data: 0.0860 max mem: 3532 +Test: [ 310/2387] eta: 0:08:10 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1137 (0.3550) acc1: 99.2188 (86.6442) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1969 data: 0.1027 max mem: 3532 +Test: [ 320/2387] eta: 0:08:06 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.2599 (0.3523) acc1: 97.6562 (86.9873) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2088 data: 0.1135 max mem: 3532 +Test: [ 330/2387] eta: 0:08:01 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.2648 (0.3496) acc1: 97.6562 (87.3088) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2059 data: 0.1091 max mem: 3532 +Test: [ 340/2387] eta: 0:07:58 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.2648 (0.3470) acc1: 97.6562 (87.6191) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2053 data: 0.1101 max mem: 3532 +Test: [ 350/2387] eta: 0:07:55 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.2460 (0.3411) acc1: 98.6979 (87.9659) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2190 data: 0.1265 max mem: 3532 +Test: [ 360/2387] eta: 0:07:50 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1623 (0.3365) acc1: 98.4375 (88.1896) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2102 data: 0.1165 max mem: 3532 +Test: [ 370/2387] eta: 0:07:47 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1793 (0.3323) acc1: 95.5729 (88.3929) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2019 data: 0.1086 max mem: 3532 +Test: [ 380/2387] eta: 0:07:41 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1793 (0.3283) acc1: 96.0938 (88.6018) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1918 data: 0.1001 max mem: 3532 +Test: [ 390/2387] eta: 0:07:36 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1797 (0.3245) acc1: 96.3542 (88.7954) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1699 data: 0.0787 max mem: 3532 +Test: [ 400/2387] eta: 0:07:32 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1793 (0.3209) acc1: 96.3542 (88.9748) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1794 data: 0.0881 max mem: 3532 +Test: [ 410/2387] eta: 0:07:27 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1775 (0.3175) acc1: 96.3542 (89.1480) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1879 data: 0.0960 max mem: 3532 +Test: [ 420/2387] eta: 0:07:24 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1759 (0.3142) acc1: 96.0938 (89.3087) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1929 data: 0.1010 max mem: 3532 +Test: [ 430/2387] eta: 0:07:21 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1841 (0.3128) acc1: 95.0521 (89.3592) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.2049 data: 0.1109 max mem: 3532 +Test: [ 440/2387] eta: 0:07:13 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1920 (0.3096) acc1: 93.2292 (89.5142) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1575 data: 0.0598 max mem: 3533 +Test: [ 450/2387] eta: 0:07:09 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.0930 (0.3048) acc1: 100.0000 (89.7467) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1418 data: 0.0469 max mem: 3533 +Test: [ 460/2387] eta: 0:07:05 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.0927 (0.3002) acc1: 100.0000 (89.9686) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1753 data: 0.0843 max mem: 3533 +Test: [ 470/2387] eta: 0:07:01 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.0925 (0.2958) acc1: 100.0000 (90.1816) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1689 data: 0.0782 max mem: 3533 +Test: [ 480/2387] eta: 0:06:56 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.0931 (0.2916) acc1: 100.0000 (90.3857) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1642 data: 0.0727 max mem: 3533 +Test: [ 490/2387] eta: 0:06:53 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.0931 (0.2876) acc1: 100.0000 (90.5815) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1811 data: 0.0894 max mem: 3533 +Test: [ 500/2387] eta: 0:06:49 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.0934 (0.2837) acc1: 100.0000 (90.7695) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1776 data: 0.0864 max mem: 3533 +Test: [ 510/2387] eta: 0:06:44 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.0933 (0.2802) acc1: 100.0000 (90.9501) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1545 data: 0.0622 max mem: 3533 +Test: [ 520/2387] eta: 0:06:41 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1206 (0.2774) acc1: 100.0000 (91.1228) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1669 data: 0.0751 max mem: 3533 +Test: [ 530/2387] eta: 0:06:38 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1202 (0.2744) acc1: 100.0000 (91.2856) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1844 data: 0.0927 max mem: 3533 +Test: [ 540/2387] eta: 0:06:35 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1180 (0.2716) acc1: 100.0000 (91.4448) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1937 data: 0.1015 max mem: 3533 +Test: [ 550/2387] eta: 0:06:32 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1186 (0.2688) acc1: 100.0000 (91.5967) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1952 data: 0.1015 max mem: 3533 +Test: [ 560/2387] eta: 0:06:30 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1180 (0.2661) acc1: 99.7396 (91.7437) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1965 data: 0.1025 max mem: 3533 +Test: [ 570/2387] eta: 0:06:27 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1176 (0.2635) acc1: 99.7396 (91.8833) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1934 data: 0.0996 max mem: 3533 +Test: [ 580/2387] eta: 0:06:24 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1182 (0.2610) acc1: 99.7396 (92.0203) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1904 data: 0.0971 max mem: 3533 +Test: [ 590/2387] eta: 0:06:21 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1193 (0.2586) acc1: 99.7396 (92.1523) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1919 data: 0.1005 max mem: 3533 +Test: [ 600/2387] eta: 0:06:18 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1260 (0.2567) acc1: 100.0000 (92.2820) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1878 data: 0.0966 max mem: 3533 +Test: [ 610/2387] eta: 0:06:15 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1256 (0.2544) acc1: 99.4792 (92.3959) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1800 data: 0.0884 max mem: 3533 +Test: [ 620/2387] eta: 0:06:12 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1182 (0.2523) acc1: 99.2188 (92.5100) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1859 data: 0.0918 max mem: 3533 +Test: [ 630/2387] eta: 0:06:09 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1185 (0.2502) acc1: 99.2188 (92.6171) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1887 data: 0.0959 max mem: 3533 +Test: [ 640/2387] eta: 0:06:07 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1204 (0.2481) acc1: 99.2188 (92.7161) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1840 data: 0.0915 max mem: 3533 +Test: [ 650/2387] eta: 0:06:05 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1207 (0.2462) acc1: 98.9583 (92.8151) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2114 data: 0.1186 max mem: 3533 +Test: [ 660/2387] eta: 0:06:02 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1199 (0.2443) acc1: 99.4792 (92.9160) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1989 data: 0.1073 max mem: 3533 +Test: [ 670/2387] eta: 0:06:00 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1196 (0.2424) acc1: 99.4792 (93.0083) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1885 data: 0.0949 max mem: 3534 +Test: [ 680/2387] eta: 0:05:57 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1221 (0.2410) acc1: 99.4792 (93.1072) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1905 data: 0.0969 max mem: 3534 +Test: [ 690/2387] eta: 0:05:54 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1602 (0.2422) acc1: 99.2188 (93.0132) acc5: 100.0000 (100.0000) acc1_neg: 77.083 acc5_neg: 100.000 time: 0.1816 data: 0.0898 max mem: 3534 +Test: [ 700/2387] eta: 0:05:53 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.3773 (0.2442) acc1: 82.5521 (92.8469) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.2157 data: 0.1222 max mem: 3534 +Test: [ 710/2387] eta: 0:05:51 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.3794 (0.2460) acc1: 81.5104 (92.6845) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.2179 data: 0.1222 max mem: 3534 +Test: [ 720/2387] eta: 0:05:48 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.3742 (0.2478) acc1: 81.5104 (92.5397) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.1939 data: 0.1001 max mem: 3534 +Test: [ 730/2387] eta: 0:05:46 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.3797 (0.2498) acc1: 81.5104 (92.3877) acc5: 100.0000 (100.0000) acc1_neg: 80.729 acc5_neg: 100.000 time: 0.1998 data: 0.1060 max mem: 3534 +Test: [ 740/2387] eta: 0:05:44 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.3876 (0.2515) acc1: 80.9896 (92.2448) acc5: 100.0000 (100.0000) acc1_neg: 80.729 acc5_neg: 100.000 time: 0.2052 data: 0.1120 max mem: 3534 +Test: [ 750/2387] eta: 0:05:41 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.3766 (0.2532) acc1: 80.9896 (92.1015) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.1970 data: 0.1058 max mem: 3534 +Test: [ 760/2387] eta: 0:05:38 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.3655 (0.2545) acc1: 82.0312 (91.9887) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.1861 data: 0.0951 max mem: 3534 +Test: [ 770/2387] eta: 0:05:36 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.3035 (0.2548) acc1: 86.9792 (91.9551) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1891 data: 0.0972 max mem: 3534 +Test: [ 780/2387] eta: 0:05:34 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.2621 (0.2542) acc1: 90.6250 (91.9871) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1941 data: 0.1007 max mem: 3534 +Test: [ 790/2387] eta: 0:05:31 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1919 (0.2540) acc1: 94.5312 (91.9732) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.1851 data: 0.0931 max mem: 3534 +Test: [ 800/2387] eta: 0:05:29 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1908 (0.2533) acc1: 94.5312 (92.0136) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1909 data: 0.0982 max mem: 3534 +Test: [ 810/2387] eta: 0:05:27 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.2013 (0.2528) acc1: 94.2708 (92.0273) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.2105 data: 0.1176 max mem: 3534 +Test: [ 820/2387] eta: 0:05:24 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.2537 (0.2532) acc1: 91.9271 (91.9950) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.1946 data: 0.1039 max mem: 3534 +Test: [ 830/2387] eta: 0:05:22 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.3160 (0.2554) acc1: 79.9479 (91.8168) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.1800 data: 0.0869 max mem: 3534 +Test: [ 840/2387] eta: 0:05:20 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.2031 (0.2541) acc1: 86.4583 (91.8949) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1932 data: 0.0997 max mem: 3534 +Test: [ 850/2387] eta: 0:05:17 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1308 (0.2527) acc1: 99.4792 (91.9806) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1827 data: 0.0916 max mem: 3534 +Test: [ 860/2387] eta: 0:05:15 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1237 (0.2512) acc1: 99.7396 (92.0680) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1830 data: 0.0917 max mem: 3534 +Test: [ 870/2387] eta: 0:05:12 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1163 (0.2498) acc1: 99.7396 (92.1543) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1851 data: 0.0940 max mem: 3534 +Test: [ 880/2387] eta: 0:05:09 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1324 (0.2485) acc1: 99.4792 (92.2354) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1743 data: 0.0836 max mem: 3534 +Test: [ 890/2387] eta: 0:05:07 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1445 (0.2477) acc1: 98.9583 (92.3038) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1791 data: 0.0873 max mem: 3534 +Test: [ 900/2387] eta: 0:05:05 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.2089 (0.2543) acc1: 95.5729 (91.8329) acc5: 100.0000 (100.0000) acc1_neg: 38.021 acc5_neg: 100.000 time: 0.1853 data: 0.0922 max mem: 3535 +Test: [ 910/2387] eta: 0:05:02 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.9982 (0.2626) acc1: 38.0208 (91.2347) acc5: 100.0000 (100.0000) acc1_neg: 36.719 acc5_neg: 100.000 time: 0.1803 data: 0.0880 max mem: 3535 +Test: [ 920/2387] eta: 0:05:00 acc1_pos: 98.630 acc5_pos: 100.000 loss: 1.0099 (0.2708) acc1: 36.9792 (90.6493) acc5: 100.0000 (100.0000) acc1_neg: 36.979 acc5_neg: 100.000 time: 0.1822 data: 0.0913 max mem: 3535 +Test: [ 930/2387] eta: 0:04:58 acc1_pos: 98.630 acc5_pos: 100.000 loss: 1.0122 (0.2787) acc1: 37.5000 (90.0854) acc5: 100.0000 (100.0000) acc1_neg: 35.677 acc5_neg: 100.000 time: 0.2103 data: 0.1197 max mem: 3535 +Test: [ 940/2387] eta: 0:04:56 acc1_pos: 98.630 acc5_pos: 100.000 loss: 1.0149 (0.2864) acc1: 37.2396 (89.5219) acc5: 100.0000 (100.0000) acc1_neg: 36.979 acc5_neg: 100.000 time: 0.2002 data: 0.1084 max mem: 3535 +Test: [ 950/2387] eta: 0:04:53 acc1_pos: 98.630 acc5_pos: 100.000 loss: 1.0071 (0.2939) acc1: 36.7188 (88.9702) acc5: 100.0000 (100.0000) acc1_neg: 36.198 acc5_neg: 100.000 time: 0.1650 data: 0.0708 max mem: 3535 +Test: [ 960/2387] eta: 0:04:51 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.9938 (0.3012) acc1: 36.9792 (88.4368) acc5: 100.0000 (100.0000) acc1_neg: 40.885 acc5_neg: 100.000 time: 0.1777 data: 0.0844 max mem: 3535 +Test: [ 970/2387] eta: 0:04:49 acc1_pos: 98.630 acc5_pos: 100.000 loss: 1.0102 (0.3086) acc1: 36.7188 (87.8996) acc5: 100.0000 (100.0000) acc1_neg: 33.854 acc5_neg: 100.000 time: 0.1990 data: 0.1070 max mem: 3535 +Test: [ 980/2387] eta: 0:04:46 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.9913 (0.3106) acc1: 37.5000 (87.7477) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.1897 data: 0.0979 max mem: 3535 +Test: [ 990/2387] eta: 0:04:44 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.2426 (0.3102) acc1: 88.2812 (87.7628) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1799 data: 0.0886 max mem: 3535 +Test: [1000/2387] eta: 0:04:42 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.2310 (0.3094) acc1: 91.6667 (87.8036) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1873 data: 0.0958 max mem: 3535 +Test: [1010/2387] eta: 0:04:39 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.2554 (0.3098) acc1: 90.6250 (87.7669) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.1919 data: 0.0991 max mem: 3535 +Test: [1020/2387] eta: 0:04:37 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.3209 (0.3102) acc1: 85.1562 (87.7163) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1894 data: 0.0953 max mem: 3535 +Test: [1030/2387] eta: 0:04:35 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.2193 (0.3091) acc1: 92.9688 (87.7862) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1791 data: 0.0863 max mem: 3535 +Test: [1040/2387] eta: 0:04:32 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1545 (0.3077) acc1: 97.1354 (87.8737) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1727 data: 0.0812 max mem: 3535 +Test: [1050/2387] eta: 0:04:30 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1545 (0.3063) acc1: 97.3958 (87.9572) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1756 data: 0.0826 max mem: 3535 +Test: [1060/2387] eta: 0:04:28 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1886 (0.3053) acc1: 95.3125 (88.0226) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1759 data: 0.0830 max mem: 3535 +Test: [1070/2387] eta: 0:04:25 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.2063 (0.3050) acc1: 94.5312 (88.0410) acc5: 100.0000 (100.0000) acc1_neg: 71.615 acc5_neg: 100.000 time: 0.1761 data: 0.0845 max mem: 3535 +Test: [1080/2387] eta: 0:04:23 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.3670 (0.3063) acc1: 79.4271 (87.9329) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.1777 data: 0.0863 max mem: 3535 +Test: [1090/2387] eta: 0:04:21 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.3421 (0.3062) acc1: 80.9896 (87.9244) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.1908 data: 0.0987 max mem: 3535 +Test: [1100/2387] eta: 0:04:19 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.2581 (0.3057) acc1: 88.5417 (87.9293) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1885 data: 0.0963 max mem: 3535 +Test: [1110/2387] eta: 0:04:17 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.2701 (0.3058) acc1: 86.9792 (87.9001) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1873 data: 0.0959 max mem: 3535 +Test: [1120/2387] eta: 0:04:14 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.2808 (0.3054) acc1: 85.9375 (87.9049) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1862 data: 0.0947 max mem: 3536 +Test: [1130/2387] eta: 0:04:12 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.2808 (0.3053) acc1: 87.7604 (87.8986) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1781 data: 0.0865 max mem: 3536 +Test: [1140/2387] eta: 0:04:10 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.3674 (0.3063) acc1: 80.4688 (87.8072) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.1793 data: 0.0878 max mem: 3536 +Test: [1150/2387] eta: 0:04:08 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.3861 (0.3058) acc1: 79.4271 (87.8274) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1813 data: 0.0884 max mem: 3536 +Test: [1160/2387] eta: 0:04:05 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1004 (0.3040) acc1: 100.0000 (87.9320) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1791 data: 0.0866 max mem: 3536 +Test: [1170/2387] eta: 0:04:03 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.0980 (0.3023) acc1: 100.0000 (88.0348) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1795 data: 0.0865 max mem: 3536 +Test: [1180/2387] eta: 0:04:01 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.0994 (0.3005) acc1: 100.0000 (88.1335) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1835 data: 0.0904 max mem: 3536 +Test: [1190/2387] eta: 0:03:59 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1004 (0.2989) acc1: 100.0000 (88.2325) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1947 data: 0.1016 max mem: 3536 +Test: [1200/2387] eta: 0:03:57 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.0985 (0.2973) acc1: 100.0000 (88.3294) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2016 data: 0.1088 max mem: 3536 +Test: [1210/2387] eta: 0:03:55 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.0985 (0.2957) acc1: 100.0000 (88.4230) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1984 data: 0.1075 max mem: 3536 +Test: [1220/2387] eta: 0:03:53 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1023 (0.2941) acc1: 99.4792 (88.5061) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1988 data: 0.1070 max mem: 3536 +Test: [1230/2387] eta: 0:03:51 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.0879 (0.2924) acc1: 99.2188 (88.5948) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2020 data: 0.1099 max mem: 3536 +Test: [1240/2387] eta: 0:03:49 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.0776 (0.2907) acc1: 99.7396 (88.6831) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1992 data: 0.1072 max mem: 3536 +Test: [1250/2387] eta: 0:03:47 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.0802 (0.2891) acc1: 99.7396 (88.7702) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1949 data: 0.1032 max mem: 3536 +Test: [1260/2387] eta: 0:03:45 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.0853 (0.2875) acc1: 99.4792 (88.8545) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1926 data: 0.1014 max mem: 3536 +Test: [1270/2387] eta: 0:03:43 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.0998 (0.2862) acc1: 99.2188 (88.9281) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1843 data: 0.0933 max mem: 3536 +Test: [1280/2387] eta: 0:03:41 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1302 (0.2851) acc1: 98.1771 (88.9962) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1986 data: 0.1075 max mem: 3536 +Test: [1290/2387] eta: 0:03:39 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1540 (0.2843) acc1: 97.6562 (89.0520) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2055 data: 0.1143 max mem: 3536 +Test: [1300/2387] eta: 0:03:37 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1670 (0.2834) acc1: 97.1354 (89.1081) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2035 data: 0.1116 max mem: 3536 +Test: [1310/2387] eta: 0:03:35 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1648 (0.2826) acc1: 97.3958 (89.1614) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2016 data: 0.1094 max mem: 3536 +Test: [1320/2387] eta: 0:03:33 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1848 (0.2820) acc1: 96.8750 (89.2151) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1819 data: 0.0912 max mem: 3536 +Test: [1330/2387] eta: 0:03:30 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.2176 (0.2821) acc1: 92.1875 (89.1979) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.1711 data: 0.0802 max mem: 3536 +Test: [1340/2387] eta: 0:03:28 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.3418 (0.2831) acc1: 84.1146 (89.1307) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1793 data: 0.0876 max mem: 3536 +Test: [1350/2387] eta: 0:03:26 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.0736 (0.2815) acc1: 99.4792 (89.2107) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1820 data: 0.0907 max mem: 3537 +Test: [1360/2387] eta: 0:03:24 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.0639 (0.2799) acc1: 100.0000 (89.2900) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1736 data: 0.0824 max mem: 3537 +Test: [1370/2387] eta: 0:03:22 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.0632 (0.2783) acc1: 100.0000 (89.3681) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1778 data: 0.0847 max mem: 3537 +Test: [1380/2387] eta: 0:03:20 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.0645 (0.2768) acc1: 100.0000 (89.4451) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1958 data: 0.1030 max mem: 3537 +Test: [1390/2387] eta: 0:03:18 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.0645 (0.2752) acc1: 100.0000 (89.5210) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1855 data: 0.0949 max mem: 3537 +Test: [1400/2387] eta: 0:03:16 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.0646 (0.2737) acc1: 100.0000 (89.5954) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1962 data: 0.1053 max mem: 3537 +Test: [1410/2387] eta: 0:03:14 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.0709 (0.2734) acc1: 100.0000 (89.6051) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1963 data: 0.1050 max mem: 3537 +Test: [1420/2387] eta: 0:03:12 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.2614 (0.2736) acc1: 89.0625 (89.5844) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1790 data: 0.0852 max mem: 3537 +Test: [1430/2387] eta: 0:03:10 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.2430 (0.2735) acc1: 89.8438 (89.5799) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1979 data: 0.1022 max mem: 3537 +Test: [1440/2387] eta: 0:03:08 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.2699 (0.2738) acc1: 86.7188 (89.5427) acc5: 100.0000 (100.0000) acc1_neg: 72.135 acc5_neg: 100.000 time: 0.2040 data: 0.1109 max mem: 3537 +Test: [1450/2387] eta: 0:03:06 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.3440 (0.2745) acc1: 82.2917 (89.4945) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1969 data: 0.1055 max mem: 3537 +Test: [1460/2387] eta: 0:03:04 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1752 (0.2737) acc1: 93.4896 (89.5386) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1913 data: 0.1002 max mem: 3537 +Test: [1470/2387] eta: 0:03:02 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1294 (0.2727) acc1: 97.1354 (89.5986) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1897 data: 0.0995 max mem: 3537 +Test: [1480/2387] eta: 0:03:00 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1281 (0.2718) acc1: 97.9167 (89.6482) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1965 data: 0.1045 max mem: 3537 +Test: [1490/2387] eta: 0:02:58 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1286 (0.2708) acc1: 96.8750 (89.7030) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1937 data: 0.1007 max mem: 3537 +Test: [1500/2387] eta: 0:02:55 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1323 (0.2699) acc1: 97.6562 (89.7554) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1869 data: 0.0932 max mem: 3537 +Test: [1510/2387] eta: 0:02:53 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1415 (0.2691) acc1: 98.4375 (89.8181) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1921 data: 0.0989 max mem: 3537 +Test: [1520/2387] eta: 0:02:51 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1380 (0.2682) acc1: 99.4792 (89.8787) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1917 data: 0.0996 max mem: 3537 +Test: [1530/2387] eta: 0:02:49 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1235 (0.2672) acc1: 99.4792 (89.9433) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1821 data: 0.0896 max mem: 3537 +Test: [1540/2387] eta: 0:02:47 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1094 (0.2662) acc1: 99.7396 (90.0055) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1928 data: 0.1014 max mem: 3537 +Test: [1550/2387] eta: 0:02:45 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1094 (0.2652) acc1: 99.7396 (90.0679) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1922 data: 0.0991 max mem: 3537 +Test: [1560/2387] eta: 0:02:43 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1127 (0.2643) acc1: 99.7396 (90.1289) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1858 data: 0.0908 max mem: 3537 +Test: [1570/2387] eta: 0:02:41 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1254 (0.2635) acc1: 99.4792 (90.1871) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1860 data: 0.0932 max mem: 3537 +Test: [1580/2387] eta: 0:02:39 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1585 (0.2632) acc1: 97.9167 (90.2078) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1860 data: 0.0948 max mem: 3538 +Test: [1590/2387] eta: 0:02:37 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1970 (0.2628) acc1: 93.7500 (90.2359) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1943 data: 0.1033 max mem: 3538 +Test: [1600/2387] eta: 0:02:35 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1903 (0.2625) acc1: 95.5729 (90.2385) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1910 data: 0.1006 max mem: 3538 +Test: [1610/2387] eta: 0:02:33 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1782 (0.2620) acc1: 95.5729 (90.2728) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1884 data: 0.0976 max mem: 3538 +Test: [1620/2387] eta: 0:02:31 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.2103 (0.2619) acc1: 91.9271 (90.2725) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1845 data: 0.0928 max mem: 3538 +Test: [1630/2387] eta: 0:02:29 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.2664 (0.2621) acc1: 87.5000 (90.2472) acc5: 100.0000 (100.0000) acc1_neg: 80.729 acc5_neg: 100.000 time: 0.1889 data: 0.0973 max mem: 3538 +Test: [1640/2387] eta: 0:02:27 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.3476 (0.2629) acc1: 83.8542 (90.1845) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1885 data: 0.0970 max mem: 3538 +Test: [1650/2387] eta: 0:02:25 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.3791 (0.2636) acc1: 80.2083 (90.1240) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.1778 data: 0.0857 max mem: 3538 +Test: [1660/2387] eta: 0:02:23 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.3791 (0.2644) acc1: 80.4688 (90.0589) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.1897 data: 0.0983 max mem: 3538 +Test: [1670/2387] eta: 0:02:21 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.3765 (0.2650) acc1: 80.9896 (90.0142) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.1999 data: 0.1086 max mem: 3538 +Test: [1680/2387] eta: 0:02:19 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.3667 (0.2656) acc1: 81.5104 (89.9674) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.2004 data: 0.1087 max mem: 3538 +Test: [1690/2387] eta: 0:02:17 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.3622 (0.2664) acc1: 78.9062 (89.9061) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1744 data: 0.0833 max mem: 3538 +Test: [1700/2387] eta: 0:02:15 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.3839 (0.2665) acc1: 78.9062 (89.8805) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1694 data: 0.0779 max mem: 3538 +Test: [1710/2387] eta: 0:02:13 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1081 (0.2655) acc1: 97.9167 (89.9320) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1856 data: 0.0944 max mem: 3538 +Test: [1720/2387] eta: 0:02:11 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.0943 (0.2645) acc1: 98.6979 (89.9842) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1809 data: 0.0893 max mem: 3538 +Test: [1730/2387] eta: 0:02:09 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.0918 (0.2635) acc1: 99.2188 (90.0342) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1972 data: 0.1031 max mem: 3538 +Test: [1740/2387] eta: 0:02:07 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.0918 (0.2625) acc1: 98.9583 (90.0844) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2051 data: 0.1114 max mem: 3538 +Test: [1750/2387] eta: 0:02:05 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.0953 (0.2616) acc1: 98.9583 (90.1336) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1950 data: 0.1016 max mem: 3538 +Test: [1760/2387] eta: 0:02:03 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1092 (0.2609) acc1: 97.9167 (90.1657) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.2104 data: 0.1143 max mem: 3538 +Test: [1770/2387] eta: 0:02:01 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1386 (0.2602) acc1: 96.8750 (90.2036) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.2162 data: 0.1214 max mem: 3538 +Test: [1780/2387] eta: 0:01:59 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1200 (0.2595) acc1: 98.1771 (90.2466) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1972 data: 0.1039 max mem: 3538 +Test: [1790/2387] eta: 0:01:57 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1065 (0.2586) acc1: 98.6979 (90.2944) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1850 data: 0.0915 max mem: 3538 +Test: [1800/2387] eta: 0:01:55 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.0954 (0.2577) acc1: 98.6979 (90.3409) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1849 data: 0.0932 max mem: 3538 +Test: [1810/2387] eta: 0:01:53 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1023 (0.2569) acc1: 98.6979 (90.3863) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1788 data: 0.0874 max mem: 3539 +Test: [1820/2387] eta: 0:01:51 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1055 (0.2562) acc1: 97.9167 (90.4226) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1670 data: 0.0755 max mem: 3539 +Test: [1830/2387] eta: 0:01:49 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1365 (0.2556) acc1: 97.1354 (90.4573) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1647 data: 0.0732 max mem: 3539 +Test: [1840/2387] eta: 0:01:47 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1365 (0.2550) acc1: 99.7396 (90.5086) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1643 data: 0.0712 max mem: 3539 +Test: [1850/2387] eta: 0:01:45 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1352 (0.2543) acc1: 100.0000 (90.5589) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1718 data: 0.0790 max mem: 3539 +Test: [1860/2387] eta: 0:01:43 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1362 (0.2537) acc1: 99.7396 (90.6079) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1767 data: 0.0861 max mem: 3539 +Test: [1870/2387] eta: 0:01:41 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1350 (0.2531) acc1: 100.0000 (90.6578) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1778 data: 0.0871 max mem: 3539 +Test: [1880/2387] eta: 0:01:39 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1337 (0.2524) acc1: 100.0000 (90.7071) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1722 data: 0.0814 max mem: 3539 +Test: [1890/2387] eta: 0:01:37 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1356 (0.2518) acc1: 100.0000 (90.7553) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1747 data: 0.0840 max mem: 3539 +Test: [1900/2387] eta: 0:01:35 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1372 (0.2512) acc1: 99.7396 (90.8024) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1827 data: 0.0908 max mem: 3539 +Test: [1910/2387] eta: 0:01:33 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1915 (0.2510) acc1: 99.2188 (90.8396) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1807 data: 0.0872 max mem: 3539 +Test: [1920/2387] eta: 0:01:31 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.2065 (0.2514) acc1: 93.7500 (90.8056) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.1739 data: 0.0807 max mem: 3539 +Test: [1930/2387] eta: 0:01:29 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.2052 (0.2511) acc1: 92.7083 (90.8242) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1739 data: 0.0823 max mem: 3539 +Test: [1940/2387] eta: 0:01:27 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1983 (0.2511) acc1: 93.7500 (90.8072) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1687 data: 0.0777 max mem: 3539 +Test: [1950/2387] eta: 0:01:25 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.2282 (0.2510) acc1: 93.2292 (90.8144) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1570 data: 0.0661 max mem: 3539 +Test: [1960/2387] eta: 0:01:23 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.2577 (0.2515) acc1: 89.0625 (90.7682) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.1587 data: 0.0680 max mem: 3539 +Test: [1970/2387] eta: 0:01:21 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.3829 (0.2527) acc1: 78.6458 (90.6749) acc5: 100.0000 (100.0000) acc1_neg: 65.625 acc5_neg: 100.000 time: 0.1632 data: 0.0725 max mem: 3539 +Test: [1980/2387] eta: 0:01:19 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.3864 (0.2529) acc1: 76.8229 (90.6550) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1669 data: 0.0754 max mem: 3539 +Test: [1990/2387] eta: 0:01:17 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1190 (0.2522) acc1: 99.7396 (90.7006) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1845 data: 0.0926 max mem: 3539 +Test: [2000/2387] eta: 0:01:15 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1170 (0.2515) acc1: 99.7396 (90.7467) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1834 data: 0.0917 max mem: 3539 +Test: [2010/2387] eta: 0:01:13 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1174 (0.2509) acc1: 100.0000 (90.7922) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1814 data: 0.0899 max mem: 3539 +Test: [2020/2387] eta: 0:01:11 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1178 (0.2502) acc1: 100.0000 (90.8366) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1981 data: 0.1061 max mem: 3539 +Test: [2030/2387] eta: 0:01:09 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1194 (0.2496) acc1: 100.0000 (90.8809) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2052 data: 0.1136 max mem: 3540 +Test: [2040/2387] eta: 0:01:07 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1175 (0.2489) acc1: 100.0000 (90.9252) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2072 data: 0.1165 max mem: 3540 +Test: [2050/2387] eta: 0:01:05 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1175 (0.2483) acc1: 100.0000 (90.9686) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2195 data: 0.1282 max mem: 3540 +Test: [2060/2387] eta: 0:01:03 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1372 (0.2478) acc1: 99.7396 (91.0108) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2005 data: 0.1092 max mem: 3540 +Test: [2070/2387] eta: 0:01:01 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1399 (0.2473) acc1: 99.7396 (91.0534) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0990 max mem: 3540 +Test: [2080/2387] eta: 0:00:59 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1384 (0.2468) acc1: 99.7396 (91.0950) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1947 data: 0.1039 max mem: 3540 +Test: [2090/2387] eta: 0:00:57 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1383 (0.2463) acc1: 99.7396 (91.1369) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1898 data: 0.0992 max mem: 3540 +Test: [2100/2387] eta: 0:00:55 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1392 (0.2458) acc1: 99.7396 (91.1781) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1956 data: 0.1038 max mem: 3540 +Test: [2110/2387] eta: 0:00:53 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1398 (0.2453) acc1: 99.7396 (91.2190) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1946 data: 0.1026 max mem: 3540 +Test: [2120/2387] eta: 0:00:51 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1476 (0.2453) acc1: 99.2188 (91.2222) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1902 data: 0.0994 max mem: 3540 +Test: [2130/2387] eta: 0:00:50 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.2319 (0.2452) acc1: 93.4896 (91.2352) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1889 data: 0.0983 max mem: 3540 +Test: [2140/2387] eta: 0:00:48 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.2252 (0.2453) acc1: 93.4896 (91.2237) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1903 data: 0.0998 max mem: 3540 +Test: [2150/2387] eta: 0:00:46 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.2204 (0.2452) acc1: 93.7500 (91.2352) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1899 data: 0.0984 max mem: 3540 +Test: [2160/2387] eta: 0:00:44 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.2333 (0.2453) acc1: 92.4479 (91.2327) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1971 data: 0.1055 max mem: 3540 +Test: [2170/2387] eta: 0:00:42 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.2961 (0.2460) acc1: 84.3750 (91.1794) acc5: 100.0000 (100.0000) acc1_neg: 73.698 acc5_neg: 100.000 time: 0.1827 data: 0.0896 max mem: 3540 +Test: [2180/2387] eta: 0:00:40 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.4443 (0.2478) acc1: 74.2188 (91.0547) acc5: 100.0000 (100.0000) acc1_neg: 49.479 acc5_neg: 100.000 time: 0.1877 data: 0.0945 max mem: 3540 +Test: [2190/2387] eta: 0:00:38 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.4190 (0.2484) acc1: 78.6458 (91.0225) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1884 data: 0.0959 max mem: 3540 +Test: [2200/2387] eta: 0:00:36 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1786 (0.2480) acc1: 96.3542 (91.0538) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1933 data: 0.1013 max mem: 3540 +Test: [2210/2387] eta: 0:00:34 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1327 (0.2475) acc1: 98.9583 (91.0878) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2000 data: 0.1089 max mem: 3540 +Test: [2220/2387] eta: 0:00:32 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1303 (0.2470) acc1: 99.2188 (91.1172) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1804 data: 0.0873 max mem: 3540 +Test: [2230/2387] eta: 0:00:30 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1317 (0.2464) acc1: 99.2188 (91.1488) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1840 data: 0.0913 max mem: 3540 +Test: [2240/2387] eta: 0:00:28 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1317 (0.2460) acc1: 99.2188 (91.1813) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1881 data: 0.0937 max mem: 3540 +Test: [2250/2387] eta: 0:00:26 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1469 (0.2456) acc1: 99.2188 (91.2154) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2057 data: 0.1110 max mem: 3540 +Test: [2260/2387] eta: 0:00:24 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1640 (0.2453) acc1: 98.4375 (91.2403) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2118 data: 0.1208 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1847 (0.2450) acc1: 96.6146 (91.2660) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2252 data: 0.1322 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1855 (0.2450) acc1: 96.6146 (91.2707) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2209 data: 0.1273 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1855 (0.2447) acc1: 96.3542 (91.2962) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2115 data: 0.1185 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.1953 (0.2445) acc1: 95.8333 (91.3098) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2123 data: 0.1190 max mem: 3541 +Test: [2310/2387] eta: 0:00:15 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.2318 (0.2448) acc1: 91.9271 (91.2980) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.1977 data: 0.1048 max mem: 3541 +Test: [2320/2387] eta: 0:00:13 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.3473 (0.2454) acc1: 83.8542 (91.2561) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.1922 data: 0.0994 max mem: 3541 +Test: [2330/2387] eta: 0:00:11 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.3490 (0.2458) acc1: 82.5521 (91.2315) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.1706 data: 0.0794 max mem: 3541 +Test: [2340/2387] eta: 0:00:09 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.2817 (0.2459) acc1: 87.7604 (91.2238) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1460 data: 0.0556 max mem: 3541 +Test: [2350/2387] eta: 0:00:07 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.2625 (0.2462) acc1: 89.0625 (91.1906) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1295 data: 0.0390 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.3062 (0.2466) acc1: 86.4583 (91.1662) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1228 data: 0.0325 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.3262 (0.2471) acc1: 84.1146 (91.1205) acc5: 100.0000 (100.0000) acc1_neg: 70.573 acc5_neg: 100.000 time: 0.1210 data: 0.0311 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.4382 (0.2482) acc1: 74.7396 (91.0427) acc5: 100.0000 (100.0000) acc1_neg: 75.781 acc5_neg: 100.000 time: 0.1173 data: 0.0276 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 98.630 acc5_pos: 100.000 loss: 0.4465 (0.2486) acc1: 73.4375 (91.0176) acc5: 100.0000 (100.0000) acc1_neg: 88.372 acc5_neg: 100.000 time: 0.1099 data: 0.0229 max mem: 3541 +Test: Total time: 0:07:41 (0.1932 s / it) +* Overall Acc@1 91.018 Acc@5 100.000 + target == real Acc@1 77.850 Acc@5 100.000 + target == fake Acc@1 91.942 Acc@5 100.000 + loss 0.249 + PR-AUC 0.9958 ROC-AUC 0.9429 EER 11.83% (thr=0.6157) +Accuracy of the network on the 916267 test images: 91.0% +[rank0]:[W726 06:06:34.878647373 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 06:06:55.270000 1202818 torch/distributed/run.py:766] +W0726 06:06:55.270000 1202818 torch/distributed/run.py:766] ***************************************** +W0726 06:06:55.270000 1202818 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:06:55.270000 1202818 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:07:22.619930287 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:07:22.872189564 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 06:07:22.228754395 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:07:22.415050230 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:07:23.580511585 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:07:23.708934685 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:07:23.821700185 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:07:23.994840398 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n23_seed9_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed9_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:10:22 acc1_pos: 78.125 acc5_pos: 100.000 loss: 0.4971 (0.4971) acc1: 78.1250 (78.1250) acc5: 100.0000 (100.0000) time: 8.1669 data: 6.3184 max mem: 3531 +Test: [ 10/517] eta: 0:06:58 acc1_pos: 52.083 acc5_pos: 100.000 loss: 0.6286 (0.6279) acc1: 67.1875 (68.4422) acc5: 100.0000 (100.0000) time: 0.8247 data: 0.5748 max mem: 3531 +Test: [ 20/517] eta: 0:03:57 acc1_pos: 64.583 acc5_pos: 100.000 loss: 0.6469 (0.6702) acc1: 64.5833 (65.4762) acc5: 100.0000 (100.0000) time: 0.0944 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:03:30 acc1_pos: 44.531 acc5_pos: 100.000 loss: 0.6855 (0.6785) acc1: 60.9375 (63.2897) acc5: 100.0000 (100.0000) time: 0.2151 data: 0.1175 max mem: 3531 +Test: [ 40/517] eta: 0:03:09 acc1_pos: 70.312 acc5_pos: 100.000 loss: 0.6612 (0.6730) acc1: 61.9792 (63.6433) acc5: 100.0000 (100.0000) time: 0.3106 data: 0.2161 max mem: 3531 +Test: [ 50/517] eta: 0:02:54 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.6296 (0.6684) acc1: 64.8438 (63.6132) acc5: 100.0000 (100.0000) acc1_neg: 43.750 acc5_neg: 100.000 time: 0.2839 data: 0.1888 max mem: 3531 +Test: [ 60/517] eta: 0:02:44 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.6397 (0.6741) acc1: 60.9375 (62.0774) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.2814 data: 0.1836 max mem: 3531 +Test: [ 70/517] eta: 0:02:34 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.5867 (0.6594) acc1: 62.5000 (63.1749) acc5: 100.0000 (100.0000) acc1_neg: 72.135 acc5_neg: 100.000 time: 0.2726 data: 0.1791 max mem: 3531 +Test: [ 80/517] eta: 0:02:28 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.5626 (0.6524) acc1: 68.2292 (63.5417) acc5: 100.0000 (100.0000) acc1_neg: 40.104 acc5_neg: 100.000 time: 0.2777 data: 0.1861 max mem: 3531 +Test: [ 90/517] eta: 0:02:21 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.5324 (0.6358) acc1: 72.3958 (65.2101) acc5: 100.0000 (100.0000) acc1_neg: 72.135 acc5_neg: 100.000 time: 0.2789 data: 0.1870 max mem: 3531 +Test: [100/517] eta: 0:02:14 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.5261 (0.6265) acc1: 72.6562 (65.8055) acc5: 100.0000 (100.0000) acc1_neg: 69.271 acc5_neg: 100.000 time: 0.2569 data: 0.1651 max mem: 3531 +Test: [110/517] eta: 0:02:08 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.5261 (0.6141) acc1: 72.6562 (66.6409) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.2486 data: 0.1568 max mem: 3531 +Test: [120/517] eta: 0:02:03 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.5278 (0.6086) acc1: 72.6562 (67.1272) acc5: 100.0000 (100.0000) acc1_neg: 69.792 acc5_neg: 100.000 time: 0.2497 data: 0.1578 max mem: 3531 +Test: [130/517] eta: 0:01:58 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.5255 (0.6008) acc1: 75.7812 (68.0463) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.2596 data: 0.1666 max mem: 3531 +Test: [140/517] eta: 0:01:54 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.5218 (0.5948) acc1: 79.4271 (68.8017) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.2541 data: 0.1612 max mem: 3531 +Test: [150/517] eta: 0:01:49 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.5044 (0.5865) acc1: 79.4271 (69.8262) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.2347 data: 0.1413 max mem: 3531 +Test: [160/517] eta: 0:01:45 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.5137 (0.5819) acc1: 75.5208 (70.0116) acc5: 100.0000 (100.0000) acc1_neg: 79.427 acc5_neg: 100.000 time: 0.2539 data: 0.1598 max mem: 3531 +Test: [170/517] eta: 0:01:42 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.4749 (0.5746) acc1: 74.2188 (70.4770) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.2699 data: 0.1744 max mem: 3531 +Test: [180/517] eta: 0:01:38 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.4474 (0.5703) acc1: 80.2083 (70.8736) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.2499 data: 0.1518 max mem: 3531 +Test: [190/517] eta: 0:01:34 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.3433 (0.5579) acc1: 90.1042 (72.0154) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.2349 data: 0.1392 max mem: 3531 +Test: [200/517] eta: 0:01:30 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.3661 (0.5500) acc1: 90.1042 (72.8532) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.2399 data: 0.1462 max mem: 3531 +Test: [210/517] eta: 0:01:27 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.4413 (0.5506) acc1: 85.4167 (72.7982) acc5: 100.0000 (100.0000) acc1_neg: 71.094 acc5_neg: 100.000 time: 0.2549 data: 0.1608 max mem: 3532 +Test: [220/517] eta: 0:01:24 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.5617 (0.5507) acc1: 70.0521 (72.7529) acc5: 100.0000 (100.0000) acc1_neg: 63.021 acc5_neg: 100.000 time: 0.2499 data: 0.1566 max mem: 3532 +Test: [230/517] eta: 0:01:21 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.5987 (0.5560) acc1: 63.8021 (72.0351) acc5: 100.0000 (100.0000) acc1_neg: 70.312 acc5_neg: 100.000 time: 0.2549 data: 0.1605 max mem: 3532 +Test: [240/517] eta: 0:01:17 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.6191 (0.5545) acc1: 61.4583 (72.1203) acc5: 100.0000 (100.0000) acc1_neg: 42.708 acc5_neg: 100.000 time: 0.2649 data: 0.1711 max mem: 3532 +Test: [250/517] eta: 0:01:14 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.5269 (0.5550) acc1: 74.2188 (72.1551) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.2599 data: 0.1672 max mem: 3532 +Test: [260/517] eta: 0:01:12 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.5433 (0.5549) acc1: 74.2188 (72.2073) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.2648 data: 0.1708 max mem: 3532 +Test: [270/517] eta: 0:01:09 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.4968 (0.5526) acc1: 79.9479 (72.5457) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.2649 data: 0.1698 max mem: 3532 +Test: [280/517] eta: 0:01:06 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.4765 (0.5496) acc1: 82.2917 (72.9102) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.2599 data: 0.1634 max mem: 3532 +Test: [290/517] eta: 0:01:02 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.4670 (0.5475) acc1: 83.0729 (73.2290) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.2499 data: 0.1531 max mem: 3532 +Test: [300/517] eta: 0:00:59 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.4699 (0.5451) acc1: 83.0729 (73.4314) acc5: 100.0000 (100.0000) acc1_neg: 57.812 acc5_neg: 100.000 time: 0.2349 data: 0.1401 max mem: 3532 +Test: [310/517] eta: 0:00:56 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.5612 (0.5485) acc1: 63.0208 (72.9058) acc5: 100.0000 (100.0000) acc1_neg: 52.344 acc5_neg: 100.000 time: 0.2399 data: 0.1445 max mem: 3532 +Test: [320/517] eta: 0:00:53 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.6092 (0.5479) acc1: 61.4583 (72.8704) acc5: 100.0000 (100.0000) acc1_neg: 79.427 acc5_neg: 100.000 time: 0.2449 data: 0.1484 max mem: 3532 +Test: [330/517] eta: 0:00:51 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.4237 (0.5444) acc1: 80.4688 (73.1826) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.2399 data: 0.1443 max mem: 3532 +Test: [340/517] eta: 0:00:48 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.4162 (0.5427) acc1: 84.8958 (73.3443) acc5: 100.0000 (100.0000) acc1_neg: 52.865 acc5_neg: 100.000 time: 0.2549 data: 0.1587 max mem: 3532 +Test: [350/517] eta: 0:00:45 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.5354 (0.5445) acc1: 70.5729 (73.1118) acc5: 100.0000 (100.0000) acc1_neg: 58.073 acc5_neg: 100.000 time: 0.2599 data: 0.1626 max mem: 3532 +Test: [360/517] eta: 0:00:42 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.5723 (0.5451) acc1: 70.3125 (73.0588) acc5: 100.0000 (100.0000) acc1_neg: 74.219 acc5_neg: 100.000 time: 0.2549 data: 0.1570 max mem: 3532 +Test: [370/517] eta: 0:00:39 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.4776 (0.5422) acc1: 81.5104 (73.4417) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2499 data: 0.1528 max mem: 3532 +Test: [380/517] eta: 0:00:37 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.4357 (0.5393) acc1: 86.7188 (73.7772) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.2407 data: 0.1434 max mem: 3532 +Test: [390/517] eta: 0:00:34 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.4396 (0.5371) acc1: 84.1146 (73.9770) acc5: 100.0000 (100.0000) acc1_neg: 70.573 acc5_neg: 100.000 time: 0.2449 data: 0.1477 max mem: 3532 +Test: [400/517] eta: 0:00:31 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.4461 (0.5355) acc1: 81.2500 (74.0142) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.2441 data: 0.1491 max mem: 3532 +Test: [410/517] eta: 0:00:28 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.5052 (0.5361) acc1: 68.7500 (73.8893) acc5: 100.0000 (100.0000) acc1_neg: 54.427 acc5_neg: 100.000 time: 0.2399 data: 0.1445 max mem: 3532 +Test: [420/517] eta: 0:00:25 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.6385 (0.5391) acc1: 60.4167 (73.4734) acc5: 100.0000 (100.0000) acc1_neg: 68.490 acc5_neg: 100.000 time: 0.2449 data: 0.1470 max mem: 3532 +Test: [430/517] eta: 0:00:23 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.6419 (0.5403) acc1: 59.1146 (73.2955) acc5: 100.0000 (100.0000) acc1_neg: 59.375 acc5_neg: 100.000 time: 0.2499 data: 0.1543 max mem: 3532 +Test: [440/517] eta: 0:00:20 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.5942 (0.5412) acc1: 64.8438 (73.1918) acc5: 100.0000 (100.0000) acc1_neg: 64.844 acc5_neg: 100.000 time: 0.2449 data: 0.1496 max mem: 3533 +Test: [450/517] eta: 0:00:17 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.5510 (0.5405) acc1: 73.1771 (73.3428) acc5: 100.0000 (100.0000) acc1_neg: 68.750 acc5_neg: 100.000 time: 0.2499 data: 0.1523 max mem: 3533 +Test: [460/517] eta: 0:00:15 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.5592 (0.5418) acc1: 72.1354 (73.2166) acc5: 100.0000 (100.0000) acc1_neg: 69.271 acc5_neg: 100.000 time: 0.2699 data: 0.1742 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.5592 (0.5415) acc1: 71.3542 (73.2539) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.2549 data: 0.1590 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.4950 (0.5405) acc1: 76.8229 (73.4294) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2349 data: 0.1397 max mem: 3533 +Test: [490/517] eta: 0:00:07 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.4915 (0.5403) acc1: 75.7812 (73.4624) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.2399 data: 0.1467 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.5616 (0.5412) acc1: 69.7917 (73.3533) acc5: 100.0000 (100.0000) acc1_neg: 56.771 acc5_neg: 100.000 time: 0.2429 data: 0.1498 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.5503 (0.5414) acc1: 71.8750 (73.3437) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.2092 data: 0.1178 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.5331 (0.5414) acc1: 72.9167 (73.3726) acc5: 100.0000 (100.0000) acc1_neg: 71.025 acc5_neg: 100.000 time: 0.1715 data: 0.0802 max mem: 3533 +Test: Total time: 0:02:15 (0.2628 s / it) +* Overall Acc@1 73.373 Acc@5 100.000 + target == real Acc@1 64.247 Acc@5 100.000 + target == fake Acc@1 74.319 Acc@5 100.000 + loss 0.541 + PR-AUC 0.9579 ROC-AUC 0.7533 EER 30.95% (thr=0.5298) +Accuracy of the network on the 198427 test images: 73.4% +[rank0]:[W726 06:09:51.687594794 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 06:10:11.857000 1203391 torch/distributed/run.py:766] +W0726 06:10:11.857000 1203391 torch/distributed/run.py:766] ***************************************** +W0726 06:10:11.857000 1203391 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:10:11.857000 1203391 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:10:42.673108827 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:10:42.893703604 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 06:10:42.060586061 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:10:42.128937564 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:10:43.501096626 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:10:43.583528499 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:10:43.599181824 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:10:43.618726075 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n23_seed9_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed9_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:37:18 acc1_pos: 5.469 acc5_pos: 100.000 loss: 1.5165 (1.5165) acc1: 5.4688 (5.4688) acc5: 100.0000 (100.0000) time: 5.7550 data: 4.0235 max mem: 3531 +Test: [ 10/389] eta: 0:03:49 acc1_pos: 19.531 acc5_pos: 100.000 loss: 1.3916 (1.4115) acc1: 7.5521 (9.5407) acc5: 100.0000 (100.0000) time: 0.6051 data: 0.3662 max mem: 3531 +Test: [ 20/389] eta: 0:02:14 acc1_pos: 0.260 acc5_pos: 100.000 loss: 1.4026 (1.5143) acc1: 3.6458 (6.8700) acc5: 100.0000 (100.0000) time: 0.0947 data: 0.0004 max mem: 3531 +Test: [ 30/389] eta: 0:01:44 acc1_pos: 1.042 acc5_pos: 100.000 loss: 1.8592 (1.6104) acc1: 0.2604 (4.9059) acc5: 100.0000 (100.0000) time: 0.1204 data: 0.0218 max mem: 3531 +Test: [ 40/389] eta: 0:01:37 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.7749 (1.6162) acc1: 0.2604 (4.1921) acc5: 100.0000 (100.0000) time: 0.1907 data: 0.0957 max mem: 3531 +Test: [ 50/389] eta: 0:01:28 acc1_pos: 26.302 acc5_pos: 100.000 loss: 1.5263 (1.5850) acc1: 0.5208 (5.4279) acc5: 100.0000 (100.0000) time: 0.2141 data: 0.1197 max mem: 3531 +Test: [ 60/389] eta: 0:01:21 acc1_pos: 0.260 acc5_pos: 100.000 loss: 1.4867 (1.5550) acc1: 1.5625 (6.3994) acc5: 100.0000 (100.0000) time: 0.1842 data: 0.0874 max mem: 3531 +Test: [ 70/389] eta: 0:01:15 acc1_pos: 10.938 acc5_pos: 100.000 loss: 1.5448 (1.5537) acc1: 1.8229 (6.3124) acc5: 100.0000 (100.0000) time: 0.1749 data: 0.0777 max mem: 3531 +Test: [ 80/389] eta: 0:01:09 acc1_pos: 39.583 acc5_pos: 100.000 loss: 1.5889 (1.5306) acc1: 1.8229 (8.4073) acc5: 100.0000 (100.0000) time: 0.1600 data: 0.0652 max mem: 3531 +Test: [ 90/389] eta: 0:01:04 acc1_pos: 19.271 acc5_pos: 100.000 loss: 1.0974 (1.4769) acc1: 19.2708 (10.7372) acc5: 100.0000 (100.0000) time: 0.1449 data: 0.0497 max mem: 3531 +Test: [100/389] eta: 0:01:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.3073 (1.4913) acc1: 11.9792 (10.3393) acc5: 100.0000 (100.0000) time: 0.1348 data: 0.0374 max mem: 3531 +Test: [110/389] eta: 0:00:57 acc1_pos: 78.125 acc5_pos: 100.000 loss: 1.6228 (1.4872) acc1: 0.5208 (11.8384) acc5: 100.0000 (100.0000) time: 0.1549 data: 0.0577 max mem: 3531 +Test: [120/389] eta: 0:00:54 acc1_pos: 20.052 acc5_pos: 100.000 loss: 1.4122 (1.4788) acc1: 15.6250 (12.4763) acc5: 100.0000 (100.0000) time: 0.1649 data: 0.0691 max mem: 3531 +Test: [130/389] eta: 0:00:51 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.9774 (1.4072) acc1: 41.6667 (17.2551) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1599 data: 0.0647 max mem: 3531 +Test: [140/389] eta: 0:00:48 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1736 (1.3228) acc1: 99.2188 (22.9037) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.1610 data: 0.0643 max mem: 3531 +Test: [150/389] eta: 0:00:46 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1560 (1.2520) acc1: 100.0000 (27.4248) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1599 data: 0.0652 max mem: 3531 +Test: [160/389] eta: 0:00:44 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1479 (1.1856) acc1: 99.7396 (31.8291) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1739 data: 0.0793 max mem: 3531 +Test: [170/389] eta: 0:00:41 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1310 (1.1252) acc1: 100.0000 (35.7959) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1551 data: 0.0609 max mem: 3531 +Test: [180/389] eta: 0:00:38 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1310 (1.0716) acc1: 100.0000 (39.2913) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.1305 data: 0.0365 max mem: 3531 +Test: [190/389] eta: 0:00:36 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1463 (1.0276) acc1: 100.0000 (42.2979) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1447 data: 0.0507 max mem: 3531 +Test: [200/389] eta: 0:00:34 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1464 (0.9876) acc1: 99.7396 (44.9873) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1715 data: 0.0791 max mem: 3531 +Test: [210/389] eta: 0:00:32 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1349 (0.9470) acc1: 100.0000 (47.5945) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1767 data: 0.0850 max mem: 3532 +Test: [220/389] eta: 0:00:30 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1188 (0.9093) acc1: 100.0000 (49.9599) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1653 data: 0.0736 max mem: 3532 +Test: [230/389] eta: 0:00:29 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1231 (0.8780) acc1: 100.0000 (52.1014) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1820 data: 0.0896 max mem: 3532 +Test: [240/389] eta: 0:00:27 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1769 (0.8494) acc1: 99.7396 (54.0651) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1866 data: 0.0933 max mem: 3532 +Test: [250/389] eta: 0:00:25 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1603 (0.8218) acc1: 100.0000 (55.8952) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1735 data: 0.0802 max mem: 3532 +Test: [260/389] eta: 0:00:23 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1601 (0.7994) acc1: 100.0000 (57.2907) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1624 data: 0.0709 max mem: 3532 +Test: [270/389] eta: 0:00:21 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1778 (0.7770) acc1: 99.7396 (58.8417) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1663 data: 0.0747 max mem: 3532 +Test: [280/389] eta: 0:00:19 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1455 (0.7569) acc1: 100.0000 (60.2369) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1805 data: 0.0888 max mem: 3532 +Test: [290/389] eta: 0:00:17 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1724 (0.7383) acc1: 100.0000 (61.5326) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1587 data: 0.0674 max mem: 3532 +Test: [300/389] eta: 0:00:15 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1707 (0.7198) acc1: 100.0000 (62.8089) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1417 data: 0.0492 max mem: 3532 +Test: [310/389] eta: 0:00:14 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1856 (0.7034) acc1: 100.0000 (63.9712) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1411 data: 0.0490 max mem: 3532 +Test: [320/389] eta: 0:00:12 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1921 (0.6876) acc1: 100.0000 (65.0417) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1380 data: 0.0473 max mem: 3532 +Test: [330/389] eta: 0:00:10 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1723 (0.6726) acc1: 100.0000 (66.0656) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1373 data: 0.0467 max mem: 3532 +Test: [340/389] eta: 0:00:08 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1708 (0.6577) acc1: 100.0000 (66.9935) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1366 data: 0.0441 max mem: 3532 +Test: [350/389] eta: 0:00:06 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1708 (0.6505) acc1: 100.0000 (67.3596) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.1339 data: 0.0407 max mem: 3532 +Test: [360/389] eta: 0:00:04 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2675 (0.6416) acc1: 94.0104 (67.9781) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1305 data: 0.0384 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2535 (0.6299) acc1: 95.8333 (68.8033) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1263 data: 0.0348 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1704 (0.6182) acc1: 100.0000 (69.6099) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1288 data: 0.0362 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1699 (0.6092) acc1: 100.0000 (70.1904) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1269 data: 0.0303 max mem: 3532 +Test: Total time: 0:01:05 (0.1688 s / it) +* Overall Acc@1 70.190 Acc@5 100.000 + target == real Acc@1 12.261 Acc@5 100.000 + target == fake Acc@1 97.118 Acc@5 100.000 + loss 0.609 + PR-AUC 0.8246 ROC-AUC 0.7102 EER 34.17% (thr=0.8345) +Accuracy of the network on the 149358 test images: 70.2% +[rank0]:[W726 06:12:11.535022140 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 06:12:30.356000 1203950 torch/distributed/run.py:766] +W0726 06:12:30.356000 1203950 torch/distributed/run.py:766] ***************************************** +W0726 06:12:30.356000 1203950 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:12:30.356000 1203950 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 06:12:58.205823043 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:12:58.279846083 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:12:59.788103966 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:12:59.912480284 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:12:59.020898352 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:12:59.139097661 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:12:59.175091990 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:12:59.203070587 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n23_seed9_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed9_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:36:58 acc1_pos: 20.573 acc5_pos: 100.000 loss: 1.2941 (1.2941) acc1: 20.5729 (20.5729) acc5: 100.0000 (100.0000) time: 6.6627 data: 5.1583 max mem: 3531 +Test: [ 10/333] eta: 0:03:55 acc1_pos: 89.323 acc5_pos: 100.000 loss: 1.0800 (0.8237) acc1: 39.8438 (51.2311) acc5: 100.0000 (100.0000) time: 0.7282 data: 0.5062 max mem: 3531 +Test: [ 20/333] eta: 0:02:14 acc1_pos: 91.927 acc5_pos: 100.000 loss: 0.3599 (0.5938) acc1: 85.1562 (68.0308) acc5: 100.0000 (100.0000) time: 0.1173 data: 0.0207 max mem: 3531 +Test: [ 30/333] eta: 0:01:50 acc1_pos: 74.740 acc5_pos: 100.000 loss: 0.3878 (0.5904) acc1: 79.4271 (68.8340) acc5: 100.0000 (100.0000) time: 0.1669 data: 0.0714 max mem: 3531 +Test: [ 40/333] eta: 0:01:34 acc1_pos: 71.354 acc5_pos: 100.000 loss: 0.5670 (0.5906) acc1: 70.5729 (68.9596) acc5: 100.0000 (100.0000) time: 0.2111 data: 0.1175 max mem: 3531 +Test: [ 50/333] eta: 0:01:24 acc1_pos: 67.188 acc5_pos: 100.000 loss: 0.6049 (0.5919) acc1: 69.2708 (69.0666) acc5: 100.0000 (100.0000) time: 0.1928 data: 0.0946 max mem: 3531 +Test: [ 60/333] eta: 0:01:17 acc1_pos: 68.490 acc5_pos: 100.000 loss: 0.6008 (0.5912) acc1: 69.5312 (69.4886) acc5: 100.0000 (100.0000) time: 0.1994 data: 0.1019 max mem: 3531 +Test: [ 70/333] eta: 0:01:12 acc1_pos: 63.802 acc5_pos: 100.000 loss: 0.5458 (0.5885) acc1: 73.9583 (69.9494) acc5: 100.0000 (100.0000) time: 0.2149 data: 0.1186 max mem: 3531 +Test: [ 80/333] eta: 0:01:08 acc1_pos: 69.010 acc5_pos: 100.000 loss: 0.5750 (0.5879) acc1: 72.1354 (70.3125) acc5: 100.0000 (100.0000) time: 0.2392 data: 0.1407 max mem: 3531 +Test: [ 90/333] eta: 0:01:05 acc1_pos: 83.854 acc5_pos: 100.000 loss: 0.5693 (0.5775) acc1: 73.1771 (71.2569) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1474 max mem: 3531 +Test: [100/333] eta: 0:00:59 acc1_pos: 82.552 acc5_pos: 100.000 loss: 0.4867 (0.5691) acc1: 77.6042 (71.8982) acc5: 100.0000 (100.0000) time: 0.1799 data: 0.0802 max mem: 3531 +Test: [110/333] eta: 0:00:56 acc1_pos: 78.125 acc5_pos: 100.000 loss: 0.4929 (0.5620) acc1: 77.8646 (72.4779) acc5: 100.0000 (100.0000) time: 0.1872 data: 0.0905 max mem: 3531 +Test: [120/333] eta: 0:00:53 acc1_pos: 78.385 acc5_pos: 100.000 loss: 0.4653 (0.5558) acc1: 79.9479 (72.9554) acc5: 100.0000 (100.0000) time: 0.2352 data: 0.1435 max mem: 3531 +Test: [130/333] eta: 0:00:50 acc1_pos: 87.240 acc5_pos: 100.000 loss: 0.4555 (0.5485) acc1: 79.9479 (73.5588) acc5: 100.0000 (100.0000) time: 0.2140 data: 0.1216 max mem: 3531 +Test: [140/333] eta: 0:00:46 acc1_pos: 81.510 acc5_pos: 100.000 loss: 0.4451 (0.5415) acc1: 81.5104 (74.2040) acc5: 100.0000 (100.0000) time: 0.1847 data: 0.0920 max mem: 3531 +Test: [150/333] eta: 0:00:43 acc1_pos: 80.469 acc5_pos: 100.000 loss: 0.4504 (0.5372) acc1: 80.9896 (74.5654) acc5: 100.0000 (100.0000) time: 0.1666 data: 0.0725 max mem: 3531 +Test: [160/333] eta: 0:00:40 acc1_pos: 78.618 acc5_pos: 100.000 loss: 0.4619 (0.5482) acc1: 79.9479 (73.7012) acc5: 100.0000 (100.0000) acc1_neg: 27.865 acc5_neg: 100.000 time: 0.2003 data: 0.1069 max mem: 3531 +Test: [170/333] eta: 0:00:38 acc1_pos: 78.618 acc5_pos: 100.000 loss: 0.8949 (0.5716) acc1: 37.7604 (71.4638) acc5: 100.0000 (100.0000) acc1_neg: 39.844 acc5_neg: 100.000 time: 0.2219 data: 0.1269 max mem: 3531 +Test: [180/333] eta: 0:00:35 acc1_pos: 78.618 acc5_pos: 100.000 loss: 0.8879 (0.5870) acc1: 38.2812 (69.8061) acc5: 100.0000 (100.0000) acc1_neg: 38.281 acc5_neg: 100.000 time: 0.2047 data: 0.1096 max mem: 3531 +Test: [190/333] eta: 0:00:32 acc1_pos: 78.618 acc5_pos: 100.000 loss: 0.8375 (0.5983) acc1: 43.2292 (68.5223) acc5: 100.0000 (100.0000) acc1_neg: 44.792 acc5_neg: 100.000 time: 0.1802 data: 0.0885 max mem: 3531 +Test: [200/333] eta: 0:00:30 acc1_pos: 78.618 acc5_pos: 100.000 loss: 0.8184 (0.6085) acc1: 44.7917 (67.4635) acc5: 100.0000 (100.0000) acc1_neg: 45.312 acc5_neg: 100.000 time: 0.1767 data: 0.0852 max mem: 3531 +Test: [210/333] eta: 0:00:27 acc1_pos: 78.618 acc5_pos: 100.000 loss: 0.8184 (0.6200) acc1: 45.3125 (66.3976) acc5: 100.0000 (100.0000) acc1_neg: 54.167 acc5_neg: 100.000 time: 0.1696 data: 0.0772 max mem: 3532 +Test: [220/333] eta: 0:00:25 acc1_pos: 78.618 acc5_pos: 100.000 loss: 0.8370 (0.6329) acc1: 43.4896 (65.2173) acc5: 100.0000 (100.0000) acc1_neg: 58.073 acc5_neg: 100.000 time: 0.1649 data: 0.0724 max mem: 3532 +Test: [230/333] eta: 0:00:22 acc1_pos: 78.618 acc5_pos: 100.000 loss: 0.6919 (0.6329) acc1: 58.8542 (65.2327) acc5: 100.0000 (100.0000) acc1_neg: 62.240 acc5_neg: 100.000 time: 0.1799 data: 0.0880 max mem: 3532 +Test: [240/333] eta: 0:00:20 acc1_pos: 78.618 acc5_pos: 100.000 loss: 0.6558 (0.6405) acc1: 59.1146 (64.3240) acc5: 100.0000 (100.0000) acc1_neg: 36.458 acc5_neg: 100.000 time: 0.1743 data: 0.0827 max mem: 3532 +Test: [250/333] eta: 0:00:17 acc1_pos: 78.618 acc5_pos: 100.000 loss: 0.8341 (0.6482) acc1: 41.4062 (63.4888) acc5: 100.0000 (100.0000) acc1_neg: 39.062 acc5_neg: 100.000 time: 0.1762 data: 0.0852 max mem: 3532 +Test: [260/333] eta: 0:00:15 acc1_pos: 78.618 acc5_pos: 100.000 loss: 0.8196 (0.6541) acc1: 45.0521 (62.8672) acc5: 100.0000 (100.0000) acc1_neg: 51.042 acc5_neg: 100.000 time: 0.1734 data: 0.0822 max mem: 3532 +Test: [270/333] eta: 0:00:13 acc1_pos: 78.618 acc5_pos: 100.000 loss: 0.7674 (0.6575) acc1: 50.7812 (62.5740) acc5: 100.0000 (100.0000) acc1_neg: 50.781 acc5_neg: 100.000 time: 0.1408 data: 0.0489 max mem: 3532 +Test: [280/333] eta: 0:00:11 acc1_pos: 78.618 acc5_pos: 100.000 loss: 0.7582 (0.6656) acc1: 50.5208 (61.7521) acc5: 100.0000 (100.0000) acc1_neg: 38.542 acc5_neg: 100.000 time: 0.1369 data: 0.0458 max mem: 3532 +Test: [290/333] eta: 0:00:08 acc1_pos: 78.618 acc5_pos: 100.000 loss: 0.9819 (0.6770) acc1: 33.0729 (60.7433) acc5: 100.0000 (100.0000) acc1_neg: 29.427 acc5_neg: 100.000 time: 0.1501 data: 0.0590 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 78.618 acc5_pos: 100.000 loss: 0.9912 (0.6876) acc1: 31.5104 (59.8007) acc5: 100.0000 (100.0000) acc1_neg: 33.333 acc5_neg: 100.000 time: 0.1435 data: 0.0522 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 78.618 acc5_pos: 100.000 loss: 0.9179 (0.6889) acc1: 36.4583 (59.5994) acc5: 100.0000 (100.0000) acc1_neg: 47.135 acc5_neg: 100.000 time: 0.1425 data: 0.0511 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 78.618 acc5_pos: 100.000 loss: 0.7570 (0.6921) acc1: 48.9583 (59.2127) acc5: 100.0000 (100.0000) acc1_neg: 36.458 acc5_neg: 100.000 time: 0.1409 data: 0.0480 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 78.618 acc5_pos: 100.000 loss: 0.9600 (0.7017) acc1: 36.4583 (58.4427) acc5: 100.0000 (100.0000) acc1_neg: 34.896 acc5_neg: 100.000 time: 0.1141 data: 0.0222 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 78.618 acc5_pos: 100.000 loss: 0.9879 (0.7035) acc1: 35.6771 (58.3082) acc5: 100.0000 (100.0000) acc1_neg: 38.028 acc5_neg: 100.000 time: 0.1247 data: 0.0305 max mem: 3532 +Test: Total time: 0:01:05 (0.1978 s / it) +* Overall Acc@1 58.308 Acc@5 100.000 + target == real Acc@1 74.980 Acc@5 100.000 + target == fake Acc@1 43.467 Acc@5 100.000 + loss 0.703 + PR-AUC 0.6468 ROC-AUC 0.6477 EER 39.33% (thr=0.3940) +Accuracy of the network on the 127843 test images: 58.3% +[rank0]:[W726 06:14:16.154507785 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 06:14:36.513000 1204509 torch/distributed/run.py:766] +W0726 06:14:36.513000 1204509 torch/distributed/run.py:766] ***************************************** +W0726 06:14:36.513000 1204509 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:14:36.513000 1204509 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:15:05.720292040 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:15:05.891771028 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 06:15:05.051376207 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:15:05.248421517 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:15:06.545074791 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +| distributed init (rank 6): env:// +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:15:06.583910916 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:15:06.586842103 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:15:06.590498375 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n23_seed9_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed9_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:25:50 acc1_pos: 96.094 acc5_pos: 100.000 loss: 0.1948 (0.1948) acc1: 96.0938 (96.0938) acc5: 100.0000 (100.0000) time: 6.5989 data: 5.1384 max mem: 3531 +Test: [ 10/235] eta: 0:02:33 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.1948 (0.1955) acc1: 96.6146 (96.5909) acc5: 100.0000 (100.0000) time: 0.6831 data: 0.4675 max mem: 3531 +Test: [ 20/235] eta: 0:01:27 acc1_pos: 96.615 acc5_pos: 100.000 loss: 0.1907 (0.1931) acc1: 96.6146 (96.5774) acc5: 100.0000 (100.0000) time: 0.0957 data: 0.0004 max mem: 3531 +Test: [ 30/235] eta: 0:01:15 acc1_pos: 97.135 acc5_pos: 100.000 loss: 0.1887 (0.1922) acc1: 96.3542 (96.4046) acc5: 100.0000 (100.0000) time: 0.1926 data: 0.0936 max mem: 3531 +Test: [ 40/235] eta: 0:01:08 acc1_pos: 95.833 acc5_pos: 100.000 loss: 0.1864 (0.1915) acc1: 96.3542 (96.4621) acc5: 100.0000 (100.0000) time: 0.2913 data: 0.1968 max mem: 3531 +Test: [ 50/235] eta: 0:00:59 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.1897 (0.1927) acc1: 96.3542 (96.4512) acc5: 100.0000 (100.0000) time: 0.2480 data: 0.1562 max mem: 3531 +Test: [ 60/235] eta: 0:00:51 acc1_pos: 96.615 acc5_pos: 100.000 loss: 0.1997 (0.1944) acc1: 96.0938 (96.3584) acc5: 100.0000 (100.0000) time: 0.1746 data: 0.0808 max mem: 3531 +Test: [ 70/235] eta: 0:00:46 acc1_pos: 98.462 acc5_pos: 100.000 loss: 0.2136 (0.3482) acc1: 94.2708 (86.7664) acc5: 100.0000 (100.0000) acc1_neg: 17.188 acc5_neg: 100.000 time: 0.1761 data: 0.0838 max mem: 3531 +Test: [ 80/235] eta: 0:00:41 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4305 (0.4821) acc1: 20.0521 (78.4722) acc5: 100.0000 (100.0000) acc1_neg: 18.229 acc5_neg: 100.000 time: 0.1938 data: 0.1035 max mem: 3531 +Test: [ 90/235] eta: 0:00:37 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4343 (0.5867) acc1: 20.0521 (72.0152) acc5: 100.0000 (100.0000) acc1_neg: 16.927 acc5_neg: 100.000 time: 0.1942 data: 0.1007 max mem: 3531 +Test: [100/235] eta: 0:00:35 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4353 (0.6685) acc1: 19.5312 (66.8987) acc5: 100.0000 (100.0000) acc1_neg: 14.844 acc5_neg: 100.000 time: 0.2249 data: 0.1306 max mem: 3531 +Test: [110/235] eta: 0:00:32 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4068 (0.7356) acc1: 19.5312 (62.6548) acc5: 100.0000 (100.0000) acc1_neg: 21.094 acc5_neg: 100.000 time: 0.2352 data: 0.1410 max mem: 3531 +Test: [120/235] eta: 0:00:28 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4776 (0.8111) acc1: 16.4062 (58.4065) acc5: 100.0000 (100.0000) acc1_neg: 13.542 acc5_neg: 100.000 time: 0.1810 data: 0.0844 max mem: 3531 +Test: [130/235] eta: 0:00:25 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.6607 (0.8760) acc1: 10.6771 (54.7670) acc5: 100.0000 (100.0000) acc1_neg: 11.458 acc5_neg: 100.000 time: 0.1533 data: 0.0591 max mem: 3531 +Test: [140/235] eta: 0:00:22 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.6921 (0.9349) acc1: 9.6354 (51.5514) acc5: 100.0000 (100.0000) acc1_neg: 6.250 acc5_neg: 100.000 time: 0.1745 data: 0.0816 max mem: 3531 +Test: [150/235] eta: 0:00:19 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.6887 (0.9827) acc1: 9.6354 (48.8204) acc5: 100.0000 (100.0000) acc1_neg: 12.240 acc5_neg: 100.000 time: 0.1824 data: 0.0886 max mem: 3531 +Test: [160/235] eta: 0:00:17 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.6667 (1.0257) acc1: 9.6354 (46.4140) acc5: 100.0000 (100.0000) acc1_neg: 11.458 acc5_neg: 100.000 time: 0.1792 data: 0.0854 max mem: 3531 +Test: [170/235] eta: 0:00:14 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.6248 (1.0333) acc1: 12.2396 (45.5531) acc5: 100.0000 (100.0000) acc1_neg: 30.208 acc5_neg: 100.000 time: 0.1774 data: 0.0850 max mem: 3531 +Test: [180/235] eta: 0:00:12 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.1098 (1.0363) acc1: 34.6354 (45.0190) acc5: 100.0000 (100.0000) acc1_neg: 36.198 acc5_neg: 100.000 time: 0.1803 data: 0.0883 max mem: 3531 +Test: [190/235] eta: 0:00:10 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.0789 (1.0387) acc1: 36.1979 (44.5285) acc5: 100.0000 (100.0000) acc1_neg: 34.115 acc5_neg: 100.000 time: 0.1900 data: 0.0983 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.0816 (1.0418) acc1: 34.8958 (43.9819) acc5: 100.0000 (100.0000) acc1_neg: 32.031 acc5_neg: 100.000 time: 0.1871 data: 0.0944 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.0990 (1.0437) acc1: 34.6354 (43.5945) acc5: 100.0000 (100.0000) acc1_neg: 38.802 acc5_neg: 100.000 time: 0.1881 data: 0.0942 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.0833 (1.0461) acc1: 34.6354 (43.1773) acc5: 100.0000 (100.0000) acc1_neg: 33.073 acc5_neg: 100.000 time: 0.1811 data: 0.0901 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.0802 (1.0472) acc1: 34.8958 (42.8921) acc5: 100.0000 (100.0000) acc1_neg: 38.802 acc5_neg: 100.000 time: 0.1473 data: 0.0574 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.0807 (1.0488) acc1: 35.1562 (42.7653) acc5: 100.0000 (100.0000) acc1_neg: 39.130 acc5_neg: 100.000 time: 0.1377 data: 0.0507 max mem: 3532 +Test: Total time: 0:00:49 (0.2121 s / it) +* Overall Acc@1 42.765 Acc@5 100.000 + target == real Acc@1 96.360 Acc@5 100.000 + target == fake Acc@1 23.381 Acc@5 100.000 + loss 1.049 + PR-AUC 0.8915 ROC-AUC 0.7594 EER 31.04% (thr=0.1674) +Accuracy of the network on the 89879 test images: 42.8% +[rank0]:[W726 06:16:05.562167566 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 06:16:25.325000 1205055 torch/distributed/run.py:766] +W0726 06:16:25.325000 1205055 torch/distributed/run.py:766] ***************************************** +W0726 06:16:25.325000 1205055 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:16:25.325000 1205055 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:16:53.796932545 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:16:54.440095589 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:16:54.747797179 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:16:54.767878291 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 06:16:54.858645844 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:16:54.933594279 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 3): env:// +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:16:54.967107594 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:16:54.975080757 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n23_seed9_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n23_classseed9_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:30:51 acc1_pos: 76.302 acc5_pos: 100.000 loss: 0.5403 (0.5403) acc1: 76.3021 (76.3021) acc5: 100.0000 (100.0000) time: 6.8844 data: 5.1818 max mem: 3531 +Test: [ 10/269] eta: 0:03:05 acc1_pos: 83.333 acc5_pos: 100.000 loss: 0.5218 (0.5105) acc1: 78.9062 (80.4688) acc5: 100.0000 (100.0000) time: 0.7170 data: 0.4716 max mem: 3531 +Test: [ 20/269] eta: 0:01:45 acc1_pos: 86.458 acc5_pos: 100.000 loss: 0.4983 (0.4989) acc1: 83.3333 (81.2128) acc5: 100.0000 (100.0000) time: 0.1001 data: 0.0005 max mem: 3531 +Test: [ 30/269] eta: 0:01:28 acc1_pos: 92.708 acc5_pos: 100.000 loss: 0.4646 (0.4858) acc1: 84.6354 (83.2409) acc5: 100.0000 (100.0000) time: 0.1764 data: 0.0799 max mem: 3531 +Test: [ 40/269] eta: 0:01:17 acc1_pos: 91.379 acc5_pos: 100.000 loss: 0.5239 (0.5479) acc1: 78.1250 (75.3430) acc5: 100.0000 (100.0000) acc1_neg: 29.948 acc5_neg: 100.000 time: 0.2499 data: 0.1553 max mem: 3531 +Test: [ 50/269] eta: 0:01:09 acc1_pos: 91.379 acc5_pos: 100.000 loss: 0.7467 (0.5962) acc1: 46.6146 (68.9236) acc5: 100.0000 (100.0000) acc1_neg: 26.823 acc5_neg: 100.000 time: 0.2334 data: 0.1394 max mem: 3531 +Test: [ 60/269] eta: 0:01:01 acc1_pos: 91.379 acc5_pos: 100.000 loss: 0.8119 (0.6420) acc1: 36.1979 (62.7860) acc5: 100.0000 (100.0000) acc1_neg: 25.781 acc5_neg: 100.000 time: 0.1964 data: 0.1022 max mem: 3531 +Test: [ 70/269] eta: 0:00:55 acc1_pos: 91.379 acc5_pos: 100.000 loss: 0.8712 (0.6689) acc1: 29.4271 (58.9312) acc5: 100.0000 (100.0000) acc1_neg: 32.552 acc5_neg: 100.000 time: 0.1909 data: 0.0970 max mem: 3531 +Test: [ 80/269] eta: 0:00:51 acc1_pos: 91.379 acc5_pos: 100.000 loss: 0.7545 (0.6754) acc1: 40.1042 (57.8221) acc5: 100.0000 (100.0000) acc1_neg: 72.656 acc5_neg: 100.000 time: 0.2040 data: 0.1106 max mem: 3531 +Test: [ 90/269] eta: 0:00:46 acc1_pos: 91.379 acc5_pos: 100.000 loss: 0.7420 (0.6881) acc1: 46.0938 (56.2242) acc5: 100.0000 (100.0000) acc1_neg: 40.885 acc5_neg: 100.000 time: 0.1990 data: 0.1053 max mem: 3531 +Test: [100/269] eta: 0:00:43 acc1_pos: 91.379 acc5_pos: 100.000 loss: 0.8360 (0.7084) acc1: 35.6771 (53.7412) acc5: 100.0000 (100.0000) acc1_neg: 18.750 acc5_neg: 100.000 time: 0.1957 data: 0.1016 max mem: 3531 +Test: [110/269] eta: 0:00:39 acc1_pos: 91.379 acc5_pos: 100.000 loss: 0.9390 (0.7323) acc1: 25.7812 (51.3959) acc5: 100.0000 (100.0000) acc1_neg: 25.781 acc5_neg: 100.000 time: 0.1949 data: 0.1005 max mem: 3531 +Test: [120/269] eta: 0:00:36 acc1_pos: 91.379 acc5_pos: 100.000 loss: 0.9336 (0.7406) acc1: 30.9896 (50.5746) acc5: 100.0000 (100.0000) acc1_neg: 33.594 acc5_neg: 100.000 time: 0.1886 data: 0.0954 max mem: 3531 +Test: [130/269] eta: 0:00:33 acc1_pos: 91.379 acc5_pos: 100.000 loss: 0.9008 (0.7532) acc1: 35.4167 (49.4175) acc5: 100.0000 (100.0000) acc1_neg: 35.417 acc5_neg: 100.000 time: 0.1762 data: 0.0823 max mem: 3531 +Test: [140/269] eta: 0:00:30 acc1_pos: 91.379 acc5_pos: 100.000 loss: 0.9705 (0.7705) acc1: 29.1667 (47.6470) acc5: 100.0000 (100.0000) acc1_neg: 20.312 acc5_neg: 100.000 time: 0.1812 data: 0.0839 max mem: 3531 +Test: [150/269] eta: 0:00:27 acc1_pos: 91.379 acc5_pos: 100.000 loss: 0.9748 (0.7788) acc1: 24.7396 (46.7577) acc5: 100.0000 (100.0000) acc1_neg: 48.698 acc5_neg: 100.000 time: 0.1842 data: 0.0864 max mem: 3531 +Test: [160/269] eta: 0:00:24 acc1_pos: 91.379 acc5_pos: 100.000 loss: 0.7459 (0.7781) acc1: 48.6979 (46.8701) acc5: 100.0000 (100.0000) acc1_neg: 50.781 acc5_neg: 100.000 time: 0.1800 data: 0.0866 max mem: 3531 +Test: [170/269] eta: 0:00:22 acc1_pos: 91.379 acc5_pos: 100.000 loss: 0.7333 (0.7763) acc1: 52.3438 (47.0121) acc5: 100.0000 (100.0000) acc1_neg: 46.094 acc5_neg: 100.000 time: 0.1869 data: 0.0960 max mem: 3531 +Test: [180/269] eta: 0:00:19 acc1_pos: 91.379 acc5_pos: 100.000 loss: 0.6782 (0.7686) acc1: 58.8542 (47.9512) acc5: 100.0000 (100.0000) acc1_neg: 41.146 acc5_neg: 100.000 time: 0.1895 data: 0.0992 max mem: 3531 +Test: [190/269] eta: 0:00:17 acc1_pos: 91.379 acc5_pos: 100.000 loss: 0.6711 (0.7726) acc1: 52.8646 (47.4558) acc5: 100.0000 (100.0000) acc1_neg: 48.698 acc5_neg: 100.000 time: 0.1873 data: 0.0948 max mem: 3531 +Test: [200/269] eta: 0:00:15 acc1_pos: 91.379 acc5_pos: 100.000 loss: 0.8757 (0.7803) acc1: 39.5833 (47.0823) acc5: 100.0000 (100.0000) acc1_neg: 52.344 acc5_neg: 100.000 time: 0.1892 data: 0.0932 max mem: 3531 +Test: [210/269] eta: 0:00:12 acc1_pos: 91.379 acc5_pos: 100.000 loss: 0.9538 (0.7869) acc1: 34.6354 (46.3443) acc5: 100.0000 (100.0000) acc1_neg: 37.240 acc5_neg: 100.000 time: 0.1952 data: 0.1016 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 91.379 acc5_pos: 100.000 loss: 0.9099 (0.7914) acc1: 28.6458 (45.9276) acc5: 100.0000 (100.0000) acc1_neg: 61.719 acc5_neg: 100.000 time: 0.1928 data: 0.1026 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 91.379 acc5_pos: 100.000 loss: 0.7694 (0.7899) acc1: 43.7500 (46.0205) acc5: 100.0000 (100.0000) acc1_neg: 44.010 acc5_neg: 100.000 time: 0.1871 data: 0.0940 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 91.379 acc5_pos: 100.000 loss: 0.8187 (0.7936) acc1: 38.0208 (45.4616) acc5: 100.0000 (100.0000) acc1_neg: 39.323 acc5_neg: 100.000 time: 0.1839 data: 0.0909 max mem: 3532 +Test: [250/269] eta: 0:00:04 acc1_pos: 91.379 acc5_pos: 100.000 loss: 0.8506 (0.7993) acc1: 33.0729 (44.8155) acc5: 100.0000 (100.0000) acc1_neg: 5.990 acc5_neg: 100.000 time: 0.1759 data: 0.0849 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 91.379 acc5_pos: 100.000 loss: 1.0121 (0.8080) acc1: 15.3646 (43.6971) acc5: 100.0000 (100.0000) acc1_neg: 32.292 acc5_neg: 100.000 time: 0.1368 data: 0.0458 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 91.379 acc5_pos: 100.000 loss: 0.9466 (0.8108) acc1: 17.1875 (43.1578) acc5: 100.0000 (100.0000) acc1_neg: 31.496 acc5_neg: 100.000 time: 0.1088 data: 0.0148 max mem: 3532 +Test: Total time: 0:00:55 (0.2074 s / it) +* Overall Acc@1 43.158 Acc@5 100.000 + target == real Acc@1 83.159 Acc@5 100.000 + target == fake Acc@1 37.699 Acc@5 100.000 + loss 0.811 + PR-AUC 0.9297 ROC-AUC 0.6589 EER 39.31% (thr=0.3965) +Accuracy of the network on the 103293 test images: 43.2% +[rank0]:[W726 06:17:59.830254859 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T06:18:06Z] eval_done tag=fixed500k_real_n23_seed9_resampled +[2026-07-26T06:18:06Z] eval_start tag=fixed500k_fake_n45_seed9_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed9_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T06:18:06Z +hostname=n116-041-023 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed9_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n45_seed9_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 06:18:19.400000 1205633 torch/distributed/run.py:766] +W0726 06:18:19.400000 1205633 torch/distributed/run.py:766] ***************************************** +W0726 06:18:19.400000 1205633 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:18:19.400000 1205633 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:18:48.853324812 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:18:48.047302639 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 06:18:48.108816685 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W726 06:18:48.143390830 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +NCCL version 2.26.2+cuda12.2 +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:18:49.437625203 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +| distributed init (rank 1): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:18:49.643679423 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank1]:[W726 06:18:49.643679583 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:18:49.647026409 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n45_seed9_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed9_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:27:00 acc1_pos: 80.208 acc5_pos: 100.000 loss: 0.5152 (0.5152) acc1: 80.2083 (80.2083) acc5: 100.0000 (100.0000) time: 6.9251 data: 5.3634 max mem: 3531 +Test: [ 10/234] eta: 0:02:44 acc1_pos: 41.927 acc5_pos: 100.000 loss: 0.6746 (0.6776) acc1: 61.1979 (59.7301) acc5: 100.0000 (100.0000) time: 0.7339 data: 0.5101 max mem: 3531 +Test: [ 20/234] eta: 0:01:35 acc1_pos: 96.354 acc5_pos: 100.000 loss: 0.6714 (0.6382) acc1: 60.1562 (62.1528) acc5: 100.0000 (100.0000) time: 0.1199 data: 0.0261 max mem: 3531 +Test: [ 30/234] eta: 0:01:24 acc1_pos: 66.406 acc5_pos: 100.000 loss: 0.2188 (0.5179) acc1: 95.3125 (71.6650) acc5: 100.0000 (100.0000) time: 0.2349 data: 0.1404 max mem: 3531 +Test: [ 40/234] eta: 0:01:11 acc1_pos: 98.698 acc5_pos: 100.000 loss: 0.2186 (0.5010) acc1: 95.5729 (72.4530) acc5: 100.0000 (100.0000) time: 0.2899 data: 0.1965 max mem: 3531 +Test: [ 50/234] eta: 0:01:02 acc1_pos: 98.698 acc5_pos: 100.000 loss: 0.1589 (0.4325) acc1: 96.8750 (77.3846) acc5: 100.0000 (100.0000) time: 0.2283 data: 0.1335 max mem: 3531 +Test: [ 60/234] eta: 0:00:56 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.1354 (0.3804) acc1: 98.1771 (81.0237) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2321 data: 0.1373 max mem: 3531 +Test: [ 70/234] eta: 0:00:51 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.1154 (0.3874) acc1: 99.4792 (80.7842) acc5: 100.0000 (100.0000) acc1_neg: 67.969 acc5_neg: 100.000 time: 0.2391 data: 0.1462 max mem: 3531 +Test: [ 80/234] eta: 0:00:47 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.5703 (0.3954) acc1: 70.8333 (80.6488) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2541 data: 0.1633 max mem: 3531 +Test: [ 90/234] eta: 0:00:43 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.2882 (0.3763) acc1: 86.7188 (81.9883) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.2549 data: 0.1616 max mem: 3531 +Test: [100/234] eta: 0:00:39 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.3019 (0.3815) acc1: 86.7188 (81.7863) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2286 data: 0.1328 max mem: 3531 +Test: [110/234] eta: 0:00:35 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.4875 (0.3894) acc1: 74.2188 (81.4142) acc5: 100.0000 (100.0000) acc1_neg: 59.115 acc5_neg: 100.000 time: 0.2219 data: 0.1276 max mem: 3531 +Test: [120/234] eta: 0:00:31 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.4429 (0.4012) acc1: 76.5625 (80.9250) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.2127 data: 0.1209 max mem: 3531 +Test: [130/234] eta: 0:00:28 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.4657 (0.4069) acc1: 80.7292 (80.9379) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.2131 data: 0.1221 max mem: 3531 +Test: [140/234] eta: 0:00:25 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.4831 (0.4356) acc1: 75.2604 (79.4622) acc5: 100.0000 (100.0000) acc1_neg: 37.500 acc5_neg: 100.000 time: 0.2144 data: 0.1238 max mem: 3531 +Test: [150/234] eta: 0:00:22 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.4971 (0.4447) acc1: 73.4375 (78.9287) acc5: 100.0000 (100.0000) acc1_neg: 76.042 acc5_neg: 100.000 time: 0.2122 data: 0.1197 max mem: 3531 +Test: [160/234] eta: 0:00:19 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.4480 (0.4466) acc1: 75.2604 (78.8836) acc5: 100.0000 (100.0000) acc1_neg: 66.146 acc5_neg: 100.000 time: 0.2076 data: 0.1149 max mem: 3531 +Test: [170/234] eta: 0:00:16 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.5959 (0.4535) acc1: 68.4896 (78.4829) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1897 data: 0.0983 max mem: 3531 +Test: [180/234] eta: 0:00:13 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.5998 (0.4640) acc1: 67.4479 (77.7883) acc5: 100.0000 (100.0000) acc1_neg: 36.458 acc5_neg: 100.000 time: 0.1699 data: 0.0788 max mem: 3531 +Test: [190/234] eta: 0:00:10 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.6542 (0.4749) acc1: 64.0625 (77.0070) acc5: 100.0000 (100.0000) acc1_neg: 64.323 acc5_neg: 100.000 time: 0.1468 data: 0.0568 max mem: 3531 +Test: [200/234] eta: 0:00:08 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.6390 (0.4666) acc1: 64.5833 (77.5238) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1384 data: 0.0480 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.1213 (0.4510) acc1: 99.7396 (78.5360) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1350 data: 0.0442 max mem: 3532 +Test: [220/234] eta: 0:00:03 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.1707 (0.4392) acc1: 97.6562 (79.3257) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1233 data: 0.0328 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.3189 (0.4469) acc1: 66.6667 (78.6909) acc5: 100.0000 (100.0000) acc1_neg: 63.802 acc5_neg: 100.000 time: 0.1131 data: 0.0231 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.5976 (0.4488) acc1: 66.4062 (78.5476) acc5: 100.0000 (100.0000) acc1_neg: 69.261 acc5_neg: 100.000 time: 0.1122 data: 0.0210 max mem: 3532 +Test: Total time: 0:00:52 (0.2235 s / it) +* Overall Acc@1 78.548 Acc@5 100.000 + target == real Acc@1 77.515 Acc@5 100.000 + target == fake Acc@1 78.839 Acc@5 100.000 + loss 0.449 + PR-AUC 0.9636 ROC-AUC 0.8772 EER 21.59% (thr=0.5068) +Accuracy of the network on the 89729 test images: 78.5% +[rank0]:[W726 06:19:49.557421202 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 06:20:08.847000 1206182 torch/distributed/run.py:766] +W0726 06:20:08.847000 1206182 torch/distributed/run.py:766] ***************************************** +W0726 06:20:08.847000 1206182 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:20:08.847000 1206182 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:20:37.633469420 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 06:20:38.583076521 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +NCCL version 2.26.2+cuda12.2 +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:20:38.834732156 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:20:38.869122465 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:20:38.949815279 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W726 06:20:38.951178502 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:20:38.166692342 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:20:38.172749529 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n45_seed9_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed9_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:26:47 acc1_pos: 37.240 acc5_pos: 100.000 loss: 0.8548 (0.8548) acc1: 37.2396 (37.2396) acc5: 100.0000 (100.0000) time: 6.7061 data: 5.2825 max mem: 3531 +Test: [ 10/2387] eta: 0:27:24 acc1_pos: 10.156 acc5_pos: 100.000 loss: 0.8677 (0.9359) acc1: 34.6354 (31.7708) acc5: 100.0000 (100.0000) time: 0.6918 data: 0.4807 max mem: 3531 +Test: [ 20/2387] eta: 0:16:17 acc1_pos: 11.979 acc5_pos: 100.000 loss: 0.9905 (1.0033) acc1: 21.3542 (26.9965) acc5: 100.0000 (100.0000) time: 0.0981 data: 0.0044 max mem: 3531 +Test: [ 30/2387] eta: 0:14:09 acc1_pos: 29.688 acc5_pos: 100.000 loss: 1.0579 (0.9990) acc1: 17.7083 (27.2177) acc5: 100.0000 (100.0000) time: 0.1779 data: 0.0818 max mem: 3531 +Test: [ 40/2387] eta: 0:13:25 acc1_pos: 38.542 acc5_pos: 100.000 loss: 0.8722 (0.9703) acc1: 29.6875 (28.9253) acc5: 100.0000 (100.0000) time: 0.2699 data: 0.1750 max mem: 3531 +Test: [ 50/2387] eta: 0:13:02 acc1_pos: 33.854 acc5_pos: 100.000 loss: 0.8693 (0.9644) acc1: 33.8542 (29.7437) acc5: 100.0000 (100.0000) time: 0.2949 data: 0.1994 max mem: 3531 +Test: [ 60/2387] eta: 0:12:31 acc1_pos: 42.708 acc5_pos: 100.000 loss: 0.9226 (0.9649) acc1: 31.7708 (30.0162) acc5: 100.0000 (100.0000) time: 0.2813 data: 0.1861 max mem: 3531 +Test: [ 70/2387] eta: 0:11:46 acc1_pos: 33.594 acc5_pos: 100.000 loss: 1.0076 (0.9769) acc1: 27.0833 (29.2950) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1362 max mem: 3531 +Test: [ 80/2387] eta: 0:11:34 acc1_pos: 71.875 acc5_pos: 100.000 loss: 0.9759 (0.9495) acc1: 32.5521 (32.0087) acc5: 100.0000 (100.0000) time: 0.2342 data: 0.1414 max mem: 3531 +Test: [ 90/2387] eta: 0:11:13 acc1_pos: 57.552 acc5_pos: 100.000 loss: 0.7030 (0.9162) acc1: 53.6458 (35.0532) acc5: 100.0000 (100.0000) time: 0.2516 data: 0.1591 max mem: 3531 +Test: [ 100/2387] eta: 0:10:52 acc1_pos: 58.594 acc5_pos: 100.000 loss: 0.6372 (0.8967) acc1: 58.3333 (37.0849) acc5: 100.0000 (100.0000) time: 0.2211 data: 0.1276 max mem: 3531 +Test: [ 110/2387] eta: 0:10:34 acc1_pos: 64.323 acc5_pos: 100.000 loss: 0.6707 (0.8750) acc1: 58.5938 (39.4379) acc5: 100.0000 (100.0000) time: 0.2108 data: 0.1181 max mem: 3531 +Test: [ 120/2387] eta: 0:10:23 acc1_pos: 42.188 acc5_pos: 100.000 loss: 0.6712 (0.8617) acc1: 58.8542 (40.5906) acc5: 100.0000 (100.0000) time: 0.2239 data: 0.1334 max mem: 3531 +Test: [ 130/2387] eta: 0:10:12 acc1_pos: 55.729 acc5_pos: 100.000 loss: 0.6963 (0.8477) acc1: 57.8125 (42.1358) acc5: 100.0000 (100.0000) time: 0.2327 data: 0.1418 max mem: 3531 +Test: [ 140/2387] eta: 0:10:00 acc1_pos: 63.281 acc5_pos: 100.000 loss: 0.6690 (0.8358) acc1: 59.8958 (43.2292) acc5: 100.0000 (100.0000) time: 0.2205 data: 0.1278 max mem: 3531 +Test: [ 150/2387] eta: 0:10:10 acc1_pos: 45.833 acc5_pos: 100.000 loss: 0.6437 (0.8226) acc1: 60.9375 (44.5088) acc5: 100.0000 (100.0000) time: 0.2813 data: 0.1876 max mem: 3531 +Test: [ 160/2387] eta: 0:09:52 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.5972 (0.8035) acc1: 68.7500 (46.3720) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2584 data: 0.1655 max mem: 3531 +Test: [ 170/2387] eta: 0:09:33 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.2500 (0.7705) acc1: 96.6146 (49.3787) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1532 data: 0.0590 max mem: 3531 +Test: [ 180/2387] eta: 0:09:17 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.2361 (0.7412) acc1: 97.9167 (52.0502) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1449 data: 0.0475 max mem: 3531 +Test: [ 190/2387] eta: 0:09:04 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.2361 (0.7148) acc1: 97.9167 (54.4394) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1561 data: 0.0601 max mem: 3531 +Test: [ 200/2387] eta: 0:08:56 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.2344 (0.6911) acc1: 97.6562 (56.6024) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1759 data: 0.0835 max mem: 3531 +Test: [ 210/2387] eta: 0:08:45 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.2413 (0.6699) acc1: 97.6562 (58.5469) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1791 data: 0.0864 max mem: 3532 +Test: [ 220/2387] eta: 0:08:38 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.2422 (0.6505) acc1: 97.6562 (60.3271) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1789 data: 0.0843 max mem: 3532 +Test: [ 230/2387] eta: 0:08:32 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.2400 (0.6327) acc1: 97.6562 (61.9465) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1945 data: 0.0997 max mem: 3532 +Test: [ 240/2387] eta: 0:08:27 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.2466 (0.6200) acc1: 96.6146 (63.2542) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.2049 data: 0.1099 max mem: 3532 +Test: [ 250/2387] eta: 0:08:20 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.2320 (0.6001) acc1: 95.5729 (64.6788) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1951 data: 0.1002 max mem: 3532 +Test: [ 260/2387] eta: 0:08:08 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.0833 (0.5800) acc1: 100.0000 (66.0321) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1499 data: 0.0549 max mem: 3532 +Test: [ 270/2387] eta: 0:08:02 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.0749 (0.5614) acc1: 100.0000 (67.2846) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1503 data: 0.0566 max mem: 3532 +Test: [ 280/2387] eta: 0:07:58 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.0761 (0.5444) acc1: 100.0000 (68.4488) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1966 data: 0.1048 max mem: 3532 +Test: [ 290/2387] eta: 0:07:52 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.0843 (0.5287) acc1: 100.0000 (69.5330) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1905 data: 0.0985 max mem: 3532 +Test: [ 300/2387] eta: 0:07:44 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.0923 (0.5143) acc1: 100.0000 (70.5444) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1532 data: 0.0610 max mem: 3532 +Test: [ 310/2387] eta: 0:07:41 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1017 (0.5060) acc1: 100.0000 (71.3960) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1742 data: 0.0807 max mem: 3532 +Test: [ 320/2387] eta: 0:07:38 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.4390 (0.5040) acc1: 93.4896 (72.0567) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.2149 data: 0.1218 max mem: 3532 +Test: [ 330/2387] eta: 0:07:32 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.4398 (0.5020) acc1: 92.9688 (72.7058) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1844 data: 0.0898 max mem: 3532 +Test: [ 340/2387] eta: 0:07:26 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.4365 (0.5000) acc1: 92.9688 (73.3000) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1549 data: 0.0603 max mem: 3532 +Test: [ 350/2387] eta: 0:07:20 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.4251 (0.4928) acc1: 95.0521 (74.0377) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1617 data: 0.0689 max mem: 3532 +Test: [ 360/2387] eta: 0:07:15 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.2137 (0.4848) acc1: 99.2188 (74.7266) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1610 data: 0.0659 max mem: 3532 +Test: [ 370/2387] eta: 0:07:10 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.2044 (0.4773) acc1: 98.9583 (75.3755) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1582 data: 0.0636 max mem: 3532 +Test: [ 380/2387] eta: 0:07:05 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.2045 (0.4701) acc1: 98.9583 (75.9911) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1639 data: 0.0707 max mem: 3532 +Test: [ 390/2387] eta: 0:07:01 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.2045 (0.4633) acc1: 98.6979 (76.5712) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1699 data: 0.0757 max mem: 3532 +Test: [ 400/2387] eta: 0:06:57 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.2036 (0.4568) acc1: 98.6979 (77.1242) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1699 data: 0.0742 max mem: 3532 +Test: [ 410/2387] eta: 0:06:52 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.2024 (0.4506) acc1: 98.9583 (77.6587) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1617 data: 0.0669 max mem: 3532 +Test: [ 420/2387] eta: 0:06:49 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.2021 (0.4447) acc1: 98.9583 (78.1621) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1699 data: 0.0748 max mem: 3532 +Test: [ 430/2387] eta: 0:06:45 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.2071 (0.4404) acc1: 98.6979 (78.5854) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1781 data: 0.0831 max mem: 3532 +Test: [ 440/2387] eta: 0:06:41 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.2120 (0.4341) acc1: 96.0938 (79.0303) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1710 data: 0.0789 max mem: 3533 +Test: [ 450/2387] eta: 0:06:38 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.0770 (0.4262) acc1: 100.0000 (79.4952) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1707 data: 0.0766 max mem: 3533 +Test: [ 460/2387] eta: 0:06:34 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.0760 (0.4186) acc1: 100.0000 (79.9400) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1707 data: 0.0766 max mem: 3533 +Test: [ 470/2387] eta: 0:06:32 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.0759 (0.4113) acc1: 100.0000 (80.3659) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1841 data: 0.0922 max mem: 3533 +Test: [ 480/2387] eta: 0:06:27 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.0768 (0.4044) acc1: 100.0000 (80.7741) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1682 data: 0.0748 max mem: 3533 +Test: [ 490/2387] eta: 0:06:25 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.0771 (0.3977) acc1: 100.0000 (81.1657) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1683 data: 0.0761 max mem: 3533 +Test: [ 500/2387] eta: 0:06:24 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.0774 (0.3913) acc1: 100.0000 (81.5411) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2149 data: 0.1245 max mem: 3533 +Test: [ 510/2387] eta: 0:06:21 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.0777 (0.3857) acc1: 100.0000 (81.9023) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2044 data: 0.1132 max mem: 3533 +Test: [ 520/2387] eta: 0:06:18 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1438 (0.3814) acc1: 100.0000 (82.2497) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1764 data: 0.0852 max mem: 3533 +Test: [ 530/2387] eta: 0:06:16 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1442 (0.3769) acc1: 100.0000 (82.5840) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1843 data: 0.0937 max mem: 3533 +Test: [ 540/2387] eta: 0:06:13 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1437 (0.3726) acc1: 100.0000 (82.9059) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1811 data: 0.0898 max mem: 3533 +Test: [ 550/2387] eta: 0:06:09 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1433 (0.3685) acc1: 100.0000 (83.2161) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1609 data: 0.0695 max mem: 3533 +Test: [ 560/2387] eta: 0:06:06 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1428 (0.3644) acc1: 100.0000 (83.5153) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1628 data: 0.0684 max mem: 3533 +Test: [ 570/2387] eta: 0:06:03 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1426 (0.3605) acc1: 100.0000 (83.8040) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1726 data: 0.0779 max mem: 3533 +Test: [ 580/2387] eta: 0:06:00 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1425 (0.3568) acc1: 100.0000 (84.0828) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1701 data: 0.0778 max mem: 3533 +Test: [ 590/2387] eta: 0:05:57 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1425 (0.3532) acc1: 100.0000 (84.3521) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1698 data: 0.0751 max mem: 3533 +Test: [ 600/2387] eta: 0:05:55 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1483 (0.3502) acc1: 100.0000 (84.6125) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1793 data: 0.0847 max mem: 3533 +Test: [ 610/2387] eta: 0:05:53 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1713 (0.3473) acc1: 100.0000 (84.8575) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1894 data: 0.0947 max mem: 3533 +Test: [ 620/2387] eta: 0:05:51 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1702 (0.3444) acc1: 99.7396 (85.0954) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1960 data: 0.1001 max mem: 3533 +Test: [ 630/2387] eta: 0:05:48 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1705 (0.3417) acc1: 99.7396 (85.3246) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1864 data: 0.0930 max mem: 3533 +Test: [ 640/2387] eta: 0:05:45 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1732 (0.3391) acc1: 99.4792 (85.5434) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1734 data: 0.0782 max mem: 3533 +Test: [ 650/2387] eta: 0:05:43 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1732 (0.3365) acc1: 99.4792 (85.7607) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1799 data: 0.0849 max mem: 3533 +Test: [ 660/2387] eta: 0:05:41 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1720 (0.3340) acc1: 99.7396 (85.9698) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1854 data: 0.0925 max mem: 3533 +Test: [ 670/2387] eta: 0:05:39 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1720 (0.3316) acc1: 99.4792 (86.1735) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1890 data: 0.0955 max mem: 3534 +Test: [ 680/2387] eta: 0:05:36 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1726 (0.3301) acc1: 99.7396 (86.3734) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1908 data: 0.0974 max mem: 3534 +Test: [ 690/2387] eta: 0:05:35 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.2596 (0.3309) acc1: 99.7396 (86.3400) acc5: 100.0000 (100.0000) acc1_neg: 75.521 acc5_neg: 100.000 time: 0.1989 data: 0.1058 max mem: 3534 +Test: [ 700/2387] eta: 0:05:33 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.4281 (0.3323) acc1: 79.1667 (86.2291) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.2041 data: 0.1099 max mem: 3534 +Test: [ 710/2387] eta: 0:05:31 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.4330 (0.3338) acc1: 79.1667 (86.1203) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.1975 data: 0.1034 max mem: 3534 +Test: [ 720/2387] eta: 0:05:28 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.4304 (0.3351) acc1: 78.6458 (86.0199) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.1843 data: 0.0921 max mem: 3534 +Test: [ 730/2387] eta: 0:05:26 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.4333 (0.3365) acc1: 78.3854 (85.9179) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.1787 data: 0.0874 max mem: 3534 +Test: [ 740/2387] eta: 0:05:24 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.4338 (0.3378) acc1: 78.1250 (85.8205) acc5: 100.0000 (100.0000) acc1_neg: 75.521 acc5_neg: 100.000 time: 0.1876 data: 0.0963 max mem: 3534 +Test: [ 750/2387] eta: 0:05:23 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.4273 (0.3390) acc1: 78.3854 (85.7256) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.2133 data: 0.1208 max mem: 3534 +Test: [ 760/2387] eta: 0:05:20 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.4255 (0.3400) acc1: 79.1667 (85.6586) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1951 data: 0.1022 max mem: 3534 +Test: [ 770/2387] eta: 0:05:19 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.3804 (0.3395) acc1: 83.0729 (85.7119) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1949 data: 0.0993 max mem: 3534 +Test: [ 780/2387] eta: 0:05:16 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.2077 (0.3375) acc1: 96.8750 (85.8718) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2021 data: 0.1041 max mem: 3534 +Test: [ 790/2387] eta: 0:05:13 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1709 (0.3354) acc1: 98.9583 (85.9915) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.1649 data: 0.0686 max mem: 3534 +Test: [ 800/2387] eta: 0:05:12 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1433 (0.3330) acc1: 99.2188 (86.1622) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1859 data: 0.0920 max mem: 3534 +Test: [ 810/2387] eta: 0:05:10 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1667 (0.3313) acc1: 99.4792 (86.3084) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2149 data: 0.1221 max mem: 3534 +Test: [ 820/2387] eta: 0:05:07 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1891 (0.3299) acc1: 97.3958 (86.4263) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1764 data: 0.0821 max mem: 3534 +Test: [ 830/2387] eta: 0:05:05 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.2611 (0.3303) acc1: 91.9271 (86.4314) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1568 data: 0.0633 max mem: 3534 +Test: [ 840/2387] eta: 0:05:03 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.2808 (0.3294) acc1: 91.9271 (86.5506) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1877 data: 0.0966 max mem: 3534 +Test: [ 850/2387] eta: 0:05:00 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.2444 (0.3282) acc1: 97.3958 (86.6882) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1835 data: 0.0919 max mem: 3534 +Test: [ 860/2387] eta: 0:04:58 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.2121 (0.3268) acc1: 99.2188 (86.8295) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1648 data: 0.0725 max mem: 3534 +Test: [ 870/2387] eta: 0:04:56 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.2083 (0.3256) acc1: 99.2188 (86.9606) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1959 data: 0.1045 max mem: 3534 +Test: [ 880/2387] eta: 0:04:55 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.2400 (0.3247) acc1: 98.4375 (87.0841) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2198 data: 0.1261 max mem: 3534 +Test: [ 890/2387] eta: 0:04:52 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.2461 (0.3239) acc1: 98.4375 (87.2042) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1805 data: 0.0854 max mem: 3534 +Test: [ 900/2387] eta: 0:04:51 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.2845 (0.3278) acc1: 95.5729 (86.9309) acc5: 100.0000 (100.0000) acc1_neg: 57.031 acc5_neg: 100.000 time: 0.2000 data: 0.1072 max mem: 3535 +Test: [ 910/2387] eta: 0:04:49 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.7735 (0.3327) acc1: 53.6458 (86.5710) acc5: 100.0000 (100.0000) acc1_neg: 53.385 acc5_neg: 100.000 time: 0.2134 data: 0.1200 max mem: 3535 +Test: [ 920/2387] eta: 0:04:47 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.7786 (0.3375) acc1: 53.3854 (86.2140) acc5: 100.0000 (100.0000) acc1_neg: 51.823 acc5_neg: 100.000 time: 0.1805 data: 0.0873 max mem: 3535 +Test: [ 930/2387] eta: 0:04:45 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.7751 (0.3422) acc1: 53.3854 (85.8656) acc5: 100.0000 (100.0000) acc1_neg: 49.219 acc5_neg: 100.000 time: 0.1904 data: 0.0983 max mem: 3535 +Test: [ 940/2387] eta: 0:04:43 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.7851 (0.3468) acc1: 53.1250 (85.5171) acc5: 100.0000 (100.0000) acc1_neg: 52.344 acc5_neg: 100.000 time: 0.1945 data: 0.1023 max mem: 3535 +Test: [ 950/2387] eta: 0:04:41 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.7738 (0.3514) acc1: 52.3438 (85.1661) acc5: 100.0000 (100.0000) acc1_neg: 54.688 acc5_neg: 100.000 time: 0.2004 data: 0.1086 max mem: 3535 +Test: [ 960/2387] eta: 0:04:39 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.7606 (0.3557) acc1: 52.8646 (84.8422) acc5: 100.0000 (100.0000) acc1_neg: 56.510 acc5_neg: 100.000 time: 0.1971 data: 0.1039 max mem: 3535 +Test: [ 970/2387] eta: 0:04:37 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.7711 (0.3601) acc1: 52.8646 (84.5096) acc5: 100.0000 (100.0000) acc1_neg: 54.167 acc5_neg: 100.000 time: 0.1938 data: 0.1006 max mem: 3535 +Test: [ 980/2387] eta: 0:04:35 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.7512 (0.3607) acc1: 54.1667 (84.4852) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.1944 data: 0.1030 max mem: 3535 +Test: [ 990/2387] eta: 0:04:33 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.2108 (0.3591) acc1: 98.4375 (84.6020) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1846 data: 0.0937 max mem: 3535 +Test: [1000/2387] eta: 0:04:31 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1517 (0.3571) acc1: 99.4792 (84.7364) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1880 data: 0.0968 max mem: 3535 +Test: [1010/2387] eta: 0:04:28 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1756 (0.3558) acc1: 98.9583 (84.8459) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1722 data: 0.0810 max mem: 3535 +Test: [1020/2387] eta: 0:04:26 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.2216 (0.3549) acc1: 96.6146 (84.9226) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1602 data: 0.0688 max mem: 3535 +Test: [1030/2387] eta: 0:04:23 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1832 (0.3531) acc1: 97.9167 (85.0562) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1645 data: 0.0724 max mem: 3535 +Test: [1040/2387] eta: 0:04:21 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1422 (0.3511) acc1: 99.7396 (85.1943) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1612 data: 0.0697 max mem: 3535 +Test: [1050/2387] eta: 0:04:19 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1366 (0.3491) acc1: 99.7396 (85.3317) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1576 data: 0.0661 max mem: 3535 +Test: [1060/2387] eta: 0:04:16 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1575 (0.3474) acc1: 99.7396 (85.4611) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1650 data: 0.0726 max mem: 3535 +Test: [1070/2387] eta: 0:04:14 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1729 (0.3462) acc1: 99.2188 (85.5655) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1703 data: 0.0774 max mem: 3535 +Test: [1080/2387] eta: 0:04:12 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.2554 (0.3460) acc1: 95.0521 (85.6108) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1708 data: 0.0784 max mem: 3535 +Test: [1090/2387] eta: 0:04:10 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.2554 (0.3449) acc1: 94.0104 (85.7098) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1703 data: 0.0788 max mem: 3535 +Test: [1100/2387] eta: 0:04:07 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.2013 (0.3435) acc1: 98.1771 (85.8188) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1629 data: 0.0722 max mem: 3535 +Test: [1110/2387] eta: 0:04:06 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1946 (0.3424) acc1: 97.6562 (85.9023) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1791 data: 0.0865 max mem: 3535 +Test: [1120/2387] eta: 0:04:04 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1948 (0.3413) acc1: 97.6562 (85.9984) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1990 data: 0.1061 max mem: 3536 +Test: [1130/2387] eta: 0:04:02 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.2208 (0.3403) acc1: 96.8750 (86.0920) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1981 data: 0.1061 max mem: 3536 +Test: [1140/2387] eta: 0:04:00 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.2528 (0.3399) acc1: 95.5729 (86.1566) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.2006 data: 0.1080 max mem: 3536 +Test: [1150/2387] eta: 0:03:58 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.2782 (0.3391) acc1: 95.0521 (86.2382) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1982 data: 0.1047 max mem: 3536 +Test: [1160/2387] eta: 0:03:56 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1639 (0.3375) acc1: 99.2188 (86.3462) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1891 data: 0.0957 max mem: 3536 +Test: [1170/2387] eta: 0:03:54 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1430 (0.3358) acc1: 99.7396 (86.4608) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1895 data: 0.0975 max mem: 3536 +Test: [1180/2387] eta: 0:03:52 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1399 (0.3342) acc1: 99.7396 (86.5717) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1949 data: 0.1021 max mem: 3536 +Test: [1190/2387] eta: 0:03:50 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1592 (0.3329) acc1: 99.7396 (86.6811) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1953 data: 0.1022 max mem: 3536 +Test: [1200/2387] eta: 0:03:49 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1628 (0.3314) acc1: 99.7396 (86.7897) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2008 data: 0.1092 max mem: 3536 +Test: [1210/2387] eta: 0:03:47 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1606 (0.3301) acc1: 99.7396 (86.8970) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2035 data: 0.1119 max mem: 3536 +Test: [1220/2387] eta: 0:03:45 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1064 (0.3282) acc1: 100.0000 (87.0028) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2020 data: 0.1085 max mem: 3536 +Test: [1230/2387] eta: 0:03:43 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.0897 (0.3263) acc1: 100.0000 (87.1084) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2052 data: 0.1109 max mem: 3536 +Test: [1240/2387] eta: 0:03:41 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.0829 (0.3243) acc1: 100.0000 (87.2119) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2073 data: 0.1143 max mem: 3536 +Test: [1250/2387] eta: 0:03:40 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.0865 (0.3225) acc1: 100.0000 (87.3137) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2164 data: 0.1251 max mem: 3536 +Test: [1260/2387] eta: 0:03:38 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.0968 (0.3207) acc1: 100.0000 (87.4133) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2004 data: 0.1099 max mem: 3536 +Test: [1270/2387] eta: 0:03:35 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.0984 (0.3191) acc1: 100.0000 (87.5115) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1730 data: 0.0815 max mem: 3536 +Test: [1280/2387] eta: 0:03:33 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1117 (0.3175) acc1: 100.0000 (87.6059) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1690 data: 0.0774 max mem: 3536 +Test: [1290/2387] eta: 0:03:31 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1161 (0.3160) acc1: 99.7396 (87.6888) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1714 data: 0.0806 max mem: 3536 +Test: [1300/2387] eta: 0:03:29 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1127 (0.3144) acc1: 99.7396 (87.7772) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1750 data: 0.0835 max mem: 3536 +Test: [1310/2387] eta: 0:03:27 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1177 (0.3130) acc1: 99.7396 (87.8605) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1759 data: 0.0837 max mem: 3536 +Test: [1320/2387] eta: 0:03:25 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1503 (0.3119) acc1: 99.2188 (87.9398) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1847 data: 0.0893 max mem: 3536 +Test: [1330/2387] eta: 0:03:23 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1741 (0.3113) acc1: 97.9167 (87.9964) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1860 data: 0.0909 max mem: 3536 +Test: [1340/2387] eta: 0:03:21 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.2600 (0.3118) acc1: 92.4479 (87.9568) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1856 data: 0.0945 max mem: 3536 +Test: [1350/2387] eta: 0:03:19 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.0639 (0.3099) acc1: 100.0000 (88.0459) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2022 data: 0.1119 max mem: 3537 +Test: [1360/2387] eta: 0:03:17 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.0517 (0.3079) acc1: 100.0000 (88.1337) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2044 data: 0.1137 max mem: 3537 +Test: [1370/2387] eta: 0:03:16 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.0451 (0.3060) acc1: 100.0000 (88.2203) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1998 data: 0.1079 max mem: 3537 +Test: [1380/2387] eta: 0:03:14 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.0456 (0.3042) acc1: 100.0000 (88.3056) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1946 data: 0.1010 max mem: 3537 +Test: [1390/2387] eta: 0:03:12 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.0503 (0.3023) acc1: 100.0000 (88.3896) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1957 data: 0.1021 max mem: 3537 +Test: [1400/2387] eta: 0:03:10 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.0522 (0.3006) acc1: 100.0000 (88.4725) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1921 data: 0.1002 max mem: 3537 +Test: [1410/2387] eta: 0:03:08 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.0634 (0.2997) acc1: 100.0000 (88.5334) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1842 data: 0.0930 max mem: 3537 +Test: [1420/2387] eta: 0:03:06 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1924 (0.2991) acc1: 96.6146 (88.5840) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1827 data: 0.0910 max mem: 3537 +Test: [1430/2387] eta: 0:03:04 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1937 (0.2984) acc1: 98.1771 (88.6456) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1843 data: 0.0919 max mem: 3537 +Test: [1440/2387] eta: 0:03:02 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.2130 (0.2980) acc1: 97.1354 (88.7009) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.1758 data: 0.0820 max mem: 3537 +Test: [1450/2387] eta: 0:03:00 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.2453 (0.2979) acc1: 96.0938 (88.7251) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1808 data: 0.0859 max mem: 3537 +Test: [1460/2387] eta: 0:02:58 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1745 (0.2970) acc1: 97.1354 (88.7905) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1916 data: 0.0988 max mem: 3537 +Test: [1470/2387] eta: 0:02:56 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1325 (0.2958) acc1: 99.4792 (88.8655) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1788 data: 0.0869 max mem: 3537 +Test: [1480/2387] eta: 0:02:54 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1105 (0.2946) acc1: 99.7396 (88.9352) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1722 data: 0.0804 max mem: 3537 +Test: [1490/2387] eta: 0:02:52 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1225 (0.2936) acc1: 99.4792 (89.0049) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1749 data: 0.0838 max mem: 3537 +Test: [1500/2387] eta: 0:02:50 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1312 (0.2926) acc1: 99.4792 (89.0734) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1826 data: 0.0902 max mem: 3537 +Test: [1510/2387] eta: 0:02:48 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1570 (0.2919) acc1: 99.7396 (89.1430) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1859 data: 0.0936 max mem: 3537 +Test: [1520/2387] eta: 0:02:46 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1997 (0.2914) acc1: 99.4792 (89.2007) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1921 data: 0.1009 max mem: 3537 +Test: [1530/2387] eta: 0:02:44 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.2077 (0.2908) acc1: 98.6979 (89.2647) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1946 data: 0.1034 max mem: 3537 +Test: [1540/2387] eta: 0:02:42 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1789 (0.2901) acc1: 99.4792 (89.3278) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1915 data: 0.1002 max mem: 3537 +Test: [1550/2387] eta: 0:02:40 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1789 (0.2895) acc1: 99.4792 (89.3887) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1938 data: 0.1017 max mem: 3537 +Test: [1560/2387] eta: 0:02:38 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.2012 (0.2890) acc1: 99.2188 (89.4479) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1984 data: 0.1064 max mem: 3537 +Test: [1570/2387] eta: 0:02:37 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.2083 (0.2885) acc1: 99.2188 (89.5056) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2066 data: 0.1158 max mem: 3537 +Test: [1580/2387] eta: 0:02:35 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.2130 (0.2882) acc1: 98.6979 (89.5557) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1965 data: 0.1044 max mem: 3538 +Test: [1590/2387] eta: 0:02:33 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1995 (0.2874) acc1: 97.6562 (89.6113) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2072 data: 0.1149 max mem: 3538 +Test: [1600/2387] eta: 0:02:31 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1661 (0.2868) acc1: 98.6979 (89.6510) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2162 data: 0.1252 max mem: 3538 +Test: [1610/2387] eta: 0:02:29 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1572 (0.2860) acc1: 99.4792 (89.7106) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1876 data: 0.0966 max mem: 3538 +Test: [1620/2387] eta: 0:02:27 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1807 (0.2856) acc1: 96.8750 (89.7429) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1843 data: 0.0915 max mem: 3538 +Test: [1630/2387] eta: 0:02:25 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.2287 (0.2855) acc1: 94.7917 (89.7668) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1941 data: 0.1012 max mem: 3538 +Test: [1640/2387] eta: 0:02:23 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.3097 (0.2859) acc1: 91.1458 (89.7582) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1925 data: 0.1009 max mem: 3538 +Test: [1650/2387] eta: 0:02:21 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.3410 (0.2861) acc1: 88.5417 (89.7564) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2103 data: 0.1176 max mem: 3538 +Test: [1660/2387] eta: 0:02:19 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.3145 (0.2864) acc1: 90.1042 (89.7481) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.2125 data: 0.1199 max mem: 3538 +Test: [1670/2387] eta: 0:02:18 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.3224 (0.2866) acc1: 89.8438 (89.7482) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1923 data: 0.0995 max mem: 3538 +Test: [1680/2387] eta: 0:02:16 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.3224 (0.2868) acc1: 89.8438 (89.7534) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.2005 data: 0.1078 max mem: 3538 +Test: [1690/2387] eta: 0:02:14 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.3180 (0.2871) acc1: 89.8438 (89.7477) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.2044 data: 0.1123 max mem: 3538 +Test: [1700/2387] eta: 0:02:12 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.3180 (0.2870) acc1: 89.8438 (89.7652) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1981 data: 0.1065 max mem: 3538 +Test: [1710/2387] eta: 0:02:10 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1174 (0.2860) acc1: 99.4792 (89.8243) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2015 data: 0.1106 max mem: 3538 +Test: [1720/2387] eta: 0:02:08 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1070 (0.2849) acc1: 100.0000 (89.8828) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1895 data: 0.0980 max mem: 3538 +Test: [1730/2387] eta: 0:02:06 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1055 (0.2839) acc1: 100.0000 (89.9400) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1849 data: 0.0927 max mem: 3538 +Test: [1740/2387] eta: 0:02:04 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1032 (0.2829) acc1: 100.0000 (89.9960) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2005 data: 0.1077 max mem: 3538 +Test: [1750/2387] eta: 0:02:02 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1102 (0.2820) acc1: 100.0000 (90.0529) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2049 data: 0.1129 max mem: 3538 +Test: [1760/2387] eta: 0:02:00 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1165 (0.2812) acc1: 100.0000 (90.1065) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2062 data: 0.1153 max mem: 3538 +Test: [1770/2387] eta: 0:01:59 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1537 (0.2805) acc1: 98.9583 (90.1536) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2049 data: 0.1142 max mem: 3538 +Test: [1780/2387] eta: 0:01:57 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1482 (0.2797) acc1: 99.4792 (90.2074) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2069 data: 0.1161 max mem: 3538 +Test: [1790/2387] eta: 0:01:55 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1349 (0.2789) acc1: 100.0000 (90.2603) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2005 data: 0.1091 max mem: 3538 +Test: [1800/2387] eta: 0:01:53 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1287 (0.2780) acc1: 99.7396 (90.3127) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1917 data: 0.0990 max mem: 3538 +Test: [1810/2387] eta: 0:01:51 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1289 (0.2772) acc1: 99.7396 (90.3644) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1943 data: 0.1025 max mem: 3539 +Test: [1820/2387] eta: 0:01:49 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1329 (0.2765) acc1: 99.7396 (90.4138) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1960 data: 0.1041 max mem: 3539 +Test: [1830/2387] eta: 0:01:47 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1411 (0.2758) acc1: 99.4792 (90.4629) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2048 data: 0.1103 max mem: 3539 +Test: [1840/2387] eta: 0:01:45 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1243 (0.2749) acc1: 100.0000 (90.5147) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2088 data: 0.1142 max mem: 3539 +Test: [1850/2387] eta: 0:01:43 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1221 (0.2741) acc1: 100.0000 (90.5659) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2119 data: 0.1189 max mem: 3539 +Test: [1860/2387] eta: 0:01:41 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1215 (0.2733) acc1: 100.0000 (90.6166) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2010 data: 0.1097 max mem: 3539 +Test: [1870/2387] eta: 0:01:39 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1231 (0.2725) acc1: 100.0000 (90.6665) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1853 data: 0.0925 max mem: 3539 +Test: [1880/2387] eta: 0:01:38 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1205 (0.2717) acc1: 100.0000 (90.7161) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1992 data: 0.1046 max mem: 3539 +Test: [1890/2387] eta: 0:01:36 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1217 (0.2709) acc1: 100.0000 (90.7651) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2101 data: 0.1175 max mem: 3539 +Test: [1900/2387] eta: 0:01:34 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1235 (0.2701) acc1: 100.0000 (90.8136) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2276 data: 0.1366 max mem: 3539 +Test: [1910/2387] eta: 0:01:32 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1584 (0.2696) acc1: 100.0000 (90.8605) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1981 data: 0.1046 max mem: 3539 +Test: [1920/2387] eta: 0:01:30 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1705 (0.2695) acc1: 99.7396 (90.8748) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1870 data: 0.0926 max mem: 3539 +Test: [1930/2387] eta: 0:01:28 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1784 (0.2689) acc1: 99.7396 (90.9201) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2156 data: 0.1240 max mem: 3539 +Test: [1940/2387] eta: 0:01:26 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1468 (0.2685) acc1: 99.7396 (90.9312) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2079 data: 0.1163 max mem: 3539 +Test: [1950/2387] eta: 0:01:24 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1705 (0.2681) acc1: 99.4792 (90.9702) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2034 data: 0.1108 max mem: 3539 +Test: [1960/2387] eta: 0:01:22 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1889 (0.2679) acc1: 98.6979 (90.9923) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2016 data: 0.1094 max mem: 3539 +Test: [1970/2387] eta: 0:01:20 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.2739 (0.2684) acc1: 91.9271 (90.9631) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.1974 data: 0.1057 max mem: 3539 +Test: [1980/2387] eta: 0:01:18 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.3203 (0.2683) acc1: 88.8021 (90.9693) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1942 data: 0.1023 max mem: 3539 +Test: [1990/2387] eta: 0:01:16 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1513 (0.2678) acc1: 98.4375 (91.0081) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1928 data: 0.1011 max mem: 3539 +Test: [2000/2387] eta: 0:01:15 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1508 (0.2672) acc1: 98.6979 (91.0473) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1941 data: 0.1023 max mem: 3539 +Test: [2010/2387] eta: 0:01:13 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1508 (0.2666) acc1: 98.6979 (91.0845) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1958 data: 0.1048 max mem: 3539 +Test: [2020/2387] eta: 0:01:11 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1529 (0.2661) acc1: 98.6979 (91.1232) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1949 data: 0.1022 max mem: 3539 +Test: [2030/2387] eta: 0:01:09 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1514 (0.2655) acc1: 98.6979 (91.1602) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2385 data: 0.1429 max mem: 3540 +Test: [2040/2387] eta: 0:01:07 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1514 (0.2650) acc1: 98.9583 (91.1979) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2209 data: 0.1250 max mem: 3540 +Test: [2050/2387] eta: 0:01:05 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1516 (0.2644) acc1: 98.9583 (91.2353) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1649 data: 0.0691 max mem: 3540 +Test: [2060/2387] eta: 0:01:03 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1633 (0.2640) acc1: 98.6979 (91.2702) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1955 data: 0.1010 max mem: 3540 +Test: [2070/2387] eta: 0:01:01 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1687 (0.2635) acc1: 98.6979 (91.3070) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2359 data: 0.1427 max mem: 3540 +Test: [2080/2387] eta: 0:00:59 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1651 (0.2630) acc1: 98.9583 (91.3429) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2029 data: 0.1088 max mem: 3540 +Test: [2090/2387] eta: 0:00:57 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1642 (0.2626) acc1: 98.6979 (91.3787) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2039 data: 0.1083 max mem: 3540 +Test: [2100/2387] eta: 0:00:55 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1647 (0.2621) acc1: 98.9583 (91.4149) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2121 data: 0.1163 max mem: 3540 +Test: [2110/2387] eta: 0:00:53 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1657 (0.2616) acc1: 98.6979 (91.4496) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1758 data: 0.0812 max mem: 3540 +Test: [2120/2387] eta: 0:00:51 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1826 (0.2615) acc1: 98.1771 (91.4755) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2027 data: 0.1080 max mem: 3540 +Test: [2130/2387] eta: 0:00:50 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1839 (0.2610) acc1: 98.4375 (91.5079) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2160 data: 0.1226 max mem: 3540 +Test: [2140/2387] eta: 0:00:48 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1639 (0.2607) acc1: 98.6979 (91.5247) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.1026 max mem: 3540 +Test: [2150/2387] eta: 0:00:46 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1684 (0.2603) acc1: 98.9583 (91.5610) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1993 data: 0.1068 max mem: 3540 +Test: [2160/2387] eta: 0:00:44 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1843 (0.2600) acc1: 99.2188 (91.5875) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2018 data: 0.1097 max mem: 3540 +Test: [2170/2387] eta: 0:00:42 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.2195 (0.2602) acc1: 95.8333 (91.5925) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.2003 data: 0.1083 max mem: 3540 +Test: [2180/2387] eta: 0:00:40 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.3525 (0.2617) acc1: 86.4583 (91.4892) acc5: 100.0000 (100.0000) acc1_neg: 40.885 acc5_neg: 100.000 time: 0.1992 data: 0.1084 max mem: 3540 +Test: [2190/2387] eta: 0:00:38 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.3201 (0.2619) acc1: 88.5417 (91.4772) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2015 data: 0.1103 max mem: 3540 +Test: [2200/2387] eta: 0:00:36 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1726 (0.2615) acc1: 97.1354 (91.5104) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1901 data: 0.0975 max mem: 3540 +Test: [2210/2387] eta: 0:00:34 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1495 (0.2610) acc1: 99.7396 (91.5476) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1788 data: 0.0868 max mem: 3540 +Test: [2220/2387] eta: 0:00:32 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1444 (0.2605) acc1: 100.0000 (91.5822) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1743 data: 0.0831 max mem: 3540 +Test: [2230/2387] eta: 0:00:30 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1492 (0.2600) acc1: 99.7396 (91.6178) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1651 data: 0.0741 max mem: 3540 +Test: [2240/2387] eta: 0:00:28 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1577 (0.2596) acc1: 99.4792 (91.6496) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1630 data: 0.0722 max mem: 3540 +Test: [2250/2387] eta: 0:00:26 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1748 (0.2594) acc1: 98.9583 (91.6774) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1483 data: 0.0578 max mem: 3540 +Test: [2260/2387] eta: 0:00:24 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1920 (0.2591) acc1: 98.9583 (91.7106) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1435 data: 0.0526 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1641 (0.2585) acc1: 99.4792 (91.7456) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1371 data: 0.0464 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1419 (0.2582) acc1: 99.7396 (91.7634) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1311 data: 0.0405 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1284 (0.2576) acc1: 100.0000 (91.7991) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1348 data: 0.0439 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1408 (0.2572) acc1: 100.0000 (91.8264) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1346 data: 0.0442 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1952 (0.2571) acc1: 98.9583 (91.8499) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1345 data: 0.0443 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.2347 (0.2572) acc1: 95.5729 (91.8526) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1544 data: 0.0644 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.2573 (0.2573) acc1: 94.0104 (91.8636) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1398 data: 0.0499 max mem: 3541 +Test: [2340/2387] eta: 0:00:09 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.2045 (0.2570) acc1: 97.9167 (91.8884) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1255 data: 0.0355 max mem: 3541 +Test: [2350/2387] eta: 0:00:07 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.1877 (0.2570) acc1: 98.6979 (91.8926) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1493 data: 0.0594 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.2244 (0.2570) acc1: 96.8750 (91.9038) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1392 data: 0.0493 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.2747 (0.2572) acc1: 95.0521 (91.9015) acc5: 100.0000 (100.0000) acc1_neg: 78.646 acc5_neg: 100.000 time: 0.1219 data: 0.0320 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.3289 (0.2578) acc1: 86.9792 (91.8754) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1289 data: 0.0390 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 68.493 acc5_pos: 100.000 loss: 0.3605 (0.2578) acc1: 86.9792 (91.8721) acc5: 100.0000 (100.0000) acc1_neg: 95.349 acc5_neg: 100.000 time: 0.1223 data: 0.0352 max mem: 3541 +Test: Total time: 0:07:35 (0.1907 s / it) +* Overall Acc@1 91.872 Acc@5 100.000 + target == real Acc@1 45.169 Acc@5 100.000 + target == fake Acc@1 95.152 Acc@5 100.000 + loss 0.258 + PR-AUC 0.9930 ROC-AUC 0.9100 EER 15.97% (thr=0.7051) +Accuracy of the network on the 916267 test images: 91.9% +[rank0]:[W726 06:28:30.912533557 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 06:28:51.165000 1206815 torch/distributed/run.py:766] +W0726 06:28:51.165000 1206815 torch/distributed/run.py:766] ***************************************** +W0726 06:28:51.165000 1206815 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:28:51.165000 1206815 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:29:21.694305909 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:29:21.805027010 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +[rank4]:[W726 06:29:21.999046345 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:29:21.001785889 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:29:21.415661261 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +[rank0]:[W726 06:29:22.423683376 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:29:22.438312549 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:29:22.539091306 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n45_seed9_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed9_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:12:16 acc1_pos: 37.240 acc5_pos: 100.000 loss: 0.8548 (0.8548) acc1: 37.2396 (37.2396) acc5: 100.0000 (100.0000) time: 8.3874 data: 6.1530 max mem: 3531 +Test: [ 10/517] eta: 0:07:09 acc1_pos: 10.156 acc5_pos: 100.000 loss: 0.8677 (0.9359) acc1: 34.6354 (31.7708) acc5: 100.0000 (100.0000) time: 0.8481 data: 0.5598 max mem: 3531 +Test: [ 20/517] eta: 0:04:04 acc1_pos: 11.979 acc5_pos: 100.000 loss: 0.9905 (1.0033) acc1: 21.3542 (26.9965) acc5: 100.0000 (100.0000) time: 0.0970 data: 0.0005 max mem: 3531 +Test: [ 30/517] eta: 0:03:24 acc1_pos: 29.688 acc5_pos: 100.000 loss: 1.0579 (0.9990) acc1: 17.7083 (27.2177) acc5: 100.0000 (100.0000) time: 0.1849 data: 0.0877 max mem: 3531 +Test: [ 40/517] eta: 0:03:05 acc1_pos: 38.542 acc5_pos: 100.000 loss: 0.8722 (0.9703) acc1: 29.6875 (28.9253) acc5: 100.0000 (100.0000) time: 0.2799 data: 0.1848 max mem: 3531 +Test: [ 50/517] eta: 0:02:50 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.8550 (0.9394) acc1: 34.6354 (32.2610) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.2799 data: 0.1867 max mem: 3531 +Test: [ 60/517] eta: 0:02:41 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.4888 (0.8605) acc1: 80.4688 (41.0775) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.2799 data: 0.1861 max mem: 3531 +Test: [ 70/517] eta: 0:02:32 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.4700 (0.8031) acc1: 87.2396 (47.9240) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2799 data: 0.1838 max mem: 3531 +Test: [ 80/517] eta: 0:02:25 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.4535 (0.7581) acc1: 91.4062 (53.4240) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.2699 data: 0.1764 max mem: 3531 +Test: [ 90/517] eta: 0:02:19 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.4263 (0.7202) acc1: 90.3646 (57.4720) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.2749 data: 0.1821 max mem: 3531 +Test: [100/517] eta: 0:02:13 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3744 (0.6824) acc1: 93.4896 (61.2572) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2749 data: 0.1803 max mem: 3531 +Test: [110/517] eta: 0:02:11 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3129 (0.6491) acc1: 97.1354 (64.5059) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.3049 data: 0.2081 max mem: 3531 +Test: [120/517] eta: 0:02:05 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3553 (0.6266) acc1: 95.0521 (66.9443) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.2949 data: 0.1965 max mem: 3531 +Test: [130/517] eta: 0:02:01 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3739 (0.6072) acc1: 94.0104 (69.0045) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2699 data: 0.1732 max mem: 3531 +Test: [140/517] eta: 0:01:56 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3176 (0.5859) acc1: 98.6979 (71.1621) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2699 data: 0.1733 max mem: 3531 +Test: [150/517] eta: 0:01:52 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3026 (0.5673) acc1: 99.4792 (73.0253) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2599 data: 0.1628 max mem: 3531 +Test: [160/517] eta: 0:01:48 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3127 (0.5521) acc1: 98.4375 (74.4339) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2699 data: 0.1733 max mem: 3531 +Test: [170/517] eta: 0:01:45 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3169 (0.5366) acc1: 96.8750 (75.8178) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2699 data: 0.1735 max mem: 3531 +Test: [180/517] eta: 0:01:41 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.2835 (0.5226) acc1: 96.8750 (76.9582) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2749 data: 0.1778 max mem: 3531 +Test: [190/517] eta: 0:01:38 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.2910 (0.5107) acc1: 98.4375 (78.1018) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2799 data: 0.1833 max mem: 3531 +Test: [200/517] eta: 0:01:34 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.2853 (0.4987) acc1: 99.4792 (79.1693) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2749 data: 0.1786 max mem: 3531 +Test: [210/517] eta: 0:01:31 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.2967 (0.4928) acc1: 98.4375 (79.9615) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2699 data: 0.1745 max mem: 3532 +Test: [220/517] eta: 0:01:28 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3685 (0.4871) acc1: 95.8333 (80.6856) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2749 data: 0.1783 max mem: 3532 +Test: [230/517] eta: 0:01:24 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3695 (0.4821) acc1: 95.5729 (81.3391) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2749 data: 0.1767 max mem: 3532 +Test: [240/517] eta: 0:01:21 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3649 (0.4772) acc1: 97.3958 (81.9794) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.2699 data: 0.1745 max mem: 3532 +Test: [250/517] eta: 0:01:18 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3466 (0.4720) acc1: 98.9583 (82.6465) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2649 data: 0.1695 max mem: 3532 +Test: [260/517] eta: 0:01:15 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3315 (0.4669) acc1: 99.4792 (83.2725) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2599 data: 0.1628 max mem: 3532 +Test: [270/517] eta: 0:01:11 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3812 (0.4649) acc1: 98.6979 (83.7581) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2649 data: 0.1669 max mem: 3532 +Test: [280/517] eta: 0:01:08 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3924 (0.4607) acc1: 97.3958 (84.2378) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2549 data: 0.1572 max mem: 3532 +Test: [290/517] eta: 0:01:05 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3075 (0.4554) acc1: 98.9583 (84.7535) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2449 data: 0.1491 max mem: 3532 +Test: [300/517] eta: 0:01:02 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3009 (0.4503) acc1: 99.2188 (85.2142) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2599 data: 0.1659 max mem: 3532 +Test: [310/517] eta: 0:00:59 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3149 (0.4467) acc1: 98.9583 (85.6419) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2749 data: 0.1789 max mem: 3532 +Test: [320/517] eta: 0:00:56 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3212 (0.4426) acc1: 99.7396 (86.0714) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2749 data: 0.1783 max mem: 3532 +Test: [330/517] eta: 0:00:53 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3022 (0.4378) acc1: 99.7396 (86.4804) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2799 data: 0.1858 max mem: 3532 +Test: [340/517] eta: 0:00:50 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.2709 (0.4360) acc1: 99.7396 (86.5217) acc5: 100.0000 (100.0000) acc1_neg: 66.406 acc5_neg: 100.000 time: 0.2600 data: 0.1660 max mem: 3532 +Test: [350/517] eta: 0:00:47 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3754 (0.4362) acc1: 93.2292 (86.5355) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2398 data: 0.1421 max mem: 3532 +Test: [360/517] eta: 0:00:44 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3892 (0.4351) acc1: 91.9271 (86.6841) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.2398 data: 0.1412 max mem: 3532 +Test: [370/517] eta: 0:00:41 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3556 (0.4326) acc1: 94.5312 (86.9771) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2499 data: 0.1524 max mem: 3532 +Test: [380/517] eta: 0:00:38 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3491 (0.4321) acc1: 92.9688 (86.9477) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.2649 data: 0.1669 max mem: 3532 +Test: [390/517] eta: 0:00:35 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.4515 (0.4324) acc1: 82.0312 (86.8187) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2599 data: 0.1618 max mem: 3532 +Test: [400/517] eta: 0:00:32 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.4082 (0.4310) acc1: 86.1979 (86.9564) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.2549 data: 0.1591 max mem: 3532 +Test: [410/517] eta: 0:00:29 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3974 (0.4310) acc1: 90.8854 (87.0698) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.2599 data: 0.1641 max mem: 3532 +Test: [420/517] eta: 0:00:27 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.4629 (0.4321) acc1: 89.5833 (87.0819) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.2699 data: 0.1714 max mem: 3532 +Test: [430/517] eta: 0:00:24 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.4399 (0.4315) acc1: 89.5833 (87.1707) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.2549 data: 0.1592 max mem: 3532 +Test: [440/517] eta: 0:00:21 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.4575 (0.4326) acc1: 85.4167 (87.1274) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.2554 data: 0.1625 max mem: 3533 +Test: [450/517] eta: 0:00:18 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.4687 (0.4330) acc1: 89.3229 (87.2194) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.2736 data: 0.1820 max mem: 3533 +Test: [460/517] eta: 0:00:15 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.4632 (0.4337) acc1: 89.8438 (87.2221) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.2533 data: 0.1607 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.4418 (0.4335) acc1: 87.5000 (87.2634) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2190 data: 0.1261 max mem: 3533 +Test: [480/517] eta: 0:00:10 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.4004 (0.4326) acc1: 92.4479 (87.3841) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1890 data: 0.0984 max mem: 3533 +Test: [490/517] eta: 0:00:07 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3737 (0.4308) acc1: 95.5729 (87.5599) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1725 data: 0.0827 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3848 (0.4315) acc1: 94.2708 (87.5676) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1779 data: 0.0881 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.4224 (0.4310) acc1: 92.4479 (87.6244) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1771 data: 0.0874 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 44.037 acc5_pos: 100.000 loss: 0.3772 (0.4302) acc1: 98.1771 (87.7381) acc5: 100.0000 (100.0000) acc1_neg: 99.647 acc5_neg: 100.000 time: 0.1717 data: 0.0804 max mem: 3533 +Test: Total time: 0:02:17 (0.2665 s / it) +* Overall Acc@1 87.738 Acc@5 100.000 + target == real Acc@1 29.528 Acc@5 100.000 + target == fake Acc@1 93.777 Acc@5 100.000 + loss 0.430 + PR-AUC 0.9506 ROC-AUC 0.7111 EER 34.91% (thr=0.6606) +Accuracy of the network on the 198427 test images: 87.7% +[rank0]:[W726 06:31:49.756624376 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 06:32:10.151000 1207392 torch/distributed/run.py:766] +W0726 06:32:10.151000 1207392 torch/distributed/run.py:766] ***************************************** +W0726 06:32:10.151000 1207392 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:32:10.151000 1207392 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:32:40.255150806 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:32:41.575241581 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:32:41.882684683 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:32:41.897760825 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:32:41.203225351 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:32:41.265525638 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:32:41.271568064 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 06:32:41.344103680 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n45_seed9_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed9_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:38:35 acc1_pos: 42.969 acc5_pos: 100.000 loss: 0.7498 (0.7498) acc1: 42.9688 (42.9688) acc5: 100.0000 (100.0000) time: 5.9512 data: 4.4175 max mem: 3531 +Test: [ 10/389] eta: 0:03:59 acc1_pos: 81.771 acc5_pos: 100.000 loss: 0.7386 (0.7168) acc1: 54.6875 (54.1667) acc5: 100.0000 (100.0000) time: 0.6324 data: 0.4112 max mem: 3531 +Test: [ 20/389] eta: 0:02:19 acc1_pos: 20.573 acc5_pos: 100.000 loss: 0.7386 (0.7511) acc1: 45.0521 (47.6066) acc5: 100.0000 (100.0000) time: 0.0999 data: 0.0055 max mem: 3531 +Test: [ 30/389] eta: 0:01:54 acc1_pos: 60.677 acc5_pos: 100.000 loss: 0.7157 (0.7480) acc1: 45.5729 (47.6142) acc5: 100.0000 (100.0000) time: 0.1446 data: 0.0413 max mem: 3531 +Test: [ 40/389] eta: 0:01:40 acc1_pos: 13.281 acc5_pos: 100.000 loss: 0.7075 (0.7518) acc1: 52.0833 (47.0973) acc5: 100.0000 (100.0000) time: 0.1899 data: 0.0908 max mem: 3531 +Test: [ 50/389] eta: 0:01:29 acc1_pos: 73.438 acc5_pos: 100.000 loss: 0.6344 (0.7296) acc1: 57.5521 (50.7864) acc5: 100.0000 (100.0000) time: 0.1826 data: 0.0910 max mem: 3531 +Test: [ 60/389] eta: 0:01:21 acc1_pos: 73.958 acc5_pos: 100.000 loss: 0.6161 (0.7163) acc1: 69.5312 (52.8859) acc5: 100.0000 (100.0000) time: 0.1665 data: 0.0738 max mem: 3531 +Test: [ 70/389] eta: 0:01:15 acc1_pos: 67.188 acc5_pos: 100.000 loss: 0.6209 (0.7022) acc1: 67.1875 (54.6765) acc5: 100.0000 (100.0000) time: 0.1689 data: 0.0774 max mem: 3531 +Test: [ 80/389] eta: 0:01:12 acc1_pos: 89.844 acc5_pos: 100.000 loss: 0.6492 (0.7054) acc1: 57.2917 (54.5460) acc5: 100.0000 (100.0000) time: 0.1933 data: 0.1019 max mem: 3531 +Test: [ 90/389] eta: 0:01:06 acc1_pos: 76.823 acc5_pos: 100.000 loss: 0.4452 (0.6753) acc1: 80.4688 (57.8354) acc5: 100.0000 (100.0000) time: 0.1682 data: 0.0761 max mem: 3531 +Test: [100/389] eta: 0:01:02 acc1_pos: 46.615 acc5_pos: 100.000 loss: 0.5747 (0.6787) acc1: 71.0938 (57.4412) acc5: 100.0000 (100.0000) time: 0.1449 data: 0.0533 max mem: 3531 +Test: [110/389] eta: 0:00:59 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.7207 (0.6816) acc1: 53.1250 (57.0359) acc5: 100.0000 (100.0000) time: 0.1699 data: 0.0773 max mem: 3531 +Test: [120/389] eta: 0:00:56 acc1_pos: 54.167 acc5_pos: 100.000 loss: 0.6566 (0.6759) acc1: 57.0312 (57.7200) acc5: 100.0000 (100.0000) time: 0.1716 data: 0.0787 max mem: 3531 +Test: [130/389] eta: 0:00:53 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.6175 (0.6710) acc1: 65.8854 (58.1107) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1699 data: 0.0778 max mem: 3531 +Test: [140/389] eta: 0:00:50 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.4662 (0.6618) acc1: 73.6979 (59.2956) acc5: 100.0000 (100.0000) acc1_neg: 46.615 acc5_neg: 100.000 time: 0.1748 data: 0.0829 max mem: 3531 +Test: [150/389] eta: 0:00:48 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.5459 (0.6548) acc1: 81.7708 (60.2356) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1655 data: 0.0722 max mem: 3531 +Test: [160/389] eta: 0:00:46 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.5459 (0.6433) acc1: 81.7708 (61.3694) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1797 data: 0.0869 max mem: 3531 +Test: [170/389] eta: 0:00:44 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.3789 (0.6339) acc1: 93.7500 (62.5426) acc5: 100.0000 (100.0000) acc1_neg: 65.365 acc5_neg: 100.000 time: 0.2007 data: 0.1098 max mem: 3531 +Test: [180/389] eta: 0:00:41 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.3899 (0.6283) acc1: 85.4167 (63.3431) acc5: 100.0000 (100.0000) acc1_neg: 71.094 acc5_neg: 100.000 time: 0.1955 data: 0.1035 max mem: 3531 +Test: [190/389] eta: 0:00:39 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.5432 (0.6248) acc1: 74.4792 (63.7844) acc5: 100.0000 (100.0000) acc1_neg: 14.323 acc5_neg: 100.000 time: 0.1907 data: 0.0990 max mem: 3531 +Test: [200/389] eta: 0:00:37 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.5650 (0.6239) acc1: 74.4792 (63.9835) acc5: 100.0000 (100.0000) acc1_neg: 70.052 acc5_neg: 100.000 time: 0.1858 data: 0.0929 max mem: 3531 +Test: [210/389] eta: 0:00:35 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.3668 (0.6135) acc1: 85.4167 (65.1054) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1953 data: 0.1019 max mem: 3532 +Test: [220/389] eta: 0:00:33 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.3086 (0.6012) acc1: 98.4375 (66.5359) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1975 data: 0.1063 max mem: 3532 +Test: [230/389] eta: 0:00:31 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.4562 (0.5999) acc1: 90.8854 (66.6576) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1893 data: 0.0989 max mem: 3532 +Test: [240/389] eta: 0:00:29 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.4805 (0.5957) acc1: 82.2917 (67.1789) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1900 data: 0.0981 max mem: 3532 +Test: [250/389] eta: 0:00:27 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.4015 (0.5863) acc1: 92.9688 (68.3869) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1793 data: 0.0860 max mem: 3532 +Test: [260/389] eta: 0:00:25 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.3741 (0.5832) acc1: 98.4375 (68.9486) acc5: 100.0000 (100.0000) acc1_neg: 19.792 acc5_neg: 100.000 time: 0.1766 data: 0.0836 max mem: 3532 +Test: [270/389] eta: 0:00:23 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.4533 (0.5889) acc1: 90.3646 (68.2157) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1814 data: 0.0895 max mem: 3532 +Test: [280/389] eta: 0:00:21 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.5082 (0.5902) acc1: 62.2396 (68.1485) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1754 data: 0.0846 max mem: 3532 +Test: [290/389] eta: 0:00:19 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.4986 (0.5869) acc1: 82.8125 (68.6211) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.1779 data: 0.0876 max mem: 3532 +Test: [300/389] eta: 0:00:17 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.5673 (0.5896) acc1: 61.1979 (68.3036) acc5: 100.0000 (100.0000) acc1_neg: 59.375 acc5_neg: 100.000 time: 0.1792 data: 0.0890 max mem: 3532 +Test: [310/389] eta: 0:00:15 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.6758 (0.5953) acc1: 57.5521 (67.7577) acc5: 100.0000 (100.0000) acc1_neg: 5.469 acc5_neg: 100.000 time: 0.1753 data: 0.0850 max mem: 3532 +Test: [320/389] eta: 0:00:13 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.6214 (0.5937) acc1: 57.5521 (67.9185) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1834 data: 0.0924 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.6324 (0.5956) acc1: 66.1458 (67.7957) acc5: 100.0000 (100.0000) acc1_neg: 65.365 acc5_neg: 100.000 time: 0.1936 data: 0.1020 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.4142 (0.5885) acc1: 67.9688 (68.6179) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1833 data: 0.0916 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.3394 (0.5925) acc1: 97.3958 (68.2410) acc5: 100.0000 (100.0000) acc1_neg: 33.854 acc5_neg: 100.000 time: 0.1754 data: 0.0842 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.5110 (0.5922) acc1: 75.5208 (68.2573) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.1688 data: 0.0779 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.5040 (0.5898) acc1: 77.6042 (68.3401) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1604 data: 0.0688 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.4847 (0.5871) acc1: 71.8750 (68.5320) acc5: 100.0000 (100.0000) acc1_neg: 76.302 acc5_neg: 100.000 time: 0.1468 data: 0.0558 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.4196 (0.5837) acc1: 82.8125 (68.8708) acc5: 100.0000 (100.0000) acc1_neg: 82.240 acc5_neg: 100.000 time: 0.1295 data: 0.0359 max mem: 3532 +Test: Total time: 0:01:13 (0.1882 s / it) +* Overall Acc@1 68.871 Acc@5 100.000 + target == real Acc@1 57.090 Acc@5 100.000 + target == fake Acc@1 74.347 Acc@5 100.000 + loss 0.584 + PR-AUC 0.8509 ROC-AUC 0.7330 EER 33.08% (thr=0.5459) +Accuracy of the network on the 149358 test images: 68.9% +[rank0]:[W726 06:34:07.172750821 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 06:34:26.485000 1207955 torch/distributed/run.py:766] +W0726 06:34:26.485000 1207955 torch/distributed/run.py:766] ***************************************** +W0726 06:34:26.485000 1207955 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:34:26.485000 1207955 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:34:55.665078460 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:34:55.763140632 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:34:55.983710526 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:34:55.021321887 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:34:55.187146049 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:34:55.231476955 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:34:55.266444699 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 06:34:55.266467639 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n45_seed9_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed9_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:38:17 acc1_pos: 84.635 acc5_pos: 100.000 loss: 0.3949 (0.3949) acc1: 84.6354 (84.6354) acc5: 100.0000 (100.0000) time: 6.8990 data: 5.3924 max mem: 3531 +Test: [ 10/333] eta: 0:03:49 acc1_pos: 85.417 acc5_pos: 100.000 loss: 0.3127 (0.3194) acc1: 86.4583 (87.6657) acc5: 100.0000 (100.0000) time: 0.7093 data: 0.4907 max mem: 3531 +Test: [ 20/333] eta: 0:02:14 acc1_pos: 94.271 acc5_pos: 100.000 loss: 0.3062 (0.3176) acc1: 86.9792 (87.1404) acc5: 100.0000 (100.0000) time: 0.1051 data: 0.0105 max mem: 3531 +Test: [ 30/333] eta: 0:01:44 acc1_pos: 59.375 acc5_pos: 100.000 loss: 0.3942 (0.4473) acc1: 76.0417 (76.7305) acc5: 100.0000 (100.0000) time: 0.1449 data: 0.0471 max mem: 3531 +Test: [ 40/333] eta: 0:01:33 acc1_pos: 55.729 acc5_pos: 100.000 loss: 0.7445 (0.5216) acc1: 53.9062 (70.5666) acc5: 100.0000 (100.0000) time: 0.2019 data: 0.1075 max mem: 3531 +Test: [ 50/333] eta: 0:01:24 acc1_pos: 55.469 acc5_pos: 100.000 loss: 0.7324 (0.5635) acc1: 52.6042 (67.1671) acc5: 100.0000 (100.0000) time: 0.2259 data: 0.1345 max mem: 3531 +Test: [ 60/333] eta: 0:01:19 acc1_pos: 46.615 acc5_pos: 100.000 loss: 0.7310 (0.5932) acc1: 52.6042 (64.7584) acc5: 100.0000 (100.0000) time: 0.2380 data: 0.1476 max mem: 3531 +Test: [ 70/333] eta: 0:01:10 acc1_pos: 51.823 acc5_pos: 100.000 loss: 0.7310 (0.6136) acc1: 53.1250 (63.4756) acc5: 100.0000 (100.0000) time: 0.1939 data: 0.1007 max mem: 3531 +Test: [ 80/333] eta: 0:01:07 acc1_pos: 60.156 acc5_pos: 100.000 loss: 0.7051 (0.6270) acc1: 53.1250 (62.1110) acc5: 100.0000 (100.0000) time: 0.1922 data: 0.0975 max mem: 3531 +Test: [ 90/333] eta: 0:01:02 acc1_pos: 64.844 acc5_pos: 100.000 loss: 0.6894 (0.6286) acc1: 60.1562 (62.2682) acc5: 100.0000 (100.0000) time: 0.2165 data: 0.1227 max mem: 3531 +Test: [100/333] eta: 0:00:58 acc1_pos: 71.615 acc5_pos: 100.000 loss: 0.6265 (0.6306) acc1: 65.6250 (62.3737) acc5: 100.0000 (100.0000) time: 0.1794 data: 0.0849 max mem: 3531 +Test: [110/333] eta: 0:00:54 acc1_pos: 72.917 acc5_pos: 100.000 loss: 0.6227 (0.6315) acc1: 63.2812 (62.5070) acc5: 100.0000 (100.0000) time: 0.1900 data: 0.0960 max mem: 3531 +Test: [120/333] eta: 0:00:51 acc1_pos: 70.052 acc5_pos: 100.000 loss: 0.6048 (0.6309) acc1: 63.0208 (62.6550) acc5: 100.0000 (100.0000) time: 0.2034 data: 0.1113 max mem: 3531 +Test: [130/333] eta: 0:00:48 acc1_pos: 76.562 acc5_pos: 100.000 loss: 0.6029 (0.6299) acc1: 66.9271 (62.8558) acc5: 100.0000 (100.0000) time: 0.2057 data: 0.1127 max mem: 3531 +Test: [140/333] eta: 0:00:45 acc1_pos: 75.260 acc5_pos: 100.000 loss: 0.6031 (0.6278) acc1: 66.9271 (63.1612) acc5: 100.0000 (100.0000) time: 0.2099 data: 0.1166 max mem: 3531 +Test: [150/333] eta: 0:00:43 acc1_pos: 68.229 acc5_pos: 100.000 loss: 0.6228 (0.6287) acc1: 64.0625 (63.1554) acc5: 100.0000 (100.0000) time: 0.2143 data: 0.1228 max mem: 3531 +Test: [160/333] eta: 0:00:40 acc1_pos: 52.961 acc5_pos: 100.000 loss: 0.6304 (0.6378) acc1: 60.6771 (62.4709) acc5: 100.0000 (100.0000) acc1_neg: 38.021 acc5_neg: 100.000 time: 0.2102 data: 0.1166 max mem: 3531 +Test: [170/333] eta: 0:00:37 acc1_pos: 52.961 acc5_pos: 100.000 loss: 0.7754 (0.6488) acc1: 49.2188 (61.6563) acc5: 100.0000 (100.0000) acc1_neg: 55.990 acc5_neg: 100.000 time: 0.2038 data: 0.1105 max mem: 3531 +Test: [180/333] eta: 0:00:35 acc1_pos: 52.961 acc5_pos: 100.000 loss: 0.7602 (0.6540) acc1: 52.6042 (61.1936) acc5: 100.0000 (100.0000) acc1_neg: 45.052 acc5_neg: 100.000 time: 0.2098 data: 0.1155 max mem: 3531 +Test: [190/333] eta: 0:00:32 acc1_pos: 52.961 acc5_pos: 100.000 loss: 0.7304 (0.6555) acc1: 54.4271 (61.0793) acc5: 100.0000 (100.0000) acc1_neg: 54.427 acc5_neg: 100.000 time: 0.2154 data: 0.1208 max mem: 3531 +Test: [200/333] eta: 0:00:30 acc1_pos: 52.961 acc5_pos: 100.000 loss: 0.7059 (0.6576) acc1: 58.3333 (61.0243) acc5: 100.0000 (100.0000) acc1_neg: 58.594 acc5_neg: 100.000 time: 0.2026 data: 0.1115 max mem: 3531 +Test: [210/333] eta: 0:00:27 acc1_pos: 52.961 acc5_pos: 100.000 loss: 0.7439 (0.6633) acc1: 55.4688 (60.6721) acc5: 100.0000 (100.0000) acc1_neg: 45.312 acc5_neg: 100.000 time: 0.1864 data: 0.0918 max mem: 3532 +Test: [220/333] eta: 0:00:25 acc1_pos: 52.961 acc5_pos: 100.000 loss: 0.7703 (0.6714) acc1: 51.0417 (60.0726) acc5: 100.0000 (100.0000) acc1_neg: 69.271 acc5_neg: 100.000 time: 0.1877 data: 0.0937 max mem: 3532 +Test: [230/333] eta: 0:00:23 acc1_pos: 52.961 acc5_pos: 100.000 loss: 0.5515 (0.6653) acc1: 69.2708 (60.7515) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.1947 data: 0.1016 max mem: 3532 +Test: [240/333] eta: 0:00:20 acc1_pos: 52.961 acc5_pos: 100.000 loss: 0.5515 (0.6634) acc1: 71.3542 (60.8835) acc5: 100.0000 (100.0000) acc1_neg: 64.062 acc5_neg: 100.000 time: 0.1889 data: 0.0947 max mem: 3532 +Test: [250/333] eta: 0:00:18 acc1_pos: 52.961 acc5_pos: 100.000 loss: 0.6436 (0.6632) acc1: 63.2812 (60.9354) acc5: 100.0000 (100.0000) acc1_neg: 57.292 acc5_neg: 100.000 time: 0.1852 data: 0.0932 max mem: 3532 +Test: [260/333] eta: 0:00:16 acc1_pos: 52.961 acc5_pos: 100.000 loss: 0.6788 (0.6651) acc1: 60.4167 (60.8477) acc5: 100.0000 (100.0000) acc1_neg: 64.844 acc5_neg: 100.000 time: 0.1958 data: 0.1029 max mem: 3532 +Test: [270/333] eta: 0:00:13 acc1_pos: 52.961 acc5_pos: 100.000 loss: 0.6675 (0.6641) acc1: 63.0208 (60.9855) acc5: 100.0000 (100.0000) acc1_neg: 64.844 acc5_neg: 100.000 time: 0.1982 data: 0.1057 max mem: 3532 +Test: [280/333] eta: 0:00:11 acc1_pos: 52.961 acc5_pos: 100.000 loss: 0.6372 (0.6677) acc1: 63.0208 (60.6465) acc5: 100.0000 (100.0000) acc1_neg: 55.208 acc5_neg: 100.000 time: 0.1751 data: 0.0843 max mem: 3532 +Test: [290/333] eta: 0:00:09 acc1_pos: 52.961 acc5_pos: 100.000 loss: 0.8692 (0.6749) acc1: 42.1875 (59.9853) acc5: 100.0000 (100.0000) acc1_neg: 46.615 acc5_neg: 100.000 time: 0.1746 data: 0.0841 max mem: 3532 +Test: [300/333] eta: 0:00:07 acc1_pos: 52.961 acc5_pos: 100.000 loss: 0.8788 (0.6817) acc1: 40.6250 (59.4278) acc5: 100.0000 (100.0000) acc1_neg: 49.219 acc5_neg: 100.000 time: 0.1662 data: 0.0754 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 52.961 acc5_pos: 100.000 loss: 0.7972 (0.6805) acc1: 49.2188 (59.5726) acc5: 100.0000 (100.0000) acc1_neg: 65.365 acc5_neg: 100.000 time: 0.1338 data: 0.0432 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 52.961 acc5_pos: 100.000 loss: 0.6093 (0.6798) acc1: 66.6667 (59.6735) acc5: 100.0000 (100.0000) acc1_neg: 40.625 acc5_neg: 100.000 time: 0.1214 data: 0.0317 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 52.961 acc5_pos: 100.000 loss: 0.8644 (0.6878) acc1: 45.3125 (59.1240) acc5: 100.0000 (100.0000) acc1_neg: 45.312 acc5_neg: 100.000 time: 0.1171 data: 0.0275 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 52.961 acc5_pos: 100.000 loss: 0.9078 (0.6893) acc1: 44.5312 (59.0279) acc5: 100.0000 (100.0000) acc1_neg: 45.352 acc5_neg: 100.000 time: 0.1183 data: 0.0254 max mem: 3532 +Test: Total time: 0:01:08 (0.2050 s / it) +* Overall Acc@1 59.028 Acc@5 100.000 + target == real Acc@1 63.171 Acc@5 100.000 + target == fake Acc@1 55.340 Acc@5 100.000 + loss 0.689 + PR-AUC 0.6516 ROC-AUC 0.6363 EER 40.63% (thr=0.4763) +Accuracy of the network on the 127843 test images: 59.0% +[rank0]:[W726 06:36:12.793100695 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 06:36:32.171000 1208503 torch/distributed/run.py:766] +W0726 06:36:32.171000 1208503 torch/distributed/run.py:766] ***************************************** +W0726 06:36:32.171000 1208503 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:36:32.171000 1208503 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 06:36:59.527426127 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:36:59.634712574 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:36:59.829672171 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:36:59.101099916 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:36:59.185837596 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:36:59.253886525 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:36:59.268420164 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:36:59.389407706 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n45_seed9_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed9_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:31:30 acc1_pos: 95.573 acc5_pos: 100.000 loss: 0.2231 (0.2231) acc1: 95.5729 (95.5729) acc5: 100.0000 (100.0000) time: 8.0452 data: 5.8475 max mem: 3531 +Test: [ 10/235] eta: 0:03:06 acc1_pos: 95.573 acc5_pos: 100.000 loss: 0.2242 (0.2273) acc1: 95.5729 (94.8864) acc5: 100.0000 (100.0000) time: 0.8283 data: 0.5425 max mem: 3531 +Test: [ 20/235] eta: 0:01:43 acc1_pos: 94.531 acc5_pos: 100.000 loss: 0.2206 (0.2219) acc1: 95.5729 (95.2753) acc5: 100.0000 (100.0000) time: 0.1030 data: 0.0062 max mem: 3531 +Test: [ 30/235] eta: 0:01:34 acc1_pos: 96.094 acc5_pos: 100.000 loss: 0.2198 (0.2210) acc1: 95.5729 (95.3461) acc5: 100.0000 (100.0000) time: 0.2561 data: 0.1607 max mem: 3531 +Test: [ 40/235] eta: 0:01:21 acc1_pos: 94.531 acc5_pos: 100.000 loss: 0.2199 (0.2206) acc1: 95.3125 (95.2807) acc5: 100.0000 (100.0000) time: 0.3560 data: 0.2639 max mem: 3531 +Test: [ 50/235] eta: 0:01:11 acc1_pos: 93.750 acc5_pos: 100.000 loss: 0.2179 (0.2217) acc1: 94.5312 (95.1593) acc5: 100.0000 (100.0000) time: 0.2706 data: 0.1783 max mem: 3531 +Test: [ 60/235] eta: 0:01:03 acc1_pos: 94.271 acc5_pos: 100.000 loss: 0.2231 (0.2228) acc1: 95.0521 (95.1545) acc5: 100.0000 (100.0000) time: 0.2491 data: 0.1556 max mem: 3531 +Test: [ 70/235] eta: 0:00:56 acc1_pos: 96.923 acc5_pos: 100.000 loss: 0.2409 (0.3664) acc1: 93.7500 (85.9778) acc5: 100.0000 (100.0000) acc1_neg: 20.573 acc5_neg: 100.000 time: 0.2427 data: 0.1472 max mem: 3531 +Test: [ 80/235] eta: 0:00:50 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.3715 (0.4926) acc1: 21.3542 (77.9385) acc5: 100.0000 (100.0000) acc1_neg: 19.010 acc5_neg: 100.000 time: 0.2070 data: 0.1110 max mem: 3531 +Test: [ 90/235] eta: 0:00:45 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.3806 (0.5906) acc1: 21.6146 (71.7634) acc5: 100.0000 (100.0000) acc1_neg: 22.396 acc5_neg: 100.000 time: 0.2005 data: 0.1071 max mem: 3531 +Test: [100/235] eta: 0:00:40 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.3901 (0.6683) acc1: 21.8750 (66.8162) acc5: 100.0000 (100.0000) acc1_neg: 17.708 acc5_neg: 100.000 time: 0.2024 data: 0.1109 max mem: 3531 +Test: [110/235] eta: 0:00:36 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.3645 (0.7320) acc1: 20.8333 (62.6572) acc5: 100.0000 (100.0000) acc1_neg: 22.396 acc5_neg: 100.000 time: 0.1993 data: 0.1075 max mem: 3531 +Test: [120/235] eta: 0:00:32 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.4232 (0.8006) acc1: 17.9688 (58.4345) acc5: 100.0000 (100.0000) acc1_neg: 11.458 acc5_neg: 100.000 time: 0.2004 data: 0.1091 max mem: 3531 +Test: [130/235] eta: 0:00:29 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.5655 (0.8594) acc1: 10.9375 (54.7789) acc5: 100.0000 (100.0000) acc1_neg: 11.719 acc5_neg: 100.000 time: 0.1853 data: 0.0910 max mem: 3531 +Test: [140/235] eta: 0:00:25 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.5939 (0.9121) acc1: 10.1562 (51.5976) acc5: 100.0000 (100.0000) acc1_neg: 6.510 acc5_neg: 100.000 time: 0.1893 data: 0.0933 max mem: 3531 +Test: [150/235] eta: 0:00:22 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.5870 (0.9556) acc1: 10.4167 (48.9066) acc5: 100.0000 (100.0000) acc1_neg: 13.802 acc5_neg: 100.000 time: 0.1949 data: 0.0996 max mem: 3531 +Test: [160/235] eta: 0:00:19 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.5704 (0.9942) acc1: 10.6771 (46.5466) acc5: 100.0000 (100.0000) acc1_neg: 11.719 acc5_neg: 100.000 time: 0.1951 data: 0.0997 max mem: 3531 +Test: [170/235] eta: 0:00:16 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.5375 (1.0042) acc1: 14.3229 (45.6567) acc5: 100.0000 (100.0000) acc1_neg: 28.906 acc5_neg: 100.000 time: 0.2075 data: 0.1132 max mem: 3531 +Test: [180/235] eta: 0:00:14 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.1264 (1.0098) acc1: 34.3750 (45.0492) acc5: 100.0000 (100.0000) acc1_neg: 35.156 acc5_neg: 100.000 time: 0.2127 data: 0.1168 max mem: 3531 +Test: [190/235] eta: 0:00:11 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.0900 (1.0144) acc1: 34.8958 (44.5558) acc5: 100.0000 (100.0000) acc1_neg: 35.417 acc5_neg: 100.000 time: 0.2056 data: 0.1105 max mem: 3531 +Test: [200/235] eta: 0:00:08 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.1018 (1.0196) acc1: 33.5938 (43.9897) acc5: 100.0000 (100.0000) acc1_neg: 30.990 acc5_neg: 100.000 time: 0.2018 data: 0.1105 max mem: 3531 +Test: [210/235] eta: 0:00:06 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.1152 (1.0230) acc1: 33.5938 (43.5747) acc5: 100.0000 (100.0000) acc1_neg: 36.198 acc5_neg: 100.000 time: 0.1727 data: 0.0827 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.0955 (1.0267) acc1: 35.1562 (43.2056) acc5: 100.0000 (100.0000) acc1_neg: 35.156 acc5_neg: 100.000 time: 0.1364 data: 0.0468 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.0921 (1.0292) acc1: 35.9375 (42.9090) acc5: 100.0000 (100.0000) acc1_neg: 35.938 acc5_neg: 100.000 time: 0.1279 data: 0.0384 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.0841 (1.0317) acc1: 35.4167 (42.7920) acc5: 100.0000 (100.0000) acc1_neg: 30.435 acc5_neg: 100.000 time: 0.1123 data: 0.0239 max mem: 3532 +Test: Total time: 0:00:54 (0.2338 s / it) +* Overall Acc@1 42.792 Acc@5 100.000 + target == real Acc@1 95.141 Acc@5 100.000 + target == fake Acc@1 23.858 Acc@5 100.000 + loss 1.032 + PR-AUC 0.8785 ROC-AUC 0.7346 EER 33.00% (thr=0.1841) +Accuracy of the network on the 89879 test images: 42.8% +[rank0]:[W726 06:38:02.668661863 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 06:38:21.673000 1209063 torch/distributed/run.py:766] +W0726 06:38:21.673000 1209063 torch/distributed/run.py:766] ***************************************** +W0726 06:38:21.673000 1209063 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:38:21.673000 1209063 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 06:38:52.901551955 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W726 06:38:52.902098360 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:38:52.359236713 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:38:53.418185955 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:38:53.588585606 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:38:53.636820368 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:38:53.761827940 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W726 06:38:53.761827848 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n45_seed9_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed9_resampled_n116-041-023_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:26:58 acc1_pos: 69.792 acc5_pos: 100.000 loss: 0.6320 (0.6320) acc1: 69.7917 (69.7917) acc5: 100.0000 (100.0000) time: 6.0174 data: 4.6673 max mem: 3531 +Test: [ 10/269] eta: 0:02:42 acc1_pos: 79.948 acc5_pos: 100.000 loss: 0.5622 (0.5636) acc1: 76.0417 (74.3608) acc5: 100.0000 (100.0000) time: 0.6293 data: 0.4247 max mem: 3531 +Test: [ 20/269] eta: 0:01:37 acc1_pos: 86.198 acc5_pos: 100.000 loss: 0.5447 (0.5457) acc1: 76.8229 (76.8849) acc5: 100.0000 (100.0000) time: 0.1102 data: 0.0156 max mem: 3531 +Test: [ 30/269] eta: 0:01:28 acc1_pos: 93.229 acc5_pos: 100.000 loss: 0.5103 (0.5471) acc1: 81.2500 (76.4617) acc5: 100.0000 (100.0000) time: 0.2261 data: 0.1308 max mem: 3531 +Test: [ 40/269] eta: 0:01:21 acc1_pos: 79.310 acc5_pos: 100.000 loss: 0.5711 (0.5617) acc1: 72.9167 (74.4538) acc5: 100.0000 (100.0000) acc1_neg: 63.802 acc5_neg: 100.000 time: 0.3165 data: 0.2252 max mem: 3531 +Test: [ 50/269] eta: 0:01:13 acc1_pos: 79.310 acc5_pos: 100.000 loss: 0.6743 (0.5990) acc1: 58.8542 (69.1330) acc5: 100.0000 (100.0000) acc1_neg: 30.208 acc5_neg: 100.000 time: 0.2787 data: 0.1865 max mem: 3531 +Test: [ 60/269] eta: 0:01:03 acc1_pos: 79.310 acc5_pos: 100.000 loss: 0.7838 (0.6359) acc1: 38.2812 (63.7124) acc5: 100.0000 (100.0000) acc1_neg: 28.906 acc5_neg: 100.000 time: 0.1983 data: 0.1023 max mem: 3531 +Test: [ 70/269] eta: 0:00:56 acc1_pos: 79.310 acc5_pos: 100.000 loss: 0.8111 (0.6563) acc1: 38.0208 (60.4570) acc5: 100.0000 (100.0000) acc1_neg: 51.302 acc5_neg: 100.000 time: 0.1550 data: 0.0587 max mem: 3531 +Test: [ 80/269] eta: 0:00:51 acc1_pos: 79.310 acc5_pos: 100.000 loss: 0.7287 (0.6603) acc1: 46.3542 (59.7865) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.1862 data: 0.0909 max mem: 3531 +Test: [ 90/269] eta: 0:00:47 acc1_pos: 79.310 acc5_pos: 100.000 loss: 0.7292 (0.6746) acc1: 46.3542 (57.6122) acc5: 100.0000 (100.0000) acc1_neg: 22.917 acc5_neg: 100.000 time: 0.2015 data: 0.1063 max mem: 3531 +Test: [100/269] eta: 0:00:43 acc1_pos: 79.310 acc5_pos: 100.000 loss: 0.8649 (0.6963) acc1: 30.2083 (54.4709) acc5: 100.0000 (100.0000) acc1_neg: 13.281 acc5_neg: 100.000 time: 0.1806 data: 0.0881 max mem: 3531 +Test: [110/269] eta: 0:00:39 acc1_pos: 79.310 acc5_pos: 100.000 loss: 0.9204 (0.7196) acc1: 22.9167 (51.3701) acc5: 100.0000 (100.0000) acc1_neg: 14.583 acc5_neg: 100.000 time: 0.1805 data: 0.0898 max mem: 3531 +Test: [120/269] eta: 0:00:36 acc1_pos: 79.310 acc5_pos: 100.000 loss: 0.8713 (0.7220) acc1: 27.0833 (51.0245) acc5: 100.0000 (100.0000) acc1_neg: 51.562 acc5_neg: 100.000 time: 0.1835 data: 0.0927 max mem: 3531 +Test: [130/269] eta: 0:00:33 acc1_pos: 79.310 acc5_pos: 100.000 loss: 0.8164 (0.7315) acc1: 39.8438 (50.0755) acc5: 100.0000 (100.0000) acc1_neg: 32.292 acc5_neg: 100.000 time: 0.1800 data: 0.0839 max mem: 3531 +Test: [140/269] eta: 0:00:30 acc1_pos: 79.310 acc5_pos: 100.000 loss: 0.8784 (0.7426) acc1: 33.5938 (48.7201) acc5: 100.0000 (100.0000) acc1_neg: 22.917 acc5_neg: 100.000 time: 0.1942 data: 0.0981 max mem: 3531 +Test: [150/269] eta: 0:00:28 acc1_pos: 79.310 acc5_pos: 100.000 loss: 0.8780 (0.7433) acc1: 32.2917 (48.6858) acc5: 100.0000 (100.0000) acc1_neg: 61.719 acc5_neg: 100.000 time: 0.2092 data: 0.1165 max mem: 3531 +Test: [160/269] eta: 0:00:25 acc1_pos: 79.310 acc5_pos: 100.000 loss: 0.7618 (0.7425) acc1: 42.4479 (48.7044) acc5: 100.0000 (100.0000) acc1_neg: 42.448 acc5_neg: 100.000 time: 0.2168 data: 0.1245 max mem: 3531 +Test: [170/269] eta: 0:00:23 acc1_pos: 79.310 acc5_pos: 100.000 loss: 0.7970 (0.7475) acc1: 37.5000 (47.8862) acc5: 100.0000 (100.0000) acc1_neg: 53.646 acc5_neg: 100.000 time: 0.2581 data: 0.1674 max mem: 3531 +Test: [180/269] eta: 0:00:21 acc1_pos: 79.310 acc5_pos: 100.000 loss: 0.7559 (0.7446) acc1: 41.1458 (48.2490) acc5: 100.0000 (100.0000) acc1_neg: 38.281 acc5_neg: 100.000 time: 0.2546 data: 0.1605 max mem: 3531 +Test: [190/269] eta: 0:00:18 acc1_pos: 79.310 acc5_pos: 100.000 loss: 0.7302 (0.7483) acc1: 45.8333 (47.7503) acc5: 100.0000 (100.0000) acc1_neg: 42.188 acc5_neg: 100.000 time: 0.1557 data: 0.0595 max mem: 3531 +Test: [200/269] eta: 0:00:15 acc1_pos: 79.310 acc5_pos: 100.000 loss: 0.8749 (0.7568) acc1: 33.3333 (46.9359) acc5: 100.0000 (100.0000) acc1_neg: 34.115 acc5_neg: 100.000 time: 0.1521 data: 0.0577 max mem: 3531 +Test: [210/269] eta: 0:00:13 acc1_pos: 79.310 acc5_pos: 100.000 loss: 0.8265 (0.7559) acc1: 36.7188 (47.1009) acc5: 100.0000 (100.0000) acc1_neg: 61.719 acc5_neg: 100.000 time: 0.2252 data: 0.1292 max mem: 3532 +Test: [220/269] eta: 0:00:11 acc1_pos: 79.310 acc5_pos: 100.000 loss: 0.7028 (0.7525) acc1: 55.2083 (47.7882) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.2211 data: 0.1260 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 79.310 acc5_pos: 100.000 loss: 0.6055 (0.7467) acc1: 65.1042 (48.5288) acc5: 100.0000 (100.0000) acc1_neg: 50.260 acc5_neg: 100.000 time: 0.1797 data: 0.0853 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 79.310 acc5_pos: 100.000 loss: 0.7345 (0.7526) acc1: 40.6250 (47.6519) acc5: 100.0000 (100.0000) acc1_neg: 37.500 acc5_neg: 100.000 time: 0.1599 data: 0.0654 max mem: 3532 +Test: [250/269] eta: 0:00:04 acc1_pos: 79.310 acc5_pos: 100.000 loss: 0.8519 (0.7549) acc1: 29.1667 (47.2329) acc5: 100.0000 (100.0000) acc1_neg: 14.062 acc5_neg: 100.000 time: 0.1659 data: 0.0747 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 79.310 acc5_pos: 100.000 loss: 0.8677 (0.7618) acc1: 20.0521 (46.2195) acc5: 100.0000 (100.0000) acc1_neg: 29.427 acc5_neg: 100.000 time: 0.1536 data: 0.0636 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 79.310 acc5_pos: 100.000 loss: 0.9167 (0.7648) acc1: 19.5312 (45.7020) acc5: 100.0000 (100.0000) acc1_neg: 50.919 acc5_neg: 100.000 time: 0.1368 data: 0.0428 max mem: 3532 +Test: Total time: 0:00:57 (0.2141 s / it) +* Overall Acc@1 45.702 Acc@5 100.000 + target == real Acc@1 76.419 Acc@5 100.000 + target == fake Acc@1 41.510 Acc@5 100.000 + loss 0.765 + PR-AUC 0.9229 ROC-AUC 0.6388 EER 40.18% (thr=0.4292) +Accuracy of the network on the 103293 test images: 45.7% +[rank0]:[W726 06:40:00.523314927 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T06:40:06Z] eval_done tag=fixed500k_fake_n45_seed9_resampled +eval_queue_done_utc=2026-07-26T06:40:06Z diff --git a/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_1958_seed5_9/eval_queue_n116-041-023.pid b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_1958_seed5_9/eval_queue_n116-041-023.pid new file mode 100644 index 0000000000000000000000000000000000000000..825ece2e1b9f8b910ff51b79df11b8e8a49c9204 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_1958_seed5_9/eval_queue_n116-041-023.pid @@ -0,0 +1 @@ +1169666 diff --git a/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_1958_seed5_9/eval_queue_n116-041-153.log b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_1958_seed5_9/eval_queue_n116-041-153.log new file mode 100644 index 0000000000000000000000000000000000000000..c1a4aa2df8bc59eb3c1e58cd05390d1db30afffb --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_1958_seed5_9/eval_queue_n116-041-153.log @@ -0,0 +1,10270 @@ +eval_queue_started_utc=2026-07-26T02:59:37Z +hostname=n116-041-153 +train_tag=20260725_1958_seed5_9 +eval_tag=20260725_1958_seed5_9 +real_level=8 +fake_level=16 +eval_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9 +batch_per_gpu=256 +seeds=5 6 7 8 9 +[2026-07-26T02:59:37Z] eval_start tag=fixed500k_real_n8_seed5_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed5_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T02:59:37Z +hostname=n116-041-153 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed5_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n8_seed5_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 02:59:50.739000 1166487 torch/distributed/run.py:766] +W0726 02:59:50.739000 1166487 torch/distributed/run.py:766] ***************************************** +W0726 02:59:50.739000 1166487 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 02:59:50.739000 1166487 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +[rank6]:[W726 03:00:20.034980041 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:00:20.044554987 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:00:21.509237022 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:00:21.575565275 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:00:21.608612528 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:00:21.613180329 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:00:21.641445291 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:00:21.670817761 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n8_seed5_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed5_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:20:58 acc1_pos: 15.885 acc5_pos: 100.000 loss: 1.5852 (1.5852) acc1: 15.8854 (15.8854) acc5: 100.0000 (100.0000) time: 5.3785 data: 3.7567 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/234] eta: 0:02:07 acc1_pos: 21.615 acc5_pos: 100.000 loss: 1.5075 (1.4526) acc1: 21.6146 (19.6970) acc5: 100.0000 (100.0000) time: 0.5707 data: 0.3418 max mem: 3531 +Test: [ 20/234] eta: 0:01:13 acc1_pos: 84.635 acc5_pos: 100.000 loss: 1.2617 (1.2460) acc1: 23.9583 (33.4697) acc5: 100.0000 (100.0000) time: 0.0933 data: 0.0004 max mem: 3531 +Test: [ 30/234] eta: 0:00:54 acc1_pos: 62.760 acc5_pos: 100.000 loss: 0.3679 (0.9777) acc1: 71.6146 (49.0087) acc5: 100.0000 (100.0000) time: 0.0983 data: 0.0004 max mem: 3531 +Test: [ 40/234] eta: 0:00:47 acc1_pos: 95.833 acc5_pos: 100.000 loss: 0.3349 (0.8556) acc1: 83.8542 (55.4560) acc5: 100.0000 (100.0000) time: 0.1399 data: 0.0436 max mem: 3531 +Test: [ 50/234] eta: 0:00:43 acc1_pos: 98.177 acc5_pos: 100.000 loss: 0.1796 (0.7211) acc1: 95.3125 (63.4140) acc5: 100.0000 (100.0000) time: 0.1949 data: 0.0996 max mem: 3531 +Test: [ 60/234] eta: 0:00:41 acc1_pos: 92.414 acc5_pos: 100.000 loss: 0.1359 (0.6172) acc1: 97.3958 (69.3605) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2299 data: 0.1356 max mem: 3531 +Test: [ 70/234] eta: 0:00:36 acc1_pos: 92.414 acc5_pos: 100.000 loss: 0.0813 (0.5847) acc1: 100.0000 (71.4789) acc5: 100.0000 (100.0000) acc1_neg: 74.219 acc5_neg: 100.000 time: 0.1934 data: 0.1028 max mem: 3531 +Test: [ 80/234] eta: 0:00:33 acc1_pos: 92.414 acc5_pos: 100.000 loss: 0.4685 (0.5580) acc1: 78.6458 (73.2157) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1499 data: 0.0575 max mem: 3531 +Test: [ 90/234] eta: 0:00:31 acc1_pos: 92.414 acc5_pos: 100.000 loss: 0.1491 (0.5154) acc1: 96.3542 (75.7326) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.1814 data: 0.0831 max mem: 3531 +Test: [100/234] eta: 0:00:28 acc1_pos: 92.414 acc5_pos: 100.000 loss: 0.1624 (0.4993) acc1: 96.0938 (76.7353) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1949 data: 0.0971 max mem: 3531 +Test: [110/234] eta: 0:00:26 acc1_pos: 92.414 acc5_pos: 100.000 loss: 0.2968 (0.4848) acc1: 87.2396 (77.6957) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1849 data: 0.0899 max mem: 3531 +Test: [120/234] eta: 0:00:23 acc1_pos: 92.414 acc5_pos: 100.000 loss: 0.2551 (0.4758) acc1: 89.8438 (78.2821) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1849 data: 0.0900 max mem: 3531 +Test: [130/234] eta: 0:00:21 acc1_pos: 92.414 acc5_pos: 100.000 loss: 0.2179 (0.4583) acc1: 91.9271 (79.3515) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1849 data: 0.0906 max mem: 3531 +Test: [140/234] eta: 0:00:19 acc1_pos: 92.414 acc5_pos: 100.000 loss: 0.2976 (0.4692) acc1: 86.7188 (78.8231) acc5: 100.0000 (100.0000) acc1_neg: 38.542 acc5_neg: 100.000 time: 0.1999 data: 0.1036 max mem: 3531 +Test: [150/234] eta: 0:00:17 acc1_pos: 92.414 acc5_pos: 100.000 loss: 0.4087 (0.4692) acc1: 80.7292 (78.9045) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.2149 data: 0.1166 max mem: 3531 +Test: [160/234] eta: 0:00:15 acc1_pos: 92.414 acc5_pos: 100.000 loss: 0.2491 (0.4528) acc1: 92.1875 (79.8848) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2099 data: 0.1112 max mem: 3531 +Test: [170/234] eta: 0:00:13 acc1_pos: 92.414 acc5_pos: 100.000 loss: 0.2487 (0.4404) acc1: 93.2292 (80.6561) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2180 data: 0.1212 max mem: 3531 +Test: [180/234] eta: 0:00:11 acc1_pos: 92.414 acc5_pos: 100.000 loss: 0.2550 (0.4443) acc1: 91.4062 (80.5263) acc5: 100.0000 (100.0000) acc1_neg: 48.958 acc5_neg: 100.000 time: 0.1999 data: 0.1058 max mem: 3531 +Test: [190/234] eta: 0:00:09 acc1_pos: 92.414 acc5_pos: 100.000 loss: 0.6097 (0.4560) acc1: 68.7500 (79.8184) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.1719 data: 0.0781 max mem: 3531 +Test: [200/234] eta: 0:00:06 acc1_pos: 92.414 acc5_pos: 100.000 loss: 0.6097 (0.4480) acc1: 69.0104 (80.2368) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0897 max mem: 3531 +Test: [210/234] eta: 0:00:04 acc1_pos: 92.414 acc5_pos: 100.000 loss: 0.0983 (0.4322) acc1: 100.0000 (81.1488) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1849 data: 0.0877 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 92.414 acc5_pos: 100.000 loss: 0.1484 (0.4201) acc1: 98.1771 (81.8698) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1799 data: 0.0826 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 92.414 acc5_pos: 100.000 loss: 0.2375 (0.4202) acc1: 84.6354 (81.8295) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.1849 data: 0.0888 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 92.414 acc5_pos: 100.000 loss: 0.3929 (0.4200) acc1: 82.2917 (81.8286) acc5: 100.0000 (100.0000) acc1_neg: 81.323 acc5_neg: 100.000 time: 0.1854 data: 0.0887 max mem: 3532 +Test: Total time: 0:00:47 (0.2009 s / it) +* Overall Acc@1 81.829 Acc@5 100.000 + target == real Acc@1 63.627 Acc@5 100.000 + target == fake Acc@1 86.959 Acc@5 100.000 + loss 0.420 + PR-AUC 0.9493 ROC-AUC 0.8506 EER 23.44% (thr=0.7358) +Accuracy of the network on the 89729 test images: 81.8% +[rank0]:[W726 03:01:20.216381476 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:01:40.676000 1167034 torch/distributed/run.py:766] +W0726 03:01:40.676000 1167034 torch/distributed/run.py:766] ***************************************** +W0726 03:01:40.676000 1167034 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:01:40.676000 1167034 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:02:09.998810677 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:02:09.343541997 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:02:10.528265079 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:02:10.849759397 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W726 03:02:10.849830381 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:02:10.869327123 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:02:10.891138482 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:02:10.956095161 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n8_seed5_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed5_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:39:48 acc1_pos: 10.938 acc5_pos: 100.000 loss: 1.5064 (1.5064) acc1: 10.9375 (10.9375) acc5: 100.0000 (100.0000) time: 7.0335 data: 5.3712 max mem: 3531 +Test: [ 10/2387] eta: 0:28:41 acc1_pos: 24.479 acc5_pos: 100.000 loss: 1.1359 (1.1664) acc1: 30.9896 (34.4460) acc5: 100.0000 (100.0000) time: 0.7244 data: 0.4887 max mem: 3531 +Test: [ 20/2387] eta: 0:16:50 acc1_pos: 19.271 acc5_pos: 100.000 loss: 1.0999 (1.1857) acc1: 30.9896 (33.3953) acc5: 100.0000 (100.0000) time: 0.0967 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:15:09 acc1_pos: 14.062 acc5_pos: 100.000 loss: 1.2501 (1.2421) acc1: 27.8646 (30.3175) acc5: 100.0000 (100.0000) time: 0.1999 data: 0.1014 max mem: 3531 +Test: [ 40/2387] eta: 0:14:56 acc1_pos: 15.625 acc5_pos: 100.000 loss: 1.3821 (1.2917) acc1: 21.3542 (27.1341) acc5: 100.0000 (100.0000) time: 0.3349 data: 0.2376 max mem: 3531 +Test: [ 50/2387] eta: 0:14:33 acc1_pos: 13.802 acc5_pos: 100.000 loss: 1.4588 (1.3337) acc1: 16.9271 (25.4085) acc5: 100.0000 (100.0000) time: 0.3549 data: 0.2587 max mem: 3531 +Test: [ 60/2387] eta: 0:13:57 acc1_pos: 27.604 acc5_pos: 100.000 loss: 1.5249 (1.3739) acc1: 16.9271 (24.2999) acc5: 100.0000 (100.0000) time: 0.3149 data: 0.2204 max mem: 3531 +Test: [ 70/2387] eta: 0:13:54 acc1_pos: 27.865 acc5_pos: 100.000 loss: 1.6392 (1.4270) acc1: 15.6250 (23.0414) acc5: 100.0000 (100.0000) time: 0.3250 data: 0.2303 max mem: 3531 +Test: [ 80/2387] eta: 0:13:16 acc1_pos: 2.604 acc5_pos: 100.000 loss: 1.8772 (1.4871) acc1: 8.5938 (21.1163) acc5: 100.0000 (100.0000) time: 0.2997 data: 0.2025 max mem: 3531 +Test: [ 90/2387] eta: 0:12:51 acc1_pos: 0.521 acc5_pos: 100.000 loss: 1.9454 (1.5375) acc1: 4.9479 (19.4568) acc5: 100.0000 (100.0000) time: 0.2496 data: 0.1505 max mem: 3531 +Test: [ 100/2387] eta: 0:12:26 acc1_pos: 4.688 acc5_pos: 100.000 loss: 1.9941 (1.5870) acc1: 2.8646 (17.8888) acc5: 100.0000 (100.0000) time: 0.2499 data: 0.1522 max mem: 3531 +Test: [ 110/2387] eta: 0:12:05 acc1_pos: 15.365 acc5_pos: 100.000 loss: 1.9585 (1.6133) acc1: 5.2083 (16.8802) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1408 max mem: 3531 +Test: [ 120/2387] eta: 0:11:43 acc1_pos: 7.812 acc5_pos: 100.000 loss: 1.9095 (1.6407) acc1: 7.0312 (16.0856) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1300 max mem: 3531 +Test: [ 130/2387] eta: 0:11:28 acc1_pos: 7.031 acc5_pos: 100.000 loss: 1.9755 (1.6680) acc1: 5.9896 (15.1419) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1328 max mem: 3531 +Test: [ 140/2387] eta: 0:11:13 acc1_pos: 1.562 acc5_pos: 100.000 loss: 2.0343 (1.6912) acc1: 2.6042 (14.3414) acc5: 100.0000 (100.0000) time: 0.2349 data: 0.1367 max mem: 3531 +Test: [ 150/2387] eta: 0:10:55 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.9666 (1.7073) acc1: 2.6042 (13.7710) acc5: 100.0000 (100.0000) time: 0.2148 data: 0.1183 max mem: 3531 +Test: [ 160/2387] eta: 0:10:42 acc1_pos: 10.046 acc5_pos: 100.000 loss: 1.9231 (1.6798) acc1: 3.9062 (15.4066) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2099 data: 0.1152 max mem: 3531 +Test: [ 170/2387] eta: 0:10:25 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1964 (1.5923) acc1: 95.5729 (20.1648) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1999 data: 0.1026 max mem: 3531 +Test: [ 180/2387] eta: 0:10:15 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1876 (1.5149) acc1: 95.8333 (24.3526) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1999 data: 0.1032 max mem: 3531 +Test: [ 190/2387] eta: 0:10:02 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1876 (1.4453) acc1: 95.8333 (28.1168) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2049 data: 0.1090 max mem: 3531 +Test: [ 200/2387] eta: 0:09:50 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1824 (1.3824) acc1: 96.6146 (31.5467) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1905 data: 0.0951 max mem: 3531 +Test: [ 210/2387] eta: 0:09:43 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1824 (1.3258) acc1: 96.6146 (34.6181) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2110 data: 0.1179 max mem: 3532 +Test: [ 220/2387] eta: 0:09:34 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1850 (1.2741) acc1: 96.6146 (37.4246) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2143 data: 0.1215 max mem: 3532 +Test: [ 230/2387] eta: 0:09:23 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1822 (1.2268) acc1: 96.6146 (39.9903) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1888 data: 0.0944 max mem: 3532 +Test: [ 240/2387] eta: 0:09:13 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1620 (1.1813) acc1: 97.9167 (42.4382) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1799 data: 0.0828 max mem: 3532 +Test: [ 250/2387] eta: 0:09:04 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1195 (1.1383) acc1: 99.7396 (44.7284) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0831 max mem: 3532 +Test: [ 260/2387] eta: 0:08:53 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.0996 (1.0984) acc1: 100.0000 (46.8461) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1649 data: 0.0687 max mem: 3532 +Test: [ 270/2387] eta: 0:08:46 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.0964 (1.0615) acc1: 100.0000 (48.8075) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0779 max mem: 3532 +Test: [ 280/2387] eta: 0:08:39 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.0953 (1.0272) acc1: 100.0000 (50.6293) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.0986 max mem: 3532 +Test: [ 290/2387] eta: 0:08:34 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.0951 (0.9952) acc1: 100.0000 (52.3259) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1959 data: 0.0976 max mem: 3532 +Test: [ 300/2387] eta: 0:08:26 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.0947 (0.9652) acc1: 100.0000 (53.9097) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1848 data: 0.0887 max mem: 3532 +Test: [ 310/2387] eta: 0:08:19 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.0969 (0.9390) acc1: 100.0000 (55.3599) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1702 data: 0.0769 max mem: 3532 +Test: [ 320/2387] eta: 0:08:16 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.2001 (0.9162) acc1: 98.4375 (56.6978) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2058 data: 0.1127 max mem: 3532 +Test: [ 330/2387] eta: 0:08:13 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.2068 (0.8947) acc1: 98.1771 (57.9447) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2363 data: 0.1432 max mem: 3532 +Test: [ 340/2387] eta: 0:08:08 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.2046 (0.8745) acc1: 98.1771 (59.1276) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2094 data: 0.1162 max mem: 3532 +Test: [ 350/2387] eta: 0:08:06 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1888 (0.8526) acc1: 98.9583 (60.2883) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2141 data: 0.1189 max mem: 3532 +Test: [ 360/2387] eta: 0:08:03 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1273 (0.8330) acc1: 99.4792 (61.3444) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2388 data: 0.1435 max mem: 3532 +Test: [ 370/2387] eta: 0:07:55 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1510 (0.8146) acc1: 98.1771 (62.3336) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1881 data: 0.0944 max mem: 3532 +Test: [ 380/2387] eta: 0:07:51 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1482 (0.7971) acc1: 98.1771 (63.2730) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1663 data: 0.0709 max mem: 3532 +Test: [ 390/2387] eta: 0:07:48 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1482 (0.7805) acc1: 98.1771 (64.1737) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2107 data: 0.1149 max mem: 3532 +Test: [ 400/2387] eta: 0:07:46 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1491 (0.7648) acc1: 98.4375 (65.0197) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2303 data: 0.1370 max mem: 3532 +Test: [ 410/2387] eta: 0:07:41 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1475 (0.7498) acc1: 98.1771 (65.8271) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2134 data: 0.1210 max mem: 3532 +Test: [ 420/2387] eta: 0:07:36 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1472 (0.7355) acc1: 97.9167 (66.5943) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1858 data: 0.0906 max mem: 3532 +Test: [ 430/2387] eta: 0:07:32 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1425 (0.7215) acc1: 98.6979 (67.3482) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1849 data: 0.0904 max mem: 3532 +Test: [ 440/2387] eta: 0:07:28 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1157 (0.7075) acc1: 99.4792 (68.0839) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1886 data: 0.0955 max mem: 3533 +Test: [ 450/2387] eta: 0:07:24 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.0984 (0.6940) acc1: 100.0000 (68.7916) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0968 max mem: 3533 +Test: [ 460/2387] eta: 0:07:20 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.0982 (0.6810) acc1: 100.0000 (69.4685) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1868 data: 0.0952 max mem: 3533 +Test: [ 470/2387] eta: 0:07:16 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.0978 (0.6687) acc1: 100.0000 (70.1168) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1848 data: 0.0929 max mem: 3533 +Test: [ 480/2387] eta: 0:07:12 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.0978 (0.6568) acc1: 100.0000 (70.7380) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1933 data: 0.0986 max mem: 3533 +Test: [ 490/2387] eta: 0:07:09 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.0981 (0.6454) acc1: 100.0000 (71.3340) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1950 data: 0.1015 max mem: 3533 +Test: [ 500/2387] eta: 0:07:05 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.0983 (0.6345) acc1: 100.0000 (71.9062) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1932 data: 0.1012 max mem: 3533 +Test: [ 510/2387] eta: 0:07:02 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.0985 (0.6241) acc1: 100.0000 (72.4560) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1901 data: 0.0985 max mem: 3533 +Test: [ 520/2387] eta: 0:06:59 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1123 (0.6143) acc1: 100.0000 (72.9846) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1964 data: 0.1035 max mem: 3533 +Test: [ 530/2387] eta: 0:06:56 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1129 (0.6048) acc1: 100.0000 (73.4934) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2024 data: 0.1089 max mem: 3533 +Test: [ 540/2387] eta: 0:06:52 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1133 (0.5958) acc1: 100.0000 (73.9834) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1996 data: 0.1089 max mem: 3533 +Test: [ 550/2387] eta: 0:06:49 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1132 (0.5870) acc1: 100.0000 (74.4555) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1932 data: 0.1021 max mem: 3533 +Test: [ 560/2387] eta: 0:06:45 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1132 (0.5786) acc1: 100.0000 (74.9109) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1863 data: 0.0947 max mem: 3533 +Test: [ 570/2387] eta: 0:06:43 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1133 (0.5704) acc1: 100.0000 (75.3503) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2002 data: 0.1087 max mem: 3533 +Test: [ 580/2387] eta: 0:06:40 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1135 (0.5626) acc1: 100.0000 (75.7745) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2024 data: 0.1114 max mem: 3533 +Test: [ 590/2387] eta: 0:06:37 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1137 (0.5550) acc1: 100.0000 (76.1844) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1956 data: 0.1027 max mem: 3533 +Test: [ 600/2387] eta: 0:06:34 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1129 (0.5476) acc1: 100.0000 (76.5790) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1977 data: 0.1036 max mem: 3533 +Test: [ 610/2387] eta: 0:06:31 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1255 (0.5410) acc1: 99.4792 (76.9427) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1979 data: 0.1039 max mem: 3533 +Test: [ 620/2387] eta: 0:06:28 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1386 (0.5345) acc1: 98.9583 (77.2955) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1991 data: 0.1051 max mem: 3533 +Test: [ 630/2387] eta: 0:06:25 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1390 (0.5283) acc1: 98.6979 (77.6355) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1966 data: 0.1051 max mem: 3533 +Test: [ 640/2387] eta: 0:06:22 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1434 (0.5223) acc1: 98.6979 (77.9609) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1840 data: 0.0929 max mem: 3533 +Test: [ 650/2387] eta: 0:06:19 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1399 (0.5164) acc1: 98.6979 (78.2842) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1795 data: 0.0878 max mem: 3533 +Test: [ 660/2387] eta: 0:06:15 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1401 (0.5107) acc1: 98.9583 (78.5954) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1762 data: 0.0837 max mem: 3533 +Test: [ 670/2387] eta: 0:06:12 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1390 (0.5052) acc1: 98.6979 (78.8934) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1701 data: 0.0775 max mem: 3534 +Test: [ 680/2387] eta: 0:06:09 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1360 (0.4995) acc1: 98.6979 (79.1969) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1874 data: 0.0961 max mem: 3534 +Test: [ 690/2387] eta: 0:06:06 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1362 (0.4982) acc1: 98.4375 (79.2643) acc5: 100.0000 (100.0000) acc1_neg: 76.302 acc5_neg: 100.000 time: 0.1882 data: 0.0960 max mem: 3534 +Test: [ 700/2387] eta: 0:06:03 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.5147 (0.4984) acc1: 79.1667 (79.2621) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.1801 data: 0.0869 max mem: 3534 +Test: [ 710/2387] eta: 0:06:01 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.5147 (0.4987) acc1: 79.1667 (79.2513) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.1949 data: 0.1001 max mem: 3534 +Test: [ 720/2387] eta: 0:05:58 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.5137 (0.4989) acc1: 79.1667 (79.2519) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.1848 data: 0.0889 max mem: 3534 +Test: [ 730/2387] eta: 0:05:54 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.5223 (0.4995) acc1: 78.3854 (79.2269) acc5: 100.0000 (100.0000) acc1_neg: 76.302 acc5_neg: 100.000 time: 0.1661 data: 0.0704 max mem: 3534 +Test: [ 740/2387] eta: 0:05:52 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.5281 (0.4999) acc1: 78.1250 (79.2215) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.1862 data: 0.0929 max mem: 3534 +Test: [ 750/2387] eta: 0:05:49 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.5118 (0.5000) acc1: 79.6875 (79.2239) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.1813 data: 0.0906 max mem: 3534 +Test: [ 760/2387] eta: 0:05:46 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.5040 (0.4990) acc1: 79.6875 (79.2772) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1798 data: 0.0880 max mem: 3534 +Test: [ 770/2387] eta: 0:05:43 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.2603 (0.4952) acc1: 90.8854 (79.4784) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1773 data: 0.0846 max mem: 3534 +Test: [ 780/2387] eta: 0:05:40 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.2255 (0.4916) acc1: 94.0104 (79.6692) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1734 data: 0.0817 max mem: 3534 +Test: [ 790/2387] eta: 0:05:38 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1741 (0.4874) acc1: 97.1354 (79.8969) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2029 data: 0.1124 max mem: 3534 +Test: [ 800/2387] eta: 0:05:36 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1641 (0.4836) acc1: 97.3958 (80.1069) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.2022 data: 0.1116 max mem: 3534 +Test: [ 810/2387] eta: 0:05:33 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1641 (0.4796) acc1: 97.9167 (80.3278) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1900 data: 0.0975 max mem: 3534 +Test: [ 820/2387] eta: 0:05:31 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1503 (0.4757) acc1: 97.9167 (80.5423) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1904 data: 0.0980 max mem: 3534 +Test: [ 830/2387] eta: 0:05:28 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1451 (0.4720) acc1: 97.6562 (80.7411) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1891 data: 0.0968 max mem: 3534 +Test: [ 840/2387] eta: 0:05:26 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1405 (0.4680) acc1: 99.4792 (80.9614) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1871 data: 0.0919 max mem: 3534 +Test: [ 850/2387] eta: 0:05:23 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1475 (0.4644) acc1: 98.4375 (81.1573) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1897 data: 0.0950 max mem: 3534 +Test: [ 860/2387] eta: 0:05:21 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1437 (0.4605) acc1: 98.4375 (81.3640) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1935 data: 0.1009 max mem: 3534 +Test: [ 870/2387] eta: 0:05:18 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1337 (0.4568) acc1: 99.2188 (81.5580) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1945 data: 0.1028 max mem: 3534 +Test: [ 880/2387] eta: 0:05:16 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1358 (0.4533) acc1: 99.2188 (81.7540) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1989 data: 0.1079 max mem: 3534 +Test: [ 890/2387] eta: 0:05:14 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1153 (0.4495) acc1: 99.4792 (81.9500) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1940 data: 0.1033 max mem: 3534 +Test: [ 900/2387] eta: 0:05:11 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1378 (0.4504) acc1: 96.8750 (81.8691) acc5: 100.0000 (100.0000) acc1_neg: 69.271 acc5_neg: 100.000 time: 0.1905 data: 0.0995 max mem: 3535 +Test: [ 910/2387] eta: 0:05:09 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.6248 (0.4523) acc1: 68.4896 (81.7260) acc5: 100.0000 (100.0000) acc1_neg: 68.490 acc5_neg: 100.000 time: 0.1901 data: 0.0983 max mem: 3535 +Test: [ 920/2387] eta: 0:05:06 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.6280 (0.4543) acc1: 67.7083 (81.5749) acc5: 100.0000 (100.0000) acc1_neg: 66.667 acc5_neg: 100.000 time: 0.1892 data: 0.0977 max mem: 3535 +Test: [ 930/2387] eta: 0:05:04 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.6280 (0.4562) acc1: 67.4479 (81.4394) acc5: 100.0000 (100.0000) acc1_neg: 65.365 acc5_neg: 100.000 time: 0.1888 data: 0.0972 max mem: 3535 +Test: [ 940/2387] eta: 0:05:02 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.6317 (0.4581) acc1: 69.0104 (81.3040) acc5: 100.0000 (100.0000) acc1_neg: 62.760 acc5_neg: 100.000 time: 0.1875 data: 0.0951 max mem: 3535 +Test: [ 950/2387] eta: 0:04:59 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.6274 (0.4598) acc1: 68.2292 (81.1681) acc5: 100.0000 (100.0000) acc1_neg: 69.531 acc5_neg: 100.000 time: 0.1849 data: 0.0911 max mem: 3535 +Test: [ 960/2387] eta: 0:04:57 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.6162 (0.4614) acc1: 68.7500 (81.0533) acc5: 100.0000 (100.0000) acc1_neg: 74.219 acc5_neg: 100.000 time: 0.1850 data: 0.0920 max mem: 3535 +Test: [ 970/2387] eta: 0:04:54 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.6184 (0.4632) acc1: 68.7500 (80.9199) acc5: 100.0000 (100.0000) acc1_neg: 69.010 acc5_neg: 100.000 time: 0.1849 data: 0.0919 max mem: 3535 +Test: [ 980/2387] eta: 0:04:52 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.6020 (0.4624) acc1: 70.3125 (80.9530) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.1767 data: 0.0836 max mem: 3535 +Test: [ 990/2387] eta: 0:04:50 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.2295 (0.4601) acc1: 92.9688 (81.0834) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1900 data: 0.0975 max mem: 3535 +Test: [1000/2387] eta: 0:04:47 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1498 (0.4571) acc1: 98.1771 (81.2544) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1997 data: 0.1068 max mem: 3535 +Test: [1010/2387] eta: 0:04:45 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1481 (0.4540) acc1: 98.6979 (81.4275) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1779 data: 0.0849 max mem: 3535 +Test: [1020/2387] eta: 0:04:42 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1332 (0.4511) acc1: 99.2188 (81.5849) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1758 data: 0.0832 max mem: 3535 +Test: [1030/2387] eta: 0:04:40 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1479 (0.4482) acc1: 98.9583 (81.7451) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1877 data: 0.0970 max mem: 3535 +Test: [1040/2387] eta: 0:04:37 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1367 (0.4451) acc1: 99.4792 (81.9157) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1801 data: 0.0881 max mem: 3535 +Test: [1050/2387] eta: 0:04:35 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1284 (0.4422) acc1: 99.7396 (82.0793) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1811 data: 0.0880 max mem: 3535 +Test: [1060/2387] eta: 0:04:33 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1271 (0.4392) acc1: 99.7396 (82.2409) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1870 data: 0.0943 max mem: 3535 +Test: [1070/2387] eta: 0:04:30 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1271 (0.4364) acc1: 99.2188 (82.3970) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1778 data: 0.0851 max mem: 3535 +Test: [1080/2387] eta: 0:04:28 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1340 (0.4338) acc1: 98.6979 (82.5410) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1812 data: 0.0888 max mem: 3535 +Test: [1090/2387] eta: 0:04:26 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1945 (0.4321) acc1: 95.0521 (82.6316) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.1758 data: 0.0837 max mem: 3535 +Test: [1100/2387] eta: 0:04:24 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.2387 (0.4306) acc1: 92.9688 (82.7113) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1886 data: 0.0965 max mem: 3535 +Test: [1110/2387] eta: 0:04:21 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.2479 (0.4291) acc1: 92.4479 (82.7952) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.1911 data: 0.0986 max mem: 3535 +Test: [1120/2387] eta: 0:04:19 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.2470 (0.4274) acc1: 92.9688 (82.8899) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1820 data: 0.0889 max mem: 3536 +Test: [1130/2387] eta: 0:04:17 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.2168 (0.4257) acc1: 95.0521 (82.9935) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1880 data: 0.0937 max mem: 3536 +Test: [1140/2387] eta: 0:04:15 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1675 (0.4233) acc1: 98.1771 (83.1288) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1882 data: 0.0934 max mem: 3536 +Test: [1150/2387] eta: 0:04:12 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1306 (0.4208) acc1: 99.4792 (83.2655) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1926 data: 0.0999 max mem: 3536 +Test: [1160/2387] eta: 0:04:10 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1231 (0.4182) acc1: 100.0000 (83.4049) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1955 data: 0.1039 max mem: 3536 +Test: [1170/2387] eta: 0:04:08 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1154 (0.4156) acc1: 100.0000 (83.5462) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1995 data: 0.1081 max mem: 3536 +Test: [1180/2387] eta: 0:04:06 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1133 (0.4131) acc1: 100.0000 (83.6828) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1989 data: 0.1085 max mem: 3536 +Test: [1190/2387] eta: 0:04:04 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1133 (0.4106) acc1: 100.0000 (83.8190) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1974 data: 0.1046 max mem: 3536 +Test: [1200/2387] eta: 0:04:01 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1039 (0.4080) acc1: 100.0000 (83.9535) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1772 data: 0.0827 max mem: 3536 +Test: [1210/2387] eta: 0:03:59 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1004 (0.4055) acc1: 100.0000 (84.0853) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1715 data: 0.0768 max mem: 3536 +Test: [1220/2387] eta: 0:03:57 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1130 (0.4033) acc1: 100.0000 (84.2044) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1903 data: 0.0960 max mem: 3536 +Test: [1230/2387] eta: 0:03:55 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1237 (0.4009) acc1: 99.4792 (84.3280) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2011 data: 0.1088 max mem: 3536 +Test: [1240/2387] eta: 0:03:53 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1198 (0.3987) acc1: 99.2188 (84.4463) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2031 data: 0.1100 max mem: 3536 +Test: [1250/2387] eta: 0:03:51 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1158 (0.3964) acc1: 99.7396 (84.5688) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1986 data: 0.1033 max mem: 3536 +Test: [1260/2387] eta: 0:03:49 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1070 (0.3941) acc1: 99.7396 (84.6885) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2098 data: 0.1160 max mem: 3536 +Test: [1270/2387] eta: 0:03:47 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1054 (0.3919) acc1: 100.0000 (84.8065) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2183 data: 0.1269 max mem: 3536 +Test: [1280/2387] eta: 0:03:45 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1046 (0.3897) acc1: 100.0000 (84.9227) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1963 data: 0.1054 max mem: 3536 +Test: [1290/2387] eta: 0:03:43 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1128 (0.3876) acc1: 99.2188 (85.0334) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1910 data: 0.0983 max mem: 3536 +Test: [1300/2387] eta: 0:03:41 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1148 (0.3855) acc1: 99.2188 (85.1418) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1905 data: 0.0963 max mem: 3536 +Test: [1310/2387] eta: 0:03:38 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1187 (0.3835) acc1: 99.2188 (85.2488) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1851 data: 0.0891 max mem: 3536 +Test: [1320/2387] eta: 0:03:36 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1187 (0.3815) acc1: 99.4792 (85.3587) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1820 data: 0.0856 max mem: 3536 +Test: [1330/2387] eta: 0:03:34 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1168 (0.3796) acc1: 99.7396 (85.4621) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1693 data: 0.0759 max mem: 3536 +Test: [1340/2387] eta: 0:03:32 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1333 (0.3783) acc1: 98.9583 (85.5243) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1726 data: 0.0812 max mem: 3536 +Test: [1350/2387] eta: 0:03:29 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.0911 (0.3761) acc1: 100.0000 (85.6314) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1752 data: 0.0846 max mem: 3537 +Test: [1360/2387] eta: 0:03:27 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.0835 (0.3740) acc1: 100.0000 (85.7370) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1637 data: 0.0700 max mem: 3537 +Test: [1370/2387] eta: 0:03:25 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.0830 (0.3718) acc1: 100.0000 (85.8408) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1588 data: 0.0641 max mem: 3537 +Test: [1380/2387] eta: 0:03:22 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.0830 (0.3698) acc1: 100.0000 (85.9433) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1512 data: 0.0595 max mem: 3537 +Test: [1390/2387] eta: 0:03:20 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.0830 (0.3677) acc1: 100.0000 (86.0444) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1487 data: 0.0580 max mem: 3537 +Test: [1400/2387] eta: 0:03:18 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.0825 (0.3657) acc1: 100.0000 (86.1440) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1599 data: 0.0694 max mem: 3537 +Test: [1410/2387] eta: 0:03:15 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.0861 (0.3644) acc1: 100.0000 (86.2114) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1642 data: 0.0736 max mem: 3537 +Test: [1420/2387] eta: 0:03:13 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1993 (0.3634) acc1: 95.0521 (86.2657) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1532 data: 0.0620 max mem: 3537 +Test: [1430/2387] eta: 0:03:11 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.2129 (0.3624) acc1: 95.0521 (86.3226) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1481 data: 0.0566 max mem: 3537 +Test: [1440/2387] eta: 0:03:08 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1902 (0.3611) acc1: 96.0938 (86.3956) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1537 data: 0.0619 max mem: 3537 +Test: [1450/2387] eta: 0:03:06 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1503 (0.3596) acc1: 98.6979 (86.4853) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1500 data: 0.0586 max mem: 3537 +Test: [1460/2387] eta: 0:03:04 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1254 (0.3580) acc1: 99.4792 (86.5733) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1588 data: 0.0682 max mem: 3537 +Test: [1470/2387] eta: 0:03:02 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1190 (0.3564) acc1: 99.7396 (86.6616) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1645 data: 0.0730 max mem: 3537 +Test: [1480/2387] eta: 0:02:59 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1176 (0.3548) acc1: 99.7396 (86.7471) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1452 data: 0.0507 max mem: 3537 +Test: [1490/2387] eta: 0:02:57 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1190 (0.3532) acc1: 99.4792 (86.8312) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1586 data: 0.0623 max mem: 3537 +Test: [1500/2387] eta: 0:02:55 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1140 (0.3516) acc1: 99.4792 (86.9171) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1798 data: 0.0849 max mem: 3537 +Test: [1510/2387] eta: 0:02:53 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1041 (0.3499) acc1: 100.0000 (87.0028) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1764 data: 0.0828 max mem: 3537 +Test: [1520/2387] eta: 0:02:50 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1047 (0.3484) acc1: 100.0000 (87.0848) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1480 data: 0.0558 max mem: 3537 +Test: [1530/2387] eta: 0:02:48 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1255 (0.3471) acc1: 99.4792 (87.1624) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1399 data: 0.0480 max mem: 3537 +Test: [1540/2387] eta: 0:02:46 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1250 (0.3456) acc1: 99.2188 (87.2384) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1643 data: 0.0723 max mem: 3537 +Test: [1550/2387] eta: 0:02:44 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1192 (0.3442) acc1: 99.7396 (87.3158) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1674 data: 0.0760 max mem: 3537 +Test: [1560/2387] eta: 0:02:42 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1149 (0.3428) acc1: 99.7396 (87.3936) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1722 data: 0.0811 max mem: 3537 +Test: [1570/2387] eta: 0:02:40 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1075 (0.3413) acc1: 99.7396 (87.4705) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0979 max mem: 3537 +Test: [1580/2387] eta: 0:02:38 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1161 (0.3403) acc1: 99.2188 (87.5231) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1866 data: 0.0930 max mem: 3538 +Test: [1590/2387] eta: 0:02:36 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.2327 (0.3398) acc1: 92.9688 (87.5545) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1672 data: 0.0737 max mem: 3538 +Test: [1600/2387] eta: 0:02:34 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.2132 (0.3387) acc1: 95.0521 (87.6160) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1682 data: 0.0766 max mem: 3538 +Test: [1610/2387] eta: 0:02:32 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1779 (0.3379) acc1: 96.8750 (87.6663) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2082 data: 0.1166 max mem: 3538 +Test: [1620/2387] eta: 0:02:30 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.2052 (0.3371) acc1: 95.0521 (87.7127) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2117 data: 0.1183 max mem: 3538 +Test: [1630/2387] eta: 0:02:28 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1911 (0.3363) acc1: 95.8333 (87.7561) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1743 data: 0.0816 max mem: 3538 +Test: [1640/2387] eta: 0:02:26 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1580 (0.3353) acc1: 97.3958 (87.8182) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1855 data: 0.0938 max mem: 3538 +Test: [1650/2387] eta: 0:02:24 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1488 (0.3341) acc1: 98.1771 (87.8801) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1873 data: 0.0964 max mem: 3538 +Test: [1660/2387] eta: 0:02:22 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1445 (0.3332) acc1: 97.9167 (87.9352) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1824 data: 0.0889 max mem: 3538 +Test: [1670/2387] eta: 0:02:20 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1522 (0.3321) acc1: 98.1771 (87.9937) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1774 data: 0.0825 max mem: 3538 +Test: [1680/2387] eta: 0:02:18 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1428 (0.3311) acc1: 99.2188 (88.0483) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1649 data: 0.0717 max mem: 3538 +Test: [1690/2387] eta: 0:02:16 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1541 (0.3301) acc1: 98.6979 (88.1063) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1759 data: 0.0814 max mem: 3538 +Test: [1700/2387] eta: 0:02:14 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1548 (0.3291) acc1: 98.1771 (88.1655) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1821 data: 0.0891 max mem: 3538 +Test: [1710/2387] eta: 0:02:12 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1665 (0.3282) acc1: 97.1354 (88.2137) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1908 data: 0.1008 max mem: 3538 +Test: [1720/2387] eta: 0:02:10 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1665 (0.3272) acc1: 96.8750 (88.2654) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2027 data: 0.1120 max mem: 3538 +Test: [1730/2387] eta: 0:02:08 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1590 (0.3263) acc1: 97.3958 (88.3178) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1942 data: 0.1029 max mem: 3538 +Test: [1740/2387] eta: 0:02:06 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1619 (0.3254) acc1: 97.3958 (88.3703) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1953 data: 0.1045 max mem: 3538 +Test: [1750/2387] eta: 0:02:04 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1554 (0.3244) acc1: 97.9167 (88.4245) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2048 data: 0.1144 max mem: 3538 +Test: [1760/2387] eta: 0:02:02 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1341 (0.3233) acc1: 98.6979 (88.4837) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1985 data: 0.1074 max mem: 3538 +Test: [1770/2387] eta: 0:02:00 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1370 (0.3224) acc1: 98.4375 (88.5342) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1909 data: 0.0975 max mem: 3538 +Test: [1780/2387] eta: 0:01:58 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1629 (0.3216) acc1: 97.1354 (88.5768) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1927 data: 0.0996 max mem: 3538 +Test: [1790/2387] eta: 0:01:56 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1629 (0.3206) acc1: 97.1354 (88.6295) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1949 data: 0.1033 max mem: 3538 +Test: [1800/2387] eta: 0:01:54 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1672 (0.3198) acc1: 96.8750 (88.6727) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1964 data: 0.1040 max mem: 3538 +Test: [1810/2387] eta: 0:01:52 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1454 (0.3187) acc1: 98.4375 (88.7300) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2008 data: 0.1084 max mem: 3539 +Test: [1820/2387] eta: 0:01:50 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1221 (0.3176) acc1: 99.2188 (88.7876) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1919 data: 0.0994 max mem: 3539 +Test: [1830/2387] eta: 0:01:48 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1072 (0.3165) acc1: 99.7396 (88.8462) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1823 data: 0.0900 max mem: 3539 +Test: [1840/2387] eta: 0:01:46 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.0887 (0.3153) acc1: 100.0000 (88.9068) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1850 data: 0.0935 max mem: 3539 +Test: [1850/2387] eta: 0:01:44 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.0886 (0.3140) acc1: 100.0000 (88.9667) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1912 data: 0.0998 max mem: 3539 +Test: [1860/2387] eta: 0:01:42 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.0891 (0.3128) acc1: 100.0000 (89.0260) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1968 data: 0.1036 max mem: 3539 +Test: [1870/2387] eta: 0:01:41 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.0887 (0.3116) acc1: 100.0000 (89.0846) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2052 data: 0.1123 max mem: 3539 +Test: [1880/2387] eta: 0:01:39 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.0887 (0.3104) acc1: 100.0000 (89.1427) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1931 data: 0.1010 max mem: 3539 +Test: [1890/2387] eta: 0:01:37 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.0885 (0.3093) acc1: 100.0000 (89.2001) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1885 data: 0.0963 max mem: 3539 +Test: [1900/2387] eta: 0:01:35 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.0886 (0.3081) acc1: 100.0000 (89.2569) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1958 data: 0.1048 max mem: 3539 +Test: [1910/2387] eta: 0:01:33 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.0889 (0.3070) acc1: 100.0000 (89.3107) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1943 data: 0.1034 max mem: 3539 +Test: [1920/2387] eta: 0:01:31 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1202 (0.3069) acc1: 97.6562 (89.3125) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2038 data: 0.1127 max mem: 3539 +Test: [1930/2387] eta: 0:01:29 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.2162 (0.3063) acc1: 94.5312 (89.3452) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1974 data: 0.1064 max mem: 3539 +Test: [1940/2387] eta: 0:01:27 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.2115 (0.3058) acc1: 97.1354 (89.3860) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1954 data: 0.1043 max mem: 3539 +Test: [1950/2387] eta: 0:01:25 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1865 (0.3054) acc1: 97.3958 (89.4087) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1990 data: 0.1079 max mem: 3539 +Test: [1960/2387] eta: 0:01:23 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1966 (0.3050) acc1: 96.0938 (89.4399) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1955 data: 0.1049 max mem: 3539 +Test: [1970/2387] eta: 0:01:21 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.2180 (0.3048) acc1: 94.7917 (89.4439) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.2030 data: 0.1103 max mem: 3539 +Test: [1980/2387] eta: 0:01:19 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1740 (0.3039) acc1: 97.3958 (89.4868) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2019 data: 0.1066 max mem: 3539 +Test: [1990/2387] eta: 0:01:17 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.0981 (0.3029) acc1: 99.7396 (89.5391) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1972 data: 0.1040 max mem: 3539 +Test: [2000/2387] eta: 0:01:15 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.0966 (0.3019) acc1: 100.0000 (89.5911) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2081 data: 0.1168 max mem: 3539 +Test: [2010/2387] eta: 0:01:13 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.0961 (0.3009) acc1: 100.0000 (89.6424) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1996 data: 0.1078 max mem: 3539 +Test: [2020/2387] eta: 0:01:11 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.0971 (0.2998) acc1: 100.0000 (89.6929) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1957 data: 0.1044 max mem: 3539 +Test: [2030/2387] eta: 0:01:09 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.0969 (0.2988) acc1: 100.0000 (89.7435) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1997 data: 0.1087 max mem: 3540 +Test: [2040/2387] eta: 0:01:07 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.0961 (0.2979) acc1: 100.0000 (89.7931) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1909 data: 0.0970 max mem: 3540 +Test: [2050/2387] eta: 0:01:05 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.0957 (0.2969) acc1: 100.0000 (89.8422) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1880 data: 0.0884 max mem: 3540 +Test: [2060/2387] eta: 0:01:03 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.0945 (0.2959) acc1: 100.0000 (89.8915) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1888 data: 0.0914 max mem: 3540 +Test: [2070/2387] eta: 0:01:01 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.0941 (0.2949) acc1: 100.0000 (89.9402) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1828 data: 0.0913 max mem: 3540 +Test: [2080/2387] eta: 0:00:59 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.0936 (0.2940) acc1: 100.0000 (89.9885) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1625 data: 0.0715 max mem: 3540 +Test: [2090/2387] eta: 0:00:57 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.0944 (0.2930) acc1: 100.0000 (90.0360) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1620 data: 0.0709 max mem: 3540 +Test: [2100/2387] eta: 0:00:55 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.0944 (0.2921) acc1: 100.0000 (90.0833) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1711 data: 0.0790 max mem: 3540 +Test: [2110/2387] eta: 0:00:53 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.0939 (0.2911) acc1: 100.0000 (90.1303) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1649 data: 0.0724 max mem: 3540 +Test: [2120/2387] eta: 0:00:51 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.0948 (0.2907) acc1: 100.0000 (90.1545) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.1588 data: 0.0651 max mem: 3540 +Test: [2130/2387] eta: 0:00:50 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1698 (0.2902) acc1: 97.3958 (90.1840) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0825 max mem: 3540 +Test: [2140/2387] eta: 0:00:48 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1698 (0.2897) acc1: 97.6562 (90.2223) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1902 data: 0.0930 max mem: 3540 +Test: [2150/2387] eta: 0:00:46 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1662 (0.2891) acc1: 98.1771 (90.2550) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1564 data: 0.0612 max mem: 3540 +Test: [2160/2387] eta: 0:00:44 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1651 (0.2885) acc1: 98.1771 (90.2923) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1393 data: 0.0449 max mem: 3540 +Test: [2170/2387] eta: 0:00:42 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1604 (0.2880) acc1: 97.9167 (90.3264) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1496 data: 0.0537 max mem: 3540 +Test: [2180/2387] eta: 0:00:40 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1739 (0.2886) acc1: 97.1354 (90.2716) acc5: 100.0000 (100.0000) acc1_neg: 50.260 acc5_neg: 100.000 time: 0.1599 data: 0.0636 max mem: 3540 +Test: [2190/2387] eta: 0:00:38 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1798 (0.2885) acc1: 95.5729 (90.2693) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1736 data: 0.0784 max mem: 3540 +Test: [2200/2387] eta: 0:00:36 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1566 (0.2879) acc1: 97.3958 (90.3081) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1648 data: 0.0674 max mem: 3540 +Test: [2210/2387] eta: 0:00:34 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1410 (0.2873) acc1: 99.2188 (90.3461) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1548 data: 0.0597 max mem: 3540 +Test: [2220/2387] eta: 0:00:32 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1410 (0.2867) acc1: 99.2188 (90.3838) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1558 data: 0.0640 max mem: 3540 +Test: [2230/2387] eta: 0:00:30 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1372 (0.2860) acc1: 99.7396 (90.4233) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1604 data: 0.0676 max mem: 3540 +Test: [2240/2387] eta: 0:00:28 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1375 (0.2854) acc1: 99.7396 (90.4629) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1694 data: 0.0766 max mem: 3540 +Test: [2250/2387] eta: 0:00:26 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1262 (0.2846) acc1: 99.7396 (90.5013) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1702 data: 0.0751 max mem: 3540 +Test: [2260/2387] eta: 0:00:24 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1139 (0.2839) acc1: 100.0000 (90.5430) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1696 data: 0.0754 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1262 (0.2833) acc1: 99.7396 (90.5830) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1690 data: 0.0765 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1307 (0.2826) acc1: 99.7396 (90.6217) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1578 data: 0.0639 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1142 (0.2819) acc1: 100.0000 (90.6615) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1706 data: 0.0778 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1211 (0.2812) acc1: 100.0000 (90.6997) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2055 data: 0.1130 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1169 (0.2805) acc1: 100.0000 (90.7367) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1906 data: 0.0986 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1294 (0.2799) acc1: 99.2188 (90.7687) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1575 data: 0.0642 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1574 (0.2797) acc1: 96.8750 (90.7826) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1606 data: 0.0671 max mem: 3541 +Test: [2340/2387] eta: 0:00:09 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.2116 (0.2795) acc1: 95.3125 (90.7985) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1867 data: 0.0946 max mem: 3541 +Test: [2350/2387] eta: 0:00:07 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1827 (0.2792) acc1: 97.1354 (90.8214) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1816 data: 0.0887 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1774 (0.2788) acc1: 97.1354 (90.8464) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1689 data: 0.0777 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1773 (0.2785) acc1: 97.1354 (90.8636) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.1718 data: 0.0813 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1666 (0.2780) acc1: 97.9167 (90.8917) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1559 data: 0.0650 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 10.046 acc5_pos: 100.000 loss: 0.1366 (0.2776) acc1: 97.3958 (90.9068) acc5: 100.0000 (100.0000) acc1_neg: 95.349 acc5_neg: 100.000 time: 0.1392 data: 0.0509 max mem: 3541 +Test: Total time: 0:07:37 (0.1917 s / it) +* Overall Acc@1 90.907 Acc@5 100.000 + target == real Acc@1 13.374 Acc@5 100.000 + target == fake Acc@1 96.352 Acc@5 100.000 + loss 0.278 + PR-AUC 0.9585 ROC-AUC 0.6842 EER 34.69% (thr=0.8848) +Accuracy of the network on the 916267 test images: 90.9% +[rank0]:[W726 03:10:06.372836324 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:10:27.929000 1167665 torch/distributed/run.py:766] +W0726 03:10:27.929000 1167665 torch/distributed/run.py:766] ***************************************** +W0726 03:10:27.929000 1167665 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:10:27.929000 1167665 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:10:58.466656789 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:10:59.967294604 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:10:59.031031343 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:11:00.534101319 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:11:00.617144587 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:11:00.620855250 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:11:00.637099475 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:11:00.732492160 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n8_seed5_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed5_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:05:26 acc1_pos: 10.938 acc5_pos: 100.000 loss: 1.5064 (1.5064) acc1: 10.9375 (10.9375) acc5: 100.0000 (100.0000) time: 7.5939 data: 5.9278 max mem: 3531 +Test: [ 10/517] eta: 0:06:32 acc1_pos: 24.479 acc5_pos: 100.000 loss: 1.1359 (1.1664) acc1: 30.9896 (34.4460) acc5: 100.0000 (100.0000) time: 0.7746 data: 0.5394 max mem: 3531 +Test: [ 20/517] eta: 0:03:45 acc1_pos: 19.271 acc5_pos: 100.000 loss: 1.0999 (1.1857) acc1: 30.9896 (33.3953) acc5: 100.0000 (100.0000) time: 0.0962 data: 0.0005 max mem: 3531 +Test: [ 30/517] eta: 0:03:07 acc1_pos: 14.062 acc5_pos: 100.000 loss: 1.2501 (1.2421) acc1: 27.8646 (30.3175) acc5: 100.0000 (100.0000) time: 0.1718 data: 0.0742 max mem: 3531 +Test: [ 40/517] eta: 0:02:51 acc1_pos: 15.625 acc5_pos: 100.000 loss: 1.3821 (1.2917) acc1: 21.3542 (27.1341) acc5: 100.0000 (100.0000) time: 0.2599 data: 0.1647 max mem: 3531 +Test: [ 50/517] eta: 0:02:40 acc1_pos: 5.505 acc5_pos: 100.000 loss: 1.3774 (1.2664) acc1: 20.0521 (29.1565) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2780 data: 0.1823 max mem: 3531 +Test: [ 60/517] eta: 0:02:29 acc1_pos: 5.505 acc5_pos: 100.000 loss: 0.2419 (1.0947) acc1: 94.7917 (40.1341) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.2649 data: 0.1691 max mem: 3531 +Test: [ 70/517] eta: 0:02:22 acc1_pos: 5.505 acc5_pos: 100.000 loss: 0.2311 (0.9830) acc1: 95.5729 (47.3408) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.2549 data: 0.1596 max mem: 3531 +Test: [ 80/517] eta: 0:02:15 acc1_pos: 5.505 acc5_pos: 100.000 loss: 0.3103 (0.9048) acc1: 89.3229 (52.5045) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.2549 data: 0.1592 max mem: 3531 +Test: [ 90/517] eta: 0:02:09 acc1_pos: 5.505 acc5_pos: 100.000 loss: 0.3579 (0.8451) acc1: 85.9375 (56.2157) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.2499 data: 0.1557 max mem: 3531 +Test: [100/517] eta: 0:02:05 acc1_pos: 5.505 acc5_pos: 100.000 loss: 0.3456 (0.7956) acc1: 87.7604 (59.4059) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.2599 data: 0.1660 max mem: 3531 +Test: [110/517] eta: 0:02:01 acc1_pos: 5.505 acc5_pos: 100.000 loss: 0.2945 (0.7456) acc1: 92.4479 (62.6150) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2749 data: 0.1814 max mem: 3531 +Test: [120/517] eta: 0:01:57 acc1_pos: 5.505 acc5_pos: 100.000 loss: 0.3268 (0.7162) acc1: 89.0625 (64.3896) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.2749 data: 0.1799 max mem: 3531 +Test: [130/517] eta: 0:01:52 acc1_pos: 5.505 acc5_pos: 100.000 loss: 0.3900 (0.6888) acc1: 84.6354 (66.1796) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2499 data: 0.1536 max mem: 3531 +Test: [140/517] eta: 0:01:48 acc1_pos: 5.505 acc5_pos: 100.000 loss: 0.3325 (0.6637) acc1: 88.5417 (67.8967) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.2400 data: 0.1448 max mem: 3531 +Test: [150/517] eta: 0:01:44 acc1_pos: 5.505 acc5_pos: 100.000 loss: 0.3072 (0.6373) acc1: 91.9271 (69.6606) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2449 data: 0.1492 max mem: 3531 +Test: [160/517] eta: 0:01:40 acc1_pos: 5.505 acc5_pos: 100.000 loss: 0.2618 (0.6139) acc1: 92.4479 (71.1520) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2399 data: 0.1431 max mem: 3531 +Test: [170/517] eta: 0:01:37 acc1_pos: 5.505 acc5_pos: 100.000 loss: 0.2374 (0.5920) acc1: 94.2708 (72.5451) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2399 data: 0.1431 max mem: 3531 +Test: [180/517] eta: 0:01:33 acc1_pos: 5.505 acc5_pos: 100.000 loss: 0.2447 (0.5730) acc1: 94.0104 (73.6591) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.2449 data: 0.1470 max mem: 3531 +Test: [190/517] eta: 0:01:30 acc1_pos: 5.505 acc5_pos: 100.000 loss: 0.1183 (0.5487) acc1: 97.1354 (75.0382) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2449 data: 0.1472 max mem: 3531 +Test: [200/517] eta: 0:01:27 acc1_pos: 5.505 acc5_pos: 100.000 loss: 0.1130 (0.5325) acc1: 100.0000 (76.0093) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.2499 data: 0.1552 max mem: 3531 +Test: [210/517] eta: 0:01:24 acc1_pos: 5.505 acc5_pos: 100.000 loss: 0.2851 (0.5253) acc1: 89.8438 (76.4378) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.2749 data: 0.1803 max mem: 3532 +Test: [220/517] eta: 0:01:21 acc1_pos: 5.505 acc5_pos: 100.000 loss: 0.3863 (0.5189) acc1: 84.6354 (76.8465) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.2799 data: 0.1868 max mem: 3532 +Test: [230/517] eta: 0:01:18 acc1_pos: 5.505 acc5_pos: 100.000 loss: 0.3452 (0.5092) acc1: 88.0208 (77.5433) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.2549 data: 0.1626 max mem: 3532 +Test: [240/517] eta: 0:01:15 acc1_pos: 5.505 acc5_pos: 100.000 loss: 0.2737 (0.5007) acc1: 94.0104 (78.0494) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.2601 data: 0.1672 max mem: 3532 +Test: [250/517] eta: 0:01:12 acc1_pos: 5.505 acc5_pos: 100.000 loss: 0.2611 (0.4919) acc1: 93.4896 (78.6230) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2499 data: 0.1581 max mem: 3532 +Test: [260/517] eta: 0:01:09 acc1_pos: 5.505 acc5_pos: 100.000 loss: 0.2558 (0.4835) acc1: 94.2708 (79.2116) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.2251 data: 0.1318 max mem: 3532 +Test: [270/517] eta: 0:01:06 acc1_pos: 5.505 acc5_pos: 100.000 loss: 0.3598 (0.4807) acc1: 89.0625 (79.4521) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.2349 data: 0.1401 max mem: 3532 +Test: [280/517] eta: 0:01:03 acc1_pos: 5.505 acc5_pos: 100.000 loss: 0.3724 (0.4756) acc1: 87.5000 (79.8163) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.2544 data: 0.1606 max mem: 3532 +Test: [290/517] eta: 0:01:00 acc1_pos: 5.505 acc5_pos: 100.000 loss: 0.2811 (0.4681) acc1: 92.7083 (80.3426) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2549 data: 0.1611 max mem: 3532 +Test: [300/517] eta: 0:00:57 acc1_pos: 5.505 acc5_pos: 100.000 loss: 0.2429 (0.4606) acc1: 95.8333 (80.8607) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2299 data: 0.1355 max mem: 3532 +Test: [310/517] eta: 0:00:55 acc1_pos: 5.505 acc5_pos: 100.000 loss: 0.2330 (0.4533) acc1: 95.8333 (81.3522) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2449 data: 0.1508 max mem: 3532 +Test: [320/517] eta: 0:00:52 acc1_pos: 5.505 acc5_pos: 100.000 loss: 0.2401 (0.4465) acc1: 95.5729 (81.7798) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.2449 data: 0.1529 max mem: 3532 +Test: [330/517] eta: 0:00:49 acc1_pos: 5.505 acc5_pos: 100.000 loss: 0.2251 (0.4396) acc1: 95.3125 (82.2059) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2449 data: 0.1525 max mem: 3532 +Test: [340/517] eta: 0:00:46 acc1_pos: 5.505 acc5_pos: 100.000 loss: 0.2010 (0.4343) acc1: 97.1354 (82.4650) acc5: 100.0000 (100.0000) acc1_neg: 69.792 acc5_neg: 100.000 time: 0.2649 data: 0.1677 max mem: 3532 +Test: [350/517] eta: 0:00:44 acc1_pos: 5.505 acc5_pos: 100.000 loss: 0.1889 (0.4292) acc1: 97.6562 (82.7799) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2550 data: 0.1579 max mem: 3532 +Test: [360/517] eta: 0:00:41 acc1_pos: 5.505 acc5_pos: 100.000 loss: 0.2074 (0.4232) acc1: 96.8750 (83.1681) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2449 data: 0.1508 max mem: 3532 +Test: [370/517] eta: 0:00:38 acc1_pos: 5.505 acc5_pos: 100.000 loss: 0.1876 (0.4165) acc1: 97.9167 (83.5860) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2398 data: 0.1444 max mem: 3532 +Test: [380/517] eta: 0:00:36 acc1_pos: 5.505 acc5_pos: 100.000 loss: 0.1984 (0.4119) acc1: 97.6562 (83.8289) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2649 data: 0.1684 max mem: 3532 +Test: [390/517] eta: 0:00:33 acc1_pos: 5.505 acc5_pos: 100.000 loss: 0.2324 (0.4075) acc1: 93.7500 (84.0773) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2649 data: 0.1695 max mem: 3532 +Test: [400/517] eta: 0:00:30 acc1_pos: 5.505 acc5_pos: 100.000 loss: 0.1995 (0.4019) acc1: 95.5729 (84.4159) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2399 data: 0.1443 max mem: 3532 +Test: [410/517] eta: 0:00:28 acc1_pos: 5.505 acc5_pos: 100.000 loss: 0.1978 (0.3984) acc1: 95.8333 (84.6519) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.2549 data: 0.1554 max mem: 3532 +Test: [420/517] eta: 0:00:25 acc1_pos: 5.505 acc5_pos: 100.000 loss: 0.2149 (0.3933) acc1: 95.5729 (84.9806) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2699 data: 0.1708 max mem: 3532 +Test: [430/517] eta: 0:00:22 acc1_pos: 5.505 acc5_pos: 100.000 loss: 0.2149 (0.3911) acc1: 97.6562 (85.1490) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.2649 data: 0.1695 max mem: 3532 +Test: [440/517] eta: 0:00:20 acc1_pos: 5.505 acc5_pos: 100.000 loss: 0.2926 (0.3899) acc1: 92.1875 (85.2236) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.2549 data: 0.1588 max mem: 3533 +Test: [450/517] eta: 0:00:17 acc1_pos: 5.505 acc5_pos: 100.000 loss: 0.2518 (0.3857) acc1: 93.4896 (85.5085) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2399 data: 0.1440 max mem: 3533 +Test: [460/517] eta: 0:00:14 acc1_pos: 5.505 acc5_pos: 100.000 loss: 0.2078 (0.3829) acc1: 97.3958 (85.6760) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.2550 data: 0.1590 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 5.505 acc5_pos: 100.000 loss: 0.1886 (0.3788) acc1: 97.3958 (85.9309) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2749 data: 0.1768 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 5.505 acc5_pos: 100.000 loss: 0.1820 (0.3745) acc1: 98.6979 (86.1930) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2548 data: 0.1582 max mem: 3533 +Test: [490/517] eta: 0:00:07 acc1_pos: 5.505 acc5_pos: 100.000 loss: 0.1722 (0.3706) acc1: 99.2188 (86.4398) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2499 data: 0.1559 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 5.505 acc5_pos: 100.000 loss: 0.2379 (0.3695) acc1: 94.2708 (86.5233) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.2597 data: 0.1673 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 5.505 acc5_pos: 100.000 loss: 0.2834 (0.3668) acc1: 91.6667 (86.6596) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2215 data: 0.1313 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 5.505 acc5_pos: 100.000 loss: 0.1420 (0.3641) acc1: 99.2188 (86.8042) acc5: 100.0000 (100.0000) acc1_neg: 99.647 acc5_neg: 100.000 time: 0.1785 data: 0.0873 max mem: 3533 +Test: Total time: 0:02:14 (0.2607 s / it) +* Overall Acc@1 86.804 Acc@5 100.000 + target == real Acc@1 25.796 Acc@5 100.000 + target == fake Acc@1 93.133 Acc@5 100.000 + loss 0.364 + PR-AUC 0.9401 ROC-AUC 0.6744 EER 37.06% (thr=0.8110) +Accuracy of the network on the 198427 test images: 86.8% +[rank0]:[W726 03:13:26.028935297 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 03:13:46.210000 1168238 torch/distributed/run.py:766] +W0726 03:13:46.210000 1168238 torch/distributed/run.py:766] ***************************************** +W0726 03:13:46.210000 1168238 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:13:46.210000 1168238 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:14:15.205199316 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:14:16.490217432 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:14:16.562505516 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:14:16.680021831 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:14:16.873787294 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +| distributed init (rank 6): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:14:16.904744733 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:14:16.908116227 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank6]:[W726 03:14:16.908120747 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n8_seed5_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed5_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:38:06 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.2343 (2.2343) acc1: 0.0000 (0.0000) acc5: 100.0000 (100.0000) time: 5.8778 data: 4.3092 max mem: 3531 +Test: [ 10/389] eta: 0:03:54 acc1_pos: 4.688 acc5_pos: 100.000 loss: 2.2094 (2.2657) acc1: 0.0000 (0.9233) acc5: 100.0000 (100.0000) time: 0.6183 data: 0.3921 max mem: 3531 +Test: [ 20/389] eta: 0:02:16 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.1972 (2.1493) acc1: 0.2604 (1.9469) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0004 max mem: 3531 +Test: [ 30/389] eta: 0:01:42 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.3632 (2.2337) acc1: 0.0000 (1.3525) acc5: 100.0000 (100.0000) time: 0.1009 data: 0.0025 max mem: 3531 +Test: [ 40/389] eta: 0:01:30 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.3581 (2.2180) acc1: 0.0000 (1.2576) acc5: 100.0000 (100.0000) time: 0.1399 data: 0.0433 max mem: 3531 +Test: [ 50/389] eta: 0:01:21 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.1867 (2.2167) acc1: 0.0000 (1.0417) acc5: 100.0000 (100.0000) time: 0.1714 data: 0.0776 max mem: 3531 +Test: [ 60/389] eta: 0:01:13 acc1_pos: 3.125 acc5_pos: 100.000 loss: 1.9172 (2.1198) acc1: 0.2604 (2.6853) acc5: 100.0000 (100.0000) time: 0.1555 data: 0.0640 max mem: 3531 +Test: [ 70/389] eta: 0:01:08 acc1_pos: 12.240 acc5_pos: 100.000 loss: 1.6777 (2.0686) acc1: 4.1667 (3.3487) acc5: 100.0000 (100.0000) time: 0.1550 data: 0.0627 max mem: 3531 +Test: [ 80/389] eta: 0:01:09 acc1_pos: 38.021 acc5_pos: 100.000 loss: 1.7036 (2.0089) acc1: 4.6875 (5.7677) acc5: 100.0000 (100.0000) time: 0.2205 data: 0.1283 max mem: 3531 +Test: [ 90/389] eta: 0:01:07 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.7522 (1.9486) acc1: 13.0208 (7.8840) acc5: 100.0000 (100.0000) time: 0.2542 data: 0.1631 max mem: 3531 +Test: [100/389] eta: 0:01:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.8958 (1.9722) acc1: 1.8229 (7.9053) acc5: 100.0000 (100.0000) time: 0.1960 data: 0.1044 max mem: 3531 +Test: [110/389] eta: 0:01:00 acc1_pos: 73.177 acc5_pos: 100.000 loss: 2.3205 (1.9510) acc1: 0.0000 (9.5603) acc5: 100.0000 (100.0000) time: 0.1807 data: 0.0875 max mem: 3531 +Test: [120/389] eta: 0:00:57 acc1_pos: 1.823 acc5_pos: 100.000 loss: 1.8154 (1.9314) acc1: 11.9792 (10.3994) acc5: 100.0000 (100.0000) time: 0.1933 data: 0.1006 max mem: 3531 +Test: [130/389] eta: 0:00:54 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.9150 (1.8203) acc1: 45.8333 (15.5852) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1818 data: 0.0901 max mem: 3531 +Test: [140/389] eta: 0:00:52 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.0935 (1.7047) acc1: 100.0000 (21.1067) acc5: 100.0000 (100.0000) acc1_neg: 69.010 acc5_neg: 100.000 time: 0.2043 data: 0.1115 max mem: 3531 +Test: [150/389] eta: 0:00:50 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.0923 (1.6057) acc1: 100.0000 (25.8002) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2070 data: 0.1134 max mem: 3531 +Test: [160/389] eta: 0:00:48 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.0923 (1.5117) acc1: 100.0000 (30.4089) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2069 data: 0.1136 max mem: 3531 +Test: [170/389] eta: 0:00:46 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.0768 (1.4281) acc1: 100.0000 (34.4725) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2068 data: 0.1154 max mem: 3531 +Test: [180/389] eta: 0:00:44 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.0764 (1.3557) acc1: 100.0000 (37.9316) acc5: 100.0000 (100.0000) acc1_neg: 70.833 acc5_neg: 100.000 time: 0.2179 data: 0.1252 max mem: 3531 +Test: [190/389] eta: 0:00:42 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.0790 (1.2942) acc1: 100.0000 (40.9086) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2214 data: 0.1265 max mem: 3531 +Test: [200/389] eta: 0:00:39 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.1034 (1.2360) acc1: 100.0000 (43.7953) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1649 data: 0.0694 max mem: 3531 +Test: [210/389] eta: 0:00:37 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.0882 (1.1816) acc1: 100.0000 (46.4591) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1656 data: 0.0722 max mem: 3532 +Test: [220/389] eta: 0:00:34 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.0880 (1.1325) acc1: 100.0000 (48.8817) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1958 data: 0.1045 max mem: 3532 +Test: [230/389] eta: 0:00:32 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.0916 (1.0910) acc1: 100.0000 (50.9425) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.1902 data: 0.0986 max mem: 3532 +Test: [240/389] eta: 0:00:30 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.1357 (1.0522) acc1: 100.0000 (52.9208) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1740 data: 0.0815 max mem: 3532 +Test: [250/389] eta: 0:00:28 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.1077 (1.0145) acc1: 100.0000 (54.7944) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1786 data: 0.0854 max mem: 3532 +Test: [260/389] eta: 0:00:26 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.0827 (0.9826) acc1: 100.0000 (56.2640) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0873 max mem: 3532 +Test: [270/389] eta: 0:00:23 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.0847 (0.9497) acc1: 100.0000 (57.8759) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1736 data: 0.0804 max mem: 3532 +Test: [280/389] eta: 0:00:21 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.0844 (0.9194) acc1: 100.0000 (59.3750) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1801 data: 0.0869 max mem: 3532 +Test: [290/389] eta: 0:00:19 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.1038 (0.8944) acc1: 100.0000 (60.6350) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1812 data: 0.0884 max mem: 3532 +Test: [300/389] eta: 0:00:17 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.1011 (0.8678) acc1: 100.0000 (61.9428) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1905 data: 0.0968 max mem: 3532 +Test: [310/389] eta: 0:00:15 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.0928 (0.8430) acc1: 100.0000 (63.1649) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2031 data: 0.1081 max mem: 3532 +Test: [320/389] eta: 0:00:13 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.0918 (0.8197) acc1: 100.0000 (64.3124) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1957 data: 0.1018 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.0915 (0.7977) acc1: 100.0000 (65.3874) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1725 data: 0.0802 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.0914 (0.7769) acc1: 100.0000 (66.3940) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1832 data: 0.0894 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.1021 (0.7647) acc1: 100.0000 (66.8967) acc5: 100.0000 (100.0000) acc1_neg: 54.167 acc5_neg: 100.000 time: 0.2162 data: 0.1219 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.2426 (0.7531) acc1: 94.7917 (67.4364) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1859 data: 0.0935 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.1974 (0.7364) acc1: 97.1354 (68.3064) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1522 data: 0.0611 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.1040 (0.7198) acc1: 100.0000 (69.1382) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1411 data: 0.0505 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.1038 (0.7077) acc1: 100.0000 (69.7351) acc5: 100.0000 (100.0000) acc1_neg: 86.339 acc5_neg: 100.000 time: 0.1287 data: 0.0350 max mem: 3532 +Test: Total time: 0:01:15 (0.1938 s / it) +* Overall Acc@1 69.735 Acc@5 100.000 + target == real Acc@1 10.410 Acc@5 100.000 + target == fake Acc@1 97.311 Acc@5 100.000 + loss 0.708 + PR-AUC 0.7760 ROC-AUC 0.6642 EER 37.68% (thr=0.9058) +Accuracy of the network on the 149358 test images: 69.7% +[rank0]:[W726 03:15:39.380910312 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:16:00.672000 1168796 torch/distributed/run.py:766] +W0726 03:16:00.672000 1168796 torch/distributed/run.py:766] ***************************************** +W0726 03:16:00.672000 1168796 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:16:00.672000 1168796 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:16:29.580717810 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:16:29.850122830 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:16:29.284429479 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:16:29.314425351 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:16:30.717763207 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:16:30.737971973 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:16:30.743149834 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:16:30.747187068 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n8_seed5_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed5_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:30:03 acc1_pos: 3.646 acc5_pos: 100.000 loss: 1.9904 (1.9904) acc1: 3.6458 (3.6458) acc5: 100.0000 (100.0000) time: 5.4146 data: 3.4700 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/333] eta: 0:03:05 acc1_pos: 10.677 acc5_pos: 100.000 loss: 1.7566 (1.7847) acc1: 7.8125 (9.3040) acc5: 100.0000 (100.0000) time: 0.5738 data: 0.3158 max mem: 3531 +Test: [ 20/333] eta: 0:01:54 acc1_pos: 27.083 acc5_pos: 100.000 loss: 1.7566 (1.7701) acc1: 11.4583 (11.9668) acc5: 100.0000 (100.0000) time: 0.1126 data: 0.0205 max mem: 3531 +Test: [ 30/333] eta: 0:01:32 acc1_pos: 32.552 acc5_pos: 100.000 loss: 1.5342 (1.5982) acc1: 23.4375 (19.1616) acc5: 100.0000 (100.0000) time: 0.1551 data: 0.0603 max mem: 3531 +Test: [ 40/333] eta: 0:01:32 acc1_pos: 32.552 acc5_pos: 100.000 loss: 1.1958 (1.5085) acc1: 33.0729 (22.4276) acc5: 100.0000 (100.0000) time: 0.2669 data: 0.1732 max mem: 3531 +Test: [ 50/333] eta: 0:01:28 acc1_pos: 29.167 acc5_pos: 100.000 loss: 1.2033 (1.4584) acc1: 30.4688 (23.8103) acc5: 100.0000 (100.0000) time: 0.3297 data: 0.2386 max mem: 3531 +Test: [ 60/333] eta: 0:01:20 acc1_pos: 26.042 acc5_pos: 100.000 loss: 1.2362 (1.4188) acc1: 30.2083 (25.4440) acc5: 100.0000 (100.0000) time: 0.2504 data: 0.1595 max mem: 3531 +Test: [ 70/333] eta: 0:01:15 acc1_pos: 34.375 acc5_pos: 100.000 loss: 1.2354 (1.3862) acc1: 33.8542 (26.9953) acc5: 100.0000 (100.0000) time: 0.2223 data: 0.1304 max mem: 3531 +Test: [ 80/333] eta: 0:01:09 acc1_pos: 32.552 acc5_pos: 100.000 loss: 1.1944 (1.3679) acc1: 33.8542 (27.8389) acc5: 100.0000 (100.0000) time: 0.2078 data: 0.1159 max mem: 3531 +Test: [ 90/333] eta: 0:01:03 acc1_pos: 58.854 acc5_pos: 100.000 loss: 1.1254 (1.3140) acc1: 35.6771 (30.5174) acc5: 100.0000 (100.0000) time: 0.1650 data: 0.0722 max mem: 3531 +Test: [100/333] eta: 0:00:58 acc1_pos: 68.750 acc5_pos: 100.000 loss: 0.7340 (1.2615) acc1: 57.0312 (33.1683) acc5: 100.0000 (100.0000) time: 0.1562 data: 0.0617 max mem: 3531 +Test: [110/333] eta: 0:00:54 acc1_pos: 65.104 acc5_pos: 100.000 loss: 0.7315 (1.2179) acc1: 59.6354 (35.3721) acc5: 100.0000 (100.0000) time: 0.1699 data: 0.0750 max mem: 3531 +Test: [120/333] eta: 0:00:52 acc1_pos: 65.885 acc5_pos: 100.000 loss: 0.7315 (1.1803) acc1: 59.6354 (37.0523) acc5: 100.0000 (100.0000) time: 0.2144 data: 0.1210 max mem: 3531 +Test: [130/333] eta: 0:00:49 acc1_pos: 54.688 acc5_pos: 100.000 loss: 0.7001 (1.1464) acc1: 58.8542 (38.7027) acc5: 100.0000 (100.0000) time: 0.2367 data: 0.1449 max mem: 3531 +Test: [140/333] eta: 0:00:46 acc1_pos: 59.896 acc5_pos: 100.000 loss: 0.7151 (1.1174) acc1: 56.7708 (40.0506) acc5: 100.0000 (100.0000) time: 0.2240 data: 0.1333 max mem: 3531 +Test: [150/333] eta: 0:00:43 acc1_pos: 56.771 acc5_pos: 100.000 loss: 0.7185 (1.0932) acc1: 57.5521 (41.2683) acc5: 100.0000 (100.0000) time: 0.1931 data: 0.1009 max mem: 3531 +Test: [160/333] eta: 0:00:40 acc1_pos: 54.934 acc5_pos: 100.000 loss: 0.7400 (1.0704) acc1: 58.8542 (42.4884) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.1819 data: 0.0875 max mem: 3531 +Test: [170/333] eta: 0:00:37 acc1_pos: 54.934 acc5_pos: 100.000 loss: 0.6863 (1.0471) acc1: 62.5000 (43.7576) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.1949 data: 0.0991 max mem: 3531 +Test: [180/333] eta: 0:00:34 acc1_pos: 54.934 acc5_pos: 100.000 loss: 0.6382 (1.0225) acc1: 66.6667 (45.1125) acc5: 100.0000 (100.0000) acc1_neg: 70.052 acc5_neg: 100.000 time: 0.1825 data: 0.0865 max mem: 3531 +Test: [190/333] eta: 0:00:32 acc1_pos: 54.934 acc5_pos: 100.000 loss: 0.5651 (0.9946) acc1: 70.0521 (46.7741) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.1770 data: 0.0816 max mem: 3531 +Test: [200/333] eta: 0:00:29 acc1_pos: 54.934 acc5_pos: 100.000 loss: 0.5251 (0.9720) acc1: 73.6979 (48.0734) acc5: 100.0000 (100.0000) acc1_neg: 67.969 acc5_neg: 100.000 time: 0.1898 data: 0.0948 max mem: 3531 +Test: [210/333] eta: 0:00:27 acc1_pos: 54.934 acc5_pos: 100.000 loss: 0.5687 (0.9556) acc1: 70.3125 (48.9954) acc5: 100.0000 (100.0000) acc1_neg: 65.104 acc5_neg: 100.000 time: 0.1853 data: 0.0917 max mem: 3532 +Test: [220/333] eta: 0:00:24 acc1_pos: 54.934 acc5_pos: 100.000 loss: 0.5861 (0.9475) acc1: 65.1042 (49.3613) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.1828 data: 0.0911 max mem: 3532 +Test: [230/333] eta: 0:00:22 acc1_pos: 54.934 acc5_pos: 100.000 loss: 0.4672 (0.9246) acc1: 75.5208 (50.7553) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.1892 data: 0.0967 max mem: 3532 +Test: [240/333] eta: 0:00:20 acc1_pos: 54.934 acc5_pos: 100.000 loss: 0.4633 (0.9081) acc1: 77.0833 (51.7019) acc5: 100.0000 (100.0000) acc1_neg: 67.969 acc5_neg: 100.000 time: 0.1921 data: 0.0975 max mem: 3532 +Test: [250/333] eta: 0:00:17 acc1_pos: 54.934 acc5_pos: 100.000 loss: 0.5274 (0.8928) acc1: 74.7396 (52.5948) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.1838 data: 0.0911 max mem: 3532 +Test: [260/333] eta: 0:00:15 acc1_pos: 54.934 acc5_pos: 100.000 loss: 0.5380 (0.8798) acc1: 71.6146 (53.2966) acc5: 100.0000 (100.0000) acc1_neg: 69.531 acc5_neg: 100.000 time: 0.1820 data: 0.0896 max mem: 3532 +Test: [270/333] eta: 0:00:13 acc1_pos: 54.934 acc5_pos: 100.000 loss: 0.5573 (0.8681) acc1: 70.0521 (53.9908) acc5: 100.0000 (100.0000) acc1_neg: 65.625 acc5_neg: 100.000 time: 0.1993 data: 0.1066 max mem: 3532 +Test: [280/333] eta: 0:00:11 acc1_pos: 54.934 acc5_pos: 100.000 loss: 0.5654 (0.8648) acc1: 69.5312 (54.1046) acc5: 100.0000 (100.0000) acc1_neg: 53.906 acc5_neg: 100.000 time: 0.1941 data: 0.1026 max mem: 3532 +Test: [290/333] eta: 0:00:09 acc1_pos: 54.934 acc5_pos: 100.000 loss: 0.9169 (0.8658) acc1: 48.6979 (53.9814) acc5: 100.0000 (100.0000) acc1_neg: 63.021 acc5_neg: 100.000 time: 0.1878 data: 0.0963 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 54.934 acc5_pos: 100.000 loss: 0.7272 (0.8603) acc1: 57.0312 (54.2445) acc5: 100.0000 (100.0000) acc1_neg: 66.146 acc5_neg: 100.000 time: 0.1809 data: 0.0899 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 54.934 acc5_pos: 100.000 loss: 0.6223 (0.8494) acc1: 66.1458 (54.8767) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.1658 data: 0.0742 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 54.934 acc5_pos: 100.000 loss: 0.5073 (0.8403) acc1: 74.2188 (55.3925) acc5: 100.0000 (100.0000) acc1_neg: 57.812 acc5_neg: 100.000 time: 0.1604 data: 0.0691 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 54.934 acc5_pos: 100.000 loss: 0.7336 (0.8383) acc1: 59.1146 (55.4239) acc5: 100.0000 (100.0000) acc1_neg: 55.990 acc5_neg: 100.000 time: 0.1455 data: 0.0553 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 54.934 acc5_pos: 100.000 loss: 0.7481 (0.8378) acc1: 57.8125 (55.4367) acc5: 100.0000 (100.0000) acc1_neg: 59.718 acc5_neg: 100.000 time: 0.1511 data: 0.0574 max mem: 3532 +Test: Total time: 0:01:08 (0.2071 s / it) +* Overall Acc@1 55.437 Acc@5 100.000 + target == real Acc@1 42.036 Acc@5 100.000 + target == fake Acc@1 67.366 Acc@5 100.000 + loss 0.838 + PR-AUC 0.5685 ROC-AUC 0.5633 EER 45.19% (thr=0.6489) +Accuracy of the network on the 127843 test images: 55.4% +[rank0]:[W726 03:17:50.035525689 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:18:10.511000 1169354 torch/distributed/run.py:766] +W0726 03:18:10.511000 1169354 torch/distributed/run.py:766] ***************************************** +W0726 03:18:10.511000 1169354 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:18:10.511000 1169354 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 03:18:40.251564134 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:18:41.060828600 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:18:41.204881633 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:18:41.305531991 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:18:41.458016515 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:18:42.500315833 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +[rank2]:[W726 03:18:42.515619471 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:18:42.524481113 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n8_seed5_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed5_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:31:53 acc1_pos: 94.271 acc5_pos: 100.000 loss: 0.2113 (0.2113) acc1: 94.2708 (94.2708) acc5: 100.0000 (100.0000) time: 8.1418 data: 5.5012 max mem: 3531 +Test: [ 10/235] eta: 0:03:06 acc1_pos: 94.271 acc5_pos: 100.000 loss: 0.2141 (0.2151) acc1: 94.2708 (94.3182) acc5: 100.0000 (100.0000) time: 0.8303 data: 0.5005 max mem: 3531 +Test: [ 20/235] eta: 0:01:43 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.2098 (0.2102) acc1: 95.0521 (94.7545) acc5: 100.0000 (100.0000) time: 0.0995 data: 0.0004 max mem: 3531 +Test: [ 30/235] eta: 0:01:25 acc1_pos: 94.531 acc5_pos: 100.000 loss: 0.2093 (0.2103) acc1: 94.7917 (94.7497) acc5: 100.0000 (100.0000) time: 0.1899 data: 0.0911 max mem: 3531 +Test: [ 40/235] eta: 0:01:13 acc1_pos: 94.531 acc5_pos: 100.000 loss: 0.2085 (0.2092) acc1: 94.5312 (94.6900) acc5: 100.0000 (100.0000) time: 0.2653 data: 0.1692 max mem: 3531 +Test: [ 50/235] eta: 0:01:05 acc1_pos: 93.229 acc5_pos: 100.000 loss: 0.2054 (0.2100) acc1: 94.5312 (94.6793) acc5: 100.0000 (100.0000) time: 0.2499 data: 0.1561 max mem: 3531 +Test: [ 60/235] eta: 0:00:56 acc1_pos: 94.792 acc5_pos: 100.000 loss: 0.2145 (0.2113) acc1: 94.2708 (94.5996) acc5: 100.0000 (100.0000) time: 0.2195 data: 0.1234 max mem: 3531 +Test: [ 70/235] eta: 0:00:50 acc1_pos: 95.385 acc5_pos: 100.000 loss: 0.2297 (0.3755) acc1: 92.7083 (85.2259) acc5: 100.0000 (100.0000) acc1_neg: 15.365 acc5_neg: 100.000 time: 0.1966 data: 0.0989 max mem: 3531 +Test: [ 80/235] eta: 0:00:45 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.5260 (0.5204) acc1: 18.4896 (76.8808) acc5: 100.0000 (100.0000) acc1_neg: 17.969 acc5_neg: 100.000 time: 0.1999 data: 0.1047 max mem: 3531 +Test: [ 90/235] eta: 0:00:40 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.5518 (0.6314) acc1: 17.9688 (70.4785) acc5: 100.0000 (100.0000) acc1_neg: 15.885 acc5_neg: 100.000 time: 0.1834 data: 0.0877 max mem: 3531 +Test: [100/235] eta: 0:00:35 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.5281 (0.7204) acc1: 18.2292 (65.3826) acc5: 100.0000 (100.0000) acc1_neg: 15.104 acc5_neg: 100.000 time: 0.1499 data: 0.0528 max mem: 3531 +Test: [110/235] eta: 0:00:32 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.5106 (0.7927) acc1: 18.7500 (61.2331) acc5: 100.0000 (100.0000) acc1_neg: 17.969 acc5_neg: 100.000 time: 0.1813 data: 0.0867 max mem: 3531 +Test: [120/235] eta: 0:00:29 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.5681 (0.8720) acc1: 17.9688 (56.9990) acc5: 100.0000 (100.0000) acc1_neg: 9.635 acc5_neg: 100.000 time: 0.2125 data: 0.1204 max mem: 3531 +Test: [130/235] eta: 0:00:26 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.7669 (0.9408) acc1: 9.1146 (53.3417) acc5: 100.0000 (100.0000) acc1_neg: 9.115 acc5_neg: 100.000 time: 0.1883 data: 0.0972 max mem: 3531 +Test: [140/235] eta: 0:00:23 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.7936 (1.0022) acc1: 8.5938 (50.1699) acc5: 100.0000 (100.0000) acc1_neg: 5.990 acc5_neg: 100.000 time: 0.1786 data: 0.0863 max mem: 3531 +Test: [150/235] eta: 0:00:20 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.7833 (1.0533) acc1: 9.3750 (47.4941) acc5: 100.0000 (100.0000) acc1_neg: 10.938 acc5_neg: 100.000 time: 0.1844 data: 0.0917 max mem: 3531 +Test: [160/235] eta: 0:00:17 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.7670 (1.0982) acc1: 9.3750 (45.1152) acc5: 100.0000 (100.0000) acc1_neg: 11.458 acc5_neg: 100.000 time: 0.1833 data: 0.0924 max mem: 3531 +Test: [170/235] eta: 0:00:15 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.7344 (1.1032) acc1: 11.4583 (44.4292) acc5: 100.0000 (100.0000) acc1_neg: 34.896 acc5_neg: 100.000 time: 0.1724 data: 0.0818 max mem: 3531 +Test: [180/235] eta: 0:00:12 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.1216 (1.1039) acc1: 35.9375 (44.0306) acc5: 100.0000 (100.0000) acc1_neg: 34.635 acc5_neg: 100.000 time: 0.1696 data: 0.0790 max mem: 3531 +Test: [190/235] eta: 0:00:10 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.1011 (1.1044) acc1: 38.0208 (43.7146) acc5: 100.0000 (100.0000) acc1_neg: 35.938 acc5_neg: 100.000 time: 0.1688 data: 0.0776 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.1011 (1.1052) acc1: 36.9792 (43.3782) acc5: 100.0000 (100.0000) acc1_neg: 35.938 acc5_neg: 100.000 time: 0.1665 data: 0.0756 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.1107 (1.1054) acc1: 36.4583 (43.0811) acc5: 100.0000 (100.0000) acc1_neg: 36.979 acc5_neg: 100.000 time: 0.1502 data: 0.0599 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.0990 (1.1057) acc1: 37.5000 (42.8391) acc5: 100.0000 (100.0000) acc1_neg: 38.021 acc5_neg: 100.000 time: 0.1422 data: 0.0520 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.1069 (1.1060) acc1: 37.5000 (42.6351) acc5: 100.0000 (100.0000) acc1_neg: 36.719 acc5_neg: 100.000 time: 0.1260 data: 0.0364 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.1105 (1.1069) acc1: 37.5000 (42.5650) acc5: 100.0000 (100.0000) acc1_neg: 39.130 acc5_neg: 100.000 time: 0.1054 data: 0.0189 max mem: 3532 +Test: Total time: 0:00:49 (0.2105 s / it) +* Overall Acc@1 42.565 Acc@5 100.000 + target == real Acc@1 94.575 Acc@5 100.000 + target == fake Acc@1 23.754 Acc@5 100.000 + loss 1.107 + PR-AUC 0.8731 ROC-AUC 0.7249 EER 33.54% (thr=0.1477) +Accuracy of the network on the 89879 test images: 42.6% +[rank0]:[W726 03:19:39.806381082 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:19:58.764000 1169900 torch/distributed/run.py:766] +W0726 03:19:58.764000 1169900 torch/distributed/run.py:766] ***************************************** +W0726 03:19:58.764000 1169900 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:19:58.764000 1169900 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 03:20:26.313096948 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:20:27.067293969 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:20:27.085390438 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:20:27.144405694 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:20:27.197988369 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:20:27.408551930 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:20:27.474993942 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:20:28.500389134 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n8_seed5_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed5_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:22:30 acc1_pos: 20.052 acc5_pos: 100.000 loss: 1.3700 (1.3700) acc1: 20.0521 (20.0521) acc5: 100.0000 (100.0000) time: 5.0187 data: 2.6117 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/269] eta: 0:02:21 acc1_pos: 42.708 acc5_pos: 100.000 loss: 0.9568 (0.9887) acc1: 42.7083 (42.1875) acc5: 100.0000 (100.0000) time: 0.5462 data: 0.2460 max mem: 3531 +Test: [ 20/269] eta: 0:01:22 acc1_pos: 37.500 acc5_pos: 100.000 loss: 0.9593 (1.0010) acc1: 42.1875 (42.3363) acc5: 100.0000 (100.0000) time: 0.0970 data: 0.0067 max mem: 3531 +Test: [ 30/269] eta: 0:01:05 acc1_pos: 61.458 acc5_pos: 100.000 loss: 1.0154 (1.0156) acc1: 39.0625 (42.8427) acc5: 100.0000 (100.0000) time: 0.1275 data: 0.0319 max mem: 3531 +Test: [ 40/269] eta: 0:01:01 acc1_pos: 61.207 acc5_pos: 100.000 loss: 0.5398 (0.8556) acc1: 61.4583 (53.7030) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.2049 data: 0.1069 max mem: 3531 +Test: [ 50/269] eta: 0:00:58 acc1_pos: 61.207 acc5_pos: 100.000 loss: 0.3516 (0.7847) acc1: 88.2812 (58.2465) acc5: 100.0000 (100.0000) acc1_neg: 51.042 acc5_neg: 100.000 time: 0.2548 data: 0.1574 max mem: 3531 +Test: [ 60/269] eta: 0:00:53 acc1_pos: 61.207 acc5_pos: 100.000 loss: 0.5606 (0.7630) acc1: 68.4896 (59.4220) acc5: 100.0000 (100.0000) acc1_neg: 56.771 acc5_neg: 100.000 time: 0.2349 data: 0.1369 max mem: 3531 +Test: [ 70/269] eta: 0:00:49 acc1_pos: 61.207 acc5_pos: 100.000 loss: 0.6675 (0.7482) acc1: 60.9375 (59.8445) acc5: 100.0000 (100.0000) acc1_neg: 59.375 acc5_neg: 100.000 time: 0.1949 data: 0.0989 max mem: 3531 +Test: [ 80/269] eta: 0:00:45 acc1_pos: 61.207 acc5_pos: 100.000 loss: 0.6630 (0.7307) acc1: 63.5417 (60.6932) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1869 data: 0.0936 max mem: 3531 +Test: [ 90/269] eta: 0:00:41 acc1_pos: 61.207 acc5_pos: 100.000 loss: 0.5311 (0.7065) acc1: 71.0938 (62.1366) acc5: 100.0000 (100.0000) acc1_neg: 71.094 acc5_neg: 100.000 time: 0.1899 data: 0.0967 max mem: 3531 +Test: [100/269] eta: 0:00:39 acc1_pos: 61.207 acc5_pos: 100.000 loss: 0.5311 (0.6973) acc1: 71.0938 (62.5258) acc5: 100.0000 (100.0000) acc1_neg: 62.760 acc5_neg: 100.000 time: 0.1997 data: 0.1070 max mem: 3531 +Test: [110/269] eta: 0:00:36 acc1_pos: 61.207 acc5_pos: 100.000 loss: 0.5569 (0.6825) acc1: 69.0104 (63.4642) acc5: 100.0000 (100.0000) acc1_neg: 60.417 acc5_neg: 100.000 time: 0.1949 data: 0.1034 max mem: 3531 +Test: [120/269] eta: 0:00:33 acc1_pos: 61.207 acc5_pos: 100.000 loss: 0.5225 (0.6645) acc1: 73.6979 (64.5726) acc5: 100.0000 (100.0000) acc1_neg: 67.708 acc5_neg: 100.000 time: 0.1702 data: 0.0787 max mem: 3531 +Test: [130/269] eta: 0:00:30 acc1_pos: 61.207 acc5_pos: 100.000 loss: 0.5280 (0.6572) acc1: 70.8333 (65.0425) acc5: 100.0000 (100.0000) acc1_neg: 63.021 acc5_neg: 100.000 time: 0.1699 data: 0.0748 max mem: 3531 +Test: [140/269] eta: 0:00:27 acc1_pos: 61.207 acc5_pos: 100.000 loss: 0.5961 (0.6536) acc1: 64.5833 (64.9527) acc5: 100.0000 (100.0000) acc1_neg: 63.542 acc5_neg: 100.000 time: 0.1827 data: 0.0864 max mem: 3531 +Test: [150/269] eta: 0:00:25 acc1_pos: 61.207 acc5_pos: 100.000 loss: 0.5513 (0.6467) acc1: 65.3646 (65.2525) acc5: 100.0000 (100.0000) acc1_neg: 76.302 acc5_neg: 100.000 time: 0.2049 data: 0.1107 max mem: 3531 +Test: [160/269] eta: 0:00:23 acc1_pos: 61.207 acc5_pos: 100.000 loss: 0.4906 (0.6381) acc1: 71.6146 (65.7237) acc5: 100.0000 (100.0000) acc1_neg: 66.406 acc5_neg: 100.000 time: 0.1949 data: 0.0997 max mem: 3531 +Test: [170/269] eta: 0:00:20 acc1_pos: 61.207 acc5_pos: 100.000 loss: 0.5438 (0.6333) acc1: 72.9167 (66.0027) acc5: 100.0000 (100.0000) acc1_neg: 70.052 acc5_neg: 100.000 time: 0.1764 data: 0.0820 max mem: 3531 +Test: [180/269] eta: 0:00:18 acc1_pos: 61.207 acc5_pos: 100.000 loss: 0.4565 (0.6212) acc1: 73.4375 (66.7199) acc5: 100.0000 (100.0000) acc1_neg: 65.104 acc5_neg: 100.000 time: 0.1868 data: 0.0926 max mem: 3531 +Test: [190/269] eta: 0:00:16 acc1_pos: 61.207 acc5_pos: 100.000 loss: 0.4193 (0.6138) acc1: 77.3438 (67.2352) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.1784 data: 0.0832 max mem: 3531 +Test: [200/269] eta: 0:00:14 acc1_pos: 61.207 acc5_pos: 100.000 loss: 0.4586 (0.6057) acc1: 76.3021 (67.6734) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1661 data: 0.0683 max mem: 3531 +Test: [210/269] eta: 0:00:12 acc1_pos: 61.207 acc5_pos: 100.000 loss: 0.4546 (0.5974) acc1: 74.2188 (68.1329) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.1717 data: 0.0761 max mem: 3532 +Test: [220/269] eta: 0:00:09 acc1_pos: 61.207 acc5_pos: 100.000 loss: 0.4242 (0.5880) acc1: 76.8229 (68.7158) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1776 data: 0.0867 max mem: 3532 +Test: [230/269] eta: 0:00:07 acc1_pos: 61.207 acc5_pos: 100.000 loss: 0.3269 (0.5757) acc1: 86.1979 (69.5606) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.1781 data: 0.0837 max mem: 3532 +Test: [240/269] eta: 0:00:05 acc1_pos: 61.207 acc5_pos: 100.000 loss: 0.3898 (0.5739) acc1: 79.9479 (69.6717) acc5: 100.0000 (100.0000) acc1_neg: 68.490 acc5_neg: 100.000 time: 0.1894 data: 0.0931 max mem: 3532 +Test: [250/269] eta: 0:00:03 acc1_pos: 61.207 acc5_pos: 100.000 loss: 0.5723 (0.5772) acc1: 67.4479 (69.3206) acc5: 100.0000 (100.0000) acc1_neg: 56.250 acc5_neg: 100.000 time: 0.1935 data: 0.1005 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 61.207 acc5_pos: 100.000 loss: 0.7122 (0.5836) acc1: 56.2500 (68.7320) acc5: 100.0000 (100.0000) acc1_neg: 69.010 acc5_neg: 100.000 time: 0.1771 data: 0.0861 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 61.207 acc5_pos: 100.000 loss: 0.6943 (0.5837) acc1: 60.9375 (68.7810) acc5: 100.0000 (100.0000) acc1_neg: 87.139 acc5_neg: 100.000 time: 0.1638 data: 0.0696 max mem: 3532 +Test: Total time: 0:00:53 (0.2001 s / it) +* Overall Acc@1 68.781 Acc@5 100.000 + target == real Acc@1 42.857 Acc@5 100.000 + target == fake Acc@1 72.319 Acc@5 100.000 + loss 0.584 + PR-AUC 0.9150 ROC-AUC 0.6181 EER 41.74% (thr=0.6606) +Accuracy of the network on the 103293 test images: 68.8% +[rank0]:[W726 03:21:33.795891937 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T03:21:40Z] eval_done tag=fixed500k_real_n8_seed5_resampled +[2026-07-26T03:21:40Z] eval_start tag=fixed500k_fake_n16_seed5_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed5_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T03:21:40Z +hostname=n116-041-153 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed5_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n16_seed5_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 03:21:53.646000 1170478 torch/distributed/run.py:766] +W0726 03:21:53.646000 1170478 torch/distributed/run.py:766] ***************************************** +W0726 03:21:53.646000 1170478 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:21:53.646000 1170478 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:22:22.569281688 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:22:22.673695967 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:22:22.685343564 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:22:22.852238333 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:22:22.126296358 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:22:22.234540933 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:22:22.284338274 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:22:22.379840071 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n16_seed5_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed5_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:19:22 acc1_pos: 69.531 acc5_pos: 100.000 loss: 0.5548 (0.5548) acc1: 69.5312 (69.5312) acc5: 100.0000 (100.0000) time: 4.9685 data: 3.2719 max mem: 3531 +Test: [ 10/234] eta: 0:02:02 acc1_pos: 45.312 acc5_pos: 100.000 loss: 0.5548 (0.5394) acc1: 74.7396 (73.6742) acc5: 100.0000 (100.0000) time: 0.5479 data: 0.3096 max mem: 3531 +Test: [ 20/234] eta: 0:01:13 acc1_pos: 97.135 acc5_pos: 100.000 loss: 0.5811 (0.5682) acc1: 65.1042 (69.6925) acc5: 100.0000 (100.0000) time: 0.1143 data: 0.0225 max mem: 3531 +Test: [ 30/234] eta: 0:00:54 acc1_pos: 94.010 acc5_pos: 100.000 loss: 0.2196 (0.4529) acc1: 96.3542 (78.4190) acc5: 100.0000 (100.0000) time: 0.1102 data: 0.0160 max mem: 3531 +Test: [ 40/234] eta: 0:00:47 acc1_pos: 98.958 acc5_pos: 100.000 loss: 0.2184 (0.4014) acc1: 96.8750 (82.3425) acc5: 100.0000 (100.0000) time: 0.1388 data: 0.0410 max mem: 3531 +Test: [ 50/234] eta: 0:00:43 acc1_pos: 99.740 acc5_pos: 100.000 loss: 0.1144 (0.3446) acc1: 98.9583 (85.6822) acc5: 100.0000 (100.0000) time: 0.1915 data: 0.0974 max mem: 3531 +Test: [ 60/234] eta: 0:00:39 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.1176 (0.3103) acc1: 99.4792 (87.9440) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1849 data: 0.0910 max mem: 3531 +Test: [ 70/234] eta: 0:00:35 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.1385 (0.3796) acc1: 99.2188 (83.5717) acc5: 100.0000 (100.0000) acc1_neg: 32.552 acc5_neg: 100.000 time: 0.1684 data: 0.0728 max mem: 3531 +Test: [ 80/234] eta: 0:00:32 acc1_pos: 99.310 acc5_pos: 100.000 loss: 1.0796 (0.4317) acc1: 37.5000 (80.3337) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1750 data: 0.0832 max mem: 3531 +Test: [ 90/234] eta: 0:00:30 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.2818 (0.4074) acc1: 90.3646 (81.8252) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.1913 data: 0.1006 max mem: 3531 +Test: [100/234] eta: 0:00:27 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.2999 (0.4156) acc1: 89.3229 (81.1778) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1898 data: 0.0989 max mem: 3531 +Test: [110/234] eta: 0:00:25 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.5461 (0.4283) acc1: 68.7500 (80.2553) acc5: 100.0000 (100.0000) acc1_neg: 63.542 acc5_neg: 100.000 time: 0.1757 data: 0.0840 max mem: 3531 +Test: [120/234] eta: 0:00:23 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.5815 (0.4470) acc1: 67.7083 (79.0332) acc5: 100.0000 (100.0000) acc1_neg: 71.094 acc5_neg: 100.000 time: 0.2055 data: 0.1133 max mem: 3531 +Test: [130/234] eta: 0:00:21 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.5356 (0.4511) acc1: 72.6562 (78.9301) acc5: 100.0000 (100.0000) acc1_neg: 72.656 acc5_neg: 100.000 time: 0.2148 data: 0.1224 max mem: 3531 +Test: [140/234] eta: 0:00:19 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.5356 (0.4767) acc1: 73.4375 (77.6023) acc5: 100.0000 (100.0000) acc1_neg: 34.375 acc5_neg: 100.000 time: 0.1860 data: 0.0937 max mem: 3531 +Test: [150/234] eta: 0:00:16 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.5794 (0.4882) acc1: 67.1875 (76.7902) acc5: 100.0000 (100.0000) acc1_neg: 76.042 acc5_neg: 100.000 time: 0.1711 data: 0.0797 max mem: 3531 +Test: [160/234] eta: 0:00:14 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.5639 (0.4883) acc1: 70.3125 (76.8811) acc5: 100.0000 (100.0000) acc1_neg: 67.708 acc5_neg: 100.000 time: 0.1750 data: 0.0828 max mem: 3531 +Test: [170/234] eta: 0:00:12 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.5642 (0.4949) acc1: 70.3125 (76.3889) acc5: 100.0000 (100.0000) acc1_neg: 59.115 acc5_neg: 100.000 time: 0.1669 data: 0.0736 max mem: 3531 +Test: [180/234] eta: 0:00:10 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.6228 (0.5082) acc1: 66.9271 (75.5554) acc5: 100.0000 (100.0000) acc1_neg: 34.635 acc5_neg: 100.000 time: 0.1581 data: 0.0649 max mem: 3531 +Test: [190/234] eta: 0:00:08 acc1_pos: 99.310 acc5_pos: 100.000 loss: 1.0776 (0.5393) acc1: 45.3125 (73.9297) acc5: 100.0000 (100.0000) acc1_neg: 42.969 acc5_neg: 100.000 time: 0.1716 data: 0.0790 max mem: 3531 +Test: [200/234] eta: 0:00:06 acc1_pos: 99.310 acc5_pos: 100.000 loss: 1.0776 (0.5354) acc1: 45.5729 (74.2317) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1806 data: 0.0893 max mem: 3531 +Test: [210/234] eta: 0:00:04 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.1079 (0.5164) acc1: 99.7396 (75.4061) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1804 data: 0.0871 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.1808 (0.5033) acc1: 97.6562 (76.2785) acc5: 100.0000 (100.0000) acc1_neg: 71.354 acc5_neg: 100.000 time: 0.1776 data: 0.0845 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.6063 (0.5455) acc1: 17.1875 (73.6190) acc5: 100.0000 (100.0000) acc1_neg: 17.188 acc5_neg: 100.000 time: 0.1725 data: 0.0812 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 99.310 acc5_pos: 100.000 loss: 1.4511 (0.5573) acc1: 16.1458 (72.9563) acc5: 100.0000 (100.0000) acc1_neg: 15.175 acc5_neg: 100.000 time: 0.1649 data: 0.0710 max mem: 3532 +Test: Total time: 0:00:45 (0.1925 s / it) +* Overall Acc@1 72.956 Acc@5 100.000 + target == real Acc@1 85.782 Acc@5 100.000 + target == fake Acc@1 69.341 Acc@5 100.000 + loss 0.557 + PR-AUC 0.9566 ROC-AUC 0.8538 EER 23.69% (thr=0.3813) +Accuracy of the network on the 89729 test images: 73.0% +[rank0]:[W726 03:23:20.721041595 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:23:40.331000 1171024 torch/distributed/run.py:766] +W0726 03:23:40.331000 1171024 torch/distributed/run.py:766] ***************************************** +W0726 03:23:40.331000 1171024 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:23:40.331000 1171024 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:24:08.423379564 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:24:09.890531613 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:24:09.991974773 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:24:09.200041339 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:24:09.261761129 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:24:09.369867289 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:24:09.454973607 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:24:09.484460074 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n16_seed5_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed5_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:44:24 acc1_pos: 61.198 acc5_pos: 100.000 loss: 0.6364 (0.6364) acc1: 61.1979 (61.1979) acc5: 100.0000 (100.0000) time: 7.1489 data: 5.6064 max mem: 3531 +Test: [ 10/2387] eta: 0:29:00 acc1_pos: 36.458 acc5_pos: 100.000 loss: 0.9134 (0.9214) acc1: 40.6250 (38.0445) acc5: 100.0000 (100.0000) time: 0.7324 data: 0.5101 max mem: 3531 +Test: [ 20/2387] eta: 0:17:00 acc1_pos: 34.375 acc5_pos: 100.000 loss: 0.9230 (0.9404) acc1: 36.4583 (35.6895) acc5: 100.0000 (100.0000) time: 0.0953 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:14:15 acc1_pos: 15.104 acc5_pos: 100.000 loss: 0.9406 (0.9632) acc1: 30.7292 (33.0225) acc5: 100.0000 (100.0000) time: 0.1599 data: 0.0635 max mem: 3531 +Test: [ 40/2387] eta: 0:13:01 acc1_pos: 39.583 acc5_pos: 100.000 loss: 1.0316 (0.9741) acc1: 26.0417 (31.4405) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1359 max mem: 3531 +Test: [ 50/2387] eta: 0:12:20 acc1_pos: 39.844 acc5_pos: 100.000 loss: 0.9534 (0.9491) acc1: 33.3333 (33.9154) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1509 max mem: 3531 +Test: [ 60/2387] eta: 0:11:43 acc1_pos: 53.906 acc5_pos: 100.000 loss: 0.9414 (0.9485) acc1: 35.1562 (34.0207) acc5: 100.0000 (100.0000) time: 0.2400 data: 0.1464 max mem: 3531 +Test: [ 70/2387] eta: 0:11:07 acc1_pos: 35.417 acc5_pos: 100.000 loss: 0.9916 (0.9688) acc1: 31.7708 (33.0032) acc5: 100.0000 (100.0000) time: 0.2149 data: 0.1198 max mem: 3531 +Test: [ 80/2387] eta: 0:10:54 acc1_pos: 60.938 acc5_pos: 100.000 loss: 0.9916 (0.9587) acc1: 29.4271 (33.9088) acc5: 100.0000 (100.0000) time: 0.2262 data: 0.1327 max mem: 3531 +Test: [ 90/2387] eta: 0:10:42 acc1_pos: 45.573 acc5_pos: 100.000 loss: 0.8259 (0.9371) acc1: 45.5729 (35.6971) acc5: 100.0000 (100.0000) time: 0.2504 data: 0.1569 max mem: 3531 +Test: [ 100/2387] eta: 0:10:27 acc1_pos: 58.333 acc5_pos: 100.000 loss: 0.7443 (0.9269) acc1: 46.8750 (36.8090) acc5: 100.0000 (100.0000) time: 0.2358 data: 0.1419 max mem: 3531 +Test: [ 110/2387] eta: 0:10:18 acc1_pos: 57.552 acc5_pos: 100.000 loss: 0.7151 (0.9082) acc1: 53.6458 (38.1358) acc5: 100.0000 (100.0000) time: 0.2345 data: 0.1413 max mem: 3531 +Test: [ 120/2387] eta: 0:10:11 acc1_pos: 35.938 acc5_pos: 100.000 loss: 0.7652 (0.8990) acc1: 51.0417 (38.8903) acc5: 100.0000 (100.0000) time: 0.2477 data: 0.1547 max mem: 3531 +Test: [ 130/2387] eta: 0:10:07 acc1_pos: 52.865 acc5_pos: 100.000 loss: 0.7764 (0.8883) acc1: 51.0417 (39.8199) acc5: 100.0000 (100.0000) time: 0.2549 data: 0.1609 max mem: 3531 +Test: [ 140/2387] eta: 0:09:59 acc1_pos: 45.052 acc5_pos: 100.000 loss: 0.7764 (0.8823) acc1: 47.9167 (40.3055) acc5: 100.0000 (100.0000) time: 0.2498 data: 0.1523 max mem: 3531 +Test: [ 150/2387] eta: 0:09:52 acc1_pos: 26.562 acc5_pos: 100.000 loss: 0.7785 (0.8744) acc1: 47.9167 (40.8389) acc5: 100.0000 (100.0000) time: 0.2358 data: 0.1411 max mem: 3531 +Test: [ 160/2387] eta: 0:09:44 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.7338 (0.8633) acc1: 49.4792 (41.8365) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2300 data: 0.1350 max mem: 3531 +Test: [ 170/2387] eta: 0:09:34 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2880 (0.8286) acc1: 92.9688 (44.9089) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2191 data: 0.1203 max mem: 3531 +Test: [ 180/2387] eta: 0:09:26 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2692 (0.7978) acc1: 94.0104 (47.6260) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2099 data: 0.1144 max mem: 3531 +Test: [ 190/2387] eta: 0:09:16 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2632 (0.7699) acc1: 93.7500 (50.0682) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1999 data: 0.1057 max mem: 3531 +Test: [ 200/2387] eta: 0:09:07 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2629 (0.7449) acc1: 94.0104 (52.2673) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1919 data: 0.0978 max mem: 3531 +Test: [ 210/2387] eta: 0:08:59 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2665 (0.7226) acc1: 94.0104 (54.2383) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1973 data: 0.1048 max mem: 3532 +Test: [ 220/2387] eta: 0:08:52 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2709 (0.7023) acc1: 94.0104 (56.0285) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1997 data: 0.1067 max mem: 3532 +Test: [ 230/2387] eta: 0:08:46 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2713 (0.6835) acc1: 94.2708 (57.7043) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2064 data: 0.1118 max mem: 3532 +Test: [ 240/2387] eta: 0:08:39 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2748 (0.6672) acc1: 94.0104 (59.1902) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2030 data: 0.1096 max mem: 3532 +Test: [ 250/2387] eta: 0:08:32 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2483 (0.6461) acc1: 94.0104 (60.7705) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1868 data: 0.0966 max mem: 3532 +Test: [ 260/2387] eta: 0:08:25 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1151 (0.6254) acc1: 99.7396 (62.2705) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1857 data: 0.0935 max mem: 3532 +Test: [ 270/2387] eta: 0:08:19 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1013 (0.6063) acc1: 100.0000 (63.6579) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1905 data: 0.0959 max mem: 3532 +Test: [ 280/2387] eta: 0:08:12 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1089 (0.5887) acc1: 100.0000 (64.9475) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1802 data: 0.0877 max mem: 3532 +Test: [ 290/2387] eta: 0:08:07 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1127 (0.5725) acc1: 100.0000 (66.1521) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1843 data: 0.0921 max mem: 3532 +Test: [ 300/2387] eta: 0:08:03 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1111 (0.5570) acc1: 100.0000 (67.2766) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2014 data: 0.1085 max mem: 3532 +Test: [ 310/2387] eta: 0:07:58 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1051 (0.5485) acc1: 100.0000 (68.0006) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.2022 data: 0.1098 max mem: 3532 +Test: [ 320/2387] eta: 0:07:53 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.5080 (0.5476) acc1: 79.4271 (68.3095) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.1927 data: 0.1010 max mem: 3532 +Test: [ 330/2387] eta: 0:07:48 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.5150 (0.5465) acc1: 78.1250 (68.6194) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.1887 data: 0.0981 max mem: 3532 +Test: [ 340/2387] eta: 0:07:45 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.5118 (0.5455) acc1: 78.1250 (68.8943) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.1995 data: 0.1065 max mem: 3532 +Test: [ 350/2387] eta: 0:07:41 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.4999 (0.5365) acc1: 79.6875 (69.6997) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2000 data: 0.1075 max mem: 3532 +Test: [ 360/2387] eta: 0:07:35 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2128 (0.5277) acc1: 97.9167 (70.4683) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1830 data: 0.0900 max mem: 3532 +Test: [ 370/2387] eta: 0:07:32 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2222 (0.5195) acc1: 97.3958 (71.1934) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1910 data: 0.0966 max mem: 3532 +Test: [ 380/2387] eta: 0:07:29 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2277 (0.5119) acc1: 97.1354 (71.8757) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2031 data: 0.1107 max mem: 3532 +Test: [ 390/2387] eta: 0:07:26 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2257 (0.5045) acc1: 96.8750 (72.5191) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2067 data: 0.1145 max mem: 3532 +Test: [ 400/2387] eta: 0:07:21 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2252 (0.4976) acc1: 96.8750 (73.1355) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1935 data: 0.1019 max mem: 3532 +Test: [ 410/2387] eta: 0:07:18 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2229 (0.4908) acc1: 97.1354 (73.7138) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1849 data: 0.0934 max mem: 3532 +Test: [ 420/2387] eta: 0:07:14 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2222 (0.4844) acc1: 96.8750 (74.2670) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1953 data: 0.1028 max mem: 3532 +Test: [ 430/2387] eta: 0:07:11 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2256 (0.4789) acc1: 96.3542 (74.7692) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2010 data: 0.1080 max mem: 3532 +Test: [ 440/2387] eta: 0:07:08 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2372 (0.4720) acc1: 96.0938 (75.3012) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2018 data: 0.1101 max mem: 3533 +Test: [ 450/2387] eta: 0:07:05 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1163 (0.4641) acc1: 100.0000 (75.8488) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1989 data: 0.1085 max mem: 3533 +Test: [ 460/2387] eta: 0:07:03 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1152 (0.4565) acc1: 100.0000 (76.3727) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2126 data: 0.1205 max mem: 3533 +Test: [ 470/2387] eta: 0:06:58 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1152 (0.4492) acc1: 100.0000 (76.8743) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1797 data: 0.0823 max mem: 3533 +Test: [ 480/2387] eta: 0:06:53 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1160 (0.4423) acc1: 100.0000 (77.3551) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1551 data: 0.0586 max mem: 3533 +Test: [ 490/2387] eta: 0:06:49 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1159 (0.4357) acc1: 100.0000 (77.8163) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1698 data: 0.0777 max mem: 3533 +Test: [ 500/2387] eta: 0:06:46 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1158 (0.4293) acc1: 100.0000 (78.2586) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1785 data: 0.0871 max mem: 3533 +Test: [ 510/2387] eta: 0:06:42 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1163 (0.4235) acc1: 100.0000 (78.6841) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1803 data: 0.0893 max mem: 3533 +Test: [ 520/2387] eta: 0:06:39 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1739 (0.4188) acc1: 100.0000 (79.0922) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1720 data: 0.0808 max mem: 3533 +Test: [ 530/2387] eta: 0:06:36 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1715 (0.4141) acc1: 100.0000 (79.4859) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1802 data: 0.0897 max mem: 3533 +Test: [ 540/2387] eta: 0:06:33 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1689 (0.4096) acc1: 100.0000 (79.8646) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1880 data: 0.0963 max mem: 3533 +Test: [ 550/2387] eta: 0:06:30 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1697 (0.4052) acc1: 100.0000 (80.2291) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1833 data: 0.0917 max mem: 3533 +Test: [ 560/2387] eta: 0:06:26 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1698 (0.4010) acc1: 100.0000 (80.5811) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1796 data: 0.0878 max mem: 3533 +Test: [ 570/2387] eta: 0:06:24 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1687 (0.3970) acc1: 100.0000 (80.9203) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1863 data: 0.0937 max mem: 3533 +Test: [ 580/2387] eta: 0:06:21 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1687 (0.3931) acc1: 100.0000 (81.2487) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.1034 max mem: 3533 +Test: [ 590/2387] eta: 0:06:19 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1696 (0.3893) acc1: 100.0000 (81.5655) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2044 data: 0.1129 max mem: 3533 +Test: [ 600/2387] eta: 0:06:17 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1746 (0.3859) acc1: 100.0000 (81.8705) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2116 data: 0.1183 max mem: 3533 +Test: [ 610/2387] eta: 0:06:15 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2008 (0.3836) acc1: 97.6562 (82.1127) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2207 data: 0.1280 max mem: 3533 +Test: [ 620/2387] eta: 0:06:12 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2479 (0.3815) acc1: 96.6146 (82.3504) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1986 data: 0.1079 max mem: 3533 +Test: [ 630/2387] eta: 0:06:10 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2494 (0.3795) acc1: 96.8750 (82.5777) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1844 data: 0.0937 max mem: 3533 +Test: [ 640/2387] eta: 0:06:08 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2519 (0.3775) acc1: 96.8750 (82.7950) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2113 data: 0.1186 max mem: 3533 +Test: [ 650/2387] eta: 0:06:05 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2527 (0.3756) acc1: 97.1354 (83.0101) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2134 data: 0.1206 max mem: 3533 +Test: [ 660/2387] eta: 0:06:03 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2499 (0.3737) acc1: 96.8750 (83.2207) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2044 data: 0.1134 max mem: 3533 +Test: [ 670/2387] eta: 0:06:01 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2499 (0.3719) acc1: 96.8750 (83.4253) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1969 data: 0.1062 max mem: 3534 +Test: [ 680/2387] eta: 0:05:58 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2470 (0.3697) acc1: 97.1354 (83.6534) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1856 data: 0.0934 max mem: 3534 +Test: [ 690/2387] eta: 0:05:55 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2470 (0.3699) acc1: 96.0938 (83.6582) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.1851 data: 0.0907 max mem: 3534 +Test: [ 700/2387] eta: 0:05:52 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.4385 (0.3710) acc1: 78.3854 (83.5726) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.1877 data: 0.0919 max mem: 3534 +Test: [ 710/2387] eta: 0:05:50 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.4453 (0.3720) acc1: 78.3854 (83.5051) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.1977 data: 0.1003 max mem: 3534 +Test: [ 720/2387] eta: 0:05:48 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.4423 (0.3729) acc1: 78.3854 (83.4247) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.1985 data: 0.1025 max mem: 3534 +Test: [ 730/2387] eta: 0:05:45 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.4486 (0.3740) acc1: 77.3438 (83.3412) acc5: 100.0000 (100.0000) acc1_neg: 77.344 acc5_neg: 100.000 time: 0.1905 data: 0.0955 max mem: 3534 +Test: [ 740/2387] eta: 0:05:43 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.4480 (0.3749) acc1: 77.8646 (83.2785) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.1867 data: 0.0938 max mem: 3534 +Test: [ 750/2387] eta: 0:05:40 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.4286 (0.3757) acc1: 78.3854 (83.2147) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.1919 data: 0.1015 max mem: 3534 +Test: [ 760/2387] eta: 0:05:40 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.4272 (0.3760) acc1: 78.3854 (83.1927) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.2368 data: 0.1468 max mem: 3534 +Test: [ 770/2387] eta: 0:05:37 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.3173 (0.3745) acc1: 86.9792 (83.3033) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2322 data: 0.1409 max mem: 3534 +Test: [ 780/2387] eta: 0:05:35 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2026 (0.3720) acc1: 96.0938 (83.4817) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1964 data: 0.1014 max mem: 3534 +Test: [ 790/2387] eta: 0:05:33 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1756 (0.3701) acc1: 97.3958 (83.6023) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.2159 data: 0.1193 max mem: 3534 +Test: [ 800/2387] eta: 0:05:31 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1708 (0.3675) acc1: 97.9167 (83.7875) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2161 data: 0.1230 max mem: 3534 +Test: [ 810/2387] eta: 0:05:28 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1725 (0.3652) acc1: 98.1771 (83.9637) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1894 data: 0.0966 max mem: 3534 +Test: [ 820/2387] eta: 0:05:26 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1848 (0.3634) acc1: 97.3958 (84.1089) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1749 data: 0.0819 max mem: 3534 +Test: [ 830/2387] eta: 0:05:23 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2441 (0.3623) acc1: 94.2708 (84.2055) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1752 data: 0.0835 max mem: 3534 +Test: [ 840/2387] eta: 0:05:20 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.3223 (0.3621) acc1: 91.1458 (84.2672) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1753 data: 0.0818 max mem: 3534 +Test: [ 850/2387] eta: 0:05:17 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.3365 (0.3619) acc1: 90.6250 (84.3325) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1691 data: 0.0766 max mem: 3534 +Test: [ 860/2387] eta: 0:05:15 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.3312 (0.3614) acc1: 91.6667 (84.4261) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1701 data: 0.0776 max mem: 3534 +Test: [ 870/2387] eta: 0:05:12 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.3102 (0.3608) acc1: 92.1875 (84.5218) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1668 data: 0.0749 max mem: 3534 +Test: [ 880/2387] eta: 0:05:10 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.3236 (0.3608) acc1: 91.1458 (84.5775) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1950 data: 0.1045 max mem: 3534 +Test: [ 890/2387] eta: 0:05:08 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.3221 (0.3595) acc1: 91.6667 (84.7094) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2206 data: 0.1299 max mem: 3534 +Test: [ 900/2387] eta: 0:05:06 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2801 (0.3630) acc1: 88.2812 (84.4791) acc5: 100.0000 (100.0000) acc1_neg: 60.677 acc5_neg: 100.000 time: 0.2047 data: 0.1147 max mem: 3535 +Test: [ 910/2387] eta: 0:05:04 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.7718 (0.3675) acc1: 55.7292 (84.1549) acc5: 100.0000 (100.0000) acc1_neg: 56.250 acc5_neg: 100.000 time: 0.1949 data: 0.1047 max mem: 3535 +Test: [ 920/2387] eta: 0:05:02 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.7811 (0.3721) acc1: 54.4271 (83.8304) acc5: 100.0000 (100.0000) acc1_neg: 52.344 acc5_neg: 100.000 time: 0.1953 data: 0.1046 max mem: 3535 +Test: [ 930/2387] eta: 0:04:59 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.7893 (0.3765) acc1: 55.4688 (83.5317) acc5: 100.0000 (100.0000) acc1_neg: 52.344 acc5_neg: 100.000 time: 0.1847 data: 0.0928 max mem: 3535 +Test: [ 940/2387] eta: 0:04:57 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.7905 (0.3808) acc1: 55.2083 (83.2273) acc5: 100.0000 (100.0000) acc1_neg: 55.729 acc5_neg: 100.000 time: 0.1811 data: 0.0884 max mem: 3535 +Test: [ 950/2387] eta: 0:04:55 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.7829 (0.3850) acc1: 54.9479 (82.9264) acc5: 100.0000 (100.0000) acc1_neg: 57.552 acc5_neg: 100.000 time: 0.1925 data: 0.0977 max mem: 3535 +Test: [ 960/2387] eta: 0:04:52 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.7705 (0.3890) acc1: 54.6875 (82.6372) acc5: 100.0000 (100.0000) acc1_neg: 57.292 acc5_neg: 100.000 time: 0.2000 data: 0.1057 max mem: 3535 +Test: [ 970/2387] eta: 0:04:50 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.7752 (0.3931) acc1: 54.4271 (82.3469) acc5: 100.0000 (100.0000) acc1_neg: 58.073 acc5_neg: 100.000 time: 0.2016 data: 0.1100 max mem: 3535 +Test: [ 980/2387] eta: 0:04:48 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.7624 (0.3933) acc1: 56.5104 (82.3339) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.1983 data: 0.1068 max mem: 3535 +Test: [ 990/2387] eta: 0:04:46 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.3189 (0.3927) acc1: 89.3229 (82.3834) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.2060 data: 0.1146 max mem: 3535 +Test: [1000/2387] eta: 0:04:44 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.3452 (0.3929) acc1: 83.3333 (82.3676) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.2143 data: 0.1215 max mem: 3535 +Test: [1010/2387] eta: 0:04:42 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.4002 (0.3932) acc1: 80.2083 (82.3486) acc5: 100.0000 (100.0000) acc1_neg: 75.781 acc5_neg: 100.000 time: 0.2070 data: 0.1138 max mem: 3535 +Test: [1020/2387] eta: 0:04:40 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.3736 (0.3924) acc1: 82.0312 (82.4077) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1972 data: 0.1062 max mem: 3535 +Test: [1030/2387] eta: 0:04:38 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1490 (0.3900) acc1: 98.6979 (82.5652) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1966 data: 0.1067 max mem: 3535 +Test: [1040/2387] eta: 0:04:36 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1340 (0.3876) acc1: 99.7396 (82.7264) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2280 data: 0.1344 max mem: 3535 +Test: [1050/2387] eta: 0:04:34 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1340 (0.3853) acc1: 99.7396 (82.8826) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2110 data: 0.1161 max mem: 3535 +Test: [1060/2387] eta: 0:04:32 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1451 (0.3831) acc1: 99.4792 (83.0346) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1845 data: 0.0926 max mem: 3535 +Test: [1070/2387] eta: 0:04:30 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1521 (0.3812) acc1: 98.9583 (83.1772) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2145 data: 0.1222 max mem: 3535 +Test: [1080/2387] eta: 0:04:28 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1988 (0.3796) acc1: 98.4375 (83.2982) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.2228 data: 0.1291 max mem: 3535 +Test: [1090/2387] eta: 0:04:26 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2248 (0.3786) acc1: 93.7500 (83.3758) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.2193 data: 0.1249 max mem: 3535 +Test: [1100/2387] eta: 0:04:24 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2516 (0.3776) acc1: 92.9688 (83.4488) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2137 data: 0.1206 max mem: 3535 +Test: [1110/2387] eta: 0:04:22 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2516 (0.3769) acc1: 91.6667 (83.4892) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2110 data: 0.1179 max mem: 3535 +Test: [1120/2387] eta: 0:04:20 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2532 (0.3758) acc1: 92.4479 (83.5749) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.2093 data: 0.1163 max mem: 3536 +Test: [1130/2387] eta: 0:04:18 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2636 (0.3753) acc1: 92.1875 (83.6193) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.2086 data: 0.1176 max mem: 3536 +Test: [1140/2387] eta: 0:04:16 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.3095 (0.3749) acc1: 88.8021 (83.6695) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2093 data: 0.1182 max mem: 3536 +Test: [1150/2387] eta: 0:04:14 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.3164 (0.3743) acc1: 90.6250 (83.7431) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2094 data: 0.1176 max mem: 3536 +Test: [1160/2387] eta: 0:04:12 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2625 (0.3731) acc1: 94.2708 (83.8344) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.2044 data: 0.1126 max mem: 3536 +Test: [1170/2387] eta: 0:04:09 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2293 (0.3719) acc1: 96.0938 (83.9425) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1756 data: 0.0847 max mem: 3536 +Test: [1180/2387] eta: 0:04:07 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2319 (0.3708) acc1: 96.3542 (84.0392) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1626 data: 0.0702 max mem: 3536 +Test: [1190/2387] eta: 0:04:05 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2404 (0.3700) acc1: 94.7917 (84.1176) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1714 data: 0.0782 max mem: 3536 +Test: [1200/2387] eta: 0:04:02 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2357 (0.3686) acc1: 97.1354 (84.2395) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1686 data: 0.0770 max mem: 3536 +Test: [1210/2387] eta: 0:04:00 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1918 (0.3671) acc1: 99.4792 (84.3615) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1621 data: 0.0699 max mem: 3536 +Test: [1220/2387] eta: 0:03:57 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1387 (0.3651) acc1: 99.4792 (84.4848) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1621 data: 0.0696 max mem: 3536 +Test: [1230/2387] eta: 0:03:55 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1262 (0.3631) acc1: 99.4792 (84.6079) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1711 data: 0.0797 max mem: 3536 +Test: [1240/2387] eta: 0:03:53 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1153 (0.3611) acc1: 99.7396 (84.7282) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1732 data: 0.0800 max mem: 3536 +Test: [1250/2387] eta: 0:03:50 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1178 (0.3592) acc1: 99.7396 (84.8457) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1681 data: 0.0721 max mem: 3536 +Test: [1260/2387] eta: 0:03:48 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1246 (0.3574) acc1: 99.7396 (84.9607) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1686 data: 0.0752 max mem: 3536 +Test: [1270/2387] eta: 0:03:46 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1095 (0.3554) acc1: 100.0000 (85.0786) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0785 max mem: 3536 +Test: [1280/2387] eta: 0:03:44 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1030 (0.3536) acc1: 100.0000 (85.1916) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1952 data: 0.1014 max mem: 3536 +Test: [1290/2387] eta: 0:03:41 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1221 (0.3519) acc1: 99.2188 (85.2827) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1976 data: 0.1034 max mem: 3536 +Test: [1300/2387] eta: 0:03:39 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1270 (0.3502) acc1: 98.9583 (85.3818) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1864 data: 0.0949 max mem: 3536 +Test: [1310/2387] eta: 0:03:37 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1270 (0.3486) acc1: 98.9583 (85.4794) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1991 data: 0.1087 max mem: 3536 +Test: [1320/2387] eta: 0:03:35 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1334 (0.3471) acc1: 99.4792 (85.5742) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1825 data: 0.0906 max mem: 3536 +Test: [1330/2387] eta: 0:03:33 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1441 (0.3457) acc1: 99.2188 (85.6681) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1980 data: 0.1064 max mem: 3536 +Test: [1340/2387] eta: 0:03:31 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1997 (0.3460) acc1: 95.3125 (85.6454) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2049 data: 0.1143 max mem: 3536 +Test: [1350/2387] eta: 0:03:29 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.0901 (0.3440) acc1: 100.0000 (85.7517) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1662 data: 0.0748 max mem: 3537 +Test: [1360/2387] eta: 0:03:27 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.0747 (0.3420) acc1: 100.0000 (85.8564) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1668 data: 0.0756 max mem: 3537 +Test: [1370/2387] eta: 0:03:24 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.0685 (0.3400) acc1: 100.0000 (85.9595) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0847 max mem: 3537 +Test: [1380/2387] eta: 0:03:22 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.0707 (0.3381) acc1: 100.0000 (86.0612) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1590 data: 0.0662 max mem: 3537 +Test: [1390/2387] eta: 0:03:20 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.0810 (0.3363) acc1: 100.0000 (86.1614) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1753 data: 0.0821 max mem: 3537 +Test: [1400/2387] eta: 0:03:18 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.0894 (0.3345) acc1: 100.0000 (86.2602) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2071 data: 0.1164 max mem: 3537 +Test: [1410/2387] eta: 0:03:16 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.0912 (0.3334) acc1: 100.0000 (86.3314) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1945 data: 0.1040 max mem: 3537 +Test: [1420/2387] eta: 0:03:14 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1843 (0.3327) acc1: 96.3542 (86.3834) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1827 data: 0.0912 max mem: 3537 +Test: [1430/2387] eta: 0:03:12 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1969 (0.3319) acc1: 95.8333 (86.4441) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1886 data: 0.0965 max mem: 3537 +Test: [1440/2387] eta: 0:03:09 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2079 (0.3312) acc1: 95.0521 (86.5107) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1795 data: 0.0888 max mem: 3537 +Test: [1450/2387] eta: 0:03:07 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2159 (0.3304) acc1: 96.3542 (86.5761) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1820 data: 0.0908 max mem: 3537 +Test: [1460/2387] eta: 0:03:05 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1835 (0.3294) acc1: 97.1354 (86.6483) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1909 data: 0.1000 max mem: 3537 +Test: [1470/2387] eta: 0:03:03 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1626 (0.3282) acc1: 98.1771 (86.7290) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1929 data: 0.1011 max mem: 3537 +Test: [1480/2387] eta: 0:03:01 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1501 (0.3271) acc1: 98.9583 (86.8053) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1843 data: 0.0916 max mem: 3537 +Test: [1490/2387] eta: 0:02:59 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1514 (0.3260) acc1: 99.2188 (86.8889) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1954 data: 0.1039 max mem: 3537 +Test: [1500/2387] eta: 0:02:57 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1638 (0.3250) acc1: 98.9583 (86.9662) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1909 data: 0.0981 max mem: 3537 +Test: [1510/2387] eta: 0:02:55 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1660 (0.3239) acc1: 99.4792 (87.0519) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1652 data: 0.0725 max mem: 3537 +Test: [1520/2387] eta: 0:02:53 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1869 (0.3236) acc1: 99.4792 (87.0905) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1728 data: 0.0815 max mem: 3537 +Test: [1530/2387] eta: 0:02:50 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2753 (0.3234) acc1: 92.7083 (87.1290) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1714 data: 0.0790 max mem: 3537 +Test: [1540/2387] eta: 0:02:48 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2621 (0.3231) acc1: 95.8333 (87.1815) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1699 data: 0.0772 max mem: 3537 +Test: [1550/2387] eta: 0:02:46 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2543 (0.3227) acc1: 96.6146 (87.2362) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1684 data: 0.0764 max mem: 3537 +Test: [1560/2387] eta: 0:02:44 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2593 (0.3225) acc1: 95.3125 (87.2746) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.1723 data: 0.0808 max mem: 3537 +Test: [1570/2387] eta: 0:02:42 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2593 (0.3218) acc1: 95.8333 (87.3435) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1811 data: 0.0900 max mem: 3537 +Test: [1580/2387] eta: 0:02:40 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1773 (0.3209) acc1: 98.4375 (87.4158) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1682 data: 0.0778 max mem: 3538 +Test: [1590/2387] eta: 0:02:38 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1630 (0.3198) acc1: 99.2188 (87.4876) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1835 data: 0.0900 max mem: 3538 +Test: [1600/2387] eta: 0:02:36 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1529 (0.3191) acc1: 98.6979 (87.5226) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2104 data: 0.1150 max mem: 3538 +Test: [1610/2387] eta: 0:02:34 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1529 (0.3181) acc1: 98.6979 (87.5933) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2056 data: 0.1125 max mem: 3538 +Test: [1620/2387] eta: 0:02:32 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1592 (0.3173) acc1: 98.6979 (87.6509) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2002 data: 0.1087 max mem: 3538 +Test: [1630/2387] eta: 0:02:30 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1804 (0.3166) acc1: 96.6146 (87.7002) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2055 data: 0.1153 max mem: 3538 +Test: [1640/2387] eta: 0:02:28 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1783 (0.3158) acc1: 97.3958 (87.7617) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2039 data: 0.1138 max mem: 3538 +Test: [1650/2387] eta: 0:02:26 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1824 (0.3150) acc1: 98.4375 (87.8251) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1989 data: 0.1077 max mem: 3538 +Test: [1660/2387] eta: 0:02:24 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1866 (0.3143) acc1: 98.4375 (87.8815) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2182 data: 0.1257 max mem: 3538 +Test: [1670/2387] eta: 0:02:22 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1918 (0.3135) acc1: 97.9167 (87.9409) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1911 data: 0.0988 max mem: 3538 +Test: [1680/2387] eta: 0:02:20 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1853 (0.3129) acc1: 97.9167 (87.9931) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1729 data: 0.0805 max mem: 3538 +Test: [1690/2387] eta: 0:02:18 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1785 (0.3122) acc1: 98.1771 (88.0462) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1954 data: 0.1039 max mem: 3538 +Test: [1700/2387] eta: 0:02:16 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1854 (0.3114) acc1: 97.9167 (88.1049) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1735 data: 0.0832 max mem: 3538 +Test: [1710/2387] eta: 0:02:14 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1787 (0.3106) acc1: 97.9167 (88.1601) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1796 data: 0.0886 max mem: 3538 +Test: [1720/2387] eta: 0:02:12 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1544 (0.3097) acc1: 98.9583 (88.2212) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2007 data: 0.1096 max mem: 3538 +Test: [1730/2387] eta: 0:02:10 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1544 (0.3089) acc1: 98.9583 (88.2746) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1985 data: 0.1076 max mem: 3538 +Test: [1740/2387] eta: 0:02:08 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1486 (0.3080) acc1: 98.6979 (88.3332) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1986 data: 0.1067 max mem: 3538 +Test: [1750/2387] eta: 0:02:06 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1455 (0.3072) acc1: 98.6979 (88.3919) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1890 data: 0.0976 max mem: 3538 +Test: [1760/2387] eta: 0:02:04 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1432 (0.3062) acc1: 99.4792 (88.4560) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1799 data: 0.0892 max mem: 3538 +Test: [1770/2387] eta: 0:02:02 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1472 (0.3055) acc1: 98.9583 (88.5034) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1747 data: 0.0822 max mem: 3538 +Test: [1780/2387] eta: 0:02:00 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1702 (0.3047) acc1: 97.9167 (88.5578) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1773 data: 0.0851 max mem: 3538 +Test: [1790/2387] eta: 0:01:58 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1679 (0.3039) acc1: 98.4375 (88.6100) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1771 data: 0.0864 max mem: 3538 +Test: [1800/2387] eta: 0:01:56 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1555 (0.3031) acc1: 98.4375 (88.6659) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1686 data: 0.0764 max mem: 3538 +Test: [1810/2387] eta: 0:01:54 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1539 (0.3023) acc1: 98.4375 (88.7207) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1839 data: 0.0917 max mem: 3539 +Test: [1820/2387] eta: 0:01:52 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1503 (0.3014) acc1: 99.2188 (88.7795) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1987 data: 0.1077 max mem: 3539 +Test: [1830/2387] eta: 0:01:50 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1537 (0.3007) acc1: 99.7396 (88.8393) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1954 data: 0.1033 max mem: 3539 +Test: [1840/2387] eta: 0:01:48 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1764 (0.3000) acc1: 99.7396 (88.8988) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2015 data: 0.1097 max mem: 3539 +Test: [1850/2387] eta: 0:01:46 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1808 (0.2994) acc1: 99.7396 (88.9578) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2021 data: 0.1086 max mem: 3539 +Test: [1860/2387] eta: 0:01:44 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1798 (0.2987) acc1: 99.7396 (89.0155) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1872 data: 0.0934 max mem: 3539 +Test: [1870/2387] eta: 0:01:42 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1798 (0.2981) acc1: 99.7396 (89.0728) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1719 data: 0.0811 max mem: 3539 +Test: [1880/2387] eta: 0:01:40 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1802 (0.2975) acc1: 99.7396 (89.1298) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1711 data: 0.0793 max mem: 3539 +Test: [1890/2387] eta: 0:01:38 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1815 (0.2969) acc1: 99.7396 (89.1856) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1802 data: 0.0875 max mem: 3539 +Test: [1900/2387] eta: 0:01:35 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1829 (0.2963) acc1: 99.7396 (89.2399) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1774 data: 0.0863 max mem: 3539 +Test: [1910/2387] eta: 0:01:33 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2395 (0.2961) acc1: 98.1771 (89.2812) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1754 data: 0.0834 max mem: 3539 +Test: [1920/2387] eta: 0:01:32 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2489 (0.2958) acc1: 97.3958 (89.3015) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1935 data: 0.1015 max mem: 3539 +Test: [1930/2387] eta: 0:01:29 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1693 (0.2951) acc1: 98.4375 (89.3473) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1781 data: 0.0868 max mem: 3539 +Test: [1940/2387] eta: 0:01:27 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1674 (0.2949) acc1: 98.4375 (89.3503) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1588 data: 0.0666 max mem: 3539 +Test: [1950/2387] eta: 0:01:25 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1826 (0.2944) acc1: 98.1771 (89.3859) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1849 data: 0.0912 max mem: 3539 +Test: [1960/2387] eta: 0:01:24 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1947 (0.2941) acc1: 96.6146 (89.4147) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2048 data: 0.1123 max mem: 3539 +Test: [1970/2387] eta: 0:01:22 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2403 (0.2942) acc1: 93.2292 (89.3997) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.2019 data: 0.1056 max mem: 3539 +Test: [1980/2387] eta: 0:01:20 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2295 (0.2939) acc1: 92.7083 (89.4302) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2105 data: 0.1134 max mem: 3539 +Test: [1990/2387] eta: 0:01:18 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1876 (0.2934) acc1: 97.3958 (89.4714) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2179 data: 0.1248 max mem: 3539 +Test: [2000/2387] eta: 0:01:16 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1870 (0.2928) acc1: 97.9167 (89.5133) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2217 data: 0.1287 max mem: 3539 +Test: [2010/2387] eta: 0:01:14 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1890 (0.2923) acc1: 97.9167 (89.5526) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2003 data: 0.1095 max mem: 3539 +Test: [2020/2387] eta: 0:01:12 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1916 (0.2918) acc1: 97.6562 (89.5929) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1838 data: 0.0935 max mem: 3539 +Test: [2030/2387] eta: 0:01:10 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1884 (0.2913) acc1: 97.6562 (89.6315) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2095 data: 0.1166 max mem: 3540 +Test: [2040/2387] eta: 0:01:08 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1862 (0.2908) acc1: 97.6562 (89.6719) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2136 data: 0.1185 max mem: 3540 +Test: [2050/2387] eta: 0:01:06 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1906 (0.2903) acc1: 97.3958 (89.7092) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2017 data: 0.1084 max mem: 3540 +Test: [2060/2387] eta: 0:01:04 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1973 (0.2899) acc1: 97.1354 (89.7454) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2008 data: 0.1080 max mem: 3540 +Test: [2070/2387] eta: 0:01:02 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2003 (0.2895) acc1: 97.6562 (89.7831) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2058 data: 0.1131 max mem: 3540 +Test: [2080/2387] eta: 0:01:00 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1966 (0.2891) acc1: 97.6562 (89.8192) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2054 data: 0.1118 max mem: 3540 +Test: [2090/2387] eta: 0:00:58 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1962 (0.2886) acc1: 97.3958 (89.8555) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2012 data: 0.1066 max mem: 3540 +Test: [2100/2387] eta: 0:00:56 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1969 (0.2882) acc1: 97.3958 (89.8932) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1916 data: 0.0993 max mem: 3540 +Test: [2110/2387] eta: 0:00:54 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1969 (0.2878) acc1: 97.6562 (89.9301) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1966 data: 0.1048 max mem: 3540 +Test: [2120/2387] eta: 0:00:52 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1981 (0.2875) acc1: 97.1354 (89.9545) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.1927 data: 0.1007 max mem: 3540 +Test: [2130/2387] eta: 0:00:50 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1989 (0.2870) acc1: 97.1354 (89.9941) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1870 data: 0.0949 max mem: 3540 +Test: [2140/2387] eta: 0:00:48 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1946 (0.2869) acc1: 97.3958 (90.0003) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1929 data: 0.1010 max mem: 3540 +Test: [2150/2387] eta: 0:00:46 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2066 (0.2865) acc1: 95.3125 (90.0289) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1725 data: 0.0806 max mem: 3540 +Test: [2160/2387] eta: 0:00:44 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1898 (0.2861) acc1: 97.3958 (90.0605) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1594 data: 0.0662 max mem: 3540 +Test: [2170/2387] eta: 0:00:42 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2306 (0.2860) acc1: 95.8333 (90.0749) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1637 data: 0.0707 max mem: 3540 +Test: [2180/2387] eta: 0:00:40 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2886 (0.2868) acc1: 92.1875 (90.0270) acc5: 100.0000 (100.0000) acc1_neg: 61.979 acc5_neg: 100.000 time: 0.1544 data: 0.0629 max mem: 3540 +Test: [2190/2387] eta: 0:00:38 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.6692 (0.2890) acc1: 60.9375 (89.8767) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1513 data: 0.0608 max mem: 3540 +Test: [2200/2387] eta: 0:00:36 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2018 (0.2884) acc1: 93.4896 (89.9118) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1595 data: 0.0691 max mem: 3540 +Test: [2210/2387] eta: 0:00:34 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1515 (0.2878) acc1: 98.6979 (89.9527) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1438 data: 0.0537 max mem: 3540 +Test: [2220/2387] eta: 0:00:32 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1443 (0.2872) acc1: 98.9583 (89.9898) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1293 data: 0.0391 max mem: 3540 +Test: [2230/2387] eta: 0:00:30 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1443 (0.2866) acc1: 98.6979 (90.0300) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1330 data: 0.0424 max mem: 3540 +Test: [2240/2387] eta: 0:00:28 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1324 (0.2860) acc1: 98.6979 (90.0662) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1303 data: 0.0403 max mem: 3540 +Test: [2250/2387] eta: 0:00:26 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1610 (0.2855) acc1: 98.4375 (90.1024) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1285 data: 0.0386 max mem: 3540 +Test: [2260/2387] eta: 0:00:24 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2002 (0.2852) acc1: 98.4375 (90.1362) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1328 data: 0.0428 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1742 (0.2846) acc1: 99.2188 (90.1755) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1323 data: 0.0424 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1697 (0.2844) acc1: 98.9583 (90.1868) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1254 data: 0.0350 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1759 (0.2839) acc1: 98.1771 (90.2234) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1237 data: 0.0335 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.1923 (0.2836) acc1: 98.1771 (90.2516) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1252 data: 0.0353 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2214 (0.2834) acc1: 96.3542 (90.2744) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1269 data: 0.0370 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2329 (0.2833) acc1: 95.3125 (90.2891) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1254 data: 0.0353 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2271 (0.2831) acc1: 95.3125 (90.3087) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1234 data: 0.0328 max mem: 3541 +Test: [2340/2387] eta: 0:00:09 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2150 (0.2829) acc1: 95.8333 (90.3309) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1193 data: 0.0291 max mem: 3541 +Test: [2350/2387] eta: 0:00:07 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2075 (0.2829) acc1: 95.8333 (90.3277) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1159 data: 0.0260 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2313 (0.2828) acc1: 95.5729 (90.3416) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1142 data: 0.0243 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2572 (0.2828) acc1: 92.7083 (90.3411) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.1144 data: 0.0250 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.2877 (0.2828) acc1: 91.9271 (90.3494) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.1149 data: 0.0254 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 62.100 acc5_pos: 100.000 loss: 0.3191 (0.2850) acc1: 88.0208 (90.2558) acc5: 100.0000 (100.0000) acc1_neg: 4.651 acc5_neg: 100.000 time: 0.1062 data: 0.0193 max mem: 3541 +Test: Total time: 0:07:34 (0.1906 s / it) +* Overall Acc@1 90.256 Acc@5 100.000 + target == real Acc@1 41.171 Acc@5 100.000 + target == fake Acc@1 93.703 Acc@5 100.000 + loss 0.285 + PR-AUC 0.9903 ROC-AUC 0.8784 EER 19.63% (thr=0.7261) +Accuracy of the network on the 916267 test images: 90.3% +[rank0]:[W726 03:32:06.694503822 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:32:26.554000 1171656 torch/distributed/run.py:766] +W0726 03:32:26.554000 1171656 torch/distributed/run.py:766] ***************************************** +W0726 03:32:26.554000 1171656 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:32:26.554000 1171656 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 03:32:58.124623288 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:32:58.205744961 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:32:58.419021058 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:32:59.094869155 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:32:59.211185766 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +[rank6]:[W726 03:32:59.242629509 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:32:59.249347838 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:32:59.427298635 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n16_seed5_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed5_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:01:53 acc1_pos: 61.198 acc5_pos: 100.000 loss: 0.6364 (0.6364) acc1: 61.1979 (61.1979) acc5: 100.0000 (100.0000) time: 7.1820 data: 5.6605 max mem: 3531 +Test: [ 10/517] eta: 0:06:12 acc1_pos: 36.458 acc5_pos: 100.000 loss: 0.9134 (0.9214) acc1: 40.6250 (38.0445) acc5: 100.0000 (100.0000) time: 0.7347 data: 0.5149 max mem: 3531 +Test: [ 20/517] eta: 0:03:32 acc1_pos: 34.375 acc5_pos: 100.000 loss: 0.9230 (0.9404) acc1: 36.4583 (35.6895) acc5: 100.0000 (100.0000) time: 0.0903 data: 0.0006 max mem: 3531 +Test: [ 30/517] eta: 0:02:59 acc1_pos: 15.104 acc5_pos: 100.000 loss: 0.9406 (0.9632) acc1: 30.7292 (33.0225) acc5: 100.0000 (100.0000) time: 0.1675 data: 0.0765 max mem: 3531 +Test: [ 40/517] eta: 0:02:47 acc1_pos: 39.583 acc5_pos: 100.000 loss: 1.0316 (0.9741) acc1: 26.0417 (31.4405) acc5: 100.0000 (100.0000) time: 0.2703 data: 0.1783 max mem: 3531 +Test: [ 50/517] eta: 0:02:32 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.9004 (0.9252) acc1: 33.3333 (36.1111) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.2637 data: 0.1697 max mem: 3531 +Test: [ 60/517] eta: 0:02:31 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.4731 (0.8390) acc1: 72.3958 (44.1769) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.2945 data: 0.1994 max mem: 3531 +Test: [ 70/517] eta: 0:02:25 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.2982 (0.7581) acc1: 95.3125 (51.6909) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.3238 data: 0.2282 max mem: 3531 +Test: [ 80/517] eta: 0:02:20 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.2609 (0.6993) acc1: 97.9167 (57.2595) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2854 data: 0.1900 max mem: 3531 +Test: [ 90/517] eta: 0:02:14 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.2399 (0.6490) acc1: 98.1771 (61.7245) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2713 data: 0.1772 max mem: 3531 +Test: [100/517] eta: 0:02:09 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.2364 (0.6084) acc1: 98.1771 (65.3311) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2644 data: 0.1678 max mem: 3531 +Test: [110/517] eta: 0:02:03 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.2020 (0.5713) acc1: 98.9583 (68.4098) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2534 data: 0.1547 max mem: 3531 +Test: [120/517] eta: 0:01:59 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.2314 (0.5465) acc1: 98.9583 (70.8054) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2498 data: 0.1495 max mem: 3531 +Test: [130/517] eta: 0:01:54 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.2670 (0.5259) acc1: 98.4375 (72.8173) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2549 data: 0.1562 max mem: 3531 +Test: [140/517] eta: 0:01:50 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.2586 (0.5058) acc1: 98.4375 (74.6454) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2499 data: 0.1532 max mem: 3531 +Test: [150/517] eta: 0:01:46 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.2586 (0.4906) acc1: 98.4375 (76.1038) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.2499 data: 0.1497 max mem: 3531 +Test: [160/517] eta: 0:01:42 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.2470 (0.4753) acc1: 98.1771 (77.4651) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2456 data: 0.1477 max mem: 3531 +Test: [170/517] eta: 0:01:39 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.2360 (0.4619) acc1: 98.4375 (78.6626) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2599 data: 0.1659 max mem: 3531 +Test: [180/517] eta: 0:01:35 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.2338 (0.4511) acc1: 98.1771 (79.5983) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.2597 data: 0.1655 max mem: 3531 +Test: [190/517] eta: 0:01:32 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.1995 (0.4376) acc1: 99.2188 (80.6296) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2449 data: 0.1474 max mem: 3531 +Test: [200/517] eta: 0:01:28 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.2112 (0.4271) acc1: 99.2188 (81.5363) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2494 data: 0.1508 max mem: 3531 +Test: [210/517] eta: 0:01:25 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.2707 (0.4238) acc1: 91.9271 (81.7918) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.2608 data: 0.1658 max mem: 3532 +Test: [220/517] eta: 0:01:23 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.3542 (0.4223) acc1: 86.9792 (81.8934) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.2749 data: 0.1803 max mem: 3532 +Test: [230/517] eta: 0:01:19 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.3482 (0.4170) acc1: 88.5417 (82.5318) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2499 data: 0.1553 max mem: 3532 +Test: [240/517] eta: 0:01:16 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.2432 (0.4076) acc1: 98.4375 (83.2296) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2355 data: 0.1428 max mem: 3532 +Test: [250/517] eta: 0:01:13 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.2112 (0.4021) acc1: 99.2188 (83.7981) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2691 data: 0.1748 max mem: 3532 +Test: [260/517] eta: 0:01:11 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.2747 (0.3987) acc1: 96.8750 (84.2134) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2793 data: 0.1813 max mem: 3532 +Test: [270/517] eta: 0:01:07 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.2520 (0.3927) acc1: 97.1354 (84.7142) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2549 data: 0.1574 max mem: 3532 +Test: [280/517] eta: 0:01:04 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.2276 (0.3864) acc1: 97.6562 (85.1970) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2399 data: 0.1444 max mem: 3532 +Test: [290/517] eta: 0:01:01 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.2276 (0.3815) acc1: 98.9583 (85.6583) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2399 data: 0.1451 max mem: 3532 +Test: [300/517] eta: 0:00:59 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.2353 (0.3764) acc1: 99.2188 (86.1157) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2499 data: 0.1541 max mem: 3532 +Test: [310/517] eta: 0:00:56 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.2353 (0.3724) acc1: 99.2188 (86.5270) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2555 data: 0.1601 max mem: 3532 +Test: [320/517] eta: 0:00:53 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.2504 (0.3693) acc1: 97.6562 (86.8550) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2561 data: 0.1622 max mem: 3532 +Test: [330/517] eta: 0:00:50 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.2535 (0.3657) acc1: 97.6562 (87.1845) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2593 data: 0.1655 max mem: 3532 +Test: [340/517] eta: 0:00:47 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.2339 (0.3611) acc1: 98.6979 (87.5336) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2536 data: 0.1565 max mem: 3532 +Test: [350/517] eta: 0:00:45 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.2417 (0.3581) acc1: 98.6979 (87.8094) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.2510 data: 0.1527 max mem: 3532 +Test: [360/517] eta: 0:00:42 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.2518 (0.3560) acc1: 97.9167 (88.0374) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2599 data: 0.1657 max mem: 3532 +Test: [370/517] eta: 0:00:39 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.2418 (0.3528) acc1: 98.4375 (88.3437) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2588 data: 0.1646 max mem: 3532 +Test: [380/517] eta: 0:00:36 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.2285 (0.3499) acc1: 99.4792 (88.6203) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2558 data: 0.1611 max mem: 3532 +Test: [390/517] eta: 0:00:34 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.2418 (0.3474) acc1: 98.9583 (88.8833) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2601 data: 0.1669 max mem: 3532 +Test: [400/517] eta: 0:00:31 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.2497 (0.3452) acc1: 98.6979 (89.1222) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2592 data: 0.1659 max mem: 3532 +Test: [410/517] eta: 0:00:28 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.2753 (0.3443) acc1: 98.9583 (89.3014) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2555 data: 0.1610 max mem: 3532 +Test: [420/517] eta: 0:00:25 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.2982 (0.3436) acc1: 98.9583 (89.4757) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2550 data: 0.1599 max mem: 3532 +Test: [430/517] eta: 0:00:23 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.3044 (0.3436) acc1: 96.0938 (89.5181) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2548 data: 0.1596 max mem: 3532 +Test: [440/517] eta: 0:00:20 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.2975 (0.3423) acc1: 95.5729 (89.6707) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2544 data: 0.1590 max mem: 3533 +Test: [450/517] eta: 0:00:17 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.2521 (0.3402) acc1: 98.9583 (89.8842) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2392 data: 0.1417 max mem: 3533 +Test: [460/517] eta: 0:00:15 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.2518 (0.3384) acc1: 99.2188 (90.0776) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2366 data: 0.1387 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.2396 (0.3366) acc1: 99.2188 (90.2645) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2622 data: 0.1676 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.2334 (0.3349) acc1: 99.2188 (90.4409) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2433 data: 0.1517 max mem: 3533 +Test: [490/517] eta: 0:00:07 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.2460 (0.3341) acc1: 98.4375 (90.5104) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2276 data: 0.1310 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.3300 (0.3351) acc1: 90.8854 (90.4504) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.2301 data: 0.1319 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.3506 (0.3356) acc1: 89.0625 (90.4354) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1960 data: 0.1042 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 66.972 acc5_pos: 100.000 loss: 0.2914 (0.3349) acc1: 95.8333 (90.4862) acc5: 100.0000 (100.0000) acc1_neg: 97.880 acc5_neg: 100.000 time: 0.1925 data: 0.1013 max mem: 3533 +Test: Total time: 0:02:14 (0.2603 s / it) +* Overall Acc@1 90.486 Acc@5 100.000 + target == real Acc@1 33.534 Acc@5 100.000 + target == fake Acc@1 96.394 Acc@5 100.000 + loss 0.335 + PR-AUC 0.9754 ROC-AUC 0.8315 EER 24.59% (thr=0.7222) +Accuracy of the network on the 198427 test images: 90.5% +[rank0]:[W726 03:35:28.324563514 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 03:35:49.839000 1172229 torch/distributed/run.py:766] +W0726 03:35:49.839000 1172229 torch/distributed/run.py:766] ***************************************** +W0726 03:35:49.839000 1172229 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:35:49.839000 1172229 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:36:19.001167454 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:36:19.406016645 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:36:19.460542832 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:36:20.719252101 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:36:20.951896178 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:36:20.993602277 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W726 03:36:20.993652127 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:36:20.024745478 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n16_seed5_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed5_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:36:59 acc1_pos: 71.094 acc5_pos: 100.000 loss: 0.5676 (0.5676) acc1: 71.0938 (71.0938) acc5: 100.0000 (100.0000) time: 5.7066 data: 3.6188 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/389] eta: 0:03:47 acc1_pos: 91.406 acc5_pos: 100.000 loss: 0.4555 (0.4681) acc1: 79.1667 (78.7642) acc5: 100.0000 (100.0000) time: 0.6007 data: 0.3294 max mem: 3531 +Test: [ 20/389] eta: 0:02:13 acc1_pos: 66.406 acc5_pos: 100.000 loss: 0.4461 (0.5004) acc1: 79.1667 (76.3021) acc5: 100.0000 (100.0000) time: 0.0943 data: 0.0045 max mem: 3531 +Test: [ 30/389] eta: 0:02:01 acc1_pos: 87.500 acc5_pos: 100.000 loss: 0.5339 (0.5176) acc1: 78.9062 (75.9745) acc5: 100.0000 (100.0000) time: 0.1942 data: 0.0993 max mem: 3531 +Test: [ 40/389] eta: 0:01:52 acc1_pos: 52.083 acc5_pos: 100.000 loss: 0.5339 (0.5177) acc1: 77.8646 (75.5335) acc5: 100.0000 (100.0000) time: 0.2798 data: 0.1831 max mem: 3531 +Test: [ 50/389] eta: 0:01:38 acc1_pos: 61.979 acc5_pos: 100.000 loss: 0.5540 (0.5173) acc1: 65.3646 (75.1277) acc5: 100.0000 (100.0000) time: 0.2198 data: 0.1266 max mem: 3531 +Test: [ 60/389] eta: 0:01:31 acc1_pos: 90.365 acc5_pos: 100.000 loss: 0.5741 (0.5211) acc1: 61.9792 (75.1238) acc5: 100.0000 (100.0000) time: 0.1900 data: 0.0980 max mem: 3531 +Test: [ 70/389] eta: 0:01:25 acc1_pos: 66.927 acc5_pos: 100.000 loss: 0.4153 (0.5041) acc1: 89.3229 (76.6359) acc5: 100.0000 (100.0000) time: 0.2099 data: 0.1170 max mem: 3531 +Test: [ 80/389] eta: 0:01:18 acc1_pos: 93.229 acc5_pos: 100.000 loss: 0.4265 (0.5070) acc1: 82.8125 (76.7136) acc5: 100.0000 (100.0000) time: 0.1786 data: 0.0852 max mem: 3531 +Test: [ 90/389] eta: 0:01:13 acc1_pos: 91.406 acc5_pos: 100.000 loss: 0.2815 (0.4832) acc1: 91.4062 (78.7946) acc5: 100.0000 (100.0000) time: 0.1623 data: 0.0702 max mem: 3531 +Test: [100/389] eta: 0:01:08 acc1_pos: 73.177 acc5_pos: 100.000 loss: 0.4444 (0.4984) acc1: 83.5938 (77.3180) acc5: 100.0000 (100.0000) time: 0.1769 data: 0.0857 max mem: 3531 +Test: [110/389] eta: 0:01:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5844 (0.5073) acc1: 69.0104 (76.2059) acc5: 100.0000 (100.0000) time: 0.1691 data: 0.0789 max mem: 3531 +Test: [120/389] eta: 0:01:00 acc1_pos: 84.375 acc5_pos: 100.000 loss: 0.5275 (0.5027) acc1: 71.3542 (76.4613) acc5: 100.0000 (100.0000) time: 0.1591 data: 0.0685 max mem: 3531 +Test: [130/389] eta: 0:00:57 acc1_pos: 49.693 acc5_pos: 100.000 loss: 0.5275 (0.5117) acc1: 74.7396 (75.7316) acc5: 100.0000 (100.0000) acc1_neg: 74.740 acc5_neg: 100.000 time: 0.1584 data: 0.0673 max mem: 3531 +Test: [140/389] eta: 0:00:53 acc1_pos: 49.693 acc5_pos: 100.000 loss: 0.6288 (0.5279) acc1: 64.3229 (74.5253) acc5: 100.0000 (100.0000) acc1_neg: 23.958 acc5_neg: 100.000 time: 0.1622 data: 0.0708 max mem: 3531 +Test: [150/389] eta: 0:00:50 acc1_pos: 49.693 acc5_pos: 100.000 loss: 0.6770 (0.5448) acc1: 53.9062 (72.8235) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1619 data: 0.0682 max mem: 3531 +Test: [160/389] eta: 0:00:48 acc1_pos: 49.693 acc5_pos: 100.000 loss: 0.4280 (0.5395) acc1: 76.5625 (73.4359) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1751 data: 0.0818 max mem: 3531 +Test: [170/389] eta: 0:00:46 acc1_pos: 49.693 acc5_pos: 100.000 loss: 0.3155 (0.5411) acc1: 94.7917 (73.2746) acc5: 100.0000 (100.0000) acc1_neg: 22.656 acc5_neg: 100.000 time: 0.1978 data: 0.1031 max mem: 3531 +Test: [180/389] eta: 0:00:43 acc1_pos: 49.693 acc5_pos: 100.000 loss: 0.4595 (0.5486) acc1: 74.2188 (72.6491) acc5: 100.0000 (100.0000) acc1_neg: 62.240 acc5_neg: 100.000 time: 0.1932 data: 0.0980 max mem: 3531 +Test: [190/389] eta: 0:00:41 acc1_pos: 49.693 acc5_pos: 100.000 loss: 0.6506 (0.5642) acc1: 59.1146 (71.2083) acc5: 100.0000 (100.0000) acc1_neg: 2.344 acc5_neg: 100.000 time: 0.1877 data: 0.0952 max mem: 3531 +Test: [200/389] eta: 0:00:39 acc1_pos: 49.693 acc5_pos: 100.000 loss: 0.6372 (0.5711) acc1: 59.1146 (70.6973) acc5: 100.0000 (100.0000) acc1_neg: 58.073 acc5_neg: 100.000 time: 0.1976 data: 0.1056 max mem: 3531 +Test: [210/389] eta: 0:00:37 acc1_pos: 49.693 acc5_pos: 100.000 loss: 0.4573 (0.5629) acc1: 83.5938 (71.3159) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1957 data: 0.1038 max mem: 3532 +Test: [220/389] eta: 0:00:35 acc1_pos: 49.693 acc5_pos: 100.000 loss: 0.2560 (0.5509) acc1: 96.6146 (72.2002) acc5: 100.0000 (100.0000) acc1_neg: 71.354 acc5_neg: 100.000 time: 0.2114 data: 0.1193 max mem: 3532 +Test: [230/389] eta: 0:00:33 acc1_pos: 49.693 acc5_pos: 100.000 loss: 0.4943 (0.5583) acc1: 72.1354 (71.5447) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.2185 data: 0.1268 max mem: 3532 +Test: [240/389] eta: 0:00:30 acc1_pos: 49.693 acc5_pos: 100.000 loss: 0.6020 (0.5579) acc1: 65.1042 (71.5011) acc5: 100.0000 (100.0000) acc1_neg: 44.531 acc5_neg: 100.000 time: 0.1699 data: 0.0786 max mem: 3532 +Test: [250/389] eta: 0:00:28 acc1_pos: 49.693 acc5_pos: 100.000 loss: 0.4371 (0.5567) acc1: 79.1667 (71.7744) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1632 data: 0.0726 max mem: 3532 +Test: [260/389] eta: 0:00:26 acc1_pos: 49.693 acc5_pos: 100.000 loss: 0.4371 (0.5550) acc1: 91.9271 (72.0137) acc5: 100.0000 (100.0000) acc1_neg: 41.406 acc5_neg: 100.000 time: 0.1866 data: 0.0962 max mem: 3532 +Test: [270/389] eta: 0:00:24 acc1_pos: 49.693 acc5_pos: 100.000 loss: 0.7104 (0.5621) acc1: 52.6042 (71.1898) acc5: 100.0000 (100.0000) acc1_neg: 54.427 acc5_neg: 100.000 time: 0.1874 data: 0.0952 max mem: 3532 +Test: [280/389] eta: 0:00:22 acc1_pos: 49.693 acc5_pos: 100.000 loss: 0.7525 (0.5743) acc1: 45.3125 (70.0576) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1845 data: 0.0916 max mem: 3532 +Test: [290/389] eta: 0:00:19 acc1_pos: 49.693 acc5_pos: 100.000 loss: 0.6781 (0.5723) acc1: 54.1667 (70.3009) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1770 data: 0.0854 max mem: 3532 +Test: [300/389] eta: 0:00:17 acc1_pos: 49.693 acc5_pos: 100.000 loss: 0.4595 (0.5697) acc1: 81.7708 (70.7157) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.1765 data: 0.0851 max mem: 3532 +Test: [310/389] eta: 0:00:15 acc1_pos: 49.693 acc5_pos: 100.000 loss: 0.6175 (0.5772) acc1: 69.5312 (70.0127) acc5: 100.0000 (100.0000) acc1_neg: 9.375 acc5_neg: 100.000 time: 0.1799 data: 0.0887 max mem: 3532 +Test: [320/389] eta: 0:00:13 acc1_pos: 49.693 acc5_pos: 100.000 loss: 0.7400 (0.5802) acc1: 52.8646 (69.5913) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1794 data: 0.0882 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 49.693 acc5_pos: 100.000 loss: 0.9521 (0.5954) acc1: 23.4375 (68.0584) acc5: 100.0000 (100.0000) acc1_neg: 31.771 acc5_neg: 100.000 time: 0.1877 data: 0.0963 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 49.693 acc5_pos: 100.000 loss: 0.4605 (0.5893) acc1: 65.6250 (68.8409) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1660 data: 0.0749 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 49.693 acc5_pos: 100.000 loss: 0.3839 (0.6016) acc1: 94.5312 (68.1164) acc5: 100.0000 (100.0000) acc1_neg: 1.302 acc5_neg: 100.000 time: 0.1404 data: 0.0496 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 49.693 acc5_pos: 100.000 loss: 1.0062 (0.6161) acc1: 25.0000 (66.7972) acc5: 100.0000 (100.0000) acc1_neg: 25.521 acc5_neg: 100.000 time: 0.1456 data: 0.0554 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 49.693 acc5_pos: 100.000 loss: 0.9293 (0.6134) acc1: 28.1250 (67.0275) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1459 data: 0.0549 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 49.693 acc5_pos: 100.000 loss: 0.5783 (0.6159) acc1: 67.4479 (66.7542) acc5: 100.0000 (100.0000) acc1_neg: 44.792 acc5_neg: 100.000 time: 0.1443 data: 0.0535 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 49.693 acc5_pos: 100.000 loss: 0.5198 (0.6142) acc1: 76.8229 (66.8428) acc5: 100.0000 (100.0000) acc1_neg: 83.060 acc5_neg: 100.000 time: 0.1352 data: 0.0411 max mem: 3532 +Test: Total time: 0:01:14 (0.1911 s / it) +* Overall Acc@1 66.843 Acc@5 100.000 + target == real Acc@1 75.953 Acc@5 100.000 + target == fake Acc@1 62.608 Acc@5 100.000 + loss 0.614 + PR-AUC 0.8689 ROC-AUC 0.7578 EER 31.74% (thr=0.4524) +Accuracy of the network on the 149358 test images: 66.8% +[rank0]:[W726 03:37:45.128798197 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:38:05.691000 1172789 torch/distributed/run.py:766] +W0726 03:38:05.691000 1172789 torch/distributed/run.py:766] ***************************************** +W0726 03:38:05.691000 1172789 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:38:05.691000 1172789 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:38:35.285195948 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:38:36.522276853 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:38:36.109253242 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:38:36.341891138 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:38:36.344376253 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:38:37.501136760 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:38:37.707436194 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:38:37.742911968 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n16_seed5_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed5_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:29:35 acc1_pos: 85.156 acc5_pos: 100.000 loss: 0.3486 (0.3486) acc1: 85.1562 (85.1562) acc5: 100.0000 (100.0000) time: 5.3308 data: 3.6104 max mem: 3531 +Test: [ 10/333] eta: 0:03:02 acc1_pos: 88.021 acc5_pos: 100.000 loss: 0.2845 (0.2848) acc1: 86.4583 (87.8078) acc5: 100.0000 (100.0000) time: 0.5662 data: 0.3285 max mem: 3531 +Test: [ 20/333] eta: 0:01:47 acc1_pos: 97.135 acc5_pos: 100.000 loss: 0.2693 (0.2740) acc1: 88.2812 (89.1741) acc5: 100.0000 (100.0000) time: 0.0934 data: 0.0004 max mem: 3531 +Test: [ 30/333] eta: 0:01:24 acc1_pos: 74.479 acc5_pos: 100.000 loss: 0.3203 (0.3882) acc1: 82.0312 (81.3172) acc5: 100.0000 (100.0000) time: 0.1207 data: 0.0249 max mem: 3531 +Test: [ 40/333] eta: 0:01:16 acc1_pos: 65.104 acc5_pos: 100.000 loss: 0.6617 (0.4645) acc1: 61.7188 (76.2386) acc5: 100.0000 (100.0000) time: 0.1749 data: 0.0805 max mem: 3531 +Test: [ 50/333] eta: 0:01:10 acc1_pos: 66.146 acc5_pos: 100.000 loss: 0.6592 (0.5065) acc1: 61.7188 (73.2026) acc5: 100.0000 (100.0000) time: 0.2026 data: 0.1081 max mem: 3531 +Test: [ 60/333] eta: 0:01:05 acc1_pos: 61.458 acc5_pos: 100.000 loss: 0.6524 (0.5334) acc1: 59.8958 (71.2901) acc5: 100.0000 (100.0000) time: 0.1953 data: 0.0979 max mem: 3531 +Test: [ 70/333] eta: 0:01:01 acc1_pos: 52.344 acc5_pos: 100.000 loss: 0.6531 (0.5533) acc1: 59.3750 (69.8834) acc5: 100.0000 (100.0000) time: 0.1899 data: 0.0917 max mem: 3531 +Test: [ 80/333] eta: 0:00:59 acc1_pos: 62.240 acc5_pos: 100.000 loss: 0.6729 (0.5681) acc1: 62.2396 (68.6536) acc5: 100.0000 (100.0000) time: 0.2200 data: 0.1247 max mem: 3531 +Test: [ 90/333] eta: 0:00:55 acc1_pos: 74.740 acc5_pos: 100.000 loss: 0.6025 (0.5696) acc1: 64.8438 (68.8072) acc5: 100.0000 (100.0000) time: 0.2199 data: 0.1247 max mem: 3531 +Test: [100/333] eta: 0:00:52 acc1_pos: 77.604 acc5_pos: 100.000 loss: 0.5681 (0.5701) acc1: 70.8333 (68.8351) acc5: 100.0000 (100.0000) time: 0.1845 data: 0.0875 max mem: 3531 +Test: [110/333] eta: 0:00:49 acc1_pos: 79.167 acc5_pos: 100.000 loss: 0.5839 (0.5709) acc1: 67.9688 (68.7875) acc5: 100.0000 (100.0000) time: 0.1849 data: 0.0880 max mem: 3531 +Test: [120/333] eta: 0:00:46 acc1_pos: 73.958 acc5_pos: 100.000 loss: 0.5569 (0.5716) acc1: 70.0521 (68.8060) acc5: 100.0000 (100.0000) time: 0.2000 data: 0.1039 max mem: 3531 +Test: [130/333] eta: 0:00:45 acc1_pos: 86.198 acc5_pos: 100.000 loss: 0.5237 (0.5699) acc1: 72.1354 (69.0263) acc5: 100.0000 (100.0000) time: 0.2298 data: 0.1341 max mem: 3531 +Test: [140/333] eta: 0:00:43 acc1_pos: 82.812 acc5_pos: 100.000 loss: 0.5606 (0.5679) acc1: 70.5729 (69.2043) acc5: 100.0000 (100.0000) time: 0.2397 data: 0.1440 max mem: 3531 +Test: [150/333] eta: 0:00:40 acc1_pos: 71.615 acc5_pos: 100.000 loss: 0.5672 (0.5692) acc1: 70.5729 (69.1932) acc5: 100.0000 (100.0000) time: 0.2199 data: 0.1252 max mem: 3531 +Test: [160/333] eta: 0:00:38 acc1_pos: 62.829 acc5_pos: 100.000 loss: 0.5877 (0.5824) acc1: 65.3646 (68.2195) acc5: 100.0000 (100.0000) acc1_neg: 42.969 acc5_neg: 100.000 time: 0.1965 data: 0.1018 max mem: 3531 +Test: [170/333] eta: 0:00:35 acc1_pos: 62.829 acc5_pos: 100.000 loss: 0.7686 (0.5972) acc1: 53.9062 (67.1296) acc5: 100.0000 (100.0000) acc1_neg: 54.948 acc5_neg: 100.000 time: 0.2049 data: 0.1099 max mem: 3531 +Test: [180/333] eta: 0:00:33 acc1_pos: 62.829 acc5_pos: 100.000 loss: 0.7542 (0.6040) acc1: 54.9479 (66.6106) acc5: 100.0000 (100.0000) acc1_neg: 42.708 acc5_neg: 100.000 time: 0.2110 data: 0.1187 max mem: 3531 +Test: [190/333] eta: 0:00:31 acc1_pos: 62.829 acc5_pos: 100.000 loss: 0.6967 (0.6095) acc1: 57.0312 (66.1854) acc5: 100.0000 (100.0000) acc1_neg: 58.594 acc5_neg: 100.000 time: 0.1899 data: 0.0971 max mem: 3531 +Test: [200/333] eta: 0:00:29 acc1_pos: 62.829 acc5_pos: 100.000 loss: 0.7048 (0.6141) acc1: 57.0312 (65.8336) acc5: 100.0000 (100.0000) acc1_neg: 57.552 acc5_neg: 100.000 time: 0.2123 data: 0.1173 max mem: 3531 +Test: [210/333] eta: 0:00:26 acc1_pos: 62.829 acc5_pos: 100.000 loss: 0.7367 (0.6226) acc1: 57.0312 (65.2807) acc5: 100.0000 (100.0000) acc1_neg: 42.969 acc5_neg: 100.000 time: 0.2262 data: 0.1300 max mem: 3532 +Test: [220/333] eta: 0:00:24 acc1_pos: 62.829 acc5_pos: 100.000 loss: 0.8206 (0.6352) acc1: 49.7396 (64.3430) acc5: 100.0000 (100.0000) acc1_neg: 69.271 acc5_neg: 100.000 time: 0.2109 data: 0.1162 max mem: 3532 +Test: [230/333] eta: 0:00:22 acc1_pos: 62.829 acc5_pos: 100.000 loss: 0.5435 (0.6291) acc1: 69.2708 (64.8753) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.2086 data: 0.1176 max mem: 3532 +Test: [240/333] eta: 0:00:20 acc1_pos: 62.829 acc5_pos: 100.000 loss: 0.5390 (0.6294) acc1: 69.7917 (64.7962) acc5: 100.0000 (100.0000) acc1_neg: 60.677 acc5_neg: 100.000 time: 0.2053 data: 0.1149 max mem: 3532 +Test: [250/333] eta: 0:00:17 acc1_pos: 62.829 acc5_pos: 100.000 loss: 0.6676 (0.6354) acc1: 60.4167 (64.3987) acc5: 100.0000 (100.0000) acc1_neg: 46.094 acc5_neg: 100.000 time: 0.2117 data: 0.1198 max mem: 3532 +Test: [260/333] eta: 0:00:15 acc1_pos: 62.829 acc5_pos: 100.000 loss: 0.8433 (0.6454) acc1: 50.5208 (63.7273) acc5: 100.0000 (100.0000) acc1_neg: 51.823 acc5_neg: 100.000 time: 0.2201 data: 0.1279 max mem: 3532 +Test: [270/333] eta: 0:00:13 acc1_pos: 62.829 acc5_pos: 100.000 loss: 0.7064 (0.6454) acc1: 55.2083 (63.7675) acc5: 100.0000 (100.0000) acc1_neg: 64.062 acc5_neg: 100.000 time: 0.2092 data: 0.1186 max mem: 3532 +Test: [280/333] eta: 0:00:11 acc1_pos: 62.829 acc5_pos: 100.000 loss: 0.6769 (0.6509) acc1: 62.2396 (63.3202) acc5: 100.0000 (100.0000) acc1_neg: 53.125 acc5_neg: 100.000 time: 0.1793 data: 0.0887 max mem: 3532 +Test: [290/333] eta: 0:00:09 acc1_pos: 62.829 acc5_pos: 100.000 loss: 0.9158 (0.6610) acc1: 42.9688 (62.5367) acc5: 100.0000 (100.0000) acc1_neg: 44.792 acc5_neg: 100.000 time: 0.1640 data: 0.0717 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 62.829 acc5_pos: 100.000 loss: 0.9473 (0.6708) acc1: 40.1042 (61.8312) acc5: 100.0000 (100.0000) acc1_neg: 41.406 acc5_neg: 100.000 time: 0.1604 data: 0.0669 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 62.829 acc5_pos: 100.000 loss: 0.8310 (0.6703) acc1: 50.2604 (61.8820) acc5: 100.0000 (100.0000) acc1_neg: 60.156 acc5_neg: 100.000 time: 0.1452 data: 0.0529 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 62.829 acc5_pos: 100.000 loss: 0.6557 (0.6715) acc1: 61.9792 (61.7861) acc5: 100.0000 (100.0000) acc1_neg: 41.406 acc5_neg: 100.000 time: 0.1332 data: 0.0421 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 62.829 acc5_pos: 100.000 loss: 0.9335 (0.6818) acc1: 43.2292 (61.1515) acc5: 100.0000 (100.0000) acc1_neg: 41.146 acc5_neg: 100.000 time: 0.1242 data: 0.0337 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 62.829 acc5_pos: 100.000 loss: 0.9793 (0.6837) acc1: 42.1875 (61.0436) acc5: 100.0000 (100.0000) acc1_neg: 42.817 acc5_neg: 100.000 time: 0.1124 data: 0.0149 max mem: 3532 +Test: Total time: 0:01:07 (0.2034 s / it) +* Overall Acc@1 61.044 Acc@5 100.000 + target == real Acc@1 69.079 Acc@5 100.000 + target == fake Acc@1 53.891 Acc@5 100.000 + loss 0.684 + PR-AUC 0.6814 ROC-AUC 0.6676 EER 38.48% (thr=0.4402) +Accuracy of the network on the 127843 test images: 61.0% +[rank0]:[W726 03:39:53.195535770 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:40:13.363000 1173347 torch/distributed/run.py:766] +W0726 03:40:13.363000 1173347 torch/distributed/run.py:766] ***************************************** +W0726 03:40:13.363000 1173347 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:40:13.363000 1173347 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:40:43.520145670 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:40:43.664862118 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:40:43.728598745 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:40:43.047135380 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +[rank6]:[W726 03:40:43.078591758 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W726 03:40:43.078641179 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:40:43.094381907 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank1]:[W726 03:40:43.096921334 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n16_seed5_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed5_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:22:12 acc1_pos: 97.135 acc5_pos: 100.000 loss: 0.1696 (0.1696) acc1: 97.1354 (97.1354) acc5: 100.0000 (100.0000) time: 5.6690 data: 3.7142 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/235] eta: 0:02:14 acc1_pos: 96.094 acc5_pos: 100.000 loss: 0.1777 (0.1754) acc1: 97.1354 (97.0644) acc5: 100.0000 (100.0000) time: 0.5972 data: 0.3380 max mem: 3531 +Test: [ 20/235] eta: 0:01:18 acc1_pos: 98.177 acc5_pos: 100.000 loss: 0.1701 (0.1715) acc1: 97.3958 (97.3958) acc5: 100.0000 (100.0000) time: 0.0976 data: 0.0078 max mem: 3531 +Test: [ 30/235] eta: 0:01:05 acc1_pos: 97.656 acc5_pos: 100.000 loss: 0.1643 (0.1699) acc1: 97.9167 (97.5050) acc5: 100.0000 (100.0000) time: 0.1640 data: 0.0737 max mem: 3531 +Test: [ 40/235] eta: 0:00:58 acc1_pos: 96.354 acc5_pos: 100.000 loss: 0.1659 (0.1701) acc1: 97.6562 (97.5356) acc5: 100.0000 (100.0000) time: 0.2387 data: 0.1440 max mem: 3531 +Test: [ 50/235] eta: 0:00:51 acc1_pos: 97.917 acc5_pos: 100.000 loss: 0.1738 (0.1709) acc1: 97.3958 (97.4929) acc5: 100.0000 (100.0000) time: 0.2242 data: 0.1298 max mem: 3531 +Test: [ 60/235] eta: 0:00:47 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.1763 (0.1725) acc1: 97.3958 (97.4129) acc5: 100.0000 (100.0000) time: 0.1999 data: 0.1075 max mem: 3531 +Test: [ 70/235] eta: 0:00:43 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1902 (0.3342) acc1: 96.0938 (87.5073) acc5: 100.0000 (100.0000) acc1_neg: 15.104 acc5_neg: 100.000 time: 0.2089 data: 0.1158 max mem: 3531 +Test: [ 80/235] eta: 0:00:39 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.4639 (0.4739) acc1: 18.4896 (78.9577) acc5: 100.0000 (100.0000) acc1_neg: 17.188 acc5_neg: 100.000 time: 0.2061 data: 0.1152 max mem: 3531 +Test: [ 90/235] eta: 0:00:35 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.4670 (0.5841) acc1: 17.9688 (72.2155) acc5: 100.0000 (100.0000) acc1_neg: 16.667 acc5_neg: 100.000 time: 0.1990 data: 0.1073 max mem: 3531 +Test: [100/235] eta: 0:00:32 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.4697 (0.6708) acc1: 17.4479 (66.8729) acc5: 100.0000 (100.0000) acc1_neg: 16.146 acc5_neg: 100.000 time: 0.1743 data: 0.0820 max mem: 3531 +Test: [110/235] eta: 0:00:29 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.4512 (0.7418) acc1: 17.9688 (62.5516) acc5: 100.0000 (100.0000) acc1_neg: 16.667 acc5_neg: 100.000 time: 0.1816 data: 0.0903 max mem: 3531 +Test: [120/235] eta: 0:00:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.5019 (0.8103) acc1: 15.8854 (58.3054) acc5: 100.0000 (100.0000) acc1_neg: 9.375 acc5_neg: 100.000 time: 0.1966 data: 0.1059 max mem: 3531 +Test: [130/235] eta: 0:00:23 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.5714 (0.8690) acc1: 11.1979 (54.7292) acc5: 100.0000 (100.0000) acc1_neg: 11.458 acc5_neg: 100.000 time: 0.1683 data: 0.0772 max mem: 3531 +Test: [140/235] eta: 0:00:21 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.5988 (0.9218) acc1: 10.4167 (51.5662) acc5: 100.0000 (100.0000) acc1_neg: 8.073 acc5_neg: 100.000 time: 0.1635 data: 0.0719 max mem: 3531 +Test: [150/235] eta: 0:00:18 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.5957 (0.9655) acc1: 10.1562 (48.8342) acc5: 100.0000 (100.0000) acc1_neg: 11.458 acc5_neg: 100.000 time: 0.1849 data: 0.0925 max mem: 3531 +Test: [160/235] eta: 0:00:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.5814 (1.0035) acc1: 10.9375 (46.5062) acc5: 100.0000 (100.0000) acc1_neg: 14.323 acc5_neg: 100.000 time: 0.1940 data: 0.1005 max mem: 3531 +Test: [170/235] eta: 0:00:13 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.5441 (1.0308) acc1: 14.3229 (44.7719) acc5: 100.0000 (100.0000) acc1_neg: 17.188 acc5_neg: 100.000 time: 0.1673 data: 0.0731 max mem: 3531 +Test: [180/235] eta: 0:00:11 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.4580 (1.0548) acc1: 17.1875 (43.2666) acc5: 100.0000 (100.0000) acc1_neg: 16.927 acc5_neg: 100.000 time: 0.1600 data: 0.0677 max mem: 3531 +Test: [190/235] eta: 0:00:09 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.4573 (1.0758) acc1: 17.4479 (41.9148) acc5: 100.0000 (100.0000) acc1_neg: 16.406 acc5_neg: 100.000 time: 0.1728 data: 0.0806 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.4627 (1.0956) acc1: 16.6667 (40.6470) acc5: 100.0000 (100.0000) acc1_neg: 15.365 acc5_neg: 100.000 time: 0.1752 data: 0.0832 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.4708 (1.1126) acc1: 16.4062 (39.5340) acc5: 100.0000 (100.0000) acc1_neg: 16.406 acc5_neg: 100.000 time: 0.2047 data: 0.1123 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.4366 (1.1280) acc1: 16.4062 (38.5205) acc5: 100.0000 (100.0000) acc1_neg: 16.146 acc5_neg: 100.000 time: 0.2063 data: 0.1120 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.4436 (1.1419) acc1: 17.9688 (37.6567) acc5: 100.0000 (100.0000) acc1_neg: 17.708 acc5_neg: 100.000 time: 0.1597 data: 0.0675 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.4530 (1.1481) acc1: 17.9688 (37.3791) acc5: 100.0000 (100.0000) acc1_neg: 13.043 acc5_neg: 100.000 time: 0.1548 data: 0.0674 max mem: 3532 +Test: Total time: 0:00:47 (0.2029 s / it) +* Overall Acc@1 37.379 Acc@5 100.000 + target == real Acc@1 97.424 Acc@5 100.000 + target == fake Acc@1 15.662 Acc@5 100.000 + loss 1.148 + PR-AUC 0.8859 ROC-AUC 0.7566 EER 31.16% (thr=0.1477) +Accuracy of the network on the 89879 test images: 37.4% +[rank0]:[W726 03:41:42.266394694 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:42:03.230000 1173895 torch/distributed/run.py:766] +W0726 03:42:03.230000 1173895 torch/distributed/run.py:766] ***************************************** +W0726 03:42:03.230000 1173895 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:42:03.230000 1173895 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:42:32.524619128 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:42:32.580980723 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:42:32.099918943 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:42:32.216065823 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:42:32.357939266 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:42:32.423074785 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:42:32.435137478 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:42:32.448301974 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n16_seed5_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed5_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:27:35 acc1_pos: 72.917 acc5_pos: 100.000 loss: 0.5414 (0.5414) acc1: 72.9167 (72.9167) acc5: 100.0000 (100.0000) time: 6.1527 data: 4.6345 max mem: 3531 +Test: [ 10/269] eta: 0:02:48 acc1_pos: 88.542 acc5_pos: 100.000 loss: 0.4605 (0.4457) acc1: 82.2917 (81.8182) acc5: 100.0000 (100.0000) time: 0.6498 data: 0.4303 max mem: 3531 +Test: [ 20/269] eta: 0:01:41 acc1_pos: 86.198 acc5_pos: 100.000 loss: 0.4451 (0.4665) acc1: 82.2917 (79.9727) acc5: 100.0000 (100.0000) time: 0.1194 data: 0.0297 max mem: 3531 +Test: [ 30/269] eta: 0:01:26 acc1_pos: 83.073 acc5_pos: 100.000 loss: 0.4509 (0.4548) acc1: 83.0729 (81.7372) acc5: 100.0000 (100.0000) time: 0.2052 data: 0.1119 max mem: 3531 +Test: [ 40/269] eta: 0:01:16 acc1_pos: 65.517 acc5_pos: 100.000 loss: 0.4872 (0.5057) acc1: 76.8229 (76.8737) acc5: 100.0000 (100.0000) acc1_neg: 46.094 acc5_neg: 100.000 time: 0.2554 data: 0.1620 max mem: 3531 +Test: [ 50/269] eta: 0:01:04 acc1_pos: 65.517 acc5_pos: 100.000 loss: 0.6610 (0.5466) acc1: 60.1562 (73.1158) acc5: 100.0000 (100.0000) acc1_neg: 32.031 acc5_neg: 100.000 time: 0.1911 data: 0.0988 max mem: 3531 +Test: [ 60/269] eta: 0:00:59 acc1_pos: 65.517 acc5_pos: 100.000 loss: 0.6929 (0.5858) acc1: 57.0312 (69.3861) acc5: 100.0000 (100.0000) acc1_neg: 25.781 acc5_neg: 100.000 time: 0.1849 data: 0.0924 max mem: 3531 +Test: [ 70/269] eta: 0:00:53 acc1_pos: 65.517 acc5_pos: 100.000 loss: 0.7967 (0.6107) acc1: 46.8750 (66.5676) acc5: 100.0000 (100.0000) acc1_neg: 50.781 acc5_neg: 100.000 time: 0.2058 data: 0.1127 max mem: 3531 +Test: [ 80/269] eta: 0:00:48 acc1_pos: 65.517 acc5_pos: 100.000 loss: 0.7424 (0.6233) acc1: 48.1771 (64.9306) acc5: 100.0000 (100.0000) acc1_neg: 48.177 acc5_neg: 100.000 time: 0.1758 data: 0.0818 max mem: 3531 +Test: [ 90/269] eta: 0:00:44 acc1_pos: 65.517 acc5_pos: 100.000 loss: 0.7611 (0.6411) acc1: 49.4792 (63.0094) acc5: 100.0000 (100.0000) acc1_neg: 28.646 acc5_neg: 100.000 time: 0.1828 data: 0.0920 max mem: 3531 +Test: [100/269] eta: 0:00:40 acc1_pos: 65.517 acc5_pos: 100.000 loss: 0.7868 (0.6577) acc1: 49.4792 (61.9534) acc5: 100.0000 (100.0000) acc1_neg: 57.552 acc5_neg: 100.000 time: 0.1583 data: 0.0669 max mem: 3531 +Test: [110/269] eta: 0:00:36 acc1_pos: 65.517 acc5_pos: 100.000 loss: 0.8529 (0.6793) acc1: 43.7500 (60.0906) acc5: 100.0000 (100.0000) acc1_neg: 43.750 acc5_neg: 100.000 time: 0.1439 data: 0.0485 max mem: 3531 +Test: [120/269] eta: 0:00:33 acc1_pos: 65.517 acc5_pos: 100.000 loss: 0.7498 (0.6759) acc1: 47.6562 (60.3629) acc5: 100.0000 (100.0000) acc1_neg: 59.635 acc5_neg: 100.000 time: 0.1649 data: 0.0680 max mem: 3531 +Test: [130/269] eta: 0:00:31 acc1_pos: 65.517 acc5_pos: 100.000 loss: 0.6508 (0.6813) acc1: 59.3750 (59.8342) acc5: 100.0000 (100.0000) acc1_neg: 56.250 acc5_neg: 100.000 time: 0.1949 data: 0.0991 max mem: 3531 +Test: [140/269] eta: 0:00:28 acc1_pos: 65.517 acc5_pos: 100.000 loss: 0.7063 (0.6814) acc1: 54.6875 (59.6280) acc5: 100.0000 (100.0000) acc1_neg: 54.427 acc5_neg: 100.000 time: 0.2176 data: 0.1244 max mem: 3531 +Test: [150/269] eta: 0:00:26 acc1_pos: 65.517 acc5_pos: 100.000 loss: 0.6207 (0.6771) acc1: 58.3333 (60.0286) acc5: 100.0000 (100.0000) acc1_neg: 74.740 acc5_neg: 100.000 time: 0.1799 data: 0.0880 max mem: 3531 +Test: [160/269] eta: 0:00:23 acc1_pos: 65.517 acc5_pos: 100.000 loss: 0.6211 (0.6787) acc1: 58.3333 (59.8975) acc5: 100.0000 (100.0000) acc1_neg: 45.573 acc5_neg: 100.000 time: 0.1673 data: 0.0728 max mem: 3531 +Test: [170/269] eta: 0:00:21 acc1_pos: 65.517 acc5_pos: 100.000 loss: 0.7945 (0.6891) acc1: 49.7396 (59.0415) acc5: 100.0000 (100.0000) acc1_neg: 36.979 acc5_neg: 100.000 time: 0.2001 data: 0.1062 max mem: 3531 +Test: [180/269] eta: 0:00:19 acc1_pos: 65.517 acc5_pos: 100.000 loss: 0.8534 (0.6939) acc1: 46.6146 (58.6398) acc5: 100.0000 (100.0000) acc1_neg: 45.312 acc5_neg: 100.000 time: 0.2098 data: 0.1143 max mem: 3531 +Test: [190/269] eta: 0:00:16 acc1_pos: 65.517 acc5_pos: 100.000 loss: 0.7437 (0.6959) acc1: 52.0833 (58.4751) acc5: 100.0000 (100.0000) acc1_neg: 51.823 acc5_neg: 100.000 time: 0.1746 data: 0.0787 max mem: 3531 +Test: [200/269] eta: 0:00:14 acc1_pos: 65.517 acc5_pos: 100.000 loss: 0.7510 (0.7010) acc1: 51.8229 (58.2012) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.1699 data: 0.0769 max mem: 3531 +Test: [210/269] eta: 0:00:12 acc1_pos: 65.517 acc5_pos: 100.000 loss: 0.7823 (0.7031) acc1: 50.0000 (57.9729) acc5: 100.0000 (100.0000) acc1_neg: 60.938 acc5_neg: 100.000 time: 0.1999 data: 0.1035 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 65.517 acc5_pos: 100.000 loss: 0.7362 (0.7058) acc1: 48.1771 (57.6935) acc5: 100.0000 (100.0000) acc1_neg: 65.885 acc5_neg: 100.000 time: 0.1899 data: 0.0910 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 65.517 acc5_pos: 100.000 loss: 0.7183 (0.7055) acc1: 48.1771 (57.5803) acc5: 100.0000 (100.0000) acc1_neg: 61.198 acc5_neg: 100.000 time: 0.1695 data: 0.0721 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 65.517 acc5_pos: 100.000 loss: 0.6940 (0.7059) acc1: 49.7396 (57.4948) acc5: 100.0000 (100.0000) acc1_neg: 59.635 acc5_neg: 100.000 time: 0.1725 data: 0.0778 max mem: 3532 +Test: [250/269] eta: 0:00:03 acc1_pos: 65.517 acc5_pos: 100.000 loss: 0.6940 (0.7040) acc1: 55.9896 (57.6527) acc5: 100.0000 (100.0000) acc1_neg: 47.656 acc5_neg: 100.000 time: 0.1750 data: 0.0817 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 65.517 acc5_pos: 100.000 loss: 0.7334 (0.7082) acc1: 52.6042 (57.2248) acc5: 100.0000 (100.0000) acc1_neg: 38.021 acc5_neg: 100.000 time: 0.1614 data: 0.0669 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 65.517 acc5_pos: 100.000 loss: 0.7720 (0.7071) acc1: 49.2188 (57.3282) acc5: 100.0000 (100.0000) acc1_neg: 85.564 acc5_neg: 100.000 time: 0.1538 data: 0.0456 max mem: 3532 +Test: Total time: 0:00:54 (0.2033 s / it) +* Overall Acc@1 57.328 Acc@5 100.000 + target == real Acc@1 81.433 Acc@5 100.000 + target == fake Acc@1 54.038 Acc@5 100.000 + loss 0.707 + PR-AUC 0.9542 ROC-AUC 0.7502 EER 32.08% (thr=0.3982) +Accuracy of the network on the 103293 test images: 57.3% +[rank0]:[W726 03:43:40.239084456 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T03:43:47Z] eval_done tag=fixed500k_fake_n16_seed5_resampled +[2026-07-26T03:43:47Z] eval_start tag=fixed500k_real_n8_seed6_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed6_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T03:43:47Z +hostname=n116-041-153 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed6_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n8_seed6_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 03:44:00.295000 1174471 torch/distributed/run.py:766] +W0726 03:44:00.295000 1174471 torch/distributed/run.py:766] ***************************************** +W0726 03:44:00.295000 1174471 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:44:00.295000 1174471 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +[rank2]:[W726 03:44:30.844019664 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:44:30.853113870 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:44:30.164477393 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:44:30.180444377 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:44:30.211570739 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:44:30.245518498 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:44:30.290543956 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +NCCL version 2.26.2+cuda12.2 +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:44:30.364791971 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n8_seed6_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed6_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:18:45 acc1_pos: 11.458 acc5_pos: 100.000 loss: 1.5448 (1.5448) acc1: 11.4583 (11.4583) acc5: 100.0000 (100.0000) time: 4.8093 data: 2.6961 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/234] eta: 0:01:56 acc1_pos: 15.885 acc5_pos: 100.000 loss: 1.6238 (1.5341) acc1: 15.8854 (20.8570) acc5: 100.0000 (100.0000) time: 0.5190 data: 0.2455 max mem: 3531 +Test: [ 20/234] eta: 0:01:11 acc1_pos: 86.198 acc5_pos: 100.000 loss: 1.3523 (1.2709) acc1: 31.2500 (35.4539) acc5: 100.0000 (100.0000) time: 0.1092 data: 0.0173 max mem: 3531 +Test: [ 30/234] eta: 0:01:00 acc1_pos: 59.896 acc5_pos: 100.000 loss: 0.3288 (0.9724) acc1: 86.1979 (51.9153) acc5: 100.0000 (100.0000) time: 0.1710 data: 0.0742 max mem: 3531 +Test: [ 40/234] eta: 0:00:58 acc1_pos: 92.708 acc5_pos: 100.000 loss: 0.2794 (0.8791) acc1: 90.1042 (56.3135) acc5: 100.0000 (100.0000) time: 0.2707 data: 0.1752 max mem: 3531 +Test: [ 50/234] eta: 0:00:54 acc1_pos: 94.010 acc5_pos: 100.000 loss: 0.2163 (0.7455) acc1: 93.4896 (63.9348) acc5: 100.0000 (100.0000) time: 0.2938 data: 0.2023 max mem: 3531 +Test: [ 60/234] eta: 0:00:48 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.1444 (0.6400) acc1: 96.6146 (69.8130) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2354 data: 0.1442 max mem: 3531 +Test: [ 70/234] eta: 0:00:44 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.0999 (0.6022) acc1: 100.0000 (71.7393) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.2208 data: 0.1294 max mem: 3531 +Test: [ 80/234] eta: 0:00:40 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.4701 (0.5696) acc1: 77.6042 (73.4021) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2008 data: 0.1101 max mem: 3531 +Test: [ 90/234] eta: 0:00:36 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.1197 (0.5223) acc1: 98.1771 (76.0417) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1782 data: 0.0875 max mem: 3531 +Test: [100/234] eta: 0:00:33 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.1207 (0.5030) acc1: 96.6146 (76.9312) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1932 data: 0.1009 max mem: 3531 +Test: [110/234] eta: 0:00:29 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.3032 (0.4815) acc1: 84.8958 (78.1227) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.1849 data: 0.0919 max mem: 3531 +Test: [120/234] eta: 0:00:27 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.2792 (0.4694) acc1: 87.2396 (78.6867) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.1844 data: 0.0871 max mem: 3531 +Test: [130/234] eta: 0:00:24 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.2347 (0.4539) acc1: 89.8438 (79.5265) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1949 data: 0.0970 max mem: 3531 +Test: [140/234] eta: 0:00:21 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.3408 (0.4651) acc1: 83.3333 (78.9524) acc5: 100.0000 (100.0000) acc1_neg: 39.583 acc5_neg: 100.000 time: 0.1763 data: 0.0817 max mem: 3531 +Test: [150/234] eta: 0:00:19 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.4041 (0.4629) acc1: 80.7292 (79.1011) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1899 data: 0.0958 max mem: 3531 +Test: [160/234] eta: 0:00:16 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.2614 (0.4473) acc1: 91.4062 (80.0029) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.2146 data: 0.1218 max mem: 3531 +Test: [170/234] eta: 0:00:14 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.2543 (0.4360) acc1: 91.6667 (80.6683) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1763 data: 0.0823 max mem: 3531 +Test: [180/234] eta: 0:00:11 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.2712 (0.4394) acc1: 88.8021 (80.3623) acc5: 100.0000 (100.0000) acc1_neg: 49.479 acc5_neg: 100.000 time: 0.1696 data: 0.0769 max mem: 3531 +Test: [190/234] eta: 0:00:09 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.4394 (0.4409) acc1: 79.6875 (80.2056) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.1935 data: 0.1025 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.4202 (0.4297) acc1: 79.9479 (80.7939) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1990 data: 0.1052 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.0854 (0.4135) acc1: 100.0000 (81.7030) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1999 data: 0.1066 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.0985 (0.3997) acc1: 100.0000 (82.5120) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1723 data: 0.0809 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.1732 (0.3955) acc1: 92.1875 (82.8226) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1524 data: 0.0609 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.2885 (0.3943) acc1: 90.3646 (82.8985) acc5: 100.0000 (100.0000) acc1_neg: 88.716 acc5_neg: 100.000 time: 0.1557 data: 0.0640 max mem: 3532 +Test: Total time: 0:00:49 (0.2109 s / it) +* Overall Acc@1 82.899 Acc@5 100.000 + target == real Acc@1 64.159 Acc@5 100.000 + target == fake Acc@1 88.180 Acc@5 100.000 + loss 0.394 + PR-AUC 0.9460 ROC-AUC 0.8546 EER 22.57% (thr=0.7534) +Accuracy of the network on the 89729 test images: 82.9% +[rank0]:[W726 03:45:32.328934056 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:45:53.840000 1175017 torch/distributed/run.py:766] +W0726 03:45:53.840000 1175017 torch/distributed/run.py:766] ***************************************** +W0726 03:45:53.840000 1175017 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:45:53.840000 1175017 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:46:21.924995151 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:46:21.230711379 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:46:22.552680682 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:46:22.682911914 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:46:22.788381860 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:46:22.929248439 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:46:22.944947326 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W726 03:46:22.945015146 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n8_seed6_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed6_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:09:39 acc1_pos: 22.917 acc5_pos: 100.000 loss: 1.3276 (1.3276) acc1: 22.9167 (22.9167) acc5: 100.0000 (100.0000) time: 6.2756 data: 4.3121 max mem: 3531 +Test: [ 10/2387] eta: 0:25:50 acc1_pos: 22.396 acc5_pos: 100.000 loss: 1.3241 (1.2183) acc1: 23.9583 (33.5938) acc5: 100.0000 (100.0000) time: 0.6522 data: 0.3923 max mem: 3531 +Test: [ 20/2387] eta: 0:15:13 acc1_pos: 10.938 acc5_pos: 100.000 loss: 1.1101 (1.2530) acc1: 25.7812 (31.3616) acc5: 100.0000 (100.0000) time: 0.0914 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:12:48 acc1_pos: 9.896 acc5_pos: 100.000 loss: 1.3894 (1.3071) acc1: 22.1354 (28.6206) acc5: 100.0000 (100.0000) time: 0.1464 data: 0.0524 max mem: 3531 +Test: [ 40/2387] eta: 0:12:12 acc1_pos: 13.021 acc5_pos: 100.000 loss: 1.4696 (1.3592) acc1: 17.7083 (25.5589) acc5: 100.0000 (100.0000) time: 0.2349 data: 0.1386 max mem: 3531 +Test: [ 50/2387] eta: 0:11:41 acc1_pos: 19.792 acc5_pos: 100.000 loss: 1.4738 (1.3767) acc1: 17.7083 (24.5609) acc5: 100.0000 (100.0000) time: 0.2599 data: 0.1636 max mem: 3531 +Test: [ 60/2387] eta: 0:11:57 acc1_pos: 32.292 acc5_pos: 100.000 loss: 1.4159 (1.3852) acc1: 23.9583 (24.4749) acc5: 100.0000 (100.0000) time: 0.2999 data: 0.2067 max mem: 3531 +Test: [ 70/2387] eta: 0:11:57 acc1_pos: 26.042 acc5_pos: 100.000 loss: 1.5041 (1.4243) acc1: 23.9583 (23.6649) acc5: 100.0000 (100.0000) time: 0.3349 data: 0.2401 max mem: 3531 +Test: [ 80/2387] eta: 0:11:35 acc1_pos: 5.729 acc5_pos: 100.000 loss: 1.7035 (1.4790) acc1: 12.5000 (21.8879) acc5: 100.0000 (100.0000) time: 0.2807 data: 0.1847 max mem: 3531 +Test: [ 90/2387] eta: 0:11:06 acc1_pos: 5.990 acc5_pos: 100.000 loss: 1.8952 (1.5226) acc1: 8.3333 (20.3840) acc5: 100.0000 (100.0000) time: 0.2202 data: 0.1268 max mem: 3531 +Test: [ 100/2387] eta: 0:10:56 acc1_pos: 3.906 acc5_pos: 100.000 loss: 1.9543 (1.5686) acc1: 5.7292 (18.8815) acc5: 100.0000 (100.0000) time: 0.2292 data: 0.1348 max mem: 3531 +Test: [ 110/2387] eta: 0:10:48 acc1_pos: 12.500 acc5_pos: 100.000 loss: 1.9579 (1.5968) acc1: 8.5938 (18.1635) acc5: 100.0000 (100.0000) time: 0.2596 data: 0.1660 max mem: 3531 +Test: [ 120/2387] eta: 0:10:39 acc1_pos: 9.115 acc5_pos: 100.000 loss: 1.9144 (1.6223) acc1: 9.3750 (17.5663) acc5: 100.0000 (100.0000) time: 0.2559 data: 0.1614 max mem: 3531 +Test: [ 130/2387] eta: 0:10:27 acc1_pos: 14.583 acc5_pos: 100.000 loss: 1.9554 (1.6499) acc1: 8.0729 (16.6905) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1437 max mem: 3531 +Test: [ 140/2387] eta: 0:10:13 acc1_pos: 4.688 acc5_pos: 100.000 loss: 2.0175 (1.6736) acc1: 4.6875 (15.8873) acc5: 100.0000 (100.0000) time: 0.2189 data: 0.1224 max mem: 3531 +Test: [ 150/2387] eta: 0:10:07 acc1_pos: 0.521 acc5_pos: 100.000 loss: 1.9652 (1.6932) acc1: 5.4688 (15.3232) acc5: 100.0000 (100.0000) time: 0.2304 data: 0.1336 max mem: 3531 +Test: [ 160/2387] eta: 0:09:50 acc1_pos: 30.137 acc5_pos: 100.000 loss: 1.9474 (1.6644) acc1: 5.9896 (17.0613) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2100 data: 0.1129 max mem: 3531 +Test: [ 170/2387] eta: 0:09:39 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1596 (1.5760) acc1: 96.3542 (21.7288) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1844 data: 0.0874 max mem: 3531 +Test: [ 180/2387] eta: 0:09:29 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1553 (1.4977) acc1: 96.6146 (25.8518) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1998 data: 0.1023 max mem: 3531 +Test: [ 190/2387] eta: 0:09:21 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1553 (1.4274) acc1: 96.3542 (29.5498) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2049 data: 0.1076 max mem: 3531 +Test: [ 200/2387] eta: 0:09:11 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1471 (1.3637) acc1: 97.1354 (32.9136) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2002 data: 0.1054 max mem: 3531 +Test: [ 210/2387] eta: 0:09:02 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1473 (1.3062) acc1: 97.1354 (35.9437) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1906 data: 0.0958 max mem: 3532 +Test: [ 220/2387] eta: 0:09:04 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1505 (1.2540) acc1: 96.8750 (38.6996) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2423 data: 0.1475 max mem: 3532 +Test: [ 230/2387] eta: 0:09:00 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1496 (1.2061) acc1: 96.6146 (41.2112) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2663 data: 0.1729 max mem: 3532 +Test: [ 240/2387] eta: 0:08:44 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1329 (1.1605) acc1: 97.3958 (43.5944) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1695 data: 0.0755 max mem: 3532 +Test: [ 250/2387] eta: 0:08:40 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0928 (1.1173) acc1: 99.4792 (45.8375) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1628 data: 0.0683 max mem: 3532 +Test: [ 260/2387] eta: 0:08:38 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0734 (1.0773) acc1: 100.0000 (47.9127) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2325 data: 0.1372 max mem: 3532 +Test: [ 270/2387] eta: 0:08:32 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0712 (1.0402) acc1: 100.0000 (49.8347) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2199 data: 0.1239 max mem: 3532 +Test: [ 280/2387] eta: 0:08:26 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0690 (1.0057) acc1: 100.0000 (51.6200) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2017 data: 0.1088 max mem: 3532 +Test: [ 290/2387] eta: 0:08:23 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0739 (0.9737) acc1: 100.0000 (53.2825) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2154 data: 0.1230 max mem: 3532 +Test: [ 300/2387] eta: 0:08:20 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0732 (0.9437) acc1: 100.0000 (54.8346) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2298 data: 0.1351 max mem: 3532 +Test: [ 310/2387] eta: 0:08:16 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0724 (0.9168) acc1: 100.0000 (56.2760) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2265 data: 0.1316 max mem: 3532 +Test: [ 320/2387] eta: 0:08:15 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1399 (0.8927) acc1: 100.0000 (57.6348) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2396 data: 0.1455 max mem: 3532 +Test: [ 330/2387] eta: 0:08:10 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1399 (0.8699) acc1: 100.0000 (58.9140) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2287 data: 0.1361 max mem: 3532 +Test: [ 340/2387] eta: 0:08:06 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1397 (0.8485) acc1: 100.0000 (60.1173) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2049 data: 0.1138 max mem: 3532 +Test: [ 350/2387] eta: 0:08:02 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1327 (0.8266) acc1: 100.0000 (61.2536) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2118 data: 0.1193 max mem: 3532 +Test: [ 360/2387] eta: 0:07:58 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0917 (0.8069) acc1: 100.0000 (62.2850) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2101 data: 0.1168 max mem: 3532 +Test: [ 370/2387] eta: 0:07:53 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1172 (0.7884) acc1: 98.4375 (63.2525) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1971 data: 0.1046 max mem: 3532 +Test: [ 380/2387] eta: 0:07:48 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1182 (0.7708) acc1: 98.1771 (64.1739) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1783 data: 0.0862 max mem: 3532 +Test: [ 390/2387] eta: 0:07:43 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1181 (0.7540) acc1: 98.4375 (65.0522) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1768 data: 0.0788 max mem: 3532 +Test: [ 400/2387] eta: 0:07:37 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1179 (0.7381) acc1: 98.4375 (65.8854) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1771 data: 0.0785 max mem: 3532 +Test: [ 410/2387] eta: 0:07:32 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1174 (0.7230) acc1: 98.4375 (66.6774) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1729 data: 0.0791 max mem: 3532 +Test: [ 420/2387] eta: 0:07:27 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1143 (0.7086) acc1: 98.4375 (67.4281) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1680 data: 0.0750 max mem: 3532 +Test: [ 430/2387] eta: 0:07:22 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1109 (0.6945) acc1: 98.6979 (68.1627) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1648 data: 0.0733 max mem: 3532 +Test: [ 440/2387] eta: 0:07:17 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0866 (0.6806) acc1: 99.4792 (68.8793) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1717 data: 0.0786 max mem: 3533 +Test: [ 450/2387] eta: 0:07:14 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0741 (0.6671) acc1: 100.0000 (69.5694) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0845 max mem: 3533 +Test: [ 460/2387] eta: 0:07:09 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0740 (0.6543) acc1: 100.0000 (70.2295) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1737 data: 0.0789 max mem: 3533 +Test: [ 470/2387] eta: 0:07:05 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0740 (0.6420) acc1: 100.0000 (70.8615) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1775 data: 0.0850 max mem: 3533 +Test: [ 480/2387] eta: 0:07:02 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0743 (0.6302) acc1: 100.0000 (71.4673) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1952 data: 0.1038 max mem: 3533 +Test: [ 490/2387] eta: 0:06:58 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0745 (0.6188) acc1: 100.0000 (72.0484) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1873 data: 0.0926 max mem: 3533 +Test: [ 500/2387] eta: 0:06:55 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0743 (0.6080) acc1: 100.0000 (72.6064) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0952 max mem: 3533 +Test: [ 510/2387] eta: 0:06:52 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0741 (0.5975) acc1: 100.0000 (73.1424) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1975 data: 0.1063 max mem: 3533 +Test: [ 520/2387] eta: 0:06:49 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0779 (0.5876) acc1: 100.0000 (73.6579) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1995 data: 0.1080 max mem: 3533 +Test: [ 530/2387] eta: 0:06:47 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0853 (0.5782) acc1: 100.0000 (74.1540) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2129 data: 0.1192 max mem: 3533 +Test: [ 540/2387] eta: 0:06:44 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0858 (0.5691) acc1: 100.0000 (74.6318) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1997 data: 0.1066 max mem: 3533 +Test: [ 550/2387] eta: 0:06:41 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0859 (0.5603) acc1: 100.0000 (75.0917) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1911 data: 0.1009 max mem: 3533 +Test: [ 560/2387] eta: 0:06:38 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0858 (0.5518) acc1: 100.0000 (75.5357) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2059 data: 0.1142 max mem: 3533 +Test: [ 570/2387] eta: 0:06:36 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0858 (0.5437) acc1: 100.0000 (75.9641) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2050 data: 0.1135 max mem: 3533 +Test: [ 580/2387] eta: 0:06:33 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0859 (0.5358) acc1: 100.0000 (76.3778) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2003 data: 0.1095 max mem: 3533 +Test: [ 590/2387] eta: 0:06:31 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0858 (0.5282) acc1: 100.0000 (76.7775) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2081 data: 0.1163 max mem: 3533 +Test: [ 600/2387] eta: 0:06:29 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0811 (0.5207) acc1: 100.0000 (77.1635) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2238 data: 0.1324 max mem: 3533 +Test: [ 610/2387] eta: 0:06:26 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0789 (0.5135) acc1: 100.0000 (77.5355) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2148 data: 0.1238 max mem: 3533 +Test: [ 620/2387] eta: 0:06:24 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0805 (0.5065) acc1: 100.0000 (77.8965) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2004 data: 0.1088 max mem: 3533 +Test: [ 630/2387] eta: 0:06:22 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0808 (0.4998) acc1: 100.0000 (78.2439) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2149 data: 0.1231 max mem: 3533 +Test: [ 640/2387] eta: 0:06:19 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0819 (0.4933) acc1: 99.7396 (78.5808) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2135 data: 0.1199 max mem: 3533 +Test: [ 650/2387] eta: 0:06:16 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0818 (0.4870) acc1: 100.0000 (78.9083) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2022 data: 0.1080 max mem: 3533 +Test: [ 660/2387] eta: 0:06:13 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0809 (0.4809) acc1: 99.7396 (79.2246) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1932 data: 0.1007 max mem: 3533 +Test: [ 670/2387] eta: 0:06:10 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0809 (0.4749) acc1: 100.0000 (79.5330) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1840 data: 0.0928 max mem: 3534 +Test: [ 680/2387] eta: 0:06:07 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0795 (0.4690) acc1: 100.0000 (79.8328) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1833 data: 0.0919 max mem: 3534 +Test: [ 690/2387] eta: 0:06:05 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0811 (0.4675) acc1: 99.7396 (79.8978) acc5: 100.0000 (100.0000) acc1_neg: 75.521 acc5_neg: 100.000 time: 0.1838 data: 0.0915 max mem: 3534 +Test: [ 700/2387] eta: 0:06:03 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.4717 (0.4676) acc1: 78.6458 (79.8755) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.2036 data: 0.1110 max mem: 3534 +Test: [ 710/2387] eta: 0:05:59 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.4756 (0.4678) acc1: 77.8646 (79.8545) acc5: 100.0000 (100.0000) acc1_neg: 77.344 acc5_neg: 100.000 time: 0.1875 data: 0.0950 max mem: 3534 +Test: [ 720/2387] eta: 0:05:56 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.4760 (0.4679) acc1: 78.6458 (79.8410) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.1716 data: 0.0756 max mem: 3534 +Test: [ 730/2387] eta: 0:05:54 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.4865 (0.4684) acc1: 78.3854 (79.8115) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.1920 data: 0.0957 max mem: 3534 +Test: [ 740/2387] eta: 0:05:51 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.4865 (0.4686) acc1: 78.3854 (79.7943) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.1878 data: 0.0944 max mem: 3534 +Test: [ 750/2387] eta: 0:05:48 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.4718 (0.4686) acc1: 79.1667 (79.7860) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.1860 data: 0.0909 max mem: 3534 +Test: [ 760/2387] eta: 0:05:47 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.4707 (0.4678) acc1: 79.1667 (79.8145) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.2149 data: 0.1207 max mem: 3534 +Test: [ 770/2387] eta: 0:05:44 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.2977 (0.4646) acc1: 87.5000 (79.9695) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2190 data: 0.1241 max mem: 3534 +Test: [ 780/2387] eta: 0:05:40 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1914 (0.4612) acc1: 91.4062 (80.1330) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1499 data: 0.0518 max mem: 3534 +Test: [ 790/2387] eta: 0:05:37 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1595 (0.4573) acc1: 95.8333 (80.3384) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1354 data: 0.0399 max mem: 3534 +Test: [ 800/2387] eta: 0:05:34 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1458 (0.4534) acc1: 96.8750 (80.5452) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1663 data: 0.0744 max mem: 3534 +Test: [ 810/2387] eta: 0:05:31 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1308 (0.4495) acc1: 97.3958 (80.7510) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1744 data: 0.0831 max mem: 3534 +Test: [ 820/2387] eta: 0:05:28 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1306 (0.4458) acc1: 97.6562 (80.9474) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1785 data: 0.0875 max mem: 3534 +Test: [ 830/2387] eta: 0:05:25 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1404 (0.4422) acc1: 96.8750 (81.1372) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1708 data: 0.0770 max mem: 3534 +Test: [ 840/2387] eta: 0:05:23 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1144 (0.4382) acc1: 98.9583 (81.3556) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1749 data: 0.0766 max mem: 3534 +Test: [ 850/2387] eta: 0:05:21 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1037 (0.4345) acc1: 99.4792 (81.5643) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1900 data: 0.0947 max mem: 3534 +Test: [ 860/2387] eta: 0:05:18 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1015 (0.4306) acc1: 99.7396 (81.7763) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2027 data: 0.1123 max mem: 3534 +Test: [ 870/2387] eta: 0:05:16 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0966 (0.4269) acc1: 99.7396 (81.9724) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2040 data: 0.1133 max mem: 3534 +Test: [ 880/2387] eta: 0:05:14 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0971 (0.4232) acc1: 99.4792 (82.1666) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2102 data: 0.1164 max mem: 3534 +Test: [ 890/2387] eta: 0:05:12 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0917 (0.4195) acc1: 99.7396 (82.3606) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2010 data: 0.1049 max mem: 3534 +Test: [ 900/2387] eta: 0:05:10 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0951 (0.4184) acc1: 98.1771 (82.4093) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.2037 data: 0.1086 max mem: 3535 +Test: [ 910/2387] eta: 0:05:08 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.3737 (0.4179) acc1: 84.3750 (82.4355) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.2135 data: 0.1139 max mem: 3535 +Test: [ 920/2387] eta: 0:05:05 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.3767 (0.4174) acc1: 84.3750 (82.4531) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1949 data: 0.0974 max mem: 3535 +Test: [ 930/2387] eta: 0:05:03 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.3768 (0.4170) acc1: 84.3750 (82.4777) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1866 data: 0.0947 max mem: 3535 +Test: [ 940/2387] eta: 0:05:01 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.3781 (0.4165) acc1: 84.3750 (82.4973) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.2032 data: 0.1106 max mem: 3535 +Test: [ 950/2387] eta: 0:04:59 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.3742 (0.4160) acc1: 84.1146 (82.5302) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.2115 data: 0.1187 max mem: 3535 +Test: [ 960/2387] eta: 0:04:56 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.3570 (0.4155) acc1: 85.4167 (82.5540) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1871 data: 0.0941 max mem: 3535 +Test: [ 970/2387] eta: 0:04:54 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.3640 (0.4151) acc1: 84.1146 (82.5709) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.1816 data: 0.0882 max mem: 3535 +Test: [ 980/2387] eta: 0:04:51 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.3649 (0.4137) acc1: 84.6354 (82.6407) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.1874 data: 0.0927 max mem: 3535 +Test: [ 990/2387] eta: 0:04:49 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1617 (0.4119) acc1: 95.0521 (82.7413) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1818 data: 0.0888 max mem: 3535 +Test: [1000/2387] eta: 0:04:46 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1207 (0.4091) acc1: 98.4375 (82.8984) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1756 data: 0.0844 max mem: 3535 +Test: [1010/2387] eta: 0:04:44 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1236 (0.4062) acc1: 98.6979 (83.0515) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1785 data: 0.0870 max mem: 3535 +Test: [1020/2387] eta: 0:04:42 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1083 (0.4035) acc1: 98.6979 (83.1928) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1809 data: 0.0884 max mem: 3535 +Test: [1030/2387] eta: 0:04:39 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1089 (0.4008) acc1: 98.9583 (83.3404) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1794 data: 0.0862 max mem: 3535 +Test: [1040/2387] eta: 0:04:37 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1004 (0.3978) acc1: 99.4792 (83.4954) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1781 data: 0.0861 max mem: 3535 +Test: [1050/2387] eta: 0:04:34 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0947 (0.3949) acc1: 99.4792 (83.6463) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1801 data: 0.0891 max mem: 3535 +Test: [1060/2387] eta: 0:04:32 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0915 (0.3921) acc1: 99.7396 (83.7945) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1752 data: 0.0838 max mem: 3535 +Test: [1070/2387] eta: 0:04:29 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0954 (0.3894) acc1: 99.7396 (83.9383) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1732 data: 0.0815 max mem: 3535 +Test: [1080/2387] eta: 0:04:27 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1028 (0.3869) acc1: 98.9583 (84.0703) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1897 data: 0.0979 max mem: 3535 +Test: [1090/2387] eta: 0:04:25 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1485 (0.3855) acc1: 96.0938 (84.1246) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.1883 data: 0.0972 max mem: 3535 +Test: [1100/2387] eta: 0:04:23 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.2061 (0.3842) acc1: 91.9271 (84.1834) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1774 data: 0.0854 max mem: 3535 +Test: [1110/2387] eta: 0:04:20 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.2186 (0.3829) acc1: 92.1875 (84.2419) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1739 data: 0.0821 max mem: 3535 +Test: [1120/2387] eta: 0:04:18 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.2294 (0.3815) acc1: 90.6250 (84.3048) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1995 data: 0.1089 max mem: 3536 +Test: [1130/2387] eta: 0:04:16 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1984 (0.3799) acc1: 91.4062 (84.3925) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1900 data: 0.0985 max mem: 3536 +Test: [1140/2387] eta: 0:04:14 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1429 (0.3776) acc1: 97.9167 (84.5106) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1855 data: 0.0911 max mem: 3536 +Test: [1150/2387] eta: 0:04:12 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0996 (0.3752) acc1: 99.2188 (84.6399) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2247 data: 0.1311 max mem: 3536 +Test: [1160/2387] eta: 0:04:09 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0836 (0.3727) acc1: 100.0000 (84.7716) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1725 data: 0.0789 max mem: 3536 +Test: [1170/2387] eta: 0:04:07 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0800 (0.3702) acc1: 100.0000 (84.9003) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1477 data: 0.0526 max mem: 3536 +Test: [1180/2387] eta: 0:04:05 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0824 (0.3678) acc1: 100.0000 (85.0262) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1852 data: 0.0921 max mem: 3536 +Test: [1190/2387] eta: 0:04:02 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0819 (0.3654) acc1: 100.0000 (85.1517) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1823 data: 0.0883 max mem: 3536 +Test: [1200/2387] eta: 0:04:00 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0751 (0.3629) acc1: 100.0000 (85.2746) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1852 data: 0.0925 max mem: 3536 +Test: [1210/2387] eta: 0:03:58 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0731 (0.3606) acc1: 100.0000 (85.3952) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1902 data: 0.0991 max mem: 3536 +Test: [1220/2387] eta: 0:03:56 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0943 (0.3585) acc1: 99.7396 (85.5033) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1892 data: 0.0973 max mem: 3536 +Test: [1230/2387] eta: 0:03:54 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1045 (0.3565) acc1: 98.9583 (85.6124) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1958 data: 0.1040 max mem: 3536 +Test: [1240/2387] eta: 0:03:52 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1017 (0.3545) acc1: 98.6979 (85.7195) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2055 data: 0.1147 max mem: 3536 +Test: [1250/2387] eta: 0:03:50 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0962 (0.3524) acc1: 99.7396 (85.8313) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2035 data: 0.1130 max mem: 3536 +Test: [1260/2387] eta: 0:03:48 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0924 (0.3503) acc1: 99.7396 (85.9389) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2040 data: 0.1130 max mem: 3536 +Test: [1270/2387] eta: 0:03:46 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0951 (0.3483) acc1: 99.7396 (86.0440) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1947 data: 0.1015 max mem: 3536 +Test: [1280/2387] eta: 0:03:44 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0805 (0.3462) acc1: 99.7396 (86.1510) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1968 data: 0.1024 max mem: 3536 +Test: [1290/2387] eta: 0:03:42 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0834 (0.3442) acc1: 99.4792 (86.2512) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2101 data: 0.1159 max mem: 3536 +Test: [1300/2387] eta: 0:03:40 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0840 (0.3422) acc1: 98.9583 (86.3510) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2047 data: 0.1121 max mem: 3536 +Test: [1310/2387] eta: 0:03:38 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0864 (0.3403) acc1: 99.2188 (86.4488) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2151 data: 0.1225 max mem: 3536 +Test: [1320/2387] eta: 0:03:36 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0924 (0.3384) acc1: 99.4792 (86.5482) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2122 data: 0.1182 max mem: 3536 +Test: [1330/2387] eta: 0:03:34 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0914 (0.3367) acc1: 99.4792 (86.6379) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1980 data: 0.1062 max mem: 3536 +Test: [1340/2387] eta: 0:03:32 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1174 (0.3353) acc1: 97.9167 (86.7061) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2009 data: 0.1104 max mem: 3536 +Test: [1350/2387] eta: 0:03:30 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0636 (0.3333) acc1: 100.0000 (86.8045) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2026 data: 0.1106 max mem: 3537 +Test: [1360/2387] eta: 0:03:28 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0603 (0.3313) acc1: 100.0000 (86.9015) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2008 data: 0.1087 max mem: 3537 +Test: [1370/2387] eta: 0:03:26 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0608 (0.3293) acc1: 100.0000 (86.9968) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1979 data: 0.1073 max mem: 3537 +Test: [1380/2387] eta: 0:03:24 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0602 (0.3273) acc1: 100.0000 (87.0910) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2040 data: 0.1135 max mem: 3537 +Test: [1390/2387] eta: 0:03:22 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0605 (0.3254) acc1: 100.0000 (87.1838) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2056 data: 0.1146 max mem: 3537 +Test: [1400/2387] eta: 0:03:19 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0590 (0.3235) acc1: 100.0000 (87.2753) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1938 data: 0.1026 max mem: 3537 +Test: [1410/2387] eta: 0:03:17 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0590 (0.3224) acc1: 100.0000 (87.3278) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1857 data: 0.0938 max mem: 3537 +Test: [1420/2387] eta: 0:03:15 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1964 (0.3218) acc1: 94.2708 (87.3616) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1923 data: 0.1004 max mem: 3537 +Test: [1430/2387] eta: 0:03:13 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1978 (0.3211) acc1: 91.9271 (87.3948) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1936 data: 0.1028 max mem: 3537 +Test: [1440/2387] eta: 0:03:11 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1646 (0.3199) acc1: 95.5729 (87.4621) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1892 data: 0.0977 max mem: 3537 +Test: [1450/2387] eta: 0:03:09 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1421 (0.3185) acc1: 98.1771 (87.5336) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1853 data: 0.0939 max mem: 3537 +Test: [1460/2387] eta: 0:03:07 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0973 (0.3170) acc1: 99.2188 (87.6127) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1805 data: 0.0894 max mem: 3537 +Test: [1470/2387] eta: 0:03:05 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0893 (0.3155) acc1: 99.7396 (87.6923) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1791 data: 0.0871 max mem: 3537 +Test: [1480/2387] eta: 0:03:02 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0881 (0.3140) acc1: 99.7396 (87.7708) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1836 data: 0.0920 max mem: 3537 +Test: [1490/2387] eta: 0:03:00 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0837 (0.3125) acc1: 99.4792 (87.8477) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1901 data: 0.0976 max mem: 3537 +Test: [1500/2387] eta: 0:02:58 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0848 (0.3110) acc1: 99.7396 (87.9252) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1900 data: 0.0976 max mem: 3537 +Test: [1510/2387] eta: 0:02:56 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0758 (0.3094) acc1: 100.0000 (88.0026) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1842 data: 0.0933 max mem: 3537 +Test: [1520/2387] eta: 0:02:54 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0758 (0.3080) acc1: 100.0000 (88.0796) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1738 data: 0.0832 max mem: 3537 +Test: [1530/2387] eta: 0:02:52 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0961 (0.3066) acc1: 99.7396 (88.1554) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1798 data: 0.0890 max mem: 3537 +Test: [1540/2387] eta: 0:02:50 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0911 (0.3052) acc1: 99.7396 (88.2309) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1822 data: 0.0912 max mem: 3537 +Test: [1550/2387] eta: 0:02:48 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0891 (0.3039) acc1: 99.7396 (88.3046) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1825 data: 0.0920 max mem: 3537 +Test: [1560/2387] eta: 0:02:46 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0875 (0.3025) acc1: 99.7396 (88.3775) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1862 data: 0.0949 max mem: 3537 +Test: [1570/2387] eta: 0:02:43 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0849 (0.3011) acc1: 99.7396 (88.4492) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1794 data: 0.0871 max mem: 3537 +Test: [1580/2387] eta: 0:02:41 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0860 (0.3003) acc1: 99.7396 (88.4843) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.1766 data: 0.0852 max mem: 3538 +Test: [1590/2387] eta: 0:02:39 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.2224 (0.2999) acc1: 91.6667 (88.4975) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1781 data: 0.0879 max mem: 3538 +Test: [1600/2387] eta: 0:02:37 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.2173 (0.2991) acc1: 92.7083 (88.5422) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1862 data: 0.0947 max mem: 3538 +Test: [1610/2387] eta: 0:02:35 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1837 (0.2984) acc1: 95.0521 (88.5819) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1796 data: 0.0877 max mem: 3538 +Test: [1620/2387] eta: 0:02:33 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1843 (0.2977) acc1: 95.0521 (88.6212) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1759 data: 0.0849 max mem: 3538 +Test: [1630/2387] eta: 0:02:31 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1695 (0.2971) acc1: 94.2708 (88.6483) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1824 data: 0.0917 max mem: 3538 +Test: [1640/2387] eta: 0:02:29 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1628 (0.2963) acc1: 97.1354 (88.6942) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1756 data: 0.0847 max mem: 3538 +Test: [1650/2387] eta: 0:02:27 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1557 (0.2954) acc1: 96.8750 (88.7409) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1724 data: 0.0796 max mem: 3538 +Test: [1660/2387] eta: 0:02:24 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1519 (0.2947) acc1: 96.3542 (88.7772) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1787 data: 0.0848 max mem: 3538 +Test: [1670/2387] eta: 0:02:22 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1570 (0.2939) acc1: 95.8333 (88.8203) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1823 data: 0.0896 max mem: 3538 +Test: [1680/2387] eta: 0:02:20 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1515 (0.2932) acc1: 96.3542 (88.8628) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1790 data: 0.0869 max mem: 3538 +Test: [1690/2387] eta: 0:02:18 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1486 (0.2924) acc1: 96.6146 (88.9074) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1825 data: 0.0901 max mem: 3538 +Test: [1700/2387] eta: 0:02:16 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1472 (0.2915) acc1: 96.6146 (88.9576) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1853 data: 0.0928 max mem: 3538 +Test: [1710/2387] eta: 0:02:14 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1351 (0.2907) acc1: 97.3958 (89.0022) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1872 data: 0.0959 max mem: 3538 +Test: [1720/2387] eta: 0:02:12 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1408 (0.2899) acc1: 96.8750 (89.0477) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1957 data: 0.1047 max mem: 3538 +Test: [1730/2387] eta: 0:02:10 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1408 (0.2890) acc1: 96.8750 (89.0936) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2007 data: 0.1089 max mem: 3538 +Test: [1740/2387] eta: 0:02:08 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1325 (0.2881) acc1: 97.3958 (89.1424) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2043 data: 0.1099 max mem: 3538 +Test: [1750/2387] eta: 0:02:06 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1257 (0.2873) acc1: 97.6562 (89.1909) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2018 data: 0.1075 max mem: 3538 +Test: [1760/2387] eta: 0:02:04 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1243 (0.2864) acc1: 97.9167 (89.2385) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1880 data: 0.0948 max mem: 3538 +Test: [1770/2387] eta: 0:02:02 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1276 (0.2856) acc1: 97.1354 (89.2812) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1909 data: 0.0983 max mem: 3538 +Test: [1780/2387] eta: 0:02:00 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1453 (0.2849) acc1: 96.3542 (89.3178) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1801 data: 0.0888 max mem: 3538 +Test: [1790/2387] eta: 0:01:58 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1403 (0.2840) acc1: 97.1354 (89.3651) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1790 data: 0.0843 max mem: 3538 +Test: [1800/2387] eta: 0:01:56 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1355 (0.2833) acc1: 97.3958 (89.4056) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2125 data: 0.1181 max mem: 3538 +Test: [1810/2387] eta: 0:01:54 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1227 (0.2823) acc1: 98.1771 (89.4569) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2022 data: 0.1074 max mem: 3539 +Test: [1820/2387] eta: 0:01:52 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1115 (0.2814) acc1: 98.6979 (89.5065) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1870 data: 0.0901 max mem: 3539 +Test: [1830/2387] eta: 0:01:50 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0954 (0.2804) acc1: 98.9583 (89.5573) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1928 data: 0.0996 max mem: 3539 +Test: [1840/2387] eta: 0:01:48 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0745 (0.2793) acc1: 100.0000 (89.6140) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2011 data: 0.1108 max mem: 3539 +Test: [1850/2387] eta: 0:01:46 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0745 (0.2782) acc1: 100.0000 (89.6701) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1996 data: 0.1090 max mem: 3539 +Test: [1860/2387] eta: 0:01:44 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0741 (0.2771) acc1: 100.0000 (89.7256) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1886 data: 0.0958 max mem: 3539 +Test: [1870/2387] eta: 0:01:42 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0740 (0.2760) acc1: 100.0000 (89.7806) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1866 data: 0.0927 max mem: 3539 +Test: [1880/2387] eta: 0:01:40 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0744 (0.2749) acc1: 100.0000 (89.8349) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0957 max mem: 3539 +Test: [1890/2387] eta: 0:01:38 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0744 (0.2739) acc1: 100.0000 (89.8886) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1953 data: 0.1020 max mem: 3539 +Test: [1900/2387] eta: 0:01:36 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0743 (0.2728) acc1: 100.0000 (89.9418) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1978 data: 0.1062 max mem: 3539 +Test: [1910/2387] eta: 0:01:34 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0732 (0.2718) acc1: 100.0000 (89.9915) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1980 data: 0.1064 max mem: 3539 +Test: [1920/2387] eta: 0:01:32 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1023 (0.2716) acc1: 98.9583 (89.9941) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1894 data: 0.0960 max mem: 3539 +Test: [1930/2387] eta: 0:01:30 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1758 (0.2710) acc1: 95.8333 (90.0264) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1894 data: 0.0943 max mem: 3539 +Test: [1940/2387] eta: 0:01:28 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1512 (0.2705) acc1: 97.1354 (90.0599) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1901 data: 0.0976 max mem: 3539 +Test: [1950/2387] eta: 0:01:26 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1553 (0.2702) acc1: 96.8750 (90.0736) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1931 data: 0.1027 max mem: 3539 +Test: [1960/2387] eta: 0:01:24 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1741 (0.2698) acc1: 95.5729 (90.1014) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1833 data: 0.0927 max mem: 3539 +Test: [1970/2387] eta: 0:01:22 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1773 (0.2696) acc1: 94.0104 (90.0998) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.1669 data: 0.0748 max mem: 3539 +Test: [1980/2387] eta: 0:01:20 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1543 (0.2689) acc1: 97.1354 (90.1402) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1705 data: 0.0769 max mem: 3539 +Test: [1990/2387] eta: 0:01:18 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0892 (0.2680) acc1: 99.4792 (90.1880) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1711 data: 0.0784 max mem: 3539 +Test: [2000/2387] eta: 0:01:16 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0867 (0.2671) acc1: 99.7396 (90.2357) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1800 data: 0.0882 max mem: 3539 +Test: [2010/2387] eta: 0:01:14 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0867 (0.2662) acc1: 99.7396 (90.2827) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1784 data: 0.0871 max mem: 3539 +Test: [2020/2387] eta: 0:01:12 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0874 (0.2653) acc1: 99.7396 (90.3302) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1736 data: 0.0831 max mem: 3539 +Test: [2030/2387] eta: 0:01:10 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0874 (0.2644) acc1: 99.7396 (90.3761) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1784 data: 0.0871 max mem: 3540 +Test: [2040/2387] eta: 0:01:08 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0874 (0.2636) acc1: 99.7396 (90.4223) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1854 data: 0.0940 max mem: 3540 +Test: [2050/2387] eta: 0:01:06 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0852 (0.2627) acc1: 99.7396 (90.4681) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1873 data: 0.0959 max mem: 3540 +Test: [2060/2387] eta: 0:01:04 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0786 (0.2618) acc1: 100.0000 (90.5137) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1793 data: 0.0874 max mem: 3540 +Test: [2070/2387] eta: 0:01:02 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0763 (0.2609) acc1: 100.0000 (90.5589) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1808 data: 0.0887 max mem: 3540 +Test: [2080/2387] eta: 0:01:00 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0768 (0.2600) acc1: 100.0000 (90.6035) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1862 data: 0.0943 max mem: 3540 +Test: [2090/2387] eta: 0:00:58 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0768 (0.2591) acc1: 100.0000 (90.6482) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1798 data: 0.0883 max mem: 3540 +Test: [2100/2387] eta: 0:00:56 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0764 (0.2583) acc1: 100.0000 (90.6922) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1850 data: 0.0941 max mem: 3540 +Test: [2110/2387] eta: 0:00:54 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0769 (0.2574) acc1: 100.0000 (90.7359) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1763 data: 0.0852 max mem: 3540 +Test: [2120/2387] eta: 0:00:52 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0791 (0.2572) acc1: 99.7396 (90.7437) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.1687 data: 0.0773 max mem: 3540 +Test: [2130/2387] eta: 0:00:50 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1994 (0.2570) acc1: 93.2292 (90.7558) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1829 data: 0.0896 max mem: 3540 +Test: [2140/2387] eta: 0:00:48 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1875 (0.2565) acc1: 95.0521 (90.7800) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1721 data: 0.0789 max mem: 3540 +Test: [2150/2387] eta: 0:00:46 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1744 (0.2562) acc1: 96.3542 (90.7995) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1659 data: 0.0732 max mem: 3540 +Test: [2160/2387] eta: 0:00:44 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1432 (0.2557) acc1: 96.8750 (90.8313) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1820 data: 0.0876 max mem: 3540 +Test: [2170/2387] eta: 0:00:42 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1333 (0.2553) acc1: 97.1354 (90.8528) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1839 data: 0.0911 max mem: 3540 +Test: [2180/2387] eta: 0:00:40 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1604 (0.2563) acc1: 96.8750 (90.7852) acc5: 100.0000 (100.0000) acc1_neg: 46.615 acc5_neg: 100.000 time: 0.1759 data: 0.0844 max mem: 3540 +Test: [2190/2387] eta: 0:00:38 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1584 (0.2562) acc1: 96.8750 (90.7851) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1887 data: 0.0942 max mem: 3540 +Test: [2200/2387] eta: 0:00:36 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1138 (0.2555) acc1: 98.4375 (90.8212) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2102 data: 0.1161 max mem: 3540 +Test: [2210/2387] eta: 0:00:34 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1098 (0.2549) acc1: 99.2188 (90.8539) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1962 data: 0.1056 max mem: 3540 +Test: [2220/2387] eta: 0:00:32 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1098 (0.2543) acc1: 98.6979 (90.8882) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1786 data: 0.0861 max mem: 3540 +Test: [2230/2387] eta: 0:00:30 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1046 (0.2537) acc1: 99.2188 (90.9222) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1835 data: 0.0901 max mem: 3540 +Test: [2240/2387] eta: 0:00:28 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0986 (0.2530) acc1: 99.4792 (90.9579) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1818 data: 0.0905 max mem: 3540 +Test: [2250/2387] eta: 0:00:26 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0927 (0.2523) acc1: 99.7396 (90.9949) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1857 data: 0.0955 max mem: 3540 +Test: [2260/2387] eta: 0:00:24 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0818 (0.2515) acc1: 100.0000 (91.0342) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1930 data: 0.1019 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0950 (0.2509) acc1: 99.7396 (91.0708) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1874 data: 0.0948 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0978 (0.2502) acc1: 99.4792 (91.1068) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1752 data: 0.0829 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0880 (0.2495) acc1: 100.0000 (91.1429) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1587 data: 0.0664 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0880 (0.2488) acc1: 99.7396 (91.1784) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1419 data: 0.0501 max mem: 3541 +Test: [2310/2387] eta: 0:00:15 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.0890 (0.2482) acc1: 99.7396 (91.2134) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1328 data: 0.0425 max mem: 3541 +Test: [2320/2387] eta: 0:00:13 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1084 (0.2476) acc1: 99.2188 (91.2428) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1266 data: 0.0365 max mem: 3541 +Test: [2330/2387] eta: 0:00:11 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1496 (0.2474) acc1: 96.8750 (91.2558) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1229 data: 0.0333 max mem: 3541 +Test: [2340/2387] eta: 0:00:09 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.2067 (0.2472) acc1: 94.0104 (91.2623) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1202 data: 0.0305 max mem: 3541 +Test: [2350/2387] eta: 0:00:07 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1682 (0.2470) acc1: 96.0938 (91.2790) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1205 data: 0.0309 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1364 (0.2466) acc1: 96.8750 (91.2942) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1191 data: 0.0294 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1436 (0.2463) acc1: 96.8750 (91.3112) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1176 data: 0.0279 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1436 (0.2459) acc1: 97.1354 (91.3335) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1159 data: 0.0262 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 30.137 acc5_pos: 100.000 loss: 0.1273 (0.2456) acc1: 97.1354 (91.3494) acc5: 100.0000 (100.0000) acc1_neg: 97.674 acc5_neg: 100.000 time: 0.1115 data: 0.0244 max mem: 3541 +Test: Total time: 0:07:40 (0.1928 s / it) +* Overall Acc@1 91.349 Acc@5 100.000 + target == real Acc@1 14.951 Acc@5 100.000 + target == fake Acc@1 96.714 Acc@5 100.000 + loss 0.246 + PR-AUC 0.9812 ROC-AUC 0.7991 EER 26.39% (thr=0.9019) +Accuracy of the network on the 916267 test images: 91.3% +[rank0]:[W726 03:54:19.849330485 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:54:39.180000 1175651 torch/distributed/run.py:766] +W0726 03:54:39.180000 1175651 torch/distributed/run.py:766] ***************************************** +W0726 03:54:39.180000 1175651 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:54:39.180000 1175651 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:55:09.618197152 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:55:09.896292622 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:55:09.007495885 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:55:09.284203062 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:55:09.331505968 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:55:09.418846262 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:55:09.453032301 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank1]:[W726 03:55:09.453100936 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n8_seed6_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed6_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:07:58 acc1_pos: 22.917 acc5_pos: 100.000 loss: 1.3276 (1.3276) acc1: 22.9167 (22.9167) acc5: 100.0000 (100.0000) time: 7.8891 data: 6.1344 max mem: 3531 +Test: [ 10/517] eta: 0:06:47 acc1_pos: 22.396 acc5_pos: 100.000 loss: 1.3241 (1.2183) acc1: 23.9583 (33.5938) acc5: 100.0000 (100.0000) time: 0.8044 data: 0.5581 max mem: 3531 +Test: [ 20/517] eta: 0:03:53 acc1_pos: 10.938 acc5_pos: 100.000 loss: 1.1101 (1.2530) acc1: 25.7812 (31.3616) acc5: 100.0000 (100.0000) time: 0.0979 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:03:26 acc1_pos: 9.896 acc5_pos: 100.000 loss: 1.3894 (1.3071) acc1: 22.1354 (28.6206) acc5: 100.0000 (100.0000) time: 0.2161 data: 0.1190 max mem: 3531 +Test: [ 40/517] eta: 0:02:59 acc1_pos: 13.021 acc5_pos: 100.000 loss: 1.4696 (1.3592) acc1: 17.7083 (25.5589) acc5: 100.0000 (100.0000) time: 0.2799 data: 0.1858 max mem: 3531 +Test: [ 50/517] eta: 0:02:45 acc1_pos: 3.670 acc5_pos: 100.000 loss: 1.4696 (1.3204) acc1: 17.7083 (27.7471) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.2437 data: 0.1489 max mem: 3531 +Test: [ 60/517] eta: 0:02:35 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.2998 (1.1448) acc1: 89.8438 (38.6100) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.2649 data: 0.1701 max mem: 3531 +Test: [ 70/517] eta: 0:02:28 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.2377 (1.0194) acc1: 94.7917 (46.5045) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2751 data: 0.1825 max mem: 3531 +Test: [ 80/517] eta: 0:02:22 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.2421 (0.9259) acc1: 96.0938 (52.4724) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2799 data: 0.1875 max mem: 3531 +Test: [ 90/517] eta: 0:02:16 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.2899 (0.8583) acc1: 92.4479 (56.6936) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.2801 data: 0.1866 max mem: 3531 +Test: [100/517] eta: 0:02:13 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.3116 (0.8085) acc1: 89.0625 (59.5065) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.3002 data: 0.2082 max mem: 3531 +Test: [110/517] eta: 0:02:08 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.2642 (0.7545) acc1: 92.1875 (62.8590) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2961 data: 0.2010 max mem: 3531 +Test: [120/517] eta: 0:02:04 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.2723 (0.7238) acc1: 89.5833 (64.6242) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.2813 data: 0.1841 max mem: 3531 +Test: [130/517] eta: 0:02:00 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.3672 (0.6950) acc1: 86.7188 (66.4321) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2804 data: 0.1862 max mem: 3531 +Test: [140/517] eta: 0:01:55 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.3406 (0.6710) acc1: 86.9792 (67.8949) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.2600 data: 0.1663 max mem: 3531 +Test: [150/517] eta: 0:01:51 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.3220 (0.6452) acc1: 89.3229 (69.4519) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.2638 data: 0.1703 max mem: 3531 +Test: [160/517] eta: 0:01:48 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.2536 (0.6189) acc1: 92.4479 (70.9918) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2781 data: 0.1849 max mem: 3531 +Test: [170/517] eta: 0:01:44 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.2022 (0.5941) acc1: 94.7917 (72.4370) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2755 data: 0.1808 max mem: 3531 +Test: [180/517] eta: 0:01:40 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1941 (0.5726) acc1: 94.7917 (73.6461) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2699 data: 0.1737 max mem: 3531 +Test: [190/517] eta: 0:01:37 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.0957 (0.5471) acc1: 99.4792 (75.0259) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2633 data: 0.1664 max mem: 3531 +Test: [200/517] eta: 0:01:33 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.0875 (0.5329) acc1: 100.0000 (75.7890) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.2555 data: 0.1577 max mem: 3531 +Test: [210/517] eta: 0:01:30 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.3695 (0.5292) acc1: 82.5521 (75.8528) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.2599 data: 0.1622 max mem: 3532 +Test: [220/517] eta: 0:01:26 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.4542 (0.5255) acc1: 77.8646 (75.9721) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.2598 data: 0.1621 max mem: 3532 +Test: [230/517] eta: 0:01:23 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.3447 (0.5141) acc1: 84.8958 (76.7068) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.2649 data: 0.1668 max mem: 3532 +Test: [240/517] eta: 0:01:20 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.2636 (0.5065) acc1: 91.9271 (77.1276) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.2744 data: 0.1760 max mem: 3532 +Test: [250/517] eta: 0:01:17 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.2357 (0.4958) acc1: 93.2292 (77.8096) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2669 data: 0.1691 max mem: 3532 +Test: [260/517] eta: 0:01:14 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1921 (0.4849) acc1: 97.6562 (78.5281) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.2658 data: 0.1704 max mem: 3532 +Test: [270/517] eta: 0:01:11 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.2955 (0.4805) acc1: 91.9271 (78.8890) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2696 data: 0.1751 max mem: 3532 +Test: [280/517] eta: 0:01:07 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.3235 (0.4747) acc1: 90.1042 (79.3353) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.2639 data: 0.1685 max mem: 3532 +Test: [290/517] eta: 0:01:04 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.2676 (0.4655) acc1: 93.4896 (79.9479) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2531 data: 0.1561 max mem: 3532 +Test: [300/517] eta: 0:01:01 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1945 (0.4564) acc1: 97.6562 (80.5328) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2499 data: 0.1511 max mem: 3532 +Test: [310/517] eta: 0:00:58 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1772 (0.4473) acc1: 97.3958 (81.0850) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2510 data: 0.1528 max mem: 3532 +Test: [320/517] eta: 0:00:55 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1932 (0.4408) acc1: 96.6146 (81.4804) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.2613 data: 0.1649 max mem: 3532 +Test: [330/517] eta: 0:00:52 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.2199 (0.4345) acc1: 93.4896 (81.8495) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2588 data: 0.1647 max mem: 3532 +Test: [340/517] eta: 0:00:49 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.2104 (0.4287) acc1: 93.4896 (82.1114) acc5: 100.0000 (100.0000) acc1_neg: 74.219 acc5_neg: 100.000 time: 0.2359 data: 0.1436 max mem: 3532 +Test: [350/517] eta: 0:00:46 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.2562 (0.4252) acc1: 92.9688 (82.3325) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2499 data: 0.1560 max mem: 3532 +Test: [360/517] eta: 0:00:44 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.2562 (0.4195) acc1: 93.4896 (82.6920) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.3189 data: 0.2234 max mem: 3532 +Test: [370/517] eta: 0:00:41 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1604 (0.4125) acc1: 98.9583 (83.1333) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2999 data: 0.2048 max mem: 3532 +Test: [380/517] eta: 0:00:38 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1846 (0.4072) acc1: 97.9167 (83.4461) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2384 data: 0.1416 max mem: 3532 +Test: [390/517] eta: 0:00:35 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.2098 (0.4027) acc1: 94.7917 (83.7336) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2372 data: 0.1391 max mem: 3532 +Test: [400/517] eta: 0:00:32 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1998 (0.3973) acc1: 95.8333 (84.0522) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2363 data: 0.1411 max mem: 3532 +Test: [410/517] eta: 0:00:29 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.2006 (0.3935) acc1: 95.8333 (84.3224) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.2385 data: 0.1448 max mem: 3532 +Test: [420/517] eta: 0:00:26 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1889 (0.3879) acc1: 96.6146 (84.6682) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2408 data: 0.1468 max mem: 3532 +Test: [430/517] eta: 0:00:23 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1516 (0.3838) acc1: 99.2188 (84.9315) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.2448 data: 0.1523 max mem: 3532 +Test: [440/517] eta: 0:00:21 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.2326 (0.3808) acc1: 96.3542 (85.1291) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2576 data: 0.1647 max mem: 3533 +Test: [450/517] eta: 0:00:18 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1960 (0.3761) acc1: 97.6562 (85.4334) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2552 data: 0.1593 max mem: 3533 +Test: [460/517] eta: 0:00:15 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1642 (0.3722) acc1: 98.9583 (85.6986) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2454 data: 0.1490 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1563 (0.3676) acc1: 98.9583 (85.9795) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2762 data: 0.1802 max mem: 3533 +Test: [480/517] eta: 0:00:10 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1517 (0.3633) acc1: 99.4792 (86.2369) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2782 data: 0.1844 max mem: 3533 +Test: [490/517] eta: 0:00:07 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1618 (0.3597) acc1: 98.6979 (86.4674) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2186 data: 0.1283 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.2491 (0.3588) acc1: 94.5312 (86.5561) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1780 data: 0.0884 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.2795 (0.3558) acc1: 92.9688 (86.7208) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1669 data: 0.0775 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 3.670 acc5_pos: 100.000 loss: 0.1091 (0.3527) acc1: 96.0938 (86.8672) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1684 data: 0.0774 max mem: 3533 +Test: Total time: 0:02:17 (0.2664 s / it) +* Overall Acc@1 86.867 Acc@5 100.000 + target == real Acc@1 24.579 Acc@5 100.000 + target == fake Acc@1 93.329 Acc@5 100.000 + loss 0.353 + PR-AUC 0.9477 ROC-AUC 0.6883 EER 36.48% (thr=0.8252) +Accuracy of the network on the 198427 test images: 86.9% +[rank0]:[W726 03:57:35.157424443 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 03:57:56.408000 1176223 torch/distributed/run.py:766] +W0726 03:57:56.408000 1176223 torch/distributed/run.py:766] ***************************************** +W0726 03:57:56.408000 1176223 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:57:56.408000 1176223 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:58:25.430078476 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:58:26.556042115 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:58:26.764752920 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +| distributed init (rank 7): env:// +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:58:26.180232525 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W726 03:58:26.180233067 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W726 03:58:26.180237469 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:58:26.286925651 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:58:26.308615071 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n8_seed6_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed6_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:34:35 acc1_pos: 4.427 acc5_pos: 100.000 loss: 1.5356 (1.5356) acc1: 4.4271 (4.4271) acc5: 100.0000 (100.0000) time: 5.3348 data: 3.4333 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/389] eta: 0:03:35 acc1_pos: 54.948 acc5_pos: 100.000 loss: 1.5338 (1.5211) acc1: 11.7188 (13.6837) acc5: 100.0000 (100.0000) time: 0.5684 data: 0.3142 max mem: 3531 +Test: [ 20/389] eta: 0:02:06 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.5338 (1.6210) acc1: 11.7188 (15.4886) acc5: 100.0000 (100.0000) time: 0.0942 data: 0.0046 max mem: 3531 +Test: [ 30/389] eta: 0:01:41 acc1_pos: 0.260 acc5_pos: 100.000 loss: 2.0236 (1.7618) acc1: 0.0000 (10.5931) acc5: 100.0000 (100.0000) time: 0.1270 data: 0.0357 max mem: 3531 +Test: [ 40/389] eta: 0:01:38 acc1_pos: 16.927 acc5_pos: 100.000 loss: 2.0050 (1.7102) acc1: 0.2604 (9.9975) acc5: 100.0000 (100.0000) time: 0.2185 data: 0.1234 max mem: 3531 +Test: [ 50/389] eta: 0:01:31 acc1_pos: 2.083 acc5_pos: 100.000 loss: 1.8131 (1.7359) acc1: 2.0833 (8.3435) acc5: 100.0000 (100.0000) time: 0.2497 data: 0.1538 max mem: 3531 +Test: [ 60/389] eta: 0:01:23 acc1_pos: 11.719 acc5_pos: 100.000 loss: 1.8741 (1.7280) acc1: 2.3438 (7.8936) acc5: 100.0000 (100.0000) time: 0.1962 data: 0.1015 max mem: 3531 +Test: [ 70/389] eta: 0:01:17 acc1_pos: 38.802 acc5_pos: 100.000 loss: 1.4648 (1.6412) acc1: 8.8542 (11.1869) acc5: 100.0000 (100.0000) time: 0.1749 data: 0.0814 max mem: 3531 +Test: [ 80/389] eta: 0:01:16 acc1_pos: 42.708 acc5_pos: 100.000 loss: 1.3274 (1.6101) acc1: 11.1979 (12.9790) acc5: 100.0000 (100.0000) time: 0.2286 data: 0.1359 max mem: 3531 +Test: [ 90/389] eta: 0:01:11 acc1_pos: 25.000 acc5_pos: 100.000 loss: 0.9690 (1.5386) acc1: 26.8229 (16.2918) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1314 max mem: 3531 +Test: [100/389] eta: 0:01:06 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.3893 (1.5599) acc1: 18.4896 (15.3981) acc5: 100.0000 (100.0000) time: 0.1664 data: 0.0729 max mem: 3531 +Test: [110/389] eta: 0:01:05 acc1_pos: 60.417 acc5_pos: 100.000 loss: 1.7164 (1.5561) acc1: 0.7812 (16.7582) acc5: 100.0000 (100.0000) time: 0.2184 data: 0.1262 max mem: 3531 +Test: [120/389] eta: 0:01:00 acc1_pos: 8.073 acc5_pos: 100.000 loss: 1.3616 (1.5297) acc1: 26.8229 (18.0139) acc5: 100.0000 (100.0000) time: 0.1976 data: 0.1063 max mem: 3531 +Test: [130/389] eta: 0:00:58 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.8816 (1.4542) acc1: 46.6146 (22.5231) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1618 data: 0.0672 max mem: 3531 +Test: [140/389] eta: 0:00:54 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.2052 (1.3693) acc1: 98.9583 (27.3973) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.1879 data: 0.0925 max mem: 3531 +Test: [150/389] eta: 0:00:52 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.2052 (1.3004) acc1: 95.3125 (31.3638) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1770 data: 0.0830 max mem: 3531 +Test: [160/389] eta: 0:00:49 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1453 (1.2298) acc1: 100.0000 (35.4765) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1842 data: 0.0909 max mem: 3531 +Test: [170/389] eta: 0:00:47 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1198 (1.1653) acc1: 100.0000 (39.2422) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1916 data: 0.0986 max mem: 3531 +Test: [180/389] eta: 0:00:44 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1118 (1.1100) acc1: 100.0000 (42.4407) acc5: 100.0000 (100.0000) acc1_neg: 73.958 acc5_neg: 100.000 time: 0.1683 data: 0.0753 max mem: 3531 +Test: [190/389] eta: 0:00:41 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1669 (1.0645) acc1: 100.0000 (45.1898) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1599 data: 0.0686 max mem: 3531 +Test: [200/389] eta: 0:00:39 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1372 (1.0193) acc1: 100.0000 (47.8817) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1742 data: 0.0818 max mem: 3531 +Test: [210/389] eta: 0:00:36 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1167 (0.9761) acc1: 100.0000 (50.3517) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1622 data: 0.0700 max mem: 3532 +Test: [220/389] eta: 0:00:34 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.0963 (0.9365) acc1: 100.0000 (52.5971) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1548 data: 0.0641 max mem: 3532 +Test: [230/389] eta: 0:00:31 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1186 (0.9072) acc1: 100.0000 (54.3786) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.1628 data: 0.0718 max mem: 3532 +Test: [240/389] eta: 0:00:29 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.2298 (0.8843) acc1: 98.1771 (55.7627) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1711 data: 0.0783 max mem: 3532 +Test: [250/389] eta: 0:00:27 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1770 (0.8548) acc1: 100.0000 (57.5251) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1585 data: 0.0656 max mem: 3532 +Test: [260/389] eta: 0:00:25 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1252 (0.8287) acc1: 100.0000 (59.0757) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1554 data: 0.0645 max mem: 3532 +Test: [270/389] eta: 0:00:23 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1734 (0.8144) acc1: 99.4792 (59.8295) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.1639 data: 0.0718 max mem: 3532 +Test: [280/389] eta: 0:00:21 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.3950 (0.7965) acc1: 86.4583 (60.8930) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1651 data: 0.0732 max mem: 3532 +Test: [290/389] eta: 0:00:19 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.2166 (0.7770) acc1: 95.5729 (62.0597) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1571 data: 0.0661 max mem: 3532 +Test: [300/389] eta: 0:00:17 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1723 (0.7566) acc1: 100.0000 (63.3202) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1590 data: 0.0660 max mem: 3532 +Test: [310/389] eta: 0:00:15 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1777 (0.7393) acc1: 100.0000 (64.4351) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1908 data: 0.0985 max mem: 3532 +Test: [320/389] eta: 0:00:13 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1790 (0.7215) acc1: 99.2188 (65.5252) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1889 data: 0.0963 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1542 (0.7045) acc1: 100.0000 (66.5463) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1656 data: 0.0729 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1392 (0.6877) acc1: 100.0000 (67.4846) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1649 data: 0.0737 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1449 (0.6832) acc1: 98.4375 (67.4947) acc5: 100.0000 (100.0000) acc1_neg: 43.750 acc5_neg: 100.000 time: 0.1637 data: 0.0708 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.5300 (0.6807) acc1: 73.6979 (67.4934) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.1555 data: 0.0635 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.4150 (0.6678) acc1: 77.3438 (68.2481) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1623 data: 0.0714 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1606 (0.6557) acc1: 100.0000 (69.0077) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1581 data: 0.0674 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1577 (0.6460) acc1: 100.0000 (69.6099) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1384 data: 0.0447 max mem: 3532 +Test: Total time: 0:01:12 (0.1860 s / it) +* Overall Acc@1 69.610 Acc@5 100.000 + target == real Acc@1 17.880 Acc@5 100.000 + target == fake Acc@1 93.656 Acc@5 100.000 + loss 0.646 + PR-AUC 0.7884 ROC-AUC 0.6569 EER 38.48% (thr=0.8521) +Accuracy of the network on the 149358 test images: 69.6% +[rank0]:[W726 03:59:51.812670425 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:00:12.376000 1176781 torch/distributed/run.py:766] +W0726 04:00:12.376000 1176781 torch/distributed/run.py:766] ***************************************** +W0726 04:00:12.376000 1176781 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:00:12.376000 1176781 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:00:42.269856122 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:00:42.472084681 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:00:43.742248658 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:00:43.852228161 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:00:43.925401096 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:00:43.960328311 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:00:43.020983815 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W726 04:00:43.020987242 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n8_seed6_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed6_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:30:26 acc1_pos: 0.521 acc5_pos: 100.000 loss: 2.1642 (2.1642) acc1: 0.5208 (0.5208) acc5: 100.0000 (100.0000) time: 5.4864 data: 3.8101 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/333] eta: 0:03:09 acc1_pos: 6.771 acc5_pos: 100.000 loss: 1.9389 (1.8666) acc1: 4.4271 (6.5814) acc5: 100.0000 (100.0000) time: 0.5855 data: 0.3480 max mem: 3531 +Test: [ 20/333] eta: 0:01:57 acc1_pos: 9.115 acc5_pos: 100.000 loss: 1.7130 (1.7562) acc1: 9.1146 (9.1642) acc5: 100.0000 (100.0000) time: 0.1184 data: 0.0248 max mem: 3531 +Test: [ 30/333] eta: 0:01:43 acc1_pos: 27.344 acc5_pos: 100.000 loss: 1.4452 (1.5910) acc1: 19.5312 (17.5487) acc5: 100.0000 (100.0000) time: 0.2061 data: 0.1123 max mem: 3531 +Test: [ 40/333] eta: 0:01:33 acc1_pos: 36.198 acc5_pos: 100.000 loss: 1.2289 (1.5010) acc1: 35.4167 (21.6590) acc5: 100.0000 (100.0000) time: 0.2581 data: 0.1650 max mem: 3531 +Test: [ 50/333] eta: 0:01:22 acc1_pos: 42.188 acc5_pos: 100.000 loss: 1.1809 (1.4473) acc1: 35.6771 (24.2851) acc5: 100.0000 (100.0000) time: 0.2116 data: 0.1202 max mem: 3531 +Test: [ 60/333] eta: 0:01:17 acc1_pos: 30.208 acc5_pos: 100.000 loss: 1.2117 (1.4121) acc1: 35.6771 (26.1399) acc5: 100.0000 (100.0000) time: 0.2117 data: 0.1205 max mem: 3531 +Test: [ 70/333] eta: 0:01:12 acc1_pos: 34.635 acc5_pos: 100.000 loss: 1.2267 (1.3855) acc1: 35.6771 (27.5455) acc5: 100.0000 (100.0000) time: 0.2341 data: 0.1409 max mem: 3531 +Test: [ 80/333] eta: 0:01:07 acc1_pos: 40.104 acc5_pos: 100.000 loss: 1.2001 (1.3656) acc1: 36.1979 (28.6651) acc5: 100.0000 (100.0000) time: 0.2257 data: 0.1323 max mem: 3531 +Test: [ 90/333] eta: 0:01:03 acc1_pos: 57.292 acc5_pos: 100.000 loss: 1.1249 (1.3137) acc1: 40.6250 (31.2758) acc5: 100.0000 (100.0000) time: 0.2192 data: 0.1258 max mem: 3531 +Test: [100/333] eta: 0:00:59 acc1_pos: 63.802 acc5_pos: 100.000 loss: 0.7727 (1.2640) acc1: 55.7292 (33.7897) acc5: 100.0000 (100.0000) time: 0.2074 data: 0.1123 max mem: 3531 +Test: [110/333] eta: 0:00:56 acc1_pos: 71.094 acc5_pos: 100.000 loss: 0.7696 (1.2208) acc1: 55.7292 (35.9328) acc5: 100.0000 (100.0000) time: 0.2072 data: 0.1134 max mem: 3531 +Test: [120/333] eta: 0:00:52 acc1_pos: 65.104 acc5_pos: 100.000 loss: 0.8027 (1.1856) acc1: 53.6458 (37.6055) acc5: 100.0000 (100.0000) time: 0.2039 data: 0.1113 max mem: 3531 +Test: [130/333] eta: 0:00:49 acc1_pos: 52.344 acc5_pos: 100.000 loss: 0.7783 (1.1538) acc1: 57.0312 (39.2613) acc5: 100.0000 (100.0000) time: 0.1975 data: 0.1055 max mem: 3531 +Test: [140/333] eta: 0:00:46 acc1_pos: 62.760 acc5_pos: 100.000 loss: 0.7110 (1.1245) acc1: 59.3750 (40.7321) acc5: 100.0000 (100.0000) time: 0.1927 data: 0.1001 max mem: 3531 +Test: [150/333] eta: 0:00:44 acc1_pos: 61.979 acc5_pos: 100.000 loss: 0.7014 (1.1004) acc1: 61.9792 (41.8719) acc5: 100.0000 (100.0000) time: 0.2221 data: 0.1285 max mem: 3531 +Test: [160/333] eta: 0:00:40 acc1_pos: 53.947 acc5_pos: 100.000 loss: 0.7104 (1.0771) acc1: 60.4167 (43.1192) acc5: 100.0000 (100.0000) acc1_neg: 60.417 acc5_neg: 100.000 time: 0.1975 data: 0.1061 max mem: 3531 +Test: [170/333] eta: 0:00:37 acc1_pos: 53.947 acc5_pos: 100.000 loss: 0.6889 (1.0534) acc1: 61.9792 (44.2587) acc5: 100.0000 (100.0000) acc1_neg: 66.146 acc5_neg: 100.000 time: 0.1621 data: 0.0693 max mem: 3531 +Test: [180/333] eta: 0:00:35 acc1_pos: 53.947 acc5_pos: 100.000 loss: 0.6368 (1.0285) acc1: 65.1042 (45.5557) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.1922 data: 0.0969 max mem: 3531 +Test: [190/333] eta: 0:00:32 acc1_pos: 53.947 acc5_pos: 100.000 loss: 0.5652 (1.0019) acc1: 69.2708 (47.0141) acc5: 100.0000 (100.0000) acc1_neg: 70.833 acc5_neg: 100.000 time: 0.1999 data: 0.1054 max mem: 3531 +Test: [200/333] eta: 0:00:30 acc1_pos: 53.947 acc5_pos: 100.000 loss: 0.5491 (0.9791) acc1: 70.5729 (48.2419) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.1964 data: 0.1021 max mem: 3531 +Test: [210/333] eta: 0:00:27 acc1_pos: 53.947 acc5_pos: 100.000 loss: 0.5817 (0.9625) acc1: 68.7500 (49.0990) acc5: 100.0000 (100.0000) acc1_neg: 60.677 acc5_neg: 100.000 time: 0.1967 data: 0.1034 max mem: 3532 +Test: [220/333] eta: 0:00:25 acc1_pos: 53.947 acc5_pos: 100.000 loss: 0.5960 (0.9519) acc1: 67.1875 (49.4686) acc5: 100.0000 (100.0000) acc1_neg: 71.354 acc5_neg: 100.000 time: 0.2033 data: 0.1069 max mem: 3532 +Test: [230/333] eta: 0:00:23 acc1_pos: 53.947 acc5_pos: 100.000 loss: 0.4940 (0.9296) acc1: 71.3542 (50.7407) acc5: 100.0000 (100.0000) acc1_neg: 77.083 acc5_neg: 100.000 time: 0.2059 data: 0.1099 max mem: 3532 +Test: [240/333] eta: 0:00:20 acc1_pos: 53.947 acc5_pos: 100.000 loss: 0.4940 (0.9149) acc1: 72.3958 (51.5344) acc5: 100.0000 (100.0000) acc1_neg: 68.490 acc5_neg: 100.000 time: 0.2218 data: 0.1312 max mem: 3532 +Test: [250/333] eta: 0:00:18 acc1_pos: 53.947 acc5_pos: 100.000 loss: 0.5665 (0.8997) acc1: 70.5729 (52.3572) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.2089 data: 0.1185 max mem: 3532 +Test: [260/333] eta: 0:00:16 acc1_pos: 53.947 acc5_pos: 100.000 loss: 0.5068 (0.8856) acc1: 72.1354 (53.1180) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.1857 data: 0.0947 max mem: 3532 +Test: [270/333] eta: 0:00:14 acc1_pos: 53.947 acc5_pos: 100.000 loss: 0.5477 (0.8731) acc1: 71.8750 (53.8015) acc5: 100.0000 (100.0000) acc1_neg: 70.312 acc5_neg: 100.000 time: 0.2231 data: 0.1318 max mem: 3532 +Test: [280/333] eta: 0:00:11 acc1_pos: 53.947 acc5_pos: 100.000 loss: 0.5609 (0.8700) acc1: 70.0521 (53.8812) acc5: 100.0000 (100.0000) acc1_neg: 51.823 acc5_neg: 100.000 time: 0.2022 data: 0.1106 max mem: 3532 +Test: [290/333] eta: 0:00:09 acc1_pos: 53.947 acc5_pos: 100.000 loss: 0.9145 (0.8707) acc1: 47.3958 (53.7228) acc5: 100.0000 (100.0000) acc1_neg: 65.104 acc5_neg: 100.000 time: 0.1483 data: 0.0559 max mem: 3532 +Test: [300/333] eta: 0:00:07 acc1_pos: 53.947 acc5_pos: 100.000 loss: 0.7042 (0.8645) acc1: 59.3750 (53.9971) acc5: 100.0000 (100.0000) acc1_neg: 64.844 acc5_neg: 100.000 time: 0.1799 data: 0.0876 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 53.947 acc5_pos: 100.000 loss: 0.6183 (0.8538) acc1: 64.8438 (54.5904) acc5: 100.0000 (100.0000) acc1_neg: 70.573 acc5_neg: 100.000 time: 0.1812 data: 0.0894 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 53.947 acc5_pos: 100.000 loss: 0.5534 (0.8453) acc1: 70.5729 (55.0412) acc5: 100.0000 (100.0000) acc1_neg: 58.073 acc5_neg: 100.000 time: 0.1402 data: 0.0498 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 53.947 acc5_pos: 100.000 loss: 0.6853 (0.8420) acc1: 60.1562 (55.1061) acc5: 100.0000 (100.0000) acc1_neg: 55.208 acc5_neg: 100.000 time: 0.1204 data: 0.0308 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 53.947 acc5_pos: 100.000 loss: 0.7225 (0.8414) acc1: 58.3333 (55.1145) acc5: 100.0000 (100.0000) acc1_neg: 58.028 acc5_neg: 100.000 time: 0.1311 data: 0.0383 max mem: 3532 +Test: Total time: 0:01:09 (0.2099 s / it) +* Overall Acc@1 55.114 Acc@5 100.000 + target == real Acc@1 42.714 Acc@5 100.000 + target == fake Acc@1 66.154 Acc@5 100.000 + loss 0.841 + PR-AUC 0.5740 ROC-AUC 0.5631 EER 45.61% (thr=0.6465) +Accuracy of the network on the 127843 test images: 55.1% +[rank0]:[W726 04:02:01.877144835 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:02:20.474000 1177340 torch/distributed/run.py:766] +W0726 04:02:20.474000 1177340 torch/distributed/run.py:766] ***************************************** +W0726 04:02:20.474000 1177340 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:02:20.474000 1177340 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:02:49.169622475 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:02:49.205496362 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:02:49.293343961 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:02:49.451422666 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:02:50.499728412 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:02:50.900322966 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:02:50.900385526 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W726 04:02:50.903912755 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n8_seed6_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed6_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:20:12 acc1_pos: 92.708 acc5_pos: 100.000 loss: 0.2658 (0.2658) acc1: 92.7083 (92.7083) acc5: 100.0000 (100.0000) time: 5.1613 data: 3.3899 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/235] eta: 0:02:11 acc1_pos: 92.708 acc5_pos: 100.000 loss: 0.2658 (0.2628) acc1: 92.7083 (92.0691) acc5: 100.0000 (100.0000) time: 0.5822 data: 0.3371 max mem: 3531 +Test: [ 20/235] eta: 0:01:22 acc1_pos: 91.667 acc5_pos: 100.000 loss: 0.2528 (0.2588) acc1: 92.1875 (92.3363) acc5: 100.0000 (100.0000) time: 0.1448 data: 0.0537 max mem: 3531 +Test: [ 30/235] eta: 0:01:04 acc1_pos: 93.750 acc5_pos: 100.000 loss: 0.2528 (0.2597) acc1: 92.1875 (92.3135) acc5: 100.0000 (100.0000) time: 0.1666 data: 0.0739 max mem: 3531 +Test: [ 40/235] eta: 0:00:57 acc1_pos: 93.490 acc5_pos: 100.000 loss: 0.2535 (0.2583) acc1: 92.1875 (92.2828) acc5: 100.0000 (100.0000) time: 0.2039 data: 0.1100 max mem: 3531 +Test: [ 50/235] eta: 0:00:49 acc1_pos: 89.583 acc5_pos: 100.000 loss: 0.2553 (0.2598) acc1: 91.9271 (92.1926) acc5: 100.0000 (100.0000) time: 0.1999 data: 0.1065 max mem: 3531 +Test: [ 60/235] eta: 0:00:43 acc1_pos: 93.229 acc5_pos: 100.000 loss: 0.2660 (0.2607) acc1: 91.6667 (92.1277) acc5: 100.0000 (100.0000) time: 0.1599 data: 0.0650 max mem: 3531 +Test: [ 70/235] eta: 0:00:40 acc1_pos: 92.308 acc5_pos: 100.000 loss: 0.2784 (0.4143) acc1: 90.6250 (83.3297) acc5: 100.0000 (100.0000) acc1_neg: 17.448 acc5_neg: 100.000 time: 0.1799 data: 0.0831 max mem: 3531 +Test: [ 80/235] eta: 0:00:36 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.4973 (0.5510) acc1: 20.3125 (75.4469) acc5: 100.0000 (100.0000) acc1_neg: 17.969 acc5_neg: 100.000 time: 0.1899 data: 0.0924 max mem: 3531 +Test: [ 90/235] eta: 0:00:34 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.5112 (0.6545) acc1: 19.7917 (69.4340) acc5: 100.0000 (100.0000) acc1_neg: 19.271 acc5_neg: 100.000 time: 0.2249 data: 0.1283 max mem: 3531 +Test: [100/235] eta: 0:00:31 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.4978 (0.7366) acc1: 21.3542 (64.6658) acc5: 100.0000 (100.0000) acc1_neg: 17.708 acc5_neg: 100.000 time: 0.2249 data: 0.1272 max mem: 3531 +Test: [110/235] eta: 0:00:28 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.4900 (0.8044) acc1: 21.0938 (60.7005) acc5: 100.0000 (100.0000) acc1_neg: 19.531 acc5_neg: 100.000 time: 0.1810 data: 0.0842 max mem: 3531 +Test: [120/235] eta: 0:00:25 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.5479 (0.8798) acc1: 17.1875 (56.6761) acc5: 100.0000 (100.0000) acc1_neg: 11.719 acc5_neg: 100.000 time: 0.1863 data: 0.0928 max mem: 3531 +Test: [130/235] eta: 0:00:23 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.7277 (0.9449) acc1: 10.6771 (53.2045) acc5: 100.0000 (100.0000) acc1_neg: 11.458 acc5_neg: 100.000 time: 0.2138 data: 0.1215 max mem: 3531 +Test: [140/235] eta: 0:00:21 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.7560 (1.0043) acc1: 10.4167 (50.1644) acc5: 100.0000 (100.0000) acc1_neg: 7.031 acc5_neg: 100.000 time: 0.2105 data: 0.1164 max mem: 3531 +Test: [150/235] eta: 0:00:18 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.7413 (1.0527) acc1: 10.9375 (47.6356) acc5: 100.0000 (100.0000) acc1_neg: 13.542 acc5_neg: 100.000 time: 0.1748 data: 0.0793 max mem: 3531 +Test: [160/235] eta: 0:00:16 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.7348 (1.0952) acc1: 11.4583 (45.4047) acc5: 100.0000 (100.0000) acc1_neg: 12.760 acc5_neg: 100.000 time: 0.1941 data: 0.0980 max mem: 3531 +Test: [170/235] eta: 0:00:14 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.6928 (1.0959) acc1: 12.7604 (45.0902) acc5: 100.0000 (100.0000) acc1_neg: 39.844 acc5_neg: 100.000 time: 0.2204 data: 0.1244 max mem: 3531 +Test: [180/235] eta: 0:00:11 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.0368 (1.0919) acc1: 42.7083 (44.9801) acc5: 100.0000 (100.0000) acc1_neg: 40.104 acc5_neg: 100.000 time: 0.1998 data: 0.1036 max mem: 3531 +Test: [190/235] eta: 0:00:09 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.0082 (1.0883) acc1: 43.4896 (44.9457) acc5: 100.0000 (100.0000) acc1_neg: 42.969 acc5_neg: 100.000 time: 0.1825 data: 0.0857 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.0248 (1.0854) acc1: 43.4896 (44.8500) acc5: 100.0000 (100.0000) acc1_neg: 40.885 acc5_neg: 100.000 time: 0.1657 data: 0.0714 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.0395 (1.0825) acc1: 43.4896 (44.8065) acc5: 100.0000 (100.0000) acc1_neg: 44.792 acc5_neg: 100.000 time: 0.1568 data: 0.0648 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.0236 (1.0804) acc1: 43.4896 (44.7339) acc5: 100.0000 (100.0000) acc1_neg: 43.750 acc5_neg: 100.000 time: 0.1680 data: 0.0754 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.0273 (1.0781) acc1: 43.7500 (44.6992) acc5: 100.0000 (100.0000) acc1_neg: 44.792 acc5_neg: 100.000 time: 0.1608 data: 0.0692 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.0314 (1.0783) acc1: 43.7500 (44.6756) acc5: 100.0000 (100.0000) acc1_neg: 34.783 acc5_neg: 100.000 time: 0.1441 data: 0.0571 max mem: 3532 +Test: Total time: 0:00:48 (0.2047 s / it) +* Overall Acc@1 44.676 Acc@5 100.000 + target == real Acc@1 92.125 Acc@5 100.000 + target == fake Acc@1 27.514 Acc@5 100.000 + loss 1.078 + PR-AUC 0.8555 ROC-AUC 0.6844 EER 36.89% (thr=0.1680) +Accuracy of the network on the 89879 test images: 44.7% +[rank0]:[W726 04:03:52.713506090 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:04:11.826000 1177886 torch/distributed/run.py:766] +W0726 04:04:11.826000 1177886 torch/distributed/run.py:766] ***************************************** +W0726 04:04:11.826000 1177886 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:04:11.826000 1177886 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 3): env:// +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:04:43.554933903 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:04:43.631613894 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:04:43.961266881 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:04:43.002782062 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:04:43.029639887 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +| distributed init (rank 1): env:// +[rank0]:[W726 04:04:43.168729153 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:04:43.170521198 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank1]:[W726 04:04:43.170635349 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n8_seed6_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed6_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:28:33 acc1_pos: 15.625 acc5_pos: 100.000 loss: 1.4758 (1.4758) acc1: 15.6250 (15.6250) acc5: 100.0000 (100.0000) time: 6.3709 data: 4.5866 max mem: 3531 +Test: [ 10/269] eta: 0:02:51 acc1_pos: 25.521 acc5_pos: 100.000 loss: 1.2697 (1.1800) acc1: 31.5104 (33.4044) acc5: 100.0000 (100.0000) time: 0.6616 data: 0.4174 max mem: 3531 +Test: [ 20/269] eta: 0:01:39 acc1_pos: 31.510 acc5_pos: 100.000 loss: 1.1983 (1.2018) acc1: 28.9062 (32.5025) acc5: 100.0000 (100.0000) time: 0.1003 data: 0.0065 max mem: 3531 +Test: [ 30/269] eta: 0:01:25 acc1_pos: 40.365 acc5_pos: 100.000 loss: 1.1807 (1.1674) acc1: 28.1250 (34.4506) acc5: 100.0000 (100.0000) time: 0.1899 data: 0.0948 max mem: 3531 +Test: [ 40/269] eta: 0:01:15 acc1_pos: 36.207 acc5_pos: 100.000 loss: 0.4807 (0.9769) acc1: 55.9896 (47.1481) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.2599 data: 0.1676 max mem: 3531 +Test: [ 50/269] eta: 0:01:08 acc1_pos: 36.207 acc5_pos: 100.000 loss: 0.2782 (0.8449) acc1: 91.9271 (55.3819) acc5: 100.0000 (100.0000) acc1_neg: 66.927 acc5_neg: 100.000 time: 0.2473 data: 0.1565 max mem: 3531 +Test: [ 60/269] eta: 0:01:02 acc1_pos: 36.207 acc5_pos: 100.000 loss: 0.3117 (0.7747) acc1: 84.6354 (59.2299) acc5: 100.0000 (100.0000) acc1_neg: 68.750 acc5_neg: 100.000 time: 0.2319 data: 0.1412 max mem: 3531 +Test: [ 70/269] eta: 0:00:54 acc1_pos: 36.207 acc5_pos: 100.000 loss: 0.4330 (0.7275) acc1: 77.6042 (62.0745) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.1725 data: 0.0778 max mem: 3531 +Test: [ 80/269] eta: 0:00:49 acc1_pos: 36.207 acc5_pos: 100.000 loss: 0.4222 (0.6881) acc1: 80.9896 (64.5608) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1432 data: 0.0477 max mem: 3531 +Test: [ 90/269] eta: 0:00:45 acc1_pos: 36.207 acc5_pos: 100.000 loss: 0.4222 (0.6671) acc1: 80.9896 (65.7652) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.1774 data: 0.0827 max mem: 3531 +Test: [100/269] eta: 0:00:41 acc1_pos: 36.207 acc5_pos: 100.000 loss: 0.4733 (0.6477) acc1: 76.0417 (66.8472) acc5: 100.0000 (100.0000) acc1_neg: 76.042 acc5_neg: 100.000 time: 0.1921 data: 0.0988 max mem: 3531 +Test: [110/269] eta: 0:00:38 acc1_pos: 36.207 acc5_pos: 100.000 loss: 0.4659 (0.6312) acc1: 74.4792 (67.6403) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.1895 data: 0.0974 max mem: 3531 +Test: [120/269] eta: 0:00:35 acc1_pos: 36.207 acc5_pos: 100.000 loss: 0.3922 (0.6063) acc1: 81.7708 (69.2386) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.2024 data: 0.1096 max mem: 3531 +Test: [130/269] eta: 0:00:32 acc1_pos: 36.207 acc5_pos: 100.000 loss: 0.3139 (0.5861) acc1: 86.7188 (70.5610) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.2085 data: 0.1155 max mem: 3531 +Test: [140/269] eta: 0:00:30 acc1_pos: 36.207 acc5_pos: 100.000 loss: 0.3540 (0.5807) acc1: 79.9479 (70.7170) acc5: 100.0000 (100.0000) acc1_neg: 58.594 acc5_neg: 100.000 time: 0.2200 data: 0.1271 max mem: 3531 +Test: [150/269] eta: 0:00:27 acc1_pos: 36.207 acc5_pos: 100.000 loss: 0.5340 (0.5749) acc1: 71.0938 (70.8747) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.2258 data: 0.1345 max mem: 3531 +Test: [160/269] eta: 0:00:25 acc1_pos: 36.207 acc5_pos: 100.000 loss: 0.3186 (0.5589) acc1: 85.9375 (71.9656) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.2112 data: 0.1192 max mem: 3531 +Test: [170/269] eta: 0:00:22 acc1_pos: 36.207 acc5_pos: 100.000 loss: 0.3060 (0.5448) acc1: 88.8021 (72.8634) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.2036 data: 0.1119 max mem: 3531 +Test: [180/269] eta: 0:00:20 acc1_pos: 36.207 acc5_pos: 100.000 loss: 0.2548 (0.5295) acc1: 89.3229 (73.7670) acc5: 100.0000 (100.0000) acc1_neg: 74.740 acc5_neg: 100.000 time: 0.1909 data: 0.0993 max mem: 3531 +Test: [190/269] eta: 0:00:17 acc1_pos: 36.207 acc5_pos: 100.000 loss: 0.3020 (0.5211) acc1: 86.7188 (74.2774) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1845 data: 0.0903 max mem: 3531 +Test: [200/269] eta: 0:00:15 acc1_pos: 36.207 acc5_pos: 100.000 loss: 0.3936 (0.5153) acc1: 79.1667 (74.5025) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1896 data: 0.0955 max mem: 3531 +Test: [210/269] eta: 0:00:13 acc1_pos: 36.207 acc5_pos: 100.000 loss: 0.4371 (0.5125) acc1: 76.3021 (74.5532) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.2016 data: 0.1098 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 36.207 acc5_pos: 100.000 loss: 0.4457 (0.5086) acc1: 76.3021 (74.7089) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2093 data: 0.1175 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 36.207 acc5_pos: 100.000 loss: 0.2387 (0.4971) acc1: 89.8438 (75.4633) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.2033 data: 0.1120 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 36.207 acc5_pos: 100.000 loss: 0.2946 (0.4960) acc1: 88.0208 (75.4830) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.2183 data: 0.1263 max mem: 3532 +Test: [250/269] eta: 0:00:04 acc1_pos: 36.207 acc5_pos: 100.000 loss: 0.5128 (0.5019) acc1: 69.2708 (74.9772) acc5: 100.0000 (100.0000) acc1_neg: 51.562 acc5_neg: 100.000 time: 0.1907 data: 0.0982 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 36.207 acc5_pos: 100.000 loss: 0.6195 (0.5079) acc1: 62.7604 (74.4383) acc5: 100.0000 (100.0000) acc1_neg: 80.729 acc5_neg: 100.000 time: 0.1359 data: 0.0447 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 36.207 acc5_pos: 100.000 loss: 0.4786 (0.5052) acc1: 73.4375 (74.6169) acc5: 100.0000 (100.0000) acc1_neg: 92.913 acc5_neg: 100.000 time: 0.1427 data: 0.0487 max mem: 3532 +Test: Total time: 0:00:57 (0.2147 s / it) +* Overall Acc@1 74.617 Acc@5 100.000 + target == real Acc@1 34.416 Acc@5 100.000 + target == fake Acc@1 80.103 Acc@5 100.000 + loss 0.505 + PR-AUC 0.9177 ROC-AUC 0.6244 EER 41.28% (thr=0.7285) +Accuracy of the network on the 103293 test images: 74.6% +[rank0]:[W726 04:05:49.900840021 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T04:05:55Z] eval_done tag=fixed500k_real_n8_seed6_resampled +[2026-07-26T04:05:55Z] eval_start tag=fixed500k_fake_n16_seed6_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed6_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T04:05:55Z +hostname=n116-041-153 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed6_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n16_seed6_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 04:06:08.892000 1178464 torch/distributed/run.py:766] +W0726 04:06:08.892000 1178464 torch/distributed/run.py:766] ***************************************** +W0726 04:06:08.892000 1178464 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:06:08.892000 1178464 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:06:39.532334040 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:06:39.619842676 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:06:39.751521773 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:06:39.435781548 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:06:40.488552466 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:06:40.537921225 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:06:40.643568025 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:06:40.662440323 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n16_seed6_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed6_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:20:11 acc1_pos: 71.875 acc5_pos: 100.000 loss: 0.6136 (0.6136) acc1: 71.8750 (71.8750) acc5: 100.0000 (100.0000) time: 5.1795 data: 2.6009 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/234] eta: 0:02:03 acc1_pos: 46.615 acc5_pos: 100.000 loss: 0.6136 (0.6300) acc1: 71.8750 (65.0331) acc5: 100.0000 (100.0000) time: 0.5527 data: 0.2368 max mem: 3531 +Test: [ 20/234] eta: 0:01:13 acc1_pos: 95.833 acc5_pos: 100.000 loss: 0.5839 (0.5904) acc1: 61.9792 (66.1582) acc5: 100.0000 (100.0000) time: 0.1020 data: 0.0125 max mem: 3531 +Test: [ 30/234] eta: 0:00:53 acc1_pos: 87.240 acc5_pos: 100.000 loss: 0.2291 (0.4697) acc1: 93.7500 (75.5124) acc5: 100.0000 (100.0000) time: 0.1019 data: 0.0125 max mem: 3531 +Test: [ 40/234] eta: 0:00:45 acc1_pos: 99.479 acc5_pos: 100.000 loss: 0.2121 (0.4177) acc1: 95.3125 (79.8399) acc5: 100.0000 (100.0000) time: 0.1151 data: 0.0232 max mem: 3531 +Test: [ 50/234] eta: 0:00:44 acc1_pos: 99.740 acc5_pos: 100.000 loss: 0.1111 (0.3572) acc1: 99.2188 (83.6703) acc5: 100.0000 (100.0000) time: 0.2088 data: 0.1129 max mem: 3531 +Test: [ 60/234] eta: 0:00:41 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.1100 (0.3170) acc1: 99.7396 (86.3388) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2465 data: 0.1516 max mem: 3531 +Test: [ 70/234] eta: 0:00:38 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.1139 (0.3611) acc1: 99.7396 (83.4250) acc5: 100.0000 (100.0000) acc1_neg: 45.312 acc5_neg: 100.000 time: 0.2173 data: 0.1230 max mem: 3531 +Test: [ 80/234] eta: 0:00:35 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.8523 (0.3957) acc1: 50.0000 (81.3111) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2000 data: 0.1043 max mem: 3531 +Test: [ 90/234] eta: 0:00:32 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.2028 (0.3700) acc1: 95.0521 (82.9441) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.2007 data: 0.1081 max mem: 3531 +Test: [100/234] eta: 0:00:30 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.2093 (0.3755) acc1: 93.7500 (82.5289) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2165 data: 0.1238 max mem: 3531 +Test: [110/234] eta: 0:00:27 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.3765 (0.3737) acc1: 82.2917 (82.6835) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.2003 data: 0.1070 max mem: 3531 +Test: [120/234] eta: 0:00:25 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.3435 (0.3973) acc1: 83.8542 (81.3985) acc5: 100.0000 (100.0000) acc1_neg: 53.385 acc5_neg: 100.000 time: 0.2032 data: 0.1117 max mem: 3531 +Test: [130/234] eta: 0:00:22 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.5008 (0.4055) acc1: 77.3438 (81.1108) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.2060 data: 0.1144 max mem: 3531 +Test: [140/234] eta: 0:00:20 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.5124 (0.4317) acc1: 74.7396 (79.8500) acc5: 100.0000 (100.0000) acc1_neg: 32.812 acc5_neg: 100.000 time: 0.2007 data: 0.1089 max mem: 3531 +Test: [150/234] eta: 0:00:17 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.5170 (0.4414) acc1: 71.8750 (79.2408) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.1783 data: 0.0866 max mem: 3531 +Test: [160/234] eta: 0:00:15 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.5016 (0.4464) acc1: 71.8750 (79.0082) acc5: 100.0000 (100.0000) acc1_neg: 62.760 acc5_neg: 100.000 time: 0.1614 data: 0.0704 max mem: 3531 +Test: [170/234] eta: 0:00:13 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.6935 (0.4584) acc1: 63.5417 (78.3108) acc5: 100.0000 (100.0000) acc1_neg: 78.646 acc5_neg: 100.000 time: 0.1743 data: 0.0835 max mem: 3531 +Test: [180/234] eta: 0:00:11 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.6460 (0.4687) acc1: 64.5833 (77.6516) acc5: 100.0000 (100.0000) acc1_neg: 33.333 acc5_neg: 100.000 time: 0.1845 data: 0.0926 max mem: 3531 +Test: [190/234] eta: 0:00:09 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.9261 (0.4948) acc1: 48.4375 (76.0376) acc5: 100.0000 (100.0000) acc1_neg: 45.573 acc5_neg: 100.000 time: 0.1922 data: 0.1000 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.9261 (0.4917) acc1: 49.2188 (76.2749) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1903 data: 0.0991 max mem: 3531 +Test: [210/234] eta: 0:00:04 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.1358 (0.4762) acc1: 99.2188 (77.3006) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1886 data: 0.0982 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.2178 (0.4666) acc1: 95.3125 (77.9836) acc5: 100.0000 (100.0000) acc1_neg: 70.052 acc5_neg: 100.000 time: 0.1764 data: 0.0857 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.5994 (0.5043) acc1: 25.0000 (75.5434) acc5: 100.0000 (100.0000) acc1_neg: 25.000 acc5_neg: 100.000 time: 0.1584 data: 0.0678 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 99.310 acc5_pos: 100.000 loss: 1.3212 (0.5149) acc1: 23.1771 (74.9278) acc5: 100.0000 (100.0000) acc1_neg: 21.401 acc5_neg: 100.000 time: 0.1520 data: 0.0613 max mem: 3532 +Test: Total time: 0:00:47 (0.2022 s / it) +* Overall Acc@1 74.928 Acc@5 100.000 + target == real Acc@1 83.785 Acc@5 100.000 + target == fake Acc@1 72.431 Acc@5 100.000 + loss 0.515 + PR-AUC 0.9609 ROC-AUC 0.8679 EER 23.02% (thr=0.4119) +Accuracy of the network on the 89729 test images: 74.9% +[rank0]:[W726 04:07:40.751620721 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:08:02.293000 1179018 torch/distributed/run.py:766] +W0726 04:08:02.293000 1179018 torch/distributed/run.py:766] ***************************************** +W0726 04:08:02.293000 1179018 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:08:02.293000 1179018 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:08:32.603499600 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:08:32.672724891 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:08:32.943157333 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:08:32.989934819 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:08:32.128266629 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:08:32.137869431 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:08:32.137923980 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:08:32.148939820 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n16_seed6_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed6_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:49:34 acc1_pos: 51.042 acc5_pos: 100.000 loss: 0.8319 (0.8319) acc1: 51.0417 (51.0417) acc5: 100.0000 (100.0000) time: 7.2787 data: 5.4949 max mem: 3531 +Test: [ 10/2387] eta: 0:29:27 acc1_pos: 25.781 acc5_pos: 100.000 loss: 0.8937 (0.9590) acc1: 42.9688 (37.4053) acc5: 100.0000 (100.0000) time: 0.7436 data: 0.4999 max mem: 3531 +Test: [ 20/2387] eta: 0:17:11 acc1_pos: 54.427 acc5_pos: 100.000 loss: 0.8937 (0.9221) acc1: 35.9375 (40.1166) acc5: 100.0000 (100.0000) time: 0.0934 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:15:46 acc1_pos: 7.552 acc5_pos: 100.000 loss: 0.9769 (0.9782) acc1: 27.8646 (35.6183) acc5: 100.0000 (100.0000) time: 0.2134 data: 0.1195 max mem: 3531 +Test: [ 40/2387] eta: 0:14:41 acc1_pos: 22.917 acc5_pos: 100.000 loss: 1.0894 (1.0159) acc1: 21.6146 (31.8661) acc5: 100.0000 (100.0000) time: 0.3127 data: 0.2206 max mem: 3531 +Test: [ 50/2387] eta: 0:14:18 acc1_pos: 44.531 acc5_pos: 100.000 loss: 1.0420 (1.0012) acc1: 23.9583 (32.6644) acc5: 100.0000 (100.0000) time: 0.3149 data: 0.2205 max mem: 3531 +Test: [ 60/2387] eta: 0:13:34 acc1_pos: 54.688 acc5_pos: 100.000 loss: 0.9328 (0.9870) acc1: 36.7188 (33.8883) acc5: 100.0000 (100.0000) time: 0.2971 data: 0.2022 max mem: 3531 +Test: [ 70/2387] eta: 0:12:51 acc1_pos: 37.240 acc5_pos: 100.000 loss: 0.9771 (0.9952) acc1: 35.4167 (33.5827) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1493 max mem: 3531 +Test: [ 80/2387] eta: 0:12:18 acc1_pos: 24.479 acc5_pos: 100.000 loss: 1.0467 (1.0038) acc1: 27.0833 (32.5360) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1321 max mem: 3531 +Test: [ 90/2387] eta: 0:12:00 acc1_pos: 27.604 acc5_pos: 100.000 loss: 1.0429 (1.0049) acc1: 24.4792 (32.1801) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1466 max mem: 3531 +Test: [ 100/2387] eta: 0:11:45 acc1_pos: 28.906 acc5_pos: 100.000 loss: 1.0051 (1.0126) acc1: 24.4792 (31.3531) acc5: 100.0000 (100.0000) time: 0.2599 data: 0.1609 max mem: 3531 +Test: [ 110/2387] eta: 0:11:32 acc1_pos: 42.188 acc5_pos: 100.000 loss: 0.9895 (1.0122) acc1: 26.8229 (31.2828) acc5: 100.0000 (100.0000) time: 0.2599 data: 0.1607 max mem: 3531 +Test: [ 120/2387] eta: 0:11:18 acc1_pos: 30.208 acc5_pos: 100.000 loss: 1.0021 (1.0124) acc1: 30.2083 (31.4007) acc5: 100.0000 (100.0000) time: 0.2549 data: 0.1577 max mem: 3531 +Test: [ 130/2387] eta: 0:11:07 acc1_pos: 34.635 acc5_pos: 100.000 loss: 1.0250 (1.0173) acc1: 28.3854 (31.0731) acc5: 100.0000 (100.0000) time: 0.2499 data: 0.1553 max mem: 3531 +Test: [ 140/2387] eta: 0:10:57 acc1_pos: 40.104 acc5_pos: 100.000 loss: 1.0384 (1.0180) acc1: 27.0833 (31.0210) acc5: 100.0000 (100.0000) time: 0.2499 data: 0.1543 max mem: 3531 +Test: [ 150/2387] eta: 0:10:44 acc1_pos: 14.844 acc5_pos: 100.000 loss: 1.0022 (1.0146) acc1: 28.6458 (31.2414) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1450 max mem: 3531 +Test: [ 160/2387] eta: 0:10:36 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.9564 (0.9990) acc1: 30.9896 (32.3984) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2399 data: 0.1469 max mem: 3531 +Test: [ 170/2387] eta: 0:10:20 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2257 (0.9526) acc1: 94.2708 (36.1218) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2149 data: 0.1214 max mem: 3531 +Test: [ 180/2387] eta: 0:10:02 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2087 (0.9114) acc1: 95.8333 (39.4136) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1708 data: 0.0787 max mem: 3531 +Test: [ 190/2387] eta: 0:09:55 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2015 (0.8742) acc1: 95.8333 (42.3770) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1959 data: 0.1055 max mem: 3531 +Test: [ 200/2387] eta: 0:09:42 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1989 (0.8408) acc1: 96.0938 (45.0352) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2040 data: 0.1095 max mem: 3531 +Test: [ 210/2387] eta: 0:09:30 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2064 (0.8109) acc1: 96.0938 (47.4403) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1789 data: 0.0817 max mem: 3532 +Test: [ 220/2387] eta: 0:09:21 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2095 (0.7836) acc1: 95.5729 (49.6147) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1863 data: 0.0898 max mem: 3532 +Test: [ 230/2387] eta: 0:09:20 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2109 (0.7588) acc1: 95.5729 (51.6155) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2349 data: 0.1394 max mem: 3532 +Test: [ 240/2387] eta: 0:09:07 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2014 (0.7355) acc1: 96.8750 (53.5129) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2094 data: 0.1151 max mem: 3532 +Test: [ 250/2387] eta: 0:08:59 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1882 (0.7101) acc1: 98.4375 (55.3411) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1652 data: 0.0714 max mem: 3532 +Test: [ 260/2387] eta: 0:08:52 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0806 (0.6859) acc1: 100.0000 (57.0502) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1940 data: 0.1008 max mem: 3532 +Test: [ 270/2387] eta: 0:08:44 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0768 (0.6634) acc1: 100.0000 (58.6332) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1904 data: 0.0943 max mem: 3532 +Test: [ 280/2387] eta: 0:08:40 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0751 (0.6427) acc1: 100.0000 (60.1053) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2028 data: 0.1088 max mem: 3532 +Test: [ 290/2387] eta: 0:08:32 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0762 (0.6233) acc1: 100.0000 (61.4744) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2001 data: 0.1094 max mem: 3532 +Test: [ 300/2387] eta: 0:08:27 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0761 (0.6051) acc1: 100.0000 (62.7544) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1920 data: 0.0969 max mem: 3532 +Test: [ 310/2387] eta: 0:08:23 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0781 (0.5910) acc1: 100.0000 (63.8950) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2138 data: 0.1152 max mem: 3532 +Test: [ 320/2387] eta: 0:08:17 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2490 (0.5805) acc1: 97.9167 (64.9476) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2049 data: 0.1096 max mem: 3532 +Test: [ 330/2387] eta: 0:08:12 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2505 (0.5705) acc1: 97.9167 (65.9436) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1929 data: 0.1016 max mem: 3532 +Test: [ 340/2387] eta: 0:08:08 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2503 (0.5611) acc1: 97.9167 (66.8828) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2028 data: 0.1113 max mem: 3532 +Test: [ 350/2387] eta: 0:08:02 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2416 (0.5491) acc1: 98.6979 (67.8181) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1951 data: 0.1039 max mem: 3532 +Test: [ 360/2387] eta: 0:07:57 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1572 (0.5384) acc1: 98.4375 (68.6512) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1828 data: 0.0889 max mem: 3532 +Test: [ 370/2387] eta: 0:07:52 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1691 (0.5285) acc1: 97.6562 (69.4358) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1887 data: 0.0943 max mem: 3532 +Test: [ 380/2387] eta: 0:07:47 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1691 (0.5190) acc1: 97.9167 (70.1861) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1862 data: 0.0950 max mem: 3532 +Test: [ 390/2387] eta: 0:07:43 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1718 (0.5101) acc1: 97.9167 (70.8853) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1887 data: 0.0979 max mem: 3532 +Test: [ 400/2387] eta: 0:07:38 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1674 (0.5015) acc1: 97.9167 (71.5652) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1890 data: 0.0976 max mem: 3532 +Test: [ 410/2387] eta: 0:07:33 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1649 (0.4934) acc1: 98.1771 (72.2045) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1791 data: 0.0838 max mem: 3532 +Test: [ 420/2387] eta: 0:07:28 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1641 (0.4856) acc1: 97.9167 (72.8196) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1762 data: 0.0825 max mem: 3532 +Test: [ 430/2387] eta: 0:07:25 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1658 (0.4782) acc1: 97.9167 (73.4031) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1900 data: 0.0973 max mem: 3532 +Test: [ 440/2387] eta: 0:07:20 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1651 (0.4702) acc1: 98.4375 (73.9926) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1875 data: 0.0939 max mem: 3533 +Test: [ 450/2387] eta: 0:07:16 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1002 (0.4620) acc1: 100.0000 (74.5692) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0885 max mem: 3533 +Test: [ 460/2387] eta: 0:07:12 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0994 (0.4542) acc1: 100.0000 (75.1209) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1847 data: 0.0922 max mem: 3533 +Test: [ 470/2387] eta: 0:07:08 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0997 (0.4467) acc1: 100.0000 (75.6491) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1798 data: 0.0864 max mem: 3533 +Test: [ 480/2387] eta: 0:07:03 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0999 (0.4395) acc1: 100.0000 (76.1554) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1717 data: 0.0792 max mem: 3533 +Test: [ 490/2387] eta: 0:07:00 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0998 (0.4325) acc1: 100.0000 (76.6410) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1708 data: 0.0778 max mem: 3533 +Test: [ 500/2387] eta: 0:06:56 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1002 (0.4259) acc1: 100.0000 (77.1072) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1798 data: 0.0856 max mem: 3533 +Test: [ 510/2387] eta: 0:06:52 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1002 (0.4197) acc1: 100.0000 (77.5552) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1811 data: 0.0881 max mem: 3533 +Test: [ 520/2387] eta: 0:06:48 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1224 (0.4140) acc1: 100.0000 (77.9860) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1690 data: 0.0743 max mem: 3533 +Test: [ 530/2387] eta: 0:06:44 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1178 (0.4084) acc1: 100.0000 (78.4006) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1628 data: 0.0663 max mem: 3533 +Test: [ 540/2387] eta: 0:06:40 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1171 (0.4030) acc1: 100.0000 (78.7989) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1640 data: 0.0678 max mem: 3533 +Test: [ 550/2387] eta: 0:06:36 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1174 (0.3979) acc1: 100.0000 (79.1837) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1702 data: 0.0764 max mem: 3533 +Test: [ 560/2387] eta: 0:06:33 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1178 (0.3929) acc1: 100.0000 (79.5538) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1764 data: 0.0861 max mem: 3533 +Test: [ 570/2387] eta: 0:06:30 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1170 (0.3880) acc1: 100.0000 (79.9114) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1877 data: 0.0962 max mem: 3533 +Test: [ 580/2387] eta: 0:06:29 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1171 (0.3834) acc1: 100.0000 (80.2567) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2201 data: 0.1280 max mem: 3533 +Test: [ 590/2387] eta: 0:06:25 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1186 (0.3789) acc1: 100.0000 (80.5908) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2087 data: 0.1172 max mem: 3533 +Test: [ 600/2387] eta: 0:06:23 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1214 (0.3748) acc1: 100.0000 (80.9120) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1970 data: 0.1027 max mem: 3533 +Test: [ 610/2387] eta: 0:06:20 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1395 (0.3714) acc1: 98.9583 (81.1933) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1970 data: 0.1026 max mem: 3533 +Test: [ 620/2387] eta: 0:06:17 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1667 (0.3681) acc1: 98.1771 (81.4660) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1859 data: 0.0935 max mem: 3533 +Test: [ 630/2387] eta: 0:06:14 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1669 (0.3649) acc1: 97.9167 (81.7312) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1849 data: 0.0921 max mem: 3533 +Test: [ 640/2387] eta: 0:06:11 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1701 (0.3620) acc1: 97.9167 (81.9829) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1793 data: 0.0864 max mem: 3533 +Test: [ 650/2387] eta: 0:06:08 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1733 (0.3591) acc1: 97.6562 (82.2241) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1770 data: 0.0849 max mem: 3533 +Test: [ 660/2387] eta: 0:06:05 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1695 (0.3562) acc1: 97.6562 (82.4666) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1846 data: 0.0918 max mem: 3533 +Test: [ 670/2387] eta: 0:06:02 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1677 (0.3534) acc1: 98.1771 (82.7019) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1878 data: 0.0946 max mem: 3534 +Test: [ 680/2387] eta: 0:06:00 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1625 (0.3504) acc1: 98.6979 (82.9490) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1820 data: 0.0912 max mem: 3534 +Test: [ 690/2387] eta: 0:05:57 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1612 (0.3503) acc1: 98.1771 (82.9855) acc5: 100.0000 (100.0000) acc1_neg: 77.083 acc5_neg: 100.000 time: 0.1759 data: 0.0851 max mem: 3534 +Test: [ 700/2387] eta: 0:05:54 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.4072 (0.3513) acc1: 80.7292 (82.9433) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.1797 data: 0.0884 max mem: 3534 +Test: [ 710/2387] eta: 0:05:51 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.4118 (0.3522) acc1: 80.2083 (82.9030) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.1739 data: 0.0829 max mem: 3534 +Test: [ 720/2387] eta: 0:05:48 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.4071 (0.3530) acc1: 80.4688 (82.8721) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.1801 data: 0.0885 max mem: 3534 +Test: [ 730/2387] eta: 0:05:45 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.4180 (0.3540) acc1: 79.4271 (82.8164) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.1775 data: 0.0861 max mem: 3534 +Test: [ 740/2387] eta: 0:05:42 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.4205 (0.3549) acc1: 79.6875 (82.7816) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.1554 data: 0.0651 max mem: 3534 +Test: [ 750/2387] eta: 0:05:39 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.4110 (0.3556) acc1: 80.4688 (82.7584) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.1678 data: 0.0772 max mem: 3534 +Test: [ 760/2387] eta: 0:05:36 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.3997 (0.3559) acc1: 81.5104 (82.7506) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.1739 data: 0.0813 max mem: 3534 +Test: [ 770/2387] eta: 0:05:34 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.3096 (0.3544) acc1: 85.4167 (82.8655) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0817 max mem: 3534 +Test: [ 780/2387] eta: 0:05:31 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2139 (0.3523) acc1: 92.9688 (83.0179) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1834 data: 0.0915 max mem: 3534 +Test: [ 790/2387] eta: 0:05:29 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1757 (0.3504) acc1: 96.0938 (83.1338) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.1903 data: 0.0980 max mem: 3534 +Test: [ 800/2387] eta: 0:05:27 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1550 (0.3480) acc1: 96.8750 (83.3050) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2200 data: 0.1280 max mem: 3534 +Test: [ 810/2387] eta: 0:05:25 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1420 (0.3454) acc1: 97.6562 (83.4894) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2290 data: 0.1371 max mem: 3534 +Test: [ 820/2387] eta: 0:05:22 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1354 (0.3432) acc1: 98.4375 (83.6531) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1613 data: 0.0671 max mem: 3534 +Test: [ 830/2387] eta: 0:05:19 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1618 (0.3416) acc1: 96.3542 (83.7746) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1349 data: 0.0391 max mem: 3534 +Test: [ 840/2387] eta: 0:05:16 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2151 (0.3404) acc1: 93.4896 (83.9000) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1558 data: 0.0603 max mem: 3534 +Test: [ 850/2387] eta: 0:05:13 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2491 (0.3396) acc1: 93.2292 (83.9962) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1569 data: 0.0629 max mem: 3534 +Test: [ 860/2387] eta: 0:05:10 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2418 (0.3383) acc1: 94.2708 (84.1258) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1681 data: 0.0754 max mem: 3534 +Test: [ 870/2387] eta: 0:05:08 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2146 (0.3370) acc1: 95.8333 (84.2545) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1779 data: 0.0843 max mem: 3534 +Test: [ 880/2387] eta: 0:05:05 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2244 (0.3358) acc1: 96.0938 (84.3747) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1794 data: 0.0856 max mem: 3534 +Test: [ 890/2387] eta: 0:05:03 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1665 (0.3338) acc1: 98.4375 (84.5343) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1802 data: 0.0865 max mem: 3534 +Test: [ 900/2387] eta: 0:05:01 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2060 (0.3378) acc1: 95.8333 (84.2585) acc5: 100.0000 (100.0000) acc1_neg: 54.167 acc5_neg: 100.000 time: 0.1797 data: 0.0856 max mem: 3535 +Test: [ 910/2387] eta: 0:04:58 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.8094 (0.3430) acc1: 50.7812 (83.8788) acc5: 100.0000 (100.0000) acc1_neg: 47.656 acc5_neg: 100.000 time: 0.1764 data: 0.0843 max mem: 3535 +Test: [ 920/2387] eta: 0:04:56 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.8219 (0.3483) acc1: 48.4375 (83.4945) acc5: 100.0000 (100.0000) acc1_neg: 45.052 acc5_neg: 100.000 time: 0.1728 data: 0.0814 max mem: 3535 +Test: [ 930/2387] eta: 0:04:53 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.8170 (0.3533) acc1: 48.6979 (83.1350) acc5: 100.0000 (100.0000) acc1_neg: 45.573 acc5_neg: 100.000 time: 0.1751 data: 0.0811 max mem: 3535 +Test: [ 940/2387] eta: 0:04:51 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.8191 (0.3583) acc1: 50.0000 (82.7740) acc5: 100.0000 (100.0000) acc1_neg: 48.177 acc5_neg: 100.000 time: 0.1870 data: 0.0929 max mem: 3535 +Test: [ 950/2387] eta: 0:04:49 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.8191 (0.3631) acc1: 48.1771 (82.4190) acc5: 100.0000 (100.0000) acc1_neg: 52.604 acc5_neg: 100.000 time: 0.1878 data: 0.0950 max mem: 3535 +Test: [ 960/2387] eta: 0:04:46 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.8041 (0.3677) acc1: 49.2188 (82.0884) acc5: 100.0000 (100.0000) acc1_neg: 53.385 acc5_neg: 100.000 time: 0.1761 data: 0.0818 max mem: 3535 +Test: [ 970/2387] eta: 0:04:45 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.8055 (0.3725) acc1: 48.9583 (81.7303) acc5: 100.0000 (100.0000) acc1_neg: 46.615 acc5_neg: 100.000 time: 0.2025 data: 0.1090 max mem: 3535 +Test: [ 980/2387] eta: 0:04:43 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.7897 (0.3735) acc1: 48.4375 (81.6540) acc5: 100.0000 (100.0000) acc1_neg: 77.083 acc5_neg: 100.000 time: 0.2112 data: 0.1200 max mem: 3535 +Test: [ 990/2387] eta: 0:04:40 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2704 (0.3720) acc1: 87.5000 (81.7598) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1574 data: 0.0667 max mem: 3535 +Test: [1000/2387] eta: 0:04:37 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1823 (0.3705) acc1: 96.0938 (81.8749) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1626 data: 0.0717 max mem: 3535 +Test: [1010/2387] eta: 0:04:35 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1826 (0.3686) acc1: 96.0938 (82.0101) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1898 data: 0.0946 max mem: 3535 +Test: [1020/2387] eta: 0:04:33 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1726 (0.3671) acc1: 96.8750 (82.1266) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1819 data: 0.0821 max mem: 3535 +Test: [1030/2387] eta: 0:04:31 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1726 (0.3652) acc1: 95.3125 (82.2568) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1799 data: 0.0829 max mem: 3535 +Test: [1040/2387] eta: 0:04:28 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1356 (0.3630) acc1: 98.1771 (82.4107) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1788 data: 0.0853 max mem: 3535 +Test: [1050/2387] eta: 0:04:26 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1266 (0.3608) acc1: 98.9583 (82.5650) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1749 data: 0.0794 max mem: 3535 +Test: [1060/2387] eta: 0:04:24 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1305 (0.3587) acc1: 98.9583 (82.7192) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1700 data: 0.0746 max mem: 3535 +Test: [1070/2387] eta: 0:04:21 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1445 (0.3568) acc1: 98.4375 (82.8541) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1677 data: 0.0759 max mem: 3535 +Test: [1080/2387] eta: 0:04:19 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1643 (0.3553) acc1: 97.1354 (82.9674) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1664 data: 0.0761 max mem: 3535 +Test: [1090/2387] eta: 0:04:17 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1524 (0.3532) acc1: 97.6562 (83.1059) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1695 data: 0.0761 max mem: 3535 +Test: [1100/2387] eta: 0:04:14 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1311 (0.3512) acc1: 97.9167 (83.2418) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1679 data: 0.0741 max mem: 3535 +Test: [1110/2387] eta: 0:04:12 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1303 (0.3497) acc1: 98.1771 (83.3369) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1643 data: 0.0728 max mem: 3535 +Test: [1120/2387] eta: 0:04:09 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1303 (0.3478) acc1: 98.6979 (83.4748) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1614 data: 0.0696 max mem: 3536 +Test: [1130/2387] eta: 0:04:07 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1213 (0.3458) acc1: 99.2188 (83.6133) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1631 data: 0.0714 max mem: 3536 +Test: [1140/2387] eta: 0:04:05 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1242 (0.3438) acc1: 99.4792 (83.7485) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1634 data: 0.0714 max mem: 3536 +Test: [1150/2387] eta: 0:04:03 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1344 (0.3426) acc1: 98.9583 (83.8524) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1768 data: 0.0847 max mem: 3536 +Test: [1160/2387] eta: 0:04:01 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1829 (0.3410) acc1: 97.6562 (83.9668) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1921 data: 0.1007 max mem: 3536 +Test: [1170/2387] eta: 0:03:59 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1829 (0.3395) acc1: 97.6562 (84.0801) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1891 data: 0.0984 max mem: 3536 +Test: [1180/2387] eta: 0:03:57 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1655 (0.3381) acc1: 97.9167 (84.1937) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1869 data: 0.0954 max mem: 3536 +Test: [1190/2387] eta: 0:03:55 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1473 (0.3365) acc1: 98.6979 (84.3153) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2052 data: 0.1108 max mem: 3536 +Test: [1200/2387] eta: 0:03:53 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1148 (0.3346) acc1: 99.7396 (84.4440) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2017 data: 0.1083 max mem: 3536 +Test: [1210/2387] eta: 0:03:51 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1114 (0.3328) acc1: 99.7396 (84.5703) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1956 data: 0.1039 max mem: 3536 +Test: [1220/2387] eta: 0:03:49 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0962 (0.3308) acc1: 100.0000 (84.6926) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2030 data: 0.1100 max mem: 3536 +Test: [1230/2387] eta: 0:03:47 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0832 (0.3288) acc1: 100.0000 (84.8150) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2193 data: 0.1274 max mem: 3536 +Test: [1240/2387] eta: 0:03:45 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0835 (0.3269) acc1: 100.0000 (84.9353) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1998 data: 0.1082 max mem: 3536 +Test: [1250/2387] eta: 0:03:44 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0864 (0.3249) acc1: 99.7396 (85.0532) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1982 data: 0.1065 max mem: 3536 +Test: [1260/2387] eta: 0:03:41 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0810 (0.3230) acc1: 100.0000 (85.1703) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2117 data: 0.1178 max mem: 3536 +Test: [1270/2387] eta: 0:03:39 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0745 (0.3210) acc1: 100.0000 (85.2870) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1808 data: 0.0870 max mem: 3536 +Test: [1280/2387] eta: 0:03:38 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0835 (0.3194) acc1: 100.0000 (85.3953) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2012 data: 0.1105 max mem: 3536 +Test: [1290/2387] eta: 0:03:35 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1174 (0.3180) acc1: 98.6979 (85.4846) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1941 data: 0.1028 max mem: 3536 +Test: [1300/2387] eta: 0:03:34 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1174 (0.3165) acc1: 98.6979 (85.5820) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2218 data: 0.1299 max mem: 3536 +Test: [1310/2387] eta: 0:03:32 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1219 (0.3151) acc1: 98.9583 (85.6761) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2197 data: 0.1279 max mem: 3536 +Test: [1320/2387] eta: 0:03:29 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1254 (0.3138) acc1: 98.9583 (85.7688) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1499 data: 0.0529 max mem: 3536 +Test: [1330/2387] eta: 0:03:27 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1262 (0.3124) acc1: 99.2188 (85.8657) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1675 data: 0.0705 max mem: 3536 +Test: [1340/2387] eta: 0:03:25 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1414 (0.3121) acc1: 97.1354 (85.8876) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1925 data: 0.0988 max mem: 3536 +Test: [1350/2387] eta: 0:03:24 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0710 (0.3103) acc1: 100.0000 (85.9921) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2123 data: 0.1192 max mem: 3537 +Test: [1360/2387] eta: 0:03:21 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0671 (0.3085) acc1: 100.0000 (86.0950) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1873 data: 0.0937 max mem: 3537 +Test: [1370/2387] eta: 0:03:19 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0664 (0.3067) acc1: 100.0000 (86.1960) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1810 data: 0.0845 max mem: 3537 +Test: [1380/2387] eta: 0:03:17 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0666 (0.3050) acc1: 100.0000 (86.2960) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2048 data: 0.1100 max mem: 3537 +Test: [1390/2387] eta: 0:03:15 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0686 (0.3033) acc1: 100.0000 (86.3945) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1915 data: 0.0991 max mem: 3537 +Test: [1400/2387] eta: 0:03:14 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0664 (0.3016) acc1: 100.0000 (86.4916) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.1011 max mem: 3537 +Test: [1410/2387] eta: 0:03:12 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0678 (0.3006) acc1: 100.0000 (86.5530) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1973 data: 0.1019 max mem: 3537 +Test: [1420/2387] eta: 0:03:10 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2104 (0.3003) acc1: 92.4479 (86.5705) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.2009 data: 0.1080 max mem: 3537 +Test: [1430/2387] eta: 0:03:08 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2185 (0.2997) acc1: 92.1875 (86.6167) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1956 data: 0.1050 max mem: 3537 +Test: [1440/2387] eta: 0:03:06 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1998 (0.2988) acc1: 93.7500 (86.6828) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1908 data: 0.1002 max mem: 3537 +Test: [1450/2387] eta: 0:03:04 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1854 (0.2980) acc1: 96.8750 (86.7460) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2165 data: 0.1253 max mem: 3537 +Test: [1460/2387] eta: 0:03:02 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1422 (0.2969) acc1: 97.6562 (86.8228) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2335 data: 0.1411 max mem: 3537 +Test: [1470/2387] eta: 0:03:00 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1174 (0.2956) acc1: 99.2188 (86.9099) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2298 data: 0.1363 max mem: 3537 +Test: [1480/2387] eta: 0:02:59 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1116 (0.2944) acc1: 99.7396 (86.9906) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2202 data: 0.1277 max mem: 3537 +Test: [1490/2387] eta: 0:02:57 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1075 (0.2932) acc1: 99.4792 (87.0744) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2230 data: 0.1317 max mem: 3537 +Test: [1500/2387] eta: 0:02:55 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1069 (0.2919) acc1: 99.4792 (87.1551) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2240 data: 0.1308 max mem: 3537 +Test: [1510/2387] eta: 0:02:53 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1149 (0.2908) acc1: 99.7396 (87.2398) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1956 data: 0.1018 max mem: 3537 +Test: [1520/2387] eta: 0:02:51 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1441 (0.2903) acc1: 99.7396 (87.2932) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1976 data: 0.1043 max mem: 3537 +Test: [1530/2387] eta: 0:02:49 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2225 (0.2901) acc1: 94.5312 (87.3275) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.2000 data: 0.1077 max mem: 3537 +Test: [1540/2387] eta: 0:02:47 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2307 (0.2896) acc1: 94.5312 (87.3819) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1905 data: 0.0999 max mem: 3537 +Test: [1550/2387] eta: 0:02:45 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2138 (0.2892) acc1: 96.3542 (87.4382) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2035 data: 0.1106 max mem: 3537 +Test: [1560/2387] eta: 0:02:43 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2059 (0.2886) acc1: 96.8750 (87.4910) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.2020 data: 0.1085 max mem: 3537 +Test: [1570/2387] eta: 0:02:41 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1482 (0.2877) acc1: 98.6979 (87.5622) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1999 data: 0.1067 max mem: 3537 +Test: [1580/2387] eta: 0:02:39 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1414 (0.2870) acc1: 98.6979 (87.6160) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2144 data: 0.1195 max mem: 3538 +Test: [1590/2387] eta: 0:02:37 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1540 (0.2861) acc1: 97.3958 (87.6728) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2249 data: 0.1303 max mem: 3538 +Test: [1600/2387] eta: 0:02:35 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1380 (0.2855) acc1: 97.3958 (87.7077) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2120 data: 0.1180 max mem: 3538 +Test: [1610/2387] eta: 0:02:34 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1380 (0.2846) acc1: 97.6562 (87.7734) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2112 data: 0.1191 max mem: 3538 +Test: [1620/2387] eta: 0:02:32 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1396 (0.2837) acc1: 97.9167 (87.8343) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2148 data: 0.1234 max mem: 3538 +Test: [1630/2387] eta: 0:02:29 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1529 (0.2831) acc1: 97.1354 (87.8775) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1736 data: 0.0806 max mem: 3538 +Test: [1640/2387] eta: 0:02:27 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1939 (0.2827) acc1: 95.3125 (87.9199) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1631 data: 0.0715 max mem: 3538 +Test: [1650/2387] eta: 0:02:26 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1939 (0.2822) acc1: 94.7917 (87.9641) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2149 data: 0.1233 max mem: 3538 +Test: [1660/2387] eta: 0:02:23 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1936 (0.2818) acc1: 95.3125 (88.0020) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1958 data: 0.1024 max mem: 3538 +Test: [1670/2387] eta: 0:02:21 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1879 (0.2813) acc1: 95.5729 (88.0469) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1699 data: 0.0754 max mem: 3538 +Test: [1680/2387] eta: 0:02:19 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1879 (0.2809) acc1: 95.5729 (88.0799) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.1891 data: 0.0961 max mem: 3538 +Test: [1690/2387] eta: 0:02:17 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2100 (0.2806) acc1: 94.2708 (88.1066) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1999 data: 0.1051 max mem: 3538 +Test: [1700/2387] eta: 0:02:15 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1986 (0.2799) acc1: 95.0521 (88.1586) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1971 data: 0.1007 max mem: 3538 +Test: [1710/2387] eta: 0:02:13 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1169 (0.2788) acc1: 98.9583 (88.2259) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1736 data: 0.0811 max mem: 3538 +Test: [1720/2387] eta: 0:02:11 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0859 (0.2777) acc1: 99.7396 (88.2923) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1878 data: 0.0936 max mem: 3538 +Test: [1730/2387] eta: 0:02:09 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0878 (0.2767) acc1: 99.4792 (88.3565) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1937 data: 0.0994 max mem: 3538 +Test: [1740/2387] eta: 0:02:07 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0878 (0.2756) acc1: 99.7396 (88.4210) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1883 data: 0.0963 max mem: 3538 +Test: [1750/2387] eta: 0:02:05 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0794 (0.2745) acc1: 100.0000 (88.4866) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2020 data: 0.1106 max mem: 3538 +Test: [1760/2387] eta: 0:02:03 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0757 (0.2734) acc1: 100.0000 (88.5516) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2049 data: 0.1137 max mem: 3538 +Test: [1770/2387] eta: 0:02:01 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0943 (0.2725) acc1: 99.4792 (88.6074) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2091 data: 0.1166 max mem: 3538 +Test: [1780/2387] eta: 0:01:59 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1091 (0.2716) acc1: 99.2188 (88.6673) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1957 data: 0.1031 max mem: 3538 +Test: [1790/2387] eta: 0:01:57 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1003 (0.2706) acc1: 99.4792 (88.7281) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1760 data: 0.0842 max mem: 3538 +Test: [1800/2387] eta: 0:01:56 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1010 (0.2697) acc1: 99.4792 (88.7862) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2091 data: 0.1181 max mem: 3538 +Test: [1810/2387] eta: 0:01:53 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0925 (0.2687) acc1: 99.7396 (88.8468) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1902 data: 0.0976 max mem: 3539 +Test: [1820/2387] eta: 0:01:51 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0785 (0.2676) acc1: 99.7396 (88.9075) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1650 data: 0.0705 max mem: 3539 +Test: [1830/2387] eta: 0:01:49 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.0914 (0.2668) acc1: 99.7396 (88.9666) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1792 data: 0.0852 max mem: 3539 +Test: [1840/2387] eta: 0:01:47 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1640 (0.2663) acc1: 99.7396 (89.0256) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1800 data: 0.0881 max mem: 3539 +Test: [1850/2387] eta: 0:01:45 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1664 (0.2657) acc1: 99.7396 (89.0840) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1871 data: 0.0955 max mem: 3539 +Test: [1860/2387] eta: 0:01:43 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1663 (0.2652) acc1: 99.7396 (89.1411) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1897 data: 0.0981 max mem: 3539 +Test: [1870/2387] eta: 0:01:41 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1663 (0.2647) acc1: 99.7396 (89.1972) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2148 data: 0.1228 max mem: 3539 +Test: [1880/2387] eta: 0:01:39 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1645 (0.2641) acc1: 99.7396 (89.2533) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2149 data: 0.1220 max mem: 3539 +Test: [1890/2387] eta: 0:01:37 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1643 (0.2636) acc1: 100.0000 (89.3096) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1592 data: 0.0654 max mem: 3539 +Test: [1900/2387] eta: 0:01:35 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1650 (0.2631) acc1: 100.0000 (89.3650) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1724 data: 0.0752 max mem: 3539 +Test: [1910/2387] eta: 0:01:33 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1672 (0.2626) acc1: 99.7396 (89.4167) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2107 data: 0.1157 max mem: 3539 +Test: [1920/2387] eta: 0:01:31 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1672 (0.2625) acc1: 99.4792 (89.4216) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1634 data: 0.0727 max mem: 3539 +Test: [1930/2387] eta: 0:01:29 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1463 (0.2618) acc1: 97.6562 (89.4678) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1534 data: 0.0606 max mem: 3539 +Test: [1940/2387] eta: 0:01:27 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1347 (0.2615) acc1: 98.6979 (89.4755) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1918 data: 0.0985 max mem: 3539 +Test: [1950/2387] eta: 0:01:25 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1578 (0.2611) acc1: 97.1354 (89.5039) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1908 data: 0.0993 max mem: 3539 +Test: [1960/2387] eta: 0:01:23 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1675 (0.2607) acc1: 96.6146 (89.5357) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1875 data: 0.0966 max mem: 3539 +Test: [1970/2387] eta: 0:01:21 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1905 (0.2606) acc1: 94.7917 (89.5407) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1991 data: 0.1066 max mem: 3539 +Test: [1980/2387] eta: 0:01:19 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2789 (0.2606) acc1: 93.4896 (89.5602) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1917 data: 0.0996 max mem: 3539 +Test: [1990/2387] eta: 0:01:17 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2948 (0.2609) acc1: 93.4896 (89.5737) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1842 data: 0.0943 max mem: 3539 +Test: [2000/2387] eta: 0:01:15 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.3010 (0.2611) acc1: 92.1875 (89.5884) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1875 data: 0.0971 max mem: 3539 +Test: [2010/2387] eta: 0:01:14 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.3111 (0.2614) acc1: 91.9271 (89.5995) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1952 data: 0.1047 max mem: 3539 +Test: [2020/2387] eta: 0:01:12 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.3141 (0.2616) acc1: 91.6667 (89.6112) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.2013 data: 0.1105 max mem: 3539 +Test: [2030/2387] eta: 0:01:10 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.3082 (0.2618) acc1: 92.1875 (89.6228) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2038 data: 0.1123 max mem: 3540 +Test: [2040/2387] eta: 0:01:08 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.3029 (0.2620) acc1: 92.4479 (89.6371) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1946 data: 0.1029 max mem: 3540 +Test: [2050/2387] eta: 0:01:06 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2985 (0.2622) acc1: 92.9688 (89.6560) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1802 data: 0.0891 max mem: 3540 +Test: [2060/2387] eta: 0:01:04 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2714 (0.2622) acc1: 94.7917 (89.6840) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1801 data: 0.0873 max mem: 3540 +Test: [2070/2387] eta: 0:01:02 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2625 (0.2621) acc1: 95.8333 (89.7149) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1727 data: 0.0801 max mem: 3540 +Test: [2080/2387] eta: 0:01:00 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2590 (0.2621) acc1: 95.8333 (89.7444) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1740 data: 0.0830 max mem: 3540 +Test: [2090/2387] eta: 0:00:58 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2590 (0.2621) acc1: 95.5729 (89.7730) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1750 data: 0.0839 max mem: 3540 +Test: [2100/2387] eta: 0:00:56 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2606 (0.2621) acc1: 95.5729 (89.8019) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1773 data: 0.0852 max mem: 3540 +Test: [2110/2387] eta: 0:00:54 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2587 (0.2621) acc1: 95.5729 (89.8303) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1815 data: 0.0894 max mem: 3540 +Test: [2120/2387] eta: 0:00:52 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2536 (0.2618) acc1: 95.3125 (89.8520) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1833 data: 0.0925 max mem: 3540 +Test: [2130/2387] eta: 0:00:50 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1528 (0.2613) acc1: 97.3958 (89.8880) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1958 data: 0.1052 max mem: 3540 +Test: [2140/2387] eta: 0:00:48 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1528 (0.2611) acc1: 97.3958 (89.8964) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1814 data: 0.0896 max mem: 3540 +Test: [2150/2387] eta: 0:00:46 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1737 (0.2606) acc1: 96.0938 (89.9272) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1809 data: 0.0883 max mem: 3540 +Test: [2160/2387] eta: 0:00:44 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1581 (0.2601) acc1: 96.8750 (89.9646) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2164 data: 0.1257 max mem: 3540 +Test: [2170/2387] eta: 0:00:42 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1601 (0.2598) acc1: 97.3958 (89.9914) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2088 data: 0.1191 max mem: 3540 +Test: [2180/2387] eta: 0:00:40 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2110 (0.2607) acc1: 94.5312 (89.9291) acc5: 100.0000 (100.0000) acc1_neg: 47.917 acc5_neg: 100.000 time: 0.1404 data: 0.0467 max mem: 3540 +Test: [2190/2387] eta: 0:00:38 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2591 (0.2610) acc1: 89.5833 (89.9127) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1378 data: 0.0425 max mem: 3540 +Test: [2200/2387] eta: 0:00:36 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1242 (0.2603) acc1: 98.1771 (89.9541) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0977 max mem: 3540 +Test: [2210/2387] eta: 0:00:34 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1177 (0.2597) acc1: 98.9583 (89.9929) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1889 data: 0.0972 max mem: 3540 +Test: [2220/2387] eta: 0:00:32 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1206 (0.2591) acc1: 98.6979 (90.0299) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1972 data: 0.1046 max mem: 3540 +Test: [2230/2387] eta: 0:00:30 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1128 (0.2584) acc1: 98.6979 (90.0653) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1887 data: 0.0957 max mem: 3540 +Test: [2240/2387] eta: 0:00:28 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1033 (0.2578) acc1: 98.6979 (90.1005) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1274 data: 0.0331 max mem: 3540 +Test: [2250/2387] eta: 0:00:26 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1181 (0.2572) acc1: 99.2188 (90.1392) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1361 data: 0.0409 max mem: 3540 +Test: [2260/2387] eta: 0:00:24 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1308 (0.2568) acc1: 99.2188 (90.1745) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1834 data: 0.0903 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1306 (0.2562) acc1: 99.2188 (90.2145) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1883 data: 0.0967 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1306 (0.2559) acc1: 99.2188 (90.2307) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2023 data: 0.1110 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1285 (0.2553) acc1: 99.2188 (90.2700) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2161 data: 0.1246 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1271 (0.2548) acc1: 99.4792 (90.3087) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1839 data: 0.0913 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1255 (0.2543) acc1: 99.2188 (90.3411) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1750 data: 0.0817 max mem: 3541 +Test: [2320/2387] eta: 0:00:13 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1574 (0.2541) acc1: 97.1354 (90.3585) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1862 data: 0.0935 max mem: 3541 +Test: [2330/2387] eta: 0:00:11 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2049 (0.2540) acc1: 92.7083 (90.3643) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1791 data: 0.0874 max mem: 3541 +Test: [2340/2387] eta: 0:00:09 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2011 (0.2538) acc1: 94.0104 (90.3826) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1954 data: 0.1051 max mem: 3541 +Test: [2350/2387] eta: 0:00:07 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1775 (0.2539) acc1: 95.5729 (90.3723) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1891 data: 0.0993 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1791 (0.2536) acc1: 96.0938 (90.3902) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1435 data: 0.0538 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.1813 (0.2535) acc1: 96.0938 (90.3951) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.1302 data: 0.0406 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2135 (0.2535) acc1: 92.7083 (90.4040) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.1374 data: 0.0478 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 21.918 acc5_pos: 100.000 loss: 0.2628 (0.2546) acc1: 85.9375 (90.3422) acc5: 100.0000 (100.0000) acc1_neg: 44.186 acc5_neg: 100.000 time: 0.1299 data: 0.0427 max mem: 3541 +Test: Total time: 0:07:41 (0.1935 s / it) +* Overall Acc@1 90.342 Acc@5 100.000 + target == real Acc@1 30.967 Acc@5 100.000 + target == fake Acc@1 94.512 Acc@5 100.000 + loss 0.255 + PR-AUC 0.9908 ROC-AUC 0.8799 EER 19.35% (thr=0.7783) +Accuracy of the network on the 916267 test images: 90.3% +[rank0]:[W726 04:16:31.349486112 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:16:52.418000 1179641 torch/distributed/run.py:766] +W0726 04:16:52.418000 1179641 torch/distributed/run.py:766] ***************************************** +W0726 04:16:52.418000 1179641 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:16:52.418000 1179641 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:17:22.286092705 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:17:23.680513670 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:17:23.793892484 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:17:23.828918007 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:17:23.845769636 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:17:23.925262823 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:17:23.944462539 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:17:23.955208033 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n16_seed6_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed6_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:05:37 acc1_pos: 51.042 acc5_pos: 100.000 loss: 0.8319 (0.8319) acc1: 51.0417 (51.0417) acc5: 100.0000 (100.0000) time: 7.6156 data: 6.1260 max mem: 3531 +Test: [ 10/517] eta: 0:06:33 acc1_pos: 25.781 acc5_pos: 100.000 loss: 0.8937 (0.9590) acc1: 42.9688 (37.4053) acc5: 100.0000 (100.0000) time: 0.7764 data: 0.5573 max mem: 3531 +Test: [ 20/517] eta: 0:03:45 acc1_pos: 54.427 acc5_pos: 100.000 loss: 0.8937 (0.9221) acc1: 35.9375 (40.1166) acc5: 100.0000 (100.0000) time: 0.0962 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:03:18 acc1_pos: 7.552 acc5_pos: 100.000 loss: 0.9769 (0.9782) acc1: 27.8646 (35.6183) acc5: 100.0000 (100.0000) time: 0.2052 data: 0.1075 max mem: 3531 +Test: [ 40/517] eta: 0:02:56 acc1_pos: 22.917 acc5_pos: 100.000 loss: 1.0894 (1.0159) acc1: 21.6146 (31.8661) acc5: 100.0000 (100.0000) time: 0.2801 data: 0.1849 max mem: 3531 +Test: [ 50/517] eta: 0:02:42 acc1_pos: 36.697 acc5_pos: 100.000 loss: 1.0420 (0.9903) acc1: 23.9583 (33.5427) acc5: 100.0000 (100.0000) acc1_neg: 61.979 acc5_neg: 100.000 time: 0.2562 data: 0.1627 max mem: 3531 +Test: [ 60/517] eta: 0:02:32 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.6356 (0.9126) acc1: 59.1146 (40.1383) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.2599 data: 0.1682 max mem: 3531 +Test: [ 70/517] eta: 0:02:25 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.5455 (0.8539) acc1: 68.7500 (45.1071) acc5: 100.0000 (100.0000) acc1_neg: 72.396 acc5_neg: 100.000 time: 0.2634 data: 0.1712 max mem: 3531 +Test: [ 80/517] eta: 0:02:16 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.5455 (0.8199) acc1: 67.1875 (47.9135) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.2501 data: 0.1567 max mem: 3531 +Test: [ 90/517] eta: 0:02:10 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.4801 (0.7758) acc1: 78.3854 (52.2636) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.2425 data: 0.1499 max mem: 3531 +Test: [100/517] eta: 0:02:06 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.4260 (0.7419) acc1: 85.9375 (55.5977) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2662 data: 0.1750 max mem: 3531 +Test: [110/517] eta: 0:02:01 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.3856 (0.7061) acc1: 86.4583 (58.7087) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2673 data: 0.1741 max mem: 3531 +Test: [120/517] eta: 0:01:58 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.4060 (0.6868) acc1: 84.8958 (60.5609) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.2681 data: 0.1713 max mem: 3531 +Test: [130/517] eta: 0:01:54 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.4615 (0.6681) acc1: 82.8125 (62.3569) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2799 data: 0.1823 max mem: 3531 +Test: [140/517] eta: 0:01:50 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.4137 (0.6508) acc1: 87.7604 (64.1622) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.2609 data: 0.1653 max mem: 3531 +Test: [150/517] eta: 0:01:46 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.4060 (0.6342) acc1: 89.5833 (65.7837) acc5: 100.0000 (100.0000) acc1_neg: 69.792 acc5_neg: 100.000 time: 0.2555 data: 0.1615 max mem: 3531 +Test: [160/517] eta: 0:01:43 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.3785 (0.6196) acc1: 89.3229 (67.0791) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.2706 data: 0.1770 max mem: 3531 +Test: [170/517] eta: 0:01:39 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.3645 (0.6034) acc1: 90.3646 (68.4942) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.2594 data: 0.1665 max mem: 3531 +Test: [180/517] eta: 0:01:36 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.3560 (0.5890) acc1: 90.8854 (69.7341) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.2532 data: 0.1573 max mem: 3531 +Test: [190/517] eta: 0:01:32 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.3569 (0.5767) acc1: 90.3646 (70.8961) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.2549 data: 0.1572 max mem: 3531 +Test: [200/517] eta: 0:01:29 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.3328 (0.5645) acc1: 92.4479 (72.0447) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.2549 data: 0.1592 max mem: 3531 +Test: [210/517] eta: 0:01:26 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.3917 (0.5645) acc1: 83.0729 (72.1083) acc5: 100.0000 (100.0000) acc1_neg: 79.427 acc5_neg: 100.000 time: 0.2599 data: 0.1632 max mem: 3532 +Test: [220/517] eta: 0:01:23 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.5479 (0.5628) acc1: 74.7396 (72.2662) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.2499 data: 0.1517 max mem: 3532 +Test: [230/517] eta: 0:01:20 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.5598 (0.5641) acc1: 71.8750 (72.0069) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.2499 data: 0.1526 max mem: 3532 +Test: [240/517] eta: 0:01:17 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.5752 (0.5632) acc1: 67.9688 (72.1322) acc5: 100.0000 (100.0000) acc1_neg: 62.760 acc5_neg: 100.000 time: 0.2749 data: 0.1789 max mem: 3532 +Test: [250/517] eta: 0:01:14 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.4762 (0.5573) acc1: 84.1146 (72.8669) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2599 data: 0.1637 max mem: 3532 +Test: [260/517] eta: 0:01:11 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.3940 (0.5510) acc1: 91.4062 (73.5542) acc5: 100.0000 (100.0000) acc1_neg: 66.146 acc5_neg: 100.000 time: 0.2449 data: 0.1480 max mem: 3532 +Test: [270/517] eta: 0:01:08 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.5072 (0.5534) acc1: 71.3542 (73.2280) acc5: 100.0000 (100.0000) acc1_neg: 70.573 acc5_neg: 100.000 time: 0.2599 data: 0.1641 max mem: 3532 +Test: [280/517] eta: 0:01:05 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.5785 (0.5521) acc1: 69.5312 (73.3356) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.2649 data: 0.1677 max mem: 3532 +Test: [290/517] eta: 0:01:02 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.5111 (0.5512) acc1: 75.0000 (73.3570) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.2599 data: 0.1618 max mem: 3532 +Test: [300/517] eta: 0:00:59 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.4872 (0.5490) acc1: 77.8646 (73.5984) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.2549 data: 0.1587 max mem: 3532 +Test: [310/517] eta: 0:00:56 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.4443 (0.5449) acc1: 85.4167 (74.0521) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.2649 data: 0.1684 max mem: 3532 +Test: [320/517] eta: 0:00:53 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.4169 (0.5401) acc1: 89.3229 (74.5522) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.2649 data: 0.1691 max mem: 3532 +Test: [330/517] eta: 0:00:51 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.3565 (0.5342) acc1: 91.4062 (75.0952) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2599 data: 0.1642 max mem: 3532 +Test: [340/517] eta: 0:00:48 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.3237 (0.5294) acc1: 94.7917 (75.4651) acc5: 100.0000 (100.0000) acc1_neg: 70.573 acc5_neg: 100.000 time: 0.2649 data: 0.1691 max mem: 3532 +Test: [350/517] eta: 0:00:45 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.4403 (0.5285) acc1: 83.5938 (75.5935) acc5: 100.0000 (100.0000) acc1_neg: 64.583 acc5_neg: 100.000 time: 0.2699 data: 0.1736 max mem: 3532 +Test: [360/517] eta: 0:00:42 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.4489 (0.5253) acc1: 83.5938 (75.9017) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.2649 data: 0.1694 max mem: 3532 +Test: [370/517] eta: 0:00:40 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2963 (0.5181) acc1: 93.7500 (76.5169) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2649 data: 0.1708 max mem: 3532 +Test: [380/517] eta: 0:00:37 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2666 (0.5120) acc1: 98.6979 (77.0621) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2649 data: 0.1674 max mem: 3532 +Test: [390/517] eta: 0:00:34 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.2941 (0.5077) acc1: 94.7917 (77.4190) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2599 data: 0.1586 max mem: 3532 +Test: [400/517] eta: 0:00:31 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.3306 (0.5031) acc1: 94.7917 (77.8691) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2601 data: 0.1597 max mem: 3532 +Test: [410/517] eta: 0:00:29 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.3516 (0.5001) acc1: 94.7917 (78.2600) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2599 data: 0.1596 max mem: 3532 +Test: [420/517] eta: 0:00:26 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.3633 (0.4969) acc1: 92.7083 (78.5376) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.2647 data: 0.1652 max mem: 3532 +Test: [430/517] eta: 0:00:23 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.3626 (0.4941) acc1: 89.8438 (78.7649) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.2749 data: 0.1787 max mem: 3532 +Test: [440/517] eta: 0:00:20 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.3731 (0.4920) acc1: 88.8021 (78.9730) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.2748 data: 0.1780 max mem: 3533 +Test: [450/517] eta: 0:00:18 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.3629 (0.4890) acc1: 92.9688 (79.3156) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2649 data: 0.1673 max mem: 3533 +Test: [460/517] eta: 0:00:15 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.3441 (0.4858) acc1: 93.4896 (79.6095) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.3049 data: 0.2084 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.3277 (0.4824) acc1: 93.4896 (79.9275) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2749 data: 0.1806 max mem: 3533 +Test: [480/517] eta: 0:00:10 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.3277 (0.4798) acc1: 93.7500 (80.2100) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2292 data: 0.1339 max mem: 3533 +Test: [490/517] eta: 0:00:07 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.3573 (0.4775) acc1: 92.9688 (80.4237) acc5: 100.0000 (100.0000) acc1_neg: 71.615 acc5_neg: 100.000 time: 0.2240 data: 0.1303 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.5250 (0.4800) acc1: 73.9583 (80.1319) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.1856 data: 0.0959 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.5533 (0.4794) acc1: 70.3125 (80.1767) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1799 data: 0.0903 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 36.697 acc5_pos: 100.000 loss: 0.4191 (0.4780) acc1: 85.9375 (80.3207) acc5: 100.0000 (100.0000) acc1_neg: 94.700 acc5_neg: 100.000 time: 0.1800 data: 0.0861 max mem: 3533 +Test: Total time: 0:02:17 (0.2662 s / it) +* Overall Acc@1 80.321 Acc@5 100.000 + target == real Acc@1 31.968 Acc@5 100.000 + target == fake Acc@1 85.337 Acc@5 100.000 + loss 0.478 + PR-AUC 0.9334 ROC-AUC 0.6335 EER 40.28% (thr=0.6714) +Accuracy of the network on the 198427 test images: 80.3% +[rank0]:[W726 04:19:51.528255957 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 04:20:12.644000 1180214 torch/distributed/run.py:766] +W0726 04:20:12.644000 1180214 torch/distributed/run.py:766] ***************************************** +W0726 04:20:12.644000 1180214 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:20:12.644000 1180214 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 3): env:// +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:20:42.239436178 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:20:43.567588126 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:20:43.636800250 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:20:43.738596744 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:20:43.936349279 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:20:43.968403039 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:20:43.971262258 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:20:43.978826782 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n16_seed6_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed6_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:34:20 acc1_pos: 89.062 acc5_pos: 100.000 loss: 0.3831 (0.3831) acc1: 89.0625 (89.0625) acc5: 100.0000 (100.0000) time: 5.2980 data: 3.7227 max mem: 3531 +Test: [ 10/389] eta: 0:03:35 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.3831 (0.3649) acc1: 89.0625 (88.8731) acc5: 100.0000 (100.0000) time: 0.5695 data: 0.3449 max mem: 3531 +Test: [ 20/389] eta: 0:02:07 acc1_pos: 74.479 acc5_pos: 100.000 loss: 0.4448 (0.4081) acc1: 85.1562 (85.6771) acc5: 100.0000 (100.0000) time: 0.0979 data: 0.0037 max mem: 3531 +Test: [ 30/389] eta: 0:01:36 acc1_pos: 84.115 acc5_pos: 100.000 loss: 0.4536 (0.4238) acc1: 84.1146 (86.4331) acc5: 100.0000 (100.0000) time: 0.1055 data: 0.0078 max mem: 3531 +Test: [ 40/389] eta: 0:01:24 acc1_pos: 81.510 acc5_pos: 100.000 loss: 0.4435 (0.4366) acc1: 84.1146 (85.8105) acc5: 100.0000 (100.0000) time: 0.1311 data: 0.0353 max mem: 3531 +Test: [ 50/389] eta: 0:01:15 acc1_pos: 61.458 acc5_pos: 100.000 loss: 0.5076 (0.4542) acc1: 81.7708 (84.0686) acc5: 100.0000 (100.0000) time: 0.1455 data: 0.0532 max mem: 3531 +Test: [ 60/389] eta: 0:01:08 acc1_pos: 73.698 acc5_pos: 100.000 loss: 0.5265 (0.4614) acc1: 73.6979 (82.7100) acc5: 100.0000 (100.0000) time: 0.1371 data: 0.0475 max mem: 3531 +Test: [ 70/389] eta: 0:01:03 acc1_pos: 93.490 acc5_pos: 100.000 loss: 0.4088 (0.4468) acc1: 84.8958 (83.6121) acc5: 100.0000 (100.0000) time: 0.1437 data: 0.0530 max mem: 3531 +Test: [ 80/389] eta: 0:00:59 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.3390 (0.4442) acc1: 89.8438 (84.1532) acc5: 100.0000 (100.0000) time: 0.1529 data: 0.0619 max mem: 3531 +Test: [ 90/389] eta: 0:00:57 acc1_pos: 83.594 acc5_pos: 100.000 loss: 0.2850 (0.4259) acc1: 96.6146 (85.3508) acc5: 100.0000 (100.0000) time: 0.1695 data: 0.0778 max mem: 3531 +Test: [100/389] eta: 0:00:54 acc1_pos: 94.792 acc5_pos: 100.000 loss: 0.4179 (0.4346) acc1: 86.9792 (84.6535) acc5: 100.0000 (100.0000) time: 0.1686 data: 0.0749 max mem: 3531 +Test: [110/389] eta: 0:00:51 acc1_pos: 96.615 acc5_pos: 100.000 loss: 0.4618 (0.4385) acc1: 81.7708 (83.8119) acc5: 100.0000 (100.0000) time: 0.1499 data: 0.0545 max mem: 3531 +Test: [120/389] eta: 0:00:49 acc1_pos: 99.740 acc5_pos: 100.000 loss: 0.3847 (0.4353) acc1: 91.6667 (84.0436) acc5: 100.0000 (100.0000) time: 0.1562 data: 0.0611 max mem: 3531 +Test: [130/389] eta: 0:00:46 acc1_pos: 94.479 acc5_pos: 100.000 loss: 0.4719 (0.4758) acc1: 61.4583 (80.5105) acc5: 100.0000 (100.0000) acc1_neg: 60.156 acc5_neg: 100.000 time: 0.1575 data: 0.0638 max mem: 3531 +Test: [140/389] eta: 0:00:45 acc1_pos: 94.479 acc5_pos: 100.000 loss: 0.9900 (0.5185) acc1: 20.3125 (76.5219) acc5: 100.0000 (100.0000) acc1_neg: 0.521 acc5_neg: 100.000 time: 0.1686 data: 0.0706 max mem: 3531 +Test: [150/389] eta: 0:00:42 acc1_pos: 94.479 acc5_pos: 100.000 loss: 0.9943 (0.5573) acc1: 20.3125 (73.1771) acc5: 100.0000 (100.0000) acc1_neg: 69.531 acc5_neg: 100.000 time: 0.1723 data: 0.0734 max mem: 3531 +Test: [160/389] eta: 0:00:41 acc1_pos: 94.479 acc5_pos: 100.000 loss: 0.9816 (0.5778) acc1: 20.8333 (70.9223) acc5: 100.0000 (100.0000) acc1_neg: 31.510 acc5_neg: 100.000 time: 0.1649 data: 0.0683 max mem: 3531 +Test: [170/389] eta: 0:00:38 acc1_pos: 94.479 acc5_pos: 100.000 loss: 0.7253 (0.5937) acc1: 47.6562 (69.3515) acc5: 100.0000 (100.0000) acc1_neg: 22.396 acc5_neg: 100.000 time: 0.1599 data: 0.0633 max mem: 3531 +Test: [180/389] eta: 0:00:37 acc1_pos: 94.479 acc5_pos: 100.000 loss: 0.8733 (0.6141) acc1: 27.6042 (66.9774) acc5: 100.0000 (100.0000) acc1_neg: 16.927 acc5_neg: 100.000 time: 0.1651 data: 0.0694 max mem: 3531 +Test: [190/389] eta: 0:00:35 acc1_pos: 94.479 acc5_pos: 100.000 loss: 0.9144 (0.6387) acc1: 27.6042 (64.8983) acc5: 100.0000 (100.0000) acc1_neg: 12.760 acc5_neg: 100.000 time: 0.1753 data: 0.0784 max mem: 3531 +Test: [200/389] eta: 0:00:33 acc1_pos: 94.479 acc5_pos: 100.000 loss: 1.1084 (0.6635) acc1: 12.7604 (62.6490) acc5: 100.0000 (100.0000) acc1_neg: 10.156 acc5_neg: 100.000 time: 0.1556 data: 0.0595 max mem: 3531 +Test: [210/389] eta: 0:00:31 acc1_pos: 94.479 acc5_pos: 100.000 loss: 1.0337 (0.6675) acc1: 14.8438 (62.0557) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1550 data: 0.0616 max mem: 3532 +Test: [220/389] eta: 0:00:29 acc1_pos: 94.479 acc5_pos: 100.000 loss: 0.5844 (0.6622) acc1: 61.9792 (62.4623) acc5: 100.0000 (100.0000) acc1_neg: 14.583 acc5_neg: 100.000 time: 0.1701 data: 0.0768 max mem: 3532 +Test: [230/389] eta: 0:00:27 acc1_pos: 94.479 acc5_pos: 100.000 loss: 0.8724 (0.6748) acc1: 23.1771 (61.0536) acc5: 100.0000 (100.0000) acc1_neg: 26.562 acc5_neg: 100.000 time: 0.1762 data: 0.0832 max mem: 3532 +Test: [240/389] eta: 0:00:25 acc1_pos: 94.479 acc5_pos: 100.000 loss: 0.9663 (0.6854) acc1: 22.3958 (59.8407) acc5: 100.0000 (100.0000) acc1_neg: 36.198 acc5_neg: 100.000 time: 0.1609 data: 0.0692 max mem: 3532 +Test: [250/389] eta: 0:00:24 acc1_pos: 94.479 acc5_pos: 100.000 loss: 0.8514 (0.6892) acc1: 36.1979 (59.1582) acc5: 100.0000 (100.0000) acc1_neg: 47.396 acc5_neg: 100.000 time: 0.1631 data: 0.0715 max mem: 3532 +Test: [260/389] eta: 0:00:22 acc1_pos: 94.479 acc5_pos: 100.000 loss: 0.7641 (0.7014) acc1: 42.9688 (58.0061) acc5: 100.0000 (100.0000) acc1_neg: 11.719 acc5_neg: 100.000 time: 0.1830 data: 0.0895 max mem: 3532 +Test: [270/389] eta: 0:00:20 acc1_pos: 94.479 acc5_pos: 100.000 loss: 1.1294 (0.7166) acc1: 11.7188 (56.4412) acc5: 100.0000 (100.0000) acc1_neg: 41.146 acc5_neg: 100.000 time: 0.1699 data: 0.0732 max mem: 3532 +Test: [280/389] eta: 0:00:19 acc1_pos: 94.479 acc5_pos: 100.000 loss: 0.9911 (0.7256) acc1: 13.0208 (55.4752) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1797 data: 0.0807 max mem: 3532 +Test: [290/389] eta: 0:00:17 acc1_pos: 94.479 acc5_pos: 100.000 loss: 0.9191 (0.7304) acc1: 27.0833 (55.0150) acc5: 100.0000 (100.0000) acc1_neg: 66.406 acc5_neg: 100.000 time: 0.1699 data: 0.0719 max mem: 3532 +Test: [300/389] eta: 0:00:15 acc1_pos: 94.479 acc5_pos: 100.000 loss: 0.7800 (0.7334) acc1: 42.1875 (54.5629) acc5: 100.0000 (100.0000) acc1_neg: 39.323 acc5_neg: 100.000 time: 0.1549 data: 0.0593 max mem: 3532 +Test: [310/389] eta: 0:00:13 acc1_pos: 94.479 acc5_pos: 100.000 loss: 0.9381 (0.7454) acc1: 25.0000 (53.4625) acc5: 100.0000 (100.0000) acc1_neg: 8.854 acc5_neg: 100.000 time: 0.1599 data: 0.0642 max mem: 3532 +Test: [320/389] eta: 0:00:11 acc1_pos: 94.479 acc5_pos: 100.000 loss: 1.0112 (0.7510) acc1: 22.3958 (52.7616) acc5: 100.0000 (100.0000) acc1_neg: 54.427 acc5_neg: 100.000 time: 0.1599 data: 0.0617 max mem: 3532 +Test: [330/389] eta: 0:00:10 acc1_pos: 94.479 acc5_pos: 100.000 loss: 1.1569 (0.7762) acc1: 7.8125 (51.3155) acc5: 100.0000 (100.0000) acc1_neg: 3.906 acc5_neg: 100.000 time: 0.1599 data: 0.0621 max mem: 3532 +Test: [340/389] eta: 0:00:08 acc1_pos: 94.479 acc5_pos: 100.000 loss: 0.9516 (0.7760) acc1: 26.0417 (51.3296) acc5: 100.0000 (100.0000) acc1_neg: 35.156 acc5_neg: 100.000 time: 0.1539 data: 0.0574 max mem: 3532 +Test: [350/389] eta: 0:00:06 acc1_pos: 94.479 acc5_pos: 100.000 loss: 0.8783 (0.7874) acc1: 32.5521 (50.3799) acc5: 100.0000 (100.0000) acc1_neg: 7.292 acc5_neg: 100.000 time: 0.1649 data: 0.0704 max mem: 3532 +Test: [360/389] eta: 0:00:04 acc1_pos: 94.479 acc5_pos: 100.000 loss: 0.8610 (0.7912) acc1: 30.2083 (49.9315) acc5: 100.0000 (100.0000) acc1_neg: 41.927 acc5_neg: 100.000 time: 0.1859 data: 0.0914 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 94.479 acc5_pos: 100.000 loss: 0.8383 (0.7908) acc1: 41.9271 (49.9137) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.1929 data: 0.0986 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 94.479 acc5_pos: 100.000 loss: 0.7064 (0.7958) acc1: 48.1771 (49.8168) acc5: 100.0000 (100.0000) acc1_neg: 31.510 acc5_neg: 100.000 time: 0.1748 data: 0.0842 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 94.479 acc5_pos: 100.000 loss: 0.8670 (0.7976) acc1: 40.3646 (49.5548) acc5: 100.0000 (100.0000) acc1_neg: 42.896 acc5_neg: 100.000 time: 0.1306 data: 0.0371 max mem: 3532 +Test: Total time: 0:01:06 (0.1717 s / it) +* Overall Acc@1 49.555 Acc@5 100.000 + target == real Acc@1 84.192 Acc@5 100.000 + target == fake Acc@1 33.454 Acc@5 100.000 + loss 0.798 + PR-AUC 0.7881 ROC-AUC 0.6571 EER 38.81% (thr=0.3425) +Accuracy of the network on the 149358 test images: 49.6% +[rank0]:[W726 04:22:08.626123228 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:22:27.667000 1180773 torch/distributed/run.py:766] +W0726 04:22:27.667000 1180773 torch/distributed/run.py:766] ***************************************** +W0726 04:22:27.667000 1180773 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:22:27.667000 1180773 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:22:56.896646372 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:22:56.219817158 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:22:57.762144086 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:22:57.837737057 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:22:57.073665019 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 1): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:22:57.169902001 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W726 04:22:57.170023423 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:22:57.322470307 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n16_seed6_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed6_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:28:57 acc1_pos: 86.458 acc5_pos: 100.000 loss: 0.3737 (0.3737) acc1: 86.4583 (86.4583) acc5: 100.0000 (100.0000) time: 5.2170 data: 2.8406 max mem: 3531 +Test: [ 10/333] eta: 0:03:00 acc1_pos: 88.021 acc5_pos: 100.000 loss: 0.2873 (0.2862) acc1: 88.0208 (90.1278) acc5: 100.0000 (100.0000) time: 0.5583 data: 0.2608 max mem: 3531 +Test: [ 20/333] eta: 0:02:01 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.2590 (0.2707) acc1: 94.0104 (91.9643) acc5: 100.0000 (100.0000) time: 0.1455 data: 0.0534 max mem: 3531 +Test: [ 30/333] eta: 0:01:42 acc1_pos: 65.625 acc5_pos: 100.000 loss: 0.3193 (0.3777) acc1: 84.6354 (83.8962) acc5: 100.0000 (100.0000) time: 0.2192 data: 0.1258 max mem: 3531 +Test: [ 40/333] eta: 0:01:30 acc1_pos: 70.052 acc5_pos: 100.000 loss: 0.6204 (0.4412) acc1: 64.0625 (79.2619) acc5: 100.0000 (100.0000) time: 0.2242 data: 0.1299 max mem: 3531 +Test: [ 50/333] eta: 0:01:22 acc1_pos: 65.625 acc5_pos: 100.000 loss: 0.6250 (0.4772) acc1: 64.3229 (76.4757) acc5: 100.0000 (100.0000) time: 0.2147 data: 0.1190 max mem: 3531 +Test: [ 60/333] eta: 0:01:15 acc1_pos: 59.635 acc5_pos: 100.000 loss: 0.6319 (0.5081) acc1: 64.0625 (74.2273) acc5: 100.0000 (100.0000) time: 0.2163 data: 0.1228 max mem: 3531 +Test: [ 70/333] eta: 0:01:10 acc1_pos: 59.375 acc5_pos: 100.000 loss: 0.6438 (0.5294) acc1: 61.9792 (72.5756) acc5: 100.0000 (100.0000) time: 0.2131 data: 0.1217 max mem: 3531 +Test: [ 80/333] eta: 0:01:06 acc1_pos: 70.573 acc5_pos: 100.000 loss: 0.6450 (0.5422) acc1: 61.9792 (71.4249) acc5: 100.0000 (100.0000) time: 0.2133 data: 0.1212 max mem: 3531 +Test: [ 90/333] eta: 0:01:01 acc1_pos: 78.385 acc5_pos: 100.000 loss: 0.5658 (0.5405) acc1: 69.5312 (71.8779) acc5: 100.0000 (100.0000) time: 0.1932 data: 0.1009 max mem: 3531 +Test: [100/333] eta: 0:00:56 acc1_pos: 73.698 acc5_pos: 100.000 loss: 0.5251 (0.5445) acc1: 73.6979 (71.5733) acc5: 100.0000 (100.0000) time: 0.1739 data: 0.0790 max mem: 3531 +Test: [110/333] eta: 0:00:53 acc1_pos: 71.875 acc5_pos: 100.000 loss: 0.5755 (0.5486) acc1: 69.2708 (71.2416) acc5: 100.0000 (100.0000) time: 0.1842 data: 0.0901 max mem: 3531 +Test: [120/333] eta: 0:00:50 acc1_pos: 72.396 acc5_pos: 100.000 loss: 0.5390 (0.5488) acc1: 71.8750 (71.2939) acc5: 100.0000 (100.0000) time: 0.1970 data: 0.1047 max mem: 3531 +Test: [130/333] eta: 0:00:47 acc1_pos: 85.938 acc5_pos: 100.000 loss: 0.5144 (0.5482) acc1: 74.4792 (71.5172) acc5: 100.0000 (100.0000) time: 0.1973 data: 0.1035 max mem: 3531 +Test: [140/333] eta: 0:00:44 acc1_pos: 83.594 acc5_pos: 100.000 loss: 0.5218 (0.5460) acc1: 74.4792 (71.7272) acc5: 100.0000 (100.0000) time: 0.1882 data: 0.0955 max mem: 3531 +Test: [150/333] eta: 0:00:41 acc1_pos: 81.250 acc5_pos: 100.000 loss: 0.5218 (0.5474) acc1: 72.9167 (71.6715) acc5: 100.0000 (100.0000) time: 0.1872 data: 0.0955 max mem: 3531 +Test: [160/333] eta: 0:00:39 acc1_pos: 63.816 acc5_pos: 100.000 loss: 0.5902 (0.5620) acc1: 66.9271 (70.6861) acc5: 100.0000 (100.0000) acc1_neg: 34.115 acc5_neg: 100.000 time: 0.1960 data: 0.1044 max mem: 3531 +Test: [170/333] eta: 0:00:36 acc1_pos: 63.816 acc5_pos: 100.000 loss: 1.0195 (0.5913) acc1: 34.1146 (68.4485) acc5: 100.0000 (100.0000) acc1_neg: 37.500 acc5_neg: 100.000 time: 0.1875 data: 0.0948 max mem: 3531 +Test: [180/333] eta: 0:00:33 acc1_pos: 63.816 acc5_pos: 100.000 loss: 1.0179 (0.6133) acc1: 33.5938 (66.6767) acc5: 100.0000 (100.0000) acc1_neg: 47.135 acc5_neg: 100.000 time: 0.1847 data: 0.0911 max mem: 3531 +Test: [190/333] eta: 0:00:31 acc1_pos: 63.816 acc5_pos: 100.000 loss: 0.9125 (0.6240) acc1: 40.3646 (65.7423) acc5: 100.0000 (100.0000) acc1_neg: 51.042 acc5_neg: 100.000 time: 0.1902 data: 0.0978 max mem: 3531 +Test: [200/333] eta: 0:00:28 acc1_pos: 63.816 acc5_pos: 100.000 loss: 0.8368 (0.6369) acc1: 46.6146 (64.7479) acc5: 100.0000 (100.0000) acc1_neg: 42.448 acc5_neg: 100.000 time: 0.1861 data: 0.0920 max mem: 3531 +Test: [210/333] eta: 0:00:26 acc1_pos: 63.816 acc5_pos: 100.000 loss: 0.9418 (0.6524) acc1: 42.1875 (63.6034) acc5: 100.0000 (100.0000) acc1_neg: 34.115 acc5_neg: 100.000 time: 0.1823 data: 0.0862 max mem: 3532 +Test: [220/333] eta: 0:00:24 acc1_pos: 63.816 acc5_pos: 100.000 loss: 0.9626 (0.6690) acc1: 38.0208 (62.3350) acc5: 100.0000 (100.0000) acc1_neg: 38.021 acc5_neg: 100.000 time: 0.1826 data: 0.0898 max mem: 3532 +Test: [230/333] eta: 0:00:22 acc1_pos: 63.816 acc5_pos: 100.000 loss: 0.9211 (0.6771) acc1: 38.0208 (61.5925) acc5: 100.0000 (100.0000) acc1_neg: 37.760 acc5_neg: 100.000 time: 0.1974 data: 0.1046 max mem: 3532 +Test: [240/333] eta: 0:00:19 acc1_pos: 63.816 acc5_pos: 100.000 loss: 0.8990 (0.6884) acc1: 39.0625 (60.5355) acc5: 100.0000 (100.0000) acc1_neg: 40.885 acc5_neg: 100.000 time: 0.1997 data: 0.1056 max mem: 3532 +Test: [250/333] eta: 0:00:17 acc1_pos: 63.816 acc5_pos: 100.000 loss: 0.9118 (0.6967) acc1: 40.1042 (59.8429) acc5: 100.0000 (100.0000) acc1_neg: 40.104 acc5_neg: 100.000 time: 0.1801 data: 0.0884 max mem: 3532 +Test: [260/333] eta: 0:00:15 acc1_pos: 63.816 acc5_pos: 100.000 loss: 0.8712 (0.7027) acc1: 43.7500 (59.3540) acc5: 100.0000 (100.0000) acc1_neg: 50.521 acc5_neg: 100.000 time: 0.1999 data: 0.1080 max mem: 3532 +Test: [270/333] eta: 0:00:13 acc1_pos: 63.816 acc5_pos: 100.000 loss: 0.8712 (0.7091) acc1: 45.8333 (58.8417) acc5: 100.0000 (100.0000) acc1_neg: 45.833 acc5_neg: 100.000 time: 0.2013 data: 0.1069 max mem: 3532 +Test: [280/333] eta: 0:00:11 acc1_pos: 63.816 acc5_pos: 100.000 loss: 0.8816 (0.7184) acc1: 42.7083 (58.0803) acc5: 100.0000 (100.0000) acc1_neg: 42.448 acc5_neg: 100.000 time: 0.1821 data: 0.0894 max mem: 3532 +Test: [290/333] eta: 0:00:08 acc1_pos: 63.816 acc5_pos: 100.000 loss: 1.0537 (0.7303) acc1: 34.3750 (57.1789) acc5: 100.0000 (100.0000) acc1_neg: 34.635 acc5_neg: 100.000 time: 0.1894 data: 0.0980 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 63.816 acc5_pos: 100.000 loss: 1.0783 (0.7421) acc1: 30.9896 (56.2993) acc5: 100.0000 (100.0000) acc1_neg: 29.427 acc5_neg: 100.000 time: 0.1784 data: 0.0860 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 63.816 acc5_pos: 100.000 loss: 0.9846 (0.7442) acc1: 33.8542 (56.1303) acc5: 100.0000 (100.0000) acc1_neg: 51.042 acc5_neg: 100.000 time: 0.1641 data: 0.0727 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 63.816 acc5_pos: 100.000 loss: 0.7902 (0.7472) acc1: 51.0417 (55.8200) acc5: 100.0000 (100.0000) acc1_neg: 32.292 acc5_neg: 100.000 time: 0.1461 data: 0.0553 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 63.816 acc5_pos: 100.000 loss: 1.0934 (0.7593) acc1: 32.5521 (55.0699) acc5: 100.0000 (100.0000) acc1_neg: 32.552 acc5_neg: 100.000 time: 0.1288 data: 0.0387 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 63.816 acc5_pos: 100.000 loss: 1.1345 (0.7617) acc1: 32.0312 (54.9299) acc5: 100.0000 (100.0000) acc1_neg: 29.577 acc5_neg: 100.000 time: 0.1294 data: 0.0360 max mem: 3532 +Test: Total time: 0:01:07 (0.2023 s / it) +* Overall Acc@1 54.930 Acc@5 100.000 + target == real Acc@1 71.770 Acc@5 100.000 + target == fake Acc@1 39.939 Acc@5 100.000 + loss 0.762 + PR-AUC 0.6074 ROC-AUC 0.6005 EER 43.07% (thr=0.3811) +Accuracy of the network on the 127843 test images: 54.9% +[rank0]:[W726 04:24:16.682354824 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:24:34.936000 1181332 torch/distributed/run.py:766] +W0726 04:24:34.936000 1181332 torch/distributed/run.py:766] ***************************************** +W0726 04:24:34.936000 1181332 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:24:34.936000 1181332 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:25:04.601333278 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:25:04.832040779 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:25:04.883676458 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:25:04.924348355 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:25:04.211380126 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:25:04.247777089 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank4]:[W726 04:25:04.248218830 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:25:04.294019658 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n16_seed6_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed6_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:27:43 acc1_pos: 97.917 acc5_pos: 100.000 loss: 0.1606 (0.1606) acc1: 97.9167 (97.9167) acc5: 100.0000 (100.0000) time: 7.0786 data: 5.4327 max mem: 3531 +Test: [ 10/235] eta: 0:02:53 acc1_pos: 97.396 acc5_pos: 100.000 loss: 0.1707 (0.1715) acc1: 97.6562 (97.2538) acc5: 100.0000 (100.0000) time: 0.7690 data: 0.5310 max mem: 3531 +Test: [ 20/235] eta: 0:01:36 acc1_pos: 97.656 acc5_pos: 100.000 loss: 0.1700 (0.1690) acc1: 97.6562 (97.3710) acc5: 100.0000 (100.0000) time: 0.1190 data: 0.0207 max mem: 3531 +Test: [ 30/235] eta: 0:01:22 acc1_pos: 97.656 acc5_pos: 100.000 loss: 0.1653 (0.1683) acc1: 97.6562 (97.4294) acc5: 100.0000 (100.0000) time: 0.1999 data: 0.1043 max mem: 3531 +Test: [ 40/235] eta: 0:01:09 acc1_pos: 97.656 acc5_pos: 100.000 loss: 0.1659 (0.1686) acc1: 97.6562 (97.4530) acc5: 100.0000 (100.0000) time: 0.2549 data: 0.1618 max mem: 3531 +Test: [ 50/235] eta: 0:01:00 acc1_pos: 95.833 acc5_pos: 100.000 loss: 0.1697 (0.1696) acc1: 97.3958 (97.4060) acc5: 100.0000 (100.0000) time: 0.2049 data: 0.1113 max mem: 3531 +Test: [ 60/235] eta: 0:00:52 acc1_pos: 97.656 acc5_pos: 100.000 loss: 0.1720 (0.1708) acc1: 97.1354 (97.3233) acc5: 100.0000 (100.0000) time: 0.1850 data: 0.0879 max mem: 3531 +Test: [ 70/235] eta: 0:00:49 acc1_pos: 96.923 acc5_pos: 100.000 loss: 0.1975 (0.3551) acc1: 95.8333 (86.8985) acc5: 100.0000 (100.0000) acc1_neg: 12.240 acc5_neg: 100.000 time: 0.2407 data: 0.1447 max mem: 3531 +Test: [ 80/235] eta: 0:00:44 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.6298 (0.5143) acc1: 13.2812 (77.8003) acc5: 100.0000 (100.0000) acc1_neg: 10.417 acc5_neg: 100.000 time: 0.2501 data: 0.1583 max mem: 3531 +Test: [ 90/235] eta: 0:00:38 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.6520 (0.6397) acc1: 13.2812 (70.7732) acc5: 100.0000 (100.0000) acc1_neg: 14.323 acc5_neg: 100.000 time: 0.1549 data: 0.0616 max mem: 3531 +Test: [100/235] eta: 0:00:35 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.6366 (0.7384) acc1: 13.2812 (65.1712) acc5: 100.0000 (100.0000) acc1_neg: 12.240 acc5_neg: 100.000 time: 0.1545 data: 0.0610 max mem: 3531 +Test: [110/235] eta: 0:00:32 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.6297 (0.8184) acc1: 14.3229 (60.6513) acc5: 100.0000 (100.0000) acc1_neg: 15.365 acc5_neg: 100.000 time: 0.2053 data: 0.1101 max mem: 3531 +Test: [120/235] eta: 0:00:29 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.6626 (0.9042) acc1: 11.4583 (56.1445) acc5: 100.0000 (100.0000) acc1_neg: 4.427 acc5_neg: 100.000 time: 0.2207 data: 0.1261 max mem: 3531 +Test: [130/235] eta: 0:00:26 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.8723 (0.9783) acc1: 5.4688 (52.3159) acc5: 100.0000 (100.0000) acc1_neg: 4.427 acc5_neg: 100.000 time: 0.2186 data: 0.1275 max mem: 3531 +Test: [140/235] eta: 0:00:23 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.8955 (1.0428) acc1: 5.7292 (49.0211) acc5: 100.0000 (100.0000) acc1_neg: 5.729 acc5_neg: 100.000 time: 0.2213 data: 0.1299 max mem: 3531 +Test: [150/235] eta: 0:00:21 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.8695 (1.0973) acc1: 5.9896 (46.1852) acc5: 100.0000 (100.0000) acc1_neg: 6.510 acc5_neg: 100.000 time: 0.2190 data: 0.1268 max mem: 3531 +Test: [160/235] eta: 0:00:18 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.8639 (1.1451) acc1: 5.9896 (43.6982) acc5: 100.0000 (100.0000) acc1_neg: 8.073 acc5_neg: 100.000 time: 0.2173 data: 0.1255 max mem: 3531 +Test: [170/235] eta: 0:00:15 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.8378 (1.1786) acc1: 6.7708 (41.9926) acc5: 100.0000 (100.0000) acc1_neg: 16.406 acc5_neg: 100.000 time: 0.2299 data: 0.1376 max mem: 3531 +Test: [180/235] eta: 0:00:13 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.7071 (1.2071) acc1: 15.3646 (40.5790) acc5: 100.0000 (100.0000) acc1_neg: 14.844 acc5_neg: 100.000 time: 0.2289 data: 0.1364 max mem: 3531 +Test: [190/235] eta: 0:00:10 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.6810 (1.2323) acc1: 16.4062 (39.2820) acc5: 100.0000 (100.0000) acc1_neg: 16.927 acc5_neg: 100.000 time: 0.2125 data: 0.1209 max mem: 3531 +Test: [200/235] eta: 0:00:08 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.6938 (1.2560) acc1: 15.1042 (38.0623) acc5: 100.0000 (100.0000) acc1_neg: 15.104 acc5_neg: 100.000 time: 0.1783 data: 0.0877 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.6993 (1.2765) acc1: 14.8438 (37.0298) acc5: 100.0000 (100.0000) acc1_neg: 15.625 acc5_neg: 100.000 time: 0.1525 data: 0.0620 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.6894 (1.2950) acc1: 15.8854 (36.0801) acc5: 100.0000 (100.0000) acc1_neg: 15.365 acc5_neg: 100.000 time: 0.1355 data: 0.0454 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.6839 (1.3114) acc1: 16.4062 (35.2487) acc5: 100.0000 (100.0000) acc1_neg: 13.021 acc5_neg: 100.000 time: 0.1228 data: 0.0332 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.6951 (1.3183) acc1: 16.4062 (34.9882) acc5: 100.0000 (100.0000) acc1_neg: 21.739 acc5_neg: 100.000 time: 0.1226 data: 0.0360 max mem: 3532 +Test: Total time: 0:00:52 (0.2224 s / it) +* Overall Acc@1 34.988 Acc@5 100.000 + target == real Acc@1 97.311 Acc@5 100.000 + target == fake Acc@1 12.447 Acc@5 100.000 + loss 1.318 + PR-AUC 0.8442 ROC-AUC 0.6628 EER 38.58% (thr=0.1292) +Accuracy of the network on the 89879 test images: 35.0% +[rank0]:[W726 04:26:04.072179470 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:26:23.987000 1181882 torch/distributed/run.py:766] +W0726 04:26:23.987000 1181882 torch/distributed/run.py:766] ***************************************** +W0726 04:26:23.987000 1181882 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:26:23.987000 1181882 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:26:51.103594165 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:26:51.444513831 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:26:52.547657999 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:26:52.686851099 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:26:52.783317086 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:26:52.972852022 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:26:52.055307368 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:26:52.099795156 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n16_seed6_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed6_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:30:55 acc1_pos: 73.958 acc5_pos: 100.000 loss: 0.5388 (0.5388) acc1: 73.9583 (73.9583) acc5: 100.0000 (100.0000) time: 6.8970 data: 5.1345 max mem: 3531 +Test: [ 10/269] eta: 0:03:06 acc1_pos: 97.135 acc5_pos: 100.000 loss: 0.4178 (0.4041) acc1: 84.8958 (86.0322) acc5: 100.0000 (100.0000) time: 0.7201 data: 0.4770 max mem: 3531 +Test: [ 20/269] eta: 0:01:45 acc1_pos: 94.531 acc5_pos: 100.000 loss: 0.4064 (0.3982) acc1: 87.2396 (87.4876) acc5: 100.0000 (100.0000) time: 0.1011 data: 0.0058 max mem: 3531 +Test: [ 30/269] eta: 0:01:32 acc1_pos: 88.281 acc5_pos: 100.000 loss: 0.4001 (0.4032) acc1: 88.0208 (86.9288) acc5: 100.0000 (100.0000) time: 0.2057 data: 0.1091 max mem: 3531 +Test: [ 40/269] eta: 0:01:21 acc1_pos: 94.828 acc5_pos: 100.000 loss: 0.5400 (0.5024) acc1: 70.3125 (77.4898) acc5: 100.0000 (100.0000) acc1_neg: 41.927 acc5_neg: 100.000 time: 0.2822 data: 0.1894 max mem: 3531 +Test: [ 50/269] eta: 0:01:12 acc1_pos: 94.828 acc5_pos: 100.000 loss: 0.9424 (0.6210) acc1: 34.8958 (67.0139) acc5: 100.0000 (100.0000) acc1_neg: 13.542 acc5_neg: 100.000 time: 0.2391 data: 0.1476 max mem: 3531 +Test: [ 60/269] eta: 0:01:04 acc1_pos: 94.828 acc5_pos: 100.000 loss: 1.1439 (0.7098) acc1: 17.7083 (59.2597) acc5: 100.0000 (100.0000) acc1_neg: 16.406 acc5_neg: 100.000 time: 0.2095 data: 0.1179 max mem: 3531 +Test: [ 70/269] eta: 0:00:59 acc1_pos: 94.828 acc5_pos: 100.000 loss: 1.1439 (0.7644) acc1: 18.2292 (53.9063) acc5: 100.0000 (100.0000) acc1_neg: 19.531 acc5_neg: 100.000 time: 0.2133 data: 0.1218 max mem: 3531 +Test: [ 80/269] eta: 0:00:54 acc1_pos: 94.828 acc5_pos: 100.000 loss: 1.0504 (0.7988) acc1: 23.1771 (50.2861) acc5: 100.0000 (100.0000) acc1_neg: 24.219 acc5_neg: 100.000 time: 0.2326 data: 0.1399 max mem: 3531 +Test: [ 90/269] eta: 0:00:49 acc1_pos: 94.828 acc5_pos: 100.000 loss: 1.0972 (0.8503) acc1: 22.6562 (46.5030) acc5: 100.0000 (100.0000) acc1_neg: 21.094 acc5_neg: 100.000 time: 0.2165 data: 0.1214 max mem: 3531 +Test: [100/269] eta: 0:00:46 acc1_pos: 94.828 acc5_pos: 100.000 loss: 1.2006 (0.8901) acc1: 14.5833 (43.3864) acc5: 100.0000 (100.0000) acc1_neg: 4.427 acc5_neg: 100.000 time: 0.2056 data: 0.1108 max mem: 3531 +Test: [110/269] eta: 0:00:42 acc1_pos: 94.828 acc5_pos: 100.000 loss: 1.1639 (0.9105) acc1: 15.3646 (41.0942) acc5: 100.0000 (100.0000) acc1_neg: 15.365 acc5_neg: 100.000 time: 0.2149 data: 0.1214 max mem: 3531 +Test: [120/269] eta: 0:00:39 acc1_pos: 94.828 acc5_pos: 100.000 loss: 1.1065 (0.9249) acc1: 20.0521 (39.9320) acc5: 100.0000 (100.0000) acc1_neg: 15.625 acc5_neg: 100.000 time: 0.2115 data: 0.1179 max mem: 3531 +Test: [130/269] eta: 0:00:35 acc1_pos: 94.828 acc5_pos: 100.000 loss: 1.1142 (0.9481) acc1: 22.3958 (38.3469) acc5: 100.0000 (100.0000) acc1_neg: 16.406 acc5_neg: 100.000 time: 0.2029 data: 0.1103 max mem: 3531 +Test: [140/269] eta: 0:00:32 acc1_pos: 94.828 acc5_pos: 100.000 loss: 1.1819 (0.9636) acc1: 16.4062 (36.7335) acc5: 100.0000 (100.0000) acc1_neg: 12.500 acc5_neg: 100.000 time: 0.2030 data: 0.1117 max mem: 3531 +Test: [150/269] eta: 0:00:29 acc1_pos: 94.828 acc5_pos: 100.000 loss: 1.1159 (0.9638) acc1: 17.9688 (36.7394) acc5: 100.0000 (100.0000) acc1_neg: 49.479 acc5_neg: 100.000 time: 0.2119 data: 0.1195 max mem: 3531 +Test: [160/269] eta: 0:00:26 acc1_pos: 94.828 acc5_pos: 100.000 loss: 0.9770 (0.9701) acc1: 33.0729 (36.2561) acc5: 100.0000 (100.0000) acc1_neg: 7.552 acc5_neg: 100.000 time: 0.1958 data: 0.1028 max mem: 3531 +Test: [170/269] eta: 0:00:23 acc1_pos: 94.828 acc5_pos: 100.000 loss: 1.2921 (0.9913) acc1: 9.3750 (34.5532) acc5: 100.0000 (100.0000) acc1_neg: 10.938 acc5_neg: 100.000 time: 0.1701 data: 0.0749 max mem: 3531 +Test: [180/269] eta: 0:00:21 acc1_pos: 94.828 acc5_pos: 100.000 loss: 1.2482 (0.9991) acc1: 10.4167 (33.7376) acc5: 100.0000 (100.0000) acc1_neg: 12.240 acc5_neg: 100.000 time: 0.1763 data: 0.0811 max mem: 3531 +Test: [190/269] eta: 0:00:18 acc1_pos: 94.828 acc5_pos: 100.000 loss: 1.1176 (1.0096) acc1: 17.7083 (32.8561) acc5: 100.0000 (100.0000) acc1_neg: 11.198 acc5_neg: 100.000 time: 0.1801 data: 0.0869 max mem: 3531 +Test: [200/269] eta: 0:00:15 acc1_pos: 94.828 acc5_pos: 100.000 loss: 1.2917 (1.0250) acc1: 11.1979 (31.7916) acc5: 100.0000 (100.0000) acc1_neg: 15.625 acc5_neg: 100.000 time: 0.1649 data: 0.0731 max mem: 3531 +Test: [210/269] eta: 0:00:13 acc1_pos: 94.828 acc5_pos: 100.000 loss: 1.1989 (1.0186) acc1: 14.5833 (32.2559) acc5: 100.0000 (100.0000) acc1_neg: 43.490 acc5_neg: 100.000 time: 0.1712 data: 0.0802 max mem: 3532 +Test: [220/269] eta: 0:00:11 acc1_pos: 94.828 acc5_pos: 100.000 loss: 0.8291 (1.0120) acc1: 42.7083 (32.6464) acc5: 100.0000 (100.0000) acc1_neg: 53.646 acc5_neg: 100.000 time: 0.1831 data: 0.0896 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 94.828 acc5_pos: 100.000 loss: 0.8291 (1.0054) acc1: 40.1042 (33.0864) acc5: 100.0000 (100.0000) acc1_neg: 51.302 acc5_neg: 100.000 time: 0.1694 data: 0.0769 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 94.828 acc5_pos: 100.000 loss: 1.0564 (1.0159) acc1: 25.2604 (32.2441) acc5: 100.0000 (100.0000) acc1_neg: 18.750 acc5_neg: 100.000 time: 0.1491 data: 0.0587 max mem: 3532 +Test: [250/269] eta: 0:00:04 acc1_pos: 94.828 acc5_pos: 100.000 loss: 1.2302 (1.0266) acc1: 9.6354 (31.7117) acc5: 100.0000 (100.0000) acc1_neg: 4.427 acc5_neg: 100.000 time: 0.1334 data: 0.0432 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 94.828 acc5_pos: 100.000 loss: 1.4292 (1.0402) acc1: 5.7292 (30.9008) acc5: 100.0000 (100.0000) acc1_neg: 30.729 acc5_neg: 100.000 time: 0.1205 data: 0.0312 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 94.828 acc5_pos: 100.000 loss: 1.2164 (1.0442) acc1: 10.4167 (30.3719) acc5: 100.0000 (100.0000) acc1_neg: 15.486 acc5_neg: 100.000 time: 0.1119 data: 0.0191 max mem: 3532 +Test: Total time: 0:00:56 (0.2107 s / it) +* Overall Acc@1 30.372 Acc@5 100.000 + target == real Acc@1 86.964 Acc@5 100.000 + target == fake Acc@1 22.649 Acc@5 100.000 + loss 1.044 + PR-AUC 0.9117 ROC-AUC 0.5948 EER 43.41% (thr=0.2974) +Accuracy of the network on the 103293 test images: 30.4% +[rank0]:[W726 04:28:01.756872117 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T04:28:07Z] eval_done tag=fixed500k_fake_n16_seed6_resampled +[2026-07-26T04:28:08Z] eval_start tag=fixed500k_real_n8_seed7_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed7_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T04:28:08Z +hostname=n116-041-153 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed7_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n8_seed7_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 04:28:21.458000 1182457 torch/distributed/run.py:766] +W0726 04:28:21.458000 1182457 torch/distributed/run.py:766] ***************************************** +W0726 04:28:21.458000 1182457 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:28:21.458000 1182457 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:28:50.436279022 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:28:51.729435368 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:28:51.217934305 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:28:51.241820732 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:28:51.297547292 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:28:52.524701739 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:28:52.535605482 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:28:52.576580811 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n8_seed7_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed7_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:22:15 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.4437 (2.4437) acc1: 0.0000 (0.0000) acc5: 100.0000 (100.0000) time: 5.7070 data: 4.3189 max mem: 3531 +Test: [ 10/234] eta: 0:02:14 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.3644 (2.3566) acc1: 0.0000 (1.9886) acc5: 100.0000 (100.0000) time: 0.6004 data: 0.3930 max mem: 3531 +Test: [ 20/234] eta: 0:01:16 acc1_pos: 94.531 acc5_pos: 100.000 loss: 2.3644 (2.0812) acc1: 0.0000 (16.7039) acc5: 100.0000 (100.0000) time: 0.0909 data: 0.0004 max mem: 3531 +Test: [ 30/234] eta: 0:00:58 acc1_pos: 67.708 acc5_pos: 100.000 loss: 0.2240 (1.4991) acc1: 92.1875 (40.7594) acc5: 100.0000 (100.0000) time: 0.1160 data: 0.0220 max mem: 3531 +Test: [ 40/234] eta: 0:00:54 acc1_pos: 85.677 acc5_pos: 100.000 loss: 0.3975 (1.2588) acc1: 83.8542 (49.4792) acc5: 100.0000 (100.0000) time: 0.2006 data: 0.1058 max mem: 3531 +Test: [ 50/234] eta: 0:00:49 acc1_pos: 84.115 acc5_pos: 100.000 loss: 0.4154 (1.0918) acc1: 82.2917 (56.1530) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1530 max mem: 3531 +Test: [ 60/234] eta: 0:00:44 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.2234 (0.9291) acc1: 86.1979 (63.2258) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2056 data: 0.1149 max mem: 3531 +Test: [ 70/234] eta: 0:00:41 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.0878 (0.8603) acc1: 100.0000 (65.1115) acc5: 100.0000 (100.0000) acc1_neg: 64.062 acc5_neg: 100.000 time: 0.1999 data: 0.1063 max mem: 3531 +Test: [ 80/234] eta: 0:00:38 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.5778 (0.8049) acc1: 66.4062 (66.7374) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2336 data: 0.1390 max mem: 3531 +Test: [ 90/234] eta: 0:00:35 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.2585 (0.7406) acc1: 89.0625 (69.6085) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.2308 data: 0.1390 max mem: 3531 +Test: [100/234] eta: 0:00:32 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.2989 (0.7044) acc1: 86.4583 (70.7843) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1909 data: 0.1005 max mem: 3531 +Test: [110/234] eta: 0:00:29 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.3113 (0.6626) acc1: 86.4583 (72.5741) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.1840 data: 0.0927 max mem: 3531 +Test: [120/234] eta: 0:00:26 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.2763 (0.6278) acc1: 88.0208 (74.0724) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.1940 data: 0.0996 max mem: 3531 +Test: [130/234] eta: 0:00:23 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.1506 (0.5932) acc1: 96.6146 (75.6759) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1857 data: 0.0921 max mem: 3531 +Test: [140/234] eta: 0:00:21 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.1559 (0.5948) acc1: 88.0208 (75.0369) acc5: 100.0000 (100.0000) acc1_neg: 51.042 acc5_neg: 100.000 time: 0.1878 data: 0.0972 max mem: 3531 +Test: [150/234] eta: 0:00:18 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.3489 (0.5809) acc1: 81.2500 (75.4743) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.1912 data: 0.1007 max mem: 3531 +Test: [160/234] eta: 0:00:16 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.1822 (0.5531) acc1: 91.9271 (76.8925) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1826 data: 0.0920 max mem: 3531 +Test: [170/234] eta: 0:00:13 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.1602 (0.5296) acc1: 97.9167 (78.1113) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1794 data: 0.0866 max mem: 3531 +Test: [180/234] eta: 0:00:11 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.1777 (0.5235) acc1: 95.8333 (78.1595) acc5: 100.0000 (100.0000) acc1_neg: 64.323 acc5_neg: 100.000 time: 0.1882 data: 0.0953 max mem: 3531 +Test: [190/234] eta: 0:00:09 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.4226 (0.5190) acc1: 80.9896 (78.2859) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.1864 data: 0.0959 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.4106 (0.5038) acc1: 81.7708 (79.0021) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1747 data: 0.0830 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.0879 (0.4854) acc1: 99.7396 (79.9627) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1563 data: 0.0648 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.1741 (0.4732) acc1: 97.6562 (80.6561) acc5: 100.0000 (100.0000) acc1_neg: 78.646 acc5_neg: 100.000 time: 0.1327 data: 0.0425 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.4465 (0.4942) acc1: 44.5312 (78.8724) acc5: 100.0000 (100.0000) acc1_neg: 39.323 acc5_neg: 100.000 time: 0.1286 data: 0.0385 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.9183 (0.4999) acc1: 42.0233 (78.4518) acc5: 100.0000 (100.0000) acc1_neg: 42.023 acc5_neg: 100.000 time: 0.1260 data: 0.0352 max mem: 3532 +Test: Total time: 0:00:47 (0.2015 s / it) +* Overall Acc@1 78.452 Acc@5 100.000 + target == real Acc@1 56.349 Acc@5 100.000 + target == fake Acc@1 84.681 Acc@5 100.000 + loss 0.500 + PR-AUC 0.8745 ROC-AUC 0.7228 EER 34.45% (thr=0.8179) +Accuracy of the network on the 89729 test images: 78.5% +[rank0]:[W726 04:29:51.533607951 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:30:12.346000 1183017 torch/distributed/run.py:766] +W0726 04:30:12.346000 1183017 torch/distributed/run.py:766] ***************************************** +W0726 04:30:12.346000 1183017 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:30:12.346000 1183017 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 04:30:42.145179509 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:30:42.359364491 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:30:43.750240030 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:30:43.782190223 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:30:43.838627158 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:30:43.033290982 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:30:43.102874934 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:30:43.170144271 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n8_seed7_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed7_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 5:08:36 acc1_pos: 2.083 acc5_pos: 100.000 loss: 2.3413 (2.3413) acc1: 2.0833 (2.0833) acc5: 100.0000 (100.0000) time: 7.7572 data: 6.1784 max mem: 3531 +Test: [ 10/2387] eta: 0:31:17 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.2909 (2.2586) acc1: 1.0417 (2.2964) acc5: 100.0000 (100.0000) time: 0.7901 data: 0.5620 max mem: 3531 +Test: [ 20/2387] eta: 0:18:12 acc1_pos: 0.260 acc5_pos: 100.000 loss: 2.3124 (2.3092) acc1: 0.7812 (2.1205) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:16:36 acc1_pos: 2.865 acc5_pos: 100.000 loss: 2.3147 (2.3036) acc1: 0.7812 (1.8481) acc5: 100.0000 (100.0000) time: 0.2210 data: 0.1259 max mem: 3531 +Test: [ 40/2387] eta: 0:15:44 acc1_pos: 1.823 acc5_pos: 100.000 loss: 2.3147 (2.3046) acc1: 0.7812 (1.6895) acc5: 100.0000 (100.0000) time: 0.3401 data: 0.2495 max mem: 3531 +Test: [ 50/2387] eta: 0:14:25 acc1_pos: 3.385 acc5_pos: 100.000 loss: 2.3390 (2.3141) acc1: 0.2604 (1.5012) acc5: 100.0000 (100.0000) time: 0.2888 data: 0.1960 max mem: 3531 +Test: [ 60/2387] eta: 0:13:43 acc1_pos: 1.042 acc5_pos: 100.000 loss: 2.3647 (2.3205) acc1: 0.0000 (1.4387) acc5: 100.0000 (100.0000) time: 0.2547 data: 0.1594 max mem: 3531 +Test: [ 70/2387] eta: 0:13:02 acc1_pos: 10.677 acc5_pos: 100.000 loss: 2.4048 (2.3344) acc1: 0.0000 (1.4451) acc5: 100.0000 (100.0000) time: 0.2548 data: 0.1593 max mem: 3531 +Test: [ 80/2387] eta: 0:12:40 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.4637 (2.3568) acc1: 0.0000 (1.3085) acc5: 100.0000 (100.0000) time: 0.2565 data: 0.1604 max mem: 3531 +Test: [ 90/2387] eta: 0:12:10 acc1_pos: 0.260 acc5_pos: 100.000 loss: 2.5367 (2.3770) acc1: 0.0000 (1.1704) acc5: 100.0000 (100.0000) time: 0.2483 data: 0.1531 max mem: 3531 +Test: [ 100/2387] eta: 0:11:55 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.5638 (2.3993) acc1: 0.0000 (1.0726) acc5: 100.0000 (100.0000) time: 0.2432 data: 0.1506 max mem: 3531 +Test: [ 110/2387] eta: 0:11:35 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.5759 (2.4126) acc1: 0.0000 (1.0252) acc5: 100.0000 (100.0000) time: 0.2474 data: 0.1530 max mem: 3531 +Test: [ 120/2387] eta: 0:11:31 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.5499 (2.4237) acc1: 0.0000 (0.9513) acc5: 100.0000 (100.0000) time: 0.2674 data: 0.1712 max mem: 3531 +Test: [ 130/2387] eta: 0:11:28 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.5390 (2.4337) acc1: 0.0000 (0.9761) acc5: 100.0000 (100.0000) time: 0.3040 data: 0.2093 max mem: 3531 +Test: [ 140/2387] eta: 0:11:11 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.5622 (2.4424) acc1: 0.0000 (0.9087) acc5: 100.0000 (100.0000) time: 0.2599 data: 0.1642 max mem: 3531 +Test: [ 150/2387] eta: 0:11:04 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.5073 (2.4459) acc1: 0.0000 (0.8985) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1498 max mem: 3531 +Test: [ 160/2387] eta: 0:10:50 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.4770 (2.3905) acc1: 0.0000 (3.3822) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2452 data: 0.1496 max mem: 3531 +Test: [ 170/2387] eta: 0:10:36 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0935 (2.2560) acc1: 99.4792 (9.0232) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2111 data: 0.1167 max mem: 3531 +Test: [ 180/2387] eta: 0:10:21 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0900 (2.1363) acc1: 100.0000 (14.0424) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1970 data: 0.1024 max mem: 3531 +Test: [ 190/2387] eta: 0:10:09 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0883 (2.0291) acc1: 100.0000 (18.5359) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1962 data: 0.0987 max mem: 3531 +Test: [ 200/2387] eta: 0:09:59 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0881 (1.9326) acc1: 100.0000 (22.5824) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2059 data: 0.1118 max mem: 3531 +Test: [ 210/2387] eta: 0:09:48 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0903 (1.8454) acc1: 100.0000 (26.2416) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1989 data: 0.1048 max mem: 3532 +Test: [ 220/2387] eta: 0:09:38 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0905 (1.7659) acc1: 100.0000 (29.5720) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1957 data: 0.0996 max mem: 3532 +Test: [ 230/2387] eta: 0:09:25 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0895 (1.6934) acc1: 100.0000 (32.6163) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1792 data: 0.0838 max mem: 3532 +Test: [ 240/2387] eta: 0:09:18 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0854 (1.6263) acc1: 100.0000 (35.4102) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1817 data: 0.0859 max mem: 3532 +Test: [ 250/2387] eta: 0:09:06 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0729 (1.5640) acc1: 100.0000 (37.9824) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1797 data: 0.0822 max mem: 3532 +Test: [ 260/2387] eta: 0:08:59 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0582 (1.5063) acc1: 100.0000 (40.3586) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1765 data: 0.0813 max mem: 3532 +Test: [ 270/2387] eta: 0:08:56 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0578 (1.4528) acc1: 100.0000 (42.5584) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2279 data: 0.1343 max mem: 3532 +Test: [ 280/2387] eta: 0:08:48 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0586 (1.4033) acc1: 100.0000 (44.6017) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2145 data: 0.1175 max mem: 3532 +Test: [ 290/2387] eta: 0:08:35 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0592 (1.3571) acc1: 100.0000 (46.5054) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1382 data: 0.0438 max mem: 3532 +Test: [ 300/2387] eta: 0:08:27 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0601 (1.3141) acc1: 100.0000 (48.2783) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1396 data: 0.0492 max mem: 3532 +Test: [ 310/2387] eta: 0:08:22 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0601 (1.2759) acc1: 100.0000 (49.9255) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1840 data: 0.0913 max mem: 3532 +Test: [ 320/2387] eta: 0:08:17 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2057 (1.2427) acc1: 99.2188 (51.4570) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2006 data: 0.1065 max mem: 3532 +Test: [ 330/2387] eta: 0:08:14 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2087 (1.2115) acc1: 99.2188 (52.9031) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2225 data: 0.1287 max mem: 3532 +Test: [ 340/2387] eta: 0:08:11 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2096 (1.1820) acc1: 99.4792 (54.2705) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2282 data: 0.1315 max mem: 3532 +Test: [ 350/2387] eta: 0:08:05 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1976 (1.1516) acc1: 99.7396 (55.5704) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2051 data: 0.1069 max mem: 3532 +Test: [ 360/2387] eta: 0:07:59 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0983 (1.1222) acc1: 100.0000 (56.7997) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0787 max mem: 3532 +Test: [ 370/2387] eta: 0:07:53 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0864 (1.0942) acc1: 100.0000 (57.9613) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1707 data: 0.0777 max mem: 3532 +Test: [ 380/2387] eta: 0:07:48 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0853 (1.0678) acc1: 100.0000 (59.0626) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1815 data: 0.0910 max mem: 3532 +Test: [ 390/2387] eta: 0:07:43 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0853 (1.0426) acc1: 100.0000 (60.1063) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1764 data: 0.0852 max mem: 3532 +Test: [ 400/2387] eta: 0:07:38 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0855 (1.0188) acc1: 100.0000 (61.0986) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1745 data: 0.0821 max mem: 3532 +Test: [ 410/2387] eta: 0:07:33 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0852 (0.9961) acc1: 100.0000 (62.0413) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1836 data: 0.0916 max mem: 3532 +Test: [ 420/2387] eta: 0:07:29 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0853 (0.9745) acc1: 99.7396 (62.9398) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1899 data: 0.0949 max mem: 3532 +Test: [ 430/2387] eta: 0:07:25 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0846 (0.9537) acc1: 100.0000 (63.7985) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1891 data: 0.0928 max mem: 3532 +Test: [ 440/2387] eta: 0:07:21 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0726 (0.9337) acc1: 100.0000 (64.6194) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1846 data: 0.0915 max mem: 3533 +Test: [ 450/2387] eta: 0:07:17 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0708 (0.9145) acc1: 100.0000 (65.4038) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1869 data: 0.0944 max mem: 3533 +Test: [ 460/2387] eta: 0:07:13 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0709 (0.8962) acc1: 100.0000 (66.1543) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1906 data: 0.0987 max mem: 3533 +Test: [ 470/2387] eta: 0:07:09 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0709 (0.8787) acc1: 100.0000 (66.8729) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1824 data: 0.0905 max mem: 3533 +Test: [ 480/2387] eta: 0:07:05 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0713 (0.8619) acc1: 100.0000 (67.5616) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1754 data: 0.0835 max mem: 3533 +Test: [ 490/2387] eta: 0:07:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0711 (0.8458) acc1: 100.0000 (68.2223) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1777 data: 0.0863 max mem: 3533 +Test: [ 500/2387] eta: 0:06:57 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0711 (0.8304) acc1: 100.0000 (68.8566) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1779 data: 0.0864 max mem: 3533 +Test: [ 510/2387] eta: 0:06:53 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0711 (0.8156) acc1: 100.0000 (69.4660) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1697 data: 0.0777 max mem: 3533 +Test: [ 520/2387] eta: 0:06:49 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0762 (0.8014) acc1: 100.0000 (70.0521) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1774 data: 0.0845 max mem: 3533 +Test: [ 530/2387] eta: 0:06:46 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0755 (0.7877) acc1: 100.0000 (70.6161) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1848 data: 0.0920 max mem: 3533 +Test: [ 540/2387] eta: 0:06:42 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0751 (0.7746) acc1: 100.0000 (71.1592) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1741 data: 0.0819 max mem: 3533 +Test: [ 550/2387] eta: 0:06:39 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0754 (0.7619) acc1: 100.0000 (71.6826) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1825 data: 0.0895 max mem: 3533 +Test: [ 560/2387] eta: 0:06:36 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0748 (0.7496) acc1: 100.0000 (72.1874) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1936 data: 0.1002 max mem: 3533 +Test: [ 570/2387] eta: 0:06:33 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0749 (0.7378) acc1: 100.0000 (72.6745) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1948 data: 0.1014 max mem: 3533 +Test: [ 580/2387] eta: 0:06:30 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0749 (0.7264) acc1: 100.0000 (73.1448) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1997 data: 0.1078 max mem: 3533 +Test: [ 590/2387] eta: 0:06:27 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0748 (0.7154) acc1: 100.0000 (73.5992) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1803 data: 0.0895 max mem: 3533 +Test: [ 600/2387] eta: 0:06:24 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0758 (0.7048) acc1: 100.0000 (74.0385) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1743 data: 0.0811 max mem: 3533 +Test: [ 610/2387] eta: 0:06:20 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0855 (0.6948) acc1: 100.0000 (74.4621) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1845 data: 0.0896 max mem: 3533 +Test: [ 620/2387] eta: 0:06:17 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0905 (0.6851) acc1: 100.0000 (74.8721) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1799 data: 0.0868 max mem: 3533 +Test: [ 630/2387] eta: 0:06:15 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0906 (0.6757) acc1: 100.0000 (75.2699) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0976 max mem: 3533 +Test: [ 640/2387] eta: 0:06:13 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0917 (0.6666) acc1: 100.0000 (75.6549) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2051 data: 0.1132 max mem: 3533 +Test: [ 650/2387] eta: 0:06:10 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0917 (0.6577) acc1: 100.0000 (76.0277) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2077 data: 0.1166 max mem: 3533 +Test: [ 660/2387] eta: 0:06:09 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0919 (0.6492) acc1: 100.0000 (76.3899) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2197 data: 0.1276 max mem: 3533 +Test: [ 670/2387] eta: 0:06:06 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0911 (0.6408) acc1: 100.0000 (76.7414) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2093 data: 0.1177 max mem: 3534 +Test: [ 680/2387] eta: 0:06:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0891 (0.6326) acc1: 100.0000 (77.0830) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1920 data: 0.1016 max mem: 3534 +Test: [ 690/2387] eta: 0:06:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0912 (0.6273) acc1: 100.0000 (77.2574) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.2011 data: 0.1101 max mem: 3534 +Test: [ 700/2387] eta: 0:05:59 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.3257 (0.6230) acc1: 85.9375 (77.3750) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.2052 data: 0.1136 max mem: 3534 +Test: [ 710/2387] eta: 0:05:56 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.3213 (0.6188) acc1: 85.9375 (77.4965) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.2095 data: 0.1179 max mem: 3534 +Test: [ 720/2387] eta: 0:05:54 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.3280 (0.6148) acc1: 85.1562 (77.6074) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.2128 data: 0.1202 max mem: 3534 +Test: [ 730/2387] eta: 0:05:52 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.3323 (0.6109) acc1: 85.1562 (77.7221) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.2077 data: 0.1143 max mem: 3534 +Test: [ 740/2387] eta: 0:05:49 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.3296 (0.6070) acc1: 85.6771 (77.8277) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.1972 data: 0.1044 max mem: 3534 +Test: [ 750/2387] eta: 0:05:47 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.3228 (0.6033) acc1: 85.6771 (77.9322) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1858 data: 0.0942 max mem: 3534 +Test: [ 760/2387] eta: 0:05:44 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.3178 (0.5987) acc1: 86.7188 (78.0963) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1746 data: 0.0837 max mem: 3534 +Test: [ 770/2387] eta: 0:05:41 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1386 (0.5925) acc1: 96.6146 (78.3540) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1732 data: 0.0820 max mem: 3534 +Test: [ 780/2387] eta: 0:05:38 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1017 (0.5861) acc1: 98.1771 (78.6188) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1826 data: 0.0900 max mem: 3534 +Test: [ 790/2387] eta: 0:05:35 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0851 (0.5803) acc1: 99.7396 (78.8453) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.1818 data: 0.0896 max mem: 3534 +Test: [ 800/2387] eta: 0:05:33 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0796 (0.5741) acc1: 99.7396 (79.1042) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1799 data: 0.0856 max mem: 3534 +Test: [ 810/2387] eta: 0:05:30 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0791 (0.5681) acc1: 99.7396 (79.3568) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1853 data: 0.0906 max mem: 3534 +Test: [ 820/2387] eta: 0:05:28 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0791 (0.5621) acc1: 99.7396 (79.6047) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1920 data: 0.1000 max mem: 3534 +Test: [ 830/2387] eta: 0:05:25 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0754 (0.5565) acc1: 100.0000 (79.8307) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1929 data: 0.1007 max mem: 3534 +Test: [ 840/2387] eta: 0:05:23 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0973 (0.5512) acc1: 99.7396 (80.0625) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1935 data: 0.1004 max mem: 3534 +Test: [ 850/2387] eta: 0:05:21 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1038 (0.5460) acc1: 100.0000 (80.2913) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2015 data: 0.1088 max mem: 3534 +Test: [ 860/2387] eta: 0:05:18 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0990 (0.5407) acc1: 100.0000 (80.5168) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1946 data: 0.1017 max mem: 3534 +Test: [ 870/2387] eta: 0:05:16 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0987 (0.5358) acc1: 100.0000 (80.7328) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1937 data: 0.1013 max mem: 3534 +Test: [ 880/2387] eta: 0:05:14 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1009 (0.5309) acc1: 99.7396 (80.9482) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1896 data: 0.0982 max mem: 3534 +Test: [ 890/2387] eta: 0:05:11 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0930 (0.5258) acc1: 100.0000 (81.1620) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1718 data: 0.0807 max mem: 3534 +Test: [ 900/2387] eta: 0:05:08 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0859 (0.5283) acc1: 100.0000 (80.8558) acc5: 100.0000 (100.0000) acc1_neg: 41.927 acc5_neg: 100.000 time: 0.1705 data: 0.0791 max mem: 3535 +Test: [ 910/2387] eta: 0:05:06 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.9033 (0.5324) acc1: 43.2292 (80.4430) acc5: 100.0000 (100.0000) acc1_neg: 43.490 acc5_neg: 100.000 time: 0.1793 data: 0.0849 max mem: 3535 +Test: [ 920/2387] eta: 0:05:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.9040 (0.5366) acc1: 42.4479 (80.0285) acc5: 100.0000 (100.0000) acc1_neg: 42.448 acc5_neg: 100.000 time: 0.1809 data: 0.0877 max mem: 3535 +Test: [ 930/2387] eta: 0:05:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.9109 (0.5406) acc1: 41.9271 (79.6268) acc5: 100.0000 (100.0000) acc1_neg: 39.583 acc5_neg: 100.000 time: 0.1746 data: 0.0816 max mem: 3535 +Test: [ 940/2387] eta: 0:04:58 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.9140 (0.5445) acc1: 41.4062 (79.2300) acc5: 100.0000 (100.0000) acc1_neg: 41.406 acc5_neg: 100.000 time: 0.1832 data: 0.0880 max mem: 3535 +Test: [ 950/2387] eta: 0:04:56 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.9144 (0.5484) acc1: 42.1875 (78.8403) acc5: 100.0000 (100.0000) acc1_neg: 42.969 acc5_neg: 100.000 time: 0.1851 data: 0.0920 max mem: 3535 +Test: [ 960/2387] eta: 0:04:54 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.9089 (0.5520) acc1: 42.9688 (78.4762) acc5: 100.0000 (100.0000) acc1_neg: 48.438 acc5_neg: 100.000 time: 0.1949 data: 0.1009 max mem: 3535 +Test: [ 970/2387] eta: 0:04:52 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.9133 (0.5558) acc1: 41.6667 (78.1027) acc5: 100.0000 (100.0000) acc1_neg: 41.667 acc5_neg: 100.000 time: 0.2154 data: 0.1207 max mem: 3535 +Test: [ 980/2387] eta: 0:04:50 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.8789 (0.5540) acc1: 43.4896 (78.1173) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2215 data: 0.1293 max mem: 3535 +Test: [ 990/2387] eta: 0:04:48 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0762 (0.5492) acc1: 100.0000 (78.3355) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2144 data: 0.1217 max mem: 3535 +Test: [1000/2387] eta: 0:04:46 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0687 (0.5445) acc1: 100.0000 (78.5384) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2109 data: 0.1171 max mem: 3535 +Test: [1010/2387] eta: 0:04:43 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0694 (0.5398) acc1: 100.0000 (78.7486) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1957 data: 0.1032 max mem: 3535 +Test: [1020/2387] eta: 0:04:41 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0698 (0.5355) acc1: 100.0000 (78.9420) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1823 data: 0.0819 max mem: 3535 +Test: [1030/2387] eta: 0:04:39 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0698 (0.5309) acc1: 100.0000 (79.1452) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1981 data: 0.0960 max mem: 3535 +Test: [1040/2387] eta: 0:04:37 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0664 (0.5265) acc1: 100.0000 (79.3453) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1976 data: 0.1043 max mem: 3535 +Test: [1050/2387] eta: 0:04:34 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0637 (0.5221) acc1: 100.0000 (79.5411) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1842 data: 0.0918 max mem: 3535 +Test: [1060/2387] eta: 0:04:32 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0637 (0.5178) acc1: 100.0000 (79.7339) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1798 data: 0.0837 max mem: 3535 +Test: [1070/2387] eta: 0:04:30 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0653 (0.5136) acc1: 100.0000 (79.9229) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1908 data: 0.0958 max mem: 3535 +Test: [1080/2387] eta: 0:04:28 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0686 (0.5096) acc1: 100.0000 (80.1009) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1931 data: 0.1027 max mem: 3535 +Test: [1090/2387] eta: 0:04:25 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0739 (0.5056) acc1: 100.0000 (80.2830) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1888 data: 0.0985 max mem: 3535 +Test: [1100/2387] eta: 0:04:23 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0758 (0.5017) acc1: 100.0000 (80.4621) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1911 data: 0.1007 max mem: 3535 +Test: [1110/2387] eta: 0:04:21 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0755 (0.4983) acc1: 100.0000 (80.5993) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1873 data: 0.0953 max mem: 3535 +Test: [1120/2387] eta: 0:04:19 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0755 (0.4945) acc1: 100.0000 (80.7701) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1883 data: 0.0944 max mem: 3536 +Test: [1130/2387] eta: 0:04:17 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0746 (0.4908) acc1: 100.0000 (80.9385) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1875 data: 0.0934 max mem: 3536 +Test: [1140/2387] eta: 0:04:14 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0696 (0.4871) acc1: 100.0000 (81.1053) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1816 data: 0.0883 max mem: 3536 +Test: [1150/2387] eta: 0:04:12 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0734 (0.4839) acc1: 100.0000 (81.2536) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1862 data: 0.0929 max mem: 3536 +Test: [1160/2387] eta: 0:04:10 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1084 (0.4807) acc1: 100.0000 (81.4131) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1812 data: 0.0881 max mem: 3536 +Test: [1170/2387] eta: 0:04:07 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1084 (0.4775) acc1: 100.0000 (81.5694) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1737 data: 0.0822 max mem: 3536 +Test: [1180/2387] eta: 0:04:05 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1115 (0.4745) acc1: 100.0000 (81.7175) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1725 data: 0.0806 max mem: 3536 +Test: [1190/2387] eta: 0:04:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1056 (0.4715) acc1: 99.7396 (81.8664) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1859 data: 0.0937 max mem: 3536 +Test: [1200/2387] eta: 0:04:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0895 (0.4682) acc1: 100.0000 (82.0172) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1714 data: 0.0803 max mem: 3536 +Test: [1210/2387] eta: 0:03:58 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0793 (0.4650) acc1: 100.0000 (82.1657) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1690 data: 0.0778 max mem: 3536 +Test: [1220/2387] eta: 0:03:56 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0668 (0.4617) acc1: 100.0000 (82.3117) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1950 data: 0.1038 max mem: 3536 +Test: [1230/2387] eta: 0:03:54 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0605 (0.4585) acc1: 100.0000 (82.4554) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1912 data: 0.1002 max mem: 3536 +Test: [1240/2387] eta: 0:03:52 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0621 (0.4553) acc1: 100.0000 (82.5968) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1886 data: 0.0970 max mem: 3536 +Test: [1250/2387] eta: 0:03:50 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0631 (0.4522) acc1: 100.0000 (82.7357) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1864 data: 0.0938 max mem: 3536 +Test: [1260/2387] eta: 0:03:47 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0628 (0.4491) acc1: 100.0000 (82.8722) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1823 data: 0.0888 max mem: 3536 +Test: [1270/2387] eta: 0:03:45 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0605 (0.4460) acc1: 100.0000 (83.0069) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1942 data: 0.1002 max mem: 3536 +Test: [1280/2387] eta: 0:03:43 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0599 (0.4430) acc1: 100.0000 (83.1394) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1819 data: 0.0894 max mem: 3536 +Test: [1290/2387] eta: 0:03:41 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0607 (0.4402) acc1: 100.0000 (83.2557) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1645 data: 0.0724 max mem: 3536 +Test: [1300/2387] eta: 0:03:38 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0607 (0.4373) acc1: 99.7396 (83.3790) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1673 data: 0.0743 max mem: 3536 +Test: [1310/2387] eta: 0:03:36 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0604 (0.4345) acc1: 100.0000 (83.4994) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1657 data: 0.0739 max mem: 3536 +Test: [1320/2387] eta: 0:03:34 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0646 (0.4318) acc1: 100.0000 (83.6133) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1768 data: 0.0836 max mem: 3536 +Test: [1330/2387] eta: 0:03:32 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0666 (0.4291) acc1: 100.0000 (83.7350) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1889 data: 0.0952 max mem: 3536 +Test: [1340/2387] eta: 0:03:30 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0710 (0.4269) acc1: 99.7396 (83.8371) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1871 data: 0.0954 max mem: 3536 +Test: [1350/2387] eta: 0:03:28 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0645 (0.4242) acc1: 100.0000 (83.9567) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1840 data: 0.0930 max mem: 3537 +Test: [1360/2387] eta: 0:03:25 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0603 (0.4215) acc1: 100.0000 (84.0746) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1877 data: 0.0961 max mem: 3537 +Test: [1370/2387] eta: 0:03:23 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0609 (0.4189) acc1: 100.0000 (84.1908) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0914 max mem: 3537 +Test: [1380/2387] eta: 0:03:21 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0619 (0.4163) acc1: 100.0000 (84.3052) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1818 data: 0.0860 max mem: 3537 +Test: [1390/2387] eta: 0:03:19 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0620 (0.4137) acc1: 100.0000 (84.4181) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1818 data: 0.0865 max mem: 3537 +Test: [1400/2387] eta: 0:03:17 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0602 (0.4112) acc1: 100.0000 (84.5293) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1844 data: 0.0923 max mem: 3537 +Test: [1410/2387] eta: 0:03:15 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0602 (0.4089) acc1: 100.0000 (84.6352) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1872 data: 0.0937 max mem: 3537 +Test: [1420/2387] eta: 0:03:13 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0729 (0.4067) acc1: 100.0000 (84.7250) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1838 data: 0.0873 max mem: 3537 +Test: [1430/2387] eta: 0:03:11 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0790 (0.4045) acc1: 99.7396 (84.8278) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1899 data: 0.0943 max mem: 3537 +Test: [1440/2387] eta: 0:03:08 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0746 (0.4022) acc1: 100.0000 (84.9329) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1809 data: 0.0872 max mem: 3537 +Test: [1450/2387] eta: 0:03:06 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0745 (0.4000) acc1: 100.0000 (85.0304) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1755 data: 0.0836 max mem: 3537 +Test: [1460/2387] eta: 0:03:04 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0797 (0.3978) acc1: 100.0000 (85.1322) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1821 data: 0.0906 max mem: 3537 +Test: [1470/2387] eta: 0:03:02 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0719 (0.3956) acc1: 100.0000 (85.2333) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1775 data: 0.0844 max mem: 3537 +Test: [1480/2387] eta: 0:03:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0711 (0.3934) acc1: 100.0000 (85.3303) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1827 data: 0.0897 max mem: 3537 +Test: [1490/2387] eta: 0:02:58 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0741 (0.3914) acc1: 100.0000 (85.4273) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1830 data: 0.0920 max mem: 3537 +Test: [1500/2387] eta: 0:02:56 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0777 (0.3893) acc1: 100.0000 (85.5234) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1735 data: 0.0803 max mem: 3537 +Test: [1510/2387] eta: 0:02:54 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0777 (0.3872) acc1: 100.0000 (85.6190) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0762 max mem: 3537 +Test: [1520/2387] eta: 0:02:52 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0858 (0.3854) acc1: 100.0000 (85.7069) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1874 data: 0.0966 max mem: 3537 +Test: [1530/2387] eta: 0:02:50 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1071 (0.3836) acc1: 99.7396 (85.7979) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1942 data: 0.1042 max mem: 3537 +Test: [1540/2387] eta: 0:02:47 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1071 (0.3818) acc1: 99.7396 (85.8876) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1756 data: 0.0837 max mem: 3537 +Test: [1550/2387] eta: 0:02:45 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1088 (0.3801) acc1: 99.7396 (85.9726) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1707 data: 0.0781 max mem: 3537 +Test: [1560/2387] eta: 0:02:43 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1062 (0.3783) acc1: 99.7396 (86.0586) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1869 data: 0.0938 max mem: 3537 +Test: [1570/2387] eta: 0:02:41 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0944 (0.3765) acc1: 100.0000 (86.1474) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0970 max mem: 3537 +Test: [1580/2387] eta: 0:02:39 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0822 (0.3747) acc1: 100.0000 (86.2295) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1808 data: 0.0893 max mem: 3538 +Test: [1590/2387] eta: 0:02:37 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0838 (0.3729) acc1: 100.0000 (86.3140) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1793 data: 0.0880 max mem: 3538 +Test: [1600/2387] eta: 0:02:35 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0820 (0.3714) acc1: 99.7396 (86.3713) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1755 data: 0.0843 max mem: 3538 +Test: [1610/2387] eta: 0:02:33 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0798 (0.3695) acc1: 100.0000 (86.4554) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1700 data: 0.0776 max mem: 3538 +Test: [1620/2387] eta: 0:02:31 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0851 (0.3678) acc1: 100.0000 (86.5346) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1714 data: 0.0778 max mem: 3538 +Test: [1630/2387] eta: 0:02:29 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0875 (0.3661) acc1: 99.7396 (86.6135) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1798 data: 0.0867 max mem: 3538 +Test: [1640/2387] eta: 0:02:27 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0800 (0.3645) acc1: 99.7396 (86.6869) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1755 data: 0.0841 max mem: 3538 +Test: [1650/2387] eta: 0:02:25 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0800 (0.3627) acc1: 99.7396 (86.7647) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1660 data: 0.0758 max mem: 3538 +Test: [1660/2387] eta: 0:02:22 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0785 (0.3611) acc1: 99.7396 (86.8406) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1582 data: 0.0667 max mem: 3538 +Test: [1670/2387] eta: 0:02:20 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0823 (0.3594) acc1: 99.7396 (86.9175) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1579 data: 0.0640 max mem: 3538 +Test: [1680/2387] eta: 0:02:18 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0754 (0.3579) acc1: 99.7396 (86.9849) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1548 data: 0.0623 max mem: 3538 +Test: [1690/2387] eta: 0:02:16 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0844 (0.3563) acc1: 99.7396 (87.0545) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1528 data: 0.0621 max mem: 3538 +Test: [1700/2387] eta: 0:02:14 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0744 (0.3547) acc1: 100.0000 (87.1300) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1686 data: 0.0772 max mem: 3538 +Test: [1710/2387] eta: 0:02:12 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0663 (0.3530) acc1: 100.0000 (87.2052) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1720 data: 0.0804 max mem: 3538 +Test: [1720/2387] eta: 0:02:10 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0636 (0.3513) acc1: 100.0000 (87.2795) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1792 data: 0.0853 max mem: 3538 +Test: [1730/2387] eta: 0:02:08 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0629 (0.3496) acc1: 100.0000 (87.3526) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0857 max mem: 3538 +Test: [1740/2387] eta: 0:02:06 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0620 (0.3480) acc1: 100.0000 (87.4251) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1789 data: 0.0871 max mem: 3538 +Test: [1750/2387] eta: 0:02:04 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0619 (0.3464) acc1: 100.0000 (87.4969) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1999 data: 0.1089 max mem: 3538 +Test: [1760/2387] eta: 0:02:02 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0614 (0.3448) acc1: 100.0000 (87.5679) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1620 data: 0.0710 max mem: 3538 +Test: [1770/2387] eta: 0:02:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0612 (0.3432) acc1: 100.0000 (87.6379) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1545 data: 0.0622 max mem: 3538 +Test: [1780/2387] eta: 0:01:58 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0627 (0.3416) acc1: 100.0000 (87.7073) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1989 data: 0.1053 max mem: 3538 +Test: [1790/2387] eta: 0:01:56 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0614 (0.3400) acc1: 100.0000 (87.7758) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1932 data: 0.0995 max mem: 3538 +Test: [1800/2387] eta: 0:01:54 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0621 (0.3385) acc1: 100.0000 (87.8437) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1913 data: 0.0985 max mem: 3538 +Test: [1810/2387] eta: 0:01:52 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0626 (0.3370) acc1: 100.0000 (87.9108) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1951 data: 0.1024 max mem: 3539 +Test: [1820/2387] eta: 0:01:50 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0621 (0.3355) acc1: 100.0000 (87.9769) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1892 data: 0.0966 max mem: 3539 +Test: [1830/2387] eta: 0:01:48 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0612 (0.3340) acc1: 100.0000 (88.0426) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1817 data: 0.0887 max mem: 3539 +Test: [1840/2387] eta: 0:01:46 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0671 (0.3325) acc1: 100.0000 (88.1075) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1761 data: 0.0833 max mem: 3539 +Test: [1850/2387] eta: 0:01:44 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0678 (0.3311) acc1: 100.0000 (88.1718) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1773 data: 0.0860 max mem: 3539 +Test: [1860/2387] eta: 0:01:42 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0679 (0.3297) acc1: 100.0000 (88.2354) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1723 data: 0.0818 max mem: 3539 +Test: [1870/2387] eta: 0:01:40 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0677 (0.3283) acc1: 100.0000 (88.2982) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1696 data: 0.0773 max mem: 3539 +Test: [1880/2387] eta: 0:01:38 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0676 (0.3269) acc1: 100.0000 (88.3604) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1745 data: 0.0820 max mem: 3539 +Test: [1890/2387] eta: 0:01:36 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0676 (0.3255) acc1: 100.0000 (88.4220) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1819 data: 0.0903 max mem: 3539 +Test: [1900/2387] eta: 0:01:34 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0676 (0.3242) acc1: 100.0000 (88.4829) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1830 data: 0.0912 max mem: 3539 +Test: [1910/2387] eta: 0:01:32 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0680 (0.3228) acc1: 100.0000 (88.5432) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0923 max mem: 3539 +Test: [1920/2387] eta: 0:01:30 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0679 (0.3215) acc1: 100.0000 (88.6028) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1868 data: 0.0936 max mem: 3539 +Test: [1930/2387] eta: 0:01:28 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0629 (0.3202) acc1: 100.0000 (88.6618) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1860 data: 0.0929 max mem: 3539 +Test: [1940/2387] eta: 0:01:26 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0609 (0.3192) acc1: 100.0000 (88.6859) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1851 data: 0.0927 max mem: 3539 +Test: [1950/2387] eta: 0:01:24 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0626 (0.3179) acc1: 100.0000 (88.7439) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1802 data: 0.0889 max mem: 3539 +Test: [1960/2387] eta: 0:01:22 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0645 (0.3166) acc1: 100.0000 (88.7970) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1815 data: 0.0895 max mem: 3539 +Test: [1970/2387] eta: 0:01:20 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0675 (0.3154) acc1: 100.0000 (88.8539) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1806 data: 0.0868 max mem: 3539 +Test: [1980/2387] eta: 0:01:18 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0791 (0.3144) acc1: 100.0000 (88.9005) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1812 data: 0.0871 max mem: 3539 +Test: [1990/2387] eta: 0:01:16 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1029 (0.3133) acc1: 99.4792 (88.9537) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1843 data: 0.0917 max mem: 3539 +Test: [2000/2387] eta: 0:01:14 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1077 (0.3123) acc1: 99.4792 (89.0055) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1791 data: 0.0875 max mem: 3539 +Test: [2010/2387] eta: 0:01:12 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1077 (0.3113) acc1: 99.4792 (89.0576) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1816 data: 0.0901 max mem: 3539 +Test: [2020/2387] eta: 0:01:11 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1084 (0.3103) acc1: 99.4792 (89.1072) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1892 data: 0.0976 max mem: 3539 +Test: [2030/2387] eta: 0:01:09 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1065 (0.3093) acc1: 99.4792 (89.1580) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1866 data: 0.0949 max mem: 3540 +Test: [2040/2387] eta: 0:01:07 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1048 (0.3083) acc1: 99.4792 (89.2091) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1891 data: 0.0959 max mem: 3540 +Test: [2050/2387] eta: 0:01:05 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1048 (0.3073) acc1: 99.7396 (89.2602) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1869 data: 0.0943 max mem: 3540 +Test: [2060/2387] eta: 0:01:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0837 (0.3062) acc1: 99.7396 (89.3117) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2062 data: 0.1148 max mem: 3540 +Test: [2070/2387] eta: 0:01:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0818 (0.3051) acc1: 100.0000 (89.3630) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2079 data: 0.1145 max mem: 3540 +Test: [2080/2387] eta: 0:00:59 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0813 (0.3040) acc1: 100.0000 (89.4136) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1686 data: 0.0736 max mem: 3540 +Test: [2090/2387] eta: 0:00:57 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0803 (0.3030) acc1: 100.0000 (89.4640) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1979 data: 0.1037 max mem: 3540 +Test: [2100/2387] eta: 0:00:55 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0804 (0.3019) acc1: 100.0000 (89.5140) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1989 data: 0.1052 max mem: 3540 +Test: [2110/2387] eta: 0:00:53 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0811 (0.3009) acc1: 100.0000 (89.5637) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1733 data: 0.0791 max mem: 3540 +Test: [2120/2387] eta: 0:00:51 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0816 (0.2999) acc1: 100.0000 (89.6097) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1959 data: 0.1033 max mem: 3540 +Test: [2130/2387] eta: 0:00:49 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0795 (0.2988) acc1: 100.0000 (89.6579) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1897 data: 0.0989 max mem: 3540 +Test: [2140/2387] eta: 0:00:47 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0758 (0.2980) acc1: 100.0000 (89.6876) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1774 data: 0.0840 max mem: 3540 +Test: [2150/2387] eta: 0:00:45 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0773 (0.2970) acc1: 100.0000 (89.7352) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1843 data: 0.0901 max mem: 3540 +Test: [2160/2387] eta: 0:00:43 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0767 (0.2960) acc1: 100.0000 (89.7802) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1963 data: 0.1024 max mem: 3540 +Test: [2170/2387] eta: 0:00:41 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0765 (0.2950) acc1: 100.0000 (89.8264) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1895 data: 0.0964 max mem: 3540 +Test: [2180/2387] eta: 0:00:40 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0855 (0.2962) acc1: 99.2188 (89.7277) acc5: 100.0000 (100.0000) acc1_neg: 27.344 acc5_neg: 100.000 time: 0.1878 data: 0.0922 max mem: 3540 +Test: [2190/2387] eta: 0:00:38 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1479 (0.2961) acc1: 95.8333 (89.7192) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1946 data: 0.0933 max mem: 3540 +Test: [2200/2387] eta: 0:00:36 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0716 (0.2951) acc1: 100.0000 (89.7659) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1933 data: 0.0970 max mem: 3540 +Test: [2210/2387] eta: 0:00:34 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0624 (0.2940) acc1: 100.0000 (89.8122) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1960 data: 0.1056 max mem: 3540 +Test: [2220/2387] eta: 0:00:32 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0629 (0.2930) acc1: 100.0000 (89.8565) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1999 data: 0.1086 max mem: 3540 +Test: [2230/2387] eta: 0:00:30 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0637 (0.2920) acc1: 100.0000 (89.9020) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1892 data: 0.0979 max mem: 3540 +Test: [2240/2387] eta: 0:00:28 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0656 (0.2910) acc1: 100.0000 (89.9471) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1664 data: 0.0759 max mem: 3540 +Test: [2250/2387] eta: 0:00:26 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0694 (0.2901) acc1: 100.0000 (89.9915) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1851 data: 0.0908 max mem: 3540 +Test: [2260/2387] eta: 0:00:24 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0741 (0.2891) acc1: 100.0000 (90.0358) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2080 data: 0.1132 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0656 (0.2881) acc1: 100.0000 (90.0796) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2094 data: 0.1167 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0620 (0.2873) acc1: 100.0000 (90.1057) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2178 data: 0.1244 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0617 (0.2864) acc1: 100.0000 (90.1488) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2299 data: 0.1371 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0627 (0.2854) acc1: 100.0000 (90.1894) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2221 data: 0.1287 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0636 (0.2845) acc1: 100.0000 (90.2315) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2074 data: 0.1099 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0673 (0.2836) acc1: 100.0000 (90.2676) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2089 data: 0.1121 max mem: 3541 +Test: [2330/2387] eta: 0:00:11 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0730 (0.2827) acc1: 100.0000 (90.3093) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1850 data: 0.0938 max mem: 3541 +Test: [2340/2387] eta: 0:00:09 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0693 (0.2818) acc1: 100.0000 (90.3503) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1596 data: 0.0699 max mem: 3541 +Test: [2350/2387] eta: 0:00:07 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0671 (0.2812) acc1: 100.0000 (90.3730) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1527 data: 0.0632 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0674 (0.2803) acc1: 100.0000 (90.4137) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1472 data: 0.0577 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0713 (0.2794) acc1: 100.0000 (90.4527) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1426 data: 0.0530 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0704 (0.2786) acc1: 100.0000 (90.4915) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1392 data: 0.0494 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0755 (0.2782) acc1: 100.0000 (90.5074) acc5: 100.0000 (100.0000) acc1_neg: 97.674 acc5_neg: 100.000 time: 0.1327 data: 0.0455 max mem: 3541 +Test: Total time: 0:07:39 (0.1927 s / it) +* Overall Acc@1 90.507 Acc@5 100.000 + target == real Acc@1 0.867 Acc@5 100.000 + target == fake Acc@1 96.802 Acc@5 100.000 + loss 0.278 + PR-AUC 0.9682 ROC-AUC 0.6576 EER 38.38% (thr=0.9268) +Accuracy of the network on the 916267 test images: 90.5% +[rank0]:[W726 04:38:39.468340416 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:39:00.410000 1183637 torch/distributed/run.py:766] +W0726 04:39:00.410000 1183637 torch/distributed/run.py:766] ***************************************** +W0726 04:39:00.410000 1183637 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:39:00.410000 1183637 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:39:30.877208001 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:39:30.099406769 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:39:30.136545189 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:39:30.189937311 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:39:30.285574569 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:39:30.334071596 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:39:30.406502095 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:39:30.433695346 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n8_seed7_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed7_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:01:59 acc1_pos: 2.083 acc5_pos: 100.000 loss: 2.3413 (2.3413) acc1: 2.0833 (2.0833) acc5: 100.0000 (100.0000) time: 7.1946 data: 5.7588 max mem: 3531 +Test: [ 10/517] eta: 0:06:13 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.2909 (2.2586) acc1: 1.0417 (2.2964) acc5: 100.0000 (100.0000) time: 0.7359 data: 0.5239 max mem: 3531 +Test: [ 20/517] eta: 0:03:34 acc1_pos: 0.260 acc5_pos: 100.000 loss: 2.3124 (2.3092) acc1: 0.7812 (2.1205) acc5: 100.0000 (100.0000) time: 0.0931 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:02:48 acc1_pos: 2.865 acc5_pos: 100.000 loss: 2.3147 (2.3036) acc1: 0.7812 (1.8481) acc5: 100.0000 (100.0000) time: 0.1331 data: 0.0389 max mem: 3531 +Test: [ 40/517] eta: 0:02:33 acc1_pos: 1.823 acc5_pos: 100.000 loss: 2.3147 (2.3046) acc1: 0.7812 (1.6895) acc5: 100.0000 (100.0000) time: 0.2049 data: 0.1116 max mem: 3531 +Test: [ 50/517] eta: 0:02:23 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.3233 (2.2065) acc1: 0.2604 (6.1989) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2450 data: 0.1521 max mem: 3531 +Test: [ 60/517] eta: 0:02:15 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1176 (1.8611) acc1: 98.4375 (21.4950) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2449 data: 0.1490 max mem: 3531 +Test: [ 70/517] eta: 0:02:08 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0947 (1.6134) acc1: 100.0000 (32.5447) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2348 data: 0.1380 max mem: 3531 +Test: [ 80/517] eta: 0:02:02 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1017 (1.4270) acc1: 100.0000 (40.8661) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2310 data: 0.1363 max mem: 3531 +Test: [ 90/517] eta: 0:01:57 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1000 (1.2813) acc1: 100.0000 (47.3644) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2349 data: 0.1419 max mem: 3531 +Test: [100/517] eta: 0:01:52 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0975 (1.1636) acc1: 100.0000 (52.5758) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2292 data: 0.1375 max mem: 3531 +Test: [110/517] eta: 0:01:48 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0894 (1.0667) acc1: 100.0000 (56.8459) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2299 data: 0.1353 max mem: 3531 +Test: [120/517] eta: 0:01:45 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0894 (0.9860) acc1: 100.0000 (60.4124) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2445 data: 0.1476 max mem: 3531 +Test: [130/517] eta: 0:01:41 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0881 (0.9174) acc1: 100.0000 (63.4343) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2403 data: 0.1454 max mem: 3531 +Test: [140/517] eta: 0:01:38 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0832 (0.8582) acc1: 100.0000 (66.0276) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2421 data: 0.1486 max mem: 3531 +Test: [150/517] eta: 0:01:35 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0824 (0.8069) acc1: 100.0000 (68.2775) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2352 data: 0.1411 max mem: 3531 +Test: [160/517] eta: 0:01:32 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0797 (0.7617) acc1: 100.0000 (70.2478) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2377 data: 0.1448 max mem: 3531 +Test: [170/517] eta: 0:01:29 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0777 (0.7217) acc1: 100.0000 (71.9877) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2392 data: 0.1431 max mem: 3531 +Test: [180/517] eta: 0:01:26 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0791 (0.6863) acc1: 100.0000 (73.5353) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2314 data: 0.1357 max mem: 3531 +Test: [190/517] eta: 0:01:23 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0825 (0.6548) acc1: 100.0000 (74.9209) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2418 data: 0.1481 max mem: 3531 +Test: [200/517] eta: 0:01:20 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0828 (0.6263) acc1: 100.0000 (76.1686) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2350 data: 0.1418 max mem: 3531 +Test: [210/517] eta: 0:01:17 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0887 (0.6016) acc1: 100.0000 (77.2734) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2336 data: 0.1420 max mem: 3532 +Test: [220/517] eta: 0:01:15 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1068 (0.5791) acc1: 99.4792 (78.2805) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2432 data: 0.1485 max mem: 3532 +Test: [230/517] eta: 0:01:13 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0963 (0.5576) acc1: 100.0000 (79.2197) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2743 data: 0.1749 max mem: 3532 +Test: [240/517] eta: 0:01:10 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0918 (0.5386) acc1: 100.0000 (80.0787) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2699 data: 0.1708 max mem: 3532 +Test: [250/517] eta: 0:01:07 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0918 (0.5204) acc1: 100.0000 (80.8713) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2356 data: 0.1398 max mem: 3532 +Test: [260/517] eta: 0:01:05 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0758 (0.5034) acc1: 100.0000 (81.6042) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2457 data: 0.1504 max mem: 3532 +Test: [270/517] eta: 0:01:02 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0869 (0.4883) acc1: 100.0000 (82.2830) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2491 data: 0.1537 max mem: 3532 +Test: [280/517] eta: 0:00:59 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0906 (0.4741) acc1: 100.0000 (82.9135) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2341 data: 0.1353 max mem: 3532 +Test: [290/517] eta: 0:00:58 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0835 (0.4606) acc1: 100.0000 (83.5007) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2899 data: 0.1930 max mem: 3532 +Test: [300/517] eta: 0:00:55 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0797 (0.4479) acc1: 100.0000 (84.0488) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.3233 data: 0.2270 max mem: 3532 +Test: [310/517] eta: 0:00:53 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0858 (0.4363) acc1: 100.0000 (84.5617) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.3334 data: 0.2361 max mem: 3532 +Test: [320/517] eta: 0:00:51 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0858 (0.4253) acc1: 100.0000 (85.0427) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.3143 data: 0.2209 max mem: 3532 +Test: [330/517] eta: 0:00:48 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0763 (0.4147) acc1: 100.0000 (85.4946) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2364 data: 0.1432 max mem: 3532 +Test: [340/517] eta: 0:00:45 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0739 (0.4052) acc1: 100.0000 (85.9176) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2376 data: 0.1420 max mem: 3532 +Test: [350/517] eta: 0:00:43 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0897 (0.3965) acc1: 100.0000 (86.3181) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2462 data: 0.1483 max mem: 3532 +Test: [360/517] eta: 0:00:40 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0932 (0.3881) acc1: 100.0000 (86.6964) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2399 data: 0.1433 max mem: 3532 +Test: [370/517] eta: 0:00:37 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0933 (0.3803) acc1: 100.0000 (87.0536) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2166 data: 0.1202 max mem: 3532 +Test: [380/517] eta: 0:00:34 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1015 (0.3732) acc1: 100.0000 (87.3913) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2043 data: 0.1104 max mem: 3532 +Test: [390/517] eta: 0:00:32 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1117 (0.3671) acc1: 99.7396 (87.6552) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2205 data: 0.1264 max mem: 3532 +Test: [400/517] eta: 0:00:29 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1088 (0.3604) acc1: 99.7396 (87.9604) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2549 data: 0.1610 max mem: 3532 +Test: [410/517] eta: 0:00:27 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1081 (0.3543) acc1: 100.0000 (88.2496) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.3013 data: 0.2076 max mem: 3532 +Test: [420/517] eta: 0:00:25 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1019 (0.3480) acc1: 100.0000 (88.5274) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.3499 data: 0.2538 max mem: 3532 +Test: [430/517] eta: 0:00:22 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0853 (0.3420) acc1: 100.0000 (88.7924) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2856 data: 0.1908 max mem: 3532 +Test: [440/517] eta: 0:00:19 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0929 (0.3365) acc1: 100.0000 (89.0454) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2299 data: 0.1356 max mem: 3533 +Test: [450/517] eta: 0:00:17 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0952 (0.3311) acc1: 100.0000 (89.2883) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2496 data: 0.1554 max mem: 3533 +Test: [460/517] eta: 0:00:14 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0916 (0.3262) acc1: 100.0000 (89.5150) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2449 data: 0.1501 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0977 (0.3213) acc1: 100.0000 (89.7354) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2694 data: 0.1754 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0963 (0.3165) acc1: 100.0000 (89.9488) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2899 data: 0.1942 max mem: 3533 +Test: [490/517] eta: 0:00:07 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0859 (0.3118) acc1: 100.0000 (90.1535) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.3049 data: 0.2086 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0850 (0.3073) acc1: 100.0000 (90.3500) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.3249 data: 0.2323 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0810 (0.3030) acc1: 100.0000 (90.5389) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2846 data: 0.1924 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0786 (0.3004) acc1: 100.0000 (90.6439) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1975 data: 0.1036 max mem: 3533 +Test: Total time: 0:02:15 (0.2612 s / it) +* Overall Acc@1 90.644 Acc@5 100.000 + target == real Acc@1 1.501 Acc@5 100.000 + target == fake Acc@1 99.892 Acc@5 100.000 + loss 0.300 + PR-AUC 0.9321 ROC-AUC 0.6244 EER 41.39% (thr=0.9180) +Accuracy of the network on the 198427 test images: 90.6% +[rank0]:[W726 04:42:02.804487137 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 04:42:23.516000 1184211 torch/distributed/run.py:766] +W0726 04:42:23.516000 1184211 torch/distributed/run.py:766] ***************************************** +W0726 04:42:23.516000 1184211 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:42:23.516000 1184211 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:42:52.627213510 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:42:52.719965874 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:42:52.913782343 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +[rank0]:[W726 04:42:52.167739080 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:42:52.178800795 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:42:52.240089011 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:42:52.270818652 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:42:52.300743681 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n8_seed7_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed7_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:36:21 acc1_pos: 0.260 acc5_pos: 100.000 loss: 1.9784 (1.9784) acc1: 0.2604 (0.2604) acc5: 100.0000 (100.0000) time: 5.6069 data: 3.3998 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/389] eta: 0:03:45 acc1_pos: 7.292 acc5_pos: 100.000 loss: 2.0974 (2.0561) acc1: 0.2604 (1.4915) acc5: 100.0000 (100.0000) time: 0.5961 data: 0.3141 max mem: 3531 +Test: [ 20/389] eta: 0:02:21 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.0974 (2.1318) acc1: 0.0000 (1.0293) acc5: 100.0000 (100.0000) time: 0.1231 data: 0.0310 max mem: 3531 +Test: [ 30/389] eta: 0:02:03 acc1_pos: 0.260 acc5_pos: 100.000 loss: 2.3403 (2.1961) acc1: 0.0000 (0.7056) acc5: 100.0000 (100.0000) time: 0.2067 data: 0.1110 max mem: 3531 +Test: [ 40/389] eta: 0:01:52 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.2969 (2.2030) acc1: 0.0000 (0.5589) acc5: 100.0000 (100.0000) time: 0.2587 data: 0.1654 max mem: 3531 +Test: [ 50/389] eta: 0:01:37 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.3084 (2.2357) acc1: 0.0000 (0.4493) acc5: 100.0000 (100.0000) time: 0.1987 data: 0.1066 max mem: 3531 +Test: [ 60/389] eta: 0:01:26 acc1_pos: 0.260 acc5_pos: 100.000 loss: 2.3362 (2.2169) acc1: 0.0000 (0.4269) acc5: 100.0000 (100.0000) time: 0.1361 data: 0.0395 max mem: 3531 +Test: [ 70/389] eta: 0:01:22 acc1_pos: 12.500 acc5_pos: 100.000 loss: 1.7712 (2.1210) acc1: 1.3021 (1.6725) acc5: 100.0000 (100.0000) time: 0.1849 data: 0.0883 max mem: 3531 +Test: [ 80/389] eta: 0:01:17 acc1_pos: 4.948 acc5_pos: 100.000 loss: 1.5646 (2.0837) acc1: 3.9062 (2.8260) acc5: 100.0000 (100.0000) time: 0.2204 data: 0.1252 max mem: 3531 +Test: [ 90/389] eta: 0:01:13 acc1_pos: 4.167 acc5_pos: 100.000 loss: 1.8022 (2.0548) acc1: 1.8229 (2.8703) acc5: 100.0000 (100.0000) time: 0.1983 data: 0.1027 max mem: 3531 +Test: [100/389] eta: 0:01:07 acc1_pos: 54.688 acc5_pos: 100.000 loss: 1.7069 (2.0147) acc1: 4.1667 (4.0377) acc5: 100.0000 (100.0000) time: 0.1569 data: 0.0642 max mem: 3531 +Test: [110/389] eta: 0:01:04 acc1_pos: 8.594 acc5_pos: 100.000 loss: 1.6554 (1.9827) acc1: 7.5521 (4.8494) acc5: 100.0000 (100.0000) time: 0.1715 data: 0.0806 max mem: 3531 +Test: [120/389] eta: 0:01:02 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.7330 (1.9691) acc1: 2.0833 (5.0189) acc5: 100.0000 (100.0000) time: 0.2356 data: 0.1405 max mem: 3531 +Test: [130/389] eta: 0:00:58 acc1_pos: 1.227 acc5_pos: 100.000 loss: 1.4095 (1.8600) acc1: 16.4062 (10.3431) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1867 data: 0.0914 max mem: 3531 +Test: [140/389] eta: 0:00:55 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1327 (1.7358) acc1: 100.0000 (16.7018) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1724 data: 0.0813 max mem: 3531 +Test: [150/389] eta: 0:00:53 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1012 (1.6304) acc1: 100.0000 (22.2130) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2130 data: 0.1227 max mem: 3531 +Test: [160/389] eta: 0:00:49 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1122 (1.5472) acc1: 100.0000 (26.2050) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1528 data: 0.0593 max mem: 3531 +Test: [170/389] eta: 0:00:47 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1013 (1.4633) acc1: 100.0000 (30.5038) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1506 data: 0.0563 max mem: 3531 +Test: [180/389] eta: 0:00:44 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1033 (1.3904) acc1: 100.0000 (34.1721) acc5: 100.0000 (100.0000) acc1_neg: 69.010 acc5_neg: 100.000 time: 0.1956 data: 0.1034 max mem: 3531 +Test: [190/389] eta: 0:00:41 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1131 (1.3315) acc1: 100.0000 (37.2110) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1584 data: 0.0638 max mem: 3531 +Test: [200/389] eta: 0:00:39 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1229 (1.2713) acc1: 99.7396 (40.2907) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1441 data: 0.0498 max mem: 3531 +Test: [210/389] eta: 0:00:36 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.0906 (1.2154) acc1: 100.0000 (43.1206) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1643 data: 0.0728 max mem: 3532 +Test: [220/389] eta: 0:00:34 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.0900 (1.1653) acc1: 100.0000 (45.6860) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1627 data: 0.0714 max mem: 3532 +Test: [230/389] eta: 0:00:31 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1320 (1.1232) acc1: 100.0000 (47.9561) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1605 data: 0.0688 max mem: 3532 +Test: [240/389] eta: 0:00:29 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1683 (1.0839) acc1: 99.7396 (50.0411) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1696 data: 0.0763 max mem: 3532 +Test: [250/389] eta: 0:00:27 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1410 (1.0461) acc1: 100.0000 (52.0294) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1838 data: 0.0908 max mem: 3532 +Test: [260/389] eta: 0:00:25 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1047 (1.0141) acc1: 100.0000 (53.6528) acc5: 100.0000 (100.0000) acc1_neg: 76.042 acc5_neg: 100.000 time: 0.1931 data: 0.1003 max mem: 3532 +Test: [270/389] eta: 0:00:23 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1258 (0.9858) acc1: 100.0000 (55.0700) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1824 data: 0.0892 max mem: 3532 +Test: [280/389] eta: 0:00:21 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1256 (0.9573) acc1: 99.7396 (56.5086) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1917 data: 0.1006 max mem: 3532 +Test: [290/389] eta: 0:00:19 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1274 (0.9302) acc1: 100.0000 (57.9002) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2025 data: 0.1103 max mem: 3532 +Test: [300/389] eta: 0:00:17 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1254 (0.9039) acc1: 100.0000 (59.2902) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2027 data: 0.1104 max mem: 3532 +Test: [310/389] eta: 0:00:15 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1338 (0.8798) acc1: 100.0000 (60.5674) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2052 data: 0.1090 max mem: 3532 +Test: [320/389] eta: 0:00:13 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1286 (0.8564) acc1: 100.0000 (61.7853) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1891 data: 0.0919 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1156 (0.8341) acc1: 100.0000 (62.9296) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1895 data: 0.0968 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1114 (0.8126) acc1: 100.0000 (64.0052) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1828 data: 0.0903 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1225 (0.7951) acc1: 100.0000 (64.9187) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1661 data: 0.0731 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1441 (0.7771) acc1: 99.7396 (65.8840) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1523 data: 0.0603 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1480 (0.7616) acc1: 99.7396 (66.7614) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1560 data: 0.0653 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1625 (0.7453) acc1: 99.7396 (67.6304) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1674 data: 0.0771 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1281 (0.7329) acc1: 100.0000 (68.2662) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1525 data: 0.0588 max mem: 3532 +Test: Total time: 0:01:15 (0.1929 s / it) +* Overall Acc@1 68.266 Acc@5 100.000 + target == real Acc@1 4.960 Acc@5 100.000 + target == fake Acc@1 97.692 Acc@5 100.000 + loss 0.733 + PR-AUC 0.7211 ROC-AUC 0.5791 EER 44.14% (thr=0.8906) +Accuracy of the network on the 149358 test images: 68.3% +[rank0]:[W726 04:44:19.543914512 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:44:38.463000 1184769 torch/distributed/run.py:766] +W0726 04:44:38.463000 1184769 torch/distributed/run.py:766] ***************************************** +W0726 04:44:38.463000 1184769 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:44:38.463000 1184769 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:45:07.076615316 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:45:07.181283603 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:45:07.319980718 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:45:07.324346738 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 6): env:// +| distributed init (rank 5): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:45:08.829602147 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W726 04:45:08.829655456 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:45:08.833775818 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:45:08.841781969 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n8_seed7_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed7_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:46:33 acc1_pos: 5.208 acc5_pos: 100.000 loss: 1.9434 (1.9434) acc1: 5.2083 (5.2083) acc5: 100.0000 (100.0000) time: 8.3899 data: 6.1613 max mem: 3531 +Test: [ 10/333] eta: 0:04:35 acc1_pos: 8.854 acc5_pos: 100.000 loss: 1.8298 (1.8244) acc1: 7.5521 (9.6591) acc5: 100.0000 (100.0000) time: 0.8518 data: 0.5606 max mem: 3531 +Test: [ 20/333] eta: 0:02:34 acc1_pos: 16.667 acc5_pos: 100.000 loss: 1.7926 (1.7941) acc1: 8.3333 (10.7887) acc5: 100.0000 (100.0000) time: 0.0989 data: 0.0005 max mem: 3531 +Test: [ 30/333] eta: 0:02:07 acc1_pos: 11.198 acc5_pos: 100.000 loss: 1.8020 (1.8160) acc1: 11.4583 (10.6855) acc5: 100.0000 (100.0000) time: 0.1849 data: 0.0855 max mem: 3531 +Test: [ 40/333] eta: 0:01:46 acc1_pos: 10.156 acc5_pos: 100.000 loss: 1.8534 (1.8185) acc1: 10.9375 (10.7597) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1284 max mem: 3531 +Test: [ 50/333] eta: 0:01:33 acc1_pos: 9.896 acc5_pos: 100.000 loss: 1.8551 (1.8266) acc1: 9.3750 (10.4065) acc5: 100.0000 (100.0000) time: 0.1899 data: 0.0928 max mem: 3531 +Test: [ 60/333] eta: 0:01:23 acc1_pos: 3.906 acc5_pos: 100.000 loss: 1.8591 (1.8296) acc1: 9.8958 (10.4850) acc5: 100.0000 (100.0000) time: 0.1849 data: 0.0893 max mem: 3531 +Test: [ 70/333] eta: 0:01:15 acc1_pos: 10.156 acc5_pos: 100.000 loss: 1.8528 (1.8266) acc1: 10.4167 (10.9375) acc5: 100.0000 (100.0000) time: 0.1701 data: 0.0779 max mem: 3531 +Test: [ 80/333] eta: 0:01:07 acc1_pos: 10.677 acc5_pos: 100.000 loss: 1.8528 (1.8263) acc1: 10.6771 (11.0211) acc5: 100.0000 (100.0000) time: 0.1499 data: 0.0580 max mem: 3531 +Test: [ 90/333] eta: 0:01:01 acc1_pos: 22.917 acc5_pos: 100.000 loss: 1.6421 (1.8042) acc1: 12.7604 (11.7531) acc5: 100.0000 (100.0000) time: 0.1409 data: 0.0476 max mem: 3531 +Test: [100/333] eta: 0:00:57 acc1_pos: 24.740 acc5_pos: 100.000 loss: 1.5671 (1.7813) acc1: 18.7500 (12.5129) acc5: 100.0000 (100.0000) time: 0.1649 data: 0.0685 max mem: 3531 +Test: [110/333] eta: 0:00:53 acc1_pos: 21.615 acc5_pos: 100.000 loss: 1.5594 (1.7630) acc1: 18.7500 (13.1898) acc5: 100.0000 (100.0000) time: 0.1787 data: 0.0837 max mem: 3531 +Test: [120/333] eta: 0:00:49 acc1_pos: 13.281 acc5_pos: 100.000 loss: 1.5573 (1.7470) acc1: 18.2292 (13.7956) acc5: 100.0000 (100.0000) time: 0.1549 data: 0.0608 max mem: 3531 +Test: [130/333] eta: 0:00:45 acc1_pos: 17.708 acc5_pos: 100.000 loss: 1.5303 (1.7300) acc1: 18.2292 (14.3627) acc5: 100.0000 (100.0000) time: 0.1454 data: 0.0520 max mem: 3531 +Test: [140/333] eta: 0:00:43 acc1_pos: 23.698 acc5_pos: 100.000 loss: 1.5242 (1.7216) acc1: 21.0938 (14.7846) acc5: 100.0000 (100.0000) time: 0.1815 data: 0.0889 max mem: 3531 +Test: [150/333] eta: 0:00:40 acc1_pos: 16.146 acc5_pos: 100.000 loss: 1.5852 (1.7122) acc1: 17.4479 (15.1231) acc5: 100.0000 (100.0000) time: 0.1744 data: 0.0812 max mem: 3531 +Test: [160/333] eta: 0:00:36 acc1_pos: 13.816 acc5_pos: 100.000 loss: 1.5402 (1.6710) acc1: 21.8750 (17.1131) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1357 data: 0.0436 max mem: 3531 +Test: [170/333] eta: 0:00:34 acc1_pos: 13.816 acc5_pos: 100.000 loss: 0.3422 (1.5913) acc1: 85.1562 (21.2719) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1649 data: 0.0708 max mem: 3531 +Test: [180/333] eta: 0:00:32 acc1_pos: 13.816 acc5_pos: 100.000 loss: 0.2838 (1.5184) acc1: 89.5833 (25.1338) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1831 data: 0.0876 max mem: 3531 +Test: [190/333] eta: 0:00:29 acc1_pos: 13.816 acc5_pos: 100.000 loss: 0.2302 (1.4497) acc1: 92.7083 (28.7958) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1749 data: 0.0805 max mem: 3531 +Test: [200/333] eta: 0:00:27 acc1_pos: 13.816 acc5_pos: 100.000 loss: 0.2090 (1.3903) acc1: 94.0104 (31.9224) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1801 data: 0.0856 max mem: 3531 +Test: [210/333] eta: 0:00:25 acc1_pos: 13.816 acc5_pos: 100.000 loss: 0.2922 (1.3396) acc1: 88.2812 (34.5429) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.1678 data: 0.0725 max mem: 3532 +Test: [220/333] eta: 0:00:23 acc1_pos: 13.816 acc5_pos: 100.000 loss: 0.3030 (1.2963) acc1: 87.7604 (36.7364) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1797 data: 0.0858 max mem: 3532 +Test: [230/333] eta: 0:00:21 acc1_pos: 13.816 acc5_pos: 100.000 loss: 0.2562 (1.2503) acc1: 92.1875 (39.1967) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2020 data: 0.1076 max mem: 3532 +Test: [240/333] eta: 0:00:19 acc1_pos: 13.816 acc5_pos: 100.000 loss: 0.2213 (1.2078) acc1: 92.9688 (41.4570) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2043 data: 0.1062 max mem: 3532 +Test: [250/333] eta: 0:00:17 acc1_pos: 13.816 acc5_pos: 100.000 loss: 0.2247 (1.1693) acc1: 92.9688 (43.4896) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2322 data: 0.1356 max mem: 3532 +Test: [260/333] eta: 0:00:15 acc1_pos: 13.816 acc5_pos: 100.000 loss: 0.2354 (1.1340) acc1: 92.1875 (45.3285) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.2349 data: 0.1417 max mem: 3532 +Test: [270/333] eta: 0:00:12 acc1_pos: 13.816 acc5_pos: 100.000 loss: 0.2451 (1.1011) acc1: 92.1875 (47.0566) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1825 data: 0.0872 max mem: 3532 +Test: [280/333] eta: 0:00:10 acc1_pos: 13.816 acc5_pos: 100.000 loss: 0.2564 (1.0743) acc1: 90.8854 (48.4255) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1806 data: 0.0846 max mem: 3532 +Test: [290/333] eta: 0:00:08 acc1_pos: 13.816 acc5_pos: 100.000 loss: 0.3895 (1.0508) acc1: 82.2917 (49.5928) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.2186 data: 0.1263 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 13.816 acc5_pos: 100.000 loss: 0.3142 (1.0258) acc1: 87.2396 (50.9188) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.2241 data: 0.1296 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 13.816 acc5_pos: 100.000 loss: 0.2751 (1.0004) acc1: 90.8854 (52.2726) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2061 data: 0.1116 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 13.816 acc5_pos: 100.000 loss: 0.2280 (0.9771) acc1: 93.4896 (53.4820) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.2074 data: 0.1163 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 13.816 acc5_pos: 100.000 loss: 0.4323 (0.9617) acc1: 79.4271 (54.1785) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.1928 data: 0.1027 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 13.816 acc5_pos: 100.000 loss: 0.4406 (0.9586) acc1: 77.6042 (54.3158) acc5: 100.0000 (100.0000) acc1_neg: 77.465 acc5_neg: 100.000 time: 0.1900 data: 0.0971 max mem: 3532 +Test: Total time: 0:01:08 (0.2057 s / it) +* Overall Acc@1 54.316 Acc@5 100.000 + target == real Acc@1 15.282 Acc@5 100.000 + target == fake Acc@1 89.063 Acc@5 100.000 + loss 0.959 + PR-AUC 0.5494 ROC-AUC 0.5341 EER 47.47% (thr=0.8584) +Accuracy of the network on the 127843 test images: 54.3% +[rank0]:[W726 04:46:25.369467887 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:46:47.994000 1185328 torch/distributed/run.py:766] +W0726 04:46:47.994000 1185328 torch/distributed/run.py:766] ***************************************** +W0726 04:46:47.994000 1185328 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:46:47.994000 1185328 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:47:23.736878542 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:47:23.763231637 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:47:23.180732241 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:47:23.307491624 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:47:23.307573678 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 3): env:// +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:47:23.417402036 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W726 04:47:23.417563669 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank1]:[W726 04:47:23.417635405 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n8_seed7_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed7_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:28:01 acc1_pos: 94.792 acc5_pos: 100.000 loss: 0.2148 (0.2148) acc1: 94.7917 (94.7917) acc5: 100.0000 (100.0000) time: 7.1559 data: 4.9520 max mem: 3531 +Test: [ 10/235] eta: 0:02:45 acc1_pos: 95.573 acc5_pos: 100.000 loss: 0.1987 (0.2009) acc1: 95.5729 (95.1468) acc5: 100.0000 (100.0000) time: 0.7354 data: 0.4506 max mem: 3531 +Test: [ 20/235] eta: 0:01:34 acc1_pos: 94.271 acc5_pos: 100.000 loss: 0.1941 (0.1978) acc1: 95.0521 (95.1761) acc5: 100.0000 (100.0000) time: 0.1016 data: 0.0077 max mem: 3531 +Test: [ 30/235] eta: 0:01:20 acc1_pos: 95.573 acc5_pos: 100.000 loss: 0.1960 (0.1988) acc1: 95.3125 (95.1277) acc5: 100.0000 (100.0000) time: 0.2054 data: 0.1100 max mem: 3531 +Test: [ 40/235] eta: 0:01:09 acc1_pos: 96.094 acc5_pos: 100.000 loss: 0.1958 (0.1978) acc1: 95.5729 (95.2236) acc5: 100.0000 (100.0000) time: 0.2700 data: 0.1772 max mem: 3531 +Test: [ 50/235] eta: 0:01:02 acc1_pos: 94.792 acc5_pos: 100.000 loss: 0.1983 (0.1990) acc1: 95.0521 (95.1900) acc5: 100.0000 (100.0000) time: 0.2552 data: 0.1641 max mem: 3531 +Test: [ 60/235] eta: 0:00:54 acc1_pos: 95.573 acc5_pos: 100.000 loss: 0.2054 (0.2004) acc1: 94.7917 (95.0948) acc5: 100.0000 (100.0000) time: 0.2248 data: 0.1318 max mem: 3531 +Test: [ 70/235] eta: 0:00:50 acc1_pos: 92.308 acc5_pos: 100.000 loss: 0.2189 (0.3909) acc1: 93.4896 (85.1379) acc5: 100.0000 (100.0000) acc1_neg: 14.583 acc5_neg: 100.000 time: 0.2197 data: 0.1279 max mem: 3531 +Test: [ 80/235] eta: 0:00:46 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.7368 (0.5583) acc1: 14.5833 (76.3535) acc5: 100.0000 (100.0000) acc1_neg: 14.583 acc5_neg: 100.000 time: 0.2549 data: 0.1627 max mem: 3531 +Test: [ 90/235] eta: 0:00:40 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.7402 (0.6888) acc1: 13.8021 (69.4626) acc5: 100.0000 (100.0000) acc1_neg: 11.458 acc5_neg: 100.000 time: 0.1793 data: 0.0831 max mem: 3531 +Test: [100/235] eta: 0:00:36 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.7330 (0.7915) acc1: 14.0625 (64.0419) acc5: 100.0000 (100.0000) acc1_neg: 13.542 acc5_neg: 100.000 time: 0.1558 data: 0.0614 max mem: 3531 +Test: [110/235] eta: 0:00:32 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.7252 (0.8756) acc1: 14.0625 (59.5603) acc5: 100.0000 (100.0000) acc1_neg: 16.667 acc5_neg: 100.000 time: 0.1999 data: 0.1073 max mem: 3531 +Test: [120/235] eta: 0:00:29 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.7771 (0.9642) acc1: 11.4583 (55.1524) acc5: 100.0000 (100.0000) acc1_neg: 5.729 acc5_neg: 100.000 time: 0.1940 data: 0.1010 max mem: 3531 +Test: [130/235] eta: 0:00:26 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.9593 (1.0402) acc1: 5.4688 (51.3836) acc5: 100.0000 (100.0000) acc1_neg: 4.948 acc5_neg: 100.000 time: 0.2178 data: 0.1266 max mem: 3531 +Test: [140/235] eta: 0:00:24 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.9775 (1.1077) acc1: 5.4688 (48.1383) acc5: 100.0000 (100.0000) acc1_neg: 4.427 acc5_neg: 100.000 time: 0.2260 data: 0.1337 max mem: 3531 +Test: [150/235] eta: 0:00:21 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.9749 (1.1646) acc1: 5.2083 (45.3194) acc5: 100.0000 (100.0000) acc1_neg: 6.510 acc5_neg: 100.000 time: 0.1920 data: 0.0993 max mem: 3531 +Test: [160/235] eta: 0:00:18 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.9611 (1.2136) acc1: 5.9896 (42.9089) acc5: 100.0000 (100.0000) acc1_neg: 7.552 acc5_neg: 100.000 time: 0.2067 data: 0.1139 max mem: 3531 +Test: [170/235] eta: 0:00:15 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.8924 (1.2311) acc1: 8.3333 (41.7687) acc5: 100.0000 (100.0000) acc1_neg: 23.438 acc5_neg: 100.000 time: 0.2211 data: 0.1291 max mem: 3531 +Test: [180/235] eta: 0:00:13 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.4522 (1.2420) acc1: 25.7812 (40.9502) acc5: 100.0000 (100.0000) acc1_neg: 27.604 acc5_neg: 100.000 time: 0.2084 data: 0.1179 max mem: 3531 +Test: [190/235] eta: 0:00:10 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.4230 (1.2517) acc1: 26.8229 (40.2173) acc5: 100.0000 (100.0000) acc1_neg: 30.469 acc5_neg: 100.000 time: 0.2095 data: 0.1183 max mem: 3531 +Test: [200/235] eta: 0:00:08 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.4272 (1.2613) acc1: 26.5625 (39.4771) acc5: 100.0000 (100.0000) acc1_neg: 23.958 acc5_neg: 100.000 time: 0.1853 data: 0.0945 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.4501 (1.2699) acc1: 25.2604 (38.8157) acc5: 100.0000 (100.0000) acc1_neg: 25.260 acc5_neg: 100.000 time: 0.1472 data: 0.0572 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.4575 (1.2780) acc1: 24.7396 (38.2023) acc5: 100.0000 (100.0000) acc1_neg: 24.219 acc5_neg: 100.000 time: 0.1243 data: 0.0338 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.4382 (1.2848) acc1: 24.7396 (37.6375) acc5: 100.0000 (100.0000) acc1_neg: 24.740 acc5_neg: 100.000 time: 0.1181 data: 0.0270 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.4373 (1.2879) acc1: 25.2604 (37.4837) acc5: 100.0000 (100.0000) acc1_neg: 26.087 acc5_neg: 100.000 time: 0.1120 data: 0.0242 max mem: 3532 +Test: Total time: 0:00:51 (0.2208 s / it) +* Overall Acc@1 37.484 Acc@5 100.000 + target == real Acc@1 95.103 Acc@5 100.000 + target == fake Acc@1 16.644 Acc@5 100.000 + loss 1.288 + PR-AUC 0.8400 ROC-AUC 0.6699 EER 37.96% (thr=0.1251) +Accuracy of the network on the 89879 test images: 37.5% +[rank0]:[W726 04:48:23.110311838 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:48:46.174000 1185877 torch/distributed/run.py:766] +W0726 04:48:46.174000 1185877 torch/distributed/run.py:766] ***************************************** +W0726 04:48:46.174000 1185877 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:48:46.174000 1185877 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:49:22.050661016 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:49:23.551503015 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:49:23.563115967 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:49:23.567621064 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +[rank0]:[W726 04:49:23.617345381 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:49:23.617933523 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +[rank4]:[W726 04:49:23.651344055 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:49:23.667483324 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n8_seed7_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed7_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:22:39 acc1_pos: 1.042 acc5_pos: 100.000 loss: 2.1593 (2.1593) acc1: 1.0417 (1.0417) acc5: 100.0000 (100.0000) time: 5.0541 data: 2.6882 max mem: 3531 +Test: [ 10/269] eta: 0:02:20 acc1_pos: 6.510 acc5_pos: 100.000 loss: 1.9317 (1.9415) acc1: 6.5104 (6.3210) acc5: 100.0000 (100.0000) time: 0.5410 data: 0.2447 max mem: 3531 +Test: [ 20/269] eta: 0:01:24 acc1_pos: 7.292 acc5_pos: 100.000 loss: 1.9286 (1.8927) acc1: 6.5104 (7.4281) acc5: 100.0000 (100.0000) time: 0.1019 data: 0.0125 max mem: 3531 +Test: [ 30/269] eta: 0:01:03 acc1_pos: 10.677 acc5_pos: 100.000 loss: 1.8483 (1.9065) acc1: 6.7708 (7.4345) acc5: 100.0000 (100.0000) time: 0.1152 data: 0.0245 max mem: 3531 +Test: [ 40/269] eta: 0:01:04 acc1_pos: 0.862 acc5_pos: 100.000 loss: 1.6286 (1.5332) acc1: 12.2396 (26.8102) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2227 data: 0.1283 max mem: 3531 +Test: [ 50/269] eta: 0:01:00 acc1_pos: 0.862 acc5_pos: 100.000 loss: 0.1274 (1.2584) acc1: 98.9583 (40.9467) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2898 data: 0.1929 max mem: 3531 +Test: [ 60/269] eta: 0:00:53 acc1_pos: 0.862 acc5_pos: 100.000 loss: 0.1285 (1.0774) acc1: 99.4792 (50.3159) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2012 data: 0.1069 max mem: 3531 +Test: [ 70/269] eta: 0:00:48 acc1_pos: 0.862 acc5_pos: 100.000 loss: 0.1394 (0.9468) acc1: 99.4792 (57.1560) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1662 data: 0.0740 max mem: 3531 +Test: [ 80/269] eta: 0:00:45 acc1_pos: 0.862 acc5_pos: 100.000 loss: 0.1277 (0.8460) acc1: 99.7396 (62.3875) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1901 data: 0.0947 max mem: 3531 +Test: [ 90/269] eta: 0:00:41 acc1_pos: 0.862 acc5_pos: 100.000 loss: 0.1221 (0.7720) acc1: 99.7396 (66.3204) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1932 data: 0.0972 max mem: 3531 +Test: [100/269] eta: 0:00:38 acc1_pos: 0.862 acc5_pos: 100.000 loss: 0.1574 (0.7139) acc1: 97.3958 (69.2734) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1728 data: 0.0805 max mem: 3531 +Test: [110/269] eta: 0:00:35 acc1_pos: 0.862 acc5_pos: 100.000 loss: 0.1923 (0.6664) acc1: 96.0938 (71.6639) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1893 data: 0.0961 max mem: 3531 +Test: [120/269] eta: 0:00:33 acc1_pos: 0.862 acc5_pos: 100.000 loss: 0.1605 (0.6250) acc1: 97.3958 (73.8421) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2060 data: 0.1117 max mem: 3531 +Test: [130/269] eta: 0:00:30 acc1_pos: 0.862 acc5_pos: 100.000 loss: 0.1613 (0.5923) acc1: 97.6562 (75.5944) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1700 data: 0.0766 max mem: 3531 +Test: [140/269] eta: 0:00:28 acc1_pos: 0.862 acc5_pos: 100.000 loss: 0.1832 (0.5647) acc1: 98.1771 (77.0926) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1838 data: 0.0912 max mem: 3531 +Test: [150/269] eta: 0:00:25 acc1_pos: 0.862 acc5_pos: 100.000 loss: 0.1832 (0.5397) acc1: 97.3958 (78.3544) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.2048 data: 0.1113 max mem: 3531 +Test: [160/269] eta: 0:00:23 acc1_pos: 0.862 acc5_pos: 100.000 loss: 0.1857 (0.5175) acc1: 95.5729 (79.4174) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1948 data: 0.1003 max mem: 3531 +Test: [170/269] eta: 0:00:21 acc1_pos: 0.862 acc5_pos: 100.000 loss: 0.1852 (0.4985) acc1: 95.5729 (80.3500) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1949 data: 0.0998 max mem: 3531 +Test: [180/269] eta: 0:00:18 acc1_pos: 0.862 acc5_pos: 100.000 loss: 0.1543 (0.4787) acc1: 97.3958 (81.3522) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1949 data: 0.1010 max mem: 3531 +Test: [190/269] eta: 0:00:16 acc1_pos: 0.862 acc5_pos: 100.000 loss: 0.1456 (0.4626) acc1: 98.9583 (82.2085) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2070 data: 0.1145 max mem: 3531 +Test: [200/269] eta: 0:00:14 acc1_pos: 0.862 acc5_pos: 100.000 loss: 0.1860 (0.4503) acc1: 96.3542 (82.7905) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2053 data: 0.1129 max mem: 3531 +Test: [210/269] eta: 0:00:12 acc1_pos: 0.862 acc5_pos: 100.000 loss: 0.1862 (0.4366) acc1: 94.7917 (83.4740) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1791 data: 0.0882 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 0.862 acc5_pos: 100.000 loss: 0.1296 (0.4240) acc1: 98.9583 (84.0733) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2411 data: 0.1463 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 0.862 acc5_pos: 100.000 loss: 0.1239 (0.4116) acc1: 98.9583 (84.6985) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2904 data: 0.1942 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 0.862 acc5_pos: 100.000 loss: 0.1481 (0.4016) acc1: 98.6979 (85.2351) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1783 data: 0.0833 max mem: 3532 +Test: [250/269] eta: 0:00:03 acc1_pos: 0.862 acc5_pos: 100.000 loss: 0.1567 (0.3928) acc1: 98.6979 (85.7196) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1330 data: 0.0369 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 0.862 acc5_pos: 100.000 loss: 0.1537 (0.3837) acc1: 98.6979 (86.2029) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1649 data: 0.0713 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 0.862 acc5_pos: 100.000 loss: 0.1547 (0.3773) acc1: 98.1771 (86.5093) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1613 data: 0.0662 max mem: 3532 +Test: Total time: 0:00:55 (0.2073 s / it) +* Overall Acc@1 86.509 Acc@5 100.000 + target == real Acc@1 7.328 Acc@5 100.000 + target == fake Acc@1 97.315 Acc@5 100.000 + loss 0.377 + PR-AUC 0.9057 ROC-AUC 0.5933 EER 43.56% (thr=0.8892) +Accuracy of the network on the 103293 test images: 86.5% +[rank0]:[W726 04:50:28.222832832 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T04:50:36Z] eval_done tag=fixed500k_real_n8_seed7_resampled +[2026-07-26T04:50:36Z] eval_start tag=fixed500k_fake_n16_seed7_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed7_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T04:50:36Z +hostname=n116-041-153 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed7_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n16_seed7_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 04:50:49.674000 1186452 torch/distributed/run.py:766] +W0726 04:50:49.674000 1186452 torch/distributed/run.py:766] ***************************************** +W0726 04:50:49.674000 1186452 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:50:49.674000 1186452 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 04:51:18.934105219 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:51:18.423398041 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:51:19.816778007 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:51:19.858587522 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:51:19.885723043 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:51:19.227251555 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:51:19.262271617 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:51:19.431973884 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n16_seed7_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed7_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:23:54 acc1_pos: 47.135 acc5_pos: 100.000 loss: 0.8395 (0.8395) acc1: 47.1354 (47.1354) acc5: 100.0000 (100.0000) time: 6.1298 data: 4.4734 max mem: 3531 +Test: [ 10/234] eta: 0:02:25 acc1_pos: 11.458 acc5_pos: 100.000 loss: 0.8478 (0.9163) acc1: 41.6667 (39.5833) acc5: 100.0000 (100.0000) time: 0.6484 data: 0.4071 max mem: 3531 +Test: [ 20/234] eta: 0:01:28 acc1_pos: 97.656 acc5_pos: 100.000 loss: 0.8831 (0.8888) acc1: 37.7604 (42.0759) acc5: 100.0000 (100.0000) time: 0.1301 data: 0.0326 max mem: 3531 +Test: [ 30/234] eta: 0:01:14 acc1_pos: 85.677 acc5_pos: 100.000 loss: 0.1627 (0.6589) acc1: 95.3125 (59.3498) acc5: 100.0000 (100.0000) time: 0.2108 data: 0.1161 max mem: 3531 +Test: [ 40/234] eta: 0:01:05 acc1_pos: 98.698 acc5_pos: 100.000 loss: 0.1461 (0.5584) acc1: 97.6562 (67.2256) acc5: 100.0000 (100.0000) time: 0.2554 data: 0.1633 max mem: 3531 +Test: [ 50/234] eta: 0:00:57 acc1_pos: 99.219 acc5_pos: 100.000 loss: 0.1226 (0.4717) acc1: 98.1771 (73.3558) acc5: 100.0000 (100.0000) time: 0.2359 data: 0.1447 max mem: 3531 +Test: [ 60/234] eta: 0:00:52 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1295 (0.4183) acc1: 99.4792 (77.7066) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2250 data: 0.1300 max mem: 3531 +Test: [ 70/234] eta: 0:00:46 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1486 (0.4356) acc1: 99.7396 (77.3254) acc5: 100.0000 (100.0000) acc1_neg: 59.115 acc5_neg: 100.000 time: 0.2029 data: 0.1069 max mem: 3531 +Test: [ 80/234] eta: 0:00:41 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.7057 (0.4489) acc1: 63.8021 (77.0833) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1793 data: 0.0858 max mem: 3531 +Test: [ 90/234] eta: 0:00:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2967 (0.4342) acc1: 84.8958 (78.4140) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.1899 data: 0.0974 max mem: 3531 +Test: [100/234] eta: 0:00:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4043 (0.4434) acc1: 79.1667 (78.0812) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2099 data: 0.1183 max mem: 3531 +Test: [110/234] eta: 0:00:31 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6037 (0.4527) acc1: 68.7500 (77.6863) acc5: 100.0000 (100.0000) acc1_neg: 55.729 acc5_neg: 100.000 time: 0.2010 data: 0.1108 max mem: 3531 +Test: [120/234] eta: 0:00:28 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6095 (0.4655) acc1: 66.9271 (77.1522) acc5: 100.0000 (100.0000) acc1_neg: 66.927 acc5_neg: 100.000 time: 0.1949 data: 0.1037 max mem: 3531 +Test: [130/234] eta: 0:00:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3846 (0.4631) acc1: 69.5312 (77.6201) acc5: 100.0000 (100.0000) acc1_neg: 69.271 acc5_neg: 100.000 time: 0.2065 data: 0.1117 max mem: 3531 +Test: [140/234] eta: 0:00:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5974 (0.5106) acc1: 69.2708 (75.3823) acc5: 100.0000 (100.0000) acc1_neg: 19.271 acc5_neg: 100.000 time: 0.2063 data: 0.1127 max mem: 3531 +Test: [150/234] eta: 0:00:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6301 (0.5257) acc1: 66.4062 (74.7361) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.2185 data: 0.1282 max mem: 3531 +Test: [160/234] eta: 0:00:17 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4912 (0.5173) acc1: 76.3021 (75.4594) acc5: 100.0000 (100.0000) acc1_neg: 80.729 acc5_neg: 100.000 time: 0.2159 data: 0.1241 max mem: 3531 +Test: [170/234] eta: 0:00:15 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4611 (0.5150) acc1: 80.7292 (75.7188) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.2057 data: 0.1143 max mem: 3531 +Test: [180/234] eta: 0:00:12 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5145 (0.5297) acc1: 76.0417 (74.8547) acc5: 100.0000 (100.0000) acc1_neg: 27.604 acc5_neg: 100.000 time: 0.1975 data: 0.1044 max mem: 3531 +Test: [190/234] eta: 0:00:10 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.2055 (0.5668) acc1: 36.7188 (72.7694) acc5: 100.0000 (100.0000) acc1_neg: 35.156 acc5_neg: 100.000 time: 0.1942 data: 0.1006 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.2231 (0.5654) acc1: 35.1562 (72.9361) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1988 data: 0.1054 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1350 (0.5472) acc1: 99.2188 (74.0842) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1737 data: 0.0805 max mem: 3532 +Test: [220/234] eta: 0:00:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2722 (0.5357) acc1: 91.9271 (74.8221) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.1383 data: 0.0475 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3939 (0.5398) acc1: 70.3125 (74.4871) acc5: 100.0000 (100.0000) acc1_neg: 64.323 acc5_neg: 100.000 time: 0.1265 data: 0.0358 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6152 (0.5406) acc1: 68.2292 (74.4141) acc5: 100.0000 (100.0000) acc1_neg: 73.152 acc5_neg: 100.000 time: 0.1241 data: 0.0315 max mem: 3532 +Test: Total time: 0:00:50 (0.2176 s / it) +* Overall Acc@1 74.414 Acc@5 100.000 + target == real Acc@1 73.552 Acc@5 100.000 + target == fake Acc@1 74.657 Acc@5 100.000 + loss 0.541 + PR-AUC 0.9442 ROC-AUC 0.8345 EER 25.90% (thr=0.5098) +Accuracy of the network on the 89729 test images: 74.4% +[rank0]:[W726 04:52:18.581329233 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:52:37.238000 1187012 torch/distributed/run.py:766] +W0726 04:52:37.238000 1187012 torch/distributed/run.py:766] ***************************************** +W0726 04:52:37.238000 1187012 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:52:37.238000 1187012 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:53:05.460544069 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +[rank7]:[W726 04:53:06.959406804 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:53:06.977996942 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:53:06.469097078 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:53:06.478663791 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +| distributed init (rank 2): env:// +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:53:07.501661555 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank1]:[W726 04:53:07.501743694 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W726 04:53:07.507480433 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n16_seed7_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed7_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:35:10 acc1_pos: 27.344 acc5_pos: 100.000 loss: 1.1033 (1.1033) acc1: 27.3438 (27.3438) acc5: 100.0000 (100.0000) time: 6.9167 data: 5.3564 max mem: 3531 +Test: [ 10/2387] eta: 0:28:09 acc1_pos: 13.281 acc5_pos: 100.000 loss: 1.1054 (1.1599) acc1: 21.3542 (21.2595) acc5: 100.0000 (100.0000) time: 0.7108 data: 0.4873 max mem: 3531 +Test: [ 20/2387] eta: 0:16:29 acc1_pos: 13.802 acc5_pos: 100.000 loss: 1.1054 (1.1287) acc1: 21.3542 (23.4871) acc5: 100.0000 (100.0000) time: 0.0930 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:14:24 acc1_pos: 2.865 acc5_pos: 100.000 loss: 1.2299 (1.1827) acc1: 14.3229 (20.9845) acc5: 100.0000 (100.0000) time: 0.1778 data: 0.0850 max mem: 3531 +Test: [ 40/2387] eta: 0:13:37 acc1_pos: 8.854 acc5_pos: 100.000 loss: 1.4008 (1.2482) acc1: 8.8542 (17.6067) acc5: 100.0000 (100.0000) time: 0.2749 data: 0.1833 max mem: 3531 +Test: [ 50/2387] eta: 0:12:54 acc1_pos: 34.115 acc5_pos: 100.000 loss: 1.3314 (1.2380) acc1: 12.5000 (18.2751) acc5: 100.0000 (100.0000) time: 0.2760 data: 0.1821 max mem: 3531 +Test: [ 60/2387] eta: 0:12:22 acc1_pos: 31.771 acc5_pos: 100.000 loss: 1.3243 (1.2556) acc1: 15.3646 (17.8535) acc5: 100.0000 (100.0000) time: 0.2599 data: 0.1621 max mem: 3531 +Test: [ 70/2387] eta: 0:11:53 acc1_pos: 26.302 acc5_pos: 100.000 loss: 1.3907 (1.2837) acc1: 13.2812 (16.9748) acc5: 100.0000 (100.0000) time: 0.2488 data: 0.1499 max mem: 3531 +Test: [ 80/2387] eta: 0:11:31 acc1_pos: 10.677 acc5_pos: 100.000 loss: 1.3790 (1.2874) acc1: 12.2396 (16.3902) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1416 max mem: 3531 +Test: [ 90/2387] eta: 0:11:13 acc1_pos: 13.281 acc5_pos: 100.000 loss: 1.2689 (1.2851) acc1: 14.3229 (16.5035) acc5: 100.0000 (100.0000) time: 0.2405 data: 0.1435 max mem: 3531 +Test: [ 100/2387] eta: 0:11:09 acc1_pos: 14.323 acc5_pos: 100.000 loss: 1.2901 (1.2949) acc1: 14.3229 (15.9679) acc5: 100.0000 (100.0000) time: 0.2649 data: 0.1713 max mem: 3531 +Test: [ 110/2387] eta: 0:10:45 acc1_pos: 13.281 acc5_pos: 100.000 loss: 1.2901 (1.2902) acc1: 13.2812 (15.9206) acc5: 100.0000 (100.0000) time: 0.2393 data: 0.1463 max mem: 3531 +Test: [ 120/2387] eta: 0:10:34 acc1_pos: 19.792 acc5_pos: 100.000 loss: 1.2728 (1.2899) acc1: 14.5833 (16.0210) acc5: 100.0000 (100.0000) time: 0.2149 data: 0.1202 max mem: 3531 +Test: [ 130/2387] eta: 0:10:23 acc1_pos: 13.021 acc5_pos: 100.000 loss: 1.2963 (1.2933) acc1: 13.0208 (15.7085) acc5: 100.0000 (100.0000) time: 0.2356 data: 0.1417 max mem: 3531 +Test: [ 140/2387] eta: 0:10:13 acc1_pos: 14.844 acc5_pos: 100.000 loss: 1.2952 (1.2933) acc1: 13.0208 (15.5918) acc5: 100.0000 (100.0000) time: 0.2303 data: 0.1376 max mem: 3531 +Test: [ 150/2387] eta: 0:10:11 acc1_pos: 2.865 acc5_pos: 100.000 loss: 1.2819 (1.2915) acc1: 14.8438 (15.5612) acc5: 100.0000 (100.0000) time: 0.2553 data: 0.1633 max mem: 3531 +Test: [ 160/2387] eta: 0:10:01 acc1_pos: 3.653 acc5_pos: 100.000 loss: 1.2213 (1.2698) acc1: 15.6250 (17.2748) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2495 data: 0.1560 max mem: 3531 +Test: [ 170/2387] eta: 0:09:52 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.2148 (1.2076) acc1: 98.4375 (22.0395) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2192 data: 0.1234 max mem: 3531 +Test: [ 180/2387] eta: 0:09:41 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.2085 (1.1524) acc1: 98.6979 (26.2575) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2111 data: 0.1165 max mem: 3531 +Test: [ 190/2387] eta: 0:09:35 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.2083 (1.1030) acc1: 98.1771 (30.0202) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2185 data: 0.1251 max mem: 3531 +Test: [ 200/2387] eta: 0:09:28 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.2059 (1.0583) acc1: 98.6979 (33.4396) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2287 data: 0.1366 max mem: 3531 +Test: [ 210/2387] eta: 0:09:13 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.2069 (1.0181) acc1: 98.6979 (36.5213) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1807 data: 0.0872 max mem: 3532 +Test: [ 220/2387] eta: 0:09:08 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.2089 (0.9815) acc1: 98.1771 (39.3135) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1818 data: 0.0860 max mem: 3532 +Test: [ 230/2387] eta: 0:09:02 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.2060 (0.9479) acc1: 98.6979 (41.8820) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2221 data: 0.1266 max mem: 3532 +Test: [ 240/2387] eta: 0:08:55 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1957 (0.9163) acc1: 98.9583 (44.2654) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2089 data: 0.1148 max mem: 3532 +Test: [ 250/2387] eta: 0:08:49 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1766 (0.8845) acc1: 99.7396 (46.4818) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2026 data: 0.1075 max mem: 3532 +Test: [ 260/2387] eta: 0:08:40 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1072 (0.8547) acc1: 100.0000 (48.5313) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1890 data: 0.0953 max mem: 3532 +Test: [ 270/2387] eta: 0:08:30 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1072 (0.8273) acc1: 100.0000 (50.4286) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1628 data: 0.0726 max mem: 3532 +Test: [ 280/2387] eta: 0:08:23 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1110 (0.8021) acc1: 100.0000 (52.1853) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1666 data: 0.0740 max mem: 3532 +Test: [ 290/2387] eta: 0:08:19 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1026 (0.7779) acc1: 100.0000 (53.8284) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1921 data: 0.0974 max mem: 3532 +Test: [ 300/2387] eta: 0:08:14 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.0942 (0.7551) acc1: 100.0000 (55.3623) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2042 data: 0.1109 max mem: 3532 +Test: [ 310/2387] eta: 0:08:09 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.0948 (0.7360) acc1: 100.0000 (56.7675) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2059 data: 0.1126 max mem: 3532 +Test: [ 320/2387] eta: 0:08:05 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.2356 (0.7205) acc1: 99.2188 (58.0770) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2090 data: 0.1157 max mem: 3532 +Test: [ 330/2387] eta: 0:07:59 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.2356 (0.7058) acc1: 99.2188 (59.3246) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1897 data: 0.0961 max mem: 3532 +Test: [ 340/2387] eta: 0:07:52 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.2347 (0.6920) acc1: 99.2188 (60.4946) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1661 data: 0.0707 max mem: 3532 +Test: [ 350/2387] eta: 0:07:49 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.2261 (0.6755) acc1: 99.7396 (61.6171) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1851 data: 0.0858 max mem: 3532 +Test: [ 360/2387] eta: 0:07:44 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1350 (0.6615) acc1: 99.7396 (62.6630) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1936 data: 0.0931 max mem: 3532 +Test: [ 370/2387] eta: 0:07:38 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1783 (0.6485) acc1: 99.2188 (63.6413) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1739 data: 0.0786 max mem: 3532 +Test: [ 380/2387] eta: 0:07:32 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1801 (0.6362) acc1: 98.9583 (64.5751) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1665 data: 0.0741 max mem: 3532 +Test: [ 390/2387] eta: 0:07:27 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1780 (0.6245) acc1: 99.2188 (65.4598) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1649 data: 0.0706 max mem: 3532 +Test: [ 400/2387] eta: 0:07:22 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1787 (0.6134) acc1: 99.2188 (66.3004) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1683 data: 0.0717 max mem: 3532 +Test: [ 410/2387] eta: 0:07:17 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1787 (0.6028) acc1: 99.2188 (67.1032) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1612 data: 0.0660 max mem: 3532 +Test: [ 420/2387] eta: 0:07:12 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1773 (0.5928) acc1: 99.2188 (67.8636) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1649 data: 0.0678 max mem: 3532 +Test: [ 430/2387] eta: 0:07:08 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1766 (0.5830) acc1: 99.2188 (68.5971) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1790 data: 0.0813 max mem: 3532 +Test: [ 440/2387] eta: 0:07:04 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1665 (0.5726) acc1: 99.7396 (69.3051) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1777 data: 0.0831 max mem: 3533 +Test: [ 450/2387] eta: 0:07:00 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.0965 (0.5620) acc1: 100.0000 (69.9857) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1758 data: 0.0821 max mem: 3533 +Test: [ 460/2387] eta: 0:06:56 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.0964 (0.5519) acc1: 100.0000 (70.6368) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1707 data: 0.0793 max mem: 3533 +Test: [ 470/2387] eta: 0:06:53 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.0964 (0.5423) acc1: 100.0000 (71.2596) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1836 data: 0.0911 max mem: 3533 +Test: [ 480/2387] eta: 0:06:51 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.0965 (0.5330) acc1: 100.0000 (71.8571) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2077 data: 0.1139 max mem: 3533 +Test: [ 490/2387] eta: 0:06:46 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.0961 (0.5241) acc1: 100.0000 (72.4303) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0830 max mem: 3533 +Test: [ 500/2387] eta: 0:06:41 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.0963 (0.5156) acc1: 100.0000 (72.9806) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1463 data: 0.0548 max mem: 3533 +Test: [ 510/2387] eta: 0:06:38 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.0967 (0.5076) acc1: 100.0000 (73.5088) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1728 data: 0.0803 max mem: 3533 +Test: [ 520/2387] eta: 0:06:35 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1380 (0.5006) acc1: 100.0000 (74.0163) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1882 data: 0.0966 max mem: 3533 +Test: [ 530/2387] eta: 0:06:32 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1468 (0.4940) acc1: 100.0000 (74.5057) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1882 data: 0.0979 max mem: 3533 +Test: [ 540/2387] eta: 0:06:28 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1482 (0.4876) acc1: 100.0000 (74.9769) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1744 data: 0.0837 max mem: 3533 +Test: [ 550/2387] eta: 0:06:25 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1495 (0.4815) acc1: 100.0000 (75.4310) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1587 data: 0.0675 max mem: 3533 +Test: [ 560/2387] eta: 0:06:21 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1501 (0.4756) acc1: 100.0000 (75.8690) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1652 data: 0.0731 max mem: 3533 +Test: [ 570/2387] eta: 0:06:18 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1496 (0.4698) acc1: 100.0000 (76.2916) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1706 data: 0.0770 max mem: 3533 +Test: [ 580/2387] eta: 0:06:15 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1496 (0.4643) acc1: 100.0000 (76.6997) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1805 data: 0.0880 max mem: 3533 +Test: [ 590/2387] eta: 0:06:12 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1503 (0.4590) acc1: 100.0000 (77.0939) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1846 data: 0.0907 max mem: 3533 +Test: [ 600/2387] eta: 0:06:09 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1448 (0.4537) acc1: 100.0000 (77.4729) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1795 data: 0.0850 max mem: 3533 +Test: [ 610/2387] eta: 0:06:06 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1459 (0.4489) acc1: 99.7396 (77.8305) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1720 data: 0.0796 max mem: 3533 +Test: [ 620/2387] eta: 0:06:02 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1566 (0.4442) acc1: 99.2188 (78.1783) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1512 data: 0.0585 max mem: 3533 +Test: [ 630/2387] eta: 0:05:59 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1572 (0.4397) acc1: 99.4792 (78.5134) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1551 data: 0.0639 max mem: 3533 +Test: [ 640/2387] eta: 0:05:56 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1614 (0.4354) acc1: 99.2188 (78.8360) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1740 data: 0.0824 max mem: 3533 +Test: [ 650/2387] eta: 0:05:53 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1610 (0.4311) acc1: 99.2188 (79.1507) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1745 data: 0.0820 max mem: 3533 +Test: [ 660/2387] eta: 0:05:50 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1602 (0.4270) acc1: 99.4792 (79.4582) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1685 data: 0.0772 max mem: 3533 +Test: [ 670/2387] eta: 0:05:48 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1593 (0.4230) acc1: 99.4792 (79.7543) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1777 data: 0.0871 max mem: 3534 +Test: [ 680/2387] eta: 0:05:46 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1550 (0.4190) acc1: 99.4792 (80.0485) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2062 data: 0.1144 max mem: 3534 +Test: [ 690/2387] eta: 0:05:45 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1585 (0.4231) acc1: 99.4792 (79.7693) acc5: 100.0000 (100.0000) acc1_neg: 44.531 acc5_neg: 100.000 time: 0.2202 data: 0.1281 max mem: 3534 +Test: [ 700/2387] eta: 0:05:43 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.8897 (0.4299) acc1: 47.1354 (79.3037) acc5: 100.0000 (100.0000) acc1_neg: 49.219 acc5_neg: 100.000 time: 0.2185 data: 0.1257 max mem: 3534 +Test: [ 710/2387] eta: 0:05:42 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.9057 (0.4366) acc1: 46.8750 (78.8392) acc5: 100.0000 (100.0000) acc1_neg: 46.354 acc5_neg: 100.000 time: 0.2198 data: 0.1261 max mem: 3534 +Test: [ 720/2387] eta: 0:05:39 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.9063 (0.4430) acc1: 46.3542 (78.4006) acc5: 100.0000 (100.0000) acc1_neg: 45.833 acc5_neg: 100.000 time: 0.2063 data: 0.1149 max mem: 3534 +Test: [ 730/2387] eta: 0:05:36 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.9084 (0.4496) acc1: 47.3958 (77.9711) acc5: 100.0000 (100.0000) acc1_neg: 48.698 acc5_neg: 100.000 time: 0.1797 data: 0.0849 max mem: 3534 +Test: [ 740/2387] eta: 0:05:33 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.9084 (0.4558) acc1: 47.3958 (77.5497) acc5: 100.0000 (100.0000) acc1_neg: 40.625 acc5_neg: 100.000 time: 0.1656 data: 0.0706 max mem: 3534 +Test: [ 750/2387] eta: 0:05:31 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.9006 (0.4616) acc1: 47.3958 (77.1496) acc5: 100.0000 (100.0000) acc1_neg: 50.260 acc5_neg: 100.000 time: 0.1810 data: 0.0901 max mem: 3534 +Test: [ 760/2387] eta: 0:05:28 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.8792 (0.4656) acc1: 48.4375 (76.8852) acc5: 100.0000 (100.0000) acc1_neg: 74.740 acc5_neg: 100.000 time: 0.1778 data: 0.0859 max mem: 3534 +Test: [ 770/2387] eta: 0:05:26 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.5493 (0.4647) acc1: 72.9167 (76.9796) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1655 data: 0.0735 max mem: 3534 +Test: [ 780/2387] eta: 0:05:23 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.2176 (0.4615) acc1: 96.0938 (77.2381) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1676 data: 0.0761 max mem: 3534 +Test: [ 790/2387] eta: 0:05:20 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1899 (0.4585) acc1: 97.6562 (77.4603) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1704 data: 0.0794 max mem: 3534 +Test: [ 800/2387] eta: 0:05:18 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.2001 (0.4556) acc1: 97.3958 (77.7014) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1742 data: 0.0839 max mem: 3534 +Test: [ 810/2387] eta: 0:05:15 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.2001 (0.4522) acc1: 98.4375 (77.9596) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1652 data: 0.0749 max mem: 3534 +Test: [ 820/2387] eta: 0:05:12 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1710 (0.4488) acc1: 98.4375 (78.2046) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1655 data: 0.0743 max mem: 3534 +Test: [ 830/2387] eta: 0:05:10 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1776 (0.4461) acc1: 98.1771 (78.4252) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1777 data: 0.0862 max mem: 3534 +Test: [ 840/2387] eta: 0:05:08 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.2589 (0.4443) acc1: 96.6146 (78.6112) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1815 data: 0.0905 max mem: 3534 +Test: [ 850/2387] eta: 0:05:05 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.2988 (0.4428) acc1: 92.1875 (78.7698) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1756 data: 0.0813 max mem: 3534 +Test: [ 860/2387] eta: 0:05:03 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.2864 (0.4410) acc1: 95.0521 (78.9619) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1749 data: 0.0799 max mem: 3534 +Test: [ 870/2387] eta: 0:05:00 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.2864 (0.4393) acc1: 95.8333 (79.1377) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1667 data: 0.0737 max mem: 3534 +Test: [ 880/2387] eta: 0:04:57 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.2909 (0.4377) acc1: 95.0521 (79.2988) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1550 data: 0.0630 max mem: 3534 +Test: [ 890/2387] eta: 0:04:55 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.2023 (0.4347) acc1: 99.2188 (79.5279) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1678 data: 0.0757 max mem: 3534 +Test: [ 900/2387] eta: 0:04:53 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1828 (0.4377) acc1: 97.9167 (79.3462) acc5: 100.0000 (100.0000) acc1_neg: 58.854 acc5_neg: 100.000 time: 0.1950 data: 0.1028 max mem: 3535 +Test: [ 910/2387] eta: 0:04:51 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.8176 (0.4419) acc1: 53.6458 (79.0649) acc5: 100.0000 (100.0000) acc1_neg: 51.562 acc5_neg: 100.000 time: 0.1984 data: 0.1064 max mem: 3535 +Test: [ 920/2387] eta: 0:04:48 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.8307 (0.4461) acc1: 53.6458 (78.7852) acc5: 100.0000 (100.0000) acc1_neg: 52.083 acc5_neg: 100.000 time: 0.1498 data: 0.0558 max mem: 3535 +Test: [ 930/2387] eta: 0:04:46 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.8188 (0.4501) acc1: 53.9062 (78.5230) acc5: 100.0000 (100.0000) acc1_neg: 46.615 acc5_neg: 100.000 time: 0.1449 data: 0.0515 max mem: 3535 +Test: [ 940/2387] eta: 0:04:44 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.8298 (0.4541) acc1: 53.9062 (78.2625) acc5: 100.0000 (100.0000) acc1_neg: 49.740 acc5_neg: 100.000 time: 0.1986 data: 0.1044 max mem: 3535 +Test: [ 950/2387] eta: 0:04:42 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.8330 (0.4580) acc1: 52.3438 (77.9936) acc5: 100.0000 (100.0000) acc1_neg: 52.083 acc5_neg: 100.000 time: 0.2156 data: 0.1205 max mem: 3535 +Test: [ 960/2387] eta: 0:04:40 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.8094 (0.4616) acc1: 53.9062 (77.7519) acc5: 100.0000 (100.0000) acc1_neg: 55.990 acc5_neg: 100.000 time: 0.2029 data: 0.1098 max mem: 3535 +Test: [ 970/2387] eta: 0:04:38 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.8183 (0.4654) acc1: 53.9062 (77.4961) acc5: 100.0000 (100.0000) acc1_neg: 54.167 acc5_neg: 100.000 time: 0.1664 data: 0.0748 max mem: 3535 +Test: [ 980/2387] eta: 0:04:35 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.7869 (0.4649) acc1: 53.9062 (77.5344) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.1645 data: 0.0728 max mem: 3535 +Test: [ 990/2387] eta: 0:04:33 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.2048 (0.4625) acc1: 96.3542 (77.7151) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1758 data: 0.0842 max mem: 3535 +Test: [1000/2387] eta: 0:04:30 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1904 (0.4599) acc1: 98.1771 (77.9046) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1595 data: 0.0682 max mem: 3535 +Test: [1010/2387] eta: 0:04:28 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.2023 (0.4574) acc1: 97.1354 (78.0877) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1642 data: 0.0714 max mem: 3535 +Test: [1020/2387] eta: 0:04:26 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.2028 (0.4551) acc1: 97.1354 (78.2594) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1697 data: 0.0746 max mem: 3535 +Test: [1030/2387] eta: 0:04:24 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1765 (0.4524) acc1: 99.2188 (78.4635) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1656 data: 0.0723 max mem: 3535 +Test: [1040/2387] eta: 0:04:21 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1590 (0.4496) acc1: 100.0000 (78.6683) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1597 data: 0.0691 max mem: 3535 +Test: [1050/2387] eta: 0:04:19 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1556 (0.4469) acc1: 99.7396 (78.8636) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1646 data: 0.0728 max mem: 3535 +Test: [1060/2387] eta: 0:04:17 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1600 (0.4442) acc1: 99.7396 (79.0597) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1782 data: 0.0851 max mem: 3535 +Test: [1070/2387] eta: 0:04:15 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1610 (0.4416) acc1: 99.7396 (79.2457) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1735 data: 0.0817 max mem: 3535 +Test: [1080/2387] eta: 0:04:12 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1610 (0.4391) acc1: 99.4792 (79.4201) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1660 data: 0.0741 max mem: 3535 +Test: [1090/2387] eta: 0:04:10 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1584 (0.4365) acc1: 99.7396 (79.6066) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1704 data: 0.0782 max mem: 3535 +Test: [1100/2387] eta: 0:04:08 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1498 (0.4340) acc1: 99.7396 (79.7871) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1777 data: 0.0860 max mem: 3535 +Test: [1110/2387] eta: 0:04:06 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1592 (0.4318) acc1: 99.4792 (79.9552) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1748 data: 0.0825 max mem: 3535 +Test: [1120/2387] eta: 0:04:04 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1592 (0.4294) acc1: 99.4792 (80.1289) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1683 data: 0.0754 max mem: 3536 +Test: [1130/2387] eta: 0:04:02 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1483 (0.4268) acc1: 99.7396 (80.3030) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1760 data: 0.0841 max mem: 3536 +Test: [1140/2387] eta: 0:04:00 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1503 (0.4246) acc1: 99.7396 (80.4660) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1818 data: 0.0903 max mem: 3536 +Test: [1150/2387] eta: 0:03:58 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1725 (0.4229) acc1: 98.9583 (80.6131) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1830 data: 0.0917 max mem: 3536 +Test: [1160/2387] eta: 0:03:55 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1983 (0.4211) acc1: 98.9583 (80.7608) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1822 data: 0.0919 max mem: 3536 +Test: [1170/2387] eta: 0:03:54 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.2142 (0.4194) acc1: 98.9583 (80.9055) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1858 data: 0.0954 max mem: 3536 +Test: [1180/2387] eta: 0:03:52 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.2443 (0.4180) acc1: 96.3542 (81.0319) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1911 data: 0.0999 max mem: 3536 +Test: [1190/2387] eta: 0:03:50 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.2128 (0.4162) acc1: 97.6562 (81.1813) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1862 data: 0.0940 max mem: 3536 +Test: [1200/2387] eta: 0:03:47 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1656 (0.4140) acc1: 100.0000 (81.3376) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1793 data: 0.0879 max mem: 3536 +Test: [1210/2387] eta: 0:03:45 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1442 (0.4118) acc1: 100.0000 (81.4889) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1777 data: 0.0873 max mem: 3536 +Test: [1220/2387] eta: 0:03:44 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1501 (0.4097) acc1: 99.7396 (81.6316) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1888 data: 0.0978 max mem: 3536 +Test: [1230/2387] eta: 0:03:42 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1512 (0.4076) acc1: 99.2188 (81.7744) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1914 data: 0.0997 max mem: 3536 +Test: [1240/2387] eta: 0:03:40 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1565 (0.4056) acc1: 99.2188 (81.9152) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1851 data: 0.0940 max mem: 3536 +Test: [1250/2387] eta: 0:03:38 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1572 (0.4036) acc1: 99.7396 (82.0566) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1921 data: 0.0981 max mem: 3536 +Test: [1260/2387] eta: 0:03:36 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1352 (0.4014) acc1: 99.7396 (82.1961) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1966 data: 0.1018 max mem: 3536 +Test: [1270/2387] eta: 0:03:34 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1072 (0.3990) acc1: 100.0000 (82.3361) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1918 data: 0.1007 max mem: 3536 +Test: [1280/2387] eta: 0:03:32 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.0928 (0.3966) acc1: 100.0000 (82.4728) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1919 data: 0.1017 max mem: 3536 +Test: [1290/2387] eta: 0:03:30 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1019 (0.3944) acc1: 100.0000 (82.6031) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1963 data: 0.1055 max mem: 3536 +Test: [1300/2387] eta: 0:03:28 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1096 (0.3923) acc1: 99.7396 (82.7318) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1987 data: 0.1076 max mem: 3536 +Test: [1310/2387] eta: 0:03:26 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1135 (0.3903) acc1: 99.7396 (82.8576) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1980 data: 0.1072 max mem: 3536 +Test: [1320/2387] eta: 0:03:24 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1120 (0.3881) acc1: 99.7396 (82.9826) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1875 data: 0.0969 max mem: 3536 +Test: [1330/2387] eta: 0:03:23 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1043 (0.3861) acc1: 100.0000 (83.1066) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1915 data: 0.1006 max mem: 3536 +Test: [1340/2387] eta: 0:03:21 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1181 (0.3855) acc1: 99.7396 (83.1457) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2002 data: 0.1094 max mem: 3536 +Test: [1350/2387] eta: 0:03:19 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.0943 (0.3833) acc1: 100.0000 (83.2705) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1848 data: 0.0945 max mem: 3537 +Test: [1360/2387] eta: 0:03:17 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.0787 (0.3811) acc1: 100.0000 (83.3934) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1723 data: 0.0775 max mem: 3537 +Test: [1370/2387] eta: 0:03:14 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.0787 (0.3789) acc1: 100.0000 (83.5144) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1673 data: 0.0723 max mem: 3537 +Test: [1380/2387] eta: 0:03:12 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.0835 (0.3768) acc1: 100.0000 (83.6337) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1624 data: 0.0714 max mem: 3537 +Test: [1390/2387] eta: 0:03:10 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.0812 (0.3746) acc1: 100.0000 (83.7514) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1614 data: 0.0693 max mem: 3537 +Test: [1400/2387] eta: 0:03:08 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.0679 (0.3724) acc1: 100.0000 (83.8674) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1745 data: 0.0804 max mem: 3537 +Test: [1410/2387] eta: 0:03:06 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.0736 (0.3709) acc1: 100.0000 (83.9660) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1941 data: 0.1007 max mem: 3537 +Test: [1420/2387] eta: 0:03:05 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1920 (0.3698) acc1: 98.4375 (84.0506) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2054 data: 0.1143 max mem: 3537 +Test: [1430/2387] eta: 0:03:03 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1994 (0.3688) acc1: 97.1354 (84.1386) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2114 data: 0.1187 max mem: 3537 +Test: [1440/2387] eta: 0:03:01 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1762 (0.3673) acc1: 98.4375 (84.2431) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2176 data: 0.1219 max mem: 3537 +Test: [1450/2387] eta: 0:02:59 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1539 (0.3659) acc1: 99.4792 (84.3405) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2254 data: 0.1308 max mem: 3537 +Test: [1460/2387] eta: 0:02:58 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1759 (0.3647) acc1: 98.6979 (84.4383) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2268 data: 0.1346 max mem: 3537 +Test: [1470/2387] eta: 0:02:56 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1842 (0.3635) acc1: 98.4375 (84.5251) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2173 data: 0.1215 max mem: 3537 +Test: [1480/2387] eta: 0:02:54 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1847 (0.3623) acc1: 97.9167 (84.6148) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1860 data: 0.0896 max mem: 3537 +Test: [1490/2387] eta: 0:02:52 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1752 (0.3611) acc1: 98.1771 (84.7006) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1912 data: 0.0994 max mem: 3537 +Test: [1500/2387] eta: 0:02:51 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1672 (0.3599) acc1: 98.6979 (84.7883) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2437 data: 0.1525 max mem: 3537 +Test: [1510/2387] eta: 0:02:49 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1359 (0.3583) acc1: 99.7396 (84.8886) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2388 data: 0.1467 max mem: 3537 +Test: [1520/2387] eta: 0:02:47 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1383 (0.3576) acc1: 100.0000 (84.9604) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2407 data: 0.1495 max mem: 3537 +Test: [1530/2387] eta: 0:02:45 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.2623 (0.3573) acc1: 94.2708 (85.0067) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2248 data: 0.1340 max mem: 3537 +Test: [1540/2387] eta: 0:02:44 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.2824 (0.3567) acc1: 93.7500 (85.0728) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2104 data: 0.1194 max mem: 3537 +Test: [1550/2387] eta: 0:02:42 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.2555 (0.3562) acc1: 96.0938 (85.1363) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2211 data: 0.1295 max mem: 3537 +Test: [1560/2387] eta: 0:02:40 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.2389 (0.3556) acc1: 96.3542 (85.1903) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.2000 data: 0.1070 max mem: 3537 +Test: [1570/2387] eta: 0:02:38 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1893 (0.3544) acc1: 98.9583 (85.2821) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1982 data: 0.1047 max mem: 3537 +Test: [1580/2387] eta: 0:02:36 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1579 (0.3534) acc1: 98.9583 (85.3600) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1942 data: 0.1011 max mem: 3538 +Test: [1590/2387] eta: 0:02:34 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1922 (0.3525) acc1: 98.4375 (85.4420) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1985 data: 0.1046 max mem: 3538 +Test: [1600/2387] eta: 0:02:32 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.2060 (0.3516) acc1: 98.1771 (85.5143) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2009 data: 0.1068 max mem: 3538 +Test: [1610/2387] eta: 0:02:30 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1962 (0.3507) acc1: 98.1771 (85.5895) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2000 data: 0.1072 max mem: 3538 +Test: [1620/2387] eta: 0:02:28 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1907 (0.3497) acc1: 98.1771 (85.6662) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1998 data: 0.1080 max mem: 3538 +Test: [1630/2387] eta: 0:02:26 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1717 (0.3487) acc1: 98.6979 (85.7440) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2161 data: 0.1213 max mem: 3538 +Test: [1640/2387] eta: 0:02:25 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1715 (0.3477) acc1: 97.9167 (85.8151) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2297 data: 0.1349 max mem: 3538 +Test: [1650/2387] eta: 0:02:23 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1655 (0.3465) acc1: 98.4375 (85.8948) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2099 data: 0.1159 max mem: 3538 +Test: [1660/2387] eta: 0:02:21 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1581 (0.3455) acc1: 98.9583 (85.9698) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1890 data: 0.0918 max mem: 3538 +Test: [1670/2387] eta: 0:02:19 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1637 (0.3444) acc1: 98.9583 (86.0435) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1902 data: 0.0956 max mem: 3538 +Test: [1680/2387] eta: 0:02:17 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1582 (0.3434) acc1: 98.9583 (86.1087) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2029 data: 0.1110 max mem: 3538 +Test: [1690/2387] eta: 0:02:15 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1572 (0.3424) acc1: 99.2188 (86.1836) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2072 data: 0.1143 max mem: 3538 +Test: [1700/2387] eta: 0:02:13 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1572 (0.3414) acc1: 99.2188 (86.2590) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2116 data: 0.1166 max mem: 3538 +Test: [1710/2387] eta: 0:02:11 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1696 (0.3405) acc1: 98.9583 (86.3299) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2172 data: 0.1218 max mem: 3538 +Test: [1720/2387] eta: 0:02:09 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1887 (0.3396) acc1: 98.6979 (86.4027) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2194 data: 0.1252 max mem: 3538 +Test: [1730/2387] eta: 0:02:08 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.2052 (0.3389) acc1: 98.6979 (86.4675) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2111 data: 0.1191 max mem: 3538 +Test: [1740/2387] eta: 0:02:06 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.2065 (0.3380) acc1: 98.9583 (86.5397) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2020 data: 0.1092 max mem: 3538 +Test: [1750/2387] eta: 0:02:04 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1644 (0.3370) acc1: 99.4792 (86.6123) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2039 data: 0.1108 max mem: 3538 +Test: [1760/2387] eta: 0:02:02 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1257 (0.3358) acc1: 100.0000 (86.6878) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2003 data: 0.1094 max mem: 3538 +Test: [1770/2387] eta: 0:02:00 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1415 (0.3351) acc1: 99.4792 (86.7442) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1961 data: 0.1051 max mem: 3538 +Test: [1780/2387] eta: 0:01:58 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.2141 (0.3345) acc1: 97.1354 (86.8062) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1934 data: 0.1025 max mem: 3538 +Test: [1790/2387] eta: 0:01:56 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.2141 (0.3338) acc1: 98.1771 (86.8681) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1940 data: 0.1023 max mem: 3538 +Test: [1800/2387] eta: 0:01:54 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.2386 (0.3334) acc1: 97.1354 (86.9190) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2062 data: 0.1128 max mem: 3538 +Test: [1810/2387] eta: 0:01:52 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.2186 (0.3325) acc1: 97.6562 (86.9856) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2014 data: 0.1090 max mem: 3539 +Test: [1820/2387] eta: 0:01:50 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1485 (0.3315) acc1: 99.7396 (87.0542) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2176 data: 0.1257 max mem: 3539 +Test: [1830/2387] eta: 0:01:48 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1329 (0.3304) acc1: 100.0000 (87.1244) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2228 data: 0.1294 max mem: 3539 +Test: [1840/2387] eta: 0:01:46 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1152 (0.3292) acc1: 100.0000 (87.1943) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1771 data: 0.0813 max mem: 3539 +Test: [1850/2387] eta: 0:01:44 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1139 (0.3280) acc1: 100.0000 (87.2635) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1815 data: 0.0863 max mem: 3539 +Test: [1860/2387] eta: 0:01:42 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1135 (0.3269) acc1: 100.0000 (87.3318) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1927 data: 0.1001 max mem: 3539 +Test: [1870/2387] eta: 0:01:40 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1135 (0.3257) acc1: 100.0000 (87.3991) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1810 data: 0.0874 max mem: 3539 +Test: [1880/2387] eta: 0:01:38 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1136 (0.3246) acc1: 100.0000 (87.4659) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1853 data: 0.0925 max mem: 3539 +Test: [1890/2387] eta: 0:01:36 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1138 (0.3235) acc1: 100.0000 (87.5322) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1779 data: 0.0875 max mem: 3539 +Test: [1900/2387] eta: 0:01:34 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1150 (0.3224) acc1: 100.0000 (87.5978) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1768 data: 0.0853 max mem: 3539 +Test: [1910/2387] eta: 0:01:32 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1329 (0.3215) acc1: 100.0000 (87.6612) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1891 data: 0.0959 max mem: 3539 +Test: [1920/2387] eta: 0:01:30 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1414 (0.3209) acc1: 100.0000 (87.7058) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1953 data: 0.1017 max mem: 3539 +Test: [1930/2387] eta: 0:01:29 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1739 (0.3201) acc1: 99.7396 (87.7680) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2189 data: 0.1267 max mem: 3539 +Test: [1940/2387] eta: 0:01:27 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1758 (0.3196) acc1: 99.4792 (87.7934) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1932 data: 0.1027 max mem: 3539 +Test: [1950/2387] eta: 0:01:25 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1997 (0.3191) acc1: 98.9583 (87.8506) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2138 data: 0.1235 max mem: 3539 +Test: [1960/2387] eta: 0:01:23 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1846 (0.3183) acc1: 99.2188 (87.9044) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2453 data: 0.1537 max mem: 3539 +Test: [1970/2387] eta: 0:01:21 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1808 (0.3178) acc1: 98.9583 (87.9431) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1519 data: 0.0573 max mem: 3539 +Test: [1980/2387] eta: 0:01:19 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1894 (0.3172) acc1: 98.1771 (87.9886) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1536 data: 0.0580 max mem: 3539 +Test: [1990/2387] eta: 0:01:17 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1634 (0.3164) acc1: 98.1771 (88.0382) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2149 data: 0.1217 max mem: 3539 +Test: [2000/2387] eta: 0:01:15 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1643 (0.3157) acc1: 97.9167 (88.0884) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2149 data: 0.1197 max mem: 3539 +Test: [2010/2387] eta: 0:01:13 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1643 (0.3149) acc1: 98.1771 (88.1389) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2249 data: 0.1274 max mem: 3539 +Test: [2020/2387] eta: 0:01:11 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1632 (0.3142) acc1: 98.1771 (88.1889) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2349 data: 0.1396 max mem: 3539 +Test: [2030/2387] eta: 0:01:09 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1656 (0.3135) acc1: 97.9167 (88.2351) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2299 data: 0.1336 max mem: 3540 +Test: [2040/2387] eta: 0:01:07 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1634 (0.3128) acc1: 98.1771 (88.2844) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2049 data: 0.1078 max mem: 3540 +Test: [2050/2387] eta: 0:01:05 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1621 (0.3120) acc1: 98.4375 (88.3345) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1719 data: 0.0772 max mem: 3540 +Test: [2060/2387] eta: 0:01:04 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1596 (0.3113) acc1: 98.1771 (88.3821) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2102 data: 0.1152 max mem: 3540 +Test: [2070/2387] eta: 0:01:02 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1557 (0.3105) acc1: 98.4375 (88.4315) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2278 data: 0.1321 max mem: 3540 +Test: [2080/2387] eta: 0:01:00 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1544 (0.3098) acc1: 98.6979 (88.4795) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1961 data: 0.1019 max mem: 3540 +Test: [2090/2387] eta: 0:00:58 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1540 (0.3090) acc1: 98.4375 (88.5277) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1788 data: 0.0847 max mem: 3540 +Test: [2100/2387] eta: 0:00:56 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1539 (0.3083) acc1: 98.6979 (88.5769) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1749 data: 0.0809 max mem: 3540 +Test: [2110/2387] eta: 0:00:54 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1537 (0.3075) acc1: 98.6979 (88.6240) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1795 data: 0.0874 max mem: 3540 +Test: [2120/2387] eta: 0:00:52 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1631 (0.3071) acc1: 98.1771 (88.6611) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1803 data: 0.0879 max mem: 3540 +Test: [2130/2387] eta: 0:00:50 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1797 (0.3065) acc1: 98.9583 (88.7107) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1864 data: 0.0935 max mem: 3540 +Test: [2140/2387] eta: 0:00:48 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1816 (0.3061) acc1: 98.9583 (88.7431) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1854 data: 0.0930 max mem: 3540 +Test: [2150/2387] eta: 0:00:46 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.2009 (0.3056) acc1: 98.1771 (88.7878) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1819 data: 0.0903 max mem: 3540 +Test: [2160/2387] eta: 0:00:44 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1712 (0.3050) acc1: 98.6979 (88.8346) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1826 data: 0.0910 max mem: 3540 +Test: [2170/2387] eta: 0:00:42 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1712 (0.3045) acc1: 98.6979 (88.8763) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1908 data: 0.0979 max mem: 3540 +Test: [2180/2387] eta: 0:00:40 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.2161 (0.3056) acc1: 96.6146 (88.8024) acc5: 100.0000 (100.0000) acc1_neg: 36.719 acc5_neg: 100.000 time: 0.1848 data: 0.0930 max mem: 3540 +Test: [2190/2387] eta: 0:00:38 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.2209 (0.3057) acc1: 95.3125 (88.8032) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1790 data: 0.0868 max mem: 3540 +Test: [2200/2387] eta: 0:00:36 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1790 (0.3051) acc1: 99.2188 (88.8523) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1899 data: 0.0971 max mem: 3540 +Test: [2210/2387] eta: 0:00:34 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1733 (0.3045) acc1: 99.7396 (88.8942) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1840 data: 0.0911 max mem: 3540 +Test: [2220/2387] eta: 0:00:32 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1811 (0.3040) acc1: 99.2188 (88.9363) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1803 data: 0.0841 max mem: 3540 +Test: [2230/2387] eta: 0:00:30 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1749 (0.3034) acc1: 99.7396 (88.9801) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1839 data: 0.0898 max mem: 3540 +Test: [2240/2387] eta: 0:00:28 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1598 (0.3028) acc1: 99.7396 (89.0219) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1834 data: 0.0923 max mem: 3540 +Test: [2250/2387] eta: 0:00:26 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1515 (0.3021) acc1: 100.0000 (89.0697) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1716 data: 0.0808 max mem: 3540 +Test: [2260/2387] eta: 0:00:24 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1387 (0.3014) acc1: 100.0000 (89.1164) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1587 data: 0.0684 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1356 (0.3007) acc1: 100.0000 (89.1640) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1580 data: 0.0670 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1390 (0.3001) acc1: 100.0000 (89.1960) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1492 data: 0.0579 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1429 (0.2995) acc1: 100.0000 (89.2416) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1514 data: 0.0604 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1349 (0.2988) acc1: 100.0000 (89.2857) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1619 data: 0.0710 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1348 (0.2981) acc1: 99.7396 (89.3278) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1580 data: 0.0667 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1578 (0.2976) acc1: 99.4792 (89.3663) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1601 data: 0.0682 max mem: 3541 +Test: [2330/2387] eta: 0:00:11 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1671 (0.2973) acc1: 98.4375 (89.3989) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1643 data: 0.0732 max mem: 3541 +Test: [2340/2387] eta: 0:00:09 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.2041 (0.2969) acc1: 98.1771 (89.4374) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1475 data: 0.0572 max mem: 3541 +Test: [2350/2387] eta: 0:00:07 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1954 (0.2967) acc1: 98.4375 (89.4566) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1368 data: 0.0461 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.2019 (0.2964) acc1: 98.6979 (89.4908) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1355 data: 0.0449 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1798 (0.2959) acc1: 98.6979 (89.5275) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1343 data: 0.0439 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1648 (0.2955) acc1: 98.6979 (89.5621) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1333 data: 0.0431 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 3.653 acc5_pos: 100.000 loss: 0.1641 (0.2951) acc1: 98.6979 (89.5823) acc5: 100.0000 (100.0000) acc1_neg: 95.349 acc5_neg: 100.000 time: 0.1248 data: 0.0370 max mem: 3541 +Test: Total time: 0:07:38 (0.1921 s / it) +* Overall Acc@1 89.582 Acc@5 100.000 + target == real Acc@1 15.302 Acc@5 100.000 + target == fake Acc@1 94.799 Acc@5 100.000 + loss 0.295 + PR-AUC 0.9828 ROC-AUC 0.8003 EER 26.54% (thr=0.8105) +Accuracy of the network on the 916267 test images: 89.6% +[rank0]:[W726 05:01:06.372328903 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:01:27.021000 1187631 torch/distributed/run.py:766] +W0726 05:01:27.021000 1187631 torch/distributed/run.py:766] ***************************************** +W0726 05:01:27.021000 1187631 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:01:27.021000 1187631 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:01:59.932546405 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:01:59.191885812 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:01:59.367244251 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:02:00.519350584 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:02:00.653245784 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:02:00.682151821 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:02:00.717139896 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W726 05:02:00.725824812 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n16_seed7_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed7_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 0:58:18 acc1_pos: 27.344 acc5_pos: 100.000 loss: 1.1033 (1.1033) acc1: 27.3438 (27.3438) acc5: 100.0000 (100.0000) time: 6.7665 data: 4.9878 max mem: 3531 +Test: [ 10/517] eta: 0:05:53 acc1_pos: 13.281 acc5_pos: 100.000 loss: 1.1054 (1.1599) acc1: 21.3542 (21.2595) acc5: 100.0000 (100.0000) time: 0.6971 data: 0.4539 max mem: 3531 +Test: [ 20/517] eta: 0:03:24 acc1_pos: 13.802 acc5_pos: 100.000 loss: 1.1054 (1.1287) acc1: 21.3542 (23.4871) acc5: 100.0000 (100.0000) time: 0.0940 data: 0.0005 max mem: 3531 +Test: [ 30/517] eta: 0:03:01 acc1_pos: 2.865 acc5_pos: 100.000 loss: 1.2299 (1.1827) acc1: 14.3229 (20.9845) acc5: 100.0000 (100.0000) time: 0.1940 data: 0.0997 max mem: 3531 +Test: [ 40/517] eta: 0:03:04 acc1_pos: 8.854 acc5_pos: 100.000 loss: 1.4008 (1.2482) acc1: 8.8542 (17.6067) acc5: 100.0000 (100.0000) time: 0.3601 data: 0.2692 max mem: 3531 +Test: [ 50/517] eta: 0:02:50 acc1_pos: 46.789 acc5_pos: 100.000 loss: 1.3314 (1.1917) acc1: 12.5000 (21.7933) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.3547 data: 0.2625 max mem: 3531 +Test: [ 60/517] eta: 0:02:36 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.2780 (1.0356) acc1: 91.4062 (34.0975) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2498 data: 0.1563 max mem: 3531 +Test: [ 70/517] eta: 0:02:30 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.2347 (0.9238) acc1: 97.9167 (42.8917) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2599 data: 0.1660 max mem: 3531 +Test: [ 80/517] eta: 0:02:23 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.2534 (0.8467) acc1: 95.5729 (49.1641) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2899 data: 0.1953 max mem: 3531 +Test: [ 90/517] eta: 0:02:16 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.3018 (0.7840) acc1: 96.6146 (54.4557) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2599 data: 0.1661 max mem: 3531 +Test: [100/517] eta: 0:02:11 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.2442 (0.7307) acc1: 97.6562 (58.7897) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1600 max mem: 3531 +Test: [110/517] eta: 0:02:05 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.1992 (0.6812) acc1: 99.2188 (62.4507) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2649 data: 0.1698 max mem: 3531 +Test: [120/517] eta: 0:02:00 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.1965 (0.6425) acc1: 100.0000 (65.4808) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2499 data: 0.1563 max mem: 3531 +Test: [130/517] eta: 0:01:56 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.2093 (0.6100) acc1: 99.4792 (68.0344) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2600 data: 0.1650 max mem: 3531 +Test: [140/517] eta: 0:01:52 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.2276 (0.5833) acc1: 99.2188 (70.2386) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2649 data: 0.1699 max mem: 3531 +Test: [150/517] eta: 0:01:48 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.2319 (0.5594) acc1: 99.2188 (72.1544) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2548 data: 0.1602 max mem: 3531 +Test: [160/517] eta: 0:01:44 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.2250 (0.5386) acc1: 99.2188 (73.8370) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2649 data: 0.1703 max mem: 3531 +Test: [170/517] eta: 0:01:41 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.2252 (0.5206) acc1: 99.2188 (75.2970) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2662 data: 0.1729 max mem: 3531 +Test: [180/517] eta: 0:01:38 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.2366 (0.5041) acc1: 98.6979 (76.5769) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2699 data: 0.1758 max mem: 3531 +Test: [190/517] eta: 0:01:34 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.1631 (0.4850) acc1: 99.4792 (77.7964) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2646 data: 0.1689 max mem: 3531 +Test: [200/517] eta: 0:01:31 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.1603 (0.4695) acc1: 100.0000 (78.8946) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2612 data: 0.1673 max mem: 3531 +Test: [210/517] eta: 0:01:28 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.1910 (0.4637) acc1: 93.7500 (79.3876) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.2689 data: 0.1742 max mem: 3532 +Test: [220/517] eta: 0:01:25 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.3488 (0.4593) acc1: 88.8021 (79.8265) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2786 data: 0.1829 max mem: 3532 +Test: [230/517] eta: 0:01:22 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.3194 (0.4515) acc1: 92.4479 (80.5443) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2749 data: 0.1787 max mem: 3532 +Test: [240/517] eta: 0:01:18 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.2376 (0.4408) acc1: 98.4375 (81.3256) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2499 data: 0.1500 max mem: 3532 +Test: [250/517] eta: 0:01:15 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.1841 (0.4315) acc1: 99.7396 (82.0385) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2399 data: 0.1411 max mem: 3532 +Test: [260/517] eta: 0:01:12 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.2011 (0.4229) acc1: 99.7396 (82.7107) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2452 data: 0.1515 max mem: 3532 +Test: [270/517] eta: 0:01:09 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.2174 (0.4151) acc1: 99.4792 (83.3228) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2449 data: 0.1534 max mem: 3532 +Test: [280/517] eta: 0:01:06 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.2061 (0.4077) acc1: 98.9583 (83.8755) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2746 data: 0.1818 max mem: 3532 +Test: [290/517] eta: 0:01:03 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.1969 (0.4007) acc1: 98.9583 (84.3992) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2899 data: 0.1920 max mem: 3532 +Test: [300/517] eta: 0:01:00 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.2046 (0.3944) acc1: 98.9583 (84.8759) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2699 data: 0.1711 max mem: 3532 +Test: [310/517] eta: 0:00:57 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.2317 (0.3907) acc1: 98.1771 (85.2534) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2699 data: 0.1728 max mem: 3532 +Test: [320/517] eta: 0:00:54 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.2598 (0.3856) acc1: 98.1771 (85.6568) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2649 data: 0.1690 max mem: 3532 +Test: [330/517] eta: 0:00:52 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.1975 (0.3796) acc1: 98.9583 (86.0618) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2608 data: 0.1689 max mem: 3532 +Test: [340/517] eta: 0:00:49 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.1769 (0.3737) acc1: 99.4792 (86.4599) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2603 data: 0.1684 max mem: 3532 +Test: [350/517] eta: 0:00:46 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.1985 (0.3695) acc1: 99.4792 (86.7722) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2690 data: 0.1751 max mem: 3532 +Test: [360/517] eta: 0:00:43 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.2252 (0.3652) acc1: 97.6562 (87.0787) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2649 data: 0.1710 max mem: 3532 +Test: [370/517] eta: 0:00:40 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.1642 (0.3596) acc1: 100.0000 (87.4256) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2699 data: 0.1752 max mem: 3532 +Test: [380/517] eta: 0:00:37 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.1579 (0.3543) acc1: 100.0000 (87.7529) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2695 data: 0.1715 max mem: 3532 +Test: [390/517] eta: 0:00:35 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.1635 (0.3496) acc1: 100.0000 (88.0621) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2599 data: 0.1621 max mem: 3532 +Test: [400/517] eta: 0:00:32 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.1635 (0.3448) acc1: 99.7396 (88.3572) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2649 data: 0.1690 max mem: 3532 +Test: [410/517] eta: 0:00:29 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.1753 (0.3411) acc1: 100.0000 (88.6380) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2649 data: 0.1692 max mem: 3532 +Test: [420/517] eta: 0:00:26 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.1777 (0.3371) acc1: 100.0000 (88.9023) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2899 data: 0.1943 max mem: 3532 +Test: [430/517] eta: 0:00:23 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.1961 (0.3344) acc1: 100.0000 (89.1477) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2606 data: 0.1657 max mem: 3532 +Test: [440/517] eta: 0:00:21 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.1829 (0.3306) acc1: 99.4792 (89.3826) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2403 data: 0.1464 max mem: 3533 +Test: [450/517] eta: 0:00:18 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.1629 (0.3266) acc1: 99.7396 (89.6168) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2891 data: 0.1942 max mem: 3533 +Test: [460/517] eta: 0:00:15 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.1487 (0.3231) acc1: 100.0000 (89.8404) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.3887 data: 0.2931 max mem: 3533 +Test: [470/517] eta: 0:00:13 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.1508 (0.3195) acc1: 100.0000 (90.0561) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2958 data: 0.2038 max mem: 3533 +Test: [480/517] eta: 0:00:10 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.1460 (0.3158) acc1: 100.0000 (90.2612) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1658 data: 0.0764 max mem: 3533 +Test: [490/517] eta: 0:00:07 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.1510 (0.3133) acc1: 100.0000 (90.4484) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1904 data: 0.1011 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.2177 (0.3112) acc1: 99.7396 (90.6172) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1638 data: 0.0735 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.2054 (0.3090) acc1: 98.9583 (90.7845) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1621 data: 0.0718 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.1800 (0.3073) acc1: 99.7396 (90.8788) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1494 data: 0.0573 max mem: 3533 +Test: Total time: 0:02:18 (0.2680 s / it) +* Overall Acc@1 90.879 Acc@5 100.000 + target == real Acc@1 18.054 Acc@5 100.000 + target == fake Acc@1 98.434 Acc@5 100.000 + loss 0.307 + PR-AUC 0.9689 ROC-AUC 0.7880 EER 28.57% (thr=0.7944) +Accuracy of the network on the 198427 test images: 90.9% +[rank0]:[W726 05:04:27.526452655 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 05:04:47.861000 1188203 torch/distributed/run.py:766] +W0726 05:04:47.861000 1188203 torch/distributed/run.py:766] ***************************************** +W0726 05:04:47.861000 1188203 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:04:47.861000 1188203 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:05:18.741605127 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:05:19.762649534 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:05:19.809594707 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:05:19.896790827 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:05:19.007382313 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:05:19.068707452 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:05:19.098208669 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank4]:[W726 05:05:19.098313424 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n16_seed7_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed7_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:38:50 acc1_pos: 30.729 acc5_pos: 100.000 loss: 0.8345 (0.8345) acc1: 30.7292 (30.7292) acc5: 100.0000 (100.0000) time: 5.9917 data: 4.4519 max mem: 3531 +Test: [ 10/389] eta: 0:03:57 acc1_pos: 80.469 acc5_pos: 100.000 loss: 0.7232 (0.7687) acc1: 49.7396 (50.9233) acc5: 100.0000 (100.0000) time: 0.6268 data: 0.4051 max mem: 3531 +Test: [ 20/389] eta: 0:02:18 acc1_pos: 12.500 acc5_pos: 100.000 loss: 0.7292 (0.8342) acc1: 41.4062 (42.5471) acc5: 100.0000 (100.0000) time: 0.0936 data: 0.0004 max mem: 3531 +Test: [ 30/389] eta: 0:01:51 acc1_pos: 26.562 acc5_pos: 100.000 loss: 0.8168 (0.8195) acc1: 30.7292 (41.3222) acc5: 100.0000 (100.0000) time: 0.1385 data: 0.0429 max mem: 3531 +Test: [ 40/389] eta: 0:01:44 acc1_pos: 16.406 acc5_pos: 100.000 loss: 0.8471 (0.8261) acc1: 30.2083 (39.4372) acc5: 100.0000 (100.0000) time: 0.2199 data: 0.1272 max mem: 3531 +Test: [ 50/389] eta: 0:01:36 acc1_pos: 66.146 acc5_pos: 100.000 loss: 0.8651 (0.8248) acc1: 26.5625 (38.6080) acc5: 100.0000 (100.0000) time: 0.2414 data: 0.1491 max mem: 3531 +Test: [ 60/389] eta: 0:01:27 acc1_pos: 17.188 acc5_pos: 100.000 loss: 0.7184 (0.8042) acc1: 51.5625 (41.0306) acc5: 100.0000 (100.0000) time: 0.1999 data: 0.1052 max mem: 3531 +Test: [ 70/389] eta: 0:01:21 acc1_pos: 56.510 acc5_pos: 100.000 loss: 0.7189 (0.7998) acc1: 51.5625 (41.4063) acc5: 100.0000 (100.0000) time: 0.1794 data: 0.0844 max mem: 3531 +Test: [ 80/389] eta: 0:01:17 acc1_pos: 56.771 acc5_pos: 100.000 loss: 0.8094 (0.7974) acc1: 39.0625 (41.8274) acc5: 100.0000 (100.0000) time: 0.2050 data: 0.1115 max mem: 3531 +Test: [ 90/389] eta: 0:01:12 acc1_pos: 62.240 acc5_pos: 100.000 loss: 0.7309 (0.7852) acc1: 42.1875 (42.8886) acc5: 100.0000 (100.0000) time: 0.1969 data: 0.1040 max mem: 3531 +Test: [100/389] eta: 0:01:07 acc1_pos: 47.135 acc5_pos: 100.000 loss: 0.7193 (0.7854) acc1: 47.1354 (43.4097) acc5: 100.0000 (100.0000) time: 0.1698 data: 0.0775 max mem: 3531 +Test: [110/389] eta: 0:01:04 acc1_pos: 90.365 acc5_pos: 100.000 loss: 0.7193 (0.7917) acc1: 47.1354 (43.3160) acc5: 100.0000 (100.0000) time: 0.1867 data: 0.0928 max mem: 3531 +Test: [120/389] eta: 0:01:01 acc1_pos: 77.083 acc5_pos: 100.000 loss: 0.7009 (0.7852) acc1: 54.6875 (43.9652) acc5: 100.0000 (100.0000) time: 0.2038 data: 0.1096 max mem: 3531 +Test: [130/389] eta: 0:00:58 acc1_pos: 6.135 acc5_pos: 100.000 loss: 0.5572 (0.7660) acc1: 68.2292 (46.0083) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1799 data: 0.0891 max mem: 3531 +Test: [140/389] eta: 0:00:54 acc1_pos: 6.135 acc5_pos: 100.000 loss: 0.4071 (0.7409) acc1: 90.3646 (48.5871) acc5: 100.0000 (100.0000) acc1_neg: 46.615 acc5_neg: 100.000 time: 0.1575 data: 0.0622 max mem: 3531 +Test: [150/389] eta: 0:00:51 acc1_pos: 6.135 acc5_pos: 100.000 loss: 0.3785 (0.7192) acc1: 91.1458 (51.2986) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1712 data: 0.0738 max mem: 3531 +Test: [160/389] eta: 0:00:49 acc1_pos: 6.135 acc5_pos: 100.000 loss: 0.3785 (0.6958) acc1: 91.1458 (53.6976) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1906 data: 0.0968 max mem: 3531 +Test: [170/389] eta: 0:00:47 acc1_pos: 6.135 acc5_pos: 100.000 loss: 0.2177 (0.6727) acc1: 99.7396 (55.6454) acc5: 100.0000 (100.0000) acc1_neg: 69.271 acc5_neg: 100.000 time: 0.2044 data: 0.1121 max mem: 3531 +Test: [180/389] eta: 0:00:44 acc1_pos: 6.135 acc5_pos: 100.000 loss: 0.3080 (0.6598) acc1: 97.9167 (57.3751) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1941 data: 0.1024 max mem: 3531 +Test: [190/389] eta: 0:00:42 acc1_pos: 6.135 acc5_pos: 100.000 loss: 0.4076 (0.6517) acc1: 88.5417 (58.4615) acc5: 100.0000 (100.0000) acc1_neg: 68.490 acc5_neg: 100.000 time: 0.1822 data: 0.0907 max mem: 3531 +Test: [200/389] eta: 0:00:39 acc1_pos: 6.135 acc5_pos: 100.000 loss: 0.4076 (0.6405) acc1: 89.5833 (59.9775) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1635 data: 0.0726 max mem: 3531 +Test: [210/389] eta: 0:00:37 acc1_pos: 6.135 acc5_pos: 100.000 loss: 0.3878 (0.6241) acc1: 96.3542 (61.7558) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1846 data: 0.0934 max mem: 3532 +Test: [220/389] eta: 0:00:35 acc1_pos: 6.135 acc5_pos: 100.000 loss: 0.2170 (0.6092) acc1: 99.4792 (63.2789) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1921 data: 0.1007 max mem: 3532 +Test: [230/389] eta: 0:00:33 acc1_pos: 6.135 acc5_pos: 100.000 loss: 0.3691 (0.6046) acc1: 88.2812 (63.7559) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1948 data: 0.1030 max mem: 3532 +Test: [240/389] eta: 0:00:30 acc1_pos: 6.135 acc5_pos: 100.000 loss: 0.3935 (0.5947) acc1: 87.7604 (64.9334) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2165 data: 0.1239 max mem: 3532 +Test: [250/389] eta: 0:00:28 acc1_pos: 6.135 acc5_pos: 100.000 loss: 0.3348 (0.5836) acc1: 95.0521 (66.2371) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1980 data: 0.1059 max mem: 3532 +Test: [260/389] eta: 0:00:26 acc1_pos: 6.135 acc5_pos: 100.000 loss: 0.2681 (0.5758) acc1: 99.7396 (67.1256) acc5: 100.0000 (100.0000) acc1_neg: 32.812 acc5_neg: 100.000 time: 0.1820 data: 0.0903 max mem: 3532 +Test: [270/389] eta: 0:00:24 acc1_pos: 6.135 acc5_pos: 100.000 loss: 0.4210 (0.5776) acc1: 94.2708 (67.0703) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1898 data: 0.0983 max mem: 3532 +Test: [280/389] eta: 0:00:22 acc1_pos: 6.135 acc5_pos: 100.000 loss: 0.5153 (0.5761) acc1: 76.5625 (67.2885) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1801 data: 0.0871 max mem: 3532 +Test: [290/389] eta: 0:00:20 acc1_pos: 6.135 acc5_pos: 100.000 loss: 0.4329 (0.5671) acc1: 93.7500 (68.1182) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1651 data: 0.0722 max mem: 3532 +Test: [300/389] eta: 0:00:17 acc1_pos: 6.135 acc5_pos: 100.000 loss: 0.2873 (0.5588) acc1: 97.6562 (69.0476) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1722 data: 0.0817 max mem: 3532 +Test: [310/389] eta: 0:00:15 acc1_pos: 6.135 acc5_pos: 100.000 loss: 0.3157 (0.5553) acc1: 94.7917 (69.5128) acc5: 100.0000 (100.0000) acc1_neg: 20.312 acc5_neg: 100.000 time: 0.1904 data: 0.0998 max mem: 3532 +Test: [320/389] eta: 0:00:13 acc1_pos: 6.135 acc5_pos: 100.000 loss: 0.4226 (0.5523) acc1: 92.1875 (69.8817) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1815 data: 0.0899 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 6.135 acc5_pos: 100.000 loss: 0.4482 (0.5497) acc1: 91.9271 (70.4706) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1602 data: 0.0687 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 6.135 acc5_pos: 100.000 loss: 0.4156 (0.5415) acc1: 92.4479 (71.2457) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1699 data: 0.0786 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 6.135 acc5_pos: 100.000 loss: 0.2556 (0.5450) acc1: 98.9583 (70.7755) acc5: 100.0000 (100.0000) acc1_neg: 33.073 acc5_neg: 100.000 time: 0.1754 data: 0.0824 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 6.135 acc5_pos: 100.000 loss: 0.6106 (0.5480) acc1: 62.7604 (70.4214) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.1680 data: 0.0758 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 6.135 acc5_pos: 100.000 loss: 0.5812 (0.5427) acc1: 68.2292 (70.9302) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1344 data: 0.0442 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 6.135 acc5_pos: 100.000 loss: 0.3488 (0.5419) acc1: 86.7188 (71.1033) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.1193 data: 0.0283 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 6.135 acc5_pos: 100.000 loss: 0.3245 (0.5378) acc1: 91.5301 (71.5422) acc5: 100.0000 (100.0000) acc1_neg: 91.530 acc5_neg: 100.000 time: 0.1229 data: 0.0283 max mem: 3532 +Test: Total time: 0:01:14 (0.1921 s / it) +* Overall Acc@1 71.542 Acc@5 100.000 + target == real Acc@1 43.452 Acc@5 100.000 + target == fake Acc@1 84.599 Acc@5 100.000 + loss 0.538 + PR-AUC 0.8835 ROC-AUC 0.7673 EER 30.38% (thr=0.6172) +Accuracy of the network on the 149358 test images: 71.5% +[rank0]:[W726 05:06:43.757127404 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:07:02.918000 1188763 torch/distributed/run.py:766] +W0726 05:07:02.918000 1188763 torch/distributed/run.py:766] ***************************************** +W0726 05:07:02.918000 1188763 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:07:02.918000 1188763 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +| distributed init (rank 4): env:// +[rank5]:[W726 05:07:32.790972686 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:07:32.795667351 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:07:32.798559111 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:07:32.958264061 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:07:32.024254733 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:07:32.157641027 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:07:32.186472195 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W726 05:07:32.186734725 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n16_seed7_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed7_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:29:58 acc1_pos: 80.469 acc5_pos: 100.000 loss: 0.3929 (0.3929) acc1: 80.4688 (80.4688) acc5: 100.0000 (100.0000) time: 5.4008 data: 3.6646 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/333] eta: 0:03:06 acc1_pos: 84.896 acc5_pos: 100.000 loss: 0.2964 (0.3001) acc1: 85.1562 (86.3636) acc5: 100.0000 (100.0000) time: 0.5779 data: 0.3388 max mem: 3531 +Test: [ 20/333] eta: 0:01:48 acc1_pos: 93.229 acc5_pos: 100.000 loss: 0.2904 (0.3012) acc1: 85.1562 (86.6816) acc5: 100.0000 (100.0000) time: 0.0948 data: 0.0053 max mem: 3531 +Test: [ 30/333] eta: 0:01:23 acc1_pos: 48.177 acc5_pos: 100.000 loss: 0.4067 (0.4990) acc1: 70.8333 (72.6563) acc5: 100.0000 (100.0000) time: 0.1115 data: 0.0209 max mem: 3531 +Test: [ 40/333] eta: 0:01:14 acc1_pos: 31.250 acc5_pos: 100.000 loss: 0.8997 (0.6043) acc1: 42.9688 (64.7548) acc5: 100.0000 (100.0000) time: 0.1551 data: 0.0643 max mem: 3531 +Test: [ 50/333] eta: 0:01:08 acc1_pos: 46.615 acc5_pos: 100.000 loss: 0.8960 (0.6752) acc1: 42.9688 (59.6558) acc5: 100.0000 (100.0000) time: 0.1846 data: 0.0942 max mem: 3531 +Test: [ 60/333] eta: 0:01:01 acc1_pos: 33.594 acc5_pos: 100.000 loss: 0.9109 (0.7197) acc1: 42.4479 (56.3311) acc5: 100.0000 (100.0000) time: 0.1698 data: 0.0790 max mem: 3531 +Test: [ 70/333] eta: 0:00:56 acc1_pos: 41.927 acc5_pos: 100.000 loss: 0.9259 (0.7530) acc1: 41.9271 (54.3537) acc5: 100.0000 (100.0000) time: 0.1490 data: 0.0581 max mem: 3531 +Test: [ 80/333] eta: 0:00:52 acc1_pos: 47.656 acc5_pos: 100.000 loss: 0.9209 (0.7736) acc1: 43.2292 (53.0671) acc5: 100.0000 (100.0000) time: 0.1497 data: 0.0587 max mem: 3531 +Test: [ 90/333] eta: 0:00:49 acc1_pos: 53.906 acc5_pos: 100.000 loss: 0.8499 (0.7770) acc1: 48.1771 (52.7902) acc5: 100.0000 (100.0000) time: 0.1595 data: 0.0676 max mem: 3531 +Test: [100/333] eta: 0:00:46 acc1_pos: 62.760 acc5_pos: 100.000 loss: 0.7991 (0.7783) acc1: 48.1771 (52.4159) acc5: 100.0000 (100.0000) time: 0.1743 data: 0.0802 max mem: 3531 +Test: [110/333] eta: 0:00:44 acc1_pos: 42.188 acc5_pos: 100.000 loss: 0.7936 (0.7808) acc1: 46.0938 (52.1748) acc5: 100.0000 (100.0000) time: 0.1749 data: 0.0779 max mem: 3531 +Test: [120/333] eta: 0:00:41 acc1_pos: 56.510 acc5_pos: 100.000 loss: 0.7597 (0.7795) acc1: 50.2604 (52.2232) acc5: 100.0000 (100.0000) time: 0.1607 data: 0.0646 max mem: 3531 +Test: [130/333] eta: 0:00:38 acc1_pos: 69.792 acc5_pos: 100.000 loss: 0.7064 (0.7776) acc1: 56.5104 (52.4352) acc5: 100.0000 (100.0000) time: 0.1609 data: 0.0687 max mem: 3531 +Test: [140/333] eta: 0:00:37 acc1_pos: 69.271 acc5_pos: 100.000 loss: 0.7461 (0.7774) acc1: 53.9062 (52.3050) acc5: 100.0000 (100.0000) time: 0.1830 data: 0.0889 max mem: 3531 +Test: [150/333] eta: 0:00:35 acc1_pos: 48.958 acc5_pos: 100.000 loss: 0.7500 (0.7793) acc1: 48.9583 (52.0092) acc5: 100.0000 (100.0000) time: 0.1939 data: 0.0977 max mem: 3531 +Test: [160/333] eta: 0:00:33 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.8164 (0.7844) acc1: 48.9583 (51.8731) acc5: 100.0000 (100.0000) acc1_neg: 50.781 acc5_neg: 100.000 time: 0.1815 data: 0.0838 max mem: 3531 +Test: [170/333] eta: 0:00:31 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.7761 (0.7790) acc1: 53.3854 (52.6103) acc5: 100.0000 (100.0000) acc1_neg: 66.667 acc5_neg: 100.000 time: 0.1866 data: 0.0908 max mem: 3531 +Test: [180/333] eta: 0:00:29 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.5920 (0.7648) acc1: 70.3125 (53.8991) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.1989 data: 0.1072 max mem: 3531 +Test: [190/333] eta: 0:00:27 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.5175 (0.7521) acc1: 76.5625 (55.0393) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.1936 data: 0.1007 max mem: 3531 +Test: [200/333] eta: 0:00:25 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.5507 (0.7431) acc1: 73.6979 (55.9235) acc5: 100.0000 (100.0000) acc1_neg: 70.312 acc5_neg: 100.000 time: 0.1955 data: 0.1012 max mem: 3531 +Test: [210/333] eta: 0:00:23 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.6098 (0.7406) acc1: 69.2708 (56.3993) acc5: 100.0000 (100.0000) acc1_neg: 53.646 acc5_neg: 100.000 time: 0.2031 data: 0.1100 max mem: 3532 +Test: [220/333] eta: 0:00:21 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.6732 (0.7430) acc1: 57.8125 (56.3985) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.1848 data: 0.0910 max mem: 3532 +Test: [230/333] eta: 0:00:19 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.3922 (0.7256) acc1: 82.2917 (57.7358) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.1780 data: 0.0849 max mem: 3532 +Test: [240/333] eta: 0:00:17 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.3693 (0.7138) acc1: 84.1146 (58.7234) acc5: 100.0000 (100.0000) acc1_neg: 79.427 acc5_neg: 100.000 time: 0.1938 data: 0.1026 max mem: 3532 +Test: [250/333] eta: 0:00:15 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.4834 (0.7063) acc1: 78.9062 (59.3968) acc5: 100.0000 (100.0000) acc1_neg: 68.490 acc5_neg: 100.000 time: 0.1953 data: 0.1043 max mem: 3532 +Test: [260/333] eta: 0:00:14 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.5445 (0.7025) acc1: 72.6562 (59.8040) acc5: 100.0000 (100.0000) acc1_neg: 72.135 acc5_neg: 100.000 time: 0.1999 data: 0.1095 max mem: 3532 +Test: [270/333] eta: 0:00:12 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.5199 (0.6947) acc1: 74.2188 (60.4599) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.2007 data: 0.1095 max mem: 3532 +Test: [280/333] eta: 0:00:10 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.5057 (0.6954) acc1: 74.2188 (60.5427) acc5: 100.0000 (100.0000) acc1_neg: 62.240 acc5_neg: 100.000 time: 0.1973 data: 0.1052 max mem: 3532 +Test: [290/333] eta: 0:00:08 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.8627 (0.7004) acc1: 51.8229 (60.3182) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.1994 data: 0.1077 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.7193 (0.7001) acc1: 62.7604 (60.4893) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.1987 data: 0.1043 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.6174 (0.6939) acc1: 70.0521 (61.0087) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.2325 data: 0.1361 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.5124 (0.6900) acc1: 75.5208 (61.3991) acc5: 100.0000 (100.0000) acc1_neg: 49.479 acc5_neg: 100.000 time: 0.2036 data: 0.1075 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.8892 (0.6990) acc1: 54.4271 (61.0618) acc5: 100.0000 (100.0000) acc1_neg: 54.427 acc5_neg: 100.000 time: 0.1530 data: 0.0577 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.9308 (0.7005) acc1: 50.2604 (60.9928) acc5: 100.0000 (100.0000) acc1_neg: 47.887 acc5_neg: 100.000 time: 0.1647 data: 0.0670 max mem: 3532 +Test: Total time: 0:01:04 (0.1939 s / it) +* Overall Acc@1 60.993 Acc@5 100.000 + target == real Acc@1 51.910 Acc@5 100.000 + target == fake Acc@1 69.078 Acc@5 100.000 + loss 0.701 + PR-AUC 0.6832 ROC-AUC 0.6599 EER 38.99% (thr=0.5811) +Accuracy of the network on the 127843 test images: 61.0% +[rank0]:[W726 05:08:52.977949625 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:09:12.058000 1189321 torch/distributed/run.py:766] +W0726 05:09:12.058000 1189321 torch/distributed/run.py:766] ***************************************** +W0726 05:09:12.058000 1189321 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:09:12.058000 1189321 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:09:42.260298158 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:09:42.266538494 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 7): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:09:43.710693394 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W726 05:09:43.710765943 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:09:43.773033596 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +| distributed init (rank 2): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:09:43.920694611 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank1]:[W726 05:09:43.920694149 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W726 05:09:43.920707151 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n16_seed7_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed7_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:25:11 acc1_pos: 95.573 acc5_pos: 100.000 loss: 0.1503 (0.1503) acc1: 95.5729 (95.5729) acc5: 100.0000 (100.0000) time: 6.4331 data: 4.8390 max mem: 3531 +Test: [ 10/235] eta: 0:02:29 acc1_pos: 97.135 acc5_pos: 100.000 loss: 0.1552 (0.1545) acc1: 96.3542 (96.4489) acc5: 100.0000 (100.0000) time: 0.6666 data: 0.4403 max mem: 3531 +Test: [ 20/235] eta: 0:01:30 acc1_pos: 96.094 acc5_pos: 100.000 loss: 0.1505 (0.1508) acc1: 96.3542 (96.3294) acc5: 100.0000 (100.0000) time: 0.1183 data: 0.0248 max mem: 3531 +Test: [ 30/235] eta: 0:01:10 acc1_pos: 97.656 acc5_pos: 100.000 loss: 0.1464 (0.1485) acc1: 96.0938 (96.3710) acc5: 100.0000 (100.0000) time: 0.1683 data: 0.0725 max mem: 3531 +Test: [ 40/235] eta: 0:01:00 acc1_pos: 95.573 acc5_pos: 100.000 loss: 0.1464 (0.1485) acc1: 96.3542 (96.3161) acc5: 100.0000 (100.0000) time: 0.1999 data: 0.1043 max mem: 3531 +Test: [ 50/235] eta: 0:00:53 acc1_pos: 96.094 acc5_pos: 100.000 loss: 0.1460 (0.1484) acc1: 96.3542 (96.2929) acc5: 100.0000 (100.0000) time: 0.2065 data: 0.1122 max mem: 3531 +Test: [ 60/235] eta: 0:00:47 acc1_pos: 97.917 acc5_pos: 100.000 loss: 0.1501 (0.1491) acc1: 96.0938 (96.3157) acc5: 100.0000 (100.0000) time: 0.1849 data: 0.0895 max mem: 3531 +Test: [ 70/235] eta: 0:00:42 acc1_pos: 96.923 acc5_pos: 100.000 loss: 0.1651 (0.3968) acc1: 94.5312 (85.7358) acc5: 100.0000 (100.0000) acc1_neg: 9.375 acc5_neg: 100.000 time: 0.1790 data: 0.0839 max mem: 3531 +Test: [ 80/235] eta: 0:00:39 acc1_pos: 96.923 acc5_pos: 100.000 loss: 2.1505 (0.6154) acc1: 10.6771 (76.3632) acc5: 100.0000 (100.0000) acc1_neg: 10.938 acc5_neg: 100.000 time: 0.2004 data: 0.1078 max mem: 3531 +Test: [ 90/235] eta: 0:00:36 acc1_pos: 96.923 acc5_pos: 100.000 loss: 2.1580 (0.7858) acc1: 9.8958 (69.1106) acc5: 100.0000 (100.0000) acc1_neg: 8.854 acc5_neg: 100.000 time: 0.2196 data: 0.1263 max mem: 3531 +Test: [100/235] eta: 0:00:33 acc1_pos: 96.923 acc5_pos: 100.000 loss: 2.1566 (0.9201) acc1: 10.1562 (63.3663) acc5: 100.0000 (100.0000) acc1_neg: 10.156 acc5_neg: 100.000 time: 0.2243 data: 0.1310 max mem: 3531 +Test: [110/235] eta: 0:00:30 acc1_pos: 96.923 acc5_pos: 100.000 loss: 2.1193 (1.0284) acc1: 10.9375 (58.6665) acc5: 100.0000 (100.0000) acc1_neg: 9.375 acc5_neg: 100.000 time: 0.2119 data: 0.1160 max mem: 3531 +Test: [120/235] eta: 0:00:27 acc1_pos: 96.923 acc5_pos: 100.000 loss: 2.0951 (1.1124) acc1: 10.4167 (54.5777) acc5: 100.0000 (100.0000) acc1_neg: 10.156 acc5_neg: 100.000 time: 0.1900 data: 0.0949 max mem: 3531 +Test: [130/235] eta: 0:00:24 acc1_pos: 96.923 acc5_pos: 100.000 loss: 2.0176 (1.1816) acc1: 9.8958 (51.2166) acc5: 100.0000 (100.0000) acc1_neg: 11.979 acc5_neg: 100.000 time: 0.1674 data: 0.0720 max mem: 3531 +Test: [140/235] eta: 0:00:22 acc1_pos: 96.923 acc5_pos: 100.000 loss: 2.0311 (1.2439) acc1: 9.6354 (48.2122) acc5: 100.0000 (100.0000) acc1_neg: 8.073 acc5_neg: 100.000 time: 0.2058 data: 0.1087 max mem: 3531 +Test: [150/235] eta: 0:00:19 acc1_pos: 96.923 acc5_pos: 100.000 loss: 2.0343 (1.2953) acc1: 8.0729 (45.5729) acc5: 100.0000 (100.0000) acc1_neg: 9.635 acc5_neg: 100.000 time: 0.2200 data: 0.1242 max mem: 3531 +Test: [160/235] eta: 0:00:16 acc1_pos: 96.923 acc5_pos: 100.000 loss: 2.0109 (1.3401) acc1: 8.3333 (43.3214) acc5: 100.0000 (100.0000) acc1_neg: 9.635 acc5_neg: 100.000 time: 0.1658 data: 0.0708 max mem: 3531 +Test: [170/235] eta: 0:00:14 acc1_pos: 96.923 acc5_pos: 100.000 loss: 2.0094 (1.3801) acc1: 10.4167 (41.5311) acc5: 100.0000 (100.0000) acc1_neg: 10.938 acc5_neg: 100.000 time: 0.1767 data: 0.0834 max mem: 3531 +Test: [180/235] eta: 0:00:12 acc1_pos: 96.923 acc5_pos: 100.000 loss: 2.0197 (1.4157) acc1: 11.9792 (39.9042) acc5: 100.0000 (100.0000) acc1_neg: 10.677 acc5_neg: 100.000 time: 0.2141 data: 0.1224 max mem: 3531 +Test: [190/235] eta: 0:00:09 acc1_pos: 96.923 acc5_pos: 100.000 loss: 2.0161 (1.4466) acc1: 11.9792 (38.4871) acc5: 100.0000 (100.0000) acc1_neg: 11.198 acc5_neg: 100.000 time: 0.1991 data: 0.1074 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 96.923 acc5_pos: 100.000 loss: 2.0162 (1.4759) acc1: 11.9792 (37.1618) acc5: 100.0000 (100.0000) acc1_neg: 9.896 acc5_neg: 100.000 time: 0.1888 data: 0.0968 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 96.923 acc5_pos: 100.000 loss: 2.0164 (1.5011) acc1: 11.9792 (35.9955) acc5: 100.0000 (100.0000) acc1_neg: 12.240 acc5_neg: 100.000 time: 0.2108 data: 0.1198 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 96.923 acc5_pos: 100.000 loss: 2.0067 (1.5242) acc1: 11.9792 (34.9253) acc5: 100.0000 (100.0000) acc1_neg: 11.198 acc5_neg: 100.000 time: 0.1874 data: 0.0974 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 96.923 acc5_pos: 100.000 loss: 2.0150 (1.5454) acc1: 11.9792 (33.9748) acc5: 100.0000 (100.0000) acc1_neg: 10.677 acc5_neg: 100.000 time: 0.1343 data: 0.0446 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 96.923 acc5_pos: 100.000 loss: 2.0312 (1.5555) acc1: 11.1979 (33.6842) acc5: 100.0000 (100.0000) acc1_neg: 4.348 acc5_neg: 100.000 time: 0.1266 data: 0.0393 max mem: 3532 +Test: Total time: 0:00:49 (0.2119 s / it) +* Overall Acc@1 33.684 Acc@5 100.000 + target == real Acc@1 96.326 Acc@5 100.000 + target == fake Acc@1 11.028 Acc@5 100.000 + loss 1.556 + PR-AUC 0.8277 ROC-AUC 0.6518 EER 39.19% (thr=0.0869) +Accuracy of the network on the 89879 test images: 33.7% +[rank0]:[W726 05:10:43.951694186 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:11:04.058000 1189882 torch/distributed/run.py:766] +W0726 05:11:04.058000 1189882 torch/distributed/run.py:766] ***************************************** +W0726 05:11:04.058000 1189882 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:11:04.058000 1189882 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:11:31.323670994 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:11:32.792367023 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:11:32.919592059 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:11:32.187794563 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:11:32.207225742 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:11:32.230635082 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:11:32.231178107 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:11:32.241601024 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n16_seed7_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed7_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:26:34 acc1_pos: 51.042 acc5_pos: 100.000 loss: 0.7648 (0.7648) acc1: 51.0417 (51.0417) acc5: 100.0000 (100.0000) time: 5.9266 data: 4.4000 max mem: 3531 +Test: [ 10/269] eta: 0:02:40 acc1_pos: 52.083 acc5_pos: 100.000 loss: 0.7169 (0.7177) acc1: 57.0312 (56.6525) acc5: 100.0000 (100.0000) time: 0.6208 data: 0.4004 max mem: 3531 +Test: [ 20/269] eta: 0:01:41 acc1_pos: 54.427 acc5_pos: 100.000 loss: 0.7169 (0.7032) acc1: 54.6875 (57.6761) acc5: 100.0000 (100.0000) time: 0.1299 data: 0.0351 max mem: 3531 +Test: [ 30/269] eta: 0:01:24 acc1_pos: 45.312 acc5_pos: 100.000 loss: 0.6845 (0.6853) acc1: 54.6875 (59.4002) acc5: 100.0000 (100.0000) time: 0.2097 data: 0.1140 max mem: 3531 +Test: [ 40/269] eta: 0:01:14 acc1_pos: 62.069 acc5_pos: 100.000 loss: 0.4459 (0.6230) acc1: 73.4375 (65.3265) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.2399 data: 0.1482 max mem: 3531 +Test: [ 50/269] eta: 0:01:04 acc1_pos: 62.069 acc5_pos: 100.000 loss: 0.4195 (0.5897) acc1: 84.1146 (68.1475) acc5: 100.0000 (100.0000) acc1_neg: 65.365 acc5_neg: 100.000 time: 0.2049 data: 0.1086 max mem: 3531 +Test: [ 60/269] eta: 0:00:58 acc1_pos: 62.069 acc5_pos: 100.000 loss: 0.4710 (0.5788) acc1: 78.6458 (69.1769) acc5: 100.0000 (100.0000) acc1_neg: 70.312 acc5_neg: 100.000 time: 0.1849 data: 0.0879 max mem: 3531 +Test: [ 70/269] eta: 0:00:53 acc1_pos: 62.069 acc5_pos: 100.000 loss: 0.4599 (0.5562) acc1: 80.2083 (71.3065) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.1999 data: 0.1066 max mem: 3531 +Test: [ 80/269] eta: 0:00:48 acc1_pos: 62.069 acc5_pos: 100.000 loss: 0.4111 (0.5410) acc1: 83.5938 (72.7141) acc5: 100.0000 (100.0000) acc1_neg: 74.219 acc5_neg: 100.000 time: 0.1949 data: 0.1020 max mem: 3531 +Test: [ 90/269] eta: 0:00:45 acc1_pos: 62.069 acc5_pos: 100.000 loss: 0.5265 (0.5457) acc1: 74.2188 (72.4388) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.1899 data: 0.0945 max mem: 3531 +Test: [100/269] eta: 0:00:41 acc1_pos: 62.069 acc5_pos: 100.000 loss: 0.5005 (0.5346) acc1: 77.6042 (73.3937) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.1999 data: 0.1034 max mem: 3531 +Test: [110/269] eta: 0:00:38 acc1_pos: 62.069 acc5_pos: 100.000 loss: 0.4509 (0.5274) acc1: 80.4688 (74.0897) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.1949 data: 0.1027 max mem: 3531 +Test: [120/269] eta: 0:00:35 acc1_pos: 62.069 acc5_pos: 100.000 loss: 0.4852 (0.5274) acc1: 77.3438 (74.0939) acc5: 100.0000 (100.0000) acc1_neg: 68.229 acc5_neg: 100.000 time: 0.1899 data: 0.0978 max mem: 3531 +Test: [130/269] eta: 0:00:32 acc1_pos: 62.069 acc5_pos: 100.000 loss: 0.5346 (0.5292) acc1: 74.7396 (73.9146) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.1857 data: 0.0915 max mem: 3531 +Test: [140/269] eta: 0:00:29 acc1_pos: 62.069 acc5_pos: 100.000 loss: 0.5780 (0.5361) acc1: 67.4479 (73.2491) acc5: 100.0000 (100.0000) acc1_neg: 53.125 acc5_neg: 100.000 time: 0.1901 data: 0.0979 max mem: 3531 +Test: [150/269] eta: 0:00:27 acc1_pos: 62.069 acc5_pos: 100.000 loss: 0.6177 (0.5392) acc1: 66.6667 (72.9874) acc5: 100.0000 (100.0000) acc1_neg: 69.271 acc5_neg: 100.000 time: 0.2091 data: 0.1149 max mem: 3531 +Test: [160/269] eta: 0:00:24 acc1_pos: 62.069 acc5_pos: 100.000 loss: 0.4926 (0.5385) acc1: 71.0938 (73.0784) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.2097 data: 0.1127 max mem: 3531 +Test: [170/269] eta: 0:00:22 acc1_pos: 62.069 acc5_pos: 100.000 loss: 0.4857 (0.5353) acc1: 77.8646 (73.2593) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.1699 data: 0.0767 max mem: 3531 +Test: [180/269] eta: 0:00:19 acc1_pos: 62.069 acc5_pos: 100.000 loss: 0.3854 (0.5259) acc1: 83.8542 (74.0346) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.1676 data: 0.0768 max mem: 3531 +Test: [190/269] eta: 0:00:17 acc1_pos: 62.069 acc5_pos: 100.000 loss: 0.3661 (0.5199) acc1: 86.7188 (74.5378) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.2006 data: 0.1058 max mem: 3531 +Test: [200/269] eta: 0:00:15 acc1_pos: 62.069 acc5_pos: 100.000 loss: 0.4355 (0.5180) acc1: 81.5104 (74.7357) acc5: 100.0000 (100.0000) acc1_neg: 75.781 acc5_neg: 100.000 time: 0.2034 data: 0.1084 max mem: 3531 +Test: [210/269] eta: 0:00:12 acc1_pos: 62.069 acc5_pos: 100.000 loss: 0.5018 (0.5202) acc1: 75.7812 (74.5742) acc5: 100.0000 (100.0000) acc1_neg: 68.490 acc5_neg: 100.000 time: 0.1942 data: 0.0983 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 62.069 acc5_pos: 100.000 loss: 0.5018 (0.5182) acc1: 74.7396 (74.6807) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1887 data: 0.0924 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 62.069 acc5_pos: 100.000 loss: 0.3905 (0.5115) acc1: 82.5521 (75.1984) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1857 data: 0.0922 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 62.069 acc5_pos: 100.000 loss: 0.3905 (0.5114) acc1: 82.5521 (75.2291) acc5: 100.0000 (100.0000) acc1_neg: 59.115 acc5_neg: 100.000 time: 0.1802 data: 0.0871 max mem: 3532 +Test: [250/269] eta: 0:00:04 acc1_pos: 62.069 acc5_pos: 100.000 loss: 0.5722 (0.5148) acc1: 70.0521 (74.9668) acc5: 100.0000 (100.0000) acc1_neg: 41.146 acc5_neg: 100.000 time: 0.1744 data: 0.0816 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 62.069 acc5_pos: 100.000 loss: 0.5767 (0.5198) acc1: 67.9688 (74.5819) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.1481 data: 0.0556 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 62.069 acc5_pos: 100.000 loss: 0.4341 (0.5154) acc1: 77.8646 (74.9354) acc5: 100.0000 (100.0000) acc1_neg: 84.252 acc5_neg: 100.000 time: 0.1339 data: 0.0318 max mem: 3532 +Test: Total time: 0:00:55 (0.2075 s / it) +* Overall Acc@1 74.935 Acc@5 100.000 + target == real Acc@1 59.183 Acc@5 100.000 + target == fake Acc@1 77.085 Acc@5 100.000 + loss 0.515 + PR-AUC 0.9548 ROC-AUC 0.7584 EER 30.94% (thr=0.5728) +Accuracy of the network on the 103293 test images: 74.9% +[rank0]:[W726 05:12:39.409394348 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T05:12:46Z] eval_done tag=fixed500k_fake_n16_seed7_resampled +[2026-07-26T05:12:46Z] eval_start tag=fixed500k_real_n8_seed8_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed8_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T05:12:46Z +hostname=n116-041-153 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed8_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n8_seed8_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 05:13:00.102000 1190445 torch/distributed/run.py:766] +W0726 05:13:00.102000 1190445 torch/distributed/run.py:766] ***************************************** +W0726 05:13:00.102000 1190445 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:13:00.102000 1190445 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:13:29.419450287 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:13:30.753711618 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:13:30.999972705 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:13:30.126420229 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:13:30.254788067 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:13:30.289563279 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:13:30.369779638 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:13:30.392418146 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n8_seed8_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed8_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:20:32 acc1_pos: 99.479 acc5_pos: 100.000 loss: 0.2590 (0.2590) acc1: 99.4792 (99.4792) acc5: 100.0000 (100.0000) time: 5.2659 data: 2.7977 max mem: 3531 +Test: [ 10/234] eta: 0:02:05 acc1_pos: 94.271 acc5_pos: 100.000 loss: 0.3469 (0.3862) acc1: 94.2708 (89.7964) acc5: 100.0000 (100.0000) time: 0.5624 data: 0.2567 max mem: 3531 +Test: [ 20/234] eta: 0:01:16 acc1_pos: 17.188 acc5_pos: 100.000 loss: 0.3880 (0.5865) acc1: 84.8958 (77.1701) acc5: 100.0000 (100.0000) time: 0.1136 data: 0.0237 max mem: 3531 +Test: [ 30/234] eta: 0:01:03 acc1_pos: 21.094 acc5_pos: 100.000 loss: 1.2432 (0.8798) acc1: 21.0938 (59.5850) acc5: 100.0000 (100.0000) time: 0.1750 data: 0.0843 max mem: 3531 +Test: [ 40/234] eta: 0:00:56 acc1_pos: 7.812 acc5_pos: 100.000 loss: 1.3790 (1.0141) acc1: 14.0625 (48.4439) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1331 max mem: 3531 +Test: [ 50/234] eta: 0:00:50 acc1_pos: 8.594 acc5_pos: 100.000 loss: 1.5384 (1.1430) acc1: 10.4167 (40.9926) acc5: 100.0000 (100.0000) time: 0.2224 data: 0.1311 max mem: 3531 +Test: [ 60/234] eta: 0:00:46 acc1_pos: 11.724 acc5_pos: 100.000 loss: 0.6583 (0.9763) acc1: 19.5312 (50.1067) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2066 data: 0.1165 max mem: 3531 +Test: [ 70/234] eta: 0:00:41 acc1_pos: 11.724 acc5_pos: 100.000 loss: 0.0697 (0.8650) acc1: 99.7396 (56.4187) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.2003 data: 0.1105 max mem: 3531 +Test: [ 80/234] eta: 0:00:38 acc1_pos: 11.724 acc5_pos: 100.000 loss: 0.2250 (0.7860) acc1: 93.7500 (61.0822) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1938 data: 0.1025 max mem: 3531 +Test: [ 90/234] eta: 0:00:34 acc1_pos: 11.724 acc5_pos: 100.000 loss: 0.1764 (0.7071) acc1: 97.1354 (65.2959) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1945 data: 0.1026 max mem: 3531 +Test: [100/234] eta: 0:00:31 acc1_pos: 11.724 acc5_pos: 100.000 loss: 0.0937 (0.6586) acc1: 98.1771 (67.9791) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1993 data: 0.1054 max mem: 3531 +Test: [110/234] eta: 0:00:29 acc1_pos: 11.724 acc5_pos: 100.000 loss: 0.1301 (0.6196) acc1: 97.9167 (70.1037) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1971 data: 0.1028 max mem: 3531 +Test: [120/234] eta: 0:00:26 acc1_pos: 11.724 acc5_pos: 100.000 loss: 0.1778 (0.5965) acc1: 97.1354 (71.3649) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.1006 max mem: 3531 +Test: [130/234] eta: 0:00:23 acc1_pos: 11.724 acc5_pos: 100.000 loss: 0.4329 (0.6021) acc1: 73.4375 (70.2549) acc5: 100.0000 (100.0000) acc1_neg: 68.490 acc5_neg: 100.000 time: 0.2027 data: 0.1032 max mem: 3531 +Test: [140/234] eta: 0:00:21 acc1_pos: 11.724 acc5_pos: 100.000 loss: 0.4066 (0.5839) acc1: 71.3542 (71.3412) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2406 data: 0.1419 max mem: 3531 +Test: [150/234] eta: 0:00:19 acc1_pos: 11.724 acc5_pos: 100.000 loss: 0.1975 (0.5570) acc1: 95.3125 (72.9598) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.2269 data: 0.1312 max mem: 3531 +Test: [160/234] eta: 0:00:16 acc1_pos: 11.724 acc5_pos: 100.000 loss: 0.3533 (0.5612) acc1: 82.0312 (72.3748) acc5: 100.0000 (100.0000) acc1_neg: 65.885 acc5_neg: 100.000 time: 0.1542 data: 0.0589 max mem: 3531 +Test: [170/234] eta: 0:00:14 acc1_pos: 11.724 acc5_pos: 100.000 loss: 0.6259 (0.5690) acc1: 61.4583 (71.5887) acc5: 100.0000 (100.0000) acc1_neg: 51.042 acc5_neg: 100.000 time: 0.1489 data: 0.0547 max mem: 3531 +Test: [180/234] eta: 0:00:11 acc1_pos: 11.724 acc5_pos: 100.000 loss: 0.5902 (0.5547) acc1: 66.4062 (72.4419) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1799 data: 0.0875 max mem: 3531 +Test: [190/234] eta: 0:00:09 acc1_pos: 11.724 acc5_pos: 100.000 loss: 0.4800 (0.5514) acc1: 75.0000 (72.5567) acc5: 100.0000 (100.0000) acc1_neg: 71.094 acc5_neg: 100.000 time: 0.1889 data: 0.0961 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 11.724 acc5_pos: 100.000 loss: 0.4805 (0.5372) acc1: 73.9583 (73.4129) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1990 data: 0.1064 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 11.724 acc5_pos: 100.000 loss: 0.1171 (0.5165) acc1: 99.4792 (74.6532) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1746 data: 0.0848 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 11.724 acc5_pos: 100.000 loss: 0.0732 (0.4969) acc1: 99.7396 (75.7636) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1164 data: 0.0260 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 11.724 acc5_pos: 100.000 loss: 0.1970 (0.4940) acc1: 86.9792 (76.0890) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1011 data: 0.0105 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 11.724 acc5_pos: 100.000 loss: 0.4190 (0.4933) acc1: 83.8542 (76.1649) acc5: 100.0000 (100.0000) acc1_neg: 80.545 acc5_neg: 100.000 time: 0.1026 data: 0.0104 max mem: 3532 +Test: Total time: 0:00:47 (0.2022 s / it) +* Overall Acc@1 76.165 Acc@5 100.000 + target == real Acc@1 40.778 Acc@5 100.000 + target == fake Acc@1 86.139 Acc@5 100.000 + loss 0.493 + PR-AUC 0.9051 ROC-AUC 0.7384 EER 33.77% (thr=0.7739) +Accuracy of the network on the 89729 test images: 76.2% +[rank0]:[W726 05:14:29.819282690 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:14:48.378000 1191005 torch/distributed/run.py:766] +W0726 05:14:48.378000 1191005 torch/distributed/run.py:766] ***************************************** +W0726 05:14:48.378000 1191005 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:14:48.378000 1191005 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:15:16.453528158 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:15:17.956682256 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:15:17.103462462 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:15:17.237514439 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:15:17.394460462 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +[rank0]:[W726 05:15:17.402473765 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:15:17.415416730 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank1]:[W726 05:15:17.415470354 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n8_seed8_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed8_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 5:17:58 acc1_pos: 75.000 acc5_pos: 100.000 loss: 0.5759 (0.5759) acc1: 75.0000 (75.0000) acc5: 100.0000 (100.0000) time: 7.9926 data: 5.9546 max mem: 3531 +Test: [ 10/2387] eta: 0:32:02 acc1_pos: 42.448 acc5_pos: 100.000 loss: 0.6564 (0.6792) acc1: 65.8854 (63.0208) acc5: 100.0000 (100.0000) time: 0.8089 data: 0.5417 max mem: 3531 +Test: [ 20/2387] eta: 0:18:33 acc1_pos: 67.708 acc5_pos: 100.000 loss: 0.6398 (0.6877) acc1: 65.6250 (62.7356) acc5: 100.0000 (100.0000) time: 0.0943 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:15:45 acc1_pos: 40.365 acc5_pos: 100.000 loss: 0.6398 (0.6939) acc1: 56.2500 (61.3995) acc5: 100.0000 (100.0000) time: 0.1766 data: 0.0825 max mem: 3531 +Test: [ 40/2387] eta: 0:14:12 acc1_pos: 74.219 acc5_pos: 100.000 loss: 0.7700 (0.7100) acc1: 50.0000 (59.5147) acc5: 100.0000 (100.0000) time: 0.2510 data: 0.1586 max mem: 3531 +Test: [ 50/2387] eta: 0:13:07 acc1_pos: 48.698 acc5_pos: 100.000 loss: 0.6838 (0.6942) acc1: 58.3333 (61.3256) acc5: 100.0000 (100.0000) time: 0.2374 data: 0.1440 max mem: 3531 +Test: [ 60/2387] eta: 0:12:30 acc1_pos: 48.698 acc5_pos: 100.000 loss: 0.7114 (0.7065) acc1: 56.7708 (60.7411) acc5: 100.0000 (100.0000) time: 0.2388 data: 0.1454 max mem: 3531 +Test: [ 70/2387] eta: 0:12:06 acc1_pos: 32.031 acc5_pos: 100.000 loss: 0.7764 (0.7233) acc1: 52.6042 (59.9178) acc5: 100.0000 (100.0000) time: 0.2549 data: 0.1551 max mem: 3531 +Test: [ 80/2387] eta: 0:11:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6412 (0.6774) acc1: 69.2708 (63.5995) acc5: 100.0000 (100.0000) time: 0.2416 data: 0.1418 max mem: 3531 +Test: [ 90/2387] eta: 0:11:31 acc1_pos: 96.354 acc5_pos: 100.000 loss: 0.2449 (0.6288) acc1: 98.6979 (67.3592) acc5: 100.0000 (100.0000) time: 0.2551 data: 0.1591 max mem: 3531 +Test: [ 100/2387] eta: 0:11:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2336 (0.5915) acc1: 98.6979 (70.1965) acc5: 100.0000 (100.0000) time: 0.2684 data: 0.1737 max mem: 3531 +Test: [ 110/2387] eta: 0:10:58 acc1_pos: 98.177 acc5_pos: 100.000 loss: 0.2402 (0.5604) acc1: 98.1771 (72.6656) acc5: 100.0000 (100.0000) time: 0.2358 data: 0.1447 max mem: 3531 +Test: [ 120/2387] eta: 0:10:49 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2435 (0.5350) acc1: 98.1771 (74.7267) acc5: 100.0000 (100.0000) time: 0.2396 data: 0.1447 max mem: 3531 +Test: [ 130/2387] eta: 0:10:42 acc1_pos: 98.958 acc5_pos: 100.000 loss: 0.2369 (0.5118) acc1: 99.2188 (76.4969) acc5: 100.0000 (100.0000) time: 0.2587 data: 0.1598 max mem: 3531 +Test: [ 140/2387] eta: 0:10:29 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2369 (0.4925) acc1: 98.9583 (77.9846) acc5: 100.0000 (100.0000) time: 0.2414 data: 0.1422 max mem: 3531 +Test: [ 150/2387] eta: 0:10:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2209 (0.4746) acc1: 99.7396 (79.3736) acc5: 100.0000 (100.0000) time: 0.2311 data: 0.1352 max mem: 3531 +Test: [ 160/2387] eta: 0:10:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2165 (0.4612) acc1: 100.0000 (80.4380) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.2541 data: 0.1629 max mem: 3531 +Test: [ 170/2387] eta: 0:10:11 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3313 (0.4547) acc1: 90.1042 (80.8647) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.2638 data: 0.1717 max mem: 3531 +Test: [ 180/2387] eta: 0:09:53 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3573 (0.4495) acc1: 87.2396 (81.2011) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.2043 data: 0.1090 max mem: 3531 +Test: [ 190/2387] eta: 0:09:45 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3525 (0.4444) acc1: 87.2396 (81.5513) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1867 data: 0.0927 max mem: 3531 +Test: [ 200/2387] eta: 0:09:40 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3503 (0.4397) acc1: 87.7604 (81.8447) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.2349 data: 0.1412 max mem: 3531 +Test: [ 210/2387] eta: 0:09:29 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3527 (0.4358) acc1: 87.5000 (82.1115) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.2131 data: 0.1153 max mem: 3532 +Test: [ 220/2387] eta: 0:09:18 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3541 (0.4321) acc1: 87.2396 (82.3577) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1799 data: 0.0838 max mem: 3532 +Test: [ 230/2387] eta: 0:09:12 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3541 (0.4285) acc1: 87.5000 (82.5870) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.2006 data: 0.1065 max mem: 3532 +Test: [ 240/2387] eta: 0:09:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3669 (0.4465) acc1: 82.0312 (80.8588) acc5: 100.0000 (100.0000) acc1_neg: 29.688 acc5_neg: 100.000 time: 0.2121 data: 0.1187 max mem: 3532 +Test: [ 250/2387] eta: 0:08:56 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3808 (0.4421) acc1: 50.2604 (81.0560) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1892 data: 0.0972 max mem: 3532 +Test: [ 260/2387] eta: 0:08:46 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2436 (0.4342) acc1: 93.4896 (81.5453) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1677 data: 0.0736 max mem: 3532 +Test: [ 270/2387] eta: 0:08:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2255 (0.4262) acc1: 94.0104 (82.0351) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1613 data: 0.0675 max mem: 3532 +Test: [ 280/2387] eta: 0:08:29 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2255 (0.4198) acc1: 94.0104 (82.3973) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1699 data: 0.0771 max mem: 3532 +Test: [ 290/2387] eta: 0:08:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2581 (0.4143) acc1: 92.1875 (82.7329) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1788 data: 0.0861 max mem: 3532 +Test: [ 300/2387] eta: 0:08:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2763 (0.4125) acc1: 90.8854 (82.7606) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.1849 data: 0.0911 max mem: 3532 +Test: [ 310/2387] eta: 0:08:10 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3529 (0.4087) acc1: 87.7604 (82.9649) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1854 data: 0.0913 max mem: 3532 +Test: [ 320/2387] eta: 0:08:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1480 (0.4005) acc1: 99.7396 (83.4932) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1811 data: 0.0881 max mem: 3532 +Test: [ 330/2387] eta: 0:08:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1463 (0.3928) acc1: 100.0000 (83.9911) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1941 data: 0.1011 max mem: 3532 +Test: [ 340/2387] eta: 0:07:56 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1450 (0.3855) acc1: 100.0000 (84.4598) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2089 data: 0.1152 max mem: 3532 +Test: [ 350/2387] eta: 0:07:54 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1470 (0.3827) acc1: 99.4792 (84.8090) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2214 data: 0.1282 max mem: 3532 +Test: [ 360/2387] eta: 0:07:49 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3102 (0.3809) acc1: 94.2708 (84.9636) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.2098 data: 0.1174 max mem: 3532 +Test: [ 370/2387] eta: 0:07:42 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3204 (0.3794) acc1: 89.3229 (85.0713) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1634 data: 0.0680 max mem: 3532 +Test: [ 380/2387] eta: 0:07:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3212 (0.3779) acc1: 89.3229 (85.1986) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1471 data: 0.0527 max mem: 3532 +Test: [ 390/2387] eta: 0:07:32 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3244 (0.3767) acc1: 89.5833 (85.3008) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1750 data: 0.0838 max mem: 3532 +Test: [ 400/2387] eta: 0:07:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3297 (0.3754) acc1: 89.0625 (85.4037) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.1793 data: 0.0856 max mem: 3532 +Test: [ 410/2387] eta: 0:07:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3214 (0.3740) acc1: 90.3646 (85.5314) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1753 data: 0.0811 max mem: 3532 +Test: [ 420/2387] eta: 0:07:17 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3150 (0.3728) acc1: 89.8438 (85.6264) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1682 data: 0.0732 max mem: 3532 +Test: [ 430/2387] eta: 0:07:11 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3312 (0.3787) acc1: 88.5417 (85.0819) acc5: 100.0000 (100.0000) acc1_neg: 42.448 acc5_neg: 100.000 time: 0.1543 data: 0.0566 max mem: 3532 +Test: [ 440/2387] eta: 0:07:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3507 (0.3798) acc1: 50.2604 (84.8893) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1603 data: 0.0631 max mem: 3533 +Test: [ 450/2387] eta: 0:07:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1719 (0.3751) acc1: 98.6979 (85.2048) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1599 data: 0.0637 max mem: 3533 +Test: [ 460/2387] eta: 0:06:57 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1620 (0.3705) acc1: 98.9583 (85.5037) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1611 data: 0.0649 max mem: 3533 +Test: [ 470/2387] eta: 0:06:52 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1629 (0.3662) acc1: 98.9583 (85.7949) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1552 data: 0.0609 max mem: 3533 +Test: [ 480/2387] eta: 0:06:48 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1633 (0.3619) acc1: 99.2188 (86.0723) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1533 data: 0.0614 max mem: 3533 +Test: [ 490/2387] eta: 0:06:45 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1652 (0.3580) acc1: 98.9583 (86.3332) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1797 data: 0.0876 max mem: 3533 +Test: [ 500/2387] eta: 0:06:41 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1665 (0.3543) acc1: 98.9583 (86.5841) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1820 data: 0.0863 max mem: 3533 +Test: [ 510/2387] eta: 0:06:39 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1687 (0.3531) acc1: 98.9583 (86.6168) acc5: 100.0000 (100.0000) acc1_neg: 62.240 acc5_neg: 100.000 time: 0.1819 data: 0.0854 max mem: 3533 +Test: [ 520/2387] eta: 0:06:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3388 (0.3565) acc1: 66.4062 (86.3199) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2252 data: 0.1330 max mem: 3533 +Test: [ 530/2387] eta: 0:06:35 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3388 (0.3560) acc1: 92.1875 (86.4554) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2259 data: 0.1338 max mem: 3533 +Test: [ 540/2387] eta: 0:06:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3309 (0.3556) acc1: 93.2292 (86.5806) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1627 data: 0.0695 max mem: 3533 +Test: [ 550/2387] eta: 0:06:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3281 (0.3551) acc1: 93.2292 (86.6961) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1374 data: 0.0442 max mem: 3533 +Test: [ 560/2387] eta: 0:06:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3224 (0.3546) acc1: 93.2292 (86.8199) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2013 data: 0.1075 max mem: 3533 +Test: [ 570/2387] eta: 0:06:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3253 (0.3541) acc1: 93.2292 (86.9304) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2198 data: 0.1271 max mem: 3533 +Test: [ 580/2387] eta: 0:06:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3260 (0.3537) acc1: 93.2292 (87.0370) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1890 data: 0.0977 max mem: 3533 +Test: [ 590/2387] eta: 0:06:18 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3364 (0.3534) acc1: 92.4479 (87.1193) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.2011 data: 0.1088 max mem: 3533 +Test: [ 600/2387] eta: 0:06:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3509 (0.3577) acc1: 70.3125 (86.7045) acc5: 100.0000 (100.0000) acc1_neg: 70.312 acc5_neg: 100.000 time: 0.2151 data: 0.1232 max mem: 3533 +Test: [ 610/2387] eta: 0:06:14 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2987 (0.3564) acc1: 70.3125 (86.8360) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2144 data: 0.1236 max mem: 3533 +Test: [ 620/2387] eta: 0:06:11 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2827 (0.3553) acc1: 94.0104 (86.9536) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1952 data: 0.1040 max mem: 3533 +Test: [ 630/2387] eta: 0:06:10 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2827 (0.3541) acc1: 94.5312 (87.0782) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2150 data: 0.1236 max mem: 3533 +Test: [ 640/2387] eta: 0:06:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2781 (0.3530) acc1: 94.7917 (87.1864) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2236 data: 0.1295 max mem: 3533 +Test: [ 650/2387] eta: 0:06:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2781 (0.3519) acc1: 94.5312 (87.3036) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2350 data: 0.1410 max mem: 3533 +Test: [ 660/2387] eta: 0:06:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2830 (0.3509) acc1: 94.5312 (87.4169) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2301 data: 0.1376 max mem: 3533 +Test: [ 670/2387] eta: 0:06:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2850 (0.3499) acc1: 94.5312 (87.5206) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2149 data: 0.1194 max mem: 3534 +Test: [ 680/2387] eta: 0:05:59 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2944 (0.3545) acc1: 93.4896 (87.0805) acc5: 100.0000 (100.0000) acc1_neg: 41.146 acc5_neg: 100.000 time: 0.1947 data: 0.1002 max mem: 3534 +Test: [ 690/2387] eta: 0:05:57 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2985 (0.3540) acc1: 59.6354 (87.0361) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1754 data: 0.0850 max mem: 3534 +Test: [ 700/2387] eta: 0:05:55 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1302 (0.3508) acc1: 99.2188 (87.2088) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2149 data: 0.1222 max mem: 3534 +Test: [ 710/2387] eta: 0:05:54 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1293 (0.3476) acc1: 99.2188 (87.3802) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2349 data: 0.1418 max mem: 3534 +Test: [ 720/2387] eta: 0:05:51 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1300 (0.3447) acc1: 99.4792 (87.5455) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2249 data: 0.1335 max mem: 3534 +Test: [ 730/2387] eta: 0:05:49 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1308 (0.3417) acc1: 99.2188 (87.7045) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2061 data: 0.1121 max mem: 3534 +Test: [ 740/2387] eta: 0:05:47 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1308 (0.3389) acc1: 99.2188 (87.8581) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2154 data: 0.1198 max mem: 3534 +Test: [ 750/2387] eta: 0:05:45 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1274 (0.3361) acc1: 99.4792 (88.0101) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2218 data: 0.1266 max mem: 3534 +Test: [ 760/2387] eta: 0:05:43 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1295 (0.3346) acc1: 99.2188 (88.0978) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.2144 data: 0.1189 max mem: 3534 +Test: [ 770/2387] eta: 0:05:41 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3441 (0.3351) acc1: 88.8021 (88.1012) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.2053 data: 0.1120 max mem: 3534 +Test: [ 780/2387] eta: 0:05:39 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3441 (0.3353) acc1: 88.2812 (88.0895) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.2073 data: 0.1165 max mem: 3534 +Test: [ 790/2387] eta: 0:05:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3086 (0.3349) acc1: 90.3646 (88.1157) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1999 data: 0.1082 max mem: 3534 +Test: [ 800/2387] eta: 0:05:33 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3086 (0.3350) acc1: 90.3646 (88.1216) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.1805 data: 0.0872 max mem: 3534 +Test: [ 810/2387] eta: 0:05:31 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3343 (0.3353) acc1: 86.7188 (88.0822) acc5: 100.0000 (100.0000) acc1_neg: 72.917 acc5_neg: 100.000 time: 0.1891 data: 0.0963 max mem: 3534 +Test: [ 820/2387] eta: 0:05:29 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3813 (0.3371) acc1: 83.0729 (87.9193) acc5: 100.0000 (100.0000) acc1_neg: 36.979 acc5_neg: 100.000 time: 0.2093 data: 0.1159 max mem: 3534 +Test: [ 830/2387] eta: 0:05:27 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6788 (0.3434) acc1: 51.5625 (87.3634) acc5: 100.0000 (100.0000) acc1_neg: 36.458 acc5_neg: 100.000 time: 0.2114 data: 0.1179 max mem: 3534 +Test: [ 840/2387] eta: 0:05:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3531 (0.3428) acc1: 57.8125 (87.4322) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2081 data: 0.1164 max mem: 3534 +Test: [ 850/2387] eta: 0:05:23 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2382 (0.3414) acc1: 96.0938 (87.5401) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2006 data: 0.1087 max mem: 3534 +Test: [ 860/2387] eta: 0:05:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2264 (0.3400) acc1: 96.6146 (87.6322) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2028 data: 0.1113 max mem: 3534 +Test: [ 870/2387] eta: 0:05:19 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2205 (0.3385) acc1: 96.6146 (87.7392) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2240 data: 0.1330 max mem: 3534 +Test: [ 880/2387] eta: 0:05:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2246 (0.3373) acc1: 96.6146 (87.8278) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2099 data: 0.1192 max mem: 3534 +Test: [ 890/2387] eta: 0:05:14 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2917 (0.3395) acc1: 81.5104 (87.6160) acc5: 100.0000 (100.0000) acc1_neg: 52.865 acc5_neg: 100.000 time: 0.1988 data: 0.1065 max mem: 3534 +Test: [ 900/2387] eta: 0:05:12 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4536 (0.3390) acc1: 77.3438 (87.6734) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2083 data: 0.1160 max mem: 3535 +Test: [ 910/2387] eta: 0:05:10 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2370 (0.3378) acc1: 97.9167 (87.7884) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2123 data: 0.1215 max mem: 3535 +Test: [ 920/2387] eta: 0:05:08 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2349 (0.3367) acc1: 98.1771 (87.9032) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2033 data: 0.1122 max mem: 3535 +Test: [ 930/2387] eta: 0:05:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2342 (0.3357) acc1: 98.1771 (88.0099) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2073 data: 0.1157 max mem: 3535 +Test: [ 940/2387] eta: 0:05:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2335 (0.3346) acc1: 97.9167 (88.1199) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2156 data: 0.1233 max mem: 3535 +Test: [ 950/2387] eta: 0:05:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2333 (0.3336) acc1: 98.4375 (88.2265) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2121 data: 0.1182 max mem: 3535 +Test: [ 960/2387] eta: 0:05:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2362 (0.3326) acc1: 98.1771 (88.3289) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2251 data: 0.1314 max mem: 3535 +Test: [ 970/2387] eta: 0:04:58 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2371 (0.3316) acc1: 97.9167 (88.4314) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2191 data: 0.1268 max mem: 3535 +Test: [ 980/2387] eta: 0:04:56 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2410 (0.3324) acc1: 97.6562 (88.3609) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.2236 data: 0.1274 max mem: 3535 +Test: [ 990/2387] eta: 0:04:54 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5241 (0.3351) acc1: 71.6146 (88.1215) acc5: 100.0000 (100.0000) acc1_neg: 71.354 acc5_neg: 100.000 time: 0.2312 data: 0.1357 max mem: 3535 +Test: [1000/2387] eta: 0:04:51 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5543 (0.3371) acc1: 68.2292 (87.9532) acc5: 100.0000 (100.0000) acc1_neg: 66.927 acc5_neg: 100.000 time: 0.2021 data: 0.1105 max mem: 3535 +Test: [1010/2387] eta: 0:04:50 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6082 (0.3419) acc1: 65.3646 (87.5520) acc5: 100.0000 (100.0000) acc1_neg: 39.583 acc5_neg: 100.000 time: 0.2034 data: 0.1075 max mem: 3535 +Test: [1020/2387] eta: 0:04:48 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6681 (0.3463) acc1: 39.5833 (87.1886) acc5: 100.0000 (100.0000) acc1_neg: 66.927 acc5_neg: 100.000 time: 0.2227 data: 0.1268 max mem: 3535 +Test: [1030/2387] eta: 0:04:45 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5360 (0.3477) acc1: 66.9271 (87.0539) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.2011 data: 0.1083 max mem: 3535 +Test: [1040/2387] eta: 0:04:43 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4092 (0.3480) acc1: 80.4688 (87.0412) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.1905 data: 0.0961 max mem: 3535 +Test: [1050/2387] eta: 0:04:41 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4178 (0.3492) acc1: 82.0312 (86.9566) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.1951 data: 0.1007 max mem: 3535 +Test: [1060/2387] eta: 0:04:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4890 (0.3508) acc1: 76.3021 (86.8361) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.2011 data: 0.1086 max mem: 3535 +Test: [1070/2387] eta: 0:04:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5497 (0.3541) acc1: 69.7917 (86.5512) acc5: 100.0000 (100.0000) acc1_neg: 27.865 acc5_neg: 100.000 time: 0.2032 data: 0.1119 max mem: 3535 +Test: [1080/2387] eta: 0:04:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.9462 (0.3606) acc1: 32.5521 (86.0360) acc5: 100.0000 (100.0000) acc1_neg: 61.458 acc5_neg: 100.000 time: 0.1911 data: 0.0980 max mem: 3535 +Test: [1090/2387] eta: 0:04:32 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5853 (0.3616) acc1: 61.4583 (85.9673) acc5: 100.0000 (100.0000) acc1_neg: 69.010 acc5_neg: 100.000 time: 0.2089 data: 0.1152 max mem: 3535 +Test: [1100/2387] eta: 0:04:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4363 (0.3624) acc1: 78.9062 (85.9023) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.2083 data: 0.1160 max mem: 3535 +Test: [1110/2387] eta: 0:04:27 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4330 (0.3630) acc1: 79.1667 (85.8381) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.1799 data: 0.0872 max mem: 3535 +Test: [1120/2387] eta: 0:04:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3923 (0.3634) acc1: 81.5104 (85.8116) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1955 data: 0.1028 max mem: 3536 +Test: [1130/2387] eta: 0:04:23 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4069 (0.3645) acc1: 82.2917 (85.7178) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.2016 data: 0.1104 max mem: 3536 +Test: [1140/2387] eta: 0:04:21 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6151 (0.3700) acc1: 61.1979 (85.2512) acc5: 100.0000 (100.0000) acc1_neg: 14.583 acc5_neg: 100.000 time: 0.1989 data: 0.1079 max mem: 3536 +Test: [1150/2387] eta: 0:04:19 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.9396 (0.3709) acc1: 31.5104 (85.1531) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2111 data: 0.1200 max mem: 3536 +Test: [1160/2387] eta: 0:04:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1603 (0.3689) acc1: 100.0000 (85.2785) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1991 data: 0.1083 max mem: 3536 +Test: [1170/2387] eta: 0:04:14 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1393 (0.3670) acc1: 100.0000 (85.3995) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1920 data: 0.1003 max mem: 3536 +Test: [1180/2387] eta: 0:04:11 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1306 (0.3649) acc1: 99.7396 (85.5190) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1756 data: 0.0831 max mem: 3536 +Test: [1190/2387] eta: 0:04:09 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1429 (0.3632) acc1: 99.2188 (85.6303) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1667 data: 0.0747 max mem: 3536 +Test: [1200/2387] eta: 0:04:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1723 (0.3622) acc1: 99.2188 (85.7165) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1895 data: 0.0983 max mem: 3536 +Test: [1210/2387] eta: 0:04:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2748 (0.3616) acc1: 94.0104 (85.7599) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2097 data: 0.1183 max mem: 3536 +Test: [1220/2387] eta: 0:04:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2442 (0.3607) acc1: 92.7083 (85.8223) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.2114 data: 0.1205 max mem: 3536 +Test: [1230/2387] eta: 0:04:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2232 (0.3594) acc1: 94.2708 (85.8971) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2052 data: 0.1147 max mem: 3536 +Test: [1240/2387] eta: 0:03:58 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1935 (0.3581) acc1: 95.0521 (85.9738) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2044 data: 0.1137 max mem: 3536 +Test: [1250/2387] eta: 0:03:57 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1927 (0.3570) acc1: 94.7917 (86.0401) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2179 data: 0.1258 max mem: 3536 +Test: [1260/2387] eta: 0:03:54 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2284 (0.3560) acc1: 94.0104 (86.1052) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.2144 data: 0.1213 max mem: 3536 +Test: [1270/2387] eta: 0:03:52 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2597 (0.3564) acc1: 91.4062 (86.0666) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.2061 data: 0.1143 max mem: 3536 +Test: [1280/2387] eta: 0:03:50 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3901 (0.3568) acc1: 83.8542 (86.0534) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.2167 data: 0.1259 max mem: 3536 +Test: [1290/2387] eta: 0:03:48 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3875 (0.3571) acc1: 85.4167 (86.0341) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1786 data: 0.0868 max mem: 3536 +Test: [1300/2387] eta: 0:03:46 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3925 (0.3574) acc1: 83.8542 (86.0172) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1741 data: 0.0810 max mem: 3536 +Test: [1310/2387] eta: 0:03:43 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3974 (0.3579) acc1: 83.3333 (85.9719) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.2012 data: 0.1093 max mem: 3536 +Test: [1320/2387] eta: 0:03:41 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4048 (0.3585) acc1: 82.0312 (85.9211) acc5: 100.0000 (100.0000) acc1_neg: 69.010 acc5_neg: 100.000 time: 0.1843 data: 0.0932 max mem: 3536 +Test: [1330/2387] eta: 0:03:39 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5072 (0.3616) acc1: 71.3542 (85.6524) acc5: 100.0000 (100.0000) acc1_neg: 35.677 acc5_neg: 100.000 time: 0.1836 data: 0.0920 max mem: 3536 +Test: [1340/2387] eta: 0:03:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.7642 (0.3639) acc1: 46.6146 (85.4250) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1836 data: 0.0917 max mem: 3536 +Test: [1350/2387] eta: 0:03:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1076 (0.3619) acc1: 98.9583 (85.5321) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1703 data: 0.0743 max mem: 3537 +Test: [1360/2387] eta: 0:03:32 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0767 (0.3597) acc1: 100.0000 (85.6384) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1723 data: 0.0756 max mem: 3537 +Test: [1370/2387] eta: 0:03:29 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0657 (0.3577) acc1: 100.0000 (85.7411) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1695 data: 0.0746 max mem: 3537 +Test: [1380/2387] eta: 0:03:27 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0669 (0.3556) acc1: 100.0000 (85.8443) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1599 data: 0.0649 max mem: 3537 +Test: [1390/2387] eta: 0:03:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0677 (0.3536) acc1: 100.0000 (85.9461) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1649 data: 0.0711 max mem: 3537 +Test: [1400/2387] eta: 0:03:23 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0891 (0.3522) acc1: 100.0000 (86.0407) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1824 data: 0.0886 max mem: 3537 +Test: [1410/2387] eta: 0:03:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2054 (0.3532) acc1: 98.4375 (85.9604) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.1851 data: 0.0916 max mem: 3537 +Test: [1420/2387] eta: 0:03:18 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4945 (0.3542) acc1: 72.3958 (85.8719) acc5: 100.0000 (100.0000) acc1_neg: 77.344 acc5_neg: 100.000 time: 0.1717 data: 0.0782 max mem: 3537 +Test: [1430/2387] eta: 0:03:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4945 (0.3552) acc1: 72.1354 (85.8003) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.1597 data: 0.0638 max mem: 3537 +Test: [1440/2387] eta: 0:03:13 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5565 (0.3575) acc1: 70.5729 (85.5965) acc5: 100.0000 (100.0000) acc1_neg: 32.552 acc5_neg: 100.000 time: 0.1397 data: 0.0423 max mem: 3537 +Test: [1450/2387] eta: 0:03:11 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.7141 (0.3609) acc1: 39.8438 (85.3133) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1865 data: 0.0912 max mem: 3537 +Test: [1460/2387] eta: 0:03:09 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2921 (0.3603) acc1: 90.1042 (85.3652) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.2084 data: 0.1146 max mem: 3537 +Test: [1470/2387] eta: 0:03:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2193 (0.3592) acc1: 94.7917 (85.4436) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1334 data: 0.0371 max mem: 3537 +Test: [1480/2387] eta: 0:03:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2173 (0.3584) acc1: 95.5729 (85.5042) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1223 data: 0.0231 max mem: 3537 +Test: [1490/2387] eta: 0:03:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2156 (0.3574) acc1: 95.3125 (85.5726) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1848 data: 0.0895 max mem: 3537 +Test: [1500/2387] eta: 0:03:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2206 (0.3566) acc1: 94.7917 (85.6276) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.2075 data: 0.1135 max mem: 3537 +Test: [1510/2387] eta: 0:02:58 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3009 (0.3572) acc1: 87.5000 (85.5847) acc5: 100.0000 (100.0000) acc1_neg: 76.042 acc5_neg: 100.000 time: 0.1699 data: 0.0741 max mem: 3537 +Test: [1520/2387] eta: 0:02:55 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3667 (0.3565) acc1: 87.5000 (85.6420) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1599 data: 0.0639 max mem: 3537 +Test: [1530/2387] eta: 0:02:53 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1782 (0.3553) acc1: 97.6562 (85.7247) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1699 data: 0.0735 max mem: 3537 +Test: [1540/2387] eta: 0:02:51 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1693 (0.3541) acc1: 98.6979 (85.8026) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2099 data: 0.1123 max mem: 3537 +Test: [1550/2387] eta: 0:02:49 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1659 (0.3529) acc1: 98.6979 (85.8843) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2199 data: 0.1223 max mem: 3537 +Test: [1560/2387] eta: 0:02:47 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1698 (0.3518) acc1: 98.4375 (85.9619) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1449 data: 0.0472 max mem: 3537 +Test: [1570/2387] eta: 0:02:45 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2145 (0.3520) acc1: 95.8333 (85.9498) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.1649 data: 0.0693 max mem: 3537 +Test: [1580/2387] eta: 0:02:43 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3878 (0.3523) acc1: 84.6354 (85.9359) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.2170 data: 0.1255 max mem: 3538 +Test: [1590/2387] eta: 0:02:41 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3670 (0.3524) acc1: 85.4167 (85.9313) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1921 data: 0.1002 max mem: 3538 +Test: [1600/2387] eta: 0:02:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3359 (0.3522) acc1: 86.4583 (85.9453) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.1628 data: 0.0693 max mem: 3538 +Test: [1610/2387] eta: 0:02:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3175 (0.3522) acc1: 87.5000 (85.9501) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1684 data: 0.0754 max mem: 3538 +Test: [1620/2387] eta: 0:02:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3420 (0.3523) acc1: 86.1979 (85.9496) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1966 data: 0.1053 max mem: 3538 +Test: [1630/2387] eta: 0:02:32 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3959 (0.3540) acc1: 78.6458 (85.7884) acc5: 100.0000 (100.0000) acc1_neg: 41.406 acc5_neg: 100.000 time: 0.1701 data: 0.0790 max mem: 3538 +Test: [1640/2387] eta: 0:02:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.8101 (0.3573) acc1: 44.5312 (85.4943) acc5: 100.0000 (100.0000) acc1_neg: 42.969 acc5_neg: 100.000 time: 0.1700 data: 0.0794 max mem: 3538 +Test: [1650/2387] eta: 0:02:28 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.8579 (0.3602) acc1: 36.7188 (85.2350) acc5: 100.0000 (100.0000) acc1_neg: 48.958 acc5_neg: 100.000 time: 0.1934 data: 0.1019 max mem: 3538 +Test: [1660/2387] eta: 0:02:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.8562 (0.3631) acc1: 39.8438 (84.9603) acc5: 100.0000 (100.0000) acc1_neg: 51.823 acc5_neg: 100.000 time: 0.1589 data: 0.0671 max mem: 3538 +Test: [1670/2387] eta: 0:02:24 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.8417 (0.3660) acc1: 41.1458 (84.6988) acc5: 100.0000 (100.0000) acc1_neg: 45.312 acc5_neg: 100.000 time: 0.2092 data: 0.1182 max mem: 3538 +Test: [1680/2387] eta: 0:02:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.7969 (0.3686) acc1: 43.7500 (84.4466) acc5: 100.0000 (100.0000) acc1_neg: 40.104 acc5_neg: 100.000 time: 0.2018 data: 0.1110 max mem: 3538 +Test: [1690/2387] eta: 0:02:19 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.8035 (0.3715) acc1: 40.1042 (84.1805) acc5: 100.0000 (100.0000) acc1_neg: 38.021 acc5_neg: 100.000 time: 0.1336 data: 0.0433 max mem: 3538 +Test: [1700/2387] eta: 0:02:17 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.8701 (0.3735) acc1: 36.4583 (83.9938) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1820 data: 0.0907 max mem: 3538 +Test: [1710/2387] eta: 0:02:15 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2718 (0.3728) acc1: 91.9271 (84.0447) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2052 data: 0.1130 max mem: 3538 +Test: [1720/2387] eta: 0:02:13 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2271 (0.3718) acc1: 93.7500 (84.1094) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1809 data: 0.0900 max mem: 3538 +Test: [1730/2387] eta: 0:02:11 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1905 (0.3708) acc1: 95.0521 (84.1754) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1829 data: 0.0925 max mem: 3538 +Test: [1740/2387] eta: 0:02:09 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2041 (0.3699) acc1: 94.7917 (84.2338) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1898 data: 0.0988 max mem: 3538 +Test: [1750/2387] eta: 0:02:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2323 (0.3692) acc1: 94.7917 (84.2895) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1640 data: 0.0727 max mem: 3538 +Test: [1760/2387] eta: 0:02:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3024 (0.3697) acc1: 86.9792 (84.2426) acc5: 100.0000 (100.0000) acc1_neg: 69.271 acc5_neg: 100.000 time: 0.1590 data: 0.0664 max mem: 3538 +Test: [1770/2387] eta: 0:02:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3971 (0.3695) acc1: 79.1667 (84.2555) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.1649 data: 0.0725 max mem: 3538 +Test: [1780/2387] eta: 0:02:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2501 (0.3689) acc1: 92.7083 (84.2925) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1711 data: 0.0802 max mem: 3538 +Test: [1790/2387] eta: 0:01:59 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2202 (0.3680) acc1: 94.0104 (84.3485) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1787 data: 0.0863 max mem: 3538 +Test: [1800/2387] eta: 0:01:56 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2015 (0.3671) acc1: 95.0521 (84.4081) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1604 data: 0.0656 max mem: 3538 +Test: [1810/2387] eta: 0:01:54 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2261 (0.3664) acc1: 94.0104 (84.4568) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1573 data: 0.0628 max mem: 3539 +Test: [1820/2387] eta: 0:01:52 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2413 (0.3666) acc1: 93.2292 (84.4331) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.1781 data: 0.0849 max mem: 3539 +Test: [1830/2387] eta: 0:01:50 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4112 (0.3669) acc1: 76.5625 (84.4090) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1837 data: 0.0909 max mem: 3539 +Test: [1840/2387] eta: 0:01:48 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2633 (0.3663) acc1: 95.0521 (84.4702) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1849 data: 0.0925 max mem: 3539 +Test: [1850/2387] eta: 0:01:46 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2626 (0.3657) acc1: 95.5729 (84.5300) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1849 data: 0.0916 max mem: 3539 +Test: [1860/2387] eta: 0:01:44 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2601 (0.3652) acc1: 95.3125 (84.5897) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1499 data: 0.0531 max mem: 3539 +Test: [1870/2387] eta: 0:01:42 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2581 (0.3646) acc1: 95.8333 (84.6518) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1894 data: 0.0935 max mem: 3539 +Test: [1880/2387] eta: 0:01:40 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2581 (0.3641) acc1: 95.8333 (84.7096) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2151 data: 0.1239 max mem: 3539 +Test: [1890/2387] eta: 0:01:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2607 (0.3635) acc1: 95.5729 (84.7680) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1450 data: 0.0516 max mem: 3539 +Test: [1900/2387] eta: 0:01:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2609 (0.3631) acc1: 95.5729 (84.8143) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.2010 data: 0.1066 max mem: 3539 +Test: [1910/2387] eta: 0:01:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5215 (0.3644) acc1: 70.0521 (84.7145) acc5: 100.0000 (100.0000) acc1_neg: 64.583 acc5_neg: 100.000 time: 0.2306 data: 0.1375 max mem: 3539 +Test: [1920/2387] eta: 0:01:32 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6048 (0.3658) acc1: 65.6250 (84.5824) acc5: 100.0000 (100.0000) acc1_neg: 29.427 acc5_neg: 100.000 time: 0.1381 data: 0.0421 max mem: 3539 +Test: [1930/2387] eta: 0:01:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5195 (0.3662) acc1: 71.8750 (84.5525) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1445 data: 0.0498 max mem: 3539 +Test: [1940/2387] eta: 0:01:28 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5086 (0.3671) acc1: 73.1771 (84.4756) acc5: 100.0000 (100.0000) acc1_neg: 72.135 acc5_neg: 100.000 time: 0.1849 data: 0.0933 max mem: 3539 +Test: [1950/2387] eta: 0:01:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5221 (0.3681) acc1: 70.3125 (84.3830) acc5: 100.0000 (100.0000) acc1_neg: 65.365 acc5_neg: 100.000 time: 0.1817 data: 0.0899 max mem: 3539 +Test: [1960/2387] eta: 0:01:24 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5431 (0.3694) acc1: 65.3646 (84.2676) acc5: 100.0000 (100.0000) acc1_neg: 53.125 acc5_neg: 100.000 time: 0.1949 data: 0.0999 max mem: 3539 +Test: [1970/2387] eta: 0:01:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.8278 (0.3724) acc1: 41.6667 (84.0159) acc5: 100.0000 (100.0000) acc1_neg: 38.281 acc5_neg: 100.000 time: 0.2078 data: 0.1123 max mem: 3539 +Test: [1980/2387] eta: 0:01:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.9231 (0.3734) acc1: 36.4583 (83.9136) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1899 data: 0.0983 max mem: 3539 +Test: [1990/2387] eta: 0:01:18 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1263 (0.3722) acc1: 99.2188 (83.9919) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1799 data: 0.0886 max mem: 3539 +Test: [2000/2387] eta: 0:01:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1255 (0.3709) acc1: 99.4792 (84.0692) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1799 data: 0.0883 max mem: 3539 +Test: [2010/2387] eta: 0:01:14 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1259 (0.3697) acc1: 99.4792 (84.1448) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1778 data: 0.0865 max mem: 3539 +Test: [2020/2387] eta: 0:01:12 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1258 (0.3685) acc1: 99.4792 (84.2211) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1787 data: 0.0868 max mem: 3539 +Test: [2030/2387] eta: 0:01:10 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1262 (0.3673) acc1: 99.4792 (84.2954) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1729 data: 0.0802 max mem: 3540 +Test: [2040/2387] eta: 0:01:08 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1244 (0.3661) acc1: 99.4792 (84.3702) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1993 data: 0.1068 max mem: 3540 +Test: [2050/2387] eta: 0:01:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1250 (0.3652) acc1: 99.4792 (84.4401) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1815 data: 0.0905 max mem: 3540 +Test: [2060/2387] eta: 0:01:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2454 (0.3646) acc1: 97.1354 (84.4973) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1617 data: 0.0687 max mem: 3540 +Test: [2070/2387] eta: 0:01:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2494 (0.3641) acc1: 96.0938 (84.5557) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1907 data: 0.0939 max mem: 3540 +Test: [2080/2387] eta: 0:01:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2513 (0.3635) acc1: 96.3542 (84.6128) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1770 data: 0.0828 max mem: 3540 +Test: [2090/2387] eta: 0:00:58 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2479 (0.3630) acc1: 96.3542 (84.6714) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1591 data: 0.0659 max mem: 3540 +Test: [2100/2387] eta: 0:00:56 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2464 (0.3624) acc1: 96.3542 (84.7285) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1696 data: 0.0742 max mem: 3540 +Test: [2110/2387] eta: 0:00:54 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2503 (0.3619) acc1: 96.0938 (84.7818) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1962 data: 0.1023 max mem: 3540 +Test: [2120/2387] eta: 0:00:52 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2647 (0.3624) acc1: 94.7917 (84.7526) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.2041 data: 0.1123 max mem: 3540 +Test: [2130/2387] eta: 0:00:50 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4815 (0.3629) acc1: 75.7812 (84.7120) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.2177 data: 0.1251 max mem: 3540 +Test: [2140/2387] eta: 0:00:48 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4413 (0.3632) acc1: 77.6042 (84.6899) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.2011 data: 0.1084 max mem: 3540 +Test: [2150/2387] eta: 0:00:46 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4230 (0.3637) acc1: 80.7292 (84.6674) acc5: 100.0000 (100.0000) acc1_neg: 75.781 acc5_neg: 100.000 time: 0.2002 data: 0.1083 max mem: 3540 +Test: [2160/2387] eta: 0:00:44 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5026 (0.3644) acc1: 75.0000 (84.6072) acc5: 100.0000 (100.0000) acc1_neg: 49.479 acc5_neg: 100.000 time: 0.2152 data: 0.1230 max mem: 3540 +Test: [2170/2387] eta: 0:00:42 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5972 (0.3667) acc1: 60.1562 (84.4125) acc5: 100.0000 (100.0000) acc1_neg: 46.354 acc5_neg: 100.000 time: 0.2150 data: 0.1227 max mem: 3540 +Test: [2180/2387] eta: 0:00:40 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.8818 (0.3681) acc1: 40.6250 (84.2979) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2138 data: 0.1221 max mem: 3540 +Test: [2190/2387] eta: 0:00:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2425 (0.3676) acc1: 98.4375 (84.3528) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1998 data: 0.1089 max mem: 3540 +Test: [2200/2387] eta: 0:00:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3642 (0.3680) acc1: 88.2812 (84.3395) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.1818 data: 0.0897 max mem: 3540 +Test: [2210/2387] eta: 0:00:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4123 (0.3679) acc1: 84.6354 (84.3633) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1687 data: 0.0763 max mem: 3540 +Test: [2220/2387] eta: 0:00:32 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3550 (0.3680) acc1: 89.5833 (84.3809) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.1914 data: 0.0990 max mem: 3540 +Test: [2230/2387] eta: 0:00:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3596 (0.3679) acc1: 89.5833 (84.4077) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1942 data: 0.1017 max mem: 3540 +Test: [2240/2387] eta: 0:00:28 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3676 (0.3681) acc1: 89.0625 (84.4127) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.1865 data: 0.0951 max mem: 3540 +Test: [2250/2387] eta: 0:00:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4298 (0.3685) acc1: 84.6354 (84.3929) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1732 data: 0.0816 max mem: 3540 +Test: [2260/2387] eta: 0:00:24 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4594 (0.3689) acc1: 77.6042 (84.3606) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1591 data: 0.0669 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4293 (0.3691) acc1: 80.2083 (84.3494) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1481 data: 0.0563 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3775 (0.3691) acc1: 80.9896 (84.3491) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.1436 data: 0.0528 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3775 (0.3693) acc1: 83.0729 (84.3424) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.1342 data: 0.0441 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4303 (0.3697) acc1: 79.9479 (84.3184) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.1243 data: 0.0344 max mem: 3541 +Test: [2310/2387] eta: 0:00:15 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5074 (0.3712) acc1: 72.3958 (84.1834) acc5: 100.0000 (100.0000) acc1_neg: 16.927 acc5_neg: 100.000 time: 0.1300 data: 0.0396 max mem: 3541 +Test: [2320/2387] eta: 0:00:13 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.7655 (0.3732) acc1: 50.0000 (84.0289) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.1418 data: 0.0516 max mem: 3541 +Test: [2330/2387] eta: 0:00:11 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6846 (0.3741) acc1: 53.1250 (83.9374) acc5: 100.0000 (100.0000) acc1_neg: 52.865 acc5_neg: 100.000 time: 0.1477 data: 0.0580 max mem: 3541 +Test: [2340/2387] eta: 0:00:09 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4915 (0.3747) acc1: 70.8333 (83.8912) acc5: 100.0000 (100.0000) acc1_neg: 79.427 acc5_neg: 100.000 time: 0.1429 data: 0.0527 max mem: 3541 +Test: [2350/2387] eta: 0:00:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4937 (0.3755) acc1: 72.1354 (83.8360) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.1350 data: 0.0443 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5721 (0.3766) acc1: 67.4479 (83.7434) acc5: 100.0000 (100.0000) acc1_neg: 63.802 acc5_neg: 100.000 time: 0.1321 data: 0.0418 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5784 (0.3779) acc1: 63.8021 (83.6348) acc5: 100.0000 (100.0000) acc1_neg: 47.135 acc5_neg: 100.000 time: 0.1336 data: 0.0432 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.8797 (0.3807) acc1: 36.7188 (83.4058) acc5: 100.0000 (100.0000) acc1_neg: 32.812 acc5_neg: 100.000 time: 0.1318 data: 0.0408 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.9152 (0.3815) acc1: 32.8125 (83.3459) acc5: 100.0000 (100.0000) acc1_neg: 93.023 acc5_neg: 100.000 time: 0.1252 data: 0.0367 max mem: 3541 +Test: Total time: 0:07:41 (0.1932 s / it) +* Overall Acc@1 83.346 Acc@5 100.000 + target == real Acc@1 80.107 Acc@5 100.000 + target == fake Acc@1 83.573 Acc@5 100.000 + loss 0.382 + PR-AUC 0.9912 ROC-AUC 0.9020 EER 18.12% (thr=0.5229) +Accuracy of the network on the 916267 test images: 83.3% +[rank0]:[W726 05:23:18.985171585 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:23:39.477000 1191623 torch/distributed/run.py:766] +W0726 05:23:39.477000 1191623 torch/distributed/run.py:766] ***************************************** +W0726 05:23:39.477000 1191623 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:23:39.477000 1191623 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 05:24:09.972023262 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:24:09.056700243 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:24:09.088938296 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:24:09.282115814 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:24:09.470796998 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:24:10.533048857 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:24:10.565556097 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank6]:[W726 05:24:10.565556194 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n8_seed8_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed8_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:06:52 acc1_pos: 75.000 acc5_pos: 100.000 loss: 0.5759 (0.5759) acc1: 75.0000 (75.0000) acc5: 100.0000 (100.0000) time: 7.7602 data: 6.3042 max mem: 3531 +Test: [ 10/517] eta: 0:06:41 acc1_pos: 42.448 acc5_pos: 100.000 loss: 0.6564 (0.6792) acc1: 65.8854 (63.0208) acc5: 100.0000 (100.0000) time: 0.7910 data: 0.5736 max mem: 3531 +Test: [ 20/517] eta: 0:03:49 acc1_pos: 67.708 acc5_pos: 100.000 loss: 0.6398 (0.6877) acc1: 65.6250 (62.7356) acc5: 100.0000 (100.0000) time: 0.0970 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:03:21 acc1_pos: 40.365 acc5_pos: 100.000 loss: 0.6398 (0.6939) acc1: 56.2500 (61.3995) acc5: 100.0000 (100.0000) time: 0.2049 data: 0.1042 max mem: 3531 +Test: [ 40/517] eta: 0:03:12 acc1_pos: 74.219 acc5_pos: 100.000 loss: 0.7700 (0.7100) acc1: 50.0000 (59.5147) acc5: 100.0000 (100.0000) time: 0.3412 data: 0.2447 max mem: 3531 +Test: [ 50/517] eta: 0:03:03 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.7616 (0.6996) acc1: 50.0000 (59.6916) acc5: 100.0000 (100.0000) acc1_neg: 25.000 acc5_neg: 100.000 time: 0.3600 data: 0.2677 max mem: 3531 +Test: [ 60/517] eta: 0:02:50 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.8004 (0.7294) acc1: 34.3750 (54.7217) acc5: 100.0000 (100.0000) acc1_neg: 36.719 acc5_neg: 100.000 time: 0.3090 data: 0.2161 max mem: 3531 +Test: [ 70/517] eta: 0:02:37 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.8018 (0.7327) acc1: 37.2396 (53.5688) acc5: 100.0000 (100.0000) acc1_neg: 51.562 acc5_neg: 100.000 time: 0.2498 data: 0.1569 max mem: 3531 +Test: [ 80/517] eta: 0:02:28 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.7932 (0.7427) acc1: 44.2708 (52.2344) acc5: 100.0000 (100.0000) acc1_neg: 35.417 acc5_neg: 100.000 time: 0.2395 data: 0.1457 max mem: 3531 +Test: [ 90/517] eta: 0:02:20 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.6879 (0.7293) acc1: 49.7396 (52.8875) acc5: 100.0000 (100.0000) acc1_neg: 52.604 acc5_neg: 100.000 time: 0.2466 data: 0.1542 max mem: 3531 +Test: [100/517] eta: 0:02:12 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.6859 (0.7366) acc1: 51.0417 (51.8410) acc5: 100.0000 (100.0000) acc1_neg: 44.010 acc5_neg: 100.000 time: 0.2311 data: 0.1406 max mem: 3531 +Test: [110/517] eta: 0:02:06 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.7992 (0.7435) acc1: 40.8854 (50.9173) acc5: 100.0000 (100.0000) acc1_neg: 42.708 acc5_neg: 100.000 time: 0.2231 data: 0.1313 max mem: 3531 +Test: [120/517] eta: 0:02:01 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.8091 (0.7504) acc1: 42.1875 (50.1291) acc5: 100.0000 (100.0000) acc1_neg: 42.188 acc5_neg: 100.000 time: 0.2487 data: 0.1562 max mem: 3531 +Test: [130/517] eta: 0:01:57 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.8065 (0.7548) acc1: 42.1875 (49.4732) acc5: 100.0000 (100.0000) acc1_neg: 44.271 acc5_neg: 100.000 time: 0.2749 data: 0.1813 max mem: 3531 +Test: [140/517] eta: 0:01:53 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.8361 (0.7620) acc1: 36.7188 (48.2103) acc5: 100.0000 (100.0000) acc1_neg: 35.677 acc5_neg: 100.000 time: 0.2699 data: 0.1753 max mem: 3531 +Test: [150/517] eta: 0:01:49 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.8361 (0.7601) acc1: 35.1562 (48.4427) acc5: 100.0000 (100.0000) acc1_neg: 68.490 acc5_neg: 100.000 time: 0.2649 data: 0.1683 max mem: 3531 +Test: [160/517] eta: 0:01:45 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.7839 (0.7609) acc1: 39.3229 (48.0703) acc5: 100.0000 (100.0000) acc1_neg: 55.469 acc5_neg: 100.000 time: 0.2599 data: 0.1633 max mem: 3531 +Test: [170/517] eta: 0:01:42 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.7517 (0.7575) acc1: 43.4896 (48.2273) acc5: 100.0000 (100.0000) acc1_neg: 69.792 acc5_neg: 100.000 time: 0.2654 data: 0.1732 max mem: 3531 +Test: [180/517] eta: 0:01:38 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.7114 (0.7532) acc1: 46.8750 (48.4735) acc5: 100.0000 (100.0000) acc1_neg: 57.812 acc5_neg: 100.000 time: 0.2599 data: 0.1678 max mem: 3531 +Test: [190/517] eta: 0:01:34 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.4587 (0.7335) acc1: 77.0833 (50.5658) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.2494 data: 0.1549 max mem: 3531 +Test: [200/517] eta: 0:01:31 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.4587 (0.7257) acc1: 80.4688 (51.3656) acc5: 100.0000 (100.0000) acc1_neg: 57.552 acc5_neg: 100.000 time: 0.2599 data: 0.1636 max mem: 3531 +Test: [210/517] eta: 0:01:28 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.6351 (0.7226) acc1: 58.3333 (51.6538) acc5: 100.0000 (100.0000) acc1_neg: 53.646 acc5_neg: 100.000 time: 0.2599 data: 0.1647 max mem: 3532 +Test: [220/517] eta: 0:01:25 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.6603 (0.7207) acc1: 57.5521 (51.8877) acc5: 100.0000 (100.0000) acc1_neg: 50.000 acc5_neg: 100.000 time: 0.2749 data: 0.1800 max mem: 3532 +Test: [230/517] eta: 0:01:22 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.7913 (0.7271) acc1: 41.6667 (51.2029) acc5: 100.0000 (100.0000) acc1_neg: 31.510 acc5_neg: 100.000 time: 0.2699 data: 0.1741 max mem: 3532 +Test: [240/517] eta: 0:01:19 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.8811 (0.7341) acc1: 33.5938 (50.2896) acc5: 100.0000 (100.0000) acc1_neg: 5.208 acc5_neg: 100.000 time: 0.2649 data: 0.1684 max mem: 3532 +Test: [250/517] eta: 0:01:16 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.8744 (0.7386) acc1: 29.4271 (49.5995) acc5: 100.0000 (100.0000) acc1_neg: 30.990 acc5_neg: 100.000 time: 0.2699 data: 0.1740 max mem: 3532 +Test: [260/517] eta: 0:01:13 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.8092 (0.7410) acc1: 32.8125 (49.1290) acc5: 100.0000 (100.0000) acc1_neg: 61.719 acc5_neg: 100.000 time: 0.2899 data: 0.1938 max mem: 3532 +Test: [270/517] eta: 0:01:10 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.7989 (0.7434) acc1: 35.6771 (48.6335) acc5: 100.0000 (100.0000) acc1_neg: 35.677 acc5_neg: 100.000 time: 0.2998 data: 0.2029 max mem: 3532 +Test: [280/517] eta: 0:01:07 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.8146 (0.7463) acc1: 35.1562 (48.1409) acc5: 100.0000 (100.0000) acc1_neg: 37.760 acc5_neg: 100.000 time: 0.2655 data: 0.1708 max mem: 3532 +Test: [290/517] eta: 0:01:04 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.8260 (0.7494) acc1: 35.1562 (47.7001) acc5: 100.0000 (100.0000) acc1_neg: 25.521 acc5_neg: 100.000 time: 0.2663 data: 0.1725 max mem: 3532 +Test: [300/517] eta: 0:01:01 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.8038 (0.7518) acc1: 38.2812 (47.5343) acc5: 100.0000 (100.0000) acc1_neg: 28.385 acc5_neg: 100.000 time: 0.2696 data: 0.1772 max mem: 3532 +Test: [310/517] eta: 0:00:58 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.9348 (0.7617) acc1: 32.5521 (46.7527) acc5: 100.0000 (100.0000) acc1_neg: 22.135 acc5_neg: 100.000 time: 0.2747 data: 0.1819 max mem: 3532 +Test: [320/517] eta: 0:00:55 acc1_pos: 70.642 acc5_pos: 100.000 loss: 1.0094 (0.7640) acc1: 25.0000 (46.6771) acc5: 100.0000 (100.0000) acc1_neg: 51.823 acc5_neg: 100.000 time: 0.2696 data: 0.1742 max mem: 3532 +Test: [330/517] eta: 0:00:52 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.6961 (0.7622) acc1: 51.8229 (46.9427) acc5: 100.0000 (100.0000) acc1_neg: 41.406 acc5_neg: 100.000 time: 0.2637 data: 0.1678 max mem: 3532 +Test: [340/517] eta: 0:00:49 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.7435 (0.7626) acc1: 46.3542 (46.8498) acc5: 100.0000 (100.0000) acc1_neg: 40.625 acc5_neg: 100.000 time: 0.2649 data: 0.1685 max mem: 3532 +Test: [350/517] eta: 0:00:46 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.7522 (0.7622) acc1: 44.7917 (46.8201) acc5: 100.0000 (100.0000) acc1_neg: 28.646 acc5_neg: 100.000 time: 0.2499 data: 0.1539 max mem: 3532 +Test: [360/517] eta: 0:00:44 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.7485 (0.7607) acc1: 46.6146 (46.9825) acc5: 100.0000 (100.0000) acc1_neg: 58.854 acc5_neg: 100.000 time: 0.2649 data: 0.1698 max mem: 3532 +Test: [370/517] eta: 0:00:41 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.5839 (0.7548) acc1: 67.1875 (47.6991) acc5: 100.0000 (100.0000) acc1_neg: 73.698 acc5_neg: 100.000 time: 0.2699 data: 0.1733 max mem: 3532 +Test: [380/517] eta: 0:00:38 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.5120 (0.7473) acc1: 77.6042 (48.6405) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.2949 data: 0.1967 max mem: 3532 +Test: [390/517] eta: 0:00:35 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.5120 (0.7436) acc1: 77.3438 (49.2567) acc5: 100.0000 (100.0000) acc1_neg: 61.719 acc5_neg: 100.000 time: 0.2799 data: 0.1823 max mem: 3532 +Test: [400/517] eta: 0:00:32 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.6025 (0.7423) acc1: 72.6562 (49.6240) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.2406 data: 0.1451 max mem: 3532 +Test: [410/517] eta: 0:00:29 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.6644 (0.7392) acc1: 59.3750 (49.9100) acc5: 100.0000 (100.0000) acc1_neg: 49.740 acc5_neg: 100.000 time: 0.2554 data: 0.1614 max mem: 3532 +Test: [420/517] eta: 0:00:27 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.7498 (0.7414) acc1: 48.9583 (49.7210) acc5: 100.0000 (100.0000) acc1_neg: 52.083 acc5_neg: 100.000 time: 0.2692 data: 0.1753 max mem: 3532 +Test: [430/517] eta: 0:00:24 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.8389 (0.7444) acc1: 42.1875 (49.5487) acc5: 100.0000 (100.0000) acc1_neg: 37.240 acc5_neg: 100.000 time: 0.2594 data: 0.1639 max mem: 3532 +Test: [440/517] eta: 0:00:21 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.8385 (0.7453) acc1: 42.9688 (49.5205) acc5: 100.0000 (100.0000) acc1_neg: 52.865 acc5_neg: 100.000 time: 0.3249 data: 0.2282 max mem: 3533 +Test: [450/517] eta: 0:00:18 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.7155 (0.7444) acc1: 53.1250 (49.6397) acc5: 100.0000 (100.0000) acc1_neg: 65.625 acc5_neg: 100.000 time: 0.3149 data: 0.2189 max mem: 3533 +Test: [460/517] eta: 0:00:15 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.6836 (0.7426) acc1: 56.5104 (49.9187) acc5: 100.0000 (100.0000) acc1_neg: 76.302 acc5_neg: 100.000 time: 0.2449 data: 0.1491 max mem: 3533 +Test: [470/517] eta: 0:00:13 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.6642 (0.7400) acc1: 62.2396 (50.1493) acc5: 100.0000 (100.0000) acc1_neg: 51.042 acc5_neg: 100.000 time: 0.2464 data: 0.1508 max mem: 3533 +Test: [480/517] eta: 0:00:10 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.6456 (0.7387) acc1: 60.6771 (50.2388) acc5: 100.0000 (100.0000) acc1_neg: 55.990 acc5_neg: 100.000 time: 0.2020 data: 0.1100 max mem: 3533 +Test: [490/517] eta: 0:00:07 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.7368 (0.7430) acc1: 45.8333 (49.8319) acc5: 100.0000 (100.0000) acc1_neg: 33.854 acc5_neg: 100.000 time: 0.1770 data: 0.0874 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.9027 (0.7464) acc1: 23.4375 (49.3191) acc5: 100.0000 (100.0000) acc1_neg: 25.000 acc5_neg: 100.000 time: 0.1752 data: 0.0857 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.8451 (0.7472) acc1: 25.0000 (49.1056) acc5: 100.0000 (100.0000) acc1_neg: 23.698 acc5_neg: 100.000 time: 0.1729 data: 0.0835 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 70.642 acc5_pos: 100.000 loss: 0.8690 (0.7496) acc1: 23.6979 (48.7817) acc5: 100.0000 (100.0000) acc1_neg: 20.141 acc5_neg: 100.000 time: 0.1532 data: 0.0558 max mem: 3533 +Test: Total time: 0:02:19 (0.2695 s / it) +* Overall Acc@1 48.782 Acc@5 100.000 + target == real Acc@1 61.405 Acc@5 100.000 + target == fake Acc@1 47.472 Acc@5 100.000 + loss 0.750 + PR-AUC 0.9219 ROC-AUC 0.5796 EER 44.97% (thr=0.4604) +Accuracy of the network on the 198427 test images: 48.8% +[rank0]:[W726 05:26:37.951259842 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 05:26:58.038000 1192196 torch/distributed/run.py:766] +W0726 05:26:58.038000 1192196 torch/distributed/run.py:766] ***************************************** +W0726 05:26:58.038000 1192196 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:26:58.038000 1192196 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:27:25.053086136 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:27:25.206333101 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:27:25.387240580 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:27:26.688573832 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:27:26.798186369 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:27:26.864174436 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:27:26.890257988 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W726 05:27:26.890271348 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n8_seed8_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed8_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:31:45 acc1_pos: 19.271 acc5_pos: 100.000 loss: 1.2948 (1.2948) acc1: 19.2708 (19.2708) acc5: 100.0000 (100.0000) time: 4.8997 data: 2.5761 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/389] eta: 0:03:41 acc1_pos: 37.500 acc5_pos: 100.000 loss: 1.2154 (1.1996) acc1: 14.8438 (21.5436) acc5: 100.0000 (100.0000) time: 0.5834 data: 0.2852 max mem: 3531 +Test: [ 20/389] eta: 0:02:19 acc1_pos: 0.521 acc5_pos: 100.000 loss: 1.2154 (1.4237) acc1: 13.2812 (16.5675) acc5: 100.0000 (100.0000) time: 0.1506 data: 0.0544 max mem: 3531 +Test: [ 30/389] eta: 0:01:51 acc1_pos: 19.792 acc5_pos: 100.000 loss: 1.4310 (1.4157) acc1: 3.3854 (14.4825) acc5: 100.0000 (100.0000) time: 0.1616 data: 0.0645 max mem: 3531 +Test: [ 40/389] eta: 0:01:42 acc1_pos: 2.604 acc5_pos: 100.000 loss: 1.3974 (1.4147) acc1: 4.6875 (14.2467) acc5: 100.0000 (100.0000) time: 0.2049 data: 0.1106 max mem: 3531 +Test: [ 50/389] eta: 0:01:33 acc1_pos: 13.281 acc5_pos: 100.000 loss: 1.3974 (1.4032) acc1: 4.6875 (12.9545) acc5: 100.0000 (100.0000) time: 0.2233 data: 0.1299 max mem: 3531 +Test: [ 60/389] eta: 0:01:22 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.1760 (1.4035) acc1: 5.4688 (12.0091) acc5: 100.0000 (100.0000) time: 0.1599 data: 0.0630 max mem: 3531 +Test: [ 70/389] eta: 0:01:17 acc1_pos: 16.667 acc5_pos: 100.000 loss: 1.6311 (1.4509) acc1: 8.5938 (12.1406) acc5: 100.0000 (100.0000) time: 0.1554 data: 0.0605 max mem: 3531 +Test: [ 80/389] eta: 0:01:12 acc1_pos: 6.510 acc5_pos: 100.000 loss: 1.7621 (1.4808) acc1: 1.3021 (12.4679) acc5: 100.0000 (100.0000) time: 0.1949 data: 0.1023 max mem: 3531 +Test: [ 90/389] eta: 0:01:08 acc1_pos: 4.688 acc5_pos: 100.000 loss: 1.4396 (1.4707) acc1: 2.3438 (11.7617) acc5: 100.0000 (100.0000) time: 0.1745 data: 0.0795 max mem: 3531 +Test: [100/389] eta: 0:01:04 acc1_pos: 1.562 acc5_pos: 100.000 loss: 1.3773 (1.4448) acc1: 6.7708 (13.7737) acc5: 100.0000 (100.0000) time: 0.1711 data: 0.0778 max mem: 3531 +Test: [110/389] eta: 0:01:01 acc1_pos: 0.781 acc5_pos: 100.000 loss: 1.4951 (1.4635) acc1: 3.1250 (12.9833) acc5: 100.0000 (100.0000) time: 0.1944 data: 0.1042 max mem: 3531 +Test: [120/389] eta: 0:00:58 acc1_pos: 31.250 acc5_pos: 100.000 loss: 1.3362 (1.4366) acc1: 7.8125 (14.3272) acc5: 100.0000 (100.0000) time: 0.1886 data: 0.0939 max mem: 3531 +Test: [130/389] eta: 0:00:56 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.8620 (1.3716) acc1: 46.3542 (18.7281) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1863 data: 0.0902 max mem: 3531 +Test: [140/389] eta: 0:00:52 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.2091 (1.2931) acc1: 98.4375 (24.1005) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1812 data: 0.0892 max mem: 3531 +Test: [150/389] eta: 0:00:50 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.2270 (1.2257) acc1: 98.4375 (28.7545) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1768 data: 0.0855 max mem: 3531 +Test: [160/389] eta: 0:00:47 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.2270 (1.1750) acc1: 97.6562 (31.8824) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1786 data: 0.0862 max mem: 3531 +Test: [170/389] eta: 0:00:45 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.2039 (1.1233) acc1: 100.0000 (35.4121) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1723 data: 0.0800 max mem: 3531 +Test: [180/389] eta: 0:00:42 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.2238 (1.0783) acc1: 96.6146 (38.5733) acc5: 100.0000 (100.0000) acc1_neg: 68.750 acc5_neg: 100.000 time: 0.1749 data: 0.0817 max mem: 3531 +Test: [190/389] eta: 0:00:40 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.3098 (1.0433) acc1: 92.7083 (40.8418) acc5: 100.0000 (100.0000) acc1_neg: 68.750 acc5_neg: 100.000 time: 0.1741 data: 0.0801 max mem: 3531 +Test: [200/389] eta: 0:00:38 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.1590 (1.0033) acc1: 97.9167 (43.4572) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1755 data: 0.0834 max mem: 3531 +Test: [210/389] eta: 0:00:36 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.1261 (0.9617) acc1: 100.0000 (46.1258) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1825 data: 0.0909 max mem: 3532 +Test: [220/389] eta: 0:00:33 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.1222 (0.9242) acc1: 100.0000 (48.4387) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1820 data: 0.0888 max mem: 3532 +Test: [230/389] eta: 0:00:31 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.1594 (0.8966) acc1: 100.0000 (50.4216) acc5: 100.0000 (100.0000) acc1_neg: 51.562 acc5_neg: 100.000 time: 0.1794 data: 0.0870 max mem: 3532 +Test: [240/389] eta: 0:00:29 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.2087 (0.8684) acc1: 98.4375 (52.3524) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1879 data: 0.0967 max mem: 3532 +Test: [250/389] eta: 0:00:27 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.2160 (0.8430) acc1: 97.6562 (54.1750) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2130 data: 0.1207 max mem: 3532 +Test: [260/389] eta: 0:00:25 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.2162 (0.8242) acc1: 97.9167 (55.2981) acc5: 100.0000 (100.0000) acc1_neg: 43.229 acc5_neg: 100.000 time: 0.1860 data: 0.0926 max mem: 3532 +Test: [270/389] eta: 0:00:23 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.2221 (0.8083) acc1: 97.6562 (56.2125) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1657 data: 0.0735 max mem: 3532 +Test: [280/389] eta: 0:00:21 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.2221 (0.7961) acc1: 97.6562 (56.9052) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1934 data: 0.1033 max mem: 3532 +Test: [290/389] eta: 0:00:19 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.2092 (0.7778) acc1: 96.0938 (58.1740) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.1888 data: 0.0977 max mem: 3532 +Test: [300/389] eta: 0:00:17 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.4262 (0.7704) acc1: 81.7708 (58.5652) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.1902 data: 0.0965 max mem: 3532 +Test: [310/389] eta: 0:00:15 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.5181 (0.7669) acc1: 67.7083 (58.6306) acc5: 100.0000 (100.0000) acc1_neg: 52.865 acc5_neg: 100.000 time: 0.1869 data: 0.0939 max mem: 3532 +Test: [320/389] eta: 0:00:13 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.4629 (0.7550) acc1: 73.4375 (59.4504) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1877 data: 0.0953 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.2234 (0.7387) acc1: 97.6562 (60.6071) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1947 data: 0.1026 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.1833 (0.7223) acc1: 99.7396 (61.6683) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2020 data: 0.1112 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.1346 (0.7092) acc1: 99.7396 (62.4666) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1874 data: 0.0963 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.0972 (0.6923) acc1: 100.0000 (63.5063) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1616 data: 0.0706 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.0972 (0.6784) acc1: 100.0000 (64.4310) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1520 data: 0.0618 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.1673 (0.6681) acc1: 99.2188 (65.1021) acc5: 100.0000 (100.0000) acc1_neg: 66.406 acc5_neg: 100.000 time: 0.1391 data: 0.0496 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 0.613 acc5_pos: 100.000 loss: 0.2557 (0.6602) acc1: 93.7500 (65.6389) acc5: 100.0000 (100.0000) acc1_neg: 98.087 acc5_neg: 100.000 time: 0.1357 data: 0.0428 max mem: 3532 +Test: Total time: 0:01:14 (0.1913 s / it) +* Overall Acc@1 65.639 Acc@5 100.000 + target == real Acc@1 14.086 Acc@5 100.000 + target == fake Acc@1 89.602 Acc@5 100.000 + loss 0.660 + PR-AUC 0.7971 ROC-AUC 0.6462 EER 38.04% (thr=0.8042) +Accuracy of the network on the 149358 test images: 65.6% +[rank0]:[W726 05:28:50.916314625 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:29:09.826000 1192755 torch/distributed/run.py:766] +W0726 05:29:09.826000 1192755 torch/distributed/run.py:766] ***************************************** +W0726 05:29:09.826000 1192755 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:29:09.826000 1192755 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 1): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:29:41.038180645 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W726 05:29:41.051904402 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:29:42.772334533 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:29:42.020671762 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank6]:[W726 05:29:42.020706710 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:29:42.037335507 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:29:42.040871031 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:29:42.067958146 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n8_seed8_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed8_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:42:19 acc1_pos: 17.969 acc5_pos: 100.000 loss: 1.3036 (1.3036) acc1: 17.9688 (17.9688) acc5: 100.0000 (100.0000) time: 7.6272 data: 5.6504 max mem: 3531 +Test: [ 10/333] eta: 0:04:13 acc1_pos: 90.625 acc5_pos: 100.000 loss: 1.0218 (0.8300) acc1: 37.5000 (51.9176) acc5: 100.0000 (100.0000) time: 0.7835 data: 0.5141 max mem: 3531 +Test: [ 20/333] eta: 0:02:23 acc1_pos: 90.104 acc5_pos: 100.000 loss: 0.3950 (0.6113) acc1: 81.7708 (67.6711) acc5: 100.0000 (100.0000) time: 0.0995 data: 0.0004 max mem: 3531 +Test: [ 30/333] eta: 0:02:04 acc1_pos: 73.438 acc5_pos: 100.000 loss: 0.4194 (0.5972) acc1: 78.6458 (68.8592) acc5: 100.0000 (100.0000) time: 0.2061 data: 0.1086 max mem: 3531 +Test: [ 40/333] eta: 0:01:48 acc1_pos: 67.188 acc5_pos: 100.000 loss: 0.6062 (0.6047) acc1: 66.4062 (68.3562) acc5: 100.0000 (100.0000) time: 0.2807 data: 0.1879 max mem: 3531 +Test: [ 50/333] eta: 0:01:36 acc1_pos: 59.115 acc5_pos: 100.000 loss: 0.6460 (0.6060) acc1: 64.8438 (68.4538) acc5: 100.0000 (100.0000) time: 0.2337 data: 0.1432 max mem: 3531 +Test: [ 60/333] eta: 0:01:27 acc1_pos: 68.750 acc5_pos: 100.000 loss: 0.6348 (0.6096) acc1: 67.1875 (68.5408) acc5: 100.0000 (100.0000) time: 0.2202 data: 0.1269 max mem: 3531 +Test: [ 70/333] eta: 0:01:20 acc1_pos: 74.740 acc5_pos: 100.000 loss: 0.5890 (0.6067) acc1: 70.0521 (68.7463) acc5: 100.0000 (100.0000) time: 0.2220 data: 0.1274 max mem: 3531 +Test: [ 80/333] eta: 0:01:14 acc1_pos: 63.802 acc5_pos: 100.000 loss: 0.5957 (0.6128) acc1: 69.5312 (68.5571) acc5: 100.0000 (100.0000) time: 0.2047 data: 0.1105 max mem: 3531 +Test: [ 90/333] eta: 0:01:07 acc1_pos: 72.917 acc5_pos: 100.000 loss: 0.5987 (0.6069) acc1: 69.0104 (68.9045) acc5: 100.0000 (100.0000) time: 0.1752 data: 0.0778 max mem: 3531 +Test: [100/333] eta: 0:01:04 acc1_pos: 75.521 acc5_pos: 100.000 loss: 0.5636 (0.6005) acc1: 69.2708 (69.4591) acc5: 100.0000 (100.0000) time: 0.2040 data: 0.1019 max mem: 3531 +Test: [110/333] eta: 0:00:59 acc1_pos: 66.927 acc5_pos: 100.000 loss: 0.5474 (0.5973) acc1: 75.0000 (69.9864) acc5: 100.0000 (100.0000) time: 0.2189 data: 0.1204 max mem: 3531 +Test: [120/333] eta: 0:00:55 acc1_pos: 72.396 acc5_pos: 100.000 loss: 0.5638 (0.5959) acc1: 71.0938 (70.0542) acc5: 100.0000 (100.0000) time: 0.1918 data: 0.0971 max mem: 3531 +Test: [130/333] eta: 0:00:52 acc1_pos: 81.250 acc5_pos: 100.000 loss: 0.5591 (0.5926) acc1: 71.6146 (70.4000) acc5: 100.0000 (100.0000) time: 0.1986 data: 0.1026 max mem: 3531 +Test: [140/333] eta: 0:00:48 acc1_pos: 72.135 acc5_pos: 100.000 loss: 0.5110 (0.5855) acc1: 74.4792 (70.9792) acc5: 100.0000 (100.0000) time: 0.2033 data: 0.1080 max mem: 3531 +Test: [150/333] eta: 0:00:45 acc1_pos: 75.260 acc5_pos: 100.000 loss: 0.5210 (0.5844) acc1: 73.1771 (71.0179) acc5: 100.0000 (100.0000) time: 0.1916 data: 0.0968 max mem: 3531 +Test: [160/333] eta: 0:00:42 acc1_pos: 64.145 acc5_pos: 100.000 loss: 0.5461 (0.5933) acc1: 69.7917 (70.2640) acc5: 100.0000 (100.0000) acc1_neg: 31.510 acc5_neg: 100.000 time: 0.1847 data: 0.0905 max mem: 3531 +Test: [170/333] eta: 0:00:39 acc1_pos: 64.145 acc5_pos: 100.000 loss: 0.9584 (0.6159) acc1: 35.4167 (68.1835) acc5: 100.0000 (100.0000) acc1_neg: 37.500 acc5_neg: 100.000 time: 0.1831 data: 0.0905 max mem: 3531 +Test: [180/333] eta: 0:00:36 acc1_pos: 64.145 acc5_pos: 100.000 loss: 0.9317 (0.6312) acc1: 36.4583 (66.6681) acc5: 100.0000 (100.0000) acc1_neg: 45.312 acc5_neg: 100.000 time: 0.1759 data: 0.0836 max mem: 3531 +Test: [190/333] eta: 0:00:33 acc1_pos: 64.145 acc5_pos: 100.000 loss: 0.8759 (0.6428) acc1: 41.9271 (65.4041) acc5: 100.0000 (100.0000) acc1_neg: 49.479 acc5_neg: 100.000 time: 0.1857 data: 0.0935 max mem: 3531 +Test: [200/333] eta: 0:00:31 acc1_pos: 64.145 acc5_pos: 100.000 loss: 0.8790 (0.6549) acc1: 41.9271 (64.1908) acc5: 100.0000 (100.0000) acc1_neg: 40.625 acc5_neg: 100.000 time: 0.1994 data: 0.1083 max mem: 3531 +Test: [210/333] eta: 0:00:28 acc1_pos: 64.145 acc5_pos: 100.000 loss: 0.8857 (0.6647) acc1: 41.9271 (63.2652) acc5: 100.0000 (100.0000) acc1_neg: 64.323 acc5_neg: 100.000 time: 0.1908 data: 0.0990 max mem: 3532 +Test: [220/333] eta: 0:00:25 acc1_pos: 64.145 acc5_pos: 100.000 loss: 0.8604 (0.6720) acc1: 43.4896 (62.5236) acc5: 100.0000 (100.0000) acc1_neg: 46.094 acc5_neg: 100.000 time: 0.1769 data: 0.0835 max mem: 3532 +Test: [230/333] eta: 0:00:23 acc1_pos: 64.145 acc5_pos: 100.000 loss: 0.7629 (0.6753) acc1: 46.3542 (61.9499) acc5: 100.0000 (100.0000) acc1_neg: 46.354 acc5_neg: 100.000 time: 0.1801 data: 0.0854 max mem: 3532 +Test: [240/333] eta: 0:00:20 acc1_pos: 64.145 acc5_pos: 100.000 loss: 0.7658 (0.6848) acc1: 41.9271 (60.8975) acc5: 100.0000 (100.0000) acc1_neg: 33.594 acc5_neg: 100.000 time: 0.1848 data: 0.0913 max mem: 3532 +Test: [250/333] eta: 0:00:18 acc1_pos: 64.145 acc5_pos: 100.000 loss: 0.8987 (0.6934) acc1: 35.9375 (59.9975) acc5: 100.0000 (100.0000) acc1_neg: 39.062 acc5_neg: 100.000 time: 0.1832 data: 0.0894 max mem: 3532 +Test: [260/333] eta: 0:00:16 acc1_pos: 64.145 acc5_pos: 100.000 loss: 0.8849 (0.6994) acc1: 41.4062 (59.4199) acc5: 100.0000 (100.0000) acc1_neg: 40.885 acc5_neg: 100.000 time: 0.1776 data: 0.0825 max mem: 3532 +Test: [270/333] eta: 0:00:13 acc1_pos: 64.145 acc5_pos: 100.000 loss: 0.8407 (0.7045) acc1: 43.7500 (58.8676) acc5: 100.0000 (100.0000) acc1_neg: 47.396 acc5_neg: 100.000 time: 0.1670 data: 0.0737 max mem: 3532 +Test: [280/333] eta: 0:00:11 acc1_pos: 64.145 acc5_pos: 100.000 loss: 0.8407 (0.7096) acc1: 43.4896 (58.2861) acc5: 100.0000 (100.0000) acc1_neg: 42.708 acc5_neg: 100.000 time: 0.1724 data: 0.0798 max mem: 3532 +Test: [290/333] eta: 0:00:09 acc1_pos: 64.145 acc5_pos: 100.000 loss: 0.8895 (0.7167) acc1: 40.1042 (57.5959) acc5: 100.0000 (100.0000) acc1_neg: 27.604 acc5_neg: 100.000 time: 0.1662 data: 0.0744 max mem: 3532 +Test: [300/333] eta: 0:00:07 acc1_pos: 64.145 acc5_pos: 100.000 loss: 0.9467 (0.7262) acc1: 35.4167 (56.7821) acc5: 100.0000 (100.0000) acc1_neg: 35.417 acc5_neg: 100.000 time: 0.1426 data: 0.0523 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 64.145 acc5_pos: 100.000 loss: 0.9227 (0.7294) acc1: 36.1979 (56.4066) acc5: 100.0000 (100.0000) acc1_neg: 48.698 acc5_neg: 100.000 time: 0.1300 data: 0.0399 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 64.145 acc5_pos: 100.000 loss: 0.8190 (0.7323) acc1: 44.2708 (56.0383) acc5: 100.0000 (100.0000) acc1_neg: 46.615 acc5_neg: 100.000 time: 0.1280 data: 0.0380 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 64.145 acc5_pos: 100.000 loss: 0.8461 (0.7362) acc1: 42.9688 (55.6505) acc5: 100.0000 (100.0000) acc1_neg: 42.708 acc5_neg: 100.000 time: 0.1177 data: 0.0281 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 64.145 acc5_pos: 100.000 loss: 0.8484 (0.7368) acc1: 43.2292 (55.5838) acc5: 100.0000 (100.0000) acc1_neg: 45.070 acc5_neg: 100.000 time: 0.1207 data: 0.0281 max mem: 3532 +Test: Total time: 0:01:08 (0.2044 s / it) +* Overall Acc@1 55.584 Acc@5 100.000 + target == real Acc@1 71.321 Acc@5 100.000 + target == fake Acc@1 41.575 Acc@5 100.000 + loss 0.737 + PR-AUC 0.6059 ROC-AUC 0.6092 EER 42.32% (thr=0.3965) +Accuracy of the network on the 127843 test images: 55.6% +[rank0]:[W726 05:31:00.779661556 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:31:20.291000 1193313 torch/distributed/run.py:766] +W0726 05:31:20.291000 1193313 torch/distributed/run.py:766] ***************************************** +W0726 05:31:20.291000 1193313 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:31:20.291000 1193313 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:31:49.522254581 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:31:49.622729166 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:31:49.830755928 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:31:49.184769755 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:31:49.221752729 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:31:49.241138798 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:31:49.263212621 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:31:49.264223338 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n8_seed8_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed8_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:21:03 acc1_pos: 36.979 acc5_pos: 100.000 loss: 1.0171 (1.0171) acc1: 36.9792 (36.9792) acc5: 100.0000 (100.0000) time: 5.3773 data: 3.6618 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/235] eta: 0:02:08 acc1_pos: 35.938 acc5_pos: 100.000 loss: 1.0546 (1.0510) acc1: 36.1979 (36.0559) acc5: 100.0000 (100.0000) time: 0.5725 data: 0.3333 max mem: 3531 +Test: [ 20/235] eta: 0:01:15 acc1_pos: 35.156 acc5_pos: 100.000 loss: 1.0546 (1.0544) acc1: 36.1979 (36.1483) acc5: 100.0000 (100.0000) time: 0.1009 data: 0.0088 max mem: 3531 +Test: [ 30/235] eta: 0:01:01 acc1_pos: 36.719 acc5_pos: 100.000 loss: 1.0483 (1.0508) acc1: 36.7188 (36.3155) acc5: 100.0000 (100.0000) time: 0.1524 data: 0.0591 max mem: 3531 +Test: [ 40/235] eta: 0:01:00 acc1_pos: 37.240 acc5_pos: 100.000 loss: 1.0488 (1.0516) acc1: 35.6771 (36.1535) acc5: 100.0000 (100.0000) time: 0.2699 data: 0.1757 max mem: 3531 +Test: [ 50/235] eta: 0:00:56 acc1_pos: 34.115 acc5_pos: 100.000 loss: 1.0623 (1.0571) acc1: 35.1562 (36.0601) acc5: 100.0000 (100.0000) time: 0.3183 data: 0.2248 max mem: 3531 +Test: [ 60/235] eta: 0:00:47 acc1_pos: 34.115 acc5_pos: 100.000 loss: 1.0718 (1.0595) acc1: 34.8958 (35.8564) acc5: 100.0000 (100.0000) time: 0.1949 data: 0.0999 max mem: 3531 +Test: [ 70/235] eta: 0:00:41 acc1_pos: 46.154 acc5_pos: 100.000 loss: 1.0327 (0.9692) acc1: 36.7188 (42.3452) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.1100 data: 0.0113 max mem: 3531 +Test: [ 80/235] eta: 0:00:37 acc1_pos: 46.154 acc5_pos: 100.000 loss: 0.3219 (0.8889) acc1: 88.5417 (48.1835) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.1418 data: 0.0469 max mem: 3531 +Test: [ 90/235] eta: 0:00:33 acc1_pos: 46.154 acc5_pos: 100.000 loss: 0.3219 (0.8273) acc1: 88.5417 (52.5698) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1616 data: 0.0700 max mem: 3531 +Test: [100/235] eta: 0:00:30 acc1_pos: 46.154 acc5_pos: 100.000 loss: 0.3276 (0.7772) acc1: 88.0208 (56.1752) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1830 data: 0.0898 max mem: 3531 +Test: [110/235] eta: 0:00:28 acc1_pos: 46.154 acc5_pos: 100.000 loss: 0.3184 (0.7356) acc1: 89.3229 (59.1052) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1947 data: 0.1011 max mem: 3531 +Test: [120/235] eta: 0:00:25 acc1_pos: 46.154 acc5_pos: 100.000 loss: 0.3313 (0.7032) acc1: 88.5417 (61.5251) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1801 data: 0.0874 max mem: 3531 +Test: [130/235] eta: 0:00:22 acc1_pos: 46.154 acc5_pos: 100.000 loss: 0.3491 (0.6767) acc1: 87.5000 (63.4741) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1743 data: 0.0826 max mem: 3531 +Test: [140/235] eta: 0:00:20 acc1_pos: 46.154 acc5_pos: 100.000 loss: 0.3545 (0.6541) acc1: 87.2396 (65.1577) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.1905 data: 0.0948 max mem: 3531 +Test: [150/235] eta: 0:00:18 acc1_pos: 46.154 acc5_pos: 100.000 loss: 0.3509 (0.6341) acc1: 87.2396 (66.6408) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1888 data: 0.0899 max mem: 3531 +Test: [160/235] eta: 0:00:15 acc1_pos: 46.154 acc5_pos: 100.000 loss: 0.3523 (0.6167) acc1: 87.7604 (67.9558) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1871 data: 0.0908 max mem: 3531 +Test: [170/235] eta: 0:00:13 acc1_pos: 46.154 acc5_pos: 100.000 loss: 0.3296 (0.5950) acc1: 89.8438 (69.4658) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2241 data: 0.1279 max mem: 3531 +Test: [180/235] eta: 0:00:11 acc1_pos: 46.154 acc5_pos: 100.000 loss: 0.2311 (0.5749) acc1: 94.5312 (70.8607) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2141 data: 0.1190 max mem: 3531 +Test: [190/235] eta: 0:00:09 acc1_pos: 46.154 acc5_pos: 100.000 loss: 0.2308 (0.5569) acc1: 95.0521 (72.1313) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1979 data: 0.1054 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 46.154 acc5_pos: 100.000 loss: 0.2304 (0.5405) acc1: 94.7917 (73.2600) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1996 data: 0.1066 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 46.154 acc5_pos: 100.000 loss: 0.2322 (0.5259) acc1: 94.5312 (74.2940) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1840 data: 0.0919 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 46.154 acc5_pos: 100.000 loss: 0.2351 (0.5128) acc1: 94.2708 (75.1944) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1806 data: 0.0893 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 46.154 acc5_pos: 100.000 loss: 0.2322 (0.5005) acc1: 94.2708 (76.0518) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1616 data: 0.0709 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 46.154 acc5_pos: 100.000 loss: 0.2303 (0.4958) acc1: 94.2708 (76.2870) acc5: 100.0000 (100.0000) acc1_neg: 95.652 acc5_neg: 100.000 time: 0.1493 data: 0.0622 max mem: 3532 +Test: Total time: 0:00:48 (0.2058 s / it) +* Overall Acc@1 76.287 Acc@5 100.000 + target == real Acc@1 35.815 Acc@5 100.000 + target == fake Acc@1 90.925 Acc@5 100.000 + loss 0.496 + PR-AUC 0.8804 ROC-AUC 0.7548 EER 30.83% (thr=0.7261) +Accuracy of the network on the 89879 test images: 76.3% +[rank0]:[W726 05:32:51.328150494 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:33:11.790000 1193862 torch/distributed/run.py:766] +W0726 05:33:11.790000 1193862 torch/distributed/run.py:766] ***************************************** +W0726 05:33:11.790000 1193862 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:33:11.790000 1193862 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:33:42.235679164 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:33:42.379133600 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W726 05:33:42.386088557 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:33:43.694796420 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:33:43.750274714 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:33:43.767715825 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:33:43.892426600 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:33:43.895191362 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n8_seed8_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed8_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:31:14 acc1_pos: 51.562 acc5_pos: 100.000 loss: 0.7172 (0.7172) acc1: 51.5625 (51.5625) acc5: 100.0000 (100.0000) time: 6.9668 data: 5.1985 max mem: 3531 +Test: [ 10/269] eta: 0:03:06 acc1_pos: 74.740 acc5_pos: 100.000 loss: 0.5808 (0.6081) acc1: 69.7917 (68.6553) acc5: 100.0000 (100.0000) time: 0.7201 data: 0.4730 max mem: 3531 +Test: [ 20/269] eta: 0:01:49 acc1_pos: 77.083 acc5_pos: 100.000 loss: 0.5733 (0.6040) acc1: 73.9583 (69.4692) acc5: 100.0000 (100.0000) time: 0.1126 data: 0.0136 max mem: 3531 +Test: [ 30/269] eta: 0:01:31 acc1_pos: 82.292 acc5_pos: 100.000 loss: 0.5464 (0.5793) acc1: 76.8229 (72.4126) acc5: 100.0000 (100.0000) time: 0.2005 data: 0.1013 max mem: 3531 +Test: [ 40/269] eta: 0:01:18 acc1_pos: 79.310 acc5_pos: 100.000 loss: 0.6100 (0.5962) acc1: 70.0521 (70.6364) acc5: 100.0000 (100.0000) acc1_neg: 49.740 acc5_neg: 100.000 time: 0.2399 data: 0.1467 max mem: 3531 +Test: [ 50/269] eta: 0:01:09 acc1_pos: 79.310 acc5_pos: 100.000 loss: 0.7006 (0.6651) acc1: 49.7396 (61.7851) acc5: 100.0000 (100.0000) acc1_neg: 12.760 acc5_neg: 100.000 time: 0.2129 data: 0.1220 max mem: 3531 +Test: [ 60/269] eta: 0:01:02 acc1_pos: 79.310 acc5_pos: 100.000 loss: 0.9636 (0.7178) acc1: 20.5729 (55.4602) acc5: 100.0000 (100.0000) acc1_neg: 36.719 acc5_neg: 100.000 time: 0.2149 data: 0.1243 max mem: 3531 +Test: [ 70/269] eta: 0:00:56 acc1_pos: 79.310 acc5_pos: 100.000 loss: 0.8076 (0.7174) acc1: 36.9792 (54.9626) acc5: 100.0000 (100.0000) acc1_neg: 54.427 acc5_neg: 100.000 time: 0.1963 data: 0.1046 max mem: 3531 +Test: [ 80/269] eta: 0:00:50 acc1_pos: 79.310 acc5_pos: 100.000 loss: 0.6829 (0.7115) acc1: 54.4271 (55.2051) acc5: 100.0000 (100.0000) acc1_neg: 61.198 acc5_neg: 100.000 time: 0.1724 data: 0.0780 max mem: 3531 +Test: [ 90/269] eta: 0:00:46 acc1_pos: 79.310 acc5_pos: 100.000 loss: 0.7422 (0.7261) acc1: 45.3125 (52.9791) acc5: 100.0000 (100.0000) acc1_neg: 45.573 acc5_neg: 100.000 time: 0.1778 data: 0.0822 max mem: 3531 +Test: [100/269] eta: 0:00:42 acc1_pos: 79.310 acc5_pos: 100.000 loss: 0.8091 (0.7322) acc1: 36.9792 (51.7920) acc5: 100.0000 (100.0000) acc1_neg: 22.917 acc5_neg: 100.000 time: 0.1874 data: 0.0910 max mem: 3531 +Test: [110/269] eta: 0:00:39 acc1_pos: 79.310 acc5_pos: 100.000 loss: 0.8048 (0.7427) acc1: 37.7604 (50.3777) acc5: 100.0000 (100.0000) acc1_neg: 37.760 acc5_neg: 100.000 time: 0.1820 data: 0.0861 max mem: 3531 +Test: [120/269] eta: 0:00:36 acc1_pos: 79.310 acc5_pos: 100.000 loss: 0.8048 (0.7445) acc1: 37.7604 (49.8687) acc5: 100.0000 (100.0000) acc1_neg: 31.510 acc5_neg: 100.000 time: 0.1899 data: 0.0951 max mem: 3531 +Test: [130/269] eta: 0:00:33 acc1_pos: 79.310 acc5_pos: 100.000 loss: 0.7703 (0.7467) acc1: 42.9688 (49.5328) acc5: 100.0000 (100.0000) acc1_neg: 53.646 acc5_neg: 100.000 time: 0.2009 data: 0.1079 max mem: 3531 +Test: [140/269] eta: 0:00:30 acc1_pos: 79.310 acc5_pos: 100.000 loss: 0.7703 (0.7473) acc1: 42.9688 (49.4644) acc5: 100.0000 (100.0000) acc1_neg: 38.542 acc5_neg: 100.000 time: 0.1999 data: 0.1072 max mem: 3531 +Test: [150/269] eta: 0:00:27 acc1_pos: 79.310 acc5_pos: 100.000 loss: 0.7408 (0.7493) acc1: 42.1875 (49.1756) acc5: 100.0000 (100.0000) acc1_neg: 52.083 acc5_neg: 100.000 time: 0.1738 data: 0.0790 max mem: 3531 +Test: [160/269] eta: 0:00:24 acc1_pos: 79.310 acc5_pos: 100.000 loss: 0.7226 (0.7466) acc1: 50.0000 (49.5212) acc5: 100.0000 (100.0000) acc1_neg: 66.927 acc5_neg: 100.000 time: 0.1662 data: 0.0732 max mem: 3531 +Test: [170/269] eta: 0:00:22 acc1_pos: 79.310 acc5_pos: 100.000 loss: 0.7139 (0.7432) acc1: 48.1771 (49.7853) acc5: 100.0000 (100.0000) acc1_neg: 26.562 acc5_neg: 100.000 time: 0.1808 data: 0.0901 max mem: 3531 +Test: [180/269] eta: 0:00:19 acc1_pos: 79.310 acc5_pos: 100.000 loss: 0.7170 (0.7436) acc1: 45.8333 (49.5094) acc5: 100.0000 (100.0000) acc1_neg: 29.688 acc5_neg: 100.000 time: 0.2016 data: 0.1114 max mem: 3531 +Test: [190/269] eta: 0:00:17 acc1_pos: 79.310 acc5_pos: 100.000 loss: 0.7355 (0.7461) acc1: 42.1875 (48.9120) acc5: 100.0000 (100.0000) acc1_neg: 57.292 acc5_neg: 100.000 time: 0.2203 data: 0.1293 max mem: 3531 +Test: [200/269] eta: 0:00:15 acc1_pos: 79.310 acc5_pos: 100.000 loss: 0.7406 (0.7478) acc1: 41.9271 (48.6759) acc5: 100.0000 (100.0000) acc1_neg: 23.438 acc5_neg: 100.000 time: 0.2280 data: 0.1342 max mem: 3531 +Test: [210/269] eta: 0:00:13 acc1_pos: 79.310 acc5_pos: 100.000 loss: 0.7923 (0.7516) acc1: 42.4479 (48.1450) acc5: 100.0000 (100.0000) acc1_neg: 42.448 acc5_neg: 100.000 time: 0.2349 data: 0.1422 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 79.310 acc5_pos: 100.000 loss: 0.8210 (0.7528) acc1: 36.9792 (48.0510) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.1992 data: 0.1087 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 79.310 acc5_pos: 100.000 loss: 0.7546 (0.7524) acc1: 44.7917 (48.1455) acc5: 100.0000 (100.0000) acc1_neg: 26.823 acc5_neg: 100.000 time: 0.1787 data: 0.0862 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 79.310 acc5_pos: 100.000 loss: 0.8352 (0.7591) acc1: 28.6458 (47.2510) acc5: 100.0000 (100.0000) acc1_neg: 44.271 acc5_neg: 100.000 time: 0.1780 data: 0.0857 max mem: 3532 +Test: [250/269] eta: 0:00:04 acc1_pos: 79.310 acc5_pos: 100.000 loss: 0.8973 (0.7636) acc1: 26.0417 (46.7962) acc5: 100.0000 (100.0000) acc1_neg: 11.979 acc5_neg: 100.000 time: 0.1557 data: 0.0651 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 79.310 acc5_pos: 100.000 loss: 0.9564 (0.7737) acc1: 19.5312 (45.7735) acc5: 100.0000 (100.0000) acc1_neg: 35.677 acc5_neg: 100.000 time: 0.1380 data: 0.0474 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 79.310 acc5_pos: 100.000 loss: 0.9481 (0.7765) acc1: 20.5729 (45.3806) acc5: 100.0000 (100.0000) acc1_neg: 46.457 acc5_neg: 100.000 time: 0.1333 data: 0.0393 max mem: 3532 +Test: Total time: 0:00:56 (0.2111 s / it) +* Overall Acc@1 45.381 Acc@5 100.000 + target == real Acc@1 72.323 Acc@5 100.000 + target == fake Acc@1 41.704 Acc@5 100.000 + loss 0.776 + PR-AUC 0.9126 ROC-AUC 0.6066 EER 42.08% (thr=0.4292) +Accuracy of the network on the 103293 test images: 45.4% +[rank0]:[W726 05:34:49.847333207 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T05:34:55Z] eval_done tag=fixed500k_real_n8_seed8_resampled +[2026-07-26T05:34:56Z] eval_start tag=fixed500k_fake_n16_seed8_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed8_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T05:34:56Z +hostname=n116-041-153 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed8_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n16_seed8_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 05:35:09.410000 1194439 torch/distributed/run.py:766] +W0726 05:35:09.410000 1194439 torch/distributed/run.py:766] ***************************************** +W0726 05:35:09.410000 1194439 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:35:09.410000 1194439 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:35:39.622727826 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:35:39.875751777 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:35:39.044008604 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:35:39.076867246 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:35:39.209617746 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:35:39.247981595 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W726 05:35:39.253473996 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:35:39.335813590 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n16_seed8_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed8_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:26:00 acc1_pos: 42.969 acc5_pos: 100.000 loss: 0.7797 (0.7797) acc1: 42.9688 (42.9688) acc5: 100.0000 (100.0000) time: 6.6676 data: 4.6518 max mem: 3531 +Test: [ 10/234] eta: 0:02:34 acc1_pos: 38.542 acc5_pos: 100.000 loss: 0.6656 (0.6558) acc1: 55.4688 (60.7955) acc5: 100.0000 (100.0000) time: 0.6880 data: 0.4233 max mem: 3531 +Test: [ 20/234] eta: 0:01:26 acc1_pos: 97.917 acc5_pos: 100.000 loss: 0.6656 (0.6690) acc1: 51.5625 (57.4033) acc5: 100.0000 (100.0000) time: 0.0906 data: 0.0005 max mem: 3531 +Test: [ 30/234] eta: 0:01:16 acc1_pos: 87.760 acc5_pos: 100.000 loss: 0.1579 (0.5073) acc1: 97.6562 (70.2201) acc5: 100.0000 (100.0000) time: 0.2006 data: 0.1094 max mem: 3531 +Test: [ 40/234] eta: 0:01:09 acc1_pos: 98.177 acc5_pos: 100.000 loss: 0.1457 (0.4489) acc1: 96.8750 (75.6987) acc5: 100.0000 (100.0000) time: 0.3099 data: 0.2172 max mem: 3531 +Test: [ 50/234] eta: 0:00:59 acc1_pos: 98.177 acc5_pos: 100.000 loss: 0.1193 (0.3840) acc1: 98.1771 (80.1420) acc5: 100.0000 (100.0000) time: 0.2415 data: 0.1483 max mem: 3531 +Test: [ 60/234] eta: 0:00:50 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.1087 (0.3378) acc1: 98.9583 (83.3846) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1449 data: 0.0507 max mem: 3531 +Test: [ 70/234] eta: 0:00:44 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.1032 (0.3707) acc1: 99.7396 (81.6791) acc5: 100.0000 (100.0000) acc1_neg: 57.552 acc5_neg: 100.000 time: 0.1333 data: 0.0365 max mem: 3531 +Test: [ 80/234] eta: 0:00:39 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.7654 (0.3954) acc1: 59.6354 (80.5459) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1549 data: 0.0588 max mem: 3531 +Test: [ 90/234] eta: 0:00:35 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.2734 (0.3754) acc1: 89.5833 (81.9254) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.1668 data: 0.0726 max mem: 3531 +Test: [100/234] eta: 0:00:32 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.2989 (0.3973) acc1: 86.7188 (80.7395) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1805 data: 0.0862 max mem: 3531 +Test: [110/234] eta: 0:00:29 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.5420 (0.4034) acc1: 72.3958 (80.4781) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.1851 data: 0.0917 max mem: 3531 +Test: [120/234] eta: 0:00:26 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.4674 (0.4180) acc1: 76.3021 (79.8403) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1799 data: 0.0886 max mem: 3531 +Test: [130/234] eta: 0:00:23 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.4046 (0.4208) acc1: 82.5521 (79.9897) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.1936 data: 0.1030 max mem: 3531 +Test: [140/234] eta: 0:00:21 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.6155 (0.4585) acc1: 68.7500 (77.9311) acc5: 100.0000 (100.0000) acc1_neg: 30.469 acc5_neg: 100.000 time: 0.1900 data: 0.0993 max mem: 3531 +Test: [150/234] eta: 0:00:18 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.6284 (0.4751) acc1: 66.1458 (76.9402) acc5: 100.0000 (100.0000) acc1_neg: 66.146 acc5_neg: 100.000 time: 0.1803 data: 0.0898 max mem: 3531 +Test: [160/234] eta: 0:00:16 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.5439 (0.4724) acc1: 72.6562 (77.2451) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.1863 data: 0.0963 max mem: 3531 +Test: [170/234] eta: 0:00:14 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.5070 (0.4742) acc1: 76.8229 (77.2585) acc5: 100.0000 (100.0000) acc1_neg: 76.302 acc5_neg: 100.000 time: 0.2009 data: 0.1105 max mem: 3531 +Test: [180/234] eta: 0:00:11 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.5127 (0.4854) acc1: 75.2604 (76.6704) acc5: 100.0000 (100.0000) acc1_neg: 38.281 acc5_neg: 100.000 time: 0.2206 data: 0.1272 max mem: 3531 +Test: [190/234] eta: 0:00:09 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.6422 (0.4952) acc1: 65.1042 (76.0730) acc5: 100.0000 (100.0000) acc1_neg: 65.104 acc5_neg: 100.000 time: 0.2130 data: 0.1179 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.6168 (0.4857) acc1: 67.9688 (76.6610) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1831 data: 0.0902 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.1125 (0.4693) acc1: 99.7396 (77.6807) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1648 data: 0.0723 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.1880 (0.4596) acc1: 95.5729 (78.3289) acc5: 100.0000 (100.0000) acc1_neg: 66.927 acc5_neg: 100.000 time: 0.1644 data: 0.0696 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.6861 (0.5121) acc1: 16.6667 (75.5547) acc5: 100.0000 (100.0000) acc1_neg: 14.583 acc5_neg: 100.000 time: 0.1658 data: 0.0726 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 98.621 acc5_pos: 100.000 loss: 1.6291 (0.5266) acc1: 15.9533 (74.8587) acc5: 100.0000 (100.0000) acc1_neg: 15.953 acc5_neg: 100.000 time: 0.1651 data: 0.0635 max mem: 3532 +Test: Total time: 0:00:49 (0.2096 s / it) +* Overall Acc@1 74.859 Acc@5 100.000 + target == real Acc@1 80.278 Acc@5 100.000 + target == fake Acc@1 73.331 Acc@5 100.000 + loss 0.527 + PR-AUC 0.9588 ROC-AUC 0.8612 EER 23.54% (thr=0.4324) +Accuracy of the network on the 89729 test images: 74.9% +[rank0]:[W726 05:36:40.806886138 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:37:00.446000 1195000 torch/distributed/run.py:766] +W0726 05:37:00.446000 1195000 torch/distributed/run.py:766] ***************************************** +W0726 05:37:00.446000 1195000 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:37:00.446000 1195000 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:37:26.866139623 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:37:26.077409629 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:37:26.145142442 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:37:26.165016279 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:37:26.258606949 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +| distributed init (rank 6): env:// +[rank0]:[W726 05:37:26.413281242 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:37:26.414729692 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank6]:[W726 05:37:26.414738106 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n16_seed8_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed8_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:45:57 acc1_pos: 41.927 acc5_pos: 100.000 loss: 0.9038 (0.9038) acc1: 41.9271 (41.9271) acc5: 100.0000 (100.0000) time: 7.1879 data: 5.7454 max mem: 3531 +Test: [ 10/2387] eta: 0:29:07 acc1_pos: 18.490 acc5_pos: 100.000 loss: 0.9349 (0.9695) acc1: 39.5833 (33.8778) acc5: 100.0000 (100.0000) time: 0.7352 data: 0.5226 max mem: 3531 +Test: [ 20/2387] eta: 0:17:02 acc1_pos: 31.771 acc5_pos: 100.000 loss: 0.9387 (0.9675) acc1: 31.7708 (34.7222) acc5: 100.0000 (100.0000) time: 0.0943 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:14:39 acc1_pos: 15.625 acc5_pos: 100.000 loss: 1.0281 (0.9967) acc1: 27.3438 (32.6193) acc5: 100.0000 (100.0000) time: 0.1743 data: 0.0794 max mem: 3531 +Test: [ 40/2387] eta: 0:13:21 acc1_pos: 26.042 acc5_pos: 100.000 loss: 1.0857 (1.0306) acc1: 22.6562 (29.1857) acc5: 100.0000 (100.0000) time: 0.2461 data: 0.1527 max mem: 3531 +Test: [ 50/2387] eta: 0:12:25 acc1_pos: 35.938 acc5_pos: 100.000 loss: 1.0716 (1.0298) acc1: 22.3958 (28.9726) acc5: 100.0000 (100.0000) time: 0.2349 data: 0.1387 max mem: 3531 +Test: [ 60/2387] eta: 0:11:52 acc1_pos: 50.260 acc5_pos: 100.000 loss: 0.9770 (1.0249) acc1: 28.9062 (29.0514) acc5: 100.0000 (100.0000) time: 0.2341 data: 0.1405 max mem: 3531 +Test: [ 70/2387] eta: 0:11:40 acc1_pos: 32.552 acc5_pos: 100.000 loss: 1.0573 (1.0403) acc1: 27.6042 (28.0957) acc5: 100.0000 (100.0000) time: 0.2597 data: 0.1683 max mem: 3531 +Test: [ 80/2387] eta: 0:11:37 acc1_pos: 30.990 acc5_pos: 100.000 loss: 1.1146 (1.0439) acc1: 25.0000 (27.4949) acc5: 100.0000 (100.0000) time: 0.2899 data: 0.1971 max mem: 3531 +Test: [ 90/2387] eta: 0:11:18 acc1_pos: 36.719 acc5_pos: 100.000 loss: 1.0178 (1.0378) acc1: 27.6042 (27.8732) acc5: 100.0000 (100.0000) time: 0.2717 data: 0.1767 max mem: 3531 +Test: [ 100/2387] eta: 0:11:02 acc1_pos: 32.552 acc5_pos: 100.000 loss: 1.0007 (1.0433) acc1: 22.9167 (27.1091) acc5: 100.0000 (100.0000) time: 0.2394 data: 0.1455 max mem: 3531 +Test: [ 110/2387] eta: 0:10:57 acc1_pos: 31.510 acc5_pos: 100.000 loss: 1.0007 (1.0373) acc1: 22.9167 (27.3836) acc5: 100.0000 (100.0000) time: 0.2581 data: 0.1660 max mem: 3531 +Test: [ 120/2387] eta: 0:10:41 acc1_pos: 28.906 acc5_pos: 100.000 loss: 1.0007 (1.0355) acc1: 29.1667 (27.4664) acc5: 100.0000 (100.0000) time: 0.2499 data: 0.1558 max mem: 3531 +Test: [ 130/2387] eta: 0:10:33 acc1_pos: 30.469 acc5_pos: 100.000 loss: 1.0118 (1.0350) acc1: 28.9062 (27.3855) acc5: 100.0000 (100.0000) time: 0.2349 data: 0.1403 max mem: 3531 +Test: [ 140/2387] eta: 0:10:22 acc1_pos: 24.740 acc5_pos: 100.000 loss: 1.0118 (1.0344) acc1: 26.3021 (27.3659) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1454 max mem: 3531 +Test: [ 150/2387] eta: 0:10:17 acc1_pos: 6.771 acc5_pos: 100.000 loss: 0.9836 (1.0318) acc1: 26.3021 (27.4058) acc5: 100.0000 (100.0000) time: 0.2461 data: 0.1522 max mem: 3531 +Test: [ 160/2387] eta: 0:09:57 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.9760 (1.0210) acc1: 32.0312 (28.3903) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.2052 data: 0.1131 max mem: 3531 +Test: [ 170/2387] eta: 0:09:40 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.3177 (0.9788) acc1: 88.8021 (32.1104) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1537 data: 0.0600 max mem: 3531 +Test: [ 180/2387] eta: 0:09:33 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2986 (0.9411) acc1: 91.6667 (35.4426) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1946 data: 0.0972 max mem: 3531 +Test: [ 190/2387] eta: 0:09:23 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2960 (0.9073) acc1: 91.6667 (38.4040) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.2111 data: 0.1143 max mem: 3531 +Test: [ 200/2387] eta: 0:09:14 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2948 (0.8768) acc1: 92.1875 (41.1070) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1954 data: 0.1023 max mem: 3531 +Test: [ 210/2387] eta: 0:09:07 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2991 (0.8498) acc1: 92.1875 (43.5044) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.2037 data: 0.1094 max mem: 3532 +Test: [ 220/2387] eta: 0:08:57 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.3032 (0.8250) acc1: 91.4062 (45.6908) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1944 data: 0.0946 max mem: 3532 +Test: [ 230/2387] eta: 0:08:48 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.3003 (0.8024) acc1: 91.1458 (47.6732) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1774 data: 0.0806 max mem: 3532 +Test: [ 240/2387] eta: 0:08:40 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2825 (0.7798) acc1: 93.2292 (49.6748) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1799 data: 0.0885 max mem: 3532 +Test: [ 250/2387] eta: 0:08:34 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2385 (0.7545) acc1: 97.1354 (51.6559) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1904 data: 0.0993 max mem: 3532 +Test: [ 260/2387] eta: 0:08:29 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1259 (0.7302) acc1: 99.7396 (53.5032) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2025 data: 0.1111 max mem: 3532 +Test: [ 270/2387] eta: 0:08:27 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1199 (0.7077) acc1: 100.0000 (55.2141) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2269 data: 0.1358 max mem: 3532 +Test: [ 280/2387] eta: 0:08:23 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1199 (0.6870) acc1: 100.0000 (56.7940) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2334 data: 0.1412 max mem: 3532 +Test: [ 290/2387] eta: 0:08:21 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1226 (0.6677) acc1: 100.0000 (58.2752) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2321 data: 0.1384 max mem: 3532 +Test: [ 300/2387] eta: 0:08:14 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1226 (0.6496) acc1: 100.0000 (59.6596) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2053 data: 0.1129 max mem: 3532 +Test: [ 310/2387] eta: 0:08:12 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1226 (0.6361) acc1: 100.0000 (60.8513) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2088 data: 0.1181 max mem: 3532 +Test: [ 320/2387] eta: 0:08:07 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.3595 (0.6276) acc1: 94.0104 (61.8802) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2206 data: 0.1282 max mem: 3532 +Test: [ 330/2387] eta: 0:08:03 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.3616 (0.6194) acc1: 94.2708 (62.8769) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1999 data: 0.1032 max mem: 3532 +Test: [ 340/2387] eta: 0:07:59 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.3570 (0.6116) acc1: 94.7917 (63.8151) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2091 data: 0.1137 max mem: 3532 +Test: [ 350/2387] eta: 0:07:57 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.3413 (0.5990) acc1: 96.6146 (64.8289) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2289 data: 0.1361 max mem: 3532 +Test: [ 360/2387] eta: 0:07:52 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2025 (0.5885) acc1: 97.9167 (65.7173) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2145 data: 0.1205 max mem: 3532 +Test: [ 370/2387] eta: 0:07:49 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2243 (0.5787) acc1: 96.6146 (66.5551) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1966 data: 0.1037 max mem: 3532 +Test: [ 380/2387] eta: 0:07:44 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2294 (0.5696) acc1: 96.3542 (67.3454) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2011 data: 0.1090 max mem: 3532 +Test: [ 390/2387] eta: 0:07:41 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2294 (0.5609) acc1: 96.3542 (68.0900) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2051 data: 0.1129 max mem: 3532 +Test: [ 400/2387] eta: 0:07:37 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2280 (0.5526) acc1: 96.3542 (68.8000) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2094 data: 0.1168 max mem: 3532 +Test: [ 410/2387] eta: 0:07:34 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2250 (0.5446) acc1: 96.6146 (69.4799) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2039 data: 0.1117 max mem: 3532 +Test: [ 420/2387] eta: 0:07:31 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2261 (0.5371) acc1: 96.6146 (70.1207) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2081 data: 0.1160 max mem: 3532 +Test: [ 430/2387] eta: 0:07:25 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2243 (0.5298) acc1: 96.8750 (70.7487) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1808 data: 0.0871 max mem: 3532 +Test: [ 440/2387] eta: 0:07:22 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2131 (0.5212) acc1: 97.6562 (71.3931) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1809 data: 0.0844 max mem: 3533 +Test: [ 450/2387] eta: 0:07:18 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1039 (0.5119) acc1: 100.0000 (72.0274) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2008 data: 0.1059 max mem: 3533 +Test: [ 460/2387] eta: 0:07:14 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1034 (0.5030) acc1: 100.0000 (72.6342) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1868 data: 0.0959 max mem: 3533 +Test: [ 470/2387] eta: 0:07:10 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1035 (0.4946) acc1: 100.0000 (73.2147) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1786 data: 0.0876 max mem: 3533 +Test: [ 480/2387] eta: 0:07:07 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1038 (0.4864) acc1: 100.0000 (73.7715) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1924 data: 0.0989 max mem: 3533 +Test: [ 490/2387] eta: 0:07:03 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1038 (0.4786) acc1: 100.0000 (74.3057) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1991 data: 0.1055 max mem: 3533 +Test: [ 500/2387] eta: 0:07:00 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1040 (0.4712) acc1: 100.0000 (74.8186) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1913 data: 0.0999 max mem: 3533 +Test: [ 510/2387] eta: 0:06:56 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1040 (0.4642) acc1: 100.0000 (75.3114) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1943 data: 0.1023 max mem: 3533 +Test: [ 520/2387] eta: 0:06:53 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1379 (0.4580) acc1: 100.0000 (75.7853) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1932 data: 0.1011 max mem: 3533 +Test: [ 530/2387] eta: 0:06:50 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1379 (0.4520) acc1: 100.0000 (76.2413) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1959 data: 0.1047 max mem: 3533 +Test: [ 540/2387] eta: 0:06:47 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1375 (0.4461) acc1: 100.0000 (76.6804) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1918 data: 0.0993 max mem: 3533 +Test: [ 550/2387] eta: 0:06:43 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1370 (0.4405) acc1: 100.0000 (77.1037) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1742 data: 0.0812 max mem: 3533 +Test: [ 560/2387] eta: 0:06:40 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1368 (0.4351) acc1: 100.0000 (77.5118) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1828 data: 0.0904 max mem: 3533 +Test: [ 570/2387] eta: 0:06:36 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1368 (0.4299) acc1: 100.0000 (77.9056) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1887 data: 0.0965 max mem: 3533 +Test: [ 580/2387] eta: 0:06:33 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1368 (0.4249) acc1: 100.0000 (78.2859) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1798 data: 0.0887 max mem: 3533 +Test: [ 590/2387] eta: 0:06:30 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1369 (0.4200) acc1: 100.0000 (78.6533) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1915 data: 0.0999 max mem: 3533 +Test: [ 600/2387] eta: 0:06:27 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1396 (0.4154) acc1: 100.0000 (79.0081) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1941 data: 0.1008 max mem: 3533 +Test: [ 610/2387] eta: 0:06:25 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1593 (0.4118) acc1: 99.4792 (79.3286) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1983 data: 0.1051 max mem: 3533 +Test: [ 620/2387] eta: 0:06:21 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1915 (0.4083) acc1: 98.6979 (79.6380) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1859 data: 0.0939 max mem: 3533 +Test: [ 630/2387] eta: 0:06:18 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1931 (0.4049) acc1: 98.4375 (79.9397) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1749 data: 0.0813 max mem: 3533 +Test: [ 640/2387] eta: 0:06:16 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1985 (0.4017) acc1: 98.4375 (80.2230) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1923 data: 0.0991 max mem: 3533 +Test: [ 650/2387] eta: 0:06:13 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2001 (0.3986) acc1: 97.9167 (80.4964) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1957 data: 0.1036 max mem: 3533 +Test: [ 660/2387] eta: 0:06:10 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1948 (0.3955) acc1: 98.4375 (80.7717) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1942 data: 0.1023 max mem: 3533 +Test: [ 670/2387] eta: 0:06:07 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1930 (0.3925) acc1: 98.6979 (81.0396) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1906 data: 0.0992 max mem: 3534 +Test: [ 680/2387] eta: 0:06:04 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1875 (0.3894) acc1: 98.9583 (81.3127) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1847 data: 0.0939 max mem: 3534 +Test: [ 690/2387] eta: 0:06:02 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1875 (0.3903) acc1: 98.9583 (81.2541) acc5: 100.0000 (100.0000) acc1_neg: 66.667 acc5_neg: 100.000 time: 0.1884 data: 0.0974 max mem: 3534 +Test: [ 700/2387] eta: 0:05:59 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.5413 (0.3925) acc1: 70.0521 (81.0903) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.1926 data: 0.1015 max mem: 3534 +Test: [ 710/2387] eta: 0:05:56 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.5474 (0.3947) acc1: 68.7500 (80.9182) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.1874 data: 0.0964 max mem: 3534 +Test: [ 720/2387] eta: 0:05:54 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.5473 (0.3967) acc1: 68.7500 (80.7707) acc5: 100.0000 (100.0000) acc1_neg: 70.833 acc5_neg: 100.000 time: 0.1912 data: 0.0994 max mem: 3534 +Test: [ 730/2387] eta: 0:05:51 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.5504 (0.3989) acc1: 69.0104 (80.6123) acc5: 100.0000 (100.0000) acc1_neg: 69.010 acc5_neg: 100.000 time: 0.1937 data: 0.1023 max mem: 3534 +Test: [ 740/2387] eta: 0:05:49 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.5513 (0.4010) acc1: 69.0104 (80.4670) acc5: 100.0000 (100.0000) acc1_neg: 64.583 acc5_neg: 100.000 time: 0.1913 data: 0.0982 max mem: 3534 +Test: [ 750/2387] eta: 0:05:46 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.5444 (0.4029) acc1: 69.5312 (80.3176) acc5: 100.0000 (100.0000) acc1_neg: 70.573 acc5_neg: 100.000 time: 0.1874 data: 0.0937 max mem: 3534 +Test: [ 760/2387] eta: 0:05:43 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.5375 (0.4037) acc1: 69.5312 (80.2515) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1798 data: 0.0880 max mem: 3534 +Test: [ 770/2387] eta: 0:05:40 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.3402 (0.4025) acc1: 84.8958 (80.3627) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1747 data: 0.0811 max mem: 3534 +Test: [ 780/2387] eta: 0:05:37 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2832 (0.4006) acc1: 91.9271 (80.5324) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1696 data: 0.0751 max mem: 3534 +Test: [ 790/2387] eta: 0:05:34 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2256 (0.3984) acc1: 95.3125 (80.7045) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1712 data: 0.0769 max mem: 3534 +Test: [ 800/2387] eta: 0:05:32 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2085 (0.3962) acc1: 96.0938 (80.8976) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1837 data: 0.0903 max mem: 3534 +Test: [ 810/2387] eta: 0:05:30 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2158 (0.3940) acc1: 96.8750 (81.0943) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1967 data: 0.1043 max mem: 3534 +Test: [ 820/2387] eta: 0:05:27 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2199 (0.3922) acc1: 96.8750 (81.2646) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1868 data: 0.0933 max mem: 3534 +Test: [ 830/2387] eta: 0:05:24 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2546 (0.3912) acc1: 93.2292 (81.3760) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1793 data: 0.0873 max mem: 3534 +Test: [ 840/2387] eta: 0:05:21 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.3266 (0.3909) acc1: 88.0208 (81.4407) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1717 data: 0.0807 max mem: 3534 +Test: [ 850/2387] eta: 0:05:19 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.3607 (0.3909) acc1: 86.1979 (81.4865) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.1673 data: 0.0755 max mem: 3534 +Test: [ 860/2387] eta: 0:05:16 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.3466 (0.3900) acc1: 88.2812 (81.5942) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1812 data: 0.0900 max mem: 3534 +Test: [ 870/2387] eta: 0:05:14 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.3167 (0.3894) acc1: 91.4062 (81.6808) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1916 data: 0.1009 max mem: 3534 +Test: [ 880/2387] eta: 0:05:11 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.3176 (0.3889) acc1: 90.1042 (81.7617) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1940 data: 0.1032 max mem: 3534 +Test: [ 890/2387] eta: 0:05:09 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2760 (0.3870) acc1: 93.4896 (81.9322) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1852 data: 0.0947 max mem: 3534 +Test: [ 900/2387] eta: 0:05:07 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2735 (0.3929) acc1: 91.6667 (81.5567) acc5: 100.0000 (100.0000) acc1_neg: 38.542 acc5_neg: 100.000 time: 0.1822 data: 0.0907 max mem: 3535 +Test: [ 910/2387] eta: 0:05:04 acc1_pos: 21.461 acc5_pos: 100.000 loss: 1.0628 (0.4003) acc1: 36.9792 (81.0579) acc5: 100.0000 (100.0000) acc1_neg: 38.281 acc5_neg: 100.000 time: 0.1707 data: 0.0792 max mem: 3535 +Test: [ 920/2387] eta: 0:05:01 acc1_pos: 21.461 acc5_pos: 100.000 loss: 1.0750 (0.4077) acc1: 35.4167 (80.5629) acc5: 100.0000 (100.0000) acc1_neg: 31.510 acc5_neg: 100.000 time: 0.1673 data: 0.0763 max mem: 3535 +Test: [ 930/2387] eta: 0:04:59 acc1_pos: 21.461 acc5_pos: 100.000 loss: 1.0750 (0.4148) acc1: 35.9375 (80.0995) acc5: 100.0000 (100.0000) acc1_neg: 34.635 acc5_neg: 100.000 time: 0.1761 data: 0.0844 max mem: 3535 +Test: [ 940/2387] eta: 0:04:56 acc1_pos: 21.461 acc5_pos: 100.000 loss: 1.0823 (0.4218) acc1: 35.6771 (79.6283) acc5: 100.0000 (100.0000) acc1_neg: 34.375 acc5_neg: 100.000 time: 0.1799 data: 0.0883 max mem: 3535 +Test: [ 950/2387] eta: 0:04:54 acc1_pos: 21.461 acc5_pos: 100.000 loss: 1.0823 (0.4287) acc1: 35.4167 (79.1658) acc5: 100.0000 (100.0000) acc1_neg: 36.458 acc5_neg: 100.000 time: 0.1736 data: 0.0826 max mem: 3535 +Test: [ 960/2387] eta: 0:04:51 acc1_pos: 21.461 acc5_pos: 100.000 loss: 1.0612 (0.4352) acc1: 36.4583 (78.7293) acc5: 100.0000 (100.0000) acc1_neg: 38.281 acc5_neg: 100.000 time: 0.1662 data: 0.0752 max mem: 3535 +Test: [ 970/2387] eta: 0:04:48 acc1_pos: 21.461 acc5_pos: 100.000 loss: 1.0612 (0.4419) acc1: 36.1979 (78.2894) acc5: 100.0000 (100.0000) acc1_neg: 34.896 acc5_neg: 100.000 time: 0.1687 data: 0.0771 max mem: 3535 +Test: [ 980/2387] eta: 0:04:46 acc1_pos: 21.461 acc5_pos: 100.000 loss: 1.0391 (0.4430) acc1: 36.7188 (78.2267) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.1586 data: 0.0675 max mem: 3535 +Test: [ 990/2387] eta: 0:04:43 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2424 (0.4403) acc1: 94.7917 (78.4046) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1655 data: 0.0747 max mem: 3535 +Test: [1000/2387] eta: 0:04:41 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1213 (0.4372) acc1: 99.7396 (78.6053) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1772 data: 0.0859 max mem: 3535 +Test: [1010/2387] eta: 0:04:38 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1260 (0.4343) acc1: 99.4792 (78.8048) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1668 data: 0.0752 max mem: 3535 +Test: [1020/2387] eta: 0:04:36 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1605 (0.4321) acc1: 98.9583 (78.9738) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1644 data: 0.0736 max mem: 3535 +Test: [1030/2387] eta: 0:04:33 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2057 (0.4301) acc1: 96.8750 (79.1414) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1743 data: 0.0840 max mem: 3535 +Test: [1040/2387] eta: 0:04:31 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1940 (0.4277) acc1: 97.9167 (79.3278) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1758 data: 0.0827 max mem: 3535 +Test: [1050/2387] eta: 0:04:29 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1745 (0.4254) acc1: 98.6979 (79.5096) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1781 data: 0.0839 max mem: 3535 +Test: [1060/2387] eta: 0:04:27 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1714 (0.4231) acc1: 98.9583 (79.6877) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1886 data: 0.0972 max mem: 3535 +Test: [1070/2387] eta: 0:04:24 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1787 (0.4211) acc1: 98.4375 (79.8480) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1813 data: 0.0906 max mem: 3535 +Test: [1080/2387] eta: 0:04:22 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2334 (0.4196) acc1: 96.3542 (79.9884) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1813 data: 0.0886 max mem: 3535 +Test: [1090/2387] eta: 0:04:20 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1950 (0.4174) acc1: 96.8750 (80.1556) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1958 data: 0.1036 max mem: 3535 +Test: [1100/2387] eta: 0:04:18 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1706 (0.4152) acc1: 99.2188 (80.3257) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1892 data: 0.0986 max mem: 3535 +Test: [1110/2387] eta: 0:04:16 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1746 (0.4134) acc1: 98.9583 (80.4580) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1926 data: 0.0990 max mem: 3535 +Test: [1120/2387] eta: 0:04:14 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1788 (0.4112) acc1: 98.6979 (80.6228) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1983 data: 0.1034 max mem: 3536 +Test: [1130/2387] eta: 0:04:11 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1651 (0.4091) acc1: 99.4792 (80.7904) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1803 data: 0.0881 max mem: 3536 +Test: [1140/2387] eta: 0:04:09 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1745 (0.4071) acc1: 99.4792 (80.9499) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1762 data: 0.0841 max mem: 3536 +Test: [1150/2387] eta: 0:04:07 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1918 (0.4057) acc1: 97.6562 (81.0670) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1815 data: 0.0885 max mem: 3536 +Test: [1160/2387] eta: 0:04:05 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2363 (0.4043) acc1: 92.7083 (81.1710) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.1792 data: 0.0861 max mem: 3536 +Test: [1170/2387] eta: 0:04:03 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2282 (0.4028) acc1: 93.7500 (81.2894) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1767 data: 0.0846 max mem: 3536 +Test: [1180/2387] eta: 0:04:00 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2398 (0.4016) acc1: 93.7500 (81.3916) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.1754 data: 0.0837 max mem: 3536 +Test: [1190/2387] eta: 0:03:58 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2236 (0.4000) acc1: 95.3125 (81.5196) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0832 max mem: 3536 +Test: [1200/2387] eta: 0:03:56 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1896 (0.3981) acc1: 97.3958 (81.6620) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1734 data: 0.0803 max mem: 3536 +Test: [1210/2387] eta: 0:03:54 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1738 (0.3962) acc1: 99.2188 (81.8029) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1770 data: 0.0835 max mem: 3536 +Test: [1220/2387] eta: 0:03:51 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1616 (0.3943) acc1: 98.9583 (81.9421) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1811 data: 0.0884 max mem: 3536 +Test: [1230/2387] eta: 0:03:49 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1576 (0.3923) acc1: 98.9583 (82.0801) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1715 data: 0.0791 max mem: 3536 +Test: [1240/2387] eta: 0:03:47 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1494 (0.3903) acc1: 99.4792 (82.2170) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1650 data: 0.0742 max mem: 3536 +Test: [1250/2387] eta: 0:03:45 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1388 (0.3884) acc1: 99.4792 (82.3552) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1862 data: 0.0947 max mem: 3536 +Test: [1260/2387] eta: 0:03:42 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1388 (0.3864) acc1: 99.4792 (82.4874) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1715 data: 0.0796 max mem: 3536 +Test: [1270/2387] eta: 0:03:40 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1336 (0.3844) acc1: 99.7396 (82.6248) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1612 data: 0.0702 max mem: 3536 +Test: [1280/2387] eta: 0:03:38 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1286 (0.3825) acc1: 100.0000 (82.7562) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1770 data: 0.0843 max mem: 3536 +Test: [1290/2387] eta: 0:03:36 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1348 (0.3807) acc1: 99.2188 (82.8718) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1767 data: 0.0814 max mem: 3536 +Test: [1300/2387] eta: 0:03:34 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1326 (0.3788) acc1: 99.2188 (82.9943) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1949 data: 0.0980 max mem: 3536 +Test: [1310/2387] eta: 0:03:32 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1396 (0.3771) acc1: 99.2188 (83.1115) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1881 data: 0.0907 max mem: 3536 +Test: [1320/2387] eta: 0:03:30 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1483 (0.3754) acc1: 99.2188 (83.2287) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1899 data: 0.0940 max mem: 3536 +Test: [1330/2387] eta: 0:03:28 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1520 (0.3739) acc1: 99.4792 (83.3437) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2048 data: 0.1115 max mem: 3536 +Test: [1340/2387] eta: 0:03:26 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1914 (0.3734) acc1: 98.1771 (83.3811) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1969 data: 0.1031 max mem: 3536 +Test: [1350/2387] eta: 0:03:24 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1025 (0.3713) acc1: 100.0000 (83.5041) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1773 data: 0.0832 max mem: 3537 +Test: [1360/2387] eta: 0:03:22 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.0894 (0.3692) acc1: 100.0000 (83.6253) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0921 max mem: 3537 +Test: [1370/2387] eta: 0:03:20 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.0879 (0.3672) acc1: 100.0000 (83.7448) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2036 data: 0.1100 max mem: 3537 +Test: [1380/2387] eta: 0:03:18 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.0879 (0.3652) acc1: 100.0000 (83.8625) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1928 data: 0.0997 max mem: 3537 +Test: [1390/2387] eta: 0:03:16 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.0882 (0.3632) acc1: 100.0000 (83.9785) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1891 data: 0.0968 max mem: 3537 +Test: [1400/2387] eta: 0:03:14 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.0878 (0.3612) acc1: 100.0000 (84.0928) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1900 data: 0.0967 max mem: 3537 +Test: [1410/2387] eta: 0:03:12 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.0891 (0.3603) acc1: 100.0000 (84.1659) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.2016 data: 0.1093 max mem: 3537 +Test: [1420/2387] eta: 0:03:10 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2434 (0.3597) acc1: 93.4896 (84.2205) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1911 data: 0.0998 max mem: 3537 +Test: [1430/2387] eta: 0:03:08 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2651 (0.3591) acc1: 92.1875 (84.2784) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1779 data: 0.0849 max mem: 3537 +Test: [1440/2387] eta: 0:03:06 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2418 (0.3583) acc1: 94.5312 (84.3600) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1862 data: 0.0928 max mem: 3537 +Test: [1450/2387] eta: 0:03:04 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2355 (0.3576) acc1: 95.5729 (84.4269) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1831 data: 0.0911 max mem: 3537 +Test: [1460/2387] eta: 0:03:02 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1917 (0.3564) acc1: 95.8333 (84.5137) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1835 data: 0.0924 max mem: 3537 +Test: [1470/2387] eta: 0:03:00 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1569 (0.3550) acc1: 98.9583 (84.6133) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1937 data: 0.1007 max mem: 3537 +Test: [1480/2387] eta: 0:02:58 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1473 (0.3536) acc1: 99.4792 (84.7072) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1887 data: 0.0958 max mem: 3537 +Test: [1490/2387] eta: 0:02:56 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1536 (0.3523) acc1: 99.2188 (84.7998) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1731 data: 0.0817 max mem: 3537 +Test: [1500/2387] eta: 0:02:54 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1536 (0.3510) acc1: 99.2188 (84.8941) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1813 data: 0.0886 max mem: 3537 +Test: [1510/2387] eta: 0:02:52 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1442 (0.3496) acc1: 99.7396 (84.9920) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1932 data: 0.1001 max mem: 3537 +Test: [1520/2387] eta: 0:02:50 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1729 (0.3493) acc1: 97.9167 (85.0340) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1974 data: 0.1044 max mem: 3537 +Test: [1530/2387] eta: 0:02:48 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2995 (0.3492) acc1: 91.1458 (85.0574) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.2012 data: 0.1075 max mem: 3537 +Test: [1540/2387] eta: 0:02:46 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2918 (0.3487) acc1: 91.4062 (85.1098) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1945 data: 0.1013 max mem: 3537 +Test: [1550/2387] eta: 0:02:44 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2705 (0.3482) acc1: 93.4896 (85.1594) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2048 data: 0.1106 max mem: 3537 +Test: [1560/2387] eta: 0:02:42 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2687 (0.3479) acc1: 93.4896 (85.2010) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.2049 data: 0.1087 max mem: 3537 +Test: [1570/2387] eta: 0:02:40 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2375 (0.3469) acc1: 96.0938 (85.2799) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2025 data: 0.1084 max mem: 3537 +Test: [1580/2387] eta: 0:02:38 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2109 (0.3462) acc1: 97.6562 (85.3455) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1904 data: 0.0983 max mem: 3538 +Test: [1590/2387] eta: 0:02:36 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2225 (0.3455) acc1: 95.0521 (85.4085) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1829 data: 0.0913 max mem: 3538 +Test: [1600/2387] eta: 0:02:34 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2225 (0.3448) acc1: 95.3125 (85.4607) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2065 data: 0.1163 max mem: 3538 +Test: [1610/2387] eta: 0:02:32 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2064 (0.3439) acc1: 97.6562 (85.5348) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1997 data: 0.1096 max mem: 3538 +Test: [1620/2387] eta: 0:02:30 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2086 (0.3432) acc1: 96.6146 (85.5964) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1953 data: 0.1051 max mem: 3538 +Test: [1630/2387] eta: 0:02:28 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2407 (0.3428) acc1: 94.5312 (85.6381) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.2022 data: 0.1104 max mem: 3538 +Test: [1640/2387] eta: 0:02:26 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2920 (0.3427) acc1: 90.3646 (85.6652) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1999 data: 0.1074 max mem: 3538 +Test: [1650/2387] eta: 0:02:24 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2960 (0.3423) acc1: 91.9271 (85.7099) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.2084 data: 0.1164 max mem: 3538 +Test: [1660/2387] eta: 0:02:22 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2980 (0.3422) acc1: 92.1875 (85.7326) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.2108 data: 0.1182 max mem: 3538 +Test: [1670/2387] eta: 0:02:20 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.3016 (0.3420) acc1: 91.9271 (85.7681) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2029 data: 0.1116 max mem: 3538 +Test: [1680/2387] eta: 0:02:19 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.3037 (0.3418) acc1: 91.4062 (85.7993) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1999 data: 0.1091 max mem: 3538 +Test: [1690/2387] eta: 0:02:17 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.3062 (0.3417) acc1: 91.1458 (85.8286) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1962 data: 0.1034 max mem: 3538 +Test: [1700/2387] eta: 0:02:15 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.3001 (0.3412) acc1: 92.1875 (85.8749) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1965 data: 0.1038 max mem: 3538 +Test: [1710/2387] eta: 0:02:13 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2177 (0.3403) acc1: 96.3542 (85.9440) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1849 data: 0.0941 max mem: 3538 +Test: [1720/2387] eta: 0:02:10 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1735 (0.3393) acc1: 98.1771 (86.0135) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1749 data: 0.0844 max mem: 3538 +Test: [1730/2387] eta: 0:02:08 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1722 (0.3384) acc1: 98.4375 (86.0800) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1687 data: 0.0778 max mem: 3538 +Test: [1740/2387] eta: 0:02:06 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1689 (0.3374) acc1: 98.6979 (86.1520) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1701 data: 0.0786 max mem: 3538 +Test: [1750/2387] eta: 0:02:04 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1663 (0.3364) acc1: 98.4375 (86.2234) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1770 data: 0.0855 max mem: 3538 +Test: [1760/2387] eta: 0:02:02 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1541 (0.3353) acc1: 99.7396 (86.2994) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0789 max mem: 3538 +Test: [1770/2387] eta: 0:02:00 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1659 (0.3347) acc1: 98.1771 (86.3485) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1922 data: 0.0978 max mem: 3538 +Test: [1780/2387] eta: 0:01:58 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2100 (0.3340) acc1: 95.8333 (86.4060) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1896 data: 0.0924 max mem: 3538 +Test: [1790/2387] eta: 0:01:56 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1967 (0.3333) acc1: 96.3542 (86.4618) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1549 data: 0.0605 max mem: 3538 +Test: [1800/2387] eta: 0:01:54 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1967 (0.3325) acc1: 97.3958 (86.5217) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1777 data: 0.0862 max mem: 3538 +Test: [1810/2387] eta: 0:01:52 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1811 (0.3317) acc1: 97.3958 (86.5850) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2049 data: 0.1121 max mem: 3539 +Test: [1820/2387] eta: 0:01:50 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1621 (0.3307) acc1: 98.9583 (86.6504) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1977 data: 0.1044 max mem: 3539 +Test: [1830/2387] eta: 0:01:48 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1475 (0.3297) acc1: 99.7396 (86.7222) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1800 data: 0.0889 max mem: 3539 +Test: [1840/2387] eta: 0:01:46 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1331 (0.3286) acc1: 100.0000 (86.7943) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1905 data: 0.0993 max mem: 3539 +Test: [1850/2387] eta: 0:01:44 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1323 (0.3276) acc1: 100.0000 (86.8655) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1913 data: 0.0996 max mem: 3539 +Test: [1860/2387] eta: 0:01:42 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1323 (0.3265) acc1: 100.0000 (86.9361) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1842 data: 0.0901 max mem: 3539 +Test: [1870/2387] eta: 0:01:41 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1330 (0.3255) acc1: 100.0000 (87.0058) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2088 data: 0.1151 max mem: 3539 +Test: [1880/2387] eta: 0:01:39 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1328 (0.3245) acc1: 100.0000 (87.0747) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1994 data: 0.1090 max mem: 3539 +Test: [1890/2387] eta: 0:01:37 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1325 (0.3235) acc1: 100.0000 (87.1429) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1704 data: 0.0789 max mem: 3539 +Test: [1900/2387] eta: 0:01:35 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1332 (0.3225) acc1: 100.0000 (87.2105) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1668 data: 0.0755 max mem: 3539 +Test: [1910/2387] eta: 0:01:32 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1510 (0.3216) acc1: 100.0000 (87.2757) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1703 data: 0.0802 max mem: 3539 +Test: [1920/2387] eta: 0:01:31 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1684 (0.3213) acc1: 99.7396 (87.3106) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1780 data: 0.0862 max mem: 3539 +Test: [1930/2387] eta: 0:01:29 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1861 (0.3205) acc1: 99.4792 (87.3723) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1936 data: 0.1004 max mem: 3539 +Test: [1940/2387] eta: 0:01:27 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1823 (0.3201) acc1: 99.4792 (87.3932) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1867 data: 0.0942 max mem: 3539 +Test: [1950/2387] eta: 0:01:25 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1966 (0.3196) acc1: 96.8750 (87.4395) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1699 data: 0.0773 max mem: 3539 +Test: [1960/2387] eta: 0:01:23 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2055 (0.3190) acc1: 98.6979 (87.4942) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1611 data: 0.0677 max mem: 3539 +Test: [1970/2387] eta: 0:01:21 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2332 (0.3189) acc1: 95.5729 (87.5186) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.1648 data: 0.0723 max mem: 3539 +Test: [1980/2387] eta: 0:01:19 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2508 (0.3184) acc1: 95.3125 (87.5614) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1646 data: 0.0735 max mem: 3539 +Test: [1990/2387] eta: 0:01:17 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1624 (0.3177) acc1: 98.1771 (87.6143) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1627 data: 0.0714 max mem: 3539 +Test: [2000/2387] eta: 0:01:15 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1644 (0.3169) acc1: 98.1771 (87.6658) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1747 data: 0.0817 max mem: 3539 +Test: [2010/2387] eta: 0:01:13 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1652 (0.3161) acc1: 98.1771 (87.7179) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1703 data: 0.0775 max mem: 3539 +Test: [2020/2387] eta: 0:01:11 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1656 (0.3154) acc1: 98.1771 (87.7692) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1632 data: 0.0718 max mem: 3539 +Test: [2030/2387] eta: 0:01:09 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1646 (0.3147) acc1: 97.9167 (87.8190) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1616 data: 0.0703 max mem: 3540 +Test: [2040/2387] eta: 0:01:07 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1624 (0.3139) acc1: 97.9167 (87.8698) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1649 data: 0.0725 max mem: 3540 +Test: [2050/2387] eta: 0:01:05 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1624 (0.3132) acc1: 98.1771 (87.9224) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1799 data: 0.0863 max mem: 3540 +Test: [2060/2387] eta: 0:01:03 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1616 (0.3124) acc1: 98.4375 (87.9728) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1848 data: 0.0927 max mem: 3540 +Test: [2070/2387] eta: 0:01:01 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1604 (0.3117) acc1: 98.6979 (88.0264) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1913 data: 0.1007 max mem: 3540 +Test: [2080/2387] eta: 0:00:59 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1605 (0.3110) acc1: 98.9583 (88.0768) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1906 data: 0.0994 max mem: 3540 +Test: [2090/2387] eta: 0:00:57 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1634 (0.3103) acc1: 98.6979 (88.1276) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1799 data: 0.0887 max mem: 3540 +Test: [2100/2387] eta: 0:00:55 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1621 (0.3096) acc1: 98.4375 (88.1770) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1909 data: 0.1005 max mem: 3540 +Test: [2110/2387] eta: 0:00:53 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1621 (0.3089) acc1: 98.4375 (88.2260) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1987 data: 0.1086 max mem: 3540 +Test: [2120/2387] eta: 0:00:51 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1730 (0.3087) acc1: 98.1771 (88.2444) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.1874 data: 0.0950 max mem: 3540 +Test: [2130/2387] eta: 0:00:49 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2091 (0.3083) acc1: 96.0938 (88.2831) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1865 data: 0.0944 max mem: 3540 +Test: [2140/2387] eta: 0:00:47 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2202 (0.3081) acc1: 96.0938 (88.3018) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1954 data: 0.1047 max mem: 3540 +Test: [2150/2387] eta: 0:00:45 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2461 (0.3078) acc1: 95.3125 (88.3321) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.2043 data: 0.1107 max mem: 3540 +Test: [2160/2387] eta: 0:00:43 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2295 (0.3074) acc1: 96.6146 (88.3734) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1950 data: 0.1020 max mem: 3540 +Test: [2170/2387] eta: 0:00:41 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2295 (0.3073) acc1: 96.0938 (88.3988) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1787 data: 0.0844 max mem: 3540 +Test: [2180/2387] eta: 0:00:39 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2984 (0.3085) acc1: 91.6667 (88.3289) acc5: 100.0000 (100.0000) acc1_neg: 43.229 acc5_neg: 100.000 time: 0.1843 data: 0.0891 max mem: 3540 +Test: [2190/2387] eta: 0:00:38 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.4667 (0.3094) acc1: 76.0417 (88.2644) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1857 data: 0.0923 max mem: 3540 +Test: [2200/2387] eta: 0:00:36 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1903 (0.3089) acc1: 95.5729 (88.3081) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1785 data: 0.0835 max mem: 3540 +Test: [2210/2387] eta: 0:00:34 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1712 (0.3082) acc1: 98.6979 (88.3552) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1683 data: 0.0744 max mem: 3540 +Test: [2220/2387] eta: 0:00:32 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1731 (0.3077) acc1: 98.9583 (88.4000) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1614 data: 0.0698 max mem: 3540 +Test: [2230/2387] eta: 0:00:30 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1720 (0.3071) acc1: 98.9583 (88.4465) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1619 data: 0.0699 max mem: 3540 +Test: [2240/2387] eta: 0:00:28 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1673 (0.3065) acc1: 98.4375 (88.4875) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1653 data: 0.0716 max mem: 3540 +Test: [2250/2387] eta: 0:00:26 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1738 (0.3060) acc1: 98.4375 (88.5299) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1648 data: 0.0725 max mem: 3540 +Test: [2260/2387] eta: 0:00:24 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1802 (0.3054) acc1: 98.6979 (88.5745) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1780 data: 0.0873 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1723 (0.3048) acc1: 99.4792 (88.6229) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1973 data: 0.1060 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1569 (0.3043) acc1: 100.0000 (88.6548) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1803 data: 0.0879 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1505 (0.3036) acc1: 99.7396 (88.7030) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1671 data: 0.0742 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1548 (0.3030) acc1: 99.7396 (88.7486) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1656 data: 0.0742 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1674 (0.3025) acc1: 99.4792 (88.7897) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1418 data: 0.0516 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.1925 (0.3023) acc1: 98.6979 (88.8200) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1234 data: 0.0333 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2661 (0.3023) acc1: 94.0104 (88.8306) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.1260 data: 0.0362 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2475 (0.3021) acc1: 95.8333 (88.8571) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1259 data: 0.0362 max mem: 3541 +Test: [2350/2387] eta: 0:00:07 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2365 (0.3022) acc1: 95.0521 (88.8535) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1229 data: 0.0332 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2863 (0.3022) acc1: 93.4896 (88.8645) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1218 data: 0.0322 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2652 (0.3021) acc1: 93.4896 (88.8784) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.1209 data: 0.0313 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.2840 (0.3022) acc1: 92.9688 (88.8897) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1211 data: 0.0313 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 21.461 acc5_pos: 100.000 loss: 0.3561 (0.3038) acc1: 88.5417 (88.8161) acc5: 100.0000 (100.0000) acc1_neg: 25.581 acc5_neg: 100.000 time: 0.1158 data: 0.0285 max mem: 3541 +Test: Total time: 0:07:32 (0.1897 s / it) +* Overall Acc@1 88.816 Acc@5 100.000 + target == real Acc@1 27.234 Acc@5 100.000 + target == fake Acc@1 93.141 Acc@5 100.000 + loss 0.304 + PR-AUC 0.9887 ROC-AUC 0.8545 EER 21.62% (thr=0.7656) +Accuracy of the network on the 916267 test images: 88.8% +[rank0]:[W726 05:45:19.338749127 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:45:39.687000 1195619 torch/distributed/run.py:766] +W0726 05:45:39.687000 1195619 torch/distributed/run.py:766] ***************************************** +W0726 05:45:39.687000 1195619 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:45:39.687000 1195619 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:46:07.097025872 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:46:07.320518370 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:46:08.585850660 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:46:08.808973042 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:46:08.865500475 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:46:08.091626512 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:46:08.097329311 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:46:08.116438922 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n16_seed8_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed8_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:07:19 acc1_pos: 41.927 acc5_pos: 100.000 loss: 0.9038 (0.9038) acc1: 41.9271 (41.9271) acc5: 100.0000 (100.0000) time: 7.8137 data: 6.2397 max mem: 3531 +Test: [ 10/517] eta: 0:06:45 acc1_pos: 18.490 acc5_pos: 100.000 loss: 0.9349 (0.9695) acc1: 39.5833 (33.8778) acc5: 100.0000 (100.0000) time: 0.8002 data: 0.5676 max mem: 3531 +Test: [ 20/517] eta: 0:03:51 acc1_pos: 31.771 acc5_pos: 100.000 loss: 0.9387 (0.9675) acc1: 31.7708 (34.7222) acc5: 100.0000 (100.0000) time: 0.0994 data: 0.0003 max mem: 3531 +Test: [ 30/517] eta: 0:03:14 acc1_pos: 15.625 acc5_pos: 100.000 loss: 1.0281 (0.9967) acc1: 27.3438 (32.6193) acc5: 100.0000 (100.0000) time: 0.1799 data: 0.0825 max mem: 3531 +Test: [ 40/517] eta: 0:02:54 acc1_pos: 26.042 acc5_pos: 100.000 loss: 1.0857 (1.0306) acc1: 22.6562 (29.1857) acc5: 100.0000 (100.0000) time: 0.2599 data: 0.1645 max mem: 3531 +Test: [ 50/517] eta: 0:02:37 acc1_pos: 24.312 acc5_pos: 100.000 loss: 1.0716 (1.0009) acc1: 22.3958 (31.7300) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.2399 data: 0.1456 max mem: 3531 +Test: [ 60/517] eta: 0:02:29 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.4609 (0.9055) acc1: 78.9062 (40.5951) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.2449 data: 0.1511 max mem: 3531 +Test: [ 70/517] eta: 0:02:22 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.4130 (0.8356) acc1: 85.9375 (47.2528) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.2757 data: 0.1829 max mem: 3531 +Test: [ 80/517] eta: 0:02:15 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.4237 (0.7874) acc1: 85.9375 (51.8872) acc5: 100.0000 (100.0000) acc1_neg: 75.521 acc5_neg: 100.000 time: 0.2649 data: 0.1720 max mem: 3531 +Test: [ 90/517] eta: 0:02:11 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.3793 (0.7418) acc1: 91.1458 (56.3273) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.2641 data: 0.1683 max mem: 3531 +Test: [100/517] eta: 0:02:07 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.3595 (0.7030) acc1: 93.2292 (60.1253) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.2799 data: 0.1824 max mem: 3531 +Test: [110/517] eta: 0:02:03 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.3472 (0.6703) acc1: 92.7083 (63.0678) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2800 data: 0.1827 max mem: 3531 +Test: [120/517] eta: 0:01:58 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.3637 (0.6458) acc1: 91.4062 (65.3861) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2599 data: 0.1652 max mem: 3531 +Test: [130/517] eta: 0:01:54 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.3703 (0.6242) acc1: 90.8854 (67.4181) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2549 data: 0.1611 max mem: 3531 +Test: [140/517] eta: 0:01:50 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.3191 (0.6015) acc1: 96.3542 (69.5774) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2649 data: 0.1692 max mem: 3531 +Test: [150/517] eta: 0:01:46 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.3050 (0.5820) acc1: 97.9167 (71.3593) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2599 data: 0.1635 max mem: 3531 +Test: [160/517] eta: 0:01:43 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.3250 (0.5669) acc1: 94.5312 (72.6029) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2649 data: 0.1656 max mem: 3531 +Test: [170/517] eta: 0:01:40 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.3274 (0.5521) acc1: 91.1458 (73.7847) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.2749 data: 0.1763 max mem: 3531 +Test: [180/517] eta: 0:01:36 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.3031 (0.5376) acc1: 95.3125 (74.9856) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2499 data: 0.1539 max mem: 3531 +Test: [190/517] eta: 0:01:32 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.2489 (0.5217) acc1: 96.8750 (76.2762) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2299 data: 0.1346 max mem: 3531 +Test: [200/517] eta: 0:01:29 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.2427 (0.5086) acc1: 99.4792 (77.4085) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2449 data: 0.1491 max mem: 3531 +Test: [210/517] eta: 0:01:25 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.3056 (0.5067) acc1: 89.8438 (77.6782) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.2499 data: 0.1521 max mem: 3532 +Test: [220/517] eta: 0:01:22 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.4399 (0.5042) acc1: 84.1146 (77.9742) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.2399 data: 0.1435 max mem: 3532 +Test: [230/517] eta: 0:01:19 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.4188 (0.4993) acc1: 87.7604 (78.5286) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.2349 data: 0.1391 max mem: 3532 +Test: [240/517] eta: 0:01:15 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.3916 (0.4967) acc1: 90.3646 (78.8976) acc5: 100.0000 (100.0000) acc1_neg: 64.583 acc5_neg: 100.000 time: 0.2399 data: 0.1444 max mem: 3532 +Test: [250/517] eta: 0:01:13 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.3909 (0.4908) acc1: 91.4062 (79.5215) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2499 data: 0.1538 max mem: 3532 +Test: [260/517] eta: 0:01:09 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.3078 (0.4842) acc1: 97.9167 (80.1954) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.2499 data: 0.1501 max mem: 3532 +Test: [270/517] eta: 0:01:06 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.3577 (0.4815) acc1: 95.0521 (80.5716) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2300 data: 0.1314 max mem: 3532 +Test: [280/517] eta: 0:01:04 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.3952 (0.4783) acc1: 91.1458 (80.9071) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.2449 data: 0.1483 max mem: 3532 +Test: [290/517] eta: 0:01:01 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.3659 (0.4743) acc1: 92.1875 (81.3386) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2647 data: 0.1647 max mem: 3532 +Test: [300/517] eta: 0:00:58 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.3659 (0.4709) acc1: 92.4479 (81.7085) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.2498 data: 0.1524 max mem: 3532 +Test: [310/517] eta: 0:00:55 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.3666 (0.4674) acc1: 92.9688 (82.0614) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.2156 data: 0.1223 max mem: 3532 +Test: [320/517] eta: 0:00:52 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.3266 (0.4624) acc1: 95.3125 (82.5131) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2349 data: 0.1380 max mem: 3532 +Test: [330/517] eta: 0:00:49 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.2830 (0.4564) acc1: 97.9167 (82.9824) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2642 data: 0.1657 max mem: 3532 +Test: [340/517] eta: 0:00:46 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.2515 (0.4527) acc1: 98.4375 (83.2165) acc5: 100.0000 (100.0000) acc1_neg: 76.042 acc5_neg: 100.000 time: 0.2408 data: 0.1459 max mem: 3532 +Test: [350/517] eta: 0:00:44 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.3990 (0.4520) acc1: 89.3229 (83.3237) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.2449 data: 0.1497 max mem: 3532 +Test: [360/517] eta: 0:00:41 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.3751 (0.4499) acc1: 89.5833 (83.5007) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.2499 data: 0.1563 max mem: 3532 +Test: [370/517] eta: 0:00:39 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.3188 (0.4456) acc1: 93.2292 (83.8984) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2799 data: 0.1879 max mem: 3532 +Test: [380/517] eta: 0:00:36 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.2952 (0.4428) acc1: 96.8750 (84.1474) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.2990 data: 0.2057 max mem: 3532 +Test: [390/517] eta: 0:00:33 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.3409 (0.4399) acc1: 94.5312 (84.4230) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2749 data: 0.1799 max mem: 3532 +Test: [400/517] eta: 0:00:31 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.3082 (0.4366) acc1: 95.5729 (84.7108) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2552 data: 0.1596 max mem: 3532 +Test: [410/517] eta: 0:00:28 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.3251 (0.4340) acc1: 96.3542 (84.9921) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2499 data: 0.1550 max mem: 3532 +Test: [420/517] eta: 0:00:25 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.2815 (0.4303) acc1: 97.3958 (85.3084) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2495 data: 0.1519 max mem: 3532 +Test: [430/517] eta: 0:00:22 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.2654 (0.4265) acc1: 99.2188 (85.6106) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2349 data: 0.1338 max mem: 3532 +Test: [440/517] eta: 0:00:20 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.3027 (0.4242) acc1: 97.6562 (85.8483) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2349 data: 0.1351 max mem: 3533 +Test: [450/517] eta: 0:00:17 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.3151 (0.4218) acc1: 96.3542 (86.0767) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.2499 data: 0.1541 max mem: 3533 +Test: [460/517] eta: 0:00:15 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.3151 (0.4201) acc1: 95.0521 (86.2154) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2599 data: 0.1646 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.2898 (0.4170) acc1: 97.3958 (86.4600) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2550 data: 0.1622 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.2999 (0.4146) acc1: 97.3958 (86.6738) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2512 data: 0.1572 max mem: 3533 +Test: [490/517] eta: 0:00:07 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.3014 (0.4122) acc1: 96.6146 (86.8673) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.2630 data: 0.1676 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.3944 (0.4137) acc1: 91.4062 (86.7520) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.2383 data: 0.1471 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.4183 (0.4129) acc1: 85.9375 (86.8074) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1857 data: 0.0961 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 24.312 acc5_pos: 100.000 loss: 0.2977 (0.4111) acc1: 92.4479 (86.9453) acc5: 100.0000 (100.0000) acc1_neg: 99.647 acc5_neg: 100.000 time: 0.1700 data: 0.0789 max mem: 3533 +Test: Total time: 0:02:13 (0.2590 s / it) +* Overall Acc@1 86.945 Acc@5 100.000 + target == real Acc@1 28.777 Acc@5 100.000 + target == fake Acc@1 92.980 Acc@5 100.000 + loss 0.411 + PR-AUC 0.9471 ROC-AUC 0.6956 EER 35.79% (thr=0.7114) +Accuracy of the network on the 198427 test images: 86.9% +[rank0]:[W726 05:48:33.381504259 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 05:48:54.755000 1196191 torch/distributed/run.py:766] +W0726 05:48:54.755000 1196191 torch/distributed/run.py:766] ***************************************** +W0726 05:48:54.755000 1196191 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:48:54.755000 1196191 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:49:23.822809641 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:49:23.070433406 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +[rank5]:[W726 05:49:23.304298474 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:49:23.311517492 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:49:23.471016420 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:49:24.522405533 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:49:24.596794492 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:49:24.700807100 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n16_seed8_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed8_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:47:22 acc1_pos: 18.750 acc5_pos: 100.000 loss: 1.2913 (1.2913) acc1: 18.7500 (18.7500) acc5: 100.0000 (100.0000) time: 7.3062 data: 5.3708 max mem: 3531 +Test: [ 10/389] eta: 0:04:43 acc1_pos: 97.917 acc5_pos: 100.000 loss: 0.4372 (0.6435) acc1: 83.3333 (65.4119) acc5: 100.0000 (100.0000) time: 0.7470 data: 0.4886 max mem: 3531 +Test: [ 20/389] eta: 0:02:41 acc1_pos: 14.583 acc5_pos: 100.000 loss: 0.5301 (0.7715) acc1: 67.7083 (57.1925) acc5: 100.0000 (100.0000) time: 0.0955 data: 0.0004 max mem: 3531 +Test: [ 30/389] eta: 0:02:12 acc1_pos: 52.865 acc5_pos: 100.000 loss: 0.7233 (0.7880) acc1: 52.8646 (53.8726) acc5: 100.0000 (100.0000) time: 0.1599 data: 0.0623 max mem: 3531 +Test: [ 40/389] eta: 0:01:55 acc1_pos: 37.500 acc5_pos: 100.000 loss: 0.7029 (0.7757) acc1: 52.8646 (54.1286) acc5: 100.0000 (100.0000) time: 0.2180 data: 0.1250 max mem: 3531 +Test: [ 50/389] eta: 0:01:43 acc1_pos: 93.750 acc5_pos: 100.000 loss: 0.4701 (0.7152) acc1: 78.6458 (58.8848) acc5: 100.0000 (100.0000) time: 0.2049 data: 0.1123 max mem: 3531 +Test: [ 60/389] eta: 0:01:32 acc1_pos: 74.219 acc5_pos: 100.000 loss: 0.3457 (0.7029) acc1: 89.5833 (59.8361) acc5: 100.0000 (100.0000) time: 0.1818 data: 0.0878 max mem: 3531 +Test: [ 70/389] eta: 0:01:25 acc1_pos: 78.906 acc5_pos: 100.000 loss: 0.5397 (0.6911) acc1: 73.4375 (60.3727) acc5: 100.0000 (100.0000) time: 0.1725 data: 0.0803 max mem: 3531 +Test: [ 80/389] eta: 0:01:18 acc1_pos: 87.240 acc5_pos: 100.000 loss: 0.5830 (0.6885) acc1: 60.1562 (60.0277) acc5: 100.0000 (100.0000) time: 0.1703 data: 0.0796 max mem: 3531 +Test: [ 90/389] eta: 0:01:13 acc1_pos: 59.635 acc5_pos: 100.000 loss: 0.4343 (0.6661) acc1: 70.8333 (61.7874) acc5: 100.0000 (100.0000) time: 0.1778 data: 0.0876 max mem: 3531 +Test: [100/389] eta: 0:01:09 acc1_pos: 74.219 acc5_pos: 100.000 loss: 0.6203 (0.6752) acc1: 63.0208 (60.9839) acc5: 100.0000 (100.0000) time: 0.1869 data: 0.0944 max mem: 3531 +Test: [110/389] eta: 0:01:05 acc1_pos: 83.333 acc5_pos: 100.000 loss: 0.6653 (0.6960) acc1: 59.8958 (60.2477) acc5: 100.0000 (100.0000) time: 0.1805 data: 0.0874 max mem: 3531 +Test: [120/389] eta: 0:01:01 acc1_pos: 77.083 acc5_pos: 100.000 loss: 0.5088 (0.6795) acc1: 68.2292 (61.4497) acc5: 100.0000 (100.0000) time: 0.1725 data: 0.0801 max mem: 3531 +Test: [130/389] eta: 0:00:57 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.5088 (0.6730) acc1: 74.2188 (61.8360) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.1588 data: 0.0662 max mem: 3531 +Test: [140/389] eta: 0:00:54 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.5327 (0.6707) acc1: 71.8750 (62.3504) acc5: 100.0000 (100.0000) acc1_neg: 0.781 acc5_neg: 100.000 time: 0.1549 data: 0.0620 max mem: 3531 +Test: [150/389] eta: 0:00:51 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.5903 (0.6772) acc1: 66.4062 (61.8584) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1593 data: 0.0656 max mem: 3531 +Test: [160/389] eta: 0:00:48 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.6173 (0.6673) acc1: 52.0833 (62.7200) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1646 data: 0.0726 max mem: 3531 +Test: [170/389] eta: 0:00:45 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.2720 (0.6617) acc1: 95.0521 (63.2310) acc5: 100.0000 (100.0000) acc1_neg: 27.865 acc5_neg: 100.000 time: 0.1555 data: 0.0640 max mem: 3531 +Test: [180/389] eta: 0:00:43 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.4740 (0.6596) acc1: 66.4062 (63.4021) acc5: 100.0000 (100.0000) acc1_neg: 66.406 acc5_neg: 100.000 time: 0.1655 data: 0.0725 max mem: 3531 +Test: [190/389] eta: 0:00:40 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.7075 (0.6569) acc1: 60.4167 (63.6235) acc5: 100.0000 (100.0000) acc1_neg: 49.479 acc5_neg: 100.000 time: 0.1763 data: 0.0842 max mem: 3531 +Test: [200/389] eta: 0:00:38 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.6431 (0.6530) acc1: 60.6771 (63.8267) acc5: 100.0000 (100.0000) acc1_neg: 42.708 acc5_neg: 100.000 time: 0.1702 data: 0.0797 max mem: 3531 +Test: [210/389] eta: 0:00:36 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.5713 (0.6462) acc1: 63.2812 (64.3192) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1748 data: 0.0845 max mem: 3532 +Test: [220/389] eta: 0:00:34 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.2640 (0.6301) acc1: 98.4375 (65.6863) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1854 data: 0.0947 max mem: 3532 +Test: [230/389] eta: 0:00:32 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.3236 (0.6275) acc1: 94.5312 (65.9237) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.2111 data: 0.1193 max mem: 3532 +Test: [240/389] eta: 0:00:30 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.5792 (0.6270) acc1: 67.4479 (65.9394) acc5: 100.0000 (100.0000) acc1_neg: 63.281 acc5_neg: 100.000 time: 0.2057 data: 0.1141 max mem: 3532 +Test: [250/389] eta: 0:00:28 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.4229 (0.6156) acc1: 75.7812 (66.8368) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2038 data: 0.1133 max mem: 3532 +Test: [260/389] eta: 0:00:25 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.2801 (0.6097) acc1: 97.6562 (67.3791) acc5: 100.0000 (100.0000) acc1_neg: 17.969 acc5_neg: 100.000 time: 0.1897 data: 0.0995 max mem: 3532 +Test: [270/389] eta: 0:00:24 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.5112 (0.6208) acc1: 73.6979 (66.4168) acc5: 100.0000 (100.0000) acc1_neg: 41.146 acc5_neg: 100.000 time: 0.2038 data: 0.1134 max mem: 3532 +Test: [280/389] eta: 0:00:22 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.8554 (0.6281) acc1: 36.4583 (65.5722) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2276 data: 0.1355 max mem: 3532 +Test: [290/389] eta: 0:00:20 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.6730 (0.6238) acc1: 52.3438 (65.9570) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.2014 data: 0.1091 max mem: 3532 +Test: [300/389] eta: 0:00:18 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.7371 (0.6319) acc1: 51.0417 (65.0773) acc5: 100.0000 (100.0000) acc1_neg: 42.969 acc5_neg: 100.000 time: 0.2103 data: 0.1193 max mem: 3532 +Test: [310/389] eta: 0:00:16 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.8927 (0.6386) acc1: 41.9271 (64.6051) acc5: 100.0000 (100.0000) acc1_neg: 5.208 acc5_neg: 100.000 time: 0.2099 data: 0.1195 max mem: 3532 +Test: [320/389] eta: 0:00:14 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.5903 (0.6357) acc1: 67.7083 (64.9427) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.2015 data: 0.1112 max mem: 3532 +Test: [330/389] eta: 0:00:12 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.4229 (0.6298) acc1: 87.5000 (65.5880) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.2129 data: 0.1221 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.3930 (0.6183) acc1: 90.1042 (66.5208) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2102 data: 0.1192 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.2734 (0.6312) acc1: 97.1354 (65.8943) acc5: 100.0000 (100.0000) acc1_neg: 0.521 acc5_neg: 100.000 time: 0.1873 data: 0.0969 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 64.417 acc5_pos: 100.000 loss: 1.1012 (0.6469) acc1: 22.6562 (64.5978) acc5: 100.0000 (100.0000) acc1_neg: 23.958 acc5_neg: 100.000 time: 0.1587 data: 0.0690 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 64.417 acc5_pos: 100.000 loss: 1.0504 (0.6452) acc1: 26.3021 (64.7314) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1446 data: 0.0551 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.5932 (0.6466) acc1: 67.7083 (64.6790) acc5: 100.0000 (100.0000) acc1_neg: 67.708 acc5_neg: 100.000 time: 0.1350 data: 0.0456 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 64.417 acc5_pos: 100.000 loss: 0.5363 (0.6438) acc1: 72.1354 (64.9145) acc5: 100.0000 (100.0000) acc1_neg: 67.760 acc5_neg: 100.000 time: 0.1295 data: 0.0367 max mem: 3532 +Test: Total time: 0:01:16 (0.1970 s / it) +* Overall Acc@1 64.915 Acc@5 100.000 + target == real Acc@1 61.604 Acc@5 100.000 + target == fake Acc@1 66.454 Acc@5 100.000 + loss 0.644 + PR-AUC 0.8238 ROC-AUC 0.6961 EER 35.94% (thr=0.5249) +Accuracy of the network on the 149358 test images: 64.9% +[rank0]:[W726 05:50:48.418475877 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:51:08.097000 1196750 torch/distributed/run.py:766] +W0726 05:51:08.097000 1196750 torch/distributed/run.py:766] ***************************************** +W0726 05:51:08.097000 1196750 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:51:08.097000 1196750 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:51:37.203963753 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:51:38.807425545 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:51:38.854224244 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:51:38.250258377 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:51:38.363499988 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:51:39.497379962 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:51:39.714945220 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:51:39.730357187 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n16_seed8_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed8_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:39:02 acc1_pos: 85.156 acc5_pos: 100.000 loss: 0.3422 (0.3422) acc1: 85.1562 (85.1562) acc5: 100.0000 (100.0000) time: 7.0355 data: 5.4041 max mem: 3531 +Test: [ 10/333] eta: 0:04:06 acc1_pos: 88.021 acc5_pos: 100.000 loss: 0.2747 (0.2751) acc1: 88.0208 (89.1335) acc5: 100.0000 (100.0000) time: 0.7641 data: 0.5307 max mem: 3531 +Test: [ 20/333] eta: 0:02:19 acc1_pos: 95.052 acc5_pos: 100.000 loss: 0.2645 (0.2653) acc1: 88.8021 (89.5833) acc5: 100.0000 (100.0000) time: 0.1175 data: 0.0219 max mem: 3531 +Test: [ 30/333] eta: 0:02:03 acc1_pos: 60.417 acc5_pos: 100.000 loss: 0.3215 (0.4042) acc1: 81.5104 (79.4019) acc5: 100.0000 (100.0000) time: 0.2089 data: 0.1132 max mem: 3531 +Test: [ 40/333] eta: 0:01:47 acc1_pos: 51.562 acc5_pos: 100.000 loss: 0.7299 (0.4879) acc1: 52.6042 (72.9738) acc5: 100.0000 (100.0000) time: 0.2799 data: 0.1865 max mem: 3531 +Test: [ 50/333] eta: 0:01:32 acc1_pos: 54.948 acc5_pos: 100.000 loss: 0.7299 (0.5407) acc1: 52.8646 (69.0002) acc5: 100.0000 (100.0000) time: 0.1999 data: 0.1066 max mem: 3531 +Test: [ 60/333] eta: 0:01:22 acc1_pos: 50.781 acc5_pos: 100.000 loss: 0.7233 (0.5778) acc1: 52.8646 (66.3977) acc5: 100.0000 (100.0000) time: 0.1700 data: 0.0772 max mem: 3531 +Test: [ 70/333] eta: 0:01:14 acc1_pos: 52.344 acc5_pos: 100.000 loss: 0.7340 (0.5982) acc1: 52.3438 (64.9098) acc5: 100.0000 (100.0000) time: 0.1731 data: 0.0807 max mem: 3531 +Test: [ 80/333] eta: 0:01:07 acc1_pos: 57.292 acc5_pos: 100.000 loss: 0.6933 (0.6122) acc1: 56.2500 (63.6317) acc5: 100.0000 (100.0000) time: 0.1666 data: 0.0746 max mem: 3531 +Test: [ 90/333] eta: 0:01:02 acc1_pos: 66.406 acc5_pos: 100.000 loss: 0.6632 (0.6149) acc1: 58.5938 (63.4129) acc5: 100.0000 (100.0000) time: 0.1729 data: 0.0817 max mem: 3531 +Test: [100/333] eta: 0:00:58 acc1_pos: 73.698 acc5_pos: 100.000 loss: 0.6281 (0.6182) acc1: 62.7604 (63.2580) acc5: 100.0000 (100.0000) time: 0.1825 data: 0.0908 max mem: 3531 +Test: [110/333] eta: 0:00:54 acc1_pos: 68.229 acc5_pos: 100.000 loss: 0.6374 (0.6202) acc1: 60.6771 (63.1217) acc5: 100.0000 (100.0000) time: 0.1831 data: 0.0915 max mem: 3531 +Test: [120/333] eta: 0:00:50 acc1_pos: 66.927 acc5_pos: 100.000 loss: 0.6092 (0.6203) acc1: 63.8021 (63.1263) acc5: 100.0000 (100.0000) time: 0.1748 data: 0.0835 max mem: 3531 +Test: [130/333] eta: 0:00:47 acc1_pos: 76.302 acc5_pos: 100.000 loss: 0.5892 (0.6187) acc1: 66.9271 (63.3946) acc5: 100.0000 (100.0000) time: 0.1846 data: 0.0902 max mem: 3531 +Test: [140/333] eta: 0:00:44 acc1_pos: 82.812 acc5_pos: 100.000 loss: 0.5921 (0.6184) acc1: 66.6667 (63.4401) acc5: 100.0000 (100.0000) time: 0.1934 data: 0.0973 max mem: 3531 +Test: [150/333] eta: 0:00:41 acc1_pos: 69.010 acc5_pos: 100.000 loss: 0.6424 (0.6207) acc1: 64.8438 (63.2071) acc5: 100.0000 (100.0000) time: 0.1787 data: 0.0822 max mem: 3531 +Test: [160/333] eta: 0:00:38 acc1_pos: 53.289 acc5_pos: 100.000 loss: 0.6555 (0.6325) acc1: 56.7708 (62.6229) acc5: 100.0000 (100.0000) acc1_neg: 39.062 acc5_neg: 100.000 time: 0.1729 data: 0.0745 max mem: 3531 +Test: [170/333] eta: 0:00:36 acc1_pos: 53.289 acc5_pos: 100.000 loss: 0.9195 (0.6514) acc1: 45.8333 (61.5695) acc5: 100.0000 (100.0000) acc1_neg: 48.698 acc5_neg: 100.000 time: 0.1899 data: 0.0949 max mem: 3531 +Test: [180/333] eta: 0:00:34 acc1_pos: 53.289 acc5_pos: 100.000 loss: 0.8979 (0.6629) acc1: 47.1354 (60.8282) acc5: 100.0000 (100.0000) acc1_neg: 40.625 acc5_neg: 100.000 time: 0.2181 data: 0.1263 max mem: 3531 +Test: [190/333] eta: 0:00:31 acc1_pos: 53.289 acc5_pos: 100.000 loss: 0.8204 (0.6663) acc1: 52.0833 (60.7698) acc5: 100.0000 (100.0000) acc1_neg: 53.125 acc5_neg: 100.000 time: 0.2249 data: 0.1303 max mem: 3531 +Test: [200/333] eta: 0:00:29 acc1_pos: 53.289 acc5_pos: 100.000 loss: 0.7511 (0.6719) acc1: 57.0312 (60.6084) acc5: 100.0000 (100.0000) acc1_neg: 53.906 acc5_neg: 100.000 time: 0.2199 data: 0.1259 max mem: 3531 +Test: [210/333] eta: 0:00:27 acc1_pos: 53.289 acc5_pos: 100.000 loss: 0.8659 (0.6829) acc1: 51.5625 (60.0649) acc5: 100.0000 (100.0000) acc1_neg: 44.010 acc5_neg: 100.000 time: 0.2053 data: 0.1143 max mem: 3532 +Test: [220/333] eta: 0:00:25 acc1_pos: 53.289 acc5_pos: 100.000 loss: 0.8845 (0.6939) acc1: 49.2188 (59.4375) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.2098 data: 0.1168 max mem: 3532 +Test: [230/333] eta: 0:00:22 acc1_pos: 53.289 acc5_pos: 100.000 loss: 0.6067 (0.6881) acc1: 68.2292 (60.0334) acc5: 100.0000 (100.0000) acc1_neg: 70.573 acc5_neg: 100.000 time: 0.2361 data: 0.1431 max mem: 3532 +Test: [240/333] eta: 0:00:20 acc1_pos: 53.289 acc5_pos: 100.000 loss: 0.5879 (0.6876) acc1: 68.7500 (60.0936) acc5: 100.0000 (100.0000) acc1_neg: 59.896 acc5_neg: 100.000 time: 0.2124 data: 0.1212 max mem: 3532 +Test: [250/333] eta: 0:00:18 acc1_pos: 53.289 acc5_pos: 100.000 loss: 0.6973 (0.6876) acc1: 60.6771 (60.1843) acc5: 100.0000 (100.0000) acc1_neg: 58.854 acc5_neg: 100.000 time: 0.1938 data: 0.1022 max mem: 3532 +Test: [260/333] eta: 0:00:16 acc1_pos: 53.289 acc5_pos: 100.000 loss: 0.6874 (0.6878) acc1: 62.5000 (60.3149) acc5: 100.0000 (100.0000) acc1_neg: 66.406 acc5_neg: 100.000 time: 0.2124 data: 0.1215 max mem: 3532 +Test: [270/333] eta: 0:00:13 acc1_pos: 53.289 acc5_pos: 100.000 loss: 0.6874 (0.6875) acc1: 62.5000 (60.4205) acc5: 100.0000 (100.0000) acc1_neg: 61.458 acc5_neg: 100.000 time: 0.2167 data: 0.1254 max mem: 3532 +Test: [280/333] eta: 0:00:11 acc1_pos: 53.289 acc5_pos: 100.000 loss: 0.6967 (0.6936) acc1: 61.1979 (60.0321) acc5: 100.0000 (100.0000) acc1_neg: 51.042 acc5_neg: 100.000 time: 0.2287 data: 0.1370 max mem: 3532 +Test: [290/333] eta: 0:00:09 acc1_pos: 53.289 acc5_pos: 100.000 loss: 0.9808 (0.7040) acc1: 38.0208 (59.2900) acc5: 100.0000 (100.0000) acc1_neg: 46.354 acc5_neg: 100.000 time: 0.2036 data: 0.1122 max mem: 3532 +Test: [300/333] eta: 0:00:07 acc1_pos: 53.289 acc5_pos: 100.000 loss: 0.9846 (0.7128) acc1: 40.3646 (58.7408) acc5: 100.0000 (100.0000) acc1_neg: 46.094 acc5_neg: 100.000 time: 0.1648 data: 0.0739 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 53.289 acc5_pos: 100.000 loss: 0.8756 (0.7127) acc1: 47.9167 (58.7997) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.1597 data: 0.0694 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 53.289 acc5_pos: 100.000 loss: 0.6587 (0.7115) acc1: 62.7604 (58.9280) acc5: 100.0000 (100.0000) acc1_neg: 40.885 acc5_neg: 100.000 time: 0.1340 data: 0.0442 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 53.289 acc5_pos: 100.000 loss: 0.9725 (0.7217) acc1: 44.0104 (58.3986) acc5: 100.0000 (100.0000) acc1_neg: 42.188 acc5_neg: 100.000 time: 0.1206 data: 0.0311 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 53.289 acc5_pos: 100.000 loss: 1.0159 (0.7235) acc1: 43.2292 (58.3168) acc5: 100.0000 (100.0000) acc1_neg: 45.352 acc5_neg: 100.000 time: 0.1303 data: 0.0374 max mem: 3532 +Test: Total time: 0:01:10 (0.2104 s / it) +* Overall Acc@1 58.317 Acc@5 100.000 + target == real Acc@1 63.339 Acc@5 100.000 + target == fake Acc@1 53.846 Acc@5 100.000 + loss 0.724 + PR-AUC 0.6440 ROC-AUC 0.6282 EER 41.19% (thr=0.4590) +Accuracy of the network on the 127843 test images: 58.3% +[rank0]:[W726 05:52:57.519300490 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:53:16.569000 1197309 torch/distributed/run.py:766] +W0726 05:53:16.569000 1197309 torch/distributed/run.py:766] ***************************************** +W0726 05:53:16.569000 1197309 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:53:16.569000 1197309 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 05:53:45.755167763 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:53:45.026328155 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:53:45.289141842 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:53:45.311659140 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:53:45.356655678 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:53:45.441222885 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:53:46.772391901 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:53:46.800232505 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n16_seed8_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed8_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:21:18 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.1603 (0.1603) acc1: 96.8750 (96.8750) acc5: 100.0000 (100.0000) time: 5.4409 data: 3.6337 max mem: 3531 +Test: [ 10/235] eta: 0:02:10 acc1_pos: 96.615 acc5_pos: 100.000 loss: 0.1607 (0.1599) acc1: 96.8750 (96.9934) acc5: 100.0000 (100.0000) time: 0.5781 data: 0.3325 max mem: 3531 +Test: [ 20/235] eta: 0:01:19 acc1_pos: 97.396 acc5_pos: 100.000 loss: 0.1519 (0.1555) acc1: 97.3958 (97.2594) acc5: 100.0000 (100.0000) time: 0.1174 data: 0.0269 max mem: 3531 +Test: [ 30/235] eta: 0:01:08 acc1_pos: 97.396 acc5_pos: 100.000 loss: 0.1495 (0.1537) acc1: 97.3958 (97.3286) acc5: 100.0000 (100.0000) time: 0.2014 data: 0.1084 max mem: 3531 +Test: [ 40/235] eta: 0:01:01 acc1_pos: 97.135 acc5_pos: 100.000 loss: 0.1500 (0.1538) acc1: 97.3958 (97.3514) acc5: 100.0000 (100.0000) time: 0.2599 data: 0.1648 max mem: 3531 +Test: [ 50/235] eta: 0:00:54 acc1_pos: 96.615 acc5_pos: 100.000 loss: 0.1561 (0.1550) acc1: 97.1354 (97.2988) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1345 max mem: 3531 +Test: [ 60/235] eta: 0:00:45 acc1_pos: 96.615 acc5_pos: 100.000 loss: 0.1588 (0.1563) acc1: 96.8750 (97.2464) acc5: 100.0000 (100.0000) time: 0.1499 data: 0.0526 max mem: 3531 +Test: [ 70/235] eta: 0:00:41 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1755 (0.3364) acc1: 95.8333 (87.4450) acc5: 100.0000 (100.0000) acc1_neg: 16.146 acc5_neg: 100.000 time: 0.1403 data: 0.0442 max mem: 3531 +Test: [ 80/235] eta: 0:00:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.5990 (0.4933) acc1: 18.7500 (78.9095) acc5: 100.0000 (100.0000) acc1_neg: 15.885 acc5_neg: 100.000 time: 0.2070 data: 0.1145 max mem: 3531 +Test: [ 90/235] eta: 0:00:35 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.6231 (0.6170) acc1: 18.2292 (72.2155) acc5: 100.0000 (100.0000) acc1_neg: 15.625 acc5_neg: 100.000 time: 0.2100 data: 0.1188 max mem: 3531 +Test: [100/235] eta: 0:00:32 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.6185 (0.7137) acc1: 18.2292 (66.9503) acc5: 100.0000 (100.0000) acc1_neg: 15.885 acc5_neg: 100.000 time: 0.1928 data: 0.1023 max mem: 3531 +Test: [110/235] eta: 0:00:29 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.5791 (0.7934) acc1: 19.2708 (62.6032) acc5: 100.0000 (100.0000) acc1_neg: 21.615 acc5_neg: 100.000 time: 0.1949 data: 0.1025 max mem: 3531 +Test: [120/235] eta: 0:00:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.6582 (0.8916) acc1: 15.8854 (57.9804) acc5: 100.0000 (100.0000) acc1_neg: 4.167 acc5_neg: 100.000 time: 0.2034 data: 0.1091 max mem: 3531 +Test: [130/235] eta: 0:00:24 acc1_pos: 100.000 acc5_pos: 100.000 loss: 2.0076 (0.9773) acc1: 5.2083 (53.9818) acc5: 100.0000 (100.0000) acc1_neg: 4.688 acc5_neg: 100.000 time: 0.2152 data: 0.1212 max mem: 3531 +Test: [140/235] eta: 0:00:21 acc1_pos: 100.000 acc5_pos: 100.000 loss: 2.0427 (1.0532) acc1: 4.9479 (50.4820) acc5: 100.0000 (100.0000) acc1_neg: 3.906 acc5_neg: 100.000 time: 0.2163 data: 0.1240 max mem: 3531 +Test: [150/235] eta: 0:00:19 acc1_pos: 100.000 acc5_pos: 100.000 loss: 2.0385 (1.1170) acc1: 4.6875 (47.4717) acc5: 100.0000 (100.0000) acc1_neg: 6.510 acc5_neg: 100.000 time: 0.2207 data: 0.1294 max mem: 3531 +Test: [160/235] eta: 0:00:17 acc1_pos: 100.000 acc5_pos: 100.000 loss: 2.0033 (1.1722) acc1: 5.2083 (44.8822) acc5: 100.0000 (100.0000) acc1_neg: 5.469 acc5_neg: 100.000 time: 0.2149 data: 0.1234 max mem: 3531 +Test: [170/235] eta: 0:00:14 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.9637 (1.1936) acc1: 6.7708 (43.5825) acc5: 100.0000 (100.0000) acc1_neg: 22.656 acc5_neg: 100.000 time: 0.2085 data: 0.1152 max mem: 3531 +Test: [180/235] eta: 0:00:12 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.4891 (1.2089) acc1: 24.2188 (42.5299) acc5: 100.0000 (100.0000) acc1_neg: 22.656 acc5_neg: 100.000 time: 0.2154 data: 0.1203 max mem: 3531 +Test: [190/235] eta: 0:00:10 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.4427 (1.2219) acc1: 23.9583 (41.5712) acc5: 100.0000 (100.0000) acc1_neg: 27.083 acc5_neg: 100.000 time: 0.2270 data: 0.1324 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.4523 (1.2351) acc1: 23.6979 (40.6470) acc5: 100.0000 (100.0000) acc1_neg: 24.740 acc5_neg: 100.000 time: 0.2145 data: 0.1198 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.4994 (1.2465) acc1: 23.9583 (39.8635) acc5: 100.0000 (100.0000) acc1_neg: 20.833 acc5_neg: 100.000 time: 0.1946 data: 0.1017 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.4633 (1.2565) acc1: 24.2188 (39.1273) acc5: 100.0000 (100.0000) acc1_neg: 21.094 acc5_neg: 100.000 time: 0.1740 data: 0.0839 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.4597 (1.2653) acc1: 25.0000 (38.5304) acc5: 100.0000 (100.0000) acc1_neg: 23.698 acc5_neg: 100.000 time: 0.1448 data: 0.0553 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.4651 (1.2699) acc1: 23.9583 (38.3338) acc5: 100.0000 (100.0000) acc1_neg: 17.391 acc5_neg: 100.000 time: 0.1334 data: 0.0472 max mem: 3532 +Test: Total time: 0:00:50 (0.2156 s / it) +* Overall Acc@1 38.334 Acc@5 100.000 + target == real Acc@1 97.265 Acc@5 100.000 + target == fake Acc@1 17.020 Acc@5 100.000 + loss 1.270 + PR-AUC 0.8729 ROC-AUC 0.7213 EER 34.10% (thr=0.1183) +Accuracy of the network on the 89879 test images: 38.3% +[rank0]:[W726 05:54:45.509137147 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:55:03.388000 1197856 torch/distributed/run.py:766] +W0726 05:55:03.388000 1197856 torch/distributed/run.py:766] ***************************************** +W0726 05:55:03.388000 1197856 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:55:03.388000 1197856 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:55:31.331170268 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:55:31.479783337 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:55:32.595884836 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:55:32.982494940 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:55:32.003323320 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:55:32.113835186 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 7): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:55:32.264696187 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W726 05:55:32.264710464 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n16_seed8_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed8_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:28:12 acc1_pos: 58.594 acc5_pos: 100.000 loss: 0.6291 (0.6291) acc1: 58.5938 (58.5938) acc5: 100.0000 (100.0000) time: 6.2907 data: 4.7202 max mem: 3531 +Test: [ 10/269] eta: 0:02:49 acc1_pos: 74.219 acc5_pos: 100.000 loss: 0.5488 (0.5587) acc1: 74.2188 (69.3892) acc5: 100.0000 (100.0000) time: 0.6537 data: 0.4295 max mem: 3531 +Test: [ 20/269] eta: 0:01:36 acc1_pos: 79.948 acc5_pos: 100.000 loss: 0.5632 (0.5683) acc1: 66.9271 (68.9112) acc5: 100.0000 (100.0000) time: 0.0916 data: 0.0004 max mem: 3531 +Test: [ 30/269] eta: 0:01:13 acc1_pos: 76.042 acc5_pos: 100.000 loss: 0.5688 (0.5658) acc1: 67.1875 (69.7833) acc5: 100.0000 (100.0000) time: 0.1179 data: 0.0232 max mem: 3531 +Test: [ 40/269] eta: 0:01:04 acc1_pos: 54.310 acc5_pos: 100.000 loss: 0.6361 (0.6043) acc1: 64.8438 (67.4225) acc5: 100.0000 (100.0000) acc1_neg: 66.927 acc5_neg: 100.000 time: 0.1711 data: 0.0765 max mem: 3531 +Test: [ 50/269] eta: 0:00:58 acc1_pos: 54.310 acc5_pos: 100.000 loss: 0.7170 (0.6358) acc1: 61.4583 (65.0633) acc5: 100.0000 (100.0000) acc1_neg: 35.156 acc5_neg: 100.000 time: 0.1986 data: 0.1019 max mem: 3531 +Test: [ 60/269] eta: 0:00:53 acc1_pos: 54.310 acc5_pos: 100.000 loss: 0.7510 (0.6710) acc1: 53.6458 (62.4530) acc5: 100.0000 (100.0000) acc1_neg: 38.542 acc5_neg: 100.000 time: 0.2037 data: 0.1050 max mem: 3531 +Test: [ 70/269] eta: 0:00:49 acc1_pos: 54.310 acc5_pos: 100.000 loss: 0.8164 (0.6928) acc1: 48.1771 (60.4937) acc5: 100.0000 (100.0000) acc1_neg: 65.365 acc5_neg: 100.000 time: 0.2048 data: 0.1091 max mem: 3531 +Test: [ 80/269] eta: 0:00:46 acc1_pos: 54.310 acc5_pos: 100.000 loss: 0.7624 (0.6982) acc1: 53.3854 (59.7319) acc5: 100.0000 (100.0000) acc1_neg: 40.625 acc5_neg: 100.000 time: 0.2099 data: 0.1144 max mem: 3531 +Test: [ 90/269] eta: 0:00:42 acc1_pos: 54.310 acc5_pos: 100.000 loss: 0.7946 (0.7172) acc1: 49.4792 (58.0930) acc5: 100.0000 (100.0000) acc1_neg: 47.917 acc5_neg: 100.000 time: 0.2099 data: 0.1137 max mem: 3531 +Test: [100/269] eta: 0:00:39 acc1_pos: 54.310 acc5_pos: 100.000 loss: 0.8715 (0.7332) acc1: 47.1354 (57.0751) acc5: 100.0000 (100.0000) acc1_neg: 33.854 acc5_neg: 100.000 time: 0.1904 data: 0.0963 max mem: 3531 +Test: [110/269] eta: 0:00:36 acc1_pos: 54.310 acc5_pos: 100.000 loss: 0.9553 (0.7586) acc1: 37.2396 (55.2506) acc5: 100.0000 (100.0000) acc1_neg: 35.156 acc5_neg: 100.000 time: 0.1855 data: 0.0938 max mem: 3531 +Test: [120/269] eta: 0:00:34 acc1_pos: 54.310 acc5_pos: 100.000 loss: 0.9553 (0.7718) acc1: 37.2396 (54.1667) acc5: 100.0000 (100.0000) acc1_neg: 30.208 acc5_neg: 100.000 time: 0.1994 data: 0.1051 max mem: 3531 +Test: [130/269] eta: 0:00:31 acc1_pos: 54.310 acc5_pos: 100.000 loss: 0.9398 (0.7849) acc1: 37.2396 (52.9242) acc5: 100.0000 (100.0000) acc1_neg: 31.250 acc5_neg: 100.000 time: 0.1957 data: 0.1001 max mem: 3531 +Test: [140/269] eta: 0:00:28 acc1_pos: 54.310 acc5_pos: 100.000 loss: 0.9550 (0.7984) acc1: 36.4583 (51.7213) acc5: 100.0000 (100.0000) acc1_neg: 32.552 acc5_neg: 100.000 time: 0.1849 data: 0.0924 max mem: 3531 +Test: [150/269] eta: 0:00:26 acc1_pos: 54.310 acc5_pos: 100.000 loss: 0.9448 (0.7984) acc1: 38.0208 (51.6349) acc5: 100.0000 (100.0000) acc1_neg: 67.708 acc5_neg: 100.000 time: 0.1996 data: 0.1078 max mem: 3531 +Test: [160/269] eta: 0:00:23 acc1_pos: 54.310 acc5_pos: 100.000 loss: 0.7325 (0.7991) acc1: 46.8750 (51.6418) acc5: 100.0000 (100.0000) acc1_neg: 42.708 acc5_neg: 100.000 time: 0.1949 data: 0.1032 max mem: 3531 +Test: [170/269] eta: 0:00:21 acc1_pos: 54.310 acc5_pos: 100.000 loss: 0.9314 (0.8111) acc1: 38.2812 (50.7645) acc5: 100.0000 (100.0000) acc1_neg: 35.677 acc5_neg: 100.000 time: 0.1738 data: 0.0823 max mem: 3531 +Test: [180/269] eta: 0:00:19 acc1_pos: 54.310 acc5_pos: 100.000 loss: 0.9630 (0.8156) acc1: 36.1979 (50.3971) acc5: 100.0000 (100.0000) acc1_neg: 40.625 acc5_neg: 100.000 time: 0.1899 data: 0.0951 max mem: 3531 +Test: [190/269] eta: 0:00:16 acc1_pos: 54.310 acc5_pos: 100.000 loss: 0.8539 (0.8189) acc1: 45.8333 (50.0095) acc5: 100.0000 (100.0000) acc1_neg: 48.698 acc5_neg: 100.000 time: 0.1899 data: 0.0944 max mem: 3531 +Test: [200/269] eta: 0:00:14 acc1_pos: 54.310 acc5_pos: 100.000 loss: 0.9230 (0.8287) acc1: 39.5833 (49.4338) acc5: 100.0000 (100.0000) acc1_neg: 64.062 acc5_neg: 100.000 time: 0.1799 data: 0.0868 max mem: 3531 +Test: [210/269] eta: 0:00:12 acc1_pos: 54.310 acc5_pos: 100.000 loss: 0.7982 (0.8240) acc1: 44.7917 (49.7655) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.1899 data: 0.0941 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 54.310 acc5_pos: 100.000 loss: 0.7187 (0.8195) acc1: 55.9896 (50.0542) acc5: 100.0000 (100.0000) acc1_neg: 71.354 acc5_neg: 100.000 time: 0.1899 data: 0.0933 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 54.310 acc5_pos: 100.000 loss: 0.6629 (0.8128) acc1: 57.0312 (50.5299) acc5: 100.0000 (100.0000) acc1_neg: 70.052 acc5_neg: 100.000 time: 0.1949 data: 0.1004 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 54.310 acc5_pos: 100.000 loss: 0.7559 (0.8159) acc1: 50.2604 (50.2734) acc5: 100.0000 (100.0000) acc1_neg: 56.250 acc5_neg: 100.000 time: 0.1849 data: 0.0863 max mem: 3532 +Test: [250/269] eta: 0:00:03 acc1_pos: 54.310 acc5_pos: 100.000 loss: 0.8294 (0.8199) acc1: 46.8750 (49.8485) acc5: 100.0000 (100.0000) acc1_neg: 20.052 acc5_neg: 100.000 time: 0.1599 data: 0.0589 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 54.310 acc5_pos: 100.000 loss: 1.0021 (0.8279) acc1: 26.5625 (49.0092) acc5: 100.0000 (100.0000) acc1_neg: 33.854 acc5_neg: 100.000 time: 0.1599 data: 0.0609 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 54.310 acc5_pos: 100.000 loss: 1.0021 (0.8307) acc1: 25.5208 (48.7216) acc5: 100.0000 (100.0000) acc1_neg: 55.381 acc5_neg: 100.000 time: 0.1579 data: 0.0578 max mem: 3532 +Test: Total time: 0:00:54 (0.2033 s / it) +* Overall Acc@1 48.722 Acc@5 100.000 + target == real Acc@1 69.470 Acc@5 100.000 + target == fake Acc@1 45.890 Acc@5 100.000 + loss 0.831 + PR-AUC 0.9141 ROC-AUC 0.6080 EER 42.55% (thr=0.4207) +Accuracy of the network on the 103293 test images: 48.7% +[rank0]:[W726 05:56:39.776256201 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T05:56:47Z] eval_done tag=fixed500k_fake_n16_seed8_resampled +[2026-07-26T05:56:48Z] eval_start tag=fixed500k_real_n8_seed9_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed9_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T05:56:48Z +hostname=n116-041-153 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed9_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n8_seed9_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 05:57:01.267000 1198431 torch/distributed/run.py:766] +W0726 05:57:01.267000 1198431 torch/distributed/run.py:766] ***************************************** +W0726 05:57:01.267000 1198431 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:57:01.267000 1198431 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:57:30.325477599 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:57:31.703198001 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:57:31.093895172 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:57:31.136427066 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:57:31.192485745 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:57:31.210668266 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:57:31.320265605 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:57:31.337200272 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n8_seed9_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed9_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:21:27 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1837 (0.1837) acc1: 100.0000 (100.0000) acc5: 100.0000 (100.0000) time: 5.5010 data: 3.8870 max mem: 3531 +Test: [ 10/234] eta: 0:02:10 acc1_pos: 97.656 acc5_pos: 100.000 loss: 0.2849 (0.3613) acc1: 94.5312 (88.2576) acc5: 100.0000 (100.0000) time: 0.5819 data: 0.3538 max mem: 3531 +Test: [ 20/234] eta: 0:01:14 acc1_pos: 48.698 acc5_pos: 100.000 loss: 0.3429 (0.4290) acc1: 89.8438 (83.4821) acc5: 100.0000 (100.0000) time: 0.0910 data: 0.0016 max mem: 3531 +Test: [ 30/234] eta: 0:00:58 acc1_pos: 59.375 acc5_pos: 100.000 loss: 0.8663 (0.6114) acc1: 50.2604 (71.0938) acc5: 100.0000 (100.0000) time: 0.1216 data: 0.0301 max mem: 3531 +Test: [ 40/234] eta: 0:00:54 acc1_pos: 2.865 acc5_pos: 100.000 loss: 1.0289 (0.7845) acc1: 40.3646 (60.8676) acc5: 100.0000 (100.0000) time: 0.2145 data: 0.1198 max mem: 3531 +Test: [ 50/234] eta: 0:00:50 acc1_pos: 1.302 acc5_pos: 100.000 loss: 1.9650 (1.0252) acc1: 3.3854 (49.5251) acc5: 100.0000 (100.0000) time: 0.2643 data: 0.1704 max mem: 3531 +Test: [ 60/234] eta: 0:00:46 acc1_pos: 5.517 acc5_pos: 100.000 loss: 0.8130 (0.8834) acc1: 5.2083 (57.2148) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2399 data: 0.1456 max mem: 3531 +Test: [ 70/234] eta: 0:00:42 acc1_pos: 5.517 acc5_pos: 100.000 loss: 0.0887 (0.7763) acc1: 100.0000 (63.1235) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2245 data: 0.1284 max mem: 3531 +Test: [ 80/234] eta: 0:00:38 acc1_pos: 5.517 acc5_pos: 100.000 loss: 0.1289 (0.6973) acc1: 98.9583 (67.5154) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2000 data: 0.1051 max mem: 3531 +Test: [ 90/234] eta: 0:00:35 acc1_pos: 5.517 acc5_pos: 100.000 loss: 0.1221 (0.6320) acc1: 99.4792 (71.0451) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2058 data: 0.1117 max mem: 3531 +Test: [100/234] eta: 0:00:33 acc1_pos: 5.517 acc5_pos: 100.000 loss: 0.1088 (0.5875) acc1: 99.2188 (73.4478) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2398 data: 0.1471 max mem: 3531 +Test: [110/234] eta: 0:00:30 acc1_pos: 5.517 acc5_pos: 100.000 loss: 0.1991 (0.5527) acc1: 95.0521 (75.3355) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2443 data: 0.1530 max mem: 3531 +Test: [120/234] eta: 0:00:27 acc1_pos: 5.517 acc5_pos: 100.000 loss: 0.2157 (0.5326) acc1: 94.7917 (76.4527) acc5: 100.0000 (100.0000) acc1_neg: 66.667 acc5_neg: 100.000 time: 0.2149 data: 0.1223 max mem: 3531 +Test: [130/234] eta: 0:00:25 acc1_pos: 5.517 acc5_pos: 100.000 loss: 0.4202 (0.5531) acc1: 76.8229 (74.4195) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.2046 data: 0.1111 max mem: 3531 +Test: [140/234] eta: 0:00:22 acc1_pos: 5.517 acc5_pos: 100.000 loss: 0.4202 (0.5350) acc1: 79.9479 (75.4968) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2060 data: 0.1136 max mem: 3531 +Test: [150/234] eta: 0:00:19 acc1_pos: 5.517 acc5_pos: 100.000 loss: 0.1824 (0.5099) acc1: 93.4896 (76.9057) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1849 data: 0.0930 max mem: 3531 +Test: [160/234] eta: 0:00:17 acc1_pos: 5.517 acc5_pos: 100.000 loss: 0.2436 (0.5104) acc1: 90.3646 (76.6175) acc5: 100.0000 (100.0000) acc1_neg: 55.729 acc5_neg: 100.000 time: 0.1953 data: 0.1013 max mem: 3531 +Test: [170/234] eta: 0:00:14 acc1_pos: 5.517 acc5_pos: 100.000 loss: 0.6591 (0.5219) acc1: 62.2396 (75.6092) acc5: 100.0000 (100.0000) acc1_neg: 58.333 acc5_neg: 100.000 time: 0.2022 data: 0.1076 max mem: 3531 +Test: [180/234] eta: 0:00:12 acc1_pos: 5.517 acc5_pos: 100.000 loss: 0.5898 (0.5093) acc1: 66.4062 (76.2992) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2008 data: 0.1080 max mem: 3531 +Test: [190/234] eta: 0:00:10 acc1_pos: 5.517 acc5_pos: 100.000 loss: 0.3007 (0.4984) acc1: 91.4062 (77.0765) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2036 data: 0.1103 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 5.517 acc5_pos: 100.000 loss: 0.2899 (0.4827) acc1: 91.6667 (78.0252) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1899 data: 0.0967 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 5.517 acc5_pos: 100.000 loss: 0.1097 (0.4648) acc1: 99.7396 (79.0605) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1511 data: 0.0602 max mem: 3532 +Test: [220/234] eta: 0:00:03 acc1_pos: 5.517 acc5_pos: 100.000 loss: 0.0936 (0.4480) acc1: 100.0000 (80.0080) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1166 data: 0.0271 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 5.517 acc5_pos: 100.000 loss: 0.1025 (0.4343) acc1: 99.2188 (80.7979) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1199 data: 0.0306 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 5.517 acc5_pos: 100.000 loss: 0.1281 (0.4304) acc1: 98.6979 (81.0061) acc5: 100.0000 (100.0000) acc1_neg: 98.444 acc5_neg: 100.000 time: 0.1210 data: 0.0305 max mem: 3532 +Test: Total time: 0:00:49 (0.2111 s / it) +* Overall Acc@1 81.006 Acc@5 100.000 + target == real Acc@1 49.202 Acc@5 100.000 + target == fake Acc@1 89.970 Acc@5 100.000 + loss 0.430 + PR-AUC 0.9301 ROC-AUC 0.8075 EER 27.78% (thr=0.8369) +Accuracy of the network on the 89729 test images: 81.0% +[rank0]:[W726 05:58:30.590573454 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:58:49.580000 1198977 torch/distributed/run.py:766] +W0726 05:58:49.580000 1198977 torch/distributed/run.py:766] ***************************************** +W0726 05:58:49.580000 1198977 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:58:49.580000 1198977 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:59:18.979067398 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:59:18.249181196 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:59:19.487781453 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:59:19.875616904 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:59:19.029052716 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +| distributed init (rank 6): env:// +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:59:19.050943432 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W726 05:59:19.050991921 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank6]:[W726 05:59:19.051123502 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n8_seed9_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed9_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:41:00 acc1_pos: 91.667 acc5_pos: 100.000 loss: 0.3522 (0.3522) acc1: 91.6667 (91.6667) acc5: 100.0000 (100.0000) time: 7.0633 data: 5.7080 max mem: 3531 +Test: [ 10/2387] eta: 0:28:40 acc1_pos: 72.917 acc5_pos: 100.000 loss: 0.5653 (0.5168) acc1: 75.7812 (77.8172) acc5: 100.0000 (100.0000) time: 0.7239 data: 0.5193 max mem: 3531 +Test: [ 20/2387] eta: 0:16:49 acc1_pos: 86.198 acc5_pos: 100.000 loss: 0.4819 (0.5032) acc1: 80.7292 (79.4643) acc5: 100.0000 (100.0000) time: 0.0948 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:13:33 acc1_pos: 68.229 acc5_pos: 100.000 loss: 0.5257 (0.5078) acc1: 78.9062 (78.4694) acc5: 100.0000 (100.0000) time: 0.1366 data: 0.0388 max mem: 3531 +Test: [ 40/2387] eta: 0:12:29 acc1_pos: 87.500 acc5_pos: 100.000 loss: 0.5259 (0.5084) acc1: 76.8229 (78.6141) acc5: 100.0000 (100.0000) time: 0.2064 data: 0.1129 max mem: 3531 +Test: [ 50/2387] eta: 0:12:34 acc1_pos: 79.688 acc5_pos: 100.000 loss: 0.5161 (0.5142) acc1: 77.8646 (78.1505) acc5: 100.0000 (100.0000) time: 0.2881 data: 0.1967 max mem: 3531 +Test: [ 60/2387] eta: 0:12:06 acc1_pos: 82.292 acc5_pos: 100.000 loss: 0.4916 (0.5020) acc1: 81.5104 (79.2606) acc5: 100.0000 (100.0000) time: 0.2985 data: 0.2058 max mem: 3531 +Test: [ 70/2387] eta: 0:11:46 acc1_pos: 84.115 acc5_pos: 100.000 loss: 0.4419 (0.4926) acc1: 83.8542 (80.0213) acc5: 100.0000 (100.0000) time: 0.2599 data: 0.1661 max mem: 3531 +Test: [ 80/2387] eta: 0:11:27 acc1_pos: 99.479 acc5_pos: 100.000 loss: 0.3480 (0.4625) acc1: 90.6250 (82.1084) acc5: 100.0000 (100.0000) time: 0.2549 data: 0.1599 max mem: 3531 +Test: [ 90/2387] eta: 0:11:12 acc1_pos: 99.219 acc5_pos: 100.000 loss: 0.1900 (0.4324) acc1: 99.4792 (84.0287) acc5: 100.0000 (100.0000) time: 0.2502 data: 0.1559 max mem: 3531 +Test: [ 100/2387] eta: 0:11:02 acc1_pos: 99.219 acc5_pos: 100.000 loss: 0.1902 (0.4111) acc1: 99.2188 (85.3909) acc5: 100.0000 (100.0000) time: 0.2549 data: 0.1605 max mem: 3531 +Test: [ 110/2387] eta: 0:10:53 acc1_pos: 95.573 acc5_pos: 100.000 loss: 0.2034 (0.3929) acc1: 99.2188 (86.5756) acc5: 100.0000 (100.0000) time: 0.2596 data: 0.1627 max mem: 3531 +Test: [ 120/2387] eta: 0:10:37 acc1_pos: 99.479 acc5_pos: 100.000 loss: 0.2097 (0.3789) acc1: 99.4792 (87.4806) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1434 max mem: 3531 +Test: [ 130/2387] eta: 0:10:28 acc1_pos: 98.698 acc5_pos: 100.000 loss: 0.2097 (0.3664) acc1: 99.4792 (88.2693) acc5: 100.0000 (100.0000) time: 0.2300 data: 0.1341 max mem: 3531 +Test: [ 140/2387] eta: 0:10:19 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.2085 (0.3556) acc1: 98.6979 (89.0256) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1449 max mem: 3531 +Test: [ 150/2387] eta: 0:10:09 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2048 (0.3454) acc1: 99.2188 (89.6954) acc5: 100.0000 (100.0000) time: 0.2348 data: 0.1419 max mem: 3531 +Test: [ 160/2387] eta: 0:09:59 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1876 (0.3343) acc1: 99.7396 (90.2336) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.2249 data: 0.1314 max mem: 3531 +Test: [ 170/2387] eta: 0:09:44 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1881 (0.3268) acc1: 93.4896 (90.3494) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1949 data: 0.1011 max mem: 3531 +Test: [ 180/2387] eta: 0:09:39 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2108 (0.3207) acc1: 91.4062 (90.4034) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2099 data: 0.1173 max mem: 3531 +Test: [ 190/2387] eta: 0:09:24 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2092 (0.3149) acc1: 91.4062 (90.4682) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1999 data: 0.1042 max mem: 3531 +Test: [ 200/2387] eta: 0:09:17 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2047 (0.3095) acc1: 92.1875 (90.5473) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1849 data: 0.0886 max mem: 3531 +Test: [ 210/2387] eta: 0:09:08 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2088 (0.3048) acc1: 91.9271 (90.6028) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.2074 data: 0.1150 max mem: 3532 +Test: [ 220/2387] eta: 0:09:00 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2107 (0.3006) acc1: 91.6667 (90.6368) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1949 data: 0.1008 max mem: 3532 +Test: [ 230/2387] eta: 0:08:52 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2076 (0.2965) acc1: 91.9271 (90.6983) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1924 data: 0.0967 max mem: 3532 +Test: [ 240/2387] eta: 0:08:49 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2135 (0.2990) acc1: 88.0208 (90.4424) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.2149 data: 0.1204 max mem: 3532 +Test: [ 250/2387] eta: 0:08:40 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2174 (0.2939) acc1: 85.1562 (90.6105) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2039 data: 0.1115 max mem: 3532 +Test: [ 260/2387] eta: 0:08:28 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1414 (0.2882) acc1: 96.8750 (90.8156) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1499 data: 0.0578 max mem: 3532 +Test: [ 270/2387] eta: 0:08:24 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1310 (0.2822) acc1: 96.8750 (91.0449) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1709 data: 0.0761 max mem: 3532 +Test: [ 280/2387] eta: 0:08:17 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1247 (0.2768) acc1: 97.1354 (91.2635) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1938 data: 0.0950 max mem: 3532 +Test: [ 290/2387] eta: 0:08:10 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1275 (0.2718) acc1: 97.1354 (91.4761) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1788 data: 0.0817 max mem: 3532 +Test: [ 300/2387] eta: 0:08:05 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1447 (0.2689) acc1: 96.3542 (91.5404) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1849 data: 0.0887 max mem: 3532 +Test: [ 310/2387] eta: 0:08:00 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1754 (0.2660) acc1: 95.5729 (91.6776) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1915 data: 0.0960 max mem: 3532 +Test: [ 320/2387] eta: 0:07:56 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1437 (0.2622) acc1: 99.7396 (91.9295) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2021 data: 0.1103 max mem: 3532 +Test: [ 330/2387] eta: 0:07:53 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1406 (0.2585) acc1: 99.7396 (92.1686) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2108 data: 0.1179 max mem: 3532 +Test: [ 340/2387] eta: 0:07:49 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1407 (0.2551) acc1: 99.7396 (92.3899) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2090 data: 0.1137 max mem: 3532 +Test: [ 350/2387] eta: 0:07:46 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1347 (0.2507) acc1: 100.0000 (92.6059) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2126 data: 0.1167 max mem: 3532 +Test: [ 360/2387] eta: 0:07:41 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1061 (0.2486) acc1: 99.7396 (92.6391) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1945 data: 0.1014 max mem: 3532 +Test: [ 370/2387] eta: 0:07:37 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1812 (0.2470) acc1: 93.2292 (92.6606) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1849 data: 0.0936 max mem: 3532 +Test: [ 380/2387] eta: 0:07:33 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1797 (0.2454) acc1: 93.2292 (92.6748) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1950 data: 0.1044 max mem: 3532 +Test: [ 390/2387] eta: 0:07:29 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1863 (0.2439) acc1: 93.2292 (92.6897) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1913 data: 0.1010 max mem: 3532 +Test: [ 400/2387] eta: 0:07:26 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1863 (0.2426) acc1: 93.2292 (92.6927) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2012 data: 0.1077 max mem: 3532 +Test: [ 410/2387] eta: 0:07:22 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1868 (0.2413) acc1: 92.9688 (92.7071) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.2002 data: 0.1062 max mem: 3532 +Test: [ 420/2387] eta: 0:07:19 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1855 (0.2399) acc1: 92.9688 (92.7213) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2027 data: 0.1110 max mem: 3532 +Test: [ 430/2387] eta: 0:07:16 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1931 (0.2412) acc1: 92.4479 (92.5923) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.2054 data: 0.1134 max mem: 3532 +Test: [ 440/2387] eta: 0:07:14 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2065 (0.2402) acc1: 89.5833 (92.6245) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2122 data: 0.1210 max mem: 3533 +Test: [ 450/2387] eta: 0:07:09 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.0975 (0.2370) acc1: 99.7396 (92.7840) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2026 data: 0.1100 max mem: 3533 +Test: [ 460/2387] eta: 0:07:06 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.0967 (0.2340) acc1: 100.0000 (92.9382) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1826 data: 0.0880 max mem: 3533 +Test: [ 470/2387] eta: 0:07:03 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.0968 (0.2311) acc1: 100.0000 (93.0860) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1959 data: 0.1017 max mem: 3533 +Test: [ 480/2387] eta: 0:07:02 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.0953 (0.2283) acc1: 99.7396 (93.2238) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2256 data: 0.1330 max mem: 3533 +Test: [ 490/2387] eta: 0:06:59 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.0969 (0.2256) acc1: 99.7396 (93.3586) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2254 data: 0.1343 max mem: 3533 +Test: [ 500/2387] eta: 0:06:55 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.0971 (0.2231) acc1: 100.0000 (93.4870) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1862 data: 0.0958 max mem: 3533 +Test: [ 510/2387] eta: 0:06:51 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.0990 (0.2208) acc1: 99.7396 (93.5961) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1762 data: 0.0857 max mem: 3533 +Test: [ 520/2387] eta: 0:06:48 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1250 (0.2192) acc1: 97.6562 (93.6685) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1888 data: 0.0966 max mem: 3533 +Test: [ 530/2387] eta: 0:06:45 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1307 (0.2175) acc1: 97.1354 (93.7343) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1941 data: 0.1004 max mem: 3533 +Test: [ 540/2387] eta: 0:06:42 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1301 (0.2159) acc1: 97.1354 (93.7977) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1911 data: 0.0980 max mem: 3533 +Test: [ 550/2387] eta: 0:06:39 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1301 (0.2144) acc1: 97.1354 (93.8578) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1892 data: 0.0971 max mem: 3533 +Test: [ 560/2387] eta: 0:06:35 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1303 (0.2129) acc1: 97.1354 (93.9102) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1874 data: 0.0936 max mem: 3533 +Test: [ 570/2387] eta: 0:06:33 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1286 (0.2115) acc1: 96.8750 (93.9630) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1936 data: 0.0976 max mem: 3533 +Test: [ 580/2387] eta: 0:06:29 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1289 (0.2101) acc1: 96.8750 (94.0122) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1866 data: 0.0930 max mem: 3533 +Test: [ 590/2387] eta: 0:06:26 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1363 (0.2089) acc1: 96.6146 (94.0518) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1807 data: 0.0898 max mem: 3533 +Test: [ 600/2387] eta: 0:06:24 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1385 (0.2077) acc1: 96.8750 (94.1075) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1982 data: 0.1053 max mem: 3533 +Test: [ 610/2387] eta: 0:06:21 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1196 (0.2061) acc1: 97.6562 (94.1741) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2056 data: 0.1126 max mem: 3533 +Test: [ 620/2387] eta: 0:06:18 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1088 (0.2045) acc1: 98.4375 (94.2390) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1909 data: 0.0995 max mem: 3533 +Test: [ 630/2387] eta: 0:06:15 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1131 (0.2031) acc1: 98.1771 (94.3006) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1746 data: 0.0833 max mem: 3533 +Test: [ 640/2387] eta: 0:06:11 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1131 (0.2017) acc1: 97.9167 (94.3517) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1660 data: 0.0724 max mem: 3533 +Test: [ 650/2387] eta: 0:06:07 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1126 (0.2003) acc1: 97.9167 (94.4096) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1513 data: 0.0558 max mem: 3533 +Test: [ 660/2387] eta: 0:06:04 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1093 (0.1990) acc1: 98.1771 (94.4678) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1634 data: 0.0709 max mem: 3533 +Test: [ 670/2387] eta: 0:06:01 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1105 (0.1976) acc1: 98.1771 (94.5227) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1728 data: 0.0801 max mem: 3534 +Test: [ 680/2387] eta: 0:05:58 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1083 (0.1963) acc1: 98.4375 (94.5940) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1644 data: 0.0690 max mem: 3534 +Test: [ 690/2387] eta: 0:05:56 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1099 (0.1967) acc1: 98.1771 (94.5723) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1844 data: 0.0911 max mem: 3534 +Test: [ 700/2387] eta: 0:05:53 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2627 (0.1976) acc1: 90.1042 (94.5119) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1923 data: 0.1004 max mem: 3534 +Test: [ 710/2387] eta: 0:05:51 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2628 (0.1986) acc1: 90.1042 (94.4543) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1957 data: 0.1036 max mem: 3534 +Test: [ 720/2387] eta: 0:05:48 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2620 (0.1994) acc1: 90.3646 (94.3980) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.1882 data: 0.0963 max mem: 3534 +Test: [ 730/2387] eta: 0:05:46 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2625 (0.2004) acc1: 90.1042 (94.3392) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1892 data: 0.0974 max mem: 3534 +Test: [ 740/2387] eta: 0:05:43 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2625 (0.2013) acc1: 90.1042 (94.2803) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.2055 data: 0.1144 max mem: 3534 +Test: [ 750/2387] eta: 0:05:41 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2619 (0.2021) acc1: 90.3646 (94.2299) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.2006 data: 0.1095 max mem: 3534 +Test: [ 760/2387] eta: 0:05:39 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2677 (0.2035) acc1: 90.1042 (94.1463) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.2073 data: 0.1137 max mem: 3534 +Test: [ 770/2387] eta: 0:05:38 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3146 (0.2052) acc1: 85.9375 (94.0148) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.2475 data: 0.1536 max mem: 3534 +Test: [ 780/2387] eta: 0:05:36 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3129 (0.2061) acc1: 86.1979 (93.9514) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.2414 data: 0.1486 max mem: 3534 +Test: [ 790/2387] eta: 0:05:34 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2551 (0.2070) acc1: 89.3229 (93.8853) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.2096 data: 0.1160 max mem: 3534 +Test: [ 800/2387] eta: 0:05:31 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2551 (0.2080) acc1: 89.3229 (93.8036) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.1885 data: 0.0954 max mem: 3534 +Test: [ 810/2387] eta: 0:05:29 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2693 (0.2091) acc1: 88.2812 (93.7217) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.1937 data: 0.1014 max mem: 3534 +Test: [ 820/2387] eta: 0:05:27 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3022 (0.2109) acc1: 85.1562 (93.5863) acc5: 100.0000 (100.0000) acc1_neg: 63.281 acc5_neg: 100.000 time: 0.2168 data: 0.1247 max mem: 3534 +Test: [ 830/2387] eta: 0:05:25 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.4923 (0.2164) acc1: 69.0104 (93.2242) acc5: 100.0000 (100.0000) acc1_neg: 64.844 acc5_neg: 100.000 time: 0.2111 data: 0.1171 max mem: 3534 +Test: [ 840/2387] eta: 0:05:24 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2250 (0.2152) acc1: 73.9583 (93.2877) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2242 data: 0.1296 max mem: 3534 +Test: [ 850/2387] eta: 0:05:21 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1012 (0.2139) acc1: 99.2188 (93.3531) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2076 data: 0.1122 max mem: 3534 +Test: [ 860/2387] eta: 0:05:19 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.0940 (0.2125) acc1: 99.4792 (93.4258) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1927 data: 0.0989 max mem: 3534 +Test: [ 870/2387] eta: 0:05:17 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.0899 (0.2111) acc1: 99.7396 (93.4968) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2075 data: 0.1171 max mem: 3534 +Test: [ 880/2387] eta: 0:05:15 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.0909 (0.2098) acc1: 99.7396 (93.5647) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2174 data: 0.1257 max mem: 3534 +Test: [ 890/2387] eta: 0:05:13 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1076 (0.2092) acc1: 98.4375 (93.6036) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2192 data: 0.1275 max mem: 3534 +Test: [ 900/2387] eta: 0:05:11 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1347 (0.2084) acc1: 98.4375 (93.6662) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2147 data: 0.1222 max mem: 3535 +Test: [ 910/2387] eta: 0:05:09 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1266 (0.2074) acc1: 99.4792 (93.7329) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2191 data: 0.1263 max mem: 3535 +Test: [ 920/2387] eta: 0:05:07 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1230 (0.2065) acc1: 99.7396 (93.7978) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2151 data: 0.1238 max mem: 3535 +Test: [ 930/2387] eta: 0:05:05 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1233 (0.2056) acc1: 99.4792 (93.8594) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2030 data: 0.1105 max mem: 3535 +Test: [ 940/2387] eta: 0:05:02 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1262 (0.2048) acc1: 99.4792 (93.9205) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1970 data: 0.1049 max mem: 3535 +Test: [ 950/2387] eta: 0:05:00 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1262 (0.2040) acc1: 99.4792 (93.9795) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0893 max mem: 3535 +Test: [ 960/2387] eta: 0:04:57 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1235 (0.2031) acc1: 99.4792 (94.0378) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1735 data: 0.0828 max mem: 3535 +Test: [ 970/2387] eta: 0:04:54 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1253 (0.2023) acc1: 99.4792 (94.0936) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1660 data: 0.0746 max mem: 3535 +Test: [ 980/2387] eta: 0:04:52 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1260 (0.2027) acc1: 99.2188 (94.0667) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1570 data: 0.0650 max mem: 3535 +Test: [ 990/2387] eta: 0:04:49 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2605 (0.2044) acc1: 88.5417 (93.9581) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1557 data: 0.0651 max mem: 3535 +Test: [1000/2387] eta: 0:04:46 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2945 (0.2053) acc1: 87.5000 (93.9019) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.1520 data: 0.0621 max mem: 3535 +Test: [1010/2387] eta: 0:04:43 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3216 (0.2079) acc1: 86.4583 (93.7340) acc5: 100.0000 (100.0000) acc1_neg: 68.490 acc5_neg: 100.000 time: 0.1548 data: 0.0645 max mem: 3535 +Test: [1020/2387] eta: 0:04:40 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.4071 (0.2102) acc1: 76.3021 (93.5697) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1570 data: 0.0659 max mem: 3535 +Test: [1030/2387] eta: 0:04:38 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2779 (0.2107) acc1: 88.5417 (93.5275) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1564 data: 0.0642 max mem: 3535 +Test: [1040/2387] eta: 0:04:35 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2141 (0.2106) acc1: 91.1458 (93.5286) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1530 data: 0.0608 max mem: 3535 +Test: [1050/2387] eta: 0:04:32 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2052 (0.2106) acc1: 92.9688 (93.5230) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1617 data: 0.0702 max mem: 3535 +Test: [1060/2387] eta: 0:04:30 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2337 (0.2109) acc1: 91.9271 (93.4999) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1646 data: 0.0727 max mem: 3535 +Test: [1070/2387] eta: 0:04:27 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2457 (0.2123) acc1: 89.8438 (93.4040) acc5: 100.0000 (100.0000) acc1_neg: 61.198 acc5_neg: 100.000 time: 0.1599 data: 0.0661 max mem: 3535 +Test: [1080/2387] eta: 0:04:25 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.5458 (0.2164) acc1: 66.9271 (93.1405) acc5: 100.0000 (100.0000) acc1_neg: 76.042 acc5_neg: 100.000 time: 0.1528 data: 0.0590 max mem: 3535 +Test: [1090/2387] eta: 0:04:22 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3780 (0.2173) acc1: 76.0417 (93.0757) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.1564 data: 0.0641 max mem: 3535 +Test: [1100/2387] eta: 0:04:20 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3037 (0.2183) acc1: 85.9375 (92.9986) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.1720 data: 0.0798 max mem: 3535 +Test: [1110/2387] eta: 0:04:17 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3098 (0.2192) acc1: 83.8542 (92.9237) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.1734 data: 0.0814 max mem: 3535 +Test: [1120/2387] eta: 0:04:15 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3116 (0.2203) acc1: 83.8542 (92.8368) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1780 data: 0.0857 max mem: 3536 +Test: [1130/2387] eta: 0:04:13 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3241 (0.2212) acc1: 84.1146 (92.7583) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1799 data: 0.0875 max mem: 3536 +Test: [1140/2387] eta: 0:04:11 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3933 (0.2247) acc1: 76.8229 (92.5353) acc5: 100.0000 (100.0000) acc1_neg: 63.542 acc5_neg: 100.000 time: 0.1838 data: 0.0917 max mem: 3536 +Test: [1150/2387] eta: 0:04:09 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.5667 (0.2253) acc1: 66.9271 (92.4821) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2066 data: 0.1152 max mem: 3536 +Test: [1160/2387] eta: 0:04:07 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.0740 (0.2240) acc1: 100.0000 (92.5464) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2029 data: 0.1110 max mem: 3536 +Test: [1170/2387] eta: 0:04:04 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.0727 (0.2227) acc1: 100.0000 (92.6080) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1832 data: 0.0913 max mem: 3536 +Test: [1180/2387] eta: 0:04:02 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.0726 (0.2215) acc1: 100.0000 (92.6689) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1861 data: 0.0943 max mem: 3536 +Test: [1190/2387] eta: 0:04:00 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.0729 (0.2202) acc1: 100.0000 (92.7287) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1899 data: 0.0953 max mem: 3536 +Test: [1200/2387] eta: 0:03:58 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.0769 (0.2191) acc1: 100.0000 (92.7862) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1822 data: 0.0885 max mem: 3536 +Test: [1210/2387] eta: 0:03:56 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.0920 (0.2181) acc1: 99.4792 (92.8371) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1799 data: 0.0877 max mem: 3536 +Test: [1220/2387] eta: 0:03:54 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1153 (0.2175) acc1: 98.1771 (92.8645) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1774 data: 0.0855 max mem: 3536 +Test: [1230/2387] eta: 0:03:51 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1217 (0.2167) acc1: 97.6562 (92.9080) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1830 data: 0.0915 max mem: 3536 +Test: [1240/2387] eta: 0:03:49 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1132 (0.2159) acc1: 98.4375 (92.9473) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1938 data: 0.1005 max mem: 3536 +Test: [1250/2387] eta: 0:03:47 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1163 (0.2151) acc1: 98.4375 (92.9879) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1847 data: 0.0921 max mem: 3536 +Test: [1260/2387] eta: 0:03:45 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1145 (0.2143) acc1: 98.4375 (93.0297) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1809 data: 0.0900 max mem: 3536 +Test: [1270/2387] eta: 0:03:43 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1286 (0.2143) acc1: 92.9688 (93.0134) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.1833 data: 0.0912 max mem: 3536 +Test: [1280/2387] eta: 0:03:41 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2049 (0.2142) acc1: 92.9688 (93.0212) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1722 data: 0.0800 max mem: 3536 +Test: [1290/2387] eta: 0:03:38 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2013 (0.2143) acc1: 94.0104 (93.0273) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1696 data: 0.0785 max mem: 3536 +Test: [1300/2387] eta: 0:03:36 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2312 (0.2144) acc1: 92.1875 (93.0166) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1767 data: 0.0823 max mem: 3536 +Test: [1310/2387] eta: 0:03:34 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2377 (0.2145) acc1: 91.1458 (93.0063) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1795 data: 0.0844 max mem: 3536 +Test: [1320/2387] eta: 0:03:32 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2200 (0.2145) acc1: 93.7500 (93.0172) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.2059 data: 0.1129 max mem: 3536 +Test: [1330/2387] eta: 0:03:30 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2379 (0.2167) acc1: 90.8854 (92.8698) acc5: 100.0000 (100.0000) acc1_neg: 67.969 acc5_neg: 100.000 time: 0.1993 data: 0.1071 max mem: 3536 +Test: [1340/2387] eta: 0:03:28 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.4897 (0.2181) acc1: 75.0000 (92.7761) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1603 data: 0.0685 max mem: 3536 +Test: [1350/2387] eta: 0:03:26 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.0945 (0.2172) acc1: 98.9583 (92.8263) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1673 data: 0.0759 max mem: 3537 +Test: [1360/2387] eta: 0:03:23 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.0789 (0.2161) acc1: 100.0000 (92.8790) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1711 data: 0.0792 max mem: 3537 +Test: [1370/2387] eta: 0:03:21 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.0748 (0.2151) acc1: 100.0000 (92.9289) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1549 data: 0.0621 max mem: 3537 +Test: [1380/2387] eta: 0:03:19 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.0749 (0.2141) acc1: 100.0000 (92.9799) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1539 data: 0.0620 max mem: 3537 +Test: [1390/2387] eta: 0:03:16 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.0737 (0.2131) acc1: 100.0000 (93.0298) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1640 data: 0.0731 max mem: 3537 +Test: [1400/2387] eta: 0:03:14 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.0794 (0.2124) acc1: 99.7396 (93.0714) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1684 data: 0.0766 max mem: 3537 +Test: [1410/2387] eta: 0:03:12 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1253 (0.2129) acc1: 97.6562 (93.0337) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.1635 data: 0.0700 max mem: 3537 +Test: [1420/2387] eta: 0:03:10 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3248 (0.2138) acc1: 83.8542 (92.9557) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1610 data: 0.0677 max mem: 3537 +Test: [1430/2387] eta: 0:03:08 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3348 (0.2148) acc1: 82.2917 (92.8741) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1638 data: 0.0714 max mem: 3537 +Test: [1440/2387] eta: 0:03:05 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3496 (0.2161) acc1: 80.7292 (92.7748) acc5: 100.0000 (100.0000) acc1_neg: 58.594 acc5_neg: 100.000 time: 0.1589 data: 0.0640 max mem: 3537 +Test: [1450/2387] eta: 0:03:03 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.4391 (0.2184) acc1: 71.0938 (92.6303) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1461 data: 0.0504 max mem: 3537 +Test: [1460/2387] eta: 0:03:01 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2154 (0.2182) acc1: 91.1458 (92.6335) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1502 data: 0.0563 max mem: 3537 +Test: [1470/2387] eta: 0:02:59 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1985 (0.2181) acc1: 92.9688 (92.6338) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1649 data: 0.0685 max mem: 3537 +Test: [1480/2387] eta: 0:02:56 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1828 (0.2179) acc1: 93.4896 (92.6394) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1512 data: 0.0562 max mem: 3537 +Test: [1490/2387] eta: 0:02:54 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1765 (0.2176) acc1: 94.5312 (92.6570) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1374 data: 0.0442 max mem: 3537 +Test: [1500/2387] eta: 0:02:52 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1543 (0.2172) acc1: 95.3125 (92.6780) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1557 data: 0.0609 max mem: 3537 +Test: [1510/2387] eta: 0:02:50 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1457 (0.2168) acc1: 97.1354 (92.7139) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1624 data: 0.0666 max mem: 3537 +Test: [1520/2387] eta: 0:02:48 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1227 (0.2160) acc1: 99.2188 (92.7571) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1576 data: 0.0595 max mem: 3537 +Test: [1530/2387] eta: 0:02:46 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.0947 (0.2152) acc1: 99.4792 (92.8000) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1549 data: 0.0592 max mem: 3537 +Test: [1540/2387] eta: 0:02:44 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.0909 (0.2144) acc1: 99.4792 (92.8444) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1599 data: 0.0645 max mem: 3537 +Test: [1550/2387] eta: 0:02:41 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.0866 (0.2135) acc1: 99.7396 (92.8888) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0711 max mem: 3537 +Test: [1560/2387] eta: 0:02:39 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.0866 (0.2127) acc1: 99.7396 (92.9317) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1499 data: 0.0536 max mem: 3537 +Test: [1570/2387] eta: 0:02:37 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.0955 (0.2122) acc1: 99.2188 (92.9666) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1375 data: 0.0435 max mem: 3537 +Test: [1580/2387] eta: 0:02:35 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1726 (0.2124) acc1: 95.5729 (92.9534) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1999 data: 0.1029 max mem: 3538 +Test: [1590/2387] eta: 0:02:34 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2698 (0.2129) acc1: 89.0625 (92.9224) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.2324 data: 0.1357 max mem: 3538 +Test: [1600/2387] eta: 0:02:31 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2409 (0.2130) acc1: 91.1458 (92.9117) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1799 data: 0.0840 max mem: 3538 +Test: [1610/2387] eta: 0:02:29 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2351 (0.2134) acc1: 90.6250 (92.8855) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.1710 data: 0.0772 max mem: 3538 +Test: [1620/2387] eta: 0:02:28 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2751 (0.2138) acc1: 87.5000 (92.8467) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1962 data: 0.1037 max mem: 3538 +Test: [1630/2387] eta: 0:02:26 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3135 (0.2154) acc1: 82.5521 (92.7392) acc5: 100.0000 (100.0000) acc1_neg: 70.573 acc5_neg: 100.000 time: 0.2022 data: 0.1077 max mem: 3538 +Test: [1640/2387] eta: 0:02:24 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.5667 (0.2178) acc1: 67.9688 (92.5703) acc5: 100.0000 (100.0000) acc1_neg: 69.271 acc5_neg: 100.000 time: 0.2013 data: 0.1091 max mem: 3538 +Test: [1650/2387] eta: 0:02:22 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.5868 (0.2201) acc1: 66.6667 (92.4154) acc5: 100.0000 (100.0000) acc1_neg: 69.271 acc5_neg: 100.000 time: 0.1981 data: 0.1080 max mem: 3538 +Test: [1660/2387] eta: 0:02:20 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.5964 (0.2224) acc1: 66.9271 (92.2631) acc5: 100.0000 (100.0000) acc1_neg: 73.958 acc5_neg: 100.000 time: 0.1939 data: 0.1018 max mem: 3538 +Test: [1670/2387] eta: 0:02:18 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.5841 (0.2244) acc1: 67.7083 (92.1238) acc5: 100.0000 (100.0000) acc1_neg: 72.135 acc5_neg: 100.000 time: 0.1910 data: 0.0985 max mem: 3538 +Test: [1680/2387] eta: 0:02:16 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.5441 (0.2265) acc1: 68.7500 (91.9768) acc5: 100.0000 (100.0000) acc1_neg: 63.542 acc5_neg: 100.000 time: 0.1859 data: 0.0948 max mem: 3538 +Test: [1690/2387] eta: 0:02:14 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.5609 (0.2286) acc1: 69.5312 (91.8371) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.1834 data: 0.0911 max mem: 3538 +Test: [1700/2387] eta: 0:02:12 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.5719 (0.2300) acc1: 69.5312 (91.7444) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1883 data: 0.0960 max mem: 3538 +Test: [1710/2387] eta: 0:02:10 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1632 (0.2295) acc1: 93.2292 (91.7688) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1889 data: 0.0965 max mem: 3538 +Test: [1720/2387] eta: 0:02:08 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1375 (0.2290) acc1: 96.3542 (91.7898) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1863 data: 0.0938 max mem: 3538 +Test: [1730/2387] eta: 0:02:06 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1309 (0.2284) acc1: 96.3542 (91.8186) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1846 data: 0.0931 max mem: 3538 +Test: [1740/2387] eta: 0:02:04 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1321 (0.2279) acc1: 97.1354 (91.8492) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1839 data: 0.0921 max mem: 3538 +Test: [1750/2387] eta: 0:02:02 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1356 (0.2274) acc1: 96.6146 (91.8759) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1771 data: 0.0857 max mem: 3538 +Test: [1760/2387] eta: 0:02:00 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1733 (0.2277) acc1: 94.0104 (91.8586) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1783 data: 0.0865 max mem: 3538 +Test: [1770/2387] eta: 0:01:58 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2394 (0.2277) acc1: 89.5833 (91.8497) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1804 data: 0.0873 max mem: 3538 +Test: [1780/2387] eta: 0:01:56 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2000 (0.2276) acc1: 90.6250 (91.8446) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1749 data: 0.0829 max mem: 3538 +Test: [1790/2387] eta: 0:01:54 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1751 (0.2273) acc1: 93.4896 (91.8611) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1794 data: 0.0855 max mem: 3538 +Test: [1800/2387] eta: 0:01:52 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1637 (0.2270) acc1: 94.0104 (91.8682) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1907 data: 0.0958 max mem: 3538 +Test: [1810/2387] eta: 0:01:51 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1718 (0.2267) acc1: 95.0521 (91.8861) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1926 data: 0.0997 max mem: 3539 +Test: [1820/2387] eta: 0:01:49 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1767 (0.2269) acc1: 93.2292 (91.8716) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1852 data: 0.0917 max mem: 3539 +Test: [1830/2387] eta: 0:01:47 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2463 (0.2271) acc1: 88.5417 (91.8582) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1912 data: 0.0978 max mem: 3539 +Test: [1840/2387] eta: 0:01:45 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1425 (0.2266) acc1: 97.3958 (91.8909) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1937 data: 0.1022 max mem: 3539 +Test: [1850/2387] eta: 0:01:43 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1412 (0.2261) acc1: 97.6562 (91.9219) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1816 data: 0.0889 max mem: 3539 +Test: [1860/2387] eta: 0:01:41 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1425 (0.2257) acc1: 97.6562 (91.9531) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1822 data: 0.0889 max mem: 3539 +Test: [1870/2387] eta: 0:01:39 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1414 (0.2252) acc1: 97.9167 (91.9860) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1818 data: 0.0907 max mem: 3539 +Test: [1880/2387] eta: 0:01:37 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1392 (0.2248) acc1: 98.1771 (92.0176) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1812 data: 0.0890 max mem: 3539 +Test: [1890/2387] eta: 0:01:35 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1399 (0.2243) acc1: 97.9167 (92.0494) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1878 data: 0.0946 max mem: 3539 +Test: [1900/2387] eta: 0:01:33 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1432 (0.2240) acc1: 97.6562 (92.0753) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1879 data: 0.0953 max mem: 3539 +Test: [1910/2387] eta: 0:01:31 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2745 (0.2245) acc1: 91.1458 (92.0529) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.1824 data: 0.0906 max mem: 3539 +Test: [1920/2387] eta: 0:01:29 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3126 (0.2254) acc1: 87.5000 (91.9783) acc5: 100.0000 (100.0000) acc1_neg: 77.083 acc5_neg: 100.000 time: 0.1741 data: 0.0834 max mem: 3539 +Test: [1930/2387] eta: 0:01:27 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3189 (0.2258) acc1: 84.8958 (91.9519) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.1799 data: 0.0886 max mem: 3539 +Test: [1940/2387] eta: 0:01:25 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3189 (0.2264) acc1: 84.8958 (91.8874) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.1798 data: 0.0882 max mem: 3539 +Test: [1950/2387] eta: 0:01:23 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3365 (0.2272) acc1: 80.9896 (91.8331) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1724 data: 0.0817 max mem: 3539 +Test: [1960/2387] eta: 0:01:21 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3472 (0.2279) acc1: 80.9896 (91.7732) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.1790 data: 0.0885 max mem: 3539 +Test: [1970/2387] eta: 0:01:19 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.4716 (0.2302) acc1: 74.4792 (91.6240) acc5: 100.0000 (100.0000) acc1_neg: 56.771 acc5_neg: 100.000 time: 0.1720 data: 0.0802 max mem: 3539 +Test: [1980/2387] eta: 0:01:17 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.5813 (0.2310) acc1: 65.6250 (91.5757) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0769 max mem: 3539 +Test: [1990/2387] eta: 0:01:15 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.0766 (0.2302) acc1: 100.0000 (91.6180) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1733 data: 0.0792 max mem: 3539 +Test: [2000/2387] eta: 0:01:14 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.0763 (0.2295) acc1: 100.0000 (91.6598) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1921 data: 0.0982 max mem: 3539 +Test: [2010/2387] eta: 0:01:12 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.0763 (0.2287) acc1: 100.0000 (91.7007) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1970 data: 0.1044 max mem: 3539 +Test: [2020/2387] eta: 0:01:10 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.0772 (0.2280) acc1: 100.0000 (91.7414) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1910 data: 0.0991 max mem: 3539 +Test: [2030/2387] eta: 0:01:08 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.0770 (0.2272) acc1: 100.0000 (91.7817) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2044 data: 0.1124 max mem: 3540 +Test: [2040/2387] eta: 0:01:06 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.0766 (0.2265) acc1: 100.0000 (91.8216) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1983 data: 0.1061 max mem: 3540 +Test: [2050/2387] eta: 0:01:04 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.0775 (0.2258) acc1: 100.0000 (91.8608) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2001 data: 0.1090 max mem: 3540 +Test: [2060/2387] eta: 0:01:02 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.0882 (0.2251) acc1: 99.7396 (91.8988) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2042 data: 0.1127 max mem: 3540 +Test: [2070/2387] eta: 0:01:00 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.0895 (0.2245) acc1: 99.7396 (91.9371) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2000 data: 0.1083 max mem: 3540 +Test: [2080/2387] eta: 0:00:58 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.0893 (0.2238) acc1: 100.0000 (91.9758) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2143 data: 0.1222 max mem: 3540 +Test: [2090/2387] eta: 0:00:56 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.0890 (0.2232) acc1: 100.0000 (92.0129) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2113 data: 0.1181 max mem: 3540 +Test: [2100/2387] eta: 0:00:55 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.0895 (0.2225) acc1: 99.7396 (92.0500) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1969 data: 0.1048 max mem: 3540 +Test: [2110/2387] eta: 0:00:53 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.0899 (0.2219) acc1: 99.7396 (92.0861) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2011 data: 0.1085 max mem: 3540 +Test: [2120/2387] eta: 0:00:51 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.0939 (0.2226) acc1: 99.2188 (92.0418) acc5: 100.0000 (100.0000) acc1_neg: 72.396 acc5_neg: 100.000 time: 0.1956 data: 0.1033 max mem: 3540 +Test: [2130/2387] eta: 0:00:49 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3968 (0.2233) acc1: 80.4688 (91.9886) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1981 data: 0.1053 max mem: 3540 +Test: [2140/2387] eta: 0:00:47 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3313 (0.2238) acc1: 83.0729 (91.9523) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.2075 data: 0.1139 max mem: 3540 +Test: [2150/2387] eta: 0:00:45 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3279 (0.2246) acc1: 82.8125 (91.8955) acc5: 100.0000 (100.0000) acc1_neg: 76.302 acc5_neg: 100.000 time: 0.2059 data: 0.1147 max mem: 3540 +Test: [2160/2387] eta: 0:00:43 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3779 (0.2252) acc1: 79.4271 (91.8441) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.2142 data: 0.1229 max mem: 3540 +Test: [2170/2387] eta: 0:00:41 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.4547 (0.2271) acc1: 73.6979 (91.7253) acc5: 100.0000 (100.0000) acc1_neg: 58.854 acc5_neg: 100.000 time: 0.2231 data: 0.1317 max mem: 3540 +Test: [2180/2387] eta: 0:00:39 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.6512 (0.2286) acc1: 64.8438 (91.6363) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.2021 data: 0.1113 max mem: 3540 +Test: [2190/2387] eta: 0:00:37 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2904 (0.2281) acc1: 90.8854 (91.6664) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1946 data: 0.1026 max mem: 3540 +Test: [2200/2387] eta: 0:00:35 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1254 (0.2278) acc1: 98.6979 (91.6903) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1984 data: 0.1055 max mem: 3540 +Test: [2210/2387] eta: 0:00:34 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1562 (0.2275) acc1: 96.6146 (91.7124) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1999 data: 0.1065 max mem: 3540 +Test: [2220/2387] eta: 0:00:32 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1522 (0.2272) acc1: 96.8750 (91.7337) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2042 data: 0.1118 max mem: 3540 +Test: [2230/2387] eta: 0:00:30 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1398 (0.2268) acc1: 97.6562 (91.7590) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1982 data: 0.1009 max mem: 3540 +Test: [2240/2387] eta: 0:00:28 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1357 (0.2265) acc1: 98.4375 (91.7844) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2019 data: 0.1040 max mem: 3540 +Test: [2250/2387] eta: 0:00:26 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1400 (0.2261) acc1: 98.6979 (91.8116) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2048 data: 0.1123 max mem: 3540 +Test: [2260/2387] eta: 0:00:24 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1782 (0.2263) acc1: 95.3125 (91.8040) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1880 data: 0.0963 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2834 (0.2267) acc1: 88.5417 (91.7750) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1732 data: 0.0819 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2793 (0.2269) acc1: 88.2812 (91.7601) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1751 data: 0.0835 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2707 (0.2273) acc1: 87.2396 (91.7290) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.1754 data: 0.0842 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3094 (0.2277) acc1: 84.3750 (91.6944) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1678 data: 0.0768 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3847 (0.2289) acc1: 75.7812 (91.6075) acc5: 100.0000 (100.0000) acc1_neg: 65.625 acc5_neg: 100.000 time: 0.1600 data: 0.0699 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.5260 (0.2303) acc1: 68.4896 (91.5078) acc5: 100.0000 (100.0000) acc1_neg: 77.083 acc5_neg: 100.000 time: 0.1548 data: 0.0652 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.4671 (0.2311) acc1: 73.6979 (91.4551) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.1508 data: 0.0612 max mem: 3541 +Test: [2340/2387] eta: 0:00:09 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.4036 (0.2319) acc1: 78.9062 (91.3971) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.1471 data: 0.0576 max mem: 3541 +Test: [2350/2387] eta: 0:00:07 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3932 (0.2325) acc1: 78.9062 (91.3514) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.1434 data: 0.0534 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.4062 (0.2335) acc1: 77.3438 (91.2848) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.1405 data: 0.0504 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.4241 (0.2344) acc1: 76.0417 (91.2149) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.1401 data: 0.0501 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.5571 (0.2363) acc1: 67.1875 (91.0941) acc5: 100.0000 (100.0000) acc1_neg: 66.146 acc5_neg: 100.000 time: 0.1382 data: 0.0483 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.5771 (0.2366) acc1: 65.6250 (91.0714) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1295 data: 0.0426 max mem: 3541 +Test: Total time: 0:07:34 (0.1906 s / it) +* Overall Acc@1 91.071 Acc@5 100.000 + target == real Acc@1 90.055 Acc@5 100.000 + target == fake Acc@1 91.143 Acc@5 100.000 + loss 0.237 + PR-AUC 0.9976 ROC-AUC 0.9672 EER 9.19% (thr=0.5137) +Accuracy of the network on the 916267 test images: 91.1% +[rank0]:[W726 06:07:10.290891585 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 06:07:30.737000 1199607 torch/distributed/run.py:766] +W0726 06:07:30.737000 1199607 torch/distributed/run.py:766] ***************************************** +W0726 06:07:30.737000 1199607 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:07:30.737000 1199607 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:08:00.680424637 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +[rank2]:[W726 06:08:00.079464585 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:08:00.096173938 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:08:00.439553407 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +[rank0]:[W726 06:08:00.467383744 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:08:00.476074030 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:08:01.559363719 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:08:01.675920271 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n8_seed9_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed9_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 0:57:06 acc1_pos: 91.667 acc5_pos: 100.000 loss: 0.3522 (0.3522) acc1: 91.6667 (91.6667) acc5: 100.0000 (100.0000) time: 6.6275 data: 4.3101 max mem: 3531 +Test: [ 10/517] eta: 0:05:46 acc1_pos: 72.917 acc5_pos: 100.000 loss: 0.5653 (0.5168) acc1: 75.7812 (77.8172) acc5: 100.0000 (100.0000) time: 0.6843 data: 0.3923 max mem: 3531 +Test: [ 20/517] eta: 0:03:26 acc1_pos: 86.198 acc5_pos: 100.000 loss: 0.4819 (0.5032) acc1: 80.7292 (79.4643) acc5: 100.0000 (100.0000) time: 0.1056 data: 0.0144 max mem: 3531 +Test: [ 30/517] eta: 0:02:47 acc1_pos: 68.229 acc5_pos: 100.000 loss: 0.5257 (0.5078) acc1: 78.9062 (78.4694) acc5: 100.0000 (100.0000) time: 0.1583 data: 0.0647 max mem: 3531 +Test: [ 40/517] eta: 0:02:47 acc1_pos: 87.500 acc5_pos: 100.000 loss: 0.5259 (0.5084) acc1: 76.8229 (78.6141) acc5: 100.0000 (100.0000) time: 0.2849 data: 0.1887 max mem: 3531 +Test: [ 50/517] eta: 0:02:37 acc1_pos: 98.624 acc5_pos: 100.000 loss: 0.5669 (0.5357) acc1: 74.2188 (75.5004) acc5: 100.0000 (100.0000) acc1_neg: 17.969 acc5_neg: 100.000 time: 0.3240 data: 0.2277 max mem: 3531 +Test: [ 60/517] eta: 0:02:25 acc1_pos: 98.624 acc5_pos: 100.000 loss: 0.7955 (0.5927) acc1: 47.1354 (69.1470) acc5: 100.0000 (100.0000) acc1_neg: 47.135 acc5_neg: 100.000 time: 0.2519 data: 0.1597 max mem: 3531 +Test: [ 70/517] eta: 0:02:19 acc1_pos: 98.624 acc5_pos: 100.000 loss: 0.6994 (0.5992) acc1: 51.8229 (67.8991) acc5: 100.0000 (100.0000) acc1_neg: 64.583 acc5_neg: 100.000 time: 0.2481 data: 0.1532 max mem: 3531 +Test: [ 80/517] eta: 0:02:12 acc1_pos: 98.624 acc5_pos: 100.000 loss: 0.6570 (0.6042) acc1: 58.0729 (67.0525) acc5: 100.0000 (100.0000) acc1_neg: 46.354 acc5_neg: 100.000 time: 0.2543 data: 0.1591 max mem: 3531 +Test: [ 90/517] eta: 0:02:06 acc1_pos: 98.624 acc5_pos: 100.000 loss: 0.7128 (0.6166) acc1: 54.1667 (65.7967) acc5: 100.0000 (100.0000) acc1_neg: 53.125 acc5_neg: 100.000 time: 0.2463 data: 0.1537 max mem: 3531 +Test: [100/517] eta: 0:02:01 acc1_pos: 98.624 acc5_pos: 100.000 loss: 0.7548 (0.6360) acc1: 43.4896 (63.1652) acc5: 100.0000 (100.0000) acc1_neg: 54.948 acc5_neg: 100.000 time: 0.2453 data: 0.1525 max mem: 3531 +Test: [110/517] eta: 0:01:57 acc1_pos: 98.624 acc5_pos: 100.000 loss: 0.7466 (0.6393) acc1: 48.9583 (62.7581) acc5: 100.0000 (100.0000) acc1_neg: 58.333 acc5_neg: 100.000 time: 0.2484 data: 0.1509 max mem: 3531 +Test: [120/517] eta: 0:01:52 acc1_pos: 98.624 acc5_pos: 100.000 loss: 0.6794 (0.6464) acc1: 58.0729 (62.1965) acc5: 100.0000 (100.0000) acc1_neg: 51.562 acc5_neg: 100.000 time: 0.2446 data: 0.1465 max mem: 3531 +Test: [130/517] eta: 0:01:49 acc1_pos: 98.624 acc5_pos: 100.000 loss: 0.6928 (0.6513) acc1: 57.0312 (61.6770) acc5: 100.0000 (100.0000) acc1_neg: 58.073 acc5_neg: 100.000 time: 0.2559 data: 0.1636 max mem: 3531 +Test: [140/517] eta: 0:01:46 acc1_pos: 98.624 acc5_pos: 100.000 loss: 0.7807 (0.6672) acc1: 47.6562 (59.9956) acc5: 100.0000 (100.0000) acc1_neg: 39.323 acc5_neg: 100.000 time: 0.2733 data: 0.1815 max mem: 3531 +Test: [150/517] eta: 0:01:42 acc1_pos: 98.624 acc5_pos: 100.000 loss: 0.8123 (0.6723) acc1: 41.9271 (59.4388) acc5: 100.0000 (100.0000) acc1_neg: 51.823 acc5_neg: 100.000 time: 0.2439 data: 0.1520 max mem: 3531 +Test: [160/517] eta: 0:01:39 acc1_pos: 98.624 acc5_pos: 100.000 loss: 0.7211 (0.6775) acc1: 45.0521 (58.5792) acc5: 100.0000 (100.0000) acc1_neg: 58.594 acc5_neg: 100.000 time: 0.2449 data: 0.1505 max mem: 3531 +Test: [170/517] eta: 0:01:35 acc1_pos: 98.624 acc5_pos: 100.000 loss: 0.6815 (0.6790) acc1: 51.3021 (58.2283) acc5: 100.0000 (100.0000) acc1_neg: 55.990 acc5_neg: 100.000 time: 0.2549 data: 0.1556 max mem: 3531 +Test: [180/517] eta: 0:01:32 acc1_pos: 98.624 acc5_pos: 100.000 loss: 0.6714 (0.6780) acc1: 53.9062 (58.3146) acc5: 100.0000 (100.0000) acc1_neg: 70.312 acc5_neg: 100.000 time: 0.2449 data: 0.1470 max mem: 3531 +Test: [190/517] eta: 0:01:28 acc1_pos: 98.624 acc5_pos: 100.000 loss: 0.4760 (0.6634) acc1: 78.3854 (59.7827) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.2403 data: 0.1476 max mem: 3531 +Test: [200/517] eta: 0:01:25 acc1_pos: 98.624 acc5_pos: 100.000 loss: 0.4578 (0.6672) acc1: 78.3854 (59.6458) acc5: 100.0000 (100.0000) acc1_neg: 43.490 acc5_neg: 100.000 time: 0.2352 data: 0.1438 max mem: 3531 +Test: [210/517] eta: 0:01:22 acc1_pos: 98.624 acc5_pos: 100.000 loss: 0.6973 (0.6696) acc1: 52.8646 (59.2738) acc5: 100.0000 (100.0000) acc1_neg: 45.573 acc5_neg: 100.000 time: 0.2357 data: 0.1432 max mem: 3532 +Test: [220/517] eta: 0:01:19 acc1_pos: 98.624 acc5_pos: 100.000 loss: 0.7066 (0.6721) acc1: 50.7812 (58.9355) acc5: 100.0000 (100.0000) acc1_neg: 41.667 acc5_neg: 100.000 time: 0.2453 data: 0.1518 max mem: 3532 +Test: [230/517] eta: 0:01:16 acc1_pos: 98.624 acc5_pos: 100.000 loss: 0.7821 (0.6802) acc1: 41.6667 (57.8689) acc5: 100.0000 (100.0000) acc1_neg: 52.604 acc5_neg: 100.000 time: 0.2537 data: 0.1588 max mem: 3532 +Test: [240/517] eta: 0:01:13 acc1_pos: 98.624 acc5_pos: 100.000 loss: 0.7821 (0.6782) acc1: 40.6250 (58.0848) acc5: 100.0000 (100.0000) acc1_neg: 34.635 acc5_neg: 100.000 time: 0.2392 data: 0.1427 max mem: 3532 +Test: [250/517] eta: 0:01:10 acc1_pos: 98.624 acc5_pos: 100.000 loss: 0.7562 (0.6829) acc1: 48.6979 (57.4763) acc5: 100.0000 (100.0000) acc1_neg: 39.583 acc5_neg: 100.000 time: 0.2402 data: 0.1467 max mem: 3532 +Test: [260/517] eta: 0:01:08 acc1_pos: 98.624 acc5_pos: 100.000 loss: 0.7773 (0.6881) acc1: 40.3646 (56.8267) acc5: 100.0000 (100.0000) acc1_neg: 64.844 acc5_neg: 100.000 time: 0.2549 data: 0.1633 max mem: 3532 +Test: [270/517] eta: 0:01:05 acc1_pos: 98.624 acc5_pos: 100.000 loss: 0.7313 (0.6901) acc1: 48.4375 (56.5344) acc5: 100.0000 (100.0000) acc1_neg: 54.948 acc5_neg: 100.000 time: 0.2446 data: 0.1472 max mem: 3532 +Test: [280/517] eta: 0:01:02 acc1_pos: 98.624 acc5_pos: 100.000 loss: 0.7313 (0.6936) acc1: 44.5312 (56.0776) acc5: 100.0000 (100.0000) acc1_neg: 30.990 acc5_neg: 100.000 time: 0.2349 data: 0.1358 max mem: 3532 +Test: [290/517] eta: 0:00:59 acc1_pos: 98.624 acc5_pos: 100.000 loss: 0.8011 (0.6976) acc1: 40.1042 (55.4893) acc5: 100.0000 (100.0000) acc1_neg: 40.104 acc5_neg: 100.000 time: 0.2399 data: 0.1429 max mem: 3532 +Test: [300/517] eta: 0:00:56 acc1_pos: 98.624 acc5_pos: 100.000 loss: 0.7169 (0.6966) acc1: 47.3958 (55.4281) acc5: 100.0000 (100.0000) acc1_neg: 33.854 acc5_neg: 100.000 time: 0.2353 data: 0.1396 max mem: 3532 +Test: [310/517] eta: 0:00:53 acc1_pos: 98.624 acc5_pos: 100.000 loss: 0.7770 (0.7048) acc1: 33.8542 (54.5418) acc5: 100.0000 (100.0000) acc1_neg: 25.521 acc5_neg: 100.000 time: 0.2449 data: 0.1515 max mem: 3532 +Test: [320/517] eta: 0:00:51 acc1_pos: 98.624 acc5_pos: 100.000 loss: 0.9065 (0.7083) acc1: 31.2500 (54.2007) acc5: 100.0000 (100.0000) acc1_neg: 42.188 acc5_neg: 100.000 time: 0.2499 data: 0.1575 max mem: 3532 +Test: [330/517] eta: 0:00:48 acc1_pos: 98.624 acc5_pos: 100.000 loss: 0.7394 (0.7079) acc1: 50.0000 (54.2099) acc5: 100.0000 (100.0000) acc1_neg: 55.990 acc5_neg: 100.000 time: 0.2401 data: 0.1497 max mem: 3532 +Test: [340/517] eta: 0:00:45 acc1_pos: 98.624 acc5_pos: 100.000 loss: 0.6759 (0.7072) acc1: 55.9896 (54.2759) acc5: 100.0000 (100.0000) acc1_neg: 39.323 acc5_neg: 100.000 time: 0.2410 data: 0.1490 max mem: 3532 +Test: [350/517] eta: 0:00:43 acc1_pos: 98.624 acc5_pos: 100.000 loss: 0.8222 (0.7146) acc1: 39.3229 (53.6087) acc5: 100.0000 (100.0000) acc1_neg: 16.667 acc5_neg: 100.000 time: 0.2448 data: 0.1500 max mem: 3532 +Test: [360/517] eta: 0:00:40 acc1_pos: 98.624 acc5_pos: 100.000 loss: 0.9146 (0.7184) acc1: 32.8125 (53.3198) acc5: 100.0000 (100.0000) acc1_neg: 51.823 acc5_neg: 100.000 time: 0.2583 data: 0.1625 max mem: 3532 +Test: [370/517] eta: 0:00:37 acc1_pos: 98.624 acc5_pos: 100.000 loss: 0.5806 (0.7115) acc1: 68.4896 (54.0129) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.2498 data: 0.1534 max mem: 3532 +Test: [380/517] eta: 0:00:35 acc1_pos: 98.624 acc5_pos: 100.000 loss: 0.4452 (0.7047) acc1: 80.9896 (54.7941) acc5: 100.0000 (100.0000) acc1_neg: 69.271 acc5_neg: 100.000 time: 0.2299 data: 0.1353 max mem: 3532 +Test: [390/517] eta: 0:00:32 acc1_pos: 98.624 acc5_pos: 100.000 loss: 0.4734 (0.6998) acc1: 78.3854 (55.3495) acc5: 100.0000 (100.0000) acc1_neg: 65.104 acc5_neg: 100.000 time: 0.2366 data: 0.1419 max mem: 3532 +Test: [400/517] eta: 0:00:30 acc1_pos: 98.624 acc5_pos: 100.000 loss: 0.5018 (0.6954) acc1: 76.5625 (55.7662) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.2449 data: 0.1497 max mem: 3532 +Test: [410/517] eta: 0:00:27 acc1_pos: 98.624 acc5_pos: 100.000 loss: 0.5882 (0.6964) acc1: 54.9479 (55.7247) acc5: 100.0000 (100.0000) acc1_neg: 26.562 acc5_neg: 100.000 time: 0.2488 data: 0.1566 max mem: 3532 +Test: [420/517] eta: 0:00:24 acc1_pos: 98.624 acc5_pos: 100.000 loss: 0.6995 (0.6965) acc1: 51.0417 (55.6853) acc5: 100.0000 (100.0000) acc1_neg: 45.573 acc5_neg: 100.000 time: 0.2409 data: 0.1505 max mem: 3532 +Test: [430/517] eta: 0:00:22 acc1_pos: 98.624 acc5_pos: 100.000 loss: 0.7584 (0.6993) acc1: 45.5729 (55.3636) acc5: 100.0000 (100.0000) acc1_neg: 31.250 acc5_neg: 100.000 time: 0.2443 data: 0.1520 max mem: 3532 +Test: [440/517] eta: 0:00:19 acc1_pos: 98.624 acc5_pos: 100.000 loss: 0.7593 (0.6996) acc1: 45.0521 (55.3436) acc5: 100.0000 (100.0000) acc1_neg: 55.208 acc5_neg: 100.000 time: 0.2540 data: 0.1580 max mem: 3533 +Test: [450/517] eta: 0:00:17 acc1_pos: 98.624 acc5_pos: 100.000 loss: 0.6740 (0.6986) acc1: 55.2083 (55.3873) acc5: 100.0000 (100.0000) acc1_neg: 70.833 acc5_neg: 100.000 time: 0.2649 data: 0.1691 max mem: 3533 +Test: [460/517] eta: 0:00:14 acc1_pos: 98.624 acc5_pos: 100.000 loss: 0.6924 (0.6998) acc1: 48.1771 (55.2326) acc5: 100.0000 (100.0000) acc1_neg: 35.677 acc5_neg: 100.000 time: 0.2599 data: 0.1650 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 98.624 acc5_pos: 100.000 loss: 0.6624 (0.6980) acc1: 55.9896 (55.3770) acc5: 100.0000 (100.0000) acc1_neg: 71.354 acc5_neg: 100.000 time: 0.2349 data: 0.1380 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 98.624 acc5_pos: 100.000 loss: 0.5791 (0.6945) acc1: 61.9792 (55.7427) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2349 data: 0.1389 max mem: 3533 +Test: [490/517] eta: 0:00:06 acc1_pos: 98.624 acc5_pos: 100.000 loss: 0.4852 (0.6903) acc1: 74.7396 (56.1816) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.2499 data: 0.1559 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 98.624 acc5_pos: 100.000 loss: 0.5890 (0.6901) acc1: 63.0208 (56.1450) acc5: 100.0000 (100.0000) acc1_neg: 52.865 acc5_neg: 100.000 time: 0.2699 data: 0.1765 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 98.624 acc5_pos: 100.000 loss: 0.6582 (0.6911) acc1: 53.1250 (56.1073) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.2203 data: 0.1285 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 98.624 acc5_pos: 100.000 loss: 0.7230 (0.6906) acc1: 52.8646 (56.1557) acc5: 100.0000 (100.0000) acc1_neg: 31.095 acc5_neg: 100.000 time: 0.1713 data: 0.0783 max mem: 3533 +Test: Total time: 0:02:11 (0.2542 s / it) +* Overall Acc@1 56.156 Acc@5 100.000 + target == real Acc@1 77.941 Acc@5 100.000 + target == fake Acc@1 53.896 Acc@5 100.000 + loss 0.691 + PR-AUC 0.9580 ROC-AUC 0.7345 EER 33.11% (thr=0.4209) +Accuracy of the network on the 198427 test images: 56.2% +[rank0]:[W726 06:10:26.045892336 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 06:10:46.028000 1200179 torch/distributed/run.py:766] +W0726 06:10:46.028000 1200179 torch/distributed/run.py:766] ***************************************** +W0726 06:10:46.028000 1200179 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:10:46.028000 1200179 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:11:13.677764880 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:11:13.058573485 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:11:13.111612194 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:11:14.533501628 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 06:11:14.537286605 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:11:14.722574744 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:11:14.749225342 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:11:14.808425546 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n8_seed9_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed9_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:35:08 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.4790 (2.4790) acc1: 0.0000 (0.0000) acc5: 100.0000 (100.0000) time: 5.4211 data: 3.5635 max mem: 3531 +Test: [ 10/389] eta: 0:03:38 acc1_pos: 1.042 acc5_pos: 100.000 loss: 2.5675 (2.4992) acc1: 0.0000 (0.3078) acc5: 100.0000 (100.0000) time: 0.5758 data: 0.3243 max mem: 3531 +Test: [ 20/389] eta: 0:02:12 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.5675 (2.5093) acc1: 0.0000 (0.1984) acc5: 100.0000 (100.0000) time: 0.1063 data: 0.0145 max mem: 3531 +Test: [ 30/389] eta: 0:01:38 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.6335 (2.5479) acc1: 0.0000 (0.1344) acc5: 100.0000 (100.0000) time: 0.1106 data: 0.0146 max mem: 3531 +Test: [ 40/389] eta: 0:01:28 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.5832 (2.5592) acc1: 0.0000 (0.1016) acc5: 100.0000 (100.0000) time: 0.1420 data: 0.0461 max mem: 3531 +Test: [ 50/389] eta: 0:01:20 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.5412 (2.5316) acc1: 0.0000 (0.0817) acc5: 100.0000 (100.0000) time: 0.1799 data: 0.0860 max mem: 3531 +Test: [ 60/389] eta: 0:01:14 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.4392 (2.5320) acc1: 0.0000 (0.0726) acc5: 100.0000 (100.0000) time: 0.1704 data: 0.0734 max mem: 3531 +Test: [ 70/389] eta: 0:01:08 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.5361 (2.5354) acc1: 0.0000 (0.0697) acc5: 100.0000 (100.0000) time: 0.1577 data: 0.0634 max mem: 3531 +Test: [ 80/389] eta: 0:01:05 acc1_pos: 0.781 acc5_pos: 100.000 loss: 2.5008 (2.5123) acc1: 0.0000 (0.0836) acc5: 100.0000 (100.0000) time: 0.1623 data: 0.0718 max mem: 3531 +Test: [ 90/389] eta: 0:00:59 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.3636 (2.5000) acc1: 0.0000 (0.1259) acc5: 100.0000 (100.0000) time: 0.1452 data: 0.0538 max mem: 3531 +Test: [100/389] eta: 0:00:56 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.4210 (2.4953) acc1: 0.0000 (0.1160) acc5: 100.0000 (100.0000) time: 0.1306 data: 0.0394 max mem: 3531 +Test: [110/389] eta: 0:00:53 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.4348 (2.4858) acc1: 0.0000 (0.1150) acc5: 100.0000 (100.0000) time: 0.1444 data: 0.0519 max mem: 3531 +Test: [120/389] eta: 0:00:49 acc1_pos: 0.260 acc5_pos: 100.000 loss: 2.3968 (2.4774) acc1: 0.0000 (0.1141) acc5: 100.0000 (100.0000) time: 0.1375 data: 0.0441 max mem: 3531 +Test: [130/389] eta: 0:00:47 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.1860 (2.3397) acc1: 0.2604 (5.8862) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1506 data: 0.0592 max mem: 3531 +Test: [140/389] eta: 0:00:44 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0680 (2.1801) acc1: 100.0000 (12.5185) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1446 data: 0.0543 max mem: 3531 +Test: [150/389] eta: 0:00:42 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0680 (2.0410) acc1: 100.0000 (18.3119) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1393 data: 0.0487 max mem: 3531 +Test: [160/389] eta: 0:00:40 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0708 (1.9202) acc1: 100.0000 (23.3631) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1632 data: 0.0721 max mem: 3531 +Test: [170/389] eta: 0:00:38 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0722 (1.8123) acc1: 100.0000 (27.8448) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1608 data: 0.0687 max mem: 3531 +Test: [180/389] eta: 0:00:36 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0721 (1.7162) acc1: 100.0000 (31.8255) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1598 data: 0.0681 max mem: 3531 +Test: [190/389] eta: 0:00:34 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0710 (1.6316) acc1: 100.0000 (35.3744) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1545 data: 0.0644 max mem: 3531 +Test: [200/389] eta: 0:00:32 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0752 (1.5539) acc1: 100.0000 (38.5896) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1601 data: 0.0691 max mem: 3531 +Test: [210/389] eta: 0:00:31 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0666 (1.4835) acc1: 100.0000 (41.5000) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1740 data: 0.0818 max mem: 3532 +Test: [220/389] eta: 0:00:29 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0673 (1.4193) acc1: 100.0000 (44.1471) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1813 data: 0.0892 max mem: 3532 +Test: [230/389] eta: 0:00:27 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0679 (1.3613) acc1: 100.0000 (46.5650) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1755 data: 0.0838 max mem: 3532 +Test: [240/389] eta: 0:00:25 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0754 (1.3084) acc1: 100.0000 (48.7800) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1677 data: 0.0770 max mem: 3532 +Test: [250/389] eta: 0:00:24 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0729 (1.2592) acc1: 100.0000 (50.8196) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1778 data: 0.0869 max mem: 3532 +Test: [260/389] eta: 0:00:22 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0697 (1.2145) acc1: 100.0000 (52.7010) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1864 data: 0.0940 max mem: 3532 +Test: [270/389] eta: 0:00:20 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0747 (1.1725) acc1: 100.0000 (54.4463) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1804 data: 0.0883 max mem: 3532 +Test: [280/389] eta: 0:00:19 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0700 (1.1334) acc1: 100.0000 (56.0674) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1701 data: 0.0787 max mem: 3532 +Test: [290/389] eta: 0:00:17 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0724 (1.0972) acc1: 100.0000 (57.5762) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1737 data: 0.0822 max mem: 3532 +Test: [300/389] eta: 0:00:15 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0734 (1.0634) acc1: 100.0000 (58.9848) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1682 data: 0.0775 max mem: 3532 +Test: [310/389] eta: 0:00:13 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0739 (1.0317) acc1: 100.0000 (60.3036) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1676 data: 0.0769 max mem: 3532 +Test: [320/389] eta: 0:00:12 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0742 (1.0019) acc1: 100.0000 (61.5403) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1743 data: 0.0837 max mem: 3532 +Test: [330/389] eta: 0:00:10 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0703 (0.9737) acc1: 100.0000 (62.7022) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1691 data: 0.0788 max mem: 3532 +Test: [340/389] eta: 0:00:08 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0690 (0.9472) acc1: 100.0000 (63.7960) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1634 data: 0.0719 max mem: 3532 +Test: [350/389] eta: 0:00:06 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0700 (0.9227) acc1: 100.0000 (64.8208) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1669 data: 0.0752 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0771 (0.8993) acc1: 100.0000 (65.7952) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1671 data: 0.0763 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0770 (0.8772) acc1: 100.0000 (66.7172) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1648 data: 0.0739 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0809 (0.8562) acc1: 100.0000 (67.5908) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1597 data: 0.0688 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0733 (0.8401) acc1: 100.0000 (68.2535) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1594 data: 0.0642 max mem: 3532 +Test: Total time: 0:01:07 (0.1734 s / it) +* Overall Acc@1 68.253 Acc@5 100.000 + target == real Acc@1 0.112 Acc@5 100.000 + target == fake Acc@1 99.927 Acc@5 100.000 + loss 0.840 + PR-AUC 0.8068 ROC-AUC 0.6902 EER 35.30% (thr=0.9277) +Accuracy of the network on the 149358 test images: 68.3% +[rank0]:[W726 06:12:36.978040341 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 06:12:55.260000 1200726 torch/distributed/run.py:766] +W0726 06:12:55.260000 1200726 torch/distributed/run.py:766] ***************************************** +W0726 06:12:55.260000 1200726 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:12:55.260000 1200726 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:13:22.435136589 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:13:22.448348285 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 06:13:22.460203762 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:13:23.783266992 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:13:23.834443581 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:13:23.892574226 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:13:23.958286673 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W726 06:13:23.958291139 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n8_seed9_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed9_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:41:39 acc1_pos: 26.042 acc5_pos: 100.000 loss: 1.2251 (1.2251) acc1: 26.0417 (26.0417) acc5: 100.0000 (100.0000) time: 7.5055 data: 5.3387 max mem: 3531 +Test: [ 10/333] eta: 0:04:07 acc1_pos: 24.740 acc5_pos: 100.000 loss: 1.1007 (1.0625) acc1: 34.1146 (36.5767) acc5: 100.0000 (100.0000) time: 0.7665 data: 0.4856 max mem: 3531 +Test: [ 20/333] eta: 0:02:20 acc1_pos: 16.667 acc5_pos: 100.000 loss: 1.0993 (1.1009) acc1: 33.8542 (34.3130) acc5: 100.0000 (100.0000) time: 0.0963 data: 0.0003 max mem: 3531 +Test: [ 30/333] eta: 0:01:55 acc1_pos: 87.760 acc5_pos: 100.000 loss: 0.8279 (0.8877) acc1: 38.2812 (49.8404) acc5: 100.0000 (100.0000) time: 0.1699 data: 0.0710 max mem: 3531 +Test: [ 40/333] eta: 0:01:40 acc1_pos: 83.854 acc5_pos: 100.000 loss: 0.3999 (0.7660) acc1: 84.6354 (58.5874) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1351 max mem: 3531 +Test: [ 50/333] eta: 0:01:28 acc1_pos: 85.938 acc5_pos: 100.000 loss: 0.3620 (0.6846) acc1: 86.9792 (64.8080) acc5: 100.0000 (100.0000) time: 0.2099 data: 0.1165 max mem: 3531 +Test: [ 60/333] eta: 0:01:20 acc1_pos: 87.500 acc5_pos: 100.000 loss: 0.3557 (0.6311) acc1: 90.1042 (68.7628) acc5: 100.0000 (100.0000) time: 0.1949 data: 0.1010 max mem: 3531 +Test: [ 70/333] eta: 0:01:13 acc1_pos: 76.823 acc5_pos: 100.000 loss: 0.3557 (0.5956) acc1: 88.2812 (71.3285) acc5: 100.0000 (100.0000) time: 0.1952 data: 0.1022 max mem: 3531 +Test: [ 80/333] eta: 0:01:08 acc1_pos: 92.188 acc5_pos: 100.000 loss: 0.3609 (0.5688) acc1: 88.2812 (73.3025) acc5: 100.0000 (100.0000) time: 0.1949 data: 0.1018 max mem: 3531 +Test: [ 90/333] eta: 0:01:03 acc1_pos: 91.667 acc5_pos: 100.000 loss: 0.3625 (0.5472) acc1: 88.2812 (74.8998) acc5: 100.0000 (100.0000) time: 0.1846 data: 0.0903 max mem: 3531 +Test: [100/333] eta: 0:00:59 acc1_pos: 89.323 acc5_pos: 100.000 loss: 0.3779 (0.5339) acc1: 85.1562 (75.8019) acc5: 100.0000 (100.0000) time: 0.2054 data: 0.1116 max mem: 3531 +Test: [110/333] eta: 0:00:55 acc1_pos: 90.885 acc5_pos: 100.000 loss: 0.3740 (0.5191) acc1: 86.9792 (76.8440) acc5: 100.0000 (100.0000) time: 0.1903 data: 0.0960 max mem: 3531 +Test: [120/333] eta: 0:00:51 acc1_pos: 92.448 acc5_pos: 100.000 loss: 0.3693 (0.5082) acc1: 86.7188 (77.5805) acc5: 100.0000 (100.0000) time: 0.1710 data: 0.0769 max mem: 3531 +Test: [130/333] eta: 0:00:49 acc1_pos: 94.792 acc5_pos: 100.000 loss: 0.3769 (0.4982) acc1: 86.1979 (78.2562) acc5: 100.0000 (100.0000) time: 0.2145 data: 0.1226 max mem: 3531 +Test: [140/333] eta: 0:00:45 acc1_pos: 78.125 acc5_pos: 100.000 loss: 0.3696 (0.4895) acc1: 87.5000 (78.9413) acc5: 100.0000 (100.0000) time: 0.2059 data: 0.1124 max mem: 3531 +Test: [150/333] eta: 0:00:42 acc1_pos: 81.771 acc5_pos: 100.000 loss: 0.3609 (0.4831) acc1: 89.3229 (79.4202) acc5: 100.0000 (100.0000) time: 0.1816 data: 0.0887 max mem: 3531 +Test: [160/333] eta: 0:00:40 acc1_pos: 77.632 acc5_pos: 100.000 loss: 0.4100 (0.4974) acc1: 82.5521 (78.2900) acc5: 100.0000 (100.0000) acc1_neg: 23.698 acc5_neg: 100.000 time: 0.1930 data: 0.1002 max mem: 3531 +Test: [170/333] eta: 0:00:37 acc1_pos: 77.632 acc5_pos: 100.000 loss: 1.0662 (0.5323) acc1: 30.2083 (75.3640) acc5: 100.0000 (100.0000) acc1_neg: 33.333 acc5_neg: 100.000 time: 0.2010 data: 0.1086 max mem: 3531 +Test: [180/333] eta: 0:00:34 acc1_pos: 77.632 acc5_pos: 100.000 loss: 1.0470 (0.5588) acc1: 30.2083 (72.9555) acc5: 100.0000 (100.0000) acc1_neg: 27.344 acc5_neg: 100.000 time: 0.1998 data: 0.1078 max mem: 3531 +Test: [190/333] eta: 0:00:32 acc1_pos: 77.632 acc5_pos: 100.000 loss: 0.9968 (0.5819) acc1: 30.9896 (70.8020) acc5: 100.0000 (100.0000) acc1_neg: 25.000 acc5_neg: 100.000 time: 0.1981 data: 0.1059 max mem: 3531 +Test: [200/333] eta: 0:00:30 acc1_pos: 77.632 acc5_pos: 100.000 loss: 0.9753 (0.6002) acc1: 35.9375 (69.1322) acc5: 100.0000 (100.0000) acc1_neg: 39.323 acc5_neg: 100.000 time: 0.2017 data: 0.1109 max mem: 3531 +Test: [210/333] eta: 0:00:27 acc1_pos: 77.632 acc5_pos: 100.000 loss: 0.9524 (0.6172) acc1: 36.9792 (67.6158) acc5: 100.0000 (100.0000) acc1_neg: 48.438 acc5_neg: 100.000 time: 0.2093 data: 0.1184 max mem: 3532 +Test: [220/333] eta: 0:00:25 acc1_pos: 77.632 acc5_pos: 100.000 loss: 0.9483 (0.6341) acc1: 36.7188 (66.0292) acc5: 100.0000 (100.0000) acc1_neg: 39.844 acc5_neg: 100.000 time: 0.2138 data: 0.1229 max mem: 3532 +Test: [230/333] eta: 0:00:23 acc1_pos: 77.632 acc5_pos: 100.000 loss: 0.8683 (0.6434) acc1: 43.2292 (65.2203) acc5: 100.0000 (100.0000) acc1_neg: 43.229 acc5_neg: 100.000 time: 0.2105 data: 0.1198 max mem: 3532 +Test: [240/333] eta: 0:00:20 acc1_pos: 77.632 acc5_pos: 100.000 loss: 0.9065 (0.6662) acc1: 37.5000 (63.3990) acc5: 100.0000 (100.0000) acc1_neg: 17.969 acc5_neg: 100.000 time: 0.2035 data: 0.1115 max mem: 3532 +Test: [250/333] eta: 0:00:18 acc1_pos: 77.632 acc5_pos: 100.000 loss: 1.2041 (0.6830) acc1: 18.4896 (61.9740) acc5: 100.0000 (100.0000) acc1_neg: 44.792 acc5_neg: 100.000 time: 0.2164 data: 0.1234 max mem: 3532 +Test: [260/333] eta: 0:00:16 acc1_pos: 77.632 acc5_pos: 100.000 loss: 1.0285 (0.6964) acc1: 30.9896 (60.8218) acc5: 100.0000 (100.0000) acc1_neg: 33.854 acc5_neg: 100.000 time: 0.2122 data: 0.1200 max mem: 3532 +Test: [270/333] eta: 0:00:13 acc1_pos: 77.632 acc5_pos: 100.000 loss: 0.9695 (0.7047) acc1: 35.4167 (60.0851) acc5: 100.0000 (100.0000) acc1_neg: 40.625 acc5_neg: 100.000 time: 0.1986 data: 0.1050 max mem: 3532 +Test: [280/333] eta: 0:00:11 acc1_pos: 77.632 acc5_pos: 100.000 loss: 0.9491 (0.7184) acc1: 36.4583 (59.0395) acc5: 100.0000 (100.0000) acc1_neg: 29.688 acc5_neg: 100.000 time: 0.2106 data: 0.1170 max mem: 3532 +Test: [290/333] eta: 0:00:09 acc1_pos: 77.632 acc5_pos: 100.000 loss: 1.1503 (0.7345) acc1: 25.2604 (57.8599) acc5: 100.0000 (100.0000) acc1_neg: 18.750 acc5_neg: 100.000 time: 0.1923 data: 0.1008 max mem: 3532 +Test: [300/333] eta: 0:00:07 acc1_pos: 77.632 acc5_pos: 100.000 loss: 1.1474 (0.7479) acc1: 25.0000 (56.7726) acc5: 100.0000 (100.0000) acc1_neg: 23.438 acc5_neg: 100.000 time: 0.1574 data: 0.0666 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 77.632 acc5_pos: 100.000 loss: 1.1282 (0.7547) acc1: 25.7812 (56.1621) acc5: 100.0000 (100.0000) acc1_neg: 23.177 acc5_neg: 100.000 time: 0.1385 data: 0.0485 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 77.632 acc5_pos: 100.000 loss: 1.1195 (0.7661) acc1: 23.9583 (55.2035) acc5: 100.0000 (100.0000) acc1_neg: 42.448 acc5_neg: 100.000 time: 0.1273 data: 0.0375 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 77.632 acc5_pos: 100.000 loss: 0.9898 (0.7720) acc1: 33.3333 (54.6309) acc5: 100.0000 (100.0000) acc1_neg: 33.333 acc5_neg: 100.000 time: 0.1186 data: 0.0288 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 77.632 acc5_pos: 100.000 loss: 0.9755 (0.7731) acc1: 35.9375 (54.5278) acc5: 100.0000 (100.0000) acc1_neg: 37.465 acc5_neg: 100.000 time: 0.1173 data: 0.0248 max mem: 3532 +Test: Total time: 0:01:09 (0.2085 s / it) +* Overall Acc@1 54.528 Acc@5 100.000 + target == real Acc@1 79.685 Acc@5 100.000 + target == fake Acc@1 32.133 Acc@5 100.000 + loss 0.773 + PR-AUC 0.6230 ROC-AUC 0.6293 EER 40.76% (thr=0.3018) +Accuracy of the network on the 127843 test images: 54.5% +[rank0]:[W726 06:14:40.876930477 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 06:14:59.842000 1201286 torch/distributed/run.py:766] +W0726 06:14:59.842000 1201286 torch/distributed/run.py:766] ***************************************** +W0726 06:14:59.842000 1201286 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:14:59.842000 1201286 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 06:15:27.642333000 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:15:27.981760146 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:15:27.292642923 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank4]:[W726 06:15:27.292660463 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:15:27.396020960 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:15:27.476706797 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:15:28.604791091 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:15:28.636020384 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n8_seed9_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed9_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:21:56 acc1_pos: 50.521 acc5_pos: 100.000 loss: 0.8455 (0.8455) acc1: 50.5208 (50.5208) acc5: 100.0000 (100.0000) time: 5.6003 data: 4.0059 max mem: 3531 +Test: [ 10/235] eta: 0:02:12 acc1_pos: 54.688 acc5_pos: 100.000 loss: 0.8021 (0.8090) acc1: 55.9896 (55.6108) acc5: 100.0000 (100.0000) time: 0.5908 data: 0.3645 max mem: 3531 +Test: [ 20/235] eta: 0:01:15 acc1_pos: 50.260 acc5_pos: 100.000 loss: 0.8015 (0.8096) acc1: 55.2083 (55.2703) acc5: 100.0000 (100.0000) time: 0.0899 data: 0.0004 max mem: 3531 +Test: [ 30/235] eta: 0:00:58 acc1_pos: 52.604 acc5_pos: 100.000 loss: 0.7998 (0.8050) acc1: 55.2083 (55.6116) acc5: 100.0000 (100.0000) time: 0.1201 data: 0.0273 max mem: 3531 +Test: [ 40/235] eta: 0:00:53 acc1_pos: 51.562 acc5_pos: 100.000 loss: 0.7925 (0.8040) acc1: 56.2500 (55.7546) acc5: 100.0000 (100.0000) time: 0.1905 data: 0.0957 max mem: 3531 +Test: [ 50/235] eta: 0:00:49 acc1_pos: 51.042 acc5_pos: 100.000 loss: 0.7937 (0.8056) acc1: 55.9896 (55.5556) acc5: 100.0000 (100.0000) time: 0.2394 data: 0.1469 max mem: 3531 +Test: [ 60/235] eta: 0:00:45 acc1_pos: 57.552 acc5_pos: 100.000 loss: 0.7895 (0.8039) acc1: 56.2500 (55.7121) acc5: 100.0000 (100.0000) time: 0.2288 data: 0.1353 max mem: 3531 +Test: [ 70/235] eta: 0:00:41 acc1_pos: 58.462 acc5_pos: 100.000 loss: 0.7611 (0.7935) acc1: 58.8542 (56.7268) acc5: 100.0000 (100.0000) acc1_neg: 59.635 acc5_neg: 100.000 time: 0.2098 data: 0.1171 max mem: 3531 +Test: [ 80/235] eta: 0:00:37 acc1_pos: 58.462 acc5_pos: 100.000 loss: 0.7182 (0.7846) acc1: 63.5417 (57.5167) acc5: 100.0000 (100.0000) acc1_neg: 63.802 acc5_neg: 100.000 time: 0.1996 data: 0.1053 max mem: 3531 +Test: [ 90/235] eta: 0:00:34 acc1_pos: 58.462 acc5_pos: 100.000 loss: 0.7284 (0.7784) acc1: 63.0208 (58.0472) acc5: 100.0000 (100.0000) acc1_neg: 58.594 acc5_neg: 100.000 time: 0.1849 data: 0.0903 max mem: 3531 +Test: [100/235] eta: 0:00:31 acc1_pos: 58.462 acc5_pos: 100.000 loss: 0.7241 (0.7709) acc1: 63.5417 (58.6041) acc5: 100.0000 (100.0000) acc1_neg: 63.281 acc5_neg: 100.000 time: 0.1899 data: 0.0993 max mem: 3531 +Test: [110/235] eta: 0:00:28 acc1_pos: 58.462 acc5_pos: 100.000 loss: 0.7129 (0.7658) acc1: 63.5417 (59.0348) acc5: 100.0000 (100.0000) acc1_neg: 63.021 acc5_neg: 100.000 time: 0.1962 data: 0.1050 max mem: 3531 +Test: [120/235] eta: 0:00:26 acc1_pos: 58.462 acc5_pos: 100.000 loss: 0.7389 (0.7712) acc1: 58.8542 (58.7186) acc5: 100.0000 (100.0000) acc1_neg: 54.688 acc5_neg: 100.000 time: 0.1982 data: 0.1044 max mem: 3531 +Test: [130/235] eta: 0:00:23 acc1_pos: 58.462 acc5_pos: 100.000 loss: 0.8330 (0.7764) acc1: 54.6875 (58.4129) acc5: 100.0000 (100.0000) acc1_neg: 55.729 acc5_neg: 100.000 time: 0.1984 data: 0.1055 max mem: 3531 +Test: [140/235] eta: 0:00:21 acc1_pos: 58.462 acc5_pos: 100.000 loss: 0.8482 (0.7827) acc1: 53.6458 (58.0341) acc5: 100.0000 (100.0000) acc1_neg: 52.865 acc5_neg: 100.000 time: 0.1938 data: 0.1028 max mem: 3531 +Test: [150/235] eta: 0:00:18 acc1_pos: 58.462 acc5_pos: 100.000 loss: 0.8448 (0.7859) acc1: 53.6458 (57.8160) acc5: 100.0000 (100.0000) acc1_neg: 54.948 acc5_neg: 100.000 time: 0.1935 data: 0.0997 max mem: 3531 +Test: [160/235] eta: 0:00:16 acc1_pos: 58.462 acc5_pos: 100.000 loss: 0.8462 (0.7901) acc1: 54.4271 (57.5230) acc5: 100.0000 (100.0000) acc1_neg: 57.031 acc5_neg: 100.000 time: 0.1943 data: 0.0975 max mem: 3531 +Test: [170/235] eta: 0:00:14 acc1_pos: 58.462 acc5_pos: 100.000 loss: 0.7954 (0.7732) acc1: 55.2083 (58.6166) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.1873 data: 0.0925 max mem: 3531 +Test: [180/235] eta: 0:00:11 acc1_pos: 58.462 acc5_pos: 100.000 loss: 0.4609 (0.7555) acc1: 78.3854 (59.8138) acc5: 100.0000 (100.0000) acc1_neg: 79.427 acc5_neg: 100.000 time: 0.1908 data: 0.0993 max mem: 3531 +Test: [190/235] eta: 0:00:09 acc1_pos: 58.462 acc5_pos: 100.000 loss: 0.4410 (0.7389) acc1: 79.6875 (60.8884) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.2098 data: 0.1183 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 58.462 acc5_pos: 100.000 loss: 0.4429 (0.7247) acc1: 79.6875 (61.7913) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.2075 data: 0.1154 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 58.462 acc5_pos: 100.000 loss: 0.4429 (0.7113) acc1: 80.4688 (62.6913) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.1907 data: 0.0971 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 58.462 acc5_pos: 100.000 loss: 0.4308 (0.6992) acc1: 80.9896 (63.4839) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.1667 data: 0.0737 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 58.462 acc5_pos: 100.000 loss: 0.4234 (0.6870) acc1: 80.9896 (64.2880) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.1378 data: 0.0475 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 58.462 acc5_pos: 100.000 loss: 0.4234 (0.6824) acc1: 81.7708 (64.5167) acc5: 100.0000 (100.0000) acc1_neg: 86.957 acc5_neg: 100.000 time: 0.1267 data: 0.0399 max mem: 3532 +Test: Total time: 0:00:48 (0.2056 s / it) +* Overall Acc@1 64.517 Acc@5 100.000 + target == real Acc@1 55.732 Acc@5 100.000 + target == fake Acc@1 67.694 Acc@5 100.000 + loss 0.682 + PR-AUC 0.8424 ROC-AUC 0.6693 EER 37.48% (thr=0.5771) +Accuracy of the network on the 89879 test images: 64.5% +[rank0]:[W726 06:16:25.972769639 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 06:16:45.771000 1201846 torch/distributed/run.py:766] +W0726 06:16:45.771000 1201846 torch/distributed/run.py:766] ***************************************** +W0726 06:16:45.771000 1201846 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:16:45.771000 1201846 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:17:13.847421177 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:17:13.058361224 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:17:14.628077367 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:17:14.655485416 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 06:17:14.695132342 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:17:14.801418086 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:17:14.890886860 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:17:14.910935181 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n8_seed9_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n8_classseed9_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:26:56 acc1_pos: 64.844 acc5_pos: 100.000 loss: 0.6351 (0.6351) acc1: 64.8438 (64.8438) acc5: 100.0000 (100.0000) time: 6.0096 data: 4.6755 max mem: 3531 +Test: [ 10/269] eta: 0:02:42 acc1_pos: 72.396 acc5_pos: 100.000 loss: 0.5367 (0.5348) acc1: 76.0417 (75.6392) acc5: 100.0000 (100.0000) time: 0.6280 data: 0.4254 max mem: 3531 +Test: [ 20/269] eta: 0:01:33 acc1_pos: 78.385 acc5_pos: 100.000 loss: 0.5585 (0.5857) acc1: 72.3958 (70.7837) acc5: 100.0000 (100.0000) time: 0.0920 data: 0.0004 max mem: 3531 +Test: [ 30/269] eta: 0:01:17 acc1_pos: 87.500 acc5_pos: 100.000 loss: 0.5610 (0.5632) acc1: 71.3542 (73.4795) acc5: 100.0000 (100.0000) time: 0.1541 data: 0.0611 max mem: 3531 +Test: [ 40/269] eta: 0:01:11 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.5901 (0.5868) acc1: 69.7917 (69.8044) acc5: 100.0000 (100.0000) acc1_neg: 59.115 acc5_neg: 100.000 time: 0.2499 data: 0.1564 max mem: 3531 +Test: [ 50/269] eta: 0:01:05 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.7341 (0.6503) acc1: 47.3958 (63.7153) acc5: 100.0000 (100.0000) acc1_neg: 30.469 acc5_neg: 100.000 time: 0.2579 data: 0.1623 max mem: 3531 +Test: [ 60/269] eta: 0:00:59 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.9120 (0.7008) acc1: 34.8958 (58.9097) acc5: 100.0000 (100.0000) acc1_neg: 48.698 acc5_neg: 100.000 time: 0.2299 data: 0.1333 max mem: 3531 +Test: [ 70/269] eta: 0:00:52 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.8010 (0.7114) acc1: 44.7917 (57.2256) acc5: 100.0000 (100.0000) acc1_neg: 47.917 acc5_neg: 100.000 time: 0.1849 data: 0.0894 max mem: 3531 +Test: [ 80/269] eta: 0:00:48 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.7431 (0.7080) acc1: 50.0000 (57.0859) acc5: 100.0000 (100.0000) acc1_neg: 68.490 acc5_neg: 100.000 time: 0.1610 data: 0.0671 max mem: 3531 +Test: [ 90/269] eta: 0:00:44 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.6812 (0.7078) acc1: 54.6875 (56.8452) acc5: 100.0000 (100.0000) acc1_neg: 63.021 acc5_neg: 100.000 time: 0.1999 data: 0.1078 max mem: 3531 +Test: [100/269] eta: 0:00:41 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.7141 (0.7145) acc1: 50.7812 (55.7782) acc5: 100.0000 (100.0000) acc1_neg: 36.719 acc5_neg: 100.000 time: 0.1991 data: 0.1082 max mem: 3531 +Test: [110/269] eta: 0:00:38 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.8394 (0.7334) acc1: 36.7188 (53.4511) acc5: 100.0000 (100.0000) acc1_neg: 30.729 acc5_neg: 100.000 time: 0.1851 data: 0.0939 max mem: 3531 +Test: [120/269] eta: 0:00:35 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.8394 (0.7399) acc1: 32.8125 (52.6946) acc5: 100.0000 (100.0000) acc1_neg: 30.729 acc5_neg: 100.000 time: 0.1896 data: 0.0975 max mem: 3531 +Test: [130/269] eta: 0:00:31 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.8028 (0.7462) acc1: 40.3646 (51.8050) acc5: 100.0000 (100.0000) acc1_neg: 33.073 acc5_neg: 100.000 time: 0.1797 data: 0.0869 max mem: 3531 +Test: [140/269] eta: 0:00:29 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.7987 (0.7475) acc1: 40.6250 (51.4942) acc5: 100.0000 (100.0000) acc1_neg: 63.542 acc5_neg: 100.000 time: 0.1699 data: 0.0752 max mem: 3531 +Test: [150/269] eta: 0:00:26 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.6840 (0.7430) acc1: 52.8646 (51.7350) acc5: 100.0000 (100.0000) acc1_neg: 68.490 acc5_neg: 100.000 time: 0.1749 data: 0.0795 max mem: 3531 +Test: [160/269] eta: 0:00:23 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.6572 (0.7378) acc1: 56.5104 (52.0914) acc5: 100.0000 (100.0000) acc1_neg: 55.469 acc5_neg: 100.000 time: 0.1699 data: 0.0772 max mem: 3531 +Test: [170/269] eta: 0:00:21 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.6572 (0.7340) acc1: 55.4688 (52.4245) acc5: 100.0000 (100.0000) acc1_neg: 69.531 acc5_neg: 100.000 time: 0.1749 data: 0.0827 max mem: 3531 +Test: [180/269] eta: 0:00:19 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.6248 (0.7278) acc1: 61.1979 (53.0228) acc5: 100.0000 (100.0000) acc1_neg: 52.865 acc5_neg: 100.000 time: 0.1851 data: 0.0922 max mem: 3531 +Test: [190/269] eta: 0:00:17 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.6540 (0.7305) acc1: 57.8125 (52.7978) acc5: 100.0000 (100.0000) acc1_neg: 54.427 acc5_neg: 100.000 time: 0.2005 data: 0.1082 max mem: 3531 +Test: [200/269] eta: 0:00:14 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.7912 (0.7317) acc1: 48.9583 (52.8633) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.2247 data: 0.1306 max mem: 3531 +Test: [210/269] eta: 0:00:12 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.7024 (0.7304) acc1: 52.3438 (52.9794) acc5: 100.0000 (100.0000) acc1_neg: 59.375 acc5_neg: 100.000 time: 0.1994 data: 0.1027 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.6820 (0.7310) acc1: 55.2083 (52.9589) acc5: 100.0000 (100.0000) acc1_neg: 64.062 acc5_neg: 100.000 time: 0.1849 data: 0.0879 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.7598 (0.7321) acc1: 46.3542 (52.6977) acc5: 100.0000 (100.0000) acc1_neg: 39.323 acc5_neg: 100.000 time: 0.1949 data: 0.1006 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.7980 (0.7358) acc1: 41.6667 (52.2822) acc5: 100.0000 (100.0000) acc1_neg: 54.167 acc5_neg: 100.000 time: 0.1899 data: 0.0971 max mem: 3532 +Test: [250/269] eta: 0:00:03 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.8038 (0.7353) acc1: 44.5312 (52.2950) acc5: 100.0000 (100.0000) acc1_neg: 47.656 acc5_neg: 100.000 time: 0.1797 data: 0.0869 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.8411 (0.7428) acc1: 38.8021 (51.5316) acc5: 100.0000 (100.0000) acc1_neg: 38.802 acc5_neg: 100.000 time: 0.1776 data: 0.0866 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.8535 (0.7453) acc1: 35.9375 (51.1864) acc5: 100.0000 (100.0000) acc1_neg: 61.680 acc5_neg: 100.000 time: 0.1501 data: 0.0560 max mem: 3532 +Test: Total time: 0:00:55 (0.2066 s / it) +* Overall Acc@1 51.186 Acc@5 100.000 + target == real Acc@1 73.525 Acc@5 100.000 + target == fake Acc@1 48.138 Acc@5 100.000 + loss 0.745 + PR-AUC 0.9300 ROC-AUC 0.6576 EER 38.91% (thr=0.4192) +Accuracy of the network on the 103293 test images: 51.2% +[rank0]:[W726 06:18:19.898333627 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T06:18:26Z] eval_done tag=fixed500k_real_n8_seed9_resampled +[2026-07-26T06:18:27Z] eval_start tag=fixed500k_fake_n16_seed9_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed9_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T06:18:27Z +hostname=n116-041-153 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed9_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n16_seed9_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 06:18:39.792000 1202409 torch/distributed/run.py:766] +W0726 06:18:39.792000 1202409 torch/distributed/run.py:766] ***************************************** +W0726 06:18:39.792000 1202409 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:18:39.792000 1202409 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:19:07.096708695 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:19:08.647919631 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:19:08.803162817 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:19:08.844228393 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:19:08.103888227 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:19:08.108950317 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +[rank0]:[W726 06:19:08.125537455 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:19:08.136066313 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n16_seed9_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed9_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:27:49 acc1_pos: 27.865 acc5_pos: 100.000 loss: 0.8909 (0.8909) acc1: 27.8646 (27.8646) acc5: 100.0000 (100.0000) time: 7.1356 data: 5.4847 max mem: 3531 +Test: [ 10/234] eta: 0:02:45 acc1_pos: 23.698 acc5_pos: 100.000 loss: 0.9313 (0.9222) acc1: 28.1250 (32.9545) acc5: 100.0000 (100.0000) time: 0.7371 data: 0.4990 max mem: 3531 +Test: [ 20/234] eta: 0:01:32 acc1_pos: 94.531 acc5_pos: 100.000 loss: 0.8954 (0.8092) acc1: 36.7188 (42.0139) acc5: 100.0000 (100.0000) time: 0.0986 data: 0.0004 max mem: 3531 +Test: [ 30/234] eta: 0:01:19 acc1_pos: 73.958 acc5_pos: 100.000 loss: 0.2159 (0.6250) acc1: 93.4896 (58.0393) acc5: 100.0000 (100.0000) time: 0.1999 data: 0.1023 max mem: 3531 +Test: [ 40/234] eta: 0:01:07 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1921 (0.5572) acc1: 94.5312 (64.5897) acc5: 100.0000 (100.0000) time: 0.2599 data: 0.1665 max mem: 3531 +Test: [ 50/234] eta: 0:00:59 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.1390 (0.4744) acc1: 96.8750 (70.9763) acc5: 100.0000 (100.0000) time: 0.2149 data: 0.1206 max mem: 3531 +Test: [ 60/234] eta: 0:00:51 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.1146 (0.4104) acc1: 97.9167 (75.6788) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0869 max mem: 3531 +Test: [ 70/234] eta: 0:00:45 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.0826 (0.4406) acc1: 99.7396 (74.5599) acc5: 100.0000 (100.0000) acc1_neg: 50.521 acc5_neg: 100.000 time: 0.1750 data: 0.0791 max mem: 3531 +Test: [ 80/234] eta: 0:00:41 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.8368 (0.4628) acc1: 53.3854 (73.9294) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2008 data: 0.1092 max mem: 3531 +Test: [ 90/234] eta: 0:00:38 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.1706 (0.4285) acc1: 97.6562 (76.4938) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.2097 data: 0.1185 max mem: 3531 +Test: [100/234] eta: 0:00:34 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.1983 (0.4263) acc1: 95.5729 (77.1452) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1990 data: 0.1056 max mem: 3531 +Test: [110/234] eta: 0:00:31 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.3272 (0.4156) acc1: 86.9792 (78.3033) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.2099 data: 0.1153 max mem: 3531 +Test: [120/234] eta: 0:00:28 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.2917 (0.4151) acc1: 89.5833 (78.7771) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2206 data: 0.1267 max mem: 3531 +Test: [130/234] eta: 0:00:26 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.3513 (0.4134) acc1: 89.3229 (79.5046) acc5: 100.0000 (100.0000) acc1_neg: 71.354 acc5_neg: 100.000 time: 0.2259 data: 0.1339 max mem: 3531 +Test: [140/234] eta: 0:00:23 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.4439 (0.4669) acc1: 79.1667 (76.7638) acc5: 100.0000 (100.0000) acc1_neg: 12.240 acc5_neg: 100.000 time: 0.2150 data: 0.1232 max mem: 3531 +Test: [150/234] eta: 0:00:20 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.4439 (0.4733) acc1: 71.3542 (76.7746) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.2170 data: 0.1257 max mem: 3531 +Test: [160/234] eta: 0:00:17 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.3657 (0.4707) acc1: 85.1562 (77.1804) acc5: 100.0000 (100.0000) acc1_neg: 73.698 acc5_neg: 100.000 time: 0.2123 data: 0.1202 max mem: 3531 +Test: [170/234] eta: 0:00:15 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.5398 (0.4729) acc1: 74.4792 (77.2554) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1749 data: 0.0828 max mem: 3531 +Test: [180/234] eta: 0:00:12 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.5558 (0.4837) acc1: 72.1354 (76.7438) acc5: 100.0000 (100.0000) acc1_neg: 32.552 acc5_neg: 100.000 time: 0.1908 data: 0.1001 max mem: 3531 +Test: [190/234] eta: 0:00:10 acc1_pos: 95.172 acc5_pos: 100.000 loss: 1.2759 (0.5370) acc1: 32.5521 (73.8288) acc5: 100.0000 (100.0000) acc1_neg: 20.312 acc5_neg: 100.000 time: 0.1905 data: 0.0999 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 95.172 acc5_pos: 100.000 loss: 1.4587 (0.5465) acc1: 20.8333 (73.6059) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1529 data: 0.0618 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.2919 (0.5350) acc1: 96.6146 (74.5433) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.1384 data: 0.0469 max mem: 3532 +Test: [220/234] eta: 0:00:03 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.3313 (0.5280) acc1: 87.5000 (74.9847) acc5: 100.0000 (100.0000) acc1_neg: 64.062 acc5_neg: 100.000 time: 0.1351 data: 0.0438 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.6823 (0.5663) acc1: 20.8333 (72.5514) acc5: 100.0000 (100.0000) acc1_neg: 19.010 acc5_neg: 100.000 time: 0.1366 data: 0.0462 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 95.172 acc5_pos: 100.000 loss: 1.3736 (0.5769) acc1: 19.7917 (71.9344) acc5: 100.0000 (100.0000) acc1_neg: 20.233 acc5_neg: 100.000 time: 0.1338 data: 0.0426 max mem: 3532 +Test: Total time: 0:00:50 (0.2158 s / it) +* Overall Acc@1 71.934 Acc@5 100.000 + target == real Acc@1 71.154 Acc@5 100.000 + target == fake Acc@1 72.154 Acc@5 100.000 + loss 0.577 + PR-AUC 0.9444 ROC-AUC 0.8183 EER 28.22% (thr=0.5068) +Accuracy of the network on the 89729 test images: 71.9% +[rank0]:[W726 06:20:06.226966783 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 06:20:26.097000 1202970 torch/distributed/run.py:766] +W0726 06:20:26.097000 1202970 torch/distributed/run.py:766] ***************************************** +W0726 06:20:26.097000 1202970 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:20:26.097000 1202970 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:20:54.050111046 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:20:54.480023368 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 06:20:55.670547372 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:20:55.709280581 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:20:55.844443599 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:20:55.007693808 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:20:55.094409770 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:20:55.151717047 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n16_seed9_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed9_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:35:24 acc1_pos: 24.219 acc5_pos: 100.000 loss: 1.0587 (1.0587) acc1: 24.2188 (24.2188) acc5: 100.0000 (100.0000) time: 6.9228 data: 4.9980 max mem: 3531 +Test: [ 10/2387] eta: 0:28:18 acc1_pos: 5.208 acc5_pos: 100.000 loss: 1.1408 (1.1426) acc1: 11.9792 (13.7547) acc5: 100.0000 (100.0000) time: 0.7146 data: 0.4547 max mem: 3531 +Test: [ 20/2387] eta: 0:16:38 acc1_pos: 1.823 acc5_pos: 100.000 loss: 1.1984 (1.1834) acc1: 10.4167 (12.2396) acc5: 100.0000 (100.0000) time: 0.0969 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:13:39 acc1_pos: 1.042 acc5_pos: 100.000 loss: 1.2467 (1.2151) acc1: 7.2917 (10.5427) acc5: 100.0000 (100.0000) time: 0.1457 data: 0.0476 max mem: 3531 +Test: [ 40/2387] eta: 0:13:30 acc1_pos: 4.427 acc5_pos: 100.000 loss: 1.2550 (1.2229) acc1: 4.9479 (9.1082) acc5: 100.0000 (100.0000) time: 0.2649 data: 0.1699 max mem: 3531 +Test: [ 50/2387] eta: 0:12:43 acc1_pos: 26.042 acc5_pos: 100.000 loss: 1.1735 (1.1922) acc1: 8.3333 (11.6115) acc5: 100.0000 (100.0000) time: 0.2942 data: 0.1991 max mem: 3531 +Test: [ 60/2387] eta: 0:11:51 acc1_pos: 20.312 acc5_pos: 100.000 loss: 1.1207 (1.1880) acc1: 13.8021 (11.9877) acc5: 100.0000 (100.0000) time: 0.2251 data: 0.1314 max mem: 3531 +Test: [ 70/2387] eta: 0:11:14 acc1_pos: 20.052 acc5_pos: 100.000 loss: 1.2018 (1.1962) acc1: 11.4583 (11.9828) acc5: 100.0000 (100.0000) time: 0.1999 data: 0.1033 max mem: 3531 +Test: [ 80/2387] eta: 0:11:05 acc1_pos: 21.354 acc5_pos: 100.000 loss: 1.2004 (1.1897) acc1: 11.4583 (12.1142) acc5: 100.0000 (100.0000) time: 0.2347 data: 0.1360 max mem: 3531 +Test: [ 90/2387] eta: 0:10:50 acc1_pos: 25.521 acc5_pos: 100.000 loss: 1.1455 (1.1827) acc1: 15.1042 (12.5658) acc5: 100.0000 (100.0000) time: 0.2549 data: 0.1595 max mem: 3531 +Test: [ 100/2387] eta: 0:10:35 acc1_pos: 5.729 acc5_pos: 100.000 loss: 1.1187 (1.1814) acc1: 13.5417 (12.4716) acc5: 100.0000 (100.0000) time: 0.2348 data: 0.1371 max mem: 3531 +Test: [ 110/2387] eta: 0:10:28 acc1_pos: 29.688 acc5_pos: 100.000 loss: 1.1097 (1.1724) acc1: 11.7188 (12.8988) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1463 max mem: 3531 +Test: [ 120/2387] eta: 0:10:19 acc1_pos: 20.573 acc5_pos: 100.000 loss: 1.0869 (1.1672) acc1: 19.5312 (13.2683) acc5: 100.0000 (100.0000) time: 0.2499 data: 0.1499 max mem: 3531 +Test: [ 130/2387] eta: 0:10:10 acc1_pos: 20.312 acc5_pos: 100.000 loss: 1.0980 (1.1636) acc1: 17.9688 (13.3965) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1391 max mem: 3531 +Test: [ 140/2387] eta: 0:10:07 acc1_pos: 19.010 acc5_pos: 100.000 loss: 1.1028 (1.1644) acc1: 11.7188 (13.2646) acc5: 100.0000 (100.0000) time: 0.2549 data: 0.1572 max mem: 3531 +Test: [ 150/2387] eta: 0:10:00 acc1_pos: 4.427 acc5_pos: 100.000 loss: 1.0821 (1.1570) acc1: 12.7604 (13.4882) acc5: 100.0000 (100.0000) time: 0.2549 data: 0.1587 max mem: 3531 +Test: [ 160/2387] eta: 0:10:00 acc1_pos: 18.721 acc5_pos: 100.000 loss: 1.0014 (1.1303) acc1: 19.2708 (15.6428) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2649 data: 0.1696 max mem: 3531 +Test: [ 170/2387] eta: 0:09:36 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.2076 (1.0757) acc1: 98.9583 (20.5562) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1949 data: 0.1005 max mem: 3531 +Test: [ 180/2387] eta: 0:09:37 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1968 (1.0272) acc1: 99.4792 (24.9137) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1949 data: 0.0981 max mem: 3531 +Test: [ 190/2387] eta: 0:09:27 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1943 (0.9836) acc1: 99.4792 (28.8204) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2449 data: 0.1462 max mem: 3531 +Test: [ 200/2387] eta: 0:09:14 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1935 (0.9444) acc1: 99.4792 (32.3396) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1799 data: 0.0818 max mem: 3531 +Test: [ 210/2387] eta: 0:09:08 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1947 (0.9090) acc1: 99.4792 (35.5166) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1899 data: 0.0941 max mem: 3532 +Test: [ 220/2387] eta: 0:09:00 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1994 (0.8769) acc1: 99.4792 (38.4073) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2099 data: 0.1151 max mem: 3532 +Test: [ 230/2387] eta: 0:08:49 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1986 (0.8474) acc1: 99.4792 (41.0489) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1800 data: 0.0849 max mem: 3532 +Test: [ 240/2387] eta: 0:08:39 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1986 (0.8205) acc1: 99.4792 (43.4680) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1599 data: 0.0649 max mem: 3532 +Test: [ 250/2387] eta: 0:08:34 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1934 (0.7931) acc1: 99.7396 (45.7192) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1860 data: 0.0930 max mem: 3532 +Test: [ 260/2387] eta: 0:08:28 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1195 (0.7670) acc1: 100.0000 (47.7989) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2059 data: 0.1152 max mem: 3532 +Test: [ 270/2387] eta: 0:08:23 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1110 (0.7428) acc1: 100.0000 (49.7242) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1991 data: 0.1071 max mem: 3532 +Test: [ 280/2387] eta: 0:08:18 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1120 (0.7205) acc1: 100.0000 (51.5134) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2007 data: 0.1088 max mem: 3532 +Test: [ 290/2387] eta: 0:08:12 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1169 (0.6999) acc1: 100.0000 (53.1796) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1946 data: 0.1031 max mem: 3532 +Test: [ 300/2387] eta: 0:08:08 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1170 (0.6805) acc1: 100.0000 (54.7351) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2007 data: 0.1073 max mem: 3532 +Test: [ 310/2387] eta: 0:08:04 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1187 (0.6656) acc1: 100.0000 (56.1143) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2121 data: 0.1192 max mem: 3532 +Test: [ 320/2387] eta: 0:08:01 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.2992 (0.6543) acc1: 98.6979 (57.4361) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2123 data: 0.1178 max mem: 3532 +Test: [ 330/2387] eta: 0:07:57 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.3011 (0.6435) acc1: 98.6979 (58.6874) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2134 data: 0.1186 max mem: 3532 +Test: [ 340/2387] eta: 0:07:53 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.2974 (0.6334) acc1: 98.9583 (59.8630) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2099 data: 0.1184 max mem: 3532 +Test: [ 350/2387] eta: 0:07:50 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.2859 (0.6212) acc1: 99.4792 (61.0020) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2122 data: 0.1190 max mem: 3532 +Test: [ 360/2387] eta: 0:07:46 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1835 (0.6088) acc1: 100.0000 (62.0773) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2034 data: 0.1104 max mem: 3532 +Test: [ 370/2387] eta: 0:07:41 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1730 (0.5971) acc1: 99.7396 (63.0959) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1872 data: 0.0957 max mem: 3532 +Test: [ 380/2387] eta: 0:07:36 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1713 (0.5859) acc1: 99.7396 (64.0584) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1836 data: 0.0922 max mem: 3532 +Test: [ 390/2387] eta: 0:07:33 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1723 (0.5753) acc1: 99.7396 (64.9696) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1931 data: 0.0990 max mem: 3532 +Test: [ 400/2387] eta: 0:07:29 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1723 (0.5653) acc1: 100.0000 (65.8406) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2027 data: 0.1071 max mem: 3532 +Test: [ 410/2387] eta: 0:07:25 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1712 (0.5557) acc1: 100.0000 (66.6673) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1952 data: 0.1026 max mem: 3532 +Test: [ 420/2387] eta: 0:07:20 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1686 (0.5465) acc1: 100.0000 (67.4560) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1825 data: 0.0906 max mem: 3532 +Test: [ 430/2387] eta: 0:07:17 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1735 (0.5379) acc1: 100.0000 (68.2062) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1862 data: 0.0946 max mem: 3532 +Test: [ 440/2387] eta: 0:07:12 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1739 (0.5283) acc1: 100.0000 (68.9254) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1778 data: 0.0865 max mem: 3533 +Test: [ 450/2387] eta: 0:07:07 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.0763 (0.5183) acc1: 100.0000 (69.6144) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1622 data: 0.0706 max mem: 3533 +Test: [ 460/2387] eta: 0:07:04 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.0761 (0.5087) acc1: 100.0000 (70.2735) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1764 data: 0.0850 max mem: 3533 +Test: [ 470/2387] eta: 0:07:00 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.0760 (0.4995) acc1: 100.0000 (70.9047) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1834 data: 0.0924 max mem: 3533 +Test: [ 480/2387] eta: 0:06:56 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.0759 (0.4907) acc1: 100.0000 (71.5096) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1793 data: 0.0887 max mem: 3533 +Test: [ 490/2387] eta: 0:06:53 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.0762 (0.4823) acc1: 100.0000 (72.0898) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1845 data: 0.0938 max mem: 3533 +Test: [ 500/2387] eta: 0:06:49 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.0763 (0.4742) acc1: 100.0000 (72.6469) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1905 data: 0.0987 max mem: 3533 +Test: [ 510/2387] eta: 0:06:47 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.0768 (0.4665) acc1: 100.0000 (73.1822) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2042 data: 0.1124 max mem: 3533 +Test: [ 520/2387] eta: 0:06:43 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.0905 (0.4593) acc1: 100.0000 (73.6969) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1885 data: 0.0962 max mem: 3533 +Test: [ 530/2387] eta: 0:06:40 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.0926 (0.4524) acc1: 100.0000 (74.1923) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1779 data: 0.0842 max mem: 3533 +Test: [ 540/2387] eta: 0:06:37 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.0926 (0.4458) acc1: 100.0000 (74.6693) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1873 data: 0.0937 max mem: 3533 +Test: [ 550/2387] eta: 0:06:34 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.0926 (0.4394) acc1: 100.0000 (75.1290) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1875 data: 0.0938 max mem: 3533 +Test: [ 560/2387] eta: 0:06:31 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.0926 (0.4332) acc1: 100.0000 (75.5724) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1876 data: 0.0951 max mem: 3533 +Test: [ 570/2387] eta: 0:06:29 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.0926 (0.4272) acc1: 100.0000 (76.0002) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1980 data: 0.1059 max mem: 3533 +Test: [ 580/2387] eta: 0:06:26 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.0926 (0.4215) acc1: 100.0000 (76.4132) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2041 data: 0.1095 max mem: 3533 +Test: [ 590/2387] eta: 0:06:23 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.0932 (0.4159) acc1: 100.0000 (76.8123) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1874 data: 0.0911 max mem: 3533 +Test: [ 600/2387] eta: 0:06:20 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.0928 (0.4106) acc1: 100.0000 (77.1982) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1852 data: 0.0907 max mem: 3533 +Test: [ 610/2387] eta: 0:06:17 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1150 (0.4067) acc1: 100.0000 (77.5713) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1847 data: 0.0926 max mem: 3533 +Test: [ 620/2387] eta: 0:06:14 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1707 (0.4029) acc1: 100.0000 (77.9308) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1853 data: 0.0938 max mem: 3533 +Test: [ 630/2387] eta: 0:06:11 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1707 (0.3992) acc1: 100.0000 (78.2802) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1901 data: 0.0991 max mem: 3533 +Test: [ 640/2387] eta: 0:06:09 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1738 (0.3957) acc1: 100.0000 (78.6182) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1942 data: 0.1007 max mem: 3533 +Test: [ 650/2387] eta: 0:06:06 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1736 (0.3923) acc1: 100.0000 (78.9455) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.0997 max mem: 3533 +Test: [ 660/2387] eta: 0:06:03 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1725 (0.3889) acc1: 100.0000 (79.2628) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1870 data: 0.0941 max mem: 3533 +Test: [ 670/2387] eta: 0:06:01 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1709 (0.3857) acc1: 100.0000 (79.5715) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1854 data: 0.0939 max mem: 3534 +Test: [ 680/2387] eta: 0:05:58 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1682 (0.3825) acc1: 100.0000 (79.8711) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1845 data: 0.0909 max mem: 3534 +Test: [ 690/2387] eta: 0:05:55 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1682 (0.3825) acc1: 100.0000 (79.9419) acc5: 100.0000 (100.0000) acc1_neg: 74.219 acc5_neg: 100.000 time: 0.1794 data: 0.0860 max mem: 3534 +Test: [ 700/2387] eta: 0:05:52 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.4543 (0.3836) acc1: 79.4271 (79.9305) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.1809 data: 0.0895 max mem: 3534 +Test: [ 710/2387] eta: 0:05:49 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.4564 (0.3847) acc1: 79.4271 (79.9226) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.1797 data: 0.0874 max mem: 3534 +Test: [ 720/2387] eta: 0:05:46 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.4496 (0.3855) acc1: 79.9479 (79.9273) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.1746 data: 0.0827 max mem: 3534 +Test: [ 730/2387] eta: 0:05:44 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.4560 (0.3867) acc1: 79.1667 (79.9166) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.1754 data: 0.0846 max mem: 3534 +Test: [ 740/2387] eta: 0:05:41 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.4655 (0.3878) acc1: 78.3854 (79.9061) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.1747 data: 0.0842 max mem: 3534 +Test: [ 750/2387] eta: 0:05:38 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.4535 (0.3887) acc1: 79.4271 (79.9077) acc5: 100.0000 (100.0000) acc1_neg: 80.729 acc5_neg: 100.000 time: 0.1743 data: 0.0810 max mem: 3534 +Test: [ 760/2387] eta: 0:05:36 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.4512 (0.3890) acc1: 80.2083 (79.9353) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1825 data: 0.0868 max mem: 3534 +Test: [ 770/2387] eta: 0:05:33 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.3490 (0.3874) acc1: 86.4583 (80.0979) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1864 data: 0.0923 max mem: 3534 +Test: [ 780/2387] eta: 0:05:30 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1344 (0.3840) acc1: 100.0000 (80.3524) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1810 data: 0.0854 max mem: 3534 +Test: [ 790/2387] eta: 0:05:28 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1170 (0.3808) acc1: 100.0000 (80.5804) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1877 data: 0.0911 max mem: 3534 +Test: [ 800/2387] eta: 0:05:26 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1145 (0.3775) acc1: 100.0000 (80.8228) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1893 data: 0.0963 max mem: 3534 +Test: [ 810/2387] eta: 0:05:23 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1164 (0.3744) acc1: 100.0000 (81.0580) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1795 data: 0.0861 max mem: 3534 +Test: [ 820/2387] eta: 0:05:21 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1343 (0.3717) acc1: 100.0000 (81.2846) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1808 data: 0.0869 max mem: 3534 +Test: [ 830/2387] eta: 0:05:18 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1696 (0.3695) acc1: 99.4792 (81.4926) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1767 data: 0.0847 max mem: 3534 +Test: [ 840/2387] eta: 0:05:15 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.2207 (0.3683) acc1: 97.9167 (81.6882) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1721 data: 0.0784 max mem: 3534 +Test: [ 850/2387] eta: 0:05:13 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.2588 (0.3669) acc1: 97.9167 (81.8850) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1906 data: 0.0956 max mem: 3534 +Test: [ 860/2387] eta: 0:05:11 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.2374 (0.3653) acc1: 98.9583 (82.0812) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2013 data: 0.1082 max mem: 3534 +Test: [ 870/2387] eta: 0:05:09 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.2248 (0.3638) acc1: 99.4792 (82.2767) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1895 data: 0.0978 max mem: 3534 +Test: [ 880/2387] eta: 0:05:06 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.2449 (0.3626) acc1: 99.4792 (82.4622) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1849 data: 0.0918 max mem: 3534 +Test: [ 890/2387] eta: 0:05:04 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.2325 (0.3610) acc1: 99.7396 (82.6512) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1855 data: 0.0909 max mem: 3534 +Test: [ 900/2387] eta: 0:05:02 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.2518 (0.3634) acc1: 96.8750 (82.5090) acc5: 100.0000 (100.0000) acc1_neg: 66.146 acc5_neg: 100.000 time: 0.1851 data: 0.0924 max mem: 3535 +Test: [ 910/2387] eta: 0:04:59 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.6458 (0.3665) acc1: 64.0625 (82.3097) acc5: 100.0000 (100.0000) acc1_neg: 64.062 acc5_neg: 100.000 time: 0.1851 data: 0.0937 max mem: 3535 +Test: [ 920/2387] eta: 0:04:57 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.6458 (0.3696) acc1: 64.0625 (82.1065) acc5: 100.0000 (100.0000) acc1_neg: 58.594 acc5_neg: 100.000 time: 0.1849 data: 0.0938 max mem: 3535 +Test: [ 930/2387] eta: 0:04:55 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.6553 (0.3726) acc1: 63.5417 (81.9104) acc5: 100.0000 (100.0000) acc1_neg: 59.896 acc5_neg: 100.000 time: 0.1884 data: 0.0977 max mem: 3535 +Test: [ 940/2387] eta: 0:04:53 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.6571 (0.3755) acc1: 63.5417 (81.7246) acc5: 100.0000 (100.0000) acc1_neg: 63.542 acc5_neg: 100.000 time: 0.1902 data: 0.0977 max mem: 3535 +Test: [ 950/2387] eta: 0:04:50 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.6379 (0.3783) acc1: 63.5417 (81.5394) acc5: 100.0000 (100.0000) acc1_neg: 64.583 acc5_neg: 100.000 time: 0.1861 data: 0.0922 max mem: 3535 +Test: [ 960/2387] eta: 0:04:48 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.6259 (0.3810) acc1: 64.8438 (81.3660) acc5: 100.0000 (100.0000) acc1_neg: 63.281 acc5_neg: 100.000 time: 0.1868 data: 0.0948 max mem: 3535 +Test: [ 970/2387] eta: 0:04:46 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.6349 (0.3838) acc1: 64.5833 (81.1819) acc5: 100.0000 (100.0000) acc1_neg: 62.760 acc5_neg: 100.000 time: 0.1849 data: 0.0942 max mem: 3535 +Test: [ 980/2387] eta: 0:04:43 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.6139 (0.3834) acc1: 66.9271 (81.2423) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1836 data: 0.0863 max mem: 3535 +Test: [ 990/2387] eta: 0:04:41 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1774 (0.3815) acc1: 98.1771 (81.3974) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1821 data: 0.0842 max mem: 3535 +Test: [1000/2387] eta: 0:04:39 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1599 (0.3793) acc1: 99.4792 (81.5674) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1797 data: 0.0870 max mem: 3535 +Test: [1010/2387] eta: 0:04:36 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1778 (0.3774) acc1: 98.6979 (81.7307) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1799 data: 0.0879 max mem: 3535 +Test: [1020/2387] eta: 0:04:34 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1857 (0.3758) acc1: 98.1771 (81.8831) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1820 data: 0.0903 max mem: 3535 +Test: [1030/2387] eta: 0:04:32 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1785 (0.3737) acc1: 99.2188 (82.0547) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1807 data: 0.0878 max mem: 3535 +Test: [1040/2387] eta: 0:04:30 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1470 (0.3715) acc1: 100.0000 (82.2264) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1750 data: 0.0817 max mem: 3535 +Test: [1050/2387] eta: 0:04:27 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1326 (0.3692) acc1: 100.0000 (82.3945) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1813 data: 0.0899 max mem: 3535 +Test: [1060/2387] eta: 0:04:26 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1435 (0.3672) acc1: 100.0000 (82.5592) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1988 data: 0.1081 max mem: 3535 +Test: [1070/2387] eta: 0:04:23 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1503 (0.3654) acc1: 99.7396 (82.7133) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2025 data: 0.1107 max mem: 3535 +Test: [1080/2387] eta: 0:04:21 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1872 (0.3639) acc1: 99.2188 (82.8636) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1851 data: 0.0932 max mem: 3535 +Test: [1090/2387] eta: 0:04:19 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1649 (0.3620) acc1: 99.4792 (83.0168) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1972 data: 0.1008 max mem: 3535 +Test: [1100/2387] eta: 0:04:17 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1498 (0.3601) acc1: 99.7396 (83.1673) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2036 data: 0.1074 max mem: 3535 +Test: [1110/2387] eta: 0:04:15 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1527 (0.3583) acc1: 99.7396 (83.3078) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2026 data: 0.1099 max mem: 3535 +Test: [1120/2387] eta: 0:04:13 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1565 (0.3564) acc1: 99.4792 (83.4502) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2044 data: 0.1112 max mem: 3536 +Test: [1130/2387] eta: 0:04:11 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1602 (0.3548) acc1: 99.2188 (83.5898) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1899 data: 0.0983 max mem: 3536 +Test: [1140/2387] eta: 0:04:09 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1648 (0.3532) acc1: 99.7396 (83.7298) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1948 data: 0.1037 max mem: 3536 +Test: [1150/2387] eta: 0:04:07 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1838 (0.3519) acc1: 99.7396 (83.8612) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1904 data: 0.0996 max mem: 3536 +Test: [1160/2387] eta: 0:04:05 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1909 (0.3504) acc1: 100.0000 (83.9977) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1993 data: 0.1074 max mem: 3536 +Test: [1170/2387] eta: 0:04:03 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1804 (0.3489) acc1: 100.0000 (84.1324) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1812 data: 0.0877 max mem: 3536 +Test: [1180/2387] eta: 0:04:00 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1852 (0.3476) acc1: 100.0000 (84.2634) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1648 data: 0.0723 max mem: 3536 +Test: [1190/2387] eta: 0:03:58 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1859 (0.3463) acc1: 100.0000 (84.3940) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1796 data: 0.0871 max mem: 3536 +Test: [1200/2387] eta: 0:03:56 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1818 (0.3448) acc1: 100.0000 (84.5235) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1753 data: 0.0806 max mem: 3536 +Test: [1210/2387] eta: 0:03:54 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1650 (0.3434) acc1: 100.0000 (84.6509) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1813 data: 0.0868 max mem: 3536 +Test: [1220/2387] eta: 0:03:52 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1513 (0.3417) acc1: 100.0000 (84.7760) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1846 data: 0.0891 max mem: 3536 +Test: [1230/2387] eta: 0:03:50 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1283 (0.3399) acc1: 100.0000 (84.8996) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1932 data: 0.0990 max mem: 3536 +Test: [1240/2387] eta: 0:03:48 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1290 (0.3382) acc1: 100.0000 (85.0209) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.1038 max mem: 3536 +Test: [1250/2387] eta: 0:03:45 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1337 (0.3365) acc1: 100.0000 (85.1404) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1846 data: 0.0936 max mem: 3536 +Test: [1260/2387] eta: 0:03:44 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1341 (0.3349) acc1: 100.0000 (85.2581) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1915 data: 0.1007 max mem: 3536 +Test: [1270/2387] eta: 0:03:42 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1310 (0.3333) acc1: 100.0000 (85.3741) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2004 data: 0.1100 max mem: 3536 +Test: [1280/2387] eta: 0:03:40 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1205 (0.3317) acc1: 100.0000 (85.4864) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2001 data: 0.1099 max mem: 3536 +Test: [1290/2387] eta: 0:03:38 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1210 (0.3300) acc1: 99.7396 (85.5940) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2033 data: 0.1120 max mem: 3536 +Test: [1300/2387] eta: 0:03:36 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1183 (0.3284) acc1: 99.7396 (85.7009) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2039 data: 0.1123 max mem: 3536 +Test: [1310/2387] eta: 0:03:34 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1195 (0.3269) acc1: 99.7396 (85.8062) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1920 data: 0.1003 max mem: 3536 +Test: [1320/2387] eta: 0:03:32 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1360 (0.3255) acc1: 99.7396 (85.9101) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1903 data: 0.0986 max mem: 3536 +Test: [1330/2387] eta: 0:03:30 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1503 (0.3243) acc1: 99.7396 (86.0117) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2057 data: 0.1148 max mem: 3536 +Test: [1340/2387] eta: 0:03:28 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1698 (0.3242) acc1: 99.2188 (86.0315) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2037 data: 0.1128 max mem: 3536 +Test: [1350/2387] eta: 0:03:26 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1032 (0.3225) acc1: 100.0000 (86.1349) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1939 data: 0.1023 max mem: 3537 +Test: [1360/2387] eta: 0:03:23 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.0824 (0.3207) acc1: 100.0000 (86.2368) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1855 data: 0.0931 max mem: 3537 +Test: [1370/2387] eta: 0:03:22 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.0785 (0.3189) acc1: 100.0000 (86.3371) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2159 data: 0.1228 max mem: 3537 +Test: [1380/2387] eta: 0:03:20 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.0815 (0.3172) acc1: 100.0000 (86.4361) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2066 data: 0.1143 max mem: 3537 +Test: [1390/2387] eta: 0:03:18 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.0923 (0.3156) acc1: 100.0000 (86.5336) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1693 data: 0.0785 max mem: 3537 +Test: [1400/2387] eta: 0:03:16 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.0913 (0.3140) acc1: 100.0000 (86.6297) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1926 data: 0.1004 max mem: 3537 +Test: [1410/2387] eta: 0:03:14 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.0945 (0.3129) acc1: 100.0000 (86.7154) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1995 data: 0.1068 max mem: 3537 +Test: [1420/2387] eta: 0:03:12 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1678 (0.3119) acc1: 99.4792 (86.7955) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2101 data: 0.1163 max mem: 3537 +Test: [1430/2387] eta: 0:03:10 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1600 (0.3109) acc1: 99.4792 (86.8842) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2080 data: 0.1133 max mem: 3537 +Test: [1440/2387] eta: 0:03:08 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1625 (0.3099) acc1: 99.7396 (86.9727) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1926 data: 0.0989 max mem: 3537 +Test: [1450/2387] eta: 0:03:06 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1799 (0.3091) acc1: 99.7396 (87.0576) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1949 data: 0.1011 max mem: 3537 +Test: [1460/2387] eta: 0:03:04 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1693 (0.3081) acc1: 99.4792 (87.1394) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1870 data: 0.0919 max mem: 3537 +Test: [1470/2387] eta: 0:03:02 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1331 (0.3068) acc1: 99.7396 (87.2263) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2107 data: 0.1151 max mem: 3537 +Test: [1480/2387] eta: 0:02:59 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1262 (0.3056) acc1: 100.0000 (87.3108) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1814 data: 0.0870 max mem: 3537 +Test: [1490/2387] eta: 0:02:57 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1281 (0.3045) acc1: 100.0000 (87.3957) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1311 data: 0.0384 max mem: 3537 +Test: [1500/2387] eta: 0:02:55 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1374 (0.3034) acc1: 100.0000 (87.4769) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1444 data: 0.0518 max mem: 3537 +Test: [1510/2387] eta: 0:02:53 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1493 (0.3025) acc1: 99.7396 (87.5586) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1735 data: 0.0815 max mem: 3537 +Test: [1520/2387] eta: 0:02:51 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1779 (0.3020) acc1: 99.7396 (87.6317) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1912 data: 0.1000 max mem: 3537 +Test: [1530/2387] eta: 0:02:49 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.2292 (0.3016) acc1: 98.9583 (87.7036) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1947 data: 0.1029 max mem: 3537 +Test: [1540/2387] eta: 0:02:47 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.2144 (0.3010) acc1: 99.2188 (87.7782) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1987 data: 0.1079 max mem: 3537 +Test: [1550/2387] eta: 0:02:45 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.2172 (0.3005) acc1: 99.2188 (87.8526) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1872 data: 0.0967 max mem: 3537 +Test: [1560/2387] eta: 0:02:43 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.2215 (0.3000) acc1: 99.4792 (87.9249) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1896 data: 0.0964 max mem: 3537 +Test: [1570/2387] eta: 0:02:41 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.2146 (0.2994) acc1: 99.7396 (88.0003) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1922 data: 0.0981 max mem: 3537 +Test: [1580/2387] eta: 0:02:39 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1780 (0.2985) acc1: 100.0000 (88.0757) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1701 data: 0.0780 max mem: 3538 +Test: [1590/2387] eta: 0:02:36 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1330 (0.2974) acc1: 100.0000 (88.1498) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1499 data: 0.0584 max mem: 3538 +Test: [1600/2387] eta: 0:02:34 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1163 (0.2963) acc1: 100.0000 (88.2177) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1487 data: 0.0567 max mem: 3538 +Test: [1610/2387] eta: 0:02:32 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1163 (0.2953) acc1: 100.0000 (88.2906) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1503 data: 0.0585 max mem: 3538 +Test: [1620/2387] eta: 0:02:30 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1354 (0.2943) acc1: 100.0000 (88.3606) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1559 data: 0.0644 max mem: 3538 +Test: [1630/2387] eta: 0:02:28 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1506 (0.2936) acc1: 99.4792 (88.4238) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1635 data: 0.0698 max mem: 3538 +Test: [1640/2387] eta: 0:02:26 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1855 (0.2929) acc1: 98.1771 (88.4831) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1565 data: 0.0628 max mem: 3538 +Test: [1650/2387] eta: 0:02:23 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1811 (0.2922) acc1: 98.9583 (88.5473) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1488 data: 0.0574 max mem: 3538 +Test: [1660/2387] eta: 0:02:21 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1811 (0.2917) acc1: 99.2188 (88.6033) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1541 data: 0.0627 max mem: 3538 +Test: [1670/2387] eta: 0:02:19 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1807 (0.2910) acc1: 99.2188 (88.6643) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1608 data: 0.0692 max mem: 3538 +Test: [1680/2387] eta: 0:02:17 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1785 (0.2904) acc1: 98.9583 (88.7214) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1749 data: 0.0830 max mem: 3538 +Test: [1690/2387] eta: 0:02:15 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1800 (0.2898) acc1: 98.9583 (88.7759) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1647 data: 0.0714 max mem: 3538 +Test: [1700/2387] eta: 0:02:13 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1737 (0.2891) acc1: 99.7396 (88.8391) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1562 data: 0.0633 max mem: 3538 +Test: [1710/2387] eta: 0:02:11 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1550 (0.2883) acc1: 99.7396 (88.9039) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1678 data: 0.0754 max mem: 3538 +Test: [1720/2387] eta: 0:02:09 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1328 (0.2874) acc1: 100.0000 (88.9678) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1637 data: 0.0700 max mem: 3538 +Test: [1730/2387] eta: 0:02:07 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1328 (0.2865) acc1: 100.0000 (89.0309) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1599 data: 0.0652 max mem: 3538 +Test: [1740/2387] eta: 0:02:05 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1362 (0.2857) acc1: 100.0000 (89.0933) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1710 data: 0.0772 max mem: 3538 +Test: [1750/2387] eta: 0:02:03 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1369 (0.2848) acc1: 100.0000 (89.1556) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.1031 max mem: 3538 +Test: [1760/2387] eta: 0:02:01 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1350 (0.2840) acc1: 100.0000 (89.2172) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2105 data: 0.1186 max mem: 3538 +Test: [1770/2387] eta: 0:01:59 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1421 (0.2833) acc1: 100.0000 (89.2742) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1950 data: 0.1032 max mem: 3538 +Test: [1780/2387] eta: 0:01:57 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1474 (0.2825) acc1: 100.0000 (89.3340) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1940 data: 0.1014 max mem: 3538 +Test: [1790/2387] eta: 0:01:55 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1376 (0.2817) acc1: 100.0000 (89.3924) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1951 data: 0.1028 max mem: 3538 +Test: [1800/2387] eta: 0:01:53 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1307 (0.2809) acc1: 100.0000 (89.4505) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1805 data: 0.0896 max mem: 3538 +Test: [1810/2387] eta: 0:01:51 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1425 (0.2801) acc1: 100.0000 (89.5081) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1945 data: 0.1027 max mem: 3539 +Test: [1820/2387] eta: 0:01:50 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1396 (0.2793) acc1: 100.0000 (89.5656) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2190 data: 0.1256 max mem: 3539 +Test: [1830/2387] eta: 0:01:48 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1240 (0.2785) acc1: 100.0000 (89.6226) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2426 data: 0.1483 max mem: 3539 +Test: [1840/2387] eta: 0:01:46 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1091 (0.2776) acc1: 100.0000 (89.6790) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2189 data: 0.1261 max mem: 3539 +Test: [1850/2387] eta: 0:01:44 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1086 (0.2766) acc1: 100.0000 (89.7347) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1519 data: 0.0569 max mem: 3539 +Test: [1860/2387] eta: 0:01:42 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1090 (0.2757) acc1: 100.0000 (89.7899) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1488 data: 0.0538 max mem: 3539 +Test: [1870/2387] eta: 0:01:40 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1091 (0.2748) acc1: 100.0000 (89.8444) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1858 data: 0.0929 max mem: 3539 +Test: [1880/2387] eta: 0:01:38 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1080 (0.2740) acc1: 100.0000 (89.8984) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1964 data: 0.1038 max mem: 3539 +Test: [1890/2387] eta: 0:01:36 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1085 (0.2731) acc1: 100.0000 (89.9517) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1909 data: 0.0998 max mem: 3539 +Test: [1900/2387] eta: 0:01:34 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1091 (0.2722) acc1: 100.0000 (90.0043) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1912 data: 0.0995 max mem: 3539 +Test: [1910/2387] eta: 0:01:32 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1300 (0.2715) acc1: 100.0000 (90.0552) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1792 data: 0.0875 max mem: 3539 +Test: [1920/2387] eta: 0:01:30 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1415 (0.2712) acc1: 99.7396 (90.0910) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1732 data: 0.0808 max mem: 3539 +Test: [1930/2387] eta: 0:01:28 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1611 (0.2705) acc1: 100.0000 (90.1423) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1847 data: 0.0894 max mem: 3539 +Test: [1940/2387] eta: 0:01:26 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1402 (0.2700) acc1: 100.0000 (90.1895) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1774 data: 0.0833 max mem: 3539 +Test: [1950/2387] eta: 0:01:24 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1569 (0.2695) acc1: 100.0000 (90.2384) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1792 data: 0.0886 max mem: 3539 +Test: [1960/2387] eta: 0:01:22 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1789 (0.2691) acc1: 100.0000 (90.2811) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1838 data: 0.0919 max mem: 3539 +Test: [1970/2387] eta: 0:01:20 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.2143 (0.2691) acc1: 98.4375 (90.2866) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1746 data: 0.0810 max mem: 3539 +Test: [1980/2387] eta: 0:01:18 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.2191 (0.2687) acc1: 97.6562 (90.3230) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1760 data: 0.0820 max mem: 3539 +Test: [1990/2387] eta: 0:01:16 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1505 (0.2682) acc1: 98.6979 (90.3637) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1856 data: 0.0931 max mem: 3539 +Test: [2000/2387] eta: 0:01:14 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1493 (0.2676) acc1: 98.4375 (90.4044) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1901 data: 0.0956 max mem: 3539 +Test: [2010/2387] eta: 0:01:12 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1483 (0.2670) acc1: 98.4375 (90.4440) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1843 data: 0.0896 max mem: 3539 +Test: [2020/2387] eta: 0:01:10 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1491 (0.2664) acc1: 98.1771 (90.4820) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1822 data: 0.0910 max mem: 3539 +Test: [2030/2387] eta: 0:01:08 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1529 (0.2659) acc1: 98.1771 (90.5202) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1910 data: 0.0984 max mem: 3540 +Test: [2040/2387] eta: 0:01:07 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1507 (0.2653) acc1: 98.4375 (90.5603) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2143 data: 0.1205 max mem: 3540 +Test: [2050/2387] eta: 0:01:05 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1507 (0.2647) acc1: 98.4375 (90.5986) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1959 data: 0.0991 max mem: 3540 +Test: [2060/2387] eta: 0:01:03 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1545 (0.2642) acc1: 98.4375 (90.6350) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1850 data: 0.0891 max mem: 3540 +Test: [2070/2387] eta: 0:01:01 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1558 (0.2637) acc1: 98.4375 (90.6749) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1931 data: 0.1004 max mem: 3540 +Test: [2080/2387] eta: 0:00:59 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1546 (0.2632) acc1: 98.6979 (90.7127) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1871 data: 0.0937 max mem: 3540 +Test: [2090/2387] eta: 0:00:57 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1541 (0.2626) acc1: 98.6979 (90.7512) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1959 data: 0.1036 max mem: 3540 +Test: [2100/2387] eta: 0:00:55 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1541 (0.2621) acc1: 98.6979 (90.7875) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1938 data: 0.1025 max mem: 3540 +Test: [2110/2387] eta: 0:00:53 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1572 (0.2616) acc1: 98.4375 (90.8251) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1945 data: 0.1028 max mem: 3540 +Test: [2120/2387] eta: 0:00:51 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1572 (0.2612) acc1: 98.6979 (90.8639) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1907 data: 0.0986 max mem: 3540 +Test: [2130/2387] eta: 0:00:49 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1423 (0.2606) acc1: 100.0000 (90.9066) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0931 max mem: 3540 +Test: [2140/2387] eta: 0:00:47 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1303 (0.2601) acc1: 99.7396 (90.9388) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1879 data: 0.0939 max mem: 3540 +Test: [2150/2387] eta: 0:00:45 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1340 (0.2595) acc1: 99.7396 (90.9806) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1862 data: 0.0916 max mem: 3540 +Test: [2160/2387] eta: 0:00:43 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1392 (0.2590) acc1: 100.0000 (91.0199) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1865 data: 0.0950 max mem: 3540 +Test: [2170/2387] eta: 0:00:41 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1595 (0.2587) acc1: 99.4792 (91.0568) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1895 data: 0.0969 max mem: 3540 +Test: [2180/2387] eta: 0:00:39 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1952 (0.2589) acc1: 98.6979 (91.0545) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.1951 data: 0.1017 max mem: 3540 +Test: [2190/2387] eta: 0:00:38 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.3139 (0.2593) acc1: 87.5000 (91.0320) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2073 data: 0.1146 max mem: 3540 +Test: [2200/2387] eta: 0:00:36 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1712 (0.2588) acc1: 99.2188 (91.0712) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2041 data: 0.1101 max mem: 3540 +Test: [2210/2387] eta: 0:00:34 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1351 (0.2583) acc1: 100.0000 (91.1116) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2071 data: 0.1142 max mem: 3540 +Test: [2220/2387] eta: 0:00:32 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1351 (0.2578) acc1: 100.0000 (91.1504) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1955 data: 0.1045 max mem: 3540 +Test: [2230/2387] eta: 0:00:30 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1379 (0.2572) acc1: 100.0000 (91.1901) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1955 data: 0.1033 max mem: 3540 +Test: [2240/2387] eta: 0:00:28 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1428 (0.2567) acc1: 100.0000 (91.2288) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2039 data: 0.1111 max mem: 3540 +Test: [2250/2387] eta: 0:00:26 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1519 (0.2564) acc1: 100.0000 (91.2652) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1977 data: 0.1044 max mem: 3540 +Test: [2260/2387] eta: 0:00:24 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1734 (0.2560) acc1: 100.0000 (91.3037) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2049 data: 0.1072 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1258 (0.2554) acc1: 100.0000 (91.3420) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1892 data: 0.0928 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1098 (0.2548) acc1: 100.0000 (91.3702) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1709 data: 0.0799 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1101 (0.2542) acc1: 100.0000 (91.4078) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1704 data: 0.0801 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1172 (0.2537) acc1: 100.0000 (91.4437) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1632 data: 0.0722 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1410 (0.2533) acc1: 100.0000 (91.4775) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1582 data: 0.0670 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1657 (0.2530) acc1: 99.4792 (91.5071) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1502 data: 0.0598 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1986 (0.2528) acc1: 99.2188 (91.5358) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1385 data: 0.0484 max mem: 3541 +Test: [2340/2387] eta: 0:00:09 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1828 (0.2524) acc1: 99.7396 (91.5690) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1350 data: 0.0446 max mem: 3541 +Test: [2350/2387] eta: 0:00:07 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1658 (0.2522) acc1: 99.7396 (91.5899) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1310 data: 0.0411 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.1820 (0.2519) acc1: 99.7396 (91.6224) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1247 data: 0.0351 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.2002 (0.2518) acc1: 99.4792 (91.6401) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.1226 data: 0.0325 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.2337 (0.2519) acc1: 96.8750 (91.6555) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1193 data: 0.0292 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 18.721 acc5_pos: 100.000 loss: 0.2689 (0.2522) acc1: 93.2292 (91.6491) acc5: 100.0000 (100.0000) acc1_neg: 79.070 acc5_neg: 100.000 time: 0.1130 data: 0.0261 max mem: 3541 +Test: Total time: 0:07:35 (0.1908 s / it) +* Overall Acc@1 91.649 Acc@5 100.000 + target == real Acc@1 13.446 Acc@5 100.000 + target == fake Acc@1 97.141 Acc@5 100.000 + loss 0.252 + PR-AUC 0.9913 ROC-AUC 0.8869 EER 18.34% (thr=0.7827) +Accuracy of the network on the 916267 test images: 91.6% +[rank0]:[W726 06:28:56.514669458 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 06:29:16.184000 1203591 torch/distributed/run.py:766] +W0726 06:29:16.184000 1203591 torch/distributed/run.py:766] ***************************************** +W0726 06:29:16.184000 1203591 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:29:16.184000 1203591 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:29:43.455786707 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:29:43.484188321 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 06:29:44.916421054 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:29:44.076536764 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:29:44.293875378 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:29:44.315754147 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:29:44.367381317 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:29:44.378056832 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n16_seed9_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed9_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:00:00 acc1_pos: 24.219 acc5_pos: 100.000 loss: 1.0587 (1.0587) acc1: 24.2188 (24.2188) acc5: 100.0000 (100.0000) time: 6.9638 data: 4.8952 max mem: 3531 +Test: [ 10/517] eta: 0:06:02 acc1_pos: 5.208 acc5_pos: 100.000 loss: 1.1408 (1.1426) acc1: 11.9792 (13.7547) acc5: 100.0000 (100.0000) time: 0.7150 data: 0.4455 max mem: 3531 +Test: [ 20/517] eta: 0:03:29 acc1_pos: 1.823 acc5_pos: 100.000 loss: 1.1984 (1.1834) acc1: 10.4167 (12.2396) acc5: 100.0000 (100.0000) time: 0.0937 data: 0.0005 max mem: 3531 +Test: [ 30/517] eta: 0:02:56 acc1_pos: 1.042 acc5_pos: 100.000 loss: 1.2467 (1.2151) acc1: 7.2917 (10.5427) acc5: 100.0000 (100.0000) time: 0.1695 data: 0.0749 max mem: 3531 +Test: [ 40/517] eta: 0:02:42 acc1_pos: 4.427 acc5_pos: 100.000 loss: 1.2550 (1.2229) acc1: 4.9479 (9.1082) acc5: 100.0000 (100.0000) time: 0.2553 data: 0.1640 max mem: 3531 +Test: [ 50/517] eta: 0:02:31 acc1_pos: 33.028 acc5_pos: 100.000 loss: 1.1207 (1.1530) acc1: 8.3333 (15.3850) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2639 data: 0.1718 max mem: 3531 +Test: [ 60/517] eta: 0:02:23 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.3587 (1.0190) acc1: 92.4479 (28.4537) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.2595 data: 0.1644 max mem: 3531 +Test: [ 70/517] eta: 0:02:14 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.3090 (0.9197) acc1: 97.9167 (38.2079) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2449 data: 0.1507 max mem: 3531 +Test: [ 80/517] eta: 0:02:09 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.3065 (0.8468) acc1: 97.9167 (45.4861) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2449 data: 0.1505 max mem: 3531 +Test: [ 90/517] eta: 0:02:05 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2786 (0.7819) acc1: 98.1771 (51.1876) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2649 data: 0.1669 max mem: 3531 +Test: [100/517] eta: 0:02:01 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2508 (0.7290) acc1: 98.9583 (55.9432) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2699 data: 0.1719 max mem: 3531 +Test: [110/517] eta: 0:01:57 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2384 (0.6829) acc1: 100.0000 (59.8747) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2599 data: 0.1620 max mem: 3531 +Test: [120/517] eta: 0:01:53 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2565 (0.6506) acc1: 99.7396 (63.1521) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1555 max mem: 3531 +Test: [130/517] eta: 0:01:50 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2857 (0.6221) acc1: 99.7396 (65.9331) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2699 data: 0.1712 max mem: 3531 +Test: [140/517] eta: 0:01:46 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2785 (0.5977) acc1: 99.7396 (68.3289) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2599 data: 0.1647 max mem: 3531 +Test: [150/517] eta: 0:01:43 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2645 (0.5748) acc1: 99.7396 (70.4074) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2649 data: 0.1733 max mem: 3531 +Test: [160/517] eta: 0:01:39 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2743 (0.5571) acc1: 99.4792 (72.1904) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2649 data: 0.1700 max mem: 3531 +Test: [170/517] eta: 0:01:36 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2822 (0.5402) acc1: 98.9583 (73.7055) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2549 data: 0.1577 max mem: 3531 +Test: [180/517] eta: 0:01:33 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2653 (0.5251) acc1: 98.4375 (75.0504) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2649 data: 0.1703 max mem: 3531 +Test: [190/517] eta: 0:01:30 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2095 (0.5079) acc1: 99.4792 (76.3525) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2601 data: 0.1664 max mem: 3531 +Test: [200/517] eta: 0:01:27 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.1995 (0.4929) acc1: 100.0000 (77.5251) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2649 data: 0.1708 max mem: 3531 +Test: [210/517] eta: 0:01:24 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2268 (0.4827) acc1: 99.7396 (78.5261) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2547 data: 0.1590 max mem: 3532 +Test: [220/517] eta: 0:01:21 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2708 (0.4728) acc1: 98.9583 (79.4577) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2449 data: 0.1480 max mem: 3532 +Test: [230/517] eta: 0:01:18 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2756 (0.4649) acc1: 99.2188 (80.2951) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2449 data: 0.1494 max mem: 3532 +Test: [240/517] eta: 0:01:15 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2481 (0.4550) acc1: 99.4792 (81.0987) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2549 data: 0.1585 max mem: 3532 +Test: [250/517] eta: 0:01:12 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2507 (0.4481) acc1: 99.7396 (81.8092) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2649 data: 0.1686 max mem: 3532 +Test: [260/517] eta: 0:01:09 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2736 (0.4415) acc1: 99.2188 (82.4693) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2649 data: 0.1703 max mem: 3532 +Test: [270/517] eta: 0:01:07 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2722 (0.4357) acc1: 99.2188 (83.1027) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2699 data: 0.1748 max mem: 3532 +Test: [280/517] eta: 0:01:04 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2691 (0.4291) acc1: 100.0000 (83.6975) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2699 data: 0.1730 max mem: 3532 +Test: [290/517] eta: 0:01:01 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2441 (0.4230) acc1: 100.0000 (84.2506) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2749 data: 0.1770 max mem: 3532 +Test: [300/517] eta: 0:00:58 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2190 (0.4162) acc1: 100.0000 (84.7721) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2649 data: 0.1673 max mem: 3532 +Test: [310/517] eta: 0:00:56 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2536 (0.4117) acc1: 100.0000 (85.2567) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2699 data: 0.1715 max mem: 3532 +Test: [320/517] eta: 0:00:53 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2679 (0.4073) acc1: 99.7396 (85.6941) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2799 data: 0.1828 max mem: 3532 +Test: [330/517] eta: 0:00:50 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2568 (0.4023) acc1: 99.7396 (86.1208) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2649 data: 0.1678 max mem: 3532 +Test: [340/517] eta: 0:00:48 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2387 (0.3975) acc1: 99.7396 (86.5179) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2749 data: 0.1745 max mem: 3532 +Test: [350/517] eta: 0:00:45 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2505 (0.3937) acc1: 99.7396 (86.8612) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.3399 data: 0.2407 max mem: 3532 +Test: [360/517] eta: 0:00:42 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2543 (0.3901) acc1: 98.9583 (87.1833) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2749 data: 0.1780 max mem: 3532 +Test: [370/517] eta: 0:00:40 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2358 (0.3855) acc1: 100.0000 (87.5288) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2299 data: 0.1340 max mem: 3532 +Test: [380/517] eta: 0:00:37 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2294 (0.3818) acc1: 100.0000 (87.8418) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2999 data: 0.2033 max mem: 3532 +Test: [390/517] eta: 0:00:34 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2554 (0.3789) acc1: 100.0000 (88.1421) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2699 data: 0.1721 max mem: 3532 +Test: [400/517] eta: 0:00:31 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2623 (0.3758) acc1: 100.0000 (88.4358) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2499 data: 0.1527 max mem: 3532 +Test: [410/517] eta: 0:00:29 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2661 (0.3736) acc1: 100.0000 (88.6950) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2649 data: 0.1648 max mem: 3532 +Test: [420/517] eta: 0:00:26 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2922 (0.3718) acc1: 99.7396 (88.9499) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2749 data: 0.1758 max mem: 3532 +Test: [430/517] eta: 0:00:23 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2922 (0.3699) acc1: 99.4792 (89.1676) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2499 data: 0.1540 max mem: 3532 +Test: [440/517] eta: 0:00:20 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2818 (0.3678) acc1: 99.4792 (89.4026) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2349 data: 0.1373 max mem: 3533 +Test: [450/517] eta: 0:00:18 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2818 (0.3659) acc1: 99.7396 (89.6347) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2449 data: 0.1487 max mem: 3533 +Test: [460/517] eta: 0:00:15 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2793 (0.3639) acc1: 100.0000 (89.8584) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2399 data: 0.1447 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2597 (0.3614) acc1: 100.0000 (90.0710) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1570 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2666 (0.3599) acc1: 100.0000 (90.2758) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2449 data: 0.1485 max mem: 3533 +Test: [490/517] eta: 0:00:07 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.3046 (0.3597) acc1: 99.7396 (90.3582) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2119 data: 0.1183 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.3293 (0.3595) acc1: 97.3958 (90.5034) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1774 data: 0.0864 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.3279 (0.3583) acc1: 98.9583 (90.6423) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1552 data: 0.0645 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 33.028 acc5_pos: 100.000 loss: 0.2566 (0.3570) acc1: 99.4792 (90.7457) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1499 data: 0.0573 max mem: 3533 +Test: Total time: 0:02:15 (0.2621 s / it) +* Overall Acc@1 90.746 Acc@5 100.000 + target == real Acc@1 11.282 Acc@5 100.000 + target == fake Acc@1 98.989 Acc@5 100.000 + loss 0.357 + PR-AUC 0.9563 ROC-AUC 0.7313 EER 33.27% (thr=0.7471) +Accuracy of the network on the 198427 test images: 90.7% +[rank0]:[W726 06:32:08.577873899 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 06:32:27.643000 1204163 torch/distributed/run.py:766] +W0726 06:32:27.643000 1204163 torch/distributed/run.py:766] ***************************************** +W0726 06:32:27.643000 1204163 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:32:27.643000 1204163 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 06:32:59.651402369 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:32:59.001645998 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:32:59.072561159 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:32:59.202270812 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:32:59.297018524 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:32:59.464753236 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:32:59.481117120 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:33:00.661576003 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n16_seed9_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed9_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:43:22 acc1_pos: 22.396 acc5_pos: 100.000 loss: 0.9389 (0.9389) acc1: 22.3958 (22.3958) acc5: 100.0000 (100.0000) time: 6.6906 data: 5.0218 max mem: 3531 +Test: [ 10/389] eta: 0:04:37 acc1_pos: 61.198 acc5_pos: 100.000 loss: 0.7970 (0.7965) acc1: 37.7604 (41.3352) acc5: 100.0000 (100.0000) time: 0.7313 data: 0.4935 max mem: 3531 +Test: [ 20/389] eta: 0:02:38 acc1_pos: 11.458 acc5_pos: 100.000 loss: 0.8349 (0.8559) acc1: 27.6042 (32.2297) acc5: 100.0000 (100.0000) time: 0.1176 data: 0.0206 max mem: 3531 +Test: [ 30/389] eta: 0:02:15 acc1_pos: 10.156 acc5_pos: 100.000 loss: 0.9083 (0.8708) acc1: 16.9271 (28.7550) acc5: 100.0000 (100.0000) time: 0.1849 data: 0.0867 max mem: 3531 +Test: [ 40/389] eta: 0:01:59 acc1_pos: 3.906 acc5_pos: 100.000 loss: 0.9019 (0.8852) acc1: 22.3958 (28.3537) acc5: 100.0000 (100.0000) time: 0.2499 data: 0.1536 max mem: 3531 +Test: [ 50/389] eta: 0:01:43 acc1_pos: 85.677 acc5_pos: 100.000 loss: 0.8497 (0.8562) acc1: 30.9896 (32.6031) acc5: 100.0000 (100.0000) time: 0.1949 data: 0.0997 max mem: 3531 +Test: [ 60/389] eta: 0:01:32 acc1_pos: 11.979 acc5_pos: 100.000 loss: 0.6748 (0.8130) acc1: 54.1667 (37.9739) acc5: 100.0000 (100.0000) time: 0.1599 data: 0.0664 max mem: 3531 +Test: [ 70/389] eta: 0:01:25 acc1_pos: 28.385 acc5_pos: 100.000 loss: 0.6750 (0.8184) acc1: 51.8229 (37.4890) acc5: 100.0000 (100.0000) time: 0.1699 data: 0.0757 max mem: 3531 +Test: [ 80/389] eta: 0:01:19 acc1_pos: 82.552 acc5_pos: 100.000 loss: 0.8840 (0.8121) acc1: 28.1250 (38.6060) acc5: 100.0000 (100.0000) time: 0.1849 data: 0.0860 max mem: 3531 +Test: [ 90/389] eta: 0:01:15 acc1_pos: 13.542 acc5_pos: 100.000 loss: 0.5888 (0.7940) acc1: 63.8021 (40.5792) acc5: 100.0000 (100.0000) time: 0.1899 data: 0.0901 max mem: 3531 +Test: [100/389] eta: 0:01:09 acc1_pos: 4.688 acc5_pos: 100.000 loss: 0.8857 (0.8104) acc1: 27.6042 (38.7686) acc5: 100.0000 (100.0000) time: 0.1749 data: 0.0774 max mem: 3531 +Test: [110/389] eta: 0:01:06 acc1_pos: 59.635 acc5_pos: 100.000 loss: 0.9719 (0.8205) acc1: 21.6146 (38.1991) acc5: 100.0000 (100.0000) time: 0.1799 data: 0.0824 max mem: 3531 +Test: [120/389] eta: 0:01:02 acc1_pos: 25.000 acc5_pos: 100.000 loss: 0.8605 (0.8186) acc1: 40.6250 (38.5374) acc5: 100.0000 (100.0000) time: 0.1814 data: 0.0849 max mem: 3531 +Test: [130/389] eta: 0:00:58 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.6759 (0.8001) acc1: 55.4688 (40.9212) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1599 data: 0.0663 max mem: 3531 +Test: [140/389] eta: 0:00:56 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.4230 (0.7765) acc1: 89.0625 (43.9753) acc5: 100.0000 (100.0000) acc1_neg: 45.573 acc5_neg: 100.000 time: 0.1825 data: 0.0883 max mem: 3531 +Test: [150/389] eta: 0:00:52 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.4100 (0.7589) acc1: 92.7083 (46.3283) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1809 data: 0.0879 max mem: 3531 +Test: [160/389] eta: 0:00:48 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.4103 (0.7362) acc1: 92.7083 (49.0829) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1358 data: 0.0430 max mem: 3531 +Test: [170/389] eta: 0:00:47 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.2733 (0.7109) acc1: 100.0000 (51.6600) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1889 data: 0.0919 max mem: 3531 +Test: [180/389] eta: 0:00:45 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.3603 (0.6952) acc1: 94.2708 (53.7710) acc5: 100.0000 (100.0000) acc1_neg: 68.490 acc5_neg: 100.000 time: 0.2698 data: 0.1731 max mem: 3531 +Test: [190/389] eta: 0:00:43 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.3712 (0.6837) acc1: 90.1042 (55.1988) acc5: 100.0000 (100.0000) acc1_neg: 42.188 acc5_neg: 100.000 time: 0.2406 data: 0.1458 max mem: 3531 +Test: [200/389] eta: 0:00:41 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.3775 (0.6746) acc1: 95.3125 (56.4534) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2056 data: 0.1117 max mem: 3531 +Test: [210/389] eta: 0:00:38 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.3582 (0.6568) acc1: 98.6979 (58.4753) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1841 data: 0.0908 max mem: 3532 +Test: [220/389] eta: 0:00:36 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.2712 (0.6400) acc1: 99.4792 (60.2270) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1661 data: 0.0713 max mem: 3532 +Test: [230/389] eta: 0:00:33 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.3837 (0.6313) acc1: 97.9167 (61.4583) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.1802 data: 0.0862 max mem: 3532 +Test: [240/389] eta: 0:00:31 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.4023 (0.6216) acc1: 89.8438 (62.7539) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1980 data: 0.1047 max mem: 3532 +Test: [250/389] eta: 0:00:29 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.3378 (0.6094) acc1: 96.0938 (64.1642) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2004 data: 0.1080 max mem: 3532 +Test: [260/389] eta: 0:00:26 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.3864 (0.6024) acc1: 97.9167 (65.2039) acc5: 100.0000 (100.0000) acc1_neg: 69.531 acc5_neg: 100.000 time: 0.1609 data: 0.0701 max mem: 3532 +Test: [270/389] eta: 0:00:24 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.4268 (0.5978) acc1: 94.5312 (65.9123) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1523 data: 0.0599 max mem: 3532 +Test: [280/389] eta: 0:00:22 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.3927 (0.5931) acc1: 89.8438 (66.3330) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1966 data: 0.1031 max mem: 3532 +Test: [290/389] eta: 0:00:20 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.4268 (0.5887) acc1: 84.8958 (66.8725) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.2131 data: 0.1192 max mem: 3532 +Test: [300/389] eta: 0:00:18 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.4514 (0.5848) acc1: 87.5000 (67.5578) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.2296 data: 0.1356 max mem: 3532 +Test: [310/389] eta: 0:00:16 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.4779 (0.5855) acc1: 82.8125 (67.5224) acc5: 100.0000 (100.0000) acc1_neg: 27.604 acc5_neg: 100.000 time: 0.2385 data: 0.1455 max mem: 3532 +Test: [320/389] eta: 0:00:14 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.5299 (0.5809) acc1: 81.5104 (68.2332) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2004 data: 0.1074 max mem: 3532 +Test: [330/389] eta: 0:00:12 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.5873 (0.5828) acc1: 72.9167 (68.0695) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.1834 data: 0.0905 max mem: 3532 +Test: [340/389] eta: 0:00:10 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.3846 (0.5750) acc1: 70.0521 (68.8974) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1757 data: 0.0844 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.3087 (0.5788) acc1: 98.1771 (68.6172) acc5: 100.0000 (100.0000) acc1_neg: 38.542 acc5_neg: 100.000 time: 0.1542 data: 0.0639 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.4508 (0.5783) acc1: 85.1562 (68.7695) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.1452 data: 0.0551 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.4340 (0.5725) acc1: 87.2396 (69.4814) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1391 data: 0.0493 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.3885 (0.5687) acc1: 92.7083 (69.9673) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1330 data: 0.0426 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 17.178 acc5_pos: 100.000 loss: 0.3885 (0.5646) acc1: 89.5833 (70.4274) acc5: 100.0000 (100.0000) acc1_neg: 88.525 acc5_neg: 100.000 time: 0.1234 data: 0.0294 max mem: 3532 +Test: Total time: 0:01:17 (0.1984 s / it) +* Overall Acc@1 70.427 Acc@5 100.000 + target == real Acc@1 38.156 Acc@5 100.000 + target == fake Acc@1 85.428 Acc@5 100.000 + loss 0.565 + PR-AUC 0.8474 ROC-AUC 0.7259 EER 34.05% (thr=0.6250) +Accuracy of the network on the 149358 test images: 70.4% +[rank0]:[W726 06:34:25.426252300 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 06:34:45.245000 1204721 torch/distributed/run.py:766] +W0726 06:34:45.245000 1204721 torch/distributed/run.py:766] ***************************************** +W0726 06:34:45.245000 1204721 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:34:45.245000 1204721 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:35:14.799526770 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 06:35:14.821274832 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:35:14.003621815 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:35:14.154237209 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +| distributed init (rank 6): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +[rank6]:[W726 06:35:14.249339740 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank1]:[W726 06:35:14.249384925 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W726 06:35:14.249835688 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:35:14.264331172 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n16_seed9_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed9_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:38:26 acc1_pos: 68.229 acc5_pos: 100.000 loss: 0.5719 (0.5719) acc1: 68.2292 (68.2292) acc5: 100.0000 (100.0000) time: 6.9265 data: 5.5627 max mem: 3531 +Test: [ 10/333] eta: 0:03:51 acc1_pos: 76.823 acc5_pos: 100.000 loss: 0.4204 (0.4347) acc1: 76.5625 (76.2784) acc5: 100.0000 (100.0000) time: 0.7156 data: 0.5061 max mem: 3531 +Test: [ 20/333] eta: 0:02:12 acc1_pos: 88.021 acc5_pos: 100.000 loss: 0.4161 (0.4195) acc1: 78.1250 (77.5794) acc5: 100.0000 (100.0000) time: 0.0972 data: 0.0004 max mem: 3531 +Test: [ 30/333] eta: 0:01:57 acc1_pos: 46.094 acc5_pos: 100.000 loss: 0.4866 (0.5722) acc1: 69.0104 (65.2722) acc5: 100.0000 (100.0000) time: 0.2058 data: 0.1108 max mem: 3531 +Test: [ 40/333] eta: 0:01:46 acc1_pos: 35.677 acc5_pos: 100.000 loss: 0.8969 (0.6514) acc1: 37.7604 (58.8415) acc5: 100.0000 (100.0000) time: 0.2999 data: 0.2087 max mem: 3531 +Test: [ 50/333] eta: 0:01:34 acc1_pos: 41.146 acc5_pos: 100.000 loss: 0.8938 (0.7043) acc1: 37.7604 (54.1054) acc5: 100.0000 (100.0000) time: 0.2540 data: 0.1616 max mem: 3531 +Test: [ 60/333] eta: 0:01:28 acc1_pos: 34.635 acc5_pos: 100.000 loss: 0.9013 (0.7413) acc1: 35.9375 (51.0459) acc5: 100.0000 (100.0000) time: 0.2499 data: 0.1544 max mem: 3531 +Test: [ 70/333] eta: 0:01:21 acc1_pos: 39.844 acc5_pos: 100.000 loss: 0.8989 (0.7614) acc1: 36.9792 (49.2884) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1486 max mem: 3531 +Test: [ 80/333] eta: 0:01:13 acc1_pos: 50.000 acc5_pos: 100.000 loss: 0.8882 (0.7786) acc1: 39.3229 (48.0228) acc5: 100.0000 (100.0000) time: 0.1849 data: 0.0913 max mem: 3531 +Test: [ 90/333] eta: 0:01:05 acc1_pos: 49.219 acc5_pos: 100.000 loss: 0.8603 (0.7875) acc1: 40.6250 (47.2241) acc5: 100.0000 (100.0000) time: 0.1310 data: 0.0381 max mem: 3531 +Test: [100/333] eta: 0:01:00 acc1_pos: 43.229 acc5_pos: 100.000 loss: 0.8407 (0.7951) acc1: 39.3229 (46.4315) acc5: 100.0000 (100.0000) time: 0.1382 data: 0.0432 max mem: 3531 +Test: [110/333] eta: 0:00:56 acc1_pos: 38.281 acc5_pos: 100.000 loss: 0.8508 (0.8009) acc1: 39.0625 (45.8474) acc5: 100.0000 (100.0000) time: 0.1798 data: 0.0850 max mem: 3531 +Test: [120/333] eta: 0:00:52 acc1_pos: 44.010 acc5_pos: 100.000 loss: 0.8427 (0.8039) acc1: 39.0625 (45.6504) acc5: 100.0000 (100.0000) time: 0.1716 data: 0.0788 max mem: 3531 +Test: [130/333] eta: 0:00:49 acc1_pos: 51.562 acc5_pos: 100.000 loss: 0.7971 (0.8050) acc1: 44.7917 (45.6465) acc5: 100.0000 (100.0000) time: 0.1790 data: 0.0867 max mem: 3531 +Test: [140/333] eta: 0:00:46 acc1_pos: 61.198 acc5_pos: 100.000 loss: 0.7931 (0.8064) acc1: 44.7917 (45.4732) acc5: 100.0000 (100.0000) time: 0.2160 data: 0.1252 max mem: 3531 +Test: [150/333] eta: 0:00:43 acc1_pos: 39.323 acc5_pos: 100.000 loss: 0.8345 (0.8098) acc1: 39.3229 (45.1245) acc5: 100.0000 (100.0000) time: 0.2002 data: 0.1085 max mem: 3531 +Test: [160/333] eta: 0:00:40 acc1_pos: 26.974 acc5_pos: 100.000 loss: 0.8412 (0.8113) acc1: 43.2292 (45.2429) acc5: 100.0000 (100.0000) acc1_neg: 57.031 acc5_neg: 100.000 time: 0.1888 data: 0.0942 max mem: 3531 +Test: [170/333] eta: 0:00:37 acc1_pos: 26.974 acc5_pos: 100.000 loss: 0.7726 (0.8040) acc1: 57.0312 (46.4227) acc5: 100.0000 (100.0000) acc1_neg: 67.708 acc5_neg: 100.000 time: 0.1995 data: 0.1019 max mem: 3531 +Test: [180/333] eta: 0:00:35 acc1_pos: 26.974 acc5_pos: 100.000 loss: 0.6267 (0.7930) acc1: 67.9688 (47.7296) acc5: 100.0000 (100.0000) acc1_neg: 75.781 acc5_neg: 100.000 time: 0.1861 data: 0.0871 max mem: 3531 +Test: [190/333] eta: 0:00:32 acc1_pos: 26.974 acc5_pos: 100.000 loss: 0.5817 (0.7792) acc1: 71.3542 (49.1901) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.1818 data: 0.0869 max mem: 3531 +Test: [200/333] eta: 0:00:30 acc1_pos: 26.974 acc5_pos: 100.000 loss: 0.5465 (0.7687) acc1: 72.3958 (50.4353) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.1938 data: 0.1013 max mem: 3531 +Test: [210/333] eta: 0:00:27 acc1_pos: 26.974 acc5_pos: 100.000 loss: 0.6252 (0.7619) acc1: 71.0938 (51.4045) acc5: 100.0000 (100.0000) acc1_neg: 71.354 acc5_neg: 100.000 time: 0.1988 data: 0.1032 max mem: 3532 +Test: [220/333] eta: 0:00:24 acc1_pos: 26.974 acc5_pos: 100.000 loss: 0.6276 (0.7548) acc1: 70.8333 (52.2283) acc5: 100.0000 (100.0000) acc1_neg: 77.083 acc5_neg: 100.000 time: 0.1549 data: 0.0579 max mem: 3532 +Test: [230/333] eta: 0:00:22 acc1_pos: 26.974 acc5_pos: 100.000 loss: 0.5249 (0.7436) acc1: 77.0833 (53.4711) acc5: 100.0000 (100.0000) acc1_neg: 77.083 acc5_neg: 100.000 time: 0.1245 data: 0.0284 max mem: 3532 +Test: [240/333] eta: 0:00:20 acc1_pos: 26.974 acc5_pos: 100.000 loss: 0.5119 (0.7344) acc1: 77.8646 (54.4401) acc5: 100.0000 (100.0000) acc1_neg: 79.427 acc5_neg: 100.000 time: 0.1753 data: 0.0798 max mem: 3532 +Test: [250/333] eta: 0:00:17 acc1_pos: 26.974 acc5_pos: 100.000 loss: 0.5266 (0.7276) acc1: 75.2604 (55.2114) acc5: 100.0000 (100.0000) acc1_neg: 64.844 acc5_neg: 100.000 time: 0.1895 data: 0.0935 max mem: 3532 +Test: [260/333] eta: 0:00:15 acc1_pos: 26.974 acc5_pos: 100.000 loss: 0.5615 (0.7223) acc1: 72.6562 (55.8818) acc5: 100.0000 (100.0000) acc1_neg: 71.615 acc5_neg: 100.000 time: 0.1657 data: 0.0715 max mem: 3532 +Test: [270/333] eta: 0:00:13 acc1_pos: 26.974 acc5_pos: 100.000 loss: 0.5561 (0.7159) acc1: 72.9167 (56.5662) acc5: 100.0000 (100.0000) acc1_neg: 72.135 acc5_neg: 100.000 time: 0.1899 data: 0.0958 max mem: 3532 +Test: [280/333] eta: 0:00:11 acc1_pos: 26.974 acc5_pos: 100.000 loss: 0.5480 (0.7116) acc1: 73.9583 (57.0859) acc5: 100.0000 (100.0000) acc1_neg: 77.344 acc5_neg: 100.000 time: 0.1988 data: 0.1016 max mem: 3532 +Test: [290/333] eta: 0:00:08 acc1_pos: 26.974 acc5_pos: 100.000 loss: 0.6497 (0.7104) acc1: 66.6667 (57.3230) acc5: 100.0000 (100.0000) acc1_neg: 63.802 acc5_neg: 100.000 time: 0.1719 data: 0.0757 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 26.974 acc5_pos: 100.000 loss: 0.6909 (0.7105) acc1: 63.2812 (57.5157) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.1622 data: 0.0695 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 26.974 acc5_pos: 100.000 loss: 0.6518 (0.7040) acc1: 65.1042 (58.1634) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.1734 data: 0.0821 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 26.974 acc5_pos: 100.000 loss: 0.4953 (0.6987) acc1: 80.4688 (58.7674) acc5: 100.0000 (100.0000) acc1_neg: 60.417 acc5_neg: 100.000 time: 0.1602 data: 0.0674 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 26.974 acc5_pos: 100.000 loss: 0.6565 (0.7000) acc1: 66.4062 (58.8786) acc5: 100.0000 (100.0000) acc1_neg: 66.406 acc5_neg: 100.000 time: 0.1342 data: 0.0420 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 26.974 acc5_pos: 100.000 loss: 0.7101 (0.7002) acc1: 64.3229 (58.9105) acc5: 100.0000 (100.0000) acc1_neg: 64.507 acc5_neg: 100.000 time: 0.1356 data: 0.0400 max mem: 3532 +Test: Total time: 0:01:07 (0.2027 s / it) +* Overall Acc@1 58.911 Acc@5 100.000 + target == real Acc@1 44.984 Acc@5 100.000 + target == fake Acc@1 71.308 Acc@5 100.000 + loss 0.700 + PR-AUC 0.6394 ROC-AUC 0.6229 EER 41.75% (thr=0.5889) +Accuracy of the network on the 127843 test images: 58.9% +[rank0]:[W726 06:36:34.858061514 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 06:36:55.199000 1205280 torch/distributed/run.py:766] +W0726 06:36:55.199000 1205280 torch/distributed/run.py:766] ***************************************** +W0726 06:36:55.199000 1205280 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:36:55.199000 1205280 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:37:25.145458771 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:37:25.319512671 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W726 06:37:25.319519082 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:37:26.540773059 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:37:26.656231719 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 06:37:26.725909686 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:37:26.776708232 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +NCCL version 2.26.2+cuda12.2 +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:37:26.870510777 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n16_seed9_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed9_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:22:21 acc1_pos: 97.135 acc5_pos: 100.000 loss: 0.1850 (0.1850) acc1: 97.1354 (97.1354) acc5: 100.0000 (100.0000) time: 5.7093 data: 4.1706 max mem: 3531 +Test: [ 10/235] eta: 0:02:15 acc1_pos: 95.833 acc5_pos: 100.000 loss: 0.1879 (0.1881) acc1: 96.3542 (96.1885) acc5: 100.0000 (100.0000) time: 0.6008 data: 0.3794 max mem: 3531 +Test: [ 20/235] eta: 0:01:16 acc1_pos: 96.354 acc5_pos: 100.000 loss: 0.1851 (0.1847) acc1: 96.3542 (96.1558) acc5: 100.0000 (100.0000) time: 0.0902 data: 0.0004 max mem: 3531 +Test: [ 30/235] eta: 0:00:58 acc1_pos: 95.573 acc5_pos: 100.000 loss: 0.1808 (0.1845) acc1: 96.3542 (96.0686) acc5: 100.0000 (100.0000) time: 0.1105 data: 0.0180 max mem: 3531 +Test: [ 40/235] eta: 0:00:52 acc1_pos: 94.792 acc5_pos: 100.000 loss: 0.1811 (0.1842) acc1: 96.0938 (96.0175) acc5: 100.0000 (100.0000) time: 0.1793 data: 0.0865 max mem: 3531 +Test: [ 50/235] eta: 0:00:49 acc1_pos: 93.229 acc5_pos: 100.000 loss: 0.1845 (0.1859) acc1: 96.0938 (95.8435) acc5: 100.0000 (100.0000) time: 0.2363 data: 0.1430 max mem: 3531 +Test: [ 60/235] eta: 0:00:45 acc1_pos: 94.271 acc5_pos: 100.000 loss: 0.1903 (0.1873) acc1: 94.7917 (95.7095) acc5: 100.0000 (100.0000) time: 0.2372 data: 0.1420 max mem: 3531 +Test: [ 70/235] eta: 0:00:41 acc1_pos: 96.923 acc5_pos: 100.000 loss: 0.2066 (0.3750) acc1: 93.4896 (85.4827) acc5: 100.0000 (100.0000) acc1_neg: 12.760 acc5_neg: 100.000 time: 0.2126 data: 0.1194 max mem: 3531 +Test: [ 80/235] eta: 0:00:37 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.7042 (0.5412) acc1: 13.5417 (76.5271) acc5: 100.0000 (100.0000) acc1_neg: 9.115 acc5_neg: 100.000 time: 0.1826 data: 0.0905 max mem: 3531 +Test: [ 90/235] eta: 0:00:35 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.7161 (0.6698) acc1: 13.2812 (69.6543) acc5: 100.0000 (100.0000) acc1_neg: 11.719 acc5_neg: 100.000 time: 0.2349 data: 0.1419 max mem: 3531 +Test: [100/235] eta: 0:00:33 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.7260 (0.7736) acc1: 13.2812 (64.0006) acc5: 100.0000 (100.0000) acc1_neg: 13.281 acc5_neg: 100.000 time: 0.2920 data: 0.1977 max mem: 3531 +Test: [110/235] eta: 0:00:29 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.6932 (0.8567) acc1: 13.2812 (59.4477) acc5: 100.0000 (100.0000) acc1_neg: 11.458 acc5_neg: 100.000 time: 0.1899 data: 0.0949 max mem: 3531 +Test: [120/235] eta: 0:00:27 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.6593 (0.9189) acc1: 15.1042 (56.0692) acc5: 100.0000 (100.0000) acc1_neg: 21.094 acc5_neg: 100.000 time: 0.1778 data: 0.0844 max mem: 3531 +Test: [130/235] eta: 0:00:24 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.5908 (0.9695) acc1: 20.3125 (53.3655) acc5: 100.0000 (100.0000) acc1_neg: 22.135 acc5_neg: 100.000 time: 0.2349 data: 0.1422 max mem: 3531 +Test: [140/235] eta: 0:00:22 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.6083 (1.0151) acc1: 19.7917 (50.9438) acc5: 100.0000 (100.0000) acc1_neg: 17.708 acc5_neg: 100.000 time: 0.1928 data: 0.0989 max mem: 3531 +Test: [150/235] eta: 0:00:19 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.5920 (1.0520) acc1: 19.7917 (48.9152) acc5: 100.0000 (100.0000) acc1_neg: 22.135 acc5_neg: 100.000 time: 0.1870 data: 0.0946 max mem: 3531 +Test: [160/235] eta: 0:00:17 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.5703 (1.0850) acc1: 20.5729 (47.1516) acc5: 100.0000 (100.0000) acc1_neg: 19.271 acc5_neg: 100.000 time: 0.2069 data: 0.1166 max mem: 3531 +Test: [170/235] eta: 0:00:14 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.6148 (1.1266) acc1: 18.2292 (44.9957) acc5: 100.0000 (100.0000) acc1_neg: 7.031 acc5_neg: 100.000 time: 0.2031 data: 0.1114 max mem: 3531 +Test: [180/235] eta: 0:00:12 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.8219 (1.1659) acc1: 10.1562 (43.1040) acc5: 100.0000 (100.0000) acc1_neg: 9.896 acc5_neg: 100.000 time: 0.1927 data: 0.0992 max mem: 3531 +Test: [190/235] eta: 0:00:10 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.8192 (1.2004) acc1: 9.8958 (41.3613) acc5: 100.0000 (100.0000) acc1_neg: 9.896 acc5_neg: 100.000 time: 0.1896 data: 0.0949 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.8301 (1.2318) acc1: 9.8958 (39.7893) acc5: 100.0000 (100.0000) acc1_neg: 8.594 acc5_neg: 100.000 time: 0.1771 data: 0.0822 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.8325 (1.2604) acc1: 9.3750 (38.3565) acc5: 100.0000 (100.0000) acc1_neg: 8.854 acc5_neg: 100.000 time: 0.1702 data: 0.0780 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.8355 (1.2868) acc1: 9.3750 (37.0522) acc5: 100.0000 (100.0000) acc1_neg: 8.333 acc5_neg: 100.000 time: 0.1571 data: 0.0669 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.8366 (1.3108) acc1: 9.3750 (35.8845) acc5: 100.0000 (100.0000) acc1_neg: 9.375 acc5_neg: 100.000 time: 0.1405 data: 0.0506 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.8366 (1.3210) acc1: 9.3750 (35.5356) acc5: 100.0000 (100.0000) acc1_neg: 8.696 acc5_neg: 100.000 time: 0.1280 data: 0.0416 max mem: 3532 +Test: Total time: 0:00:49 (0.2098 s / it) +* Overall Acc@1 35.536 Acc@5 100.000 + target == real Acc@1 95.715 Acc@5 100.000 + target == fake Acc@1 13.770 Acc@5 100.000 + loss 1.321 + PR-AUC 0.8358 ROC-AUC 0.6605 EER 38.46% (thr=0.1262) +Accuracy of the network on the 89879 test images: 35.5% +[rank0]:[W726 06:38:24.383935416 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 06:38:44.182000 1205826 torch/distributed/run.py:766] +W0726 06:38:44.182000 1205826 torch/distributed/run.py:766] ***************************************** +W0726 06:38:44.182000 1205826 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:38:44.182000 1205826 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:39:12.171168747 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:39:12.436706831 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:39:13.655089210 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:39:13.686598993 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:39:13.715293627 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:39:13.725851428 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 06:39:13.736990267 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:39:13.767447182 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n16_seed9_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed9_resampled_n116-041-153_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:28:41 acc1_pos: 71.615 acc5_pos: 100.000 loss: 0.5525 (0.5525) acc1: 71.6146 (71.6146) acc5: 100.0000 (100.0000) time: 6.3991 data: 4.9078 max mem: 3531 +Test: [ 10/269] eta: 0:02:51 acc1_pos: 79.948 acc5_pos: 100.000 loss: 0.5310 (0.5309) acc1: 72.9167 (75.0000) acc5: 100.0000 (100.0000) time: 0.6635 data: 0.4465 max mem: 3531 +Test: [ 20/269] eta: 0:01:38 acc1_pos: 75.781 acc5_pos: 100.000 loss: 0.5304 (0.5313) acc1: 75.7812 (75.1860) acc5: 100.0000 (100.0000) time: 0.0945 data: 0.0011 max mem: 3531 +Test: [ 30/269] eta: 0:01:27 acc1_pos: 70.833 acc5_pos: 100.000 loss: 0.5249 (0.5262) acc1: 77.3438 (75.7897) acc5: 100.0000 (100.0000) time: 0.1995 data: 0.1046 max mem: 3531 +Test: [ 40/269] eta: 0:01:19 acc1_pos: 80.172 acc5_pos: 100.000 loss: 0.6088 (0.5702) acc1: 64.5833 (71.4431) acc5: 100.0000 (100.0000) acc1_neg: 63.021 acc5_neg: 100.000 time: 0.2965 data: 0.2035 max mem: 3531 +Test: [ 50/269] eta: 0:01:09 acc1_pos: 80.172 acc5_pos: 100.000 loss: 0.6501 (0.5948) acc1: 63.0208 (69.1330) acc5: 100.0000 (100.0000) acc1_neg: 45.312 acc5_neg: 100.000 time: 0.2506 data: 0.1585 max mem: 3531 +Test: [ 60/269] eta: 0:01:04 acc1_pos: 80.172 acc5_pos: 100.000 loss: 0.7108 (0.6237) acc1: 54.9479 (65.9964) acc5: 100.0000 (100.0000) acc1_neg: 32.031 acc5_neg: 100.000 time: 0.2283 data: 0.1367 max mem: 3531 +Test: [ 70/269] eta: 0:00:58 acc1_pos: 80.172 acc5_pos: 100.000 loss: 0.7538 (0.6460) acc1: 45.3125 (63.4243) acc5: 100.0000 (100.0000) acc1_neg: 47.656 acc5_neg: 100.000 time: 0.2309 data: 0.1378 max mem: 3531 +Test: [ 80/269] eta: 0:00:52 acc1_pos: 80.172 acc5_pos: 100.000 loss: 0.7498 (0.6574) acc1: 48.9583 (61.8795) acc5: 100.0000 (100.0000) acc1_neg: 42.448 acc5_neg: 100.000 time: 0.1905 data: 0.0970 max mem: 3531 +Test: [ 90/269] eta: 0:00:47 acc1_pos: 80.172 acc5_pos: 100.000 loss: 0.7972 (0.6887) acc1: 42.4479 (58.5995) acc5: 100.0000 (100.0000) acc1_neg: 38.281 acc5_neg: 100.000 time: 0.1741 data: 0.0809 max mem: 3531 +Test: [100/269] eta: 0:00:44 acc1_pos: 80.172 acc5_pos: 100.000 loss: 0.8860 (0.7074) acc1: 30.9896 (56.3557) acc5: 100.0000 (100.0000) acc1_neg: 30.729 acc5_neg: 100.000 time: 0.2143 data: 0.1222 max mem: 3531 +Test: [110/269] eta: 0:00:41 acc1_pos: 80.172 acc5_pos: 100.000 loss: 0.9598 (0.7372) acc1: 27.8646 (53.3596) acc5: 100.0000 (100.0000) acc1_neg: 19.010 acc5_neg: 100.000 time: 0.2200 data: 0.1270 max mem: 3531 +Test: [120/269] eta: 0:00:38 acc1_pos: 80.172 acc5_pos: 100.000 loss: 0.9280 (0.7425) acc1: 26.3021 (52.8581) acc5: 100.0000 (100.0000) acc1_neg: 53.385 acc5_neg: 100.000 time: 0.1999 data: 0.1061 max mem: 3531 +Test: [130/269] eta: 0:00:35 acc1_pos: 80.172 acc5_pos: 100.000 loss: 0.8098 (0.7532) acc1: 39.3229 (51.8289) acc5: 100.0000 (100.0000) acc1_neg: 32.552 acc5_neg: 100.000 time: 0.2403 data: 0.1463 max mem: 3531 +Test: [140/269] eta: 0:00:32 acc1_pos: 80.172 acc5_pos: 100.000 loss: 0.9140 (0.7694) acc1: 31.7708 (50.0351) acc5: 100.0000 (100.0000) acc1_neg: 14.323 acc5_neg: 100.000 time: 0.2405 data: 0.1461 max mem: 3531 +Test: [150/269] eta: 0:00:29 acc1_pos: 80.172 acc5_pos: 100.000 loss: 0.9378 (0.7750) acc1: 30.4688 (49.3722) acc5: 100.0000 (100.0000) acc1_neg: 68.490 acc5_neg: 100.000 time: 0.1701 data: 0.0772 max mem: 3531 +Test: [160/269] eta: 0:00:26 acc1_pos: 80.172 acc5_pos: 100.000 loss: 0.7406 (0.7730) acc1: 51.3021 (49.6280) acc5: 100.0000 (100.0000) acc1_neg: 69.271 acc5_neg: 100.000 time: 0.1806 data: 0.0895 max mem: 3531 +Test: [170/269] eta: 0:00:24 acc1_pos: 80.172 acc5_pos: 100.000 loss: 0.7687 (0.7735) acc1: 48.1771 (49.5660) acc5: 100.0000 (100.0000) acc1_neg: 47.396 acc5_neg: 100.000 time: 0.2358 data: 0.1457 max mem: 3531 +Test: [180/269] eta: 0:00:21 acc1_pos: 80.172 acc5_pos: 100.000 loss: 0.6901 (0.7665) acc1: 58.8542 (50.3597) acc5: 100.0000 (100.0000) acc1_neg: 50.781 acc5_neg: 100.000 time: 0.2350 data: 0.1442 max mem: 3531 +Test: [190/269] eta: 0:00:18 acc1_pos: 80.172 acc5_pos: 100.000 loss: 0.6606 (0.7679) acc1: 60.4167 (50.1091) acc5: 100.0000 (100.0000) acc1_neg: 57.031 acc5_neg: 100.000 time: 0.1974 data: 0.1059 max mem: 3531 +Test: [200/269] eta: 0:00:16 acc1_pos: 80.172 acc5_pos: 100.000 loss: 0.7768 (0.7688) acc1: 45.5729 (50.1464) acc5: 100.0000 (100.0000) acc1_neg: 72.917 acc5_neg: 100.000 time: 0.1712 data: 0.0797 max mem: 3531 +Test: [210/269] eta: 0:00:13 acc1_pos: 80.172 acc5_pos: 100.000 loss: 0.7208 (0.7668) acc1: 48.6979 (50.1098) acc5: 100.0000 (100.0000) acc1_neg: 54.948 acc5_neg: 100.000 time: 0.1896 data: 0.0990 max mem: 3532 +Test: [220/269] eta: 0:00:11 acc1_pos: 80.172 acc5_pos: 100.000 loss: 0.7049 (0.7649) acc1: 51.3021 (50.2380) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.1696 data: 0.0794 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 80.172 acc5_pos: 100.000 loss: 0.6890 (0.7611) acc1: 53.1250 (50.5637) acc5: 100.0000 (100.0000) acc1_neg: 51.042 acc5_neg: 100.000 time: 0.1257 data: 0.0355 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 80.172 acc5_pos: 100.000 loss: 0.7460 (0.7668) acc1: 50.2604 (50.0378) acc5: 100.0000 (100.0000) acc1_neg: 35.156 acc5_neg: 100.000 time: 0.1135 data: 0.0237 max mem: 3532 +Test: [250/269] eta: 0:00:04 acc1_pos: 80.172 acc5_pos: 100.000 loss: 0.8815 (0.7719) acc1: 35.1562 (49.4771) acc5: 100.0000 (100.0000) acc1_neg: 15.625 acc5_neg: 100.000 time: 0.1191 data: 0.0284 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 80.172 acc5_pos: 100.000 loss: 0.9086 (0.7789) acc1: 28.1250 (48.7209) acc5: 100.0000 (100.0000) acc1_neg: 39.583 acc5_neg: 100.000 time: 0.1231 data: 0.0321 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 80.172 acc5_pos: 100.000 loss: 0.8641 (0.7807) acc1: 32.2917 (48.3953) acc5: 100.0000 (100.0000) acc1_neg: 39.108 acc5_neg: 100.000 time: 0.1089 data: 0.0150 max mem: 3532 +Test: Total time: 0:00:56 (0.2101 s / it) +* Overall Acc@1 48.395 Acc@5 100.000 + target == real Acc@1 75.508 Acc@5 100.000 + target == fake Acc@1 44.695 Acc@5 100.000 + loss 0.781 + PR-AUC 0.9266 ROC-AUC 0.6497 EER 39.58% (thr=0.4233) +Accuracy of the network on the 103293 test images: 48.4% +[rank0]:[W726 06:40:19.708011203 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T06:40:26Z] eval_done tag=fixed500k_fake_n16_seed9_resampled +eval_queue_done_utc=2026-07-26T06:40:26Z diff --git a/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_1958_seed5_9/eval_queue_n116-041-153.pid b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_1958_seed5_9/eval_queue_n116-041-153.pid new file mode 100644 index 0000000000000000000000000000000000000000..076668fc2072ab552f37be56849a8ac543b9bf77 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_1958_seed5_9/eval_queue_n116-041-153.pid @@ -0,0 +1 @@ +1166466 diff --git a/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_1958_seed5_9/eval_queue_n116-041-188.log b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_1958_seed5_9/eval_queue_n116-041-188.log new file mode 100644 index 0000000000000000000000000000000000000000..b5a04c31bfac815c62ba2521700ee12016be1539 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_1958_seed5_9/eval_queue_n116-041-188.log @@ -0,0 +1,10270 @@ +eval_queue_started_utc=2026-07-26T02:59:37Z +hostname=n116-041-188 +train_tag=20260725_1958_seed5_9 +eval_tag=20260725_1958_seed5_9 +real_level=11 +fake_level=22 +eval_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9 +batch_per_gpu=256 +seeds=5 6 7 8 9 +[2026-07-26T02:59:38Z] eval_start tag=fixed500k_real_n11_seed5_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n11_classseed5_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T02:59:38Z +hostname=n116-041-188 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n11_classseed5_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n11_seed5_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 02:59:51.072000 1163824 torch/distributed/run.py:766] +W0726 02:59:51.072000 1163824 torch/distributed/run.py:766] ***************************************** +W0726 02:59:51.072000 1163824 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 02:59:51.072000 1163824 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 03:00:20.152900488 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:00:20.539380156 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W726 03:00:20.539380250 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:00:20.641895457 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:00:20.818407171 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:00:21.926587731 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:00:21.000044233 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:00:21.022302390 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n11_seed5_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n11_classseed5_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:25:02 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.3018 (0.3018) acc1: 95.3125 (95.3125) acc5: 100.0000 (100.0000) time: 6.4188 data: 4.5023 max mem: 3531 +Test: [ 10/234] eta: 0:02:32 acc1_pos: 86.979 acc5_pos: 100.000 loss: 0.4481 (0.4976) acc1: 88.5417 (79.1667) acc5: 100.0000 (100.0000) time: 0.6808 data: 0.4225 max mem: 3531 +Test: [ 20/234] eta: 0:01:33 acc1_pos: 85.677 acc5_pos: 100.000 loss: 0.4520 (0.5488) acc1: 76.0417 (73.7103) acc5: 100.0000 (100.0000) time: 0.1369 data: 0.0425 max mem: 3531 +Test: [ 30/234] eta: 0:01:19 acc1_pos: 68.229 acc5_pos: 100.000 loss: 0.3764 (0.4933) acc1: 85.6771 (77.3522) acc5: 100.0000 (100.0000) time: 0.2283 data: 0.1317 max mem: 3531 +Test: [ 40/234] eta: 0:01:06 acc1_pos: 93.490 acc5_pos: 100.000 loss: 0.3579 (0.4964) acc1: 86.9792 (76.0417) acc5: 100.0000 (100.0000) time: 0.2499 data: 0.1550 max mem: 3531 +Test: [ 50/234] eta: 0:00:58 acc1_pos: 93.750 acc5_pos: 100.000 loss: 0.2040 (0.4361) acc1: 93.7500 (79.8458) acc5: 100.0000 (100.0000) time: 0.2028 data: 0.1106 max mem: 3531 +Test: [ 60/234] eta: 0:00:52 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.1262 (0.3778) acc1: 96.8750 (83.1199) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2111 data: 0.1193 max mem: 3531 +Test: [ 70/234] eta: 0:00:48 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.0769 (0.3891) acc1: 100.0000 (82.6328) acc5: 100.0000 (100.0000) acc1_neg: 68.490 acc5_neg: 100.000 time: 0.2470 data: 0.1556 max mem: 3531 +Test: [ 80/234] eta: 0:00:43 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.5964 (0.3978) acc1: 71.6146 (82.4203) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2207 data: 0.1283 max mem: 3531 +Test: [ 90/234] eta: 0:00:40 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.1239 (0.3688) acc1: 98.1771 (84.0631) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.2199 data: 0.1270 max mem: 3531 +Test: [100/234] eta: 0:00:35 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.1464 (0.3662) acc1: 94.0104 (84.1894) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2039 data: 0.1105 max mem: 3531 +Test: [110/234] eta: 0:00:32 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.2743 (0.3610) acc1: 90.6250 (84.5955) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1649 data: 0.0682 max mem: 3531 +Test: [120/234] eta: 0:00:28 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.3225 (0.3650) acc1: 88.0208 (84.4546) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.1943 data: 0.0981 max mem: 3531 +Test: [130/234] eta: 0:00:26 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.4398 (0.3744) acc1: 80.7292 (83.8144) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.2006 data: 0.1077 max mem: 3531 +Test: [140/234] eta: 0:00:22 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.4785 (0.3956) acc1: 77.8646 (82.8310) acc5: 100.0000 (100.0000) acc1_neg: 36.198 acc5_neg: 100.000 time: 0.1810 data: 0.0896 max mem: 3531 +Test: [150/234] eta: 0:00:20 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.4159 (0.4005) acc1: 78.9062 (82.6055) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1716 data: 0.0743 max mem: 3531 +Test: [160/234] eta: 0:00:17 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.3947 (0.4030) acc1: 82.2917 (82.3936) acc5: 100.0000 (100.0000) acc1_neg: 68.229 acc5_neg: 100.000 time: 0.1839 data: 0.0865 max mem: 3531 +Test: [170/234] eta: 0:00:14 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.5639 (0.4139) acc1: 70.0521 (81.5302) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.1878 data: 0.0967 max mem: 3531 +Test: [180/234] eta: 0:00:12 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.5639 (0.4264) acc1: 69.0104 (80.6831) acc5: 100.0000 (100.0000) acc1_neg: 39.844 acc5_neg: 100.000 time: 0.1744 data: 0.0801 max mem: 3531 +Test: [190/234] eta: 0:00:09 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.4975 (0.4315) acc1: 76.3021 (80.4169) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.1681 data: 0.0740 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.4782 (0.4225) acc1: 78.1250 (80.9598) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1541 data: 0.0631 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.1092 (0.4075) acc1: 99.7396 (81.8523) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1381 data: 0.0478 max mem: 3532 +Test: [220/234] eta: 0:00:03 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.1000 (0.3938) acc1: 99.7396 (82.6322) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1520 data: 0.0622 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.1806 (0.3925) acc1: 88.8021 (82.7967) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1346 data: 0.0449 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.3516 (0.3921) acc1: 87.5000 (82.8305) acc5: 100.0000 (100.0000) acc1_neg: 85.992 acc5_neg: 100.000 time: 0.1303 data: 0.0394 max mem: 3532 +Test: Total time: 0:00:49 (0.2109 s / it) +* Overall Acc@1 82.831 Acc@5 100.000 + target == real Acc@1 79.958 Acc@5 100.000 + target == fake Acc@1 83.640 Acc@5 100.000 + loss 0.392 + PR-AUC 0.9703 ROC-AUC 0.8999 EER 17.82% (thr=0.5283) +Accuracy of the network on the 89729 test images: 82.8% +[rank0]:[W726 03:01:20.023170707 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:01:39.819000 1164370 torch/distributed/run.py:766] +W0726 03:01:39.819000 1164370 torch/distributed/run.py:766] ***************************************** +W0726 03:01:39.819000 1164370 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:01:39.819000 1164370 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:02:09.350323946 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 03:02:09.588525273 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W726 03:02:09.589897588 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:02:10.953918083 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:02:10.101701865 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:02:10.253265044 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:02:10.282631849 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W726 03:02:10.282846688 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n11_seed5_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n11_classseed5_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:00:34 acc1_pos: 76.823 acc5_pos: 100.000 loss: 0.5224 (0.5224) acc1: 76.8229 (76.8229) acc5: 100.0000 (100.0000) time: 6.0473 data: 3.5959 max mem: 3531 +Test: [ 10/2387] eta: 0:25:01 acc1_pos: 41.406 acc5_pos: 100.000 loss: 0.5691 (0.6026) acc1: 72.6562 (69.2945) acc5: 100.0000 (100.0000) time: 0.6315 data: 0.3272 max mem: 3531 +Test: [ 20/2387] eta: 0:14:48 acc1_pos: 65.104 acc5_pos: 100.000 loss: 0.6146 (0.6390) acc1: 71.0938 (66.4187) acc5: 100.0000 (100.0000) time: 0.0919 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:12:08 acc1_pos: 38.542 acc5_pos: 100.000 loss: 0.6256 (0.6410) acc1: 61.4583 (65.1546) acc5: 100.0000 (100.0000) time: 0.1319 data: 0.0388 max mem: 3531 +Test: [ 40/2387] eta: 0:11:38 acc1_pos: 74.479 acc5_pos: 100.000 loss: 0.6649 (0.6528) acc1: 58.5938 (63.9799) acc5: 100.0000 (100.0000) time: 0.2155 data: 0.1218 max mem: 3531 +Test: [ 50/2387] eta: 0:10:54 acc1_pos: 60.677 acc5_pos: 100.000 loss: 0.6773 (0.6530) acc1: 60.9375 (63.7255) acc5: 100.0000 (100.0000) time: 0.2349 data: 0.1421 max mem: 3531 +Test: [ 60/2387] eta: 0:10:44 acc1_pos: 71.094 acc5_pos: 100.000 loss: 0.6505 (0.6589) acc1: 60.9375 (63.0849) acc5: 100.0000 (100.0000) time: 0.2345 data: 0.1441 max mem: 3531 +Test: [ 70/2387] eta: 0:10:29 acc1_pos: 51.042 acc5_pos: 100.000 loss: 0.7028 (0.6682) acc1: 59.1146 (62.6541) acc5: 100.0000 (100.0000) time: 0.2502 data: 0.1599 max mem: 3531 +Test: [ 80/2387] eta: 0:10:17 acc1_pos: 99.479 acc5_pos: 100.000 loss: 0.5856 (0.6317) acc1: 64.8438 (66.0944) acc5: 100.0000 (100.0000) time: 0.2398 data: 0.1487 max mem: 3531 +Test: [ 90/2387] eta: 0:10:12 acc1_pos: 98.177 acc5_pos: 100.000 loss: 0.2968 (0.5947) acc1: 96.0938 (69.4397) acc5: 100.0000 (100.0000) time: 0.2496 data: 0.1576 max mem: 3531 +Test: [ 100/2387] eta: 0:10:15 acc1_pos: 95.833 acc5_pos: 100.000 loss: 0.2958 (0.5695) acc1: 95.3125 (71.5733) acc5: 100.0000 (100.0000) time: 0.2749 data: 0.1798 max mem: 3531 +Test: [ 110/2387] eta: 0:10:06 acc1_pos: 85.677 acc5_pos: 100.000 loss: 0.3066 (0.5469) acc1: 92.7083 (73.5572) acc5: 100.0000 (100.0000) time: 0.2649 data: 0.1674 max mem: 3531 +Test: [ 120/2387] eta: 0:10:10 acc1_pos: 98.958 acc5_pos: 100.000 loss: 0.3062 (0.5295) acc1: 94.0104 (75.1442) acc5: 100.0000 (100.0000) time: 0.2699 data: 0.1726 max mem: 3531 +Test: [ 130/2387] eta: 0:10:00 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.2871 (0.5107) acc1: 96.3542 (76.7911) acc5: 100.0000 (100.0000) time: 0.2649 data: 0.1676 max mem: 3531 +Test: [ 140/2387] eta: 0:09:46 acc1_pos: 98.177 acc5_pos: 100.000 loss: 0.2783 (0.4969) acc1: 97.6562 (78.0013) acc5: 100.0000 (100.0000) time: 0.2109 data: 0.1138 max mem: 3531 +Test: [ 150/2387] eta: 0:09:42 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2935 (0.4830) acc1: 94.7917 (79.1925) acc5: 100.0000 (100.0000) time: 0.2222 data: 0.1262 max mem: 3531 +Test: [ 160/2387] eta: 0:09:28 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2876 (0.4700) acc1: 95.0521 (80.1905) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2159 data: 0.1224 max mem: 3531 +Test: [ 170/2387] eta: 0:09:17 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2347 (0.4557) acc1: 95.0521 (81.0475) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1830 data: 0.0905 max mem: 3531 +Test: [ 180/2387] eta: 0:09:06 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2260 (0.4433) acc1: 94.5312 (81.8111) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1833 data: 0.0888 max mem: 3531 +Test: [ 190/2387] eta: 0:08:49 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2255 (0.4318) acc1: 95.0521 (82.5139) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1545 data: 0.0579 max mem: 3531 +Test: [ 200/2387] eta: 0:08:36 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2227 (0.4214) acc1: 95.0521 (83.1558) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1344 data: 0.0360 max mem: 3531 +Test: [ 210/2387] eta: 0:08:26 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2254 (0.4123) acc1: 95.3125 (83.7270) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1499 data: 0.0540 max mem: 3532 +Test: [ 220/2387] eta: 0:08:20 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2271 (0.4039) acc1: 95.0521 (84.2277) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1810 data: 0.0885 max mem: 3532 +Test: [ 230/2387] eta: 0:08:16 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2236 (0.3962) acc1: 95.3125 (84.7211) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2049 data: 0.1114 max mem: 3532 +Test: [ 240/2387] eta: 0:08:09 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2356 (0.3997) acc1: 93.2292 (84.5047) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.1945 data: 0.1013 max mem: 3532 +Test: [ 250/2387] eta: 0:08:02 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2224 (0.3894) acc1: 80.4688 (85.0183) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1758 data: 0.0836 max mem: 3532 +Test: [ 260/2387] eta: 0:07:56 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.0971 (0.3780) acc1: 100.0000 (85.5873) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1744 data: 0.0824 max mem: 3532 +Test: [ 270/2387] eta: 0:07:51 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.0812 (0.3672) acc1: 100.0000 (86.1057) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1815 data: 0.0894 max mem: 3532 +Test: [ 280/2387] eta: 0:07:44 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.0812 (0.3572) acc1: 100.0000 (86.5983) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1756 data: 0.0827 max mem: 3532 +Test: [ 290/2387] eta: 0:07:37 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.0833 (0.3482) acc1: 100.0000 (87.0436) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1599 data: 0.0677 max mem: 3532 +Test: [ 300/2387] eta: 0:07:32 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1036 (0.3406) acc1: 100.0000 (87.4689) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1642 data: 0.0717 max mem: 3532 +Test: [ 310/2387] eta: 0:07:27 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1276 (0.3345) acc1: 100.0000 (87.8433) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1724 data: 0.0799 max mem: 3532 +Test: [ 320/2387] eta: 0:07:21 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1328 (0.3282) acc1: 100.0000 (88.2204) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1651 data: 0.0741 max mem: 3532 +Test: [ 330/2387] eta: 0:07:16 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1322 (0.3222) acc1: 100.0000 (88.5763) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1649 data: 0.0706 max mem: 3532 +Test: [ 340/2387] eta: 0:07:12 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1307 (0.3166) acc1: 100.0000 (88.9113) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1746 data: 0.0799 max mem: 3532 +Test: [ 350/2387] eta: 0:07:08 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1317 (0.3118) acc1: 100.0000 (89.2272) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1805 data: 0.0875 max mem: 3532 +Test: [ 360/2387] eta: 0:07:04 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1554 (0.3085) acc1: 100.0000 (89.4427) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1807 data: 0.0868 max mem: 3532 +Test: [ 370/2387] eta: 0:07:00 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1984 (0.3055) acc1: 96.8750 (89.6332) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1743 data: 0.0807 max mem: 3532 +Test: [ 380/2387] eta: 0:06:55 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1957 (0.3027) acc1: 96.3542 (89.8109) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1642 data: 0.0680 max mem: 3532 +Test: [ 390/2387] eta: 0:06:51 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2017 (0.3002) acc1: 96.6146 (89.9889) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1567 data: 0.0619 max mem: 3532 +Test: [ 400/2387] eta: 0:06:47 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2017 (0.2977) acc1: 96.6146 (90.1496) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1650 data: 0.0720 max mem: 3532 +Test: [ 410/2387] eta: 0:06:44 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2009 (0.2952) acc1: 96.6146 (90.3164) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1836 data: 0.0906 max mem: 3532 +Test: [ 420/2387] eta: 0:06:41 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1995 (0.2930) acc1: 97.1354 (90.4710) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1798 data: 0.0859 max mem: 3532 +Test: [ 430/2387] eta: 0:06:37 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2043 (0.2943) acc1: 95.8333 (90.4075) acc5: 100.0000 (100.0000) acc1_neg: 80.729 acc5_neg: 100.000 time: 0.1699 data: 0.0729 max mem: 3532 +Test: [ 440/2387] eta: 0:06:33 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2132 (0.2929) acc1: 86.4583 (90.4732) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1649 data: 0.0684 max mem: 3533 +Test: [ 450/2387] eta: 0:06:30 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1021 (0.2886) acc1: 100.0000 (90.6845) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1714 data: 0.0758 max mem: 3533 +Test: [ 460/2387] eta: 0:06:27 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1012 (0.2846) acc1: 100.0000 (90.8860) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0816 max mem: 3533 +Test: [ 470/2387] eta: 0:06:24 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1012 (0.2807) acc1: 100.0000 (91.0778) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1753 data: 0.0847 max mem: 3533 +Test: [ 480/2387] eta: 0:06:21 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1014 (0.2770) acc1: 100.0000 (91.2633) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0918 max mem: 3533 +Test: [ 490/2387] eta: 0:06:19 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1026 (0.2734) acc1: 100.0000 (91.4407) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1876 data: 0.0893 max mem: 3533 +Test: [ 500/2387] eta: 0:06:17 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1025 (0.2700) acc1: 100.0000 (91.6105) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1901 data: 0.0934 max mem: 3533 +Test: [ 510/2387] eta: 0:06:14 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1029 (0.2678) acc1: 100.0000 (91.7660) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1900 data: 0.0962 max mem: 3533 +Test: [ 520/2387] eta: 0:06:12 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1954 (0.2678) acc1: 99.2188 (91.8976) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1859 data: 0.0917 max mem: 3533 +Test: [ 530/2387] eta: 0:06:10 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1964 (0.2664) acc1: 98.9583 (92.0227) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1930 data: 0.1000 max mem: 3533 +Test: [ 540/2387] eta: 0:06:06 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1947 (0.2651) acc1: 98.6979 (92.1437) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1788 data: 0.0874 max mem: 3533 +Test: [ 550/2387] eta: 0:06:05 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1927 (0.2638) acc1: 98.6979 (92.2626) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1826 data: 0.0900 max mem: 3533 +Test: [ 560/2387] eta: 0:06:03 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1912 (0.2625) acc1: 98.6979 (92.3778) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2097 data: 0.1173 max mem: 3533 +Test: [ 570/2387] eta: 0:06:01 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1914 (0.2613) acc1: 98.6979 (92.4871) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1996 data: 0.1088 max mem: 3533 +Test: [ 580/2387] eta: 0:05:59 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1914 (0.2601) acc1: 98.4375 (92.5887) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1899 data: 0.0977 max mem: 3533 +Test: [ 590/2387] eta: 0:05:57 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1963 (0.2591) acc1: 98.4375 (92.6889) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1950 data: 0.1019 max mem: 3533 +Test: [ 600/2387] eta: 0:05:55 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2046 (0.2597) acc1: 98.6979 (92.7855) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2013 data: 0.1088 max mem: 3533 +Test: [ 610/2387] eta: 0:05:52 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2223 (0.2589) acc1: 98.1771 (92.8699) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1947 data: 0.1034 max mem: 3533 +Test: [ 620/2387] eta: 0:05:50 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2125 (0.2582) acc1: 97.9167 (92.9469) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1835 data: 0.0911 max mem: 3533 +Test: [ 630/2387] eta: 0:05:48 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2159 (0.2575) acc1: 97.6562 (93.0216) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1844 data: 0.0897 max mem: 3533 +Test: [ 640/2387] eta: 0:05:45 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2148 (0.2568) acc1: 97.3958 (93.0882) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1863 data: 0.0923 max mem: 3533 +Test: [ 650/2387] eta: 0:05:43 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2146 (0.2562) acc1: 97.6562 (93.1612) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1922 data: 0.0951 max mem: 3533 +Test: [ 660/2387] eta: 0:05:41 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2160 (0.2556) acc1: 97.6562 (93.2272) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1912 data: 0.0952 max mem: 3533 +Test: [ 670/2387] eta: 0:05:38 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2143 (0.2549) acc1: 97.6562 (93.2955) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1799 data: 0.0884 max mem: 3534 +Test: [ 680/2387] eta: 0:05:36 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2172 (0.2570) acc1: 97.3958 (93.1898) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.1842 data: 0.0915 max mem: 3534 +Test: [ 690/2387] eta: 0:05:34 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2903 (0.2580) acc1: 84.3750 (93.1101) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1888 data: 0.0940 max mem: 3534 +Test: [ 700/2387] eta: 0:05:32 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2754 (0.2582) acc1: 89.5833 (93.0702) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1835 data: 0.0882 max mem: 3534 +Test: [ 710/2387] eta: 0:05:29 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2694 (0.2583) acc1: 89.8438 (93.0285) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1786 data: 0.0861 max mem: 3534 +Test: [ 720/2387] eta: 0:05:27 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2632 (0.2584) acc1: 90.3646 (92.9904) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.1770 data: 0.0852 max mem: 3534 +Test: [ 730/2387] eta: 0:05:25 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2700 (0.2587) acc1: 90.3646 (92.9495) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.1849 data: 0.0906 max mem: 3534 +Test: [ 740/2387] eta: 0:05:22 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2696 (0.2588) acc1: 90.1042 (92.9108) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.1839 data: 0.0901 max mem: 3534 +Test: [ 750/2387] eta: 0:05:20 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2636 (0.2588) acc1: 90.3646 (92.8755) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.1775 data: 0.0853 max mem: 3534 +Test: [ 760/2387] eta: 0:05:18 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2736 (0.2598) acc1: 90.1042 (92.7980) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.1798 data: 0.0880 max mem: 3534 +Test: [ 770/2387] eta: 0:05:15 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.3647 (0.2613) acc1: 85.1562 (92.7023) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1776 data: 0.0864 max mem: 3534 +Test: [ 780/2387] eta: 0:05:13 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.3450 (0.2622) acc1: 85.1562 (92.6410) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1748 data: 0.0833 max mem: 3534 +Test: [ 790/2387] eta: 0:05:11 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2472 (0.2619) acc1: 92.7083 (92.6622) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1846 data: 0.0936 max mem: 3534 +Test: [ 800/2387] eta: 0:05:09 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2411 (0.2624) acc1: 94.0104 (92.6290) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1845 data: 0.0931 max mem: 3534 +Test: [ 810/2387] eta: 0:05:07 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.3018 (0.2633) acc1: 87.2396 (92.5571) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.1833 data: 0.0919 max mem: 3534 +Test: [ 820/2387] eta: 0:05:05 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.3419 (0.2650) acc1: 86.4583 (92.4172) acc5: 100.0000 (100.0000) acc1_neg: 71.615 acc5_neg: 100.000 time: 0.1957 data: 0.1039 max mem: 3534 +Test: [ 830/2387] eta: 0:05:03 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.5266 (0.2697) acc1: 71.3542 (92.0712) acc5: 100.0000 (100.0000) acc1_neg: 64.323 acc5_neg: 100.000 time: 0.1915 data: 0.0981 max mem: 3534 +Test: [ 840/2387] eta: 0:05:01 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2401 (0.2686) acc1: 76.0417 (92.1380) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1886 data: 0.0939 max mem: 3534 +Test: [ 850/2387] eta: 0:04:58 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1322 (0.2669) acc1: 99.2188 (92.2218) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1863 data: 0.0921 max mem: 3534 +Test: [ 860/2387] eta: 0:04:56 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1192 (0.2651) acc1: 100.0000 (92.3100) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1795 data: 0.0877 max mem: 3534 +Test: [ 870/2387] eta: 0:04:54 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1192 (0.2635) acc1: 99.7396 (92.3887) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1805 data: 0.0893 max mem: 3534 +Test: [ 880/2387] eta: 0:04:52 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1269 (0.2621) acc1: 99.7396 (92.4680) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1896 data: 0.0950 max mem: 3534 +Test: [ 890/2387] eta: 0:04:50 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1626 (0.2621) acc1: 98.9583 (92.5043) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1928 data: 0.0987 max mem: 3534 +Test: [ 900/2387] eta: 0:04:48 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.3167 (0.2637) acc1: 91.4062 (92.4236) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1949 data: 0.1028 max mem: 3535 +Test: [ 910/2387] eta: 0:04:46 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.4204 (0.2654) acc1: 83.3333 (92.3267) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.1963 data: 0.1044 max mem: 3535 +Test: [ 920/2387] eta: 0:04:44 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.4254 (0.2671) acc1: 83.5938 (92.2293) acc5: 100.0000 (100.0000) acc1_neg: 80.729 acc5_neg: 100.000 time: 0.1899 data: 0.0988 max mem: 3535 +Test: [ 930/2387] eta: 0:04:42 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.4243 (0.2688) acc1: 83.3333 (92.1324) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.1940 data: 0.1022 max mem: 3535 +Test: [ 940/2387] eta: 0:04:40 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.4252 (0.2705) acc1: 83.3333 (92.0334) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.1968 data: 0.1046 max mem: 3535 +Test: [ 950/2387] eta: 0:04:38 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.4203 (0.2721) acc1: 83.5938 (91.9498) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1912 data: 0.0994 max mem: 3535 +Test: [ 960/2387] eta: 0:04:36 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.4159 (0.2736) acc1: 83.8542 (91.8593) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1863 data: 0.0948 max mem: 3535 +Test: [ 970/2387] eta: 0:04:34 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.4169 (0.2752) acc1: 83.3333 (91.7793) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1852 data: 0.0939 max mem: 3535 +Test: [ 980/2387] eta: 0:04:32 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.4165 (0.2761) acc1: 84.8958 (91.7200) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1792 data: 0.0872 max mem: 3535 +Test: [ 990/2387] eta: 0:04:30 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.3347 (0.2767) acc1: 86.7188 (91.6593) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1731 data: 0.0804 max mem: 3535 +Test: [1000/2387] eta: 0:04:28 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.3337 (0.2775) acc1: 86.9792 (91.6006) acc5: 100.0000 (100.0000) acc1_neg: 80.729 acc5_neg: 100.000 time: 0.1793 data: 0.0879 max mem: 3535 +Test: [1010/2387] eta: 0:04:26 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.3819 (0.2796) acc1: 82.2917 (91.4446) acc5: 100.0000 (100.0000) acc1_neg: 67.708 acc5_neg: 100.000 time: 0.1774 data: 0.0856 max mem: 3535 +Test: [1020/2387] eta: 0:04:23 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.4805 (0.2817) acc1: 74.4792 (91.2945) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.1717 data: 0.0791 max mem: 3535 +Test: [1030/2387] eta: 0:04:21 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.3111 (0.2820) acc1: 86.9792 (91.2691) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1837 data: 0.0906 max mem: 3535 +Test: [1040/2387] eta: 0:04:19 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2257 (0.2814) acc1: 93.4896 (91.3049) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1887 data: 0.0960 max mem: 3535 +Test: [1050/2387] eta: 0:04:17 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2150 (0.2813) acc1: 94.2708 (91.3094) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1803 data: 0.0892 max mem: 3535 +Test: [1060/2387] eta: 0:04:15 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2766 (0.2814) acc1: 92.4479 (91.3022) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1794 data: 0.0881 max mem: 3535 +Test: [1070/2387] eta: 0:04:13 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.3138 (0.2824) acc1: 88.8021 (91.2193) acc5: 100.0000 (100.0000) acc1_neg: 61.719 acc5_neg: 100.000 time: 0.1778 data: 0.0836 max mem: 3535 +Test: [1080/2387] eta: 0:04:11 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.4981 (0.2854) acc1: 73.9583 (91.0078) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.1779 data: 0.0833 max mem: 3535 +Test: [1090/2387] eta: 0:04:09 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.3797 (0.2855) acc1: 81.7708 (90.9993) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.1827 data: 0.0912 max mem: 3535 +Test: [1100/2387] eta: 0:04:07 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2599 (0.2850) acc1: 91.9271 (91.0259) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1775 data: 0.0855 max mem: 3535 +Test: [1110/2387] eta: 0:04:05 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2246 (0.2849) acc1: 92.7083 (91.0504) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1820 data: 0.0884 max mem: 3535 +Test: [1120/2387] eta: 0:04:03 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2485 (0.2847) acc1: 93.2292 (91.0594) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1821 data: 0.0888 max mem: 3536 +Test: [1130/2387] eta: 0:04:01 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2665 (0.2847) acc1: 91.1458 (91.0666) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1766 data: 0.0842 max mem: 3536 +Test: [1140/2387] eta: 0:03:59 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.3492 (0.2870) acc1: 83.3333 (90.9064) acc5: 100.0000 (100.0000) acc1_neg: 66.667 acc5_neg: 100.000 time: 0.1816 data: 0.0893 max mem: 3536 +Test: [1150/2387] eta: 0:03:57 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.5174 (0.2871) acc1: 74.7396 (90.8673) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1782 data: 0.0862 max mem: 3536 +Test: [1160/2387] eta: 0:03:55 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.0847 (0.2853) acc1: 100.0000 (90.9460) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1781 data: 0.0852 max mem: 3536 +Test: [1170/2387] eta: 0:03:53 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.0773 (0.2836) acc1: 100.0000 (91.0229) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1831 data: 0.0901 max mem: 3536 +Test: [1180/2387] eta: 0:03:51 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.0712 (0.2817) acc1: 100.0000 (91.0986) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2014 data: 0.1095 max mem: 3536 +Test: [1190/2387] eta: 0:03:48 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.0729 (0.2801) acc1: 100.0000 (91.1729) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1802 data: 0.0887 max mem: 3536 +Test: [1200/2387] eta: 0:03:47 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.0939 (0.2788) acc1: 100.0000 (91.2428) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1684 data: 0.0763 max mem: 3536 +Test: [1210/2387] eta: 0:03:44 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1219 (0.2778) acc1: 100.0000 (91.3073) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1790 data: 0.0871 max mem: 3536 +Test: [1220/2387] eta: 0:03:42 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.0936 (0.2762) acc1: 100.0000 (91.3766) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1762 data: 0.0856 max mem: 3536 +Test: [1230/2387] eta: 0:03:41 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.0775 (0.2746) acc1: 100.0000 (91.4445) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1893 data: 0.0976 max mem: 3536 +Test: [1240/2387] eta: 0:03:39 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.0688 (0.2729) acc1: 100.0000 (91.5126) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1843 data: 0.0913 max mem: 3536 +Test: [1250/2387] eta: 0:03:36 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.0731 (0.2714) acc1: 100.0000 (91.5792) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1756 data: 0.0837 max mem: 3536 +Test: [1260/2387] eta: 0:03:34 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.0805 (0.2699) acc1: 100.0000 (91.6450) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1746 data: 0.0833 max mem: 3536 +Test: [1270/2387] eta: 0:03:32 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.0990 (0.2690) acc1: 99.7396 (91.7031) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1760 data: 0.0843 max mem: 3536 +Test: [1280/2387] eta: 0:03:30 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1601 (0.2684) acc1: 98.6979 (91.7433) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1747 data: 0.0835 max mem: 3536 +Test: [1290/2387] eta: 0:03:28 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1797 (0.2677) acc1: 97.6562 (91.7907) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1738 data: 0.0826 max mem: 3536 +Test: [1300/2387] eta: 0:03:26 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1828 (0.2671) acc1: 97.3958 (91.8290) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1780 data: 0.0849 max mem: 3536 +Test: [1310/2387] eta: 0:03:24 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1795 (0.2665) acc1: 96.6146 (91.8641) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1759 data: 0.0825 max mem: 3536 +Test: [1320/2387] eta: 0:03:22 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1824 (0.2660) acc1: 97.6562 (91.9074) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1775 data: 0.0855 max mem: 3536 +Test: [1330/2387] eta: 0:03:20 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2209 (0.2671) acc1: 93.4896 (91.8369) acc5: 100.0000 (100.0000) acc1_neg: 72.135 acc5_neg: 100.000 time: 0.1850 data: 0.0918 max mem: 3536 +Test: [1340/2387] eta: 0:03:18 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.4674 (0.2692) acc1: 75.7812 (91.6964) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1836 data: 0.0900 max mem: 3536 +Test: [1350/2387] eta: 0:03:16 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1008 (0.2678) acc1: 98.6979 (91.7567) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1858 data: 0.0938 max mem: 3537 +Test: [1360/2387] eta: 0:03:15 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.0770 (0.2663) acc1: 100.0000 (91.8173) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1895 data: 0.0980 max mem: 3537 +Test: [1370/2387] eta: 0:03:13 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.0630 (0.2649) acc1: 100.0000 (91.8767) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1890 data: 0.0961 max mem: 3537 +Test: [1380/2387] eta: 0:03:11 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.0630 (0.2634) acc1: 100.0000 (91.9356) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1896 data: 0.0941 max mem: 3537 +Test: [1390/2387] eta: 0:03:09 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.0643 (0.2620) acc1: 100.0000 (91.9935) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1916 data: 0.0972 max mem: 3537 +Test: [1400/2387] eta: 0:03:07 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.0838 (0.2611) acc1: 100.0000 (92.0449) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1909 data: 0.0986 max mem: 3537 +Test: [1410/2387] eta: 0:03:05 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1517 (0.2615) acc1: 98.4375 (92.0223) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.1894 data: 0.0980 max mem: 3537 +Test: [1420/2387] eta: 0:03:03 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2717 (0.2617) acc1: 91.1458 (92.0158) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1906 data: 0.0998 max mem: 3537 +Test: [1430/2387] eta: 0:03:01 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.3060 (0.2623) acc1: 89.5833 (91.9780) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1857 data: 0.0945 max mem: 3537 +Test: [1440/2387] eta: 0:02:59 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.3715 (0.2633) acc1: 84.3750 (91.9092) acc5: 100.0000 (100.0000) acc1_neg: 57.812 acc5_neg: 100.000 time: 0.1839 data: 0.0929 max mem: 3537 +Test: [1450/2387] eta: 0:02:57 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.4752 (0.2652) acc1: 77.0833 (91.7753) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1779 data: 0.0861 max mem: 3537 +Test: [1460/2387] eta: 0:02:55 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2211 (0.2648) acc1: 91.6667 (91.8005) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1723 data: 0.0800 max mem: 3537 +Test: [1470/2387] eta: 0:02:53 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1622 (0.2640) acc1: 97.3958 (91.8425) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1811 data: 0.0888 max mem: 3537 +Test: [1480/2387] eta: 0:02:51 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1422 (0.2633) acc1: 98.1771 (91.8812) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1769 data: 0.0842 max mem: 3537 +Test: [1490/2387] eta: 0:02:49 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1422 (0.2626) acc1: 98.1771 (91.9173) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1702 data: 0.0786 max mem: 3537 +Test: [1500/2387] eta: 0:02:47 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1557 (0.2620) acc1: 98.1771 (91.9564) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1724 data: 0.0811 max mem: 3537 +Test: [1510/2387] eta: 0:02:45 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1746 (0.2617) acc1: 98.6979 (92.0045) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1793 data: 0.0868 max mem: 3537 +Test: [1520/2387] eta: 0:02:43 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1771 (0.2609) acc1: 99.2188 (92.0505) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1810 data: 0.0888 max mem: 3537 +Test: [1530/2387] eta: 0:02:41 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1106 (0.2598) acc1: 99.7396 (92.1013) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1690 data: 0.0745 max mem: 3537 +Test: [1540/2387] eta: 0:02:39 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.0963 (0.2588) acc1: 100.0000 (92.1510) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1641 data: 0.0700 max mem: 3537 +Test: [1550/2387] eta: 0:02:37 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.0931 (0.2578) acc1: 100.0000 (92.2001) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1681 data: 0.0775 max mem: 3537 +Test: [1560/2387] eta: 0:02:35 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1015 (0.2568) acc1: 99.7396 (92.2484) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1729 data: 0.0786 max mem: 3537 +Test: [1570/2387] eta: 0:02:33 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1174 (0.2565) acc1: 99.4792 (92.2888) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1717 data: 0.0748 max mem: 3537 +Test: [1580/2387] eta: 0:02:31 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2526 (0.2568) acc1: 97.1354 (92.2712) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.1668 data: 0.0710 max mem: 3538 +Test: [1590/2387] eta: 0:02:30 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2823 (0.2569) acc1: 89.0625 (92.2630) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1668 data: 0.0710 max mem: 3538 +Test: [1600/2387] eta: 0:02:28 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2347 (0.2570) acc1: 94.0104 (92.2656) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1725 data: 0.0789 max mem: 3538 +Test: [1610/2387] eta: 0:02:26 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2426 (0.2570) acc1: 93.2292 (92.2635) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1757 data: 0.0843 max mem: 3538 +Test: [1620/2387] eta: 0:02:24 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2813 (0.2573) acc1: 89.8438 (92.2362) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1757 data: 0.0842 max mem: 3538 +Test: [1630/2387] eta: 0:02:22 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.3373 (0.2584) acc1: 83.8542 (92.1562) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.1772 data: 0.0841 max mem: 3538 +Test: [1640/2387] eta: 0:02:20 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.5323 (0.2605) acc1: 72.9167 (92.0099) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.1761 data: 0.0835 max mem: 3538 +Test: [1650/2387] eta: 0:02:18 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.5798 (0.2624) acc1: 69.0104 (91.8787) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.1737 data: 0.0815 max mem: 3538 +Test: [1660/2387] eta: 0:02:16 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.5476 (0.2642) acc1: 69.5312 (91.7468) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.1726 data: 0.0803 max mem: 3538 +Test: [1670/2387] eta: 0:02:14 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.5532 (0.2661) acc1: 70.0521 (91.6190) acc5: 100.0000 (100.0000) acc1_neg: 70.573 acc5_neg: 100.000 time: 0.1712 data: 0.0797 max mem: 3538 +Test: [1680/2387] eta: 0:02:12 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.5814 (0.2680) acc1: 69.7917 (91.4819) acc5: 100.0000 (100.0000) acc1_neg: 59.375 acc5_neg: 100.000 time: 0.1740 data: 0.0819 max mem: 3538 +Test: [1690/2387] eta: 0:02:10 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.5814 (0.2698) acc1: 69.7917 (91.3568) acc5: 100.0000 (100.0000) acc1_neg: 76.042 acc5_neg: 100.000 time: 0.1773 data: 0.0852 max mem: 3538 +Test: [1700/2387] eta: 0:02:08 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.5851 (0.2709) acc1: 69.2708 (91.2783) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1797 data: 0.0872 max mem: 3538 +Test: [1710/2387] eta: 0:02:06 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1070 (0.2699) acc1: 99.4792 (91.3270) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1817 data: 0.0898 max mem: 3538 +Test: [1720/2387] eta: 0:02:04 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.0826 (0.2688) acc1: 99.7396 (91.3754) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1805 data: 0.0893 max mem: 3538 +Test: [1730/2387] eta: 0:02:02 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.0737 (0.2677) acc1: 100.0000 (91.4231) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1733 data: 0.0816 max mem: 3538 +Test: [1740/2387] eta: 0:02:01 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.0790 (0.2666) acc1: 100.0000 (91.4712) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1696 data: 0.0782 max mem: 3538 +Test: [1750/2387] eta: 0:01:59 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.0894 (0.2657) acc1: 99.4792 (91.5160) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1770 data: 0.0855 max mem: 3538 +Test: [1760/2387] eta: 0:01:57 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1191 (0.2653) acc1: 98.6979 (91.5383) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1827 data: 0.0908 max mem: 3538 +Test: [1770/2387] eta: 0:01:55 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1548 (0.2646) acc1: 97.1354 (91.5712) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1747 data: 0.0815 max mem: 3538 +Test: [1780/2387] eta: 0:01:53 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1188 (0.2638) acc1: 99.2188 (91.6096) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1785 data: 0.0859 max mem: 3538 +Test: [1790/2387] eta: 0:01:51 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1047 (0.2630) acc1: 99.7396 (91.6526) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1892 data: 0.0986 max mem: 3538 +Test: [1800/2387] eta: 0:01:49 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.0948 (0.2620) acc1: 99.7396 (91.6978) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1799 data: 0.0889 max mem: 3538 +Test: [1810/2387] eta: 0:01:47 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.0948 (0.2612) acc1: 99.7396 (91.7364) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1807 data: 0.0892 max mem: 3539 +Test: [1820/2387] eta: 0:01:45 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1201 (0.2607) acc1: 97.9167 (91.7631) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1846 data: 0.0915 max mem: 3539 +Test: [1830/2387] eta: 0:01:44 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1757 (0.2605) acc1: 97.1354 (91.7826) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1859 data: 0.0926 max mem: 3539 +Test: [1840/2387] eta: 0:01:42 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1478 (0.2598) acc1: 99.2188 (91.8242) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1840 data: 0.0922 max mem: 3539 +Test: [1850/2387] eta: 0:01:40 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1449 (0.2592) acc1: 99.4792 (91.8656) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1830 data: 0.0907 max mem: 3539 +Test: [1860/2387] eta: 0:01:38 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1447 (0.2586) acc1: 99.4792 (91.9051) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1947 data: 0.1009 max mem: 3539 +Test: [1870/2387] eta: 0:01:36 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1436 (0.2580) acc1: 99.4792 (91.9463) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2008 data: 0.1064 max mem: 3539 +Test: [1880/2387] eta: 0:01:34 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1419 (0.2574) acc1: 99.4792 (91.9868) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1955 data: 0.1027 max mem: 3539 +Test: [1890/2387] eta: 0:01:32 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1446 (0.2568) acc1: 99.4792 (92.0253) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1951 data: 0.1013 max mem: 3539 +Test: [1900/2387] eta: 0:01:31 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1479 (0.2563) acc1: 99.2188 (92.0605) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1957 data: 0.1006 max mem: 3539 +Test: [1910/2387] eta: 0:01:29 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2302 (0.2564) acc1: 95.5729 (92.0737) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1896 data: 0.0957 max mem: 3539 +Test: [1920/2387] eta: 0:01:27 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2795 (0.2569) acc1: 94.0104 (92.0439) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1892 data: 0.0967 max mem: 3539 +Test: [1930/2387] eta: 0:01:25 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2344 (0.2566) acc1: 94.7917 (92.0673) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1906 data: 0.0999 max mem: 3539 +Test: [1940/2387] eta: 0:01:23 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2068 (0.2567) acc1: 96.6146 (92.0796) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1970 data: 0.1049 max mem: 3539 +Test: [1950/2387] eta: 0:01:21 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2362 (0.2567) acc1: 94.7917 (92.0859) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.2034 data: 0.1112 max mem: 3539 +Test: [1960/2387] eta: 0:01:19 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2627 (0.2569) acc1: 92.7083 (92.0817) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1986 data: 0.1071 max mem: 3539 +Test: [1970/2387] eta: 0:01:18 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.3916 (0.2586) acc1: 82.0312 (91.9548) acc5: 100.0000 (100.0000) acc1_neg: 61.719 acc5_neg: 100.000 time: 0.1810 data: 0.0888 max mem: 3539 +Test: [1980/2387] eta: 0:01:16 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.5746 (0.2592) acc1: 67.4479 (91.8999) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1756 data: 0.0838 max mem: 3539 +Test: [1990/2387] eta: 0:01:14 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.0915 (0.2584) acc1: 99.4792 (91.9389) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1936 data: 0.1012 max mem: 3539 +Test: [2000/2387] eta: 0:01:12 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.0909 (0.2575) acc1: 99.7396 (91.9778) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2034 data: 0.1115 max mem: 3539 +Test: [2010/2387] eta: 0:01:10 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.0895 (0.2567) acc1: 99.7396 (92.0166) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2023 data: 0.1108 max mem: 3539 +Test: [2020/2387] eta: 0:01:08 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.0897 (0.2559) acc1: 99.7396 (92.0549) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2037 data: 0.1114 max mem: 3539 +Test: [2030/2387] eta: 0:01:07 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.0909 (0.2551) acc1: 99.7396 (92.0931) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2112 data: 0.1177 max mem: 3540 +Test: [2040/2387] eta: 0:01:05 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.0899 (0.2543) acc1: 100.0000 (92.1312) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2017 data: 0.1088 max mem: 3540 +Test: [2050/2387] eta: 0:01:03 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.0890 (0.2535) acc1: 100.0000 (92.1685) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1888 data: 0.0967 max mem: 3540 +Test: [2060/2387] eta: 0:01:01 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1151 (0.2529) acc1: 99.7396 (92.2034) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1691 data: 0.0767 max mem: 3540 +Test: [2070/2387] eta: 0:00:59 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1177 (0.2522) acc1: 99.4792 (92.2391) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1732 data: 0.0803 max mem: 3540 +Test: [2080/2387] eta: 0:00:57 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1161 (0.2516) acc1: 99.7396 (92.2745) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2027 data: 0.1073 max mem: 3540 +Test: [2090/2387] eta: 0:00:55 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1146 (0.2509) acc1: 99.7396 (92.3102) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2103 data: 0.1162 max mem: 3540 +Test: [2100/2387] eta: 0:00:53 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1156 (0.2503) acc1: 99.7396 (92.3445) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2053 data: 0.1140 max mem: 3540 +Test: [2110/2387] eta: 0:00:52 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1180 (0.2497) acc1: 99.4792 (92.3782) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2048 data: 0.1129 max mem: 3540 +Test: [2120/2387] eta: 0:00:50 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1269 (0.2504) acc1: 98.6979 (92.3340) acc5: 100.0000 (100.0000) acc1_neg: 69.792 acc5_neg: 100.000 time: 0.2158 data: 0.1221 max mem: 3540 +Test: [2130/2387] eta: 0:00:48 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.3174 (0.2506) acc1: 84.6354 (92.3147) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2041 data: 0.1105 max mem: 3540 +Test: [2140/2387] eta: 0:00:46 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2968 (0.2511) acc1: 91.4062 (92.2938) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2094 data: 0.1173 max mem: 3540 +Test: [2150/2387] eta: 0:00:44 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.3190 (0.2514) acc1: 86.9792 (92.2716) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.2133 data: 0.1216 max mem: 3540 +Test: [2160/2387] eta: 0:00:42 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.3591 (0.2520) acc1: 85.1562 (92.2364) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.1874 data: 0.0956 max mem: 3540 +Test: [2170/2387] eta: 0:00:40 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.4538 (0.2537) acc1: 77.8646 (92.1160) acc5: 100.0000 (100.0000) acc1_neg: 63.021 acc5_neg: 100.000 time: 0.1853 data: 0.0941 max mem: 3540 +Test: [2180/2387] eta: 0:00:39 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.6681 (0.2559) acc1: 61.4583 (91.9473) acc5: 100.0000 (100.0000) acc1_neg: 52.083 acc5_neg: 100.000 time: 0.2030 data: 0.1118 max mem: 3540 +Test: [2190/2387] eta: 0:00:37 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.6500 (0.2558) acc1: 61.7188 (91.9474) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2300 data: 0.1376 max mem: 3540 +Test: [2200/2387] eta: 0:00:35 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1784 (0.2556) acc1: 97.9167 (91.9704) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1964 data: 0.1030 max mem: 3540 +Test: [2210/2387] eta: 0:00:33 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1730 (0.2551) acc1: 98.9583 (92.0035) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1867 data: 0.0940 max mem: 3540 +Test: [2220/2387] eta: 0:00:31 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1557 (0.2546) acc1: 99.2188 (92.0345) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2046 data: 0.1138 max mem: 3540 +Test: [2230/2387] eta: 0:00:29 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1557 (0.2542) acc1: 98.9583 (92.0619) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1735 data: 0.0825 max mem: 3540 +Test: [2240/2387] eta: 0:00:27 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1722 (0.2539) acc1: 98.6979 (92.0891) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1744 data: 0.0832 max mem: 3540 +Test: [2250/2387] eta: 0:00:25 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.1878 (0.2538) acc1: 98.1771 (92.1146) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1910 data: 0.1006 max mem: 3540 +Test: [2260/2387] eta: 0:00:23 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2478 (0.2539) acc1: 96.8750 (92.1103) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.1955 data: 0.1041 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2650 (0.2540) acc1: 93.2292 (92.1162) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1938 data: 0.1022 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2381 (0.2540) acc1: 95.0521 (92.1225) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.1930 data: 0.1023 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.2274 (0.2539) acc1: 94.5312 (92.1276) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.1861 data: 0.0944 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.3007 (0.2542) acc1: 89.5833 (92.1139) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1862 data: 0.0925 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.3482 (0.2552) acc1: 85.9375 (92.0492) acc5: 100.0000 (100.0000) acc1_neg: 64.583 acc5_neg: 100.000 time: 0.1800 data: 0.0876 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.4903 (0.2564) acc1: 75.5208 (91.9620) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.1768 data: 0.0863 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.4859 (0.2573) acc1: 77.6042 (91.9071) acc5: 100.0000 (100.0000) acc1_neg: 79.427 acc5_neg: 100.000 time: 0.1615 data: 0.0712 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.3408 (0.2575) acc1: 84.3750 (91.8943) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1562 data: 0.0663 max mem: 3541 +Test: [2350/2387] eta: 0:00:06 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.3184 (0.2580) acc1: 89.3229 (91.8668) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1469 data: 0.0571 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.3844 (0.2588) acc1: 84.3750 (91.8266) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.1281 data: 0.0383 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.4269 (0.2596) acc1: 81.2500 (91.7729) acc5: 100.0000 (100.0000) acc1_neg: 63.802 acc5_neg: 100.000 time: 0.1359 data: 0.0461 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.5590 (0.2613) acc1: 65.8854 (91.6496) acc5: 100.0000 (100.0000) acc1_neg: 63.281 acc5_neg: 100.000 time: 0.1272 data: 0.0372 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 94.521 acc5_pos: 100.000 loss: 0.6118 (0.2616) acc1: 65.3646 (91.6177) acc5: 100.0000 (100.0000) acc1_neg: 97.674 acc5_neg: 100.000 time: 0.1229 data: 0.0319 max mem: 3541 +Test: Total time: 0:07:26 (0.1872 s / it) +* Overall Acc@1 91.618 Acc@5 100.000 + target == real Acc@1 79.831 Acc@5 100.000 + target == fake Acc@1 92.445 Acc@5 100.000 + loss 0.262 + PR-AUC 0.9959 ROC-AUC 0.9477 EER 12.02% (thr=0.6016) +Accuracy of the network on the 916267 test images: 91.6% +[rank0]:[W726 03:10:00.297032550 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:10:19.490000 1165003 torch/distributed/run.py:766] +W0726 03:10:19.490000 1165003 torch/distributed/run.py:766] ***************************************** +W0726 03:10:19.490000 1165003 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:10:19.490000 1165003 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:10:49.369755450 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:10:49.685586791 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:10:50.057410300 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +[rank0]:[W726 03:10:50.359396976 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:10:50.372974373 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:10:50.404273986 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 1): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:10:50.621518215 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank1]:[W726 03:10:50.622693988 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n11_seed5_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n11_classseed5_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:03:12 acc1_pos: 76.823 acc5_pos: 100.000 loss: 0.5224 (0.5224) acc1: 76.8229 (76.8229) acc5: 100.0000 (100.0000) time: 7.3353 data: 5.6237 max mem: 3531 +Test: [ 10/517] eta: 0:06:20 acc1_pos: 41.406 acc5_pos: 100.000 loss: 0.5691 (0.6026) acc1: 72.6562 (69.2945) acc5: 100.0000 (100.0000) time: 0.7504 data: 0.5116 max mem: 3531 +Test: [ 20/517] eta: 0:03:38 acc1_pos: 65.104 acc5_pos: 100.000 loss: 0.6146 (0.6390) acc1: 71.0938 (66.4187) acc5: 100.0000 (100.0000) time: 0.0959 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:03:06 acc1_pos: 38.542 acc5_pos: 100.000 loss: 0.6256 (0.6410) acc1: 61.4583 (65.1546) acc5: 100.0000 (100.0000) time: 0.1804 data: 0.0851 max mem: 3531 +Test: [ 40/517] eta: 0:02:49 acc1_pos: 74.479 acc5_pos: 100.000 loss: 0.6649 (0.6528) acc1: 58.5938 (63.9799) acc5: 100.0000 (100.0000) time: 0.2649 data: 0.1713 max mem: 3531 +Test: [ 50/517] eta: 0:02:37 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.6649 (0.6523) acc1: 63.2812 (63.4242) acc5: 100.0000 (100.0000) acc1_neg: 42.708 acc5_neg: 100.000 time: 0.2650 data: 0.1719 max mem: 3531 +Test: [ 60/517] eta: 0:02:25 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.7009 (0.6662) acc1: 52.6042 (61.4583) acc5: 100.0000 (100.0000) acc1_neg: 38.281 acc5_neg: 100.000 time: 0.2467 data: 0.1550 max mem: 3531 +Test: [ 70/517] eta: 0:02:18 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.7107 (0.6712) acc1: 52.6042 (60.5267) acc5: 100.0000 (100.0000) acc1_neg: 60.417 acc5_neg: 100.000 time: 0.2446 data: 0.1522 max mem: 3531 +Test: [ 80/517] eta: 0:02:13 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.6880 (0.6839) acc1: 52.3438 (58.8831) acc5: 100.0000 (100.0000) acc1_neg: 22.396 acc5_neg: 100.000 time: 0.2589 data: 0.1662 max mem: 3531 +Test: [ 90/517] eta: 0:02:08 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.6835 (0.6824) acc1: 52.3438 (58.8227) acc5: 100.0000 (100.0000) acc1_neg: 58.073 acc5_neg: 100.000 time: 0.2615 data: 0.1691 max mem: 3531 +Test: [100/517] eta: 0:02:03 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.6351 (0.6795) acc1: 58.0729 (58.7459) acc5: 100.0000 (100.0000) acc1_neg: 51.042 acc5_neg: 100.000 time: 0.2641 data: 0.1729 max mem: 3531 +Test: [110/517] eta: 0:02:01 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.6020 (0.6745) acc1: 63.5417 (59.0184) acc5: 100.0000 (100.0000) acc1_neg: 64.844 acc5_neg: 100.000 time: 0.2884 data: 0.1944 max mem: 3531 +Test: [120/517] eta: 0:01:56 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.6020 (0.6712) acc1: 64.8438 (59.3190) acc5: 100.0000 (100.0000) acc1_neg: 64.323 acc5_neg: 100.000 time: 0.2849 data: 0.1899 max mem: 3531 +Test: [130/517] eta: 0:01:52 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.6007 (0.6662) acc1: 65.6250 (59.9435) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.2503 data: 0.1575 max mem: 3531 +Test: [140/517] eta: 0:01:48 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.5702 (0.6544) acc1: 69.5312 (61.5544) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.2451 data: 0.1539 max mem: 3531 +Test: [150/517] eta: 0:01:45 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.4864 (0.6417) acc1: 85.9375 (63.1071) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2608 data: 0.1688 max mem: 3531 +Test: [160/517] eta: 0:01:41 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.4383 (0.6288) acc1: 84.1146 (64.2679) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.2654 data: 0.1725 max mem: 3531 +Test: [170/517] eta: 0:01:38 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.4123 (0.6162) acc1: 84.1146 (65.3737) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.2537 data: 0.1603 max mem: 3531 +Test: [180/517] eta: 0:01:34 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.4009 (0.6075) acc1: 85.6771 (66.2696) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.2455 data: 0.1526 max mem: 3531 +Test: [190/517] eta: 0:01:31 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.3577 (0.5935) acc1: 87.5000 (67.5693) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.2444 data: 0.1524 max mem: 3531 +Test: [200/517] eta: 0:01:28 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.3804 (0.5872) acc1: 87.5000 (68.4235) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.2537 data: 0.1559 max mem: 3531 +Test: [210/517] eta: 0:01:25 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.4492 (0.5808) acc1: 84.8958 (69.3424) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.2612 data: 0.1630 max mem: 3532 +Test: [220/517] eta: 0:01:22 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.4451 (0.5756) acc1: 89.0625 (70.0745) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.2661 data: 0.1742 max mem: 3532 +Test: [230/517] eta: 0:01:19 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.5133 (0.5758) acc1: 79.1667 (70.1231) acc5: 100.0000 (100.0000) acc1_neg: 48.438 acc5_neg: 100.000 time: 0.2702 data: 0.1780 max mem: 3532 +Test: [240/517] eta: 0:01:16 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.6276 (0.5843) acc1: 63.5417 (69.2741) acc5: 100.0000 (100.0000) acc1_neg: 14.583 acc5_neg: 100.000 time: 0.2537 data: 0.1590 max mem: 3532 +Test: [250/517] eta: 0:01:13 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.6613 (0.5883) acc1: 60.6771 (68.9554) acc5: 100.0000 (100.0000) acc1_neg: 66.667 acc5_neg: 100.000 time: 0.2454 data: 0.1495 max mem: 3532 +Test: [260/517] eta: 0:01:10 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.6008 (0.5906) acc1: 66.6667 (68.6841) acc5: 100.0000 (100.0000) acc1_neg: 33.333 acc5_neg: 100.000 time: 0.2569 data: 0.1623 max mem: 3532 +Test: [270/517] eta: 0:01:07 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.7401 (0.5980) acc1: 45.5729 (67.6055) acc5: 100.0000 (100.0000) acc1_neg: 43.490 acc5_neg: 100.000 time: 0.2594 data: 0.1669 max mem: 3532 +Test: [280/517] eta: 0:01:04 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.7401 (0.6003) acc1: 45.5729 (67.3515) acc5: 100.0000 (100.0000) acc1_neg: 75.521 acc5_neg: 100.000 time: 0.2579 data: 0.1642 max mem: 3532 +Test: [290/517] eta: 0:01:01 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.5234 (0.5971) acc1: 75.7812 (67.7432) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.2497 data: 0.1540 max mem: 3532 +Test: [300/517] eta: 0:00:58 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.5219 (0.5959) acc1: 75.7812 (67.7931) acc5: 100.0000 (100.0000) acc1_neg: 46.354 acc5_neg: 100.000 time: 0.2559 data: 0.1606 max mem: 3532 +Test: [310/517] eta: 0:00:56 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.6261 (0.5977) acc1: 61.4583 (67.4194) acc5: 100.0000 (100.0000) acc1_neg: 53.646 acc5_neg: 100.000 time: 0.2649 data: 0.1706 max mem: 3532 +Test: [320/517] eta: 0:00:53 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.6303 (0.5986) acc1: 55.2083 (67.2524) acc5: 100.0000 (100.0000) acc1_neg: 48.438 acc5_neg: 100.000 time: 0.2639 data: 0.1688 max mem: 3532 +Test: [330/517] eta: 0:00:50 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.6065 (0.5975) acc1: 64.5833 (67.3496) acc5: 100.0000 (100.0000) acc1_neg: 64.583 acc5_neg: 100.000 time: 0.2636 data: 0.1669 max mem: 3532 +Test: [340/517] eta: 0:00:47 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.5350 (0.5969) acc1: 73.6979 (67.4120) acc5: 100.0000 (100.0000) acc1_neg: 41.406 acc5_neg: 100.000 time: 0.2558 data: 0.1586 max mem: 3532 +Test: [350/517] eta: 0:00:44 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.6575 (0.5993) acc1: 59.8958 (67.1103) acc5: 100.0000 (100.0000) acc1_neg: 34.115 acc5_neg: 100.000 time: 0.2514 data: 0.1557 max mem: 3532 +Test: [360/517] eta: 0:00:42 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.6595 (0.6005) acc1: 59.8958 (66.9632) acc5: 100.0000 (100.0000) acc1_neg: 50.521 acc5_neg: 100.000 time: 0.2589 data: 0.1621 max mem: 3532 +Test: [370/517] eta: 0:00:39 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.5072 (0.5970) acc1: 76.5625 (67.3960) acc5: 100.0000 (100.0000) acc1_neg: 76.042 acc5_neg: 100.000 time: 0.2541 data: 0.1564 max mem: 3532 +Test: [380/517] eta: 0:00:36 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.4922 (0.5950) acc1: 77.8646 (67.5921) acc5: 100.0000 (100.0000) acc1_neg: 68.490 acc5_neg: 100.000 time: 0.2607 data: 0.1648 max mem: 3532 +Test: [390/517] eta: 0:00:34 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.5030 (0.5931) acc1: 77.3438 (67.8329) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.2691 data: 0.1726 max mem: 3532 +Test: [400/517] eta: 0:00:31 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.4763 (0.5889) acc1: 79.6875 (68.2681) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.2541 data: 0.1567 max mem: 3532 +Test: [410/517] eta: 0:00:28 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.4452 (0.5878) acc1: 78.3854 (68.3299) acc5: 100.0000 (100.0000) acc1_neg: 58.333 acc5_neg: 100.000 time: 0.2549 data: 0.1574 max mem: 3532 +Test: [420/517] eta: 0:00:25 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.6372 (0.5904) acc1: 61.4583 (68.0634) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.2549 data: 0.1573 max mem: 3532 +Test: [430/517] eta: 0:00:23 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.6658 (0.5909) acc1: 60.4167 (68.0328) acc5: 100.0000 (100.0000) acc1_neg: 64.062 acc5_neg: 100.000 time: 0.2499 data: 0.1502 max mem: 3532 +Test: [440/517] eta: 0:00:20 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.5779 (0.5910) acc1: 67.9688 (68.0514) acc5: 100.0000 (100.0000) acc1_neg: 63.281 acc5_neg: 100.000 time: 0.2461 data: 0.1467 max mem: 3533 +Test: [450/517] eta: 0:00:17 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.5286 (0.5886) acc1: 73.1771 (68.3383) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.2503 data: 0.1550 max mem: 3533 +Test: [460/517] eta: 0:00:15 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.4785 (0.5862) acc1: 80.2083 (68.6076) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.2587 data: 0.1630 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.5187 (0.5856) acc1: 73.4375 (68.6228) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.2595 data: 0.1610 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.5397 (0.5846) acc1: 70.3125 (68.7235) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.2377 data: 0.1396 max mem: 3533 +Test: [490/517] eta: 0:00:07 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.5015 (0.5820) acc1: 77.8646 (69.0088) acc5: 100.0000 (100.0000) acc1_neg: 71.094 acc5_neg: 100.000 time: 0.1991 data: 0.1058 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.5269 (0.5854) acc1: 66.9271 (68.5681) acc5: 100.0000 (100.0000) acc1_neg: 44.792 acc5_neg: 100.000 time: 0.1791 data: 0.0892 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.7308 (0.5882) acc1: 46.8750 (68.2241) acc5: 100.0000 (100.0000) acc1_neg: 64.583 acc5_neg: 100.000 time: 0.1710 data: 0.0813 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 95.413 acc5_pos: 100.000 loss: 0.7174 (0.5897) acc1: 49.7396 (68.0759) acc5: 100.0000 (100.0000) acc1_neg: 48.763 acc5_neg: 100.000 time: 0.1727 data: 0.0812 max mem: 3533 +Test: Total time: 0:02:13 (0.2582 s / it) +* Overall Acc@1 68.076 Acc@5 100.000 + target == real Acc@1 63.909 Acc@5 100.000 + target == fake Acc@1 68.508 Acc@5 100.000 + loss 0.590 + PR-AUC 0.9515 ROC-AUC 0.7202 EER 34.00% (thr=0.5137) +Accuracy of the network on the 198427 test images: 68.1% +[rank0]:[W726 03:13:14.353918522 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 03:13:34.269000 1165563 torch/distributed/run.py:766] +W0726 03:13:34.269000 1165563 torch/distributed/run.py:766] ***************************************** +W0726 03:13:34.269000 1165563 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:13:34.269000 1165563 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:14:02.198450430 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:14:02.598073493 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:14:02.788072933 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:14:02.884031856 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 03:14:03.167558329 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W726 03:14:03.168061052 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:14:03.204780607 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W726 03:14:03.204838154 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n11_seed5_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n11_classseed5_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:41:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.4711 (2.4711) acc1: 0.0000 (0.0000) acc5: 100.0000 (100.0000) time: 6.3335 data: 4.9515 max mem: 3531 +Test: [ 10/389] eta: 0:04:11 acc1_pos: 63.802 acc5_pos: 100.000 loss: 2.0193 (1.9429) acc1: 0.5208 (6.9129) acc5: 100.0000 (100.0000) time: 0.6641 data: 0.4506 max mem: 3531 +Test: [ 20/389] eta: 0:02:25 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.8705 (1.8470) acc1: 2.0833 (16.2946) acc5: 100.0000 (100.0000) time: 0.0982 data: 0.0005 max mem: 3531 +Test: [ 30/389] eta: 0:02:03 acc1_pos: 0.260 acc5_pos: 100.000 loss: 1.6121 (1.8650) acc1: 0.7812 (11.4247) acc5: 100.0000 (100.0000) time: 0.1665 data: 0.0696 max mem: 3531 +Test: [ 40/389] eta: 0:01:53 acc1_pos: 0.781 acc5_pos: 100.000 loss: 1.7719 (1.8124) acc1: 0.7812 (13.7322) acc5: 100.0000 (100.0000) time: 0.2504 data: 0.1578 max mem: 3531 +Test: [ 50/389] eta: 0:01:38 acc1_pos: 23.177 acc5_pos: 100.000 loss: 1.2467 (1.7191) acc1: 12.5000 (14.5323) acc5: 100.0000 (100.0000) time: 0.2080 data: 0.1148 max mem: 3531 +Test: [ 60/389] eta: 0:01:25 acc1_pos: 26.562 acc5_pos: 100.000 loss: 1.2102 (1.7511) acc1: 20.8333 (14.3101) acc5: 100.0000 (100.0000) time: 0.1294 data: 0.0325 max mem: 3531 +Test: [ 70/389] eta: 0:01:19 acc1_pos: 8.854 acc5_pos: 100.000 loss: 1.8726 (1.7887) acc1: 2.0833 (13.7397) acc5: 100.0000 (100.0000) time: 0.1472 data: 0.0526 max mem: 3531 +Test: [ 80/389] eta: 0:01:14 acc1_pos: 60.677 acc5_pos: 100.000 loss: 2.0195 (1.7680) acc1: 0.2604 (14.9434) acc5: 100.0000 (100.0000) time: 0.1799 data: 0.0867 max mem: 3531 +Test: [ 90/389] eta: 0:01:09 acc1_pos: 24.740 acc5_pos: 100.000 loss: 1.1978 (1.6881) acc1: 37.2396 (18.5611) acc5: 100.0000 (100.0000) time: 0.1779 data: 0.0836 max mem: 3531 +Test: [100/389] eta: 0:01:05 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.5185 (1.7265) acc1: 22.6562 (17.5768) acc5: 100.0000 (100.0000) time: 0.1705 data: 0.0787 max mem: 3531 +Test: [110/389] eta: 0:01:01 acc1_pos: 98.438 acc5_pos: 100.000 loss: 2.0342 (1.7279) acc1: 0.2604 (19.0057) acc5: 100.0000 (100.0000) time: 0.1599 data: 0.0674 max mem: 3531 +Test: [120/389] eta: 0:00:58 acc1_pos: 45.052 acc5_pos: 100.000 loss: 1.6098 (1.7053) acc1: 32.0312 (20.6741) acc5: 100.0000 (100.0000) time: 0.1600 data: 0.0670 max mem: 3531 +Test: [130/389] eta: 0:00:55 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.7551 (1.6186) acc1: 62.2396 (24.9284) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1765 data: 0.0850 max mem: 3531 +Test: [140/389] eta: 0:00:53 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1029 (1.5220) acc1: 99.2188 (29.2646) acc5: 100.0000 (100.0000) acc1_neg: 37.760 acc5_neg: 100.000 time: 0.2007 data: 0.1066 max mem: 3531 +Test: [150/389] eta: 0:00:50 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1370 (1.4408) acc1: 99.7396 (33.1471) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1944 data: 0.1004 max mem: 3531 +Test: [160/389] eta: 0:00:48 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1517 (1.3594) acc1: 99.7396 (37.2218) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1884 data: 0.0966 max mem: 3531 +Test: [170/389] eta: 0:00:45 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1739 (1.2942) acc1: 98.1771 (40.7118) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1968 data: 0.1051 max mem: 3531 +Test: [180/389] eta: 0:00:43 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2238 (1.2323) acc1: 96.3542 (43.8450) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1720 data: 0.0812 max mem: 3531 +Test: [190/389] eta: 0:00:40 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1114 (1.1763) acc1: 100.0000 (46.5478) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1671 data: 0.0760 max mem: 3531 +Test: [200/389] eta: 0:00:38 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1008 (1.1233) acc1: 100.0000 (49.1812) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1980 data: 0.1069 max mem: 3531 +Test: [210/389] eta: 0:00:36 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1044 (1.0755) acc1: 100.0000 (51.5822) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1994 data: 0.1066 max mem: 3532 +Test: [220/389] eta: 0:00:33 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0629 (1.0297) acc1: 100.0000 (53.7731) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1574 data: 0.0606 max mem: 3532 +Test: [230/389] eta: 0:00:31 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0630 (0.9930) acc1: 100.0000 (55.7089) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1469 data: 0.0509 max mem: 3532 +Test: [240/389] eta: 0:00:29 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1798 (0.9619) acc1: 99.2188 (57.1955) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1696 data: 0.0758 max mem: 3532 +Test: [250/389] eta: 0:00:27 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0924 (0.9265) acc1: 100.0000 (58.9009) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1679 data: 0.0737 max mem: 3532 +Test: [260/389] eta: 0:00:25 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0666 (0.8967) acc1: 100.0000 (60.4616) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1662 data: 0.0734 max mem: 3532 +Test: [270/389] eta: 0:00:23 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1270 (0.8719) acc1: 100.0000 (61.6928) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1733 data: 0.0818 max mem: 3532 +Test: [280/389] eta: 0:00:21 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1266 (0.8454) acc1: 100.0000 (63.0561) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1734 data: 0.0812 max mem: 3532 +Test: [290/389] eta: 0:00:19 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1311 (0.8251) acc1: 100.0000 (64.1395) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1645 data: 0.0720 max mem: 3532 +Test: [300/389] eta: 0:00:17 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1384 (0.8021) acc1: 100.0000 (65.3300) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1631 data: 0.0714 max mem: 3532 +Test: [310/389] eta: 0:00:15 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1309 (0.7808) acc1: 100.0000 (66.4004) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1649 data: 0.0728 max mem: 3532 +Test: [320/389] eta: 0:00:13 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0817 (0.7590) acc1: 100.0000 (67.4471) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1628 data: 0.0705 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0803 (0.7386) acc1: 100.0000 (68.4306) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1641 data: 0.0727 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0816 (0.7195) acc1: 100.0000 (69.3518) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1620 data: 0.0705 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1131 (0.7198) acc1: 99.4792 (69.0000) acc5: 100.0000 (100.0000) acc1_neg: 1.042 acc5_neg: 100.000 time: 0.1600 data: 0.0684 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.8083 (0.7272) acc1: 50.5208 (68.1859) acc5: 100.0000 (100.0000) acc1_neg: 50.521 acc5_neg: 100.000 time: 0.1649 data: 0.0740 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.4967 (0.7128) acc1: 54.4271 (68.9402) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1559 data: 0.0656 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0952 (0.6972) acc1: 100.0000 (69.7418) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1405 data: 0.0501 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0952 (0.6858) acc1: 100.0000 (70.3290) acc5: 100.0000 (100.0000) acc1_neg: 87.432 acc5_neg: 100.000 time: 0.1379 data: 0.0442 max mem: 3532 +Test: Total time: 0:01:11 (0.1846 s / it) +* Overall Acc@1 70.329 Acc@5 100.000 + target == real Acc@1 20.553 Acc@5 100.000 + target == fake Acc@1 93.466 Acc@5 100.000 + loss 0.686 + PR-AUC 0.7688 ROC-AUC 0.6533 EER 37.78% (thr=0.8877) +Accuracy of the network on the 149358 test images: 70.3% +[rank0]:[W726 03:15:27.631795609 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:15:46.946000 1166123 torch/distributed/run.py:766] +W0726 03:15:46.946000 1166123 torch/distributed/run.py:766] ***************************************** +W0726 03:15:46.946000 1166123 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:15:46.946000 1166123 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:16:14.076479347 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:16:14.349901081 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:16:14.442426148 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:16:14.759277999 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:16:14.783750309 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:16:14.900623292 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +[rank0]:[W726 03:16:15.925077858 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:16:15.932485144 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n11_seed5_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n11_classseed5_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:34:28 acc1_pos: 92.188 acc5_pos: 100.000 loss: 0.2903 (0.2903) acc1: 92.1875 (92.1875) acc5: 100.0000 (100.0000) time: 6.2129 data: 4.7007 max mem: 3531 +Test: [ 10/333] eta: 0:03:28 acc1_pos: 32.552 acc5_pos: 100.000 loss: 0.2903 (0.5423) acc1: 91.1458 (70.2415) acc5: 100.0000 (100.0000) time: 0.6469 data: 0.4277 max mem: 3531 +Test: [ 20/333] eta: 0:02:00 acc1_pos: 40.365 acc5_pos: 100.000 loss: 1.0521 (0.8199) acc1: 32.5521 (50.8557) acc5: 100.0000 (100.0000) time: 0.0922 data: 0.0022 max mem: 3531 +Test: [ 30/333] eta: 0:01:38 acc1_pos: 68.750 acc5_pos: 100.000 loss: 0.9536 (0.7743) acc1: 37.7604 (54.4019) acc5: 100.0000 (100.0000) time: 0.1467 data: 0.0553 max mem: 3531 +Test: [ 40/333] eta: 0:01:26 acc1_pos: 66.406 acc5_pos: 100.000 loss: 0.6582 (0.7402) acc1: 65.3646 (57.3679) acc5: 100.0000 (100.0000) time: 0.1998 data: 0.1078 max mem: 3531 +Test: [ 50/333] eta: 0:01:19 acc1_pos: 72.917 acc5_pos: 100.000 loss: 0.6238 (0.7170) acc1: 66.1458 (59.2729) acc5: 100.0000 (100.0000) time: 0.2099 data: 0.1179 max mem: 3531 +Test: [ 60/333] eta: 0:01:13 acc1_pos: 70.052 acc5_pos: 100.000 loss: 0.6120 (0.7058) acc1: 66.4062 (60.2929) acc5: 100.0000 (100.0000) time: 0.2148 data: 0.1208 max mem: 3531 +Test: [ 70/333] eta: 0:01:07 acc1_pos: 50.260 acc5_pos: 100.000 loss: 0.6151 (0.6982) acc1: 64.0625 (60.8641) acc5: 100.0000 (100.0000) time: 0.1949 data: 0.0972 max mem: 3531 +Test: [ 80/333] eta: 0:01:03 acc1_pos: 74.740 acc5_pos: 100.000 loss: 0.6190 (0.6912) acc1: 65.6250 (61.4648) acc5: 100.0000 (100.0000) time: 0.2052 data: 0.1094 max mem: 3531 +Test: [ 90/333] eta: 0:00:59 acc1_pos: 73.438 acc5_pos: 100.000 loss: 0.5781 (0.6750) acc1: 70.8333 (62.6774) acc5: 100.0000 (100.0000) time: 0.2149 data: 0.1213 max mem: 3531 +Test: [100/333] eta: 0:00:56 acc1_pos: 77.865 acc5_pos: 100.000 loss: 0.5181 (0.6602) acc1: 73.9583 (63.9078) acc5: 100.0000 (100.0000) time: 0.1946 data: 0.0994 max mem: 3531 +Test: [110/333] eta: 0:00:51 acc1_pos: 80.729 acc5_pos: 100.000 loss: 0.5088 (0.6462) acc1: 75.2604 (65.0361) acc5: 100.0000 (100.0000) time: 0.1700 data: 0.0754 max mem: 3531 +Test: [120/333] eta: 0:00:47 acc1_pos: 76.823 acc5_pos: 100.000 loss: 0.4896 (0.6330) acc1: 79.4271 (66.1910) acc5: 100.0000 (100.0000) time: 0.1399 data: 0.0460 max mem: 3531 +Test: [130/333] eta: 0:00:44 acc1_pos: 87.500 acc5_pos: 100.000 loss: 0.4896 (0.6223) acc1: 79.4271 (67.0006) acc5: 100.0000 (100.0000) time: 0.1562 data: 0.0638 max mem: 3531 +Test: [140/333] eta: 0:00:41 acc1_pos: 87.760 acc5_pos: 100.000 loss: 0.4848 (0.6123) acc1: 78.9062 (67.8395) acc5: 100.0000 (100.0000) time: 0.1699 data: 0.0749 max mem: 3531 +Test: [150/333] eta: 0:00:39 acc1_pos: 77.865 acc5_pos: 100.000 loss: 0.5046 (0.6075) acc1: 72.9167 (68.1222) acc5: 100.0000 (100.0000) time: 0.1635 data: 0.0658 max mem: 3531 +Test: [160/333] eta: 0:00:36 acc1_pos: 56.579 acc5_pos: 100.000 loss: 0.5280 (0.6137) acc1: 70.3125 (67.6873) acc5: 100.0000 (100.0000) acc1_neg: 35.417 acc5_neg: 100.000 time: 0.1706 data: 0.0764 max mem: 3531 +Test: [170/333] eta: 0:00:34 acc1_pos: 56.579 acc5_pos: 100.000 loss: 0.8239 (0.6287) acc1: 47.1354 (66.3499) acc5: 100.0000 (100.0000) acc1_neg: 51.042 acc5_neg: 100.000 time: 0.1765 data: 0.0830 max mem: 3531 +Test: [180/333] eta: 0:00:31 acc1_pos: 56.579 acc5_pos: 100.000 loss: 0.7949 (0.6350) acc1: 48.1771 (65.6567) acc5: 100.0000 (100.0000) acc1_neg: 61.198 acc5_neg: 100.000 time: 0.1842 data: 0.0902 max mem: 3531 +Test: [190/333] eta: 0:00:29 acc1_pos: 56.579 acc5_pos: 100.000 loss: 0.7456 (0.6406) acc1: 53.1250 (64.9937) acc5: 100.0000 (100.0000) acc1_neg: 45.312 acc5_neg: 100.000 time: 0.1983 data: 0.1057 max mem: 3531 +Test: [200/333] eta: 0:00:27 acc1_pos: 56.579 acc5_pos: 100.000 loss: 0.7594 (0.6457) acc1: 52.6042 (64.4875) acc5: 100.0000 (100.0000) acc1_neg: 52.344 acc5_neg: 100.000 time: 0.2125 data: 0.1187 max mem: 3531 +Test: [210/333] eta: 0:00:25 acc1_pos: 56.579 acc5_pos: 100.000 loss: 0.7619 (0.6525) acc1: 52.3438 (63.9156) acc5: 100.0000 (100.0000) acc1_neg: 52.083 acc5_neg: 100.000 time: 0.2049 data: 0.1106 max mem: 3532 +Test: [220/333] eta: 0:00:23 acc1_pos: 56.579 acc5_pos: 100.000 loss: 0.8008 (0.6607) acc1: 51.5625 (63.1599) acc5: 100.0000 (100.0000) acc1_neg: 63.021 acc5_neg: 100.000 time: 0.1923 data: 0.0983 max mem: 3532 +Test: [230/333] eta: 0:00:21 acc1_pos: 56.579 acc5_pos: 100.000 loss: 0.5962 (0.6559) acc1: 65.3646 (63.5665) acc5: 100.0000 (100.0000) acc1_neg: 71.615 acc5_neg: 100.000 time: 0.2301 data: 0.1380 max mem: 3532 +Test: [240/333] eta: 0:00:19 acc1_pos: 56.579 acc5_pos: 100.000 loss: 0.5962 (0.6586) acc1: 66.4062 (63.1797) acc5: 100.0000 (100.0000) acc1_neg: 52.604 acc5_neg: 100.000 time: 0.2508 data: 0.1594 max mem: 3532 +Test: [250/333] eta: 0:00:17 acc1_pos: 56.579 acc5_pos: 100.000 loss: 0.7590 (0.6633) acc1: 51.8229 (62.6785) acc5: 100.0000 (100.0000) acc1_neg: 52.604 acc5_neg: 100.000 time: 0.1820 data: 0.0866 max mem: 3532 +Test: [260/333] eta: 0:00:15 acc1_pos: 56.579 acc5_pos: 100.000 loss: 0.7634 (0.6683) acc1: 52.6042 (62.2176) acc5: 100.0000 (100.0000) acc1_neg: 49.740 acc5_neg: 100.000 time: 0.1802 data: 0.0851 max mem: 3532 +Test: [270/333] eta: 0:00:13 acc1_pos: 56.579 acc5_pos: 100.000 loss: 0.7357 (0.6692) acc1: 54.4271 (62.1185) acc5: 100.0000 (100.0000) acc1_neg: 61.198 acc5_neg: 100.000 time: 0.2126 data: 0.1205 max mem: 3532 +Test: [280/333] eta: 0:00:10 acc1_pos: 56.579 acc5_pos: 100.000 loss: 0.7050 (0.6762) acc1: 53.9062 (61.5158) acc5: 100.0000 (100.0000) acc1_neg: 50.000 acc5_neg: 100.000 time: 0.1936 data: 0.1013 max mem: 3532 +Test: [290/333] eta: 0:00:08 acc1_pos: 56.579 acc5_pos: 100.000 loss: 0.9952 (0.6865) acc1: 36.1979 (60.6708) acc5: 100.0000 (100.0000) acc1_neg: 42.448 acc5_neg: 100.000 time: 0.1907 data: 0.0933 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 56.579 acc5_pos: 100.000 loss: 0.8738 (0.6928) acc1: 41.4062 (60.1277) acc5: 100.0000 (100.0000) acc1_neg: 47.917 acc5_neg: 100.000 time: 0.1911 data: 0.0941 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 56.579 acc5_pos: 100.000 loss: 0.8313 (0.6931) acc1: 47.9167 (60.0600) acc5: 100.0000 (100.0000) acc1_neg: 49.219 acc5_neg: 100.000 time: 0.1841 data: 0.0923 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 56.579 acc5_pos: 100.000 loss: 0.7453 (0.6958) acc1: 54.9479 (59.7969) acc5: 100.0000 (100.0000) acc1_neg: 40.365 acc5_neg: 100.000 time: 0.1677 data: 0.0758 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 56.579 acc5_pos: 100.000 loss: 0.9418 (0.7054) acc1: 40.3646 (59.1240) acc5: 100.0000 (100.0000) acc1_neg: 39.062 acc5_neg: 100.000 time: 0.1444 data: 0.0532 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 56.579 acc5_pos: 100.000 loss: 0.9740 (0.7069) acc1: 40.3646 (59.0318) acc5: 100.0000 (100.0000) acc1_neg: 41.972 acc5_neg: 100.000 time: 0.1409 data: 0.0468 max mem: 3532 +Test: Total time: 0:01:07 (0.2020 s / it) +* Overall Acc@1 59.032 Acc@5 100.000 + target == real Acc@1 68.529 Acc@5 100.000 + target == fake Acc@1 50.577 Acc@5 100.000 + loss 0.707 + PR-AUC 0.6400 ROC-AUC 0.6409 EER 39.56% (thr=0.4329) +Accuracy of the network on the 127843 test images: 59.0% +[rank0]:[W726 03:17:31.668067388 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:17:51.094000 1166681 torch/distributed/run.py:766] +W0726 03:17:51.094000 1166681 torch/distributed/run.py:766] ***************************************** +W0726 03:17:51.094000 1166681 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:17:51.094000 1166681 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 03:18:19.569081575 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:18:20.251258936 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:18:20.591950386 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:18:20.604856422 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:18:20.677820900 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:18:20.737794510 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:18:20.817738119 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:18:20.847094764 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n11_seed5_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n11_classseed5_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:24:43 acc1_pos: 94.531 acc5_pos: 100.000 loss: 0.2484 (0.2484) acc1: 94.5312 (94.5312) acc5: 100.0000 (100.0000) time: 6.3107 data: 4.9436 max mem: 3531 +Test: [ 10/235] eta: 0:02:27 acc1_pos: 92.969 acc5_pos: 100.000 loss: 0.2516 (0.2578) acc1: 93.2292 (93.2765) acc5: 100.0000 (100.0000) time: 0.6567 data: 0.4498 max mem: 3531 +Test: [ 20/235] eta: 0:01:24 acc1_pos: 91.927 acc5_pos: 100.000 loss: 0.2514 (0.2537) acc1: 93.2292 (93.4772) acc5: 100.0000 (100.0000) time: 0.0956 data: 0.0004 max mem: 3531 +Test: [ 30/235] eta: 0:01:07 acc1_pos: 94.531 acc5_pos: 100.000 loss: 0.2490 (0.2536) acc1: 93.2292 (93.2880) acc5: 100.0000 (100.0000) time: 0.1500 data: 0.0542 max mem: 3531 +Test: [ 40/235] eta: 0:00:58 acc1_pos: 96.094 acc5_pos: 100.000 loss: 0.2490 (0.2524) acc1: 93.2292 (93.3816) acc5: 100.0000 (100.0000) time: 0.2003 data: 0.1079 max mem: 3531 +Test: [ 50/235] eta: 0:00:52 acc1_pos: 92.969 acc5_pos: 100.000 loss: 0.2485 (0.2533) acc1: 93.4896 (93.3977) acc5: 100.0000 (100.0000) time: 0.2071 data: 0.1156 max mem: 3531 +Test: [ 60/235] eta: 0:00:47 acc1_pos: 94.010 acc5_pos: 100.000 loss: 0.2549 (0.2543) acc1: 93.4896 (93.3999) acc5: 100.0000 (100.0000) time: 0.2100 data: 0.1179 max mem: 3531 +Test: [ 70/235] eta: 0:00:43 acc1_pos: 92.308 acc5_pos: 100.000 loss: 0.2785 (0.3947) acc1: 91.9271 (84.4630) acc5: 100.0000 (100.0000) acc1_neg: 17.969 acc5_neg: 100.000 time: 0.2177 data: 0.1231 max mem: 3531 +Test: [ 80/235] eta: 0:00:38 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.3842 (0.5180) acc1: 21.0938 (76.6043) acc5: 100.0000 (100.0000) acc1_neg: 17.448 acc5_neg: 100.000 time: 0.1944 data: 0.0984 max mem: 3531 +Test: [ 90/235] eta: 0:00:35 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.3910 (0.6133) acc1: 20.8333 (70.5157) acc5: 100.0000 (100.0000) acc1_neg: 18.750 acc5_neg: 100.000 time: 0.1768 data: 0.0817 max mem: 3531 +Test: [100/235] eta: 0:00:32 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.3832 (0.6884) acc1: 20.8333 (65.6044) acc5: 100.0000 (100.0000) acc1_neg: 19.271 acc5_neg: 100.000 time: 0.2049 data: 0.1108 max mem: 3531 +Test: [110/235] eta: 0:00:29 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.3766 (0.7505) acc1: 21.0938 (61.6038) acc5: 100.0000 (100.0000) acc1_neg: 21.875 acc5_neg: 100.000 time: 0.1981 data: 0.1041 max mem: 3531 +Test: [120/235] eta: 0:00:26 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.4194 (0.8178) acc1: 18.4896 (57.5370) acc5: 100.0000 (100.0000) acc1_neg: 13.021 acc5_neg: 100.000 time: 0.1873 data: 0.0918 max mem: 3531 +Test: [130/235] eta: 0:00:24 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.5733 (0.8756) acc1: 11.9792 (54.0712) acc5: 100.0000 (100.0000) acc1_neg: 13.021 acc5_neg: 100.000 time: 0.1975 data: 0.1025 max mem: 3531 +Test: [140/235] eta: 0:00:21 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.6020 (0.9281) acc1: 11.1979 (51.0214) acc5: 100.0000 (100.0000) acc1_neg: 9.375 acc5_neg: 100.000 time: 0.1904 data: 0.0994 max mem: 3531 +Test: [150/235] eta: 0:00:19 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.5935 (0.9711) acc1: 10.9375 (48.3909) acc5: 100.0000 (100.0000) acc1_neg: 13.542 acc5_neg: 100.000 time: 0.1889 data: 0.0985 max mem: 3531 +Test: [160/235] eta: 0:00:16 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.5784 (1.0093) acc1: 11.1979 (46.1035) acc5: 100.0000 (100.0000) acc1_neg: 12.500 acc5_neg: 100.000 time: 0.2034 data: 0.1122 max mem: 3531 +Test: [170/235] eta: 0:00:14 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.5195 (1.0247) acc1: 12.5000 (44.9790) acc5: 100.0000 (100.0000) acc1_neg: 24.740 acc5_neg: 100.000 time: 0.2118 data: 0.1198 max mem: 3531 +Test: [180/235] eta: 0:00:12 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.2356 (1.0359) acc1: 29.1667 (44.1442) acc5: 100.0000 (100.0000) acc1_neg: 29.167 acc5_neg: 100.000 time: 0.2138 data: 0.1221 max mem: 3531 +Test: [190/235] eta: 0:00:09 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.2040 (1.0455) acc1: 30.4688 (43.4037) acc5: 100.0000 (100.0000) acc1_neg: 30.729 acc5_neg: 100.000 time: 0.2081 data: 0.1143 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.2142 (1.0552) acc1: 29.1667 (42.6837) acc5: 100.0000 (100.0000) acc1_neg: 28.646 acc5_neg: 100.000 time: 0.2474 data: 0.1527 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.2465 (1.0637) acc1: 28.6458 (42.0591) acc5: 100.0000 (100.0000) acc1_neg: 26.302 acc5_neg: 100.000 time: 0.2099 data: 0.1180 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.2376 (1.0713) acc1: 29.4271 (41.4758) acc5: 100.0000 (100.0000) acc1_neg: 29.948 acc5_neg: 100.000 time: 0.1317 data: 0.0412 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.2252 (1.0779) acc1: 29.9479 (40.9835) acc5: 100.0000 (100.0000) acc1_neg: 29.167 acc5_neg: 100.000 time: 0.1272 data: 0.0374 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.2287 (1.0808) acc1: 29.9479 (40.8293) acc5: 100.0000 (100.0000) acc1_neg: 34.783 acc5_neg: 100.000 time: 0.1240 data: 0.0373 max mem: 3532 +Test: Total time: 0:00:49 (0.2109 s / it) +* Overall Acc@1 40.829 Acc@5 100.000 + target == real Acc@1 93.411 Acc@5 100.000 + target == fake Acc@1 21.812 Acc@5 100.000 + loss 1.081 + PR-AUC 0.8496 ROC-AUC 0.6798 EER 36.94% (thr=0.1848) +Accuracy of the network on the 89879 test images: 40.8% +[rank0]:[W726 03:19:18.846519515 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:19:37.757000 1167240 torch/distributed/run.py:766] +W0726 03:19:37.757000 1167240 torch/distributed/run.py:766] ***************************************** +W0726 03:19:37.757000 1167240 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:19:37.757000 1167240 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 03:20:06.122067181 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:20:06.529343772 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +NCCL version 2.26.2+cuda12.2 +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:20:06.723216229 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:20:07.021813916 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:20:07.052789960 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +| distributed init (rank 7): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:20:07.135793554 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W726 03:20:07.135836817 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank6]:[W726 03:20:07.135846760 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n11_seed5_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n11_classseed5_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:26:32 acc1_pos: 52.344 acc5_pos: 100.000 loss: 0.7151 (0.7151) acc1: 52.3438 (52.3438) acc5: 100.0000 (100.0000) time: 5.9185 data: 4.4590 max mem: 3531 +Test: [ 10/269] eta: 0:02:40 acc1_pos: 54.948 acc5_pos: 100.000 loss: 0.6257 (0.6524) acc1: 66.4062 (63.1155) acc5: 100.0000 (100.0000) time: 0.6212 data: 0.4058 max mem: 3531 +Test: [ 20/269] eta: 0:01:32 acc1_pos: 67.448 acc5_pos: 100.000 loss: 0.6257 (0.6585) acc1: 63.8021 (61.5327) acc5: 100.0000 (100.0000) time: 0.0943 data: 0.0005 max mem: 3531 +Test: [ 30/269] eta: 0:01:12 acc1_pos: 83.854 acc5_pos: 100.000 loss: 0.6187 (0.6533) acc1: 59.8958 (63.0628) acc5: 100.0000 (100.0000) time: 0.1290 data: 0.0314 max mem: 3531 +Test: [ 40/269] eta: 0:01:04 acc1_pos: 78.448 acc5_pos: 100.000 loss: 0.5896 (0.6397) acc1: 69.5312 (65.0661) acc5: 100.0000 (100.0000) acc1_neg: 63.281 acc5_neg: 100.000 time: 0.1857 data: 0.0890 max mem: 3531 +Test: [ 50/269] eta: 0:00:59 acc1_pos: 78.448 acc5_pos: 100.000 loss: 0.6357 (0.6545) acc1: 64.8438 (63.4702) acc5: 100.0000 (100.0000) acc1_neg: 32.812 acc5_neg: 100.000 time: 0.2257 data: 0.1299 max mem: 3531 +Test: [ 60/269] eta: 0:00:54 acc1_pos: 78.448 acc5_pos: 100.000 loss: 0.7666 (0.6846) acc1: 49.4792 (59.8788) acc5: 100.0000 (100.0000) acc1_neg: 40.104 acc5_neg: 100.000 time: 0.2141 data: 0.1175 max mem: 3531 +Test: [ 70/269] eta: 0:00:49 acc1_pos: 78.448 acc5_pos: 100.000 loss: 0.7168 (0.6835) acc1: 53.1250 (59.6941) acc5: 100.0000 (100.0000) acc1_neg: 56.510 acc5_neg: 100.000 time: 0.1937 data: 0.0975 max mem: 3531 +Test: [ 80/269] eta: 0:00:45 acc1_pos: 78.448 acc5_pos: 100.000 loss: 0.6591 (0.6806) acc1: 57.5521 (59.5968) acc5: 100.0000 (100.0000) acc1_neg: 65.625 acc5_neg: 100.000 time: 0.1877 data: 0.0933 max mem: 3531 +Test: [ 90/269] eta: 0:00:42 acc1_pos: 78.448 acc5_pos: 100.000 loss: 0.6748 (0.6862) acc1: 55.9896 (58.6853) acc5: 100.0000 (100.0000) acc1_neg: 70.312 acc5_neg: 100.000 time: 0.1855 data: 0.0912 max mem: 3531 +Test: [100/269] eta: 0:00:39 acc1_pos: 78.448 acc5_pos: 100.000 loss: 0.6456 (0.6771) acc1: 59.1146 (59.7643) acc5: 100.0000 (100.0000) acc1_neg: 59.115 acc5_neg: 100.000 time: 0.1944 data: 0.0994 max mem: 3531 +Test: [110/269] eta: 0:00:36 acc1_pos: 78.448 acc5_pos: 100.000 loss: 0.6395 (0.6776) acc1: 62.5000 (59.4337) acc5: 100.0000 (100.0000) acc1_neg: 54.948 acc5_neg: 100.000 time: 0.1993 data: 0.1030 max mem: 3531 +Test: [120/269] eta: 0:00:33 acc1_pos: 78.448 acc5_pos: 100.000 loss: 0.6395 (0.6691) acc1: 62.2396 (60.2553) acc5: 100.0000 (100.0000) acc1_neg: 69.531 acc5_neg: 100.000 time: 0.1903 data: 0.0960 max mem: 3531 +Test: [130/269] eta: 0:00:30 acc1_pos: 78.448 acc5_pos: 100.000 loss: 0.5581 (0.6638) acc1: 69.0104 (60.8321) acc5: 100.0000 (100.0000) acc1_neg: 69.010 acc5_neg: 100.000 time: 0.1808 data: 0.0897 max mem: 3531 +Test: [140/269] eta: 0:00:28 acc1_pos: 78.448 acc5_pos: 100.000 loss: 0.6206 (0.6617) acc1: 65.3646 (61.0557) acc5: 100.0000 (100.0000) acc1_neg: 65.365 acc5_neg: 100.000 time: 0.1781 data: 0.0867 max mem: 3531 +Test: [150/269] eta: 0:00:25 acc1_pos: 78.448 acc5_pos: 100.000 loss: 0.6179 (0.6581) acc1: 64.3229 (61.5032) acc5: 100.0000 (100.0000) acc1_neg: 78.646 acc5_neg: 100.000 time: 0.1840 data: 0.0925 max mem: 3531 +Test: [160/269] eta: 0:00:23 acc1_pos: 78.448 acc5_pos: 100.000 loss: 0.5612 (0.6532) acc1: 72.3958 (62.0325) acc5: 100.0000 (100.0000) acc1_neg: 58.594 acc5_neg: 100.000 time: 0.1858 data: 0.0891 max mem: 3531 +Test: [170/269] eta: 0:00:21 acc1_pos: 78.448 acc5_pos: 100.000 loss: 0.6031 (0.6515) acc1: 65.3646 (62.2533) acc5: 100.0000 (100.0000) acc1_neg: 65.365 acc5_neg: 100.000 time: 0.1837 data: 0.0861 max mem: 3531 +Test: [180/269] eta: 0:00:18 acc1_pos: 78.448 acc5_pos: 100.000 loss: 0.5999 (0.6450) acc1: 68.7500 (62.9503) acc5: 100.0000 (100.0000) acc1_neg: 62.760 acc5_neg: 100.000 time: 0.1802 data: 0.0880 max mem: 3531 +Test: [190/269] eta: 0:00:16 acc1_pos: 78.448 acc5_pos: 100.000 loss: 0.5559 (0.6427) acc1: 71.3542 (63.1013) acc5: 100.0000 (100.0000) acc1_neg: 72.396 acc5_neg: 100.000 time: 0.1761 data: 0.0847 max mem: 3531 +Test: [200/269] eta: 0:00:14 acc1_pos: 78.448 acc5_pos: 100.000 loss: 0.6244 (0.6444) acc1: 63.8021 (62.9379) acc5: 100.0000 (100.0000) acc1_neg: 71.354 acc5_neg: 100.000 time: 0.1828 data: 0.0905 max mem: 3531 +Test: [210/269] eta: 0:00:12 acc1_pos: 78.448 acc5_pos: 100.000 loss: 0.6642 (0.6425) acc1: 58.8542 (63.1048) acc5: 100.0000 (100.0000) acc1_neg: 76.302 acc5_neg: 100.000 time: 0.1799 data: 0.0849 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 78.448 acc5_pos: 100.000 loss: 0.5704 (0.6403) acc1: 68.7500 (63.3626) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1810 data: 0.0859 max mem: 3532 +Test: [230/269] eta: 0:00:07 acc1_pos: 78.448 acc5_pos: 100.000 loss: 0.5174 (0.6353) acc1: 73.1771 (63.8314) acc5: 100.0000 (100.0000) acc1_neg: 47.656 acc5_neg: 100.000 time: 0.1877 data: 0.0925 max mem: 3532 +Test: [240/269] eta: 0:00:05 acc1_pos: 78.448 acc5_pos: 100.000 loss: 0.5846 (0.6370) acc1: 68.4896 (63.6540) acc5: 100.0000 (100.0000) acc1_neg: 70.312 acc5_neg: 100.000 time: 0.1859 data: 0.0910 max mem: 3532 +Test: [250/269] eta: 0:00:03 acc1_pos: 78.448 acc5_pos: 100.000 loss: 0.6370 (0.6362) acc1: 63.2812 (63.7378) acc5: 100.0000 (100.0000) acc1_neg: 59.635 acc5_neg: 100.000 time: 0.1705 data: 0.0778 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 78.448 acc5_pos: 100.000 loss: 0.6651 (0.6412) acc1: 58.8542 (63.1266) acc5: 100.0000 (100.0000) acc1_neg: 40.625 acc5_neg: 100.000 time: 0.1482 data: 0.0561 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 78.448 acc5_pos: 100.000 loss: 0.7589 (0.6438) acc1: 46.8750 (62.7196) acc5: 100.0000 (100.0000) acc1_neg: 56.168 acc5_neg: 100.000 time: 0.1317 data: 0.0375 max mem: 3532 +Test: Total time: 0:00:53 (0.1980 s / it) +* Overall Acc@1 62.720 Acc@5 100.000 + target == real Acc@1 63.246 Acc@5 100.000 + target == fake Acc@1 62.648 Acc@5 100.000 + loss 0.644 + PR-AUC 0.9320 ROC-AUC 0.6763 EER 37.01% (thr=0.4980) +Accuracy of the network on the 103293 test images: 62.7% +[rank0]:[W726 03:21:10.313865893 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T03:21:17Z] eval_done tag=fixed500k_real_n11_seed5_resampled +[2026-07-26T03:21:17Z] eval_start tag=fixed500k_fake_n22_seed5_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed5_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T03:21:17Z +hostname=n116-041-188 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed5_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n22_seed5_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 03:21:29.928000 1167804 torch/distributed/run.py:766] +W0726 03:21:29.928000 1167804 torch/distributed/run.py:766] ***************************************** +W0726 03:21:29.928000 1167804 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:21:29.928000 1167804 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 03:21:56.238811531 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:21:56.404394989 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:21:56.439920443 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:21:57.922749501 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:21:57.063837611 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:21:57.074170851 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:21:57.101523944 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:21:57.227452292 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n22_seed5_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed5_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:19:36 acc1_pos: 77.865 acc5_pos: 100.000 loss: 0.4964 (0.4964) acc1: 77.8646 (77.8646) acc5: 100.0000 (100.0000) time: 5.0274 data: 3.1039 max mem: 3531 +Test: [ 10/234] eta: 0:02:00 acc1_pos: 40.625 acc5_pos: 100.000 loss: 0.6354 (0.6422) acc1: 64.8438 (64.2045) acc5: 100.0000 (100.0000) time: 0.5394 data: 0.2830 max mem: 3531 +Test: [ 20/234] eta: 0:01:22 acc1_pos: 95.573 acc5_pos: 100.000 loss: 0.6367 (0.5915) acc1: 64.8438 (67.0883) acc5: 100.0000 (100.0000) time: 0.1524 data: 0.0585 max mem: 3531 +Test: [ 30/234] eta: 0:01:03 acc1_pos: 77.604 acc5_pos: 100.000 loss: 0.2303 (0.4786) acc1: 94.7917 (75.6972) acc5: 100.0000 (100.0000) time: 0.1832 data: 0.0858 max mem: 3531 +Test: [ 40/234] eta: 0:00:58 acc1_pos: 99.219 acc5_pos: 100.000 loss: 0.2231 (0.4472) acc1: 95.0521 (77.7630) acc5: 100.0000 (100.0000) time: 0.2099 data: 0.1163 max mem: 3531 +Test: [ 50/234] eta: 0:00:52 acc1_pos: 99.740 acc5_pos: 100.000 loss: 0.1198 (0.3827) acc1: 98.9583 (82.0159) acc5: 100.0000 (100.0000) time: 0.2438 data: 0.1531 max mem: 3531 +Test: [ 60/234] eta: 0:00:47 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.1003 (0.3351) acc1: 99.7396 (84.9556) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2249 data: 0.1322 max mem: 3531 +Test: [ 70/234] eta: 0:00:43 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.0949 (0.3538) acc1: 100.0000 (83.8138) acc5: 100.0000 (100.0000) acc1_neg: 65.885 acc5_neg: 100.000 time: 0.2199 data: 0.1252 max mem: 3531 +Test: [ 80/234] eta: 0:00:40 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.6243 (0.3698) acc1: 65.8854 (82.9154) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2366 data: 0.1429 max mem: 3531 +Test: [ 90/234] eta: 0:00:37 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.1939 (0.3481) acc1: 92.9688 (84.3035) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.2430 data: 0.1513 max mem: 3531 +Test: [100/234] eta: 0:00:34 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.1878 (0.3478) acc1: 92.9688 (84.2564) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2232 data: 0.1322 max mem: 3531 +Test: [110/234] eta: 0:00:31 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.3751 (0.3517) acc1: 79.1667 (84.1451) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.2287 data: 0.1363 max mem: 3531 +Test: [120/234] eta: 0:00:27 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.3420 (0.3624) acc1: 83.0729 (83.7745) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1811 data: 0.0881 max mem: 3531 +Test: [130/234] eta: 0:00:25 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.3923 (0.3665) acc1: 85.4167 (83.8780) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.1640 data: 0.0713 max mem: 3531 +Test: [140/234] eta: 0:00:22 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.4489 (0.3947) acc1: 80.7292 (82.3120) acc5: 100.0000 (100.0000) acc1_neg: 34.635 acc5_neg: 100.000 time: 0.2290 data: 0.1368 max mem: 3531 +Test: [150/234] eta: 0:00:20 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.3885 (0.3990) acc1: 77.3438 (82.0226) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.2215 data: 0.1286 max mem: 3531 +Test: [160/234] eta: 0:00:17 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.3761 (0.3983) acc1: 81.2500 (82.1461) acc5: 100.0000 (100.0000) acc1_neg: 72.396 acc5_neg: 100.000 time: 0.1897 data: 0.0954 max mem: 3531 +Test: [170/234] eta: 0:00:14 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.4895 (0.4035) acc1: 76.8229 (81.8957) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1972 data: 0.1037 max mem: 3531 +Test: [180/234] eta: 0:00:12 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.4895 (0.4145) acc1: 76.8229 (81.1651) acc5: 100.0000 (100.0000) acc1_neg: 39.583 acc5_neg: 100.000 time: 0.1919 data: 0.0998 max mem: 3531 +Test: [190/234] eta: 0:00:09 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.7225 (0.4323) acc1: 60.4167 (80.0406) acc5: 100.0000 (100.0000) acc1_neg: 60.417 acc5_neg: 100.000 time: 0.1631 data: 0.0721 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.7216 (0.4279) acc1: 61.4583 (80.3016) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1408 data: 0.0505 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.1187 (0.4135) acc1: 99.7396 (81.2154) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1350 data: 0.0446 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.1271 (0.4023) acc1: 99.2188 (81.9111) acc5: 100.0000 (100.0000) acc1_neg: 74.219 acc5_neg: 100.000 time: 0.1263 data: 0.0362 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.4719 (0.4331) acc1: 31.5104 (79.5657) acc5: 100.0000 (100.0000) acc1_neg: 27.083 acc5_neg: 100.000 time: 0.1233 data: 0.0335 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 99.310 acc5_pos: 100.000 loss: 1.0796 (0.4416) acc1: 29.4271 (78.9834) acc5: 100.0000 (100.0000) acc1_neg: 29.183 acc5_neg: 100.000 time: 0.1207 data: 0.0299 max mem: 3532 +Test: Total time: 0:00:48 (0.2086 s / it) +* Overall Acc@1 78.983 Acc@5 100.000 + target == real Acc@1 82.143 Acc@5 100.000 + target == fake Acc@1 78.093 Acc@5 100.000 + loss 0.442 + PR-AUC 0.9683 ROC-AUC 0.8893 EER 20.47% (thr=0.4741) +Accuracy of the network on the 89729 test images: 79.0% +[rank0]:[W726 03:22:53.641798974 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:23:12.400000 1168362 torch/distributed/run.py:766] +W0726 03:23:12.400000 1168362 torch/distributed/run.py:766] ***************************************** +W0726 03:23:12.400000 1168362 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:23:12.400000 1168362 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 03:23:40.496059811 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:23:40.545681517 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:23:40.810686779 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:23:41.924165270 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:23:41.932221176 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:23:41.981862818 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:23:41.006278923 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:23:41.078157458 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n22_seed5_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed5_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:14:02 acc1_pos: 34.375 acc5_pos: 100.000 loss: 0.8607 (0.8607) acc1: 34.3750 (34.3750) acc5: 100.0000 (100.0000) time: 6.3857 data: 4.6900 max mem: 3531 +Test: [ 10/2387] eta: 0:26:26 acc1_pos: 10.156 acc5_pos: 100.000 loss: 0.9865 (0.9830) acc1: 25.2604 (28.2907) acc5: 100.0000 (100.0000) time: 0.6674 data: 0.4266 max mem: 3531 +Test: [ 20/2387] eta: 0:15:30 acc1_pos: 27.865 acc5_pos: 100.000 loss: 0.9980 (1.0142) acc1: 25.2604 (26.3021) acc5: 100.0000 (100.0000) time: 0.0936 data: 0.0003 max mem: 3531 +Test: [ 30/2387] eta: 0:12:38 acc1_pos: 10.677 acc5_pos: 100.000 loss: 1.0212 (1.0266) acc1: 22.3958 (26.3693) acc5: 100.0000 (100.0000) time: 0.1319 data: 0.0389 max mem: 3531 +Test: [ 40/2387] eta: 0:12:10 acc1_pos: 31.250 acc5_pos: 100.000 loss: 1.0468 (1.0341) acc1: 21.3542 (24.7840) acc5: 100.0000 (100.0000) time: 0.2251 data: 0.1300 max mem: 3531 +Test: [ 50/2387] eta: 0:11:31 acc1_pos: 26.823 acc5_pos: 100.000 loss: 1.0037 (1.0283) acc1: 21.3542 (25.6230) acc5: 100.0000 (100.0000) time: 0.2559 data: 0.1618 max mem: 3531 +Test: [ 60/2387] eta: 0:11:02 acc1_pos: 37.500 acc5_pos: 100.000 loss: 0.9789 (1.0241) acc1: 26.8229 (26.3917) acc5: 100.0000 (100.0000) time: 0.2297 data: 0.1363 max mem: 3531 +Test: [ 70/2387] eta: 0:10:34 acc1_pos: 32.031 acc5_pos: 100.000 loss: 1.0819 (1.0359) acc1: 26.0417 (25.8619) acc5: 100.0000 (100.0000) time: 0.2178 data: 0.1228 max mem: 3531 +Test: [ 80/2387] eta: 0:10:19 acc1_pos: 79.167 acc5_pos: 100.000 loss: 1.0361 (0.9967) acc1: 28.6458 (29.5332) acc5: 100.0000 (100.0000) time: 0.2199 data: 0.1260 max mem: 3531 +Test: [ 90/2387] eta: 0:10:09 acc1_pos: 69.010 acc5_pos: 100.000 loss: 0.6204 (0.9507) acc1: 67.1875 (34.0602) acc5: 100.0000 (100.0000) time: 0.2349 data: 0.1424 max mem: 3531 +Test: [ 100/2387] eta: 0:10:05 acc1_pos: 64.583 acc5_pos: 100.000 loss: 0.5698 (0.9185) acc1: 69.0104 (37.2061) acc5: 100.0000 (100.0000) time: 0.2499 data: 0.1555 max mem: 3531 +Test: [ 110/2387] eta: 0:09:53 acc1_pos: 72.917 acc5_pos: 100.000 loss: 0.5865 (0.8883) acc1: 68.4896 (40.3575) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1464 max mem: 3531 +Test: [ 120/2387] eta: 0:09:51 acc1_pos: 50.260 acc5_pos: 100.000 loss: 0.6167 (0.8687) acc1: 68.4896 (42.3339) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1446 max mem: 3531 +Test: [ 130/2387] eta: 0:09:43 acc1_pos: 60.156 acc5_pos: 100.000 loss: 0.5941 (0.8474) acc1: 68.7500 (44.6048) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1486 max mem: 3531 +Test: [ 140/2387] eta: 0:09:44 acc1_pos: 73.177 acc5_pos: 100.000 loss: 0.5870 (0.8313) acc1: 69.0104 (46.1529) acc5: 100.0000 (100.0000) time: 0.2549 data: 0.1586 max mem: 3531 +Test: [ 150/2387] eta: 0:09:34 acc1_pos: 75.000 acc5_pos: 100.000 loss: 0.5743 (0.8124) acc1: 68.2292 (48.1012) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1479 max mem: 3531 +Test: [ 160/2387] eta: 0:09:33 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.5543 (0.7947) acc1: 72.1354 (49.7493) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2399 data: 0.1471 max mem: 3531 +Test: [ 170/2387] eta: 0:09:11 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.2272 (0.7607) acc1: 95.5729 (52.5204) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1892 data: 0.0949 max mem: 3531 +Test: [ 180/2387] eta: 0:09:09 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.2136 (0.7305) acc1: 97.1354 (54.9896) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1792 data: 0.0807 max mem: 3531 +Test: [ 190/2387] eta: 0:09:02 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.2119 (0.7033) acc1: 97.1354 (57.2058) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2299 data: 0.1290 max mem: 3531 +Test: [ 200/2387] eta: 0:08:56 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.2105 (0.6788) acc1: 97.6562 (59.2014) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2149 data: 0.1172 max mem: 3531 +Test: [ 210/2387] eta: 0:08:50 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.2160 (0.6570) acc1: 96.8750 (60.9819) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2149 data: 0.1203 max mem: 3532 +Test: [ 220/2387] eta: 0:08:44 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.2164 (0.6371) acc1: 96.8750 (62.6178) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2099 data: 0.1132 max mem: 3532 +Test: [ 230/2387] eta: 0:08:33 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.2164 (0.6188) acc1: 97.1354 (64.0974) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1808 data: 0.0846 max mem: 3532 +Test: [ 240/2387] eta: 0:08:29 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.2240 (0.6075) acc1: 95.8333 (65.1398) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.1829 data: 0.0892 max mem: 3532 +Test: [ 250/2387] eta: 0:08:25 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.2047 (0.5875) acc1: 89.8438 (66.4685) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2140 data: 0.1218 max mem: 3532 +Test: [ 260/2387] eta: 0:08:19 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.0733 (0.5677) acc1: 100.0000 (67.7532) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2047 data: 0.1124 max mem: 3532 +Test: [ 270/2387] eta: 0:08:14 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.0685 (0.5491) acc1: 100.0000 (68.9432) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2006 data: 0.1091 max mem: 3532 +Test: [ 280/2387] eta: 0:08:07 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.0685 (0.5321) acc1: 100.0000 (70.0484) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1891 data: 0.0976 max mem: 3532 +Test: [ 290/2387] eta: 0:08:02 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.0729 (0.5164) acc1: 100.0000 (71.0776) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1841 data: 0.0916 max mem: 3532 +Test: [ 300/2387] eta: 0:07:58 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.0764 (0.5019) acc1: 100.0000 (72.0385) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1941 data: 0.0976 max mem: 3532 +Test: [ 310/2387] eta: 0:07:53 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.0820 (0.4906) acc1: 100.0000 (72.9217) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1927 data: 0.0974 max mem: 3532 +Test: [ 320/2387] eta: 0:07:49 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.2149 (0.4822) acc1: 99.7396 (73.7563) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1986 data: 0.1059 max mem: 3532 +Test: [ 330/2387] eta: 0:07:45 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.2160 (0.4741) acc1: 99.7396 (74.5421) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2021 data: 0.1077 max mem: 3532 +Test: [ 340/2387] eta: 0:07:40 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.2138 (0.4665) acc1: 99.7396 (75.2818) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1905 data: 0.0982 max mem: 3532 +Test: [ 350/2387] eta: 0:07:36 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.2084 (0.4577) acc1: 100.0000 (75.9845) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1851 data: 0.0945 max mem: 3532 +Test: [ 360/2387] eta: 0:07:32 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1663 (0.4499) acc1: 99.4792 (76.6144) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1953 data: 0.1041 max mem: 3532 +Test: [ 370/2387] eta: 0:07:30 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1761 (0.4426) acc1: 98.6979 (77.2055) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2097 data: 0.1175 max mem: 3532 +Test: [ 380/2387] eta: 0:07:25 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1775 (0.4356) acc1: 98.4375 (77.7668) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1953 data: 0.1021 max mem: 3532 +Test: [ 390/2387] eta: 0:07:24 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1811 (0.4291) acc1: 98.4375 (78.2955) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2155 data: 0.1233 max mem: 3532 +Test: [ 400/2387] eta: 0:07:20 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1813 (0.4229) acc1: 98.4375 (78.7959) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2189 data: 0.1260 max mem: 3532 +Test: [ 410/2387] eta: 0:07:15 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1757 (0.4168) acc1: 98.6979 (79.2845) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1718 data: 0.0768 max mem: 3532 +Test: [ 420/2387] eta: 0:07:11 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1757 (0.4111) acc1: 98.9583 (79.7494) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1715 data: 0.0779 max mem: 3532 +Test: [ 430/2387] eta: 0:07:07 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1811 (0.4074) acc1: 98.4375 (80.0978) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1799 data: 0.0882 max mem: 3532 +Test: [ 440/2387] eta: 0:07:02 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1847 (0.4018) acc1: 95.0521 (80.4947) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1665 data: 0.0735 max mem: 3533 +Test: [ 450/2387] eta: 0:06:59 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.0754 (0.3945) acc1: 100.0000 (80.9272) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1723 data: 0.0750 max mem: 3533 +Test: [ 460/2387] eta: 0:06:56 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.0745 (0.3876) acc1: 100.0000 (81.3409) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1964 data: 0.0998 max mem: 3533 +Test: [ 470/2387] eta: 0:06:55 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.0745 (0.3810) acc1: 100.0000 (81.7371) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2199 data: 0.1230 max mem: 3533 +Test: [ 480/2387] eta: 0:06:50 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.0749 (0.3746) acc1: 100.0000 (82.1168) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2017 data: 0.1056 max mem: 3533 +Test: [ 490/2387] eta: 0:06:49 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.0752 (0.3685) acc1: 100.0000 (82.4810) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1999 data: 0.1049 max mem: 3533 +Test: [ 500/2387] eta: 0:06:46 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.0753 (0.3627) acc1: 100.0000 (82.8307) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2181 data: 0.1212 max mem: 3533 +Test: [ 510/2387] eta: 0:06:42 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.0756 (0.3577) acc1: 100.0000 (83.1662) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1849 data: 0.0881 max mem: 3533 +Test: [ 520/2387] eta: 0:06:40 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1342 (0.3542) acc1: 100.0000 (83.4883) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1918 data: 0.0967 max mem: 3533 +Test: [ 530/2387] eta: 0:06:38 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1321 (0.3500) acc1: 100.0000 (83.7992) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2071 data: 0.1155 max mem: 3533 +Test: [ 540/2387] eta: 0:06:36 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1315 (0.3460) acc1: 100.0000 (84.0987) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2088 data: 0.1162 max mem: 3533 +Test: [ 550/2387] eta: 0:06:33 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1313 (0.3421) acc1: 100.0000 (84.3868) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2101 data: 0.1171 max mem: 3533 +Test: [ 560/2387] eta: 0:06:31 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1309 (0.3384) acc1: 100.0000 (84.6651) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2092 data: 0.1132 max mem: 3533 +Test: [ 570/2387] eta: 0:06:28 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1305 (0.3347) acc1: 100.0000 (84.9337) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2048 data: 0.1078 max mem: 3533 +Test: [ 580/2387] eta: 0:06:25 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1308 (0.3312) acc1: 100.0000 (85.1926) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1877 data: 0.0952 max mem: 3533 +Test: [ 590/2387] eta: 0:06:23 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1319 (0.3279) acc1: 100.0000 (85.4431) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1929 data: 0.1018 max mem: 3533 +Test: [ 600/2387] eta: 0:06:20 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1357 (0.3257) acc1: 100.0000 (85.6840) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1991 data: 0.1076 max mem: 3533 +Test: [ 610/2387] eta: 0:06:17 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1464 (0.3226) acc1: 100.0000 (85.9170) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1967 data: 0.1056 max mem: 3533 +Test: [ 620/2387] eta: 0:06:15 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1369 (0.3196) acc1: 100.0000 (86.1438) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1982 data: 0.1067 max mem: 3533 +Test: [ 630/2387] eta: 0:06:12 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1377 (0.3168) acc1: 100.0000 (86.3630) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1947 data: 0.1027 max mem: 3533 +Test: [ 640/2387] eta: 0:06:10 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1375 (0.3140) acc1: 100.0000 (86.5749) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1957 data: 0.1012 max mem: 3533 +Test: [ 650/2387] eta: 0:06:07 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1392 (0.3113) acc1: 100.0000 (86.7800) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1931 data: 0.0977 max mem: 3533 +Test: [ 660/2387] eta: 0:06:04 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1400 (0.3087) acc1: 100.0000 (86.9796) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1838 data: 0.0905 max mem: 3533 +Test: [ 670/2387] eta: 0:06:01 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1398 (0.3062) acc1: 100.0000 (87.1728) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1867 data: 0.0948 max mem: 3534 +Test: [ 680/2387] eta: 0:05:59 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1408 (0.3049) acc1: 100.0000 (87.3597) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1951 data: 0.1038 max mem: 3534 +Test: [ 690/2387] eta: 0:05:56 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.2571 (0.3061) acc1: 99.7396 (87.3165) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.1853 data: 0.0923 max mem: 3534 +Test: [ 700/2387] eta: 0:05:53 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.4255 (0.3079) acc1: 79.1667 (87.1917) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.1756 data: 0.0824 max mem: 3534 +Test: [ 710/2387] eta: 0:05:50 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.4278 (0.3096) acc1: 79.1667 (87.0751) acc5: 100.0000 (100.0000) acc1_neg: 80.729 acc5_neg: 100.000 time: 0.1779 data: 0.0858 max mem: 3534 +Test: [ 720/2387] eta: 0:05:48 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.4256 (0.3112) acc1: 78.9062 (86.9622) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.1827 data: 0.0910 max mem: 3534 +Test: [ 730/2387] eta: 0:05:45 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.4295 (0.3128) acc1: 78.6458 (86.8470) acc5: 100.0000 (100.0000) acc1_neg: 79.427 acc5_neg: 100.000 time: 0.1888 data: 0.0972 max mem: 3534 +Test: [ 740/2387] eta: 0:05:43 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.4295 (0.3144) acc1: 78.6458 (86.7381) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.1920 data: 0.1006 max mem: 3534 +Test: [ 750/2387] eta: 0:05:40 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.4221 (0.3158) acc1: 78.6458 (86.6341) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.1916 data: 0.0948 max mem: 3534 +Test: [ 760/2387] eta: 0:05:38 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.4151 (0.3171) acc1: 79.1667 (86.5535) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.2001 data: 0.1020 max mem: 3534 +Test: [ 770/2387] eta: 0:05:35 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.3723 (0.3167) acc1: 83.0729 (86.6022) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1891 data: 0.0956 max mem: 3534 +Test: [ 780/2387] eta: 0:05:33 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1770 (0.3148) acc1: 97.1354 (86.7454) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1807 data: 0.0875 max mem: 3534 +Test: [ 790/2387] eta: 0:05:30 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1432 (0.3128) acc1: 98.9583 (86.8639) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.1887 data: 0.0954 max mem: 3534 +Test: [ 800/2387] eta: 0:05:30 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1322 (0.3106) acc1: 99.4792 (87.0218) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2318 data: 0.1385 max mem: 3534 +Test: [ 810/2387] eta: 0:05:26 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1468 (0.3087) acc1: 99.2188 (87.1638) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1998 data: 0.1038 max mem: 3534 +Test: [ 820/2387] eta: 0:05:23 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1680 (0.3073) acc1: 98.4375 (87.2894) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.1449 data: 0.0496 max mem: 3534 +Test: [ 830/2387] eta: 0:05:21 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.2693 (0.3080) acc1: 92.4479 (87.2643) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.1832 data: 0.0909 max mem: 3534 +Test: [ 840/2387] eta: 0:05:19 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.2394 (0.3063) acc1: 93.2292 (87.4052) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2073 data: 0.1139 max mem: 3534 +Test: [ 850/2387] eta: 0:05:16 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1543 (0.3046) acc1: 100.0000 (87.5480) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1869 data: 0.0943 max mem: 3534 +Test: [ 860/2387] eta: 0:05:14 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1488 (0.3027) acc1: 100.0000 (87.6902) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1738 data: 0.0831 max mem: 3534 +Test: [ 870/2387] eta: 0:05:11 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1481 (0.3010) acc1: 100.0000 (87.8271) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1922 data: 0.1001 max mem: 3534 +Test: [ 880/2387] eta: 0:05:09 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1512 (0.2993) acc1: 99.7396 (87.9605) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0979 max mem: 3534 +Test: [ 890/2387] eta: 0:05:07 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1666 (0.2981) acc1: 99.7396 (88.0857) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1938 data: 0.1031 max mem: 3534 +Test: [ 900/2387] eta: 0:05:05 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.2317 (0.3015) acc1: 95.3125 (87.8653) acc5: 100.0000 (100.0000) acc1_neg: 61.198 acc5_neg: 100.000 time: 0.1972 data: 0.1050 max mem: 3535 +Test: [ 910/2387] eta: 0:05:03 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.6727 (0.3056) acc1: 61.1979 (87.5826) acc5: 100.0000 (100.0000) acc1_neg: 64.323 acc5_neg: 100.000 time: 0.1993 data: 0.1052 max mem: 3535 +Test: [ 920/2387] eta: 0:05:00 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.6818 (0.3096) acc1: 60.9375 (87.2956) acc5: 100.0000 (100.0000) acc1_neg: 61.198 acc5_neg: 100.000 time: 0.2013 data: 0.1078 max mem: 3535 +Test: [ 930/2387] eta: 0:04:58 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.6770 (0.3135) acc1: 61.1979 (87.0234) acc5: 100.0000 (100.0000) acc1_neg: 57.292 acc5_neg: 100.000 time: 0.2051 data: 0.1125 max mem: 3535 +Test: [ 940/2387] eta: 0:04:56 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.6820 (0.3174) acc1: 60.9375 (86.7545) acc5: 100.0000 (100.0000) acc1_neg: 61.979 acc5_neg: 100.000 time: 0.1973 data: 0.1049 max mem: 3535 +Test: [ 950/2387] eta: 0:04:54 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.6820 (0.3211) acc1: 61.1979 (86.4898) acc5: 100.0000 (100.0000) acc1_neg: 61.458 acc5_neg: 100.000 time: 0.1918 data: 0.0985 max mem: 3535 +Test: [ 960/2387] eta: 0:04:52 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.6638 (0.3247) acc1: 62.2396 (86.2413) acc5: 100.0000 (100.0000) acc1_neg: 64.583 acc5_neg: 100.000 time: 0.2129 data: 0.1200 max mem: 3535 +Test: [ 970/2387] eta: 0:04:50 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.6722 (0.3284) acc1: 61.4583 (85.9780) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.2083 data: 0.1162 max mem: 3535 +Test: [ 980/2387] eta: 0:04:48 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.6501 (0.3296) acc1: 63.2812 (85.9216) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.1945 data: 0.1002 max mem: 3535 +Test: [ 990/2387] eta: 0:04:46 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.2795 (0.3285) acc1: 93.4896 (86.0027) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2039 data: 0.1103 max mem: 3535 +Test: [1000/2387] eta: 0:04:43 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.2018 (0.3273) acc1: 94.7917 (86.0928) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1851 data: 0.0920 max mem: 3535 +Test: [1010/2387] eta: 0:04:41 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.2237 (0.3267) acc1: 94.5312 (86.1456) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1748 data: 0.0800 max mem: 3535 +Test: [1020/2387] eta: 0:04:38 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.2216 (0.3263) acc1: 93.4896 (86.1747) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1823 data: 0.0882 max mem: 3535 +Test: [1030/2387] eta: 0:04:36 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1612 (0.3249) acc1: 97.6562 (86.2876) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1811 data: 0.0875 max mem: 3535 +Test: [1040/2387] eta: 0:04:34 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1414 (0.3229) acc1: 99.7396 (86.4178) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1880 data: 0.0952 max mem: 3535 +Test: [1050/2387] eta: 0:04:32 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1243 (0.3212) acc1: 99.7396 (86.5366) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1847 data: 0.0930 max mem: 3535 +Test: [1060/2387] eta: 0:04:29 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1376 (0.3195) acc1: 99.7396 (86.6569) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1850 data: 0.0937 max mem: 3535 +Test: [1070/2387] eta: 0:04:27 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1541 (0.3185) acc1: 99.2188 (86.7433) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1844 data: 0.0933 max mem: 3535 +Test: [1080/2387] eta: 0:04:25 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.2464 (0.3187) acc1: 92.1875 (86.7510) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1866 data: 0.0933 max mem: 3535 +Test: [1090/2387] eta: 0:04:23 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.2464 (0.3179) acc1: 92.1875 (86.8274) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2089 data: 0.1146 max mem: 3535 +Test: [1100/2387] eta: 0:04:21 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.2223 (0.3169) acc1: 95.8333 (86.9167) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1922 data: 0.0978 max mem: 3535 +Test: [1110/2387] eta: 0:04:19 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.2126 (0.3162) acc1: 96.0938 (86.9693) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1770 data: 0.0829 max mem: 3535 +Test: [1120/2387] eta: 0:04:16 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.2181 (0.3154) acc1: 95.5729 (87.0410) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1849 data: 0.0933 max mem: 3536 +Test: [1130/2387] eta: 0:04:14 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.2134 (0.3145) acc1: 95.8333 (87.1169) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1859 data: 0.0938 max mem: 3536 +Test: [1140/2387] eta: 0:04:12 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.2701 (0.3148) acc1: 91.9271 (87.1166) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.2131 data: 0.1202 max mem: 3536 +Test: [1150/2387] eta: 0:04:10 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.3295 (0.3141) acc1: 89.0625 (87.1688) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1991 data: 0.1062 max mem: 3536 +Test: [1160/2387] eta: 0:04:08 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1248 (0.3123) acc1: 100.0000 (87.2759) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1839 data: 0.0902 max mem: 3536 +Test: [1170/2387] eta: 0:04:06 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1082 (0.3106) acc1: 100.0000 (87.3844) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1926 data: 0.0994 max mem: 3536 +Test: [1180/2387] eta: 0:04:04 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1034 (0.3089) acc1: 100.0000 (87.4903) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1979 data: 0.1060 max mem: 3536 +Test: [1190/2387] eta: 0:04:02 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1095 (0.3073) acc1: 100.0000 (87.5953) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2001 data: 0.1081 max mem: 3536 +Test: [1200/2387] eta: 0:04:00 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1122 (0.3057) acc1: 100.0000 (87.6986) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1954 data: 0.1035 max mem: 3536 +Test: [1210/2387] eta: 0:03:57 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1179 (0.3042) acc1: 100.0000 (87.7981) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1932 data: 0.1022 max mem: 3536 +Test: [1220/2387] eta: 0:03:55 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.0796 (0.3023) acc1: 100.0000 (87.8976) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1852 data: 0.0938 max mem: 3536 +Test: [1230/2387] eta: 0:03:53 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.0701 (0.3004) acc1: 100.0000 (87.9959) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1864 data: 0.0938 max mem: 3536 +Test: [1240/2387] eta: 0:03:51 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.0649 (0.2985) acc1: 100.0000 (88.0926) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1891 data: 0.0971 max mem: 3536 +Test: [1250/2387] eta: 0:03:49 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.0664 (0.2967) acc1: 100.0000 (88.1878) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1917 data: 0.1013 max mem: 3536 +Test: [1260/2387] eta: 0:03:46 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.0693 (0.2949) acc1: 100.0000 (88.2815) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1720 data: 0.0808 max mem: 3536 +Test: [1270/2387] eta: 0:03:44 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.0788 (0.2933) acc1: 100.0000 (88.3735) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1539 data: 0.0624 max mem: 3536 +Test: [1280/2387] eta: 0:03:42 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1042 (0.2920) acc1: 100.0000 (88.4608) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1568 data: 0.0656 max mem: 3536 +Test: [1290/2387] eta: 0:03:39 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1254 (0.2907) acc1: 99.4792 (88.5310) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1526 data: 0.0614 max mem: 3536 +Test: [1300/2387] eta: 0:03:37 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1254 (0.2895) acc1: 99.4792 (88.6103) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1465 data: 0.0554 max mem: 3536 +Test: [1310/2387] eta: 0:03:34 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1265 (0.2883) acc1: 99.2188 (88.6843) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1504 data: 0.0599 max mem: 3536 +Test: [1320/2387] eta: 0:03:32 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1518 (0.2874) acc1: 98.9583 (88.7556) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1510 data: 0.0604 max mem: 3536 +Test: [1330/2387] eta: 0:03:29 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1728 (0.2871) acc1: 95.8333 (88.7833) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1472 data: 0.0564 max mem: 3536 +Test: [1340/2387] eta: 0:03:27 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.2907 (0.2878) acc1: 88.5417 (88.7339) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1554 data: 0.0646 max mem: 3536 +Test: [1350/2387] eta: 0:03:25 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.0478 (0.2860) acc1: 100.0000 (88.8173) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1521 data: 0.0590 max mem: 3537 +Test: [1360/2387] eta: 0:03:22 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.0398 (0.2842) acc1: 100.0000 (88.8995) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1463 data: 0.0538 max mem: 3537 +Test: [1370/2387] eta: 0:03:20 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.0380 (0.2824) acc1: 100.0000 (88.9804) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1507 data: 0.0584 max mem: 3537 +Test: [1380/2387] eta: 0:03:18 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.0403 (0.2806) acc1: 100.0000 (89.0602) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1516 data: 0.0592 max mem: 3537 +Test: [1390/2387] eta: 0:03:16 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.0463 (0.2790) acc1: 100.0000 (89.1389) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1543 data: 0.0618 max mem: 3537 +Test: [1400/2387] eta: 0:03:13 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.0481 (0.2773) acc1: 100.0000 (89.2164) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1530 data: 0.0589 max mem: 3537 +Test: [1410/2387] eta: 0:03:11 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.0557 (0.2767) acc1: 100.0000 (89.2604) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1507 data: 0.0558 max mem: 3537 +Test: [1420/2387] eta: 0:03:09 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.2082 (0.2764) acc1: 94.7917 (89.2936) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1772 data: 0.0838 max mem: 3537 +Test: [1430/2387] eta: 0:03:07 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.2252 (0.2760) acc1: 95.5729 (89.3340) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2088 data: 0.1154 max mem: 3537 +Test: [1440/2387] eta: 0:03:05 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.2316 (0.2759) acc1: 95.5729 (89.3672) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.2077 data: 0.1131 max mem: 3537 +Test: [1450/2387] eta: 0:03:03 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.2790 (0.2763) acc1: 91.9271 (89.3471) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1960 data: 0.1021 max mem: 3537 +Test: [1460/2387] eta: 0:03:01 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1619 (0.2754) acc1: 97.1354 (89.4095) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1968 data: 0.1033 max mem: 3537 +Test: [1470/2387] eta: 0:03:00 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1162 (0.2743) acc1: 99.7396 (89.4807) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2074 data: 0.1133 max mem: 3537 +Test: [1480/2387] eta: 0:02:58 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1117 (0.2732) acc1: 99.7396 (89.5480) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2041 data: 0.1108 max mem: 3537 +Test: [1490/2387] eta: 0:02:56 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1140 (0.2722) acc1: 99.7396 (89.6155) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1978 data: 0.1047 max mem: 3537 +Test: [1500/2387] eta: 0:02:54 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1184 (0.2712) acc1: 99.7396 (89.6814) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1994 data: 0.1061 max mem: 3537 +Test: [1510/2387] eta: 0:02:52 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1296 (0.2705) acc1: 99.7396 (89.7428) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2023 data: 0.1106 max mem: 3537 +Test: [1520/2387] eta: 0:02:50 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1514 (0.2697) acc1: 99.4792 (89.8033) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1960 data: 0.1032 max mem: 3537 +Test: [1530/2387] eta: 0:02:48 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1389 (0.2688) acc1: 99.7396 (89.8693) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1780 data: 0.0852 max mem: 3537 +Test: [1540/2387] eta: 0:02:46 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1279 (0.2679) acc1: 100.0000 (89.9342) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2052 data: 0.1118 max mem: 3537 +Test: [1550/2387] eta: 0:02:44 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1268 (0.2670) acc1: 100.0000 (89.9970) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2353 data: 0.1430 max mem: 3537 +Test: [1560/2387] eta: 0:02:42 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1268 (0.2661) acc1: 100.0000 (90.0601) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1945 data: 0.1038 max mem: 3537 +Test: [1570/2387] eta: 0:02:40 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1349 (0.2654) acc1: 99.7396 (90.1191) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1876 data: 0.0955 max mem: 3537 +Test: [1580/2387] eta: 0:02:38 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1676 (0.2650) acc1: 99.2188 (90.1625) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2181 data: 0.1218 max mem: 3538 +Test: [1590/2387] eta: 0:02:36 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1767 (0.2644) acc1: 98.1771 (90.2120) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1997 data: 0.1041 max mem: 3538 +Test: [1600/2387] eta: 0:02:34 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1730 (0.2639) acc1: 98.6979 (90.2382) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1924 data: 0.1009 max mem: 3538 +Test: [1610/2387] eta: 0:02:33 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1489 (0.2632) acc1: 98.9583 (90.2928) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2249 data: 0.1329 max mem: 3538 +Test: [1620/2387] eta: 0:02:31 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1537 (0.2626) acc1: 98.4375 (90.3373) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2327 data: 0.1401 max mem: 3538 +Test: [1630/2387] eta: 0:02:29 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1908 (0.2628) acc1: 96.6146 (90.3406) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.2427 data: 0.1482 max mem: 3538 +Test: [1640/2387] eta: 0:02:27 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.3288 (0.2636) acc1: 86.1979 (90.2911) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.1922 data: 0.0985 max mem: 3538 +Test: [1650/2387] eta: 0:02:25 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.3860 (0.2642) acc1: 82.5521 (90.2524) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.1688 data: 0.0761 max mem: 3538 +Test: [1660/2387] eta: 0:02:23 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.3644 (0.2649) acc1: 83.5938 (90.2081) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.2099 data: 0.1172 max mem: 3538 +Test: [1670/2387] eta: 0:02:21 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.3521 (0.2654) acc1: 85.4167 (90.1855) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1798 data: 0.0884 max mem: 3538 +Test: [1680/2387] eta: 0:02:19 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.3457 (0.2660) acc1: 85.4167 (90.1520) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.1776 data: 0.0858 max mem: 3538 +Test: [1690/2387] eta: 0:02:17 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.3846 (0.2667) acc1: 83.5938 (90.1106) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.1894 data: 0.0959 max mem: 3538 +Test: [1700/2387] eta: 0:02:15 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.3703 (0.2668) acc1: 83.0729 (90.1077) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1645 data: 0.0713 max mem: 3538 +Test: [1710/2387] eta: 0:02:12 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.0971 (0.2658) acc1: 99.7396 (90.1649) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1567 data: 0.0631 max mem: 3538 +Test: [1720/2387] eta: 0:02:10 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.0866 (0.2647) acc1: 100.0000 (90.2217) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1702 data: 0.0734 max mem: 3538 +Test: [1730/2387] eta: 0:02:08 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.0826 (0.2637) acc1: 100.0000 (90.2781) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1672 data: 0.0709 max mem: 3538 +Test: [1740/2387] eta: 0:02:06 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.0826 (0.2627) acc1: 100.0000 (90.3338) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1846 data: 0.0904 max mem: 3538 +Test: [1750/2387] eta: 0:02:04 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.0885 (0.2617) acc1: 100.0000 (90.3888) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2078 data: 0.1148 max mem: 3538 +Test: [1760/2387] eta: 0:02:03 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.0947 (0.2609) acc1: 100.0000 (90.4428) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2020 data: 0.1108 max mem: 3538 +Test: [1770/2387] eta: 0:02:01 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1143 (0.2601) acc1: 99.7396 (90.4909) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1927 data: 0.1020 max mem: 3538 +Test: [1780/2387] eta: 0:01:59 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1142 (0.2592) acc1: 99.7396 (90.5433) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1954 data: 0.1030 max mem: 3538 +Test: [1790/2387] eta: 0:01:57 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.0978 (0.2583) acc1: 100.0000 (90.5955) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1991 data: 0.1067 max mem: 3538 +Test: [1800/2387] eta: 0:01:55 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.0978 (0.2575) acc1: 100.0000 (90.6476) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1746 data: 0.0833 max mem: 3538 +Test: [1810/2387] eta: 0:01:52 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1003 (0.2566) acc1: 100.0000 (90.6988) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1510 data: 0.0588 max mem: 3539 +Test: [1820/2387] eta: 0:01:50 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.0909 (0.2558) acc1: 100.0000 (90.7483) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1500 data: 0.0585 max mem: 3539 +Test: [1830/2387] eta: 0:01:48 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1301 (0.2552) acc1: 99.7396 (90.7948) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1538 data: 0.0628 max mem: 3539 +Test: [1840/2387] eta: 0:01:46 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1642 (0.2547) acc1: 99.7396 (90.8433) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1476 data: 0.0551 max mem: 3539 +Test: [1850/2387] eta: 0:01:44 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1663 (0.2543) acc1: 99.7396 (90.8916) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1450 data: 0.0522 max mem: 3539 +Test: [1860/2387] eta: 0:01:42 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1660 (0.2538) acc1: 99.7396 (90.9396) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1466 data: 0.0550 max mem: 3539 +Test: [1870/2387] eta: 0:01:40 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1673 (0.2533) acc1: 99.7396 (90.9870) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1502 data: 0.0575 max mem: 3539 +Test: [1880/2387] eta: 0:01:38 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1666 (0.2529) acc1: 99.7396 (91.0342) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1553 data: 0.0630 max mem: 3539 +Test: [1890/2387] eta: 0:01:36 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1674 (0.2524) acc1: 100.0000 (91.0810) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1568 data: 0.0650 max mem: 3539 +Test: [1900/2387] eta: 0:01:34 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1691 (0.2520) acc1: 100.0000 (91.1254) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1540 data: 0.0618 max mem: 3539 +Test: [1910/2387] eta: 0:01:32 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1729 (0.2522) acc1: 97.6562 (91.1535) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1444 data: 0.0532 max mem: 3539 +Test: [1920/2387] eta: 0:01:30 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.2885 (0.2520) acc1: 96.8750 (91.1667) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1549 data: 0.0639 max mem: 3539 +Test: [1930/2387] eta: 0:01:28 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1496 (0.2514) acc1: 99.7396 (91.2115) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1641 data: 0.0729 max mem: 3539 +Test: [1940/2387] eta: 0:01:26 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1335 (0.2512) acc1: 99.7396 (91.2184) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1461 data: 0.0552 max mem: 3539 +Test: [1950/2387] eta: 0:01:24 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1667 (0.2509) acc1: 98.6979 (91.2466) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1400 data: 0.0493 max mem: 3539 +Test: [1960/2387] eta: 0:01:22 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1667 (0.2505) acc1: 98.6979 (91.2792) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1529 data: 0.0608 max mem: 3539 +Test: [1970/2387] eta: 0:01:20 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.2307 (0.2510) acc1: 92.7083 (91.2544) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.1563 data: 0.0643 max mem: 3539 +Test: [1980/2387] eta: 0:01:18 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.2959 (0.2511) acc1: 90.8854 (91.2534) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1500 data: 0.0591 max mem: 3539 +Test: [1990/2387] eta: 0:01:16 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1603 (0.2506) acc1: 97.6562 (91.2867) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1469 data: 0.0550 max mem: 3539 +Test: [2000/2387] eta: 0:01:14 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1611 (0.2502) acc1: 97.9167 (91.3215) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1482 data: 0.0530 max mem: 3539 +Test: [2010/2387] eta: 0:01:12 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1652 (0.2498) acc1: 97.9167 (91.3537) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1503 data: 0.0538 max mem: 3539 +Test: [2020/2387] eta: 0:01:10 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1659 (0.2494) acc1: 97.9167 (91.3873) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1470 data: 0.0530 max mem: 3539 +Test: [2030/2387] eta: 0:01:08 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1635 (0.2490) acc1: 97.9167 (91.4197) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1390 data: 0.0450 max mem: 3540 +Test: [2040/2387] eta: 0:01:06 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1634 (0.2485) acc1: 97.9167 (91.4530) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1335 data: 0.0369 max mem: 3540 +Test: [2050/2387] eta: 0:01:04 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1643 (0.2482) acc1: 98.1771 (91.4845) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1379 data: 0.0415 max mem: 3540 +Test: [2060/2387] eta: 0:01:02 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1982 (0.2480) acc1: 97.3958 (91.5104) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1403 data: 0.0469 max mem: 3540 +Test: [2070/2387] eta: 0:01:00 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.2072 (0.2478) acc1: 97.3958 (91.5405) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1351 data: 0.0430 max mem: 3540 +Test: [2080/2387] eta: 0:00:58 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.2072 (0.2476) acc1: 97.3958 (91.5663) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1493 data: 0.0555 max mem: 3540 +Test: [2090/2387] eta: 0:00:56 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.2090 (0.2474) acc1: 96.6146 (91.5929) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1616 data: 0.0667 max mem: 3540 +Test: [2100/2387] eta: 0:00:54 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.2064 (0.2472) acc1: 97.3958 (91.6219) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1769 data: 0.0841 max mem: 3540 +Test: [2110/2387] eta: 0:00:52 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.2004 (0.2470) acc1: 97.3958 (91.6495) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1899 data: 0.0982 max mem: 3540 +Test: [2120/2387] eta: 0:00:50 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.2112 (0.2470) acc1: 97.1354 (91.6607) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1882 data: 0.0958 max mem: 3540 +Test: [2130/2387] eta: 0:00:48 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1994 (0.2467) acc1: 97.6562 (91.6883) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1952 data: 0.1032 max mem: 3540 +Test: [2140/2387] eta: 0:00:46 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1836 (0.2466) acc1: 98.4375 (91.7004) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1954 data: 0.1044 max mem: 3540 +Test: [2150/2387] eta: 0:00:44 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1899 (0.2463) acc1: 97.9167 (91.7287) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1896 data: 0.0986 max mem: 3540 +Test: [2160/2387] eta: 0:00:43 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1923 (0.2461) acc1: 98.1771 (91.7554) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1860 data: 0.0947 max mem: 3540 +Test: [2170/2387] eta: 0:00:41 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.2363 (0.2465) acc1: 94.7917 (91.7409) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1849 data: 0.0935 max mem: 3540 +Test: [2180/2387] eta: 0:00:39 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.3551 (0.2482) acc1: 84.8958 (91.6214) acc5: 100.0000 (100.0000) acc1_neg: 40.104 acc5_neg: 100.000 time: 0.1907 data: 0.0991 max mem: 3540 +Test: [2190/2387] eta: 0:00:37 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.4147 (0.2490) acc1: 76.3021 (91.5612) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1956 data: 0.1043 max mem: 3540 +Test: [2200/2387] eta: 0:00:35 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1524 (0.2485) acc1: 97.1354 (91.5948) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1936 data: 0.1017 max mem: 3540 +Test: [2210/2387] eta: 0:00:33 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1194 (0.2479) acc1: 99.7396 (91.6312) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1968 data: 0.1051 max mem: 3540 +Test: [2220/2387] eta: 0:00:31 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1169 (0.2473) acc1: 100.0000 (91.6681) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2031 data: 0.1112 max mem: 3540 +Test: [2230/2387] eta: 0:00:29 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1269 (0.2468) acc1: 100.0000 (91.7038) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1983 data: 0.1059 max mem: 3540 +Test: [2240/2387] eta: 0:00:27 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1177 (0.2462) acc1: 99.7396 (91.7383) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2002 data: 0.1069 max mem: 3540 +Test: [2250/2387] eta: 0:00:25 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1353 (0.2458) acc1: 99.7396 (91.7697) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2045 data: 0.1076 max mem: 3540 +Test: [2260/2387] eta: 0:00:24 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1641 (0.2455) acc1: 99.7396 (91.8023) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2026 data: 0.1068 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1710 (0.2452) acc1: 99.4792 (91.8340) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1975 data: 0.1044 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1400 (0.2448) acc1: 99.4792 (91.8514) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1781 data: 0.0856 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1319 (0.2444) acc1: 100.0000 (91.8864) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1450 data: 0.0534 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1550 (0.2440) acc1: 99.7396 (91.9166) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1287 data: 0.0366 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.1780 (0.2440) acc1: 98.9583 (91.9249) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.1290 data: 0.0354 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.2696 (0.2443) acc1: 90.3646 (91.9096) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1348 data: 0.0422 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.2823 (0.2445) acc1: 91.1458 (91.9146) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1369 data: 0.0464 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.2104 (0.2443) acc1: 96.8750 (91.9345) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1279 data: 0.0377 max mem: 3541 +Test: [2350/2387] eta: 0:00:06 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.2049 (0.2444) acc1: 97.9167 (91.9333) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1139 data: 0.0238 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.2212 (0.2445) acc1: 96.0938 (91.9400) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1055 data: 0.0152 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.2644 (0.2446) acc1: 94.7917 (91.9408) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.1109 data: 0.0196 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.3462 (0.2454) acc1: 84.8958 (91.8958) acc5: 100.0000 (100.0000) acc1_neg: 80.729 acc5_neg: 100.000 time: 0.1104 data: 0.0194 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 67.580 acc5_pos: 100.000 loss: 0.4292 (0.2471) acc1: 79.9479 (91.7920) acc5: 100.0000 (100.0000) acc1_neg: 16.279 acc5_neg: 100.000 time: 0.1067 data: 0.0193 max mem: 3541 +Test: Total time: 0:07:25 (0.1868 s / it) +* Overall Acc@1 91.792 Acc@5 100.000 + target == real Acc@1 48.983 Acc@5 100.000 + target == fake Acc@1 94.798 Acc@5 100.000 + loss 0.247 + PR-AUC 0.9931 ROC-AUC 0.9120 EER 15.84% (thr=0.7197) +Accuracy of the network on the 916267 test images: 91.8% +[rank0]:[W726 03:31:27.424378885 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:31:47.185000 1168982 torch/distributed/run.py:766] +W0726 03:31:47.185000 1168982 torch/distributed/run.py:766] ***************************************** +W0726 03:31:47.185000 1168982 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:31:47.185000 1168982 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:32:14.572588236 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:32:14.718729702 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:32:15.974971704 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:32:15.009883487 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:32:15.039702203 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:32:15.295565698 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:32:15.365643225 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:32:15.397731909 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n22_seed5_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed5_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:00:37 acc1_pos: 34.375 acc5_pos: 100.000 loss: 0.8607 (0.8607) acc1: 34.3750 (34.3750) acc5: 100.0000 (100.0000) time: 7.0354 data: 5.3365 max mem: 3531 +Test: [ 10/517] eta: 0:06:05 acc1_pos: 10.156 acc5_pos: 100.000 loss: 0.9865 (0.9830) acc1: 25.2604 (28.2907) acc5: 100.0000 (100.0000) time: 0.7216 data: 0.4854 max mem: 3531 +Test: [ 20/517] eta: 0:03:30 acc1_pos: 27.865 acc5_pos: 100.000 loss: 0.9980 (1.0142) acc1: 25.2604 (26.3021) acc5: 100.0000 (100.0000) time: 0.0927 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:02:52 acc1_pos: 10.677 acc5_pos: 100.000 loss: 1.0212 (1.0266) acc1: 22.3958 (26.3693) acc5: 100.0000 (100.0000) time: 0.1524 data: 0.0585 max mem: 3531 +Test: [ 40/517] eta: 0:02:48 acc1_pos: 31.250 acc5_pos: 100.000 loss: 1.0468 (1.0341) acc1: 21.3542 (24.7840) acc5: 100.0000 (100.0000) time: 0.2798 data: 0.1843 max mem: 3531 +Test: [ 50/517] eta: 0:02:36 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.9711 (0.9987) acc1: 21.3542 (28.4314) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.3049 data: 0.2076 max mem: 3531 +Test: [ 60/517] eta: 0:02:27 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.4709 (0.9060) acc1: 83.0729 (38.2684) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.2599 data: 0.1634 max mem: 3531 +Test: [ 70/517] eta: 0:02:21 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.4347 (0.8364) acc1: 86.9792 (45.4812) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.2699 data: 0.1757 max mem: 3531 +Test: [ 80/517] eta: 0:02:16 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.4103 (0.7852) acc1: 89.8438 (50.8198) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.2849 data: 0.1874 max mem: 3531 +Test: [ 90/517] eta: 0:02:11 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.3783 (0.7379) acc1: 90.6250 (55.3199) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.2799 data: 0.1780 max mem: 3531 +Test: [100/517] eta: 0:02:07 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.3566 (0.7006) acc1: 94.0104 (59.3905) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2749 data: 0.1747 max mem: 3531 +Test: [110/517] eta: 0:02:02 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.3151 (0.6638) acc1: 97.9167 (62.8285) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2649 data: 0.1664 max mem: 3531 +Test: [120/517] eta: 0:01:58 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.3360 (0.6443) acc1: 94.0104 (64.8954) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.2599 data: 0.1638 max mem: 3531 +Test: [130/517] eta: 0:01:55 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.4289 (0.6267) acc1: 86.7188 (66.6846) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2849 data: 0.1872 max mem: 3531 +Test: [140/517] eta: 0:01:51 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.3583 (0.6057) acc1: 95.5729 (68.9439) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2749 data: 0.1732 max mem: 3531 +Test: [150/517] eta: 0:01:47 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.3192 (0.5850) acc1: 99.4792 (70.8782) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1524 max mem: 3531 +Test: [160/517] eta: 0:01:43 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.3231 (0.5690) acc1: 97.6562 (72.4880) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2599 data: 0.1601 max mem: 3531 +Test: [170/517] eta: 0:01:39 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.3142 (0.5524) acc1: 97.3958 (73.9857) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2549 data: 0.1591 max mem: 3531 +Test: [180/517] eta: 0:01:36 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.2755 (0.5362) acc1: 98.1771 (75.3180) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2549 data: 0.1596 max mem: 3531 +Test: [190/517] eta: 0:01:33 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.2290 (0.5201) acc1: 99.2188 (76.5843) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2749 data: 0.1771 max mem: 3531 +Test: [200/517] eta: 0:01:30 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.2437 (0.5084) acc1: 99.4792 (77.7065) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2599 data: 0.1614 max mem: 3531 +Test: [210/517] eta: 0:01:26 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.3088 (0.4999) acc1: 98.6979 (78.6631) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2449 data: 0.1483 max mem: 3532 +Test: [220/517] eta: 0:01:23 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.3269 (0.4919) acc1: 98.1771 (79.5402) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2599 data: 0.1649 max mem: 3532 +Test: [230/517] eta: 0:01:20 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.3556 (0.4870) acc1: 96.6146 (80.2151) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.2599 data: 0.1648 max mem: 3532 +Test: [240/517] eta: 0:01:17 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.3796 (0.4827) acc1: 94.7917 (80.7097) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.2549 data: 0.1581 max mem: 3532 +Test: [250/517] eta: 0:01:14 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.3469 (0.4775) acc1: 96.8750 (81.3112) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2649 data: 0.1689 max mem: 3532 +Test: [260/517] eta: 0:01:11 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.3240 (0.4720) acc1: 98.4375 (81.9375) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.2649 data: 0.1666 max mem: 3532 +Test: [270/517] eta: 0:01:08 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.4064 (0.4708) acc1: 94.2708 (82.2792) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2499 data: 0.1498 max mem: 3532 +Test: [280/517] eta: 0:01:05 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.4132 (0.4682) acc1: 91.6667 (82.6160) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2649 data: 0.1670 max mem: 3532 +Test: [290/517] eta: 0:01:02 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.3438 (0.4636) acc1: 94.5312 (83.0738) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2499 data: 0.1541 max mem: 3532 +Test: [300/517] eta: 0:00:59 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.3204 (0.4588) acc1: 96.0938 (83.5116) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2399 data: 0.1453 max mem: 3532 +Test: [310/517] eta: 0:00:56 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.3310 (0.4555) acc1: 95.8333 (83.9237) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2499 data: 0.1534 max mem: 3532 +Test: [320/517] eta: 0:00:53 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.3393 (0.4519) acc1: 95.8333 (84.2939) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.2449 data: 0.1494 max mem: 3532 +Test: [330/517] eta: 0:00:50 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.3128 (0.4473) acc1: 96.6146 (84.6685) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2549 data: 0.1593 max mem: 3532 +Test: [340/517] eta: 0:00:48 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.3045 (0.4447) acc1: 97.1354 (84.8309) acc5: 100.0000 (100.0000) acc1_neg: 71.094 acc5_neg: 100.000 time: 0.2549 data: 0.1575 max mem: 3532 +Test: [350/517] eta: 0:00:45 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.4009 (0.4442) acc1: 90.3646 (84.9129) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.2499 data: 0.1530 max mem: 3532 +Test: [360/517] eta: 0:00:42 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.4009 (0.4430) acc1: 90.1042 (85.0524) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.2499 data: 0.1530 max mem: 3532 +Test: [370/517] eta: 0:00:39 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.3392 (0.4394) acc1: 94.7917 (85.4026) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2549 data: 0.1586 max mem: 3532 +Test: [380/517] eta: 0:00:36 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.3241 (0.4378) acc1: 96.6146 (85.5329) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.2399 data: 0.1431 max mem: 3532 +Test: [390/517] eta: 0:00:34 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.3516 (0.4352) acc1: 91.1458 (85.7197) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2299 data: 0.1341 max mem: 3532 +Test: [400/517] eta: 0:00:31 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.3387 (0.4330) acc1: 94.0104 (85.9784) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2399 data: 0.1441 max mem: 3532 +Test: [410/517] eta: 0:00:28 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.3699 (0.4325) acc1: 96.0938 (86.1808) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2499 data: 0.1500 max mem: 3532 +Test: [420/517] eta: 0:00:26 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.4091 (0.4312) acc1: 96.8750 (86.4299) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2769 data: 0.1791 max mem: 3532 +Test: [430/517] eta: 0:00:23 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.3191 (0.4285) acc1: 97.3958 (86.6511) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.2669 data: 0.1715 max mem: 3532 +Test: [440/517] eta: 0:00:20 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.3867 (0.4280) acc1: 94.7917 (86.8115) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2492 data: 0.1526 max mem: 3533 +Test: [450/517] eta: 0:00:17 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.3619 (0.4263) acc1: 96.3542 (87.0358) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2529 data: 0.1561 max mem: 3533 +Test: [460/517] eta: 0:00:15 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.3583 (0.4250) acc1: 96.3542 (87.1887) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.2437 data: 0.1471 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.3711 (0.4239) acc1: 95.3125 (87.3557) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2649 data: 0.1685 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.3786 (0.4228) acc1: 94.5312 (87.4989) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2517 data: 0.1578 max mem: 3533 +Test: [490/517] eta: 0:00:07 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.3725 (0.4209) acc1: 94.2708 (87.6692) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.2030 data: 0.1124 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.3806 (0.4215) acc1: 92.4479 (87.5863) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.1829 data: 0.0929 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.4374 (0.4220) acc1: 83.0729 (87.5663) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1779 data: 0.0880 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 46.789 acc5_pos: 100.000 loss: 0.3996 (0.4215) acc1: 93.7500 (87.6494) acc5: 100.0000 (100.0000) acc1_neg: 95.053 acc5_neg: 100.000 time: 0.1765 data: 0.0849 max mem: 3533 +Test: Total time: 0:02:14 (0.2592 s / it) +* Overall Acc@1 87.649 Acc@5 100.000 + target == real Acc@1 25.560 Acc@5 100.000 + target == fake Acc@1 94.090 Acc@5 100.000 + loss 0.421 + PR-AUC 0.9468 ROC-AUC 0.6930 EER 36.01% (thr=0.6758) +Accuracy of the network on the 198427 test images: 87.6% +[rank0]:[W726 03:34:38.542206841 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 03:34:59.207000 1169554 torch/distributed/run.py:766] +W0726 03:34:59.207000 1169554 torch/distributed/run.py:766] ***************************************** +W0726 03:34:59.207000 1169554 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:34:59.207000 1169554 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 03:35:27.316404025 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:35:27.784694622 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:35:27.871860616 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:35:28.063891693 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:35:28.191569080 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:35:28.197344888 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:35:28.303243545 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W726 03:35:28.303464672 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n22_seed5_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed5_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:37:29 acc1_pos: 26.042 acc5_pos: 100.000 loss: 0.9310 (0.9310) acc1: 26.0417 (26.0417) acc5: 100.0000 (100.0000) time: 5.7824 data: 4.4876 max mem: 3531 +Test: [ 10/389] eta: 0:03:50 acc1_pos: 67.708 acc5_pos: 100.000 loss: 0.7656 (0.7309) acc1: 39.3229 (50.0473) acc5: 100.0000 (100.0000) time: 0.6076 data: 0.4083 max mem: 3531 +Test: [ 20/389] eta: 0:02:15 acc1_pos: 55.469 acc5_pos: 100.000 loss: 0.6962 (0.7177) acc1: 52.6042 (51.5873) acc5: 100.0000 (100.0000) time: 0.0951 data: 0.0003 max mem: 3531 +Test: [ 30/389] eta: 0:01:45 acc1_pos: 58.854 acc5_pos: 100.000 loss: 0.6589 (0.6830) acc1: 57.2917 (56.0736) acc5: 100.0000 (100.0000) time: 0.1225 data: 0.0237 max mem: 3531 +Test: [ 40/389] eta: 0:01:36 acc1_pos: 5.729 acc5_pos: 100.000 loss: 0.6589 (0.7147) acc1: 55.4688 (52.7693) acc5: 100.0000 (100.0000) time: 0.1810 data: 0.0834 max mem: 3531 +Test: [ 50/389] eta: 0:01:24 acc1_pos: 53.385 acc5_pos: 100.000 loss: 0.7690 (0.7069) acc1: 51.0417 (54.9632) acc5: 100.0000 (100.0000) time: 0.1824 data: 0.0875 max mem: 3531 +Test: [ 60/389] eta: 0:01:19 acc1_pos: 14.583 acc5_pos: 100.000 loss: 0.7514 (0.6939) acc1: 51.0417 (56.3695) acc5: 100.0000 (100.0000) time: 0.1741 data: 0.0805 max mem: 3531 +Test: [ 70/389] eta: 0:01:16 acc1_pos: 72.396 acc5_pos: 100.000 loss: 0.7601 (0.7142) acc1: 50.0000 (54.0933) acc5: 100.0000 (100.0000) time: 0.2071 data: 0.1147 max mem: 3531 +Test: [ 80/389] eta: 0:01:11 acc1_pos: 92.188 acc5_pos: 100.000 loss: 0.7286 (0.7048) acc1: 54.9479 (56.1118) acc5: 100.0000 (100.0000) time: 0.2001 data: 0.1092 max mem: 3531 +Test: [ 90/389] eta: 0:01:08 acc1_pos: 27.604 acc5_pos: 100.000 loss: 0.4441 (0.6819) acc1: 83.5938 (58.0844) acc5: 100.0000 (100.0000) time: 0.1961 data: 0.1047 max mem: 3531 +Test: [100/389] eta: 0:01:05 acc1_pos: 23.438 acc5_pos: 100.000 loss: 0.7022 (0.6942) acc1: 53.6458 (56.6677) acc5: 100.0000 (100.0000) time: 0.1969 data: 0.1050 max mem: 3531 +Test: [110/389] eta: 0:01:01 acc1_pos: 99.219 acc5_pos: 100.000 loss: 0.8632 (0.7062) acc1: 38.2812 (55.6424) acc5: 100.0000 (100.0000) time: 0.1903 data: 0.0976 max mem: 3531 +Test: [120/389] eta: 0:00:58 acc1_pos: 58.854 acc5_pos: 100.000 loss: 0.7300 (0.7090) acc1: 54.4271 (55.5161) acc5: 100.0000 (100.0000) time: 0.1888 data: 0.0969 max mem: 3531 +Test: [130/389] eta: 0:00:56 acc1_pos: 55.828 acc5_pos: 100.000 loss: 0.6556 (0.7019) acc1: 59.8958 (56.1884) acc5: 100.0000 (100.0000) acc1_neg: 68.490 acc5_neg: 100.000 time: 0.1963 data: 0.1042 max mem: 3531 +Test: [140/389] eta: 0:00:53 acc1_pos: 55.828 acc5_pos: 100.000 loss: 0.5521 (0.6955) acc1: 69.5312 (57.0035) acc5: 100.0000 (100.0000) acc1_neg: 18.490 acc5_neg: 100.000 time: 0.2023 data: 0.1101 max mem: 3531 +Test: [150/389] eta: 0:00:50 acc1_pos: 55.828 acc5_pos: 100.000 loss: 0.5335 (0.6886) acc1: 70.5729 (57.4555) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1621 data: 0.0712 max mem: 3531 +Test: [160/389] eta: 0:00:47 acc1_pos: 55.828 acc5_pos: 100.000 loss: 0.5389 (0.6778) acc1: 57.2917 (58.5857) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1463 data: 0.0522 max mem: 3531 +Test: [170/389] eta: 0:00:44 acc1_pos: 55.828 acc5_pos: 100.000 loss: 0.3345 (0.6622) acc1: 97.1354 (60.2705) acc5: 100.0000 (100.0000) acc1_neg: 76.042 acc5_neg: 100.000 time: 0.1619 data: 0.0680 max mem: 3531 +Test: [180/389] eta: 0:00:42 acc1_pos: 55.828 acc5_pos: 100.000 loss: 0.4939 (0.6556) acc1: 83.8542 (61.3217) acc5: 100.0000 (100.0000) acc1_neg: 64.323 acc5_neg: 100.000 time: 0.1540 data: 0.0634 max mem: 3531 +Test: [190/389] eta: 0:00:39 acc1_pos: 55.828 acc5_pos: 100.000 loss: 0.5205 (0.6521) acc1: 79.1667 (61.6165) acc5: 100.0000 (100.0000) acc1_neg: 55.469 acc5_neg: 100.000 time: 0.1605 data: 0.0681 max mem: 3531 +Test: [200/389] eta: 0:00:37 acc1_pos: 55.828 acc5_pos: 100.000 loss: 0.4881 (0.6459) acc1: 76.0417 (62.3989) acc5: 100.0000 (100.0000) acc1_neg: 60.677 acc5_neg: 100.000 time: 0.1627 data: 0.0700 max mem: 3531 +Test: [210/389] eta: 0:00:35 acc1_pos: 55.828 acc5_pos: 100.000 loss: 0.3286 (0.6334) acc1: 87.2396 (63.5034) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1573 data: 0.0651 max mem: 3532 +Test: [220/389] eta: 0:00:33 acc1_pos: 55.828 acc5_pos: 100.000 loss: 0.2721 (0.6173) acc1: 100.0000 (64.8532) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.1751 data: 0.0816 max mem: 3532 +Test: [230/389] eta: 0:00:31 acc1_pos: 55.828 acc5_pos: 100.000 loss: 0.4432 (0.6128) acc1: 83.3333 (65.3747) acc5: 100.0000 (100.0000) acc1_neg: 64.323 acc5_neg: 100.000 time: 0.1873 data: 0.0948 max mem: 3532 +Test: [240/389] eta: 0:00:29 acc1_pos: 55.828 acc5_pos: 100.000 loss: 0.4580 (0.6065) acc1: 81.7708 (66.0788) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1958 data: 0.1017 max mem: 3532 +Test: [250/389] eta: 0:00:27 acc1_pos: 55.828 acc5_pos: 100.000 loss: 0.4580 (0.6000) acc1: 88.8021 (67.0485) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1993 data: 0.1039 max mem: 3532 +Test: [260/389] eta: 0:00:25 acc1_pos: 55.828 acc5_pos: 100.000 loss: 0.4812 (0.5986) acc1: 88.2812 (67.4010) acc5: 100.0000 (100.0000) acc1_neg: 25.521 acc5_neg: 100.000 time: 0.1965 data: 0.1040 max mem: 3532 +Test: [270/389] eta: 0:00:23 acc1_pos: 55.828 acc5_pos: 100.000 loss: 0.5397 (0.6021) acc1: 75.5208 (66.9617) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.2004 data: 0.1083 max mem: 3532 +Test: [280/389] eta: 0:00:21 acc1_pos: 55.828 acc5_pos: 100.000 loss: 0.5839 (0.6049) acc1: 68.4896 (66.7779) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1994 data: 0.1061 max mem: 3532 +Test: [290/389] eta: 0:00:19 acc1_pos: 55.828 acc5_pos: 100.000 loss: 0.5344 (0.6018) acc1: 78.1250 (67.2859) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.1993 data: 0.1037 max mem: 3532 +Test: [300/389] eta: 0:00:17 acc1_pos: 55.828 acc5_pos: 100.000 loss: 0.4974 (0.5995) acc1: 86.9792 (67.8096) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.2049 data: 0.1105 max mem: 3532 +Test: [310/389] eta: 0:00:15 acc1_pos: 55.828 acc5_pos: 100.000 loss: 0.5320 (0.6012) acc1: 80.7292 (67.7326) acc5: 100.0000 (100.0000) acc1_neg: 40.104 acc5_neg: 100.000 time: 0.2064 data: 0.1151 max mem: 3532 +Test: [320/389] eta: 0:00:13 acc1_pos: 55.828 acc5_pos: 100.000 loss: 0.5193 (0.5984) acc1: 78.1250 (68.1797) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1910 data: 0.1001 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 55.828 acc5_pos: 100.000 loss: 0.7188 (0.6047) acc1: 51.3021 (67.2300) acc5: 100.0000 (100.0000) acc1_neg: 33.594 acc5_neg: 100.000 time: 0.1872 data: 0.0957 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 55.828 acc5_pos: 100.000 loss: 0.5159 (0.5974) acc1: 65.1042 (68.0505) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1906 data: 0.0988 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 55.828 acc5_pos: 100.000 loss: 0.3755 (0.6026) acc1: 95.8333 (67.6987) acc5: 100.0000 (100.0000) acc1_neg: 34.635 acc5_neg: 100.000 time: 0.1770 data: 0.0862 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 55.828 acc5_pos: 100.000 loss: 0.6071 (0.6040) acc1: 64.3229 (67.4782) acc5: 100.0000 (100.0000) acc1_neg: 72.396 acc5_neg: 100.000 time: 0.1491 data: 0.0587 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 55.828 acc5_pos: 100.000 loss: 0.5878 (0.6026) acc1: 65.8854 (67.5687) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1237 data: 0.0330 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 55.828 acc5_pos: 100.000 loss: 0.4449 (0.6018) acc1: 78.9062 (67.5894) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1150 data: 0.0249 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 55.828 acc5_pos: 100.000 loss: 0.4027 (0.5991) acc1: 84.1146 (67.8424) acc5: 100.0000 (100.0000) acc1_neg: 58.743 acc5_neg: 100.000 time: 0.1127 data: 0.0197 max mem: 3532 +Test: Total time: 0:01:13 (0.1883 s / it) +* Overall Acc@1 67.842 Acc@5 100.000 + target == real Acc@1 55.656 Acc@5 100.000 + target == fake Acc@1 73.507 Acc@5 100.000 + loss 0.599 + PR-AUC 0.8216 ROC-AUC 0.7055 EER 35.28% (thr=0.5513) +Accuracy of the network on the 149358 test images: 67.8% +[rank0]:[W726 03:36:51.995157471 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:37:10.430000 1170115 torch/distributed/run.py:766] +W0726 03:37:10.430000 1170115 torch/distributed/run.py:766] ***************************************** +W0726 03:37:10.430000 1170115 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:37:10.430000 1170115 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:37:36.571700215 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:37:37.132563852 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:37:37.440250749 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:37:37.462394966 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:37:37.524065901 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 6): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:37:37.668482138 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:37:37.676028405 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:37:37.700589233 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n22_seed5_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed5_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:33:06 acc1_pos: 78.385 acc5_pos: 100.000 loss: 0.4731 (0.4731) acc1: 78.3854 (78.3854) acc5: 100.0000 (100.0000) time: 5.9650 data: 4.5136 max mem: 3531 +Test: [ 10/333] eta: 0:03:21 acc1_pos: 86.719 acc5_pos: 100.000 loss: 0.3720 (0.3838) acc1: 85.4167 (82.5284) acc5: 100.0000 (100.0000) time: 0.6242 data: 0.4107 max mem: 3531 +Test: [ 20/333] eta: 0:01:56 acc1_pos: 93.750 acc5_pos: 100.000 loss: 0.3638 (0.3657) acc1: 86.1979 (84.2262) acc5: 100.0000 (100.0000) time: 0.0942 data: 0.0004 max mem: 3531 +Test: [ 30/333] eta: 0:01:42 acc1_pos: 56.250 acc5_pos: 100.000 loss: 0.4662 (0.4860) acc1: 71.0938 (74.7984) acc5: 100.0000 (100.0000) time: 0.1791 data: 0.0830 max mem: 3531 +Test: [ 40/333] eta: 0:01:35 acc1_pos: 46.875 acc5_pos: 100.000 loss: 0.7489 (0.5562) acc1: 51.8229 (68.8961) acc5: 100.0000 (100.0000) time: 0.2767 data: 0.1841 max mem: 3531 +Test: [ 50/333] eta: 0:01:25 acc1_pos: 52.344 acc5_pos: 100.000 loss: 0.7494 (0.5942) acc1: 51.8229 (65.4973) acc5: 100.0000 (100.0000) time: 0.2505 data: 0.1600 max mem: 3531 +Test: [ 60/333] eta: 0:01:20 acc1_pos: 39.062 acc5_pos: 100.000 loss: 0.7494 (0.6265) acc1: 52.3438 (62.7732) acc5: 100.0000 (100.0000) time: 0.2280 data: 0.1357 max mem: 3531 +Test: [ 70/333] eta: 0:01:12 acc1_pos: 54.688 acc5_pos: 100.000 loss: 0.7616 (0.6440) acc1: 53.1250 (61.5940) acc5: 100.0000 (100.0000) time: 0.2043 data: 0.1085 max mem: 3531 +Test: [ 80/333] eta: 0:01:04 acc1_pos: 64.844 acc5_pos: 100.000 loss: 0.7563 (0.6586) acc1: 52.8646 (60.2109) acc5: 100.0000 (100.0000) time: 0.1311 data: 0.0339 max mem: 3531 +Test: [ 90/333] eta: 0:01:00 acc1_pos: 66.667 acc5_pos: 100.000 loss: 0.7054 (0.6576) acc1: 54.1667 (60.3394) acc5: 100.0000 (100.0000) time: 0.1517 data: 0.0575 max mem: 3531 +Test: [100/333] eta: 0:00:56 acc1_pos: 67.969 acc5_pos: 100.000 loss: 0.6454 (0.6571) acc1: 62.5000 (60.4321) acc5: 100.0000 (100.0000) time: 0.2041 data: 0.1135 max mem: 3531 +Test: [110/333] eta: 0:00:53 acc1_pos: 67.448 acc5_pos: 100.000 loss: 0.6505 (0.6564) acc1: 59.1146 (60.5387) acc5: 100.0000 (100.0000) time: 0.1988 data: 0.1072 max mem: 3531 +Test: [120/333] eta: 0:00:50 acc1_pos: 65.104 acc5_pos: 100.000 loss: 0.6103 (0.6539) acc1: 59.1146 (60.7503) acc5: 100.0000 (100.0000) time: 0.1898 data: 0.0970 max mem: 3531 +Test: [130/333] eta: 0:00:47 acc1_pos: 71.094 acc5_pos: 100.000 loss: 0.6081 (0.6508) acc1: 65.6250 (61.2039) acc5: 100.0000 (100.0000) time: 0.2044 data: 0.1117 max mem: 3531 +Test: [140/333] eta: 0:00:44 acc1_pos: 76.823 acc5_pos: 100.000 loss: 0.5873 (0.6459) acc1: 66.6667 (61.6689) acc5: 100.0000 (100.0000) time: 0.1909 data: 0.0959 max mem: 3531 +Test: [150/333] eta: 0:00:40 acc1_pos: 63.542 acc5_pos: 100.000 loss: 0.6118 (0.6457) acc1: 64.3229 (61.7050) acc5: 100.0000 (100.0000) time: 0.1517 data: 0.0553 max mem: 3531 +Test: [160/333] eta: 0:00:37 acc1_pos: 54.605 acc5_pos: 100.000 loss: 0.6461 (0.6547) acc1: 58.3333 (61.0831) acc5: 100.0000 (100.0000) acc1_neg: 38.802 acc5_neg: 100.000 time: 0.1490 data: 0.0543 max mem: 3531 +Test: [170/333] eta: 0:00:35 acc1_pos: 54.605 acc5_pos: 100.000 loss: 0.8249 (0.6665) acc1: 46.8750 (60.2446) acc5: 100.0000 (100.0000) acc1_neg: 52.083 acc5_neg: 100.000 time: 0.1944 data: 0.1021 max mem: 3531 +Test: [180/333] eta: 0:00:33 acc1_pos: 54.605 acc5_pos: 100.000 loss: 0.8162 (0.6729) acc1: 48.1771 (59.7592) acc5: 100.0000 (100.0000) acc1_neg: 61.458 acc5_neg: 100.000 time: 0.2044 data: 0.1118 max mem: 3531 +Test: [190/333] eta: 0:00:30 acc1_pos: 54.605 acc5_pos: 100.000 loss: 0.7238 (0.6724) acc1: 55.4688 (59.8222) acc5: 100.0000 (100.0000) acc1_neg: 55.469 acc5_neg: 100.000 time: 0.1812 data: 0.0868 max mem: 3531 +Test: [200/333] eta: 0:00:28 acc1_pos: 54.605 acc5_pos: 100.000 loss: 0.6767 (0.6751) acc1: 57.5521 (59.7235) acc5: 100.0000 (100.0000) acc1_neg: 54.167 acc5_neg: 100.000 time: 0.1827 data: 0.0898 max mem: 3531 +Test: [210/333] eta: 0:00:26 acc1_pos: 54.605 acc5_pos: 100.000 loss: 0.7864 (0.6815) acc1: 52.8646 (59.3158) acc5: 100.0000 (100.0000) acc1_neg: 52.083 acc5_neg: 100.000 time: 0.1921 data: 0.1003 max mem: 3532 +Test: [220/333] eta: 0:00:24 acc1_pos: 54.605 acc5_pos: 100.000 loss: 0.7952 (0.6863) acc1: 52.0833 (59.0345) acc5: 100.0000 (100.0000) acc1_neg: 65.885 acc5_neg: 100.000 time: 0.1979 data: 0.1056 max mem: 3532 +Test: [230/333] eta: 0:00:21 acc1_pos: 54.605 acc5_pos: 100.000 loss: 0.5490 (0.6796) acc1: 69.5312 (59.7628) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.1748 data: 0.0819 max mem: 3532 +Test: [240/333] eta: 0:00:19 acc1_pos: 54.605 acc5_pos: 100.000 loss: 0.5490 (0.6775) acc1: 70.3125 (59.9445) acc5: 100.0000 (100.0000) acc1_neg: 63.542 acc5_neg: 100.000 time: 0.1748 data: 0.0800 max mem: 3532 +Test: [250/333] eta: 0:00:17 acc1_pos: 54.605 acc5_pos: 100.000 loss: 0.6466 (0.6772) acc1: 62.7604 (60.0390) acc5: 100.0000 (100.0000) acc1_neg: 54.167 acc5_neg: 100.000 time: 0.1999 data: 0.1055 max mem: 3532 +Test: [260/333] eta: 0:00:15 acc1_pos: 54.605 acc5_pos: 100.000 loss: 0.6827 (0.6784) acc1: 60.1562 (60.0186) acc5: 100.0000 (100.0000) acc1_neg: 57.552 acc5_neg: 100.000 time: 0.2340 data: 0.1382 max mem: 3532 +Test: [270/333] eta: 0:00:13 acc1_pos: 54.605 acc5_pos: 100.000 loss: 0.6893 (0.6793) acc1: 58.0729 (59.9525) acc5: 100.0000 (100.0000) acc1_neg: 56.771 acc5_neg: 100.000 time: 0.2449 data: 0.1470 max mem: 3532 +Test: [280/333] eta: 0:00:11 acc1_pos: 54.605 acc5_pos: 100.000 loss: 0.6992 (0.6825) acc1: 55.2083 (59.6614) acc5: 100.0000 (100.0000) acc1_neg: 59.375 acc5_neg: 100.000 time: 0.2405 data: 0.1462 max mem: 3532 +Test: [290/333] eta: 0:00:09 acc1_pos: 54.605 acc5_pos: 100.000 loss: 0.8341 (0.6887) acc1: 45.3125 (59.1361) acc5: 100.0000 (100.0000) acc1_neg: 43.229 acc5_neg: 100.000 time: 0.1849 data: 0.0923 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 54.605 acc5_pos: 100.000 loss: 0.8737 (0.6961) acc1: 43.2292 (58.5860) acc5: 100.0000 (100.0000) acc1_neg: 46.875 acc5_neg: 100.000 time: 0.1668 data: 0.0748 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 54.605 acc5_pos: 100.000 loss: 0.8353 (0.6939) acc1: 46.8750 (58.8391) acc5: 100.0000 (100.0000) acc1_neg: 69.792 acc5_neg: 100.000 time: 0.1992 data: 0.1084 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 54.605 acc5_pos: 100.000 loss: 0.5795 (0.6914) acc1: 70.0521 (59.1170) acc5: 100.0000 (100.0000) acc1_neg: 39.323 acc5_neg: 100.000 time: 0.1673 data: 0.0768 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 54.605 acc5_pos: 100.000 loss: 0.8838 (0.6994) acc1: 45.3125 (58.6480) acc5: 100.0000 (100.0000) acc1_neg: 45.312 acc5_neg: 100.000 time: 0.1400 data: 0.0499 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 54.605 acc5_pos: 100.000 loss: 0.9224 (0.7008) acc1: 45.3125 (58.5734) acc5: 100.0000 (100.0000) acc1_neg: 45.634 acc5_neg: 100.000 time: 0.1333 data: 0.0399 max mem: 3532 +Test: Total time: 0:01:08 (0.2054 s / it) +* Overall Acc@1 58.573 Acc@5 100.000 + target == real Acc@1 61.713 Acc@5 100.000 + target == fake Acc@1 55.779 Acc@5 100.000 + loss 0.701 + PR-AUC 0.6335 ROC-AUC 0.6260 EER 41.22% (thr=0.4810) +Accuracy of the network on the 127843 test images: 58.6% +[rank0]:[W726 03:38:55.211737056 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:39:14.360000 1170674 torch/distributed/run.py:766] +W0726 03:39:14.360000 1170674 torch/distributed/run.py:766] ***************************************** +W0726 03:39:14.360000 1170674 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:39:14.360000 1170674 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:39:42.392228731 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:39:42.708757949 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:39:43.226881552 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:39:43.263080214 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:39:43.361424060 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:39:43.431464347 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:39:43.470129260 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:39:43.529395743 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n22_seed5_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed5_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:25:05 acc1_pos: 96.615 acc5_pos: 100.000 loss: 0.1806 (0.1806) acc1: 96.6146 (96.6146) acc5: 100.0000 (100.0000) time: 6.4048 data: 4.6331 max mem: 3531 +Test: [ 10/235] eta: 0:02:29 acc1_pos: 96.094 acc5_pos: 100.000 loss: 0.1878 (0.1895) acc1: 96.3542 (96.3068) acc5: 100.0000 (100.0000) time: 0.6647 data: 0.4216 max mem: 3531 +Test: [ 20/235] eta: 0:01:34 acc1_pos: 95.052 acc5_pos: 100.000 loss: 0.1860 (0.1870) acc1: 96.3542 (96.4038) acc5: 100.0000 (100.0000) time: 0.1432 data: 0.0504 max mem: 3531 +Test: [ 30/235] eta: 0:01:26 acc1_pos: 97.396 acc5_pos: 100.000 loss: 0.1813 (0.1854) acc1: 96.6146 (96.5642) acc5: 100.0000 (100.0000) time: 0.2863 data: 0.1915 max mem: 3531 +Test: [ 40/235] eta: 0:01:13 acc1_pos: 95.052 acc5_pos: 100.000 loss: 0.1847 (0.1859) acc1: 96.6146 (96.5384) acc5: 100.0000 (100.0000) time: 0.3090 data: 0.2162 max mem: 3531 +Test: [ 50/235] eta: 0:01:01 acc1_pos: 95.833 acc5_pos: 100.000 loss: 0.1861 (0.1866) acc1: 96.3542 (96.5278) acc5: 100.0000 (100.0000) time: 0.1924 data: 0.1013 max mem: 3531 +Test: [ 60/235] eta: 0:00:54 acc1_pos: 95.833 acc5_pos: 100.000 loss: 0.1918 (0.1885) acc1: 96.3542 (96.4652) acc5: 100.0000 (100.0000) time: 0.1709 data: 0.0778 max mem: 3531 +Test: [ 70/235] eta: 0:00:48 acc1_pos: 95.385 acc5_pos: 100.000 loss: 0.2157 (0.3519) acc1: 94.5312 (86.7114) acc5: 100.0000 (100.0000) acc1_neg: 15.625 acc5_neg: 100.000 time: 0.1890 data: 0.0924 max mem: 3531 +Test: [ 80/235] eta: 0:00:42 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.4913 (0.4935) acc1: 17.9688 (78.1925) acc5: 100.0000 (100.0000) acc1_neg: 14.323 acc5_neg: 100.000 time: 0.1549 data: 0.0563 max mem: 3531 +Test: [ 90/235] eta: 0:00:38 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.5084 (0.6039) acc1: 17.1875 (71.5430) acc5: 100.0000 (100.0000) acc1_neg: 17.188 acc5_neg: 100.000 time: 0.1799 data: 0.0806 max mem: 3531 +Test: [100/235] eta: 0:00:34 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.5084 (0.6903) acc1: 17.4479 (66.3315) acc5: 100.0000 (100.0000) acc1_neg: 17.188 acc5_neg: 100.000 time: 0.1849 data: 0.0883 max mem: 3531 +Test: [110/235] eta: 0:00:30 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.4770 (0.7617) acc1: 17.9688 (62.0237) acc5: 100.0000 (100.0000) acc1_neg: 20.052 acc5_neg: 100.000 time: 0.1549 data: 0.0570 max mem: 3531 +Test: [120/235] eta: 0:00:27 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.5262 (0.8425) acc1: 16.1458 (57.5672) acc5: 100.0000 (100.0000) acc1_neg: 5.208 acc5_neg: 100.000 time: 0.1699 data: 0.0724 max mem: 3531 +Test: [130/235] eta: 0:00:24 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.7542 (0.9113) acc1: 7.2917 (53.7651) acc5: 100.0000 (100.0000) acc1_neg: 7.031 acc5_neg: 100.000 time: 0.1701 data: 0.0766 max mem: 3531 +Test: [140/235] eta: 0:00:22 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.7605 (0.9730) acc1: 7.8125 (50.4987) acc5: 100.0000 (100.0000) acc1_neg: 5.469 acc5_neg: 100.000 time: 0.1849 data: 0.0891 max mem: 3531 +Test: [150/235] eta: 0:00:19 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.7700 (1.0244) acc1: 7.8125 (47.6494) acc5: 100.0000 (100.0000) acc1_neg: 7.292 acc5_neg: 100.000 time: 0.1848 data: 0.0863 max mem: 3531 +Test: [160/235] eta: 0:00:16 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.7497 (1.0696) acc1: 7.2917 (45.1766) acc5: 100.0000 (100.0000) acc1_neg: 9.115 acc5_neg: 100.000 time: 0.1399 data: 0.0414 max mem: 3531 +Test: [170/235] eta: 0:00:14 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.7113 (1.0809) acc1: 9.1146 (44.1109) acc5: 100.0000 (100.0000) acc1_neg: 27.083 acc5_neg: 100.000 time: 0.1254 data: 0.0296 max mem: 3531 +Test: [180/235] eta: 0:00:11 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.2126 (1.0876) acc1: 28.6458 (43.3227) acc5: 100.0000 (100.0000) acc1_neg: 30.990 acc5_neg: 100.000 time: 0.1603 data: 0.0670 max mem: 3531 +Test: [190/235] eta: 0:00:09 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.1897 (1.0927) acc1: 29.4271 (42.6129) acc5: 100.0000 (100.0000) acc1_neg: 30.469 acc5_neg: 100.000 time: 0.1794 data: 0.0841 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.2013 (1.0987) acc1: 29.4271 (41.9012) acc5: 100.0000 (100.0000) acc1_neg: 28.385 acc5_neg: 100.000 time: 0.1795 data: 0.0845 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.2141 (1.1029) acc1: 28.3854 (41.3248) acc5: 100.0000 (100.0000) acc1_neg: 27.865 acc5_neg: 100.000 time: 0.1749 data: 0.0807 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.1929 (1.1070) acc1: 28.1250 (40.7711) acc5: 100.0000 (100.0000) acc1_neg: 27.344 acc5_neg: 100.000 time: 0.1699 data: 0.0740 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.1936 (1.1107) acc1: 30.2083 (40.3319) acc5: 100.0000 (100.0000) acc1_neg: 27.604 acc5_neg: 100.000 time: 0.1749 data: 0.0813 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.1938 (1.1127) acc1: 30.2083 (40.1896) acc5: 100.0000 (100.0000) acc1_neg: 30.435 acc5_neg: 100.000 time: 0.1358 data: 0.0445 max mem: 3532 +Test: Total time: 0:00:48 (0.2049 s / it) +* Overall Acc@1 40.190 Acc@5 100.000 + target == real Acc@1 96.477 Acc@5 100.000 + target == fake Acc@1 19.832 Acc@5 100.000 + loss 1.113 + PR-AUC 0.8816 ROC-AUC 0.7424 EER 32.67% (thr=0.1549) +Accuracy of the network on the 89879 test images: 40.2% +[rank0]:[W726 03:40:46.032957384 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:41:05.786000 1171222 torch/distributed/run.py:766] +W0726 03:41:05.786000 1171222 torch/distributed/run.py:766] ***************************************** +W0726 03:41:05.786000 1171222 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:41:05.786000 1171222 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:41:33.217099863 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:41:33.328014036 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:41:33.632123459 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:41:33.788446235 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:41:33.833665071 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:41:33.877134170 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +| distributed init (rank 7): env:// +NCCL version 2.26.2+cuda12.2 +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:41:34.029195660 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W726 03:41:34.029225049 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n22_seed5_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed5_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:23:58 acc1_pos: 43.229 acc5_pos: 100.000 loss: 0.7646 (0.7646) acc1: 43.2292 (43.2292) acc5: 100.0000 (100.0000) time: 5.3469 data: 2.9564 max mem: 3531 +Test: [ 10/269] eta: 0:02:27 acc1_pos: 63.802 acc5_pos: 100.000 loss: 0.6128 (0.6157) acc1: 66.6667 (66.0985) acc5: 100.0000 (100.0000) time: 0.5684 data: 0.2691 max mem: 3531 +Test: [ 20/269] eta: 0:01:34 acc1_pos: 70.573 acc5_pos: 100.000 loss: 0.5964 (0.6023) acc1: 69.2708 (68.9236) acc5: 100.0000 (100.0000) time: 0.1299 data: 0.0380 max mem: 3531 +Test: [ 30/269] eta: 0:01:11 acc1_pos: 76.823 acc5_pos: 100.000 loss: 0.5831 (0.6036) acc1: 71.8750 (68.9516) acc5: 100.0000 (100.0000) time: 0.1495 data: 0.0553 max mem: 3531 +Test: [ 40/269] eta: 0:01:05 acc1_pos: 85.345 acc5_pos: 100.000 loss: 0.5667 (0.5979) acc1: 73.6979 (70.2490) acc5: 100.0000 (100.0000) acc1_neg: 75.521 acc5_neg: 100.000 time: 0.1899 data: 0.0951 max mem: 3531 +Test: [ 50/269] eta: 0:00:59 acc1_pos: 85.345 acc5_pos: 100.000 loss: 0.6383 (0.6273) acc1: 61.1979 (66.2377) acc5: 100.0000 (100.0000) acc1_neg: 26.042 acc5_neg: 100.000 time: 0.2349 data: 0.1409 max mem: 3531 +Test: [ 60/269] eta: 0:00:53 acc1_pos: 85.345 acc5_pos: 100.000 loss: 0.7732 (0.6572) acc1: 41.6667 (61.7529) acc5: 100.0000 (100.0000) acc1_neg: 34.115 acc5_neg: 100.000 time: 0.1949 data: 0.0999 max mem: 3531 +Test: [ 70/269] eta: 0:00:48 acc1_pos: 85.345 acc5_pos: 100.000 loss: 0.7732 (0.6712) acc1: 41.4062 (59.5401) acc5: 100.0000 (100.0000) acc1_neg: 42.708 acc5_neg: 100.000 time: 0.1649 data: 0.0664 max mem: 3531 +Test: [ 80/269] eta: 0:00:45 acc1_pos: 85.345 acc5_pos: 100.000 loss: 0.6992 (0.6733) acc1: 51.8229 (59.0696) acc5: 100.0000 (100.0000) acc1_neg: 70.312 acc5_neg: 100.000 time: 0.1963 data: 0.1004 max mem: 3531 +Test: [ 90/269] eta: 0:00:42 acc1_pos: 85.345 acc5_pos: 100.000 loss: 0.7244 (0.6825) acc1: 51.5625 (57.8840) acc5: 100.0000 (100.0000) acc1_neg: 50.781 acc5_neg: 100.000 time: 0.2098 data: 0.1183 max mem: 3531 +Test: [100/269] eta: 0:00:39 acc1_pos: 85.345 acc5_pos: 100.000 loss: 0.7764 (0.6962) acc1: 39.5833 (55.6647) acc5: 100.0000 (100.0000) acc1_neg: 21.094 acc5_neg: 100.000 time: 0.2035 data: 0.1097 max mem: 3531 +Test: [110/269] eta: 0:00:36 acc1_pos: 85.345 acc5_pos: 100.000 loss: 0.8250 (0.7082) acc1: 34.3750 (53.7725) acc5: 100.0000 (100.0000) acc1_neg: 29.948 acc5_neg: 100.000 time: 0.2135 data: 0.1179 max mem: 3531 +Test: [120/269] eta: 0:00:34 acc1_pos: 85.345 acc5_pos: 100.000 loss: 0.8125 (0.7106) acc1: 37.5000 (53.4780) acc5: 100.0000 (100.0000) acc1_neg: 37.500 acc5_neg: 100.000 time: 0.2051 data: 0.1114 max mem: 3531 +Test: [130/269] eta: 0:00:31 acc1_pos: 85.345 acc5_pos: 100.000 loss: 0.7022 (0.7081) acc1: 53.6458 (53.7989) acc5: 100.0000 (100.0000) acc1_neg: 66.927 acc5_neg: 100.000 time: 0.1913 data: 0.0976 max mem: 3531 +Test: [140/269] eta: 0:00:28 acc1_pos: 85.345 acc5_pos: 100.000 loss: 0.7026 (0.7094) acc1: 52.6042 (53.5258) acc5: 100.0000 (100.0000) acc1_neg: 42.188 acc5_neg: 100.000 time: 0.1805 data: 0.0848 max mem: 3531 +Test: [150/269] eta: 0:00:26 acc1_pos: 85.345 acc5_pos: 100.000 loss: 0.7026 (0.7069) acc1: 50.2604 (53.7786) acc5: 100.0000 (100.0000) acc1_neg: 71.094 acc5_neg: 100.000 time: 0.1975 data: 0.1022 max mem: 3531 +Test: [160/269] eta: 0:00:23 acc1_pos: 85.345 acc5_pos: 100.000 loss: 0.6404 (0.7029) acc1: 58.5938 (54.2265) acc5: 100.0000 (100.0000) acc1_neg: 46.875 acc5_neg: 100.000 time: 0.1858 data: 0.0913 max mem: 3531 +Test: [170/269] eta: 0:00:21 acc1_pos: 85.345 acc5_pos: 100.000 loss: 0.7103 (0.7060) acc1: 49.7396 (53.8149) acc5: 100.0000 (100.0000) acc1_neg: 49.740 acc5_neg: 100.000 time: 0.1857 data: 0.0929 max mem: 3531 +Test: [180/269] eta: 0:00:19 acc1_pos: 85.345 acc5_pos: 100.000 loss: 0.7069 (0.7035) acc1: 51.5625 (54.1321) acc5: 100.0000 (100.0000) acc1_neg: 40.625 acc5_neg: 100.000 time: 0.2004 data: 0.1068 max mem: 3531 +Test: [190/269] eta: 0:00:16 acc1_pos: 85.345 acc5_pos: 100.000 loss: 0.6712 (0.7062) acc1: 59.3750 (53.7604) acc5: 100.0000 (100.0000) acc1_neg: 66.406 acc5_neg: 100.000 time: 0.1841 data: 0.0911 max mem: 3531 +Test: [200/269] eta: 0:00:14 acc1_pos: 85.345 acc5_pos: 100.000 loss: 0.7061 (0.7076) acc1: 46.8750 (53.6134) acc5: 100.0000 (100.0000) acc1_neg: 36.198 acc5_neg: 100.000 time: 0.1877 data: 0.0941 max mem: 3531 +Test: [210/269] eta: 0:00:12 acc1_pos: 85.345 acc5_pos: 100.000 loss: 0.6712 (0.7044) acc1: 54.1667 (54.0173) acc5: 100.0000 (100.0000) acc1_neg: 66.406 acc5_neg: 100.000 time: 0.1933 data: 0.0992 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 85.345 acc5_pos: 100.000 loss: 0.6013 (0.6983) acc1: 65.1042 (54.7146) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1908 data: 0.0976 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 85.345 acc5_pos: 100.000 loss: 0.5498 (0.6904) acc1: 71.6146 (55.5928) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.1914 data: 0.0984 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 85.345 acc5_pos: 100.000 loss: 0.5823 (0.6935) acc1: 58.0729 (55.1943) acc5: 100.0000 (100.0000) acc1_neg: 50.521 acc5_neg: 100.000 time: 0.1908 data: 0.1004 max mem: 3532 +Test: [250/269] eta: 0:00:03 acc1_pos: 85.345 acc5_pos: 100.000 loss: 0.7919 (0.6979) acc1: 42.4479 (54.5443) acc5: 100.0000 (100.0000) acc1_neg: 15.365 acc5_neg: 100.000 time: 0.1759 data: 0.0849 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 85.345 acc5_pos: 100.000 loss: 0.8814 (0.7059) acc1: 28.6458 (53.4343) acc5: 100.0000 (100.0000) acc1_neg: 40.365 acc5_neg: 100.000 time: 0.1474 data: 0.0564 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 85.345 acc5_pos: 100.000 loss: 0.8402 (0.7090) acc1: 29.4271 (52.8661) acc5: 100.0000 (100.0000) acc1_neg: 34.383 acc5_neg: 100.000 time: 0.1325 data: 0.0379 max mem: 3532 +Test: Total time: 0:00:54 (0.2031 s / it) +* Overall Acc@1 52.866 Acc@5 100.000 + target == real Acc@1 69.373 Acc@5 100.000 + target == fake Acc@1 50.613 Acc@5 100.000 + loss 0.709 + PR-AUC 0.9256 ROC-AUC 0.6441 EER 40.05% (thr=0.4592) +Accuracy of the network on the 103293 test images: 52.9% +[rank0]:[W726 03:42:38.910999019 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T03:42:45Z] eval_done tag=fixed500k_fake_n22_seed5_resampled +[2026-07-26T03:42:45Z] eval_start tag=fixed500k_real_n11_seed6_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n11_classseed6_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T03:42:45Z +hostname=n116-041-188 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n11_classseed6_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n11_seed6_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 03:42:58.006000 1171798 torch/distributed/run.py:766] +W0726 03:42:58.006000 1171798 torch/distributed/run.py:766] ***************************************** +W0726 03:42:58.006000 1171798 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:42:58.006000 1171798 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:43:26.097782860 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:43:26.322623166 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:43:26.657585277 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:43:26.889077058 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:43:26.898087844 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:43:27.912355929 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:43:27.953052436 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:43:27.064934895 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n11_seed6_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n11_classseed6_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:22:07 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.1469 (2.1469) acc1: 0.0000 (0.0000) acc5: 100.0000 (100.0000) time: 5.6734 data: 4.0798 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/234] eta: 0:02:13 acc1_pos: 5.729 acc5_pos: 100.000 loss: 1.7605 (1.7267) acc1: 5.4688 (9.2093) acc5: 100.0000 (100.0000) time: 0.5977 data: 0.3712 max mem: 3531 +Test: [ 20/234] eta: 0:01:18 acc1_pos: 84.896 acc5_pos: 100.000 loss: 1.5748 (1.4945) acc1: 8.0729 (23.1895) acc5: 100.0000 (100.0000) time: 0.1001 data: 0.0075 max mem: 3531 +Test: [ 30/234] eta: 0:00:56 acc1_pos: 10.677 acc5_pos: 100.000 loss: 0.3627 (1.1955) acc1: 83.0729 (39.3481) acc5: 100.0000 (100.0000) time: 0.1030 data: 0.0076 max mem: 3531 +Test: [ 40/234] eta: 0:00:51 acc1_pos: 65.365 acc5_pos: 100.000 loss: 0.6483 (1.1924) acc1: 65.3646 (37.7668) acc5: 100.0000 (100.0000) time: 0.1629 data: 0.0701 max mem: 3531 +Test: [ 50/234] eta: 0:00:48 acc1_pos: 69.271 acc5_pos: 100.000 loss: 0.6556 (1.0818) acc1: 65.3646 (43.8470) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1431 max mem: 3531 +Test: [ 60/234] eta: 0:00:43 acc1_pos: 60.690 acc5_pos: 100.000 loss: 0.3480 (0.9245) acc1: 75.0000 (52.8048) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2199 data: 0.1189 max mem: 3531 +Test: [ 70/234] eta: 0:00:39 acc1_pos: 60.690 acc5_pos: 100.000 loss: 0.0989 (0.8349) acc1: 99.7396 (57.8125) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1859 data: 0.0911 max mem: 3531 +Test: [ 80/234] eta: 0:00:36 acc1_pos: 60.690 acc5_pos: 100.000 loss: 0.3501 (0.7660) acc1: 85.1562 (61.6770) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0958 max mem: 3531 +Test: [ 90/234] eta: 0:00:33 acc1_pos: 60.690 acc5_pos: 100.000 loss: 0.1825 (0.6982) acc1: 94.5312 (65.5477) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1943 data: 0.0998 max mem: 3531 +Test: [100/234] eta: 0:00:30 acc1_pos: 60.690 acc5_pos: 100.000 loss: 0.1882 (0.6578) acc1: 94.5312 (67.8115) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1921 data: 0.1006 max mem: 3531 +Test: [110/234] eta: 0:00:28 acc1_pos: 60.690 acc5_pos: 100.000 loss: 0.2595 (0.6180) acc1: 90.1042 (70.1201) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.2018 data: 0.1109 max mem: 3531 +Test: [120/234] eta: 0:00:25 acc1_pos: 60.690 acc5_pos: 100.000 loss: 0.2342 (0.5871) acc1: 91.9271 (71.9525) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2176 data: 0.1247 max mem: 3531 +Test: [130/234] eta: 0:00:23 acc1_pos: 60.690 acc5_pos: 100.000 loss: 0.2054 (0.5597) acc1: 96.0938 (73.6224) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2086 data: 0.1134 max mem: 3531 +Test: [140/234] eta: 0:00:20 acc1_pos: 60.690 acc5_pos: 100.000 loss: 0.2183 (0.5519) acc1: 93.4896 (74.0525) acc5: 100.0000 (100.0000) acc1_neg: 68.490 acc5_neg: 100.000 time: 0.1985 data: 0.1017 max mem: 3531 +Test: [150/234] eta: 0:00:18 acc1_pos: 60.690 acc5_pos: 100.000 loss: 0.2715 (0.5360) acc1: 89.3229 (74.9500) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1963 data: 0.1016 max mem: 3531 +Test: [160/234] eta: 0:00:16 acc1_pos: 60.690 acc5_pos: 100.000 loss: 0.2497 (0.5152) acc1: 92.7083 (76.2228) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1868 data: 0.0949 max mem: 3531 +Test: [170/234] eta: 0:00:13 acc1_pos: 60.690 acc5_pos: 100.000 loss: 0.2421 (0.4989) acc1: 93.7500 (77.2524) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1939 data: 0.0996 max mem: 3531 +Test: [180/234] eta: 0:00:11 acc1_pos: 60.690 acc5_pos: 100.000 loss: 0.2474 (0.4908) acc1: 91.9271 (77.6632) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.1910 data: 0.0976 max mem: 3531 +Test: [190/234] eta: 0:00:09 acc1_pos: 60.690 acc5_pos: 100.000 loss: 0.2958 (0.4812) acc1: 91.1458 (78.3377) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1896 data: 0.0979 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 60.690 acc5_pos: 100.000 loss: 0.2914 (0.4658) acc1: 91.4062 (79.2444) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1898 data: 0.0977 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 60.690 acc5_pos: 100.000 loss: 0.0973 (0.4491) acc1: 99.7396 (80.2096) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1793 data: 0.0882 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 60.690 acc5_pos: 100.000 loss: 0.1460 (0.4368) acc1: 98.9583 (80.9696) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.1719 data: 0.0814 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 60.690 acc5_pos: 100.000 loss: 0.3531 (0.4522) acc1: 55.4688 (79.7743) acc5: 100.0000 (100.0000) acc1_neg: 50.260 acc5_neg: 100.000 time: 0.1515 data: 0.0604 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 60.690 acc5_pos: 100.000 loss: 0.7685 (0.4564) acc1: 55.4688 (79.4927) acc5: 100.0000 (100.0000) acc1_neg: 56.420 acc5_neg: 100.000 time: 0.1464 data: 0.0543 max mem: 3532 +Test: Total time: 0:00:48 (0.2053 s / it) +* Overall Acc@1 79.493 Acc@5 100.000 + target == real Acc@1 43.971 Acc@5 100.000 + target == fake Acc@1 89.504 Acc@5 100.000 + loss 0.456 + PR-AUC 0.9151 ROC-AUC 0.7781 EER 29.88% (thr=0.8145) +Accuracy of the network on the 89729 test images: 79.5% +[rank0]:[W726 03:44:25.736581548 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:44:45.052000 1172346 torch/distributed/run.py:766] +W0726 03:44:45.052000 1172346 torch/distributed/run.py:766] ***************************************** +W0726 03:44:45.052000 1172346 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:44:45.052000 1172346 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:45:13.033649209 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:45:13.199619125 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:45:13.338445071 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:45:13.404840631 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:45:13.476889754 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:45:13.509092403 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:45:13.510323452 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:45:13.520891035 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n11_seed6_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n11_classseed6_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:39:26 acc1_pos: 7.552 acc5_pos: 100.000 loss: 1.7129 (1.7129) acc1: 7.5521 (7.5521) acc5: 100.0000 (100.0000) time: 7.0240 data: 5.3184 max mem: 3531 +Test: [ 10/2387] eta: 0:28:32 acc1_pos: 1.302 acc5_pos: 100.000 loss: 1.8995 (1.9168) acc1: 2.0833 (3.6932) acc5: 100.0000 (100.0000) time: 0.7206 data: 0.4840 max mem: 3531 +Test: [ 20/2387] eta: 0:16:43 acc1_pos: 0.781 acc5_pos: 100.000 loss: 1.9493 (1.9163) acc1: 2.6042 (3.7326) acc5: 100.0000 (100.0000) time: 0.0938 data: 0.0005 max mem: 3531 +Test: [ 30/2387] eta: 0:15:06 acc1_pos: 0.521 acc5_pos: 100.000 loss: 1.9794 (1.9496) acc1: 2.0833 (3.2342) acc5: 100.0000 (100.0000) time: 0.1998 data: 0.1058 max mem: 3531 +Test: [ 40/2387] eta: 0:14:53 acc1_pos: 0.781 acc5_pos: 100.000 loss: 2.0280 (1.9599) acc1: 0.7812 (3.0805) acc5: 100.0000 (100.0000) time: 0.3358 data: 0.2419 max mem: 3531 +Test: [ 50/2387] eta: 0:13:58 acc1_pos: 0.521 acc5_pos: 100.000 loss: 1.8811 (1.9198) acc1: 2.8646 (3.7990) acc5: 100.0000 (100.0000) time: 0.3188 data: 0.2237 max mem: 3531 +Test: [ 60/2387] eta: 0:13:09 acc1_pos: 0.521 acc5_pos: 100.000 loss: 1.9954 (1.9492) acc1: 1.8229 (3.5391) acc5: 100.0000 (100.0000) time: 0.2540 data: 0.1612 max mem: 3531 +Test: [ 70/2387] eta: 0:12:27 acc1_pos: 8.333 acc5_pos: 100.000 loss: 2.1678 (1.9950) acc1: 0.7812 (3.3231) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1355 max mem: 3531 +Test: [ 80/2387] eta: 0:12:23 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.1944 (2.0103) acc1: 0.7812 (3.1186) acc5: 100.0000 (100.0000) time: 0.2699 data: 0.1719 max mem: 3531 +Test: [ 90/2387] eta: 0:11:56 acc1_pos: 2.083 acc5_pos: 100.000 loss: 2.1407 (2.0262) acc1: 1.0417 (3.0334) acc5: 100.0000 (100.0000) time: 0.2749 data: 0.1786 max mem: 3531 +Test: [ 100/2387] eta: 0:11:39 acc1_pos: 0.260 acc5_pos: 100.000 loss: 2.1407 (2.0403) acc1: 1.0417 (2.9136) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1441 max mem: 3531 +Test: [ 110/2387] eta: 0:11:25 acc1_pos: 5.469 acc5_pos: 100.000 loss: 2.1071 (2.0438) acc1: 1.5625 (2.9373) acc5: 100.0000 (100.0000) time: 0.2499 data: 0.1492 max mem: 3531 +Test: [ 120/2387] eta: 0:11:10 acc1_pos: 0.521 acc5_pos: 100.000 loss: 2.0946 (2.0502) acc1: 1.5625 (2.8775) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1464 max mem: 3531 +Test: [ 130/2387] eta: 0:10:59 acc1_pos: 0.781 acc5_pos: 100.000 loss: 2.1357 (2.0596) acc1: 0.5208 (2.8069) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1501 max mem: 3531 +Test: [ 140/2387] eta: 0:10:48 acc1_pos: 1.302 acc5_pos: 100.000 loss: 2.1023 (2.0631) acc1: 0.7812 (2.7778) acc5: 100.0000 (100.0000) time: 0.2459 data: 0.1518 max mem: 3531 +Test: [ 150/2387] eta: 0:10:34 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.1691 (2.0679) acc1: 1.3021 (2.8059) acc5: 100.0000 (100.0000) time: 0.2261 data: 0.1346 max mem: 3531 +Test: [ 160/2387] eta: 0:10:26 acc1_pos: 2.740 acc5_pos: 100.000 loss: 2.0882 (2.0171) acc1: 1.8229 (5.4138) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2289 data: 0.1351 max mem: 3531 +Test: [ 170/2387] eta: 0:10:10 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1197 (1.9060) acc1: 98.6979 (10.8964) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2137 data: 0.1165 max mem: 3531 +Test: [ 180/2387] eta: 0:09:57 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1166 (1.8072) acc1: 99.2188 (15.7674) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1849 data: 0.0851 max mem: 3531 +Test: [ 190/2387] eta: 0:09:46 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1155 (1.7186) acc1: 98.9583 (20.1243) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1949 data: 0.0945 max mem: 3531 +Test: [ 200/2387] eta: 0:09:35 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1148 (1.6387) acc1: 99.2188 (24.0672) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1949 data: 0.0948 max mem: 3531 +Test: [ 210/2387] eta: 0:09:25 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1169 (1.5667) acc1: 98.9583 (27.6042) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1892 data: 0.0916 max mem: 3532 +Test: [ 220/2387] eta: 0:09:15 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1187 (1.5011) acc1: 98.6979 (30.8376) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1892 data: 0.0947 max mem: 3532 +Test: [ 230/2387] eta: 0:09:07 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1180 (1.4412) acc1: 98.9583 (33.7922) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1949 data: 0.0991 max mem: 3532 +Test: [ 240/2387] eta: 0:09:00 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1086 (1.3855) acc1: 99.4792 (36.5199) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1999 data: 0.1043 max mem: 3532 +Test: [ 250/2387] eta: 0:08:52 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0927 (1.3334) acc1: 99.7396 (39.0366) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.1003 max mem: 3532 +Test: [ 260/2387] eta: 0:08:46 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0707 (1.2850) acc1: 100.0000 (41.3703) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2001 data: 0.1064 max mem: 3532 +Test: [ 270/2387] eta: 0:08:43 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0688 (1.2400) acc1: 100.0000 (43.5328) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2199 data: 0.1258 max mem: 3532 +Test: [ 280/2387] eta: 0:08:39 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0685 (1.1984) acc1: 100.0000 (45.5396) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2290 data: 0.1346 max mem: 3532 +Test: [ 290/2387] eta: 0:08:33 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0654 (1.1595) acc1: 100.0000 (47.4093) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2142 data: 0.1161 max mem: 3532 +Test: [ 300/2387] eta: 0:08:29 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0624 (1.1231) acc1: 100.0000 (49.1478) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2099 data: 0.1096 max mem: 3532 +Test: [ 310/2387] eta: 0:08:26 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0634 (1.0902) acc1: 100.0000 (50.7695) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2261 data: 0.1292 max mem: 3532 +Test: [ 320/2387] eta: 0:08:25 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1418 (1.0608) acc1: 99.4792 (52.2788) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2544 data: 0.1589 max mem: 3532 +Test: [ 330/2387] eta: 0:08:21 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1474 (1.0331) acc1: 99.4792 (53.7009) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2439 data: 0.1475 max mem: 3532 +Test: [ 340/2387] eta: 0:08:14 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1443 (1.0071) acc1: 99.4792 (55.0350) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1950 data: 0.0977 max mem: 3532 +Test: [ 350/2387] eta: 0:08:09 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1358 (0.9801) acc1: 99.4792 (56.3123) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1893 data: 0.0920 max mem: 3532 +Test: [ 360/2387] eta: 0:08:05 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0794 (0.9557) acc1: 100.0000 (57.5059) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2049 data: 0.1086 max mem: 3532 +Test: [ 370/2387] eta: 0:08:01 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1034 (0.9328) acc1: 99.2188 (58.6274) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2069 data: 0.1127 max mem: 3532 +Test: [ 380/2387] eta: 0:07:55 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1034 (0.9111) acc1: 99.2188 (59.6949) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1849 data: 0.0901 max mem: 3532 +Test: [ 390/2387] eta: 0:07:48 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1023 (0.8904) acc1: 99.4792 (60.7077) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1581 data: 0.0634 max mem: 3532 +Test: [ 400/2387] eta: 0:07:45 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1032 (0.8708) acc1: 99.4792 (61.6746) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1834 data: 0.0895 max mem: 3532 +Test: [ 410/2387] eta: 0:07:41 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1013 (0.8521) acc1: 99.7396 (62.5957) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2118 data: 0.1181 max mem: 3532 +Test: [ 420/2387] eta: 0:07:37 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1012 (0.8343) acc1: 99.4792 (63.4705) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1980 data: 0.1049 max mem: 3532 +Test: [ 430/2387] eta: 0:07:33 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1000 (0.8171) acc1: 99.4792 (64.3133) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1902 data: 0.0958 max mem: 3532 +Test: [ 440/2387] eta: 0:07:29 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0878 (0.8005) acc1: 100.0000 (65.1213) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2002 data: 0.1064 max mem: 3533 +Test: [ 450/2387] eta: 0:07:25 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0873 (0.7847) acc1: 100.0000 (65.8947) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1955 data: 0.1026 max mem: 3533 +Test: [ 460/2387] eta: 0:07:21 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0874 (0.7696) acc1: 100.0000 (66.6345) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0905 max mem: 3533 +Test: [ 470/2387] eta: 0:07:17 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0875 (0.7551) acc1: 100.0000 (67.3429) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1919 data: 0.0971 max mem: 3533 +Test: [ 480/2387] eta: 0:07:14 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0877 (0.7413) acc1: 100.0000 (68.0218) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1987 data: 0.1060 max mem: 3533 +Test: [ 490/2387] eta: 0:07:10 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0877 (0.7280) acc1: 100.0000 (68.6726) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1970 data: 0.1056 max mem: 3533 +Test: [ 500/2387] eta: 0:07:06 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0878 (0.7152) acc1: 100.0000 (69.2973) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1793 data: 0.0879 max mem: 3533 +Test: [ 510/2387] eta: 0:07:01 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0878 (0.7029) acc1: 100.0000 (69.8982) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1690 data: 0.0774 max mem: 3533 +Test: [ 520/2387] eta: 0:06:57 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0850 (0.6910) acc1: 100.0000 (70.4759) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1726 data: 0.0816 max mem: 3533 +Test: [ 530/2387] eta: 0:06:54 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0872 (0.6797) acc1: 100.0000 (71.0320) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1804 data: 0.0898 max mem: 3533 +Test: [ 540/2387] eta: 0:06:50 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0883 (0.6688) acc1: 100.0000 (71.5674) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1833 data: 0.0908 max mem: 3533 +Test: [ 550/2387] eta: 0:06:46 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0881 (0.6582) acc1: 100.0000 (72.0834) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1789 data: 0.0857 max mem: 3533 +Test: [ 560/2387] eta: 0:06:43 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0883 (0.6481) acc1: 100.0000 (72.5806) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1805 data: 0.0884 max mem: 3533 +Test: [ 570/2387] eta: 0:06:39 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0884 (0.6383) acc1: 100.0000 (73.0608) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1753 data: 0.0815 max mem: 3533 +Test: [ 580/2387] eta: 0:06:36 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0876 (0.6288) acc1: 100.0000 (73.5245) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1880 data: 0.0940 max mem: 3533 +Test: [ 590/2387] eta: 0:06:32 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0878 (0.6196) acc1: 100.0000 (73.9724) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1794 data: 0.0857 max mem: 3533 +Test: [ 600/2387] eta: 0:06:28 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0855 (0.6107) acc1: 100.0000 (74.4051) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1566 data: 0.0639 max mem: 3533 +Test: [ 610/2387] eta: 0:06:25 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0924 (0.6026) acc1: 100.0000 (74.8155) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1760 data: 0.0853 max mem: 3533 +Test: [ 620/2387] eta: 0:06:22 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1111 (0.5946) acc1: 99.4792 (75.2126) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1768 data: 0.0854 max mem: 3533 +Test: [ 630/2387] eta: 0:06:19 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1121 (0.5870) acc1: 99.4792 (75.5935) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1845 data: 0.0926 max mem: 3533 +Test: [ 640/2387] eta: 0:06:16 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1133 (0.5796) acc1: 99.4792 (75.9653) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1920 data: 0.0988 max mem: 3533 +Test: [ 650/2387] eta: 0:06:13 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1136 (0.5725) acc1: 99.2188 (76.3197) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1769 data: 0.0844 max mem: 3533 +Test: [ 660/2387] eta: 0:06:11 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1139 (0.5655) acc1: 99.2188 (76.6685) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1993 data: 0.1066 max mem: 3533 +Test: [ 670/2387] eta: 0:06:08 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1131 (0.5588) acc1: 99.4792 (77.0057) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1982 data: 0.1049 max mem: 3534 +Test: [ 680/2387] eta: 0:06:04 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1086 (0.5519) acc1: 99.4792 (77.3403) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1747 data: 0.0834 max mem: 3534 +Test: [ 690/2387] eta: 0:06:02 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1124 (0.5470) acc1: 99.4792 (77.5699) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1856 data: 0.0930 max mem: 3534 +Test: [ 700/2387] eta: 0:05:59 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.2601 (0.5430) acc1: 90.6250 (77.7517) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1937 data: 0.1002 max mem: 3534 +Test: [ 710/2387] eta: 0:05:57 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.2620 (0.5391) acc1: 90.6250 (77.9272) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1937 data: 0.1015 max mem: 3534 +Test: [ 720/2387] eta: 0:05:54 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.2568 (0.5351) acc1: 90.6250 (78.1019) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1989 data: 0.1065 max mem: 3534 +Test: [ 730/2387] eta: 0:05:52 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.2576 (0.5315) acc1: 90.3646 (78.2654) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.2092 data: 0.1156 max mem: 3534 +Test: [ 740/2387] eta: 0:05:50 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.2699 (0.5279) acc1: 89.5833 (78.4202) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.2070 data: 0.1133 max mem: 3534 +Test: [ 750/2387] eta: 0:05:47 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.2608 (0.5243) acc1: 89.8438 (78.5831) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1994 data: 0.1063 max mem: 3534 +Test: [ 760/2387] eta: 0:05:45 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.2572 (0.5205) acc1: 90.6250 (78.7553) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2010 data: 0.1095 max mem: 3534 +Test: [ 770/2387] eta: 0:05:42 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1760 (0.5156) acc1: 93.7500 (78.9870) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1912 data: 0.1004 max mem: 3534 +Test: [ 780/2387] eta: 0:05:40 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1100 (0.5106) acc1: 99.2188 (79.2147) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1893 data: 0.0977 max mem: 3534 +Test: [ 790/2387] eta: 0:05:37 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0905 (0.5057) acc1: 99.7396 (79.4340) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1943 data: 0.1027 max mem: 3534 +Test: [ 800/2387] eta: 0:05:34 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0847 (0.5005) acc1: 99.7396 (79.6885) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1768 data: 0.0852 max mem: 3534 +Test: [ 810/2387] eta: 0:05:32 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0857 (0.4955) acc1: 100.0000 (79.9341) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1806 data: 0.0868 max mem: 3534 +Test: [ 820/2387] eta: 0:05:29 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0857 (0.4905) acc1: 100.0000 (80.1766) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1875 data: 0.0921 max mem: 3534 +Test: [ 830/2387] eta: 0:05:25 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0836 (0.4858) acc1: 100.0000 (80.4011) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1498 data: 0.0530 max mem: 3534 +Test: [ 840/2387] eta: 0:05:23 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1282 (0.4820) acc1: 97.1354 (80.5929) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1465 data: 0.0518 max mem: 3534 +Test: [ 850/2387] eta: 0:05:20 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1771 (0.4786) acc1: 95.8333 (80.7668) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1716 data: 0.0778 max mem: 3534 +Test: [ 860/2387] eta: 0:05:17 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1597 (0.4748) acc1: 96.3542 (80.9615) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1633 data: 0.0692 max mem: 3534 +Test: [ 870/2387] eta: 0:05:14 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1449 (0.4713) acc1: 97.3958 (81.1349) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1568 data: 0.0649 max mem: 3534 +Test: [ 880/2387] eta: 0:05:11 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1624 (0.4678) acc1: 96.8750 (81.3150) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1629 data: 0.0680 max mem: 3534 +Test: [ 890/2387] eta: 0:05:08 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1168 (0.4636) acc1: 99.4792 (81.5189) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1699 data: 0.0732 max mem: 3534 +Test: [ 900/2387] eta: 0:05:07 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1241 (0.4657) acc1: 97.6562 (81.3023) acc5: 100.0000 (100.0000) acc1_neg: 52.604 acc5_neg: 100.000 time: 0.1983 data: 0.1047 max mem: 3535 +Test: [ 910/2387] eta: 0:05:04 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.7796 (0.4692) acc1: 52.8646 (80.9973) acc5: 100.0000 (100.0000) acc1_neg: 52.083 acc5_neg: 100.000 time: 0.2061 data: 0.1141 max mem: 3535 +Test: [ 920/2387] eta: 0:05:02 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.7903 (0.4727) acc1: 52.0833 (80.6828) acc5: 100.0000 (100.0000) acc1_neg: 54.167 acc5_neg: 100.000 time: 0.1984 data: 0.1068 max mem: 3535 +Test: [ 930/2387] eta: 0:05:00 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.7903 (0.4762) acc1: 52.8646 (80.3941) acc5: 100.0000 (100.0000) acc1_neg: 50.521 acc5_neg: 100.000 time: 0.1939 data: 0.1029 max mem: 3535 +Test: [ 940/2387] eta: 0:04:57 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.8015 (0.4795) acc1: 52.8646 (80.1082) acc5: 100.0000 (100.0000) acc1_neg: 51.302 acc5_neg: 100.000 time: 0.1612 data: 0.0704 max mem: 3535 +Test: [ 950/2387] eta: 0:04:54 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.7985 (0.4828) acc1: 52.6042 (79.8241) acc5: 100.0000 (100.0000) acc1_neg: 56.250 acc5_neg: 100.000 time: 0.1597 data: 0.0671 max mem: 3535 +Test: [ 960/2387] eta: 0:04:52 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.7731 (0.4858) acc1: 54.4271 (79.5569) acc5: 100.0000 (100.0000) acc1_neg: 55.729 acc5_neg: 100.000 time: 0.1848 data: 0.0913 max mem: 3535 +Test: [ 970/2387] eta: 0:04:49 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.7781 (0.4891) acc1: 53.3854 (79.2804) acc5: 100.0000 (100.0000) acc1_neg: 53.385 acc5_neg: 100.000 time: 0.1778 data: 0.0841 max mem: 3535 +Test: [ 980/2387] eta: 0:04:47 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.7762 (0.4877) acc1: 53.3854 (79.3066) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1586 data: 0.0645 max mem: 3535 +Test: [ 990/2387] eta: 0:04:45 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1139 (0.4838) acc1: 99.4792 (79.5091) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1952 data: 0.0970 max mem: 3535 +Test: [1000/2387] eta: 0:04:43 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0978 (0.4801) acc1: 99.7396 (79.6992) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2149 data: 0.1151 max mem: 3535 +Test: [1010/2387] eta: 0:04:40 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0947 (0.4764) acc1: 99.7396 (79.8907) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1603 data: 0.0656 max mem: 3535 +Test: [1020/2387] eta: 0:04:37 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0950 (0.4727) acc1: 99.7396 (80.0841) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1431 data: 0.0467 max mem: 3535 +Test: [1030/2387] eta: 0:04:35 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0981 (0.4693) acc1: 100.0000 (80.2654) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1612 data: 0.0647 max mem: 3535 +Test: [1040/2387] eta: 0:04:32 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0970 (0.4657) acc1: 100.0000 (80.4545) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1761 data: 0.0843 max mem: 3535 +Test: [1050/2387] eta: 0:04:30 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0931 (0.4621) acc1: 100.0000 (80.6387) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1858 data: 0.0931 max mem: 3535 +Test: [1060/2387] eta: 0:04:28 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0931 (0.4587) acc1: 100.0000 (80.8205) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1795 data: 0.0863 max mem: 3535 +Test: [1070/2387] eta: 0:04:26 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0891 (0.4553) acc1: 100.0000 (80.9974) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1926 data: 0.0998 max mem: 3535 +Test: [1080/2387] eta: 0:04:23 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0877 (0.4520) acc1: 100.0000 (81.1671) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1890 data: 0.0953 max mem: 3535 +Test: [1090/2387] eta: 0:04:21 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1112 (0.4494) acc1: 99.2188 (81.3030) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.1846 data: 0.0907 max mem: 3535 +Test: [1100/2387] eta: 0:04:19 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1318 (0.4467) acc1: 98.4375 (81.4397) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1949 data: 0.1022 max mem: 3535 +Test: [1110/2387] eta: 0:04:17 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1441 (0.4443) acc1: 98.4375 (81.5678) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1874 data: 0.0956 max mem: 3535 +Test: [1120/2387] eta: 0:04:15 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1399 (0.4415) acc1: 99.2188 (81.7251) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1858 data: 0.0933 max mem: 3536 +Test: [1130/2387] eta: 0:04:12 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1301 (0.4389) acc1: 99.4792 (81.8788) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1827 data: 0.0908 max mem: 3536 +Test: [1140/2387] eta: 0:04:10 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1067 (0.4359) acc1: 99.4792 (82.0306) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1792 data: 0.0874 max mem: 3536 +Test: [1150/2387] eta: 0:04:08 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1194 (0.4334) acc1: 99.2188 (82.1738) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1753 data: 0.0826 max mem: 3536 +Test: [1160/2387] eta: 0:04:06 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1385 (0.4308) acc1: 98.9583 (82.3119) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1815 data: 0.0889 max mem: 3536 +Test: [1170/2387] eta: 0:04:03 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1500 (0.4285) acc1: 97.9167 (82.4402) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1803 data: 0.0891 max mem: 3536 +Test: [1180/2387] eta: 0:04:01 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1500 (0.4264) acc1: 97.6562 (82.5503) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1690 data: 0.0772 max mem: 3536 +Test: [1190/2387] eta: 0:03:59 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1360 (0.4239) acc1: 97.6562 (82.6756) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1693 data: 0.0766 max mem: 3536 +Test: [1200/2387] eta: 0:03:57 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0924 (0.4212) acc1: 99.4792 (82.8173) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1746 data: 0.0831 max mem: 3536 +Test: [1210/2387] eta: 0:03:54 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0818 (0.4184) acc1: 99.7396 (82.9546) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1797 data: 0.0890 max mem: 3536 +Test: [1220/2387] eta: 0:03:52 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0820 (0.4157) acc1: 100.0000 (83.0919) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1794 data: 0.0888 max mem: 3536 +Test: [1230/2387] eta: 0:03:50 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0816 (0.4129) acc1: 100.0000 (83.2284) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1819 data: 0.0911 max mem: 3536 +Test: [1240/2387] eta: 0:03:48 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0793 (0.4102) acc1: 100.0000 (83.3619) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1819 data: 0.0886 max mem: 3536 +Test: [1250/2387] eta: 0:03:46 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0772 (0.4076) acc1: 100.0000 (83.4924) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1738 data: 0.0787 max mem: 3536 +Test: [1260/2387] eta: 0:03:43 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0754 (0.4049) acc1: 99.7396 (83.6204) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1678 data: 0.0740 max mem: 3536 +Test: [1270/2387] eta: 0:03:41 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0673 (0.4022) acc1: 100.0000 (83.7482) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1645 data: 0.0715 max mem: 3536 +Test: [1280/2387] eta: 0:03:39 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0718 (0.3998) acc1: 100.0000 (83.8747) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1707 data: 0.0784 max mem: 3536 +Test: [1290/2387] eta: 0:03:37 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0866 (0.3975) acc1: 100.0000 (83.9885) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1832 data: 0.0900 max mem: 3536 +Test: [1300/2387] eta: 0:03:35 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0898 (0.3951) acc1: 99.7396 (84.1054) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1755 data: 0.0825 max mem: 3536 +Test: [1310/2387] eta: 0:03:32 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0892 (0.3929) acc1: 99.7396 (84.2187) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1705 data: 0.0786 max mem: 3536 +Test: [1320/2387] eta: 0:03:30 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0906 (0.3907) acc1: 99.4792 (84.3285) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1797 data: 0.0869 max mem: 3536 +Test: [1330/2387] eta: 0:03:28 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0878 (0.3885) acc1: 99.4792 (84.4405) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1993 data: 0.1068 max mem: 3536 +Test: [1340/2387] eta: 0:03:27 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0878 (0.3865) acc1: 99.4792 (84.5459) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2130 data: 0.1188 max mem: 3536 +Test: [1350/2387] eta: 0:03:25 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0634 (0.3841) acc1: 100.0000 (84.6603) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.0984 max mem: 3537 +Test: [1360/2387] eta: 0:03:23 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0605 (0.3817) acc1: 100.0000 (84.7730) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0921 max mem: 3537 +Test: [1370/2387] eta: 0:03:20 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0619 (0.3794) acc1: 100.0000 (84.8841) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0915 max mem: 3537 +Test: [1380/2387] eta: 0:03:18 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0618 (0.3771) acc1: 100.0000 (84.9935) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0716 max mem: 3537 +Test: [1390/2387] eta: 0:03:16 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0551 (0.3748) acc1: 100.0000 (85.1014) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0718 max mem: 3537 +Test: [1400/2387] eta: 0:03:14 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0499 (0.3725) acc1: 100.0000 (85.2077) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0828 max mem: 3537 +Test: [1410/2387] eta: 0:03:12 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0508 (0.3704) acc1: 100.0000 (85.3109) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1599 data: 0.0646 max mem: 3537 +Test: [1420/2387] eta: 0:03:10 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0932 (0.3689) acc1: 100.0000 (85.3773) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1649 data: 0.0697 max mem: 3537 +Test: [1430/2387] eta: 0:03:08 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1045 (0.3671) acc1: 99.7396 (85.4765) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1949 data: 0.0982 max mem: 3537 +Test: [1440/2387] eta: 0:03:06 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0948 (0.3652) acc1: 100.0000 (85.5755) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.0994 max mem: 3537 +Test: [1450/2387] eta: 0:03:03 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0948 (0.3633) acc1: 100.0000 (85.6688) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1599 data: 0.0646 max mem: 3537 +Test: [1460/2387] eta: 0:03:01 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0961 (0.3615) acc1: 99.7396 (85.7603) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1499 data: 0.0561 max mem: 3537 +Test: [1470/2387] eta: 0:02:59 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0850 (0.3597) acc1: 100.0000 (85.8559) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1777 data: 0.0843 max mem: 3537 +Test: [1480/2387] eta: 0:02:57 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0835 (0.3578) acc1: 100.0000 (85.9463) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1971 data: 0.1039 max mem: 3537 +Test: [1490/2387] eta: 0:02:55 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0871 (0.3561) acc1: 99.4792 (86.0358) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1771 data: 0.0821 max mem: 3537 +Test: [1500/2387] eta: 0:02:53 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0913 (0.3544) acc1: 99.7396 (86.1250) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1383 data: 0.0428 max mem: 3537 +Test: [1510/2387] eta: 0:02:51 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0888 (0.3526) acc1: 99.7396 (86.2110) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1550 data: 0.0614 max mem: 3537 +Test: [1520/2387] eta: 0:02:49 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1192 (0.3515) acc1: 98.1771 (86.2750) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1805 data: 0.0851 max mem: 3537 +Test: [1530/2387] eta: 0:02:46 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1774 (0.3505) acc1: 95.3125 (86.3246) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.1699 data: 0.0761 max mem: 3537 +Test: [1540/2387] eta: 0:02:44 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1751 (0.3493) acc1: 95.3125 (86.3897) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1669 data: 0.0758 max mem: 3537 +Test: [1550/2387] eta: 0:02:42 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1686 (0.3483) acc1: 95.5729 (86.4467) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1748 data: 0.0819 max mem: 3537 +Test: [1560/2387] eta: 0:02:40 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1570 (0.3471) acc1: 96.0938 (86.5087) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1700 data: 0.0772 max mem: 3537 +Test: [1570/2387] eta: 0:02:38 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1138 (0.3455) acc1: 98.9583 (86.5911) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1665 data: 0.0761 max mem: 3537 +Test: [1580/2387] eta: 0:02:36 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0945 (0.3440) acc1: 99.7396 (86.6716) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1726 data: 0.0812 max mem: 3538 +Test: [1590/2387] eta: 0:02:34 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0976 (0.3427) acc1: 99.7396 (86.7305) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1747 data: 0.0812 max mem: 3538 +Test: [1600/2387] eta: 0:02:32 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0965 (0.3414) acc1: 99.4792 (86.7942) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1739 data: 0.0774 max mem: 3538 +Test: [1610/2387] eta: 0:02:30 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0935 (0.3399) acc1: 99.7396 (86.8739) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1706 data: 0.0752 max mem: 3538 +Test: [1620/2387] eta: 0:02:28 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0974 (0.3384) acc1: 99.7396 (86.9509) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1800 data: 0.0876 max mem: 3538 +Test: [1630/2387] eta: 0:02:26 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1000 (0.3370) acc1: 99.4792 (87.0248) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1928 data: 0.0986 max mem: 3538 +Test: [1640/2387] eta: 0:02:24 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0928 (0.3355) acc1: 99.4792 (87.0977) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1812 data: 0.0871 max mem: 3538 +Test: [1650/2387] eta: 0:02:22 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0853 (0.3341) acc1: 99.4792 (87.1699) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1728 data: 0.0805 max mem: 3538 +Test: [1660/2387] eta: 0:02:20 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0949 (0.3327) acc1: 99.2188 (87.2399) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1787 data: 0.0862 max mem: 3538 +Test: [1670/2387] eta: 0:02:18 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0851 (0.3312) acc1: 99.4792 (87.3105) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1794 data: 0.0879 max mem: 3538 +Test: [1680/2387] eta: 0:02:16 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0930 (0.3299) acc1: 99.4792 (87.3770) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1820 data: 0.0907 max mem: 3538 +Test: [1690/2387] eta: 0:02:14 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0964 (0.3286) acc1: 99.4792 (87.4410) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1897 data: 0.0983 max mem: 3538 +Test: [1700/2387] eta: 0:02:12 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0907 (0.3272) acc1: 99.2188 (87.5089) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1989 data: 0.1079 max mem: 3538 +Test: [1710/2387] eta: 0:02:10 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0831 (0.3257) acc1: 99.4792 (87.5805) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1800 data: 0.0892 max mem: 3538 +Test: [1720/2387] eta: 0:02:08 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0806 (0.3243) acc1: 99.7396 (87.6510) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1636 data: 0.0731 max mem: 3538 +Test: [1730/2387] eta: 0:02:06 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0750 (0.3229) acc1: 99.7396 (87.7190) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1740 data: 0.0833 max mem: 3538 +Test: [1740/2387] eta: 0:02:04 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0749 (0.3215) acc1: 100.0000 (87.7861) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1805 data: 0.0897 max mem: 3538 +Test: [1750/2387] eta: 0:02:02 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0705 (0.3201) acc1: 100.0000 (87.8538) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1890 data: 0.0976 max mem: 3538 +Test: [1760/2387] eta: 0:02:00 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0646 (0.3186) acc1: 100.0000 (87.9213) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1803 data: 0.0865 max mem: 3538 +Test: [1770/2387] eta: 0:01:58 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0663 (0.3173) acc1: 99.7396 (87.9848) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1746 data: 0.0784 max mem: 3538 +Test: [1780/2387] eta: 0:01:56 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0781 (0.3160) acc1: 99.7396 (88.0515) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1768 data: 0.0818 max mem: 3538 +Test: [1790/2387] eta: 0:01:54 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0748 (0.3146) acc1: 100.0000 (88.1165) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1947 data: 0.1017 max mem: 3538 +Test: [1800/2387] eta: 0:01:52 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0736 (0.3133) acc1: 99.7396 (88.1803) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1819 data: 0.0870 max mem: 3538 +Test: [1810/2387] eta: 0:01:50 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0736 (0.3119) acc1: 99.7396 (88.2434) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1702 data: 0.0742 max mem: 3539 +Test: [1820/2387] eta: 0:01:48 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0620 (0.3106) acc1: 100.0000 (88.3073) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1712 data: 0.0760 max mem: 3539 +Test: [1830/2387] eta: 0:01:46 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0621 (0.3093) acc1: 100.0000 (88.3710) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1570 data: 0.0629 max mem: 3539 +Test: [1840/2387] eta: 0:01:44 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0947 (0.3081) acc1: 100.0000 (88.4342) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1707 data: 0.0789 max mem: 3539 +Test: [1850/2387] eta: 0:01:42 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0957 (0.3070) acc1: 100.0000 (88.4965) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1675 data: 0.0768 max mem: 3539 +Test: [1860/2387] eta: 0:01:40 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0959 (0.3059) acc1: 100.0000 (88.5583) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1592 data: 0.0675 max mem: 3539 +Test: [1870/2387] eta: 0:01:38 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0960 (0.3047) acc1: 100.0000 (88.6192) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0765 max mem: 3539 +Test: [1880/2387] eta: 0:01:36 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0953 (0.3036) acc1: 100.0000 (88.6797) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1586 data: 0.0656 max mem: 3539 +Test: [1890/2387] eta: 0:01:35 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0948 (0.3025) acc1: 100.0000 (88.7396) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0867 max mem: 3539 +Test: [1900/2387] eta: 0:01:33 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0952 (0.3014) acc1: 100.0000 (88.7988) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2116 data: 0.1171 max mem: 3539 +Test: [1910/2387] eta: 0:01:31 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0941 (0.3003) acc1: 100.0000 (88.8567) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2073 data: 0.1136 max mem: 3539 +Test: [1920/2387] eta: 0:01:29 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0918 (0.2994) acc1: 100.0000 (88.9092) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1832 data: 0.0909 max mem: 3539 +Test: [1930/2387] eta: 0:01:27 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0983 (0.2983) acc1: 100.0000 (88.9645) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1647 data: 0.0735 max mem: 3539 +Test: [1940/2387] eta: 0:01:25 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0952 (0.2975) acc1: 100.0000 (88.9785) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1999 data: 0.1088 max mem: 3539 +Test: [1950/2387] eta: 0:01:23 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0915 (0.2965) acc1: 100.0000 (89.0343) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2188 data: 0.1259 max mem: 3539 +Test: [1960/2387] eta: 0:01:21 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0943 (0.2956) acc1: 100.0000 (89.0837) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2249 data: 0.1320 max mem: 3539 +Test: [1970/2387] eta: 0:01:19 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1166 (0.2948) acc1: 99.7396 (89.1283) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1788 data: 0.0830 max mem: 3539 +Test: [1980/2387] eta: 0:01:17 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1281 (0.2940) acc1: 99.2188 (89.1790) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1719 data: 0.0758 max mem: 3539 +Test: [1990/2387] eta: 0:01:15 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1362 (0.2932) acc1: 98.9583 (89.2277) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1910 data: 0.0994 max mem: 3539 +Test: [2000/2387] eta: 0:01:14 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1423 (0.2925) acc1: 98.6979 (89.2748) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1703 data: 0.0755 max mem: 3539 +Test: [2010/2387] eta: 0:01:12 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1440 (0.2917) acc1: 98.6979 (89.3206) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1937 data: 0.0988 max mem: 3539 +Test: [2020/2387] eta: 0:01:10 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1441 (0.2910) acc1: 98.4375 (89.3663) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1925 data: 0.1015 max mem: 3539 +Test: [2030/2387] eta: 0:01:08 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1451 (0.2903) acc1: 98.4375 (89.4127) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1399 data: 0.0451 max mem: 3540 +Test: [2040/2387] eta: 0:01:06 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1446 (0.2896) acc1: 98.9583 (89.4596) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1449 data: 0.0458 max mem: 3540 +Test: [2050/2387] eta: 0:01:04 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1382 (0.2888) acc1: 98.9583 (89.5054) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1799 data: 0.0847 max mem: 3540 +Test: [2060/2387] eta: 0:01:02 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1332 (0.2881) acc1: 98.6979 (89.5499) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1699 data: 0.0782 max mem: 3540 +Test: [2070/2387] eta: 0:01:00 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1289 (0.2873) acc1: 98.9583 (89.5977) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1762 data: 0.0839 max mem: 3540 +Test: [2080/2387] eta: 0:00:58 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1253 (0.2865) acc1: 99.2188 (89.6434) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1810 data: 0.0875 max mem: 3540 +Test: [2090/2387] eta: 0:00:56 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1267 (0.2858) acc1: 98.9583 (89.6884) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1786 data: 0.0843 max mem: 3540 +Test: [2100/2387] eta: 0:00:54 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1278 (0.2850) acc1: 98.9583 (89.7333) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1713 data: 0.0780 max mem: 3540 +Test: [2110/2387] eta: 0:00:52 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1280 (0.2843) acc1: 98.9583 (89.7778) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1730 data: 0.0812 max mem: 3540 +Test: [2120/2387] eta: 0:00:50 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1260 (0.2835) acc1: 99.2188 (89.8225) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1778 data: 0.0868 max mem: 3540 +Test: [2130/2387] eta: 0:00:48 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1217 (0.2828) acc1: 100.0000 (89.8552) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1830 data: 0.0915 max mem: 3540 +Test: [2140/2387] eta: 0:00:46 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1172 (0.2823) acc1: 99.4792 (89.8827) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1860 data: 0.0929 max mem: 3540 +Test: [2150/2387] eta: 0:00:45 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1163 (0.2815) acc1: 99.4792 (89.9278) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1811 data: 0.0871 max mem: 3540 +Test: [2160/2387] eta: 0:00:43 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1123 (0.2807) acc1: 99.4792 (89.9706) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1882 data: 0.0954 max mem: 3540 +Test: [2170/2387] eta: 0:00:41 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1097 (0.2799) acc1: 100.0000 (90.0140) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1836 data: 0.0883 max mem: 3540 +Test: [2180/2387] eta: 0:00:39 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1183 (0.2808) acc1: 99.4792 (89.9583) acc5: 100.0000 (100.0000) acc1_neg: 44.792 acc5_neg: 100.000 time: 0.1770 data: 0.0816 max mem: 3540 +Test: [2190/2387] eta: 0:00:37 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1064 (0.2805) acc1: 99.2188 (89.9683) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1837 data: 0.0886 max mem: 3540 +Test: [2200/2387] eta: 0:00:35 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0896 (0.2796) acc1: 99.7396 (90.0132) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1735 data: 0.0778 max mem: 3540 +Test: [2210/2387] eta: 0:00:33 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0841 (0.2788) acc1: 100.0000 (90.0571) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1664 data: 0.0747 max mem: 3540 +Test: [2220/2387] eta: 0:00:31 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0902 (0.2780) acc1: 99.7396 (90.0975) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1663 data: 0.0759 max mem: 3540 +Test: [2230/2387] eta: 0:00:29 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0902 (0.2771) acc1: 99.7396 (90.1397) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1610 data: 0.0705 max mem: 3540 +Test: [2240/2387] eta: 0:00:27 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0862 (0.2763) acc1: 99.7396 (90.1806) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1710 data: 0.0760 max mem: 3540 +Test: [2250/2387] eta: 0:00:25 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0831 (0.2754) acc1: 99.4792 (90.2210) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1691 data: 0.0716 max mem: 3540 +Test: [2260/2387] eta: 0:00:24 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0943 (0.2746) acc1: 99.7396 (90.2632) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1582 data: 0.0621 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0955 (0.2740) acc1: 100.0000 (90.2977) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1681 data: 0.0741 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0925 (0.2734) acc1: 100.0000 (90.3237) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1771 data: 0.0845 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0907 (0.2726) acc1: 100.0000 (90.3655) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0858 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0922 (0.2718) acc1: 100.0000 (90.4057) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1889 data: 0.0915 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0917 (0.2710) acc1: 100.0000 (90.4444) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1955 data: 0.1003 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0901 (0.2703) acc1: 100.0000 (90.4832) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1936 data: 0.1027 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0999 (0.2697) acc1: 100.0000 (90.5134) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1992 data: 0.1070 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0999 (0.2690) acc1: 100.0000 (90.5516) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1942 data: 0.1018 max mem: 3541 +Test: [2350/2387] eta: 0:00:07 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1035 (0.2686) acc1: 100.0000 (90.5711) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1944 data: 0.1002 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1004 (0.2679) acc1: 100.0000 (90.6075) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1796 data: 0.0862 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1004 (0.2673) acc1: 100.0000 (90.6454) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1502 data: 0.0600 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.1070 (0.2667) acc1: 100.0000 (90.6805) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1471 data: 0.0566 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 2.740 acc5_pos: 100.000 loss: 0.0921 (0.2662) acc1: 99.7396 (90.6988) acc5: 100.0000 (100.0000) acc1_neg: 97.674 acc5_neg: 100.000 time: 0.1294 data: 0.0417 max mem: 3541 +Test: Total time: 0:07:30 (0.1888 s / it) +* Overall Acc@1 90.699 Acc@5 100.000 + target == real Acc@1 2.774 Acc@5 100.000 + target == fake Acc@1 96.873 Acc@5 100.000 + loss 0.266 + PR-AUC 0.9703 ROC-AUC 0.7024 EER 34.26% (thr=0.9062) +Accuracy of the network on the 916267 test images: 90.7% +[rank0]:[W726 03:53:11.929466492 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:53:30.877000 1172976 torch/distributed/run.py:766] +W0726 03:53:30.877000 1172976 torch/distributed/run.py:766] ***************************************** +W0726 03:53:30.877000 1172976 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:53:30.877000 1172976 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:53:58.514191588 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:53:58.758354311 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:53:59.917259275 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:53:59.409516846 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:53:59.423582532 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank1]:[W726 03:53:59.423729992 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +[rank0]:[W726 03:53:59.445216166 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 03:53:59.452358998 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n11_seed6_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n11_classseed6_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:09:10 acc1_pos: 7.552 acc5_pos: 100.000 loss: 1.7129 (1.7129) acc1: 7.5521 (7.5521) acc5: 100.0000 (100.0000) time: 8.0286 data: 6.4250 max mem: 3531 +Test: [ 10/517] eta: 0:06:54 acc1_pos: 1.302 acc5_pos: 100.000 loss: 1.8995 (1.9168) acc1: 2.0833 (3.6932) acc5: 100.0000 (100.0000) time: 0.8168 data: 0.5845 max mem: 3531 +Test: [ 20/517] eta: 0:03:56 acc1_pos: 0.781 acc5_pos: 100.000 loss: 1.9493 (1.9163) acc1: 2.6042 (3.7326) acc5: 100.0000 (100.0000) time: 0.0977 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:03:10 acc1_pos: 0.521 acc5_pos: 100.000 loss: 1.9794 (1.9496) acc1: 2.0833 (3.2342) acc5: 100.0000 (100.0000) time: 0.1571 data: 0.0621 max mem: 3531 +Test: [ 40/517] eta: 0:02:53 acc1_pos: 0.781 acc5_pos: 100.000 loss: 2.0280 (1.9599) acc1: 0.7812 (3.0805) acc5: 100.0000 (100.0000) time: 0.2453 data: 0.1550 max mem: 3531 +Test: [ 50/517] eta: 0:02:34 acc1_pos: 1.376 acc5_pos: 100.000 loss: 1.7961 (1.8353) acc1: 3.9062 (8.3895) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2388 data: 0.1470 max mem: 3531 +Test: [ 60/517] eta: 0:02:25 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.2021 (1.5658) acc1: 97.6562 (23.1728) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2256 data: 0.1296 max mem: 3531 +Test: [ 70/517] eta: 0:02:17 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1815 (1.3686) acc1: 98.9583 (33.7551) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2495 data: 0.1539 max mem: 3531 +Test: [ 80/517] eta: 0:02:11 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1721 (1.2220) acc1: 99.4792 (41.7117) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2447 data: 0.1527 max mem: 3531 +Test: [ 90/517] eta: 0:02:06 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1456 (1.1022) acc1: 99.7396 (48.0855) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2492 data: 0.1577 max mem: 3531 +Test: [100/517] eta: 0:02:01 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1319 (1.0080) acc1: 100.0000 (53.0631) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.2541 data: 0.1625 max mem: 3531 +Test: [110/517] eta: 0:01:56 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1777 (0.9379) acc1: 92.4479 (56.3345) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2499 data: 0.1557 max mem: 3531 +Test: [120/517] eta: 0:01:52 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1773 (0.8757) acc1: 95.0521 (59.7086) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2399 data: 0.1431 max mem: 3531 +Test: [130/517] eta: 0:01:48 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1609 (0.8209) acc1: 99.7396 (62.7584) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2499 data: 0.1520 max mem: 3531 +Test: [140/517] eta: 0:01:45 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1414 (0.7725) acc1: 99.7396 (65.3923) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1580 max mem: 3531 +Test: [150/517] eta: 0:01:41 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1402 (0.7308) acc1: 100.0000 (67.6773) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2407 data: 0.1458 max mem: 3531 +Test: [160/517] eta: 0:01:37 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1428 (0.6943) acc1: 100.0000 (69.6849) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2402 data: 0.1477 max mem: 3531 +Test: [170/517] eta: 0:01:34 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1400 (0.6618) acc1: 100.0000 (71.4516) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2375 data: 0.1459 max mem: 3531 +Test: [180/517] eta: 0:01:31 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1374 (0.6331) acc1: 99.7396 (72.9757) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2430 data: 0.1470 max mem: 3531 +Test: [190/517] eta: 0:01:28 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1161 (0.6056) acc1: 100.0000 (74.3905) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1579 max mem: 3531 +Test: [200/517] eta: 0:01:25 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1038 (0.5807) acc1: 100.0000 (75.6646) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1601 max mem: 3531 +Test: [210/517] eta: 0:01:22 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1173 (0.5630) acc1: 99.2188 (76.6612) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2499 data: 0.1549 max mem: 3532 +Test: [220/517] eta: 0:01:19 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.2014 (0.5468) acc1: 97.6562 (77.6065) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2399 data: 0.1463 max mem: 3532 +Test: [230/517] eta: 0:01:16 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1912 (0.5307) acc1: 98.9583 (78.5410) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2453 data: 0.1540 max mem: 3532 +Test: [240/517] eta: 0:01:13 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1649 (0.5153) acc1: 99.7396 (79.4303) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2500 data: 0.1576 max mem: 3532 +Test: [250/517] eta: 0:01:10 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1439 (0.5004) acc1: 100.0000 (80.2478) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2445 data: 0.1493 max mem: 3532 +Test: [260/517] eta: 0:01:07 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1325 (0.4866) acc1: 100.0000 (81.0006) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2398 data: 0.1432 max mem: 3532 +Test: [270/517] eta: 0:01:04 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1430 (0.4744) acc1: 100.0000 (81.6978) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2399 data: 0.1428 max mem: 3532 +Test: [280/517] eta: 0:01:02 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1574 (0.4630) acc1: 100.0000 (82.3223) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2549 data: 0.1582 max mem: 3532 +Test: [290/517] eta: 0:00:59 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1509 (0.4521) acc1: 99.7396 (82.9145) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2499 data: 0.1534 max mem: 3532 +Test: [300/517] eta: 0:00:56 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1329 (0.4409) acc1: 100.0000 (83.4787) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2399 data: 0.1415 max mem: 3532 +Test: [310/517] eta: 0:00:53 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1117 (0.4303) acc1: 100.0000 (84.0099) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2356 data: 0.1383 max mem: 3532 +Test: [320/517] eta: 0:00:51 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1188 (0.4208) acc1: 100.0000 (84.5080) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2406 data: 0.1466 max mem: 3532 +Test: [330/517] eta: 0:00:48 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1257 (0.4118) acc1: 100.0000 (84.9761) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2344 data: 0.1411 max mem: 3532 +Test: [340/517] eta: 0:00:45 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1209 (0.4035) acc1: 100.0000 (85.4136) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2341 data: 0.1372 max mem: 3532 +Test: [350/517] eta: 0:00:43 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1311 (0.3959) acc1: 100.0000 (85.8240) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2447 data: 0.1455 max mem: 3532 +Test: [360/517] eta: 0:00:40 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1347 (0.3888) acc1: 100.0000 (86.2116) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2599 data: 0.1632 max mem: 3532 +Test: [370/517] eta: 0:00:37 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1226 (0.3814) acc1: 100.0000 (86.5833) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2453 data: 0.1494 max mem: 3532 +Test: [380/517] eta: 0:00:35 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1192 (0.3747) acc1: 100.0000 (86.9347) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2399 data: 0.1431 max mem: 3532 +Test: [390/517] eta: 0:00:32 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1286 (0.3686) acc1: 100.0000 (87.2656) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2900 data: 0.1938 max mem: 3532 +Test: [400/517] eta: 0:00:30 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1411 (0.3633) acc1: 99.7396 (87.5623) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2504 data: 0.1561 max mem: 3532 +Test: [410/517] eta: 0:00:27 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1377 (0.3577) acc1: 99.7396 (87.8618) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2295 data: 0.1363 max mem: 3532 +Test: [420/517] eta: 0:00:24 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1295 (0.3527) acc1: 100.0000 (88.1464) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2593 data: 0.1669 max mem: 3532 +Test: [430/517] eta: 0:00:22 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1735 (0.3489) acc1: 99.4792 (88.4045) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2549 data: 0.1627 max mem: 3532 +Test: [440/517] eta: 0:00:19 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1810 (0.3446) acc1: 99.4792 (88.6574) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2458 data: 0.1544 max mem: 3533 +Test: [450/517] eta: 0:00:17 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1489 (0.3409) acc1: 99.7396 (88.8870) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2699 data: 0.1767 max mem: 3533 +Test: [460/517] eta: 0:00:14 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1662 (0.3374) acc1: 99.4792 (89.0941) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2548 data: 0.1598 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1171 (0.3327) acc1: 99.7396 (89.3235) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2349 data: 0.1403 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1122 (0.3284) acc1: 100.0000 (89.5444) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2490 data: 0.1521 max mem: 3533 +Test: [490/517] eta: 0:00:06 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1393 (0.3274) acc1: 99.7396 (89.5929) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.2349 data: 0.1382 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1174 (0.3231) acc1: 100.0000 (89.8006) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2366 data: 0.1420 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1134 (0.3196) acc1: 100.0000 (89.9977) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2105 data: 0.1172 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 1.376 acc5_pos: 100.000 loss: 0.1417 (0.3177) acc1: 100.0000 (90.1047) acc5: 100.0000 (100.0000) acc1_neg: 99.647 acc5_neg: 100.000 time: 0.1971 data: 0.1048 max mem: 3533 +Test: Total time: 0:02:10 (0.2528 s / it) +* Overall Acc@1 90.105 Acc@5 100.000 + target == real Acc@1 3.914 Acc@5 100.000 + target == fake Acc@1 99.046 Acc@5 100.000 + loss 0.318 + PR-AUC 0.9212 ROC-AUC 0.5978 EER 42.67% (thr=0.8789) +Accuracy of the network on the 198427 test images: 90.1% +[rank0]:[W726 03:56:23.399446080 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 03:56:42.996000 1173552 torch/distributed/run.py:766] +W0726 03:56:42.996000 1173552 torch/distributed/run.py:766] ***************************************** +W0726 03:56:42.996000 1173552 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:56:42.996000 1173552 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:57:10.515596635 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:57:10.522488469 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:57:10.805288517 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 03:57:11.939113122 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:57:11.124091605 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:57:11.247229146 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 03:57:11.315695115 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W726 03:57:11.315703068 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n11_seed6_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n11_classseed6_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:37:52 acc1_pos: 4.688 acc5_pos: 100.000 loss: 1.6183 (1.6183) acc1: 4.6875 (4.6875) acc5: 100.0000 (100.0000) time: 5.8421 data: 3.3696 max mem: 3531 +Test: [ 10/389] eta: 0:03:52 acc1_pos: 42.969 acc5_pos: 100.000 loss: 0.9751 (1.0869) acc1: 29.6875 (28.9773) acc5: 100.0000 (100.0000) time: 0.6134 data: 0.3068 max mem: 3531 +Test: [ 20/389] eta: 0:02:15 acc1_pos: 0.521 acc5_pos: 100.000 loss: 1.0047 (1.2671) acc1: 21.0938 (23.5615) acc5: 100.0000 (100.0000) time: 0.0939 data: 0.0010 max mem: 3531 +Test: [ 30/389] eta: 0:01:48 acc1_pos: 2.344 acc5_pos: 100.000 loss: 1.3190 (1.3058) acc1: 3.6458 (17.2127) acc5: 100.0000 (100.0000) time: 0.1306 data: 0.0356 max mem: 3531 +Test: [ 40/389] eta: 0:01:40 acc1_pos: 0.781 acc5_pos: 100.000 loss: 1.3538 (1.3517) acc1: 2.8646 (14.1705) acc5: 100.0000 (100.0000) time: 0.2049 data: 0.1113 max mem: 3531 +Test: [ 50/389] eta: 0:01:34 acc1_pos: 34.115 acc5_pos: 100.000 loss: 1.3428 (1.3249) acc1: 7.2917 (14.9510) acc5: 100.0000 (100.0000) time: 0.2439 data: 0.1518 max mem: 3531 +Test: [ 60/389] eta: 0:01:27 acc1_pos: 18.490 acc5_pos: 100.000 loss: 1.0498 (1.3347) acc1: 20.8333 (15.0359) acc5: 100.0000 (100.0000) time: 0.2162 data: 0.1255 max mem: 3531 +Test: [ 70/389] eta: 0:01:20 acc1_pos: 41.146 acc5_pos: 100.000 loss: 1.0123 (1.2698) acc1: 26.8229 (19.1608) acc5: 100.0000 (100.0000) time: 0.1794 data: 0.0872 max mem: 3531 +Test: [ 80/389] eta: 0:01:16 acc1_pos: 71.615 acc5_pos: 100.000 loss: 0.9220 (1.2656) acc1: 41.1458 (21.2577) acc5: 100.0000 (100.0000) time: 0.1995 data: 0.1064 max mem: 3531 +Test: [ 90/389] eta: 0:01:11 acc1_pos: 31.250 acc5_pos: 100.000 loss: 0.7020 (1.1978) acc1: 56.7708 (26.0073) acc5: 100.0000 (100.0000) time: 0.1923 data: 0.1011 max mem: 3531 +Test: [100/389] eta: 0:01:07 acc1_pos: 75.521 acc5_pos: 100.000 loss: 0.9242 (1.1996) acc1: 39.3229 (25.8199) acc5: 100.0000 (100.0000) time: 0.1763 data: 0.0856 max mem: 3531 +Test: [110/389] eta: 0:01:04 acc1_pos: 67.708 acc5_pos: 100.000 loss: 1.1238 (1.1901) acc1: 31.7708 (27.5127) acc5: 100.0000 (100.0000) time: 0.1937 data: 0.1026 max mem: 3531 +Test: [120/389] eta: 0:01:00 acc1_pos: 17.188 acc5_pos: 100.000 loss: 1.0537 (1.1785) acc1: 35.1562 (28.2907) acc5: 100.0000 (100.0000) time: 0.1884 data: 0.0976 max mem: 3531 +Test: [130/389] eta: 0:00:57 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.6010 (1.1323) acc1: 53.1250 (31.6535) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1794 data: 0.0889 max mem: 3531 +Test: [140/389] eta: 0:00:54 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.2155 (1.0708) acc1: 97.1354 (35.9116) acc5: 100.0000 (100.0000) acc1_neg: 55.729 acc5_neg: 100.000 time: 0.1854 data: 0.0944 max mem: 3531 +Test: [150/389] eta: 0:00:52 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.3087 (1.0286) acc1: 94.7917 (39.0556) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1862 data: 0.0936 max mem: 3531 +Test: [160/389] eta: 0:00:49 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.3087 (0.9791) acc1: 94.7917 (42.6517) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1763 data: 0.0827 max mem: 3531 +Test: [170/389] eta: 0:00:46 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.1827 (0.9351) acc1: 99.7396 (45.9080) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1776 data: 0.0859 max mem: 3531 +Test: [180/389] eta: 0:00:44 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.2067 (0.9007) acc1: 99.4792 (48.5166) acc5: 100.0000 (100.0000) acc1_neg: 75.521 acc5_neg: 100.000 time: 0.1823 data: 0.0919 max mem: 3531 +Test: [190/389] eta: 0:00:41 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.2542 (0.8692) acc1: 97.1354 (50.8917) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1650 data: 0.0703 max mem: 3531 +Test: [200/389] eta: 0:00:38 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.2118 (0.8381) acc1: 99.7396 (53.0810) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1349 data: 0.0392 max mem: 3531 +Test: [210/389] eta: 0:00:36 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.1588 (0.8082) acc1: 100.0000 (55.1034) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1547 data: 0.0620 max mem: 3532 +Test: [220/389] eta: 0:00:33 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.1390 (0.7780) acc1: 100.0000 (57.1279) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1702 data: 0.0764 max mem: 3532 +Test: [230/389] eta: 0:00:31 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.1690 (0.7574) acc1: 99.4792 (58.7279) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1349 data: 0.0400 max mem: 3532 +Test: [240/389] eta: 0:00:28 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.2894 (0.7403) acc1: 96.6146 (60.0903) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1197 data: 0.0260 max mem: 3532 +Test: [250/389] eta: 0:00:26 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.2404 (0.7189) acc1: 98.9583 (61.6441) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1552 data: 0.0625 max mem: 3532 +Test: [260/389] eta: 0:00:24 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.1484 (0.7008) acc1: 99.7396 (62.8213) acc5: 100.0000 (100.0000) acc1_neg: 59.115 acc5_neg: 100.000 time: 0.1807 data: 0.0879 max mem: 3532 +Test: [270/389] eta: 0:00:22 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.2002 (0.6919) acc1: 95.5729 (63.3764) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1596 data: 0.0657 max mem: 3532 +Test: [280/389] eta: 0:00:20 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.2802 (0.6775) acc1: 93.7500 (64.3794) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1390 data: 0.0423 max mem: 3532 +Test: [290/389] eta: 0:00:18 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.2034 (0.6631) acc1: 99.7396 (65.4147) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1606 data: 0.0657 max mem: 3532 +Test: [300/389] eta: 0:00:16 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.2270 (0.6482) acc1: 99.7396 (66.5559) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1799 data: 0.0880 max mem: 3532 +Test: [310/389] eta: 0:00:14 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.2380 (0.6369) acc1: 99.4792 (67.4521) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.1792 data: 0.0850 max mem: 3532 +Test: [320/389] eta: 0:00:12 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.2499 (0.6249) acc1: 97.9167 (68.3476) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1722 data: 0.0766 max mem: 3532 +Test: [330/389] eta: 0:00:10 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.1983 (0.6124) acc1: 99.7396 (69.2669) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1464 data: 0.0530 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.1971 (0.6000) acc1: 99.7396 (70.1185) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1377 data: 0.0464 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.2270 (0.5957) acc1: 99.2188 (70.3778) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1561 data: 0.0654 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.3321 (0.5889) acc1: 92.7083 (70.9993) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1655 data: 0.0750 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.3312 (0.5830) acc1: 92.9688 (71.3724) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1741 data: 0.0826 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.2146 (0.5733) acc1: 95.3125 (72.0568) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1776 data: 0.0856 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.2034 (0.5662) acc1: 98.9583 (72.5626) acc5: 100.0000 (100.0000) acc1_neg: 95.902 acc5_neg: 100.000 time: 0.1692 data: 0.0739 max mem: 3532 +Test: Total time: 0:01:11 (0.1833 s / it) +* Overall Acc@1 72.563 Acc@5 100.000 + target == real Acc@1 27.832 Acc@5 100.000 + target == fake Acc@1 93.354 Acc@5 100.000 + loss 0.566 + PR-AUC 0.8042 ROC-AUC 0.7078 EER 33.92% (thr=0.7769) +Accuracy of the network on the 149358 test images: 72.6% +[rank0]:[W726 03:58:37.016070495 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 03:58:55.505000 1174110 torch/distributed/run.py:766] +W0726 03:58:55.505000 1174110 torch/distributed/run.py:766] ***************************************** +W0726 03:58:55.505000 1174110 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 03:58:55.505000 1174110 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 03:59:21.597868368 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 03:59:21.705275987 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 03:59:22.139453450 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank1]:[W726 03:59:22.141106806 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 03:59:22.453010723 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +| distributed init (rank 3): env:// +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +NCCL version 2.26.2+cuda12.2 +[rank3]:[W726 03:59:22.607010616 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank4]:[W726 03:59:22.607165956 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 03:59:22.612808939 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n11_seed6_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n11_classseed6_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:36:12 acc1_pos: 1.042 acc5_pos: 100.000 loss: 2.3643 (2.3643) acc1: 1.0417 (1.0417) acc5: 100.0000 (100.0000) time: 6.5240 data: 4.7953 max mem: 3531 +Test: [ 10/333] eta: 0:03:38 acc1_pos: 78.906 acc5_pos: 100.000 loss: 2.0911 (1.5868) acc1: 2.0833 (30.3741) acc5: 100.0000 (100.0000) time: 0.6752 data: 0.4363 max mem: 3531 +Test: [ 20/333] eta: 0:02:05 acc1_pos: 92.188 acc5_pos: 100.000 loss: 0.4566 (1.0339) acc1: 77.6042 (54.8611) acc5: 100.0000 (100.0000) time: 0.0932 data: 0.0004 max mem: 3531 +Test: [ 30/333] eta: 0:01:39 acc1_pos: 15.104 acc5_pos: 100.000 loss: 0.5800 (1.1982) acc1: 59.1146 (43.0360) acc5: 100.0000 (100.0000) time: 0.1388 data: 0.0455 max mem: 3531 +Test: [ 40/333] eta: 0:01:25 acc1_pos: 14.583 acc5_pos: 100.000 loss: 1.6123 (1.2828) acc1: 15.1042 (36.6870) acc5: 100.0000 (100.0000) time: 0.1798 data: 0.0883 max mem: 3531 +Test: [ 50/333] eta: 0:01:16 acc1_pos: 19.271 acc5_pos: 100.000 loss: 1.5902 (1.3529) acc1: 13.8021 (31.7606) acc5: 100.0000 (100.0000) time: 0.1790 data: 0.0868 max mem: 3531 +Test: [ 60/333] eta: 0:01:11 acc1_pos: 9.635 acc5_pos: 100.000 loss: 1.6523 (1.3967) acc1: 11.1979 (28.9148) acc5: 100.0000 (100.0000) time: 0.1999 data: 0.1041 max mem: 3531 +Test: [ 70/333] eta: 0:01:07 acc1_pos: 21.875 acc5_pos: 100.000 loss: 1.6360 (1.4220) acc1: 13.8021 (27.2117) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1281 max mem: 3531 +Test: [ 80/333] eta: 0:01:04 acc1_pos: 15.885 acc5_pos: 100.000 loss: 1.6085 (1.4450) acc1: 14.5833 (25.5626) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1365 max mem: 3531 +Test: [ 90/333] eta: 0:00:58 acc1_pos: 11.979 acc5_pos: 100.000 loss: 1.6085 (1.4609) acc1: 13.5417 (24.1443) acc5: 100.0000 (100.0000) time: 0.1858 data: 0.0931 max mem: 3531 +Test: [100/333] eta: 0:00:55 acc1_pos: 6.250 acc5_pos: 100.000 loss: 1.6036 (1.4746) acc1: 11.1979 (23.0250) acc5: 100.0000 (100.0000) time: 0.1799 data: 0.0856 max mem: 3531 +Test: [110/333] eta: 0:00:52 acc1_pos: 7.552 acc5_pos: 100.000 loss: 1.6143 (1.4897) acc1: 10.4167 (21.9266) acc5: 100.0000 (100.0000) time: 0.2053 data: 0.1119 max mem: 3531 +Test: [120/333] eta: 0:00:48 acc1_pos: 10.417 acc5_pos: 100.000 loss: 1.6143 (1.4986) acc1: 10.6771 (21.1777) acc5: 100.0000 (100.0000) time: 0.1807 data: 0.0890 max mem: 3531 +Test: [130/333] eta: 0:00:45 acc1_pos: 19.531 acc5_pos: 100.000 loss: 1.5448 (1.5026) acc1: 13.8021 (20.6007) acc5: 100.0000 (100.0000) time: 0.1686 data: 0.0741 max mem: 3531 +Test: [140/333] eta: 0:00:43 acc1_pos: 9.896 acc5_pos: 100.000 loss: 1.5471 (1.5050) acc1: 13.2812 (20.1297) acc5: 100.0000 (100.0000) time: 0.1965 data: 0.1027 max mem: 3531 +Test: [150/333] eta: 0:00:40 acc1_pos: 8.854 acc5_pos: 100.000 loss: 1.5656 (1.5090) acc1: 13.0208 (19.6985) acc5: 100.0000 (100.0000) time: 0.2061 data: 0.1150 max mem: 3531 +Test: [160/333] eta: 0:00:38 acc1_pos: 7.895 acc5_pos: 100.000 loss: 1.4875 (1.4830) acc1: 15.3646 (21.1099) acc5: 100.0000 (100.0000) acc1_neg: 80.729 acc5_neg: 100.000 time: 0.1935 data: 0.1021 max mem: 3531 +Test: [170/333] eta: 0:00:35 acc1_pos: 7.895 acc5_pos: 100.000 loss: 0.4304 (1.4204) acc1: 80.7292 (24.7061) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.1845 data: 0.0932 max mem: 3531 +Test: [180/333] eta: 0:00:33 acc1_pos: 7.895 acc5_pos: 100.000 loss: 0.3999 (1.3636) acc1: 83.3333 (27.9437) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.1889 data: 0.0952 max mem: 3531 +Test: [190/333] eta: 0:00:30 acc1_pos: 7.895 acc5_pos: 100.000 loss: 0.3837 (1.3100) acc1: 83.8542 (31.0155) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1791 data: 0.0814 max mem: 3531 +Test: [200/333] eta: 0:00:28 acc1_pos: 7.895 acc5_pos: 100.000 loss: 0.3634 (1.2627) acc1: 85.1562 (33.7492) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.1899 data: 0.0930 max mem: 3531 +Test: [210/333] eta: 0:00:26 acc1_pos: 7.895 acc5_pos: 100.000 loss: 0.3739 (1.2205) acc1: 84.1146 (36.1374) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.2301 data: 0.1359 max mem: 3532 +Test: [220/333] eta: 0:00:24 acc1_pos: 7.895 acc5_pos: 100.000 loss: 0.3675 (1.1821) acc1: 84.8958 (38.4356) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.2098 data: 0.1157 max mem: 3532 +Test: [230/333] eta: 0:00:21 acc1_pos: 7.895 acc5_pos: 100.000 loss: 0.3295 (1.1436) acc1: 90.1042 (40.6870) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1465 data: 0.0501 max mem: 3532 +Test: [240/333] eta: 0:00:19 acc1_pos: 7.895 acc5_pos: 100.000 loss: 0.3087 (1.1091) acc1: 89.0625 (42.6910) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1477 data: 0.0515 max mem: 3532 +Test: [250/333] eta: 0:00:17 acc1_pos: 7.895 acc5_pos: 100.000 loss: 0.3134 (1.0777) acc1: 89.0625 (44.5240) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.2179 data: 0.1242 max mem: 3532 +Test: [260/333] eta: 0:00:15 acc1_pos: 7.895 acc5_pos: 100.000 loss: 0.3019 (1.0479) acc1: 90.1042 (46.2694) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.2121 data: 0.1161 max mem: 3532 +Test: [270/333] eta: 0:00:13 acc1_pos: 7.895 acc5_pos: 100.000 loss: 0.3207 (1.0222) acc1: 87.7604 (47.7437) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1808 data: 0.0835 max mem: 3532 +Test: [280/333] eta: 0:00:11 acc1_pos: 7.895 acc5_pos: 100.000 loss: 0.3616 (0.9987) acc1: 85.6771 (49.1094) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1937 data: 0.0985 max mem: 3532 +Test: [290/333] eta: 0:00:08 acc1_pos: 7.895 acc5_pos: 100.000 loss: 0.3700 (0.9782) acc1: 84.1146 (50.2765) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.2007 data: 0.1085 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 7.895 acc5_pos: 100.000 loss: 0.4030 (0.9596) acc1: 82.2917 (51.3272) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.1750 data: 0.0827 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 7.895 acc5_pos: 100.000 loss: 0.3845 (0.9392) acc1: 84.6354 (52.5020) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1471 data: 0.0525 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 7.895 acc5_pos: 100.000 loss: 0.3210 (0.9206) acc1: 87.5000 (53.5720) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.1385 data: 0.0446 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 7.895 acc5_pos: 100.000 loss: 0.3636 (0.9047) acc1: 85.4167 (54.4845) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.1217 data: 0.0303 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 7.895 acc5_pos: 100.000 loss: 0.3815 (0.9017) acc1: 84.1146 (54.6522) acc5: 100.0000 (100.0000) acc1_neg: 83.099 acc5_neg: 100.000 time: 0.1239 data: 0.0303 max mem: 3532 +Test: Total time: 0:01:06 (0.1990 s / it) +* Overall Acc@1 54.652 Acc@5 100.000 + target == real Acc@1 19.532 Acc@5 100.000 + target == fake Acc@1 85.916 Acc@5 100.000 + loss 0.902 + PR-AUC 0.5523 ROC-AUC 0.5442 EER 46.70% (thr=0.8037) +Accuracy of the network on the 127843 test images: 54.7% +[rank0]:[W726 04:00:40.279882590 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:01:00.428000 1174656 torch/distributed/run.py:766] +W0726 04:01:00.428000 1174656 torch/distributed/run.py:766] ***************************************** +W0726 04:01:00.428000 1174656 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:01:00.428000 1174656 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:01:29.583333273 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:01:31.028041500 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:01:31.174983160 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:01:31.194587837 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:01:31.291727522 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:01:31.333833775 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +[rank0]:[W726 04:01:31.351745685 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:01:31.352520460 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n11_seed6_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n11_classseed6_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:21:57 acc1_pos: 44.271 acc5_pos: 100.000 loss: 1.0237 (1.0237) acc1: 44.2708 (44.2708) acc5: 100.0000 (100.0000) time: 5.6062 data: 3.7561 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/235] eta: 0:02:13 acc1_pos: 48.177 acc5_pos: 100.000 loss: 0.9914 (0.9934) acc1: 44.7917 (44.8390) acc5: 100.0000 (100.0000) time: 0.5940 data: 0.3433 max mem: 3531 +Test: [ 20/235] eta: 0:01:16 acc1_pos: 45.052 acc5_pos: 100.000 loss: 0.9851 (0.9900) acc1: 44.7917 (44.7917) acc5: 100.0000 (100.0000) time: 0.0957 data: 0.0043 max mem: 3531 +Test: [ 30/235] eta: 0:00:59 acc1_pos: 41.667 acc5_pos: 100.000 loss: 0.9837 (0.9876) acc1: 45.0521 (45.0101) acc5: 100.0000 (100.0000) time: 0.1261 data: 0.0334 max mem: 3531 +Test: [ 40/235] eta: 0:00:54 acc1_pos: 42.188 acc5_pos: 100.000 loss: 0.9873 (0.9883) acc1: 44.5312 (44.9441) acc5: 100.0000 (100.0000) time: 0.2009 data: 0.1071 max mem: 3531 +Test: [ 50/235] eta: 0:00:49 acc1_pos: 42.188 acc5_pos: 100.000 loss: 0.9787 (0.9871) acc1: 44.5312 (45.0470) acc5: 100.0000 (100.0000) time: 0.2329 data: 0.1380 max mem: 3531 +Test: [ 60/235] eta: 0:00:45 acc1_pos: 47.917 acc5_pos: 100.000 loss: 0.9712 (0.9852) acc1: 47.1354 (45.3296) acc5: 100.0000 (100.0000) time: 0.2142 data: 0.1203 max mem: 3531 +Test: [ 70/235] eta: 0:00:41 acc1_pos: 47.692 acc5_pos: 100.000 loss: 0.9402 (0.9522) acc1: 48.1771 (47.2528) acc5: 100.0000 (100.0000) acc1_neg: 59.635 acc5_neg: 100.000 time: 0.1999 data: 0.1059 max mem: 3531 +Test: [ 80/235] eta: 0:00:37 acc1_pos: 47.692 acc5_pos: 100.000 loss: 0.7134 (0.9231) acc1: 61.1979 (48.9423) acc5: 100.0000 (100.0000) acc1_neg: 56.250 acc5_neg: 100.000 time: 0.1946 data: 0.1001 max mem: 3531 +Test: [ 90/235] eta: 0:00:34 acc1_pos: 47.692 acc5_pos: 100.000 loss: 0.7205 (0.9024) acc1: 60.6771 (50.1746) acc5: 100.0000 (100.0000) acc1_neg: 59.896 acc5_neg: 100.000 time: 0.2004 data: 0.1069 max mem: 3531 +Test: [100/235] eta: 0:00:31 acc1_pos: 47.692 acc5_pos: 100.000 loss: 0.7206 (0.8835) acc1: 59.8958 (51.2660) acc5: 100.0000 (100.0000) acc1_neg: 62.240 acc5_neg: 100.000 time: 0.1949 data: 0.1008 max mem: 3531 +Test: [110/235] eta: 0:00:28 acc1_pos: 47.692 acc5_pos: 100.000 loss: 0.7100 (0.8685) acc1: 60.1562 (52.0388) acc5: 100.0000 (100.0000) acc1_neg: 61.198 acc5_neg: 100.000 time: 0.1944 data: 0.1006 max mem: 3531 +Test: [120/235] eta: 0:00:25 acc1_pos: 47.692 acc5_pos: 100.000 loss: 0.7152 (0.8555) acc1: 59.8958 (52.7398) acc5: 100.0000 (100.0000) acc1_neg: 61.719 acc5_neg: 100.000 time: 0.1756 data: 0.0814 max mem: 3531 +Test: [130/235] eta: 0:00:22 acc1_pos: 47.692 acc5_pos: 100.000 loss: 0.7025 (0.8435) acc1: 60.9375 (53.4371) acc5: 100.0000 (100.0000) acc1_neg: 63.021 acc5_neg: 100.000 time: 0.1499 data: 0.0552 max mem: 3531 +Test: [140/235] eta: 0:00:20 acc1_pos: 47.692 acc5_pos: 100.000 loss: 0.7025 (0.8342) acc1: 60.6771 (53.8841) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.1693 data: 0.0755 max mem: 3531 +Test: [150/235] eta: 0:00:18 acc1_pos: 47.692 acc5_pos: 100.000 loss: 0.7045 (0.8251) acc1: 59.8958 (54.3426) acc5: 100.0000 (100.0000) acc1_neg: 64.583 acc5_neg: 100.000 time: 0.2054 data: 0.1132 max mem: 3531 +Test: [160/235] eta: 0:00:15 acc1_pos: 47.692 acc5_pos: 100.000 loss: 0.6967 (0.8178) acc1: 59.8958 (54.7037) acc5: 100.0000 (100.0000) acc1_neg: 59.115 acc5_neg: 100.000 time: 0.1799 data: 0.0859 max mem: 3531 +Test: [170/235] eta: 0:00:13 acc1_pos: 47.692 acc5_pos: 100.000 loss: 0.6927 (0.8099) acc1: 61.9792 (55.3104) acc5: 100.0000 (100.0000) acc1_neg: 62.760 acc5_neg: 100.000 time: 0.1544 data: 0.0602 max mem: 3531 +Test: [180/235] eta: 0:00:11 acc1_pos: 47.692 acc5_pos: 100.000 loss: 0.6935 (0.8037) acc1: 64.3229 (55.7910) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.1649 data: 0.0722 max mem: 3531 +Test: [190/235] eta: 0:00:09 acc1_pos: 47.692 acc5_pos: 100.000 loss: 0.6940 (0.7979) acc1: 64.3229 (56.2555) acc5: 100.0000 (100.0000) acc1_neg: 66.667 acc5_neg: 100.000 time: 0.1899 data: 0.0964 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 47.692 acc5_pos: 100.000 loss: 0.6940 (0.7937) acc1: 63.5417 (56.5933) acc5: 100.0000 (100.0000) acc1_neg: 60.677 acc5_neg: 100.000 time: 0.2116 data: 0.1169 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 47.692 acc5_pos: 100.000 loss: 0.7010 (0.7892) acc1: 63.2812 (56.9461) acc5: 100.0000 (100.0000) acc1_neg: 61.458 acc5_neg: 100.000 time: 0.1699 data: 0.0724 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 47.692 acc5_pos: 100.000 loss: 0.6953 (0.7853) acc1: 63.2812 (57.2646) acc5: 100.0000 (100.0000) acc1_neg: 64.844 acc5_neg: 100.000 time: 0.1344 data: 0.0364 max mem: 3532 +Test: [230/235] eta: 0:00:00 acc1_pos: 47.692 acc5_pos: 100.000 loss: 0.6847 (0.7805) acc1: 65.1042 (57.6276) acc5: 100.0000 (100.0000) acc1_neg: 68.750 acc5_neg: 100.000 time: 0.1467 data: 0.0522 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 47.692 acc5_pos: 100.000 loss: 0.6847 (0.7809) acc1: 65.1042 (57.7042) acc5: 100.0000 (100.0000) acc1_neg: 52.174 acc5_neg: 100.000 time: 0.1423 data: 0.0521 max mem: 3532 +Test: Total time: 0:00:46 (0.1976 s / it) +* Overall Acc@1 57.704 Acc@5 100.000 + target == real Acc@1 45.298 Acc@5 100.000 + target == fake Acc@1 62.191 Acc@5 100.000 + loss 0.781 + PR-AUC 0.7719 ROC-AUC 0.5595 EER 45.39% (thr=0.5879) +Accuracy of the network on the 89879 test images: 57.7% +[rank0]:[W726 04:02:29.229544823 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:02:48.150000 1175216 torch/distributed/run.py:766] +W0726 04:02:48.150000 1175216 torch/distributed/run.py:766] ***************************************** +W0726 04:02:48.150000 1175216 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:02:48.150000 1175216 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:03:15.438827323 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:03:15.861733431 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:03:16.024604385 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:03:16.205882418 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +[rank7]:[W726 04:03:16.210048216 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:03:16.360980651 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:03:16.428059000 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W726 04:03:16.428058507 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n11_seed6_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n11_classseed6_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:30:32 acc1_pos: 8.333 acc5_pos: 100.000 loss: 1.6383 (1.6383) acc1: 8.3333 (8.3333) acc5: 100.0000 (100.0000) time: 6.8117 data: 5.2348 max mem: 3531 +Test: [ 10/269] eta: 0:03:01 acc1_pos: 17.969 acc5_pos: 100.000 loss: 1.2867 (1.2686) acc1: 20.3125 (24.3608) acc5: 100.0000 (100.0000) time: 0.7014 data: 0.4763 max mem: 3531 +Test: [ 20/269] eta: 0:01:43 acc1_pos: 27.604 acc5_pos: 100.000 loss: 1.1136 (1.1829) acc1: 27.6042 (28.8566) acc5: 100.0000 (100.0000) time: 0.0949 data: 0.0004 max mem: 3531 +Test: [ 30/269] eta: 0:01:24 acc1_pos: 8.854 acc5_pos: 100.000 loss: 1.0496 (1.1772) acc1: 31.7708 (31.3508) acc5: 100.0000 (100.0000) time: 0.1611 data: 0.0645 max mem: 3531 +Test: [ 40/269] eta: 0:01:11 acc1_pos: 32.759 acc5_pos: 100.000 loss: 0.4915 (0.9849) acc1: 76.5625 (44.6011) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2049 data: 0.1098 max mem: 3531 +Test: [ 50/269] eta: 0:01:04 acc1_pos: 32.759 acc5_pos: 100.000 loss: 0.2554 (0.8533) acc1: 93.7500 (53.6560) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.1995 data: 0.1038 max mem: 3531 +Test: [ 60/269] eta: 0:00:57 acc1_pos: 32.759 acc5_pos: 100.000 loss: 0.3297 (0.7701) acc1: 89.5833 (59.3110) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.2008 data: 0.1074 max mem: 3531 +Test: [ 70/269] eta: 0:00:53 acc1_pos: 32.759 acc5_pos: 100.000 loss: 0.3032 (0.7022) acc1: 90.1042 (63.7874) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1999 data: 0.1081 max mem: 3531 +Test: [ 80/269] eta: 0:00:48 acc1_pos: 32.759 acc5_pos: 100.000 loss: 0.2853 (0.6593) acc1: 89.8438 (66.4995) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.2066 data: 0.1137 max mem: 3531 +Test: [ 90/269] eta: 0:00:45 acc1_pos: 32.759 acc5_pos: 100.000 loss: 0.4444 (0.6524) acc1: 77.6042 (66.5150) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.2142 data: 0.1221 max mem: 3531 +Test: [100/269] eta: 0:00:42 acc1_pos: 32.759 acc5_pos: 100.000 loss: 0.4748 (0.6292) acc1: 77.3438 (68.1621) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.2333 data: 0.1389 max mem: 3531 +Test: [110/269] eta: 0:00:40 acc1_pos: 32.759 acc5_pos: 100.000 loss: 0.3563 (0.5993) acc1: 88.8021 (70.4580) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2447 data: 0.1475 max mem: 3531 +Test: [120/269] eta: 0:00:36 acc1_pos: 32.759 acc5_pos: 100.000 loss: 0.3342 (0.5827) acc1: 89.5833 (71.6253) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.2047 data: 0.1078 max mem: 3531 +Test: [130/269] eta: 0:00:33 acc1_pos: 32.759 acc5_pos: 100.000 loss: 0.4088 (0.5710) acc1: 80.9896 (72.3123) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.1749 data: 0.0795 max mem: 3531 +Test: [140/269] eta: 0:00:30 acc1_pos: 32.759 acc5_pos: 100.000 loss: 0.4709 (0.5724) acc1: 78.9062 (72.0209) acc5: 100.0000 (100.0000) acc1_neg: 50.000 acc5_neg: 100.000 time: 0.2040 data: 0.1085 max mem: 3531 +Test: [150/269] eta: 0:00:27 acc1_pos: 32.759 acc5_pos: 100.000 loss: 0.6144 (0.5754) acc1: 66.6667 (71.6801) acc5: 100.0000 (100.0000) acc1_neg: 67.708 acc5_neg: 100.000 time: 0.1709 data: 0.0734 max mem: 3531 +Test: [160/269] eta: 0:00:25 acc1_pos: 32.759 acc5_pos: 100.000 loss: 0.4738 (0.5661) acc1: 77.6042 (72.3829) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1671 data: 0.0728 max mem: 3531 +Test: [170/269] eta: 0:00:22 acc1_pos: 32.759 acc5_pos: 100.000 loss: 0.3504 (0.5513) acc1: 90.1042 (73.5715) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2106 data: 0.1181 max mem: 3531 +Test: [180/269] eta: 0:00:20 acc1_pos: 32.759 acc5_pos: 100.000 loss: 0.3306 (0.5408) acc1: 89.8438 (74.2777) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.2135 data: 0.1199 max mem: 3531 +Test: [190/269] eta: 0:00:18 acc1_pos: 32.759 acc5_pos: 100.000 loss: 0.3560 (0.5356) acc1: 86.4583 (74.6305) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2331 data: 0.1399 max mem: 3531 +Test: [200/269] eta: 0:00:15 acc1_pos: 32.759 acc5_pos: 100.000 loss: 0.3560 (0.5283) acc1: 84.1146 (75.0441) acc5: 100.0000 (100.0000) acc1_neg: 72.396 acc5_neg: 100.000 time: 0.2142 data: 0.1183 max mem: 3531 +Test: [210/269] eta: 0:00:13 acc1_pos: 32.759 acc5_pos: 100.000 loss: 0.3932 (0.5246) acc1: 83.3333 (75.2111) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.2122 data: 0.1170 max mem: 3532 +Test: [220/269] eta: 0:00:11 acc1_pos: 32.759 acc5_pos: 100.000 loss: 0.4157 (0.5192) acc1: 82.0312 (75.5138) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2116 data: 0.1209 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 32.759 acc5_pos: 100.000 loss: 0.2443 (0.5073) acc1: 92.9688 (76.2750) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1789 data: 0.0885 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 32.759 acc5_pos: 100.000 loss: 0.2895 (0.5114) acc1: 90.1042 (76.0503) acc5: 100.0000 (100.0000) acc1_neg: 41.927 acc5_neg: 100.000 time: 0.1674 data: 0.0769 max mem: 3532 +Test: [250/269] eta: 0:00:04 acc1_pos: 32.759 acc5_pos: 100.000 loss: 0.6529 (0.5204) acc1: 55.9896 (75.2864) acc5: 100.0000 (100.0000) acc1_neg: 47.135 acc5_neg: 100.000 time: 0.1518 data: 0.0615 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 32.759 acc5_pos: 100.000 loss: 0.6529 (0.5235) acc1: 58.0729 (74.9940) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1392 data: 0.0493 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 32.759 acc5_pos: 100.000 loss: 0.3713 (0.5181) acc1: 85.6771 (75.4146) acc5: 100.0000 (100.0000) acc1_neg: 86.877 acc5_neg: 100.000 time: 0.1285 data: 0.0352 max mem: 3532 +Test: Total time: 0:00:57 (0.2130 s / it) +* Overall Acc@1 75.415 Acc@5 100.000 + target == real Acc@1 30.676 Acc@5 100.000 + target == fake Acc@1 81.520 Acc@5 100.000 + loss 0.518 + PR-AUC 0.9040 ROC-AUC 0.5977 EER 42.38% (thr=0.7241) +Accuracy of the network on the 103293 test images: 75.4% +[rank0]:[W726 04:04:21.083840439 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T04:04:29Z] eval_done tag=fixed500k_real_n11_seed6_resampled +[2026-07-26T04:04:29Z] eval_start tag=fixed500k_fake_n22_seed6_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed6_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T04:04:29Z +hostname=n116-041-188 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed6_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n22_seed6_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 04:04:41.816000 1175791 torch/distributed/run.py:766] +W0726 04:04:41.816000 1175791 torch/distributed/run.py:766] ***************************************** +W0726 04:04:41.816000 1175791 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:04:41.816000 1175791 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 7): env:// +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:05:10.947250655 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:05:10.289017158 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:05:10.326199407 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:05:10.358145094 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:05:10.447680872 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:05:10.591045336 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:05:10.605714144 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:05:10.616202916 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n22_seed6_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed6_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:25:31 acc1_pos: 17.969 acc5_pos: 100.000 loss: 1.0884 (1.0884) acc1: 17.9688 (17.9688) acc5: 100.0000 (100.0000) time: 6.5466 data: 4.7447 max mem: 3531 +Test: [ 10/234] eta: 0:02:31 acc1_pos: 8.073 acc5_pos: 100.000 loss: 0.9956 (0.9814) acc1: 37.2396 (34.9432) acc5: 100.0000 (100.0000) time: 0.6771 data: 0.4318 max mem: 3531 +Test: [ 20/234] eta: 0:01:25 acc1_pos: 95.833 acc5_pos: 100.000 loss: 0.9956 (0.9571) acc1: 34.6354 (37.6860) acc5: 100.0000 (100.0000) time: 0.0929 data: 0.0024 max mem: 3531 +Test: [ 30/234] eta: 0:01:03 acc1_pos: 59.375 acc5_pos: 100.000 loss: 0.2353 (0.7418) acc1: 94.0104 (54.4691) acc5: 100.0000 (100.0000) time: 0.1123 data: 0.0187 max mem: 3531 +Test: [ 40/234] eta: 0:00:56 acc1_pos: 99.479 acc5_pos: 100.000 loss: 0.2271 (0.6876) acc1: 95.0521 (58.5302) acc5: 100.0000 (100.0000) time: 0.1750 data: 0.0820 max mem: 3531 +Test: [ 50/234] eta: 0:00:50 acc1_pos: 99.740 acc5_pos: 100.000 loss: 0.1073 (0.5731) acc1: 99.2188 (66.5799) acc5: 100.0000 (100.0000) time: 0.2150 data: 0.1229 max mem: 3531 +Test: [ 60/234] eta: 0:00:46 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.1077 (0.4977) acc1: 99.7396 (72.0458) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2143 data: 0.1195 max mem: 3531 +Test: [ 70/234] eta: 0:00:41 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.1172 (0.5195) acc1: 99.7396 (71.2992) acc5: 100.0000 (100.0000) acc1_neg: 46.354 acc5_neg: 100.000 time: 0.2049 data: 0.1107 max mem: 3531 +Test: [ 80/234] eta: 0:00:38 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.9025 (0.5357) acc1: 51.0417 (70.7305) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1998 data: 0.1071 max mem: 3531 +Test: [ 90/234] eta: 0:00:34 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.2166 (0.4990) acc1: 92.7083 (73.3889) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.1948 data: 0.1021 max mem: 3531 +Test: [100/234] eta: 0:00:31 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.2309 (0.4868) acc1: 92.7083 (74.3219) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1749 data: 0.0797 max mem: 3531 +Test: [110/234] eta: 0:00:28 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.3656 (0.4784) acc1: 78.6458 (75.0587) acc5: 100.0000 (100.0000) acc1_neg: 71.094 acc5_neg: 100.000 time: 0.1712 data: 0.0767 max mem: 3531 +Test: [120/234] eta: 0:00:25 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.3081 (0.4813) acc1: 84.8958 (75.3121) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1702 data: 0.0782 max mem: 3531 +Test: [130/234] eta: 0:00:22 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.2926 (0.4694) acc1: 92.1875 (76.4253) acc5: 100.0000 (100.0000) acc1_neg: 79.427 acc5_neg: 100.000 time: 0.1653 data: 0.0732 max mem: 3531 +Test: [140/234] eta: 0:00:20 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.3889 (0.4878) acc1: 80.7292 (75.7517) acc5: 100.0000 (100.0000) acc1_neg: 43.750 acc5_neg: 100.000 time: 0.1637 data: 0.0728 max mem: 3531 +Test: [150/234] eta: 0:00:17 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.4841 (0.4899) acc1: 74.4792 (75.7381) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.1523 data: 0.0620 max mem: 3531 +Test: [160/234] eta: 0:00:15 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.3879 (0.4801) acc1: 83.8542 (76.5625) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.1379 data: 0.0475 max mem: 3531 +Test: [170/234] eta: 0:00:12 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.4004 (0.4750) acc1: 84.8958 (77.1321) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1448 data: 0.0521 max mem: 3531 +Test: [180/234] eta: 0:00:10 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.4247 (0.4893) acc1: 82.2917 (76.3265) acc5: 100.0000 (100.0000) acc1_neg: 23.958 acc5_neg: 100.000 time: 0.1469 data: 0.0521 max mem: 3531 +Test: [190/234] eta: 0:00:08 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.7324 (0.5042) acc1: 59.1146 (75.3504) acc5: 100.0000 (100.0000) acc1_neg: 60.156 acc5_neg: 100.000 time: 0.1425 data: 0.0494 max mem: 3531 +Test: [200/234] eta: 0:00:06 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.7182 (0.4961) acc1: 60.4167 (75.8396) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1414 data: 0.0496 max mem: 3531 +Test: [210/234] eta: 0:00:04 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.1145 (0.4796) acc1: 99.7396 (76.8982) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1354 data: 0.0434 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.2017 (0.4695) acc1: 96.0938 (77.6112) acc5: 100.0000 (100.0000) acc1_neg: 68.229 acc5_neg: 100.000 time: 0.1584 data: 0.0644 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.6320 (0.5099) acc1: 19.5312 (74.9752) acc5: 100.0000 (100.0000) acc1_neg: 17.708 acc5_neg: 100.000 time: 0.1853 data: 0.0914 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 99.310 acc5_pos: 100.000 loss: 1.3769 (0.5212) acc1: 18.7500 (74.3260) acc5: 100.0000 (100.0000) acc1_neg: 20.233 acc5_neg: 100.000 time: 0.1850 data: 0.0893 max mem: 3532 +Test: Total time: 0:00:44 (0.1917 s / it) +* Overall Acc@1 74.326 Acc@5 100.000 + target == real Acc@1 66.820 Acc@5 100.000 + target == fake Acc@1 76.441 Acc@5 100.000 + loss 0.521 + PR-AUC 0.9459 ROC-AUC 0.8279 EER 27.69% (thr=0.5894) +Accuracy of the network on the 89729 test images: 74.3% +[rank0]:[W726 04:06:09.662381188 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:06:28.680000 1176340 torch/distributed/run.py:766] +W0726 04:06:28.680000 1176340 torch/distributed/run.py:766] ***************************************** +W0726 04:06:28.680000 1176340 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:06:28.680000 1176340 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:06:57.240469601 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:06:57.348418490 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:06:57.693236978 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:06:57.900471583 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +| distributed init (rank 6): env:// +[rank0]:[W726 04:06:58.925259142 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:06:58.927143326 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:06:58.936141806 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W726 04:06:58.936152427 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n22_seed6_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed6_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:53:20 acc1_pos: 8.073 acc5_pos: 100.000 loss: 1.3984 (1.3984) acc1: 8.0729 (8.0729) acc5: 100.0000 (100.0000) time: 7.3735 data: 5.7699 max mem: 3531 +Test: [ 10/2387] eta: 0:30:02 acc1_pos: 4.427 acc5_pos: 100.000 loss: 1.4226 (1.4452) acc1: 9.1146 (10.1799) acc5: 100.0000 (100.0000) time: 0.7582 data: 0.5249 max mem: 3531 +Test: [ 20/2387] eta: 0:17:32 acc1_pos: 4.948 acc5_pos: 100.000 loss: 1.4226 (1.4296) acc1: 9.1146 (9.6354) acc5: 100.0000 (100.0000) time: 0.0983 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:14:45 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.4909 (1.4681) acc1: 5.7292 (8.8626) acc5: 100.0000 (100.0000) time: 0.1650 data: 0.0667 max mem: 3531 +Test: [ 40/2387] eta: 0:13:41 acc1_pos: 5.990 acc5_pos: 100.000 loss: 1.6341 (1.5159) acc1: 2.3438 (7.2853) acc5: 100.0000 (100.0000) time: 0.2508 data: 0.1548 max mem: 3531 +Test: [ 50/2387] eta: 0:12:28 acc1_pos: 23.177 acc5_pos: 100.000 loss: 1.5740 (1.4961) acc1: 3.9062 (7.7359) acc5: 100.0000 (100.0000) time: 0.2348 data: 0.1409 max mem: 3531 +Test: [ 60/2387] eta: 0:11:54 acc1_pos: 25.260 acc5_pos: 100.000 loss: 1.4027 (1.4854) acc1: 7.2917 (8.1156) acc5: 100.0000 (100.0000) time: 0.2190 data: 0.1257 max mem: 3531 +Test: [ 70/2387] eta: 0:11:47 acc1_pos: 22.396 acc5_pos: 100.000 loss: 1.4891 (1.4992) acc1: 6.7708 (8.0326) acc5: 100.0000 (100.0000) time: 0.2665 data: 0.1722 max mem: 3531 +Test: [ 80/2387] eta: 0:11:27 acc1_pos: 16.146 acc5_pos: 100.000 loss: 1.5261 (1.4967) acc1: 3.9062 (7.8382) acc5: 100.0000 (100.0000) time: 0.2699 data: 0.1748 max mem: 3531 +Test: [ 90/2387] eta: 0:11:07 acc1_pos: 5.208 acc5_pos: 100.000 loss: 1.4589 (1.4877) acc1: 7.0312 (8.1330) acc5: 100.0000 (100.0000) time: 0.2384 data: 0.1404 max mem: 3531 +Test: [ 100/2387] eta: 0:10:52 acc1_pos: 5.469 acc5_pos: 100.000 loss: 1.4410 (1.4885) acc1: 6.7708 (8.0291) acc5: 100.0000 (100.0000) time: 0.2349 data: 0.1375 max mem: 3531 +Test: [ 110/2387] eta: 0:10:48 acc1_pos: 13.281 acc5_pos: 100.000 loss: 1.4410 (1.4829) acc1: 7.0312 (8.0706) acc5: 100.0000 (100.0000) time: 0.2599 data: 0.1635 max mem: 3531 +Test: [ 120/2387] eta: 0:10:48 acc1_pos: 13.542 acc5_pos: 100.000 loss: 1.4267 (1.4793) acc1: 8.8542 (8.3032) acc5: 100.0000 (100.0000) time: 0.2899 data: 0.1913 max mem: 3531 +Test: [ 130/2387] eta: 0:10:34 acc1_pos: 15.104 acc5_pos: 100.000 loss: 1.4560 (1.4788) acc1: 8.8542 (8.2618) acc5: 100.0000 (100.0000) time: 0.2599 data: 0.1613 max mem: 3531 +Test: [ 140/2387] eta: 0:10:26 acc1_pos: 8.073 acc5_pos: 100.000 loss: 1.4510 (1.4773) acc1: 8.0729 (8.2465) acc5: 100.0000 (100.0000) time: 0.2349 data: 0.1364 max mem: 3531 +Test: [ 150/2387] eta: 0:10:16 acc1_pos: 0.260 acc5_pos: 100.000 loss: 1.4255 (1.4719) acc1: 8.3333 (8.3471) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1428 max mem: 3531 +Test: [ 160/2387] eta: 0:10:14 acc1_pos: 11.872 acc5_pos: 100.000 loss: 1.3812 (1.4432) acc1: 9.1146 (10.2582) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2549 data: 0.1573 max mem: 3531 +Test: [ 170/2387] eta: 0:09:54 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1830 (1.3693) acc1: 98.1771 (15.4285) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2099 data: 0.1130 max mem: 3531 +Test: [ 180/2387] eta: 0:09:49 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1789 (1.3033) acc1: 98.6979 (20.0219) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1949 data: 0.0969 max mem: 3531 +Test: [ 190/2387] eta: 0:09:38 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1751 (1.2442) acc1: 98.9583 (24.1560) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2249 data: 0.1227 max mem: 3531 +Test: [ 200/2387] eta: 0:09:29 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1718 (1.1909) acc1: 98.9583 (27.8685) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1998 data: 0.1011 max mem: 3531 +Test: [ 210/2387] eta: 0:09:20 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1764 (1.1430) acc1: 98.6979 (31.2253) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1999 data: 0.1037 max mem: 3532 +Test: [ 220/2387] eta: 0:09:07 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1764 (1.0992) acc1: 98.6979 (34.2854) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1749 data: 0.0791 max mem: 3532 +Test: [ 230/2387] eta: 0:08:57 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1765 (1.0593) acc1: 98.6979 (37.0694) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1599 data: 0.0656 max mem: 3532 +Test: [ 240/2387] eta: 0:08:52 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1743 (1.0225) acc1: 98.9583 (39.6449) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1949 data: 0.0994 max mem: 3532 +Test: [ 250/2387] eta: 0:08:52 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1658 (0.9855) acc1: 99.4792 (42.0443) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2499 data: 0.1551 max mem: 3532 +Test: [ 260/2387] eta: 0:08:42 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0834 (0.9509) acc1: 100.0000 (44.2648) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2199 data: 0.1270 max mem: 3532 +Test: [ 270/2387] eta: 0:08:33 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0818 (0.9188) acc1: 100.0000 (46.3205) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1618 data: 0.0667 max mem: 3532 +Test: [ 280/2387] eta: 0:08:27 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0822 (0.8892) acc1: 100.0000 (48.2308) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1811 data: 0.0867 max mem: 3532 +Test: [ 290/2387] eta: 0:08:22 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0812 (0.8614) acc1: 100.0000 (50.0098) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1997 data: 0.1083 max mem: 3532 +Test: [ 300/2387] eta: 0:08:18 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0804 (0.8355) acc1: 100.0000 (51.6706) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2086 data: 0.1161 max mem: 3532 +Test: [ 310/2387] eta: 0:08:14 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0841 (0.8128) acc1: 100.0000 (53.2054) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2149 data: 0.1222 max mem: 3532 +Test: [ 320/2387] eta: 0:08:10 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1728 (0.7930) acc1: 99.7396 (54.6559) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2108 data: 0.1169 max mem: 3532 +Test: [ 330/2387] eta: 0:08:05 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1740 (0.7743) acc1: 99.7396 (56.0195) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2057 data: 0.1104 max mem: 3532 +Test: [ 340/2387] eta: 0:08:01 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1737 (0.7567) acc1: 99.7396 (57.3062) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2045 data: 0.1099 max mem: 3532 +Test: [ 350/2387] eta: 0:07:58 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1701 (0.7380) acc1: 100.0000 (58.5218) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2124 data: 0.1168 max mem: 3532 +Test: [ 360/2387] eta: 0:07:53 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1127 (0.7216) acc1: 99.7396 (59.6563) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2044 data: 0.1087 max mem: 3532 +Test: [ 370/2387] eta: 0:07:48 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1523 (0.7063) acc1: 99.4792 (60.7297) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1848 data: 0.0896 max mem: 3532 +Test: [ 380/2387] eta: 0:07:43 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1525 (0.6918) acc1: 99.4792 (61.7386) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1859 data: 0.0918 max mem: 3532 +Test: [ 390/2387] eta: 0:07:38 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1534 (0.6780) acc1: 99.2188 (62.6998) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1869 data: 0.0949 max mem: 3532 +Test: [ 400/2387] eta: 0:07:34 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1534 (0.6650) acc1: 99.2188 (63.6144) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1838 data: 0.0931 max mem: 3532 +Test: [ 410/2387] eta: 0:07:29 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1526 (0.6525) acc1: 99.2188 (64.4794) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1829 data: 0.0909 max mem: 3532 +Test: [ 420/2387] eta: 0:07:25 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1526 (0.6407) acc1: 99.2188 (65.3071) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1799 data: 0.0870 max mem: 3532 +Test: [ 430/2387] eta: 0:07:19 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1538 (0.6294) acc1: 99.2188 (66.0999) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1659 data: 0.0733 max mem: 3532 +Test: [ 440/2387] eta: 0:07:15 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1501 (0.6180) acc1: 99.4792 (66.8627) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1655 data: 0.0716 max mem: 3533 +Test: [ 450/2387] eta: 0:07:11 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1109 (0.6068) acc1: 100.0000 (67.5975) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1816 data: 0.0889 max mem: 3533 +Test: [ 460/2387] eta: 0:07:07 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1109 (0.5960) acc1: 100.0000 (68.3003) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1793 data: 0.0861 max mem: 3533 +Test: [ 470/2387] eta: 0:07:03 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1118 (0.5857) acc1: 100.0000 (68.9734) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1823 data: 0.0884 max mem: 3533 +Test: [ 480/2387] eta: 0:07:00 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1123 (0.5759) acc1: 100.0000 (69.6184) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1905 data: 0.0977 max mem: 3533 +Test: [ 490/2387] eta: 0:06:56 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1119 (0.5664) acc1: 100.0000 (70.2372) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1803 data: 0.0859 max mem: 3533 +Test: [ 500/2387] eta: 0:06:52 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1114 (0.5574) acc1: 100.0000 (70.8302) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1716 data: 0.0780 max mem: 3533 +Test: [ 510/2387] eta: 0:06:49 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1117 (0.5487) acc1: 100.0000 (71.4011) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1865 data: 0.0925 max mem: 3533 +Test: [ 520/2387] eta: 0:06:46 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1250 (0.5406) acc1: 100.0000 (71.9500) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1936 data: 0.0997 max mem: 3533 +Test: [ 530/2387] eta: 0:06:43 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1234 (0.5328) acc1: 100.0000 (72.4782) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1928 data: 0.0995 max mem: 3533 +Test: [ 540/2387] eta: 0:06:40 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1220 (0.5252) acc1: 100.0000 (72.9869) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1947 data: 0.1020 max mem: 3533 +Test: [ 550/2387] eta: 0:06:37 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1222 (0.5178) acc1: 100.0000 (73.4772) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1922 data: 0.1013 max mem: 3533 +Test: [ 560/2387] eta: 0:06:34 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1226 (0.5108) acc1: 100.0000 (73.9495) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.1032 max mem: 3533 +Test: [ 570/2387] eta: 0:06:31 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1226 (0.5040) acc1: 100.0000 (74.4057) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1926 data: 0.0951 max mem: 3533 +Test: [ 580/2387] eta: 0:06:30 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1219 (0.4974) acc1: 100.0000 (74.8458) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2158 data: 0.1186 max mem: 3533 +Test: [ 590/2387] eta: 0:06:26 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1223 (0.4911) acc1: 100.0000 (75.2714) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2067 data: 0.1152 max mem: 3533 +Test: [ 600/2387] eta: 0:06:22 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1249 (0.4851) acc1: 100.0000 (75.6829) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1682 data: 0.0759 max mem: 3533 +Test: [ 610/2387] eta: 0:06:22 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1197 (0.4790) acc1: 100.0000 (76.0796) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2131 data: 0.1179 max mem: 3533 +Test: [ 620/2387] eta: 0:06:19 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1158 (0.4732) acc1: 100.0000 (76.4640) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2361 data: 0.1414 max mem: 3533 +Test: [ 630/2387] eta: 0:06:16 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1158 (0.4675) acc1: 100.0000 (76.8361) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1827 data: 0.0908 max mem: 3533 +Test: [ 640/2387] eta: 0:06:14 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1168 (0.4621) acc1: 100.0000 (77.1963) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1894 data: 0.0949 max mem: 3533 +Test: [ 650/2387] eta: 0:06:11 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1172 (0.4568) acc1: 100.0000 (77.5462) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2183 data: 0.1232 max mem: 3533 +Test: [ 660/2387] eta: 0:06:09 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1166 (0.4516) acc1: 100.0000 (77.8859) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2006 data: 0.1089 max mem: 3533 +Test: [ 670/2387] eta: 0:06:06 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1157 (0.4466) acc1: 100.0000 (78.2154) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1892 data: 0.0987 max mem: 3534 +Test: [ 680/2387] eta: 0:06:04 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1138 (0.4417) acc1: 100.0000 (78.5353) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1961 data: 0.1048 max mem: 3534 +Test: [ 690/2387] eta: 0:06:02 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1152 (0.4405) acc1: 100.0000 (78.6104) acc5: 100.0000 (100.0000) acc1_neg: 72.917 acc5_neg: 100.000 time: 0.2194 data: 0.1236 max mem: 3534 +Test: [ 700/2387] eta: 0:05:59 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.4466 (0.4406) acc1: 77.8646 (78.5927) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.2177 data: 0.1215 max mem: 3534 +Test: [ 710/2387] eta: 0:05:56 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.4478 (0.4408) acc1: 77.3438 (78.5759) acc5: 100.0000 (100.0000) acc1_neg: 78.646 acc5_neg: 100.000 time: 0.1799 data: 0.0869 max mem: 3534 +Test: [ 720/2387] eta: 0:05:54 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.4445 (0.4408) acc1: 77.8646 (78.5696) acc5: 100.0000 (100.0000) acc1_neg: 77.344 acc5_neg: 100.000 time: 0.1846 data: 0.0916 max mem: 3534 +Test: [ 730/2387] eta: 0:05:51 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.4455 (0.4410) acc1: 76.5625 (78.5397) acc5: 100.0000 (100.0000) acc1_neg: 76.042 acc5_neg: 100.000 time: 0.2057 data: 0.1095 max mem: 3534 +Test: [ 740/2387] eta: 0:05:49 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.4467 (0.4411) acc1: 76.5625 (78.5323) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.2007 data: 0.1054 max mem: 3534 +Test: [ 750/2387] eta: 0:05:47 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.4463 (0.4411) acc1: 78.3854 (78.5290) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.2015 data: 0.1091 max mem: 3534 +Test: [ 760/2387] eta: 0:05:44 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.4342 (0.4404) acc1: 78.1250 (78.5736) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.2002 data: 0.1073 max mem: 3534 +Test: [ 770/2387] eta: 0:05:42 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.2928 (0.4376) acc1: 88.8021 (78.7634) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1953 data: 0.1014 max mem: 3534 +Test: [ 780/2387] eta: 0:05:39 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1649 (0.4341) acc1: 97.9167 (79.0110) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1956 data: 0.1016 max mem: 3534 +Test: [ 790/2387] eta: 0:05:38 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1398 (0.4307) acc1: 98.9583 (79.2440) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.2124 data: 0.1204 max mem: 3534 +Test: [ 800/2387] eta: 0:05:35 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1398 (0.4270) acc1: 99.4792 (79.4954) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2179 data: 0.1241 max mem: 3534 +Test: [ 810/2387] eta: 0:05:33 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1245 (0.4233) acc1: 99.7396 (79.7405) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2070 data: 0.1119 max mem: 3534 +Test: [ 820/2387] eta: 0:05:31 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1235 (0.4198) acc1: 100.0000 (79.9781) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1949 data: 0.1023 max mem: 3534 +Test: [ 830/2387] eta: 0:05:28 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1406 (0.4170) acc1: 99.4792 (80.1911) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1932 data: 0.0937 max mem: 3534 +Test: [ 840/2387] eta: 0:05:25 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1584 (0.4139) acc1: 99.4792 (80.4189) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1871 data: 0.0868 max mem: 3534 +Test: [ 850/2387] eta: 0:05:23 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1587 (0.4109) acc1: 99.4792 (80.6371) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1765 data: 0.0846 max mem: 3534 +Test: [ 860/2387] eta: 0:05:20 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1528 (0.4079) acc1: 99.7396 (80.8577) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1849 data: 0.0928 max mem: 3534 +Test: [ 870/2387] eta: 0:05:17 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1528 (0.4049) acc1: 99.7396 (81.0727) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1786 data: 0.0868 max mem: 3534 +Test: [ 880/2387] eta: 0:05:15 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1541 (0.4021) acc1: 99.7396 (81.2828) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1731 data: 0.0818 max mem: 3534 +Test: [ 890/2387] eta: 0:05:12 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1249 (0.3989) acc1: 100.0000 (81.4923) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1810 data: 0.0901 max mem: 3534 +Test: [ 900/2387] eta: 0:05:10 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1364 (0.4034) acc1: 99.7396 (81.1780) acc5: 100.0000 (100.0000) acc1_neg: 45.573 acc5_neg: 100.000 time: 0.1869 data: 0.0965 max mem: 3535 +Test: [ 910/2387] eta: 0:05:08 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.9668 (0.4097) acc1: 41.9271 (80.7386) acc5: 100.0000 (100.0000) acc1_neg: 41.927 acc5_neg: 100.000 time: 0.1975 data: 0.1065 max mem: 3535 +Test: [ 920/2387] eta: 0:05:05 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.9827 (0.4159) acc1: 41.6667 (80.3065) acc5: 100.0000 (100.0000) acc1_neg: 35.677 acc5_neg: 100.000 time: 0.1926 data: 0.1000 max mem: 3535 +Test: [ 930/2387] eta: 0:05:03 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.9711 (0.4219) acc1: 41.9271 (79.8951) acc5: 100.0000 (100.0000) acc1_neg: 36.458 acc5_neg: 100.000 time: 0.1775 data: 0.0842 max mem: 3535 +Test: [ 940/2387] eta: 0:05:00 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.9751 (0.4278) acc1: 41.1458 (79.4841) acc5: 100.0000 (100.0000) acc1_neg: 42.969 acc5_neg: 100.000 time: 0.1774 data: 0.0857 max mem: 3535 +Test: [ 950/2387] eta: 0:04:57 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.9677 (0.4335) acc1: 40.3646 (79.0782) acc5: 100.0000 (100.0000) acc1_neg: 44.010 acc5_neg: 100.000 time: 0.1754 data: 0.0852 max mem: 3535 +Test: [ 960/2387] eta: 0:04:55 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.9628 (0.4389) acc1: 40.6250 (78.6895) acc5: 100.0000 (100.0000) acc1_neg: 44.792 acc5_neg: 100.000 time: 0.1874 data: 0.0963 max mem: 3535 +Test: [ 970/2387] eta: 0:04:53 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.9675 (0.4446) acc1: 40.1042 (78.2932) acc5: 100.0000 (100.0000) acc1_neg: 39.844 acc5_neg: 100.000 time: 0.1971 data: 0.1035 max mem: 3535 +Test: [ 980/2387] eta: 0:04:50 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.9403 (0.4445) acc1: 41.1458 (78.2883) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.1767 data: 0.0814 max mem: 3535 +Test: [ 990/2387] eta: 0:04:48 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.2077 (0.4422) acc1: 95.5729 (78.4648) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1766 data: 0.0833 max mem: 3535 +Test: [1000/2387] eta: 0:04:46 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.2077 (0.4401) acc1: 97.1354 (78.6367) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2081 data: 0.1145 max mem: 3535 +Test: [1010/2387] eta: 0:04:44 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.2155 (0.4378) acc1: 96.8750 (78.8169) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1980 data: 0.1022 max mem: 3535 +Test: [1020/2387] eta: 0:04:42 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1908 (0.4355) acc1: 98.1771 (78.9950) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2119 data: 0.1170 max mem: 3535 +Test: [1030/2387] eta: 0:04:40 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1587 (0.4327) acc1: 99.2188 (79.1917) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2424 data: 0.1483 max mem: 3535 +Test: [1040/2387] eta: 0:04:38 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1397 (0.4299) acc1: 99.7396 (79.3888) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2039 data: 0.1118 max mem: 3535 +Test: [1050/2387] eta: 0:04:35 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1307 (0.4270) acc1: 100.0000 (79.5810) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1688 data: 0.0779 max mem: 3535 +Test: [1060/2387] eta: 0:04:33 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1305 (0.4243) acc1: 99.7396 (79.7712) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1898 data: 0.0940 max mem: 3535 +Test: [1070/2387] eta: 0:04:31 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1337 (0.4216) acc1: 99.7396 (79.9569) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2083 data: 0.1120 max mem: 3535 +Test: [1080/2387] eta: 0:04:29 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1413 (0.4192) acc1: 99.7396 (80.1317) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1902 data: 0.0986 max mem: 3535 +Test: [1090/2387] eta: 0:04:27 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1530 (0.4169) acc1: 99.7396 (80.3048) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1917 data: 0.0972 max mem: 3535 +Test: [1100/2387] eta: 0:04:25 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1606 (0.4146) acc1: 99.7396 (80.4749) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2037 data: 0.1089 max mem: 3535 +Test: [1110/2387] eta: 0:04:23 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1621 (0.4125) acc1: 99.2188 (80.6288) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2016 data: 0.1095 max mem: 3535 +Test: [1120/2387] eta: 0:04:21 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1653 (0.4103) acc1: 99.2188 (80.7947) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2062 data: 0.1139 max mem: 3536 +Test: [1130/2387] eta: 0:04:18 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1515 (0.4080) acc1: 99.4792 (80.9617) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1815 data: 0.0884 max mem: 3536 +Test: [1140/2387] eta: 0:04:16 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1556 (0.4059) acc1: 99.7396 (81.1229) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1636 data: 0.0687 max mem: 3536 +Test: [1150/2387] eta: 0:04:13 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1552 (0.4037) acc1: 99.7396 (81.2767) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0847 max mem: 3536 +Test: [1160/2387] eta: 0:04:11 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1279 (0.4012) acc1: 100.0000 (81.4362) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1914 data: 0.0978 max mem: 3536 +Test: [1170/2387] eta: 0:04:09 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1145 (0.3988) acc1: 100.0000 (81.5945) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2076 data: 0.1150 max mem: 3536 +Test: [1180/2387] eta: 0:04:08 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1179 (0.3964) acc1: 100.0000 (81.7486) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2384 data: 0.1464 max mem: 3536 +Test: [1190/2387] eta: 0:04:05 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1146 (0.3940) acc1: 100.0000 (81.9018) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2096 data: 0.1165 max mem: 3536 +Test: [1200/2387] eta: 0:04:02 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1020 (0.3916) acc1: 100.0000 (82.0525) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1435 data: 0.0477 max mem: 3536 +Test: [1210/2387] eta: 0:04:00 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0961 (0.3892) acc1: 100.0000 (82.2007) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1581 data: 0.0642 max mem: 3536 +Test: [1220/2387] eta: 0:03:58 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0917 (0.3867) acc1: 100.0000 (82.3461) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1863 data: 0.0918 max mem: 3536 +Test: [1230/2387] eta: 0:03:56 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0834 (0.3842) acc1: 100.0000 (82.4895) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1754 data: 0.0806 max mem: 3536 +Test: [1240/2387] eta: 0:03:54 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0832 (0.3818) acc1: 100.0000 (82.6306) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1867 data: 0.0949 max mem: 3536 +Test: [1250/2387] eta: 0:03:51 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0831 (0.3794) acc1: 100.0000 (82.7694) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1987 data: 0.1068 max mem: 3536 +Test: [1260/2387] eta: 0:03:49 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0815 (0.3771) acc1: 100.0000 (82.9061) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1930 data: 0.1004 max mem: 3536 +Test: [1270/2387] eta: 0:03:47 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0800 (0.3747) acc1: 100.0000 (83.0405) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2049 data: 0.1127 max mem: 3536 +Test: [1280/2387] eta: 0:03:45 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0826 (0.3725) acc1: 100.0000 (83.1711) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2125 data: 0.1208 max mem: 3536 +Test: [1290/2387] eta: 0:03:43 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0940 (0.3704) acc1: 100.0000 (83.2958) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2080 data: 0.1163 max mem: 3536 +Test: [1300/2387] eta: 0:03:41 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0962 (0.3684) acc1: 100.0000 (83.4210) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1926 data: 0.1015 max mem: 3536 +Test: [1310/2387] eta: 0:03:39 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0977 (0.3664) acc1: 100.0000 (83.5439) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1899 data: 0.0982 max mem: 3536 +Test: [1320/2387] eta: 0:03:37 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1004 (0.3644) acc1: 100.0000 (83.6643) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1905 data: 0.0983 max mem: 3536 +Test: [1330/2387] eta: 0:03:35 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1023 (0.3625) acc1: 100.0000 (83.7831) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1857 data: 0.0933 max mem: 3536 +Test: [1340/2387] eta: 0:03:33 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1209 (0.3615) acc1: 99.7396 (83.8501) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2061 data: 0.1119 max mem: 3536 +Test: [1350/2387] eta: 0:03:31 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0708 (0.3593) acc1: 100.0000 (83.9696) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2060 data: 0.1118 max mem: 3537 +Test: [1360/2387] eta: 0:03:29 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0667 (0.3572) acc1: 100.0000 (84.0874) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1927 data: 0.0998 max mem: 3537 +Test: [1370/2387] eta: 0:03:26 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0647 (0.3551) acc1: 100.0000 (84.2035) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1976 data: 0.1020 max mem: 3537 +Test: [1380/2387] eta: 0:03:24 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0670 (0.3530) acc1: 100.0000 (84.3179) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1910 data: 0.0959 max mem: 3537 +Test: [1390/2387] eta: 0:03:22 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0669 (0.3509) acc1: 100.0000 (84.4306) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1823 data: 0.0887 max mem: 3537 +Test: [1400/2387] eta: 0:03:20 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0655 (0.3489) acc1: 100.0000 (84.5417) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1913 data: 0.0958 max mem: 3537 +Test: [1410/2387] eta: 0:03:18 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0652 (0.3473) acc1: 100.0000 (84.6443) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2067 data: 0.1124 max mem: 3537 +Test: [1420/2387] eta: 0:03:16 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1381 (0.3459) acc1: 99.7396 (84.7373) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2023 data: 0.1099 max mem: 3537 +Test: [1430/2387] eta: 0:03:14 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1490 (0.3445) acc1: 99.7396 (84.8391) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1868 data: 0.0936 max mem: 3537 +Test: [1440/2387] eta: 0:03:12 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1336 (0.3430) acc1: 99.7396 (84.9432) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1758 data: 0.0832 max mem: 3537 +Test: [1450/2387] eta: 0:03:09 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1248 (0.3416) acc1: 100.0000 (85.0428) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1712 data: 0.0796 max mem: 3537 +Test: [1460/2387] eta: 0:03:07 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1215 (0.3401) acc1: 99.7396 (85.1420) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1819 data: 0.0890 max mem: 3537 +Test: [1470/2387] eta: 0:03:05 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1084 (0.3385) acc1: 100.0000 (85.2426) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1860 data: 0.0937 max mem: 3537 +Test: [1480/2387] eta: 0:03:03 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1052 (0.3369) acc1: 100.0000 (85.3400) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1882 data: 0.0975 max mem: 3537 +Test: [1490/2387] eta: 0:03:01 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1065 (0.3354) acc1: 100.0000 (85.4382) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1838 data: 0.0917 max mem: 3537 +Test: [1500/2387] eta: 0:02:59 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1037 (0.3339) acc1: 100.0000 (85.5340) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1739 data: 0.0816 max mem: 3537 +Test: [1510/2387] eta: 0:02:56 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1007 (0.3323) acc1: 100.0000 (85.6297) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1720 data: 0.0814 max mem: 3537 +Test: [1520/2387] eta: 0:02:54 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1137 (0.3310) acc1: 100.0000 (85.7207) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1708 data: 0.0797 max mem: 3537 +Test: [1530/2387] eta: 0:02:52 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1356 (0.3298) acc1: 99.7396 (85.8116) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1728 data: 0.0795 max mem: 3537 +Test: [1540/2387] eta: 0:02:50 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1370 (0.3286) acc1: 100.0000 (85.9020) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1700 data: 0.0767 max mem: 3537 +Test: [1550/2387] eta: 0:02:48 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1338 (0.3273) acc1: 100.0000 (85.9916) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1668 data: 0.0752 max mem: 3537 +Test: [1560/2387] eta: 0:02:45 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1252 (0.3261) acc1: 99.7396 (86.0796) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1707 data: 0.0767 max mem: 3537 +Test: [1570/2387] eta: 0:02:43 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1134 (0.3247) acc1: 100.0000 (86.1682) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1779 data: 0.0841 max mem: 3537 +Test: [1580/2387] eta: 0:02:41 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1154 (0.3235) acc1: 100.0000 (86.2526) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1685 data: 0.0771 max mem: 3538 +Test: [1590/2387] eta: 0:02:39 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1344 (0.3223) acc1: 99.7396 (86.3354) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1632 data: 0.0715 max mem: 3538 +Test: [1600/2387] eta: 0:02:37 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1323 (0.3213) acc1: 99.7396 (86.4094) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1742 data: 0.0815 max mem: 3538 +Test: [1610/2387] eta: 0:02:35 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1239 (0.3201) acc1: 99.7396 (86.4931) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1720 data: 0.0775 max mem: 3538 +Test: [1620/2387] eta: 0:02:33 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1239 (0.3189) acc1: 100.0000 (86.5729) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1660 data: 0.0722 max mem: 3538 +Test: [1630/2387] eta: 0:02:31 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1335 (0.3178) acc1: 99.4792 (86.6487) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1867 data: 0.0953 max mem: 3538 +Test: [1640/2387] eta: 0:02:29 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1345 (0.3168) acc1: 99.4792 (86.7213) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2005 data: 0.1067 max mem: 3538 +Test: [1650/2387] eta: 0:02:27 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1345 (0.3157) acc1: 99.4792 (86.7935) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2027 data: 0.1065 max mem: 3538 +Test: [1660/2387] eta: 0:02:25 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1330 (0.3147) acc1: 99.4792 (86.8691) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2064 data: 0.1122 max mem: 3538 +Test: [1670/2387] eta: 0:02:23 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1286 (0.3136) acc1: 99.7396 (86.9429) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2025 data: 0.1083 max mem: 3538 +Test: [1680/2387] eta: 0:02:21 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1286 (0.3126) acc1: 99.7396 (87.0112) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2120 data: 0.1152 max mem: 3538 +Test: [1690/2387] eta: 0:02:19 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1388 (0.3117) acc1: 99.4792 (87.0767) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2211 data: 0.1248 max mem: 3538 +Test: [1700/2387] eta: 0:02:17 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1388 (0.3106) acc1: 99.4792 (87.1486) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2044 data: 0.1095 max mem: 3538 +Test: [1710/2387] eta: 0:02:15 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1033 (0.3093) acc1: 100.0000 (87.2230) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1967 data: 0.1026 max mem: 3538 +Test: [1720/2387] eta: 0:02:13 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0924 (0.3081) acc1: 100.0000 (87.2969) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2005 data: 0.1091 max mem: 3538 +Test: [1730/2387] eta: 0:02:11 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0936 (0.3069) acc1: 100.0000 (87.3697) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1994 data: 0.1082 max mem: 3538 +Test: [1740/2387] eta: 0:02:09 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0966 (0.3056) acc1: 100.0000 (87.4417) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1950 data: 0.1035 max mem: 3538 +Test: [1750/2387] eta: 0:02:07 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0911 (0.3044) acc1: 100.0000 (87.5131) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1929 data: 0.1014 max mem: 3538 +Test: [1760/2387] eta: 0:02:05 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0878 (0.3032) acc1: 100.0000 (87.5840) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1856 data: 0.0944 max mem: 3538 +Test: [1770/2387] eta: 0:02:03 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0915 (0.3021) acc1: 100.0000 (87.6506) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1832 data: 0.0907 max mem: 3538 +Test: [1780/2387] eta: 0:02:01 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1014 (0.3009) acc1: 100.0000 (87.7199) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1798 data: 0.0875 max mem: 3538 +Test: [1790/2387] eta: 0:01:59 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0965 (0.2998) acc1: 100.0000 (87.7882) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1796 data: 0.0888 max mem: 3538 +Test: [1800/2387] eta: 0:01:56 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0951 (0.2987) acc1: 100.0000 (87.8556) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1835 data: 0.0910 max mem: 3538 +Test: [1810/2387] eta: 0:01:54 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0929 (0.2975) acc1: 100.0000 (87.9220) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1713 data: 0.0784 max mem: 3539 +Test: [1820/2387] eta: 0:01:52 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0888 (0.2964) acc1: 100.0000 (87.9884) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1669 data: 0.0743 max mem: 3539 +Test: [1830/2387] eta: 0:01:50 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0926 (0.2953) acc1: 100.0000 (88.0540) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1698 data: 0.0777 max mem: 3539 +Test: [1840/2387] eta: 0:01:48 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1249 (0.2944) acc1: 100.0000 (88.1187) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1741 data: 0.0815 max mem: 3539 +Test: [1850/2387] eta: 0:01:46 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1273 (0.2935) acc1: 100.0000 (88.1826) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1729 data: 0.0801 max mem: 3539 +Test: [1860/2387] eta: 0:01:44 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1264 (0.2926) acc1: 100.0000 (88.2461) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1648 data: 0.0736 max mem: 3539 +Test: [1870/2387] eta: 0:01:42 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1258 (0.2917) acc1: 100.0000 (88.3089) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1796 data: 0.0855 max mem: 3539 +Test: [1880/2387] eta: 0:01:40 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1261 (0.2909) acc1: 100.0000 (88.3711) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1888 data: 0.0932 max mem: 3539 +Test: [1890/2387] eta: 0:01:38 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1264 (0.2900) acc1: 100.0000 (88.4326) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1903 data: 0.0967 max mem: 3539 +Test: [1900/2387] eta: 0:01:36 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1277 (0.2892) acc1: 100.0000 (88.4934) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1871 data: 0.0938 max mem: 3539 +Test: [1910/2387] eta: 0:01:34 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1303 (0.2883) acc1: 100.0000 (88.5537) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1795 data: 0.0859 max mem: 3539 +Test: [1920/2387] eta: 0:01:32 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1366 (0.2876) acc1: 100.0000 (88.6040) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1738 data: 0.0818 max mem: 3539 +Test: [1930/2387] eta: 0:01:30 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1102 (0.2867) acc1: 100.0000 (88.6630) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1757 data: 0.0844 max mem: 3539 +Test: [1940/2387] eta: 0:01:28 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1102 (0.2860) acc1: 100.0000 (88.6863) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1894 data: 0.0974 max mem: 3539 +Test: [1950/2387] eta: 0:01:26 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1209 (0.2852) acc1: 100.0000 (88.7442) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1919 data: 0.1000 max mem: 3539 +Test: [1960/2387] eta: 0:01:24 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1207 (0.2844) acc1: 100.0000 (88.8000) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1847 data: 0.0930 max mem: 3539 +Test: [1970/2387] eta: 0:01:22 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1284 (0.2838) acc1: 100.0000 (88.8465) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1695 data: 0.0766 max mem: 3539 +Test: [1980/2387] eta: 0:01:20 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1403 (0.2832) acc1: 99.2188 (88.8909) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1673 data: 0.0738 max mem: 3539 +Test: [1990/2387] eta: 0:01:18 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1399 (0.2825) acc1: 98.6979 (88.9411) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1739 data: 0.0824 max mem: 3539 +Test: [2000/2387] eta: 0:01:16 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1399 (0.2818) acc1: 98.9583 (88.9901) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1737 data: 0.0830 max mem: 3539 +Test: [2010/2387] eta: 0:01:14 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1406 (0.2811) acc1: 98.9583 (89.0392) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1796 data: 0.0886 max mem: 3539 +Test: [2020/2387] eta: 0:01:12 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1417 (0.2804) acc1: 98.9583 (89.0878) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1904 data: 0.0985 max mem: 3539 +Test: [2030/2387] eta: 0:01:10 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1413 (0.2797) acc1: 98.9583 (89.1365) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1904 data: 0.0973 max mem: 3540 +Test: [2040/2387] eta: 0:01:08 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1390 (0.2790) acc1: 98.9583 (89.1852) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1879 data: 0.0945 max mem: 3540 +Test: [2050/2387] eta: 0:01:06 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1373 (0.2783) acc1: 99.2188 (89.2348) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1837 data: 0.0917 max mem: 3540 +Test: [2060/2387] eta: 0:01:04 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1371 (0.2776) acc1: 99.2188 (89.2817) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1825 data: 0.0915 max mem: 3540 +Test: [2070/2387] eta: 0:01:02 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1347 (0.2769) acc1: 98.9583 (89.3298) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1945 data: 0.1013 max mem: 3540 +Test: [2080/2387] eta: 0:01:00 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1323 (0.2763) acc1: 99.2188 (89.3767) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2172 data: 0.1233 max mem: 3540 +Test: [2090/2387] eta: 0:00:58 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1334 (0.2756) acc1: 99.2188 (89.4233) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2062 data: 0.1125 max mem: 3540 +Test: [2100/2387] eta: 0:00:56 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1321 (0.2749) acc1: 99.2188 (89.4704) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1849 data: 0.0922 max mem: 3540 +Test: [2110/2387] eta: 0:00:54 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1321 (0.2742) acc1: 99.2188 (89.5165) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2036 data: 0.1129 max mem: 3540 +Test: [2120/2387] eta: 0:00:52 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1364 (0.2737) acc1: 99.2188 (89.5583) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.1017 max mem: 3540 +Test: [2130/2387] eta: 0:00:50 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1497 (0.2732) acc1: 99.4792 (89.6025) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1725 data: 0.0789 max mem: 3540 +Test: [2140/2387] eta: 0:00:48 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1583 (0.2727) acc1: 98.9583 (89.6375) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1911 data: 0.0971 max mem: 3540 +Test: [2150/2387] eta: 0:00:46 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1594 (0.2722) acc1: 99.2188 (89.6827) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1926 data: 0.0974 max mem: 3540 +Test: [2160/2387] eta: 0:00:44 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1408 (0.2716) acc1: 99.7396 (89.7269) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1937 data: 0.0994 max mem: 3540 +Test: [2170/2387] eta: 0:00:42 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1488 (0.2711) acc1: 99.7396 (89.7697) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2095 data: 0.1166 max mem: 3540 +Test: [2180/2387] eta: 0:00:40 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1866 (0.2727) acc1: 97.9167 (89.6556) acc5: 100.0000 (100.0000) acc1_neg: 17.448 acc5_neg: 100.000 time: 0.2025 data: 0.1115 max mem: 3540 +Test: [2190/2387] eta: 0:00:38 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.2552 (0.2732) acc1: 90.8854 (89.6218) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1959 data: 0.1023 max mem: 3540 +Test: [2200/2387] eta: 0:00:36 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1143 (0.2725) acc1: 99.2188 (89.6684) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1852 data: 0.0904 max mem: 3540 +Test: [2210/2387] eta: 0:00:34 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1046 (0.2717) acc1: 100.0000 (89.7150) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1889 data: 0.0962 max mem: 3540 +Test: [2220/2387] eta: 0:00:32 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1034 (0.2709) acc1: 100.0000 (89.7600) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1884 data: 0.0963 max mem: 3540 +Test: [2230/2387] eta: 0:00:30 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1031 (0.2702) acc1: 100.0000 (89.8055) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1781 data: 0.0866 max mem: 3540 +Test: [2240/2387] eta: 0:00:28 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0977 (0.2694) acc1: 100.0000 (89.8496) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1785 data: 0.0866 max mem: 3540 +Test: [2250/2387] eta: 0:00:26 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.0980 (0.2687) acc1: 100.0000 (89.8943) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1847 data: 0.0932 max mem: 3540 +Test: [2260/2387] eta: 0:00:24 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1115 (0.2681) acc1: 100.0000 (89.9390) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1618 data: 0.0715 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1246 (0.2675) acc1: 100.0000 (89.9833) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1451 data: 0.0551 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1246 (0.2670) acc1: 100.0000 (90.0164) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1538 data: 0.0624 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1281 (0.2663) acc1: 100.0000 (90.0598) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1432 data: 0.0510 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1277 (0.2657) acc1: 100.0000 (90.1025) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1405 data: 0.0490 max mem: 3541 +Test: [2310/2387] eta: 0:00:15 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1261 (0.2652) acc1: 100.0000 (90.1437) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1369 data: 0.0463 max mem: 3541 +Test: [2320/2387] eta: 0:00:13 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1325 (0.2647) acc1: 99.7396 (90.1792) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1329 data: 0.0426 max mem: 3541 +Test: [2330/2387] eta: 0:00:11 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1437 (0.2643) acc1: 99.7396 (90.2173) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1320 data: 0.0415 max mem: 3541 +Test: [2340/2387] eta: 0:00:09 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1409 (0.2638) acc1: 100.0000 (90.2566) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1297 data: 0.0397 max mem: 3541 +Test: [2350/2387] eta: 0:00:07 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1366 (0.2635) acc1: 99.7396 (90.2842) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1249 data: 0.0347 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1395 (0.2630) acc1: 99.7396 (90.3180) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1215 data: 0.0313 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1473 (0.2626) acc1: 100.0000 (90.3574) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1199 data: 0.0300 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1466 (0.2621) acc1: 99.4792 (90.3917) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1201 data: 0.0303 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 11.872 acc5_pos: 100.000 loss: 0.1666 (0.2632) acc1: 99.2188 (90.3359) acc5: 100.0000 (100.0000) acc1_neg: 34.884 acc5_neg: 100.000 time: 0.1150 data: 0.0280 max mem: 3541 +Test: Total time: 0:07:40 (0.1929 s / it) +* Overall Acc@1 90.336 Acc@5 100.000 + target == real Acc@1 8.241 Acc@5 100.000 + target == fake Acc@1 96.101 Acc@5 100.000 + loss 0.263 + PR-AUC 0.9876 ROC-AUC 0.8422 EER 22.42% (thr=0.8462) +Accuracy of the network on the 916267 test images: 90.3% +[rank0]:[W726 04:14:52.853239277 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:15:12.760000 1176974 torch/distributed/run.py:766] +W0726 04:15:12.760000 1176974 torch/distributed/run.py:766] ***************************************** +W0726 04:15:12.760000 1176974 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:15:12.760000 1176974 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:15:40.704625588 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:15:40.815959598 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:15:41.142879762 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:15:41.301015382 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:15:41.525208574 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:15:41.546885611 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:15:41.560036384 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank1]:[W726 04:15:41.560057972 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n22_seed6_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed6_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:01:34 acc1_pos: 8.073 acc5_pos: 100.000 loss: 1.3984 (1.3984) acc1: 8.0729 (8.0729) acc5: 100.0000 (100.0000) time: 7.1457 data: 5.8167 max mem: 3531 +Test: [ 10/517] eta: 0:06:10 acc1_pos: 4.427 acc5_pos: 100.000 loss: 1.4226 (1.4452) acc1: 9.1146 (10.1799) acc5: 100.0000 (100.0000) time: 0.7314 data: 0.5291 max mem: 3531 +Test: [ 20/517] eta: 0:03:33 acc1_pos: 4.948 acc5_pos: 100.000 loss: 1.4226 (1.4296) acc1: 9.1146 (9.6354) acc5: 100.0000 (100.0000) time: 0.0937 data: 0.0003 max mem: 3531 +Test: [ 30/517] eta: 0:02:57 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.4909 (1.4681) acc1: 5.7292 (8.8626) acc5: 100.0000 (100.0000) time: 0.1637 data: 0.0690 max mem: 3531 +Test: [ 40/517] eta: 0:02:39 acc1_pos: 5.990 acc5_pos: 100.000 loss: 1.6341 (1.5159) acc1: 2.3438 (7.2853) acc5: 100.0000 (100.0000) time: 0.2353 data: 0.1439 max mem: 3531 +Test: [ 50/517] eta: 0:02:27 acc1_pos: 8.257 acc5_pos: 100.000 loss: 1.5740 (1.4424) acc1: 3.9062 (11.5554) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2414 data: 0.1487 max mem: 3531 +Test: [ 60/517] eta: 0:02:19 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1984 (1.2344) acc1: 99.4792 (26.0331) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2445 data: 0.1496 max mem: 3531 +Test: [ 70/517] eta: 0:02:12 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1679 (1.0844) acc1: 100.0000 (36.4473) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2435 data: 0.1486 max mem: 3531 +Test: [ 80/517] eta: 0:02:07 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1752 (0.9731) acc1: 100.0000 (44.2837) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2464 data: 0.1527 max mem: 3531 +Test: [ 90/517] eta: 0:02:02 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1795 (0.8861) acc1: 100.0000 (50.3806) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2497 data: 0.1579 max mem: 3531 +Test: [100/517] eta: 0:01:57 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1747 (0.8150) acc1: 100.0000 (55.2883) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2434 data: 0.1506 max mem: 3531 +Test: [110/517] eta: 0:01:52 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1449 (0.7544) acc1: 100.0000 (59.3163) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2399 data: 0.1440 max mem: 3531 +Test: [120/517] eta: 0:01:48 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1582 (0.7064) acc1: 100.0000 (62.6743) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2399 data: 0.1435 max mem: 3531 +Test: [130/517] eta: 0:01:45 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1767 (0.6662) acc1: 100.0000 (65.5177) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2399 data: 0.1464 max mem: 3531 +Test: [140/517] eta: 0:01:41 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1754 (0.6311) acc1: 100.0000 (67.9632) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2349 data: 0.1403 max mem: 3531 +Test: [150/517] eta: 0:01:37 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1672 (0.6006) acc1: 100.0000 (70.0849) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2349 data: 0.1354 max mem: 3531 +Test: [160/517] eta: 0:01:34 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1708 (0.5742) acc1: 100.0000 (71.9397) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2464 data: 0.1499 max mem: 3531 +Test: [170/517] eta: 0:01:32 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1722 (0.5509) acc1: 100.0000 (73.5730) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1629 max mem: 3531 +Test: [180/517] eta: 0:01:29 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1617 (0.5292) acc1: 99.7396 (75.0201) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2583 data: 0.1639 max mem: 3531 +Test: [190/517] eta: 0:01:26 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1397 (0.5083) acc1: 100.0000 (76.3280) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2649 data: 0.1665 max mem: 3531 +Test: [200/517] eta: 0:01:23 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1368 (0.4902) acc1: 100.0000 (77.5057) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2501 data: 0.1516 max mem: 3531 +Test: [210/517] eta: 0:01:20 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1634 (0.4771) acc1: 99.7396 (78.4558) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2466 data: 0.1501 max mem: 3532 +Test: [220/517] eta: 0:01:18 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.2172 (0.4658) acc1: 97.1354 (79.2810) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2555 data: 0.1607 max mem: 3532 +Test: [230/517] eta: 0:01:14 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.2251 (0.4552) acc1: 97.3958 (80.1339) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2339 data: 0.1396 max mem: 3532 +Test: [240/517] eta: 0:01:11 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1910 (0.4431) acc1: 99.7396 (80.9561) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2192 data: 0.1223 max mem: 3532 +Test: [250/517] eta: 0:01:09 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1590 (0.4322) acc1: 100.0000 (81.7075) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2464 data: 0.1504 max mem: 3532 +Test: [260/517] eta: 0:01:06 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1644 (0.4221) acc1: 99.7396 (82.4004) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2615 data: 0.1690 max mem: 3532 +Test: [270/517] eta: 0:01:03 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1703 (0.4129) acc1: 99.7396 (83.0470) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2487 data: 0.1578 max mem: 3532 +Test: [280/517] eta: 0:01:01 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1662 (0.4043) acc1: 100.0000 (83.6438) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2435 data: 0.1515 max mem: 3532 +Test: [290/517] eta: 0:00:58 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1721 (0.3967) acc1: 100.0000 (84.2032) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2390 data: 0.1434 max mem: 3532 +Test: [300/517] eta: 0:00:56 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1754 (0.3894) acc1: 100.0000 (84.7271) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2947 data: 0.1963 max mem: 3532 +Test: [310/517] eta: 0:00:53 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1804 (0.3827) acc1: 100.0000 (85.2182) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2761 data: 0.1782 max mem: 3532 +Test: [320/517] eta: 0:00:50 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1796 (0.3763) acc1: 100.0000 (85.6763) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2249 data: 0.1292 max mem: 3532 +Test: [330/517] eta: 0:00:48 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1646 (0.3698) acc1: 100.0000 (86.1051) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2387 data: 0.1401 max mem: 3532 +Test: [340/517] eta: 0:00:45 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1534 (0.3634) acc1: 100.0000 (86.5103) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2349 data: 0.1343 max mem: 3532 +Test: [350/517] eta: 0:00:42 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1536 (0.3577) acc1: 100.0000 (86.8938) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2401 data: 0.1425 max mem: 3532 +Test: [360/517] eta: 0:00:40 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1577 (0.3522) acc1: 100.0000 (87.2547) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2399 data: 0.1424 max mem: 3532 +Test: [370/517] eta: 0:00:37 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1367 (0.3462) acc1: 100.0000 (87.5983) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2400 data: 0.1425 max mem: 3532 +Test: [380/517] eta: 0:00:34 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1275 (0.3406) acc1: 100.0000 (87.9238) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2399 data: 0.1425 max mem: 3532 +Test: [390/517] eta: 0:00:32 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1361 (0.3355) acc1: 100.0000 (88.2326) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2347 data: 0.1368 max mem: 3532 +Test: [400/517] eta: 0:00:29 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1367 (0.3304) acc1: 100.0000 (88.5261) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2309 data: 0.1337 max mem: 3532 +Test: [410/517] eta: 0:00:27 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1412 (0.3260) acc1: 100.0000 (88.8053) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2357 data: 0.1400 max mem: 3532 +Test: [420/517] eta: 0:00:24 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1473 (0.3218) acc1: 100.0000 (89.0712) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2303 data: 0.1353 max mem: 3532 +Test: [430/517] eta: 0:00:21 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1534 (0.3182) acc1: 100.0000 (89.3247) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2252 data: 0.1307 max mem: 3532 +Test: [440/517] eta: 0:00:19 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1527 (0.3142) acc1: 100.0000 (89.5668) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2337 data: 0.1412 max mem: 3533 +Test: [450/517] eta: 0:00:16 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1381 (0.3103) acc1: 100.0000 (89.7981) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2438 data: 0.1489 max mem: 3533 +Test: [460/517] eta: 0:00:14 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1400 (0.3069) acc1: 100.0000 (90.0194) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2313 data: 0.1335 max mem: 3533 +Test: [470/517] eta: 0:00:11 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1400 (0.3033) acc1: 100.0000 (90.2313) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2265 data: 0.1315 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1352 (0.2999) acc1: 100.0000 (90.4344) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2351 data: 0.1414 max mem: 3533 +Test: [490/517] eta: 0:00:06 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1468 (0.2973) acc1: 100.0000 (90.6266) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2397 data: 0.1465 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1762 (0.2949) acc1: 100.0000 (90.8121) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2334 data: 0.1422 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1784 (0.2926) acc1: 100.0000 (90.9756) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2149 data: 0.1236 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 8.257 acc5_pos: 100.000 loss: 0.1466 (0.2908) acc1: 100.0000 (91.0758) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2093 data: 0.1157 max mem: 3533 +Test: Total time: 0:02:08 (0.2478 s / it) +* Overall Acc@1 91.076 Acc@5 100.000 + target == real Acc@1 7.126 Acc@5 100.000 + target == fake Acc@1 99.785 Acc@5 100.000 + loss 0.291 + PR-AUC 0.9646 ROC-AUC 0.7779 EER 29.56% (thr=0.8398) +Accuracy of the network on the 198427 test images: 91.1% +[rank0]:[W726 04:18:05.280098506 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 04:18:25.000000 1177536 torch/distributed/run.py:766] +W0726 04:18:25.000000 1177536 torch/distributed/run.py:766] ***************************************** +W0726 04:18:25.000000 1177536 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:18:25.000000 1177536 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 04:18:53.869684545 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:18:54.415574019 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +NCCL version 2.26.2+cuda12.2 +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:18:54.518814719 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:18:54.569984744 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:18:54.812978362 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:18:54.873299872 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:18:55.933429434 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:18:55.953254294 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n22_seed6_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed6_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:39:40 acc1_pos: 31.250 acc5_pos: 100.000 loss: 0.9649 (0.9649) acc1: 31.2500 (31.2500) acc5: 100.0000 (100.0000) time: 6.1202 data: 4.5153 max mem: 3531 +Test: [ 10/389] eta: 0:04:04 acc1_pos: 76.823 acc5_pos: 100.000 loss: 0.8966 (0.8600) acc1: 38.5417 (40.8144) acc5: 100.0000 (100.0000) time: 0.6448 data: 0.4108 max mem: 3531 +Test: [ 20/389] eta: 0:02:22 acc1_pos: 4.688 acc5_pos: 100.000 loss: 0.8689 (0.8805) acc1: 29.9479 (36.6443) acc5: 100.0000 (100.0000) time: 0.0986 data: 0.0004 max mem: 3531 +Test: [ 30/389] eta: 0:02:01 acc1_pos: 17.708 acc5_pos: 100.000 loss: 1.0358 (0.9528) acc1: 11.4583 (28.6290) acc5: 100.0000 (100.0000) time: 0.1699 data: 0.0740 max mem: 3531 +Test: [ 40/389] eta: 0:01:50 acc1_pos: 7.552 acc5_pos: 100.000 loss: 1.0345 (0.9569) acc1: 12.5000 (27.7820) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1516 max mem: 3531 +Test: [ 50/389] eta: 0:01:41 acc1_pos: 65.104 acc5_pos: 100.000 loss: 0.8828 (0.9237) acc1: 28.9062 (31.2704) acc5: 100.0000 (100.0000) time: 0.2398 data: 0.1425 max mem: 3531 +Test: [ 60/389] eta: 0:01:34 acc1_pos: 22.656 acc5_pos: 100.000 loss: 0.6589 (0.8854) acc1: 60.4167 (35.6259) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1280 max mem: 3531 +Test: [ 70/389] eta: 0:01:29 acc1_pos: 57.812 acc5_pos: 100.000 loss: 0.6600 (0.8840) acc1: 59.3750 (35.9522) acc5: 100.0000 (100.0000) time: 0.2286 data: 0.1347 max mem: 3531 +Test: [ 80/389] eta: 0:01:23 acc1_pos: 78.125 acc5_pos: 100.000 loss: 0.7362 (0.8716) acc1: 49.4792 (38.2427) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1308 max mem: 3531 +Test: [ 90/389] eta: 0:01:16 acc1_pos: 34.115 acc5_pos: 100.000 loss: 0.5955 (0.8416) acc1: 55.4688 (41.0886) acc5: 100.0000 (100.0000) time: 0.1731 data: 0.0778 max mem: 3531 +Test: [100/389] eta: 0:01:13 acc1_pos: 41.406 acc5_pos: 100.000 loss: 0.8040 (0.8503) acc1: 42.7083 (40.4651) acc5: 100.0000 (100.0000) time: 0.1817 data: 0.0882 max mem: 3531 +Test: [110/389] eta: 0:01:08 acc1_pos: 99.740 acc5_pos: 100.000 loss: 0.8267 (0.8471) acc1: 39.3229 (40.9980) acc5: 100.0000 (100.0000) time: 0.1981 data: 0.1076 max mem: 3531 +Test: [120/389] eta: 0:01:04 acc1_pos: 64.844 acc5_pos: 100.000 loss: 0.7532 (0.8357) acc1: 49.4792 (42.0907) acc5: 100.0000 (100.0000) time: 0.1732 data: 0.0815 max mem: 3531 +Test: [130/389] eta: 0:01:01 acc1_pos: 42.945 acc5_pos: 100.000 loss: 0.6006 (0.8140) acc1: 64.8438 (44.3623) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1909 data: 0.0989 max mem: 3531 +Test: [140/389] eta: 0:00:58 acc1_pos: 42.945 acc5_pos: 100.000 loss: 0.4118 (0.7943) acc1: 84.6354 (46.6294) acc5: 100.0000 (100.0000) acc1_neg: 4.427 acc5_neg: 100.000 time: 0.1968 data: 0.1049 max mem: 3531 +Test: [150/389] eta: 0:00:55 acc1_pos: 42.945 acc5_pos: 100.000 loss: 0.3386 (0.7788) acc1: 88.8021 (48.0305) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1901 data: 0.0984 max mem: 3531 +Test: [160/389] eta: 0:00:52 acc1_pos: 42.945 acc5_pos: 100.000 loss: 0.3386 (0.7502) acc1: 88.8021 (50.7473) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1892 data: 0.0978 max mem: 3531 +Test: [170/389] eta: 0:00:49 acc1_pos: 42.945 acc5_pos: 100.000 loss: 0.2025 (0.7240) acc1: 99.7396 (52.9560) acc5: 100.0000 (100.0000) acc1_neg: 61.458 acc5_neg: 100.000 time: 0.2008 data: 0.1093 max mem: 3531 +Test: [180/389] eta: 0:00:46 acc1_pos: 42.945 acc5_pos: 100.000 loss: 0.2025 (0.7031) acc1: 97.3958 (54.9868) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1940 data: 0.1032 max mem: 3531 +Test: [190/389] eta: 0:00:44 acc1_pos: 42.945 acc5_pos: 100.000 loss: 0.3207 (0.6907) acc1: 87.2396 (56.1723) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.1817 data: 0.0896 max mem: 3531 +Test: [200/389] eta: 0:00:41 acc1_pos: 42.945 acc5_pos: 100.000 loss: 0.3745 (0.6756) acc1: 88.0208 (57.8540) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.1857 data: 0.0932 max mem: 3531 +Test: [210/389] eta: 0:00:39 acc1_pos: 42.945 acc5_pos: 100.000 loss: 0.2703 (0.6577) acc1: 95.8333 (59.5157) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1910 data: 0.0990 max mem: 3532 +Test: [220/389] eta: 0:00:36 acc1_pos: 42.945 acc5_pos: 100.000 loss: 0.2126 (0.6365) acc1: 99.7396 (61.2981) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1920 data: 0.1002 max mem: 3532 +Test: [230/389] eta: 0:00:34 acc1_pos: 42.945 acc5_pos: 100.000 loss: 0.3144 (0.6249) acc1: 96.8750 (62.4707) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.1783 data: 0.0867 max mem: 3532 +Test: [240/389] eta: 0:00:32 acc1_pos: 42.945 acc5_pos: 100.000 loss: 0.3472 (0.6146) acc1: 89.3229 (63.5136) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1823 data: 0.0906 max mem: 3532 +Test: [250/389] eta: 0:00:29 acc1_pos: 42.945 acc5_pos: 100.000 loss: 0.2828 (0.5995) acc1: 98.9583 (64.9454) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1869 data: 0.0958 max mem: 3532 +Test: [260/389] eta: 0:00:27 acc1_pos: 42.945 acc5_pos: 100.000 loss: 0.2647 (0.5914) acc1: 99.2188 (65.9014) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1761 data: 0.0834 max mem: 3532 +Test: [270/389] eta: 0:00:25 acc1_pos: 42.945 acc5_pos: 100.000 loss: 0.3716 (0.5849) acc1: 92.4479 (66.6676) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.1755 data: 0.0824 max mem: 3532 +Test: [280/389] eta: 0:00:22 acc1_pos: 42.945 acc5_pos: 100.000 loss: 0.4432 (0.5837) acc1: 82.2917 (66.7519) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1751 data: 0.0835 max mem: 3532 +Test: [290/389] eta: 0:00:20 acc1_pos: 42.945 acc5_pos: 100.000 loss: 0.4849 (0.5758) acc1: 80.4688 (67.4846) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1748 data: 0.0829 max mem: 3532 +Test: [300/389] eta: 0:00:18 acc1_pos: 42.945 acc5_pos: 100.000 loss: 0.5428 (0.5745) acc1: 79.9479 (67.8130) acc5: 100.0000 (100.0000) acc1_neg: 74.219 acc5_neg: 100.000 time: 0.1728 data: 0.0784 max mem: 3532 +Test: [310/389] eta: 0:00:16 acc1_pos: 42.945 acc5_pos: 100.000 loss: 0.5553 (0.5729) acc1: 77.0833 (68.1086) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1640 data: 0.0689 max mem: 3532 +Test: [320/389] eta: 0:00:14 acc1_pos: 42.945 acc5_pos: 100.000 loss: 0.3939 (0.5674) acc1: 89.8438 (68.7119) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1617 data: 0.0690 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 42.945 acc5_pos: 100.000 loss: 0.3081 (0.5600) acc1: 97.3958 (69.5690) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1599 data: 0.0681 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 42.945 acc5_pos: 100.000 loss: 0.2766 (0.5514) acc1: 99.7396 (70.3774) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1673 data: 0.0762 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 42.945 acc5_pos: 100.000 loss: 0.2582 (0.5607) acc1: 98.9583 (69.6937) acc5: 100.0000 (100.0000) acc1_neg: 1.823 acc5_neg: 100.000 time: 0.1560 data: 0.0658 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 42.945 acc5_pos: 100.000 loss: 0.8247 (0.5714) acc1: 40.3646 (68.6714) acc5: 100.0000 (100.0000) acc1_neg: 42.188 acc5_neg: 100.000 time: 0.1447 data: 0.0546 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 42.945 acc5_pos: 100.000 loss: 0.8027 (0.5664) acc1: 43.4896 (69.0483) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1436 data: 0.0536 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 42.945 acc5_pos: 100.000 loss: 0.3907 (0.5622) acc1: 91.9271 (69.4629) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1296 data: 0.0399 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 42.945 acc5_pos: 100.000 loss: 0.3736 (0.5575) acc1: 92.7083 (69.9427) acc5: 100.0000 (100.0000) acc1_neg: 90.710 acc5_neg: 100.000 time: 0.1309 data: 0.0379 max mem: 3532 +Test: Total time: 0:01:15 (0.1946 s / it) +* Overall Acc@1 69.943 Acc@5 100.000 + target == real Acc@1 42.015 Acc@5 100.000 + target == fake Acc@1 82.924 Acc@5 100.000 + loss 0.557 + PR-AUC 0.8689 ROC-AUC 0.7460 EER 31.96% (thr=0.6479) +Accuracy of the network on the 149358 test images: 69.9% +[rank0]:[W726 04:20:18.117473634 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:20:36.243000 1178095 torch/distributed/run.py:766] +W0726 04:20:36.243000 1178095 torch/distributed/run.py:766] ***************************************** +W0726 04:20:36.243000 1178095 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:20:36.243000 1178095 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:21:03.545036442 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:21:03.690971375 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:21:04.016368212 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +[rank0]:[W726 04:21:04.267107805 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:21:04.275170227 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:21:04.301915890 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W726 04:21:04.307204731 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:21:04.327744474 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n22_seed6_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed6_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:40:26 acc1_pos: 71.875 acc5_pos: 100.000 loss: 0.5506 (0.5506) acc1: 71.8750 (71.8750) acc5: 100.0000 (100.0000) time: 7.2882 data: 5.5255 max mem: 3531 +Test: [ 10/333] eta: 0:04:01 acc1_pos: 77.083 acc5_pos: 100.000 loss: 0.4777 (0.4528) acc1: 76.0417 (76.5152) acc5: 100.0000 (100.0000) time: 0.7469 data: 0.5028 max mem: 3531 +Test: [ 20/333] eta: 0:02:17 acc1_pos: 89.323 acc5_pos: 100.000 loss: 0.4265 (0.4466) acc1: 76.8229 (77.9018) acc5: 100.0000 (100.0000) time: 0.0963 data: 0.0005 max mem: 3531 +Test: [ 30/333] eta: 0:01:53 acc1_pos: 45.312 acc5_pos: 100.000 loss: 0.5683 (0.6104) acc1: 67.4479 (65.3058) acc5: 100.0000 (100.0000) time: 0.1704 data: 0.0745 max mem: 3531 +Test: [ 40/333] eta: 0:01:46 acc1_pos: 31.510 acc5_pos: 100.000 loss: 1.0110 (0.7130) acc1: 33.5938 (57.3425) acc5: 100.0000 (100.0000) time: 0.2854 data: 0.1927 max mem: 3531 +Test: [ 50/333] eta: 0:01:34 acc1_pos: 39.323 acc5_pos: 100.000 loss: 1.0090 (0.7734) acc1: 34.6354 (52.9361) acc5: 100.0000 (100.0000) time: 0.2678 data: 0.1753 max mem: 3531 +Test: [ 60/333] eta: 0:01:23 acc1_pos: 23.177 acc5_pos: 100.000 loss: 1.0041 (0.8152) acc1: 32.2917 (49.7609) acc5: 100.0000 (100.0000) time: 0.1844 data: 0.0923 max mem: 3531 +Test: [ 70/333] eta: 0:01:15 acc1_pos: 22.656 acc5_pos: 100.000 loss: 1.0250 (0.8472) acc1: 31.5104 (47.4692) acc5: 100.0000 (100.0000) time: 0.1765 data: 0.0830 max mem: 3531 +Test: [ 80/333] eta: 0:01:09 acc1_pos: 42.448 acc5_pos: 100.000 loss: 1.0099 (0.8666) acc1: 35.9375 (45.9941) acc5: 100.0000 (100.0000) time: 0.1849 data: 0.0889 max mem: 3531 +Test: [ 90/333] eta: 0:01:03 acc1_pos: 49.740 acc5_pos: 100.000 loss: 0.9205 (0.8689) acc1: 39.5833 (45.7303) acc5: 100.0000 (100.0000) time: 0.1704 data: 0.0737 max mem: 3531 +Test: [100/333] eta: 0:00:58 acc1_pos: 46.094 acc5_pos: 100.000 loss: 0.8759 (0.8711) acc1: 40.6250 (45.4105) acc5: 100.0000 (100.0000) time: 0.1602 data: 0.0658 max mem: 3531 +Test: [110/333] eta: 0:00:55 acc1_pos: 48.698 acc5_pos: 100.000 loss: 0.8915 (0.8729) acc1: 40.6250 (45.3289) acc5: 100.0000 (100.0000) time: 0.1859 data: 0.0922 max mem: 3531 +Test: [120/333] eta: 0:00:52 acc1_pos: 40.885 acc5_pos: 100.000 loss: 0.8425 (0.8710) acc1: 43.7500 (45.2200) acc5: 100.0000 (100.0000) time: 0.2156 data: 0.1207 max mem: 3531 +Test: [130/333] eta: 0:00:49 acc1_pos: 58.333 acc5_pos: 100.000 loss: 0.8425 (0.8684) acc1: 44.2708 (45.4735) acc5: 100.0000 (100.0000) time: 0.2288 data: 0.1355 max mem: 3531 +Test: [140/333] eta: 0:00:46 acc1_pos: 58.333 acc5_pos: 100.000 loss: 0.8536 (0.8671) acc1: 47.3958 (45.3328) acc5: 100.0000 (100.0000) time: 0.2112 data: 0.1201 max mem: 3531 +Test: [150/333] eta: 0:00:44 acc1_pos: 39.323 acc5_pos: 100.000 loss: 0.8659 (0.8686) acc1: 42.7083 (45.1849) acc5: 100.0000 (100.0000) time: 0.2345 data: 0.1435 max mem: 3531 +Test: [160/333] eta: 0:00:42 acc1_pos: 37.500 acc5_pos: 100.000 loss: 0.8455 (0.8661) acc1: 46.0938 (45.5050) acc5: 100.0000 (100.0000) acc1_neg: 63.021 acc5_neg: 100.000 time: 0.2639 data: 0.1706 max mem: 3531 +Test: [170/333] eta: 0:00:38 acc1_pos: 37.500 acc5_pos: 100.000 loss: 0.7229 (0.8518) acc1: 60.4167 (46.8491) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.1712 data: 0.0758 max mem: 3531 +Test: [180/333] eta: 0:00:35 acc1_pos: 37.500 acc5_pos: 100.000 loss: 0.5756 (0.8357) acc1: 70.8333 (48.2648) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.1535 data: 0.0584 max mem: 3531 +Test: [190/333] eta: 0:00:33 acc1_pos: 37.500 acc5_pos: 100.000 loss: 0.5528 (0.8176) acc1: 73.1771 (49.8609) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.2157 data: 0.1211 max mem: 3531 +Test: [200/333] eta: 0:00:30 acc1_pos: 37.500 acc5_pos: 100.000 loss: 0.4974 (0.8011) acc1: 77.6042 (51.2969) acc5: 100.0000 (100.0000) acc1_neg: 78.646 acc5_neg: 100.000 time: 0.2152 data: 0.1191 max mem: 3531 +Test: [210/333] eta: 0:00:28 acc1_pos: 37.500 acc5_pos: 100.000 loss: 0.5043 (0.7911) acc1: 76.0417 (52.2191) acc5: 100.0000 (100.0000) acc1_neg: 57.812 acc5_neg: 100.000 time: 0.1981 data: 0.1029 max mem: 3532 +Test: [220/333] eta: 0:00:26 acc1_pos: 37.500 acc5_pos: 100.000 loss: 0.6432 (0.7890) acc1: 63.0208 (52.5370) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.2096 data: 0.1166 max mem: 3532 +Test: [230/333] eta: 0:00:23 acc1_pos: 37.500 acc5_pos: 100.000 loss: 0.3777 (0.7692) acc1: 85.6771 (54.1610) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.2272 data: 0.1333 max mem: 3532 +Test: [240/333] eta: 0:00:21 acc1_pos: 37.500 acc5_pos: 100.000 loss: 0.3598 (0.7546) acc1: 86.7188 (55.4018) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.1949 data: 0.1010 max mem: 3532 +Test: [250/333] eta: 0:00:18 acc1_pos: 37.500 acc5_pos: 100.000 loss: 0.4392 (0.7446) acc1: 80.9896 (56.2604) acc5: 100.0000 (100.0000) acc1_neg: 72.656 acc5_neg: 100.000 time: 0.1673 data: 0.0737 max mem: 3532 +Test: [260/333] eta: 0:00:16 acc1_pos: 37.500 acc5_pos: 100.000 loss: 0.5309 (0.7376) acc1: 75.0000 (56.9055) acc5: 100.0000 (100.0000) acc1_neg: 72.917 acc5_neg: 100.000 time: 0.1917 data: 0.0974 max mem: 3532 +Test: [270/333] eta: 0:00:14 acc1_pos: 37.500 acc5_pos: 100.000 loss: 0.5000 (0.7276) acc1: 77.0833 (57.7241) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.2132 data: 0.1194 max mem: 3532 +Test: [280/333] eta: 0:00:11 acc1_pos: 37.500 acc5_pos: 100.000 loss: 0.4757 (0.7228) acc1: 77.3438 (58.0840) acc5: 100.0000 (100.0000) acc1_neg: 69.531 acc5_neg: 100.000 time: 0.2060 data: 0.1122 max mem: 3532 +Test: [290/333] eta: 0:00:09 acc1_pos: 37.500 acc5_pos: 100.000 loss: 0.6896 (0.7224) acc1: 60.9375 (58.1329) acc5: 100.0000 (100.0000) acc1_neg: 60.938 acc5_neg: 100.000 time: 0.1786 data: 0.0859 max mem: 3532 +Test: [300/333] eta: 0:00:07 acc1_pos: 37.500 acc5_pos: 100.000 loss: 0.7171 (0.7231) acc1: 58.5938 (58.1473) acc5: 100.0000 (100.0000) acc1_neg: 60.938 acc5_neg: 100.000 time: 0.1467 data: 0.0559 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 37.500 acc5_pos: 100.000 loss: 0.6793 (0.7143) acc1: 60.9375 (58.8910) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.1329 data: 0.0426 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 37.500 acc5_pos: 100.000 loss: 0.4090 (0.7068) acc1: 84.3750 (59.5575) acc5: 100.0000 (100.0000) acc1_neg: 53.646 acc5_neg: 100.000 time: 0.1290 data: 0.0391 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 37.500 acc5_pos: 100.000 loss: 0.8055 (0.7116) acc1: 54.4271 (59.3302) acc5: 100.0000 (100.0000) acc1_neg: 53.906 acc5_neg: 100.000 time: 0.1230 data: 0.0333 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 37.500 acc5_pos: 100.000 loss: 0.8324 (0.7126) acc1: 53.6458 (59.2907) acc5: 100.0000 (100.0000) acc1_neg: 52.113 acc5_neg: 100.000 time: 0.1247 data: 0.0319 max mem: 3532 +Test: Total time: 0:01:09 (0.2100 s / it) +* Overall Acc@1 59.291 Acc@5 100.000 + target == real Acc@1 45.140 Acc@5 100.000 + target == fake Acc@1 71.887 Acc@5 100.000 + loss 0.713 + PR-AUC 0.6442 ROC-AUC 0.6271 EER 41.32% (thr=0.6128) +Accuracy of the network on the 127843 test images: 59.3% +[rank0]:[W726 04:22:21.803892790 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:22:41.766000 1178653 torch/distributed/run.py:766] +W0726 04:22:41.766000 1178653 torch/distributed/run.py:766] ***************************************** +W0726 04:22:41.766000 1178653 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:22:41.766000 1178653 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:23:07.742556427 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:23:08.964128536 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:23:08.130243134 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W726 04:23:08.132688438 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:23:08.257606118 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:23:08.295914262 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +[rank0]:[W726 04:23:08.374106785 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:23:08.382431275 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n22_seed6_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed6_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:23:49 acc1_pos: 96.615 acc5_pos: 100.000 loss: 0.1848 (0.1848) acc1: 96.6146 (96.6146) acc5: 100.0000 (100.0000) time: 6.0834 data: 4.7192 max mem: 3531 +Test: [ 10/235] eta: 0:02:22 acc1_pos: 95.833 acc5_pos: 100.000 loss: 0.1917 (0.1911) acc1: 96.6146 (96.5199) acc5: 100.0000 (100.0000) time: 0.6351 data: 0.4293 max mem: 3531 +Test: [ 20/235] eta: 0:01:21 acc1_pos: 96.094 acc5_pos: 100.000 loss: 0.1865 (0.1876) acc1: 96.6146 (96.6642) acc5: 100.0000 (100.0000) time: 0.0950 data: 0.0003 max mem: 3531 +Test: [ 30/235] eta: 0:00:59 acc1_pos: 96.094 acc5_pos: 100.000 loss: 0.1834 (0.1876) acc1: 96.8750 (96.6230) acc5: 100.0000 (100.0000) time: 0.0999 data: 0.0026 max mem: 3531 +Test: [ 40/235] eta: 0:00:50 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.1863 (0.1881) acc1: 96.0938 (96.5574) acc5: 100.0000 (100.0000) time: 0.1350 data: 0.0390 max mem: 3531 +Test: [ 50/235] eta: 0:00:44 acc1_pos: 94.531 acc5_pos: 100.000 loss: 0.1895 (0.1889) acc1: 96.0938 (96.4971) acc5: 100.0000 (100.0000) time: 0.1653 data: 0.0717 max mem: 3531 +Test: [ 60/235] eta: 0:00:39 acc1_pos: 96.615 acc5_pos: 100.000 loss: 0.1913 (0.1902) acc1: 96.0938 (96.4396) acc5: 100.0000 (100.0000) time: 0.1549 data: 0.0614 max mem: 3531 +Test: [ 70/235] eta: 0:00:36 acc1_pos: 93.846 acc5_pos: 100.000 loss: 0.2095 (0.3518) acc1: 94.7917 (86.8068) acc5: 100.0000 (100.0000) acc1_neg: 16.667 acc5_neg: 100.000 time: 0.1606 data: 0.0655 max mem: 3531 +Test: [ 80/235] eta: 0:00:32 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.4803 (0.4914) acc1: 18.7500 (78.3533) acc5: 100.0000 (100.0000) acc1_neg: 17.448 acc5_neg: 100.000 time: 0.1659 data: 0.0725 max mem: 3531 +Test: [ 90/235] eta: 0:00:29 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.4887 (0.6018) acc1: 17.7083 (71.7119) acc5: 100.0000 (100.0000) acc1_neg: 17.188 acc5_neg: 100.000 time: 0.1592 data: 0.0676 max mem: 3531 +Test: [100/235] eta: 0:00:27 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.4844 (0.6873) acc1: 19.2708 (66.5764) acc5: 100.0000 (100.0000) acc1_neg: 19.531 acc5_neg: 100.000 time: 0.1638 data: 0.0710 max mem: 3531 +Test: [110/235] eta: 0:00:24 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.4737 (0.7576) acc1: 19.5312 (62.3475) acc5: 100.0000 (100.0000) acc1_neg: 22.656 acc5_neg: 100.000 time: 0.1695 data: 0.0727 max mem: 3531 +Test: [120/235] eta: 0:00:22 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.5047 (0.8387) acc1: 16.9271 (57.9050) acc5: 100.0000 (100.0000) acc1_neg: 6.510 acc5_neg: 100.000 time: 0.1599 data: 0.0654 max mem: 3531 +Test: [130/235] eta: 0:00:20 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.7531 (0.9084) acc1: 7.5521 (54.0772) acc5: 100.0000 (100.0000) acc1_neg: 7.552 acc5_neg: 100.000 time: 0.1649 data: 0.0736 max mem: 3531 +Test: [140/235] eta: 0:00:18 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.7659 (0.9706) acc1: 7.5521 (50.7683) acc5: 100.0000 (100.0000) acc1_neg: 5.469 acc5_neg: 100.000 time: 0.1749 data: 0.0795 max mem: 3531 +Test: [150/235] eta: 0:00:16 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.7663 (1.0227) acc1: 7.2917 (47.8718) acc5: 100.0000 (100.0000) acc1_neg: 9.115 acc5_neg: 100.000 time: 0.1649 data: 0.0671 max mem: 3531 +Test: [160/235] eta: 0:00:14 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.7454 (1.0680) acc1: 7.0312 (45.3982) acc5: 100.0000 (100.0000) acc1_neg: 8.854 acc5_neg: 100.000 time: 0.1654 data: 0.0683 max mem: 3531 +Test: [170/235] eta: 0:00:12 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.7063 (1.0775) acc1: 9.6354 (44.5191) acc5: 100.0000 (100.0000) acc1_neg: 33.333 acc5_neg: 100.000 time: 0.1749 data: 0.0770 max mem: 3531 +Test: [180/235] eta: 0:00:10 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.1751 (1.0827) acc1: 32.8125 (43.9140) acc5: 100.0000 (100.0000) acc1_neg: 30.729 acc5_neg: 100.000 time: 0.2045 data: 0.1074 max mem: 3531 +Test: [190/235] eta: 0:00:08 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.1636 (1.0864) acc1: 34.1146 (43.3996) acc5: 100.0000 (100.0000) acc1_neg: 37.760 acc5_neg: 100.000 time: 0.2049 data: 0.1082 max mem: 3531 +Test: [200/235] eta: 0:00:06 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.1636 (1.0910) acc1: 32.2917 (42.8314) acc5: 100.0000 (100.0000) acc1_neg: 32.292 acc5_neg: 100.000 time: 0.2000 data: 0.1040 max mem: 3531 +Test: [210/235] eta: 0:00:04 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.1583 (1.0939) acc1: 32.2917 (42.4368) acc5: 100.0000 (100.0000) acc1_neg: 35.417 acc5_neg: 100.000 time: 0.1970 data: 0.1047 max mem: 3532 +Test: [220/235] eta: 0:00:02 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.1580 (1.0969) acc1: 33.5938 (42.0390) acc5: 100.0000 (100.0000) acc1_neg: 32.292 acc5_neg: 100.000 time: 0.1748 data: 0.0827 max mem: 3532 +Test: [230/235] eta: 0:00:00 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.1576 (1.0993) acc1: 33.8542 (41.7174) acc5: 100.0000 (100.0000) acc1_neg: 32.292 acc5_neg: 100.000 time: 0.1738 data: 0.0798 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.1576 (1.1009) acc1: 33.8542 (41.6093) acc5: 100.0000 (100.0000) acc1_neg: 39.130 acc5_neg: 100.000 time: 0.1516 data: 0.0607 max mem: 3532 +Test: Total time: 0:00:44 (0.1896 s / it) +* Overall Acc@1 41.609 Acc@5 100.000 + target == real Acc@1 96.439 Acc@5 100.000 + target == fake Acc@1 21.778 Acc@5 100.000 + loss 1.101 + PR-AUC 0.8833 ROC-AUC 0.7415 EER 32.64% (thr=0.1539) +Accuracy of the network on the 89879 test images: 41.6% +[rank0]:[W726 04:24:06.210179942 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:24:25.399000 1179212 torch/distributed/run.py:766] +W0726 04:24:25.399000 1179212 torch/distributed/run.py:766] ***************************************** +W0726 04:24:25.399000 1179212 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:24:25.399000 1179212 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:24:53.299174480 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:24:53.621914001 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:24:54.912115034 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:24:54.060218952 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:24:54.106054118 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:24:54.132926333 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W726 04:24:54.136500075 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:24:54.177268367 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n22_seed6_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed6_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:31:36 acc1_pos: 34.896 acc5_pos: 100.000 loss: 0.9421 (0.9421) acc1: 34.8958 (34.8958) acc5: 100.0000 (100.0000) time: 7.0515 data: 5.4055 max mem: 3531 +Test: [ 10/269] eta: 0:03:08 acc1_pos: 30.990 acc5_pos: 100.000 loss: 0.8514 (0.8661) acc1: 42.4479 (42.4479) acc5: 100.0000 (100.0000) time: 0.7265 data: 0.4918 max mem: 3531 +Test: [ 20/269] eta: 0:01:46 acc1_pos: 45.052 acc5_pos: 100.000 loss: 0.8131 (0.8357) acc1: 42.4479 (44.9901) acc5: 100.0000 (100.0000) time: 0.0970 data: 0.0004 max mem: 3531 +Test: [ 30/269] eta: 0:01:24 acc1_pos: 44.010 acc5_pos: 100.000 loss: 0.7518 (0.8155) acc1: 45.3125 (47.1186) acc5: 100.0000 (100.0000) time: 0.1467 data: 0.0504 max mem: 3531 +Test: [ 40/269] eta: 0:01:15 acc1_pos: 4.310 acc5_pos: 100.000 loss: 0.6002 (0.7250) acc1: 67.7083 (55.8498) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2299 data: 0.1369 max mem: 3531 +Test: [ 50/269] eta: 0:01:06 acc1_pos: 4.310 acc5_pos: 100.000 loss: 0.3412 (0.6521) acc1: 92.7083 (62.7655) acc5: 100.0000 (100.0000) acc1_neg: 76.042 acc5_neg: 100.000 time: 0.2331 data: 0.1398 max mem: 3531 +Test: [ 60/269] eta: 0:00:58 acc1_pos: 4.310 acc5_pos: 100.000 loss: 0.3674 (0.6146) acc1: 87.2396 (66.0861) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.1767 data: 0.0840 max mem: 3531 +Test: [ 70/269] eta: 0:00:54 acc1_pos: 4.310 acc5_pos: 100.000 loss: 0.4178 (0.5896) acc1: 84.1146 (68.3979) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1899 data: 0.0968 max mem: 3531 +Test: [ 80/269] eta: 0:00:50 acc1_pos: 4.310 acc5_pos: 100.000 loss: 0.3771 (0.5631) acc1: 88.0208 (70.8591) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.2245 data: 0.1311 max mem: 3531 +Test: [ 90/269] eta: 0:00:44 acc1_pos: 4.310 acc5_pos: 100.000 loss: 0.3955 (0.5539) acc1: 85.1562 (71.9923) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.1731 data: 0.0794 max mem: 3531 +Test: [100/269] eta: 0:00:40 acc1_pos: 4.310 acc5_pos: 100.000 loss: 0.4732 (0.5446) acc1: 82.5521 (72.9966) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.1436 data: 0.0500 max mem: 3531 +Test: [110/269] eta: 0:00:37 acc1_pos: 4.310 acc5_pos: 100.000 loss: 0.4398 (0.5341) acc1: 83.8542 (74.1296) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.1792 data: 0.0875 max mem: 3531 +Test: [120/269] eta: 0:00:35 acc1_pos: 4.310 acc5_pos: 100.000 loss: 0.4398 (0.5296) acc1: 83.8542 (74.6234) acc5: 100.0000 (100.0000) acc1_neg: 69.271 acc5_neg: 100.000 time: 0.2276 data: 0.1357 max mem: 3531 +Test: [130/269] eta: 0:00:32 acc1_pos: 4.310 acc5_pos: 100.000 loss: 0.4865 (0.5251) acc1: 78.3854 (74.9980) acc5: 100.0000 (100.0000) acc1_neg: 80.729 acc5_neg: 100.000 time: 0.2162 data: 0.1231 max mem: 3531 +Test: [140/269] eta: 0:00:29 acc1_pos: 4.310 acc5_pos: 100.000 loss: 0.5234 (0.5333) acc1: 70.0521 (74.0304) acc5: 100.0000 (100.0000) acc1_neg: 56.250 acc5_neg: 100.000 time: 0.1779 data: 0.0855 max mem: 3531 +Test: [150/269] eta: 0:00:27 acc1_pos: 4.310 acc5_pos: 100.000 loss: 0.6120 (0.5357) acc1: 64.5833 (73.6634) acc5: 100.0000 (100.0000) acc1_neg: 70.833 acc5_neg: 100.000 time: 0.1811 data: 0.0860 max mem: 3531 +Test: [160/269] eta: 0:00:24 acc1_pos: 4.310 acc5_pos: 100.000 loss: 0.4120 (0.5279) acc1: 83.5938 (74.4565) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1592 data: 0.0616 max mem: 3531 +Test: [170/269] eta: 0:00:21 acc1_pos: 4.310 acc5_pos: 100.000 loss: 0.3989 (0.5217) acc1: 87.5000 (75.0624) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1455 data: 0.0511 max mem: 3531 +Test: [180/269] eta: 0:00:19 acc1_pos: 4.310 acc5_pos: 100.000 loss: 0.3502 (0.5112) acc1: 91.6667 (76.0273) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1658 data: 0.0723 max mem: 3531 +Test: [190/269] eta: 0:00:16 acc1_pos: 4.310 acc5_pos: 100.000 loss: 0.3579 (0.5064) acc1: 90.8854 (76.4534) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.1681 data: 0.0750 max mem: 3531 +Test: [200/269] eta: 0:00:14 acc1_pos: 4.310 acc5_pos: 100.000 loss: 0.4086 (0.5035) acc1: 84.3750 (76.7232) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.1657 data: 0.0744 max mem: 3531 +Test: [210/269] eta: 0:00:12 acc1_pos: 4.310 acc5_pos: 100.000 loss: 0.4561 (0.5016) acc1: 82.5521 (76.9081) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.1631 data: 0.0714 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 4.310 acc5_pos: 100.000 loss: 0.4340 (0.4988) acc1: 83.8542 (77.1470) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1730 data: 0.0796 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 4.310 acc5_pos: 100.000 loss: 0.3575 (0.4931) acc1: 85.9375 (77.5782) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1905 data: 0.0959 max mem: 3532 +Test: [240/269] eta: 0:00:05 acc1_pos: 4.310 acc5_pos: 100.000 loss: 0.3920 (0.4900) acc1: 87.2396 (77.8786) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.1911 data: 0.0982 max mem: 3532 +Test: [250/269] eta: 0:00:03 acc1_pos: 4.310 acc5_pos: 100.000 loss: 0.5082 (0.4944) acc1: 73.4375 (77.4153) acc5: 100.0000 (100.0000) acc1_neg: 39.583 acc5_neg: 100.000 time: 0.1744 data: 0.0811 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 4.310 acc5_pos: 100.000 loss: 0.5570 (0.4968) acc1: 65.6250 (77.1701) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1316 data: 0.0378 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 4.310 acc5_pos: 100.000 loss: 0.4121 (0.4921) acc1: 83.8542 (77.6093) acc5: 100.0000 (100.0000) acc1_neg: 98.163 acc5_neg: 100.000 time: 0.1435 data: 0.0485 max mem: 3532 +Test: Total time: 0:00:53 (0.1990 s / it) +* Overall Acc@1 77.609 Acc@5 100.000 + target == real Acc@1 46.090 Acc@5 100.000 + target == fake Acc@1 81.911 Acc@5 100.000 + loss 0.492 + PR-AUC 0.9431 ROC-AUC 0.7187 EER 34.10% (thr=0.6196) +Accuracy of the network on the 103293 test images: 77.6% +[rank0]:[W726 04:25:56.494470861 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T04:26:04Z] eval_done tag=fixed500k_fake_n22_seed6_resampled +[2026-07-26T04:26:04Z] eval_start tag=fixed500k_real_n11_seed7_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n11_classseed7_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T04:26:05Z +hostname=n116-041-188 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n11_classseed7_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n11_seed7_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 04:26:17.104000 1179777 torch/distributed/run.py:766] +W0726 04:26:17.104000 1179777 torch/distributed/run.py:766] ***************************************** +W0726 04:26:17.104000 1179777 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:26:17.104000 1179777 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:26:43.888636909 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:26:43.904342520 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:26:44.320234072 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:26:44.384761868 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:26:44.487718334 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:26:44.536980999 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:26:44.662765538 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:26:44.742263218 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n11_seed7_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n11_classseed7_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:27:24 acc1_pos: 70.052 acc5_pos: 100.000 loss: 0.5960 (0.5960) acc1: 70.0521 (70.0521) acc5: 100.0000 (100.0000) time: 7.0296 data: 5.4423 max mem: 3531 +Test: [ 10/234] eta: 0:02:47 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4563 (0.4871) acc1: 80.9896 (76.0180) acc5: 100.0000 (100.0000) time: 0.7472 data: 0.5180 max mem: 3531 +Test: [ 20/234] eta: 0:01:33 acc1_pos: 89.323 acc5_pos: 100.000 loss: 0.4557 (0.5196) acc1: 80.4688 (73.2515) acc5: 100.0000 (100.0000) time: 0.1068 data: 0.0130 max mem: 3531 +Test: [ 30/234] eta: 0:01:27 acc1_pos: 57.812 acc5_pos: 100.000 loss: 0.3149 (0.4671) acc1: 89.3229 (76.8145) acc5: 100.0000 (100.0000) time: 0.2533 data: 0.1604 max mem: 3531 +Test: [ 40/234] eta: 0:01:16 acc1_pos: 98.698 acc5_pos: 100.000 loss: 0.2755 (0.4873) acc1: 89.3229 (74.5490) acc5: 100.0000 (100.0000) time: 0.3506 data: 0.2587 max mem: 3531 +Test: [ 50/234] eta: 0:01:04 acc1_pos: 97.917 acc5_pos: 100.000 loss: 0.1785 (0.4259) acc1: 96.6146 (78.9931) acc5: 100.0000 (100.0000) time: 0.2305 data: 0.1390 max mem: 3531 +Test: [ 60/234] eta: 0:00:54 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.1573 (0.3765) acc1: 98.1771 (82.3429) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1526 data: 0.0606 max mem: 3531 +Test: [ 70/234] eta: 0:00:49 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.1223 (0.3735) acc1: 99.4792 (82.5851) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.1793 data: 0.0864 max mem: 3531 +Test: [ 80/234] eta: 0:00:43 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.4357 (0.3715) acc1: 79.4271 (82.8254) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1795 data: 0.0854 max mem: 3531 +Test: [ 90/234] eta: 0:00:37 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.2161 (0.3480) acc1: 95.0521 (84.2863) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.1250 data: 0.0305 max mem: 3531 +Test: [100/234] eta: 0:00:34 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.2523 (0.3474) acc1: 90.8854 (84.3647) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.1495 data: 0.0556 max mem: 3531 +Test: [110/234] eta: 0:00:30 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.3117 (0.3438) acc1: 85.9375 (84.6730) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.1846 data: 0.0913 max mem: 3531 +Test: [120/234] eta: 0:00:27 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.3276 (0.3542) acc1: 85.1562 (84.1619) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.1630 data: 0.0687 max mem: 3531 +Test: [130/234] eta: 0:00:24 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.4597 (0.3735) acc1: 77.3438 (82.9536) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.1292 data: 0.0323 max mem: 3531 +Test: [140/234] eta: 0:00:21 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.4929 (0.3916) acc1: 74.2188 (82.0405) acc5: 100.0000 (100.0000) acc1_neg: 40.104 acc5_neg: 100.000 time: 0.1661 data: 0.0711 max mem: 3531 +Test: [150/234] eta: 0:00:19 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.3517 (0.3957) acc1: 83.3333 (81.8191) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.2010 data: 0.1090 max mem: 3531 +Test: [160/234] eta: 0:00:16 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.3590 (0.4046) acc1: 81.7708 (81.2371) acc5: 100.0000 (100.0000) acc1_neg: 59.375 acc5_neg: 100.000 time: 0.1892 data: 0.0982 max mem: 3531 +Test: [170/234] eta: 0:00:14 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.6567 (0.4200) acc1: 64.5833 (80.2312) acc5: 100.0000 (100.0000) acc1_neg: 73.698 acc5_neg: 100.000 time: 0.2041 data: 0.1135 max mem: 3531 +Test: [180/234] eta: 0:00:11 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.6357 (0.4281) acc1: 65.1042 (79.7220) acc5: 100.0000 (100.0000) acc1_neg: 41.146 acc5_neg: 100.000 time: 0.1872 data: 0.0955 max mem: 3531 +Test: [190/234] eta: 0:00:09 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.5013 (0.4335) acc1: 77.3438 (79.5457) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.1926 data: 0.1010 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.4886 (0.4243) acc1: 79.4271 (80.1539) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1784 data: 0.0873 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.1007 (0.4110) acc1: 99.4792 (81.0291) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1302 data: 0.0390 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.2275 (0.4039) acc1: 97.1354 (81.6554) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.1401 data: 0.0495 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.3906 (0.4184) acc1: 60.6771 (80.4778) acc5: 100.0000 (100.0000) acc1_neg: 54.948 acc5_neg: 100.000 time: 0.1519 data: 0.0618 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.7151 (0.4224) acc1: 55.9896 (80.1904) acc5: 100.0000 (100.0000) acc1_neg: 54.475 acc5_neg: 100.000 time: 0.1478 data: 0.0566 max mem: 3532 +Test: Total time: 0:00:48 (0.2069 s / it) +* Overall Acc@1 80.190 Acc@5 100.000 + target == real Acc@1 79.107 Acc@5 100.000 + target == fake Acc@1 80.496 Acc@5 100.000 + loss 0.422 + PR-AUC 0.9657 ROC-AUC 0.8893 EER 20.14% (thr=0.5107) +Accuracy of the network on the 89729 test images: 80.2% +[rank0]:[W726 04:27:45.817050216 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:28:05.741000 1180336 torch/distributed/run.py:766] +W0726 04:28:05.741000 1180336 torch/distributed/run.py:766] ***************************************** +W0726 04:28:05.741000 1180336 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:28:05.741000 1180336 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:28:34.278795167 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:28:34.303593121 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:28:34.529370548 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 6): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:28:34.651419791 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:28:34.658141230 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:28:34.876520406 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:28:34.897216889 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:28:35.914572353 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n11_seed7_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n11_classseed7_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:23:52 acc1_pos: 25.260 acc5_pos: 100.000 loss: 1.1677 (1.1677) acc1: 25.2604 (25.2604) acc5: 100.0000 (100.0000) time: 6.6329 data: 5.0802 max mem: 3531 +Test: [ 10/2387] eta: 0:27:08 acc1_pos: 14.062 acc5_pos: 100.000 loss: 1.7181 (1.6799) acc1: 5.2083 (8.7358) acc5: 100.0000 (100.0000) time: 0.6849 data: 0.4622 max mem: 3531 +Test: [ 20/2387] eta: 0:15:52 acc1_pos: 14.844 acc5_pos: 100.000 loss: 1.6325 (1.5808) acc1: 8.5938 (11.1235) acc5: 100.0000 (100.0000) time: 0.0908 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:13:25 acc1_pos: 0.260 acc5_pos: 100.000 loss: 1.5457 (1.6227) acc1: 7.0312 (10.5091) acc5: 100.0000 (100.0000) time: 0.1533 data: 0.0606 max mem: 3531 +Test: [ 40/2387] eta: 0:13:24 acc1_pos: 13.281 acc5_pos: 100.000 loss: 1.6099 (1.6247) acc1: 5.4688 (9.9784) acc5: 100.0000 (100.0000) time: 0.2799 data: 0.1845 max mem: 3531 +Test: [ 50/2387] eta: 0:12:10 acc1_pos: 47.135 acc5_pos: 100.000 loss: 1.4584 (1.5547) acc1: 10.4167 (12.2498) acc5: 100.0000 (100.0000) time: 0.2673 data: 0.1733 max mem: 3531 +Test: [ 60/2387] eta: 0:11:28 acc1_pos: 32.552 acc5_pos: 100.000 loss: 1.1514 (1.4940) acc1: 26.8229 (15.3646) acc5: 100.0000 (100.0000) time: 0.1999 data: 0.1076 max mem: 3531 +Test: [ 70/2387] eta: 0:11:16 acc1_pos: 38.281 acc5_pos: 100.000 loss: 1.1514 (1.4520) acc1: 26.8229 (16.7877) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1447 max mem: 3531 +Test: [ 80/2387] eta: 0:10:59 acc1_pos: 46.615 acc5_pos: 100.000 loss: 1.0650 (1.3960) acc1: 32.2917 (19.3158) acc5: 100.0000 (100.0000) time: 0.2549 data: 0.1590 max mem: 3531 +Test: [ 90/2387] eta: 0:10:32 acc1_pos: 36.719 acc5_pos: 100.000 loss: 0.9055 (1.3405) acc1: 38.5417 (22.0582) acc5: 100.0000 (100.0000) time: 0.2149 data: 0.1209 max mem: 3531 +Test: [ 100/2387] eta: 0:10:19 acc1_pos: 54.167 acc5_pos: 100.000 loss: 0.8951 (1.2977) acc1: 44.5312 (24.2677) acc5: 100.0000 (100.0000) time: 0.2099 data: 0.1118 max mem: 3531 +Test: [ 110/2387] eta: 0:10:05 acc1_pos: 48.698 acc5_pos: 100.000 loss: 0.8827 (1.2619) acc1: 42.7083 (25.8469) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1254 max mem: 3531 +Test: [ 120/2387] eta: 0:10:00 acc1_pos: 58.854 acc5_pos: 100.000 loss: 0.8420 (1.2244) acc1: 46.0938 (28.0023) acc5: 100.0000 (100.0000) time: 0.2349 data: 0.1388 max mem: 3531 +Test: [ 130/2387] eta: 0:09:48 acc1_pos: 50.781 acc5_pos: 100.000 loss: 0.8500 (1.2030) acc1: 46.0938 (29.2005) acc5: 100.0000 (100.0000) time: 0.2302 data: 0.1368 max mem: 3531 +Test: [ 140/2387] eta: 0:09:39 acc1_pos: 59.115 acc5_pos: 100.000 loss: 0.9152 (1.1827) acc1: 41.4062 (30.3173) acc5: 100.0000 (100.0000) time: 0.2159 data: 0.1249 max mem: 3531 +Test: [ 150/2387] eta: 0:09:32 acc1_pos: 37.240 acc5_pos: 100.000 loss: 0.8410 (1.1597) acc1: 47.1354 (31.5380) acc5: 100.0000 (100.0000) time: 0.2246 data: 0.1329 max mem: 3531 +Test: [ 160/2387] eta: 0:09:20 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.7921 (1.1231) acc1: 50.7812 (33.8445) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2089 data: 0.1150 max mem: 3531 +Test: [ 170/2387] eta: 0:09:06 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1428 (1.0654) acc1: 97.6562 (37.6371) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1749 data: 0.0781 max mem: 3531 +Test: [ 180/2387] eta: 0:08:53 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1360 (1.0142) acc1: 98.6979 (41.0005) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1599 data: 0.0642 max mem: 3531 +Test: [ 190/2387] eta: 0:08:41 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1357 (0.9681) acc1: 98.6979 (44.0172) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1599 data: 0.0677 max mem: 3531 +Test: [ 200/2387] eta: 0:08:31 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1366 (0.9267) acc1: 98.6979 (46.7338) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1649 data: 0.0731 max mem: 3531 +Test: [ 210/2387] eta: 0:08:24 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1364 (0.8892) acc1: 98.6979 (49.2002) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1799 data: 0.0859 max mem: 3532 +Test: [ 220/2387] eta: 0:08:17 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1345 (0.8552) acc1: 98.6979 (51.4270) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1848 data: 0.0898 max mem: 3532 +Test: [ 230/2387] eta: 0:08:07 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1361 (0.8241) acc1: 98.4375 (53.4610) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1648 data: 0.0699 max mem: 3532 +Test: [ 240/2387] eta: 0:07:59 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1433 (0.7975) acc1: 97.6562 (55.2462) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1549 data: 0.0602 max mem: 3532 +Test: [ 250/2387] eta: 0:07:51 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1391 (0.7691) acc1: 97.6562 (57.0053) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1599 data: 0.0639 max mem: 3532 +Test: [ 260/2387] eta: 0:07:47 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0669 (0.7422) acc1: 100.0000 (58.6426) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0762 max mem: 3532 +Test: [ 270/2387] eta: 0:07:42 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0609 (0.7172) acc1: 100.0000 (60.1649) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0907 max mem: 3532 +Test: [ 280/2387] eta: 0:07:40 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0626 (0.6939) acc1: 100.0000 (61.5816) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2060 data: 0.1091 max mem: 3532 +Test: [ 290/2387] eta: 0:07:36 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0626 (0.6722) acc1: 100.0000 (62.9018) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2067 data: 0.1139 max mem: 3532 +Test: [ 300/2387] eta: 0:07:34 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0659 (0.6527) acc1: 100.0000 (64.1135) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2051 data: 0.1122 max mem: 3532 +Test: [ 310/2387] eta: 0:07:27 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0834 (0.6359) acc1: 99.4792 (65.2474) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1832 data: 0.0884 max mem: 3532 +Test: [ 320/2387] eta: 0:07:24 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1907 (0.6222) acc1: 99.4792 (66.3097) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1744 data: 0.0816 max mem: 3532 +Test: [ 330/2387] eta: 0:07:21 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1919 (0.6091) acc1: 99.2188 (67.3165) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1999 data: 0.1075 max mem: 3532 +Test: [ 340/2387] eta: 0:07:20 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1921 (0.5969) acc1: 99.7396 (68.2681) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2195 data: 0.1253 max mem: 3532 +Test: [ 350/2387] eta: 0:07:20 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1840 (0.5830) acc1: 100.0000 (69.1714) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2469 data: 0.1542 max mem: 3532 +Test: [ 360/2387] eta: 0:07:17 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1185 (0.5704) acc1: 99.7396 (69.9929) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2249 data: 0.1334 max mem: 3532 +Test: [ 370/2387] eta: 0:07:14 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1281 (0.5586) acc1: 98.6979 (70.7709) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.1031 max mem: 3532 +Test: [ 380/2387] eta: 0:07:12 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1303 (0.5473) acc1: 98.6979 (71.5100) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2060 data: 0.1146 max mem: 3532 +Test: [ 390/2387] eta: 0:07:09 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1267 (0.5366) acc1: 98.6979 (72.2073) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2133 data: 0.1217 max mem: 3532 +Test: [ 400/2387] eta: 0:07:07 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1248 (0.5264) acc1: 98.9583 (72.8719) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2084 data: 0.1148 max mem: 3532 +Test: [ 410/2387] eta: 0:07:04 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1255 (0.5167) acc1: 98.9583 (73.5047) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2018 data: 0.1078 max mem: 3532 +Test: [ 420/2387] eta: 0:07:02 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1269 (0.5075) acc1: 98.9583 (74.1099) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2058 data: 0.1119 max mem: 3532 +Test: [ 430/2387] eta: 0:06:59 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1280 (0.4994) acc1: 98.6979 (74.6520) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2074 data: 0.1129 max mem: 3532 +Test: [ 440/2387] eta: 0:06:56 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1309 (0.4906) acc1: 97.9167 (75.1972) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2051 data: 0.1106 max mem: 3533 +Test: [ 450/2387] eta: 0:06:55 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0720 (0.4813) acc1: 100.0000 (75.7472) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2199 data: 0.1247 max mem: 3533 +Test: [ 460/2387] eta: 0:06:52 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0721 (0.4725) acc1: 100.0000 (76.2727) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2146 data: 0.1210 max mem: 3533 +Test: [ 470/2387] eta: 0:06:50 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0723 (0.4640) acc1: 100.0000 (76.7765) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2020 data: 0.1112 max mem: 3533 +Test: [ 480/2387] eta: 0:06:48 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0722 (0.4558) acc1: 100.0000 (77.2593) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2127 data: 0.1213 max mem: 3533 +Test: [ 490/2387] eta: 0:06:45 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0723 (0.4480) acc1: 100.0000 (77.7219) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2102 data: 0.1180 max mem: 3533 +Test: [ 500/2387] eta: 0:06:42 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0726 (0.4405) acc1: 100.0000 (78.1666) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1885 data: 0.0963 max mem: 3533 +Test: [ 510/2387] eta: 0:06:40 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0728 (0.4334) acc1: 100.0000 (78.5933) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2057 data: 0.1133 max mem: 3533 +Test: [ 520/2387] eta: 0:06:36 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0837 (0.4268) acc1: 100.0000 (79.0037) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1944 data: 0.1031 max mem: 3533 +Test: [ 530/2387] eta: 0:06:34 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0894 (0.4205) acc1: 100.0000 (79.3991) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1882 data: 0.0951 max mem: 3533 +Test: [ 540/2387] eta: 0:06:31 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0925 (0.4144) acc1: 100.0000 (79.7790) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2043 data: 0.1104 max mem: 3533 +Test: [ 550/2387] eta: 0:06:28 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0933 (0.4086) acc1: 100.0000 (80.1459) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1872 data: 0.0949 max mem: 3533 +Test: [ 560/2387] eta: 0:06:26 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0935 (0.4030) acc1: 100.0000 (80.4999) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1952 data: 0.1030 max mem: 3533 +Test: [ 570/2387] eta: 0:06:24 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0931 (0.3975) acc1: 100.0000 (80.8409) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2061 data: 0.1144 max mem: 3533 +Test: [ 580/2387] eta: 0:06:21 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0921 (0.3923) acc1: 100.0000 (81.1702) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1959 data: 0.1050 max mem: 3533 +Test: [ 590/2387] eta: 0:06:18 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0931 (0.3872) acc1: 100.0000 (81.4888) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0993 max mem: 3533 +Test: [ 600/2387] eta: 0:06:15 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0885 (0.3822) acc1: 100.0000 (81.7960) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1947 data: 0.1032 max mem: 3533 +Test: [ 610/2387] eta: 0:06:12 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0859 (0.3774) acc1: 100.0000 (82.0939) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1764 data: 0.0843 max mem: 3533 +Test: [ 620/2387] eta: 0:06:09 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0868 (0.3727) acc1: 100.0000 (82.3822) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1654 data: 0.0721 max mem: 3533 +Test: [ 630/2387] eta: 0:06:05 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0868 (0.3682) acc1: 100.0000 (82.6615) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1641 data: 0.0704 max mem: 3533 +Test: [ 640/2387] eta: 0:06:02 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0873 (0.3638) acc1: 100.0000 (82.9319) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1623 data: 0.0701 max mem: 3533 +Test: [ 650/2387] eta: 0:05:59 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0880 (0.3596) acc1: 100.0000 (83.1941) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1642 data: 0.0731 max mem: 3533 +Test: [ 660/2387] eta: 0:05:55 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0869 (0.3554) acc1: 100.0000 (83.4484) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1629 data: 0.0717 max mem: 3533 +Test: [ 670/2387] eta: 0:05:53 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0869 (0.3514) acc1: 100.0000 (83.6950) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1762 data: 0.0832 max mem: 3534 +Test: [ 680/2387] eta: 0:05:50 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0844 (0.3474) acc1: 100.0000 (83.9345) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1776 data: 0.0842 max mem: 3534 +Test: [ 690/2387] eta: 0:05:47 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0867 (0.3448) acc1: 100.0000 (84.1044) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1697 data: 0.0762 max mem: 3534 +Test: [ 700/2387] eta: 0:05:44 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1901 (0.3426) acc1: 94.7917 (84.2539) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1748 data: 0.0799 max mem: 3534 +Test: [ 710/2387] eta: 0:05:42 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1923 (0.3406) acc1: 94.5312 (84.3893) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1798 data: 0.0863 max mem: 3534 +Test: [ 720/2387] eta: 0:05:39 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1903 (0.3384) acc1: 94.5312 (84.5339) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1808 data: 0.0893 max mem: 3534 +Test: [ 730/2387] eta: 0:05:37 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1870 (0.3364) acc1: 94.7917 (84.6710) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1815 data: 0.0885 max mem: 3534 +Test: [ 740/2387] eta: 0:05:34 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1858 (0.3344) acc1: 94.5312 (84.8076) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1794 data: 0.0866 max mem: 3534 +Test: [ 750/2387] eta: 0:05:32 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1844 (0.3324) acc1: 94.5312 (84.9402) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1823 data: 0.0893 max mem: 3534 +Test: [ 760/2387] eta: 0:05:29 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1786 (0.3302) acc1: 95.0521 (85.0912) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1838 data: 0.0907 max mem: 3534 +Test: [ 770/2387] eta: 0:05:27 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1296 (0.3274) acc1: 97.9167 (85.2691) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0980 max mem: 3534 +Test: [ 780/2387] eta: 0:05:24 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0852 (0.3242) acc1: 100.0000 (85.4550) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1799 data: 0.0870 max mem: 3534 +Test: [ 790/2387] eta: 0:05:21 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0834 (0.3216) acc1: 100.0000 (85.6162) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1611 data: 0.0693 max mem: 3534 +Test: [ 800/2387] eta: 0:05:19 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0822 (0.3185) acc1: 99.7396 (85.7951) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1620 data: 0.0704 max mem: 3534 +Test: [ 810/2387] eta: 0:05:16 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0793 (0.3156) acc1: 100.0000 (85.9664) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1668 data: 0.0749 max mem: 3534 +Test: [ 820/2387] eta: 0:05:13 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0917 (0.3130) acc1: 100.0000 (86.1335) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1692 data: 0.0786 max mem: 3534 +Test: [ 830/2387] eta: 0:05:11 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1135 (0.3114) acc1: 99.7396 (86.2550) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1718 data: 0.0780 max mem: 3534 +Test: [ 840/2387] eta: 0:05:09 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1513 (0.3092) acc1: 98.1771 (86.3983) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1838 data: 0.0870 max mem: 3534 +Test: [ 850/2387] eta: 0:05:07 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1195 (0.3070) acc1: 98.9583 (86.5459) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1914 data: 0.0975 max mem: 3534 +Test: [ 860/2387] eta: 0:05:05 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1139 (0.3048) acc1: 99.2188 (86.6876) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2161 data: 0.1234 max mem: 3534 +Test: [ 870/2387] eta: 0:05:04 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1116 (0.3028) acc1: 99.2188 (86.8204) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2351 data: 0.1424 max mem: 3534 +Test: [ 880/2387] eta: 0:05:03 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1116 (0.3006) acc1: 99.4792 (86.9614) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2389 data: 0.1478 max mem: 3534 +Test: [ 890/2387] eta: 0:05:00 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1124 (0.2986) acc1: 99.4792 (87.1008) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2047 data: 0.1123 max mem: 3534 +Test: [ 900/2387] eta: 0:04:58 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1327 (0.3009) acc1: 97.9167 (86.9506) acc5: 100.0000 (100.0000) acc1_neg: 69.271 acc5_neg: 100.000 time: 0.1910 data: 0.0973 max mem: 3535 +Test: [ 910/2387] eta: 0:04:56 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.5895 (0.3040) acc1: 66.4062 (86.7273) acc5: 100.0000 (100.0000) acc1_neg: 64.323 acc5_neg: 100.000 time: 0.2186 data: 0.1253 max mem: 3535 +Test: [ 920/2387] eta: 0:04:55 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.5999 (0.3072) acc1: 66.4062 (86.5112) acc5: 100.0000 (100.0000) acc1_neg: 64.583 acc5_neg: 100.000 time: 0.2296 data: 0.1379 max mem: 3535 +Test: [ 930/2387] eta: 0:04:53 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.5924 (0.3102) acc1: 66.6667 (86.2967) acc5: 100.0000 (100.0000) acc1_neg: 62.760 acc5_neg: 100.000 time: 0.2250 data: 0.1345 max mem: 3535 +Test: [ 940/2387] eta: 0:04:51 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.5952 (0.3132) acc1: 66.1458 (86.0900) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.2043 data: 0.1129 max mem: 3535 +Test: [ 950/2387] eta: 0:04:49 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.5909 (0.3161) acc1: 66.4062 (85.8877) acc5: 100.0000 (100.0000) acc1_neg: 69.010 acc5_neg: 100.000 time: 0.2011 data: 0.1096 max mem: 3535 +Test: [ 960/2387] eta: 0:04:47 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.5813 (0.3188) acc1: 67.7083 (85.7020) acc5: 100.0000 (100.0000) acc1_neg: 65.885 acc5_neg: 100.000 time: 0.1996 data: 0.1082 max mem: 3535 +Test: [ 970/2387] eta: 0:04:45 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.5802 (0.3216) acc1: 67.7083 (85.5146) acc5: 100.0000 (100.0000) acc1_neg: 71.354 acc5_neg: 100.000 time: 0.1933 data: 0.1010 max mem: 3535 +Test: [ 980/2387] eta: 0:04:43 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.5681 (0.3211) acc1: 68.7500 (85.5412) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1864 data: 0.0932 max mem: 3535 +Test: [ 990/2387] eta: 0:04:41 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1008 (0.3189) acc1: 98.6979 (85.6779) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1965 data: 0.1011 max mem: 3535 +Test: [1000/2387] eta: 0:04:39 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0871 (0.3167) acc1: 99.4792 (85.8074) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2079 data: 0.1127 max mem: 3535 +Test: [1010/2387] eta: 0:04:36 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0969 (0.3151) acc1: 99.2188 (85.9156) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.1900 data: 0.0930 max mem: 3535 +Test: [1020/2387] eta: 0:04:35 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1329 (0.3134) acc1: 98.1771 (86.0331) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1969 data: 0.1011 max mem: 3535 +Test: [1030/2387] eta: 0:04:32 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0839 (0.3111) acc1: 100.0000 (86.1684) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2032 data: 0.1089 max mem: 3535 +Test: [1040/2387] eta: 0:04:30 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0803 (0.3091) acc1: 100.0000 (86.2977) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1793 data: 0.0844 max mem: 3535 +Test: [1050/2387] eta: 0:04:28 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0830 (0.3069) acc1: 100.0000 (86.4274) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1813 data: 0.0897 max mem: 3535 +Test: [1060/2387] eta: 0:04:26 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0833 (0.3048) acc1: 100.0000 (86.5538) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1943 data: 0.1030 max mem: 3535 +Test: [1070/2387] eta: 0:04:24 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0887 (0.3032) acc1: 100.0000 (86.6594) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1960 data: 0.1028 max mem: 3535 +Test: [1080/2387] eta: 0:04:21 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1786 (0.3025) acc1: 95.8333 (86.7214) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1809 data: 0.0883 max mem: 3535 +Test: [1090/2387] eta: 0:04:19 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1631 (0.3008) acc1: 97.3958 (86.8362) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1860 data: 0.0952 max mem: 3535 +Test: [1100/2387] eta: 0:04:17 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1115 (0.2992) acc1: 99.7396 (86.9461) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1856 data: 0.0945 max mem: 3535 +Test: [1110/2387] eta: 0:04:15 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1241 (0.2978) acc1: 98.9583 (87.0396) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1830 data: 0.0920 max mem: 3535 +Test: [1120/2387] eta: 0:04:13 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1223 (0.2963) acc1: 99.2188 (87.1474) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2025 data: 0.1093 max mem: 3536 +Test: [1130/2387] eta: 0:04:11 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1105 (0.2947) acc1: 99.4792 (87.2541) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2040 data: 0.1102 max mem: 3536 +Test: [1140/2387] eta: 0:04:09 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1609 (0.2943) acc1: 95.8333 (87.2998) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2076 data: 0.1142 max mem: 3536 +Test: [1150/2387] eta: 0:04:07 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.2074 (0.2933) acc1: 94.7917 (87.3817) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2015 data: 0.1083 max mem: 3536 +Test: [1160/2387] eta: 0:04:05 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1188 (0.2916) acc1: 98.6979 (87.4836) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1899 data: 0.0976 max mem: 3536 +Test: [1170/2387] eta: 0:04:03 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1083 (0.2902) acc1: 99.4792 (87.5816) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1976 data: 0.1052 max mem: 3536 +Test: [1180/2387] eta: 0:04:01 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1179 (0.2889) acc1: 99.2188 (87.6676) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1930 data: 0.1010 max mem: 3536 +Test: [1190/2387] eta: 0:03:59 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1154 (0.2875) acc1: 99.2188 (87.7633) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1836 data: 0.0920 max mem: 3536 +Test: [1200/2387] eta: 0:03:57 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0940 (0.2858) acc1: 100.0000 (87.8651) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1911 data: 0.1004 max mem: 3536 +Test: [1210/2387] eta: 0:03:55 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0806 (0.2842) acc1: 100.0000 (87.9645) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1938 data: 0.0997 max mem: 3536 +Test: [1220/2387] eta: 0:03:53 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0651 (0.2824) acc1: 100.0000 (88.0616) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1938 data: 0.0991 max mem: 3536 +Test: [1230/2387] eta: 0:03:50 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0565 (0.2805) acc1: 100.0000 (88.1586) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1886 data: 0.0961 max mem: 3536 +Test: [1240/2387] eta: 0:03:48 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0559 (0.2787) acc1: 100.0000 (88.2536) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1933 data: 0.1013 max mem: 3536 +Test: [1250/2387] eta: 0:03:46 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0566 (0.2770) acc1: 100.0000 (88.3464) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2045 data: 0.1140 max mem: 3536 +Test: [1260/2387] eta: 0:03:44 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0588 (0.2752) acc1: 100.0000 (88.4382) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2018 data: 0.1107 max mem: 3536 +Test: [1270/2387] eta: 0:03:42 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0691 (0.2737) acc1: 100.0000 (88.5253) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1946 data: 0.1031 max mem: 3536 +Test: [1280/2387] eta: 0:03:40 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0745 (0.2722) acc1: 99.7396 (88.6132) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1823 data: 0.0902 max mem: 3536 +Test: [1290/2387] eta: 0:03:38 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0793 (0.2708) acc1: 99.7396 (88.6919) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1652 data: 0.0709 max mem: 3536 +Test: [1300/2387] eta: 0:03:36 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0896 (0.2695) acc1: 99.2188 (88.7717) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1650 data: 0.0701 max mem: 3536 +Test: [1310/2387] eta: 0:03:33 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0896 (0.2681) acc1: 99.7396 (88.8519) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1681 data: 0.0755 max mem: 3536 +Test: [1320/2387] eta: 0:03:31 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0825 (0.2667) acc1: 99.7396 (88.9308) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1612 data: 0.0699 max mem: 3536 +Test: [1330/2387] eta: 0:03:29 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0961 (0.2658) acc1: 99.7396 (88.9977) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1603 data: 0.0667 max mem: 3536 +Test: [1340/2387] eta: 0:03:26 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1522 (0.2651) acc1: 98.6979 (89.0493) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1514 data: 0.0582 max mem: 3536 +Test: [1350/2387] eta: 0:03:24 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0538 (0.2635) acc1: 100.0000 (89.1304) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1409 data: 0.0503 max mem: 3537 +Test: [1360/2387] eta: 0:03:22 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0510 (0.2620) acc1: 100.0000 (89.2102) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1443 data: 0.0528 max mem: 3537 +Test: [1370/2387] eta: 0:03:19 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0495 (0.2604) acc1: 100.0000 (89.2889) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1539 data: 0.0606 max mem: 3537 +Test: [1380/2387] eta: 0:03:17 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0483 (0.2589) acc1: 100.0000 (89.3665) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1500 data: 0.0571 max mem: 3537 +Test: [1390/2387] eta: 0:03:15 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0474 (0.2574) acc1: 100.0000 (89.4429) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1512 data: 0.0602 max mem: 3537 +Test: [1400/2387] eta: 0:03:12 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0469 (0.2559) acc1: 100.0000 (89.5183) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1546 data: 0.0634 max mem: 3537 +Test: [1410/2387] eta: 0:03:10 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0481 (0.2547) acc1: 100.0000 (89.5881) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1654 data: 0.0725 max mem: 3537 +Test: [1420/2387] eta: 0:03:09 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1052 (0.2539) acc1: 99.4792 (89.6471) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1950 data: 0.1026 max mem: 3537 +Test: [1430/2387] eta: 0:03:06 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1146 (0.2529) acc1: 98.4375 (89.7100) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1772 data: 0.0854 max mem: 3537 +Test: [1440/2387] eta: 0:03:04 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1146 (0.2523) acc1: 98.4375 (89.7622) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1429 data: 0.0508 max mem: 3537 +Test: [1450/2387] eta: 0:03:02 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1716 (0.2518) acc1: 97.3958 (89.8035) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1673 data: 0.0753 max mem: 3537 +Test: [1460/2387] eta: 0:03:00 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1036 (0.2507) acc1: 99.2188 (89.8707) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1946 data: 0.1026 max mem: 3537 +Test: [1470/2387] eta: 0:02:58 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0843 (0.2496) acc1: 100.0000 (89.9363) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1925 data: 0.0976 max mem: 3537 +Test: [1480/2387] eta: 0:02:56 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0823 (0.2485) acc1: 99.7396 (90.0001) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1921 data: 0.0957 max mem: 3537 +Test: [1490/2387] eta: 0:02:54 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0811 (0.2474) acc1: 99.4792 (90.0638) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1899 data: 0.0963 max mem: 3537 +Test: [1500/2387] eta: 0:02:52 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0810 (0.2463) acc1: 99.7396 (90.1285) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1788 data: 0.0867 max mem: 3537 +Test: [1510/2387] eta: 0:02:50 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0917 (0.2454) acc1: 99.7396 (90.1914) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1599 data: 0.0656 max mem: 3537 +Test: [1520/2387] eta: 0:02:48 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1182 (0.2447) acc1: 99.2188 (90.2381) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1459 data: 0.0464 max mem: 3537 +Test: [1530/2387] eta: 0:02:46 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1190 (0.2439) acc1: 98.6979 (90.2938) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1658 data: 0.0688 max mem: 3537 +Test: [1540/2387] eta: 0:02:44 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1252 (0.2432) acc1: 98.4375 (90.3443) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1937 data: 0.1025 max mem: 3537 +Test: [1550/2387] eta: 0:02:42 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1151 (0.2425) acc1: 98.1771 (90.3923) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2004 data: 0.1078 max mem: 3537 +Test: [1560/2387] eta: 0:02:40 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1150 (0.2417) acc1: 98.9583 (90.4450) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1939 data: 0.0985 max mem: 3537 +Test: [1570/2387] eta: 0:02:38 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1075 (0.2408) acc1: 99.4792 (90.5032) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1935 data: 0.0894 max mem: 3537 +Test: [1580/2387] eta: 0:02:36 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0990 (0.2399) acc1: 99.7396 (90.5609) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1977 data: 0.0943 max mem: 3538 +Test: [1590/2387] eta: 0:02:34 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0763 (0.2389) acc1: 100.0000 (90.6196) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1876 data: 0.0948 max mem: 3538 +Test: [1600/2387] eta: 0:02:32 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0791 (0.2381) acc1: 100.0000 (90.6649) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1897 data: 0.0921 max mem: 3538 +Test: [1610/2387] eta: 0:02:30 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0749 (0.2371) acc1: 100.0000 (90.7225) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1998 data: 0.1021 max mem: 3538 +Test: [1620/2387] eta: 0:02:28 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0769 (0.2362) acc1: 100.0000 (90.7767) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2026 data: 0.1090 max mem: 3538 +Test: [1630/2387] eta: 0:02:26 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0955 (0.2355) acc1: 99.4792 (90.8209) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2003 data: 0.1066 max mem: 3538 +Test: [1640/2387] eta: 0:02:24 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1427 (0.2350) acc1: 98.4375 (90.8645) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1945 data: 0.1016 max mem: 3538 +Test: [1650/2387] eta: 0:02:23 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1626 (0.2346) acc1: 98.1771 (90.9029) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2033 data: 0.1104 max mem: 3538 +Test: [1660/2387] eta: 0:02:21 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1626 (0.2342) acc1: 97.1354 (90.9353) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1919 data: 0.1010 max mem: 3538 +Test: [1670/2387] eta: 0:02:18 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1575 (0.2338) acc1: 97.1354 (90.9746) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1683 data: 0.0773 max mem: 3538 +Test: [1680/2387] eta: 0:02:16 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1575 (0.2334) acc1: 97.6562 (91.0063) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1630 data: 0.0713 max mem: 3538 +Test: [1690/2387] eta: 0:02:14 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1655 (0.2330) acc1: 97.3958 (91.0442) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1617 data: 0.0701 max mem: 3538 +Test: [1700/2387] eta: 0:02:12 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1510 (0.2324) acc1: 98.4375 (91.0875) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1660 data: 0.0725 max mem: 3538 +Test: [1710/2387] eta: 0:02:10 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0613 (0.2314) acc1: 99.7396 (91.1393) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1655 data: 0.0715 max mem: 3538 +Test: [1720/2387] eta: 0:02:08 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0582 (0.2304) acc1: 100.0000 (91.1899) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1841 data: 0.0905 max mem: 3538 +Test: [1730/2387] eta: 0:02:06 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0548 (0.2294) acc1: 100.0000 (91.2399) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1982 data: 0.1054 max mem: 3538 +Test: [1740/2387] eta: 0:02:04 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0548 (0.2284) acc1: 100.0000 (91.2891) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1875 data: 0.0960 max mem: 3538 +Test: [1750/2387] eta: 0:02:02 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0597 (0.2274) acc1: 100.0000 (91.3384) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1864 data: 0.0949 max mem: 3538 +Test: [1760/2387] eta: 0:02:01 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0668 (0.2266) acc1: 100.0000 (91.3836) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1942 data: 0.1034 max mem: 3538 +Test: [1770/2387] eta: 0:01:59 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0710 (0.2257) acc1: 99.7396 (91.4314) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1787 data: 0.0878 max mem: 3538 +Test: [1780/2387] eta: 0:01:57 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0585 (0.2248) acc1: 100.0000 (91.4785) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1614 data: 0.0707 max mem: 3538 +Test: [1790/2387] eta: 0:01:54 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0577 (0.2239) acc1: 100.0000 (91.5258) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1591 data: 0.0689 max mem: 3538 +Test: [1800/2387] eta: 0:01:53 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0570 (0.2230) acc1: 100.0000 (91.5722) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1727 data: 0.0817 max mem: 3538 +Test: [1810/2387] eta: 0:01:50 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0568 (0.2220) acc1: 100.0000 (91.6188) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1738 data: 0.0821 max mem: 3539 +Test: [1820/2387] eta: 0:01:48 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0623 (0.2212) acc1: 100.0000 (91.6621) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1564 data: 0.0645 max mem: 3539 +Test: [1830/2387] eta: 0:01:47 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0869 (0.2206) acc1: 99.7396 (91.7056) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1764 data: 0.0841 max mem: 3539 +Test: [1840/2387] eta: 0:01:45 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1135 (0.2200) acc1: 99.7396 (91.7496) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1930 data: 0.1014 max mem: 3539 +Test: [1850/2387] eta: 0:01:43 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1149 (0.2194) acc1: 100.0000 (91.7936) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1854 data: 0.0933 max mem: 3539 +Test: [1860/2387] eta: 0:01:41 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1166 (0.2189) acc1: 100.0000 (91.8367) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1871 data: 0.0952 max mem: 3539 +Test: [1870/2387] eta: 0:01:39 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1169 (0.2183) acc1: 99.7396 (91.8792) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1819 data: 0.0896 max mem: 3539 +Test: [1880/2387] eta: 0:01:37 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1153 (0.2178) acc1: 99.7396 (91.9215) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1901 data: 0.0969 max mem: 3539 +Test: [1890/2387] eta: 0:01:35 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1145 (0.2172) acc1: 99.7396 (91.9632) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2139 data: 0.1211 max mem: 3539 +Test: [1900/2387] eta: 0:01:33 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1170 (0.2167) acc1: 100.0000 (92.0045) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1595 data: 0.0670 max mem: 3539 +Test: [1910/2387] eta: 0:01:31 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1358 (0.2164) acc1: 99.7396 (92.0425) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1672 data: 0.0748 max mem: 3539 +Test: [1920/2387] eta: 0:01:29 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1358 (0.2157) acc1: 99.7396 (92.0827) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2298 data: 0.1381 max mem: 3539 +Test: [1930/2387] eta: 0:01:27 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0807 (0.2152) acc1: 100.0000 (92.1185) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1936 data: 0.1012 max mem: 3539 +Test: [1940/2387] eta: 0:01:25 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1042 (0.2147) acc1: 99.4792 (92.1545) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1651 data: 0.0697 max mem: 3539 +Test: [1950/2387] eta: 0:01:23 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0935 (0.2141) acc1: 99.7396 (92.1940) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2049 data: 0.1105 max mem: 3539 +Test: [1960/2387] eta: 0:01:22 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0935 (0.2137) acc1: 100.0000 (92.2242) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2299 data: 0.1372 max mem: 3539 +Test: [1970/2387] eta: 0:01:20 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1598 (0.2138) acc1: 96.0938 (92.2160) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.1969 data: 0.1036 max mem: 3539 +Test: [1980/2387] eta: 0:01:18 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1706 (0.2136) acc1: 96.0938 (92.2419) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2006 data: 0.1070 max mem: 3539 +Test: [1990/2387] eta: 0:01:16 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1706 (0.2135) acc1: 98.4375 (92.2715) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2055 data: 0.1133 max mem: 3539 +Test: [2000/2387] eta: 0:01:14 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1750 (0.2133) acc1: 98.4375 (92.3011) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1889 data: 0.0978 max mem: 3539 +Test: [2010/2387] eta: 0:01:12 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1750 (0.2131) acc1: 98.1771 (92.3279) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1999 data: 0.1082 max mem: 3539 +Test: [2020/2387] eta: 0:01:10 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1747 (0.2129) acc1: 97.9167 (92.3558) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1872 data: 0.0945 max mem: 3539 +Test: [2030/2387] eta: 0:01:08 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1761 (0.2128) acc1: 97.9167 (92.3829) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1789 data: 0.0869 max mem: 3540 +Test: [2040/2387] eta: 0:01:06 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1759 (0.2126) acc1: 97.9167 (92.4109) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1832 data: 0.0919 max mem: 3540 +Test: [2050/2387] eta: 0:01:04 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1693 (0.2123) acc1: 98.4375 (92.4423) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1752 data: 0.0831 max mem: 3540 +Test: [2060/2387] eta: 0:01:02 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1572 (0.2120) acc1: 98.6979 (92.4739) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1844 data: 0.0927 max mem: 3540 +Test: [2070/2387] eta: 0:01:00 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1542 (0.2117) acc1: 98.9583 (92.5068) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1845 data: 0.0931 max mem: 3540 +Test: [2080/2387] eta: 0:00:58 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1496 (0.2115) acc1: 99.2188 (92.5384) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1801 data: 0.0889 max mem: 3540 +Test: [2090/2387] eta: 0:00:57 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1493 (0.2112) acc1: 98.9583 (92.5698) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1985 data: 0.1077 max mem: 3540 +Test: [2100/2387] eta: 0:00:55 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1510 (0.2109) acc1: 98.6979 (92.6001) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2053 data: 0.1133 max mem: 3540 +Test: [2110/2387] eta: 0:00:53 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1501 (0.2106) acc1: 98.9583 (92.6311) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1966 data: 0.1040 max mem: 3540 +Test: [2120/2387] eta: 0:00:51 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1401 (0.2101) acc1: 99.2188 (92.6634) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1744 data: 0.0820 max mem: 3540 +Test: [2130/2387] eta: 0:00:49 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0905 (0.2095) acc1: 100.0000 (92.6973) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1705 data: 0.0788 max mem: 3540 +Test: [2140/2387] eta: 0:00:47 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0940 (0.2091) acc1: 100.0000 (92.7213) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1856 data: 0.0953 max mem: 3540 +Test: [2150/2387] eta: 0:00:45 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0913 (0.2086) acc1: 100.0000 (92.7548) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1824 data: 0.0897 max mem: 3540 +Test: [2160/2387] eta: 0:00:43 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0928 (0.2081) acc1: 100.0000 (92.7861) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1895 data: 0.0923 max mem: 3540 +Test: [2170/2387] eta: 0:00:41 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1149 (0.2079) acc1: 99.4792 (92.8027) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1960 data: 0.1011 max mem: 3540 +Test: [2180/2387] eta: 0:00:39 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1903 (0.2083) acc1: 95.0521 (92.7882) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.1924 data: 0.1002 max mem: 3540 +Test: [2190/2387] eta: 0:00:37 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.2150 (0.2084) acc1: 94.0104 (92.7892) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1938 data: 0.1013 max mem: 3540 +Test: [2200/2387] eta: 0:00:35 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0865 (0.2078) acc1: 98.4375 (92.8196) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1998 data: 0.1089 max mem: 3540 +Test: [2210/2387] eta: 0:00:33 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0763 (0.2072) acc1: 100.0000 (92.8510) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1919 data: 0.1008 max mem: 3540 +Test: [2220/2387] eta: 0:00:32 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0820 (0.2066) acc1: 100.0000 (92.8807) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1777 data: 0.0841 max mem: 3540 +Test: [2230/2387] eta: 0:00:30 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0799 (0.2061) acc1: 100.0000 (92.9123) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1707 data: 0.0773 max mem: 3540 +Test: [2240/2387] eta: 0:00:28 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0675 (0.2055) acc1: 100.0000 (92.9425) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1659 data: 0.0743 max mem: 3540 +Test: [2250/2387] eta: 0:00:26 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0818 (0.2050) acc1: 99.7396 (92.9718) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1822 data: 0.0911 max mem: 3540 +Test: [2260/2387] eta: 0:00:24 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0976 (0.2046) acc1: 99.7396 (93.0012) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1825 data: 0.0913 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0840 (0.2040) acc1: 100.0000 (93.0320) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1700 data: 0.0779 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0864 (0.2037) acc1: 100.0000 (93.0494) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1656 data: 0.0740 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0919 (0.2032) acc1: 100.0000 (93.0788) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1603 data: 0.0692 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.0919 (0.2027) acc1: 100.0000 (93.1065) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1594 data: 0.0687 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1019 (0.2024) acc1: 99.7396 (93.1316) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1605 data: 0.0685 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1344 (0.2022) acc1: 98.6979 (93.1531) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1510 data: 0.0578 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1159 (0.2017) acc1: 99.4792 (93.1817) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1489 data: 0.0574 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1042 (0.2015) acc1: 99.7396 (93.2032) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1502 data: 0.0585 max mem: 3541 +Test: [2350/2387] eta: 0:00:07 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1331 (0.2013) acc1: 98.4375 (93.2169) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1356 data: 0.0436 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1271 (0.2009) acc1: 98.9583 (93.2422) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1337 data: 0.0426 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1271 (0.2007) acc1: 99.2188 (93.2588) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1348 data: 0.0444 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.1853 (0.2009) acc1: 96.0938 (93.2576) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1312 data: 0.0406 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 70.776 acc5_pos: 100.000 loss: 0.2317 (0.2010) acc1: 93.2292 (93.2564) acc5: 100.0000 (100.0000) acc1_neg: 93.023 acc5_neg: 100.000 time: 0.1241 data: 0.0360 max mem: 3541 +Test: Total time: 0:07:31 (0.1893 s / it) +* Overall Acc@1 93.256 Acc@5 100.000 + target == real Acc@1 32.202 Acc@5 100.000 + target == fake Acc@1 97.544 Acc@5 100.000 + loss 0.201 + PR-AUC 0.9921 ROC-AUC 0.9009 EER 18.11% (thr=0.8467) +Accuracy of the network on the 916267 test images: 93.3% +[rank0]:[W726 04:36:21.448078412 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:36:42.219000 1180957 torch/distributed/run.py:766] +W0726 04:36:42.219000 1180957 torch/distributed/run.py:766] ***************************************** +W0726 04:36:42.219000 1180957 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:36:42.219000 1180957 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:37:11.915727556 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:37:11.106311141 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:37:11.161578190 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:37:11.216781902 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:37:11.315023588 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:37:11.539756613 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:37:11.571150266 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W726 04:37:11.571172582 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n11_seed7_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n11_classseed7_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:06:33 acc1_pos: 25.260 acc5_pos: 100.000 loss: 1.1677 (1.1677) acc1: 25.2604 (25.2604) acc5: 100.0000 (100.0000) time: 7.7235 data: 5.7613 max mem: 3531 +Test: [ 10/517] eta: 0:06:38 acc1_pos: 14.062 acc5_pos: 100.000 loss: 1.7181 (1.6799) acc1: 5.2083 (8.7358) acc5: 100.0000 (100.0000) time: 0.7853 data: 0.5241 max mem: 3531 +Test: [ 20/517] eta: 0:03:47 acc1_pos: 14.844 acc5_pos: 100.000 loss: 1.6325 (1.5808) acc1: 8.5938 (11.1235) acc5: 100.0000 (100.0000) time: 0.0942 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:03:13 acc1_pos: 0.260 acc5_pos: 100.000 loss: 1.5457 (1.6227) acc1: 7.0312 (10.5091) acc5: 100.0000 (100.0000) time: 0.1845 data: 0.0901 max mem: 3531 +Test: [ 40/517] eta: 0:02:53 acc1_pos: 13.281 acc5_pos: 100.000 loss: 1.6099 (1.6247) acc1: 5.4688 (9.9784) acc5: 100.0000 (100.0000) time: 0.2661 data: 0.1751 max mem: 3531 +Test: [ 50/517] eta: 0:02:48 acc1_pos: 28.899 acc5_pos: 100.000 loss: 1.4584 (1.5219) acc1: 10.4167 (14.8335) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.3038 data: 0.2110 max mem: 3531 +Test: [ 60/517] eta: 0:02:35 acc1_pos: 28.899 acc5_pos: 100.000 loss: 0.3600 (1.3214) acc1: 87.7604 (27.7365) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2888 data: 0.1941 max mem: 3531 +Test: [ 70/517] eta: 0:02:27 acc1_pos: 28.899 acc5_pos: 100.000 loss: 0.1614 (1.1541) acc1: 98.1771 (37.8154) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1588 max mem: 3531 +Test: [ 80/517] eta: 0:02:21 acc1_pos: 28.899 acc5_pos: 100.000 loss: 0.1250 (1.0277) acc1: 100.0000 (45.4925) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2749 data: 0.1758 max mem: 3531 +Test: [ 90/517] eta: 0:02:17 acc1_pos: 28.899 acc5_pos: 100.000 loss: 0.1297 (0.9305) acc1: 100.0000 (51.4824) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2849 data: 0.1884 max mem: 3531 +Test: [100/517] eta: 0:02:10 acc1_pos: 28.899 acc5_pos: 100.000 loss: 0.1335 (0.8517) acc1: 100.0000 (56.2835) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2649 data: 0.1714 max mem: 3531 +Test: [110/517] eta: 0:02:05 acc1_pos: 28.899 acc5_pos: 100.000 loss: 0.1468 (0.7897) acc1: 100.0000 (60.1750) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1584 max mem: 3531 +Test: [120/517] eta: 0:02:00 acc1_pos: 28.899 acc5_pos: 100.000 loss: 0.1515 (0.7369) acc1: 99.7396 (63.4470) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2649 data: 0.1687 max mem: 3531 +Test: [130/517] eta: 0:01:56 acc1_pos: 28.899 acc5_pos: 100.000 loss: 0.1487 (0.6916) acc1: 99.7396 (66.2134) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2649 data: 0.1685 max mem: 3531 +Test: [140/517] eta: 0:01:52 acc1_pos: 28.899 acc5_pos: 100.000 loss: 0.1548 (0.6549) acc1: 99.7396 (68.5875) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2649 data: 0.1674 max mem: 3531 +Test: [150/517] eta: 0:01:48 acc1_pos: 28.899 acc5_pos: 100.000 loss: 0.1756 (0.6231) acc1: 99.7396 (70.6488) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1560 max mem: 3531 +Test: [160/517] eta: 0:01:44 acc1_pos: 28.899 acc5_pos: 100.000 loss: 0.1803 (0.5964) acc1: 100.0000 (72.4654) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1563 max mem: 3531 +Test: [170/517] eta: 0:01:41 acc1_pos: 28.899 acc5_pos: 100.000 loss: 0.1874 (0.5726) acc1: 100.0000 (74.0451) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2549 data: 0.1581 max mem: 3531 +Test: [180/517] eta: 0:01:37 acc1_pos: 28.899 acc5_pos: 100.000 loss: 0.1892 (0.5520) acc1: 99.2188 (75.3741) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2549 data: 0.1559 max mem: 3531 +Test: [190/517] eta: 0:01:33 acc1_pos: 28.899 acc5_pos: 100.000 loss: 0.1937 (0.5339) acc1: 98.9583 (76.6048) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2499 data: 0.1532 max mem: 3531 +Test: [200/517] eta: 0:01:30 acc1_pos: 28.899 acc5_pos: 100.000 loss: 0.1917 (0.5157) acc1: 99.4792 (77.7324) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2549 data: 0.1593 max mem: 3531 +Test: [210/517] eta: 0:01:27 acc1_pos: 28.899 acc5_pos: 100.000 loss: 0.1855 (0.5002) acc1: 99.4792 (78.7532) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2599 data: 0.1633 max mem: 3532 +Test: [220/517] eta: 0:01:24 acc1_pos: 28.899 acc5_pos: 100.000 loss: 0.1880 (0.4861) acc1: 98.9583 (79.6651) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2649 data: 0.1673 max mem: 3532 +Test: [230/517] eta: 0:01:20 acc1_pos: 28.899 acc5_pos: 100.000 loss: 0.2090 (0.4760) acc1: 98.9583 (80.4338) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2599 data: 0.1613 max mem: 3532 +Test: [240/517] eta: 0:01:17 acc1_pos: 28.899 acc5_pos: 100.000 loss: 0.2256 (0.4649) acc1: 99.2188 (81.2262) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2559 data: 0.1588 max mem: 3532 +Test: [250/517] eta: 0:01:15 acc1_pos: 28.899 acc5_pos: 100.000 loss: 0.1902 (0.4537) acc1: 99.7396 (81.9493) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2749 data: 0.1812 max mem: 3532 +Test: [260/517] eta: 0:01:12 acc1_pos: 28.899 acc5_pos: 100.000 loss: 0.1801 (0.4433) acc1: 99.7396 (82.6269) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2788 data: 0.1861 max mem: 3532 +Test: [270/517] eta: 0:01:09 acc1_pos: 28.899 acc5_pos: 100.000 loss: 0.1836 (0.4341) acc1: 99.4792 (83.2219) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2749 data: 0.1783 max mem: 3532 +Test: [280/517] eta: 0:01:06 acc1_pos: 28.899 acc5_pos: 100.000 loss: 0.2003 (0.4261) acc1: 97.9167 (83.7133) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2700 data: 0.1713 max mem: 3532 +Test: [290/517] eta: 0:01:03 acc1_pos: 28.899 acc5_pos: 100.000 loss: 0.2247 (0.4195) acc1: 97.9167 (84.2068) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2700 data: 0.1741 max mem: 3532 +Test: [300/517] eta: 0:01:00 acc1_pos: 28.899 acc5_pos: 100.000 loss: 0.2219 (0.4124) acc1: 98.1771 (84.6683) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2649 data: 0.1694 max mem: 3532 +Test: [310/517] eta: 0:00:57 acc1_pos: 28.899 acc5_pos: 100.000 loss: 0.2081 (0.4061) acc1: 98.1771 (85.1035) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2598 data: 0.1621 max mem: 3532 +Test: [320/517] eta: 0:00:54 acc1_pos: 28.899 acc5_pos: 100.000 loss: 0.1993 (0.3990) acc1: 98.9583 (85.5262) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2764 data: 0.1814 max mem: 3532 +Test: [330/517] eta: 0:00:52 acc1_pos: 28.899 acc5_pos: 100.000 loss: 0.1455 (0.3913) acc1: 100.0000 (85.9595) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2654 data: 0.1743 max mem: 3532 +Test: [340/517] eta: 0:00:49 acc1_pos: 28.899 acc5_pos: 100.000 loss: 0.1395 (0.3839) acc1: 100.0000 (86.3560) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2634 data: 0.1722 max mem: 3532 +Test: [350/517] eta: 0:00:46 acc1_pos: 28.899 acc5_pos: 100.000 loss: 0.1488 (0.3784) acc1: 99.2188 (86.6972) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2644 data: 0.1687 max mem: 3532 +Test: [360/517] eta: 0:00:43 acc1_pos: 28.899 acc5_pos: 100.000 loss: 0.1835 (0.3728) acc1: 99.2188 (87.0268) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2599 data: 0.1607 max mem: 3532 +Test: [370/517] eta: 0:00:40 acc1_pos: 28.899 acc5_pos: 100.000 loss: 0.1336 (0.3659) acc1: 100.0000 (87.3751) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2500 data: 0.1522 max mem: 3532 +Test: [380/517] eta: 0:00:37 acc1_pos: 28.899 acc5_pos: 100.000 loss: 0.1148 (0.3591) acc1: 100.0000 (87.7057) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2449 data: 0.1477 max mem: 3532 +Test: [390/517] eta: 0:00:34 acc1_pos: 28.899 acc5_pos: 100.000 loss: 0.1092 (0.3528) acc1: 100.0000 (88.0195) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2563 data: 0.1591 max mem: 3532 +Test: [400/517] eta: 0:00:32 acc1_pos: 28.899 acc5_pos: 100.000 loss: 0.1131 (0.3474) acc1: 100.0000 (88.3066) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2556 data: 0.1606 max mem: 3532 +Test: [410/517] eta: 0:00:29 acc1_pos: 28.899 acc5_pos: 100.000 loss: 0.1320 (0.3423) acc1: 100.0000 (88.5898) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2584 data: 0.1630 max mem: 3532 +Test: [420/517] eta: 0:00:26 acc1_pos: 28.899 acc5_pos: 100.000 loss: 0.1780 (0.3392) acc1: 100.0000 (88.8225) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2592 data: 0.1616 max mem: 3532 +Test: [430/517] eta: 0:00:23 acc1_pos: 28.899 acc5_pos: 100.000 loss: 0.2195 (0.3372) acc1: 96.8750 (88.9350) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2652 data: 0.1682 max mem: 3532 +Test: [440/517] eta: 0:00:21 acc1_pos: 28.899 acc5_pos: 100.000 loss: 0.2081 (0.3335) acc1: 97.3958 (89.1694) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2560 data: 0.1600 max mem: 3533 +Test: [450/517] eta: 0:00:18 acc1_pos: 28.899 acc5_pos: 100.000 loss: 0.1788 (0.3303) acc1: 98.9583 (89.3835) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2596 data: 0.1630 max mem: 3533 +Test: [460/517] eta: 0:00:15 acc1_pos: 28.899 acc5_pos: 100.000 loss: 0.1958 (0.3275) acc1: 98.9583 (89.5879) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2545 data: 0.1576 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 28.899 acc5_pos: 100.000 loss: 0.1876 (0.3240) acc1: 99.4792 (89.8034) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2358 data: 0.1398 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 28.899 acc5_pos: 100.000 loss: 0.1690 (0.3217) acc1: 99.4792 (89.9932) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2038 data: 0.1112 max mem: 3533 +Test: [490/517] eta: 0:00:07 acc1_pos: 28.899 acc5_pos: 100.000 loss: 0.2124 (0.3195) acc1: 98.6979 (90.1689) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1721 data: 0.0823 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 28.899 acc5_pos: 100.000 loss: 0.1658 (0.3160) acc1: 99.7396 (90.3630) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1776 data: 0.0878 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 28.899 acc5_pos: 100.000 loss: 0.1755 (0.3143) acc1: 100.0000 (90.5119) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1761 data: 0.0864 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 28.899 acc5_pos: 100.000 loss: 0.2266 (0.3135) acc1: 98.9583 (90.5653) acc5: 100.0000 (100.0000) acc1_neg: 89.753 acc5_neg: 100.000 time: 0.1776 data: 0.0863 max mem: 3533 +Test: Total time: 0:02:15 (0.2623 s / it) +* Overall Acc@1 90.565 Acc@5 100.000 + target == real Acc@1 10.810 Acc@5 100.000 + target == fake Acc@1 98.839 Acc@5 100.000 + loss 0.313 + PR-AUC 0.9435 ROC-AUC 0.6766 EER 37.63% (thr=0.8452) +Accuracy of the network on the 198427 test images: 90.6% +[rank0]:[W726 04:39:35.313702927 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 04:39:55.078000 1181531 torch/distributed/run.py:766] +W0726 04:39:55.078000 1181531 torch/distributed/run.py:766] ***************************************** +W0726 04:39:55.078000 1181531 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:39:55.078000 1181531 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:40:24.055799874 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:40:24.661881118 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:40:24.691473619 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:40:24.749448638 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:40:24.792195378 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W726 04:40:24.796731382 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:40:25.926162413 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:40:25.073155877 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n11_seed7_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n11_classseed7_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:44:12 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.0694 (2.0694) acc1: 0.0000 (0.0000) acc5: 100.0000 (100.0000) time: 6.8177 data: 5.2458 max mem: 3531 +Test: [ 10/389] eta: 0:04:27 acc1_pos: 52.083 acc5_pos: 100.000 loss: 1.4564 (1.5508) acc1: 6.5104 (10.4640) acc5: 100.0000 (100.0000) time: 0.7070 data: 0.4773 max mem: 3531 +Test: [ 20/389] eta: 0:02:34 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.4219 (1.6078) acc1: 7.2917 (17.2495) acc5: 100.0000 (100.0000) time: 0.0979 data: 0.0004 max mem: 3531 +Test: [ 30/389] eta: 0:02:06 acc1_pos: 0.521 acc5_pos: 100.000 loss: 1.9488 (1.7531) acc1: 0.2604 (12.1220) acc5: 100.0000 (100.0000) time: 0.1562 data: 0.0588 max mem: 3531 +Test: [ 40/389] eta: 0:01:57 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.8660 (1.7309) acc1: 0.2604 (11.6743) acc5: 100.0000 (100.0000) time: 0.2499 data: 0.1521 max mem: 3531 +Test: [ 50/389] eta: 0:01:45 acc1_pos: 33.073 acc5_pos: 100.000 loss: 1.6316 (1.6838) acc1: 1.8229 (11.9792) acc5: 100.0000 (100.0000) time: 0.2486 data: 0.1511 max mem: 3531 +Test: [ 60/389] eta: 0:01:35 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.4696 (1.6900) acc1: 2.3438 (12.4232) acc5: 100.0000 (100.0000) time: 0.1949 data: 0.0998 max mem: 3531 +Test: [ 70/389] eta: 0:01:26 acc1_pos: 27.083 acc5_pos: 100.000 loss: 1.6012 (1.7006) acc1: 2.3438 (12.2836) acc5: 100.0000 (100.0000) time: 0.1657 data: 0.0694 max mem: 3531 +Test: [ 80/389] eta: 0:01:20 acc1_pos: 50.521 acc5_pos: 100.000 loss: 1.4575 (1.6656) acc1: 20.3125 (14.4965) acc5: 100.0000 (100.0000) time: 0.1680 data: 0.0726 max mem: 3531 +Test: [ 90/389] eta: 0:01:15 acc1_pos: 23.438 acc5_pos: 100.000 loss: 1.2211 (1.6080) acc1: 25.0000 (16.4749) acc5: 100.0000 (100.0000) time: 0.1944 data: 0.1014 max mem: 3531 +Test: [100/389] eta: 0:01:11 acc1_pos: 88.021 acc5_pos: 100.000 loss: 1.3293 (1.6043) acc1: 16.4062 (17.0147) acc5: 100.0000 (100.0000) time: 0.1921 data: 0.0986 max mem: 3531 +Test: [110/389] eta: 0:01:07 acc1_pos: 85.156 acc5_pos: 100.000 loss: 1.4146 (1.5585) acc1: 31.5104 (20.3055) acc5: 100.0000 (100.0000) time: 0.1979 data: 0.1037 max mem: 3531 +Test: [120/389] eta: 0:01:03 acc1_pos: 34.375 acc5_pos: 100.000 loss: 0.8986 (1.5262) acc1: 39.3229 (21.5694) acc5: 100.0000 (100.0000) time: 0.1869 data: 0.0938 max mem: 3531 +Test: [130/389] eta: 0:01:00 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.8914 (1.4597) acc1: 48.9583 (25.0318) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1768 data: 0.0849 max mem: 3531 +Test: [140/389] eta: 0:00:56 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1560 (1.3735) acc1: 98.6979 (29.5490) acc5: 100.0000 (100.0000) acc1_neg: 45.573 acc5_neg: 100.000 time: 0.1595 data: 0.0653 max mem: 3531 +Test: [150/389] eta: 0:00:53 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.2033 (1.3043) acc1: 99.4792 (33.2764) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1706 data: 0.0762 max mem: 3531 +Test: [160/389] eta: 0:00:50 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.2235 (1.2350) acc1: 99.2188 (37.2412) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1900 data: 0.0975 max mem: 3531 +Test: [170/389] eta: 0:00:47 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1660 (1.1726) acc1: 100.0000 (40.9052) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1697 data: 0.0760 max mem: 3531 +Test: [180/389] eta: 0:00:46 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1636 (1.1173) acc1: 100.0000 (44.0104) acc5: 100.0000 (100.0000) acc1_neg: 75.521 acc5_neg: 100.000 time: 0.2166 data: 0.1228 max mem: 3531 +Test: [190/389] eta: 0:00:43 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1596 (1.0720) acc1: 99.2188 (46.6146) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1971 data: 0.1044 max mem: 3531 +Test: [200/389] eta: 0:00:40 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1505 (1.0281) acc1: 99.4792 (49.1034) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1361 data: 0.0442 max mem: 3531 +Test: [210/389] eta: 0:00:38 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1165 (0.9867) acc1: 100.0000 (51.4625) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1835 data: 0.0907 max mem: 3532 +Test: [220/389] eta: 0:00:35 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1180 (0.9472) acc1: 100.0000 (53.6517) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2199 data: 0.1264 max mem: 3532 +Test: [230/389] eta: 0:00:33 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1224 (0.9136) acc1: 100.0000 (55.6480) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1934 data: 0.0991 max mem: 3532 +Test: [240/389] eta: 0:00:31 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.2056 (0.8875) acc1: 98.9583 (57.2085) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1835 data: 0.0908 max mem: 3532 +Test: [250/389] eta: 0:00:29 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1939 (0.8595) acc1: 98.4375 (58.8811) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1871 data: 0.0961 max mem: 3532 +Test: [260/389] eta: 0:00:26 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1212 (0.8359) acc1: 99.7396 (60.2720) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1933 data: 0.1012 max mem: 3532 +Test: [270/389] eta: 0:00:24 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.2197 (0.8132) acc1: 98.6979 (61.6563) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1890 data: 0.0969 max mem: 3532 +Test: [280/389] eta: 0:00:22 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1294 (0.7893) acc1: 99.7396 (63.0153) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1885 data: 0.0957 max mem: 3532 +Test: [290/389] eta: 0:00:20 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1560 (0.7703) acc1: 99.7396 (64.0571) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2036 data: 0.1105 max mem: 3532 +Test: [300/389] eta: 0:00:18 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1958 (0.7516) acc1: 99.7396 (65.2331) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1923 data: 0.1004 max mem: 3532 +Test: [310/389] eta: 0:00:16 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.2016 (0.7336) acc1: 99.7396 (66.2915) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1988 data: 0.1062 max mem: 3532 +Test: [320/389] eta: 0:00:14 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1423 (0.7154) acc1: 99.7396 (67.3279) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2019 data: 0.1096 max mem: 3532 +Test: [330/389] eta: 0:00:12 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1335 (0.6979) acc1: 100.0000 (68.3118) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1719 data: 0.0807 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1337 (0.6819) acc1: 100.0000 (69.2296) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1601 data: 0.0689 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1979 (0.6770) acc1: 99.2188 (69.3035) acc5: 100.0000 (100.0000) acc1_neg: 46.354 acc5_neg: 100.000 time: 0.1341 data: 0.0429 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.5096 (0.6742) acc1: 73.6979 (69.2918) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.1114 data: 0.0205 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.4892 (0.6647) acc1: 77.6042 (69.7650) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1232 data: 0.0329 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1652 (0.6509) acc1: 99.7396 (70.5551) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1191 data: 0.0292 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1606 (0.6415) acc1: 100.0000 (71.1090) acc5: 100.0000 (100.0000) acc1_neg: 96.175 acc5_neg: 100.000 time: 0.1012 data: 0.0077 max mem: 3532 +Test: Total time: 0:01:15 (0.1929 s / it) +* Overall Acc@1 71.109 Acc@5 100.000 + target == real Acc@1 21.466 Acc@5 100.000 + target == fake Acc@1 94.184 Acc@5 100.000 + loss 0.641 + PR-AUC 0.7541 ROC-AUC 0.6408 EER 39.77% (thr=0.8403) +Accuracy of the network on the 149358 test images: 71.1% +[rank0]:[W726 04:41:50.999871842 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:42:08.936000 1182089 torch/distributed/run.py:766] +W0726 04:42:08.936000 1182089 torch/distributed/run.py:766] ***************************************** +W0726 04:42:08.936000 1182089 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:42:08.936000 1182089 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:42:36.137308280 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:42:36.156065904 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:42:36.618990999 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:42:36.649684360 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:42:37.914210967 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:42:37.060895189 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:42:37.127105431 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:42:37.182448802 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n11_seed7_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n11_classseed7_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:29:50 acc1_pos: 15.625 acc5_pos: 100.000 loss: 1.2936 (1.2936) acc1: 15.6250 (15.6250) acc5: 100.0000 (100.0000) time: 5.3764 data: 2.8294 max mem: 3531 +Test: [ 10/333] eta: 0:03:07 acc1_pos: 19.792 acc5_pos: 100.000 loss: 1.1588 (1.1645) acc1: 26.3021 (25.4735) acc5: 100.0000 (100.0000) time: 0.5795 data: 0.2662 max mem: 3531 +Test: [ 20/333] eta: 0:01:52 acc1_pos: 7.552 acc5_pos: 100.000 loss: 1.2200 (1.2372) acc1: 20.3125 (21.5154) acc5: 100.0000 (100.0000) time: 0.1074 data: 0.0143 max mem: 3531 +Test: [ 30/333] eta: 0:01:37 acc1_pos: 82.031 acc5_pos: 100.000 loss: 1.0265 (1.0417) acc1: 28.9062 (35.0806) acc5: 100.0000 (100.0000) time: 0.1775 data: 0.0813 max mem: 3531 +Test: [ 40/333] eta: 0:01:33 acc1_pos: 83.594 acc5_pos: 100.000 loss: 0.5510 (0.9207) acc1: 67.9688 (43.8135) acc5: 100.0000 (100.0000) time: 0.2799 data: 0.1846 max mem: 3531 +Test: [ 50/333] eta: 0:01:24 acc1_pos: 64.844 acc5_pos: 100.000 loss: 0.5226 (0.8460) acc1: 73.4375 (49.3362) acc5: 100.0000 (100.0000) time: 0.2658 data: 0.1701 max mem: 3531 +Test: [ 60/333] eta: 0:01:12 acc1_pos: 70.312 acc5_pos: 100.000 loss: 0.5226 (0.7928) acc1: 73.4375 (53.3683) acc5: 100.0000 (100.0000) time: 0.1549 data: 0.0577 max mem: 3531 +Test: [ 70/333] eta: 0:01:08 acc1_pos: 71.875 acc5_pos: 100.000 loss: 0.5111 (0.7553) acc1: 73.4375 (56.1803) acc5: 100.0000 (100.0000) time: 0.1590 data: 0.0645 max mem: 3531 +Test: [ 80/333] eta: 0:01:04 acc1_pos: 72.656 acc5_pos: 100.000 loss: 0.4929 (0.7228) acc1: 73.4375 (58.5520) acc5: 100.0000 (100.0000) time: 0.2199 data: 0.1265 max mem: 3531 +Test: [ 90/333] eta: 0:00:59 acc1_pos: 71.875 acc5_pos: 100.000 loss: 0.5442 (0.7129) acc1: 71.8750 (59.3063) acc5: 100.0000 (100.0000) time: 0.1949 data: 0.1002 max mem: 3531 +Test: [100/333] eta: 0:00:55 acc1_pos: 63.542 acc5_pos: 100.000 loss: 0.6474 (0.7086) acc1: 63.5417 (59.6122) acc5: 100.0000 (100.0000) time: 0.1759 data: 0.0816 max mem: 3531 +Test: [110/333] eta: 0:00:53 acc1_pos: 65.625 acc5_pos: 100.000 loss: 0.6474 (0.7055) acc1: 62.5000 (59.7738) acc5: 100.0000 (100.0000) time: 0.2155 data: 0.1227 max mem: 3531 +Test: [120/333] eta: 0:00:50 acc1_pos: 56.771 acc5_pos: 100.000 loss: 0.6466 (0.7021) acc1: 61.7188 (60.0336) acc5: 100.0000 (100.0000) time: 0.2259 data: 0.1328 max mem: 3531 +Test: [130/333] eta: 0:00:45 acc1_pos: 80.208 acc5_pos: 100.000 loss: 0.6354 (0.6994) acc1: 61.9792 (60.1225) acc5: 100.0000 (100.0000) time: 0.1492 data: 0.0539 max mem: 3531 +Test: [140/333] eta: 0:00:42 acc1_pos: 72.135 acc5_pos: 100.000 loss: 0.6354 (0.6987) acc1: 62.2396 (60.2652) acc5: 100.0000 (100.0000) time: 0.1134 data: 0.0180 max mem: 3531 +Test: [150/333] eta: 0:00:39 acc1_pos: 55.990 acc5_pos: 100.000 loss: 0.6674 (0.6994) acc1: 62.2396 (60.2735) acc5: 100.0000 (100.0000) time: 0.1649 data: 0.0697 max mem: 3531 +Test: [160/333] eta: 0:00:37 acc1_pos: 65.461 acc5_pos: 100.000 loss: 0.7014 (0.7048) acc1: 59.8958 (59.8942) acc5: 100.0000 (100.0000) acc1_neg: 41.667 acc5_neg: 100.000 time: 0.1843 data: 0.0855 max mem: 3531 +Test: [170/333] eta: 0:00:34 acc1_pos: 65.461 acc5_pos: 100.000 loss: 0.7685 (0.7115) acc1: 53.6458 (59.4466) acc5: 100.0000 (100.0000) acc1_neg: 58.594 acc5_neg: 100.000 time: 0.1799 data: 0.0812 max mem: 3531 +Test: [180/333] eta: 0:00:32 acc1_pos: 65.461 acc5_pos: 100.000 loss: 0.7685 (0.7131) acc1: 54.9479 (59.2973) acc5: 100.0000 (100.0000) acc1_neg: 64.062 acc5_neg: 100.000 time: 0.2149 data: 0.1213 max mem: 3531 +Test: [190/333] eta: 0:00:30 acc1_pos: 65.461 acc5_pos: 100.000 loss: 0.7119 (0.7103) acc1: 55.9896 (59.4732) acc5: 100.0000 (100.0000) acc1_neg: 58.594 acc5_neg: 100.000 time: 0.2264 data: 0.1316 max mem: 3531 +Test: [200/333] eta: 0:00:28 acc1_pos: 65.461 acc5_pos: 100.000 loss: 0.6745 (0.7087) acc1: 60.4167 (59.5823) acc5: 100.0000 (100.0000) acc1_neg: 60.417 acc5_neg: 100.000 time: 0.1763 data: 0.0813 max mem: 3531 +Test: [210/333] eta: 0:00:26 acc1_pos: 65.461 acc5_pos: 100.000 loss: 0.6911 (0.7076) acc1: 60.4167 (59.6811) acc5: 100.0000 (100.0000) acc1_neg: 72.656 acc5_neg: 100.000 time: 0.1857 data: 0.0931 max mem: 3532 +Test: [220/333] eta: 0:00:23 acc1_pos: 65.461 acc5_pos: 100.000 loss: 0.6911 (0.7092) acc1: 59.6354 (59.4398) acc5: 100.0000 (100.0000) acc1_neg: 54.427 acc5_neg: 100.000 time: 0.2188 data: 0.1264 max mem: 3532 +Test: [230/333] eta: 0:00:21 acc1_pos: 65.461 acc5_pos: 100.000 loss: 0.7783 (0.7119) acc1: 53.3854 (59.1946) acc5: 100.0000 (100.0000) acc1_neg: 56.250 acc5_neg: 100.000 time: 0.1788 data: 0.0868 max mem: 3532 +Test: [240/333] eta: 0:00:19 acc1_pos: 65.461 acc5_pos: 100.000 loss: 0.8055 (0.7242) acc1: 48.6979 (58.2977) acc5: 100.0000 (100.0000) acc1_neg: 33.333 acc5_neg: 100.000 time: 0.1600 data: 0.0641 max mem: 3532 +Test: [250/333] eta: 0:00:17 acc1_pos: 65.461 acc5_pos: 100.000 loss: 1.0242 (0.7298) acc1: 36.1979 (57.9173) acc5: 100.0000 (100.0000) acc1_neg: 61.979 acc5_neg: 100.000 time: 0.1787 data: 0.0820 max mem: 3532 +Test: [260/333] eta: 0:00:15 acc1_pos: 65.461 acc5_pos: 100.000 loss: 0.6804 (0.7277) acc1: 60.4167 (58.0769) acc5: 100.0000 (100.0000) acc1_neg: 71.354 acc5_neg: 100.000 time: 0.1872 data: 0.0943 max mem: 3532 +Test: [270/333] eta: 0:00:12 acc1_pos: 65.461 acc5_pos: 100.000 loss: 0.6372 (0.7236) acc1: 65.8854 (58.4160) acc5: 100.0000 (100.0000) acc1_neg: 65.885 acc5_neg: 100.000 time: 0.1999 data: 0.1043 max mem: 3532 +Test: [280/333] eta: 0:00:10 acc1_pos: 65.461 acc5_pos: 100.000 loss: 0.6372 (0.7255) acc1: 65.6250 (58.2685) acc5: 100.0000 (100.0000) acc1_neg: 49.219 acc5_neg: 100.000 time: 0.1896 data: 0.0944 max mem: 3532 +Test: [290/333] eta: 0:00:08 acc1_pos: 65.461 acc5_pos: 100.000 loss: 0.8591 (0.7313) acc1: 47.1354 (57.8519) acc5: 100.0000 (100.0000) acc1_neg: 38.281 acc5_neg: 100.000 time: 0.1811 data: 0.0900 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 65.461 acc5_pos: 100.000 loss: 0.9016 (0.7378) acc1: 44.7917 (57.3695) acc5: 100.0000 (100.0000) acc1_neg: 45.833 acc5_neg: 100.000 time: 0.2079 data: 0.1163 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 65.461 acc5_pos: 100.000 loss: 0.8964 (0.7369) acc1: 45.3125 (57.4583) acc5: 100.0000 (100.0000) acc1_neg: 39.323 acc5_neg: 100.000 time: 0.1957 data: 0.1038 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 65.461 acc5_pos: 100.000 loss: 0.8363 (0.7405) acc1: 47.6562 (57.1870) acc5: 100.0000 (100.0000) acc1_neg: 60.417 acc5_neg: 100.000 time: 0.1586 data: 0.0678 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 65.461 acc5_pos: 100.000 loss: 0.7855 (0.7408) acc1: 52.6042 (57.1626) acc5: 100.0000 (100.0000) acc1_neg: 52.604 acc5_neg: 100.000 time: 0.1375 data: 0.0465 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 65.461 acc5_pos: 100.000 loss: 0.7623 (0.7407) acc1: 54.4271 (57.1733) acc5: 100.0000 (100.0000) acc1_neg: 59.155 acc5_neg: 100.000 time: 0.1369 data: 0.0427 max mem: 3532 +Test: Total time: 0:01:06 (0.2002 s / it) +* Overall Acc@1 57.173 Acc@5 100.000 + target == real Acc@1 60.336 Acc@5 100.000 + target == fake Acc@1 54.358 Acc@5 100.000 + loss 0.741 + PR-AUC 0.6118 ROC-AUC 0.6089 EER 42.45% (thr=0.4749) +Accuracy of the network on the 127843 test images: 57.2% +[rank0]:[W726 04:43:53.230155366 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:44:12.221000 1182649 torch/distributed/run.py:766] +W0726 04:44:12.221000 1182649 torch/distributed/run.py:766] ***************************************** +W0726 04:44:12.221000 1182649 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:44:12.221000 1182649 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:44:40.071664282 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:44:40.420854201 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:44:40.504845323 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:44:40.854048702 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:44:41.127073154 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:44:41.177271922 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank4]:[W726 04:44:41.178051012 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:44:41.206216484 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n11_seed7_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n11_classseed7_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:25:39 acc1_pos: 87.240 acc5_pos: 100.000 loss: 0.3058 (0.3058) acc1: 87.2396 (87.2396) acc5: 100.0000 (100.0000) time: 6.5518 data: 5.0912 max mem: 3531 +Test: [ 10/235] eta: 0:02:34 acc1_pos: 88.542 acc5_pos: 100.000 loss: 0.3061 (0.3088) acc1: 88.2812 (87.6420) acc5: 100.0000 (100.0000) time: 0.6873 data: 0.4685 max mem: 3531 +Test: [ 20/235] eta: 0:01:29 acc1_pos: 85.417 acc5_pos: 100.000 loss: 0.3061 (0.3055) acc1: 87.7604 (87.8968) acc5: 100.0000 (100.0000) time: 0.1080 data: 0.0141 max mem: 3531 +Test: [ 30/235] eta: 0:01:18 acc1_pos: 91.927 acc5_pos: 100.000 loss: 0.2965 (0.3021) acc1: 88.0208 (88.1552) acc5: 100.0000 (100.0000) time: 0.2175 data: 0.1217 max mem: 3531 +Test: [ 40/235] eta: 0:01:07 acc1_pos: 87.760 acc5_pos: 100.000 loss: 0.2947 (0.3027) acc1: 88.0208 (88.0971) acc5: 100.0000 (100.0000) time: 0.2753 data: 0.1802 max mem: 3531 +Test: [ 50/235] eta: 0:01:00 acc1_pos: 86.198 acc5_pos: 100.000 loss: 0.2977 (0.3046) acc1: 88.0208 (87.9034) acc5: 100.0000 (100.0000) time: 0.2349 data: 0.1409 max mem: 3531 +Test: [ 60/235] eta: 0:00:54 acc1_pos: 86.458 acc5_pos: 100.000 loss: 0.3137 (0.3054) acc1: 87.2396 (87.8629) acc5: 100.0000 (100.0000) time: 0.2419 data: 0.1475 max mem: 3531 +Test: [ 70/235] eta: 0:00:49 acc1_pos: 86.154 acc5_pos: 100.000 loss: 0.3315 (0.4331) acc1: 85.4167 (80.7255) acc5: 100.0000 (100.0000) acc1_neg: 28.646 acc5_neg: 100.000 time: 0.2299 data: 0.1349 max mem: 3531 +Test: [ 80/235] eta: 0:00:44 acc1_pos: 86.154 acc5_pos: 100.000 loss: 1.3152 (0.5444) acc1: 31.7708 (74.5885) acc5: 100.0000 (100.0000) acc1_neg: 30.208 acc5_neg: 100.000 time: 0.2175 data: 0.1202 max mem: 3531 +Test: [ 90/235] eta: 0:00:40 acc1_pos: 86.154 acc5_pos: 100.000 loss: 1.3425 (0.6332) acc1: 30.7292 (69.7344) acc5: 100.0000 (100.0000) acc1_neg: 30.469 acc5_neg: 100.000 time: 0.2049 data: 0.1077 max mem: 3531 +Test: [100/235] eta: 0:00:36 acc1_pos: 86.154 acc5_pos: 100.000 loss: 1.3355 (0.7002) acc1: 30.7292 (65.9963) acc5: 100.0000 (100.0000) acc1_neg: 30.208 acc5_neg: 100.000 time: 0.1999 data: 0.1026 max mem: 3531 +Test: [110/235] eta: 0:00:32 acc1_pos: 86.154 acc5_pos: 100.000 loss: 1.3233 (0.7574) acc1: 31.2500 (62.8378) acc5: 100.0000 (100.0000) acc1_neg: 33.333 acc5_neg: 100.000 time: 0.1999 data: 0.1019 max mem: 3531 +Test: [120/235] eta: 0:00:29 acc1_pos: 86.154 acc5_pos: 100.000 loss: 1.3725 (0.8199) acc1: 25.5208 (59.6182) acc5: 100.0000 (100.0000) acc1_neg: 25.260 acc5_neg: 100.000 time: 0.1564 data: 0.0591 max mem: 3531 +Test: [130/235] eta: 0:00:25 acc1_pos: 86.154 acc5_pos: 100.000 loss: 1.5195 (0.8725) acc1: 23.6979 (56.9000) acc5: 100.0000 (100.0000) acc1_neg: 23.177 acc5_neg: 100.000 time: 0.1464 data: 0.0501 max mem: 3531 +Test: [140/235] eta: 0:00:23 acc1_pos: 86.154 acc5_pos: 100.000 loss: 1.5335 (0.9218) acc1: 22.3958 (54.4012) acc5: 100.0000 (100.0000) acc1_neg: 20.312 acc5_neg: 100.000 time: 0.1834 data: 0.0895 max mem: 3531 +Test: [150/235] eta: 0:00:20 acc1_pos: 86.154 acc5_pos: 100.000 loss: 1.5414 (0.9612) acc1: 21.8750 (52.3403) acc5: 100.0000 (100.0000) acc1_neg: 26.042 acc5_neg: 100.000 time: 0.1840 data: 0.0933 max mem: 3531 +Test: [160/235] eta: 0:00:17 acc1_pos: 86.154 acc5_pos: 100.000 loss: 1.5162 (0.9969) acc1: 21.8750 (50.4691) acc5: 100.0000 (100.0000) acc1_neg: 23.177 acc5_neg: 100.000 time: 0.1754 data: 0.0843 max mem: 3531 +Test: [170/235] eta: 0:00:14 acc1_pos: 86.154 acc5_pos: 100.000 loss: 1.4502 (0.9982) acc1: 23.6979 (50.1523) acc5: 100.0000 (100.0000) acc1_neg: 43.490 acc5_neg: 100.000 time: 0.1743 data: 0.0828 max mem: 3531 +Test: [180/235] eta: 0:00:12 acc1_pos: 86.154 acc5_pos: 100.000 loss: 0.9643 (0.9963) acc1: 47.9167 (50.0576) acc5: 100.0000 (100.0000) acc1_neg: 47.917 acc5_neg: 100.000 time: 0.1911 data: 0.0978 max mem: 3531 +Test: [190/235] eta: 0:00:10 acc1_pos: 86.154 acc5_pos: 100.000 loss: 0.9527 (0.9942) acc1: 48.1771 (49.9073) acc5: 100.0000 (100.0000) acc1_neg: 50.521 acc5_neg: 100.000 time: 0.1999 data: 0.1060 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 86.154 acc5_pos: 100.000 loss: 0.9538 (0.9937) acc1: 46.8750 (49.7461) acc5: 100.0000 (100.0000) acc1_neg: 46.875 acc5_neg: 100.000 time: 0.1951 data: 0.1026 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 86.154 acc5_pos: 100.000 loss: 0.9685 (0.9919) acc1: 46.8750 (49.6840) acc5: 100.0000 (100.0000) acc1_neg: 47.917 acc5_neg: 100.000 time: 0.1894 data: 0.0971 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 86.154 acc5_pos: 100.000 loss: 0.9544 (0.9902) acc1: 47.9167 (49.6359) acc5: 100.0000 (100.0000) acc1_neg: 46.875 acc5_neg: 100.000 time: 0.1653 data: 0.0743 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 86.154 acc5_pos: 100.000 loss: 0.9544 (0.9887) acc1: 48.1771 (49.5896) acc5: 100.0000 (100.0000) acc1_neg: 50.521 acc5_neg: 100.000 time: 0.1402 data: 0.0492 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 86.154 acc5_pos: 100.000 loss: 0.9600 (0.9893) acc1: 48.1771 (49.5633) acc5: 100.0000 (100.0000) acc1_neg: 43.478 acc5_neg: 100.000 time: 0.1370 data: 0.0492 max mem: 3532 +Test: Total time: 0:00:50 (0.2145 s / it) +* Overall Acc@1 49.563 Acc@5 100.000 + target == real Acc@1 87.865 Acc@5 100.000 + target == fake Acc@1 35.710 Acc@5 100.000 + loss 0.989 + PR-AUC 0.8614 ROC-AUC 0.7052 EER 35.10% (thr=0.2137) +Accuracy of the network on the 89879 test images: 49.6% +[rank0]:[W726 04:45:42.983495036 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:46:00.826000 1183197 torch/distributed/run.py:766] +W0726 04:46:00.826000 1183197 torch/distributed/run.py:766] ***************************************** +W0726 04:46:00.826000 1183197 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:46:00.826000 1183197 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 04:46:31.474567070 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:46:31.768886119 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:46:31.803824754 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:46:32.229642954 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:46:32.485716775 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:46:32.556211852 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank6]:[W726 04:46:32.556254955 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:46:32.585356111 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n11_seed7_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n11_classseed7_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:35:31 acc1_pos: 65.104 acc5_pos: 100.000 loss: 0.6032 (0.6032) acc1: 65.1042 (65.1042) acc5: 100.0000 (100.0000) time: 7.9240 data: 5.7298 max mem: 3531 +Test: [ 10/269] eta: 0:03:29 acc1_pos: 88.542 acc5_pos: 100.000 loss: 0.5076 (0.5048) acc1: 75.5208 (76.9650) acc5: 100.0000 (100.0000) time: 0.8103 data: 0.5212 max mem: 3531 +Test: [ 20/269] eta: 0:01:57 acc1_pos: 76.823 acc5_pos: 100.000 loss: 0.4931 (0.5054) acc1: 75.5208 (76.6989) acc5: 100.0000 (100.0000) time: 0.0994 data: 0.0004 max mem: 3531 +Test: [ 30/269] eta: 0:01:40 acc1_pos: 70.052 acc5_pos: 100.000 loss: 0.4931 (0.4933) acc1: 76.8229 (77.9234) acc5: 100.0000 (100.0000) time: 0.2049 data: 0.1064 max mem: 3531 +Test: [ 40/269] eta: 0:01:28 acc1_pos: 33.621 acc5_pos: 100.000 loss: 0.5680 (0.5694) acc1: 66.1458 (70.4840) acc5: 100.0000 (100.0000) acc1_neg: 41.406 acc5_neg: 100.000 time: 0.2999 data: 0.2032 max mem: 3531 +Test: [ 50/269] eta: 0:01:19 acc1_pos: 33.621 acc5_pos: 100.000 loss: 0.8524 (0.6378) acc1: 41.4062 (64.4097) acc5: 100.0000 (100.0000) acc1_neg: 19.271 acc5_neg: 100.000 time: 0.2699 data: 0.1748 max mem: 3531 +Test: [ 60/269] eta: 0:01:11 acc1_pos: 33.621 acc5_pos: 100.000 loss: 0.8748 (0.6945) acc1: 35.6771 (59.5287) acc5: 100.0000 (100.0000) acc1_neg: 29.948 acc5_neg: 100.000 time: 0.2449 data: 0.1512 max mem: 3531 +Test: [ 70/269] eta: 0:01:03 acc1_pos: 33.621 acc5_pos: 100.000 loss: 0.8748 (0.7189) acc1: 37.2396 (57.1706) acc5: 100.0000 (100.0000) acc1_neg: 53.125 acc5_neg: 100.000 time: 0.2199 data: 0.1242 max mem: 3531 +Test: [ 80/269] eta: 0:00:57 acc1_pos: 33.621 acc5_pos: 100.000 loss: 0.8519 (0.7279) acc1: 40.6250 (56.0024) acc5: 100.0000 (100.0000) acc1_neg: 34.635 acc5_neg: 100.000 time: 0.1853 data: 0.0866 max mem: 3531 +Test: [ 90/269] eta: 0:00:53 acc1_pos: 33.621 acc5_pos: 100.000 loss: 0.7855 (0.7377) acc1: 47.9167 (54.8735) acc5: 100.0000 (100.0000) acc1_neg: 50.781 acc5_neg: 100.000 time: 0.2249 data: 0.1282 max mem: 3531 +Test: [100/269] eta: 0:00:49 acc1_pos: 33.621 acc5_pos: 100.000 loss: 0.8337 (0.7540) acc1: 41.4062 (53.2049) acc5: 100.0000 (100.0000) acc1_neg: 34.635 acc5_neg: 100.000 time: 0.2395 data: 0.1433 max mem: 3531 +Test: [110/269] eta: 0:00:44 acc1_pos: 33.621 acc5_pos: 100.000 loss: 0.9182 (0.7718) acc1: 38.5417 (51.8839) acc5: 100.0000 (100.0000) acc1_neg: 44.271 acc5_neg: 100.000 time: 0.1925 data: 0.0970 max mem: 3531 +Test: [120/269] eta: 0:00:40 acc1_pos: 33.621 acc5_pos: 100.000 loss: 0.8515 (0.7756) acc1: 39.0625 (51.3968) acc5: 100.0000 (100.0000) acc1_neg: 34.115 acc5_neg: 100.000 time: 0.1976 data: 0.1045 max mem: 3531 +Test: [130/269] eta: 0:00:37 acc1_pos: 33.621 acc5_pos: 100.000 loss: 0.9320 (0.7955) acc1: 35.6771 (49.9424) acc5: 100.0000 (100.0000) acc1_neg: 22.656 acc5_neg: 100.000 time: 0.2023 data: 0.1074 max mem: 3531 +Test: [140/269] eta: 0:00:33 acc1_pos: 33.621 acc5_pos: 100.000 loss: 1.0525 (0.8161) acc1: 23.9583 (48.1771) acc5: 100.0000 (100.0000) acc1_neg: 22.917 acc5_neg: 100.000 time: 0.1945 data: 0.0993 max mem: 3531 +Test: [150/269] eta: 0:00:30 acc1_pos: 33.621 acc5_pos: 100.000 loss: 1.0736 (0.8308) acc1: 23.9583 (46.7526) acc5: 100.0000 (100.0000) acc1_neg: 35.677 acc5_neg: 100.000 time: 0.1920 data: 0.0991 max mem: 3531 +Test: [160/269] eta: 0:00:27 acc1_pos: 33.621 acc5_pos: 100.000 loss: 0.9351 (0.8325) acc1: 33.0729 (46.5822) acc5: 100.0000 (100.0000) acc1_neg: 62.760 acc5_neg: 100.000 time: 0.1928 data: 0.1006 max mem: 3531 +Test: [170/269] eta: 0:00:24 acc1_pos: 33.621 acc5_pos: 100.000 loss: 0.8076 (0.8310) acc1: 45.5729 (46.7791) acc5: 100.0000 (100.0000) acc1_neg: 47.656 acc5_neg: 100.000 time: 0.1948 data: 0.1018 max mem: 3531 +Test: [180/269] eta: 0:00:22 acc1_pos: 33.621 acc5_pos: 100.000 loss: 0.7281 (0.8232) acc1: 50.7812 (47.4059) acc5: 100.0000 (100.0000) acc1_neg: 44.792 acc5_neg: 100.000 time: 0.1965 data: 0.1023 max mem: 3531 +Test: [190/269] eta: 0:00:19 acc1_pos: 33.621 acc5_pos: 100.000 loss: 0.7715 (0.8272) acc1: 50.7812 (47.2240) acc5: 100.0000 (100.0000) acc1_neg: 42.969 acc5_neg: 100.000 time: 0.1829 data: 0.0896 max mem: 3531 +Test: [200/269] eta: 0:00:16 acc1_pos: 33.621 acc5_pos: 100.000 loss: 0.9841 (0.8344) acc1: 40.3646 (47.1924) acc5: 100.0000 (100.0000) acc1_neg: 78.646 acc5_neg: 100.000 time: 0.1768 data: 0.0836 max mem: 3531 +Test: [210/269] eta: 0:00:14 acc1_pos: 33.621 acc5_pos: 100.000 loss: 1.0246 (0.8404) acc1: 36.7188 (46.6775) acc5: 100.0000 (100.0000) acc1_neg: 35.417 acc5_neg: 100.000 time: 0.1826 data: 0.0876 max mem: 3532 +Test: [220/269] eta: 0:00:11 acc1_pos: 33.621 acc5_pos: 100.000 loss: 1.0152 (0.8486) acc1: 29.9479 (45.8145) acc5: 100.0000 (100.0000) acc1_neg: 29.427 acc5_neg: 100.000 time: 0.1668 data: 0.0742 max mem: 3532 +Test: [230/269] eta: 0:00:09 acc1_pos: 33.621 acc5_pos: 100.000 loss: 1.0266 (0.8502) acc1: 22.3958 (45.4827) acc5: 100.0000 (100.0000) acc1_neg: 57.552 acc5_neg: 100.000 time: 0.1566 data: 0.0660 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 33.621 acc5_pos: 100.000 loss: 0.8418 (0.8491) acc1: 44.2708 (45.5232) acc5: 100.0000 (100.0000) acc1_neg: 64.844 acc5_neg: 100.000 time: 0.1473 data: 0.0564 max mem: 3532 +Test: [250/269] eta: 0:00:04 acc1_pos: 33.621 acc5_pos: 100.000 loss: 0.8418 (0.8538) acc1: 44.2708 (45.2015) acc5: 100.0000 (100.0000) acc1_neg: 18.490 acc5_neg: 100.000 time: 0.1365 data: 0.0454 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 33.621 acc5_pos: 100.000 loss: 1.0039 (0.8604) acc1: 27.6042 (44.6230) acc5: 100.0000 (100.0000) acc1_neg: 57.552 acc5_neg: 100.000 time: 0.1273 data: 0.0351 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 33.621 acc5_pos: 100.000 loss: 0.8806 (0.8564) acc1: 37.2396 (44.8656) acc5: 100.0000 (100.0000) acc1_neg: 82.415 acc5_neg: 100.000 time: 0.1165 data: 0.0214 max mem: 3532 +Test: Total time: 0:00:58 (0.2168 s / it) +* Overall Acc@1 44.866 Acc@5 100.000 + target == real Acc@1 77.144 Acc@5 100.000 + target == fake Acc@1 40.460 Acc@5 100.000 + loss 0.856 + PR-AUC 0.9208 ROC-AUC 0.6389 EER 39.80% (thr=0.3801) +Accuracy of the network on the 103293 test images: 44.9% +[rank0]:[W726 04:47:38.257969189 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T04:47:45Z] eval_done tag=fixed500k_real_n11_seed7_resampled +[2026-07-26T04:47:46Z] eval_start tag=fixed500k_fake_n22_seed7_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed7_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T04:47:46Z +hostname=n116-041-188 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed7_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n22_seed7_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 04:47:58.821000 1183772 torch/distributed/run.py:766] +W0726 04:47:58.821000 1183772 torch/distributed/run.py:766] ***************************************** +W0726 04:47:58.821000 1183772 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:47:58.821000 1183772 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:48:26.295006952 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:48:27.041104553 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:48:27.068486321 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:48:27.256527341 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:48:27.503977207 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +| distributed init (rank 4): env:// +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:48:27.617839948 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W726 04:48:27.617906927 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank4]:[W726 04:48:27.619669004 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n22_seed7_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed7_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:26:50 acc1_pos: 57.031 acc5_pos: 100.000 loss: 0.6786 (0.6786) acc1: 57.0312 (57.0312) acc5: 100.0000 (100.0000) time: 6.8838 data: 5.3097 max mem: 3531 +Test: [ 10/234] eta: 0:02:39 acc1_pos: 28.646 acc5_pos: 100.000 loss: 0.6064 (0.6523) acc1: 64.3229 (62.0975) acc5: 100.0000 (100.0000) time: 0.7138 data: 0.4831 max mem: 3531 +Test: [ 20/234] eta: 0:01:30 acc1_pos: 94.792 acc5_pos: 100.000 loss: 0.6064 (0.6624) acc1: 53.6458 (60.2679) acc5: 100.0000 (100.0000) time: 0.0984 data: 0.0004 max mem: 3531 +Test: [ 30/234] eta: 0:01:12 acc1_pos: 69.531 acc5_pos: 100.000 loss: 0.2376 (0.5370) acc1: 94.5312 (70.3545) acc5: 100.0000 (100.0000) time: 0.1554 data: 0.0587 max mem: 3531 +Test: [ 40/234] eta: 0:01:03 acc1_pos: 97.656 acc5_pos: 100.000 loss: 0.2271 (0.5097) acc1: 94.7917 (72.2879) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1307 max mem: 3531 +Test: [ 50/234] eta: 0:00:56 acc1_pos: 97.917 acc5_pos: 100.000 loss: 0.1722 (0.4427) acc1: 96.8750 (77.1446) acc5: 100.0000 (100.0000) time: 0.2325 data: 0.1388 max mem: 3531 +Test: [ 60/234] eta: 0:00:50 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.1212 (0.3846) acc1: 98.1771 (80.8658) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2149 data: 0.1207 max mem: 3531 +Test: [ 70/234] eta: 0:00:44 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.0862 (0.3809) acc1: 100.0000 (81.2023) acc5: 100.0000 (100.0000) acc1_neg: 72.917 acc5_neg: 100.000 time: 0.1901 data: 0.0931 max mem: 3531 +Test: [ 80/234] eta: 0:00:40 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.4463 (0.3777) acc1: 77.8646 (81.6390) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1884 data: 0.0928 max mem: 3531 +Test: [ 90/234] eta: 0:00:36 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.2039 (0.3559) acc1: 93.2292 (83.1130) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.1846 data: 0.0907 max mem: 3531 +Test: [100/234] eta: 0:00:33 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.2091 (0.3667) acc1: 91.6667 (82.5263) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2094 data: 0.1138 max mem: 3531 +Test: [110/234] eta: 0:00:30 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.4849 (0.3733) acc1: 73.9583 (82.2682) acc5: 100.0000 (100.0000) acc1_neg: 67.708 acc5_neg: 100.000 time: 0.1959 data: 0.0978 max mem: 3531 +Test: [120/234] eta: 0:00:27 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.4656 (0.3833) acc1: 76.3021 (81.8010) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.1566 data: 0.0610 max mem: 3531 +Test: [130/234] eta: 0:00:24 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.4507 (0.3877) acc1: 81.7708 (81.8583) acc5: 100.0000 (100.0000) acc1_neg: 79.427 acc5_neg: 100.000 time: 0.1818 data: 0.0903 max mem: 3531 +Test: [140/234] eta: 0:00:21 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.4741 (0.4150) acc1: 79.4271 (80.6313) acc5: 100.0000 (100.0000) acc1_neg: 39.062 acc5_neg: 100.000 time: 0.1857 data: 0.0937 max mem: 3531 +Test: [150/234] eta: 0:00:19 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.5405 (0.4270) acc1: 71.3542 (79.9600) acc5: 100.0000 (100.0000) acc1_neg: 71.354 acc5_neg: 100.000 time: 0.2034 data: 0.1118 max mem: 3531 +Test: [160/234] eta: 0:00:16 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.4916 (0.4305) acc1: 71.3542 (79.7943) acc5: 100.0000 (100.0000) acc1_neg: 63.281 acc5_neg: 100.000 time: 0.2030 data: 0.1119 max mem: 3531 +Test: [170/234] eta: 0:00:14 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.6072 (0.4399) acc1: 65.3646 (79.1499) acc5: 100.0000 (100.0000) acc1_neg: 75.781 acc5_neg: 100.000 time: 0.1687 data: 0.0777 max mem: 3531 +Test: [180/234] eta: 0:00:11 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.5578 (0.4459) acc1: 67.9688 (78.7768) acc5: 100.0000 (100.0000) acc1_neg: 47.917 acc5_neg: 100.000 time: 0.1811 data: 0.0894 max mem: 3531 +Test: [190/234] eta: 0:00:09 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.6506 (0.4577) acc1: 65.1042 (77.9927) acc5: 100.0000 (100.0000) acc1_neg: 64.844 acc5_neg: 100.000 time: 0.2002 data: 0.1065 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.6506 (0.4505) acc1: 65.1042 (78.3854) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1885 data: 0.0949 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.1023 (0.4338) acc1: 99.7396 (79.4049) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1781 data: 0.0854 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.0881 (0.4185) acc1: 100.0000 (80.3014) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1378 data: 0.0463 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.1947 (0.4189) acc1: 82.0312 (80.3121) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.1156 data: 0.0256 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 95.862 acc5_pos: 100.000 loss: 0.4212 (0.4189) acc1: 82.0312 (80.3185) acc5: 100.0000 (100.0000) acc1_neg: 83.268 acc5_neg: 100.000 time: 0.1243 data: 0.0274 max mem: 3532 +Test: Total time: 0:00:48 (0.2078 s / it) +* Overall Acc@1 80.319 Acc@5 100.000 + target == real Acc@1 77.282 Acc@5 100.000 + target == fake Acc@1 81.174 Acc@5 100.000 + loss 0.419 + PR-AUC 0.9661 ROC-AUC 0.8872 EER 20.44% (thr=0.5244) +Accuracy of the network on the 89729 test images: 80.3% +[rank0]:[W726 04:49:26.099935908 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:49:46.599000 1184318 torch/distributed/run.py:766] +W0726 04:49:46.599000 1184318 torch/distributed/run.py:766] ***************************************** +W0726 04:49:46.599000 1184318 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:49:46.599000 1184318 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:50:13.659088771 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:50:13.902709601 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:50:14.193271247 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:50:14.280090351 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:50:14.320735265 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:50:14.570804692 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:50:14.581464647 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:50:14.594560367 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n22_seed7_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed7_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 5:02:20 acc1_pos: 20.833 acc5_pos: 100.000 loss: 1.1286 (1.1286) acc1: 20.8333 (20.8333) acc5: 100.0000 (100.0000) time: 7.5998 data: 5.7936 max mem: 3531 +Test: [ 10/2387] eta: 0:30:40 acc1_pos: 14.844 acc5_pos: 100.000 loss: 1.1286 (1.1014) acc1: 20.0521 (21.8750) acc5: 100.0000 (100.0000) time: 0.7744 data: 0.5271 max mem: 3531 +Test: [ 20/2387] eta: 0:18:04 acc1_pos: 17.448 acc5_pos: 100.000 loss: 1.1222 (1.1142) acc1: 19.2708 (21.7262) acc5: 100.0000 (100.0000) time: 0.1009 data: 0.0057 max mem: 3531 +Test: [ 30/2387] eta: 0:17:00 acc1_pos: 13.021 acc5_pos: 100.000 loss: 1.1266 (1.1412) acc1: 17.9688 (20.8249) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1454 max mem: 3531 +Test: [ 40/2387] eta: 0:15:24 acc1_pos: 14.844 acc5_pos: 100.000 loss: 1.2262 (1.1795) acc1: 13.0208 (18.7818) acc5: 100.0000 (100.0000) time: 0.3267 data: 0.2310 max mem: 3531 +Test: [ 50/2387] eta: 0:14:54 acc1_pos: 37.500 acc5_pos: 100.000 loss: 1.2135 (1.1678) acc1: 13.0208 (19.9857) acc5: 100.0000 (100.0000) time: 0.3053 data: 0.2146 max mem: 3531 +Test: [ 60/2387] eta: 0:14:11 acc1_pos: 53.646 acc5_pos: 100.000 loss: 1.1350 (1.1584) acc1: 21.8750 (20.8675) acc5: 100.0000 (100.0000) time: 0.3081 data: 0.2171 max mem: 3531 +Test: [ 70/2387] eta: 0:13:42 acc1_pos: 24.479 acc5_pos: 100.000 loss: 1.1914 (1.1695) acc1: 19.0104 (20.3125) acc5: 100.0000 (100.0000) time: 0.2845 data: 0.1918 max mem: 3531 +Test: [ 80/2387] eta: 0:13:23 acc1_pos: 43.490 acc5_pos: 100.000 loss: 1.1326 (1.1474) acc1: 19.7917 (21.4538) acc5: 100.0000 (100.0000) time: 0.2949 data: 0.2016 max mem: 3531 +Test: [ 90/2387] eta: 0:12:50 acc1_pos: 45.833 acc5_pos: 100.000 loss: 0.9136 (1.1179) acc1: 33.3333 (23.4060) acc5: 100.0000 (100.0000) time: 0.2654 data: 0.1731 max mem: 3531 +Test: [ 100/2387] eta: 0:12:27 acc1_pos: 36.719 acc5_pos: 100.000 loss: 0.8884 (1.0987) acc1: 36.7188 (24.6932) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1467 max mem: 3531 +Test: [ 110/2387] eta: 0:12:28 acc1_pos: 34.635 acc5_pos: 100.000 loss: 0.8574 (1.0777) acc1: 36.7188 (26.1543) acc5: 100.0000 (100.0000) time: 0.2994 data: 0.2036 max mem: 3531 +Test: [ 120/2387] eta: 0:12:11 acc1_pos: 29.427 acc5_pos: 100.000 loss: 0.9246 (1.0670) acc1: 35.1562 (26.7992) acc5: 100.0000 (100.0000) time: 0.3004 data: 0.2045 max mem: 3531 +Test: [ 130/2387] eta: 0:11:55 acc1_pos: 45.833 acc5_pos: 100.000 loss: 0.8869 (1.0534) acc1: 35.1562 (27.4253) acc5: 100.0000 (100.0000) time: 0.2503 data: 0.1559 max mem: 3531 +Test: [ 140/2387] eta: 0:11:41 acc1_pos: 34.896 acc5_pos: 100.000 loss: 0.8729 (1.0452) acc1: 32.8125 (27.6393) acc5: 100.0000 (100.0000) time: 0.2495 data: 0.1556 max mem: 3531 +Test: [ 150/2387] eta: 0:11:33 acc1_pos: 12.760 acc5_pos: 100.000 loss: 0.8943 (1.0333) acc1: 32.2917 (28.2026) acc5: 100.0000 (100.0000) time: 0.2645 data: 0.1685 max mem: 3531 +Test: [ 160/2387] eta: 0:11:20 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.8259 (1.0093) acc1: 41.9271 (30.1339) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2598 data: 0.1619 max mem: 3531 +Test: [ 170/2387] eta: 0:11:06 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2401 (0.9635) acc1: 95.0521 (34.0476) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2307 data: 0.1344 max mem: 3531 +Test: [ 180/2387] eta: 0:10:57 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2255 (0.9228) acc1: 96.8750 (37.5144) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2349 data: 0.1391 max mem: 3531 +Test: [ 190/2387] eta: 0:10:46 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2234 (0.8861) acc1: 96.8750 (40.6195) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2391 data: 0.1433 max mem: 3531 +Test: [ 200/2387] eta: 0:10:32 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2210 (0.8531) acc1: 97.1354 (43.4339) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2099 data: 0.1139 max mem: 3531 +Test: [ 210/2387] eta: 0:10:11 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2232 (0.8234) acc1: 97.1354 (45.9728) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1519 data: 0.0560 max mem: 3532 +Test: [ 220/2387] eta: 0:10:01 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2239 (0.7962) acc1: 97.1354 (48.2914) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1614 data: 0.0686 max mem: 3532 +Test: [ 230/2387] eta: 0:09:53 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2239 (0.7715) acc1: 97.3958 (50.4070) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2178 data: 0.1250 max mem: 3532 +Test: [ 240/2387] eta: 0:09:46 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2305 (0.7505) acc1: 96.3542 (52.2778) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2234 data: 0.1289 max mem: 3532 +Test: [ 250/2387] eta: 0:09:35 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2158 (0.7244) acc1: 96.8750 (54.1573) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2002 data: 0.1027 max mem: 3532 +Test: [ 260/2387] eta: 0:09:27 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.0772 (0.6993) acc1: 100.0000 (55.9138) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0935 max mem: 3532 +Test: [ 270/2387] eta: 0:09:18 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.0675 (0.6761) acc1: 100.0000 (57.5406) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1972 data: 0.1055 max mem: 3532 +Test: [ 280/2387] eta: 0:09:13 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.0712 (0.6547) acc1: 100.0000 (59.0516) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2103 data: 0.1185 max mem: 3532 +Test: [ 290/2387] eta: 0:09:07 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.0687 (0.6345) acc1: 100.0000 (60.4587) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2190 data: 0.1247 max mem: 3532 +Test: [ 300/2387] eta: 0:09:00 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.0656 (0.6157) acc1: 100.0000 (61.7724) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2050 data: 0.1111 max mem: 3532 +Test: [ 310/2387] eta: 0:08:53 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.0713 (0.6016) acc1: 100.0000 (62.9555) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2016 data: 0.1095 max mem: 3532 +Test: [ 320/2387] eta: 0:08:47 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.3055 (0.5926) acc1: 97.9167 (64.0357) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2043 data: 0.1101 max mem: 3532 +Test: [ 330/2387] eta: 0:08:40 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.3096 (0.5840) acc1: 97.9167 (65.0530) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1916 data: 0.0980 max mem: 3532 +Test: [ 340/2387] eta: 0:08:34 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.3078 (0.5759) acc1: 97.9167 (66.0168) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1868 data: 0.0940 max mem: 3532 +Test: [ 350/2387] eta: 0:08:29 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2966 (0.5637) acc1: 99.2188 (66.9775) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1963 data: 0.1036 max mem: 3532 +Test: [ 360/2387] eta: 0:08:22 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1759 (0.5531) acc1: 99.2188 (67.8519) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1892 data: 0.0956 max mem: 3532 +Test: [ 370/2387] eta: 0:08:16 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1866 (0.5433) acc1: 98.6979 (68.6798) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1807 data: 0.0858 max mem: 3532 +Test: [ 380/2387] eta: 0:08:10 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1876 (0.5339) acc1: 98.4375 (69.4574) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1846 data: 0.0911 max mem: 3532 +Test: [ 390/2387] eta: 0:08:04 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1863 (0.5251) acc1: 98.4375 (70.2039) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1741 data: 0.0796 max mem: 3532 +Test: [ 400/2387] eta: 0:07:59 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1838 (0.5167) acc1: 98.6979 (70.9100) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1805 data: 0.0866 max mem: 3532 +Test: [ 410/2387] eta: 0:07:54 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1838 (0.5086) acc1: 98.6979 (71.5816) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1933 data: 0.1020 max mem: 3532 +Test: [ 420/2387] eta: 0:07:49 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1839 (0.5010) acc1: 98.6979 (72.2208) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1834 data: 0.0913 max mem: 3532 +Test: [ 430/2387] eta: 0:07:44 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1900 (0.4942) acc1: 98.1771 (72.8115) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1811 data: 0.0843 max mem: 3532 +Test: [ 440/2387] eta: 0:07:39 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1940 (0.4861) acc1: 97.9167 (73.3991) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1822 data: 0.0862 max mem: 3533 +Test: [ 450/2387] eta: 0:07:34 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.0796 (0.4771) acc1: 100.0000 (73.9889) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1798 data: 0.0861 max mem: 3533 +Test: [ 460/2387] eta: 0:07:29 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.0796 (0.4685) acc1: 100.0000 (74.5532) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1738 data: 0.0809 max mem: 3533 +Test: [ 470/2387] eta: 0:07:25 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.0801 (0.4602) acc1: 100.0000 (75.0934) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1755 data: 0.0811 max mem: 3533 +Test: [ 480/2387] eta: 0:07:20 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.0804 (0.4523) acc1: 100.0000 (75.6112) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1800 data: 0.0853 max mem: 3533 +Test: [ 490/2387] eta: 0:07:16 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.0798 (0.4447) acc1: 100.0000 (76.1080) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0868 max mem: 3533 +Test: [ 500/2387] eta: 0:07:11 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.0791 (0.4375) acc1: 100.0000 (76.5849) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1746 data: 0.0819 max mem: 3533 +Test: [ 510/2387] eta: 0:07:08 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.0798 (0.4306) acc1: 100.0000 (77.0431) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1764 data: 0.0842 max mem: 3533 +Test: [ 520/2387] eta: 0:07:04 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1072 (0.4245) acc1: 100.0000 (77.4837) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1892 data: 0.0969 max mem: 3533 +Test: [ 530/2387] eta: 0:07:00 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1075 (0.4185) acc1: 100.0000 (77.9077) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1867 data: 0.0941 max mem: 3533 +Test: [ 540/2387] eta: 0:06:57 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1049 (0.4127) acc1: 100.0000 (78.3161) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1894 data: 0.0931 max mem: 3533 +Test: [ 550/2387] eta: 0:06:53 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1051 (0.4071) acc1: 100.0000 (78.7096) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1883 data: 0.0908 max mem: 3533 +Test: [ 560/2387] eta: 0:06:50 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1055 (0.4017) acc1: 100.0000 (79.0891) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1878 data: 0.0902 max mem: 3533 +Test: [ 570/2387] eta: 0:06:48 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1052 (0.3965) acc1: 100.0000 (79.4554) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2173 data: 0.1211 max mem: 3533 +Test: [ 580/2387] eta: 0:06:44 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1043 (0.3915) acc1: 100.0000 (79.8090) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2070 data: 0.1126 max mem: 3533 +Test: [ 590/2387] eta: 0:06:42 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1045 (0.3867) acc1: 100.0000 (80.1506) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1992 data: 0.1048 max mem: 3533 +Test: [ 600/2387] eta: 0:06:37 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1078 (0.3821) acc1: 100.0000 (80.4809) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0832 max mem: 3533 +Test: [ 610/2387] eta: 0:06:35 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1137 (0.3778) acc1: 100.0000 (80.8003) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1818 data: 0.0862 max mem: 3533 +Test: [ 620/2387] eta: 0:06:31 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1161 (0.3735) acc1: 100.0000 (81.1091) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1977 data: 0.1061 max mem: 3533 +Test: [ 630/2387] eta: 0:06:27 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1161 (0.3695) acc1: 100.0000 (81.4085) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1530 data: 0.0589 max mem: 3533 +Test: [ 640/2387] eta: 0:06:23 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1161 (0.3655) acc1: 100.0000 (81.6981) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1571 data: 0.0611 max mem: 3533 +Test: [ 650/2387] eta: 0:06:20 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1164 (0.3617) acc1: 100.0000 (81.9788) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1835 data: 0.0899 max mem: 3533 +Test: [ 660/2387] eta: 0:06:17 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1166 (0.3580) acc1: 100.0000 (82.2511) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1856 data: 0.0942 max mem: 3533 +Test: [ 670/2387] eta: 0:06:14 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1156 (0.3544) acc1: 100.0000 (82.5156) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1834 data: 0.0926 max mem: 3534 +Test: [ 680/2387] eta: 0:06:11 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1161 (0.3509) acc1: 100.0000 (82.7723) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1896 data: 0.0986 max mem: 3534 +Test: [ 690/2387] eta: 0:06:08 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1176 (0.3498) acc1: 100.0000 (82.8619) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1784 data: 0.0864 max mem: 3534 +Test: [ 700/2387] eta: 0:06:05 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.3380 (0.3497) acc1: 84.8958 (82.8939) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.1695 data: 0.0764 max mem: 3534 +Test: [ 710/2387] eta: 0:06:02 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.3396 (0.3495) acc1: 84.8958 (82.9249) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1843 data: 0.0904 max mem: 3534 +Test: [ 720/2387] eta: 0:05:59 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.3296 (0.3493) acc1: 85.4167 (82.9620) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.1848 data: 0.0917 max mem: 3534 +Test: [ 730/2387] eta: 0:05:56 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.3296 (0.3492) acc1: 84.8958 (82.9885) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.1899 data: 0.0969 max mem: 3534 +Test: [ 740/2387] eta: 0:05:54 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.3296 (0.3489) acc1: 84.8958 (83.0209) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.2019 data: 0.1076 max mem: 3534 +Test: [ 750/2387] eta: 0:05:51 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.3275 (0.3486) acc1: 84.8958 (83.0538) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.1917 data: 0.0981 max mem: 3534 +Test: [ 760/2387] eta: 0:05:48 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.3253 (0.3478) acc1: 84.8958 (83.1154) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1784 data: 0.0866 max mem: 3534 +Test: [ 770/2387] eta: 0:05:45 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2133 (0.3455) acc1: 92.9688 (83.2864) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1681 data: 0.0770 max mem: 3534 +Test: [ 780/2387] eta: 0:05:42 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1259 (0.3426) acc1: 99.7396 (83.4964) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1747 data: 0.0790 max mem: 3534 +Test: [ 790/2387] eta: 0:05:39 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1043 (0.3400) acc1: 100.0000 (83.6632) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.1699 data: 0.0732 max mem: 3534 +Test: [ 800/2387] eta: 0:05:36 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1028 (0.3371) acc1: 100.0000 (83.8613) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1573 data: 0.0644 max mem: 3534 +Test: [ 810/2387] eta: 0:05:33 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1132 (0.3343) acc1: 100.0000 (84.0574) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1707 data: 0.0785 max mem: 3534 +Test: [ 820/2387] eta: 0:05:32 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1133 (0.3316) acc1: 100.0000 (84.2446) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2186 data: 0.1265 max mem: 3534 +Test: [ 830/2387] eta: 0:05:28 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1375 (0.3298) acc1: 99.2188 (84.4004) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2052 data: 0.1132 max mem: 3534 +Test: [ 840/2387] eta: 0:05:27 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2061 (0.3287) acc1: 98.4375 (84.5500) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1859 data: 0.0932 max mem: 3534 +Test: [ 850/2387] eta: 0:05:24 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2270 (0.3274) acc1: 98.4375 (84.7037) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2164 data: 0.1235 max mem: 3534 +Test: [ 860/2387] eta: 0:05:22 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2108 (0.3260) acc1: 98.4375 (84.8611) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2088 data: 0.1136 max mem: 3534 +Test: [ 870/2387] eta: 0:05:20 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1989 (0.3246) acc1: 98.9583 (85.0160) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2055 data: 0.1101 max mem: 3534 +Test: [ 880/2387] eta: 0:05:18 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1989 (0.3231) acc1: 98.9583 (85.1681) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2041 data: 0.1110 max mem: 3534 +Test: [ 890/2387] eta: 0:05:16 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1730 (0.3215) acc1: 99.2188 (85.3164) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2118 data: 0.1187 max mem: 3534 +Test: [ 900/2387] eta: 0:05:14 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2186 (0.3247) acc1: 95.5729 (85.1158) acc5: 100.0000 (100.0000) acc1_neg: 60.938 acc5_neg: 100.000 time: 0.2179 data: 0.1263 max mem: 3535 +Test: [ 910/2387] eta: 0:05:11 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.6898 (0.3287) acc1: 60.1562 (84.8450) acc5: 100.0000 (100.0000) acc1_neg: 60.417 acc5_neg: 100.000 time: 0.2088 data: 0.1124 max mem: 3535 +Test: [ 920/2387] eta: 0:05:09 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.7003 (0.3327) acc1: 60.4167 (84.5772) acc5: 100.0000 (100.0000) acc1_neg: 60.938 acc5_neg: 100.000 time: 0.1981 data: 0.1003 max mem: 3535 +Test: [ 930/2387] eta: 0:05:07 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.7016 (0.3366) acc1: 60.6771 (84.3314) acc5: 100.0000 (100.0000) acc1_neg: 59.896 acc5_neg: 100.000 time: 0.2031 data: 0.1100 max mem: 3535 +Test: [ 940/2387] eta: 0:05:05 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.6975 (0.3403) acc1: 60.6771 (84.0731) acc5: 100.0000 (100.0000) acc1_neg: 58.594 acc5_neg: 100.000 time: 0.2122 data: 0.1194 max mem: 3535 +Test: [ 950/2387] eta: 0:05:03 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.6881 (0.3439) acc1: 59.3750 (83.8238) acc5: 100.0000 (100.0000) acc1_neg: 58.333 acc5_neg: 100.000 time: 0.2045 data: 0.1115 max mem: 3535 +Test: [ 960/2387] eta: 0:05:01 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.6823 (0.3475) acc1: 61.1979 (83.5864) acc5: 100.0000 (100.0000) acc1_neg: 63.542 acc5_neg: 100.000 time: 0.2077 data: 0.1149 max mem: 3535 +Test: [ 970/2387] eta: 0:04:58 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.6895 (0.3512) acc1: 59.3750 (83.3392) acc5: 100.0000 (100.0000) acc1_neg: 63.802 acc5_neg: 100.000 time: 0.2040 data: 0.1123 max mem: 3535 +Test: [ 980/2387] eta: 0:04:56 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.6755 (0.3513) acc1: 59.8958 (83.3469) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.1951 data: 0.1045 max mem: 3535 +Test: [ 990/2387] eta: 0:04:54 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1560 (0.3495) acc1: 99.2188 (83.4739) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2048 data: 0.1143 max mem: 3535 +Test: [1000/2387] eta: 0:04:52 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1206 (0.3473) acc1: 99.4792 (83.6190) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2179 data: 0.1268 max mem: 3535 +Test: [1010/2387] eta: 0:04:50 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1342 (0.3455) acc1: 99.2188 (83.7545) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2400 data: 0.1478 max mem: 3535 +Test: [1020/2387] eta: 0:04:48 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1451 (0.3439) acc1: 98.6979 (83.8835) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2175 data: 0.1252 max mem: 3535 +Test: [1030/2387] eta: 0:04:46 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1449 (0.3420) acc1: 99.7396 (84.0312) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1930 data: 0.0980 max mem: 3535 +Test: [1040/2387] eta: 0:04:43 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1394 (0.3400) acc1: 99.7396 (84.1809) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1838 data: 0.0873 max mem: 3535 +Test: [1050/2387] eta: 0:04:40 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1261 (0.3381) acc1: 99.7396 (84.3279) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1722 data: 0.0780 max mem: 3535 +Test: [1060/2387] eta: 0:04:38 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1361 (0.3362) acc1: 99.7396 (84.4707) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1782 data: 0.0847 max mem: 3535 +Test: [1070/2387] eta: 0:04:35 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1426 (0.3345) acc1: 99.7396 (84.6019) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1776 data: 0.0822 max mem: 3535 +Test: [1080/2387] eta: 0:04:33 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1673 (0.3333) acc1: 98.6979 (84.7130) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1789 data: 0.0807 max mem: 3535 +Test: [1090/2387] eta: 0:04:30 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2412 (0.3327) acc1: 94.7917 (84.7913) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1723 data: 0.0764 max mem: 3535 +Test: [1100/2387] eta: 0:04:28 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2678 (0.3322) acc1: 93.7500 (84.8729) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1768 data: 0.0832 max mem: 3535 +Test: [1110/2387] eta: 0:04:26 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2678 (0.3317) acc1: 93.7500 (84.9352) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2025 data: 0.1076 max mem: 3535 +Test: [1120/2387] eta: 0:04:24 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2560 (0.3311) acc1: 93.4896 (85.0143) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1930 data: 0.0976 max mem: 3536 +Test: [1130/2387] eta: 0:04:21 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2560 (0.3303) acc1: 95.3125 (85.1026) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1905 data: 0.0957 max mem: 3536 +Test: [1140/2387] eta: 0:04:19 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2590 (0.3297) acc1: 94.2708 (85.1880) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1898 data: 0.0971 max mem: 3536 +Test: [1150/2387] eta: 0:04:17 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2423 (0.3286) acc1: 96.0938 (85.2861) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1752 data: 0.0842 max mem: 3536 +Test: [1160/2387] eta: 0:04:14 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1544 (0.3271) acc1: 99.2188 (85.3967) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1854 data: 0.0947 max mem: 3536 +Test: [1170/2387] eta: 0:04:12 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1425 (0.3255) acc1: 99.7396 (85.5183) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1904 data: 0.0993 max mem: 3536 +Test: [1180/2387] eta: 0:04:10 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1425 (0.3241) acc1: 99.7396 (85.6341) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1821 data: 0.0894 max mem: 3536 +Test: [1190/2387] eta: 0:04:08 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1405 (0.3226) acc1: 99.4792 (85.7508) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1933 data: 0.0988 max mem: 3536 +Test: [1200/2387] eta: 0:04:06 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1196 (0.3209) acc1: 100.0000 (85.8692) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2090 data: 0.1140 max mem: 3536 +Test: [1210/2387] eta: 0:04:03 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1254 (0.3193) acc1: 100.0000 (85.9846) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.0990 max mem: 3536 +Test: [1220/2387] eta: 0:04:01 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.0880 (0.3174) acc1: 100.0000 (86.0977) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1933 data: 0.0989 max mem: 3536 +Test: [1230/2387] eta: 0:03:59 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.0799 (0.3155) acc1: 100.0000 (86.2104) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1925 data: 0.0992 max mem: 3536 +Test: [1240/2387] eta: 0:03:57 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.0778 (0.3136) acc1: 100.0000 (86.3215) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1852 data: 0.0923 max mem: 3536 +Test: [1250/2387] eta: 0:03:54 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.0793 (0.3117) acc1: 100.0000 (86.4306) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1896 data: 0.0966 max mem: 3536 +Test: [1260/2387] eta: 0:03:52 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.0745 (0.3098) acc1: 100.0000 (86.5383) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1971 data: 0.1029 max mem: 3536 +Test: [1270/2387] eta: 0:03:50 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.0732 (0.3080) acc1: 100.0000 (86.6440) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1951 data: 0.1028 max mem: 3536 +Test: [1280/2387] eta: 0:03:48 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.0765 (0.3062) acc1: 100.0000 (86.7452) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1900 data: 0.0989 max mem: 3536 +Test: [1290/2387] eta: 0:03:46 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.0859 (0.3046) acc1: 99.7396 (86.8321) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1903 data: 0.0978 max mem: 3536 +Test: [1300/2387] eta: 0:03:43 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.0948 (0.3031) acc1: 99.4792 (86.9255) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1842 data: 0.0916 max mem: 3536 +Test: [1310/2387] eta: 0:03:41 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.0951 (0.3016) acc1: 99.4792 (87.0155) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1722 data: 0.0805 max mem: 3536 +Test: [1320/2387] eta: 0:03:39 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.0978 (0.3001) acc1: 99.4792 (87.1024) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1730 data: 0.0812 max mem: 3536 +Test: [1330/2387] eta: 0:03:37 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1068 (0.2989) acc1: 99.2188 (87.1830) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1871 data: 0.0939 max mem: 3536 +Test: [1340/2387] eta: 0:03:34 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1696 (0.2987) acc1: 96.8750 (87.1990) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1846 data: 0.0914 max mem: 3536 +Test: [1350/2387] eta: 0:03:32 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.0588 (0.2969) acc1: 100.0000 (87.2937) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1751 data: 0.0837 max mem: 3537 +Test: [1360/2387] eta: 0:03:30 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.0533 (0.2951) acc1: 100.0000 (87.3871) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1764 data: 0.0848 max mem: 3537 +Test: [1370/2387] eta: 0:03:28 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.0537 (0.2933) acc1: 100.0000 (87.4791) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1839 data: 0.0919 max mem: 3537 +Test: [1380/2387] eta: 0:03:25 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.0536 (0.2916) acc1: 100.0000 (87.5698) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1844 data: 0.0923 max mem: 3537 +Test: [1390/2387] eta: 0:03:23 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.0497 (0.2898) acc1: 100.0000 (87.6591) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1760 data: 0.0842 max mem: 3537 +Test: [1400/2387] eta: 0:03:21 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.0466 (0.2881) acc1: 100.0000 (87.7472) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1847 data: 0.0926 max mem: 3537 +Test: [1410/2387] eta: 0:03:19 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.0523 (0.2870) acc1: 100.0000 (87.8210) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1771 data: 0.0849 max mem: 3537 +Test: [1420/2387] eta: 0:03:16 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1614 (0.2864) acc1: 98.4375 (87.8777) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1703 data: 0.0776 max mem: 3537 +Test: [1430/2387] eta: 0:03:14 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1794 (0.2856) acc1: 98.4375 (87.9513) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1829 data: 0.0908 max mem: 3537 +Test: [1440/2387] eta: 0:03:12 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1662 (0.2848) acc1: 98.9583 (88.0250) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1722 data: 0.0807 max mem: 3537 +Test: [1450/2387] eta: 0:03:10 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1568 (0.2841) acc1: 98.9583 (88.0915) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1693 data: 0.0771 max mem: 3537 +Test: [1460/2387] eta: 0:03:08 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1372 (0.2831) acc1: 98.4375 (88.1670) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1792 data: 0.0872 max mem: 3537 +Test: [1470/2387] eta: 0:03:05 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1076 (0.2818) acc1: 99.7396 (88.2467) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1801 data: 0.0892 max mem: 3537 +Test: [1480/2387] eta: 0:03:03 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.0971 (0.2806) acc1: 100.0000 (88.3233) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1773 data: 0.0849 max mem: 3537 +Test: [1490/2387] eta: 0:03:01 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1025 (0.2794) acc1: 99.7396 (88.3998) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1886 data: 0.0953 max mem: 3537 +Test: [1500/2387] eta: 0:02:59 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1021 (0.2782) acc1: 99.7396 (88.4757) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1998 data: 0.1067 max mem: 3537 +Test: [1510/2387] eta: 0:02:57 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1050 (0.2772) acc1: 100.0000 (88.5517) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1962 data: 0.1037 max mem: 3537 +Test: [1520/2387] eta: 0:02:55 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1329 (0.2766) acc1: 99.7396 (88.6173) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2006 data: 0.1097 max mem: 3537 +Test: [1530/2387] eta: 0:02:53 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1938 (0.2760) acc1: 98.9583 (88.6811) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1987 data: 0.1077 max mem: 3537 +Test: [1540/2387] eta: 0:02:51 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1733 (0.2753) acc1: 99.4792 (88.7489) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1958 data: 0.1045 max mem: 3537 +Test: [1550/2387] eta: 0:02:49 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1690 (0.2747) acc1: 99.4792 (88.8167) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1942 data: 0.1029 max mem: 3537 +Test: [1560/2387] eta: 0:02:47 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1541 (0.2739) acc1: 99.4792 (88.8837) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1890 data: 0.0969 max mem: 3537 +Test: [1570/2387] eta: 0:02:45 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1405 (0.2731) acc1: 99.7396 (88.9461) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1896 data: 0.0964 max mem: 3537 +Test: [1580/2387] eta: 0:02:43 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1421 (0.2724) acc1: 99.7396 (89.0126) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1963 data: 0.1031 max mem: 3538 +Test: [1590/2387] eta: 0:02:41 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1462 (0.2716) acc1: 99.4792 (89.0787) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2008 data: 0.1087 max mem: 3538 +Test: [1600/2387] eta: 0:02:39 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1404 (0.2709) acc1: 99.4792 (89.1202) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2076 data: 0.1162 max mem: 3538 +Test: [1610/2387] eta: 0:02:37 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1315 (0.2701) acc1: 99.7396 (89.1854) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1990 data: 0.1056 max mem: 3538 +Test: [1620/2387] eta: 0:02:34 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1398 (0.2693) acc1: 99.7396 (89.2432) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0809 max mem: 3538 +Test: [1630/2387] eta: 0:02:32 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1523 (0.2686) acc1: 98.6979 (89.2993) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1720 data: 0.0790 max mem: 3538 +Test: [1640/2387] eta: 0:02:30 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1576 (0.2680) acc1: 98.1771 (89.3513) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1826 data: 0.0873 max mem: 3538 +Test: [1650/2387] eta: 0:02:28 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1449 (0.2672) acc1: 98.6979 (89.4067) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1893 data: 0.0939 max mem: 3538 +Test: [1660/2387] eta: 0:02:26 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1445 (0.2666) acc1: 98.9583 (89.4617) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1883 data: 0.0955 max mem: 3538 +Test: [1670/2387] eta: 0:02:24 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1386 (0.2658) acc1: 99.4792 (89.5169) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1987 data: 0.1061 max mem: 3538 +Test: [1680/2387] eta: 0:02:22 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1339 (0.2652) acc1: 99.4792 (89.5647) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2056 data: 0.1131 max mem: 3538 +Test: [1690/2387] eta: 0:02:20 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1539 (0.2646) acc1: 99.2188 (89.6124) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2054 data: 0.1134 max mem: 3538 +Test: [1700/2387] eta: 0:02:18 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1511 (0.2638) acc1: 98.6979 (89.6682) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2038 data: 0.1115 max mem: 3538 +Test: [1710/2387] eta: 0:02:16 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1137 (0.2629) acc1: 99.7396 (89.7267) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1990 data: 0.1065 max mem: 3538 +Test: [1720/2387] eta: 0:02:14 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.0927 (0.2619) acc1: 100.0000 (89.7846) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1964 data: 0.1031 max mem: 3538 +Test: [1730/2387] eta: 0:02:12 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.0931 (0.2610) acc1: 100.0000 (89.8418) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1933 data: 0.0995 max mem: 3538 +Test: [1740/2387] eta: 0:02:10 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.0931 (0.2600) acc1: 100.0000 (89.8995) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2049 data: 0.1125 max mem: 3538 +Test: [1750/2387] eta: 0:02:08 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.0901 (0.2591) acc1: 100.0000 (89.9572) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1961 data: 0.1049 max mem: 3538 +Test: [1760/2387] eta: 0:02:06 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.0958 (0.2582) acc1: 100.0000 (90.0141) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1896 data: 0.0986 max mem: 3538 +Test: [1770/2387] eta: 0:02:04 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1110 (0.2575) acc1: 100.0000 (90.0648) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1667 data: 0.0753 max mem: 3538 +Test: [1780/2387] eta: 0:02:01 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1259 (0.2567) acc1: 99.7396 (90.1185) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1400 data: 0.0489 max mem: 3538 +Test: [1790/2387] eta: 0:01:59 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1228 (0.2559) acc1: 100.0000 (90.1727) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1541 data: 0.0624 max mem: 3538 +Test: [1800/2387] eta: 0:01:57 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1183 (0.2552) acc1: 99.7396 (90.2252) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1551 data: 0.0630 max mem: 3538 +Test: [1810/2387] eta: 0:01:55 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1019 (0.2543) acc1: 100.0000 (90.2786) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1648 data: 0.0736 max mem: 3539 +Test: [1820/2387] eta: 0:01:53 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.0936 (0.2534) acc1: 100.0000 (90.3318) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0929 max mem: 3539 +Test: [1830/2387] eta: 0:01:51 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1000 (0.2527) acc1: 100.0000 (90.3839) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1802 data: 0.0878 max mem: 3539 +Test: [1840/2387] eta: 0:01:49 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1367 (0.2521) acc1: 100.0000 (90.4362) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1827 data: 0.0904 max mem: 3539 +Test: [1850/2387] eta: 0:01:47 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1384 (0.2514) acc1: 100.0000 (90.4874) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1835 data: 0.0895 max mem: 3539 +Test: [1860/2387] eta: 0:01:45 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1387 (0.2508) acc1: 100.0000 (90.5385) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1665 data: 0.0733 max mem: 3539 +Test: [1870/2387] eta: 0:01:43 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1398 (0.2503) acc1: 100.0000 (90.5884) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1773 data: 0.0866 max mem: 3539 +Test: [1880/2387] eta: 0:01:41 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1397 (0.2497) acc1: 100.0000 (90.6383) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2040 data: 0.1134 max mem: 3539 +Test: [1890/2387] eta: 0:01:39 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1398 (0.2491) acc1: 100.0000 (90.6875) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2101 data: 0.1196 max mem: 3539 +Test: [1900/2387] eta: 0:01:37 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1412 (0.2486) acc1: 100.0000 (90.7356) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1843 data: 0.0929 max mem: 3539 +Test: [1910/2387] eta: 0:01:35 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1808 (0.2483) acc1: 99.4792 (90.7797) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1742 data: 0.0800 max mem: 3539 +Test: [1920/2387] eta: 0:01:33 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2016 (0.2482) acc1: 98.9583 (90.7954) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2012 data: 0.1080 max mem: 3539 +Test: [1930/2387] eta: 0:01:31 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1665 (0.2478) acc1: 98.9583 (90.8397) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2142 data: 0.1213 max mem: 3539 +Test: [1940/2387] eta: 0:01:29 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1654 (0.2477) acc1: 99.4792 (90.8461) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2136 data: 0.1189 max mem: 3539 +Test: [1950/2387] eta: 0:01:27 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2013 (0.2475) acc1: 98.9583 (90.8857) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2219 data: 0.1293 max mem: 3539 +Test: [1960/2387] eta: 0:01:25 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1916 (0.2472) acc1: 98.9583 (90.9121) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2179 data: 0.1262 max mem: 3539 +Test: [1970/2387] eta: 0:01:23 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2265 (0.2474) acc1: 94.5312 (90.9159) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1998 data: 0.1083 max mem: 3539 +Test: [1980/2387] eta: 0:01:21 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2556 (0.2472) acc1: 94.5312 (90.9431) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1776 data: 0.0865 max mem: 3539 +Test: [1990/2387] eta: 0:01:19 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1347 (0.2467) acc1: 99.2188 (90.9856) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1676 data: 0.0757 max mem: 3539 +Test: [2000/2387] eta: 0:01:17 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1345 (0.2461) acc1: 99.2188 (91.0271) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1792 data: 0.0874 max mem: 3539 +Test: [2010/2387] eta: 0:01:15 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1346 (0.2456) acc1: 99.4792 (91.0693) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1947 data: 0.1038 max mem: 3539 +Test: [2020/2387] eta: 0:01:13 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1347 (0.2450) acc1: 99.4792 (91.1105) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1949 data: 0.1033 max mem: 3539 +Test: [2030/2387] eta: 0:01:11 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1347 (0.2445) acc1: 99.2188 (91.1502) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1862 data: 0.0944 max mem: 3540 +Test: [2040/2387] eta: 0:01:09 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1350 (0.2439) acc1: 99.2188 (91.1906) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1825 data: 0.0907 max mem: 3540 +Test: [2050/2387] eta: 0:01:07 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1335 (0.2434) acc1: 99.4792 (91.2307) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2029 data: 0.1102 max mem: 3540 +Test: [2060/2387] eta: 0:01:05 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1361 (0.2429) acc1: 99.2188 (91.2690) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1980 data: 0.1011 max mem: 3540 +Test: [2070/2387] eta: 0:01:03 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1361 (0.2424) acc1: 99.2188 (91.3088) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1705 data: 0.0747 max mem: 3540 +Test: [2080/2387] eta: 0:01:01 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1341 (0.2419) acc1: 99.4792 (91.3473) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1801 data: 0.0888 max mem: 3540 +Test: [2090/2387] eta: 0:00:59 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1332 (0.2414) acc1: 99.4792 (91.3860) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1917 data: 0.0925 max mem: 3540 +Test: [2100/2387] eta: 0:00:57 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1342 (0.2408) acc1: 99.4792 (91.4243) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1889 data: 0.0886 max mem: 3540 +Test: [2110/2387] eta: 0:00:55 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1348 (0.2403) acc1: 99.4792 (91.4624) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1892 data: 0.0967 max mem: 3540 +Test: [2120/2387] eta: 0:00:53 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1431 (0.2401) acc1: 99.4792 (91.4945) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1768 data: 0.0848 max mem: 3540 +Test: [2130/2387] eta: 0:00:51 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1657 (0.2397) acc1: 98.9583 (91.5293) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1686 data: 0.0761 max mem: 3540 +Test: [2140/2387] eta: 0:00:49 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1595 (0.2395) acc1: 98.9583 (91.5488) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1836 data: 0.0923 max mem: 3540 +Test: [2150/2387] eta: 0:00:47 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1624 (0.2392) acc1: 99.2188 (91.5836) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1906 data: 0.0996 max mem: 3540 +Test: [2160/2387] eta: 0:00:45 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1552 (0.2388) acc1: 99.2188 (91.6164) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1844 data: 0.0921 max mem: 3540 +Test: [2170/2387] eta: 0:00:43 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1623 (0.2386) acc1: 98.4375 (91.6438) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1724 data: 0.0794 max mem: 3540 +Test: [2180/2387] eta: 0:00:40 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2311 (0.2398) acc1: 95.8333 (91.5674) acc5: 100.0000 (100.0000) acc1_neg: 45.312 acc5_neg: 100.000 time: 0.1700 data: 0.0783 max mem: 3540 +Test: [2190/2387] eta: 0:00:38 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2801 (0.2403) acc1: 91.4062 (91.5490) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1742 data: 0.0839 max mem: 3540 +Test: [2200/2387] eta: 0:00:36 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1903 (0.2400) acc1: 97.1354 (91.5797) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1760 data: 0.0852 max mem: 3540 +Test: [2210/2387] eta: 0:00:34 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1573 (0.2396) acc1: 99.2188 (91.6158) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1731 data: 0.0817 max mem: 3540 +Test: [2220/2387] eta: 0:00:32 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1437 (0.2392) acc1: 99.4792 (91.6485) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1586 data: 0.0656 max mem: 3540 +Test: [2230/2387] eta: 0:00:30 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1536 (0.2388) acc1: 99.2188 (91.6836) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1624 data: 0.0697 max mem: 3540 +Test: [2240/2387] eta: 0:00:28 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1467 (0.2384) acc1: 99.4792 (91.7145) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1602 data: 0.0697 max mem: 3540 +Test: [2250/2387] eta: 0:00:26 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1453 (0.2380) acc1: 99.7396 (91.7436) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1407 data: 0.0506 max mem: 3540 +Test: [2260/2387] eta: 0:00:24 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1427 (0.2376) acc1: 99.7396 (91.7790) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1377 data: 0.0476 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1287 (0.2371) acc1: 100.0000 (91.8147) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1395 data: 0.0495 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1187 (0.2368) acc1: 100.0000 (91.8348) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1396 data: 0.0495 max mem: 3541 +Test: [2290/2387] eta: 0:00:19 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1209 (0.2362) acc1: 100.0000 (91.8698) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1419 data: 0.0520 max mem: 3541 +Test: [2300/2387] eta: 0:00:17 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1236 (0.2358) acc1: 99.7396 (91.9007) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1402 data: 0.0504 max mem: 3541 +Test: [2310/2387] eta: 0:00:15 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1265 (0.2354) acc1: 99.7396 (91.9290) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1302 data: 0.0403 max mem: 3541 +Test: [2320/2387] eta: 0:00:13 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1569 (0.2352) acc1: 98.6979 (91.9486) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1253 data: 0.0355 max mem: 3541 +Test: [2330/2387] eta: 0:00:11 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1916 (0.2351) acc1: 98.4375 (91.9738) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1231 data: 0.0334 max mem: 3541 +Test: [2340/2387] eta: 0:00:09 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1932 (0.2349) acc1: 98.6979 (92.0008) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1198 data: 0.0301 max mem: 3541 +Test: [2350/2387] eta: 0:00:07 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.1932 (0.2350) acc1: 98.1771 (92.0076) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1158 data: 0.0260 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2018 (0.2349) acc1: 98.6979 (92.0278) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1135 data: 0.0237 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2058 (0.2348) acc1: 98.1771 (92.0457) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1130 data: 0.0232 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2277 (0.2350) acc1: 94.2708 (92.0482) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.1133 data: 0.0234 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 45.205 acc5_pos: 100.000 loss: 0.2620 (0.2351) acc1: 93.7500 (92.0505) acc5: 100.0000 (100.0000) acc1_neg: 95.349 acc5_neg: 100.000 time: 0.1082 data: 0.0209 max mem: 3541 +Test: Total time: 0:07:40 (0.1930 s / it) +* Overall Acc@1 92.050 Acc@5 100.000 + target == real Acc@1 28.595 Acc@5 100.000 + target == fake Acc@1 96.507 Acc@5 100.000 + loss 0.235 + PR-AUC 0.9920 ROC-AUC 0.8976 EER 17.53% (thr=0.7803) +Accuracy of the network on the 916267 test images: 92.1% +[rank0]:[W726 04:58:12.340846117 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 04:58:31.691000 1184949 torch/distributed/run.py:766] +W0726 04:58:31.691000 1184949 torch/distributed/run.py:766] ***************************************** +W0726 04:58:31.691000 1184949 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 04:58:31.691000 1184949 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 04:58:58.228951589 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 04:58:58.549649228 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 04:58:58.631675620 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 04:58:58.721697782 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 04:58:59.915243340 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 04:58:59.946543191 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 04:58:59.983922801 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 04:58:59.017325970 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n22_seed7_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed7_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:01:33 acc1_pos: 20.833 acc5_pos: 100.000 loss: 1.1286 (1.1286) acc1: 20.8333 (20.8333) acc5: 100.0000 (100.0000) time: 7.1450 data: 5.5061 max mem: 3531 +Test: [ 10/517] eta: 0:06:12 acc1_pos: 14.844 acc5_pos: 100.000 loss: 1.1286 (1.1014) acc1: 20.0521 (21.8750) acc5: 100.0000 (100.0000) time: 0.7339 data: 0.5010 max mem: 3531 +Test: [ 20/517] eta: 0:03:32 acc1_pos: 17.448 acc5_pos: 100.000 loss: 1.1222 (1.1142) acc1: 19.2708 (21.7262) acc5: 100.0000 (100.0000) time: 0.0925 data: 0.0005 max mem: 3531 +Test: [ 30/517] eta: 0:02:56 acc1_pos: 13.021 acc5_pos: 100.000 loss: 1.1266 (1.1412) acc1: 17.9688 (20.8249) acc5: 100.0000 (100.0000) time: 0.1570 data: 0.0657 max mem: 3531 +Test: [ 40/517] eta: 0:02:35 acc1_pos: 14.844 acc5_pos: 100.000 loss: 1.2262 (1.1795) acc1: 13.0208 (18.7818) acc5: 100.0000 (100.0000) time: 0.2203 data: 0.1292 max mem: 3531 +Test: [ 50/517] eta: 0:02:24 acc1_pos: 47.248 acc5_pos: 100.000 loss: 1.2135 (1.1323) acc1: 13.0208 (23.1209) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2308 data: 0.1400 max mem: 3531 +Test: [ 60/517] eta: 0:02:18 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.3294 (0.9931) acc1: 91.4062 (35.2246) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2555 data: 0.1655 max mem: 3531 +Test: [ 70/517] eta: 0:02:16 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.2630 (0.8912) acc1: 98.6979 (44.1461) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2931 data: 0.2012 max mem: 3531 +Test: [ 80/517] eta: 0:02:10 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.2629 (0.8178) acc1: 98.9583 (50.6237) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2839 data: 0.1907 max mem: 3531 +Test: [ 90/517] eta: 0:02:09 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.2782 (0.7598) acc1: 98.6979 (55.8951) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2949 data: 0.2012 max mem: 3531 +Test: [100/517] eta: 0:02:03 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.2438 (0.7060) acc1: 99.4792 (60.2027) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2899 data: 0.1937 max mem: 3531 +Test: [110/517] eta: 0:01:58 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.1888 (0.6594) acc1: 100.0000 (63.7645) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2399 data: 0.1418 max mem: 3531 +Test: [120/517] eta: 0:01:53 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.2147 (0.6246) acc1: 99.7396 (66.6882) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2349 data: 0.1385 max mem: 3531 +Test: [130/517] eta: 0:01:49 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.2312 (0.5952) acc1: 98.9583 (69.1376) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2349 data: 0.1403 max mem: 3531 +Test: [140/517] eta: 0:01:45 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.2262 (0.5685) acc1: 99.7396 (71.3117) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2399 data: 0.1437 max mem: 3531 +Test: [150/517] eta: 0:01:41 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.2224 (0.5458) acc1: 99.7396 (73.1719) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2399 data: 0.1449 max mem: 3531 +Test: [160/517] eta: 0:01:38 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.2606 (0.5293) acc1: 99.4792 (74.7380) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2449 data: 0.1524 max mem: 3531 +Test: [170/517] eta: 0:01:35 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.2724 (0.5138) acc1: 98.6979 (76.1270) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2499 data: 0.1527 max mem: 3531 +Test: [180/517] eta: 0:01:31 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.2383 (0.4973) acc1: 98.9583 (77.4027) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2399 data: 0.1405 max mem: 3531 +Test: [190/517] eta: 0:01:28 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.1842 (0.4801) acc1: 99.7396 (78.5858) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2349 data: 0.1400 max mem: 3531 +Test: [200/517] eta: 0:01:25 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.1835 (0.4659) acc1: 100.0000 (79.6396) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2399 data: 0.1451 max mem: 3531 +Test: [210/517] eta: 0:01:22 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.2054 (0.4560) acc1: 99.2188 (80.5021) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2449 data: 0.1476 max mem: 3532 +Test: [220/517] eta: 0:01:19 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.2612 (0.4481) acc1: 97.9167 (81.2394) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2449 data: 0.1483 max mem: 3532 +Test: [230/517] eta: 0:01:16 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.2589 (0.4393) acc1: 98.4375 (82.0234) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2399 data: 0.1442 max mem: 3532 +Test: [240/517] eta: 0:01:13 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.2307 (0.4303) acc1: 99.4792 (82.7455) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2399 data: 0.1449 max mem: 3532 +Test: [250/517] eta: 0:01:10 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.2148 (0.4218) acc1: 99.7396 (83.4111) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2399 data: 0.1465 max mem: 3532 +Test: [260/517] eta: 0:01:07 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.2150 (0.4142) acc1: 99.4792 (84.0088) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2399 data: 0.1457 max mem: 3532 +Test: [270/517] eta: 0:01:04 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.2077 (0.4063) acc1: 99.4792 (84.5864) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2449 data: 0.1496 max mem: 3532 +Test: [280/517] eta: 0:01:01 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.1991 (0.3992) acc1: 99.7396 (85.1229) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2499 data: 0.1555 max mem: 3532 +Test: [290/517] eta: 0:00:59 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.2129 (0.3937) acc1: 99.4792 (85.5545) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2449 data: 0.1509 max mem: 3532 +Test: [300/517] eta: 0:00:56 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.2189 (0.3878) acc1: 99.4792 (86.0214) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2349 data: 0.1408 max mem: 3532 +Test: [310/517] eta: 0:00:53 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.2214 (0.3833) acc1: 99.7396 (86.4634) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2349 data: 0.1405 max mem: 3532 +Test: [320/517] eta: 0:00:50 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.2412 (0.3787) acc1: 99.4792 (86.8664) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2399 data: 0.1443 max mem: 3532 +Test: [330/517] eta: 0:00:48 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.2138 (0.3735) acc1: 99.4792 (87.2506) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2349 data: 0.1394 max mem: 3532 +Test: [340/517] eta: 0:00:45 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.1964 (0.3687) acc1: 99.4792 (87.5840) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2349 data: 0.1380 max mem: 3532 +Test: [350/517] eta: 0:00:42 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.2298 (0.3655) acc1: 98.9583 (87.8917) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2299 data: 0.1322 max mem: 3532 +Test: [360/517] eta: 0:00:40 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.2354 (0.3619) acc1: 99.2188 (88.1925) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2249 data: 0.1303 max mem: 3532 +Test: [370/517] eta: 0:00:37 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.1982 (0.3570) acc1: 100.0000 (88.5080) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2399 data: 0.1473 max mem: 3532 +Test: [380/517] eta: 0:00:34 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.1783 (0.3527) acc1: 100.0000 (88.7987) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2448 data: 0.1510 max mem: 3532 +Test: [390/517] eta: 0:00:32 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.2044 (0.3492) acc1: 99.7396 (89.0319) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2297 data: 0.1355 max mem: 3532 +Test: [400/517] eta: 0:00:29 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.1986 (0.3452) acc1: 99.4792 (89.2924) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2249 data: 0.1325 max mem: 3532 +Test: [410/517] eta: 0:00:27 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.2187 (0.3428) acc1: 99.4792 (89.5136) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2349 data: 0.1421 max mem: 3532 +Test: [420/517] eta: 0:00:24 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.2187 (0.3396) acc1: 99.7396 (89.7411) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2399 data: 0.1470 max mem: 3532 +Test: [430/517] eta: 0:00:21 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.2301 (0.3382) acc1: 98.6979 (89.8957) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2349 data: 0.1418 max mem: 3532 +Test: [440/517] eta: 0:00:19 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.1971 (0.3345) acc1: 99.7396 (90.1225) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2249 data: 0.1281 max mem: 3533 +Test: [450/517] eta: 0:00:16 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.1813 (0.3314) acc1: 100.0000 (90.3328) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2349 data: 0.1378 max mem: 3533 +Test: [460/517] eta: 0:00:14 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.1919 (0.3288) acc1: 99.7396 (90.5228) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2499 data: 0.1514 max mem: 3533 +Test: [470/517] eta: 0:00:11 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.1919 (0.3259) acc1: 99.7396 (90.7184) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2399 data: 0.1412 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.1833 (0.3228) acc1: 100.0000 (90.9103) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2249 data: 0.1303 max mem: 3533 +Test: [490/517] eta: 0:00:06 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.1870 (0.3206) acc1: 100.0000 (91.0764) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2299 data: 0.1351 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.2336 (0.3193) acc1: 99.4792 (91.2217) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2449 data: 0.1488 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.2228 (0.3174) acc1: 99.4792 (91.3690) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2449 data: 0.1475 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 47.248 acc5_pos: 100.000 loss: 0.1937 (0.3159) acc1: 99.7396 (91.4639) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2070 data: 0.1078 max mem: 3533 +Test: Total time: 0:02:09 (0.2498 s / it) +* Overall Acc@1 91.464 Acc@5 100.000 + target == real Acc@1 19.340 Acc@5 100.000 + target == fake Acc@1 98.946 Acc@5 100.000 + loss 0.316 + PR-AUC 0.9654 ROC-AUC 0.7887 EER 28.35% (thr=0.7729) +Accuracy of the network on the 198427 test images: 91.5% +[rank0]:[W726 05:01:21.514725285 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 05:01:41.823000 1185522 torch/distributed/run.py:766] +W0726 05:01:41.823000 1185522 torch/distributed/run.py:766] ***************************************** +W0726 05:01:41.823000 1185522 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:01:41.823000 1185522 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:02:09.876910649 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:02:09.894636589 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:02:10.248865667 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:02:10.418212926 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:02:10.477033626 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:02:10.490151568 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W726 05:02:10.490209961 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:02:10.515422940 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n22_seed7_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed7_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:36:30 acc1_pos: 34.115 acc5_pos: 100.000 loss: 0.9785 (0.9785) acc1: 34.1146 (34.1146) acc5: 100.0000 (100.0000) time: 5.6315 data: 3.5877 max mem: 3531 +Test: [ 10/389] eta: 0:03:45 acc1_pos: 82.552 acc5_pos: 100.000 loss: 0.6904 (0.7443) acc1: 52.0833 (53.5275) acc5: 100.0000 (100.0000) time: 0.5939 data: 0.3265 max mem: 3531 +Test: [ 20/389] eta: 0:02:19 acc1_pos: 16.667 acc5_pos: 100.000 loss: 0.7446 (0.7981) acc1: 51.3021 (45.2381) acc5: 100.0000 (100.0000) time: 0.1163 data: 0.0227 max mem: 3531 +Test: [ 30/389] eta: 0:01:53 acc1_pos: 79.167 acc5_pos: 100.000 loss: 0.7446 (0.8171) acc1: 47.3958 (43.2040) acc5: 100.0000 (100.0000) time: 0.1633 data: 0.0667 max mem: 3531 +Test: [ 40/389] eta: 0:01:53 acc1_pos: 17.708 acc5_pos: 100.000 loss: 0.7162 (0.7898) acc1: 50.0000 (47.2625) acc5: 100.0000 (100.0000) time: 0.2668 data: 0.1737 max mem: 3531 +Test: [ 50/389] eta: 0:01:41 acc1_pos: 79.688 acc5_pos: 100.000 loss: 0.5763 (0.7479) acc1: 76.5625 (53.0688) acc5: 100.0000 (100.0000) time: 0.2729 data: 0.1800 max mem: 3531 +Test: [ 60/389] eta: 0:01:28 acc1_pos: 59.896 acc5_pos: 100.000 loss: 0.5795 (0.7252) acc1: 76.8229 (55.9298) acc5: 100.0000 (100.0000) time: 0.1566 data: 0.0596 max mem: 3531 +Test: [ 70/389] eta: 0:01:21 acc1_pos: 72.135 acc5_pos: 100.000 loss: 0.6212 (0.7130) acc1: 63.5417 (56.8332) acc5: 100.0000 (100.0000) time: 0.1426 data: 0.0472 max mem: 3531 +Test: [ 80/389] eta: 0:01:18 acc1_pos: 88.281 acc5_pos: 100.000 loss: 0.6753 (0.7073) acc1: 60.1562 (57.8447) acc5: 100.0000 (100.0000) time: 0.2053 data: 0.1133 max mem: 3531 +Test: [ 90/389] eta: 0:01:12 acc1_pos: 82.292 acc5_pos: 100.000 loss: 0.4457 (0.6765) acc1: 82.2917 (60.7200) acc5: 100.0000 (100.0000) time: 0.1927 data: 0.1014 max mem: 3531 +Test: [100/389] eta: 0:01:07 acc1_pos: 17.448 acc5_pos: 100.000 loss: 0.6645 (0.7009) acc1: 52.0833 (58.1039) acc5: 100.0000 (100.0000) time: 0.1559 data: 0.0654 max mem: 3531 +Test: [110/389] eta: 0:01:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.8922 (0.7122) acc1: 37.2396 (56.9468) acc5: 100.0000 (100.0000) time: 0.1846 data: 0.0926 max mem: 3531 +Test: [120/389] eta: 0:01:01 acc1_pos: 56.510 acc5_pos: 100.000 loss: 0.6582 (0.7050) acc1: 63.2812 (57.5155) acc5: 100.0000 (100.0000) time: 0.1912 data: 0.0991 max mem: 3531 +Test: [130/389] eta: 0:00:58 acc1_pos: 23.313 acc5_pos: 100.000 loss: 0.5199 (0.6920) acc1: 73.4375 (58.6872) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1848 data: 0.0921 max mem: 3531 +Test: [140/389] eta: 0:00:55 acc1_pos: 23.313 acc5_pos: 100.000 loss: 0.3720 (0.6724) acc1: 88.8021 (60.6863) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.1888 data: 0.0959 max mem: 3531 +Test: [150/389] eta: 0:00:52 acc1_pos: 23.313 acc5_pos: 100.000 loss: 0.3720 (0.6561) acc1: 87.7604 (62.2999) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1857 data: 0.0946 max mem: 3531 +Test: [160/389] eta: 0:00:49 acc1_pos: 23.313 acc5_pos: 100.000 loss: 0.4334 (0.6346) acc1: 87.7604 (64.1288) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1660 data: 0.0743 max mem: 3531 +Test: [170/389] eta: 0:00:46 acc1_pos: 23.313 acc5_pos: 100.000 loss: 0.1831 (0.6185) acc1: 100.0000 (65.0417) acc5: 100.0000 (100.0000) acc1_neg: 46.094 acc5_neg: 100.000 time: 0.1590 data: 0.0659 max mem: 3531 +Test: [180/389] eta: 0:00:44 acc1_pos: 23.313 acc5_pos: 100.000 loss: 0.2639 (0.6018) acc1: 99.2188 (66.6408) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1844 data: 0.0895 max mem: 3531 +Test: [190/389] eta: 0:00:41 acc1_pos: 23.313 acc5_pos: 100.000 loss: 0.3772 (0.5985) acc1: 89.8438 (66.8535) acc5: 100.0000 (100.0000) acc1_neg: 69.792 acc5_neg: 100.000 time: 0.1649 data: 0.0689 max mem: 3531 +Test: [200/389] eta: 0:00:38 acc1_pos: 23.313 acc5_pos: 100.000 loss: 0.4932 (0.5903) acc1: 85.4167 (67.8483) acc5: 100.0000 (100.0000) acc1_neg: 72.656 acc5_neg: 100.000 time: 0.1362 data: 0.0402 max mem: 3531 +Test: [210/389] eta: 0:00:36 acc1_pos: 23.313 acc5_pos: 100.000 loss: 0.2545 (0.5774) acc1: 98.1771 (68.8463) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1718 data: 0.0784 max mem: 3532 +Test: [220/389] eta: 0:00:34 acc1_pos: 23.313 acc5_pos: 100.000 loss: 0.1966 (0.5595) acc1: 100.0000 (70.2430) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1789 data: 0.0878 max mem: 3532 +Test: [230/389] eta: 0:00:31 acc1_pos: 23.313 acc5_pos: 100.000 loss: 0.2975 (0.5532) acc1: 99.2188 (70.9821) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1529 data: 0.0608 max mem: 3532 +Test: [240/389] eta: 0:00:29 acc1_pos: 23.313 acc5_pos: 100.000 loss: 0.3306 (0.5459) acc1: 95.3125 (71.6092) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1500 data: 0.0574 max mem: 3532 +Test: [250/389] eta: 0:00:27 acc1_pos: 23.313 acc5_pos: 100.000 loss: 0.2461 (0.5323) acc1: 99.4792 (72.7351) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1719 data: 0.0791 max mem: 3532 +Test: [260/389] eta: 0:00:25 acc1_pos: 23.313 acc5_pos: 100.000 loss: 0.2162 (0.5265) acc1: 99.7396 (73.1232) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1936 data: 0.1002 max mem: 3532 +Test: [270/389] eta: 0:00:23 acc1_pos: 23.313 acc5_pos: 100.000 loss: 0.4600 (0.5260) acc1: 88.2812 (73.1886) acc5: 100.0000 (100.0000) acc1_neg: 70.573 acc5_neg: 100.000 time: 0.1692 data: 0.0765 max mem: 3532 +Test: [280/389] eta: 0:00:21 acc1_pos: 23.313 acc5_pos: 100.000 loss: 0.6080 (0.5286) acc1: 69.2708 (72.8333) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1530 data: 0.0612 max mem: 3532 +Test: [290/389] eta: 0:00:19 acc1_pos: 23.313 acc5_pos: 100.000 loss: 0.3818 (0.5205) acc1: 84.1146 (73.5359) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1685 data: 0.0767 max mem: 3532 +Test: [300/389] eta: 0:00:17 acc1_pos: 23.313 acc5_pos: 100.000 loss: 0.3434 (0.5176) acc1: 94.5312 (74.2162) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.1686 data: 0.0758 max mem: 3532 +Test: [310/389] eta: 0:00:15 acc1_pos: 23.313 acc5_pos: 100.000 loss: 0.4706 (0.5162) acc1: 92.4479 (74.4817) acc5: 100.0000 (100.0000) acc1_neg: 76.042 acc5_neg: 100.000 time: 0.1619 data: 0.0695 max mem: 3532 +Test: [320/389] eta: 0:00:13 acc1_pos: 23.313 acc5_pos: 100.000 loss: 0.3924 (0.5104) acc1: 89.8438 (75.0584) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1667 data: 0.0753 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 23.313 acc5_pos: 100.000 loss: 0.3740 (0.5072) acc1: 97.6562 (75.6349) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.1630 data: 0.0711 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 23.313 acc5_pos: 100.000 loss: 0.3723 (0.5003) acc1: 97.9167 (76.2326) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1538 data: 0.0618 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 23.313 acc5_pos: 100.000 loss: 0.3012 (0.5087) acc1: 97.3958 (75.4081) acc5: 100.0000 (100.0000) acc1_neg: 1.042 acc5_neg: 100.000 time: 0.1549 data: 0.0640 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 23.313 acc5_pos: 100.000 loss: 0.7947 (0.5198) acc1: 39.0625 (74.2180) acc5: 100.0000 (100.0000) acc1_neg: 44.010 acc5_neg: 100.000 time: 0.1510 data: 0.0603 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 23.313 acc5_pos: 100.000 loss: 0.7665 (0.5164) acc1: 44.0104 (74.4448) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1438 data: 0.0533 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 23.313 acc5_pos: 100.000 loss: 0.3577 (0.5124) acc1: 85.6771 (74.7656) acc5: 100.0000 (100.0000) acc1_neg: 79.427 acc5_neg: 100.000 time: 0.1386 data: 0.0481 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 23.313 acc5_pos: 100.000 loss: 0.3326 (0.5092) acc1: 88.2812 (75.0110) acc5: 100.0000 (100.0000) acc1_neg: 91.257 acc5_neg: 100.000 time: 0.1220 data: 0.0227 max mem: 3532 +Test: Total time: 0:01:11 (0.1830 s / it) +* Overall Acc@1 75.011 Acc@5 100.000 + target == real Acc@1 56.778 Acc@5 100.000 + target == fake Acc@1 83.486 Acc@5 100.000 + loss 0.509 + PR-AUC 0.8971 ROC-AUC 0.8034 EER 27.15% (thr=0.5977) +Accuracy of the network on the 149358 test images: 75.0% +[rank0]:[W726 05:03:35.486644979 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:03:54.783000 1186070 torch/distributed/run.py:766] +W0726 05:03:54.783000 1186070 torch/distributed/run.py:766] ***************************************** +W0726 05:03:54.783000 1186070 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:03:54.783000 1186070 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:04:21.386752515 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:04:21.550063035 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:04:21.691695989 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:04:21.817135737 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 2): env:// +[rank0]:[W726 05:04:22.219046879 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:04:22.225187674 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:04:22.227090631 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:04:22.256294542 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n22_seed7_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed7_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:36:51 acc1_pos: 80.729 acc5_pos: 100.000 loss: 0.4022 (0.4022) acc1: 80.7292 (80.7292) acc5: 100.0000 (100.0000) time: 6.6423 data: 4.7163 max mem: 3531 +Test: [ 10/333] eta: 0:03:43 acc1_pos: 88.021 acc5_pos: 100.000 loss: 0.3232 (0.3218) acc1: 84.6354 (86.0795) acc5: 100.0000 (100.0000) time: 0.6932 data: 0.4291 max mem: 3531 +Test: [ 20/333] eta: 0:02:08 acc1_pos: 95.052 acc5_pos: 100.000 loss: 0.3141 (0.3139) acc1: 87.5000 (86.7808) acc5: 100.0000 (100.0000) time: 0.0991 data: 0.0004 max mem: 3531 +Test: [ 30/333] eta: 0:01:51 acc1_pos: 64.062 acc5_pos: 100.000 loss: 0.3860 (0.4396) acc1: 77.3438 (77.7722) acc5: 100.0000 (100.0000) time: 0.1901 data: 0.0913 max mem: 3531 +Test: [ 40/333] eta: 0:01:38 acc1_pos: 64.062 acc5_pos: 100.000 loss: 0.7078 (0.5069) acc1: 57.5521 (72.8150) acc5: 100.0000 (100.0000) time: 0.2555 data: 0.1595 max mem: 3531 +Test: [ 50/333] eta: 0:01:27 acc1_pos: 64.323 acc5_pos: 100.000 loss: 0.6937 (0.5489) acc1: 57.5521 (69.5721) acc5: 100.0000 (100.0000) time: 0.2199 data: 0.1260 max mem: 3531 +Test: [ 60/333] eta: 0:01:19 acc1_pos: 51.562 acc5_pos: 100.000 loss: 0.6942 (0.5769) acc1: 58.0729 (67.7510) acc5: 100.0000 (100.0000) time: 0.2052 data: 0.1111 max mem: 3531 +Test: [ 70/333] eta: 0:01:15 acc1_pos: 45.052 acc5_pos: 100.000 loss: 0.7026 (0.5973) acc1: 58.0729 (66.0835) acc5: 100.0000 (100.0000) time: 0.2216 data: 0.1284 max mem: 3531 +Test: [ 80/333] eta: 0:01:09 acc1_pos: 61.458 acc5_pos: 100.000 loss: 0.7114 (0.6128) acc1: 57.0312 (64.6058) acc5: 100.0000 (100.0000) time: 0.2149 data: 0.1235 max mem: 3531 +Test: [ 90/333] eta: 0:01:04 acc1_pos: 66.927 acc5_pos: 100.000 loss: 0.6740 (0.6180) acc1: 59.1146 (64.2914) acc5: 100.0000 (100.0000) time: 0.1881 data: 0.0960 max mem: 3531 +Test: [100/333] eta: 0:00:59 acc1_pos: 71.875 acc5_pos: 100.000 loss: 0.6348 (0.6209) acc1: 64.8438 (64.2327) acc5: 100.0000 (100.0000) time: 0.1839 data: 0.0861 max mem: 3531 +Test: [110/333] eta: 0:00:55 acc1_pos: 63.281 acc5_pos: 100.000 loss: 0.6307 (0.6253) acc1: 64.8438 (63.9288) acc5: 100.0000 (100.0000) time: 0.1906 data: 0.0919 max mem: 3531 +Test: [120/333] eta: 0:00:52 acc1_pos: 69.531 acc5_pos: 100.000 loss: 0.6228 (0.6248) acc1: 63.2812 (64.0001) acc5: 100.0000 (100.0000) time: 0.1951 data: 0.1009 max mem: 3531 +Test: [130/333] eta: 0:00:49 acc1_pos: 75.521 acc5_pos: 100.000 loss: 0.6024 (0.6250) acc1: 63.8021 (64.0287) acc5: 100.0000 (100.0000) time: 0.2042 data: 0.1065 max mem: 3531 +Test: [140/333] eta: 0:00:46 acc1_pos: 79.167 acc5_pos: 100.000 loss: 0.5983 (0.6243) acc1: 64.8438 (64.0921) acc5: 100.0000 (100.0000) time: 0.2099 data: 0.1115 max mem: 3531 +Test: [150/333] eta: 0:00:43 acc1_pos: 62.500 acc5_pos: 100.000 loss: 0.6191 (0.6264) acc1: 66.6667 (64.0504) acc5: 100.0000 (100.0000) time: 0.1999 data: 0.1043 max mem: 3531 +Test: [160/333] eta: 0:00:40 acc1_pos: 55.921 acc5_pos: 100.000 loss: 0.6545 (0.6366) acc1: 61.1979 (63.3314) acc5: 100.0000 (100.0000) acc1_neg: 38.021 acc5_neg: 100.000 time: 0.1897 data: 0.0919 max mem: 3531 +Test: [170/333] eta: 0:00:37 acc1_pos: 55.921 acc5_pos: 100.000 loss: 0.6861 (0.6434) acc1: 57.8125 (62.8137) acc5: 100.0000 (100.0000) acc1_neg: 63.281 acc5_neg: 100.000 time: 0.1719 data: 0.0735 max mem: 3531 +Test: [180/333] eta: 0:00:34 acc1_pos: 55.921 acc5_pos: 100.000 loss: 0.6887 (0.6467) acc1: 58.8542 (62.5791) acc5: 100.0000 (100.0000) acc1_neg: 47.135 acc5_neg: 100.000 time: 0.1560 data: 0.0605 max mem: 3531 +Test: [190/333] eta: 0:00:31 acc1_pos: 55.921 acc5_pos: 100.000 loss: 0.6782 (0.6450) acc1: 59.3750 (62.8109) acc5: 100.0000 (100.0000) acc1_neg: 55.729 acc5_neg: 100.000 time: 0.1692 data: 0.0773 max mem: 3531 +Test: [200/333] eta: 0:00:29 acc1_pos: 55.921 acc5_pos: 100.000 loss: 0.6241 (0.6430) acc1: 64.5833 (63.0169) acc5: 100.0000 (100.0000) acc1_neg: 66.406 acc5_neg: 100.000 time: 0.1705 data: 0.0802 max mem: 3531 +Test: [210/333] eta: 0:00:26 acc1_pos: 55.921 acc5_pos: 100.000 loss: 0.6332 (0.6453) acc1: 64.3229 (62.9196) acc5: 100.0000 (100.0000) acc1_neg: 46.094 acc5_neg: 100.000 time: 0.1685 data: 0.0751 max mem: 3532 +Test: [220/333] eta: 0:00:24 acc1_pos: 55.921 acc5_pos: 100.000 loss: 0.7705 (0.6532) acc1: 56.5104 (62.2997) acc5: 100.0000 (100.0000) acc1_neg: 71.354 acc5_neg: 100.000 time: 0.1845 data: 0.0895 max mem: 3532 +Test: [230/333] eta: 0:00:22 acc1_pos: 55.921 acc5_pos: 100.000 loss: 0.5193 (0.6459) acc1: 73.1771 (62.9859) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.1660 data: 0.0736 max mem: 3532 +Test: [240/333] eta: 0:00:19 acc1_pos: 55.921 acc5_pos: 100.000 loss: 0.5193 (0.6436) acc1: 73.6979 (63.1548) acc5: 100.0000 (100.0000) acc1_neg: 64.062 acc5_neg: 100.000 time: 0.1497 data: 0.0569 max mem: 3532 +Test: [250/333] eta: 0:00:17 acc1_pos: 55.921 acc5_pos: 100.000 loss: 0.6139 (0.6441) acc1: 64.0625 (63.1692) acc5: 100.0000 (100.0000) acc1_neg: 58.333 acc5_neg: 100.000 time: 0.1691 data: 0.0752 max mem: 3532 +Test: [260/333] eta: 0:00:15 acc1_pos: 55.921 acc5_pos: 100.000 loss: 0.6963 (0.6474) acc1: 60.1562 (62.9759) acc5: 100.0000 (100.0000) acc1_neg: 68.229 acc5_neg: 100.000 time: 0.1801 data: 0.0864 max mem: 3532 +Test: [270/333] eta: 0:00:13 acc1_pos: 55.921 acc5_pos: 100.000 loss: 0.6486 (0.6461) acc1: 61.7188 (63.1150) acc5: 100.0000 (100.0000) acc1_neg: 66.667 acc5_neg: 100.000 time: 0.1713 data: 0.0774 max mem: 3532 +Test: [280/333] eta: 0:00:10 acc1_pos: 55.921 acc5_pos: 100.000 loss: 0.6266 (0.6488) acc1: 64.5833 (62.9004) acc5: 100.0000 (100.0000) acc1_neg: 59.115 acc5_neg: 100.000 time: 0.1596 data: 0.0643 max mem: 3532 +Test: [290/333] eta: 0:00:08 acc1_pos: 55.921 acc5_pos: 100.000 loss: 0.8289 (0.6559) acc1: 46.8750 (62.2861) acc5: 100.0000 (100.0000) acc1_neg: 41.927 acc5_neg: 100.000 time: 0.1637 data: 0.0697 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 55.921 acc5_pos: 100.000 loss: 0.8839 (0.6659) acc1: 41.9271 (61.4912) acc5: 100.0000 (100.0000) acc1_neg: 41.146 acc5_neg: 100.000 time: 0.1738 data: 0.0812 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 55.921 acc5_pos: 100.000 loss: 0.8734 (0.6634) acc1: 43.7500 (61.7104) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.1701 data: 0.0770 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 55.921 acc5_pos: 100.000 loss: 0.6264 (0.6632) acc1: 62.5000 (61.7269) acc5: 100.0000 (100.0000) acc1_neg: 57.031 acc5_neg: 100.000 time: 0.1682 data: 0.0770 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 55.921 acc5_pos: 100.000 loss: 0.7425 (0.6677) acc1: 56.7708 (61.4261) acc5: 100.0000 (100.0000) acc1_neg: 54.948 acc5_neg: 100.000 time: 0.1562 data: 0.0658 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 55.921 acc5_pos: 100.000 loss: 0.7764 (0.6687) acc1: 54.1667 (61.3643) acc5: 100.0000 (100.0000) acc1_neg: 51.549 acc5_neg: 100.000 time: 0.1589 data: 0.0652 max mem: 3532 +Test: Total time: 0:01:06 (0.2000 s / it) +* Overall Acc@1 61.364 Acc@5 100.000 + target == real Acc@1 64.036 Acc@5 100.000 + target == fake Acc@1 58.986 Acc@5 100.000 + loss 0.669 + PR-AUC 0.6873 ROC-AUC 0.6658 EER 38.43% (thr=0.4822) +Accuracy of the network on the 127843 test images: 61.4% +[rank0]:[W726 05:05:40.789753176 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:05:59.798000 1186630 torch/distributed/run.py:766] +W0726 05:05:59.798000 1186630 torch/distributed/run.py:766] ***************************************** +W0726 05:05:59.798000 1186630 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:05:59.798000 1186630 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:06:29.178488764 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:06:29.557018396 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:06:29.903875818 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:06:30.974093134 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 3): env:// +[rank0]:[W726 05:06:30.048135187 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:06:30.061899773 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W726 05:06:30.061962608 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:06:30.155176471 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n22_seed7_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed7_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:22:58 acc1_pos: 94.010 acc5_pos: 100.000 loss: 0.2418 (0.2418) acc1: 94.0104 (94.0104) acc5: 100.0000 (100.0000) time: 5.8675 data: 3.9251 max mem: 3531 +Test: [ 10/235] eta: 0:02:19 acc1_pos: 93.750 acc5_pos: 100.000 loss: 0.2418 (0.2399) acc1: 94.2708 (94.0104) acc5: 100.0000 (100.0000) time: 0.6178 data: 0.3572 max mem: 3531 +Test: [ 20/235] eta: 0:01:22 acc1_pos: 94.010 acc5_pos: 100.000 loss: 0.2330 (0.2341) acc1: 94.2708 (94.5189) acc5: 100.0000 (100.0000) time: 0.1114 data: 0.0173 max mem: 3531 +Test: [ 30/235] eta: 0:01:14 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.2236 (0.2320) acc1: 95.0521 (94.5649) acc5: 100.0000 (100.0000) time: 0.2204 data: 0.1202 max mem: 3531 +Test: [ 40/235] eta: 0:01:07 acc1_pos: 95.052 acc5_pos: 100.000 loss: 0.2261 (0.2326) acc1: 94.7917 (94.5058) acc5: 100.0000 (100.0000) time: 0.3000 data: 0.2005 max mem: 3531 +Test: [ 50/235] eta: 0:00:55 acc1_pos: 93.229 acc5_pos: 100.000 loss: 0.2323 (0.2339) acc1: 94.7917 (94.4802) acc5: 100.0000 (100.0000) time: 0.2094 data: 0.1154 max mem: 3531 +Test: [ 60/235] eta: 0:00:49 acc1_pos: 94.531 acc5_pos: 100.000 loss: 0.2399 (0.2345) acc1: 94.5312 (94.5441) acc5: 100.0000 (100.0000) time: 0.1498 data: 0.0557 max mem: 3531 +Test: [ 70/235] eta: 0:00:44 acc1_pos: 95.385 acc5_pos: 100.000 loss: 0.2502 (0.3842) acc1: 92.4479 (85.1122) acc5: 100.0000 (100.0000) acc1_neg: 17.969 acc5_neg: 100.000 time: 0.1963 data: 0.1025 max mem: 3531 +Test: [ 80/235] eta: 0:00:41 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.4374 (0.5162) acc1: 19.2708 (76.8904) acc5: 100.0000 (100.0000) acc1_neg: 16.146 acc5_neg: 100.000 time: 0.2249 data: 0.1273 max mem: 3531 +Test: [ 90/235] eta: 0:00:37 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.4593 (0.6201) acc1: 17.9688 (70.4699) acc5: 100.0000 (100.0000) acc1_neg: 17.969 acc5_neg: 100.000 time: 0.2152 data: 0.1183 max mem: 3531 +Test: [100/235] eta: 0:00:34 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.4549 (0.7007) acc1: 18.4896 (65.4187) acc5: 100.0000 (100.0000) acc1_neg: 16.667 acc5_neg: 100.000 time: 0.2149 data: 0.1218 max mem: 3531 +Test: [110/235] eta: 0:00:31 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.4339 (0.7682) acc1: 18.7500 (61.2167) acc5: 100.0000 (100.0000) acc1_neg: 20.052 acc5_neg: 100.000 time: 0.2159 data: 0.1228 max mem: 3531 +Test: [120/235] eta: 0:00:28 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.4456 (0.8241) acc1: 17.9688 (57.6274) acc5: 100.0000 (100.0000) acc1_neg: 18.490 acc5_neg: 100.000 time: 0.2054 data: 0.1120 max mem: 3531 +Test: [130/235] eta: 0:00:25 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.4376 (0.8696) acc1: 18.4896 (54.6696) acc5: 100.0000 (100.0000) acc1_neg: 16.927 acc5_neg: 100.000 time: 0.2038 data: 0.1110 max mem: 3531 +Test: [140/235] eta: 0:00:22 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.4405 (0.9115) acc1: 17.9688 (51.9670) acc5: 100.0000 (100.0000) acc1_neg: 12.500 acc5_neg: 100.000 time: 0.1994 data: 0.1078 max mem: 3531 +Test: [150/235] eta: 0:00:20 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.4491 (0.9454) acc1: 17.9688 (49.7292) acc5: 100.0000 (100.0000) acc1_neg: 20.052 acc5_neg: 100.000 time: 0.1934 data: 0.1002 max mem: 3531 +Test: [160/235] eta: 0:00:17 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.4251 (0.9757) acc1: 18.4896 (47.8051) acc5: 100.0000 (100.0000) acc1_neg: 18.229 acc5_neg: 100.000 time: 0.1917 data: 0.0997 max mem: 3531 +Test: [170/235] eta: 0:00:15 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.3828 (0.9857) acc1: 20.3125 (46.8582) acc5: 100.0000 (100.0000) acc1_neg: 32.292 acc5_neg: 100.000 time: 0.1949 data: 0.1034 max mem: 3531 +Test: [180/235] eta: 0:00:12 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.1153 (0.9920) acc1: 33.0729 (46.1628) acc5: 100.0000 (100.0000) acc1_neg: 36.979 acc5_neg: 100.000 time: 0.1962 data: 0.1032 max mem: 3531 +Test: [190/235] eta: 0:00:10 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.0942 (0.9981) acc1: 33.8542 (45.5102) acc5: 100.0000 (100.0000) acc1_neg: 32.812 acc5_neg: 100.000 time: 0.2120 data: 0.1194 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.1214 (1.0039) acc1: 32.8125 (44.8953) acc5: 100.0000 (100.0000) acc1_neg: 32.031 acc5_neg: 100.000 time: 0.2010 data: 0.1097 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.1214 (1.0081) acc1: 32.8125 (44.4066) acc5: 100.0000 (100.0000) acc1_neg: 32.031 acc5_neg: 100.000 time: 0.1573 data: 0.0670 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.0880 (1.0117) acc1: 34.1146 (43.9810) acc5: 100.0000 (100.0000) acc1_neg: 33.073 acc5_neg: 100.000 time: 0.1394 data: 0.0476 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.0880 (1.0149) acc1: 34.8958 (43.5989) acc5: 100.0000 (100.0000) acc1_neg: 35.938 acc5_neg: 100.000 time: 0.1259 data: 0.0342 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.0944 (1.0175) acc1: 34.6354 (43.4740) acc5: 100.0000 (100.0000) acc1_neg: 30.435 acc5_neg: 100.000 time: 0.1227 data: 0.0340 max mem: 3532 +Test: Total time: 0:00:50 (0.2141 s / it) +* Overall Acc@1 43.474 Acc@5 100.000 + target == real Acc@1 94.525 Acc@5 100.000 + target == fake Acc@1 25.010 Acc@5 100.000 + loss 1.018 + PR-AUC 0.8755 ROC-AUC 0.7284 EER 33.30% (thr=0.1951) +Accuracy of the network on the 89879 test images: 43.5% +[rank0]:[W726 05:07:29.872418346 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:07:49.310000 1187188 torch/distributed/run.py:766] +W0726 05:07:49.310000 1187188 torch/distributed/run.py:766] ***************************************** +W0726 05:07:49.310000 1187188 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:07:49.310000 1187188 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:08:17.515335097 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:08:18.926327608 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:08:18.056542138 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:08:18.065278313 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 1): env:// +| distributed init (rank 6): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:08:18.415548415 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W726 05:08:18.415635125 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank6]:[W726 05:08:18.415666825 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:08:18.434686519 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n22_seed7_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed7_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:25:08 acc1_pos: 70.312 acc5_pos: 100.000 loss: 0.5830 (0.5830) acc1: 70.3125 (70.3125) acc5: 100.0000 (100.0000) time: 5.6071 data: 4.0154 max mem: 3531 +Test: [ 10/269] eta: 0:02:34 acc1_pos: 80.729 acc5_pos: 100.000 loss: 0.6335 (0.6199) acc1: 64.0625 (66.2642) acc5: 100.0000 (100.0000) time: 0.5951 data: 0.3654 max mem: 3531 +Test: [ 20/269] eta: 0:01:29 acc1_pos: 70.052 acc5_pos: 100.000 loss: 0.6083 (0.5916) acc1: 64.0625 (68.7128) acc5: 100.0000 (100.0000) time: 0.0957 data: 0.0037 max mem: 3531 +Test: [ 30/269] eta: 0:01:18 acc1_pos: 83.594 acc5_pos: 100.000 loss: 0.5255 (0.5672) acc1: 74.4792 (70.8837) acc5: 100.0000 (100.0000) time: 0.1788 data: 0.0874 max mem: 3531 +Test: [ 40/269] eta: 0:01:13 acc1_pos: 39.655 acc5_pos: 100.000 loss: 0.5370 (0.5707) acc1: 72.3958 (70.4205) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.2838 data: 0.1911 max mem: 3531 +Test: [ 50/269] eta: 0:01:03 acc1_pos: 39.655 acc5_pos: 100.000 loss: 0.6034 (0.5832) acc1: 64.5833 (68.8317) acc5: 100.0000 (100.0000) acc1_neg: 37.240 acc5_neg: 100.000 time: 0.2386 data: 0.1453 max mem: 3531 +Test: [ 60/269] eta: 0:00:58 acc1_pos: 39.655 acc5_pos: 100.000 loss: 0.6630 (0.6097) acc1: 57.2917 (66.2910) acc5: 100.0000 (100.0000) acc1_neg: 34.635 acc5_neg: 100.000 time: 0.1999 data: 0.1032 max mem: 3531 +Test: [ 70/269] eta: 0:00:54 acc1_pos: 39.655 acc5_pos: 100.000 loss: 0.6630 (0.6116) acc1: 56.7708 (65.9845) acc5: 100.0000 (100.0000) acc1_neg: 58.854 acc5_neg: 100.000 time: 0.2249 data: 0.1274 max mem: 3531 +Test: [ 80/269] eta: 0:00:49 acc1_pos: 39.655 acc5_pos: 100.000 loss: 0.5771 (0.6037) acc1: 67.1875 (66.7503) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.2049 data: 0.1104 max mem: 3531 +Test: [ 90/269] eta: 0:00:45 acc1_pos: 39.655 acc5_pos: 100.000 loss: 0.5825 (0.6090) acc1: 63.8021 (66.0342) acc5: 100.0000 (100.0000) acc1_neg: 63.542 acc5_neg: 100.000 time: 0.1749 data: 0.0782 max mem: 3531 +Test: [100/269] eta: 0:00:40 acc1_pos: 39.655 acc5_pos: 100.000 loss: 0.6781 (0.6193) acc1: 58.0729 (65.1454) acc5: 100.0000 (100.0000) acc1_neg: 51.042 acc5_neg: 100.000 time: 0.1599 data: 0.0640 max mem: 3531 +Test: [110/269] eta: 0:00:37 acc1_pos: 39.655 acc5_pos: 100.000 loss: 0.7625 (0.6343) acc1: 51.0417 (63.8115) acc5: 100.0000 (100.0000) acc1_neg: 45.573 acc5_neg: 100.000 time: 0.1807 data: 0.0887 max mem: 3531 +Test: [120/269] eta: 0:00:35 acc1_pos: 39.655 acc5_pos: 100.000 loss: 0.7181 (0.6355) acc1: 54.1667 (63.7311) acc5: 100.0000 (100.0000) acc1_neg: 63.021 acc5_neg: 100.000 time: 0.1999 data: 0.1070 max mem: 3531 +Test: [130/269] eta: 0:00:32 acc1_pos: 39.655 acc5_pos: 100.000 loss: 0.7018 (0.6453) acc1: 57.8125 (62.9672) acc5: 100.0000 (100.0000) acc1_neg: 51.302 acc5_neg: 100.000 time: 0.1942 data: 0.1013 max mem: 3531 +Test: [140/269] eta: 0:00:29 acc1_pos: 39.655 acc5_pos: 100.000 loss: 0.8107 (0.6609) acc1: 49.7396 (61.7188) acc5: 100.0000 (100.0000) acc1_neg: 35.417 acc5_neg: 100.000 time: 0.1791 data: 0.0863 max mem: 3531 +Test: [150/269] eta: 0:00:26 acc1_pos: 39.655 acc5_pos: 100.000 loss: 0.7931 (0.6695) acc1: 47.1354 (60.9496) acc5: 100.0000 (100.0000) acc1_neg: 53.125 acc5_neg: 100.000 time: 0.1935 data: 0.1002 max mem: 3531 +Test: [160/269] eta: 0:00:24 acc1_pos: 39.655 acc5_pos: 100.000 loss: 0.7108 (0.6741) acc1: 54.4271 (60.6399) acc5: 100.0000 (100.0000) acc1_neg: 63.802 acc5_neg: 100.000 time: 0.1960 data: 0.1039 max mem: 3531 +Test: [170/269] eta: 0:00:22 acc1_pos: 39.655 acc5_pos: 100.000 loss: 0.7083 (0.6782) acc1: 54.6875 (60.3451) acc5: 100.0000 (100.0000) acc1_neg: 69.792 acc5_neg: 100.000 time: 0.1899 data: 0.0978 max mem: 3531 +Test: [180/269] eta: 0:00:19 acc1_pos: 39.655 acc5_pos: 100.000 loss: 0.5852 (0.6690) acc1: 68.7500 (61.2555) acc5: 100.0000 (100.0000) acc1_neg: 66.146 acc5_neg: 100.000 time: 0.1982 data: 0.1012 max mem: 3531 +Test: [190/269] eta: 0:00:17 acc1_pos: 39.655 acc5_pos: 100.000 loss: 0.6001 (0.6693) acc1: 68.7500 (61.1843) acc5: 100.0000 (100.0000) acc1_neg: 57.552 acc5_neg: 100.000 time: 0.1779 data: 0.0819 max mem: 3531 +Test: [200/269] eta: 0:00:14 acc1_pos: 39.655 acc5_pos: 100.000 loss: 0.6593 (0.6716) acc1: 57.5521 (61.1163) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.1699 data: 0.0787 max mem: 3531 +Test: [210/269] eta: 0:00:12 acc1_pos: 39.655 acc5_pos: 100.000 loss: 0.7135 (0.6734) acc1: 56.2500 (60.9548) acc5: 100.0000 (100.0000) acc1_neg: 66.927 acc5_neg: 100.000 time: 0.1728 data: 0.0816 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 39.655 acc5_pos: 100.000 loss: 0.6557 (0.6732) acc1: 61.9792 (61.0801) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1717 data: 0.0814 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 39.655 acc5_pos: 100.000 loss: 0.5497 (0.6686) acc1: 68.4896 (61.4538) acc5: 100.0000 (100.0000) acc1_neg: 76.302 acc5_neg: 100.000 time: 0.1698 data: 0.0792 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 39.655 acc5_pos: 100.000 loss: 0.6660 (0.6732) acc1: 58.8542 (60.9645) acc5: 100.0000 (100.0000) acc1_neg: 63.542 acc5_neg: 100.000 time: 0.1719 data: 0.0797 max mem: 3532 +Test: [250/269] eta: 0:00:03 acc1_pos: 39.655 acc5_pos: 100.000 loss: 0.7615 (0.6772) acc1: 51.8229 (60.6242) acc5: 100.0000 (100.0000) acc1_neg: 37.500 acc5_neg: 100.000 time: 0.1499 data: 0.0563 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 39.655 acc5_pos: 100.000 loss: 0.8047 (0.6844) acc1: 42.7083 (59.8489) acc5: 100.0000 (100.0000) acc1_neg: 48.698 acc5_neg: 100.000 time: 0.1221 data: 0.0301 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 39.655 acc5_pos: 100.000 loss: 0.8047 (0.6849) acc1: 44.5312 (59.7311) acc5: 100.0000 (100.0000) acc1_neg: 83.727 acc5_neg: 100.000 time: 0.1132 data: 0.0193 max mem: 3532 +Test: Total time: 0:00:53 (0.1997 s / it) +* Overall Acc@1 59.731 Acc@5 100.000 + target == real Acc@1 70.518 Acc@5 100.000 + target == fake Acc@1 58.259 Acc@5 100.000 + loss 0.685 + PR-AUC 0.9390 ROC-AUC 0.6964 EER 35.69% (thr=0.4619) +Accuracy of the network on the 103293 test images: 59.7% +[rank0]:[W726 05:09:22.376544766 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T05:09:29Z] eval_done tag=fixed500k_fake_n22_seed7_resampled +[2026-07-26T05:09:29Z] eval_start tag=fixed500k_real_n11_seed8_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n11_classseed8_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T05:09:29Z +hostname=n116-041-188 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n11_classseed8_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n11_seed8_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 05:09:42.620000 1187751 torch/distributed/run.py:766] +W0726 05:09:42.620000 1187751 torch/distributed/run.py:766] ***************************************** +W0726 05:09:42.620000 1187751 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:09:42.620000 1187751 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 05:10:10.918561701 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:10:10.993975500 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:10:10.379775815 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:10:10.436811065 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +[rank4]:[W726 05:10:10.703528148 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:10:10.710714527 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:10:10.788392436 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:10:10.804239141 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n11_seed8_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n11_classseed8_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:23:18 acc1_pos: 0.781 acc5_pos: 100.000 loss: 2.1834 (2.1834) acc1: 0.7812 (0.7812) acc5: 100.0000 (100.0000) time: 5.9758 data: 4.3094 max mem: 3531 +Test: [ 10/234] eta: 0:02:20 acc1_pos: 0.260 acc5_pos: 100.000 loss: 1.9242 (1.9172) acc1: 3.6458 (7.1970) acc5: 100.0000 (100.0000) time: 0.6252 data: 0.3921 max mem: 3531 +Test: [ 20/234] eta: 0:01:20 acc1_pos: 91.146 acc5_pos: 100.000 loss: 1.9242 (1.8247) acc1: 2.3438 (18.9856) acc5: 100.0000 (100.0000) time: 0.0942 data: 0.0004 max mem: 3531 +Test: [ 30/234] eta: 0:01:08 acc1_pos: 55.729 acc5_pos: 100.000 loss: 0.2844 (1.3516) acc1: 90.1042 (40.7426) acc5: 100.0000 (100.0000) time: 0.1794 data: 0.0845 max mem: 3531 +Test: [ 40/234] eta: 0:01:03 acc1_pos: 86.979 acc5_pos: 100.000 loss: 0.3317 (1.1909) acc1: 86.9792 (46.1001) acc5: 100.0000 (100.0000) time: 0.2749 data: 0.1806 max mem: 3531 +Test: [ 50/234] eta: 0:00:56 acc1_pos: 88.021 acc5_pos: 100.000 loss: 0.3698 (1.0261) acc1: 85.1562 (54.0492) acc5: 100.0000 (100.0000) time: 0.2596 data: 0.1637 max mem: 3531 +Test: [ 60/234] eta: 0:00:47 acc1_pos: 87.586 acc5_pos: 100.000 loss: 0.1739 (0.8728) acc1: 90.1042 (61.4968) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1666 data: 0.0698 max mem: 3531 +Test: [ 70/234] eta: 0:00:42 acc1_pos: 87.586 acc5_pos: 100.000 loss: 0.0832 (0.7955) acc1: 100.0000 (64.8217) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.1471 data: 0.0522 max mem: 3531 +Test: [ 80/234] eta: 0:00:39 acc1_pos: 87.586 acc5_pos: 100.000 loss: 0.4117 (0.7341) acc1: 79.6875 (67.5219) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1956 data: 0.1042 max mem: 3531 +Test: [ 90/234] eta: 0:00:35 acc1_pos: 87.586 acc5_pos: 100.000 loss: 0.2054 (0.6737) acc1: 92.4479 (70.4985) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1977 data: 0.1051 max mem: 3531 +Test: [100/234] eta: 0:00:32 acc1_pos: 87.586 acc5_pos: 100.000 loss: 0.2329 (0.6393) acc1: 90.3646 (72.0091) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1777 data: 0.0854 max mem: 3531 +Test: [110/234] eta: 0:00:29 acc1_pos: 87.586 acc5_pos: 100.000 loss: 0.3260 (0.6056) acc1: 86.1979 (73.5525) acc5: 100.0000 (100.0000) acc1_neg: 78.646 acc5_neg: 100.000 time: 0.1949 data: 0.1039 max mem: 3531 +Test: [120/234] eta: 0:00:27 acc1_pos: 87.586 acc5_pos: 100.000 loss: 0.2926 (0.5814) acc1: 88.2812 (74.6169) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.2263 data: 0.1348 max mem: 3531 +Test: [130/234] eta: 0:00:24 acc1_pos: 87.586 acc5_pos: 100.000 loss: 0.1564 (0.5515) acc1: 93.2292 (76.1331) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2085 data: 0.1154 max mem: 3531 +Test: [140/234] eta: 0:00:21 acc1_pos: 87.586 acc5_pos: 100.000 loss: 0.2211 (0.5524) acc1: 88.5417 (75.8071) acc5: 100.0000 (100.0000) acc1_neg: 56.771 acc5_neg: 100.000 time: 0.1890 data: 0.0969 max mem: 3531 +Test: [150/234] eta: 0:00:19 acc1_pos: 87.586 acc5_pos: 100.000 loss: 0.3637 (0.5423) acc1: 80.2083 (76.1107) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.1981 data: 0.1078 max mem: 3531 +Test: [160/234] eta: 0:00:16 acc1_pos: 87.586 acc5_pos: 100.000 loss: 0.3292 (0.5234) acc1: 85.9375 (77.0720) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.2002 data: 0.1072 max mem: 3531 +Test: [170/234] eta: 0:00:14 acc1_pos: 87.586 acc5_pos: 100.000 loss: 0.3128 (0.5105) acc1: 86.7188 (77.7382) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1837 data: 0.0881 max mem: 3531 +Test: [180/234] eta: 0:00:12 acc1_pos: 87.586 acc5_pos: 100.000 loss: 0.3518 (0.5062) acc1: 84.6354 (77.7437) acc5: 100.0000 (100.0000) acc1_neg: 60.417 acc5_neg: 100.000 time: 0.1939 data: 0.0998 max mem: 3531 +Test: [190/234] eta: 0:00:09 acc1_pos: 87.586 acc5_pos: 100.000 loss: 0.5802 (0.5117) acc1: 68.4896 (77.1965) acc5: 100.0000 (100.0000) acc1_neg: 66.406 acc5_neg: 100.000 time: 0.1998 data: 0.1072 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 87.586 acc5_pos: 100.000 loss: 0.5787 (0.5000) acc1: 67.9688 (77.7363) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1868 data: 0.0940 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 87.586 acc5_pos: 100.000 loss: 0.0888 (0.4809) acc1: 100.0000 (78.7791) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1921 data: 0.0995 max mem: 3532 +Test: [220/234] eta: 0:00:03 acc1_pos: 87.586 acc5_pos: 100.000 loss: 0.1153 (0.4663) acc1: 99.7396 (79.5991) acc5: 100.0000 (100.0000) acc1_neg: 74.740 acc5_neg: 100.000 time: 0.1672 data: 0.0762 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 87.586 acc5_pos: 100.000 loss: 0.4618 (0.4944) acc1: 29.9479 (77.3156) acc5: 100.0000 (100.0000) acc1_neg: 24.219 acc5_neg: 100.000 time: 0.1273 data: 0.0374 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 87.586 acc5_pos: 100.000 loss: 1.0974 (0.5022) acc1: 28.3854 (76.7522) acc5: 100.0000 (100.0000) acc1_neg: 28.405 acc5_neg: 100.000 time: 0.1190 data: 0.0281 max mem: 3532 +Test: Total time: 0:00:49 (0.2106 s / it) +* Overall Acc@1 76.752 Acc@5 100.000 + target == real Acc@1 54.296 Acc@5 100.000 + target == fake Acc@1 83.081 Acc@5 100.000 + loss 0.502 + PR-AUC 0.8983 ROC-AUC 0.7505 EER 31.92% (thr=0.7886) +Accuracy of the network on the 89729 test images: 76.8% +[rank0]:[W726 05:11:08.341063370 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:11:27.017000 1188309 torch/distributed/run.py:766] +W0726 05:11:27.017000 1188309 torch/distributed/run.py:766] ***************************************** +W0726 05:11:27.017000 1188309 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:11:27.017000 1188309 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:11:55.614432802 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:11:55.689743937 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:11:55.726138298 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:11:55.872580180 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:11:56.916036522 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +[rank4]:[W726 05:11:56.962900441 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:11:56.973622364 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:11:56.114509013 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n11_seed8_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n11_classseed8_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 3:17:51 acc1_pos: 1.302 acc5_pos: 100.000 loss: 2.0701 (2.0701) acc1: 1.3021 (1.3021) acc5: 100.0000 (100.0000) time: 4.9735 data: 2.8913 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/2387] eta: 0:21:50 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.2080 (2.1874) acc1: 0.5208 (2.0360) acc5: 100.0000 (100.0000) time: 0.5513 data: 0.2805 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 20/2387] eta: 0:14:10 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.2999 (2.2455) acc1: 0.2604 (1.4137) acc5: 100.0000 (100.0000) time: 0.1287 data: 0.0391 max mem: 3531 +Test: [ 30/2387] eta: 0:11:26 acc1_pos: 3.646 acc5_pos: 100.000 loss: 2.3014 (2.2621) acc1: 0.2604 (1.2601) acc5: 100.0000 (100.0000) time: 0.1482 data: 0.0566 max mem: 3531 +Test: [ 40/2387] eta: 0:11:24 acc1_pos: 4.688 acc5_pos: 100.000 loss: 2.2359 (2.2532) acc1: 0.5208 (1.2957) acc5: 100.0000 (100.0000) time: 0.2204 data: 0.1266 max mem: 3531 +Test: [ 50/2387] eta: 0:12:23 acc1_pos: 11.979 acc5_pos: 100.000 loss: 2.1597 (2.2025) acc1: 1.8229 (2.3540) acc5: 100.0000 (100.0000) time: 0.3599 data: 0.2668 max mem: 3531 +Test: [ 60/2387] eta: 0:12:01 acc1_pos: 11.719 acc5_pos: 100.000 loss: 2.0488 (2.1997) acc1: 3.3854 (2.7408) acc5: 100.0000 (100.0000) time: 0.3485 data: 0.2550 max mem: 3531 +Test: [ 70/2387] eta: 0:11:29 acc1_pos: 8.073 acc5_pos: 100.000 loss: 2.3102 (2.2178) acc1: 1.3021 (2.6592) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1491 max mem: 3531 +Test: [ 80/2387] eta: 0:11:15 acc1_pos: 3.385 acc5_pos: 100.000 loss: 2.2985 (2.2249) acc1: 0.5208 (2.4884) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1426 max mem: 3531 +Test: [ 90/2387] eta: 0:11:10 acc1_pos: 3.646 acc5_pos: 100.000 loss: 2.2479 (2.2255) acc1: 1.5625 (2.5126) acc5: 100.0000 (100.0000) time: 0.2723 data: 0.1769 max mem: 3531 +Test: [ 100/2387] eta: 0:10:53 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.2354 (2.2297) acc1: 1.3021 (2.5088) acc5: 100.0000 (100.0000) time: 0.2569 data: 0.1641 max mem: 3531 +Test: [ 110/2387] eta: 0:10:42 acc1_pos: 3.906 acc5_pos: 100.000 loss: 2.2442 (2.2285) acc1: 0.7812 (2.6370) acc5: 100.0000 (100.0000) time: 0.2377 data: 0.1446 max mem: 3531 +Test: [ 120/2387] eta: 0:10:35 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.2049 (2.2246) acc1: 2.3438 (2.7118) acc5: 100.0000 (100.0000) time: 0.2530 data: 0.1553 max mem: 3531 +Test: [ 130/2387] eta: 0:10:25 acc1_pos: 15.885 acc5_pos: 100.000 loss: 2.1713 (2.2235) acc1: 2.0833 (2.8467) acc5: 100.0000 (100.0000) time: 0.2501 data: 0.1500 max mem: 3531 +Test: [ 140/2387] eta: 0:10:17 acc1_pos: 2.083 acc5_pos: 100.000 loss: 2.2153 (2.2218) acc1: 1.0417 (2.8313) acc5: 100.0000 (100.0000) time: 0.2398 data: 0.1423 max mem: 3531 +Test: [ 150/2387] eta: 0:10:23 acc1_pos: 0.521 acc5_pos: 100.000 loss: 2.1291 (2.2150) acc1: 2.3438 (2.9784) acc5: 100.0000 (100.0000) time: 0.2896 data: 0.1920 max mem: 3531 +Test: [ 160/2387] eta: 0:09:56 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.0654 (2.1669) acc1: 3.6458 (4.9738) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2199 data: 0.1210 max mem: 3531 +Test: [ 170/2387] eta: 0:09:38 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0990 (2.0459) acc1: 99.4792 (10.5172) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1266 data: 0.0282 max mem: 3531 +Test: [ 180/2387] eta: 0:09:32 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0972 (1.9382) acc1: 99.7396 (15.4466) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1904 data: 0.0947 max mem: 3531 +Test: [ 190/2387] eta: 0:09:20 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0960 (1.8418) acc1: 99.7396 (19.8598) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2035 data: 0.1084 max mem: 3531 +Test: [ 200/2387] eta: 0:09:09 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0958 (1.7550) acc1: 99.7396 (23.8327) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1795 data: 0.0830 max mem: 3531 +Test: [ 210/2387] eta: 0:09:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0995 (1.6766) acc1: 99.7396 (27.4227) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1846 data: 0.0876 max mem: 3532 +Test: [ 220/2387] eta: 0:08:50 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0982 (1.6051) acc1: 99.7396 (30.6974) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1813 data: 0.0840 max mem: 3532 +Test: [ 230/2387] eta: 0:08:44 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0972 (1.5398) acc1: 99.7396 (33.6873) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1898 data: 0.0917 max mem: 3532 +Test: [ 240/2387] eta: 0:08:36 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0942 (1.4796) acc1: 99.7396 (36.4270) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1934 data: 0.0949 max mem: 3532 +Test: [ 250/2387] eta: 0:08:30 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0851 (1.4234) acc1: 99.7396 (38.9546) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1902 data: 0.0923 max mem: 3532 +Test: [ 260/2387] eta: 0:08:26 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0623 (1.3712) acc1: 100.0000 (41.2935) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2061 data: 0.1088 max mem: 3532 +Test: [ 270/2387] eta: 0:08:21 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0602 (1.3228) acc1: 100.0000 (43.4588) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2105 data: 0.1145 max mem: 3532 +Test: [ 280/2387] eta: 0:08:14 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0609 (1.2780) acc1: 100.0000 (45.4700) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1903 data: 0.0950 max mem: 3532 +Test: [ 290/2387] eta: 0:08:07 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0613 (1.2362) acc1: 100.0000 (47.3430) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1717 data: 0.0766 max mem: 3532 +Test: [ 300/2387] eta: 0:08:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0596 (1.1972) acc1: 100.0000 (49.0872) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1889 data: 0.0951 max mem: 3532 +Test: [ 310/2387] eta: 0:08:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0608 (1.1621) acc1: 100.0000 (50.7067) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2192 data: 0.1229 max mem: 3532 +Test: [ 320/2387] eta: 0:07:57 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1494 (1.1307) acc1: 99.4792 (52.2221) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2215 data: 0.1266 max mem: 3532 +Test: [ 330/2387] eta: 0:07:49 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1520 (1.1012) acc1: 99.4792 (53.6521) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1728 data: 0.0824 max mem: 3532 +Test: [ 340/2387] eta: 0:07:47 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1544 (1.0734) acc1: 99.4792 (54.9976) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1885 data: 0.0980 max mem: 3532 +Test: [ 350/2387] eta: 0:07:46 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1429 (1.0452) acc1: 99.7396 (56.2767) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2424 data: 0.1515 max mem: 3532 +Test: [ 360/2387] eta: 0:07:42 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0890 (1.0187) acc1: 100.0000 (57.4843) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2256 data: 0.1331 max mem: 3532 +Test: [ 370/2387] eta: 0:07:34 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0897 (0.9937) acc1: 100.0000 (58.6274) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1594 data: 0.0653 max mem: 3532 +Test: [ 380/2387] eta: 0:07:30 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0901 (0.9700) acc1: 100.0000 (59.7045) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1510 data: 0.0559 max mem: 3532 +Test: [ 390/2387] eta: 0:07:27 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0904 (0.9475) acc1: 100.0000 (60.7310) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2036 data: 0.1092 max mem: 3532 +Test: [ 400/2387] eta: 0:07:23 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0905 (0.9262) acc1: 100.0000 (61.7058) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2043 data: 0.1118 max mem: 3532 +Test: [ 410/2387] eta: 0:07:19 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0907 (0.9058) acc1: 99.7396 (62.6337) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1911 data: 0.0984 max mem: 3532 +Test: [ 420/2387] eta: 0:07:16 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0879 (0.8864) acc1: 99.7396 (63.5151) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1898 data: 0.0952 max mem: 3532 +Test: [ 430/2387] eta: 0:07:12 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0885 (0.8679) acc1: 99.7396 (64.3561) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1908 data: 0.0967 max mem: 3532 +Test: [ 440/2387] eta: 0:07:07 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0845 (0.8501) acc1: 100.0000 (65.1632) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1779 data: 0.0839 max mem: 3533 +Test: [ 450/2387] eta: 0:07:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0762 (0.8329) acc1: 100.0000 (65.9357) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1691 data: 0.0728 max mem: 3533 +Test: [ 460/2387] eta: 0:06:59 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0758 (0.8165) acc1: 100.0000 (66.6746) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1772 data: 0.0803 max mem: 3533 +Test: [ 470/2387] eta: 0:06:56 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0761 (0.8007) acc1: 100.0000 (67.3821) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1835 data: 0.0897 max mem: 3533 +Test: [ 480/2387] eta: 0:06:52 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0764 (0.7857) acc1: 100.0000 (68.0602) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1789 data: 0.0883 max mem: 3533 +Test: [ 490/2387] eta: 0:06:48 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0760 (0.7712) acc1: 100.0000 (68.7108) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1726 data: 0.0799 max mem: 3533 +Test: [ 500/2387] eta: 0:06:45 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0763 (0.7574) acc1: 100.0000 (69.3353) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1767 data: 0.0836 max mem: 3533 +Test: [ 510/2387] eta: 0:06:40 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0763 (0.7441) acc1: 100.0000 (69.9354) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1692 data: 0.0785 max mem: 3533 +Test: [ 520/2387] eta: 0:06:37 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0778 (0.7313) acc1: 100.0000 (70.5124) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1678 data: 0.0767 max mem: 3533 +Test: [ 530/2387] eta: 0:06:33 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0774 (0.7190) acc1: 100.0000 (71.0678) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1755 data: 0.0839 max mem: 3533 +Test: [ 540/2387] eta: 0:06:30 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0769 (0.7071) acc1: 100.0000 (71.6026) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1723 data: 0.0809 max mem: 3533 +Test: [ 550/2387] eta: 0:06:27 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0769 (0.6957) acc1: 100.0000 (72.1179) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1730 data: 0.0807 max mem: 3533 +Test: [ 560/2387] eta: 0:06:23 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0770 (0.6847) acc1: 100.0000 (72.6149) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1781 data: 0.0860 max mem: 3533 +Test: [ 570/2387] eta: 0:06:20 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0768 (0.6740) acc1: 100.0000 (73.0945) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1687 data: 0.0756 max mem: 3533 +Test: [ 580/2387] eta: 0:06:17 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0765 (0.6637) acc1: 100.0000 (73.5576) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1707 data: 0.0776 max mem: 3533 +Test: [ 590/2387] eta: 0:06:13 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0767 (0.6538) acc1: 100.0000 (74.0050) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1755 data: 0.0826 max mem: 3533 +Test: [ 600/2387] eta: 0:06:11 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0784 (0.6443) acc1: 100.0000 (74.4376) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1804 data: 0.0865 max mem: 3533 +Test: [ 610/2387] eta: 0:06:07 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0813 (0.6350) acc1: 100.0000 (74.8555) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1769 data: 0.0839 max mem: 3533 +Test: [ 620/2387] eta: 0:06:04 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0777 (0.6261) acc1: 100.0000 (75.2600) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1651 data: 0.0729 max mem: 3533 +Test: [ 630/2387] eta: 0:06:02 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0782 (0.6174) acc1: 100.0000 (75.6521) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1758 data: 0.0839 max mem: 3533 +Test: [ 640/2387] eta: 0:05:58 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0789 (0.6090) acc1: 100.0000 (76.0311) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1585 data: 0.0670 max mem: 3533 +Test: [ 650/2387] eta: 0:05:55 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0780 (0.6008) acc1: 100.0000 (76.3993) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1682 data: 0.0763 max mem: 3533 +Test: [ 660/2387] eta: 0:05:54 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0784 (0.5929) acc1: 100.0000 (76.7555) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2134 data: 0.1218 max mem: 3533 +Test: [ 670/2387] eta: 0:05:51 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0784 (0.5852) acc1: 100.0000 (77.1012) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1945 data: 0.1020 max mem: 3534 +Test: [ 680/2387] eta: 0:05:48 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0771 (0.5777) acc1: 100.0000 (77.4371) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1752 data: 0.0828 max mem: 3534 +Test: [ 690/2387] eta: 0:05:46 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0776 (0.5722) acc1: 100.0000 (77.6618) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.1941 data: 0.1034 max mem: 3534 +Test: [ 700/2387] eta: 0:05:45 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2420 (0.5676) acc1: 90.8854 (77.8490) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.2174 data: 0.1266 max mem: 3534 +Test: [ 710/2387] eta: 0:05:43 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2420 (0.5631) acc1: 90.8854 (78.0345) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.2215 data: 0.1306 max mem: 3534 +Test: [ 720/2387] eta: 0:05:41 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2437 (0.5587) acc1: 91.4062 (78.2193) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.2169 data: 0.1237 max mem: 3534 +Test: [ 730/2387] eta: 0:05:39 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2433 (0.5543) acc1: 91.4062 (78.3986) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.2048 data: 0.1100 max mem: 3534 +Test: [ 740/2387] eta: 0:05:38 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2431 (0.5501) acc1: 91.4062 (78.5741) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.2187 data: 0.1254 max mem: 3534 +Test: [ 750/2387] eta: 0:05:35 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2397 (0.5460) acc1: 91.4062 (78.7460) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.2192 data: 0.1276 max mem: 3534 +Test: [ 760/2387] eta: 0:05:33 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2330 (0.5414) acc1: 91.6667 (78.9401) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1889 data: 0.0968 max mem: 3534 +Test: [ 770/2387] eta: 0:05:33 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1232 (0.5358) acc1: 97.9167 (79.1944) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2393 data: 0.1462 max mem: 3534 +Test: [ 780/2387] eta: 0:05:30 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0878 (0.5300) acc1: 99.7396 (79.4598) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2273 data: 0.1344 max mem: 3534 +Test: [ 790/2387] eta: 0:05:27 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0781 (0.5247) acc1: 100.0000 (79.6931) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1740 data: 0.0812 max mem: 3534 +Test: [ 800/2387] eta: 0:05:25 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0763 (0.5190) acc1: 100.0000 (79.9460) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1807 data: 0.0890 max mem: 3534 +Test: [ 810/2387] eta: 0:05:22 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0747 (0.5136) acc1: 100.0000 (80.1910) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1858 data: 0.0944 max mem: 3534 +Test: [ 820/2387] eta: 0:05:20 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0783 (0.5084) acc1: 100.0000 (80.4310) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1816 data: 0.0879 max mem: 3534 +Test: [ 830/2387] eta: 0:05:17 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0875 (0.5035) acc1: 99.7396 (80.6505) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1797 data: 0.0863 max mem: 3534 +Test: [ 840/2387] eta: 0:05:15 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1218 (0.4993) acc1: 97.9167 (80.8521) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1829 data: 0.0911 max mem: 3534 +Test: [ 850/2387] eta: 0:05:13 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1355 (0.4950) acc1: 97.9167 (81.0542) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1870 data: 0.0953 max mem: 3534 +Test: [ 860/2387] eta: 0:05:10 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1247 (0.4907) acc1: 98.6979 (81.2588) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1812 data: 0.0899 max mem: 3534 +Test: [ 870/2387] eta: 0:05:08 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1223 (0.4866) acc1: 98.9583 (81.4527) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1845 data: 0.0919 max mem: 3534 +Test: [ 880/2387] eta: 0:05:06 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1253 (0.4826) acc1: 99.2188 (81.6505) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2098 data: 0.1170 max mem: 3534 +Test: [ 890/2387] eta: 0:05:04 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1025 (0.4781) acc1: 99.7396 (81.8553) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1994 data: 0.1075 max mem: 3534 +Test: [ 900/2387] eta: 0:05:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0954 (0.4811) acc1: 99.4792 (81.5723) acc5: 100.0000 (100.0000) acc1_neg: 45.833 acc5_neg: 100.000 time: 0.1781 data: 0.0846 max mem: 3535 +Test: [ 910/2387] eta: 0:04:59 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.8962 (0.4856) acc1: 46.3542 (81.1905) acc5: 100.0000 (100.0000) acc1_neg: 48.177 acc5_neg: 100.000 time: 0.1846 data: 0.0918 max mem: 3535 +Test: [ 920/2387] eta: 0:04:57 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.9035 (0.4902) acc1: 46.3542 (80.8083) acc5: 100.0000 (100.0000) acc1_neg: 43.750 acc5_neg: 100.000 time: 0.2116 data: 0.1200 max mem: 3535 +Test: [ 930/2387] eta: 0:04:57 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.9047 (0.4946) acc1: 46.0938 (80.4397) acc5: 100.0000 (100.0000) acc1_neg: 44.531 acc5_neg: 100.000 time: 0.2534 data: 0.1614 max mem: 3535 +Test: [ 940/2387] eta: 0:04:53 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.9047 (0.4990) acc1: 45.8333 (80.0788) acc5: 100.0000 (100.0000) acc1_neg: 45.573 acc5_neg: 100.000 time: 0.2043 data: 0.1109 max mem: 3535 +Test: [ 950/2387] eta: 0:04:52 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.8979 (0.5032) acc1: 46.8750 (79.7231) acc5: 100.0000 (100.0000) acc1_neg: 46.875 acc5_neg: 100.000 time: 0.1967 data: 0.1009 max mem: 3535 +Test: [ 960/2387] eta: 0:04:50 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.8886 (0.5071) acc1: 46.8750 (79.3881) acc5: 100.0000 (100.0000) acc1_neg: 51.042 acc5_neg: 100.000 time: 0.2387 data: 0.1450 max mem: 3535 +Test: [ 970/2387] eta: 0:04:48 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.8921 (0.5112) acc1: 46.0938 (79.0462) acc5: 100.0000 (100.0000) acc1_neg: 45.573 acc5_neg: 100.000 time: 0.1891 data: 0.0945 max mem: 3535 +Test: [ 980/2387] eta: 0:04:46 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.8736 (0.5098) acc1: 47.1354 (79.0698) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1763 data: 0.0821 max mem: 3535 +Test: [ 990/2387] eta: 0:04:43 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0945 (0.5056) acc1: 99.7396 (79.2773) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1952 data: 0.1040 max mem: 3535 +Test: [1000/2387] eta: 0:04:41 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0830 (0.5015) acc1: 100.0000 (79.4760) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1945 data: 0.1035 max mem: 3535 +Test: [1010/2387] eta: 0:04:39 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0892 (0.4975) acc1: 100.0000 (79.6738) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1886 data: 0.0948 max mem: 3535 +Test: [1020/2387] eta: 0:04:37 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0964 (0.4937) acc1: 100.0000 (79.8650) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1896 data: 0.0947 max mem: 3535 +Test: [1030/2387] eta: 0:04:35 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0816 (0.4896) acc1: 100.0000 (80.0603) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1910 data: 0.0979 max mem: 3535 +Test: [1040/2387] eta: 0:04:32 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0722 (0.4856) acc1: 100.0000 (80.2516) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1831 data: 0.0886 max mem: 3535 +Test: [1050/2387] eta: 0:04:30 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0697 (0.4817) acc1: 100.0000 (80.4393) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1761 data: 0.0829 max mem: 3535 +Test: [1060/2387] eta: 0:04:28 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0708 (0.4779) acc1: 100.0000 (80.6236) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1864 data: 0.0949 max mem: 3535 +Test: [1070/2387] eta: 0:04:26 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0757 (0.4742) acc1: 100.0000 (80.8043) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1952 data: 0.1038 max mem: 3535 +Test: [1080/2387] eta: 0:04:24 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0862 (0.4707) acc1: 100.0000 (80.9771) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2002 data: 0.1092 max mem: 3535 +Test: [1090/2387] eta: 0:04:21 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0957 (0.4673) acc1: 99.7396 (81.1490) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1900 data: 0.0981 max mem: 3535 +Test: [1100/2387] eta: 0:04:20 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0909 (0.4638) acc1: 99.7396 (81.3176) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2018 data: 0.1098 max mem: 3535 +Test: [1110/2387] eta: 0:04:17 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0909 (0.4607) acc1: 99.7396 (81.4652) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1902 data: 0.0988 max mem: 3535 +Test: [1120/2387] eta: 0:04:15 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0869 (0.4574) acc1: 99.7396 (81.6266) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1876 data: 0.0947 max mem: 3536 +Test: [1130/2387] eta: 0:04:13 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0863 (0.4542) acc1: 100.0000 (81.7874) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1972 data: 0.1027 max mem: 3536 +Test: [1140/2387] eta: 0:04:11 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0887 (0.4510) acc1: 100.0000 (81.9459) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1749 data: 0.0795 max mem: 3536 +Test: [1150/2387] eta: 0:04:08 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0965 (0.4484) acc1: 99.7396 (82.0860) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1776 data: 0.0829 max mem: 3536 +Test: [1160/2387] eta: 0:04:06 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1419 (0.4458) acc1: 98.6979 (82.2282) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1776 data: 0.0836 max mem: 3536 +Test: [1170/2387] eta: 0:04:04 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1460 (0.4432) acc1: 98.6979 (82.3675) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1842 data: 0.0912 max mem: 3536 +Test: [1180/2387] eta: 0:04:02 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1501 (0.4409) acc1: 98.6979 (82.4897) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1834 data: 0.0919 max mem: 3536 +Test: [1190/2387] eta: 0:04:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1491 (0.4385) acc1: 97.1354 (82.6074) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1885 data: 0.0957 max mem: 3536 +Test: [1200/2387] eta: 0:03:58 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1213 (0.4357) acc1: 98.4375 (82.7485) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2011 data: 0.1082 max mem: 3536 +Test: [1210/2387] eta: 0:03:55 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0854 (0.4328) acc1: 100.0000 (82.8903) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1893 data: 0.0956 max mem: 3536 +Test: [1220/2387] eta: 0:03:53 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0792 (0.4298) acc1: 100.0000 (83.0300) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1746 data: 0.0805 max mem: 3536 +Test: [1230/2387] eta: 0:03:51 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0630 (0.4268) acc1: 100.0000 (83.1677) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1607 data: 0.0692 max mem: 3536 +Test: [1240/2387] eta: 0:03:49 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0647 (0.4239) acc1: 100.0000 (83.3027) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1612 data: 0.0687 max mem: 3536 +Test: [1250/2387] eta: 0:03:46 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0656 (0.4211) acc1: 100.0000 (83.4353) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1744 data: 0.0807 max mem: 3536 +Test: [1260/2387] eta: 0:03:44 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0650 (0.4183) acc1: 100.0000 (83.5644) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1596 data: 0.0680 max mem: 3536 +Test: [1270/2387] eta: 0:03:42 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0613 (0.4155) acc1: 100.0000 (83.6933) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1612 data: 0.0694 max mem: 3536 +Test: [1280/2387] eta: 0:03:39 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0618 (0.4128) acc1: 100.0000 (83.8192) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1745 data: 0.0809 max mem: 3536 +Test: [1290/2387] eta: 0:03:37 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0675 (0.4102) acc1: 100.0000 (83.9371) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1786 data: 0.0856 max mem: 3536 +Test: [1300/2387] eta: 0:03:35 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0671 (0.4076) acc1: 100.0000 (84.0575) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1744 data: 0.0830 max mem: 3536 +Test: [1310/2387] eta: 0:03:33 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0670 (0.4051) acc1: 100.0000 (84.1752) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1747 data: 0.0808 max mem: 3536 +Test: [1320/2387] eta: 0:03:31 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0708 (0.4026) acc1: 100.0000 (84.2894) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1792 data: 0.0830 max mem: 3536 +Test: [1330/2387] eta: 0:03:29 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0719 (0.4002) acc1: 100.0000 (84.4057) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1714 data: 0.0762 max mem: 3536 +Test: [1340/2387] eta: 0:03:26 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0786 (0.3991) acc1: 100.0000 (84.4467) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1759 data: 0.0831 max mem: 3536 +Test: [1350/2387] eta: 0:03:24 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0615 (0.3966) acc1: 100.0000 (84.5618) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1687 data: 0.0770 max mem: 3537 +Test: [1360/2387] eta: 0:03:22 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0562 (0.3941) acc1: 100.0000 (84.6752) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1807 data: 0.0886 max mem: 3537 +Test: [1370/2387] eta: 0:03:20 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0563 (0.3916) acc1: 100.0000 (84.7870) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1759 data: 0.0840 max mem: 3537 +Test: [1380/2387] eta: 0:03:18 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0563 (0.3892) acc1: 100.0000 (84.8972) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1525 data: 0.0608 max mem: 3537 +Test: [1390/2387] eta: 0:03:15 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0558 (0.3868) acc1: 100.0000 (85.0057) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1586 data: 0.0667 max mem: 3537 +Test: [1400/2387] eta: 0:03:13 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0548 (0.3844) acc1: 100.0000 (85.1128) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1569 data: 0.0650 max mem: 3537 +Test: [1410/2387] eta: 0:03:11 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0555 (0.3823) acc1: 100.0000 (85.2140) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1659 data: 0.0745 max mem: 3537 +Test: [1420/2387] eta: 0:03:09 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0809 (0.3804) acc1: 100.0000 (85.3058) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1718 data: 0.0787 max mem: 3537 +Test: [1430/2387] eta: 0:03:07 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0867 (0.3783) acc1: 100.0000 (85.4061) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1671 data: 0.0734 max mem: 3537 +Test: [1440/2387] eta: 0:03:05 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0814 (0.3763) acc1: 100.0000 (85.5065) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1700 data: 0.0785 max mem: 3537 +Test: [1450/2387] eta: 0:03:02 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0849 (0.3744) acc1: 100.0000 (85.6022) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1687 data: 0.0778 max mem: 3537 +Test: [1460/2387] eta: 0:03:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0824 (0.3724) acc1: 99.7396 (85.6983) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1651 data: 0.0742 max mem: 3537 +Test: [1470/2387] eta: 0:02:58 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0723 (0.3703) acc1: 100.0000 (85.7950) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1693 data: 0.0778 max mem: 3537 +Test: [1480/2387] eta: 0:02:56 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0703 (0.3684) acc1: 100.0000 (85.8872) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1765 data: 0.0848 max mem: 3537 +Test: [1490/2387] eta: 0:02:54 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0737 (0.3664) acc1: 99.7396 (85.9799) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1854 data: 0.0936 max mem: 3537 +Test: [1500/2387] eta: 0:02:52 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0805 (0.3645) acc1: 99.7396 (86.0714) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1890 data: 0.0977 max mem: 3537 +Test: [1510/2387] eta: 0:02:50 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0786 (0.3626) acc1: 100.0000 (86.1634) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1941 data: 0.1029 max mem: 3537 +Test: [1520/2387] eta: 0:02:48 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0864 (0.3612) acc1: 99.7396 (86.2325) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1788 data: 0.0868 max mem: 3537 +Test: [1530/2387] eta: 0:02:46 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1406 (0.3598) acc1: 97.9167 (86.3102) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1667 data: 0.0732 max mem: 3537 +Test: [1540/2387] eta: 0:02:44 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1402 (0.3583) acc1: 98.1771 (86.3884) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1799 data: 0.0853 max mem: 3537 +Test: [1550/2387] eta: 0:02:42 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1286 (0.3569) acc1: 98.9583 (86.4635) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1799 data: 0.0867 max mem: 3537 +Test: [1560/2387] eta: 0:02:40 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1306 (0.3554) acc1: 98.6979 (86.5416) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1749 data: 0.0836 max mem: 3537 +Test: [1570/2387] eta: 0:02:38 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1062 (0.3537) acc1: 99.7396 (86.6256) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1687 data: 0.0773 max mem: 3537 +Test: [1580/2387] eta: 0:02:36 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0847 (0.3521) acc1: 100.0000 (86.7079) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1709 data: 0.0782 max mem: 3538 +Test: [1590/2387] eta: 0:02:34 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0800 (0.3503) acc1: 100.0000 (86.7909) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1772 data: 0.0829 max mem: 3538 +Test: [1600/2387] eta: 0:02:32 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0742 (0.3488) acc1: 100.0000 (86.8612) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1797 data: 0.0852 max mem: 3538 +Test: [1610/2387] eta: 0:02:30 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0714 (0.3471) acc1: 100.0000 (86.9423) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1856 data: 0.0920 max mem: 3538 +Test: [1620/2387] eta: 0:02:28 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0763 (0.3455) acc1: 99.7396 (87.0200) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2021 data: 0.1076 max mem: 3538 +Test: [1630/2387] eta: 0:02:26 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0864 (0.3440) acc1: 99.7396 (87.0954) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2004 data: 0.1048 max mem: 3538 +Test: [1640/2387] eta: 0:02:24 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0864 (0.3424) acc1: 99.7396 (87.1706) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1620 data: 0.0683 max mem: 3538 +Test: [1650/2387] eta: 0:02:22 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0857 (0.3409) acc1: 99.7396 (87.2445) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1633 data: 0.0702 max mem: 3538 +Test: [1660/2387] eta: 0:02:20 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0858 (0.3395) acc1: 99.4792 (87.3161) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1808 data: 0.0869 max mem: 3538 +Test: [1670/2387] eta: 0:02:18 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0961 (0.3380) acc1: 99.7396 (87.3872) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1992 data: 0.1071 max mem: 3538 +Test: [1680/2387] eta: 0:02:16 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0845 (0.3366) acc1: 99.7396 (87.4537) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2031 data: 0.1101 max mem: 3538 +Test: [1690/2387] eta: 0:02:14 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0901 (0.3353) acc1: 99.7396 (87.5196) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2049 data: 0.1066 max mem: 3538 +Test: [1700/2387] eta: 0:02:12 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0865 (0.3338) acc1: 99.7396 (87.5917) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1948 data: 0.0978 max mem: 3538 +Test: [1710/2387] eta: 0:02:10 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0757 (0.3322) acc1: 100.0000 (87.6630) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1859 data: 0.0911 max mem: 3538 +Test: [1720/2387] eta: 0:02:08 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0678 (0.3307) acc1: 100.0000 (87.7333) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2000 data: 0.1063 max mem: 3538 +Test: [1730/2387] eta: 0:02:07 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0658 (0.3292) acc1: 100.0000 (87.8024) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1947 data: 0.1009 max mem: 3538 +Test: [1740/2387] eta: 0:02:05 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0654 (0.3277) acc1: 100.0000 (87.8701) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2015 data: 0.1078 max mem: 3538 +Test: [1750/2387] eta: 0:02:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0654 (0.3263) acc1: 100.0000 (87.9381) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1872 data: 0.0956 max mem: 3538 +Test: [1760/2387] eta: 0:02:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0654 (0.3248) acc1: 100.0000 (88.0065) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1672 data: 0.0756 max mem: 3538 +Test: [1770/2387] eta: 0:01:59 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0698 (0.3234) acc1: 100.0000 (88.0717) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1532 data: 0.0601 max mem: 3538 +Test: [1780/2387] eta: 0:01:56 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0701 (0.3219) acc1: 100.0000 (88.1382) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1414 data: 0.0476 max mem: 3538 +Test: [1790/2387] eta: 0:01:54 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0647 (0.3205) acc1: 100.0000 (88.2032) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1485 data: 0.0565 max mem: 3538 +Test: [1800/2387] eta: 0:01:52 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0661 (0.3191) acc1: 100.0000 (88.2669) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1681 data: 0.0739 max mem: 3538 +Test: [1810/2387] eta: 0:01:50 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0669 (0.3177) acc1: 100.0000 (88.3310) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1649 data: 0.0674 max mem: 3539 +Test: [1820/2387] eta: 0:01:48 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0621 (0.3163) acc1: 100.0000 (88.3931) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1453 data: 0.0486 max mem: 3539 +Test: [1830/2387] eta: 0:01:46 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0629 (0.3150) acc1: 100.0000 (88.4563) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1449 data: 0.0493 max mem: 3539 +Test: [1840/2387] eta: 0:01:44 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0819 (0.3137) acc1: 100.0000 (88.5190) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1695 data: 0.0727 max mem: 3539 +Test: [1850/2387] eta: 0:01:42 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0828 (0.3125) acc1: 100.0000 (88.5811) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1821 data: 0.0860 max mem: 3539 +Test: [1860/2387] eta: 0:01:40 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0831 (0.3113) acc1: 100.0000 (88.6424) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1467 data: 0.0537 max mem: 3539 +Test: [1870/2387] eta: 0:01:38 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0833 (0.3100) acc1: 100.0000 (88.7030) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1348 data: 0.0415 max mem: 3539 +Test: [1880/2387] eta: 0:01:36 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0824 (0.3088) acc1: 100.0000 (88.7630) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1555 data: 0.0619 max mem: 3539 +Test: [1890/2387] eta: 0:01:34 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0824 (0.3076) acc1: 100.0000 (88.8225) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1679 data: 0.0763 max mem: 3539 +Test: [1900/2387] eta: 0:01:32 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0829 (0.3065) acc1: 100.0000 (88.8813) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1775 data: 0.0851 max mem: 3539 +Test: [1910/2387] eta: 0:01:30 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0844 (0.3053) acc1: 100.0000 (88.9394) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1649 data: 0.0699 max mem: 3539 +Test: [1920/2387] eta: 0:01:28 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0893 (0.3041) acc1: 100.0000 (88.9970) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1473 data: 0.0518 max mem: 3539 +Test: [1930/2387] eta: 0:01:26 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0708 (0.3029) acc1: 100.0000 (89.0540) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1622 data: 0.0689 max mem: 3539 +Test: [1940/2387] eta: 0:01:24 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0669 (0.3019) acc1: 100.0000 (89.1072) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1725 data: 0.0810 max mem: 3539 +Test: [1950/2387] eta: 0:01:22 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0686 (0.3006) acc1: 100.0000 (89.1630) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1626 data: 0.0673 max mem: 3539 +Test: [1960/2387] eta: 0:01:21 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0687 (0.2996) acc1: 100.0000 (89.2130) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1670 data: 0.0707 max mem: 3539 +Test: [1970/2387] eta: 0:01:19 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0916 (0.2987) acc1: 100.0000 (89.2615) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1754 data: 0.0831 max mem: 3539 +Test: [1980/2387] eta: 0:01:17 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1074 (0.2977) acc1: 99.7396 (89.3103) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1756 data: 0.0841 max mem: 3539 +Test: [1990/2387] eta: 0:01:15 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1108 (0.2968) acc1: 99.7396 (89.3623) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1743 data: 0.0828 max mem: 3539 +Test: [2000/2387] eta: 0:01:13 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1127 (0.2959) acc1: 99.7396 (89.4131) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1742 data: 0.0816 max mem: 3539 +Test: [2010/2387] eta: 0:01:11 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1133 (0.2950) acc1: 99.4792 (89.4638) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1749 data: 0.0828 max mem: 3539 +Test: [2020/2387] eta: 0:01:09 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1148 (0.2941) acc1: 99.4792 (89.5140) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1651 data: 0.0745 max mem: 3539 +Test: [2030/2387] eta: 0:01:07 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1142 (0.2932) acc1: 99.7396 (89.5637) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1680 data: 0.0768 max mem: 3540 +Test: [2040/2387] eta: 0:01:05 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1121 (0.2923) acc1: 99.7396 (89.6132) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1982 data: 0.1065 max mem: 3540 +Test: [2050/2387] eta: 0:01:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1111 (0.2914) acc1: 99.7396 (89.6629) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2239 data: 0.1326 max mem: 3540 +Test: [2060/2387] eta: 0:01:02 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0968 (0.2905) acc1: 100.0000 (89.7128) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2416 data: 0.1498 max mem: 3540 +Test: [2070/2387] eta: 0:01:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0931 (0.2895) acc1: 100.0000 (89.7625) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2394 data: 0.1476 max mem: 3540 +Test: [2080/2387] eta: 0:00:58 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0931 (0.2886) acc1: 100.0000 (89.8112) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2028 data: 0.1121 max mem: 3540 +Test: [2090/2387] eta: 0:00:56 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0930 (0.2876) acc1: 100.0000 (89.8599) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1717 data: 0.0815 max mem: 3540 +Test: [2100/2387] eta: 0:00:54 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0929 (0.2867) acc1: 100.0000 (89.9081) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1846 data: 0.0940 max mem: 3540 +Test: [2110/2387] eta: 0:00:52 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0935 (0.2858) acc1: 100.0000 (89.9559) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2064 data: 0.1134 max mem: 3540 +Test: [2120/2387] eta: 0:00:50 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0922 (0.2849) acc1: 100.0000 (90.0024) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1931 data: 0.0997 max mem: 3540 +Test: [2130/2387] eta: 0:00:48 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0770 (0.2839) acc1: 100.0000 (90.0493) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1868 data: 0.0929 max mem: 3540 +Test: [2140/2387] eta: 0:00:46 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0775 (0.2831) acc1: 100.0000 (90.0837) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2268 data: 0.1327 max mem: 3540 +Test: [2150/2387] eta: 0:00:45 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0760 (0.2821) acc1: 100.0000 (90.1297) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2180 data: 0.1228 max mem: 3540 +Test: [2160/2387] eta: 0:00:43 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0800 (0.2812) acc1: 100.0000 (90.1743) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1810 data: 0.0852 max mem: 3540 +Test: [2170/2387] eta: 0:00:41 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0859 (0.2804) acc1: 100.0000 (90.2182) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1869 data: 0.0948 max mem: 3540 +Test: [2180/2387] eta: 0:00:39 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0994 (0.2814) acc1: 99.4792 (90.1397) acc5: 100.0000 (100.0000) acc1_neg: 34.896 acc5_neg: 100.000 time: 0.1987 data: 0.1056 max mem: 3540 +Test: [2190/2387] eta: 0:00:37 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2601 (0.2819) acc1: 88.8021 (90.1060) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2035 data: 0.1099 max mem: 3540 +Test: [2200/2387] eta: 0:00:35 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0949 (0.2809) acc1: 100.0000 (90.1509) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2199 data: 0.1284 max mem: 3540 +Test: [2210/2387] eta: 0:00:33 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0695 (0.2800) acc1: 100.0000 (90.1953) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2068 data: 0.1143 max mem: 3540 +Test: [2220/2387] eta: 0:00:31 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0698 (0.2791) acc1: 100.0000 (90.2371) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0819 max mem: 3540 +Test: [2230/2387] eta: 0:00:29 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0698 (0.2781) acc1: 100.0000 (90.2803) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1774 data: 0.0821 max mem: 3540 +Test: [2240/2387] eta: 0:00:27 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0704 (0.2772) acc1: 100.0000 (90.3223) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1953 data: 0.1003 max mem: 3540 +Test: [2250/2387] eta: 0:00:26 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0727 (0.2764) acc1: 100.0000 (90.3640) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1978 data: 0.1060 max mem: 3540 +Test: [2260/2387] eta: 0:00:24 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0769 (0.2755) acc1: 100.0000 (90.4066) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1950 data: 0.1016 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0723 (0.2746) acc1: 100.0000 (90.4489) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2220 data: 0.1291 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0680 (0.2739) acc1: 100.0000 (90.4782) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.1027 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0683 (0.2730) acc1: 100.0000 (90.5197) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1578 data: 0.0655 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0693 (0.2721) acc1: 100.0000 (90.5602) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1620 data: 0.0709 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0792 (0.2713) acc1: 100.0000 (90.6003) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1569 data: 0.0652 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0820 (0.2706) acc1: 100.0000 (90.6387) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1510 data: 0.0592 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0873 (0.2698) acc1: 100.0000 (90.6787) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1509 data: 0.0594 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0749 (0.2689) acc1: 100.0000 (90.7184) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1452 data: 0.0542 max mem: 3541 +Test: [2350/2387] eta: 0:00:07 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0730 (0.2683) acc1: 100.0000 (90.7471) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1366 data: 0.0466 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0732 (0.2675) acc1: 100.0000 (90.7860) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1370 data: 0.0470 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0870 (0.2668) acc1: 100.0000 (90.8225) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1371 data: 0.0472 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1017 (0.2662) acc1: 100.0000 (90.8592) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1305 data: 0.0402 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1151 (0.2662) acc1: 99.2188 (90.8603) acc5: 100.0000 (100.0000) acc1_neg: 83.721 acc5_neg: 100.000 time: 0.1197 data: 0.0320 max mem: 3541 +Test: Total time: 0:07:30 (0.1886 s / it) +* Overall Acc@1 90.860 Acc@5 100.000 + target == real Acc@1 2.912 Acc@5 100.000 + target == fake Acc@1 97.036 Acc@5 100.000 + loss 0.266 + PR-AUC 0.9735 ROC-AUC 0.7162 EER 33.45% (thr=0.9185) +Accuracy of the network on the 916267 test images: 90.9% +[rank0]:[W726 05:19:43.757448447 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:20:03.863000 1188931 torch/distributed/run.py:766] +W0726 05:20:03.863000 1188931 torch/distributed/run.py:766] ***************************************** +W0726 05:20:03.863000 1188931 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:20:03.863000 1188931 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 5): env:// +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:20:32.895614071 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:20:33.922667214 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:20:33.084421286 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:20:33.255691220 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:20:33.497264805 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:20:33.534350484 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +[rank0]:[W726 05:20:33.556631317 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:20:33.558078979 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n11_seed8_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n11_classseed8_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:09:47 acc1_pos: 1.302 acc5_pos: 100.000 loss: 2.0701 (2.0701) acc1: 1.3021 (1.3021) acc5: 100.0000 (100.0000) time: 8.0989 data: 5.8764 max mem: 3531 +Test: [ 10/517] eta: 0:06:57 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.2080 (2.1874) acc1: 0.5208 (2.0360) acc5: 100.0000 (100.0000) time: 0.8231 data: 0.5349 max mem: 3531 +Test: [ 20/517] eta: 0:03:57 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.2999 (2.2455) acc1: 0.2604 (1.4137) acc5: 100.0000 (100.0000) time: 0.0977 data: 0.0006 max mem: 3531 +Test: [ 30/517] eta: 0:03:31 acc1_pos: 3.646 acc5_pos: 100.000 loss: 2.3014 (2.2621) acc1: 0.2604 (1.2601) acc5: 100.0000 (100.0000) time: 0.2199 data: 0.1217 max mem: 3531 +Test: [ 40/517] eta: 0:03:17 acc1_pos: 4.688 acc5_pos: 100.000 loss: 2.2359 (2.2532) acc1: 0.5208 (1.2957) acc5: 100.0000 (100.0000) time: 0.3449 data: 0.2498 max mem: 3531 +Test: [ 50/517] eta: 0:03:05 acc1_pos: 19.266 acc5_pos: 100.000 loss: 2.1597 (2.1119) acc1: 1.8229 (6.7147) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.3409 data: 0.2487 max mem: 3531 +Test: [ 60/517] eta: 0:02:51 acc1_pos: 19.266 acc5_pos: 100.000 loss: 0.1526 (1.7864) acc1: 99.4792 (21.9775) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2949 data: 0.2027 max mem: 3531 +Test: [ 70/517] eta: 0:02:41 acc1_pos: 19.266 acc5_pos: 100.000 loss: 0.1080 (1.5496) acc1: 100.0000 (32.9629) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2689 data: 0.1745 max mem: 3531 +Test: [ 80/517] eta: 0:02:30 acc1_pos: 19.266 acc5_pos: 100.000 loss: 0.1042 (1.3715) acc1: 100.0000 (41.2391) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2499 data: 0.1541 max mem: 3531 +Test: [ 90/517] eta: 0:02:22 acc1_pos: 19.266 acc5_pos: 100.000 loss: 0.0913 (1.2301) acc1: 100.0000 (47.6963) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2402 data: 0.1447 max mem: 3531 +Test: [100/517] eta: 0:02:16 acc1_pos: 19.266 acc5_pos: 100.000 loss: 0.0885 (1.1175) acc1: 100.0000 (52.8749) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2650 data: 0.1705 max mem: 3531 +Test: [110/517] eta: 0:02:10 acc1_pos: 19.266 acc5_pos: 100.000 loss: 0.0928 (1.0255) acc1: 100.0000 (57.1204) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2596 data: 0.1640 max mem: 3531 +Test: [120/517] eta: 0:02:05 acc1_pos: 19.266 acc5_pos: 100.000 loss: 0.0937 (0.9486) acc1: 100.0000 (60.6642) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2548 data: 0.1569 max mem: 3531 +Test: [130/517] eta: 0:02:00 acc1_pos: 19.266 acc5_pos: 100.000 loss: 0.0894 (0.8831) acc1: 100.0000 (63.6669) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2599 data: 0.1616 max mem: 3531 +Test: [140/517] eta: 0:01:56 acc1_pos: 19.266 acc5_pos: 100.000 loss: 0.0864 (0.8264) acc1: 100.0000 (66.2437) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2599 data: 0.1621 max mem: 3531 +Test: [150/517] eta: 0:01:51 acc1_pos: 19.266 acc5_pos: 100.000 loss: 0.0832 (0.7774) acc1: 100.0000 (68.4792) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2499 data: 0.1535 max mem: 3531 +Test: [160/517] eta: 0:01:47 acc1_pos: 19.266 acc5_pos: 100.000 loss: 0.0845 (0.7345) acc1: 100.0000 (70.4370) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1584 max mem: 3531 +Test: [170/517] eta: 0:01:43 acc1_pos: 19.266 acc5_pos: 100.000 loss: 0.0845 (0.6965) acc1: 100.0000 (72.1659) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2603 data: 0.1641 max mem: 3531 +Test: [180/517] eta: 0:01:40 acc1_pos: 19.266 acc5_pos: 100.000 loss: 0.0821 (0.6623) acc1: 100.0000 (73.7037) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2599 data: 0.1643 max mem: 3531 +Test: [190/517] eta: 0:01:36 acc1_pos: 19.266 acc5_pos: 100.000 loss: 0.0830 (0.6324) acc1: 100.0000 (75.0804) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2596 data: 0.1630 max mem: 3531 +Test: [200/517] eta: 0:01:32 acc1_pos: 19.266 acc5_pos: 100.000 loss: 0.0879 (0.6051) acc1: 100.0000 (76.3189) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2499 data: 0.1532 max mem: 3531 +Test: [210/517] eta: 0:01:29 acc1_pos: 19.266 acc5_pos: 100.000 loss: 0.0912 (0.5808) acc1: 100.0000 (77.4413) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1568 max mem: 3532 +Test: [220/517] eta: 0:01:25 acc1_pos: 19.266 acc5_pos: 100.000 loss: 0.0920 (0.5587) acc1: 100.0000 (78.4620) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1568 max mem: 3532 +Test: [230/517] eta: 0:01:22 acc1_pos: 19.266 acc5_pos: 100.000 loss: 0.0928 (0.5387) acc1: 100.0000 (79.3944) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2548 data: 0.1566 max mem: 3532 +Test: [240/517] eta: 0:01:19 acc1_pos: 19.266 acc5_pos: 100.000 loss: 0.0960 (0.5206) acc1: 100.0000 (80.2494) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2598 data: 0.1586 max mem: 3532 +Test: [250/517] eta: 0:01:16 acc1_pos: 19.266 acc5_pos: 100.000 loss: 0.0958 (0.5036) acc1: 100.0000 (81.0363) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1551 max mem: 3532 +Test: [260/517] eta: 0:01:12 acc1_pos: 19.266 acc5_pos: 100.000 loss: 0.0869 (0.4877) acc1: 100.0000 (81.7629) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2499 data: 0.1538 max mem: 3532 +Test: [270/517] eta: 0:01:09 acc1_pos: 19.266 acc5_pos: 100.000 loss: 0.0943 (0.4735) acc1: 100.0000 (82.4358) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1594 max mem: 3532 +Test: [280/517] eta: 0:01:06 acc1_pos: 19.266 acc5_pos: 100.000 loss: 0.0983 (0.4599) acc1: 100.0000 (83.0609) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2649 data: 0.1683 max mem: 3532 +Test: [290/517] eta: 0:01:04 acc1_pos: 19.266 acc5_pos: 100.000 loss: 0.0888 (0.4471) acc1: 100.0000 (83.6430) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2799 data: 0.1829 max mem: 3532 +Test: [300/517] eta: 0:01:01 acc1_pos: 19.266 acc5_pos: 100.000 loss: 0.0834 (0.4350) acc1: 100.0000 (84.1864) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2649 data: 0.1680 max mem: 3532 +Test: [310/517] eta: 0:00:58 acc1_pos: 19.266 acc5_pos: 100.000 loss: 0.0839 (0.4238) acc1: 100.0000 (84.6949) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2499 data: 0.1536 max mem: 3532 +Test: [320/517] eta: 0:00:55 acc1_pos: 19.266 acc5_pos: 100.000 loss: 0.0846 (0.4132) acc1: 100.0000 (85.1717) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2599 data: 0.1641 max mem: 3532 +Test: [330/517] eta: 0:00:52 acc1_pos: 19.266 acc5_pos: 100.000 loss: 0.0830 (0.4032) acc1: 100.0000 (85.6197) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2699 data: 0.1730 max mem: 3532 +Test: [340/517] eta: 0:00:49 acc1_pos: 19.266 acc5_pos: 100.000 loss: 0.0801 (0.3938) acc1: 100.0000 (86.0414) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2499 data: 0.1539 max mem: 3532 +Test: [350/517] eta: 0:00:46 acc1_pos: 19.266 acc5_pos: 100.000 loss: 0.0786 (0.3850) acc1: 100.0000 (86.4390) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2349 data: 0.1391 max mem: 3532 +Test: [360/517] eta: 0:00:43 acc1_pos: 19.266 acc5_pos: 100.000 loss: 0.0818 (0.3769) acc1: 100.0000 (86.8147) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1582 max mem: 3532 +Test: [370/517] eta: 0:00:40 acc1_pos: 19.266 acc5_pos: 100.000 loss: 0.0948 (0.3694) acc1: 100.0000 (87.1680) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2598 data: 0.1622 max mem: 3532 +Test: [380/517] eta: 0:00:37 acc1_pos: 19.266 acc5_pos: 100.000 loss: 0.0945 (0.3621) acc1: 100.0000 (87.5041) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2601 data: 0.1634 max mem: 3532 +Test: [390/517] eta: 0:00:34 acc1_pos: 19.266 acc5_pos: 100.000 loss: 0.1019 (0.3561) acc1: 100.0000 (87.7904) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1591 max mem: 3532 +Test: [400/517] eta: 0:00:32 acc1_pos: 19.266 acc5_pos: 100.000 loss: 0.1128 (0.3499) acc1: 100.0000 (88.0949) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2546 data: 0.1578 max mem: 3532 +Test: [410/517] eta: 0:00:29 acc1_pos: 19.266 acc5_pos: 100.000 loss: 0.1102 (0.3441) acc1: 100.0000 (88.3845) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1587 max mem: 3532 +Test: [420/517] eta: 0:00:26 acc1_pos: 19.266 acc5_pos: 100.000 loss: 0.1050 (0.3383) acc1: 100.0000 (88.6592) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2599 data: 0.1632 max mem: 3532 +Test: [430/517] eta: 0:00:23 acc1_pos: 19.266 acc5_pos: 100.000 loss: 0.1011 (0.3331) acc1: 100.0000 (88.9217) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2603 data: 0.1631 max mem: 3532 +Test: [440/517] eta: 0:00:21 acc1_pos: 19.266 acc5_pos: 100.000 loss: 0.1051 (0.3278) acc1: 100.0000 (89.1723) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2491 data: 0.1551 max mem: 3533 +Test: [450/517] eta: 0:00:18 acc1_pos: 19.266 acc5_pos: 100.000 loss: 0.0950 (0.3227) acc1: 100.0000 (89.4124) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2533 data: 0.1593 max mem: 3533 +Test: [460/517] eta: 0:00:15 acc1_pos: 19.266 acc5_pos: 100.000 loss: 0.1103 (0.3182) acc1: 100.0000 (89.6381) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2545 data: 0.1586 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 19.266 acc5_pos: 100.000 loss: 0.0980 (0.3134) acc1: 100.0000 (89.8570) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2195 data: 0.1244 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 19.266 acc5_pos: 100.000 loss: 0.0934 (0.3092) acc1: 100.0000 (90.0668) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1860 data: 0.0936 max mem: 3533 +Test: [490/517] eta: 0:00:07 acc1_pos: 19.266 acc5_pos: 100.000 loss: 0.1092 (0.3050) acc1: 100.0000 (90.2691) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1801 data: 0.0900 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 19.266 acc5_pos: 100.000 loss: 0.0864 (0.3006) acc1: 100.0000 (90.4633) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1797 data: 0.0896 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 19.266 acc5_pos: 100.000 loss: 0.0864 (0.2969) acc1: 100.0000 (90.6490) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1788 data: 0.0890 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 19.266 acc5_pos: 100.000 loss: 0.0903 (0.2945) acc1: 100.0000 (90.7528) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1530 data: 0.0616 max mem: 3533 +Test: Total time: 0:02:15 (0.2627 s / it) +* Overall Acc@1 90.753 Acc@5 100.000 + target == real Acc@1 2.043 Acc@5 100.000 + target == fake Acc@1 99.956 Acc@5 100.000 + loss 0.295 + PR-AUC 0.9337 ROC-AUC 0.6437 EER 39.85% (thr=0.9121) +Accuracy of the network on the 198427 test images: 90.8% +[rank0]:[W726 05:22:57.407284998 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 05:23:17.441000 1189503 torch/distributed/run.py:766] +W0726 05:23:17.441000 1189503 torch/distributed/run.py:766] ***************************************** +W0726 05:23:17.441000 1189503 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:23:17.441000 1189503 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:23:46.319364768 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:23:46.817258377 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:23:47.968593967 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:23:47.186069578 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:23:47.350521858 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:23:47.568822846 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:23:47.734860653 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:23:47.807049724 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n11_seed8_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n11_classseed8_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:32:01 acc1_pos: 15.104 acc5_pos: 100.000 loss: 1.1738 (1.1738) acc1: 15.1042 (15.1042) acc5: 100.0000 (100.0000) time: 4.9389 data: 2.9810 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/389] eta: 0:03:21 acc1_pos: 23.177 acc5_pos: 100.000 loss: 1.2829 (1.2724) acc1: 23.1771 (22.3011) acc5: 100.0000 (100.0000) time: 0.5309 data: 0.2713 max mem: 3531 +Test: [ 20/389] eta: 0:01:58 acc1_pos: 0.260 acc5_pos: 100.000 loss: 1.4010 (1.5678) acc1: 7.0312 (13.8517) acc5: 100.0000 (100.0000) time: 0.0915 data: 0.0017 max mem: 3531 +Test: [ 30/389] eta: 0:01:29 acc1_pos: 1.042 acc5_pos: 100.000 loss: 1.7946 (1.6439) acc1: 0.0000 (9.4254) acc5: 100.0000 (100.0000) time: 0.0952 data: 0.0017 max mem: 3531 +Test: [ 40/389] eta: 0:01:21 acc1_pos: 2.865 acc5_pos: 100.000 loss: 1.7880 (1.6479) acc1: 0.0000 (8.8097) acc5: 100.0000 (100.0000) time: 0.1386 data: 0.0426 max mem: 3531 +Test: [ 50/389] eta: 0:01:13 acc1_pos: 0.260 acc5_pos: 100.000 loss: 1.7126 (1.6813) acc1: 0.5208 (7.3427) acc5: 100.0000 (100.0000) time: 0.1649 data: 0.0707 max mem: 3531 +Test: [ 60/389] eta: 0:01:07 acc1_pos: 9.635 acc5_pos: 100.000 loss: 1.8667 (1.6735) acc1: 0.2604 (7.3514) acc5: 100.0000 (100.0000) time: 0.1500 data: 0.0583 max mem: 3531 +Test: [ 70/389] eta: 0:01:04 acc1_pos: 38.281 acc5_pos: 100.000 loss: 0.9420 (1.5479) acc1: 20.8333 (13.3583) acc5: 100.0000 (100.0000) time: 0.1606 data: 0.0705 max mem: 3531 +Test: [ 80/389] eta: 0:01:01 acc1_pos: 23.698 acc5_pos: 100.000 loss: 0.9161 (1.5251) acc1: 38.2812 (14.6508) acc5: 100.0000 (100.0000) time: 0.1749 data: 0.0839 max mem: 3531 +Test: [ 90/389] eta: 0:00:59 acc1_pos: 35.156 acc5_pos: 100.000 loss: 0.9902 (1.4656) acc1: 24.7396 (16.6953) acc5: 100.0000 (100.0000) time: 0.1904 data: 0.0985 max mem: 3531 +Test: [100/389] eta: 0:00:57 acc1_pos: 65.625 acc5_pos: 100.000 loss: 1.0439 (1.4416) acc1: 24.7396 (17.4943) acc5: 100.0000 (100.0000) time: 0.1999 data: 0.1038 max mem: 3531 +Test: [110/389] eta: 0:00:54 acc1_pos: 22.135 acc5_pos: 100.000 loss: 1.0927 (1.4206) acc1: 22.6562 (19.2051) acc5: 100.0000 (100.0000) time: 0.1887 data: 0.0923 max mem: 3531 +Test: [120/389] eta: 0:00:51 acc1_pos: 0.781 acc5_pos: 100.000 loss: 0.9868 (1.3995) acc1: 32.8125 (19.7594) acc5: 100.0000 (100.0000) time: 0.1654 data: 0.0724 max mem: 3531 +Test: [130/389] eta: 0:00:50 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.7855 (1.3391) acc1: 48.1771 (23.6959) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1861 data: 0.0944 max mem: 3531 +Test: [140/389] eta: 0:00:48 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2545 (1.2595) acc1: 96.6146 (28.8989) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2093 data: 0.1170 max mem: 3531 +Test: [150/389] eta: 0:00:46 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2205 (1.2000) acc1: 99.2188 (32.5383) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1764 data: 0.0816 max mem: 3531 +Test: [160/389] eta: 0:00:44 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1793 (1.1415) acc1: 96.6146 (36.2497) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1859 data: 0.0927 max mem: 3531 +Test: [170/389] eta: 0:00:42 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1392 (1.0859) acc1: 99.7396 (39.8102) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2172 data: 0.1217 max mem: 3531 +Test: [180/389] eta: 0:00:40 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1239 (1.0368) acc1: 100.0000 (42.9486) acc5: 100.0000 (100.0000) acc1_neg: 68.229 acc5_neg: 100.000 time: 0.1899 data: 0.0936 max mem: 3531 +Test: [190/389] eta: 0:00:38 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2376 (1.0046) acc1: 97.9167 (44.9866) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.1818 data: 0.0876 max mem: 3531 +Test: [200/389] eta: 0:00:36 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2469 (0.9654) acc1: 97.9167 (47.5617) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2039 data: 0.1106 max mem: 3531 +Test: [210/389] eta: 0:00:34 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1180 (0.9251) acc1: 100.0000 (50.0469) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1954 data: 0.1035 max mem: 3532 +Test: [220/389] eta: 0:00:32 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1010 (0.8902) acc1: 100.0000 (52.2884) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2000 data: 0.1083 max mem: 3532 +Test: [230/389] eta: 0:00:31 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2322 (0.8668) acc1: 99.2188 (53.9085) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.2136 data: 0.1214 max mem: 3532 +Test: [240/389] eta: 0:00:29 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2845 (0.8449) acc1: 97.6562 (55.3769) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.2002 data: 0.1065 max mem: 3532 +Test: [250/389] eta: 0:00:27 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2354 (0.8235) acc1: 97.6562 (56.8372) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1988 data: 0.1050 max mem: 3532 +Test: [260/389] eta: 0:00:25 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1873 (0.8014) acc1: 99.4792 (58.2385) acc5: 100.0000 (100.0000) acc1_neg: 76.302 acc5_neg: 100.000 time: 0.2153 data: 0.1225 max mem: 3532 +Test: [270/389] eta: 0:00:23 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2251 (0.7868) acc1: 98.1771 (59.2366) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2163 data: 0.1242 max mem: 3532 +Test: [280/389] eta: 0:00:21 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.3109 (0.7710) acc1: 91.6667 (60.2545) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1928 data: 0.1005 max mem: 3532 +Test: [290/389] eta: 0:00:19 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2519 (0.7537) acc1: 99.4792 (61.3581) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1765 data: 0.0847 max mem: 3532 +Test: [300/389] eta: 0:00:17 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2143 (0.7360) acc1: 99.7396 (62.6194) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1740 data: 0.0832 max mem: 3532 +Test: [310/389] eta: 0:00:15 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2444 (0.7224) acc1: 98.6979 (63.5819) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1819 data: 0.0886 max mem: 3532 +Test: [320/389] eta: 0:00:13 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2896 (0.7082) acc1: 96.8750 (64.5501) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1871 data: 0.0935 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2984 (0.6959) acc1: 98.4375 (65.5455) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1968 data: 0.1055 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2091 (0.6803) acc1: 99.4792 (66.4834) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1761 data: 0.0850 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1606 (0.6692) acc1: 99.7396 (67.1830) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1614 data: 0.0707 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1981 (0.6560) acc1: 99.2188 (68.0697) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1683 data: 0.0778 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2012 (0.6484) acc1: 98.9583 (68.5008) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1477 data: 0.0575 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.2083 (0.6372) acc1: 98.9583 (69.2080) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1361 data: 0.0462 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1924 (0.6290) acc1: 99.7396 (69.7753) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1320 data: 0.0386 max mem: 3532 +Test: Total time: 0:01:13 (0.1887 s / it) +* Overall Acc@1 69.775 Acc@5 100.000 + target == real Acc@1 19.399 Acc@5 100.000 + target == fake Acc@1 93.192 Acc@5 100.000 + loss 0.629 + PR-AUC 0.7561 ROC-AUC 0.6356 EER 39.91% (thr=0.8057) +Accuracy of the network on the 149358 test images: 69.8% +[rank0]:[W726 05:25:10.767016891 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:25:29.278000 1190061 torch/distributed/run.py:766] +W0726 05:25:29.278000 1190061 torch/distributed/run.py:766] ***************************************** +W0726 05:25:29.278000 1190061 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:25:29.278000 1190061 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:25:56.652759112 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:25:56.730349139 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:25:57.990652306 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +[rank0]:[W726 05:25:57.130647177 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:25:57.134075454 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +[rank1]:[W726 05:25:57.361681334 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:25:57.364541048 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:25:57.393872977 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n11_seed8_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n11_classseed8_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:38:43 acc1_pos: 0.781 acc5_pos: 100.000 loss: 2.1955 (2.1955) acc1: 0.7812 (0.7812) acc5: 100.0000 (100.0000) time: 6.9776 data: 5.2977 max mem: 3531 +Test: [ 10/333] eta: 0:03:52 acc1_pos: 20.312 acc5_pos: 100.000 loss: 1.9180 (1.8838) acc1: 12.5000 (11.8845) acc5: 100.0000 (100.0000) time: 0.7210 data: 0.4821 max mem: 3531 +Test: [ 20/333] eta: 0:02:13 acc1_pos: 13.021 acc5_pos: 100.000 loss: 1.6604 (1.7450) acc1: 16.6667 (15.3398) acc5: 100.0000 (100.0000) time: 0.0976 data: 0.0005 max mem: 3531 +Test: [ 30/333] eta: 0:02:01 acc1_pos: 16.146 acc5_pos: 100.000 loss: 1.6536 (1.7289) acc1: 16.1458 (15.1210) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1267 max mem: 3531 +Test: [ 40/333] eta: 0:01:46 acc1_pos: 21.875 acc5_pos: 100.000 loss: 1.6663 (1.7169) acc1: 14.8438 (15.3392) acc5: 100.0000 (100.0000) time: 0.2999 data: 0.2026 max mem: 3531 +Test: [ 50/333] eta: 0:01:38 acc1_pos: 16.406 acc5_pos: 100.000 loss: 1.6849 (1.7104) acc1: 14.8438 (15.5127) acc5: 100.0000 (100.0000) time: 0.2649 data: 0.1690 max mem: 3531 +Test: [ 60/333] eta: 0:01:28 acc1_pos: 7.031 acc5_pos: 100.000 loss: 1.7020 (1.7095) acc1: 14.8438 (15.3689) acc5: 100.0000 (100.0000) time: 0.2456 data: 0.1526 max mem: 3531 +Test: [ 70/333] eta: 0:01:21 acc1_pos: 11.458 acc5_pos: 100.000 loss: 1.7111 (1.7022) acc1: 14.8438 (15.5260) acc5: 100.0000 (100.0000) time: 0.2167 data: 0.1242 max mem: 3531 +Test: [ 80/333] eta: 0:01:15 acc1_pos: 16.406 acc5_pos: 100.000 loss: 1.6504 (1.6953) acc1: 16.4062 (15.7247) acc5: 100.0000 (100.0000) time: 0.2192 data: 0.1262 max mem: 3531 +Test: [ 90/333] eta: 0:01:09 acc1_pos: 21.875 acc5_pos: 100.000 loss: 1.5285 (1.6702) acc1: 17.7083 (16.2431) acc5: 100.0000 (100.0000) time: 0.2065 data: 0.1112 max mem: 3531 +Test: [100/333] eta: 0:01:04 acc1_pos: 31.250 acc5_pos: 100.000 loss: 1.4521 (1.6525) acc1: 20.8333 (16.9606) acc5: 100.0000 (100.0000) time: 0.1899 data: 0.0932 max mem: 3531 +Test: [110/333] eta: 0:01:02 acc1_pos: 22.396 acc5_pos: 100.000 loss: 1.4886 (1.6391) acc1: 20.8333 (17.4409) acc5: 100.0000 (100.0000) time: 0.2394 data: 0.1423 max mem: 3531 +Test: [120/333] eta: 0:00:57 acc1_pos: 15.625 acc5_pos: 100.000 loss: 1.4498 (1.6236) acc1: 22.3958 (17.9515) acc5: 100.0000 (100.0000) time: 0.2317 data: 0.1369 max mem: 3531 +Test: [130/333] eta: 0:00:53 acc1_pos: 24.740 acc5_pos: 100.000 loss: 1.4185 (1.6087) acc1: 24.7396 (18.5492) acc5: 100.0000 (100.0000) time: 0.1881 data: 0.0950 max mem: 3531 +Test: [140/333] eta: 0:00:50 acc1_pos: 27.344 acc5_pos: 100.000 loss: 1.4185 (1.5980) acc1: 25.5208 (19.0012) acc5: 100.0000 (100.0000) time: 0.2161 data: 0.1225 max mem: 3531 +Test: [150/333] eta: 0:00:46 acc1_pos: 27.604 acc5_pos: 100.000 loss: 1.4481 (1.5906) acc1: 25.5208 (19.2536) acc5: 100.0000 (100.0000) time: 0.1796 data: 0.0873 max mem: 3531 +Test: [160/333] eta: 0:00:43 acc1_pos: 16.776 acc5_pos: 100.000 loss: 1.4021 (1.5539) acc1: 26.8229 (21.1779) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.1625 data: 0.0689 max mem: 3531 +Test: [170/333] eta: 0:00:40 acc1_pos: 16.776 acc5_pos: 100.000 loss: 0.3774 (1.4804) acc1: 82.2917 (25.1782) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1944 data: 0.0994 max mem: 3531 +Test: [180/333] eta: 0:00:37 acc1_pos: 16.776 acc5_pos: 100.000 loss: 0.2651 (1.4127) acc1: 91.9271 (28.9106) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.2044 data: 0.1101 max mem: 3531 +Test: [190/333] eta: 0:00:34 acc1_pos: 16.776 acc5_pos: 100.000 loss: 0.2256 (1.3485) acc1: 93.4896 (32.4812) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1917 data: 0.0951 max mem: 3531 +Test: [200/333] eta: 0:00:31 acc1_pos: 16.776 acc5_pos: 100.000 loss: 0.2065 (1.2931) acc1: 94.7917 (35.5268) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1902 data: 0.0928 max mem: 3531 +Test: [210/333] eta: 0:00:29 acc1_pos: 16.776 acc5_pos: 100.000 loss: 0.2607 (1.2450) acc1: 91.4062 (38.1430) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.2009 data: 0.1067 max mem: 3532 +Test: [220/333] eta: 0:00:26 acc1_pos: 16.776 acc5_pos: 100.000 loss: 0.2803 (1.2031) acc1: 91.4062 (40.4329) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1874 data: 0.0952 max mem: 3532 +Test: [230/333] eta: 0:00:23 acc1_pos: 16.776 acc5_pos: 100.000 loss: 0.2444 (1.1606) acc1: 93.7500 (42.7591) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1856 data: 0.0945 max mem: 3532 +Test: [240/333] eta: 0:00:21 acc1_pos: 16.776 acc5_pos: 100.000 loss: 0.2332 (1.1224) acc1: 93.7500 (44.8511) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1891 data: 0.0976 max mem: 3532 +Test: [250/333] eta: 0:00:18 acc1_pos: 16.776 acc5_pos: 100.000 loss: 0.2380 (1.0876) acc1: 92.7083 (46.7360) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1667 data: 0.0750 max mem: 3532 +Test: [260/333] eta: 0:00:16 acc1_pos: 16.776 acc5_pos: 100.000 loss: 0.2371 (1.0551) acc1: 92.7083 (48.5233) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1847 data: 0.0930 max mem: 3532 +Test: [270/333] eta: 0:00:14 acc1_pos: 16.776 acc5_pos: 100.000 loss: 0.2328 (1.0246) acc1: 92.9688 (50.1883) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1823 data: 0.0910 max mem: 3532 +Test: [280/333] eta: 0:00:11 acc1_pos: 16.776 acc5_pos: 100.000 loss: 0.2400 (0.9992) acc1: 91.9271 (51.5523) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1603 data: 0.0691 max mem: 3532 +Test: [290/333] eta: 0:00:09 acc1_pos: 16.776 acc5_pos: 100.000 loss: 0.3461 (0.9777) acc1: 85.1562 (52.6588) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.1619 data: 0.0690 max mem: 3532 +Test: [300/333] eta: 0:00:07 acc1_pos: 16.776 acc5_pos: 100.000 loss: 0.3561 (0.9564) acc1: 85.1562 (53.8042) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1179 data: 0.0258 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 16.776 acc5_pos: 100.000 loss: 0.2971 (0.9330) acc1: 89.8438 (55.0878) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.0906 data: 0.0004 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 16.776 acc5_pos: 100.000 loss: 0.2225 (0.9119) acc1: 93.2292 (56.2224) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.0973 data: 0.0069 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 16.776 acc5_pos: 100.000 loss: 0.3882 (0.8973) acc1: 83.3333 (56.9274) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.1050 data: 0.0148 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 16.776 acc5_pos: 100.000 loss: 0.4052 (0.8944) acc1: 80.2083 (57.0622) acc5: 100.0000 (100.0000) acc1_neg: 77.746 acc5_neg: 100.000 time: 0.1091 data: 0.0157 max mem: 3532 +Test: Total time: 0:01:08 (0.2055 s / it) +* Overall Acc@1 57.062 Acc@5 100.000 + target == real Acc@1 19.537 Acc@5 100.000 + target == fake Acc@1 90.466 Acc@5 100.000 + loss 0.894 + PR-AUC 0.5760 ROC-AUC 0.5825 EER 43.96% (thr=0.8457) +Accuracy of the network on the 127843 test images: 57.1% +[rank0]:[W726 05:27:13.710490868 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:27:33.551000 1190620 torch/distributed/run.py:766] +W0726 05:27:33.551000 1190620 torch/distributed/run.py:766] ***************************************** +W0726 05:27:33.551000 1190620 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:27:33.551000 1190620 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:27:59.877635541 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:28:00.914033594 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:28:00.239971421 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:28:00.444868425 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:28:00.566765194 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:28:00.714959171 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:28:00.739886809 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W726 05:28:00.739889809 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n11_seed8_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n11_classseed8_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:25:04 acc1_pos: 93.490 acc5_pos: 100.000 loss: 0.2525 (0.2525) acc1: 93.4896 (93.4896) acc5: 100.0000 (100.0000) time: 6.4030 data: 4.6489 max mem: 3531 +Test: [ 10/235] eta: 0:02:30 acc1_pos: 95.833 acc5_pos: 100.000 loss: 0.2224 (0.2298) acc1: 94.2708 (94.0104) acc5: 100.0000 (100.0000) time: 0.6676 data: 0.4230 max mem: 3531 +Test: [ 20/235] eta: 0:01:25 acc1_pos: 94.010 acc5_pos: 100.000 loss: 0.2224 (0.2260) acc1: 94.0104 (94.0476) acc5: 100.0000 (100.0000) time: 0.0970 data: 0.0004 max mem: 3531 +Test: [ 30/235] eta: 0:01:11 acc1_pos: 94.531 acc5_pos: 100.000 loss: 0.2227 (0.2247) acc1: 94.0104 (94.1532) acc5: 100.0000 (100.0000) time: 0.1720 data: 0.0763 max mem: 3531 +Test: [ 40/235] eta: 0:01:02 acc1_pos: 93.750 acc5_pos: 100.000 loss: 0.2180 (0.2241) acc1: 94.5312 (94.2391) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1505 max mem: 3531 +Test: [ 50/235] eta: 0:00:55 acc1_pos: 94.792 acc5_pos: 100.000 loss: 0.2193 (0.2253) acc1: 94.5312 (94.2351) acc5: 100.0000 (100.0000) time: 0.2228 data: 0.1291 max mem: 3531 +Test: [ 60/235] eta: 0:00:49 acc1_pos: 93.750 acc5_pos: 100.000 loss: 0.2336 (0.2266) acc1: 94.0104 (94.1855) acc5: 100.0000 (100.0000) time: 0.2049 data: 0.1105 max mem: 3531 +Test: [ 70/235] eta: 0:00:44 acc1_pos: 93.846 acc5_pos: 100.000 loss: 0.2475 (0.3824) acc1: 91.9271 (84.9142) acc5: 100.0000 (100.0000) acc1_neg: 17.708 acc5_neg: 100.000 time: 0.1949 data: 0.0983 max mem: 3531 +Test: [ 80/235] eta: 0:00:41 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.4688 (0.5170) acc1: 20.3125 (76.8165) acc5: 100.0000 (100.0000) acc1_neg: 16.667 acc5_neg: 100.000 time: 0.2099 data: 0.1130 max mem: 3531 +Test: [ 90/235] eta: 0:00:36 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.4764 (0.6236) acc1: 19.5312 (70.4813) acc5: 100.0000 (100.0000) acc1_neg: 17.188 acc5_neg: 100.000 time: 0.1799 data: 0.0834 max mem: 3531 +Test: [100/235] eta: 0:00:33 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.4717 (0.7060) acc1: 19.0104 (65.4729) acc5: 100.0000 (100.0000) acc1_neg: 16.146 acc5_neg: 100.000 time: 0.1855 data: 0.0916 max mem: 3531 +Test: [110/235] eta: 0:00:30 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.4597 (0.7747) acc1: 19.5312 (61.3833) acc5: 100.0000 (100.0000) acc1_neg: 21.094 acc5_neg: 100.000 time: 0.2299 data: 0.1344 max mem: 3531 +Test: [120/235] eta: 0:00:27 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.5298 (0.8434) acc1: 18.4896 (57.4660) acc5: 100.0000 (100.0000) acc1_neg: 14.583 acc5_neg: 100.000 time: 0.1892 data: 0.0923 max mem: 3531 +Test: [130/235] eta: 0:00:25 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.6191 (0.9020) acc1: 13.8021 (54.1488) acc5: 100.0000 (100.0000) acc1_neg: 12.760 acc5_neg: 100.000 time: 0.1949 data: 0.0976 max mem: 3531 +Test: [140/235] eta: 0:00:22 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.6318 (0.9552) acc1: 13.2812 (51.2467) acc5: 100.0000 (100.0000) acc1_neg: 11.458 acc5_neg: 100.000 time: 0.2149 data: 0.1188 max mem: 3531 +Test: [150/235] eta: 0:00:19 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.6361 (0.9987) acc1: 13.0208 (48.7824) acc5: 100.0000 (100.0000) acc1_neg: 15.365 acc5_neg: 100.000 time: 0.2099 data: 0.1146 max mem: 3531 +Test: [160/235] eta: 0:00:17 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.6153 (1.0372) acc1: 14.0625 (46.6308) acc5: 100.0000 (100.0000) acc1_neg: 14.062 acc5_neg: 100.000 time: 0.1949 data: 0.0953 max mem: 3531 +Test: [170/235] eta: 0:00:14 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.5629 (1.0468) acc1: 15.6250 (45.7618) acc5: 100.0000 (100.0000) acc1_neg: 29.427 acc5_neg: 100.000 time: 0.1999 data: 0.1002 max mem: 3531 +Test: [180/235] eta: 0:00:12 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.1635 (1.0523) acc1: 34.3750 (45.1442) acc5: 100.0000 (100.0000) acc1_neg: 33.073 acc5_neg: 100.000 time: 0.2149 data: 0.1202 max mem: 3531 +Test: [190/235] eta: 0:00:10 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.1635 (1.0580) acc1: 34.3750 (44.5394) acc5: 100.0000 (100.0000) acc1_neg: 34.635 acc5_neg: 100.000 time: 0.1799 data: 0.0839 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.1652 (1.0626) acc1: 33.3333 (44.0039) acc5: 100.0000 (100.0000) acc1_neg: 32.031 acc5_neg: 100.000 time: 0.1310 data: 0.0323 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.1535 (1.0665) acc1: 33.3333 (43.5254) acc5: 100.0000 (100.0000) acc1_neg: 34.896 acc5_neg: 100.000 time: 0.1437 data: 0.0486 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.1472 (1.0705) acc1: 34.8958 (43.1113) acc5: 100.0000 (100.0000) acc1_neg: 32.552 acc5_neg: 100.000 time: 0.1539 data: 0.0630 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.1351 (1.0728) acc1: 35.1562 (42.7636) acc5: 100.0000 (100.0000) acc1_neg: 34.375 acc5_neg: 100.000 time: 0.1270 data: 0.0357 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.1351 (1.0747) acc1: 34.8958 (42.6384) acc5: 100.0000 (100.0000) acc1_neg: 34.783 acc5_neg: 100.000 time: 0.1121 data: 0.0241 max mem: 3532 +Test: Total time: 0:00:48 (0.2084 s / it) +* Overall Acc@1 42.638 Acc@5 100.000 + target == real Acc@1 94.203 Acc@5 100.000 + target == fake Acc@1 23.989 Acc@5 100.000 + loss 1.075 + PR-AUC 0.8706 ROC-AUC 0.7289 EER 33.02% (thr=0.1663) +Accuracy of the network on the 89879 test images: 42.6% +[rank0]:[W726 05:28:57.176708345 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:29:16.708000 1191167 torch/distributed/run.py:766] +W0726 05:29:16.708000 1191167 torch/distributed/run.py:766] ***************************************** +W0726 05:29:16.708000 1191167 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:29:16.708000 1191167 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:29:44.595066703 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:29:44.623683482 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +| distributed init (rank 6): env:// +[rank0]:[W726 05:29:45.577177634 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:29:45.581291348 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:29:45.590210403 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:29:45.601475521 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:29:45.712860923 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:29:45.780411456 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n11_seed8_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n11_classseed8_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:24:59 acc1_pos: 2.344 acc5_pos: 100.000 loss: 2.0980 (2.0980) acc1: 2.3438 (2.3438) acc5: 100.0000 (100.0000) time: 5.5754 data: 3.8131 max mem: 3531 +Test: [ 10/269] eta: 0:02:32 acc1_pos: 21.094 acc5_pos: 100.000 loss: 1.4802 (1.5980) acc1: 13.8021 (16.0748) acc5: 100.0000 (100.0000) time: 0.5889 data: 0.3470 max mem: 3531 +Test: [ 20/269] eta: 0:01:28 acc1_pos: 20.052 acc5_pos: 100.000 loss: 1.4802 (1.5461) acc1: 13.8021 (16.1582) acc5: 100.0000 (100.0000) time: 0.0952 data: 0.0036 max mem: 3531 +Test: [ 30/269] eta: 0:01:14 acc1_pos: 9.635 acc5_pos: 100.000 loss: 1.5524 (1.5858) acc1: 13.5417 (16.0702) acc5: 100.0000 (100.0000) time: 0.1598 data: 0.0635 max mem: 3531 +Test: [ 40/269] eta: 0:01:07 acc1_pos: 2.586 acc5_pos: 100.000 loss: 1.1978 (1.2879) acc1: 29.6875 (33.3587) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2347 data: 0.1393 max mem: 3531 +Test: [ 50/269] eta: 0:01:01 acc1_pos: 2.586 acc5_pos: 100.000 loss: 0.1456 (1.0664) acc1: 98.6979 (46.0018) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2323 data: 0.1415 max mem: 3531 +Test: [ 60/269] eta: 0:00:54 acc1_pos: 2.586 acc5_pos: 100.000 loss: 0.1712 (0.9234) acc1: 97.3958 (54.2777) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1899 data: 0.0978 max mem: 3531 +Test: [ 70/269] eta: 0:00:49 acc1_pos: 2.586 acc5_pos: 100.000 loss: 0.1840 (0.8232) acc1: 96.8750 (60.2479) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1597 data: 0.0662 max mem: 3531 +Test: [ 80/269] eta: 0:00:44 acc1_pos: 2.586 acc5_pos: 100.000 loss: 0.1770 (0.7424) acc1: 97.9167 (64.9402) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1610 data: 0.0664 max mem: 3531 +Test: [ 90/269] eta: 0:00:41 acc1_pos: 2.586 acc5_pos: 100.000 loss: 0.1888 (0.6939) acc1: 97.1354 (67.7742) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1826 data: 0.0871 max mem: 3531 +Test: [100/269] eta: 0:00:38 acc1_pos: 2.586 acc5_pos: 100.000 loss: 0.2445 (0.6521) acc1: 93.4896 (70.2119) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1893 data: 0.0953 max mem: 3531 +Test: [110/269] eta: 0:00:35 acc1_pos: 2.586 acc5_pos: 100.000 loss: 0.2451 (0.6169) acc1: 93.7500 (72.2551) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1658 data: 0.0732 max mem: 3531 +Test: [120/269] eta: 0:00:32 acc1_pos: 2.586 acc5_pos: 100.000 loss: 0.2376 (0.5876) acc1: 94.2708 (73.8895) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1644 data: 0.0701 max mem: 3531 +Test: [130/269] eta: 0:00:29 acc1_pos: 2.586 acc5_pos: 100.000 loss: 0.2786 (0.5671) acc1: 91.9271 (75.0358) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1893 data: 0.0948 max mem: 3531 +Test: [140/269] eta: 0:00:27 acc1_pos: 2.586 acc5_pos: 100.000 loss: 0.2975 (0.5476) acc1: 92.4479 (76.3501) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1999 data: 0.1058 max mem: 3531 +Test: [150/269] eta: 0:00:24 acc1_pos: 2.586 acc5_pos: 100.000 loss: 0.2698 (0.5257) acc1: 94.5312 (77.6335) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1557 data: 0.0605 max mem: 3531 +Test: [160/269] eta: 0:00:22 acc1_pos: 2.586 acc5_pos: 100.000 loss: 0.1874 (0.5052) acc1: 96.3542 (78.8173) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1255 data: 0.0331 max mem: 3531 +Test: [170/269] eta: 0:00:20 acc1_pos: 2.586 acc5_pos: 100.000 loss: 0.2197 (0.4898) acc1: 95.8333 (79.7012) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1743 data: 0.0816 max mem: 3531 +Test: [180/269] eta: 0:00:18 acc1_pos: 2.586 acc5_pos: 100.000 loss: 0.2002 (0.4738) acc1: 96.0938 (80.6299) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1943 data: 0.1008 max mem: 3531 +Test: [190/269] eta: 0:00:16 acc1_pos: 2.586 acc5_pos: 100.000 loss: 0.2006 (0.4617) acc1: 95.3125 (81.3577) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.2044 data: 0.1114 max mem: 3531 +Test: [200/269] eta: 0:00:13 acc1_pos: 2.586 acc5_pos: 100.000 loss: 0.2631 (0.4547) acc1: 93.2292 (81.5985) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1849 data: 0.0882 max mem: 3531 +Test: [210/269] eta: 0:00:11 acc1_pos: 2.586 acc5_pos: 100.000 loss: 0.2539 (0.4427) acc1: 94.5312 (82.3275) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1275 data: 0.0291 max mem: 3532 +Test: [220/269] eta: 0:00:09 acc1_pos: 2.586 acc5_pos: 100.000 loss: 0.1651 (0.4303) acc1: 97.9167 (83.0152) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1417 data: 0.0473 max mem: 3532 +Test: [230/269] eta: 0:00:07 acc1_pos: 2.586 acc5_pos: 100.000 loss: 0.1507 (0.4181) acc1: 98.6979 (83.6682) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1798 data: 0.0862 max mem: 3532 +Test: [240/269] eta: 0:00:05 acc1_pos: 2.586 acc5_pos: 100.000 loss: 0.1684 (0.4136) acc1: 97.6562 (83.8682) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1883 data: 0.0953 max mem: 3532 +Test: [250/269] eta: 0:00:03 acc1_pos: 2.586 acc5_pos: 100.000 loss: 0.2361 (0.4072) acc1: 95.3125 (84.2982) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1782 data: 0.0861 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 2.586 acc5_pos: 100.000 loss: 0.2209 (0.3995) acc1: 96.0938 (84.7921) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1475 data: 0.0550 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 2.586 acc5_pos: 100.000 loss: 0.1986 (0.3935) acc1: 97.3958 (85.1665) acc5: 100.0000 (100.0000) acc1_neg: 96.850 acc5_neg: 100.000 time: 0.1310 data: 0.0368 max mem: 3532 +Test: Total time: 0:00:51 (0.1900 s / it) +* Overall Acc@1 85.166 Acc@5 100.000 + target == real Acc@1 15.705 Acc@5 100.000 + target == fake Acc@1 94.646 Acc@5 100.000 + loss 0.393 + PR-AUC 0.9099 ROC-AUC 0.6178 EER 41.59% (thr=0.8447) +Accuracy of the network on the 103293 test images: 85.2% +[rank0]:[W726 05:30:50.954337723 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T05:30:56Z] eval_done tag=fixed500k_real_n11_seed8_resampled +[2026-07-26T05:30:57Z] eval_start tag=fixed500k_fake_n22_seed8_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed8_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T05:30:57Z +hostname=n116-041-188 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed8_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n22_seed8_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 05:31:09.347000 1191742 torch/distributed/run.py:766] +W0726 05:31:09.347000 1191742 torch/distributed/run.py:766] ***************************************** +W0726 05:31:09.347000 1191742 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:31:09.347000 1191742 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:31:35.598636321 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:31:35.872487681 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:31:36.947451221 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W726 05:31:36.947504958 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:31:36.063167576 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:31:36.203055632 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +[rank0]:[W726 05:31:36.227030311 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:31:36.258616745 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n22_seed8_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed8_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:24:52 acc1_pos: 22.135 acc5_pos: 100.000 loss: 1.0852 (1.0852) acc1: 22.1354 (22.1354) acc5: 100.0000 (100.0000) time: 6.3785 data: 4.8415 max mem: 3531 +Test: [ 10/234] eta: 0:02:28 acc1_pos: 11.458 acc5_pos: 100.000 loss: 0.8874 (0.8782) acc1: 36.4583 (38.4943) acc5: 100.0000 (100.0000) time: 0.6618 data: 0.4405 max mem: 3531 +Test: [ 20/234] eta: 0:01:24 acc1_pos: 96.094 acc5_pos: 100.000 loss: 0.8874 (0.8750) acc1: 36.4583 (40.2778) acc5: 100.0000 (100.0000) time: 0.0934 data: 0.0004 max mem: 3531 +Test: [ 30/234] eta: 0:01:14 acc1_pos: 70.573 acc5_pos: 100.000 loss: 0.2142 (0.6764) acc1: 95.3125 (56.7792) acc5: 100.0000 (100.0000) time: 0.2033 data: 0.1091 max mem: 3531 +Test: [ 40/234] eta: 0:01:04 acc1_pos: 98.958 acc5_pos: 100.000 loss: 0.2112 (0.6150) acc1: 95.5729 (62.0617) acc5: 100.0000 (100.0000) time: 0.2649 data: 0.1708 max mem: 3531 +Test: [ 50/234] eta: 0:00:55 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1489 (0.5226) acc1: 98.1771 (69.1381) acc5: 100.0000 (100.0000) time: 0.1999 data: 0.1048 max mem: 3531 +Test: [ 60/234] eta: 0:00:48 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.1342 (0.4580) acc1: 98.9583 (74.1163) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1749 data: 0.0781 max mem: 3531 +Test: [ 70/234] eta: 0:00:43 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.1320 (0.4941) acc1: 99.4792 (72.0291) acc5: 100.0000 (100.0000) acc1_neg: 36.198 acc5_neg: 100.000 time: 0.1849 data: 0.0876 max mem: 3531 +Test: [ 80/234] eta: 0:00:40 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.9424 (0.5197) acc1: 41.6667 (70.6694) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2068 data: 0.1116 max mem: 3531 +Test: [ 90/234] eta: 0:00:36 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.2671 (0.4903) acc1: 91.1458 (73.1084) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.1949 data: 0.0988 max mem: 3531 +Test: [100/234] eta: 0:00:32 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.3283 (0.4911) acc1: 83.8542 (73.4839) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1735 data: 0.0782 max mem: 3531 +Test: [110/234] eta: 0:00:29 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.4786 (0.4869) acc1: 75.0000 (74.1390) acc5: 100.0000 (100.0000) acc1_neg: 71.354 acc5_neg: 100.000 time: 0.1849 data: 0.0906 max mem: 3531 +Test: [120/234] eta: 0:00:27 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.4468 (0.4877) acc1: 81.5104 (74.4232) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.2045 data: 0.1101 max mem: 3531 +Test: [130/234] eta: 0:00:23 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.3614 (0.4786) acc1: 84.8958 (75.6103) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.1668 data: 0.0736 max mem: 3531 +Test: [140/234] eta: 0:00:21 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.3753 (0.4963) acc1: 84.8958 (75.1976) acc5: 100.0000 (100.0000) acc1_neg: 46.354 acc5_neg: 100.000 time: 0.1467 data: 0.0549 max mem: 3531 +Test: [150/234] eta: 0:00:18 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.4885 (0.5035) acc1: 72.6562 (74.9931) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.1892 data: 0.0965 max mem: 3531 +Test: [160/234] eta: 0:00:16 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.4205 (0.4942) acc1: 84.8958 (75.9026) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.1886 data: 0.0951 max mem: 3531 +Test: [170/234] eta: 0:00:13 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.4016 (0.4896) acc1: 86.1979 (76.4483) acc5: 100.0000 (100.0000) acc1_neg: 73.958 acc5_neg: 100.000 time: 0.1706 data: 0.0757 max mem: 3531 +Test: [180/234] eta: 0:00:11 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.4328 (0.5022) acc1: 83.0729 (75.5856) acc5: 100.0000 (100.0000) acc1_neg: 22.396 acc5_neg: 100.000 time: 0.1593 data: 0.0648 max mem: 3531 +Test: [190/234] eta: 0:00:09 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.5637 (0.5066) acc1: 70.3125 (75.2891) acc5: 100.0000 (100.0000) acc1_neg: 70.312 acc5_neg: 100.000 time: 0.1745 data: 0.0812 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.5394 (0.4951) acc1: 73.4375 (76.0171) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1852 data: 0.0912 max mem: 3531 +Test: [210/234] eta: 0:00:04 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.1177 (0.4800) acc1: 99.7396 (77.0315) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1495 data: 0.0568 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.2859 (0.4718) acc1: 93.4896 (77.7126) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1222 data: 0.0304 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.3626 (0.4752) acc1: 72.6562 (77.3911) acc5: 100.0000 (100.0000) acc1_neg: 68.490 acc5_neg: 100.000 time: 0.1147 data: 0.0238 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.5351 (0.4760) acc1: 71.6146 (77.2972) acc5: 100.0000 (100.0000) acc1_neg: 75.486 acc5_neg: 100.000 time: 0.1204 data: 0.0237 max mem: 3532 +Test: Total time: 0:00:46 (0.1988 s / it) +* Overall Acc@1 77.297 Acc@5 100.000 + target == real Acc@1 69.355 Acc@5 100.000 + target == fake Acc@1 79.536 Acc@5 100.000 + loss 0.476 + PR-AUC 0.9504 ROC-AUC 0.8448 EER 25.30% (thr=0.5674) +Accuracy of the network on the 89729 test images: 77.3% +[rank0]:[W726 05:32:33.051918365 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:32:52.500000 1192290 torch/distributed/run.py:766] +W0726 05:32:52.500000 1192290 torch/distributed/run.py:766] ***************************************** +W0726 05:32:52.500000 1192290 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:32:52.500000 1192290 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:33:20.266755841 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:33:20.590853965 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:33:20.637291356 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:33:20.769639465 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:33:20.789553945 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +| distributed init (rank 7): env:// +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:33:21.009341371 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:33:21.019380156 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W726 05:33:21.019415036 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n22_seed8_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed8_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 5:02:38 acc1_pos: 10.677 acc5_pos: 100.000 loss: 1.2030 (1.2030) acc1: 10.6771 (10.6771) acc5: 100.0000 (100.0000) time: 7.6074 data: 5.9961 max mem: 3531 +Test: [ 10/2387] eta: 0:30:57 acc1_pos: 4.688 acc5_pos: 100.000 loss: 1.3181 (1.2816) acc1: 6.5104 (8.8542) acc5: 100.0000 (100.0000) time: 0.7816 data: 0.5455 max mem: 3531 +Test: [ 20/2387] eta: 0:18:01 acc1_pos: 1.042 acc5_pos: 100.000 loss: 1.2997 (1.2820) acc1: 5.9896 (7.7133) acc5: 100.0000 (100.0000) time: 0.0995 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:15:23 acc1_pos: 0.781 acc5_pos: 100.000 loss: 1.2997 (1.3023) acc1: 5.2083 (7.3757) acc5: 100.0000 (100.0000) time: 0.1773 data: 0.0823 max mem: 3531 +Test: [ 40/2387] eta: 0:13:50 acc1_pos: 6.771 acc5_pos: 100.000 loss: 1.3764 (1.3280) acc1: 1.8229 (6.3326) acc5: 100.0000 (100.0000) time: 0.2455 data: 0.1538 max mem: 3531 +Test: [ 50/2387] eta: 0:12:54 acc1_pos: 30.469 acc5_pos: 100.000 loss: 1.3492 (1.3088) acc1: 4.1667 (7.3121) acc5: 100.0000 (100.0000) time: 0.2377 data: 0.1462 max mem: 3531 +Test: [ 60/2387] eta: 0:12:12 acc1_pos: 30.990 acc5_pos: 100.000 loss: 1.2129 (1.2912) acc1: 10.1562 (8.4315) acc5: 100.0000 (100.0000) time: 0.2343 data: 0.1422 max mem: 3531 +Test: [ 70/2387] eta: 0:11:47 acc1_pos: 17.969 acc5_pos: 100.000 loss: 1.2433 (1.2923) acc1: 9.6354 (8.1573) acc5: 100.0000 (100.0000) time: 0.2387 data: 0.1445 max mem: 3531 +Test: [ 80/2387] eta: 0:11:31 acc1_pos: 1.562 acc5_pos: 100.000 loss: 1.3425 (1.3015) acc1: 3.6458 (7.8447) acc5: 100.0000 (100.0000) time: 0.2540 data: 0.1603 max mem: 3531 +Test: [ 90/2387] eta: 0:11:20 acc1_pos: 4.948 acc5_pos: 100.000 loss: 1.3445 (1.3077) acc1: 4.9479 (7.7009) acc5: 100.0000 (100.0000) time: 0.2649 data: 0.1673 max mem: 3531 +Test: [ 100/2387] eta: 0:11:00 acc1_pos: 0.260 acc5_pos: 100.000 loss: 1.4042 (1.3180) acc1: 5.2083 (7.3974) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1476 max mem: 3531 +Test: [ 110/2387] eta: 0:10:47 acc1_pos: 7.031 acc5_pos: 100.000 loss: 1.3856 (1.3204) acc1: 5.2083 (7.3222) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1378 max mem: 3531 +Test: [ 120/2387] eta: 0:10:38 acc1_pos: 7.812 acc5_pos: 100.000 loss: 1.3355 (1.3245) acc1: 7.0312 (7.4208) acc5: 100.0000 (100.0000) time: 0.2451 data: 0.1523 max mem: 3531 +Test: [ 130/2387] eta: 0:10:28 acc1_pos: 14.844 acc5_pos: 100.000 loss: 1.3446 (1.3293) acc1: 7.5521 (7.4050) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1505 max mem: 3531 +Test: [ 140/2387] eta: 0:10:19 acc1_pos: 1.302 acc5_pos: 100.000 loss: 1.4297 (1.3354) acc1: 3.9062 (7.2215) acc5: 100.0000 (100.0000) time: 0.2397 data: 0.1461 max mem: 3531 +Test: [ 150/2387] eta: 0:10:17 acc1_pos: 3.646 acc5_pos: 100.000 loss: 1.3573 (1.3350) acc1: 5.7292 (7.2486) acc5: 100.0000 (100.0000) time: 0.2605 data: 0.1686 max mem: 3531 +Test: [ 160/2387] eta: 0:09:59 acc1_pos: 0.457 acc5_pos: 100.000 loss: 1.2900 (1.3122) acc1: 6.5104 (9.3410) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2249 data: 0.1339 max mem: 3531 +Test: [ 170/2387] eta: 0:09:53 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.2081 (1.2471) acc1: 98.6979 (14.6107) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2045 data: 0.1137 max mem: 3531 +Test: [ 180/2387] eta: 0:09:36 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1988 (1.1892) acc1: 99.4792 (19.2939) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1949 data: 0.1033 max mem: 3531 +Test: [ 190/2387] eta: 0:09:26 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1957 (1.1373) acc1: 99.4792 (23.4934) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1747 data: 0.0813 max mem: 3531 +Test: [ 200/2387] eta: 0:09:17 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1952 (1.0905) acc1: 99.4792 (27.2764) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1999 data: 0.1038 max mem: 3531 +Test: [ 210/2387] eta: 0:09:08 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1977 (1.0484) acc1: 99.4792 (30.6884) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.0981 max mem: 3532 +Test: [ 220/2387] eta: 0:08:55 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1991 (1.0099) acc1: 99.4792 (33.7988) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1699 data: 0.0731 max mem: 3532 +Test: [ 230/2387] eta: 0:08:46 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1999 (0.9748) acc1: 99.4792 (36.6410) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1604 data: 0.0648 max mem: 3532 +Test: [ 240/2387] eta: 0:08:38 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1918 (0.9422) acc1: 99.7396 (39.2602) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1769 data: 0.0845 max mem: 3532 +Test: [ 250/2387] eta: 0:08:32 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1848 (0.9091) acc1: 99.7396 (41.6729) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1900 data: 0.0969 max mem: 3532 +Test: [ 260/2387] eta: 0:08:26 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0952 (0.8777) acc1: 100.0000 (43.9067) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.1017 max mem: 3532 +Test: [ 270/2387] eta: 0:08:19 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0890 (0.8486) acc1: 100.0000 (45.9746) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1893 data: 0.0975 max mem: 3532 +Test: [ 280/2387] eta: 0:08:13 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0894 (0.8217) acc1: 100.0000 (47.8907) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1850 data: 0.0928 max mem: 3532 +Test: [ 290/2387] eta: 0:08:07 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0929 (0.7967) acc1: 100.0000 (49.6814) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1874 data: 0.0957 max mem: 3532 +Test: [ 300/2387] eta: 0:08:02 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0950 (0.7735) acc1: 100.0000 (51.3523) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1902 data: 0.0971 max mem: 3532 +Test: [ 310/2387] eta: 0:07:57 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1000 (0.7568) acc1: 100.0000 (52.7725) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1887 data: 0.0948 max mem: 3532 +Test: [ 320/2387] eta: 0:07:52 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.4456 (0.7473) acc1: 91.1458 (53.9314) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.1928 data: 0.0992 max mem: 3532 +Test: [ 330/2387] eta: 0:07:49 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.4456 (0.7382) acc1: 90.1042 (55.0337) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.2013 data: 0.1056 max mem: 3532 +Test: [ 340/2387] eta: 0:07:45 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.4452 (0.7297) acc1: 90.6250 (56.0682) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.2009 data: 0.1053 max mem: 3532 +Test: [ 350/2387] eta: 0:07:39 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.4384 (0.7152) acc1: 91.4062 (57.2872) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1826 data: 0.0901 max mem: 3532 +Test: [ 360/2387] eta: 0:07:34 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1844 (0.7003) acc1: 100.0000 (58.4610) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1693 data: 0.0751 max mem: 3532 +Test: [ 370/2387] eta: 0:07:28 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1768 (0.6862) acc1: 99.7396 (59.5694) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1712 data: 0.0739 max mem: 3532 +Test: [ 380/2387] eta: 0:07:24 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1766 (0.6728) acc1: 99.7396 (60.6224) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1758 data: 0.0817 max mem: 3532 +Test: [ 390/2387] eta: 0:07:20 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1766 (0.6602) acc1: 99.7396 (61.6202) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1786 data: 0.0865 max mem: 3532 +Test: [ 400/2387] eta: 0:07:15 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1774 (0.6482) acc1: 99.7396 (62.5682) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1783 data: 0.0853 max mem: 3532 +Test: [ 410/2387] eta: 0:07:10 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1767 (0.6367) acc1: 99.7396 (63.4675) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1710 data: 0.0790 max mem: 3532 +Test: [ 420/2387] eta: 0:07:06 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1751 (0.6257) acc1: 99.7396 (64.3303) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1725 data: 0.0804 max mem: 3532 +Test: [ 430/2387] eta: 0:07:03 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1744 (0.6152) acc1: 99.7396 (65.1507) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1814 data: 0.0890 max mem: 3532 +Test: [ 440/2387] eta: 0:06:58 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1684 (0.6043) acc1: 99.7396 (65.9362) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1752 data: 0.0833 max mem: 3533 +Test: [ 450/2387] eta: 0:06:54 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1061 (0.5932) acc1: 100.0000 (66.6915) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1668 data: 0.0759 max mem: 3533 +Test: [ 460/2387] eta: 0:06:50 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1057 (0.5826) acc1: 100.0000 (67.4140) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1680 data: 0.0775 max mem: 3533 +Test: [ 470/2387] eta: 0:06:46 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1062 (0.5725) acc1: 100.0000 (68.1059) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1702 data: 0.0774 max mem: 3533 +Test: [ 480/2387] eta: 0:06:42 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1070 (0.5629) acc1: 100.0000 (68.7690) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1669 data: 0.0741 max mem: 3533 +Test: [ 490/2387] eta: 0:06:39 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1069 (0.5536) acc1: 100.0000 (69.4050) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1675 data: 0.0764 max mem: 3533 +Test: [ 500/2387] eta: 0:06:35 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1066 (0.5447) acc1: 100.0000 (70.0157) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1650 data: 0.0725 max mem: 3533 +Test: [ 510/2387] eta: 0:06:31 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1069 (0.5361) acc1: 100.0000 (70.6025) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1630 data: 0.0704 max mem: 3533 +Test: [ 520/2387] eta: 0:06:27 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1092 (0.5279) acc1: 100.0000 (71.1667) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1629 data: 0.0717 max mem: 3533 +Test: [ 530/2387] eta: 0:06:23 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1075 (0.5200) acc1: 100.0000 (71.7097) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1611 data: 0.0698 max mem: 3533 +Test: [ 540/2387] eta: 0:06:20 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1058 (0.5123) acc1: 100.0000 (72.2327) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1642 data: 0.0724 max mem: 3533 +Test: [ 550/2387] eta: 0:06:16 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1059 (0.5050) acc1: 100.0000 (72.7366) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1625 data: 0.0706 max mem: 3533 +Test: [ 560/2387] eta: 0:06:13 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1058 (0.4979) acc1: 100.0000 (73.2226) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1601 data: 0.0687 max mem: 3533 +Test: [ 570/2387] eta: 0:06:10 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1063 (0.4910) acc1: 100.0000 (73.6915) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1632 data: 0.0718 max mem: 3533 +Test: [ 580/2387] eta: 0:06:06 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1062 (0.4844) acc1: 100.0000 (74.1443) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1642 data: 0.0707 max mem: 3533 +Test: [ 590/2387] eta: 0:06:03 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1062 (0.4780) acc1: 100.0000 (74.5818) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1684 data: 0.0752 max mem: 3533 +Test: [ 600/2387] eta: 0:06:00 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1092 (0.4719) acc1: 100.0000 (75.0048) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1715 data: 0.0803 max mem: 3533 +Test: [ 610/2387] eta: 0:05:58 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1219 (0.4665) acc1: 100.0000 (75.4134) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1732 data: 0.0806 max mem: 3533 +Test: [ 620/2387] eta: 0:05:55 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1420 (0.4612) acc1: 100.0000 (75.8093) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1757 data: 0.0830 max mem: 3533 +Test: [ 630/2387] eta: 0:05:52 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1412 (0.4562) acc1: 100.0000 (76.1927) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1716 data: 0.0794 max mem: 3533 +Test: [ 640/2387] eta: 0:05:49 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1417 (0.4513) acc1: 100.0000 (76.5633) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1691 data: 0.0765 max mem: 3533 +Test: [ 650/2387] eta: 0:05:46 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1413 (0.4465) acc1: 100.0000 (76.9229) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1664 data: 0.0750 max mem: 3533 +Test: [ 660/2387] eta: 0:05:43 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1408 (0.4419) acc1: 100.0000 (77.2720) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1666 data: 0.0760 max mem: 3533 +Test: [ 670/2387] eta: 0:05:41 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1413 (0.4374) acc1: 100.0000 (77.6108) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1866 data: 0.0955 max mem: 3534 +Test: [ 680/2387] eta: 0:05:39 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1391 (0.4329) acc1: 100.0000 (77.9395) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1850 data: 0.0926 max mem: 3534 +Test: [ 690/2387] eta: 0:05:37 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1397 (0.4311) acc1: 100.0000 (78.1046) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1935 data: 0.1015 max mem: 3534 +Test: [ 700/2387] eta: 0:05:34 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.3697 (0.4303) acc1: 85.9375 (78.2038) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1945 data: 0.1033 max mem: 3534 +Test: [ 710/2387] eta: 0:05:32 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.3742 (0.4295) acc1: 85.1562 (78.3015) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1832 data: 0.0914 max mem: 3534 +Test: [ 720/2387] eta: 0:05:30 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.3686 (0.4287) acc1: 85.4167 (78.4035) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1970 data: 0.1047 max mem: 3534 +Test: [ 730/2387] eta: 0:05:28 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.3642 (0.4279) acc1: 85.1562 (78.4884) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1948 data: 0.1024 max mem: 3534 +Test: [ 740/2387] eta: 0:05:26 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.3689 (0.4271) acc1: 84.8958 (78.5826) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.2002 data: 0.1047 max mem: 3534 +Test: [ 750/2387] eta: 0:05:25 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.3678 (0.4263) acc1: 85.1562 (78.6659) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.2045 data: 0.1090 max mem: 3534 +Test: [ 760/2387] eta: 0:05:22 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.3640 (0.4250) acc1: 85.1562 (78.7961) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1898 data: 0.0966 max mem: 3534 +Test: [ 770/2387] eta: 0:05:20 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.2455 (0.4224) acc1: 94.0104 (79.0255) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1884 data: 0.0951 max mem: 3534 +Test: [ 780/2387] eta: 0:05:18 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.2170 (0.4194) acc1: 98.9583 (79.2844) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1946 data: 0.1032 max mem: 3534 +Test: [ 790/2387] eta: 0:05:16 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1802 (0.4167) acc1: 99.2188 (79.4956) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.1980 data: 0.1071 max mem: 3534 +Test: [ 800/2387] eta: 0:05:14 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1704 (0.4138) acc1: 99.7396 (79.7463) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.1037 max mem: 3534 +Test: [ 810/2387] eta: 0:05:11 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1763 (0.4109) acc1: 100.0000 (79.9932) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1726 data: 0.0813 max mem: 3534 +Test: [ 820/2387] eta: 0:05:09 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1858 (0.4082) acc1: 100.0000 (80.2315) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1675 data: 0.0761 max mem: 3534 +Test: [ 830/2387] eta: 0:05:06 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1946 (0.4060) acc1: 99.2188 (80.4305) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1699 data: 0.0767 max mem: 3534 +Test: [ 840/2387] eta: 0:05:04 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.2278 (0.4040) acc1: 98.1771 (80.6369) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1674 data: 0.0743 max mem: 3534 +Test: [ 850/2387] eta: 0:05:01 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.2459 (0.4022) acc1: 97.9167 (80.8155) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1696 data: 0.0789 max mem: 3534 +Test: [ 860/2387] eta: 0:04:59 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.2335 (0.4001) acc1: 98.4375 (81.0238) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1677 data: 0.0765 max mem: 3534 +Test: [ 870/2387] eta: 0:04:56 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.2242 (0.3981) acc1: 99.2188 (81.2261) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1676 data: 0.0756 max mem: 3534 +Test: [ 880/2387] eta: 0:04:54 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.2285 (0.3964) acc1: 98.4375 (81.4117) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1726 data: 0.0795 max mem: 3534 +Test: [ 890/2387] eta: 0:04:52 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.2061 (0.3940) acc1: 99.2188 (81.6177) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1775 data: 0.0831 max mem: 3534 +Test: [ 900/2387] eta: 0:04:50 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.2107 (0.3975) acc1: 98.9583 (81.3581) acc5: 100.0000 (100.0000) acc1_neg: 51.042 acc5_neg: 100.000 time: 0.1859 data: 0.0893 max mem: 3535 +Test: [ 910/2387] eta: 0:04:48 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.8214 (0.4022) acc1: 48.4375 (80.9933) acc5: 100.0000 (100.0000) acc1_neg: 47.917 acc5_neg: 100.000 time: 0.2041 data: 0.1074 max mem: 3535 +Test: [ 920/2387] eta: 0:04:46 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.8251 (0.4068) acc1: 47.9167 (80.6265) acc5: 100.0000 (100.0000) acc1_neg: 46.354 acc5_neg: 100.000 time: 0.1843 data: 0.0890 max mem: 3535 +Test: [ 930/2387] eta: 0:04:44 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.8271 (0.4113) acc1: 47.1354 (80.2780) acc5: 100.0000 (100.0000) acc1_neg: 45.833 acc5_neg: 100.000 time: 0.1719 data: 0.0786 max mem: 3535 +Test: [ 940/2387] eta: 0:04:42 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.8271 (0.4158) acc1: 46.6146 (79.9247) acc5: 100.0000 (100.0000) acc1_neg: 49.479 acc5_neg: 100.000 time: 0.2002 data: 0.1052 max mem: 3535 +Test: [ 950/2387] eta: 0:04:40 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.8237 (0.4201) acc1: 46.6146 (79.5829) acc5: 100.0000 (100.0000) acc1_neg: 46.354 acc5_neg: 100.000 time: 0.2128 data: 0.1139 max mem: 3535 +Test: [ 960/2387] eta: 0:04:39 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.8130 (0.4242) acc1: 48.4375 (79.2631) acc5: 100.0000 (100.0000) acc1_neg: 50.781 acc5_neg: 100.000 time: 0.2243 data: 0.1297 max mem: 3535 +Test: [ 970/2387] eta: 0:04:37 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.8173 (0.4283) acc1: 47.9167 (78.9349) acc5: 100.0000 (100.0000) acc1_neg: 47.917 acc5_neg: 100.000 time: 0.2299 data: 0.1381 max mem: 3535 +Test: [ 980/2387] eta: 0:04:35 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.7876 (0.4282) acc1: 49.7396 (78.9580) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.1940 data: 0.1004 max mem: 3535 +Test: [ 990/2387] eta: 0:04:33 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.2535 (0.4266) acc1: 96.8750 (79.1304) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1883 data: 0.0953 max mem: 3535 +Test: [1000/2387] eta: 0:04:31 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.2625 (0.4251) acc1: 97.9167 (79.3046) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2006 data: 0.1090 max mem: 3535 +Test: [1010/2387] eta: 0:04:29 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.2456 (0.4234) acc1: 98.1771 (79.4886) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1916 data: 0.0988 max mem: 3535 +Test: [1020/2387] eta: 0:04:27 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.2406 (0.4216) acc1: 98.4375 (79.6584) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2057 data: 0.1125 max mem: 3535 +Test: [1030/2387] eta: 0:04:25 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.2172 (0.4196) acc1: 99.2188 (79.8456) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2018 data: 0.1104 max mem: 3535 +Test: [1040/2387] eta: 0:04:24 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1996 (0.4174) acc1: 99.7396 (80.0345) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2034 data: 0.1106 max mem: 3535 +Test: [1050/2387] eta: 0:04:21 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1917 (0.4153) acc1: 99.7396 (80.2215) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1888 data: 0.0958 max mem: 3535 +Test: [1060/2387] eta: 0:04:19 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1917 (0.4132) acc1: 99.7396 (80.4030) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1815 data: 0.0905 max mem: 3535 +Test: [1070/2387] eta: 0:04:17 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1895 (0.4112) acc1: 99.7396 (80.5835) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1988 data: 0.1079 max mem: 3535 +Test: [1080/2387] eta: 0:04:15 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1990 (0.4094) acc1: 99.4792 (80.7429) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1758 data: 0.0855 max mem: 3535 +Test: [1090/2387] eta: 0:04:13 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1990 (0.4074) acc1: 99.2188 (80.9161) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1593 data: 0.0665 max mem: 3535 +Test: [1100/2387] eta: 0:04:10 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1916 (0.4054) acc1: 99.7396 (81.0821) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1641 data: 0.0688 max mem: 3535 +Test: [1110/2387] eta: 0:04:08 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1886 (0.4038) acc1: 99.7396 (81.2116) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1602 data: 0.0672 max mem: 3535 +Test: [1120/2387] eta: 0:04:05 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1893 (0.4020) acc1: 99.2188 (81.3694) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1524 data: 0.0614 max mem: 3536 +Test: [1130/2387] eta: 0:04:03 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1864 (0.4001) acc1: 99.7396 (81.5316) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1564 data: 0.0637 max mem: 3536 +Test: [1140/2387] eta: 0:04:01 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1829 (0.3981) acc1: 99.7396 (81.6891) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1572 data: 0.0641 max mem: 3536 +Test: [1150/2387] eta: 0:03:59 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1814 (0.3964) acc1: 99.7396 (81.8340) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1582 data: 0.0658 max mem: 3536 +Test: [1160/2387] eta: 0:03:56 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1781 (0.3944) acc1: 99.7396 (81.9864) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1526 data: 0.0586 max mem: 3536 +Test: [1170/2387] eta: 0:03:54 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1680 (0.3924) acc1: 100.0000 (82.1356) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1449 data: 0.0483 max mem: 3536 +Test: [1180/2387] eta: 0:03:52 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1755 (0.3906) acc1: 100.0000 (82.2773) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1671 data: 0.0722 max mem: 3536 +Test: [1190/2387] eta: 0:03:50 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1686 (0.3888) acc1: 99.7396 (82.4226) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1755 data: 0.0836 max mem: 3536 +Test: [1200/2387] eta: 0:03:47 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1552 (0.3868) acc1: 100.0000 (82.5688) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1702 data: 0.0753 max mem: 3536 +Test: [1210/2387] eta: 0:03:45 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1533 (0.3848) acc1: 100.0000 (82.7112) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1793 data: 0.0844 max mem: 3536 +Test: [1220/2387] eta: 0:03:43 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1284 (0.3826) acc1: 100.0000 (82.8505) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1783 data: 0.0855 max mem: 3536 +Test: [1230/2387] eta: 0:03:42 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1079 (0.3804) acc1: 100.0000 (82.9875) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2048 data: 0.1108 max mem: 3536 +Test: [1240/2387] eta: 0:03:40 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1080 (0.3782) acc1: 100.0000 (83.1229) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2089 data: 0.1165 max mem: 3536 +Test: [1250/2387] eta: 0:03:38 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1118 (0.3761) acc1: 100.0000 (83.2565) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1810 data: 0.0892 max mem: 3536 +Test: [1260/2387] eta: 0:03:36 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1111 (0.3739) acc1: 100.0000 (83.3874) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1961 data: 0.1036 max mem: 3536 +Test: [1270/2387] eta: 0:03:34 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1087 (0.3719) acc1: 100.0000 (83.5179) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1945 data: 0.1031 max mem: 3536 +Test: [1280/2387] eta: 0:03:32 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1126 (0.3700) acc1: 100.0000 (83.6456) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1938 data: 0.1018 max mem: 3536 +Test: [1290/2387] eta: 0:03:31 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1266 (0.3682) acc1: 100.0000 (83.7594) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2152 data: 0.1221 max mem: 3536 +Test: [1300/2387] eta: 0:03:29 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1297 (0.3664) acc1: 100.0000 (83.8796) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2147 data: 0.1222 max mem: 3536 +Test: [1310/2387] eta: 0:03:27 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1315 (0.3647) acc1: 100.0000 (83.9966) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1888 data: 0.0954 max mem: 3536 +Test: [1320/2387] eta: 0:03:25 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1383 (0.3630) acc1: 100.0000 (84.1077) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2099 data: 0.1167 max mem: 3536 +Test: [1330/2387] eta: 0:03:23 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1433 (0.3615) acc1: 100.0000 (84.2232) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2099 data: 0.1142 max mem: 3536 +Test: [1340/2387] eta: 0:03:21 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1664 (0.3610) acc1: 99.2188 (84.2571) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1854 data: 0.0871 max mem: 3536 +Test: [1350/2387] eta: 0:03:19 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0940 (0.3590) acc1: 100.0000 (84.3737) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1906 data: 0.0965 max mem: 3537 +Test: [1360/2387] eta: 0:03:17 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0832 (0.3570) acc1: 100.0000 (84.4885) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1950 data: 0.1003 max mem: 3537 +Test: [1370/2387] eta: 0:03:15 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0831 (0.3550) acc1: 100.0000 (84.6016) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1964 data: 0.1005 max mem: 3537 +Test: [1380/2387] eta: 0:03:13 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0873 (0.3530) acc1: 100.0000 (84.7131) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1900 data: 0.0936 max mem: 3537 +Test: [1390/2387] eta: 0:03:12 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0822 (0.3511) acc1: 100.0000 (84.8230) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1958 data: 0.0992 max mem: 3537 +Test: [1400/2387] eta: 0:03:10 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0813 (0.3492) acc1: 100.0000 (84.9313) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1924 data: 0.0997 max mem: 3537 +Test: [1410/2387] eta: 0:03:08 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.0879 (0.3478) acc1: 100.0000 (85.0328) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1983 data: 0.1060 max mem: 3537 +Test: [1420/2387] eta: 0:03:06 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1654 (0.3467) acc1: 99.7396 (85.1154) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1871 data: 0.0951 max mem: 3537 +Test: [1430/2387] eta: 0:03:04 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1712 (0.3456) acc1: 99.4792 (85.2150) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1742 data: 0.0828 max mem: 3537 +Test: [1440/2387] eta: 0:03:02 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1686 (0.3444) acc1: 99.7396 (85.3162) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1814 data: 0.0875 max mem: 3537 +Test: [1450/2387] eta: 0:03:00 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1737 (0.3432) acc1: 99.7396 (85.4088) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1847 data: 0.0913 max mem: 3537 +Test: [1460/2387] eta: 0:02:58 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1497 (0.3419) acc1: 99.4792 (85.5049) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1891 data: 0.0976 max mem: 3537 +Test: [1470/2387] eta: 0:02:56 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1326 (0.3404) acc1: 100.0000 (85.6031) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1825 data: 0.0894 max mem: 3537 +Test: [1480/2387] eta: 0:02:54 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1235 (0.3390) acc1: 100.0000 (85.6987) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1786 data: 0.0826 max mem: 3537 +Test: [1490/2387] eta: 0:02:52 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1297 (0.3376) acc1: 100.0000 (85.7941) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1778 data: 0.0833 max mem: 3537 +Test: [1500/2387] eta: 0:02:50 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1312 (0.3363) acc1: 100.0000 (85.8868) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1829 data: 0.0910 max mem: 3537 +Test: [1510/2387] eta: 0:02:48 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1379 (0.3350) acc1: 100.0000 (85.9802) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1791 data: 0.0864 max mem: 3537 +Test: [1520/2387] eta: 0:02:46 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1493 (0.3342) acc1: 100.0000 (86.0620) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1868 data: 0.0933 max mem: 3537 +Test: [1530/2387] eta: 0:02:44 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.2323 (0.3336) acc1: 98.1771 (86.1316) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1754 data: 0.0807 max mem: 3537 +Test: [1540/2387] eta: 0:02:42 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.2195 (0.3328) acc1: 98.6979 (86.2131) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1728 data: 0.0787 max mem: 3537 +Test: [1550/2387] eta: 0:02:40 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.2149 (0.3320) acc1: 99.2188 (86.2943) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1974 data: 0.1047 max mem: 3537 +Test: [1560/2387] eta: 0:02:38 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.2210 (0.3314) acc1: 98.6979 (86.3674) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1834 data: 0.0924 max mem: 3537 +Test: [1570/2387] eta: 0:02:36 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1963 (0.3304) acc1: 99.7396 (86.4520) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1857 data: 0.0931 max mem: 3537 +Test: [1580/2387] eta: 0:02:34 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1752 (0.3294) acc1: 99.7396 (86.5361) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1834 data: 0.0903 max mem: 3538 +Test: [1590/2387] eta: 0:02:32 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1610 (0.3283) acc1: 99.7396 (86.6197) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1725 data: 0.0811 max mem: 3538 +Test: [1600/2387] eta: 0:02:30 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1552 (0.3274) acc1: 100.0000 (86.6789) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1735 data: 0.0799 max mem: 3538 +Test: [1610/2387] eta: 0:02:28 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1552 (0.3263) acc1: 100.0000 (86.7611) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1777 data: 0.0829 max mem: 3538 +Test: [1620/2387] eta: 0:02:26 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1572 (0.3253) acc1: 100.0000 (86.8394) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1805 data: 0.0881 max mem: 3538 +Test: [1630/2387] eta: 0:02:24 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1612 (0.3244) acc1: 99.7396 (86.9137) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1764 data: 0.0857 max mem: 3538 +Test: [1640/2387] eta: 0:02:22 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1809 (0.3236) acc1: 99.4792 (86.9855) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1776 data: 0.0837 max mem: 3538 +Test: [1650/2387] eta: 0:02:20 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1797 (0.3227) acc1: 99.7396 (87.0607) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1850 data: 0.0906 max mem: 3538 +Test: [1660/2387] eta: 0:02:18 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1780 (0.3218) acc1: 99.7396 (87.1333) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1766 data: 0.0842 max mem: 3538 +Test: [1670/2387] eta: 0:02:16 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1749 (0.3210) acc1: 99.4792 (87.2056) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1737 data: 0.0793 max mem: 3538 +Test: [1680/2387] eta: 0:02:14 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1734 (0.3202) acc1: 99.7396 (87.2692) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0901 max mem: 3538 +Test: [1690/2387] eta: 0:02:12 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1791 (0.3195) acc1: 99.7396 (87.3374) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1818 data: 0.0870 max mem: 3538 +Test: [1700/2387] eta: 0:02:10 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1705 (0.3185) acc1: 99.7396 (87.4088) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1815 data: 0.0871 max mem: 3538 +Test: [1710/2387] eta: 0:02:09 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1374 (0.3174) acc1: 99.4792 (87.4802) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1918 data: 0.0974 max mem: 3538 +Test: [1720/2387] eta: 0:02:07 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1174 (0.3162) acc1: 99.7396 (87.5516) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1987 data: 0.1047 max mem: 3538 +Test: [1730/2387] eta: 0:02:05 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1155 (0.3151) acc1: 99.7396 (87.6207) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1848 data: 0.0924 max mem: 3538 +Test: [1740/2387] eta: 0:02:03 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1162 (0.3139) acc1: 99.7396 (87.6901) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1769 data: 0.0852 max mem: 3538 +Test: [1750/2387] eta: 0:02:01 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1162 (0.3128) acc1: 100.0000 (87.7592) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1829 data: 0.0905 max mem: 3538 +Test: [1760/2387] eta: 0:01:59 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1171 (0.3117) acc1: 100.0000 (87.8287) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1774 data: 0.0843 max mem: 3538 +Test: [1770/2387] eta: 0:01:57 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1266 (0.3107) acc1: 100.0000 (87.8908) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1725 data: 0.0788 max mem: 3538 +Test: [1780/2387] eta: 0:01:55 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1294 (0.3097) acc1: 99.7396 (87.9584) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0810 max mem: 3538 +Test: [1790/2387] eta: 0:01:53 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1212 (0.3086) acc1: 100.0000 (88.0240) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1756 data: 0.0827 max mem: 3538 +Test: [1800/2387] eta: 0:01:51 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1203 (0.3076) acc1: 99.7396 (88.0868) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1758 data: 0.0822 max mem: 3538 +Test: [1810/2387] eta: 0:01:49 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1200 (0.3065) acc1: 99.7396 (88.1505) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1807 data: 0.0861 max mem: 3539 +Test: [1820/2387] eta: 0:01:47 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1155 (0.3054) acc1: 100.0000 (88.2152) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1770 data: 0.0833 max mem: 3539 +Test: [1830/2387] eta: 0:01:45 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1155 (0.3044) acc1: 100.0000 (88.2795) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1693 data: 0.0772 max mem: 3539 +Test: [1840/2387] eta: 0:01:43 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1288 (0.3035) acc1: 100.0000 (88.3432) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1676 data: 0.0761 max mem: 3539 +Test: [1850/2387] eta: 0:01:41 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1306 (0.3025) acc1: 100.0000 (88.4062) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1743 data: 0.0807 max mem: 3539 +Test: [1860/2387] eta: 0:01:39 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1306 (0.3016) acc1: 100.0000 (88.4685) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1873 data: 0.0928 max mem: 3539 +Test: [1870/2387] eta: 0:01:38 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1309 (0.3007) acc1: 100.0000 (88.5301) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1890 data: 0.0972 max mem: 3539 +Test: [1880/2387] eta: 0:01:36 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1303 (0.2998) acc1: 100.0000 (88.5911) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1839 data: 0.0915 max mem: 3539 +Test: [1890/2387] eta: 0:01:34 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1305 (0.2989) acc1: 100.0000 (88.6514) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1852 data: 0.0931 max mem: 3539 +Test: [1900/2387] eta: 0:01:32 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1321 (0.2981) acc1: 100.0000 (88.7111) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1885 data: 0.0945 max mem: 3539 +Test: [1910/2387] eta: 0:01:30 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1515 (0.2973) acc1: 100.0000 (88.7698) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1920 data: 0.0952 max mem: 3539 +Test: [1920/2387] eta: 0:01:28 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1567 (0.2968) acc1: 100.0000 (88.8240) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1847 data: 0.0915 max mem: 3539 +Test: [1930/2387] eta: 0:01:26 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1462 (0.2959) acc1: 100.0000 (88.8819) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1763 data: 0.0837 max mem: 3539 +Test: [1940/2387] eta: 0:01:24 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1440 (0.2955) acc1: 100.0000 (88.9024) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0858 max mem: 3539 +Test: [1950/2387] eta: 0:01:22 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1565 (0.2948) acc1: 100.0000 (88.9593) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1795 data: 0.0872 max mem: 3539 +Test: [1960/2387] eta: 0:01:20 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1566 (0.2942) acc1: 100.0000 (89.0079) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1759 data: 0.0840 max mem: 3539 +Test: [1970/2387] eta: 0:01:18 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1868 (0.2939) acc1: 100.0000 (89.0550) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1741 data: 0.0818 max mem: 3539 +Test: [1980/2387] eta: 0:01:16 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1891 (0.2934) acc1: 99.2188 (89.0958) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1746 data: 0.0831 max mem: 3539 +Test: [1990/2387] eta: 0:01:15 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1729 (0.2928) acc1: 98.9583 (89.1462) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1788 data: 0.0877 max mem: 3539 +Test: [2000/2387] eta: 0:01:13 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1735 (0.2922) acc1: 98.9583 (89.1960) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1812 data: 0.0889 max mem: 3539 +Test: [2010/2387] eta: 0:01:11 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1760 (0.2917) acc1: 99.2188 (89.2455) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1789 data: 0.0861 max mem: 3539 +Test: [2020/2387] eta: 0:01:09 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1760 (0.2911) acc1: 99.2188 (89.2944) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1789 data: 0.0877 max mem: 3539 +Test: [2030/2387] eta: 0:01:07 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1743 (0.2905) acc1: 99.2188 (89.3425) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1896 data: 0.0990 max mem: 3540 +Test: [2040/2387] eta: 0:01:05 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1743 (0.2900) acc1: 98.9583 (89.3904) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2253 data: 0.1342 max mem: 3540 +Test: [2050/2387] eta: 0:01:03 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1738 (0.2894) acc1: 99.2188 (89.4393) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2145 data: 0.1218 max mem: 3540 +Test: [2060/2387] eta: 0:01:01 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1680 (0.2888) acc1: 99.4792 (89.4882) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2079 data: 0.1130 max mem: 3540 +Test: [2070/2387] eta: 0:01:00 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1630 (0.2882) acc1: 99.7396 (89.5373) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2058 data: 0.1112 max mem: 3540 +Test: [2080/2387] eta: 0:00:58 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1620 (0.2876) acc1: 99.4792 (89.5855) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1748 data: 0.0808 max mem: 3540 +Test: [2090/2387] eta: 0:00:56 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1613 (0.2870) acc1: 99.4792 (89.6336) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1895 data: 0.0969 max mem: 3540 +Test: [2100/2387] eta: 0:00:54 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1611 (0.2864) acc1: 99.7396 (89.6813) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1915 data: 0.0985 max mem: 3540 +Test: [2110/2387] eta: 0:00:52 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1629 (0.2858) acc1: 99.7396 (89.7282) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1815 data: 0.0879 max mem: 3540 +Test: [2120/2387] eta: 0:00:50 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1651 (0.2855) acc1: 99.4792 (89.7722) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1801 data: 0.0881 max mem: 3540 +Test: [2130/2387] eta: 0:00:48 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1979 (0.2850) acc1: 99.7396 (89.8183) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1820 data: 0.0878 max mem: 3540 +Test: [2140/2387] eta: 0:00:46 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.2079 (0.2848) acc1: 99.2188 (89.8435) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1771 data: 0.0832 max mem: 3540 +Test: [2150/2387] eta: 0:00:44 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.2103 (0.2845) acc1: 99.4792 (89.8890) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1751 data: 0.0834 max mem: 3540 +Test: [2160/2387] eta: 0:00:42 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1956 (0.2841) acc1: 99.7396 (89.9311) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1859 data: 0.0939 max mem: 3540 +Test: [2170/2387] eta: 0:00:41 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.2154 (0.2839) acc1: 99.7396 (89.9723) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1796 data: 0.0882 max mem: 3540 +Test: [2180/2387] eta: 0:00:39 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.2465 (0.2854) acc1: 96.8750 (89.8748) acc5: 100.0000 (100.0000) acc1_neg: 27.344 acc5_neg: 100.000 time: 0.1668 data: 0.0740 max mem: 3540 +Test: [2190/2387] eta: 0:00:37 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.2757 (0.2858) acc1: 92.4479 (89.8569) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1762 data: 0.0831 max mem: 3540 +Test: [2200/2387] eta: 0:00:35 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1677 (0.2852) acc1: 98.9583 (89.9018) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1757 data: 0.0845 max mem: 3540 +Test: [2210/2387] eta: 0:00:33 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1510 (0.2846) acc1: 100.0000 (89.9473) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1782 data: 0.0853 max mem: 3540 +Test: [2220/2387] eta: 0:00:31 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1510 (0.2840) acc1: 100.0000 (89.9898) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1861 data: 0.0906 max mem: 3540 +Test: [2230/2387] eta: 0:00:29 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1510 (0.2834) acc1: 100.0000 (90.0335) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1755 data: 0.0795 max mem: 3540 +Test: [2240/2387] eta: 0:00:27 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1475 (0.2828) acc1: 100.0000 (90.0759) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1802 data: 0.0867 max mem: 3540 +Test: [2250/2387] eta: 0:00:25 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1582 (0.2823) acc1: 100.0000 (90.1183) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1903 data: 0.0993 max mem: 3540 +Test: [2260/2387] eta: 0:00:23 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1647 (0.2818) acc1: 100.0000 (90.1618) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1834 data: 0.0923 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1640 (0.2812) acc1: 100.0000 (90.2048) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1886 data: 0.0971 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1703 (0.2809) acc1: 100.0000 (90.2306) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1950 data: 0.1035 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1673 (0.2804) acc1: 100.0000 (90.2732) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1861 data: 0.0942 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1673 (0.2799) acc1: 100.0000 (90.3130) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1956 data: 0.1041 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1714 (0.2795) acc1: 100.0000 (90.3534) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1879 data: 0.0961 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1824 (0.2791) acc1: 99.4792 (90.3871) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1712 data: 0.0792 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1990 (0.2788) acc1: 99.7396 (90.4258) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1706 data: 0.0794 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1988 (0.2784) acc1: 100.0000 (90.4645) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1649 data: 0.0742 max mem: 3541 +Test: [2350/2387] eta: 0:00:06 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1857 (0.2783) acc1: 100.0000 (90.4871) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1698 data: 0.0794 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1900 (0.2780) acc1: 100.0000 (90.5251) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1625 data: 0.0720 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.1933 (0.2777) acc1: 100.0000 (90.5607) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1454 data: 0.0551 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.2084 (0.2774) acc1: 99.2188 (90.5920) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1364 data: 0.0460 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 0.457 acc5_pos: 100.000 loss: 0.2113 (0.2771) acc1: 99.2188 (90.6102) acc5: 100.0000 (100.0000) acc1_neg: 97.674 acc5_neg: 100.000 time: 0.1264 data: 0.0388 max mem: 3541 +Test: Total time: 0:07:28 (0.1878 s / it) +* Overall Acc@1 90.610 Acc@5 100.000 + target == real Acc@1 7.109 Acc@5 100.000 + target == fake Acc@1 96.474 Acc@5 100.000 + loss 0.277 + PR-AUC 0.9859 ROC-AUC 0.8243 EER 24.31% (thr=0.8062) +Accuracy of the network on the 916267 test images: 90.6% +[rank0]:[W726 05:41:11.176038104 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:41:30.540000 1192920 torch/distributed/run.py:766] +W0726 05:41:30.540000 1192920 torch/distributed/run.py:766] ***************************************** +W0726 05:41:30.540000 1192920 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:41:30.540000 1192920 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:41:57.193553558 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:41:57.494643236 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:41:57.501451695 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:41:58.976271995 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +[rank0]:[W726 05:41:58.060747404 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:41:58.065938745 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:41:58.076907977 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank6]:[W726 05:41:58.076926391 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n22_seed8_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed8_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:03:37 acc1_pos: 10.677 acc5_pos: 100.000 loss: 1.2030 (1.2030) acc1: 10.6771 (10.6771) acc5: 100.0000 (100.0000) time: 7.3841 data: 5.7633 max mem: 3531 +Test: [ 10/517] eta: 0:06:23 acc1_pos: 4.688 acc5_pos: 100.000 loss: 1.3181 (1.2816) acc1: 6.5104 (8.8542) acc5: 100.0000 (100.0000) time: 0.7558 data: 0.5243 max mem: 3531 +Test: [ 20/517] eta: 0:03:40 acc1_pos: 1.042 acc5_pos: 100.000 loss: 1.2997 (1.2820) acc1: 5.9896 (7.7133) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:02:56 acc1_pos: 0.781 acc5_pos: 100.000 loss: 1.2997 (1.3023) acc1: 5.2083 (7.3757) acc5: 100.0000 (100.0000) time: 0.1472 data: 0.0505 max mem: 3531 +Test: [ 40/517] eta: 0:02:36 acc1_pos: 6.771 acc5_pos: 100.000 loss: 1.3764 (1.3280) acc1: 1.8229 (6.3326) acc5: 100.0000 (100.0000) time: 0.2049 data: 0.1126 max mem: 3531 +Test: [ 50/517] eta: 0:02:23 acc1_pos: 16.514 acc5_pos: 100.000 loss: 1.3492 (1.2670) acc1: 4.1667 (10.9324) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2182 data: 0.1274 max mem: 3531 +Test: [ 60/517] eta: 0:02:12 acc1_pos: 16.514 acc5_pos: 100.000 loss: 0.3434 (1.1047) acc1: 97.3958 (25.3330) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2167 data: 0.1243 max mem: 3531 +Test: [ 70/517] eta: 0:02:05 acc1_pos: 16.514 acc5_pos: 100.000 loss: 0.2559 (0.9844) acc1: 99.7396 (35.8128) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2163 data: 0.1244 max mem: 3531 +Test: [ 80/517] eta: 0:01:59 acc1_pos: 16.514 acc5_pos: 100.000 loss: 0.2561 (0.8971) acc1: 100.0000 (43.6664) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2202 data: 0.1287 max mem: 3531 +Test: [ 90/517] eta: 0:01:55 acc1_pos: 16.514 acc5_pos: 100.000 loss: 0.2754 (0.8292) acc1: 99.7396 (49.7653) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2298 data: 0.1347 max mem: 3531 +Test: [100/517] eta: 0:01:50 acc1_pos: 16.514 acc5_pos: 100.000 loss: 0.2644 (0.7702) acc1: 99.7396 (54.7262) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2329 data: 0.1365 max mem: 3531 +Test: [110/517] eta: 0:01:47 acc1_pos: 16.514 acc5_pos: 100.000 loss: 0.2140 (0.7201) acc1: 100.0000 (58.7932) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2330 data: 0.1347 max mem: 3531 +Test: [120/517] eta: 0:01:43 acc1_pos: 16.514 acc5_pos: 100.000 loss: 0.2420 (0.6815) acc1: 99.7396 (62.1556) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2349 data: 0.1366 max mem: 3531 +Test: [130/517] eta: 0:01:40 acc1_pos: 16.514 acc5_pos: 100.000 loss: 0.2494 (0.6485) acc1: 99.4792 (64.9909) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2349 data: 0.1405 max mem: 3531 +Test: [140/517] eta: 0:01:37 acc1_pos: 16.514 acc5_pos: 100.000 loss: 0.2417 (0.6180) acc1: 99.7396 (67.4627) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2449 data: 0.1500 max mem: 3531 +Test: [150/517] eta: 0:01:34 acc1_pos: 16.514 acc5_pos: 100.000 loss: 0.2147 (0.5918) acc1: 100.0000 (69.6158) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2499 data: 0.1526 max mem: 3531 +Test: [160/517] eta: 0:01:31 acc1_pos: 16.514 acc5_pos: 100.000 loss: 0.2424 (0.5710) acc1: 100.0000 (71.4917) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2499 data: 0.1546 max mem: 3531 +Test: [170/517] eta: 0:01:28 acc1_pos: 16.514 acc5_pos: 100.000 loss: 0.2520 (0.5525) acc1: 100.0000 (73.1390) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2349 data: 0.1411 max mem: 3531 +Test: [180/517] eta: 0:01:25 acc1_pos: 16.514 acc5_pos: 100.000 loss: 0.2464 (0.5352) acc1: 99.7396 (74.6000) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2299 data: 0.1311 max mem: 3531 +Test: [190/517] eta: 0:01:22 acc1_pos: 16.514 acc5_pos: 100.000 loss: 0.2233 (0.5183) acc1: 99.7396 (75.9244) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2349 data: 0.1349 max mem: 3531 +Test: [200/517] eta: 0:01:19 acc1_pos: 16.514 acc5_pos: 100.000 loss: 0.2220 (0.5038) acc1: 100.0000 (77.1067) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2249 data: 0.1293 max mem: 3531 +Test: [210/517] eta: 0:01:17 acc1_pos: 16.514 acc5_pos: 100.000 loss: 0.2485 (0.4939) acc1: 98.9583 (78.0608) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2399 data: 0.1414 max mem: 3532 +Test: [220/517] eta: 0:01:14 acc1_pos: 16.514 acc5_pos: 100.000 loss: 0.3065 (0.4858) acc1: 95.3125 (78.8026) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2449 data: 0.1438 max mem: 3532 +Test: [230/517] eta: 0:01:11 acc1_pos: 16.514 acc5_pos: 100.000 loss: 0.2897 (0.4765) acc1: 96.3542 (79.6672) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2349 data: 0.1366 max mem: 3532 +Test: [240/517] eta: 0:01:09 acc1_pos: 16.514 acc5_pos: 100.000 loss: 0.2472 (0.4665) acc1: 100.0000 (80.5087) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2449 data: 0.1497 max mem: 3532 +Test: [250/517] eta: 0:01:06 acc1_pos: 16.514 acc5_pos: 100.000 loss: 0.2234 (0.4570) acc1: 99.7396 (81.2749) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2499 data: 0.1521 max mem: 3532 +Test: [260/517] eta: 0:01:04 acc1_pos: 16.514 acc5_pos: 100.000 loss: 0.2218 (0.4484) acc1: 99.7396 (81.9824) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2449 data: 0.1450 max mem: 3532 +Test: [270/517] eta: 0:01:01 acc1_pos: 16.514 acc5_pos: 100.000 loss: 0.2405 (0.4408) acc1: 99.7396 (82.6443) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2349 data: 0.1390 max mem: 3532 +Test: [280/517] eta: 0:00:58 acc1_pos: 16.514 acc5_pos: 100.000 loss: 0.2346 (0.4334) acc1: 99.7396 (83.2546) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2254 data: 0.1327 max mem: 3532 +Test: [290/517] eta: 0:00:56 acc1_pos: 16.514 acc5_pos: 100.000 loss: 0.2305 (0.4271) acc1: 99.7396 (83.8220) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2399 data: 0.1475 max mem: 3532 +Test: [300/517] eta: 0:00:53 acc1_pos: 16.514 acc5_pos: 100.000 loss: 0.2388 (0.4212) acc1: 99.7396 (84.3516) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2544 data: 0.1600 max mem: 3532 +Test: [310/517] eta: 0:00:51 acc1_pos: 16.514 acc5_pos: 100.000 loss: 0.2609 (0.4166) acc1: 99.7396 (84.8397) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2549 data: 0.1588 max mem: 3532 +Test: [320/517] eta: 0:00:48 acc1_pos: 16.514 acc5_pos: 100.000 loss: 0.2671 (0.4116) acc1: 99.7396 (85.3007) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2507 data: 0.1556 max mem: 3532 +Test: [330/517] eta: 0:00:46 acc1_pos: 16.514 acc5_pos: 100.000 loss: 0.2393 (0.4061) acc1: 99.7396 (85.7290) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2413 data: 0.1476 max mem: 3532 +Test: [340/517] eta: 0:00:43 acc1_pos: 16.514 acc5_pos: 100.000 loss: 0.2150 (0.4003) acc1: 99.7396 (86.1406) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2441 data: 0.1474 max mem: 3532 +Test: [350/517] eta: 0:00:41 acc1_pos: 16.514 acc5_pos: 100.000 loss: 0.2224 (0.3957) acc1: 100.0000 (86.5325) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2487 data: 0.1529 max mem: 3532 +Test: [360/517] eta: 0:00:38 acc1_pos: 16.514 acc5_pos: 100.000 loss: 0.2387 (0.3915) acc1: 100.0000 (86.9027) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2449 data: 0.1523 max mem: 3532 +Test: [370/517] eta: 0:00:36 acc1_pos: 16.514 acc5_pos: 100.000 loss: 0.2233 (0.3863) acc1: 100.0000 (87.2543) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2361 data: 0.1442 max mem: 3532 +Test: [380/517] eta: 0:00:33 acc1_pos: 16.514 acc5_pos: 100.000 loss: 0.1958 (0.3815) acc1: 100.0000 (87.5868) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2399 data: 0.1478 max mem: 3532 +Test: [390/517] eta: 0:00:31 acc1_pos: 16.514 acc5_pos: 100.000 loss: 0.2079 (0.3772) acc1: 100.0000 (87.9036) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2436 data: 0.1506 max mem: 3532 +Test: [400/517] eta: 0:00:28 acc1_pos: 16.514 acc5_pos: 100.000 loss: 0.2024 (0.3726) acc1: 100.0000 (88.2020) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2407 data: 0.1477 max mem: 3532 +Test: [410/517] eta: 0:00:26 acc1_pos: 16.514 acc5_pos: 100.000 loss: 0.2132 (0.3691) acc1: 100.0000 (88.4891) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2449 data: 0.1514 max mem: 3532 +Test: [420/517] eta: 0:00:24 acc1_pos: 16.514 acc5_pos: 100.000 loss: 0.2187 (0.3656) acc1: 100.0000 (88.7625) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2542 data: 0.1597 max mem: 3532 +Test: [430/517] eta: 0:00:21 acc1_pos: 16.514 acc5_pos: 100.000 loss: 0.2313 (0.3631) acc1: 100.0000 (89.0190) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2511 data: 0.1577 max mem: 3532 +Test: [440/517] eta: 0:00:19 acc1_pos: 16.514 acc5_pos: 100.000 loss: 0.2323 (0.3598) acc1: 100.0000 (89.2662) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2412 data: 0.1481 max mem: 3533 +Test: [450/517] eta: 0:00:16 acc1_pos: 16.514 acc5_pos: 100.000 loss: 0.2232 (0.3569) acc1: 100.0000 (89.5042) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2393 data: 0.1461 max mem: 3533 +Test: [460/517] eta: 0:00:14 acc1_pos: 16.514 acc5_pos: 100.000 loss: 0.2256 (0.3543) acc1: 100.0000 (89.7150) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2486 data: 0.1562 max mem: 3533 +Test: [470/517] eta: 0:00:11 acc1_pos: 16.514 acc5_pos: 100.000 loss: 0.2170 (0.3512) acc1: 100.0000 (89.9317) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2501 data: 0.1575 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 16.514 acc5_pos: 100.000 loss: 0.2082 (0.3484) acc1: 100.0000 (90.1361) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2360 data: 0.1446 max mem: 3533 +Test: [490/517] eta: 0:00:06 acc1_pos: 16.514 acc5_pos: 100.000 loss: 0.2175 (0.3462) acc1: 100.0000 (90.3253) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2396 data: 0.1495 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 16.514 acc5_pos: 100.000 loss: 0.2509 (0.3442) acc1: 100.0000 (90.5184) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2542 data: 0.1637 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 16.514 acc5_pos: 100.000 loss: 0.2560 (0.3425) acc1: 100.0000 (90.6795) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2447 data: 0.1536 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 16.514 acc5_pos: 100.000 loss: 0.2188 (0.3409) acc1: 100.0000 (90.7815) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2167 data: 0.1227 max mem: 3533 +Test: Total time: 0:02:07 (0.2467 s / it) +* Overall Acc@1 90.781 Acc@5 100.000 + target == real Acc@1 6.499 Acc@5 100.000 + target == fake Acc@1 99.525 Acc@5 100.000 + loss 0.341 + PR-AUC 0.9532 ROC-AUC 0.7213 EER 33.66% (thr=0.7769) +Accuracy of the network on the 198427 test images: 90.8% +[rank0]:[W726 05:44:22.449676876 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 05:44:42.795000 1193483 torch/distributed/run.py:766] +W0726 05:44:42.795000 1193483 torch/distributed/run.py:766] ***************************************** +W0726 05:44:42.795000 1193483 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:44:42.795000 1193483 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:45:10.903094556 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:45:11.924622302 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:45:11.456679806 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:45:11.569524662 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:45:11.598873431 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 05:45:11.620329534 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W726 05:45:11.620378028 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:45:11.627659957 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n22_seed8_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed8_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:43:29 acc1_pos: 65.625 acc5_pos: 100.000 loss: 0.6116 (0.6116) acc1: 65.6250 (65.6250) acc5: 100.0000 (100.0000) time: 6.7077 data: 4.3625 max mem: 3531 +Test: [ 10/389] eta: 0:04:23 acc1_pos: 89.323 acc5_pos: 100.000 loss: 0.6519 (0.6767) acc1: 57.5521 (54.7112) acc5: 100.0000 (100.0000) time: 0.6959 data: 0.3970 max mem: 3531 +Test: [ 20/389] eta: 0:02:32 acc1_pos: 0.781 acc5_pos: 100.000 loss: 0.6574 (0.7935) acc1: 48.4375 (46.2674) acc5: 100.0000 (100.0000) time: 0.0973 data: 0.0004 max mem: 3531 +Test: [ 30/389] eta: 0:02:19 acc1_pos: 35.417 acc5_pos: 100.000 loss: 0.9294 (0.8469) acc1: 17.7083 (36.2819) acc5: 100.0000 (100.0000) time: 0.2215 data: 0.1268 max mem: 3531 +Test: [ 40/389] eta: 0:02:02 acc1_pos: 58.073 acc5_pos: 100.000 loss: 0.8241 (0.8121) acc1: 27.8646 (40.4408) acc5: 100.0000 (100.0000) time: 0.2862 data: 0.1946 max mem: 3531 +Test: [ 50/389] eta: 0:01:49 acc1_pos: 55.208 acc5_pos: 100.000 loss: 0.7023 (0.7796) acc1: 55.2083 (45.0725) acc5: 100.0000 (100.0000) time: 0.2186 data: 0.1266 max mem: 3531 +Test: [ 60/389] eta: 0:01:38 acc1_pos: 67.969 acc5_pos: 100.000 loss: 0.7124 (0.7681) acc1: 52.6042 (46.4908) acc5: 100.0000 (100.0000) time: 0.1986 data: 0.1072 max mem: 3531 +Test: [ 70/389] eta: 0:01:30 acc1_pos: 71.615 acc5_pos: 100.000 loss: 0.7126 (0.7458) acc1: 52.6042 (48.9693) acc5: 100.0000 (100.0000) time: 0.1896 data: 0.0942 max mem: 3531 +Test: [ 80/389] eta: 0:01:21 acc1_pos: 90.885 acc5_pos: 100.000 loss: 0.7305 (0.7442) acc1: 48.9583 (49.1898) acc5: 100.0000 (100.0000) time: 0.1516 data: 0.0543 max mem: 3531 +Test: [ 90/389] eta: 0:01:15 acc1_pos: 63.542 acc5_pos: 100.000 loss: 0.5555 (0.7190) acc1: 63.5417 (52.1577) acc5: 100.0000 (100.0000) time: 0.1371 data: 0.0424 max mem: 3531 +Test: [100/389] eta: 0:01:11 acc1_pos: 26.823 acc5_pos: 100.000 loss: 0.6344 (0.7228) acc1: 62.7604 (51.6347) acc5: 100.0000 (100.0000) time: 0.1732 data: 0.0794 max mem: 3531 +Test: [110/389] eta: 0:01:06 acc1_pos: 99.479 acc5_pos: 100.000 loss: 0.7366 (0.7307) acc1: 54.4271 (50.9478) acc5: 100.0000 (100.0000) time: 0.1777 data: 0.0824 max mem: 3531 +Test: [120/389] eta: 0:01:02 acc1_pos: 74.479 acc5_pos: 100.000 loss: 0.6570 (0.7207) acc1: 58.0729 (52.2921) acc5: 100.0000 (100.0000) time: 0.1549 data: 0.0561 max mem: 3531 +Test: [130/389] eta: 0:00:57 acc1_pos: 57.055 acc5_pos: 100.000 loss: 0.6124 (0.7099) acc1: 63.5417 (53.6896) acc5: 100.0000 (100.0000) acc1_neg: 77.083 acc5_neg: 100.000 time: 0.1349 data: 0.0359 max mem: 3531 +Test: [140/389] eta: 0:00:54 acc1_pos: 57.055 acc5_pos: 100.000 loss: 0.5369 (0.6986) acc1: 74.7396 (55.1825) acc5: 100.0000 (100.0000) acc1_neg: 29.948 acc5_neg: 100.000 time: 0.1399 data: 0.0453 max mem: 3531 +Test: [150/389] eta: 0:00:51 acc1_pos: 57.055 acc5_pos: 100.000 loss: 0.5244 (0.6974) acc1: 70.0521 (55.1807) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1599 data: 0.0665 max mem: 3531 +Test: [160/389] eta: 0:00:48 acc1_pos: 57.055 acc5_pos: 100.000 loss: 0.4383 (0.6802) acc1: 78.6458 (57.0943) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1516 data: 0.0588 max mem: 3531 +Test: [170/389] eta: 0:00:45 acc1_pos: 57.055 acc5_pos: 100.000 loss: 0.2763 (0.6659) acc1: 95.0521 (58.0881) acc5: 100.0000 (100.0000) acc1_neg: 26.042 acc5_neg: 100.000 time: 0.1452 data: 0.0525 max mem: 3531 +Test: [180/389] eta: 0:00:42 acc1_pos: 57.055 acc5_pos: 100.000 loss: 0.3272 (0.6550) acc1: 79.6875 (59.3707) acc5: 100.0000 (100.0000) acc1_neg: 74.740 acc5_neg: 100.000 time: 0.1632 data: 0.0690 max mem: 3531 +Test: [190/389] eta: 0:00:40 acc1_pos: 57.055 acc5_pos: 100.000 loss: 0.5242 (0.6519) acc1: 76.3021 (59.8358) acc5: 100.0000 (100.0000) acc1_neg: 53.906 acc5_neg: 100.000 time: 0.1646 data: 0.0719 max mem: 3531 +Test: [200/389] eta: 0:00:37 acc1_pos: 57.055 acc5_pos: 100.000 loss: 0.4458 (0.6432) acc1: 86.9792 (61.2044) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1657 data: 0.0734 max mem: 3531 +Test: [210/389] eta: 0:00:35 acc1_pos: 57.055 acc5_pos: 100.000 loss: 0.3633 (0.6285) acc1: 97.6562 (62.9159) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1605 data: 0.0683 max mem: 3532 +Test: [220/389] eta: 0:00:33 acc1_pos: 57.055 acc5_pos: 100.000 loss: 0.2983 (0.6124) acc1: 98.9583 (64.5598) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1532 data: 0.0608 max mem: 3532 +Test: [230/389] eta: 0:00:31 acc1_pos: 57.055 acc5_pos: 100.000 loss: 0.3772 (0.6083) acc1: 97.6562 (65.1064) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.1692 data: 0.0769 max mem: 3532 +Test: [240/389] eta: 0:00:28 acc1_pos: 57.055 acc5_pos: 100.000 loss: 0.5450 (0.6061) acc1: 76.8229 (65.5623) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1656 data: 0.0750 max mem: 3532 +Test: [250/389] eta: 0:00:26 acc1_pos: 57.055 acc5_pos: 100.000 loss: 0.4483 (0.5948) acc1: 95.5729 (66.8669) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1401 data: 0.0484 max mem: 3532 +Test: [260/389] eta: 0:00:24 acc1_pos: 57.055 acc5_pos: 100.000 loss: 0.3631 (0.5907) acc1: 98.9583 (67.3751) acc5: 100.0000 (100.0000) acc1_neg: 77.344 acc5_neg: 100.000 time: 0.1358 data: 0.0425 max mem: 3532 +Test: [270/389] eta: 0:00:22 acc1_pos: 57.055 acc5_pos: 100.000 loss: 0.5406 (0.5899) acc1: 80.2083 (67.5728) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.1540 data: 0.0589 max mem: 3532 +Test: [280/389] eta: 0:00:20 acc1_pos: 57.055 acc5_pos: 100.000 loss: 0.6213 (0.5927) acc1: 64.3229 (67.1430) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1612 data: 0.0655 max mem: 3532 +Test: [290/389] eta: 0:00:18 acc1_pos: 57.055 acc5_pos: 100.000 loss: 0.5812 (0.5881) acc1: 64.3229 (67.6609) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1586 data: 0.0659 max mem: 3532 +Test: [300/389] eta: 0:00:16 acc1_pos: 57.055 acc5_pos: 100.000 loss: 0.5812 (0.5928) acc1: 55.9896 (66.9167) acc5: 100.0000 (100.0000) acc1_neg: 37.760 acc5_neg: 100.000 time: 0.1707 data: 0.0786 max mem: 3532 +Test: [310/389] eta: 0:00:14 acc1_pos: 57.055 acc5_pos: 100.000 loss: 0.7968 (0.5990) acc1: 35.6771 (66.2589) acc5: 100.0000 (100.0000) acc1_neg: 36.198 acc5_neg: 100.000 time: 0.1772 data: 0.0820 max mem: 3532 +Test: [320/389] eta: 0:00:12 acc1_pos: 57.055 acc5_pos: 100.000 loss: 0.6615 (0.5978) acc1: 53.3854 (66.3787) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1767 data: 0.0816 max mem: 3532 +Test: [330/389] eta: 0:00:10 acc1_pos: 57.055 acc5_pos: 100.000 loss: 0.6046 (0.5983) acc1: 74.2188 (66.6014) acc5: 100.0000 (100.0000) acc1_neg: 77.083 acc5_neg: 100.000 time: 0.1718 data: 0.0780 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 57.055 acc5_pos: 100.000 loss: 0.4318 (0.5914) acc1: 77.3438 (67.4288) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1750 data: 0.0823 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 57.055 acc5_pos: 100.000 loss: 0.3883 (0.5983) acc1: 96.0938 (66.8232) acc5: 100.0000 (100.0000) acc1_neg: 0.781 acc5_neg: 100.000 time: 0.1899 data: 0.0982 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 57.055 acc5_pos: 100.000 loss: 0.8493 (0.6074) acc1: 33.3333 (65.6971) acc5: 100.0000 (100.0000) acc1_neg: 33.333 acc5_neg: 100.000 time: 0.1891 data: 0.0972 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 57.055 acc5_pos: 100.000 loss: 0.6231 (0.6022) acc1: 35.6771 (66.3396) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1808 data: 0.0894 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 57.055 acc5_pos: 100.000 loss: 0.4188 (0.6009) acc1: 89.5833 (66.5682) acc5: 100.0000 (100.0000) acc1_neg: 56.771 acc5_neg: 100.000 time: 0.1695 data: 0.0792 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 57.055 acc5_pos: 100.000 loss: 0.4998 (0.5993) acc1: 80.2083 (66.7423) acc5: 100.0000 (100.0000) acc1_neg: 85.246 acc5_neg: 100.000 time: 0.1617 data: 0.0682 max mem: 3532 +Test: Total time: 0:01:11 (0.1846 s / it) +* Overall Acc@1 66.742 Acc@5 100.000 + target == real Acc@1 52.341 Acc@5 100.000 + target == fake Acc@1 73.436 Acc@5 100.000 + loss 0.599 + PR-AUC 0.8390 ROC-AUC 0.7025 EER 35.42% (thr=0.5562) +Accuracy of the network on the 149358 test images: 66.7% +[rank0]:[W726 05:46:37.642761074 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:46:58.371000 1194046 torch/distributed/run.py:766] +W0726 05:46:58.371000 1194046 torch/distributed/run.py:766] ***************************************** +W0726 05:46:58.371000 1194046 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:46:58.371000 1194046 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 05:47:28.458341471 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:47:28.705189154 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:47:29.128361311 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:47:29.355690341 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:47:29.472966021 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:47:29.759847105 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank4]:[W726 05:47:29.768062525 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:47:30.945997576 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n22_seed8_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed8_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:42:19 acc1_pos: 82.812 acc5_pos: 100.000 loss: 0.3823 (0.3823) acc1: 82.8125 (82.8125) acc5: 100.0000 (100.0000) time: 7.6260 data: 5.5597 max mem: 3531 +Test: [ 10/333] eta: 0:04:11 acc1_pos: 86.719 acc5_pos: 100.000 loss: 0.2949 (0.3032) acc1: 90.1042 (89.7017) acc5: 100.0000 (100.0000) time: 0.7795 data: 0.5058 max mem: 3531 +Test: [ 20/333] eta: 0:02:22 acc1_pos: 96.094 acc5_pos: 100.000 loss: 0.3182 (0.3165) acc1: 89.8438 (89.0129) acc5: 100.0000 (100.0000) time: 0.0974 data: 0.0004 max mem: 3531 +Test: [ 30/333] eta: 0:02:00 acc1_pos: 51.823 acc5_pos: 100.000 loss: 0.4202 (0.4765) acc1: 75.2604 (75.0420) acc5: 100.0000 (100.0000) time: 0.1861 data: 0.0894 max mem: 3531 +Test: [ 40/333] eta: 0:01:44 acc1_pos: 44.792 acc5_pos: 100.000 loss: 0.8472 (0.5724) acc1: 40.1042 (66.6667) acc5: 100.0000 (100.0000) time: 0.2556 data: 0.1615 max mem: 3531 +Test: [ 50/333] eta: 0:01:34 acc1_pos: 47.656 acc5_pos: 100.000 loss: 0.8554 (0.6302) acc1: 39.0625 (61.4481) acc5: 100.0000 (100.0000) time: 0.2387 data: 0.1427 max mem: 3531 +Test: [ 60/333] eta: 0:01:24 acc1_pos: 35.417 acc5_pos: 100.000 loss: 0.8554 (0.6672) acc1: 38.8021 (58.0260) acc5: 100.0000 (100.0000) time: 0.2108 data: 0.1140 max mem: 3531 +Test: [ 70/333] eta: 0:01:18 acc1_pos: 33.333 acc5_pos: 100.000 loss: 0.8641 (0.6938) acc1: 38.8021 (55.6191) acc5: 100.0000 (100.0000) time: 0.2060 data: 0.1111 max mem: 3531 +Test: [ 80/333] eta: 0:01:11 acc1_pos: 45.833 acc5_pos: 100.000 loss: 0.8406 (0.7111) acc1: 41.6667 (53.8645) acc5: 100.0000 (100.0000) time: 0.1984 data: 0.1064 max mem: 3531 +Test: [ 90/333] eta: 0:01:08 acc1_pos: 50.260 acc5_pos: 100.000 loss: 0.8019 (0.7223) acc1: 44.7917 (52.7845) acc5: 100.0000 (100.0000) time: 0.2238 data: 0.1328 max mem: 3531 +Test: [100/333] eta: 0:01:02 acc1_pos: 47.396 acc5_pos: 100.000 loss: 0.8061 (0.7341) acc1: 44.7917 (51.7507) acc5: 100.0000 (100.0000) time: 0.2179 data: 0.1244 max mem: 3531 +Test: [110/333] eta: 0:00:57 acc1_pos: 43.490 acc5_pos: 100.000 loss: 0.8234 (0.7440) acc1: 44.0104 (50.8986) acc5: 100.0000 (100.0000) time: 0.1556 data: 0.0622 max mem: 3531 +Test: [120/333] eta: 0:00:55 acc1_pos: 50.781 acc5_pos: 100.000 loss: 0.8233 (0.7497) acc1: 44.0104 (50.4756) acc5: 100.0000 (100.0000) time: 0.2219 data: 0.1284 max mem: 3531 +Test: [130/333] eta: 0:00:52 acc1_pos: 57.812 acc5_pos: 100.000 loss: 0.7656 (0.7534) acc1: 47.9167 (50.2326) acc5: 100.0000 (100.0000) time: 0.2412 data: 0.1472 max mem: 3531 +Test: [140/333] eta: 0:00:48 acc1_pos: 64.583 acc5_pos: 100.000 loss: 0.7761 (0.7581) acc1: 45.3125 (49.8319) acc5: 100.0000 (100.0000) time: 0.1803 data: 0.0888 max mem: 3531 +Test: [150/333] eta: 0:00:45 acc1_pos: 41.406 acc5_pos: 100.000 loss: 0.8251 (0.7634) acc1: 41.6667 (49.3498) acc5: 100.0000 (100.0000) time: 0.1779 data: 0.0862 max mem: 3531 +Test: [160/333] eta: 0:00:42 acc1_pos: 40.132 acc5_pos: 100.000 loss: 0.8165 (0.7653) acc1: 44.7917 (49.3498) acc5: 100.0000 (100.0000) acc1_neg: 63.802 acc5_neg: 100.000 time: 0.1964 data: 0.1021 max mem: 3531 +Test: [170/333] eta: 0:00:39 acc1_pos: 40.132 acc5_pos: 100.000 loss: 0.7453 (0.7616) acc1: 53.9062 (50.0396) acc5: 100.0000 (100.0000) acc1_neg: 61.458 acc5_neg: 100.000 time: 0.2009 data: 0.1078 max mem: 3531 +Test: [180/333] eta: 0:00:36 acc1_pos: 40.132 acc5_pos: 100.000 loss: 0.6625 (0.7557) acc1: 63.2812 (50.8172) acc5: 100.0000 (100.0000) acc1_neg: 59.375 acc5_neg: 100.000 time: 0.2029 data: 0.1091 max mem: 3531 +Test: [190/333] eta: 0:00:33 acc1_pos: 40.132 acc5_pos: 100.000 loss: 0.6333 (0.7483) acc1: 64.5833 (51.7057) acc5: 100.0000 (100.0000) acc1_neg: 65.625 acc5_neg: 100.000 time: 0.1995 data: 0.1049 max mem: 3531 +Test: [200/333] eta: 0:00:31 acc1_pos: 40.132 acc5_pos: 100.000 loss: 0.6043 (0.7404) acc1: 69.0104 (52.6599) acc5: 100.0000 (100.0000) acc1_neg: 70.833 acc5_neg: 100.000 time: 0.2103 data: 0.1157 max mem: 3531 +Test: [210/333] eta: 0:00:28 acc1_pos: 40.132 acc5_pos: 100.000 loss: 0.6109 (0.7368) acc1: 68.4896 (53.2299) acc5: 100.0000 (100.0000) acc1_neg: 45.833 acc5_neg: 100.000 time: 0.2006 data: 0.1055 max mem: 3532 +Test: [220/333] eta: 0:00:26 acc1_pos: 40.132 acc5_pos: 100.000 loss: 0.6661 (0.7377) acc1: 55.9896 (53.3666) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.1752 data: 0.0828 max mem: 3532 +Test: [230/333] eta: 0:00:23 acc1_pos: 40.132 acc5_pos: 100.000 loss: 0.5221 (0.7267) acc1: 76.5625 (54.5387) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1885 data: 0.0962 max mem: 3532 +Test: [240/333] eta: 0:00:21 acc1_pos: 40.132 acc5_pos: 100.000 loss: 0.5013 (0.7189) acc1: 77.3438 (55.4428) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.1748 data: 0.0823 max mem: 3532 +Test: [250/333] eta: 0:00:18 acc1_pos: 40.132 acc5_pos: 100.000 loss: 0.5582 (0.7145) acc1: 72.3958 (55.9917) acc5: 100.0000 (100.0000) acc1_neg: 58.854 acc5_neg: 100.000 time: 0.1507 data: 0.0582 max mem: 3532 +Test: [260/333] eta: 0:00:16 acc1_pos: 40.132 acc5_pos: 100.000 loss: 0.6403 (0.7131) acc1: 65.1042 (56.2590) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.1945 data: 0.1009 max mem: 3532 +Test: [270/333] eta: 0:00:13 acc1_pos: 40.132 acc5_pos: 100.000 loss: 0.5951 (0.7079) acc1: 69.5312 (56.8487) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.2086 data: 0.1143 max mem: 3532 +Test: [280/333] eta: 0:00:11 acc1_pos: 40.132 acc5_pos: 100.000 loss: 0.5821 (0.7065) acc1: 70.5729 (57.0609) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.1361 data: 0.0420 max mem: 3532 +Test: [290/333] eta: 0:00:09 acc1_pos: 40.132 acc5_pos: 100.000 loss: 0.7361 (0.7079) acc1: 56.2500 (57.0250) acc5: 100.0000 (100.0000) acc1_neg: 57.292 acc5_neg: 100.000 time: 0.1109 data: 0.0174 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 40.132 acc5_pos: 100.000 loss: 0.7512 (0.7100) acc1: 55.7292 (56.9586) acc5: 100.0000 (100.0000) acc1_neg: 55.990 acc5_neg: 100.000 time: 0.1377 data: 0.0456 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 40.132 acc5_pos: 100.000 loss: 0.6910 (0.7051) acc1: 59.6354 (57.4717) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.1354 data: 0.0454 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 40.132 acc5_pos: 100.000 loss: 0.5435 (0.7014) acc1: 73.9583 (57.8798) acc5: 100.0000 (100.0000) acc1_neg: 52.865 acc5_neg: 100.000 time: 0.1263 data: 0.0366 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 40.132 acc5_pos: 100.000 loss: 0.7747 (0.7055) acc1: 55.4688 (57.6780) acc5: 100.0000 (100.0000) acc1_neg: 52.865 acc5_neg: 100.000 time: 0.1145 data: 0.0247 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 40.132 acc5_pos: 100.000 loss: 0.8093 (0.7063) acc1: 52.8646 (57.6418) acc5: 100.0000 (100.0000) acc1_neg: 52.394 acc5_neg: 100.000 time: 0.1176 data: 0.0247 max mem: 3532 +Test: Total time: 0:01:07 (0.2035 s / it) +* Overall Acc@1 57.642 Acc@5 100.000 + target == real Acc@1 49.146 Acc@5 100.000 + target == fake Acc@1 65.204 Acc@5 100.000 + loss 0.706 + PR-AUC 0.6137 ROC-AUC 0.6035 EER 42.87% (thr=0.5513) +Accuracy of the network on the 127843 test images: 57.6% +[rank0]:[W726 05:48:47.622191720 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:49:05.949000 1194604 torch/distributed/run.py:766] +W0726 05:49:05.949000 1194604 torch/distributed/run.py:766] ***************************************** +W0726 05:49:05.949000 1194604 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:49:05.949000 1194604 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:49:34.156156368 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 0): env:// +[rank7]:[W726 05:49:34.436096939 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:49:34.778736942 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +[rank6]:[W726 05:49:34.898832584 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:49:34.905051709 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:49:35.008552701 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:49:35.083396424 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:49:35.103200400 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n22_seed8_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed8_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:19:48 acc1_pos: 91.667 acc5_pos: 100.000 loss: 0.2662 (0.2662) acc1: 91.6667 (91.6667) acc5: 100.0000 (100.0000) time: 5.0587 data: 2.5489 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/235] eta: 0:02:01 acc1_pos: 91.667 acc5_pos: 100.000 loss: 0.2662 (0.2674) acc1: 91.6667 (91.8087) acc5: 100.0000 (100.0000) time: 0.5419 data: 0.2321 max mem: 3531 +Test: [ 20/235] eta: 0:01:13 acc1_pos: 92.448 acc5_pos: 100.000 loss: 0.2629 (0.2632) acc1: 92.4479 (92.2619) acc5: 100.0000 (100.0000) time: 0.1057 data: 0.0159 max mem: 3531 +Test: [ 30/235] eta: 0:00:55 acc1_pos: 94.010 acc5_pos: 100.000 loss: 0.2532 (0.2599) acc1: 92.4479 (92.5487) acc5: 100.0000 (100.0000) time: 0.1239 data: 0.0341 max mem: 3531 +Test: [ 40/235] eta: 0:00:54 acc1_pos: 94.010 acc5_pos: 100.000 loss: 0.2520 (0.2590) acc1: 92.9688 (92.6512) acc5: 100.0000 (100.0000) time: 0.2181 data: 0.1215 max mem: 3531 +Test: [ 50/235] eta: 0:00:53 acc1_pos: 90.625 acc5_pos: 100.000 loss: 0.2590 (0.2596) acc1: 92.7083 (92.5551) acc5: 100.0000 (100.0000) time: 0.3124 data: 0.2141 max mem: 3531 +Test: [ 60/235] eta: 0:00:49 acc1_pos: 93.229 acc5_pos: 100.000 loss: 0.2618 (0.2601) acc1: 92.7083 (92.6144) acc5: 100.0000 (100.0000) time: 0.2799 data: 0.1864 max mem: 3531 +Test: [ 70/235] eta: 0:00:45 acc1_pos: 93.846 acc5_pos: 100.000 loss: 0.2810 (0.3856) acc1: 90.8854 (84.6171) acc5: 100.0000 (100.0000) acc1_neg: 24.219 acc5_neg: 100.000 time: 0.2488 data: 0.1558 max mem: 3531 +Test: [ 80/235] eta: 0:00:42 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.2638 (0.4963) acc1: 28.6458 (77.4724) acc5: 100.0000 (100.0000) acc1_neg: 26.562 acc5_neg: 100.000 time: 0.2459 data: 0.1526 max mem: 3531 +Test: [ 90/235] eta: 0:00:38 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.2801 (0.5835) acc1: 26.5625 (71.9609) acc5: 100.0000 (100.0000) acc1_neg: 29.427 acc5_neg: 100.000 time: 0.2305 data: 0.1365 max mem: 3531 +Test: [100/235] eta: 0:00:35 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.2798 (0.6519) acc1: 27.8646 (67.6593) acc5: 100.0000 (100.0000) acc1_neg: 25.260 acc5_neg: 100.000 time: 0.2344 data: 0.1416 max mem: 3531 +Test: [110/235] eta: 0:00:32 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.2691 (0.7081) acc1: 26.8229 (63.9968) acc5: 100.0000 (100.0000) acc1_neg: 26.823 acc5_neg: 100.000 time: 0.2326 data: 0.1394 max mem: 3531 +Test: [120/235] eta: 0:00:29 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.3372 (0.7774) acc1: 23.9583 (59.8657) acc5: 100.0000 (100.0000) acc1_neg: 13.281 acc5_neg: 100.000 time: 0.2068 data: 0.1107 max mem: 3531 +Test: [130/235] eta: 0:00:26 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.5572 (0.8372) acc1: 12.7604 (56.3037) acc5: 100.0000 (100.0000) acc1_neg: 12.240 acc5_neg: 100.000 time: 0.2200 data: 0.1257 max mem: 3531 +Test: [140/235] eta: 0:00:23 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.5874 (0.8910) acc1: 12.2396 (53.1933) acc5: 100.0000 (100.0000) acc1_neg: 9.896 acc5_neg: 100.000 time: 0.2294 data: 0.1383 max mem: 3531 +Test: [150/235] eta: 0:00:21 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.5874 (0.9359) acc1: 12.2396 (50.4950) acc5: 100.0000 (100.0000) acc1_neg: 13.802 acc5_neg: 100.000 time: 0.2265 data: 0.1353 max mem: 3531 +Test: [160/235] eta: 0:00:18 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.5620 (0.9750) acc1: 13.2812 (48.1690) acc5: 100.0000 (100.0000) acc1_neg: 15.104 acc5_neg: 100.000 time: 0.2281 data: 0.1352 max mem: 3531 +Test: [170/235] eta: 0:00:16 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.5240 (0.9812) acc1: 15.1042 (47.5009) acc5: 100.0000 (100.0000) acc1_neg: 34.115 acc5_neg: 100.000 time: 0.2141 data: 0.1190 max mem: 3531 +Test: [180/235] eta: 0:00:13 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.0239 (0.9832) acc1: 39.3229 (47.0635) acc5: 100.0000 (100.0000) acc1_neg: 36.458 acc5_neg: 100.000 time: 0.2109 data: 0.1164 max mem: 3531 +Test: [190/235] eta: 0:00:10 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.0141 (0.9853) acc1: 39.8438 (46.6828) acc5: 100.0000 (100.0000) acc1_neg: 41.927 acc5_neg: 100.000 time: 0.2098 data: 0.1176 max mem: 3531 +Test: [200/235] eta: 0:00:08 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.0261 (0.9880) acc1: 39.0625 (46.3127) acc5: 100.0000 (100.0000) acc1_neg: 38.542 acc5_neg: 100.000 time: 0.2188 data: 0.1264 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.0334 (0.9894) acc1: 39.0625 (46.0172) acc5: 100.0000 (100.0000) acc1_neg: 40.365 acc5_neg: 100.000 time: 0.1874 data: 0.0953 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.0233 (0.9909) acc1: 40.1042 (45.7697) acc5: 100.0000 (100.0000) acc1_neg: 39.062 acc5_neg: 100.000 time: 0.1433 data: 0.0524 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.0167 (0.9918) acc1: 40.3646 (45.5774) acc5: 100.0000 (100.0000) acc1_neg: 42.448 acc5_neg: 100.000 time: 0.1327 data: 0.0420 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 93.846 acc5_pos: 100.000 loss: 1.0167 (0.9933) acc1: 40.3646 (45.4967) acc5: 100.0000 (100.0000) acc1_neg: 30.435 acc5_neg: 100.000 time: 0.1275 data: 0.0397 max mem: 3532 +Test: Total time: 0:00:53 (0.2266 s / it) +* Overall Acc@1 45.497 Acc@5 100.000 + target == real Acc@1 92.611 Acc@5 100.000 + target == fake Acc@1 28.457 Acc@5 100.000 + loss 0.993 + PR-AUC 0.8698 ROC-AUC 0.7192 EER 33.52% (thr=0.2100) +Accuracy of the network on the 89879 test images: 45.5% +[rank0]:[W726 05:50:35.872512422 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:50:55.434000 1195167 torch/distributed/run.py:766] +W0726 05:50:55.434000 1195167 torch/distributed/run.py:766] ***************************************** +W0726 05:50:55.434000 1195167 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:50:55.434000 1195167 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:51:22.713513937 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:51:22.741076272 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:51:22.791266324 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:51:23.077538039 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:51:23.339264401 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:51:23.380568032 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank6]:[W726 05:51:23.380615292 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:51:23.417116624 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n22_seed8_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed8_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:23:34 acc1_pos: 35.677 acc5_pos: 100.000 loss: 0.8911 (0.8911) acc1: 35.6771 (35.6771) acc5: 100.0000 (100.0000) time: 5.2578 data: 3.2867 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/269] eta: 0:02:32 acc1_pos: 72.135 acc5_pos: 100.000 loss: 0.6943 (0.7207) acc1: 57.2917 (55.1373) acc5: 100.0000 (100.0000) time: 0.5902 data: 0.3295 max mem: 3531 +Test: [ 20/269] eta: 0:01:33 acc1_pos: 55.729 acc5_pos: 100.000 loss: 0.6830 (0.7027) acc1: 55.7292 (56.7584) acc5: 100.0000 (100.0000) time: 0.1303 data: 0.0377 max mem: 3531 +Test: [ 30/269] eta: 0:01:13 acc1_pos: 55.990 acc5_pos: 100.000 loss: 0.6283 (0.6753) acc1: 60.4167 (59.4758) acc5: 100.0000 (100.0000) time: 0.1492 data: 0.0549 max mem: 3531 +Test: [ 40/269] eta: 0:01:07 acc1_pos: 21.552 acc5_pos: 100.000 loss: 0.5410 (0.6319) acc1: 75.5208 (64.6469) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2144 data: 0.1221 max mem: 3531 +Test: [ 50/269] eta: 0:01:03 acc1_pos: 21.552 acc5_pos: 100.000 loss: 0.4554 (0.6002) acc1: 85.6771 (68.6581) acc5: 100.0000 (100.0000) acc1_neg: 65.885 acc5_neg: 100.000 time: 0.2676 data: 0.1745 max mem: 3531 +Test: [ 60/269] eta: 0:00:57 acc1_pos: 21.552 acc5_pos: 100.000 loss: 0.4894 (0.5898) acc1: 83.8542 (69.9838) acc5: 100.0000 (100.0000) acc1_neg: 70.312 acc5_neg: 100.000 time: 0.2374 data: 0.1410 max mem: 3531 +Test: [ 70/269] eta: 0:00:53 acc1_pos: 21.552 acc5_pos: 100.000 loss: 0.5361 (0.5817) acc1: 75.7812 (70.9544) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.2136 data: 0.1184 max mem: 3531 +Test: [ 80/269] eta: 0:00:48 acc1_pos: 21.552 acc5_pos: 100.000 loss: 0.5107 (0.5716) acc1: 80.2083 (72.1804) acc5: 100.0000 (100.0000) acc1_neg: 66.146 acc5_neg: 100.000 time: 0.2009 data: 0.1081 max mem: 3531 +Test: [ 90/269] eta: 0:00:45 acc1_pos: 21.552 acc5_pos: 100.000 loss: 0.5771 (0.5813) acc1: 71.3542 (71.0823) acc5: 100.0000 (100.0000) acc1_neg: 73.958 acc5_neg: 100.000 time: 0.1967 data: 0.1047 max mem: 3531 +Test: [100/269] eta: 0:00:42 acc1_pos: 21.552 acc5_pos: 100.000 loss: 0.6079 (0.5820) acc1: 64.8438 (70.8359) acc5: 100.0000 (100.0000) acc1_neg: 63.281 acc5_neg: 100.000 time: 0.2189 data: 0.1275 max mem: 3531 +Test: [110/269] eta: 0:00:38 acc1_pos: 21.552 acc5_pos: 100.000 loss: 0.6158 (0.5845) acc1: 66.6667 (70.4932) acc5: 100.0000 (100.0000) acc1_neg: 59.635 acc5_neg: 100.000 time: 0.1982 data: 0.1056 max mem: 3531 +Test: [120/269] eta: 0:00:36 acc1_pos: 21.552 acc5_pos: 100.000 loss: 0.5888 (0.5861) acc1: 66.9271 (70.3276) acc5: 100.0000 (100.0000) acc1_neg: 72.135 acc5_neg: 100.000 time: 0.2017 data: 0.1095 max mem: 3531 +Test: [130/269] eta: 0:00:33 acc1_pos: 21.552 acc5_pos: 100.000 loss: 0.5980 (0.5968) acc1: 65.8854 (69.0263) acc5: 100.0000 (100.0000) acc1_neg: 41.406 acc5_neg: 100.000 time: 0.2066 data: 0.1157 max mem: 3531 +Test: [140/269] eta: 0:00:30 acc1_pos: 21.552 acc5_pos: 100.000 loss: 0.6766 (0.6027) acc1: 51.3021 (68.2403) acc5: 100.0000 (100.0000) acc1_neg: 61.458 acc5_neg: 100.000 time: 0.2144 data: 0.1244 max mem: 3531 +Test: [150/269] eta: 0:00:28 acc1_pos: 21.552 acc5_pos: 100.000 loss: 0.6440 (0.6043) acc1: 61.1979 (67.9084) acc5: 100.0000 (100.0000) acc1_neg: 65.885 acc5_neg: 100.000 time: 0.2231 data: 0.1320 max mem: 3531 +Test: [160/269] eta: 0:00:25 acc1_pos: 21.552 acc5_pos: 100.000 loss: 0.6114 (0.6040) acc1: 63.8021 (67.8652) acc5: 100.0000 (100.0000) acc1_neg: 68.490 acc5_neg: 100.000 time: 0.1886 data: 0.0954 max mem: 3531 +Test: [170/269] eta: 0:00:22 acc1_pos: 21.552 acc5_pos: 100.000 loss: 0.6114 (0.6070) acc1: 64.5833 (67.4875) acc5: 100.0000 (100.0000) acc1_neg: 65.885 acc5_neg: 100.000 time: 0.1767 data: 0.0838 max mem: 3531 +Test: [180/269] eta: 0:00:20 acc1_pos: 21.552 acc5_pos: 100.000 loss: 0.6006 (0.6024) acc1: 67.4479 (68.0623) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.2049 data: 0.1135 max mem: 3531 +Test: [190/269] eta: 0:00:17 acc1_pos: 21.552 acc5_pos: 100.000 loss: 0.5458 (0.6017) acc1: 72.3958 (68.0846) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.2181 data: 0.1247 max mem: 3531 +Test: [200/269] eta: 0:00:15 acc1_pos: 21.552 acc5_pos: 100.000 loss: 0.5794 (0.6022) acc1: 68.2292 (68.0271) acc5: 100.0000 (100.0000) acc1_neg: 75.781 acc5_neg: 100.000 time: 0.1799 data: 0.0839 max mem: 3531 +Test: [210/269] eta: 0:00:13 acc1_pos: 21.552 acc5_pos: 100.000 loss: 0.6380 (0.6051) acc1: 63.0208 (67.5960) acc5: 100.0000 (100.0000) acc1_neg: 58.333 acc5_neg: 100.000 time: 0.1688 data: 0.0737 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 21.552 acc5_pos: 100.000 loss: 0.6178 (0.6079) acc1: 63.0208 (67.2370) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.1856 data: 0.0918 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 21.552 acc5_pos: 100.000 loss: 0.5499 (0.6047) acc1: 74.2188 (67.6159) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.1843 data: 0.0918 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 21.552 acc5_pos: 100.000 loss: 0.5781 (0.6068) acc1: 69.0104 (67.3863) acc5: 100.0000 (100.0000) acc1_neg: 66.927 acc5_neg: 100.000 time: 0.1525 data: 0.0614 max mem: 3532 +Test: [250/269] eta: 0:00:04 acc1_pos: 21.552 acc5_pos: 100.000 loss: 0.6520 (0.6098) acc1: 59.8958 (67.0236) acc5: 100.0000 (100.0000) acc1_neg: 38.542 acc5_neg: 100.000 time: 0.1355 data: 0.0449 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 21.552 acc5_pos: 100.000 loss: 0.6520 (0.6123) acc1: 59.8958 (66.7196) acc5: 100.0000 (100.0000) acc1_neg: 72.396 acc5_neg: 100.000 time: 0.1301 data: 0.0401 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 21.552 acc5_pos: 100.000 loss: 0.6176 (0.6107) acc1: 64.5833 (66.8835) acc5: 100.0000 (100.0000) acc1_neg: 92.913 acc5_neg: 100.000 time: 0.1221 data: 0.0281 max mem: 3532 +Test: Total time: 0:00:55 (0.2061 s / it) +* Overall Acc@1 66.884 Acc@5 100.000 + target == real Acc@1 58.223 Acc@5 100.000 + target == fake Acc@1 68.065 Acc@5 100.000 + loss 0.611 + PR-AUC 0.9312 ROC-AUC 0.6801 EER 36.73% (thr=0.5244) +Accuracy of the network on the 103293 test images: 66.9% +[rank0]:[W726 05:52:27.897433940 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T05:52:34Z] eval_done tag=fixed500k_fake_n22_seed8_resampled +[2026-07-26T05:52:35Z] eval_start tag=fixed500k_real_n11_seed9_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n11_classseed9_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T05:52:35Z +hostname=n116-041-188 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n11_classseed9_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n11_seed9_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 05:52:48.182000 1195732 torch/distributed/run.py:766] +W0726 05:52:48.182000 1195732 torch/distributed/run.py:766] ***************************************** +W0726 05:52:48.182000 1195732 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:52:48.182000 1195732 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:53:17.381372916 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:53:17.721925197 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:53:17.758699307 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:53:17.807600843 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:53:18.958147132 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:53:18.038432652 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +[rank0]:[W726 05:53:18.051319235 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 05:53:18.053537523 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n11_seed9_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n11_classseed9_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:28:05 acc1_pos: 19.531 acc5_pos: 100.000 loss: 1.3381 (1.3381) acc1: 19.5312 (19.5312) acc5: 100.0000 (100.0000) time: 7.2027 data: 5.5094 max mem: 3531 +Test: [ 10/234] eta: 0:02:45 acc1_pos: 15.885 acc5_pos: 100.000 loss: 1.3381 (1.2975) acc1: 22.6562 (24.7869) acc5: 100.0000 (100.0000) time: 0.7400 data: 0.5013 max mem: 3531 +Test: [ 20/234] eta: 0:01:33 acc1_pos: 72.917 acc5_pos: 100.000 loss: 1.0222 (1.0494) acc1: 38.0208 (40.0050) acc5: 100.0000 (100.0000) time: 0.0968 data: 0.0004 max mem: 3531 +Test: [ 30/234] eta: 0:01:11 acc1_pos: 69.010 acc5_pos: 100.000 loss: 0.5204 (0.8602) acc1: 75.7812 (52.8730) acc5: 100.0000 (100.0000) time: 0.1399 data: 0.0420 max mem: 3531 +Test: [ 40/234] eta: 0:01:01 acc1_pos: 94.010 acc5_pos: 100.000 loss: 0.4896 (0.7598) acc1: 77.8646 (59.4449) acc5: 100.0000 (100.0000) time: 0.1906 data: 0.0966 max mem: 3531 +Test: [ 50/234] eta: 0:00:53 acc1_pos: 92.969 acc5_pos: 100.000 loss: 0.2295 (0.6518) acc1: 92.7083 (66.1918) acc5: 100.0000 (100.0000) time: 0.1999 data: 0.1043 max mem: 3531 +Test: [ 60/234] eta: 0:00:47 acc1_pos: 93.103 acc5_pos: 100.000 loss: 0.1484 (0.5631) acc1: 96.3542 (71.6914) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1892 data: 0.0933 max mem: 3531 +Test: [ 70/234] eta: 0:00:44 acc1_pos: 93.103 acc5_pos: 100.000 loss: 0.1072 (0.5395) acc1: 100.0000 (73.4448) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.2163 data: 0.1231 max mem: 3531 +Test: [ 80/234] eta: 0:00:41 acc1_pos: 93.103 acc5_pos: 100.000 loss: 0.5021 (0.5200) acc1: 77.3438 (74.8007) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2675 data: 0.1742 max mem: 3531 +Test: [ 90/234] eta: 0:00:37 acc1_pos: 93.103 acc5_pos: 100.000 loss: 0.1347 (0.4791) acc1: 98.4375 (77.2779) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.2290 data: 0.1362 max mem: 3531 +Test: [100/234] eta: 0:00:33 acc1_pos: 93.103 acc5_pos: 100.000 loss: 0.1370 (0.4683) acc1: 97.6562 (77.9909) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1397 data: 0.0443 max mem: 3531 +Test: [110/234] eta: 0:00:30 acc1_pos: 93.103 acc5_pos: 100.000 loss: 0.2552 (0.4520) acc1: 89.3229 (79.0752) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1670 data: 0.0729 max mem: 3531 +Test: [120/234] eta: 0:00:27 acc1_pos: 93.103 acc5_pos: 100.000 loss: 0.2094 (0.4459) acc1: 93.2292 (79.4680) acc5: 100.0000 (100.0000) acc1_neg: 66.406 acc5_neg: 100.000 time: 0.2327 data: 0.1418 max mem: 3531 +Test: [130/234] eta: 0:00:24 acc1_pos: 93.103 acc5_pos: 100.000 loss: 0.3341 (0.4392) acc1: 84.6354 (79.8823) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.2037 data: 0.1126 max mem: 3531 +Test: [140/234] eta: 0:00:22 acc1_pos: 93.103 acc5_pos: 100.000 loss: 0.3657 (0.4516) acc1: 84.1146 (79.3994) acc5: 100.0000 (100.0000) acc1_neg: 33.333 acc5_neg: 100.000 time: 0.1747 data: 0.0835 max mem: 3531 +Test: [150/234] eta: 0:00:19 acc1_pos: 93.103 acc5_pos: 100.000 loss: 0.4381 (0.4547) acc1: 79.9479 (79.2908) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1708 data: 0.0768 max mem: 3531 +Test: [160/234] eta: 0:00:16 acc1_pos: 93.103 acc5_pos: 100.000 loss: 0.3157 (0.4461) acc1: 83.0729 (79.8104) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1855 data: 0.0898 max mem: 3531 +Test: [170/234] eta: 0:00:14 acc1_pos: 93.103 acc5_pos: 100.000 loss: 0.4144 (0.4433) acc1: 82.8125 (80.0515) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2094 data: 0.1139 max mem: 3531 +Test: [180/234] eta: 0:00:12 acc1_pos: 93.103 acc5_pos: 100.000 loss: 0.4180 (0.4476) acc1: 82.8125 (79.8731) acc5: 100.0000 (100.0000) acc1_neg: 46.094 acc5_neg: 100.000 time: 0.2002 data: 0.1065 max mem: 3531 +Test: [190/234] eta: 0:00:09 acc1_pos: 93.103 acc5_pos: 100.000 loss: 0.6374 (0.4596) acc1: 65.6250 (79.0331) acc5: 100.0000 (100.0000) acc1_neg: 65.625 acc5_neg: 100.000 time: 0.2132 data: 0.1223 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 93.103 acc5_pos: 100.000 loss: 0.6243 (0.4516) acc1: 67.4479 (79.4685) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2176 data: 0.1253 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 93.103 acc5_pos: 100.000 loss: 0.1072 (0.4356) acc1: 99.7396 (80.4231) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1567 data: 0.0646 max mem: 3532 +Test: [220/234] eta: 0:00:03 acc1_pos: 93.103 acc5_pos: 100.000 loss: 0.1243 (0.4217) acc1: 99.4792 (81.2783) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1420 data: 0.0517 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 93.103 acc5_pos: 100.000 loss: 0.1484 (0.4118) acc1: 97.1354 (81.9129) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1287 data: 0.0386 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 93.103 acc5_pos: 100.000 loss: 0.1881 (0.4091) acc1: 96.3542 (82.0771) acc5: 100.0000 (100.0000) acc1_neg: 96.109 acc5_neg: 100.000 time: 0.0945 data: 0.0034 max mem: 3532 +Test: Total time: 0:00:49 (0.2111 s / it) +* Overall Acc@1 82.077 Acc@5 100.000 + target == real Acc@1 66.390 Acc@5 100.000 + target == fake Acc@1 86.499 Acc@5 100.000 + loss 0.409 + PR-AUC 0.9602 ROC-AUC 0.8709 EER 21.46% (thr=0.7119) +Accuracy of the network on the 89729 test images: 82.1% +[rank0]:[W726 05:54:15.706852975 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 05:54:35.029000 1196292 torch/distributed/run.py:766] +W0726 05:54:35.029000 1196292 torch/distributed/run.py:766] ***************************************** +W0726 05:54:35.029000 1196292 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 05:54:35.029000 1196292 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 05:55:03.879429309 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 05:55:04.062425399 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 05:55:04.193112813 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 05:55:04.359872183 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 05:55:04.536402348 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 05:55:04.559519947 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +NCCL version 2.26.2+cuda12.2 +[rank2]:[W726 05:55:04.673196565 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 05:55:04.676845121 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n11_seed9_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n11_classseed9_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:04:44 acc1_pos: 34.896 acc5_pos: 100.000 loss: 1.0287 (1.0287) acc1: 34.8958 (34.8958) acc5: 100.0000 (100.0000) time: 6.1517 data: 4.3326 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/2387] eta: 0:27:15 acc1_pos: 28.385 acc5_pos: 100.000 loss: 1.0511 (1.0471) acc1: 34.8958 (36.6477) acc5: 100.0000 (100.0000) time: 0.6881 data: 0.4412 max mem: 3531 +Test: [ 20/2387] eta: 0:15:58 acc1_pos: 15.625 acc5_pos: 100.000 loss: 1.0511 (1.0765) acc1: 34.8958 (34.8338) acc5: 100.0000 (100.0000) time: 0.1176 data: 0.0263 max mem: 3531 +Test: [ 30/2387] eta: 0:13:13 acc1_pos: 14.062 acc5_pos: 100.000 loss: 1.1351 (1.1067) acc1: 32.0312 (33.2157) acc5: 100.0000 (100.0000) time: 0.1431 data: 0.0473 max mem: 3531 +Test: [ 40/2387] eta: 0:13:06 acc1_pos: 20.833 acc5_pos: 100.000 loss: 1.1302 (1.1152) acc1: 28.1250 (31.9804) acc5: 100.0000 (100.0000) time: 0.2616 data: 0.1663 max mem: 3531 +Test: [ 50/2387] eta: 0:12:14 acc1_pos: 16.927 acc5_pos: 100.000 loss: 1.0997 (1.1071) acc1: 29.9479 (32.6695) acc5: 100.0000 (100.0000) time: 0.2801 data: 0.1866 max mem: 3531 +Test: [ 60/2387] eta: 0:11:30 acc1_pos: 26.823 acc5_pos: 100.000 loss: 1.2151 (1.1419) acc1: 26.5625 (31.2628) acc5: 100.0000 (100.0000) time: 0.2185 data: 0.1234 max mem: 3531 +Test: [ 70/2387] eta: 0:11:03 acc1_pos: 22.135 acc5_pos: 100.000 loss: 1.3783 (1.1926) acc1: 22.9167 (29.7609) acc5: 100.0000 (100.0000) time: 0.2146 data: 0.1181 max mem: 3531 +Test: [ 80/2387] eta: 0:10:49 acc1_pos: 10.156 acc5_pos: 100.000 loss: 1.5151 (1.2287) acc1: 13.8021 (27.8967) acc5: 100.0000 (100.0000) time: 0.2349 data: 0.1381 max mem: 3531 +Test: [ 90/2387] eta: 0:10:28 acc1_pos: 9.375 acc5_pos: 100.000 loss: 1.5738 (1.2659) acc1: 10.1562 (26.0846) acc5: 100.0000 (100.0000) time: 0.2293 data: 0.1326 max mem: 3531 +Test: [ 100/2387] eta: 0:10:18 acc1_pos: 7.812 acc5_pos: 100.000 loss: 1.5738 (1.3001) acc1: 9.3750 (24.8350) acc5: 100.0000 (100.0000) time: 0.2250 data: 0.1278 max mem: 3531 +Test: [ 110/2387] eta: 0:10:11 acc1_pos: 22.396 acc5_pos: 100.000 loss: 1.5187 (1.3192) acc1: 12.7604 (24.1085) acc5: 100.0000 (100.0000) time: 0.2443 data: 0.1472 max mem: 3531 +Test: [ 120/2387] eta: 0:10:01 acc1_pos: 8.854 acc5_pos: 100.000 loss: 1.5193 (1.3392) acc1: 13.5417 (23.2675) acc5: 100.0000 (100.0000) time: 0.2403 data: 0.1426 max mem: 3531 +Test: [ 130/2387] eta: 0:09:47 acc1_pos: 7.552 acc5_pos: 100.000 loss: 1.6227 (1.3617) acc1: 9.1146 (22.2646) acc5: 100.0000 (100.0000) time: 0.2161 data: 0.1188 max mem: 3531 +Test: [ 140/2387] eta: 0:09:43 acc1_pos: 6.771 acc5_pos: 100.000 loss: 1.6179 (1.3775) acc1: 8.8542 (21.3893) acc5: 100.0000 (100.0000) time: 0.2242 data: 0.1280 max mem: 3531 +Test: [ 150/2387] eta: 0:09:34 acc1_pos: 11.719 acc5_pos: 100.000 loss: 1.5386 (1.3843) acc1: 11.7188 (20.9230) acc5: 100.0000 (100.0000) time: 0.2342 data: 0.1373 max mem: 3531 +Test: [ 160/2387] eta: 0:09:25 acc1_pos: 37.900 acc5_pos: 100.000 loss: 1.5026 (1.3585) acc1: 13.8021 (22.6805) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2147 data: 0.1167 max mem: 3531 +Test: [ 170/2387] eta: 0:09:13 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1778 (1.2890) acc1: 96.0938 (26.9935) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1942 data: 0.0988 max mem: 3531 +Test: [ 180/2387] eta: 0:09:03 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1724 (1.2275) acc1: 96.0938 (30.8025) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1824 data: 0.0867 max mem: 3531 +Test: [ 190/2387] eta: 0:08:50 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1758 (1.1723) acc1: 95.8333 (34.2141) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1717 data: 0.0755 max mem: 3531 +Test: [ 200/2387] eta: 0:08:42 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1701 (1.1226) acc1: 96.3542 (37.3031) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1759 data: 0.0823 max mem: 3531 +Test: [ 210/2387] eta: 0:08:37 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1702 (1.0775) acc1: 96.3542 (40.1079) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2007 data: 0.1085 max mem: 3532 +Test: [ 220/2387] eta: 0:08:27 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1705 (1.0365) acc1: 96.3542 (42.6553) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1864 data: 0.0947 max mem: 3532 +Test: [ 230/2387] eta: 0:08:20 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1687 (0.9989) acc1: 96.3542 (44.9720) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1745 data: 0.0809 max mem: 3532 +Test: [ 240/2387] eta: 0:08:15 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1506 (0.9631) acc1: 96.8750 (47.1754) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1920 data: 0.0979 max mem: 3532 +Test: [ 250/2387] eta: 0:08:10 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1237 (0.9291) acc1: 98.9583 (49.2623) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1977 data: 0.1043 max mem: 3532 +Test: [ 260/2387] eta: 0:08:03 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1082 (0.8976) acc1: 99.7396 (51.1973) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1853 data: 0.0916 max mem: 3532 +Test: [ 270/2387] eta: 0:07:59 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1052 (0.8683) acc1: 100.0000 (52.9943) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1898 data: 0.0974 max mem: 3532 +Test: [ 280/2387] eta: 0:07:52 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1052 (0.8413) acc1: 100.0000 (54.6560) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1874 data: 0.0927 max mem: 3532 +Test: [ 290/2387] eta: 0:07:48 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0997 (0.8158) acc1: 100.0000 (56.2062) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1853 data: 0.0896 max mem: 3532 +Test: [ 300/2387] eta: 0:07:43 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0934 (0.7918) acc1: 100.0000 (57.6602) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1874 data: 0.0941 max mem: 3532 +Test: [ 310/2387] eta: 0:07:40 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0992 (0.7700) acc1: 100.0000 (59.0108) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1961 data: 0.1000 max mem: 3532 +Test: [ 320/2387] eta: 0:07:36 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1206 (0.7498) acc1: 100.0000 (60.2852) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2035 data: 0.1090 max mem: 3532 +Test: [ 330/2387] eta: 0:07:34 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1209 (0.7308) acc1: 100.0000 (61.4843) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2055 data: 0.1152 max mem: 3532 +Test: [ 340/2387] eta: 0:07:30 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1207 (0.7129) acc1: 100.0000 (62.6130) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2094 data: 0.1188 max mem: 3532 +Test: [ 350/2387] eta: 0:07:25 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1165 (0.6952) acc1: 100.0000 (63.6774) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1783 data: 0.0865 max mem: 3532 +Test: [ 360/2387] eta: 0:07:20 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0971 (0.6800) acc1: 99.7396 (64.6194) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1708 data: 0.0774 max mem: 3532 +Test: [ 370/2387] eta: 0:07:16 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1501 (0.6658) acc1: 97.3958 (65.4937) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1827 data: 0.0887 max mem: 3532 +Test: [ 380/2387] eta: 0:07:11 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1501 (0.6522) acc1: 97.3958 (66.3400) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1673 data: 0.0750 max mem: 3532 +Test: [ 390/2387] eta: 0:07:07 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1494 (0.6394) acc1: 97.6562 (67.1415) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1680 data: 0.0769 max mem: 3532 +Test: [ 400/2387] eta: 0:07:03 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1477 (0.6271) acc1: 97.9167 (67.9071) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1835 data: 0.0924 max mem: 3532 +Test: [ 410/2387] eta: 0:07:00 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1455 (0.6154) acc1: 97.9167 (68.6340) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1835 data: 0.0902 max mem: 3532 +Test: [ 420/2387] eta: 0:06:56 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1455 (0.6043) acc1: 97.6562 (69.3234) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1846 data: 0.0898 max mem: 3532 +Test: [ 430/2387] eta: 0:06:53 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1390 (0.5933) acc1: 97.9167 (70.0050) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1853 data: 0.0928 max mem: 3532 +Test: [ 440/2387] eta: 0:06:51 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1166 (0.5821) acc1: 99.2188 (70.6780) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2029 data: 0.1114 max mem: 3533 +Test: [ 450/2387] eta: 0:06:48 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0906 (0.5712) acc1: 100.0000 (71.3282) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2002 data: 0.1076 max mem: 3533 +Test: [ 460/2387] eta: 0:06:44 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0906 (0.5608) acc1: 100.0000 (71.9501) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1801 data: 0.0873 max mem: 3533 +Test: [ 470/2387] eta: 0:06:41 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0905 (0.5508) acc1: 100.0000 (72.5451) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1844 data: 0.0900 max mem: 3533 +Test: [ 480/2387] eta: 0:06:39 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0902 (0.5413) acc1: 100.0000 (73.1159) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1959 data: 0.0976 max mem: 3533 +Test: [ 490/2387] eta: 0:06:37 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0905 (0.5321) acc1: 100.0000 (73.6634) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2021 data: 0.1062 max mem: 3533 +Test: [ 500/2387] eta: 0:06:34 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0906 (0.5233) acc1: 100.0000 (74.1891) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1954 data: 0.1038 max mem: 3533 +Test: [ 510/2387] eta: 0:06:31 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0909 (0.5148) acc1: 100.0000 (74.6942) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1827 data: 0.0910 max mem: 3533 +Test: [ 520/2387] eta: 0:06:28 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0925 (0.5068) acc1: 100.0000 (75.1799) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1816 data: 0.0892 max mem: 3533 +Test: [ 530/2387] eta: 0:06:25 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1004 (0.4991) acc1: 100.0000 (75.6474) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1820 data: 0.0892 max mem: 3533 +Test: [ 540/2387] eta: 0:06:21 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1013 (0.4918) acc1: 100.0000 (76.0975) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1742 data: 0.0814 max mem: 3533 +Test: [ 550/2387] eta: 0:06:18 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1013 (0.4847) acc1: 100.0000 (76.5313) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1725 data: 0.0805 max mem: 3533 +Test: [ 560/2387] eta: 0:06:15 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1014 (0.4779) acc1: 100.0000 (76.9496) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1720 data: 0.0810 max mem: 3533 +Test: [ 570/2387] eta: 0:06:12 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1014 (0.4713) acc1: 100.0000 (77.3529) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1726 data: 0.0801 max mem: 3533 +Test: [ 580/2387] eta: 0:06:09 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1014 (0.4649) acc1: 100.0000 (77.7427) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1764 data: 0.0835 max mem: 3533 +Test: [ 590/2387] eta: 0:06:06 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1014 (0.4588) acc1: 100.0000 (78.1193) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1752 data: 0.0829 max mem: 3533 +Test: [ 600/2387] eta: 0:06:04 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0970 (0.4527) acc1: 100.0000 (78.4829) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1859 data: 0.0911 max mem: 3533 +Test: [ 610/2387] eta: 0:06:02 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0970 (0.4470) acc1: 100.0000 (78.8321) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2019 data: 0.1070 max mem: 3533 +Test: [ 620/2387] eta: 0:06:00 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1034 (0.4414) acc1: 100.0000 (79.1709) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2028 data: 0.1110 max mem: 3533 +Test: [ 630/2387] eta: 0:05:57 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1036 (0.4361) acc1: 100.0000 (79.4985) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1912 data: 0.0994 max mem: 3533 +Test: [ 640/2387] eta: 0:05:56 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1039 (0.4309) acc1: 99.7396 (79.8159) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2100 data: 0.1151 max mem: 3533 +Test: [ 650/2387] eta: 0:05:53 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1038 (0.4259) acc1: 100.0000 (80.1251) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2093 data: 0.1140 max mem: 3533 +Test: [ 660/2387] eta: 0:05:53 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1018 (0.4210) acc1: 100.0000 (80.4254) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2191 data: 0.1232 max mem: 3533 +Test: [ 670/2387] eta: 0:05:50 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1018 (0.4162) acc1: 100.0000 (80.7133) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2141 data: 0.1172 max mem: 3534 +Test: [ 680/2387] eta: 0:05:47 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1002 (0.4115) acc1: 100.0000 (80.9953) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1717 data: 0.0756 max mem: 3534 +Test: [ 690/2387] eta: 0:05:45 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1011 (0.4108) acc1: 99.7396 (81.0303) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.1849 data: 0.0863 max mem: 3534 +Test: [ 700/2387] eta: 0:05:42 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.4653 (0.4116) acc1: 77.0833 (80.9706) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.1891 data: 0.0926 max mem: 3534 +Test: [ 710/2387] eta: 0:05:40 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.4674 (0.4124) acc1: 76.8229 (80.9097) acc5: 100.0000 (100.0000) acc1_neg: 76.302 acc5_neg: 100.000 time: 0.1849 data: 0.0933 max mem: 3534 +Test: [ 720/2387] eta: 0:05:38 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.4667 (0.4131) acc1: 77.0833 (80.8653) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.1939 data: 0.1000 max mem: 3534 +Test: [ 730/2387] eta: 0:05:36 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.4706 (0.4140) acc1: 77.0833 (80.8061) acc5: 100.0000 (100.0000) acc1_neg: 77.083 acc5_neg: 100.000 time: 0.2059 data: 0.1109 max mem: 3534 +Test: [ 740/2387] eta: 0:05:33 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.4740 (0.4149) acc1: 77.0833 (80.7467) acc5: 100.0000 (100.0000) acc1_neg: 75.521 acc5_neg: 100.000 time: 0.1955 data: 0.1033 max mem: 3534 +Test: [ 750/2387] eta: 0:05:31 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.4664 (0.4155) acc1: 77.3438 (80.7042) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.1765 data: 0.0830 max mem: 3534 +Test: [ 760/2387] eta: 0:05:28 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.4554 (0.4150) acc1: 77.6042 (80.7299) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1809 data: 0.0845 max mem: 3534 +Test: [ 770/2387] eta: 0:05:26 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.2351 (0.4122) acc1: 90.3646 (80.8967) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1878 data: 0.0940 max mem: 3534 +Test: [ 780/2387] eta: 0:05:24 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.2081 (0.4097) acc1: 94.0104 (81.0696) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1861 data: 0.0950 max mem: 3534 +Test: [ 790/2387] eta: 0:05:21 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1829 (0.4077) acc1: 96.0938 (81.2085) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.1848 data: 0.0900 max mem: 3534 +Test: [ 800/2387] eta: 0:05:19 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1799 (0.4049) acc1: 97.1354 (81.3970) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1901 data: 0.0932 max mem: 3534 +Test: [ 810/2387] eta: 0:05:17 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1799 (0.4022) acc1: 97.1354 (81.5811) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1910 data: 0.0981 max mem: 3534 +Test: [ 820/2387] eta: 0:05:14 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1843 (0.3997) acc1: 97.1354 (81.7689) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1838 data: 0.0923 max mem: 3534 +Test: [ 830/2387] eta: 0:05:12 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1733 (0.3971) acc1: 97.6562 (81.9294) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1802 data: 0.0883 max mem: 3534 +Test: [ 840/2387] eta: 0:05:10 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1306 (0.3938) acc1: 98.9583 (82.1381) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1829 data: 0.0909 max mem: 3534 +Test: [ 850/2387] eta: 0:05:08 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1130 (0.3905) acc1: 99.7396 (82.3461) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1931 data: 0.1002 max mem: 3534 +Test: [ 860/2387] eta: 0:05:05 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1090 (0.3872) acc1: 100.0000 (82.5497) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0962 max mem: 3534 +Test: [ 870/2387] eta: 0:05:03 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1059 (0.3840) acc1: 99.7396 (82.7461) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1722 data: 0.0793 max mem: 3534 +Test: [ 880/2387] eta: 0:05:01 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1112 (0.3809) acc1: 99.7396 (82.9378) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1799 data: 0.0887 max mem: 3534 +Test: [ 890/2387] eta: 0:04:59 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1012 (0.3778) acc1: 99.7396 (83.1244) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1962 data: 0.1032 max mem: 3534 +Test: [ 900/2387] eta: 0:04:56 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0997 (0.3756) acc1: 99.4792 (83.2732) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1915 data: 0.0977 max mem: 3535 +Test: [ 910/2387] eta: 0:04:54 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1900 (0.3735) acc1: 96.0938 (83.4194) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1870 data: 0.0955 max mem: 3535 +Test: [ 920/2387] eta: 0:04:52 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1855 (0.3715) acc1: 96.8750 (83.5655) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1938 data: 0.1036 max mem: 3535 +Test: [ 930/2387] eta: 0:04:50 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1880 (0.3695) acc1: 96.8750 (83.7054) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1807 data: 0.0899 max mem: 3535 +Test: [ 940/2387] eta: 0:04:47 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1881 (0.3676) acc1: 96.8750 (83.8461) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1665 data: 0.0751 max mem: 3535 +Test: [ 950/2387] eta: 0:04:45 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1849 (0.3657) acc1: 97.1354 (83.9859) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1884 data: 0.0954 max mem: 3535 +Test: [ 960/2387] eta: 0:04:43 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1820 (0.3638) acc1: 97.1354 (84.1214) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1848 data: 0.0908 max mem: 3535 +Test: [ 970/2387] eta: 0:04:40 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1847 (0.3620) acc1: 96.8750 (84.2500) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1699 data: 0.0758 max mem: 3535 +Test: [ 980/2387] eta: 0:04:38 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1861 (0.3606) acc1: 96.8750 (84.3559) acc5: 100.0000 (100.0000) acc1_neg: 79.427 acc5_neg: 100.000 time: 0.1785 data: 0.0854 max mem: 3535 +Test: [ 990/2387] eta: 0:04:36 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1868 (0.3595) acc1: 96.8750 (84.4339) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1850 data: 0.0923 max mem: 3535 +Test: [1000/2387] eta: 0:04:34 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1527 (0.3578) acc1: 97.6562 (84.5503) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1959 data: 0.1031 max mem: 3535 +Test: [1010/2387] eta: 0:04:32 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1655 (0.3558) acc1: 97.3958 (84.6777) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1767 data: 0.0855 max mem: 3535 +Test: [1020/2387] eta: 0:04:29 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1680 (0.3540) acc1: 97.3958 (84.7892) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1606 data: 0.0693 max mem: 3535 +Test: [1030/2387] eta: 0:04:27 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1618 (0.3522) acc1: 97.9167 (84.9160) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1686 data: 0.0767 max mem: 3535 +Test: [1040/2387] eta: 0:04:24 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1485 (0.3502) acc1: 98.6979 (85.0517) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1665 data: 0.0735 max mem: 3535 +Test: [1050/2387] eta: 0:04:22 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1443 (0.3483) acc1: 98.6979 (85.1795) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1658 data: 0.0735 max mem: 3535 +Test: [1060/2387] eta: 0:04:20 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1466 (0.3464) acc1: 98.4375 (85.3030) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1687 data: 0.0762 max mem: 3535 +Test: [1070/2387] eta: 0:04:18 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1449 (0.3446) acc1: 98.6979 (85.4237) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1747 data: 0.0824 max mem: 3535 +Test: [1080/2387] eta: 0:04:15 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1449 (0.3430) acc1: 98.6979 (85.5246) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1772 data: 0.0858 max mem: 3535 +Test: [1090/2387] eta: 0:04:13 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.2180 (0.3423) acc1: 93.7500 (85.5730) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.1661 data: 0.0740 max mem: 3535 +Test: [1100/2387] eta: 0:04:11 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.2678 (0.3419) acc1: 90.8854 (85.6014) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1599 data: 0.0665 max mem: 3535 +Test: [1110/2387] eta: 0:04:08 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.2831 (0.3419) acc1: 88.2812 (85.6051) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.1623 data: 0.0678 max mem: 3535 +Test: [1120/2387] eta: 0:04:06 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.2831 (0.3413) acc1: 88.2812 (85.6476) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1610 data: 0.0686 max mem: 3536 +Test: [1130/2387] eta: 0:04:04 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.2697 (0.3406) acc1: 90.8854 (85.6941) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1658 data: 0.0753 max mem: 3536 +Test: [1140/2387] eta: 0:04:01 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1996 (0.3392) acc1: 93.2292 (85.7828) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1654 data: 0.0735 max mem: 3536 +Test: [1150/2387] eta: 0:03:59 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1359 (0.3374) acc1: 98.6979 (85.8920) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1700 data: 0.0770 max mem: 3536 +Test: [1160/2387] eta: 0:03:57 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0968 (0.3353) acc1: 100.0000 (86.0135) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1750 data: 0.0831 max mem: 3536 +Test: [1170/2387] eta: 0:03:55 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0972 (0.3333) acc1: 100.0000 (86.1330) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1793 data: 0.0880 max mem: 3536 +Test: [1180/2387] eta: 0:03:53 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0972 (0.3313) acc1: 100.0000 (86.2482) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1939 data: 0.1005 max mem: 3536 +Test: [1190/2387] eta: 0:03:51 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0912 (0.3292) acc1: 100.0000 (86.3637) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1794 data: 0.0850 max mem: 3536 +Test: [1200/2387] eta: 0:03:49 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0834 (0.3272) acc1: 100.0000 (86.4772) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1654 data: 0.0730 max mem: 3536 +Test: [1210/2387] eta: 0:03:47 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0848 (0.3253) acc1: 100.0000 (86.5863) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1803 data: 0.0890 max mem: 3536 +Test: [1220/2387] eta: 0:03:45 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1186 (0.3237) acc1: 99.2188 (86.6808) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1746 data: 0.0837 max mem: 3536 +Test: [1230/2387] eta: 0:03:43 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1198 (0.3220) acc1: 99.2188 (86.7829) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1649 data: 0.0736 max mem: 3536 +Test: [1240/2387] eta: 0:03:40 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1155 (0.3204) acc1: 99.4792 (86.8837) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1646 data: 0.0735 max mem: 3536 +Test: [1250/2387] eta: 0:03:38 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1147 (0.3187) acc1: 99.2188 (86.9829) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1720 data: 0.0813 max mem: 3536 +Test: [1260/2387] eta: 0:03:36 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1123 (0.3171) acc1: 99.4792 (87.0789) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1791 data: 0.0875 max mem: 3536 +Test: [1270/2387] eta: 0:03:34 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1123 (0.3155) acc1: 99.4792 (87.1757) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1836 data: 0.0907 max mem: 3536 +Test: [1280/2387] eta: 0:03:32 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1117 (0.3139) acc1: 99.4792 (87.2705) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1663 data: 0.0740 max mem: 3536 +Test: [1290/2387] eta: 0:03:30 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1238 (0.3127) acc1: 98.9583 (87.3455) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1595 data: 0.0686 max mem: 3536 +Test: [1300/2387] eta: 0:03:28 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1332 (0.3114) acc1: 98.6979 (87.4285) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1686 data: 0.0770 max mem: 3536 +Test: [1310/2387] eta: 0:03:26 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1358 (0.3102) acc1: 98.9583 (87.5101) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1708 data: 0.0784 max mem: 3536 +Test: [1320/2387] eta: 0:03:24 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1377 (0.3090) acc1: 99.2188 (87.5893) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1803 data: 0.0870 max mem: 3536 +Test: [1330/2387] eta: 0:03:22 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1398 (0.3078) acc1: 98.6979 (87.6667) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1746 data: 0.0809 max mem: 3536 +Test: [1340/2387] eta: 0:03:20 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1527 (0.3074) acc1: 97.1354 (87.6915) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1697 data: 0.0770 max mem: 3536 +Test: [1350/2387] eta: 0:03:17 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0872 (0.3057) acc1: 100.0000 (87.7826) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1647 data: 0.0731 max mem: 3537 +Test: [1360/2387] eta: 0:03:15 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0817 (0.3041) acc1: 100.0000 (87.8724) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1768 data: 0.0850 max mem: 3537 +Test: [1370/2387] eta: 0:03:14 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0821 (0.3025) acc1: 100.0000 (87.9606) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1910 data: 0.0970 max mem: 3537 +Test: [1380/2387] eta: 0:03:11 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0808 (0.3008) acc1: 100.0000 (88.0478) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1753 data: 0.0822 max mem: 3537 +Test: [1390/2387] eta: 0:03:10 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0786 (0.2992) acc1: 100.0000 (88.1337) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1844 data: 0.0938 max mem: 3537 +Test: [1400/2387] eta: 0:03:08 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0753 (0.2976) acc1: 100.0000 (88.2184) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2149 data: 0.1239 max mem: 3537 +Test: [1410/2387] eta: 0:03:06 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0803 (0.2968) acc1: 100.0000 (88.2676) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.2094 data: 0.1176 max mem: 3537 +Test: [1420/2387] eta: 0:03:04 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.2290 (0.2969) acc1: 92.9688 (88.2688) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1908 data: 0.0978 max mem: 3537 +Test: [1430/2387] eta: 0:03:02 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.2414 (0.2965) acc1: 91.9271 (88.3036) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1884 data: 0.0946 max mem: 3537 +Test: [1440/2387] eta: 0:03:00 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.2106 (0.2959) acc1: 94.0104 (88.3407) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1889 data: 0.0959 max mem: 3537 +Test: [1450/2387] eta: 0:02:58 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1854 (0.2951) acc1: 96.6146 (88.3963) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1881 data: 0.0956 max mem: 3537 +Test: [1460/2387] eta: 0:02:56 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1447 (0.2941) acc1: 97.3958 (88.4606) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1871 data: 0.0942 max mem: 3537 +Test: [1470/2387] eta: 0:02:55 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1328 (0.2929) acc1: 97.9167 (88.5312) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1869 data: 0.0945 max mem: 3537 +Test: [1480/2387] eta: 0:02:52 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1187 (0.2918) acc1: 99.2188 (88.6009) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1803 data: 0.0873 max mem: 3537 +Test: [1490/2387] eta: 0:02:51 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1187 (0.2907) acc1: 99.2188 (88.6690) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1799 data: 0.0867 max mem: 3537 +Test: [1500/2387] eta: 0:02:49 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1181 (0.2896) acc1: 98.9583 (88.7367) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1799 data: 0.0857 max mem: 3537 +Test: [1510/2387] eta: 0:02:47 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1030 (0.2882) acc1: 100.0000 (88.8112) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1850 data: 0.0909 max mem: 3537 +Test: [1520/2387] eta: 0:02:45 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0945 (0.2870) acc1: 100.0000 (88.8819) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1923 data: 0.1000 max mem: 3537 +Test: [1530/2387] eta: 0:02:43 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1034 (0.2858) acc1: 99.7396 (88.9540) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1781 data: 0.0863 max mem: 3537 +Test: [1540/2387] eta: 0:02:41 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1013 (0.2847) acc1: 100.0000 (89.0246) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1818 data: 0.0891 max mem: 3537 +Test: [1550/2387] eta: 0:02:39 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1013 (0.2835) acc1: 100.0000 (89.0944) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1847 data: 0.0912 max mem: 3537 +Test: [1560/2387] eta: 0:02:37 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1027 (0.2823) acc1: 99.7396 (89.1628) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1707 data: 0.0787 max mem: 3537 +Test: [1570/2387] eta: 0:02:35 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0942 (0.2811) acc1: 99.7396 (89.2296) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1698 data: 0.0789 max mem: 3537 +Test: [1580/2387] eta: 0:02:33 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1018 (0.2804) acc1: 99.7396 (89.2717) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1753 data: 0.0846 max mem: 3538 +Test: [1590/2387] eta: 0:02:31 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.2016 (0.2802) acc1: 94.2708 (89.2938) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1841 data: 0.0931 max mem: 3538 +Test: [1600/2387] eta: 0:02:29 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.2041 (0.2800) acc1: 93.7500 (89.3091) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1974 data: 0.1048 max mem: 3538 +Test: [1610/2387] eta: 0:02:27 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1804 (0.2795) acc1: 95.5729 (89.3464) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1829 data: 0.0907 max mem: 3538 +Test: [1620/2387] eta: 0:02:25 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.2137 (0.2792) acc1: 92.9688 (89.3666) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1653 data: 0.0740 max mem: 3538 +Test: [1630/2387] eta: 0:02:23 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.2177 (0.2789) acc1: 93.4896 (89.3925) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1641 data: 0.0713 max mem: 3538 +Test: [1640/2387] eta: 0:02:21 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1937 (0.2783) acc1: 95.3125 (89.4286) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1640 data: 0.0709 max mem: 3538 +Test: [1650/2387] eta: 0:02:19 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1785 (0.2777) acc1: 96.6146 (89.4720) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1672 data: 0.0744 max mem: 3538 +Test: [1660/2387] eta: 0:02:17 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1698 (0.2772) acc1: 96.8750 (89.5093) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1707 data: 0.0784 max mem: 3538 +Test: [1670/2387] eta: 0:02:15 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1725 (0.2766) acc1: 96.6146 (89.5523) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1863 data: 0.0945 max mem: 3538 +Test: [1680/2387] eta: 0:02:13 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1659 (0.2760) acc1: 97.3958 (89.5898) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1850 data: 0.0930 max mem: 3538 +Test: [1690/2387] eta: 0:02:11 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1726 (0.2756) acc1: 97.3958 (89.6215) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1751 data: 0.0826 max mem: 3538 +Test: [1700/2387] eta: 0:02:09 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1726 (0.2750) acc1: 96.0938 (89.6611) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1748 data: 0.0822 max mem: 3538 +Test: [1710/2387] eta: 0:02:07 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1536 (0.2742) acc1: 96.6146 (89.7037) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1715 data: 0.0796 max mem: 3538 +Test: [1720/2387] eta: 0:02:05 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1398 (0.2735) acc1: 97.3958 (89.7478) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1704 data: 0.0784 max mem: 3538 +Test: [1730/2387] eta: 0:02:03 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1389 (0.2727) acc1: 97.1354 (89.7908) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1652 data: 0.0733 max mem: 3538 +Test: [1740/2387] eta: 0:02:02 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1400 (0.2720) acc1: 96.8750 (89.8309) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1650 data: 0.0734 max mem: 3538 +Test: [1750/2387] eta: 0:02:00 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1392 (0.2712) acc1: 97.3958 (89.8750) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1737 data: 0.0820 max mem: 3538 +Test: [1760/2387] eta: 0:01:58 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1295 (0.2704) acc1: 98.1771 (89.9204) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1760 data: 0.0840 max mem: 3538 +Test: [1770/2387] eta: 0:01:56 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1213 (0.2697) acc1: 98.1771 (89.9632) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1765 data: 0.0843 max mem: 3538 +Test: [1780/2387] eta: 0:01:54 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1328 (0.2690) acc1: 98.1771 (90.0087) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1848 data: 0.0905 max mem: 3538 +Test: [1790/2387] eta: 0:01:52 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1242 (0.2682) acc1: 98.4375 (90.0576) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1791 data: 0.0858 max mem: 3538 +Test: [1800/2387] eta: 0:01:50 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1285 (0.2675) acc1: 98.4375 (90.1016) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1832 data: 0.0918 max mem: 3538 +Test: [1810/2387] eta: 0:01:48 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1285 (0.2666) acc1: 98.6979 (90.1506) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1991 data: 0.1046 max mem: 3539 +Test: [1820/2387] eta: 0:01:46 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1107 (0.2658) acc1: 99.2188 (90.1998) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1742 data: 0.0802 max mem: 3539 +Test: [1830/2387] eta: 0:01:44 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1093 (0.2649) acc1: 99.4792 (90.2505) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1740 data: 0.0826 max mem: 3539 +Test: [1840/2387] eta: 0:01:42 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0956 (0.2640) acc1: 100.0000 (90.3035) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1908 data: 0.0994 max mem: 3539 +Test: [1850/2387] eta: 0:01:41 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0956 (0.2631) acc1: 100.0000 (90.3559) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1934 data: 0.0999 max mem: 3539 +Test: [1860/2387] eta: 0:01:39 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0959 (0.2622) acc1: 100.0000 (90.4077) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2084 data: 0.1143 max mem: 3539 +Test: [1870/2387] eta: 0:01:37 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0955 (0.2613) acc1: 100.0000 (90.4590) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2265 data: 0.1344 max mem: 3539 +Test: [1880/2387] eta: 0:01:35 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0952 (0.2604) acc1: 100.0000 (90.5097) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2049 data: 0.1121 max mem: 3539 +Test: [1890/2387] eta: 0:01:33 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0955 (0.2595) acc1: 100.0000 (90.5599) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1797 data: 0.0871 max mem: 3539 +Test: [1900/2387] eta: 0:01:31 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0964 (0.2587) acc1: 100.0000 (90.6095) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2011 data: 0.1095 max mem: 3539 +Test: [1910/2387] eta: 0:01:30 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1006 (0.2579) acc1: 100.0000 (90.6565) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1986 data: 0.1067 max mem: 3539 +Test: [1920/2387] eta: 0:01:28 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1252 (0.2579) acc1: 97.6562 (90.6570) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1666 data: 0.0743 max mem: 3539 +Test: [1930/2387] eta: 0:01:26 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1916 (0.2575) acc1: 95.8333 (90.6878) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1630 data: 0.0676 max mem: 3539 +Test: [1940/2387] eta: 0:01:24 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1775 (0.2577) acc1: 97.6562 (90.6867) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1731 data: 0.0791 max mem: 3539 +Test: [1950/2387] eta: 0:01:22 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1798 (0.2574) acc1: 97.9167 (90.7060) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1617 data: 0.0711 max mem: 3539 +Test: [1960/2387] eta: 0:01:20 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.2288 (0.2576) acc1: 92.4479 (90.7010) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1701 data: 0.0772 max mem: 3539 +Test: [1970/2387] eta: 0:01:18 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.2663 (0.2579) acc1: 89.8438 (90.6775) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1764 data: 0.0833 max mem: 3539 +Test: [1980/2387] eta: 0:01:16 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.2211 (0.2574) acc1: 92.7083 (90.7032) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1741 data: 0.0830 max mem: 3539 +Test: [1990/2387] eta: 0:01:14 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1039 (0.2566) acc1: 99.4792 (90.7485) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1733 data: 0.0813 max mem: 3539 +Test: [2000/2387] eta: 0:01:12 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1009 (0.2559) acc1: 99.7396 (90.7938) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1726 data: 0.0792 max mem: 3539 +Test: [2010/2387] eta: 0:01:10 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1009 (0.2551) acc1: 99.7396 (90.8383) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1764 data: 0.0837 max mem: 3539 +Test: [2020/2387] eta: 0:01:08 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1020 (0.2543) acc1: 99.7396 (90.8826) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1732 data: 0.0818 max mem: 3539 +Test: [2030/2387] eta: 0:01:06 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1010 (0.2536) acc1: 99.7396 (90.9264) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1714 data: 0.0806 max mem: 3540 +Test: [2040/2387] eta: 0:01:05 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0996 (0.2528) acc1: 99.7396 (90.9701) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1741 data: 0.0830 max mem: 3540 +Test: [2050/2387] eta: 0:01:03 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0993 (0.2521) acc1: 99.7396 (91.0134) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1775 data: 0.0862 max mem: 3540 +Test: [2060/2387] eta: 0:01:01 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0990 (0.2513) acc1: 99.7396 (91.0561) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1897 data: 0.0989 max mem: 3540 +Test: [2070/2387] eta: 0:00:59 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0968 (0.2506) acc1: 100.0000 (91.0987) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1843 data: 0.0936 max mem: 3540 +Test: [2080/2387] eta: 0:00:57 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0974 (0.2498) acc1: 100.0000 (91.1407) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1744 data: 0.0825 max mem: 3540 +Test: [2090/2387] eta: 0:00:55 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0958 (0.2491) acc1: 100.0000 (91.1825) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1762 data: 0.0838 max mem: 3540 +Test: [2100/2387] eta: 0:00:53 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0957 (0.2484) acc1: 99.7396 (91.2238) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1770 data: 0.0850 max mem: 3540 +Test: [2110/2387] eta: 0:00:51 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.0973 (0.2477) acc1: 99.7396 (91.2644) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1710 data: 0.0796 max mem: 3540 +Test: [2120/2387] eta: 0:00:49 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1007 (0.2474) acc1: 99.4792 (91.2820) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1628 data: 0.0715 max mem: 3540 +Test: [2130/2387] eta: 0:00:48 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1974 (0.2473) acc1: 94.2708 (91.2947) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1745 data: 0.0816 max mem: 3540 +Test: [2140/2387] eta: 0:00:46 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1984 (0.2474) acc1: 94.2708 (91.2928) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1770 data: 0.0839 max mem: 3540 +Test: [2150/2387] eta: 0:00:44 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1966 (0.2473) acc1: 95.5729 (91.3094) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1751 data: 0.0835 max mem: 3540 +Test: [2160/2387] eta: 0:00:42 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1869 (0.2470) acc1: 95.8333 (91.3324) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1778 data: 0.0846 max mem: 3540 +Test: [2170/2387] eta: 0:00:40 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1882 (0.2469) acc1: 94.7917 (91.3476) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1781 data: 0.0845 max mem: 3540 +Test: [2180/2387] eta: 0:00:38 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.2202 (0.2471) acc1: 94.0104 (91.3381) acc5: 100.0000 (100.0000) acc1_neg: 79.427 acc5_neg: 100.000 time: 0.1859 data: 0.0916 max mem: 3540 +Test: [2190/2387] eta: 0:00:36 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1798 (0.2467) acc1: 95.5729 (91.3618) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2065 data: 0.1130 max mem: 3540 +Test: [2200/2387] eta: 0:00:34 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1302 (0.2463) acc1: 98.9583 (91.3924) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1653 data: 0.0750 max mem: 3540 +Test: [2210/2387] eta: 0:00:33 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1185 (0.2457) acc1: 98.9583 (91.4246) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1427 data: 0.0507 max mem: 3540 +Test: [2220/2387] eta: 0:00:31 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1252 (0.2453) acc1: 99.4792 (91.4551) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1978 data: 0.1034 max mem: 3540 +Test: [2230/2387] eta: 0:00:29 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1259 (0.2448) acc1: 99.4792 (91.4868) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1712 data: 0.0784 max mem: 3540 +Test: [2240/2387] eta: 0:00:27 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1280 (0.2443) acc1: 99.2188 (91.5170) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1256 data: 0.0314 max mem: 3540 +Test: [2250/2387] eta: 0:00:25 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1126 (0.2436) acc1: 99.7396 (91.5519) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1597 data: 0.0648 max mem: 3540 +Test: [2260/2387] eta: 0:00:23 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1046 (0.2431) acc1: 100.0000 (91.5873) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1932 data: 0.1012 max mem: 3541 +Test: [2270/2387] eta: 0:00:21 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1393 (0.2428) acc1: 99.4792 (91.6122) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1555 data: 0.0636 max mem: 3541 +Test: [2280/2387] eta: 0:00:19 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1393 (0.2427) acc1: 99.4792 (91.6219) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1388 data: 0.0451 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1349 (0.2423) acc1: 99.4792 (91.6496) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1787 data: 0.0827 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1475 (0.2419) acc1: 98.6979 (91.6774) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1889 data: 0.0937 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1445 (0.2415) acc1: 98.9583 (91.7061) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1594 data: 0.0627 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1595 (0.2412) acc1: 97.6562 (91.7271) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1548 data: 0.0577 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1851 (0.2412) acc1: 96.0938 (91.7319) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1859 data: 0.0926 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.2127 (0.2411) acc1: 94.5312 (91.7415) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.1725 data: 0.0814 max mem: 3541 +Test: [2350/2387] eta: 0:00:06 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1979 (0.2413) acc1: 95.3125 (91.7337) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1289 data: 0.0381 max mem: 3541 +Test: [2360/2387] eta: 0:00:04 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1953 (0.2413) acc1: 94.7917 (91.7392) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1313 data: 0.0401 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.2304 (0.2414) acc1: 93.2292 (91.7388) acc5: 100.0000 (100.0000) acc1_neg: 74.740 acc5_neg: 100.000 time: 0.1390 data: 0.0486 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.2304 (0.2412) acc1: 93.4896 (91.7524) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1295 data: 0.0392 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 37.900 acc5_pos: 100.000 loss: 0.1777 (0.2409) acc1: 96.6146 (91.7652) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1246 data: 0.0368 max mem: 3541 +Test: Total time: 0:07:20 (0.1847 s / it) +* Overall Acc@1 91.765 Acc@5 100.000 + target == real Acc@1 20.583 Acc@5 100.000 + target == fake Acc@1 96.764 Acc@5 100.000 + loss 0.241 + PR-AUC 0.9869 ROC-AUC 0.8538 EER 21.20% (thr=0.8687) +Accuracy of the network on the 916267 test images: 91.8% +[rank0]:[W726 06:02:53.995641734 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 06:03:12.487000 1196912 torch/distributed/run.py:766] +W0726 06:03:12.487000 1196912 torch/distributed/run.py:766] ***************************************** +W0726 06:03:12.487000 1196912 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:03:12.487000 1196912 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:03:40.576833764 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:03:40.625577185 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:03:40.802626190 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W726 06:03:40.802780556 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 06:03:41.162023825 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +| distributed init (rank 5): env:// +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:03:41.180579898 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank1]:[W726 06:03:41.180658547 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:03:41.191012802 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n11_seed9_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n11_classseed9_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:01:02 acc1_pos: 34.896 acc5_pos: 100.000 loss: 1.0287 (1.0287) acc1: 34.8958 (34.8958) acc5: 100.0000 (100.0000) time: 7.0840 data: 5.5697 max mem: 3531 +Test: [ 10/517] eta: 0:06:09 acc1_pos: 28.385 acc5_pos: 100.000 loss: 1.0511 (1.0471) acc1: 34.8958 (36.6477) acc5: 100.0000 (100.0000) time: 0.7289 data: 0.5068 max mem: 3531 +Test: [ 20/517] eta: 0:03:33 acc1_pos: 15.625 acc5_pos: 100.000 loss: 1.0511 (1.0765) acc1: 34.8958 (34.8338) acc5: 100.0000 (100.0000) time: 0.0967 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:03:13 acc1_pos: 14.062 acc5_pos: 100.000 loss: 1.1351 (1.1067) acc1: 32.0312 (33.2157) acc5: 100.0000 (100.0000) time: 0.2149 data: 0.1163 max mem: 3531 +Test: [ 40/517] eta: 0:03:02 acc1_pos: 20.833 acc5_pos: 100.000 loss: 1.1302 (1.1152) acc1: 28.1250 (31.9804) acc5: 100.0000 (100.0000) time: 0.3349 data: 0.2390 max mem: 3531 +Test: [ 50/517] eta: 0:02:44 acc1_pos: 48.624 acc5_pos: 100.000 loss: 1.0529 (1.0664) acc1: 32.5521 (35.1205) acc5: 100.0000 (100.0000) acc1_neg: 76.302 acc5_neg: 100.000 time: 0.2849 data: 0.1904 max mem: 3531 +Test: [ 60/517] eta: 0:02:40 acc1_pos: 48.624 acc5_pos: 100.000 loss: 0.5074 (0.9574) acc1: 76.0417 (43.2676) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.2849 data: 0.1886 max mem: 3531 +Test: [ 70/517] eta: 0:02:33 acc1_pos: 48.624 acc5_pos: 100.000 loss: 0.4051 (0.8767) acc1: 84.6354 (49.3141) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.3149 data: 0.2180 max mem: 3531 +Test: [ 80/517] eta: 0:02:23 acc1_pos: 48.624 acc5_pos: 100.000 loss: 0.4250 (0.8258) acc1: 84.3750 (53.2440) acc5: 100.0000 (100.0000) acc1_neg: 66.146 acc5_neg: 100.000 time: 0.2549 data: 0.1596 max mem: 3531 +Test: [ 90/517] eta: 0:02:16 acc1_pos: 48.624 acc5_pos: 100.000 loss: 0.3184 (0.7667) acc1: 89.0625 (57.5721) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.2349 data: 0.1387 max mem: 3531 +Test: [100/517] eta: 0:02:09 acc1_pos: 48.624 acc5_pos: 100.000 loss: 0.3110 (0.7286) acc1: 90.3646 (60.2852) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.2399 data: 0.1421 max mem: 3531 +Test: [110/517] eta: 0:02:02 acc1_pos: 48.624 acc5_pos: 100.000 loss: 0.3479 (0.6934) acc1: 83.3333 (62.5868) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.2253 data: 0.1291 max mem: 3531 +Test: [120/517] eta: 0:01:58 acc1_pos: 48.624 acc5_pos: 100.000 loss: 0.4283 (0.6767) acc1: 80.2083 (63.8430) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.2349 data: 0.1380 max mem: 3531 +Test: [130/517] eta: 0:01:53 acc1_pos: 48.624 acc5_pos: 100.000 loss: 0.4784 (0.6592) acc1: 79.6875 (65.2115) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2401 data: 0.1430 max mem: 3531 +Test: [140/517] eta: 0:01:49 acc1_pos: 48.624 acc5_pos: 100.000 loss: 0.4152 (0.6427) acc1: 83.3333 (66.4838) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.2399 data: 0.1425 max mem: 3531 +Test: [150/517] eta: 0:01:45 acc1_pos: 48.624 acc5_pos: 100.000 loss: 0.4100 (0.6245) acc1: 84.3750 (67.8946) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2443 data: 0.1476 max mem: 3531 +Test: [160/517] eta: 0:01:41 acc1_pos: 48.624 acc5_pos: 100.000 loss: 0.3660 (0.6086) acc1: 87.7604 (69.0832) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.2549 data: 0.1602 max mem: 3531 +Test: [170/517] eta: 0:01:38 acc1_pos: 48.624 acc5_pos: 100.000 loss: 0.3404 (0.5925) acc1: 88.8021 (70.3186) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2650 data: 0.1714 max mem: 3531 +Test: [180/517] eta: 0:01:34 acc1_pos: 48.624 acc5_pos: 100.000 loss: 0.3287 (0.5784) acc1: 89.5833 (71.2851) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.2499 data: 0.1576 max mem: 3531 +Test: [190/517] eta: 0:01:31 acc1_pos: 48.624 acc5_pos: 100.000 loss: 0.1987 (0.5576) acc1: 98.1771 (72.7231) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2398 data: 0.1420 max mem: 3531 +Test: [200/517] eta: 0:01:27 acc1_pos: 48.624 acc5_pos: 100.000 loss: 0.1934 (0.5410) acc1: 98.4375 (73.9726) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2357 data: 0.1373 max mem: 3531 +Test: [210/517] eta: 0:01:24 acc1_pos: 48.624 acc5_pos: 100.000 loss: 0.2512 (0.5415) acc1: 88.2812 (73.9719) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.2399 data: 0.1434 max mem: 3532 +Test: [220/517] eta: 0:01:21 acc1_pos: 48.624 acc5_pos: 100.000 loss: 0.5410 (0.5416) acc1: 73.9583 (73.9937) acc5: 100.0000 (100.0000) acc1_neg: 74.740 acc5_neg: 100.000 time: 0.2491 data: 0.1516 max mem: 3532 +Test: [230/517] eta: 0:01:18 acc1_pos: 48.624 acc5_pos: 100.000 loss: 0.4992 (0.5375) acc1: 75.2604 (74.3653) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.2449 data: 0.1508 max mem: 3532 +Test: [240/517] eta: 0:01:15 acc1_pos: 48.624 acc5_pos: 100.000 loss: 0.3862 (0.5310) acc1: 84.1146 (74.8228) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.2399 data: 0.1460 max mem: 3532 +Test: [250/517] eta: 0:01:12 acc1_pos: 48.624 acc5_pos: 100.000 loss: 0.3324 (0.5243) acc1: 91.9271 (75.4804) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2453 data: 0.1499 max mem: 3532 +Test: [260/517] eta: 0:01:09 acc1_pos: 48.624 acc5_pos: 100.000 loss: 0.3211 (0.5175) acc1: 96.0938 (76.1923) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.2649 data: 0.1688 max mem: 3532 +Test: [270/517] eta: 0:01:06 acc1_pos: 48.624 acc5_pos: 100.000 loss: 0.3541 (0.5117) acc1: 89.5833 (76.6596) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.2546 data: 0.1575 max mem: 3532 +Test: [280/517] eta: 0:01:03 acc1_pos: 48.624 acc5_pos: 100.000 loss: 0.3541 (0.5055) acc1: 89.5833 (77.1667) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.2350 data: 0.1393 max mem: 3532 +Test: [290/517] eta: 0:01:00 acc1_pos: 48.624 acc5_pos: 100.000 loss: 0.3223 (0.4991) acc1: 92.1875 (77.7142) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2398 data: 0.1434 max mem: 3532 +Test: [300/517] eta: 0:00:57 acc1_pos: 48.624 acc5_pos: 100.000 loss: 0.3023 (0.4924) acc1: 93.4896 (78.2236) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.2498 data: 0.1511 max mem: 3532 +Test: [310/517] eta: 0:00:55 acc1_pos: 48.624 acc5_pos: 100.000 loss: 0.2965 (0.4861) acc1: 92.1875 (78.6919) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.2649 data: 0.1659 max mem: 3532 +Test: [320/517] eta: 0:00:52 acc1_pos: 48.624 acc5_pos: 100.000 loss: 0.3295 (0.4827) acc1: 90.8854 (78.9922) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.2549 data: 0.1568 max mem: 3532 +Test: [330/517] eta: 0:00:49 acc1_pos: 48.624 acc5_pos: 100.000 loss: 0.3781 (0.4794) acc1: 88.8021 (79.3091) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2449 data: 0.1476 max mem: 3532 +Test: [340/517] eta: 0:00:46 acc1_pos: 48.624 acc5_pos: 100.000 loss: 0.3531 (0.4751) acc1: 89.3229 (79.5554) acc5: 100.0000 (100.0000) acc1_neg: 67.969 acc5_neg: 100.000 time: 0.2499 data: 0.1523 max mem: 3532 +Test: [350/517] eta: 0:00:44 acc1_pos: 48.624 acc5_pos: 100.000 loss: 0.3098 (0.4724) acc1: 87.5000 (79.6823) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2449 data: 0.1472 max mem: 3532 +Test: [360/517] eta: 0:00:41 acc1_pos: 48.624 acc5_pos: 100.000 loss: 0.3311 (0.4686) acc1: 87.7604 (79.9840) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2449 data: 0.1469 max mem: 3532 +Test: [370/517] eta: 0:00:38 acc1_pos: 48.624 acc5_pos: 100.000 loss: 0.2697 (0.4628) acc1: 94.5312 (80.4273) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2499 data: 0.1511 max mem: 3532 +Test: [380/517] eta: 0:00:36 acc1_pos: 48.624 acc5_pos: 100.000 loss: 0.2732 (0.4594) acc1: 93.2292 (80.6424) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.2499 data: 0.1514 max mem: 3532 +Test: [390/517] eta: 0:00:33 acc1_pos: 48.624 acc5_pos: 100.000 loss: 0.3234 (0.4561) acc1: 89.3229 (80.8677) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.2399 data: 0.1435 max mem: 3532 +Test: [400/517] eta: 0:00:30 acc1_pos: 48.624 acc5_pos: 100.000 loss: 0.2931 (0.4522) acc1: 90.3646 (81.1000) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.2349 data: 0.1363 max mem: 3532 +Test: [410/517] eta: 0:00:27 acc1_pos: 48.624 acc5_pos: 100.000 loss: 0.3109 (0.4499) acc1: 90.1042 (81.3026) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.2399 data: 0.1374 max mem: 3532 +Test: [420/517] eta: 0:00:25 acc1_pos: 48.624 acc5_pos: 100.000 loss: 0.2996 (0.4463) acc1: 90.8854 (81.5964) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.2499 data: 0.1505 max mem: 3532 +Test: [430/517] eta: 0:00:22 acc1_pos: 48.624 acc5_pos: 100.000 loss: 0.3718 (0.4462) acc1: 89.3229 (81.6440) acc5: 100.0000 (100.0000) acc1_neg: 72.656 acc5_neg: 100.000 time: 0.2549 data: 0.1571 max mem: 3532 +Test: [440/517] eta: 0:00:20 acc1_pos: 48.624 acc5_pos: 100.000 loss: 0.4087 (0.4452) acc1: 84.1146 (81.7407) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.2499 data: 0.1489 max mem: 3533 +Test: [450/517] eta: 0:00:17 acc1_pos: 48.624 acc5_pos: 100.000 loss: 0.3914 (0.4438) acc1: 87.2396 (81.8430) acc5: 100.0000 (100.0000) acc1_neg: 77.344 acc5_neg: 100.000 time: 0.2549 data: 0.1539 max mem: 3533 +Test: [460/517] eta: 0:00:14 acc1_pos: 48.624 acc5_pos: 100.000 loss: 0.3914 (0.4438) acc1: 84.6354 (81.8053) acc5: 100.0000 (100.0000) acc1_neg: 69.271 acc5_neg: 100.000 time: 0.2549 data: 0.1555 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 48.624 acc5_pos: 100.000 loss: 0.2673 (0.4397) acc1: 94.0104 (82.1026) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2449 data: 0.1466 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 48.624 acc5_pos: 100.000 loss: 0.2395 (0.4363) acc1: 95.8333 (82.3556) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2449 data: 0.1467 max mem: 3533 +Test: [490/517] eta: 0:00:07 acc1_pos: 48.624 acc5_pos: 100.000 loss: 0.3006 (0.4357) acc1: 91.9271 (82.3787) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2449 data: 0.1462 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 48.624 acc5_pos: 100.000 loss: 0.2939 (0.4326) acc1: 96.3542 (82.6633) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.2123 data: 0.1170 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 48.624 acc5_pos: 100.000 loss: 0.2965 (0.4313) acc1: 95.3125 (82.7728) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1754 data: 0.0843 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 48.624 acc5_pos: 100.000 loss: 0.2835 (0.4294) acc1: 94.5312 (82.9252) acc5: 100.0000 (100.0000) acc1_neg: 93.993 acc5_neg: 100.000 time: 0.1450 data: 0.0535 max mem: 3533 +Test: Total time: 0:02:12 (0.2555 s / it) +* Overall Acc@1 82.925 Acc@5 100.000 + target == real Acc@1 32.906 Acc@5 100.000 + target == fake Acc@1 88.114 Acc@5 100.000 + loss 0.429 + PR-AUC 0.9379 ROC-AUC 0.6668 EER 37.03% (thr=0.7378) +Accuracy of the network on the 198427 test images: 82.9% +[rank0]:[W726 06:06:04.030176368 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 06:06:23.394000 1197488 torch/distributed/run.py:766] +W0726 06:06:23.394000 1197488 torch/distributed/run.py:766] ***************************************** +W0726 06:06:23.394000 1197488 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:06:23.394000 1197488 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:06:51.828045028 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:06:52.471327822 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:06:52.659828664 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:06:52.838883415 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:06:52.882428209 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:06:52.892045064 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 06:06:52.892047134 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:06:53.932252751 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n11_seed9_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n11_classseed9_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:43:24 acc1_pos: 0.521 acc5_pos: 100.000 loss: 1.8335 (1.8335) acc1: 0.5208 (0.5208) acc5: 100.0000 (100.0000) time: 6.6942 data: 5.0982 max mem: 3531 +Test: [ 10/389] eta: 0:04:24 acc1_pos: 30.469 acc5_pos: 100.000 loss: 1.7741 (1.7539) acc1: 0.7812 (4.6165) acc5: 100.0000 (100.0000) time: 0.6983 data: 0.4638 max mem: 3531 +Test: [ 20/389] eta: 0:02:32 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.7271 (1.7462) acc1: 0.7812 (8.5441) acc5: 100.0000 (100.0000) time: 0.0993 data: 0.0004 max mem: 3531 +Test: [ 30/389] eta: 0:01:54 acc1_pos: 0.260 acc5_pos: 100.000 loss: 1.9446 (1.8387) acc1: 0.0000 (5.9308) acc5: 100.0000 (100.0000) time: 0.1099 data: 0.0134 max mem: 3531 +Test: [ 40/389] eta: 0:01:36 acc1_pos: 0.260 acc5_pos: 100.000 loss: 1.9385 (1.7755) acc1: 0.2604 (7.8189) acc5: 100.0000 (100.0000) time: 0.1338 data: 0.0390 max mem: 3531 +Test: [ 50/389] eta: 0:01:27 acc1_pos: 3.906 acc5_pos: 100.000 loss: 1.5650 (1.7640) acc1: 2.0833 (7.0108) acc5: 100.0000 (100.0000) time: 0.1638 data: 0.0686 max mem: 3531 +Test: [ 60/389] eta: 0:01:21 acc1_pos: 15.365 acc5_pos: 100.000 loss: 1.5650 (1.7667) acc1: 2.8646 (6.4037) acc5: 100.0000 (100.0000) time: 0.1899 data: 0.0971 max mem: 3531 +Test: [ 70/389] eta: 0:01:17 acc1_pos: 3.906 acc5_pos: 100.000 loss: 1.5625 (1.7378) acc1: 1.0417 (7.0863) acc5: 100.0000 (100.0000) time: 0.2054 data: 0.1143 max mem: 3531 +Test: [ 80/389] eta: 0:01:13 acc1_pos: 62.240 acc5_pos: 100.000 loss: 1.4753 (1.6922) acc1: 0.2604 (9.9601) acc5: 100.0000 (100.0000) time: 0.1999 data: 0.1064 max mem: 3531 +Test: [ 90/389] eta: 0:01:06 acc1_pos: 15.885 acc5_pos: 100.000 loss: 0.9102 (1.6117) acc1: 41.1458 (13.8479) acc5: 100.0000 (100.0000) time: 0.1520 data: 0.0584 max mem: 3531 +Test: [100/389] eta: 0:01:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.3947 (1.6386) acc1: 15.8854 (13.3045) acc5: 100.0000 (100.0000) time: 0.1580 data: 0.0666 max mem: 3531 +Test: [110/389] eta: 0:01:00 acc1_pos: 97.917 acc5_pos: 100.000 loss: 1.9112 (1.6315) acc1: 1.3021 (15.2144) acc5: 100.0000 (100.0000) time: 0.1974 data: 0.1052 max mem: 3531 +Test: [120/389] eta: 0:00:57 acc1_pos: 20.573 acc5_pos: 100.000 loss: 1.4607 (1.6089) acc1: 29.6875 (16.8367) acc5: 100.0000 (100.0000) time: 0.1868 data: 0.0937 max mem: 3531 +Test: [130/389] eta: 0:00:54 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.7550 (1.5231) acc1: 56.7708 (21.5132) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1549 data: 0.0617 max mem: 3531 +Test: [140/389] eta: 0:00:52 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.1292 (1.4305) acc1: 99.4792 (26.6715) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.1699 data: 0.0782 max mem: 3531 +Test: [150/389] eta: 0:00:49 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.1292 (1.3553) acc1: 100.0000 (30.6378) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0930 max mem: 3531 +Test: [160/389] eta: 0:00:47 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.1266 (1.2833) acc1: 100.0000 (34.6678) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1868 data: 0.0934 max mem: 3531 +Test: [170/389] eta: 0:00:44 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.1518 (1.2166) acc1: 100.0000 (38.4838) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1822 data: 0.0895 max mem: 3531 +Test: [180/389] eta: 0:00:42 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.1415 (1.1577) acc1: 100.0000 (41.7400) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.1680 data: 0.0747 max mem: 3531 +Test: [190/389] eta: 0:00:40 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.1352 (1.1091) acc1: 100.0000 (44.4808) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1926 data: 0.0991 max mem: 3531 +Test: [200/389] eta: 0:00:37 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.1418 (1.0638) acc1: 100.0000 (47.0486) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1761 data: 0.0830 max mem: 3531 +Test: [210/389] eta: 0:00:35 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.1418 (1.0199) acc1: 100.0000 (49.5557) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1926 data: 0.1003 max mem: 3532 +Test: [220/389] eta: 0:00:33 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.1029 (0.9785) acc1: 100.0000 (51.8382) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1887 data: 0.0967 max mem: 3532 +Test: [230/389] eta: 0:00:31 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.1101 (0.9457) acc1: 100.0000 (53.7270) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.1622 data: 0.0695 max mem: 3532 +Test: [240/389] eta: 0:00:29 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.1533 (0.9176) acc1: 96.6146 (55.3002) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1838 data: 0.0912 max mem: 3532 +Test: [250/389] eta: 0:00:27 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.1357 (0.8857) acc1: 100.0000 (57.0811) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1869 data: 0.0945 max mem: 3532 +Test: [260/389] eta: 0:00:25 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.1250 (0.8609) acc1: 100.0000 (58.3473) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.1810 data: 0.0898 max mem: 3532 +Test: [270/389] eta: 0:00:23 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.1482 (0.8411) acc1: 98.1771 (59.4922) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1750 data: 0.0842 max mem: 3532 +Test: [280/389] eta: 0:00:21 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.1601 (0.8171) acc1: 99.2188 (60.9180) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1762 data: 0.0841 max mem: 3532 +Test: [290/389] eta: 0:00:19 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.1453 (0.7957) acc1: 100.0000 (62.1447) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1746 data: 0.0824 max mem: 3532 +Test: [300/389] eta: 0:00:17 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.1493 (0.7742) acc1: 100.0000 (63.4015) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1847 data: 0.0922 max mem: 3532 +Test: [310/389] eta: 0:00:15 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.1508 (0.7547) acc1: 100.0000 (64.5507) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1881 data: 0.0952 max mem: 3532 +Test: [320/389] eta: 0:00:13 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.1508 (0.7358) acc1: 100.0000 (65.6339) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1738 data: 0.0819 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.1204 (0.7173) acc1: 100.0000 (66.6698) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1706 data: 0.0794 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.1186 (0.6998) acc1: 100.0000 (67.6350) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1649 data: 0.0737 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.1298 (0.7026) acc1: 99.2188 (67.2788) acc5: 100.0000 (100.0000) acc1_neg: 1.042 acc5_neg: 100.000 time: 0.1600 data: 0.0675 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.7191 (0.7089) acc1: 56.5104 (66.6840) acc5: 100.0000 (100.0000) acc1_neg: 59.896 acc5_neg: 100.000 time: 0.1522 data: 0.0607 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.4084 (0.6951) acc1: 62.2396 (67.5279) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1375 data: 0.0465 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.1357 (0.6809) acc1: 100.0000 (68.3529) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1362 data: 0.0448 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 3.067 acc5_pos: 100.000 loss: 0.1357 (0.6702) acc1: 100.0000 (68.9859) acc5: 100.0000 (100.0000) acc1_neg: 95.082 acc5_neg: 100.000 time: 0.1340 data: 0.0401 max mem: 3532 +Test: Total time: 0:01:12 (0.1856 s / it) +* Overall Acc@1 68.986 Acc@5 100.000 + target == real Acc@1 16.803 Acc@5 100.000 + target == fake Acc@1 93.242 Acc@5 100.000 + loss 0.670 + PR-AUC 0.7773 ROC-AUC 0.6519 EER 38.15% (thr=0.8652) +Accuracy of the network on the 149358 test images: 69.0% +[rank0]:[W726 06:08:15.454965770 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 06:08:34.604000 1198048 torch/distributed/run.py:766] +W0726 06:08:34.604000 1198048 torch/distributed/run.py:766] ***************************************** +W0726 06:08:34.604000 1198048 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:08:34.604000 1198048 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:09:04.030418389 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +[rank4]:[W726 06:09:04.055936916 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:09:04.059481971 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:09:04.213043394 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:09:04.237232059 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 06:09:04.473454042 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:09:04.484214142 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:09:04.499204658 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n11_seed9_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n11_classseed9_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:34:02 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.2127 (2.2127) acc1: 0.0000 (0.0000) acc5: 100.0000 (100.0000) time: 6.1342 data: 4.5723 max mem: 3531 +Test: [ 10/333] eta: 0:03:26 acc1_pos: 25.260 acc5_pos: 100.000 loss: 1.9388 (1.7479) acc1: 2.8646 (10.2036) acc5: 100.0000 (100.0000) time: 0.6396 data: 0.4160 max mem: 3531 +Test: [ 20/333] eta: 0:01:59 acc1_pos: 26.042 acc5_pos: 100.000 loss: 1.3118 (1.5225) acc1: 20.5729 (16.1334) acc5: 100.0000 (100.0000) time: 0.0930 data: 0.0004 max mem: 3531 +Test: [ 30/333] eta: 0:01:32 acc1_pos: 44.010 acc5_pos: 100.000 loss: 1.1833 (1.3786) acc1: 26.3021 (23.4039) acc5: 100.0000 (100.0000) time: 0.1229 data: 0.0277 max mem: 3531 +Test: [ 40/333] eta: 0:01:21 acc1_pos: 48.177 acc5_pos: 100.000 loss: 1.0819 (1.3030) acc1: 37.2396 (27.1596) acc5: 100.0000 (100.0000) time: 0.1671 data: 0.0725 max mem: 3531 +Test: [ 50/333] eta: 0:01:13 acc1_pos: 34.375 acc5_pos: 100.000 loss: 1.0773 (1.2627) acc1: 34.3750 (28.5999) acc5: 100.0000 (100.0000) time: 0.1868 data: 0.0942 max mem: 3531 +Test: [ 60/333] eta: 0:01:07 acc1_pos: 31.250 acc5_pos: 100.000 loss: 1.0773 (1.2340) acc1: 34.3750 (30.1913) acc5: 100.0000 (100.0000) time: 0.1877 data: 0.0926 max mem: 3531 +Test: [ 70/333] eta: 0:01:02 acc1_pos: 30.469 acc5_pos: 100.000 loss: 1.0694 (1.2110) acc1: 38.0208 (31.4591) acc5: 100.0000 (100.0000) time: 0.1803 data: 0.0849 max mem: 3531 +Test: [ 80/333] eta: 0:00:59 acc1_pos: 44.531 acc5_pos: 100.000 loss: 1.0319 (1.1922) acc1: 39.0625 (32.5071) acc5: 100.0000 (100.0000) time: 0.2010 data: 0.1090 max mem: 3531 +Test: [ 90/333] eta: 0:00:57 acc1_pos: 64.323 acc5_pos: 100.000 loss: 0.9532 (1.1479) acc1: 42.4479 (34.9101) acc5: 100.0000 (100.0000) time: 0.2288 data: 0.1376 max mem: 3531 +Test: [100/333] eta: 0:00:53 acc1_pos: 61.979 acc5_pos: 100.000 loss: 0.6913 (1.1057) acc1: 61.1979 (37.4510) acc5: 100.0000 (100.0000) time: 0.2009 data: 0.1092 max mem: 3531 +Test: [110/333] eta: 0:00:49 acc1_pos: 66.406 acc5_pos: 100.000 loss: 0.7045 (1.0704) acc1: 60.9375 (39.4168) acc5: 100.0000 (100.0000) time: 0.1713 data: 0.0786 max mem: 3531 +Test: [120/333] eta: 0:00:47 acc1_pos: 77.604 acc5_pos: 100.000 loss: 0.7265 (1.0385) acc1: 58.5938 (41.1910) acc5: 100.0000 (100.0000) time: 0.1980 data: 0.1061 max mem: 3531 +Test: [130/333] eta: 0:00:44 acc1_pos: 65.365 acc5_pos: 100.000 loss: 0.6985 (1.0109) acc1: 59.1146 (42.8892) acc5: 100.0000 (100.0000) time: 0.1982 data: 0.1036 max mem: 3531 +Test: [140/333] eta: 0:00:42 acc1_pos: 65.885 acc5_pos: 100.000 loss: 0.6281 (0.9832) acc1: 65.8854 (44.6605) acc5: 100.0000 (100.0000) time: 0.1837 data: 0.0884 max mem: 3531 +Test: [150/333] eta: 0:00:39 acc1_pos: 59.635 acc5_pos: 100.000 loss: 0.6404 (0.9639) acc1: 64.5833 (45.7557) acc5: 100.0000 (100.0000) time: 0.1831 data: 0.0911 max mem: 3531 +Test: [160/333] eta: 0:00:36 acc1_pos: 59.211 acc5_pos: 100.000 loss: 0.6595 (0.9466) acc1: 60.1562 (46.8249) acc5: 100.0000 (100.0000) acc1_neg: 57.552 acc5_neg: 100.000 time: 0.1825 data: 0.0874 max mem: 3531 +Test: [170/333] eta: 0:00:34 acc1_pos: 59.211 acc5_pos: 100.000 loss: 0.6492 (0.9302) acc1: 62.2396 (47.7431) acc5: 100.0000 (100.0000) acc1_neg: 66.406 acc5_neg: 100.000 time: 0.1863 data: 0.0907 max mem: 3531 +Test: [180/333] eta: 0:00:32 acc1_pos: 59.211 acc5_pos: 100.000 loss: 0.6395 (0.9129) acc1: 65.1042 (48.7886) acc5: 100.0000 (100.0000) acc1_neg: 72.396 acc5_neg: 100.000 time: 0.1751 data: 0.0814 max mem: 3531 +Test: [190/333] eta: 0:00:29 acc1_pos: 59.211 acc5_pos: 100.000 loss: 0.5924 (0.8944) acc1: 68.4896 (49.9618) acc5: 100.0000 (100.0000) acc1_neg: 65.365 acc5_neg: 100.000 time: 0.1801 data: 0.0863 max mem: 3531 +Test: [200/333] eta: 0:00:27 acc1_pos: 59.211 acc5_pos: 100.000 loss: 0.5709 (0.8779) acc1: 70.3125 (51.0326) acc5: 100.0000 (100.0000) acc1_neg: 65.365 acc5_neg: 100.000 time: 0.1839 data: 0.0927 max mem: 3531 +Test: [210/333] eta: 0:00:25 acc1_pos: 59.211 acc5_pos: 100.000 loss: 0.5728 (0.8658) acc1: 69.7917 (51.7476) acc5: 100.0000 (100.0000) acc1_neg: 69.531 acc5_neg: 100.000 time: 0.1783 data: 0.0844 max mem: 3532 +Test: [220/333] eta: 0:00:23 acc1_pos: 59.211 acc5_pos: 100.000 loss: 0.5959 (0.8582) acc1: 66.6667 (52.0998) acc5: 100.0000 (100.0000) acc1_neg: 75.781 acc5_neg: 100.000 time: 0.1738 data: 0.0795 max mem: 3532 +Test: [230/333] eta: 0:00:20 acc1_pos: 59.211 acc5_pos: 100.000 loss: 0.4991 (0.8413) acc1: 74.2188 (53.2772) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.1747 data: 0.0834 max mem: 3532 +Test: [240/333] eta: 0:00:18 acc1_pos: 59.211 acc5_pos: 100.000 loss: 0.4991 (0.8309) acc1: 72.3958 (53.9019) acc5: 100.0000 (100.0000) acc1_neg: 64.062 acc5_neg: 100.000 time: 0.1913 data: 0.0981 max mem: 3532 +Test: [250/333] eta: 0:00:16 acc1_pos: 59.211 acc5_pos: 100.000 loss: 0.5775 (0.8210) acc1: 69.2708 (54.5225) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.2049 data: 0.1112 max mem: 3532 +Test: [260/333] eta: 0:00:14 acc1_pos: 59.211 acc5_pos: 100.000 loss: 0.5820 (0.8132) acc1: 67.1875 (54.9759) acc5: 100.0000 (100.0000) acc1_neg: 66.406 acc5_neg: 100.000 time: 0.1989 data: 0.1071 max mem: 3532 +Test: [270/333] eta: 0:00:12 acc1_pos: 59.211 acc5_pos: 100.000 loss: 0.5732 (0.8034) acc1: 69.7917 (55.5994) acc5: 100.0000 (100.0000) acc1_neg: 69.792 acc5_neg: 100.000 time: 0.1923 data: 0.1002 max mem: 3532 +Test: [280/333] eta: 0:00:10 acc1_pos: 59.211 acc5_pos: 100.000 loss: 0.5599 (0.8005) acc1: 69.7917 (55.6856) acc5: 100.0000 (100.0000) acc1_neg: 54.167 acc5_neg: 100.000 time: 0.1804 data: 0.0878 max mem: 3532 +Test: [290/333] eta: 0:00:08 acc1_pos: 59.211 acc5_pos: 100.000 loss: 0.8112 (0.8022) acc1: 50.7812 (55.4589) acc5: 100.0000 (100.0000) acc1_neg: 56.250 acc5_neg: 100.000 time: 0.1696 data: 0.0771 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 59.211 acc5_pos: 100.000 loss: 0.7589 (0.7993) acc1: 53.9062 (55.5977) acc5: 100.0000 (100.0000) acc1_neg: 60.938 acc5_neg: 100.000 time: 0.1671 data: 0.0744 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 59.211 acc5_pos: 100.000 loss: 0.6624 (0.7919) acc1: 61.9792 (56.0398) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.1603 data: 0.0679 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 59.211 acc5_pos: 100.000 loss: 0.5896 (0.7865) acc1: 67.4479 (56.3538) acc5: 100.0000 (100.0000) acc1_neg: 60.938 acc5_neg: 100.000 time: 0.1556 data: 0.0636 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 59.211 acc5_pos: 100.000 loss: 0.7032 (0.7852) acc1: 60.9375 (56.3790) acc5: 100.0000 (100.0000) acc1_neg: 58.594 acc5_neg: 100.000 time: 0.1408 data: 0.0492 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 59.211 acc5_pos: 100.000 loss: 0.7165 (0.7849) acc1: 58.5938 (56.3793) acc5: 100.0000 (100.0000) acc1_neg: 55.211 acc5_neg: 100.000 time: 0.1459 data: 0.0511 max mem: 3532 +Test: Total time: 0:01:04 (0.1947 s / it) +* Overall Acc@1 56.379 Acc@5 100.000 + target == real Acc@1 46.590 Acc@5 100.000 + target == fake Acc@1 65.094 Acc@5 100.000 + loss 0.785 + PR-AUC 0.5829 ROC-AUC 0.5801 EER 44.26% (thr=0.5986) +Accuracy of the network on the 127843 test images: 56.4% +[rank0]:[W726 06:10:19.679009497 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 06:10:39.148000 1198594 torch/distributed/run.py:766] +W0726 06:10:39.148000 1198594 torch/distributed/run.py:766] ***************************************** +W0726 06:10:39.148000 1198594 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:10:39.148000 1198594 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 06:11:07.726666140 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:11:07.774016819 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:11:07.860918106 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:11:08.326493307 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:11:08.547286812 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:11:08.582106683 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:11:08.603759982 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:11:08.690125487 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n11_seed9_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n11_classseed9_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:23:43 acc1_pos: 86.458 acc5_pos: 100.000 loss: 0.3706 (0.3706) acc1: 86.4583 (86.4583) acc5: 100.0000 (100.0000) time: 6.0591 data: 4.4587 max mem: 3531 +Test: [ 10/235] eta: 0:02:22 acc1_pos: 87.240 acc5_pos: 100.000 loss: 0.3706 (0.3652) acc1: 86.4583 (86.2926) acc5: 100.0000 (100.0000) time: 0.6350 data: 0.4057 max mem: 3531 +Test: [ 20/235] eta: 0:01:22 acc1_pos: 85.417 acc5_pos: 100.000 loss: 0.3530 (0.3580) acc1: 87.2396 (87.0660) acc5: 100.0000 (100.0000) time: 0.1012 data: 0.0061 max mem: 3531 +Test: [ 30/235] eta: 0:01:11 acc1_pos: 88.021 acc5_pos: 100.000 loss: 0.3441 (0.3558) acc1: 87.5000 (87.2480) acc5: 100.0000 (100.0000) time: 0.1899 data: 0.0935 max mem: 3531 +Test: [ 40/235] eta: 0:01:02 acc1_pos: 86.198 acc5_pos: 100.000 loss: 0.3511 (0.3552) acc1: 87.5000 (87.1443) acc5: 100.0000 (100.0000) time: 0.2549 data: 0.1605 max mem: 3531 +Test: [ 50/235] eta: 0:00:56 acc1_pos: 83.854 acc5_pos: 100.000 loss: 0.3512 (0.3573) acc1: 87.2396 (87.1324) acc5: 100.0000 (100.0000) time: 0.2418 data: 0.1497 max mem: 3531 +Test: [ 60/235] eta: 0:00:50 acc1_pos: 87.240 acc5_pos: 100.000 loss: 0.3640 (0.3590) acc1: 86.7188 (86.9920) acc5: 100.0000 (100.0000) time: 0.2216 data: 0.1309 max mem: 3531 +Test: [ 70/235] eta: 0:00:45 acc1_pos: 89.231 acc5_pos: 100.000 loss: 0.3882 (0.4445) acc1: 84.3750 (80.9896) acc5: 100.0000 (100.0000) acc1_neg: 33.073 acc5_neg: 100.000 time: 0.2047 data: 0.1136 max mem: 3531 +Test: [ 80/235] eta: 0:00:41 acc1_pos: 89.231 acc5_pos: 100.000 loss: 1.0528 (0.5210) acc1: 37.7604 (75.6591) acc5: 100.0000 (100.0000) acc1_neg: 35.417 acc5_neg: 100.000 time: 0.2083 data: 0.1169 max mem: 3531 +Test: [ 90/235] eta: 0:00:37 acc1_pos: 89.231 acc5_pos: 100.000 loss: 1.0663 (0.5802) acc1: 37.5000 (71.5030) acc5: 100.0000 (100.0000) acc1_neg: 36.979 acc5_neg: 100.000 time: 0.1917 data: 0.1001 max mem: 3531 +Test: [100/235] eta: 0:00:33 acc1_pos: 89.231 acc5_pos: 100.000 loss: 1.0458 (0.6254) acc1: 38.2812 (68.3143) acc5: 100.0000 (100.0000) acc1_neg: 36.979 acc5_neg: 100.000 time: 0.1612 data: 0.0683 max mem: 3531 +Test: [110/235] eta: 0:00:30 acc1_pos: 89.231 acc5_pos: 100.000 loss: 1.0437 (0.6641) acc1: 38.2812 (65.5757) acc5: 100.0000 (100.0000) acc1_neg: 35.938 acc5_neg: 100.000 time: 0.1564 data: 0.0632 max mem: 3531 +Test: [120/235] eta: 0:00:26 acc1_pos: 89.231 acc5_pos: 100.000 loss: 1.0739 (0.7202) acc1: 33.8542 (62.0028) acc5: 100.0000 (100.0000) acc1_neg: 21.875 acc5_neg: 100.000 time: 0.1690 data: 0.0760 max mem: 3531 +Test: [130/235] eta: 0:00:24 acc1_pos: 89.231 acc5_pos: 100.000 loss: 1.3600 (0.7698) acc1: 19.7917 (58.7687) acc5: 100.0000 (100.0000) acc1_neg: 18.490 acc5_neg: 100.000 time: 0.1650 data: 0.0719 max mem: 3531 +Test: [140/235] eta: 0:00:21 acc1_pos: 89.231 acc5_pos: 100.000 loss: 1.3978 (0.8145) acc1: 19.0104 (55.9453) acc5: 100.0000 (100.0000) acc1_neg: 15.365 acc5_neg: 100.000 time: 0.1663 data: 0.0724 max mem: 3531 +Test: [150/235] eta: 0:00:18 acc1_pos: 89.231 acc5_pos: 100.000 loss: 1.3743 (0.8501) acc1: 19.2708 (53.5665) acc5: 100.0000 (100.0000) acc1_neg: 20.312 acc5_neg: 100.000 time: 0.1726 data: 0.0785 max mem: 3531 +Test: [160/235] eta: 0:00:16 acc1_pos: 89.231 acc5_pos: 100.000 loss: 1.3664 (0.8830) acc1: 19.7917 (51.4897) acc5: 100.0000 (100.0000) acc1_neg: 20.833 acc5_neg: 100.000 time: 0.1936 data: 0.1021 max mem: 3531 +Test: [170/235] eta: 0:00:14 acc1_pos: 89.231 acc5_pos: 100.000 loss: 1.3410 (0.8834) acc1: 21.3542 (51.4605) acc5: 100.0000 (100.0000) acc1_neg: 51.302 acc5_neg: 100.000 time: 0.2025 data: 0.1103 max mem: 3531 +Test: [180/235] eta: 0:00:11 acc1_pos: 89.231 acc5_pos: 100.000 loss: 0.8316 (0.8796) acc1: 55.2083 (51.7452) acc5: 100.0000 (100.0000) acc1_neg: 53.646 acc5_neg: 100.000 time: 0.1502 data: 0.0554 max mem: 3531 +Test: [190/235] eta: 0:00:09 acc1_pos: 89.231 acc5_pos: 100.000 loss: 0.8006 (0.8759) acc1: 57.0312 (51.9906) acc5: 100.0000 (100.0000) acc1_neg: 54.688 acc5_neg: 100.000 time: 0.1545 data: 0.0574 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 89.231 acc5_pos: 100.000 loss: 0.7988 (0.8728) acc1: 57.0312 (52.1909) acc5: 100.0000 (100.0000) acc1_neg: 57.031 acc5_neg: 100.000 time: 0.1897 data: 0.0921 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 89.231 acc5_pos: 100.000 loss: 0.8275 (0.8705) acc1: 55.7292 (52.3907) acc5: 100.0000 (100.0000) acc1_neg: 59.115 acc5_neg: 100.000 time: 0.1721 data: 0.0777 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 89.231 acc5_pos: 100.000 loss: 0.8207 (0.8687) acc1: 55.7292 (52.5323) acc5: 100.0000 (100.0000) acc1_neg: 54.948 acc5_neg: 100.000 time: 0.1639 data: 0.0716 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 89.231 acc5_pos: 100.000 loss: 0.8114 (0.8659) acc1: 56.7708 (52.7135) acc5: 100.0000 (100.0000) acc1_neg: 59.635 acc5_neg: 100.000 time: 0.1300 data: 0.0382 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 89.231 acc5_pos: 100.000 loss: 0.8207 (0.8653) acc1: 56.5104 (52.7665) acc5: 100.0000 (100.0000) acc1_neg: 52.174 acc5_neg: 100.000 time: 0.1115 data: 0.0234 max mem: 3532 +Test: Total time: 0:00:47 (0.2004 s / it) +* Overall Acc@1 52.766 Acc@5 100.000 + target == real Acc@1 86.994 Acc@5 100.000 + target == fake Acc@1 40.387 Acc@5 100.000 + loss 0.865 + PR-AUC 0.8671 ROC-AUC 0.7094 EER 34.51% (thr=0.2534) +Accuracy of the network on the 89879 test images: 52.8% +[rank0]:[W726 06:12:06.018828410 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 06:12:25.132000 1199153 torch/distributed/run.py:766] +W0726 06:12:25.132000 1199153 torch/distributed/run.py:766] ***************************************** +W0726 06:12:25.132000 1199153 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:12:25.132000 1199153 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:12:53.681055065 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:12:53.836668080 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +[rank0]:[W726 06:12:54.986446422 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:12:54.010075122 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +| distributed init (rank 7): env:// +| distributed init (rank 6): env:// +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:12:54.490554875 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank4]:[W726 06:12:54.490590632 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W726 06:12:54.490605569 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W726 06:12:54.490620049 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_real_n11_seed9_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n11_classseed9_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:21:37 acc1_pos: 24.219 acc5_pos: 100.000 loss: 1.2939 (1.2939) acc1: 24.2188 (24.2188) acc5: 100.0000 (100.0000) time: 4.8221 data: 3.2614 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/269] eta: 0:02:14 acc1_pos: 36.719 acc5_pos: 100.000 loss: 0.9497 (0.9640) acc1: 36.9792 (41.2169) acc5: 100.0000 (100.0000) time: 0.5202 data: 0.2969 max mem: 3531 +Test: [ 20/269] eta: 0:01:19 acc1_pos: 41.927 acc5_pos: 100.000 loss: 0.9497 (0.9755) acc1: 36.9792 (40.4886) acc5: 100.0000 (100.0000) time: 0.0959 data: 0.0055 max mem: 3531 +Test: [ 30/269] eta: 0:01:05 acc1_pos: 34.115 acc5_pos: 100.000 loss: 0.9737 (0.9815) acc1: 37.5000 (41.0618) acc5: 100.0000 (100.0000) time: 0.1359 data: 0.0445 max mem: 3531 +Test: [ 40/269] eta: 0:00:58 acc1_pos: 68.966 acc5_pos: 100.000 loss: 0.5938 (0.8436) acc1: 70.3125 (51.3465) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.1854 data: 0.0937 max mem: 3531 +Test: [ 50/269] eta: 0:00:55 acc1_pos: 68.966 acc5_pos: 100.000 loss: 0.4111 (0.7691) acc1: 83.5938 (56.8474) acc5: 100.0000 (100.0000) acc1_neg: 60.156 acc5_neg: 100.000 time: 0.2199 data: 0.1273 max mem: 3531 +Test: [ 60/269] eta: 0:00:49 acc1_pos: 68.966 acc5_pos: 100.000 loss: 0.5172 (0.7337) acc1: 71.6146 (58.9694) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.2044 data: 0.1099 max mem: 3531 +Test: [ 70/269] eta: 0:00:44 acc1_pos: 68.966 acc5_pos: 100.000 loss: 0.5206 (0.6988) acc1: 71.6146 (61.3116) acc5: 100.0000 (100.0000) acc1_neg: 74.740 acc5_neg: 100.000 time: 0.1601 data: 0.0654 max mem: 3531 +Test: [ 80/269] eta: 0:00:42 acc1_pos: 68.966 acc5_pos: 100.000 loss: 0.4495 (0.6661) acc1: 79.6875 (63.6510) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1749 data: 0.0789 max mem: 3531 +Test: [ 90/269] eta: 0:00:39 acc1_pos: 68.966 acc5_pos: 100.000 loss: 0.4495 (0.6558) acc1: 79.6875 (64.2170) acc5: 100.0000 (100.0000) acc1_neg: 73.698 acc5_neg: 100.000 time: 0.2047 data: 0.1083 max mem: 3531 +Test: [100/269] eta: 0:00:36 acc1_pos: 68.966 acc5_pos: 100.000 loss: 0.4886 (0.6405) acc1: 74.7396 (65.3594) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.1899 data: 0.0950 max mem: 3531 +Test: [110/269] eta: 0:00:33 acc1_pos: 68.966 acc5_pos: 100.000 loss: 0.5437 (0.6359) acc1: 69.5312 (65.4185) acc5: 100.0000 (100.0000) acc1_neg: 53.125 acc5_neg: 100.000 time: 0.1687 data: 0.0747 max mem: 3531 +Test: [120/269] eta: 0:00:31 acc1_pos: 68.966 acc5_pos: 100.000 loss: 0.5466 (0.6213) acc1: 68.2292 (66.4579) acc5: 100.0000 (100.0000) acc1_neg: 68.229 acc5_neg: 100.000 time: 0.1716 data: 0.0751 max mem: 3531 +Test: [130/269] eta: 0:00:28 acc1_pos: 68.966 acc5_pos: 100.000 loss: 0.4079 (0.6096) acc1: 82.2917 (67.3247) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.1761 data: 0.0769 max mem: 3531 +Test: [140/269] eta: 0:00:26 acc1_pos: 68.966 acc5_pos: 100.000 loss: 0.5127 (0.6099) acc1: 73.6979 (66.9677) acc5: 100.0000 (100.0000) acc1_neg: 42.188 acc5_neg: 100.000 time: 0.1932 data: 0.0956 max mem: 3531 +Test: [150/269] eta: 0:00:24 acc1_pos: 68.966 acc5_pos: 100.000 loss: 0.6017 (0.6069) acc1: 54.1667 (66.8529) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1849 data: 0.0882 max mem: 3531 +Test: [160/269] eta: 0:00:21 acc1_pos: 68.966 acc5_pos: 100.000 loss: 0.4039 (0.5944) acc1: 82.0312 (67.8571) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1599 data: 0.0648 max mem: 3531 +Test: [170/269] eta: 0:00:19 acc1_pos: 68.966 acc5_pos: 100.000 loss: 0.4072 (0.5856) acc1: 83.5938 (68.6358) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1724 data: 0.0772 max mem: 3531 +Test: [180/269] eta: 0:00:17 acc1_pos: 68.966 acc5_pos: 100.000 loss: 0.3908 (0.5720) acc1: 84.8958 (69.6679) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.1711 data: 0.0773 max mem: 3531 +Test: [190/269] eta: 0:00:15 acc1_pos: 68.966 acc5_pos: 100.000 loss: 0.3887 (0.5665) acc1: 83.3333 (70.0766) acc5: 100.0000 (100.0000) acc1_neg: 77.344 acc5_neg: 100.000 time: 0.1594 data: 0.0675 max mem: 3531 +Test: [200/269] eta: 0:00:13 acc1_pos: 68.966 acc5_pos: 100.000 loss: 0.4796 (0.5644) acc1: 75.5208 (70.1739) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.1638 data: 0.0705 max mem: 3531 +Test: [210/269] eta: 0:00:11 acc1_pos: 68.966 acc5_pos: 100.000 loss: 0.4879 (0.5603) acc1: 71.0938 (70.3421) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.2079 data: 0.1117 max mem: 3532 +Test: [220/269] eta: 0:00:09 acc1_pos: 68.966 acc5_pos: 100.000 loss: 0.4719 (0.5547) acc1: 74.7396 (70.7320) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1948 data: 0.0988 max mem: 3532 +Test: [230/269] eta: 0:00:07 acc1_pos: 68.966 acc5_pos: 100.000 loss: 0.2910 (0.5415) acc1: 92.1875 (71.7611) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1299 data: 0.0367 max mem: 3532 +Test: [240/269] eta: 0:00:05 acc1_pos: 68.966 acc5_pos: 100.000 loss: 0.3466 (0.5450) acc1: 83.5938 (71.5433) acc5: 100.0000 (100.0000) acc1_neg: 57.552 acc5_neg: 100.000 time: 0.1499 data: 0.0554 max mem: 3532 +Test: [250/269] eta: 0:00:03 acc1_pos: 68.966 acc5_pos: 100.000 loss: 0.6106 (0.5482) acc1: 63.5417 (71.1923) acc5: 100.0000 (100.0000) acc1_neg: 46.354 acc5_neg: 100.000 time: 0.1799 data: 0.0841 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 68.966 acc5_pos: 100.000 loss: 0.6344 (0.5534) acc1: 59.8958 (70.6717) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.1749 data: 0.0802 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 68.966 acc5_pos: 100.000 loss: 0.6283 (0.5549) acc1: 60.4167 (70.5159) acc5: 100.0000 (100.0000) acc1_neg: 76.378 acc5_neg: 100.000 time: 0.1568 data: 0.0600 max mem: 3532 +Test: Total time: 0:00:50 (0.1889 s / it) +* Overall Acc@1 70.516 Acc@5 100.000 + target == real Acc@1 41.366 Acc@5 100.000 + target == fake Acc@1 74.494 Acc@5 100.000 + loss 0.555 + PR-AUC 0.9125 ROC-AUC 0.6199 EER 41.04% (thr=0.6367) +Accuracy of the network on the 103293 test images: 70.5% +[rank0]:[W726 06:13:59.150964057 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T06:14:06Z] eval_done tag=fixed500k_real_n11_seed9_resampled +[2026-07-26T06:14:06Z] eval_start tag=fixed500k_fake_n22_seed9_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed9_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth +started_utc=2026-07-26T06:14:06Z +hostname=n116-041-188 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed9_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n22_seed9_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0726 06:14:19.441000 1199729 torch/distributed/run.py:766] +W0726 06:14:19.441000 1199729 torch/distributed/run.py:766] ***************************************** +W0726 06:14:19.441000 1199729 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:14:19.441000 1199729 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 06:14:49.833120663 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:14:50.158422126 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:14:50.241107934 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:14:50.489217066 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:14:50.501810427 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:14:50.637943893 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:14:50.659903958 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:14:50.809395979 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n22_seed9_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed9_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:22:16 acc1_pos: 60.677 acc5_pos: 100.000 loss: 0.6557 (0.6557) acc1: 60.6771 (60.6771) acc5: 100.0000 (100.0000) time: 5.7112 data: 4.0944 max mem: 3531 +Test: [ 10/234] eta: 0:02:14 acc1_pos: 28.646 acc5_pos: 100.000 loss: 0.8172 (0.7972) acc1: 47.6562 (45.2888) acc5: 100.0000 (100.0000) time: 0.6013 data: 0.3727 max mem: 3531 +Test: [ 20/234] eta: 0:01:19 acc1_pos: 94.010 acc5_pos: 100.000 loss: 0.7445 (0.7207) acc1: 46.6146 (49.6652) acc5: 100.0000 (100.0000) time: 0.1041 data: 0.0114 max mem: 3531 +Test: [ 30/234] eta: 0:01:04 acc1_pos: 36.979 acc5_pos: 100.000 loss: 0.2300 (0.5956) acc1: 94.0104 (61.0551) acc5: 100.0000 (100.0000) time: 0.1562 data: 0.0613 max mem: 3531 +Test: [ 40/234] eta: 0:00:56 acc1_pos: 97.135 acc5_pos: 100.000 loss: 0.2173 (0.6154) acc1: 94.2708 (60.3595) acc5: 100.0000 (100.0000) time: 0.2054 data: 0.1097 max mem: 3531 +Test: [ 50/234] eta: 0:00:51 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1393 (0.5218) acc1: 97.6562 (67.7849) acc5: 100.0000 (100.0000) time: 0.2255 data: 0.1315 max mem: 3531 +Test: [ 60/234] eta: 0:00:46 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.1376 (0.4591) acc1: 98.4375 (73.0063) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2150 data: 0.1245 max mem: 3531 +Test: [ 70/234] eta: 0:00:41 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.1443 (0.4954) acc1: 99.4792 (71.5889) acc5: 100.0000 (100.0000) acc1_neg: 40.885 acc5_neg: 100.000 time: 0.1820 data: 0.0885 max mem: 3531 +Test: [ 80/234] eta: 0:00:37 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.9824 (0.5238) acc1: 46.3542 (70.5761) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1757 data: 0.0820 max mem: 3531 +Test: [ 90/234] eta: 0:00:34 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.2322 (0.4885) acc1: 92.7083 (73.2772) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.1913 data: 0.1005 max mem: 3531 +Test: [100/234] eta: 0:00:30 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.2322 (0.4787) acc1: 94.0104 (74.3116) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1790 data: 0.0873 max mem: 3531 +Test: [110/234] eta: 0:00:27 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.4423 (0.4718) acc1: 79.4271 (75.1830) acc5: 100.0000 (100.0000) acc1_neg: 72.135 acc5_neg: 100.000 time: 0.1584 data: 0.0632 max mem: 3531 +Test: [120/234] eta: 0:00:25 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.2997 (0.4706) acc1: 85.4167 (75.8114) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1696 data: 0.0729 max mem: 3531 +Test: [130/234] eta: 0:00:22 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.3815 (0.4668) acc1: 90.1042 (76.6420) acc5: 100.0000 (100.0000) acc1_neg: 74.219 acc5_neg: 100.000 time: 0.1918 data: 0.0967 max mem: 3531 +Test: [140/234] eta: 0:00:20 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.4244 (0.4941) acc1: 83.5938 (75.4229) acc5: 100.0000 (100.0000) acc1_neg: 31.250 acc5_neg: 100.000 time: 0.1906 data: 0.0962 max mem: 3531 +Test: [150/234] eta: 0:00:18 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.4345 (0.4968) acc1: 75.7812 (75.4415) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.2093 data: 0.1173 max mem: 3531 +Test: [160/234] eta: 0:00:15 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.4175 (0.4901) acc1: 82.0312 (76.1678) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.1939 data: 0.0978 max mem: 3531 +Test: [170/234] eta: 0:00:13 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.4710 (0.4882) acc1: 80.7292 (76.5001) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1441 data: 0.0452 max mem: 3531 +Test: [180/234] eta: 0:00:11 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.4837 (0.4989) acc1: 77.6042 (75.9481) acc5: 100.0000 (100.0000) acc1_neg: 35.677 acc5_neg: 100.000 time: 0.1620 data: 0.0652 max mem: 3531 +Test: [190/234] eta: 0:00:09 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.4815 (0.4997) acc1: 77.6042 (75.9899) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.1803 data: 0.0858 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.4614 (0.4871) acc1: 80.2083 (76.8216) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1888 data: 0.0957 max mem: 3531 +Test: [210/234] eta: 0:00:04 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.1205 (0.4710) acc1: 100.0000 (77.8510) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1896 data: 0.0956 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.1899 (0.4593) acc1: 96.8750 (78.6411) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.1539 data: 0.0591 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.3586 (0.4697) acc1: 61.4583 (77.7440) acc5: 100.0000 (100.0000) acc1_neg: 55.729 acc5_neg: 100.000 time: 0.1295 data: 0.0338 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 97.241 acc5_pos: 100.000 loss: 0.6588 (0.4725) acc1: 60.4167 (77.5301) acc5: 100.0000 (100.0000) acc1_neg: 59.922 acc5_neg: 100.000 time: 0.1297 data: 0.0338 max mem: 3532 +Test: Total time: 0:00:46 (0.1977 s / it) +* Overall Acc@1 77.530 Acc@5 100.000 + target == real Acc@1 68.001 Acc@5 100.000 + target == fake Acc@1 80.216 Acc@5 100.000 + loss 0.472 + PR-AUC 0.9539 ROC-AUC 0.8520 EER 24.00% (thr=0.5718) +Accuracy of the network on the 89729 test images: 77.5% +[rank0]:[W726 06:15:48.140463097 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 06:16:07.025000 1200278 torch/distributed/run.py:766] +W0726 06:16:07.025000 1200278 torch/distributed/run.py:766] ***************************************** +W0726 06:16:07.025000 1200278 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:16:07.025000 1200278 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:16:35.480250311 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:16:35.820502287 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:16:35.884687971 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:16:36.350406002 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +[rank0]:[W726 06:16:36.420964348 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +[rank5]:[W726 06:16:36.424136958 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:16:36.434883239 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W726 06:16:36.435289238 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n22_seed9_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed9_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:50:50 acc1_pos: 23.698 acc5_pos: 100.000 loss: 0.9981 (0.9981) acc1: 23.6979 (23.6979) acc5: 100.0000 (100.0000) time: 7.3108 data: 5.8767 max mem: 3531 +Test: [ 10/2387] eta: 0:29:44 acc1_pos: 1.302 acc5_pos: 100.000 loss: 1.1853 (1.1507) acc1: 9.6354 (11.7188) acc5: 100.0000 (100.0000) time: 0.7508 data: 0.5346 max mem: 3531 +Test: [ 20/2387] eta: 0:17:23 acc1_pos: 5.729 acc5_pos: 100.000 loss: 1.1986 (1.1965) acc1: 9.6354 (11.1359) acc5: 100.0000 (100.0000) time: 0.0972 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:15:02 acc1_pos: 2.083 acc5_pos: 100.000 loss: 1.2146 (1.2005) acc1: 9.6354 (10.4587) acc5: 100.0000 (100.0000) time: 0.1808 data: 0.0863 max mem: 3531 +Test: [ 40/2387] eta: 0:13:42 acc1_pos: 9.115 acc5_pos: 100.000 loss: 1.1558 (1.1843) acc1: 7.5521 (9.8260) acc5: 100.0000 (100.0000) time: 0.2553 data: 0.1653 max mem: 3531 +Test: [ 50/2387] eta: 0:12:38 acc1_pos: 27.604 acc5_pos: 100.000 loss: 1.1078 (1.1668) acc1: 8.0729 (10.9528) acc5: 100.0000 (100.0000) time: 0.2338 data: 0.1437 max mem: 3531 +Test: [ 60/2387] eta: 0:11:56 acc1_pos: 18.490 acc5_pos: 100.000 loss: 1.0794 (1.1526) acc1: 15.1042 (11.9536) acc5: 100.0000 (100.0000) time: 0.2207 data: 0.1262 max mem: 3531 +Test: [ 70/2387] eta: 0:11:27 acc1_pos: 25.000 acc5_pos: 100.000 loss: 1.1210 (1.1532) acc1: 14.3229 (12.1992) acc5: 100.0000 (100.0000) time: 0.2250 data: 0.1302 max mem: 3531 +Test: [ 80/2387] eta: 0:11:10 acc1_pos: 32.812 acc5_pos: 100.000 loss: 1.0995 (1.1306) acc1: 18.2292 (13.6124) acc5: 100.0000 (100.0000) time: 0.2388 data: 0.1469 max mem: 3531 +Test: [ 90/2387] eta: 0:10:49 acc1_pos: 37.240 acc5_pos: 100.000 loss: 0.8711 (1.0979) acc1: 28.9062 (15.8597) acc5: 100.0000 (100.0000) time: 0.2348 data: 0.1417 max mem: 3531 +Test: [ 100/2387] eta: 0:10:40 acc1_pos: 29.688 acc5_pos: 100.000 loss: 0.8344 (1.0809) acc1: 29.1667 (16.8265) acc5: 100.0000 (100.0000) time: 0.2366 data: 0.1405 max mem: 3531 +Test: [ 110/2387] eta: 0:10:24 acc1_pos: 32.552 acc5_pos: 100.000 loss: 0.8704 (1.0620) acc1: 28.3854 (18.0251) acc5: 100.0000 (100.0000) time: 0.2349 data: 0.1377 max mem: 3531 +Test: [ 120/2387] eta: 0:10:15 acc1_pos: 20.573 acc5_pos: 100.000 loss: 0.8875 (1.0503) acc1: 30.9896 (18.9760) acc5: 100.0000 (100.0000) time: 0.2283 data: 0.1352 max mem: 3531 +Test: [ 130/2387] eta: 0:10:06 acc1_pos: 23.438 acc5_pos: 100.000 loss: 0.9056 (1.0396) acc1: 26.3021 (19.3523) acc5: 100.0000 (100.0000) time: 0.2365 data: 0.1447 max mem: 3531 +Test: [ 140/2387] eta: 0:10:00 acc1_pos: 25.260 acc5_pos: 100.000 loss: 0.8982 (1.0300) acc1: 25.2604 (19.8748) acc5: 100.0000 (100.0000) time: 0.2405 data: 0.1480 max mem: 3531 +Test: [ 150/2387] eta: 0:09:51 acc1_pos: 9.115 acc5_pos: 100.000 loss: 0.8519 (1.0179) acc1: 28.3854 (20.5971) acc5: 100.0000 (100.0000) time: 0.2386 data: 0.1451 max mem: 3531 +Test: [ 160/2387] eta: 0:09:47 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.8390 (0.9953) acc1: 31.5104 (22.6223) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2393 data: 0.1460 max mem: 3531 +Test: [ 170/2387] eta: 0:09:27 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2469 (0.9511) acc1: 97.6562 (27.0529) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1896 data: 0.0969 max mem: 3531 +Test: [ 180/2387] eta: 0:09:16 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2370 (0.9116) acc1: 98.1771 (30.9910) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1599 data: 0.0653 max mem: 3531 +Test: [ 190/2387] eta: 0:09:03 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2370 (0.8763) acc1: 98.4375 (34.5250) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1749 data: 0.0779 max mem: 3531 +Test: [ 200/2387] eta: 0:08:53 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2333 (0.8444) acc1: 98.6979 (37.7125) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1699 data: 0.0742 max mem: 3531 +Test: [ 210/2387] eta: 0:08:47 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2378 (0.8158) acc1: 98.4375 (40.5793) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1962 data: 0.1031 max mem: 3532 +Test: [ 220/2387] eta: 0:08:41 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2405 (0.7897) acc1: 98.4375 (43.2032) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2049 data: 0.1124 max mem: 3532 +Test: [ 230/2387] eta: 0:08:33 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2377 (0.7658) acc1: 98.9583 (45.6124) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1898 data: 0.0980 max mem: 3532 +Test: [ 240/2387] eta: 0:08:25 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2477 (0.7461) acc1: 98.1771 (47.7535) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1779 data: 0.0857 max mem: 3532 +Test: [ 250/2387] eta: 0:08:16 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2368 (0.7220) acc1: 97.9167 (49.8205) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1686 data: 0.0763 max mem: 3532 +Test: [ 260/2387] eta: 0:08:10 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1114 (0.6984) acc1: 100.0000 (51.7431) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1718 data: 0.0794 max mem: 3532 +Test: [ 270/2387] eta: 0:08:05 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1081 (0.6766) acc1: 100.0000 (53.5228) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0969 max mem: 3532 +Test: [ 280/2387] eta: 0:07:56 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1120 (0.6566) acc1: 100.0000 (55.1768) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1706 data: 0.0783 max mem: 3532 +Test: [ 290/2387] eta: 0:07:52 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1168 (0.6382) acc1: 100.0000 (56.7171) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0821 max mem: 3532 +Test: [ 300/2387] eta: 0:07:48 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1231 (0.6210) acc1: 100.0000 (58.1542) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2042 data: 0.1090 max mem: 3532 +Test: [ 310/2387] eta: 0:07:44 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1307 (0.6088) acc1: 100.0000 (59.4403) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1949 data: 0.1010 max mem: 3532 +Test: [ 320/2387] eta: 0:07:43 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.3680 (0.6015) acc1: 96.3542 (60.5635) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2158 data: 0.1249 max mem: 3532 +Test: [ 330/2387] eta: 0:07:38 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.3736 (0.5945) acc1: 95.8333 (61.6291) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2110 data: 0.1176 max mem: 3532 +Test: [ 340/2387] eta: 0:07:33 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.3681 (0.5879) acc1: 95.3125 (62.6138) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1796 data: 0.0841 max mem: 3532 +Test: [ 350/2387] eta: 0:07:29 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.3593 (0.5777) acc1: 96.6146 (63.6626) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1888 data: 0.0947 max mem: 3532 +Test: [ 360/2387] eta: 0:07:25 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2113 (0.5675) acc1: 99.7396 (64.6504) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1899 data: 0.0972 max mem: 3532 +Test: [ 370/2387] eta: 0:07:20 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2099 (0.5579) acc1: 98.9583 (65.5745) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1749 data: 0.0821 max mem: 3532 +Test: [ 380/2387] eta: 0:07:14 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2099 (0.5487) acc1: 99.2188 (66.4616) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1593 data: 0.0631 max mem: 3532 +Test: [ 390/2387] eta: 0:07:09 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2101 (0.5401) acc1: 99.4792 (67.3034) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1569 data: 0.0591 max mem: 3532 +Test: [ 400/2387] eta: 0:07:06 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2089 (0.5318) acc1: 99.2188 (68.1038) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1788 data: 0.0836 max mem: 3532 +Test: [ 410/2387] eta: 0:07:03 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2075 (0.5239) acc1: 99.2188 (68.8634) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1901 data: 0.0978 max mem: 3532 +Test: [ 420/2387] eta: 0:07:01 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2058 (0.5163) acc1: 99.2188 (69.5906) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2011 data: 0.1064 max mem: 3532 +Test: [ 430/2387] eta: 0:06:56 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2084 (0.5099) acc1: 99.2188 (70.2702) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1886 data: 0.0926 max mem: 3532 +Test: [ 440/2387] eta: 0:06:54 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2108 (0.5024) acc1: 98.9583 (70.9308) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1864 data: 0.0939 max mem: 3533 +Test: [ 450/2387] eta: 0:06:50 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1208 (0.4939) acc1: 100.0000 (71.5753) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1948 data: 0.1041 max mem: 3533 +Test: [ 460/2387] eta: 0:06:46 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1196 (0.4857) acc1: 100.0000 (72.1919) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1755 data: 0.0843 max mem: 3533 +Test: [ 470/2387] eta: 0:06:44 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1202 (0.4780) acc1: 100.0000 (72.7823) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1842 data: 0.0908 max mem: 3533 +Test: [ 480/2387] eta: 0:06:40 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1205 (0.4706) acc1: 100.0000 (73.3482) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1879 data: 0.0927 max mem: 3533 +Test: [ 490/2387] eta: 0:06:38 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1202 (0.4634) acc1: 100.0000 (73.8910) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1854 data: 0.0913 max mem: 3533 +Test: [ 500/2387] eta: 0:06:34 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1202 (0.4566) acc1: 100.0000 (74.4116) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0917 max mem: 3533 +Test: [ 510/2387] eta: 0:06:31 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1213 (0.4503) acc1: 100.0000 (74.9123) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1806 data: 0.0884 max mem: 3533 +Test: [ 520/2387] eta: 0:06:28 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1612 (0.4450) acc1: 100.0000 (75.3939) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1839 data: 0.0914 max mem: 3533 +Test: [ 530/2387] eta: 0:06:25 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1612 (0.4396) acc1: 100.0000 (75.8573) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0918 max mem: 3533 +Test: [ 540/2387] eta: 0:06:22 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1609 (0.4345) acc1: 100.0000 (76.3035) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1760 data: 0.0826 max mem: 3533 +Test: [ 550/2387] eta: 0:06:19 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1601 (0.4295) acc1: 100.0000 (76.7336) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1789 data: 0.0864 max mem: 3533 +Test: [ 560/2387] eta: 0:06:17 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1597 (0.4247) acc1: 100.0000 (77.1483) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1894 data: 0.0976 max mem: 3533 +Test: [ 570/2387] eta: 0:06:14 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1596 (0.4200) acc1: 100.0000 (77.5485) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1905 data: 0.0957 max mem: 3533 +Test: [ 580/2387] eta: 0:06:12 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1596 (0.4156) acc1: 100.0000 (77.9345) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1926 data: 0.0983 max mem: 3533 +Test: [ 590/2387] eta: 0:06:09 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1609 (0.4113) acc1: 100.0000 (78.3079) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1942 data: 0.1011 max mem: 3533 +Test: [ 600/2387] eta: 0:06:07 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1660 (0.4074) acc1: 100.0000 (78.6688) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1890 data: 0.0959 max mem: 3533 +Test: [ 610/2387] eta: 0:06:04 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1709 (0.4034) acc1: 100.0000 (79.0166) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1844 data: 0.0928 max mem: 3533 +Test: [ 620/2387] eta: 0:06:01 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1651 (0.3996) acc1: 100.0000 (79.3524) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1876 data: 0.0960 max mem: 3533 +Test: [ 630/2387] eta: 0:05:59 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1660 (0.3959) acc1: 100.0000 (79.6772) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1868 data: 0.0960 max mem: 3533 +Test: [ 640/2387] eta: 0:05:56 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1665 (0.3923) acc1: 99.7396 (79.9910) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1808 data: 0.0895 max mem: 3533 +Test: [ 650/2387] eta: 0:05:54 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1657 (0.3888) acc1: 100.0000 (80.2971) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1904 data: 0.0984 max mem: 3533 +Test: [ 660/2387] eta: 0:05:51 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1671 (0.3855) acc1: 100.0000 (80.5952) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1900 data: 0.0983 max mem: 3533 +Test: [ 670/2387] eta: 0:05:49 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1666 (0.3822) acc1: 100.0000 (80.8844) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1819 data: 0.0902 max mem: 3534 +Test: [ 680/2387] eta: 0:05:46 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1666 (0.3801) acc1: 100.0000 (81.1643) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1772 data: 0.0858 max mem: 3534 +Test: [ 690/2387] eta: 0:05:43 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2753 (0.3796) acc1: 99.7396 (81.2726) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1719 data: 0.0804 max mem: 3534 +Test: [ 700/2387] eta: 0:05:41 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.3677 (0.3795) acc1: 84.3750 (81.3198) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.1826 data: 0.0907 max mem: 3534 +Test: [ 710/2387] eta: 0:05:38 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.3674 (0.3794) acc1: 84.6354 (81.3617) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.1856 data: 0.0924 max mem: 3534 +Test: [ 720/2387] eta: 0:05:36 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.3656 (0.3793) acc1: 84.6354 (81.4013) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.1841 data: 0.0911 max mem: 3534 +Test: [ 730/2387] eta: 0:05:33 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.3748 (0.3793) acc1: 84.1146 (81.4288) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.1832 data: 0.0912 max mem: 3534 +Test: [ 740/2387] eta: 0:05:31 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.3793 (0.3793) acc1: 83.3333 (81.4623) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.1847 data: 0.0929 max mem: 3534 +Test: [ 750/2387] eta: 0:05:29 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.3724 (0.3792) acc1: 84.3750 (81.5042) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1890 data: 0.0981 max mem: 3534 +Test: [ 760/2387] eta: 0:05:27 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.3726 (0.3791) acc1: 84.6354 (81.5392) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.1949 data: 0.1029 max mem: 3534 +Test: [ 770/2387] eta: 0:05:24 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.3601 (0.3779) acc1: 84.8958 (81.6634) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1874 data: 0.0934 max mem: 3534 +Test: [ 780/2387] eta: 0:05:22 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2084 (0.3756) acc1: 96.3542 (81.8582) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0862 max mem: 3534 +Test: [ 790/2387] eta: 0:05:19 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1829 (0.3736) acc1: 98.4375 (82.0125) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.1778 data: 0.0854 max mem: 3534 +Test: [ 800/2387] eta: 0:05:17 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1519 (0.3709) acc1: 99.2188 (82.2279) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1770 data: 0.0860 max mem: 3534 +Test: [ 810/2387] eta: 0:05:15 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1731 (0.3686) acc1: 99.2188 (82.4333) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1814 data: 0.0907 max mem: 3534 +Test: [ 820/2387] eta: 0:05:12 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1958 (0.3667) acc1: 99.4792 (82.6225) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1835 data: 0.0921 max mem: 3534 +Test: [ 830/2387] eta: 0:05:10 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2354 (0.3661) acc1: 96.6146 (82.7379) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1855 data: 0.0922 max mem: 3534 +Test: [ 840/2387] eta: 0:05:08 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2333 (0.3643) acc1: 97.1354 (82.9354) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1845 data: 0.0879 max mem: 3534 +Test: [ 850/2387] eta: 0:05:06 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2069 (0.3623) acc1: 99.7396 (83.1298) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1874 data: 0.0905 max mem: 3534 +Test: [ 860/2387] eta: 0:05:03 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1908 (0.3603) acc1: 100.0000 (83.3212) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1895 data: 0.0945 max mem: 3534 +Test: [ 870/2387] eta: 0:05:01 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1922 (0.3585) acc1: 99.7396 (83.5073) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1818 data: 0.0888 max mem: 3534 +Test: [ 880/2387] eta: 0:04:59 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2120 (0.3570) acc1: 99.4792 (83.6875) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0864 max mem: 3534 +Test: [ 890/2387] eta: 0:04:57 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2244 (0.3556) acc1: 99.4792 (83.8588) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1942 data: 0.1007 max mem: 3534 +Test: [ 900/2387] eta: 0:04:55 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2685 (0.3610) acc1: 94.7917 (83.5391) acc5: 100.0000 (100.0000) acc1_neg: 46.354 acc5_neg: 100.000 time: 0.1899 data: 0.0979 max mem: 3535 +Test: [ 910/2387] eta: 0:04:52 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.9683 (0.3678) acc1: 45.5729 (83.1192) acc5: 100.0000 (100.0000) acc1_neg: 42.969 acc5_neg: 100.000 time: 0.1853 data: 0.0936 max mem: 3535 +Test: [ 920/2387] eta: 0:04:50 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.9787 (0.3744) acc1: 44.7917 (82.6943) acc5: 100.0000 (100.0000) acc1_neg: 37.760 acc5_neg: 100.000 time: 0.1919 data: 0.1010 max mem: 3535 +Test: [ 930/2387] eta: 0:04:49 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.9717 (0.3809) acc1: 44.5312 (82.2869) acc5: 100.0000 (100.0000) acc1_neg: 40.625 acc5_neg: 100.000 time: 0.2000 data: 0.1069 max mem: 3535 +Test: [ 940/2387] eta: 0:04:47 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.9803 (0.3873) acc1: 44.5312 (81.8807) acc5: 100.0000 (100.0000) acc1_neg: 43.490 acc5_neg: 100.000 time: 0.2033 data: 0.1075 max mem: 3535 +Test: [ 950/2387] eta: 0:04:44 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.9867 (0.3936) acc1: 43.4896 (81.4797) acc5: 100.0000 (100.0000) acc1_neg: 48.177 acc5_neg: 100.000 time: 0.1838 data: 0.0894 max mem: 3535 +Test: [ 960/2387] eta: 0:04:42 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.9635 (0.3995) acc1: 44.2708 (81.1096) acc5: 100.0000 (100.0000) acc1_neg: 49.219 acc5_neg: 100.000 time: 0.1758 data: 0.0836 max mem: 3535 +Test: [ 970/2387] eta: 0:04:40 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.9646 (0.4055) acc1: 44.2708 (80.7243) acc5: 100.0000 (100.0000) acc1_neg: 43.229 acc5_neg: 100.000 time: 0.1762 data: 0.0842 max mem: 3535 +Test: [ 980/2387] eta: 0:04:37 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.9516 (0.4066) acc1: 45.5729 (80.6907) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1735 data: 0.0815 max mem: 3535 +Test: [ 990/2387] eta: 0:04:35 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2478 (0.4049) acc1: 93.7500 (80.8366) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1679 data: 0.0739 max mem: 3535 +Test: [1000/2387] eta: 0:04:33 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2076 (0.4031) acc1: 97.3958 (80.9888) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1821 data: 0.0877 max mem: 3535 +Test: [1010/2387] eta: 0:04:31 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2168 (0.4016) acc1: 96.8750 (81.1325) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1968 data: 0.1037 max mem: 3535 +Test: [1020/2387] eta: 0:04:29 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2331 (0.4004) acc1: 97.1354 (81.2607) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1943 data: 0.1021 max mem: 3535 +Test: [1030/2387] eta: 0:04:26 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2019 (0.3984) acc1: 99.2188 (81.4351) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1839 data: 0.0928 max mem: 3535 +Test: [1040/2387] eta: 0:04:24 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1738 (0.3961) acc1: 99.4792 (81.6082) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1761 data: 0.0852 max mem: 3535 +Test: [1050/2387] eta: 0:04:22 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1604 (0.3939) acc1: 99.7396 (81.7795) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1756 data: 0.0845 max mem: 3535 +Test: [1060/2387] eta: 0:04:20 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1759 (0.3919) acc1: 100.0000 (81.9471) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1741 data: 0.0799 max mem: 3535 +Test: [1070/2387] eta: 0:04:18 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1871 (0.3902) acc1: 99.4792 (82.1035) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1804 data: 0.0855 max mem: 3535 +Test: [1080/2387] eta: 0:04:15 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2298 (0.3891) acc1: 98.6979 (82.2283) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1736 data: 0.0816 max mem: 3535 +Test: [1090/2387] eta: 0:04:13 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2285 (0.3873) acc1: 97.6562 (82.3704) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1790 data: 0.0873 max mem: 3535 +Test: [1100/2387] eta: 0:04:11 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1957 (0.3855) acc1: 99.2188 (82.5213) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1896 data: 0.0960 max mem: 3535 +Test: [1110/2387] eta: 0:04:09 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1894 (0.3842) acc1: 99.4792 (82.6245) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1903 data: 0.0967 max mem: 3535 +Test: [1120/2387] eta: 0:04:07 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1909 (0.3825) acc1: 98.6979 (82.7605) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1951 data: 0.1029 max mem: 3536 +Test: [1130/2387] eta: 0:04:06 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1990 (0.3810) acc1: 99.2188 (82.9021) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2010 data: 0.1076 max mem: 3536 +Test: [1140/2387] eta: 0:04:04 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2299 (0.3798) acc1: 98.1771 (83.0325) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2091 data: 0.1145 max mem: 3536 +Test: [1150/2387] eta: 0:04:02 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2327 (0.3785) acc1: 98.9583 (83.1616) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2070 data: 0.1109 max mem: 3536 +Test: [1160/2387] eta: 0:04:00 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1683 (0.3765) acc1: 100.0000 (83.3066) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2023 data: 0.1077 max mem: 3536 +Test: [1170/2387] eta: 0:03:58 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1438 (0.3746) acc1: 100.0000 (83.4490) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2024 data: 0.1105 max mem: 3536 +Test: [1180/2387] eta: 0:03:56 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1438 (0.3727) acc1: 100.0000 (83.5856) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2031 data: 0.1114 max mem: 3536 +Test: [1190/2387] eta: 0:03:54 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1679 (0.3711) acc1: 100.0000 (83.7228) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2045 data: 0.1113 max mem: 3536 +Test: [1200/2387] eta: 0:03:52 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1699 (0.3694) acc1: 100.0000 (83.8583) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1920 data: 0.0969 max mem: 3536 +Test: [1210/2387] eta: 0:03:50 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1725 (0.3678) acc1: 100.0000 (83.9916) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1944 data: 0.1009 max mem: 3536 +Test: [1220/2387] eta: 0:03:48 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1204 (0.3657) acc1: 100.0000 (84.1225) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2045 data: 0.1122 max mem: 3536 +Test: [1230/2387] eta: 0:03:47 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1006 (0.3635) acc1: 100.0000 (84.2515) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2002 data: 0.1080 max mem: 3536 +Test: [1240/2387] eta: 0:03:45 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1000 (0.3614) acc1: 100.0000 (84.3781) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1959 data: 0.1045 max mem: 3536 +Test: [1250/2387] eta: 0:03:43 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1046 (0.3594) acc1: 100.0000 (84.5028) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1980 data: 0.1069 max mem: 3536 +Test: [1260/2387] eta: 0:03:41 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1130 (0.3575) acc1: 100.0000 (84.6255) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2032 data: 0.1117 max mem: 3536 +Test: [1270/2387] eta: 0:03:39 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1191 (0.3557) acc1: 100.0000 (84.7463) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2008 data: 0.1094 max mem: 3536 +Test: [1280/2387] eta: 0:03:37 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1428 (0.3541) acc1: 100.0000 (84.8645) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1934 data: 0.1012 max mem: 3536 +Test: [1290/2387] eta: 0:03:35 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1456 (0.3527) acc1: 99.7396 (84.9634) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1952 data: 0.1000 max mem: 3536 +Test: [1300/2387] eta: 0:03:33 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1504 (0.3511) acc1: 99.7396 (85.0714) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2007 data: 0.1046 max mem: 3536 +Test: [1310/2387] eta: 0:03:31 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1505 (0.3497) acc1: 99.4792 (85.1765) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1966 data: 0.1028 max mem: 3536 +Test: [1320/2387] eta: 0:03:29 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1728 (0.3485) acc1: 99.4792 (85.2751) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1896 data: 0.0975 max mem: 3536 +Test: [1330/2387] eta: 0:03:27 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2000 (0.3475) acc1: 99.4792 (85.3738) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1868 data: 0.0929 max mem: 3536 +Test: [1340/2387] eta: 0:03:25 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2305 (0.3474) acc1: 97.6562 (85.3955) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1891 data: 0.0937 max mem: 3536 +Test: [1350/2387] eta: 0:03:23 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.0855 (0.3454) acc1: 100.0000 (85.5036) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1957 data: 0.1030 max mem: 3537 +Test: [1360/2387] eta: 0:03:21 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.0767 (0.3434) acc1: 100.0000 (85.6101) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0884 max mem: 3537 +Test: [1370/2387] eta: 0:03:18 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.0739 (0.3414) acc1: 100.0000 (85.7151) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1622 data: 0.0692 max mem: 3537 +Test: [1380/2387] eta: 0:03:16 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.0775 (0.3395) acc1: 100.0000 (85.8185) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1752 data: 0.0830 max mem: 3537 +Test: [1390/2387] eta: 0:03:14 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.0801 (0.3377) acc1: 100.0000 (85.9205) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1837 data: 0.0930 max mem: 3537 +Test: [1400/2387] eta: 0:03:13 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.0831 (0.3359) acc1: 100.0000 (86.0210) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1996 data: 0.1072 max mem: 3537 +Test: [1410/2387] eta: 0:03:10 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.0948 (0.3347) acc1: 100.0000 (86.1110) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1830 data: 0.0883 max mem: 3537 +Test: [1420/2387] eta: 0:03:08 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1814 (0.3339) acc1: 98.9583 (86.1787) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1722 data: 0.0795 max mem: 3537 +Test: [1430/2387] eta: 0:03:06 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1814 (0.3328) acc1: 99.4792 (86.2683) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1899 data: 0.0988 max mem: 3537 +Test: [1440/2387] eta: 0:03:04 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1874 (0.3320) acc1: 99.4792 (86.3524) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1726 data: 0.0804 max mem: 3537 +Test: [1450/2387] eta: 0:03:02 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2186 (0.3314) acc1: 98.9583 (86.4250) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1700 data: 0.0765 max mem: 3537 +Test: [1460/2387] eta: 0:03:00 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1525 (0.3302) acc1: 99.2188 (86.5163) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1810 data: 0.0886 max mem: 3537 +Test: [1470/2387] eta: 0:02:58 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1268 (0.3287) acc1: 100.0000 (86.6076) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1826 data: 0.0900 max mem: 3537 +Test: [1480/2387] eta: 0:02:56 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1217 (0.3274) acc1: 100.0000 (86.6969) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1838 data: 0.0903 max mem: 3537 +Test: [1490/2387] eta: 0:02:54 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1335 (0.3261) acc1: 100.0000 (86.7860) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1779 data: 0.0861 max mem: 3537 +Test: [1500/2387] eta: 0:02:52 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1410 (0.3249) acc1: 100.0000 (86.8730) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0828 max mem: 3537 +Test: [1510/2387] eta: 0:02:50 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1531 (0.3240) acc1: 100.0000 (86.9597) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1770 data: 0.0800 max mem: 3537 +Test: [1520/2387] eta: 0:02:48 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1871 (0.3231) acc1: 100.0000 (87.0422) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1576 data: 0.0648 max mem: 3537 +Test: [1530/2387] eta: 0:02:46 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1870 (0.3222) acc1: 99.7396 (87.1251) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1550 data: 0.0618 max mem: 3537 +Test: [1540/2387] eta: 0:02:44 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1723 (0.3212) acc1: 100.0000 (87.2075) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1678 data: 0.0755 max mem: 3537 +Test: [1550/2387] eta: 0:02:42 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1730 (0.3203) acc1: 100.0000 (87.2891) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1869 data: 0.0942 max mem: 3537 +Test: [1560/2387] eta: 0:02:40 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1837 (0.3195) acc1: 99.7396 (87.3680) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2015 data: 0.1092 max mem: 3537 +Test: [1570/2387] eta: 0:02:38 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2010 (0.3188) acc1: 99.7396 (87.4446) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1999 data: 0.1075 max mem: 3537 +Test: [1580/2387] eta: 0:02:36 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2010 (0.3180) acc1: 99.4792 (87.5166) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1899 data: 0.0959 max mem: 3538 +Test: [1590/2387] eta: 0:02:34 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1664 (0.3170) acc1: 99.2188 (87.5905) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1949 data: 0.1015 max mem: 3538 +Test: [1600/2387] eta: 0:02:32 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1546 (0.3163) acc1: 99.4792 (87.6324) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2041 data: 0.1107 max mem: 3538 +Test: [1610/2387] eta: 0:02:30 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1452 (0.3152) acc1: 99.7396 (87.7072) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1926 data: 0.0999 max mem: 3538 +Test: [1620/2387] eta: 0:02:28 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1649 (0.3144) acc1: 99.7396 (87.7750) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1854 data: 0.0946 max mem: 3538 +Test: [1630/2387] eta: 0:02:26 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2018 (0.3140) acc1: 98.1771 (87.8235) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1839 data: 0.0909 max mem: 3538 +Test: [1640/2387] eta: 0:02:24 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2710 (0.3139) acc1: 94.0104 (87.8510) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1740 data: 0.0806 max mem: 3538 +Test: [1650/2387] eta: 0:02:22 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2813 (0.3136) acc1: 94.0104 (87.8924) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1734 data: 0.0823 max mem: 3538 +Test: [1660/2387] eta: 0:02:20 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2682 (0.3134) acc1: 94.2708 (87.9293) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1849 data: 0.0920 max mem: 3538 +Test: [1670/2387] eta: 0:02:18 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2655 (0.3132) acc1: 94.2708 (87.9652) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1999 data: 0.1047 max mem: 3538 +Test: [1680/2387] eta: 0:02:16 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2698 (0.3130) acc1: 95.3125 (88.0026) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2188 data: 0.1243 max mem: 3538 +Test: [1690/2387] eta: 0:02:14 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2867 (0.3129) acc1: 94.0104 (88.0248) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2095 data: 0.1161 max mem: 3538 +Test: [1700/2387] eta: 0:02:13 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2838 (0.3124) acc1: 94.5312 (88.0786) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1954 data: 0.1028 max mem: 3538 +Test: [1710/2387] eta: 0:02:11 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1324 (0.3113) acc1: 99.7396 (88.1475) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2051 data: 0.1131 max mem: 3538 +Test: [1720/2387] eta: 0:02:09 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1201 (0.3102) acc1: 100.0000 (88.2159) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2075 data: 0.1162 max mem: 3538 +Test: [1730/2387] eta: 0:02:07 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1163 (0.3091) acc1: 100.0000 (88.2831) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1951 data: 0.1029 max mem: 3538 +Test: [1740/2387] eta: 0:02:05 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1199 (0.3080) acc1: 100.0000 (88.3493) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1967 data: 0.1021 max mem: 3538 +Test: [1750/2387] eta: 0:02:03 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1211 (0.3070) acc1: 100.0000 (88.4157) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1945 data: 0.0990 max mem: 3538 +Test: [1760/2387] eta: 0:02:01 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1282 (0.3061) acc1: 100.0000 (88.4810) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1899 data: 0.0945 max mem: 3538 +Test: [1770/2387] eta: 0:01:59 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1425 (0.3052) acc1: 100.0000 (88.5439) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1948 data: 0.1020 max mem: 3538 +Test: [1780/2387] eta: 0:01:57 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1278 (0.3041) acc1: 100.0000 (88.6081) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1877 data: 0.0970 max mem: 3538 +Test: [1790/2387] eta: 0:01:55 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1207 (0.3031) acc1: 100.0000 (88.6708) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1880 data: 0.0968 max mem: 3538 +Test: [1800/2387] eta: 0:01:53 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1152 (0.3020) acc1: 100.0000 (88.7327) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1994 data: 0.1087 max mem: 3538 +Test: [1810/2387] eta: 0:01:51 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1228 (0.3011) acc1: 100.0000 (88.7943) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1885 data: 0.0983 max mem: 3539 +Test: [1820/2387] eta: 0:01:49 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1270 (0.3001) acc1: 100.0000 (88.8551) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1730 data: 0.0823 max mem: 3539 +Test: [1830/2387] eta: 0:01:47 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1407 (0.2993) acc1: 100.0000 (88.9156) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1933 data: 0.1004 max mem: 3539 +Test: [1840/2387] eta: 0:01:45 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1444 (0.2985) acc1: 100.0000 (88.9758) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2095 data: 0.1141 max mem: 3539 +Test: [1850/2387] eta: 0:01:43 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1432 (0.2976) acc1: 100.0000 (89.0353) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0828 max mem: 3539 +Test: [1860/2387] eta: 0:01:41 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1421 (0.2968) acc1: 100.0000 (89.0943) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1557 data: 0.0591 max mem: 3539 +Test: [1870/2387] eta: 0:01:39 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1421 (0.2960) acc1: 100.0000 (89.1524) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0870 max mem: 3539 +Test: [1880/2387] eta: 0:01:38 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1421 (0.2952) acc1: 100.0000 (89.2101) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2142 data: 0.1194 max mem: 3539 +Test: [1890/2387] eta: 0:01:36 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1419 (0.2944) acc1: 100.0000 (89.2671) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1999 data: 0.1036 max mem: 3539 +Test: [1900/2387] eta: 0:01:34 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1435 (0.2936) acc1: 100.0000 (89.3236) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1465 data: 0.0507 max mem: 3539 +Test: [1910/2387] eta: 0:01:32 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1567 (0.2929) acc1: 100.0000 (89.3788) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1500 data: 0.0534 max mem: 3539 +Test: [1920/2387] eta: 0:01:30 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1644 (0.2924) acc1: 100.0000 (89.4207) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1760 data: 0.0819 max mem: 3539 +Test: [1930/2387] eta: 0:01:28 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1607 (0.2917) acc1: 100.0000 (89.4753) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1915 data: 0.0995 max mem: 3539 +Test: [1940/2387] eta: 0:01:26 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1533 (0.2913) acc1: 100.0000 (89.4891) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1823 data: 0.0900 max mem: 3539 +Test: [1950/2387] eta: 0:01:24 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1579 (0.2906) acc1: 100.0000 (89.5406) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1633 data: 0.0705 max mem: 3539 +Test: [1960/2387] eta: 0:01:22 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1783 (0.2902) acc1: 100.0000 (89.5865) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1754 data: 0.0841 max mem: 3539 +Test: [1970/2387] eta: 0:01:20 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2293 (0.2902) acc1: 98.6979 (89.6131) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1873 data: 0.0966 max mem: 3539 +Test: [1980/2387] eta: 0:01:18 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2409 (0.2900) acc1: 97.9167 (89.6479) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1850 data: 0.0933 max mem: 3539 +Test: [1990/2387] eta: 0:01:16 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1758 (0.2894) acc1: 98.9583 (89.6950) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1742 data: 0.0825 max mem: 3539 +Test: [2000/2387] eta: 0:01:14 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1742 (0.2888) acc1: 98.9583 (89.7420) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1761 data: 0.0839 max mem: 3539 +Test: [2010/2387] eta: 0:01:12 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1740 (0.2883) acc1: 98.9583 (89.7872) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1956 data: 0.1018 max mem: 3539 +Test: [2020/2387] eta: 0:01:10 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1762 (0.2877) acc1: 98.9583 (89.8320) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2135 data: 0.1209 max mem: 3539 +Test: [2030/2387] eta: 0:01:08 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1764 (0.2872) acc1: 98.9583 (89.8766) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1878 data: 0.0958 max mem: 3540 +Test: [2040/2387] eta: 0:01:06 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1746 (0.2866) acc1: 98.9583 (89.9217) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1894 data: 0.0963 max mem: 3540 +Test: [2050/2387] eta: 0:01:04 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1736 (0.2860) acc1: 99.2188 (89.9672) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2152 data: 0.1223 max mem: 3540 +Test: [2060/2387] eta: 0:01:02 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1736 (0.2855) acc1: 99.2188 (90.0121) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1999 data: 0.1082 max mem: 3540 +Test: [2070/2387] eta: 0:01:01 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1698 (0.2849) acc1: 99.2188 (90.0574) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2211 data: 0.1299 max mem: 3540 +Test: [2080/2387] eta: 0:00:59 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1686 (0.2844) acc1: 99.2188 (90.1012) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2299 data: 0.1381 max mem: 3540 +Test: [2090/2387] eta: 0:00:57 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1686 (0.2838) acc1: 99.4792 (90.1459) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2030 data: 0.1071 max mem: 3540 +Test: [2100/2387] eta: 0:00:55 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1671 (0.2833) acc1: 99.4792 (90.1902) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1549 data: 0.0564 max mem: 3540 +Test: [2110/2387] eta: 0:00:53 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1693 (0.2827) acc1: 99.4792 (90.2336) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1876 data: 0.0936 max mem: 3540 +Test: [2120/2387] eta: 0:00:51 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1739 (0.2824) acc1: 99.2188 (90.2653) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2400 data: 0.1492 max mem: 3540 +Test: [2130/2387] eta: 0:00:49 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1939 (0.2820) acc1: 98.9583 (90.3043) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1572 data: 0.0620 max mem: 3540 +Test: [2140/2387] eta: 0:00:47 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1806 (0.2818) acc1: 97.9167 (90.3157) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1403 data: 0.0446 max mem: 3540 +Test: [2150/2387] eta: 0:00:45 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1736 (0.2813) acc1: 98.4375 (90.3568) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1908 data: 0.0990 max mem: 3540 +Test: [2160/2387] eta: 0:00:43 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1906 (0.2809) acc1: 98.9583 (90.3933) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2044 data: 0.1110 max mem: 3540 +Test: [2170/2387] eta: 0:00:41 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2268 (0.2809) acc1: 97.1354 (90.4142) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2046 data: 0.1116 max mem: 3540 +Test: [2180/2387] eta: 0:00:39 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.3119 (0.2826) acc1: 92.9688 (90.3126) acc5: 100.0000 (100.0000) acc1_neg: 28.385 acc5_neg: 100.000 time: 0.1849 data: 0.0920 max mem: 3540 +Test: [2190/2387] eta: 0:00:37 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2957 (0.2827) acc1: 92.9688 (90.3044) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1798 data: 0.0849 max mem: 3540 +Test: [2200/2387] eta: 0:00:36 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1537 (0.2821) acc1: 98.9583 (90.3472) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1885 data: 0.0938 max mem: 3540 +Test: [2210/2387] eta: 0:00:34 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1339 (0.2814) acc1: 100.0000 (90.3909) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1722 data: 0.0791 max mem: 3540 +Test: [2220/2387] eta: 0:00:32 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1358 (0.2808) acc1: 100.0000 (90.4327) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1630 data: 0.0720 max mem: 3540 +Test: [2230/2387] eta: 0:00:30 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1376 (0.2801) acc1: 100.0000 (90.4755) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1729 data: 0.0822 max mem: 3540 +Test: [2240/2387] eta: 0:00:28 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1417 (0.2795) acc1: 100.0000 (90.5166) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1802 data: 0.0896 max mem: 3540 +Test: [2250/2387] eta: 0:00:26 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1539 (0.2792) acc1: 100.0000 (90.5554) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1719 data: 0.0814 max mem: 3540 +Test: [2260/2387] eta: 0:00:24 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1882 (0.2788) acc1: 100.0000 (90.5956) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1738 data: 0.0829 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1734 (0.2782) acc1: 100.0000 (90.6366) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1821 data: 0.0885 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1610 (0.2779) acc1: 100.0000 (90.6530) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1795 data: 0.0865 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1476 (0.2774) acc1: 100.0000 (90.6932) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1718 data: 0.0812 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1639 (0.2770) acc1: 100.0000 (90.7308) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1526 data: 0.0622 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1922 (0.2767) acc1: 100.0000 (90.7657) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1412 data: 0.0512 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2205 (0.2766) acc1: 99.2188 (90.7908) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1320 data: 0.0421 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2273 (0.2763) acc1: 98.4375 (90.8197) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1248 data: 0.0349 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1833 (0.2759) acc1: 99.4792 (90.8532) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1230 data: 0.0327 max mem: 3541 +Test: [2350/2387] eta: 0:00:07 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1643 (0.2759) acc1: 99.7396 (90.8587) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1213 data: 0.0305 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.1808 (0.2755) acc1: 99.4792 (90.8888) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1202 data: 0.0294 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2178 (0.2754) acc1: 98.9583 (90.9156) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.1171 data: 0.0270 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2641 (0.2755) acc1: 97.6562 (90.9283) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1149 data: 0.0251 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 39.726 acc5_pos: 100.000 loss: 0.2740 (0.2754) acc1: 95.5729 (90.9357) acc5: 100.0000 (100.0000) acc1_neg: 93.023 acc5_neg: 100.000 time: 0.1106 data: 0.0235 max mem: 3541 +Test: Total time: 0:07:31 (0.1892 s / it) +* Overall Acc@1 90.936 Acc@5 100.000 + target == real Acc@1 20.746 Acc@5 100.000 + target == fake Acc@1 95.865 Acc@5 100.000 + loss 0.275 + PR-AUC 0.9913 ROC-AUC 0.8878 EER 17.55% (thr=0.7471) +Accuracy of the network on the 916267 test images: 90.9% +[rank0]:[W726 06:24:22.696433554 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 06:24:43.677000 1200896 torch/distributed/run.py:766] +W0726 06:24:43.677000 1200896 torch/distributed/run.py:766] ***************************************** +W0726 06:24:43.677000 1200896 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:24:43.677000 1200896 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:25:12.226183946 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 06:25:12.549531642 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:25:12.599376130 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:25:12.633075257 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:25:12.671380936 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +| distributed init (rank 3): env:// +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:25:13.932677849 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W726 06:25:13.932693770 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W726 06:25:13.932703587 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n22_seed9_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed9_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:03:34 acc1_pos: 23.698 acc5_pos: 100.000 loss: 0.9981 (0.9981) acc1: 23.6979 (23.6979) acc5: 100.0000 (100.0000) time: 7.3791 data: 5.4352 max mem: 3531 +Test: [ 10/517] eta: 0:06:23 acc1_pos: 1.302 acc5_pos: 100.000 loss: 1.1853 (1.1507) acc1: 9.6354 (11.7188) acc5: 100.0000 (100.0000) time: 0.7558 data: 0.4945 max mem: 3531 +Test: [ 20/517] eta: 0:03:40 acc1_pos: 5.729 acc5_pos: 100.000 loss: 1.1986 (1.1965) acc1: 9.6354 (11.1359) acc5: 100.0000 (100.0000) time: 0.0967 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:02:57 acc1_pos: 2.083 acc5_pos: 100.000 loss: 1.2146 (1.2005) acc1: 9.6354 (10.4587) acc5: 100.0000 (100.0000) time: 0.1499 data: 0.0552 max mem: 3531 +Test: [ 40/517] eta: 0:02:43 acc1_pos: 9.115 acc5_pos: 100.000 loss: 1.1558 (1.1843) acc1: 7.5521 (9.8260) acc5: 100.0000 (100.0000) time: 0.2349 data: 0.1427 max mem: 3531 +Test: [ 50/517] eta: 0:02:30 acc1_pos: 18.349 acc5_pos: 100.000 loss: 1.0945 (1.1366) acc1: 8.0729 (14.2157) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2549 data: 0.1620 max mem: 3531 +Test: [ 60/517] eta: 0:02:22 acc1_pos: 18.349 acc5_pos: 100.000 loss: 0.4358 (1.0147) acc1: 89.0625 (27.4291) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2499 data: 0.1568 max mem: 3531 +Test: [ 70/517] eta: 0:02:16 acc1_pos: 18.349 acc5_pos: 100.000 loss: 0.3778 (0.9247) acc1: 95.8333 (37.0599) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2599 data: 0.1637 max mem: 3531 +Test: [ 80/517] eta: 0:02:09 acc1_pos: 18.349 acc5_pos: 100.000 loss: 0.3817 (0.8587) acc1: 95.0521 (44.2130) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.2499 data: 0.1524 max mem: 3531 +Test: [ 90/517] eta: 0:02:03 acc1_pos: 18.349 acc5_pos: 100.000 loss: 0.3601 (0.8020) acc1: 94.2708 (49.8054) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2399 data: 0.1427 max mem: 3531 +Test: [100/517] eta: 0:01:58 acc1_pos: 18.349 acc5_pos: 100.000 loss: 0.3113 (0.7522) acc1: 96.8750 (54.5328) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2299 data: 0.1349 max mem: 3531 +Test: [110/517] eta: 0:01:53 acc1_pos: 18.349 acc5_pos: 100.000 loss: 0.2908 (0.7109) acc1: 98.1771 (58.4858) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2249 data: 0.1318 max mem: 3531 +Test: [120/517] eta: 0:01:48 acc1_pos: 18.349 acc5_pos: 100.000 loss: 0.3330 (0.6827) acc1: 97.6562 (61.6047) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2299 data: 0.1358 max mem: 3531 +Test: [130/517] eta: 0:01:45 acc1_pos: 18.349 acc5_pos: 100.000 loss: 0.3661 (0.6578) acc1: 97.3958 (64.3527) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2399 data: 0.1454 max mem: 3531 +Test: [140/517] eta: 0:01:41 acc1_pos: 18.349 acc5_pos: 100.000 loss: 0.3170 (0.6321) acc1: 99.7396 (66.8791) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2349 data: 0.1416 max mem: 3531 +Test: [150/517] eta: 0:01:37 acc1_pos: 18.349 acc5_pos: 100.000 loss: 0.2872 (0.6091) acc1: 100.0000 (69.0484) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2199 data: 0.1260 max mem: 3531 +Test: [160/517] eta: 0:01:34 acc1_pos: 18.349 acc5_pos: 100.000 loss: 0.2840 (0.5889) acc1: 100.0000 (70.9546) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2449 data: 0.1507 max mem: 3531 +Test: [170/517] eta: 0:01:32 acc1_pos: 18.349 acc5_pos: 100.000 loss: 0.2749 (0.5697) acc1: 100.0000 (72.5999) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2649 data: 0.1679 max mem: 3531 +Test: [180/517] eta: 0:01:28 acc1_pos: 18.349 acc5_pos: 100.000 loss: 0.2667 (0.5542) acc1: 99.7396 (74.0504) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2449 data: 0.1477 max mem: 3531 +Test: [190/517] eta: 0:01:25 acc1_pos: 18.349 acc5_pos: 100.000 loss: 0.3204 (0.5422) acc1: 98.6979 (75.3013) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2299 data: 0.1371 max mem: 3531 +Test: [200/517] eta: 0:01:22 acc1_pos: 18.349 acc5_pos: 100.000 loss: 0.3000 (0.5288) acc1: 99.2188 (76.5120) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2349 data: 0.1420 max mem: 3531 +Test: [210/517] eta: 0:01:19 acc1_pos: 18.349 acc5_pos: 100.000 loss: 0.3232 (0.5219) acc1: 93.2292 (77.0192) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.2399 data: 0.1467 max mem: 3532 +Test: [220/517] eta: 0:01:16 acc1_pos: 18.349 acc5_pos: 100.000 loss: 0.3740 (0.5150) acc1: 88.8021 (77.5947) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.2399 data: 0.1454 max mem: 3532 +Test: [230/517] eta: 0:01:14 acc1_pos: 18.349 acc5_pos: 100.000 loss: 0.3620 (0.5082) acc1: 91.1458 (78.3922) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2399 data: 0.1426 max mem: 3532 +Test: [240/517] eta: 0:01:11 acc1_pos: 18.349 acc5_pos: 100.000 loss: 0.3288 (0.4998) acc1: 97.3958 (79.2283) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2399 data: 0.1415 max mem: 3532 +Test: [250/517] eta: 0:01:08 acc1_pos: 18.349 acc5_pos: 100.000 loss: 0.3151 (0.4928) acc1: 99.4792 (80.0195) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2449 data: 0.1482 max mem: 3532 +Test: [260/517] eta: 0:01:05 acc1_pos: 18.349 acc5_pos: 100.000 loss: 0.3141 (0.4864) acc1: 99.7396 (80.7591) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2449 data: 0.1514 max mem: 3532 +Test: [270/517] eta: 0:01:03 acc1_pos: 18.349 acc5_pos: 100.000 loss: 0.3614 (0.4822) acc1: 99.2188 (81.4047) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2499 data: 0.1547 max mem: 3532 +Test: [280/517] eta: 0:01:00 acc1_pos: 18.349 acc5_pos: 100.000 loss: 0.3636 (0.4770) acc1: 98.4375 (81.9970) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2449 data: 0.1475 max mem: 3532 +Test: [290/517] eta: 0:00:58 acc1_pos: 18.349 acc5_pos: 100.000 loss: 0.3155 (0.4712) acc1: 98.9583 (82.5736) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2449 data: 0.1475 max mem: 3532 +Test: [300/517] eta: 0:00:55 acc1_pos: 18.349 acc5_pos: 100.000 loss: 0.3143 (0.4660) acc1: 98.9583 (83.1110) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2399 data: 0.1419 max mem: 3532 +Test: [310/517] eta: 0:00:52 acc1_pos: 18.349 acc5_pos: 100.000 loss: 0.2989 (0.4606) acc1: 99.4792 (83.6432) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2299 data: 0.1338 max mem: 3532 +Test: [320/517] eta: 0:00:49 acc1_pos: 18.349 acc5_pos: 100.000 loss: 0.3083 (0.4561) acc1: 99.4792 (84.1284) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2399 data: 0.1445 max mem: 3532 +Test: [330/517] eta: 0:00:47 acc1_pos: 18.349 acc5_pos: 100.000 loss: 0.3103 (0.4515) acc1: 99.2188 (84.5796) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2449 data: 0.1477 max mem: 3532 +Test: [340/517] eta: 0:00:44 acc1_pos: 18.349 acc5_pos: 100.000 loss: 0.3012 (0.4482) acc1: 99.4792 (84.8645) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.2449 data: 0.1483 max mem: 3532 +Test: [350/517] eta: 0:00:42 acc1_pos: 18.349 acc5_pos: 100.000 loss: 0.2808 (0.4450) acc1: 98.1771 (85.1496) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2449 data: 0.1508 max mem: 3532 +Test: [360/517] eta: 0:00:39 acc1_pos: 18.349 acc5_pos: 100.000 loss: 0.2912 (0.4415) acc1: 97.6562 (85.4823) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2449 data: 0.1516 max mem: 3532 +Test: [370/517] eta: 0:00:37 acc1_pos: 18.349 acc5_pos: 100.000 loss: 0.2946 (0.4377) acc1: 99.2188 (85.8652) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2449 data: 0.1491 max mem: 3532 +Test: [380/517] eta: 0:00:34 acc1_pos: 18.349 acc5_pos: 100.000 loss: 0.3197 (0.4354) acc1: 98.9583 (86.1350) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2391 data: 0.1435 max mem: 3532 +Test: [390/517] eta: 0:00:31 acc1_pos: 18.349 acc5_pos: 100.000 loss: 0.3412 (0.4331) acc1: 96.3542 (86.4317) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2391 data: 0.1448 max mem: 3532 +Test: [400/517] eta: 0:00:29 acc1_pos: 18.349 acc5_pos: 100.000 loss: 0.3332 (0.4300) acc1: 98.6979 (86.7558) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2399 data: 0.1449 max mem: 3532 +Test: [410/517] eta: 0:00:26 acc1_pos: 18.349 acc5_pos: 100.000 loss: 0.3075 (0.4280) acc1: 99.4792 (87.0552) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2399 data: 0.1428 max mem: 3532 +Test: [420/517] eta: 0:00:24 acc1_pos: 18.349 acc5_pos: 100.000 loss: 0.3781 (0.4269) acc1: 99.2188 (87.3262) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2349 data: 0.1366 max mem: 3532 +Test: [430/517] eta: 0:00:21 acc1_pos: 18.349 acc5_pos: 100.000 loss: 0.3668 (0.4252) acc1: 98.9583 (87.5278) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.2299 data: 0.1332 max mem: 3532 +Test: [440/517] eta: 0:00:19 acc1_pos: 18.349 acc5_pos: 100.000 loss: 0.3845 (0.4252) acc1: 96.3542 (87.6364) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2399 data: 0.1450 max mem: 3533 +Test: [450/517] eta: 0:00:16 acc1_pos: 18.349 acc5_pos: 100.000 loss: 0.3907 (0.4241) acc1: 97.6562 (87.8851) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2449 data: 0.1474 max mem: 3533 +Test: [460/517] eta: 0:00:14 acc1_pos: 18.349 acc5_pos: 100.000 loss: 0.3717 (0.4228) acc1: 98.6979 (88.0666) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.2449 data: 0.1457 max mem: 3533 +Test: [470/517] eta: 0:00:11 acc1_pos: 18.349 acc5_pos: 100.000 loss: 0.3480 (0.4212) acc1: 98.6979 (88.2763) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2399 data: 0.1425 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 18.349 acc5_pos: 100.000 loss: 0.3398 (0.4196) acc1: 99.4792 (88.5092) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2349 data: 0.1394 max mem: 3533 +Test: [490/517] eta: 0:00:06 acc1_pos: 18.349 acc5_pos: 100.000 loss: 0.3328 (0.4177) acc1: 99.4792 (88.7087) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2399 data: 0.1464 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 18.349 acc5_pos: 100.000 loss: 0.3749 (0.4178) acc1: 97.9167 (88.8203) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.2499 data: 0.1542 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 18.349 acc5_pos: 100.000 loss: 0.3996 (0.4178) acc1: 94.7917 (88.9264) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2239 data: 0.1275 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 18.349 acc5_pos: 100.000 loss: 0.3772 (0.4171) acc1: 98.1771 (89.0423) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1813 data: 0.0812 max mem: 3533 +Test: Total time: 0:02:08 (0.2482 s / it) +* Overall Acc@1 89.042 Acc@5 100.000 + target == real Acc@1 10.177 Acc@5 100.000 + target == fake Acc@1 97.224 Acc@5 100.000 + loss 0.417 + PR-AUC 0.9320 ROC-AUC 0.6222 EER 41.51% (thr=0.7109) +Accuracy of the network on the 198427 test images: 89.0% +[rank0]:[W726 06:27:35.283471204 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0726 06:27:55.885000 1201468 torch/distributed/run.py:766] +W0726 06:27:55.885000 1201468 torch/distributed/run.py:766] ***************************************** +W0726 06:27:55.885000 1201468 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:27:55.885000 1201468 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:28:25.934926791 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:28:25.056918708 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:28:25.548619696 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:28:25.703253646 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:28:25.788498982 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:28:25.839892659 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:28:25.872114697 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 06:28:25.893345430 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n22_seed9_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed9_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:46:48 acc1_pos: 48.698 acc5_pos: 100.000 loss: 0.6706 (0.6706) acc1: 48.6979 (48.6979) acc5: 100.0000 (100.0000) time: 7.2188 data: 5.4989 max mem: 3531 +Test: [ 10/389] eta: 0:04:42 acc1_pos: 52.344 acc5_pos: 100.000 loss: 0.7302 (0.8456) acc1: 52.0833 (42.7320) acc5: 100.0000 (100.0000) time: 0.7463 data: 0.5026 max mem: 3531 +Test: [ 20/389] eta: 0:02:41 acc1_pos: 16.927 acc5_pos: 100.000 loss: 0.8840 (0.8623) acc1: 29.4271 (34.8586) acc5: 100.0000 (100.0000) time: 0.0987 data: 0.0017 max mem: 3531 +Test: [ 30/389] eta: 0:02:34 acc1_pos: 27.083 acc5_pos: 100.000 loss: 0.8852 (0.8595) acc1: 20.0521 (32.2413) acc5: 100.0000 (100.0000) time: 0.2550 data: 0.1603 max mem: 3531 +Test: [ 40/389] eta: 0:02:17 acc1_pos: 36.979 acc5_pos: 100.000 loss: 0.7961 (0.8169) acc1: 35.6771 (37.0871) acc5: 100.0000 (100.0000) time: 0.3499 data: 0.2577 max mem: 3531 +Test: [ 50/389] eta: 0:02:03 acc1_pos: 11.719 acc5_pos: 100.000 loss: 0.7908 (0.8390) acc1: 35.6771 (35.3094) acc5: 100.0000 (100.0000) time: 0.2640 data: 0.1701 max mem: 3531 +Test: [ 60/389] eta: 0:01:47 acc1_pos: 45.052 acc5_pos: 100.000 loss: 0.8913 (0.8367) acc1: 15.6250 (36.9493) acc5: 100.0000 (100.0000) time: 0.1904 data: 0.0950 max mem: 3531 +Test: [ 70/389] eta: 0:01:40 acc1_pos: 75.521 acc5_pos: 100.000 loss: 0.7055 (0.8017) acc1: 49.2188 (40.9258) acc5: 100.0000 (100.0000) time: 0.1849 data: 0.0894 max mem: 3531 +Test: [ 80/389] eta: 0:01:32 acc1_pos: 91.406 acc5_pos: 100.000 loss: 0.7055 (0.7879) acc1: 49.2188 (42.2389) acc5: 100.0000 (100.0000) time: 0.2094 data: 0.1158 max mem: 3531 +Test: [ 90/389] eta: 0:01:26 acc1_pos: 45.833 acc5_pos: 100.000 loss: 0.4642 (0.7522) acc1: 65.1042 (46.1481) acc5: 100.0000 (100.0000) time: 0.2004 data: 0.1082 max mem: 3531 +Test: [100/389] eta: 0:01:20 acc1_pos: 52.604 acc5_pos: 100.000 loss: 0.6748 (0.7558) acc1: 52.8646 (45.9030) acc5: 100.0000 (100.0000) time: 0.1905 data: 0.0978 max mem: 3531 +Test: [110/389] eta: 0:01:15 acc1_pos: 99.740 acc5_pos: 100.000 loss: 0.7272 (0.7451) acc1: 48.9583 (46.4809) acc5: 100.0000 (100.0000) time: 0.1824 data: 0.0893 max mem: 3531 +Test: [120/389] eta: 0:01:10 acc1_pos: 61.198 acc5_pos: 100.000 loss: 0.6401 (0.7358) acc1: 54.6875 (47.6111) acc5: 100.0000 (100.0000) time: 0.1823 data: 0.0896 max mem: 3531 +Test: [130/389] eta: 0:01:06 acc1_pos: 28.834 acc5_pos: 100.000 loss: 0.6336 (0.7288) acc1: 61.1979 (48.6502) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.1749 data: 0.0824 max mem: 3531 +Test: [140/389] eta: 0:01:02 acc1_pos: 28.834 acc5_pos: 100.000 loss: 0.5646 (0.7216) acc1: 72.3958 (50.0166) acc5: 100.0000 (100.0000) acc1_neg: 24.219 acc5_neg: 100.000 time: 0.1840 data: 0.0908 max mem: 3531 +Test: [150/389] eta: 0:00:58 acc1_pos: 28.834 acc5_pos: 100.000 loss: 0.5496 (0.7139) acc1: 76.0417 (51.0917) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1841 data: 0.0911 max mem: 3531 +Test: [160/389] eta: 0:00:55 acc1_pos: 28.834 acc5_pos: 100.000 loss: 0.5496 (0.6966) acc1: 76.0417 (53.1978) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1728 data: 0.0813 max mem: 3531 +Test: [170/389] eta: 0:00:51 acc1_pos: 28.834 acc5_pos: 100.000 loss: 0.3681 (0.6815) acc1: 98.1771 (55.4094) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.1678 data: 0.0775 max mem: 3531 +Test: [180/389] eta: 0:00:48 acc1_pos: 28.834 acc5_pos: 100.000 loss: 0.4094 (0.6697) acc1: 93.4896 (57.2197) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.1659 data: 0.0757 max mem: 3531 +Test: [190/389] eta: 0:00:45 acc1_pos: 28.834 acc5_pos: 100.000 loss: 0.4927 (0.6658) acc1: 84.1146 (57.6339) acc5: 100.0000 (100.0000) acc1_neg: 21.094 acc5_neg: 100.000 time: 0.1698 data: 0.0788 max mem: 3531 +Test: [200/389] eta: 0:00:42 acc1_pos: 28.834 acc5_pos: 100.000 loss: 0.6112 (0.6654) acc1: 66.6667 (57.8915) acc5: 100.0000 (100.0000) acc1_neg: 66.667 acc5_neg: 100.000 time: 0.1753 data: 0.0843 max mem: 3531 +Test: [210/389] eta: 0:00:40 acc1_pos: 28.834 acc5_pos: 100.000 loss: 0.5371 (0.6521) acc1: 73.9583 (59.3219) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1771 data: 0.0861 max mem: 3532 +Test: [220/389] eta: 0:00:37 acc1_pos: 28.834 acc5_pos: 100.000 loss: 0.2907 (0.6362) acc1: 99.2188 (60.8750) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.1844 data: 0.0929 max mem: 3532 +Test: [230/389] eta: 0:00:35 acc1_pos: 28.834 acc5_pos: 100.000 loss: 0.5440 (0.6345) acc1: 82.2917 (61.1630) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1827 data: 0.0909 max mem: 3532 +Test: [240/389] eta: 0:00:32 acc1_pos: 28.834 acc5_pos: 100.000 loss: 0.5500 (0.6295) acc1: 71.3542 (61.8690) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1805 data: 0.0887 max mem: 3532 +Test: [250/389] eta: 0:00:30 acc1_pos: 28.834 acc5_pos: 100.000 loss: 0.4930 (0.6235) acc1: 89.3229 (63.0364) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1876 data: 0.0970 max mem: 3532 +Test: [260/389] eta: 0:00:28 acc1_pos: 28.834 acc5_pos: 100.000 loss: 0.5026 (0.6226) acc1: 90.1042 (63.5846) acc5: 100.0000 (100.0000) acc1_neg: 41.146 acc5_neg: 100.000 time: 0.1896 data: 0.0989 max mem: 3532 +Test: [270/389] eta: 0:00:25 acc1_pos: 28.834 acc5_pos: 100.000 loss: 0.5587 (0.6241) acc1: 70.3125 (63.3927) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.1796 data: 0.0873 max mem: 3532 +Test: [280/389] eta: 0:00:23 acc1_pos: 28.834 acc5_pos: 100.000 loss: 0.6133 (0.6252) acc1: 60.1562 (63.2952) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1771 data: 0.0836 max mem: 3532 +Test: [290/389] eta: 0:00:21 acc1_pos: 28.834 acc5_pos: 100.000 loss: 0.5861 (0.6217) acc1: 69.5312 (63.8746) acc5: 100.0000 (100.0000) acc1_neg: 78.646 acc5_neg: 100.000 time: 0.1800 data: 0.0865 max mem: 3532 +Test: [300/389] eta: 0:00:18 acc1_pos: 28.834 acc5_pos: 100.000 loss: 0.3765 (0.6165) acc1: 85.4167 (64.5513) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1815 data: 0.0896 max mem: 3532 +Test: [310/389] eta: 0:00:16 acc1_pos: 28.834 acc5_pos: 100.000 loss: 0.5766 (0.6192) acc1: 65.8854 (64.2308) acc5: 100.0000 (100.0000) acc1_neg: 29.688 acc5_neg: 100.000 time: 0.1756 data: 0.0847 max mem: 3532 +Test: [320/389] eta: 0:00:14 acc1_pos: 28.834 acc5_pos: 100.000 loss: 0.6211 (0.6176) acc1: 62.7604 (64.5631) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1706 data: 0.0781 max mem: 3532 +Test: [330/389] eta: 0:00:12 acc1_pos: 28.834 acc5_pos: 100.000 loss: 0.6905 (0.6199) acc1: 54.6875 (64.2859) acc5: 100.0000 (100.0000) acc1_neg: 73.698 acc5_neg: 100.000 time: 0.1829 data: 0.0903 max mem: 3532 +Test: [340/389] eta: 0:00:10 acc1_pos: 28.834 acc5_pos: 100.000 loss: 0.4554 (0.6124) acc1: 73.6979 (65.2141) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1645 data: 0.0737 max mem: 3532 +Test: [350/389] eta: 0:00:08 acc1_pos: 28.834 acc5_pos: 100.000 loss: 0.3655 (0.6191) acc1: 96.0938 (64.7488) acc5: 100.0000 (100.0000) acc1_neg: 18.490 acc5_neg: 100.000 time: 0.1520 data: 0.0618 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 28.834 acc5_pos: 100.000 loss: 0.5305 (0.6198) acc1: 76.0417 (64.7319) acc5: 100.0000 (100.0000) acc1_neg: 79.427 acc5_neg: 100.000 time: 0.1416 data: 0.0516 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 28.834 acc5_pos: 100.000 loss: 0.5160 (0.6165) acc1: 78.1250 (65.0319) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1132 data: 0.0229 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 28.834 acc5_pos: 100.000 loss: 0.4519 (0.6142) acc1: 78.6458 (65.2873) acc5: 100.0000 (100.0000) acc1_neg: 49.740 acc5_neg: 100.000 time: 0.1145 data: 0.0244 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 28.834 acc5_pos: 100.000 loss: 0.4664 (0.6118) acc1: 86.1979 (65.6483) acc5: 100.0000 (100.0000) acc1_neg: 86.612 acc5_neg: 100.000 time: 0.1195 data: 0.0264 max mem: 3532 +Test: Total time: 0:01:16 (0.1975 s / it) +* Overall Acc@1 65.648 Acc@5 100.000 + target == real Acc@1 47.435 Acc@5 100.000 + target == fake Acc@1 74.114 Acc@5 100.000 + loss 0.612 + PR-AUC 0.8076 ROC-AUC 0.6781 EER 37.96% (thr=0.5630) +Accuracy of the network on the 149358 test images: 65.6% +[rank0]:[W726 06:29:52.304997015 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 06:30:11.944000 1202026 torch/distributed/run.py:766] +W0726 06:30:11.944000 1202026 torch/distributed/run.py:766] ***************************************** +W0726 06:30:11.944000 1202026 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:30:11.944000 1202026 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W726 06:30:41.057110607 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:30:41.077829891 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:30:41.400471734 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +[rank5]:[W726 06:30:41.767845969 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:30:41.776991143 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W726 06:30:41.777018659 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:30:41.782944738 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:30:41.832165412 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n22_seed9_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed9_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:31:02 acc1_pos: 70.052 acc5_pos: 100.000 loss: 0.5488 (0.5488) acc1: 70.0521 (70.0521) acc5: 100.0000 (100.0000) time: 5.5941 data: 3.6279 max mem: 3531 +Test: [ 10/333] eta: 0:03:10 acc1_pos: 83.594 acc5_pos: 100.000 loss: 0.3959 (0.3844) acc1: 78.1250 (80.3267) acc5: 100.0000 (100.0000) time: 0.5904 data: 0.3301 max mem: 3531 +Test: [ 20/333] eta: 0:01:54 acc1_pos: 94.010 acc5_pos: 100.000 loss: 0.3390 (0.3487) acc1: 83.5938 (83.1597) acc5: 100.0000 (100.0000) time: 0.1055 data: 0.0140 max mem: 3531 +Test: [ 30/333] eta: 0:01:31 acc1_pos: 53.125 acc5_pos: 100.000 loss: 0.3828 (0.4827) acc1: 75.2604 (72.5218) acc5: 100.0000 (100.0000) time: 0.1454 data: 0.0475 max mem: 3531 +Test: [ 40/333] eta: 0:01:22 acc1_pos: 50.521 acc5_pos: 100.000 loss: 0.7897 (0.5617) acc1: 47.6562 (66.0887) acc5: 100.0000 (100.0000) time: 0.1899 data: 0.0929 max mem: 3531 +Test: [ 50/333] eta: 0:01:16 acc1_pos: 45.833 acc5_pos: 100.000 loss: 0.7897 (0.6085) acc1: 45.8333 (61.8311) acc5: 100.0000 (100.0000) time: 0.2199 data: 0.1266 max mem: 3531 +Test: [ 60/333] eta: 0:01:09 acc1_pos: 33.594 acc5_pos: 100.000 loss: 0.7786 (0.6384) acc1: 45.0521 (59.2896) acc5: 100.0000 (100.0000) time: 0.2049 data: 0.1100 max mem: 3531 +Test: [ 70/333] eta: 0:01:04 acc1_pos: 51.823 acc5_pos: 100.000 loss: 0.7552 (0.6583) acc1: 47.3958 (57.6695) acc5: 100.0000 (100.0000) time: 0.1849 data: 0.0899 max mem: 3531 +Test: [ 80/333] eta: 0:00:59 acc1_pos: 52.083 acc5_pos: 100.000 loss: 0.7552 (0.6730) acc1: 48.1771 (56.2564) acc5: 100.0000 (100.0000) time: 0.1751 data: 0.0767 max mem: 3531 +Test: [ 90/333] eta: 0:00:57 acc1_pos: 56.510 acc5_pos: 100.000 loss: 0.7119 (0.6741) acc1: 52.0833 (56.4188) acc5: 100.0000 (100.0000) time: 0.1949 data: 0.0957 max mem: 3531 +Test: [100/333] eta: 0:00:53 acc1_pos: 65.625 acc5_pos: 100.000 loss: 0.6704 (0.6765) acc1: 58.0729 (56.3712) acc5: 100.0000 (100.0000) time: 0.1997 data: 0.1032 max mem: 3531 +Test: [110/333] eta: 0:00:50 acc1_pos: 65.365 acc5_pos: 100.000 loss: 0.6929 (0.6785) acc1: 54.9479 (56.2031) acc5: 100.0000 (100.0000) time: 0.1865 data: 0.0930 max mem: 3531 +Test: [120/333] eta: 0:00:47 acc1_pos: 61.458 acc5_pos: 100.000 loss: 0.6619 (0.6778) acc1: 53.9062 (56.2220) acc5: 100.0000 (100.0000) time: 0.2006 data: 0.1087 max mem: 3531 +Test: [130/333] eta: 0:00:45 acc1_pos: 68.229 acc5_pos: 100.000 loss: 0.6459 (0.6775) acc1: 58.0729 (56.4726) acc5: 100.0000 (100.0000) time: 0.2083 data: 0.1134 max mem: 3531 +Test: [140/333] eta: 0:00:42 acc1_pos: 75.000 acc5_pos: 100.000 loss: 0.6448 (0.6761) acc1: 58.0729 (56.5603) acc5: 100.0000 (100.0000) time: 0.1953 data: 0.1007 max mem: 3531 +Test: [150/333] eta: 0:00:40 acc1_pos: 61.198 acc5_pos: 100.000 loss: 0.6448 (0.6778) acc1: 55.7292 (56.4259) acc5: 100.0000 (100.0000) time: 0.1953 data: 0.1031 max mem: 3531 +Test: [160/333] eta: 0:00:37 acc1_pos: 50.987 acc5_pos: 100.000 loss: 0.7416 (0.6852) acc1: 50.7812 (56.1190) acc5: 100.0000 (100.0000) acc1_neg: 44.792 acc5_neg: 100.000 time: 0.2070 data: 0.1149 max mem: 3531 +Test: [170/333] eta: 0:00:35 acc1_pos: 50.987 acc5_pos: 100.000 loss: 0.8561 (0.6974) acc1: 46.0938 (55.4307) acc5: 100.0000 (100.0000) acc1_neg: 46.615 acc5_neg: 100.000 time: 0.1975 data: 0.1059 max mem: 3531 +Test: [180/333] eta: 0:00:33 acc1_pos: 50.987 acc5_pos: 100.000 loss: 0.8570 (0.7055) acc1: 44.7917 (54.9450) acc5: 100.0000 (100.0000) acc1_neg: 54.427 acc5_neg: 100.000 time: 0.1992 data: 0.1048 max mem: 3531 +Test: [190/333] eta: 0:00:30 acc1_pos: 50.987 acc5_pos: 100.000 loss: 0.7668 (0.7054) acc1: 50.5208 (55.1443) acc5: 100.0000 (100.0000) acc1_neg: 61.198 acc5_neg: 100.000 time: 0.1926 data: 0.0963 max mem: 3531 +Test: [200/333] eta: 0:00:28 acc1_pos: 50.987 acc5_pos: 100.000 loss: 0.7169 (0.7067) acc1: 58.5938 (55.3936) acc5: 100.0000 (100.0000) acc1_neg: 58.854 acc5_neg: 100.000 time: 0.1879 data: 0.0937 max mem: 3531 +Test: [210/333] eta: 0:00:26 acc1_pos: 50.987 acc5_pos: 100.000 loss: 0.7963 (0.7123) acc1: 54.1667 (55.2441) acc5: 100.0000 (100.0000) acc1_neg: 54.167 acc5_neg: 100.000 time: 0.1997 data: 0.1068 max mem: 3532 +Test: [220/333] eta: 0:00:24 acc1_pos: 50.987 acc5_pos: 100.000 loss: 0.8138 (0.7206) acc1: 51.3021 (54.8336) acc5: 100.0000 (100.0000) acc1_neg: 62.240 acc5_neg: 100.000 time: 0.2143 data: 0.1209 max mem: 3532 +Test: [230/333] eta: 0:00:21 acc1_pos: 50.987 acc5_pos: 100.000 loss: 0.6590 (0.7162) acc1: 63.2812 (55.4947) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.1994 data: 0.1055 max mem: 3532 +Test: [240/333] eta: 0:00:19 acc1_pos: 50.987 acc5_pos: 100.000 loss: 0.6333 (0.7143) acc1: 64.3229 (55.7411) acc5: 100.0000 (100.0000) acc1_neg: 60.156 acc5_neg: 100.000 time: 0.1849 data: 0.0904 max mem: 3532 +Test: [250/333] eta: 0:00:17 acc1_pos: 50.987 acc5_pos: 100.000 loss: 0.6707 (0.7131) acc1: 62.7604 (55.9771) acc5: 100.0000 (100.0000) acc1_neg: 52.083 acc5_neg: 100.000 time: 0.1949 data: 0.0997 max mem: 3532 +Test: [260/333] eta: 0:00:15 acc1_pos: 50.987 acc5_pos: 100.000 loss: 0.6875 (0.7130) acc1: 60.4167 (56.1343) acc5: 100.0000 (100.0000) acc1_neg: 64.844 acc5_neg: 100.000 time: 0.1972 data: 0.1037 max mem: 3532 +Test: [270/333] eta: 0:00:13 acc1_pos: 50.987 acc5_pos: 100.000 loss: 0.7267 (0.7140) acc1: 58.0729 (56.1510) acc5: 100.0000 (100.0000) acc1_neg: 58.333 acc5_neg: 100.000 time: 0.2004 data: 0.1083 max mem: 3532 +Test: [280/333] eta: 0:00:11 acc1_pos: 50.987 acc5_pos: 100.000 loss: 0.7337 (0.7186) acc1: 53.3854 (55.8793) acc5: 100.0000 (100.0000) acc1_neg: 52.083 acc5_neg: 100.000 time: 0.1876 data: 0.0952 max mem: 3532 +Test: [290/333] eta: 0:00:08 acc1_pos: 50.987 acc5_pos: 100.000 loss: 0.8985 (0.7258) acc1: 41.4062 (55.3381) acc5: 100.0000 (100.0000) acc1_neg: 40.625 acc5_neg: 100.000 time: 0.1850 data: 0.0898 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 50.987 acc5_pos: 100.000 loss: 0.9200 (0.7325) acc1: 40.6250 (54.9211) acc5: 100.0000 (100.0000) acc1_neg: 48.958 acc5_neg: 100.000 time: 0.1949 data: 0.1000 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 50.987 acc5_pos: 100.000 loss: 0.8240 (0.7308) acc1: 48.9583 (55.2075) acc5: 100.0000 (100.0000) acc1_neg: 70.312 acc5_neg: 100.000 time: 0.1651 data: 0.0719 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 50.987 acc5_pos: 100.000 loss: 0.6410 (0.7286) acc1: 66.4062 (55.4898) acc5: 100.0000 (100.0000) acc1_neg: 38.542 acc5_neg: 100.000 time: 0.1400 data: 0.0482 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 50.987 acc5_pos: 100.000 loss: 0.9391 (0.7375) acc1: 42.7083 (54.9983) acc5: 100.0000 (100.0000) acc1_neg: 41.406 acc5_neg: 100.000 time: 0.1270 data: 0.0367 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 50.987 acc5_pos: 100.000 loss: 0.9850 (0.7393) acc1: 41.4062 (54.9111) acc5: 100.0000 (100.0000) acc1_neg: 41.690 acc5_neg: 100.000 time: 0.1303 data: 0.0367 max mem: 3532 +Test: Total time: 0:01:06 (0.2007 s / it) +* Overall Acc@1 54.911 Acc@5 100.000 + target == real Acc@1 56.473 Acc@5 100.000 + target == fake Acc@1 53.521 Acc@5 100.000 + loss 0.739 + PR-AUC 0.5858 ROC-AUC 0.5764 EER 44.96% (thr=0.4905) +Accuracy of the network on the 127843 test images: 54.9% +[rank0]:[W726 06:31:59.534146446 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 06:32:19.550000 1202585 torch/distributed/run.py:766] +W0726 06:32:19.550000 1202585 torch/distributed/run.py:766] ***************************************** +W0726 06:32:19.550000 1202585 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:32:19.550000 1202585 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:32:49.268102044 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W726 06:32:49.481466481 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 06:32:50.978654866 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:32:50.063532805 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +NCCL version 2.26.2+cuda12.2 +[rank7]:[W726 06:32:50.318737254 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +[rank3]:[W726 06:32:50.431172231 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:32:50.439851144 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:32:50.510605164 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n22_seed9_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed9_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:20:39 acc1_pos: 95.052 acc5_pos: 100.000 loss: 0.2130 (0.2130) acc1: 95.0521 (95.0521) acc5: 100.0000 (100.0000) time: 5.2758 data: 3.0579 max mem: 3531 +Test: [ 10/235] eta: 0:02:09 acc1_pos: 94.271 acc5_pos: 100.000 loss: 0.2235 (0.2243) acc1: 94.7917 (94.2945) acc5: 100.0000 (100.0000) time: 0.5738 data: 0.2901 max mem: 3531 +Test: [ 20/235] eta: 0:01:18 acc1_pos: 93.490 acc5_pos: 100.000 loss: 0.2143 (0.2182) acc1: 94.7917 (94.5189) acc5: 100.0000 (100.0000) time: 0.1218 data: 0.0286 max mem: 3531 +Test: [ 30/235] eta: 0:01:02 acc1_pos: 96.354 acc5_pos: 100.000 loss: 0.2104 (0.2168) acc1: 95.0521 (94.5649) acc5: 100.0000 (100.0000) time: 0.1566 data: 0.0606 max mem: 3531 +Test: [ 40/235] eta: 0:00:59 acc1_pos: 93.750 acc5_pos: 100.000 loss: 0.2127 (0.2165) acc1: 94.7917 (94.5948) acc5: 100.0000 (100.0000) time: 0.2368 data: 0.1412 max mem: 3531 +Test: [ 50/235] eta: 0:00:53 acc1_pos: 94.271 acc5_pos: 100.000 loss: 0.2131 (0.2169) acc1: 94.7917 (94.6078) acc5: 100.0000 (100.0000) time: 0.2648 data: 0.1711 max mem: 3531 +Test: [ 60/235] eta: 0:00:49 acc1_pos: 93.750 acc5_pos: 100.000 loss: 0.2187 (0.2181) acc1: 94.2708 (94.5611) acc5: 100.0000 (100.0000) time: 0.2386 data: 0.1475 max mem: 3531 +Test: [ 70/235] eta: 0:00:44 acc1_pos: 95.385 acc5_pos: 100.000 loss: 0.2385 (0.3820) acc1: 92.4479 (85.1452) acc5: 100.0000 (100.0000) acc1_neg: 16.406 acc5_neg: 100.000 time: 0.2153 data: 0.1220 max mem: 3531 +Test: [ 80/235] eta: 0:00:39 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.5369 (0.5262) acc1: 19.2708 (76.9419) acc5: 100.0000 (100.0000) acc1_neg: 17.448 acc5_neg: 100.000 time: 0.1611 data: 0.0661 max mem: 3531 +Test: [ 90/235] eta: 0:00:35 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.5513 (0.6384) acc1: 18.7500 (70.6158) acc5: 100.0000 (100.0000) acc1_neg: 16.927 acc5_neg: 100.000 time: 0.1639 data: 0.0717 max mem: 3531 +Test: [100/235] eta: 0:00:32 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.5481 (0.7276) acc1: 19.0104 (65.6044) acc5: 100.0000 (100.0000) acc1_neg: 17.708 acc5_neg: 100.000 time: 0.1791 data: 0.0846 max mem: 3531 +Test: [110/235] eta: 0:00:30 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.5342 (0.8004) acc1: 19.2708 (61.4114) acc5: 100.0000 (100.0000) acc1_neg: 19.792 acc5_neg: 100.000 time: 0.2339 data: 0.1379 max mem: 3531 +Test: [120/235] eta: 0:00:27 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.5721 (0.8912) acc1: 17.1875 (56.8655) acc5: 100.0000 (100.0000) acc1_neg: 4.688 acc5_neg: 100.000 time: 0.2339 data: 0.1381 max mem: 3531 +Test: [130/235] eta: 0:00:24 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.9414 (0.9709) acc1: 5.2083 (52.9501) acc5: 100.0000 (100.0000) acc1_neg: 6.510 acc5_neg: 100.000 time: 0.1516 data: 0.0539 max mem: 3531 +Test: [140/235] eta: 0:00:22 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.9463 (1.0407) acc1: 5.2083 (49.5918) acc5: 100.0000 (100.0000) acc1_neg: 4.948 acc5_neg: 100.000 time: 0.2028 data: 0.1065 max mem: 3531 +Test: [150/235] eta: 0:00:19 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.9435 (1.0999) acc1: 5.4688 (46.6646) acc5: 100.0000 (100.0000) acc1_neg: 5.469 acc5_neg: 100.000 time: 0.2347 data: 0.1412 max mem: 3531 +Test: [160/235] eta: 0:00:16 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.9353 (1.1518) acc1: 5.4688 (44.1479) acc5: 100.0000 (100.0000) acc1_neg: 8.073 acc5_neg: 100.000 time: 0.1687 data: 0.0755 max mem: 3531 +Test: [170/235] eta: 0:00:14 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.8870 (1.1632) acc1: 8.0729 (43.1576) acc5: 100.0000 (100.0000) acc1_neg: 26.562 acc5_neg: 100.000 time: 0.1285 data: 0.0341 max mem: 3531 +Test: [180/235] eta: 0:00:11 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.2912 (1.1690) acc1: 29.4271 (42.4177) acc5: 100.0000 (100.0000) acc1_neg: 30.208 acc5_neg: 100.000 time: 0.1500 data: 0.0562 max mem: 3531 +Test: [190/235] eta: 0:00:09 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.2462 (1.1733) acc1: 30.2083 (41.7730) acc5: 100.0000 (100.0000) acc1_neg: 30.990 acc5_neg: 100.000 time: 0.1970 data: 0.1050 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.2462 (1.1783) acc1: 29.9479 (41.1575) acc5: 100.0000 (100.0000) acc1_neg: 28.646 acc5_neg: 100.000 time: 0.2132 data: 0.1208 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.2619 (1.1818) acc1: 29.9479 (40.6583) acc5: 100.0000 (100.0000) acc1_neg: 33.333 acc5_neg: 100.000 time: 0.2049 data: 0.1094 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.2601 (1.1854) acc1: 29.9479 (40.1949) acc5: 100.0000 (100.0000) acc1_neg: 29.948 acc5_neg: 100.000 time: 0.1705 data: 0.0771 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.2411 (1.1879) acc1: 31.2500 (39.8404) acc5: 100.0000 (100.0000) acc1_neg: 32.812 acc5_neg: 100.000 time: 0.1384 data: 0.0485 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.2509 (1.1903) acc1: 30.7292 (39.7023) acc5: 100.0000 (100.0000) acc1_neg: 34.783 acc5_neg: 100.000 time: 0.1160 data: 0.0293 max mem: 3532 +Test: Total time: 0:00:48 (0.2067 s / it) +* Overall Acc@1 39.702 Acc@5 100.000 + target == real Acc@1 94.575 Acc@5 100.000 + target == fake Acc@1 19.856 Acc@5 100.000 + loss 1.190 + PR-AUC 0.8462 ROC-AUC 0.6689 EER 38.54% (thr=0.1511) +Accuracy of the network on the 89879 test images: 39.7% +[rank0]:[W726 06:33:47.265066839 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0726 06:34:06.385000 1203132 torch/distributed/run.py:766] +W0726 06:34:06.385000 1203132 torch/distributed/run.py:766] ***************************************** +W0726 06:34:06.385000 1203132 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0726 06:34:06.385000 1203132 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W726 06:34:33.277617111 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W726 06:34:33.550055232 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W726 06:34:33.640881658 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W726 06:34:33.698943098 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W726 06:34:33.889679651 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank4]:[W726 06:34:33.889698867 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W726 06:34:34.923186785 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W726 06:34:34.953393358 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed5_9_benchmark_eval_20260725_1958_seed5_9/fixed500k_fake_n22_seed9_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed9_resampled_n116-041-188_20260725_1958_seed5_9/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:26:26 acc1_pos: 59.375 acc5_pos: 100.000 loss: 0.6705 (0.6705) acc1: 59.3750 (59.3750) acc5: 100.0000 (100.0000) time: 5.8963 data: 3.9752 max mem: 3531 +Test: [ 10/269] eta: 0:02:40 acc1_pos: 93.229 acc5_pos: 100.000 loss: 0.5939 (0.5999) acc1: 64.0625 (67.5663) acc5: 100.0000 (100.0000) time: 0.6180 data: 0.3618 max mem: 3531 +Test: [ 20/269] eta: 0:01:31 acc1_pos: 83.333 acc5_pos: 100.000 loss: 0.5939 (0.6114) acc1: 65.1042 (67.0139) acc5: 100.0000 (100.0000) time: 0.0921 data: 0.0004 max mem: 3531 +Test: [ 30/269] eta: 0:01:13 acc1_pos: 70.312 acc5_pos: 100.000 loss: 0.6064 (0.6260) acc1: 65.1042 (65.0454) acc5: 100.0000 (100.0000) time: 0.1378 data: 0.0438 max mem: 3531 +Test: [ 40/269] eta: 0:01:09 acc1_pos: 67.241 acc5_pos: 100.000 loss: 0.6421 (0.6282) acc1: 63.0208 (65.0025) acc5: 100.0000 (100.0000) acc1_neg: 65.104 acc5_neg: 100.000 time: 0.2310 data: 0.1328 max mem: 3531 +Test: [ 50/269] eta: 0:01:02 acc1_pos: 67.241 acc5_pos: 100.000 loss: 0.6928 (0.6611) acc1: 50.5208 (60.5647) acc5: 100.0000 (100.0000) acc1_neg: 19.531 acc5_neg: 100.000 time: 0.2446 data: 0.1477 max mem: 3531 +Test: [ 60/269] eta: 0:00:55 acc1_pos: 67.241 acc5_pos: 100.000 loss: 0.8088 (0.6878) acc1: 40.1042 (57.2020) acc5: 100.0000 (100.0000) acc1_neg: 40.104 acc5_neg: 100.000 time: 0.1988 data: 0.1059 max mem: 3531 +Test: [ 70/269] eta: 0:00:52 acc1_pos: 67.241 acc5_pos: 100.000 loss: 0.8088 (0.7003) acc1: 42.9688 (55.6448) acc5: 100.0000 (100.0000) acc1_neg: 59.635 acc5_neg: 100.000 time: 0.2060 data: 0.1133 max mem: 3531 +Test: [ 80/269] eta: 0:00:48 acc1_pos: 67.241 acc5_pos: 100.000 loss: 0.7509 (0.7020) acc1: 49.2188 (55.3659) acc5: 100.0000 (100.0000) acc1_neg: 65.365 acc5_neg: 100.000 time: 0.2209 data: 0.1295 max mem: 3531 +Test: [ 90/269] eta: 0:00:44 acc1_pos: 67.241 acc5_pos: 100.000 loss: 0.7608 (0.7106) acc1: 47.1354 (54.2926) acc5: 100.0000 (100.0000) acc1_neg: 50.260 acc5_neg: 100.000 time: 0.1987 data: 0.1074 max mem: 3531 +Test: [100/269] eta: 0:00:40 acc1_pos: 67.241 acc5_pos: 100.000 loss: 0.7993 (0.7211) acc1: 41.9271 (52.5913) acc5: 100.0000 (100.0000) acc1_neg: 21.615 acc5_neg: 100.000 time: 0.1868 data: 0.0929 max mem: 3531 +Test: [110/269] eta: 0:00:38 acc1_pos: 67.241 acc5_pos: 100.000 loss: 0.8109 (0.7301) acc1: 39.3229 (51.2012) acc5: 100.0000 (100.0000) acc1_neg: 30.729 acc5_neg: 100.000 time: 0.2001 data: 0.1060 max mem: 3531 +Test: [120/269] eta: 0:00:35 acc1_pos: 67.241 acc5_pos: 100.000 loss: 0.7799 (0.7271) acc1: 40.6250 (51.7218) acc5: 100.0000 (100.0000) acc1_neg: 59.375 acc5_neg: 100.000 time: 0.2020 data: 0.1091 max mem: 3531 +Test: [130/269] eta: 0:00:32 acc1_pos: 67.241 acc5_pos: 100.000 loss: 0.7017 (0.7365) acc1: 52.6042 (50.9125) acc5: 100.0000 (100.0000) acc1_neg: 52.604 acc5_neg: 100.000 time: 0.2008 data: 0.1063 max mem: 3531 +Test: [140/269] eta: 0:00:30 acc1_pos: 67.241 acc5_pos: 100.000 loss: 0.7363 (0.7385) acc1: 48.1771 (50.5061) acc5: 100.0000 (100.0000) acc1_neg: 41.667 acc5_neg: 100.000 time: 0.2199 data: 0.1258 max mem: 3531 +Test: [150/269] eta: 0:00:27 acc1_pos: 67.241 acc5_pos: 100.000 loss: 0.7297 (0.7328) acc1: 48.1771 (51.1814) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1993 data: 0.1049 max mem: 3531 +Test: [160/269] eta: 0:00:25 acc1_pos: 67.241 acc5_pos: 100.000 loss: 0.6081 (0.7270) acc1: 57.0312 (51.8326) acc5: 100.0000 (100.0000) acc1_neg: 46.354 acc5_neg: 100.000 time: 0.2219 data: 0.1253 max mem: 3531 +Test: [170/269] eta: 0:00:22 acc1_pos: 67.241 acc5_pos: 100.000 loss: 0.7888 (0.7343) acc1: 41.6667 (50.7097) acc5: 100.0000 (100.0000) acc1_neg: 34.896 acc5_neg: 100.000 time: 0.2511 data: 0.1565 max mem: 3531 +Test: [180/269] eta: 0:00:19 acc1_pos: 67.241 acc5_pos: 100.000 loss: 0.7664 (0.7342) acc1: 41.6667 (50.7525) acc5: 100.0000 (100.0000) acc1_neg: 27.344 acc5_neg: 100.000 time: 0.1632 data: 0.0709 max mem: 3531 +Test: [190/269] eta: 0:00:17 acc1_pos: 67.241 acc5_pos: 100.000 loss: 0.7363 (0.7370) acc1: 46.3542 (50.2931) acc5: 100.0000 (100.0000) acc1_neg: 31.250 acc5_neg: 100.000 time: 0.1496 data: 0.0546 max mem: 3531 +Test: [200/269] eta: 0:00:15 acc1_pos: 67.241 acc5_pos: 100.000 loss: 0.8887 (0.7473) acc1: 32.0312 (49.1747) acc5: 100.0000 (100.0000) acc1_neg: 33.594 acc5_neg: 100.000 time: 0.1748 data: 0.0807 max mem: 3531 +Test: [210/269] eta: 0:00:12 acc1_pos: 67.241 acc5_pos: 100.000 loss: 0.8324 (0.7412) acc1: 33.5938 (49.9877) acc5: 100.0000 (100.0000) acc1_neg: 70.312 acc5_neg: 100.000 time: 0.1631 data: 0.0714 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 67.241 acc5_pos: 100.000 loss: 0.5676 (0.7335) acc1: 70.3125 (50.9721) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1797 data: 0.0839 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 67.241 acc5_pos: 100.000 loss: 0.5180 (0.7243) acc1: 76.3021 (52.1172) acc5: 100.0000 (100.0000) acc1_neg: 64.844 acc5_neg: 100.000 time: 0.1699 data: 0.0740 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 67.241 acc5_pos: 100.000 loss: 0.6377 (0.7313) acc1: 43.2292 (51.2686) acc5: 100.0000 (100.0000) acc1_neg: 43.229 acc5_neg: 100.000 time: 0.1601 data: 0.0689 max mem: 3532 +Test: [250/269] eta: 0:00:03 acc1_pos: 67.241 acc5_pos: 100.000 loss: 0.8937 (0.7331) acc1: 28.9062 (51.0749) acc5: 100.0000 (100.0000) acc1_neg: 23.177 acc5_neg: 100.000 time: 0.1537 data: 0.0630 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 67.241 acc5_pos: 100.000 loss: 0.8353 (0.7389) acc1: 34.8958 (50.4121) acc5: 100.0000 (100.0000) acc1_neg: 53.906 acc5_neg: 100.000 time: 0.1305 data: 0.0397 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 67.241 acc5_pos: 100.000 loss: 0.8246 (0.7411) acc1: 39.3229 (50.1321) acc5: 100.0000 (100.0000) acc1_neg: 55.906 acc5_neg: 100.000 time: 0.1356 data: 0.0417 max mem: 3532 +Test: Total time: 0:00:54 (0.2037 s / it) +* Overall Acc@1 50.132 Acc@5 100.000 + target == real Acc@1 64.616 Acc@5 100.000 + target == fake Acc@1 48.155 Acc@5 100.000 + loss 0.741 + PR-AUC 0.9102 ROC-AUC 0.5946 EER 43.53% (thr=0.4646) +Accuracy of the network on the 103293 test images: 50.1% +[rank0]:[W726 06:35:38.383568539 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-26T06:35:45Z] eval_done tag=fixed500k_fake_n22_seed9_resampled +eval_queue_done_utc=2026-07-26T06:35:45Z diff --git a/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_1958_seed5_9/eval_queue_n116-041-188.pid b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_1958_seed5_9/eval_queue_n116-041-188.pid new file mode 100644 index 0000000000000000000000000000000000000000..5a920b536fc876363ce9dd409572d46399fff07b --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_20260725_1958_seed5_9/eval_queue_n116-041-188.pid @@ -0,0 +1 @@ +1163803 diff --git a/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_early_20260725_112559_anchor/early_eval_watch.log b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_early_20260725_112559_anchor/early_eval_watch.log new file mode 100644 index 0000000000000000000000000000000000000000..cd491a7fc98924e272b87615c163fa22bd105bb3 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_early_20260725_112559_anchor/early_eval_watch.log @@ -0,0 +1,10407 @@ +[2026-07-25T14:47:38Z] early_eval_watch_started host=n116-042-186 train_tag=20260725_112559 +[2026-07-25T14:47:38Z] train_log_dir=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_seed1_4_20260725_112559 +[2026-07-25T14:47:38Z] eval_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559 +[2026-07-25T14:47:39Z] scan completed_train=28/48 eval_complete=0/48 queue_done=0/6 +[2026-07-25T14:47:39Z] eval_start tag=fixed500k_fake_n11_seed1_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed1_resampled_n116-040-168_20260725_112559/best_checkpoint.pth +started_utc=2026-07-25T14:47:39Z +hostname=n116-042-186 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed1_resampled_n116-040-168_20260725_112559/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed1_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0725 14:47:51.810000 64168 torch/distributed/run.py:766] +W0725 14:47:51.810000 64168 torch/distributed/run.py:766] ***************************************** +W0725 14:47:51.810000 64168 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 14:47:51.810000 64168 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 14:48:18.960711158 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 14:48:20.090650106 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +[rank2]:[W725 14:48:20.167828190 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 14:48:20.175939356 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 14:48:20.240651875 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 14:48:20.276330692 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +[rank7]:[W725 14:48:20.354195593 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 14:48:20.359164643 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed1_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed1_resampled_n116-040-168_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:25:58 acc1_pos: 91.406 acc5_pos: 100.000 loss: 0.2901 (0.2901) acc1: 91.4062 (91.4062) acc5: 100.0000 (100.0000) time: 6.6586 data: 5.1801 max mem: 3531 +Test: [ 10/234] eta: 0:02:33 acc1_pos: 93.229 acc5_pos: 100.000 loss: 0.3782 (0.4255) acc1: 86.7188 (84.5407) acc5: 100.0000 (100.0000) time: 0.6874 data: 0.4713 max mem: 3531 +Test: [ 20/234] eta: 0:01:27 acc1_pos: 96.094 acc5_pos: 100.000 loss: 0.3673 (0.3660) acc1: 88.5417 (87.9712) acc5: 100.0000 (100.0000) time: 0.0947 data: 0.0004 max mem: 3531 +Test: [ 30/234] eta: 0:01:09 acc1_pos: 92.448 acc5_pos: 100.000 loss: 0.2060 (0.3106) acc1: 96.0938 (90.7426) acc5: 100.0000 (100.0000) time: 0.1511 data: 0.0552 max mem: 3531 +Test: [ 40/234] eta: 0:01:00 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1996 (0.2878) acc1: 97.1354 (91.9843) acc5: 100.0000 (100.0000) time: 0.2066 data: 0.1115 max mem: 3531 +Test: [ 50/234] eta: 0:00:52 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.1381 (0.2574) acc1: 98.4375 (93.3211) acc5: 100.0000 (100.0000) time: 0.2033 data: 0.1094 max mem: 3531 +Test: [ 60/234] eta: 0:00:46 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.1070 (0.2302) acc1: 99.4792 (94.4117) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1882 data: 0.0959 max mem: 3531 +Test: [ 70/234] eta: 0:00:43 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.0920 (0.2915) acc1: 100.0000 (89.7961) acc5: 100.0000 (100.0000) acc1_neg: 41.146 acc5_neg: 100.000 time: 0.2049 data: 0.1098 max mem: 3531 +Test: [ 80/234] eta: 0:00:40 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.8540 (0.3386) acc1: 44.7917 (86.2847) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2400 data: 0.1452 max mem: 3531 +Test: [ 90/234] eta: 0:00:36 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.4011 (0.3426) acc1: 81.5104 (86.1292) acc5: 100.0000 (100.0000) acc1_neg: 70.833 acc5_neg: 100.000 time: 0.2299 data: 0.1365 max mem: 3531 +Test: [100/234] eta: 0:00:33 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.4175 (0.3625) acc1: 81.5104 (84.6999) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1998 data: 0.1033 max mem: 3531 +Test: [110/234] eta: 0:00:30 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.4776 (0.3771) acc1: 74.2188 (83.7838) acc5: 100.0000 (100.0000) acc1_neg: 77.083 acc5_neg: 100.000 time: 0.1899 data: 0.0941 max mem: 3531 +Test: [120/234] eta: 0:00:27 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.4748 (0.3965) acc1: 74.7396 (82.3842) acc5: 100.0000 (100.0000) acc1_neg: 35.417 acc5_neg: 100.000 time: 0.1949 data: 0.1012 max mem: 3531 +Test: [130/234] eta: 0:00:24 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.7076 (0.4293) acc1: 48.4375 (79.9300) acc5: 100.0000 (100.0000) acc1_neg: 67.969 acc5_neg: 100.000 time: 0.2010 data: 0.1074 max mem: 3531 +Test: [140/234] eta: 0:00:21 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.7718 (0.4570) acc1: 52.6042 (78.7825) acc5: 100.0000 (100.0000) acc1_neg: 42.969 acc5_neg: 100.000 time: 0.1806 data: 0.0897 max mem: 3531 +Test: [150/234] eta: 0:00:19 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.5914 (0.4707) acc1: 70.0521 (78.0871) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.1789 data: 0.0887 max mem: 3531 +Test: [160/234] eta: 0:00:16 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.5914 (0.4782) acc1: 69.7917 (77.4829) acc5: 100.0000 (100.0000) acc1_neg: 54.948 acc5_neg: 100.000 time: 0.1951 data: 0.1048 max mem: 3531 +Test: [170/234] eta: 0:00:14 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.6961 (0.4905) acc1: 58.3333 (76.4955) acc5: 100.0000 (100.0000) acc1_neg: 67.708 acc5_neg: 100.000 time: 0.1916 data: 0.1014 max mem: 3531 +Test: [180/234] eta: 0:00:12 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.6825 (0.5087) acc1: 60.1562 (75.3798) acc5: 100.0000 (100.0000) acc1_neg: 19.792 acc5_neg: 100.000 time: 0.2054 data: 0.1147 max mem: 3531 +Test: [190/234] eta: 0:00:09 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.6329 (0.5175) acc1: 68.4896 (74.9427) acc5: 100.0000 (100.0000) acc1_neg: 71.615 acc5_neg: 100.000 time: 0.1881 data: 0.0956 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.6101 (0.5074) acc1: 69.7917 (75.5960) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1599 data: 0.0674 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.1294 (0.4899) acc1: 99.2188 (76.7044) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1577 data: 0.0668 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.1515 (0.4766) acc1: 98.4375 (77.5441) acc5: 100.0000 (100.0000) acc1_neg: 70.052 acc5_neg: 100.000 time: 0.1561 data: 0.0660 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.5682 (0.5124) acc1: 22.6562 (74.9763) acc5: 100.0000 (100.0000) acc1_neg: 15.365 acc5_neg: 100.000 time: 0.1270 data: 0.0373 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 99.310 acc5_pos: 100.000 loss: 1.2865 (0.5225) acc1: 19.4553 (74.3127) acc5: 100.0000 (100.0000) acc1_neg: 19.455 acc5_neg: 100.000 time: 0.1182 data: 0.0267 max mem: 3532 +Test: Total time: 0:00:48 (0.2073 s / it) +* Overall Acc@1 74.313 Acc@5 100.000 + target == real Acc@1 93.365 Acc@5 100.000 + target == fake Acc@1 68.943 Acc@5 100.000 + loss 0.523 + PR-AUC 0.9699 ROC-AUC 0.8963 EER 18.79% (thr=0.3215) +Accuracy of the network on the 89729 test images: 74.3% +[rank0]:[W725 14:49:22.617386657 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 14:49:41.930000 64753 torch/distributed/run.py:766] +W0725 14:49:41.930000 64753 torch/distributed/run.py:766] ***************************************** +W0725 14:49:41.930000 64753 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 14:49:41.930000 64753 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 14:50:09.076412656 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 14:50:09.636384770 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 14:50:09.868112842 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 14:50:09.921428184 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 14:50:09.932119152 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 4): env:// +NCCL version 2.26.2+cuda12.2 +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 14:50:10.055849025 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 14:50:10.062684909 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 14:50:10.119304144 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed1_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed1_resampled_n116-040-168_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 3:39:22 acc1_pos: 84.896 acc5_pos: 100.000 loss: 0.4130 (0.4130) acc1: 84.8958 (84.8958) acc5: 100.0000 (100.0000) time: 5.5143 data: 3.0892 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/2387] eta: 0:23:05 acc1_pos: 67.969 acc5_pos: 100.000 loss: 0.4994 (0.5054) acc1: 80.7292 (78.1487) acc5: 100.0000 (100.0000) time: 0.5830 data: 0.2812 max mem: 3531 +Test: [ 20/2387] eta: 0:14:49 acc1_pos: 78.906 acc5_pos: 100.000 loss: 0.4768 (0.4981) acc1: 79.9479 (78.5714) acc5: 100.0000 (100.0000) time: 0.1187 data: 0.0288 max mem: 3531 +Test: [ 30/2387] eta: 0:13:46 acc1_pos: 74.479 acc5_pos: 100.000 loss: 0.4491 (0.4893) acc1: 81.2500 (79.3683) acc5: 100.0000 (100.0000) time: 0.2230 data: 0.1294 max mem: 3531 +Test: [ 40/2387] eta: 0:14:05 acc1_pos: 83.594 acc5_pos: 100.000 loss: 0.4288 (0.4786) acc1: 83.5938 (80.1766) acc5: 100.0000 (100.0000) time: 0.3442 data: 0.2440 max mem: 3531 +Test: [ 50/2387] eta: 0:13:38 acc1_pos: 65.365 acc5_pos: 100.000 loss: 0.4234 (0.4887) acc1: 83.3333 (79.0697) acc5: 100.0000 (100.0000) time: 0.3499 data: 0.2515 max mem: 3531 +Test: [ 60/2387] eta: 0:13:16 acc1_pos: 82.031 acc5_pos: 100.000 loss: 0.4620 (0.4875) acc1: 81.7708 (79.0258) acc5: 100.0000 (100.0000) time: 0.3049 data: 0.2099 max mem: 3531 +Test: [ 70/2387] eta: 0:12:42 acc1_pos: 60.156 acc5_pos: 100.000 loss: 0.4824 (0.4973) acc1: 79.1667 (78.1910) acc5: 100.0000 (100.0000) time: 0.2749 data: 0.1770 max mem: 3531 +Test: [ 80/2387] eta: 0:12:06 acc1_pos: 94.792 acc5_pos: 100.000 loss: 0.4727 (0.4863) acc1: 79.1667 (79.0766) acc5: 100.0000 (100.0000) time: 0.2316 data: 0.1342 max mem: 3531 +Test: [ 90/2387] eta: 0:11:33 acc1_pos: 95.573 acc5_pos: 100.000 loss: 0.3434 (0.4678) acc1: 87.5000 (80.2970) acc5: 100.0000 (100.0000) time: 0.2049 data: 0.1102 max mem: 3531 +Test: [ 100/2387] eta: 0:11:08 acc1_pos: 94.531 acc5_pos: 100.000 loss: 0.3118 (0.4575) acc1: 89.5833 (81.0901) acc5: 100.0000 (100.0000) time: 0.2002 data: 0.1083 max mem: 3531 +Test: [ 110/2387] eta: 0:11:04 acc1_pos: 76.562 acc5_pos: 100.000 loss: 0.3161 (0.4481) acc1: 89.3229 (81.6324) acc5: 100.0000 (100.0000) time: 0.2456 data: 0.1540 max mem: 3531 +Test: [ 120/2387] eta: 0:10:50 acc1_pos: 84.896 acc5_pos: 100.000 loss: 0.3504 (0.4432) acc1: 86.7188 (81.9409) acc5: 100.0000 (100.0000) time: 0.2597 data: 0.1667 max mem: 3531 +Test: [ 130/2387] eta: 0:10:41 acc1_pos: 82.031 acc5_pos: 100.000 loss: 0.3806 (0.4367) acc1: 88.8021 (82.4686) acc5: 100.0000 (100.0000) time: 0.2414 data: 0.1480 max mem: 3531 +Test: [ 140/2387] eta: 0:10:32 acc1_pos: 83.594 acc5_pos: 100.000 loss: 0.3823 (0.4327) acc1: 85.6771 (82.7331) acc5: 100.0000 (100.0000) time: 0.2501 data: 0.1558 max mem: 3531 +Test: [ 150/2387] eta: 0:10:20 acc1_pos: 89.844 acc5_pos: 100.000 loss: 0.3505 (0.4250) acc1: 89.8438 (83.3713) acc5: 100.0000 (100.0000) time: 0.2344 data: 0.1386 max mem: 3531 +Test: [ 160/2387] eta: 0:10:20 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.3240 (0.4251) acc1: 90.3646 (83.2120) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.2579 data: 0.1628 max mem: 3531 +Test: [ 170/2387] eta: 0:10:03 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2916 (0.4165) acc1: 90.1042 (83.6486) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.2332 data: 0.1374 max mem: 3531 +Test: [ 180/2387] eta: 0:09:50 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2875 (0.4099) acc1: 89.8438 (83.9563) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1798 data: 0.0839 max mem: 3531 +Test: [ 190/2387] eta: 0:09:41 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2941 (0.4038) acc1: 88.8021 (84.2278) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.1998 data: 0.1066 max mem: 3531 +Test: [ 200/2387] eta: 0:09:24 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2854 (0.3980) acc1: 89.5833 (84.5357) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1716 data: 0.0784 max mem: 3531 +Test: [ 210/2387] eta: 0:09:16 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2897 (0.3933) acc1: 90.1042 (84.7699) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1699 data: 0.0761 max mem: 3532 +Test: [ 220/2387] eta: 0:09:09 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2966 (0.3887) acc1: 88.8021 (84.9854) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.2094 data: 0.1145 max mem: 3532 +Test: [ 230/2387] eta: 0:09:03 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2922 (0.3844) acc1: 89.3229 (85.1946) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.2098 data: 0.1141 max mem: 3532 +Test: [ 240/2387] eta: 0:08:54 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.3032 (0.3838) acc1: 85.6771 (85.1833) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.1986 data: 0.1045 max mem: 3532 +Test: [ 250/2387] eta: 0:08:45 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2789 (0.3775) acc1: 85.6771 (85.5038) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1799 data: 0.0855 max mem: 3532 +Test: [ 260/2387] eta: 0:08:38 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.1729 (0.3696) acc1: 96.0938 (85.9275) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1775 data: 0.0827 max mem: 3532 +Test: [ 270/2387] eta: 0:08:30 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.1625 (0.3620) acc1: 96.8750 (86.3305) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1815 data: 0.0896 max mem: 3532 +Test: [ 280/2387] eta: 0:08:24 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.1714 (0.3554) acc1: 96.6146 (86.6872) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1873 data: 0.0967 max mem: 3532 +Test: [ 290/2387] eta: 0:08:18 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.1802 (0.3494) acc1: 95.3125 (86.9926) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1909 data: 0.0995 max mem: 3532 +Test: [ 300/2387] eta: 0:08:14 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.1790 (0.3438) acc1: 95.3125 (87.2820) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1959 data: 0.1038 max mem: 3532 +Test: [ 310/2387] eta: 0:08:08 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.1918 (0.3423) acc1: 94.7917 (87.4958) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1978 data: 0.1042 max mem: 3532 +Test: [ 320/2387] eta: 0:08:05 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.4464 (0.3456) acc1: 90.6250 (87.5803) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.2039 data: 0.1087 max mem: 3532 +Test: [ 330/2387] eta: 0:08:03 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.4490 (0.3487) acc1: 90.3646 (87.6668) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.2344 data: 0.1382 max mem: 3532 +Test: [ 340/2387] eta: 0:08:00 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.4431 (0.3515) acc1: 90.8854 (87.7627) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.2355 data: 0.1397 max mem: 3532 +Test: [ 350/2387] eta: 0:07:57 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.4326 (0.3483) acc1: 91.9271 (88.0616) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2209 data: 0.1243 max mem: 3532 +Test: [ 360/2387] eta: 0:07:47 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2238 (0.3453) acc1: 94.2708 (88.2026) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1643 data: 0.0647 max mem: 3532 +Test: [ 370/2387] eta: 0:07:45 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2388 (0.3425) acc1: 92.7083 (88.3213) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1699 data: 0.0694 max mem: 3532 +Test: [ 380/2387] eta: 0:07:41 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2382 (0.3398) acc1: 92.7083 (88.4364) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.2099 data: 0.1144 max mem: 3532 +Test: [ 390/2387] eta: 0:07:36 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2382 (0.3375) acc1: 92.7083 (88.5244) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1888 data: 0.0955 max mem: 3532 +Test: [ 400/2387] eta: 0:07:31 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2403 (0.3352) acc1: 92.4479 (88.6215) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1821 data: 0.0878 max mem: 3532 +Test: [ 410/2387] eta: 0:07:30 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2435 (0.3329) acc1: 92.4479 (88.7108) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.2099 data: 0.1177 max mem: 3532 +Test: [ 420/2387] eta: 0:07:26 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2444 (0.3307) acc1: 92.1875 (88.7971) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.2226 data: 0.1280 max mem: 3532 +Test: [ 430/2387] eta: 0:07:20 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2529 (0.3304) acc1: 91.1458 (88.7652) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.1649 data: 0.0666 max mem: 3532 +Test: [ 440/2387] eta: 0:07:14 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2776 (0.3272) acc1: 90.1042 (88.9078) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1354 data: 0.0385 max mem: 3533 +Test: [ 450/2387] eta: 0:07:11 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.1031 (0.3223) acc1: 100.0000 (89.1514) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1814 data: 0.0863 max mem: 3533 +Test: [ 460/2387] eta: 0:07:09 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.1034 (0.3176) acc1: 100.0000 (89.3845) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2263 data: 0.1324 max mem: 3533 +Test: [ 470/2387] eta: 0:07:07 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.1026 (0.3130) acc1: 100.0000 (89.6066) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2234 data: 0.1309 max mem: 3533 +Test: [ 480/2387] eta: 0:07:04 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.1045 (0.3087) acc1: 100.0000 (89.8199) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2086 data: 0.1171 max mem: 3533 +Test: [ 490/2387] eta: 0:06:59 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.1051 (0.3046) acc1: 100.0000 (90.0251) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1817 data: 0.0909 max mem: 3533 +Test: [ 500/2387] eta: 0:06:57 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.1038 (0.3005) acc1: 100.0000 (90.2242) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1842 data: 0.0902 max mem: 3533 +Test: [ 510/2387] eta: 0:06:52 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.1035 (0.2967) acc1: 100.0000 (90.4135) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1880 data: 0.0916 max mem: 3533 +Test: [ 520/2387] eta: 0:06:48 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.1066 (0.2931) acc1: 100.0000 (90.5910) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1705 data: 0.0755 max mem: 3533 +Test: [ 530/2387] eta: 0:06:45 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.1015 (0.2894) acc1: 99.4792 (90.7594) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1773 data: 0.0845 max mem: 3533 +Test: [ 540/2387] eta: 0:06:41 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.1009 (0.2860) acc1: 99.4792 (90.9230) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1700 data: 0.0792 max mem: 3533 +Test: [ 550/2387] eta: 0:06:37 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.1007 (0.2826) acc1: 99.4792 (91.0811) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1627 data: 0.0721 max mem: 3533 +Test: [ 560/2387] eta: 0:06:33 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.0993 (0.2794) acc1: 99.7396 (91.2340) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1701 data: 0.0775 max mem: 3533 +Test: [ 570/2387] eta: 0:06:30 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.0979 (0.2762) acc1: 99.7396 (91.3830) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1755 data: 0.0786 max mem: 3533 +Test: [ 580/2387] eta: 0:06:27 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.1000 (0.2732) acc1: 99.7396 (91.5246) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1790 data: 0.0830 max mem: 3533 +Test: [ 590/2387] eta: 0:06:24 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.1022 (0.2703) acc1: 99.4792 (91.6609) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1790 data: 0.0819 max mem: 3533 +Test: [ 600/2387] eta: 0:06:20 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.1059 (0.2677) acc1: 99.7396 (91.7949) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1749 data: 0.0764 max mem: 3533 +Test: [ 610/2387] eta: 0:06:17 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.1213 (0.2660) acc1: 98.9583 (91.8934) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1670 data: 0.0732 max mem: 3533 +Test: [ 620/2387] eta: 0:06:13 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.1660 (0.2644) acc1: 97.9167 (91.9891) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1668 data: 0.0727 max mem: 3533 +Test: [ 630/2387] eta: 0:06:10 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.1664 (0.2629) acc1: 97.6562 (92.0765) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1795 data: 0.0855 max mem: 3533 +Test: [ 640/2387] eta: 0:06:07 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.1701 (0.2615) acc1: 97.3958 (92.1611) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1829 data: 0.0904 max mem: 3533 +Test: [ 650/2387] eta: 0:06:04 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.1711 (0.2601) acc1: 97.3958 (92.2439) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1682 data: 0.0740 max mem: 3533 +Test: [ 660/2387] eta: 0:06:00 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.1696 (0.2587) acc1: 97.6562 (92.3266) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1569 data: 0.0624 max mem: 3533 +Test: [ 670/2387] eta: 0:05:59 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.1680 (0.2574) acc1: 97.6562 (92.4002) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1856 data: 0.0906 max mem: 3534 +Test: [ 680/2387] eta: 0:05:56 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.1663 (0.2560) acc1: 97.9167 (92.4999) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2045 data: 0.1101 max mem: 3534 +Test: [ 690/2387] eta: 0:05:54 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.1715 (0.2605) acc1: 97.1354 (92.1928) acc5: 100.0000 (100.0000) acc1_neg: 58.854 acc5_neg: 100.000 time: 0.2103 data: 0.1185 max mem: 3534 +Test: [ 700/2387] eta: 0:05:52 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.7118 (0.2669) acc1: 60.4167 (91.7425) acc5: 100.0000 (100.0000) acc1_neg: 65.625 acc5_neg: 100.000 time: 0.2143 data: 0.1227 max mem: 3534 +Test: [ 710/2387] eta: 0:05:50 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.7221 (0.2733) acc1: 60.4167 (91.3140) acc5: 100.0000 (100.0000) acc1_neg: 62.760 acc5_neg: 100.000 time: 0.2092 data: 0.1175 max mem: 3534 +Test: [ 720/2387] eta: 0:05:48 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.7126 (0.2793) acc1: 61.7188 (90.9009) acc5: 100.0000 (100.0000) acc1_neg: 61.979 acc5_neg: 100.000 time: 0.2188 data: 0.1253 max mem: 3534 +Test: [ 730/2387] eta: 0:05:46 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.7098 (0.2854) acc1: 60.9375 (90.4882) acc5: 100.0000 (100.0000) acc1_neg: 60.938 acc5_neg: 100.000 time: 0.2044 data: 0.1064 max mem: 3534 +Test: [ 740/2387] eta: 0:05:43 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.7028 (0.2910) acc1: 61.7188 (90.1073) acc5: 100.0000 (100.0000) acc1_neg: 60.417 acc5_neg: 100.000 time: 0.1921 data: 0.0936 max mem: 3534 +Test: [ 750/2387] eta: 0:05:41 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.6979 (0.2964) acc1: 62.5000 (89.7425) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.1968 data: 0.1027 max mem: 3534 +Test: [ 760/2387] eta: 0:05:39 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.6958 (0.3011) acc1: 61.7188 (89.4311) acc5: 100.0000 (100.0000) acc1_neg: 76.302 acc5_neg: 100.000 time: 0.2001 data: 0.1082 max mem: 3534 +Test: [ 770/2387] eta: 0:05:37 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.5602 (0.3039) acc1: 72.1354 (89.2381) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.2006 data: 0.1090 max mem: 3534 +Test: [ 780/2387] eta: 0:05:34 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.4870 (0.3057) acc1: 73.9583 (89.0968) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1924 data: 0.0997 max mem: 3534 +Test: [ 790/2387] eta: 0:05:31 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.3905 (0.3068) acc1: 80.9896 (89.0019) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.1797 data: 0.0870 max mem: 3534 +Test: [ 800/2387] eta: 0:05:30 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.3929 (0.3084) acc1: 80.9896 (88.8775) acc5: 100.0000 (100.0000) acc1_neg: 73.698 acc5_neg: 100.000 time: 0.2017 data: 0.1078 max mem: 3534 +Test: [ 810/2387] eta: 0:05:27 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.4138 (0.3103) acc1: 77.8646 (88.7318) acc5: 100.0000 (100.0000) acc1_neg: 72.917 acc5_neg: 100.000 time: 0.2026 data: 0.1079 max mem: 3534 +Test: [ 820/2387] eta: 0:05:24 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.4440 (0.3122) acc1: 77.3438 (88.5839) acc5: 100.0000 (100.0000) acc1_neg: 63.802 acc5_neg: 100.000 time: 0.1690 data: 0.0736 max mem: 3534 +Test: [ 830/2387] eta: 0:05:21 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.5435 (0.3171) acc1: 69.5312 (88.2421) acc5: 100.0000 (100.0000) acc1_neg: 66.927 acc5_neg: 100.000 time: 0.1625 data: 0.0680 max mem: 3534 +Test: [ 840/2387] eta: 0:05:19 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.5174 (0.3178) acc1: 74.2188 (88.2181) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.1752 data: 0.0847 max mem: 3534 +Test: [ 850/2387] eta: 0:05:16 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.3910 (0.3186) acc1: 84.1146 (88.1720) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.1822 data: 0.0876 max mem: 3534 +Test: [ 860/2387] eta: 0:05:13 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.3477 (0.3188) acc1: 86.1979 (88.1899) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1645 data: 0.0654 max mem: 3534 +Test: [ 870/2387] eta: 0:05:11 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.3147 (0.3188) acc1: 90.6250 (88.2244) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1649 data: 0.0698 max mem: 3534 +Test: [ 880/2387] eta: 0:05:08 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.3250 (0.3190) acc1: 90.6250 (88.2298) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1740 data: 0.0794 max mem: 3534 +Test: [ 890/2387] eta: 0:05:06 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.3263 (0.3191) acc1: 89.5833 (88.2435) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1809 data: 0.0841 max mem: 3534 +Test: [ 900/2387] eta: 0:05:04 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.3703 (0.3238) acc1: 84.6354 (87.8766) acc5: 100.0000 (100.0000) acc1_neg: 50.260 acc5_neg: 100.000 time: 0.1981 data: 0.1019 max mem: 3535 +Test: [ 910/2387] eta: 0:05:01 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.8425 (0.3296) acc1: 47.3958 (87.4254) acc5: 100.0000 (100.0000) acc1_neg: 51.302 acc5_neg: 100.000 time: 0.1917 data: 0.0979 max mem: 3535 +Test: [ 920/2387] eta: 0:04:59 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.8493 (0.3353) acc1: 46.8750 (86.9843) acc5: 100.0000 (100.0000) acc1_neg: 44.792 acc5_neg: 100.000 time: 0.1772 data: 0.0859 max mem: 3535 +Test: [ 930/2387] eta: 0:04:56 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.8498 (0.3407) acc1: 46.8750 (86.5624) acc5: 100.0000 (100.0000) acc1_neg: 44.271 acc5_neg: 100.000 time: 0.1671 data: 0.0738 max mem: 3535 +Test: [ 940/2387] eta: 0:04:53 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.8434 (0.3461) acc1: 47.3958 (86.1517) acc5: 100.0000 (100.0000) acc1_neg: 50.781 acc5_neg: 100.000 time: 0.1557 data: 0.0603 max mem: 3535 +Test: [ 950/2387] eta: 0:04:50 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.8478 (0.3514) acc1: 47.1354 (85.7319) acc5: 100.0000 (100.0000) acc1_neg: 48.438 acc5_neg: 100.000 time: 0.1564 data: 0.0630 max mem: 3535 +Test: [ 960/2387] eta: 0:04:48 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.8435 (0.3564) acc1: 47.3958 (85.3416) acc5: 100.0000 (100.0000) acc1_neg: 54.167 acc5_neg: 100.000 time: 0.1856 data: 0.0917 max mem: 3535 +Test: [ 970/2387] eta: 0:04:46 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.8374 (0.3615) acc1: 47.3958 (84.9481) acc5: 100.0000 (100.0000) acc1_neg: 50.260 acc5_neg: 100.000 time: 0.1988 data: 0.1023 max mem: 3535 +Test: [ 980/2387] eta: 0:04:44 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.8144 (0.3630) acc1: 48.9583 (84.8327) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.1841 data: 0.0908 max mem: 3535 +Test: [ 990/2387] eta: 0:04:42 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.3767 (0.3633) acc1: 82.2917 (84.8191) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.2006 data: 0.1064 max mem: 3535 +Test: [1000/2387] eta: 0:04:39 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.3962 (0.3638) acc1: 82.5521 (84.7910) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.1884 data: 0.0911 max mem: 3535 +Test: [1010/2387] eta: 0:04:37 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.4195 (0.3652) acc1: 79.1667 (84.7003) acc5: 100.0000 (100.0000) acc1_neg: 63.542 acc5_neg: 100.000 time: 0.1787 data: 0.0821 max mem: 3535 +Test: [1020/2387] eta: 0:04:35 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.4593 (0.3655) acc1: 73.6979 (84.6602) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1749 data: 0.0798 max mem: 3535 +Test: [1030/2387] eta: 0:04:32 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.1370 (0.3634) acc1: 97.1354 (84.7887) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1599 data: 0.0660 max mem: 3535 +Test: [1040/2387] eta: 0:04:30 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.1045 (0.3609) acc1: 99.4792 (84.9254) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0818 max mem: 3535 +Test: [1050/2387] eta: 0:04:28 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.1045 (0.3586) acc1: 99.4792 (85.0621) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1749 data: 0.0820 max mem: 3535 +Test: [1060/2387] eta: 0:04:25 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.1269 (0.3565) acc1: 98.9583 (85.1852) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1655 data: 0.0735 max mem: 3535 +Test: [1070/2387] eta: 0:04:23 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.1347 (0.3549) acc1: 98.1771 (85.2786) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.1603 data: 0.0674 max mem: 3535 +Test: [1080/2387] eta: 0:04:20 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.3235 (0.3555) acc1: 85.1562 (85.2331) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.1696 data: 0.0768 max mem: 3535 +Test: [1090/2387] eta: 0:04:18 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.4135 (0.3559) acc1: 80.7292 (85.2061) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.1856 data: 0.0935 max mem: 3535 +Test: [1100/2387] eta: 0:04:16 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.3782 (0.3558) acc1: 82.8125 (85.2064) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1751 data: 0.0824 max mem: 3535 +Test: [1110/2387] eta: 0:04:13 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.3776 (0.3561) acc1: 82.0312 (85.1783) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.1656 data: 0.0742 max mem: 3535 +Test: [1120/2387] eta: 0:04:11 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.3878 (0.3565) acc1: 82.0312 (85.1590) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.1718 data: 0.0796 max mem: 3536 +Test: [1130/2387] eta: 0:04:09 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.3804 (0.3566) acc1: 82.5521 (85.1362) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1740 data: 0.0815 max mem: 3536 +Test: [1140/2387] eta: 0:04:07 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.4425 (0.3583) acc1: 76.0417 (85.0252) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.1834 data: 0.0917 max mem: 3536 +Test: [1150/2387] eta: 0:04:05 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.5159 (0.3586) acc1: 74.4792 (85.0237) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1896 data: 0.0980 max mem: 3536 +Test: [1160/2387] eta: 0:04:03 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2312 (0.3573) acc1: 95.0521 (85.1278) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2145 data: 0.1207 max mem: 3536 +Test: [1170/2387] eta: 0:04:00 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2150 (0.3561) acc1: 96.8750 (85.2299) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1812 data: 0.0871 max mem: 3536 +Test: [1180/2387] eta: 0:03:58 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2196 (0.3549) acc1: 97.1354 (85.3296) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1198 data: 0.0277 max mem: 3536 +Test: [1190/2387] eta: 0:03:56 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2196 (0.3537) acc1: 97.3958 (85.4261) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1515 data: 0.0584 max mem: 3536 +Test: [1200/2387] eta: 0:03:53 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.1927 (0.3523) acc1: 97.9167 (85.5335) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1844 data: 0.0916 max mem: 3536 +Test: [1210/2387] eta: 0:03:51 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.1939 (0.3514) acc1: 97.3958 (85.5926) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1814 data: 0.0884 max mem: 3536 +Test: [1220/2387] eta: 0:03:49 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2323 (0.3503) acc1: 93.2292 (85.6547) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1899 data: 0.0951 max mem: 3536 +Test: [1230/2387] eta: 0:03:47 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2075 (0.3492) acc1: 94.0104 (85.7315) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1999 data: 0.1036 max mem: 3536 +Test: [1240/2387] eta: 0:03:45 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2017 (0.3479) acc1: 94.7917 (85.8116) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1756 data: 0.0819 max mem: 3536 +Test: [1250/2387] eta: 0:03:43 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2032 (0.3469) acc1: 94.7917 (85.8798) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1528 data: 0.0619 max mem: 3536 +Test: [1260/2387] eta: 0:03:40 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.1984 (0.3456) acc1: 95.0521 (85.9573) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1605 data: 0.0682 max mem: 3536 +Test: [1270/2387] eta: 0:03:39 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2000 (0.3447) acc1: 94.5312 (86.0076) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1883 data: 0.0958 max mem: 3536 +Test: [1280/2387] eta: 0:03:37 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2132 (0.3437) acc1: 92.9688 (86.0609) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.2154 data: 0.1237 max mem: 3536 +Test: [1290/2387] eta: 0:03:35 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.1990 (0.3427) acc1: 94.5312 (86.1223) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1846 data: 0.0934 max mem: 3536 +Test: [1300/2387] eta: 0:03:32 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2082 (0.3418) acc1: 93.7500 (86.1783) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1282 data: 0.0361 max mem: 3536 +Test: [1310/2387] eta: 0:03:30 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2177 (0.3410) acc1: 93.2292 (86.2208) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.1639 data: 0.0688 max mem: 3536 +Test: [1320/2387] eta: 0:03:28 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2296 (0.3407) acc1: 89.3229 (86.2346) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.2060 data: 0.1123 max mem: 3536 +Test: [1330/2387] eta: 0:03:26 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.3360 (0.3415) acc1: 84.8958 (86.1688) acc5: 100.0000 (100.0000) acc1_neg: 73.698 acc5_neg: 100.000 time: 0.2049 data: 0.1116 max mem: 3536 +Test: [1340/2387] eta: 0:03:24 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.5132 (0.3433) acc1: 73.6979 (86.0410) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2088 data: 0.1117 max mem: 3536 +Test: [1350/2387] eta: 0:03:22 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.1073 (0.3414) acc1: 98.4375 (86.1403) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1699 data: 0.0711 max mem: 3537 +Test: [1360/2387] eta: 0:03:20 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.0842 (0.3395) acc1: 99.7396 (86.2417) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1372 data: 0.0378 max mem: 3537 +Test: [1370/2387] eta: 0:03:17 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.0791 (0.3376) acc1: 100.0000 (86.3413) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1527 data: 0.0572 max mem: 3537 +Test: [1380/2387] eta: 0:03:15 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.0768 (0.3357) acc1: 100.0000 (86.4395) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1737 data: 0.0828 max mem: 3537 +Test: [1390/2387] eta: 0:03:13 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.0786 (0.3339) acc1: 100.0000 (86.5355) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1886 data: 0.0971 max mem: 3537 +Test: [1400/2387] eta: 0:03:12 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.0796 (0.3321) acc1: 100.0000 (86.6280) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2037 data: 0.1125 max mem: 3537 +Test: [1410/2387] eta: 0:03:09 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.1059 (0.3324) acc1: 99.2188 (86.6034) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1858 data: 0.0944 max mem: 3537 +Test: [1420/2387] eta: 0:03:07 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.3652 (0.3328) acc1: 81.2500 (86.5604) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1702 data: 0.0757 max mem: 3537 +Test: [1430/2387] eta: 0:03:05 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.4289 (0.3336) acc1: 78.1250 (86.4927) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.1774 data: 0.0819 max mem: 3537 +Test: [1440/2387] eta: 0:03:04 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.4345 (0.3344) acc1: 77.6042 (86.4314) acc5: 100.0000 (100.0000) acc1_neg: 65.885 acc5_neg: 100.000 time: 0.2047 data: 0.1123 max mem: 3537 +Test: [1450/2387] eta: 0:03:02 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.4482 (0.3357) acc1: 75.2604 (86.3445) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.2049 data: 0.1125 max mem: 3537 +Test: [1460/2387] eta: 0:02:59 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2811 (0.3354) acc1: 82.5521 (86.3601) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1697 data: 0.0756 max mem: 3537 +Test: [1470/2387] eta: 0:02:57 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2470 (0.3347) acc1: 90.6250 (86.3969) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1699 data: 0.0772 max mem: 3537 +Test: [1480/2387] eta: 0:02:56 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2269 (0.3341) acc1: 92.4479 (86.4337) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1910 data: 0.0978 max mem: 3537 +Test: [1490/2387] eta: 0:02:54 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2322 (0.3335) acc1: 92.1875 (86.4742) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2105 data: 0.1163 max mem: 3537 +Test: [1500/2387] eta: 0:02:52 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2236 (0.3329) acc1: 93.2292 (86.5142) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1965 data: 0.1046 max mem: 3537 +Test: [1510/2387] eta: 0:02:50 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2236 (0.3321) acc1: 93.7500 (86.5735) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1806 data: 0.0904 max mem: 3537 +Test: [1520/2387] eta: 0:02:48 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2603 (0.3319) acc1: 93.4896 (86.6020) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1972 data: 0.1034 max mem: 3537 +Test: [1530/2387] eta: 0:02:46 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.3340 (0.3320) acc1: 89.0625 (86.6138) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.2006 data: 0.1064 max mem: 3537 +Test: [1540/2387] eta: 0:02:44 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.3265 (0.3318) acc1: 89.5833 (86.6469) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1898 data: 0.0960 max mem: 3537 +Test: [1550/2387] eta: 0:02:42 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2817 (0.3314) acc1: 92.7083 (86.6936) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1979 data: 0.1038 max mem: 3537 +Test: [1560/2387] eta: 0:02:40 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2777 (0.3312) acc1: 92.9688 (86.7259) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.2248 data: 0.1315 max mem: 3537 +Test: [1570/2387] eta: 0:02:38 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2754 (0.3309) acc1: 92.9688 (86.7680) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2090 data: 0.1150 max mem: 3537 +Test: [1580/2387] eta: 0:02:36 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.3143 (0.3314) acc1: 89.5833 (86.7385) acc5: 100.0000 (100.0000) acc1_neg: 63.802 acc5_neg: 100.000 time: 0.1701 data: 0.0776 max mem: 3538 +Test: [1590/2387] eta: 0:02:34 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.3422 (0.3316) acc1: 83.8542 (86.7168) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1962 data: 0.1048 max mem: 3538 +Test: [1600/2387] eta: 0:02:32 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.3400 (0.3317) acc1: 85.1562 (86.7098) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.1900 data: 0.0994 max mem: 3538 +Test: [1610/2387] eta: 0:02:30 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.3518 (0.3319) acc1: 83.8542 (86.6898) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.1800 data: 0.0886 max mem: 3538 +Test: [1620/2387] eta: 0:02:28 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.4080 (0.3326) acc1: 80.7292 (86.6386) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1848 data: 0.0930 max mem: 3538 +Test: [1630/2387] eta: 0:02:26 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.4917 (0.3339) acc1: 74.2188 (86.5460) acc5: 100.0000 (100.0000) acc1_neg: 60.677 acc5_neg: 100.000 time: 0.1709 data: 0.0790 max mem: 3538 +Test: [1640/2387] eta: 0:02:24 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.6041 (0.3362) acc1: 67.1875 (86.4014) acc5: 100.0000 (100.0000) acc1_neg: 73.958 acc5_neg: 100.000 time: 0.1884 data: 0.0961 max mem: 3538 +Test: [1650/2387] eta: 0:02:22 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.6788 (0.3383) acc1: 63.8021 (86.2650) acc5: 100.0000 (100.0000) acc1_neg: 63.802 acc5_neg: 100.000 time: 0.2000 data: 0.1084 max mem: 3538 +Test: [1660/2387] eta: 0:02:21 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.6460 (0.3403) acc1: 64.3229 (86.1369) acc5: 100.0000 (100.0000) acc1_neg: 71.615 acc5_neg: 100.000 time: 0.1999 data: 0.1050 max mem: 3538 +Test: [1670/2387] eta: 0:02:19 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.6450 (0.3421) acc1: 67.1875 (86.0198) acc5: 100.0000 (100.0000) acc1_neg: 73.698 acc5_neg: 100.000 time: 0.2122 data: 0.1135 max mem: 3538 +Test: [1680/2387] eta: 0:02:17 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.6276 (0.3438) acc1: 66.1458 (85.8980) acc5: 100.0000 (100.0000) acc1_neg: 57.552 acc5_neg: 100.000 time: 0.2057 data: 0.1095 max mem: 3538 +Test: [1690/2387] eta: 0:02:15 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.6119 (0.3456) acc1: 66.1458 (85.7812) acc5: 100.0000 (100.0000) acc1_neg: 69.531 acc5_neg: 100.000 time: 0.1795 data: 0.0822 max mem: 3538 +Test: [1700/2387] eta: 0:02:13 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.6119 (0.3467) acc1: 67.7083 (85.7157) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1848 data: 0.0872 max mem: 3538 +Test: [1710/2387] eta: 0:02:11 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2633 (0.3461) acc1: 90.8854 (85.7536) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1978 data: 0.1040 max mem: 3538 +Test: [1720/2387] eta: 0:02:09 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2362 (0.3455) acc1: 92.4479 (85.7928) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2044 data: 0.1115 max mem: 3538 +Test: [1730/2387] eta: 0:02:07 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2295 (0.3448) acc1: 93.7500 (85.8379) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2369 data: 0.1446 max mem: 3538 +Test: [1740/2387] eta: 0:02:05 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2318 (0.3442) acc1: 93.2292 (85.8786) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2317 data: 0.1397 max mem: 3538 +Test: [1750/2387] eta: 0:02:03 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2363 (0.3436) acc1: 92.4479 (85.9185) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1860 data: 0.0947 max mem: 3538 +Test: [1760/2387] eta: 0:02:02 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2517 (0.3432) acc1: 92.1875 (85.9419) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.2128 data: 0.1212 max mem: 3538 +Test: [1770/2387] eta: 0:02:00 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.3315 (0.3435) acc1: 85.9375 (85.9172) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.2150 data: 0.1238 max mem: 3538 +Test: [1780/2387] eta: 0:01:58 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.3640 (0.3436) acc1: 82.5521 (85.9099) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.1777 data: 0.0841 max mem: 3538 +Test: [1790/2387] eta: 0:01:56 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.3324 (0.3434) acc1: 86.7188 (85.9224) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1986 data: 0.1032 max mem: 3538 +Test: [1800/2387] eta: 0:01:54 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.3180 (0.3433) acc1: 87.5000 (85.9307) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.2072 data: 0.1143 max mem: 3538 +Test: [1810/2387] eta: 0:01:52 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.3077 (0.3430) acc1: 88.0208 (85.9510) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.2049 data: 0.1117 max mem: 3539 +Test: [1820/2387] eta: 0:01:50 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.3014 (0.3428) acc1: 88.5417 (85.9637) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.2330 data: 0.1391 max mem: 3539 +Test: [1830/2387] eta: 0:01:48 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.3114 (0.3424) acc1: 88.0208 (85.9864) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1853 data: 0.0933 max mem: 3539 +Test: [1840/2387] eta: 0:01:46 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.0941 (0.3410) acc1: 99.2188 (86.0587) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1617 data: 0.0700 max mem: 3539 +Test: [1850/2387] eta: 0:01:44 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.0910 (0.3397) acc1: 99.2188 (86.1301) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1995 data: 0.1039 max mem: 3539 +Test: [1860/2387] eta: 0:01:42 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.0922 (0.3384) acc1: 99.2188 (86.2010) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1962 data: 0.1014 max mem: 3539 +Test: [1870/2387] eta: 0:01:40 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.0922 (0.3371) acc1: 99.4792 (86.2718) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2150 data: 0.1239 max mem: 3539 +Test: [1880/2387] eta: 0:01:38 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.0912 (0.3358) acc1: 99.2188 (86.3412) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2090 data: 0.1171 max mem: 3539 +Test: [1890/2387] eta: 0:01:36 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.0918 (0.3345) acc1: 99.2188 (86.4097) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2018 data: 0.1094 max mem: 3539 +Test: [1900/2387] eta: 0:01:34 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.0923 (0.3332) acc1: 99.2188 (86.4768) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1852 data: 0.0937 max mem: 3539 +Test: [1910/2387] eta: 0:01:32 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.1197 (0.3323) acc1: 98.6979 (86.5325) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1640 data: 0.0710 max mem: 3539 +Test: [1920/2387] eta: 0:01:30 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2651 (0.3329) acc1: 89.0625 (86.4852) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.1812 data: 0.0876 max mem: 3539 +Test: [1930/2387] eta: 0:01:29 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2774 (0.3324) acc1: 88.2812 (86.5159) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1911 data: 0.0992 max mem: 3539 +Test: [1940/2387] eta: 0:01:26 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2567 (0.3322) acc1: 92.4479 (86.5379) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1851 data: 0.0928 max mem: 3539 +Test: [1950/2387] eta: 0:01:25 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2707 (0.3321) acc1: 90.1042 (86.5455) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1749 data: 0.0795 max mem: 3539 +Test: [1960/2387] eta: 0:01:23 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.3239 (0.3325) acc1: 82.0312 (86.4957) acc5: 100.0000 (100.0000) acc1_neg: 79.427 acc5_neg: 100.000 time: 0.1826 data: 0.0825 max mem: 3539 +Test: [1970/2387] eta: 0:01:21 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.4313 (0.3337) acc1: 73.4375 (86.4109) acc5: 100.0000 (100.0000) acc1_neg: 63.542 acc5_neg: 100.000 time: 0.1751 data: 0.0770 max mem: 3539 +Test: [1980/2387] eta: 0:01:19 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.4950 (0.3337) acc1: 75.0000 (86.3992) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1779 data: 0.0834 max mem: 3539 +Test: [1990/2387] eta: 0:01:17 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.0924 (0.3325) acc1: 99.4792 (86.4653) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1751 data: 0.0813 max mem: 3539 +Test: [2000/2387] eta: 0:01:15 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.0923 (0.3313) acc1: 99.4792 (86.5302) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1719 data: 0.0797 max mem: 3539 +Test: [2010/2387] eta: 0:01:13 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.0918 (0.3301) acc1: 99.4792 (86.5946) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1833 data: 0.0904 max mem: 3539 +Test: [2020/2387] eta: 0:01:11 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.0914 (0.3290) acc1: 99.4792 (86.6578) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1749 data: 0.0831 max mem: 3539 +Test: [2030/2387] eta: 0:01:09 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.0939 (0.3278) acc1: 99.4792 (86.7207) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1786 data: 0.0854 max mem: 3540 +Test: [2040/2387] eta: 0:01:07 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.0915 (0.3266) acc1: 99.4792 (86.7837) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1975 data: 0.1030 max mem: 3540 +Test: [2050/2387] eta: 0:01:05 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.0872 (0.3255) acc1: 99.7396 (86.8464) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1964 data: 0.1046 max mem: 3540 +Test: [2060/2387] eta: 0:01:03 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.0827 (0.3243) acc1: 99.4792 (86.9077) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1929 data: 0.1023 max mem: 3540 +Test: [2070/2387] eta: 0:01:01 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.0793 (0.3231) acc1: 99.7396 (86.9699) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1950 data: 0.1026 max mem: 3540 +Test: [2080/2387] eta: 0:00:59 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.0784 (0.3219) acc1: 99.7396 (87.0306) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1943 data: 0.0996 max mem: 3540 +Test: [2090/2387] eta: 0:00:57 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.0798 (0.3208) acc1: 99.7396 (87.0911) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1950 data: 0.1024 max mem: 3540 +Test: [2100/2387] eta: 0:00:55 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.0797 (0.3196) acc1: 99.7396 (87.1520) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1951 data: 0.1044 max mem: 3540 +Test: [2110/2387] eta: 0:00:53 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.0794 (0.3185) acc1: 99.7396 (87.2115) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2131 data: 0.1204 max mem: 3540 +Test: [2120/2387] eta: 0:00:51 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.0862 (0.3186) acc1: 99.2188 (87.2002) acc5: 100.0000 (100.0000) acc1_neg: 68.750 acc5_neg: 100.000 time: 0.2154 data: 0.1202 max mem: 3540 +Test: [2130/2387] eta: 0:00:49 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2893 (0.3183) acc1: 88.2812 (87.2167) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2052 data: 0.1096 max mem: 3540 +Test: [2140/2387] eta: 0:00:47 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2814 (0.3182) acc1: 89.5833 (87.2288) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1791 data: 0.0819 max mem: 3540 +Test: [2150/2387] eta: 0:00:45 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2940 (0.3183) acc1: 88.2812 (87.2254) acc5: 100.0000 (100.0000) acc1_neg: 80.729 acc5_neg: 100.000 time: 0.1657 data: 0.0708 max mem: 3540 +Test: [2160/2387] eta: 0:00:44 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2940 (0.3182) acc1: 87.7604 (87.2254) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1799 data: 0.0883 max mem: 3540 +Test: [2170/2387] eta: 0:00:42 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.3414 (0.3188) acc1: 83.0729 (87.1812) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.1799 data: 0.0869 max mem: 3540 +Test: [2180/2387] eta: 0:00:40 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.4983 (0.3210) acc1: 72.3958 (87.0244) acc5: 100.0000 (100.0000) acc1_neg: 28.646 acc5_neg: 100.000 time: 0.1952 data: 0.1027 max mem: 3540 +Test: [2190/2387] eta: 0:00:38 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.4983 (0.3215) acc1: 72.3958 (86.9901) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2037 data: 0.1110 max mem: 3540 +Test: [2200/2387] eta: 0:00:36 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2603 (0.3212) acc1: 90.6250 (87.0231) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1895 data: 0.0938 max mem: 3540 +Test: [2210/2387] eta: 0:00:34 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2543 (0.3209) acc1: 92.9688 (87.0495) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1863 data: 0.0918 max mem: 3540 +Test: [2220/2387] eta: 0:00:32 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2491 (0.3206) acc1: 92.9688 (87.0818) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1749 data: 0.0822 max mem: 3540 +Test: [2230/2387] eta: 0:00:30 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2487 (0.3202) acc1: 94.2708 (87.1154) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1710 data: 0.0777 max mem: 3540 +Test: [2240/2387] eta: 0:00:28 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2487 (0.3199) acc1: 94.7917 (87.1442) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1722 data: 0.0804 max mem: 3540 +Test: [2250/2387] eta: 0:00:26 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2410 (0.3197) acc1: 94.2708 (87.1697) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1644 data: 0.0725 max mem: 3540 +Test: [2260/2387] eta: 0:00:24 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2410 (0.3194) acc1: 92.9688 (87.1922) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1738 data: 0.0823 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2025 (0.3190) acc1: 95.3125 (87.2220) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1743 data: 0.0838 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.1984 (0.3186) acc1: 96.0938 (87.2523) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1752 data: 0.0844 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2011 (0.3181) acc1: 94.5312 (87.2874) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1736 data: 0.0804 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2299 (0.3178) acc1: 94.2708 (87.3033) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1690 data: 0.0756 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.2784 (0.3180) acc1: 88.5417 (87.2913) acc5: 100.0000 (100.0000) acc1_neg: 70.833 acc5_neg: 100.000 time: 0.1607 data: 0.0693 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.3884 (0.3187) acc1: 79.4271 (87.2373) acc5: 100.0000 (100.0000) acc1_neg: 69.531 acc5_neg: 100.000 time: 0.1392 data: 0.0485 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.4646 (0.3195) acc1: 75.5208 (87.1856) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.1294 data: 0.0397 max mem: 3541 +Test: [2340/2387] eta: 0:00:09 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.3923 (0.3197) acc1: 82.2917 (87.1734) acc5: 100.0000 (100.0000) acc1_neg: 79.427 acc5_neg: 100.000 time: 0.1289 data: 0.0392 max mem: 3541 +Test: [2350/2387] eta: 0:00:07 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.3739 (0.3202) acc1: 83.3333 (87.1409) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.1264 data: 0.0367 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.4367 (0.3211) acc1: 77.3438 (87.0833) acc5: 100.0000 (100.0000) acc1_neg: 73.698 acc5_neg: 100.000 time: 0.1280 data: 0.0384 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.5391 (0.3220) acc1: 70.8333 (87.0187) acc5: 100.0000 (100.0000) acc1_neg: 70.833 acc5_neg: 100.000 time: 0.1247 data: 0.0350 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.5466 (0.3233) acc1: 67.9688 (86.9350) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.1175 data: 0.0278 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 85.388 acc5_pos: 100.000 loss: 0.6633 (0.3251) acc1: 63.0208 (86.8283) acc5: 100.0000 (100.0000) acc1_neg: 18.605 acc5_neg: 100.000 time: 0.1128 data: 0.0247 max mem: 3541 +Test: Total time: 0:07:35 (0.1908 s / it) +* Overall Acc@1 86.828 Acc@5 100.000 + target == real Acc@1 83.610 Acc@5 100.000 + target == fake Acc@1 87.054 Acc@5 100.000 + loss 0.325 + PR-AUC 0.9944 ROC-AUC 0.9279 EER 14.23% (thr=0.5273) +Accuracy of the network on the 916267 test images: 86.8% +[rank0]:[W725 14:58:07.467633894 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 14:58:24.823000 65396 torch/distributed/run.py:766] +W0725 14:58:24.823000 65396 torch/distributed/run.py:766] ***************************************** +W0725 14:58:24.823000 65396 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 14:58:24.823000 65396 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W725 14:58:52.005513645 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 14:58:53.096677815 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 14:58:53.312786781 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 14:58:53.715850587 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 14:58:53.765612246 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 14:58:53.868998706 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 14:58:53.909164407 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 14:58:53.030125706 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed1_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed1_resampled_n116-040-168_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 0:59:36 acc1_pos: 84.896 acc5_pos: 100.000 loss: 0.4130 (0.4130) acc1: 84.8958 (84.8958) acc5: 100.0000 (100.0000) time: 6.9184 data: 4.8316 max mem: 3531 +Test: [ 10/517] eta: 0:06:02 acc1_pos: 67.969 acc5_pos: 100.000 loss: 0.4994 (0.5054) acc1: 80.7292 (78.1487) acc5: 100.0000 (100.0000) time: 0.7141 data: 0.4396 max mem: 3531 +Test: [ 20/517] eta: 0:03:29 acc1_pos: 78.906 acc5_pos: 100.000 loss: 0.4768 (0.4981) acc1: 79.9479 (78.5714) acc5: 100.0000 (100.0000) time: 0.0968 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:03:04 acc1_pos: 74.479 acc5_pos: 100.000 loss: 0.4491 (0.4893) acc1: 81.2500 (79.3683) acc5: 100.0000 (100.0000) time: 0.1959 data: 0.1005 max mem: 3531 +Test: [ 40/517] eta: 0:03:04 acc1_pos: 83.594 acc5_pos: 100.000 loss: 0.4288 (0.4786) acc1: 83.5938 (80.1766) acc5: 100.0000 (100.0000) time: 0.3499 data: 0.2562 max mem: 3531 +Test: [ 50/517] eta: 0:02:51 acc1_pos: 88.073 acc5_pos: 100.000 loss: 0.4618 (0.5174) acc1: 73.6979 (76.9557) acc5: 100.0000 (100.0000) acc1_neg: 28.125 acc5_neg: 100.000 time: 0.3489 data: 0.2530 max mem: 3531 +Test: [ 60/517] eta: 0:02:39 acc1_pos: 88.073 acc5_pos: 100.000 loss: 0.9363 (0.6112) acc1: 36.7188 (69.0745) acc5: 100.0000 (100.0000) acc1_neg: 22.656 acc5_neg: 100.000 time: 0.2699 data: 0.1738 max mem: 3531 +Test: [ 70/517] eta: 0:02:30 acc1_pos: 88.073 acc5_pos: 100.000 loss: 0.9658 (0.6538) acc1: 33.0729 (65.2545) acc5: 100.0000 (100.0000) acc1_neg: 41.406 acc5_neg: 100.000 time: 0.2599 data: 0.1618 max mem: 3531 +Test: [ 80/517] eta: 0:02:21 acc1_pos: 88.073 acc5_pos: 100.000 loss: 0.9218 (0.6829) acc1: 40.8854 (62.2942) acc5: 100.0000 (100.0000) acc1_neg: 52.344 acc5_neg: 100.000 time: 0.2499 data: 0.1511 max mem: 3531 +Test: [ 90/517] eta: 0:02:15 acc1_pos: 88.073 acc5_pos: 100.000 loss: 0.9841 (0.7176) acc1: 34.8958 (59.0917) acc5: 100.0000 (100.0000) acc1_neg: 26.823 acc5_neg: 100.000 time: 0.2499 data: 0.1498 max mem: 3531 +Test: [100/517] eta: 0:02:10 acc1_pos: 88.073 acc5_pos: 100.000 loss: 1.0107 (0.7437) acc1: 30.7292 (56.8662) acc5: 100.0000 (100.0000) acc1_neg: 50.521 acc5_neg: 100.000 time: 0.2649 data: 0.1660 max mem: 3531 +Test: [110/517] eta: 0:02:05 acc1_pos: 88.073 acc5_pos: 100.000 loss: 0.7452 (0.7391) acc1: 50.5208 (57.2072) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.2649 data: 0.1685 max mem: 3531 +Test: [120/517] eta: 0:02:00 acc1_pos: 88.073 acc5_pos: 100.000 loss: 0.7381 (0.7476) acc1: 53.3854 (56.1768) acc5: 100.0000 (100.0000) acc1_neg: 39.583 acc5_neg: 100.000 time: 0.2649 data: 0.1673 max mem: 3531 +Test: [130/517] eta: 0:01:56 acc1_pos: 88.073 acc5_pos: 100.000 loss: 0.8408 (0.7493) acc1: 46.3542 (55.9121) acc5: 100.0000 (100.0000) acc1_neg: 51.302 acc5_neg: 100.000 time: 0.2549 data: 0.1564 max mem: 3531 +Test: [140/517] eta: 0:01:51 acc1_pos: 88.073 acc5_pos: 100.000 loss: 0.8494 (0.7602) acc1: 43.7500 (54.5120) acc5: 100.0000 (100.0000) acc1_neg: 46.615 acc5_neg: 100.000 time: 0.2502 data: 0.1553 max mem: 3531 +Test: [150/517] eta: 0:01:48 acc1_pos: 88.073 acc5_pos: 100.000 loss: 0.9072 (0.7710) acc1: 33.0729 (53.2216) acc5: 100.0000 (100.0000) acc1_neg: 20.312 acc5_neg: 100.000 time: 0.2549 data: 0.1602 max mem: 3531 +Test: [160/517] eta: 0:01:43 acc1_pos: 88.073 acc5_pos: 100.000 loss: 0.7807 (0.7635) acc1: 47.3958 (53.5633) acc5: 100.0000 (100.0000) acc1_neg: 74.740 acc5_neg: 100.000 time: 0.2496 data: 0.1537 max mem: 3531 +Test: [170/517] eta: 0:01:39 acc1_pos: 88.073 acc5_pos: 100.000 loss: 0.5780 (0.7493) acc1: 64.3229 (54.5733) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.2299 data: 0.1352 max mem: 3531 +Test: [180/517] eta: 0:01:35 acc1_pos: 88.073 acc5_pos: 100.000 loss: 0.5645 (0.7468) acc1: 65.8854 (54.8702) acc5: 100.0000 (100.0000) acc1_neg: 40.365 acc5_neg: 100.000 time: 0.2249 data: 0.1272 max mem: 3531 +Test: [190/517] eta: 0:01:33 acc1_pos: 88.073 acc5_pos: 100.000 loss: 0.7178 (0.7443) acc1: 55.7292 (55.0570) acc5: 100.0000 (100.0000) acc1_neg: 49.740 acc5_neg: 100.000 time: 0.2649 data: 0.1685 max mem: 3531 +Test: [200/517] eta: 0:01:29 acc1_pos: 88.073 acc5_pos: 100.000 loss: 0.7032 (0.7448) acc1: 54.4271 (54.8793) acc5: 100.0000 (100.0000) acc1_neg: 35.938 acc5_neg: 100.000 time: 0.2599 data: 0.1629 max mem: 3531 +Test: [210/517] eta: 0:01:26 acc1_pos: 88.073 acc5_pos: 100.000 loss: 0.8710 (0.7549) acc1: 43.2292 (53.8556) acc5: 100.0000 (100.0000) acc1_neg: 35.677 acc5_neg: 100.000 time: 0.2453 data: 0.1489 max mem: 3532 +Test: [220/517] eta: 0:01:22 acc1_pos: 88.073 acc5_pos: 100.000 loss: 0.9229 (0.7618) acc1: 32.0312 (53.0814) acc5: 100.0000 (100.0000) acc1_neg: 52.083 acc5_neg: 100.000 time: 0.2549 data: 0.1604 max mem: 3532 +Test: [230/517] eta: 0:01:19 acc1_pos: 88.073 acc5_pos: 100.000 loss: 0.9733 (0.7741) acc1: 31.5104 (52.0563) acc5: 100.0000 (100.0000) acc1_neg: 36.719 acc5_neg: 100.000 time: 0.2356 data: 0.1397 max mem: 3532 +Test: [240/517] eta: 0:01:16 acc1_pos: 88.073 acc5_pos: 100.000 loss: 0.8591 (0.7749) acc1: 37.7604 (51.8683) acc5: 100.0000 (100.0000) acc1_neg: 43.229 acc5_neg: 100.000 time: 0.2599 data: 0.1660 max mem: 3532 +Test: [250/517] eta: 0:01:13 acc1_pos: 88.073 acc5_pos: 100.000 loss: 0.8325 (0.7784) acc1: 43.2292 (51.3239) acc5: 100.0000 (100.0000) acc1_neg: 30.990 acc5_neg: 100.000 time: 0.2588 data: 0.1658 max mem: 3532 +Test: [260/517] eta: 0:01:10 acc1_pos: 88.073 acc5_pos: 100.000 loss: 0.8383 (0.7824) acc1: 36.9792 (50.6815) acc5: 100.0000 (100.0000) acc1_neg: 25.260 acc5_neg: 100.000 time: 0.2349 data: 0.1418 max mem: 3532 +Test: [270/517] eta: 0:01:07 acc1_pos: 88.073 acc5_pos: 100.000 loss: 0.8902 (0.7872) acc1: 33.3333 (50.1682) acc5: 100.0000 (100.0000) acc1_neg: 44.531 acc5_neg: 100.000 time: 0.2449 data: 0.1494 max mem: 3532 +Test: [280/517] eta: 0:01:04 acc1_pos: 88.073 acc5_pos: 100.000 loss: 0.9058 (0.7912) acc1: 34.6354 (49.6413) acc5: 100.0000 (100.0000) acc1_neg: 28.125 acc5_neg: 100.000 time: 0.2450 data: 0.1499 max mem: 3532 +Test: [290/517] eta: 0:01:01 acc1_pos: 88.073 acc5_pos: 100.000 loss: 0.8410 (0.7921) acc1: 39.3229 (49.4881) acc5: 100.0000 (100.0000) acc1_neg: 45.052 acc5_neg: 100.000 time: 0.2406 data: 0.1495 max mem: 3532 +Test: [300/517] eta: 0:00:58 acc1_pos: 88.073 acc5_pos: 100.000 loss: 0.8074 (0.7922) acc1: 45.8333 (49.4048) acc5: 100.0000 (100.0000) acc1_neg: 33.594 acc5_neg: 100.000 time: 0.2448 data: 0.1528 max mem: 3532 +Test: [310/517] eta: 0:00:56 acc1_pos: 88.073 acc5_pos: 100.000 loss: 0.8007 (0.7942) acc1: 46.3542 (49.1409) acc5: 100.0000 (100.0000) acc1_neg: 50.000 acc5_neg: 100.000 time: 0.2915 data: 0.1964 max mem: 3532 +Test: [320/517] eta: 0:00:53 acc1_pos: 88.073 acc5_pos: 100.000 loss: 0.7804 (0.7936) acc1: 49.2188 (49.1709) acc5: 100.0000 (100.0000) acc1_neg: 61.719 acc5_neg: 100.000 time: 0.2999 data: 0.2026 max mem: 3532 +Test: [330/517] eta: 0:00:50 acc1_pos: 88.073 acc5_pos: 100.000 loss: 0.7112 (0.7905) acc1: 54.4271 (49.4296) acc5: 100.0000 (100.0000) acc1_neg: 58.854 acc5_neg: 100.000 time: 0.2545 data: 0.1580 max mem: 3532 +Test: [340/517] eta: 0:00:47 acc1_pos: 88.073 acc5_pos: 100.000 loss: 0.7004 (0.7942) acc1: 55.4688 (49.2317) acc5: 100.0000 (100.0000) acc1_neg: 15.625 acc5_neg: 100.000 time: 0.2400 data: 0.1477 max mem: 3532 +Test: [350/517] eta: 0:00:45 acc1_pos: 88.073 acc5_pos: 100.000 loss: 1.1781 (0.8068) acc1: 19.5312 (48.3151) acc5: 100.0000 (100.0000) acc1_neg: 8.594 acc5_neg: 100.000 time: 0.2432 data: 0.1500 max mem: 3532 +Test: [360/517] eta: 0:00:42 acc1_pos: 88.073 acc5_pos: 100.000 loss: 1.1594 (0.8137) acc1: 19.5312 (47.7508) acc5: 100.0000 (100.0000) acc1_neg: 42.969 acc5_neg: 100.000 time: 0.2498 data: 0.1540 max mem: 3532 +Test: [370/517] eta: 0:00:39 acc1_pos: 88.073 acc5_pos: 100.000 loss: 0.8832 (0.8132) acc1: 41.4062 (47.7784) acc5: 100.0000 (100.0000) acc1_neg: 53.385 acc5_neg: 100.000 time: 0.2460 data: 0.1499 max mem: 3532 +Test: [380/517] eta: 0:00:36 acc1_pos: 88.073 acc5_pos: 100.000 loss: 0.7327 (0.8098) acc1: 53.3854 (48.0705) acc5: 100.0000 (100.0000) acc1_neg: 35.938 acc5_neg: 100.000 time: 0.2711 data: 0.1773 max mem: 3532 +Test: [390/517] eta: 0:00:34 acc1_pos: 88.073 acc5_pos: 100.000 loss: 0.7501 (0.8127) acc1: 51.0417 (47.8254) acc5: 100.0000 (100.0000) acc1_neg: 54.688 acc5_neg: 100.000 time: 0.2638 data: 0.1698 max mem: 3532 +Test: [400/517] eta: 0:00:31 acc1_pos: 88.073 acc5_pos: 100.000 loss: 0.9016 (0.8150) acc1: 39.8438 (47.7296) acc5: 100.0000 (100.0000) acc1_neg: 26.823 acc5_neg: 100.000 time: 0.2347 data: 0.1364 max mem: 3532 +Test: [410/517] eta: 0:00:28 acc1_pos: 88.073 acc5_pos: 100.000 loss: 0.9560 (0.8223) acc1: 35.6771 (47.1823) acc5: 100.0000 (100.0000) acc1_neg: 12.240 acc5_neg: 100.000 time: 0.2412 data: 0.1457 max mem: 3532 +Test: [420/517] eta: 0:00:25 acc1_pos: 88.073 acc5_pos: 100.000 loss: 1.0378 (0.8262) acc1: 23.6979 (46.8453) acc5: 100.0000 (100.0000) acc1_neg: 62.240 acc5_neg: 100.000 time: 0.2440 data: 0.1526 max mem: 3532 +Test: [430/517] eta: 0:00:23 acc1_pos: 88.073 acc5_pos: 100.000 loss: 0.9175 (0.8268) acc1: 35.4167 (46.7826) acc5: 100.0000 (100.0000) acc1_neg: 14.062 acc5_neg: 100.000 time: 0.2436 data: 0.1504 max mem: 3532 +Test: [440/517] eta: 0:00:20 acc1_pos: 88.073 acc5_pos: 100.000 loss: 0.9224 (0.8335) acc1: 35.1562 (46.2957) acc5: 100.0000 (100.0000) acc1_neg: 30.729 acc5_neg: 100.000 time: 0.2469 data: 0.1494 max mem: 3533 +Test: [450/517] eta: 0:00:17 acc1_pos: 88.073 acc5_pos: 100.000 loss: 0.9224 (0.8347) acc1: 35.1562 (46.1492) acc5: 100.0000 (100.0000) acc1_neg: 45.312 acc5_neg: 100.000 time: 0.2496 data: 0.1516 max mem: 3533 +Test: [460/517] eta: 0:00:15 acc1_pos: 88.073 acc5_pos: 100.000 loss: 0.9481 (0.8383) acc1: 32.5521 (45.7791) acc5: 100.0000 (100.0000) acc1_neg: 28.385 acc5_neg: 100.000 time: 0.2529 data: 0.1566 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 88.073 acc5_pos: 100.000 loss: 0.9861 (0.8408) acc1: 29.1667 (45.5646) acc5: 100.0000 (100.0000) acc1_neg: 28.646 acc5_neg: 100.000 time: 0.2411 data: 0.1465 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 88.073 acc5_pos: 100.000 loss: 0.8766 (0.8394) acc1: 41.1458 (45.5946) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.2315 data: 0.1380 max mem: 3533 +Test: [490/517] eta: 0:00:07 acc1_pos: 88.073 acc5_pos: 100.000 loss: 0.7081 (0.8365) acc1: 53.1250 (45.8275) acc5: 100.0000 (100.0000) acc1_neg: 40.104 acc5_neg: 100.000 time: 0.2288 data: 0.1361 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 88.073 acc5_pos: 100.000 loss: 0.9033 (0.8423) acc1: 34.3750 (45.3983) acc5: 100.0000 (100.0000) acc1_neg: 25.260 acc5_neg: 100.000 time: 0.2229 data: 0.1305 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 88.073 acc5_pos: 100.000 loss: 1.0600 (0.8434) acc1: 29.1667 (45.3212) acc5: 100.0000 (100.0000) acc1_neg: 45.052 acc5_neg: 100.000 time: 0.1903 data: 0.0989 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 88.073 acc5_pos: 100.000 loss: 0.8715 (0.8431) acc1: 35.4167 (45.3210) acc5: 100.0000 (100.0000) acc1_neg: 60.777 acc5_neg: 100.000 time: 0.1741 data: 0.0823 max mem: 3533 +Test: Total time: 0:02:13 (0.2583 s / it) +* Overall Acc@1 45.321 Acc@5 100.000 + target == real Acc@1 79.271 Acc@5 100.000 + target == fake Acc@1 41.799 Acc@5 100.000 + loss 0.843 + PR-AUC 0.9446 ROC-AUC 0.6715 EER 37.58% (thr=0.3672) +Accuracy of the network on the 198427 test images: 45.3% +[rank0]:[W725 15:01:17.938467926 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0725 15:01:36.962000 65955 torch/distributed/run.py:766] +W0725 15:01:36.962000 65955 torch/distributed/run.py:766] ***************************************** +W0725 15:01:36.962000 65955 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 15:01:36.962000 65955 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 15:02:04.153040659 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 15:02:04.569693905 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 15:02:04.957232651 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 15:02:04.996558134 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 15:02:05.143954377 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +[rank7]:[W725 15:02:05.175179001 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 15:02:05.182148624 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 15:02:05.204870152 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed1_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed1_resampled_n116-040-168_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:29:48 acc1_pos: 16.406 acc5_pos: 100.000 loss: 1.2530 (1.2530) acc1: 16.4062 (16.4062) acc5: 100.0000 (100.0000) time: 4.5985 data: 2.4501 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/389] eta: 0:03:09 acc1_pos: 3.646 acc5_pos: 100.000 loss: 1.3683 (1.3214) acc1: 16.4062 (14.9384) acc5: 100.0000 (100.0000) time: 0.5000 data: 0.2232 max mem: 3531 +Test: [ 20/389] eta: 0:01:52 acc1_pos: 2.865 acc5_pos: 100.000 loss: 1.4556 (1.6068) acc1: 2.8646 (8.6310) acc5: 100.0000 (100.0000) time: 0.0902 data: 0.0004 max mem: 3531 +Test: [ 30/389] eta: 0:01:26 acc1_pos: 5.729 acc5_pos: 100.000 loss: 1.5827 (1.5674) acc1: 1.8229 (8.8290) acc5: 100.0000 (100.0000) time: 0.1001 data: 0.0103 max mem: 3531 +Test: [ 40/389] eta: 0:01:22 acc1_pos: 0.781 acc5_pos: 100.000 loss: 1.7374 (1.5930) acc1: 0.7812 (8.0539) acc5: 100.0000 (100.0000) time: 0.1621 data: 0.0704 max mem: 3531 +Test: [ 50/389] eta: 0:01:17 acc1_pos: 13.021 acc5_pos: 100.000 loss: 1.5171 (1.5467) acc1: 2.3438 (7.6593) acc5: 100.0000 (100.0000) time: 0.2093 data: 0.1146 max mem: 3531 +Test: [ 60/389] eta: 0:01:14 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.3992 (1.5672) acc1: 3.3854 (7.2575) acc5: 100.0000 (100.0000) time: 0.2080 data: 0.1123 max mem: 3531 +Test: [ 70/389] eta: 0:01:09 acc1_pos: 16.927 acc5_pos: 100.000 loss: 1.4448 (1.5660) acc1: 1.5625 (7.9555) acc5: 100.0000 (100.0000) time: 0.1873 data: 0.0916 max mem: 3531 +Test: [ 80/389] eta: 0:01:06 acc1_pos: 28.385 acc5_pos: 100.000 loss: 1.7353 (1.5639) acc1: 1.5625 (9.3011) acc5: 100.0000 (100.0000) time: 0.1811 data: 0.0873 max mem: 3531 +Test: [ 90/389] eta: 0:01:03 acc1_pos: 9.635 acc5_pos: 100.000 loss: 1.2599 (1.5258) acc1: 9.8958 (10.4510) acc5: 100.0000 (100.0000) time: 0.1878 data: 0.0959 max mem: 3531 +Test: [100/389] eta: 0:00:59 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.2964 (1.5283) acc1: 10.1562 (10.2362) acc5: 100.0000 (100.0000) time: 0.1745 data: 0.0787 max mem: 3531 +Test: [110/389] eta: 0:00:57 acc1_pos: 54.427 acc5_pos: 100.000 loss: 1.4124 (1.5189) acc1: 9.3750 (12.0003) acc5: 100.0000 (100.0000) time: 0.1896 data: 0.0941 max mem: 3531 +Test: [120/389] eta: 0:00:55 acc1_pos: 14.323 acc5_pos: 100.000 loss: 1.1741 (1.4892) acc1: 22.6562 (13.5029) acc5: 100.0000 (100.0000) time: 0.2001 data: 0.1067 max mem: 3531 +Test: [130/389] eta: 0:00:52 acc1_pos: 2.454 acc5_pos: 100.000 loss: 0.7629 (1.4153) acc1: 48.6979 (18.2153) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1879 data: 0.0947 max mem: 3531 +Test: [140/389] eta: 0:00:50 acc1_pos: 2.454 acc5_pos: 100.000 loss: 0.1906 (1.3340) acc1: 97.9167 (23.3747) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.1785 data: 0.0879 max mem: 3531 +Test: [150/389] eta: 0:00:48 acc1_pos: 2.454 acc5_pos: 100.000 loss: 0.1906 (1.2606) acc1: 98.9583 (28.3302) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1818 data: 0.0909 max mem: 3531 +Test: [160/389] eta: 0:00:46 acc1_pos: 2.454 acc5_pos: 100.000 loss: 0.1846 (1.1933) acc1: 99.2188 (32.5181) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1945 data: 0.1018 max mem: 3531 +Test: [170/389] eta: 0:00:43 acc1_pos: 2.454 acc5_pos: 100.000 loss: 0.0826 (1.1364) acc1: 100.0000 (36.0167) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1817 data: 0.0871 max mem: 3531 +Test: [180/389] eta: 0:00:41 acc1_pos: 2.454 acc5_pos: 100.000 loss: 0.1538 (1.0843) acc1: 99.7396 (39.3733) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.1609 data: 0.0660 max mem: 3531 +Test: [190/389] eta: 0:00:39 acc1_pos: 2.454 acc5_pos: 100.000 loss: 0.1653 (1.0431) acc1: 98.4375 (41.9298) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1748 data: 0.0810 max mem: 3531 +Test: [200/389] eta: 0:00:36 acc1_pos: 2.454 acc5_pos: 100.000 loss: 0.2252 (1.0025) acc1: 99.2188 (44.6828) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1841 data: 0.0902 max mem: 3531 +Test: [210/389] eta: 0:00:34 acc1_pos: 2.454 acc5_pos: 100.000 loss: 0.1893 (0.9631) acc1: 99.7396 (47.2971) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1732 data: 0.0796 max mem: 3532 +Test: [220/389] eta: 0:00:32 acc1_pos: 2.454 acc5_pos: 100.000 loss: 0.1094 (0.9250) acc1: 100.0000 (49.6760) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0832 max mem: 3532 +Test: [230/389] eta: 0:00:30 acc1_pos: 2.454 acc5_pos: 100.000 loss: 0.1429 (0.8926) acc1: 100.0000 (51.7970) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1857 data: 0.0943 max mem: 3532 +Test: [240/389] eta: 0:00:28 acc1_pos: 2.454 acc5_pos: 100.000 loss: 0.1505 (0.8626) acc1: 99.4792 (53.7388) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1917 data: 0.0975 max mem: 3532 +Test: [250/389] eta: 0:00:26 acc1_pos: 2.454 acc5_pos: 100.000 loss: 0.1381 (0.8349) acc1: 99.4792 (55.5580) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1740 data: 0.0798 max mem: 3532 +Test: [260/389] eta: 0:00:24 acc1_pos: 2.454 acc5_pos: 100.000 loss: 0.1173 (0.8091) acc1: 100.0000 (57.2118) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1614 data: 0.0698 max mem: 3532 +Test: [270/389] eta: 0:00:22 acc1_pos: 2.454 acc5_pos: 100.000 loss: 0.1659 (0.7864) acc1: 100.0000 (58.7552) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1655 data: 0.0731 max mem: 3532 +Test: [280/389] eta: 0:00:20 acc1_pos: 2.454 acc5_pos: 100.000 loss: 0.1702 (0.7660) acc1: 100.0000 (60.0793) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1665 data: 0.0737 max mem: 3532 +Test: [290/389] eta: 0:00:18 acc1_pos: 2.454 acc5_pos: 100.000 loss: 0.1480 (0.7463) acc1: 100.0000 (61.3635) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1717 data: 0.0809 max mem: 3532 +Test: [300/389] eta: 0:00:16 acc1_pos: 2.454 acc5_pos: 100.000 loss: 0.1485 (0.7285) acc1: 99.4792 (62.5502) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1920 data: 0.0997 max mem: 3532 +Test: [310/389] eta: 0:00:14 acc1_pos: 2.454 acc5_pos: 100.000 loss: 0.2330 (0.7165) acc1: 96.3542 (63.3198) acc5: 100.0000 (100.0000) acc1_neg: 52.865 acc5_neg: 100.000 time: 0.1681 data: 0.0756 max mem: 3532 +Test: [320/389] eta: 0:00:12 acc1_pos: 2.454 acc5_pos: 100.000 loss: 0.2543 (0.7019) acc1: 95.0521 (64.3367) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1455 data: 0.0533 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 2.454 acc5_pos: 100.000 loss: 0.1960 (0.6870) acc1: 97.1354 (65.3567) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1767 data: 0.0839 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 2.454 acc5_pos: 100.000 loss: 0.1654 (0.6698) acc1: 99.4792 (66.3459) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1749 data: 0.0822 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 2.454 acc5_pos: 100.000 loss: 0.1042 (0.6628) acc1: 99.2188 (66.6496) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.1796 data: 0.0875 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 2.454 acc5_pos: 100.000 loss: 0.2377 (0.6533) acc1: 97.3958 (67.3722) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1829 data: 0.0907 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 2.454 acc5_pos: 100.000 loss: 0.1763 (0.6393) acc1: 99.4792 (68.2453) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1624 data: 0.0707 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 2.454 acc5_pos: 100.000 loss: 0.1564 (0.6280) acc1: 99.7396 (68.9687) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1540 data: 0.0633 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 2.454 acc5_pos: 100.000 loss: 0.1743 (0.6197) acc1: 98.9583 (69.4881) acc5: 100.0000 (100.0000) acc1_neg: 92.350 acc5_neg: 100.000 time: 0.1442 data: 0.0501 max mem: 3532 +Test: Total time: 0:01:11 (0.1836 s / it) +* Overall Acc@1 69.488 Acc@5 100.000 + target == real Acc@1 13.314 Acc@5 100.000 + target == fake Acc@1 95.599 Acc@5 100.000 + loss 0.620 + PR-AUC 0.8291 ROC-AUC 0.6980 EER 35.31% (thr=0.8340) +Accuracy of the network on the 149358 test images: 69.5% +[rank0]:[W725 15:03:29.848621187 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 15:03:48.649000 66511 torch/distributed/run.py:766] +W0725 15:03:48.649000 66511 torch/distributed/run.py:766] ***************************************** +W0725 15:03:48.649000 66511 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 15:03:48.649000 66511 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W725 15:04:18.488440088 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 15:04:18.510681579 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 15:04:18.549505864 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 15:04:18.766392974 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 15:04:18.933504451 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 15:04:18.976585172 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 15:04:19.053789724 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 15:04:19.075611866 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed1_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed1_resampled_n116-040-168_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:37:23 acc1_pos: 93.490 acc5_pos: 100.000 loss: 0.2277 (0.2277) acc1: 93.4896 (93.4896) acc5: 100.0000 (100.0000) time: 6.7358 data: 4.9460 max mem: 3531 +Test: [ 10/333] eta: 0:03:45 acc1_pos: 94.792 acc5_pos: 100.000 loss: 0.2381 (0.2321) acc1: 92.1875 (92.9214) acc5: 100.0000 (100.0000) time: 0.6997 data: 0.4501 max mem: 3531 +Test: [ 20/333] eta: 0:02:09 acc1_pos: 95.833 acc5_pos: 100.000 loss: 0.2393 (0.2382) acc1: 92.9688 (92.8571) acc5: 100.0000 (100.0000) time: 0.0980 data: 0.0004 max mem: 3531 +Test: [ 30/333] eta: 0:01:51 acc1_pos: 81.510 acc5_pos: 100.000 loss: 0.3037 (0.3108) acc1: 87.5000 (88.3485) acc5: 100.0000 (100.0000) time: 0.1850 data: 0.0840 max mem: 3531 +Test: [ 40/333] eta: 0:01:38 acc1_pos: 76.823 acc5_pos: 100.000 loss: 0.4588 (0.3581) acc1: 79.6875 (85.4103) acc5: 100.0000 (100.0000) time: 0.2554 data: 0.1555 max mem: 3531 +Test: [ 50/333] eta: 0:01:30 acc1_pos: 77.344 acc5_pos: 100.000 loss: 0.4530 (0.3752) acc1: 80.7292 (84.6456) acc5: 100.0000 (100.0000) time: 0.2448 data: 0.1477 max mem: 3531 +Test: [ 60/333] eta: 0:01:23 acc1_pos: 77.865 acc5_pos: 100.000 loss: 0.4530 (0.3918) acc1: 78.3854 (83.8841) acc5: 100.0000 (100.0000) time: 0.2412 data: 0.1467 max mem: 3531 +Test: [ 70/333] eta: 0:01:15 acc1_pos: 82.552 acc5_pos: 100.000 loss: 0.4615 (0.4050) acc1: 77.8646 (83.2783) acc5: 100.0000 (100.0000) time: 0.2085 data: 0.1162 max mem: 3531 +Test: [ 80/333] eta: 0:01:08 acc1_pos: 85.417 acc5_pos: 100.000 loss: 0.4664 (0.4140) acc1: 80.4688 (82.7514) acc5: 100.0000 (100.0000) time: 0.1731 data: 0.0788 max mem: 3531 +Test: [ 90/333] eta: 0:01:04 acc1_pos: 83.854 acc5_pos: 100.000 loss: 0.4470 (0.4145) acc1: 82.2917 (82.7524) acc5: 100.0000 (100.0000) time: 0.1814 data: 0.0834 max mem: 3531 +Test: [100/333] eta: 0:00:59 acc1_pos: 90.365 acc5_pos: 100.000 loss: 0.4091 (0.4143) acc1: 83.5938 (82.8512) acc5: 100.0000 (100.0000) time: 0.1945 data: 0.0968 max mem: 3531 +Test: [110/333] eta: 0:00:56 acc1_pos: 86.458 acc5_pos: 100.000 loss: 0.3940 (0.4141) acc1: 85.9375 (83.0659) acc5: 100.0000 (100.0000) time: 0.2007 data: 0.1052 max mem: 3531 +Test: [120/333] eta: 0:00:52 acc1_pos: 88.802 acc5_pos: 100.000 loss: 0.4208 (0.4157) acc1: 83.0729 (82.9180) acc5: 100.0000 (100.0000) time: 0.1848 data: 0.0892 max mem: 3531 +Test: [130/333] eta: 0:00:48 acc1_pos: 91.667 acc5_pos: 100.000 loss: 0.4022 (0.4136) acc1: 83.0729 (83.0968) acc5: 100.0000 (100.0000) time: 0.1686 data: 0.0745 max mem: 3531 +Test: [140/333] eta: 0:00:46 acc1_pos: 86.198 acc5_pos: 100.000 loss: 0.3807 (0.4112) acc1: 86.1979 (83.3149) acc5: 100.0000 (100.0000) time: 0.2317 data: 0.1354 max mem: 3531 +Test: [150/333] eta: 0:00:43 acc1_pos: 83.073 acc5_pos: 100.000 loss: 0.3891 (0.4110) acc1: 85.4167 (83.3437) acc5: 100.0000 (100.0000) time: 0.2366 data: 0.1391 max mem: 3531 +Test: [160/333] eta: 0:00:40 acc1_pos: 82.566 acc5_pos: 100.000 loss: 0.4089 (0.4357) acc1: 82.2917 (81.6819) acc5: 100.0000 (100.0000) acc1_neg: 17.188 acc5_neg: 100.000 time: 0.1681 data: 0.0753 max mem: 3531 +Test: [170/333] eta: 0:00:38 acc1_pos: 82.566 acc5_pos: 100.000 loss: 1.1688 (0.4809) acc1: 26.8229 (78.3123) acc5: 100.0000 (100.0000) acc1_neg: 26.823 acc5_neg: 100.000 time: 0.1835 data: 0.0904 max mem: 3531 +Test: [180/333] eta: 0:00:35 acc1_pos: 82.566 acc5_pos: 100.000 loss: 1.1449 (0.5153) acc1: 26.8229 (75.6086) acc5: 100.0000 (100.0000) acc1_neg: 23.177 acc5_neg: 100.000 time: 0.2205 data: 0.1272 max mem: 3531 +Test: [190/333] eta: 0:00:32 acc1_pos: 82.566 acc5_pos: 100.000 loss: 1.0944 (0.5455) acc1: 30.4688 (73.3407) acc5: 100.0000 (100.0000) acc1_neg: 22.656 acc5_neg: 100.000 time: 0.2096 data: 0.1174 max mem: 3531 +Test: [200/333] eta: 0:00:30 acc1_pos: 82.566 acc5_pos: 100.000 loss: 1.1394 (0.5734) acc1: 29.4271 (71.1754) acc5: 100.0000 (100.0000) acc1_neg: 29.427 acc5_neg: 100.000 time: 0.1953 data: 0.1031 max mem: 3531 +Test: [210/333] eta: 0:00:27 acc1_pos: 82.566 acc5_pos: 100.000 loss: 1.1708 (0.6047) acc1: 27.3438 (68.9833) acc5: 100.0000 (100.0000) acc1_neg: 22.135 acc5_neg: 100.000 time: 0.1971 data: 0.1034 max mem: 3532 +Test: [220/333] eta: 0:00:25 acc1_pos: 82.566 acc5_pos: 100.000 loss: 1.2026 (0.6339) acc1: 23.6979 (66.8917) acc5: 100.0000 (100.0000) acc1_neg: 36.198 acc5_neg: 100.000 time: 0.2239 data: 0.1280 max mem: 3532 +Test: [230/333] eta: 0:00:23 acc1_pos: 82.566 acc5_pos: 100.000 loss: 0.9420 (0.6446) acc1: 36.1979 (65.9204) acc5: 100.0000 (100.0000) acc1_neg: 45.312 acc5_neg: 100.000 time: 0.2278 data: 0.1333 max mem: 3532 +Test: [240/333] eta: 0:00:20 acc1_pos: 82.566 acc5_pos: 100.000 loss: 0.9183 (0.6608) acc1: 38.8021 (64.5055) acc5: 100.0000 (100.0000) acc1_neg: 28.125 acc5_neg: 100.000 time: 0.1949 data: 0.1015 max mem: 3532 +Test: [250/333] eta: 0:00:18 acc1_pos: 82.566 acc5_pos: 100.000 loss: 1.0607 (0.6767) acc1: 30.2083 (63.1931) acc5: 100.0000 (100.0000) acc1_neg: 30.208 acc5_neg: 100.000 time: 0.1962 data: 0.1041 max mem: 3532 +Test: [260/333] eta: 0:00:16 acc1_pos: 82.566 acc5_pos: 100.000 loss: 1.0853 (0.6936) acc1: 32.5521 (61.9602) acc5: 100.0000 (100.0000) acc1_neg: 40.104 acc5_neg: 100.000 time: 0.2199 data: 0.1285 max mem: 3532 +Test: [270/333] eta: 0:00:14 acc1_pos: 82.566 acc5_pos: 100.000 loss: 1.0832 (0.7060) acc1: 33.3333 (60.9721) acc5: 100.0000 (100.0000) acc1_neg: 35.938 acc5_neg: 100.000 time: 0.1935 data: 0.0973 max mem: 3532 +Test: [280/333] eta: 0:00:11 acc1_pos: 82.566 acc5_pos: 100.000 loss: 1.0534 (0.7232) acc1: 30.2083 (59.7540) acc5: 100.0000 (100.0000) acc1_neg: 29.688 acc5_neg: 100.000 time: 0.1799 data: 0.0821 max mem: 3532 +Test: [290/333] eta: 0:00:09 acc1_pos: 82.566 acc5_pos: 100.000 loss: 1.2991 (0.7426) acc1: 23.4375 (58.5069) acc5: 100.0000 (100.0000) acc1_neg: 28.906 acc5_neg: 100.000 time: 0.2020 data: 0.1066 max mem: 3532 +Test: [300/333] eta: 0:00:07 acc1_pos: 82.566 acc5_pos: 100.000 loss: 1.2165 (0.7572) acc1: 23.6979 (57.4102) acc5: 100.0000 (100.0000) acc1_neg: 25.781 acc5_neg: 100.000 time: 0.1732 data: 0.0802 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 82.566 acc5_pos: 100.000 loss: 1.0576 (0.7650) acc1: 29.6875 (56.7147) acc5: 100.0000 (100.0000) acc1_neg: 30.208 acc5_neg: 100.000 time: 0.1277 data: 0.0374 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 82.566 acc5_pos: 100.000 loss: 0.9726 (0.7731) acc1: 35.6771 (55.9985) acc5: 100.0000 (100.0000) acc1_neg: 23.177 acc5_neg: 100.000 time: 0.1177 data: 0.0276 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 82.566 acc5_pos: 100.000 loss: 1.3498 (0.7919) acc1: 23.4375 (54.9637) acc5: 100.0000 (100.0000) acc1_neg: 22.656 acc5_neg: 100.000 time: 0.1220 data: 0.0321 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 82.566 acc5_pos: 100.000 loss: 1.3562 (0.7952) acc1: 23.4375 (54.7828) acc5: 100.0000 (100.0000) acc1_neg: 23.944 acc5_neg: 100.000 time: 0.1287 data: 0.0265 max mem: 3532 +Test: Total time: 0:01:09 (0.2101 s / it) +* Overall Acc@1 54.783 Acc@5 100.000 + target == real Acc@1 83.398 Acc@5 100.000 + target == fake Acc@1 29.310 Acc@5 100.000 + loss 0.795 + PR-AUC 0.6338 ROC-AUC 0.6238 EER 41.48% (thr=0.2949) +Accuracy of the network on the 127843 test images: 54.8% +[rank0]:[W725 15:05:36.656558431 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 15:05:55.417000 67069 torch/distributed/run.py:766] +W0725 15:05:55.417000 67069 torch/distributed/run.py:766] ***************************************** +W0725 15:05:55.417000 67069 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 15:05:55.417000 67069 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 15:06:24.589137628 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W725 15:06:24.034699774 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W725 15:06:24.035069053 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 15:06:25.068109866 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 15:06:25.143043833 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 15:06:25.377467050 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 15:06:25.410074222 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 15:06:25.481895490 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed1_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed1_resampled_n116-040-168_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:22:50 acc1_pos: 97.135 acc5_pos: 100.000 loss: 0.1424 (0.1424) acc1: 97.1354 (97.1354) acc5: 100.0000 (100.0000) time: 5.8309 data: 3.4668 max mem: 3531 +Test: [ 10/235] eta: 0:02:17 acc1_pos: 96.354 acc5_pos: 100.000 loss: 0.1519 (0.1483) acc1: 97.3958 (97.4669) acc5: 100.0000 (100.0000) time: 0.6120 data: 0.3155 max mem: 3531 +Test: [ 20/235] eta: 0:01:18 acc1_pos: 96.354 acc5_pos: 100.000 loss: 0.1452 (0.1466) acc1: 97.3958 (97.4454) acc5: 100.0000 (100.0000) time: 0.0901 data: 0.0004 max mem: 3531 +Test: [ 30/235] eta: 0:01:07 acc1_pos: 97.917 acc5_pos: 100.000 loss: 0.1439 (0.1451) acc1: 97.6562 (97.5302) acc5: 100.0000 (100.0000) time: 0.1721 data: 0.0761 max mem: 3531 +Test: [ 40/235] eta: 0:01:06 acc1_pos: 97.135 acc5_pos: 100.000 loss: 0.1439 (0.1449) acc1: 97.6562 (97.5800) acc5: 100.0000 (100.0000) time: 0.3149 data: 0.2184 max mem: 3531 +Test: [ 50/235] eta: 0:01:02 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.1484 (0.1466) acc1: 97.3958 (97.4929) acc5: 100.0000 (100.0000) time: 0.3459 data: 0.2547 max mem: 3531 +Test: [ 60/235] eta: 0:00:53 acc1_pos: 97.656 acc5_pos: 100.000 loss: 0.1552 (0.1482) acc1: 97.3958 (97.4343) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1375 max mem: 3531 +Test: [ 70/235] eta: 0:00:48 acc1_pos: 98.462 acc5_pos: 100.000 loss: 0.1693 (0.3337) acc1: 95.8333 (87.5293) acc5: 100.0000 (100.0000) acc1_neg: 17.708 acc5_neg: 100.000 time: 0.1876 data: 0.0932 max mem: 3531 +Test: [ 80/235] eta: 0:00:44 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.6387 (0.4956) acc1: 17.7083 (78.9320) acc5: 100.0000 (100.0000) acc1_neg: 16.667 acc5_neg: 100.000 time: 0.2316 data: 0.1376 max mem: 3531 +Test: [ 90/235] eta: 0:00:40 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.6470 (0.6225) acc1: 17.1875 (72.0782) acc5: 100.0000 (100.0000) acc1_neg: 14.062 acc5_neg: 100.000 time: 0.2162 data: 0.1251 max mem: 3531 +Test: [100/235] eta: 0:00:35 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.6497 (0.7224) acc1: 17.1875 (66.7518) acc5: 100.0000 (100.0000) acc1_neg: 17.188 acc5_neg: 100.000 time: 0.1782 data: 0.0851 max mem: 3531 +Test: [110/235] eta: 0:00:32 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.6189 (0.8029) acc1: 18.2292 (62.4155) acc5: 100.0000 (100.0000) acc1_neg: 19.792 acc5_neg: 100.000 time: 0.1787 data: 0.0844 max mem: 3531 +Test: [120/235] eta: 0:00:28 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.6666 (0.8794) acc1: 14.5833 (58.2365) acc5: 100.0000 (100.0000) acc1_neg: 11.198 acc5_neg: 100.000 time: 0.1873 data: 0.0955 max mem: 3531 +Test: [130/235] eta: 0:00:26 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.7278 (0.9428) acc1: 11.1979 (54.6895) acc5: 100.0000 (100.0000) acc1_neg: 10.417 acc5_neg: 100.000 time: 0.2159 data: 0.1214 max mem: 3531 +Test: [140/235] eta: 0:00:23 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.7242 (0.9998) acc1: 11.1979 (51.6105) acc5: 100.0000 (100.0000) acc1_neg: 9.115 acc5_neg: 100.000 time: 0.2296 data: 0.1348 max mem: 3531 +Test: [150/235] eta: 0:00:20 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.7212 (1.0470) acc1: 11.1979 (48.9532) acc5: 100.0000 (100.0000) acc1_neg: 12.240 acc5_neg: 100.000 time: 0.1548 data: 0.0604 max mem: 3531 +Test: [160/235] eta: 0:00:18 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.7072 (1.0882) acc1: 11.7188 (46.6243) acc5: 100.0000 (100.0000) acc1_neg: 11.719 acc5_neg: 100.000 time: 0.1910 data: 0.0965 max mem: 3531 +Test: [170/235] eta: 0:00:15 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.6781 (1.1067) acc1: 12.7604 (45.4526) acc5: 100.0000 (100.0000) acc1_neg: 26.302 acc5_neg: 100.000 time: 0.2480 data: 0.1532 max mem: 3531 +Test: [180/235] eta: 0:00:12 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.3659 (1.1203) acc1: 28.6458 (44.5571) acc5: 100.0000 (100.0000) acc1_neg: 28.385 acc5_neg: 100.000 time: 0.1766 data: 0.0791 max mem: 3531 +Test: [190/235] eta: 0:00:10 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.3403 (1.1324) acc1: 29.6875 (43.7582) acc5: 100.0000 (100.0000) acc1_neg: 31.771 acc5_neg: 100.000 time: 0.1511 data: 0.0569 max mem: 3531 +Test: [200/235] eta: 0:00:08 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.3457 (1.1444) acc1: 27.8646 (42.9584) acc5: 100.0000 (100.0000) acc1_neg: 26.823 acc5_neg: 100.000 time: 0.1856 data: 0.0941 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.3656 (1.1545) acc1: 27.6042 (42.2825) acc5: 100.0000 (100.0000) acc1_neg: 27.604 acc5_neg: 100.000 time: 0.1720 data: 0.0804 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.3537 (1.1638) acc1: 28.1250 (41.6596) acc5: 100.0000 (100.0000) acc1_neg: 28.125 acc5_neg: 100.000 time: 0.1371 data: 0.0460 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.3480 (1.1721) acc1: 28.9062 (41.1605) acc5: 100.0000 (100.0000) acc1_neg: 28.125 acc5_neg: 100.000 time: 0.1239 data: 0.0336 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.3414 (1.1744) acc1: 29.1667 (41.0307) acc5: 100.0000 (100.0000) acc1_neg: 39.130 acc5_neg: 100.000 time: 0.1134 data: 0.0263 max mem: 3532 +Test: Total time: 0:00:50 (0.2154 s / it) +* Overall Acc@1 41.031 Acc@5 100.000 + target == real Acc@1 97.415 Acc@5 100.000 + target == fake Acc@1 20.638 Acc@5 100.000 + loss 1.174 + PR-AUC 0.8973 ROC-AUC 0.7710 EER 29.99% (thr=0.1261) +Accuracy of the network on the 89879 test images: 41.0% +[rank0]:[W725 15:07:23.705474558 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 15:07:42.414000 67636 torch/distributed/run.py:766] +W0725 15:07:42.414000 67636 torch/distributed/run.py:766] ***************************************** +W0725 15:07:42.414000 67636 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 15:07:42.414000 67636 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 15:08:09.781552620 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 15:08:09.816413089 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 15:08:10.145197587 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 15:08:10.280039250 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 15:08:10.364762395 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 15:08:10.401078257 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 15:08:10.446075255 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W725 15:08:10.446139413 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed1_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed1_resampled_n116-040-168_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:30:37 acc1_pos: 92.188 acc5_pos: 100.000 loss: 0.2526 (0.2526) acc1: 92.1875 (92.1875) acc5: 100.0000 (100.0000) time: 6.8290 data: 5.2187 max mem: 3531 +Test: [ 10/269] eta: 0:03:03 acc1_pos: 95.573 acc5_pos: 100.000 loss: 0.2605 (0.2686) acc1: 95.8333 (95.4545) acc5: 100.0000 (100.0000) time: 0.7089 data: 0.4757 max mem: 3531 +Test: [ 20/269] eta: 0:01:44 acc1_pos: 97.917 acc5_pos: 100.000 loss: 0.2518 (0.2684) acc1: 96.8750 (95.5729) acc5: 100.0000 (100.0000) time: 0.0984 data: 0.0009 max mem: 3531 +Test: [ 30/269] eta: 0:01:28 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2281 (0.2538) acc1: 97.9167 (96.4634) acc5: 100.0000 (100.0000) time: 0.1857 data: 0.0908 max mem: 3531 +Test: [ 40/269] eta: 0:01:19 acc1_pos: 98.276 acc5_pos: 100.000 loss: 0.2651 (0.4733) acc1: 94.7917 (80.1512) acc5: 100.0000 (100.0000) acc1_neg: 4.948 acc5_neg: 100.000 time: 0.2754 data: 0.1839 max mem: 3531 +Test: [ 50/269] eta: 0:01:11 acc1_pos: 98.276 acc5_pos: 100.000 loss: 1.4376 (0.6843) acc1: 12.7604 (65.8854) acc5: 100.0000 (100.0000) acc1_neg: 2.344 acc5_neg: 100.000 time: 0.2591 data: 0.1657 max mem: 3531 +Test: [ 60/269] eta: 0:01:01 acc1_pos: 98.276 acc5_pos: 100.000 loss: 1.5746 (0.8281) acc1: 4.4271 (55.9341) acc5: 100.0000 (100.0000) acc1_neg: 7.552 acc5_neg: 100.000 time: 0.1821 data: 0.0877 max mem: 3531 +Test: [ 70/269] eta: 0:00:58 acc1_pos: 98.276 acc5_pos: 100.000 loss: 1.4427 (0.9035) acc1: 8.3333 (49.8460) acc5: 100.0000 (100.0000) acc1_neg: 10.938 acc5_neg: 100.000 time: 0.2116 data: 0.1185 max mem: 3531 +Test: [ 80/269] eta: 0:00:53 acc1_pos: 98.276 acc5_pos: 100.000 loss: 1.3413 (0.9438) acc1: 13.2812 (46.0005) acc5: 100.0000 (100.0000) acc1_neg: 32.812 acc5_neg: 100.000 time: 0.2576 data: 0.1652 max mem: 3531 +Test: [ 90/269] eta: 0:00:47 acc1_pos: 98.276 acc5_pos: 100.000 loss: 1.1798 (0.9712) acc1: 19.0104 (43.1891) acc5: 100.0000 (100.0000) acc1_neg: 7.552 acc5_neg: 100.000 time: 0.1732 data: 0.0777 max mem: 3531 +Test: [100/269] eta: 0:00:43 acc1_pos: 98.276 acc5_pos: 100.000 loss: 1.3227 (1.0229) acc1: 8.8542 (39.7561) acc5: 100.0000 (100.0000) acc1_neg: 2.344 acc5_neg: 100.000 time: 0.1506 data: 0.0559 max mem: 3531 +Test: [110/269] eta: 0:00:40 acc1_pos: 98.276 acc5_pos: 100.000 loss: 1.4815 (1.0659) acc1: 5.4688 (36.6319) acc5: 100.0000 (100.0000) acc1_neg: 7.031 acc5_neg: 100.000 time: 0.1916 data: 0.0959 max mem: 3531 +Test: [120/269] eta: 0:00:37 acc1_pos: 98.276 acc5_pos: 100.000 loss: 1.4673 (1.0988) acc1: 7.0312 (34.9712) acc5: 100.0000 (100.0000) acc1_neg: 11.719 acc5_neg: 100.000 time: 0.2073 data: 0.1077 max mem: 3531 +Test: [130/269] eta: 0:00:34 acc1_pos: 98.276 acc5_pos: 100.000 loss: 1.4372 (1.1295) acc1: 11.7188 (33.0431) acc5: 100.0000 (100.0000) acc1_neg: 9.896 acc5_neg: 100.000 time: 0.2036 data: 0.1065 max mem: 3531 +Test: [140/269] eta: 0:00:31 acc1_pos: 98.276 acc5_pos: 100.000 loss: 1.3492 (1.1436) acc1: 11.4583 (31.8152) acc5: 100.0000 (100.0000) acc1_neg: 21.094 acc5_neg: 100.000 time: 0.2164 data: 0.1199 max mem: 3531 +Test: [150/269] eta: 0:00:28 acc1_pos: 98.276 acc5_pos: 100.000 loss: 1.2386 (1.1511) acc1: 16.1458 (30.8085) acc5: 100.0000 (100.0000) acc1_neg: 16.927 acc5_neg: 100.000 time: 0.2094 data: 0.1125 max mem: 3531 +Test: [160/269] eta: 0:00:26 acc1_pos: 98.276 acc5_pos: 100.000 loss: 1.3113 (1.1663) acc1: 13.8021 (29.5791) acc5: 100.0000 (100.0000) acc1_neg: 8.333 acc5_neg: 100.000 time: 0.2024 data: 0.1065 max mem: 3531 +Test: [170/269] eta: 0:00:23 acc1_pos: 98.276 acc5_pos: 100.000 loss: 1.5534 (1.1872) acc1: 7.0312 (28.2438) acc5: 100.0000 (100.0000) acc1_neg: 9.115 acc5_neg: 100.000 time: 0.2259 data: 0.1319 max mem: 3531 +Test: [180/269] eta: 0:00:20 acc1_pos: 98.276 acc5_pos: 100.000 loss: 1.4954 (1.2055) acc1: 5.7292 (27.0459) acc5: 100.0000 (100.0000) acc1_neg: 6.771 acc5_neg: 100.000 time: 0.1923 data: 0.0978 max mem: 3531 +Test: [190/269] eta: 0:00:18 acc1_pos: 98.276 acc5_pos: 100.000 loss: 1.4587 (1.2205) acc1: 6.7708 (26.0771) acc5: 100.0000 (100.0000) acc1_neg: 1.823 acc5_neg: 100.000 time: 0.1741 data: 0.0798 max mem: 3531 +Test: [200/269] eta: 0:00:15 acc1_pos: 98.276 acc5_pos: 100.000 loss: 1.5424 (1.2398) acc1: 5.2083 (25.0466) acc5: 100.0000 (100.0000) acc1_neg: 6.250 acc5_neg: 100.000 time: 0.1962 data: 0.1049 max mem: 3531 +Test: [210/269] eta: 0:00:13 acc1_pos: 98.276 acc5_pos: 100.000 loss: 1.4615 (1.2479) acc1: 6.2500 (24.4804) acc5: 100.0000 (100.0000) acc1_neg: 12.240 acc5_neg: 100.000 time: 0.2047 data: 0.1097 max mem: 3532 +Test: [220/269] eta: 0:00:11 acc1_pos: 98.276 acc5_pos: 100.000 loss: 1.3300 (1.2524) acc1: 13.2812 (24.0986) acc5: 100.0000 (100.0000) acc1_neg: 30.729 acc5_neg: 100.000 time: 0.2048 data: 0.1092 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 98.276 acc5_pos: 100.000 loss: 1.3465 (1.2573) acc1: 14.3229 (23.7340) acc5: 100.0000 (100.0000) acc1_neg: 10.938 acc5_neg: 100.000 time: 0.1921 data: 0.0994 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 98.276 acc5_pos: 100.000 loss: 1.4211 (1.2667) acc1: 9.6354 (23.0139) acc5: 100.0000 (100.0000) acc1_neg: 19.792 acc5_neg: 100.000 time: 0.1532 data: 0.0606 max mem: 3532 +Test: [250/269] eta: 0:00:04 acc1_pos: 98.276 acc5_pos: 100.000 loss: 1.4292 (1.2705) acc1: 6.2500 (22.7590) acc5: 100.0000 (100.0000) acc1_neg: 11.458 acc5_neg: 100.000 time: 0.1129 data: 0.0220 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 98.276 acc5_pos: 100.000 loss: 1.5375 (1.2839) acc1: 6.2500 (22.0556) acc5: 100.0000 (100.0000) acc1_neg: 7.552 acc5_neg: 100.000 time: 0.1147 data: 0.0251 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 98.276 acc5_pos: 100.000 loss: 1.5375 (1.2917) acc1: 4.4271 (21.5416) acc5: 100.0000 (100.0000) acc1_neg: 13.648 acc5_neg: 100.000 time: 0.1221 data: 0.0289 max mem: 3532 +Test: Total time: 0:00:57 (0.2122 s / it) +* Overall Acc@1 21.542 Acc@5 100.000 + target == real Acc@1 96.566 Acc@5 100.000 + target == fake Acc@1 11.303 Acc@5 100.000 + loss 1.292 + PR-AUC 0.9229 ROC-AUC 0.6316 EER 41.04% (thr=0.2052) +Accuracy of the network on the 103293 test images: 21.5% +[rank0]:[W725 15:09:14.937328638 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-25T15:09:20Z] eval_done tag=fixed500k_fake_n11_seed1_resampled +[2026-07-25T15:09:20Z] eval_start tag=fixed500k_fake_n11_seed2_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed2_resampled_n116-040-168_20260725_112559/best_checkpoint.pth +started_utc=2026-07-25T15:09:20Z +hostname=n116-042-186 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed2_resampled_n116-040-168_20260725_112559/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed2_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0725 15:09:33.338000 68194 torch/distributed/run.py:766] +W0725 15:09:33.338000 68194 torch/distributed/run.py:766] ***************************************** +W0725 15:09:33.338000 68194 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 15:09:33.338000 68194 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 15:10:01.734291456 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 15:10:02.420687706 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 15:10:02.653540149 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 15:10:02.828924297 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 15:10:02.898908931 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 15:10:02.946231089 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 15:10:03.053089396 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 15:10:03.084096080 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed2_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed2_resampled_n116-040-168_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:19:01 acc1_pos: 76.562 acc5_pos: 100.000 loss: 0.5558 (0.5558) acc1: 76.5625 (76.5625) acc5: 100.0000 (100.0000) time: 4.8798 data: 2.8913 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/234] eta: 0:01:59 acc1_pos: 52.604 acc5_pos: 100.000 loss: 0.6685 (0.7493) acc1: 62.2396 (52.8646) acc5: 100.0000 (100.0000) time: 0.5346 data: 0.2685 max mem: 3531 +Test: [ 20/234] eta: 0:01:10 acc1_pos: 88.542 acc5_pos: 100.000 loss: 0.6158 (0.6715) acc1: 63.2812 (61.1359) acc5: 100.0000 (100.0000) time: 0.1019 data: 0.0100 max mem: 3531 +Test: [ 30/234] eta: 0:00:57 acc1_pos: 33.854 acc5_pos: 100.000 loss: 0.3720 (0.5894) acc1: 84.8958 (67.8931) acc5: 100.0000 (100.0000) time: 0.1420 data: 0.0513 max mem: 3531 +Test: [ 40/234] eta: 0:00:55 acc1_pos: 99.740 acc5_pos: 100.000 loss: 0.3426 (0.6300) acc1: 88.2812 (64.8819) acc5: 100.0000 (100.0000) time: 0.2433 data: 0.1514 max mem: 3531 +Test: [ 50/234] eta: 0:00:50 acc1_pos: 98.958 acc5_pos: 100.000 loss: 0.1251 (0.5303) acc1: 98.6979 (71.5278) acc5: 100.0000 (100.0000) time: 0.2581 data: 0.1645 max mem: 3531 +Test: [ 60/234] eta: 0:00:43 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1080 (0.4604) acc1: 99.4792 (76.1868) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0751 max mem: 3531 +Test: [ 70/234] eta: 0:00:40 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1080 (0.4548) acc1: 100.0000 (76.8156) acc5: 100.0000 (100.0000) acc1_neg: 68.750 acc5_neg: 100.000 time: 0.1729 data: 0.0781 max mem: 3531 +Test: [ 80/234] eta: 0:00:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5716 (0.4566) acc1: 71.3542 (77.0898) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2108 data: 0.1170 max mem: 3531 +Test: [ 90/234] eta: 0:00:33 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2928 (0.4362) acc1: 92.1875 (78.7832) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.2070 data: 0.1135 max mem: 3531 +Test: [100/234] eta: 0:00:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2883 (0.4433) acc1: 92.1875 (78.4653) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1963 data: 0.1028 max mem: 3531 +Test: [110/234] eta: 0:00:28 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5021 (0.4497) acc1: 73.4375 (78.1907) acc5: 100.0000 (100.0000) acc1_neg: 71.354 acc5_neg: 100.000 time: 0.2200 data: 0.1271 max mem: 3531 +Test: [120/234] eta: 0:00:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4452 (0.4547) acc1: 75.2604 (78.1702) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2076 data: 0.1131 max mem: 3531 +Test: [130/234] eta: 0:00:23 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4234 (0.4572) acc1: 82.2917 (78.0972) acc5: 100.0000 (100.0000) acc1_neg: 52.865 acc5_neg: 100.000 time: 0.1878 data: 0.0925 max mem: 3531 +Test: [140/234] eta: 0:00:21 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6116 (0.5016) acc1: 60.1562 (75.6704) acc5: 100.0000 (100.0000) acc1_neg: 15.885 acc5_neg: 100.000 time: 0.2060 data: 0.1121 max mem: 3531 +Test: [150/234] eta: 0:00:18 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.7366 (0.5177) acc1: 57.5521 (74.9276) acc5: 100.0000 (100.0000) acc1_neg: 74.219 acc5_neg: 100.000 time: 0.1998 data: 0.1063 max mem: 3531 +Test: [160/234] eta: 0:00:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5499 (0.5166) acc1: 73.1771 (75.0404) acc5: 100.0000 (100.0000) acc1_neg: 69.792 acc5_neg: 100.000 time: 0.2053 data: 0.1106 max mem: 3531 +Test: [170/234] eta: 0:00:14 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5402 (0.5166) acc1: 72.9167 (75.0396) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.2270 data: 0.1330 max mem: 3531 +Test: [180/234] eta: 0:00:11 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5397 (0.5243) acc1: 71.8750 (74.6230) acc5: 100.0000 (100.0000) acc1_neg: 30.729 acc5_neg: 100.000 time: 0.2140 data: 0.1211 max mem: 3531 +Test: [190/234] eta: 0:00:09 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.7513 (0.5389) acc1: 57.8125 (73.6611) acc5: 100.0000 (100.0000) acc1_neg: 54.948 acc5_neg: 100.000 time: 0.2252 data: 0.1336 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.7513 (0.5300) acc1: 58.5938 (74.1864) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2143 data: 0.1211 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1228 (0.5103) acc1: 99.4792 (75.3937) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1684 data: 0.0744 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1106 (0.4937) acc1: 99.4792 (76.3563) acc5: 100.0000 (100.0000) acc1_neg: 74.219 acc5_neg: 100.000 time: 0.1455 data: 0.0548 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4502 (0.5183) acc1: 32.8125 (74.3698) acc5: 100.0000 (100.0000) acc1_neg: 30.729 acc5_neg: 100.000 time: 0.1099 data: 0.0202 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.0535 (0.5254) acc1: 30.7292 (73.8446) acc5: 100.0000 (100.0000) acc1_neg: 27.237 acc5_neg: 100.000 time: 0.1137 data: 0.0230 max mem: 3532 +Test: Total time: 0:00:48 (0.2083 s / it) +* Overall Acc@1 73.845 Acc@5 100.000 + target == real Acc@1 71.737 Acc@5 100.000 + target == fake Acc@1 74.439 Acc@5 100.000 + loss 0.525 + PR-AUC 0.9419 ROC-AUC 0.8194 EER 26.66% (thr=0.5171) +Accuracy of the network on the 89729 test images: 73.8% +[rank0]:[W725 15:11:06.574372428 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 15:11:25.925000 68751 torch/distributed/run.py:766] +W0725 15:11:25.925000 68751 torch/distributed/run.py:766] ***************************************** +W0725 15:11:25.925000 68751 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 15:11:25.925000 68751 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 15:11:56.072880413 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 15:11:56.413087748 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W725 15:11:56.550600082 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W725 15:11:56.550986902 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 15:11:56.774426322 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 15:11:56.777472250 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 15:11:56.810087526 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 15:11:56.897860228 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed2_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed2_resampled_n116-040-168_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:48:43 acc1_pos: 35.417 acc5_pos: 100.000 loss: 0.9038 (0.9038) acc1: 35.4167 (35.4167) acc5: 100.0000 (100.0000) time: 7.2572 data: 5.5634 max mem: 3531 +Test: [ 10/2387] eta: 0:29:24 acc1_pos: 17.448 acc5_pos: 100.000 loss: 0.8689 (0.8564) acc1: 40.8854 (44.0341) acc5: 100.0000 (100.0000) time: 0.7422 data: 0.5062 max mem: 3531 +Test: [ 20/2387] eta: 0:17:12 acc1_pos: 16.927 acc5_pos: 100.000 loss: 0.8658 (0.8952) acc1: 40.6250 (39.1369) acc5: 100.0000 (100.0000) time: 0.0953 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:14:32 acc1_pos: 32.552 acc5_pos: 100.000 loss: 0.8759 (0.9022) acc1: 35.9375 (37.6344) acc5: 100.0000 (100.0000) time: 0.1658 data: 0.0686 max mem: 3531 +Test: [ 40/2387] eta: 0:13:39 acc1_pos: 32.031 acc5_pos: 100.000 loss: 0.8759 (0.9042) acc1: 35.9375 (37.5254) acc5: 100.0000 (100.0000) time: 0.2572 data: 0.1642 max mem: 3531 +Test: [ 50/2387] eta: 0:12:40 acc1_pos: 35.677 acc5_pos: 100.000 loss: 0.8282 (0.8924) acc1: 42.4479 (39.3076) acc5: 100.0000 (100.0000) time: 0.2558 data: 0.1643 max mem: 3531 +Test: [ 60/2387] eta: 0:12:15 acc1_pos: 60.156 acc5_pos: 100.000 loss: 0.8242 (0.8803) acc1: 43.2292 (40.4500) acc5: 100.0000 (100.0000) time: 0.2483 data: 0.1551 max mem: 3531 +Test: [ 70/2387] eta: 0:11:47 acc1_pos: 31.771 acc5_pos: 100.000 loss: 0.8442 (0.8842) acc1: 42.9688 (40.5443) acc5: 100.0000 (100.0000) time: 0.2538 data: 0.1593 max mem: 3531 +Test: [ 80/2387] eta: 0:11:28 acc1_pos: 85.677 acc5_pos: 100.000 loss: 0.7723 (0.8567) acc1: 45.8333 (43.1359) acc5: 100.0000 (100.0000) time: 0.2442 data: 0.1475 max mem: 3531 +Test: [ 90/2387] eta: 0:11:15 acc1_pos: 77.865 acc5_pos: 100.000 loss: 0.6050 (0.8182) acc1: 67.4479 (46.9494) acc5: 100.0000 (100.0000) time: 0.2542 data: 0.1544 max mem: 3531 +Test: [ 100/2387] eta: 0:11:25 acc1_pos: 84.635 acc5_pos: 100.000 loss: 0.4789 (0.7915) acc1: 74.2188 (49.4869) acc5: 100.0000 (100.0000) time: 0.3049 data: 0.2052 max mem: 3531 +Test: [ 110/2387] eta: 0:11:04 acc1_pos: 64.844 acc5_pos: 100.000 loss: 0.5227 (0.7686) acc1: 74.2188 (51.8558) acc5: 100.0000 (100.0000) time: 0.2813 data: 0.1836 max mem: 3531 +Test: [ 120/2387] eta: 0:10:50 acc1_pos: 66.406 acc5_pos: 100.000 loss: 0.5680 (0.7551) acc1: 71.6146 (53.4069) acc5: 100.0000 (100.0000) time: 0.2227 data: 0.1283 max mem: 3531 +Test: [ 130/2387] eta: 0:10:36 acc1_pos: 41.927 acc5_pos: 100.000 loss: 0.5418 (0.7379) acc1: 73.6979 (55.0573) acc5: 100.0000 (100.0000) time: 0.2285 data: 0.1371 max mem: 3531 +Test: [ 140/2387] eta: 0:10:27 acc1_pos: 68.750 acc5_pos: 100.000 loss: 0.5303 (0.7259) acc1: 76.3021 (56.2094) acc5: 100.0000 (100.0000) time: 0.2321 data: 0.1376 max mem: 3531 +Test: [ 150/2387] eta: 0:10:20 acc1_pos: 66.146 acc5_pos: 100.000 loss: 0.5167 (0.7113) acc1: 76.3021 (57.6452) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1474 max mem: 3531 +Test: [ 160/2387] eta: 0:10:10 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.5027 (0.7006) acc1: 77.3438 (58.6843) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.2404 data: 0.1446 max mem: 3531 +Test: [ 170/2387] eta: 0:09:59 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.3744 (0.6812) acc1: 89.0625 (60.5217) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.2199 data: 0.1232 max mem: 3531 +Test: [ 180/2387] eta: 0:09:52 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.3696 (0.6641) acc1: 89.5833 (62.1302) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.2217 data: 0.1239 max mem: 3531 +Test: [ 190/2387] eta: 0:09:43 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.3714 (0.6488) acc1: 89.8438 (63.5703) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.2249 data: 0.1300 max mem: 3531 +Test: [ 200/2387] eta: 0:09:35 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.3730 (0.6352) acc1: 89.5833 (64.8450) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.2149 data: 0.1202 max mem: 3531 +Test: [ 210/2387] eta: 0:09:27 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.3747 (0.6231) acc1: 89.3229 (65.9903) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.2148 data: 0.1209 max mem: 3532 +Test: [ 220/2387] eta: 0:09:17 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.3751 (0.6117) acc1: 89.3229 (67.0449) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.1976 data: 0.1034 max mem: 3532 +Test: [ 230/2387] eta: 0:09:12 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.3731 (0.6014) acc1: 89.5833 (68.0060) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.2054 data: 0.1090 max mem: 3532 +Test: [ 240/2387] eta: 0:09:08 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.3923 (0.6003) acc1: 86.1979 (68.0185) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.2349 data: 0.1394 max mem: 3532 +Test: [ 250/2387] eta: 0:08:57 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.3777 (0.5848) acc1: 72.1354 (69.1349) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1996 data: 0.1032 max mem: 3532 +Test: [ 260/2387] eta: 0:08:42 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.1665 (0.5687) acc1: 99.7396 (70.2945) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1311 data: 0.0336 max mem: 3532 +Test: [ 270/2387] eta: 0:08:35 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.1434 (0.5531) acc1: 100.0000 (71.3705) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1446 data: 0.0491 max mem: 3532 +Test: [ 280/2387] eta: 0:08:28 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.1457 (0.5390) acc1: 100.0000 (72.3838) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1837 data: 0.0885 max mem: 3532 +Test: [ 290/2387] eta: 0:08:22 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.1587 (0.5257) acc1: 100.0000 (73.3122) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0840 max mem: 3532 +Test: [ 300/2387] eta: 0:08:15 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.1501 (0.5138) acc1: 100.0000 (74.1712) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1802 data: 0.0817 max mem: 3532 +Test: [ 310/2387] eta: 0:08:11 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.1938 (0.5067) acc1: 99.7396 (74.9431) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1908 data: 0.0983 max mem: 3532 +Test: [ 320/2387] eta: 0:08:05 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.3780 (0.5028) acc1: 97.6562 (75.6506) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2001 data: 0.1085 max mem: 3532 +Test: [ 330/2387] eta: 0:07:59 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.3813 (0.4991) acc1: 97.6562 (76.3280) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1840 data: 0.0930 max mem: 3532 +Test: [ 340/2387] eta: 0:07:56 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.3803 (0.4956) acc1: 98.1771 (76.9650) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1962 data: 0.1036 max mem: 3532 +Test: [ 350/2387] eta: 0:07:52 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.3710 (0.4893) acc1: 98.6979 (77.5982) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2099 data: 0.1178 max mem: 3532 +Test: [ 360/2387] eta: 0:07:49 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2871 (0.4839) acc1: 98.1771 (78.0853) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2131 data: 0.1213 max mem: 3532 +Test: [ 370/2387] eta: 0:07:44 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2957 (0.4788) acc1: 94.7917 (78.5398) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2049 data: 0.1120 max mem: 3532 +Test: [ 380/2387] eta: 0:07:41 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2937 (0.4739) acc1: 95.0521 (78.9732) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1949 data: 0.1011 max mem: 3532 +Test: [ 390/2387] eta: 0:07:37 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2959 (0.4696) acc1: 94.7917 (79.3658) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2008 data: 0.1083 max mem: 3532 +Test: [ 400/2387] eta: 0:07:34 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2976 (0.4653) acc1: 94.7917 (79.7609) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2099 data: 0.1182 max mem: 3532 +Test: [ 410/2387] eta: 0:07:30 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2953 (0.4611) acc1: 95.0521 (80.1329) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2043 data: 0.1111 max mem: 3532 +Test: [ 420/2387] eta: 0:07:26 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2951 (0.4572) acc1: 94.7917 (80.4879) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1954 data: 0.1018 max mem: 3532 +Test: [ 430/2387] eta: 0:07:23 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.3017 (0.4560) acc1: 94.2708 (80.5968) acc5: 100.0000 (100.0000) acc1_neg: 76.302 acc5_neg: 100.000 time: 0.2013 data: 0.1077 max mem: 3532 +Test: [ 440/2387] eta: 0:07:20 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.3134 (0.4516) acc1: 82.0312 (80.8467) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2044 data: 0.1095 max mem: 3533 +Test: [ 450/2387] eta: 0:07:17 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.1297 (0.4444) acc1: 100.0000 (81.2714) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2099 data: 0.1161 max mem: 3533 +Test: [ 460/2387] eta: 0:07:13 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.1274 (0.4376) acc1: 100.0000 (81.6776) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2024 data: 0.1092 max mem: 3533 +Test: [ 470/2387] eta: 0:07:09 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.1272 (0.4310) acc1: 100.0000 (82.0661) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1889 data: 0.0939 max mem: 3533 +Test: [ 480/2387] eta: 0:07:06 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.1280 (0.4247) acc1: 100.0000 (82.4384) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1897 data: 0.0968 max mem: 3533 +Test: [ 490/2387] eta: 0:07:03 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.1298 (0.4187) acc1: 100.0000 (82.7955) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2050 data: 0.1119 max mem: 3533 +Test: [ 500/2387] eta: 0:07:01 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.1311 (0.4130) acc1: 100.0000 (83.1389) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2137 data: 0.1208 max mem: 3533 +Test: [ 510/2387] eta: 0:06:58 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.1300 (0.4081) acc1: 100.0000 (83.4638) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2043 data: 0.1130 max mem: 3533 +Test: [ 520/2387] eta: 0:06:56 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2146 (0.4050) acc1: 99.4792 (83.7677) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2158 data: 0.1245 max mem: 3533 +Test: [ 530/2387] eta: 0:06:52 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2096 (0.4012) acc1: 99.4792 (84.0665) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1997 data: 0.1084 max mem: 3533 +Test: [ 540/2387] eta: 0:06:49 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2073 (0.3977) acc1: 99.7396 (84.3557) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1932 data: 0.0988 max mem: 3533 +Test: [ 550/2387] eta: 0:06:47 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2091 (0.3942) acc1: 99.7396 (84.6331) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2157 data: 0.1221 max mem: 3533 +Test: [ 560/2387] eta: 0:06:44 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2076 (0.3909) acc1: 99.7396 (84.9019) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2149 data: 0.1234 max mem: 3533 +Test: [ 570/2387] eta: 0:06:41 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2058 (0.3877) acc1: 99.7396 (85.1626) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2090 data: 0.1155 max mem: 3533 +Test: [ 580/2387] eta: 0:06:40 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2073 (0.3847) acc1: 99.7396 (85.4117) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2196 data: 0.1270 max mem: 3533 +Test: [ 590/2387] eta: 0:06:36 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2098 (0.3817) acc1: 99.7396 (85.6546) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1999 data: 0.1087 max mem: 3533 +Test: [ 600/2387] eta: 0:06:32 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2131 (0.3797) acc1: 99.4792 (85.8768) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1652 data: 0.0707 max mem: 3533 +Test: [ 610/2387] eta: 0:06:29 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2803 (0.3785) acc1: 96.0938 (86.0232) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1759 data: 0.0818 max mem: 3533 +Test: [ 620/2387] eta: 0:06:26 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.3046 (0.3773) acc1: 94.5312 (86.1610) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1953 data: 0.1036 max mem: 3533 +Test: [ 630/2387] eta: 0:06:24 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.3047 (0.3761) acc1: 94.5312 (86.3015) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2116 data: 0.1190 max mem: 3533 +Test: [ 640/2387] eta: 0:06:22 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.3068 (0.3750) acc1: 94.5312 (86.4283) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2168 data: 0.1254 max mem: 3533 +Test: [ 650/2387] eta: 0:06:19 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.3068 (0.3740) acc1: 94.5312 (86.5603) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2108 data: 0.1157 max mem: 3533 +Test: [ 660/2387] eta: 0:06:16 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.3051 (0.3729) acc1: 94.2708 (86.6762) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1943 data: 0.0981 max mem: 3533 +Test: [ 670/2387] eta: 0:06:13 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.3057 (0.3719) acc1: 94.5312 (86.7913) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1923 data: 0.1011 max mem: 3534 +Test: [ 680/2387] eta: 0:06:11 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.3107 (0.3727) acc1: 94.0104 (86.7119) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.1978 data: 0.1051 max mem: 3534 +Test: [ 690/2387] eta: 0:06:09 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.4825 (0.3783) acc1: 73.6979 (86.2085) acc5: 100.0000 (100.0000) acc1_neg: 41.146 acc5_neg: 100.000 time: 0.2187 data: 0.1250 max mem: 3534 +Test: [ 700/2387] eta: 0:06:07 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.8597 (0.3853) acc1: 44.0104 (85.6017) acc5: 100.0000 (100.0000) acc1_neg: 45.573 acc5_neg: 100.000 time: 0.2222 data: 0.1291 max mem: 3534 +Test: [ 710/2387] eta: 0:06:02 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.8598 (0.3921) acc1: 43.2292 (85.0090) acc5: 100.0000 (100.0000) acc1_neg: 42.708 acc5_neg: 100.000 time: 0.1656 data: 0.0727 max mem: 3534 +Test: [ 720/2387] eta: 0:05:59 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.8619 (0.3986) acc1: 42.7083 (84.4155) acc5: 100.0000 (100.0000) acc1_neg: 41.146 acc5_neg: 100.000 time: 0.1526 data: 0.0601 max mem: 3534 +Test: [ 730/2387] eta: 0:05:57 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.8662 (0.4050) acc1: 42.7083 (83.8502) acc5: 100.0000 (100.0000) acc1_neg: 45.833 acc5_neg: 100.000 time: 0.1853 data: 0.0927 max mem: 3534 +Test: [ 740/2387] eta: 0:05:54 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.8619 (0.4112) acc1: 44.0104 (83.3189) acc5: 100.0000 (100.0000) acc1_neg: 44.010 acc5_neg: 100.000 time: 0.1860 data: 0.0947 max mem: 3534 +Test: [ 750/2387] eta: 0:05:50 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.8553 (0.4171) acc1: 44.5312 (82.7938) acc5: 100.0000 (100.0000) acc1_neg: 46.875 acc5_neg: 100.000 time: 0.1706 data: 0.0797 max mem: 3534 +Test: [ 760/2387] eta: 0:05:48 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.8425 (0.4227) acc1: 45.0521 (82.3156) acc5: 100.0000 (100.0000) acc1_neg: 50.781 acc5_neg: 100.000 time: 0.1900 data: 0.0977 max mem: 3534 +Test: [ 770/2387] eta: 0:05:45 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.8112 (0.4253) acc1: 49.7396 (82.1042) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.1986 data: 0.1055 max mem: 3534 +Test: [ 780/2387] eta: 0:05:43 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.3855 (0.4248) acc1: 85.1562 (82.1906) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1800 data: 0.0870 max mem: 3534 +Test: [ 790/2387] eta: 0:05:39 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.3040 (0.4232) acc1: 97.6562 (82.3167) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.1597 data: 0.0672 max mem: 3534 +Test: [ 800/2387] eta: 0:05:36 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2858 (0.4218) acc1: 95.8333 (82.4793) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1561 data: 0.0645 max mem: 3534 +Test: [ 810/2387] eta: 0:05:33 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.3249 (0.4208) acc1: 95.3125 (82.5864) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.1807 data: 0.0885 max mem: 3534 +Test: [ 820/2387] eta: 0:05:30 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.3363 (0.4207) acc1: 90.8854 (82.6244) acc5: 100.0000 (100.0000) acc1_neg: 57.292 acc5_neg: 100.000 time: 0.1725 data: 0.0800 max mem: 3534 +Test: [ 830/2387] eta: 0:05:27 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.5583 (0.4252) acc1: 67.9688 (82.2089) acc5: 100.0000 (100.0000) acc1_neg: 49.740 acc5_neg: 100.000 time: 0.1609 data: 0.0689 max mem: 3534 +Test: [ 840/2387] eta: 0:05:25 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.5156 (0.4250) acc1: 68.7500 (82.2818) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1824 data: 0.0910 max mem: 3534 +Test: [ 850/2387] eta: 0:05:23 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.3693 (0.4242) acc1: 90.3646 (82.3783) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1979 data: 0.1065 max mem: 3534 +Test: [ 860/2387] eta: 0:05:20 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.3332 (0.4232) acc1: 93.2292 (82.5094) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1944 data: 0.1031 max mem: 3534 +Test: [ 870/2387] eta: 0:05:18 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.3332 (0.4221) acc1: 94.0104 (82.6256) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1907 data: 0.0996 max mem: 3534 +Test: [ 880/2387] eta: 0:05:15 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.3531 (0.4215) acc1: 91.4062 (82.6949) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1927 data: 0.1011 max mem: 3534 +Test: [ 890/2387] eta: 0:05:14 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.3607 (0.4216) acc1: 84.6354 (82.6728) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.2107 data: 0.1181 max mem: 3534 +Test: [ 900/2387] eta: 0:05:11 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.5266 (0.4240) acc1: 68.2292 (82.4665) acc5: 100.0000 (100.0000) acc1_neg: 58.594 acc5_neg: 100.000 time: 0.2027 data: 0.1104 max mem: 3535 +Test: [ 910/2387] eta: 0:05:09 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.6629 (0.4266) acc1: 61.4583 (82.2345) acc5: 100.0000 (100.0000) acc1_neg: 59.115 acc5_neg: 100.000 time: 0.1954 data: 0.1034 max mem: 3535 +Test: [ 920/2387] eta: 0:05:07 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.6697 (0.4293) acc1: 60.4167 (81.9942) acc5: 100.0000 (100.0000) acc1_neg: 57.552 acc5_neg: 100.000 time: 0.2146 data: 0.1210 max mem: 3535 +Test: [ 930/2387] eta: 0:05:04 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.6667 (0.4318) acc1: 60.1562 (81.7787) acc5: 100.0000 (100.0000) acc1_neg: 58.594 acc5_neg: 100.000 time: 0.2028 data: 0.1085 max mem: 3535 +Test: [ 940/2387] eta: 0:05:02 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.6689 (0.4343) acc1: 60.9375 (81.5500) acc5: 100.0000 (100.0000) acc1_neg: 60.417 acc5_neg: 100.000 time: 0.1936 data: 0.1003 max mem: 3535 +Test: [ 950/2387] eta: 0:05:00 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.6765 (0.4368) acc1: 60.4167 (81.3250) acc5: 100.0000 (100.0000) acc1_neg: 61.979 acc5_neg: 100.000 time: 0.1949 data: 0.1025 max mem: 3535 +Test: [ 960/2387] eta: 0:04:57 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.6560 (0.4391) acc1: 61.1979 (81.1159) acc5: 100.0000 (100.0000) acc1_neg: 64.583 acc5_neg: 100.000 time: 0.1851 data: 0.0930 max mem: 3535 +Test: [ 970/2387] eta: 0:04:55 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.6711 (0.4416) acc1: 60.6771 (80.8971) acc5: 100.0000 (100.0000) acc1_neg: 61.458 acc5_neg: 100.000 time: 0.1753 data: 0.0835 max mem: 3535 +Test: [ 980/2387] eta: 0:04:52 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.6524 (0.4425) acc1: 61.7188 (80.8255) acc5: 100.0000 (100.0000) acc1_neg: 61.719 acc5_neg: 100.000 time: 0.1747 data: 0.0839 max mem: 3535 +Test: [ 990/2387] eta: 0:04:50 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.4003 (0.4417) acc1: 82.2917 (80.9021) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1815 data: 0.0885 max mem: 3535 +Test: [1000/2387] eta: 0:04:47 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2990 (0.4405) acc1: 94.7917 (81.0229) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1802 data: 0.0860 max mem: 3535 +Test: [1010/2387] eta: 0:04:45 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.3398 (0.4409) acc1: 91.4062 (80.9909) acc5: 100.0000 (100.0000) acc1_neg: 67.708 acc5_neg: 100.000 time: 0.1740 data: 0.0825 max mem: 3535 +Test: [1020/2387] eta: 0:04:42 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.4120 (0.4415) acc1: 78.1250 (80.9138) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1754 data: 0.0846 max mem: 3535 +Test: [1030/2387] eta: 0:04:40 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2897 (0.4400) acc1: 94.2708 (81.0146) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2004 data: 0.1089 max mem: 3535 +Test: [1040/2387] eta: 0:04:38 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2124 (0.4377) acc1: 98.9583 (81.1797) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2189 data: 0.1279 max mem: 3535 +Test: [1050/2387] eta: 0:04:36 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2036 (0.4357) acc1: 99.2188 (81.3365) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2002 data: 0.1099 max mem: 3535 +Test: [1060/2387] eta: 0:04:34 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2107 (0.4340) acc1: 98.4375 (81.4802) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1994 data: 0.1081 max mem: 3535 +Test: [1070/2387] eta: 0:04:32 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2765 (0.4334) acc1: 96.8750 (81.5272) acc5: 100.0000 (100.0000) acc1_neg: 36.198 acc5_neg: 100.000 time: 0.1948 data: 0.1022 max mem: 3535 +Test: [1080/2387] eta: 0:04:29 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.4008 (0.4348) acc1: 75.0000 (81.3823) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.1643 data: 0.0721 max mem: 3535 +Test: [1090/2387] eta: 0:04:27 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.4008 (0.4335) acc1: 74.4792 (81.4784) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1711 data: 0.0793 max mem: 3535 +Test: [1100/2387] eta: 0:04:24 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2360 (0.4318) acc1: 98.1771 (81.6277) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1758 data: 0.0849 max mem: 3535 +Test: [1110/2387] eta: 0:04:22 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2383 (0.4303) acc1: 98.6979 (81.7354) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1776 data: 0.0848 max mem: 3535 +Test: [1120/2387] eta: 0:04:20 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2396 (0.4288) acc1: 97.9167 (81.8658) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1890 data: 0.0951 max mem: 3536 +Test: [1130/2387] eta: 0:04:17 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2428 (0.4271) acc1: 97.9167 (82.0050) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1892 data: 0.0970 max mem: 3536 +Test: [1140/2387] eta: 0:04:15 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2746 (0.4280) acc1: 95.8333 (81.9313) acc5: 100.0000 (100.0000) acc1_neg: 66.406 acc5_neg: 100.000 time: 0.1833 data: 0.0912 max mem: 3536 +Test: [1150/2387] eta: 0:04:13 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.4379 (0.4277) acc1: 75.2604 (81.9394) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1814 data: 0.0892 max mem: 3536 +Test: [1160/2387] eta: 0:04:10 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2309 (0.4258) acc1: 97.9167 (82.0804) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1838 data: 0.0924 max mem: 3536 +Test: [1170/2387] eta: 0:04:08 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2139 (0.4239) acc1: 99.7396 (82.2216) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1911 data: 0.0998 max mem: 3536 +Test: [1180/2387] eta: 0:04:06 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.1999 (0.4220) acc1: 99.4792 (82.3541) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1946 data: 0.1033 max mem: 3536 +Test: [1190/2387] eta: 0:04:04 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2023 (0.4202) acc1: 99.2188 (82.4933) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1920 data: 0.1013 max mem: 3536 +Test: [1200/2387] eta: 0:04:02 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2192 (0.4187) acc1: 99.2188 (82.6121) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2063 data: 0.1148 max mem: 3536 +Test: [1210/2387] eta: 0:04:00 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2558 (0.4176) acc1: 97.1354 (82.7157) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.1029 max mem: 3536 +Test: [1220/2387] eta: 0:03:58 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2365 (0.4160) acc1: 97.3958 (82.8462) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1783 data: 0.0856 max mem: 3536 +Test: [1230/2387] eta: 0:03:55 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2090 (0.4142) acc1: 99.4792 (82.9765) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1800 data: 0.0875 max mem: 3536 +Test: [1240/2387] eta: 0:03:53 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.1883 (0.4124) acc1: 99.7396 (83.1067) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1945 data: 0.1032 max mem: 3536 +Test: [1250/2387] eta: 0:03:51 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.1924 (0.4107) acc1: 99.7396 (83.2353) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2105 data: 0.1199 max mem: 3536 +Test: [1260/2387] eta: 0:03:49 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.1847 (0.4090) acc1: 99.7396 (83.3556) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2058 data: 0.1114 max mem: 3536 +Test: [1270/2387] eta: 0:03:47 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.1939 (0.4075) acc1: 99.2188 (83.4743) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2105 data: 0.1149 max mem: 3536 +Test: [1280/2387] eta: 0:03:46 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2374 (0.4064) acc1: 96.6146 (83.5551) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2229 data: 0.1310 max mem: 3536 +Test: [1290/2387] eta: 0:03:43 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2518 (0.4052) acc1: 94.2708 (83.6414) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2051 data: 0.1136 max mem: 3536 +Test: [1300/2387] eta: 0:03:42 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2518 (0.4039) acc1: 95.3125 (83.7357) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2147 data: 0.1237 max mem: 3536 +Test: [1310/2387] eta: 0:03:40 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2448 (0.4029) acc1: 95.5729 (83.8246) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.2385 data: 0.1454 max mem: 3536 +Test: [1320/2387] eta: 0:03:38 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2596 (0.4019) acc1: 94.0104 (83.9046) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2056 data: 0.1119 max mem: 3536 +Test: [1330/2387] eta: 0:03:35 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2949 (0.4019) acc1: 90.8854 (83.9105) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.1799 data: 0.0852 max mem: 3536 +Test: [1340/2387] eta: 0:03:34 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.4490 (0.4030) acc1: 77.0833 (83.7946) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2090 data: 0.1117 max mem: 3536 +Test: [1350/2387] eta: 0:03:32 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.1492 (0.4008) acc1: 100.0000 (83.9145) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2399 data: 0.1451 max mem: 3537 +Test: [1360/2387] eta: 0:03:29 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.0934 (0.3985) acc1: 100.0000 (84.0327) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1951 data: 0.1017 max mem: 3537 +Test: [1370/2387] eta: 0:03:27 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.0859 (0.3963) acc1: 100.0000 (84.1492) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0791 max mem: 3537 +Test: [1380/2387] eta: 0:03:25 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.0872 (0.3941) acc1: 100.0000 (84.2639) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2047 data: 0.1073 max mem: 3537 +Test: [1390/2387] eta: 0:03:23 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.0999 (0.3920) acc1: 100.0000 (84.3771) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2099 data: 0.1126 max mem: 3537 +Test: [1400/2387] eta: 0:03:21 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.1173 (0.3902) acc1: 100.0000 (84.4876) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0932 max mem: 3537 +Test: [1410/2387] eta: 0:03:19 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.1481 (0.3902) acc1: 100.0000 (84.4865) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.1699 data: 0.0743 max mem: 3537 +Test: [1420/2387] eta: 0:03:17 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.4023 (0.3903) acc1: 85.6771 (84.5049) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1909 data: 0.0964 max mem: 3537 +Test: [1430/2387] eta: 0:03:15 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.4002 (0.3904) acc1: 88.5417 (84.5219) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.2050 data: 0.1111 max mem: 3537 +Test: [1440/2387] eta: 0:03:12 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.4053 (0.3909) acc1: 85.1562 (84.5044) acc5: 100.0000 (100.0000) acc1_neg: 38.021 acc5_neg: 100.000 time: 0.1879 data: 0.0938 max mem: 3537 +Test: [1450/2387] eta: 0:03:10 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.4617 (0.3921) acc1: 73.6979 (84.3851) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1750 data: 0.0823 max mem: 3537 +Test: [1460/2387] eta: 0:03:08 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2642 (0.3913) acc1: 86.4583 (84.4561) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1618 data: 0.0708 max mem: 3537 +Test: [1470/2387] eta: 0:03:05 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2487 (0.3902) acc1: 97.1354 (84.5430) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1515 data: 0.0584 max mem: 3537 +Test: [1480/2387] eta: 0:03:03 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2194 (0.3890) acc1: 98.4375 (84.6331) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1741 data: 0.0778 max mem: 3537 +Test: [1490/2387] eta: 0:03:01 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2057 (0.3879) acc1: 98.9583 (84.7248) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1897 data: 0.0948 max mem: 3537 +Test: [1500/2387] eta: 0:02:59 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2156 (0.3868) acc1: 97.9167 (84.8035) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1946 data: 0.0997 max mem: 3537 +Test: [1510/2387] eta: 0:02:57 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2442 (0.3862) acc1: 96.6146 (84.8588) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.2233 data: 0.1268 max mem: 3537 +Test: [1520/2387] eta: 0:02:55 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.3042 (0.3859) acc1: 91.4062 (84.9042) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1971 data: 0.1011 max mem: 3537 +Test: [1530/2387] eta: 0:02:53 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.3394 (0.3855) acc1: 91.9271 (84.9540) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1799 data: 0.0838 max mem: 3537 +Test: [1540/2387] eta: 0:02:51 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2905 (0.3849) acc1: 95.0521 (85.0246) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2077 data: 0.1093 max mem: 3537 +Test: [1550/2387] eta: 0:02:49 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2836 (0.3843) acc1: 96.0938 (85.0941) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2090 data: 0.1124 max mem: 3537 +Test: [1560/2387] eta: 0:02:47 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2896 (0.3838) acc1: 94.5312 (85.1409) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1687 data: 0.0750 max mem: 3537 +Test: [1570/2387] eta: 0:02:45 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.3129 (0.3837) acc1: 90.8854 (85.1556) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1799 data: 0.0863 max mem: 3537 +Test: [1580/2387] eta: 0:02:43 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.3413 (0.3839) acc1: 89.5833 (85.1523) acc5: 100.0000 (100.0000) acc1_neg: 55.729 acc5_neg: 100.000 time: 0.2369 data: 0.1431 max mem: 3538 +Test: [1590/2387] eta: 0:02:41 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.3404 (0.3837) acc1: 91.4062 (85.1995) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2296 data: 0.1340 max mem: 3538 +Test: [1600/2387] eta: 0:02:39 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.3027 (0.3833) acc1: 93.7500 (85.2316) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2297 data: 0.1341 max mem: 3538 +Test: [1610/2387] eta: 0:02:37 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.3112 (0.3828) acc1: 95.3125 (85.2940) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2253 data: 0.1305 max mem: 3538 +Test: [1620/2387] eta: 0:02:35 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.3109 (0.3824) acc1: 93.2292 (85.3312) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1843 data: 0.0899 max mem: 3538 +Test: [1630/2387] eta: 0:02:33 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.3807 (0.3832) acc1: 87.7604 (85.2792) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.2214 data: 0.1269 max mem: 3538 +Test: [1640/2387] eta: 0:02:32 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.5829 (0.3854) acc1: 63.5417 (85.0777) acc5: 100.0000 (100.0000) acc1_neg: 68.229 acc5_neg: 100.000 time: 0.2741 data: 0.1802 max mem: 3538 +Test: [1650/2387] eta: 0:02:29 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.6614 (0.3871) acc1: 56.7708 (84.9184) acc5: 100.0000 (100.0000) acc1_neg: 62.760 acc5_neg: 100.000 time: 0.1830 data: 0.0863 max mem: 3538 +Test: [1660/2387] eta: 0:02:27 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.6420 (0.3887) acc1: 60.6771 (84.7627) acc5: 100.0000 (100.0000) acc1_neg: 69.010 acc5_neg: 100.000 time: 0.1345 data: 0.0355 max mem: 3538 +Test: [1670/2387] eta: 0:02:25 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.6575 (0.3903) acc1: 59.6354 (84.6175) acc5: 100.0000 (100.0000) acc1_neg: 74.740 acc5_neg: 100.000 time: 0.1715 data: 0.0762 max mem: 3538 +Test: [1680/2387] eta: 0:02:23 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.6521 (0.3916) acc1: 59.6354 (84.4869) acc5: 100.0000 (100.0000) acc1_neg: 54.688 acc5_neg: 100.000 time: 0.1878 data: 0.0963 max mem: 3538 +Test: [1690/2387] eta: 0:02:21 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.6531 (0.3933) acc1: 59.6354 (84.3333) acc5: 100.0000 (100.0000) acc1_neg: 51.042 acc5_neg: 100.000 time: 0.1851 data: 0.0932 max mem: 3538 +Test: [1700/2387] eta: 0:02:18 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.6709 (0.3943) acc1: 59.6354 (84.2430) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1649 data: 0.0735 max mem: 3538 +Test: [1710/2387] eta: 0:02:16 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2778 (0.3935) acc1: 94.0104 (84.3102) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1835 data: 0.0924 max mem: 3538 +Test: [1720/2387] eta: 0:02:14 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2556 (0.3927) acc1: 96.8750 (84.3821) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1975 data: 0.1060 max mem: 3538 +Test: [1730/2387] eta: 0:02:12 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2355 (0.3918) acc1: 98.9583 (84.4573) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1865 data: 0.0944 max mem: 3538 +Test: [1740/2387] eta: 0:02:10 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2429 (0.3910) acc1: 98.9583 (84.5342) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1944 data: 0.1019 max mem: 3538 +Test: [1750/2387] eta: 0:02:08 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2511 (0.3903) acc1: 96.8750 (84.5972) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2188 data: 0.1233 max mem: 3538 +Test: [1760/2387] eta: 0:02:06 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2683 (0.3899) acc1: 95.5729 (84.6351) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.2110 data: 0.1166 max mem: 3538 +Test: [1770/2387] eta: 0:02:04 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.3511 (0.3898) acc1: 91.6667 (84.6575) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.1565 data: 0.0652 max mem: 3538 +Test: [1780/2387] eta: 0:02:02 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.3630 (0.3896) acc1: 90.1042 (84.6901) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1639 data: 0.0718 max mem: 3538 +Test: [1790/2387] eta: 0:02:00 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.3027 (0.3890) acc1: 94.7917 (84.7567) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2023 data: 0.1091 max mem: 3538 +Test: [1800/2387] eta: 0:01:58 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2975 (0.3886) acc1: 97.1354 (84.8195) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1920 data: 0.0971 max mem: 3538 +Test: [1810/2387] eta: 0:01:56 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2975 (0.3880) acc1: 96.0938 (84.8802) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.2316 data: 0.1380 max mem: 3539 +Test: [1820/2387] eta: 0:01:54 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.3049 (0.3877) acc1: 93.4896 (84.9190) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2526 data: 0.1604 max mem: 3539 +Test: [1830/2387] eta: 0:01:52 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.3312 (0.3873) acc1: 89.8438 (84.9412) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2331 data: 0.1385 max mem: 3539 +Test: [1840/2387] eta: 0:01:50 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.1470 (0.3860) acc1: 99.4792 (85.0216) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2199 data: 0.1231 max mem: 3539 +Test: [1850/2387] eta: 0:01:48 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.1422 (0.3846) acc1: 99.7396 (85.1008) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1653 data: 0.0681 max mem: 3539 +Test: [1860/2387] eta: 0:01:46 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.1427 (0.3833) acc1: 99.7396 (85.1795) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1651 data: 0.0699 max mem: 3539 +Test: [1870/2387] eta: 0:01:44 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.1438 (0.3821) acc1: 99.7396 (85.2563) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2216 data: 0.1297 max mem: 3539 +Test: [1880/2387] eta: 0:01:42 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.1431 (0.3808) acc1: 99.4792 (85.3319) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2522 data: 0.1597 max mem: 3539 +Test: [1890/2387] eta: 0:01:40 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.1418 (0.3795) acc1: 99.4792 (85.4079) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2642 data: 0.1702 max mem: 3539 +Test: [1900/2387] eta: 0:01:38 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.1442 (0.3783) acc1: 99.4792 (85.4819) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2131 data: 0.1191 max mem: 3539 +Test: [1910/2387] eta: 0:01:36 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2019 (0.3776) acc1: 98.4375 (85.5362) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.2035 data: 0.1098 max mem: 3539 +Test: [1920/2387] eta: 0:01:34 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2569 (0.3785) acc1: 96.0938 (85.4971) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2197 data: 0.1250 max mem: 3539 +Test: [1930/2387] eta: 0:01:32 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.3917 (0.3785) acc1: 92.4479 (85.5300) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1914 data: 0.0980 max mem: 3539 +Test: [1940/2387] eta: 0:01:30 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.3779 (0.3788) acc1: 91.4062 (85.5291) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.2315 data: 0.1396 max mem: 3539 +Test: [1950/2387] eta: 0:01:28 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.4263 (0.3790) acc1: 89.5833 (85.5301) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.2143 data: 0.1218 max mem: 3539 +Test: [1960/2387] eta: 0:01:26 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.4255 (0.3792) acc1: 89.0625 (85.5310) acc5: 100.0000 (100.0000) acc1_neg: 65.885 acc5_neg: 100.000 time: 0.1749 data: 0.0829 max mem: 3539 +Test: [1970/2387] eta: 0:01:24 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.5246 (0.3806) acc1: 70.5729 (85.3990) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.2090 data: 0.1150 max mem: 3539 +Test: [1980/2387] eta: 0:01:22 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.5709 (0.3809) acc1: 65.1042 (85.3505) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2078 data: 0.1115 max mem: 3539 +Test: [1990/2387] eta: 0:01:20 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.1431 (0.3797) acc1: 98.1771 (85.4178) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2020 data: 0.1086 max mem: 3539 +Test: [2000/2387] eta: 0:01:18 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.1405 (0.3785) acc1: 98.4375 (85.4832) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1798 data: 0.0878 max mem: 3539 +Test: [2010/2387] eta: 0:01:16 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.1457 (0.3773) acc1: 98.4375 (85.5480) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1505 data: 0.0563 max mem: 3539 +Test: [2020/2387] eta: 0:01:14 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.1447 (0.3762) acc1: 98.6979 (85.6145) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1724 data: 0.0787 max mem: 3539 +Test: [2030/2387] eta: 0:01:12 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.1414 (0.3750) acc1: 98.9583 (85.6800) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2021 data: 0.1074 max mem: 3540 +Test: [2040/2387] eta: 0:01:10 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.1410 (0.3739) acc1: 98.9583 (85.7443) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1925 data: 0.0974 max mem: 3540 +Test: [2050/2387] eta: 0:01:08 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.1407 (0.3728) acc1: 98.6979 (85.8060) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1657 data: 0.0719 max mem: 3540 +Test: [2060/2387] eta: 0:01:06 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.1721 (0.3719) acc1: 97.1354 (85.8599) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1708 data: 0.0787 max mem: 3540 +Test: [2070/2387] eta: 0:01:03 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.1780 (0.3709) acc1: 96.8750 (85.9161) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1841 data: 0.0929 max mem: 3540 +Test: [2080/2387] eta: 0:01:01 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.1743 (0.3700) acc1: 97.1354 (85.9694) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2090 data: 0.1157 max mem: 3540 +Test: [2090/2387] eta: 0:00:59 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.1785 (0.3691) acc1: 97.1354 (86.0227) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1849 data: 0.0910 max mem: 3540 +Test: [2100/2387] eta: 0:00:57 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.1759 (0.3681) acc1: 97.3958 (86.0787) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1607 data: 0.0678 max mem: 3540 +Test: [2110/2387] eta: 0:00:55 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.1705 (0.3672) acc1: 97.6562 (86.1332) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2099 data: 0.1165 max mem: 3540 +Test: [2120/2387] eta: 0:00:53 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2008 (0.3671) acc1: 97.6562 (86.1499) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.2191 data: 0.1222 max mem: 3540 +Test: [2130/2387] eta: 0:00:51 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2365 (0.3665) acc1: 97.9167 (86.2000) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1499 data: 0.0505 max mem: 3540 +Test: [2140/2387] eta: 0:00:49 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2222 (0.3661) acc1: 96.8750 (86.2265) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1469 data: 0.0516 max mem: 3540 +Test: [2150/2387] eta: 0:00:47 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2554 (0.3656) acc1: 96.8750 (86.2815) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1999 data: 0.1074 max mem: 3540 +Test: [2160/2387] eta: 0:00:45 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2605 (0.3652) acc1: 97.3958 (86.3228) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2029 data: 0.1082 max mem: 3540 +Test: [2170/2387] eta: 0:00:43 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.3105 (0.3657) acc1: 91.4062 (86.2860) acc5: 100.0000 (100.0000) acc1_neg: 71.615 acc5_neg: 100.000 time: 0.1799 data: 0.0846 max mem: 3540 +Test: [2180/2387] eta: 0:00:41 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.5849 (0.3673) acc1: 65.1042 (86.1468) acc5: 100.0000 (100.0000) acc1_neg: 46.875 acc5_neg: 100.000 time: 0.1714 data: 0.0751 max mem: 3540 +Test: [2190/2387] eta: 0:00:39 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.5384 (0.3674) acc1: 69.2708 (86.1372) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1920 data: 0.0981 max mem: 3540 +Test: [2200/2387] eta: 0:00:37 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.3835 (0.3676) acc1: 87.7604 (86.1349) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.2090 data: 0.1164 max mem: 3540 +Test: [2210/2387] eta: 0:00:35 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.3877 (0.3677) acc1: 88.2812 (86.1421) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1878 data: 0.0927 max mem: 3540 +Test: [2220/2387] eta: 0:00:33 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.3789 (0.3677) acc1: 89.0625 (86.1640) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1898 data: 0.0930 max mem: 3540 +Test: [2230/2387] eta: 0:00:31 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.3564 (0.3676) acc1: 90.6250 (86.1882) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1999 data: 0.1049 max mem: 3540 +Test: [2240/2387] eta: 0:00:29 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.3609 (0.3677) acc1: 90.6250 (86.2041) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1772 data: 0.0864 max mem: 3540 +Test: [2250/2387] eta: 0:00:27 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.3837 (0.3678) acc1: 88.5417 (86.1946) acc5: 100.0000 (100.0000) acc1_neg: 73.958 acc5_neg: 100.000 time: 0.1592 data: 0.0693 max mem: 3540 +Test: [2260/2387] eta: 0:00:25 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.3408 (0.3676) acc1: 87.7604 (86.2181) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1411 data: 0.0512 max mem: 3541 +Test: [2270/2387] eta: 0:00:23 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2325 (0.3670) acc1: 97.9167 (86.2484) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1324 data: 0.0426 max mem: 3541 +Test: [2280/2387] eta: 0:00:21 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.1764 (0.3664) acc1: 99.2188 (86.2822) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1254 data: 0.0357 max mem: 3541 +Test: [2290/2387] eta: 0:00:19 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.1769 (0.3656) acc1: 99.4792 (86.3363) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1263 data: 0.0366 max mem: 3541 +Test: [2300/2387] eta: 0:00:17 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2158 (0.3651) acc1: 97.6562 (86.3772) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1261 data: 0.0364 max mem: 3541 +Test: [2310/2387] eta: 0:00:15 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.2928 (0.3652) acc1: 93.2292 (86.3787) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.1245 data: 0.0348 max mem: 3541 +Test: [2320/2387] eta: 0:00:13 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.4246 (0.3658) acc1: 78.9062 (86.3273) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.1227 data: 0.0330 max mem: 3541 +Test: [2330/2387] eta: 0:00:11 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.4942 (0.3669) acc1: 74.4792 (86.2507) acc5: 100.0000 (100.0000) acc1_neg: 53.646 acc5_neg: 100.000 time: 0.1209 data: 0.0312 max mem: 3541 +Test: [2340/2387] eta: 0:00:09 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.4810 (0.3672) acc1: 76.5625 (86.2354) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.1185 data: 0.0289 max mem: 3541 +Test: [2350/2387] eta: 0:00:07 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.4029 (0.3678) acc1: 86.9792 (86.2002) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1147 data: 0.0251 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.5103 (0.3685) acc1: 77.6042 (86.1405) acc5: 100.0000 (100.0000) acc1_neg: 69.531 acc5_neg: 100.000 time: 0.1113 data: 0.0218 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.5491 (0.3693) acc1: 72.3958 (86.0714) acc5: 100.0000 (100.0000) acc1_neg: 52.604 acc5_neg: 100.000 time: 0.1078 data: 0.0183 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.6752 (0.3710) acc1: 52.8646 (85.9239) acc5: 100.0000 (100.0000) acc1_neg: 47.135 acc5_neg: 100.000 time: 0.1065 data: 0.0171 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 67.123 acc5_pos: 100.000 loss: 0.8680 (0.3726) acc1: 39.5833 (85.8081) acc5: 100.0000 (100.0000) acc1_neg: 30.233 acc5_neg: 100.000 time: 0.1020 data: 0.0152 max mem: 3541 +Test: Total time: 0:07:47 (0.1960 s / it) +* Overall Acc@1 85.808 Acc@5 100.000 + target == real Acc@1 58.252 Acc@5 100.000 + target == fake Acc@1 87.743 Acc@5 100.000 + loss 0.373 + PR-AUC 0.9866 ROC-AUC 0.8464 EER 23.07% (thr=0.6294) +Accuracy of the network on the 916267 test images: 85.8% +[rank0]:[W725 15:19:58.397844283 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 15:20:17.028000 69376 torch/distributed/run.py:766] +W0725 15:20:17.028000 69376 torch/distributed/run.py:766] ***************************************** +W0725 15:20:17.028000 69376 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 15:20:17.028000 69376 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 15:20:46.260049516 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 15:20:46.655823722 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 15:20:46.014698008 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 15:20:47.195906263 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 15:20:47.241918240 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 15:20:47.381093204 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank1]:[W725 15:20:47.384270851 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 15:20:47.496660111 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed2_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed2_resampled_n116-040-168_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 0:53:21 acc1_pos: 35.417 acc5_pos: 100.000 loss: 0.9038 (0.9038) acc1: 35.4167 (35.4167) acc5: 100.0000 (100.0000) time: 6.1916 data: 4.4708 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/517] eta: 0:05:26 acc1_pos: 17.448 acc5_pos: 100.000 loss: 0.8689 (0.8564) acc1: 40.8854 (44.0341) acc5: 100.0000 (100.0000) time: 0.6448 data: 0.4068 max mem: 3531 +Test: [ 20/517] eta: 0:03:09 acc1_pos: 16.927 acc5_pos: 100.000 loss: 0.8658 (0.8952) acc1: 40.6250 (39.1369) acc5: 100.0000 (100.0000) time: 0.0907 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:02:37 acc1_pos: 32.552 acc5_pos: 100.000 loss: 0.8759 (0.9022) acc1: 35.9375 (37.6344) acc5: 100.0000 (100.0000) time: 0.1462 data: 0.0558 max mem: 3531 +Test: [ 40/517] eta: 0:02:33 acc1_pos: 32.031 acc5_pos: 100.000 loss: 0.8759 (0.9042) acc1: 35.9375 (37.5254) acc5: 100.0000 (100.0000) time: 0.2612 data: 0.1706 max mem: 3531 +Test: [ 50/517] eta: 0:02:28 acc1_pos: 67.431 acc5_pos: 100.000 loss: 0.7993 (0.8830) acc1: 42.7083 (39.8386) acc5: 100.0000 (100.0000) acc1_neg: 61.719 acc5_neg: 100.000 time: 0.3093 data: 0.2142 max mem: 3531 +Test: [ 60/517] eta: 0:02:33 acc1_pos: 67.431 acc5_pos: 100.000 loss: 0.6137 (0.8303) acc1: 59.8958 (44.8258) acc5: 100.0000 (100.0000) acc1_neg: 67.969 acc5_neg: 100.000 time: 0.3636 data: 0.2681 max mem: 3531 +Test: [ 70/517] eta: 0:02:17 acc1_pos: 67.431 acc5_pos: 100.000 loss: 0.5427 (0.7886) acc1: 69.2708 (49.0317) acc5: 100.0000 (100.0000) acc1_neg: 76.042 acc5_neg: 100.000 time: 0.2841 data: 0.1892 max mem: 3531 +Test: [ 80/517] eta: 0:02:10 acc1_pos: 67.431 acc5_pos: 100.000 loss: 0.5200 (0.7565) acc1: 74.2188 (52.1991) acc5: 100.0000 (100.0000) acc1_neg: 68.750 acc5_neg: 100.000 time: 0.1841 data: 0.0885 max mem: 3531 +Test: [ 90/517] eta: 0:02:05 acc1_pos: 67.431 acc5_pos: 100.000 loss: 0.5483 (0.7363) acc1: 78.3854 (55.2913) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.2399 data: 0.1456 max mem: 3531 +Test: [100/517] eta: 0:01:59 acc1_pos: 67.431 acc5_pos: 100.000 loss: 0.5177 (0.7101) acc1: 81.7708 (58.2637) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.2399 data: 0.1458 max mem: 3531 +Test: [110/517] eta: 0:01:56 acc1_pos: 67.431 acc5_pos: 100.000 loss: 0.3665 (0.6768) acc1: 88.5417 (61.2965) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2499 data: 0.1577 max mem: 3531 +Test: [120/517] eta: 0:01:51 acc1_pos: 67.431 acc5_pos: 100.000 loss: 0.3782 (0.6656) acc1: 85.9375 (62.4247) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.2449 data: 0.1524 max mem: 3531 +Test: [130/517] eta: 0:01:46 acc1_pos: 67.431 acc5_pos: 100.000 loss: 0.5435 (0.6538) acc1: 75.0000 (63.6609) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.2249 data: 0.1284 max mem: 3531 +Test: [140/517] eta: 0:01:43 acc1_pos: 67.431 acc5_pos: 100.000 loss: 0.4648 (0.6383) acc1: 86.9792 (65.6601) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2348 data: 0.1376 max mem: 3531 +Test: [150/517] eta: 0:01:39 acc1_pos: 67.431 acc5_pos: 100.000 loss: 0.4042 (0.6213) acc1: 91.9271 (67.4376) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.2398 data: 0.1410 max mem: 3531 +Test: [160/517] eta: 0:01:36 acc1_pos: 67.431 acc5_pos: 100.000 loss: 0.3563 (0.6040) acc1: 94.0104 (69.1447) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2399 data: 0.1399 max mem: 3531 +Test: [170/517] eta: 0:01:32 acc1_pos: 67.431 acc5_pos: 100.000 loss: 0.3323 (0.5870) acc1: 95.8333 (70.7755) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2349 data: 0.1376 max mem: 3531 +Test: [180/517] eta: 0:01:29 acc1_pos: 67.431 acc5_pos: 100.000 loss: 0.3255 (0.5766) acc1: 95.5729 (71.8304) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.2399 data: 0.1421 max mem: 3531 +Test: [190/517] eta: 0:01:26 acc1_pos: 67.431 acc5_pos: 100.000 loss: 0.3392 (0.5643) acc1: 86.4583 (72.5226) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.2449 data: 0.1465 max mem: 3531 +Test: [200/517] eta: 0:01:23 acc1_pos: 67.431 acc5_pos: 100.000 loss: 0.3539 (0.5563) acc1: 85.1562 (73.1628) acc5: 100.0000 (100.0000) acc1_neg: 77.344 acc5_neg: 100.000 time: 0.2276 data: 0.1322 max mem: 3531 +Test: [210/517] eta: 0:01:20 acc1_pos: 67.431 acc5_pos: 100.000 loss: 0.4768 (0.5542) acc1: 80.4688 (73.3980) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.2449 data: 0.1532 max mem: 3532 +Test: [220/517] eta: 0:01:18 acc1_pos: 67.431 acc5_pos: 100.000 loss: 0.4863 (0.5508) acc1: 78.3854 (73.7015) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.2722 data: 0.1784 max mem: 3532 +Test: [230/517] eta: 0:01:15 acc1_pos: 67.431 acc5_pos: 100.000 loss: 0.4863 (0.5503) acc1: 77.3438 (73.7126) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.2649 data: 0.1705 max mem: 3532 +Test: [240/517] eta: 0:01:12 acc1_pos: 67.431 acc5_pos: 100.000 loss: 0.4543 (0.5445) acc1: 82.0312 (74.4079) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.2549 data: 0.1626 max mem: 3532 +Test: [250/517] eta: 0:01:09 acc1_pos: 67.431 acc5_pos: 100.000 loss: 0.4026 (0.5396) acc1: 91.9271 (75.0809) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2499 data: 0.1574 max mem: 3532 +Test: [260/517] eta: 0:01:07 acc1_pos: 67.431 acc5_pos: 100.000 loss: 0.3997 (0.5348) acc1: 94.2708 (75.7244) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.2649 data: 0.1718 max mem: 3532 +Test: [270/517] eta: 0:01:05 acc1_pos: 67.431 acc5_pos: 100.000 loss: 0.4409 (0.5333) acc1: 85.1562 (75.8476) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.2799 data: 0.1853 max mem: 3532 +Test: [280/517] eta: 0:01:02 acc1_pos: 67.431 acc5_pos: 100.000 loss: 0.4807 (0.5312) acc1: 82.0312 (76.0463) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.2899 data: 0.1944 max mem: 3532 +Test: [290/517] eta: 0:01:00 acc1_pos: 67.431 acc5_pos: 100.000 loss: 0.4112 (0.5261) acc1: 86.4583 (76.5043) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.2849 data: 0.1879 max mem: 3532 +Test: [300/517] eta: 0:00:57 acc1_pos: 67.431 acc5_pos: 100.000 loss: 0.3768 (0.5219) acc1: 91.1458 (77.0063) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.2999 data: 0.2026 max mem: 3532 +Test: [310/517] eta: 0:00:54 acc1_pos: 67.431 acc5_pos: 100.000 loss: 0.4102 (0.5198) acc1: 90.6250 (77.2835) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.2700 data: 0.1736 max mem: 3532 +Test: [320/517] eta: 0:00:52 acc1_pos: 67.431 acc5_pos: 100.000 loss: 0.4325 (0.5156) acc1: 88.5417 (77.6642) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2349 data: 0.1391 max mem: 3532 +Test: [330/517] eta: 0:00:49 acc1_pos: 67.431 acc5_pos: 100.000 loss: 0.3313 (0.5099) acc1: 92.9688 (78.1549) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2549 data: 0.1594 max mem: 3532 +Test: [340/517] eta: 0:00:46 acc1_pos: 67.431 acc5_pos: 100.000 loss: 0.3372 (0.5081) acc1: 93.4896 (78.3197) acc5: 100.0000 (100.0000) acc1_neg: 55.208 acc5_neg: 100.000 time: 0.2509 data: 0.1565 max mem: 3532 +Test: [350/517] eta: 0:00:44 acc1_pos: 67.431 acc5_pos: 100.000 loss: 0.5340 (0.5103) acc1: 72.9167 (78.0716) acc5: 100.0000 (100.0000) acc1_neg: 58.854 acc5_neg: 100.000 time: 0.2549 data: 0.1616 max mem: 3532 +Test: [360/517] eta: 0:00:41 acc1_pos: 67.431 acc5_pos: 100.000 loss: 0.5596 (0.5107) acc1: 72.3958 (78.0420) acc5: 100.0000 (100.0000) acc1_neg: 69.271 acc5_neg: 100.000 time: 0.2489 data: 0.1538 max mem: 3532 +Test: [370/517] eta: 0:00:38 acc1_pos: 67.431 acc5_pos: 100.000 loss: 0.4526 (0.5070) acc1: 83.3333 (78.4668) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2599 data: 0.1620 max mem: 3532 +Test: [380/517] eta: 0:00:36 acc1_pos: 67.431 acc5_pos: 100.000 loss: 0.3834 (0.5050) acc1: 91.9271 (78.6909) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.2563 data: 0.1590 max mem: 3532 +Test: [390/517] eta: 0:00:33 acc1_pos: 67.431 acc5_pos: 100.000 loss: 0.4152 (0.5027) acc1: 86.1979 (78.8723) acc5: 100.0000 (100.0000) acc1_neg: 79.427 acc5_neg: 100.000 time: 0.2454 data: 0.1509 max mem: 3532 +Test: [400/517] eta: 0:00:30 acc1_pos: 67.431 acc5_pos: 100.000 loss: 0.4423 (0.5018) acc1: 84.1146 (78.9309) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.2534 data: 0.1569 max mem: 3532 +Test: [410/517] eta: 0:00:28 acc1_pos: 67.431 acc5_pos: 100.000 loss: 0.5139 (0.5034) acc1: 75.7812 (78.7688) acc5: 100.0000 (100.0000) acc1_neg: 63.021 acc5_neg: 100.000 time: 0.2448 data: 0.1474 max mem: 3532 +Test: [420/517] eta: 0:00:25 acc1_pos: 67.431 acc5_pos: 100.000 loss: 0.6070 (0.5041) acc1: 67.1875 (78.6489) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2449 data: 0.1496 max mem: 3532 +Test: [430/517] eta: 0:00:22 acc1_pos: 67.431 acc5_pos: 100.000 loss: 0.4007 (0.5015) acc1: 84.6354 (78.8579) acc5: 100.0000 (100.0000) acc1_neg: 57.812 acc5_neg: 100.000 time: 0.2395 data: 0.1426 max mem: 3532 +Test: [440/517] eta: 0:00:20 acc1_pos: 67.431 acc5_pos: 100.000 loss: 0.4936 (0.5025) acc1: 78.9062 (78.7120) acc5: 100.0000 (100.0000) acc1_neg: 76.042 acc5_neg: 100.000 time: 0.2356 data: 0.1393 max mem: 3533 +Test: [450/517] eta: 0:00:17 acc1_pos: 67.431 acc5_pos: 100.000 loss: 0.4936 (0.5015) acc1: 79.1667 (78.8497) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.2549 data: 0.1592 max mem: 3533 +Test: [460/517] eta: 0:00:14 acc1_pos: 67.431 acc5_pos: 100.000 loss: 0.4784 (0.5008) acc1: 83.3333 (78.9266) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.2442 data: 0.1470 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 67.431 acc5_pos: 100.000 loss: 0.4784 (0.4998) acc1: 81.5104 (79.0450) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.2505 data: 0.1532 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 67.431 acc5_pos: 100.000 loss: 0.4796 (0.4991) acc1: 81.7708 (79.1488) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2564 data: 0.1611 max mem: 3533 +Test: [490/517] eta: 0:00:07 acc1_pos: 67.431 acc5_pos: 100.000 loss: 0.4045 (0.4969) acc1: 88.0208 (79.3555) acc5: 100.0000 (100.0000) acc1_neg: 78.646 acc5_neg: 100.000 time: 0.2395 data: 0.1466 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 67.431 acc5_pos: 100.000 loss: 0.4957 (0.4999) acc1: 77.6042 (78.9951) acc5: 100.0000 (100.0000) acc1_neg: 55.469 acc5_neg: 100.000 time: 0.2434 data: 0.1486 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 67.431 acc5_pos: 100.000 loss: 0.6067 (0.5022) acc1: 62.5000 (78.7284) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.2154 data: 0.1187 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 67.431 acc5_pos: 100.000 loss: 0.5353 (0.5020) acc1: 75.5208 (78.7615) acc5: 100.0000 (100.0000) acc1_neg: 89.399 acc5_neg: 100.000 time: 0.1703 data: 0.0723 max mem: 3533 +Test: Total time: 0:02:12 (0.2568 s / it) +* Overall Acc@1 78.761 Acc@5 100.000 + target == real Acc@1 38.767 Acc@5 100.000 + target == fake Acc@1 82.910 Acc@5 100.000 + loss 0.502 + PR-AUC 0.9415 ROC-AUC 0.6629 EER 37.98% (thr=0.6167) +Accuracy of the network on the 198427 test images: 78.8% +[rank0]:[W725 15:23:13.060459445 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0725 15:23:31.383000 69948 torch/distributed/run.py:766] +W0725 15:23:31.383000 69948 torch/distributed/run.py:766] ***************************************** +W0725 15:23:31.383000 69948 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 15:23:31.383000 69948 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 15:23:59.643026089 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 15:23:59.898334639 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 15:24:00.131657741 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 15:24:00.243640512 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 15:24:00.319215839 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 15:24:00.337412135 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +[rank2]:[W725 15:24:00.339028416 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 15:24:00.351406464 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed2_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed2_resampled_n116-040-168_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:32:30 acc1_pos: 62.500 acc5_pos: 100.000 loss: 0.6577 (0.6577) acc1: 62.5000 (62.5000) acc5: 100.0000 (100.0000) time: 5.0148 data: 3.4467 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/389] eta: 0:03:23 acc1_pos: 62.500 acc5_pos: 100.000 loss: 0.6729 (0.7324) acc1: 55.4688 (54.4034) acc5: 100.0000 (100.0000) time: 0.5377 data: 0.3138 max mem: 3531 +Test: [ 20/389] eta: 0:02:03 acc1_pos: 39.844 acc5_pos: 100.000 loss: 0.6836 (0.7706) acc1: 54.6875 (50.6448) acc5: 100.0000 (100.0000) time: 0.1012 data: 0.0116 max mem: 3531 +Test: [ 30/389] eta: 0:01:55 acc1_pos: 90.625 acc5_pos: 100.000 loss: 0.5408 (0.6454) acc1: 76.0417 (63.2056) acc5: 100.0000 (100.0000) time: 0.2048 data: 0.1131 max mem: 3531 +Test: [ 40/389] eta: 0:01:43 acc1_pos: 31.771 acc5_pos: 100.000 loss: 0.4487 (0.6284) acc1: 79.6875 (64.7866) acc5: 100.0000 (100.0000) time: 0.2536 data: 0.1608 max mem: 3531 +Test: [ 50/389] eta: 0:01:34 acc1_pos: 97.917 acc5_pos: 100.000 loss: 0.4433 (0.5913) acc1: 86.1979 (69.4598) acc5: 100.0000 (100.0000) time: 0.2066 data: 0.1153 max mem: 3531 +Test: [ 60/389] eta: 0:01:26 acc1_pos: 51.562 acc5_pos: 100.000 loss: 0.3618 (0.5549) acc1: 98.1771 (73.4674) acc5: 100.0000 (100.0000) time: 0.1999 data: 0.1092 max mem: 3531 +Test: [ 70/389] eta: 0:01:17 acc1_pos: 47.917 acc5_pos: 100.000 loss: 0.5464 (0.5849) acc1: 69.2708 (69.9531) acc5: 100.0000 (100.0000) time: 0.1510 data: 0.0575 max mem: 3531 +Test: [ 80/389] eta: 0:01:12 acc1_pos: 89.583 acc5_pos: 100.000 loss: 0.7782 (0.5995) acc1: 47.9167 (68.7532) acc5: 100.0000 (100.0000) time: 0.1480 data: 0.0543 max mem: 3531 +Test: [ 90/389] eta: 0:01:09 acc1_pos: 71.615 acc5_pos: 100.000 loss: 0.4918 (0.5836) acc1: 71.6146 (69.9491) acc5: 100.0000 (100.0000) time: 0.1972 data: 0.1055 max mem: 3531 +Test: [100/389] eta: 0:01:05 acc1_pos: 19.531 acc5_pos: 100.000 loss: 0.5228 (0.5956) acc1: 71.6146 (68.8093) acc5: 100.0000 (100.0000) time: 0.1869 data: 0.0942 max mem: 3531 +Test: [110/389] eta: 0:01:02 acc1_pos: 99.479 acc5_pos: 100.000 loss: 0.7170 (0.6163) acc1: 59.6354 (66.8661) acc5: 100.0000 (100.0000) time: 0.1899 data: 0.0969 max mem: 3531 +Test: [120/389] eta: 0:00:59 acc1_pos: 80.208 acc5_pos: 100.000 loss: 0.5560 (0.6098) acc1: 65.1042 (67.4092) acc5: 100.0000 (100.0000) time: 0.1952 data: 0.1033 max mem: 3531 +Test: [130/389] eta: 0:00:56 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.5560 (0.6188) acc1: 63.8021 (66.3506) acc5: 100.0000 (100.0000) acc1_neg: 79.427 acc5_neg: 100.000 time: 0.1799 data: 0.0869 max mem: 3531 +Test: [140/389] eta: 0:00:53 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.7750 (0.6349) acc1: 34.8958 (63.9406) acc5: 100.0000 (100.0000) acc1_neg: 19.531 acc5_neg: 100.000 time: 0.1796 data: 0.0835 max mem: 3531 +Test: [150/389] eta: 0:00:51 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.9160 (0.6442) acc1: 25.5208 (62.9243) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2099 data: 0.1130 max mem: 3531 +Test: [160/389] eta: 0:00:49 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.6953 (0.6398) acc1: 48.9583 (63.5303) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.2099 data: 0.1104 max mem: 3531 +Test: [170/389] eta: 0:00:45 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.7088 (0.6478) acc1: 48.9583 (62.2548) acc5: 100.0000 (100.0000) acc1_neg: 54.427 acc5_neg: 100.000 time: 0.1576 data: 0.0570 max mem: 3531 +Test: [180/389] eta: 0:00:43 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.8446 (0.6615) acc1: 29.4271 (60.3721) acc5: 100.0000 (100.0000) acc1_neg: 16.927 acc5_neg: 100.000 time: 0.1549 data: 0.0583 max mem: 3531 +Test: [190/389] eta: 0:00:41 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.8073 (0.6628) acc1: 26.8229 (60.0417) acc5: 100.0000 (100.0000) acc1_neg: 40.104 acc5_neg: 100.000 time: 0.1935 data: 0.0998 max mem: 3531 +Test: [200/389] eta: 0:00:39 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.5868 (0.6573) acc1: 68.4896 (60.8002) acc5: 100.0000 (100.0000) acc1_neg: 69.531 acc5_neg: 100.000 time: 0.2160 data: 0.1247 max mem: 3531 +Test: [210/389] eta: 0:00:36 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.5345 (0.6474) acc1: 76.0417 (62.0001) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1792 data: 0.0883 max mem: 3532 +Test: [220/389] eta: 0:00:34 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.3444 (0.6369) acc1: 96.0938 (63.2329) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.1811 data: 0.0898 max mem: 3532 +Test: [230/389] eta: 0:00:32 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.5086 (0.6340) acc1: 78.9062 (63.7119) acc5: 100.0000 (100.0000) acc1_neg: 65.885 acc5_neg: 100.000 time: 0.1994 data: 0.1064 max mem: 3532 +Test: [240/389] eta: 0:00:30 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.5190 (0.6275) acc1: 75.5208 (64.2721) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1841 data: 0.0914 max mem: 3532 +Test: [250/389] eta: 0:00:28 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.4816 (0.6186) acc1: 82.2917 (65.1592) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.2054 data: 0.1124 max mem: 3532 +Test: [260/389] eta: 0:00:26 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.5244 (0.6268) acc1: 76.0417 (64.2990) acc5: 100.0000 (100.0000) acc1_neg: 14.323 acc5_neg: 100.000 time: 0.1939 data: 0.0995 max mem: 3532 +Test: [270/389] eta: 0:00:24 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.9179 (0.6438) acc1: 17.4479 (62.8921) acc5: 100.0000 (100.0000) acc1_neg: 54.948 acc5_neg: 100.000 time: 0.1869 data: 0.0913 max mem: 3532 +Test: [280/389] eta: 0:00:22 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.8733 (0.6503) acc1: 30.9896 (62.1793) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2105 data: 0.1160 max mem: 3532 +Test: [290/389] eta: 0:00:20 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.6911 (0.6497) acc1: 45.0521 (62.1447) acc5: 100.0000 (100.0000) acc1_neg: 24.219 acc5_neg: 100.000 time: 0.2090 data: 0.1177 max mem: 3532 +Test: [300/389] eta: 0:00:18 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.4483 (0.6459) acc1: 77.6042 (62.7025) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1987 data: 0.1036 max mem: 3532 +Test: [310/389] eta: 0:00:16 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.5918 (0.6494) acc1: 59.1146 (62.3032) acc5: 100.0000 (100.0000) acc1_neg: 34.896 acc5_neg: 100.000 time: 0.1884 data: 0.0923 max mem: 3532 +Test: [320/389] eta: 0:00:14 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.6247 (0.6478) acc1: 62.2396 (62.4968) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.2045 data: 0.1120 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.8320 (0.6573) acc1: 38.2812 (61.3018) acc5: 100.0000 (100.0000) acc1_neg: 27.604 acc5_neg: 100.000 time: 0.1949 data: 0.1037 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.7784 (0.6496) acc1: 45.3125 (62.1739) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1681 data: 0.0767 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.3984 (0.6580) acc1: 90.1042 (61.7054) acc5: 100.0000 (100.0000) acc1_neg: 1.042 acc5_neg: 100.000 time: 0.1800 data: 0.0886 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.8834 (0.6687) acc1: 43.2292 (60.9527) acc5: 100.0000 (100.0000) acc1_neg: 43.229 acc5_neg: 100.000 time: 0.1671 data: 0.0762 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.8689 (0.6698) acc1: 43.2292 (60.7887) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1330 data: 0.0428 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.5393 (0.6641) acc1: 69.0104 (61.3011) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1251 data: 0.0337 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 72.393 acc5_pos: 100.000 loss: 0.5393 (0.6624) acc1: 76.3021 (61.4463) acc5: 100.0000 (100.0000) acc1_neg: 46.995 acc5_neg: 100.000 time: 0.1305 data: 0.0343 max mem: 3532 +Test: Total time: 0:01:15 (0.1950 s / it) +* Overall Acc@1 61.446 Acc@5 100.000 + target == real Acc@1 67.180 Acc@5 100.000 + target == fake Acc@1 58.781 Acc@5 100.000 + loss 0.662 + PR-AUC 0.8096 ROC-AUC 0.6843 EER 36.87% (thr=0.4697) +Accuracy of the network on the 149358 test images: 61.4% +[rank0]:[W725 15:25:24.923807597 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 15:25:42.889000 70493 torch/distributed/run.py:766] +W0725 15:25:42.889000 70493 torch/distributed/run.py:766] ***************************************** +W0725 15:25:42.889000 70493 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 15:25:42.889000 70493 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 15:26:09.556507558 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 15:26:09.890573904 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 15:26:09.917236262 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 15:26:09.014353333 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 15:26:10.274525051 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +[rank0]:[W725 15:26:10.323093486 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 15:26:10.328549499 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 15:26:10.350241213 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed2_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed2_resampled_n116-040-168_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:25:15 acc1_pos: 67.448 acc5_pos: 100.000 loss: 0.6799 (0.6799) acc1: 67.4479 (67.4479) acc5: 100.0000 (100.0000) time: 4.5512 data: 2.4751 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/333] eta: 0:02:53 acc1_pos: 84.115 acc5_pos: 100.000 loss: 0.4993 (0.5051) acc1: 78.3854 (77.1307) acc5: 100.0000 (100.0000) time: 0.5367 data: 0.2664 max mem: 3531 +Test: [ 20/333] eta: 0:01:42 acc1_pos: 92.448 acc5_pos: 100.000 loss: 0.4168 (0.4492) acc1: 80.9896 (79.3403) acc5: 100.0000 (100.0000) time: 0.1179 data: 0.0281 max mem: 3531 +Test: [ 30/333] eta: 0:01:23 acc1_pos: 58.073 acc5_pos: 100.000 loss: 0.4961 (0.5367) acc1: 69.5312 (72.5638) acc5: 100.0000 (100.0000) time: 0.1298 data: 0.0392 max mem: 3531 +Test: [ 40/333] eta: 0:01:22 acc1_pos: 57.031 acc5_pos: 100.000 loss: 0.7265 (0.5876) acc1: 57.0312 (68.8516) acc5: 100.0000 (100.0000) time: 0.2330 data: 0.1419 max mem: 3531 +Test: [ 50/333] eta: 0:01:18 acc1_pos: 55.990 acc5_pos: 100.000 loss: 0.7239 (0.6199) acc1: 55.9896 (65.9467) acc5: 100.0000 (100.0000) time: 0.2786 data: 0.1866 max mem: 3531 +Test: [ 60/333] eta: 0:01:09 acc1_pos: 57.552 acc5_pos: 100.000 loss: 0.7239 (0.6371) acc1: 56.2500 (64.4254) acc5: 100.0000 (100.0000) time: 0.2000 data: 0.1059 max mem: 3531 +Test: [ 70/333] eta: 0:01:05 acc1_pos: 49.740 acc5_pos: 100.000 loss: 0.7260 (0.6523) acc1: 57.5521 (63.2042) acc5: 100.0000 (100.0000) time: 0.1747 data: 0.0819 max mem: 3531 +Test: [ 80/333] eta: 0:01:01 acc1_pos: 53.385 acc5_pos: 100.000 loss: 0.7201 (0.6635) acc1: 54.9479 (62.0949) acc5: 100.0000 (100.0000) time: 0.2068 data: 0.1155 max mem: 3531 +Test: [ 90/333] eta: 0:00:58 acc1_pos: 73.698 acc5_pos: 100.000 loss: 0.6847 (0.6578) acc1: 57.5521 (62.8234) acc5: 100.0000 (100.0000) time: 0.2214 data: 0.1280 max mem: 3531 +Test: [100/333] eta: 0:00:54 acc1_pos: 74.219 acc5_pos: 100.000 loss: 0.6072 (0.6515) acc1: 67.1875 (63.2709) acc5: 100.0000 (100.0000) time: 0.1943 data: 0.0996 max mem: 3531 +Test: [110/333] eta: 0:00:51 acc1_pos: 82.031 acc5_pos: 100.000 loss: 0.6126 (0.6462) acc1: 65.3646 (63.7669) acc5: 100.0000 (100.0000) time: 0.1833 data: 0.0903 max mem: 3531 +Test: [120/333] eta: 0:00:48 acc1_pos: 66.927 acc5_pos: 100.000 loss: 0.6112 (0.6429) acc1: 66.9271 (63.9915) acc5: 100.0000 (100.0000) time: 0.2109 data: 0.1189 max mem: 3531 +Test: [130/333] eta: 0:00:46 acc1_pos: 72.396 acc5_pos: 100.000 loss: 0.5843 (0.6374) acc1: 68.4896 (64.5714) acc5: 100.0000 (100.0000) time: 0.2055 data: 0.1133 max mem: 3531 +Test: [140/333] eta: 0:00:43 acc1_pos: 79.427 acc5_pos: 100.000 loss: 0.5484 (0.6308) acc1: 71.3542 (65.1688) acc5: 100.0000 (100.0000) time: 0.1925 data: 0.1015 max mem: 3531 +Test: [150/333] eta: 0:00:40 acc1_pos: 81.510 acc5_pos: 100.000 loss: 0.5507 (0.6272) acc1: 71.0938 (65.5232) acc5: 100.0000 (100.0000) time: 0.1960 data: 0.1044 max mem: 3531 +Test: [160/333] eta: 0:00:38 acc1_pos: 55.592 acc5_pos: 100.000 loss: 0.5903 (0.6334) acc1: 67.7083 (64.9554) acc5: 100.0000 (100.0000) acc1_neg: 42.448 acc5_neg: 100.000 time: 0.2154 data: 0.1235 max mem: 3531 +Test: [170/333] eta: 0:00:36 acc1_pos: 55.592 acc5_pos: 100.000 loss: 0.7924 (0.6446) acc1: 48.4375 (63.8371) acc5: 100.0000 (100.0000) acc1_neg: 49.479 acc5_neg: 100.000 time: 0.2120 data: 0.1213 max mem: 3531 +Test: [180/333] eta: 0:00:33 acc1_pos: 55.592 acc5_pos: 100.000 loss: 0.7513 (0.6459) acc1: 51.3021 (63.5230) acc5: 100.0000 (100.0000) acc1_neg: 58.594 acc5_neg: 100.000 time: 0.1733 data: 0.0781 max mem: 3531 +Test: [190/333] eta: 0:00:31 acc1_pos: 55.592 acc5_pos: 100.000 loss: 0.6636 (0.6452) acc1: 58.8542 (63.4653) acc5: 100.0000 (100.0000) acc1_neg: 55.469 acc5_neg: 100.000 time: 0.2160 data: 0.1184 max mem: 3531 +Test: [200/333] eta: 0:00:28 acc1_pos: 55.592 acc5_pos: 100.000 loss: 0.6899 (0.6481) acc1: 56.2500 (63.1478) acc5: 100.0000 (100.0000) acc1_neg: 53.646 acc5_neg: 100.000 time: 0.2053 data: 0.1111 max mem: 3531 +Test: [210/333] eta: 0:00:26 acc1_pos: 55.592 acc5_pos: 100.000 loss: 0.7737 (0.6545) acc1: 51.0417 (62.5222) acc5: 100.0000 (100.0000) acc1_neg: 58.594 acc5_neg: 100.000 time: 0.1357 data: 0.0408 max mem: 3532 +Test: [220/333] eta: 0:00:23 acc1_pos: 55.592 acc5_pos: 100.000 loss: 0.7914 (0.6608) acc1: 48.1771 (61.9014) acc5: 100.0000 (100.0000) acc1_neg: 56.771 acc5_neg: 100.000 time: 0.1700 data: 0.0757 max mem: 3532 +Test: [230/333] eta: 0:00:21 acc1_pos: 55.592 acc5_pos: 100.000 loss: 0.6562 (0.6593) acc1: 57.8125 (61.9938) acc5: 100.0000 (100.0000) acc1_neg: 57.812 acc5_neg: 100.000 time: 0.2154 data: 0.1228 max mem: 3532 +Test: [240/333] eta: 0:00:19 acc1_pos: 55.592 acc5_pos: 100.000 loss: 0.6572 (0.6610) acc1: 58.5938 (61.7285) acc5: 100.0000 (100.0000) acc1_neg: 55.469 acc5_neg: 100.000 time: 0.1995 data: 0.1033 max mem: 3532 +Test: [250/333] eta: 0:00:17 acc1_pos: 55.592 acc5_pos: 100.000 loss: 0.6984 (0.6617) acc1: 56.7708 (61.5693) acc5: 100.0000 (100.0000) acc1_neg: 56.771 acc5_neg: 100.000 time: 0.1727 data: 0.0772 max mem: 3532 +Test: [260/333] eta: 0:00:15 acc1_pos: 55.592 acc5_pos: 100.000 loss: 0.6702 (0.6619) acc1: 58.0729 (61.4913) acc5: 100.0000 (100.0000) acc1_neg: 56.250 acc5_neg: 100.000 time: 0.2049 data: 0.1116 max mem: 3532 +Test: [270/333] eta: 0:00:13 acc1_pos: 55.592 acc5_pos: 100.000 loss: 0.6944 (0.6638) acc1: 56.7708 (61.2902) acc5: 100.0000 (100.0000) acc1_neg: 57.552 acc5_neg: 100.000 time: 0.2197 data: 0.1258 max mem: 3532 +Test: [280/333] eta: 0:00:10 acc1_pos: 55.592 acc5_pos: 100.000 loss: 0.7402 (0.6702) acc1: 51.5625 (60.6715) acc5: 100.0000 (100.0000) acc1_neg: 47.135 acc5_neg: 100.000 time: 0.1548 data: 0.0582 max mem: 3532 +Test: [290/333] eta: 0:00:08 acc1_pos: 55.592 acc5_pos: 100.000 loss: 0.9165 (0.6784) acc1: 37.7604 (59.9307) acc5: 100.0000 (100.0000) acc1_neg: 53.906 acc5_neg: 100.000 time: 0.1449 data: 0.0489 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 55.592 acc5_pos: 100.000 loss: 0.8009 (0.6816) acc1: 48.4375 (59.6527) acc5: 100.0000 (100.0000) acc1_neg: 51.302 acc5_neg: 100.000 time: 0.2049 data: 0.1115 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 55.592 acc5_pos: 100.000 loss: 0.7180 (0.6804) acc1: 53.6458 (59.7091) acc5: 100.0000 (100.0000) acc1_neg: 56.510 acc5_neg: 100.000 time: 0.1969 data: 0.1040 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 55.592 acc5_pos: 100.000 loss: 0.6339 (0.6801) acc1: 60.6771 (59.6744) acc5: 100.0000 (100.0000) acc1_neg: 39.323 acc5_neg: 100.000 time: 0.1583 data: 0.0676 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 55.592 acc5_pos: 100.000 loss: 0.8811 (0.6874) acc1: 44.0104 (59.1272) acc5: 100.0000 (100.0000) acc1_neg: 44.010 acc5_neg: 100.000 time: 0.1163 data: 0.0265 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 55.592 acc5_pos: 100.000 loss: 0.9064 (0.6888) acc1: 41.9718 (59.0232) acc5: 100.0000 (100.0000) acc1_neg: 41.972 acc5_neg: 100.000 time: 0.1195 data: 0.0265 max mem: 3532 +Test: Total time: 0:01:06 (0.2004 s / it) +* Overall Acc@1 59.023 Acc@5 100.000 + target == real Acc@1 65.719 Acc@5 100.000 + target == fake Acc@1 53.063 Acc@5 100.000 + loss 0.689 + PR-AUC 0.6454 ROC-AUC 0.6395 EER 40.27% (thr=0.4617) +Accuracy of the network on the 127843 test images: 59.0% +[rank0]:[W725 15:27:26.492802328 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 15:27:45.210000 71050 torch/distributed/run.py:766] +W0725 15:27:45.210000 71050 torch/distributed/run.py:766] ***************************************** +W0725 15:27:45.210000 71050 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 15:27:45.210000 71050 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 15:28:12.553560941 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 15:28:12.761672569 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 15:28:13.222473517 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 15:28:13.231347342 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 15:28:13.280275053 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +| distributed init (rank 6): env:// +[rank0]:[W725 15:28:13.381062088 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 15:28:13.391906936 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank1]:[W725 15:28:13.393985599 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed2_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed2_resampled_n116-040-168_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:20:02 acc1_pos: 96.094 acc5_pos: 100.000 loss: 0.2088 (0.2088) acc1: 96.0938 (96.0938) acc5: 100.0000 (100.0000) time: 5.1160 data: 3.1718 max mem: 3531 +Test: [ 10/235] eta: 0:02:03 acc1_pos: 95.052 acc5_pos: 100.000 loss: 0.2141 (0.2165) acc1: 95.8333 (95.8097) acc5: 100.0000 (100.0000) time: 0.5492 data: 0.2909 max mem: 3531 +Test: [ 20/235] eta: 0:01:15 acc1_pos: 94.792 acc5_pos: 100.000 loss: 0.2142 (0.2148) acc1: 95.8333 (95.8581) acc5: 100.0000 (100.0000) time: 0.1111 data: 0.0212 max mem: 3531 +Test: [ 30/235] eta: 0:01:04 acc1_pos: 96.615 acc5_pos: 100.000 loss: 0.2142 (0.2136) acc1: 95.8333 (95.8585) acc5: 100.0000 (100.0000) time: 0.1844 data: 0.0907 max mem: 3531 +Test: [ 40/235] eta: 0:00:56 acc1_pos: 94.010 acc5_pos: 100.000 loss: 0.2147 (0.2144) acc1: 95.8333 (95.8079) acc5: 100.0000 (100.0000) time: 0.2295 data: 0.1347 max mem: 3531 +Test: [ 50/235] eta: 0:00:50 acc1_pos: 93.750 acc5_pos: 100.000 loss: 0.2178 (0.2163) acc1: 95.3125 (95.6342) acc5: 100.0000 (100.0000) time: 0.2035 data: 0.1120 max mem: 3531 +Test: [ 60/235] eta: 0:00:43 acc1_pos: 93.490 acc5_pos: 100.000 loss: 0.2272 (0.2183) acc1: 95.0521 (95.5217) acc5: 100.0000 (100.0000) time: 0.1621 data: 0.0675 max mem: 3531 +Test: [ 70/235] eta: 0:00:39 acc1_pos: 98.462 acc5_pos: 100.000 loss: 0.2361 (0.3738) acc1: 93.2292 (86.0842) acc5: 100.0000 (100.0000) acc1_neg: 18.490 acc5_neg: 100.000 time: 0.1580 data: 0.0632 max mem: 3531 +Test: [ 80/235] eta: 0:00:35 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4552 (0.5091) acc1: 19.7917 (77.9096) acc5: 100.0000 (100.0000) acc1_neg: 16.667 acc5_neg: 100.000 time: 0.1777 data: 0.0860 max mem: 3531 +Test: [ 90/235] eta: 0:00:33 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4706 (0.6151) acc1: 19.7917 (71.4801) acc5: 100.0000 (100.0000) acc1_neg: 17.188 acc5_neg: 100.000 time: 0.2132 data: 0.1207 max mem: 3531 +Test: [100/235] eta: 0:00:31 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4706 (0.6970) acc1: 20.3125 (66.5017) acc5: 100.0000 (100.0000) acc1_neg: 20.312 acc5_neg: 100.000 time: 0.2399 data: 0.1450 max mem: 3531 +Test: [110/235] eta: 0:00:28 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4579 (0.7653) acc1: 20.3125 (62.3358) acc5: 100.0000 (100.0000) acc1_neg: 23.958 acc5_neg: 100.000 time: 0.2049 data: 0.1075 max mem: 3531 +Test: [120/235] eta: 0:00:26 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4420 (0.8201) acc1: 18.4896 (58.6906) acc5: 100.0000 (100.0000) acc1_neg: 18.490 acc5_neg: 100.000 time: 0.2001 data: 0.1057 max mem: 3531 +Test: [130/235] eta: 0:00:23 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4137 (0.8643) acc1: 17.7083 (55.5741) acc5: 100.0000 (100.0000) acc1_neg: 16.667 acc5_neg: 100.000 time: 0.2149 data: 0.1221 max mem: 3531 +Test: [140/235] eta: 0:00:21 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4141 (0.9050) acc1: 16.9271 (52.7999) acc5: 100.0000 (100.0000) acc1_neg: 16.927 acc5_neg: 100.000 time: 0.2368 data: 0.1417 max mem: 3531 +Test: [150/235] eta: 0:00:19 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4141 (0.9375) acc1: 17.7083 (50.5777) acc5: 100.0000 (100.0000) acc1_neg: 20.833 acc5_neg: 100.000 time: 0.2417 data: 0.1478 max mem: 3531 +Test: [160/235] eta: 0:00:16 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4050 (0.9672) acc1: 18.7500 (48.5782) acc5: 100.0000 (100.0000) acc1_neg: 20.312 acc5_neg: 100.000 time: 0.2027 data: 0.1086 max mem: 3531 +Test: [170/235] eta: 0:00:14 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.3626 (0.9879) acc1: 20.3125 (47.1324) acc5: 100.0000 (100.0000) acc1_neg: 24.479 acc5_neg: 100.000 time: 0.2107 data: 0.1168 max mem: 3531 +Test: [180/235] eta: 0:00:12 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.3052 (1.0054) acc1: 25.0000 (45.9700) acc5: 100.0000 (100.0000) acc1_neg: 25.260 acc5_neg: 100.000 time: 0.2116 data: 0.1187 max mem: 3531 +Test: [190/235] eta: 0:00:10 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.2882 (1.0199) acc1: 25.7812 (44.9239) acc5: 100.0000 (100.0000) acc1_neg: 28.906 acc5_neg: 100.000 time: 0.2070 data: 0.1141 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.3025 (1.0348) acc1: 25.2604 (43.9107) acc5: 100.0000 (100.0000) acc1_neg: 23.438 acc5_neg: 100.000 time: 0.2147 data: 0.1226 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.3020 (1.0471) acc1: 25.2604 (43.0231) acc5: 100.0000 (100.0000) acc1_neg: 25.521 acc5_neg: 100.000 time: 0.1925 data: 0.1007 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.3020 (1.0593) acc1: 25.2604 (42.2040) acc5: 100.0000 (100.0000) acc1_neg: 24.740 acc5_neg: 100.000 time: 0.1689 data: 0.0778 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.2954 (1.0693) acc1: 26.3021 (41.5618) acc5: 100.0000 (100.0000) acc1_neg: 29.427 acc5_neg: 100.000 time: 0.1388 data: 0.0485 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.2941 (1.0740) acc1: 26.3021 (41.3411) acc5: 100.0000 (100.0000) acc1_neg: 30.435 acc5_neg: 100.000 time: 0.1213 data: 0.0350 max mem: 3532 +Test: Total time: 0:00:49 (0.2126 s / it) +* Overall Acc@1 41.341 Acc@5 100.000 + target == real Acc@1 95.535 Acc@5 100.000 + target == fake Acc@1 21.740 Acc@5 100.000 + loss 1.074 + PR-AUC 0.8712 ROC-AUC 0.7197 EER 34.10% (thr=0.1808) +Accuracy of the network on the 89879 test images: 41.3% +[rank0]:[W725 15:29:11.430157398 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 15:29:28.828000 71606 torch/distributed/run.py:766] +W0725 15:29:28.828000 71606 torch/distributed/run.py:766] ***************************************** +W0725 15:29:28.828000 71606 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 15:29:28.828000 71606 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 15:29:56.440311415 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 15:29:56.826494480 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 15:29:56.980401382 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 15:29:57.127917105 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +| distributed init (rank 1): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 15:29:57.362036050 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 15:29:57.368259639 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank1]:[W725 15:29:57.368283249 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 15:29:57.373317580 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed2_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed2_resampled_n116-040-168_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:26:12 acc1_pos: 76.823 acc5_pos: 100.000 loss: 0.4979 (0.4979) acc1: 76.8229 (76.8229) acc5: 100.0000 (100.0000) time: 5.8461 data: 4.2537 max mem: 3531 +Test: [ 10/269] eta: 0:02:39 acc1_pos: 63.542 acc5_pos: 100.000 loss: 0.4913 (0.5103) acc1: 82.0312 (78.4091) acc5: 100.0000 (100.0000) time: 0.6173 data: 0.3871 max mem: 3531 +Test: [ 20/269] eta: 0:01:31 acc1_pos: 89.844 acc5_pos: 100.000 loss: 0.4626 (0.4775) acc1: 82.5521 (82.0933) acc5: 100.0000 (100.0000) time: 0.0943 data: 0.0005 max mem: 3531 +Test: [ 30/269] eta: 0:01:24 acc1_pos: 97.396 acc5_pos: 100.000 loss: 0.3988 (0.4587) acc1: 90.1042 (83.6778) acc5: 100.0000 (100.0000) time: 0.2076 data: 0.1140 max mem: 3531 +Test: [ 40/269] eta: 0:01:15 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.4973 (0.5050) acc1: 73.1771 (78.2584) acc5: 100.0000 (100.0000) acc1_neg: 57.292 acc5_neg: 100.000 time: 0.2930 data: 0.2004 max mem: 3531 +Test: [ 50/269] eta: 0:01:06 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.7024 (0.5684) acc1: 53.1250 (70.8027) acc5: 100.0000 (100.0000) acc1_neg: 37.760 acc5_neg: 100.000 time: 0.2298 data: 0.1382 max mem: 3531 +Test: [ 60/269] eta: 0:00:56 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.9195 (0.6290) acc1: 29.6875 (63.9174) acc5: 100.0000 (100.0000) acc1_neg: 21.875 acc5_neg: 100.000 time: 0.1468 data: 0.0537 max mem: 3531 +Test: [ 70/269] eta: 0:00:51 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.8519 (0.6614) acc1: 30.2083 (60.0059) acc5: 100.0000 (100.0000) acc1_neg: 40.885 acc5_neg: 100.000 time: 0.1494 data: 0.0550 max mem: 3531 +Test: [ 80/269] eta: 0:00:47 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.8121 (0.6761) acc1: 40.1042 (58.0183) acc5: 100.0000 (100.0000) acc1_neg: 47.917 acc5_neg: 100.000 time: 0.1916 data: 0.0983 max mem: 3531 +Test: [ 90/269] eta: 0:00:44 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.7804 (0.6888) acc1: 41.6667 (56.3845) acc5: 100.0000 (100.0000) acc1_neg: 29.688 acc5_neg: 100.000 time: 0.1925 data: 0.0988 max mem: 3531 +Test: [100/269] eta: 0:00:41 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.7961 (0.7050) acc1: 39.3229 (54.4116) acc5: 100.0000 (100.0000) acc1_neg: 21.094 acc5_neg: 100.000 time: 0.2099 data: 0.1174 max mem: 3531 +Test: [110/269] eta: 0:00:37 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.8933 (0.7304) acc1: 31.2500 (51.6352) acc5: 100.0000 (100.0000) acc1_neg: 24.219 acc5_neg: 100.000 time: 0.1893 data: 0.0974 max mem: 3531 +Test: [120/269] eta: 0:00:34 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.9582 (0.7462) acc1: 27.6042 (50.2690) acc5: 100.0000 (100.0000) acc1_neg: 21.354 acc5_neg: 100.000 time: 0.1581 data: 0.0638 max mem: 3531 +Test: [130/269] eta: 0:00:31 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.8581 (0.7513) acc1: 32.2917 (49.6283) acc5: 100.0000 (100.0000) acc1_neg: 43.490 acc5_neg: 100.000 time: 0.1805 data: 0.0866 max mem: 3531 +Test: [140/269] eta: 0:00:29 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.8440 (0.7600) acc1: 34.8958 (48.5409) acc5: 100.0000 (100.0000) acc1_neg: 34.896 acc5_neg: 100.000 time: 0.1999 data: 0.1081 max mem: 3531 +Test: [150/269] eta: 0:00:26 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.8821 (0.7707) acc1: 34.3750 (47.4907) acc5: 100.0000 (100.0000) acc1_neg: 28.125 acc5_neg: 100.000 time: 0.1978 data: 0.1067 max mem: 3531 +Test: [160/269] eta: 0:00:24 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.8780 (0.7760) acc1: 35.1562 (46.8653) acc5: 100.0000 (100.0000) acc1_neg: 54.427 acc5_neg: 100.000 time: 0.1876 data: 0.0954 max mem: 3531 +Test: [170/269] eta: 0:00:21 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.7564 (0.7749) acc1: 43.4896 (46.8537) acc5: 100.0000 (100.0000) acc1_neg: 32.552 acc5_neg: 100.000 time: 0.1821 data: 0.0888 max mem: 3531 +Test: [180/269] eta: 0:00:19 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.7730 (0.7777) acc1: 43.7500 (46.4880) acc5: 100.0000 (100.0000) acc1_neg: 52.604 acc5_neg: 100.000 time: 0.1886 data: 0.0960 max mem: 3531 +Test: [190/269] eta: 0:00:16 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.8448 (0.7834) acc1: 34.1146 (45.9097) acc5: 100.0000 (100.0000) acc1_neg: 34.115 acc5_neg: 100.000 time: 0.1836 data: 0.0923 max mem: 3531 +Test: [200/269] eta: 0:00:14 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.9020 (0.7917) acc1: 25.5208 (44.9148) acc5: 100.0000 (100.0000) acc1_neg: 18.490 acc5_neg: 100.000 time: 0.1809 data: 0.0887 max mem: 3531 +Test: [210/269] eta: 0:00:12 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.9085 (0.8010) acc1: 25.7812 (44.2104) acc5: 100.0000 (100.0000) acc1_neg: 31.510 acc5_neg: 100.000 time: 0.1822 data: 0.0894 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.9020 (0.8029) acc1: 28.6458 (44.1082) acc5: 100.0000 (100.0000) acc1_neg: 58.594 acc5_neg: 100.000 time: 0.1899 data: 0.0983 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.7146 (0.7970) acc1: 48.1771 (44.6237) acc5: 100.0000 (100.0000) acc1_neg: 56.510 acc5_neg: 100.000 time: 0.1819 data: 0.0910 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.7562 (0.8036) acc1: 40.3646 (43.8332) acc5: 100.0000 (100.0000) acc1_neg: 28.906 acc5_neg: 100.000 time: 0.1823 data: 0.0882 max mem: 3532 +Test: [250/269] eta: 0:00:03 acc1_pos: 87.931 acc5_pos: 100.000 loss: 0.9677 (0.8174) acc1: 19.0104 (42.8463) acc5: 100.0000 (100.0000) acc1_neg: 7.031 acc5_neg: 100.000 time: 0.1895 data: 0.0947 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 87.931 acc5_pos: 100.000 loss: 1.1781 (0.8309) acc1: 9.8958 (41.6567) acc5: 100.0000 (100.0000) acc1_neg: 21.875 acc5_neg: 100.000 time: 0.1490 data: 0.0580 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 87.931 acc5_pos: 100.000 loss: 1.0882 (0.8343) acc1: 14.8438 (41.1616) acc5: 100.0000 (100.0000) acc1_neg: 19.948 acc5_neg: 100.000 time: 0.1170 data: 0.0235 max mem: 3532 +Test: Total time: 0:00:54 (0.2023 s / it) +* Overall Acc@1 41.162 Acc@5 100.000 + target == real Acc@1 83.884 Acc@5 100.000 + target == fake Acc@1 35.331 Acc@5 100.000 + loss 0.834 + PR-AUC 0.9317 ROC-AUC 0.6731 EER 37.58% (thr=0.3723) +Accuracy of the network on the 103293 test images: 41.2% +[rank0]:[W725 15:31:01.445374726 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-25T15:31:08Z] eval_done tag=fixed500k_fake_n11_seed2_resampled +[2026-07-25T15:31:08Z] eval_start tag=fixed500k_fake_n16_seed1_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed1_resampled_n116-041-153_20260725_112559/best_checkpoint.pth +started_utc=2026-07-25T15:31:08Z +hostname=n116-042-186 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed1_resampled_n116-041-153_20260725_112559/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n16_seed1_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0725 15:31:20.323000 72164 torch/distributed/run.py:766] +W0725 15:31:20.323000 72164 torch/distributed/run.py:766] ***************************************** +W0725 15:31:20.323000 72164 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 15:31:20.323000 72164 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W725 15:31:47.078402623 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 15:31:47.532230647 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 15:31:47.709739931 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 15:31:47.025850006 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 15:31:48.064692297 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W725 15:31:48.064743771 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 15:31:48.315676852 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W725 15:31:48.315702047 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n16_seed1_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed1_resampled_n116-041-153_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:17:02 acc1_pos: 86.198 acc5_pos: 100.000 loss: 0.3305 (0.3305) acc1: 86.1979 (86.1979) acc5: 100.0000 (100.0000) time: 4.3701 data: 2.3608 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/234] eta: 0:01:51 acc1_pos: 92.708 acc5_pos: 100.000 loss: 0.3569 (0.4288) acc1: 92.4479 (84.5170) acc5: 100.0000 (100.0000) time: 0.4969 data: 0.2231 max mem: 3531 +Test: [ 20/234] eta: 0:01:05 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.3569 (0.4167) acc1: 92.1875 (84.3998) acc5: 100.0000 (100.0000) time: 0.1010 data: 0.0061 max mem: 3531 +Test: [ 30/234] eta: 0:00:48 acc1_pos: 74.479 acc5_pos: 100.000 loss: 0.2203 (0.3641) acc1: 95.3125 (87.0464) acc5: 100.0000 (100.0000) time: 0.0931 data: 0.0016 max mem: 3531 +Test: [ 40/234] eta: 0:00:45 acc1_pos: 99.219 acc5_pos: 100.000 loss: 0.2099 (0.3738) acc1: 96.3542 (85.0356) acc5: 100.0000 (100.0000) time: 0.1591 data: 0.0651 max mem: 3531 +Test: [ 50/234] eta: 0:00:46 acc1_pos: 98.958 acc5_pos: 100.000 loss: 0.1171 (0.3225) acc1: 98.9583 (87.7911) acc5: 100.0000 (100.0000) time: 0.2749 data: 0.1770 max mem: 3531 +Test: [ 60/234] eta: 0:00:43 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.1078 (0.2872) acc1: 99.4792 (89.7584) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2778 data: 0.1801 max mem: 3531 +Test: [ 70/234] eta: 0:00:37 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.1115 (0.3155) acc1: 99.7396 (87.6834) acc5: 100.0000 (100.0000) acc1_neg: 60.417 acc5_neg: 100.000 time: 0.1748 data: 0.0800 max mem: 3531 +Test: [ 80/234] eta: 0:00:35 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.6515 (0.3481) acc1: 65.1042 (85.5999) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.1798 data: 0.0853 max mem: 3531 +Test: [ 90/234] eta: 0:00:32 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.4162 (0.3409) acc1: 79.1667 (86.0634) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.2111 data: 0.1171 max mem: 3531 +Test: [100/234] eta: 0:00:29 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.3603 (0.3598) acc1: 83.3333 (84.7979) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1813 data: 0.0844 max mem: 3531 +Test: [110/234] eta: 0:00:27 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.5550 (0.3786) acc1: 70.3125 (83.5797) acc5: 100.0000 (100.0000) acc1_neg: 56.510 acc5_neg: 100.000 time: 0.1909 data: 0.0963 max mem: 3531 +Test: [120/234] eta: 0:00:24 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.5515 (0.4040) acc1: 71.0938 (81.9861) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1835 data: 0.0923 max mem: 3531 +Test: [130/234] eta: 0:00:21 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.5795 (0.4207) acc1: 67.9688 (80.7431) acc5: 100.0000 (100.0000) acc1_neg: 71.354 acc5_neg: 100.000 time: 0.1629 data: 0.0709 max mem: 3531 +Test: [140/234] eta: 0:00:19 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.5903 (0.4416) acc1: 67.9688 (79.7540) acc5: 100.0000 (100.0000) acc1_neg: 43.229 acc5_neg: 100.000 time: 0.1649 data: 0.0730 max mem: 3531 +Test: [150/234] eta: 0:00:17 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.5623 (0.4555) acc1: 68.4896 (78.9269) acc5: 100.0000 (100.0000) acc1_neg: 70.573 acc5_neg: 100.000 time: 0.1915 data: 0.0993 max mem: 3531 +Test: [160/234] eta: 0:00:15 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.5623 (0.4620) acc1: 69.7917 (78.4129) acc5: 100.0000 (100.0000) acc1_neg: 67.969 acc5_neg: 100.000 time: 0.2049 data: 0.1105 max mem: 3531 +Test: [170/234] eta: 0:00:13 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.5725 (0.4695) acc1: 69.5312 (77.8021) acc5: 100.0000 (100.0000) acc1_neg: 63.542 acc5_neg: 100.000 time: 0.1765 data: 0.0819 max mem: 3531 +Test: [180/234] eta: 0:00:11 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.5913 (0.4851) acc1: 67.9688 (76.8805) acc5: 100.0000 (100.0000) acc1_neg: 32.552 acc5_neg: 100.000 time: 0.1748 data: 0.0828 max mem: 3531 +Test: [190/234] eta: 0:00:08 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.7993 (0.5038) acc1: 58.0729 (75.8140) acc5: 100.0000 (100.0000) acc1_neg: 58.073 acc5_neg: 100.000 time: 0.1884 data: 0.0972 max mem: 3531 +Test: [200/234] eta: 0:00:06 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.7993 (0.4972) acc1: 59.1146 (76.2490) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1787 data: 0.0869 max mem: 3531 +Test: [210/234] eta: 0:00:04 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.1113 (0.4796) acc1: 99.7396 (77.3573) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1778 data: 0.0839 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.1515 (0.4669) acc1: 98.9583 (78.1910) acc5: 100.0000 (100.0000) acc1_neg: 69.531 acc5_neg: 100.000 time: 0.1628 data: 0.0685 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.5931 (0.5085) acc1: 16.4062 (75.4216) acc5: 100.0000 (100.0000) acc1_neg: 14.062 acc5_neg: 100.000 time: 0.1392 data: 0.0476 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 99.310 acc5_pos: 100.000 loss: 1.4193 (0.5203) acc1: 15.3646 (74.7306) acc5: 100.0000 (100.0000) acc1_neg: 15.175 acc5_neg: 100.000 time: 0.1415 data: 0.0490 max mem: 3532 +Test: Total time: 0:00:45 (0.1950 s / it) +* Overall Acc@1 74.731 Acc@5 100.000 + target == real Acc@1 87.876 Acc@5 100.000 + target == fake Acc@1 71.026 Acc@5 100.000 + loss 0.520 + PR-AUC 0.9628 ROC-AUC 0.8761 EER 21.07% (thr=0.3728) +Accuracy of the network on the 89729 test images: 74.7% +[rank0]:[W725 15:32:52.223629523 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 15:33:09.573000 72720 torch/distributed/run.py:766] +W0725 15:33:09.573000 72720 torch/distributed/run.py:766] ***************************************** +W0725 15:33:09.573000 72720 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 15:33:09.573000 72720 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 15:33:35.838476735 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 15:33:36.461584002 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 15:33:36.738494175 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 15:33:36.791749989 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 15:33:36.920103939 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 15:33:36.038734691 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W725 15:33:37.191898341 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank6]:[W725 15:33:37.195108323 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n16_seed1_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed1_resampled_n116-041-153_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:22:45 acc1_pos: 65.104 acc5_pos: 100.000 loss: 0.6117 (0.6117) acc1: 65.1042 (65.1042) acc5: 100.0000 (100.0000) time: 6.6045 data: 4.5188 max mem: 3531 +Test: [ 10/2387] eta: 0:27:02 acc1_pos: 53.906 acc5_pos: 100.000 loss: 0.7843 (0.7658) acc1: 53.9062 (52.7699) acc5: 100.0000 (100.0000) time: 0.6824 data: 0.4112 max mem: 3531 +Test: [ 20/2387] eta: 0:15:57 acc1_pos: 50.781 acc5_pos: 100.000 loss: 0.7717 (0.7530) acc1: 52.8646 (54.5759) acc5: 100.0000 (100.0000) time: 0.0943 data: 0.0005 max mem: 3531 +Test: [ 30/2387] eta: 0:13:40 acc1_pos: 32.552 acc5_pos: 100.000 loss: 0.8024 (0.7838) acc1: 47.1354 (51.4617) acc5: 100.0000 (100.0000) time: 0.1641 data: 0.0684 max mem: 3531 +Test: [ 40/2387] eta: 0:12:29 acc1_pos: 54.948 acc5_pos: 100.000 loss: 0.8207 (0.7930) acc1: 45.5729 (50.4637) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1358 max mem: 3531 +Test: [ 50/2387] eta: 0:12:03 acc1_pos: 58.333 acc5_pos: 100.000 loss: 0.7127 (0.7697) acc1: 52.0833 (52.8901) acc5: 100.0000 (100.0000) time: 0.2499 data: 0.1554 max mem: 3531 +Test: [ 60/2387] eta: 0:11:29 acc1_pos: 66.406 acc5_pos: 100.000 loss: 0.6632 (0.7585) acc1: 58.3333 (53.9788) acc5: 100.0000 (100.0000) time: 0.2499 data: 0.1537 max mem: 3531 +Test: [ 70/2387] eta: 0:11:01 acc1_pos: 65.365 acc5_pos: 100.000 loss: 0.7117 (0.7520) acc1: 58.3333 (54.8782) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1280 max mem: 3531 +Test: [ 80/2387] eta: 0:10:46 acc1_pos: 96.354 acc5_pos: 100.000 loss: 0.6205 (0.7123) acc1: 63.0208 (58.1983) acc5: 100.0000 (100.0000) time: 0.2309 data: 0.1370 max mem: 3531 +Test: [ 90/2387] eta: 0:10:38 acc1_pos: 95.573 acc5_pos: 100.000 loss: 0.3288 (0.6656) acc1: 90.6250 (62.1394) acc5: 100.0000 (100.0000) time: 0.2499 data: 0.1566 max mem: 3531 +Test: [ 100/2387] eta: 0:10:27 acc1_pos: 99.219 acc5_pos: 100.000 loss: 0.2908 (0.6341) acc1: 90.6250 (64.7355) acc5: 100.0000 (100.0000) time: 0.2497 data: 0.1565 max mem: 3531 +Test: [ 110/2387] eta: 0:10:17 acc1_pos: 82.552 acc5_pos: 100.000 loss: 0.3306 (0.6056) acc1: 89.0625 (67.0890) acc5: 100.0000 (100.0000) time: 0.2413 data: 0.1459 max mem: 3531 +Test: [ 120/2387] eta: 0:10:14 acc1_pos: 86.458 acc5_pos: 100.000 loss: 0.3473 (0.5878) acc1: 86.4583 (68.5778) acc5: 100.0000 (100.0000) time: 0.2542 data: 0.1579 max mem: 3531 +Test: [ 130/2387] eta: 0:10:09 acc1_pos: 85.677 acc5_pos: 100.000 loss: 0.3542 (0.5690) acc1: 86.4583 (70.1455) acc5: 100.0000 (100.0000) time: 0.2643 data: 0.1697 max mem: 3531 +Test: [ 140/2387] eta: 0:10:05 acc1_pos: 95.052 acc5_pos: 100.000 loss: 0.3457 (0.5532) acc1: 89.0625 (71.5333) acc5: 100.0000 (100.0000) time: 0.2611 data: 0.1666 max mem: 3531 +Test: [ 150/2387] eta: 0:09:55 acc1_pos: 98.698 acc5_pos: 100.000 loss: 0.3397 (0.5364) acc1: 91.1458 (73.0288) acc5: 100.0000 (100.0000) time: 0.2391 data: 0.1455 max mem: 3531 +Test: [ 160/2387] eta: 0:09:37 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.3346 (0.5272) acc1: 91.1458 (73.7513) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.1887 data: 0.0942 max mem: 3531 +Test: [ 170/2387] eta: 0:09:22 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.3785 (0.5192) acc1: 83.0729 (74.3071) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.1613 data: 0.0677 max mem: 3531 +Test: [ 180/2387] eta: 0:09:10 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.3787 (0.5117) acc1: 83.0729 (74.8360) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1699 data: 0.0772 max mem: 3531 +Test: [ 190/2387] eta: 0:09:03 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.3811 (0.5049) acc1: 84.1146 (75.3327) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.1935 data: 0.0998 max mem: 3531 +Test: [ 200/2387] eta: 0:08:55 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.3811 (0.4988) acc1: 83.8542 (75.7281) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.1999 data: 0.1028 max mem: 3531 +Test: [ 210/2387] eta: 0:08:46 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.3834 (0.4938) acc1: 83.0729 (76.0836) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.1899 data: 0.0906 max mem: 3532 +Test: [ 220/2387] eta: 0:08:41 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.3915 (0.4893) acc1: 82.8125 (76.3669) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.2012 data: 0.1038 max mem: 3532 +Test: [ 230/2387] eta: 0:08:36 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.3915 (0.4849) acc1: 82.8125 (76.6516) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.2125 data: 0.1170 max mem: 3532 +Test: [ 240/2387] eta: 0:08:23 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.4060 (0.4946) acc1: 79.4271 (75.9044) acc5: 100.0000 (100.0000) acc1_neg: 55.208 acc5_neg: 100.000 time: 0.1663 data: 0.0728 max mem: 3532 +Test: [ 250/2387] eta: 0:08:09 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.3612 (0.4821) acc1: 64.8438 (76.6465) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1073 data: 0.0142 max mem: 3532 +Test: [ 260/2387] eta: 0:08:07 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1002 (0.4678) acc1: 99.4792 (77.5074) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1640 data: 0.0704 max mem: 3532 +Test: [ 270/2387] eta: 0:08:03 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.0969 (0.4542) acc1: 99.7396 (78.3230) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2212 data: 0.1277 max mem: 3532 +Test: [ 280/2387] eta: 0:07:56 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.0969 (0.4419) acc1: 99.7396 (79.0749) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1893 data: 0.0949 max mem: 3532 +Test: [ 290/2387] eta: 0:07:53 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1022 (0.4303) acc1: 99.7396 (79.7645) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1886 data: 0.0934 max mem: 3532 +Test: [ 300/2387] eta: 0:07:48 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1047 (0.4202) acc1: 99.4792 (80.4082) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1987 data: 0.1023 max mem: 3532 +Test: [ 310/2387] eta: 0:07:49 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1437 (0.4156) acc1: 98.1771 (80.8146) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.2349 data: 0.1366 max mem: 3532 +Test: [ 320/2387] eta: 0:07:44 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.4317 (0.4164) acc1: 88.5417 (81.0301) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.2324 data: 0.1340 max mem: 3532 +Test: [ 330/2387] eta: 0:07:43 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.4396 (0.4169) acc1: 87.7604 (81.2406) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.2099 data: 0.1147 max mem: 3532 +Test: [ 340/2387] eta: 0:07:38 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.4302 (0.4173) acc1: 88.2812 (81.4600) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.2086 data: 0.1135 max mem: 3532 +Test: [ 350/2387] eta: 0:07:36 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.4168 (0.4152) acc1: 90.8854 (81.7827) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2049 data: 0.1084 max mem: 3532 +Test: [ 360/2387] eta: 0:07:34 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.3146 (0.4119) acc1: 91.4062 (82.0262) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.2237 data: 0.1268 max mem: 3532 +Test: [ 370/2387] eta: 0:07:31 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.2896 (0.4085) acc1: 90.8854 (82.2650) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2199 data: 0.1214 max mem: 3532 +Test: [ 380/2387] eta: 0:07:22 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.2887 (0.4053) acc1: 91.4062 (82.5022) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1599 data: 0.0608 max mem: 3532 +Test: [ 390/2387] eta: 0:07:15 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.2904 (0.4025) acc1: 90.3646 (82.7053) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1065 data: 0.0092 max mem: 3532 +Test: [ 400/2387] eta: 0:07:12 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.2904 (0.3996) acc1: 90.8854 (82.9106) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1599 data: 0.0617 max mem: 3532 +Test: [ 410/2387] eta: 0:07:08 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.2824 (0.3968) acc1: 91.1458 (83.1135) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1954 data: 0.0974 max mem: 3532 +Test: [ 420/2387] eta: 0:07:06 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.2816 (0.3940) acc1: 91.4062 (83.3203) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.2049 data: 0.1092 max mem: 3532 +Test: [ 430/2387] eta: 0:07:02 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.2973 (0.3955) acc1: 89.8438 (83.1992) acc5: 100.0000 (100.0000) acc1_neg: 66.406 acc5_neg: 100.000 time: 0.1978 data: 0.1035 max mem: 3532 +Test: [ 440/2387] eta: 0:07:00 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.3013 (0.3927) acc1: 73.1771 (83.3079) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1950 data: 0.1022 max mem: 3533 +Test: [ 450/2387] eta: 0:06:58 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.0857 (0.3859) acc1: 100.0000 (83.6781) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2202 data: 0.1272 max mem: 3533 +Test: [ 460/2387] eta: 0:06:56 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.0857 (0.3794) acc1: 100.0000 (84.0321) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2155 data: 0.1218 max mem: 3533 +Test: [ 470/2387] eta: 0:06:52 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.0858 (0.3732) acc1: 100.0000 (84.3711) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1896 data: 0.0940 max mem: 3533 +Test: [ 480/2387] eta: 0:06:48 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.0864 (0.3672) acc1: 100.0000 (84.6961) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1791 data: 0.0856 max mem: 3533 +Test: [ 490/2387] eta: 0:06:44 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.0868 (0.3615) acc1: 100.0000 (85.0077) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0818 max mem: 3533 +Test: [ 500/2387] eta: 0:06:42 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.0869 (0.3560) acc1: 100.0000 (85.3070) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0965 max mem: 3533 +Test: [ 510/2387] eta: 0:06:40 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.0869 (0.3517) acc1: 100.0000 (85.5925) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2150 data: 0.1241 max mem: 3533 +Test: [ 520/2387] eta: 0:06:38 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1730 (0.3493) acc1: 99.7396 (85.8560) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2099 data: 0.1165 max mem: 3533 +Test: [ 530/2387] eta: 0:06:35 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1720 (0.3459) acc1: 99.7396 (86.1204) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2048 data: 0.1094 max mem: 3533 +Test: [ 540/2387] eta: 0:06:33 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1670 (0.3426) acc1: 100.0000 (86.3741) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2099 data: 0.1135 max mem: 3533 +Test: [ 550/2387] eta: 0:06:31 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1675 (0.3394) acc1: 100.0000 (86.6186) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2206 data: 0.1245 max mem: 3533 +Test: [ 560/2387] eta: 0:06:28 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1686 (0.3364) acc1: 100.0000 (86.8538) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.1009 max mem: 3533 +Test: [ 570/2387] eta: 0:06:27 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1684 (0.3335) acc1: 99.7396 (87.0795) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2142 data: 0.1203 max mem: 3533 +Test: [ 580/2387] eta: 0:06:24 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1682 (0.3306) acc1: 99.7396 (87.2983) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2205 data: 0.1268 max mem: 3533 +Test: [ 590/2387] eta: 0:06:20 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1692 (0.3279) acc1: 99.7396 (87.5110) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0756 max mem: 3533 +Test: [ 600/2387] eta: 0:06:19 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1750 (0.3267) acc1: 99.7396 (87.7015) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1993 data: 0.1058 max mem: 3533 +Test: [ 610/2387] eta: 0:06:17 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.2849 (0.3269) acc1: 90.8854 (87.7080) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.2299 data: 0.1377 max mem: 3533 +Test: [ 620/2387] eta: 0:06:13 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.3335 (0.3270) acc1: 88.0208 (87.7172) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1899 data: 0.0953 max mem: 3533 +Test: [ 630/2387] eta: 0:06:12 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.3306 (0.3271) acc1: 88.2812 (87.7332) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1975 data: 0.1029 max mem: 3533 +Test: [ 640/2387] eta: 0:06:09 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.3360 (0.3272) acc1: 88.2812 (87.7393) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.2106 data: 0.1185 max mem: 3533 +Test: [ 650/2387] eta: 0:06:07 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.3383 (0.3274) acc1: 88.0208 (87.7432) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1996 data: 0.1076 max mem: 3533 +Test: [ 660/2387] eta: 0:06:06 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.3385 (0.3276) acc1: 87.5000 (87.7462) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.2343 data: 0.1407 max mem: 3533 +Test: [ 670/2387] eta: 0:06:01 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.3341 (0.3277) acc1: 87.5000 (87.7538) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.1826 data: 0.0866 max mem: 3534 +Test: [ 680/2387] eta: 0:06:01 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.3412 (0.3328) acc1: 86.9792 (87.3119) acc5: 100.0000 (100.0000) acc1_neg: 40.625 acc5_neg: 100.000 time: 0.1978 data: 0.1029 max mem: 3534 +Test: [ 690/2387] eta: 0:05:58 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.5950 (0.3371) acc1: 50.7812 (86.9471) acc5: 100.0000 (100.0000) acc1_neg: 65.104 acc5_neg: 100.000 time: 0.2312 data: 0.1394 max mem: 3534 +Test: [ 700/2387] eta: 0:05:55 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.5550 (0.3400) acc1: 69.7917 (86.7124) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.1697 data: 0.0779 max mem: 3534 +Test: [ 710/2387] eta: 0:05:51 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.5363 (0.3429) acc1: 70.0521 (86.4763) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.1566 data: 0.0608 max mem: 3534 +Test: [ 720/2387] eta: 0:05:49 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.5363 (0.3457) acc1: 69.5312 (86.2460) acc5: 100.0000 (100.0000) acc1_neg: 71.094 acc5_neg: 100.000 time: 0.1721 data: 0.0771 max mem: 3534 +Test: [ 730/2387] eta: 0:05:46 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.5461 (0.3485) acc1: 69.2708 (86.0152) acc5: 100.0000 (100.0000) acc1_neg: 67.708 acc5_neg: 100.000 time: 0.1920 data: 0.1006 max mem: 3534 +Test: [ 740/2387] eta: 0:05:44 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.5464 (0.3511) acc1: 68.4896 (85.8004) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.1899 data: 0.0970 max mem: 3534 +Test: [ 750/2387] eta: 0:05:41 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.5329 (0.3535) acc1: 70.0521 (85.6011) acc5: 100.0000 (100.0000) acc1_neg: 69.531 acc5_neg: 100.000 time: 0.1896 data: 0.0963 max mem: 3534 +Test: [ 760/2387] eta: 0:05:38 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.5329 (0.3559) acc1: 70.3125 (85.4160) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.1817 data: 0.0897 max mem: 3534 +Test: [ 770/2387] eta: 0:05:36 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.4957 (0.3560) acc1: 74.2188 (85.4025) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1800 data: 0.0865 max mem: 3534 +Test: [ 780/2387] eta: 0:05:33 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.2017 (0.3540) acc1: 95.3125 (85.5307) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1845 data: 0.0915 max mem: 3534 +Test: [ 790/2387] eta: 0:05:31 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1601 (0.3515) acc1: 97.1354 (85.6751) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1803 data: 0.0897 max mem: 3534 +Test: [ 800/2387] eta: 0:05:28 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1601 (0.3490) acc1: 97.9167 (85.8367) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1838 data: 0.0934 max mem: 3534 +Test: [ 810/2387] eta: 0:05:25 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1653 (0.3471) acc1: 97.9167 (85.9632) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1761 data: 0.0859 max mem: 3534 +Test: [ 820/2387] eta: 0:05:23 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.2056 (0.3459) acc1: 94.2708 (86.0327) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.1700 data: 0.0798 max mem: 3534 +Test: [ 830/2387] eta: 0:05:20 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.4711 (0.3490) acc1: 75.2604 (85.8002) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.1860 data: 0.0946 max mem: 3534 +Test: [ 840/2387] eta: 0:05:18 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.5721 (0.3510) acc1: 67.4479 (85.6477) acc5: 100.0000 (100.0000) acc1_neg: 69.010 acc5_neg: 100.000 time: 0.1958 data: 0.1008 max mem: 3534 +Test: [ 850/2387] eta: 0:05:16 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.5226 (0.3531) acc1: 70.8333 (85.5057) acc5: 100.0000 (100.0000) acc1_neg: 70.833 acc5_neg: 100.000 time: 0.1971 data: 0.1014 max mem: 3534 +Test: [ 860/2387] eta: 0:05:14 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.4712 (0.3546) acc1: 75.0000 (85.3994) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.1959 data: 0.1026 max mem: 3534 +Test: [ 870/2387] eta: 0:05:12 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.4661 (0.3558) acc1: 77.3438 (85.2938) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.2023 data: 0.1092 max mem: 3534 +Test: [ 880/2387] eta: 0:05:09 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.4362 (0.3574) acc1: 79.1667 (85.1846) acc5: 100.0000 (100.0000) acc1_neg: 65.625 acc5_neg: 100.000 time: 0.1925 data: 0.0987 max mem: 3534 +Test: [ 890/2387] eta: 0:05:07 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.5946 (0.3609) acc1: 65.6250 (84.8996) acc5: 100.0000 (100.0000) acc1_neg: 51.823 acc5_neg: 100.000 time: 0.1837 data: 0.0914 max mem: 3534 +Test: [ 900/2387] eta: 0:05:04 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.7531 (0.3664) acc1: 50.5208 (84.4958) acc5: 100.0000 (100.0000) acc1_neg: 50.521 acc5_neg: 100.000 time: 0.1835 data: 0.0902 max mem: 3535 +Test: [ 910/2387] eta: 0:05:02 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.8862 (0.3722) acc1: 45.8333 (84.0703) acc5: 100.0000 (100.0000) acc1_neg: 47.135 acc5_neg: 100.000 time: 0.1825 data: 0.0885 max mem: 3535 +Test: [ 920/2387] eta: 0:05:00 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.8928 (0.3779) acc1: 45.3125 (83.6458) acc5: 100.0000 (100.0000) acc1_neg: 41.927 acc5_neg: 100.000 time: 0.1923 data: 0.1012 max mem: 3535 +Test: [ 930/2387] eta: 0:04:57 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.8902 (0.3833) acc1: 45.3125 (83.2491) acc5: 100.0000 (100.0000) acc1_neg: 39.583 acc5_neg: 100.000 time: 0.1806 data: 0.0876 max mem: 3535 +Test: [ 940/2387] eta: 0:04:55 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.8916 (0.3887) acc1: 46.3542 (82.8499) acc5: 100.0000 (100.0000) acc1_neg: 47.396 acc5_neg: 100.000 time: 0.1712 data: 0.0748 max mem: 3535 +Test: [ 950/2387] eta: 0:04:52 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.8862 (0.3939) acc1: 45.3125 (82.4568) acc5: 100.0000 (100.0000) acc1_neg: 48.958 acc5_neg: 100.000 time: 0.1798 data: 0.0839 max mem: 3535 +Test: [ 960/2387] eta: 0:04:50 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.8740 (0.3990) acc1: 46.8750 (82.0892) acc5: 100.0000 (100.0000) acc1_neg: 49.479 acc5_neg: 100.000 time: 0.1862 data: 0.0907 max mem: 3535 +Test: [ 970/2387] eta: 0:04:48 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.8847 (0.4041) acc1: 45.8333 (81.7035) acc5: 100.0000 (100.0000) acc1_neg: 45.833 acc5_neg: 100.000 time: 0.1808 data: 0.0869 max mem: 3535 +Test: [ 980/2387] eta: 0:04:45 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.8633 (0.4059) acc1: 45.8333 (81.5622) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.1799 data: 0.0850 max mem: 3535 +Test: [ 990/2387] eta: 0:04:43 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.4259 (0.4055) acc1: 79.6875 (81.5893) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1809 data: 0.0866 max mem: 3535 +Test: [1000/2387] eta: 0:04:41 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.3052 (0.4046) acc1: 86.1979 (81.6486) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1772 data: 0.0859 max mem: 3535 +Test: [1010/2387] eta: 0:04:38 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.3403 (0.4051) acc1: 83.5938 (81.6119) acc5: 100.0000 (100.0000) acc1_neg: 66.146 acc5_neg: 100.000 time: 0.1790 data: 0.0859 max mem: 3535 +Test: [1020/2387] eta: 0:04:36 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.3403 (0.4050) acc1: 76.3021 (81.6025) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1776 data: 0.0853 max mem: 3535 +Test: [1030/2387] eta: 0:04:33 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1515 (0.4026) acc1: 97.9167 (81.7549) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1746 data: 0.0836 max mem: 3535 +Test: [1040/2387] eta: 0:04:31 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1201 (0.3999) acc1: 98.9583 (81.9199) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1700 data: 0.0784 max mem: 3535 +Test: [1050/2387] eta: 0:04:29 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1136 (0.3972) acc1: 99.4792 (82.0840) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1694 data: 0.0783 max mem: 3535 +Test: [1060/2387] eta: 0:04:26 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1193 (0.3949) acc1: 98.9583 (82.2377) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1713 data: 0.0805 max mem: 3535 +Test: [1070/2387] eta: 0:04:24 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1450 (0.3934) acc1: 98.4375 (82.3342) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.1704 data: 0.0781 max mem: 3535 +Test: [1080/2387] eta: 0:04:21 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.3736 (0.3942) acc1: 80.4688 (82.2743) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.1686 data: 0.0749 max mem: 3535 +Test: [1090/2387] eta: 0:04:19 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.3736 (0.3937) acc1: 78.3854 (82.3024) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.1692 data: 0.0765 max mem: 3535 +Test: [1100/2387] eta: 0:04:17 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.3052 (0.3930) acc1: 88.2812 (82.3551) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1812 data: 0.0878 max mem: 3535 +Test: [1110/2387] eta: 0:04:15 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.3062 (0.3923) acc1: 87.5000 (82.4021) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.1947 data: 0.1011 max mem: 3535 +Test: [1120/2387] eta: 0:04:13 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.3294 (0.3917) acc1: 86.7188 (82.4410) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.1834 data: 0.0892 max mem: 3536 +Test: [1130/2387] eta: 0:04:10 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.3290 (0.3911) acc1: 85.9375 (82.4842) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1658 data: 0.0695 max mem: 3536 +Test: [1140/2387] eta: 0:04:08 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.3940 (0.3929) acc1: 81.5104 (82.3638) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.1699 data: 0.0754 max mem: 3536 +Test: [1150/2387] eta: 0:04:06 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.5962 (0.3939) acc1: 67.1875 (82.3023) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.1863 data: 0.0939 max mem: 3536 +Test: [1160/2387] eta: 0:04:03 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.3996 (0.3939) acc1: 79.1667 (82.3060) acc5: 100.0000 (100.0000) acc1_neg: 72.396 acc5_neg: 100.000 time: 0.1814 data: 0.0847 max mem: 3536 +Test: [1170/2387] eta: 0:04:01 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.3672 (0.3936) acc1: 83.5938 (82.3268) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1776 data: 0.0815 max mem: 3536 +Test: [1180/2387] eta: 0:03:59 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.3664 (0.3933) acc1: 84.3750 (82.3596) acc5: 100.0000 (100.0000) acc1_neg: 75.781 acc5_neg: 100.000 time: 0.1803 data: 0.0872 max mem: 3536 +Test: [1190/2387] eta: 0:03:57 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.3096 (0.3930) acc1: 89.3229 (82.3704) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1792 data: 0.0852 max mem: 3536 +Test: [1200/2387] eta: 0:03:55 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.3480 (0.3933) acc1: 85.6771 (82.3437) acc5: 100.0000 (100.0000) acc1_neg: 66.146 acc5_neg: 100.000 time: 0.1980 data: 0.1018 max mem: 3536 +Test: [1210/2387] eta: 0:03:53 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.4353 (0.3936) acc1: 76.3021 (82.2992) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1912 data: 0.0937 max mem: 3536 +Test: [1220/2387] eta: 0:03:51 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1772 (0.3916) acc1: 95.5729 (82.4262) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2013 data: 0.1076 max mem: 3536 +Test: [1230/2387] eta: 0:03:49 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1210 (0.3895) acc1: 98.6979 (82.5531) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1850 data: 0.0932 max mem: 3536 +Test: [1240/2387] eta: 0:03:47 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1242 (0.3874) acc1: 98.4375 (82.6795) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1657 data: 0.0743 max mem: 3536 +Test: [1250/2387] eta: 0:03:44 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1266 (0.3854) acc1: 98.1771 (82.8021) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1845 data: 0.0940 max mem: 3536 +Test: [1260/2387] eta: 0:03:42 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1306 (0.3834) acc1: 97.6562 (82.9193) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1730 data: 0.0797 max mem: 3536 +Test: [1270/2387] eta: 0:03:40 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1537 (0.3819) acc1: 97.1354 (83.0201) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1619 data: 0.0685 max mem: 3536 +Test: [1280/2387] eta: 0:03:38 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1678 (0.3802) acc1: 96.3542 (83.1292) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1646 data: 0.0727 max mem: 3536 +Test: [1290/2387] eta: 0:03:36 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1662 (0.3785) acc1: 96.8750 (83.2349) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1728 data: 0.0810 max mem: 3536 +Test: [1300/2387] eta: 0:03:34 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1740 (0.3770) acc1: 96.8750 (83.3335) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1813 data: 0.0881 max mem: 3536 +Test: [1310/2387] eta: 0:03:31 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1754 (0.3755) acc1: 96.3542 (83.4295) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1798 data: 0.0863 max mem: 3536 +Test: [1320/2387] eta: 0:03:29 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1754 (0.3740) acc1: 96.6146 (83.5320) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1735 data: 0.0800 max mem: 3536 +Test: [1330/2387] eta: 0:03:27 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.2157 (0.3740) acc1: 94.2708 (83.5386) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.1745 data: 0.0808 max mem: 3536 +Test: [1340/2387] eta: 0:03:25 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.4113 (0.3748) acc1: 79.9479 (83.4765) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1719 data: 0.0806 max mem: 3536 +Test: [1350/2387] eta: 0:03:23 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.0615 (0.3725) acc1: 100.0000 (83.5988) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1815 data: 0.0906 max mem: 3537 +Test: [1360/2387] eta: 0:03:21 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.0563 (0.3701) acc1: 100.0000 (83.7193) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1852 data: 0.0938 max mem: 3537 +Test: [1370/2387] eta: 0:03:19 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.0563 (0.3679) acc1: 100.0000 (83.8380) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1827 data: 0.0909 max mem: 3537 +Test: [1380/2387] eta: 0:03:17 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.0538 (0.3656) acc1: 100.0000 (83.9551) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1867 data: 0.0957 max mem: 3537 +Test: [1390/2387] eta: 0:03:15 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.0527 (0.3634) acc1: 100.0000 (84.0704) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1944 data: 0.1038 max mem: 3537 +Test: [1400/2387] eta: 0:03:13 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.0527 (0.3612) acc1: 100.0000 (84.1835) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.1042 max mem: 3537 +Test: [1410/2387] eta: 0:03:11 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.0621 (0.3603) acc1: 100.0000 (84.2355) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.1707 data: 0.0800 max mem: 3537 +Test: [1420/2387] eta: 0:03:08 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.2521 (0.3597) acc1: 91.4062 (84.2775) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1649 data: 0.0743 max mem: 3537 +Test: [1430/2387] eta: 0:03:06 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.2817 (0.3593) acc1: 91.4062 (84.3140) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1785 data: 0.0882 max mem: 3537 +Test: [1440/2387] eta: 0:03:04 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.2833 (0.3591) acc1: 90.6250 (84.3266) acc5: 100.0000 (100.0000) acc1_neg: 57.552 acc5_neg: 100.000 time: 0.1893 data: 0.0970 max mem: 3537 +Test: [1450/2387] eta: 0:03:02 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.3751 (0.3603) acc1: 79.9479 (84.2363) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1857 data: 0.0915 max mem: 3537 +Test: [1460/2387] eta: 0:03:00 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.2242 (0.3594) acc1: 87.5000 (84.3019) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1717 data: 0.0792 max mem: 3537 +Test: [1470/2387] eta: 0:02:58 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1913 (0.3582) acc1: 95.8333 (84.3835) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1895 data: 0.0969 max mem: 3537 +Test: [1480/2387] eta: 0:02:56 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1705 (0.3569) acc1: 96.8750 (84.4652) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1988 data: 0.1065 max mem: 3537 +Test: [1490/2387] eta: 0:02:55 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1692 (0.3558) acc1: 97.9167 (84.5486) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2001 data: 0.1092 max mem: 3537 +Test: [1500/2387] eta: 0:02:53 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1784 (0.3547) acc1: 96.8750 (84.6165) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.2191 data: 0.1254 max mem: 3537 +Test: [1510/2387] eta: 0:02:51 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.2804 (0.3550) acc1: 90.3646 (84.6094) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.1893 data: 0.0915 max mem: 3537 +Test: [1520/2387] eta: 0:02:49 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.4163 (0.3556) acc1: 82.8125 (84.5775) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.2061 data: 0.1092 max mem: 3537 +Test: [1530/2387] eta: 0:02:47 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.4429 (0.3562) acc1: 77.8646 (84.5429) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.2253 data: 0.1316 max mem: 3537 +Test: [1540/2387] eta: 0:02:45 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.3904 (0.3565) acc1: 83.0729 (84.5342) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.1845 data: 0.0899 max mem: 3537 +Test: [1550/2387] eta: 0:02:43 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.3904 (0.3567) acc1: 84.3750 (84.5261) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.2149 data: 0.1201 max mem: 3537 +Test: [1560/2387] eta: 0:02:41 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.3978 (0.3570) acc1: 82.0312 (84.5185) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.2215 data: 0.1277 max mem: 3537 +Test: [1570/2387] eta: 0:02:39 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.4610 (0.3582) acc1: 73.6979 (84.4365) acc5: 100.0000 (100.0000) acc1_neg: 73.698 acc5_neg: 100.000 time: 0.1744 data: 0.0821 max mem: 3537 +Test: [1580/2387] eta: 0:02:37 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.4296 (0.3580) acc1: 78.9062 (84.4471) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.1681 data: 0.0758 max mem: 3538 +Test: [1590/2387] eta: 0:02:35 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1829 (0.3568) acc1: 95.5729 (84.5274) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1949 data: 0.1002 max mem: 3538 +Test: [1600/2387] eta: 0:02:33 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1406 (0.3555) acc1: 98.1771 (84.6052) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1942 data: 0.0991 max mem: 3538 +Test: [1610/2387] eta: 0:02:31 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1484 (0.3543) acc1: 98.1771 (84.6907) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1635 data: 0.0685 max mem: 3538 +Test: [1620/2387] eta: 0:02:29 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1667 (0.3533) acc1: 96.3542 (84.7537) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1549 data: 0.0615 max mem: 3538 +Test: [1630/2387] eta: 0:02:27 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.2151 (0.3535) acc1: 92.9688 (84.7390) acc5: 100.0000 (100.0000) acc1_neg: 71.615 acc5_neg: 100.000 time: 0.1662 data: 0.0723 max mem: 3538 +Test: [1640/2387] eta: 0:02:25 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.4949 (0.3551) acc1: 73.4375 (84.6294) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.1799 data: 0.0857 max mem: 3538 +Test: [1650/2387] eta: 0:02:23 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.6158 (0.3565) acc1: 67.4479 (84.5356) acc5: 100.0000 (100.0000) acc1_neg: 70.312 acc5_neg: 100.000 time: 0.1814 data: 0.0889 max mem: 3538 +Test: [1660/2387] eta: 0:02:21 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.5526 (0.3578) acc1: 68.2292 (84.4412) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.1654 data: 0.0694 max mem: 3538 +Test: [1670/2387] eta: 0:02:19 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.5136 (0.3587) acc1: 73.6979 (84.3826) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.1565 data: 0.0585 max mem: 3538 +Test: [1680/2387] eta: 0:02:17 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.5057 (0.3596) acc1: 73.6979 (84.3174) acc5: 100.0000 (100.0000) acc1_neg: 62.240 acc5_neg: 100.000 time: 0.1549 data: 0.0610 max mem: 3538 +Test: [1690/2387] eta: 0:02:14 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.5099 (0.3606) acc1: 73.6979 (84.2557) acc5: 100.0000 (100.0000) acc1_neg: 75.521 acc5_neg: 100.000 time: 0.1518 data: 0.0594 max mem: 3538 +Test: [1700/2387] eta: 0:02:12 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.5099 (0.3612) acc1: 73.6979 (84.2054) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1513 data: 0.0591 max mem: 3538 +Test: [1710/2387] eta: 0:02:10 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.2249 (0.3602) acc1: 91.1458 (84.2672) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1656 data: 0.0742 max mem: 3538 +Test: [1720/2387] eta: 0:02:08 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1709 (0.3592) acc1: 95.8333 (84.3319) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1901 data: 0.0988 max mem: 3538 +Test: [1730/2387] eta: 0:02:06 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1624 (0.3581) acc1: 96.8750 (84.3970) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1797 data: 0.0885 max mem: 3538 +Test: [1740/2387] eta: 0:02:05 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1624 (0.3571) acc1: 96.8750 (84.4634) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1889 data: 0.0970 max mem: 3538 +Test: [1750/2387] eta: 0:02:03 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1722 (0.3561) acc1: 95.8333 (84.5246) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2044 data: 0.1113 max mem: 3538 +Test: [1760/2387] eta: 0:02:01 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1942 (0.3557) acc1: 94.2708 (84.5477) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.1771 data: 0.0831 max mem: 3538 +Test: [1770/2387] eta: 0:01:59 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.2962 (0.3555) acc1: 87.2396 (84.5550) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.1901 data: 0.0954 max mem: 3538 +Test: [1780/2387] eta: 0:01:57 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.3155 (0.3553) acc1: 87.2396 (84.5717) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1826 data: 0.0880 max mem: 3538 +Test: [1790/2387] eta: 0:01:55 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.2919 (0.3547) acc1: 88.5417 (84.6091) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1804 data: 0.0877 max mem: 3538 +Test: [1800/2387] eta: 0:01:53 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.2648 (0.3542) acc1: 91.9271 (84.6474) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1817 data: 0.0902 max mem: 3538 +Test: [1810/2387] eta: 0:01:51 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.2581 (0.3535) acc1: 91.9271 (84.6954) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1702 data: 0.0790 max mem: 3539 +Test: [1820/2387] eta: 0:01:49 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.2743 (0.3534) acc1: 89.3229 (84.7043) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1991 data: 0.1079 max mem: 3539 +Test: [1830/2387] eta: 0:01:47 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.2975 (0.3532) acc1: 85.4167 (84.7183) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1817 data: 0.0908 max mem: 3539 +Test: [1840/2387] eta: 0:01:45 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1986 (0.3524) acc1: 97.6562 (84.7931) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1694 data: 0.0768 max mem: 3539 +Test: [1850/2387] eta: 0:01:43 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1926 (0.3515) acc1: 98.1771 (84.8656) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1820 data: 0.0877 max mem: 3539 +Test: [1860/2387] eta: 0:01:41 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1926 (0.3507) acc1: 98.1771 (84.9380) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1736 data: 0.0795 max mem: 3539 +Test: [1870/2387] eta: 0:01:39 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1975 (0.3499) acc1: 98.1771 (85.0073) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1749 data: 0.0820 max mem: 3539 +Test: [1880/2387] eta: 0:01:37 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1982 (0.3491) acc1: 98.1771 (85.0783) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1714 data: 0.0801 max mem: 3539 +Test: [1890/2387] eta: 0:01:35 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1943 (0.3483) acc1: 98.4375 (85.1477) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1749 data: 0.0832 max mem: 3539 +Test: [1900/2387] eta: 0:01:33 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1947 (0.3476) acc1: 98.1771 (85.2126) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1781 data: 0.0840 max mem: 3539 +Test: [1910/2387] eta: 0:01:31 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.3296 (0.3476) acc1: 90.6250 (85.2300) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1949 data: 0.1011 max mem: 3539 +Test: [1920/2387] eta: 0:01:29 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.3542 (0.3480) acc1: 87.7604 (85.2067) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.1878 data: 0.0945 max mem: 3539 +Test: [1930/2387] eta: 0:01:27 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.2674 (0.3475) acc1: 90.6250 (85.2339) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1762 data: 0.0810 max mem: 3539 +Test: [1940/2387] eta: 0:01:25 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.2745 (0.3473) acc1: 91.9271 (85.2684) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1921 data: 0.0982 max mem: 3539 +Test: [1950/2387] eta: 0:01:23 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.2983 (0.3472) acc1: 91.4062 (85.2868) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1850 data: 0.0937 max mem: 3539 +Test: [1960/2387] eta: 0:01:21 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.2878 (0.3471) acc1: 88.8021 (85.2946) acc5: 100.0000 (100.0000) acc1_neg: 75.781 acc5_neg: 100.000 time: 0.1898 data: 0.0979 max mem: 3539 +Test: [1970/2387] eta: 0:01:19 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.4550 (0.3487) acc1: 76.3021 (85.1732) acc5: 100.0000 (100.0000) acc1_neg: 61.979 acc5_neg: 100.000 time: 0.1742 data: 0.0808 max mem: 3539 +Test: [1980/2387] eta: 0:01:18 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.6245 (0.3493) acc1: 61.9792 (85.1302) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1776 data: 0.0853 max mem: 3539 +Test: [1990/2387] eta: 0:01:16 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1899 (0.3485) acc1: 97.3958 (85.1924) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1943 data: 0.1034 max mem: 3539 +Test: [2000/2387] eta: 0:01:14 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1895 (0.3477) acc1: 97.6562 (85.2545) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1783 data: 0.0849 max mem: 3539 +Test: [2010/2387] eta: 0:01:12 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1911 (0.3470) acc1: 97.6562 (85.3155) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1619 data: 0.0677 max mem: 3539 +Test: [2020/2387] eta: 0:01:10 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1907 (0.3462) acc1: 97.3958 (85.3749) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1692 data: 0.0779 max mem: 3539 +Test: [2030/2387] eta: 0:01:08 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1907 (0.3455) acc1: 97.3958 (85.4362) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2027 data: 0.1101 max mem: 3540 +Test: [2040/2387] eta: 0:01:06 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1913 (0.3447) acc1: 97.6562 (85.4968) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1962 data: 0.1009 max mem: 3540 +Test: [2050/2387] eta: 0:01:04 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1897 (0.3439) acc1: 97.6562 (85.5544) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1799 data: 0.0856 max mem: 3540 +Test: [2060/2387] eta: 0:01:02 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1968 (0.3433) acc1: 96.6146 (85.6080) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1908 data: 0.0970 max mem: 3540 +Test: [2070/2387] eta: 0:01:00 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.2081 (0.3426) acc1: 96.6146 (85.6634) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2024 data: 0.1095 max mem: 3540 +Test: [2080/2387] eta: 0:00:58 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.2085 (0.3420) acc1: 96.8750 (85.7176) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1935 data: 0.1013 max mem: 3540 +Test: [2090/2387] eta: 0:00:56 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.2013 (0.3413) acc1: 97.1354 (85.7732) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1657 data: 0.0710 max mem: 3540 +Test: [2100/2387] eta: 0:00:54 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1979 (0.3406) acc1: 97.3958 (85.8283) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1613 data: 0.0680 max mem: 3540 +Test: [2110/2387] eta: 0:00:53 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1985 (0.3400) acc1: 97.1354 (85.8806) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1973 data: 0.1067 max mem: 3540 +Test: [2120/2387] eta: 0:00:51 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1999 (0.3394) acc1: 97.1354 (85.9241) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.1985 data: 0.1052 max mem: 3540 +Test: [2130/2387] eta: 0:00:49 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1477 (0.3384) acc1: 98.1771 (85.9832) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1594 data: 0.0643 max mem: 3540 +Test: [2140/2387] eta: 0:00:47 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1434 (0.3376) acc1: 98.4375 (86.0386) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1553 data: 0.0609 max mem: 3540 +Test: [2150/2387] eta: 0:00:45 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1434 (0.3367) acc1: 98.4375 (86.0978) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1915 data: 0.0984 max mem: 3540 +Test: [2160/2387] eta: 0:00:43 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1687 (0.3360) acc1: 98.4375 (86.1454) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1886 data: 0.0959 max mem: 3540 +Test: [2170/2387] eta: 0:00:41 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.2220 (0.3361) acc1: 94.2708 (86.1407) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.1281 data: 0.0319 max mem: 3540 +Test: [2180/2387] eta: 0:00:39 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.4505 (0.3376) acc1: 77.8646 (86.0245) acc5: 100.0000 (100.0000) acc1_neg: 39.323 acc5_neg: 100.000 time: 0.1468 data: 0.0517 max mem: 3540 +Test: [2190/2387] eta: 0:00:37 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.4900 (0.3384) acc1: 70.3125 (85.9660) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2265 data: 0.1350 max mem: 3540 +Test: [2200/2387] eta: 0:00:35 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.3139 (0.3382) acc1: 86.4583 (85.9913) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2076 data: 0.1163 max mem: 3540 +Test: [2210/2387] eta: 0:00:33 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.2604 (0.3379) acc1: 90.8854 (86.0103) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1552 data: 0.0623 max mem: 3540 +Test: [2220/2387] eta: 0:00:31 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.2739 (0.3376) acc1: 90.8854 (86.0312) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1726 data: 0.0789 max mem: 3540 +Test: [2230/2387] eta: 0:00:29 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.2679 (0.3372) acc1: 91.4062 (86.0588) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1742 data: 0.0825 max mem: 3540 +Test: [2240/2387] eta: 0:00:27 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.2057 (0.3368) acc1: 92.9688 (86.0853) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1672 data: 0.0760 max mem: 3540 +Test: [2250/2387] eta: 0:00:26 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.3204 (0.3377) acc1: 83.3333 (86.0187) acc5: 100.0000 (100.0000) acc1_neg: 55.469 acc5_neg: 100.000 time: 0.1718 data: 0.0802 max mem: 3540 +Test: [2260/2387] eta: 0:00:24 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.3655 (0.3376) acc1: 74.4792 (86.0255) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1760 data: 0.0853 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1264 (0.3367) acc1: 99.7396 (86.0781) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1945 data: 0.1028 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1162 (0.3358) acc1: 99.7396 (86.1308) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1815 data: 0.0885 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1085 (0.3348) acc1: 99.7396 (86.1897) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1795 data: 0.0850 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1315 (0.3340) acc1: 99.2188 (86.2409) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1932 data: 0.0998 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.1616 (0.3338) acc1: 95.8333 (86.2623) acc5: 100.0000 (100.0000) acc1_neg: 80.729 acc5_neg: 100.000 time: 0.1831 data: 0.0926 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.3295 (0.3340) acc1: 87.2396 (86.2458) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1810 data: 0.0899 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.3557 (0.3344) acc1: 84.6354 (86.2284) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.1869 data: 0.0945 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.3064 (0.3343) acc1: 88.2812 (86.2307) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1750 data: 0.0834 max mem: 3541 +Test: [2350/2387] eta: 0:00:07 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.3066 (0.3344) acc1: 88.2812 (86.2237) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1723 data: 0.0817 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.3335 (0.3347) acc1: 87.5000 (86.2116) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1871 data: 0.0955 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.3706 (0.3350) acc1: 80.2083 (86.1867) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.1588 data: 0.0650 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.5580 (0.3368) acc1: 66.1458 (86.0651) acc5: 100.0000 (100.0000) acc1_neg: 54.427 acc5_neg: 100.000 time: 0.1302 data: 0.0371 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 85.845 acc5_pos: 100.000 loss: 0.7674 (0.3392) acc1: 54.4271 (85.9397) acc5: 100.0000 (100.0000) acc1_neg: 6.977 acc5_neg: 100.000 time: 0.1337 data: 0.0462 max mem: 3541 +Test: Total time: 0:07:32 (0.1895 s / it) +* Overall Acc@1 85.940 Acc@5 100.000 + target == real Acc@1 73.770 Acc@5 100.000 + target == fake Acc@1 86.794 Acc@5 100.000 + loss 0.339 + PR-AUC 0.9920 ROC-AUC 0.9000 EER 18.31% (thr=0.5874) +Accuracy of the network on the 916267 test images: 85.9% +[rank0]:[W725 15:41:32.932520823 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 15:41:51.922000 73345 torch/distributed/run.py:766] +W0725 15:41:51.922000 73345 torch/distributed/run.py:766] ***************************************** +W0725 15:41:51.922000 73345 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 15:41:51.922000 73345 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 15:42:16.954972303 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 15:42:17.298934992 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 15:42:17.846624504 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +[rank0]:[W725 15:42:17.985199030 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 15:42:17.988603563 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +[rank4]:[W725 15:42:18.090135252 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 15:42:18.103256201 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 15:42:18.148263445 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n16_seed1_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed1_resampled_n116-041-153_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 0:51:39 acc1_pos: 65.104 acc5_pos: 100.000 loss: 0.6117 (0.6117) acc1: 65.1042 (65.1042) acc5: 100.0000 (100.0000) time: 5.9948 data: 3.8605 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/517] eta: 0:05:22 acc1_pos: 53.906 acc5_pos: 100.000 loss: 0.7843 (0.7658) acc1: 53.9062 (52.7699) acc5: 100.0000 (100.0000) time: 0.6360 data: 0.3607 max mem: 3531 +Test: [ 20/517] eta: 0:03:12 acc1_pos: 50.781 acc5_pos: 100.000 loss: 0.7717 (0.7530) acc1: 52.8646 (54.5759) acc5: 100.0000 (100.0000) time: 0.1060 data: 0.0162 max mem: 3531 +Test: [ 30/517] eta: 0:02:31 acc1_pos: 32.552 acc5_pos: 100.000 loss: 0.8024 (0.7838) acc1: 47.1354 (51.4617) acc5: 100.0000 (100.0000) time: 0.1311 data: 0.0404 max mem: 3531 +Test: [ 40/517] eta: 0:02:25 acc1_pos: 54.948 acc5_pos: 100.000 loss: 0.8207 (0.7930) acc1: 45.5729 (50.4637) acc5: 100.0000 (100.0000) time: 0.2210 data: 0.1303 max mem: 3531 +Test: [ 50/517] eta: 0:02:18 acc1_pos: 72.936 acc5_pos: 100.000 loss: 0.7127 (0.7732) acc1: 52.0833 (52.4101) acc5: 100.0000 (100.0000) acc1_neg: 59.896 acc5_neg: 100.000 time: 0.2749 data: 0.1848 max mem: 3531 +Test: [ 60/517] eta: 0:02:12 acc1_pos: 72.936 acc5_pos: 100.000 loss: 0.6753 (0.7509) acc1: 59.8958 (54.3588) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.2595 data: 0.1668 max mem: 3531 +Test: [ 70/517] eta: 0:02:11 acc1_pos: 72.936 acc5_pos: 100.000 loss: 0.6645 (0.7273) acc1: 59.6354 (56.6094) acc5: 100.0000 (100.0000) acc1_neg: 57.552 acc5_neg: 100.000 time: 0.2847 data: 0.1917 max mem: 3531 +Test: [ 80/517] eta: 0:02:05 acc1_pos: 72.936 acc5_pos: 100.000 loss: 0.6371 (0.7154) acc1: 62.7604 (57.5585) acc5: 100.0000 (100.0000) acc1_neg: 63.281 acc5_neg: 100.000 time: 0.2807 data: 0.1887 max mem: 3531 +Test: [ 90/517] eta: 0:02:01 acc1_pos: 72.936 acc5_pos: 100.000 loss: 0.6535 (0.7076) acc1: 64.5833 (58.9057) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.2504 data: 0.1575 max mem: 3531 +Test: [100/517] eta: 0:01:57 acc1_pos: 72.936 acc5_pos: 100.000 loss: 0.6260 (0.6940) acc1: 65.8854 (60.0196) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.2590 data: 0.1676 max mem: 3531 +Test: [110/517] eta: 0:01:53 acc1_pos: 72.936 acc5_pos: 100.000 loss: 0.4234 (0.6665) acc1: 79.6875 (62.4062) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2560 data: 0.1641 max mem: 3531 +Test: [120/517] eta: 0:01:50 acc1_pos: 72.936 acc5_pos: 100.000 loss: 0.4234 (0.6569) acc1: 77.8646 (63.1715) acc5: 100.0000 (100.0000) acc1_neg: 70.312 acc5_neg: 100.000 time: 0.2508 data: 0.1585 max mem: 3531 +Test: [130/517] eta: 0:01:47 acc1_pos: 72.936 acc5_pos: 100.000 loss: 0.5298 (0.6425) acc1: 74.2188 (64.4521) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2643 data: 0.1715 max mem: 3531 +Test: [140/517] eta: 0:01:43 acc1_pos: 72.936 acc5_pos: 100.000 loss: 0.4067 (0.6246) acc1: 85.1562 (66.0295) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2585 data: 0.1633 max mem: 3531 +Test: [150/517] eta: 0:01:40 acc1_pos: 72.936 acc5_pos: 100.000 loss: 0.4027 (0.6098) acc1: 85.1562 (67.3755) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.2546 data: 0.1611 max mem: 3531 +Test: [160/517] eta: 0:01:37 acc1_pos: 72.936 acc5_pos: 100.000 loss: 0.4193 (0.5996) acc1: 84.1146 (68.2373) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2609 data: 0.1689 max mem: 3531 +Test: [170/517] eta: 0:01:34 acc1_pos: 72.936 acc5_pos: 100.000 loss: 0.3965 (0.5858) acc1: 86.7188 (69.4384) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2591 data: 0.1654 max mem: 3531 +Test: [180/517] eta: 0:01:31 acc1_pos: 72.936 acc5_pos: 100.000 loss: 0.3774 (0.5752) acc1: 86.7188 (70.3643) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.2489 data: 0.1508 max mem: 3531 +Test: [190/517] eta: 0:01:27 acc1_pos: 72.936 acc5_pos: 100.000 loss: 0.3278 (0.5626) acc1: 92.9688 (71.5628) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2419 data: 0.1430 max mem: 3531 +Test: [200/517] eta: 0:01:24 acc1_pos: 72.936 acc5_pos: 100.000 loss: 0.3244 (0.5545) acc1: 93.2292 (72.3634) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.2365 data: 0.1414 max mem: 3531 +Test: [210/517] eta: 0:01:21 acc1_pos: 72.936 acc5_pos: 100.000 loss: 0.4821 (0.5533) acc1: 81.2500 (72.6094) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.2442 data: 0.1504 max mem: 3532 +Test: [220/517] eta: 0:01:18 acc1_pos: 72.936 acc5_pos: 100.000 loss: 0.4904 (0.5490) acc1: 80.2083 (73.1512) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.2533 data: 0.1580 max mem: 3532 +Test: [230/517] eta: 0:01:15 acc1_pos: 72.936 acc5_pos: 100.000 loss: 0.5019 (0.5488) acc1: 78.3854 (73.1275) acc5: 100.0000 (100.0000) acc1_neg: 74.219 acc5_neg: 100.000 time: 0.2397 data: 0.1459 max mem: 3532 +Test: [240/517] eta: 0:01:13 acc1_pos: 72.936 acc5_pos: 100.000 loss: 0.5483 (0.5489) acc1: 71.0938 (73.0734) acc5: 100.0000 (100.0000) acc1_neg: 64.062 acc5_neg: 100.000 time: 0.2456 data: 0.1537 max mem: 3532 +Test: [250/517] eta: 0:01:10 acc1_pos: 72.936 acc5_pos: 100.000 loss: 0.4761 (0.5450) acc1: 76.3021 (73.5195) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.2489 data: 0.1550 max mem: 3532 +Test: [260/517] eta: 0:01:07 acc1_pos: 72.936 acc5_pos: 100.000 loss: 0.4196 (0.5393) acc1: 87.7604 (74.1589) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.2457 data: 0.1510 max mem: 3532 +Test: [270/517] eta: 0:01:04 acc1_pos: 72.936 acc5_pos: 100.000 loss: 0.4690 (0.5396) acc1: 80.4688 (74.1179) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.2599 data: 0.1661 max mem: 3532 +Test: [280/517] eta: 0:01:01 acc1_pos: 72.936 acc5_pos: 100.000 loss: 0.5567 (0.5404) acc1: 71.3542 (74.0158) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.2389 data: 0.1455 max mem: 3532 +Test: [290/517] eta: 0:00:59 acc1_pos: 72.936 acc5_pos: 100.000 loss: 0.5452 (0.5395) acc1: 73.6979 (74.1633) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.2310 data: 0.1372 max mem: 3532 +Test: [300/517] eta: 0:00:56 acc1_pos: 72.936 acc5_pos: 100.000 loss: 0.4695 (0.5369) acc1: 80.2083 (74.4342) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.2411 data: 0.1490 max mem: 3532 +Test: [310/517] eta: 0:00:53 acc1_pos: 72.936 acc5_pos: 100.000 loss: 0.4536 (0.5338) acc1: 81.5104 (74.7337) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.2493 data: 0.1563 max mem: 3532 +Test: [320/517] eta: 0:00:51 acc1_pos: 72.936 acc5_pos: 100.000 loss: 0.3929 (0.5292) acc1: 86.7188 (75.1493) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.2632 data: 0.1668 max mem: 3532 +Test: [330/517] eta: 0:00:48 acc1_pos: 72.936 acc5_pos: 100.000 loss: 0.3572 (0.5238) acc1: 90.3646 (75.5893) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.2745 data: 0.1762 max mem: 3532 +Test: [340/517] eta: 0:00:45 acc1_pos: 72.936 acc5_pos: 100.000 loss: 0.3440 (0.5216) acc1: 91.1458 (75.7568) acc5: 100.0000 (100.0000) acc1_neg: 53.385 acc5_neg: 100.000 time: 0.2599 data: 0.1614 max mem: 3532 +Test: [350/517] eta: 0:00:43 acc1_pos: 72.936 acc5_pos: 100.000 loss: 0.5684 (0.5260) acc1: 66.1458 (75.2597) acc5: 100.0000 (100.0000) acc1_neg: 49.479 acc5_neg: 100.000 time: 0.2301 data: 0.1329 max mem: 3532 +Test: [360/517] eta: 0:00:40 acc1_pos: 72.936 acc5_pos: 100.000 loss: 0.6189 (0.5264) acc1: 65.1042 (75.1998) acc5: 100.0000 (100.0000) acc1_neg: 75.521 acc5_neg: 100.000 time: 0.2516 data: 0.1552 max mem: 3532 +Test: [370/517] eta: 0:00:38 acc1_pos: 72.936 acc5_pos: 100.000 loss: 0.3544 (0.5207) acc1: 88.2812 (75.7300) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2560 data: 0.1613 max mem: 3532 +Test: [380/517] eta: 0:00:35 acc1_pos: 72.936 acc5_pos: 100.000 loss: 0.3337 (0.5174) acc1: 93.7500 (76.0082) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.3110 data: 0.2174 max mem: 3532 +Test: [390/517] eta: 0:00:32 acc1_pos: 72.936 acc5_pos: 100.000 loss: 0.3911 (0.5148) acc1: 83.5938 (76.1982) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.2686 data: 0.1757 max mem: 3532 +Test: [400/517] eta: 0:00:30 acc1_pos: 72.936 acc5_pos: 100.000 loss: 0.3785 (0.5112) acc1: 85.1562 (76.4755) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.2176 data: 0.1234 max mem: 3532 +Test: [410/517] eta: 0:00:27 acc1_pos: 72.936 acc5_pos: 100.000 loss: 0.4124 (0.5110) acc1: 84.1146 (76.5029) acc5: 100.0000 (100.0000) acc1_neg: 69.010 acc5_neg: 100.000 time: 0.2660 data: 0.1698 max mem: 3532 +Test: [420/517] eta: 0:00:25 acc1_pos: 72.936 acc5_pos: 100.000 loss: 0.4946 (0.5107) acc1: 79.9479 (76.6070) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.2459 data: 0.1508 max mem: 3532 +Test: [430/517] eta: 0:00:22 acc1_pos: 72.936 acc5_pos: 100.000 loss: 0.4885 (0.5099) acc1: 77.8646 (76.6320) acc5: 100.0000 (100.0000) acc1_neg: 62.240 acc5_neg: 100.000 time: 0.2389 data: 0.1435 max mem: 3532 +Test: [440/517] eta: 0:00:19 acc1_pos: 72.936 acc5_pos: 100.000 loss: 0.5132 (0.5116) acc1: 73.9583 (76.4450) acc5: 100.0000 (100.0000) acc1_neg: 63.802 acc5_neg: 100.000 time: 0.2398 data: 0.1435 max mem: 3533 +Test: [450/517] eta: 0:00:17 acc1_pos: 72.936 acc5_pos: 100.000 loss: 0.5000 (0.5104) acc1: 77.0833 (76.6099) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.2399 data: 0.1436 max mem: 3533 +Test: [460/517] eta: 0:00:14 acc1_pos: 72.936 acc5_pos: 100.000 loss: 0.4789 (0.5098) acc1: 81.2500 (76.6890) acc5: 100.0000 (100.0000) acc1_neg: 79.427 acc5_neg: 100.000 time: 0.2396 data: 0.1429 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 72.936 acc5_pos: 100.000 loss: 0.4822 (0.5093) acc1: 79.4271 (76.7333) acc5: 100.0000 (100.0000) acc1_neg: 73.698 acc5_neg: 100.000 time: 0.2505 data: 0.1548 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 72.936 acc5_pos: 100.000 loss: 0.4782 (0.5089) acc1: 76.0417 (76.7693) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.2438 data: 0.1491 max mem: 3533 +Test: [490/517] eta: 0:00:06 acc1_pos: 72.936 acc5_pos: 100.000 loss: 0.4884 (0.5086) acc1: 80.9896 (76.7959) acc5: 100.0000 (100.0000) acc1_neg: 60.156 acc5_neg: 100.000 time: 0.2152 data: 0.1193 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 72.936 acc5_pos: 100.000 loss: 0.6597 (0.5146) acc1: 58.3333 (76.1732) acc5: 100.0000 (100.0000) acc1_neg: 45.052 acc5_neg: 100.000 time: 0.2261 data: 0.1303 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 72.936 acc5_pos: 100.000 loss: 0.7387 (0.5167) acc1: 50.2604 (75.9678) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.2317 data: 0.1371 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 72.936 acc5_pos: 100.000 loss: 0.5576 (0.5166) acc1: 69.0104 (76.0270) acc5: 100.0000 (100.0000) acc1_neg: 72.438 acc5_neg: 100.000 time: 0.2364 data: 0.1370 max mem: 3533 +Test: Total time: 0:02:11 (0.2536 s / it) +* Overall Acc@1 76.027 Acc@5 100.000 + target == real Acc@1 52.043 Acc@5 100.000 + target == fake Acc@1 78.515 Acc@5 100.000 + loss 0.517 + PR-AUC 0.9528 ROC-AUC 0.7134 EER 34.21% (thr=0.5889) +Accuracy of the network on the 198427 test images: 76.0% +[rank0]:[W725 15:44:41.913827067 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0725 15:45:00.293000 73905 torch/distributed/run.py:766] +W0725 15:45:00.293000 73905 torch/distributed/run.py:766] ***************************************** +W0725 15:45:00.293000 73905 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 15:45:00.293000 73905 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W725 15:45:27.757517902 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 15:45:27.862499493 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 15:45:28.108540207 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 15:45:28.333498731 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 15:45:28.362814109 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 15:45:28.405634966 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 15:45:28.427357496 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 15:45:28.538467551 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n16_seed1_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed1_resampled_n116-041-153_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:28:47 acc1_pos: 98.177 acc5_pos: 100.000 loss: 0.2390 (0.2390) acc1: 98.1771 (98.1771) acc5: 100.0000 (100.0000) time: 4.4416 data: 2.4718 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/389] eta: 0:03:30 acc1_pos: 98.958 acc5_pos: 100.000 loss: 0.2390 (0.2312) acc1: 98.1771 (96.2595) acc5: 100.0000 (100.0000) time: 0.5551 data: 0.2893 max mem: 3531 +Test: [ 20/389] eta: 0:02:04 acc1_pos: 98.958 acc5_pos: 100.000 loss: 0.2467 (0.2487) acc1: 98.1771 (95.4489) acc5: 100.0000 (100.0000) time: 0.1311 data: 0.0375 max mem: 3531 +Test: [ 30/389] eta: 0:01:35 acc1_pos: 96.354 acc5_pos: 100.000 loss: 0.1940 (0.2222) acc1: 98.1771 (96.1610) acc5: 100.0000 (100.0000) time: 0.1081 data: 0.0139 max mem: 3531 +Test: [ 40/389] eta: 0:01:30 acc1_pos: 84.896 acc5_pos: 100.000 loss: 0.1991 (0.2510) acc1: 96.3542 (93.9469) acc5: 100.0000 (100.0000) time: 0.1757 data: 0.0818 max mem: 3531 +Test: [ 50/389] eta: 0:01:23 acc1_pos: 85.156 acc5_pos: 100.000 loss: 0.3183 (0.2589) acc1: 85.9375 (93.6275) acc5: 100.0000 (100.0000) time: 0.2159 data: 0.1239 max mem: 3531 +Test: [ 60/389] eta: 0:01:15 acc1_pos: 84.375 acc5_pos: 100.000 loss: 0.2962 (0.2673) acc1: 86.4583 (93.2036) acc5: 100.0000 (100.0000) time: 0.1692 data: 0.0777 max mem: 3531 +Test: [ 70/389] eta: 0:01:10 acc1_pos: 91.927 acc5_pos: 100.000 loss: 0.2178 (0.2646) acc1: 98.1771 (93.5299) acc5: 100.0000 (100.0000) time: 0.1540 data: 0.0617 max mem: 3531 +Test: [ 80/389] eta: 0:01:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2348 (0.2742) acc1: 97.3958 (93.2517) acc5: 100.0000 (100.0000) time: 0.1865 data: 0.0934 max mem: 3531 +Test: [ 90/389] eta: 0:01:03 acc1_pos: 93.490 acc5_pos: 100.000 loss: 0.1875 (0.2650) acc1: 97.1354 (93.7214) acc5: 100.0000 (100.0000) time: 0.1852 data: 0.0942 max mem: 3531 +Test: [100/389] eta: 0:01:00 acc1_pos: 83.073 acc5_pos: 100.000 loss: 0.2005 (0.2753) acc1: 95.3125 (93.0642) acc5: 100.0000 (100.0000) time: 0.1649 data: 0.0719 max mem: 3531 +Test: [110/389] eta: 0:00:56 acc1_pos: 98.958 acc5_pos: 100.000 loss: 0.3340 (0.2904) acc1: 85.9375 (91.8989) acc5: 100.0000 (100.0000) time: 0.1631 data: 0.0685 max mem: 3531 +Test: [120/389] eta: 0:00:53 acc1_pos: 84.896 acc5_pos: 100.000 loss: 0.2723 (0.2902) acc1: 90.8854 (91.9486) acc5: 100.0000 (100.0000) time: 0.1632 data: 0.0708 max mem: 3531 +Test: [130/389] eta: 0:00:51 acc1_pos: 98.160 acc5_pos: 100.000 loss: 0.4730 (0.3539) acc1: 73.1771 (87.9175) acc5: 100.0000 (100.0000) acc1_neg: 48.438 acc5_neg: 100.000 time: 0.1610 data: 0.0690 max mem: 3531 +Test: [140/389] eta: 0:00:47 acc1_pos: 98.160 acc5_pos: 100.000 loss: 1.3641 (0.4317) acc1: 11.4583 (82.5724) acc5: 100.0000 (100.0000) acc1_neg: 4.688 acc5_neg: 100.000 time: 0.1406 data: 0.0483 max mem: 3531 +Test: [150/389] eta: 0:00:45 acc1_pos: 98.160 acc5_pos: 100.000 loss: 1.4239 (0.4899) acc1: 7.5521 (78.7183) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.1449 data: 0.0535 max mem: 3531 +Test: [160/389] eta: 0:00:43 acc1_pos: 98.160 acc5_pos: 100.000 loss: 1.2393 (0.5250) acc1: 15.6250 (76.4929) acc5: 100.0000 (100.0000) acc1_neg: 47.656 acc5_neg: 100.000 time: 0.1623 data: 0.0700 max mem: 3531 +Test: [170/389] eta: 0:00:40 acc1_pos: 98.160 acc5_pos: 100.000 loss: 0.9576 (0.5622) acc1: 34.6354 (73.6294) acc5: 100.0000 (100.0000) acc1_neg: 22.396 acc5_neg: 100.000 time: 0.1615 data: 0.0693 max mem: 3531 +Test: [180/389] eta: 0:00:38 acc1_pos: 98.160 acc5_pos: 100.000 loss: 0.9974 (0.5918) acc1: 33.8542 (71.2635) acc5: 100.0000 (100.0000) acc1_neg: 13.021 acc5_neg: 100.000 time: 0.1568 data: 0.0652 max mem: 3531 +Test: [190/389] eta: 0:00:36 acc1_pos: 98.160 acc5_pos: 100.000 loss: 1.1444 (0.6398) acc1: 19.0104 (68.3314) acc5: 100.0000 (100.0000) acc1_neg: 0.521 acc5_neg: 100.000 time: 0.1565 data: 0.0635 max mem: 3531 +Test: [200/389] eta: 0:00:34 acc1_pos: 98.160 acc5_pos: 100.000 loss: 1.5695 (0.6791) acc1: 2.6042 (65.8206) acc5: 100.0000 (100.0000) acc1_neg: 3.125 acc5_neg: 100.000 time: 0.1855 data: 0.0911 max mem: 3531 +Test: [210/389] eta: 0:00:32 acc1_pos: 98.160 acc5_pos: 100.000 loss: 0.9220 (0.6922) acc1: 16.6667 (64.7487) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.1874 data: 0.0944 max mem: 3532 +Test: [220/389] eta: 0:00:30 acc1_pos: 98.160 acc5_pos: 100.000 loss: 0.6797 (0.7032) acc1: 54.6875 (64.2004) acc5: 100.0000 (100.0000) acc1_neg: 4.948 acc5_neg: 100.000 time: 0.1643 data: 0.0734 max mem: 3532 +Test: [230/389] eta: 0:00:28 acc1_pos: 98.160 acc5_pos: 100.000 loss: 0.8277 (0.7261) acc1: 47.9167 (63.0208) acc5: 100.0000 (100.0000) acc1_neg: 50.521 acc5_neg: 100.000 time: 0.1606 data: 0.0693 max mem: 3532 +Test: [240/389] eta: 0:00:27 acc1_pos: 98.160 acc5_pos: 100.000 loss: 1.0210 (0.7363) acc1: 28.6458 (61.8409) acc5: 100.0000 (100.0000) acc1_neg: 16.146 acc5_neg: 100.000 time: 0.1682 data: 0.0765 max mem: 3532 +Test: [250/389] eta: 0:00:25 acc1_pos: 98.160 acc5_pos: 100.000 loss: 1.0452 (0.7506) acc1: 24.4792 (60.1822) acc5: 100.0000 (100.0000) acc1_neg: 19.792 acc5_neg: 100.000 time: 0.1745 data: 0.0832 max mem: 3532 +Test: [260/389] eta: 0:00:23 acc1_pos: 98.160 acc5_pos: 100.000 loss: 1.1287 (0.7734) acc1: 16.9271 (58.3184) acc5: 100.0000 (100.0000) acc1_neg: 0.000 acc5_neg: 100.000 time: 0.1594 data: 0.0690 max mem: 3532 +Test: [270/389] eta: 0:00:21 acc1_pos: 98.160 acc5_pos: 100.000 loss: 1.3256 (0.8043) acc1: 10.6771 (56.7382) acc5: 100.0000 (100.0000) acc1_neg: 20.833 acc5_neg: 100.000 time: 0.1612 data: 0.0709 max mem: 3532 +Test: [280/389] eta: 0:00:19 acc1_pos: 98.160 acc5_pos: 100.000 loss: 1.4823 (0.8268) acc1: 11.4583 (55.2760) acc5: 100.0000 (100.0000) acc1_neg: 41.406 acc5_neg: 100.000 time: 0.1909 data: 0.1005 max mem: 3532 +Test: [290/389] eta: 0:00:17 acc1_pos: 98.160 acc5_pos: 100.000 loss: 1.1046 (0.8299) acc1: 23.6979 (54.9336) acc5: 100.0000 (100.0000) acc1_neg: 76.042 acc5_neg: 100.000 time: 0.1676 data: 0.0766 max mem: 3532 +Test: [300/389] eta: 0:00:15 acc1_pos: 98.160 acc5_pos: 100.000 loss: 1.3705 (0.8559) acc1: 4.1667 (53.1536) acc5: 100.0000 (100.0000) acc1_neg: 0.260 acc5_neg: 100.000 time: 0.1468 data: 0.0525 max mem: 3532 +Test: [310/389] eta: 0:00:14 acc1_pos: 98.160 acc5_pos: 100.000 loss: 1.6185 (0.8816) acc1: 0.7812 (51.8187) acc5: 100.0000 (100.0000) acc1_neg: 0.000 acc5_neg: 100.000 time: 0.1749 data: 0.0814 max mem: 3532 +Test: [320/389] eta: 0:00:12 acc1_pos: 98.160 acc5_pos: 100.000 loss: 1.5100 (0.8944) acc1: 4.9479 (50.9070) acc5: 100.0000 (100.0000) acc1_neg: 73.958 acc5_neg: 100.000 time: 0.1694 data: 0.0778 max mem: 3532 +Test: [330/389] eta: 0:00:10 acc1_pos: 98.160 acc5_pos: 100.000 loss: 1.7812 (0.9351) acc1: 1.5625 (49.4965) acc5: 100.0000 (100.0000) acc1_neg: 0.000 acc5_neg: 100.000 time: 0.1629 data: 0.0704 max mem: 3532 +Test: [340/389] eta: 0:00:08 acc1_pos: 98.160 acc5_pos: 100.000 loss: 1.0777 (0.9293) acc1: 27.0833 (49.7342) acc5: 100.0000 (100.0000) acc1_neg: 40.625 acc5_neg: 100.000 time: 0.1911 data: 0.0996 max mem: 3532 +Test: [350/389] eta: 0:00:06 acc1_pos: 98.160 acc5_pos: 100.000 loss: 0.8670 (0.9412) acc1: 34.6354 (48.9732) acc5: 100.0000 (100.0000) acc1_neg: 1.042 acc5_neg: 100.000 time: 0.1719 data: 0.0792 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 98.160 acc5_pos: 100.000 loss: 1.1040 (0.9458) acc1: 22.9167 (48.3113) acc5: 100.0000 (100.0000) acc1_neg: 30.729 acc5_neg: 100.000 time: 0.1644 data: 0.0715 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 98.160 acc5_pos: 100.000 loss: 0.9845 (0.9498) acc1: 31.7708 (47.9904) acc5: 100.0000 (100.0000) acc1_neg: 70.312 acc5_neg: 100.000 time: 0.1821 data: 0.0913 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 98.160 acc5_pos: 100.000 loss: 0.9040 (0.9503) acc1: 35.4167 (47.7738) acc5: 100.0000 (100.0000) acc1_neg: 43.490 acc5_neg: 100.000 time: 0.1532 data: 0.0626 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 98.160 acc5_pos: 100.000 loss: 0.8893 (0.9512) acc1: 43.4896 (47.5080) acc5: 100.0000 (100.0000) acc1_neg: 46.995 acc5_neg: 100.000 time: 0.1639 data: 0.0675 max mem: 3532 +Test: Total time: 0:01:08 (0.1773 s / it) +* Overall Acc@1 47.508 Acc@5 100.000 + target == real Acc@1 91.531 Acc@5 100.000 + target == fake Acc@1 27.045 Acc@5 100.000 + loss 0.951 + PR-AUC 0.8099 ROC-AUC 0.6786 EER 37.83% (thr=0.2245) +Accuracy of the network on the 149358 test images: 47.5% +[rank0]:[W725 15:46:52.401788626 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 15:47:09.964000 74462 torch/distributed/run.py:766] +W0725 15:47:09.964000 74462 torch/distributed/run.py:766] ***************************************** +W0725 15:47:09.964000 74462 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 15:47:09.964000 74462 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 15:47:37.838481801 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 15:47:38.238764484 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 15:47:38.260059478 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 15:47:38.693837748 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 15:47:38.729956792 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 15:47:38.923346855 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 15:47:39.134900473 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 15:47:39.245229420 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n16_seed1_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed1_resampled_n116-041-153_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:27:58 acc1_pos: 84.635 acc5_pos: 100.000 loss: 0.4056 (0.4056) acc1: 84.6354 (84.6354) acc5: 100.0000 (100.0000) time: 5.0418 data: 3.0466 max mem: 3531 +Test: [ 10/333] eta: 0:02:54 acc1_pos: 85.677 acc5_pos: 100.000 loss: 0.3491 (0.3439) acc1: 84.6354 (85.1799) acc5: 100.0000 (100.0000) time: 0.5414 data: 0.2784 max mem: 3531 +Test: [ 20/333] eta: 0:01:45 acc1_pos: 94.271 acc5_pos: 100.000 loss: 0.2959 (0.3201) acc1: 87.5000 (86.7436) acc5: 100.0000 (100.0000) time: 0.1031 data: 0.0099 max mem: 3531 +Test: [ 30/333] eta: 0:01:31 acc1_pos: 67.188 acc5_pos: 100.000 loss: 0.3679 (0.4052) acc1: 79.4271 (80.7208) acc5: 100.0000 (100.0000) time: 0.1723 data: 0.0759 max mem: 3531 +Test: [ 40/333] eta: 0:01:25 acc1_pos: 70.312 acc5_pos: 100.000 loss: 0.5631 (0.4526) acc1: 67.1875 (77.3692) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1505 max mem: 3531 +Test: [ 50/333] eta: 0:01:17 acc1_pos: 70.052 acc5_pos: 100.000 loss: 0.5593 (0.4735) acc1: 70.3125 (76.0110) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1350 max mem: 3531 +Test: [ 60/333] eta: 0:01:16 acc1_pos: 62.760 acc5_pos: 100.000 loss: 0.5500 (0.4926) acc1: 70.0521 (74.9915) acc5: 100.0000 (100.0000) time: 0.2599 data: 0.1655 max mem: 3531 +Test: [ 70/333] eta: 0:01:09 acc1_pos: 64.844 acc5_pos: 100.000 loss: 0.5648 (0.5021) acc1: 69.2708 (74.3215) acc5: 100.0000 (100.0000) time: 0.2349 data: 0.1406 max mem: 3531 +Test: [ 80/333] eta: 0:01:05 acc1_pos: 70.052 acc5_pos: 100.000 loss: 0.5669 (0.5117) acc1: 68.4896 (73.6143) acc5: 100.0000 (100.0000) time: 0.1852 data: 0.0908 max mem: 3531 +Test: [ 90/333] eta: 0:01:00 acc1_pos: 77.083 acc5_pos: 100.000 loss: 0.5465 (0.5111) acc1: 73.1771 (73.8897) acc5: 100.0000 (100.0000) time: 0.2049 data: 0.1108 max mem: 3531 +Test: [100/333] eta: 0:00:56 acc1_pos: 79.167 acc5_pos: 100.000 loss: 0.4862 (0.5107) acc1: 75.0000 (73.8552) acc5: 100.0000 (100.0000) time: 0.1895 data: 0.0944 max mem: 3531 +Test: [110/333] eta: 0:00:53 acc1_pos: 82.812 acc5_pos: 100.000 loss: 0.5096 (0.5118) acc1: 73.1771 (73.7566) acc5: 100.0000 (100.0000) time: 0.1899 data: 0.0938 max mem: 3531 +Test: [120/333] eta: 0:00:49 acc1_pos: 78.646 acc5_pos: 100.000 loss: 0.5325 (0.5132) acc1: 73.1771 (73.6979) acc5: 100.0000 (100.0000) time: 0.1849 data: 0.0887 max mem: 3531 +Test: [130/333] eta: 0:00:46 acc1_pos: 88.021 acc5_pos: 100.000 loss: 0.5197 (0.5107) acc1: 74.2188 (74.0041) acc5: 100.0000 (100.0000) time: 0.1849 data: 0.0910 max mem: 3531 +Test: [140/333] eta: 0:00:44 acc1_pos: 85.417 acc5_pos: 100.000 loss: 0.4640 (0.5067) acc1: 77.8646 (74.4552) acc5: 100.0000 (100.0000) time: 0.1899 data: 0.0939 max mem: 3531 +Test: [150/333] eta: 0:00:41 acc1_pos: 76.562 acc5_pos: 100.000 loss: 0.4849 (0.5074) acc1: 76.5625 (74.3360) acc5: 100.0000 (100.0000) time: 0.1799 data: 0.0847 max mem: 3531 +Test: [160/333] eta: 0:00:38 acc1_pos: 73.355 acc5_pos: 100.000 loss: 0.5180 (0.5253) acc1: 72.3958 (73.1269) acc5: 100.0000 (100.0000) acc1_neg: 26.823 acc5_neg: 100.000 time: 0.1699 data: 0.0776 max mem: 3531 +Test: [170/333] eta: 0:00:35 acc1_pos: 73.355 acc5_pos: 100.000 loss: 1.0302 (0.5570) acc1: 33.5938 (70.7694) acc5: 100.0000 (100.0000) acc1_neg: 37.500 acc5_neg: 100.000 time: 0.1849 data: 0.0907 max mem: 3531 +Test: [180/333] eta: 0:00:33 acc1_pos: 73.355 acc5_pos: 100.000 loss: 0.9886 (0.5779) acc1: 36.9792 (69.0694) acc5: 100.0000 (100.0000) acc1_neg: 38.281 acc5_neg: 100.000 time: 0.1975 data: 0.1014 max mem: 3531 +Test: [190/333] eta: 0:00:30 acc1_pos: 73.355 acc5_pos: 100.000 loss: 0.8636 (0.5890) acc1: 42.7083 (68.1092) acc5: 100.0000 (100.0000) acc1_neg: 43.229 acc5_neg: 100.000 time: 0.1849 data: 0.0903 max mem: 3531 +Test: [200/333] eta: 0:00:28 acc1_pos: 73.355 acc5_pos: 100.000 loss: 0.8203 (0.6032) acc1: 43.2292 (66.9634) acc5: 100.0000 (100.0000) acc1_neg: 39.062 acc5_neg: 100.000 time: 0.1723 data: 0.0770 max mem: 3531 +Test: [210/333] eta: 0:00:26 acc1_pos: 73.355 acc5_pos: 100.000 loss: 0.9455 (0.6202) acc1: 40.6250 (65.6633) acc5: 100.0000 (100.0000) acc1_neg: 44.010 acc5_neg: 100.000 time: 0.1899 data: 0.0908 max mem: 3532 +Test: [220/333] eta: 0:00:24 acc1_pos: 73.355 acc5_pos: 100.000 loss: 0.9560 (0.6342) acc1: 40.1042 (64.5963) acc5: 100.0000 (100.0000) acc1_neg: 43.750 acc5_neg: 100.000 time: 0.2149 data: 0.1183 max mem: 3532 +Test: [230/333] eta: 0:00:22 acc1_pos: 73.355 acc5_pos: 100.000 loss: 0.8852 (0.6443) acc1: 41.1458 (63.6071) acc5: 100.0000 (100.0000) acc1_neg: 38.021 acc5_neg: 100.000 time: 0.2199 data: 0.1257 max mem: 3532 +Test: [240/333] eta: 0:00:19 acc1_pos: 73.355 acc5_pos: 100.000 loss: 0.8922 (0.6573) acc1: 38.0208 (62.4525) acc5: 100.0000 (100.0000) acc1_neg: 36.198 acc5_neg: 100.000 time: 0.1836 data: 0.0894 max mem: 3532 +Test: [250/333] eta: 0:00:17 acc1_pos: 73.355 acc5_pos: 100.000 loss: 0.9380 (0.6664) acc1: 38.0208 (61.7156) acc5: 100.0000 (100.0000) acc1_neg: 42.969 acc5_neg: 100.000 time: 0.1799 data: 0.0862 max mem: 3532 +Test: [260/333] eta: 0:00:15 acc1_pos: 73.355 acc5_pos: 100.000 loss: 0.8606 (0.6732) acc1: 44.7917 (61.2328) acc5: 100.0000 (100.0000) acc1_neg: 46.875 acc5_neg: 100.000 time: 0.2326 data: 0.1382 max mem: 3532 +Test: [270/333] eta: 0:00:13 acc1_pos: 73.355 acc5_pos: 100.000 loss: 0.8652 (0.6815) acc1: 45.5729 (60.5992) acc5: 100.0000 (100.0000) acc1_neg: 43.750 acc5_neg: 100.000 time: 0.2299 data: 0.1361 max mem: 3532 +Test: [280/333] eta: 0:00:11 acc1_pos: 73.355 acc5_pos: 100.000 loss: 0.9095 (0.6923) acc1: 40.8854 (59.7262) acc5: 100.0000 (100.0000) acc1_neg: 40.885 acc5_neg: 100.000 time: 0.1603 data: 0.0654 max mem: 3532 +Test: [290/333] eta: 0:00:08 acc1_pos: 73.355 acc5_pos: 100.000 loss: 1.0530 (0.7057) acc1: 31.5104 (58.7495) acc5: 100.0000 (100.0000) acc1_neg: 30.990 acc5_neg: 100.000 time: 0.1505 data: 0.0571 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 73.355 acc5_pos: 100.000 loss: 1.0628 (0.7169) acc1: 31.7708 (57.9224) acc5: 100.0000 (100.0000) acc1_neg: 34.115 acc5_neg: 100.000 time: 0.1737 data: 0.0835 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 73.355 acc5_pos: 100.000 loss: 0.9516 (0.7188) acc1: 38.2812 (57.7731) acc5: 100.0000 (100.0000) acc1_neg: 43.490 acc5_neg: 100.000 time: 0.1581 data: 0.0680 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 73.355 acc5_pos: 100.000 loss: 0.8165 (0.7230) acc1: 45.8333 (57.3468) acc5: 100.0000 (100.0000) acc1_neg: 41.667 acc5_neg: 100.000 time: 0.1295 data: 0.0395 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 73.355 acc5_pos: 100.000 loss: 0.9903 (0.7326) acc1: 40.3646 (56.7685) acc5: 100.0000 (100.0000) acc1_neg: 39.323 acc5_neg: 100.000 time: 0.1057 data: 0.0158 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 73.355 acc5_pos: 100.000 loss: 1.0230 (0.7344) acc1: 39.8438 (56.6742) acc5: 100.0000 (100.0000) acc1_neg: 39.718 acc5_neg: 100.000 time: 0.1106 data: 0.0176 max mem: 3532 +Test: Total time: 0:01:06 (0.1994 s / it) +* Overall Acc@1 56.674 Acc@5 100.000 + target == real Acc@1 74.404 Acc@5 100.000 + target == fake Acc@1 40.892 Acc@5 100.000 + loss 0.734 + PR-AUC 0.6374 ROC-AUC 0.6333 EER 40.71% (thr=0.3735) +Accuracy of the network on the 127843 test images: 56.7% +[rank0]:[W725 15:48:58.054331507 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 15:49:15.037000 75019 torch/distributed/run.py:766] +W0725 15:49:15.037000 75019 torch/distributed/run.py:766] ***************************************** +W0725 15:49:15.037000 75019 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 15:49:15.037000 75019 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 15:49:43.307329787 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 15:49:43.756966124 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 15:49:43.905680309 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 15:49:44.107508452 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +| distributed init (rank 4): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 15:49:44.297077261 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank4]:[W725 15:49:44.297196860 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W725 15:49:44.297334544 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 15:49:44.422463001 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n16_seed1_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed1_resampled_n116-041-153_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:26:06 acc1_pos: 97.656 acc5_pos: 100.000 loss: 0.1695 (0.1695) acc1: 97.6562 (97.6562) acc5: 100.0000 (100.0000) time: 6.6661 data: 5.2925 max mem: 3531 +Test: [ 10/235] eta: 0:02:40 acc1_pos: 95.833 acc5_pos: 100.000 loss: 0.1906 (0.1886) acc1: 96.6146 (96.3068) acc5: 100.0000 (100.0000) time: 0.7124 data: 0.5027 max mem: 3531 +Test: [ 20/235] eta: 0:01:30 acc1_pos: 96.615 acc5_pos: 100.000 loss: 0.1880 (0.1880) acc1: 96.6146 (96.4162) acc5: 100.0000 (100.0000) time: 0.1070 data: 0.0121 max mem: 3531 +Test: [ 30/235] eta: 0:01:17 acc1_pos: 96.615 acc5_pos: 100.000 loss: 0.1861 (0.1871) acc1: 96.8750 (96.6146) acc5: 100.0000 (100.0000) time: 0.1954 data: 0.1022 max mem: 3531 +Test: [ 40/235] eta: 0:01:07 acc1_pos: 96.354 acc5_pos: 100.000 loss: 0.1855 (0.1878) acc1: 96.8750 (96.5892) acc5: 100.0000 (100.0000) time: 0.2649 data: 0.1726 max mem: 3531 +Test: [ 50/235] eta: 0:01:00 acc1_pos: 94.271 acc5_pos: 100.000 loss: 0.1879 (0.1892) acc1: 96.3542 (96.4920) acc5: 100.0000 (100.0000) time: 0.2429 data: 0.1484 max mem: 3531 +Test: [ 60/235] eta: 0:00:54 acc1_pos: 96.354 acc5_pos: 100.000 loss: 0.1953 (0.1908) acc1: 96.0938 (96.4481) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1439 max mem: 3531 +Test: [ 70/235] eta: 0:00:47 acc1_pos: 96.923 acc5_pos: 100.000 loss: 0.2079 (0.3393) acc1: 95.0521 (87.1699) acc5: 100.0000 (100.0000) acc1_neg: 20.573 acc5_neg: 100.000 time: 0.1902 data: 0.0953 max mem: 3531 +Test: [ 80/235] eta: 0:00:43 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.3723 (0.4671) acc1: 23.1771 (79.2374) acc5: 100.0000 (100.0000) acc1_neg: 19.792 acc5_neg: 100.000 time: 0.1899 data: 0.0975 max mem: 3531 +Test: [ 90/235] eta: 0:00:39 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.3719 (0.5666) acc1: 22.3958 (72.9625) acc5: 100.0000 (100.0000) acc1_neg: 22.396 acc5_neg: 100.000 time: 0.2246 data: 0.1294 max mem: 3531 +Test: [100/235] eta: 0:00:35 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.3698 (0.6460) acc1: 22.1354 (68.0487) acc5: 100.0000 (100.0000) acc1_neg: 22.135 acc5_neg: 100.000 time: 0.1970 data: 0.1025 max mem: 3531 +Test: [110/235] eta: 0:00:32 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.3566 (0.7100) acc1: 22.1354 (64.0038) acc5: 100.0000 (100.0000) acc1_neg: 25.781 acc5_neg: 100.000 time: 0.1915 data: 0.1005 max mem: 3531 +Test: [120/235] eta: 0:00:29 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.4064 (0.7752) acc1: 19.7917 (60.0077) acc5: 100.0000 (100.0000) acc1_neg: 16.146 acc5_neg: 100.000 time: 0.2203 data: 0.1294 max mem: 3531 +Test: [130/235] eta: 0:00:26 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.4940 (0.8296) acc1: 15.8854 (56.6834) acc5: 100.0000 (100.0000) acc1_neg: 16.927 acc5_neg: 100.000 time: 0.2333 data: 0.1426 max mem: 3531 +Test: [140/235] eta: 0:00:23 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.4940 (0.8785) acc1: 15.6250 (53.7326) acc5: 100.0000 (100.0000) acc1_neg: 11.719 acc5_neg: 100.000 time: 0.2203 data: 0.1287 max mem: 3531 +Test: [150/235] eta: 0:00:21 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.5005 (0.9189) acc1: 15.1042 (51.1848) acc5: 100.0000 (100.0000) acc1_neg: 15.885 acc5_neg: 100.000 time: 0.2050 data: 0.1133 max mem: 3531 +Test: [160/235] eta: 0:00:18 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.4959 (0.9546) acc1: 15.3646 (48.9891) acc5: 100.0000 (100.0000) acc1_neg: 15.365 acc5_neg: 100.000 time: 0.2220 data: 0.1316 max mem: 3531 +Test: [170/235] eta: 0:00:15 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.4278 (0.9675) acc1: 19.0104 (48.0537) acc5: 100.0000 (100.0000) acc1_neg: 34.115 acc5_neg: 100.000 time: 0.2202 data: 0.1298 max mem: 3531 +Test: [180/235] eta: 0:00:13 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.1413 (0.9763) acc1: 34.8958 (47.3743) acc5: 100.0000 (100.0000) acc1_neg: 37.240 acc5_neg: 100.000 time: 0.2115 data: 0.1210 max mem: 3531 +Test: [190/235] eta: 0:00:10 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.1276 (0.9844) acc1: 34.8958 (46.7305) acc5: 100.0000 (100.0000) acc1_neg: 36.198 acc5_neg: 100.000 time: 0.1995 data: 0.1076 max mem: 3531 +Test: [200/235] eta: 0:00:08 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.1380 (0.9926) acc1: 34.1146 (46.0938) acc5: 100.0000 (100.0000) acc1_neg: 32.812 acc5_neg: 100.000 time: 0.1602 data: 0.0682 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.1443 (0.9989) acc1: 34.1146 (45.5939) acc5: 100.0000 (100.0000) acc1_neg: 34.375 acc5_neg: 100.000 time: 0.1506 data: 0.0596 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.1184 (1.0046) acc1: 35.1562 (45.1357) acc5: 100.0000 (100.0000) acc1_neg: 34.375 acc5_neg: 100.000 time: 0.1353 data: 0.0453 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.1244 (1.0099) acc1: 35.4167 (44.7556) acc5: 100.0000 (100.0000) acc1_neg: 35.417 acc5_neg: 100.000 time: 0.1188 data: 0.0287 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.1244 (1.0124) acc1: 35.6771 (44.6400) acc5: 100.0000 (100.0000) acc1_neg: 34.783 acc5_neg: 100.000 time: 0.1088 data: 0.0215 max mem: 3532 +Test: Total time: 0:00:51 (0.2207 s / it) +* Overall Acc@1 44.640 Acc@5 100.000 + target == real Acc@1 96.435 Acc@5 100.000 + target == fake Acc@1 25.907 Acc@5 100.000 + loss 1.012 + PR-AUC 0.9022 ROC-AUC 0.7799 EER 29.35% (thr=0.1652) +Accuracy of the network on the 89879 test images: 44.6% +[rank0]:[W725 15:50:45.659640572 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 15:51:02.420000 75576 torch/distributed/run.py:766] +W0725 15:51:02.420000 75576 torch/distributed/run.py:766] ***************************************** +W0725 15:51:02.420000 75576 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 15:51:02.420000 75576 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 15:51:27.272427568 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 15:51:27.653468067 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 15:51:28.185941409 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 15:51:28.253539844 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 15:51:28.293194086 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 15:51:28.426057253 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 15:51:28.454336883 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 15:51:28.531941982 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n16_seed1_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed1_resampled_n116-041-153_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:25:13 acc1_pos: 88.021 acc5_pos: 100.000 loss: 0.3746 (0.3746) acc1: 88.0208 (88.0208) acc5: 100.0000 (100.0000) time: 5.6269 data: 4.0971 max mem: 3531 +Test: [ 10/269] eta: 0:02:33 acc1_pos: 94.792 acc5_pos: 100.000 loss: 0.3746 (0.3733) acc1: 89.3229 (89.1099) acc5: 100.0000 (100.0000) time: 0.5933 data: 0.3728 max mem: 3531 +Test: [ 20/269] eta: 0:01:28 acc1_pos: 96.615 acc5_pos: 100.000 loss: 0.3438 (0.3501) acc1: 93.2292 (91.6171) acc5: 100.0000 (100.0000) time: 0.0918 data: 0.0004 max mem: 3531 +Test: [ 30/269] eta: 0:01:07 acc1_pos: 91.146 acc5_pos: 100.000 loss: 0.2990 (0.3399) acc1: 94.7917 (92.1539) acc5: 100.0000 (100.0000) time: 0.1085 data: 0.0143 max mem: 3531 +Test: [ 40/269] eta: 0:00:58 acc1_pos: 92.241 acc5_pos: 100.000 loss: 0.3986 (0.4880) acc1: 77.8646 (78.5061) acc5: 100.0000 (100.0000) acc1_neg: 24.219 acc5_neg: 100.000 time: 0.1473 data: 0.0543 max mem: 3531 +Test: [ 50/269] eta: 0:00:53 acc1_pos: 92.241 acc5_pos: 100.000 loss: 1.0521 (0.6142) acc1: 24.2188 (67.0905) acc5: 100.0000 (100.0000) acc1_neg: 8.594 acc5_neg: 100.000 time: 0.1832 data: 0.0897 max mem: 3531 +Test: [ 60/269] eta: 0:00:47 acc1_pos: 92.241 acc5_pos: 100.000 loss: 1.1539 (0.7161) acc1: 13.8021 (58.0900) acc5: 100.0000 (100.0000) acc1_neg: 11.458 acc5_neg: 100.000 time: 0.1717 data: 0.0742 max mem: 3531 +Test: [ 70/269] eta: 0:00:44 acc1_pos: 92.241 acc5_pos: 100.000 loss: 1.1795 (0.7743) acc1: 14.0625 (52.6665) acc5: 100.0000 (100.0000) acc1_neg: 13.802 acc5_neg: 100.000 time: 0.1691 data: 0.0712 max mem: 3531 +Test: [ 80/269] eta: 0:00:41 acc1_pos: 92.241 acc5_pos: 100.000 loss: 1.0632 (0.8013) acc1: 18.7500 (49.5724) acc5: 100.0000 (100.0000) acc1_neg: 48.698 acc5_neg: 100.000 time: 0.1899 data: 0.0938 max mem: 3531 +Test: [ 90/269] eta: 0:00:38 acc1_pos: 92.241 acc5_pos: 100.000 loss: 1.0340 (0.8295) acc1: 23.4375 (46.9437) acc5: 100.0000 (100.0000) acc1_neg: 20.312 acc5_neg: 100.000 time: 0.1899 data: 0.0909 max mem: 3531 +Test: [100/269] eta: 0:00:35 acc1_pos: 92.241 acc5_pos: 100.000 loss: 1.1169 (0.8666) acc1: 19.2708 (43.9460) acc5: 100.0000 (100.0000) acc1_neg: 7.552 acc5_neg: 100.000 time: 0.1749 data: 0.0750 max mem: 3531 +Test: [110/269] eta: 0:00:33 acc1_pos: 92.241 acc5_pos: 100.000 loss: 1.2024 (0.9008) acc1: 14.3229 (41.0895) acc5: 100.0000 (100.0000) acc1_neg: 13.802 acc5_neg: 100.000 time: 0.1899 data: 0.0936 max mem: 3531 +Test: [120/269] eta: 0:00:30 acc1_pos: 92.241 acc5_pos: 100.000 loss: 1.2529 (0.9309) acc1: 15.1042 (39.2627) acc5: 100.0000 (100.0000) acc1_neg: 14.062 acc5_neg: 100.000 time: 0.1954 data: 0.0997 max mem: 3531 +Test: [130/269] eta: 0:00:28 acc1_pos: 92.241 acc5_pos: 100.000 loss: 1.2746 (0.9611) acc1: 15.1042 (37.3191) acc5: 100.0000 (100.0000) acc1_neg: 16.927 acc5_neg: 100.000 time: 0.1899 data: 0.0908 max mem: 3531 +Test: [140/269] eta: 0:00:26 acc1_pos: 92.241 acc5_pos: 100.000 loss: 1.2520 (0.9741) acc1: 15.8854 (35.9098) acc5: 100.0000 (100.0000) acc1_neg: 19.271 acc5_neg: 100.000 time: 0.1994 data: 0.1008 max mem: 3531 +Test: [150/269] eta: 0:00:24 acc1_pos: 92.241 acc5_pos: 100.000 loss: 1.1058 (0.9827) acc1: 17.9688 (34.7061) acc5: 100.0000 (100.0000) acc1_neg: 27.344 acc5_neg: 100.000 time: 0.1799 data: 0.0858 max mem: 3531 +Test: [160/269] eta: 0:00:21 acc1_pos: 92.241 acc5_pos: 100.000 loss: 1.1166 (0.9964) acc1: 15.8854 (33.5565) acc5: 100.0000 (100.0000) acc1_neg: 20.573 acc5_neg: 100.000 time: 0.1699 data: 0.0765 max mem: 3531 +Test: [170/269] eta: 0:00:19 acc1_pos: 92.241 acc5_pos: 100.000 loss: 1.2692 (1.0144) acc1: 13.0208 (32.3374) acc5: 100.0000 (100.0000) acc1_neg: 17.448 acc5_neg: 100.000 time: 0.1749 data: 0.0815 max mem: 3531 +Test: [180/269] eta: 0:00:17 acc1_pos: 92.241 acc5_pos: 100.000 loss: 1.2299 (1.0239) acc1: 14.0625 (31.3982) acc5: 100.0000 (100.0000) acc1_neg: 15.365 acc5_neg: 100.000 time: 0.1799 data: 0.0856 max mem: 3531 +Test: [190/269] eta: 0:00:15 acc1_pos: 92.241 acc5_pos: 100.000 loss: 1.2179 (1.0357) acc1: 12.7604 (30.4456) acc5: 100.0000 (100.0000) acc1_neg: 9.115 acc5_neg: 100.000 time: 0.1799 data: 0.0841 max mem: 3531 +Test: [200/269] eta: 0:00:13 acc1_pos: 92.241 acc5_pos: 100.000 loss: 1.3225 (1.0549) acc1: 9.1146 (29.3597) acc5: 100.0000 (100.0000) acc1_neg: 14.844 acc5_neg: 100.000 time: 0.1899 data: 0.0939 max mem: 3531 +Test: [210/269] eta: 0:00:11 acc1_pos: 92.241 acc5_pos: 100.000 loss: 1.2994 (1.0644) acc1: 9.6354 (28.6162) acc5: 100.0000 (100.0000) acc1_neg: 14.583 acc5_neg: 100.000 time: 0.2199 data: 0.1239 max mem: 3532 +Test: [220/269] eta: 0:00:09 acc1_pos: 92.241 acc5_pos: 100.000 loss: 1.2118 (1.0707) acc1: 14.5833 (28.2016) acc5: 100.0000 (100.0000) acc1_neg: 46.615 acc5_neg: 100.000 time: 0.1949 data: 0.0988 max mem: 3532 +Test: [230/269] eta: 0:00:07 acc1_pos: 92.241 acc5_pos: 100.000 loss: 1.1244 (1.0720) acc1: 19.0104 (28.1442) acc5: 100.0000 (100.0000) acc1_neg: 25.781 acc5_neg: 100.000 time: 0.1615 data: 0.0657 max mem: 3532 +Test: [240/269] eta: 0:00:05 acc1_pos: 92.241 acc5_pos: 100.000 loss: 1.1805 (1.0790) acc1: 18.4896 (27.4875) acc5: 100.0000 (100.0000) acc1_neg: 21.615 acc5_neg: 100.000 time: 0.1899 data: 0.0950 max mem: 3532 +Test: [250/269] eta: 0:00:03 acc1_pos: 92.241 acc5_pos: 100.000 loss: 1.2782 (1.0902) acc1: 9.3750 (26.9723) acc5: 100.0000 (100.0000) acc1_neg: 6.510 acc5_neg: 100.000 time: 0.1984 data: 0.1038 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 92.241 acc5_pos: 100.000 loss: 1.4442 (1.1019) acc1: 5.9896 (26.2442) acc5: 100.0000 (100.0000) acc1_neg: 16.927 acc5_neg: 100.000 time: 0.1804 data: 0.0863 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 92.241 acc5_pos: 100.000 loss: 1.3285 (1.1047) acc1: 6.5104 (25.8749) acc5: 100.0000 (100.0000) acc1_neg: 21.522 acc5_neg: 100.000 time: 0.1407 data: 0.0441 max mem: 3532 +Test: Total time: 0:00:52 (0.1943 s / it) +* Overall Acc@1 25.875 Acc@5 100.000 + target == real Acc@1 91.970 Acc@5 100.000 + target == fake Acc@1 16.855 Acc@5 100.000 + loss 1.105 + PR-AUC 0.9173 ROC-AUC 0.6171 EER 42.09% (thr=0.2673) +Accuracy of the network on the 103293 test images: 25.9% +[rank0]:[W725 15:52:32.859842042 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-25T15:52:38Z] eval_done tag=fixed500k_fake_n16_seed1_resampled +[2026-07-25T15:52:38Z] eval_start tag=fixed500k_fake_n16_seed2_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed2_resampled_n116-041-153_20260725_112559/best_checkpoint.pth +started_utc=2026-07-25T15:52:38Z +hostname=n116-042-186 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed2_resampled_n116-041-153_20260725_112559/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n16_seed2_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0725 15:52:49.909000 76137 torch/distributed/run.py:766] +W0725 15:52:49.909000 76137 torch/distributed/run.py:766] ***************************************** +W0725 15:52:49.909000 76137 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 15:52:49.909000 76137 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 15:53:14.007124548 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 15:53:15.492019451 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 15:53:15.555230560 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W725 15:53:15.772659731 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank6]:[W725 15:53:15.777322235 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 15:53:15.889196364 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 15:53:15.996702980 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 15:53:16.065143006 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n16_seed2_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed2_resampled_n116-041-153_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:19:19 acc1_pos: 45.833 acc5_pos: 100.000 loss: 0.7780 (0.7780) acc1: 45.8333 (45.8333) acc5: 100.0000 (100.0000) time: 4.9570 data: 3.1519 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/234] eta: 0:02:09 acc1_pos: 47.917 acc5_pos: 100.000 loss: 0.7550 (0.6864) acc1: 50.7812 (58.8778) acc5: 100.0000 (100.0000) time: 0.5776 data: 0.3321 max mem: 3531 +Test: [ 20/234] eta: 0:01:15 acc1_pos: 94.271 acc5_pos: 100.000 loss: 0.7191 (0.6855) acc1: 50.7812 (58.0729) acc5: 100.0000 (100.0000) time: 0.1207 data: 0.0290 max mem: 3531 +Test: [ 30/234] eta: 0:00:56 acc1_pos: 54.427 acc5_pos: 100.000 loss: 0.2076 (0.5541) acc1: 94.2708 (67.9688) acc5: 100.0000 (100.0000) time: 0.1128 data: 0.0173 max mem: 3531 +Test: [ 40/234] eta: 0:00:58 acc1_pos: 95.052 acc5_pos: 100.000 loss: 0.2016 (0.5390) acc1: 94.7917 (68.7119) acc5: 100.0000 (100.0000) time: 0.2469 data: 0.1522 max mem: 3531 +Test: [ 50/234] eta: 0:00:56 acc1_pos: 96.094 acc5_pos: 100.000 loss: 0.1549 (0.4626) acc1: 95.5729 (74.1268) acc5: 100.0000 (100.0000) time: 0.3509 data: 0.2591 max mem: 3531 +Test: [ 60/234] eta: 0:00:48 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.1434 (0.4100) acc1: 97.3958 (78.3086) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2418 data: 0.1483 max mem: 3531 +Test: [ 70/234] eta: 0:00:42 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.1414 (0.4452) acc1: 99.7396 (76.4488) acc5: 100.0000 (100.0000) acc1_neg: 42.969 acc5_neg: 100.000 time: 0.1398 data: 0.0450 max mem: 3531 +Test: [ 80/234] eta: 0:00:38 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.8653 (0.4689) acc1: 51.3021 (75.3311) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1630 data: 0.0682 max mem: 3531 +Test: [ 90/234] eta: 0:00:35 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.2824 (0.4449) acc1: 90.8854 (77.2379) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.1893 data: 0.0945 max mem: 3531 +Test: [100/234] eta: 0:00:32 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.2874 (0.4560) acc1: 86.4583 (76.9157) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2049 data: 0.1097 max mem: 3531 +Test: [110/234] eta: 0:00:29 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.5510 (0.4589) acc1: 70.0521 (77.0505) acc5: 100.0000 (100.0000) acc1_neg: 68.490 acc5_neg: 100.000 time: 0.1946 data: 0.0996 max mem: 3531 +Test: [120/234] eta: 0:00:26 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.5510 (0.4690) acc1: 73.1771 (76.5647) acc5: 100.0000 (100.0000) acc1_neg: 48.698 acc5_neg: 100.000 time: 0.1776 data: 0.0840 max mem: 3531 +Test: [130/234] eta: 0:00:23 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.4597 (0.4737) acc1: 74.2188 (76.4552) acc5: 100.0000 (100.0000) acc1_neg: 65.885 acc5_neg: 100.000 time: 0.2049 data: 0.1104 max mem: 3531 +Test: [140/234] eta: 0:00:21 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.5293 (0.5045) acc1: 68.7500 (75.2992) acc5: 100.0000 (100.0000) acc1_neg: 34.115 acc5_neg: 100.000 time: 0.2082 data: 0.1153 max mem: 3531 +Test: [150/234] eta: 0:00:19 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.6386 (0.5207) acc1: 67.9688 (74.6723) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.2011 data: 0.1103 max mem: 3531 +Test: [160/234] eta: 0:00:16 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.4955 (0.5139) acc1: 76.5625 (75.2766) acc5: 100.0000 (100.0000) acc1_neg: 78.646 acc5_neg: 100.000 time: 0.1989 data: 0.1062 max mem: 3531 +Test: [170/234] eta: 0:00:14 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.4669 (0.5124) acc1: 78.6458 (75.4249) acc5: 100.0000 (100.0000) acc1_neg: 68.750 acc5_neg: 100.000 time: 0.2104 data: 0.1160 max mem: 3531 +Test: [180/234] eta: 0:00:12 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.5011 (0.5285) acc1: 76.0417 (74.6619) acc5: 100.0000 (100.0000) acc1_neg: 26.823 acc5_neg: 100.000 time: 0.2049 data: 0.1124 max mem: 3531 +Test: [190/234] eta: 0:00:09 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.6284 (0.5350) acc1: 67.4479 (74.2651) acc5: 100.0000 (100.0000) acc1_neg: 70.052 acc5_neg: 100.000 time: 0.1894 data: 0.0946 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.5845 (0.5234) acc1: 70.3125 (74.9728) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1855 data: 0.0912 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.1194 (0.5059) acc1: 99.7396 (76.0910) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1600 data: 0.0691 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.2129 (0.4928) acc1: 96.8750 (77.0056) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1393 data: 0.0485 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.2336 (0.4826) acc1: 95.5729 (77.7440) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1351 data: 0.0455 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 93.793 acc5_pos: 100.000 loss: 0.2432 (0.4797) acc1: 95.0521 (77.9302) acc5: 100.0000 (100.0000) acc1_neg: 95.331 acc5_neg: 100.000 time: 0.1332 data: 0.0424 max mem: 3532 +Test: Total time: 0:00:48 (0.2083 s / it) +* Overall Acc@1 77.930 Acc@5 100.000 + target == real Acc@1 74.271 Acc@5 100.000 + target == fake Acc@1 78.961 Acc@5 100.000 + loss 0.480 + PR-AUC 0.9566 ROC-AUC 0.8609 EER 23.11% (thr=0.5366) +Accuracy of the network on the 89729 test images: 77.9% +[rank0]:[W725 15:54:20.378757248 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 15:54:37.995000 76695 torch/distributed/run.py:766] +W0725 15:54:37.995000 76695 torch/distributed/run.py:766] ***************************************** +W0725 15:54:37.995000 76695 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 15:54:37.995000 76695 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 15:55:04.474804532 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 15:55:04.624943474 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 15:55:04.825337574 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 15:55:04.858627043 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 15:55:04.990958731 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +NCCL version 2.26.2+cuda12.2 +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 15:55:05.042186956 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 15:55:05.144825782 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 15:55:05.190065876 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n16_seed2_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed2_resampled_n116-041-153_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:45:14 acc1_pos: 41.667 acc5_pos: 100.000 loss: 0.8722 (0.8722) acc1: 41.6667 (41.6667) acc5: 100.0000 (100.0000) time: 7.1698 data: 5.6796 max mem: 3531 +Test: [ 10/2387] eta: 0:29:21 acc1_pos: 17.448 acc5_pos: 100.000 loss: 1.1540 (1.0830) acc1: 26.3021 (28.1723) acc5: 100.0000 (100.0000) time: 0.7410 data: 0.5167 max mem: 3531 +Test: [ 20/2387] eta: 0:17:11 acc1_pos: 12.500 acc5_pos: 100.000 loss: 1.1361 (1.0946) acc1: 23.6979 (27.7406) acc5: 100.0000 (100.0000) time: 0.0989 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:14:53 acc1_pos: 1.823 acc5_pos: 100.000 loss: 1.1361 (1.1411) acc1: 21.6146 (25.1428) acc5: 100.0000 (100.0000) time: 0.1798 data: 0.0800 max mem: 3531 +Test: [ 40/2387] eta: 0:13:37 acc1_pos: 20.052 acc5_pos: 100.000 loss: 1.1749 (1.1525) acc1: 17.4479 (23.3168) acc5: 100.0000 (100.0000) time: 0.2565 data: 0.1596 max mem: 3531 +Test: [ 50/2387] eta: 0:12:51 acc1_pos: 40.625 acc5_pos: 100.000 loss: 1.1365 (1.1399) acc1: 19.7917 (24.0605) acc5: 100.0000 (100.0000) time: 0.2549 data: 0.1596 max mem: 3531 +Test: [ 60/2387] eta: 0:12:11 acc1_pos: 57.292 acc5_pos: 100.000 loss: 0.9898 (1.1138) acc1: 30.9896 (25.7727) acc5: 100.0000 (100.0000) time: 0.2448 data: 0.1491 max mem: 3531 +Test: [ 70/2387] eta: 0:11:43 acc1_pos: 32.031 acc5_pos: 100.000 loss: 1.0263 (1.1123) acc1: 29.1667 (25.7336) acc5: 100.0000 (100.0000) time: 0.2350 data: 0.1426 max mem: 3531 +Test: [ 80/2387] eta: 0:11:30 acc1_pos: 43.229 acc5_pos: 100.000 loss: 1.1048 (1.1033) acc1: 25.2604 (26.5111) acc5: 100.0000 (100.0000) time: 0.2535 data: 0.1605 max mem: 3531 +Test: [ 90/2387] eta: 0:11:17 acc1_pos: 36.198 acc5_pos: 100.000 loss: 0.9140 (1.0794) acc1: 35.9375 (27.9190) acc5: 100.0000 (100.0000) time: 0.2648 data: 0.1721 max mem: 3531 +Test: [ 100/2387] eta: 0:11:06 acc1_pos: 35.938 acc5_pos: 100.000 loss: 0.8929 (1.0675) acc1: 35.9375 (28.9114) acc5: 100.0000 (100.0000) time: 0.2599 data: 0.1672 max mem: 3531 +Test: [ 110/2387] eta: 0:10:55 acc1_pos: 37.500 acc5_pos: 100.000 loss: 0.9425 (1.0544) acc1: 35.9375 (29.6734) acc5: 100.0000 (100.0000) time: 0.2549 data: 0.1600 max mem: 3531 +Test: [ 120/2387] eta: 0:10:48 acc1_pos: 35.938 acc5_pos: 100.000 loss: 0.9466 (1.0492) acc1: 35.9375 (29.8812) acc5: 100.0000 (100.0000) time: 0.2599 data: 0.1670 max mem: 3531 +Test: [ 130/2387] eta: 0:10:45 acc1_pos: 32.552 acc5_pos: 100.000 loss: 0.9780 (1.0472) acc1: 32.5521 (29.6974) acc5: 100.0000 (100.0000) time: 0.2749 data: 0.1826 max mem: 3531 +Test: [ 140/2387] eta: 0:10:36 acc1_pos: 19.531 acc5_pos: 100.000 loss: 1.0321 (1.0446) acc1: 27.6042 (29.7743) acc5: 100.0000 (100.0000) time: 0.2649 data: 0.1700 max mem: 3531 +Test: [ 150/2387] eta: 0:10:24 acc1_pos: 15.104 acc5_pos: 100.000 loss: 1.0311 (1.0413) acc1: 30.2083 (29.9048) acc5: 100.0000 (100.0000) time: 0.2349 data: 0.1415 max mem: 3531 +Test: [ 160/2387] eta: 0:10:14 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.9223 (1.0176) acc1: 34.1146 (31.6786) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2249 data: 0.1305 max mem: 3531 +Test: [ 170/2387] eta: 0:09:56 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2806 (0.9739) acc1: 94.7917 (35.4136) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1952 data: 0.0999 max mem: 3531 +Test: [ 180/2387] eta: 0:09:42 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2677 (0.9351) acc1: 95.0521 (38.6870) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1649 data: 0.0717 max mem: 3531 +Test: [ 190/2387] eta: 0:09:35 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2677 (0.9002) acc1: 94.7917 (41.6489) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2008 data: 0.1081 max mem: 3531 +Test: [ 200/2387] eta: 0:09:25 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2635 (0.8687) acc1: 94.7917 (44.3253) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2099 data: 0.1173 max mem: 3531 +Test: [ 210/2387] eta: 0:09:16 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2705 (0.8405) acc1: 94.7917 (46.7220) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1937 data: 0.1002 max mem: 3532 +Test: [ 220/2387] eta: 0:09:09 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2680 (0.8145) acc1: 95.3125 (48.9418) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2049 data: 0.1104 max mem: 3532 +Test: [ 230/2387] eta: 0:09:02 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2669 (0.7909) acc1: 95.5729 (50.9391) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2069 data: 0.1112 max mem: 3532 +Test: [ 240/2387] eta: 0:08:53 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2730 (0.7696) acc1: 94.0104 (52.7220) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1916 data: 0.0983 max mem: 3532 +Test: [ 250/2387] eta: 0:08:48 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2677 (0.7439) acc1: 95.0521 (54.5775) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1987 data: 0.1072 max mem: 3532 +Test: [ 260/2387] eta: 0:08:44 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1022 (0.7193) acc1: 100.0000 (56.3178) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2240 data: 0.1314 max mem: 3532 +Test: [ 270/2387] eta: 0:08:33 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1000 (0.6966) acc1: 100.0000 (57.9297) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1841 data: 0.0908 max mem: 3532 +Test: [ 280/2387] eta: 0:08:20 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1030 (0.6755) acc1: 100.0000 (59.4269) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1191 data: 0.0226 max mem: 3532 +Test: [ 290/2387] eta: 0:08:19 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1027 (0.6559) acc1: 100.0000 (60.8212) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1771 data: 0.0800 max mem: 3532 +Test: [ 300/2387] eta: 0:08:16 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1027 (0.6375) acc1: 100.0000 (62.1228) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2453 data: 0.1527 max mem: 3532 +Test: [ 310/2387] eta: 0:08:10 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1065 (0.6242) acc1: 100.0000 (63.2612) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2076 data: 0.1129 max mem: 3532 +Test: [ 320/2387] eta: 0:08:06 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.3639 (0.6163) acc1: 95.8333 (64.2718) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1943 data: 0.0965 max mem: 3532 +Test: [ 330/2387] eta: 0:08:00 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.3658 (0.6087) acc1: 95.8333 (65.2308) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1964 data: 0.1005 max mem: 3532 +Test: [ 340/2387] eta: 0:07:56 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.3634 (0.6014) acc1: 95.8333 (66.1351) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1949 data: 0.1010 max mem: 3532 +Test: [ 350/2387] eta: 0:07:53 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.3494 (0.5891) acc1: 96.8750 (67.0836) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2086 data: 0.1148 max mem: 3532 +Test: [ 360/2387] eta: 0:07:46 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1983 (0.5785) acc1: 97.9167 (67.9341) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1855 data: 0.0880 max mem: 3532 +Test: [ 370/2387] eta: 0:07:41 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2124 (0.5686) acc1: 97.6562 (68.7296) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1706 data: 0.0722 max mem: 3532 +Test: [ 380/2387] eta: 0:07:36 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2138 (0.5592) acc1: 97.6562 (69.4978) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1794 data: 0.0834 max mem: 3532 +Test: [ 390/2387] eta: 0:07:33 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2113 (0.5503) acc1: 97.9167 (70.2253) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1890 data: 0.0899 max mem: 3532 +Test: [ 400/2387] eta: 0:07:28 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2113 (0.5419) acc1: 98.1771 (70.9236) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1898 data: 0.0887 max mem: 3532 +Test: [ 410/2387] eta: 0:07:24 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2106 (0.5337) acc1: 98.1771 (71.5823) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1822 data: 0.0835 max mem: 3532 +Test: [ 420/2387] eta: 0:07:19 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2100 (0.5260) acc1: 98.1771 (72.2115) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1748 data: 0.0769 max mem: 3532 +Test: [ 430/2387] eta: 0:07:14 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2122 (0.5189) acc1: 97.3958 (72.7868) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1626 data: 0.0648 max mem: 3532 +Test: [ 440/2387] eta: 0:07:09 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2117 (0.5108) acc1: 97.3958 (73.3749) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1664 data: 0.0702 max mem: 3533 +Test: [ 450/2387] eta: 0:07:04 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1206 (0.5021) acc1: 100.0000 (73.9653) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1654 data: 0.0711 max mem: 3533 +Test: [ 460/2387] eta: 0:06:59 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1207 (0.4939) acc1: 100.0000 (74.5294) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1494 data: 0.0547 max mem: 3533 +Test: [ 470/2387] eta: 0:06:54 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1213 (0.4859) acc1: 100.0000 (75.0702) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1500 data: 0.0560 max mem: 3533 +Test: [ 480/2387] eta: 0:06:51 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1213 (0.4784) acc1: 100.0000 (75.5885) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1767 data: 0.0854 max mem: 3533 +Test: [ 490/2387] eta: 0:06:47 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1208 (0.4711) acc1: 100.0000 (76.0857) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1843 data: 0.0920 max mem: 3533 +Test: [ 500/2387] eta: 0:06:44 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1217 (0.4641) acc1: 100.0000 (76.5620) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1791 data: 0.0805 max mem: 3533 +Test: [ 510/2387] eta: 0:06:41 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1219 (0.4576) acc1: 100.0000 (77.0207) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0869 max mem: 3533 +Test: [ 520/2387] eta: 0:06:39 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1509 (0.4518) acc1: 100.0000 (77.4617) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1953 data: 0.1033 max mem: 3533 +Test: [ 530/2387] eta: 0:06:36 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1509 (0.4461) acc1: 100.0000 (77.8857) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2048 data: 0.1113 max mem: 3533 +Test: [ 540/2387] eta: 0:06:34 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1495 (0.4406) acc1: 100.0000 (78.2944) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2161 data: 0.1227 max mem: 3533 +Test: [ 550/2387] eta: 0:06:32 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1501 (0.4354) acc1: 100.0000 (78.6879) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2223 data: 0.1291 max mem: 3533 +Test: [ 560/2387] eta: 0:06:28 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1502 (0.4303) acc1: 100.0000 (79.0678) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1812 data: 0.0865 max mem: 3533 +Test: [ 570/2387] eta: 0:06:26 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1497 (0.4254) acc1: 100.0000 (79.4344) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1875 data: 0.0928 max mem: 3533 +Test: [ 580/2387] eta: 0:06:24 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1494 (0.4206) acc1: 100.0000 (79.7875) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2099 data: 0.1151 max mem: 3533 +Test: [ 590/2387] eta: 0:06:20 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1504 (0.4161) acc1: 100.0000 (80.1290) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1754 data: 0.0823 max mem: 3533 +Test: [ 600/2387] eta: 0:06:17 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1532 (0.4118) acc1: 100.0000 (80.4597) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0864 max mem: 3533 +Test: [ 610/2387] eta: 0:06:14 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1499 (0.4074) acc1: 100.0000 (80.7782) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1854 data: 0.0918 max mem: 3533 +Test: [ 620/2387] eta: 0:06:11 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1456 (0.4032) acc1: 100.0000 (81.0865) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1824 data: 0.0893 max mem: 3533 +Test: [ 630/2387] eta: 0:06:09 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1459 (0.3992) acc1: 100.0000 (81.3837) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1939 data: 0.0981 max mem: 3533 +Test: [ 640/2387] eta: 0:06:06 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1480 (0.3952) acc1: 100.0000 (81.6717) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1874 data: 0.0873 max mem: 3533 +Test: [ 650/2387] eta: 0:06:04 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1463 (0.3914) acc1: 100.0000 (81.9520) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1980 data: 0.1013 max mem: 3533 +Test: [ 660/2387] eta: 0:06:02 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1452 (0.3877) acc1: 100.0000 (82.2247) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2055 data: 0.1116 max mem: 3533 +Test: [ 670/2387] eta: 0:05:59 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1445 (0.3841) acc1: 100.0000 (82.4888) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1934 data: 0.0997 max mem: 3534 +Test: [ 680/2387] eta: 0:05:57 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1445 (0.3806) acc1: 100.0000 (82.7460) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1993 data: 0.1076 max mem: 3534 +Test: [ 690/2387] eta: 0:05:54 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1474 (0.3818) acc1: 100.0000 (82.6606) acc5: 100.0000 (100.0000) acc1_neg: 66.667 acc5_neg: 100.000 time: 0.1842 data: 0.0914 max mem: 3534 +Test: [ 700/2387] eta: 0:05:51 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.5773 (0.3847) acc1: 68.2292 (82.4529) acc5: 100.0000 (100.0000) acc1_neg: 70.312 acc5_neg: 100.000 time: 0.1668 data: 0.0740 max mem: 3534 +Test: [ 710/2387] eta: 0:05:48 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.5827 (0.3876) acc1: 67.9688 (82.2437) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.1746 data: 0.0819 max mem: 3534 +Test: [ 720/2387] eta: 0:05:45 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.5827 (0.3903) acc1: 67.1875 (82.0367) acc5: 100.0000 (100.0000) acc1_neg: 68.750 acc5_neg: 100.000 time: 0.1780 data: 0.0848 max mem: 3534 +Test: [ 730/2387] eta: 0:05:42 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.5931 (0.3931) acc1: 67.1875 (81.8253) acc5: 100.0000 (100.0000) acc1_neg: 65.625 acc5_neg: 100.000 time: 0.1760 data: 0.0834 max mem: 3534 +Test: [ 740/2387] eta: 0:05:40 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.5815 (0.3956) acc1: 67.7083 (81.6412) acc5: 100.0000 (100.0000) acc1_neg: 67.708 acc5_neg: 100.000 time: 0.1776 data: 0.0856 max mem: 3534 +Test: [ 750/2387] eta: 0:05:37 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.5814 (0.3982) acc1: 67.4479 (81.4504) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.1818 data: 0.0890 max mem: 3534 +Test: [ 760/2387] eta: 0:05:34 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.5828 (0.3994) acc1: 67.4479 (81.3715) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1772 data: 0.0853 max mem: 3534 +Test: [ 770/2387] eta: 0:05:32 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.3344 (0.3979) acc1: 86.7188 (81.4905) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1745 data: 0.0825 max mem: 3534 +Test: [ 780/2387] eta: 0:05:29 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.3136 (0.3969) acc1: 89.0625 (81.5924) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1703 data: 0.0774 max mem: 3534 +Test: [ 790/2387] eta: 0:05:26 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2325 (0.3956) acc1: 93.4896 (81.7073) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.1745 data: 0.0809 max mem: 3534 +Test: [ 800/2387] eta: 0:05:24 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2059 (0.3933) acc1: 97.3958 (81.9071) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1826 data: 0.0902 max mem: 3534 +Test: [ 810/2387] eta: 0:05:21 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2033 (0.3916) acc1: 97.3958 (82.0746) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1689 data: 0.0764 max mem: 3534 +Test: [ 820/2387] eta: 0:05:18 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2181 (0.3896) acc1: 96.6146 (82.2460) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.1718 data: 0.0782 max mem: 3534 +Test: [ 830/2387] eta: 0:05:15 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2537 (0.3888) acc1: 94.5312 (82.3412) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1632 data: 0.0693 max mem: 3534 +Test: [ 840/2387] eta: 0:05:12 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2503 (0.3869) acc1: 94.7917 (82.5174) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.1499 data: 0.0568 max mem: 3534 +Test: [ 850/2387] eta: 0:05:10 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2204 (0.3850) acc1: 97.1354 (82.6803) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1607 data: 0.0695 max mem: 3534 +Test: [ 860/2387] eta: 0:05:08 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2070 (0.3828) acc1: 97.3958 (82.8564) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1849 data: 0.0934 max mem: 3534 +Test: [ 870/2387] eta: 0:05:05 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1838 (0.3807) acc1: 98.4375 (83.0376) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1819 data: 0.0875 max mem: 3534 +Test: [ 880/2387] eta: 0:05:02 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2095 (0.3789) acc1: 97.9167 (83.1917) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1655 data: 0.0691 max mem: 3534 +Test: [ 890/2387] eta: 0:05:00 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2095 (0.3769) acc1: 97.9167 (83.3517) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1885 data: 0.0926 max mem: 3534 +Test: [ 900/2387] eta: 0:04:57 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2392 (0.3789) acc1: 93.4896 (83.2278) acc5: 100.0000 (100.0000) acc1_neg: 68.750 acc5_neg: 100.000 time: 0.1746 data: 0.0807 max mem: 3535 +Test: [ 910/2387] eta: 0:04:55 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.6335 (0.3817) acc1: 66.1458 (83.0409) acc5: 100.0000 (100.0000) acc1_neg: 66.146 acc5_neg: 100.000 time: 0.1554 data: 0.0637 max mem: 3535 +Test: [ 920/2387] eta: 0:04:53 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.6349 (0.3846) acc1: 66.4062 (82.8575) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.1854 data: 0.0941 max mem: 3535 +Test: [ 930/2387] eta: 0:04:50 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.6451 (0.3873) acc1: 65.8854 (82.6732) acc5: 100.0000 (100.0000) acc1_neg: 64.062 acc5_neg: 100.000 time: 0.1830 data: 0.0901 max mem: 3535 +Test: [ 940/2387] eta: 0:04:48 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.6419 (0.3900) acc1: 65.3646 (82.4879) acc5: 100.0000 (100.0000) acc1_neg: 65.625 acc5_neg: 100.000 time: 0.1771 data: 0.0825 max mem: 3535 +Test: [ 950/2387] eta: 0:04:47 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.6419 (0.3926) acc1: 65.3646 (82.3147) acc5: 100.0000 (100.0000) acc1_neg: 66.927 acc5_neg: 100.000 time: 0.2099 data: 0.1156 max mem: 3535 +Test: [ 960/2387] eta: 0:04:44 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.6231 (0.3951) acc1: 66.9271 (82.1640) acc5: 100.0000 (100.0000) acc1_neg: 69.792 acc5_neg: 100.000 time: 0.2024 data: 0.1076 max mem: 3535 +Test: [ 970/2387] eta: 0:04:42 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.6258 (0.3976) acc1: 66.1458 (82.0007) acc5: 100.0000 (100.0000) acc1_neg: 65.365 acc5_neg: 100.000 time: 0.1750 data: 0.0818 max mem: 3535 +Test: [ 980/2387] eta: 0:04:40 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.6191 (0.3974) acc1: 66.9271 (82.0291) acc5: 100.0000 (100.0000) acc1_neg: 66.667 acc5_neg: 100.000 time: 0.2009 data: 0.1095 max mem: 3535 +Test: [ 990/2387] eta: 0:04:38 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.3029 (0.3972) acc1: 91.1458 (82.0654) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.2098 data: 0.1184 max mem: 3535 +Test: [1000/2387] eta: 0:04:37 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.3551 (0.3973) acc1: 87.5000 (82.0802) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.2046 data: 0.1120 max mem: 3535 +Test: [1010/2387] eta: 0:04:34 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.3830 (0.3973) acc1: 84.6354 (82.1116) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.2064 data: 0.1132 max mem: 3535 +Test: [1020/2387] eta: 0:04:33 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.3486 (0.3960) acc1: 88.2812 (82.2154) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2000 data: 0.1073 max mem: 3535 +Test: [1030/2387] eta: 0:04:31 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1756 (0.3938) acc1: 99.7396 (82.3816) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2036 data: 0.1122 max mem: 3535 +Test: [1040/2387] eta: 0:04:29 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1421 (0.3914) acc1: 99.7396 (82.5478) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2008 data: 0.1096 max mem: 3535 +Test: [1050/2387] eta: 0:04:27 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1374 (0.3890) acc1: 99.7396 (82.7099) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2051 data: 0.1092 max mem: 3535 +Test: [1060/2387] eta: 0:04:25 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1381 (0.3868) acc1: 100.0000 (82.8699) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2126 data: 0.1143 max mem: 3535 +Test: [1070/2387] eta: 0:04:22 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1561 (0.3847) acc1: 99.7396 (83.0192) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1820 data: 0.0863 max mem: 3535 +Test: [1080/2387] eta: 0:04:20 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1870 (0.3833) acc1: 98.9583 (83.1372) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.1673 data: 0.0747 max mem: 3535 +Test: [1090/2387] eta: 0:04:18 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.3016 (0.3843) acc1: 91.4062 (83.0844) acc5: 100.0000 (100.0000) acc1_neg: 55.469 acc5_neg: 100.000 time: 0.1963 data: 0.1057 max mem: 3535 +Test: [1100/2387] eta: 0:04:16 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.4463 (0.3848) acc1: 79.9479 (83.0793) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1877 data: 0.0964 max mem: 3535 +Test: [1110/2387] eta: 0:04:14 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.4489 (0.3857) acc1: 79.9479 (83.0342) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1766 data: 0.0836 max mem: 3535 +Test: [1120/2387] eta: 0:04:12 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.4273 (0.3860) acc1: 82.5521 (83.0525) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.1811 data: 0.0892 max mem: 3536 +Test: [1130/2387] eta: 0:04:09 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.3946 (0.3863) acc1: 85.6771 (83.0556) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.1893 data: 0.0989 max mem: 3536 +Test: [1140/2387] eta: 0:04:07 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.3690 (0.3863) acc1: 86.7188 (83.0891) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1899 data: 0.0986 max mem: 3536 +Test: [1150/2387] eta: 0:04:05 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2834 (0.3850) acc1: 91.4062 (83.1919) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1973 data: 0.1052 max mem: 3536 +Test: [1160/2387] eta: 0:04:03 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1474 (0.3829) acc1: 99.7396 (83.3318) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1880 data: 0.0966 max mem: 3536 +Test: [1170/2387] eta: 0:04:01 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1440 (0.3808) acc1: 99.7396 (83.4697) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1764 data: 0.0841 max mem: 3536 +Test: [1180/2387] eta: 0:03:59 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1388 (0.3788) acc1: 99.7396 (83.6061) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1868 data: 0.0944 max mem: 3536 +Test: [1190/2387] eta: 0:03:57 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1470 (0.3770) acc1: 99.7396 (83.7402) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1953 data: 0.1043 max mem: 3536 +Test: [1200/2387] eta: 0:03:55 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1447 (0.3750) acc1: 99.7396 (83.8726) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1941 data: 0.1032 max mem: 3536 +Test: [1210/2387] eta: 0:03:53 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1429 (0.3731) acc1: 100.0000 (84.0038) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1895 data: 0.0984 max mem: 3536 +Test: [1220/2387] eta: 0:03:51 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1356 (0.3711) acc1: 99.7396 (84.1319) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1832 data: 0.0911 max mem: 3536 +Test: [1230/2387] eta: 0:03:48 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1272 (0.3691) acc1: 100.0000 (84.2599) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0819 max mem: 3536 +Test: [1240/2387] eta: 0:03:46 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1274 (0.3672) acc1: 100.0000 (84.3857) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1814 data: 0.0884 max mem: 3536 +Test: [1250/2387] eta: 0:03:44 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1274 (0.3652) acc1: 100.0000 (84.5101) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0826 max mem: 3536 +Test: [1260/2387] eta: 0:03:42 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1259 (0.3634) acc1: 100.0000 (84.6319) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1940 data: 0.1008 max mem: 3536 +Test: [1270/2387] eta: 0:03:40 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1236 (0.3615) acc1: 100.0000 (84.7518) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1951 data: 0.1025 max mem: 3536 +Test: [1280/2387] eta: 0:03:38 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1233 (0.3596) acc1: 99.7396 (84.8668) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1769 data: 0.0829 max mem: 3536 +Test: [1290/2387] eta: 0:03:36 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1285 (0.3581) acc1: 99.2188 (84.9614) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1746 data: 0.0802 max mem: 3536 +Test: [1300/2387] eta: 0:03:34 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1367 (0.3565) acc1: 98.9583 (85.0652) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1738 data: 0.0829 max mem: 3536 +Test: [1310/2387] eta: 0:03:32 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1366 (0.3549) acc1: 98.9583 (85.1666) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1817 data: 0.0910 max mem: 3536 +Test: [1320/2387] eta: 0:03:29 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1324 (0.3534) acc1: 99.4792 (85.2653) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1700 data: 0.0767 max mem: 3536 +Test: [1330/2387] eta: 0:03:27 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1493 (0.3523) acc1: 99.2188 (85.3535) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1688 data: 0.0748 max mem: 3536 +Test: [1340/2387] eta: 0:03:25 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2435 (0.3526) acc1: 94.5312 (85.3378) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0927 max mem: 3536 +Test: [1350/2387] eta: 0:03:23 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1013 (0.3507) acc1: 100.0000 (85.4464) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1998 data: 0.1076 max mem: 3537 +Test: [1360/2387] eta: 0:03:22 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.0909 (0.3488) acc1: 100.0000 (85.5533) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2188 data: 0.1255 max mem: 3537 +Test: [1370/2387] eta: 0:03:20 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.0913 (0.3470) acc1: 100.0000 (85.6585) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2101 data: 0.1181 max mem: 3537 +Test: [1380/2387] eta: 0:03:17 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.0913 (0.3451) acc1: 100.0000 (85.7623) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1822 data: 0.0914 max mem: 3537 +Test: [1390/2387] eta: 0:03:16 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.0895 (0.3433) acc1: 100.0000 (85.8647) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1913 data: 0.0992 max mem: 3537 +Test: [1400/2387] eta: 0:03:14 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.0933 (0.3415) acc1: 100.0000 (85.9656) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2019 data: 0.1102 max mem: 3537 +Test: [1410/2387] eta: 0:03:12 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.0998 (0.3405) acc1: 100.0000 (86.0418) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1900 data: 0.0976 max mem: 3537 +Test: [1420/2387] eta: 0:03:09 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2182 (0.3401) acc1: 97.1354 (86.0929) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1849 data: 0.0916 max mem: 3537 +Test: [1430/2387] eta: 0:03:07 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2248 (0.3393) acc1: 97.6562 (86.1635) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1875 data: 0.0923 max mem: 3537 +Test: [1440/2387] eta: 0:03:05 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2186 (0.3385) acc1: 98.1771 (86.2462) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1710 data: 0.0738 max mem: 3537 +Test: [1450/2387] eta: 0:03:04 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2167 (0.3378) acc1: 97.9167 (86.3130) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1949 data: 0.0994 max mem: 3537 +Test: [1460/2387] eta: 0:03:01 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1907 (0.3366) acc1: 97.9167 (86.3929) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2102 data: 0.1175 max mem: 3537 +Test: [1470/2387] eta: 0:02:59 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1462 (0.3352) acc1: 99.2188 (86.4813) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1758 data: 0.0850 max mem: 3537 +Test: [1480/2387] eta: 0:02:57 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1312 (0.3339) acc1: 99.4792 (86.5682) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1897 data: 0.0976 max mem: 3537 +Test: [1490/2387] eta: 0:02:55 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1312 (0.3326) acc1: 99.4792 (86.6534) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1846 data: 0.0908 max mem: 3537 +Test: [1500/2387] eta: 0:02:53 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1332 (0.3313) acc1: 99.2188 (86.7368) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1837 data: 0.0895 max mem: 3537 +Test: [1510/2387] eta: 0:02:51 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1318 (0.3300) acc1: 99.7396 (86.8225) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1993 data: 0.1042 max mem: 3537 +Test: [1520/2387] eta: 0:02:49 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1495 (0.3290) acc1: 99.4792 (86.9009) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1916 data: 0.0984 max mem: 3537 +Test: [1530/2387] eta: 0:02:47 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1740 (0.3280) acc1: 98.9583 (86.9778) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1900 data: 0.0992 max mem: 3537 +Test: [1540/2387] eta: 0:02:46 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1710 (0.3270) acc1: 98.9583 (87.0547) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2007 data: 0.1103 max mem: 3537 +Test: [1550/2387] eta: 0:02:44 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1551 (0.3260) acc1: 99.4792 (87.1345) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1966 data: 0.1056 max mem: 3537 +Test: [1560/2387] eta: 0:02:42 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1582 (0.3250) acc1: 99.4792 (87.2061) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1846 data: 0.0932 max mem: 3537 +Test: [1570/2387] eta: 0:02:40 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1646 (0.3240) acc1: 99.2188 (87.2782) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2130 data: 0.1186 max mem: 3537 +Test: [1580/2387] eta: 0:02:38 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1646 (0.3234) acc1: 98.9583 (87.3284) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.1980 data: 0.1016 max mem: 3538 +Test: [1590/2387] eta: 0:02:35 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2166 (0.3230) acc1: 95.8333 (87.3702) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1620 data: 0.0659 max mem: 3538 +Test: [1600/2387] eta: 0:02:33 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2171 (0.3227) acc1: 95.5729 (87.4019) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1708 data: 0.0765 max mem: 3538 +Test: [1610/2387] eta: 0:02:31 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2018 (0.3220) acc1: 98.4375 (87.4652) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1678 data: 0.0756 max mem: 3538 +Test: [1620/2387] eta: 0:02:29 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2210 (0.3216) acc1: 96.3542 (87.5095) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1686 data: 0.0735 max mem: 3538 +Test: [1630/2387] eta: 0:02:27 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2376 (0.3211) acc1: 94.5312 (87.5600) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1622 data: 0.0670 max mem: 3538 +Test: [1640/2387] eta: 0:02:25 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2376 (0.3206) acc1: 96.8750 (87.6109) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1453 data: 0.0524 max mem: 3538 +Test: [1650/2387] eta: 0:02:23 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2340 (0.3201) acc1: 96.6146 (87.6656) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1546 data: 0.0633 max mem: 3538 +Test: [1660/2387] eta: 0:02:21 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2173 (0.3195) acc1: 97.3958 (87.7240) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1645 data: 0.0732 max mem: 3538 +Test: [1670/2387] eta: 0:02:19 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2115 (0.3190) acc1: 98.1771 (87.7729) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1679 data: 0.0738 max mem: 3538 +Test: [1680/2387] eta: 0:02:17 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2245 (0.3186) acc1: 96.8750 (87.8199) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1658 data: 0.0726 max mem: 3538 +Test: [1690/2387] eta: 0:02:15 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2245 (0.3181) acc1: 95.8333 (87.8654) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1749 data: 0.0802 max mem: 3538 +Test: [1700/2387] eta: 0:02:13 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2325 (0.3176) acc1: 95.8333 (87.9176) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1840 data: 0.0851 max mem: 3538 +Test: [1710/2387] eta: 0:02:11 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1865 (0.3167) acc1: 99.2188 (87.9805) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1651 data: 0.0670 max mem: 3538 +Test: [1720/2387] eta: 0:02:08 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1618 (0.3158) acc1: 99.4792 (88.0434) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1506 data: 0.0553 max mem: 3538 +Test: [1730/2387] eta: 0:02:06 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1618 (0.3150) acc1: 99.4792 (88.1031) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1656 data: 0.0737 max mem: 3538 +Test: [1740/2387] eta: 0:02:05 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1575 (0.3142) acc1: 99.4792 (88.1640) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1992 data: 0.1066 max mem: 3538 +Test: [1750/2387] eta: 0:02:03 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1604 (0.3134) acc1: 99.4792 (88.2255) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2056 data: 0.1113 max mem: 3538 +Test: [1760/2387] eta: 0:02:01 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1657 (0.3125) acc1: 99.4792 (88.2864) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1899 data: 0.0939 max mem: 3538 +Test: [1770/2387] eta: 0:01:59 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1674 (0.3118) acc1: 99.4792 (88.3418) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.2355 data: 0.1409 max mem: 3538 +Test: [1780/2387] eta: 0:01:57 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1687 (0.3109) acc1: 98.9583 (88.4033) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2015 data: 0.1100 max mem: 3538 +Test: [1790/2387] eta: 0:01:55 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1570 (0.3100) acc1: 99.2188 (88.4623) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1888 data: 0.0958 max mem: 3538 +Test: [1800/2387] eta: 0:01:53 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1606 (0.3092) acc1: 99.4792 (88.5223) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2393 data: 0.1448 max mem: 3538 +Test: [1810/2387] eta: 0:01:51 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1517 (0.3083) acc1: 99.4792 (88.5829) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1587 data: 0.0625 max mem: 3539 +Test: [1820/2387] eta: 0:01:49 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1489 (0.3074) acc1: 99.7396 (88.6426) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1389 data: 0.0445 max mem: 3539 +Test: [1830/2387] eta: 0:01:47 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1404 (0.3064) acc1: 99.7396 (88.7039) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0918 max mem: 3539 +Test: [1840/2387] eta: 0:01:45 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.0994 (0.3053) acc1: 100.0000 (88.7650) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1923 data: 0.0974 max mem: 3539 +Test: [1850/2387] eta: 0:01:43 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.0982 (0.3042) acc1: 100.0000 (88.8256) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1999 data: 0.1062 max mem: 3539 +Test: [1860/2387] eta: 0:01:41 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.0981 (0.3031) acc1: 100.0000 (88.8856) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1977 data: 0.1054 max mem: 3539 +Test: [1870/2387] eta: 0:01:40 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.0981 (0.3020) acc1: 100.0000 (88.9450) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1966 data: 0.1053 max mem: 3539 +Test: [1880/2387] eta: 0:01:38 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.0975 (0.3009) acc1: 100.0000 (89.0037) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1996 data: 0.1083 max mem: 3539 +Test: [1890/2387] eta: 0:01:36 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.0978 (0.2998) acc1: 100.0000 (89.0613) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2018 data: 0.1046 max mem: 3539 +Test: [1900/2387] eta: 0:01:34 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1001 (0.2988) acc1: 100.0000 (89.1187) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2019 data: 0.1045 max mem: 3539 +Test: [1910/2387] eta: 0:01:32 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1156 (0.2980) acc1: 100.0000 (89.1684) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.2073 data: 0.1150 max mem: 3539 +Test: [1920/2387] eta: 0:01:30 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1548 (0.2981) acc1: 99.2188 (89.1627) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.2128 data: 0.1214 max mem: 3539 +Test: [1930/2387] eta: 0:01:28 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2315 (0.2977) acc1: 96.0938 (89.2064) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1987 data: 0.1073 max mem: 3539 +Test: [1940/2387] eta: 0:01:26 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2273 (0.2979) acc1: 97.3958 (89.2086) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1958 data: 0.1031 max mem: 3539 +Test: [1950/2387] eta: 0:01:24 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2115 (0.2974) acc1: 97.9167 (89.2522) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2012 data: 0.1080 max mem: 3539 +Test: [1960/2387] eta: 0:01:22 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2024 (0.2972) acc1: 97.3958 (89.2786) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2348 data: 0.1408 max mem: 3539 +Test: [1970/2387] eta: 0:01:20 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2736 (0.2975) acc1: 92.4479 (89.2632) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.2294 data: 0.1367 max mem: 3539 +Test: [1980/2387] eta: 0:01:18 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2858 (0.2972) acc1: 90.6250 (89.2895) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1597 data: 0.0667 max mem: 3539 +Test: [1990/2387] eta: 0:01:16 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1176 (0.2963) acc1: 98.9583 (89.3389) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1521 data: 0.0545 max mem: 3539 +Test: [2000/2387] eta: 0:01:14 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1164 (0.2954) acc1: 98.9583 (89.3880) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1770 data: 0.0819 max mem: 3539 +Test: [2010/2387] eta: 0:01:13 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1147 (0.2945) acc1: 99.2188 (89.4377) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1942 data: 0.1007 max mem: 3539 +Test: [2020/2387] eta: 0:01:11 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1147 (0.2936) acc1: 99.2188 (89.4870) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2171 data: 0.1214 max mem: 3539 +Test: [2030/2387] eta: 0:01:09 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1172 (0.2928) acc1: 99.2188 (89.5345) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2234 data: 0.1308 max mem: 3540 +Test: [2040/2387] eta: 0:01:07 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1172 (0.2919) acc1: 99.2188 (89.5823) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2014 data: 0.1106 max mem: 3540 +Test: [2050/2387] eta: 0:01:05 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1136 (0.2910) acc1: 99.2188 (89.6306) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1990 data: 0.1070 max mem: 3540 +Test: [2060/2387] eta: 0:01:03 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1115 (0.2902) acc1: 99.4792 (89.6780) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2042 data: 0.1093 max mem: 3540 +Test: [2070/2387] eta: 0:01:01 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1095 (0.2893) acc1: 99.4792 (89.7261) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2049 data: 0.1102 max mem: 3540 +Test: [2080/2387] eta: 0:00:59 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1082 (0.2884) acc1: 99.7396 (89.7733) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2017 data: 0.1103 max mem: 3540 +Test: [2090/2387] eta: 0:00:57 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1081 (0.2876) acc1: 99.4792 (89.8201) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2027 data: 0.1121 max mem: 3540 +Test: [2100/2387] eta: 0:00:55 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1089 (0.2867) acc1: 99.7396 (89.8667) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2100 data: 0.1195 max mem: 3540 +Test: [2110/2387] eta: 0:00:53 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1089 (0.2859) acc1: 99.7396 (89.9122) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2129 data: 0.1211 max mem: 3540 +Test: [2120/2387] eta: 0:00:51 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1204 (0.2862) acc1: 98.9583 (89.8932) acc5: 100.0000 (100.0000) acc1_neg: 77.083 acc5_neg: 100.000 time: 0.1991 data: 0.1072 max mem: 3540 +Test: [2130/2387] eta: 0:00:49 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.3266 (0.2864) acc1: 88.8021 (89.8921) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1814 data: 0.0904 max mem: 3540 +Test: [2140/2387] eta: 0:00:48 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.3371 (0.2870) acc1: 89.5833 (89.8738) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1980 data: 0.1069 max mem: 3540 +Test: [2150/2387] eta: 0:00:46 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2912 (0.2872) acc1: 92.9688 (89.8731) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.2226 data: 0.1297 max mem: 3540 +Test: [2160/2387] eta: 0:00:44 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2575 (0.2871) acc1: 94.7917 (89.8922) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1901 data: 0.0960 max mem: 3540 +Test: [2170/2387] eta: 0:00:42 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2944 (0.2874) acc1: 92.7083 (89.8893) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1909 data: 0.0985 max mem: 3540 +Test: [2180/2387] eta: 0:00:40 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.3547 (0.2895) acc1: 87.5000 (89.7617) acc5: 100.0000 (100.0000) acc1_neg: 24.740 acc5_neg: 100.000 time: 0.1946 data: 0.1041 max mem: 3540 +Test: [2190/2387] eta: 0:00:38 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.3429 (0.2897) acc1: 90.3646 (89.7509) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1746 data: 0.0823 max mem: 3540 +Test: [2200/2387] eta: 0:00:36 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2156 (0.2894) acc1: 96.8750 (89.7838) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2125 data: 0.1194 max mem: 3540 +Test: [2210/2387] eta: 0:00:34 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2164 (0.2890) acc1: 97.6562 (89.8142) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1971 data: 0.1056 max mem: 3540 +Test: [2220/2387] eta: 0:00:32 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2053 (0.2887) acc1: 97.9167 (89.8412) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1772 data: 0.0850 max mem: 3540 +Test: [2230/2387] eta: 0:00:30 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1874 (0.2883) acc1: 98.1771 (89.8792) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1820 data: 0.0890 max mem: 3540 +Test: [2240/2387] eta: 0:00:28 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1986 (0.2879) acc1: 97.9167 (89.9103) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1800 data: 0.0868 max mem: 3540 +Test: [2250/2387] eta: 0:00:26 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1959 (0.2875) acc1: 97.6562 (89.9443) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1854 data: 0.0935 max mem: 3540 +Test: [2260/2387] eta: 0:00:24 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1688 (0.2870) acc1: 98.9583 (89.9822) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1898 data: 0.0995 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1902 (0.2867) acc1: 99.2188 (90.0194) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1745 data: 0.0838 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1954 (0.2867) acc1: 98.9583 (90.0303) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1848 data: 0.0908 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.1918 (0.2863) acc1: 98.6979 (90.0676) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1955 data: 0.1006 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2004 (0.2860) acc1: 98.6979 (90.1009) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1768 data: 0.0847 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2109 (0.2857) acc1: 98.4375 (90.1327) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1773 data: 0.0862 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2201 (0.2856) acc1: 97.3958 (90.1526) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1713 data: 0.0805 max mem: 3541 +Test: [2330/2387] eta: 0:00:11 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2375 (0.2855) acc1: 97.1354 (90.1776) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1567 data: 0.0657 max mem: 3541 +Test: [2340/2387] eta: 0:00:09 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2247 (0.2854) acc1: 97.6562 (90.2019) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1429 data: 0.0522 max mem: 3541 +Test: [2350/2387] eta: 0:00:07 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2186 (0.2856) acc1: 97.9167 (90.1965) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1268 data: 0.0367 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2266 (0.2856) acc1: 96.6146 (90.2092) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1176 data: 0.0278 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2266 (0.2854) acc1: 96.3542 (90.2308) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1144 data: 0.0247 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2639 (0.2855) acc1: 94.7917 (90.2359) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.1138 data: 0.0241 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 46.119 acc5_pos: 100.000 loss: 0.2628 (0.2853) acc1: 94.2708 (90.2468) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1084 data: 0.0211 max mem: 3541 +Test: Total time: 0:07:38 (0.1920 s / it) +* Overall Acc@1 90.247 Acc@5 100.000 + target == real Acc@1 29.970 Acc@5 100.000 + target == fake Acc@1 94.480 Acc@5 100.000 + loss 0.285 + PR-AUC 0.9882 ROC-AUC 0.8550 EER 22.05% (thr=0.7749) +Accuracy of the network on the 916267 test images: 90.2% +[rank0]:[W725 16:02:59.473910760 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 16:03:18.742000 77309 torch/distributed/run.py:766] +W0725 16:03:18.742000 77309 torch/distributed/run.py:766] ***************************************** +W0725 16:03:18.742000 77309 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 16:03:18.742000 77309 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 1): env:// +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 16:03:46.923122294 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 16:03:47.278024747 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 16:03:47.525112376 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 16:03:47.681434310 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 16:03:47.004283104 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 16:03:47.017239796 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 16:03:48.066525028 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank6]:[W725 16:03:48.079103945 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n16_seed2_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed2_resampled_n116-041-153_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:04:42 acc1_pos: 41.667 acc5_pos: 100.000 loss: 0.8722 (0.8722) acc1: 41.6667 (41.6667) acc5: 100.0000 (100.0000) time: 7.5090 data: 5.9499 max mem: 3531 +Test: [ 10/517] eta: 0:06:32 acc1_pos: 17.448 acc5_pos: 100.000 loss: 1.1540 (1.0830) acc1: 26.3021 (28.1723) acc5: 100.0000 (100.0000) time: 0.7732 data: 0.5412 max mem: 3531 +Test: [ 20/517] eta: 0:03:44 acc1_pos: 12.500 acc5_pos: 100.000 loss: 1.1361 (1.0946) acc1: 23.6979 (27.7406) acc5: 100.0000 (100.0000) time: 0.0998 data: 0.0003 max mem: 3531 +Test: [ 30/517] eta: 0:03:05 acc1_pos: 1.823 acc5_pos: 100.000 loss: 1.1361 (1.1411) acc1: 21.6146 (25.1428) acc5: 100.0000 (100.0000) time: 0.1662 data: 0.0692 max mem: 3531 +Test: [ 40/517] eta: 0:02:45 acc1_pos: 20.052 acc5_pos: 100.000 loss: 1.1749 (1.1525) acc1: 17.4479 (23.3168) acc5: 100.0000 (100.0000) time: 0.2349 data: 0.1414 max mem: 3531 +Test: [ 50/517] eta: 0:02:43 acc1_pos: 54.587 acc5_pos: 100.000 loss: 1.1365 (1.1100) acc1: 19.7917 (26.7259) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.2987 data: 0.2027 max mem: 3531 +Test: [ 60/517] eta: 0:02:42 acc1_pos: 54.587 acc5_pos: 100.000 loss: 0.4246 (0.9791) acc1: 84.3750 (37.6452) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.3753 data: 0.2770 max mem: 3531 +Test: [ 70/517] eta: 0:02:36 acc1_pos: 54.587 acc5_pos: 100.000 loss: 0.2825 (0.8766) acc1: 96.3542 (46.0351) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.3552 data: 0.2582 max mem: 3531 +Test: [ 80/517] eta: 0:02:27 acc1_pos: 54.587 acc5_pos: 100.000 loss: 0.2626 (0.8031) acc1: 98.4375 (52.2987) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2797 data: 0.1830 max mem: 3531 +Test: [ 90/517] eta: 0:02:20 acc1_pos: 54.587 acc5_pos: 100.000 loss: 0.2940 (0.7498) acc1: 94.5312 (56.9025) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2495 data: 0.1519 max mem: 3531 +Test: [100/517] eta: 0:02:13 acc1_pos: 54.587 acc5_pos: 100.000 loss: 0.2856 (0.6989) acc1: 96.0938 (60.9375) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2499 data: 0.1518 max mem: 3531 +Test: [110/517] eta: 0:02:07 acc1_pos: 54.587 acc5_pos: 100.000 loss: 0.1997 (0.6554) acc1: 98.6979 (64.3088) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2499 data: 0.1519 max mem: 3531 +Test: [120/517] eta: 0:02:02 acc1_pos: 54.587 acc5_pos: 100.000 loss: 0.2298 (0.6221) acc1: 97.6562 (66.9594) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2546 data: 0.1542 max mem: 3531 +Test: [130/517] eta: 0:01:59 acc1_pos: 54.587 acc5_pos: 100.000 loss: 0.2411 (0.5925) acc1: 97.3958 (69.2848) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2807 data: 0.1808 max mem: 3531 +Test: [140/517] eta: 0:01:54 acc1_pos: 54.587 acc5_pos: 100.000 loss: 0.2083 (0.5642) acc1: 99.7396 (71.4613) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2707 data: 0.1752 max mem: 3531 +Test: [150/517] eta: 0:01:49 acc1_pos: 54.587 acc5_pos: 100.000 loss: 0.1978 (0.5412) acc1: 100.0000 (73.2857) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2293 data: 0.1345 max mem: 3531 +Test: [160/517] eta: 0:01:45 acc1_pos: 54.587 acc5_pos: 100.000 loss: 0.2316 (0.5224) acc1: 98.6979 (74.8269) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2301 data: 0.1336 max mem: 3531 +Test: [170/517] eta: 0:01:40 acc1_pos: 54.587 acc5_pos: 100.000 loss: 0.2400 (0.5051) acc1: 98.6979 (76.2183) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2315 data: 0.1352 max mem: 3531 +Test: [180/517] eta: 0:01:36 acc1_pos: 54.587 acc5_pos: 100.000 loss: 0.2251 (0.4892) acc1: 99.2188 (77.4143) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2299 data: 0.1352 max mem: 3531 +Test: [190/517] eta: 0:01:32 acc1_pos: 54.587 acc5_pos: 100.000 loss: 0.1725 (0.4726) acc1: 99.7396 (78.5954) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2296 data: 0.1349 max mem: 3531 +Test: [200/517] eta: 0:01:29 acc1_pos: 54.587 acc5_pos: 100.000 loss: 0.1760 (0.4585) acc1: 100.0000 (79.6409) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2391 data: 0.1441 max mem: 3531 +Test: [210/517] eta: 0:01:25 acc1_pos: 54.587 acc5_pos: 100.000 loss: 0.2178 (0.4547) acc1: 91.4062 (79.9578) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.2325 data: 0.1377 max mem: 3532 +Test: [220/517] eta: 0:01:22 acc1_pos: 54.587 acc5_pos: 100.000 loss: 0.3890 (0.4521) acc1: 84.3750 (80.1977) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.2286 data: 0.1333 max mem: 3532 +Test: [230/517] eta: 0:01:18 acc1_pos: 54.587 acc5_pos: 100.000 loss: 0.3156 (0.4443) acc1: 90.6250 (80.9355) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2299 data: 0.1343 max mem: 3532 +Test: [240/517] eta: 0:01:15 acc1_pos: 54.587 acc5_pos: 100.000 loss: 0.2439 (0.4354) acc1: 98.1771 (81.6790) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2299 data: 0.1354 max mem: 3532 +Test: [250/517] eta: 0:01:12 acc1_pos: 54.587 acc5_pos: 100.000 loss: 0.2293 (0.4278) acc1: 98.9583 (82.3529) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2360 data: 0.1389 max mem: 3532 +Test: [260/517] eta: 0:01:09 acc1_pos: 54.587 acc5_pos: 100.000 loss: 0.2407 (0.4208) acc1: 98.9583 (82.9771) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2250 data: 0.1293 max mem: 3532 +Test: [270/517] eta: 0:01:06 acc1_pos: 54.587 acc5_pos: 100.000 loss: 0.2556 (0.4151) acc1: 98.9583 (83.5390) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2303 data: 0.1368 max mem: 3532 +Test: [280/517] eta: 0:01:03 acc1_pos: 54.587 acc5_pos: 100.000 loss: 0.2588 (0.4085) acc1: 98.9583 (84.0859) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2548 data: 0.1618 max mem: 3532 +Test: [290/517] eta: 0:01:00 acc1_pos: 54.587 acc5_pos: 100.000 loss: 0.2257 (0.4023) acc1: 99.2188 (84.6005) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2635 data: 0.1732 max mem: 3532 +Test: [300/517] eta: 0:00:58 acc1_pos: 54.587 acc5_pos: 100.000 loss: 0.2078 (0.3955) acc1: 99.4792 (85.0974) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2726 data: 0.1823 max mem: 3532 +Test: [310/517] eta: 0:00:55 acc1_pos: 54.587 acc5_pos: 100.000 loss: 0.1844 (0.3886) acc1: 99.7396 (85.5724) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2650 data: 0.1731 max mem: 3532 +Test: [320/517] eta: 0:00:52 acc1_pos: 54.587 acc5_pos: 100.000 loss: 0.1844 (0.3831) acc1: 99.7396 (85.9975) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2572 data: 0.1639 max mem: 3532 +Test: [330/517] eta: 0:00:50 acc1_pos: 54.587 acc5_pos: 100.000 loss: 0.2086 (0.3779) acc1: 99.2188 (86.3915) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2653 data: 0.1725 max mem: 3532 +Test: [340/517] eta: 0:00:47 acc1_pos: 54.587 acc5_pos: 100.000 loss: 0.1926 (0.3724) acc1: 99.4792 (86.7760) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2501 data: 0.1579 max mem: 3532 +Test: [350/517] eta: 0:00:44 acc1_pos: 54.587 acc5_pos: 100.000 loss: 0.2120 (0.3691) acc1: 99.2188 (87.0756) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2544 data: 0.1607 max mem: 3532 +Test: [360/517] eta: 0:00:41 acc1_pos: 54.587 acc5_pos: 100.000 loss: 0.2368 (0.3655) acc1: 98.4375 (87.3774) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2599 data: 0.1651 max mem: 3532 +Test: [370/517] eta: 0:00:39 acc1_pos: 54.587 acc5_pos: 100.000 loss: 0.1884 (0.3600) acc1: 99.7396 (87.7169) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1589 max mem: 3532 +Test: [380/517] eta: 0:00:36 acc1_pos: 54.587 acc5_pos: 100.000 loss: 0.1744 (0.3559) acc1: 100.0000 (88.0270) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2647 data: 0.1681 max mem: 3532 +Test: [390/517] eta: 0:00:33 acc1_pos: 54.587 acc5_pos: 100.000 loss: 0.1916 (0.3515) acc1: 99.7396 (88.3259) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2749 data: 0.1800 max mem: 3532 +Test: [400/517] eta: 0:00:31 acc1_pos: 54.587 acc5_pos: 100.000 loss: 0.1794 (0.3469) acc1: 99.7396 (88.6125) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2603 data: 0.1644 max mem: 3532 +Test: [410/517] eta: 0:00:28 acc1_pos: 54.587 acc5_pos: 100.000 loss: 0.1794 (0.3435) acc1: 99.7396 (88.8781) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2549 data: 0.1571 max mem: 3532 +Test: [420/517] eta: 0:00:25 acc1_pos: 54.587 acc5_pos: 100.000 loss: 0.2241 (0.3417) acc1: 99.4792 (89.0767) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2696 data: 0.1728 max mem: 3532 +Test: [430/517] eta: 0:00:23 acc1_pos: 54.587 acc5_pos: 100.000 loss: 0.2542 (0.3407) acc1: 97.6562 (89.2190) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2649 data: 0.1678 max mem: 3532 +Test: [440/517] eta: 0:00:20 acc1_pos: 54.587 acc5_pos: 100.000 loss: 0.2482 (0.3380) acc1: 97.6562 (89.4227) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2599 data: 0.1627 max mem: 3533 +Test: [450/517] eta: 0:00:17 acc1_pos: 54.587 acc5_pos: 100.000 loss: 0.2179 (0.3353) acc1: 98.9583 (89.6388) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2649 data: 0.1681 max mem: 3533 +Test: [460/517] eta: 0:00:15 acc1_pos: 54.587 acc5_pos: 100.000 loss: 0.2202 (0.3329) acc1: 99.2188 (89.8172) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.2704 data: 0.1740 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 54.587 acc5_pos: 100.000 loss: 0.2170 (0.3303) acc1: 99.2188 (90.0179) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2549 data: 0.1615 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 54.587 acc5_pos: 100.000 loss: 0.2154 (0.3277) acc1: 99.4792 (90.2000) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2505 data: 0.1573 max mem: 3533 +Test: [490/517] eta: 0:00:07 acc1_pos: 54.587 acc5_pos: 100.000 loss: 0.2164 (0.3254) acc1: 99.2188 (90.3879) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2255 data: 0.1334 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 54.587 acc5_pos: 100.000 loss: 0.2276 (0.3234) acc1: 99.2188 (90.5543) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1919 data: 0.1017 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 54.587 acc5_pos: 100.000 loss: 0.2388 (0.3221) acc1: 98.1771 (90.6765) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1818 data: 0.0923 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 54.587 acc5_pos: 100.000 loss: 0.2355 (0.3208) acc1: 98.4375 (90.7679) acc5: 100.0000 (100.0000) acc1_neg: 99.293 acc5_neg: 100.000 time: 0.1783 data: 0.0847 max mem: 3533 +Test: Total time: 0:02:15 (0.2614 s / it) +* Overall Acc@1 90.768 Acc@5 100.000 + target == real Acc@1 23.458 Acc@5 100.000 + target == fake Acc@1 97.751 Acc@5 100.000 + loss 0.321 + PR-AUC 0.9664 ROC-AUC 0.7880 EER 27.99% (thr=0.7783) +Accuracy of the network on the 198427 test images: 90.8% +[rank0]:[W725 16:06:13.334858183 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0725 16:06:31.978000 77879 torch/distributed/run.py:766] +W0725 16:06:31.978000 77879 torch/distributed/run.py:766] ***************************************** +W0725 16:06:31.978000 77879 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 16:06:31.978000 77879 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 16:06:58.758876734 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 16:06:59.330545385 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 16:06:59.583197927 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 16:06:59.695414161 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 16:06:59.841179093 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 16:06:59.848069424 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +[rank3]:[W725 16:06:59.867970497 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 16:06:59.884598936 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n16_seed2_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed2_resampled_n116-041-153_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:32:55 acc1_pos: 48.698 acc5_pos: 100.000 loss: 0.7627 (0.7627) acc1: 48.6979 (48.6979) acc5: 100.0000 (100.0000) time: 5.0796 data: 3.2303 max mem: 3531 +Test: [ 10/389] eta: 0:03:27 acc1_pos: 40.885 acc5_pos: 100.000 loss: 0.7627 (0.7593) acc1: 48.6979 (50.6392) acc5: 100.0000 (100.0000) time: 0.5486 data: 0.2989 max mem: 3531 +Test: [ 20/389] eta: 0:02:08 acc1_pos: 4.167 acc5_pos: 100.000 loss: 0.9828 (0.9571) acc1: 21.6146 (32.1181) acc5: 100.0000 (100.0000) time: 0.1111 data: 0.0159 max mem: 3531 +Test: [ 30/389] eta: 0:01:44 acc1_pos: 14.844 acc5_pos: 100.000 loss: 1.2650 (1.0470) acc1: 9.3750 (25.4116) acc5: 100.0000 (100.0000) time: 0.1482 data: 0.0460 max mem: 3531 +Test: [ 40/389] eta: 0:01:37 acc1_pos: 32.812 acc5_pos: 100.000 loss: 1.1227 (1.0399) acc1: 13.5417 (24.9301) acc5: 100.0000 (100.0000) time: 0.2049 data: 0.1053 max mem: 3531 +Test: [ 50/389] eta: 0:01:30 acc1_pos: 15.104 acc5_pos: 100.000 loss: 1.0063 (0.9974) acc1: 15.3646 (28.6816) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1362 max mem: 3531 +Test: [ 60/389] eta: 0:01:18 acc1_pos: 17.969 acc5_pos: 100.000 loss: 0.8737 (0.9796) acc1: 17.9688 (29.6662) acc5: 100.0000 (100.0000) time: 0.1599 data: 0.0642 max mem: 3531 +Test: [ 70/389] eta: 0:01:11 acc1_pos: 46.615 acc5_pos: 100.000 loss: 0.7285 (0.9347) acc1: 48.9583 (33.7148) acc5: 100.0000 (100.0000) time: 0.1107 data: 0.0122 max mem: 3531 +Test: [ 80/389] eta: 0:01:06 acc1_pos: 84.896 acc5_pos: 100.000 loss: 0.7285 (0.9302) acc1: 48.9583 (34.3396) acc5: 100.0000 (100.0000) time: 0.1472 data: 0.0502 max mem: 3531 +Test: [ 90/389] eta: 0:01:03 acc1_pos: 68.750 acc5_pos: 100.000 loss: 0.5637 (0.8874) acc1: 63.5417 (38.1239) acc5: 100.0000 (100.0000) time: 0.1721 data: 0.0785 max mem: 3531 +Test: [100/389] eta: 0:01:00 acc1_pos: 8.073 acc5_pos: 100.000 loss: 0.7286 (0.9035) acc1: 55.4688 (37.3608) acc5: 100.0000 (100.0000) time: 0.1787 data: 0.0874 max mem: 3531 +Test: [110/389] eta: 0:00:57 acc1_pos: 97.135 acc5_pos: 100.000 loss: 1.1213 (0.9135) acc1: 18.2292 (37.4179) acc5: 100.0000 (100.0000) time: 0.1795 data: 0.0883 max mem: 3531 +Test: [120/389] eta: 0:00:54 acc1_pos: 44.792 acc5_pos: 100.000 loss: 0.8045 (0.8999) acc1: 44.7917 (38.5890) acc5: 100.0000 (100.0000) time: 0.1649 data: 0.0742 max mem: 3531 +Test: [130/389] eta: 0:00:51 acc1_pos: 58.282 acc5_pos: 100.000 loss: 0.6064 (0.8705) acc1: 63.8021 (41.2591) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1596 data: 0.0688 max mem: 3531 +Test: [140/389] eta: 0:00:49 acc1_pos: 58.282 acc5_pos: 100.000 loss: 0.3428 (0.8354) acc1: 90.3646 (44.7012) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.1754 data: 0.0834 max mem: 3531 +Test: [150/389] eta: 0:00:46 acc1_pos: 58.282 acc5_pos: 100.000 loss: 0.3135 (0.8082) acc1: 90.6250 (47.1475) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1758 data: 0.0842 max mem: 3531 +Test: [160/389] eta: 0:00:44 acc1_pos: 58.282 acc5_pos: 100.000 loss: 0.2730 (0.7749) acc1: 92.7083 (49.9935) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1587 data: 0.0679 max mem: 3531 +Test: [170/389] eta: 0:00:41 acc1_pos: 58.282 acc5_pos: 100.000 loss: 0.1389 (0.7483) acc1: 99.7396 (51.9767) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.1567 data: 0.0649 max mem: 3531 +Test: [180/389] eta: 0:00:39 acc1_pos: 58.282 acc5_pos: 100.000 loss: 0.2241 (0.7217) acc1: 99.7396 (54.4314) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1573 data: 0.0649 max mem: 3531 +Test: [190/389] eta: 0:00:37 acc1_pos: 58.282 acc5_pos: 100.000 loss: 0.3125 (0.7052) acc1: 92.7083 (56.0046) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1482 data: 0.0567 max mem: 3531 +Test: [200/389] eta: 0:00:34 acc1_pos: 58.282 acc5_pos: 100.000 loss: 0.3926 (0.6902) acc1: 91.9271 (57.6752) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1425 data: 0.0520 max mem: 3531 +Test: [210/389] eta: 0:00:32 acc1_pos: 58.282 acc5_pos: 100.000 loss: 0.3219 (0.6694) acc1: 97.6562 (59.4589) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1435 data: 0.0530 max mem: 3532 +Test: [220/389] eta: 0:00:30 acc1_pos: 58.282 acc5_pos: 100.000 loss: 0.1703 (0.6487) acc1: 100.0000 (61.1319) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1602 data: 0.0690 max mem: 3532 +Test: [230/389] eta: 0:00:28 acc1_pos: 58.282 acc5_pos: 100.000 loss: 0.2598 (0.6393) acc1: 90.1042 (62.0096) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1642 data: 0.0728 max mem: 3532 +Test: [240/389] eta: 0:00:26 acc1_pos: 58.282 acc5_pos: 100.000 loss: 0.2891 (0.6246) acc1: 90.3646 (63.3526) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1598 data: 0.0681 max mem: 3532 +Test: [250/389] eta: 0:00:25 acc1_pos: 58.282 acc5_pos: 100.000 loss: 0.2151 (0.6082) acc1: 100.0000 (64.8126) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1759 data: 0.0843 max mem: 3532 +Test: [260/389] eta: 0:00:23 acc1_pos: 58.282 acc5_pos: 100.000 loss: 0.2137 (0.5997) acc1: 100.0000 (65.5133) acc5: 100.0000 (100.0000) acc1_neg: 38.021 acc5_neg: 100.000 time: 0.1749 data: 0.0846 max mem: 3532 +Test: [270/389] eta: 0:00:21 acc1_pos: 58.282 acc5_pos: 100.000 loss: 0.3722 (0.5979) acc1: 88.2812 (65.8239) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1748 data: 0.0845 max mem: 3532 +Test: [280/389] eta: 0:00:19 acc1_pos: 58.282 acc5_pos: 100.000 loss: 0.3724 (0.5948) acc1: 85.6771 (66.2339) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1667 data: 0.0762 max mem: 3532 +Test: [290/389] eta: 0:00:17 acc1_pos: 58.282 acc5_pos: 100.000 loss: 0.3533 (0.5837) acc1: 97.3958 (67.1159) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1582 data: 0.0673 max mem: 3532 +Test: [300/389] eta: 0:00:15 acc1_pos: 58.282 acc5_pos: 100.000 loss: 0.2697 (0.5743) acc1: 97.3958 (68.0605) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1734 data: 0.0823 max mem: 3532 +Test: [310/389] eta: 0:00:14 acc1_pos: 58.282 acc5_pos: 100.000 loss: 0.3496 (0.5741) acc1: 85.9375 (68.1253) acc5: 100.0000 (100.0000) acc1_neg: 16.667 acc5_neg: 100.000 time: 0.1823 data: 0.0902 max mem: 3532 +Test: [320/389] eta: 0:00:12 acc1_pos: 58.282 acc5_pos: 100.000 loss: 0.4745 (0.5680) acc1: 79.9479 (68.7330) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1909 data: 0.0987 max mem: 3532 +Test: [330/389] eta: 0:00:10 acc1_pos: 58.282 acc5_pos: 100.000 loss: 0.4757 (0.5655) acc1: 84.1146 (69.1473) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.1969 data: 0.1049 max mem: 3532 +Test: [340/389] eta: 0:00:08 acc1_pos: 58.282 acc5_pos: 100.000 loss: 0.2600 (0.5552) acc1: 85.9375 (69.9612) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1874 data: 0.0929 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 58.282 acc5_pos: 100.000 loss: 0.1771 (0.5599) acc1: 99.4792 (69.6626) acc5: 100.0000 (100.0000) acc1_neg: 42.969 acc5_neg: 100.000 time: 0.1907 data: 0.0972 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 58.282 acc5_pos: 100.000 loss: 0.3080 (0.5565) acc1: 94.2708 (70.0341) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2022 data: 0.1105 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 58.282 acc5_pos: 100.000 loss: 0.3022 (0.5514) acc1: 95.5729 (70.3616) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2088 data: 0.1152 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 58.282 acc5_pos: 100.000 loss: 0.2323 (0.5474) acc1: 95.8333 (70.7534) acc5: 100.0000 (100.0000) acc1_neg: 64.844 acc5_neg: 100.000 time: 0.1878 data: 0.0934 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 58.282 acc5_pos: 100.000 loss: 0.3074 (0.5424) acc1: 94.2708 (71.1947) acc5: 100.0000 (100.0000) acc1_neg: 96.995 acc5_neg: 100.000 time: 0.1463 data: 0.0478 max mem: 3532 +Test: Total time: 0:01:10 (0.1814 s / it) +* Overall Acc@1 71.195 Acc@5 100.000 + target == real Acc@1 38.367 Acc@5 100.000 + target == fake Acc@1 86.454 Acc@5 100.000 + loss 0.542 + PR-AUC 0.8742 ROC-AUC 0.7563 EER 31.83% (thr=0.6914) +Accuracy of the network on the 149358 test images: 71.2% +[rank0]:[W725 16:08:24.212394104 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 16:08:42.599000 78449 torch/distributed/run.py:766] +W0725 16:08:42.599000 78449 torch/distributed/run.py:766] ***************************************** +W0725 16:08:42.599000 78449 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 16:08:42.599000 78449 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 16:09:09.218096859 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 16:09:09.254820026 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 16:09:09.905576745 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 16:09:09.979206384 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 16:09:10.084433773 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 16:09:10.119027557 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 16:09:10.140672524 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 16:09:10.229018034 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n16_seed2_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed2_resampled_n116-041-153_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:34:46 acc1_pos: 94.271 acc5_pos: 100.000 loss: 0.2584 (0.2584) acc1: 94.2708 (94.2708) acc5: 100.0000 (100.0000) time: 6.2664 data: 4.6588 max mem: 3531 +Test: [ 10/333] eta: 0:03:30 acc1_pos: 90.625 acc5_pos: 100.000 loss: 0.2375 (0.2236) acc1: 92.9688 (93.3002) acc5: 100.0000 (100.0000) time: 0.6519 data: 0.4239 max mem: 3531 +Test: [ 20/333] eta: 0:02:01 acc1_pos: 98.698 acc5_pos: 100.000 loss: 0.2005 (0.2138) acc1: 94.7917 (94.3948) acc5: 100.0000 (100.0000) time: 0.0946 data: 0.0004 max mem: 3531 +Test: [ 30/333] eta: 0:01:41 acc1_pos: 63.281 acc5_pos: 100.000 loss: 0.2868 (0.3662) acc1: 86.4583 (83.4929) acc5: 100.0000 (100.0000) time: 0.1593 data: 0.0615 max mem: 3531 +Test: [ 40/333] eta: 0:01:29 acc1_pos: 64.583 acc5_pos: 100.000 loss: 0.6848 (0.4549) acc1: 57.2917 (77.0008) acc5: 100.0000 (100.0000) time: 0.2149 data: 0.1203 max mem: 3531 +Test: [ 50/333] eta: 0:01:20 acc1_pos: 68.229 acc5_pos: 100.000 loss: 0.6797 (0.5072) acc1: 57.2917 (72.9320) acc5: 100.0000 (100.0000) time: 0.2099 data: 0.1166 max mem: 3531 +Test: [ 60/333] eta: 0:01:15 acc1_pos: 53.125 acc5_pos: 100.000 loss: 0.6881 (0.5405) acc1: 56.2500 (70.2357) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1318 max mem: 3531 +Test: [ 70/333] eta: 0:01:08 acc1_pos: 56.250 acc5_pos: 100.000 loss: 0.7009 (0.5639) acc1: 56.2500 (68.3869) acc5: 100.0000 (100.0000) time: 0.1949 data: 0.1014 max mem: 3531 +Test: [ 80/333] eta: 0:01:02 acc1_pos: 63.021 acc5_pos: 100.000 loss: 0.6977 (0.5808) acc1: 56.5104 (67.0878) acc5: 100.0000 (100.0000) time: 0.1549 data: 0.0557 max mem: 3531 +Test: [ 90/333] eta: 0:00:58 acc1_pos: 66.146 acc5_pos: 100.000 loss: 0.6650 (0.5861) acc1: 58.3333 (66.8298) acc5: 100.0000 (100.0000) time: 0.1749 data: 0.0750 max mem: 3531 +Test: [100/333] eta: 0:00:55 acc1_pos: 68.490 acc5_pos: 100.000 loss: 0.6212 (0.5917) acc1: 63.5417 (66.4295) acc5: 100.0000 (100.0000) time: 0.1899 data: 0.0932 max mem: 3531 +Test: [110/333] eta: 0:00:51 acc1_pos: 69.010 acc5_pos: 100.000 loss: 0.6212 (0.5964) acc1: 62.5000 (66.0426) acc5: 100.0000 (100.0000) time: 0.1799 data: 0.0820 max mem: 3531 +Test: [120/333] eta: 0:00:47 acc1_pos: 69.792 acc5_pos: 100.000 loss: 0.6101 (0.5993) acc1: 64.0625 (65.9112) acc5: 100.0000 (100.0000) time: 0.1450 data: 0.0470 max mem: 3531 +Test: [130/333] eta: 0:00:44 acc1_pos: 73.177 acc5_pos: 100.000 loss: 0.5775 (0.6005) acc1: 67.1875 (65.9570) acc5: 100.0000 (100.0000) time: 0.1475 data: 0.0500 max mem: 3531 +Test: [140/333] eta: 0:00:41 acc1_pos: 81.510 acc5_pos: 100.000 loss: 0.5838 (0.6028) acc1: 63.5417 (65.6786) acc5: 100.0000 (100.0000) time: 0.1829 data: 0.0869 max mem: 3531 +Test: [150/333] eta: 0:00:39 acc1_pos: 70.052 acc5_pos: 100.000 loss: 0.6150 (0.6060) acc1: 63.2812 (65.4491) acc5: 100.0000 (100.0000) time: 0.1845 data: 0.0906 max mem: 3531 +Test: [160/333] eta: 0:00:36 acc1_pos: 56.908 acc5_pos: 100.000 loss: 0.6597 (0.6190) acc1: 61.9792 (64.6254) acc5: 100.0000 (100.0000) acc1_neg: 38.542 acc5_neg: 100.000 time: 0.1569 data: 0.0639 max mem: 3531 +Test: [170/333] eta: 0:00:33 acc1_pos: 56.908 acc5_pos: 100.000 loss: 0.8727 (0.6366) acc1: 47.6562 (63.5219) acc5: 100.0000 (100.0000) acc1_neg: 46.875 acc5_neg: 100.000 time: 0.1357 data: 0.0415 max mem: 3531 +Test: [180/333] eta: 0:00:30 acc1_pos: 56.908 acc5_pos: 100.000 loss: 0.8863 (0.6511) acc1: 44.5312 (62.4727) acc5: 100.0000 (100.0000) acc1_neg: 40.885 acc5_neg: 100.000 time: 0.1538 data: 0.0590 max mem: 3531 +Test: [190/333] eta: 0:00:28 acc1_pos: 56.908 acc5_pos: 100.000 loss: 0.8851 (0.6597) acc1: 46.0938 (61.9669) acc5: 100.0000 (100.0000) acc1_neg: 44.531 acc5_neg: 100.000 time: 0.1770 data: 0.0812 max mem: 3531 +Test: [200/333] eta: 0:00:26 acc1_pos: 56.908 acc5_pos: 100.000 loss: 0.7790 (0.6640) acc1: 55.9896 (61.8470) acc5: 100.0000 (100.0000) acc1_neg: 59.375 acc5_neg: 100.000 time: 0.1710 data: 0.0704 max mem: 3531 +Test: [210/333] eta: 0:00:24 acc1_pos: 56.908 acc5_pos: 100.000 loss: 0.8004 (0.6729) acc1: 55.9896 (61.3497) acc5: 100.0000 (100.0000) acc1_neg: 40.104 acc5_neg: 100.000 time: 0.1799 data: 0.0794 max mem: 3532 +Test: [220/333] eta: 0:00:22 acc1_pos: 56.908 acc5_pos: 100.000 loss: 0.8515 (0.6894) acc1: 45.0521 (60.3707) acc5: 100.0000 (100.0000) acc1_neg: 64.062 acc5_neg: 100.000 time: 0.1900 data: 0.0944 max mem: 3532 +Test: [230/333] eta: 0:00:20 acc1_pos: 56.908 acc5_pos: 100.000 loss: 0.6057 (0.6837) acc1: 61.1979 (60.8744) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.1749 data: 0.0802 max mem: 3532 +Test: [240/333] eta: 0:00:18 acc1_pos: 56.908 acc5_pos: 100.000 loss: 0.6048 (0.6830) acc1: 64.5833 (60.9548) acc5: 100.0000 (100.0000) acc1_neg: 61.979 acc5_neg: 100.000 time: 0.1699 data: 0.0734 max mem: 3532 +Test: [250/333] eta: 0:00:16 acc1_pos: 56.908 acc5_pos: 100.000 loss: 0.6747 (0.6866) acc1: 60.4167 (60.7912) acc5: 100.0000 (100.0000) acc1_neg: 52.604 acc5_neg: 100.000 time: 0.1749 data: 0.0800 max mem: 3532 +Test: [260/333] eta: 0:00:14 acc1_pos: 56.908 acc5_pos: 100.000 loss: 0.8646 (0.6955) acc1: 51.5625 (60.2900) acc5: 100.0000 (100.0000) acc1_neg: 55.469 acc5_neg: 100.000 time: 0.1685 data: 0.0734 max mem: 3532 +Test: [270/333] eta: 0:00:12 acc1_pos: 56.908 acc5_pos: 100.000 loss: 0.7940 (0.6975) acc1: 54.9479 (60.2600) acc5: 100.0000 (100.0000) acc1_neg: 62.760 acc5_neg: 100.000 time: 0.1899 data: 0.0910 max mem: 3532 +Test: [280/333] eta: 0:00:10 acc1_pos: 56.908 acc5_pos: 100.000 loss: 0.7628 (0.7053) acc1: 57.2917 (59.7401) acc5: 100.0000 (100.0000) acc1_neg: 45.052 acc5_neg: 100.000 time: 0.2113 data: 0.1124 max mem: 3532 +Test: [290/333] eta: 0:00:08 acc1_pos: 56.908 acc5_pos: 100.000 loss: 1.0319 (0.7177) acc1: 36.9792 (58.8855) acc5: 100.0000 (100.0000) acc1_neg: 36.979 acc5_neg: 100.000 time: 0.1699 data: 0.0732 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 56.908 acc5_pos: 100.000 loss: 1.0696 (0.7301) acc1: 34.8958 (58.0807) acc5: 100.0000 (100.0000) acc1_neg: 36.979 acc5_neg: 100.000 time: 0.1549 data: 0.0587 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 56.908 acc5_pos: 100.000 loss: 0.9732 (0.7305) acc1: 37.7604 (58.0738) acc5: 100.0000 (100.0000) acc1_neg: 52.865 acc5_neg: 100.000 time: 0.1708 data: 0.0771 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 56.908 acc5_pos: 100.000 loss: 0.7272 (0.7318) acc1: 57.0312 (57.9545) acc5: 100.0000 (100.0000) acc1_neg: 41.667 acc5_neg: 100.000 time: 0.1836 data: 0.0927 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 56.908 acc5_pos: 100.000 loss: 0.9576 (0.7412) acc1: 41.6667 (57.3892) acc5: 100.0000 (100.0000) acc1_neg: 40.104 acc5_neg: 100.000 time: 0.1771 data: 0.0866 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 56.908 acc5_pos: 100.000 loss: 1.0082 (0.7432) acc1: 40.3646 (57.2789) acc5: 100.0000 (100.0000) acc1_neg: 42.254 acc5_neg: 100.000 time: 0.1803 data: 0.0866 max mem: 3532 +Test: Total time: 0:01:03 (0.1911 s / it) +* Overall Acc@1 57.279 Acc@5 100.000 + target == real Acc@1 65.362 Acc@5 100.000 + target == fake Acc@1 50.084 Acc@5 100.000 + loss 0.743 + PR-AUC 0.6314 ROC-AUC 0.6155 EER 42.09% (thr=0.4363) +Accuracy of the network on the 127843 test images: 57.3% +[rank0]:[W725 16:10:25.932190667 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 16:10:44.154000 79006 torch/distributed/run.py:766] +W0725 16:10:44.154000 79006 torch/distributed/run.py:766] ***************************************** +W0725 16:10:44.154000 79006 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 16:10:44.154000 79006 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 16:11:10.515697226 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 16:11:11.072346144 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 16:11:11.217364056 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 16:11:11.321173065 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 16:11:11.635269752 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +[rank7]:[W725 16:11:11.654061372 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 16:11:11.658611922 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W725 16:11:11.662187419 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n16_seed2_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed2_resampled_n116-041-153_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:21:50 acc1_pos: 91.667 acc5_pos: 100.000 loss: 0.2385 (0.2385) acc1: 91.6667 (91.6667) acc5: 100.0000 (100.0000) time: 5.5785 data: 4.0429 max mem: 3531 +Test: [ 10/235] eta: 0:02:19 acc1_pos: 92.448 acc5_pos: 100.000 loss: 0.2432 (0.2417) acc1: 92.4479 (92.0691) acc5: 100.0000 (100.0000) time: 0.6182 data: 0.3961 max mem: 3531 +Test: [ 20/235] eta: 0:01:23 acc1_pos: 91.146 acc5_pos: 100.000 loss: 0.2388 (0.2374) acc1: 92.4479 (92.3735) acc5: 100.0000 (100.0000) time: 0.1310 data: 0.0383 max mem: 3531 +Test: [ 30/235] eta: 0:01:08 acc1_pos: 95.052 acc5_pos: 100.000 loss: 0.2292 (0.2352) acc1: 92.7083 (92.5571) acc5: 100.0000 (100.0000) time: 0.1749 data: 0.0820 max mem: 3531 +Test: [ 40/235] eta: 0:01:01 acc1_pos: 92.448 acc5_pos: 100.000 loss: 0.2297 (0.2353) acc1: 92.4479 (92.5305) acc5: 100.0000 (100.0000) time: 0.2349 data: 0.1417 max mem: 3531 +Test: [ 50/235] eta: 0:00:54 acc1_pos: 91.146 acc5_pos: 100.000 loss: 0.2314 (0.2352) acc1: 92.4479 (92.4224) acc5: 100.0000 (100.0000) time: 0.2409 data: 0.1475 max mem: 3531 +Test: [ 60/235] eta: 0:00:49 acc1_pos: 94.271 acc5_pos: 100.000 loss: 0.2314 (0.2357) acc1: 92.4479 (92.4863) acc5: 100.0000 (100.0000) time: 0.2199 data: 0.1277 max mem: 3531 +Test: [ 70/235] eta: 0:00:44 acc1_pos: 90.769 acc5_pos: 100.000 loss: 0.2577 (0.4339) acc1: 90.1042 (82.4824) acc5: 100.0000 (100.0000) acc1_neg: 13.281 acc5_neg: 100.000 time: 0.2090 data: 0.1160 max mem: 3531 +Test: [ 80/235] eta: 0:00:39 acc1_pos: 90.769 acc5_pos: 100.000 loss: 1.8115 (0.6072) acc1: 12.2396 (73.7429) acc5: 100.0000 (100.0000) acc1_neg: 12.240 acc5_neg: 100.000 time: 0.1750 data: 0.0806 max mem: 3531 +Test: [ 90/235] eta: 0:00:36 acc1_pos: 90.769 acc5_pos: 100.000 loss: 1.8515 (0.7436) acc1: 11.4583 (66.9385) acc5: 100.0000 (100.0000) acc1_neg: 11.458 acc5_neg: 100.000 time: 0.1798 data: 0.0836 max mem: 3531 +Test: [100/235] eta: 0:00:33 acc1_pos: 90.769 acc5_pos: 100.000 loss: 1.8554 (0.8525) acc1: 11.9792 (61.5537) acc5: 100.0000 (100.0000) acc1_neg: 10.417 acc5_neg: 100.000 time: 0.2052 data: 0.1078 max mem: 3531 +Test: [110/235] eta: 0:00:30 acc1_pos: 90.769 acc5_pos: 100.000 loss: 1.8464 (0.9405) acc1: 11.7188 (57.0969) acc5: 100.0000 (100.0000) acc1_neg: 13.542 acc5_neg: 100.000 time: 0.2249 data: 0.1273 max mem: 3531 +Test: [120/235] eta: 0:00:27 acc1_pos: 90.769 acc5_pos: 100.000 loss: 1.8347 (1.0155) acc1: 10.6771 (53.1272) acc5: 100.0000 (100.0000) acc1_neg: 9.115 acc5_neg: 100.000 time: 0.2244 data: 0.1293 max mem: 3531 +Test: [130/235] eta: 0:00:25 acc1_pos: 90.769 acc5_pos: 100.000 loss: 1.8347 (1.0777) acc1: 8.5938 (49.7396) acc5: 100.0000 (100.0000) acc1_neg: 7.812 acc5_neg: 100.000 time: 0.1999 data: 0.1067 max mem: 3531 +Test: [140/235] eta: 0:00:22 acc1_pos: 90.769 acc5_pos: 100.000 loss: 1.8463 (1.1332) acc1: 8.3333 (46.8048) acc5: 100.0000 (100.0000) acc1_neg: 6.510 acc5_neg: 100.000 time: 0.1949 data: 0.1020 max mem: 3531 +Test: [150/235] eta: 0:00:19 acc1_pos: 90.769 acc5_pos: 100.000 loss: 1.8456 (1.1793) acc1: 8.3333 (44.2674) acc5: 100.0000 (100.0000) acc1_neg: 10.156 acc5_neg: 100.000 time: 0.2068 data: 0.1114 max mem: 3531 +Test: [160/235] eta: 0:00:17 acc1_pos: 90.769 acc5_pos: 100.000 loss: 1.8172 (1.2197) acc1: 8.3333 (42.0824) acc5: 100.0000 (100.0000) acc1_neg: 11.198 acc5_neg: 100.000 time: 0.2299 data: 0.1331 max mem: 3531 +Test: [170/235] eta: 0:00:15 acc1_pos: 90.769 acc5_pos: 100.000 loss: 1.7517 (1.2414) acc1: 11.1979 (40.7149) acc5: 100.0000 (100.0000) acc1_neg: 16.927 acc5_neg: 100.000 time: 0.2231 data: 0.1257 max mem: 3531 +Test: [180/235] eta: 0:00:12 acc1_pos: 90.769 acc5_pos: 100.000 loss: 1.5832 (1.2594) acc1: 19.0104 (39.5632) acc5: 100.0000 (100.0000) acc1_neg: 19.010 acc5_neg: 100.000 time: 0.2072 data: 0.1114 max mem: 3531 +Test: [190/235] eta: 0:00:10 acc1_pos: 90.769 acc5_pos: 100.000 loss: 1.5521 (1.2757) acc1: 19.5312 (38.5280) acc5: 100.0000 (100.0000) acc1_neg: 19.010 acc5_neg: 100.000 time: 0.2069 data: 0.1145 max mem: 3531 +Test: [200/235] eta: 0:00:08 acc1_pos: 90.769 acc5_pos: 100.000 loss: 1.5588 (1.2911) acc1: 19.0104 (37.5376) acc5: 100.0000 (100.0000) acc1_neg: 16.406 acc5_neg: 100.000 time: 0.1970 data: 0.1052 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 90.769 acc5_pos: 100.000 loss: 1.5761 (1.3043) acc1: 19.0104 (36.6657) acc5: 100.0000 (100.0000) acc1_neg: 18.229 acc5_neg: 100.000 time: 0.1643 data: 0.0733 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 90.769 acc5_pos: 100.000 loss: 1.5503 (1.3160) acc1: 19.2708 (35.9092) acc5: 100.0000 (100.0000) acc1_neg: 16.146 acc5_neg: 100.000 time: 0.1394 data: 0.0498 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 90.769 acc5_pos: 100.000 loss: 1.5509 (1.3263) acc1: 19.7917 (35.2397) acc5: 100.0000 (100.0000) acc1_neg: 21.094 acc5_neg: 100.000 time: 0.1303 data: 0.0408 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 90.769 acc5_pos: 100.000 loss: 1.5564 (1.3309) acc1: 19.7917 (35.0182) acc5: 100.0000 (100.0000) acc1_neg: 13.043 acc5_neg: 100.000 time: 0.1224 data: 0.0361 max mem: 3532 +Test: Total time: 0:00:50 (0.2152 s / it) +* Overall Acc@1 35.018 Acc@5 100.000 + target == real Acc@1 92.460 Acc@5 100.000 + target == fake Acc@1 14.243 Acc@5 100.000 + loss 1.331 + PR-AUC 0.8017 ROC-AUC 0.6095 EER 41.87% (thr=0.1400) +Accuracy of the network on the 89879 test images: 35.0% +[rank0]:[W725 16:12:09.588822177 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 16:12:27.431000 79551 torch/distributed/run.py:766] +W0725 16:12:27.431000 79551 torch/distributed/run.py:766] ***************************************** +W0725 16:12:27.431000 79551 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 16:12:27.431000 79551 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 16:12:53.175805382 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 16:12:53.832901029 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 16:12:53.032359852 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 16:12:54.213957809 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 16:12:54.259245411 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W725 16:12:54.310732321 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank6]:[W725 16:12:54.322331054 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 16:12:54.391110133 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n16_seed2_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n16_classseed2_resampled_n116-041-153_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:22:37 acc1_pos: 67.969 acc5_pos: 100.000 loss: 0.5926 (0.5926) acc1: 67.9688 (67.9688) acc5: 100.0000 (100.0000) time: 5.0464 data: 3.0017 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/269] eta: 0:02:19 acc1_pos: 90.625 acc5_pos: 100.000 loss: 0.5926 (0.5775) acc1: 66.1458 (67.8030) acc5: 100.0000 (100.0000) time: 0.5404 data: 0.2732 max mem: 3531 +Test: [ 20/269] eta: 0:01:22 acc1_pos: 82.552 acc5_pos: 100.000 loss: 0.5636 (0.5720) acc1: 66.1458 (68.8988) acc5: 100.0000 (100.0000) time: 0.0974 data: 0.0076 max mem: 3531 +Test: [ 30/269] eta: 0:01:01 acc1_pos: 68.750 acc5_pos: 100.000 loss: 0.5198 (0.5557) acc1: 74.7396 (70.8417) acc5: 100.0000 (100.0000) time: 0.1019 data: 0.0077 max mem: 3531 +Test: [ 40/269] eta: 0:00:59 acc1_pos: 32.759 acc5_pos: 100.000 loss: 0.5294 (0.5783) acc1: 69.0104 (69.6900) acc5: 100.0000 (100.0000) acc1_neg: 54.948 acc5_neg: 100.000 time: 0.1810 data: 0.0839 max mem: 3531 +Test: [ 50/269] eta: 0:00:54 acc1_pos: 32.759 acc5_pos: 100.000 loss: 0.6561 (0.6083) acc1: 64.3229 (68.1066) acc5: 100.0000 (100.0000) acc1_neg: 33.073 acc5_neg: 100.000 time: 0.2361 data: 0.1405 max mem: 3531 +Test: [ 60/269] eta: 0:00:51 acc1_pos: 32.759 acc5_pos: 100.000 loss: 0.6773 (0.6339) acc1: 64.0625 (66.3337) acc5: 100.0000 (100.0000) acc1_neg: 64.583 acc5_neg: 100.000 time: 0.2255 data: 0.1302 max mem: 3531 +Test: [ 70/269] eta: 0:00:48 acc1_pos: 32.759 acc5_pos: 100.000 loss: 0.6997 (0.6395) acc1: 56.7708 (65.4013) acc5: 100.0000 (100.0000) acc1_neg: 53.906 acc5_neg: 100.000 time: 0.2387 data: 0.1437 max mem: 3531 +Test: [ 80/269] eta: 0:00:44 acc1_pos: 32.759 acc5_pos: 100.000 loss: 0.7127 (0.6486) acc1: 53.9062 (64.2168) acc5: 100.0000 (100.0000) acc1_neg: 36.979 acc5_neg: 100.000 time: 0.2001 data: 0.1034 max mem: 3531 +Test: [ 90/269] eta: 0:00:40 acc1_pos: 32.759 acc5_pos: 100.000 loss: 0.7760 (0.6623) acc1: 51.5625 (62.7890) acc5: 100.0000 (100.0000) acc1_neg: 65.104 acc5_neg: 100.000 time: 0.1553 data: 0.0592 max mem: 3531 +Test: [100/269] eta: 0:00:38 acc1_pos: 32.759 acc5_pos: 100.000 loss: 0.7803 (0.6773) acc1: 51.5625 (61.8013) acc5: 100.0000 (100.0000) acc1_neg: 53.385 acc5_neg: 100.000 time: 0.1925 data: 0.0982 max mem: 3531 +Test: [110/269] eta: 0:00:36 acc1_pos: 32.759 acc5_pos: 100.000 loss: 0.8148 (0.6920) acc1: 52.3438 (60.6489) acc5: 100.0000 (100.0000) acc1_neg: 46.615 acc5_neg: 100.000 time: 0.2295 data: 0.1334 max mem: 3531 +Test: [120/269] eta: 0:00:33 acc1_pos: 32.759 acc5_pos: 100.000 loss: 0.8673 (0.7024) acc1: 47.6562 (59.9281) acc5: 100.0000 (100.0000) acc1_neg: 47.656 acc5_neg: 100.000 time: 0.2157 data: 0.1177 max mem: 3531 +Test: [130/269] eta: 0:00:30 acc1_pos: 32.759 acc5_pos: 100.000 loss: 0.8224 (0.7168) acc1: 47.1354 (58.7389) acc5: 100.0000 (100.0000) acc1_neg: 47.917 acc5_neg: 100.000 time: 0.1974 data: 0.0995 max mem: 3531 +Test: [140/269] eta: 0:00:28 acc1_pos: 32.759 acc5_pos: 100.000 loss: 0.7381 (0.7153) acc1: 54.6875 (58.7304) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.1756 data: 0.0804 max mem: 3531 +Test: [150/269] eta: 0:00:25 acc1_pos: 32.759 acc5_pos: 100.000 loss: 0.6432 (0.7088) acc1: 62.5000 (59.2301) acc5: 100.0000 (100.0000) acc1_neg: 72.135 acc5_neg: 100.000 time: 0.1532 data: 0.0587 max mem: 3531 +Test: [160/269] eta: 0:00:23 acc1_pos: 32.759 acc5_pos: 100.000 loss: 0.6632 (0.7093) acc1: 59.3750 (59.1566) acc5: 100.0000 (100.0000) acc1_neg: 54.427 acc5_neg: 100.000 time: 0.1934 data: 0.0988 max mem: 3531 +Test: [170/269] eta: 0:00:21 acc1_pos: 32.759 acc5_pos: 100.000 loss: 0.8273 (0.7209) acc1: 46.6146 (58.1719) acc5: 100.0000 (100.0000) acc1_neg: 51.042 acc5_neg: 100.000 time: 0.2512 data: 0.1551 max mem: 3531 +Test: [180/269] eta: 0:00:19 acc1_pos: 32.759 acc5_pos: 100.000 loss: 0.8566 (0.7269) acc1: 44.7917 (57.7075) acc5: 100.0000 (100.0000) acc1_neg: 40.625 acc5_neg: 100.000 time: 0.2180 data: 0.1214 max mem: 3531 +Test: [190/269] eta: 0:00:16 acc1_pos: 32.759 acc5_pos: 100.000 loss: 0.7718 (0.7319) acc1: 49.7396 (57.2876) acc5: 100.0000 (100.0000) acc1_neg: 38.802 acc5_neg: 100.000 time: 0.1378 data: 0.0439 max mem: 3531 +Test: [200/269] eta: 0:00:14 acc1_pos: 32.759 acc5_pos: 100.000 loss: 0.8878 (0.7407) acc1: 45.5729 (56.7086) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.1127 data: 0.0168 max mem: 3531 +Test: [210/269] eta: 0:00:12 acc1_pos: 32.759 acc5_pos: 100.000 loss: 0.7056 (0.7385) acc1: 52.6042 (56.8671) acc5: 100.0000 (100.0000) acc1_neg: 63.281 acc5_neg: 100.000 time: 0.1749 data: 0.0796 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 32.759 acc5_pos: 100.000 loss: 0.6493 (0.7372) acc1: 63.2812 (57.0454) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1990 data: 0.1053 max mem: 3532 +Test: [230/269] eta: 0:00:07 acc1_pos: 32.759 acc5_pos: 100.000 loss: 0.6455 (0.7348) acc1: 61.9792 (57.2533) acc5: 100.0000 (100.0000) acc1_neg: 54.167 acc5_neg: 100.000 time: 0.1928 data: 0.0970 max mem: 3532 +Test: [240/269] eta: 0:00:05 acc1_pos: 32.759 acc5_pos: 100.000 loss: 0.7908 (0.7399) acc1: 53.3854 (56.7892) acc5: 100.0000 (100.0000) acc1_neg: 53.646 acc5_neg: 100.000 time: 0.1957 data: 0.1010 max mem: 3532 +Test: [250/269] eta: 0:00:03 acc1_pos: 32.759 acc5_pos: 100.000 loss: 0.8364 (0.7423) acc1: 48.9583 (56.5737) acc5: 100.0000 (100.0000) acc1_neg: 45.052 acc5_neg: 100.000 time: 0.1778 data: 0.0850 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 32.759 acc5_pos: 100.000 loss: 0.8549 (0.7467) acc1: 45.5729 (56.1652) acc5: 100.0000 (100.0000) acc1_neg: 45.312 acc5_neg: 100.000 time: 0.1617 data: 0.0703 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 32.759 acc5_pos: 100.000 loss: 0.8549 (0.7500) acc1: 45.0521 (55.9631) acc5: 100.0000 (100.0000) acc1_neg: 71.129 acc5_neg: 100.000 time: 0.1507 data: 0.0570 max mem: 3532 +Test: Total time: 0:00:53 (0.1984 s / it) +* Overall Acc@1 55.963 Acc@5 100.000 + target == real Acc@1 70.268 Acc@5 100.000 + target == fake Acc@1 54.011 Acc@5 100.000 + loss 0.750 + PR-AUC 0.9319 ROC-AUC 0.6634 EER 38.25% (thr=0.4485) +Accuracy of the network on the 103293 test images: 56.0% +[rank0]:[W725 16:13:58.916238691 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-25T16:14:05Z] eval_done tag=fixed500k_fake_n16_seed2_resampled +[2026-07-25T16:14:05Z] eval_start tag=fixed500k_fake_n22_seed1_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed1_resampled_n116-041-188_20260725_112559/best_checkpoint.pth +started_utc=2026-07-25T16:14:05Z +hostname=n116-042-186 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed1_resampled_n116-041-188_20260725_112559/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n22_seed1_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0725 16:14:17.028000 80120 torch/distributed/run.py:766] +W0725 16:14:17.028000 80120 torch/distributed/run.py:766] ***************************************** +W0725 16:14:17.028000 80120 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 16:14:17.028000 80120 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 16:14:45.302607063 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 16:14:45.616486140 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 16:14:45.634443423 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W725 16:14:45.642015268 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 16:14:45.735976201 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 16:14:45.784688881 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 16:14:45.839156669 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 16:14:45.860896632 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n22_seed1_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed1_resampled_n116-041-188_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:28:01 acc1_pos: 77.083 acc5_pos: 100.000 loss: 0.5133 (0.5133) acc1: 77.0833 (77.0833) acc5: 100.0000 (100.0000) time: 7.1839 data: 5.2145 max mem: 3531 +Test: [ 10/234] eta: 0:02:46 acc1_pos: 86.198 acc5_pos: 100.000 loss: 0.4212 (0.4665) acc1: 83.8542 (79.7112) acc5: 100.0000 (100.0000) time: 0.7427 data: 0.4744 max mem: 3531 +Test: [ 20/234] eta: 0:01:33 acc1_pos: 96.094 acc5_pos: 100.000 loss: 0.4212 (0.4902) acc1: 81.2500 (77.5298) acc5: 100.0000 (100.0000) time: 0.0993 data: 0.0004 max mem: 3531 +Test: [ 30/234] eta: 0:01:20 acc1_pos: 94.531 acc5_pos: 100.000 loss: 0.2059 (0.3958) acc1: 95.5729 (83.8374) acc5: 100.0000 (100.0000) time: 0.1999 data: 0.1044 max mem: 3531 +Test: [ 40/234] eta: 0:01:09 acc1_pos: 98.958 acc5_pos: 100.000 loss: 0.1891 (0.3526) acc1: 97.1354 (86.9474) acc5: 100.0000 (100.0000) time: 0.2749 data: 0.1833 max mem: 3531 +Test: [ 50/234] eta: 0:01:01 acc1_pos: 98.698 acc5_pos: 100.000 loss: 0.1461 (0.3110) acc1: 98.1771 (89.1544) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1460 max mem: 3531 +Test: [ 60/234] eta: 0:00:53 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.1251 (0.2768) acc1: 98.6979 (90.8982) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2008 data: 0.1041 max mem: 3531 +Test: [ 70/234] eta: 0:00:47 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.1039 (0.2911) acc1: 99.7396 (89.6714) acc5: 100.0000 (100.0000) acc1_neg: 72.656 acc5_neg: 100.000 time: 0.1749 data: 0.0791 max mem: 3531 +Test: [ 80/234] eta: 0:00:42 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.4843 (0.3058) acc1: 74.4792 (88.7217) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1852 data: 0.0902 max mem: 3531 +Test: [ 90/234] eta: 0:00:38 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.2799 (0.2949) acc1: 89.3229 (89.1798) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.1910 data: 0.0967 max mem: 3531 +Test: [100/234] eta: 0:00:34 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.2979 (0.3110) acc1: 85.6771 (88.0389) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1987 data: 0.1062 max mem: 3531 +Test: [110/234] eta: 0:00:31 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.4932 (0.3279) acc1: 73.1771 (86.8009) acc5: 100.0000 (100.0000) acc1_neg: 65.625 acc5_neg: 100.000 time: 0.1841 data: 0.0910 max mem: 3531 +Test: [120/234] eta: 0:00:27 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.5430 (0.3548) acc1: 69.2708 (85.0508) acc5: 100.0000 (100.0000) acc1_neg: 52.865 acc5_neg: 100.000 time: 0.1562 data: 0.0639 max mem: 3531 +Test: [130/234] eta: 0:00:24 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.5744 (0.3704) acc1: 67.9688 (83.9337) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.1717 data: 0.0779 max mem: 3531 +Test: [140/234] eta: 0:00:22 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.5744 (0.3912) acc1: 67.9688 (82.9085) acc5: 100.0000 (100.0000) acc1_neg: 45.833 acc5_neg: 100.000 time: 0.1986 data: 0.1044 max mem: 3531 +Test: [150/234] eta: 0:00:19 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.5065 (0.4029) acc1: 72.1354 (82.0899) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.1884 data: 0.0956 max mem: 3531 +Test: [160/234] eta: 0:00:17 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.4947 (0.4100) acc1: 72.1354 (81.6269) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.1961 data: 0.1028 max mem: 3531 +Test: [170/234] eta: 0:00:14 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.6293 (0.4240) acc1: 63.8021 (80.6150) acc5: 100.0000 (100.0000) acc1_neg: 63.802 acc5_neg: 100.000 time: 0.2307 data: 0.1391 max mem: 3531 +Test: [180/234] eta: 0:00:12 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.6381 (0.4373) acc1: 63.2812 (79.7695) acc5: 100.0000 (100.0000) acc1_neg: 40.104 acc5_neg: 100.000 time: 0.2064 data: 0.1158 max mem: 3531 +Test: [190/234] eta: 0:00:09 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.8780 (0.4621) acc1: 50.2604 (78.1959) acc5: 100.0000 (100.0000) acc1_neg: 48.438 acc5_neg: 100.000 time: 0.1693 data: 0.0768 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.8780 (0.4592) acc1: 50.2604 (78.3543) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1598 data: 0.0669 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.1099 (0.4443) acc1: 99.7396 (79.2901) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1338 data: 0.0410 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.2031 (0.4348) acc1: 95.3125 (79.9361) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.1132 data: 0.0214 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.4279 (0.4513) acc1: 53.6458 (78.6503) acc5: 100.0000 (100.0000) acc1_neg: 49.479 acc5_neg: 100.000 time: 0.1096 data: 0.0197 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 96.552 acc5_pos: 100.000 loss: 0.7982 (0.4559) acc1: 51.8229 (78.3258) acc5: 100.0000 (100.0000) acc1_neg: 49.805 acc5_neg: 100.000 time: 0.1073 data: 0.0164 max mem: 3532 +Test: Total time: 0:00:48 (0.2072 s / it) +* Overall Acc@1 78.326 Acc@5 100.000 + target == real Acc@1 89.209 Acc@5 100.000 + target == fake Acc@1 75.259 Acc@5 100.000 + loss 0.456 + PR-AUC 0.9716 ROC-AUC 0.9040 EER 18.09% (thr=0.3960) +Accuracy of the network on the 89729 test images: 78.3% +[rank0]:[W725 16:15:50.943905871 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 16:16:07.976000 80676 torch/distributed/run.py:766] +W0725 16:16:07.976000 80676 torch/distributed/run.py:766] ***************************************** +W0725 16:16:07.976000 80676 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 16:16:07.976000 80676 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 16:16:34.318427943 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 16:16:34.544115401 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 16:16:34.812212883 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank1]:[W725 16:16:34.814989480 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 16:16:34.979019434 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 16:16:35.124917196 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 16:16:35.149687967 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 16:16:35.201402462 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n22_seed1_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed1_resampled_n116-041-188_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:18:52 acc1_pos: 57.552 acc5_pos: 100.000 loss: 0.6550 (0.6550) acc1: 57.5521 (57.5521) acc5: 100.0000 (100.0000) time: 6.5071 data: 4.8030 max mem: 3531 +Test: [ 10/2387] eta: 0:26:58 acc1_pos: 41.927 acc5_pos: 100.000 loss: 0.8264 (0.7901) acc1: 48.9583 (50.2841) acc5: 100.0000 (100.0000) time: 0.6809 data: 0.4370 max mem: 3531 +Test: [ 20/2387] eta: 0:15:56 acc1_pos: 40.885 acc5_pos: 100.000 loss: 0.7712 (0.7752) acc1: 47.3958 (51.6989) acc5: 100.0000 (100.0000) time: 0.0991 data: 0.0030 max mem: 3531 +Test: [ 30/2387] eta: 0:14:27 acc1_pos: 26.042 acc5_pos: 100.000 loss: 0.7820 (0.7972) acc1: 47.3958 (49.5548) acc5: 100.0000 (100.0000) time: 0.1957 data: 0.1027 max mem: 3531 +Test: [ 40/2387] eta: 0:14:07 acc1_pos: 40.104 acc5_pos: 100.000 loss: 0.8183 (0.8002) acc1: 45.8333 (48.5582) acc5: 100.0000 (100.0000) time: 0.3157 data: 0.2236 max mem: 3531 +Test: [ 50/2387] eta: 0:13:12 acc1_pos: 55.469 acc5_pos: 100.000 loss: 0.7335 (0.7816) acc1: 50.0000 (50.4289) acc5: 100.0000 (100.0000) time: 0.2942 data: 0.2002 max mem: 3531 +Test: [ 60/2387] eta: 0:12:42 acc1_pos: 64.062 acc5_pos: 100.000 loss: 0.7490 (0.7856) acc1: 47.3958 (50.1580) acc5: 100.0000 (100.0000) time: 0.2591 data: 0.1642 max mem: 3531 +Test: [ 70/2387] eta: 0:12:21 acc1_pos: 52.344 acc5_pos: 100.000 loss: 0.8387 (0.7982) acc1: 43.4896 (48.9473) acc5: 100.0000 (100.0000) time: 0.2714 data: 0.1767 max mem: 3531 +Test: [ 80/2387] eta: 0:12:00 acc1_pos: 58.594 acc5_pos: 100.000 loss: 0.8274 (0.7941) acc1: 45.8333 (49.1191) acc5: 100.0000 (100.0000) time: 0.2650 data: 0.1705 max mem: 3531 +Test: [ 90/2387] eta: 0:11:45 acc1_pos: 52.083 acc5_pos: 100.000 loss: 0.7313 (0.7830) acc1: 54.4271 (49.8827) acc5: 100.0000 (100.0000) time: 0.2608 data: 0.1684 max mem: 3531 +Test: [ 100/2387] eta: 0:11:27 acc1_pos: 57.812 acc5_pos: 100.000 loss: 0.7313 (0.7827) acc1: 53.1250 (49.8814) acc5: 100.0000 (100.0000) time: 0.2536 data: 0.1603 max mem: 3531 +Test: [ 110/2387] eta: 0:11:11 acc1_pos: 56.250 acc5_pos: 100.000 loss: 0.6735 (0.7713) acc1: 56.2500 (50.7390) acc5: 100.0000 (100.0000) time: 0.2396 data: 0.1443 max mem: 3531 +Test: [ 120/2387] eta: 0:10:56 acc1_pos: 51.302 acc5_pos: 100.000 loss: 0.6714 (0.7660) acc1: 60.4167 (51.1794) acc5: 100.0000 (100.0000) time: 0.2340 data: 0.1381 max mem: 3531 +Test: [ 130/2387] eta: 0:10:40 acc1_pos: 55.990 acc5_pos: 100.000 loss: 0.6961 (0.7617) acc1: 55.9896 (51.5128) acc5: 100.0000 (100.0000) time: 0.2216 data: 0.1255 max mem: 3531 +Test: [ 140/2387] eta: 0:10:31 acc1_pos: 55.469 acc5_pos: 100.000 loss: 0.6875 (0.7564) acc1: 56.7708 (51.8008) acc5: 100.0000 (100.0000) time: 0.2307 data: 0.1364 max mem: 3531 +Test: [ 150/2387] eta: 0:10:18 acc1_pos: 33.854 acc5_pos: 100.000 loss: 0.6761 (0.7514) acc1: 56.7708 (52.1713) acc5: 100.0000 (100.0000) time: 0.2282 data: 0.1350 max mem: 3531 +Test: [ 160/2387] eta: 0:10:01 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.6535 (0.7389) acc1: 59.8958 (53.3789) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1935 data: 0.0978 max mem: 3531 +Test: [ 170/2387] eta: 0:09:45 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2790 (0.7112) acc1: 93.2292 (55.7596) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1710 data: 0.0738 max mem: 3531 +Test: [ 180/2387] eta: 0:09:38 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2637 (0.6864) acc1: 94.0104 (57.8816) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1952 data: 0.1021 max mem: 3531 +Test: [ 190/2387] eta: 0:09:26 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2591 (0.6641) acc1: 94.0104 (59.7718) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.2038 data: 0.1113 max mem: 3531 +Test: [ 200/2387] eta: 0:09:19 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2581 (0.6440) acc1: 94.2708 (61.4661) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1996 data: 0.1043 max mem: 3531 +Test: [ 210/2387] eta: 0:09:10 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2656 (0.6262) acc1: 94.0104 (63.0060) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2055 data: 0.1123 max mem: 3532 +Test: [ 220/2387] eta: 0:09:00 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2670 (0.6099) acc1: 94.0104 (64.4137) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1849 data: 0.0921 max mem: 3532 +Test: [ 230/2387] eta: 0:08:44 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2670 (0.5949) acc1: 94.2708 (65.7005) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1411 data: 0.0445 max mem: 3532 +Test: [ 240/2387] eta: 0:08:38 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2609 (0.5809) acc1: 94.5312 (66.9303) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1564 data: 0.0619 max mem: 3532 +Test: [ 250/2387] eta: 0:08:31 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2397 (0.5621) acc1: 96.3542 (68.2271) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1984 data: 0.1077 max mem: 3532 +Test: [ 260/2387] eta: 0:08:26 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.0814 (0.5438) acc1: 100.0000 (69.4444) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1934 data: 0.0988 max mem: 3532 +Test: [ 270/2387] eta: 0:08:23 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.0808 (0.5269) acc1: 100.0000 (70.5710) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2145 data: 0.1180 max mem: 3532 +Test: [ 280/2387] eta: 0:08:14 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.0831 (0.5113) acc1: 100.0000 (71.6183) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0947 max mem: 3532 +Test: [ 290/2387] eta: 0:08:10 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.0785 (0.4964) acc1: 100.0000 (72.5936) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1813 data: 0.0864 max mem: 3532 +Test: [ 300/2387] eta: 0:08:08 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.0763 (0.4825) acc1: 100.0000 (73.5041) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2303 data: 0.1382 max mem: 3532 +Test: [ 310/2387] eta: 0:08:02 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.0776 (0.4750) acc1: 100.0000 (74.1752) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.2105 data: 0.1172 max mem: 3532 +Test: [ 320/2387] eta: 0:07:56 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.4647 (0.4750) acc1: 87.7604 (74.5538) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.1763 data: 0.0814 max mem: 3532 +Test: [ 330/2387] eta: 0:07:53 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.4713 (0.4749) acc1: 85.6771 (74.8922) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.1979 data: 0.1056 max mem: 3532 +Test: [ 340/2387] eta: 0:07:50 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.4689 (0.4747) acc1: 85.4167 (75.2108) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.2131 data: 0.1179 max mem: 3532 +Test: [ 350/2387] eta: 0:07:44 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.4560 (0.4673) acc1: 87.7604 (75.8733) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.0984 max mem: 3532 +Test: [ 360/2387] eta: 0:07:42 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1963 (0.4602) acc1: 97.3958 (76.4572) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2011 data: 0.1073 max mem: 3532 +Test: [ 370/2387] eta: 0:07:37 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2107 (0.4535) acc1: 96.8750 (77.0054) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1999 data: 0.1070 max mem: 3532 +Test: [ 380/2387] eta: 0:07:32 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2107 (0.4471) acc1: 97.1354 (77.5303) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1837 data: 0.0866 max mem: 3532 +Test: [ 390/2387] eta: 0:07:26 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2104 (0.4411) acc1: 96.8750 (78.0264) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1663 data: 0.0669 max mem: 3532 +Test: [ 400/2387] eta: 0:07:20 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2123 (0.4354) acc1: 96.6146 (78.4874) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1499 data: 0.0552 max mem: 3532 +Test: [ 410/2387] eta: 0:07:17 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2103 (0.4299) acc1: 96.8750 (78.9392) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1734 data: 0.0784 max mem: 3532 +Test: [ 420/2387] eta: 0:07:12 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2085 (0.4246) acc1: 96.8750 (79.3621) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1809 data: 0.0874 max mem: 3532 +Test: [ 430/2387] eta: 0:07:09 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2098 (0.4198) acc1: 96.8750 (79.7618) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1865 data: 0.0932 max mem: 3532 +Test: [ 440/2387] eta: 0:07:04 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2095 (0.4132) acc1: 97.3958 (80.2012) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1798 data: 0.0846 max mem: 3533 +Test: [ 450/2387] eta: 0:07:02 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.0805 (0.4059) acc1: 100.0000 (80.6402) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1833 data: 0.0879 max mem: 3533 +Test: [ 460/2387] eta: 0:06:59 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.0804 (0.3988) acc1: 100.0000 (81.0596) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2040 data: 0.1078 max mem: 3533 +Test: [ 470/2387] eta: 0:06:55 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.0819 (0.3921) acc1: 100.0000 (81.4612) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0971 max mem: 3533 +Test: [ 480/2387] eta: 0:06:51 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.0815 (0.3857) acc1: 100.0000 (81.8466) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1758 data: 0.0836 max mem: 3533 +Test: [ 490/2387] eta: 0:06:48 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.0815 (0.3795) acc1: 100.0000 (82.2164) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1768 data: 0.0845 max mem: 3533 +Test: [ 500/2387] eta: 0:06:44 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.0818 (0.3735) acc1: 100.0000 (82.5713) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1767 data: 0.0850 max mem: 3533 +Test: [ 510/2387] eta: 0:06:40 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.0822 (0.3681) acc1: 100.0000 (82.9124) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1758 data: 0.0827 max mem: 3533 +Test: [ 520/2387] eta: 0:06:37 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1284 (0.3636) acc1: 100.0000 (83.2404) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1779 data: 0.0847 max mem: 3533 +Test: [ 530/2387] eta: 0:06:33 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1329 (0.3592) acc1: 100.0000 (83.5545) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1702 data: 0.0780 max mem: 3533 +Test: [ 540/2387] eta: 0:06:30 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1326 (0.3551) acc1: 100.0000 (83.8566) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1704 data: 0.0796 max mem: 3533 +Test: [ 550/2387] eta: 0:06:27 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1325 (0.3510) acc1: 100.0000 (84.1486) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1777 data: 0.0866 max mem: 3533 +Test: [ 560/2387] eta: 0:06:24 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1337 (0.3471) acc1: 100.0000 (84.4298) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1856 data: 0.0942 max mem: 3533 +Test: [ 570/2387] eta: 0:06:21 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1332 (0.3434) acc1: 100.0000 (84.7020) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1950 data: 0.1043 max mem: 3533 +Test: [ 580/2387] eta: 0:06:18 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1319 (0.3398) acc1: 100.0000 (84.9640) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1879 data: 0.0951 max mem: 3533 +Test: [ 590/2387] eta: 0:06:16 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1332 (0.3363) acc1: 100.0000 (85.2184) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1864 data: 0.0924 max mem: 3533 +Test: [ 600/2387] eta: 0:06:13 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1332 (0.3329) acc1: 100.0000 (85.4630) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2003 data: 0.1076 max mem: 3533 +Test: [ 610/2387] eta: 0:06:09 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1511 (0.3306) acc1: 98.9583 (85.6605) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1673 data: 0.0718 max mem: 3533 +Test: [ 620/2387] eta: 0:06:05 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1895 (0.3284) acc1: 97.6562 (85.8524) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1395 data: 0.0416 max mem: 3533 +Test: [ 630/2387] eta: 0:06:02 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1917 (0.3262) acc1: 97.6562 (86.0316) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1516 data: 0.0551 max mem: 3533 +Test: [ 640/2387] eta: 0:05:58 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1982 (0.3242) acc1: 97.1354 (86.2073) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1549 data: 0.0569 max mem: 3533 +Test: [ 650/2387] eta: 0:05:56 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1982 (0.3222) acc1: 97.6562 (86.3787) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1682 data: 0.0716 max mem: 3533 +Test: [ 660/2387] eta: 0:05:54 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1929 (0.3203) acc1: 97.6562 (86.5458) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1949 data: 0.0986 max mem: 3533 +Test: [ 670/2387] eta: 0:05:53 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1943 (0.3184) acc1: 97.1354 (86.7044) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2349 data: 0.1357 max mem: 3534 +Test: [ 680/2387] eta: 0:05:49 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1873 (0.3162) acc1: 97.6562 (86.8905) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.0957 max mem: 3534 +Test: [ 690/2387] eta: 0:05:47 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1873 (0.3175) acc1: 98.1771 (86.8058) acc5: 100.0000 (100.0000) acc1_neg: 69.271 acc5_neg: 100.000 time: 0.1667 data: 0.0715 max mem: 3534 +Test: [ 700/2387] eta: 0:05:46 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.4950 (0.3200) acc1: 75.2604 (86.6333) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.2216 data: 0.1312 max mem: 3534 +Test: [ 710/2387] eta: 0:05:43 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.4988 (0.3226) acc1: 73.9583 (86.4591) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.2131 data: 0.1214 max mem: 3534 +Test: [ 720/2387] eta: 0:05:41 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.4941 (0.3249) acc1: 75.2604 (86.3092) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.1932 data: 0.0964 max mem: 3534 +Test: [ 730/2387] eta: 0:05:38 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.4964 (0.3274) acc1: 74.7396 (86.1327) acc5: 100.0000 (100.0000) acc1_neg: 71.615 acc5_neg: 100.000 time: 0.1876 data: 0.0923 max mem: 3534 +Test: [ 740/2387] eta: 0:05:36 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.4974 (0.3297) acc1: 73.9583 (85.9818) acc5: 100.0000 (100.0000) acc1_neg: 73.958 acc5_neg: 100.000 time: 0.1812 data: 0.0909 max mem: 3534 +Test: [ 750/2387] eta: 0:05:34 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.4898 (0.3318) acc1: 75.0000 (85.8310) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.1972 data: 0.1055 max mem: 3534 +Test: [ 760/2387] eta: 0:05:33 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.4770 (0.3331) acc1: 75.0000 (85.7472) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.2186 data: 0.1261 max mem: 3534 +Test: [ 770/2387] eta: 0:05:31 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.3414 (0.3322) acc1: 85.6771 (85.8075) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2199 data: 0.1245 max mem: 3534 +Test: [ 780/2387] eta: 0:05:28 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1780 (0.3299) acc1: 95.8333 (85.9538) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1899 data: 0.0917 max mem: 3534 +Test: [ 790/2387] eta: 0:05:25 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1444 (0.3280) acc1: 96.8750 (86.0458) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1770 data: 0.0827 max mem: 3534 +Test: [ 800/2387] eta: 0:05:24 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1441 (0.3257) acc1: 98.4375 (86.2012) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2050 data: 0.1144 max mem: 3534 +Test: [ 810/2387] eta: 0:05:21 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1440 (0.3236) acc1: 98.4375 (86.3386) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2029 data: 0.1116 max mem: 3534 +Test: [ 820/2387] eta: 0:05:20 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1668 (0.3219) acc1: 97.1354 (86.4615) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.2013 data: 0.1092 max mem: 3534 +Test: [ 830/2387] eta: 0:05:19 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2020 (0.3212) acc1: 94.7917 (86.5154) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2466 data: 0.1554 max mem: 3534 +Test: [ 840/2387] eta: 0:05:16 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2647 (0.3206) acc1: 92.7083 (86.6042) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2312 data: 0.1400 max mem: 3534 +Test: [ 850/2387] eta: 0:05:13 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2756 (0.3202) acc1: 94.0104 (86.6878) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1681 data: 0.0733 max mem: 3534 +Test: [ 860/2387] eta: 0:05:11 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2652 (0.3193) acc1: 95.3125 (86.8010) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1784 data: 0.0825 max mem: 3534 +Test: [ 870/2387] eta: 0:05:09 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2368 (0.3185) acc1: 96.3542 (86.9020) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2071 data: 0.1146 max mem: 3534 +Test: [ 880/2387] eta: 0:05:07 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2465 (0.3179) acc1: 95.8333 (86.9925) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1896 data: 0.0987 max mem: 3534 +Test: [ 890/2387] eta: 0:05:04 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2178 (0.3165) acc1: 97.3958 (87.1215) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1705 data: 0.0766 max mem: 3534 +Test: [ 900/2387] eta: 0:05:02 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2338 (0.3195) acc1: 96.0938 (86.9326) acc5: 100.0000 (100.0000) acc1_neg: 66.146 acc5_neg: 100.000 time: 0.1695 data: 0.0752 max mem: 3535 +Test: [ 910/2387] eta: 0:04:59 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.6694 (0.3233) acc1: 63.0208 (86.6776) acc5: 100.0000 (100.0000) acc1_neg: 64.323 acc5_neg: 100.000 time: 0.1838 data: 0.0925 max mem: 3535 +Test: [ 920/2387] eta: 0:04:57 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.6803 (0.3272) acc1: 63.0208 (86.4091) acc5: 100.0000 (100.0000) acc1_neg: 59.115 acc5_neg: 100.000 time: 0.1914 data: 0.1004 max mem: 3535 +Test: [ 930/2387] eta: 0:04:55 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.6832 (0.3310) acc1: 62.2396 (86.1562) acc5: 100.0000 (100.0000) acc1_neg: 58.594 acc5_neg: 100.000 time: 0.2032 data: 0.1117 max mem: 3535 +Test: [ 940/2387] eta: 0:04:54 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.6801 (0.3347) acc1: 62.2396 (85.9018) acc5: 100.0000 (100.0000) acc1_neg: 63.542 acc5_neg: 100.000 time: 0.2181 data: 0.1266 max mem: 3535 +Test: [ 950/2387] eta: 0:04:51 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.6781 (0.3382) acc1: 63.0208 (85.6524) acc5: 100.0000 (100.0000) acc1_neg: 63.281 acc5_neg: 100.000 time: 0.2025 data: 0.1096 max mem: 3535 +Test: [ 960/2387] eta: 0:04:49 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.6642 (0.3416) acc1: 63.0208 (85.4224) acc5: 100.0000 (100.0000) acc1_neg: 65.625 acc5_neg: 100.000 time: 0.1895 data: 0.0946 max mem: 3535 +Test: [ 970/2387] eta: 0:04:47 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.6731 (0.3452) acc1: 62.2396 (85.1731) acc5: 100.0000 (100.0000) acc1_neg: 62.760 acc5_neg: 100.000 time: 0.1923 data: 0.0984 max mem: 3535 +Test: [ 980/2387] eta: 0:04:44 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.6455 (0.3456) acc1: 62.7604 (85.1448) acc5: 100.0000 (100.0000) acc1_neg: 76.302 acc5_neg: 100.000 time: 0.1801 data: 0.0848 max mem: 3535 +Test: [ 990/2387] eta: 0:04:42 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1909 (0.3439) acc1: 95.8333 (85.2422) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1756 data: 0.0806 max mem: 3535 +Test: [1000/2387] eta: 0:04:39 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1198 (0.3418) acc1: 98.9583 (85.3644) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1705 data: 0.0794 max mem: 3535 +Test: [1010/2387] eta: 0:04:37 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1313 (0.3401) acc1: 98.6979 (85.4787) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1747 data: 0.0836 max mem: 3535 +Test: [1020/2387] eta: 0:04:35 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1638 (0.3385) acc1: 97.9167 (85.5697) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1748 data: 0.0830 max mem: 3535 +Test: [1030/2387] eta: 0:04:33 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1152 (0.3364) acc1: 99.7396 (85.7018) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1906 data: 0.0990 max mem: 3535 +Test: [1040/2387] eta: 0:04:31 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.0987 (0.3341) acc1: 99.7396 (85.8329) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2101 data: 0.1182 max mem: 3535 +Test: [1050/2387] eta: 0:04:28 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.0925 (0.3319) acc1: 99.7396 (85.9633) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1845 data: 0.0919 max mem: 3535 +Test: [1060/2387] eta: 0:04:26 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.0950 (0.3297) acc1: 99.7396 (86.0919) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1676 data: 0.0741 max mem: 3535 +Test: [1070/2387] eta: 0:04:24 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1122 (0.3280) acc1: 99.4792 (86.2033) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1870 data: 0.0941 max mem: 3535 +Test: [1080/2387] eta: 0:04:22 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1714 (0.3269) acc1: 97.1354 (86.2784) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2045 data: 0.1110 max mem: 3535 +Test: [1090/2387] eta: 0:04:21 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1721 (0.3252) acc1: 96.8750 (86.3817) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2195 data: 0.1268 max mem: 3535 +Test: [1100/2387] eta: 0:04:19 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1379 (0.3237) acc1: 98.1771 (86.4796) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2370 data: 0.1448 max mem: 3535 +Test: [1110/2387] eta: 0:04:16 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1522 (0.3226) acc1: 97.3958 (86.5350) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1924 data: 0.0991 max mem: 3535 +Test: [1120/2387] eta: 0:04:14 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1534 (0.3210) acc1: 97.3958 (86.6335) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1812 data: 0.0898 max mem: 3536 +Test: [1130/2387] eta: 0:04:12 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1519 (0.3196) acc1: 97.6562 (86.7298) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2068 data: 0.1160 max mem: 3536 +Test: [1140/2387] eta: 0:04:10 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1666 (0.3183) acc1: 96.6146 (86.8144) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1853 data: 0.0934 max mem: 3536 +Test: [1150/2387] eta: 0:04:08 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1784 (0.3175) acc1: 96.6146 (86.8932) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1733 data: 0.0798 max mem: 3536 +Test: [1160/2387] eta: 0:04:05 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2000 (0.3164) acc1: 97.1354 (86.9742) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1583 data: 0.0656 max mem: 3536 +Test: [1170/2387] eta: 0:04:03 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1825 (0.3154) acc1: 97.9167 (87.0670) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1516 data: 0.0578 max mem: 3536 +Test: [1180/2387] eta: 0:04:01 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1849 (0.3143) acc1: 98.1771 (87.1518) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1705 data: 0.0763 max mem: 3536 +Test: [1190/2387] eta: 0:03:58 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1835 (0.3133) acc1: 98.1771 (87.2442) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1727 data: 0.0795 max mem: 3536 +Test: [1200/2387] eta: 0:03:56 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1571 (0.3119) acc1: 99.2188 (87.3450) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1670 data: 0.0727 max mem: 3536 +Test: [1210/2387] eta: 0:03:53 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1466 (0.3105) acc1: 99.7396 (87.4475) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1607 data: 0.0684 max mem: 3536 +Test: [1220/2387] eta: 0:03:51 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1144 (0.3088) acc1: 99.7396 (87.5471) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1667 data: 0.0749 max mem: 3536 +Test: [1230/2387] eta: 0:03:49 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.0952 (0.3071) acc1: 100.0000 (87.6479) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1611 data: 0.0694 max mem: 3536 +Test: [1240/2387] eta: 0:03:46 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.0928 (0.3054) acc1: 100.0000 (87.7455) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1525 data: 0.0599 max mem: 3536 +Test: [1250/2387] eta: 0:03:44 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.0931 (0.3037) acc1: 99.7396 (87.8420) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1680 data: 0.0741 max mem: 3536 +Test: [1260/2387] eta: 0:03:42 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.0892 (0.3020) acc1: 99.7396 (87.9362) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1679 data: 0.0753 max mem: 3536 +Test: [1270/2387] eta: 0:03:40 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.0860 (0.3003) acc1: 100.0000 (88.0305) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1653 data: 0.0738 max mem: 3536 +Test: [1280/2387] eta: 0:03:38 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.0876 (0.2988) acc1: 100.0000 (88.1182) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1770 data: 0.0854 max mem: 3536 +Test: [1290/2387] eta: 0:03:35 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1108 (0.2975) acc1: 98.9583 (88.1881) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1666 data: 0.0710 max mem: 3536 +Test: [1300/2387] eta: 0:03:33 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1226 (0.2962) acc1: 98.6979 (88.2636) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1678 data: 0.0716 max mem: 3536 +Test: [1310/2387] eta: 0:03:31 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1197 (0.2949) acc1: 98.9583 (88.3365) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1777 data: 0.0828 max mem: 3536 +Test: [1320/2387] eta: 0:03:29 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1294 (0.2937) acc1: 98.9583 (88.4074) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1504 data: 0.0543 max mem: 3536 +Test: [1330/2387] eta: 0:03:27 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1389 (0.2927) acc1: 98.4375 (88.4759) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1586 data: 0.0616 max mem: 3536 +Test: [1340/2387] eta: 0:03:24 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1691 (0.2931) acc1: 95.8333 (88.4564) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1787 data: 0.0843 max mem: 3536 +Test: [1350/2387] eta: 0:03:22 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.0698 (0.2913) acc1: 100.0000 (88.5419) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1662 data: 0.0745 max mem: 3537 +Test: [1360/2387] eta: 0:03:20 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.0591 (0.2896) acc1: 100.0000 (88.6261) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1549 data: 0.0621 max mem: 3537 +Test: [1370/2387] eta: 0:03:18 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.0578 (0.2880) acc1: 100.0000 (88.7088) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1499 data: 0.0567 max mem: 3537 +Test: [1380/2387] eta: 0:03:15 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.0579 (0.2863) acc1: 100.0000 (88.7906) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1560 data: 0.0632 max mem: 3537 +Test: [1390/2387] eta: 0:03:13 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.0576 (0.2847) acc1: 100.0000 (88.8712) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1667 data: 0.0749 max mem: 3537 +Test: [1400/2387] eta: 0:03:11 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.0576 (0.2830) acc1: 100.0000 (88.9506) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1689 data: 0.0772 max mem: 3537 +Test: [1410/2387] eta: 0:03:09 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.0584 (0.2824) acc1: 100.0000 (88.9863) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1630 data: 0.0698 max mem: 3537 +Test: [1420/2387] eta: 0:03:07 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2291 (0.2821) acc1: 92.7083 (88.9930) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1648 data: 0.0716 max mem: 3537 +Test: [1430/2387] eta: 0:03:05 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2333 (0.2819) acc1: 91.6667 (89.0134) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1625 data: 0.0704 max mem: 3537 +Test: [1440/2387] eta: 0:03:03 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2449 (0.2816) acc1: 92.1875 (89.0314) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.1654 data: 0.0723 max mem: 3537 +Test: [1450/2387] eta: 0:03:01 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2360 (0.2814) acc1: 93.4896 (89.0564) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1788 data: 0.0865 max mem: 3537 +Test: [1460/2387] eta: 0:02:58 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1838 (0.2806) acc1: 96.3542 (89.1087) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1745 data: 0.0811 max mem: 3537 +Test: [1470/2387] eta: 0:02:57 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1490 (0.2797) acc1: 97.6562 (89.1685) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1784 data: 0.0843 max mem: 3537 +Test: [1480/2387] eta: 0:02:54 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1380 (0.2788) acc1: 98.6979 (89.2280) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1749 data: 0.0781 max mem: 3537 +Test: [1490/2387] eta: 0:02:52 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1472 (0.2779) acc1: 98.9583 (89.2903) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1675 data: 0.0716 max mem: 3537 +Test: [1500/2387] eta: 0:02:50 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1480 (0.2771) acc1: 98.1771 (89.3481) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1763 data: 0.0860 max mem: 3537 +Test: [1510/2387] eta: 0:02:48 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1381 (0.2762) acc1: 98.9583 (89.4163) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1876 data: 0.0966 max mem: 3537 +Test: [1520/2387] eta: 0:02:46 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1590 (0.2758) acc1: 98.4375 (89.4570) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1768 data: 0.0851 max mem: 3537 +Test: [1530/2387] eta: 0:02:44 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2279 (0.2756) acc1: 97.1354 (89.4944) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1702 data: 0.0776 max mem: 3537 +Test: [1540/2387] eta: 0:02:42 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2279 (0.2753) acc1: 97.3958 (89.5433) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1735 data: 0.0807 max mem: 3537 +Test: [1550/2387] eta: 0:02:40 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2094 (0.2749) acc1: 97.3958 (89.5946) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1701 data: 0.0783 max mem: 3537 +Test: [1560/2387] eta: 0:02:38 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2109 (0.2746) acc1: 97.3958 (89.6359) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.1799 data: 0.0860 max mem: 3537 +Test: [1570/2387] eta: 0:02:36 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1781 (0.2740) acc1: 98.9583 (89.6942) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1756 data: 0.0815 max mem: 3537 +Test: [1580/2387] eta: 0:02:34 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1768 (0.2734) acc1: 98.4375 (89.7344) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1689 data: 0.0766 max mem: 3538 +Test: [1590/2387] eta: 0:02:32 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1661 (0.2726) acc1: 97.3958 (89.7847) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1696 data: 0.0757 max mem: 3538 +Test: [1600/2387] eta: 0:02:30 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1569 (0.2721) acc1: 97.3958 (89.8050) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1700 data: 0.0756 max mem: 3538 +Test: [1610/2387] eta: 0:02:28 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1528 (0.2713) acc1: 98.1771 (89.8594) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1738 data: 0.0789 max mem: 3538 +Test: [1620/2387] eta: 0:02:26 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1573 (0.2708) acc1: 96.6146 (89.8850) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1674 data: 0.0735 max mem: 3538 +Test: [1630/2387] eta: 0:02:24 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2046 (0.2705) acc1: 94.5312 (89.9084) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1702 data: 0.0788 max mem: 3538 +Test: [1640/2387] eta: 0:02:22 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2272 (0.2704) acc1: 92.1875 (89.9187) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1664 data: 0.0738 max mem: 3538 +Test: [1650/2387] eta: 0:02:20 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2333 (0.2702) acc1: 92.1875 (89.9338) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1610 data: 0.0682 max mem: 3538 +Test: [1660/2387] eta: 0:02:18 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2438 (0.2702) acc1: 92.1875 (89.9386) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1725 data: 0.0804 max mem: 3538 +Test: [1670/2387] eta: 0:02:16 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2567 (0.2700) acc1: 91.4062 (89.9441) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1704 data: 0.0770 max mem: 3538 +Test: [1680/2387] eta: 0:02:14 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2441 (0.2700) acc1: 91.4062 (89.9505) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1738 data: 0.0818 max mem: 3538 +Test: [1690/2387] eta: 0:02:12 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2374 (0.2698) acc1: 92.9688 (89.9633) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1729 data: 0.0811 max mem: 3538 +Test: [1700/2387] eta: 0:02:10 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2356 (0.2695) acc1: 93.4896 (89.9884) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1612 data: 0.0686 max mem: 3538 +Test: [1710/2387] eta: 0:02:08 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1494 (0.2686) acc1: 98.1771 (90.0415) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1667 data: 0.0755 max mem: 3538 +Test: [1720/2387] eta: 0:02:06 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1176 (0.2677) acc1: 98.9583 (90.0943) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1693 data: 0.0778 max mem: 3538 +Test: [1730/2387] eta: 0:02:04 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1190 (0.2669) acc1: 99.2188 (90.1437) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1619 data: 0.0690 max mem: 3538 +Test: [1740/2387] eta: 0:02:02 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1186 (0.2661) acc1: 99.4792 (90.1963) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1646 data: 0.0713 max mem: 3538 +Test: [1750/2387] eta: 0:02:00 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1044 (0.2652) acc1: 99.7396 (90.2496) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1727 data: 0.0798 max mem: 3538 +Test: [1760/2387] eta: 0:01:58 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1035 (0.2643) acc1: 99.7396 (90.3022) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1632 data: 0.0708 max mem: 3538 +Test: [1770/2387] eta: 0:01:56 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1152 (0.2636) acc1: 99.2188 (90.3455) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1522 data: 0.0606 max mem: 3538 +Test: [1780/2387] eta: 0:01:54 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1534 (0.2629) acc1: 98.1771 (90.3911) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1541 data: 0.0633 max mem: 3538 +Test: [1790/2387] eta: 0:01:52 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1351 (0.2622) acc1: 98.4375 (90.4366) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1548 data: 0.0632 max mem: 3538 +Test: [1800/2387] eta: 0:01:50 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1345 (0.2615) acc1: 98.6979 (90.4837) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1592 data: 0.0673 max mem: 3538 +Test: [1810/2387] eta: 0:01:48 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1233 (0.2607) acc1: 99.2188 (90.5323) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1660 data: 0.0746 max mem: 3539 +Test: [1820/2387] eta: 0:01:46 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.0985 (0.2599) acc1: 99.4792 (90.5808) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1707 data: 0.0769 max mem: 3539 +Test: [1830/2387] eta: 0:01:44 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1132 (0.2591) acc1: 99.4792 (90.6303) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1504 data: 0.0572 max mem: 3539 +Test: [1840/2387] eta: 0:01:42 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1404 (0.2585) acc1: 99.7396 (90.6807) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1417 data: 0.0507 max mem: 3539 +Test: [1850/2387] eta: 0:01:41 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1417 (0.2579) acc1: 100.0000 (90.7300) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1944 data: 0.1038 max mem: 3539 +Test: [1860/2387] eta: 0:01:39 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1415 (0.2573) acc1: 99.7396 (90.7791) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1941 data: 0.1038 max mem: 3539 +Test: [1870/2387] eta: 0:01:36 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1435 (0.2567) acc1: 100.0000 (90.8274) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1420 data: 0.0514 max mem: 3539 +Test: [1880/2387] eta: 0:01:35 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1424 (0.2560) acc1: 99.7396 (90.8755) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1555 data: 0.0643 max mem: 3539 +Test: [1890/2387] eta: 0:01:33 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1410 (0.2554) acc1: 100.0000 (90.9234) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1728 data: 0.0796 max mem: 3539 +Test: [1900/2387] eta: 0:01:31 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1445 (0.2549) acc1: 100.0000 (90.9694) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1678 data: 0.0737 max mem: 3539 +Test: [1910/2387] eta: 0:01:29 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1765 (0.2546) acc1: 99.2188 (91.0090) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1648 data: 0.0708 max mem: 3539 +Test: [1920/2387] eta: 0:01:27 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1893 (0.2546) acc1: 98.1771 (91.0067) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1377 data: 0.0441 max mem: 3539 +Test: [1930/2387] eta: 0:01:25 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1795 (0.2541) acc1: 96.3542 (91.0412) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1369 data: 0.0449 max mem: 3539 +Test: [1940/2387] eta: 0:01:23 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1775 (0.2540) acc1: 97.6562 (91.0405) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1568 data: 0.0627 max mem: 3539 +Test: [1950/2387] eta: 0:01:21 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1814 (0.2538) acc1: 96.6146 (91.0595) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1569 data: 0.0602 max mem: 3539 +Test: [1960/2387] eta: 0:01:19 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1888 (0.2538) acc1: 95.8333 (91.0604) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1623 data: 0.0652 max mem: 3539 +Test: [1970/2387] eta: 0:01:17 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2765 (0.2545) acc1: 88.8021 (91.0134) acc5: 100.0000 (100.0000) acc1_neg: 75.521 acc5_neg: 100.000 time: 0.1649 data: 0.0667 max mem: 3539 +Test: [1980/2387] eta: 0:01:15 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2590 (0.2545) acc1: 90.3646 (91.0265) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1675 data: 0.0719 max mem: 3539 +Test: [1990/2387] eta: 0:01:13 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2160 (0.2544) acc1: 97.6562 (91.0591) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1649 data: 0.0695 max mem: 3539 +Test: [2000/2387] eta: 0:01:11 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2165 (0.2542) acc1: 97.9167 (91.0949) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1699 data: 0.0738 max mem: 3539 +Test: [2010/2387] eta: 0:01:10 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2165 (0.2540) acc1: 97.9167 (91.1280) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1666 data: 0.0728 max mem: 3539 +Test: [2020/2387] eta: 0:01:08 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2159 (0.2538) acc1: 97.9167 (91.1609) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1603 data: 0.0677 max mem: 3539 +Test: [2030/2387] eta: 0:01:06 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2157 (0.2536) acc1: 97.9167 (91.1942) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1832 data: 0.0920 max mem: 3540 +Test: [2040/2387] eta: 0:01:04 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2138 (0.2534) acc1: 97.9167 (91.2284) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2095 data: 0.1169 max mem: 3540 +Test: [2050/2387] eta: 0:01:02 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2125 (0.2532) acc1: 98.1771 (91.2625) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2049 data: 0.1094 max mem: 3540 +Test: [2060/2387] eta: 0:01:00 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1971 (0.2529) acc1: 98.1771 (91.2963) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1921 data: 0.0986 max mem: 3540 +Test: [2070/2387] eta: 0:00:58 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1937 (0.2526) acc1: 98.4375 (91.3314) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1927 data: 0.1022 max mem: 3540 +Test: [2080/2387] eta: 0:00:57 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1923 (0.2523) acc1: 98.4375 (91.3655) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1958 data: 0.1033 max mem: 3540 +Test: [2090/2387] eta: 0:00:55 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1894 (0.2520) acc1: 98.6979 (91.4001) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2024 data: 0.1087 max mem: 3540 +Test: [2100/2387] eta: 0:00:53 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1885 (0.2517) acc1: 98.6979 (91.4353) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1985 data: 0.1064 max mem: 3540 +Test: [2110/2387] eta: 0:00:51 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1865 (0.2514) acc1: 98.6979 (91.4685) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1960 data: 0.1050 max mem: 3540 +Test: [2120/2387] eta: 0:00:49 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1887 (0.2512) acc1: 98.1771 (91.4864) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1941 data: 0.1031 max mem: 3540 +Test: [2130/2387] eta: 0:00:47 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1357 (0.2506) acc1: 98.4375 (91.5210) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1995 data: 0.1084 max mem: 3540 +Test: [2140/2387] eta: 0:00:46 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1320 (0.2503) acc1: 97.9167 (91.5248) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2090 data: 0.1164 max mem: 3540 +Test: [2150/2387] eta: 0:00:44 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1523 (0.2499) acc1: 97.3958 (91.5556) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1974 data: 0.1048 max mem: 3540 +Test: [2160/2387] eta: 0:00:42 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1519 (0.2495) acc1: 97.3958 (91.5797) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1754 data: 0.0824 max mem: 3540 +Test: [2170/2387] eta: 0:00:40 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1978 (0.2495) acc1: 95.3125 (91.5851) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1714 data: 0.0785 max mem: 3540 +Test: [2180/2387] eta: 0:00:38 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2476 (0.2503) acc1: 91.1458 (91.5290) acc5: 100.0000 (100.0000) acc1_neg: 58.854 acc5_neg: 100.000 time: 0.1697 data: 0.0790 max mem: 3540 +Test: [2190/2387] eta: 0:00:36 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2583 (0.2506) acc1: 91.1458 (91.5180) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1667 data: 0.0757 max mem: 3540 +Test: [2200/2387] eta: 0:00:34 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1746 (0.2502) acc1: 96.6146 (91.5488) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1755 data: 0.0842 max mem: 3540 +Test: [2210/2387] eta: 0:00:32 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1608 (0.2498) acc1: 98.4375 (91.5702) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1638 data: 0.0724 max mem: 3540 +Test: [2220/2387] eta: 0:00:31 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1768 (0.2495) acc1: 97.1354 (91.5916) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1805 data: 0.0886 max mem: 3540 +Test: [2230/2387] eta: 0:00:29 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1657 (0.2491) acc1: 97.6562 (91.6188) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2098 data: 0.1181 max mem: 3540 +Test: [2240/2387] eta: 0:00:27 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1477 (0.2487) acc1: 98.1771 (91.6409) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1994 data: 0.1087 max mem: 3540 +Test: [2250/2387] eta: 0:00:25 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1517 (0.2483) acc1: 98.4375 (91.6708) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1919 data: 0.0972 max mem: 3540 +Test: [2260/2387] eta: 0:00:23 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1612 (0.2479) acc1: 98.9583 (91.7005) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1960 data: 0.0997 max mem: 3541 +Test: [2270/2387] eta: 0:00:21 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1315 (0.2473) acc1: 99.2188 (91.7351) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1945 data: 0.1013 max mem: 3541 +Test: [2280/2387] eta: 0:00:19 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1137 (0.2470) acc1: 99.7396 (91.7466) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1932 data: 0.1005 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1137 (0.2463) acc1: 99.7396 (91.7812) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1989 data: 0.1075 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1204 (0.2459) acc1: 99.4792 (91.8094) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1995 data: 0.1082 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1381 (0.2455) acc1: 98.6979 (91.8354) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2044 data: 0.1130 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.1769 (0.2455) acc1: 96.6146 (91.8405) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1848 data: 0.0922 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2268 (0.2456) acc1: 94.0104 (91.8367) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1539 data: 0.0597 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2306 (0.2456) acc1: 93.4896 (91.8382) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1708 data: 0.0774 max mem: 3541 +Test: [2350/2387] eta: 0:00:06 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2186 (0.2459) acc1: 92.4479 (91.8147) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1926 data: 0.0998 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2679 (0.2460) acc1: 90.8854 (91.8109) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1891 data: 0.0964 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.2874 (0.2464) acc1: 89.0625 (91.7830) acc5: 100.0000 (100.0000) acc1_neg: 69.792 acc5_neg: 100.000 time: 0.1796 data: 0.0870 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.3292 (0.2469) acc1: 85.4167 (91.7580) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1706 data: 0.0781 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 65.753 acc5_pos: 100.000 loss: 0.4124 (0.2475) acc1: 82.0312 (91.7230) acc5: 100.0000 (100.0000) acc1_neg: 60.465 acc5_neg: 100.000 time: 0.1580 data: 0.0693 max mem: 3541 +Test: Total time: 0:07:24 (0.1863 s / it) +* Overall Acc@1 91.723 Acc@5 100.000 + target == real Acc@1 52.394 Acc@5 100.000 + target == fake Acc@1 94.485 Acc@5 100.000 + loss 0.248 + PR-AUC 0.9938 ROC-AUC 0.9184 EER 15.36% (thr=0.7070) +Accuracy of the network on the 916267 test images: 91.7% +[rank0]:[W725 16:24:32.536967662 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 16:24:50.924000 81287 torch/distributed/run.py:766] +W0725 16:24:50.924000 81287 torch/distributed/run.py:766] ***************************************** +W0725 16:24:50.924000 81287 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 16:24:50.924000 81287 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 16:25:17.185553036 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 16:25:17.927372502 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 16:25:18.143399671 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 16:25:18.234976371 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank6]:[W725 16:25:18.235195874 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +[rank7]:[W725 16:25:18.260302475 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 16:25:18.261955770 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 16:25:18.267727280 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n22_seed1_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed1_resampled_n116-041-188_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 0:58:19 acc1_pos: 57.552 acc5_pos: 100.000 loss: 0.6550 (0.6550) acc1: 57.5521 (57.5521) acc5: 100.0000 (100.0000) time: 6.7692 data: 4.8216 max mem: 3531 +Test: [ 10/517] eta: 0:05:53 acc1_pos: 41.927 acc5_pos: 100.000 loss: 0.8264 (0.7901) acc1: 48.9583 (50.2841) acc5: 100.0000 (100.0000) time: 0.6971 data: 0.4386 max mem: 3531 +Test: [ 20/517] eta: 0:03:23 acc1_pos: 40.885 acc5_pos: 100.000 loss: 0.7712 (0.7752) acc1: 47.3958 (51.6989) acc5: 100.0000 (100.0000) time: 0.0913 data: 0.0003 max mem: 3531 +Test: [ 30/517] eta: 0:02:42 acc1_pos: 26.042 acc5_pos: 100.000 loss: 0.7820 (0.7972) acc1: 47.3958 (49.5548) acc5: 100.0000 (100.0000) time: 0.1326 data: 0.0407 max mem: 3531 +Test: [ 40/517] eta: 0:02:33 acc1_pos: 40.104 acc5_pos: 100.000 loss: 0.8183 (0.8002) acc1: 45.8333 (48.5582) acc5: 100.0000 (100.0000) time: 0.2311 data: 0.1403 max mem: 3531 +Test: [ 50/517] eta: 0:02:29 acc1_pos: 79.817 acc5_pos: 100.000 loss: 0.7265 (0.7834) acc1: 50.0000 (50.5515) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.2987 data: 0.2062 max mem: 3531 +Test: [ 60/517] eta: 0:02:26 acc1_pos: 79.817 acc5_pos: 100.000 loss: 0.6873 (0.7599) acc1: 62.5000 (53.3000) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.3137 data: 0.2177 max mem: 3531 +Test: [ 70/517] eta: 0:02:17 acc1_pos: 79.817 acc5_pos: 100.000 loss: 0.6246 (0.7317) acc1: 65.3646 (55.6888) acc5: 100.0000 (100.0000) acc1_neg: 68.750 acc5_neg: 100.000 time: 0.2806 data: 0.1841 max mem: 3531 +Test: [ 80/517] eta: 0:02:11 acc1_pos: 79.817 acc5_pos: 100.000 loss: 0.5800 (0.7187) acc1: 67.7083 (56.7033) acc5: 100.0000 (100.0000) acc1_neg: 66.406 acc5_neg: 100.000 time: 0.2399 data: 0.1446 max mem: 3531 +Test: [ 90/517] eta: 0:02:05 acc1_pos: 79.817 acc5_pos: 100.000 loss: 0.5559 (0.6949) acc1: 70.8333 (59.0516) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.2392 data: 0.1423 max mem: 3531 +Test: [100/517] eta: 0:02:00 acc1_pos: 79.817 acc5_pos: 100.000 loss: 0.4264 (0.6667) acc1: 83.8542 (61.8760) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.2449 data: 0.1492 max mem: 3531 +Test: [110/517] eta: 0:01:57 acc1_pos: 79.817 acc5_pos: 100.000 loss: 0.3919 (0.6389) acc1: 87.7604 (64.4449) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2699 data: 0.1739 max mem: 3531 +Test: [120/517] eta: 0:01:52 acc1_pos: 79.817 acc5_pos: 100.000 loss: 0.4318 (0.6264) acc1: 82.5521 (65.5454) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.2607 data: 0.1630 max mem: 3531 +Test: [130/517] eta: 0:01:48 acc1_pos: 79.817 acc5_pos: 100.000 loss: 0.4708 (0.6110) acc1: 79.1667 (67.0106) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.2399 data: 0.1437 max mem: 3531 +Test: [140/517] eta: 0:01:44 acc1_pos: 79.817 acc5_pos: 100.000 loss: 0.4093 (0.5961) acc1: 86.4583 (68.5487) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.2402 data: 0.1459 max mem: 3531 +Test: [150/517] eta: 0:01:41 acc1_pos: 79.817 acc5_pos: 100.000 loss: 0.4001 (0.5850) acc1: 88.5417 (69.6209) acc5: 100.0000 (100.0000) acc1_neg: 61.458 acc5_neg: 100.000 time: 0.2502 data: 0.1572 max mem: 3531 +Test: [160/517] eta: 0:01:38 acc1_pos: 79.817 acc5_pos: 100.000 loss: 0.4356 (0.5764) acc1: 81.7708 (70.1572) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.2641 data: 0.1704 max mem: 3531 +Test: [170/517] eta: 0:01:35 acc1_pos: 79.817 acc5_pos: 100.000 loss: 0.3960 (0.5648) acc1: 81.7708 (71.0755) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.2496 data: 0.1544 max mem: 3531 +Test: [180/517] eta: 0:01:31 acc1_pos: 79.817 acc5_pos: 100.000 loss: 0.3949 (0.5559) acc1: 84.1146 (71.8203) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.2395 data: 0.1427 max mem: 3531 +Test: [190/517] eta: 0:01:28 acc1_pos: 79.817 acc5_pos: 100.000 loss: 0.3835 (0.5471) acc1: 89.0625 (72.7626) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.2449 data: 0.1481 max mem: 3531 +Test: [200/517] eta: 0:01:25 acc1_pos: 79.817 acc5_pos: 100.000 loss: 0.3815 (0.5380) acc1: 91.1458 (73.7290) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.2399 data: 0.1427 max mem: 3531 +Test: [210/517] eta: 0:01:22 acc1_pos: 79.817 acc5_pos: 100.000 loss: 0.4356 (0.5454) acc1: 76.3021 (72.9623) acc5: 100.0000 (100.0000) acc1_neg: 67.708 acc5_neg: 100.000 time: 0.2399 data: 0.1423 max mem: 3532 +Test: [220/517] eta: 0:01:19 acc1_pos: 79.817 acc5_pos: 100.000 loss: 0.6586 (0.5485) acc1: 60.6771 (72.6044) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.2407 data: 0.1441 max mem: 3532 +Test: [230/517] eta: 0:01:16 acc1_pos: 79.817 acc5_pos: 100.000 loss: 0.5374 (0.5482) acc1: 70.3125 (72.6957) acc5: 100.0000 (100.0000) acc1_neg: 64.323 acc5_neg: 100.000 time: 0.2399 data: 0.1438 max mem: 3532 +Test: [240/517] eta: 0:01:13 acc1_pos: 79.817 acc5_pos: 100.000 loss: 0.5909 (0.5510) acc1: 65.6250 (72.4056) acc5: 100.0000 (100.0000) acc1_neg: 52.865 acc5_neg: 100.000 time: 0.2345 data: 0.1384 max mem: 3532 +Test: [250/517] eta: 0:01:10 acc1_pos: 79.817 acc5_pos: 100.000 loss: 0.5331 (0.5472) acc1: 73.6979 (72.8918) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2306 data: 0.1355 max mem: 3532 +Test: [260/517] eta: 0:01:07 acc1_pos: 79.817 acc5_pos: 100.000 loss: 0.4032 (0.5422) acc1: 89.8438 (73.5093) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.2395 data: 0.1449 max mem: 3532 +Test: [270/517] eta: 0:01:04 acc1_pos: 79.817 acc5_pos: 100.000 loss: 0.5205 (0.5441) acc1: 76.3021 (73.3289) acc5: 100.0000 (100.0000) acc1_neg: 76.042 acc5_neg: 100.000 time: 0.2542 data: 0.1582 max mem: 3532 +Test: [280/517] eta: 0:01:01 acc1_pos: 79.817 acc5_pos: 100.000 loss: 0.5497 (0.5439) acc1: 70.8333 (73.3217) acc5: 100.0000 (100.0000) acc1_neg: 74.740 acc5_neg: 100.000 time: 0.2409 data: 0.1451 max mem: 3532 +Test: [290/517] eta: 0:00:58 acc1_pos: 79.817 acc5_pos: 100.000 loss: 0.4945 (0.5411) acc1: 79.4271 (73.6979) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.2149 data: 0.1204 max mem: 3532 +Test: [300/517] eta: 0:00:55 acc1_pos: 79.817 acc5_pos: 100.000 loss: 0.4289 (0.5371) acc1: 85.6771 (74.2023) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.2239 data: 0.1286 max mem: 3532 +Test: [310/517] eta: 0:00:53 acc1_pos: 79.817 acc5_pos: 100.000 loss: 0.4148 (0.5341) acc1: 89.0625 (74.5780) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.2449 data: 0.1488 max mem: 3532 +Test: [320/517] eta: 0:00:50 acc1_pos: 79.817 acc5_pos: 100.000 loss: 0.4026 (0.5287) acc1: 89.5833 (75.1322) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2399 data: 0.1429 max mem: 3532 +Test: [330/517] eta: 0:00:47 acc1_pos: 79.817 acc5_pos: 100.000 loss: 0.3176 (0.5219) acc1: 95.0521 (75.7639) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2299 data: 0.1316 max mem: 3532 +Test: [340/517] eta: 0:00:45 acc1_pos: 79.817 acc5_pos: 100.000 loss: 0.3025 (0.5196) acc1: 95.3125 (75.9164) acc5: 100.0000 (100.0000) acc1_neg: 67.969 acc5_neg: 100.000 time: 0.2399 data: 0.1426 max mem: 3532 +Test: [350/517] eta: 0:00:42 acc1_pos: 79.817 acc5_pos: 100.000 loss: 0.6037 (0.5240) acc1: 63.2812 (75.4170) acc5: 100.0000 (100.0000) acc1_neg: 41.146 acc5_neg: 100.000 time: 0.2405 data: 0.1453 max mem: 3532 +Test: [360/517] eta: 0:00:40 acc1_pos: 79.817 acc5_pos: 100.000 loss: 0.6108 (0.5250) acc1: 63.2812 (75.3160) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.2399 data: 0.1456 max mem: 3532 +Test: [370/517] eta: 0:00:37 acc1_pos: 79.817 acc5_pos: 100.000 loss: 0.4109 (0.5204) acc1: 86.1979 (75.8121) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2345 data: 0.1422 max mem: 3532 +Test: [380/517] eta: 0:00:34 acc1_pos: 79.817 acc5_pos: 100.000 loss: 0.3517 (0.5168) acc1: 91.9271 (76.1421) acc5: 100.0000 (100.0000) acc1_neg: 78.646 acc5_neg: 100.000 time: 0.2349 data: 0.1431 max mem: 3532 +Test: [390/517] eta: 0:00:32 acc1_pos: 79.817 acc5_pos: 100.000 loss: 0.4013 (0.5143) acc1: 86.4583 (76.3381) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.2447 data: 0.1489 max mem: 3532 +Test: [400/517] eta: 0:00:29 acc1_pos: 79.817 acc5_pos: 100.000 loss: 0.4171 (0.5120) acc1: 84.3750 (76.5255) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.2449 data: 0.1462 max mem: 3532 +Test: [410/517] eta: 0:00:27 acc1_pos: 79.817 acc5_pos: 100.000 loss: 0.4282 (0.5107) acc1: 83.5938 (76.7228) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.2399 data: 0.1451 max mem: 3532 +Test: [420/517] eta: 0:00:24 acc1_pos: 79.817 acc5_pos: 100.000 loss: 0.4718 (0.5103) acc1: 85.1562 (76.8483) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.2500 data: 0.1590 max mem: 3532 +Test: [430/517] eta: 0:00:22 acc1_pos: 79.817 acc5_pos: 100.000 loss: 0.4921 (0.5111) acc1: 75.7812 (76.7528) acc5: 100.0000 (100.0000) acc1_neg: 64.583 acc5_neg: 100.000 time: 0.2549 data: 0.1616 max mem: 3532 +Test: [440/517] eta: 0:00:19 acc1_pos: 79.817 acc5_pos: 100.000 loss: 0.5251 (0.5115) acc1: 73.4375 (76.7757) acc5: 100.0000 (100.0000) acc1_neg: 80.729 acc5_neg: 100.000 time: 0.2398 data: 0.1443 max mem: 3533 +Test: [450/517] eta: 0:00:16 acc1_pos: 79.817 acc5_pos: 100.000 loss: 0.5049 (0.5114) acc1: 77.0833 (76.8062) acc5: 100.0000 (100.0000) acc1_neg: 77.083 acc5_neg: 100.000 time: 0.2349 data: 0.1403 max mem: 3533 +Test: [460/517] eta: 0:00:14 acc1_pos: 79.817 acc5_pos: 100.000 loss: 0.5049 (0.5107) acc1: 77.8646 (76.8873) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.2501 data: 0.1577 max mem: 3533 +Test: [470/517] eta: 0:00:11 acc1_pos: 79.817 acc5_pos: 100.000 loss: 0.4355 (0.5088) acc1: 85.9375 (77.1082) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.2499 data: 0.1553 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 79.817 acc5_pos: 100.000 loss: 0.4333 (0.5074) acc1: 86.1979 (77.2116) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.2447 data: 0.1473 max mem: 3533 +Test: [490/517] eta: 0:00:06 acc1_pos: 79.817 acc5_pos: 100.000 loss: 0.4173 (0.5061) acc1: 85.1562 (77.3363) acc5: 100.0000 (100.0000) acc1_neg: 60.938 acc5_neg: 100.000 time: 0.2499 data: 0.1516 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 79.817 acc5_pos: 100.000 loss: 0.6295 (0.5107) acc1: 61.7188 (76.7943) acc5: 100.0000 (100.0000) acc1_neg: 52.083 acc5_neg: 100.000 time: 0.2499 data: 0.1518 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 79.817 acc5_pos: 100.000 loss: 0.6311 (0.5103) acc1: 52.0833 (76.7872) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.2243 data: 0.1308 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 79.817 acc5_pos: 100.000 loss: 0.5342 (0.5102) acc1: 74.7396 (76.8157) acc5: 100.0000 (100.0000) acc1_neg: 84.099 acc5_neg: 100.000 time: 0.1967 data: 0.0977 max mem: 3533 +Test: Total time: 0:02:10 (0.2516 s / it) +* Overall Acc@1 76.816 Acc@5 100.000 + target == real Acc@1 49.979 Acc@5 100.000 + target == fake Acc@1 79.600 Acc@5 100.000 + loss 0.510 + PR-AUC 0.9525 ROC-AUC 0.7091 EER 35.14% (thr=0.5923) +Accuracy of the network on the 198427 test images: 76.8% +[rank0]:[W725 16:27:42.251775130 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0725 16:28:01.187000 81856 torch/distributed/run.py:766] +W0725 16:28:01.187000 81856 torch/distributed/run.py:766] ***************************************** +W0725 16:28:01.187000 81856 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 16:28:01.187000 81856 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 16:28:30.301440868 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 16:28:30.917417088 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 16:28:31.059845380 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +[rank0]:[W725 16:28:31.088982020 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 16:28:31.093265647 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 16:28:31.110672931 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 16:28:31.314636359 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 16:28:31.381733093 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n22_seed1_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed1_resampled_n116-041-188_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:42:05 acc1_pos: 79.948 acc5_pos: 100.000 loss: 0.5402 (0.5402) acc1: 79.9479 (79.9479) acc5: 100.0000 (100.0000) time: 6.4931 data: 4.1394 max mem: 3531 +Test: [ 10/389] eta: 0:04:14 acc1_pos: 98.177 acc5_pos: 100.000 loss: 0.4191 (0.4824) acc1: 84.6354 (82.2917) acc5: 100.0000 (100.0000) time: 0.6721 data: 0.3766 max mem: 3531 +Test: [ 20/389] eta: 0:02:26 acc1_pos: 16.667 acc5_pos: 100.000 loss: 0.4893 (0.5986) acc1: 79.4271 (67.6215) acc5: 100.0000 (100.0000) time: 0.0931 data: 0.0004 max mem: 3531 +Test: [ 30/389] eta: 0:02:10 acc1_pos: 53.646 acc5_pos: 100.000 loss: 0.7086 (0.6287) acc1: 47.3958 (62.4076) acc5: 100.0000 (100.0000) time: 0.1930 data: 0.0982 max mem: 3531 +Test: [ 40/389] eta: 0:02:00 acc1_pos: 40.885 acc5_pos: 100.000 loss: 0.7242 (0.6574) acc1: 46.6146 (58.7525) acc5: 100.0000 (100.0000) time: 0.2901 data: 0.1944 max mem: 3531 +Test: [ 50/389] eta: 0:01:49 acc1_pos: 39.583 acc5_pos: 100.000 loss: 0.7242 (0.6507) acc1: 46.6146 (59.7937) acc5: 100.0000 (100.0000) time: 0.2611 data: 0.1671 max mem: 3531 +Test: [ 60/389] eta: 0:01:37 acc1_pos: 78.646 acc5_pos: 100.000 loss: 0.5969 (0.6568) acc1: 60.6771 (59.4049) acc5: 100.0000 (100.0000) time: 0.1999 data: 0.1094 max mem: 3531 +Test: [ 70/389] eta: 0:01:30 acc1_pos: 75.260 acc5_pos: 100.000 loss: 0.5287 (0.6349) acc1: 78.3854 (62.5330) acc5: 100.0000 (100.0000) time: 0.1887 data: 0.0965 max mem: 3531 +Test: [ 80/389] eta: 0:01:24 acc1_pos: 93.750 acc5_pos: 100.000 loss: 0.5323 (0.6410) acc1: 75.2604 (61.7188) acc5: 100.0000 (100.0000) time: 0.1951 data: 0.1030 max mem: 3531 +Test: [ 90/389] eta: 0:01:19 acc1_pos: 73.698 acc5_pos: 100.000 loss: 0.4261 (0.6102) acc1: 73.6979 (64.7035) acc5: 100.0000 (100.0000) time: 0.1958 data: 0.1043 max mem: 3531 +Test: [100/389] eta: 0:01:15 acc1_pos: 75.000 acc5_pos: 100.000 loss: 0.5083 (0.6154) acc1: 77.0833 (64.6710) acc5: 100.0000 (100.0000) time: 0.2195 data: 0.1247 max mem: 3531 +Test: [110/389] eta: 0:01:11 acc1_pos: 93.490 acc5_pos: 100.000 loss: 0.6087 (0.6216) acc1: 68.4896 (64.0367) acc5: 100.0000 (100.0000) time: 0.2144 data: 0.1207 max mem: 3531 +Test: [120/389] eta: 0:01:07 acc1_pos: 83.854 acc5_pos: 100.000 loss: 0.5328 (0.6144) acc1: 70.5729 (64.8179) acc5: 100.0000 (100.0000) time: 0.1949 data: 0.1030 max mem: 3531 +Test: [130/389] eta: 0:01:04 acc1_pos: 18.405 acc5_pos: 100.000 loss: 0.5354 (0.6210) acc1: 67.1875 (64.1798) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.2044 data: 0.1117 max mem: 3531 +Test: [140/389] eta: 0:01:00 acc1_pos: 18.405 acc5_pos: 100.000 loss: 0.5923 (0.6220) acc1: 67.1875 (64.4042) acc5: 100.0000 (100.0000) acc1_neg: 14.062 acc5_neg: 100.000 time: 0.1966 data: 0.1032 max mem: 3531 +Test: [150/389] eta: 0:00:58 acc1_pos: 18.405 acc5_pos: 100.000 loss: 0.6419 (0.6339) acc1: 66.1458 (63.3037) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2177 data: 0.1249 max mem: 3531 +Test: [160/389] eta: 0:00:55 acc1_pos: 18.405 acc5_pos: 100.000 loss: 0.6514 (0.6281) acc1: 50.2604 (63.9363) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2234 data: 0.1313 max mem: 3531 +Test: [170/389] eta: 0:00:51 acc1_pos: 18.405 acc5_pos: 100.000 loss: 0.4214 (0.6262) acc1: 89.8438 (64.2315) acc5: 100.0000 (100.0000) acc1_neg: 38.281 acc5_neg: 100.000 time: 0.1587 data: 0.0648 max mem: 3531 +Test: [180/389] eta: 0:00:48 acc1_pos: 18.405 acc5_pos: 100.000 loss: 0.4621 (0.6274) acc1: 84.1146 (64.2913) acc5: 100.0000 (100.0000) acc1_neg: 57.552 acc5_neg: 100.000 time: 0.1697 data: 0.0710 max mem: 3531 +Test: [190/389] eta: 0:00:45 acc1_pos: 18.405 acc5_pos: 100.000 loss: 0.6156 (0.6380) acc1: 57.5521 (63.3399) acc5: 100.0000 (100.0000) acc1_neg: 29.427 acc5_neg: 100.000 time: 0.2033 data: 0.1032 max mem: 3531 +Test: [200/389] eta: 0:00:43 acc1_pos: 18.405 acc5_pos: 100.000 loss: 0.6767 (0.6389) acc1: 55.2083 (63.3162) acc5: 100.0000 (100.0000) acc1_neg: 55.208 acc5_neg: 100.000 time: 0.1799 data: 0.0845 max mem: 3531 +Test: [210/389] eta: 0:00:40 acc1_pos: 18.405 acc5_pos: 100.000 loss: 0.5547 (0.6316) acc1: 73.1771 (64.1008) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1762 data: 0.0837 max mem: 3532 +Test: [220/389] eta: 0:00:37 acc1_pos: 18.405 acc5_pos: 100.000 loss: 0.3387 (0.6206) acc1: 95.0521 (65.3328) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.1875 data: 0.0955 max mem: 3532 +Test: [230/389] eta: 0:00:35 acc1_pos: 18.405 acc5_pos: 100.000 loss: 0.5064 (0.6217) acc1: 79.6875 (65.1887) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.1751 data: 0.0801 max mem: 3532 +Test: [240/389] eta: 0:00:32 acc1_pos: 18.405 acc5_pos: 100.000 loss: 0.5885 (0.6219) acc1: 66.9271 (65.2155) acc5: 100.0000 (100.0000) acc1_neg: 55.208 acc5_neg: 100.000 time: 0.1823 data: 0.0869 max mem: 3532 +Test: [250/389] eta: 0:00:30 acc1_pos: 18.405 acc5_pos: 100.000 loss: 0.4846 (0.6171) acc1: 83.0729 (65.8605) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1685 data: 0.0752 max mem: 3532 +Test: [260/389] eta: 0:00:27 acc1_pos: 18.405 acc5_pos: 100.000 loss: 0.5247 (0.6166) acc1: 84.8958 (66.1708) acc5: 100.0000 (100.0000) acc1_neg: 21.094 acc5_neg: 100.000 time: 0.1649 data: 0.0712 max mem: 3532 +Test: [270/389] eta: 0:00:25 acc1_pos: 18.405 acc5_pos: 100.000 loss: 0.5933 (0.6273) acc1: 57.0312 (64.9485) acc5: 100.0000 (100.0000) acc1_neg: 71.615 acc5_neg: 100.000 time: 0.1703 data: 0.0768 max mem: 3532 +Test: [280/389] eta: 0:00:23 acc1_pos: 18.405 acc5_pos: 100.000 loss: 0.7605 (0.6311) acc1: 46.6146 (64.6844) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1551 data: 0.0608 max mem: 3532 +Test: [290/389] eta: 0:00:20 acc1_pos: 18.405 acc5_pos: 100.000 loss: 0.6054 (0.6305) acc1: 68.2292 (64.8115) acc5: 100.0000 (100.0000) acc1_neg: 73.958 acc5_neg: 100.000 time: 0.1493 data: 0.0548 max mem: 3532 +Test: [300/389] eta: 0:00:18 acc1_pos: 18.405 acc5_pos: 100.000 loss: 0.5937 (0.6304) acc1: 72.3958 (64.8810) acc5: 100.0000 (100.0000) acc1_neg: 64.062 acc5_neg: 100.000 time: 0.1501 data: 0.0579 max mem: 3532 +Test: [310/389] eta: 0:00:16 acc1_pos: 18.405 acc5_pos: 100.000 loss: 0.6662 (0.6380) acc1: 59.3750 (64.0851) acc5: 100.0000 (100.0000) acc1_neg: 9.896 acc5_neg: 100.000 time: 0.1599 data: 0.0690 max mem: 3532 +Test: [320/389] eta: 0:00:14 acc1_pos: 18.405 acc5_pos: 100.000 loss: 0.7482 (0.6382) acc1: 50.5208 (64.0560) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.1618 data: 0.0690 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 18.405 acc5_pos: 100.000 loss: 0.8203 (0.6466) acc1: 42.9688 (62.8328) acc5: 100.0000 (100.0000) acc1_neg: 42.969 acc5_neg: 100.000 time: 0.1552 data: 0.0621 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 18.405 acc5_pos: 100.000 loss: 0.6588 (0.6385) acc1: 59.3750 (63.7318) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.1455 data: 0.0546 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 18.405 acc5_pos: 100.000 loss: 0.3901 (0.6442) acc1: 88.5417 (63.3117) acc5: 100.0000 (100.0000) acc1_neg: 22.917 acc5_neg: 100.000 time: 0.1488 data: 0.0585 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 18.405 acc5_pos: 100.000 loss: 0.5524 (0.6445) acc1: 65.6250 (63.2041) acc5: 100.0000 (100.0000) acc1_neg: 69.792 acc5_neg: 100.000 time: 0.1384 data: 0.0480 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 18.405 acc5_pos: 100.000 loss: 0.5459 (0.6419) acc1: 69.7917 (63.4862) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1273 data: 0.0371 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 18.405 acc5_pos: 100.000 loss: 0.5276 (0.6407) acc1: 61.7188 (63.4781) acc5: 100.0000 (100.0000) acc1_neg: 61.719 acc5_neg: 100.000 time: 0.1240 data: 0.0340 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 18.405 acc5_pos: 100.000 loss: 0.4367 (0.6378) acc1: 82.2917 (63.8379) acc5: 100.0000 (100.0000) acc1_neg: 75.137 acc5_neg: 100.000 time: 0.1232 data: 0.0301 max mem: 3532 +Test: Total time: 0:01:15 (0.1931 s / it) +* Overall Acc@1 63.838 Acc@5 100.000 + target == real Acc@1 64.313 Acc@5 100.000 + target == fake Acc@1 63.617 Acc@5 100.000 + loss 0.638 + PR-AUC 0.8192 ROC-AUC 0.6968 EER 36.02% (thr=0.4978) +Accuracy of the network on the 149358 test images: 63.8% +[rank0]:[W725 16:29:58.512277451 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 16:30:17.632000 82414 torch/distributed/run.py:766] +W0725 16:30:17.632000 82414 torch/distributed/run.py:766] ***************************************** +W0725 16:30:17.632000 82414 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 16:30:17.632000 82414 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 16:30:47.269333155 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 16:30:47.700814806 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 16:30:47.932420090 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 16:30:48.244680669 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 16:30:48.341434477 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 16:30:48.351928894 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 16:30:48.572755293 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 16:30:48.618940768 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n22_seed1_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed1_resampled_n116-041-188_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:31:12 acc1_pos: 91.927 acc5_pos: 100.000 loss: 0.2534 (0.2534) acc1: 91.9271 (91.9271) acc5: 100.0000 (100.0000) time: 5.6245 data: 4.0346 max mem: 3531 +Test: [ 10/333] eta: 0:03:11 acc1_pos: 95.833 acc5_pos: 100.000 loss: 0.2098 (0.2104) acc1: 95.8333 (94.5549) acc5: 100.0000 (100.0000) time: 0.5932 data: 0.3672 max mem: 3531 +Test: [ 20/333] eta: 0:01:50 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.2098 (0.2145) acc1: 94.5312 (94.5437) acc5: 100.0000 (100.0000) time: 0.0901 data: 0.0004 max mem: 3531 +Test: [ 30/333] eta: 0:01:31 acc1_pos: 78.906 acc5_pos: 100.000 loss: 0.2896 (0.3112) acc1: 88.8021 (87.9872) acc5: 100.0000 (100.0000) time: 0.1400 data: 0.0500 max mem: 3531 +Test: [ 40/333] eta: 0:01:24 acc1_pos: 74.219 acc5_pos: 100.000 loss: 0.5392 (0.3711) acc1: 72.9167 (84.3242) acc5: 100.0000 (100.0000) time: 0.2194 data: 0.1281 max mem: 3531 +Test: [ 50/333] eta: 0:01:17 acc1_pos: 74.479 acc5_pos: 100.000 loss: 0.5392 (0.4042) acc1: 74.2188 (82.1078) acc5: 100.0000 (100.0000) time: 0.2344 data: 0.1393 max mem: 3531 +Test: [ 60/333] eta: 0:01:13 acc1_pos: 69.531 acc5_pos: 100.000 loss: 0.5298 (0.4291) acc1: 72.6562 (80.2852) acc5: 100.0000 (100.0000) time: 0.2257 data: 0.1292 max mem: 3531 +Test: [ 70/333] eta: 0:01:07 acc1_pos: 63.802 acc5_pos: 100.000 loss: 0.5573 (0.4476) acc1: 69.5312 (79.1006) acc5: 100.0000 (100.0000) time: 0.2150 data: 0.1206 max mem: 3531 +Test: [ 80/333] eta: 0:01:03 acc1_pos: 72.135 acc5_pos: 100.000 loss: 0.5494 (0.4581) acc1: 70.0521 (78.2825) acc5: 100.0000 (100.0000) time: 0.1966 data: 0.1040 max mem: 3531 +Test: [ 90/333] eta: 0:00:59 acc1_pos: 79.688 acc5_pos: 100.000 loss: 0.4924 (0.4594) acc1: 78.1250 (78.4856) acc5: 100.0000 (100.0000) time: 0.1961 data: 0.1029 max mem: 3531 +Test: [100/333] eta: 0:00:56 acc1_pos: 86.198 acc5_pos: 100.000 loss: 0.4778 (0.4644) acc1: 78.3854 (78.0399) acc5: 100.0000 (100.0000) time: 0.2047 data: 0.1118 max mem: 3531 +Test: [110/333] eta: 0:00:53 acc1_pos: 80.729 acc5_pos: 100.000 loss: 0.4881 (0.4692) acc1: 73.6979 (77.7262) acc5: 100.0000 (100.0000) time: 0.2102 data: 0.1199 max mem: 3531 +Test: [120/333] eta: 0:00:50 acc1_pos: 79.948 acc5_pos: 100.000 loss: 0.4881 (0.4738) acc1: 73.6979 (77.4471) acc5: 100.0000 (100.0000) time: 0.2062 data: 0.1109 max mem: 3531 +Test: [130/333] eta: 0:00:47 acc1_pos: 90.625 acc5_pos: 100.000 loss: 0.4704 (0.4741) acc1: 75.2604 (77.5425) acc5: 100.0000 (100.0000) time: 0.1983 data: 0.0995 max mem: 3531 +Test: [140/333] eta: 0:00:44 acc1_pos: 87.760 acc5_pos: 100.000 loss: 0.4835 (0.4743) acc1: 78.3854 (77.6522) acc5: 100.0000 (100.0000) time: 0.1940 data: 0.0965 max mem: 3531 +Test: [150/333] eta: 0:00:41 acc1_pos: 85.938 acc5_pos: 100.000 loss: 0.4838 (0.4767) acc1: 78.1250 (77.4455) acc5: 100.0000 (100.0000) time: 0.2022 data: 0.1080 max mem: 3531 +Test: [160/333] eta: 0:00:39 acc1_pos: 68.092 acc5_pos: 100.000 loss: 0.5268 (0.4972) acc1: 73.6979 (76.0223) acc5: 100.0000 (100.0000) acc1_neg: 30.990 acc5_neg: 100.000 time: 0.2322 data: 0.1365 max mem: 3531 +Test: [170/333] eta: 0:00:36 acc1_pos: 68.092 acc5_pos: 100.000 loss: 1.0104 (0.5307) acc1: 33.3333 (73.3964) acc5: 100.0000 (100.0000) acc1_neg: 35.938 acc5_neg: 100.000 time: 0.1974 data: 0.1006 max mem: 3531 +Test: [180/333] eta: 0:00:33 acc1_pos: 68.092 acc5_pos: 100.000 loss: 0.9744 (0.5521) acc1: 35.9375 (71.5657) acc5: 100.0000 (100.0000) acc1_neg: 43.750 acc5_neg: 100.000 time: 0.1549 data: 0.0616 max mem: 3531 +Test: [190/333] eta: 0:00:31 acc1_pos: 68.092 acc5_pos: 100.000 loss: 0.8953 (0.5668) acc1: 42.1875 (70.3480) acc5: 100.0000 (100.0000) acc1_neg: 40.104 acc5_neg: 100.000 time: 0.1649 data: 0.0724 max mem: 3531 +Test: [200/333] eta: 0:00:28 acc1_pos: 68.092 acc5_pos: 100.000 loss: 0.8456 (0.5830) acc1: 46.3542 (69.1076) acc5: 100.0000 (100.0000) acc1_neg: 39.062 acc5_neg: 100.000 time: 0.1720 data: 0.0799 max mem: 3531 +Test: [210/333] eta: 0:00:26 acc1_pos: 68.092 acc5_pos: 100.000 loss: 0.9883 (0.6030) acc1: 39.0625 (67.6540) acc5: 100.0000 (100.0000) acc1_neg: 36.719 acc5_neg: 100.000 time: 0.1676 data: 0.0748 max mem: 3532 +Test: [220/333] eta: 0:00:23 acc1_pos: 68.092 acc5_pos: 100.000 loss: 0.9931 (0.6209) acc1: 38.2812 (66.3026) acc5: 100.0000 (100.0000) acc1_neg: 39.062 acc5_neg: 100.000 time: 0.1591 data: 0.0654 max mem: 3532 +Test: [230/333] eta: 0:00:21 acc1_pos: 68.092 acc5_pos: 100.000 loss: 0.8708 (0.6298) acc1: 44.7917 (65.3883) acc5: 100.0000 (100.0000) acc1_neg: 40.885 acc5_neg: 100.000 time: 0.1683 data: 0.0758 max mem: 3532 +Test: [240/333] eta: 0:00:19 acc1_pos: 68.092 acc5_pos: 100.000 loss: 0.8708 (0.6426) acc1: 40.8854 (64.1695) acc5: 100.0000 (100.0000) acc1_neg: 34.115 acc5_neg: 100.000 time: 0.1655 data: 0.0734 max mem: 3532 +Test: [250/333] eta: 0:00:17 acc1_pos: 68.092 acc5_pos: 100.000 loss: 0.9209 (0.6515) acc1: 37.2396 (63.3819) acc5: 100.0000 (100.0000) acc1_neg: 47.917 acc5_neg: 100.000 time: 0.1668 data: 0.0725 max mem: 3532 +Test: [260/333] eta: 0:00:14 acc1_pos: 68.092 acc5_pos: 100.000 loss: 0.8392 (0.6580) acc1: 48.6979 (62.9340) acc5: 100.0000 (100.0000) acc1_neg: 50.521 acc5_neg: 100.000 time: 0.1781 data: 0.0856 max mem: 3532 +Test: [270/333] eta: 0:00:12 acc1_pos: 68.092 acc5_pos: 100.000 loss: 0.8046 (0.6633) acc1: 50.5208 (62.4827) acc5: 100.0000 (100.0000) acc1_neg: 50.781 acc5_neg: 100.000 time: 0.1931 data: 0.1014 max mem: 3532 +Test: [280/333] eta: 0:00:10 acc1_pos: 68.092 acc5_pos: 100.000 loss: 0.8201 (0.6742) acc1: 47.6562 (61.5594) acc5: 100.0000 (100.0000) acc1_neg: 39.844 acc5_neg: 100.000 time: 0.1848 data: 0.0916 max mem: 3532 +Test: [290/333] eta: 0:00:08 acc1_pos: 68.092 acc5_pos: 100.000 loss: 1.0976 (0.6886) acc1: 28.9062 (60.4909) acc5: 100.0000 (100.0000) acc1_neg: 36.458 acc5_neg: 100.000 time: 0.1592 data: 0.0653 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 68.092 acc5_pos: 100.000 loss: 1.0571 (0.6999) acc1: 33.3333 (59.7055) acc5: 100.0000 (100.0000) acc1_neg: 38.021 acc5_neg: 100.000 time: 0.1459 data: 0.0510 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 68.092 acc5_pos: 100.000 loss: 0.9163 (0.7033) acc1: 39.8438 (59.3867) acc5: 100.0000 (100.0000) acc1_neg: 41.406 acc5_neg: 100.000 time: 0.1496 data: 0.0554 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 68.092 acc5_pos: 100.000 loss: 0.8375 (0.7087) acc1: 42.9688 (58.8647) acc5: 100.0000 (100.0000) acc1_neg: 38.281 acc5_neg: 100.000 time: 0.1343 data: 0.0425 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 68.092 acc5_pos: 100.000 loss: 1.0190 (0.7200) acc1: 38.2812 (58.1500) acc5: 100.0000 (100.0000) acc1_neg: 36.198 acc5_neg: 100.000 time: 0.1054 data: 0.0144 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 68.092 acc5_pos: 100.000 loss: 1.0430 (0.7221) acc1: 36.7188 (58.0282) acc5: 100.0000 (100.0000) acc1_neg: 38.028 acc5_neg: 100.000 time: 0.1100 data: 0.0150 max mem: 3532 +Test: Total time: 0:01:04 (0.1932 s / it) +* Overall Acc@1 58.028 Acc@5 100.000 + target == real Acc@1 77.363 Acc@5 100.000 + target == fake Acc@1 40.816 Acc@5 100.000 + loss 0.722 + PR-AUC 0.6628 ROC-AUC 0.6507 EER 39.61% (thr=0.3716) +Accuracy of the network on the 127843 test images: 58.0% +[rank0]:[W725 16:32:05.608703939 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 16:32:25.016000 82970 torch/distributed/run.py:766] +W0725 16:32:25.016000 82970 torch/distributed/run.py:766] ***************************************** +W0725 16:32:25.016000 82970 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 16:32:25.016000 82970 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 16:32:54.309693359 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 16:32:54.567377222 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 16:32:54.926484213 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 16:32:54.945318822 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 16:32:54.953104514 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 16:32:55.146397047 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 16:32:55.167721846 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W725 16:32:55.171172943 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n22_seed1_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed1_resampled_n116-041-188_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:22:18 acc1_pos: 98.177 acc5_pos: 100.000 loss: 0.1523 (0.1523) acc1: 98.1771 (98.1771) acc5: 100.0000 (100.0000) time: 5.6976 data: 4.2060 max mem: 3531 +Test: [ 10/235] eta: 0:02:15 acc1_pos: 96.354 acc5_pos: 100.000 loss: 0.1677 (0.1659) acc1: 97.6562 (97.2538) acc5: 100.0000 (100.0000) time: 0.6020 data: 0.3828 max mem: 3531 +Test: [ 20/235] eta: 0:01:22 acc1_pos: 97.917 acc5_pos: 100.000 loss: 0.1636 (0.1636) acc1: 97.6562 (97.3710) acc5: 100.0000 (100.0000) time: 0.1193 data: 0.0233 max mem: 3531 +Test: [ 30/235] eta: 0:01:09 acc1_pos: 97.135 acc5_pos: 100.000 loss: 0.1619 (0.1631) acc1: 97.6562 (97.5134) acc5: 100.0000 (100.0000) time: 0.1980 data: 0.1013 max mem: 3531 +Test: [ 40/235] eta: 0:01:00 acc1_pos: 97.396 acc5_pos: 100.000 loss: 0.1619 (0.1633) acc1: 97.3958 (97.5229) acc5: 100.0000 (100.0000) time: 0.2268 data: 0.1343 max mem: 3531 +Test: [ 50/235] eta: 0:00:50 acc1_pos: 95.573 acc5_pos: 100.000 loss: 0.1636 (0.1641) acc1: 97.3958 (97.4980) acc5: 100.0000 (100.0000) time: 0.1699 data: 0.0782 max mem: 3531 +Test: [ 60/235] eta: 0:00:42 acc1_pos: 97.656 acc5_pos: 100.000 loss: 0.1698 (0.1653) acc1: 97.1354 (97.4385) acc5: 100.0000 (100.0000) time: 0.1181 data: 0.0252 max mem: 3531 +Test: [ 70/235] eta: 0:00:38 acc1_pos: 98.462 acc5_pos: 100.000 loss: 0.1811 (0.3220) acc1: 96.3542 (87.9438) acc5: 100.0000 (100.0000) acc1_neg: 21.354 acc5_neg: 100.000 time: 0.1299 data: 0.0317 max mem: 3531 +Test: [ 80/235] eta: 0:00:36 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4256 (0.4593) acc1: 21.3542 (79.6328) acc5: 100.0000 (100.0000) acc1_neg: 17.188 acc5_neg: 100.000 time: 0.1967 data: 0.0977 max mem: 3531 +Test: [ 90/235] eta: 0:00:33 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4327 (0.5667) acc1: 20.5729 (73.1284) acc5: 100.0000 (100.0000) acc1_neg: 19.271 acc5_neg: 100.000 time: 0.2199 data: 0.1253 max mem: 3531 +Test: [100/235] eta: 0:00:30 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4329 (0.6513) acc1: 20.8333 (67.9997) acc5: 100.0000 (100.0000) acc1_neg: 21.875 acc5_neg: 100.000 time: 0.2082 data: 0.1135 max mem: 3531 +Test: [110/235] eta: 0:00:28 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4118 (0.7203) acc1: 21.0938 (63.7880) acc5: 100.0000 (100.0000) acc1_neg: 22.656 acc5_neg: 100.000 time: 0.1999 data: 0.1042 max mem: 3531 +Test: [120/235] eta: 0:00:25 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4779 (0.7999) acc1: 19.2708 (59.4073) acc5: 100.0000 (100.0000) acc1_neg: 10.938 acc5_neg: 100.000 time: 0.2018 data: 0.1019 max mem: 3531 +Test: [130/235] eta: 0:00:23 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.6933 (0.8684) acc1: 9.3750 (55.5940) acc5: 100.0000 (100.0000) acc1_neg: 7.552 acc5_neg: 100.000 time: 0.2070 data: 0.1077 max mem: 3531 +Test: [140/235] eta: 0:00:20 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.7121 (0.9297) acc1: 8.5938 (52.2865) acc5: 100.0000 (100.0000) acc1_neg: 7.031 acc5_neg: 100.000 time: 0.1933 data: 0.0981 max mem: 3531 +Test: [150/235] eta: 0:00:18 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.7121 (0.9803) acc1: 8.5938 (49.4016) acc5: 100.0000 (100.0000) acc1_neg: 9.115 acc5_neg: 100.000 time: 0.2278 data: 0.1290 max mem: 3531 +Test: [160/235] eta: 0:00:16 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.6987 (1.0252) acc1: 9.1146 (46.9138) acc5: 100.0000 (100.0000) acc1_neg: 10.938 acc5_neg: 100.000 time: 0.2444 data: 0.1450 max mem: 3531 +Test: [170/235] eta: 0:00:14 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.6611 (1.0470) acc1: 10.9375 (45.4633) acc5: 100.0000 (100.0000) acc1_neg: 22.396 acc5_neg: 100.000 time: 0.1851 data: 0.0893 max mem: 3531 +Test: [180/235] eta: 0:00:12 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.3723 (1.0647) acc1: 23.1771 (44.3140) acc5: 100.0000 (100.0000) acc1_neg: 23.958 acc5_neg: 100.000 time: 0.1985 data: 0.1045 max mem: 3531 +Test: [190/235] eta: 0:00:09 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.3637 (1.0807) acc1: 24.4792 (43.2251) acc5: 100.0000 (100.0000) acc1_neg: 23.958 acc5_neg: 100.000 time: 0.2297 data: 0.1337 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.3686 (1.0960) acc1: 23.1771 (42.2030) acc5: 100.0000 (100.0000) acc1_neg: 22.917 acc5_neg: 100.000 time: 0.1888 data: 0.0923 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.3686 (1.1087) acc1: 22.9167 (41.3248) acc5: 100.0000 (100.0000) acc1_neg: 22.135 acc5_neg: 100.000 time: 0.1750 data: 0.0822 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.3584 (1.1201) acc1: 23.1771 (40.5131) acc5: 100.0000 (100.0000) acc1_neg: 20.833 acc5_neg: 100.000 time: 0.1882 data: 0.0969 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.3536 (1.1301) acc1: 24.7396 (39.8550) acc5: 100.0000 (100.0000) acc1_neg: 22.656 acc5_neg: 100.000 time: 0.1633 data: 0.0726 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.3584 (1.1349) acc1: 25.5208 (39.6544) acc5: 100.0000 (100.0000) acc1_neg: 26.087 acc5_neg: 100.000 time: 0.1201 data: 0.0328 max mem: 3532 +Test: Total time: 0:00:49 (0.2097 s / it) +* Overall Acc@1 39.654 Acc@5 100.000 + target == real Acc@1 97.441 Acc@5 100.000 + target == fake Acc@1 18.754 Acc@5 100.000 + loss 1.135 + PR-AUC 0.8923 ROC-AUC 0.7625 EER 30.78% (thr=0.1444) +Accuracy of the network on the 89879 test images: 39.7% +[rank0]:[W725 16:33:53.329743260 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 16:34:11.310000 83517 torch/distributed/run.py:766] +W0725 16:34:11.310000 83517 torch/distributed/run.py:766] ***************************************** +W0725 16:34:11.310000 83517 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 16:34:11.310000 83517 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 16:34:39.921180298 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 16:34:40.107281313 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 16:34:40.161922121 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 16:34:40.257142928 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 16:34:40.694343013 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 16:34:40.727121005 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 16:34:40.858906294 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 16:34:40.888643394 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n22_seed1_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed1_resampled_n116-041-188_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:22:03 acc1_pos: 89.844 acc5_pos: 100.000 loss: 0.3816 (0.3816) acc1: 89.8438 (89.8438) acc5: 100.0000 (100.0000) time: 4.9216 data: 3.0895 max mem: 3531 +Test: [ 10/269] eta: 0:02:20 acc1_pos: 94.531 acc5_pos: 100.000 loss: 0.3830 (0.3671) acc1: 89.8438 (90.2936) acc5: 100.0000 (100.0000) time: 0.5426 data: 0.2945 max mem: 3531 +Test: [ 20/269] eta: 0:01:21 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.3579 (0.3579) acc1: 91.4062 (91.6543) acc5: 100.0000 (100.0000) time: 0.0975 data: 0.0077 max mem: 3531 +Test: [ 30/269] eta: 0:01:03 acc1_pos: 96.354 acc5_pos: 100.000 loss: 0.3088 (0.3414) acc1: 96.0938 (93.0444) acc5: 100.0000 (100.0000) time: 0.1164 data: 0.0236 max mem: 3531 +Test: [ 40/269] eta: 0:00:59 acc1_pos: 86.207 acc5_pos: 100.000 loss: 0.3869 (0.4819) acc1: 88.5417 (79.4525) acc5: 100.0000 (100.0000) acc1_neg: 20.833 acc5_neg: 100.000 time: 0.1851 data: 0.0873 max mem: 3531 +Test: [ 50/269] eta: 0:00:55 acc1_pos: 86.207 acc5_pos: 100.000 loss: 1.0476 (0.6068) acc1: 22.3958 (67.5041) acc5: 100.0000 (100.0000) acc1_neg: 8.073 acc5_neg: 100.000 time: 0.2260 data: 0.1305 max mem: 3531 +Test: [ 60/269] eta: 0:00:51 acc1_pos: 86.207 acc5_pos: 100.000 loss: 1.1402 (0.7049) acc1: 14.0625 (58.4230) acc5: 100.0000 (100.0000) acc1_neg: 13.281 acc5_neg: 100.000 time: 0.2260 data: 0.1333 max mem: 3531 +Test: [ 70/269] eta: 0:00:47 acc1_pos: 86.207 acc5_pos: 100.000 loss: 1.1418 (0.7581) acc1: 14.0625 (53.0333) acc5: 100.0000 (100.0000) acc1_neg: 16.406 acc5_neg: 100.000 time: 0.1996 data: 0.1057 max mem: 3531 +Test: [ 80/269] eta: 0:00:45 acc1_pos: 86.207 acc5_pos: 100.000 loss: 0.9958 (0.7866) acc1: 22.6562 (49.5177) acc5: 100.0000 (100.0000) acc1_neg: 32.292 acc5_neg: 100.000 time: 0.2141 data: 0.1196 max mem: 3531 +Test: [ 90/269] eta: 0:00:42 acc1_pos: 86.207 acc5_pos: 100.000 loss: 0.9928 (0.8131) acc1: 22.6562 (46.5173) acc5: 100.0000 (100.0000) acc1_neg: 15.625 acc5_neg: 100.000 time: 0.2501 data: 0.1556 max mem: 3531 +Test: [100/269] eta: 0:00:39 acc1_pos: 86.207 acc5_pos: 100.000 loss: 1.0921 (0.8516) acc1: 17.7083 (43.4638) acc5: 100.0000 (100.0000) acc1_neg: 6.510 acc5_neg: 100.000 time: 0.2194 data: 0.1268 max mem: 3531 +Test: [110/269] eta: 0:00:36 acc1_pos: 86.207 acc5_pos: 100.000 loss: 1.0956 (0.8744) acc1: 15.6250 (41.0801) acc5: 100.0000 (100.0000) acc1_neg: 15.625 acc5_neg: 100.000 time: 0.1840 data: 0.0915 max mem: 3531 +Test: [120/269] eta: 0:00:34 acc1_pos: 86.207 acc5_pos: 100.000 loss: 1.0765 (0.9042) acc1: 17.1875 (39.4693) acc5: 100.0000 (100.0000) acc1_neg: 14.844 acc5_neg: 100.000 time: 0.2268 data: 0.1328 max mem: 3531 +Test: [130/269] eta: 0:00:31 acc1_pos: 86.207 acc5_pos: 100.000 loss: 1.2414 (0.9301) acc1: 17.1875 (37.6769) acc5: 100.0000 (100.0000) acc1_neg: 24.219 acc5_neg: 100.000 time: 0.2242 data: 0.1309 max mem: 3531 +Test: [140/269] eta: 0:00:29 acc1_pos: 86.207 acc5_pos: 100.000 loss: 1.1601 (0.9449) acc1: 17.7083 (36.5876) acc5: 100.0000 (100.0000) acc1_neg: 17.188 acc5_neg: 100.000 time: 0.1732 data: 0.0817 max mem: 3531 +Test: [150/269] eta: 0:00:26 acc1_pos: 86.207 acc5_pos: 100.000 loss: 1.1057 (0.9534) acc1: 23.1771 (35.8047) acc5: 100.0000 (100.0000) acc1_neg: 32.031 acc5_neg: 100.000 time: 0.1922 data: 0.0999 max mem: 3531 +Test: [160/269] eta: 0:00:24 acc1_pos: 86.207 acc5_pos: 100.000 loss: 1.1181 (0.9677) acc1: 17.7083 (34.6710) acc5: 100.0000 (100.0000) acc1_neg: 10.417 acc5_neg: 100.000 time: 0.2106 data: 0.1187 max mem: 3531 +Test: [170/269] eta: 0:00:21 acc1_pos: 86.207 acc5_pos: 100.000 loss: 1.2936 (0.9889) acc1: 9.3750 (33.0836) acc5: 100.0000 (100.0000) acc1_neg: 12.760 acc5_neg: 100.000 time: 0.1989 data: 0.1054 max mem: 3531 +Test: [180/269] eta: 0:00:19 acc1_pos: 86.207 acc5_pos: 100.000 loss: 1.2889 (1.0013) acc1: 8.0729 (31.9751) acc5: 100.0000 (100.0000) acc1_neg: 8.073 acc5_neg: 100.000 time: 0.1790 data: 0.0847 max mem: 3531 +Test: [190/269] eta: 0:00:17 acc1_pos: 86.207 acc5_pos: 100.000 loss: 1.1177 (1.0102) acc1: 10.1562 (31.0032) acc5: 100.0000 (100.0000) acc1_neg: 7.552 acc5_neg: 100.000 time: 0.1779 data: 0.0836 max mem: 3531 +Test: [200/269] eta: 0:00:14 acc1_pos: 86.207 acc5_pos: 100.000 loss: 1.2493 (1.0261) acc1: 9.6354 (29.9609) acc5: 100.0000 (100.0000) acc1_neg: 20.573 acc5_neg: 100.000 time: 0.1667 data: 0.0714 max mem: 3531 +Test: [210/269] eta: 0:00:12 acc1_pos: 86.207 acc5_pos: 100.000 loss: 1.2730 (1.0335) acc1: 9.6354 (29.3234) acc5: 100.0000 (100.0000) acc1_neg: 15.625 acc5_neg: 100.000 time: 0.1752 data: 0.0830 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 86.207 acc5_pos: 100.000 loss: 1.1189 (1.0400) acc1: 17.1875 (28.9180) acc5: 100.0000 (100.0000) acc1_neg: 39.583 acc5_neg: 100.000 time: 0.2157 data: 0.1243 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 86.207 acc5_pos: 100.000 loss: 1.0793 (1.0422) acc1: 17.7083 (28.6684) acc5: 100.0000 (100.0000) acc1_neg: 29.688 acc5_neg: 100.000 time: 0.1967 data: 0.1052 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 86.207 acc5_pos: 100.000 loss: 1.1598 (1.0492) acc1: 14.8438 (27.9197) acc5: 100.0000 (100.0000) acc1_neg: 28.906 acc5_neg: 100.000 time: 0.1755 data: 0.0843 max mem: 3532 +Test: [250/269] eta: 0:00:03 acc1_pos: 86.207 acc5_pos: 100.000 loss: 1.1631 (1.0541) acc1: 8.3333 (27.6394) acc5: 100.0000 (100.0000) acc1_neg: 8.073 acc5_neg: 100.000 time: 0.2022 data: 0.1100 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 86.207 acc5_pos: 100.000 loss: 1.2467 (1.0637) acc1: 8.0729 (26.8848) acc5: 100.0000 (100.0000) acc1_neg: 19.010 acc5_neg: 100.000 time: 0.1557 data: 0.0629 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 86.207 acc5_pos: 100.000 loss: 1.2513 (1.0671) acc1: 7.5521 (26.4035) acc5: 100.0000 (100.0000) acc1_neg: 20.472 acc5_neg: 100.000 time: 0.1037 data: 0.0068 max mem: 3532 +Test: Total time: 0:00:54 (0.2041 s / it) +* Overall Acc@1 26.404 Acc@5 100.000 + target == real Acc@1 92.841 Acc@5 100.000 + target == fake Acc@1 17.337 Acc@5 100.000 + loss 1.067 + PR-AUC 0.9232 ROC-AUC 0.6344 EER 40.82% (thr=0.2874) +Accuracy of the network on the 103293 test images: 26.4% +[rank0]:[W725 16:35:45.278704199 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-25T16:35:50Z] eval_done tag=fixed500k_fake_n22_seed1_resampled +[2026-07-25T16:35:50Z] eval_start tag=fixed500k_fake_n22_seed2_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed2_resampled_n116-041-188_20260725_112559/best_checkpoint.pth +started_utc=2026-07-25T16:35:50Z +hostname=n116-042-186 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed2_resampled_n116-041-188_20260725_112559/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n22_seed2_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0725 16:36:03.518000 84085 torch/distributed/run.py:766] +W0725 16:36:03.518000 84085 torch/distributed/run.py:766] ***************************************** +W0725 16:36:03.518000 84085 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 16:36:03.518000 84085 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W725 16:36:31.861871234 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 16:36:31.041187527 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank1]:[W725 16:36:31.041198852 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +NCCL version 2.26.2+cuda12.2 +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 16:36:32.213956232 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 16:36:32.354782427 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 16:36:32.622374111 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 16:36:32.670980548 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W725 16:36:32.670975805 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n22_seed2_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed2_resampled_n116-041-188_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:24:46 acc1_pos: 34.896 acc5_pos: 100.000 loss: 0.9147 (0.9147) acc1: 34.8958 (34.8958) acc5: 100.0000 (100.0000) time: 6.3516 data: 4.9723 max mem: 3531 +Test: [ 10/234] eta: 0:02:29 acc1_pos: 15.365 acc5_pos: 100.000 loss: 0.8888 (0.9162) acc1: 38.2812 (38.8494) acc5: 100.0000 (100.0000) time: 0.6677 data: 0.4532 max mem: 3531 +Test: [ 20/234] eta: 0:01:25 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.8888 (0.8716) acc1: 38.2812 (42.6587) acc5: 100.0000 (100.0000) time: 0.0996 data: 0.0009 max mem: 3531 +Test: [ 30/234] eta: 0:01:13 acc1_pos: 85.677 acc5_pos: 100.000 loss: 0.2171 (0.6594) acc1: 95.0521 (59.4254) acc5: 100.0000 (100.0000) time: 0.1927 data: 0.0918 max mem: 3531 +Test: [ 40/234] eta: 0:01:08 acc1_pos: 99.479 acc5_pos: 100.000 loss: 0.1946 (0.5756) acc1: 95.5729 (66.0252) acc5: 100.0000 (100.0000) time: 0.3026 data: 0.2060 max mem: 3531 +Test: [ 50/234] eta: 0:01:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0981 (0.4810) acc1: 99.2188 (72.5848) acc5: 100.0000 (100.0000) time: 0.2971 data: 0.2049 max mem: 3531 +Test: [ 60/234] eta: 0:00:52 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1011 (0.4264) acc1: 99.4792 (76.9809) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2072 data: 0.1124 max mem: 3531 +Test: [ 70/234] eta: 0:00:47 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1546 (0.4550) acc1: 98.9583 (75.8546) acc5: 100.0000 (100.0000) acc1_neg: 52.344 acc5_neg: 100.000 time: 0.1729 data: 0.0774 max mem: 3531 +Test: [ 80/234] eta: 0:00:42 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.8581 (0.4790) acc1: 53.9062 (74.9871) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2018 data: 0.1092 max mem: 3531 +Test: [ 90/234] eta: 0:00:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3636 (0.4590) acc1: 86.4583 (76.7685) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.1985 data: 0.1075 max mem: 3531 +Test: [100/234] eta: 0:00:35 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3663 (0.4658) acc1: 82.0312 (76.6579) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1952 data: 0.1024 max mem: 3531 +Test: [110/234] eta: 0:00:31 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5905 (0.4733) acc1: 68.7500 (76.4382) acc5: 100.0000 (100.0000) acc1_neg: 61.198 acc5_neg: 100.000 time: 0.1933 data: 0.1008 max mem: 3531 +Test: [120/234] eta: 0:00:28 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5601 (0.4811) acc1: 68.7500 (76.2181) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.1903 data: 0.0998 max mem: 3531 +Test: [130/234] eta: 0:00:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3698 (0.4739) acc1: 82.2917 (77.0436) acc5: 100.0000 (100.0000) acc1_neg: 75.521 acc5_neg: 100.000 time: 0.1899 data: 0.0974 max mem: 3531 +Test: [140/234] eta: 0:00:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4782 (0.5024) acc1: 78.6458 (75.6667) acc5: 100.0000 (100.0000) acc1_neg: 33.073 acc5_neg: 100.000 time: 0.1951 data: 0.1010 max mem: 3531 +Test: [150/234] eta: 0:00:19 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6259 (0.5170) acc1: 66.1458 (74.8689) acc5: 100.0000 (100.0000) acc1_neg: 71.354 acc5_neg: 100.000 time: 0.1858 data: 0.0938 max mem: 3531 +Test: [160/234] eta: 0:00:17 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5537 (0.5127) acc1: 72.3958 (75.3833) acc5: 100.0000 (100.0000) acc1_neg: 76.042 acc5_neg: 100.000 time: 0.1780 data: 0.0853 max mem: 3531 +Test: [170/234] eta: 0:00:14 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5271 (0.5124) acc1: 76.3021 (75.5696) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1901 data: 0.0973 max mem: 3531 +Test: [180/234] eta: 0:00:12 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5534 (0.5222) acc1: 74.4792 (75.0763) acc5: 100.0000 (100.0000) acc1_neg: 35.156 acc5_neg: 100.000 time: 0.1960 data: 0.1035 max mem: 3531 +Test: [190/234] eta: 0:00:10 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.0492 (0.5514) acc1: 48.6979 (73.6175) acc5: 100.0000 (100.0000) acc1_neg: 46.094 acc5_neg: 100.000 time: 0.1854 data: 0.0931 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.0492 (0.5468) acc1: 50.0000 (73.9959) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2024 data: 0.1118 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1180 (0.5300) acc1: 99.7396 (75.0037) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.2013 data: 0.1113 max mem: 3532 +Test: [220/234] eta: 0:00:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3478 (0.5245) acc1: 86.1979 (75.3641) acc5: 100.0000 (100.0000) acc1_neg: 62.240 acc5_neg: 100.000 time: 0.1351 data: 0.0454 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.8128 (0.5767) acc1: 11.4583 (72.5086) acc5: 100.0000 (100.0000) acc1_neg: 8.594 acc5_neg: 100.000 time: 0.0964 data: 0.0069 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.6905 (0.5910) acc1: 10.6771 (71.7995) acc5: 100.0000 (100.0000) acc1_neg: 11.673 acc5_neg: 100.000 time: 0.1027 data: 0.0051 max mem: 3532 +Test: Total time: 0:00:49 (0.2134 s / it) +* Overall Acc@1 71.800 Acc@5 100.000 + target == real Acc@1 72.786 Acc@5 100.000 + target == fake Acc@1 71.521 Acc@5 100.000 + loss 0.591 + PR-AUC 0.9393 ROC-AUC 0.8139 EER 27.91% (thr=0.4897) +Accuracy of the network on the 89729 test images: 71.8% +[rank0]:[W725 16:37:36.832465168 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 16:37:56.651000 84643 torch/distributed/run.py:766] +W0725 16:37:56.651000 84643 torch/distributed/run.py:766] ***************************************** +W0725 16:37:56.651000 84643 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 16:37:56.651000 84643 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 16:38:25.515685778 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 16:38:25.994837259 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 16:38:26.043921559 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 16:38:26.312219137 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 16:38:26.414277198 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W725 16:38:26.414303270 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 16:38:26.488683284 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 16:38:26.501510700 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n22_seed2_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed2_resampled_n116-041-188_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:04:16 acc1_pos: 23.177 acc5_pos: 100.000 loss: 1.1332 (1.1332) acc1: 23.1771 (23.1771) acc5: 100.0000 (100.0000) time: 6.1402 data: 4.0367 max mem: 3531 +Test: [ 10/2387] eta: 0:26:00 acc1_pos: 5.729 acc5_pos: 100.000 loss: 1.3057 (1.2633) acc1: 7.8125 (12.9025) acc5: 100.0000 (100.0000) time: 0.6565 data: 0.3836 max mem: 3531 +Test: [ 20/2387] eta: 0:15:15 acc1_pos: 3.385 acc5_pos: 100.000 loss: 1.3057 (1.2913) acc1: 9.6354 (12.2024) acc5: 100.0000 (100.0000) time: 0.0991 data: 0.0094 max mem: 3531 +Test: [ 30/2387] eta: 0:12:02 acc1_pos: 1.042 acc5_pos: 100.000 loss: 1.3654 (1.3084) acc1: 8.5938 (10.3327) acc5: 100.0000 (100.0000) time: 0.1140 data: 0.0221 max mem: 3531 +Test: [ 40/2387] eta: 0:11:37 acc1_pos: 5.990 acc5_pos: 100.000 loss: 1.3619 (1.3178) acc1: 5.9896 (9.2988) acc5: 100.0000 (100.0000) time: 0.2031 data: 0.1055 max mem: 3531 +Test: [ 50/2387] eta: 0:11:04 acc1_pos: 20.573 acc5_pos: 100.000 loss: 1.3229 (1.3164) acc1: 7.2917 (9.5078) acc5: 100.0000 (100.0000) time: 0.2498 data: 0.1507 max mem: 3531 +Test: [ 60/2387] eta: 0:10:32 acc1_pos: 24.740 acc5_pos: 100.000 loss: 1.3493 (1.3230) acc1: 9.3750 (9.7763) acc5: 100.0000 (100.0000) time: 0.2199 data: 0.1264 max mem: 3531 +Test: [ 70/2387] eta: 0:10:25 acc1_pos: 14.844 acc5_pos: 100.000 loss: 1.4193 (1.3490) acc1: 7.2917 (9.3163) acc5: 100.0000 (100.0000) time: 0.2342 data: 0.1428 max mem: 3531 +Test: [ 80/2387] eta: 0:10:31 acc1_pos: 22.917 acc5_pos: 100.000 loss: 1.3697 (1.3355) acc1: 7.2917 (9.8765) acc5: 100.0000 (100.0000) time: 0.2800 data: 0.1838 max mem: 3531 +Test: [ 90/2387] eta: 0:10:22 acc1_pos: 22.656 acc5_pos: 100.000 loss: 1.1691 (1.3121) acc1: 17.1875 (10.8202) acc5: 100.0000 (100.0000) time: 0.2749 data: 0.1751 max mem: 3531 +Test: [ 100/2387] eta: 0:10:17 acc1_pos: 10.677 acc5_pos: 100.000 loss: 1.1546 (1.3012) acc1: 15.3646 (11.0742) acc5: 100.0000 (100.0000) time: 0.2547 data: 0.1560 max mem: 3531 +Test: [ 110/2387] eta: 0:10:09 acc1_pos: 22.396 acc5_pos: 100.000 loss: 1.1327 (1.2820) acc1: 15.1042 (11.8243) acc5: 100.0000 (100.0000) time: 0.2509 data: 0.1552 max mem: 3531 +Test: [ 120/2387] eta: 0:10:03 acc1_pos: 14.583 acc5_pos: 100.000 loss: 1.1153 (1.2735) acc1: 15.1042 (12.3020) acc5: 100.0000 (100.0000) time: 0.2464 data: 0.1534 max mem: 3531 +Test: [ 130/2387] eta: 0:09:54 acc1_pos: 21.875 acc5_pos: 100.000 loss: 1.1588 (1.2670) acc1: 15.1042 (12.5099) acc5: 100.0000 (100.0000) time: 0.2395 data: 0.1473 max mem: 3531 +Test: [ 140/2387] eta: 0:09:47 acc1_pos: 14.844 acc5_pos: 100.000 loss: 1.1510 (1.2593) acc1: 15.8854 (12.7733) acc5: 100.0000 (100.0000) time: 0.2334 data: 0.1395 max mem: 3531 +Test: [ 150/2387] eta: 0:09:40 acc1_pos: 4.948 acc5_pos: 100.000 loss: 1.0908 (1.2471) acc1: 17.9688 (13.1536) acc5: 100.0000 (100.0000) time: 0.2347 data: 0.1399 max mem: 3531 +Test: [ 160/2387] eta: 0:09:28 acc1_pos: 9.589 acc5_pos: 100.000 loss: 1.0453 (1.2249) acc1: 18.4896 (14.7030) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2102 data: 0.1165 max mem: 3531 +Test: [ 170/2387] eta: 0:09:18 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2693 (1.1684) acc1: 96.0938 (19.5297) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1945 data: 0.0985 max mem: 3531 +Test: [ 180/2387] eta: 0:09:04 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2556 (1.1178) acc1: 97.3958 (23.8317) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1803 data: 0.0842 max mem: 3531 +Test: [ 190/2387] eta: 0:08:56 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2522 (1.0725) acc1: 97.1354 (27.6655) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1799 data: 0.0852 max mem: 3531 +Test: [ 200/2387] eta: 0:08:49 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2529 (1.0319) acc1: 97.1354 (31.1088) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1992 data: 0.0999 max mem: 3531 +Test: [ 210/2387] eta: 0:08:33 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2595 (0.9955) acc1: 96.6146 (34.2220) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1552 data: 0.0555 max mem: 3532 +Test: [ 220/2387] eta: 0:08:27 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2608 (0.9620) acc1: 97.3958 (37.0640) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1526 data: 0.0590 max mem: 3532 +Test: [ 230/2387] eta: 0:08:21 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2568 (0.9314) acc1: 97.3958 (39.6803) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1973 data: 0.1070 max mem: 3532 +Test: [ 240/2387] eta: 0:08:19 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2591 (0.9036) acc1: 97.3958 (42.0838) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2152 data: 0.1232 max mem: 3532 +Test: [ 250/2387] eta: 0:08:11 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2427 (0.8734) acc1: 98.1771 (44.3715) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2021 data: 0.1052 max mem: 3532 +Test: [ 260/2387] eta: 0:08:07 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1261 (0.8447) acc1: 100.0000 (46.5008) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1900 data: 0.0920 max mem: 3532 +Test: [ 270/2387] eta: 0:08:01 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1238 (0.8184) acc1: 100.0000 (48.4702) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.0992 max mem: 3532 +Test: [ 280/2387] eta: 0:07:57 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1292 (0.7941) acc1: 99.7396 (50.2929) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1931 data: 0.0998 max mem: 3532 +Test: [ 290/2387] eta: 0:07:53 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1248 (0.7710) acc1: 100.0000 (52.0010) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1999 data: 0.1057 max mem: 3532 +Test: [ 300/2387] eta: 0:07:48 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1187 (0.7492) acc1: 100.0000 (53.5853) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1907 data: 0.0949 max mem: 3532 +Test: [ 310/2387] eta: 0:07:41 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1219 (0.7325) acc1: 100.0000 (55.0082) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1722 data: 0.0798 max mem: 3532 +Test: [ 320/2387] eta: 0:07:35 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.3482 (0.7207) acc1: 96.6146 (56.2987) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1637 data: 0.0710 max mem: 3532 +Test: [ 330/2387] eta: 0:07:30 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.3533 (0.7096) acc1: 96.3542 (57.5253) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1732 data: 0.0791 max mem: 3532 +Test: [ 340/2387] eta: 0:07:26 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.3500 (0.6990) acc1: 96.8750 (58.6770) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1840 data: 0.0881 max mem: 3532 +Test: [ 350/2387] eta: 0:07:22 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.3362 (0.6844) acc1: 97.6562 (59.8446) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1865 data: 0.0882 max mem: 3532 +Test: [ 360/2387] eta: 0:07:17 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2117 (0.6713) acc1: 99.2188 (60.9202) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1762 data: 0.0808 max mem: 3532 +Test: [ 370/2387] eta: 0:07:14 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2158 (0.6591) acc1: 98.4375 (61.9371) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1776 data: 0.0817 max mem: 3532 +Test: [ 380/2387] eta: 0:07:10 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2156 (0.6475) acc1: 98.6979 (62.8992) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1835 data: 0.0879 max mem: 3532 +Test: [ 390/2387] eta: 0:07:06 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2158 (0.6364) acc1: 98.6979 (63.8161) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1849 data: 0.0915 max mem: 3532 +Test: [ 400/2387] eta: 0:07:04 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2168 (0.6260) acc1: 98.6979 (64.6788) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2049 data: 0.1104 max mem: 3532 +Test: [ 410/2387] eta: 0:07:00 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2164 (0.6160) acc1: 98.4375 (65.5059) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1949 data: 0.1012 max mem: 3532 +Test: [ 420/2387] eta: 0:06:57 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2156 (0.6065) acc1: 98.6979 (66.2906) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1849 data: 0.0915 max mem: 3532 +Test: [ 430/2387] eta: 0:06:56 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2199 (0.5977) acc1: 98.6979 (67.0401) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2099 data: 0.1153 max mem: 3532 +Test: [ 440/2387] eta: 0:06:53 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2206 (0.5875) acc1: 98.6979 (67.7709) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2111 data: 0.1165 max mem: 3533 +Test: [ 450/2387] eta: 0:06:51 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0949 (0.5766) acc1: 100.0000 (68.4855) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2098 data: 0.1165 max mem: 3533 +Test: [ 460/2387] eta: 0:06:50 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0942 (0.5661) acc1: 100.0000 (69.1692) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2286 data: 0.1341 max mem: 3533 +Test: [ 470/2387] eta: 0:06:47 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0945 (0.5561) acc1: 100.0000 (69.8237) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2199 data: 0.1231 max mem: 3533 +Test: [ 480/2387] eta: 0:06:45 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0952 (0.5465) acc1: 100.0000 (70.4511) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2054 data: 0.1093 max mem: 3533 +Test: [ 490/2387] eta: 0:06:42 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0949 (0.5373) acc1: 100.0000 (71.0529) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2001 data: 0.1070 max mem: 3533 +Test: [ 500/2387] eta: 0:06:40 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0948 (0.5285) acc1: 100.0000 (71.6302) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1967 data: 0.1056 max mem: 3533 +Test: [ 510/2387] eta: 0:06:39 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0953 (0.5203) acc1: 100.0000 (72.1854) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2297 data: 0.1391 max mem: 3533 +Test: [ 520/2387] eta: 0:06:37 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1369 (0.5131) acc1: 100.0000 (72.7192) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2325 data: 0.1382 max mem: 3533 +Test: [ 530/2387] eta: 0:06:34 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1406 (0.5060) acc1: 100.0000 (73.2330) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1957 data: 0.0997 max mem: 3533 +Test: [ 540/2387] eta: 0:06:34 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1396 (0.4992) acc1: 100.0000 (73.7273) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2350 data: 0.1421 max mem: 3533 +Test: [ 550/2387] eta: 0:06:32 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1395 (0.4927) acc1: 100.0000 (74.2041) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2544 data: 0.1627 max mem: 3533 +Test: [ 560/2387] eta: 0:06:30 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1399 (0.4864) acc1: 100.0000 (74.6639) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2148 data: 0.1213 max mem: 3533 +Test: [ 570/2387] eta: 0:06:28 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1394 (0.4804) acc1: 100.0000 (75.1076) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2095 data: 0.1154 max mem: 3533 +Test: [ 580/2387] eta: 0:06:24 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1381 (0.4745) acc1: 100.0000 (75.5361) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1850 data: 0.0911 max mem: 3533 +Test: [ 590/2387] eta: 0:06:25 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1379 (0.4688) acc1: 100.0000 (75.9500) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2309 data: 0.1384 max mem: 3533 +Test: [ 600/2387] eta: 0:06:21 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1421 (0.4634) acc1: 100.0000 (76.3493) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2398 data: 0.1469 max mem: 3533 +Test: [ 610/2387] eta: 0:06:16 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1571 (0.4589) acc1: 100.0000 (76.7185) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1439 data: 0.0465 max mem: 3533 +Test: [ 620/2387] eta: 0:06:14 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1860 (0.4545) acc1: 98.9583 (77.0791) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1527 data: 0.0560 max mem: 3533 +Test: [ 630/2387] eta: 0:06:13 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1863 (0.4502) acc1: 98.9583 (77.4234) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2249 data: 0.1309 max mem: 3533 +Test: [ 640/2387] eta: 0:06:10 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1881 (0.4462) acc1: 98.9583 (77.7606) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2271 data: 0.1337 max mem: 3533 +Test: [ 650/2387] eta: 0:06:07 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1881 (0.4422) acc1: 99.2188 (78.0890) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1780 data: 0.0849 max mem: 3533 +Test: [ 660/2387] eta: 0:06:04 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1892 (0.4384) acc1: 99.2188 (78.4055) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1764 data: 0.0828 max mem: 3533 +Test: [ 670/2387] eta: 0:06:02 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1865 (0.4346) acc1: 99.2188 (78.7145) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1991 data: 0.1070 max mem: 3534 +Test: [ 680/2387] eta: 0:06:00 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1849 (0.4309) acc1: 99.4792 (79.0221) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2135 data: 0.1206 max mem: 3534 +Test: [ 690/2387] eta: 0:05:58 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1868 (0.4320) acc1: 99.2188 (78.9262) acc5: 100.0000 (100.0000) acc1_neg: 58.333 acc5_neg: 100.000 time: 0.2176 data: 0.1199 max mem: 3534 +Test: [ 700/2387] eta: 0:05:55 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.6280 (0.4348) acc1: 62.7604 (78.7045) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.2052 data: 0.1036 max mem: 3534 +Test: [ 710/2387] eta: 0:05:54 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.6306 (0.4377) acc1: 62.5000 (78.4777) acc5: 100.0000 (100.0000) acc1_neg: 61.979 acc5_neg: 100.000 time: 0.2154 data: 0.1178 max mem: 3534 +Test: [ 720/2387] eta: 0:05:52 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.6351 (0.4403) acc1: 62.5000 (78.2633) acc5: 100.0000 (100.0000) acc1_neg: 61.198 acc5_neg: 100.000 time: 0.2196 data: 0.1253 max mem: 3534 +Test: [ 730/2387] eta: 0:05:47 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.6317 (0.4430) acc1: 62.5000 (78.0488) acc5: 100.0000 (100.0000) acc1_neg: 59.375 acc5_neg: 100.000 time: 0.1644 data: 0.0671 max mem: 3534 +Test: [ 740/2387] eta: 0:05:46 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.6317 (0.4456) acc1: 62.5000 (77.8424) acc5: 100.0000 (100.0000) acc1_neg: 59.115 acc5_neg: 100.000 time: 0.1702 data: 0.0731 max mem: 3534 +Test: [ 750/2387] eta: 0:05:44 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.6235 (0.4480) acc1: 62.5000 (77.6361) acc5: 100.0000 (100.0000) acc1_neg: 62.760 acc5_neg: 100.000 time: 0.2321 data: 0.1384 max mem: 3534 +Test: [ 760/2387] eta: 0:05:41 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.6135 (0.4490) acc1: 62.7604 (77.5508) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1899 data: 0.0987 max mem: 3534 +Test: [ 770/2387] eta: 0:05:38 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.3601 (0.4469) acc1: 85.4167 (77.7285) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1577 data: 0.0643 max mem: 3534 +Test: [ 780/2387] eta: 0:05:36 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2076 (0.4436) acc1: 99.2188 (78.0100) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1897 data: 0.0934 max mem: 3534 +Test: [ 790/2387] eta: 0:05:34 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1753 (0.4405) acc1: 99.7396 (78.2340) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.2068 data: 0.1131 max mem: 3534 +Test: [ 800/2387] eta: 0:05:30 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1763 (0.4372) acc1: 99.4792 (78.4995) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1811 data: 0.0884 max mem: 3534 +Test: [ 810/2387] eta: 0:05:28 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1785 (0.4341) acc1: 99.4792 (78.7531) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1693 data: 0.0749 max mem: 3534 +Test: [ 820/2387] eta: 0:05:25 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1785 (0.4310) acc1: 99.7396 (79.0005) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1909 data: 0.0982 max mem: 3534 +Test: [ 830/2387] eta: 0:05:23 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2088 (0.4290) acc1: 97.9167 (79.2074) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1844 data: 0.0926 max mem: 3534 +Test: [ 840/2387] eta: 0:05:20 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.3252 (0.4282) acc1: 92.7083 (79.3249) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1796 data: 0.0879 max mem: 3534 +Test: [ 850/2387] eta: 0:05:18 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.3678 (0.4277) acc1: 88.5417 (79.4188) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.1858 data: 0.0943 max mem: 3534 +Test: [ 860/2387] eta: 0:05:16 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.3634 (0.4269) acc1: 89.5833 (79.5450) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.2029 data: 0.1076 max mem: 3534 +Test: [ 870/2387] eta: 0:05:13 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.3466 (0.4259) acc1: 90.6250 (79.6779) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1946 data: 0.0982 max mem: 3534 +Test: [ 880/2387] eta: 0:05:11 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.3480 (0.4253) acc1: 90.3646 (79.7809) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.1726 data: 0.0779 max mem: 3534 +Test: [ 890/2387] eta: 0:05:08 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2926 (0.4234) acc1: 93.4896 (79.9652) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1855 data: 0.0905 max mem: 3534 +Test: [ 900/2387] eta: 0:05:07 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2926 (0.4268) acc1: 90.1042 (79.7439) acc5: 100.0000 (100.0000) acc1_neg: 51.302 acc5_neg: 100.000 time: 0.2127 data: 0.1179 max mem: 3535 +Test: [ 910/2387] eta: 0:05:04 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.8385 (0.4312) acc1: 51.3021 (79.4368) acc5: 100.0000 (100.0000) acc1_neg: 52.083 acc5_neg: 100.000 time: 0.2087 data: 0.1159 max mem: 3535 +Test: [ 920/2387] eta: 0:05:02 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.8398 (0.4357) acc1: 50.7812 (79.1209) acc5: 100.0000 (100.0000) acc1_neg: 48.698 acc5_neg: 100.000 time: 0.1897 data: 0.0988 max mem: 3535 +Test: [ 930/2387] eta: 0:05:00 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.8395 (0.4400) acc1: 50.5208 (78.8198) acc5: 100.0000 (100.0000) acc1_neg: 44.531 acc5_neg: 100.000 time: 0.2001 data: 0.1090 max mem: 3535 +Test: [ 940/2387] eta: 0:04:57 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.8398 (0.4443) acc1: 50.5208 (78.5124) acc5: 100.0000 (100.0000) acc1_neg: 51.823 acc5_neg: 100.000 time: 0.1909 data: 0.0995 max mem: 3535 +Test: [ 950/2387] eta: 0:04:55 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.8373 (0.4485) acc1: 48.9583 (78.2069) acc5: 100.0000 (100.0000) acc1_neg: 54.167 acc5_neg: 100.000 time: 0.1768 data: 0.0854 max mem: 3535 +Test: [ 960/2387] eta: 0:04:53 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.8263 (0.4524) acc1: 50.7812 (77.9348) acc5: 100.0000 (100.0000) acc1_neg: 57.031 acc5_neg: 100.000 time: 0.1945 data: 0.1015 max mem: 3535 +Test: [ 970/2387] eta: 0:04:51 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.8305 (0.4565) acc1: 51.0417 (77.6490) acc5: 100.0000 (100.0000) acc1_neg: 50.260 acc5_neg: 100.000 time: 0.2034 data: 0.1100 max mem: 3535 +Test: [ 980/2387] eta: 0:04:49 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.8041 (0.4561) acc1: 52.0833 (77.6883) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.1954 data: 0.1041 max mem: 3535 +Test: [ 990/2387] eta: 0:04:46 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2109 (0.4540) acc1: 98.4375 (77.8656) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1885 data: 0.0959 max mem: 3535 +Test: [1000/2387] eta: 0:04:44 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2109 (0.4518) acc1: 99.2188 (78.0519) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1908 data: 0.0954 max mem: 3535 +Test: [1010/2387] eta: 0:04:42 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2269 (0.4499) acc1: 97.3958 (78.2200) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1970 data: 0.1005 max mem: 3535 +Test: [1020/2387] eta: 0:04:39 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2299 (0.4480) acc1: 97.6562 (78.3857) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1802 data: 0.0845 max mem: 3535 +Test: [1030/2387] eta: 0:04:37 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1842 (0.4454) acc1: 99.4792 (78.5872) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1746 data: 0.0819 max mem: 3535 +Test: [1040/2387] eta: 0:04:35 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1691 (0.4427) acc1: 99.7396 (78.7902) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1885 data: 0.0977 max mem: 3535 +Test: [1050/2387] eta: 0:04:33 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1486 (0.4400) acc1: 100.0000 (78.9890) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1940 data: 0.1029 max mem: 3535 +Test: [1060/2387] eta: 0:04:31 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1486 (0.4375) acc1: 99.7396 (79.1826) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2056 data: 0.1125 max mem: 3535 +Test: [1070/2387] eta: 0:04:28 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1609 (0.4350) acc1: 100.0000 (79.3690) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1991 data: 0.1059 max mem: 3535 +Test: [1080/2387] eta: 0:04:26 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1999 (0.4331) acc1: 99.2188 (79.5389) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1756 data: 0.0823 max mem: 3535 +Test: [1090/2387] eta: 0:04:24 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2197 (0.4312) acc1: 98.9583 (79.7190) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1754 data: 0.0810 max mem: 3535 +Test: [1100/2387] eta: 0:04:21 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2111 (0.4294) acc1: 99.2188 (79.8869) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1877 data: 0.0948 max mem: 3535 +Test: [1110/2387] eta: 0:04:19 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2003 (0.4278) acc1: 98.9583 (80.0164) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1905 data: 0.0989 max mem: 3535 +Test: [1120/2387] eta: 0:04:17 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2223 (0.4261) acc1: 98.4375 (80.1772) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1872 data: 0.0957 max mem: 3536 +Test: [1130/2387] eta: 0:04:15 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2183 (0.4243) acc1: 98.6979 (80.3414) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1925 data: 0.1011 max mem: 3536 +Test: [1140/2387] eta: 0:04:13 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2240 (0.4228) acc1: 98.6979 (80.4907) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1921 data: 0.0997 max mem: 3536 +Test: [1150/2387] eta: 0:04:11 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2798 (0.4221) acc1: 95.8333 (80.5893) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.1864 data: 0.0924 max mem: 3536 +Test: [1160/2387] eta: 0:04:08 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2999 (0.4209) acc1: 93.2292 (80.6919) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.1761 data: 0.0832 max mem: 3536 +Test: [1170/2387] eta: 0:04:06 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2812 (0.4199) acc1: 92.7083 (80.7905) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1646 data: 0.0731 max mem: 3536 +Test: [1180/2387] eta: 0:04:03 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.3119 (0.4190) acc1: 92.7083 (80.8813) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1693 data: 0.0783 max mem: 3536 +Test: [1190/2387] eta: 0:04:01 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2821 (0.4180) acc1: 94.0104 (80.9721) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1725 data: 0.0801 max mem: 3536 +Test: [1200/2387] eta: 0:03:59 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2462 (0.4163) acc1: 95.5729 (81.1160) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1818 data: 0.0878 max mem: 3536 +Test: [1210/2387] eta: 0:03:57 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2168 (0.4146) acc1: 99.2188 (81.2601) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2008 data: 0.1077 max mem: 3536 +Test: [1220/2387] eta: 0:03:55 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1755 (0.4125) acc1: 99.2188 (81.4061) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2005 data: 0.1083 max mem: 3536 +Test: [1230/2387] eta: 0:03:53 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1588 (0.4105) acc1: 99.2188 (81.5500) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1927 data: 0.1005 max mem: 3536 +Test: [1240/2387] eta: 0:03:51 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1645 (0.4085) acc1: 99.2188 (81.6936) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1968 data: 0.1048 max mem: 3536 +Test: [1250/2387] eta: 0:03:49 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1624 (0.4065) acc1: 99.4792 (81.8345) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2005 data: 0.1081 max mem: 3536 +Test: [1260/2387] eta: 0:03:47 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1434 (0.4045) acc1: 99.4792 (81.9712) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1959 data: 0.1003 max mem: 3536 +Test: [1270/2387] eta: 0:03:45 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1306 (0.4023) acc1: 100.0000 (82.1124) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1955 data: 0.1019 max mem: 3536 +Test: [1280/2387] eta: 0:03:42 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1182 (0.4000) acc1: 100.0000 (82.2498) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1943 data: 0.1024 max mem: 3536 +Test: [1290/2387] eta: 0:03:40 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1121 (0.3978) acc1: 99.7396 (82.3713) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1938 data: 0.1016 max mem: 3536 +Test: [1300/2387] eta: 0:03:38 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1135 (0.3957) acc1: 99.7396 (82.5008) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2006 data: 0.1094 max mem: 3536 +Test: [1310/2387] eta: 0:03:36 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1166 (0.3936) acc1: 99.7396 (82.6264) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1905 data: 0.0997 max mem: 3536 +Test: [1320/2387] eta: 0:03:34 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1195 (0.3916) acc1: 100.0000 (82.7465) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1939 data: 0.1036 max mem: 3536 +Test: [1330/2387] eta: 0:03:32 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1239 (0.3898) acc1: 99.7396 (82.8665) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1837 data: 0.0922 max mem: 3536 +Test: [1340/2387] eta: 0:03:30 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1653 (0.3895) acc1: 98.4375 (82.8830) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1712 data: 0.0768 max mem: 3536 +Test: [1350/2387] eta: 0:03:28 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0868 (0.3872) acc1: 100.0000 (83.0097) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1859 data: 0.0924 max mem: 3537 +Test: [1360/2387] eta: 0:03:26 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0786 (0.3849) acc1: 100.0000 (83.1345) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1887 data: 0.0964 max mem: 3537 +Test: [1370/2387] eta: 0:03:23 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0758 (0.3827) acc1: 100.0000 (83.2575) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1880 data: 0.0960 max mem: 3537 +Test: [1380/2387] eta: 0:03:21 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0765 (0.3805) acc1: 100.0000 (83.3788) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1746 data: 0.0801 max mem: 3537 +Test: [1390/2387] eta: 0:03:19 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0825 (0.3784) acc1: 100.0000 (83.4983) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1915 data: 0.0951 max mem: 3537 +Test: [1400/2387] eta: 0:03:17 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0792 (0.3762) acc1: 100.0000 (83.6161) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2014 data: 0.1075 max mem: 3537 +Test: [1410/2387] eta: 0:03:15 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.0792 (0.3747) acc1: 100.0000 (83.7229) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1700 data: 0.0781 max mem: 3537 +Test: [1420/2387] eta: 0:03:13 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1876 (0.3736) acc1: 99.7396 (83.8124) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1599 data: 0.0672 max mem: 3537 +Test: [1430/2387] eta: 0:03:11 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2065 (0.3725) acc1: 98.6979 (83.9120) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1835 data: 0.0907 max mem: 3537 +Test: [1440/2387] eta: 0:03:09 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1993 (0.3712) acc1: 98.9583 (84.0175) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1957 data: 0.1017 max mem: 3537 +Test: [1450/2387] eta: 0:03:07 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1949 (0.3700) acc1: 99.4792 (84.1148) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1961 data: 0.1018 max mem: 3537 +Test: [1460/2387] eta: 0:03:05 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1903 (0.3689) acc1: 98.6979 (84.2101) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1977 data: 0.1044 max mem: 3537 +Test: [1470/2387] eta: 0:03:02 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1811 (0.3676) acc1: 99.2188 (84.3107) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1651 data: 0.0719 max mem: 3537 +Test: [1480/2387] eta: 0:03:00 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1797 (0.3664) acc1: 99.2188 (84.4056) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1707 data: 0.0798 max mem: 3537 +Test: [1490/2387] eta: 0:02:58 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1797 (0.3652) acc1: 98.9583 (84.5016) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1942 data: 0.1017 max mem: 3537 +Test: [1500/2387] eta: 0:02:56 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1760 (0.3639) acc1: 99.2188 (84.5946) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1842 data: 0.0913 max mem: 3537 +Test: [1510/2387] eta: 0:02:54 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1756 (0.3627) acc1: 99.4792 (84.6887) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1760 data: 0.0843 max mem: 3537 +Test: [1520/2387] eta: 0:02:52 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2055 (0.3623) acc1: 97.9167 (84.7327) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1862 data: 0.0953 max mem: 3537 +Test: [1530/2387] eta: 0:02:50 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.3307 (0.3623) acc1: 89.5833 (84.7557) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.1975 data: 0.1074 max mem: 3537 +Test: [1540/2387] eta: 0:02:48 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.3241 (0.3620) acc1: 89.5833 (84.8051) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1899 data: 0.0980 max mem: 3537 +Test: [1550/2387] eta: 0:02:46 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.3157 (0.3616) acc1: 92.7083 (84.8612) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1749 data: 0.0800 max mem: 3537 +Test: [1560/2387] eta: 0:02:44 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.3082 (0.3614) acc1: 94.0104 (84.9087) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1783 data: 0.0830 max mem: 3537 +Test: [1570/2387] eta: 0:02:42 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2664 (0.3606) acc1: 95.3125 (84.9830) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1904 data: 0.0979 max mem: 3537 +Test: [1580/2387] eta: 0:02:40 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1874 (0.3595) acc1: 98.6979 (85.0732) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1863 data: 0.0953 max mem: 3538 +Test: [1590/2387] eta: 0:02:37 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1866 (0.3584) acc1: 99.4792 (85.1639) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1754 data: 0.0767 max mem: 3538 +Test: [1600/2387] eta: 0:02:35 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1796 (0.3574) acc1: 99.7396 (85.2255) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1684 data: 0.0693 max mem: 3538 +Test: [1610/2387] eta: 0:02:33 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1576 (0.3562) acc1: 99.7396 (85.3156) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1610 data: 0.0687 max mem: 3538 +Test: [1620/2387] eta: 0:02:31 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1749 (0.3552) acc1: 99.7396 (85.4004) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1501 data: 0.0589 max mem: 3538 +Test: [1630/2387] eta: 0:02:29 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1939 (0.3541) acc1: 99.4792 (85.4783) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1599 data: 0.0672 max mem: 3538 +Test: [1640/2387] eta: 0:02:27 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1972 (0.3532) acc1: 98.9583 (85.5519) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1704 data: 0.0773 max mem: 3538 +Test: [1650/2387] eta: 0:02:25 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1860 (0.3522) acc1: 99.2188 (85.6340) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1876 data: 0.0950 max mem: 3538 +Test: [1660/2387] eta: 0:02:23 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1860 (0.3512) acc1: 99.2188 (85.7119) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1962 data: 0.1011 max mem: 3538 +Test: [1670/2387] eta: 0:02:21 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1826 (0.3503) acc1: 98.9583 (85.7885) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1649 data: 0.0709 max mem: 3538 +Test: [1680/2387] eta: 0:02:18 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1688 (0.3494) acc1: 99.2188 (85.8597) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1527 data: 0.0606 max mem: 3538 +Test: [1690/2387] eta: 0:02:16 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1867 (0.3485) acc1: 98.9583 (85.9298) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1477 data: 0.0540 max mem: 3538 +Test: [1700/2387] eta: 0:02:14 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2016 (0.3476) acc1: 98.9583 (86.0072) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1506 data: 0.0557 max mem: 3538 +Test: [1710/2387] eta: 0:02:12 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1861 (0.3467) acc1: 99.2188 (86.0809) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1792 data: 0.0863 max mem: 3538 +Test: [1720/2387] eta: 0:02:10 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1861 (0.3458) acc1: 98.9583 (86.1546) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2042 data: 0.1128 max mem: 3538 +Test: [1730/2387] eta: 0:02:08 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1950 (0.3449) acc1: 98.6979 (86.2239) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1991 data: 0.1049 max mem: 3538 +Test: [1740/2387] eta: 0:02:06 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1902 (0.3440) acc1: 99.2188 (86.2990) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1865 data: 0.0926 max mem: 3538 +Test: [1750/2387] eta: 0:02:04 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1730 (0.3431) acc1: 99.2188 (86.3727) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1946 data: 0.1036 max mem: 3538 +Test: [1760/2387] eta: 0:02:02 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1568 (0.3420) acc1: 100.0000 (86.4480) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1954 data: 0.1047 max mem: 3538 +Test: [1770/2387] eta: 0:02:00 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1734 (0.3414) acc1: 98.4375 (86.5049) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1991 data: 0.1077 max mem: 3538 +Test: [1780/2387] eta: 0:01:59 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2237 (0.3407) acc1: 96.8750 (86.5667) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2229 data: 0.1312 max mem: 3538 +Test: [1790/2387] eta: 0:01:57 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2151 (0.3400) acc1: 98.1771 (86.6334) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2060 data: 0.1145 max mem: 3538 +Test: [1800/2387] eta: 0:01:54 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2174 (0.3394) acc1: 98.1771 (86.6947) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1648 data: 0.0723 max mem: 3538 +Test: [1810/2387] eta: 0:01:52 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2075 (0.3386) acc1: 98.1771 (86.7617) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1613 data: 0.0688 max mem: 3539 +Test: [1820/2387] eta: 0:01:51 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1781 (0.3377) acc1: 99.4792 (86.8297) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1966 data: 0.1054 max mem: 3539 +Test: [1830/2387] eta: 0:01:49 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1529 (0.3366) acc1: 100.0000 (86.9012) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2172 data: 0.1255 max mem: 3539 +Test: [1840/2387] eta: 0:01:46 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1073 (0.3354) acc1: 100.0000 (86.9724) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1638 data: 0.0686 max mem: 3539 +Test: [1850/2387] eta: 0:01:44 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1059 (0.3341) acc1: 100.0000 (87.0426) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1555 data: 0.0586 max mem: 3539 +Test: [1860/2387] eta: 0:01:43 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1060 (0.3329) acc1: 100.0000 (87.1122) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1892 data: 0.0939 max mem: 3539 +Test: [1870/2387] eta: 0:01:41 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1065 (0.3317) acc1: 100.0000 (87.1807) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1814 data: 0.0864 max mem: 3539 +Test: [1880/2387] eta: 0:01:39 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1064 (0.3305) acc1: 100.0000 (87.2487) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1799 data: 0.0839 max mem: 3539 +Test: [1890/2387] eta: 0:01:37 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1062 (0.3293) acc1: 100.0000 (87.3162) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1953 data: 0.0978 max mem: 3539 +Test: [1900/2387] eta: 0:01:35 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1075 (0.3282) acc1: 100.0000 (87.3827) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1857 data: 0.0912 max mem: 3539 +Test: [1910/2387] eta: 0:01:33 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1253 (0.3272) acc1: 100.0000 (87.4441) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1681 data: 0.0766 max mem: 3539 +Test: [1920/2387] eta: 0:01:31 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1487 (0.3267) acc1: 100.0000 (87.4958) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1851 data: 0.0927 max mem: 3539 +Test: [1930/2387] eta: 0:01:29 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2029 (0.3260) acc1: 99.7396 (87.5592) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1914 data: 0.0970 max mem: 3539 +Test: [1940/2387] eta: 0:01:27 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2027 (0.3257) acc1: 99.7396 (87.5863) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1889 data: 0.0941 max mem: 3539 +Test: [1950/2387] eta: 0:01:25 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2101 (0.3252) acc1: 98.9583 (87.6416) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1927 data: 0.0992 max mem: 3539 +Test: [1960/2387] eta: 0:01:23 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2141 (0.3246) acc1: 99.7396 (87.6953) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1814 data: 0.0875 max mem: 3539 +Test: [1970/2387] eta: 0:01:21 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2191 (0.3242) acc1: 99.7396 (87.7349) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1867 data: 0.0938 max mem: 3539 +Test: [1980/2387] eta: 0:01:19 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2405 (0.3236) acc1: 97.1354 (87.7786) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2034 data: 0.1103 max mem: 3539 +Test: [1990/2387] eta: 0:01:17 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1291 (0.3226) acc1: 98.4375 (87.8324) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1931 data: 0.0990 max mem: 3539 +Test: [2000/2387] eta: 0:01:15 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1271 (0.3217) acc1: 98.4375 (87.8863) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1967 data: 0.1033 max mem: 3539 +Test: [2010/2387] eta: 0:01:13 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1282 (0.3207) acc1: 98.4375 (87.9394) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2098 data: 0.1176 max mem: 3539 +Test: [2020/2387] eta: 0:01:11 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1300 (0.3198) acc1: 98.4375 (87.9927) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1703 data: 0.0752 max mem: 3539 +Test: [2030/2387] eta: 0:01:09 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1303 (0.3188) acc1: 98.4375 (88.0442) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1850 data: 0.0886 max mem: 3540 +Test: [2040/2387] eta: 0:01:07 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1281 (0.3179) acc1: 98.6979 (88.0966) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2197 data: 0.1249 max mem: 3540 +Test: [2050/2387] eta: 0:01:05 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1235 (0.3169) acc1: 98.9583 (88.1492) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1685 data: 0.0734 max mem: 3540 +Test: [2060/2387] eta: 0:01:03 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1235 (0.3160) acc1: 98.6979 (88.2000) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1842 data: 0.0911 max mem: 3540 +Test: [2070/2387] eta: 0:01:01 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1210 (0.3150) acc1: 98.6979 (88.2528) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2100 data: 0.1186 max mem: 3540 +Test: [2080/2387] eta: 0:00:59 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1168 (0.3141) acc1: 98.9583 (88.3038) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1822 data: 0.0871 max mem: 3540 +Test: [2090/2387] eta: 0:00:57 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1155 (0.3132) acc1: 98.9583 (88.3552) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1713 data: 0.0738 max mem: 3540 +Test: [2100/2387] eta: 0:00:55 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1175 (0.3122) acc1: 98.9583 (88.4066) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1848 data: 0.0891 max mem: 3540 +Test: [2110/2387] eta: 0:00:53 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1187 (0.3113) acc1: 98.9583 (88.4565) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1993 data: 0.1064 max mem: 3540 +Test: [2120/2387] eta: 0:00:51 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1306 (0.3107) acc1: 98.9583 (88.5058) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1908 data: 0.1001 max mem: 3540 +Test: [2130/2387] eta: 0:00:50 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1748 (0.3100) acc1: 99.7396 (88.5580) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1890 data: 0.0956 max mem: 3540 +Test: [2140/2387] eta: 0:00:48 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1712 (0.3096) acc1: 99.7396 (88.5861) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2120 data: 0.1175 max mem: 3540 +Test: [2150/2387] eta: 0:00:46 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1618 (0.3089) acc1: 99.7396 (88.6351) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2176 data: 0.1239 max mem: 3540 +Test: [2160/2387] eta: 0:00:44 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1664 (0.3083) acc1: 100.0000 (88.6837) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1698 data: 0.0773 max mem: 3540 +Test: [2170/2387] eta: 0:00:42 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1795 (0.3078) acc1: 99.7396 (88.7288) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1584 data: 0.0658 max mem: 3540 +Test: [2180/2387] eta: 0:00:40 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2291 (0.3087) acc1: 96.0938 (88.6675) acc5: 100.0000 (100.0000) acc1_neg: 47.917 acc5_neg: 100.000 time: 0.1785 data: 0.0851 max mem: 3540 +Test: [2190/2387] eta: 0:00:38 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.5089 (0.3098) acc1: 71.6146 (88.5936) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1811 data: 0.0897 max mem: 3540 +Test: [2200/2387] eta: 0:00:36 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2186 (0.3092) acc1: 97.9167 (88.6431) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1716 data: 0.0814 max mem: 3540 +Test: [2210/2387] eta: 0:00:34 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1877 (0.3087) acc1: 99.2188 (88.6878) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1774 data: 0.0845 max mem: 3540 +Test: [2220/2387] eta: 0:00:32 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1958 (0.3083) acc1: 98.4375 (88.7288) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1670 data: 0.0720 max mem: 3540 +Test: [2230/2387] eta: 0:00:30 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1956 (0.3077) acc1: 99.2188 (88.7737) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1564 data: 0.0619 max mem: 3540 +Test: [2240/2387] eta: 0:00:28 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1805 (0.3071) acc1: 99.2188 (88.8127) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1661 data: 0.0737 max mem: 3540 +Test: [2250/2387] eta: 0:00:26 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1858 (0.3067) acc1: 97.9167 (88.8513) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1685 data: 0.0759 max mem: 3540 +Test: [2260/2387] eta: 0:00:24 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1713 (0.3061) acc1: 99.4792 (88.8983) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1650 data: 0.0705 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1328 (0.3053) acc1: 100.0000 (88.9470) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1765 data: 0.0842 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1262 (0.3047) acc1: 100.0000 (88.9753) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2094 data: 0.1185 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1228 (0.3039) acc1: 100.0000 (89.0232) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1633 data: 0.0699 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1368 (0.3032) acc1: 100.0000 (89.0684) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1593 data: 0.0661 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1523 (0.3026) acc1: 100.0000 (89.1102) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1876 data: 0.0971 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.1859 (0.3023) acc1: 99.4792 (89.1454) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1281 data: 0.0378 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2043 (0.3019) acc1: 99.4792 (89.1838) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.0983 data: 0.0078 max mem: 3541 +Test: [2340/2387] eta: 0:00:09 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2086 (0.3015) acc1: 99.7396 (89.2264) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1230 data: 0.0327 max mem: 3541 +Test: [2350/2387] eta: 0:00:07 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2086 (0.3014) acc1: 99.4792 (89.2442) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1437 data: 0.0527 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2165 (0.3012) acc1: 99.4792 (89.2741) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1309 data: 0.0403 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2161 (0.3008) acc1: 99.4792 (89.3119) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1226 data: 0.0325 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2161 (0.3007) acc1: 99.2188 (89.3382) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1325 data: 0.0420 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 9.589 acc5_pos: 100.000 loss: 0.2946 (0.3018) acc1: 94.0104 (89.2845) acc5: 100.0000 (100.0000) acc1_neg: 30.233 acc5_neg: 100.000 time: 0.1199 data: 0.0323 max mem: 3541 +Test: Total time: 0:07:36 (0.1911 s / it) +* Overall Acc@1 89.284 Acc@5 100.000 + target == real Acc@1 13.025 Acc@5 100.000 + target == fake Acc@1 94.640 Acc@5 100.000 + loss 0.302 + PR-AUC 0.9826 ROC-AUC 0.7968 EER 26.75% (thr=0.7876) +Accuracy of the network on the 916267 test images: 89.3% +[rank0]:[W725 16:46:16.701791393 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 16:46:36.648000 85255 torch/distributed/run.py:766] +W0725 16:46:36.648000 85255 torch/distributed/run.py:766] ***************************************** +W0725 16:46:36.648000 85255 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 16:46:36.648000 85255 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 16:47:04.322648089 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 16:47:05.314651774 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 16:47:05.464711926 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 16:47:05.496649631 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 16:47:05.541592849 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 16:47:05.574040822 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 16:47:05.731102439 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 16:47:05.758578054 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n22_seed2_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed2_resampled_n116-041-188_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 0:52:14 acc1_pos: 23.177 acc5_pos: 100.000 loss: 1.1332 (1.1332) acc1: 23.1771 (23.1771) acc5: 100.0000 (100.0000) time: 6.0621 data: 3.8541 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/517] eta: 0:05:20 acc1_pos: 5.729 acc5_pos: 100.000 loss: 1.3057 (1.2633) acc1: 7.8125 (12.9025) acc5: 100.0000 (100.0000) time: 0.6329 data: 0.3507 max mem: 3531 +Test: [ 20/517] eta: 0:03:06 acc1_pos: 3.385 acc5_pos: 100.000 loss: 1.3057 (1.2913) acc1: 9.6354 (12.2024) acc5: 100.0000 (100.0000) time: 0.0901 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:02:36 acc1_pos: 1.042 acc5_pos: 100.000 loss: 1.3654 (1.3084) acc1: 8.5938 (10.3327) acc5: 100.0000 (100.0000) time: 0.1505 data: 0.0569 max mem: 3531 +Test: [ 40/517] eta: 0:02:26 acc1_pos: 5.990 acc5_pos: 100.000 loss: 1.3619 (1.3178) acc1: 5.9896 (9.2988) acc5: 100.0000 (100.0000) time: 0.2358 data: 0.1405 max mem: 3531 +Test: [ 50/517] eta: 0:02:20 acc1_pos: 14.220 acc5_pos: 100.000 loss: 1.2915 (1.2640) acc1: 7.2917 (13.5723) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2706 data: 0.1749 max mem: 3531 +Test: [ 60/517] eta: 0:02:13 acc1_pos: 14.220 acc5_pos: 100.000 loss: 0.2742 (1.0996) acc1: 94.2708 (27.3480) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2640 data: 0.1690 max mem: 3531 +Test: [ 70/517] eta: 0:02:08 acc1_pos: 14.220 acc5_pos: 100.000 loss: 0.2329 (0.9741) acc1: 99.2188 (37.5550) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2490 data: 0.1549 max mem: 3531 +Test: [ 80/517] eta: 0:02:03 acc1_pos: 14.220 acc5_pos: 100.000 loss: 0.2051 (0.8813) acc1: 100.0000 (45.2096) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2546 data: 0.1608 max mem: 3531 +Test: [ 90/517] eta: 0:02:02 acc1_pos: 14.220 acc5_pos: 100.000 loss: 0.2442 (0.8112) acc1: 99.2188 (51.1218) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2841 data: 0.1902 max mem: 3531 +Test: [100/517] eta: 0:01:58 acc1_pos: 14.220 acc5_pos: 100.000 loss: 0.2352 (0.7531) acc1: 99.2188 (55.9251) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2847 data: 0.1870 max mem: 3531 +Test: [110/517] eta: 0:01:54 acc1_pos: 14.220 acc5_pos: 100.000 loss: 0.1908 (0.7006) acc1: 100.0000 (59.8935) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2609 data: 0.1649 max mem: 3531 +Test: [120/517] eta: 0:01:54 acc1_pos: 14.220 acc5_pos: 100.000 loss: 0.1908 (0.6623) acc1: 99.7396 (63.0897) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.3099 data: 0.2108 max mem: 3531 +Test: [130/517] eta: 0:01:55 acc1_pos: 14.220 acc5_pos: 100.000 loss: 0.2410 (0.6300) acc1: 98.4375 (65.7343) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.3889 data: 0.2882 max mem: 3531 +Test: [140/517] eta: 0:01:51 acc1_pos: 14.220 acc5_pos: 100.000 loss: 0.2187 (0.5998) acc1: 99.7396 (68.1590) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.3360 data: 0.2401 max mem: 3531 +Test: [150/517] eta: 0:01:47 acc1_pos: 14.220 acc5_pos: 100.000 loss: 0.1910 (0.5724) acc1: 100.0000 (70.2608) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2549 data: 0.1604 max mem: 3531 +Test: [160/517] eta: 0:01:43 acc1_pos: 14.220 acc5_pos: 100.000 loss: 0.1946 (0.5494) acc1: 100.0000 (72.1031) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2538 data: 0.1574 max mem: 3531 +Test: [170/517] eta: 0:01:39 acc1_pos: 14.220 acc5_pos: 100.000 loss: 0.2040 (0.5294) acc1: 100.0000 (73.6964) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2549 data: 0.1578 max mem: 3531 +Test: [180/517] eta: 0:01:35 acc1_pos: 14.220 acc5_pos: 100.000 loss: 0.1946 (0.5103) acc1: 99.7396 (75.1324) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2399 data: 0.1443 max mem: 3531 +Test: [190/517] eta: 0:01:32 acc1_pos: 14.220 acc5_pos: 100.000 loss: 0.1326 (0.4898) acc1: 100.0000 (76.4343) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2349 data: 0.1379 max mem: 3531 +Test: [200/517] eta: 0:01:29 acc1_pos: 14.220 acc5_pos: 100.000 loss: 0.1317 (0.4735) acc1: 100.0000 (77.6003) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2649 data: 0.1698 max mem: 3531 +Test: [210/517] eta: 0:01:26 acc1_pos: 14.220 acc5_pos: 100.000 loss: 0.1912 (0.4637) acc1: 99.2188 (78.4730) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2599 data: 0.1655 max mem: 3532 +Test: [220/517] eta: 0:01:23 acc1_pos: 14.220 acc5_pos: 100.000 loss: 0.2782 (0.4557) acc1: 95.5729 (79.2279) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2557 data: 0.1618 max mem: 3532 +Test: [230/517] eta: 0:01:19 acc1_pos: 14.220 acc5_pos: 100.000 loss: 0.2587 (0.4461) acc1: 97.3958 (80.0967) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2449 data: 0.1534 max mem: 3532 +Test: [240/517] eta: 0:01:16 acc1_pos: 14.220 acc5_pos: 100.000 loss: 0.1818 (0.4341) acc1: 100.0000 (80.9215) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2341 data: 0.1361 max mem: 3532 +Test: [250/517] eta: 0:01:13 acc1_pos: 14.220 acc5_pos: 100.000 loss: 0.1708 (0.4235) acc1: 100.0000 (81.6754) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2453 data: 0.1443 max mem: 3532 +Test: [260/517] eta: 0:01:10 acc1_pos: 14.220 acc5_pos: 100.000 loss: 0.1741 (0.4143) acc1: 99.7396 (82.3715) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2701 data: 0.1754 max mem: 3532 +Test: [270/517] eta: 0:01:07 acc1_pos: 14.220 acc5_pos: 100.000 loss: 0.1856 (0.4060) acc1: 100.0000 (83.0210) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2796 data: 0.1855 max mem: 3532 +Test: [280/517] eta: 0:01:04 acc1_pos: 14.220 acc5_pos: 100.000 loss: 0.1845 (0.3974) acc1: 100.0000 (83.6234) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2420 data: 0.1477 max mem: 3532 +Test: [290/517] eta: 0:01:01 acc1_pos: 14.220 acc5_pos: 100.000 loss: 0.1566 (0.3892) acc1: 100.0000 (84.1853) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2399 data: 0.1482 max mem: 3532 +Test: [300/517] eta: 0:00:59 acc1_pos: 14.220 acc5_pos: 100.000 loss: 0.1509 (0.3815) acc1: 100.0000 (84.7098) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2941 data: 0.2012 max mem: 3532 +Test: [310/517] eta: 0:00:56 acc1_pos: 14.220 acc5_pos: 100.000 loss: 0.2032 (0.3768) acc1: 100.0000 (85.1914) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2714 data: 0.1792 max mem: 3532 +Test: [320/517] eta: 0:00:53 acc1_pos: 14.220 acc5_pos: 100.000 loss: 0.2153 (0.3710) acc1: 99.7396 (85.6479) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2385 data: 0.1467 max mem: 3532 +Test: [330/517] eta: 0:00:51 acc1_pos: 14.220 acc5_pos: 100.000 loss: 0.1653 (0.3646) acc1: 100.0000 (86.0815) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.3038 data: 0.2094 max mem: 3532 +Test: [340/517] eta: 0:00:49 acc1_pos: 14.220 acc5_pos: 100.000 loss: 0.1526 (0.3589) acc1: 100.0000 (86.4874) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.3599 data: 0.2653 max mem: 3532 +Test: [350/517] eta: 0:00:45 acc1_pos: 14.220 acc5_pos: 100.000 loss: 0.2000 (0.3550) acc1: 100.0000 (86.8545) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2895 data: 0.1911 max mem: 3532 +Test: [360/517] eta: 0:00:42 acc1_pos: 14.220 acc5_pos: 100.000 loss: 0.2169 (0.3512) acc1: 99.4792 (87.2014) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2099 data: 0.1096 max mem: 3532 +Test: [370/517] eta: 0:00:40 acc1_pos: 14.220 acc5_pos: 100.000 loss: 0.1929 (0.3463) acc1: 100.0000 (87.5456) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2299 data: 0.1316 max mem: 3532 +Test: [380/517] eta: 0:00:37 acc1_pos: 14.220 acc5_pos: 100.000 loss: 0.1736 (0.3423) acc1: 100.0000 (87.8657) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2349 data: 0.1375 max mem: 3532 +Test: [390/517] eta: 0:00:34 acc1_pos: 14.220 acc5_pos: 100.000 loss: 0.1872 (0.3383) acc1: 100.0000 (88.1747) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2350 data: 0.1390 max mem: 3532 +Test: [400/517] eta: 0:00:31 acc1_pos: 14.220 acc5_pos: 100.000 loss: 0.1881 (0.3351) acc1: 100.0000 (88.4696) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2499 data: 0.1528 max mem: 3532 +Test: [410/517] eta: 0:00:28 acc1_pos: 14.220 acc5_pos: 100.000 loss: 0.2259 (0.3325) acc1: 100.0000 (88.7438) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2548 data: 0.1563 max mem: 3532 +Test: [420/517] eta: 0:00:26 acc1_pos: 14.220 acc5_pos: 100.000 loss: 0.2063 (0.3293) acc1: 100.0000 (89.0105) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2499 data: 0.1523 max mem: 3532 +Test: [430/517] eta: 0:00:23 acc1_pos: 14.220 acc5_pos: 100.000 loss: 0.2046 (0.3270) acc1: 100.0000 (89.2546) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2349 data: 0.1377 max mem: 3532 +Test: [440/517] eta: 0:00:20 acc1_pos: 14.220 acc5_pos: 100.000 loss: 0.2205 (0.3246) acc1: 99.7396 (89.4894) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2402 data: 0.1436 max mem: 3533 +Test: [450/517] eta: 0:00:17 acc1_pos: 14.220 acc5_pos: 100.000 loss: 0.2030 (0.3216) acc1: 100.0000 (89.7225) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2710 data: 0.1756 max mem: 3533 +Test: [460/517] eta: 0:00:15 acc1_pos: 14.220 acc5_pos: 100.000 loss: 0.2030 (0.3195) acc1: 100.0000 (89.9149) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2596 data: 0.1643 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 14.220 acc5_pos: 100.000 loss: 0.2030 (0.3169) acc1: 99.4792 (90.1196) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2488 data: 0.1529 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 14.220 acc5_pos: 100.000 loss: 0.1957 (0.3141) acc1: 99.7396 (90.3240) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2349 data: 0.1385 max mem: 3533 +Test: [490/517] eta: 0:00:07 acc1_pos: 14.220 acc5_pos: 100.000 loss: 0.1738 (0.3116) acc1: 100.0000 (90.5168) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2062 data: 0.1101 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 14.220 acc5_pos: 100.000 loss: 0.2139 (0.3098) acc1: 100.0000 (90.6999) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1924 data: 0.0981 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 14.220 acc5_pos: 100.000 loss: 0.2248 (0.3078) acc1: 99.7396 (90.8564) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1774 data: 0.0861 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 14.220 acc5_pos: 100.000 loss: 0.1622 (0.3058) acc1: 100.0000 (90.9579) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1773 data: 0.0860 max mem: 3533 +Test: Total time: 0:02:14 (0.2603 s / it) +* Overall Acc@1 90.958 Acc@5 100.000 + target == real Acc@1 9.340 Acc@5 100.000 + target == fake Acc@1 99.425 Acc@5 100.000 + loss 0.306 + PR-AUC 0.9687 ROC-AUC 0.7900 EER 28.27% (thr=0.7954) +Accuracy of the network on the 198427 test images: 91.0% +[rank0]:[W725 16:49:33.020088030 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0725 16:49:55.800000 85824 torch/distributed/run.py:766] +W0725 16:49:55.800000 85824 torch/distributed/run.py:766] ***************************************** +W0725 16:49:55.800000 85824 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 16:49:55.800000 85824 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 16:50:28.284688184 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 16:50:28.749744013 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 16:50:29.191236765 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 16:50:29.310974352 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 16:50:29.473021943 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 16:50:29.473039378 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 16:50:29.501128402 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank4]:[W725 16:50:29.510693446 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n22_seed2_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed2_resampled_n116-041-188_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:31:25 acc1_pos: 87.240 acc5_pos: 100.000 loss: 0.4438 (0.4438) acc1: 87.2396 (87.2396) acc5: 100.0000 (100.0000) time: 4.8471 data: 2.9206 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/389] eta: 0:03:18 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.4438 (0.4841) acc1: 82.0312 (74.5502) acc5: 100.0000 (100.0000) time: 0.5226 data: 0.2659 max mem: 3531 +Test: [ 20/389] eta: 0:01:59 acc1_pos: 36.458 acc5_pos: 100.000 loss: 0.4872 (0.5614) acc1: 62.7604 (66.9767) acc5: 100.0000 (100.0000) time: 0.0985 data: 0.0088 max mem: 3531 +Test: [ 30/389] eta: 0:01:37 acc1_pos: 79.167 acc5_pos: 100.000 loss: 0.6336 (0.5653) acc1: 63.0208 (68.9264) acc5: 100.0000 (100.0000) time: 0.1332 data: 0.0357 max mem: 3531 +Test: [ 40/389] eta: 0:01:33 acc1_pos: 48.698 acc5_pos: 100.000 loss: 0.5607 (0.5683) acc1: 70.3125 (69.0422) acc5: 100.0000 (100.0000) time: 0.2062 data: 0.1078 max mem: 3531 +Test: [ 50/389] eta: 0:01:29 acc1_pos: 95.833 acc5_pos: 100.000 loss: 0.5616 (0.5665) acc1: 69.7917 (68.7602) acc5: 100.0000 (100.0000) time: 0.2507 data: 0.1595 max mem: 3531 +Test: [ 60/389] eta: 0:01:20 acc1_pos: 76.042 acc5_pos: 100.000 loss: 0.3336 (0.5235) acc1: 92.9688 (72.9380) acc5: 100.0000 (100.0000) time: 0.1952 data: 0.1032 max mem: 3531 +Test: [ 70/389] eta: 0:01:15 acc1_pos: 85.938 acc5_pos: 100.000 loss: 0.3375 (0.5109) acc1: 90.6250 (74.1747) acc5: 100.0000 (100.0000) time: 0.1641 data: 0.0695 max mem: 3531 +Test: [ 80/389] eta: 0:01:10 acc1_pos: 83.594 acc5_pos: 100.000 loss: 0.4209 (0.5212) acc1: 80.7292 (73.0292) acc5: 100.0000 (100.0000) time: 0.1845 data: 0.0897 max mem: 3531 +Test: [ 90/389] eta: 0:01:06 acc1_pos: 78.906 acc5_pos: 100.000 loss: 0.4826 (0.5131) acc1: 76.0417 (73.5978) acc5: 100.0000 (100.0000) time: 0.1749 data: 0.0829 max mem: 3531 +Test: [100/389] eta: 0:01:02 acc1_pos: 94.792 acc5_pos: 100.000 loss: 0.4718 (0.5194) acc1: 79.9479 (73.2570) acc5: 100.0000 (100.0000) time: 0.1614 data: 0.0695 max mem: 3531 +Test: [110/389] eta: 0:00:59 acc1_pos: 99.740 acc5_pos: 100.000 loss: 0.4643 (0.5279) acc1: 79.9479 (72.3982) acc5: 100.0000 (100.0000) time: 0.1673 data: 0.0753 max mem: 3531 +Test: [120/389] eta: 0:00:55 acc1_pos: 79.688 acc5_pos: 100.000 loss: 0.4297 (0.5185) acc1: 82.8125 (73.4289) acc5: 100.0000 (100.0000) time: 0.1684 data: 0.0756 max mem: 3531 +Test: [130/389] eta: 0:00:52 acc1_pos: 77.914 acc5_pos: 100.000 loss: 0.4609 (0.5273) acc1: 79.6875 (72.7656) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.1491 data: 0.0557 max mem: 3531 +Test: [140/389] eta: 0:00:49 acc1_pos: 77.914 acc5_pos: 100.000 loss: 0.6204 (0.5366) acc1: 65.3646 (71.9821) acc5: 100.0000 (100.0000) acc1_neg: 66.146 acc5_neg: 100.000 time: 0.1386 data: 0.0453 max mem: 3531 +Test: [150/389] eta: 0:00:46 acc1_pos: 77.914 acc5_pos: 100.000 loss: 0.6622 (0.5510) acc1: 59.8958 (70.7764) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1433 data: 0.0507 max mem: 3531 +Test: [160/389] eta: 0:00:44 acc1_pos: 77.914 acc5_pos: 100.000 loss: 0.5587 (0.5475) acc1: 64.8438 (70.9579) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1575 data: 0.0670 max mem: 3531 +Test: [170/389] eta: 0:00:42 acc1_pos: 77.914 acc5_pos: 100.000 loss: 0.2633 (0.5401) acc1: 99.2188 (71.4577) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.1698 data: 0.0791 max mem: 3531 +Test: [180/389] eta: 0:00:39 acc1_pos: 77.914 acc5_pos: 100.000 loss: 0.3783 (0.5462) acc1: 76.5625 (71.0679) acc5: 100.0000 (100.0000) acc1_neg: 42.708 acc5_neg: 100.000 time: 0.1643 data: 0.0702 max mem: 3531 +Test: [190/389] eta: 0:00:37 acc1_pos: 77.914 acc5_pos: 100.000 loss: 0.7241 (0.5634) acc1: 54.1667 (69.5190) acc5: 100.0000 (100.0000) acc1_neg: 2.865 acc5_neg: 100.000 time: 0.1558 data: 0.0606 max mem: 3531 +Test: [200/389] eta: 0:00:35 acc1_pos: 77.914 acc5_pos: 100.000 loss: 0.8349 (0.5795) acc1: 33.8542 (68.0335) acc5: 100.0000 (100.0000) acc1_neg: 33.333 acc5_neg: 100.000 time: 0.1553 data: 0.0629 max mem: 3531 +Test: [210/389] eta: 0:00:33 acc1_pos: 77.914 acc5_pos: 100.000 loss: 0.6330 (0.5735) acc1: 36.7188 (68.5093) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1518 data: 0.0600 max mem: 3532 +Test: [220/389] eta: 0:00:31 acc1_pos: 77.914 acc5_pos: 100.000 loss: 0.3207 (0.5726) acc1: 95.8333 (68.6522) acc5: 100.0000 (100.0000) acc1_neg: 17.448 acc5_neg: 100.000 time: 0.1592 data: 0.0678 max mem: 3532 +Test: [230/389] eta: 0:00:28 acc1_pos: 77.914 acc5_pos: 100.000 loss: 0.5227 (0.5854) acc1: 65.3646 (67.6790) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.1549 data: 0.0641 max mem: 3532 +Test: [240/389] eta: 0:00:26 acc1_pos: 77.914 acc5_pos: 100.000 loss: 0.6866 (0.5869) acc1: 52.3438 (67.5635) acc5: 100.0000 (100.0000) acc1_neg: 46.875 acc5_neg: 100.000 time: 0.1432 data: 0.0517 max mem: 3532 +Test: [250/389] eta: 0:00:24 acc1_pos: 77.914 acc5_pos: 100.000 loss: 0.4865 (0.5834) acc1: 77.6042 (67.9988) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1494 data: 0.0578 max mem: 3532 +Test: [260/389] eta: 0:00:23 acc1_pos: 77.914 acc5_pos: 100.000 loss: 0.4749 (0.5872) acc1: 75.5208 (67.9518) acc5: 100.0000 (100.0000) acc1_neg: 11.458 acc5_neg: 100.000 time: 0.1580 data: 0.0661 max mem: 3532 +Test: [270/389] eta: 0:00:21 acc1_pos: 77.914 acc5_pos: 100.000 loss: 0.7681 (0.6025) acc1: 41.4062 (66.7291) acc5: 100.0000 (100.0000) acc1_neg: 40.625 acc5_neg: 100.000 time: 0.1606 data: 0.0693 max mem: 3532 +Test: [280/389] eta: 0:00:19 acc1_pos: 77.914 acc5_pos: 100.000 loss: 0.8527 (0.6157) acc1: 36.1979 (65.7344) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1668 data: 0.0754 max mem: 3532 +Test: [290/389] eta: 0:00:17 acc1_pos: 77.914 acc5_pos: 100.000 loss: 0.8000 (0.6121) acc1: 42.7083 (66.0912) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1719 data: 0.0791 max mem: 3532 +Test: [300/389] eta: 0:00:15 acc1_pos: 77.914 acc5_pos: 100.000 loss: 0.5100 (0.6122) acc1: 77.0833 (66.2523) acc5: 100.0000 (100.0000) acc1_neg: 74.740 acc5_neg: 100.000 time: 0.1549 data: 0.0617 max mem: 3532 +Test: [310/389] eta: 0:00:13 acc1_pos: 77.914 acc5_pos: 100.000 loss: 0.6863 (0.6253) acc1: 56.5104 (65.4182) acc5: 100.0000 (100.0000) acc1_neg: 1.562 acc5_neg: 100.000 time: 0.1475 data: 0.0543 max mem: 3532 +Test: [320/389] eta: 0:00:12 acc1_pos: 77.914 acc5_pos: 100.000 loss: 0.8774 (0.6299) acc1: 42.4479 (65.0733) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1554 data: 0.0623 max mem: 3532 +Test: [330/389] eta: 0:00:10 acc1_pos: 77.914 acc5_pos: 100.000 loss: 1.1093 (0.6463) acc1: 9.8958 (63.5802) acc5: 100.0000 (100.0000) acc1_neg: 32.552 acc5_neg: 100.000 time: 0.1550 data: 0.0633 max mem: 3532 +Test: [340/389] eta: 0:00:08 acc1_pos: 77.914 acc5_pos: 100.000 loss: 0.4938 (0.6364) acc1: 64.3229 (64.4963) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1584 data: 0.0673 max mem: 3532 +Test: [350/389] eta: 0:00:06 acc1_pos: 77.914 acc5_pos: 100.000 loss: 0.3067 (0.6442) acc1: 96.0938 (64.0506) acc5: 100.0000 (100.0000) acc1_neg: 20.312 acc5_neg: 100.000 time: 0.1583 data: 0.0664 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 77.914 acc5_pos: 100.000 loss: 0.5789 (0.6460) acc1: 71.8750 (63.9716) acc5: 100.0000 (100.0000) acc1_neg: 77.083 acc5_neg: 100.000 time: 0.1562 data: 0.0647 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 77.914 acc5_pos: 100.000 loss: 0.5649 (0.6473) acc1: 72.6562 (63.9249) acc5: 100.0000 (100.0000) acc1_neg: 78.646 acc5_neg: 100.000 time: 0.1553 data: 0.0637 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 77.914 acc5_pos: 100.000 loss: 0.6179 (0.6498) acc1: 69.5312 (63.6920) acc5: 100.0000 (100.0000) acc1_neg: 56.250 acc5_neg: 100.000 time: 0.1555 data: 0.0621 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 77.914 acc5_pos: 100.000 loss: 0.6179 (0.6484) acc1: 69.5312 (63.7937) acc5: 100.0000 (100.0000) acc1_neg: 78.962 acc5_neg: 100.000 time: 0.1740 data: 0.0699 max mem: 3532 +Test: Total time: 0:01:07 (0.1732 s / it) +* Overall Acc@1 63.794 Acc@5 100.000 + target == real Acc@1 72.900 Acc@5 100.000 + target == fake Acc@1 59.561 Acc@5 100.000 + loss 0.648 + PR-AUC 0.8567 ROC-AUC 0.7261 EER 34.24% (thr=0.4421) +Accuracy of the network on the 149358 test images: 63.8% +[rank0]:[W725 16:51:52.214563348 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 16:52:12.749000 86380 torch/distributed/run.py:766] +W0725 16:52:12.749000 86380 torch/distributed/run.py:766] ***************************************** +W0725 16:52:12.749000 86380 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 16:52:12.749000 86380 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 16:52:44.658511700 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 16:52:45.674530011 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 16:52:45.958513793 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 16:52:45.986141047 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 16:52:46.103178923 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W725 16:52:46.103250969 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 16:52:46.235598124 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 16:52:46.264927542 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n22_seed2_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed2_resampled_n116-041-188_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:31:04 acc1_pos: 68.229 acc5_pos: 100.000 loss: 0.5543 (0.5543) acc1: 68.2292 (68.2292) acc5: 100.0000 (100.0000) time: 5.5996 data: 3.0691 max mem: 3531 +Test: [ 10/333] eta: 0:03:10 acc1_pos: 82.812 acc5_pos: 100.000 loss: 0.4422 (0.4276) acc1: 77.3438 (78.7169) acc5: 100.0000 (100.0000) time: 0.5910 data: 0.2794 max mem: 3531 +Test: [ 20/333] eta: 0:01:51 acc1_pos: 90.104 acc5_pos: 100.000 loss: 0.3833 (0.3944) acc1: 82.2917 (81.1756) acc5: 100.0000 (100.0000) time: 0.0934 data: 0.0004 max mem: 3531 +Test: [ 30/333] eta: 0:01:37 acc1_pos: 41.927 acc5_pos: 100.000 loss: 0.4804 (0.5878) acc1: 71.0938 (66.4231) acc5: 100.0000 (100.0000) time: 0.1750 data: 0.0738 max mem: 3531 +Test: [ 40/333] eta: 0:01:31 acc1_pos: 28.906 acc5_pos: 100.000 loss: 1.0297 (0.6908) acc1: 30.9896 (57.8760) acc5: 100.0000 (100.0000) time: 0.2673 data: 0.1661 max mem: 3531 +Test: [ 50/333] eta: 0:01:21 acc1_pos: 34.115 acc5_pos: 100.000 loss: 1.0236 (0.7572) acc1: 29.6875 (52.2212) acc5: 100.0000 (100.0000) time: 0.2328 data: 0.1396 max mem: 3531 +Test: [ 60/333] eta: 0:01:13 acc1_pos: 23.177 acc5_pos: 100.000 loss: 0.9981 (0.8013) acc1: 31.7708 (48.9071) acc5: 100.0000 (100.0000) time: 0.1844 data: 0.0926 max mem: 3531 +Test: [ 70/333] eta: 0:01:07 acc1_pos: 27.865 acc5_pos: 100.000 loss: 1.0073 (0.8329) acc1: 31.7708 (46.6806) acc5: 100.0000 (100.0000) time: 0.1830 data: 0.0908 max mem: 3531 +Test: [ 80/333] eta: 0:01:03 acc1_pos: 38.281 acc5_pos: 100.000 loss: 0.9896 (0.8510) acc1: 34.1146 (45.3543) acc5: 100.0000 (100.0000) time: 0.1885 data: 0.0965 max mem: 3531 +Test: [ 90/333] eta: 0:00:59 acc1_pos: 51.302 acc5_pos: 100.000 loss: 0.9221 (0.8517) acc1: 38.5417 (45.1093) acc5: 100.0000 (100.0000) time: 0.1935 data: 0.1014 max mem: 3531 +Test: [100/333] eta: 0:00:55 acc1_pos: 47.917 acc5_pos: 100.000 loss: 0.8225 (0.8521) acc1: 40.6250 (44.6963) acc5: 100.0000 (100.0000) time: 0.1911 data: 0.1003 max mem: 3531 +Test: [110/333] eta: 0:00:54 acc1_pos: 48.698 acc5_pos: 100.000 loss: 0.8448 (0.8520) acc1: 40.3646 (44.6486) acc5: 100.0000 (100.0000) time: 0.2435 data: 0.1498 max mem: 3531 +Test: [120/333] eta: 0:00:50 acc1_pos: 47.135 acc5_pos: 100.000 loss: 0.8104 (0.8513) acc1: 43.2292 (44.4322) acc5: 100.0000 (100.0000) time: 0.2232 data: 0.1255 max mem: 3531 +Test: [130/333] eta: 0:00:46 acc1_pos: 59.375 acc5_pos: 100.000 loss: 0.7934 (0.8481) acc1: 44.0104 (44.5193) acc5: 100.0000 (100.0000) time: 0.1568 data: 0.0577 max mem: 3531 +Test: [140/333] eta: 0:00:44 acc1_pos: 57.812 acc5_pos: 100.000 loss: 0.7857 (0.8452) acc1: 46.8750 (44.6291) acc5: 100.0000 (100.0000) time: 0.1864 data: 0.0913 max mem: 3531 +Test: [150/333] eta: 0:00:41 acc1_pos: 47.396 acc5_pos: 100.000 loss: 0.8027 (0.8457) acc1: 44.2708 (44.3933) acc5: 100.0000 (100.0000) time: 0.2005 data: 0.1073 max mem: 3531 +Test: [160/333] eta: 0:00:39 acc1_pos: 38.487 acc5_pos: 100.000 loss: 0.7931 (0.8430) acc1: 42.9688 (44.7318) acc5: 100.0000 (100.0000) acc1_neg: 65.625 acc5_neg: 100.000 time: 0.2311 data: 0.1365 max mem: 3531 +Test: [170/333] eta: 0:00:36 acc1_pos: 38.487 acc5_pos: 100.000 loss: 0.7131 (0.8268) acc1: 61.9792 (46.4364) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.2290 data: 0.1351 max mem: 3531 +Test: [180/333] eta: 0:00:34 acc1_pos: 38.487 acc5_pos: 100.000 loss: 0.5044 (0.8081) acc1: 78.3854 (48.2907) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.1788 data: 0.0866 max mem: 3531 +Test: [190/333] eta: 0:00:32 acc1_pos: 38.487 acc5_pos: 100.000 loss: 0.4694 (0.7895) acc1: 80.9896 (50.0900) acc5: 100.0000 (100.0000) acc1_neg: 76.042 acc5_neg: 100.000 time: 0.2090 data: 0.1184 max mem: 3531 +Test: [200/333] eta: 0:00:29 acc1_pos: 38.487 acc5_pos: 100.000 loss: 0.4518 (0.7737) acc1: 80.9896 (51.6130) acc5: 100.0000 (100.0000) acc1_neg: 77.344 acc5_neg: 100.000 time: 0.1998 data: 0.1083 max mem: 3531 +Test: [210/333] eta: 0:00:27 acc1_pos: 38.487 acc5_pos: 100.000 loss: 0.5011 (0.7639) acc1: 77.3438 (52.6461) acc5: 100.0000 (100.0000) acc1_neg: 57.031 acc5_neg: 100.000 time: 0.1703 data: 0.0782 max mem: 3532 +Test: [220/333] eta: 0:00:24 acc1_pos: 38.487 acc5_pos: 100.000 loss: 0.5938 (0.7591) acc1: 66.9271 (53.2393) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.2042 data: 0.1129 max mem: 3532 +Test: [230/333] eta: 0:00:22 acc1_pos: 38.487 acc5_pos: 100.000 loss: 0.3639 (0.7410) acc1: 86.1979 (54.7969) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1949 data: 0.1039 max mem: 3532 +Test: [240/333] eta: 0:00:20 acc1_pos: 38.487 acc5_pos: 100.000 loss: 0.3628 (0.7269) acc1: 88.0208 (56.1225) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.1991 data: 0.1028 max mem: 3532 +Test: [250/333] eta: 0:00:18 acc1_pos: 38.487 acc5_pos: 100.000 loss: 0.4229 (0.7181) acc1: 83.3333 (57.0333) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.2098 data: 0.1138 max mem: 3532 +Test: [260/333] eta: 0:00:15 acc1_pos: 38.487 acc5_pos: 100.000 loss: 0.5469 (0.7119) acc1: 75.0000 (57.6598) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.1869 data: 0.0950 max mem: 3532 +Test: [270/333] eta: 0:00:13 acc1_pos: 38.487 acc5_pos: 100.000 loss: 0.4610 (0.7009) acc1: 79.4271 (58.6504) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1917 data: 0.0960 max mem: 3532 +Test: [280/333] eta: 0:00:11 acc1_pos: 38.487 acc5_pos: 100.000 loss: 0.4153 (0.6947) acc1: 82.8125 (59.2471) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.1957 data: 0.1012 max mem: 3532 +Test: [290/333] eta: 0:00:09 acc1_pos: 38.487 acc5_pos: 100.000 loss: 0.6187 (0.6924) acc1: 67.4479 (59.5262) acc5: 100.0000 (100.0000) acc1_neg: 65.104 acc5_neg: 100.000 time: 0.1848 data: 0.0932 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 38.487 acc5_pos: 100.000 loss: 0.6414 (0.6917) acc1: 66.4062 (59.7514) acc5: 100.0000 (100.0000) acc1_neg: 68.750 acc5_neg: 100.000 time: 0.1825 data: 0.0898 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 38.487 acc5_pos: 100.000 loss: 0.5826 (0.6833) acc1: 68.7500 (60.5155) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.1684 data: 0.0764 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 38.487 acc5_pos: 100.000 loss: 0.4067 (0.6761) acc1: 85.9375 (61.2336) acc5: 100.0000 (100.0000) acc1_neg: 63.021 acc5_neg: 100.000 time: 0.1380 data: 0.0469 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 38.487 acc5_pos: 100.000 loss: 0.6759 (0.6783) acc1: 64.5833 (61.2121) acc5: 100.0000 (100.0000) acc1_neg: 64.583 acc5_neg: 100.000 time: 0.1173 data: 0.0272 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 38.487 acc5_pos: 100.000 loss: 0.7155 (0.6788) acc1: 61.9792 (61.2048) acc5: 100.0000 (100.0000) acc1_neg: 57.746 acc5_neg: 100.000 time: 0.1166 data: 0.0233 max mem: 3532 +Test: Total time: 0:01:08 (0.2047 s / it) +* Overall Acc@1 61.205 Acc@5 100.000 + target == real Acc@1 44.303 Acc@5 100.000 + target == fake Acc@1 76.250 Acc@5 100.000 + loss 0.679 + PR-AUC 0.6726 ROC-AUC 0.6576 EER 38.97% (thr=0.6143) +Accuracy of the network on the 127843 test images: 61.2% +[rank0]:[W725 16:54:04.073306787 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 16:54:24.579000 86939 torch/distributed/run.py:766] +W0725 16:54:24.579000 86939 torch/distributed/run.py:766] ***************************************** +W0725 16:54:24.579000 86939 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 16:54:24.579000 86939 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 16:54:59.481780023 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 16:54:59.588757069 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 16:54:59.853692976 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 16:54:59.919757789 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 16:54:59.002631609 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +[rank0]:[W725 16:55:00.064519034 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 16:55:00.077980826 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 16:55:00.079249399 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n22_seed2_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed2_resampled_n116-041-188_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:19:29 acc1_pos: 96.615 acc5_pos: 100.000 loss: 0.1714 (0.1714) acc1: 96.6146 (96.6146) acc5: 100.0000 (100.0000) time: 4.9766 data: 3.2768 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/235] eta: 0:02:00 acc1_pos: 96.615 acc5_pos: 100.000 loss: 0.1714 (0.1719) acc1: 96.6146 (96.9224) acc5: 100.0000 (100.0000) time: 0.5343 data: 0.2983 max mem: 3531 +Test: [ 20/235] eta: 0:01:09 acc1_pos: 96.615 acc5_pos: 100.000 loss: 0.1650 (0.1678) acc1: 96.8750 (96.9742) acc5: 100.0000 (100.0000) time: 0.0900 data: 0.0005 max mem: 3531 +Test: [ 30/235] eta: 0:00:51 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.1638 (0.1674) acc1: 96.8750 (96.9422) acc5: 100.0000 (100.0000) time: 0.0932 data: 0.0004 max mem: 3531 +Test: [ 40/235] eta: 0:00:45 acc1_pos: 95.833 acc5_pos: 100.000 loss: 0.1649 (0.1676) acc1: 97.1354 (96.9766) acc5: 100.0000 (100.0000) time: 0.1381 data: 0.0443 max mem: 3531 +Test: [ 50/235] eta: 0:00:41 acc1_pos: 96.615 acc5_pos: 100.000 loss: 0.1725 (0.1688) acc1: 96.6146 (96.8495) acc5: 100.0000 (100.0000) time: 0.1799 data: 0.0842 max mem: 3531 +Test: [ 60/235] eta: 0:00:37 acc1_pos: 95.833 acc5_pos: 100.000 loss: 0.1768 (0.1706) acc1: 96.6146 (96.8195) acc5: 100.0000 (100.0000) time: 0.1699 data: 0.0720 max mem: 3531 +Test: [ 70/235] eta: 0:00:34 acc1_pos: 98.462 acc5_pos: 100.000 loss: 0.1903 (0.3584) acc1: 94.7917 (86.7004) acc5: 100.0000 (100.0000) acc1_neg: 14.323 acc5_neg: 100.000 time: 0.1749 data: 0.0798 max mem: 3531 +Test: [ 80/235] eta: 0:00:31 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.6620 (0.5211) acc1: 16.6667 (78.0414) acc5: 100.0000 (100.0000) acc1_neg: 14.583 acc5_neg: 100.000 time: 0.1761 data: 0.0824 max mem: 3531 +Test: [ 90/235] eta: 0:00:28 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.6639 (0.6498) acc1: 16.1458 (71.1882) acc5: 100.0000 (100.0000) acc1_neg: 16.146 acc5_neg: 100.000 time: 0.1649 data: 0.0705 max mem: 3531 +Test: [100/235] eta: 0:00:26 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.6724 (0.7499) acc1: 16.1458 (65.7849) acc5: 100.0000 (100.0000) acc1_neg: 15.625 acc5_neg: 100.000 time: 0.1687 data: 0.0721 max mem: 3531 +Test: [110/235] eta: 0:00:24 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.6605 (0.8334) acc1: 16.4062 (61.3410) acc5: 100.0000 (100.0000) acc1_neg: 17.448 acc5_neg: 100.000 time: 0.1711 data: 0.0759 max mem: 3531 +Test: [120/235] eta: 0:00:22 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.6187 (0.8953) acc1: 14.8438 (57.4466) acc5: 100.0000 (100.0000) acc1_neg: 13.802 acc5_neg: 100.000 time: 0.1705 data: 0.0790 max mem: 3531 +Test: [130/235] eta: 0:00:20 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.5647 (0.9455) acc1: 14.5833 (54.2144) acc5: 100.0000 (100.0000) acc1_neg: 14.583 acc5_neg: 100.000 time: 0.1887 data: 0.0975 max mem: 3531 +Test: [140/235] eta: 0:00:18 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.5772 (0.9916) acc1: 14.5833 (51.3852) acc5: 100.0000 (100.0000) acc1_neg: 11.979 acc5_neg: 100.000 time: 0.1993 data: 0.1068 max mem: 3531 +Test: [150/235] eta: 0:00:16 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.5749 (1.0287) acc1: 14.8438 (48.9963) acc5: 100.0000 (100.0000) acc1_neg: 14.844 acc5_neg: 100.000 time: 0.1799 data: 0.0840 max mem: 3531 +Test: [160/235] eta: 0:00:13 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.5519 (1.0620) acc1: 15.1042 (46.9090) acc5: 100.0000 (100.0000) acc1_neg: 15.365 acc5_neg: 100.000 time: 0.1360 data: 0.0398 max mem: 3531 +Test: [170/235] eta: 0:00:11 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.5846 (1.0951) acc1: 15.3646 (45.1343) acc5: 100.0000 (100.0000) acc1_neg: 15.365 acc5_neg: 100.000 time: 0.1309 data: 0.0382 max mem: 3531 +Test: [180/235] eta: 0:00:10 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.6472 (1.1258) acc1: 16.9271 (43.5428) acc5: 100.0000 (100.0000) acc1_neg: 16.927 acc5_neg: 100.000 time: 0.1637 data: 0.0695 max mem: 3531 +Test: [190/235] eta: 0:00:08 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.6282 (1.1524) acc1: 16.4062 (42.1261) acc5: 100.0000 (100.0000) acc1_neg: 14.062 acc5_neg: 100.000 time: 0.1739 data: 0.0798 max mem: 3531 +Test: [200/235] eta: 0:00:06 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.6311 (1.1777) acc1: 15.1042 (40.7934) acc5: 100.0000 (100.0000) acc1_neg: 13.281 acc5_neg: 100.000 time: 0.1799 data: 0.0851 max mem: 3531 +Test: [210/235] eta: 0:00:04 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.6365 (1.1997) acc1: 15.3646 (39.6080) acc5: 100.0000 (100.0000) acc1_neg: 16.667 acc5_neg: 100.000 time: 0.1749 data: 0.0789 max mem: 3532 +Test: [220/235] eta: 0:00:02 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.6397 (1.2199) acc1: 15.3646 (38.5275) acc5: 100.0000 (100.0000) acc1_neg: 14.844 acc5_neg: 100.000 time: 0.1749 data: 0.0798 max mem: 3532 +Test: [230/235] eta: 0:00:00 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.6405 (1.2379) acc1: 16.1458 (37.5992) acc5: 100.0000 (100.0000) acc1_neg: 16.927 acc5_neg: 100.000 time: 0.1649 data: 0.0673 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.6460 (1.2454) acc1: 16.4062 (37.3090) acc5: 100.0000 (100.0000) acc1_neg: 21.739 acc5_neg: 100.000 time: 0.1200 data: 0.0256 max mem: 3532 +Test: Total time: 0:00:42 (0.1804 s / it) +* Overall Acc@1 37.309 Acc@5 100.000 + target == real Acc@1 96.842 Acc@5 100.000 + target == fake Acc@1 15.777 Acc@5 100.000 + loss 1.245 + PR-AUC 0.8656 ROC-AUC 0.7115 EER 34.65% (thr=0.1315) +Accuracy of the network on the 89879 test images: 37.3% +[rank0]:[W725 16:55:59.966447847 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 16:56:20.725000 87495 torch/distributed/run.py:766] +W0725 16:56:20.725000 87495 torch/distributed/run.py:766] ***************************************** +W0725 16:56:20.725000 87495 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 16:56:20.725000 87495 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 16:56:55.180101179 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 16:56:55.245329831 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 16:56:55.473623941 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 16:56:55.533380303 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 16:56:55.873114226 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 16:56:55.938660945 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W725 16:56:55.938747691 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 16:56:55.976931095 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n22_seed2_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n22_classseed2_resampled_n116-041-188_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:28:33 acc1_pos: 32.292 acc5_pos: 100.000 loss: 0.9539 (0.9539) acc1: 32.2917 (32.2917) acc5: 100.0000 (100.0000) time: 6.3712 data: 4.6998 max mem: 3531 +Test: [ 10/269] eta: 0:02:53 acc1_pos: 54.427 acc5_pos: 100.000 loss: 0.7773 (0.7535) acc1: 48.9583 (50.9706) acc5: 100.0000 (100.0000) time: 0.6682 data: 0.4276 max mem: 3531 +Test: [ 20/269] eta: 0:01:38 acc1_pos: 55.990 acc5_pos: 100.000 loss: 0.7011 (0.7285) acc1: 48.9583 (53.5218) acc5: 100.0000 (100.0000) time: 0.0989 data: 0.0004 max mem: 3531 +Test: [ 30/269] eta: 0:01:12 acc1_pos: 48.698 acc5_pos: 100.000 loss: 0.6624 (0.7095) acc1: 58.3333 (55.6452) acc5: 100.0000 (100.0000) time: 0.0999 data: 0.0004 max mem: 3531 +Test: [ 40/269] eta: 0:01:01 acc1_pos: 43.103 acc5_pos: 100.000 loss: 0.4870 (0.6334) acc1: 75.2604 (63.1606) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1299 data: 0.0350 max mem: 3531 +Test: [ 50/269] eta: 0:00:53 acc1_pos: 43.103 acc5_pos: 100.000 loss: 0.3900 (0.5966) acc1: 86.7188 (66.8607) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.1524 data: 0.0600 max mem: 3531 +Test: [ 60/269] eta: 0:00:47 acc1_pos: 43.103 acc5_pos: 100.000 loss: 0.4227 (0.5821) acc1: 76.5625 (68.2847) acc5: 100.0000 (100.0000) acc1_neg: 67.708 acc5_neg: 100.000 time: 0.1413 data: 0.0485 max mem: 3531 +Test: [ 70/269] eta: 0:00:43 acc1_pos: 43.103 acc5_pos: 100.000 loss: 0.5352 (0.5740) acc1: 72.6562 (69.0911) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.1538 data: 0.0605 max mem: 3531 +Test: [ 80/269] eta: 0:00:39 acc1_pos: 43.103 acc5_pos: 100.000 loss: 0.4954 (0.5630) acc1: 77.0833 (70.2514) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.1586 data: 0.0657 max mem: 3531 +Test: [ 90/269] eta: 0:00:36 acc1_pos: 43.103 acc5_pos: 100.000 loss: 0.4954 (0.5573) acc1: 77.3438 (71.1538) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.1636 data: 0.0705 max mem: 3531 +Test: [100/269] eta: 0:00:34 acc1_pos: 43.103 acc5_pos: 100.000 loss: 0.5190 (0.5548) acc1: 76.0417 (71.4109) acc5: 100.0000 (100.0000) acc1_neg: 76.042 acc5_neg: 100.000 time: 0.1799 data: 0.0849 max mem: 3531 +Test: [110/269] eta: 0:00:32 acc1_pos: 43.103 acc5_pos: 100.000 loss: 0.5190 (0.5520) acc1: 74.7396 (71.8445) acc5: 100.0000 (100.0000) acc1_neg: 68.229 acc5_neg: 100.000 time: 0.1799 data: 0.0850 max mem: 3531 +Test: [120/269] eta: 0:00:29 acc1_pos: 43.103 acc5_pos: 100.000 loss: 0.5031 (0.5455) acc1: 75.5208 (72.4023) acc5: 100.0000 (100.0000) acc1_neg: 75.521 acc5_neg: 100.000 time: 0.1799 data: 0.0833 max mem: 3531 +Test: [130/269] eta: 0:00:27 acc1_pos: 43.103 acc5_pos: 100.000 loss: 0.5137 (0.5470) acc1: 71.6146 (72.2010) acc5: 100.0000 (100.0000) acc1_neg: 58.073 acc5_neg: 100.000 time: 0.1749 data: 0.0765 max mem: 3531 +Test: [140/269] eta: 0:00:25 acc1_pos: 43.103 acc5_pos: 100.000 loss: 0.5973 (0.5517) acc1: 63.5417 (71.5906) acc5: 100.0000 (100.0000) acc1_neg: 59.896 acc5_neg: 100.000 time: 0.2099 data: 0.1102 max mem: 3531 +Test: [150/269] eta: 0:00:23 acc1_pos: 43.103 acc5_pos: 100.000 loss: 0.5939 (0.5486) acc1: 63.8021 (71.7112) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.2300 data: 0.1317 max mem: 3531 +Test: [160/269] eta: 0:00:21 acc1_pos: 43.103 acc5_pos: 100.000 loss: 0.4691 (0.5449) acc1: 74.2188 (72.0691) acc5: 100.0000 (100.0000) acc1_neg: 70.052 acc5_neg: 100.000 time: 0.1863 data: 0.0909 max mem: 3531 +Test: [170/269] eta: 0:00:19 acc1_pos: 43.103 acc5_pos: 100.000 loss: 0.4916 (0.5422) acc1: 76.3021 (72.2679) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.1498 data: 0.0544 max mem: 3531 +Test: [180/269] eta: 0:00:17 acc1_pos: 43.103 acc5_pos: 100.000 loss: 0.4483 (0.5355) acc1: 83.5938 (72.9526) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.1707 data: 0.0752 max mem: 3531 +Test: [190/269] eta: 0:00:15 acc1_pos: 43.103 acc5_pos: 100.000 loss: 0.4209 (0.5305) acc1: 84.3750 (73.4907) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.1999 data: 0.1074 max mem: 3531 +Test: [200/269] eta: 0:00:13 acc1_pos: 43.103 acc5_pos: 100.000 loss: 0.4677 (0.5293) acc1: 79.9479 (73.5930) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.2127 data: 0.1213 max mem: 3531 +Test: [210/269] eta: 0:00:11 acc1_pos: 43.103 acc5_pos: 100.000 loss: 0.5379 (0.5307) acc1: 70.8333 (73.4190) acc5: 100.0000 (100.0000) acc1_neg: 69.792 acc5_neg: 100.000 time: 0.2099 data: 0.1170 max mem: 3532 +Test: [220/269] eta: 0:00:09 acc1_pos: 43.103 acc5_pos: 100.000 loss: 0.5421 (0.5313) acc1: 69.5312 (73.3114) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1904 data: 0.0979 max mem: 3532 +Test: [230/269] eta: 0:00:07 acc1_pos: 43.103 acc5_pos: 100.000 loss: 0.4227 (0.5262) acc1: 81.7708 (73.8073) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.1966 data: 0.1028 max mem: 3532 +Test: [240/269] eta: 0:00:05 acc1_pos: 43.103 acc5_pos: 100.000 loss: 0.4263 (0.5256) acc1: 81.7708 (73.8762) acc5: 100.0000 (100.0000) acc1_neg: 69.010 acc5_neg: 100.000 time: 0.2065 data: 0.1132 max mem: 3532 +Test: [250/269] eta: 0:00:03 acc1_pos: 43.103 acc5_pos: 100.000 loss: 0.4955 (0.5258) acc1: 73.6979 (73.8743) acc5: 100.0000 (100.0000) acc1_neg: 57.292 acc5_neg: 100.000 time: 0.1947 data: 0.1035 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 43.103 acc5_pos: 100.000 loss: 0.5360 (0.5272) acc1: 72.9167 (73.7269) acc5: 100.0000 (100.0000) acc1_neg: 70.312 acc5_neg: 100.000 time: 0.1602 data: 0.0690 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 43.103 acc5_pos: 100.000 loss: 0.5250 (0.5255) acc1: 73.6979 (73.9750) acc5: 100.0000 (100.0000) acc1_neg: 95.013 acc5_neg: 100.000 time: 0.1380 data: 0.0432 max mem: 3532 +Test: Total time: 0:00:52 (0.1942 s / it) +* Overall Acc@1 73.975 Acc@5 100.000 + target == real Acc@1 55.152 Acc@5 100.000 + target == fake Acc@1 76.544 Acc@5 100.000 + loss 0.526 + PR-AUC 0.9518 ROC-AUC 0.7395 EER 32.64% (thr=0.5679) +Accuracy of the network on the 103293 test images: 74.0% +[rank0]:[W725 16:57:59.418173279 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-25T16:58:06Z] eval_done tag=fixed500k_fake_n22_seed2_resampled +[2026-07-25T16:58:06Z] eval_start tag=fixed500k_fake_n32_seed1_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed1_resampled_n111-046-081_20260725_112559/best_checkpoint.pth +started_utc=2026-07-25T16:58:06Z +hostname=n116-042-186 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed1_resampled_n111-046-081_20260725_112559/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed1_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0725 16:58:21.807000 88054 torch/distributed/run.py:766] +W0725 16:58:21.807000 88054 torch/distributed/run.py:766] ***************************************** +W0725 16:58:21.807000 88054 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 16:58:21.807000 88054 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 6): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 16:58:56.248766990 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W725 16:58:56.249220933 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 16:58:56.655409036 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W725 16:58:56.657497306 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 16:58:56.756172483 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 16:58:56.891985999 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +[rank0]:[W725 16:58:56.940439105 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 16:58:56.942254273 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed1_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed1_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:26:27 acc1_pos: 59.115 acc5_pos: 100.000 loss: 0.6906 (0.6906) acc1: 59.1146 (59.1146) acc5: 100.0000 (100.0000) time: 6.7843 data: 5.0634 max mem: 3531 +Test: [ 10/234] eta: 0:02:38 acc1_pos: 26.562 acc5_pos: 100.000 loss: 0.7892 (0.8256) acc1: 47.3958 (45.6676) acc5: 100.0000 (100.0000) time: 0.7057 data: 0.4606 max mem: 3531 +Test: [ 20/234] eta: 0:01:29 acc1_pos: 92.969 acc5_pos: 100.000 loss: 0.7892 (0.7877) acc1: 45.3125 (49.4668) acc5: 100.0000 (100.0000) time: 0.0989 data: 0.0004 max mem: 3531 +Test: [ 30/234] eta: 0:01:14 acc1_pos: 76.562 acc5_pos: 100.000 loss: 0.3036 (0.6326) acc1: 91.9271 (62.6932) acc5: 100.0000 (100.0000) time: 0.1749 data: 0.0783 max mem: 3531 +Test: [ 40/234] eta: 0:01:02 acc1_pos: 98.958 acc5_pos: 100.000 loss: 0.3012 (0.5868) acc1: 92.1875 (66.0569) acc5: 100.0000 (100.0000) time: 0.2201 data: 0.1270 max mem: 3531 +Test: [ 50/234] eta: 0:00:54 acc1_pos: 99.479 acc5_pos: 100.000 loss: 0.1703 (0.5039) acc1: 97.6562 (72.3601) acc5: 100.0000 (100.0000) time: 0.1902 data: 0.0969 max mem: 3531 +Test: [ 60/234] eta: 0:00:49 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.1703 (0.4501) acc1: 99.2188 (76.8272) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2008 data: 0.1081 max mem: 3531 +Test: [ 70/234] eta: 0:00:44 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.1759 (0.4664) acc1: 99.2188 (76.3021) acc5: 100.0000 (100.0000) acc1_neg: 58.854 acc5_neg: 100.000 time: 0.2123 data: 0.1208 max mem: 3531 +Test: [ 80/234] eta: 0:00:40 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.6863 (0.4770) acc1: 60.9375 (75.9549) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2052 data: 0.1139 max mem: 3531 +Test: [ 90/234] eta: 0:00:35 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.3343 (0.4643) acc1: 86.1979 (77.2350) acc5: 100.0000 (100.0000) acc1_neg: 65.885 acc5_neg: 100.000 time: 0.1572 data: 0.0645 max mem: 3531 +Test: [100/234] eta: 0:00:32 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.3343 (0.4725) acc1: 86.1979 (76.7688) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1536 data: 0.0576 max mem: 3531 +Test: [110/234] eta: 0:00:29 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.5463 (0.4849) acc1: 72.1354 (76.0417) acc5: 100.0000 (100.0000) acc1_neg: 54.688 acc5_neg: 100.000 time: 0.2009 data: 0.1063 max mem: 3531 +Test: [120/234] eta: 0:00:26 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.5123 (0.4925) acc1: 72.1354 (75.6909) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.1962 data: 0.1035 max mem: 3531 +Test: [130/234] eta: 0:00:24 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.3749 (0.4868) acc1: 84.3750 (76.3995) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1939 data: 0.1002 max mem: 3531 +Test: [140/234] eta: 0:00:21 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.3749 (0.5002) acc1: 83.5938 (75.9752) acc5: 100.0000 (100.0000) acc1_neg: 37.500 acc5_neg: 100.000 time: 0.2101 data: 0.1153 max mem: 3531 +Test: [150/234] eta: 0:00:18 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.5528 (0.5112) acc1: 65.8854 (75.4674) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.1648 data: 0.0703 max mem: 3531 +Test: [160/234] eta: 0:00:16 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.4854 (0.5035) acc1: 77.6042 (76.1177) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.1834 data: 0.0911 max mem: 3531 +Test: [170/234] eta: 0:00:14 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.4800 (0.5020) acc1: 78.9062 (76.3158) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.2163 data: 0.1227 max mem: 3531 +Test: [180/234] eta: 0:00:11 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.5039 (0.5174) acc1: 75.7812 (75.2949) acc5: 100.0000 (100.0000) acc1_neg: 22.396 acc5_neg: 100.000 time: 0.1966 data: 0.1032 max mem: 3531 +Test: [190/234] eta: 0:00:09 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.5751 (0.5219) acc1: 69.2708 (74.9386) acc5: 100.0000 (100.0000) acc1_neg: 72.396 acc5_neg: 100.000 time: 0.2001 data: 0.1091 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.5405 (0.5088) acc1: 72.3958 (75.6646) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2032 data: 0.1112 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.0864 (0.4910) acc1: 99.7396 (76.7810) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1797 data: 0.0886 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.2227 (0.4796) acc1: 97.9167 (77.6749) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1342 data: 0.0445 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.3702 (0.4874) acc1: 64.5833 (76.9920) acc5: 100.0000 (100.0000) acc1_neg: 61.198 acc5_neg: 100.000 time: 0.1298 data: 0.0400 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.6393 (0.4896) acc1: 63.0208 (76.8291) acc5: 100.0000 (100.0000) acc1_neg: 61.868 acc5_neg: 100.000 time: 0.1310 data: 0.0402 max mem: 3532 +Test: Total time: 0:00:48 (0.2083 s / it) +* Overall Acc@1 76.829 Acc@5 100.000 + target == real Acc@1 72.553 Acc@5 100.000 + target == fake Acc@1 78.034 Acc@5 100.000 + loss 0.490 + PR-AUC 0.9493 ROC-AUC 0.8420 EER 24.35% (thr=0.5303) +Accuracy of the network on the 89729 test images: 76.8% +[rank0]:[W725 16:59:58.768838047 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 17:00:19.943000 88611 torch/distributed/run.py:766] +W0725 17:00:19.943000 88611 torch/distributed/run.py:766] ***************************************** +W0725 17:00:19.943000 88611 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:00:19.943000 88611 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:00:53.135088427 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 17:00:53.177968809 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:00:53.428364520 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:00:53.478355796 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:00:53.598624388 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:00:53.686772894 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:00:53.722818679 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank6]:[W725 17:00:53.723688201 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed1_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed1_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:33:04 acc1_pos: 26.302 acc5_pos: 100.000 loss: 1.1336 (1.1336) acc1: 26.3021 (26.3021) acc5: 100.0000 (100.0000) time: 6.8639 data: 5.1912 max mem: 3531 +Test: [ 10/2387] eta: 0:28:00 acc1_pos: 12.240 acc5_pos: 100.000 loss: 1.1401 (1.1180) acc1: 21.0938 (22.8693) acc5: 100.0000 (100.0000) time: 0.7071 data: 0.4724 max mem: 3531 +Test: [ 20/2387] eta: 0:16:29 acc1_pos: 24.219 acc5_pos: 100.000 loss: 1.1401 (1.0901) acc1: 19.7917 (24.6900) acc5: 100.0000 (100.0000) time: 0.0957 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:14:03 acc1_pos: 15.104 acc5_pos: 100.000 loss: 1.1086 (1.0862) acc1: 19.0104 (24.6724) acc5: 100.0000 (100.0000) time: 0.1658 data: 0.0682 max mem: 3531 +Test: [ 40/2387] eta: 0:12:45 acc1_pos: 17.969 acc5_pos: 100.000 loss: 1.1086 (1.0875) acc1: 18.7500 (23.0818) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1359 max mem: 3531 +Test: [ 50/2387] eta: 0:12:07 acc1_pos: 41.927 acc5_pos: 100.000 loss: 1.0200 (1.0765) acc1: 20.3125 (24.4690) acc5: 100.0000 (100.0000) time: 0.2390 data: 0.1454 max mem: 3531 +Test: [ 60/2387] eta: 0:11:44 acc1_pos: 39.844 acc5_pos: 100.000 loss: 1.0709 (1.0815) acc1: 22.6562 (24.1078) acc5: 100.0000 (100.0000) time: 0.2549 data: 0.1609 max mem: 3531 +Test: [ 70/2387] eta: 0:11:31 acc1_pos: 33.594 acc5_pos: 100.000 loss: 1.1287 (1.0915) acc1: 19.5312 (23.5329) acc5: 100.0000 (100.0000) time: 0.2664 data: 0.1731 max mem: 3531 +Test: [ 80/2387] eta: 0:11:19 acc1_pos: 36.198 acc5_pos: 100.000 loss: 1.1175 (1.0886) acc1: 16.1458 (23.2382) acc5: 100.0000 (100.0000) time: 0.2699 data: 0.1765 max mem: 3531 +Test: [ 90/2387] eta: 0:11:15 acc1_pos: 36.979 acc5_pos: 100.000 loss: 1.0483 (1.0783) acc1: 25.0000 (24.0642) acc5: 100.0000 (100.0000) time: 0.2783 data: 0.1820 max mem: 3531 +Test: [ 100/2387] eta: 0:10:55 acc1_pos: 22.135 acc5_pos: 100.000 loss: 1.0451 (1.0815) acc1: 25.0000 (23.8114) acc5: 100.0000 (100.0000) time: 0.2549 data: 0.1582 max mem: 3531 +Test: [ 110/2387] eta: 0:10:47 acc1_pos: 38.281 acc5_pos: 100.000 loss: 1.0238 (1.0707) acc1: 24.7396 (24.4369) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1455 max mem: 3531 +Test: [ 120/2387] eta: 0:10:38 acc1_pos: 21.875 acc5_pos: 100.000 loss: 1.0082 (1.0709) acc1: 25.7812 (24.6018) acc5: 100.0000 (100.0000) time: 0.2549 data: 0.1612 max mem: 3531 +Test: [ 130/2387] eta: 0:10:31 acc1_pos: 15.885 acc5_pos: 100.000 loss: 1.0660 (1.0703) acc1: 20.3125 (24.2684) acc5: 100.0000 (100.0000) time: 0.2549 data: 0.1589 max mem: 3531 +Test: [ 140/2387] eta: 0:10:22 acc1_pos: 26.302 acc5_pos: 100.000 loss: 1.0432 (1.0666) acc1: 20.5729 (24.4219) acc5: 100.0000 (100.0000) time: 0.2499 data: 0.1504 max mem: 3531 +Test: [ 150/2387] eta: 0:10:09 acc1_pos: 8.854 acc5_pos: 100.000 loss: 0.9972 (1.0626) acc1: 26.3021 (24.6896) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1242 max mem: 3531 +Test: [ 160/2387] eta: 0:10:02 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.9487 (1.0433) acc1: 29.4271 (26.3263) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2249 data: 0.1257 max mem: 3531 +Test: [ 170/2387] eta: 0:09:48 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2050 (0.9937) acc1: 97.1354 (30.5266) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2099 data: 0.1127 max mem: 3531 +Test: [ 180/2387] eta: 0:09:37 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1933 (0.9496) acc1: 98.1771 (34.2728) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1899 data: 0.0914 max mem: 3531 +Test: [ 190/2387] eta: 0:09:25 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1911 (0.9100) acc1: 98.4375 (37.6323) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1899 data: 0.0920 max mem: 3531 +Test: [ 200/2387] eta: 0:09:19 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1909 (0.8742) acc1: 98.4375 (40.6639) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2049 data: 0.1078 max mem: 3531 +Test: [ 210/2387] eta: 0:09:12 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1935 (0.8422) acc1: 98.4375 (43.4020) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2199 data: 0.1237 max mem: 3532 +Test: [ 220/2387] eta: 0:09:05 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1953 (0.8129) acc1: 98.4375 (45.8852) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2099 data: 0.1156 max mem: 3532 +Test: [ 230/2387] eta: 0:08:59 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1953 (0.7863) acc1: 98.1771 (48.1444) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2099 data: 0.1149 max mem: 3532 +Test: [ 240/2387] eta: 0:08:52 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2042 (0.7624) acc1: 98.1771 (50.2140) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2049 data: 0.1079 max mem: 3532 +Test: [ 250/2387] eta: 0:08:44 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1916 (0.7355) acc1: 98.6979 (52.1871) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0930 max mem: 3532 +Test: [ 260/2387] eta: 0:08:35 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0637 (0.7096) acc1: 100.0000 (54.0190) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0773 max mem: 3532 +Test: [ 270/2387] eta: 0:08:33 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0611 (0.6857) acc1: 100.0000 (55.7157) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2067 data: 0.1095 max mem: 3532 +Test: [ 280/2387] eta: 0:08:27 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0611 (0.6636) acc1: 100.0000 (57.2917) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2199 data: 0.1264 max mem: 3532 +Test: [ 290/2387] eta: 0:08:21 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0649 (0.6430) acc1: 100.0000 (58.7593) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1931 data: 0.1008 max mem: 3532 +Test: [ 300/2387] eta: 0:08:18 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0649 (0.6238) acc1: 100.0000 (60.1294) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2105 data: 0.1168 max mem: 3532 +Test: [ 310/2387] eta: 0:08:14 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0692 (0.6102) acc1: 100.0000 (61.3779) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2249 data: 0.1269 max mem: 3532 +Test: [ 320/2387] eta: 0:08:12 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.3646 (0.6026) acc1: 98.1771 (62.5073) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2304 data: 0.1336 max mem: 3532 +Test: [ 330/2387] eta: 0:08:07 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.3652 (0.5954) acc1: 97.6562 (63.5794) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2200 data: 0.1284 max mem: 3532 +Test: [ 340/2387] eta: 0:08:01 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.3621 (0.5885) acc1: 97.6562 (64.5826) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1888 data: 0.0943 max mem: 3532 +Test: [ 350/2387] eta: 0:07:59 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.3555 (0.5770) acc1: 98.6979 (65.5835) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2121 data: 0.1169 max mem: 3532 +Test: [ 360/2387] eta: 0:07:55 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1607 (0.5656) acc1: 99.7396 (66.5166) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2202 data: 0.1289 max mem: 3532 +Test: [ 370/2387] eta: 0:07:50 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1711 (0.5550) acc1: 98.9583 (67.3883) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1929 data: 0.1026 max mem: 3532 +Test: [ 380/2387] eta: 0:07:46 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1718 (0.5449) acc1: 98.6979 (68.2107) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1973 data: 0.1070 max mem: 3532 +Test: [ 390/2387] eta: 0:07:43 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1696 (0.5354) acc1: 98.9583 (68.9958) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2148 data: 0.1246 max mem: 3532 +Test: [ 400/2387] eta: 0:07:40 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1699 (0.5263) acc1: 98.9583 (69.7410) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2124 data: 0.1220 max mem: 3532 +Test: [ 410/2387] eta: 0:07:35 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1694 (0.5177) acc1: 98.9583 (70.4500) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1915 data: 0.1001 max mem: 3532 +Test: [ 420/2387] eta: 0:07:31 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1695 (0.5094) acc1: 98.9583 (71.1253) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1875 data: 0.0949 max mem: 3532 +Test: [ 430/2387] eta: 0:07:27 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1737 (0.5019) acc1: 98.6979 (71.7656) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1887 data: 0.0967 max mem: 3532 +Test: [ 440/2387] eta: 0:07:23 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1743 (0.4932) acc1: 98.9583 (72.3905) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1875 data: 0.0968 max mem: 3533 +Test: [ 450/2387] eta: 0:07:19 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0701 (0.4838) acc1: 100.0000 (73.0027) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1907 data: 0.1000 max mem: 3533 +Test: [ 460/2387] eta: 0:07:14 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0695 (0.4748) acc1: 100.0000 (73.5883) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1848 data: 0.0929 max mem: 3533 +Test: [ 470/2387] eta: 0:07:11 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0695 (0.4662) acc1: 100.0000 (74.1480) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1912 data: 0.0956 max mem: 3533 +Test: [ 480/2387] eta: 0:07:08 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0699 (0.4580) acc1: 100.0000 (74.6854) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1972 data: 0.1009 max mem: 3533 +Test: [ 490/2387] eta: 0:07:04 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0706 (0.4501) acc1: 100.0000 (75.2010) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0971 max mem: 3533 +Test: [ 500/2387] eta: 0:07:00 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0704 (0.4425) acc1: 100.0000 (75.6955) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1876 data: 0.0960 max mem: 3533 +Test: [ 510/2387] eta: 0:06:57 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0704 (0.4354) acc1: 100.0000 (76.1711) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1884 data: 0.0971 max mem: 3533 +Test: [ 520/2387] eta: 0:06:53 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0990 (0.4291) acc1: 100.0000 (76.6285) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1796 data: 0.0871 max mem: 3533 +Test: [ 530/2387] eta: 0:06:50 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1059 (0.4230) acc1: 100.0000 (77.0686) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1842 data: 0.0896 max mem: 3533 +Test: [ 540/2387] eta: 0:06:46 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1078 (0.4172) acc1: 100.0000 (77.4920) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0904 max mem: 3533 +Test: [ 550/2387] eta: 0:06:42 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1074 (0.4116) acc1: 100.0000 (77.9005) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1721 data: 0.0787 max mem: 3533 +Test: [ 560/2387] eta: 0:06:39 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1078 (0.4062) acc1: 100.0000 (78.2944) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1780 data: 0.0858 max mem: 3533 +Test: [ 570/2387] eta: 0:06:35 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1080 (0.4009) acc1: 100.0000 (78.6746) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1733 data: 0.0802 max mem: 3533 +Test: [ 580/2387] eta: 0:06:31 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1080 (0.3959) acc1: 100.0000 (79.0416) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1685 data: 0.0739 max mem: 3533 +Test: [ 590/2387] eta: 0:06:28 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1087 (0.3910) acc1: 100.0000 (79.3962) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1689 data: 0.0756 max mem: 3533 +Test: [ 600/2387] eta: 0:06:24 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1052 (0.3862) acc1: 100.0000 (79.7391) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1735 data: 0.0816 max mem: 3533 +Test: [ 610/2387] eta: 0:06:21 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1190 (0.3824) acc1: 100.0000 (80.0634) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1777 data: 0.0846 max mem: 3533 +Test: [ 620/2387] eta: 0:06:18 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1494 (0.3786) acc1: 99.7396 (80.3786) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1866 data: 0.0939 max mem: 3533 +Test: [ 630/2387] eta: 0:06:15 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1486 (0.3750) acc1: 99.7396 (80.6825) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1921 data: 0.0943 max mem: 3533 +Test: [ 640/2387] eta: 0:06:13 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1527 (0.3715) acc1: 99.4792 (80.9750) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1928 data: 0.0950 max mem: 3533 +Test: [ 650/2387] eta: 0:06:10 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1531 (0.3681) acc1: 99.4792 (81.2584) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1908 data: 0.0999 max mem: 3533 +Test: [ 660/2387] eta: 0:06:07 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1503 (0.3648) acc1: 99.4792 (81.5341) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1768 data: 0.0863 max mem: 3533 +Test: [ 670/2387] eta: 0:06:04 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1509 (0.3617) acc1: 99.4792 (81.8050) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1813 data: 0.0885 max mem: 3534 +Test: [ 680/2387] eta: 0:06:02 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1537 (0.3588) acc1: 100.0000 (82.0695) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1941 data: 0.0951 max mem: 3534 +Test: [ 690/2387] eta: 0:05:59 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1711 (0.3588) acc1: 99.7396 (82.1183) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.1892 data: 0.0908 max mem: 3534 +Test: [ 700/2387] eta: 0:05:57 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.4276 (0.3597) acc1: 80.9896 (82.1096) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.1998 data: 0.1065 max mem: 3534 +Test: [ 710/2387] eta: 0:05:53 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.4264 (0.3607) acc1: 81.2500 (82.0917) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1848 data: 0.0917 max mem: 3534 +Test: [ 720/2387] eta: 0:05:50 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.4258 (0.3616) acc1: 80.9896 (82.0822) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.1621 data: 0.0696 max mem: 3534 +Test: [ 730/2387] eta: 0:05:48 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.4327 (0.3627) acc1: 80.7292 (82.0462) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.1796 data: 0.0882 max mem: 3534 +Test: [ 740/2387] eta: 0:05:45 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.4341 (0.3636) acc1: 79.6875 (82.0225) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.1895 data: 0.0984 max mem: 3534 +Test: [ 750/2387] eta: 0:05:42 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.4253 (0.3644) acc1: 80.7292 (82.0132) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.1814 data: 0.0860 max mem: 3534 +Test: [ 760/2387] eta: 0:05:39 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.4140 (0.3648) acc1: 81.2500 (82.0141) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1771 data: 0.0792 max mem: 3534 +Test: [ 770/2387] eta: 0:05:37 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.3595 (0.3638) acc1: 85.4167 (82.1272) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1851 data: 0.0915 max mem: 3534 +Test: [ 780/2387] eta: 0:05:34 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2252 (0.3619) acc1: 95.5729 (82.3093) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1755 data: 0.0843 max mem: 3534 +Test: [ 790/2387] eta: 0:05:31 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1965 (0.3596) acc1: 96.8750 (82.5053) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1771 data: 0.0843 max mem: 3534 +Test: [ 800/2387] eta: 0:05:29 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1712 (0.3576) acc1: 97.9167 (82.6831) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1921 data: 0.0993 max mem: 3534 +Test: [ 810/2387] eta: 0:05:27 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1881 (0.3558) acc1: 97.9167 (82.8549) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2005 data: 0.1087 max mem: 3534 +Test: [ 820/2387] eta: 0:05:25 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2083 (0.3545) acc1: 95.0521 (82.9889) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.2099 data: 0.1178 max mem: 3534 +Test: [ 830/2387] eta: 0:05:23 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2704 (0.3538) acc1: 93.2292 (83.0870) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2089 data: 0.1161 max mem: 3534 +Test: [ 840/2387] eta: 0:05:21 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2330 (0.3521) acc1: 96.3542 (83.2692) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2049 data: 0.1109 max mem: 3534 +Test: [ 850/2387] eta: 0:05:19 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1959 (0.3502) acc1: 99.2188 (83.4545) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2041 data: 0.1110 max mem: 3534 +Test: [ 860/2387] eta: 0:05:17 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1782 (0.3481) acc1: 99.7396 (83.6418) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2056 data: 0.1139 max mem: 3534 +Test: [ 870/2387] eta: 0:05:15 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1788 (0.3464) acc1: 99.2188 (83.8168) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2119 data: 0.1199 max mem: 3534 +Test: [ 880/2387] eta: 0:05:13 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1946 (0.3447) acc1: 98.9583 (83.9893) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2180 data: 0.1256 max mem: 3534 +Test: [ 890/2387] eta: 0:05:11 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1881 (0.3428) acc1: 99.2188 (84.1581) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2154 data: 0.1234 max mem: 3534 +Test: [ 900/2387] eta: 0:05:08 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2116 (0.3453) acc1: 95.5729 (84.0076) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.1988 data: 0.1074 max mem: 3535 +Test: [ 910/2387] eta: 0:05:07 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.6526 (0.3487) acc1: 64.3229 (83.7930) acc5: 100.0000 (100.0000) acc1_neg: 63.802 acc5_neg: 100.000 time: 0.2078 data: 0.1164 max mem: 3535 +Test: [ 920/2387] eta: 0:05:04 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.6647 (0.3521) acc1: 63.8021 (83.5697) acc5: 100.0000 (100.0000) acc1_neg: 59.896 acc5_neg: 100.000 time: 0.2171 data: 0.1226 max mem: 3535 +Test: [ 930/2387] eta: 0:05:01 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.6542 (0.3553) acc1: 64.0625 (83.3613) acc5: 100.0000 (100.0000) acc1_neg: 58.594 acc5_neg: 100.000 time: 0.1784 data: 0.0819 max mem: 3535 +Test: [ 940/2387] eta: 0:04:59 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.6573 (0.3586) acc1: 64.0625 (83.1457) acc5: 100.0000 (100.0000) acc1_neg: 59.896 acc5_neg: 100.000 time: 0.1620 data: 0.0689 max mem: 3535 +Test: [ 950/2387] eta: 0:04:57 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.6573 (0.3617) acc1: 63.8021 (82.9390) acc5: 100.0000 (100.0000) acc1_neg: 64.844 acc5_neg: 100.000 time: 0.1853 data: 0.0932 max mem: 3535 +Test: [ 960/2387] eta: 0:04:54 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.6429 (0.3646) acc1: 64.5833 (82.7499) acc5: 100.0000 (100.0000) acc1_neg: 65.625 acc5_neg: 100.000 time: 0.1829 data: 0.0907 max mem: 3535 +Test: [ 970/2387] eta: 0:04:52 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.6516 (0.3676) acc1: 64.5833 (82.5507) acc5: 100.0000 (100.0000) acc1_neg: 65.885 acc5_neg: 100.000 time: 0.1773 data: 0.0847 max mem: 3535 +Test: [ 980/2387] eta: 0:04:50 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.6199 (0.3679) acc1: 66.9271 (82.5667) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.2003 data: 0.1076 max mem: 3535 +Test: [ 990/2387] eta: 0:04:47 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2712 (0.3667) acc1: 92.9688 (82.6801) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1971 data: 0.1017 max mem: 3535 +Test: [1000/2387] eta: 0:04:45 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2212 (0.3653) acc1: 97.3958 (82.8159) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1794 data: 0.0830 max mem: 3535 +Test: [1010/2387] eta: 0:04:42 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2294 (0.3644) acc1: 94.7917 (82.9209) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1729 data: 0.0792 max mem: 3535 +Test: [1020/2387] eta: 0:04:40 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2433 (0.3632) acc1: 95.8333 (83.0303) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1726 data: 0.0799 max mem: 3535 +Test: [1030/2387] eta: 0:04:38 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1751 (0.3612) acc1: 97.6562 (83.1798) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1826 data: 0.0894 max mem: 3535 +Test: [1040/2387] eta: 0:04:35 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1440 (0.3590) acc1: 99.4792 (83.3376) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1896 data: 0.0949 max mem: 3535 +Test: [1050/2387] eta: 0:04:33 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1309 (0.3569) acc1: 100.0000 (83.4897) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1992 data: 0.1058 max mem: 3535 +Test: [1060/2387] eta: 0:04:31 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1357 (0.3550) acc1: 99.4792 (83.6367) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2124 data: 0.1194 max mem: 3535 +Test: [1070/2387] eta: 0:04:29 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1620 (0.3534) acc1: 98.9583 (83.7715) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1999 data: 0.1048 max mem: 3535 +Test: [1080/2387] eta: 0:04:26 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1958 (0.3521) acc1: 97.6562 (83.8838) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1617 data: 0.0657 max mem: 3535 +Test: [1090/2387] eta: 0:04:25 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1775 (0.3504) acc1: 97.6562 (84.0143) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1851 data: 0.0898 max mem: 3535 +Test: [1100/2387] eta: 0:04:23 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1521 (0.3486) acc1: 98.6979 (84.1508) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2246 data: 0.1310 max mem: 3535 +Test: [1110/2387] eta: 0:04:20 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1530 (0.3471) acc1: 98.6979 (84.2756) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1865 data: 0.0946 max mem: 3535 +Test: [1120/2387] eta: 0:04:19 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1853 (0.3457) acc1: 97.6562 (84.3943) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2060 data: 0.1138 max mem: 3536 +Test: [1130/2387] eta: 0:04:16 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1813 (0.3442) acc1: 98.1771 (84.5180) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2100 data: 0.1189 max mem: 3536 +Test: [1140/2387] eta: 0:04:14 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1990 (0.3431) acc1: 97.6562 (84.6249) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1779 data: 0.0870 max mem: 3536 +Test: [1150/2387] eta: 0:04:12 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1876 (0.3417) acc1: 98.6979 (84.7418) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1906 data: 0.0995 max mem: 3536 +Test: [1160/2387] eta: 0:04:10 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1244 (0.3396) acc1: 100.0000 (84.8725) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1971 data: 0.1064 max mem: 3536 +Test: [1170/2387] eta: 0:04:08 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1069 (0.3377) acc1: 100.0000 (85.0012) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2134 data: 0.1228 max mem: 3536 +Test: [1180/2387] eta: 0:04:06 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1085 (0.3358) acc1: 100.0000 (85.1263) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2090 data: 0.1160 max mem: 3536 +Test: [1190/2387] eta: 0:04:04 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1118 (0.3340) acc1: 100.0000 (85.2501) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1965 data: 0.1035 max mem: 3536 +Test: [1200/2387] eta: 0:04:02 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1122 (0.3321) acc1: 100.0000 (85.3724) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1959 data: 0.1054 max mem: 3536 +Test: [1210/2387] eta: 0:03:59 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1152 (0.3304) acc1: 100.0000 (85.4926) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1946 data: 0.1044 max mem: 3536 +Test: [1220/2387] eta: 0:03:57 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0846 (0.3284) acc1: 100.0000 (85.6114) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1941 data: 0.1038 max mem: 3536 +Test: [1230/2387] eta: 0:03:55 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0722 (0.3263) acc1: 100.0000 (85.7283) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1944 data: 0.1020 max mem: 3536 +Test: [1240/2387] eta: 0:03:53 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0694 (0.3242) acc1: 100.0000 (85.8429) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2006 data: 0.1060 max mem: 3536 +Test: [1250/2387] eta: 0:03:51 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0721 (0.3223) acc1: 100.0000 (85.9560) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2047 data: 0.1121 max mem: 3536 +Test: [1260/2387] eta: 0:03:49 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0745 (0.3203) acc1: 100.0000 (86.0674) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1952 data: 0.1051 max mem: 3536 +Test: [1270/2387] eta: 0:03:46 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0745 (0.3184) acc1: 100.0000 (86.1768) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1748 data: 0.0842 max mem: 3536 +Test: [1280/2387] eta: 0:03:44 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0806 (0.3166) acc1: 100.0000 (86.2821) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1668 data: 0.0733 max mem: 3536 +Test: [1290/2387] eta: 0:03:42 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0955 (0.3149) acc1: 99.4792 (86.3849) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1693 data: 0.0726 max mem: 3536 +Test: [1300/2387] eta: 0:03:40 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0925 (0.3133) acc1: 99.4792 (86.4864) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0806 max mem: 3536 +Test: [1310/2387] eta: 0:03:37 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0990 (0.3117) acc1: 99.7396 (86.5853) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1751 data: 0.0808 max mem: 3536 +Test: [1320/2387] eta: 0:03:35 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1099 (0.3103) acc1: 99.4792 (86.6831) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1802 data: 0.0844 max mem: 3536 +Test: [1330/2387] eta: 0:03:33 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1278 (0.3090) acc1: 99.4792 (86.7753) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1819 data: 0.0893 max mem: 3536 +Test: [1340/2387] eta: 0:03:31 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1652 (0.3088) acc1: 97.9167 (86.7955) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1750 data: 0.0828 max mem: 3536 +Test: [1350/2387] eta: 0:03:29 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0544 (0.3069) acc1: 100.0000 (86.8932) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1887 data: 0.0969 max mem: 3537 +Test: [1360/2387] eta: 0:03:26 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0410 (0.3049) acc1: 100.0000 (86.9895) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1819 data: 0.0896 max mem: 3537 +Test: [1370/2387] eta: 0:03:24 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0410 (0.3031) acc1: 100.0000 (87.0844) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1729 data: 0.0802 max mem: 3537 +Test: [1380/2387] eta: 0:03:22 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0444 (0.3012) acc1: 100.0000 (87.1779) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1678 data: 0.0755 max mem: 3537 +Test: [1390/2387] eta: 0:03:20 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0444 (0.2994) acc1: 100.0000 (87.2701) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1752 data: 0.0802 max mem: 3537 +Test: [1400/2387] eta: 0:03:18 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0451 (0.2976) acc1: 100.0000 (87.3610) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1750 data: 0.0818 max mem: 3537 +Test: [1410/2387] eta: 0:03:15 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0465 (0.2964) acc1: 100.0000 (87.4347) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1697 data: 0.0773 max mem: 3537 +Test: [1420/2387] eta: 0:03:13 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1633 (0.2955) acc1: 98.9583 (87.5090) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1858 data: 0.0913 max mem: 3537 +Test: [1430/2387] eta: 0:03:11 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1778 (0.2948) acc1: 97.9167 (87.5766) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1704 data: 0.0773 max mem: 3537 +Test: [1440/2387] eta: 0:03:09 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1849 (0.2941) acc1: 97.9167 (87.6458) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1737 data: 0.0802 max mem: 3537 +Test: [1450/2387] eta: 0:03:07 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1804 (0.2935) acc1: 98.1771 (87.7064) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1896 data: 0.0959 max mem: 3537 +Test: [1460/2387] eta: 0:03:04 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1627 (0.2926) acc1: 98.1771 (87.7798) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1549 data: 0.0595 max mem: 3537 +Test: [1470/2387] eta: 0:03:02 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1343 (0.2914) acc1: 98.9583 (87.8578) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1456 data: 0.0509 max mem: 3537 +Test: [1480/2387] eta: 0:03:00 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1236 (0.2903) acc1: 99.7396 (87.9363) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1761 data: 0.0822 max mem: 3537 +Test: [1490/2387] eta: 0:02:58 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1317 (0.2893) acc1: 99.7396 (88.0105) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1989 data: 0.1043 max mem: 3537 +Test: [1500/2387] eta: 0:02:56 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1336 (0.2882) acc1: 99.7396 (88.0852) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1737 data: 0.0804 max mem: 3537 +Test: [1510/2387] eta: 0:02:54 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1290 (0.2872) acc1: 99.7396 (88.1594) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1512 data: 0.0582 max mem: 3537 +Test: [1520/2387] eta: 0:02:52 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1574 (0.2865) acc1: 99.4792 (88.2302) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2018 data: 0.1096 max mem: 3537 +Test: [1530/2387] eta: 0:02:50 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1683 (0.2857) acc1: 99.4792 (88.3015) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2035 data: 0.1108 max mem: 3537 +Test: [1540/2387] eta: 0:02:47 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1570 (0.2848) acc1: 99.7396 (88.3752) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1414 data: 0.0470 max mem: 3537 +Test: [1550/2387] eta: 0:02:45 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1535 (0.2841) acc1: 99.7396 (88.4468) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1510 data: 0.0576 max mem: 3537 +Test: [1560/2387] eta: 0:02:43 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1606 (0.2834) acc1: 99.4792 (88.5175) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1816 data: 0.0903 max mem: 3537 +Test: [1570/2387] eta: 0:02:41 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1606 (0.2826) acc1: 99.4792 (88.5871) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1788 data: 0.0854 max mem: 3537 +Test: [1580/2387] eta: 0:02:39 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1737 (0.2820) acc1: 98.9583 (88.6385) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1760 data: 0.0826 max mem: 3538 +Test: [1590/2387] eta: 0:02:37 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1845 (0.2814) acc1: 96.3542 (88.6891) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1758 data: 0.0847 max mem: 3538 +Test: [1600/2387] eta: 0:02:35 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1663 (0.2808) acc1: 97.6562 (88.7465) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1736 data: 0.0818 max mem: 3538 +Test: [1610/2387] eta: 0:02:33 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1663 (0.2802) acc1: 97.9167 (88.8021) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1691 data: 0.0755 max mem: 3538 +Test: [1620/2387] eta: 0:02:31 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1958 (0.2797) acc1: 96.8750 (88.8467) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1719 data: 0.0784 max mem: 3538 +Test: [1630/2387] eta: 0:02:29 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2354 (0.2797) acc1: 95.0521 (88.8679) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1811 data: 0.0858 max mem: 3538 +Test: [1640/2387] eta: 0:02:26 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2831 (0.2800) acc1: 91.6667 (88.8686) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1780 data: 0.0807 max mem: 3538 +Test: [1650/2387] eta: 0:02:24 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2904 (0.2800) acc1: 91.1458 (88.8906) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1737 data: 0.0783 max mem: 3538 +Test: [1660/2387] eta: 0:02:22 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.3125 (0.2802) acc1: 90.3646 (88.8886) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1864 data: 0.0934 max mem: 3538 +Test: [1670/2387] eta: 0:02:21 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.3123 (0.2804) acc1: 89.8438 (88.8996) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2018 data: 0.1083 max mem: 3538 +Test: [1680/2387] eta: 0:02:18 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2957 (0.2805) acc1: 91.1458 (88.9158) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1859 data: 0.0907 max mem: 3538 +Test: [1690/2387] eta: 0:02:16 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2999 (0.2807) acc1: 91.1458 (88.9251) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1701 data: 0.0764 max mem: 3538 +Test: [1700/2387] eta: 0:02:14 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2925 (0.2804) acc1: 91.4062 (88.9625) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1791 data: 0.0849 max mem: 3538 +Test: [1710/2387] eta: 0:02:12 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1097 (0.2793) acc1: 99.7396 (89.0257) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1640 data: 0.0706 max mem: 3538 +Test: [1720/2387] eta: 0:02:10 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0938 (0.2782) acc1: 100.0000 (89.0891) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1582 data: 0.0651 max mem: 3538 +Test: [1730/2387] eta: 0:02:08 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0931 (0.2772) acc1: 100.0000 (89.1516) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1787 data: 0.0848 max mem: 3538 +Test: [1740/2387] eta: 0:02:06 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0976 (0.2762) acc1: 100.0000 (89.2130) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1808 data: 0.0879 max mem: 3538 +Test: [1750/2387] eta: 0:02:04 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0982 (0.2752) acc1: 100.0000 (89.2740) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1769 data: 0.0849 max mem: 3538 +Test: [1760/2387] eta: 0:02:02 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1003 (0.2742) acc1: 100.0000 (89.3345) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1810 data: 0.0887 max mem: 3538 +Test: [1770/2387] eta: 0:02:00 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1192 (0.2735) acc1: 99.7396 (89.3888) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1839 data: 0.0902 max mem: 3538 +Test: [1780/2387] eta: 0:01:58 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1354 (0.2727) acc1: 99.4792 (89.4475) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1832 data: 0.0896 max mem: 3538 +Test: [1790/2387] eta: 0:01:56 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1244 (0.2718) acc1: 100.0000 (89.5045) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1755 data: 0.0808 max mem: 3538 +Test: [1800/2387] eta: 0:01:54 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1231 (0.2710) acc1: 99.7396 (89.5614) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1814 data: 0.0883 max mem: 3538 +Test: [1810/2387] eta: 0:01:52 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1156 (0.2701) acc1: 99.7396 (89.6177) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1833 data: 0.0907 max mem: 3539 +Test: [1820/2387] eta: 0:01:50 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1085 (0.2693) acc1: 99.7396 (89.6734) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1889 data: 0.0952 max mem: 3539 +Test: [1830/2387] eta: 0:01:48 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1190 (0.2685) acc1: 100.0000 (89.7293) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2064 data: 0.1149 max mem: 3539 +Test: [1840/2387] eta: 0:01:46 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1150 (0.2676) acc1: 100.0000 (89.7849) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1995 data: 0.1082 max mem: 3539 +Test: [1850/2387] eta: 0:01:45 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1138 (0.2668) acc1: 100.0000 (89.8397) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2347 data: 0.1427 max mem: 3539 +Test: [1860/2387] eta: 0:01:42 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1153 (0.2660) acc1: 100.0000 (89.8943) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2154 data: 0.1222 max mem: 3539 +Test: [1870/2387] eta: 0:01:41 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1163 (0.2652) acc1: 100.0000 (89.9479) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1887 data: 0.0958 max mem: 3539 +Test: [1880/2387] eta: 0:01:39 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1143 (0.2644) acc1: 100.0000 (90.0010) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2146 data: 0.1206 max mem: 3539 +Test: [1890/2387] eta: 0:01:37 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1135 (0.2636) acc1: 100.0000 (90.0538) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1703 data: 0.0719 max mem: 3539 +Test: [1900/2387] eta: 0:01:35 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1153 (0.2628) acc1: 100.0000 (90.1055) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1691 data: 0.0726 max mem: 3539 +Test: [1910/2387] eta: 0:01:33 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1293 (0.2622) acc1: 100.0000 (90.1553) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2001 data: 0.1080 max mem: 3539 +Test: [1920/2387] eta: 0:01:31 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1512 (0.2618) acc1: 99.7396 (90.1927) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1949 data: 0.1038 max mem: 3539 +Test: [1930/2387] eta: 0:01:29 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1292 (0.2610) acc1: 100.0000 (90.2435) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1966 data: 0.1058 max mem: 3539 +Test: [1940/2387] eta: 0:01:27 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1173 (0.2605) acc1: 100.0000 (90.2919) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1966 data: 0.1049 max mem: 3539 +Test: [1950/2387] eta: 0:01:25 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1402 (0.2599) acc1: 100.0000 (90.3374) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1949 data: 0.1029 max mem: 3539 +Test: [1960/2387] eta: 0:01:23 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1478 (0.2595) acc1: 99.7396 (90.3748) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1948 data: 0.1043 max mem: 3539 +Test: [1970/2387] eta: 0:01:21 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2088 (0.2597) acc1: 97.1354 (90.3680) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1990 data: 0.1051 max mem: 3539 +Test: [1980/2387] eta: 0:01:19 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2178 (0.2594) acc1: 95.3125 (90.3989) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2033 data: 0.1082 max mem: 3539 +Test: [1990/2387] eta: 0:01:17 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1476 (0.2589) acc1: 98.4375 (90.4398) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1933 data: 0.0999 max mem: 3539 +Test: [2000/2387] eta: 0:01:15 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1472 (0.2583) acc1: 98.4375 (90.4807) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1821 data: 0.0876 max mem: 3539 +Test: [2010/2387] eta: 0:01:13 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1478 (0.2578) acc1: 98.4375 (90.5209) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2107 data: 0.1172 max mem: 3539 +Test: [2020/2387] eta: 0:01:11 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1474 (0.2572) acc1: 98.9583 (90.5617) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1977 data: 0.1032 max mem: 3539 +Test: [2030/2387] eta: 0:01:09 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1453 (0.2567) acc1: 98.9583 (90.6015) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1446 data: 0.0493 max mem: 3540 +Test: [2040/2387] eta: 0:01:07 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1465 (0.2561) acc1: 98.9583 (90.6418) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1707 data: 0.0780 max mem: 3540 +Test: [2050/2387] eta: 0:01:05 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1467 (0.2556) acc1: 98.9583 (90.6828) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2194 data: 0.1261 max mem: 3540 +Test: [2060/2387] eta: 0:01:03 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1455 (0.2551) acc1: 98.6979 (90.7215) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1941 data: 0.0979 max mem: 3540 +Test: [2070/2387] eta: 0:01:01 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1447 (0.2545) acc1: 98.6979 (90.7613) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1727 data: 0.0786 max mem: 3540 +Test: [2080/2387] eta: 0:00:59 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1439 (0.2540) acc1: 98.6979 (90.7989) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1763 data: 0.0858 max mem: 3540 +Test: [2090/2387] eta: 0:00:57 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1458 (0.2535) acc1: 98.6979 (90.8378) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1751 data: 0.0844 max mem: 3540 +Test: [2100/2387] eta: 0:00:55 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1465 (0.2530) acc1: 98.9583 (90.8761) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1885 data: 0.0944 max mem: 3540 +Test: [2110/2387] eta: 0:00:53 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1436 (0.2525) acc1: 98.6979 (90.9133) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1934 data: 0.0988 max mem: 3540 +Test: [2120/2387] eta: 0:00:52 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1485 (0.2523) acc1: 98.4375 (90.9329) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.1949 data: 0.0988 max mem: 3540 +Test: [2130/2387] eta: 0:00:50 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1675 (0.2519) acc1: 97.1354 (90.9631) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1884 data: 0.0918 max mem: 3540 +Test: [2140/2387] eta: 0:00:48 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1699 (0.2516) acc1: 98.1771 (90.9910) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1877 data: 0.0950 max mem: 3540 +Test: [2150/2387] eta: 0:00:46 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1878 (0.2514) acc1: 97.1354 (91.0213) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1899 data: 0.0982 max mem: 3540 +Test: [2160/2387] eta: 0:00:44 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1805 (0.2511) acc1: 97.3958 (91.0540) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1895 data: 0.0983 max mem: 3540 +Test: [2170/2387] eta: 0:00:42 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2255 (0.2512) acc1: 95.5729 (91.0585) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1722 data: 0.0809 max mem: 3540 +Test: [2180/2387] eta: 0:00:40 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2985 (0.2523) acc1: 91.4062 (90.9858) acc5: 100.0000 (100.0000) acc1_neg: 46.094 acc5_neg: 100.000 time: 0.1787 data: 0.0853 max mem: 3540 +Test: [2190/2387] eta: 0:00:38 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2822 (0.2526) acc1: 93.4896 (90.9809) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1999 data: 0.1069 max mem: 3540 +Test: [2200/2387] eta: 0:00:36 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1488 (0.2520) acc1: 97.1354 (91.0145) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1808 data: 0.0902 max mem: 3540 +Test: [2210/2387] eta: 0:00:34 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1019 (0.2513) acc1: 99.7396 (91.0541) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1927 data: 0.1021 max mem: 3540 +Test: [2220/2387] eta: 0:00:32 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0984 (0.2506) acc1: 100.0000 (91.0925) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1787 data: 0.0850 max mem: 3540 +Test: [2230/2387] eta: 0:00:30 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1023 (0.2500) acc1: 100.0000 (91.1314) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1497 data: 0.0549 max mem: 3540 +Test: [2240/2387] eta: 0:00:28 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1057 (0.2494) acc1: 99.7396 (91.1651) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1740 data: 0.0820 max mem: 3540 +Test: [2250/2387] eta: 0:00:26 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1166 (0.2489) acc1: 99.7396 (91.2013) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1771 data: 0.0848 max mem: 3540 +Test: [2260/2387] eta: 0:00:24 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1447 (0.2485) acc1: 99.7396 (91.2366) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0911 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1332 (0.2479) acc1: 99.7396 (91.2733) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2127 data: 0.1190 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1201 (0.2474) acc1: 100.0000 (91.3074) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2099 data: 0.1144 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1151 (0.2468) acc1: 100.0000 (91.3437) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1950 data: 0.1012 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1346 (0.2464) acc1: 99.7396 (91.3782) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1894 data: 0.0975 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1498 (0.2461) acc1: 99.2188 (91.4065) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1510 data: 0.0593 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1722 (0.2459) acc1: 98.1771 (91.4286) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1246 data: 0.0343 max mem: 3541 +Test: [2330/2387] eta: 0:00:11 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2003 (0.2459) acc1: 96.3542 (91.4450) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.1344 data: 0.0437 max mem: 3541 +Test: [2340/2387] eta: 0:00:09 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1910 (0.2456) acc1: 96.3542 (91.4723) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1463 data: 0.0560 max mem: 3541 +Test: [2350/2387] eta: 0:00:07 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1619 (0.2455) acc1: 98.9583 (91.4923) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1344 data: 0.0445 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2060 (0.2455) acc1: 96.8750 (91.5081) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1255 data: 0.0356 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2159 (0.2456) acc1: 96.3542 (91.5066) acc5: 100.0000 (100.0000) acc1_neg: 68.229 acc5_neg: 100.000 time: 0.1311 data: 0.0414 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2396 (0.2457) acc1: 92.4479 (91.5134) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1328 data: 0.0431 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.3021 (0.2459) acc1: 90.8854 (91.5170) acc5: 100.0000 (100.0000) acc1_neg: 88.372 acc5_neg: 100.000 time: 0.1239 data: 0.0357 max mem: 3541 +Test: Total time: 0:07:38 (0.1920 s / it) +* Overall Acc@1 91.517 Acc@5 100.000 + target == real Acc@1 24.472 Acc@5 100.000 + target == fake Acc@1 96.225 Acc@5 100.000 + loss 0.246 + PR-AUC 0.9903 ROC-AUC 0.8769 EER 19.96% (thr=0.7710) +Accuracy of the network on the 916267 test images: 91.5% +[rank0]:[W725 17:08:47.840802145 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 17:09:09.252000 89234 torch/distributed/run.py:766] +W0725 17:09:09.252000 89234 torch/distributed/run.py:766] ***************************************** +W0725 17:09:09.252000 89234 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:09:09.252000 89234 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:09:43.937027884 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:09:44.118731136 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:09:44.599588549 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:09:44.837631677 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 17:09:44.948151742 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:09:44.977556495 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W725 17:09:44.997815279 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:09:45.225891611 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed1_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed1_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 0:59:55 acc1_pos: 26.302 acc5_pos: 100.000 loss: 1.1336 (1.1336) acc1: 26.3021 (26.3021) acc5: 100.0000 (100.0000) time: 6.9544 data: 5.3465 max mem: 3531 +Test: [ 10/517] eta: 0:06:02 acc1_pos: 12.240 acc5_pos: 100.000 loss: 1.1401 (1.1180) acc1: 21.0938 (22.8693) acc5: 100.0000 (100.0000) time: 0.7142 data: 0.4865 max mem: 3531 +Test: [ 20/517] eta: 0:03:28 acc1_pos: 24.219 acc5_pos: 100.000 loss: 1.1401 (1.0901) acc1: 19.7917 (24.6900) acc5: 100.0000 (100.0000) time: 0.0920 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:02:52 acc1_pos: 15.104 acc5_pos: 100.000 loss: 1.1086 (1.0862) acc1: 19.0104 (24.6724) acc5: 100.0000 (100.0000) time: 0.1568 data: 0.0646 max mem: 3531 +Test: [ 40/517] eta: 0:02:41 acc1_pos: 17.969 acc5_pos: 100.000 loss: 1.1086 (1.0875) acc1: 18.7500 (23.0818) acc5: 100.0000 (100.0000) time: 0.2549 data: 0.1626 max mem: 3531 +Test: [ 50/517] eta: 0:02:34 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.9860 (1.0462) acc1: 20.3125 (26.8689) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.2949 data: 0.2004 max mem: 3531 +Test: [ 60/517] eta: 0:02:26 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.4359 (0.9405) acc1: 77.0833 (36.5010) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.2849 data: 0.1889 max mem: 3531 +Test: [ 70/517] eta: 0:02:20 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.4021 (0.8653) acc1: 88.8021 (43.8637) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.2699 data: 0.1752 max mem: 3531 +Test: [ 80/517] eta: 0:02:13 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.4174 (0.8127) acc1: 86.1979 (48.8265) acc5: 100.0000 (100.0000) acc1_neg: 72.396 acc5_neg: 100.000 time: 0.2549 data: 0.1625 max mem: 3531 +Test: [ 90/517] eta: 0:02:07 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3973 (0.7653) acc1: 89.3229 (53.6659) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.2449 data: 0.1523 max mem: 3531 +Test: [100/517] eta: 0:02:02 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3770 (0.7270) acc1: 90.8854 (57.2710) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.2452 data: 0.1513 max mem: 3531 +Test: [110/517] eta: 0:01:58 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3499 (0.6907) acc1: 90.8854 (60.4824) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2505 data: 0.1563 max mem: 3531 +Test: [120/517] eta: 0:01:54 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3527 (0.6651) acc1: 91.9271 (62.9498) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.2666 data: 0.1743 max mem: 3531 +Test: [130/517] eta: 0:01:51 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3874 (0.6428) acc1: 89.5833 (64.9889) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2747 data: 0.1833 max mem: 3531 +Test: [140/517] eta: 0:01:46 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.2529 (0.6143) acc1: 99.4792 (67.4664) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2500 data: 0.1581 max mem: 3531 +Test: [150/517] eta: 0:01:43 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.2461 (0.5910) acc1: 100.0000 (69.5261) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2494 data: 0.1549 max mem: 3531 +Test: [160/517] eta: 0:01:40 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.2641 (0.5701) acc1: 98.4375 (71.3315) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2734 data: 0.1795 max mem: 3531 +Test: [170/517] eta: 0:01:37 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.2594 (0.5512) acc1: 98.4375 (72.9075) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2714 data: 0.1802 max mem: 3531 +Test: [180/517] eta: 0:01:34 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.2606 (0.5356) acc1: 98.6979 (74.3065) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2604 data: 0.1674 max mem: 3531 +Test: [190/517] eta: 0:01:30 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.2769 (0.5220) acc1: 98.6979 (75.5836) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2490 data: 0.1562 max mem: 3531 +Test: [200/517] eta: 0:01:27 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.2686 (0.5086) acc1: 98.9583 (76.7322) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2538 data: 0.1623 max mem: 3531 +Test: [210/517] eta: 0:01:24 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3083 (0.5048) acc1: 95.8333 (77.3326) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2614 data: 0.1689 max mem: 3532 +Test: [220/517] eta: 0:01:21 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.4304 (0.5014) acc1: 88.2812 (77.8717) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.2599 data: 0.1657 max mem: 3532 +Test: [230/517] eta: 0:01:18 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.4304 (0.4987) acc1: 89.3229 (78.2896) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.2634 data: 0.1696 max mem: 3532 +Test: [240/517] eta: 0:01:16 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.4125 (0.4951) acc1: 90.1042 (78.8295) acc5: 100.0000 (100.0000) acc1_neg: 77.083 acc5_neg: 100.000 time: 0.2649 data: 0.1694 max mem: 3532 +Test: [250/517] eta: 0:01:13 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3776 (0.4880) acc1: 95.0521 (79.5101) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2549 data: 0.1607 max mem: 3532 +Test: [260/517] eta: 0:01:10 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.2807 (0.4813) acc1: 99.4792 (80.1555) acc5: 100.0000 (100.0000) acc1_neg: 77.083 acc5_neg: 100.000 time: 0.2549 data: 0.1623 max mem: 3532 +Test: [270/517] eta: 0:01:07 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.4796 (0.4824) acc1: 79.4271 (80.0594) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.2644 data: 0.1695 max mem: 3532 +Test: [280/517] eta: 0:01:04 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.4935 (0.4810) acc1: 78.3854 (80.3177) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2649 data: 0.1684 max mem: 3532 +Test: [290/517] eta: 0:01:01 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3589 (0.4763) acc1: 96.3542 (80.8858) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2699 data: 0.1718 max mem: 3532 +Test: [300/517] eta: 0:00:59 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3589 (0.4734) acc1: 95.5729 (81.2985) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.2649 data: 0.1674 max mem: 3532 +Test: [310/517] eta: 0:00:56 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3970 (0.4705) acc1: 91.9271 (81.6561) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.2649 data: 0.1692 max mem: 3532 +Test: [320/517] eta: 0:00:53 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3613 (0.4651) acc1: 92.1875 (82.1059) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2699 data: 0.1750 max mem: 3532 +Test: [330/517] eta: 0:00:50 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.2441 (0.4583) acc1: 99.4792 (82.6292) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2599 data: 0.1643 max mem: 3532 +Test: [340/517] eta: 0:00:48 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.2314 (0.4544) acc1: 99.2188 (82.9232) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.2651 data: 0.1681 max mem: 3532 +Test: [350/517] eta: 0:00:45 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.4805 (0.4560) acc1: 83.8542 (82.8993) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.2955 data: 0.2008 max mem: 3532 +Test: [360/517] eta: 0:00:42 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.4901 (0.4551) acc1: 83.8542 (83.0130) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2895 data: 0.1957 max mem: 3532 +Test: [370/517] eta: 0:00:40 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.2828 (0.4501) acc1: 99.2188 (83.4576) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2655 data: 0.1700 max mem: 3532 +Test: [380/517] eta: 0:00:37 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.2941 (0.4474) acc1: 99.2188 (83.8015) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.2598 data: 0.1665 max mem: 3532 +Test: [390/517] eta: 0:00:34 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3638 (0.4471) acc1: 92.9688 (83.8562) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2534 data: 0.1586 max mem: 3532 +Test: [400/517] eta: 0:00:31 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3037 (0.4429) acc1: 96.0938 (84.2328) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2512 data: 0.1566 max mem: 3532 +Test: [410/517] eta: 0:00:28 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.2967 (0.4409) acc1: 98.9583 (84.5492) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2615 data: 0.1684 max mem: 3532 +Test: [420/517] eta: 0:00:26 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3352 (0.4381) acc1: 98.9583 (84.8958) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.3019 data: 0.2066 max mem: 3532 +Test: [430/517] eta: 0:00:23 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3162 (0.4352) acc1: 99.4792 (85.1738) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.2638 data: 0.1671 max mem: 3532 +Test: [440/517] eta: 0:00:20 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3756 (0.4349) acc1: 96.3542 (85.3074) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.2291 data: 0.1349 max mem: 3533 +Test: [450/517] eta: 0:00:18 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3955 (0.4336) acc1: 93.2292 (85.5154) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2599 data: 0.1647 max mem: 3533 +Test: [460/517] eta: 0:00:15 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3812 (0.4327) acc1: 94.7917 (85.6771) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.2774 data: 0.1788 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3712 (0.4304) acc1: 95.5729 (85.9248) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2647 data: 0.1697 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3016 (0.4270) acc1: 96.3542 (86.1043) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2179 data: 0.1269 max mem: 3533 +Test: [490/517] eta: 0:00:07 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3016 (0.4247) acc1: 95.0521 (86.2785) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.1875 data: 0.0977 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.4667 (0.4277) acc1: 83.0729 (85.9250) acc5: 100.0000 (100.0000) acc1_neg: 56.510 acc5_neg: 100.000 time: 0.1835 data: 0.0938 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.5331 (0.4280) acc1: 72.6562 (85.9013) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1838 data: 0.0941 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3862 (0.4268) acc1: 86.7188 (86.0402) acc5: 100.0000 (100.0000) acc1_neg: 99.647 acc5_neg: 100.000 time: 0.1768 data: 0.0849 max mem: 3533 +Test: Total time: 0:02:15 (0.2628 s / it) +* Overall Acc@1 86.040 Acc@5 100.000 + target == real Acc@1 24.054 Acc@5 100.000 + target == fake Acc@1 92.471 Acc@5 100.000 + loss 0.427 + PR-AUC 0.9417 ROC-AUC 0.6594 EER 38.76% (thr=0.6914) +Accuracy of the network on the 198427 test images: 86.0% +[rank0]:[W725 17:12:11.349184809 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0725 17:12:32.916000 89806 torch/distributed/run.py:766] +W0725 17:12:32.916000 89806 torch/distributed/run.py:766] ***************************************** +W0725 17:12:32.916000 89806 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:12:32.916000 89806 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:13:07.708852686 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:13:08.742973168 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:13:08.879365633 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +[rank1]:[W725 17:13:09.103060027 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:13:09.107985147 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +[rank0]:[W725 17:13:09.143613671 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:13:09.148404412 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:13:09.180448794 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed1_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed1_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:31:38 acc1_pos: 24.479 acc5_pos: 100.000 loss: 0.9853 (0.9853) acc1: 24.4792 (24.4792) acc5: 100.0000 (100.0000) time: 4.8811 data: 2.2988 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/389] eta: 0:03:36 acc1_pos: 82.812 acc5_pos: 100.000 loss: 0.7133 (0.8052) acc1: 48.4375 (48.6269) acc5: 100.0000 (100.0000) time: 0.5703 data: 0.2540 max mem: 3531 +Test: [ 20/389] eta: 0:02:22 acc1_pos: 6.510 acc5_pos: 100.000 loss: 0.7133 (0.8309) acc1: 45.5729 (44.7297) acc5: 100.0000 (100.0000) time: 0.1628 data: 0.0709 max mem: 3531 +Test: [ 30/389] eta: 0:01:59 acc1_pos: 48.958 acc5_pos: 100.000 loss: 0.8546 (0.8399) acc1: 22.1354 (39.4909) acc5: 100.0000 (100.0000) time: 0.2039 data: 0.1079 max mem: 3531 +Test: [ 40/389] eta: 0:01:46 acc1_pos: 7.031 acc5_pos: 100.000 loss: 0.8589 (0.8490) acc1: 23.6979 (37.6905) acc5: 100.0000 (100.0000) time: 0.2199 data: 0.1257 max mem: 3531 +Test: [ 50/389] eta: 0:01:33 acc1_pos: 90.365 acc5_pos: 100.000 loss: 0.8589 (0.8429) acc1: 25.0000 (39.7518) acc5: 100.0000 (100.0000) time: 0.1841 data: 0.0922 max mem: 3531 +Test: [ 60/389] eta: 0:01:26 acc1_pos: 3.385 acc5_pos: 100.000 loss: 0.7465 (0.8108) acc1: 48.6979 (43.8397) acc5: 100.0000 (100.0000) time: 0.1763 data: 0.0816 max mem: 3531 +Test: [ 70/389] eta: 0:01:20 acc1_pos: 53.906 acc5_pos: 100.000 loss: 0.7664 (0.8397) acc1: 47.1354 (41.2962) acc5: 100.0000 (100.0000) time: 0.1899 data: 0.0961 max mem: 3531 +Test: [ 80/389] eta: 0:01:14 acc1_pos: 77.604 acc5_pos: 100.000 loss: 1.0543 (0.8515) acc1: 10.1562 (40.5961) acc5: 100.0000 (100.0000) time: 0.1790 data: 0.0871 max mem: 3531 +Test: [ 90/389] eta: 0:01:11 acc1_pos: 13.021 acc5_pos: 100.000 loss: 0.6499 (0.8281) acc1: 57.5521 (42.5166) acc5: 100.0000 (100.0000) time: 0.2005 data: 0.1092 max mem: 3531 +Test: [100/389] eta: 0:01:06 acc1_pos: 21.354 acc5_pos: 100.000 loss: 0.7771 (0.8293) acc1: 41.4062 (41.8472) acc5: 100.0000 (100.0000) time: 0.1817 data: 0.0895 max mem: 3531 +Test: [110/389] eta: 0:01:03 acc1_pos: 99.479 acc5_pos: 100.000 loss: 0.8251 (0.8270) acc1: 34.6354 (42.1077) acc5: 100.0000 (100.0000) time: 0.1711 data: 0.0783 max mem: 3531 +Test: [120/389] eta: 0:01:00 acc1_pos: 65.885 acc5_pos: 100.000 loss: 0.7818 (0.8217) acc1: 40.6250 (42.7320) acc5: 100.0000 (100.0000) time: 0.2037 data: 0.1120 max mem: 3531 +Test: [130/389] eta: 0:00:57 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.6056 (0.8044) acc1: 65.1042 (44.5670) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1895 data: 0.0981 max mem: 3531 +Test: [140/389] eta: 0:00:54 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.4371 (0.7782) acc1: 86.1979 (47.4217) acc5: 100.0000 (100.0000) acc1_neg: 44.792 acc5_neg: 100.000 time: 0.1808 data: 0.0889 max mem: 3531 +Test: [150/389] eta: 0:00:51 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.4091 (0.7646) acc1: 91.1458 (49.0635) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1837 data: 0.0895 max mem: 3531 +Test: [160/389] eta: 0:00:49 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.4368 (0.7415) acc1: 86.9792 (51.5253) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1748 data: 0.0813 max mem: 3531 +Test: [170/389] eta: 0:00:46 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.3589 (0.7210) acc1: 97.6562 (54.0753) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1766 data: 0.0829 max mem: 3531 +Test: [180/389] eta: 0:00:43 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.3614 (0.7068) acc1: 95.0521 (55.8227) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1787 data: 0.0851 max mem: 3531 +Test: [190/389] eta: 0:00:41 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.4510 (0.6958) acc1: 89.5833 (57.2303) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1700 data: 0.0787 max mem: 3531 +Test: [200/389] eta: 0:00:38 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.4454 (0.6835) acc1: 90.3646 (58.7155) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.1617 data: 0.0706 max mem: 3531 +Test: [210/389] eta: 0:00:36 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.4032 (0.6694) acc1: 94.2708 (60.2019) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1644 data: 0.0733 max mem: 3532 +Test: [220/389] eta: 0:00:34 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.2539 (0.6519) acc1: 99.4792 (61.9532) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1824 data: 0.0879 max mem: 3532 +Test: [230/389] eta: 0:00:32 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.3202 (0.6410) acc1: 98.4375 (63.3917) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1920 data: 0.0962 max mem: 3532 +Test: [240/389] eta: 0:00:30 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.3878 (0.6303) acc1: 96.6146 (64.6395) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1843 data: 0.0916 max mem: 3532 +Test: [250/389] eta: 0:00:28 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.3472 (0.6183) acc1: 98.4375 (66.0141) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1900 data: 0.0961 max mem: 3532 +Test: [260/389] eta: 0:00:25 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.3433 (0.6146) acc1: 99.2188 (66.5858) acc5: 100.0000 (100.0000) acc1_neg: 25.521 acc5_neg: 100.000 time: 0.1746 data: 0.0801 max mem: 3532 +Test: [270/389] eta: 0:00:23 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.4932 (0.6181) acc1: 75.5208 (66.0642) acc5: 100.0000 (100.0000) acc1_neg: 75.521 acc5_neg: 100.000 time: 0.1650 data: 0.0719 max mem: 3532 +Test: [280/389] eta: 0:00:21 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.6467 (0.6186) acc1: 59.3750 (65.9762) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1774 data: 0.0834 max mem: 3532 +Test: [290/389] eta: 0:00:19 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.6055 (0.6133) acc1: 69.5312 (66.6497) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1799 data: 0.0866 max mem: 3532 +Test: [300/389] eta: 0:00:17 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.4189 (0.6074) acc1: 92.4479 (67.5569) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.1953 data: 0.1035 max mem: 3532 +Test: [310/389] eta: 0:00:15 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.4811 (0.6079) acc1: 85.9375 (67.5250) acc5: 100.0000 (100.0000) acc1_neg: 54.948 acc5_neg: 100.000 time: 0.1979 data: 0.1057 max mem: 3532 +Test: [320/389] eta: 0:00:13 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.4719 (0.6018) acc1: 85.4167 (68.2162) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1948 data: 0.1017 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.4693 (0.5981) acc1: 89.5833 (68.8208) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1918 data: 0.0999 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.4287 (0.5893) acc1: 91.1458 (69.6351) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1996 data: 0.1082 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.2791 (0.5967) acc1: 97.6562 (68.9444) acc5: 100.0000 (100.0000) acc1_neg: 2.344 acc5_neg: 100.000 time: 0.1909 data: 0.0990 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.6213 (0.6003) acc1: 61.7188 (68.4788) acc5: 100.0000 (100.0000) acc1_neg: 70.573 acc5_neg: 100.000 time: 0.1617 data: 0.0711 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.5822 (0.5960) acc1: 68.2292 (68.8321) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1469 data: 0.0565 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.3976 (0.5914) acc1: 82.2917 (69.1403) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1339 data: 0.0437 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.3344 (0.5880) acc1: 88.8021 (69.5115) acc5: 100.0000 (100.0000) acc1_neg: 74.863 acc5_neg: 100.000 time: 0.1319 data: 0.0385 max mem: 3532 +Test: Total time: 0:01:14 (0.1918 s / it) +* Overall Acc@1 69.512 Acc@5 100.000 + target == real Acc@1 42.853 Acc@5 100.000 + target == fake Acc@1 81.903 Acc@5 100.000 + loss 0.588 + PR-AUC 0.8086 ROC-AUC 0.6924 EER 35.78% (thr=0.6147) +Accuracy of the network on the 149358 test images: 69.5% +[rank0]:[W725 17:14:32.274289177 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 17:14:53.720000 90382 torch/distributed/run.py:766] +W0725 17:14:53.720000 90382 torch/distributed/run.py:766] ***************************************** +W0725 17:14:53.720000 90382 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:14:53.720000 90382 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:15:27.187799819 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 17:15:27.227604794 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:15:27.504310240 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:15:27.650046893 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:15:27.755133213 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:15:27.858050843 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:15:27.887328437 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:15:27.890835144 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed1_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed1_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:33:18 acc1_pos: 80.208 acc5_pos: 100.000 loss: 0.4223 (0.4223) acc1: 80.2083 (80.2083) acc5: 100.0000 (100.0000) time: 6.0025 data: 4.5376 max mem: 3531 +Test: [ 10/333] eta: 0:03:22 acc1_pos: 82.812 acc5_pos: 100.000 loss: 0.3405 (0.3341) acc1: 87.5000 (87.2159) acc5: 100.0000 (100.0000) time: 0.6277 data: 0.4129 max mem: 3531 +Test: [ 20/333] eta: 0:01:57 acc1_pos: 92.708 acc5_pos: 100.000 loss: 0.3340 (0.3332) acc1: 87.2396 (86.0987) acc5: 100.0000 (100.0000) time: 0.0943 data: 0.0004 max mem: 3531 +Test: [ 30/333] eta: 0:01:36 acc1_pos: 40.365 acc5_pos: 100.000 loss: 0.4042 (0.4939) acc1: 72.1354 (73.1183) acc5: 100.0000 (100.0000) time: 0.1504 data: 0.0553 max mem: 3531 +Test: [ 40/333] eta: 0:01:30 acc1_pos: 45.573 acc5_pos: 100.000 loss: 0.8626 (0.5783) acc1: 42.4479 (66.1331) acc5: 100.0000 (100.0000) time: 0.2409 data: 0.1484 max mem: 3531 +Test: [ 50/333] eta: 0:01:24 acc1_pos: 49.740 acc5_pos: 100.000 loss: 0.8460 (0.6316) acc1: 43.7500 (61.7392) acc5: 100.0000 (100.0000) time: 0.2634 data: 0.1681 max mem: 3531 +Test: [ 60/333] eta: 0:01:17 acc1_pos: 33.333 acc5_pos: 100.000 loss: 0.8429 (0.6650) acc1: 45.8333 (59.1359) acc5: 100.0000 (100.0000) time: 0.2337 data: 0.1366 max mem: 3531 +Test: [ 70/333] eta: 0:01:11 acc1_pos: 37.500 acc5_pos: 100.000 loss: 0.8429 (0.6902) acc1: 46.6146 (57.4090) acc5: 100.0000 (100.0000) time: 0.2007 data: 0.1064 max mem: 3531 +Test: [ 80/333] eta: 0:01:07 acc1_pos: 47.135 acc5_pos: 100.000 loss: 0.8195 (0.7097) acc1: 44.5312 (55.9992) acc5: 100.0000 (100.0000) time: 0.2049 data: 0.1125 max mem: 3531 +Test: [ 90/333] eta: 0:01:02 acc1_pos: 53.646 acc5_pos: 100.000 loss: 0.7761 (0.7156) acc1: 51.3021 (55.5746) acc5: 100.0000 (100.0000) time: 0.2191 data: 0.1244 max mem: 3531 +Test: [100/333] eta: 0:00:59 acc1_pos: 58.333 acc5_pos: 100.000 loss: 0.7711 (0.7243) acc1: 53.3854 (55.0278) acc5: 100.0000 (100.0000) time: 0.2168 data: 0.1205 max mem: 3531 +Test: [110/333] eta: 0:00:56 acc1_pos: 52.083 acc5_pos: 100.000 loss: 0.7694 (0.7287) acc1: 51.5625 (54.6734) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1312 max mem: 3531 +Test: [120/333] eta: 0:00:52 acc1_pos: 42.708 acc5_pos: 100.000 loss: 0.7685 (0.7321) acc1: 51.5625 (54.3862) acc5: 100.0000 (100.0000) time: 0.2003 data: 0.1060 max mem: 3531 +Test: [130/333] eta: 0:00:50 acc1_pos: 75.260 acc5_pos: 100.000 loss: 0.7581 (0.7333) acc1: 50.7812 (54.4092) acc5: 100.0000 (100.0000) time: 0.2199 data: 0.1244 max mem: 3531 +Test: [140/333] eta: 0:00:47 acc1_pos: 67.188 acc5_pos: 100.000 loss: 0.7423 (0.7342) acc1: 52.3438 (54.2978) acc5: 100.0000 (100.0000) time: 0.2386 data: 0.1430 max mem: 3531 +Test: [150/333] eta: 0:00:44 acc1_pos: 54.948 acc5_pos: 100.000 loss: 0.7593 (0.7383) acc1: 51.3021 (53.8114) acc5: 100.0000 (100.0000) time: 0.1968 data: 0.0976 max mem: 3531 +Test: [160/333] eta: 0:00:41 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.7842 (0.7441) acc1: 48.9583 (53.5019) acc5: 100.0000 (100.0000) acc1_neg: 52.344 acc5_neg: 100.000 time: 0.2013 data: 0.1051 max mem: 3531 +Test: [170/333] eta: 0:00:38 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.8174 (0.7475) acc1: 49.2188 (53.4555) acc5: 100.0000 (100.0000) acc1_neg: 59.375 acc5_neg: 100.000 time: 0.2181 data: 0.1270 max mem: 3531 +Test: [180/333] eta: 0:00:35 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.7515 (0.7450) acc1: 56.5104 (53.8372) acc5: 100.0000 (100.0000) acc1_neg: 65.885 acc5_neg: 100.000 time: 0.1775 data: 0.0829 max mem: 3531 +Test: [190/333] eta: 0:00:33 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.6718 (0.7400) acc1: 59.1146 (54.3575) acc5: 100.0000 (100.0000) acc1_neg: 56.250 acc5_neg: 100.000 time: 0.1754 data: 0.0804 max mem: 3531 +Test: [200/333] eta: 0:00:30 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.6405 (0.7364) acc1: 59.8958 (54.7821) acc5: 100.0000 (100.0000) acc1_neg: 59.896 acc5_neg: 100.000 time: 0.2249 data: 0.1311 max mem: 3531 +Test: [210/333] eta: 0:00:28 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.7020 (0.7373) acc1: 59.6354 (54.9973) acc5: 100.0000 (100.0000) acc1_neg: 57.031 acc5_neg: 100.000 time: 0.2047 data: 0.1091 max mem: 3532 +Test: [220/333] eta: 0:00:25 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.7577 (0.7401) acc1: 58.8542 (54.9515) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.1349 data: 0.0378 max mem: 3532 +Test: [230/333] eta: 0:00:22 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.5294 (0.7297) acc1: 75.0000 (55.9918) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.1446 data: 0.0474 max mem: 3532 +Test: [240/333] eta: 0:00:20 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.5294 (0.7235) acc1: 76.5625 (56.5936) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.2449 data: 0.1498 max mem: 3532 +Test: [250/333] eta: 0:00:18 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.6050 (0.7197) acc1: 66.6667 (56.9752) acc5: 100.0000 (100.0000) acc1_neg: 58.073 acc5_neg: 100.000 time: 0.2453 data: 0.1507 max mem: 3532 +Test: [260/333] eta: 0:00:16 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.6479 (0.7180) acc1: 64.8438 (57.2148) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.1799 data: 0.0851 max mem: 3532 +Test: [270/333] eta: 0:00:13 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.6195 (0.7135) acc1: 65.3646 (57.6280) acc5: 100.0000 (100.0000) acc1_neg: 66.927 acc5_neg: 100.000 time: 0.1995 data: 0.1049 max mem: 3532 +Test: [280/333] eta: 0:00:11 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.6139 (0.7137) acc1: 65.3646 (57.5947) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.2199 data: 0.1248 max mem: 3532 +Test: [290/333] eta: 0:00:09 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.7915 (0.7165) acc1: 49.7396 (57.3821) acc5: 100.0000 (100.0000) acc1_neg: 58.073 acc5_neg: 100.000 time: 0.1689 data: 0.0729 max mem: 3532 +Test: [300/333] eta: 0:00:07 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.7844 (0.7187) acc1: 52.6042 (57.2865) acc5: 100.0000 (100.0000) acc1_neg: 59.896 acc5_neg: 100.000 time: 0.1541 data: 0.0613 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.7186 (0.7143) acc1: 57.8125 (57.7263) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.1565 data: 0.0667 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.5509 (0.7103) acc1: 73.1771 (58.1678) acc5: 100.0000 (100.0000) acc1_neg: 44.792 acc5_neg: 100.000 time: 0.1125 data: 0.0228 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.8247 (0.7157) acc1: 50.5208 (57.8345) acc5: 100.0000 (100.0000) acc1_neg: 48.177 acc5_neg: 100.000 time: 0.1010 data: 0.0113 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.8606 (0.7168) acc1: 48.1771 (57.7638) acc5: 100.0000 (100.0000) acc1_neg: 46.197 acc5_neg: 100.000 time: 0.1096 data: 0.0113 max mem: 3532 +Test: Total time: 0:01:09 (0.2079 s / it) +* Overall Acc@1 57.764 Acc@5 100.000 + target == real Acc@1 53.651 Acc@5 100.000 + target == fake Acc@1 61.425 Acc@5 100.000 + loss 0.717 + PR-AUC 0.6121 ROC-AUC 0.6052 EER 42.53% (thr=0.5244) +Accuracy of the network on the 127843 test images: 57.8% +[rank0]:[W725 17:16:45.060510976 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 17:17:06.125000 90940 torch/distributed/run.py:766] +W0725 17:17:06.125000 90940 torch/distributed/run.py:766] ***************************************** +W0725 17:17:06.125000 90940 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:17:06.125000 90940 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:17:44.278644024 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:17:44.473268824 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:17:44.963977719 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:17:45.130839910 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 17:17:45.149929774 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:17:45.215952474 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank4]:[W725 17:17:45.223966712 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:17:45.329484088 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed1_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed1_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:19:34 acc1_pos: 94.531 acc5_pos: 100.000 loss: 0.2751 (0.2751) acc1: 94.5312 (94.5312) acc5: 100.0000 (100.0000) time: 4.9994 data: 2.9445 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/235] eta: 0:02:00 acc1_pos: 93.750 acc5_pos: 100.000 loss: 0.2744 (0.2752) acc1: 93.7500 (93.4186) acc5: 100.0000 (100.0000) time: 0.5362 data: 0.2680 max mem: 3531 +Test: [ 20/235] eta: 0:01:09 acc1_pos: 91.927 acc5_pos: 100.000 loss: 0.2675 (0.2732) acc1: 93.4896 (93.4524) acc5: 100.0000 (100.0000) time: 0.0910 data: 0.0004 max mem: 3531 +Test: [ 30/235] eta: 0:01:00 acc1_pos: 94.792 acc5_pos: 100.000 loss: 0.2648 (0.2712) acc1: 93.7500 (93.5484) acc5: 100.0000 (100.0000) time: 0.1635 data: 0.0719 max mem: 3531 +Test: [ 40/235] eta: 0:01:00 acc1_pos: 93.229 acc5_pos: 100.000 loss: 0.2678 (0.2714) acc1: 93.7500 (93.6039) acc5: 100.0000 (100.0000) time: 0.2934 data: 0.2017 max mem: 3531 +Test: [ 50/235] eta: 0:00:54 acc1_pos: 91.667 acc5_pos: 100.000 loss: 0.2715 (0.2721) acc1: 93.4896 (93.5815) acc5: 100.0000 (100.0000) time: 0.2897 data: 0.1985 max mem: 3531 +Test: [ 60/235] eta: 0:00:47 acc1_pos: 94.271 acc5_pos: 100.000 loss: 0.2756 (0.2734) acc1: 93.2292 (93.5280) acc5: 100.0000 (100.0000) time: 0.2018 data: 0.1097 max mem: 3531 +Test: [ 70/235] eta: 0:00:42 acc1_pos: 92.308 acc5_pos: 100.000 loss: 0.2957 (0.3868) acc1: 90.1042 (85.5560) acc5: 100.0000 (100.0000) acc1_neg: 26.042 acc5_neg: 100.000 time: 0.1628 data: 0.0687 max mem: 3531 +Test: [ 80/235] eta: 0:00:39 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.1831 (0.4866) acc1: 29.6875 (78.6233) acc5: 100.0000 (100.0000) acc1_neg: 28.125 acc5_neg: 100.000 time: 0.2070 data: 0.1143 max mem: 3531 +Test: [ 90/235] eta: 0:00:36 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.2004 (0.5643) acc1: 29.1667 (73.2000) acc5: 100.0000 (100.0000) acc1_neg: 29.948 acc5_neg: 100.000 time: 0.2412 data: 0.1481 max mem: 3531 +Test: [100/235] eta: 0:00:33 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.2004 (0.6248) acc1: 29.6875 (68.9485) acc5: 100.0000 (100.0000) acc1_neg: 27.604 acc5_neg: 100.000 time: 0.1937 data: 0.1006 max mem: 3531 +Test: [110/235] eta: 0:00:29 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.1733 (0.6756) acc1: 29.6875 (65.3411) acc5: 100.0000 (100.0000) acc1_neg: 30.990 acc5_neg: 100.000 time: 0.1697 data: 0.0758 max mem: 3531 +Test: [120/235] eta: 0:00:27 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.2344 (0.7460) acc1: 24.7396 (60.9483) acc5: 100.0000 (100.0000) acc1_neg: 13.281 acc5_neg: 100.000 time: 0.1887 data: 0.0949 max mem: 3531 +Test: [130/235] eta: 0:00:24 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.5439 (0.8078) acc1: 10.9375 (57.1028) acc5: 100.0000 (100.0000) acc1_neg: 11.458 acc5_neg: 100.000 time: 0.1937 data: 0.1009 max mem: 3531 +Test: [140/235] eta: 0:00:21 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.5732 (0.8635) acc1: 10.1562 (53.7622) acc5: 100.0000 (100.0000) acc1_neg: 7.812 acc5_neg: 100.000 time: 0.1766 data: 0.0831 max mem: 3531 +Test: [150/235] eta: 0:00:19 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.5732 (0.9094) acc1: 9.8958 (50.8744) acc5: 100.0000 (100.0000) acc1_neg: 11.719 acc5_neg: 100.000 time: 0.1799 data: 0.0865 max mem: 3531 +Test: [160/235] eta: 0:00:16 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.5594 (0.9501) acc1: 10.6771 (48.4019) acc5: 100.0000 (100.0000) acc1_neg: 12.760 acc5_neg: 100.000 time: 0.1849 data: 0.0924 max mem: 3531 +Test: [170/235] eta: 0:00:14 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.5072 (0.9526) acc1: 12.7604 (47.8710) acc5: 100.0000 (100.0000) acc1_neg: 38.542 acc5_neg: 100.000 time: 0.1802 data: 0.0896 max mem: 3531 +Test: [180/235] eta: 0:00:12 acc1_pos: 92.308 acc5_pos: 100.000 loss: 0.9272 (0.9508) acc1: 42.4479 (47.6275) acc5: 100.0000 (100.0000) acc1_neg: 43.750 acc5_neg: 100.000 time: 0.1882 data: 0.0965 max mem: 3531 +Test: [190/235] eta: 0:00:09 acc1_pos: 92.308 acc5_pos: 100.000 loss: 0.9193 (0.9486) acc1: 43.7500 (47.4408) acc5: 100.0000 (100.0000) acc1_neg: 45.312 acc5_neg: 100.000 time: 0.1897 data: 0.0975 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 92.308 acc5_pos: 100.000 loss: 0.9193 (0.9478) acc1: 42.9688 (47.1743) acc5: 100.0000 (100.0000) acc1_neg: 41.146 acc5_neg: 100.000 time: 0.1663 data: 0.0740 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 92.308 acc5_pos: 100.000 loss: 0.9263 (0.9462) acc1: 42.7083 (46.9910) acc5: 100.0000 (100.0000) acc1_neg: 42.969 acc5_neg: 100.000 time: 0.1556 data: 0.0639 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 92.308 acc5_pos: 100.000 loss: 0.9117 (0.9447) acc1: 43.7500 (46.8691) acc5: 100.0000 (100.0000) acc1_neg: 43.490 acc5_neg: 100.000 time: 0.1632 data: 0.0728 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 92.308 acc5_pos: 100.000 loss: 0.9032 (0.9429) acc1: 44.2708 (46.7938) acc5: 100.0000 (100.0000) acc1_neg: 46.354 acc5_neg: 100.000 time: 0.1511 data: 0.0611 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 92.308 acc5_pos: 100.000 loss: 0.9084 (0.9431) acc1: 44.0104 (46.7417) acc5: 100.0000 (100.0000) acc1_neg: 52.174 acc5_neg: 100.000 time: 0.1292 data: 0.0419 max mem: 3532 +Test: Total time: 0:00:48 (0.2044 s / it) +* Overall Acc@1 46.742 Acc@5 100.000 + target == real Acc@1 93.516 Acc@5 100.000 + target == fake Acc@1 29.825 Acc@5 100.000 + loss 0.943 + PR-AUC 0.8761 ROC-AUC 0.7187 EER 34.00% (thr=0.2389) +Accuracy of the network on the 89879 test images: 46.7% +[rank0]:[W725 17:18:47.233945134 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 17:19:04.114000 91498 torch/distributed/run.py:766] +W0725 17:19:04.114000 91498 torch/distributed/run.py:766] ***************************************** +W0725 17:19:04.114000 91498 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:19:04.114000 91498 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W725 17:19:32.062855961 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:19:32.197221764 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:19:32.642390609 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:19:32.905721588 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:19:33.055756402 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:19:33.103582079 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W725 17:19:33.103628186 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:19:33.139593624 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed1_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed1_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:31:26 acc1_pos: 35.417 acc5_pos: 100.000 loss: 0.8183 (0.8183) acc1: 35.4167 (35.4167) acc5: 100.0000 (100.0000) time: 7.0135 data: 5.4260 max mem: 3531 +Test: [ 10/269] eta: 0:03:07 acc1_pos: 61.458 acc5_pos: 100.000 loss: 0.6768 (0.6813) acc1: 55.7292 (57.0786) acc5: 100.0000 (100.0000) time: 0.7252 data: 0.4937 max mem: 3531 +Test: [ 20/269] eta: 0:01:46 acc1_pos: 79.167 acc5_pos: 100.000 loss: 0.6529 (0.6398) acc1: 59.3750 (62.1032) acc5: 100.0000 (100.0000) time: 0.0981 data: 0.0005 max mem: 3531 +Test: [ 30/269] eta: 0:01:36 acc1_pos: 65.104 acc5_pos: 100.000 loss: 0.5707 (0.6191) acc1: 65.1042 (64.5497) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1345 max mem: 3531 +Test: [ 40/269] eta: 0:01:23 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.5230 (0.6034) acc1: 71.6146 (67.6321) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.3015 data: 0.2105 max mem: 3531 +Test: [ 50/269] eta: 0:01:13 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.5851 (0.6083) acc1: 71.6146 (67.3356) acc5: 100.0000 (100.0000) acc1_neg: 44.271 acc5_neg: 100.000 time: 0.2249 data: 0.1333 max mem: 3531 +Test: [ 60/269] eta: 0:01:06 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.6283 (0.6168) acc1: 61.1979 (66.1629) acc5: 100.0000 (100.0000) acc1_neg: 59.896 acc5_neg: 100.000 time: 0.2183 data: 0.1243 max mem: 3531 +Test: [ 70/269] eta: 0:00:59 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.6570 (0.6238) acc1: 60.4167 (65.2472) acc5: 100.0000 (100.0000) acc1_neg: 60.938 acc5_neg: 100.000 time: 0.2099 data: 0.1138 max mem: 3531 +Test: [ 80/269] eta: 0:00:54 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.6356 (0.6213) acc1: 63.5417 (65.7118) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.1951 data: 0.1008 max mem: 3531 +Test: [ 90/269] eta: 0:00:50 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.6414 (0.6233) acc1: 63.5417 (65.4504) acc5: 100.0000 (100.0000) acc1_neg: 66.146 acc5_neg: 100.000 time: 0.2278 data: 0.1347 max mem: 3531 +Test: [100/269] eta: 0:00:46 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.6631 (0.6327) acc1: 58.3333 (64.3410) acc5: 100.0000 (100.0000) acc1_neg: 36.979 acc5_neg: 100.000 time: 0.2247 data: 0.1316 max mem: 3531 +Test: [110/269] eta: 0:00:41 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.7178 (0.6428) acc1: 52.3438 (62.9106) acc5: 100.0000 (100.0000) acc1_neg: 47.396 acc5_neg: 100.000 time: 0.1591 data: 0.0644 max mem: 3531 +Test: [120/269] eta: 0:00:37 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.7504 (0.6584) acc1: 46.0938 (61.6004) acc5: 100.0000 (100.0000) acc1_neg: 36.458 acc5_neg: 100.000 time: 0.1415 data: 0.0462 max mem: 3531 +Test: [130/269] eta: 0:00:35 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.8144 (0.6696) acc1: 43.2292 (60.3531) acc5: 100.0000 (100.0000) acc1_neg: 43.490 acc5_neg: 100.000 time: 0.2036 data: 0.1116 max mem: 3531 +Test: [140/269] eta: 0:00:32 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.8147 (0.6858) acc1: 39.0625 (58.5217) acc5: 100.0000 (100.0000) acc1_neg: 40.625 acc5_neg: 100.000 time: 0.2233 data: 0.1292 max mem: 3531 +Test: [150/269] eta: 0:00:29 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.7994 (0.6910) acc1: 40.3646 (57.7849) acc5: 100.0000 (100.0000) acc1_neg: 46.875 acc5_neg: 100.000 time: 0.2005 data: 0.1030 max mem: 3531 +Test: [160/269] eta: 0:00:26 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.7316 (0.6900) acc1: 51.8229 (57.9645) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.1913 data: 0.0951 max mem: 3531 +Test: [170/269] eta: 0:00:23 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.6659 (0.6909) acc1: 59.6354 (57.9785) acc5: 100.0000 (100.0000) acc1_neg: 71.094 acc5_neg: 100.000 time: 0.1673 data: 0.0714 max mem: 3531 +Test: [180/269] eta: 0:00:20 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.6495 (0.6882) acc1: 60.9375 (58.4844) acc5: 100.0000 (100.0000) acc1_neg: 60.938 acc5_neg: 100.000 time: 0.1735 data: 0.0795 max mem: 3531 +Test: [190/269] eta: 0:00:18 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.6349 (0.6893) acc1: 60.9375 (58.4056) acc5: 100.0000 (100.0000) acc1_neg: 47.656 acc5_neg: 100.000 time: 0.2340 data: 0.1396 max mem: 3531 +Test: [200/269] eta: 0:00:15 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.7370 (0.6937) acc1: 50.7812 (57.8993) acc5: 100.0000 (100.0000) acc1_neg: 56.771 acc5_neg: 100.000 time: 0.2119 data: 0.1171 max mem: 3531 +Test: [210/269] eta: 0:00:13 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.7379 (0.6958) acc1: 48.1771 (57.6200) acc5: 100.0000 (100.0000) acc1_neg: 53.385 acc5_neg: 100.000 time: 0.1622 data: 0.0670 max mem: 3532 +Test: [220/269] eta: 0:00:11 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.6067 (0.6943) acc1: 53.3854 (57.8490) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.1790 data: 0.0811 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.6000 (0.6921) acc1: 64.8438 (58.0673) acc5: 100.0000 (100.0000) acc1_neg: 58.594 acc5_neg: 100.000 time: 0.1973 data: 0.1008 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.6597 (0.6938) acc1: 58.5938 (57.8979) acc5: 100.0000 (100.0000) acc1_neg: 61.198 acc5_neg: 100.000 time: 0.1932 data: 0.0991 max mem: 3532 +Test: [250/269] eta: 0:00:04 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.7723 (0.7013) acc1: 48.4375 (57.0053) acc5: 100.0000 (100.0000) acc1_neg: 15.365 acc5_neg: 100.000 time: 0.1574 data: 0.0662 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.9100 (0.7087) acc1: 25.5208 (56.1103) acc5: 100.0000 (100.0000) acc1_neg: 56.250 acc5_neg: 100.000 time: 0.1344 data: 0.0446 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.7695 (0.7089) acc1: 37.5000 (55.9835) acc5: 100.0000 (100.0000) acc1_neg: 65.879 acc5_neg: 100.000 time: 0.1380 data: 0.0445 max mem: 3532 +Test: Total time: 0:00:57 (0.2144 s / it) +* Overall Acc@1 55.983 Acc@5 100.000 + target == real Acc@1 64.036 Acc@5 100.000 + target == fake Acc@1 54.885 Acc@5 100.000 + loss 0.709 + PR-AUC 0.9225 ROC-AUC 0.6322 EER 40.63% (thr=0.4792) +Accuracy of the network on the 103293 test images: 56.0% +[rank0]:[W725 17:20:41.935617527 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-25T17:20:48Z] eval_done tag=fixed500k_fake_n32_seed1_resampled +[2026-07-25T17:20:48Z] eval_start tag=fixed500k_fake_n32_seed2_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed2_resampled_n111-046-081_20260725_112559/best_checkpoint.pth +started_utc=2026-07-25T17:20:48Z +hostname=n116-042-186 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed2_resampled_n111-046-081_20260725_112559/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed2_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0725 17:21:02.751000 92058 torch/distributed/run.py:766] +W0725 17:21:02.751000 92058 torch/distributed/run.py:766] ***************************************** +W0725 17:21:02.751000 92058 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:21:02.751000 92058 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:21:38.631674182 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 17:21:39.190222400 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:21:39.219363316 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:21:39.763419776 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:21:39.002652785 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:21:39.036226971 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:21:40.061903159 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:21:40.077412655 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed2_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed2_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:19:05 acc1_pos: 44.792 acc5_pos: 100.000 loss: 0.7534 (0.7534) acc1: 44.7917 (44.7917) acc5: 100.0000 (100.0000) time: 4.8945 data: 3.0401 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/234] eta: 0:01:58 acc1_pos: 25.781 acc5_pos: 100.000 loss: 0.7534 (0.7889) acc1: 44.7917 (46.3305) acc5: 100.0000 (100.0000) time: 0.5269 data: 0.2767 max mem: 3531 +Test: [ 20/234] eta: 0:01:08 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.8349 (0.7521) acc1: 39.3229 (47.9787) acc5: 100.0000 (100.0000) time: 0.0901 data: 0.0004 max mem: 3531 +Test: [ 30/234] eta: 0:00:52 acc1_pos: 80.208 acc5_pos: 100.000 loss: 0.1623 (0.5707) acc1: 96.8750 (63.0796) acc5: 100.0000 (100.0000) time: 0.1121 data: 0.0212 max mem: 3531 +Test: [ 40/234] eta: 0:00:49 acc1_pos: 99.219 acc5_pos: 100.000 loss: 0.1429 (0.5183) acc1: 97.6562 (67.8227) acc5: 100.0000 (100.0000) time: 0.1870 data: 0.0957 max mem: 3531 +Test: [ 50/234] eta: 0:00:45 acc1_pos: 98.958 acc5_pos: 100.000 loss: 0.1172 (0.4386) acc1: 97.9167 (73.8205) acc5: 100.0000 (100.0000) time: 0.2265 data: 0.1335 max mem: 3531 +Test: [ 60/234] eta: 0:00:42 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.1011 (0.3818) acc1: 98.9583 (78.0951) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2249 data: 0.1304 max mem: 3531 +Test: [ 70/234] eta: 0:00:38 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.0935 (0.3941) acc1: 100.0000 (78.0773) acc5: 100.0000 (100.0000) acc1_neg: 66.146 acc5_neg: 100.000 time: 0.1983 data: 0.1001 max mem: 3531 +Test: [ 80/234] eta: 0:00:34 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.6258 (0.4051) acc1: 68.7500 (78.2054) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1716 data: 0.0736 max mem: 3531 +Test: [ 90/234] eta: 0:00:32 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.2336 (0.3877) acc1: 88.2812 (79.6961) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.2084 data: 0.1151 max mem: 3531 +Test: [100/234] eta: 0:00:29 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.3148 (0.4043) acc1: 84.3750 (79.0713) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2031 data: 0.1096 max mem: 3531 +Test: [110/234] eta: 0:00:27 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.5946 (0.4122) acc1: 69.7917 (79.0353) acc5: 100.0000 (100.0000) acc1_neg: 62.760 acc5_neg: 100.000 time: 0.1838 data: 0.0914 max mem: 3531 +Test: [120/234] eta: 0:00:24 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4578 (0.4260) acc1: 74.7396 (78.6372) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.1966 data: 0.1059 max mem: 3531 +Test: [130/234] eta: 0:00:22 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4276 (0.4276) acc1: 83.3333 (78.9778) acc5: 100.0000 (100.0000) acc1_neg: 69.010 acc5_neg: 100.000 time: 0.1932 data: 0.1028 max mem: 3531 +Test: [140/234] eta: 0:00:20 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.5573 (0.4759) acc1: 70.5729 (76.6548) acc5: 100.0000 (100.0000) acc1_neg: 20.833 acc5_neg: 100.000 time: 0.1932 data: 0.1005 max mem: 3531 +Test: [150/234] eta: 0:00:17 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.6276 (0.4905) acc1: 65.1042 (76.0900) acc5: 100.0000 (100.0000) acc1_neg: 77.083 acc5_neg: 100.000 time: 0.1910 data: 0.0939 max mem: 3531 +Test: [160/234] eta: 0:00:15 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4879 (0.4852) acc1: 77.0833 (76.6046) acc5: 100.0000 (100.0000) acc1_neg: 73.958 acc5_neg: 100.000 time: 0.1756 data: 0.0795 max mem: 3531 +Test: [170/234] eta: 0:00:13 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4879 (0.4859) acc1: 76.8229 (76.7087) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1682 data: 0.0755 max mem: 3531 +Test: [180/234] eta: 0:00:11 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.5412 (0.4994) acc1: 73.9583 (76.0258) acc5: 100.0000 (100.0000) acc1_neg: 32.031 acc5_neg: 100.000 time: 0.1809 data: 0.0898 max mem: 3531 +Test: [190/234] eta: 0:00:09 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.8633 (0.5214) acc1: 50.2604 (74.5937) acc5: 100.0000 (100.0000) acc1_neg: 50.781 acc5_neg: 100.000 time: 0.2071 data: 0.1134 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.8633 (0.5158) acc1: 50.2604 (74.9521) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2046 data: 0.1112 max mem: 3531 +Test: [210/234] eta: 0:00:04 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.1291 (0.4980) acc1: 99.4792 (76.0849) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1877 data: 0.0957 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.1551 (0.4836) acc1: 97.3958 (76.9608) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.1878 data: 0.0957 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.3238 (0.4920) acc1: 63.5417 (76.2897) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.1651 data: 0.0745 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.6551 (0.4941) acc1: 63.0208 (76.1337) acc5: 100.0000 (100.0000) acc1_neg: 65.370 acc5_neg: 100.000 time: 0.1521 data: 0.0606 max mem: 3532 +Test: Total time: 0:00:47 (0.2015 s / it) +* Overall Acc@1 76.134 Acc@5 100.000 + target == real Acc@1 73.998 Acc@5 100.000 + target == fake Acc@1 76.736 Acc@5 100.000 + loss 0.494 + PR-AUC 0.9583 ROC-AUC 0.8616 EER 24.39% (thr=0.5210) +Accuracy of the network on the 89729 test images: 76.1% +[rank0]:[W725 17:22:43.624251285 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 17:23:05.065000 92615 torch/distributed/run.py:766] +W0725 17:23:05.065000 92615 torch/distributed/run.py:766] ***************************************** +W0725 17:23:05.065000 92615 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:23:05.065000 92615 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W725 17:23:42.063077499 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:23:42.259774574 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:23:42.417516050 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:23:42.881019244 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:23:42.039308995 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:23:43.293043596 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:23:43.355782428 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:23:43.476084711 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed2_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed2_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:44:09 acc1_pos: 39.844 acc5_pos: 100.000 loss: 0.8949 (0.8949) acc1: 39.8438 (39.8438) acc5: 100.0000 (100.0000) time: 7.1425 data: 5.3954 max mem: 3531 +Test: [ 10/2387] eta: 0:29:06 acc1_pos: 19.010 acc5_pos: 100.000 loss: 0.8949 (0.9268) acc1: 37.5000 (35.4640) acc5: 100.0000 (100.0000) time: 0.7348 data: 0.4909 max mem: 3531 +Test: [ 20/2387] eta: 0:17:03 acc1_pos: 23.698 acc5_pos: 100.000 loss: 0.9423 (0.9533) acc1: 27.0833 (33.5193) acc5: 100.0000 (100.0000) time: 0.0970 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:14:40 acc1_pos: 16.667 acc5_pos: 100.000 loss: 1.0040 (0.9779) acc1: 25.7812 (31.5608) acc5: 100.0000 (100.0000) time: 0.1749 data: 0.0791 max mem: 3531 +Test: [ 40/2387] eta: 0:13:14 acc1_pos: 29.948 acc5_pos: 100.000 loss: 1.0518 (1.0016) acc1: 21.8750 (28.8745) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1459 max mem: 3531 +Test: [ 50/2387] eta: 0:12:16 acc1_pos: 39.583 acc5_pos: 100.000 loss: 1.0289 (1.0053) acc1: 23.6979 (28.5897) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1297 max mem: 3531 +Test: [ 60/2387] eta: 0:11:41 acc1_pos: 50.260 acc5_pos: 100.000 loss: 1.0186 (1.0096) acc1: 26.5625 (28.2403) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1296 max mem: 3531 +Test: [ 70/2387] eta: 0:11:21 acc1_pos: 30.990 acc5_pos: 100.000 loss: 1.0624 (1.0236) acc1: 22.6562 (27.2741) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1446 max mem: 3531 +Test: [ 80/2387] eta: 0:10:54 acc1_pos: 35.938 acc5_pos: 100.000 loss: 1.1069 (1.0264) acc1: 23.6979 (26.8422) acc5: 100.0000 (100.0000) time: 0.2300 data: 0.1341 max mem: 3531 +Test: [ 90/2387] eta: 0:10:35 acc1_pos: 36.979 acc5_pos: 100.000 loss: 0.9909 (1.0221) acc1: 25.0000 (26.9546) acc5: 100.0000 (100.0000) time: 0.2155 data: 0.1186 max mem: 3531 +Test: [ 100/2387] eta: 0:10:22 acc1_pos: 20.573 acc5_pos: 100.000 loss: 0.9985 (1.0307) acc1: 22.3958 (26.0468) acc5: 100.0000 (100.0000) time: 0.2248 data: 0.1303 max mem: 3531 +Test: [ 110/2387] eta: 0:10:14 acc1_pos: 25.521 acc5_pos: 100.000 loss: 0.9985 (1.0268) acc1: 20.5729 (26.0206) acc5: 100.0000 (100.0000) time: 0.2393 data: 0.1448 max mem: 3531 +Test: [ 120/2387] eta: 0:10:04 acc1_pos: 29.427 acc5_pos: 100.000 loss: 1.0038 (1.0283) acc1: 25.7812 (25.9556) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1437 max mem: 3531 +Test: [ 130/2387] eta: 0:09:59 acc1_pos: 19.531 acc5_pos: 100.000 loss: 1.0258 (1.0292) acc1: 21.0938 (25.6182) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1441 max mem: 3531 +Test: [ 140/2387] eta: 0:09:47 acc1_pos: 20.052 acc5_pos: 100.000 loss: 1.0367 (1.0288) acc1: 21.0938 (25.5559) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1364 max mem: 3531 +Test: [ 150/2387] eta: 0:09:44 acc1_pos: 9.115 acc5_pos: 100.000 loss: 0.9958 (1.0277) acc1: 21.0938 (25.4018) acc5: 100.0000 (100.0000) time: 0.2349 data: 0.1414 max mem: 3531 +Test: [ 160/2387] eta: 0:09:28 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.9733 (1.0096) acc1: 23.1771 (26.9264) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2104 data: 0.1165 max mem: 3531 +Test: [ 170/2387] eta: 0:09:18 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.2247 (0.9633) acc1: 97.1354 (31.0764) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1799 data: 0.0884 max mem: 3531 +Test: [ 180/2387] eta: 0:09:07 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.2197 (0.9222) acc1: 97.6562 (34.7606) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1901 data: 0.0977 max mem: 3531 +Test: [ 190/2387] eta: 0:09:01 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.2168 (0.8851) acc1: 97.6562 (38.0672) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1999 data: 0.1061 max mem: 3531 +Test: [ 200/2387] eta: 0:08:50 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.2116 (0.8516) acc1: 97.9167 (41.0538) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1942 data: 0.1000 max mem: 3531 +Test: [ 210/2387] eta: 0:08:37 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.2144 (0.8217) acc1: 97.9167 (43.7451) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1552 data: 0.0592 max mem: 3532 +Test: [ 220/2387] eta: 0:08:30 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.2174 (0.7944) acc1: 97.9167 (46.1868) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1649 data: 0.0688 max mem: 3532 +Test: [ 230/2387] eta: 0:08:30 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.2174 (0.7693) acc1: 97.6562 (48.4161) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2234 data: 0.1286 max mem: 3532 +Test: [ 240/2387] eta: 0:08:22 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.2194 (0.7466) acc1: 97.9167 (50.4906) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2158 data: 0.1227 max mem: 3532 +Test: [ 250/2387] eta: 0:08:10 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.2099 (0.7215) acc1: 98.9583 (52.4568) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1473 data: 0.0550 max mem: 3532 +Test: [ 260/2387] eta: 0:08:04 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1002 (0.6977) acc1: 100.0000 (54.2784) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1562 data: 0.0626 max mem: 3532 +Test: [ 270/2387] eta: 0:07:59 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.0989 (0.6756) acc1: 100.0000 (55.9656) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1893 data: 0.0950 max mem: 3532 +Test: [ 280/2387] eta: 0:07:52 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.0989 (0.6551) acc1: 100.0000 (57.5326) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1747 data: 0.0783 max mem: 3532 +Test: [ 290/2387] eta: 0:07:46 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.0980 (0.6359) acc1: 100.0000 (58.9920) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1714 data: 0.0756 max mem: 3532 +Test: [ 300/2387] eta: 0:07:43 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.0972 (0.6181) acc1: 100.0000 (60.3544) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1931 data: 0.0989 max mem: 3532 +Test: [ 310/2387] eta: 0:07:38 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1071 (0.6045) acc1: 100.0000 (61.5948) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1929 data: 0.0974 max mem: 3532 +Test: [ 320/2387] eta: 0:07:33 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.2924 (0.5950) acc1: 98.6979 (62.7474) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1755 data: 0.0804 max mem: 3532 +Test: [ 330/2387] eta: 0:07:30 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.2979 (0.5859) acc1: 98.9583 (63.8469) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1919 data: 0.0971 max mem: 3532 +Test: [ 340/2387] eta: 0:07:25 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.2929 (0.5773) acc1: 98.9583 (64.8789) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1957 data: 0.1025 max mem: 3532 +Test: [ 350/2387] eta: 0:07:21 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.2872 (0.5654) acc1: 99.7396 (65.8758) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1781 data: 0.0861 max mem: 3532 +Test: [ 360/2387] eta: 0:07:19 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1725 (0.5546) acc1: 99.7396 (66.7994) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1963 data: 0.1023 max mem: 3532 +Test: [ 370/2387] eta: 0:07:17 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1764 (0.5444) acc1: 99.2188 (67.6711) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2197 data: 0.1254 max mem: 3532 +Test: [ 380/2387] eta: 0:07:13 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1766 (0.5348) acc1: 98.9583 (68.4916) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2020 data: 0.1053 max mem: 3532 +Test: [ 390/2387] eta: 0:07:12 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1785 (0.5257) acc1: 99.2188 (69.2822) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2149 data: 0.1187 max mem: 3532 +Test: [ 400/2387] eta: 0:07:07 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1785 (0.5170) acc1: 99.2188 (70.0255) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2003 data: 0.1079 max mem: 3532 +Test: [ 410/2387] eta: 0:07:03 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1773 (0.5088) acc1: 98.9583 (70.7351) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1629 data: 0.0713 max mem: 3532 +Test: [ 420/2387] eta: 0:07:00 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1775 (0.5009) acc1: 99.2188 (71.4117) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1894 data: 0.0965 max mem: 3532 +Test: [ 430/2387] eta: 0:06:58 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1802 (0.4936) acc1: 99.4792 (72.0599) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2089 data: 0.1141 max mem: 3532 +Test: [ 440/2387] eta: 0:06:57 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1831 (0.4854) acc1: 99.4792 (72.6834) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2211 data: 0.1269 max mem: 3533 +Test: [ 450/2387] eta: 0:06:52 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.0863 (0.4765) acc1: 100.0000 (73.2891) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0965 max mem: 3533 +Test: [ 460/2387] eta: 0:06:50 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.0855 (0.4680) acc1: 100.0000 (73.8685) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1863 data: 0.0921 max mem: 3533 +Test: [ 470/2387] eta: 0:06:48 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.0855 (0.4599) acc1: 100.0000 (74.4233) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2178 data: 0.1229 max mem: 3533 +Test: [ 480/2387] eta: 0:06:45 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.0860 (0.4521) acc1: 100.0000 (74.9551) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2073 data: 0.1139 max mem: 3533 +Test: [ 490/2387] eta: 0:06:44 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.0862 (0.4447) acc1: 100.0000 (75.4651) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2183 data: 0.1270 max mem: 3533 +Test: [ 500/2387] eta: 0:06:39 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.0863 (0.4375) acc1: 100.0000 (75.9549) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1869 data: 0.0964 max mem: 3533 +Test: [ 510/2387] eta: 0:06:37 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.0862 (0.4310) acc1: 100.0000 (76.4254) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1726 data: 0.0819 max mem: 3533 +Test: [ 520/2387] eta: 0:06:34 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1337 (0.4254) acc1: 100.0000 (76.8779) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1990 data: 0.1070 max mem: 3533 +Test: [ 530/2387] eta: 0:06:31 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1336 (0.4199) acc1: 100.0000 (77.3133) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1961 data: 0.1046 max mem: 3533 +Test: [ 540/2387] eta: 0:06:29 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1324 (0.4146) acc1: 100.0000 (77.7327) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2061 data: 0.1151 max mem: 3533 +Test: [ 550/2387] eta: 0:06:27 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1324 (0.4095) acc1: 100.0000 (78.1368) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2031 data: 0.1118 max mem: 3533 +Test: [ 560/2387] eta: 0:06:25 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1319 (0.4045) acc1: 100.0000 (78.5265) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2033 data: 0.1117 max mem: 3533 +Test: [ 570/2387] eta: 0:06:22 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1319 (0.3998) acc1: 100.0000 (78.9026) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2044 data: 0.1092 max mem: 3533 +Test: [ 580/2387] eta: 0:06:19 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1315 (0.3952) acc1: 100.0000 (79.2657) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1864 data: 0.0898 max mem: 3533 +Test: [ 590/2387] eta: 0:06:16 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1317 (0.3907) acc1: 100.0000 (79.6166) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0801 max mem: 3533 +Test: [ 600/2387] eta: 0:06:13 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1346 (0.3867) acc1: 100.0000 (79.9553) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1747 data: 0.0816 max mem: 3533 +Test: [ 610/2387] eta: 0:06:09 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1639 (0.3836) acc1: 99.7396 (80.2676) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1699 data: 0.0781 max mem: 3533 +Test: [ 620/2387] eta: 0:06:07 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1926 (0.3805) acc1: 98.9583 (80.5702) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1777 data: 0.0854 max mem: 3533 +Test: [ 630/2387] eta: 0:06:03 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1931 (0.3776) acc1: 99.2188 (80.8641) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1778 data: 0.0855 max mem: 3533 +Test: [ 640/2387] eta: 0:06:01 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1952 (0.3748) acc1: 98.9583 (81.1423) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1775 data: 0.0858 max mem: 3533 +Test: [ 650/2387] eta: 0:05:58 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1964 (0.3720) acc1: 98.6979 (81.4152) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1769 data: 0.0849 max mem: 3533 +Test: [ 660/2387] eta: 0:05:55 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1966 (0.3694) acc1: 98.9583 (81.6834) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1693 data: 0.0769 max mem: 3533 +Test: [ 670/2387] eta: 0:05:53 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1956 (0.3668) acc1: 99.2188 (81.9439) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1898 data: 0.0971 max mem: 3534 +Test: [ 680/2387] eta: 0:05:51 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1880 (0.3641) acc1: 99.2188 (82.2049) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2038 data: 0.1122 max mem: 3534 +Test: [ 690/2387] eta: 0:05:48 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1938 (0.3681) acc1: 98.9583 (81.9495) acc5: 100.0000 (100.0000) acc1_neg: 49.219 acc5_neg: 100.000 time: 0.2037 data: 0.1131 max mem: 3534 +Test: [ 700/2387] eta: 0:05:46 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.8087 (0.3744) acc1: 51.5625 (81.5130) acc5: 100.0000 (100.0000) acc1_neg: 55.729 acc5_neg: 100.000 time: 0.2052 data: 0.1110 max mem: 3534 +Test: [ 710/2387] eta: 0:05:44 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.8087 (0.3806) acc1: 51.0417 (81.0804) acc5: 100.0000 (100.0000) acc1_neg: 50.521 acc5_neg: 100.000 time: 0.1961 data: 0.1014 max mem: 3534 +Test: [ 720/2387] eta: 0:05:42 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.8081 (0.3865) acc1: 50.5208 (80.6692) acc5: 100.0000 (100.0000) acc1_neg: 51.302 acc5_neg: 100.000 time: 0.1895 data: 0.0984 max mem: 3534 +Test: [ 730/2387] eta: 0:05:38 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.8097 (0.3924) acc1: 50.5208 (80.2539) acc5: 100.0000 (100.0000) acc1_neg: 51.302 acc5_neg: 100.000 time: 0.1704 data: 0.0795 max mem: 3534 +Test: [ 740/2387] eta: 0:05:35 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.8107 (0.3981) acc1: 50.7812 (79.8604) acc5: 100.0000 (100.0000) acc1_neg: 45.312 acc5_neg: 100.000 time: 0.1613 data: 0.0695 max mem: 3534 +Test: [ 750/2387] eta: 0:05:32 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.8108 (0.4035) acc1: 51.5625 (79.4892) acc5: 100.0000 (100.0000) acc1_neg: 54.688 acc5_neg: 100.000 time: 0.1657 data: 0.0733 max mem: 3534 +Test: [ 760/2387] eta: 0:05:30 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.8037 (0.4071) acc1: 52.3438 (79.2526) acc5: 100.0000 (100.0000) acc1_neg: 80.729 acc5_neg: 100.000 time: 0.1601 data: 0.0640 max mem: 3534 +Test: [ 770/2387] eta: 0:05:27 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.4538 (0.4063) acc1: 77.3438 (79.3457) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1618 data: 0.0663 max mem: 3534 +Test: [ 780/2387] eta: 0:05:24 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.2141 (0.4035) acc1: 97.6562 (79.5985) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1630 data: 0.0680 max mem: 3534 +Test: [ 790/2387] eta: 0:05:21 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1710 (0.4007) acc1: 99.7396 (79.8143) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.1631 data: 0.0686 max mem: 3534 +Test: [ 800/2387] eta: 0:05:18 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1659 (0.3977) acc1: 99.7396 (80.0630) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1647 data: 0.0731 max mem: 3534 +Test: [ 810/2387] eta: 0:05:16 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1683 (0.3950) acc1: 99.7396 (80.3002) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1699 data: 0.0786 max mem: 3534 +Test: [ 820/2387] eta: 0:05:14 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1773 (0.3926) acc1: 99.7396 (80.5147) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1845 data: 0.0888 max mem: 3534 +Test: [ 830/2387] eta: 0:05:12 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.2093 (0.3910) acc1: 97.6562 (80.6931) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2035 data: 0.1061 max mem: 3534 +Test: [ 840/2387] eta: 0:05:09 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.2408 (0.3892) acc1: 97.6562 (80.8896) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1849 data: 0.0889 max mem: 3534 +Test: [ 850/2387] eta: 0:05:05 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.2408 (0.3876) acc1: 96.6146 (81.0548) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1299 data: 0.0310 max mem: 3534 +Test: [ 860/2387] eta: 0:05:02 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.2175 (0.3856) acc1: 96.8750 (81.2551) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1099 data: 0.0117 max mem: 3534 +Test: [ 870/2387] eta: 0:04:59 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.2148 (0.3836) acc1: 98.9583 (81.4443) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1349 data: 0.0395 max mem: 3534 +Test: [ 880/2387] eta: 0:04:57 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.2168 (0.3819) acc1: 98.1771 (81.6210) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1611 data: 0.0677 max mem: 3534 +Test: [ 890/2387] eta: 0:04:54 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.2081 (0.3798) acc1: 99.4792 (81.8199) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1766 data: 0.0846 max mem: 3534 +Test: [ 900/2387] eta: 0:04:52 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.2175 (0.3822) acc1: 97.1354 (81.6807) acc5: 100.0000 (100.0000) acc1_neg: 63.542 acc5_neg: 100.000 time: 0.1786 data: 0.0852 max mem: 3535 +Test: [ 910/2387] eta: 0:04:49 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.6845 (0.3855) acc1: 62.5000 (81.4753) acc5: 100.0000 (100.0000) acc1_neg: 61.979 acc5_neg: 100.000 time: 0.1632 data: 0.0674 max mem: 3535 +Test: [ 920/2387] eta: 0:04:48 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.6881 (0.3888) acc1: 61.9792 (81.2641) acc5: 100.0000 (100.0000) acc1_neg: 61.458 acc5_neg: 100.000 time: 0.1849 data: 0.0895 max mem: 3535 +Test: [ 930/2387] eta: 0:04:46 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.6940 (0.3919) acc1: 61.4583 (81.0536) acc5: 100.0000 (100.0000) acc1_neg: 57.031 acc5_neg: 100.000 time: 0.1999 data: 0.1041 max mem: 3535 +Test: [ 940/2387] eta: 0:04:43 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.6940 (0.3951) acc1: 61.4583 (80.8410) acc5: 100.0000 (100.0000) acc1_neg: 59.635 acc5_neg: 100.000 time: 0.1751 data: 0.0793 max mem: 3535 +Test: [ 950/2387] eta: 0:04:41 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.6968 (0.3981) acc1: 61.4583 (80.6407) acc5: 100.0000 (100.0000) acc1_neg: 60.938 acc5_neg: 100.000 time: 0.1903 data: 0.0955 max mem: 3535 +Test: [ 960/2387] eta: 0:04:40 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.6681 (0.4010) acc1: 62.7604 (80.4612) acc5: 100.0000 (100.0000) acc1_neg: 64.323 acc5_neg: 100.000 time: 0.2047 data: 0.1091 max mem: 3535 +Test: [ 970/2387] eta: 0:04:36 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.6692 (0.4040) acc1: 62.2396 (80.2633) acc5: 100.0000 (100.0000) acc1_neg: 62.240 acc5_neg: 100.000 time: 0.1495 data: 0.0515 max mem: 3535 +Test: [ 980/2387] eta: 0:04:36 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.6540 (0.4041) acc1: 62.5000 (80.2851) acc5: 100.0000 (100.0000) acc1_neg: 70.052 acc5_neg: 100.000 time: 0.2022 data: 0.1070 max mem: 3535 +Test: [ 990/2387] eta: 0:04:33 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.2357 (0.4022) acc1: 98.1771 (80.4280) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2265 data: 0.1359 max mem: 3535 +Test: [1000/2387] eta: 0:04:31 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1559 (0.3999) acc1: 99.2188 (80.6012) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1776 data: 0.0863 max mem: 3535 +Test: [1010/2387] eta: 0:04:29 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1594 (0.3978) acc1: 98.6979 (80.7709) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2047 data: 0.1086 max mem: 3535 +Test: [1020/2387] eta: 0:04:28 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1896 (0.3961) acc1: 98.4375 (80.9240) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2066 data: 0.1104 max mem: 3535 +Test: [1030/2387] eta: 0:04:26 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1919 (0.3941) acc1: 98.9583 (81.0984) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2004 data: 0.1086 max mem: 3535 +Test: [1040/2387] eta: 0:04:23 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1668 (0.3919) acc1: 99.7396 (81.2785) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1805 data: 0.0894 max mem: 3535 +Test: [1050/2387] eta: 0:04:21 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1598 (0.3897) acc1: 100.0000 (81.4510) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1744 data: 0.0811 max mem: 3535 +Test: [1060/2387] eta: 0:04:19 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1610 (0.3876) acc1: 99.7396 (81.6231) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1847 data: 0.0871 max mem: 3535 +Test: [1070/2387] eta: 0:04:17 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1746 (0.3859) acc1: 99.7396 (81.7730) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1706 data: 0.0738 max mem: 3535 +Test: [1080/2387] eta: 0:04:14 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.2167 (0.3846) acc1: 98.6979 (81.9094) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1642 data: 0.0718 max mem: 3535 +Test: [1090/2387] eta: 0:04:12 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.2259 (0.3833) acc1: 97.9167 (82.0463) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1785 data: 0.0880 max mem: 3535 +Test: [1100/2387] eta: 0:04:10 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.2254 (0.3817) acc1: 97.9167 (82.1904) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1690 data: 0.0742 max mem: 3535 +Test: [1110/2387] eta: 0:04:08 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.2140 (0.3803) acc1: 97.1354 (82.3123) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1660 data: 0.0695 max mem: 3535 +Test: [1120/2387] eta: 0:04:05 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.2038 (0.3788) acc1: 97.3958 (82.4513) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1611 data: 0.0691 max mem: 3536 +Test: [1130/2387] eta: 0:04:03 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1954 (0.3772) acc1: 98.4375 (82.5961) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1537 data: 0.0628 max mem: 3536 +Test: [1140/2387] eta: 0:04:01 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1988 (0.3758) acc1: 98.4375 (82.7356) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1559 data: 0.0628 max mem: 3536 +Test: [1150/2387] eta: 0:03:58 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1979 (0.3742) acc1: 99.4792 (82.8675) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1542 data: 0.0602 max mem: 3536 +Test: [1160/2387] eta: 0:03:56 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1445 (0.3722) acc1: 99.2188 (83.0045) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1522 data: 0.0585 max mem: 3536 +Test: [1170/2387] eta: 0:03:54 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1377 (0.3702) acc1: 99.4792 (83.1447) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1517 data: 0.0579 max mem: 3536 +Test: [1180/2387] eta: 0:03:51 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1377 (0.3683) acc1: 99.4792 (83.2811) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1581 data: 0.0659 max mem: 3536 +Test: [1190/2387] eta: 0:03:49 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1374 (0.3663) acc1: 100.0000 (83.4206) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1548 data: 0.0641 max mem: 3536 +Test: [1200/2387] eta: 0:03:46 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1314 (0.3644) acc1: 100.0000 (83.5573) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1384 data: 0.0480 max mem: 3536 +Test: [1210/2387] eta: 0:03:44 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1314 (0.3625) acc1: 100.0000 (83.6929) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1410 data: 0.0502 max mem: 3536 +Test: [1220/2387] eta: 0:03:42 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1219 (0.3605) acc1: 100.0000 (83.8249) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1683 data: 0.0770 max mem: 3536 +Test: [1230/2387] eta: 0:03:40 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1169 (0.3585) acc1: 100.0000 (83.9563) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1708 data: 0.0766 max mem: 3536 +Test: [1240/2387] eta: 0:03:38 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1093 (0.3565) acc1: 100.0000 (84.0856) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1467 data: 0.0504 max mem: 3536 +Test: [1250/2387] eta: 0:03:36 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1093 (0.3545) acc1: 100.0000 (84.2126) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1569 data: 0.0631 max mem: 3536 +Test: [1260/2387] eta: 0:03:34 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1085 (0.3525) acc1: 100.0000 (84.3376) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1889 data: 0.0969 max mem: 3536 +Test: [1270/2387] eta: 0:03:32 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1070 (0.3507) acc1: 100.0000 (84.4606) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1962 data: 0.1044 max mem: 3536 +Test: [1280/2387] eta: 0:03:30 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1221 (0.3490) acc1: 100.0000 (84.5765) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1960 data: 0.1038 max mem: 3536 +Test: [1290/2387] eta: 0:03:28 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1483 (0.3475) acc1: 98.9583 (84.6816) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2003 data: 0.1083 max mem: 3536 +Test: [1300/2387] eta: 0:03:27 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1473 (0.3460) acc1: 99.2188 (84.7917) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2235 data: 0.1318 max mem: 3536 +Test: [1310/2387] eta: 0:03:24 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1423 (0.3445) acc1: 99.2188 (84.8978) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1972 data: 0.1041 max mem: 3536 +Test: [1320/2387] eta: 0:03:22 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1482 (0.3431) acc1: 99.4792 (85.0029) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1603 data: 0.0648 max mem: 3536 +Test: [1330/2387] eta: 0:03:21 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1676 (0.3418) acc1: 99.4792 (85.1089) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1848 data: 0.0900 max mem: 3536 +Test: [1340/2387] eta: 0:03:19 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1791 (0.3421) acc1: 98.9583 (85.1058) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1848 data: 0.0915 max mem: 3536 +Test: [1350/2387] eta: 0:03:17 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.0902 (0.3402) acc1: 100.0000 (85.2160) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0891 max mem: 3537 +Test: [1360/2387] eta: 0:03:15 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.0836 (0.3383) acc1: 100.0000 (85.3246) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1763 data: 0.0788 max mem: 3537 +Test: [1370/2387] eta: 0:03:12 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.0792 (0.3364) acc1: 100.0000 (85.4317) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1450 data: 0.0486 max mem: 3537 +Test: [1380/2387] eta: 0:03:10 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.0790 (0.3345) acc1: 100.0000 (85.5372) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1445 data: 0.0515 max mem: 3537 +Test: [1390/2387] eta: 0:03:08 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.0802 (0.3327) acc1: 100.0000 (85.6411) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1599 data: 0.0681 max mem: 3537 +Test: [1400/2387] eta: 0:03:06 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.0810 (0.3309) acc1: 100.0000 (85.7436) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1634 data: 0.0696 max mem: 3537 +Test: [1410/2387] eta: 0:03:04 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.0870 (0.3301) acc1: 100.0000 (85.8222) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1604 data: 0.0677 max mem: 3537 +Test: [1420/2387] eta: 0:03:02 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.2129 (0.3294) acc1: 98.1771 (85.8939) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1591 data: 0.0668 max mem: 3537 +Test: [1430/2387] eta: 0:03:00 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.2191 (0.3287) acc1: 97.3958 (85.9653) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1555 data: 0.0613 max mem: 3537 +Test: [1440/2387] eta: 0:02:58 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.2276 (0.3280) acc1: 97.3958 (86.0503) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1649 data: 0.0717 max mem: 3537 +Test: [1450/2387] eta: 0:02:56 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.2069 (0.3273) acc1: 98.6979 (86.1312) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1649 data: 0.0727 max mem: 3537 +Test: [1460/2387] eta: 0:02:53 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1836 (0.3264) acc1: 98.4375 (86.2116) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1509 data: 0.0586 max mem: 3537 +Test: [1470/2387] eta: 0:02:51 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1663 (0.3252) acc1: 98.6979 (86.3004) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1537 data: 0.0593 max mem: 3537 +Test: [1480/2387] eta: 0:02:49 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1454 (0.3240) acc1: 99.4792 (86.3845) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1696 data: 0.0748 max mem: 3537 +Test: [1490/2387] eta: 0:02:48 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1488 (0.3229) acc1: 99.4792 (86.4716) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1799 data: 0.0845 max mem: 3537 +Test: [1500/2387] eta: 0:02:45 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1494 (0.3218) acc1: 99.2188 (86.5522) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1650 data: 0.0700 max mem: 3537 +Test: [1510/2387] eta: 0:02:43 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1482 (0.3206) acc1: 99.7396 (86.6409) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1549 data: 0.0615 max mem: 3537 +Test: [1520/2387] eta: 0:02:41 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1624 (0.3198) acc1: 99.7396 (86.7158) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1640 data: 0.0682 max mem: 3537 +Test: [1530/2387] eta: 0:02:39 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.2053 (0.3192) acc1: 98.4375 (86.7781) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1672 data: 0.0713 max mem: 3537 +Test: [1540/2387] eta: 0:02:37 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.2015 (0.3184) acc1: 99.2188 (86.8548) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1549 data: 0.0614 max mem: 3537 +Test: [1550/2387] eta: 0:02:35 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1881 (0.3176) acc1: 99.4792 (86.9318) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1476 data: 0.0552 max mem: 3537 +Test: [1560/2387] eta: 0:02:33 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1881 (0.3169) acc1: 99.2188 (86.9994) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1549 data: 0.0628 max mem: 3537 +Test: [1570/2387] eta: 0:02:31 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1764 (0.3160) acc1: 99.7396 (87.0800) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1599 data: 0.0646 max mem: 3537 +Test: [1580/2387] eta: 0:02:29 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1887 (0.3154) acc1: 99.4792 (87.1417) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1511 data: 0.0564 max mem: 3538 +Test: [1590/2387] eta: 0:02:27 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.2086 (0.3147) acc1: 98.6979 (87.2150) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1619 data: 0.0711 max mem: 3538 +Test: [1600/2387] eta: 0:02:26 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1848 (0.3140) acc1: 99.2188 (87.2739) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1942 data: 0.1025 max mem: 3538 +Test: [1610/2387] eta: 0:02:24 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1759 (0.3132) acc1: 99.2188 (87.3458) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1884 data: 0.0916 max mem: 3538 +Test: [1620/2387] eta: 0:02:22 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1918 (0.3125) acc1: 98.4375 (87.4047) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1862 data: 0.0904 max mem: 3538 +Test: [1630/2387] eta: 0:02:20 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.2168 (0.3121) acc1: 97.6562 (87.4618) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1944 data: 0.1024 max mem: 3538 +Test: [1640/2387] eta: 0:02:18 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.2525 (0.3119) acc1: 96.3542 (87.5037) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1981 data: 0.1045 max mem: 3538 +Test: [1650/2387] eta: 0:02:17 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.2538 (0.3115) acc1: 95.8333 (87.5565) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2065 data: 0.1141 max mem: 3538 +Test: [1660/2387] eta: 0:02:15 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.2500 (0.3113) acc1: 95.8333 (87.6002) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2055 data: 0.1146 max mem: 3538 +Test: [1670/2387] eta: 0:02:13 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.2661 (0.3110) acc1: 95.5729 (87.6487) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1833 data: 0.0904 max mem: 3538 +Test: [1680/2387] eta: 0:02:11 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.2669 (0.3108) acc1: 95.8333 (87.6888) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1718 data: 0.0782 max mem: 3538 +Test: [1690/2387] eta: 0:02:09 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.2675 (0.3107) acc1: 96.0938 (87.7330) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1981 data: 0.1040 max mem: 3538 +Test: [1700/2387] eta: 0:02:07 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.2586 (0.3101) acc1: 96.3542 (87.7875) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1942 data: 0.0990 max mem: 3538 +Test: [1710/2387] eta: 0:02:05 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1409 (0.3091) acc1: 99.4792 (87.8563) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1824 data: 0.0895 max mem: 3538 +Test: [1720/2387] eta: 0:02:04 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1354 (0.3080) acc1: 99.7396 (87.9247) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1996 data: 0.1077 max mem: 3538 +Test: [1730/2387] eta: 0:02:02 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1292 (0.3070) acc1: 99.7396 (87.9907) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1892 data: 0.0967 max mem: 3538 +Test: [1740/2387] eta: 0:02:00 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1269 (0.3060) acc1: 99.7396 (88.0582) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1798 data: 0.0876 max mem: 3538 +Test: [1750/2387] eta: 0:01:58 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1274 (0.3050) acc1: 99.7396 (88.1248) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1965 data: 0.1039 max mem: 3538 +Test: [1760/2387] eta: 0:01:56 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1337 (0.3040) acc1: 100.0000 (88.1913) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1986 data: 0.1048 max mem: 3538 +Test: [1770/2387] eta: 0:01:54 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1523 (0.3034) acc1: 99.4792 (88.2465) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1985 data: 0.1031 max mem: 3538 +Test: [1780/2387] eta: 0:01:53 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1926 (0.3027) acc1: 98.4375 (88.3013) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1927 data: 0.0988 max mem: 3538 +Test: [1790/2387] eta: 0:01:51 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1844 (0.3020) acc1: 98.4375 (88.3599) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1881 data: 0.0966 max mem: 3538 +Test: [1800/2387] eta: 0:01:49 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1784 (0.3013) acc1: 98.6979 (88.4173) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1993 data: 0.1077 max mem: 3538 +Test: [1810/2387] eta: 0:01:47 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1574 (0.3004) acc1: 99.2188 (88.4793) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1718 data: 0.0809 max mem: 3539 +Test: [1820/2387] eta: 0:01:45 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1432 (0.2996) acc1: 100.0000 (88.5412) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1655 data: 0.0746 max mem: 3539 +Test: [1830/2387] eta: 0:01:43 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1432 (0.2987) acc1: 100.0000 (88.6037) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1774 data: 0.0859 max mem: 3539 +Test: [1840/2387] eta: 0:01:41 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1393 (0.2979) acc1: 100.0000 (88.6656) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1572 data: 0.0635 max mem: 3539 +Test: [1850/2387] eta: 0:01:39 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1382 (0.2970) acc1: 100.0000 (88.7264) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1745 data: 0.0804 max mem: 3539 +Test: [1860/2387] eta: 0:01:37 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1381 (0.2962) acc1: 100.0000 (88.7868) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1811 data: 0.0890 max mem: 3539 +Test: [1870/2387] eta: 0:01:35 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1381 (0.2953) acc1: 100.0000 (88.8463) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1395 data: 0.0475 max mem: 3539 +Test: [1880/2387] eta: 0:01:34 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1376 (0.2945) acc1: 100.0000 (88.9054) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1514 data: 0.0596 max mem: 3539 +Test: [1890/2387] eta: 0:01:32 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1375 (0.2936) acc1: 100.0000 (88.9640) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1801 data: 0.0876 max mem: 3539 +Test: [1900/2387] eta: 0:01:30 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1399 (0.2928) acc1: 100.0000 (89.0221) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1770 data: 0.0854 max mem: 3539 +Test: [1910/2387] eta: 0:01:28 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1546 (0.2922) acc1: 100.0000 (89.0774) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1778 data: 0.0860 max mem: 3539 +Test: [1920/2387] eta: 0:01:26 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1854 (0.2922) acc1: 99.7396 (89.0877) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1711 data: 0.0786 max mem: 3539 +Test: [1930/2387] eta: 0:01:24 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1936 (0.2916) acc1: 99.4792 (89.1429) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1599 data: 0.0687 max mem: 3539 +Test: [1940/2387] eta: 0:01:22 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1863 (0.2914) acc1: 99.7396 (89.1611) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1609 data: 0.0680 max mem: 3539 +Test: [1950/2387] eta: 0:01:20 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.2010 (0.2910) acc1: 99.2188 (89.2077) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1699 data: 0.0767 max mem: 3539 +Test: [1960/2387] eta: 0:01:18 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.2023 (0.2907) acc1: 99.2188 (89.2466) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1725 data: 0.0796 max mem: 3539 +Test: [1970/2387] eta: 0:01:17 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.2439 (0.2908) acc1: 96.0938 (89.2553) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.1701 data: 0.0769 max mem: 3539 +Test: [1980/2387] eta: 0:01:15 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.2635 (0.2905) acc1: 95.0521 (89.2886) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1698 data: 0.0757 max mem: 3539 +Test: [1990/2387] eta: 0:01:13 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1677 (0.2899) acc1: 97.6562 (89.3304) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1699 data: 0.0730 max mem: 3539 +Test: [2000/2387] eta: 0:01:11 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1677 (0.2893) acc1: 97.9167 (89.3741) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1599 data: 0.0642 max mem: 3539 +Test: [2010/2387] eta: 0:01:09 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1690 (0.2887) acc1: 97.9167 (89.4169) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1748 data: 0.0827 max mem: 3539 +Test: [2020/2387] eta: 0:01:07 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1686 (0.2881) acc1: 97.6562 (89.4595) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1604 data: 0.0664 max mem: 3539 +Test: [2030/2387] eta: 0:01:05 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1678 (0.2875) acc1: 97.9167 (89.5004) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1261 data: 0.0292 max mem: 3540 +Test: [2040/2387] eta: 0:01:03 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1673 (0.2869) acc1: 97.9167 (89.5428) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1396 data: 0.0466 max mem: 3540 +Test: [2050/2387] eta: 0:01:01 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1649 (0.2864) acc1: 98.1771 (89.5863) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1686 data: 0.0751 max mem: 3540 +Test: [2060/2387] eta: 0:01:00 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1710 (0.2858) acc1: 98.1771 (89.6255) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1646 data: 0.0684 max mem: 3540 +Test: [2070/2387] eta: 0:00:58 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1738 (0.2853) acc1: 97.9167 (89.6675) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1549 data: 0.0592 max mem: 3540 +Test: [2080/2387] eta: 0:00:56 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1691 (0.2847) acc1: 98.1771 (89.7086) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1649 data: 0.0693 max mem: 3540 +Test: [2090/2387] eta: 0:00:54 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1691 (0.2842) acc1: 98.1771 (89.7495) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1549 data: 0.0583 max mem: 3540 +Test: [2100/2387] eta: 0:00:52 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1694 (0.2836) acc1: 97.9167 (89.7895) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1399 data: 0.0442 max mem: 3540 +Test: [2110/2387] eta: 0:00:50 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1718 (0.2831) acc1: 98.1771 (89.8289) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1705 data: 0.0781 max mem: 3540 +Test: [2120/2387] eta: 0:00:48 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1785 (0.2829) acc1: 98.1771 (89.8602) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.1899 data: 0.0985 max mem: 3540 +Test: [2130/2387] eta: 0:00:47 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1858 (0.2824) acc1: 98.9583 (89.9031) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1827 data: 0.0900 max mem: 3540 +Test: [2140/2387] eta: 0:00:45 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1840 (0.2820) acc1: 99.4792 (89.9311) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1773 data: 0.0851 max mem: 3540 +Test: [2150/2387] eta: 0:00:43 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1972 (0.2816) acc1: 98.6979 (89.9723) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1818 data: 0.0910 max mem: 3540 +Test: [2160/2387] eta: 0:00:41 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1936 (0.2812) acc1: 99.2188 (90.0134) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1996 data: 0.1073 max mem: 3540 +Test: [2170/2387] eta: 0:00:39 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.2018 (0.2810) acc1: 98.6979 (90.0489) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1762 data: 0.0840 max mem: 3540 +Test: [2180/2387] eta: 0:00:37 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.2641 (0.2823) acc1: 95.0521 (89.9702) acc5: 100.0000 (100.0000) acc1_neg: 39.844 acc5_neg: 100.000 time: 0.1631 data: 0.0708 max mem: 3540 +Test: [2190/2387] eta: 0:00:36 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.2347 (0.2823) acc1: 94.7917 (89.9667) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1812 data: 0.0870 max mem: 3540 +Test: [2200/2387] eta: 0:00:34 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1770 (0.2819) acc1: 97.9167 (90.0058) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1828 data: 0.0904 max mem: 3540 +Test: [2210/2387] eta: 0:00:32 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1770 (0.2814) acc1: 99.2188 (90.0437) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1821 data: 0.0896 max mem: 3540 +Test: [2220/2387] eta: 0:00:30 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1742 (0.2809) acc1: 99.7396 (90.0819) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1840 data: 0.0892 max mem: 3540 +Test: [2230/2387] eta: 0:00:28 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1684 (0.2804) acc1: 99.4792 (90.1220) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0905 max mem: 3540 +Test: [2240/2387] eta: 0:00:26 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1647 (0.2800) acc1: 98.9583 (90.1588) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1898 data: 0.0951 max mem: 3540 +Test: [2250/2387] eta: 0:00:25 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1673 (0.2795) acc1: 98.9583 (90.1985) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1960 data: 0.1028 max mem: 3540 +Test: [2260/2387] eta: 0:00:23 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1679 (0.2790) acc1: 99.7396 (90.2410) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1851 data: 0.0935 max mem: 3541 +Test: [2270/2387] eta: 0:00:21 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1572 (0.2785) acc1: 100.0000 (90.2839) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1796 data: 0.0875 max mem: 3541 +Test: [2280/2387] eta: 0:00:19 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1487 (0.2780) acc1: 100.0000 (90.3099) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1874 data: 0.0929 max mem: 3541 +Test: [2290/2387] eta: 0:00:17 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1364 (0.2774) acc1: 100.0000 (90.3511) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1897 data: 0.0929 max mem: 3541 +Test: [2300/2387] eta: 0:00:15 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1455 (0.2769) acc1: 100.0000 (90.3902) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1849 data: 0.0902 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1584 (0.2765) acc1: 100.0000 (90.4255) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1796 data: 0.0870 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.1910 (0.2764) acc1: 98.6979 (90.4508) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1799 data: 0.0878 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.3001 (0.2767) acc1: 92.1875 (90.4530) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1745 data: 0.0830 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.2826 (0.2765) acc1: 92.4479 (90.4774) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1723 data: 0.0802 max mem: 3541 +Test: [2350/2387] eta: 0:00:06 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.2380 (0.2767) acc1: 97.6562 (90.4804) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1849 data: 0.0926 max mem: 3541 +Test: [2360/2387] eta: 0:00:04 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.2752 (0.2768) acc1: 94.5312 (90.4907) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1926 data: 0.0991 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.2848 (0.2768) acc1: 94.5312 (90.5035) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.1899 data: 0.0960 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.2892 (0.2770) acc1: 93.2292 (90.5085) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.1821 data: 0.0906 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 21.005 acc5_pos: 100.000 loss: 0.2939 (0.2768) acc1: 93.2292 (90.5209) acc5: 100.0000 (100.0000) acc1_neg: 95.349 acc5_neg: 100.000 time: 0.1668 data: 0.0734 max mem: 3541 +Test: Total time: 0:07:17 (0.1832 s / it) +* Overall Acc@1 90.521 Acc@5 100.000 + target == real Acc@1 25.145 Acc@5 100.000 + target == fake Acc@1 95.112 Acc@5 100.000 + loss 0.277 + PR-AUC 0.9906 ROC-AUC 0.8776 EER 19.32% (thr=0.7646) +Accuracy of the network on the 916267 test images: 90.5% +[rank0]:[W725 17:31:35.173219700 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 17:31:57.468000 93238 torch/distributed/run.py:766] +W0725 17:31:57.468000 93238 torch/distributed/run.py:766] ***************************************** +W0725 17:31:57.468000 93238 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:31:57.468000 93238 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:32:31.226494873 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W725 17:32:31.232056664 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:32:31.439681421 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:32:31.511738140 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:32:31.743650113 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 17:32:31.755711531 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:32:31.789079425 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:32:31.874552268 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed2_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed2_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:05:17 acc1_pos: 39.844 acc5_pos: 100.000 loss: 0.8949 (0.8949) acc1: 39.8438 (39.8438) acc5: 100.0000 (100.0000) time: 7.5770 data: 5.8426 max mem: 3531 +Test: [ 10/517] eta: 0:06:31 acc1_pos: 19.010 acc5_pos: 100.000 loss: 0.8949 (0.9268) acc1: 37.5000 (35.4640) acc5: 100.0000 (100.0000) time: 0.7713 data: 0.5315 max mem: 3531 +Test: [ 20/517] eta: 0:03:44 acc1_pos: 23.698 acc5_pos: 100.000 loss: 0.9423 (0.9533) acc1: 27.0833 (33.5193) acc5: 100.0000 (100.0000) time: 0.0953 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:03:06 acc1_pos: 16.667 acc5_pos: 100.000 loss: 1.0040 (0.9779) acc1: 25.7812 (31.5608) acc5: 100.0000 (100.0000) time: 0.1700 data: 0.0749 max mem: 3531 +Test: [ 40/517] eta: 0:02:47 acc1_pos: 29.948 acc5_pos: 100.000 loss: 1.0518 (1.0016) acc1: 21.8750 (28.8745) acc5: 100.0000 (100.0000) time: 0.2468 data: 0.1546 max mem: 3531 +Test: [ 50/517] eta: 0:02:36 acc1_pos: 40.826 acc5_pos: 100.000 loss: 1.0289 (0.9789) acc1: 23.6979 (31.1428) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.2616 data: 0.1665 max mem: 3531 +Test: [ 60/517] eta: 0:02:29 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.4516 (0.8844) acc1: 85.1562 (40.7702) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.2783 data: 0.1822 max mem: 3531 +Test: [ 70/517] eta: 0:02:22 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.4106 (0.8213) acc1: 90.1042 (47.2895) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.2741 data: 0.1784 max mem: 3531 +Test: [ 80/517] eta: 0:02:16 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.4359 (0.7762) acc1: 88.0208 (52.0222) acc5: 100.0000 (100.0000) acc1_neg: 72.656 acc5_neg: 100.000 time: 0.2646 data: 0.1691 max mem: 3531 +Test: [ 90/517] eta: 0:02:10 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.4085 (0.7326) acc1: 91.9271 (56.6506) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2652 data: 0.1721 max mem: 3531 +Test: [100/517] eta: 0:02:07 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.3564 (0.6960) acc1: 95.0521 (60.4734) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2849 data: 0.1910 max mem: 3531 +Test: [110/517] eta: 0:02:02 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.3353 (0.6613) acc1: 96.6146 (63.7786) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2786 data: 0.1838 max mem: 3531 +Test: [120/517] eta: 0:01:57 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.3130 (0.6329) acc1: 98.4375 (66.5913) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2449 data: 0.1528 max mem: 3531 +Test: [130/517] eta: 0:01:53 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.3057 (0.6070) acc1: 98.6979 (69.0542) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2499 data: 0.1563 max mem: 3531 +Test: [140/517] eta: 0:01:49 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.2872 (0.5841) acc1: 99.4792 (71.2175) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2599 data: 0.1638 max mem: 3531 +Test: [150/517] eta: 0:01:46 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.2735 (0.5631) acc1: 99.7396 (73.0857) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2749 data: 0.1805 max mem: 3531 +Test: [160/517] eta: 0:01:43 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.3016 (0.5507) acc1: 97.3958 (74.3255) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2799 data: 0.1864 max mem: 3531 +Test: [170/517] eta: 0:01:39 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.3546 (0.5373) acc1: 94.2708 (75.5452) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2599 data: 0.1671 max mem: 3531 +Test: [180/517] eta: 0:01:36 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.2696 (0.5217) acc1: 97.6562 (76.8229) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2549 data: 0.1631 max mem: 3531 +Test: [190/517] eta: 0:01:32 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.2326 (0.5065) acc1: 99.4792 (78.0159) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2549 data: 0.1603 max mem: 3531 +Test: [200/517] eta: 0:01:29 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.2362 (0.4937) acc1: 99.7396 (79.0889) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2599 data: 0.1650 max mem: 3531 +Test: [210/517] eta: 0:01:26 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.2662 (0.4888) acc1: 96.3542 (79.6776) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2549 data: 0.1569 max mem: 3532 +Test: [220/517] eta: 0:01:23 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.3865 (0.4837) acc1: 91.1458 (80.2449) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2699 data: 0.1711 max mem: 3532 +Test: [230/517] eta: 0:01:20 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.3528 (0.4779) acc1: 94.2708 (80.9400) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2599 data: 0.1670 max mem: 3532 +Test: [240/517] eta: 0:01:17 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.3536 (0.4742) acc1: 96.3542 (81.5061) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.2449 data: 0.1498 max mem: 3532 +Test: [250/517] eta: 0:01:14 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.3415 (0.4682) acc1: 97.3958 (82.1298) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2649 data: 0.1656 max mem: 3532 +Test: [260/517] eta: 0:01:11 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.2920 (0.4613) acc1: 99.2188 (82.7816) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2599 data: 0.1636 max mem: 3532 +Test: [270/517] eta: 0:01:08 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.3157 (0.4573) acc1: 98.6979 (83.3084) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2599 data: 0.1683 max mem: 3532 +Test: [280/517] eta: 0:01:05 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.3386 (0.4532) acc1: 97.1354 (83.7420) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.2502 data: 0.1603 max mem: 3532 +Test: [290/517] eta: 0:01:02 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.3197 (0.4482) acc1: 97.9167 (84.2318) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2499 data: 0.1573 max mem: 3532 +Test: [300/517] eta: 0:00:59 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.3098 (0.4437) acc1: 98.6979 (84.7098) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2648 data: 0.1715 max mem: 3532 +Test: [310/517] eta: 0:00:56 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.3178 (0.4403) acc1: 98.6979 (85.1336) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2556 data: 0.1617 max mem: 3532 +Test: [320/517] eta: 0:00:53 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.3156 (0.4353) acc1: 98.9583 (85.5586) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2513 data: 0.1565 max mem: 3532 +Test: [330/517] eta: 0:00:51 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.2493 (0.4295) acc1: 99.2188 (85.9784) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2595 data: 0.1681 max mem: 3532 +Test: [340/517] eta: 0:00:48 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.2388 (0.4257) acc1: 99.4792 (86.2460) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.2591 data: 0.1689 max mem: 3532 +Test: [350/517] eta: 0:00:45 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.3661 (0.4247) acc1: 92.9688 (86.4324) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.2754 data: 0.1816 max mem: 3532 +Test: [360/517] eta: 0:00:42 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.3661 (0.4225) acc1: 92.9688 (86.6610) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2792 data: 0.1795 max mem: 3532 +Test: [370/517] eta: 0:00:40 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.3000 (0.4184) acc1: 97.9167 (86.9995) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2601 data: 0.1600 max mem: 3532 +Test: [380/517] eta: 0:00:37 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.2763 (0.4154) acc1: 97.9167 (87.2252) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2664 data: 0.1706 max mem: 3532 +Test: [390/517] eta: 0:00:34 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.3066 (0.4127) acc1: 95.5729 (87.4447) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2589 data: 0.1636 max mem: 3532 +Test: [400/517] eta: 0:00:31 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.2849 (0.4094) acc1: 96.6146 (87.7124) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2497 data: 0.1515 max mem: 3532 +Test: [410/517] eta: 0:00:28 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.3025 (0.4073) acc1: 97.1354 (87.9302) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2549 data: 0.1597 max mem: 3532 +Test: [420/517] eta: 0:00:26 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.2939 (0.4044) acc1: 98.4375 (88.1792) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2586 data: 0.1635 max mem: 3532 +Test: [430/517] eta: 0:00:23 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.2856 (0.4020) acc1: 98.4375 (88.3894) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2661 data: 0.1708 max mem: 3532 +Test: [440/517] eta: 0:00:20 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.3081 (0.4001) acc1: 97.3958 (88.5824) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2606 data: 0.1664 max mem: 3533 +Test: [450/517] eta: 0:00:18 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.3098 (0.3979) acc1: 97.3958 (88.7570) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.2651 data: 0.1716 max mem: 3533 +Test: [460/517] eta: 0:00:15 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.3098 (0.3963) acc1: 96.0938 (88.8620) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2463 data: 0.1546 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.2986 (0.3941) acc1: 97.3958 (89.0581) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2285 data: 0.1345 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.2976 (0.3921) acc1: 98.6979 (89.2607) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2120 data: 0.1183 max mem: 3533 +Test: [490/517] eta: 0:00:07 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.2820 (0.3898) acc1: 99.2188 (89.4449) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1888 data: 0.0977 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.3099 (0.3898) acc1: 98.4375 (89.5126) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.1764 data: 0.0868 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.3442 (0.3893) acc1: 96.6146 (89.5696) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1701 data: 0.0805 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 40.826 acc5_pos: 100.000 loss: 0.3227 (0.3882) acc1: 97.6562 (89.6728) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1808 data: 0.0895 max mem: 3533 +Test: Total time: 0:02:15 (0.2616 s / it) +* Overall Acc@1 89.673 Acc@5 100.000 + target == real Acc@1 28.311 Acc@5 100.000 + target == fake Acc@1 96.038 Acc@5 100.000 + loss 0.388 + PR-AUC 0.9559 ROC-AUC 0.7373 EER 32.75% (thr=0.7090) +Accuracy of the network on the 198427 test images: 89.7% +[rank0]:[W725 17:34:56.711087025 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0725 17:35:18.780000 93808 torch/distributed/run.py:766] +W0725 17:35:18.780000 93808 torch/distributed/run.py:766] ***************************************** +W0725 17:35:18.780000 93808 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:35:18.780000 93808 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:35:55.794520199 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:35:56.096565843 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 17:35:56.664871548 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:35:56.018839639 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:35:57.404581293 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W725 17:35:57.404693648 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:35:57.440394345 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank6]:[W725 17:35:57.441827323 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed2_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed2_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:41:40 acc1_pos: 27.344 acc5_pos: 100.000 loss: 1.0467 (1.0467) acc1: 27.3438 (27.3438) acc5: 100.0000 (100.0000) time: 6.4290 data: 4.9726 max mem: 3531 +Test: [ 10/389] eta: 0:04:14 acc1_pos: 92.708 acc5_pos: 100.000 loss: 0.5382 (0.6958) acc1: 78.9062 (60.7244) acc5: 100.0000 (100.0000) time: 0.6707 data: 0.4525 max mem: 3531 +Test: [ 20/389] eta: 0:02:26 acc1_pos: 26.042 acc5_pos: 100.000 loss: 0.6450 (0.7436) acc1: 57.8125 (54.5387) acc5: 100.0000 (100.0000) time: 0.0959 data: 0.0005 max mem: 3531 +Test: [ 30/389] eta: 0:02:11 acc1_pos: 59.635 acc5_pos: 100.000 loss: 0.6521 (0.7213) acc1: 58.8542 (56.2920) acc5: 100.0000 (100.0000) time: 0.1994 data: 0.1058 max mem: 3531 +Test: [ 40/389] eta: 0:01:57 acc1_pos: 2.344 acc5_pos: 100.000 loss: 0.7198 (0.7638) acc1: 51.8229 (52.3247) acc5: 100.0000 (100.0000) time: 0.2710 data: 0.1801 max mem: 3531 +Test: [ 50/389] eta: 0:01:46 acc1_pos: 89.323 acc5_pos: 100.000 loss: 0.7058 (0.7198) acc1: 52.8646 (57.1181) acc5: 100.0000 (100.0000) time: 0.2312 data: 0.1397 max mem: 3531 +Test: [ 60/389] eta: 0:01:35 acc1_pos: 29.167 acc5_pos: 100.000 loss: 0.4207 (0.7041) acc1: 87.2396 (58.4273) acc5: 100.0000 (100.0000) time: 0.2003 data: 0.1089 max mem: 3531 +Test: [ 70/389] eta: 0:01:26 acc1_pos: 40.104 acc5_pos: 100.000 loss: 0.8023 (0.7305) acc1: 41.1458 (55.0873) acc5: 100.0000 (100.0000) time: 0.1675 data: 0.0744 max mem: 3531 +Test: [ 80/389] eta: 0:01:20 acc1_pos: 91.406 acc5_pos: 100.000 loss: 0.8937 (0.7324) acc1: 27.3438 (54.5878) acc5: 100.0000 (100.0000) time: 0.1648 data: 0.0704 max mem: 3531 +Test: [ 90/389] eta: 0:01:14 acc1_pos: 63.281 acc5_pos: 100.000 loss: 0.4081 (0.6994) acc1: 77.6042 (57.3346) acc5: 100.0000 (100.0000) time: 0.1716 data: 0.0792 max mem: 3531 +Test: [100/389] eta: 0:01:09 acc1_pos: 25.781 acc5_pos: 100.000 loss: 0.6998 (0.7182) acc1: 52.8646 (55.6467) acc5: 100.0000 (100.0000) time: 0.1590 data: 0.0681 max mem: 3531 +Test: [110/389] eta: 0:01:06 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.8843 (0.7354) acc1: 44.7917 (54.7954) acc5: 100.0000 (100.0000) time: 0.1841 data: 0.0933 max mem: 3531 +Test: [120/389] eta: 0:01:03 acc1_pos: 64.323 acc5_pos: 100.000 loss: 0.6486 (0.7319) acc1: 64.3229 (55.5376) acc5: 100.0000 (100.0000) time: 0.2070 data: 0.1153 max mem: 3531 +Test: [130/389] eta: 0:01:00 acc1_pos: 54.601 acc5_pos: 100.000 loss: 0.6214 (0.7231) acc1: 66.1458 (56.3395) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.2060 data: 0.1134 max mem: 3531 +Test: [140/389] eta: 0:00:56 acc1_pos: 54.601 acc5_pos: 100.000 loss: 0.3867 (0.7060) acc1: 79.4271 (57.7368) acc5: 100.0000 (100.0000) acc1_neg: 16.146 acc5_neg: 100.000 time: 0.1924 data: 0.0990 max mem: 3531 +Test: [150/389] eta: 0:00:54 acc1_pos: 54.601 acc5_pos: 100.000 loss: 0.4173 (0.7020) acc1: 73.4375 (57.8384) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1834 data: 0.0903 max mem: 3531 +Test: [160/389] eta: 0:00:51 acc1_pos: 54.601 acc5_pos: 100.000 loss: 0.5599 (0.6868) acc1: 63.5417 (59.2505) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2019 data: 0.1085 max mem: 3531 +Test: [170/389] eta: 0:00:49 acc1_pos: 54.601 acc5_pos: 100.000 loss: 0.2189 (0.6656) acc1: 98.1771 (60.9954) acc5: 100.0000 (100.0000) acc1_neg: 67.969 acc5_neg: 100.000 time: 0.2008 data: 0.1085 max mem: 3531 +Test: [180/389] eta: 0:00:46 acc1_pos: 54.601 acc5_pos: 100.000 loss: 0.3730 (0.6612) acc1: 84.1146 (61.4713) acc5: 100.0000 (100.0000) acc1_neg: 64.583 acc5_neg: 100.000 time: 0.1840 data: 0.0919 max mem: 3531 +Test: [190/389] eta: 0:00:43 acc1_pos: 54.601 acc5_pos: 100.000 loss: 0.4966 (0.6555) acc1: 71.6146 (62.1005) acc5: 100.0000 (100.0000) acc1_neg: 46.094 acc5_neg: 100.000 time: 0.1785 data: 0.0801 max mem: 3531 +Test: [200/389] eta: 0:00:41 acc1_pos: 54.601 acc5_pos: 100.000 loss: 0.4966 (0.6481) acc1: 71.6146 (62.7656) acc5: 100.0000 (100.0000) acc1_neg: 63.802 acc5_neg: 100.000 time: 0.1887 data: 0.0900 max mem: 3531 +Test: [210/389] eta: 0:00:38 acc1_pos: 54.601 acc5_pos: 100.000 loss: 0.4700 (0.6364) acc1: 75.2604 (63.6416) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1827 data: 0.0897 max mem: 3532 +Test: [220/389] eta: 0:00:36 acc1_pos: 54.601 acc5_pos: 100.000 loss: 0.2130 (0.6185) acc1: 99.7396 (65.1053) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1799 data: 0.0888 max mem: 3532 +Test: [230/389] eta: 0:00:33 acc1_pos: 54.601 acc5_pos: 100.000 loss: 0.3327 (0.6142) acc1: 91.1458 (65.4311) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.1921 data: 0.0997 max mem: 3532 +Test: [240/389] eta: 0:00:31 acc1_pos: 54.601 acc5_pos: 100.000 loss: 0.4575 (0.6076) acc1: 82.2917 (66.0886) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1849 data: 0.0909 max mem: 3532 +Test: [250/389] eta: 0:00:29 acc1_pos: 54.601 acc5_pos: 100.000 loss: 0.3196 (0.5936) acc1: 95.0521 (67.3722) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1700 data: 0.0775 max mem: 3532 +Test: [260/389] eta: 0:00:26 acc1_pos: 54.601 acc5_pos: 100.000 loss: 0.3070 (0.5912) acc1: 96.6146 (67.7053) acc5: 100.0000 (100.0000) acc1_neg: 25.781 acc5_neg: 100.000 time: 0.1754 data: 0.0845 max mem: 3532 +Test: [270/389] eta: 0:00:24 acc1_pos: 54.601 acc5_pos: 100.000 loss: 0.5369 (0.5966) acc1: 75.2604 (67.1635) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1875 data: 0.0961 max mem: 3532 +Test: [280/389] eta: 0:00:22 acc1_pos: 54.601 acc5_pos: 100.000 loss: 0.5038 (0.5944) acc1: 75.0000 (67.3636) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1874 data: 0.0956 max mem: 3532 +Test: [290/389] eta: 0:00:20 acc1_pos: 54.601 acc5_pos: 100.000 loss: 0.4098 (0.5881) acc1: 88.8021 (68.0287) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.2022 data: 0.1099 max mem: 3532 +Test: [300/389] eta: 0:00:18 acc1_pos: 54.601 acc5_pos: 100.000 loss: 0.4236 (0.5833) acc1: 88.8021 (68.6618) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1944 data: 0.1016 max mem: 3532 +Test: [310/389] eta: 0:00:16 acc1_pos: 54.601 acc5_pos: 100.000 loss: 0.5071 (0.5840) acc1: 79.6875 (68.5407) acc5: 100.0000 (100.0000) acc1_neg: 36.458 acc5_neg: 100.000 time: 0.1754 data: 0.0840 max mem: 3532 +Test: [320/389] eta: 0:00:14 acc1_pos: 54.601 acc5_pos: 100.000 loss: 0.5210 (0.5778) acc1: 69.5312 (69.0867) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1755 data: 0.0833 max mem: 3532 +Test: [330/389] eta: 0:00:12 acc1_pos: 54.601 acc5_pos: 100.000 loss: 0.6149 (0.5819) acc1: 62.2396 (68.6194) acc5: 100.0000 (100.0000) acc1_neg: 62.240 acc5_neg: 100.000 time: 0.1765 data: 0.0822 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 54.601 acc5_pos: 100.000 loss: 0.4364 (0.5723) acc1: 69.7917 (69.4358) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1737 data: 0.0806 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 54.601 acc5_pos: 100.000 loss: 0.2821 (0.5770) acc1: 98.1771 (69.2107) acc5: 100.0000 (100.0000) acc1_neg: 47.917 acc5_neg: 100.000 time: 0.1628 data: 0.0709 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 54.601 acc5_pos: 100.000 loss: 0.4207 (0.5745) acc1: 82.5521 (69.4303) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1525 data: 0.0612 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 54.601 acc5_pos: 100.000 loss: 0.4207 (0.5742) acc1: 84.8958 (69.4168) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1303 data: 0.0402 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 54.601 acc5_pos: 100.000 loss: 0.4007 (0.5703) acc1: 83.8542 (69.7028) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1230 data: 0.0334 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 54.601 acc5_pos: 100.000 loss: 0.3365 (0.5669) acc1: 91.9271 (70.0043) acc5: 100.0000 (100.0000) acc1_neg: 59.563 acc5_neg: 100.000 time: 0.1186 data: 0.0256 max mem: 3532 +Test: Total time: 0:01:15 (0.1948 s / it) +* Overall Acc@1 70.004 Acc@5 100.000 + target == real Acc@1 55.535 Acc@5 100.000 + target == fake Acc@1 76.730 Acc@5 100.000 + loss 0.567 + PR-AUC 0.8523 ROC-AUC 0.7389 EER 33.04% (thr=0.5894) +Accuracy of the network on the 149358 test images: 70.0% +[rank0]:[W725 17:37:23.784289512 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 17:37:43.783000 94365 torch/distributed/run.py:766] +W0725 17:37:43.783000 94365 torch/distributed/run.py:766] ***************************************** +W0725 17:37:43.783000 94365 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:37:43.783000 94365 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:38:15.226194216 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 17:38:15.430876512 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:38:15.606458105 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:38:16.136989002 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:38:16.150428581 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:38:16.154446795 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:38:16.275512818 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:38:16.342602690 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed2_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed2_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:32:59 acc1_pos: 83.854 acc5_pos: 100.000 loss: 0.3726 (0.3726) acc1: 83.8542 (83.8542) acc5: 100.0000 (100.0000) time: 5.9448 data: 4.4455 max mem: 3531 +Test: [ 10/333] eta: 0:03:21 acc1_pos: 88.281 acc5_pos: 100.000 loss: 0.2925 (0.3092) acc1: 87.2396 (87.1922) acc5: 100.0000 (100.0000) time: 0.6224 data: 0.4045 max mem: 3531 +Test: [ 20/333] eta: 0:01:56 acc1_pos: 94.010 acc5_pos: 100.000 loss: 0.2871 (0.2950) acc1: 89.8438 (88.0704) acc5: 100.0000 (100.0000) time: 0.0943 data: 0.0004 max mem: 3531 +Test: [ 30/333] eta: 0:01:29 acc1_pos: 50.781 acc5_pos: 100.000 loss: 0.3799 (0.4514) acc1: 73.6979 (75.1512) acc5: 100.0000 (100.0000) time: 0.1141 data: 0.0157 max mem: 3531 +Test: [ 40/333] eta: 0:01:18 acc1_pos: 44.271 acc5_pos: 100.000 loss: 0.7892 (0.5347) acc1: 45.3125 (67.9815) acc5: 100.0000 (100.0000) time: 0.1605 data: 0.0638 max mem: 3531 +Test: [ 50/333] eta: 0:01:13 acc1_pos: 48.177 acc5_pos: 100.000 loss: 0.7892 (0.5870) acc1: 46.3542 (63.8174) acc5: 100.0000 (100.0000) time: 0.2050 data: 0.1106 max mem: 3531 +Test: [ 60/333] eta: 0:01:08 acc1_pos: 36.979 acc5_pos: 100.000 loss: 0.8068 (0.6243) acc1: 46.3542 (60.8052) acc5: 100.0000 (100.0000) time: 0.2093 data: 0.1118 max mem: 3531 +Test: [ 70/333] eta: 0:01:02 acc1_pos: 47.396 acc5_pos: 100.000 loss: 0.8053 (0.6493) acc1: 47.1354 (59.0302) acc5: 100.0000 (100.0000) time: 0.1813 data: 0.0839 max mem: 3531 +Test: [ 80/333] eta: 0:00:58 acc1_pos: 50.781 acc5_pos: 100.000 loss: 0.7720 (0.6679) acc1: 47.6562 (57.5135) acc5: 100.0000 (100.0000) time: 0.1799 data: 0.0839 max mem: 3531 +Test: [ 90/333] eta: 0:00:55 acc1_pos: 59.635 acc5_pos: 100.000 loss: 0.7323 (0.6707) acc1: 52.6042 (57.4290) acc5: 100.0000 (100.0000) time: 0.1890 data: 0.0926 max mem: 3531 +Test: [100/333] eta: 0:00:51 acc1_pos: 62.760 acc5_pos: 100.000 loss: 0.6751 (0.6739) acc1: 56.7708 (57.2968) acc5: 100.0000 (100.0000) time: 0.1762 data: 0.0820 max mem: 3531 +Test: [110/333] eta: 0:00:48 acc1_pos: 55.469 acc5_pos: 100.000 loss: 0.7068 (0.6758) acc1: 54.1667 (57.2776) acc5: 100.0000 (100.0000) time: 0.1744 data: 0.0808 max mem: 3531 +Test: [120/333] eta: 0:00:45 acc1_pos: 63.281 acc5_pos: 100.000 loss: 0.6663 (0.6758) acc1: 55.4688 (57.3196) acc5: 100.0000 (100.0000) time: 0.1736 data: 0.0780 max mem: 3531 +Test: [130/333] eta: 0:00:42 acc1_pos: 72.396 acc5_pos: 100.000 loss: 0.6326 (0.6738) acc1: 59.8958 (57.6356) acc5: 100.0000 (100.0000) time: 0.1699 data: 0.0716 max mem: 3531 +Test: [140/333] eta: 0:00:40 acc1_pos: 74.479 acc5_pos: 100.000 loss: 0.6514 (0.6734) acc1: 59.8958 (57.6278) acc5: 100.0000 (100.0000) time: 0.1749 data: 0.0765 max mem: 3531 +Test: [150/333] eta: 0:00:37 acc1_pos: 60.938 acc5_pos: 100.000 loss: 0.6673 (0.6755) acc1: 55.2083 (57.4607) acc5: 100.0000 (100.0000) time: 0.1865 data: 0.0893 max mem: 3531 +Test: [160/333] eta: 0:00:35 acc1_pos: 45.724 acc5_pos: 100.000 loss: 0.7021 (0.6846) acc1: 54.4271 (57.0102) acc5: 100.0000 (100.0000) acc1_neg: 42.448 acc5_neg: 100.000 time: 0.1849 data: 0.0893 max mem: 3531 +Test: [170/333] eta: 0:00:33 acc1_pos: 45.724 acc5_pos: 100.000 loss: 0.7874 (0.6934) acc1: 53.6458 (56.7754) acc5: 100.0000 (100.0000) acc1_neg: 57.812 acc5_neg: 100.000 time: 0.1798 data: 0.0834 max mem: 3531 +Test: [180/333] eta: 0:00:30 acc1_pos: 45.724 acc5_pos: 100.000 loss: 0.7663 (0.6950) acc1: 57.0312 (56.8974) acc5: 100.0000 (100.0000) acc1_neg: 59.635 acc5_neg: 100.000 time: 0.1699 data: 0.0738 max mem: 3531 +Test: [190/333] eta: 0:00:28 acc1_pos: 45.724 acc5_pos: 100.000 loss: 0.6881 (0.6928) acc1: 60.1562 (57.3394) acc5: 100.0000 (100.0000) acc1_neg: 61.198 acc5_neg: 100.000 time: 0.1635 data: 0.0668 max mem: 3531 +Test: [200/333] eta: 0:00:26 acc1_pos: 45.724 acc5_pos: 100.000 loss: 0.6685 (0.6917) acc1: 64.3229 (57.7853) acc5: 100.0000 (100.0000) acc1_neg: 63.802 acc5_neg: 100.000 time: 0.1699 data: 0.0712 max mem: 3531 +Test: [210/333] eta: 0:00:24 acc1_pos: 45.724 acc5_pos: 100.000 loss: 0.7111 (0.6948) acc1: 62.7604 (57.9248) acc5: 100.0000 (100.0000) acc1_neg: 51.302 acc5_neg: 100.000 time: 0.1749 data: 0.0765 max mem: 3532 +Test: [220/333] eta: 0:00:22 acc1_pos: 45.724 acc5_pos: 100.000 loss: 0.7447 (0.7002) acc1: 60.6771 (57.8750) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.1799 data: 0.0818 max mem: 3532 +Test: [230/333] eta: 0:00:20 acc1_pos: 45.724 acc5_pos: 100.000 loss: 0.5297 (0.6917) acc1: 76.8229 (58.8012) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.1699 data: 0.0717 max mem: 3532 +Test: [240/333] eta: 0:00:17 acc1_pos: 45.724 acc5_pos: 100.000 loss: 0.5297 (0.6876) acc1: 75.2604 (59.2723) acc5: 100.0000 (100.0000) acc1_neg: 71.354 acc5_neg: 100.000 time: 0.1599 data: 0.0613 max mem: 3532 +Test: [250/333] eta: 0:00:16 acc1_pos: 45.724 acc5_pos: 100.000 loss: 0.6062 (0.6853) acc1: 69.0104 (59.6157) acc5: 100.0000 (100.0000) acc1_neg: 59.635 acc5_neg: 100.000 time: 0.1699 data: 0.0709 max mem: 3532 +Test: [260/333] eta: 0:00:14 acc1_pos: 45.724 acc5_pos: 100.000 loss: 0.6381 (0.6845) acc1: 67.4479 (59.8280) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.1756 data: 0.0780 max mem: 3532 +Test: [270/333] eta: 0:00:12 acc1_pos: 45.724 acc5_pos: 100.000 loss: 0.6370 (0.6823) acc1: 67.1875 (60.1611) acc5: 100.0000 (100.0000) acc1_neg: 68.490 acc5_neg: 100.000 time: 0.1757 data: 0.0806 max mem: 3532 +Test: [280/333] eta: 0:00:10 acc1_pos: 45.724 acc5_pos: 100.000 loss: 0.6370 (0.6852) acc1: 65.1042 (60.0070) acc5: 100.0000 (100.0000) acc1_neg: 60.156 acc5_neg: 100.000 time: 0.1882 data: 0.0945 max mem: 3532 +Test: [290/333] eta: 0:00:08 acc1_pos: 45.724 acc5_pos: 100.000 loss: 0.8763 (0.6918) acc1: 48.6979 (59.5889) acc5: 100.0000 (100.0000) acc1_neg: 52.083 acc5_neg: 100.000 time: 0.1981 data: 0.1025 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 45.724 acc5_pos: 100.000 loss: 0.8731 (0.6968) acc1: 50.2604 (59.3473) acc5: 100.0000 (100.0000) acc1_neg: 53.646 acc5_neg: 100.000 time: 0.2066 data: 0.1096 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 45.724 acc5_pos: 100.000 loss: 0.7424 (0.6933) acc1: 60.1562 (59.7342) acc5: 100.0000 (100.0000) acc1_neg: 71.615 acc5_neg: 100.000 time: 0.2100 data: 0.1157 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 45.724 acc5_pos: 100.000 loss: 0.5708 (0.6906) acc1: 72.3958 (60.0589) acc5: 100.0000 (100.0000) acc1_neg: 50.260 acc5_neg: 100.000 time: 0.1978 data: 0.1035 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 45.724 acc5_pos: 100.000 loss: 0.8358 (0.6976) acc1: 51.5625 (59.7401) acc5: 100.0000 (100.0000) acc1_neg: 51.302 acc5_neg: 100.000 time: 0.1548 data: 0.0615 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 45.724 acc5_pos: 100.000 loss: 0.8954 (0.6989) acc1: 50.7812 (59.6810) acc5: 100.0000 (100.0000) acc1_neg: 50.141 acc5_neg: 100.000 time: 0.1571 data: 0.0615 max mem: 3532 +Test: Total time: 0:01:03 (0.1911 s / it) +* Overall Acc@1 59.681 Acc@5 100.000 + target == real Acc@1 57.441 Acc@5 100.000 + target == fake Acc@1 61.675 Acc@5 100.000 + loss 0.699 + PR-AUC 0.6476 ROC-AUC 0.6360 EER 40.50% (thr=0.5156) +Accuracy of the network on the 127843 test images: 59.7% +[rank0]:[W725 17:39:32.968253415 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 17:39:50.785000 94925 torch/distributed/run.py:766] +W0725 17:39:50.785000 94925 torch/distributed/run.py:766] ***************************************** +W0725 17:39:50.785000 94925 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:39:50.785000 94925 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W725 17:40:17.563637337 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:40:17.579788283 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:40:17.007082202 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:40:18.093898045 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:40:18.118141091 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:40:18.306414952 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:40:18.354878881 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:40:18.548645929 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed2_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed2_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:23:52 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.1599 (0.1599) acc1: 96.8750 (96.8750) acc5: 100.0000 (100.0000) time: 6.0945 data: 4.7106 max mem: 3531 +Test: [ 10/235] eta: 0:02:23 acc1_pos: 96.615 acc5_pos: 100.000 loss: 0.1666 (0.1676) acc1: 96.6146 (96.4725) acc5: 100.0000 (100.0000) time: 0.6393 data: 0.4287 max mem: 3531 +Test: [ 20/235] eta: 0:01:31 acc1_pos: 95.833 acc5_pos: 100.000 loss: 0.1638 (0.1651) acc1: 96.6146 (96.5774) acc5: 100.0000 (100.0000) time: 0.1422 data: 0.0480 max mem: 3531 +Test: [ 30/235] eta: 0:01:20 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.1590 (0.1632) acc1: 96.6146 (96.6230) acc5: 100.0000 (100.0000) time: 0.2599 data: 0.1659 max mem: 3531 +Test: [ 40/235] eta: 0:01:09 acc1_pos: 95.833 acc5_pos: 100.000 loss: 0.1601 (0.1632) acc1: 96.6146 (96.5193) acc5: 100.0000 (100.0000) time: 0.2813 data: 0.1883 max mem: 3531 +Test: [ 50/235] eta: 0:00:57 acc1_pos: 95.052 acc5_pos: 100.000 loss: 0.1605 (0.1639) acc1: 96.3542 (96.4665) acc5: 100.0000 (100.0000) time: 0.1849 data: 0.0898 max mem: 3531 +Test: [ 60/235] eta: 0:00:50 acc1_pos: 96.615 acc5_pos: 100.000 loss: 0.1681 (0.1649) acc1: 96.3542 (96.4353) acc5: 100.0000 (100.0000) time: 0.1482 data: 0.0526 max mem: 3531 +Test: [ 70/235] eta: 0:00:45 acc1_pos: 95.385 acc5_pos: 100.000 loss: 0.1838 (0.3603) acc1: 94.7917 (86.3850) acc5: 100.0000 (100.0000) acc1_neg: 15.104 acc5_neg: 100.000 time: 0.1844 data: 0.0911 max mem: 3531 +Test: [ 80/235] eta: 0:00:41 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.7247 (0.5318) acc1: 15.6250 (77.5913) acc5: 100.0000 (100.0000) acc1_neg: 13.021 acc5_neg: 100.000 time: 0.1958 data: 0.1013 max mem: 3531 +Test: [ 90/235] eta: 0:00:37 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.7515 (0.6655) acc1: 15.6250 (70.7561) acc5: 100.0000 (100.0000) acc1_neg: 16.146 acc5_neg: 100.000 time: 0.1899 data: 0.0945 max mem: 3531 +Test: [100/235] eta: 0:00:33 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.7515 (0.7711) acc1: 15.8854 (65.3775) acc5: 100.0000 (100.0000) acc1_neg: 14.583 acc5_neg: 100.000 time: 0.1842 data: 0.0907 max mem: 3531 +Test: [110/235] eta: 0:00:30 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.7192 (0.8571) acc1: 15.8854 (60.9445) acc5: 100.0000 (100.0000) acc1_neg: 17.448 acc5_neg: 100.000 time: 0.1749 data: 0.0830 max mem: 3531 +Test: [120/235] eta: 0:00:27 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.7830 (0.9428) acc1: 13.5417 (56.7450) acc5: 100.0000 (100.0000) acc1_neg: 8.854 acc5_neg: 100.000 time: 0.1757 data: 0.0850 max mem: 3531 +Test: [130/235] eta: 0:00:24 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.8986 (1.0155) acc1: 10.1562 (53.1846) acc5: 100.0000 (100.0000) acc1_neg: 10.156 acc5_neg: 100.000 time: 0.1705 data: 0.0794 max mem: 3531 +Test: [140/235] eta: 0:00:21 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.9228 (1.0809) acc1: 8.8542 (50.0536) acc5: 100.0000 (100.0000) acc1_neg: 6.510 acc5_neg: 100.000 time: 0.1934 data: 0.0980 max mem: 3531 +Test: [150/235] eta: 0:00:19 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.9168 (1.1347) acc1: 9.1146 (47.3734) acc5: 100.0000 (100.0000) acc1_neg: 10.417 acc5_neg: 100.000 time: 0.2102 data: 0.1159 max mem: 3531 +Test: [160/235] eta: 0:00:17 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.8900 (1.1819) acc1: 9.3750 (45.0731) acc5: 100.0000 (100.0000) acc1_neg: 11.979 acc5_neg: 100.000 time: 0.2099 data: 0.1178 max mem: 3531 +Test: [170/235] eta: 0:00:14 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.8204 (1.2010) acc1: 11.9792 (43.7226) acc5: 100.0000 (100.0000) acc1_neg: 22.396 acc5_neg: 100.000 time: 0.2250 data: 0.1302 max mem: 3531 +Test: [180/235] eta: 0:00:12 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.4781 (1.2147) acc1: 22.9167 (42.6177) acc5: 100.0000 (100.0000) acc1_neg: 24.479 acc5_neg: 100.000 time: 0.2449 data: 0.1489 max mem: 3531 +Test: [190/235] eta: 0:00:10 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.4355 (1.2265) acc1: 23.9583 (41.6476) acc5: 100.0000 (100.0000) acc1_neg: 23.438 acc5_neg: 100.000 time: 0.2189 data: 0.1216 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.4420 (1.2385) acc1: 23.1771 (40.6976) acc5: 100.0000 (100.0000) acc1_neg: 22.396 acc5_neg: 100.000 time: 0.1678 data: 0.0707 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.4627 (1.2479) acc1: 22.3958 (39.8931) acc5: 100.0000 (100.0000) acc1_neg: 22.656 acc5_neg: 100.000 time: 0.1864 data: 0.0918 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.4392 (1.2569) acc1: 23.6979 (39.1627) acc5: 100.0000 (100.0000) acc1_neg: 24.479 acc5_neg: 100.000 time: 0.2065 data: 0.1149 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.4358 (1.2646) acc1: 24.2188 (38.5439) acc5: 100.0000 (100.0000) acc1_neg: 24.740 acc5_neg: 100.000 time: 0.1665 data: 0.0759 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 95.385 acc5_pos: 100.000 loss: 1.4371 (1.2686) acc1: 24.4792 (38.3627) acc5: 100.0000 (100.0000) acc1_neg: 26.087 acc5_neg: 100.000 time: 0.1114 data: 0.0231 max mem: 3532 +Test: Total time: 0:00:50 (0.2169 s / it) +* Overall Acc@1 38.363 Acc@5 100.000 + target == real Acc@1 96.431 Acc@5 100.000 + target == fake Acc@1 17.361 Acc@5 100.000 + loss 1.269 + PR-AUC 0.8679 ROC-AUC 0.7157 EER 34.49% (thr=0.1235) +Accuracy of the network on the 89879 test images: 38.4% +[rank0]:[W725 17:41:19.933290860 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 17:41:38.429000 95481 torch/distributed/run.py:766] +W0725 17:41:38.429000 95481 torch/distributed/run.py:766] ***************************************** +W0725 17:41:38.429000 95481 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:41:38.429000 95481 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:42:05.236316504 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:42:05.475724404 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:42:06.042843154 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:42:06.065480919 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +| distributed init (rank 5): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:42:06.238145820 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W725 17:42:06.238182936 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W725 17:42:06.238215144 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 17:42:06.250964075 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed2_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed2_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:22:46 acc1_pos: 63.802 acc5_pos: 100.000 loss: 0.6346 (0.6346) acc1: 63.8021 (63.8021) acc5: 100.0000 (100.0000) time: 5.0804 data: 3.1602 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/269] eta: 0:02:25 acc1_pos: 51.562 acc5_pos: 100.000 loss: 0.6346 (0.5891) acc1: 62.2396 (66.3826) acc5: 100.0000 (100.0000) time: 0.5602 data: 0.3041 max mem: 3531 +Test: [ 20/269] eta: 0:01:24 acc1_pos: 72.135 acc5_pos: 100.000 loss: 0.5775 (0.5574) acc1: 67.4479 (70.5605) acc5: 100.0000 (100.0000) time: 0.1007 data: 0.0095 max mem: 3531 +Test: [ 30/269] eta: 0:01:02 acc1_pos: 75.781 acc5_pos: 100.000 loss: 0.5315 (0.5560) acc1: 73.9583 (71.2702) acc5: 100.0000 (100.0000) time: 0.0966 data: 0.0004 max mem: 3531 +Test: [ 40/269] eta: 0:00:56 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.5823 (0.5716) acc1: 72.9167 (70.6110) acc5: 100.0000 (100.0000) acc1_neg: 72.917 acc5_neg: 100.000 time: 0.1503 data: 0.0547 max mem: 3531 +Test: [ 50/269] eta: 0:00:53 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.5866 (0.5873) acc1: 70.3125 (69.2198) acc5: 100.0000 (100.0000) acc1_neg: 44.271 acc5_neg: 100.000 time: 0.2199 data: 0.1272 max mem: 3531 +Test: [ 60/269] eta: 0:00:50 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.6634 (0.6149) acc1: 59.1146 (66.4746) acc5: 100.0000 (100.0000) acc1_neg: 44.010 acc5_neg: 100.000 time: 0.2345 data: 0.1425 max mem: 3531 +Test: [ 70/269] eta: 0:00:47 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.7317 (0.6248) acc1: 51.0417 (65.1042) acc5: 100.0000 (100.0000) acc1_neg: 54.167 acc5_neg: 100.000 time: 0.2179 data: 0.1269 max mem: 3531 +Test: [ 80/269] eta: 0:00:43 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.6789 (0.6305) acc1: 55.2083 (64.3004) acc5: 100.0000 (100.0000) acc1_neg: 54.688 acc5_neg: 100.000 time: 0.1879 data: 0.0938 max mem: 3531 +Test: [ 90/269] eta: 0:00:39 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.7456 (0.6546) acc1: 49.4792 (61.5814) acc5: 100.0000 (100.0000) acc1_neg: 49.479 acc5_neg: 100.000 time: 0.1649 data: 0.0677 max mem: 3531 +Test: [100/269] eta: 0:00:37 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.8059 (0.6670) acc1: 44.0104 (60.1279) acc5: 100.0000 (100.0000) acc1_neg: 38.802 acc5_neg: 100.000 time: 0.1817 data: 0.0874 max mem: 3531 +Test: [110/269] eta: 0:00:35 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.8059 (0.6816) acc1: 44.2708 (58.6688) acc5: 100.0000 (100.0000) acc1_neg: 36.198 acc5_neg: 100.000 time: 0.2206 data: 0.1302 max mem: 3531 +Test: [120/269] eta: 0:00:32 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.8247 (0.6946) acc1: 44.2708 (57.5090) acc5: 100.0000 (100.0000) acc1_neg: 35.156 acc5_neg: 100.000 time: 0.1981 data: 0.1063 max mem: 3531 +Test: [130/269] eta: 0:00:30 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.8485 (0.7075) acc1: 43.4896 (56.4528) acc5: 100.0000 (100.0000) acc1_neg: 36.979 acc5_neg: 100.000 time: 0.1942 data: 0.1008 max mem: 3531 +Test: [140/269] eta: 0:00:27 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.8883 (0.7203) acc1: 38.2812 (55.1474) acc5: 100.0000 (100.0000) acc1_neg: 34.896 acc5_neg: 100.000 time: 0.2109 data: 0.1165 max mem: 3531 +Test: [150/269] eta: 0:00:25 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.7951 (0.7224) acc1: 42.4479 (54.8841) acc5: 100.0000 (100.0000) acc1_neg: 53.125 acc5_neg: 100.000 time: 0.1821 data: 0.0852 max mem: 3531 +Test: [160/269] eta: 0:00:23 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.7275 (0.7234) acc1: 53.1250 (54.7975) acc5: 100.0000 (100.0000) acc1_neg: 57.812 acc5_neg: 100.000 time: 0.1859 data: 0.0914 max mem: 3531 +Test: [170/269] eta: 0:00:20 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.7529 (0.7280) acc1: 48.1771 (54.3159) acc5: 100.0000 (100.0000) acc1_neg: 48.177 acc5_neg: 100.000 time: 0.1909 data: 0.0989 max mem: 3531 +Test: [180/269] eta: 0:00:18 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.7437 (0.7260) acc1: 50.7812 (54.4674) acc5: 100.0000 (100.0000) acc1_neg: 51.302 acc5_neg: 100.000 time: 0.1628 data: 0.0702 max mem: 3531 +Test: [190/269] eta: 0:00:16 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.6971 (0.7278) acc1: 57.2917 (54.3180) acc5: 100.0000 (100.0000) acc1_neg: 57.292 acc5_neg: 100.000 time: 0.1567 data: 0.0622 max mem: 3531 +Test: [200/269] eta: 0:00:14 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.7857 (0.7319) acc1: 46.8750 (53.9892) acc5: 100.0000 (100.0000) acc1_neg: 56.510 acc5_neg: 100.000 time: 0.1769 data: 0.0835 max mem: 3531 +Test: [210/269] eta: 0:00:11 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.7212 (0.7300) acc1: 51.5625 (54.1889) acc5: 100.0000 (100.0000) acc1_neg: 69.010 acc5_neg: 100.000 time: 0.1774 data: 0.0875 max mem: 3532 +Test: [220/269] eta: 0:00:09 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.6405 (0.7271) acc1: 63.2812 (54.5143) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.1779 data: 0.0874 max mem: 3532 +Test: [230/269] eta: 0:00:07 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.6071 (0.7216) acc1: 66.4062 (55.0945) acc5: 100.0000 (100.0000) acc1_neg: 68.229 acc5_neg: 100.000 time: 0.1728 data: 0.0809 max mem: 3532 +Test: [240/269] eta: 0:00:05 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.6833 (0.7242) acc1: 59.1146 (54.8766) acc5: 100.0000 (100.0000) acc1_neg: 47.396 acc5_neg: 100.000 time: 0.2102 data: 0.1182 max mem: 3532 +Test: [250/269] eta: 0:00:03 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.7953 (0.7322) acc1: 42.9688 (54.1158) acc5: 100.0000 (100.0000) acc1_neg: 15.365 acc5_neg: 100.000 time: 0.2245 data: 0.1307 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.9433 (0.7416) acc1: 27.6042 (53.1679) acc5: 100.0000 (100.0000) acc1_neg: 39.844 acc5_neg: 100.000 time: 0.1783 data: 0.0828 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 82.759 acc5_pos: 100.000 loss: 0.8187 (0.7433) acc1: 33.5938 (52.9562) acc5: 100.0000 (100.0000) acc1_neg: 48.031 acc5_neg: 100.000 time: 0.1341 data: 0.0364 max mem: 3532 +Test: Total time: 0:00:53 (0.1979 s / it) +* Overall Acc@1 52.956 Acc@5 100.000 + target == real Acc@1 71.348 Acc@5 100.000 + target == fake Acc@1 50.446 Acc@5 100.000 + loss 0.743 + PR-AUC 0.9279 ROC-AUC 0.6579 EER 38.92% (thr=0.4438) +Accuracy of the network on the 103293 test images: 53.0% +[rank0]:[W725 17:43:09.000067502 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-25T17:43:15Z] eval_done tag=fixed500k_fake_n32_seed2_resampled +[2026-07-25T17:43:15Z] eval_wait tag=fixed500k_fake_n45_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed1_resampled +[2026-07-25T17:44:16Z] eval_wait tag=fixed500k_fake_n45_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed1_resampled +[2026-07-25T17:45:16Z] eval_wait tag=fixed500k_fake_n45_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed1_resampled +[2026-07-25T17:46:16Z] eval_wait tag=fixed500k_fake_n45_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed1_resampled +[2026-07-25T17:47:16Z] eval_skip tag=fixed500k_fake_n45_seed1_resampled reason=complete_after_lock output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed1_resampled +[2026-07-25T17:47:16Z] eval_start tag=fixed500k_fake_n45_seed2_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed2_resampled_n116-041-023_20260725_112559/best_checkpoint.pth +started_utc=2026-07-25T17:47:16Z +hostname=n116-042-186 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed2_resampled_n116-041-023_20260725_112559/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed2_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0725 17:47:28.339000 96110 torch/distributed/run.py:766] +W0725 17:47:28.339000 96110 torch/distributed/run.py:766] ***************************************** +W0725 17:47:28.339000 96110 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:47:28.339000 96110 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:47:55.918283694 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:47:56.227785949 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:47:56.569825653 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:47:56.676787169 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank1]:[W725 17:47:56.677942904 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:47:56.844441046 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +[rank0]:[W725 17:47:56.857573062 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:47:56.869028748 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed2_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed2_resampled_n116-041-023_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:20:10 acc1_pos: 74.219 acc5_pos: 100.000 loss: 0.5518 (0.5518) acc1: 74.2188 (74.2188) acc5: 100.0000 (100.0000) time: 5.1743 data: 3.1103 max mem: 3531 +Test: [ 10/234] eta: 0:02:04 acc1_pos: 67.969 acc5_pos: 100.000 loss: 0.5922 (0.6115) acc1: 67.9688 (65.6013) acc5: 100.0000 (100.0000) time: 0.5550 data: 0.2831 max mem: 3531 +Test: [ 20/234] eta: 0:01:16 acc1_pos: 97.396 acc5_pos: 100.000 loss: 0.6075 (0.6042) acc1: 63.0208 (63.4301) acc5: 100.0000 (100.0000) time: 0.1154 data: 0.0199 max mem: 3531 +Test: [ 30/234] eta: 0:01:01 acc1_pos: 90.885 acc5_pos: 100.000 loss: 0.1636 (0.4660) acc1: 97.3958 (74.1936) acc5: 100.0000 (100.0000) time: 0.1600 data: 0.0645 max mem: 3531 +Test: [ 40/234] eta: 0:00:57 acc1_pos: 99.740 acc5_pos: 100.000 loss: 0.1486 (0.4148) acc1: 97.6562 (78.9825) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1378 max mem: 3531 +Test: [ 50/234] eta: 0:00:52 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1122 (0.3546) acc1: 98.6979 (82.8942) acc5: 100.0000 (100.0000) time: 0.2647 data: 0.1739 max mem: 3531 +Test: [ 60/234] eta: 0:00:46 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.1063 (0.3138) acc1: 99.4792 (85.6856) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2049 data: 0.1125 max mem: 3531 +Test: [ 70/234] eta: 0:00:43 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.1075 (0.3447) acc1: 100.0000 (84.0926) acc5: 100.0000 (100.0000) acc1_neg: 58.333 acc5_neg: 100.000 time: 0.2039 data: 0.1077 max mem: 3531 +Test: [ 80/234] eta: 0:00:40 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.7307 (0.3735) acc1: 62.2396 (82.8286) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2599 data: 0.1624 max mem: 3531 +Test: [ 90/234] eta: 0:00:35 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.3253 (0.3646) acc1: 84.3750 (83.6596) acc5: 100.0000 (100.0000) acc1_neg: 78.646 acc5_neg: 100.000 time: 0.1950 data: 0.0972 max mem: 3531 +Test: [100/234] eta: 0:00:32 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.3857 (0.3854) acc1: 78.6458 (82.4309) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1550 data: 0.0572 max mem: 3531 +Test: [110/234] eta: 0:00:29 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.6140 (0.3994) acc1: 68.4896 (81.7943) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.2061 data: 0.1109 max mem: 3531 +Test: [120/234] eta: 0:00:27 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.6140 (0.4205) acc1: 68.4896 (80.5828) acc5: 100.0000 (100.0000) acc1_neg: 60.677 acc5_neg: 100.000 time: 0.2110 data: 0.1184 max mem: 3531 +Test: [130/234] eta: 0:00:24 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.5399 (0.4296) acc1: 69.5312 (80.2382) acc5: 100.0000 (100.0000) acc1_neg: 69.531 acc5_neg: 100.000 time: 0.2046 data: 0.1111 max mem: 3531 +Test: [140/234] eta: 0:00:21 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.5648 (0.4693) acc1: 67.7083 (78.2598) acc5: 100.0000 (100.0000) acc1_neg: 26.823 acc5_neg: 100.000 time: 0.1900 data: 0.0975 max mem: 3531 +Test: [150/234] eta: 0:00:19 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.6166 (0.4855) acc1: 67.7083 (77.3455) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.1803 data: 0.0902 max mem: 3531 +Test: [160/234] eta: 0:00:16 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.5480 (0.4856) acc1: 71.8750 (77.4537) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.1840 data: 0.0940 max mem: 3531 +Test: [170/234] eta: 0:00:14 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.5690 (0.4913) acc1: 70.0521 (77.0651) acc5: 100.0000 (100.0000) acc1_neg: 80.729 acc5_neg: 100.000 time: 0.1841 data: 0.0935 max mem: 3531 +Test: [180/234] eta: 0:00:11 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.6220 (0.5076) acc1: 67.7083 (76.1870) acc5: 100.0000 (100.0000) acc1_neg: 29.427 acc5_neg: 100.000 time: 0.2011 data: 0.1105 max mem: 3531 +Test: [190/234] eta: 0:00:09 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.7288 (0.5207) acc1: 60.9375 (75.3218) acc5: 100.0000 (100.0000) acc1_neg: 59.115 acc5_neg: 100.000 time: 0.1952 data: 0.1052 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.7121 (0.5120) acc1: 61.4583 (75.8525) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1782 data: 0.0869 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.1281 (0.4955) acc1: 99.7396 (76.8920) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1661 data: 0.0726 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.2302 (0.4850) acc1: 94.0104 (77.5582) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.1420 data: 0.0499 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.4628 (0.5042) acc1: 49.7396 (76.1364) acc5: 100.0000 (100.0000) acc1_neg: 47.656 acc5_neg: 100.000 time: 0.1322 data: 0.0421 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 97.931 acc5_pos: 100.000 loss: 0.9021 (0.5093) acc1: 46.6146 (75.7938) acc5: 100.0000 (100.0000) acc1_neg: 48.638 acc5_neg: 100.000 time: 0.1288 data: 0.0380 max mem: 3532 +Test: Total time: 0:00:48 (0.2069 s / it) +* Overall Acc@1 75.794 Acc@5 100.000 + target == real Acc@1 83.005 Acc@5 100.000 + target == fake Acc@1 73.761 Acc@5 100.000 + loss 0.509 + PR-AUC 0.9624 ROC-AUC 0.8740 EER 22.34% (thr=0.4377) +Accuracy of the network on the 89729 test images: 75.8% +[rank0]:[W725 17:48:59.628976750 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 17:49:16.740000 96659 torch/distributed/run.py:766] +W0725 17:49:16.740000 96659 torch/distributed/run.py:766] ***************************************** +W0725 17:49:16.740000 96659 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:49:16.740000 96659 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:49:43.732713449 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 17:49:44.297018395 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:49:44.547581011 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:49:44.761947249 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:49:44.794970103 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:49:44.940349803 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W725 17:49:44.944277989 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:49:44.019909181 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed2_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed2_resampled_n116-041-023_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:20:10 acc1_pos: 51.042 acc5_pos: 100.000 loss: 0.7346 (0.7346) acc1: 51.0417 (51.0417) acc5: 100.0000 (100.0000) time: 6.5397 data: 4.8698 max mem: 3531 +Test: [ 10/2387] eta: 0:26:48 acc1_pos: 28.385 acc5_pos: 100.000 loss: 0.7978 (0.7858) acc1: 48.4375 (48.2718) acc5: 100.0000 (100.0000) time: 0.6765 data: 0.4431 max mem: 3531 +Test: [ 20/2387] eta: 0:15:49 acc1_pos: 25.260 acc5_pos: 100.000 loss: 0.7978 (0.8049) acc1: 47.6562 (45.3497) acc5: 100.0000 (100.0000) time: 0.0941 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:13:27 acc1_pos: 36.458 acc5_pos: 100.000 loss: 0.8570 (0.8215) acc1: 36.7188 (43.7332) acc5: 100.0000 (100.0000) time: 0.1590 data: 0.0649 max mem: 3531 +Test: [ 40/2387] eta: 0:12:48 acc1_pos: 43.490 acc5_pos: 100.000 loss: 0.8394 (0.8266) acc1: 37.7604 (42.9306) acc5: 100.0000 (100.0000) time: 0.2499 data: 0.1547 max mem: 3531 +Test: [ 50/2387] eta: 0:11:46 acc1_pos: 44.792 acc5_pos: 100.000 loss: 0.7822 (0.8214) acc1: 43.4896 (43.5509) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1422 max mem: 3531 +Test: [ 60/2387] eta: 0:11:19 acc1_pos: 59.896 acc5_pos: 100.000 loss: 0.8246 (0.8239) acc1: 44.5312 (43.2761) acc5: 100.0000 (100.0000) time: 0.2199 data: 0.1251 max mem: 3531 +Test: [ 70/2387] eta: 0:10:59 acc1_pos: 38.542 acc5_pos: 100.000 loss: 0.8818 (0.8364) acc1: 38.2812 (42.2058) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1459 max mem: 3531 +Test: [ 80/2387] eta: 0:10:44 acc1_pos: 50.521 acc5_pos: 100.000 loss: 0.8469 (0.8331) acc1: 38.5417 (42.4126) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1451 max mem: 3531 +Test: [ 90/2387] eta: 0:10:28 acc1_pos: 51.042 acc5_pos: 100.000 loss: 0.7653 (0.8217) acc1: 47.1354 (43.0889) acc5: 100.0000 (100.0000) time: 0.2349 data: 0.1389 max mem: 3531 +Test: [ 100/2387] eta: 0:10:16 acc1_pos: 42.969 acc5_pos: 100.000 loss: 0.7653 (0.8239) acc1: 43.2292 (42.7599) acc5: 100.0000 (100.0000) time: 0.2302 data: 0.1356 max mem: 3531 +Test: [ 110/2387] eta: 0:10:10 acc1_pos: 34.375 acc5_pos: 100.000 loss: 0.7587 (0.8177) acc1: 43.2292 (43.2221) acc5: 100.0000 (100.0000) time: 0.2415 data: 0.1476 max mem: 3531 +Test: [ 120/2387] eta: 0:10:05 acc1_pos: 43.750 acc5_pos: 100.000 loss: 0.7587 (0.8155) acc1: 45.0521 (43.3992) acc5: 100.0000 (100.0000) time: 0.2546 data: 0.1616 max mem: 3531 +Test: [ 130/2387] eta: 0:09:59 acc1_pos: 44.531 acc5_pos: 100.000 loss: 0.7743 (0.8134) acc1: 46.3542 (43.5234) acc5: 100.0000 (100.0000) time: 0.2533 data: 0.1598 max mem: 3531 +Test: [ 140/2387] eta: 0:09:53 acc1_pos: 44.010 acc5_pos: 100.000 loss: 0.7776 (0.8108) acc1: 44.5312 (43.5893) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1461 max mem: 3531 +Test: [ 150/2387] eta: 0:09:46 acc1_pos: 37.240 acc5_pos: 100.000 loss: 0.7618 (0.8069) acc1: 44.5312 (43.7638) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1384 max mem: 3531 +Test: [ 160/2387] eta: 0:09:32 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.6960 (0.7949) acc1: 48.6979 (44.9874) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2106 data: 0.1141 max mem: 3531 +Test: [ 170/2387] eta: 0:09:17 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2936 (0.7652) acc1: 94.0104 (47.9045) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1705 data: 0.0786 max mem: 3531 +Test: [ 180/2387] eta: 0:09:12 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2854 (0.7388) acc1: 94.7917 (50.4877) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1942 data: 0.1020 max mem: 3531 +Test: [ 190/2387] eta: 0:09:03 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2808 (0.7149) acc1: 94.5312 (52.8032) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2106 data: 0.1180 max mem: 3531 +Test: [ 200/2387] eta: 0:08:53 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2800 (0.6934) acc1: 94.7917 (54.9129) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1849 data: 0.0891 max mem: 3531 +Test: [ 210/2387] eta: 0:08:44 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2852 (0.6743) acc1: 94.7917 (56.7819) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1787 data: 0.0833 max mem: 3532 +Test: [ 220/2387] eta: 0:08:45 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2886 (0.6568) acc1: 94.5312 (58.5101) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2300 data: 0.1385 max mem: 3532 +Test: [ 230/2387] eta: 0:08:31 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2915 (0.6409) acc1: 94.2708 (60.0616) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.2011 data: 0.1078 max mem: 3532 +Test: [ 240/2387] eta: 0:08:20 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2939 (0.6264) acc1: 95.3125 (61.5437) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1267 data: 0.0312 max mem: 3532 +Test: [ 250/2387] eta: 0:08:16 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2724 (0.6074) acc1: 96.0938 (63.0551) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1740 data: 0.0779 max mem: 3532 +Test: [ 260/2387] eta: 0:08:10 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1313 (0.5889) acc1: 100.0000 (64.4706) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2030 data: 0.1058 max mem: 3532 +Test: [ 270/2387] eta: 0:08:04 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1243 (0.5719) acc1: 100.0000 (65.7778) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1857 data: 0.0889 max mem: 3532 +Test: [ 280/2387] eta: 0:08:05 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1254 (0.5562) acc1: 100.0000 (66.9957) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2278 data: 0.1337 max mem: 3532 +Test: [ 290/2387] eta: 0:07:59 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1227 (0.5412) acc1: 100.0000 (68.1289) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2257 data: 0.1328 max mem: 3532 +Test: [ 300/2387] eta: 0:07:48 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1164 (0.5272) acc1: 100.0000 (69.1878) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1430 data: 0.0488 max mem: 3532 +Test: [ 310/2387] eta: 0:07:48 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1363 (0.5189) acc1: 100.0000 (70.0412) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1842 data: 0.0912 max mem: 3532 +Test: [ 320/2387] eta: 0:07:45 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.4323 (0.5164) acc1: 91.9271 (70.7060) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.2356 data: 0.1421 max mem: 3532 +Test: [ 330/2387] eta: 0:07:41 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.4385 (0.5140) acc1: 91.9271 (71.3479) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2038 data: 0.1086 max mem: 3532 +Test: [ 340/2387] eta: 0:07:39 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.4341 (0.5116) acc1: 91.9271 (71.9453) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.2095 data: 0.1136 max mem: 3532 +Test: [ 350/2387] eta: 0:07:40 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.4215 (0.5039) acc1: 93.4896 (72.7186) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2550 data: 0.1599 max mem: 3532 +Test: [ 360/2387] eta: 0:07:32 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2349 (0.4966) acc1: 98.1771 (73.4022) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2053 data: 0.1098 max mem: 3532 +Test: [ 370/2387] eta: 0:07:28 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2394 (0.4898) acc1: 97.3958 (74.0503) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1574 data: 0.0609 max mem: 3532 +Test: [ 380/2387] eta: 0:07:25 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2411 (0.4833) acc1: 97.3958 (74.6500) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1947 data: 0.1006 max mem: 3532 +Test: [ 390/2387] eta: 0:07:20 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2416 (0.4772) acc1: 97.1354 (75.2198) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1832 data: 0.0889 max mem: 3532 +Test: [ 400/2387] eta: 0:07:14 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2419 (0.4713) acc1: 97.1354 (75.7644) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1625 data: 0.0659 max mem: 3532 +Test: [ 410/2387] eta: 0:07:11 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2412 (0.4657) acc1: 97.3958 (76.2913) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1751 data: 0.0761 max mem: 3532 +Test: [ 420/2387] eta: 0:07:07 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2385 (0.4603) acc1: 97.3958 (76.7938) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1885 data: 0.0904 max mem: 3532 +Test: [ 430/2387] eta: 0:07:04 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2460 (0.4555) acc1: 97.1354 (77.2646) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1894 data: 0.0949 max mem: 3532 +Test: [ 440/2387] eta: 0:07:00 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2460 (0.4489) acc1: 97.3958 (77.7607) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1823 data: 0.0855 max mem: 3533 +Test: [ 450/2387] eta: 0:06:56 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1009 (0.4412) acc1: 100.0000 (78.2538) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1718 data: 0.0724 max mem: 3533 +Test: [ 460/2387] eta: 0:06:52 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1008 (0.4338) acc1: 100.0000 (78.7255) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1719 data: 0.0760 max mem: 3533 +Test: [ 470/2387] eta: 0:06:48 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1011 (0.4267) acc1: 100.0000 (79.1772) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1732 data: 0.0793 max mem: 3533 +Test: [ 480/2387] eta: 0:06:44 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1011 (0.4199) acc1: 100.0000 (79.6101) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1751 data: 0.0796 max mem: 3533 +Test: [ 490/2387] eta: 0:06:42 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1011 (0.4135) acc1: 100.0000 (80.0254) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1890 data: 0.0932 max mem: 3533 +Test: [ 500/2387] eta: 0:06:38 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1018 (0.4072) acc1: 100.0000 (80.4240) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1804 data: 0.0836 max mem: 3533 +Test: [ 510/2387] eta: 0:06:31 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1020 (0.4016) acc1: 100.0000 (80.8071) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1249 data: 0.0286 max mem: 3533 +Test: [ 520/2387] eta: 0:06:28 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1577 (0.3970) acc1: 100.0000 (81.1755) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1374 data: 0.0434 max mem: 3533 +Test: [ 530/2387] eta: 0:06:24 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1591 (0.3925) acc1: 100.0000 (81.5300) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0745 max mem: 3533 +Test: [ 540/2387] eta: 0:06:21 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1581 (0.3882) acc1: 100.0000 (81.8714) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1696 data: 0.0745 max mem: 3533 +Test: [ 550/2387] eta: 0:06:19 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1586 (0.3840) acc1: 100.0000 (82.2005) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1811 data: 0.0888 max mem: 3533 +Test: [ 560/2387] eta: 0:06:16 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1586 (0.3800) acc1: 100.0000 (82.5177) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1860 data: 0.0942 max mem: 3533 +Test: [ 570/2387] eta: 0:06:13 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1591 (0.3762) acc1: 100.0000 (82.8239) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1806 data: 0.0887 max mem: 3533 +Test: [ 580/2387] eta: 0:06:10 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1584 (0.3724) acc1: 100.0000 (83.1195) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1723 data: 0.0792 max mem: 3533 +Test: [ 590/2387] eta: 0:06:07 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1584 (0.3688) acc1: 100.0000 (83.4052) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1830 data: 0.0882 max mem: 3533 +Test: [ 600/2387] eta: 0:06:05 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1621 (0.3654) acc1: 100.0000 (83.6808) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1893 data: 0.0954 max mem: 3533 +Test: [ 610/2387] eta: 0:06:02 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1868 (0.3632) acc1: 99.2188 (83.9168) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1849 data: 0.0923 max mem: 3533 +Test: [ 620/2387] eta: 0:06:00 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2283 (0.3611) acc1: 98.1771 (84.1481) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1890 data: 0.0931 max mem: 3533 +Test: [ 630/2387] eta: 0:05:57 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2290 (0.3590) acc1: 98.1771 (84.3659) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1922 data: 0.0966 max mem: 3533 +Test: [ 640/2387] eta: 0:05:55 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2328 (0.3571) acc1: 97.9167 (84.5761) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1854 data: 0.0935 max mem: 3533 +Test: [ 650/2387] eta: 0:05:52 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2328 (0.3552) acc1: 97.9167 (84.7858) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1778 data: 0.0863 max mem: 3533 +Test: [ 660/2387] eta: 0:05:49 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2334 (0.3533) acc1: 98.4375 (84.9904) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1801 data: 0.0883 max mem: 3533 +Test: [ 670/2387] eta: 0:05:47 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2309 (0.3515) acc1: 98.4375 (85.1869) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1796 data: 0.0860 max mem: 3534 +Test: [ 680/2387] eta: 0:05:45 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2316 (0.3498) acc1: 98.6979 (85.3960) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1873 data: 0.0933 max mem: 3534 +Test: [ 690/2387] eta: 0:05:42 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2418 (0.3521) acc1: 98.1771 (85.2414) acc5: 100.0000 (100.0000) acc1_neg: 63.281 acc5_neg: 100.000 time: 0.1905 data: 0.0968 max mem: 3534 +Test: [ 700/2387] eta: 0:05:40 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.5963 (0.3557) acc1: 65.6250 (84.9605) acc5: 100.0000 (100.0000) acc1_neg: 70.312 acc5_neg: 100.000 time: 0.1948 data: 0.1022 max mem: 3534 +Test: [ 710/2387] eta: 0:05:37 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.6051 (0.3592) acc1: 64.8438 (84.6775) acc5: 100.0000 (100.0000) acc1_neg: 65.104 acc5_neg: 100.000 time: 0.1843 data: 0.0886 max mem: 3534 +Test: [ 720/2387] eta: 0:05:34 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.6013 (0.3626) acc1: 65.3646 (84.4169) acc5: 100.0000 (100.0000) acc1_neg: 64.062 acc5_neg: 100.000 time: 0.1653 data: 0.0693 max mem: 3534 +Test: [ 730/2387] eta: 0:05:32 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.6068 (0.3660) acc1: 65.8854 (84.1609) acc5: 100.0000 (100.0000) acc1_neg: 64.583 acc5_neg: 100.000 time: 0.1750 data: 0.0835 max mem: 3534 +Test: [ 740/2387] eta: 0:05:30 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.6057 (0.3693) acc1: 65.3646 (83.9107) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.1900 data: 0.0988 max mem: 3534 +Test: [ 750/2387] eta: 0:05:27 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.6022 (0.3723) acc1: 65.8854 (83.6732) acc5: 100.0000 (100.0000) acc1_neg: 68.490 acc5_neg: 100.000 time: 0.1865 data: 0.0950 max mem: 3534 +Test: [ 760/2387] eta: 0:05:25 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.5908 (0.3744) acc1: 65.8854 (83.5120) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.1745 data: 0.0825 max mem: 3534 +Test: [ 770/2387] eta: 0:05:22 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.4089 (0.3740) acc1: 81.2500 (83.5566) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1796 data: 0.0861 max mem: 3534 +Test: [ 780/2387] eta: 0:05:21 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2816 (0.3723) acc1: 91.4062 (83.7128) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1936 data: 0.0974 max mem: 3534 +Test: [ 790/2387] eta: 0:05:18 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2201 (0.3704) acc1: 97.1354 (83.8591) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1896 data: 0.0936 max mem: 3534 +Test: [ 800/2387] eta: 0:05:16 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2004 (0.3683) acc1: 97.9167 (84.0369) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1815 data: 0.0865 max mem: 3534 +Test: [ 810/2387] eta: 0:05:13 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2188 (0.3666) acc1: 97.9167 (84.1920) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1801 data: 0.0843 max mem: 3534 +Test: [ 820/2387] eta: 0:05:11 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2353 (0.3655) acc1: 95.5729 (84.2960) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1796 data: 0.0841 max mem: 3534 +Test: [ 830/2387] eta: 0:05:09 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.3039 (0.3654) acc1: 90.6250 (84.3327) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.1959 data: 0.1011 max mem: 3534 +Test: [ 840/2387] eta: 0:05:07 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.3435 (0.3650) acc1: 90.6250 (84.4351) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1849 data: 0.0888 max mem: 3534 +Test: [ 850/2387] eta: 0:05:04 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.3187 (0.3645) acc1: 92.9688 (84.5298) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1787 data: 0.0833 max mem: 3534 +Test: [ 860/2387] eta: 0:05:03 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.3122 (0.3637) acc1: 94.7917 (84.6590) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2054 data: 0.1140 max mem: 3534 +Test: [ 870/2387] eta: 0:05:01 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2890 (0.3629) acc1: 95.8333 (84.7673) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2069 data: 0.1144 max mem: 3534 +Test: [ 880/2387] eta: 0:04:59 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2939 (0.3624) acc1: 95.0521 (84.8621) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2005 data: 0.1061 max mem: 3534 +Test: [ 890/2387] eta: 0:04:58 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2771 (0.3613) acc1: 96.0938 (85.0087) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2218 data: 0.1281 max mem: 3534 +Test: [ 900/2387] eta: 0:04:55 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2917 (0.3649) acc1: 95.0521 (84.7539) acc5: 100.0000 (100.0000) acc1_neg: 54.948 acc5_neg: 100.000 time: 0.2088 data: 0.1140 max mem: 3535 +Test: [ 910/2387] eta: 0:04:53 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.7827 (0.3695) acc1: 53.9062 (84.4165) acc5: 100.0000 (100.0000) acc1_neg: 51.042 acc5_neg: 100.000 time: 0.1721 data: 0.0772 max mem: 3535 +Test: [ 920/2387] eta: 0:04:51 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.7920 (0.3742) acc1: 52.8646 (84.0722) acc5: 100.0000 (100.0000) acc1_neg: 52.344 acc5_neg: 100.000 time: 0.1949 data: 0.0982 max mem: 3535 +Test: [ 930/2387] eta: 0:04:48 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.7992 (0.3786) acc1: 53.9062 (83.7549) acc5: 100.0000 (100.0000) acc1_neg: 48.438 acc5_neg: 100.000 time: 0.1889 data: 0.0915 max mem: 3535 +Test: [ 940/2387] eta: 0:04:46 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.7992 (0.3830) acc1: 53.9062 (83.4332) acc5: 100.0000 (100.0000) acc1_neg: 53.385 acc5_neg: 100.000 time: 0.1601 data: 0.0668 max mem: 3535 +Test: [ 950/2387] eta: 0:04:45 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.7842 (0.3872) acc1: 53.1250 (83.1115) acc5: 100.0000 (100.0000) acc1_neg: 52.344 acc5_neg: 100.000 time: 0.1987 data: 0.1058 max mem: 3535 +Test: [ 960/2387] eta: 0:04:42 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.7777 (0.3913) acc1: 53.9062 (82.8177) acc5: 100.0000 (100.0000) acc1_neg: 55.729 acc5_neg: 100.000 time: 0.2049 data: 0.1109 max mem: 3535 +Test: [ 970/2387] eta: 0:04:40 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.7795 (0.3954) acc1: 53.9062 (82.5127) acc5: 100.0000 (100.0000) acc1_neg: 54.427 acc5_neg: 100.000 time: 0.1749 data: 0.0823 max mem: 3535 +Test: [ 980/2387] eta: 0:04:38 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.7544 (0.3966) acc1: 54.4271 (82.4523) acc5: 100.0000 (100.0000) acc1_neg: 70.312 acc5_neg: 100.000 time: 0.1847 data: 0.0924 max mem: 3535 +Test: [ 990/2387] eta: 0:04:36 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2954 (0.3949) acc1: 91.6667 (82.5605) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1849 data: 0.0918 max mem: 3535 +Test: [1000/2387] eta: 0:04:33 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1574 (0.3926) acc1: 99.4792 (82.7149) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1725 data: 0.0808 max mem: 3535 +Test: [1010/2387] eta: 0:04:32 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1683 (0.3908) acc1: 98.9583 (82.8509) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2021 data: 0.1101 max mem: 3535 +Test: [1020/2387] eta: 0:04:30 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2237 (0.3895) acc1: 96.8750 (82.9622) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2109 data: 0.1172 max mem: 3535 +Test: [1030/2387] eta: 0:04:28 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2158 (0.3877) acc1: 97.9167 (83.1070) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1892 data: 0.0959 max mem: 3535 +Test: [1040/2387] eta: 0:04:26 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1815 (0.3857) acc1: 98.9583 (83.2598) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1933 data: 0.1013 max mem: 3535 +Test: [1050/2387] eta: 0:04:23 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1737 (0.3837) acc1: 99.4792 (83.4119) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1934 data: 0.1022 max mem: 3535 +Test: [1060/2387] eta: 0:04:22 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1737 (0.3818) acc1: 99.4792 (83.5601) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2143 data: 0.1225 max mem: 3535 +Test: [1070/2387] eta: 0:04:20 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1841 (0.3803) acc1: 99.2188 (83.6730) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.2049 data: 0.1102 max mem: 3535 +Test: [1080/2387] eta: 0:04:17 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2589 (0.3795) acc1: 95.0521 (83.7602) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1684 data: 0.0706 max mem: 3535 +Test: [1090/2387] eta: 0:04:15 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2459 (0.3781) acc1: 95.5729 (83.8778) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1825 data: 0.0866 max mem: 3535 +Test: [1100/2387] eta: 0:04:13 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2039 (0.3766) acc1: 96.6146 (83.9942) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.1014 max mem: 3535 +Test: [1110/2387] eta: 0:04:11 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2073 (0.3754) acc1: 97.3958 (84.0963) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1848 data: 0.0908 max mem: 3535 +Test: [1120/2387] eta: 0:04:10 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2105 (0.3739) acc1: 97.3958 (84.2094) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.2108 data: 0.1172 max mem: 3536 +Test: [1130/2387] eta: 0:04:07 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2061 (0.3726) acc1: 96.8750 (84.3197) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2084 data: 0.1160 max mem: 3536 +Test: [1140/2387] eta: 0:04:05 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2281 (0.3714) acc1: 96.3542 (84.4282) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1806 data: 0.0887 max mem: 3536 +Test: [1150/2387] eta: 0:04:03 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2423 (0.3706) acc1: 97.9167 (84.5273) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1796 data: 0.0886 max mem: 3536 +Test: [1160/2387] eta: 0:04:01 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2423 (0.3693) acc1: 97.9167 (84.6231) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1682 data: 0.0763 max mem: 3536 +Test: [1170/2387] eta: 0:03:59 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2156 (0.3681) acc1: 96.6146 (84.7302) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1742 data: 0.0812 max mem: 3536 +Test: [1180/2387] eta: 0:03:57 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2303 (0.3669) acc1: 96.6146 (84.8279) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1816 data: 0.0890 max mem: 3536 +Test: [1190/2387] eta: 0:03:54 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2177 (0.3657) acc1: 97.1354 (84.9295) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1775 data: 0.0860 max mem: 3536 +Test: [1200/2387] eta: 0:03:53 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1900 (0.3642) acc1: 98.6979 (85.0465) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1855 data: 0.0952 max mem: 3536 +Test: [1210/2387] eta: 0:03:50 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1900 (0.3629) acc1: 99.4792 (85.1646) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1743 data: 0.0830 max mem: 3536 +Test: [1220/2387] eta: 0:03:48 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1537 (0.3610) acc1: 99.7396 (85.2821) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1675 data: 0.0750 max mem: 3536 +Test: [1230/2387] eta: 0:03:46 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1386 (0.3592) acc1: 99.7396 (85.3997) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1809 data: 0.0880 max mem: 3536 +Test: [1240/2387] eta: 0:03:44 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1361 (0.3574) acc1: 100.0000 (85.5151) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0832 max mem: 3536 +Test: [1250/2387] eta: 0:03:42 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1314 (0.3556) acc1: 100.0000 (85.6294) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1769 data: 0.0843 max mem: 3536 +Test: [1260/2387] eta: 0:03:40 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1273 (0.3538) acc1: 100.0000 (85.7417) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1899 data: 0.0965 max mem: 3536 +Test: [1270/2387] eta: 0:03:38 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1252 (0.3520) acc1: 100.0000 (85.8531) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1878 data: 0.0937 max mem: 3536 +Test: [1280/2387] eta: 0:03:36 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1331 (0.3505) acc1: 100.0000 (85.9558) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1822 data: 0.0893 max mem: 3536 +Test: [1290/2387] eta: 0:03:34 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1522 (0.3490) acc1: 98.9583 (86.0511) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1991 data: 0.1085 max mem: 3536 +Test: [1300/2387] eta: 0:03:32 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1567 (0.3475) acc1: 98.6979 (86.1457) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1976 data: 0.1036 max mem: 3536 +Test: [1310/2387] eta: 0:03:30 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1567 (0.3461) acc1: 98.6979 (86.2374) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1895 data: 0.0946 max mem: 3536 +Test: [1320/2387] eta: 0:03:28 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1635 (0.3448) acc1: 98.6979 (86.3272) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1880 data: 0.0955 max mem: 3536 +Test: [1330/2387] eta: 0:03:26 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1811 (0.3438) acc1: 98.1771 (86.4120) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1997 data: 0.1072 max mem: 3536 +Test: [1340/2387] eta: 0:03:24 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2382 (0.3439) acc1: 95.8333 (86.4139) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0929 max mem: 3536 +Test: [1350/2387] eta: 0:03:22 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.0859 (0.3419) acc1: 100.0000 (86.5144) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1635 data: 0.0720 max mem: 3537 +Test: [1360/2387] eta: 0:03:20 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.0768 (0.3399) acc1: 100.0000 (86.6135) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1831 data: 0.0913 max mem: 3537 +Test: [1370/2387] eta: 0:03:18 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.0728 (0.3380) acc1: 100.0000 (86.7112) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1882 data: 0.0962 max mem: 3537 +Test: [1380/2387] eta: 0:03:16 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.0739 (0.3361) acc1: 100.0000 (86.8074) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1898 data: 0.0985 max mem: 3537 +Test: [1390/2387] eta: 0:03:14 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.0752 (0.3342) acc1: 100.0000 (86.9022) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1936 data: 0.1014 max mem: 3537 +Test: [1400/2387] eta: 0:03:12 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.0764 (0.3324) acc1: 100.0000 (86.9957) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1886 data: 0.0940 max mem: 3537 +Test: [1410/2387] eta: 0:03:10 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.0795 (0.3319) acc1: 100.0000 (87.0294) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1881 data: 0.0934 max mem: 3537 +Test: [1420/2387] eta: 0:03:08 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2861 (0.3317) acc1: 90.3646 (87.0492) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.2028 data: 0.1095 max mem: 3537 +Test: [1430/2387] eta: 0:03:06 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2883 (0.3316) acc1: 90.3646 (87.0713) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2011 data: 0.1095 max mem: 3537 +Test: [1440/2387] eta: 0:03:04 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2913 (0.3314) acc1: 91.4062 (87.1003) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1732 data: 0.0828 max mem: 3537 +Test: [1450/2387] eta: 0:03:02 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2913 (0.3313) acc1: 92.7083 (87.1210) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1669 data: 0.0769 max mem: 3537 +Test: [1460/2387] eta: 0:03:00 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2241 (0.3306) acc1: 94.0104 (87.1763) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1854 data: 0.0939 max mem: 3537 +Test: [1470/2387] eta: 0:02:58 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1947 (0.3297) acc1: 97.6562 (87.2518) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1944 data: 0.1026 max mem: 3537 +Test: [1480/2387] eta: 0:02:56 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1780 (0.3288) acc1: 98.6979 (87.3199) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1883 data: 0.0976 max mem: 3537 +Test: [1490/2387] eta: 0:02:53 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1814 (0.3279) acc1: 98.4375 (87.3915) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1693 data: 0.0787 max mem: 3537 +Test: [1500/2387] eta: 0:02:51 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1862 (0.3270) acc1: 97.9167 (87.4599) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1655 data: 0.0732 max mem: 3537 +Test: [1510/2387] eta: 0:02:49 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1879 (0.3261) acc1: 99.4792 (87.5390) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1797 data: 0.0871 max mem: 3537 +Test: [1520/2387] eta: 0:02:47 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2126 (0.3258) acc1: 97.9167 (87.5846) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1705 data: 0.0778 max mem: 3537 +Test: [1530/2387] eta: 0:02:45 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2873 (0.3257) acc1: 94.2708 (87.6225) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1633 data: 0.0704 max mem: 3537 +Test: [1540/2387] eta: 0:02:43 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2825 (0.3253) acc1: 96.3542 (87.6813) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1655 data: 0.0739 max mem: 3537 +Test: [1550/2387] eta: 0:02:41 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2693 (0.3250) acc1: 97.6562 (87.7327) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1643 data: 0.0725 max mem: 3537 +Test: [1560/2387] eta: 0:02:39 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2693 (0.3248) acc1: 96.3542 (87.7774) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1781 data: 0.0847 max mem: 3537 +Test: [1570/2387] eta: 0:02:37 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2522 (0.3242) acc1: 97.3958 (87.8418) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2074 data: 0.1148 max mem: 3537 +Test: [1580/2387] eta: 0:02:35 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2442 (0.3238) acc1: 97.1354 (87.8805) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.2022 data: 0.1119 max mem: 3538 +Test: [1590/2387] eta: 0:02:33 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2398 (0.3232) acc1: 95.3125 (87.9336) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1627 data: 0.0722 max mem: 3538 +Test: [1600/2387] eta: 0:02:31 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2255 (0.3226) acc1: 97.1354 (87.9811) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1757 data: 0.0842 max mem: 3538 +Test: [1610/2387] eta: 0:02:29 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1997 (0.3219) acc1: 97.3958 (88.0415) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1845 data: 0.0925 max mem: 3538 +Test: [1620/2387] eta: 0:02:27 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2229 (0.3214) acc1: 95.3125 (88.0729) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1647 data: 0.0728 max mem: 3538 +Test: [1630/2387] eta: 0:02:25 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2558 (0.3212) acc1: 92.9688 (88.0959) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1860 data: 0.0929 max mem: 3538 +Test: [1640/2387] eta: 0:02:23 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.3272 (0.3215) acc1: 90.1042 (88.0905) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1738 data: 0.0796 max mem: 3538 +Test: [1650/2387] eta: 0:02:21 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.3470 (0.3215) acc1: 89.0625 (88.1036) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1195 data: 0.0254 max mem: 3538 +Test: [1660/2387] eta: 0:02:19 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.3366 (0.3217) acc1: 89.0625 (88.0994) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1241 data: 0.0314 max mem: 3538 +Test: [1670/2387] eta: 0:02:17 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.3372 (0.3218) acc1: 88.5417 (88.1000) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1793 data: 0.0885 max mem: 3538 +Test: [1680/2387] eta: 0:02:15 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.3430 (0.3220) acc1: 87.5000 (88.0890) acc5: 100.0000 (100.0000) acc1_neg: 79.427 acc5_neg: 100.000 time: 0.2010 data: 0.1102 max mem: 3538 +Test: [1690/2387] eta: 0:02:13 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.3582 (0.3223) acc1: 87.5000 (88.0834) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1816 data: 0.0906 max mem: 3538 +Test: [1700/2387] eta: 0:02:11 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.3396 (0.3220) acc1: 88.8021 (88.1083) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1622 data: 0.0687 max mem: 3538 +Test: [1710/2387] eta: 0:02:09 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1672 (0.3210) acc1: 98.4375 (88.1724) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1601 data: 0.0648 max mem: 3538 +Test: [1720/2387] eta: 0:02:07 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1534 (0.3201) acc1: 99.2188 (88.2378) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1727 data: 0.0791 max mem: 3538 +Test: [1730/2387] eta: 0:02:05 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1517 (0.3191) acc1: 99.4792 (88.2989) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1673 data: 0.0739 max mem: 3538 +Test: [1740/2387] eta: 0:02:03 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1442 (0.3181) acc1: 99.7396 (88.3625) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1673 data: 0.0752 max mem: 3538 +Test: [1750/2387] eta: 0:02:01 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1485 (0.3172) acc1: 99.4792 (88.4246) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1749 data: 0.0843 max mem: 3538 +Test: [1760/2387] eta: 0:01:59 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1469 (0.3162) acc1: 99.7396 (88.4884) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1647 data: 0.0731 max mem: 3538 +Test: [1770/2387] eta: 0:01:57 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1600 (0.3155) acc1: 98.9583 (88.5423) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1560 data: 0.0618 max mem: 3538 +Test: [1780/2387] eta: 0:01:55 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1966 (0.3149) acc1: 98.1771 (88.5952) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1573 data: 0.0638 max mem: 3538 +Test: [1790/2387] eta: 0:01:53 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1947 (0.3141) acc1: 98.6979 (88.6522) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1744 data: 0.0830 max mem: 3538 +Test: [1800/2387] eta: 0:01:51 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1779 (0.3133) acc1: 98.9583 (88.7082) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1682 data: 0.0769 max mem: 3538 +Test: [1810/2387] eta: 0:01:49 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1691 (0.3125) acc1: 98.9583 (88.7657) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1661 data: 0.0747 max mem: 3539 +Test: [1820/2387] eta: 0:01:47 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1539 (0.3116) acc1: 99.4792 (88.8253) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1805 data: 0.0888 max mem: 3539 +Test: [1830/2387] eta: 0:01:45 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1494 (0.3108) acc1: 99.7396 (88.8840) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1944 data: 0.1014 max mem: 3539 +Test: [1840/2387] eta: 0:01:43 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1418 (0.3099) acc1: 100.0000 (88.9444) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1852 data: 0.0912 max mem: 3539 +Test: [1850/2387] eta: 0:01:41 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1411 (0.3089) acc1: 100.0000 (89.0040) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1441 data: 0.0493 max mem: 3539 +Test: [1860/2387] eta: 0:01:39 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1402 (0.3080) acc1: 100.0000 (89.0631) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1655 data: 0.0713 max mem: 3539 +Test: [1870/2387] eta: 0:01:38 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1402 (0.3071) acc1: 100.0000 (89.1212) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2095 data: 0.1153 max mem: 3539 +Test: [1880/2387] eta: 0:01:36 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1393 (0.3062) acc1: 100.0000 (89.1788) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1852 data: 0.0902 max mem: 3539 +Test: [1890/2387] eta: 0:01:34 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1393 (0.3054) acc1: 100.0000 (89.2360) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0906 max mem: 3539 +Test: [1900/2387] eta: 0:01:32 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1408 (0.3045) acc1: 100.0000 (89.2926) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1982 data: 0.1059 max mem: 3539 +Test: [1910/2387] eta: 0:01:30 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1585 (0.3038) acc1: 100.0000 (89.3455) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2049 data: 0.1125 max mem: 3539 +Test: [1920/2387] eta: 0:01:28 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1815 (0.3040) acc1: 98.9583 (89.3384) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.2090 data: 0.1158 max mem: 3539 +Test: [1930/2387] eta: 0:01:26 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2188 (0.3035) acc1: 97.3958 (89.3839) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1819 data: 0.0889 max mem: 3539 +Test: [1940/2387] eta: 0:01:24 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2188 (0.3033) acc1: 97.9167 (89.3940) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1904 data: 0.0976 max mem: 3539 +Test: [1950/2387] eta: 0:01:22 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2334 (0.3030) acc1: 97.1354 (89.4221) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1947 data: 0.1027 max mem: 3539 +Test: [1960/2387] eta: 0:01:20 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2518 (0.3028) acc1: 95.0521 (89.4420) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1796 data: 0.0857 max mem: 3539 +Test: [1970/2387] eta: 0:01:19 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2973 (0.3032) acc1: 90.8854 (89.4097) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.1831 data: 0.0900 max mem: 3539 +Test: [1980/2387] eta: 0:01:17 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2979 (0.3030) acc1: 90.8854 (89.4306) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1849 data: 0.0935 max mem: 3539 +Test: [1990/2387] eta: 0:01:15 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1701 (0.3023) acc1: 97.9167 (89.4733) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1998 data: 0.1067 max mem: 3539 +Test: [2000/2387] eta: 0:01:13 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1701 (0.3016) acc1: 98.1771 (89.5168) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1746 data: 0.0798 max mem: 3539 +Test: [2010/2387] eta: 0:01:11 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1709 (0.3010) acc1: 98.4375 (89.5602) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1762 data: 0.0828 max mem: 3539 +Test: [2020/2387] eta: 0:01:09 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1733 (0.3003) acc1: 98.1771 (89.6028) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2083 data: 0.1167 max mem: 3539 +Test: [2030/2387] eta: 0:01:07 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1687 (0.2997) acc1: 98.1771 (89.6437) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2010 data: 0.1093 max mem: 3540 +Test: [2040/2387] eta: 0:01:05 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1687 (0.2991) acc1: 97.9167 (89.6855) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1980 data: 0.1042 max mem: 3540 +Test: [2050/2387] eta: 0:01:03 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1683 (0.2984) acc1: 98.4375 (89.7286) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1698 data: 0.0759 max mem: 3540 +Test: [2060/2387] eta: 0:01:01 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1695 (0.2978) acc1: 98.1771 (89.7691) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1712 data: 0.0795 max mem: 3540 +Test: [2070/2387] eta: 0:01:00 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1663 (0.2971) acc1: 98.4375 (89.8126) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1901 data: 0.0979 max mem: 3540 +Test: [2080/2387] eta: 0:00:58 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1623 (0.2965) acc1: 98.6979 (89.8554) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1992 data: 0.1074 max mem: 3540 +Test: [2090/2387] eta: 0:00:56 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1632 (0.2959) acc1: 98.6979 (89.8979) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2164 data: 0.1242 max mem: 3540 +Test: [2100/2387] eta: 0:00:54 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1640 (0.2952) acc1: 98.6979 (89.9397) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1850 data: 0.0925 max mem: 3540 +Test: [2110/2387] eta: 0:00:52 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1640 (0.2946) acc1: 98.6979 (89.9814) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1568 data: 0.0642 max mem: 3540 +Test: [2120/2387] eta: 0:00:50 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1755 (0.2944) acc1: 98.1771 (90.0053) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1826 data: 0.0865 max mem: 3540 +Test: [2130/2387] eta: 0:00:48 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1904 (0.2939) acc1: 98.1771 (90.0471) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1739 data: 0.0739 max mem: 3540 +Test: [2140/2387] eta: 0:00:46 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1839 (0.2935) acc1: 98.4375 (90.0719) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1968 data: 0.0999 max mem: 3540 +Test: [2150/2387] eta: 0:00:44 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2096 (0.2931) acc1: 97.3958 (90.1083) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2349 data: 0.1432 max mem: 3540 +Test: [2160/2387] eta: 0:00:43 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2129 (0.2927) acc1: 97.6562 (90.1421) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1787 data: 0.0851 max mem: 3540 +Test: [2170/2387] eta: 0:00:41 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2336 (0.2927) acc1: 96.8750 (90.1592) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1450 data: 0.0512 max mem: 3540 +Test: [2180/2387] eta: 0:00:39 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.3131 (0.2938) acc1: 90.6250 (90.0934) acc5: 100.0000 (100.0000) acc1_neg: 55.729 acc5_neg: 100.000 time: 0.1691 data: 0.0777 max mem: 3540 +Test: [2190/2387] eta: 0:00:37 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2922 (0.2939) acc1: 89.5833 (90.0852) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1848 data: 0.0923 max mem: 3540 +Test: [2200/2387] eta: 0:00:35 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2169 (0.2935) acc1: 94.5312 (90.1152) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2102 data: 0.1181 max mem: 3540 +Test: [2210/2387] eta: 0:00:33 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2013 (0.2931) acc1: 97.6562 (90.1481) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1763 data: 0.0843 max mem: 3540 +Test: [2220/2387] eta: 0:00:31 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2013 (0.2927) acc1: 97.9167 (90.1811) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1598 data: 0.0663 max mem: 3540 +Test: [2230/2387] eta: 0:00:29 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1986 (0.2923) acc1: 97.9167 (90.2141) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2035 data: 0.1096 max mem: 3540 +Test: [2240/2387] eta: 0:00:27 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1902 (0.2918) acc1: 97.6562 (90.2432) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2004 data: 0.1071 max mem: 3540 +Test: [2250/2387] eta: 0:00:25 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1939 (0.2915) acc1: 98.1771 (90.2761) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1968 data: 0.1048 max mem: 3540 +Test: [2260/2387] eta: 0:00:24 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2071 (0.2912) acc1: 98.4375 (90.3129) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2053 data: 0.1132 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1824 (0.2906) acc1: 99.2188 (90.3539) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1980 data: 0.1037 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1664 (0.2902) acc1: 99.7396 (90.3811) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1888 data: 0.0924 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1528 (0.2896) acc1: 99.7396 (90.4218) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2049 data: 0.1098 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1643 (0.2891) acc1: 99.7396 (90.4563) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1899 data: 0.0961 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.1794 (0.2888) acc1: 98.6979 (90.4839) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1815 data: 0.0875 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.2251 (0.2889) acc1: 95.3125 (90.4913) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1924 data: 0.1001 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.3460 (0.2893) acc1: 90.6250 (90.4741) acc5: 100.0000 (100.0000) acc1_neg: 72.135 acc5_neg: 100.000 time: 0.1764 data: 0.0850 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.3222 (0.2893) acc1: 90.6250 (90.4769) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1556 data: 0.0645 max mem: 3541 +Test: [2350/2387] eta: 0:00:06 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.3010 (0.2896) acc1: 92.1875 (90.4635) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1396 data: 0.0485 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.3306 (0.2899) acc1: 89.5833 (90.4518) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1298 data: 0.0385 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.3306 (0.2903) acc1: 84.3750 (90.4233) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.1176 data: 0.0274 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.3896 (0.2908) acc1: 82.2917 (90.3966) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.1132 data: 0.0232 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 50.685 acc5_pos: 100.000 loss: 0.4050 (0.2907) acc1: 82.0312 (90.3983) acc5: 100.0000 (100.0000) acc1_neg: 95.349 acc5_neg: 100.000 time: 0.1128 data: 0.0240 max mem: 3541 +Test: Total time: 0:07:28 (0.1880 s / it) +* Overall Acc@1 90.398 Acc@5 100.000 + target == real Acc@1 43.853 Acc@5 100.000 + target == fake Acc@1 93.667 Acc@5 100.000 + loss 0.291 + PR-AUC 0.9920 ROC-AUC 0.8951 EER 17.77% (thr=0.6938) +Accuracy of the network on the 916267 test images: 90.4% +[rank0]:[W725 17:57:38.091033602 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 17:57:59.160000 97281 torch/distributed/run.py:766] +W0725 17:57:59.160000 97281 torch/distributed/run.py:766] ***************************************** +W0725 17:57:59.160000 97281 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:57:59.160000 97281 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:58:34.529355433 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +[rank2]:[W725 17:58:34.886919516 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 17:58:34.887077412 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:58:34.903147953 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:58:35.135646842 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:58:35.297954224 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:58:35.424931856 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:58:35.439811878 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed2_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed2_resampled_n116-041-023_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 0:56:31 acc1_pos: 51.042 acc5_pos: 100.000 loss: 0.7346 (0.7346) acc1: 51.0417 (51.0417) acc5: 100.0000 (100.0000) time: 6.5596 data: 4.2712 max mem: 3531 +Test: [ 10/517] eta: 0:05:43 acc1_pos: 28.385 acc5_pos: 100.000 loss: 0.7978 (0.7858) acc1: 48.4375 (48.2718) acc5: 100.0000 (100.0000) time: 0.6784 data: 0.3886 max mem: 3531 +Test: [ 20/517] eta: 0:03:19 acc1_pos: 25.260 acc5_pos: 100.000 loss: 0.7978 (0.8049) acc1: 47.6562 (45.3497) acc5: 100.0000 (100.0000) time: 0.0928 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:02:54 acc1_pos: 36.458 acc5_pos: 100.000 loss: 0.8570 (0.8215) acc1: 36.7188 (43.7332) acc5: 100.0000 (100.0000) time: 0.1826 data: 0.0893 max mem: 3531 +Test: [ 40/517] eta: 0:02:42 acc1_pos: 43.490 acc5_pos: 100.000 loss: 0.8394 (0.8266) acc1: 37.7604 (42.9306) acc5: 100.0000 (100.0000) time: 0.2766 data: 0.1848 max mem: 3531 +Test: [ 50/517] eta: 0:02:27 acc1_pos: 77.064 acc5_pos: 100.000 loss: 0.7794 (0.8108) acc1: 43.4896 (44.7100) acc5: 100.0000 (100.0000) acc1_neg: 70.312 acc5_neg: 100.000 time: 0.2499 data: 0.1559 max mem: 3531 +Test: [ 60/517] eta: 0:02:20 acc1_pos: 77.064 acc5_pos: 100.000 loss: 0.6095 (0.7718) acc1: 65.6250 (48.9498) acc5: 100.0000 (100.0000) acc1_neg: 70.833 acc5_neg: 100.000 time: 0.2382 data: 0.1420 max mem: 3531 +Test: [ 70/517] eta: 0:02:13 acc1_pos: 77.064 acc5_pos: 100.000 loss: 0.5897 (0.7447) acc1: 68.4896 (51.8743) acc5: 100.0000 (100.0000) acc1_neg: 73.958 acc5_neg: 100.000 time: 0.2549 data: 0.1611 max mem: 3531 +Test: [ 80/517] eta: 0:02:12 acc1_pos: 77.064 acc5_pos: 100.000 loss: 0.6051 (0.7268) acc1: 66.4062 (53.9223) acc5: 100.0000 (100.0000) acc1_neg: 58.333 acc5_neg: 100.000 time: 0.2917 data: 0.2009 max mem: 3531 +Test: [ 90/517] eta: 0:02:08 acc1_pos: 77.064 acc5_pos: 100.000 loss: 0.5639 (0.7042) acc1: 71.3542 (56.4961) acc5: 100.0000 (100.0000) acc1_neg: 71.354 acc5_neg: 100.000 time: 0.3049 data: 0.2137 max mem: 3531 +Test: [100/517] eta: 0:02:03 acc1_pos: 77.064 acc5_pos: 100.000 loss: 0.4800 (0.6792) acc1: 80.9896 (59.3724) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.2681 data: 0.1738 max mem: 3531 +Test: [110/517] eta: 0:01:57 acc1_pos: 77.064 acc5_pos: 100.000 loss: 0.4479 (0.6570) acc1: 85.1562 (61.8501) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2308 data: 0.1362 max mem: 3531 +Test: [120/517] eta: 0:01:52 acc1_pos: 77.064 acc5_pos: 100.000 loss: 0.4806 (0.6443) acc1: 79.9479 (63.3501) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.2220 data: 0.1276 max mem: 3531 +Test: [130/517] eta: 0:01:48 acc1_pos: 77.064 acc5_pos: 100.000 loss: 0.4806 (0.6296) acc1: 81.5104 (65.0843) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.2453 data: 0.1498 max mem: 3531 +Test: [140/517] eta: 0:01:45 acc1_pos: 77.064 acc5_pos: 100.000 loss: 0.4287 (0.6155) acc1: 88.0208 (66.6962) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.2552 data: 0.1584 max mem: 3531 +Test: [150/517] eta: 0:01:42 acc1_pos: 77.064 acc5_pos: 100.000 loss: 0.4101 (0.6015) acc1: 88.8021 (68.1930) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.2652 data: 0.1666 max mem: 3531 +Test: [160/517] eta: 0:01:38 acc1_pos: 77.064 acc5_pos: 100.000 loss: 0.4162 (0.5906) acc1: 88.8021 (69.2126) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.2588 data: 0.1613 max mem: 3531 +Test: [170/517] eta: 0:01:34 acc1_pos: 77.064 acc5_pos: 100.000 loss: 0.3948 (0.5786) acc1: 86.9792 (70.3293) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.2191 data: 0.1226 max mem: 3531 +Test: [180/517] eta: 0:01:30 acc1_pos: 77.064 acc5_pos: 100.000 loss: 0.3833 (0.5676) acc1: 88.0208 (71.3887) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.2108 data: 0.1139 max mem: 3531 +Test: [190/517] eta: 0:01:27 acc1_pos: 77.064 acc5_pos: 100.000 loss: 0.3296 (0.5541) acc1: 91.9271 (72.7708) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2391 data: 0.1439 max mem: 3531 +Test: [200/517] eta: 0:01:24 acc1_pos: 77.064 acc5_pos: 100.000 loss: 0.3201 (0.5440) acc1: 97.3958 (73.9713) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2426 data: 0.1491 max mem: 3531 +Test: [210/517] eta: 0:01:22 acc1_pos: 77.064 acc5_pos: 100.000 loss: 0.4090 (0.5455) acc1: 88.5417 (73.9830) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.2549 data: 0.1599 max mem: 3532 +Test: [220/517] eta: 0:01:19 acc1_pos: 77.064 acc5_pos: 100.000 loss: 0.5407 (0.5446) acc1: 78.3854 (74.2694) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.2699 data: 0.1742 max mem: 3532 +Test: [230/517] eta: 0:01:16 acc1_pos: 77.064 acc5_pos: 100.000 loss: 0.5149 (0.5432) acc1: 79.9479 (74.5491) acc5: 100.0000 (100.0000) acc1_neg: 67.708 acc5_neg: 100.000 time: 0.2599 data: 0.1624 max mem: 3532 +Test: [240/517] eta: 0:01:13 acc1_pos: 77.064 acc5_pos: 100.000 loss: 0.5295 (0.5456) acc1: 71.3542 (74.3322) acc5: 100.0000 (100.0000) acc1_neg: 55.208 acc5_neg: 100.000 time: 0.2449 data: 0.1465 max mem: 3532 +Test: [250/517] eta: 0:01:10 acc1_pos: 77.064 acc5_pos: 100.000 loss: 0.5173 (0.5428) acc1: 73.6979 (74.8817) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2549 data: 0.1565 max mem: 3532 +Test: [260/517] eta: 0:01:08 acc1_pos: 77.064 acc5_pos: 100.000 loss: 0.4187 (0.5380) acc1: 95.5729 (75.6436) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.2549 data: 0.1564 max mem: 3532 +Test: [270/517] eta: 0:01:05 acc1_pos: 77.064 acc5_pos: 100.000 loss: 0.4685 (0.5375) acc1: 86.9792 (75.7880) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.2499 data: 0.1513 max mem: 3532 +Test: [280/517] eta: 0:01:02 acc1_pos: 77.064 acc5_pos: 100.000 loss: 0.5037 (0.5354) acc1: 82.8125 (76.0973) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.2599 data: 0.1603 max mem: 3532 +Test: [290/517] eta: 0:00:59 acc1_pos: 77.064 acc5_pos: 100.000 loss: 0.4401 (0.5316) acc1: 88.5417 (76.6448) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2599 data: 0.1603 max mem: 3532 +Test: [300/517] eta: 0:00:57 acc1_pos: 77.064 acc5_pos: 100.000 loss: 0.4200 (0.5284) acc1: 90.6250 (77.0323) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.2449 data: 0.1484 max mem: 3532 +Test: [310/517] eta: 0:00:54 acc1_pos: 77.064 acc5_pos: 100.000 loss: 0.4593 (0.5273) acc1: 85.1562 (77.2106) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.2449 data: 0.1484 max mem: 3532 +Test: [320/517] eta: 0:00:51 acc1_pos: 77.064 acc5_pos: 100.000 loss: 0.4567 (0.5235) acc1: 88.5417 (77.6618) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2549 data: 0.1583 max mem: 3532 +Test: [330/517] eta: 0:00:49 acc1_pos: 77.064 acc5_pos: 100.000 loss: 0.3579 (0.5180) acc1: 94.5312 (78.2241) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2599 data: 0.1645 max mem: 3532 +Test: [340/517] eta: 0:00:46 acc1_pos: 77.064 acc5_pos: 100.000 loss: 0.3353 (0.5159) acc1: 96.3542 (78.4297) acc5: 100.0000 (100.0000) acc1_neg: 64.062 acc5_neg: 100.000 time: 0.2599 data: 0.1622 max mem: 3532 +Test: [350/517] eta: 0:00:43 acc1_pos: 77.064 acc5_pos: 100.000 loss: 0.5298 (0.5176) acc1: 79.6875 (78.2645) acc5: 100.0000 (100.0000) acc1_neg: 64.062 acc5_neg: 100.000 time: 0.2499 data: 0.1520 max mem: 3532 +Test: [360/517] eta: 0:00:41 acc1_pos: 77.064 acc5_pos: 100.000 loss: 0.5117 (0.5172) acc1: 79.6875 (78.3277) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.2499 data: 0.1514 max mem: 3532 +Test: [370/517] eta: 0:00:38 acc1_pos: 77.064 acc5_pos: 100.000 loss: 0.4493 (0.5136) acc1: 87.5000 (78.7181) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2499 data: 0.1513 max mem: 3532 +Test: [380/517] eta: 0:00:35 acc1_pos: 77.064 acc5_pos: 100.000 loss: 0.3993 (0.5111) acc1: 90.6250 (78.8721) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.2449 data: 0.1466 max mem: 3532 +Test: [390/517] eta: 0:00:33 acc1_pos: 77.064 acc5_pos: 100.000 loss: 0.4297 (0.5095) acc1: 82.5521 (78.9642) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.2449 data: 0.1464 max mem: 3532 +Test: [400/517] eta: 0:00:30 acc1_pos: 77.064 acc5_pos: 100.000 loss: 0.4188 (0.5069) acc1: 83.0729 (79.1387) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.3299 data: 0.2349 max mem: 3532 +Test: [410/517] eta: 0:00:28 acc1_pos: 77.064 acc5_pos: 100.000 loss: 0.4169 (0.5071) acc1: 76.8229 (79.0780) acc5: 100.0000 (100.0000) acc1_neg: 68.490 acc5_neg: 100.000 time: 0.2849 data: 0.1896 max mem: 3532 +Test: [420/517] eta: 0:00:25 acc1_pos: 77.064 acc5_pos: 100.000 loss: 0.4898 (0.5065) acc1: 80.7292 (79.1716) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.2299 data: 0.1324 max mem: 3532 +Test: [430/517] eta: 0:00:22 acc1_pos: 77.064 acc5_pos: 100.000 loss: 0.4691 (0.5053) acc1: 82.2917 (79.2658) acc5: 100.0000 (100.0000) acc1_neg: 72.396 acc5_neg: 100.000 time: 0.2499 data: 0.1528 max mem: 3532 +Test: [440/517] eta: 0:00:20 acc1_pos: 77.064 acc5_pos: 100.000 loss: 0.4772 (0.5050) acc1: 80.9896 (79.2830) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.2199 data: 0.1225 max mem: 3533 +Test: [450/517] eta: 0:00:17 acc1_pos: 77.064 acc5_pos: 100.000 loss: 0.4772 (0.5044) acc1: 82.0312 (79.4011) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.2549 data: 0.1601 max mem: 3533 +Test: [460/517] eta: 0:00:14 acc1_pos: 77.064 acc5_pos: 100.000 loss: 0.4695 (0.5040) acc1: 82.0312 (79.4587) acc5: 100.0000 (100.0000) acc1_neg: 80.729 acc5_neg: 100.000 time: 0.2549 data: 0.1601 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 77.064 acc5_pos: 100.000 loss: 0.4533 (0.5025) acc1: 85.4167 (79.6466) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.2649 data: 0.1653 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 77.064 acc5_pos: 100.000 loss: 0.4378 (0.5020) acc1: 87.5000 (79.6626) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2749 data: 0.1747 max mem: 3533 +Test: [490/517] eta: 0:00:07 acc1_pos: 77.064 acc5_pos: 100.000 loss: 0.3849 (0.4997) acc1: 90.3646 (79.9129) acc5: 100.0000 (100.0000) acc1_neg: 77.083 acc5_neg: 100.000 time: 0.2499 data: 0.1542 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 77.064 acc5_pos: 100.000 loss: 0.5117 (0.5016) acc1: 77.0833 (79.7135) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.2375 data: 0.1438 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 77.064 acc5_pos: 100.000 loss: 0.5657 (0.5020) acc1: 72.9167 (79.6946) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2086 data: 0.1166 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 77.064 acc5_pos: 100.000 loss: 0.4562 (0.5011) acc1: 87.7604 (79.8490) acc5: 100.0000 (100.0000) acc1_neg: 96.466 acc5_neg: 100.000 time: 0.1846 data: 0.0929 max mem: 3533 +Test: Total time: 0:02:13 (0.2589 s / it) +* Overall Acc@1 79.849 Acc@5 100.000 + target == real Acc@1 43.416 Acc@5 100.000 + target == fake Acc@1 83.629 Acc@5 100.000 + loss 0.501 + PR-AUC 0.9484 ROC-AUC 0.6957 EER 35.95% (thr=0.6069) +Accuracy of the network on the 198427 test images: 79.8% +[rank0]:[W725 18:01:03.399281804 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0725 18:01:25.059000 97850 torch/distributed/run.py:766] +W0725 18:01:25.059000 97850 torch/distributed/run.py:766] ***************************************** +W0725 18:01:25.059000 97850 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:01:25.059000 97850 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:02:01.195223108 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:02:01.729809895 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:02:02.057890830 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:02:02.239037932 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:02:02.272603146 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:02:02.341462198 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:02:02.383183324 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +NCCL version 2.26.2+cuda12.2 +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:02:02.497548446 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed2_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed2_resampled_n116-041-023_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:32:35 acc1_pos: 60.417 acc5_pos: 100.000 loss: 0.6663 (0.6663) acc1: 60.4167 (60.4167) acc5: 100.0000 (100.0000) time: 5.0280 data: 2.9896 max mem: 3531 +Test: [ 10/389] eta: 0:03:24 acc1_pos: 93.229 acc5_pos: 100.000 loss: 0.6461 (0.6112) acc1: 60.4167 (64.2756) acc5: 100.0000 (100.0000) time: 0.5390 data: 0.2721 max mem: 3531 +Test: [ 20/389] eta: 0:02:06 acc1_pos: 22.135 acc5_pos: 100.000 loss: 0.6461 (0.6686) acc1: 55.2083 (57.7257) acc5: 100.0000 (100.0000) time: 0.1080 data: 0.0160 max mem: 3531 +Test: [ 30/389] eta: 0:01:49 acc1_pos: 60.677 acc5_pos: 100.000 loss: 0.6607 (0.6472) acc1: 62.2396 (61.3995) acc5: 100.0000 (100.0000) time: 0.1779 data: 0.0834 max mem: 3531 +Test: [ 40/389] eta: 0:01:41 acc1_pos: 11.458 acc5_pos: 100.000 loss: 0.6104 (0.6695) acc1: 61.9792 (57.8379) acc5: 100.0000 (100.0000) time: 0.2349 data: 0.1389 max mem: 3531 +Test: [ 50/389] eta: 0:01:32 acc1_pos: 82.812 acc5_pos: 100.000 loss: 0.6065 (0.6507) acc1: 66.1458 (60.4116) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1280 max mem: 3531 +Test: [ 60/389] eta: 0:01:24 acc1_pos: 60.156 acc5_pos: 100.000 loss: 0.4988 (0.6343) acc1: 75.5208 (62.6238) acc5: 100.0000 (100.0000) time: 0.1849 data: 0.0902 max mem: 3531 +Test: [ 70/389] eta: 0:01:18 acc1_pos: 73.958 acc5_pos: 100.000 loss: 0.5691 (0.6270) acc1: 68.7500 (63.5197) acc5: 100.0000 (100.0000) time: 0.1722 data: 0.0784 max mem: 3531 +Test: [ 80/389] eta: 0:01:14 acc1_pos: 91.406 acc5_pos: 100.000 loss: 0.5959 (0.6366) acc1: 62.5000 (62.0788) acc5: 100.0000 (100.0000) time: 0.2025 data: 0.1090 max mem: 3531 +Test: [ 90/389] eta: 0:01:09 acc1_pos: 67.188 acc5_pos: 100.000 loss: 0.5358 (0.6213) acc1: 67.1875 (63.3270) acc5: 100.0000 (100.0000) time: 0.1926 data: 0.0985 max mem: 3531 +Test: [100/389] eta: 0:01:05 acc1_pos: 84.375 acc5_pos: 100.000 loss: 0.5473 (0.6292) acc1: 67.1875 (62.8094) acc5: 100.0000 (100.0000) time: 0.1701 data: 0.0746 max mem: 3531 +Test: [110/389] eta: 0:01:03 acc1_pos: 99.740 acc5_pos: 100.000 loss: 0.6219 (0.6367) acc1: 65.1042 (62.4179) acc5: 100.0000 (100.0000) time: 0.1910 data: 0.0972 max mem: 3531 +Test: [120/389] eta: 0:00:59 acc1_pos: 76.302 acc5_pos: 100.000 loss: 0.5268 (0.6254) acc1: 76.3021 (63.6148) acc5: 100.0000 (100.0000) time: 0.1952 data: 0.1037 max mem: 3531 +Test: [130/389] eta: 0:00:56 acc1_pos: 50.307 acc5_pos: 100.000 loss: 0.5543 (0.6322) acc1: 67.9688 (63.1361) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1756 data: 0.0828 max mem: 3531 +Test: [140/389] eta: 0:00:53 acc1_pos: 50.307 acc5_pos: 100.000 loss: 0.6443 (0.6343) acc1: 60.9375 (62.8417) acc5: 100.0000 (100.0000) acc1_neg: 29.427 acc5_neg: 100.000 time: 0.1629 data: 0.0704 max mem: 3531 +Test: [150/389] eta: 0:00:50 acc1_pos: 50.307 acc5_pos: 100.000 loss: 0.6341 (0.6435) acc1: 61.4583 (62.2724) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1731 data: 0.0797 max mem: 3531 +Test: [160/389] eta: 0:00:48 acc1_pos: 50.307 acc5_pos: 100.000 loss: 0.5309 (0.6396) acc1: 62.7604 (62.9400) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1875 data: 0.0922 max mem: 3531 +Test: [170/389] eta: 0:00:45 acc1_pos: 50.307 acc5_pos: 100.000 loss: 0.3246 (0.6294) acc1: 95.8333 (63.9894) acc5: 100.0000 (100.0000) acc1_neg: 53.385 acc5_neg: 100.000 time: 0.1778 data: 0.0855 max mem: 3531 +Test: [180/389] eta: 0:00:43 acc1_pos: 50.307 acc5_pos: 100.000 loss: 0.6869 (0.6384) acc1: 53.3854 (63.0021) acc5: 100.0000 (100.0000) acc1_neg: 28.125 acc5_neg: 100.000 time: 0.1840 data: 0.0927 max mem: 3531 +Test: [190/389] eta: 0:00:41 acc1_pos: 50.307 acc5_pos: 100.000 loss: 0.7384 (0.6422) acc1: 47.6562 (62.8177) acc5: 100.0000 (100.0000) acc1_neg: 13.281 acc5_neg: 100.000 time: 0.1993 data: 0.1072 max mem: 3531 +Test: [200/389] eta: 0:00:38 acc1_pos: 50.307 acc5_pos: 100.000 loss: 0.7361 (0.6515) acc1: 46.8750 (61.6669) acc5: 100.0000 (100.0000) acc1_neg: 20.052 acc5_neg: 100.000 time: 0.1838 data: 0.0896 max mem: 3531 +Test: [210/389] eta: 0:00:36 acc1_pos: 50.307 acc5_pos: 100.000 loss: 0.5609 (0.6475) acc1: 33.8542 (62.0248) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1852 data: 0.0914 max mem: 3532 +Test: [220/389] eta: 0:00:34 acc1_pos: 50.307 acc5_pos: 100.000 loss: 0.3534 (0.6373) acc1: 96.3542 (63.2094) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.1882 data: 0.0969 max mem: 3532 +Test: [230/389] eta: 0:00:32 acc1_pos: 50.307 acc5_pos: 100.000 loss: 0.5465 (0.6404) acc1: 72.1354 (62.7514) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.1788 data: 0.0868 max mem: 3532 +Test: [240/389] eta: 0:00:30 acc1_pos: 50.307 acc5_pos: 100.000 loss: 0.6423 (0.6414) acc1: 51.8229 (62.6621) acc5: 100.0000 (100.0000) acc1_neg: 57.812 acc5_neg: 100.000 time: 0.1959 data: 0.1037 max mem: 3532 +Test: [250/389] eta: 0:00:27 acc1_pos: 50.307 acc5_pos: 100.000 loss: 0.5014 (0.6326) acc1: 78.3854 (63.6610) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1803 data: 0.0885 max mem: 3532 +Test: [260/389] eta: 0:00:25 acc1_pos: 50.307 acc5_pos: 100.000 loss: 0.4277 (0.6325) acc1: 87.2396 (63.8470) acc5: 100.0000 (100.0000) acc1_neg: 16.406 acc5_neg: 100.000 time: 0.1719 data: 0.0802 max mem: 3532 +Test: [270/389] eta: 0:00:23 acc1_pos: 50.307 acc5_pos: 100.000 loss: 0.6523 (0.6431) acc1: 57.5521 (62.9689) acc5: 100.0000 (100.0000) acc1_neg: 71.615 acc5_neg: 100.000 time: 0.1781 data: 0.0870 max mem: 3532 +Test: [280/389] eta: 0:00:21 acc1_pos: 50.307 acc5_pos: 100.000 loss: 0.7304 (0.6464) acc1: 53.9062 (62.6668) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1580 data: 0.0662 max mem: 3532 +Test: [290/389] eta: 0:00:19 acc1_pos: 50.307 acc5_pos: 100.000 loss: 0.6636 (0.6483) acc1: 54.1667 (62.4016) acc5: 100.0000 (100.0000) acc1_neg: 16.927 acc5_neg: 100.000 time: 0.1749 data: 0.0800 max mem: 3532 +Test: [300/389] eta: 0:00:17 acc1_pos: 50.307 acc5_pos: 100.000 loss: 0.6483 (0.6502) acc1: 55.4688 (62.1894) acc5: 100.0000 (100.0000) acc1_neg: 55.469 acc5_neg: 100.000 time: 0.1919 data: 0.0972 max mem: 3532 +Test: [310/389] eta: 0:00:15 acc1_pos: 50.307 acc5_pos: 100.000 loss: 0.7608 (0.6605) acc1: 45.5729 (61.2934) acc5: 100.0000 (100.0000) acc1_neg: 5.729 acc5_neg: 100.000 time: 0.1806 data: 0.0891 max mem: 3532 +Test: [320/389] eta: 0:00:13 acc1_pos: 50.307 acc5_pos: 100.000 loss: 0.8336 (0.6581) acc1: 45.5729 (61.5776) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.1798 data: 0.0876 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 50.307 acc5_pos: 100.000 loss: 0.8938 (0.6664) acc1: 31.7708 (60.4025) acc5: 100.0000 (100.0000) acc1_neg: 41.927 acc5_neg: 100.000 time: 0.1804 data: 0.0883 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 50.307 acc5_pos: 100.000 loss: 0.5754 (0.6566) acc1: 64.8438 (61.4308) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1747 data: 0.0843 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 50.307 acc5_pos: 100.000 loss: 0.3312 (0.6652) acc1: 96.3542 (60.9019) acc5: 100.0000 (100.0000) acc1_neg: 7.031 acc5_neg: 100.000 time: 0.1515 data: 0.0615 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 50.307 acc5_pos: 100.000 loss: 0.7033 (0.6694) acc1: 50.2604 (60.4859) acc5: 100.0000 (100.0000) acc1_neg: 57.552 acc5_neg: 100.000 time: 0.1406 data: 0.0500 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 50.307 acc5_pos: 100.000 loss: 0.6864 (0.6703) acc1: 57.5521 (60.4237) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1480 data: 0.0573 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 50.307 acc5_pos: 100.000 loss: 0.5868 (0.6687) acc1: 65.3646 (60.5602) acc5: 100.0000 (100.0000) acc1_neg: 54.948 acc5_neg: 100.000 time: 0.1379 data: 0.0482 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 50.307 acc5_pos: 100.000 loss: 0.5868 (0.6677) acc1: 65.6250 (60.6804) acc5: 100.0000 (100.0000) acc1_neg: 66.393 acc5_neg: 100.000 time: 0.1285 data: 0.0349 max mem: 3532 +Test: Total time: 0:01:13 (0.1883 s / it) +* Overall Acc@1 60.680 Acc@5 100.000 + target == real Acc@1 63.469 Acc@5 100.000 + target == fake Acc@1 59.384 Acc@5 100.000 + loss 0.668 + PR-AUC 0.8063 ROC-AUC 0.6670 EER 38.72% (thr=0.4861) +Accuracy of the network on the 149358 test images: 60.7% +[rank0]:[W725 18:03:23.801620951 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:03:45.227000 98407 torch/distributed/run.py:766] +W0725 18:03:45.227000 98407 torch/distributed/run.py:766] ***************************************** +W0725 18:03:45.227000 98407 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:03:45.227000 98407 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:04:19.698867585 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:04:19.844914223 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:04:20.196756820 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +[rank7]:[W725 18:04:20.217454170 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:04:20.220739474 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +| distributed init (rank 5): env:// +NCCL version 2.26.2+cuda12.2 +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:04:20.292757094 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +[rank5]:[W725 18:04:20.298695904 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:04:20.306699787 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed2_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed2_resampled_n116-041-023_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:37:14 acc1_pos: 87.240 acc5_pos: 100.000 loss: 0.3114 (0.3114) acc1: 87.2396 (87.2396) acc5: 100.0000 (100.0000) time: 6.7097 data: 4.9039 max mem: 3531 +Test: [ 10/333] eta: 0:03:45 acc1_pos: 91.927 acc5_pos: 100.000 loss: 0.2287 (0.2421) acc1: 91.9271 (91.3352) acc5: 100.0000 (100.0000) time: 0.6993 data: 0.4463 max mem: 3531 +Test: [ 20/333] eta: 0:02:09 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.2270 (0.2396) acc1: 91.9271 (92.0511) acc5: 100.0000 (100.0000) time: 0.0991 data: 0.0005 max mem: 3531 +Test: [ 30/333] eta: 0:01:57 acc1_pos: 63.802 acc5_pos: 100.000 loss: 0.3008 (0.3662) acc1: 82.0312 (82.1993) acc5: 100.0000 (100.0000) time: 0.2149 data: 0.1163 max mem: 3531 +Test: [ 40/333] eta: 0:01:41 acc1_pos: 63.542 acc5_pos: 100.000 loss: 0.6552 (0.4380) acc1: 59.3750 (76.6451) acc5: 100.0000 (100.0000) time: 0.2749 data: 0.1796 max mem: 3531 +Test: [ 50/333] eta: 0:01:29 acc1_pos: 61.719 acc5_pos: 100.000 loss: 0.6458 (0.4795) acc1: 61.4583 (73.6009) acc5: 100.0000 (100.0000) time: 0.2049 data: 0.1088 max mem: 3531 +Test: [ 60/333] eta: 0:01:21 acc1_pos: 58.594 acc5_pos: 100.000 loss: 0.6291 (0.5087) acc1: 61.4583 (71.4737) acc5: 100.0000 (100.0000) time: 0.1999 data: 0.1013 max mem: 3531 +Test: [ 70/333] eta: 0:01:12 acc1_pos: 60.156 acc5_pos: 100.000 loss: 0.6374 (0.5283) acc1: 60.9375 (70.4189) acc5: 100.0000 (100.0000) time: 0.1806 data: 0.0867 max mem: 3531 +Test: [ 80/333] eta: 0:01:08 acc1_pos: 62.240 acc5_pos: 100.000 loss: 0.6509 (0.5432) acc1: 58.5938 (69.0329) acc5: 100.0000 (100.0000) time: 0.1799 data: 0.0892 max mem: 3531 +Test: [ 90/333] eta: 0:01:04 acc1_pos: 71.354 acc5_pos: 100.000 loss: 0.6053 (0.5445) acc1: 62.2396 (69.2909) acc5: 100.0000 (100.0000) time: 0.2142 data: 0.1214 max mem: 3531 +Test: [100/333] eta: 0:01:00 acc1_pos: 76.042 acc5_pos: 100.000 loss: 0.5670 (0.5482) acc1: 70.3125 (69.0568) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1293 max mem: 3531 +Test: [110/333] eta: 0:00:56 acc1_pos: 72.656 acc5_pos: 100.000 loss: 0.5688 (0.5509) acc1: 67.9688 (68.9870) acc5: 100.0000 (100.0000) time: 0.1999 data: 0.1036 max mem: 3531 +Test: [120/333] eta: 0:00:52 acc1_pos: 77.604 acc5_pos: 100.000 loss: 0.5447 (0.5515) acc1: 69.7917 (69.0384) acc5: 100.0000 (100.0000) time: 0.1849 data: 0.0916 max mem: 3531 +Test: [130/333] eta: 0:00:49 acc1_pos: 86.979 acc5_pos: 100.000 loss: 0.5347 (0.5510) acc1: 70.3125 (69.2828) acc5: 100.0000 (100.0000) time: 0.1953 data: 0.1044 max mem: 3531 +Test: [140/333] eta: 0:00:46 acc1_pos: 81.510 acc5_pos: 100.000 loss: 0.5096 (0.5500) acc1: 73.6979 (69.5220) acc5: 100.0000 (100.0000) time: 0.1899 data: 0.0988 max mem: 3531 +Test: [150/333] eta: 0:00:43 acc1_pos: 79.688 acc5_pos: 100.000 loss: 0.5432 (0.5520) acc1: 73.6979 (69.4657) acc5: 100.0000 (100.0000) time: 0.1845 data: 0.0899 max mem: 3531 +Test: [160/333] eta: 0:00:40 acc1_pos: 60.197 acc5_pos: 100.000 loss: 0.5856 (0.5669) acc1: 66.4062 (68.5899) acc5: 100.0000 (100.0000) acc1_neg: 35.156 acc5_neg: 100.000 time: 0.1799 data: 0.0853 max mem: 3531 +Test: [170/333] eta: 0:00:37 acc1_pos: 60.197 acc5_pos: 100.000 loss: 0.9566 (0.5924) acc1: 38.8021 (66.7717) acc5: 100.0000 (100.0000) acc1_neg: 42.448 acc5_neg: 100.000 time: 0.1899 data: 0.0983 max mem: 3531 +Test: [180/333] eta: 0:00:34 acc1_pos: 60.197 acc5_pos: 100.000 loss: 0.9069 (0.6065) acc1: 42.4479 (65.6739) acc5: 100.0000 (100.0000) acc1_neg: 42.969 acc5_neg: 100.000 time: 0.1972 data: 0.1060 max mem: 3531 +Test: [190/333] eta: 0:00:32 acc1_pos: 60.197 acc5_pos: 100.000 loss: 0.8319 (0.6162) acc1: 47.6562 (64.9965) acc5: 100.0000 (100.0000) acc1_neg: 47.135 acc5_neg: 100.000 time: 0.1999 data: 0.1083 max mem: 3531 +Test: [200/333] eta: 0:00:30 acc1_pos: 60.197 acc5_pos: 100.000 loss: 0.8304 (0.6282) acc1: 47.9167 (64.2465) acc5: 100.0000 (100.0000) acc1_neg: 45.312 acc5_neg: 100.000 time: 0.2025 data: 0.1093 max mem: 3531 +Test: [210/333] eta: 0:00:27 acc1_pos: 60.197 acc5_pos: 100.000 loss: 0.9341 (0.6436) acc1: 44.5312 (63.2936) acc5: 100.0000 (100.0000) acc1_neg: 49.740 acc5_neg: 100.000 time: 0.2013 data: 0.1084 max mem: 3532 +Test: [220/333] eta: 0:00:25 acc1_pos: 60.197 acc5_pos: 100.000 loss: 0.9442 (0.6555) acc1: 43.4896 (62.4340) acc5: 100.0000 (100.0000) acc1_neg: 53.125 acc5_neg: 100.000 time: 0.2108 data: 0.1199 max mem: 3532 +Test: [230/333] eta: 0:00:22 acc1_pos: 60.197 acc5_pos: 100.000 loss: 0.7202 (0.6568) acc1: 55.9896 (62.3670) acc5: 100.0000 (100.0000) acc1_neg: 58.854 acc5_neg: 100.000 time: 0.2085 data: 0.1168 max mem: 3532 +Test: [240/333] eta: 0:00:20 acc1_pos: 60.197 acc5_pos: 100.000 loss: 0.7132 (0.6629) acc1: 57.0312 (61.7220) acc5: 100.0000 (100.0000) acc1_neg: 46.875 acc5_neg: 100.000 time: 0.1782 data: 0.0859 max mem: 3532 +Test: [250/333] eta: 0:00:18 acc1_pos: 60.197 acc5_pos: 100.000 loss: 0.7966 (0.6680) acc1: 47.6562 (61.2913) acc5: 100.0000 (100.0000) acc1_neg: 46.615 acc5_neg: 100.000 time: 0.1889 data: 0.0955 max mem: 3532 +Test: [260/333] eta: 0:00:15 acc1_pos: 60.197 acc5_pos: 100.000 loss: 0.7782 (0.6717) acc1: 52.3438 (61.0363) acc5: 100.0000 (100.0000) acc1_neg: 59.115 acc5_neg: 100.000 time: 0.1997 data: 0.1001 max mem: 3532 +Test: [270/333] eta: 0:00:13 acc1_pos: 60.197 acc5_pos: 100.000 loss: 0.7430 (0.6743) acc1: 56.2500 (60.8923) acc5: 100.0000 (100.0000) acc1_neg: 56.510 acc5_neg: 100.000 time: 0.1899 data: 0.0892 max mem: 3532 +Test: [280/333] eta: 0:00:11 acc1_pos: 60.197 acc5_pos: 100.000 loss: 0.7524 (0.6820) acc1: 53.9062 (60.2758) acc5: 100.0000 (100.0000) acc1_neg: 45.052 acc5_neg: 100.000 time: 0.2007 data: 0.1064 max mem: 3532 +Test: [290/333] eta: 0:00:09 acc1_pos: 60.197 acc5_pos: 100.000 loss: 0.9828 (0.6925) acc1: 35.9375 (59.4278) acc5: 100.0000 (100.0000) acc1_neg: 43.490 acc5_neg: 100.000 time: 0.1849 data: 0.0938 max mem: 3532 +Test: [300/333] eta: 0:00:07 acc1_pos: 60.197 acc5_pos: 100.000 loss: 0.9372 (0.6999) acc1: 42.7083 (58.9450) acc5: 100.0000 (100.0000) acc1_neg: 46.094 acc5_neg: 100.000 time: 0.1571 data: 0.0657 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 60.197 acc5_pos: 100.000 loss: 0.8341 (0.7011) acc1: 48.6979 (58.8684) acc5: 100.0000 (100.0000) acc1_neg: 57.292 acc5_neg: 100.000 time: 0.1272 data: 0.0358 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 60.197 acc5_pos: 100.000 loss: 0.7084 (0.7023) acc1: 57.0312 (58.7390) acc5: 100.0000 (100.0000) acc1_neg: 41.406 acc5_neg: 100.000 time: 0.1213 data: 0.0307 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 60.197 acc5_pos: 100.000 loss: 0.9366 (0.7119) acc1: 43.2292 (58.1587) acc5: 100.0000 (100.0000) acc1_neg: 42.448 acc5_neg: 100.000 time: 0.1198 data: 0.0299 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 60.197 acc5_pos: 100.000 loss: 0.9895 (0.7137) acc1: 41.4062 (58.0571) acc5: 100.0000 (100.0000) acc1_neg: 39.718 acc5_neg: 100.000 time: 0.1281 data: 0.0266 max mem: 3532 +Test: Total time: 0:01:08 (0.2053 s / it) +* Overall Acc@1 58.057 Acc@5 100.000 + target == real Acc@1 69.627 Acc@5 100.000 + target == fake Acc@1 47.758 Acc@5 100.000 + loss 0.714 + PR-AUC 0.6449 ROC-AUC 0.6325 EER 40.74% (thr=0.4275) +Accuracy of the network on the 127843 test images: 58.1% +[rank0]:[W725 18:05:36.664159290 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:05:58.165000 98965 torch/distributed/run.py:766] +W0725 18:05:58.165000 98965 torch/distributed/run.py:766] ***************************************** +W0725 18:05:58.165000 98965 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:05:58.165000 98965 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:06:34.611376406 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:06:34.029913229 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank4]:[W725 18:06:35.051243527 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:06:35.460896182 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:06:35.591092844 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:06:35.592237263 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +[rank0]:[W725 18:06:35.619656470 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:06:35.627115687 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed2_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed2_resampled_n116-041-023_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:21:45 acc1_pos: 96.615 acc5_pos: 100.000 loss: 0.1797 (0.1797) acc1: 96.6146 (96.6146) acc5: 100.0000 (100.0000) time: 5.5566 data: 3.7652 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/235] eta: 0:02:12 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.1808 (0.1814) acc1: 96.6146 (96.1884) acc5: 100.0000 (100.0000) time: 0.5869 data: 0.3426 max mem: 3531 +Test: [ 20/235] eta: 0:01:15 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.1784 (0.1787) acc1: 96.3542 (96.2922) acc5: 100.0000 (100.0000) time: 0.0932 data: 0.0035 max mem: 3531 +Test: [ 30/235] eta: 0:01:10 acc1_pos: 96.354 acc5_pos: 100.000 loss: 0.1736 (0.1766) acc1: 96.6146 (96.2954) acc5: 100.0000 (100.0000) time: 0.2080 data: 0.1131 max mem: 3531 +Test: [ 40/235] eta: 0:01:05 acc1_pos: 95.833 acc5_pos: 100.000 loss: 0.1724 (0.1760) acc1: 96.6146 (96.3478) acc5: 100.0000 (100.0000) time: 0.3152 data: 0.2192 max mem: 3531 +Test: [ 50/235] eta: 0:00:58 acc1_pos: 95.833 acc5_pos: 100.000 loss: 0.1724 (0.1765) acc1: 96.3542 (96.3082) acc5: 100.0000 (100.0000) time: 0.2749 data: 0.1811 max mem: 3531 +Test: [ 60/235] eta: 0:00:50 acc1_pos: 96.094 acc5_pos: 100.000 loss: 0.1780 (0.1775) acc1: 96.0938 (96.2176) acc5: 100.0000 (100.0000) time: 0.1860 data: 0.0896 max mem: 3531 +Test: [ 70/235] eta: 0:00:46 acc1_pos: 96.923 acc5_pos: 100.000 loss: 0.1948 (0.3633) acc1: 94.0104 (86.3556) acc5: 100.0000 (100.0000) acc1_neg: 15.104 acc5_neg: 100.000 time: 0.1867 data: 0.0930 max mem: 3531 +Test: [ 80/235] eta: 0:00:42 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.6638 (0.5262) acc1: 16.9271 (77.7167) acc5: 100.0000 (100.0000) acc1_neg: 15.104 acc5_neg: 100.000 time: 0.2334 data: 0.1423 max mem: 3531 +Test: [ 90/235] eta: 0:00:38 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.6822 (0.6534) acc1: 16.6667 (70.9965) acc5: 100.0000 (100.0000) acc1_neg: 15.885 acc5_neg: 100.000 time: 0.2188 data: 0.1277 max mem: 3531 +Test: [100/235] eta: 0:00:35 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.6842 (0.7534) acc1: 16.9271 (65.7101) acc5: 100.0000 (100.0000) acc1_neg: 14.583 acc5_neg: 100.000 time: 0.2249 data: 0.1321 max mem: 3531 +Test: [110/235] eta: 0:00:31 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.6520 (0.8350) acc1: 17.1875 (61.3504) acc5: 100.0000 (100.0000) acc1_neg: 19.792 acc5_neg: 100.000 time: 0.2092 data: 0.1165 max mem: 3531 +Test: [120/235] eta: 0:00:29 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.7190 (0.9278) acc1: 12.2396 (56.8569) acc5: 100.0000 (100.0000) acc1_neg: 5.469 acc5_neg: 100.000 time: 0.2007 data: 0.1051 max mem: 3531 +Test: [130/235] eta: 0:00:26 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.9747 (1.0079) acc1: 6.5104 (53.0216) acc5: 100.0000 (100.0000) acc1_neg: 5.469 acc5_neg: 100.000 time: 0.2209 data: 0.1222 max mem: 3531 +Test: [140/235] eta: 0:00:23 acc1_pos: 96.923 acc5_pos: 100.000 loss: 2.0000 (1.0792) acc1: 6.2500 (49.7230) acc5: 100.0000 (100.0000) acc1_neg: 6.250 acc5_neg: 100.000 time: 0.2146 data: 0.1189 max mem: 3531 +Test: [150/235] eta: 0:00:20 acc1_pos: 96.923 acc5_pos: 100.000 loss: 2.0057 (1.1386) acc1: 6.2500 (46.8664) acc5: 100.0000 (100.0000) acc1_neg: 7.292 acc5_neg: 100.000 time: 0.2142 data: 0.1211 max mem: 3531 +Test: [160/235] eta: 0:00:18 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.9723 (1.1904) acc1: 6.7708 (44.3970) acc5: 100.0000 (100.0000) acc1_neg: 9.115 acc5_neg: 100.000 time: 0.2245 data: 0.1314 max mem: 3531 +Test: [170/235] eta: 0:00:15 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.9054 (1.2077) acc1: 8.8542 (43.1210) acc5: 100.0000 (100.0000) acc1_neg: 20.312 acc5_neg: 100.000 time: 0.2096 data: 0.1171 max mem: 3531 +Test: [180/235] eta: 0:00:13 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.4323 (1.2194) acc1: 23.9583 (42.1098) acc5: 100.0000 (100.0000) acc1_neg: 23.698 acc5_neg: 100.000 time: 0.2056 data: 0.1150 max mem: 3531 +Test: [190/235] eta: 0:00:10 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.3902 (1.2296) acc1: 24.2188 (41.1908) acc5: 100.0000 (100.0000) acc1_neg: 23.698 acc5_neg: 100.000 time: 0.1999 data: 0.1062 max mem: 3531 +Test: [200/235] eta: 0:00:08 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.4163 (1.2398) acc1: 24.2188 (40.3413) acc5: 100.0000 (100.0000) acc1_neg: 23.438 acc5_neg: 100.000 time: 0.1879 data: 0.0936 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.4233 (1.2477) acc1: 23.9583 (39.5957) acc5: 100.0000 (100.0000) acc1_neg: 26.562 acc5_neg: 100.000 time: 0.1727 data: 0.0814 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.4042 (1.2556) acc1: 24.7396 (38.9270) acc5: 100.0000 (100.0000) acc1_neg: 25.781 acc5_neg: 100.000 time: 0.1291 data: 0.0388 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.4129 (1.2621) acc1: 25.0000 (38.3613) acc5: 100.0000 (100.0000) acc1_neg: 25.000 acc5_neg: 100.000 time: 0.1227 data: 0.0322 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.4219 (1.2661) acc1: 25.5208 (38.1680) acc5: 100.0000 (100.0000) acc1_neg: 30.435 acc5_neg: 100.000 time: 0.1251 data: 0.0376 max mem: 3532 +Test: Total time: 0:00:51 (0.2203 s / it) +* Overall Acc@1 38.168 Acc@5 100.000 + target == real Acc@1 96.217 Acc@5 100.000 + target == fake Acc@1 17.173 Acc@5 100.000 + loss 1.266 + PR-AUC 0.8582 ROC-AUC 0.6946 EER 36.00% (thr=0.1302) +Accuracy of the network on the 89879 test images: 38.2% +[rank0]:[W725 18:07:36.725555296 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:07:58.838000 99524 torch/distributed/run.py:766] +W0725 18:07:58.838000 99524 torch/distributed/run.py:766] ***************************************** +W0725 18:07:58.838000 99524 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:07:58.838000 99524 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:08:35.194400064 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:08:35.494968402 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:08:35.902201214 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:08:36.081663906 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:08:36.299349417 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:08:36.317806385 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:08:36.362782721 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:08:36.715738549 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed2_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n45_classseed2_resampled_n116-041-023_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:27:47 acc1_pos: 70.312 acc5_pos: 100.000 loss: 0.5888 (0.5888) acc1: 70.3125 (70.3125) acc5: 100.0000 (100.0000) time: 6.1999 data: 4.5397 max mem: 3531 +Test: [ 10/269] eta: 0:02:48 acc1_pos: 80.208 acc5_pos: 100.000 loss: 0.4856 (0.4993) acc1: 80.2083 (79.6875) acc5: 100.0000 (100.0000) time: 0.6493 data: 0.4131 max mem: 3531 +Test: [ 20/269] eta: 0:01:36 acc1_pos: 90.365 acc5_pos: 100.000 loss: 0.4700 (0.4686) acc1: 82.2917 (82.2545) acc5: 100.0000 (100.0000) time: 0.0970 data: 0.0004 max mem: 3531 +Test: [ 30/269] eta: 0:01:24 acc1_pos: 84.375 acc5_pos: 100.000 loss: 0.4554 (0.4726) acc1: 83.8542 (81.7372) acc5: 100.0000 (100.0000) time: 0.1917 data: 0.0953 max mem: 3531 +Test: [ 40/269] eta: 0:01:19 acc1_pos: 85.345 acc5_pos: 100.000 loss: 0.5885 (0.5249) acc1: 69.5312 (76.7022) acc5: 100.0000 (100.0000) acc1_neg: 55.990 acc5_neg: 100.000 time: 0.3019 data: 0.2083 max mem: 3531 +Test: [ 50/269] eta: 0:01:10 acc1_pos: 85.345 acc5_pos: 100.000 loss: 0.7556 (0.5829) acc1: 53.1250 (70.4453) acc5: 100.0000 (100.0000) acc1_neg: 29.688 acc5_neg: 100.000 time: 0.2700 data: 0.1779 max mem: 3531 +Test: [ 60/269] eta: 0:00:59 acc1_pos: 85.345 acc5_pos: 100.000 loss: 0.8471 (0.6340) acc1: 39.0625 (64.6986) acc5: 100.0000 (100.0000) acc1_neg: 27.604 acc5_neg: 100.000 time: 0.1579 data: 0.0650 max mem: 3531 +Test: [ 70/269] eta: 0:00:53 acc1_pos: 85.345 acc5_pos: 100.000 loss: 0.8572 (0.6622) acc1: 32.8125 (60.8421) acc5: 100.0000 (100.0000) acc1_neg: 32.292 acc5_neg: 100.000 time: 0.1430 data: 0.0489 max mem: 3531 +Test: [ 80/269] eta: 0:00:50 acc1_pos: 85.345 acc5_pos: 100.000 loss: 0.8300 (0.6809) acc1: 35.1562 (58.1726) acc5: 100.0000 (100.0000) acc1_neg: 26.562 acc5_neg: 100.000 time: 0.2149 data: 0.1188 max mem: 3531 +Test: [ 90/269] eta: 0:00:45 acc1_pos: 85.345 acc5_pos: 100.000 loss: 0.9071 (0.7163) acc1: 27.8646 (54.2497) acc5: 100.0000 (100.0000) acc1_neg: 21.875 acc5_neg: 100.000 time: 0.2062 data: 0.1106 max mem: 3531 +Test: [100/269] eta: 0:00:43 acc1_pos: 85.345 acc5_pos: 100.000 loss: 0.9889 (0.7434) acc1: 20.3125 (50.9824) acc5: 100.0000 (100.0000) acc1_neg: 14.583 acc5_neg: 100.000 time: 0.2058 data: 0.1146 max mem: 3531 +Test: [110/269] eta: 0:00:39 acc1_pos: 85.345 acc5_pos: 100.000 loss: 1.0172 (0.7712) acc1: 19.5312 (48.0551) acc5: 100.0000 (100.0000) acc1_neg: 15.625 acc5_neg: 100.000 time: 0.2264 data: 0.1349 max mem: 3531 +Test: [120/269] eta: 0:00:36 acc1_pos: 85.345 acc5_pos: 100.000 loss: 1.0172 (0.7900) acc1: 22.9167 (46.6469) acc5: 100.0000 (100.0000) acc1_neg: 20.052 acc5_neg: 100.000 time: 0.2107 data: 0.1186 max mem: 3531 +Test: [130/269] eta: 0:00:33 acc1_pos: 85.345 acc5_pos: 100.000 loss: 1.0265 (0.8084) acc1: 29.4271 (45.0461) acc5: 100.0000 (100.0000) acc1_neg: 15.365 acc5_neg: 100.000 time: 0.1921 data: 0.1006 max mem: 3531 +Test: [140/269] eta: 0:00:30 acc1_pos: 85.345 acc5_pos: 100.000 loss: 1.0890 (0.8308) acc1: 19.0104 (43.0556) acc5: 100.0000 (100.0000) acc1_neg: 14.583 acc5_neg: 100.000 time: 0.1834 data: 0.0913 max mem: 3531 +Test: [150/269] eta: 0:00:28 acc1_pos: 85.345 acc5_pos: 100.000 loss: 1.0960 (0.8391) acc1: 16.6667 (42.3738) acc5: 100.0000 (100.0000) acc1_neg: 39.062 acc5_neg: 100.000 time: 0.2099 data: 0.1145 max mem: 3531 +Test: [160/269] eta: 0:00:25 acc1_pos: 85.345 acc5_pos: 100.000 loss: 0.8750 (0.8419) acc1: 34.1146 (42.0306) acc5: 100.0000 (100.0000) acc1_neg: 42.708 acc5_neg: 100.000 time: 0.1997 data: 0.1024 max mem: 3531 +Test: [170/269] eta: 0:00:22 acc1_pos: 85.345 acc5_pos: 100.000 loss: 0.9285 (0.8481) acc1: 26.3021 (41.2022) acc5: 100.0000 (100.0000) acc1_neg: 38.281 acc5_neg: 100.000 time: 0.1723 data: 0.0753 max mem: 3531 +Test: [180/269] eta: 0:00:20 acc1_pos: 85.345 acc5_pos: 100.000 loss: 0.9051 (0.8495) acc1: 29.6875 (41.0135) acc5: 100.0000 (100.0000) acc1_neg: 29.688 acc5_neg: 100.000 time: 0.2219 data: 0.1281 max mem: 3531 +Test: [190/269] eta: 0:00:18 acc1_pos: 85.345 acc5_pos: 100.000 loss: 0.9051 (0.8547) acc1: 32.8125 (40.4396) acc5: 100.0000 (100.0000) acc1_neg: 30.469 acc5_neg: 100.000 time: 0.2075 data: 0.1144 max mem: 3531 +Test: [200/269] eta: 0:00:15 acc1_pos: 85.345 acc5_pos: 100.000 loss: 0.9629 (0.8616) acc1: 29.9479 (39.8956) acc5: 100.0000 (100.0000) acc1_neg: 29.948 acc5_neg: 100.000 time: 0.1528 data: 0.0589 max mem: 3531 +Test: [210/269] eta: 0:00:13 acc1_pos: 85.345 acc5_pos: 100.000 loss: 0.9453 (0.8602) acc1: 34.1146 (39.9956) acc5: 100.0000 (100.0000) acc1_neg: 50.521 acc5_neg: 100.000 time: 0.1867 data: 0.0951 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 85.345 acc5_pos: 100.000 loss: 0.7531 (0.8564) acc1: 45.3125 (40.4259) acc5: 100.0000 (100.0000) acc1_neg: 80.729 acc5_neg: 100.000 time: 0.2120 data: 0.1212 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 85.345 acc5_pos: 100.000 loss: 0.7005 (0.8497) acc1: 55.9896 (41.0861) acc5: 100.0000 (100.0000) acc1_neg: 57.031 acc5_neg: 100.000 time: 0.2007 data: 0.1091 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 85.345 acc5_pos: 100.000 loss: 0.8540 (0.8561) acc1: 35.1562 (40.3743) acc5: 100.0000 (100.0000) acc1_neg: 28.906 acc5_neg: 100.000 time: 0.2101 data: 0.1173 max mem: 3532 +Test: [250/269] eta: 0:00:04 acc1_pos: 85.345 acc5_pos: 100.000 loss: 0.9885 (0.8634) acc1: 21.0938 (39.8126) acc5: 100.0000 (100.0000) acc1_neg: 8.073 acc5_neg: 100.000 time: 0.1920 data: 0.1005 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 85.345 acc5_pos: 100.000 loss: 1.0569 (0.8720) acc1: 18.4896 (38.9478) acc5: 100.0000 (100.0000) acc1_neg: 27.083 acc5_neg: 100.000 time: 0.1392 data: 0.0495 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 85.345 acc5_pos: 100.000 loss: 1.0106 (0.8749) acc1: 15.6250 (38.4992) acc5: 100.0000 (100.0000) acc1_neg: 27.559 acc5_neg: 100.000 time: 0.1101 data: 0.0170 max mem: 3532 +Test: Total time: 0:00:57 (0.2129 s / it) +* Overall Acc@1 38.499 Acc@5 100.000 + target == real Acc@1 81.683 Acc@5 100.000 + target == fake Acc@1 32.606 Acc@5 100.000 + loss 0.875 + PR-AUC 0.9191 ROC-AUC 0.6238 EER 41.56% (thr=0.3762) +Accuracy of the network on the 103293 test images: 38.5% +[rank0]:[W725 18:09:42.975732924 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-25T18:09:49Z] eval_done tag=fixed500k_fake_n45_seed2_resampled +[2026-07-25T18:09:49Z] eval_skip tag=fixed500k_fake_n64_seed1_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed1_resampled +[2026-07-25T18:09:49Z] eval_wait tag=fixed500k_fake_n64_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed2_resampled +[2026-07-25T18:10:49Z] eval_wait tag=fixed500k_fake_n64_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed2_resampled +[2026-07-25T18:11:49Z] eval_wait tag=fixed500k_fake_n64_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed2_resampled +[2026-07-25T18:12:49Z] eval_wait tag=fixed500k_fake_n64_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed2_resampled +[2026-07-25T18:13:49Z] eval_wait tag=fixed500k_fake_n64_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed2_resampled +[2026-07-25T18:14:49Z] eval_wait tag=fixed500k_fake_n64_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed2_resampled +[2026-07-25T18:15:50Z] eval_wait tag=fixed500k_fake_n64_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed2_resampled +[2026-07-25T18:16:50Z] eval_skip tag=fixed500k_fake_n64_seed2_resampled reason=complete_after_lock output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed2_resampled +[2026-07-25T18:16:50Z] eval_skip tag=fixed500k_real_n11_seed1_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed1_resampled +[2026-07-25T18:16:50Z] eval_skip tag=fixed500k_real_n11_seed2_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed2_resampled +[2026-07-25T18:16:50Z] eval_skip tag=fixed500k_real_n11_seed3_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed3_resampled +[2026-07-25T18:16:50Z] eval_skip tag=fixed500k_real_n16_seed1_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed1_resampled +[2026-07-25T18:16:50Z] eval_skip tag=fixed500k_real_n16_seed2_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed2_resampled +[2026-07-25T18:16:50Z] eval_wait tag=fixed500k_real_n16_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed3_resampled +[2026-07-25T18:17:50Z] eval_wait tag=fixed500k_real_n16_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed3_resampled +[2026-07-25T18:18:50Z] eval_wait tag=fixed500k_real_n16_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed3_resampled +[2026-07-25T18:19:50Z] eval_wait tag=fixed500k_real_n16_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed3_resampled +[2026-07-25T18:20:51Z] eval_wait tag=fixed500k_real_n16_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed3_resampled +[2026-07-25T18:21:51Z] eval_wait tag=fixed500k_real_n16_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed3_resampled +[2026-07-25T18:22:51Z] eval_wait tag=fixed500k_real_n16_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed3_resampled +[2026-07-25T18:23:51Z] eval_wait tag=fixed500k_real_n16_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed3_resampled +[2026-07-25T18:24:51Z] eval_wait tag=fixed500k_real_n16_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed3_resampled +[2026-07-25T18:25:52Z] eval_wait tag=fixed500k_real_n16_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed3_resampled +[2026-07-25T18:26:52Z] eval_wait tag=fixed500k_real_n16_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed3_resampled +[2026-07-25T18:27:52Z] eval_wait tag=fixed500k_real_n16_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed3_resampled +[2026-07-25T18:28:52Z] eval_wait tag=fixed500k_real_n16_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed3_resampled +[2026-07-25T18:29:52Z] eval_wait tag=fixed500k_real_n16_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed3_resampled +[2026-07-25T18:30:52Z] eval_wait tag=fixed500k_real_n16_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed3_resampled +[2026-07-25T18:31:53Z] eval_wait tag=fixed500k_real_n16_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed3_resampled +[2026-07-25T18:32:53Z] eval_wait tag=fixed500k_real_n16_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed3_resampled +[2026-07-25T18:33:53Z] eval_wait tag=fixed500k_real_n16_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed3_resampled +[2026-07-25T18:34:53Z] eval_wait tag=fixed500k_real_n16_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed3_resampled +[2026-07-25T18:35:53Z] eval_wait tag=fixed500k_real_n16_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed3_resampled +[2026-07-25T18:36:54Z] eval_wait tag=fixed500k_real_n16_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed3_resampled +[2026-07-25T18:37:54Z] eval_skip tag=fixed500k_real_n16_seed3_resampled reason=complete_after_lock output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed3_resampled +[2026-07-25T18:37:54Z] eval_skip tag=fixed500k_real_n23_seed1_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed1_resampled +[2026-07-25T18:37:54Z] eval_skip tag=fixed500k_real_n23_seed2_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed2_resampled +[2026-07-25T18:37:54Z] eval_skip tag=fixed500k_real_n32_seed1_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed1_resampled +[2026-07-25T18:37:54Z] eval_skip tag=fixed500k_real_n32_seed2_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed2_resampled +[2026-07-25T18:37:54Z] eval_wait tag=fixed500k_real_n32_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed3_resampled +[2026-07-25T18:38:54Z] eval_skip tag=fixed500k_real_n32_seed3_resampled reason=complete_after_lock output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed3_resampled +[2026-07-25T18:38:54Z] eval_skip tag=fixed500k_real_n5_seed1_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed1_resampled +[2026-07-25T18:38:54Z] eval_skip tag=fixed500k_real_n5_seed2_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed2_resampled +[2026-07-25T18:38:54Z] eval_skip tag=fixed500k_real_n5_seed3_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed3_resampled +[2026-07-25T18:38:54Z] eval_skip tag=fixed500k_real_n8_seed1_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed1_resampled +[2026-07-25T18:38:54Z] eval_skip tag=fixed500k_real_n8_seed2_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed2_resampled +[2026-07-25T18:39:56Z] scan completed_train=48/48 eval_complete=35/48 queue_done=6/6 +[2026-07-25T18:39:56Z] eval_skip tag=fixed500k_fake_n11_seed1_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed1_resampled +[2026-07-25T18:39:56Z] eval_skip tag=fixed500k_fake_n11_seed2_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed2_resampled +[2026-07-25T18:39:56Z] eval_skip tag=fixed500k_fake_n11_seed3_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed3_resampled +[2026-07-25T18:39:56Z] eval_wait tag=fixed500k_fake_n11_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed4_resampled +[2026-07-25T18:40:56Z] eval_wait tag=fixed500k_fake_n11_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed4_resampled +[2026-07-25T18:41:57Z] eval_wait tag=fixed500k_fake_n11_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed4_resampled +[2026-07-25T18:42:57Z] eval_wait tag=fixed500k_fake_n11_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed4_resampled +[2026-07-25T18:43:57Z] eval_wait tag=fixed500k_fake_n11_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed4_resampled +[2026-07-25T18:44:57Z] eval_wait tag=fixed500k_fake_n11_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed4_resampled +[2026-07-25T18:45:57Z] eval_wait tag=fixed500k_fake_n11_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed4_resampled +[2026-07-25T18:46:57Z] eval_wait tag=fixed500k_fake_n11_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed4_resampled +[2026-07-25T18:47:58Z] eval_wait tag=fixed500k_fake_n11_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed4_resampled +[2026-07-25T18:48:58Z] eval_wait tag=fixed500k_fake_n11_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed4_resampled +[2026-07-25T18:49:58Z] eval_wait tag=fixed500k_fake_n11_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed4_resampled +[2026-07-25T18:50:58Z] eval_wait tag=fixed500k_fake_n11_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed4_resampled +[2026-07-25T18:51:58Z] eval_wait tag=fixed500k_fake_n11_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed4_resampled +[2026-07-25T18:52:58Z] eval_wait tag=fixed500k_fake_n11_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed4_resampled +[2026-07-25T18:53:59Z] eval_wait tag=fixed500k_fake_n11_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed4_resampled +[2026-07-25T18:54:59Z] eval_wait tag=fixed500k_fake_n11_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed4_resampled +[2026-07-25T18:55:59Z] eval_wait tag=fixed500k_fake_n11_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed4_resampled +[2026-07-25T18:56:59Z] eval_wait tag=fixed500k_fake_n11_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed4_resampled +[2026-07-25T18:57:59Z] eval_wait tag=fixed500k_fake_n11_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed4_resampled +[2026-07-25T18:58:59Z] eval_skip tag=fixed500k_fake_n11_seed4_resampled reason=complete_after_wait output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed4_resampled +[2026-07-25T18:58:59Z] eval_skip tag=fixed500k_fake_n16_seed1_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n16_seed1_resampled +[2026-07-25T18:58:59Z] eval_skip tag=fixed500k_fake_n16_seed2_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n16_seed2_resampled +[2026-07-25T18:58:59Z] eval_skip tag=fixed500k_fake_n16_seed3_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n16_seed3_resampled +[2026-07-25T18:59:00Z] eval_wait tag=fixed500k_fake_n16_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n16_seed4_resampled +[2026-07-25T19:00:00Z] eval_wait tag=fixed500k_fake_n16_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n16_seed4_resampled +[2026-07-25T19:01:00Z] eval_wait tag=fixed500k_fake_n16_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n16_seed4_resampled +[2026-07-25T19:02:00Z] eval_wait tag=fixed500k_fake_n16_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n16_seed4_resampled +[2026-07-25T19:03:00Z] eval_wait tag=fixed500k_fake_n16_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n16_seed4_resampled +[2026-07-25T19:04:00Z] eval_wait tag=fixed500k_fake_n16_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n16_seed4_resampled +[2026-07-25T19:05:00Z] eval_wait tag=fixed500k_fake_n16_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n16_seed4_resampled +[2026-07-25T19:06:01Z] eval_wait tag=fixed500k_fake_n16_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n16_seed4_resampled +[2026-07-25T19:07:01Z] eval_wait tag=fixed500k_fake_n16_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n16_seed4_resampled +[2026-07-25T19:08:01Z] eval_wait tag=fixed500k_fake_n16_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n16_seed4_resampled +[2026-07-25T19:09:01Z] eval_wait tag=fixed500k_fake_n16_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n16_seed4_resampled +[2026-07-25T19:10:01Z] eval_wait tag=fixed500k_fake_n16_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n16_seed4_resampled +[2026-07-25T19:11:01Z] eval_skip tag=fixed500k_fake_n16_seed4_resampled reason=complete_after_lock output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n16_seed4_resampled +[2026-07-25T19:11:02Z] eval_skip tag=fixed500k_fake_n22_seed1_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n22_seed1_resampled +[2026-07-25T19:11:02Z] eval_skip tag=fixed500k_fake_n22_seed2_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n22_seed2_resampled +[2026-07-25T19:11:02Z] eval_skip tag=fixed500k_fake_n22_seed3_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n22_seed3_resampled +[2026-07-25T19:11:02Z] eval_skip tag=fixed500k_fake_n22_seed4_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n22_seed4_resampled +[2026-07-25T19:11:02Z] eval_skip tag=fixed500k_fake_n32_seed1_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed1_resampled +[2026-07-25T19:11:02Z] eval_skip tag=fixed500k_fake_n32_seed2_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed2_resampled +[2026-07-25T19:11:02Z] eval_skip tag=fixed500k_fake_n32_seed3_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed3_resampled +[2026-07-25T19:11:02Z] eval_start tag=fixed500k_fake_n32_seed4_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed4_resampled_n111-046-081_20260725_112559/best_checkpoint.pth +started_utc=2026-07-25T19:11:02Z +hostname=n116-042-186 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed4_resampled_n111-046-081_20260725_112559/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed4_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0725 19:11:17.732000 101301 torch/distributed/run.py:766] +W0725 19:11:17.732000 101301 torch/distributed/run.py:766] ***************************************** +W0725 19:11:17.732000 101301 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 19:11:17.732000 101301 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 19:11:55.266378884 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 19:11:55.606242186 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 19:11:55.730058015 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 19:11:55.744629183 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 19:11:55.935239350 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 19:11:56.078142363 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 19:11:56.116455996 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 19:11:56.127082071 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed4_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed4_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:28:11 acc1_pos: 34.375 acc5_pos: 100.000 loss: 0.8428 (0.8428) acc1: 34.3750 (34.3750) acc5: 100.0000 (100.0000) time: 7.2304 data: 5.6890 max mem: 3531 +Test: [ 10/234] eta: 0:02:46 acc1_pos: 25.000 acc5_pos: 100.000 loss: 0.7894 (0.8171) acc1: 38.5417 (45.3835) acc5: 100.0000 (100.0000) time: 0.7435 data: 0.5176 max mem: 3531 +Test: [ 20/234] eta: 0:01:33 acc1_pos: 97.396 acc5_pos: 100.000 loss: 0.7894 (0.7886) acc1: 38.2812 (45.1761) acc5: 100.0000 (100.0000) time: 0.0973 data: 0.0004 max mem: 3531 +Test: [ 30/234] eta: 0:01:20 acc1_pos: 77.604 acc5_pos: 100.000 loss: 0.1863 (0.6000) acc1: 97.1354 (61.1643) acc5: 100.0000 (100.0000) time: 0.2009 data: 0.1056 max mem: 3531 +Test: [ 40/234] eta: 0:01:06 acc1_pos: 98.958 acc5_pos: 100.000 loss: 0.1578 (0.5399) acc1: 97.6562 (66.8001) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1477 max mem: 3531 +Test: [ 50/234] eta: 0:00:56 acc1_pos: 99.479 acc5_pos: 100.000 loss: 0.1298 (0.4588) acc1: 98.9583 (73.1516) acc5: 100.0000 (100.0000) time: 0.1744 data: 0.0826 max mem: 3531 +Test: [ 60/234] eta: 0:00:49 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.0938 (0.3957) acc1: 99.7396 (77.5487) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1753 data: 0.0825 max mem: 3531 +Test: [ 70/234] eta: 0:00:44 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.0740 (0.4154) acc1: 100.0000 (76.8559) acc5: 100.0000 (100.0000) acc1_neg: 55.990 acc5_neg: 100.000 time: 0.1894 data: 0.0938 max mem: 3531 +Test: [ 80/234] eta: 0:00:40 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.7468 (0.4351) acc1: 59.3750 (76.3085) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1921 data: 0.0977 max mem: 3531 +Test: [ 90/234] eta: 0:00:37 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.3022 (0.4191) acc1: 84.1146 (77.7616) acc5: 100.0000 (100.0000) acc1_neg: 74.219 acc5_neg: 100.000 time: 0.2020 data: 0.1104 max mem: 3531 +Test: [100/234] eta: 0:00:33 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.3757 (0.4338) acc1: 83.3333 (77.1452) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2125 data: 0.1202 max mem: 3531 +Test: [110/234] eta: 0:00:30 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.6282 (0.4424) acc1: 65.1042 (76.9238) acc5: 100.0000 (100.0000) acc1_neg: 57.031 acc5_neg: 100.000 time: 0.2042 data: 0.1104 max mem: 3531 +Test: [120/234] eta: 0:00:27 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.4533 (0.4503) acc1: 78.3854 (76.6830) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.1958 data: 0.1016 max mem: 3531 +Test: [130/234] eta: 0:00:24 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.3931 (0.4472) acc1: 84.8958 (77.3517) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.1785 data: 0.0850 max mem: 3531 +Test: [140/234] eta: 0:00:22 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.4385 (0.4794) acc1: 76.8229 (75.8699) acc5: 100.0000 (100.0000) acc1_neg: 29.427 acc5_neg: 100.000 time: 0.1788 data: 0.0840 max mem: 3531 +Test: [150/234] eta: 0:00:19 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.5655 (0.4917) acc1: 69.2708 (75.3001) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.1949 data: 0.0970 max mem: 3531 +Test: [160/234] eta: 0:00:17 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.5119 (0.4868) acc1: 76.3021 (75.8832) acc5: 100.0000 (100.0000) acc1_neg: 76.302 acc5_neg: 100.000 time: 0.2152 data: 0.1191 max mem: 3531 +Test: [170/234] eta: 0:00:14 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.5119 (0.4872) acc1: 77.3438 (76.0341) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.2317 data: 0.1390 max mem: 3531 +Test: [180/234] eta: 0:00:12 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.5310 (0.4978) acc1: 74.7396 (75.4331) acc5: 100.0000 (100.0000) acc1_neg: 34.115 acc5_neg: 100.000 time: 0.2146 data: 0.1231 max mem: 3531 +Test: [190/234] eta: 0:00:10 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.7806 (0.5141) acc1: 58.3333 (74.4764) acc5: 100.0000 (100.0000) acc1_neg: 59.115 acc5_neg: 100.000 time: 0.2056 data: 0.1146 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.7671 (0.5065) acc1: 59.1146 (74.9663) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2000 data: 0.1100 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.1171 (0.4884) acc1: 99.7396 (76.1145) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1671 data: 0.0770 max mem: 3532 +Test: [220/234] eta: 0:00:03 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.1366 (0.4746) acc1: 98.4375 (76.9855) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.1392 data: 0.0494 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 99.310 acc5_pos: 100.000 loss: 0.5232 (0.5076) acc1: 33.3333 (74.7892) acc5: 100.0000 (100.0000) acc1_neg: 24.740 acc5_neg: 100.000 time: 0.1323 data: 0.0427 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 99.310 acc5_pos: 100.000 loss: 1.1990 (0.5168) acc1: 26.0700 (74.2224) acc5: 100.0000 (100.0000) acc1_neg: 26.070 acc5_neg: 100.000 time: 0.1270 data: 0.0364 max mem: 3532 +Test: Total time: 0:00:50 (0.2147 s / it) +* Overall Acc@1 74.222 Acc@5 100.000 + target == real Acc@1 73.344 Acc@5 100.000 + target == fake Acc@1 74.470 Acc@5 100.000 + loss 0.517 + PR-AUC 0.9527 ROC-AUC 0.8446 EER 26.01% (thr=0.5083) +Accuracy of the network on the 89729 test images: 74.2% +[rank0]:[W725 19:12:59.601934449 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 19:13:20.603000 101857 torch/distributed/run.py:766] +W0725 19:13:20.603000 101857 torch/distributed/run.py:766] ***************************************** +W0725 19:13:20.603000 101857 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 19:13:20.603000 101857 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 19:13:57.123643797 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 19:13:57.793440149 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 19:13:57.008107595 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 19:13:57.039147680 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 19:13:58.141911274 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 19:13:58.155497025 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +[rank0]:[W725 19:13:58.171726131 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 19:13:58.172646379 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed4_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed4_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:30:52 acc1_pos: 24.740 acc5_pos: 100.000 loss: 1.0875 (1.0875) acc1: 24.7396 (24.7396) acc5: 100.0000 (100.0000) time: 6.8086 data: 5.1416 max mem: 3531 +Test: [ 10/2387] eta: 0:27:46 acc1_pos: 8.854 acc5_pos: 100.000 loss: 1.2275 (1.2202) acc1: 14.0625 (16.0511) acc5: 100.0000 (100.0000) time: 0.7010 data: 0.4678 max mem: 3531 +Test: [ 20/2387] eta: 0:16:15 acc1_pos: 13.021 acc5_pos: 100.000 loss: 1.2052 (1.1936) acc1: 14.8438 (17.5099) acc5: 100.0000 (100.0000) time: 0.0921 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:14:17 acc1_pos: 5.990 acc5_pos: 100.000 loss: 1.1958 (1.2098) acc1: 14.0625 (15.5494) acc5: 100.0000 (100.0000) time: 0.1784 data: 0.0855 max mem: 3531 +Test: [ 40/2387] eta: 0:13:13 acc1_pos: 9.115 acc5_pos: 100.000 loss: 1.2567 (1.2267) acc1: 8.5938 (13.3892) acc5: 100.0000 (100.0000) time: 0.2609 data: 0.1682 max mem: 3531 +Test: [ 50/2387] eta: 0:12:47 acc1_pos: 34.375 acc5_pos: 100.000 loss: 1.2184 (1.2263) acc1: 8.5938 (13.3374) acc5: 100.0000 (100.0000) time: 0.2735 data: 0.1793 max mem: 3531 +Test: [ 60/2387] eta: 0:11:58 acc1_pos: 40.365 acc5_pos: 100.000 loss: 1.1840 (1.2159) acc1: 13.5417 (14.3058) acc5: 100.0000 (100.0000) time: 0.2489 data: 0.1547 max mem: 3531 +Test: [ 70/2387] eta: 0:11:23 acc1_pos: 31.250 acc5_pos: 100.000 loss: 1.2004 (1.2206) acc1: 13.5417 (14.2972) acc5: 100.0000 (100.0000) time: 0.2102 data: 0.1175 max mem: 3531 +Test: [ 80/2387] eta: 0:11:19 acc1_pos: 15.885 acc5_pos: 100.000 loss: 1.2004 (1.2149) acc1: 11.4583 (14.1975) acc5: 100.0000 (100.0000) time: 0.2499 data: 0.1528 max mem: 3531 +Test: [ 90/2387] eta: 0:11:10 acc1_pos: 23.177 acc5_pos: 100.000 loss: 1.1100 (1.2001) acc1: 17.1875 (14.8123) acc5: 100.0000 (100.0000) time: 0.2809 data: 0.1838 max mem: 3531 +Test: [ 100/2387] eta: 0:10:50 acc1_pos: 13.021 acc5_pos: 100.000 loss: 1.0987 (1.1974) acc1: 15.1042 (14.5833) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1502 max mem: 3531 +Test: [ 110/2387] eta: 0:10:40 acc1_pos: 24.219 acc5_pos: 100.000 loss: 1.1052 (1.1885) acc1: 14.3229 (14.8438) acc5: 100.0000 (100.0000) time: 0.2336 data: 0.1372 max mem: 3531 +Test: [ 120/2387] eta: 0:10:32 acc1_pos: 16.927 acc5_pos: 100.000 loss: 1.0931 (1.1821) acc1: 15.8854 (15.2806) acc5: 100.0000 (100.0000) time: 0.2499 data: 0.1552 max mem: 3531 +Test: [ 130/2387] eta: 0:10:24 acc1_pos: 22.396 acc5_pos: 100.000 loss: 1.0926 (1.1783) acc1: 14.3229 (15.1718) acc5: 100.0000 (100.0000) time: 0.2499 data: 0.1540 max mem: 3531 +Test: [ 140/2387] eta: 0:10:18 acc1_pos: 16.927 acc5_pos: 100.000 loss: 1.0927 (1.1730) acc1: 14.3229 (15.3794) acc5: 100.0000 (100.0000) time: 0.2549 data: 0.1547 max mem: 3531 +Test: [ 150/2387] eta: 0:10:03 acc1_pos: 1.823 acc5_pos: 100.000 loss: 1.0793 (1.1672) acc1: 17.4479 (15.4681) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1261 max mem: 3531 +Test: [ 160/2387] eta: 0:09:59 acc1_pos: 1.826 acc5_pos: 100.000 loss: 1.0431 (1.1467) acc1: 17.4479 (17.0532) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2249 data: 0.1299 max mem: 3531 +Test: [ 170/2387] eta: 0:09:43 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.2303 (1.0927) acc1: 97.3958 (21.7973) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2149 data: 0.1201 max mem: 3531 +Test: [ 180/2387] eta: 0:09:27 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.2230 (1.0446) acc1: 98.1771 (26.0115) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1599 data: 0.0636 max mem: 3531 +Test: [ 190/2387] eta: 0:09:18 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.2212 (1.0014) acc1: 98.1771 (29.7993) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1749 data: 0.0776 max mem: 3531 +Test: [ 200/2387] eta: 0:09:12 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.2177 (0.9625) acc1: 98.1771 (33.2193) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2099 data: 0.1104 max mem: 3531 +Test: [ 210/2387] eta: 0:09:07 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.2242 (0.9277) acc1: 98.1771 (36.2769) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2265 data: 0.1250 max mem: 3532 +Test: [ 220/2387] eta: 0:08:58 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.2251 (0.8958) acc1: 98.1771 (39.0967) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2099 data: 0.1122 max mem: 3532 +Test: [ 230/2387] eta: 0:08:44 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.2232 (0.8667) acc1: 98.1771 (41.6599) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1538 data: 0.0587 max mem: 3532 +Test: [ 240/2387] eta: 0:08:36 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.2289 (0.8405) acc1: 98.1771 (44.0180) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1510 data: 0.0568 max mem: 3532 +Test: [ 250/2387] eta: 0:08:31 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.2150 (0.8119) acc1: 98.9583 (46.2411) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1953 data: 0.1031 max mem: 3532 +Test: [ 260/2387] eta: 0:08:24 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1058 (0.7848) acc1: 100.0000 (48.3008) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2003 data: 0.1096 max mem: 3532 +Test: [ 270/2387] eta: 0:08:18 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1010 (0.7595) acc1: 100.0000 (50.2076) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1891 data: 0.0947 max mem: 3532 +Test: [ 280/2387] eta: 0:08:18 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1025 (0.7363) acc1: 100.0000 (51.9786) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2284 data: 0.1300 max mem: 3532 +Test: [ 290/2387] eta: 0:08:10 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1094 (0.7149) acc1: 100.0000 (53.6288) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2152 data: 0.1182 max mem: 3532 +Test: [ 300/2387] eta: 0:08:08 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1143 (0.6950) acc1: 100.0000 (55.1694) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1999 data: 0.1043 max mem: 3532 +Test: [ 310/2387] eta: 0:08:04 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1196 (0.6785) acc1: 100.0000 (56.5866) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2196 data: 0.1233 max mem: 3532 +Test: [ 320/2387] eta: 0:07:56 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.2533 (0.6653) acc1: 99.2188 (57.9099) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1715 data: 0.0749 max mem: 3532 +Test: [ 330/2387] eta: 0:07:49 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.2551 (0.6528) acc1: 99.2188 (59.1586) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1499 data: 0.0531 max mem: 3532 +Test: [ 340/2387] eta: 0:07:44 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.2517 (0.6411) acc1: 99.2188 (60.3311) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1748 data: 0.0762 max mem: 3532 +Test: [ 350/2387] eta: 0:07:46 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.2463 (0.6274) acc1: 99.4792 (61.4576) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2425 data: 0.1443 max mem: 3532 +Test: [ 360/2387] eta: 0:07:42 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1762 (0.6151) acc1: 99.7396 (62.4993) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2449 data: 0.1500 max mem: 3532 +Test: [ 370/2387] eta: 0:07:32 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1856 (0.6036) acc1: 99.2188 (63.4848) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1472 data: 0.0526 max mem: 3532 +Test: [ 380/2387] eta: 0:07:28 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1876 (0.5927) acc1: 98.9583 (64.4166) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1434 data: 0.0487 max mem: 3532 +Test: [ 390/2387] eta: 0:07:25 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1893 (0.5824) acc1: 98.9583 (65.3006) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1908 data: 0.0986 max mem: 3532 +Test: [ 400/2387] eta: 0:07:21 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1886 (0.5726) acc1: 98.9583 (66.1419) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1912 data: 0.0995 max mem: 3532 +Test: [ 410/2387] eta: 0:07:17 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1863 (0.5632) acc1: 99.2188 (66.9467) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1940 data: 0.1007 max mem: 3532 +Test: [ 420/2387] eta: 0:07:14 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1862 (0.5543) acc1: 98.9583 (67.7102) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1990 data: 0.1035 max mem: 3532 +Test: [ 430/2387] eta: 0:07:10 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1917 (0.5460) acc1: 99.2188 (68.4406) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1871 data: 0.0882 max mem: 3532 +Test: [ 440/2387] eta: 0:07:08 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1924 (0.5367) acc1: 99.4792 (69.1486) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1970 data: 0.0980 max mem: 3533 +Test: [ 450/2387] eta: 0:07:04 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.0880 (0.5268) acc1: 100.0000 (69.8327) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2027 data: 0.1073 max mem: 3533 +Test: [ 460/2387] eta: 0:07:02 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.0868 (0.5172) acc1: 100.0000 (70.4871) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2027 data: 0.1086 max mem: 3533 +Test: [ 470/2387] eta: 0:06:57 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.0870 (0.5081) acc1: 100.0000 (71.1137) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1866 data: 0.0932 max mem: 3533 +Test: [ 480/2387] eta: 0:06:53 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.0877 (0.4994) acc1: 100.0000 (71.7142) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1605 data: 0.0671 max mem: 3533 +Test: [ 490/2387] eta: 0:06:49 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.0879 (0.4910) acc1: 100.0000 (72.2903) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1706 data: 0.0779 max mem: 3533 +Test: [ 500/2387] eta: 0:06:45 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.0882 (0.4830) acc1: 100.0000 (72.8434) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1761 data: 0.0834 max mem: 3533 +Test: [ 510/2387] eta: 0:06:41 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.0882 (0.4755) acc1: 100.0000 (73.3748) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1695 data: 0.0764 max mem: 3533 +Test: [ 520/2387] eta: 0:06:37 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1262 (0.4691) acc1: 100.0000 (73.8859) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1588 data: 0.0639 max mem: 3533 +Test: [ 530/2387] eta: 0:06:33 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1253 (0.4626) acc1: 100.0000 (74.3776) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1547 data: 0.0592 max mem: 3533 +Test: [ 540/2387] eta: 0:06:30 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1244 (0.4563) acc1: 100.0000 (74.8513) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1690 data: 0.0757 max mem: 3533 +Test: [ 550/2387] eta: 0:06:28 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1240 (0.4503) acc1: 100.0000 (75.3077) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1982 data: 0.1050 max mem: 3533 +Test: [ 560/2387] eta: 0:06:22 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1236 (0.4445) acc1: 100.0000 (75.7478) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1649 data: 0.0698 max mem: 3533 +Test: [ 570/2387] eta: 0:06:17 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1240 (0.4388) acc1: 100.0000 (76.1726) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1099 data: 0.0117 max mem: 3533 +Test: [ 580/2387] eta: 0:06:14 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1240 (0.4334) acc1: 100.0000 (76.5827) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1359 data: 0.0391 max mem: 3533 +Test: [ 590/2387] eta: 0:06:11 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1241 (0.4282) acc1: 100.0000 (76.9789) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0825 max mem: 3533 +Test: [ 600/2387] eta: 0:06:08 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1268 (0.4235) acc1: 100.0000 (77.3615) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1739 data: 0.0797 max mem: 3533 +Test: [ 610/2387] eta: 0:06:04 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1548 (0.4194) acc1: 100.0000 (77.7261) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1649 data: 0.0687 max mem: 3533 +Test: [ 620/2387] eta: 0:06:01 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1731 (0.4155) acc1: 99.7396 (78.0785) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1599 data: 0.0638 max mem: 3533 +Test: [ 630/2387] eta: 0:05:58 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1727 (0.4116) acc1: 99.7396 (78.4209) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1649 data: 0.0672 max mem: 3533 +Test: [ 640/2387] eta: 0:05:55 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1765 (0.4080) acc1: 99.4792 (78.7494) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1799 data: 0.0846 max mem: 3533 +Test: [ 650/2387] eta: 0:05:53 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1775 (0.4044) acc1: 99.4792 (79.0699) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1862 data: 0.0938 max mem: 3533 +Test: [ 660/2387] eta: 0:05:51 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1773 (0.4010) acc1: 99.7396 (79.3826) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1957 data: 0.1041 max mem: 3533 +Test: [ 670/2387] eta: 0:05:48 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1761 (0.3976) acc1: 99.7396 (79.6844) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1847 data: 0.0927 max mem: 3534 +Test: [ 680/2387] eta: 0:05:45 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1738 (0.3944) acc1: 99.7396 (79.9812) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1642 data: 0.0714 max mem: 3534 +Test: [ 690/2387] eta: 0:05:42 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1787 (0.3951) acc1: 99.7396 (79.9641) acc5: 100.0000 (100.0000) acc1_neg: 68.229 acc5_neg: 100.000 time: 0.1738 data: 0.0792 max mem: 3534 +Test: [ 700/2387] eta: 0:05:41 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.5330 (0.3970) acc1: 71.0938 (79.8413) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.2006 data: 0.1061 max mem: 3534 +Test: [ 710/2387] eta: 0:05:38 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.5323 (0.3989) acc1: 71.0938 (79.7175) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.2008 data: 0.1094 max mem: 3534 +Test: [ 720/2387] eta: 0:05:36 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.5315 (0.4007) acc1: 71.6146 (79.6095) acc5: 100.0000 (100.0000) acc1_neg: 73.958 acc5_neg: 100.000 time: 0.1941 data: 0.1019 max mem: 3534 +Test: [ 730/2387] eta: 0:05:34 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.5316 (0.4026) acc1: 71.0938 (79.4948) acc5: 100.0000 (100.0000) acc1_neg: 69.531 acc5_neg: 100.000 time: 0.1893 data: 0.0966 max mem: 3534 +Test: [ 740/2387] eta: 0:05:31 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.5316 (0.4044) acc1: 71.3542 (79.3933) acc5: 100.0000 (100.0000) acc1_neg: 66.927 acc5_neg: 100.000 time: 0.1827 data: 0.0914 max mem: 3534 +Test: [ 750/2387] eta: 0:05:29 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.5299 (0.4060) acc1: 72.1354 (79.2998) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.1860 data: 0.0947 max mem: 3534 +Test: [ 760/2387] eta: 0:05:27 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.5294 (0.4068) acc1: 72.1354 (79.2642) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.1825 data: 0.0912 max mem: 3534 +Test: [ 770/2387] eta: 0:05:24 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.3575 (0.4053) acc1: 85.1562 (79.4176) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1866 data: 0.0929 max mem: 3534 +Test: [ 780/2387] eta: 0:05:22 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.2080 (0.4025) acc1: 98.4375 (79.6682) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1904 data: 0.0970 max mem: 3534 +Test: [ 790/2387] eta: 0:05:20 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1749 (0.3997) acc1: 98.9583 (79.8821) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1997 data: 0.1091 max mem: 3534 +Test: [ 800/2387] eta: 0:05:17 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1804 (0.3971) acc1: 98.4375 (80.1108) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1852 data: 0.0910 max mem: 3534 +Test: [ 810/2387] eta: 0:05:15 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1945 (0.3946) acc1: 98.9583 (80.3458) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1731 data: 0.0762 max mem: 3534 +Test: [ 820/2387] eta: 0:05:13 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1974 (0.3922) acc1: 98.9583 (80.5655) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1844 data: 0.0871 max mem: 3534 +Test: [ 830/2387] eta: 0:05:11 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.2229 (0.3910) acc1: 96.6146 (80.7185) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1887 data: 0.0935 max mem: 3534 +Test: [ 840/2387] eta: 0:05:08 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.2530 (0.3891) acc1: 96.6146 (80.9181) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1850 data: 0.0916 max mem: 3534 +Test: [ 850/2387] eta: 0:05:06 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.2332 (0.3873) acc1: 97.6562 (81.1080) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1755 data: 0.0821 max mem: 3534 +Test: [ 860/2387] eta: 0:05:03 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.2249 (0.3853) acc1: 98.6979 (81.3126) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1673 data: 0.0754 max mem: 3534 +Test: [ 870/2387] eta: 0:05:01 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.2187 (0.3835) acc1: 99.4792 (81.4967) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1747 data: 0.0829 max mem: 3534 +Test: [ 880/2387] eta: 0:04:59 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.2202 (0.3818) acc1: 98.6979 (81.6904) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1821 data: 0.0898 max mem: 3534 +Test: [ 890/2387] eta: 0:04:56 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.2124 (0.3799) acc1: 99.4792 (81.8907) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1821 data: 0.0880 max mem: 3534 +Test: [ 900/2387] eta: 0:04:54 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.2316 (0.3830) acc1: 97.3958 (81.7191) acc5: 100.0000 (100.0000) acc1_neg: 60.677 acc5_neg: 100.000 time: 0.1817 data: 0.0851 max mem: 3535 +Test: [ 910/2387] eta: 0:04:52 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.7468 (0.3870) acc1: 58.5938 (81.4658) acc5: 100.0000 (100.0000) acc1_neg: 60.156 acc5_neg: 100.000 time: 0.1872 data: 0.0931 max mem: 3535 +Test: [ 920/2387] eta: 0:04:50 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.7619 (0.3911) acc1: 57.2917 (81.2031) acc5: 100.0000 (100.0000) acc1_neg: 54.427 acc5_neg: 100.000 time: 0.1876 data: 0.0969 max mem: 3535 +Test: [ 930/2387] eta: 0:04:48 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.7620 (0.3950) acc1: 57.5521 (80.9644) acc5: 100.0000 (100.0000) acc1_neg: 54.167 acc5_neg: 100.000 time: 0.2002 data: 0.1085 max mem: 3535 +Test: [ 940/2387] eta: 0:04:46 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.7735 (0.3990) acc1: 57.5521 (80.6976) acc5: 100.0000 (100.0000) acc1_neg: 53.646 acc5_neg: 100.000 time: 0.1998 data: 0.1037 max mem: 3535 +Test: [ 950/2387] eta: 0:04:44 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.7632 (0.4028) acc1: 56.2500 (80.4515) acc5: 100.0000 (100.0000) acc1_neg: 58.333 acc5_neg: 100.000 time: 0.1854 data: 0.0872 max mem: 3535 +Test: [ 960/2387] eta: 0:04:42 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.7500 (0.4064) acc1: 58.3333 (80.2219) acc5: 100.0000 (100.0000) acc1_neg: 60.677 acc5_neg: 100.000 time: 0.1909 data: 0.0967 max mem: 3535 +Test: [ 970/2387] eta: 0:04:40 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.7505 (0.4101) acc1: 57.5521 (79.9731) acc5: 100.0000 (100.0000) acc1_neg: 56.771 acc5_neg: 100.000 time: 0.1992 data: 0.1077 max mem: 3535 +Test: [ 980/2387] eta: 0:04:38 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.7499 (0.4104) acc1: 57.0312 (79.9920) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.2089 data: 0.1157 max mem: 3535 +Test: [ 990/2387] eta: 0:04:36 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.2270 (0.4079) acc1: 97.1354 (80.1645) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2008 data: 0.1076 max mem: 3535 +Test: [1000/2387] eta: 0:04:34 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1195 (0.4051) acc1: 100.0000 (80.3486) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1952 data: 0.1039 max mem: 3535 +Test: [1010/2387] eta: 0:04:32 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1340 (0.4027) acc1: 99.7396 (80.5344) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1889 data: 0.0966 max mem: 3535 +Test: [1020/2387] eta: 0:04:30 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1684 (0.4007) acc1: 99.7396 (80.7026) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1896 data: 0.0970 max mem: 3535 +Test: [1030/2387] eta: 0:04:28 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1827 (0.3987) acc1: 99.2188 (80.8784) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2160 data: 0.1244 max mem: 3535 +Test: [1040/2387] eta: 0:04:25 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1766 (0.3964) acc1: 99.7396 (81.0576) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1768 data: 0.0847 max mem: 3535 +Test: [1050/2387] eta: 0:04:23 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1742 (0.3943) acc1: 99.7396 (81.2272) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1404 data: 0.0457 max mem: 3535 +Test: [1060/2387] eta: 0:04:21 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1762 (0.3923) acc1: 99.7396 (81.4009) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1647 data: 0.0711 max mem: 3535 +Test: [1070/2387] eta: 0:04:18 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1805 (0.3905) acc1: 99.7396 (81.5663) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1633 data: 0.0693 max mem: 3535 +Test: [1080/2387] eta: 0:04:16 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.2083 (0.3891) acc1: 99.2188 (81.7176) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1546 data: 0.0606 max mem: 3535 +Test: [1090/2387] eta: 0:04:13 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.2127 (0.3873) acc1: 99.2188 (81.8775) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1599 data: 0.0691 max mem: 3535 +Test: [1100/2387] eta: 0:04:11 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1992 (0.3855) acc1: 99.4792 (82.0263) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1758 data: 0.0821 max mem: 3535 +Test: [1110/2387] eta: 0:04:09 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.2010 (0.3841) acc1: 98.9583 (82.1632) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1848 data: 0.0907 max mem: 3535 +Test: [1120/2387] eta: 0:04:07 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.2079 (0.3825) acc1: 98.6979 (82.3116) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1775 data: 0.0863 max mem: 3536 +Test: [1130/2387] eta: 0:04:05 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.2082 (0.3809) acc1: 98.9583 (82.4609) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1675 data: 0.0767 max mem: 3536 +Test: [1140/2387] eta: 0:04:02 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.2056 (0.3794) acc1: 99.2188 (82.6062) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1659 data: 0.0750 max mem: 3536 +Test: [1150/2387] eta: 0:04:00 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.2027 (0.3779) acc1: 99.4792 (82.7410) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1584 data: 0.0650 max mem: 3536 +Test: [1160/2387] eta: 0:03:58 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1801 (0.3761) acc1: 99.4792 (82.8753) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1502 data: 0.0565 max mem: 3536 +Test: [1170/2387] eta: 0:03:55 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1583 (0.3743) acc1: 99.7396 (83.0173) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1538 data: 0.0625 max mem: 3536 +Test: [1180/2387] eta: 0:03:53 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1583 (0.3726) acc1: 99.4792 (83.1470) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1636 data: 0.0701 max mem: 3536 +Test: [1190/2387] eta: 0:03:51 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1782 (0.3709) acc1: 99.2188 (83.2868) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1555 data: 0.0610 max mem: 3536 +Test: [1200/2387] eta: 0:03:48 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1600 (0.3691) acc1: 100.0000 (83.4259) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1499 data: 0.0579 max mem: 3536 +Test: [1210/2387] eta: 0:03:46 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1600 (0.3674) acc1: 100.0000 (83.5624) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1510 data: 0.0590 max mem: 3536 +Test: [1220/2387] eta: 0:03:44 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1186 (0.3653) acc1: 100.0000 (83.6963) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1549 data: 0.0595 max mem: 3536 +Test: [1230/2387] eta: 0:03:41 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1027 (0.3631) acc1: 100.0000 (83.8288) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1583 data: 0.0625 max mem: 3536 +Test: [1240/2387] eta: 0:03:39 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1011 (0.3610) acc1: 100.0000 (83.9589) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1499 data: 0.0545 max mem: 3536 +Test: [1250/2387] eta: 0:03:37 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1039 (0.3590) acc1: 100.0000 (84.0871) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1557 data: 0.0623 max mem: 3536 +Test: [1260/2387] eta: 0:03:35 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1073 (0.3570) acc1: 100.0000 (84.2133) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1714 data: 0.0799 max mem: 3536 +Test: [1270/2387] eta: 0:03:33 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1151 (0.3552) acc1: 100.0000 (84.3373) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1763 data: 0.0846 max mem: 3536 +Test: [1280/2387] eta: 0:03:31 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1189 (0.3534) acc1: 100.0000 (84.4575) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1858 data: 0.0953 max mem: 3536 +Test: [1290/2387] eta: 0:03:29 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1254 (0.3517) acc1: 100.0000 (84.5711) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1988 data: 0.1072 max mem: 3536 +Test: [1300/2387] eta: 0:03:27 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1254 (0.3500) acc1: 99.7396 (84.6861) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1936 data: 0.1020 max mem: 3536 +Test: [1310/2387] eta: 0:03:25 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1324 (0.3484) acc1: 100.0000 (84.7977) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1704 data: 0.0798 max mem: 3536 +Test: [1320/2387] eta: 0:03:23 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1425 (0.3469) acc1: 100.0000 (84.9077) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1591 data: 0.0683 max mem: 3536 +Test: [1330/2387] eta: 0:03:21 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1576 (0.3456) acc1: 99.7396 (85.0166) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1798 data: 0.0872 max mem: 3536 +Test: [1340/2387] eta: 0:03:19 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1899 (0.3452) acc1: 98.9583 (85.0508) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2124 data: 0.1203 max mem: 3536 +Test: [1350/2387] eta: 0:03:17 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.0691 (0.3431) acc1: 100.0000 (85.1615) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1886 data: 0.0950 max mem: 3537 +Test: [1360/2387] eta: 0:03:15 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.0593 (0.3410) acc1: 100.0000 (85.2705) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1533 data: 0.0586 max mem: 3537 +Test: [1370/2387] eta: 0:03:13 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.0557 (0.3389) acc1: 100.0000 (85.3779) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1713 data: 0.0787 max mem: 3537 +Test: [1380/2387] eta: 0:03:11 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.0562 (0.3369) acc1: 100.0000 (85.4838) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1938 data: 0.1023 max mem: 3537 +Test: [1390/2387] eta: 0:03:09 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.0606 (0.3350) acc1: 100.0000 (85.5882) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2054 data: 0.1126 max mem: 3537 +Test: [1400/2387] eta: 0:03:08 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.0641 (0.3331) acc1: 100.0000 (85.6910) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1948 data: 0.1019 max mem: 3537 +Test: [1410/2387] eta: 0:03:06 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.0694 (0.3319) acc1: 100.0000 (85.7808) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1829 data: 0.0920 max mem: 3537 +Test: [1420/2387] eta: 0:03:03 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1900 (0.3310) acc1: 98.9583 (85.8570) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1599 data: 0.0661 max mem: 3537 +Test: [1430/2387] eta: 0:03:01 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.2166 (0.3303) acc1: 97.6562 (85.9353) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1574 data: 0.0640 max mem: 3537 +Test: [1440/2387] eta: 0:02:59 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.2114 (0.3294) acc1: 98.1771 (86.0232) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1768 data: 0.0861 max mem: 3537 +Test: [1450/2387] eta: 0:02:57 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.2066 (0.3288) acc1: 98.6979 (86.1041) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1777 data: 0.0859 max mem: 3537 +Test: [1460/2387] eta: 0:02:55 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1874 (0.3276) acc1: 98.9583 (86.1904) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1731 data: 0.0788 max mem: 3537 +Test: [1470/2387] eta: 0:02:53 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1418 (0.3263) acc1: 99.7396 (86.2829) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1623 data: 0.0692 max mem: 3537 +Test: [1480/2387] eta: 0:02:51 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1401 (0.3251) acc1: 99.7396 (86.3688) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1717 data: 0.0806 max mem: 3537 +Test: [1490/2387] eta: 0:02:49 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1433 (0.3240) acc1: 99.7396 (86.4555) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1822 data: 0.0909 max mem: 3537 +Test: [1500/2387] eta: 0:02:47 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1470 (0.3228) acc1: 99.7396 (86.5437) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1752 data: 0.0819 max mem: 3537 +Test: [1510/2387] eta: 0:02:45 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1565 (0.3218) acc1: 100.0000 (86.6321) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1713 data: 0.0780 max mem: 3537 +Test: [1520/2387] eta: 0:02:43 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1761 (0.3210) acc1: 100.0000 (86.7112) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1779 data: 0.0831 max mem: 3537 +Test: [1530/2387] eta: 0:02:42 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.2078 (0.3203) acc1: 98.9583 (86.7870) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1905 data: 0.0929 max mem: 3537 +Test: [1540/2387] eta: 0:02:40 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.2076 (0.3195) acc1: 99.2188 (86.8639) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1768 data: 0.0832 max mem: 3537 +Test: [1550/2387] eta: 0:02:38 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1992 (0.3188) acc1: 99.2188 (86.9385) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1810 data: 0.0875 max mem: 3537 +Test: [1560/2387] eta: 0:02:36 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.2002 (0.3180) acc1: 99.2188 (87.0129) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2129 data: 0.1184 max mem: 3537 +Test: [1570/2387] eta: 0:02:34 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1978 (0.3173) acc1: 99.7396 (87.0939) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1918 data: 0.0963 max mem: 3537 +Test: [1580/2387] eta: 0:02:32 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1953 (0.3164) acc1: 99.7396 (87.1697) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1762 data: 0.0799 max mem: 3538 +Test: [1590/2387] eta: 0:02:30 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1786 (0.3155) acc1: 99.4792 (87.2440) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1702 data: 0.0770 max mem: 3538 +Test: [1600/2387] eta: 0:02:28 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1479 (0.3146) acc1: 98.9583 (87.3082) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1655 data: 0.0733 max mem: 3538 +Test: [1610/2387] eta: 0:02:26 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1559 (0.3137) acc1: 99.4792 (87.3844) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2002 data: 0.1082 max mem: 3538 +Test: [1620/2387] eta: 0:02:24 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1799 (0.3129) acc1: 99.4792 (87.4560) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1953 data: 0.1024 max mem: 3538 +Test: [1630/2387] eta: 0:02:22 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1959 (0.3123) acc1: 98.9583 (87.5171) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1768 data: 0.0839 max mem: 3538 +Test: [1640/2387] eta: 0:02:21 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.2504 (0.3121) acc1: 96.6146 (87.5657) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1863 data: 0.0924 max mem: 3538 +Test: [1650/2387] eta: 0:02:19 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.2619 (0.3117) acc1: 96.6146 (87.6197) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1811 data: 0.0858 max mem: 3538 +Test: [1660/2387] eta: 0:02:17 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.2553 (0.3114) acc1: 96.3542 (87.6656) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1858 data: 0.0924 max mem: 3538 +Test: [1670/2387] eta: 0:02:15 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.2540 (0.3111) acc1: 96.0938 (87.7194) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1949 data: 0.1025 max mem: 3538 +Test: [1680/2387] eta: 0:02:13 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.2482 (0.3109) acc1: 96.3542 (87.7632) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1845 data: 0.0912 max mem: 3538 +Test: [1690/2387] eta: 0:02:11 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.2675 (0.3107) acc1: 96.0938 (87.8083) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2005 data: 0.1071 max mem: 3538 +Test: [1700/2387] eta: 0:02:09 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.2602 (0.3100) acc1: 97.1354 (87.8706) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1914 data: 0.0992 max mem: 3538 +Test: [1710/2387] eta: 0:02:07 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1251 (0.3089) acc1: 100.0000 (87.9408) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1693 data: 0.0765 max mem: 3538 +Test: [1720/2387] eta: 0:02:05 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1200 (0.3078) acc1: 100.0000 (88.0105) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1757 data: 0.0831 max mem: 3538 +Test: [1730/2387] eta: 0:02:03 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1165 (0.3067) acc1: 100.0000 (88.0789) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1766 data: 0.0850 max mem: 3538 +Test: [1740/2387] eta: 0:02:02 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1190 (0.3057) acc1: 100.0000 (88.1466) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1842 data: 0.0927 max mem: 3538 +Test: [1750/2387] eta: 0:02:00 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1257 (0.3047) acc1: 100.0000 (88.2143) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1885 data: 0.0957 max mem: 3538 +Test: [1760/2387] eta: 0:01:58 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1281 (0.3037) acc1: 100.0000 (88.2811) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1877 data: 0.0947 max mem: 3538 +Test: [1770/2387] eta: 0:01:56 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1480 (0.3029) acc1: 100.0000 (88.3427) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1858 data: 0.0948 max mem: 3538 +Test: [1780/2387] eta: 0:01:54 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1437 (0.3020) acc1: 100.0000 (88.4076) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.1020 max mem: 3538 +Test: [1790/2387] eta: 0:01:52 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1380 (0.3010) acc1: 100.0000 (88.4711) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1937 data: 0.0999 max mem: 3538 +Test: [1800/2387] eta: 0:01:50 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1352 (0.3001) acc1: 99.7396 (88.5336) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1870 data: 0.0954 max mem: 3538 +Test: [1810/2387] eta: 0:01:48 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1368 (0.2992) acc1: 100.0000 (88.5960) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1898 data: 0.0986 max mem: 3539 +Test: [1820/2387] eta: 0:01:46 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1324 (0.2983) acc1: 100.0000 (88.6586) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1928 data: 0.0961 max mem: 3539 +Test: [1830/2387] eta: 0:01:45 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1361 (0.2974) acc1: 100.0000 (88.7206) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2082 data: 0.1114 max mem: 3539 +Test: [1840/2387] eta: 0:01:43 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1279 (0.2965) acc1: 100.0000 (88.7819) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1959 data: 0.1048 max mem: 3539 +Test: [1850/2387] eta: 0:01:41 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1274 (0.2956) acc1: 100.0000 (88.8423) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1877 data: 0.0949 max mem: 3539 +Test: [1860/2387] eta: 0:01:39 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1270 (0.2947) acc1: 100.0000 (88.9023) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1940 data: 0.0996 max mem: 3539 +Test: [1870/2387] eta: 0:01:37 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1265 (0.2938) acc1: 100.0000 (88.9613) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1938 data: 0.1005 max mem: 3539 +Test: [1880/2387] eta: 0:01:35 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1272 (0.2929) acc1: 100.0000 (89.0199) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1899 data: 0.0935 max mem: 3539 +Test: [1890/2387] eta: 0:01:33 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1270 (0.2920) acc1: 100.0000 (89.0779) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2149 data: 0.1176 max mem: 3539 +Test: [1900/2387] eta: 0:01:31 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1270 (0.2911) acc1: 100.0000 (89.1354) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.0989 max mem: 3539 +Test: [1910/2387] eta: 0:01:30 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1550 (0.2905) acc1: 100.0000 (89.1922) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1660 data: 0.0693 max mem: 3539 +Test: [1920/2387] eta: 0:01:28 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1615 (0.2900) acc1: 100.0000 (89.2360) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1907 data: 0.0976 max mem: 3539 +Test: [1930/2387] eta: 0:01:26 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1513 (0.2892) acc1: 100.0000 (89.2895) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1848 data: 0.0942 max mem: 3539 +Test: [1940/2387] eta: 0:01:24 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1546 (0.2888) acc1: 100.0000 (89.3065) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2047 data: 0.1138 max mem: 3539 +Test: [1950/2387] eta: 0:01:22 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1684 (0.2883) acc1: 100.0000 (89.3602) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2046 data: 0.1136 max mem: 3539 +Test: [1960/2387] eta: 0:01:20 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1840 (0.2878) acc1: 100.0000 (89.4075) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2024 data: 0.1112 max mem: 3539 +Test: [1970/2387] eta: 0:01:18 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1996 (0.2876) acc1: 99.4792 (89.4454) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.2128 data: 0.1210 max mem: 3539 +Test: [1980/2387] eta: 0:01:17 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.2182 (0.2872) acc1: 98.6979 (89.4828) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2197 data: 0.1257 max mem: 3539 +Test: [1990/2387] eta: 0:01:15 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1491 (0.2865) acc1: 98.4375 (89.5285) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2130 data: 0.1196 max mem: 3539 +Test: [2000/2387] eta: 0:01:13 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1489 (0.2858) acc1: 98.4375 (89.5742) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1981 data: 0.1062 max mem: 3539 +Test: [2010/2387] eta: 0:01:11 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1496 (0.2852) acc1: 98.6979 (89.6195) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2032 data: 0.1118 max mem: 3539 +Test: [2020/2387] eta: 0:01:09 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1504 (0.2845) acc1: 98.6979 (89.6641) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1991 data: 0.1075 max mem: 3539 +Test: [2030/2387] eta: 0:01:07 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1504 (0.2838) acc1: 98.6979 (89.7073) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1898 data: 0.0977 max mem: 3540 +Test: [2040/2387] eta: 0:01:05 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1471 (0.2832) acc1: 98.6979 (89.7511) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1767 data: 0.0857 max mem: 3540 +Test: [2050/2387] eta: 0:01:03 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1463 (0.2825) acc1: 98.6979 (89.7969) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1803 data: 0.0837 max mem: 3540 +Test: [2060/2387] eta: 0:01:01 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1487 (0.2819) acc1: 99.2188 (89.8414) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1828 data: 0.0860 max mem: 3540 +Test: [2070/2387] eta: 0:00:59 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1487 (0.2812) acc1: 99.2188 (89.8871) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1603 data: 0.0689 max mem: 3540 +Test: [2080/2387] eta: 0:00:58 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1448 (0.2805) acc1: 99.2188 (89.9322) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1649 data: 0.0726 max mem: 3540 +Test: [2090/2387] eta: 0:00:56 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1448 (0.2799) acc1: 99.2188 (89.9766) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2227 data: 0.1272 max mem: 3540 +Test: [2100/2387] eta: 0:00:54 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1447 (0.2792) acc1: 99.4792 (90.0215) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2110 data: 0.1162 max mem: 3540 +Test: [2110/2387] eta: 0:00:52 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1447 (0.2786) acc1: 99.2188 (90.0652) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1628 data: 0.0694 max mem: 3540 +Test: [2120/2387] eta: 0:00:50 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1517 (0.2783) acc1: 98.9583 (90.0962) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1798 data: 0.0863 max mem: 3540 +Test: [2130/2387] eta: 0:00:48 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1684 (0.2778) acc1: 98.9583 (90.1378) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1933 data: 0.1021 max mem: 3540 +Test: [2140/2387] eta: 0:00:46 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1659 (0.2776) acc1: 98.1771 (90.1624) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1748 data: 0.0826 max mem: 3540 +Test: [2150/2387] eta: 0:00:44 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.2070 (0.2773) acc1: 98.4375 (90.2030) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1773 data: 0.0843 max mem: 3540 +Test: [2160/2387] eta: 0:00:42 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.2038 (0.2769) acc1: 98.9583 (90.2430) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1817 data: 0.0902 max mem: 3540 +Test: [2170/2387] eta: 0:00:41 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.2069 (0.2767) acc1: 98.6979 (90.2747) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1798 data: 0.0888 max mem: 3540 +Test: [2180/2387] eta: 0:00:39 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.2658 (0.2782) acc1: 96.0938 (90.1880) acc5: 100.0000 (100.0000) acc1_neg: 34.375 acc5_neg: 100.000 time: 0.1890 data: 0.0946 max mem: 3540 +Test: [2190/2387] eta: 0:00:37 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.2667 (0.2786) acc1: 91.1458 (90.1656) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1925 data: 0.0971 max mem: 3540 +Test: [2200/2387] eta: 0:00:35 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1693 (0.2780) acc1: 99.2188 (90.2092) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1759 data: 0.0825 max mem: 3540 +Test: [2210/2387] eta: 0:00:33 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1442 (0.2774) acc1: 100.0000 (90.2527) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1679 data: 0.0737 max mem: 3540 +Test: [2220/2387] eta: 0:00:31 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1573 (0.2769) acc1: 100.0000 (90.2934) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1886 data: 0.0951 max mem: 3540 +Test: [2230/2387] eta: 0:00:29 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1589 (0.2764) acc1: 100.0000 (90.3358) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1788 data: 0.0878 max mem: 3540 +Test: [2240/2387] eta: 0:00:27 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1584 (0.2758) acc1: 100.0000 (90.3760) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1732 data: 0.0821 max mem: 3540 +Test: [2250/2387] eta: 0:00:25 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1675 (0.2754) acc1: 100.0000 (90.4171) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1923 data: 0.0952 max mem: 3540 +Test: [2260/2387] eta: 0:00:23 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1811 (0.2750) acc1: 99.7396 (90.4573) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1897 data: 0.0927 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1559 (0.2744) acc1: 100.0000 (90.4991) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1862 data: 0.0947 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1321 (0.2739) acc1: 100.0000 (90.5273) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1856 data: 0.0942 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1436 (0.2734) acc1: 100.0000 (90.5682) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1786 data: 0.0837 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1534 (0.2729) acc1: 100.0000 (90.6066) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1698 data: 0.0750 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1647 (0.2725) acc1: 100.0000 (90.6431) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1899 data: 0.0975 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.1921 (0.2723) acc1: 99.2188 (90.6709) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1859 data: 0.0918 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.2015 (0.2721) acc1: 99.2188 (90.6991) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.1724 data: 0.0798 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.2004 (0.2718) acc1: 99.2188 (90.7321) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1779 data: 0.0860 max mem: 3541 +Test: [2350/2387] eta: 0:00:06 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.2125 (0.2719) acc1: 98.6979 (90.7383) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1543 data: 0.0605 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.2462 (0.2718) acc1: 98.4375 (90.7678) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1677 data: 0.0742 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.2389 (0.2716) acc1: 98.1771 (90.7973) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1813 data: 0.0880 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.2603 (0.2716) acc1: 97.1354 (90.8171) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1521 data: 0.0579 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 1.826 acc5_pos: 100.000 loss: 0.2908 (0.2723) acc1: 94.2708 (90.7851) acc5: 100.0000 (100.0000) acc1_neg: 65.116 acc5_neg: 100.000 time: 0.1286 data: 0.0330 max mem: 3541 +Test: Total time: 0:07:29 (0.1882 s / it) +* Overall Acc@1 90.785 Acc@5 100.000 + target == real Acc@1 15.227 Acc@5 100.000 + target == fake Acc@1 96.091 Acc@5 100.000 + loss 0.272 + PR-AUC 0.9892 ROC-AUC 0.8628 EER 20.45% (thr=0.7842) +Accuracy of the network on the 916267 test images: 90.8% +[rank0]:[W725 19:21:48.774531700 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 19:22:11.341000 102469 torch/distributed/run.py:766] +W0725 19:22:11.341000 102469 torch/distributed/run.py:766] ***************************************** +W0725 19:22:11.341000 102469 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 19:22:11.341000 102469 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 19:22:49.201881754 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 19:22:49.385366026 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W725 19:22:49.603416147 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W725 19:22:49.619377072 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 19:22:49.949389900 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 19:22:50.092575490 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 19:22:50.161026881 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 19:22:50.281077494 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed4_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed4_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 0:58:38 acc1_pos: 24.740 acc5_pos: 100.000 loss: 1.0875 (1.0875) acc1: 24.7396 (24.7396) acc5: 100.0000 (100.0000) time: 6.8057 data: 5.0558 max mem: 3531 +Test: [ 10/517] eta: 0:05:55 acc1_pos: 8.854 acc5_pos: 100.000 loss: 1.2275 (1.2202) acc1: 14.0625 (16.0511) acc5: 100.0000 (100.0000) time: 0.7006 data: 0.4599 max mem: 3531 +Test: [ 20/517] eta: 0:03:24 acc1_pos: 13.021 acc5_pos: 100.000 loss: 1.2052 (1.1936) acc1: 14.8438 (17.5099) acc5: 100.0000 (100.0000) time: 0.0910 data: 0.0003 max mem: 3531 +Test: [ 30/517] eta: 0:03:18 acc1_pos: 5.990 acc5_pos: 100.000 loss: 1.1958 (1.2098) acc1: 14.0625 (15.5494) acc5: 100.0000 (100.0000) time: 0.2459 data: 0.1542 max mem: 3531 +Test: [ 40/517] eta: 0:03:00 acc1_pos: 9.115 acc5_pos: 100.000 loss: 1.2567 (1.2267) acc1: 8.5938 (13.3892) acc5: 100.0000 (100.0000) time: 0.3449 data: 0.2522 max mem: 3531 +Test: [ 50/517] eta: 0:02:49 acc1_pos: 18.807 acc5_pos: 100.000 loss: 1.2184 (1.1897) acc1: 8.5938 (16.3858) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.2949 data: 0.2012 max mem: 3531 +Test: [ 60/517] eta: 0:02:36 acc1_pos: 18.807 acc5_pos: 100.000 loss: 0.3800 (1.0516) acc1: 88.8021 (28.8892) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.2699 data: 0.1753 max mem: 3531 +Test: [ 70/517] eta: 0:02:28 acc1_pos: 18.807 acc5_pos: 100.000 loss: 0.3261 (0.9470) acc1: 94.7917 (38.4610) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2499 data: 0.1537 max mem: 3531 +Test: [ 80/517] eta: 0:02:18 acc1_pos: 18.807 acc5_pos: 100.000 loss: 0.3261 (0.8732) acc1: 96.8750 (45.4765) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2399 data: 0.1441 max mem: 3531 +Test: [ 90/517] eta: 0:02:12 acc1_pos: 18.807 acc5_pos: 100.000 loss: 0.3038 (0.8098) acc1: 98.1771 (51.3164) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2349 data: 0.1396 max mem: 3531 +Test: [100/517] eta: 0:02:08 acc1_pos: 18.807 acc5_pos: 100.000 loss: 0.2878 (0.7569) acc1: 98.9583 (56.0618) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2699 data: 0.1753 max mem: 3531 +Test: [110/517] eta: 0:02:02 acc1_pos: 18.807 acc5_pos: 100.000 loss: 0.2466 (0.7096) acc1: 99.4792 (59.9428) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2599 data: 0.1663 max mem: 3531 +Test: [120/517] eta: 0:01:58 acc1_pos: 18.807 acc5_pos: 100.000 loss: 0.2599 (0.6779) acc1: 98.1771 (62.8228) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2449 data: 0.1486 max mem: 3531 +Test: [130/517] eta: 0:01:54 acc1_pos: 18.807 acc5_pos: 100.000 loss: 0.3330 (0.6516) acc1: 93.4896 (65.2135) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2649 data: 0.1653 max mem: 3531 +Test: [140/517] eta: 0:01:50 acc1_pos: 18.807 acc5_pos: 100.000 loss: 0.2610 (0.6229) acc1: 99.4792 (67.6695) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2699 data: 0.1715 max mem: 3531 +Test: [150/517] eta: 0:01:46 acc1_pos: 18.807 acc5_pos: 100.000 loss: 0.2559 (0.5984) acc1: 100.0000 (69.7417) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2549 data: 0.1595 max mem: 3531 +Test: [160/517] eta: 0:01:43 acc1_pos: 18.807 acc5_pos: 100.000 loss: 0.2587 (0.5780) acc1: 98.9583 (71.4658) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2549 data: 0.1577 max mem: 3531 +Test: [170/517] eta: 0:01:39 acc1_pos: 18.807 acc5_pos: 100.000 loss: 0.2514 (0.5589) acc1: 98.4375 (72.9578) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2549 data: 0.1559 max mem: 3531 +Test: [180/517] eta: 0:01:35 acc1_pos: 18.807 acc5_pos: 100.000 loss: 0.2322 (0.5399) acc1: 98.9583 (74.4044) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2348 data: 0.1377 max mem: 3531 +Test: [190/517] eta: 0:01:31 acc1_pos: 18.807 acc5_pos: 100.000 loss: 0.2309 (0.5237) acc1: 99.2188 (75.7090) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2398 data: 0.1436 max mem: 3531 +Test: [200/517] eta: 0:01:28 acc1_pos: 18.807 acc5_pos: 100.000 loss: 0.2316 (0.5090) acc1: 99.7396 (76.9071) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2449 data: 0.1484 max mem: 3531 +Test: [210/517] eta: 0:01:25 acc1_pos: 18.807 acc5_pos: 100.000 loss: 0.2478 (0.5013) acc1: 98.4375 (77.7930) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2449 data: 0.1508 max mem: 3532 +Test: [220/517] eta: 0:01:22 acc1_pos: 18.807 acc5_pos: 100.000 loss: 0.3488 (0.4949) acc1: 95.0521 (78.5339) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2599 data: 0.1625 max mem: 3532 +Test: [230/517] eta: 0:01:19 acc1_pos: 18.807 acc5_pos: 100.000 loss: 0.3170 (0.4863) acc1: 95.5729 (79.3865) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2499 data: 0.1501 max mem: 3532 +Test: [240/517] eta: 0:01:16 acc1_pos: 18.807 acc5_pos: 100.000 loss: 0.3163 (0.4803) acc1: 97.3958 (80.0484) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.2549 data: 0.1578 max mem: 3532 +Test: [250/517] eta: 0:01:12 acc1_pos: 18.807 acc5_pos: 100.000 loss: 0.3150 (0.4721) acc1: 97.6562 (80.7987) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2499 data: 0.1549 max mem: 3532 +Test: [260/517] eta: 0:01:10 acc1_pos: 18.807 acc5_pos: 100.000 loss: 0.2624 (0.4645) acc1: 99.7396 (81.4964) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2399 data: 0.1468 max mem: 3532 +Test: [270/517] eta: 0:01:07 acc1_pos: 18.807 acc5_pos: 100.000 loss: 0.3501 (0.4614) acc1: 96.6146 (81.9726) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2649 data: 0.1673 max mem: 3532 +Test: [280/517] eta: 0:01:04 acc1_pos: 18.807 acc5_pos: 100.000 loss: 0.3528 (0.4565) acc1: 95.8333 (82.4724) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2449 data: 0.1450 max mem: 3532 +Test: [290/517] eta: 0:01:01 acc1_pos: 18.807 acc5_pos: 100.000 loss: 0.2972 (0.4510) acc1: 97.3958 (83.0156) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2249 data: 0.1292 max mem: 3532 +Test: [300/517] eta: 0:00:58 acc1_pos: 18.807 acc5_pos: 100.000 loss: 0.2850 (0.4452) acc1: 98.9583 (83.5444) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2399 data: 0.1419 max mem: 3532 +Test: [310/517] eta: 0:00:55 acc1_pos: 18.807 acc5_pos: 100.000 loss: 0.2850 (0.4403) acc1: 99.7396 (84.0685) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2449 data: 0.1440 max mem: 3532 +Test: [320/517] eta: 0:00:52 acc1_pos: 18.807 acc5_pos: 100.000 loss: 0.2807 (0.4350) acc1: 100.0000 (84.5575) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2399 data: 0.1427 max mem: 3532 +Test: [330/517] eta: 0:00:49 acc1_pos: 18.807 acc5_pos: 100.000 loss: 0.2565 (0.4294) acc1: 99.7396 (85.0162) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2449 data: 0.1482 max mem: 3532 +Test: [340/517] eta: 0:00:47 acc1_pos: 18.807 acc5_pos: 100.000 loss: 0.2264 (0.4241) acc1: 100.0000 (85.4251) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2505 data: 0.1550 max mem: 3532 +Test: [350/517] eta: 0:00:44 acc1_pos: 18.807 acc5_pos: 100.000 loss: 0.3213 (0.4216) acc1: 98.6979 (85.7580) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2449 data: 0.1523 max mem: 3532 +Test: [360/517] eta: 0:00:41 acc1_pos: 18.807 acc5_pos: 100.000 loss: 0.3221 (0.4188) acc1: 96.8750 (86.0508) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2511 data: 0.1585 max mem: 3532 +Test: [370/517] eta: 0:00:38 acc1_pos: 18.807 acc5_pos: 100.000 loss: 0.2537 (0.4136) acc1: 99.4792 (86.4246) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2505 data: 0.1590 max mem: 3532 +Test: [380/517] eta: 0:00:36 acc1_pos: 18.807 acc5_pos: 100.000 loss: 0.2355 (0.4096) acc1: 100.0000 (86.7639) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2630 data: 0.1672 max mem: 3532 +Test: [390/517] eta: 0:00:33 acc1_pos: 18.807 acc5_pos: 100.000 loss: 0.2618 (0.4060) acc1: 99.4792 (87.0844) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2443 data: 0.1466 max mem: 3532 +Test: [400/517] eta: 0:00:30 acc1_pos: 18.807 acc5_pos: 100.000 loss: 0.2558 (0.4021) acc1: 99.4792 (87.3818) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2201 data: 0.1262 max mem: 3532 +Test: [410/517] eta: 0:00:28 acc1_pos: 18.807 acc5_pos: 100.000 loss: 0.2558 (0.3989) acc1: 99.4792 (87.6799) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2548 data: 0.1604 max mem: 3532 +Test: [420/517] eta: 0:00:25 acc1_pos: 18.807 acc5_pos: 100.000 loss: 0.2674 (0.3958) acc1: 100.0000 (87.9652) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2546 data: 0.1559 max mem: 3532 +Test: [430/517] eta: 0:00:22 acc1_pos: 18.807 acc5_pos: 100.000 loss: 0.2643 (0.3931) acc1: 99.7396 (88.2148) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2351 data: 0.1375 max mem: 3532 +Test: [440/517] eta: 0:00:20 acc1_pos: 18.807 acc5_pos: 100.000 loss: 0.2943 (0.3913) acc1: 98.6979 (88.4360) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2449 data: 0.1501 max mem: 3533 +Test: [450/517] eta: 0:00:17 acc1_pos: 18.807 acc5_pos: 100.000 loss: 0.2891 (0.3889) acc1: 98.9583 (88.6814) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2697 data: 0.1729 max mem: 3533 +Test: [460/517] eta: 0:00:14 acc1_pos: 18.807 acc5_pos: 100.000 loss: 0.2891 (0.3871) acc1: 99.4792 (88.8789) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2549 data: 0.1579 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 18.807 acc5_pos: 100.000 loss: 0.2924 (0.3849) acc1: 99.2188 (89.1001) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2349 data: 0.1381 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 18.807 acc5_pos: 100.000 loss: 0.2792 (0.3825) acc1: 99.4792 (89.3202) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2349 data: 0.1395 max mem: 3533 +Test: [490/517] eta: 0:00:07 acc1_pos: 18.807 acc5_pos: 100.000 loss: 0.2645 (0.3798) acc1: 99.7396 (89.5048) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2354 data: 0.1397 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 18.807 acc5_pos: 100.000 loss: 0.2900 (0.3789) acc1: 99.2188 (89.6712) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2499 data: 0.1527 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 18.807 acc5_pos: 100.000 loss: 0.3097 (0.3770) acc1: 98.1771 (89.8086) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2230 data: 0.1280 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 18.807 acc5_pos: 100.000 loss: 0.2482 (0.3754) acc1: 99.2188 (89.9197) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1769 data: 0.0819 max mem: 3533 +Test: Total time: 0:02:13 (0.2582 s / it) +* Overall Acc@1 89.920 Acc@5 100.000 + target == real Acc@1 12.751 Acc@5 100.000 + target == fake Acc@1 97.925 Acc@5 100.000 + loss 0.375 + PR-AUC 0.9432 ROC-AUC 0.6695 EER 38.02% (thr=0.7515) +Accuracy of the network on the 198427 test images: 89.9% +[rank0]:[W725 19:25:17.992708870 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0725 19:25:40.863000 103039 torch/distributed/run.py:766] +W0725 19:25:40.863000 103039 torch/distributed/run.py:766] ***************************************** +W0725 19:25:40.863000 103039 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 19:25:40.863000 103039 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 19:26:16.130528600 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 19:26:16.195100851 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 19:26:16.624073762 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 19:26:16.726851425 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 19:26:16.776333732 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 19:26:16.792053723 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 19:26:16.815557400 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W725 19:26:16.815589870 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed4_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed4_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:33:33 acc1_pos: 19.531 acc5_pos: 100.000 loss: 0.9405 (0.9405) acc1: 19.5312 (19.5312) acc5: 100.0000 (100.0000) time: 5.1768 data: 3.3837 max mem: 3531 +Test: [ 10/389] eta: 0:03:29 acc1_pos: 78.125 acc5_pos: 100.000 loss: 0.8101 (0.8108) acc1: 40.3646 (43.2765) acc5: 100.0000 (100.0000) time: 0.5526 data: 0.3080 max mem: 3531 +Test: [ 20/389] eta: 0:02:04 acc1_pos: 2.344 acc5_pos: 100.000 loss: 0.8042 (0.8807) acc1: 38.2812 (39.1865) acc5: 100.0000 (100.0000) time: 0.0950 data: 0.0053 max mem: 3531 +Test: [ 30/389] eta: 0:01:44 acc1_pos: 10.938 acc5_pos: 100.000 loss: 1.0155 (0.9373) acc1: 7.0312 (30.0991) acc5: 100.0000 (100.0000) time: 0.1463 data: 0.0557 max mem: 3531 +Test: [ 40/389] eta: 0:01:33 acc1_pos: 4.167 acc5_pos: 100.000 loss: 1.0127 (0.9292) acc1: 13.2812 (29.5478) acc5: 100.0000 (100.0000) time: 0.1956 data: 0.1044 max mem: 3531 +Test: [ 50/389] eta: 0:01:25 acc1_pos: 55.729 acc5_pos: 100.000 loss: 0.8883 (0.9164) acc1: 26.5625 (30.3666) acc5: 100.0000 (100.0000) time: 0.1946 data: 0.1039 max mem: 3531 +Test: [ 60/389] eta: 0:01:20 acc1_pos: 6.250 acc5_pos: 100.000 loss: 0.8136 (0.9024) acc1: 33.3333 (31.6726) acc5: 100.0000 (100.0000) time: 0.1961 data: 0.1057 max mem: 3531 +Test: [ 70/389] eta: 0:01:15 acc1_pos: 67.448 acc5_pos: 100.000 loss: 0.8647 (0.8929) acc1: 30.4688 (32.7538) acc5: 100.0000 (100.0000) time: 0.1979 data: 0.1055 max mem: 3531 +Test: [ 80/389] eta: 0:01:10 acc1_pos: 78.125 acc5_pos: 100.000 loss: 0.8450 (0.8775) acc1: 33.8542 (34.4072) acc5: 100.0000 (100.0000) time: 0.1744 data: 0.0780 max mem: 3531 +Test: [ 90/389] eta: 0:01:07 acc1_pos: 52.865 acc5_pos: 100.000 loss: 0.5722 (0.8434) acc1: 63.8021 (38.1095) acc5: 100.0000 (100.0000) time: 0.1805 data: 0.0833 max mem: 3531 +Test: [100/389] eta: 0:01:04 acc1_pos: 26.562 acc5_pos: 100.000 loss: 0.7182 (0.8477) acc1: 45.0521 (37.7553) acc5: 100.0000 (100.0000) time: 0.1996 data: 0.1064 max mem: 3531 +Test: [110/389] eta: 0:01:01 acc1_pos: 93.750 acc5_pos: 100.000 loss: 0.8692 (0.8521) acc1: 40.6250 (38.0560) acc5: 100.0000 (100.0000) time: 0.2009 data: 0.1095 max mem: 3531 +Test: [120/389] eta: 0:00:58 acc1_pos: 57.812 acc5_pos: 100.000 loss: 0.7507 (0.8408) acc1: 49.7396 (39.4370) acc5: 100.0000 (100.0000) time: 0.2028 data: 0.1110 max mem: 3531 +Test: [130/389] eta: 0:00:55 acc1_pos: 38.037 acc5_pos: 100.000 loss: 0.6060 (0.8228) acc1: 57.8125 (41.3069) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1895 data: 0.0966 max mem: 3531 +Test: [140/389] eta: 0:00:53 acc1_pos: 38.037 acc5_pos: 100.000 loss: 0.4214 (0.7968) acc1: 80.2083 (44.0972) acc5: 100.0000 (100.0000) acc1_neg: 40.625 acc5_neg: 100.000 time: 0.1849 data: 0.0926 max mem: 3531 +Test: [150/389] eta: 0:00:50 acc1_pos: 38.037 acc5_pos: 100.000 loss: 0.4193 (0.7801) acc1: 87.2396 (46.0696) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1708 data: 0.0786 max mem: 3531 +Test: [160/389] eta: 0:00:47 acc1_pos: 38.037 acc5_pos: 100.000 loss: 0.4401 (0.7562) acc1: 87.2396 (48.4294) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1798 data: 0.0862 max mem: 3531 +Test: [170/389] eta: 0:00:45 acc1_pos: 38.037 acc5_pos: 100.000 loss: 0.4041 (0.7418) acc1: 90.8854 (50.1858) acc5: 100.0000 (100.0000) acc1_neg: 59.896 acc5_neg: 100.000 time: 0.1990 data: 0.1066 max mem: 3531 +Test: [180/389] eta: 0:00:43 acc1_pos: 38.037 acc5_pos: 100.000 loss: 0.4042 (0.7243) acc1: 90.3646 (52.2272) acc5: 100.0000 (100.0000) acc1_neg: 65.885 acc5_neg: 100.000 time: 0.1868 data: 0.0940 max mem: 3531 +Test: [190/389] eta: 0:00:41 acc1_pos: 38.037 acc5_pos: 100.000 loss: 0.4104 (0.7139) acc1: 86.9792 (53.5463) acc5: 100.0000 (100.0000) acc1_neg: 66.667 acc5_neg: 100.000 time: 0.1914 data: 0.0980 max mem: 3531 +Test: [200/389] eta: 0:00:38 acc1_pos: 38.037 acc5_pos: 100.000 loss: 0.5028 (0.7043) acc1: 74.4792 (54.7005) acc5: 100.0000 (100.0000) acc1_neg: 63.802 acc5_neg: 100.000 time: 0.1980 data: 0.1060 max mem: 3531 +Test: [210/389] eta: 0:00:36 acc1_pos: 38.037 acc5_pos: 100.000 loss: 0.4196 (0.6890) acc1: 89.5833 (56.2414) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1840 data: 0.0913 max mem: 3532 +Test: [220/389] eta: 0:00:34 acc1_pos: 38.037 acc5_pos: 100.000 loss: 0.2516 (0.6694) acc1: 98.9583 (58.1318) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1949 data: 0.1012 max mem: 3532 +Test: [230/389] eta: 0:00:32 acc1_pos: 38.037 acc5_pos: 100.000 loss: 0.3563 (0.6588) acc1: 95.0521 (59.4776) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1789 data: 0.0829 max mem: 3532 +Test: [240/389] eta: 0:00:29 acc1_pos: 38.037 acc5_pos: 100.000 loss: 0.4309 (0.6526) acc1: 86.9792 (60.1617) acc5: 100.0000 (100.0000) acc1_neg: 72.917 acc5_neg: 100.000 time: 0.1503 data: 0.0551 max mem: 3532 +Test: [250/389] eta: 0:00:27 acc1_pos: 38.037 acc5_pos: 100.000 loss: 0.4041 (0.6421) acc1: 89.8438 (61.4002) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1752 data: 0.0818 max mem: 3532 +Test: [260/389] eta: 0:00:25 acc1_pos: 38.037 acc5_pos: 100.000 loss: 0.3637 (0.6339) acc1: 99.4792 (62.4441) acc5: 100.0000 (100.0000) acc1_neg: 64.062 acc5_neg: 100.000 time: 0.1860 data: 0.0922 max mem: 3532 +Test: [270/389] eta: 0:00:23 acc1_pos: 38.037 acc5_pos: 100.000 loss: 0.4366 (0.6310) acc1: 87.5000 (62.8325) acc5: 100.0000 (100.0000) acc1_neg: 71.354 acc5_neg: 100.000 time: 0.1844 data: 0.0903 max mem: 3532 +Test: [280/389] eta: 0:00:21 acc1_pos: 38.037 acc5_pos: 100.000 loss: 0.6236 (0.6302) acc1: 65.6250 (62.8753) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1931 data: 0.0982 max mem: 3532 +Test: [290/389] eta: 0:00:19 acc1_pos: 38.037 acc5_pos: 100.000 loss: 0.5679 (0.6237) acc1: 70.3125 (63.5900) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1939 data: 0.1011 max mem: 3532 +Test: [300/389] eta: 0:00:17 acc1_pos: 38.037 acc5_pos: 100.000 loss: 0.6169 (0.6242) acc1: 68.2292 (63.6368) acc5: 100.0000 (100.0000) acc1_neg: 60.677 acc5_neg: 100.000 time: 0.1598 data: 0.0674 max mem: 3532 +Test: [310/389] eta: 0:00:15 acc1_pos: 38.037 acc5_pos: 100.000 loss: 0.6427 (0.6276) acc1: 60.6771 (63.4856) acc5: 100.0000 (100.0000) acc1_neg: 48.698 acc5_neg: 100.000 time: 0.1568 data: 0.0632 max mem: 3532 +Test: [320/389] eta: 0:00:13 acc1_pos: 38.037 acc5_pos: 100.000 loss: 0.4895 (0.6225) acc1: 71.8750 (64.1104) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1740 data: 0.0811 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 38.037 acc5_pos: 100.000 loss: 0.4799 (0.6190) acc1: 93.2292 (64.8886) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.1720 data: 0.0790 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 38.037 acc5_pos: 100.000 loss: 0.4393 (0.6091) acc1: 94.7917 (65.8121) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1704 data: 0.0772 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 38.037 acc5_pos: 100.000 loss: 0.3041 (0.6120) acc1: 96.8750 (65.5597) acc5: 100.0000 (100.0000) acc1_neg: 38.802 acc5_neg: 100.000 time: 0.1699 data: 0.0780 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 38.037 acc5_pos: 100.000 loss: 0.4521 (0.6099) acc1: 83.8542 (65.7902) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1619 data: 0.0693 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 38.037 acc5_pos: 100.000 loss: 0.4366 (0.6057) acc1: 84.1146 (66.2329) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1429 data: 0.0510 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 38.037 acc5_pos: 100.000 loss: 0.4235 (0.6027) acc1: 73.9583 (66.4083) acc5: 100.0000 (100.0000) acc1_neg: 73.958 acc5_neg: 100.000 time: 0.1299 data: 0.0394 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 38.037 acc5_pos: 100.000 loss: 0.4081 (0.5988) acc1: 90.3646 (66.8896) acc5: 100.0000 (100.0000) acc1_neg: 88.525 acc5_neg: 100.000 time: 0.1363 data: 0.0326 max mem: 3532 +Test: Total time: 0:01:13 (0.1884 s / it) +* Overall Acc@1 66.890 Acc@5 100.000 + target == real Acc@1 39.238 Acc@5 100.000 + target == fake Acc@1 79.743 Acc@5 100.000 + loss 0.599 + PR-AUC 0.8188 ROC-AUC 0.6738 EER 38.41% (thr=0.6143) +Accuracy of the network on the 149358 test images: 66.9% +[rank0]:[W725 19:27:39.522039007 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 19:28:00.386000 103596 torch/distributed/run.py:766] +W0725 19:28:00.386000 103596 torch/distributed/run.py:766] ***************************************** +W0725 19:28:00.386000 103596 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 19:28:00.386000 103596 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 19:28:31.067311317 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 19:28:31.125987267 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 19:28:31.476439676 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 19:28:31.554627766 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 19:28:31.617970307 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 19:28:31.695557121 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +[rank0]:[W725 19:28:31.705522029 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 19:28:31.714855826 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed4_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed4_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:39:42 acc1_pos: 83.594 acc5_pos: 100.000 loss: 0.3834 (0.3834) acc1: 83.5938 (83.5938) acc5: 100.0000 (100.0000) time: 7.1547 data: 4.9541 max mem: 3531 +Test: [ 10/333] eta: 0:03:56 acc1_pos: 84.635 acc5_pos: 100.000 loss: 0.3112 (0.3271) acc1: 85.6771 (86.5767) acc5: 100.0000 (100.0000) time: 0.7326 data: 0.4508 max mem: 3531 +Test: [ 20/333] eta: 0:02:15 acc1_pos: 92.969 acc5_pos: 100.000 loss: 0.3042 (0.3165) acc1: 87.5000 (87.2148) acc5: 100.0000 (100.0000) time: 0.0952 data: 0.0004 max mem: 3531 +Test: [ 30/333] eta: 0:01:51 acc1_pos: 49.219 acc5_pos: 100.000 loss: 0.4096 (0.4793) acc1: 74.4792 (73.8995) acc5: 100.0000 (100.0000) time: 0.1700 data: 0.0699 max mem: 3531 +Test: [ 40/333] eta: 0:01:38 acc1_pos: 39.583 acc5_pos: 100.000 loss: 0.8735 (0.5733) acc1: 39.5833 (65.6822) acc5: 100.0000 (100.0000) time: 0.2369 data: 0.1414 max mem: 3531 +Test: [ 50/333] eta: 0:01:26 acc1_pos: 44.271 acc5_pos: 100.000 loss: 0.8788 (0.6301) acc1: 41.1458 (61.2081) acc5: 100.0000 (100.0000) time: 0.2099 data: 0.1169 max mem: 3531 +Test: [ 60/333] eta: 0:01:20 acc1_pos: 28.906 acc5_pos: 100.000 loss: 0.8750 (0.6717) acc1: 42.7083 (57.7826) acc5: 100.0000 (100.0000) time: 0.2137 data: 0.1189 max mem: 3531 +Test: [ 70/333] eta: 0:01:14 acc1_pos: 38.281 acc5_pos: 100.000 loss: 0.8785 (0.6987) acc1: 42.7083 (55.6375) acc5: 100.0000 (100.0000) time: 0.2262 data: 0.1341 max mem: 3531 +Test: [ 80/333] eta: 0:01:08 acc1_pos: 48.698 acc5_pos: 100.000 loss: 0.8013 (0.7147) acc1: 45.3125 (54.2856) acc5: 100.0000 (100.0000) time: 0.1941 data: 0.1018 max mem: 3531 +Test: [ 90/333] eta: 0:01:03 acc1_pos: 55.990 acc5_pos: 100.000 loss: 0.7851 (0.7198) acc1: 47.9167 (53.8261) acc5: 100.0000 (100.0000) time: 0.1786 data: 0.0831 max mem: 3531 +Test: [100/333] eta: 0:00:58 acc1_pos: 55.208 acc5_pos: 100.000 loss: 0.7598 (0.7277) acc1: 50.2604 (53.0863) acc5: 100.0000 (100.0000) time: 0.1717 data: 0.0780 max mem: 3531 +Test: [110/333] eta: 0:00:54 acc1_pos: 53.125 acc5_pos: 100.000 loss: 0.7630 (0.7334) acc1: 48.6979 (52.7426) acc5: 100.0000 (100.0000) time: 0.1749 data: 0.0838 max mem: 3531 +Test: [120/333] eta: 0:00:51 acc1_pos: 49.740 acc5_pos: 100.000 loss: 0.7442 (0.7362) acc1: 49.7396 (52.4750) acc5: 100.0000 (100.0000) time: 0.1835 data: 0.0901 max mem: 3531 +Test: [130/333] eta: 0:00:48 acc1_pos: 64.583 acc5_pos: 100.000 loss: 0.7356 (0.7364) acc1: 51.3021 (52.5763) acc5: 100.0000 (100.0000) time: 0.1999 data: 0.1038 max mem: 3531 +Test: [140/333] eta: 0:00:45 acc1_pos: 70.052 acc5_pos: 100.000 loss: 0.7356 (0.7385) acc1: 50.0000 (52.3696) acc5: 100.0000 (100.0000) time: 0.1934 data: 0.0976 max mem: 3531 +Test: [150/333] eta: 0:00:41 acc1_pos: 57.552 acc5_pos: 100.000 loss: 0.7508 (0.7426) acc1: 48.1771 (51.9919) acc5: 100.0000 (100.0000) time: 0.1688 data: 0.0755 max mem: 3531 +Test: [160/333] eta: 0:00:39 acc1_pos: 42.763 acc5_pos: 100.000 loss: 0.8053 (0.7480) acc1: 47.9167 (51.7760) acc5: 100.0000 (100.0000) acc1_neg: 52.604 acc5_neg: 100.000 time: 0.1786 data: 0.0877 max mem: 3531 +Test: [170/333] eta: 0:00:36 acc1_pos: 42.763 acc5_pos: 100.000 loss: 0.7877 (0.7516) acc1: 50.5208 (51.8960) acc5: 100.0000 (100.0000) acc1_neg: 57.552 acc5_neg: 100.000 time: 0.2079 data: 0.1148 max mem: 3531 +Test: [180/333] eta: 0:00:33 acc1_pos: 42.763 acc5_pos: 100.000 loss: 0.7532 (0.7504) acc1: 56.2500 (52.2804) acc5: 100.0000 (100.0000) acc1_neg: 60.417 acc5_neg: 100.000 time: 0.1854 data: 0.0891 max mem: 3531 +Test: [190/333] eta: 0:00:31 acc1_pos: 42.763 acc5_pos: 100.000 loss: 0.6666 (0.7443) acc1: 60.1562 (53.0377) acc5: 100.0000 (100.0000) acc1_neg: 65.885 acc5_neg: 100.000 time: 0.1773 data: 0.0817 max mem: 3531 +Test: [200/333] eta: 0:00:29 acc1_pos: 42.763 acc5_pos: 100.000 loss: 0.6317 (0.7402) acc1: 65.6250 (53.7106) acc5: 100.0000 (100.0000) acc1_neg: 64.583 acc5_neg: 100.000 time: 0.1849 data: 0.0917 max mem: 3531 +Test: [210/333] eta: 0:00:26 acc1_pos: 42.763 acc5_pos: 100.000 loss: 0.7183 (0.7414) acc1: 61.4583 (54.0247) acc5: 100.0000 (100.0000) acc1_neg: 53.906 acc5_neg: 100.000 time: 0.1887 data: 0.0977 max mem: 3532 +Test: [220/333] eta: 0:00:24 acc1_pos: 42.763 acc5_pos: 100.000 loss: 0.7542 (0.7441) acc1: 59.6354 (54.0571) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.1936 data: 0.1009 max mem: 3532 +Test: [230/333] eta: 0:00:22 acc1_pos: 42.763 acc5_pos: 100.000 loss: 0.5336 (0.7339) acc1: 73.4375 (55.1136) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.1697 data: 0.0738 max mem: 3532 +Test: [240/333] eta: 0:00:19 acc1_pos: 42.763 acc5_pos: 100.000 loss: 0.5316 (0.7274) acc1: 75.2604 (55.7983) acc5: 100.0000 (100.0000) acc1_neg: 71.354 acc5_neg: 100.000 time: 0.1757 data: 0.0824 max mem: 3532 +Test: [250/333] eta: 0:00:17 acc1_pos: 42.763 acc5_pos: 100.000 loss: 0.5958 (0.7230) acc1: 69.5312 (56.2884) acc5: 100.0000 (100.0000) acc1_neg: 59.375 acc5_neg: 100.000 time: 0.1949 data: 0.1027 max mem: 3532 +Test: [260/333] eta: 0:00:15 acc1_pos: 42.763 acc5_pos: 100.000 loss: 0.6198 (0.7199) acc1: 67.1875 (56.6790) acc5: 100.0000 (100.0000) acc1_neg: 70.573 acc5_neg: 100.000 time: 0.2012 data: 0.1086 max mem: 3532 +Test: [270/333] eta: 0:00:13 acc1_pos: 42.763 acc5_pos: 100.000 loss: 0.6198 (0.7160) acc1: 65.8854 (57.0937) acc5: 100.0000 (100.0000) acc1_neg: 65.365 acc5_neg: 100.000 time: 0.1840 data: 0.0932 max mem: 3532 +Test: [280/333] eta: 0:00:11 acc1_pos: 42.763 acc5_pos: 100.000 loss: 0.6356 (0.7163) acc1: 65.3646 (57.1174) acc5: 100.0000 (100.0000) acc1_neg: 63.021 acc5_neg: 100.000 time: 0.1628 data: 0.0698 max mem: 3532 +Test: [290/333] eta: 0:00:08 acc1_pos: 42.763 acc5_pos: 100.000 loss: 0.8065 (0.7197) acc1: 51.3021 (56.8997) acc5: 100.0000 (100.0000) acc1_neg: 53.646 acc5_neg: 100.000 time: 0.1599 data: 0.0644 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 42.763 acc5_pos: 100.000 loss: 0.8157 (0.7227) acc1: 51.5625 (56.8089) acc5: 100.0000 (100.0000) acc1_neg: 56.510 acc5_neg: 100.000 time: 0.1617 data: 0.0675 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 42.763 acc5_pos: 100.000 loss: 0.7115 (0.7178) acc1: 59.8958 (57.3252) acc5: 100.0000 (100.0000) acc1_neg: 74.740 acc5_neg: 100.000 time: 0.1504 data: 0.0561 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 42.763 acc5_pos: 100.000 loss: 0.5264 (0.7134) acc1: 74.7396 (57.7955) acc5: 100.0000 (100.0000) acc1_neg: 48.698 acc5_neg: 100.000 time: 0.1502 data: 0.0560 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 42.763 acc5_pos: 100.000 loss: 0.8129 (0.7182) acc1: 52.8646 (57.5718) acc5: 100.0000 (100.0000) acc1_neg: 52.865 acc5_neg: 100.000 time: 0.1467 data: 0.0552 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 42.763 acc5_pos: 100.000 loss: 0.8533 (0.7190) acc1: 52.8646 (57.5511) acc5: 100.0000 (100.0000) acc1_neg: 53.803 acc5_neg: 100.000 time: 0.1535 data: 0.0552 max mem: 3532 +Test: Total time: 0:01:06 (0.2005 s / it) +* Overall Acc@1 57.551 Acc@5 100.000 + target == real Acc@1 51.850 Acc@5 100.000 + target == fake Acc@1 62.626 Acc@5 100.000 + loss 0.719 + PR-AUC 0.6163 ROC-AUC 0.6037 EER 42.91% (thr=0.5400) +Accuracy of the network on the 127843 test images: 57.6% +[rank0]:[W725 19:29:50.879891404 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 19:30:09.410000 104154 torch/distributed/run.py:766] +W0725 19:30:09.410000 104154 torch/distributed/run.py:766] ***************************************** +W0725 19:30:09.410000 104154 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 19:30:09.410000 104154 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 19:30:37.525906496 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 19:30:37.881779461 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 19:30:38.328856682 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 19:30:38.916119751 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 19:30:38.942389157 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 19:30:38.961485439 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W725 19:30:38.961738250 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +NCCL version 2.26.2+cuda12.2 +[rank4]:[W725 19:30:39.169664745 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed4_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed4_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:24:02 acc1_pos: 97.656 acc5_pos: 100.000 loss: 0.1629 (0.1629) acc1: 97.6562 (97.6562) acc5: 100.0000 (100.0000) time: 6.1381 data: 4.7596 max mem: 3531 +Test: [ 10/235] eta: 0:02:24 acc1_pos: 96.094 acc5_pos: 100.000 loss: 0.1631 (0.1645) acc1: 97.3958 (97.1117) acc5: 100.0000 (100.0000) time: 0.6422 data: 0.4331 max mem: 3531 +Test: [ 20/235] eta: 0:01:22 acc1_pos: 97.917 acc5_pos: 100.000 loss: 0.1623 (0.1607) acc1: 97.3958 (97.2594) acc5: 100.0000 (100.0000) time: 0.0949 data: 0.0005 max mem: 3531 +Test: [ 30/235] eta: 0:01:11 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.1557 (0.1599) acc1: 97.6562 (97.3286) acc5: 100.0000 (100.0000) time: 0.1896 data: 0.0955 max mem: 3531 +Test: [ 40/235] eta: 0:01:04 acc1_pos: 96.094 acc5_pos: 100.000 loss: 0.1533 (0.1600) acc1: 97.6562 (97.3577) acc5: 100.0000 (100.0000) time: 0.2798 data: 0.1869 max mem: 3531 +Test: [ 50/235] eta: 0:00:59 acc1_pos: 96.094 acc5_pos: 100.000 loss: 0.1608 (0.1613) acc1: 97.1354 (97.2937) acc5: 100.0000 (100.0000) time: 0.2738 data: 0.1776 max mem: 3531 +Test: [ 60/235] eta: 0:00:57 acc1_pos: 97.135 acc5_pos: 100.000 loss: 0.1647 (0.1625) acc1: 96.8750 (97.2251) acc5: 100.0000 (100.0000) time: 0.3242 data: 0.2298 max mem: 3531 +Test: [ 70/235] eta: 0:00:51 acc1_pos: 98.462 acc5_pos: 100.000 loss: 0.1752 (0.3576) acc1: 95.8333 (86.8068) acc5: 100.0000 (100.0000) acc1_neg: 13.542 acc5_neg: 100.000 time: 0.2855 data: 0.1939 max mem: 3531 +Test: [ 80/235] eta: 0:00:44 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.7170 (0.5262) acc1: 13.5417 (77.6749) acc5: 100.0000 (100.0000) acc1_neg: 10.417 acc5_neg: 100.000 time: 0.1656 data: 0.0709 max mem: 3531 +Test: [ 90/235] eta: 0:00:41 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.7173 (0.6599) acc1: 12.5000 (70.5701) acc5: 100.0000 (100.0000) acc1_neg: 12.240 acc5_neg: 100.000 time: 0.1894 data: 0.0957 max mem: 3531 +Test: [100/235] eta: 0:00:37 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.7318 (0.7647) acc1: 12.5000 (64.9314) acc5: 100.0000 (100.0000) acc1_neg: 12.240 acc5_neg: 100.000 time: 0.2254 data: 0.1352 max mem: 3531 +Test: [110/235] eta: 0:00:34 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.7151 (0.8505) acc1: 13.2812 (60.3627) acc5: 100.0000 (100.0000) acc1_neg: 16.146 acc5_neg: 100.000 time: 0.2351 data: 0.1444 max mem: 3531 +Test: [120/235] eta: 0:00:30 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.7621 (0.9409) acc1: 11.1979 (55.8346) acc5: 100.0000 (100.0000) acc1_neg: 3.906 acc5_neg: 100.000 time: 0.2308 data: 0.1396 max mem: 3531 +Test: [130/235] eta: 0:00:27 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.9470 (1.0183) acc1: 4.9479 (51.9998) acc5: 100.0000 (100.0000) acc1_neg: 4.688 acc5_neg: 100.000 time: 0.2102 data: 0.1193 max mem: 3531 +Test: [140/235] eta: 0:00:24 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.9753 (1.0870) acc1: 4.9479 (48.6610) acc5: 100.0000 (100.0000) acc1_neg: 5.729 acc5_neg: 100.000 time: 0.2244 data: 0.1324 max mem: 3531 +Test: [150/235] eta: 0:00:21 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.9754 (1.1446) acc1: 4.6875 (45.7454) acc5: 100.0000 (100.0000) acc1_neg: 5.990 acc5_neg: 100.000 time: 0.2080 data: 0.1133 max mem: 3531 +Test: [160/235] eta: 0:00:19 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.9483 (1.1944) acc1: 5.2083 (43.2712) acc5: 100.0000 (100.0000) acc1_neg: 6.250 acc5_neg: 100.000 time: 0.2092 data: 0.1152 max mem: 3531 +Test: [170/235] eta: 0:00:16 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.8886 (1.2195) acc1: 6.2500 (41.7763) acc5: 100.0000 (100.0000) acc1_neg: 17.708 acc5_neg: 100.000 time: 0.2149 data: 0.1222 max mem: 3531 +Test: [180/235] eta: 0:00:13 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.5864 (1.2390) acc1: 18.4896 (40.5401) acc5: 100.0000 (100.0000) acc1_neg: 17.708 acc5_neg: 100.000 time: 0.1865 data: 0.0927 max mem: 3531 +Test: [190/235] eta: 0:00:10 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.5562 (1.2554) acc1: 18.4896 (39.3911) acc5: 100.0000 (100.0000) acc1_neg: 20.052 acc5_neg: 100.000 time: 0.1702 data: 0.0777 max mem: 3531 +Test: [200/235] eta: 0:00:08 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.5624 (1.2718) acc1: 17.9688 (38.3085) acc5: 100.0000 (100.0000) acc1_neg: 16.667 acc5_neg: 100.000 time: 0.1545 data: 0.0635 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.5675 (1.2858) acc1: 17.9688 (37.3815) acc5: 100.0000 (100.0000) acc1_neg: 18.229 acc5_neg: 100.000 time: 0.1405 data: 0.0492 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.5614 (1.2982) acc1: 18.7500 (36.5349) acc5: 100.0000 (100.0000) acc1_neg: 18.750 acc5_neg: 100.000 time: 0.1324 data: 0.0409 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.5651 (1.3098) acc1: 19.2708 (35.8462) acc5: 100.0000 (100.0000) acc1_neg: 17.448 acc5_neg: 100.000 time: 0.1248 data: 0.0340 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.5651 (1.3158) acc1: 19.2708 (35.6268) acc5: 100.0000 (100.0000) acc1_neg: 17.391 acc5_neg: 100.000 time: 0.1101 data: 0.0234 max mem: 3532 +Test: Total time: 0:00:52 (0.2228 s / it) +* Overall Acc@1 35.627 Acc@5 100.000 + target == real Acc@1 97.231 Acc@5 100.000 + target == fake Acc@1 13.346 Acc@5 100.000 + loss 1.316 + PR-AUC 0.8543 ROC-AUC 0.6885 EER 36.72% (thr=0.1221) +Accuracy of the network on the 89879 test images: 35.6% +[rank0]:[W725 19:31:41.423052584 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 19:31:59.950000 104711 torch/distributed/run.py:766] +W0725 19:31:59.950000 104711 torch/distributed/run.py:766] ***************************************** +W0725 19:31:59.950000 104711 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 19:31:59.950000 104711 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 19:32:28.783742830 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 19:32:29.061426598 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 19:32:29.301306094 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 19:32:29.301740674 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 19:32:29.332830417 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 19:32:29.355745568 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 1): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 19:32:29.581351781 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W725 19:32:29.581872382 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed4_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed4_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:25:57 acc1_pos: 35.677 acc5_pos: 100.000 loss: 0.8545 (0.8545) acc1: 35.6771 (35.6771) acc5: 100.0000 (100.0000) time: 5.7899 data: 4.3591 max mem: 3531 +Test: [ 10/269] eta: 0:02:37 acc1_pos: 70.312 acc5_pos: 100.000 loss: 0.6655 (0.6662) acc1: 60.6771 (59.9432) acc5: 100.0000 (100.0000) time: 0.6084 data: 0.3967 max mem: 3531 +Test: [ 20/269] eta: 0:01:30 acc1_pos: 70.573 acc5_pos: 100.000 loss: 0.6697 (0.6722) acc1: 56.2500 (59.3626) acc5: 100.0000 (100.0000) time: 0.0932 data: 0.0004 max mem: 3531 +Test: [ 30/269] eta: 0:01:14 acc1_pos: 53.385 acc5_pos: 100.000 loss: 0.6904 (0.6754) acc1: 55.7292 (58.9298) acc5: 100.0000 (100.0000) time: 0.1480 data: 0.0522 max mem: 3531 +Test: [ 40/269] eta: 0:01:10 acc1_pos: 69.828 acc5_pos: 100.000 loss: 0.5814 (0.6413) acc1: 70.5729 (63.1606) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.2499 data: 0.1563 max mem: 3531 +Test: [ 50/269] eta: 0:01:03 acc1_pos: 69.828 acc5_pos: 100.000 loss: 0.6022 (0.6450) acc1: 70.3125 (62.8932) acc5: 100.0000 (100.0000) acc1_neg: 39.844 acc5_neg: 100.000 time: 0.2599 data: 0.1657 max mem: 3531 +Test: [ 60/269] eta: 0:00:59 acc1_pos: 69.828 acc5_pos: 100.000 loss: 0.6654 (0.6564) acc1: 59.6354 (61.7743) acc5: 100.0000 (100.0000) acc1_neg: 54.167 acc5_neg: 100.000 time: 0.2349 data: 0.1379 max mem: 3531 +Test: [ 70/269] eta: 0:00:54 acc1_pos: 69.828 acc5_pos: 100.000 loss: 0.6991 (0.6572) acc1: 54.9479 (61.6381) acc5: 100.0000 (100.0000) acc1_neg: 68.750 acc5_neg: 100.000 time: 0.2249 data: 0.1283 max mem: 3531 +Test: [ 80/269] eta: 0:00:49 acc1_pos: 69.828 acc5_pos: 100.000 loss: 0.6228 (0.6525) acc1: 65.3646 (62.1110) acc5: 100.0000 (100.0000) acc1_neg: 68.229 acc5_neg: 100.000 time: 0.1949 data: 0.0990 max mem: 3531 +Test: [ 90/269] eta: 0:00:43 acc1_pos: 69.828 acc5_pos: 100.000 loss: 0.6466 (0.6659) acc1: 60.4167 (60.4024) acc5: 100.0000 (100.0000) acc1_neg: 53.906 acc5_neg: 100.000 time: 0.1460 data: 0.0490 max mem: 3531 +Test: [100/269] eta: 0:00:41 acc1_pos: 69.828 acc5_pos: 100.000 loss: 0.7588 (0.6745) acc1: 48.4375 (59.6148) acc5: 100.0000 (100.0000) acc1_neg: 44.792 acc5_neg: 100.000 time: 0.1749 data: 0.0766 max mem: 3531 +Test: [110/269] eta: 0:00:38 acc1_pos: 69.828 acc5_pos: 100.000 loss: 0.7958 (0.6853) acc1: 47.1354 (58.4272) acc5: 100.0000 (100.0000) acc1_neg: 39.062 acc5_neg: 100.000 time: 0.2388 data: 0.1423 max mem: 3531 +Test: [120/269] eta: 0:00:35 acc1_pos: 69.828 acc5_pos: 100.000 loss: 0.7294 (0.6858) acc1: 48.9583 (58.3441) acc5: 100.0000 (100.0000) acc1_neg: 48.698 acc5_neg: 100.000 time: 0.2049 data: 0.1105 max mem: 3531 +Test: [130/269] eta: 0:00:32 acc1_pos: 69.828 acc5_pos: 100.000 loss: 0.7294 (0.6935) acc1: 48.4375 (57.4606) acc5: 100.0000 (100.0000) acc1_neg: 39.844 acc5_neg: 100.000 time: 0.1699 data: 0.0742 max mem: 3531 +Test: [140/269] eta: 0:00:29 acc1_pos: 69.828 acc5_pos: 100.000 loss: 0.7669 (0.6964) acc1: 48.4375 (57.1642) acc5: 100.0000 (100.0000) acc1_neg: 62.240 acc5_neg: 100.000 time: 0.1801 data: 0.0831 max mem: 3531 +Test: [150/269] eta: 0:00:27 acc1_pos: 69.828 acc5_pos: 100.000 loss: 0.6547 (0.6867) acc1: 59.1146 (58.0350) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.1910 data: 0.0970 max mem: 3531 +Test: [160/269] eta: 0:00:24 acc1_pos: 69.828 acc5_pos: 100.000 loss: 0.5774 (0.6809) acc1: 67.4479 (58.5840) acc5: 100.0000 (100.0000) acc1_neg: 54.167 acc5_neg: 100.000 time: 0.1917 data: 0.0998 max mem: 3531 +Test: [170/269] eta: 0:00:22 acc1_pos: 69.828 acc5_pos: 100.000 loss: 0.7110 (0.6856) acc1: 51.5625 (58.1049) acc5: 100.0000 (100.0000) acc1_neg: 50.000 acc5_neg: 100.000 time: 0.2038 data: 0.1129 max mem: 3531 +Test: [180/269] eta: 0:00:19 acc1_pos: 69.828 acc5_pos: 100.000 loss: 0.7058 (0.6849) acc1: 51.3021 (58.2125) acc5: 100.0000 (100.0000) acc1_neg: 41.146 acc5_neg: 100.000 time: 0.2050 data: 0.1146 max mem: 3531 +Test: [190/269] eta: 0:00:17 acc1_pos: 69.828 acc5_pos: 100.000 loss: 0.6722 (0.6871) acc1: 58.5938 (57.9216) acc5: 100.0000 (100.0000) acc1_neg: 46.094 acc5_neg: 100.000 time: 0.2057 data: 0.1139 max mem: 3531 +Test: [200/269] eta: 0:00:15 acc1_pos: 69.828 acc5_pos: 100.000 loss: 0.7629 (0.6927) acc1: 51.3021 (57.4173) acc5: 100.0000 (100.0000) acc1_neg: 58.594 acc5_neg: 100.000 time: 0.2129 data: 0.1190 max mem: 3531 +Test: [210/269] eta: 0:00:12 acc1_pos: 69.828 acc5_pos: 100.000 loss: 0.6845 (0.6874) acc1: 56.2500 (57.9088) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.1996 data: 0.1038 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 69.828 acc5_pos: 100.000 loss: 0.5430 (0.6804) acc1: 71.3542 (58.6621) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2099 data: 0.1147 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 69.828 acc5_pos: 100.000 loss: 0.4474 (0.6715) acc1: 82.0312 (59.5971) acc5: 100.0000 (100.0000) acc1_neg: 72.135 acc5_neg: 100.000 time: 0.2214 data: 0.1288 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 69.828 acc5_pos: 100.000 loss: 0.6049 (0.6762) acc1: 58.5938 (59.0768) acc5: 100.0000 (100.0000) acc1_neg: 48.958 acc5_neg: 100.000 time: 0.1837 data: 0.0930 max mem: 3532 +Test: [250/269] eta: 0:00:04 acc1_pos: 69.828 acc5_pos: 100.000 loss: 0.7634 (0.6778) acc1: 48.4375 (58.9123) acc5: 100.0000 (100.0000) acc1_neg: 42.708 acc5_neg: 100.000 time: 0.1436 data: 0.0539 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 69.828 acc5_pos: 100.000 loss: 0.7398 (0.6823) acc1: 48.4375 (58.4591) acc5: 100.0000 (100.0000) acc1_neg: 63.542 acc5_neg: 100.000 time: 0.1355 data: 0.0459 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 69.828 acc5_pos: 100.000 loss: 0.7441 (0.6829) acc1: 48.4375 (58.4454) acc5: 100.0000 (100.0000) acc1_neg: 69.554 acc5_neg: 100.000 time: 0.1275 data: 0.0344 max mem: 3532 +Test: Total time: 0:00:56 (0.2089 s / it) +* Overall Acc@1 58.445 Acc@5 100.000 + target == real Acc@1 58.763 Acc@5 100.000 + target == fake Acc@1 58.402 Acc@5 100.000 + loss 0.683 + PR-AUC 0.9195 ROC-AUC 0.6222 EER 41.40% (thr=0.4988) +Accuracy of the network on the 103293 test images: 58.4% +[rank0]:[W725 19:33:33.533860656 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-25T19:33:40Z] eval_done tag=fixed500k_fake_n32_seed4_resampled +[2026-07-25T19:33:40Z] eval_skip tag=fixed500k_fake_n45_seed1_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed1_resampled +[2026-07-25T19:33:40Z] eval_skip tag=fixed500k_fake_n45_seed2_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed2_resampled +[2026-07-25T19:33:40Z] eval_skip tag=fixed500k_fake_n45_seed3_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed3_resampled +[2026-07-25T19:33:40Z] eval_wait tag=fixed500k_fake_n45_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed4_resampled +[2026-07-25T19:34:41Z] eval_wait tag=fixed500k_fake_n45_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed4_resampled +[2026-07-25T19:35:41Z] eval_wait tag=fixed500k_fake_n45_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed4_resampled +[2026-07-25T19:36:41Z] eval_wait tag=fixed500k_fake_n45_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed4_resampled +[2026-07-25T19:37:41Z] eval_wait tag=fixed500k_fake_n45_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed4_resampled +[2026-07-25T19:38:41Z] eval_skip tag=fixed500k_fake_n45_seed4_resampled reason=complete_after_lock output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n45_seed4_resampled +[2026-07-25T19:38:41Z] eval_skip tag=fixed500k_fake_n64_seed1_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed1_resampled +[2026-07-25T19:38:41Z] eval_skip tag=fixed500k_fake_n64_seed2_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed2_resampled +[2026-07-25T19:38:41Z] eval_skip tag=fixed500k_fake_n64_seed3_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed3_resampled +[2026-07-25T19:38:41Z] eval_wait tag=fixed500k_fake_n64_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed4_resampled +[2026-07-25T19:39:42Z] eval_wait tag=fixed500k_fake_n64_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed4_resampled +[2026-07-25T19:40:42Z] eval_wait tag=fixed500k_fake_n64_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed4_resampled +[2026-07-25T19:41:42Z] eval_wait tag=fixed500k_fake_n64_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed4_resampled +[2026-07-25T19:42:42Z] eval_wait tag=fixed500k_fake_n64_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed4_resampled +[2026-07-25T19:43:42Z] eval_wait tag=fixed500k_fake_n64_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed4_resampled +[2026-07-25T19:44:42Z] eval_skip tag=fixed500k_fake_n64_seed4_resampled reason=complete_after_lock output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n64_seed4_resampled +[2026-07-25T19:44:42Z] eval_skip tag=fixed500k_real_n11_seed1_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed1_resampled +[2026-07-25T19:44:42Z] eval_skip tag=fixed500k_real_n11_seed2_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed2_resampled +[2026-07-25T19:44:42Z] eval_skip tag=fixed500k_real_n11_seed3_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed3_resampled +[2026-07-25T19:44:42Z] eval_skip tag=fixed500k_real_n11_seed4_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n11_seed4_resampled +[2026-07-25T19:44:43Z] eval_skip tag=fixed500k_real_n16_seed1_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed1_resampled +[2026-07-25T19:44:43Z] eval_skip tag=fixed500k_real_n16_seed2_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed2_resampled +[2026-07-25T19:44:43Z] eval_skip tag=fixed500k_real_n16_seed3_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed3_resampled +[2026-07-25T19:44:43Z] eval_skip tag=fixed500k_real_n16_seed4_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed4_resampled +[2026-07-25T19:44:43Z] eval_skip tag=fixed500k_real_n23_seed1_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed1_resampled +[2026-07-25T19:44:43Z] eval_skip tag=fixed500k_real_n23_seed2_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed2_resampled +[2026-07-25T19:44:43Z] eval_skip tag=fixed500k_real_n23_seed3_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed3_resampled +[2026-07-25T19:44:43Z] eval_skip tag=fixed500k_real_n23_seed4_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n23_seed4_resampled +[2026-07-25T19:44:43Z] eval_skip tag=fixed500k_real_n32_seed1_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed1_resampled +[2026-07-25T19:44:43Z] eval_skip tag=fixed500k_real_n32_seed2_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed2_resampled +[2026-07-25T19:44:43Z] eval_skip tag=fixed500k_real_n32_seed3_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed3_resampled +[2026-07-25T19:44:43Z] eval_skip tag=fixed500k_real_n32_seed4_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n32_seed4_resampled +[2026-07-25T19:44:43Z] eval_skip tag=fixed500k_real_n5_seed1_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed1_resampled +[2026-07-25T19:44:43Z] eval_skip tag=fixed500k_real_n5_seed2_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed2_resampled +[2026-07-25T19:44:43Z] eval_skip tag=fixed500k_real_n5_seed3_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed3_resampled +[2026-07-25T19:44:43Z] eval_skip tag=fixed500k_real_n5_seed4_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed4_resampled +[2026-07-25T19:44:43Z] eval_skip tag=fixed500k_real_n8_seed1_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed1_resampled +[2026-07-25T19:44:43Z] eval_skip tag=fixed500k_real_n8_seed2_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed2_resampled +[2026-07-25T19:44:43Z] eval_skip tag=fixed500k_real_n8_seed3_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed3_resampled +[2026-07-25T19:44:43Z] eval_skip tag=fixed500k_real_n8_seed4_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n8_seed4_resampled +[2026-07-25T19:44:44Z] early_eval_watch_done eval_complete=48/48 diff --git a/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_early_20260725_112559_anchor/early_eval_watch.pid b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_early_20260725_112559_anchor/early_eval_watch.pid new file mode 100644 index 0000000000000000000000000000000000000000..3177a6616435795199510991bc41854c75893924 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_early_20260725_112559_anchor/early_eval_watch.pid @@ -0,0 +1 @@ +64131 diff --git a/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_host_done_20260725_112559/eval_queue_n111-046-081.log b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_host_done_20260725_112559/eval_queue_n111-046-081.log new file mode 100644 index 0000000000000000000000000000000000000000..b169fa613aaf7177af172720de8b4a40c6d36411 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_host_done_20260725_112559/eval_queue_n111-046-081.log @@ -0,0 +1,2179 @@ +eval_queue_started_utc=2026-07-25T16:41:49Z +hostname=n111-046-081 +train_tag=20260725_112559 +eval_tag=20260725_112559 +real_level=16 +fake_level=32 +eval_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559 +batch_per_gpu=256 +[2026-07-25T16:41:49Z] eval_start tag=fixed500k_real_n16_seed1_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed1_resampled_n111-046-081_20260725_112559/best_checkpoint.pth +started_utc=2026-07-25T16:41:49Z +hostname=n111-046-081 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed1_resampled_n111-046-081_20260725_112559/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed1_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0725 16:42:02.383000 2557722 torch/distributed/run.py:766] +W0725 16:42:02.383000 2557722 torch/distributed/run.py:766] ***************************************** +W0725 16:42:02.383000 2557722 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 16:42:02.383000 2557722 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 16:42:31.909452613 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 16:42:31.024326604 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 16:42:31.421341808 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 16:42:31.558044236 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 16:42:32.594691821 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 16:42:32.643710548 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 16:42:32.714914272 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 16:42:32.731324444 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed1_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed1_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:17:54 acc1_pos: 90.365 acc5_pos: 100.000 loss: 0.3576 (0.3576) acc1: 90.3646 (90.3646) acc5: 100.0000 (100.0000) time: 4.5920 data: 2.7642 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/234] eta: 0:02:02 acc1_pos: 77.604 acc5_pos: 100.000 loss: 0.4725 (0.5462) acc1: 81.5104 (72.8693) acc5: 100.0000 (100.0000) time: 0.5465 data: 0.2986 max mem: 3531 +Test: [ 20/234] eta: 0:01:15 acc1_pos: 73.438 acc5_pos: 100.000 loss: 0.4725 (0.5348) acc1: 79.1667 (74.1940) acc5: 100.0000 (100.0000) time: 0.1425 data: 0.0518 max mem: 3531 +Test: [ 30/234] eta: 0:00:55 acc1_pos: 38.021 acc5_pos: 100.000 loss: 0.4607 (0.5320) acc1: 78.1250 (73.9163) acc5: 100.0000 (100.0000) time: 0.1215 data: 0.0260 max mem: 3531 +Test: [ 40/234] eta: 0:00:50 acc1_pos: 96.354 acc5_pos: 100.000 loss: 0.4568 (0.5644) acc1: 76.3021 (70.9858) acc5: 100.0000 (100.0000) time: 0.1656 data: 0.0707 max mem: 3531 +Test: [ 50/234] eta: 0:00:46 acc1_pos: 97.396 acc5_pos: 100.000 loss: 0.1906 (0.4888) acc1: 95.0521 (75.8068) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1310 max mem: 3531 +Test: [ 60/234] eta: 0:00:42 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.1242 (0.4210) acc1: 97.3958 (79.7302) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2003 data: 0.1052 max mem: 3531 +Test: [ 70/234] eta: 0:00:38 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.0710 (0.4333) acc1: 99.7396 (78.8806) acc5: 100.0000 (100.0000) acc1_neg: 60.677 acc5_neg: 100.000 time: 0.1799 data: 0.0864 max mem: 3531 +Test: [ 80/234] eta: 0:00:35 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.6572 (0.4457) acc1: 64.8438 (78.4304) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2038 data: 0.1107 max mem: 3531 +Test: [ 90/234] eta: 0:00:32 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.2406 (0.4175) acc1: 93.7500 (80.2312) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.2001 data: 0.1079 max mem: 3531 +Test: [100/234] eta: 0:00:29 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.2627 (0.4197) acc1: 90.6250 (80.0923) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1710 data: 0.0791 max mem: 3531 +Test: [110/234] eta: 0:00:26 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.3692 (0.4154) acc1: 81.5104 (80.4359) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.1759 data: 0.0853 max mem: 3531 +Test: [120/234] eta: 0:00:24 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.3490 (0.4203) acc1: 84.6354 (80.2342) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1848 data: 0.0949 max mem: 3531 +Test: [130/234] eta: 0:00:22 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.4514 (0.4295) acc1: 78.9062 (79.5444) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.1912 data: 0.0993 max mem: 3531 +Test: [140/234] eta: 0:00:19 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.4850 (0.4461) acc1: 70.3125 (78.8342) acc5: 100.0000 (100.0000) acc1_neg: 38.021 acc5_neg: 100.000 time: 0.1989 data: 0.1057 max mem: 3531 +Test: [150/234] eta: 0:00:17 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.4850 (0.4537) acc1: 73.1771 (78.4147) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.1974 data: 0.1043 max mem: 3531 +Test: [160/234] eta: 0:00:15 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.5252 (0.4576) acc1: 73.1771 (78.1331) acc5: 100.0000 (100.0000) acc1_neg: 65.365 acc5_neg: 100.000 time: 0.1853 data: 0.0932 max mem: 3531 +Test: [170/234] eta: 0:00:13 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.5788 (0.4658) acc1: 68.7500 (77.4671) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.1880 data: 0.0963 max mem: 3531 +Test: [180/234] eta: 0:00:11 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.5788 (0.4726) acc1: 68.7500 (76.9984) acc5: 100.0000 (100.0000) acc1_neg: 42.708 acc5_neg: 100.000 time: 0.1995 data: 0.1074 max mem: 3531 +Test: [190/234] eta: 0:00:09 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.5202 (0.4767) acc1: 72.1354 (76.7002) acc5: 100.0000 (100.0000) acc1_neg: 72.917 acc5_neg: 100.000 time: 0.1968 data: 0.1053 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.5173 (0.4667) acc1: 73.4375 (77.3515) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1949 data: 0.1026 max mem: 3531 +Test: [210/234] eta: 0:00:04 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.1200 (0.4509) acc1: 99.4792 (78.3632) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1906 data: 0.0987 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.1563 (0.4385) acc1: 97.1354 (79.1643) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1613 data: 0.0708 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.2848 (0.4448) acc1: 70.8333 (78.6977) acc5: 100.0000 (100.0000) acc1_neg: 68.229 acc5_neg: 100.000 time: 0.1381 data: 0.0483 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 95.172 acc5_pos: 100.000 loss: 0.5662 (0.4467) acc1: 69.5312 (78.5565) acc5: 100.0000 (100.0000) acc1_neg: 65.759 acc5_neg: 100.000 time: 0.1367 data: 0.0456 max mem: 3532 +Test: Total time: 0:00:46 (0.1990 s / it) +* Overall Acc@1 78.557 Acc@5 100.000 + target == real Acc@1 75.949 Acc@5 100.000 + target == fake Acc@1 79.291 Acc@5 100.000 + loss 0.447 + PR-AUC 0.9577 ROC-AUC 0.8641 EER 22.07% (thr=0.5215) +Accuracy of the network on the 89729 test images: 78.6% +[rank0]:[W725 16:43:27.667723742 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 16:43:45.088000 2558282 torch/distributed/run.py:766] +W0725 16:43:45.088000 2558282 torch/distributed/run.py:766] ***************************************** +W0725 16:43:45.088000 2558282 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 16:43:45.088000 2558282 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 16:44:11.794392274 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 16:44:11.285564759 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 16:44:11.420300346 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 16:44:11.542605016 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 16:44:12.635175453 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 16:44:12.700265849 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 16:44:12.728090339 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 16:44:12.741694129 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed1_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed1_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:23:50 acc1_pos: 87.760 acc5_pos: 100.000 loss: 0.4116 (0.4116) acc1: 87.7604 (87.7604) acc5: 100.0000 (100.0000) time: 6.6320 data: 4.9448 max mem: 3531 +Test: [ 10/2387] eta: 0:27:07 acc1_pos: 75.260 acc5_pos: 100.000 loss: 0.4630 (0.4701) acc1: 79.9479 (81.6525) acc5: 100.0000 (100.0000) time: 0.6848 data: 0.4500 max mem: 3531 +Test: [ 20/2387] eta: 0:15:58 acc1_pos: 80.208 acc5_pos: 100.000 loss: 0.4630 (0.4894) acc1: 79.9479 (81.3616) acc5: 100.0000 (100.0000) time: 0.0937 data: 0.0005 max mem: 3531 +Test: [ 30/2387] eta: 0:13:33 acc1_pos: 62.500 acc5_pos: 100.000 loss: 0.4848 (0.4900) acc1: 80.2083 (80.6620) acc5: 100.0000 (100.0000) time: 0.1586 data: 0.0632 max mem: 3531 +Test: [ 40/2387] eta: 0:12:24 acc1_pos: 67.448 acc5_pos: 100.000 loss: 0.5262 (0.5027) acc1: 74.4792 (79.2429) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1276 max mem: 3531 +Test: [ 50/2387] eta: 0:11:51 acc1_pos: 63.542 acc5_pos: 100.000 loss: 0.5302 (0.5075) acc1: 73.9583 (78.5692) acc5: 100.0000 (100.0000) time: 0.2411 data: 0.1420 max mem: 3531 +Test: [ 60/2387] eta: 0:11:22 acc1_pos: 68.229 acc5_pos: 100.000 loss: 0.6346 (0.5376) acc1: 68.2292 (76.1612) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1505 max mem: 3531 +Test: [ 70/2387] eta: 0:11:12 acc1_pos: 61.458 acc5_pos: 100.000 loss: 0.6739 (0.5617) acc1: 64.8438 (74.2738) acc5: 100.0000 (100.0000) time: 0.2537 data: 0.1611 max mem: 3531 +Test: [ 80/2387] eta: 0:10:55 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.5946 (0.5377) acc1: 73.1771 (76.2314) acc5: 100.0000 (100.0000) time: 0.2559 data: 0.1615 max mem: 3531 +Test: [ 90/2387] eta: 0:10:41 acc1_pos: 98.958 acc5_pos: 100.000 loss: 0.3156 (0.5107) acc1: 94.0104 (78.3854) acc5: 100.0000 (100.0000) time: 0.2399 data: 0.1424 max mem: 3531 +Test: [ 100/2387] eta: 0:10:36 acc1_pos: 99.740 acc5_pos: 100.000 loss: 0.2883 (0.4920) acc1: 95.0521 (79.8010) acc5: 100.0000 (100.0000) time: 0.2539 data: 0.1574 max mem: 3531 +Test: [ 110/2387] eta: 0:10:26 acc1_pos: 86.458 acc5_pos: 100.000 loss: 0.3129 (0.4756) acc1: 92.9688 (80.9990) acc5: 100.0000 (100.0000) time: 0.2565 data: 0.1642 max mem: 3531 +Test: [ 120/2387] eta: 0:10:18 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.3129 (0.4647) acc1: 93.7500 (81.9538) acc5: 100.0000 (100.0000) time: 0.2451 data: 0.1519 max mem: 3531 +Test: [ 130/2387] eta: 0:10:06 acc1_pos: 95.312 acc5_pos: 100.000 loss: 0.3129 (0.4527) acc1: 95.3125 (82.9318) acc5: 100.0000 (100.0000) time: 0.2333 data: 0.1377 max mem: 3531 +Test: [ 140/2387] eta: 0:09:59 acc1_pos: 98.958 acc5_pos: 100.000 loss: 0.3084 (0.4424) acc1: 95.5729 (83.8505) acc5: 100.0000 (100.0000) time: 0.2300 data: 0.1307 max mem: 3531 +Test: [ 150/2387] eta: 0:09:48 acc1_pos: 98.958 acc5_pos: 100.000 loss: 0.2976 (0.4322) acc1: 96.8750 (84.6941) acc5: 100.0000 (100.0000) time: 0.2278 data: 0.1314 max mem: 3531 +Test: [ 160/2387] eta: 0:09:39 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2822 (0.4229) acc1: 96.8750 (85.3180) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.2149 data: 0.1226 max mem: 3531 +Test: [ 170/2387] eta: 0:09:31 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3424 (0.4189) acc1: 89.5833 (85.4532) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.2170 data: 0.1214 max mem: 3531 +Test: [ 180/2387] eta: 0:09:23 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3615 (0.4156) acc1: 86.9792 (85.5001) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.2147 data: 0.1195 max mem: 3531 +Test: [ 190/2387] eta: 0:09:16 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3615 (0.4122) acc1: 87.2396 (85.6171) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.2149 data: 0.1174 max mem: 3531 +Test: [ 200/2387] eta: 0:09:12 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3509 (0.4092) acc1: 87.7604 (85.7160) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.2300 data: 0.1327 max mem: 3531 +Test: [ 210/2387] eta: 0:09:01 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3582 (0.4071) acc1: 87.2396 (85.7363) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.2065 data: 0.1111 max mem: 3532 +Test: [ 220/2387] eta: 0:08:53 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3598 (0.4050) acc1: 87.2396 (85.7831) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.1824 data: 0.0844 max mem: 3532 +Test: [ 230/2387] eta: 0:08:45 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3612 (0.4030) acc1: 86.7188 (85.8135) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.1855 data: 0.0903 max mem: 3532 +Test: [ 240/2387] eta: 0:08:36 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3714 (0.4099) acc1: 82.0312 (85.1130) acc5: 100.0000 (100.0000) acc1_neg: 68.490 acc5_neg: 100.000 time: 0.1773 data: 0.0841 max mem: 3532 +Test: [ 250/2387] eta: 0:08:30 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3442 (0.4020) acc1: 71.0938 (85.4924) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1837 data: 0.0902 max mem: 3532 +Test: [ 260/2387] eta: 0:08:23 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1555 (0.3920) acc1: 98.1771 (85.9934) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1924 data: 0.0996 max mem: 3532 +Test: [ 270/2387] eta: 0:08:16 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1346 (0.3828) acc1: 98.9583 (86.4603) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1805 data: 0.0853 max mem: 3532 +Test: [ 280/2387] eta: 0:08:06 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1408 (0.3744) acc1: 98.9583 (86.9060) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1526 data: 0.0562 max mem: 3532 +Test: [ 290/2387] eta: 0:07:58 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1420 (0.3662) acc1: 98.6979 (87.3067) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1440 data: 0.0476 max mem: 3532 +Test: [ 300/2387] eta: 0:07:53 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1397 (0.3591) acc1: 98.9583 (87.6869) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1688 data: 0.0682 max mem: 3532 +Test: [ 310/2387] eta: 0:07:50 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1569 (0.3561) acc1: 98.6979 (87.9137) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.2035 data: 0.1053 max mem: 3532 +Test: [ 320/2387] eta: 0:07:44 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3655 (0.3567) acc1: 91.4062 (88.0038) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1926 data: 0.1010 max mem: 3532 +Test: [ 330/2387] eta: 0:07:41 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3756 (0.3572) acc1: 90.6250 (88.0885) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1859 data: 0.0913 max mem: 3532 +Test: [ 340/2387] eta: 0:07:34 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3726 (0.3576) acc1: 90.6250 (88.1636) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1821 data: 0.0849 max mem: 3532 +Test: [ 350/2387] eta: 0:07:29 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3510 (0.3525) acc1: 92.1875 (88.4608) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1635 data: 0.0681 max mem: 3532 +Test: [ 360/2387] eta: 0:07:25 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2226 (0.3504) acc1: 92.9688 (88.5640) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1785 data: 0.0831 max mem: 3532 +Test: [ 370/2387] eta: 0:07:25 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2977 (0.3490) acc1: 91.4062 (88.6498) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2248 data: 0.1309 max mem: 3532 +Test: [ 380/2387] eta: 0:07:22 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2981 (0.3475) acc1: 91.6667 (88.7269) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.2387 data: 0.1473 max mem: 3532 +Test: [ 390/2387] eta: 0:07:17 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2923 (0.3464) acc1: 91.1458 (88.7741) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1809 data: 0.0906 max mem: 3532 +Test: [ 400/2387] eta: 0:07:12 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2974 (0.3452) acc1: 91.1458 (88.8313) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1591 data: 0.0688 max mem: 3532 +Test: [ 410/2387] eta: 0:07:10 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2971 (0.3439) acc1: 91.4062 (88.9079) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1991 data: 0.1087 max mem: 3532 +Test: [ 420/2387] eta: 0:07:06 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2944 (0.3428) acc1: 90.8854 (88.9487) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.2000 data: 0.1073 max mem: 3532 +Test: [ 430/2387] eta: 0:07:03 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3067 (0.3445) acc1: 89.8438 (88.7676) acc5: 100.0000 (100.0000) acc1_neg: 72.396 acc5_neg: 100.000 time: 0.1897 data: 0.0961 max mem: 3532 +Test: [ 440/2387] eta: 0:07:01 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3074 (0.3425) acc1: 81.2500 (88.8163) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2083 data: 0.1148 max mem: 3533 +Test: [ 450/2387] eta: 0:06:58 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1109 (0.3374) acc1: 100.0000 (89.0613) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2051 data: 0.1114 max mem: 3533 +Test: [ 460/2387] eta: 0:06:54 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1090 (0.3325) acc1: 99.7396 (89.2941) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1907 data: 0.0976 max mem: 3533 +Test: [ 470/2387] eta: 0:06:52 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1090 (0.3278) acc1: 99.7396 (89.5186) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1966 data: 0.1038 max mem: 3533 +Test: [ 480/2387] eta: 0:06:49 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1095 (0.3232) acc1: 100.0000 (89.7344) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1995 data: 0.1082 max mem: 3533 +Test: [ 490/2387] eta: 0:06:47 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1102 (0.3189) acc1: 100.0000 (89.9413) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2029 data: 0.1121 max mem: 3533 +Test: [ 500/2387] eta: 0:06:44 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1114 (0.3148) acc1: 100.0000 (90.1385) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2068 data: 0.1134 max mem: 3533 +Test: [ 510/2387] eta: 0:06:42 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1114 (0.3115) acc1: 99.7396 (90.3259) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2056 data: 0.1112 max mem: 3533 +Test: [ 520/2387] eta: 0:06:41 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1974 (0.3098) acc1: 99.4792 (90.4905) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2358 data: 0.1418 max mem: 3533 +Test: [ 530/2387] eta: 0:06:37 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1971 (0.3076) acc1: 98.6979 (90.6407) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1993 data: 0.1072 max mem: 3533 +Test: [ 540/2387] eta: 0:06:33 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1931 (0.3055) acc1: 98.6979 (90.7867) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1643 data: 0.0720 max mem: 3533 +Test: [ 550/2387] eta: 0:06:31 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1923 (0.3035) acc1: 98.4375 (90.9237) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1997 data: 0.1054 max mem: 3533 +Test: [ 560/2387] eta: 0:06:29 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1923 (0.3015) acc1: 98.4375 (91.0549) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2138 data: 0.1211 max mem: 3533 +Test: [ 570/2387] eta: 0:06:27 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1890 (0.2995) acc1: 98.4375 (91.1841) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2159 data: 0.1245 max mem: 3533 +Test: [ 580/2387] eta: 0:06:25 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1889 (0.2977) acc1: 98.4375 (91.3121) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2134 data: 0.1219 max mem: 3533 +Test: [ 590/2387] eta: 0:06:22 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1946 (0.2960) acc1: 98.4375 (91.4270) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1989 data: 0.1043 max mem: 3533 +Test: [ 600/2387] eta: 0:06:20 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2081 (0.2950) acc1: 98.4375 (91.5492) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2049 data: 0.1107 max mem: 3533 +Test: [ 610/2387] eta: 0:06:18 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2846 (0.2965) acc1: 88.5417 (91.4561) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.2166 data: 0.1253 max mem: 3533 +Test: [ 620/2387] eta: 0:06:16 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3814 (0.2978) acc1: 85.6771 (91.3563) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.2131 data: 0.1216 max mem: 3533 +Test: [ 630/2387] eta: 0:06:15 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3775 (0.2991) acc1: 85.6771 (91.2767) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.2293 data: 0.1379 max mem: 3533 +Test: [ 640/2387] eta: 0:06:12 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3828 (0.3005) acc1: 85.9375 (91.1873) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.2179 data: 0.1262 max mem: 3533 +Test: [ 650/2387] eta: 0:06:09 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3838 (0.3017) acc1: 85.6771 (91.1098) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1954 data: 0.1039 max mem: 3533 +Test: [ 660/2387] eta: 0:06:06 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3826 (0.3029) acc1: 86.1979 (91.0493) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1886 data: 0.0961 max mem: 3533 +Test: [ 670/2387] eta: 0:06:03 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3788 (0.3041) acc1: 86.1979 (90.9724) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1733 data: 0.0806 max mem: 3534 +Test: [ 680/2387] eta: 0:06:01 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3925 (0.3067) acc1: 84.6354 (90.7317) acc5: 100.0000 (100.0000) acc1_neg: 66.927 acc5_neg: 100.000 time: 0.1929 data: 0.0979 max mem: 3534 +Test: [ 690/2387] eta: 0:05:59 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3922 (0.3072) acc1: 74.7396 (90.6638) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.2174 data: 0.1200 max mem: 3534 +Test: [ 700/2387] eta: 0:05:56 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2596 (0.3065) acc1: 91.6667 (90.6811) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2031 data: 0.1054 max mem: 3534 +Test: [ 710/2387] eta: 0:05:55 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2580 (0.3059) acc1: 92.1875 (90.7048) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2087 data: 0.1138 max mem: 3534 +Test: [ 720/2387] eta: 0:05:52 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2543 (0.3051) acc1: 92.9688 (90.7366) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.2110 data: 0.1186 max mem: 3534 +Test: [ 730/2387] eta: 0:05:50 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2564 (0.3045) acc1: 92.7083 (90.7607) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1984 data: 0.1040 max mem: 3534 +Test: [ 740/2387] eta: 0:05:48 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2553 (0.3038) acc1: 92.9688 (90.7983) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.2226 data: 0.1266 max mem: 3534 +Test: [ 750/2387] eta: 0:05:47 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2511 (0.3031) acc1: 92.9688 (90.8261) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.2309 data: 0.1351 max mem: 3534 +Test: [ 760/2387] eta: 0:05:44 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2605 (0.3027) acc1: 92.1875 (90.8283) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.2114 data: 0.1163 max mem: 3534 +Test: [ 770/2387] eta: 0:05:43 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2847 (0.3028) acc1: 89.8438 (90.7996) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.2242 data: 0.1313 max mem: 3534 +Test: [ 780/2387] eta: 0:05:41 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3059 (0.3033) acc1: 88.0208 (90.7467) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2284 data: 0.1353 max mem: 3534 +Test: [ 790/2387] eta: 0:05:38 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2938 (0.3034) acc1: 88.0208 (90.6912) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.2057 data: 0.1118 max mem: 3534 +Test: [ 800/2387] eta: 0:05:36 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3089 (0.3038) acc1: 88.8021 (90.6465) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.1920 data: 0.0990 max mem: 3534 +Test: [ 810/2387] eta: 0:05:33 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3237 (0.3044) acc1: 84.8958 (90.5662) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.1858 data: 0.0937 max mem: 3534 +Test: [ 820/2387] eta: 0:05:30 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3251 (0.3051) acc1: 84.1146 (90.4877) acc5: 100.0000 (100.0000) acc1_neg: 76.302 acc5_neg: 100.000 time: 0.1851 data: 0.0942 max mem: 3534 +Test: [ 830/2387] eta: 0:05:27 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.4369 (0.3080) acc1: 76.5625 (90.2486) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.1751 data: 0.0834 max mem: 3534 +Test: [ 840/2387] eta: 0:05:25 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.5247 (0.3103) acc1: 72.9167 (90.0769) acc5: 100.0000 (100.0000) acc1_neg: 75.781 acc5_neg: 100.000 time: 0.1755 data: 0.0847 max mem: 3534 +Test: [ 850/2387] eta: 0:05:22 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.4975 (0.3124) acc1: 75.7812 (89.9166) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.1727 data: 0.0815 max mem: 3534 +Test: [ 860/2387] eta: 0:05:19 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.4591 (0.3139) acc1: 77.8646 (89.8247) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1596 data: 0.0672 max mem: 3534 +Test: [ 870/2387] eta: 0:05:16 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.4265 (0.3153) acc1: 80.2083 (89.7083) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.1749 data: 0.0800 max mem: 3534 +Test: [ 880/2387] eta: 0:05:14 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.4604 (0.3170) acc1: 78.1250 (89.5754) acc5: 100.0000 (100.0000) acc1_neg: 72.917 acc5_neg: 100.000 time: 0.1785 data: 0.0826 max mem: 3534 +Test: [ 890/2387] eta: 0:05:11 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.4328 (0.3184) acc1: 76.5625 (89.4749) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.1599 data: 0.0670 max mem: 3534 +Test: [ 900/2387] eta: 0:05:08 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.5164 (0.3229) acc1: 68.7500 (89.1443) acc5: 100.0000 (100.0000) acc1_neg: 58.333 acc5_neg: 100.000 time: 0.1787 data: 0.0867 max mem: 3535 +Test: [ 910/2387] eta: 0:05:06 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.7693 (0.3279) acc1: 55.2083 (88.7592) acc5: 100.0000 (100.0000) acc1_neg: 52.865 acc5_neg: 100.000 time: 0.1866 data: 0.0917 max mem: 3535 +Test: [ 920/2387] eta: 0:05:03 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.7860 (0.3330) acc1: 53.3854 (88.3819) acc5: 100.0000 (100.0000) acc1_neg: 53.385 acc5_neg: 100.000 time: 0.1799 data: 0.0852 max mem: 3535 +Test: [ 930/2387] eta: 0:05:01 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.7874 (0.3379) acc1: 53.9062 (88.0186) acc5: 100.0000 (100.0000) acc1_neg: 51.302 acc5_neg: 100.000 time: 0.1782 data: 0.0851 max mem: 3535 +Test: [ 940/2387] eta: 0:04:58 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.7929 (0.3427) acc1: 54.1667 (87.6644) acc5: 100.0000 (100.0000) acc1_neg: 51.302 acc5_neg: 100.000 time: 0.1678 data: 0.0741 max mem: 3535 +Test: [ 950/2387] eta: 0:04:56 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.7823 (0.3474) acc1: 54.1667 (87.3198) acc5: 100.0000 (100.0000) acc1_neg: 55.208 acc5_neg: 100.000 time: 0.1821 data: 0.0906 max mem: 3535 +Test: [ 960/2387] eta: 0:04:53 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.7684 (0.3518) acc1: 55.2083 (86.9946) acc5: 100.0000 (100.0000) acc1_neg: 58.854 acc5_neg: 100.000 time: 0.1782 data: 0.0867 max mem: 3535 +Test: [ 970/2387] eta: 0:04:50 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.7777 (0.3564) acc1: 54.4271 (86.6477) acc5: 100.0000 (100.0000) acc1_neg: 55.990 acc5_neg: 100.000 time: 0.1428 data: 0.0487 max mem: 3535 +Test: [ 980/2387] eta: 0:04:48 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.7664 (0.3581) acc1: 53.9062 (86.5175) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.1659 data: 0.0728 max mem: 3535 +Test: [ 990/2387] eta: 0:04:45 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.4723 (0.3593) acc1: 77.6042 (86.4071) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.1926 data: 0.1003 max mem: 3535 +Test: [1000/2387] eta: 0:04:43 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.4792 (0.3603) acc1: 74.7396 (86.3166) acc5: 100.0000 (100.0000) acc1_neg: 72.917 acc5_neg: 100.000 time: 0.1647 data: 0.0721 max mem: 3535 +Test: [1010/2387] eta: 0:04:40 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.5113 (0.3623) acc1: 73.1771 (86.1505) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.1476 data: 0.0549 max mem: 3535 +Test: [1020/2387] eta: 0:04:38 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.5308 (0.3640) acc1: 69.7917 (86.0117) acc5: 100.0000 (100.0000) acc1_neg: 76.302 acc5_neg: 100.000 time: 0.1818 data: 0.0860 max mem: 3535 +Test: [1030/2387] eta: 0:04:35 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.4019 (0.3643) acc1: 80.9896 (85.9706) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.1844 data: 0.0878 max mem: 3535 +Test: [1040/2387] eta: 0:04:33 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3352 (0.3637) acc1: 87.5000 (86.0138) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1666 data: 0.0701 max mem: 3535 +Test: [1050/2387] eta: 0:04:30 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3263 (0.3637) acc1: 88.2812 (86.0121) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1699 data: 0.0708 max mem: 3535 +Test: [1060/2387] eta: 0:04:28 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3737 (0.3641) acc1: 84.1146 (85.9768) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.1667 data: 0.0673 max mem: 3535 +Test: [1070/2387] eta: 0:04:25 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.4069 (0.3650) acc1: 80.9896 (85.8947) acc5: 100.0000 (100.0000) acc1_neg: 61.979 acc5_neg: 100.000 time: 0.1698 data: 0.0741 max mem: 3535 +Test: [1080/2387] eta: 0:04:23 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.5508 (0.3671) acc1: 68.7500 (85.7139) acc5: 100.0000 (100.0000) acc1_neg: 70.052 acc5_neg: 100.000 time: 0.1696 data: 0.0741 max mem: 3535 +Test: [1090/2387] eta: 0:04:21 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.5524 (0.3684) acc1: 69.7917 (85.6169) acc5: 100.0000 (100.0000) acc1_neg: 64.323 acc5_neg: 100.000 time: 0.1649 data: 0.0685 max mem: 3535 +Test: [1100/2387] eta: 0:04:18 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.5048 (0.3693) acc1: 75.2604 (85.5425) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1747 data: 0.0783 max mem: 3535 +Test: [1110/2387] eta: 0:04:16 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.4710 (0.3707) acc1: 75.7812 (85.4214) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.1799 data: 0.0821 max mem: 3535 +Test: [1120/2387] eta: 0:04:14 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.4710 (0.3717) acc1: 75.5208 (85.3319) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.1649 data: 0.0670 max mem: 3536 +Test: [1130/2387] eta: 0:04:11 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.4697 (0.3727) acc1: 75.5208 (85.2525) acc5: 100.0000 (100.0000) acc1_neg: 72.917 acc5_neg: 100.000 time: 0.1770 data: 0.0812 max mem: 3536 +Test: [1140/2387] eta: 0:04:09 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.5744 (0.3750) acc1: 68.7500 (85.0631) acc5: 100.0000 (100.0000) acc1_neg: 55.729 acc5_neg: 100.000 time: 0.1801 data: 0.0866 max mem: 3536 +Test: [1150/2387] eta: 0:04:07 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.6217 (0.3756) acc1: 65.6250 (85.0234) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.1834 data: 0.0900 max mem: 3536 +Test: [1160/2387] eta: 0:04:05 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2856 (0.3745) acc1: 88.2812 (85.1002) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1828 data: 0.0876 max mem: 3536 +Test: [1170/2387] eta: 0:04:03 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2557 (0.3737) acc1: 92.7083 (85.1645) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1749 data: 0.0776 max mem: 3536 +Test: [1180/2387] eta: 0:04:00 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2488 (0.3726) acc1: 92.4479 (85.2418) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1622 data: 0.0652 max mem: 3536 +Test: [1190/2387] eta: 0:03:58 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2273 (0.3716) acc1: 94.7917 (85.3115) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1709 data: 0.0762 max mem: 3536 +Test: [1200/2387] eta: 0:03:56 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2140 (0.3706) acc1: 93.4896 (85.3759) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.1929 data: 0.0991 max mem: 3536 +Test: [1210/2387] eta: 0:03:54 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2483 (0.3696) acc1: 92.9688 (85.4420) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1888 data: 0.0931 max mem: 3536 +Test: [1220/2387] eta: 0:03:51 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2032 (0.3681) acc1: 96.3542 (85.5395) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1660 data: 0.0693 max mem: 3536 +Test: [1230/2387] eta: 0:03:49 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1876 (0.3664) acc1: 97.6562 (85.6415) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1702 data: 0.0746 max mem: 3536 +Test: [1240/2387] eta: 0:03:47 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1559 (0.3648) acc1: 98.1771 (85.7413) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1960 data: 0.1007 max mem: 3536 +Test: [1250/2387] eta: 0:03:45 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1522 (0.3632) acc1: 97.9167 (85.8386) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1860 data: 0.0915 max mem: 3536 +Test: [1260/2387] eta: 0:03:43 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1522 (0.3615) acc1: 97.9167 (85.9328) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1638 data: 0.0701 max mem: 3536 +Test: [1270/2387] eta: 0:03:40 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1650 (0.3601) acc1: 97.6562 (86.0205) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1479 data: 0.0534 max mem: 3536 +Test: [1280/2387] eta: 0:03:38 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1737 (0.3590) acc1: 97.1354 (86.0948) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1801 data: 0.0840 max mem: 3536 +Test: [1290/2387] eta: 0:03:36 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2165 (0.3580) acc1: 94.7917 (86.1533) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1914 data: 0.0955 max mem: 3536 +Test: [1300/2387] eta: 0:03:34 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2257 (0.3571) acc1: 93.7500 (86.2035) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1534 data: 0.0590 max mem: 3536 +Test: [1310/2387] eta: 0:03:32 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2526 (0.3564) acc1: 93.4896 (86.2460) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.1640 data: 0.0681 max mem: 3536 +Test: [1320/2387] eta: 0:03:29 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2513 (0.3556) acc1: 93.2292 (86.3063) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1748 data: 0.0772 max mem: 3536 +Test: [1330/2387] eta: 0:03:27 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2833 (0.3559) acc1: 89.8438 (86.2789) acc5: 100.0000 (100.0000) acc1_neg: 74.219 acc5_neg: 100.000 time: 0.1611 data: 0.0630 max mem: 3536 +Test: [1340/2387] eta: 0:03:25 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.4251 (0.3567) acc1: 78.1250 (86.2061) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1676 data: 0.0713 max mem: 3536 +Test: [1350/2387] eta: 0:03:23 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1146 (0.3547) acc1: 98.9583 (86.3059) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1701 data: 0.0770 max mem: 3537 +Test: [1360/2387] eta: 0:03:21 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.0755 (0.3526) acc1: 100.0000 (86.4065) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1673 data: 0.0740 max mem: 3537 +Test: [1370/2387] eta: 0:03:19 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.0692 (0.3506) acc1: 100.0000 (86.5047) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1697 data: 0.0738 max mem: 3537 +Test: [1380/2387] eta: 0:03:16 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.0751 (0.3487) acc1: 100.0000 (86.6011) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1703 data: 0.0745 max mem: 3537 +Test: [1390/2387] eta: 0:03:14 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.0723 (0.3467) acc1: 100.0000 (86.6974) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1752 data: 0.0795 max mem: 3537 +Test: [1400/2387] eta: 0:03:13 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.0783 (0.3449) acc1: 100.0000 (86.7898) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2047 data: 0.1093 max mem: 3537 +Test: [1410/2387] eta: 0:03:11 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1083 (0.3449) acc1: 99.4792 (86.7839) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.2255 data: 0.1301 max mem: 3537 +Test: [1420/2387] eta: 0:03:09 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3811 (0.3453) acc1: 80.9896 (86.7347) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.2147 data: 0.1193 max mem: 3537 +Test: [1430/2387] eta: 0:03:06 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.4216 (0.3461) acc1: 79.1667 (86.6736) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1623 data: 0.0664 max mem: 3537 +Test: [1440/2387] eta: 0:03:04 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.4612 (0.3467) acc1: 76.0417 (86.6141) acc5: 100.0000 (100.0000) acc1_neg: 66.146 acc5_neg: 100.000 time: 0.1452 data: 0.0495 max mem: 3537 +Test: [1450/2387] eta: 0:03:03 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.4666 (0.3478) acc1: 75.0000 (86.5253) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.2070 data: 0.1148 max mem: 3537 +Test: [1460/2387] eta: 0:03:01 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3497 (0.3477) acc1: 84.6354 (86.5291) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.2297 data: 0.1385 max mem: 3537 +Test: [1470/2387] eta: 0:02:59 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2637 (0.3469) acc1: 91.1458 (86.5709) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2097 data: 0.1178 max mem: 3537 +Test: [1480/2387] eta: 0:02:57 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2405 (0.3463) acc1: 93.2292 (86.6090) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1951 data: 0.1031 max mem: 3537 +Test: [1490/2387] eta: 0:02:55 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2405 (0.3457) acc1: 92.1875 (86.6536) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1942 data: 0.1030 max mem: 3537 +Test: [1500/2387] eta: 0:02:53 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2388 (0.3451) acc1: 93.7500 (86.6905) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1997 data: 0.1071 max mem: 3537 +Test: [1510/2387] eta: 0:02:51 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2648 (0.3448) acc1: 91.1458 (86.7206) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1838 data: 0.0877 max mem: 3537 +Test: [1520/2387] eta: 0:02:49 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3412 (0.3451) acc1: 89.5833 (86.7107) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.1755 data: 0.0803 max mem: 3537 +Test: [1530/2387] eta: 0:02:47 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3871 (0.3454) acc1: 85.9375 (86.6958) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.1806 data: 0.0882 max mem: 3537 +Test: [1540/2387] eta: 0:02:45 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3756 (0.3456) acc1: 86.4583 (86.7044) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1743 data: 0.0822 max mem: 3537 +Test: [1550/2387] eta: 0:02:43 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3522 (0.3456) acc1: 88.5417 (86.7092) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.1813 data: 0.0880 max mem: 3537 +Test: [1560/2387] eta: 0:02:41 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3324 (0.3457) acc1: 88.5417 (86.7051) acc5: 100.0000 (100.0000) acc1_neg: 72.396 acc5_neg: 100.000 time: 0.1949 data: 0.1019 max mem: 3537 +Test: [1570/2387] eta: 0:02:39 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3423 (0.3458) acc1: 88.2812 (86.7055) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.1891 data: 0.0960 max mem: 3537 +Test: [1580/2387] eta: 0:02:37 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3933 (0.3463) acc1: 84.1146 (86.6731) acc5: 100.0000 (100.0000) acc1_neg: 70.312 acc5_neg: 100.000 time: 0.1826 data: 0.0901 max mem: 3538 +Test: [1590/2387] eta: 0:02:35 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3556 (0.3465) acc1: 83.0729 (86.6498) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.1786 data: 0.0866 max mem: 3538 +Test: [1600/2387] eta: 0:02:33 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3551 (0.3466) acc1: 83.0729 (86.6246) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.1822 data: 0.0856 max mem: 3538 +Test: [1610/2387] eta: 0:02:31 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3651 (0.3468) acc1: 82.8125 (86.6093) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.1699 data: 0.0724 max mem: 3538 +Test: [1620/2387] eta: 0:02:29 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3618 (0.3471) acc1: 82.8125 (86.5796) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.1623 data: 0.0686 max mem: 3538 +Test: [1630/2387] eta: 0:02:27 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.4416 (0.3478) acc1: 77.8646 (86.5147) acc5: 100.0000 (100.0000) acc1_neg: 77.344 acc5_neg: 100.000 time: 0.1799 data: 0.0866 max mem: 3538 +Test: [1640/2387] eta: 0:02:25 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.4659 (0.3488) acc1: 75.0000 (86.4310) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.1801 data: 0.0854 max mem: 3538 +Test: [1650/2387] eta: 0:02:23 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.4909 (0.3497) acc1: 72.6562 (86.3522) acc5: 100.0000 (100.0000) acc1_neg: 77.344 acc5_neg: 100.000 time: 0.1775 data: 0.0852 max mem: 3538 +Test: [1660/2387] eta: 0:02:21 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.4815 (0.3506) acc1: 72.6562 (86.2722) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.1775 data: 0.0872 max mem: 3538 +Test: [1670/2387] eta: 0:02:18 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.4809 (0.3513) acc1: 75.0000 (86.2121) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.1682 data: 0.0759 max mem: 3538 +Test: [1680/2387] eta: 0:02:16 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.4776 (0.3521) acc1: 75.0000 (86.1449) acc5: 100.0000 (100.0000) acc1_neg: 69.531 acc5_neg: 100.000 time: 0.1696 data: 0.0764 max mem: 3538 +Test: [1690/2387] eta: 0:02:15 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.4833 (0.3530) acc1: 75.0000 (86.0746) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.1800 data: 0.0878 max mem: 3538 +Test: [1700/2387] eta: 0:02:12 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.5099 (0.3535) acc1: 72.9167 (86.0291) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1754 data: 0.0835 max mem: 3538 +Test: [1710/2387] eta: 0:02:10 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2454 (0.3527) acc1: 92.4479 (86.0806) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1615 data: 0.0684 max mem: 3538 +Test: [1720/2387] eta: 0:02:08 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1885 (0.3517) acc1: 95.5729 (86.1383) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1612 data: 0.0682 max mem: 3538 +Test: [1730/2387] eta: 0:02:06 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1735 (0.3507) acc1: 96.3542 (86.1928) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1682 data: 0.0769 max mem: 3538 +Test: [1740/2387] eta: 0:02:04 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1709 (0.3497) acc1: 96.6146 (86.2518) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1689 data: 0.0777 max mem: 3538 +Test: [1750/2387] eta: 0:02:02 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1689 (0.3488) acc1: 96.8750 (86.3109) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1789 data: 0.0879 max mem: 3538 +Test: [1760/2387] eta: 0:02:00 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2052 (0.3481) acc1: 94.5312 (86.3505) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.1749 data: 0.0827 max mem: 3538 +Test: [1770/2387] eta: 0:01:58 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2794 (0.3479) acc1: 89.0625 (86.3557) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.1670 data: 0.0740 max mem: 3538 +Test: [1780/2387] eta: 0:01:56 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3161 (0.3477) acc1: 87.2396 (86.3686) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1731 data: 0.0807 max mem: 3538 +Test: [1790/2387] eta: 0:01:54 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2835 (0.3472) acc1: 89.5833 (86.3971) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1738 data: 0.0822 max mem: 3538 +Test: [1800/2387] eta: 0:01:52 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2630 (0.3468) acc1: 91.9271 (86.4284) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1763 data: 0.0857 max mem: 3538 +Test: [1810/2387] eta: 0:01:51 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2339 (0.3460) acc1: 93.2292 (86.4681) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.2060 data: 0.1150 max mem: 3539 +Test: [1820/2387] eta: 0:01:49 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2361 (0.3456) acc1: 92.7083 (86.4951) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1969 data: 0.1058 max mem: 3539 +Test: [1830/2387] eta: 0:01:46 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2648 (0.3452) acc1: 90.8854 (86.5256) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1499 data: 0.0589 max mem: 3539 +Test: [1840/2387] eta: 0:01:45 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1669 (0.3442) acc1: 99.2188 (86.5950) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1644 data: 0.0718 max mem: 3539 +Test: [1850/2387] eta: 0:01:43 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1650 (0.3432) acc1: 99.2188 (86.6628) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1877 data: 0.0954 max mem: 3539 +Test: [1860/2387] eta: 0:01:41 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1650 (0.3423) acc1: 99.2188 (86.7302) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1979 data: 0.1038 max mem: 3539 +Test: [1870/2387] eta: 0:01:39 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1662 (0.3413) acc1: 99.2188 (86.7977) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1864 data: 0.0920 max mem: 3539 +Test: [1880/2387] eta: 0:01:37 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1632 (0.3404) acc1: 99.4792 (86.8647) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1699 data: 0.0748 max mem: 3539 +Test: [1890/2387] eta: 0:01:35 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1652 (0.3395) acc1: 99.2188 (86.9295) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1834 data: 0.0881 max mem: 3539 +Test: [1900/2387] eta: 0:01:33 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1698 (0.3386) acc1: 99.2188 (86.9912) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1724 data: 0.0794 max mem: 3539 +Test: [1910/2387] eta: 0:01:31 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2673 (0.3386) acc1: 95.0521 (87.0166) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.1737 data: 0.0807 max mem: 3539 +Test: [1920/2387] eta: 0:01:29 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3481 (0.3400) acc1: 85.4167 (86.9134) acc5: 100.0000 (100.0000) acc1_neg: 61.198 acc5_neg: 100.000 time: 0.1896 data: 0.0979 max mem: 3539 +Test: [1930/2387] eta: 0:01:27 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.4085 (0.3401) acc1: 81.2500 (86.9090) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1762 data: 0.0846 max mem: 3539 +Test: [1940/2387] eta: 0:01:25 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.4085 (0.3410) acc1: 80.4688 (86.8380) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.1800 data: 0.0871 max mem: 3539 +Test: [1950/2387] eta: 0:01:23 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.5060 (0.3419) acc1: 75.2604 (86.7807) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.1923 data: 0.1005 max mem: 3539 +Test: [1960/2387] eta: 0:01:21 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.4854 (0.3429) acc1: 74.7396 (86.6929) acc5: 100.0000 (100.0000) acc1_neg: 68.490 acc5_neg: 100.000 time: 0.1706 data: 0.0787 max mem: 3539 +Test: [1970/2387] eta: 0:01:19 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.5712 (0.3446) acc1: 65.3646 (86.5586) acc5: 100.0000 (100.0000) acc1_neg: 63.542 acc5_neg: 100.000 time: 0.1825 data: 0.0891 max mem: 3539 +Test: [1980/2387] eta: 0:01:17 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.6058 (0.3449) acc1: 61.1979 (86.5147) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2169 data: 0.1239 max mem: 3539 +Test: [1990/2387] eta: 0:01:15 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1338 (0.3438) acc1: 98.9583 (86.5774) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1871 data: 0.0937 max mem: 3539 +Test: [2000/2387] eta: 0:01:14 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1324 (0.3428) acc1: 98.9583 (86.6398) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1709 data: 0.0778 max mem: 3539 +Test: [2010/2387] eta: 0:01:12 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1323 (0.3418) acc1: 99.2188 (86.7017) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2054 data: 0.1093 max mem: 3539 +Test: [2020/2387] eta: 0:01:10 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1323 (0.3407) acc1: 99.2188 (86.7632) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1958 data: 0.1001 max mem: 3539 +Test: [2030/2387] eta: 0:01:08 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1334 (0.3397) acc1: 99.2188 (86.8235) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1595 data: 0.0684 max mem: 3540 +Test: [2040/2387] eta: 0:01:06 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1319 (0.3387) acc1: 98.9583 (86.8841) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1796 data: 0.0887 max mem: 3540 +Test: [2050/2387] eta: 0:01:04 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1319 (0.3377) acc1: 99.2188 (86.9444) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2037 data: 0.1112 max mem: 3540 +Test: [2060/2387] eta: 0:01:02 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1591 (0.3369) acc1: 98.6979 (86.9994) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1886 data: 0.0944 max mem: 3540 +Test: [2070/2387] eta: 0:01:00 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1656 (0.3361) acc1: 98.6979 (87.0569) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1639 data: 0.0680 max mem: 3540 +Test: [2080/2387] eta: 0:00:58 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1639 (0.3353) acc1: 98.6979 (87.1113) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1703 data: 0.0754 max mem: 3540 +Test: [2090/2387] eta: 0:00:56 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1652 (0.3345) acc1: 98.4375 (87.1670) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2049 data: 0.1120 max mem: 3540 +Test: [2100/2387] eta: 0:00:54 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1652 (0.3337) acc1: 98.6979 (87.2227) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1764 data: 0.0828 max mem: 3540 +Test: [2110/2387] eta: 0:00:52 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1652 (0.3329) acc1: 98.6979 (87.2765) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2000 data: 0.1068 max mem: 3540 +Test: [2120/2387] eta: 0:00:51 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.1813 (0.3332) acc1: 97.6562 (87.2497) acc5: 100.0000 (100.0000) acc1_neg: 71.354 acc5_neg: 100.000 time: 0.2432 data: 0.1482 max mem: 3540 +Test: [2130/2387] eta: 0:00:49 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3592 (0.3334) acc1: 83.8542 (87.2299) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1750 data: 0.0780 max mem: 3540 +Test: [2140/2387] eta: 0:00:47 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3592 (0.3338) acc1: 82.8125 (87.1919) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1445 data: 0.0490 max mem: 3540 +Test: [2150/2387] eta: 0:00:45 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3649 (0.3341) acc1: 82.5521 (87.1708) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.1849 data: 0.0920 max mem: 3540 +Test: [2160/2387] eta: 0:00:43 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3634 (0.3343) acc1: 83.3333 (87.1496) acc5: 100.0000 (100.0000) acc1_neg: 68.229 acc5_neg: 100.000 time: 0.2212 data: 0.1295 max mem: 3540 +Test: [2170/2387] eta: 0:00:41 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.4807 (0.3353) acc1: 71.8750 (87.0670) acc5: 100.0000 (100.0000) acc1_neg: 70.312 acc5_neg: 100.000 time: 0.2069 data: 0.1141 max mem: 3540 +Test: [2180/2387] eta: 0:00:39 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.5515 (0.3370) acc1: 69.0104 (86.9466) acc5: 100.0000 (100.0000) acc1_neg: 50.781 acc5_neg: 100.000 time: 0.1836 data: 0.0918 max mem: 3540 +Test: [2190/2387] eta: 0:00:37 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.4963 (0.3368) acc1: 74.4792 (86.9629) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.1877 data: 0.0963 max mem: 3540 +Test: [2200/2387] eta: 0:00:35 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3054 (0.3369) acc1: 94.0104 (86.9760) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1905 data: 0.0956 max mem: 3540 +Test: [2210/2387] eta: 0:00:33 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3128 (0.3368) acc1: 92.7083 (87.0026) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1806 data: 0.0865 max mem: 3540 +Test: [2220/2387] eta: 0:00:31 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3144 (0.3367) acc1: 92.4479 (87.0269) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1802 data: 0.0899 max mem: 3540 +Test: [2230/2387] eta: 0:00:30 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3224 (0.3366) acc1: 92.4479 (87.0531) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1960 data: 0.1059 max mem: 3540 +Test: [2240/2387] eta: 0:00:28 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3022 (0.3364) acc1: 92.9688 (87.0807) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1888 data: 0.0979 max mem: 3540 +Test: [2250/2387] eta: 0:00:26 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3391 (0.3366) acc1: 90.8854 (87.0811) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.1691 data: 0.0778 max mem: 3540 +Test: [2260/2387] eta: 0:00:24 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3482 (0.3366) acc1: 89.3229 (87.0856) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.1708 data: 0.0800 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3024 (0.3365) acc1: 89.3229 (87.0932) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1650 data: 0.0736 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2657 (0.3365) acc1: 87.2396 (87.0827) acc5: 100.0000 (100.0000) acc1_neg: 78.646 acc5_neg: 100.000 time: 0.1390 data: 0.0475 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.2983 (0.3364) acc1: 88.8021 (87.0874) acc5: 100.0000 (100.0000) acc1_neg: 75.521 acc5_neg: 100.000 time: 0.1341 data: 0.0436 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3155 (0.3364) acc1: 88.5417 (87.0820) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.1377 data: 0.0477 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.3641 (0.3369) acc1: 84.1146 (87.0450) acc5: 100.0000 (100.0000) acc1_neg: 67.969 acc5_neg: 100.000 time: 0.1311 data: 0.0415 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.4397 (0.3376) acc1: 76.5625 (86.9894) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1124 data: 0.0228 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.5240 (0.3386) acc1: 69.7917 (86.9034) acc5: 100.0000 (100.0000) acc1_neg: 74.219 acc5_neg: 100.000 time: 0.1046 data: 0.0150 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.5077 (0.3391) acc1: 74.2188 (86.8746) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.1178 data: 0.0282 max mem: 3541 +Test: [2350/2387] eta: 0:00:06 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.4696 (0.3400) acc1: 75.2604 (86.7943) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.1300 data: 0.0404 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.5196 (0.3410) acc1: 72.6562 (86.7234) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.1276 data: 0.0380 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.5391 (0.3421) acc1: 70.8333 (86.6407) acc5: 100.0000 (100.0000) acc1_neg: 63.021 acc5_neg: 100.000 time: 0.1288 data: 0.0393 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.6116 (0.3434) acc1: 63.0208 (86.5349) acc5: 100.0000 (100.0000) acc1_neg: 66.927 acc5_neg: 100.000 time: 0.1142 data: 0.0247 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 99.543 acc5_pos: 100.000 loss: 0.6441 (0.3435) acc1: 62.7604 (86.5157) acc5: 100.0000 (100.0000) acc1_neg: 97.674 acc5_neg: 100.000 time: 0.1114 data: 0.0246 max mem: 3541 +Test: Total time: 0:07:27 (0.1874 s / it) +* Overall Acc@1 86.516 Acc@5 100.000 + target == real Acc@1 85.160 Acc@5 100.000 + target == fake Acc@1 86.611 Acc@5 100.000 + loss 0.343 + PR-AUC 0.9942 ROC-AUC 0.9309 EER 14.03% (thr=0.5098) +Accuracy of the network on the 916267 test images: 86.5% +[rank0]:[W725 16:52:02.976589783 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 16:52:24.629000 2558900 torch/distributed/run.py:766] +W0725 16:52:24.629000 2558900 torch/distributed/run.py:766] ***************************************** +W0725 16:52:24.629000 2558900 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 16:52:24.629000 2558900 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +[rank7]:[W725 16:52:59.029330696 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 16:52:59.043326088 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 16:53:00.648580448 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 16:53:00.677455961 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 16:53:00.738388887 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +[rank4]:[W725 16:53:00.871668520 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 16:53:00.883075728 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 16:53:00.928640007 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed1_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed1_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 0:51:51 acc1_pos: 87.760 acc5_pos: 100.000 loss: 0.4116 (0.4116) acc1: 87.7604 (87.7604) acc5: 100.0000 (100.0000) time: 6.0177 data: 3.7272 max mem: 3531 +Test: [ 10/517] eta: 0:05:21 acc1_pos: 75.260 acc5_pos: 100.000 loss: 0.4630 (0.4701) acc1: 79.9479 (81.6525) acc5: 100.0000 (100.0000) time: 0.6340 data: 0.3446 max mem: 3531 +Test: [ 20/517] eta: 0:03:09 acc1_pos: 80.208 acc5_pos: 100.000 loss: 0.4630 (0.4894) acc1: 79.9479 (81.3616) acc5: 100.0000 (100.0000) time: 0.0993 data: 0.0100 max mem: 3531 +Test: [ 30/517] eta: 0:02:31 acc1_pos: 62.500 acc5_pos: 100.000 loss: 0.4848 (0.4900) acc1: 80.2083 (80.6620) acc5: 100.0000 (100.0000) time: 0.1321 data: 0.0401 max mem: 3531 +Test: [ 40/517] eta: 0:02:25 acc1_pos: 67.448 acc5_pos: 100.000 loss: 0.5262 (0.5027) acc1: 74.4792 (79.2429) acc5: 100.0000 (100.0000) time: 0.2263 data: 0.1324 max mem: 3531 +Test: [ 50/517] eta: 0:02:19 acc1_pos: 97.706 acc5_pos: 100.000 loss: 0.5527 (0.5206) acc1: 73.1771 (76.5982) acc5: 100.0000 (100.0000) acc1_neg: 20.573 acc5_neg: 100.000 time: 0.2799 data: 0.1839 max mem: 3531 +Test: [ 60/517] eta: 0:02:13 acc1_pos: 97.706 acc5_pos: 100.000 loss: 0.7257 (0.5699) acc1: 43.4896 (70.3338) acc5: 100.0000 (100.0000) acc1_neg: 40.885 acc5_neg: 100.000 time: 0.2642 data: 0.1672 max mem: 3531 +Test: [ 70/517] eta: 0:02:05 acc1_pos: 97.706 acc5_pos: 100.000 loss: 0.8937 (0.6192) acc1: 27.8646 (64.8951) acc5: 100.0000 (100.0000) acc1_neg: 24.219 acc5_neg: 100.000 time: 0.2399 data: 0.1457 max mem: 3531 +Test: [ 80/517] eta: 0:02:02 acc1_pos: 97.706 acc5_pos: 100.000 loss: 0.9691 (0.6680) acc1: 21.0938 (59.4329) acc5: 100.0000 (100.0000) acc1_neg: 17.448 acc5_neg: 100.000 time: 0.2449 data: 0.1508 max mem: 3531 +Test: [ 90/517] eta: 0:01:59 acc1_pos: 97.706 acc5_pos: 100.000 loss: 0.9204 (0.6850) acc1: 25.0000 (56.7966) acc5: 100.0000 (100.0000) acc1_neg: 27.865 acc5_neg: 100.000 time: 0.2749 data: 0.1804 max mem: 3531 +Test: [100/517] eta: 0:01:56 acc1_pos: 97.706 acc5_pos: 100.000 loss: 0.7565 (0.6961) acc1: 36.4583 (55.0459) acc5: 100.0000 (100.0000) acc1_neg: 47.396 acc5_neg: 100.000 time: 0.2699 data: 0.1759 max mem: 3531 +Test: [110/517] eta: 0:01:51 acc1_pos: 97.706 acc5_pos: 100.000 loss: 0.7274 (0.6948) acc1: 46.8750 (54.9010) acc5: 100.0000 (100.0000) acc1_neg: 51.823 acc5_neg: 100.000 time: 0.2499 data: 0.1569 max mem: 3531 +Test: [120/517] eta: 0:01:48 acc1_pos: 97.706 acc5_pos: 100.000 loss: 0.7150 (0.6979) acc1: 49.7396 (54.4766) acc5: 100.0000 (100.0000) acc1_neg: 49.479 acc5_neg: 100.000 time: 0.2499 data: 0.1573 max mem: 3531 +Test: [130/517] eta: 0:01:46 acc1_pos: 97.706 acc5_pos: 100.000 loss: 0.7150 (0.6976) acc1: 51.3021 (54.5424) acc5: 100.0000 (100.0000) acc1_neg: 58.333 acc5_neg: 100.000 time: 0.2799 data: 0.1872 max mem: 3531 +Test: [140/517] eta: 0:01:42 acc1_pos: 97.706 acc5_pos: 100.000 loss: 0.7165 (0.7001) acc1: 51.3021 (54.2332) acc5: 100.0000 (100.0000) acc1_neg: 41.927 acc5_neg: 100.000 time: 0.2699 data: 0.1747 max mem: 3531 +Test: [150/517] eta: 0:01:40 acc1_pos: 97.706 acc5_pos: 100.000 loss: 0.7155 (0.6990) acc1: 52.0833 (54.3443) acc5: 100.0000 (100.0000) acc1_neg: 52.083 acc5_neg: 100.000 time: 0.2549 data: 0.1566 max mem: 3531 +Test: [160/517] eta: 0:01:36 acc1_pos: 97.706 acc5_pos: 100.000 loss: 0.7362 (0.7016) acc1: 50.2604 (54.0615) acc5: 100.0000 (100.0000) acc1_neg: 58.594 acc5_neg: 100.000 time: 0.2549 data: 0.1585 max mem: 3531 +Test: [170/517] eta: 0:01:33 acc1_pos: 97.706 acc5_pos: 100.000 loss: 0.7428 (0.7017) acc1: 50.0000 (54.0677) acc5: 100.0000 (100.0000) acc1_neg: 61.979 acc5_neg: 100.000 time: 0.2499 data: 0.1547 max mem: 3531 +Test: [180/517] eta: 0:01:30 acc1_pos: 97.706 acc5_pos: 100.000 loss: 0.6932 (0.7002) acc1: 56.2500 (54.1480) acc5: 100.0000 (100.0000) acc1_neg: 62.760 acc5_neg: 100.000 time: 0.2599 data: 0.1646 max mem: 3531 +Test: [190/517] eta: 0:01:27 acc1_pos: 97.706 acc5_pos: 100.000 loss: 0.3708 (0.6793) acc1: 86.1979 (56.1955) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.2549 data: 0.1583 max mem: 3531 +Test: [200/517] eta: 0:01:24 acc1_pos: 97.706 acc5_pos: 100.000 loss: 0.3607 (0.6705) acc1: 86.9792 (57.1751) acc5: 100.0000 (100.0000) acc1_neg: 69.271 acc5_neg: 100.000 time: 0.2499 data: 0.1534 max mem: 3531 +Test: [210/517] eta: 0:01:22 acc1_pos: 97.706 acc5_pos: 100.000 loss: 0.5681 (0.6763) acc1: 59.8958 (56.5203) acc5: 100.0000 (100.0000) acc1_neg: 48.438 acc5_neg: 100.000 time: 0.2678 data: 0.1730 max mem: 3532 +Test: [220/517] eta: 0:01:19 acc1_pos: 97.706 acc5_pos: 100.000 loss: 0.8084 (0.6837) acc1: 39.5833 (55.6844) acc5: 100.0000 (100.0000) acc1_neg: 27.344 acc5_neg: 100.000 time: 0.2628 data: 0.1659 max mem: 3532 +Test: [230/517] eta: 0:01:16 acc1_pos: 97.706 acc5_pos: 100.000 loss: 0.8504 (0.6908) acc1: 34.1146 (54.6717) acc5: 100.0000 (100.0000) acc1_neg: 26.562 acc5_neg: 100.000 time: 0.2449 data: 0.1486 max mem: 3532 +Test: [240/517] eta: 0:01:13 acc1_pos: 97.706 acc5_pos: 100.000 loss: 0.8372 (0.6977) acc1: 30.7292 (53.8112) acc5: 100.0000 (100.0000) acc1_neg: 18.229 acc5_neg: 100.000 time: 0.2549 data: 0.1594 max mem: 3532 +Test: [250/517] eta: 0:01:10 acc1_pos: 97.706 acc5_pos: 100.000 loss: 0.8246 (0.7034) acc1: 32.5521 (53.0181) acc5: 100.0000 (100.0000) acc1_neg: 35.938 acc5_neg: 100.000 time: 0.2549 data: 0.1589 max mem: 3532 +Test: [260/517] eta: 0:01:08 acc1_pos: 97.706 acc5_pos: 100.000 loss: 0.8246 (0.7077) acc1: 33.5938 (52.4505) acc5: 100.0000 (100.0000) acc1_neg: 61.458 acc5_neg: 100.000 time: 0.2549 data: 0.1592 max mem: 3532 +Test: [270/517] eta: 0:01:05 acc1_pos: 97.706 acc5_pos: 100.000 loss: 0.7781 (0.7100) acc1: 44.2708 (52.1410) acc5: 100.0000 (100.0000) acc1_neg: 46.615 acc5_neg: 100.000 time: 0.2599 data: 0.1632 max mem: 3532 +Test: [280/517] eta: 0:01:02 acc1_pos: 97.706 acc5_pos: 100.000 loss: 0.7513 (0.7115) acc1: 46.6146 (52.0305) acc5: 100.0000 (100.0000) acc1_neg: 34.896 acc5_neg: 100.000 time: 0.2599 data: 0.1621 max mem: 3532 +Test: [290/517] eta: 0:01:00 acc1_pos: 97.706 acc5_pos: 100.000 loss: 0.7450 (0.7138) acc1: 47.9167 (51.8140) acc5: 100.0000 (100.0000) acc1_neg: 42.188 acc5_neg: 100.000 time: 0.2549 data: 0.1595 max mem: 3532 +Test: [300/517] eta: 0:00:57 acc1_pos: 97.706 acc5_pos: 100.000 loss: 0.7307 (0.7132) acc1: 49.7396 (51.8030) acc5: 100.0000 (100.0000) acc1_neg: 26.823 acc5_neg: 100.000 time: 0.2549 data: 0.1597 max mem: 3532 +Test: [310/517] eta: 0:00:54 acc1_pos: 97.706 acc5_pos: 100.000 loss: 0.8299 (0.7196) acc1: 34.1146 (50.9965) acc5: 100.0000 (100.0000) acc1_neg: 29.948 acc5_neg: 100.000 time: 0.2599 data: 0.1623 max mem: 3532 +Test: [320/517] eta: 0:00:52 acc1_pos: 97.706 acc5_pos: 100.000 loss: 0.8395 (0.7208) acc1: 32.8125 (50.9021) acc5: 100.0000 (100.0000) acc1_neg: 51.562 acc5_neg: 100.000 time: 0.2599 data: 0.1614 max mem: 3532 +Test: [330/517] eta: 0:00:49 acc1_pos: 97.706 acc5_pos: 100.000 loss: 0.6872 (0.7193) acc1: 57.2917 (51.1550) acc5: 100.0000 (100.0000) acc1_neg: 57.292 acc5_neg: 100.000 time: 0.2599 data: 0.1623 max mem: 3532 +Test: [340/517] eta: 0:00:46 acc1_pos: 97.706 acc5_pos: 100.000 loss: 0.6543 (0.7194) acc1: 57.8125 (51.1089) acc5: 100.0000 (100.0000) acc1_neg: 26.823 acc5_neg: 100.000 time: 0.2549 data: 0.1605 max mem: 3532 +Test: [350/517] eta: 0:00:44 acc1_pos: 97.706 acc5_pos: 100.000 loss: 0.8263 (0.7245) acc1: 35.1562 (50.4926) acc5: 100.0000 (100.0000) acc1_neg: 12.760 acc5_neg: 100.000 time: 0.2599 data: 0.1620 max mem: 3532 +Test: [360/517] eta: 0:00:41 acc1_pos: 97.706 acc5_pos: 100.000 loss: 0.8446 (0.7266) acc1: 32.8125 (50.3116) acc5: 100.0000 (100.0000) acc1_neg: 48.698 acc5_neg: 100.000 time: 0.2599 data: 0.1602 max mem: 3532 +Test: [370/517] eta: 0:00:38 acc1_pos: 97.706 acc5_pos: 100.000 loss: 0.5708 (0.7215) acc1: 69.7917 (50.9687) acc5: 100.0000 (100.0000) acc1_neg: 69.792 acc5_neg: 100.000 time: 0.2549 data: 0.1581 max mem: 3532 +Test: [380/517] eta: 0:00:36 acc1_pos: 97.706 acc5_pos: 100.000 loss: 0.5708 (0.7189) acc1: 69.7917 (51.2891) acc5: 100.0000 (100.0000) acc1_neg: 66.927 acc5_neg: 100.000 time: 0.2549 data: 0.1576 max mem: 3532 +Test: [390/517] eta: 0:00:33 acc1_pos: 97.706 acc5_pos: 100.000 loss: 0.6061 (0.7161) acc1: 64.5833 (51.6491) acc5: 100.0000 (100.0000) acc1_neg: 66.927 acc5_neg: 100.000 time: 0.2499 data: 0.1537 max mem: 3532 +Test: [400/517] eta: 0:00:30 acc1_pos: 97.706 acc5_pos: 100.000 loss: 0.5840 (0.7130) acc1: 69.0104 (51.9911) acc5: 100.0000 (100.0000) acc1_neg: 70.833 acc5_neg: 100.000 time: 0.2499 data: 0.1536 max mem: 3532 +Test: [410/517] eta: 0:00:28 acc1_pos: 97.706 acc5_pos: 100.000 loss: 0.6014 (0.7113) acc1: 63.5417 (52.2386) acc5: 100.0000 (100.0000) acc1_neg: 49.479 acc5_neg: 100.000 time: 0.2549 data: 0.1584 max mem: 3532 +Test: [420/517] eta: 0:00:25 acc1_pos: 97.706 acc5_pos: 100.000 loss: 0.7054 (0.7136) acc1: 51.5625 (51.9448) acc5: 100.0000 (100.0000) acc1_neg: 30.208 acc5_neg: 100.000 time: 0.2549 data: 0.1597 max mem: 3532 +Test: [430/517] eta: 0:00:22 acc1_pos: 97.706 acc5_pos: 100.000 loss: 0.8364 (0.7169) acc1: 37.2396 (51.6048) acc5: 100.0000 (100.0000) acc1_neg: 42.708 acc5_neg: 100.000 time: 0.2549 data: 0.1559 max mem: 3532 +Test: [440/517] eta: 0:00:20 acc1_pos: 97.706 acc5_pos: 100.000 loss: 0.7570 (0.7166) acc1: 44.0104 (51.5932) acc5: 100.0000 (100.0000) acc1_neg: 49.219 acc5_neg: 100.000 time: 0.2549 data: 0.1551 max mem: 3533 +Test: [450/517] eta: 0:00:17 acc1_pos: 97.706 acc5_pos: 100.000 loss: 0.7069 (0.7157) acc1: 51.5625 (51.7063) acc5: 100.0000 (100.0000) acc1_neg: 73.698 acc5_neg: 100.000 time: 0.2499 data: 0.1552 max mem: 3533 +Test: [460/517] eta: 0:00:14 acc1_pos: 97.706 acc5_pos: 100.000 loss: 0.6146 (0.7137) acc1: 61.4583 (52.0020) acc5: 100.0000 (100.0000) acc1_neg: 66.927 acc5_neg: 100.000 time: 0.2549 data: 0.1595 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 97.706 acc5_pos: 100.000 loss: 0.6146 (0.7119) acc1: 61.4583 (52.1701) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.2549 data: 0.1572 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 97.706 acc5_pos: 100.000 loss: 0.5834 (0.7082) acc1: 65.6250 (52.6166) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.2315 data: 0.1375 max mem: 3533 +Test: [490/517] eta: 0:00:06 acc1_pos: 97.706 acc5_pos: 100.000 loss: 0.5834 (0.7085) acc1: 67.1875 (52.6180) acc5: 100.0000 (100.0000) acc1_neg: 52.604 acc5_neg: 100.000 time: 0.1949 data: 0.1049 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 97.706 acc5_pos: 100.000 loss: 0.8660 (0.7133) acc1: 33.5938 (52.0844) acc5: 100.0000 (100.0000) acc1_neg: 36.198 acc5_neg: 100.000 time: 0.1820 data: 0.0925 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 97.706 acc5_pos: 100.000 loss: 0.8135 (0.7129) acc1: 33.3333 (52.1827) acc5: 100.0000 (100.0000) acc1_neg: 54.427 acc5_neg: 100.000 time: 0.1800 data: 0.0905 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 97.706 acc5_pos: 100.000 loss: 0.8095 (0.7140) acc1: 40.3646 (52.1063) acc5: 100.0000 (100.0000) acc1_neg: 24.028 acc5_neg: 100.000 time: 0.1765 data: 0.0853 max mem: 3533 +Test: Total time: 0:02:12 (0.2555 s / it) +* Overall Acc@1 52.106 Acc@5 100.000 + target == real Acc@1 79.078 Acc@5 100.000 + target == fake Acc@1 49.308 Acc@5 100.000 + loss 0.714 + PR-AUC 0.9545 ROC-AUC 0.7397 EER 32.59% (thr=0.4119) +Accuracy of the network on the 198427 test images: 52.1% +[rank0]:[W725 16:55:22.910480663 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0725 16:55:45.262000 2559472 torch/distributed/run.py:766] +W0725 16:55:45.262000 2559472 torch/distributed/run.py:766] ***************************************** +W0725 16:55:45.262000 2559472 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 16:55:45.262000 2559472 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W725 16:56:22.115687052 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 16:56:22.282208166 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 16:56:22.387116046 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 16:56:23.645620330 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 16:56:23.695281861 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 16:56:23.807685507 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 16:56:23.818166590 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 16:56:23.859177405 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed1_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed1_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:31:05 acc1_pos: 65.885 acc5_pos: 100.000 loss: 0.5822 (0.5822) acc1: 65.8854 (65.8854) acc5: 100.0000 (100.0000) time: 4.7961 data: 2.9615 max mem: 3531 +Test: [ 10/389] eta: 0:03:18 acc1_pos: 98.958 acc5_pos: 100.000 loss: 0.3013 (0.3443) acc1: 98.9583 (89.2282) acc5: 100.0000 (100.0000) time: 0.5236 data: 0.2755 max mem: 3531 +Test: [ 20/389] eta: 0:02:14 acc1_pos: 84.375 acc5_pos: 100.000 loss: 0.3672 (0.3836) acc1: 88.5417 (88.0332) acc5: 100.0000 (100.0000) time: 0.1432 data: 0.0471 max mem: 3531 +Test: [ 30/389] eta: 0:01:50 acc1_pos: 96.094 acc5_pos: 100.000 loss: 0.3926 (0.3703) acc1: 89.3229 (89.6253) acc5: 100.0000 (100.0000) time: 0.1910 data: 0.0927 max mem: 3531 +Test: [ 40/389] eta: 0:01:40 acc1_pos: 88.281 acc5_pos: 100.000 loss: 0.3967 (0.4043) acc1: 89.3229 (86.7886) acc5: 100.0000 (100.0000) time: 0.2086 data: 0.1161 max mem: 3531 +Test: [ 50/389] eta: 0:01:32 acc1_pos: 95.833 acc5_pos: 100.000 loss: 0.3580 (0.3802) acc1: 94.7917 (88.5825) acc5: 100.0000 (100.0000) time: 0.2203 data: 0.1281 max mem: 3531 +Test: [ 60/389] eta: 0:01:22 acc1_pos: 92.188 acc5_pos: 100.000 loss: 0.3029 (0.3694) acc1: 96.0938 (89.7712) acc5: 100.0000 (100.0000) time: 0.1738 data: 0.0787 max mem: 3531 +Test: [ 70/389] eta: 0:01:21 acc1_pos: 61.458 acc5_pos: 100.000 loss: 0.3265 (0.3780) acc1: 95.8333 (89.2532) acc5: 100.0000 (100.0000) time: 0.2085 data: 0.1124 max mem: 3531 +Test: [ 80/389] eta: 0:01:15 acc1_pos: 99.219 acc5_pos: 100.000 loss: 0.4276 (0.4101) acc1: 84.6354 (86.2043) acc5: 100.0000 (100.0000) time: 0.2189 data: 0.1223 max mem: 3531 +Test: [ 90/389] eta: 0:01:08 acc1_pos: 97.135 acc5_pos: 100.000 loss: 0.3344 (0.3940) acc1: 94.0104 (87.4742) acc5: 100.0000 (100.0000) time: 0.1417 data: 0.0470 max mem: 3531 +Test: [100/389] eta: 0:01:05 acc1_pos: 86.719 acc5_pos: 100.000 loss: 0.3363 (0.4025) acc1: 94.0104 (87.0539) acc5: 100.0000 (100.0000) time: 0.1633 data: 0.0684 max mem: 3531 +Test: [110/389] eta: 0:01:02 acc1_pos: 96.615 acc5_pos: 100.000 loss: 0.4790 (0.4134) acc1: 84.6354 (85.5856) acc5: 100.0000 (100.0000) time: 0.1884 data: 0.0926 max mem: 3531 +Test: [120/389] eta: 0:00:58 acc1_pos: 85.417 acc5_pos: 100.000 loss: 0.3650 (0.4141) acc1: 85.6771 (85.3564) acc5: 100.0000 (100.0000) time: 0.1718 data: 0.0790 max mem: 3531 +Test: [130/389] eta: 0:00:55 acc1_pos: 74.847 acc5_pos: 100.000 loss: 0.5260 (0.4429) acc1: 69.0104 (82.5024) acc5: 100.0000 (100.0000) acc1_neg: 70.573 acc5_neg: 100.000 time: 0.1713 data: 0.0782 max mem: 3531 +Test: [140/389] eta: 0:00:52 acc1_pos: 74.847 acc5_pos: 100.000 loss: 0.6852 (0.4657) acc1: 53.1250 (79.9276) acc5: 100.0000 (100.0000) acc1_neg: 27.344 acc5_neg: 100.000 time: 0.1703 data: 0.0763 max mem: 3531 +Test: [150/389] eta: 0:00:49 acc1_pos: 74.847 acc5_pos: 100.000 loss: 0.8233 (0.5051) acc1: 28.9062 (76.7781) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1681 data: 0.0751 max mem: 3531 +Test: [160/389] eta: 0:00:47 acc1_pos: 74.847 acc5_pos: 100.000 loss: 0.7564 (0.5119) acc1: 30.9896 (76.3102) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.1627 data: 0.0708 max mem: 3531 +Test: [170/389] eta: 0:00:44 acc1_pos: 74.847 acc5_pos: 100.000 loss: 0.5916 (0.5247) acc1: 65.3646 (74.9543) acc5: 100.0000 (100.0000) acc1_neg: 41.146 acc5_neg: 100.000 time: 0.1602 data: 0.0673 max mem: 3531 +Test: [180/389] eta: 0:00:41 acc1_pos: 74.847 acc5_pos: 100.000 loss: 0.6355 (0.5401) acc1: 60.6771 (73.2375) acc5: 100.0000 (100.0000) acc1_neg: 60.677 acc5_neg: 100.000 time: 0.1605 data: 0.0657 max mem: 3531 +Test: [190/389] eta: 0:00:39 acc1_pos: 74.847 acc5_pos: 100.000 loss: 0.7683 (0.5562) acc1: 43.4896 (71.4128) acc5: 100.0000 (100.0000) acc1_neg: 0.000 acc5_neg: 100.000 time: 0.1482 data: 0.0530 max mem: 3531 +Test: [200/389] eta: 0:00:36 acc1_pos: 74.847 acc5_pos: 100.000 loss: 0.9351 (0.5826) acc1: 16.9271 (69.2462) acc5: 100.0000 (100.0000) acc1_neg: 5.208 acc5_neg: 100.000 time: 0.1542 data: 0.0595 max mem: 3531 +Test: [210/389] eta: 0:00:34 acc1_pos: 74.847 acc5_pos: 100.000 loss: 0.5869 (0.5869) acc1: 16.9271 (68.9882) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1711 data: 0.0786 max mem: 3532 +Test: [220/389] eta: 0:00:32 acc1_pos: 74.847 acc5_pos: 100.000 loss: 0.4240 (0.5839) acc1: 83.5938 (69.2956) acc5: 100.0000 (100.0000) acc1_neg: 30.729 acc5_neg: 100.000 time: 0.1495 data: 0.0578 max mem: 3532 +Test: [230/389] eta: 0:00:30 acc1_pos: 74.847 acc5_pos: 100.000 loss: 0.6289 (0.5960) acc1: 65.1042 (68.3047) acc5: 100.0000 (100.0000) acc1_neg: 65.885 acc5_neg: 100.000 time: 0.1455 data: 0.0529 max mem: 3532 +Test: [240/389] eta: 0:00:28 acc1_pos: 74.847 acc5_pos: 100.000 loss: 0.7214 (0.5990) acc1: 51.8229 (67.9806) acc5: 100.0000 (100.0000) acc1_neg: 53.646 acc5_neg: 100.000 time: 0.1825 data: 0.0906 max mem: 3532 +Test: [250/389] eta: 0:00:26 acc1_pos: 74.847 acc5_pos: 100.000 loss: 0.6262 (0.5998) acc1: 63.8021 (67.8484) acc5: 100.0000 (100.0000) acc1_neg: 63.542 acc5_neg: 100.000 time: 0.1943 data: 0.1031 max mem: 3532 +Test: [260/389] eta: 0:00:24 acc1_pos: 74.847 acc5_pos: 100.000 loss: 0.6085 (0.6008) acc1: 65.1042 (67.8750) acc5: 100.0000 (100.0000) acc1_neg: 14.844 acc5_neg: 100.000 time: 0.1934 data: 0.1026 max mem: 3532 +Test: [270/389] eta: 0:00:22 acc1_pos: 74.847 acc5_pos: 100.000 loss: 0.7461 (0.6192) acc1: 44.0104 (66.2660) acc5: 100.0000 (100.0000) acc1_neg: 48.958 acc5_neg: 100.000 time: 0.1907 data: 0.0978 max mem: 3532 +Test: [280/389] eta: 0:00:20 acc1_pos: 74.847 acc5_pos: 100.000 loss: 0.9735 (0.6299) acc1: 27.8646 (65.2423) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1902 data: 0.0968 max mem: 3532 +Test: [290/389] eta: 0:00:18 acc1_pos: 74.847 acc5_pos: 100.000 loss: 0.6639 (0.6281) acc1: 60.9375 (65.4451) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1913 data: 0.1003 max mem: 3532 +Test: [300/389] eta: 0:00:17 acc1_pos: 74.847 acc5_pos: 100.000 loss: 0.5086 (0.6251) acc1: 76.3021 (65.8768) acc5: 100.0000 (100.0000) acc1_neg: 68.229 acc5_neg: 100.000 time: 0.1949 data: 0.1024 max mem: 3532 +Test: [310/389] eta: 0:00:15 acc1_pos: 74.847 acc5_pos: 100.000 loss: 0.6061 (0.6361) acc1: 62.7604 (64.9057) acc5: 100.0000 (100.0000) acc1_neg: 2.344 acc5_neg: 100.000 time: 0.1889 data: 0.0960 max mem: 3532 +Test: [320/389] eta: 0:00:13 acc1_pos: 74.847 acc5_pos: 100.000 loss: 0.8868 (0.6424) acc1: 35.6771 (64.2499) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1835 data: 0.0915 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 74.847 acc5_pos: 100.000 loss: 1.1485 (0.6577) acc1: 7.5521 (62.6802) acc5: 100.0000 (100.0000) acc1_neg: 14.583 acc5_neg: 100.000 time: 0.1753 data: 0.0831 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 74.847 acc5_pos: 100.000 loss: 0.5415 (0.6511) acc1: 61.9792 (63.5294) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1651 data: 0.0741 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 74.847 acc5_pos: 100.000 loss: 0.4232 (0.6576) acc1: 91.9271 (63.1076) acc5: 100.0000 (100.0000) acc1_neg: 21.354 acc5_neg: 100.000 time: 0.1519 data: 0.0615 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 74.847 acc5_pos: 100.000 loss: 0.5771 (0.6575) acc1: 67.1875 (63.0439) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.1364 data: 0.0460 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 74.847 acc5_pos: 100.000 loss: 0.5690 (0.6595) acc1: 71.6146 (62.9071) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1218 data: 0.0310 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 74.847 acc5_pos: 100.000 loss: 0.6205 (0.6594) acc1: 53.9062 (62.7939) acc5: 100.0000 (100.0000) acc1_neg: 52.865 acc5_neg: 100.000 time: 0.1155 data: 0.0253 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 74.847 acc5_pos: 100.000 loss: 0.5048 (0.6572) acc1: 79.5082 (63.0679) acc5: 100.0000 (100.0000) acc1_neg: 79.508 acc5_neg: 100.000 time: 0.1241 data: 0.0309 max mem: 3532 +Test: Total time: 0:01:10 (0.1821 s / it) +* Overall Acc@1 63.068 Acc@5 100.000 + target == real Acc@1 84.730 Acc@5 100.000 + target == fake Acc@1 52.999 Acc@5 100.000 + loss 0.657 + PR-AUC 0.8691 ROC-AUC 0.7700 EER 30.10% (thr=0.3787) +Accuracy of the network on the 149358 test images: 63.1% +[rank0]:[W725 16:57:43.965110899 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 16:58:04.574000 2560030 torch/distributed/run.py:766] +W0725 16:58:04.574000 2560030 torch/distributed/run.py:766] ***************************************** +W0725 16:58:04.574000 2560030 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 16:58:04.574000 2560030 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 16:58:40.359244955 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +[rank1]:[W725 16:58:41.840589018 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 16:58:41.852463092 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 16:58:41.934200702 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 16:58:41.116225555 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 16:58:41.165224358 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 16:58:41.214199672 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 16:58:41.256895052 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed1_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed1_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:31:37 acc1_pos: 3.906 acc5_pos: 100.000 loss: 1.7749 (1.7749) acc1: 3.9062 (3.9062) acc5: 100.0000 (100.0000) time: 5.6975 data: 4.2398 max mem: 3531 +Test: [ 10/333] eta: 0:03:14 acc1_pos: 22.396 acc5_pos: 100.000 loss: 1.4728 (1.4536) acc1: 11.4583 (16.0511) acc5: 100.0000 (100.0000) time: 0.6013 data: 0.3857 max mem: 3531 +Test: [ 20/333] eta: 0:01:53 acc1_pos: 28.906 acc5_pos: 100.000 loss: 1.3503 (1.3749) acc1: 20.3125 (19.5437) acc5: 100.0000 (100.0000) time: 0.0958 data: 0.0003 max mem: 3531 +Test: [ 30/333] eta: 0:01:26 acc1_pos: 52.604 acc5_pos: 100.000 loss: 1.0492 (1.1902) acc1: 30.2083 (28.8558) acc5: 100.0000 (100.0000) time: 0.1108 data: 0.0147 max mem: 3531 +Test: [ 40/333] eta: 0:01:16 acc1_pos: 59.896 acc5_pos: 100.000 loss: 0.7500 (1.0803) acc1: 52.3438 (35.3913) acc5: 100.0000 (100.0000) time: 0.1549 data: 0.0630 max mem: 3531 +Test: [ 50/333] eta: 0:01:08 acc1_pos: 54.948 acc5_pos: 100.000 loss: 0.7437 (1.0149) acc1: 54.9479 (39.1902) acc5: 100.0000 (100.0000) time: 0.1740 data: 0.0807 max mem: 3531 +Test: [ 60/333] eta: 0:01:01 acc1_pos: 59.896 acc5_pos: 100.000 loss: 0.7340 (0.9734) acc1: 55.7292 (41.6539) acc5: 100.0000 (100.0000) time: 0.1549 data: 0.0599 max mem: 3531 +Test: [ 70/333] eta: 0:00:57 acc1_pos: 55.208 acc5_pos: 100.000 loss: 0.7356 (0.9444) acc1: 52.0833 (43.2585) acc5: 100.0000 (100.0000) time: 0.1649 data: 0.0697 max mem: 3531 +Test: [ 80/333] eta: 0:00:54 acc1_pos: 51.042 acc5_pos: 100.000 loss: 0.7888 (0.9260) acc1: 51.0417 (44.0940) acc5: 100.0000 (100.0000) time: 0.1749 data: 0.0796 max mem: 3531 +Test: [ 90/333] eta: 0:00:52 acc1_pos: 72.396 acc5_pos: 100.000 loss: 0.7049 (0.8924) acc1: 57.8125 (46.5860) acc5: 100.0000 (100.0000) time: 0.1949 data: 0.0987 max mem: 3531 +Test: [100/333] eta: 0:00:49 acc1_pos: 69.271 acc5_pos: 100.000 loss: 0.5734 (0.8596) acc1: 69.5312 (48.9171) acc5: 100.0000 (100.0000) time: 0.1999 data: 0.1033 max mem: 3531 +Test: [110/333] eta: 0:00:46 acc1_pos: 57.812 acc5_pos: 100.000 loss: 0.5823 (0.8380) acc1: 65.3646 (50.4528) acc5: 100.0000 (100.0000) time: 0.1853 data: 0.0901 max mem: 3531 +Test: [120/333] eta: 0:00:44 acc1_pos: 63.542 acc5_pos: 100.000 loss: 0.5823 (0.8164) acc1: 67.7083 (52.0790) acc5: 100.0000 (100.0000) time: 0.1872 data: 0.0913 max mem: 3531 +Test: [130/333] eta: 0:00:43 acc1_pos: 84.115 acc5_pos: 100.000 loss: 0.5472 (0.7959) acc1: 72.6562 (53.6836) acc5: 100.0000 (100.0000) time: 0.2395 data: 0.1442 max mem: 3531 +Test: [140/333] eta: 0:00:39 acc1_pos: 62.240 acc5_pos: 100.000 loss: 0.5192 (0.7759) acc1: 73.1771 (55.2416) acc5: 100.0000 (100.0000) time: 0.1977 data: 0.1011 max mem: 3531 +Test: [150/333] eta: 0:00:38 acc1_pos: 68.750 acc5_pos: 100.000 loss: 0.5455 (0.7631) acc1: 68.7500 (56.1189) acc5: 100.0000 (100.0000) time: 0.1854 data: 0.0883 max mem: 3531 +Test: [160/333] eta: 0:00:36 acc1_pos: 60.197 acc5_pos: 100.000 loss: 0.6015 (0.7557) acc1: 64.0625 (56.5509) acc5: 100.0000 (100.0000) acc1_neg: 51.823 acc5_neg: 100.000 time: 0.2599 data: 0.1643 max mem: 3531 +Test: [170/333] eta: 0:00:33 acc1_pos: 60.197 acc5_pos: 100.000 loss: 0.6740 (0.7528) acc1: 58.8542 (56.5744) acc5: 100.0000 (100.0000) acc1_neg: 63.021 acc5_neg: 100.000 time: 0.1894 data: 0.0919 max mem: 3531 +Test: [180/333] eta: 0:00:31 acc1_pos: 60.197 acc5_pos: 100.000 loss: 0.6412 (0.7439) acc1: 63.0208 (57.1564) acc5: 100.0000 (100.0000) acc1_neg: 64.062 acc5_neg: 100.000 time: 0.1503 data: 0.0529 max mem: 3531 +Test: [190/333] eta: 0:00:30 acc1_pos: 60.197 acc5_pos: 100.000 loss: 0.5643 (0.7347) acc1: 67.7083 (57.7743) acc5: 100.0000 (100.0000) acc1_neg: 63.802 acc5_neg: 100.000 time: 0.2299 data: 0.1331 max mem: 3531 +Test: [200/333] eta: 0:00:27 acc1_pos: 60.197 acc5_pos: 100.000 loss: 0.5596 (0.7265) acc1: 68.7500 (58.3437) acc5: 100.0000 (100.0000) acc1_neg: 66.146 acc5_neg: 100.000 time: 0.2318 data: 0.1360 max mem: 3531 +Test: [210/333] eta: 0:00:25 acc1_pos: 60.197 acc5_pos: 100.000 loss: 0.5597 (0.7198) acc1: 70.0521 (58.8036) acc5: 100.0000 (100.0000) acc1_neg: 71.094 acc5_neg: 100.000 time: 0.1821 data: 0.0899 max mem: 3532 +Test: [220/333] eta: 0:00:23 acc1_pos: 60.197 acc5_pos: 100.000 loss: 0.5716 (0.7169) acc1: 67.9688 (58.8989) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.1826 data: 0.0909 max mem: 3532 +Test: [230/333] eta: 0:00:20 acc1_pos: 60.197 acc5_pos: 100.000 loss: 0.5113 (0.7067) acc1: 77.0833 (59.8801) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1527 data: 0.0570 max mem: 3532 +Test: [240/333] eta: 0:00:18 acc1_pos: 60.197 acc5_pos: 100.000 loss: 0.4990 (0.7030) acc1: 77.0833 (60.1098) acc5: 100.0000 (100.0000) acc1_neg: 60.156 acc5_neg: 100.000 time: 0.1661 data: 0.0704 max mem: 3532 +Test: [250/333] eta: 0:00:16 acc1_pos: 60.197 acc5_pos: 100.000 loss: 0.6368 (0.7001) acc1: 62.7604 (60.2631) acc5: 100.0000 (100.0000) acc1_neg: 61.719 acc5_neg: 100.000 time: 0.2001 data: 0.1071 max mem: 3532 +Test: [260/333] eta: 0:00:14 acc1_pos: 60.197 acc5_pos: 100.000 loss: 0.6099 (0.6962) acc1: 64.8438 (60.5314) acc5: 100.0000 (100.0000) acc1_neg: 66.146 acc5_neg: 100.000 time: 0.1911 data: 0.0959 max mem: 3532 +Test: [270/333] eta: 0:00:12 acc1_pos: 60.197 acc5_pos: 100.000 loss: 0.5720 (0.6911) acc1: 68.7500 (60.8856) acc5: 100.0000 (100.0000) acc1_neg: 64.583 acc5_neg: 100.000 time: 0.1547 data: 0.0564 max mem: 3532 +Test: [280/333] eta: 0:00:10 acc1_pos: 60.197 acc5_pos: 100.000 loss: 0.5720 (0.6910) acc1: 66.4062 (60.7994) acc5: 100.0000 (100.0000) acc1_neg: 60.156 acc5_neg: 100.000 time: 0.1774 data: 0.0810 max mem: 3532 +Test: [290/333] eta: 0:00:08 acc1_pos: 60.197 acc5_pos: 100.000 loss: 0.7483 (0.6939) acc1: 52.3438 (60.4274) acc5: 100.0000 (100.0000) acc1_neg: 54.427 acc5_neg: 100.000 time: 0.1969 data: 0.1041 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 60.197 acc5_pos: 100.000 loss: 0.7087 (0.6935) acc1: 55.2083 (60.3526) acc5: 100.0000 (100.0000) acc1_neg: 57.031 acc5_neg: 100.000 time: 0.1578 data: 0.0662 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 60.197 acc5_pos: 100.000 loss: 0.6312 (0.6883) acc1: 62.7604 (60.7734) acc5: 100.0000 (100.0000) acc1_neg: 65.365 acc5_neg: 100.000 time: 0.1600 data: 0.0688 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 60.197 acc5_pos: 100.000 loss: 0.6006 (0.6863) acc1: 66.4062 (60.9010) acc5: 100.0000 (100.0000) acc1_neg: 59.635 acc5_neg: 100.000 time: 0.1518 data: 0.0606 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 60.197 acc5_pos: 100.000 loss: 0.6850 (0.6872) acc1: 59.6354 (60.8061) acc5: 100.0000 (100.0000) acc1_neg: 57.812 acc5_neg: 100.000 time: 0.1272 data: 0.0363 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 60.197 acc5_pos: 100.000 loss: 0.6870 (0.6871) acc1: 59.6354 (60.7996) acc5: 100.0000 (100.0000) acc1_neg: 59.718 acc5_neg: 100.000 time: 0.1225 data: 0.0287 max mem: 3532 +Test: Total time: 0:01:03 (0.1922 s / it) +* Overall Acc@1 60.800 Acc@5 100.000 + target == real Acc@1 56.722 Acc@5 100.000 + target == fake Acc@1 64.430 Acc@5 100.000 + loss 0.687 + PR-AUC 0.6287 ROC-AUC 0.6421 EER 39.71% (thr=0.5273) +Accuracy of the network on the 127843 test images: 60.8% +[rank0]:[W725 16:59:57.875333992 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 17:00:19.619000 2560588 torch/distributed/run.py:766] +W0725 17:00:19.619000 2560588 torch/distributed/run.py:766] ***************************************** +W0725 17:00:19.619000 2560588 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:00:19.619000 2560588 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:00:52.208646004 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:00:52.549412544 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:00:53.693368057 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 17:00:53.773017120 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:00:53.862311683 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank4]:[W725 17:00:53.866841079 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:00:53.006120667 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W725 17:00:53.006138841 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed1_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed1_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:22:52 acc1_pos: 79.427 acc5_pos: 100.000 loss: 0.4751 (0.4751) acc1: 79.4271 (79.4271) acc5: 100.0000 (100.0000) time: 5.8415 data: 4.1948 max mem: 3531 +Test: [ 10/235] eta: 0:02:18 acc1_pos: 76.302 acc5_pos: 100.000 loss: 0.5162 (0.5137) acc1: 76.3021 (76.5388) acc5: 100.0000 (100.0000) time: 0.6156 data: 0.3817 max mem: 3531 +Test: [ 20/235] eta: 0:01:19 acc1_pos: 80.208 acc5_pos: 100.000 loss: 0.5162 (0.5175) acc1: 76.0417 (76.1285) acc5: 100.0000 (100.0000) time: 0.0965 data: 0.0004 max mem: 3531 +Test: [ 30/235] eta: 0:00:58 acc1_pos: 77.604 acc5_pos: 100.000 loss: 0.5045 (0.5130) acc1: 76.5625 (76.3777) acc5: 100.0000 (100.0000) time: 0.1069 data: 0.0098 max mem: 3531 +Test: [ 40/235] eta: 0:00:51 acc1_pos: 76.823 acc5_pos: 100.000 loss: 0.5095 (0.5160) acc1: 76.5625 (76.0607) acc5: 100.0000 (100.0000) time: 0.1499 data: 0.0550 max mem: 3531 +Test: [ 50/235] eta: 0:00:44 acc1_pos: 70.312 acc5_pos: 100.000 loss: 0.5172 (0.5174) acc1: 75.5208 (75.8630) acc5: 100.0000 (100.0000) time: 0.1651 data: 0.0717 max mem: 3531 +Test: [ 60/235] eta: 0:00:39 acc1_pos: 73.698 acc5_pos: 100.000 loss: 0.5183 (0.5196) acc1: 75.0000 (75.5934) acc5: 100.0000 (100.0000) time: 0.1549 data: 0.0611 max mem: 3531 +Test: [ 70/235] eta: 0:00:35 acc1_pos: 78.462 acc5_pos: 100.000 loss: 0.5578 (0.5589) acc1: 70.3125 (72.3958) acc5: 100.0000 (100.0000) acc1_neg: 46.875 acc5_neg: 100.000 time: 0.1628 data: 0.0680 max mem: 3531 +Test: [ 80/235] eta: 0:00:33 acc1_pos: 78.462 acc5_pos: 100.000 loss: 0.8224 (0.5924) acc1: 49.2188 (69.5023) acc5: 100.0000 (100.0000) acc1_neg: 48.698 acc5_neg: 100.000 time: 0.1799 data: 0.0857 max mem: 3531 +Test: [ 90/235] eta: 0:00:30 acc1_pos: 78.462 acc5_pos: 100.000 loss: 0.8304 (0.6194) acc1: 48.9583 (67.2276) acc5: 100.0000 (100.0000) acc1_neg: 45.573 acc5_neg: 100.000 time: 0.1807 data: 0.0878 max mem: 3531 +Test: [100/235] eta: 0:00:27 acc1_pos: 78.462 acc5_pos: 100.000 loss: 0.8330 (0.6398) acc1: 48.9583 (65.4162) acc5: 100.0000 (100.0000) acc1_neg: 44.271 acc5_neg: 100.000 time: 0.1699 data: 0.0777 max mem: 3531 +Test: [110/235] eta: 0:00:25 acc1_pos: 78.462 acc5_pos: 100.000 loss: 0.8236 (0.6559) acc1: 49.7396 (63.9358) acc5: 100.0000 (100.0000) acc1_neg: 46.615 acc5_neg: 100.000 time: 0.1755 data: 0.0819 max mem: 3531 +Test: [120/235] eta: 0:00:23 acc1_pos: 78.462 acc5_pos: 100.000 loss: 0.8536 (0.6765) acc1: 46.6146 (62.2224) acc5: 100.0000 (100.0000) acc1_neg: 43.229 acc5_neg: 100.000 time: 0.1949 data: 0.1008 max mem: 3531 +Test: [130/235] eta: 0:00:21 acc1_pos: 78.462 acc5_pos: 100.000 loss: 0.9190 (0.6951) acc1: 43.2292 (60.7367) acc5: 100.0000 (100.0000) acc1_neg: 43.750 acc5_neg: 100.000 time: 0.2136 data: 0.1186 max mem: 3531 +Test: [140/235] eta: 0:00:19 acc1_pos: 78.462 acc5_pos: 100.000 loss: 0.9121 (0.7103) acc1: 42.4479 (59.3695) acc5: 100.0000 (100.0000) acc1_neg: 36.979 acc5_neg: 100.000 time: 0.1960 data: 0.0997 max mem: 3531 +Test: [150/235] eta: 0:00:17 acc1_pos: 78.462 acc5_pos: 100.000 loss: 0.8989 (0.7231) acc1: 42.4479 (58.2695) acc5: 100.0000 (100.0000) acc1_neg: 40.625 acc5_neg: 100.000 time: 0.1799 data: 0.0844 max mem: 3531 +Test: [160/235] eta: 0:00:15 acc1_pos: 78.462 acc5_pos: 100.000 loss: 0.8994 (0.7349) acc1: 43.2292 (57.3192) acc5: 100.0000 (100.0000) acc1_neg: 38.542 acc5_neg: 100.000 time: 0.1993 data: 0.1037 max mem: 3531 +Test: [170/235] eta: 0:00:13 acc1_pos: 78.462 acc5_pos: 100.000 loss: 0.9660 (0.7631) acc1: 38.5417 (55.8190) acc5: 100.0000 (100.0000) acc1_neg: 26.302 acc5_neg: 100.000 time: 0.2249 data: 0.1257 max mem: 3531 +Test: [180/235] eta: 0:00:11 acc1_pos: 78.462 acc5_pos: 100.000 loss: 1.2261 (0.7895) acc1: 30.7292 (54.4904) acc5: 100.0000 (100.0000) acc1_neg: 32.552 acc5_neg: 100.000 time: 0.2093 data: 0.1109 max mem: 3531 +Test: [190/235] eta: 0:00:09 acc1_pos: 78.462 acc5_pos: 100.000 loss: 1.2247 (0.8130) acc1: 31.7708 (53.3200) acc5: 100.0000 (100.0000) acc1_neg: 30.729 acc5_neg: 100.000 time: 0.1928 data: 0.0979 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 78.462 acc5_pos: 100.000 loss: 1.2601 (0.8358) acc1: 30.4688 (52.1416) acc5: 100.0000 (100.0000) acc1_neg: 28.385 acc5_neg: 100.000 time: 0.1951 data: 0.1033 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 78.462 acc5_pos: 100.000 loss: 1.2696 (0.8557) acc1: 28.9062 (51.0972) acc5: 100.0000 (100.0000) acc1_neg: 27.604 acc5_neg: 100.000 time: 0.1971 data: 0.1048 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 78.462 acc5_pos: 100.000 loss: 1.2545 (0.8738) acc1: 29.9479 (50.1756) acc5: 100.0000 (100.0000) acc1_neg: 28.125 acc5_neg: 100.000 time: 0.1975 data: 0.1034 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 78.462 acc5_pos: 100.000 loss: 1.2431 (0.8896) acc1: 30.2083 (49.3777) acc5: 100.0000 (100.0000) acc1_neg: 30.208 acc5_neg: 100.000 time: 0.1738 data: 0.0810 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 78.462 acc5_pos: 100.000 loss: 1.2431 (0.8970) acc1: 30.2083 (49.1249) acc5: 100.0000 (100.0000) acc1_neg: 34.783 acc5_neg: 100.000 time: 0.1638 data: 0.0760 max mem: 3532 +Test: Total time: 0:00:46 (0.1988 s / it) +* Overall Acc@1 49.125 Acc@5 100.000 + target == real Acc@1 75.650 Acc@5 100.000 + target == fake Acc@1 39.531 Acc@5 100.000 + loss 0.897 + PR-AUC 0.8054 ROC-AUC 0.5880 EER 43.77% (thr=0.3608) +Accuracy of the network on the 89879 test images: 49.1% +[rank0]:[W725 17:01:51.923741015 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 17:02:12.784000 2561146 torch/distributed/run.py:766] +W0725 17:02:12.784000 2561146 torch/distributed/run.py:766] ***************************************** +W0725 17:02:12.784000 2561146 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:02:12.784000 2561146 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:02:46.941114198 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:02:46.211723061 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:02:46.486937621 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:02:47.752170846 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:02:47.766176206 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:02:47.775936380 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:02:47.794284876 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 17:02:47.794453362 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed1_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n16_classseed1_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:24:24 acc1_pos: 58.854 acc5_pos: 100.000 loss: 0.6485 (0.6485) acc1: 58.8542 (58.8542) acc5: 100.0000 (100.0000) time: 5.4458 data: 3.3908 max mem: 3531 +Test: [ 10/269] eta: 0:02:29 acc1_pos: 80.208 acc5_pos: 100.000 loss: 0.6510 (0.6906) acc1: 58.8542 (58.1676) acc5: 100.0000 (100.0000) time: 0.5767 data: 0.3085 max mem: 3531 +Test: [ 20/269] eta: 0:01:26 acc1_pos: 68.229 acc5_pos: 100.000 loss: 0.6134 (0.6241) acc1: 67.1875 (65.5878) acc5: 100.0000 (100.0000) time: 0.0909 data: 0.0004 max mem: 3531 +Test: [ 30/269] eta: 0:01:14 acc1_pos: 81.250 acc5_pos: 100.000 loss: 0.5160 (0.5956) acc1: 79.1667 (69.5817) acc5: 100.0000 (100.0000) time: 0.1665 data: 0.0721 max mem: 3531 +Test: [ 40/269] eta: 0:01:04 acc1_pos: 73.276 acc5_pos: 100.000 loss: 0.5861 (0.5973) acc1: 67.9688 (68.9723) acc5: 100.0000 (100.0000) acc1_neg: 53.125 acc5_neg: 100.000 time: 0.2149 data: 0.1211 max mem: 3531 +Test: [ 50/269] eta: 0:00:58 acc1_pos: 73.276 acc5_pos: 100.000 loss: 0.6358 (0.6221) acc1: 61.9792 (65.5944) acc5: 100.0000 (100.0000) acc1_neg: 22.917 acc5_neg: 100.000 time: 0.2011 data: 0.1086 max mem: 3531 +Test: [ 60/269] eta: 0:00:53 acc1_pos: 73.276 acc5_pos: 100.000 loss: 0.7852 (0.6530) acc1: 42.4479 (61.0400) acc5: 100.0000 (100.0000) acc1_neg: 58.594 acc5_neg: 100.000 time: 0.2076 data: 0.1149 max mem: 3531 +Test: [ 70/269] eta: 0:00:50 acc1_pos: 73.276 acc5_pos: 100.000 loss: 0.7305 (0.6544) acc1: 45.3125 (60.7504) acc5: 100.0000 (100.0000) acc1_neg: 57.292 acc5_neg: 100.000 time: 0.2133 data: 0.1230 max mem: 3531 +Test: [ 80/269] eta: 0:00:46 acc1_pos: 73.276 acc5_pos: 100.000 loss: 0.6246 (0.6440) acc1: 61.9792 (61.8924) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2128 data: 0.1184 max mem: 3531 +Test: [ 90/269] eta: 0:00:43 acc1_pos: 73.276 acc5_pos: 100.000 loss: 0.5338 (0.6373) acc1: 66.4062 (62.6860) acc5: 100.0000 (100.0000) acc1_neg: 66.406 acc5_neg: 100.000 time: 0.2046 data: 0.1078 max mem: 3531 +Test: [100/269] eta: 0:00:41 acc1_pos: 73.276 acc5_pos: 100.000 loss: 0.6145 (0.6366) acc1: 66.6667 (62.9925) acc5: 100.0000 (100.0000) acc1_neg: 53.646 acc5_neg: 100.000 time: 0.2342 data: 0.1411 max mem: 3531 +Test: [110/269] eta: 0:00:38 acc1_pos: 73.276 acc5_pos: 100.000 loss: 0.6227 (0.6357) acc1: 64.8438 (63.0138) acc5: 100.0000 (100.0000) acc1_neg: 58.854 acc5_neg: 100.000 time: 0.2499 data: 0.1558 max mem: 3531 +Test: [120/269] eta: 0:00:35 acc1_pos: 73.276 acc5_pos: 100.000 loss: 0.6674 (0.6510) acc1: 55.2083 (61.3314) acc5: 100.0000 (100.0000) acc1_neg: 29.688 acc5_neg: 100.000 time: 0.2003 data: 0.1042 max mem: 3531 +Test: [130/269] eta: 0:00:32 acc1_pos: 73.276 acc5_pos: 100.000 loss: 0.8696 (0.6691) acc1: 34.3750 (59.1702) acc5: 100.0000 (100.0000) acc1_neg: 29.167 acc5_neg: 100.000 time: 0.1811 data: 0.0843 max mem: 3531 +Test: [140/269] eta: 0:00:30 acc1_pos: 73.276 acc5_pos: 100.000 loss: 0.8696 (0.6782) acc1: 34.3750 (58.0637) acc5: 100.0000 (100.0000) acc1_neg: 51.302 acc5_neg: 100.000 time: 0.2096 data: 0.1152 max mem: 3531 +Test: [150/269] eta: 0:00:27 acc1_pos: 73.276 acc5_pos: 100.000 loss: 0.7309 (0.6843) acc1: 44.2708 (57.1968) acc5: 100.0000 (100.0000) acc1_neg: 33.073 acc5_neg: 100.000 time: 0.1991 data: 0.1078 max mem: 3531 +Test: [160/269] eta: 0:00:24 acc1_pos: 73.276 acc5_pos: 100.000 loss: 0.8195 (0.6969) acc1: 35.6771 (55.7696) acc5: 100.0000 (100.0000) acc1_neg: 39.583 acc5_neg: 100.000 time: 0.1958 data: 0.1041 max mem: 3531 +Test: [170/269] eta: 0:00:22 acc1_pos: 73.276 acc5_pos: 100.000 loss: 0.8908 (0.7070) acc1: 29.4271 (54.6677) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.2159 data: 0.1207 max mem: 3531 +Test: [180/269] eta: 0:00:20 acc1_pos: 73.276 acc5_pos: 100.000 loss: 0.6258 (0.6963) acc1: 66.9271 (55.9766) acc5: 100.0000 (100.0000) acc1_neg: 70.052 acc5_neg: 100.000 time: 0.2093 data: 0.1146 max mem: 3531 +Test: [190/269] eta: 0:00:17 acc1_pos: 73.276 acc5_pos: 100.000 loss: 0.5990 (0.6950) acc1: 68.7500 (56.0796) acc5: 100.0000 (100.0000) acc1_neg: 55.208 acc5_neg: 100.000 time: 0.2059 data: 0.1152 max mem: 3531 +Test: [200/269] eta: 0:00:15 acc1_pos: 73.276 acc5_pos: 100.000 loss: 0.7336 (0.6994) acc1: 48.1771 (55.5620) acc5: 100.0000 (100.0000) acc1_neg: 58.594 acc5_neg: 100.000 time: 0.2111 data: 0.1200 max mem: 3531 +Test: [210/269] eta: 0:00:13 acc1_pos: 73.276 acc5_pos: 100.000 loss: 0.7670 (0.7045) acc1: 42.7083 (54.9035) acc5: 100.0000 (100.0000) acc1_neg: 38.281 acc5_neg: 100.000 time: 0.2005 data: 0.1086 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 73.276 acc5_pos: 100.000 loss: 0.7569 (0.7084) acc1: 41.1458 (54.4365) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.2114 data: 0.1193 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 73.276 acc5_pos: 100.000 loss: 0.7323 (0.7102) acc1: 41.1458 (54.1216) acc5: 100.0000 (100.0000) acc1_neg: 33.594 acc5_neg: 100.000 time: 0.1916 data: 0.1008 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 73.276 acc5_pos: 100.000 loss: 0.8074 (0.7142) acc1: 41.1458 (53.5961) acc5: 100.0000 (100.0000) acc1_neg: 42.188 acc5_neg: 100.000 time: 0.1301 data: 0.0398 max mem: 3532 +Test: [250/269] eta: 0:00:04 acc1_pos: 73.276 acc5_pos: 100.000 loss: 0.7698 (0.7156) acc1: 43.7500 (53.2931) acc5: 100.0000 (100.0000) acc1_neg: 33.073 acc5_neg: 100.000 time: 0.1079 data: 0.0176 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 73.276 acc5_pos: 100.000 loss: 0.7800 (0.7203) acc1: 41.9271 (52.6052) acc5: 100.0000 (100.0000) acc1_neg: 41.927 acc5_neg: 100.000 time: 0.1154 data: 0.0257 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 73.276 acc5_pos: 100.000 loss: 0.8040 (0.7221) acc1: 36.4583 (52.3424) acc5: 100.0000 (100.0000) acc1_neg: 56.168 acc5_neg: 100.000 time: 0.1248 data: 0.0319 max mem: 3532 +Test: Total time: 0:00:55 (0.2057 s / it) +* Overall Acc@1 52.342 Acc@5 100.000 + target == real Acc@1 69.558 Acc@5 100.000 + target == fake Acc@1 49.993 Acc@5 100.000 + loss 0.722 + PR-AUC 0.9220 ROC-AUC 0.6463 EER 39.79% (thr=0.4529) +Accuracy of the network on the 103293 test images: 52.3% +[rank0]:[W725 17:03:49.332222971 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-25T17:03:56Z] eval_done tag=fixed500k_real_n16_seed1_resampled +[2026-07-25T17:03:56Z] eval_wait tag=fixed500k_fake_n32_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed1_resampled +[2026-07-25T17:04:56Z] eval_wait tag=fixed500k_fake_n32_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed1_resampled +[2026-07-25T17:05:56Z] eval_wait tag=fixed500k_fake_n32_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed1_resampled +[2026-07-25T17:06:56Z] eval_wait tag=fixed500k_fake_n32_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed1_resampled +[2026-07-25T17:07:56Z] eval_wait tag=fixed500k_fake_n32_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed1_resampled +[2026-07-25T17:08:57Z] eval_wait tag=fixed500k_fake_n32_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed1_resampled +[2026-07-25T17:09:57Z] eval_wait tag=fixed500k_fake_n32_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed1_resampled +[2026-07-25T17:10:57Z] eval_wait tag=fixed500k_fake_n32_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed1_resampled +[2026-07-25T17:11:57Z] eval_wait tag=fixed500k_fake_n32_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed1_resampled +[2026-07-25T17:12:57Z] eval_wait tag=fixed500k_fake_n32_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed1_resampled +[2026-07-25T17:13:57Z] eval_wait tag=fixed500k_fake_n32_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed1_resampled +[2026-07-25T17:14:57Z] eval_wait tag=fixed500k_fake_n32_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed1_resampled +[2026-07-25T17:15:58Z] eval_wait tag=fixed500k_fake_n32_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed1_resampled +[2026-07-25T17:16:58Z] eval_wait tag=fixed500k_fake_n32_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed1_resampled +[2026-07-25T17:17:58Z] eval_wait tag=fixed500k_fake_n32_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed1_resampled +[2026-07-25T17:18:58Z] eval_wait tag=fixed500k_fake_n32_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed1_resampled +[2026-07-25T17:19:58Z] eval_wait tag=fixed500k_fake_n32_seed1_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed1_resampled +[2026-07-25T17:20:58Z] eval_start tag=fixed500k_fake_n32_seed1_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed1_resampled_n111-046-081_20260725_112559/best_checkpoint.pth +started_utc=2026-07-25T17:20:58Z +hostname=n111-046-081 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed1_resampled_n111-046-081_20260725_112559/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed1_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0725 17:21:14.950000 2563304 torch/distributed/run.py:766] +W0725 17:21:14.950000 2563304 torch/distributed/run.py:766] ***************************************** +W0725 17:21:14.950000 2563304 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:21:14.950000 2563304 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:21:55.051946114 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:21:55.216074943 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +[rank1]:[W725 17:21:56.660835804 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:21:56.668075759 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:21:57.123481259 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:21:57.238226981 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 17:21:57.360548928 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:21:57.424049735 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed1_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed1_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 1:08:14 acc1_pos: 59.115 acc5_pos: 100.000 loss: 0.6906 (0.6906) acc1: 59.1146 (59.1146) acc5: 100.0000 (100.0000) time: 17.4979 data: 12.6980 max mem: 3531 +Test: [ 10/234] eta: 0:06:18 acc1_pos: 26.562 acc5_pos: 100.000 loss: 0.7892 (0.8256) acc1: 47.3958 (45.6676) acc5: 100.0000 (100.0000) time: 1.6902 data: 1.1548 max mem: 3531 +Test: [ 20/234] eta: 0:03:31 acc1_pos: 92.969 acc5_pos: 100.000 loss: 0.7892 (0.7877) acc1: 45.3125 (49.4668) acc5: 100.0000 (100.0000) time: 0.1647 data: 0.0505 max mem: 3531 +Test: [ 30/234] eta: 0:03:08 acc1_pos: 76.562 acc5_pos: 100.000 loss: 0.3036 (0.6326) acc1: 91.9271 (62.6932) acc5: 100.0000 (100.0000) time: 0.5049 data: 0.3805 max mem: 3531 +Test: [ 40/234] eta: 0:02:38 acc1_pos: 98.958 acc5_pos: 100.000 loss: 0.3012 (0.5868) acc1: 92.1875 (66.0569) acc5: 100.0000 (100.0000) time: 0.6399 data: 0.5037 max mem: 3531 +Test: [ 50/234] eta: 0:02:13 acc1_pos: 99.479 acc5_pos: 100.000 loss: 0.1703 (0.5039) acc1: 97.6562 (72.3601) acc5: 100.0000 (100.0000) time: 0.4209 data: 0.2903 max mem: 3531 +Test: [ 60/234] eta: 0:02:03 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.1703 (0.4501) acc1: 99.2188 (76.8272) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.4818 data: 0.3576 max mem: 3531 +Test: [ 70/234] eta: 0:01:52 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.1759 (0.4664) acc1: 99.2188 (76.3021) acc5: 100.0000 (100.0000) acc1_neg: 58.854 acc5_neg: 100.000 time: 0.5890 data: 0.4712 max mem: 3531 +Test: [ 80/234] eta: 0:01:43 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.6863 (0.4770) acc1: 60.9375 (75.9549) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.5639 data: 0.4482 max mem: 3531 +Test: [ 90/234] eta: 0:01:36 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.3343 (0.4643) acc1: 86.1979 (77.2350) acc5: 100.0000 (100.0000) acc1_neg: 65.885 acc5_neg: 100.000 time: 0.5897 data: 0.4619 max mem: 3531 +Test: [100/234] eta: 0:01:25 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.3343 (0.4725) acc1: 86.1979 (76.7688) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.4989 data: 0.3799 max mem: 3531 +Test: [110/234] eta: 0:01:18 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.5463 (0.4849) acc1: 72.1354 (76.0417) acc5: 100.0000 (100.0000) acc1_neg: 54.688 acc5_neg: 100.000 time: 0.4698 data: 0.3545 max mem: 3531 +Test: [120/234] eta: 0:01:11 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.5123 (0.4925) acc1: 72.1354 (75.6909) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.5910 data: 0.4658 max mem: 3531 +Test: [130/234] eta: 0:01:04 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.3749 (0.4868) acc1: 84.3750 (76.3995) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.5555 data: 0.4284 max mem: 3531 +Test: [140/234] eta: 0:00:58 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.3749 (0.5002) acc1: 83.5938 (75.9752) acc5: 100.0000 (100.0000) acc1_neg: 37.500 acc5_neg: 100.000 time: 0.5738 data: 0.4399 max mem: 3531 +Test: [150/234] eta: 0:00:51 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.5528 (0.5112) acc1: 65.8854 (75.4674) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.5545 data: 0.4162 max mem: 3531 +Test: [160/234] eta: 0:00:45 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.4854 (0.5035) acc1: 77.6042 (76.1177) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.5348 data: 0.4063 max mem: 3531 +Test: [170/234] eta: 0:00:38 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.4800 (0.5020) acc1: 78.9062 (76.3158) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.5680 data: 0.4479 max mem: 3531 +Test: [180/234] eta: 0:00:32 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.5039 (0.5174) acc1: 75.7812 (75.2949) acc5: 100.0000 (100.0000) acc1_neg: 22.396 acc5_neg: 100.000 time: 0.5833 data: 0.4628 max mem: 3531 +Test: [190/234] eta: 0:00:26 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.5751 (0.5219) acc1: 69.2708 (74.9386) acc5: 100.0000 (100.0000) acc1_neg: 72.396 acc5_neg: 100.000 time: 0.5950 data: 0.4604 max mem: 3531 +Test: [200/234] eta: 0:00:20 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.5405 (0.5088) acc1: 72.3958 (75.6646) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.4808 data: 0.3356 max mem: 3531 +Test: [210/234] eta: 0:00:14 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.0864 (0.4910) acc1: 99.7396 (76.7810) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.4147 data: 0.2965 max mem: 3532 +Test: [220/234] eta: 0:00:08 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.2227 (0.4796) acc1: 97.9167 (77.6749) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.3412 data: 0.2256 max mem: 3532 +Test: [230/234] eta: 0:00:02 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.3702 (0.4874) acc1: 64.5833 (76.9920) acc5: 100.0000 (100.0000) acc1_neg: 61.198 acc5_neg: 100.000 time: 0.2652 data: 0.1279 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 98.621 acc5_pos: 100.000 loss: 0.6393 (0.4896) acc1: 63.0208 (76.8291) acc5: 100.0000 (100.0000) acc1_neg: 61.868 acc5_neg: 100.000 time: 0.2198 data: 0.0781 max mem: 3532 +Test: Total time: 0:02:10 (0.5560 s / it) +* Overall Acc@1 76.829 Acc@5 100.000 + target == real Acc@1 72.553 Acc@5 100.000 + target == fake Acc@1 78.034 Acc@5 100.000 + loss 0.490 + PR-AUC 0.9493 ROC-AUC 0.8420 EER 24.35% (thr=0.5303) +Accuracy of the network on the 89729 test images: 76.8% +[rank0]:[W725 17:24:26.306141775 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 17:24:57.255000 2563894 torch/distributed/run.py:766] +W0725 17:24:57.255000 2563894 torch/distributed/run.py:766] ***************************************** +W0725 17:24:57.255000 2563894 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:24:57.255000 2563894 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:25:35.329101848 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:25:36.058278429 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +[rank0]:[W725 17:25:37.845504599 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:25:37.856955559 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:25:37.130860295 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:25:37.442703766 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:25:37.521126037 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:25:37.564309772 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed1_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed1_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 11:48:13 acc1_pos: 26.302 acc5_pos: 100.000 loss: 1.1336 (1.1336) acc1: 26.3021 (26.3021) acc5: 100.0000 (100.0000) time: 17.8020 data: 13.8019 max mem: 3531 +Test: [ 10/2387] eta: 1:10:32 acc1_pos: 12.240 acc5_pos: 100.000 loss: 1.1401 (1.1180) acc1: 21.0938 (22.8693) acc5: 100.0000 (100.0000) time: 1.7804 data: 1.3011 max mem: 3531 +Test: [ 20/2387] eta: 0:39:02 acc1_pos: 24.219 acc5_pos: 100.000 loss: 1.1401 (1.0901) acc1: 19.7917 (24.6900) acc5: 100.0000 (100.0000) time: 0.1491 data: 0.0259 max mem: 3531 +Test: [ 30/2387] eta: 0:36:59 acc1_pos: 15.104 acc5_pos: 100.000 loss: 1.1086 (1.0862) acc1: 19.0104 (24.6724) acc5: 100.0000 (100.0000) time: 0.4804 data: 0.3646 max mem: 3531 +Test: [ 40/2387] eta: 0:34:19 acc1_pos: 17.969 acc5_pos: 100.000 loss: 1.1086 (1.0875) acc1: 18.7500 (23.0818) acc5: 100.0000 (100.0000) time: 0.7599 data: 0.6410 max mem: 3531 +Test: [ 50/2387] eta: 0:31:45 acc1_pos: 41.927 acc5_pos: 100.000 loss: 1.0200 (1.0765) acc1: 20.3125 (24.4690) acc5: 100.0000 (100.0000) time: 0.6193 data: 0.4943 max mem: 3531 +Test: [ 60/2387] eta: 0:30:22 acc1_pos: 39.844 acc5_pos: 100.000 loss: 1.0709 (1.0815) acc1: 22.6562 (24.1078) acc5: 100.0000 (100.0000) time: 0.5899 data: 0.4662 max mem: 3531 +Test: [ 70/2387] eta: 0:29:11 acc1_pos: 33.594 acc5_pos: 100.000 loss: 1.1287 (1.0915) acc1: 19.5312 (23.5329) acc5: 100.0000 (100.0000) time: 0.6049 data: 0.4811 max mem: 3531 +Test: [ 80/2387] eta: 0:28:33 acc1_pos: 36.198 acc5_pos: 100.000 loss: 1.1175 (1.0886) acc1: 16.1458 (23.2382) acc5: 100.0000 (100.0000) time: 0.6197 data: 0.5026 max mem: 3531 +Test: [ 90/2387] eta: 0:28:15 acc1_pos: 36.979 acc5_pos: 100.000 loss: 1.0483 (1.0783) acc1: 25.0000 (24.0642) acc5: 100.0000 (100.0000) time: 0.6749 data: 0.5476 max mem: 3531 +Test: [ 100/2387] eta: 0:27:39 acc1_pos: 22.135 acc5_pos: 100.000 loss: 1.0451 (1.0815) acc1: 25.0000 (23.8114) acc5: 100.0000 (100.0000) time: 0.6548 data: 0.5235 max mem: 3531 +Test: [ 110/2387] eta: 0:27:02 acc1_pos: 38.281 acc5_pos: 100.000 loss: 1.0238 (1.0707) acc1: 24.7396 (24.4369) acc5: 100.0000 (100.0000) time: 0.5949 data: 0.4716 max mem: 3531 +Test: [ 120/2387] eta: 0:26:33 acc1_pos: 21.875 acc5_pos: 100.000 loss: 1.0082 (1.0709) acc1: 25.7812 (24.6018) acc5: 100.0000 (100.0000) time: 0.5899 data: 0.4628 max mem: 3531 +Test: [ 130/2387] eta: 0:26:22 acc1_pos: 15.885 acc5_pos: 100.000 loss: 1.0660 (1.0703) acc1: 20.3125 (24.2684) acc5: 100.0000 (100.0000) time: 0.6399 data: 0.5285 max mem: 3531 +Test: [ 140/2387] eta: 0:25:43 acc1_pos: 26.302 acc5_pos: 100.000 loss: 1.0432 (1.0666) acc1: 20.5729 (24.4219) acc5: 100.0000 (100.0000) time: 0.5899 data: 0.4817 max mem: 3531 +Test: [ 150/2387] eta: 0:25:20 acc1_pos: 8.854 acc5_pos: 100.000 loss: 0.9972 (1.0626) acc1: 26.3021 (24.6896) acc5: 100.0000 (100.0000) time: 0.5398 data: 0.4128 max mem: 3531 +Test: [ 160/2387] eta: 0:24:45 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.9487 (1.0433) acc1: 29.4271 (26.3263) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.5249 data: 0.3986 max mem: 3531 +Test: [ 170/2387] eta: 0:24:28 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2050 (0.9937) acc1: 97.1354 (30.5266) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.5299 data: 0.3925 max mem: 3531 +Test: [ 180/2387] eta: 0:24:20 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1933 (0.9496) acc1: 98.1771 (34.2728) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.6198 data: 0.4776 max mem: 3531 +Test: [ 190/2387] eta: 0:23:49 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1911 (0.9100) acc1: 98.4375 (37.6323) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.5505 data: 0.4372 max mem: 3531 +Test: [ 200/2387] eta: 0:23:28 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1909 (0.8742) acc1: 98.4375 (40.6639) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.4849 data: 0.3736 max mem: 3531 +Test: [ 210/2387] eta: 0:23:00 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1935 (0.8422) acc1: 98.4375 (43.4020) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.4743 data: 0.3546 max mem: 3532 +Test: [ 220/2387] eta: 0:22:49 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1953 (0.8129) acc1: 98.4375 (45.8852) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.5099 data: 0.3945 max mem: 3532 +Test: [ 230/2387] eta: 0:22:35 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1953 (0.7863) acc1: 98.1771 (48.1444) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.5698 data: 0.4564 max mem: 3532 +Test: [ 240/2387] eta: 0:22:16 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2042 (0.7624) acc1: 98.1771 (50.2140) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.5203 data: 0.4043 max mem: 3532 +Test: [ 250/2387] eta: 0:21:57 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1916 (0.7355) acc1: 98.6979 (52.1871) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.4819 data: 0.3682 max mem: 3532 +Test: [ 260/2387] eta: 0:21:34 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0637 (0.7096) acc1: 100.0000 (54.0190) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.4413 data: 0.3204 max mem: 3532 +Test: [ 270/2387] eta: 0:21:20 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0611 (0.6857) acc1: 100.0000 (55.7157) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.4540 data: 0.3277 max mem: 3532 +Test: [ 280/2387] eta: 0:20:59 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0611 (0.6636) acc1: 100.0000 (57.2917) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.4529 data: 0.3377 max mem: 3532 +Test: [ 290/2387] eta: 0:20:56 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0649 (0.6430) acc1: 100.0000 (58.7593) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5236 data: 0.3983 max mem: 3532 +Test: [ 300/2387] eta: 0:20:42 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0649 (0.6238) acc1: 100.0000 (60.1294) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5648 data: 0.4394 max mem: 3532 +Test: [ 310/2387] eta: 0:20:26 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0692 (0.6102) acc1: 100.0000 (61.3779) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.4656 data: 0.3546 max mem: 3532 +Test: [ 320/2387] eta: 0:20:19 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.3646 (0.6026) acc1: 98.1771 (62.5073) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.5099 data: 0.3962 max mem: 3532 +Test: [ 330/2387] eta: 0:20:17 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.3652 (0.5954) acc1: 97.6562 (63.5794) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.6091 data: 0.4855 max mem: 3532 +Test: [ 340/2387] eta: 0:20:06 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.3621 (0.5885) acc1: 97.6562 (64.5826) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.5750 data: 0.4459 max mem: 3532 +Test: [ 350/2387] eta: 0:19:59 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.3555 (0.5770) acc1: 98.6979 (65.5835) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5458 data: 0.4210 max mem: 3532 +Test: [ 360/2387] eta: 0:19:43 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1607 (0.5656) acc1: 99.7396 (66.5166) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.4947 data: 0.3648 max mem: 3532 +Test: [ 370/2387] eta: 0:19:36 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1711 (0.5550) acc1: 98.9583 (67.3883) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.4839 data: 0.3531 max mem: 3532 +Test: [ 380/2387] eta: 0:19:19 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1718 (0.5449) acc1: 98.6979 (68.2107) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.4653 data: 0.3365 max mem: 3532 +Test: [ 390/2387] eta: 0:19:11 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1696 (0.5354) acc1: 98.9583 (68.9958) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.4549 data: 0.3363 max mem: 3532 +Test: [ 400/2387] eta: 0:19:03 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1699 (0.5263) acc1: 98.9583 (69.7410) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.5345 data: 0.4052 max mem: 3532 +Test: [ 410/2387] eta: 0:18:52 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1694 (0.5177) acc1: 98.9583 (70.4500) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.4998 data: 0.3597 max mem: 3532 +Test: [ 420/2387] eta: 0:18:42 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1695 (0.5094) acc1: 98.9583 (71.1253) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.4666 data: 0.3386 max mem: 3532 +Test: [ 430/2387] eta: 0:18:26 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1737 (0.5019) acc1: 98.6979 (71.7656) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.4105 data: 0.2436 max mem: 3532 +Test: [ 440/2387] eta: 0:18:08 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1743 (0.4932) acc1: 98.9583 (72.3905) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.3246 data: 0.1210 max mem: 3533 +Test: [ 450/2387] eta: 0:17:50 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0701 (0.4838) acc1: 100.0000 (73.0027) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2712 data: 0.0660 max mem: 3533 +Test: [ 460/2387] eta: 0:17:31 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0695 (0.4748) acc1: 100.0000 (73.5883) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2447 data: 0.0419 max mem: 3533 +Test: [ 470/2387] eta: 0:17:14 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0695 (0.4662) acc1: 100.0000 (74.1480) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2489 data: 0.0441 max mem: 3533 +Test: [ 480/2387] eta: 0:16:55 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0699 (0.4580) acc1: 100.0000 (74.6854) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2286 data: 0.0379 max mem: 3533 +Test: [ 490/2387] eta: 0:16:38 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0706 (0.4501) acc1: 100.0000 (75.2010) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2184 data: 0.0863 max mem: 3533 +Test: [ 500/2387] eta: 0:16:26 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0704 (0.4425) acc1: 100.0000 (75.6955) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2843 data: 0.1934 max mem: 3533 +Test: [ 510/2387] eta: 0:16:10 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0704 (0.4354) acc1: 100.0000 (76.1711) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2849 data: 0.1907 max mem: 3533 +Test: [ 520/2387] eta: 0:15:52 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0990 (0.4291) acc1: 100.0000 (76.6285) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1965 data: 0.1012 max mem: 3533 +Test: [ 530/2387] eta: 0:15:35 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1059 (0.4230) acc1: 100.0000 (77.0686) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0763 max mem: 3533 +Test: [ 540/2387] eta: 0:15:21 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1078 (0.4172) acc1: 100.0000 (77.4920) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2044 data: 0.1108 max mem: 3533 +Test: [ 550/2387] eta: 0:15:05 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1074 (0.4116) acc1: 100.0000 (77.9005) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2021 data: 0.1099 max mem: 3533 +Test: [ 560/2387] eta: 0:14:50 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1078 (0.4062) acc1: 100.0000 (78.2944) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1738 data: 0.0820 max mem: 3533 +Test: [ 570/2387] eta: 0:14:34 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1080 (0.4009) acc1: 100.0000 (78.6746) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1627 data: 0.0713 max mem: 3533 +Test: [ 580/2387] eta: 0:14:20 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1080 (0.3959) acc1: 100.0000 (79.0416) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1713 data: 0.0779 max mem: 3533 +Test: [ 590/2387] eta: 0:14:06 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1087 (0.3910) acc1: 100.0000 (79.3962) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1832 data: 0.0896 max mem: 3533 +Test: [ 600/2387] eta: 0:13:54 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1052 (0.3862) acc1: 100.0000 (79.7391) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1905 data: 0.0991 max mem: 3533 +Test: [ 610/2387] eta: 0:13:41 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1190 (0.3824) acc1: 100.0000 (80.0634) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1866 data: 0.0926 max mem: 3533 +Test: [ 620/2387] eta: 0:13:27 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1494 (0.3786) acc1: 99.7396 (80.3786) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1687 data: 0.0745 max mem: 3533 +Test: [ 630/2387] eta: 0:13:15 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1486 (0.3750) acc1: 99.7396 (80.6825) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1723 data: 0.0800 max mem: 3533 +Test: [ 640/2387] eta: 0:13:03 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1527 (0.3715) acc1: 99.4792 (80.9750) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1801 data: 0.0871 max mem: 3533 +Test: [ 650/2387] eta: 0:12:51 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1531 (0.3681) acc1: 99.4792 (81.2584) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1735 data: 0.0817 max mem: 3533 +Test: [ 660/2387] eta: 0:12:40 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1503 (0.3648) acc1: 99.4792 (81.5341) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1712 data: 0.0814 max mem: 3533 +Test: [ 670/2387] eta: 0:12:28 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1509 (0.3617) acc1: 99.4792 (81.8050) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1739 data: 0.0820 max mem: 3534 +Test: [ 680/2387] eta: 0:12:17 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1537 (0.3588) acc1: 100.0000 (82.0695) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1725 data: 0.0776 max mem: 3534 +Test: [ 690/2387] eta: 0:12:07 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1711 (0.3588) acc1: 99.7396 (82.1183) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.1799 data: 0.0853 max mem: 3534 +Test: [ 700/2387] eta: 0:11:57 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.4276 (0.3597) acc1: 80.9896 (82.1096) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.1916 data: 0.0978 max mem: 3534 +Test: [ 710/2387] eta: 0:11:47 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.4264 (0.3607) acc1: 81.2500 (82.0917) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1907 data: 0.0966 max mem: 3534 +Test: [ 720/2387] eta: 0:11:37 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.4258 (0.3616) acc1: 80.9896 (82.0822) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.1792 data: 0.0845 max mem: 3534 +Test: [ 730/2387] eta: 0:11:28 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.4327 (0.3627) acc1: 80.7292 (82.0462) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.1753 data: 0.0804 max mem: 3534 +Test: [ 740/2387] eta: 0:11:18 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.4341 (0.3636) acc1: 79.6875 (82.0225) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.1789 data: 0.0860 max mem: 3534 +Test: [ 750/2387] eta: 0:11:09 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.4253 (0.3644) acc1: 80.7292 (82.0132) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.1744 data: 0.0828 max mem: 3534 +Test: [ 760/2387] eta: 0:11:00 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.4140 (0.3648) acc1: 81.2500 (82.0141) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1806 data: 0.0875 max mem: 3534 +Test: [ 770/2387] eta: 0:10:52 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.3595 (0.3638) acc1: 85.4167 (82.1272) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1941 data: 0.0988 max mem: 3534 +Test: [ 780/2387] eta: 0:10:43 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2252 (0.3619) acc1: 95.5729 (82.3093) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1942 data: 0.1001 max mem: 3534 +Test: [ 790/2387] eta: 0:10:35 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1965 (0.3596) acc1: 96.8750 (82.5053) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1899 data: 0.0982 max mem: 3534 +Test: [ 800/2387] eta: 0:10:27 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1712 (0.3576) acc1: 97.9167 (82.6831) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1871 data: 0.0954 max mem: 3534 +Test: [ 810/2387] eta: 0:10:19 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1881 (0.3558) acc1: 97.9167 (82.8549) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1875 data: 0.0968 max mem: 3534 +Test: [ 820/2387] eta: 0:10:12 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2083 (0.3545) acc1: 95.0521 (82.9889) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.2027 data: 0.1115 max mem: 3534 +Test: [ 830/2387] eta: 0:10:04 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2704 (0.3538) acc1: 93.2292 (83.0870) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2173 data: 0.1251 max mem: 3534 +Test: [ 840/2387] eta: 0:09:57 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2330 (0.3521) acc1: 96.3542 (83.2692) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2063 data: 0.1120 max mem: 3534 +Test: [ 850/2387] eta: 0:09:49 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1959 (0.3502) acc1: 99.2188 (83.4545) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1803 data: 0.0847 max mem: 3534 +Test: [ 860/2387] eta: 0:09:43 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1782 (0.3481) acc1: 99.7396 (83.6418) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2044 data: 0.1110 max mem: 3534 +Test: [ 870/2387] eta: 0:09:36 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1788 (0.3464) acc1: 99.2188 (83.8168) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2161 data: 0.1227 max mem: 3534 +Test: [ 880/2387] eta: 0:09:28 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1946 (0.3447) acc1: 98.9583 (83.9893) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1846 data: 0.0906 max mem: 3534 +Test: [ 890/2387] eta: 0:09:22 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1881 (0.3428) acc1: 99.2188 (84.1581) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2006 data: 0.1088 max mem: 3534 +Test: [ 900/2387] eta: 0:09:16 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2116 (0.3453) acc1: 95.5729 (84.0076) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.2192 data: 0.1283 max mem: 3535 +Test: [ 910/2387] eta: 0:09:10 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.6526 (0.3487) acc1: 64.3229 (83.7930) acc5: 100.0000 (100.0000) acc1_neg: 63.802 acc5_neg: 100.000 time: 0.2501 data: 0.1578 max mem: 3535 +Test: [ 920/2387] eta: 0:09:04 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.6647 (0.3521) acc1: 63.8021 (83.5697) acc5: 100.0000 (100.0000) acc1_neg: 59.896 acc5_neg: 100.000 time: 0.2399 data: 0.1472 max mem: 3535 +Test: [ 930/2387] eta: 0:08:57 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.6542 (0.3553) acc1: 64.0625 (83.3613) acc5: 100.0000 (100.0000) acc1_neg: 58.594 acc5_neg: 100.000 time: 0.1979 data: 0.0988 max mem: 3535 +Test: [ 940/2387] eta: 0:08:52 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.6573 (0.3586) acc1: 64.0625 (83.1457) acc5: 100.0000 (100.0000) acc1_neg: 59.896 acc5_neg: 100.000 time: 0.2199 data: 0.1217 max mem: 3535 +Test: [ 950/2387] eta: 0:08:50 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.6573 (0.3617) acc1: 63.8021 (82.9390) acc5: 100.0000 (100.0000) acc1_neg: 64.844 acc5_neg: 100.000 time: 0.3699 data: 0.2744 max mem: 3535 +Test: [ 960/2387] eta: 0:08:51 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.6429 (0.3646) acc1: 64.5833 (82.7499) acc5: 100.0000 (100.0000) acc1_neg: 65.625 acc5_neg: 100.000 time: 0.5949 data: 0.4943 max mem: 3535 +Test: [ 970/2387] eta: 0:08:49 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.6516 (0.3676) acc1: 64.5833 (82.5507) acc5: 100.0000 (100.0000) acc1_neg: 65.885 acc5_neg: 100.000 time: 0.5858 data: 0.4872 max mem: 3535 +Test: [ 980/2387] eta: 0:08:44 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.6199 (0.3679) acc1: 66.9271 (82.5667) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.3951 data: 0.3009 max mem: 3535 +Test: [ 990/2387] eta: 0:08:46 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2712 (0.3667) acc1: 92.9688 (82.6801) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.5184 data: 0.4073 max mem: 3535 +Test: [1000/2387] eta: 0:08:44 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2212 (0.3653) acc1: 97.3958 (82.8159) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.6446 data: 0.5257 max mem: 3535 +Test: [1010/2387] eta: 0:08:43 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2294 (0.3644) acc1: 94.7917 (82.9209) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.5600 data: 0.4555 max mem: 3535 +Test: [1020/2387] eta: 0:08:42 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2433 (0.3632) acc1: 95.8333 (83.0303) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.5699 data: 0.4607 max mem: 3535 +Test: [1030/2387] eta: 0:08:42 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1751 (0.3612) acc1: 97.6562 (83.1798) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.6100 data: 0.4854 max mem: 3535 +Test: [1040/2387] eta: 0:08:40 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1440 (0.3590) acc1: 99.4792 (83.3376) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.5949 data: 0.4649 max mem: 3535 +Test: [1050/2387] eta: 0:08:39 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1309 (0.3569) acc1: 100.0000 (83.4897) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5996 data: 0.4691 max mem: 3535 +Test: [1060/2387] eta: 0:08:37 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1357 (0.3550) acc1: 99.4792 (83.6367) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.5949 data: 0.4694 max mem: 3535 +Test: [1070/2387] eta: 0:08:37 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1620 (0.3534) acc1: 98.9583 (83.7715) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.6050 data: 0.4877 max mem: 3535 +Test: [1080/2387] eta: 0:08:36 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1958 (0.3521) acc1: 97.6562 (83.8838) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.6699 data: 0.5495 max mem: 3535 +Test: [1090/2387] eta: 0:08:34 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1775 (0.3504) acc1: 97.6562 (84.0143) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.5947 data: 0.4851 max mem: 3535 +Test: [1100/2387] eta: 0:08:32 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1521 (0.3486) acc1: 98.6979 (84.1508) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5549 data: 0.4477 max mem: 3535 +Test: [1110/2387] eta: 0:08:30 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1530 (0.3471) acc1: 98.6979 (84.2756) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.5752 data: 0.4396 max mem: 3535 +Test: [1120/2387] eta: 0:08:30 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1853 (0.3457) acc1: 97.6562 (84.3943) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.6498 data: 0.5117 max mem: 3536 +Test: [1130/2387] eta: 0:08:29 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1813 (0.3442) acc1: 98.1771 (84.5180) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.7246 data: 0.5907 max mem: 3536 +Test: [1140/2387] eta: 0:08:26 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1990 (0.3431) acc1: 97.6562 (84.6249) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.5848 data: 0.4553 max mem: 3536 +Test: [1150/2387] eta: 0:08:26 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1876 (0.3417) acc1: 98.6979 (84.7418) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.6200 data: 0.5053 max mem: 3536 +Test: [1160/2387] eta: 0:08:24 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1244 (0.3396) acc1: 100.0000 (84.8725) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.6961 data: 0.5776 max mem: 3536 +Test: [1170/2387] eta: 0:08:21 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1069 (0.3377) acc1: 100.0000 (85.0012) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.5752 data: 0.4512 max mem: 3536 +Test: [1180/2387] eta: 0:08:21 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1085 (0.3358) acc1: 100.0000 (85.1263) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.6844 data: 0.5719 max mem: 3536 +Test: [1190/2387] eta: 0:08:18 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1118 (0.3340) acc1: 100.0000 (85.2501) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.6494 data: 0.5447 max mem: 3536 +Test: [1200/2387] eta: 0:08:16 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1122 (0.3321) acc1: 100.0000 (85.3724) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5549 data: 0.4443 max mem: 3536 +Test: [1210/2387] eta: 0:08:13 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1152 (0.3304) acc1: 100.0000 (85.4926) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.6154 data: 0.4965 max mem: 3536 +Test: [1220/2387] eta: 0:08:09 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0846 (0.3284) acc1: 100.0000 (85.6114) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5141 data: 0.3913 max mem: 3536 +Test: [1230/2387] eta: 0:08:07 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0722 (0.3263) acc1: 100.0000 (85.7283) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5660 data: 0.4501 max mem: 3536 +Test: [1240/2387] eta: 0:08:05 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0694 (0.3242) acc1: 100.0000 (85.8429) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.6448 data: 0.5225 max mem: 3536 +Test: [1250/2387] eta: 0:08:03 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0721 (0.3223) acc1: 100.0000 (85.9560) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.6482 data: 0.5121 max mem: 3536 +Test: [1260/2387] eta: 0:08:00 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0745 (0.3203) acc1: 100.0000 (86.0674) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.6048 data: 0.4542 max mem: 3536 +Test: [1270/2387] eta: 0:07:59 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0745 (0.3184) acc1: 100.0000 (86.1768) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.6549 data: 0.5146 max mem: 3536 +Test: [1280/2387] eta: 0:07:54 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0806 (0.3166) acc1: 100.0000 (86.2821) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.5998 data: 0.4697 max mem: 3536 +Test: [1290/2387] eta: 0:07:54 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0955 (0.3149) acc1: 99.4792 (86.3849) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.6448 data: 0.5136 max mem: 3536 +Test: [1300/2387] eta: 0:07:51 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0925 (0.3133) acc1: 99.4792 (86.4864) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.7649 data: 0.6485 max mem: 3536 +Test: [1310/2387] eta: 0:07:49 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0990 (0.3117) acc1: 99.7396 (86.5853) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.6549 data: 0.5365 max mem: 3536 +Test: [1320/2387] eta: 0:07:44 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1099 (0.3103) acc1: 99.4792 (86.6831) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5199 data: 0.3954 max mem: 3536 +Test: [1330/2387] eta: 0:07:41 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1278 (0.3090) acc1: 99.4792 (86.7753) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.5250 data: 0.4023 max mem: 3536 +Test: [1340/2387] eta: 0:07:38 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1652 (0.3088) acc1: 97.9167 (86.7955) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.6149 data: 0.4900 max mem: 3536 +Test: [1350/2387] eta: 0:07:36 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0544 (0.3069) acc1: 100.0000 (86.8932) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.6208 data: 0.5041 max mem: 3537 +Test: [1360/2387] eta: 0:07:34 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0410 (0.3049) acc1: 100.0000 (86.9895) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.7349 data: 0.6255 max mem: 3537 +Test: [1370/2387] eta: 0:07:31 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0410 (0.3031) acc1: 100.0000 (87.0844) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.7037 data: 0.5897 max mem: 3537 +Test: [1380/2387] eta: 0:07:28 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0444 (0.3012) acc1: 100.0000 (87.1779) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.6799 data: 0.5691 max mem: 3537 +Test: [1390/2387] eta: 0:07:25 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0444 (0.2994) acc1: 100.0000 (87.2701) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.6849 data: 0.5716 max mem: 3537 +Test: [1400/2387] eta: 0:07:22 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0451 (0.2976) acc1: 100.0000 (87.3610) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.6598 data: 0.5340 max mem: 3537 +Test: [1410/2387] eta: 0:07:19 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0465 (0.2964) acc1: 100.0000 (87.4347) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.6398 data: 0.5262 max mem: 3537 +Test: [1420/2387] eta: 0:07:15 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1633 (0.2955) acc1: 98.9583 (87.5090) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.5999 data: 0.5051 max mem: 3537 +Test: [1430/2387] eta: 0:07:12 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1778 (0.2948) acc1: 97.9167 (87.5766) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.6049 data: 0.5031 max mem: 3537 +Test: [1440/2387] eta: 0:07:09 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1849 (0.2941) acc1: 97.9167 (87.6458) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.6104 data: 0.4900 max mem: 3537 +Test: [1450/2387] eta: 0:07:04 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1804 (0.2935) acc1: 98.1771 (87.7064) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.5649 data: 0.4308 max mem: 3537 +Test: [1460/2387] eta: 0:07:01 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1627 (0.2926) acc1: 98.1771 (87.7798) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.5593 data: 0.4241 max mem: 3537 +Test: [1470/2387] eta: 0:06:57 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1343 (0.2914) acc1: 98.9583 (87.8578) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.5510 data: 0.4262 max mem: 3537 +Test: [1480/2387] eta: 0:06:52 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1236 (0.2903) acc1: 99.7396 (87.9363) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.4699 data: 0.3545 max mem: 3537 +Test: [1490/2387] eta: 0:06:47 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1317 (0.2893) acc1: 99.7396 (88.0105) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.4141 data: 0.3001 max mem: 3537 +Test: [1500/2387] eta: 0:06:42 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1336 (0.2882) acc1: 99.7396 (88.0852) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.3549 data: 0.2178 max mem: 3537 +Test: [1510/2387] eta: 0:06:36 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1290 (0.2872) acc1: 99.7396 (88.1594) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2946 data: 0.1538 max mem: 3537 +Test: [1520/2387] eta: 0:06:31 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1574 (0.2865) acc1: 99.4792 (88.2302) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2902 data: 0.1655 max mem: 3537 +Test: [1530/2387] eta: 0:06:25 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1683 (0.2857) acc1: 99.4792 (88.3015) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2800 data: 0.1536 max mem: 3537 +Test: [1540/2387] eta: 0:06:19 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1570 (0.2848) acc1: 99.7396 (88.3752) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2247 data: 0.1105 max mem: 3537 +Test: [1550/2387] eta: 0:06:13 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1535 (0.2841) acc1: 99.7396 (88.4468) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2054 data: 0.1045 max mem: 3537 +Test: [1560/2387] eta: 0:06:08 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1606 (0.2834) acc1: 99.4792 (88.5175) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2498 data: 0.1547 max mem: 3537 +Test: [1570/2387] eta: 0:06:02 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1606 (0.2826) acc1: 99.4792 (88.5871) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2442 data: 0.1494 max mem: 3537 +Test: [1580/2387] eta: 0:05:57 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1737 (0.2820) acc1: 98.9583 (88.6385) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1999 data: 0.1037 max mem: 3538 +Test: [1590/2387] eta: 0:05:51 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1845 (0.2814) acc1: 96.3542 (88.6891) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1949 data: 0.0999 max mem: 3538 +Test: [1600/2387] eta: 0:05:45 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1663 (0.2808) acc1: 97.6562 (88.7465) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1955 data: 0.1018 max mem: 3538 +Test: [1610/2387] eta: 0:05:40 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1663 (0.2802) acc1: 97.9167 (88.8021) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2325 data: 0.1401 max mem: 3538 +Test: [1620/2387] eta: 0:05:34 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1958 (0.2797) acc1: 96.8750 (88.8467) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2095 data: 0.1188 max mem: 3538 +Test: [1630/2387] eta: 0:05:29 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2354 (0.2797) acc1: 95.0521 (88.8679) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1773 data: 0.0835 max mem: 3538 +Test: [1640/2387] eta: 0:05:23 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2831 (0.2800) acc1: 91.6667 (88.8686) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1860 data: 0.0905 max mem: 3538 +Test: [1650/2387] eta: 0:05:18 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2904 (0.2800) acc1: 91.1458 (88.8906) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1722 data: 0.0802 max mem: 3538 +Test: [1660/2387] eta: 0:05:13 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.3125 (0.2802) acc1: 90.3646 (88.8886) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1736 data: 0.0813 max mem: 3538 +Test: [1670/2387] eta: 0:05:07 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.3123 (0.2804) acc1: 89.8438 (88.8996) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1903 data: 0.0978 max mem: 3538 +Test: [1680/2387] eta: 0:05:02 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2957 (0.2805) acc1: 91.1458 (88.9158) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1959 data: 0.1052 max mem: 3538 +Test: [1690/2387] eta: 0:04:57 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2999 (0.2807) acc1: 91.1458 (88.9251) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1795 data: 0.0881 max mem: 3538 +Test: [1700/2387] eta: 0:04:51 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2925 (0.2804) acc1: 91.4062 (88.9625) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1846 data: 0.0932 max mem: 3538 +Test: [1710/2387] eta: 0:04:46 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1097 (0.2793) acc1: 99.7396 (89.0257) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1834 data: 0.0927 max mem: 3538 +Test: [1720/2387] eta: 0:04:41 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0938 (0.2782) acc1: 100.0000 (89.0891) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1917 data: 0.1006 max mem: 3538 +Test: [1730/2387] eta: 0:04:36 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0931 (0.2772) acc1: 100.0000 (89.1516) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2018 data: 0.1112 max mem: 3538 +Test: [1740/2387] eta: 0:04:31 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0976 (0.2762) acc1: 100.0000 (89.2130) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1788 data: 0.0881 max mem: 3538 +Test: [1750/2387] eta: 0:04:26 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0982 (0.2752) acc1: 100.0000 (89.2740) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1829 data: 0.0924 max mem: 3538 +Test: [1760/2387] eta: 0:04:21 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1003 (0.2742) acc1: 100.0000 (89.3345) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1861 data: 0.0958 max mem: 3538 +Test: [1770/2387] eta: 0:04:16 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1192 (0.2735) acc1: 99.7396 (89.3888) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1648 data: 0.0727 max mem: 3538 +Test: [1780/2387] eta: 0:04:11 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1354 (0.2727) acc1: 99.4792 (89.4475) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1681 data: 0.0765 max mem: 3538 +Test: [1790/2387] eta: 0:04:06 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1244 (0.2718) acc1: 100.0000 (89.5045) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1805 data: 0.0896 max mem: 3538 +Test: [1800/2387] eta: 0:04:01 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1231 (0.2710) acc1: 99.7396 (89.5614) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1892 data: 0.0951 max mem: 3538 +Test: [1810/2387] eta: 0:03:56 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1156 (0.2701) acc1: 99.7396 (89.6177) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1837 data: 0.0879 max mem: 3539 +Test: [1820/2387] eta: 0:03:51 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1085 (0.2693) acc1: 99.7396 (89.6734) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1521 data: 0.0553 max mem: 3539 +Test: [1830/2387] eta: 0:03:46 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1190 (0.2685) acc1: 100.0000 (89.7293) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1449 data: 0.0492 max mem: 3539 +Test: [1840/2387] eta: 0:03:42 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1150 (0.2676) acc1: 100.0000 (89.7849) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1778 data: 0.0828 max mem: 3539 +Test: [1850/2387] eta: 0:03:37 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1138 (0.2668) acc1: 100.0000 (89.8397) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0760 max mem: 3539 +Test: [1860/2387] eta: 0:03:32 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1153 (0.2660) acc1: 100.0000 (89.8943) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1682 data: 0.0735 max mem: 3539 +Test: [1870/2387] eta: 0:03:28 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1163 (0.2652) acc1: 100.0000 (89.9479) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1966 data: 0.1021 max mem: 3539 +Test: [1880/2387] eta: 0:03:23 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1143 (0.2644) acc1: 100.0000 (90.0010) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1967 data: 0.1039 max mem: 3539 +Test: [1890/2387] eta: 0:03:18 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1135 (0.2636) acc1: 100.0000 (90.0538) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1850 data: 0.0890 max mem: 3539 +Test: [1900/2387] eta: 0:03:14 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1153 (0.2628) acc1: 100.0000 (90.1055) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1512 data: 0.0557 max mem: 3539 +Test: [1910/2387] eta: 0:03:09 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1293 (0.2622) acc1: 100.0000 (90.1553) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1591 data: 0.0647 max mem: 3539 +Test: [1920/2387] eta: 0:03:05 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1512 (0.2618) acc1: 99.7396 (90.1927) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2065 data: 0.1127 max mem: 3539 +Test: [1930/2387] eta: 0:03:00 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1292 (0.2610) acc1: 100.0000 (90.2435) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1810 data: 0.0890 max mem: 3539 +Test: [1940/2387] eta: 0:02:56 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1173 (0.2605) acc1: 100.0000 (90.2919) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1370 data: 0.0441 max mem: 3539 +Test: [1950/2387] eta: 0:02:51 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1402 (0.2599) acc1: 100.0000 (90.3374) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1748 data: 0.0819 max mem: 3539 +Test: [1960/2387] eta: 0:02:47 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1478 (0.2595) acc1: 99.7396 (90.3748) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2199 data: 0.1257 max mem: 3539 +Test: [1970/2387] eta: 0:02:43 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2088 (0.2597) acc1: 97.1354 (90.3680) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2052 data: 0.1104 max mem: 3539 +Test: [1980/2387] eta: 0:02:38 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2178 (0.2594) acc1: 95.3125 (90.3989) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1749 data: 0.0822 max mem: 3539 +Test: [1990/2387] eta: 0:02:34 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1476 (0.2589) acc1: 98.4375 (90.4398) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1927 data: 0.1004 max mem: 3539 +Test: [2000/2387] eta: 0:02:30 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1472 (0.2583) acc1: 98.4375 (90.4807) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2099 data: 0.1176 max mem: 3539 +Test: [2010/2387] eta: 0:02:26 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1478 (0.2578) acc1: 98.4375 (90.5209) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2018 data: 0.1078 max mem: 3539 +Test: [2020/2387] eta: 0:02:21 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1474 (0.2572) acc1: 98.9583 (90.5617) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2058 data: 0.0986 max mem: 3539 +Test: [2030/2387] eta: 0:02:18 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1453 (0.2567) acc1: 98.9583 (90.6015) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.3178 data: 0.2079 max mem: 3540 +Test: [2040/2387] eta: 0:02:14 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1465 (0.2561) acc1: 98.9583 (90.6418) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.5540 data: 0.4546 max mem: 3540 +Test: [2050/2387] eta: 0:02:10 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1467 (0.2556) acc1: 98.9583 (90.6828) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.4765 data: 0.3809 max mem: 3540 +Test: [2060/2387] eta: 0:02:06 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1455 (0.2551) acc1: 98.6979 (90.7215) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2794 data: 0.1760 max mem: 3540 +Test: [2070/2387] eta: 0:02:02 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1447 (0.2545) acc1: 98.6979 (90.7613) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.3775 data: 0.2470 max mem: 3540 +Test: [2080/2387] eta: 0:01:59 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1439 (0.2540) acc1: 98.6979 (90.7989) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.5196 data: 0.3937 max mem: 3540 +Test: [2090/2387] eta: 0:01:55 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1458 (0.2535) acc1: 98.6979 (90.8378) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.5906 data: 0.4777 max mem: 3540 +Test: [2100/2387] eta: 0:01:51 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1465 (0.2530) acc1: 98.9583 (90.8761) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.5599 data: 0.4406 max mem: 3540 +Test: [2110/2387] eta: 0:01:48 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1436 (0.2525) acc1: 98.6979 (90.9133) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.4897 data: 0.3694 max mem: 3540 +Test: [2120/2387] eta: 0:01:44 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1485 (0.2523) acc1: 98.4375 (90.9329) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.4948 data: 0.3761 max mem: 3540 +Test: [2130/2387] eta: 0:01:40 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1675 (0.2519) acc1: 97.1354 (90.9631) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.4961 data: 0.3854 max mem: 3540 +Test: [2140/2387] eta: 0:01:36 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1699 (0.2516) acc1: 98.1771 (90.9910) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5258 data: 0.3931 max mem: 3540 +Test: [2150/2387] eta: 0:01:33 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1878 (0.2514) acc1: 97.1354 (91.0213) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.5580 data: 0.4227 max mem: 3540 +Test: [2160/2387] eta: 0:01:29 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1805 (0.2511) acc1: 97.3958 (91.0540) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.5489 data: 0.4312 max mem: 3540 +Test: [2170/2387] eta: 0:01:25 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2255 (0.2512) acc1: 95.5729 (91.0585) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.5748 data: 0.4546 max mem: 3540 +Test: [2180/2387] eta: 0:01:21 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2985 (0.2523) acc1: 91.4062 (90.9858) acc5: 100.0000 (100.0000) acc1_neg: 46.094 acc5_neg: 100.000 time: 0.5199 data: 0.4025 max mem: 3540 +Test: [2190/2387] eta: 0:01:17 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2822 (0.2526) acc1: 93.4896 (90.9809) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.4650 data: 0.3501 max mem: 3540 +Test: [2200/2387] eta: 0:01:13 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1488 (0.2520) acc1: 97.1354 (91.0145) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.4198 data: 0.3162 max mem: 3540 +Test: [2210/2387] eta: 0:01:10 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1019 (0.2513) acc1: 99.7396 (91.0541) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.4802 data: 0.3732 max mem: 3540 +Test: [2220/2387] eta: 0:01:06 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.0984 (0.2506) acc1: 100.0000 (91.0925) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5400 data: 0.4242 max mem: 3540 +Test: [2230/2387] eta: 0:01:02 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1023 (0.2500) acc1: 100.0000 (91.1314) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.4797 data: 0.3733 max mem: 3540 +Test: [2240/2387] eta: 0:00:58 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1057 (0.2494) acc1: 99.7396 (91.1651) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.5498 data: 0.4354 max mem: 3540 +Test: [2250/2387] eta: 0:00:54 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1166 (0.2489) acc1: 99.7396 (91.2013) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5745 data: 0.4490 max mem: 3540 +Test: [2260/2387] eta: 0:00:50 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1447 (0.2485) acc1: 99.7396 (91.2366) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5498 data: 0.4115 max mem: 3541 +Test: [2270/2387] eta: 0:00:46 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1332 (0.2479) acc1: 99.7396 (91.2733) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.5648 data: 0.4291 max mem: 3541 +Test: [2280/2387] eta: 0:00:42 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1201 (0.2474) acc1: 100.0000 (91.3074) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.5057 data: 0.3896 max mem: 3541 +Test: [2290/2387] eta: 0:00:38 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1151 (0.2468) acc1: 100.0000 (91.3437) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.4599 data: 0.3400 max mem: 3541 +Test: [2300/2387] eta: 0:00:34 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1346 (0.2464) acc1: 99.7396 (91.3782) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.4093 data: 0.2866 max mem: 3541 +Test: [2310/2387] eta: 0:00:30 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1498 (0.2461) acc1: 99.2188 (91.4065) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.4450 data: 0.3210 max mem: 3541 +Test: [2320/2387] eta: 0:00:26 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1722 (0.2459) acc1: 98.1771 (91.4286) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.4896 data: 0.3726 max mem: 3541 +Test: [2330/2387] eta: 0:00:22 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2003 (0.2459) acc1: 96.3542 (91.4450) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.3706 data: 0.2598 max mem: 3541 +Test: [2340/2387] eta: 0:00:18 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1910 (0.2456) acc1: 96.3542 (91.4723) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.3449 data: 0.2181 max mem: 3541 +Test: [2350/2387] eta: 0:00:14 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.1619 (0.2455) acc1: 98.9583 (91.4923) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.3406 data: 0.1934 max mem: 3541 +Test: [2360/2387] eta: 0:00:10 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2060 (0.2455) acc1: 96.8750 (91.5081) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.3352 data: 0.1779 max mem: 3541 +Test: [2370/2387] eta: 0:00:06 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2159 (0.2456) acc1: 96.3542 (91.5066) acc5: 100.0000 (100.0000) acc1_neg: 68.229 acc5_neg: 100.000 time: 0.2943 data: 0.1397 max mem: 3541 +Test: [2380/2387] eta: 0:00:02 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.2396 (0.2457) acc1: 92.4479 (91.5134) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.2561 data: 0.1077 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 15.525 acc5_pos: 100.000 loss: 0.3021 (0.2459) acc1: 90.8854 (91.5170) acc5: 100.0000 (100.0000) acc1_neg: 88.372 acc5_neg: 100.000 time: 0.2448 data: 0.1118 max mem: 3541 +Test: Total time: 0:15:50 (0.3984 s / it) +* Overall Acc@1 91.517 Acc@5 100.000 + target == real Acc@1 24.472 Acc@5 100.000 + target == fake Acc@1 96.225 Acc@5 100.000 + loss 0.246 + PR-AUC 0.9903 ROC-AUC 0.8769 EER 19.96% (thr=0.7710) +Accuracy of the network on the 916267 test images: 91.5% +[rank0]:[W725 17:41:57.428238515 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 17:42:17.696000 2565745 torch/distributed/run.py:766] +W0725 17:42:17.696000 2565745 torch/distributed/run.py:766] ***************************************** +W0725 17:42:17.696000 2565745 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:42:17.696000 2565745 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:42:49.269562767 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:42:49.318367869 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:42:50.704584599 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:42:51.626219016 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:42:51.257495221 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:42:51.287095590 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 17:42:52.593261322 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:42:52.673819422 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed1_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed1_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 2:40:03 acc1_pos: 26.302 acc5_pos: 100.000 loss: 1.1336 (1.1336) acc1: 26.3021 (26.3021) acc5: 100.0000 (100.0000) time: 18.5755 data: 13.3723 max mem: 3531 +Test: [ 10/517] eta: 0:15:15 acc1_pos: 12.240 acc5_pos: 100.000 loss: 1.1401 (1.1180) acc1: 21.0938 (22.8693) acc5: 100.0000 (100.0000) time: 1.8053 data: 1.2363 max mem: 3531 +Test: [ 20/517] eta: 0:08:18 acc1_pos: 24.219 acc5_pos: 100.000 loss: 1.1401 (1.0901) acc1: 19.7917 (24.6900) acc5: 100.0000 (100.0000) time: 0.1235 data: 0.0116 max mem: 3531 +Test: [ 30/517] eta: 0:08:17 acc1_pos: 15.104 acc5_pos: 100.000 loss: 1.1086 (1.0862) acc1: 19.0104 (24.6724) acc5: 100.0000 (100.0000) time: 0.5893 data: 0.4629 max mem: 3531 +Test: [ 40/517] eta: 0:07:29 acc1_pos: 17.969 acc5_pos: 100.000 loss: 1.1086 (1.0875) acc1: 18.7500 (23.0818) acc5: 100.0000 (100.0000) time: 0.8817 data: 0.7607 max mem: 3531 +Test: [ 50/517] eta: 0:07:03 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.9860 (1.0462) acc1: 20.3125 (26.8689) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.7298 data: 0.6143 max mem: 3531 +Test: [ 60/517] eta: 0:06:38 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.4359 (0.9405) acc1: 77.0833 (36.5010) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.7231 data: 0.5833 max mem: 3531 +Test: [ 70/517] eta: 0:06:12 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.4021 (0.8653) acc1: 88.8021 (43.8637) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.6501 data: 0.4976 max mem: 3531 +Test: [ 80/517] eta: 0:05:52 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.4174 (0.8127) acc1: 86.1979 (48.8265) acc5: 100.0000 (100.0000) acc1_neg: 72.396 acc5_neg: 100.000 time: 0.6098 data: 0.4758 max mem: 3531 +Test: [ 90/517] eta: 0:05:33 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3973 (0.7653) acc1: 89.3229 (53.6659) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.5896 data: 0.4812 max mem: 3531 +Test: [100/517] eta: 0:05:21 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3770 (0.7270) acc1: 90.8854 (57.2710) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.6262 data: 0.5189 max mem: 3531 +Test: [110/517] eta: 0:05:10 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3499 (0.6907) acc1: 90.8854 (60.4824) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.6799 data: 0.5667 max mem: 3531 +Test: [120/517] eta: 0:05:02 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3527 (0.6651) acc1: 91.9271 (62.9498) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.7235 data: 0.5996 max mem: 3531 +Test: [130/517] eta: 0:04:51 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3874 (0.6428) acc1: 89.5833 (64.9889) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.7098 data: 0.5725 max mem: 3531 +Test: [140/517] eta: 0:04:43 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.2529 (0.6143) acc1: 99.4792 (67.4664) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.6798 data: 0.5532 max mem: 3531 +Test: [150/517] eta: 0:04:32 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.2461 (0.5910) acc1: 100.0000 (69.5261) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.6549 data: 0.5424 max mem: 3531 +Test: [160/517] eta: 0:04:30 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.2641 (0.5701) acc1: 98.4375 (71.3315) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.7999 data: 0.6932 max mem: 3531 +Test: [170/517] eta: 0:04:15 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.2594 (0.5512) acc1: 98.4375 (72.9075) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.6949 data: 0.5794 max mem: 3531 +Test: [180/517] eta: 0:04:07 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.2606 (0.5356) acc1: 98.6979 (74.3065) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.5598 data: 0.4477 max mem: 3531 +Test: [190/517] eta: 0:03:58 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.2769 (0.5220) acc1: 98.6979 (75.5836) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.6798 data: 0.5818 max mem: 3531 +Test: [200/517] eta: 0:03:50 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.2686 (0.5086) acc1: 98.9583 (76.7322) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.6448 data: 0.5467 max mem: 3531 +Test: [210/517] eta: 0:03:46 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3083 (0.5048) acc1: 95.8333 (77.3326) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.8249 data: 0.7228 max mem: 3532 +Test: [220/517] eta: 0:03:36 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.4304 (0.5014) acc1: 88.2812 (77.8717) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.7449 data: 0.6379 max mem: 3532 +Test: [230/517] eta: 0:03:28 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.4304 (0.4987) acc1: 89.3229 (78.2896) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.6149 data: 0.4973 max mem: 3532 +Test: [240/517] eta: 0:03:21 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.4125 (0.4951) acc1: 90.1042 (78.8295) acc5: 100.0000 (100.0000) acc1_neg: 77.083 acc5_neg: 100.000 time: 0.7149 data: 0.5980 max mem: 3532 +Test: [250/517] eta: 0:03:13 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3776 (0.4880) acc1: 95.0521 (79.5101) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.7048 data: 0.6017 max mem: 3532 +Test: [260/517] eta: 0:03:05 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.2807 (0.4813) acc1: 99.4792 (80.1555) acc5: 100.0000 (100.0000) acc1_neg: 77.083 acc5_neg: 100.000 time: 0.6606 data: 0.5310 max mem: 3532 +Test: [270/517] eta: 0:02:56 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.4796 (0.4824) acc1: 79.4271 (80.0594) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.5798 data: 0.4457 max mem: 3532 +Test: [280/517] eta: 0:02:48 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.4935 (0.4810) acc1: 78.3854 (80.3177) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.5941 data: 0.4615 max mem: 3532 +Test: [290/517] eta: 0:02:38 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3589 (0.4763) acc1: 96.3542 (80.8858) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.4404 data: 0.2795 max mem: 3532 +Test: [300/517] eta: 0:02:28 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3589 (0.4734) acc1: 95.5729 (81.2985) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.2660 data: 0.0813 max mem: 3532 +Test: [310/517] eta: 0:02:18 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3970 (0.4705) acc1: 91.9271 (81.6561) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.2595 data: 0.1077 max mem: 3532 +Test: [320/517] eta: 0:02:09 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3613 (0.4651) acc1: 92.1875 (82.1059) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2687 data: 0.1689 max mem: 3532 +Test: [330/517] eta: 0:02:00 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.2441 (0.4583) acc1: 99.4792 (82.6292) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2708 data: 0.1748 max mem: 3532 +Test: [340/517] eta: 0:01:52 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.2314 (0.4544) acc1: 99.2188 (82.9232) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.2402 data: 0.1486 max mem: 3532 +Test: [350/517] eta: 0:01:43 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.4805 (0.4560) acc1: 83.8542 (82.8993) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.2188 data: 0.1271 max mem: 3532 +Test: [360/517] eta: 0:01:35 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.4901 (0.4551) acc1: 83.8542 (83.0130) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2146 data: 0.1209 max mem: 3532 +Test: [370/517] eta: 0:01:28 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.2828 (0.4501) acc1: 99.2188 (83.4576) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2954 data: 0.1165 max mem: 3532 +Test: [380/517] eta: 0:01:21 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.2941 (0.4474) acc1: 99.2188 (83.8015) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.2699 data: 0.0890 max mem: 3532 +Test: [390/517] eta: 0:01:13 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3638 (0.4471) acc1: 92.9688 (83.8562) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1700 data: 0.0729 max mem: 3532 +Test: [400/517] eta: 0:01:06 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3037 (0.4429) acc1: 96.0938 (84.2328) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1949 data: 0.1016 max mem: 3532 +Test: [410/517] eta: 0:01:00 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.2967 (0.4409) acc1: 98.9583 (84.5492) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2342 data: 0.1428 max mem: 3532 +Test: [420/517] eta: 0:00:53 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3352 (0.4381) acc1: 98.9583 (84.8958) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2354 data: 0.1446 max mem: 3532 +Test: [430/517] eta: 0:00:47 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3162 (0.4352) acc1: 99.4792 (85.1738) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.2455 data: 0.1552 max mem: 3532 +Test: [440/517] eta: 0:00:41 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3756 (0.4349) acc1: 96.3542 (85.3074) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.2501 data: 0.1593 max mem: 3533 +Test: [450/517] eta: 0:00:35 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3955 (0.4336) acc1: 93.2292 (85.5154) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2493 data: 0.1565 max mem: 3533 +Test: [460/517] eta: 0:00:30 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3812 (0.4327) acc1: 94.7917 (85.6771) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.2663 data: 0.1731 max mem: 3533 +Test: [470/517] eta: 0:00:24 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3712 (0.4304) acc1: 95.5729 (85.9248) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2669 data: 0.1758 max mem: 3533 +Test: [480/517] eta: 0:00:19 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3016 (0.4270) acc1: 96.3542 (86.1043) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2453 data: 0.1551 max mem: 3533 +Test: [490/517] eta: 0:00:13 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3016 (0.4247) acc1: 95.0521 (86.2785) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.2400 data: 0.1496 max mem: 3533 +Test: [500/517] eta: 0:00:08 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.4667 (0.4277) acc1: 83.0729 (85.9250) acc5: 100.0000 (100.0000) acc1_neg: 56.510 acc5_neg: 100.000 time: 0.2009 data: 0.1110 max mem: 3533 +Test: [510/517] eta: 0:00:03 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.5331 (0.4280) acc1: 72.6562 (85.9013) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1703 data: 0.0807 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 52.752 acc5_pos: 100.000 loss: 0.3862 (0.4268) acc1: 86.7188 (86.0402) acc5: 100.0000 (100.0000) acc1_neg: 99.647 acc5_neg: 100.000 time: 0.1631 data: 0.0719 max mem: 3533 +Test: Total time: 0:04:16 (0.4955 s / it) +* Overall Acc@1 86.040 Acc@5 100.000 + target == real Acc@1 24.055 Acc@5 100.000 + target == fake Acc@1 92.470 Acc@5 100.000 + loss 0.427 + PR-AUC 0.9417 ROC-AUC 0.6594 EER 38.77% (thr=0.6914) +Accuracy of the network on the 198427 test images: 86.0% +[rank0]:[W725 17:47:22.362651144 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0725 17:47:38.404000 2566964 torch/distributed/run.py:766] +W0725 17:47:38.404000 2566964 torch/distributed/run.py:766] ***************************************** +W0725 17:47:38.404000 2566964 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:47:38.404000 2566964 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:48:06.648162946 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:48:07.819825753 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:48:07.865894636 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:48:07.435081609 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 17:48:08.667613936 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:48:08.333739985 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:48:08.442693537 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:48:08.539064138 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed1_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed1_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 1:38:32 acc1_pos: 24.479 acc5_pos: 100.000 loss: 0.9853 (0.9853) acc1: 24.4792 (24.4792) acc5: 100.0000 (100.0000) time: 15.1986 data: 12.1598 max mem: 3531 +Test: [ 10/389] eta: 0:10:43 acc1_pos: 82.812 acc5_pos: 100.000 loss: 0.7133 (0.8052) acc1: 48.4375 (48.6269) acc5: 100.0000 (100.0000) time: 1.6984 data: 1.3186 max mem: 3531 +Test: [ 20/389] eta: 0:05:50 acc1_pos: 6.510 acc5_pos: 100.000 loss: 0.7133 (0.8309) acc1: 45.5729 (44.7297) acc5: 100.0000 (100.0000) time: 0.2376 data: 0.1196 max mem: 3531 +Test: [ 30/389] eta: 0:04:26 acc1_pos: 48.958 acc5_pos: 100.000 loss: 0.8546 (0.8399) acc1: 22.1354 (39.4909) acc5: 100.0000 (100.0000) time: 0.2184 data: 0.0982 max mem: 3531 +Test: [ 40/389] eta: 0:03:41 acc1_pos: 7.031 acc5_pos: 100.000 loss: 0.8589 (0.8490) acc1: 23.6979 (37.6905) acc5: 100.0000 (100.0000) time: 0.3049 data: 0.1851 max mem: 3531 +Test: [ 50/389] eta: 0:03:12 acc1_pos: 90.365 acc5_pos: 100.000 loss: 0.8589 (0.8429) acc1: 25.0000 (39.7518) acc5: 100.0000 (100.0000) time: 0.2959 data: 0.1379 max mem: 3531 +Test: [ 60/389] eta: 0:02:48 acc1_pos: 3.385 acc5_pos: 100.000 loss: 0.7465 (0.8108) acc1: 48.6979 (43.8397) acc5: 100.0000 (100.0000) time: 0.2566 data: 0.0582 max mem: 3531 +Test: [ 70/389] eta: 0:02:33 acc1_pos: 53.906 acc5_pos: 100.000 loss: 0.7664 (0.8397) acc1: 47.1354 (41.2962) acc5: 100.0000 (100.0000) time: 0.2568 data: 0.0553 max mem: 3531 +Test: [ 80/389] eta: 0:02:16 acc1_pos: 77.604 acc5_pos: 100.000 loss: 1.0543 (0.8515) acc1: 10.1562 (40.5961) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.0488 max mem: 3531 +Test: [ 90/389] eta: 0:02:04 acc1_pos: 13.021 acc5_pos: 100.000 loss: 0.6499 (0.8281) acc1: 57.5521 (42.5166) acc5: 100.0000 (100.0000) time: 0.1935 data: 0.0671 max mem: 3531 +Test: [100/389] eta: 0:01:55 acc1_pos: 21.354 acc5_pos: 100.000 loss: 0.7771 (0.8293) acc1: 41.4062 (41.8472) acc5: 100.0000 (100.0000) time: 0.2296 data: 0.1366 max mem: 3531 +Test: [110/389] eta: 0:01:46 acc1_pos: 99.479 acc5_pos: 100.000 loss: 0.8251 (0.8270) acc1: 34.6354 (42.1077) acc5: 100.0000 (100.0000) time: 0.2113 data: 0.1164 max mem: 3531 +Test: [120/389] eta: 0:01:37 acc1_pos: 65.885 acc5_pos: 100.000 loss: 0.7818 (0.8217) acc1: 40.6250 (42.7320) acc5: 100.0000 (100.0000) time: 0.1849 data: 0.0924 max mem: 3531 +Test: [130/389] eta: 0:01:30 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.6056 (0.8044) acc1: 65.1042 (44.5670) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1667 data: 0.0753 max mem: 3531 +Test: [140/389] eta: 0:01:23 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.4371 (0.7782) acc1: 86.1979 (47.4217) acc5: 100.0000 (100.0000) acc1_neg: 44.792 acc5_neg: 100.000 time: 0.1621 data: 0.0699 max mem: 3531 +Test: [150/389] eta: 0:01:17 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.4091 (0.7646) acc1: 91.1458 (49.0635) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1766 data: 0.0848 max mem: 3531 +Test: [160/389] eta: 0:01:12 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.4368 (0.7415) acc1: 86.9792 (51.5253) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1899 data: 0.0977 max mem: 3531 +Test: [170/389] eta: 0:01:07 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.3589 (0.7210) acc1: 97.6562 (54.0753) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1799 data: 0.0878 max mem: 3531 +Test: [180/389] eta: 0:01:03 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.3614 (0.7068) acc1: 95.0521 (55.8227) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1747 data: 0.0823 max mem: 3531 +Test: [190/389] eta: 0:00:58 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.4510 (0.6958) acc1: 89.5833 (57.2303) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1881 data: 0.0956 max mem: 3531 +Test: [200/389] eta: 0:00:54 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.4454 (0.6835) acc1: 90.3646 (58.7155) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.1900 data: 0.0980 max mem: 3531 +Test: [210/389] eta: 0:00:51 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.4032 (0.6694) acc1: 94.2708 (60.2019) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1918 data: 0.1003 max mem: 3532 +Test: [220/389] eta: 0:00:47 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.2539 (0.6519) acc1: 99.4792 (61.9532) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1805 data: 0.0894 max mem: 3532 +Test: [230/389] eta: 0:00:43 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.3202 (0.6410) acc1: 98.4375 (63.3917) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1800 data: 0.0870 max mem: 3532 +Test: [240/389] eta: 0:00:40 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.3878 (0.6303) acc1: 96.6146 (64.6395) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1901 data: 0.0960 max mem: 3532 +Test: [250/389] eta: 0:00:37 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.3472 (0.6183) acc1: 98.4375 (66.0141) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1848 data: 0.0922 max mem: 3532 +Test: [260/389] eta: 0:00:34 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.3433 (0.6146) acc1: 99.2188 (66.5858) acc5: 100.0000 (100.0000) acc1_neg: 25.521 acc5_neg: 100.000 time: 0.1739 data: 0.0827 max mem: 3532 +Test: [270/389] eta: 0:00:31 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.4932 (0.6181) acc1: 75.5208 (66.0642) acc5: 100.0000 (100.0000) acc1_neg: 75.521 acc5_neg: 100.000 time: 0.1806 data: 0.0895 max mem: 3532 +Test: [280/389] eta: 0:00:28 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.6467 (0.6186) acc1: 59.3750 (65.9762) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1831 data: 0.0918 max mem: 3532 +Test: [290/389] eta: 0:00:25 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.6055 (0.6133) acc1: 69.5312 (66.6497) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1792 data: 0.0857 max mem: 3532 +Test: [300/389] eta: 0:00:22 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.4189 (0.6074) acc1: 92.4479 (67.5569) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.1907 data: 0.0940 max mem: 3532 +Test: [310/389] eta: 0:00:19 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.4811 (0.6079) acc1: 85.9375 (67.5250) acc5: 100.0000 (100.0000) acc1_neg: 54.948 acc5_neg: 100.000 time: 0.1901 data: 0.0963 max mem: 3532 +Test: [320/389] eta: 0:00:17 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.4719 (0.6018) acc1: 85.4167 (68.2162) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1857 data: 0.0937 max mem: 3532 +Test: [330/389] eta: 0:00:14 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.4693 (0.5981) acc1: 89.5833 (68.8208) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1924 data: 0.1002 max mem: 3532 +Test: [340/389] eta: 0:00:12 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.4287 (0.5893) acc1: 91.1458 (69.6351) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1943 data: 0.1023 max mem: 3532 +Test: [350/389] eta: 0:00:09 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.2791 (0.5967) acc1: 97.6562 (68.9444) acc5: 100.0000 (100.0000) acc1_neg: 2.344 acc5_neg: 100.000 time: 0.1855 data: 0.0934 max mem: 3532 +Test: [360/389] eta: 0:00:07 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.6213 (0.6003) acc1: 61.7188 (68.4788) acc5: 100.0000 (100.0000) acc1_neg: 70.573 acc5_neg: 100.000 time: 0.1748 data: 0.0811 max mem: 3532 +Test: [370/389] eta: 0:00:04 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.5822 (0.5960) acc1: 68.2292 (68.8321) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1675 data: 0.0744 max mem: 3532 +Test: [380/389] eta: 0:00:02 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.3976 (0.5914) acc1: 82.2917 (69.1403) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1514 data: 0.0619 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 41.718 acc5_pos: 100.000 loss: 0.3344 (0.5880) acc1: 88.8021 (69.5115) acc5: 100.0000 (100.0000) acc1_neg: 74.863 acc5_neg: 100.000 time: 0.1416 data: 0.0466 max mem: 3532 +Test: Total time: 0:01:32 (0.2374 s / it) +* Overall Acc@1 69.511 Acc@5 100.000 + target == real Acc@1 42.851 Acc@5 100.000 + target == fake Acc@1 81.904 Acc@5 100.000 + loss 0.588 + PR-AUC 0.8086 ROC-AUC 0.6924 EER 35.78% (thr=0.6147) +Accuracy of the network on the 149358 test images: 69.5% +[rank0]:[W725 17:49:58.147738057 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 17:50:11.788000 2568059 torch/distributed/run.py:766] +W0725 17:50:11.788000 2568059 torch/distributed/run.py:766] ***************************************** +W0725 17:50:11.788000 2568059 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:50:11.788000 2568059 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:50:39.866516573 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:50:40.649526118 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:50:40.043934760 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:50:41.968549160 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:50:41.201556965 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 17:50:41.320630815 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:50:41.340936695 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:50:41.363131237 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed1_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed1_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 1:36:37 acc1_pos: 80.208 acc5_pos: 100.000 loss: 0.4223 (0.4223) acc1: 80.2083 (80.2083) acc5: 100.0000 (100.0000) time: 17.4090 data: 12.7675 max mem: 3531 +Test: [ 10/333] eta: 0:09:08 acc1_pos: 82.812 acc5_pos: 100.000 loss: 0.3405 (0.3341) acc1: 87.5000 (87.2159) acc5: 100.0000 (100.0000) time: 1.6969 data: 1.1613 max mem: 3531 +Test: [ 20/333] eta: 0:05:45 acc1_pos: 92.708 acc5_pos: 100.000 loss: 0.3340 (0.3332) acc1: 87.2396 (86.0987) acc5: 100.0000 (100.0000) time: 0.2878 data: 0.1677 max mem: 3531 +Test: [ 30/333] eta: 0:04:46 acc1_pos: 40.365 acc5_pos: 100.000 loss: 0.4042 (0.4939) acc1: 72.1354 (73.1183) acc5: 100.0000 (100.0000) time: 0.5348 data: 0.4151 max mem: 3531 +Test: [ 40/333] eta: 0:04:08 acc1_pos: 45.573 acc5_pos: 100.000 loss: 0.8626 (0.5783) acc1: 42.4479 (66.1331) acc5: 100.0000 (100.0000) time: 0.5808 data: 0.4439 max mem: 3531 +Test: [ 50/333] eta: 0:03:35 acc1_pos: 49.740 acc5_pos: 100.000 loss: 0.8460 (0.6316) acc1: 43.7500 (61.7392) acc5: 100.0000 (100.0000) time: 0.4712 data: 0.2961 max mem: 3531 +Test: [ 60/333] eta: 0:03:19 acc1_pos: 33.333 acc5_pos: 100.000 loss: 0.8429 (0.6650) acc1: 45.8333 (59.1359) acc5: 100.0000 (100.0000) time: 0.4846 data: 0.2806 max mem: 3531 +Test: [ 70/333] eta: 0:02:52 acc1_pos: 37.500 acc5_pos: 100.000 loss: 0.8429 (0.6902) acc1: 46.6146 (57.4090) acc5: 100.0000 (100.0000) time: 0.3857 data: 0.1810 max mem: 3531 +Test: [ 80/333] eta: 0:02:31 acc1_pos: 47.135 acc5_pos: 100.000 loss: 0.8195 (0.7097) acc1: 44.5312 (55.9992) acc5: 100.0000 (100.0000) time: 0.2047 data: 0.0005 max mem: 3531 +Test: [ 90/333] eta: 0:02:14 acc1_pos: 53.646 acc5_pos: 100.000 loss: 0.7761 (0.7156) acc1: 51.3021 (55.5746) acc5: 100.0000 (100.0000) time: 0.1928 data: 0.0366 max mem: 3531 +Test: [100/333] eta: 0:02:01 acc1_pos: 58.333 acc5_pos: 100.000 loss: 0.7711 (0.7243) acc1: 53.3854 (55.0278) acc5: 100.0000 (100.0000) time: 0.1997 data: 0.1005 max mem: 3531 +Test: [110/333] eta: 0:01:49 acc1_pos: 52.083 acc5_pos: 100.000 loss: 0.7694 (0.7287) acc1: 51.5625 (54.6734) acc5: 100.0000 (100.0000) time: 0.2147 data: 0.1211 max mem: 3531 +Test: [120/333] eta: 0:01:40 acc1_pos: 42.708 acc5_pos: 100.000 loss: 0.7685 (0.7321) acc1: 51.5625 (54.3862) acc5: 100.0000 (100.0000) time: 0.2195 data: 0.1249 max mem: 3531 +Test: [130/333] eta: 0:01:31 acc1_pos: 75.260 acc5_pos: 100.000 loss: 0.7581 (0.7333) acc1: 50.7812 (54.4092) acc5: 100.0000 (100.0000) time: 0.2049 data: 0.1126 max mem: 3531 +Test: [140/333] eta: 0:01:23 acc1_pos: 67.188 acc5_pos: 100.000 loss: 0.7423 (0.7342) acc1: 52.3438 (54.2978) acc5: 100.0000 (100.0000) time: 0.1858 data: 0.0941 max mem: 3531 +Test: [150/333] eta: 0:01:16 acc1_pos: 54.948 acc5_pos: 100.000 loss: 0.7593 (0.7383) acc1: 51.3021 (53.8114) acc5: 100.0000 (100.0000) time: 0.2099 data: 0.1155 max mem: 3531 +Test: [160/333] eta: 0:01:09 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.7842 (0.7441) acc1: 48.9583 (53.5019) acc5: 100.0000 (100.0000) acc1_neg: 52.344 acc5_neg: 100.000 time: 0.2094 data: 0.1166 max mem: 3531 +Test: [170/333] eta: 0:01:04 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.8174 (0.7475) acc1: 49.2188 (53.4555) acc5: 100.0000 (100.0000) acc1_neg: 59.375 acc5_neg: 100.000 time: 0.2199 data: 0.1290 max mem: 3531 +Test: [180/333] eta: 0:00:58 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.7515 (0.7450) acc1: 56.5104 (53.8372) acc5: 100.0000 (100.0000) acc1_neg: 65.885 acc5_neg: 100.000 time: 0.2399 data: 0.1483 max mem: 3531 +Test: [190/333] eta: 0:00:53 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.6718 (0.7400) acc1: 59.1146 (54.3575) acc5: 100.0000 (100.0000) acc1_neg: 56.250 acc5_neg: 100.000 time: 0.2118 data: 0.1171 max mem: 3531 +Test: [200/333] eta: 0:00:48 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.6405 (0.7364) acc1: 59.8958 (54.7821) acc5: 100.0000 (100.0000) acc1_neg: 59.896 acc5_neg: 100.000 time: 0.2070 data: 0.1122 max mem: 3531 +Test: [210/333] eta: 0:00:44 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.7020 (0.7373) acc1: 59.6354 (54.9973) acc5: 100.0000 (100.0000) acc1_neg: 57.031 acc5_neg: 100.000 time: 0.2146 data: 0.1240 max mem: 3532 +Test: [220/333] eta: 0:00:39 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.7577 (0.7401) acc1: 58.8542 (54.9515) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.2078 data: 0.1173 max mem: 3532 +Test: [230/333] eta: 0:00:35 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.5294 (0.7297) acc1: 75.0000 (55.9918) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.1793 data: 0.0882 max mem: 3532 +Test: [240/333] eta: 0:00:31 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.5294 (0.7235) acc1: 76.5625 (56.5936) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.1634 data: 0.0700 max mem: 3532 +Test: [250/333] eta: 0:00:27 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.6050 (0.7197) acc1: 66.6667 (56.9752) acc5: 100.0000 (100.0000) acc1_neg: 58.073 acc5_neg: 100.000 time: 0.1679 data: 0.0728 max mem: 3532 +Test: [260/333] eta: 0:00:23 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.6479 (0.7180) acc1: 64.8438 (57.2148) acc5: 100.0000 (100.0000) acc1_neg: 67.448 acc5_neg: 100.000 time: 0.1627 data: 0.0704 max mem: 3532 +Test: [270/333] eta: 0:00:20 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.6195 (0.7135) acc1: 65.3646 (57.6280) acc5: 100.0000 (100.0000) acc1_neg: 66.927 acc5_neg: 100.000 time: 0.1633 data: 0.0717 max mem: 3532 +Test: [280/333] eta: 0:00:16 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.6139 (0.7137) acc1: 65.3646 (57.5947) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.1671 data: 0.0752 max mem: 3532 +Test: [290/333] eta: 0:00:13 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.7915 (0.7165) acc1: 49.7396 (57.3821) acc5: 100.0000 (100.0000) acc1_neg: 58.073 acc5_neg: 100.000 time: 0.1865 data: 0.0941 max mem: 3532 +Test: [300/333] eta: 0:00:09 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.7844 (0.7187) acc1: 52.6042 (57.2865) acc5: 100.0000 (100.0000) acc1_neg: 59.896 acc5_neg: 100.000 time: 0.1699 data: 0.0746 max mem: 3532 +Test: [310/333] eta: 0:00:06 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.7186 (0.7143) acc1: 57.8125 (57.7263) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.1519 data: 0.0572 max mem: 3532 +Test: [320/333] eta: 0:00:03 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.5509 (0.7103) acc1: 73.1771 (58.1678) acc5: 100.0000 (100.0000) acc1_neg: 44.792 acc5_neg: 100.000 time: 0.1801 data: 0.0891 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.8247 (0.7157) acc1: 50.5208 (57.8345) acc5: 100.0000 (100.0000) acc1_neg: 48.177 acc5_neg: 100.000 time: 0.1715 data: 0.0822 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 41.776 acc5_pos: 100.000 loss: 0.8606 (0.7168) acc1: 48.1771 (57.7638) acc5: 100.0000 (100.0000) acc1_neg: 46.197 acc5_neg: 100.000 time: 0.1747 data: 0.0822 max mem: 3532 +Test: Total time: 0:01:36 (0.2903 s / it) +* Overall Acc@1 57.764 Acc@5 100.000 + target == real Acc@1 53.651 Acc@5 100.000 + target == fake Acc@1 61.425 Acc@5 100.000 + loss 0.717 + PR-AUC 0.6121 ROC-AUC 0.6052 EER 42.53% (thr=0.5244) +Accuracy of the network on the 127843 test images: 57.8% +[rank0]:[W725 17:52:35.321085930 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 17:52:51.229000 2569176 torch/distributed/run.py:766] +W0725 17:52:51.229000 2569176 torch/distributed/run.py:766] ***************************************** +W0725 17:52:51.229000 2569176 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:52:51.229000 2569176 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:53:15.046285971 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:53:15.507472365 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank6]:[W725 17:53:15.509150911 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W725 17:53:15.515520132 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 17:53:16.577508639 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:53:16.420715899 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:53:17.709650440 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:53:17.832043064 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed1_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed1_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 1:06:45 acc1_pos: 94.531 acc5_pos: 100.000 loss: 0.2751 (0.2751) acc1: 94.5312 (94.5312) acc5: 100.0000 (100.0000) time: 17.0444 data: 13.8632 max mem: 3531 +Test: [ 10/235] eta: 0:06:19 acc1_pos: 93.750 acc5_pos: 100.000 loss: 0.2744 (0.2752) acc1: 93.7500 (93.4186) acc5: 100.0000 (100.0000) time: 1.6865 data: 1.2607 max mem: 3531 +Test: [ 20/235] eta: 0:03:32 acc1_pos: 91.927 acc5_pos: 100.000 loss: 0.2675 (0.2732) acc1: 93.4896 (93.4524) acc5: 100.0000 (100.0000) time: 0.1848 data: 0.0562 max mem: 3531 +Test: [ 30/235] eta: 0:03:08 acc1_pos: 94.792 acc5_pos: 100.000 loss: 0.2648 (0.2712) acc1: 93.7500 (93.5484) acc5: 100.0000 (100.0000) time: 0.4994 data: 0.3828 max mem: 3531 +Test: [ 40/235] eta: 0:02:40 acc1_pos: 93.229 acc5_pos: 100.000 loss: 0.2678 (0.2714) acc1: 93.7500 (93.6039) acc5: 100.0000 (100.0000) time: 0.6499 data: 0.5321 max mem: 3531 +Test: [ 50/235] eta: 0:02:20 acc1_pos: 91.667 acc5_pos: 100.000 loss: 0.2715 (0.2721) acc1: 93.4896 (93.5815) acc5: 100.0000 (100.0000) time: 0.5149 data: 0.4037 max mem: 3531 +Test: [ 60/235] eta: 0:02:05 acc1_pos: 94.271 acc5_pos: 100.000 loss: 0.2756 (0.2734) acc1: 93.2292 (93.5280) acc5: 100.0000 (100.0000) time: 0.4998 data: 0.3832 max mem: 3531 +Test: [ 70/235] eta: 0:01:52 acc1_pos: 92.308 acc5_pos: 100.000 loss: 0.2957 (0.3868) acc1: 90.1042 (85.5560) acc5: 100.0000 (100.0000) acc1_neg: 26.042 acc5_neg: 100.000 time: 0.4749 data: 0.3517 max mem: 3531 +Test: [ 80/235] eta: 0:01:42 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.1831 (0.4866) acc1: 29.6875 (78.6233) acc5: 100.0000 (100.0000) acc1_neg: 28.125 acc5_neg: 100.000 time: 0.4998 data: 0.3644 max mem: 3531 +Test: [ 90/235] eta: 0:01:31 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.2004 (0.5643) acc1: 29.1667 (73.2000) acc5: 100.0000 (100.0000) acc1_neg: 29.948 acc5_neg: 100.000 time: 0.4514 data: 0.2998 max mem: 3531 +Test: [100/235] eta: 0:01:21 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.2004 (0.6248) acc1: 29.6875 (68.9485) acc5: 100.0000 (100.0000) acc1_neg: 27.604 acc5_neg: 100.000 time: 0.3523 data: 0.1699 max mem: 3531 +Test: [110/235] eta: 0:01:10 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.1733 (0.6756) acc1: 29.6875 (65.3411) acc5: 100.0000 (100.0000) acc1_neg: 30.990 acc5_neg: 100.000 time: 0.2752 data: 0.0679 max mem: 3531 +Test: [120/235] eta: 0:01:01 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.2344 (0.7460) acc1: 24.7396 (60.9483) acc5: 100.0000 (100.0000) acc1_neg: 13.281 acc5_neg: 100.000 time: 0.2140 data: 0.0364 max mem: 3531 +Test: [130/235] eta: 0:00:53 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.5439 (0.8078) acc1: 10.9375 (57.1028) acc5: 100.0000 (100.0000) acc1_neg: 11.458 acc5_neg: 100.000 time: 0.2187 data: 0.0983 max mem: 3531 +Test: [140/235] eta: 0:00:46 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.5732 (0.8635) acc1: 10.1562 (53.7622) acc5: 100.0000 (100.0000) acc1_neg: 7.812 acc5_neg: 100.000 time: 0.2233 data: 0.1281 max mem: 3531 +Test: [150/235] eta: 0:00:40 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.5732 (0.9094) acc1: 9.8958 (50.8744) acc5: 100.0000 (100.0000) acc1_neg: 11.719 acc5_neg: 100.000 time: 0.2167 data: 0.1219 max mem: 3531 +Test: [160/235] eta: 0:00:34 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.5594 (0.9501) acc1: 10.6771 (48.4019) acc5: 100.0000 (100.0000) acc1_neg: 12.760 acc5_neg: 100.000 time: 0.2056 data: 0.1088 max mem: 3531 +Test: [170/235] eta: 0:00:28 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.5072 (0.9526) acc1: 12.7604 (47.8710) acc5: 100.0000 (100.0000) acc1_neg: 38.542 acc5_neg: 100.000 time: 0.2293 data: 0.1315 max mem: 3531 +Test: [180/235] eta: 0:00:23 acc1_pos: 92.308 acc5_pos: 100.000 loss: 0.9272 (0.9508) acc1: 42.4479 (47.6275) acc5: 100.0000 (100.0000) acc1_neg: 43.750 acc5_neg: 100.000 time: 0.2272 data: 0.1343 max mem: 3531 +Test: [190/235] eta: 0:00:18 acc1_pos: 92.308 acc5_pos: 100.000 loss: 0.9193 (0.9486) acc1: 43.7500 (47.4408) acc5: 100.0000 (100.0000) acc1_neg: 45.312 acc5_neg: 100.000 time: 0.2157 data: 0.1242 max mem: 3531 +Test: [200/235] eta: 0:00:14 acc1_pos: 92.308 acc5_pos: 100.000 loss: 0.9193 (0.9478) acc1: 42.9688 (47.1743) acc5: 100.0000 (100.0000) acc1_neg: 41.146 acc5_neg: 100.000 time: 0.2208 data: 0.1272 max mem: 3531 +Test: [210/235] eta: 0:00:10 acc1_pos: 92.308 acc5_pos: 100.000 loss: 0.9263 (0.9462) acc1: 42.7083 (46.9910) acc5: 100.0000 (100.0000) acc1_neg: 42.969 acc5_neg: 100.000 time: 0.1985 data: 0.1061 max mem: 3532 +Test: [220/235] eta: 0:00:05 acc1_pos: 92.308 acc5_pos: 100.000 loss: 0.9117 (0.9447) acc1: 43.7500 (46.8691) acc5: 100.0000 (100.0000) acc1_neg: 43.490 acc5_neg: 100.000 time: 0.1591 data: 0.0696 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 92.308 acc5_pos: 100.000 loss: 0.9032 (0.9429) acc1: 44.2708 (46.7938) acc5: 100.0000 (100.0000) acc1_neg: 46.354 acc5_neg: 100.000 time: 0.1288 data: 0.0392 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 92.308 acc5_pos: 100.000 loss: 0.9084 (0.9431) acc1: 44.0104 (46.7417) acc5: 100.0000 (100.0000) acc1_neg: 52.174 acc5_neg: 100.000 time: 0.1391 data: 0.0461 max mem: 3532 +Test: Total time: 0:01:27 (0.3737 s / it) +* Overall Acc@1 46.742 Acc@5 100.000 + target == real Acc@1 93.516 Acc@5 100.000 + target == fake Acc@1 29.825 Acc@5 100.000 + loss 0.943 + PR-AUC 0.8761 ROC-AUC 0.7187 EER 34.00% (thr=0.2389) +Accuracy of the network on the 89879 test images: 46.7% +[rank0]:[W725 17:54:59.734439999 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 17:55:09.436000 2570258 torch/distributed/run.py:766] +W0725 17:55:09.436000 2570258 torch/distributed/run.py:766] ***************************************** +W0725 17:55:09.436000 2570258 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:55:09.436000 2570258 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:55:41.876547669 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:55:41.989074934 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:55:41.229665472 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:55:42.657421111 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W725 17:55:42.661398096 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:55:43.679066028 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 17:55:43.740048038 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:55:43.774583889 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed1_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n32_classseed1_resampled_n111-046-081_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:25:15 acc1_pos: 35.417 acc5_pos: 100.000 loss: 0.8183 (0.8183) acc1: 35.4167 (35.4167) acc5: 100.0000 (100.0000) time: 5.6323 data: 3.6579 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/269] eta: 0:02:37 acc1_pos: 61.458 acc5_pos: 100.000 loss: 0.6768 (0.6813) acc1: 55.7292 (57.0786) acc5: 100.0000 (100.0000) time: 0.6079 data: 0.3329 max mem: 3531 +Test: [ 20/269] eta: 0:01:30 acc1_pos: 79.167 acc5_pos: 100.000 loss: 0.6529 (0.6398) acc1: 59.3750 (62.1032) acc5: 100.0000 (100.0000) time: 0.1008 data: 0.0004 max mem: 3531 +Test: [ 30/269] eta: 0:01:06 acc1_pos: 65.104 acc5_pos: 100.000 loss: 0.5707 (0.6191) acc1: 65.1042 (64.5497) acc5: 100.0000 (100.0000) time: 0.0957 data: 0.0004 max mem: 3531 +Test: [ 40/269] eta: 0:01:20 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.5230 (0.6034) acc1: 71.6146 (67.6321) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.3377 data: 0.2429 max mem: 3531 +Test: [ 50/269] eta: 0:01:30 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.5851 (0.6083) acc1: 71.6146 (67.3356) acc5: 100.0000 (100.0000) acc1_neg: 44.271 acc5_neg: 100.000 time: 0.6249 data: 0.5293 max mem: 3531 +Test: [ 60/269] eta: 0:01:25 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.6283 (0.6168) acc1: 61.1979 (66.1629) acc5: 100.0000 (100.0000) acc1_neg: 59.896 acc5_neg: 100.000 time: 0.5348 data: 0.4398 max mem: 3531 +Test: [ 70/269] eta: 0:01:27 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.6570 (0.6238) acc1: 60.4167 (65.2472) acc5: 100.0000 (100.0000) acc1_neg: 60.938 acc5_neg: 100.000 time: 0.5098 data: 0.3907 max mem: 3531 +Test: [ 80/269] eta: 0:01:30 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.6356 (0.6213) acc1: 63.5417 (65.7118) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.6903 data: 0.5557 max mem: 3531 +Test: [ 90/269] eta: 0:01:29 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.6414 (0.6233) acc1: 63.5417 (65.4504) acc5: 100.0000 (100.0000) acc1_neg: 66.146 acc5_neg: 100.000 time: 0.7050 data: 0.5898 max mem: 3531 +Test: [100/269] eta: 0:01:21 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.6631 (0.6327) acc1: 58.3333 (64.3410) acc5: 100.0000 (100.0000) acc1_neg: 36.979 acc5_neg: 100.000 time: 0.5044 data: 0.3865 max mem: 3531 +Test: [110/269] eta: 0:01:23 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.7178 (0.6428) acc1: 52.3438 (62.9106) acc5: 100.0000 (100.0000) acc1_neg: 47.396 acc5_neg: 100.000 time: 0.6602 data: 0.5435 max mem: 3531 +Test: [120/269] eta: 0:01:20 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.7504 (0.6584) acc1: 46.0938 (61.6004) acc5: 100.0000 (100.0000) acc1_neg: 36.458 acc5_neg: 100.000 time: 0.8312 data: 0.6976 max mem: 3531 +Test: [130/269] eta: 0:01:15 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.8144 (0.6696) acc1: 43.2292 (60.3531) acc5: 100.0000 (100.0000) acc1_neg: 43.490 acc5_neg: 100.000 time: 0.6394 data: 0.4865 max mem: 3531 +Test: [140/269] eta: 0:01:10 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.8147 (0.6858) acc1: 39.0625 (58.5217) acc5: 100.0000 (100.0000) acc1_neg: 40.625 acc5_neg: 100.000 time: 0.5741 data: 0.4416 max mem: 3531 +Test: [150/269] eta: 0:01:06 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.7994 (0.6910) acc1: 40.3646 (57.7849) acc5: 100.0000 (100.0000) acc1_neg: 46.875 acc5_neg: 100.000 time: 0.6547 data: 0.5368 max mem: 3531 +Test: [160/269] eta: 0:00:59 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.7316 (0.6900) acc1: 51.8229 (57.9645) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.5746 data: 0.4653 max mem: 3531 +Test: [170/269] eta: 0:00:55 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.6659 (0.6909) acc1: 59.6354 (57.9785) acc5: 100.0000 (100.0000) acc1_neg: 71.094 acc5_neg: 100.000 time: 0.5448 data: 0.4253 max mem: 3531 +Test: [180/269] eta: 0:00:49 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.6495 (0.6882) acc1: 60.9375 (58.4844) acc5: 100.0000 (100.0000) acc1_neg: 60.938 acc5_neg: 100.000 time: 0.6398 data: 0.5107 max mem: 3531 +Test: [190/269] eta: 0:00:44 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.6349 (0.6893) acc1: 60.9375 (58.4056) acc5: 100.0000 (100.0000) acc1_neg: 47.656 acc5_neg: 100.000 time: 0.6399 data: 0.5143 max mem: 3531 +Test: [200/269] eta: 0:00:38 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.7370 (0.6937) acc1: 50.7812 (57.8993) acc5: 100.0000 (100.0000) acc1_neg: 56.771 acc5_neg: 100.000 time: 0.5902 data: 0.4635 max mem: 3531 +Test: [210/269] eta: 0:00:32 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.7379 (0.6958) acc1: 48.1771 (57.6200) acc5: 100.0000 (100.0000) acc1_neg: 53.385 acc5_neg: 100.000 time: 0.4499 data: 0.3181 max mem: 3532 +Test: [220/269] eta: 0:00:27 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.6067 (0.6943) acc1: 53.3854 (57.8490) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.4343 data: 0.2946 max mem: 3532 +Test: [230/269] eta: 0:00:21 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.6000 (0.6921) acc1: 64.8438 (58.0673) acc5: 100.0000 (100.0000) acc1_neg: 58.594 acc5_neg: 100.000 time: 0.5208 data: 0.3719 max mem: 3532 +Test: [240/269] eta: 0:00:16 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.6597 (0.6938) acc1: 58.5938 (57.8979) acc5: 100.0000 (100.0000) acc1_neg: 61.198 acc5_neg: 100.000 time: 0.5701 data: 0.4275 max mem: 3532 +Test: [250/269] eta: 0:00:10 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.7723 (0.7013) acc1: 48.4375 (57.0053) acc5: 100.0000 (100.0000) acc1_neg: 15.365 acc5_neg: 100.000 time: 0.6198 data: 0.4810 max mem: 3532 +Test: [260/269] eta: 0:00:04 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.9100 (0.7087) acc1: 25.5208 (56.1103) acc5: 100.0000 (100.0000) acc1_neg: 56.250 acc5_neg: 100.000 time: 0.4697 data: 0.3368 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 62.931 acc5_pos: 100.000 loss: 0.7695 (0.7089) acc1: 37.5000 (55.9835) acc5: 100.0000 (100.0000) acc1_neg: 65.879 acc5_neg: 100.000 time: 0.3587 data: 0.2152 max mem: 3532 +Test: Total time: 0:02:25 (0.5408 s / it) +* Overall Acc@1 55.982 Acc@5 100.000 + target == real Acc@1 64.036 Acc@5 100.000 + target == fake Acc@1 54.883 Acc@5 100.000 + loss 0.709 + PR-AUC 0.9225 ROC-AUC 0.6322 EER 40.63% (thr=0.4792) +Accuracy of the network on the 103293 test images: 56.0% +[rank0]:[W725 17:58:22.246358568 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-25T17:58:45Z] eval_done tag=fixed500k_fake_n32_seed1_resampled +[2026-07-25T17:58:46Z] eval_wait tag=fixed500k_real_n16_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed2_resampled +[2026-07-25T17:59:46Z] eval_wait tag=fixed500k_real_n16_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed2_resampled +[2026-07-25T18:00:46Z] eval_wait tag=fixed500k_real_n16_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed2_resampled +[2026-07-25T18:01:46Z] eval_wait tag=fixed500k_real_n16_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed2_resampled +[2026-07-25T18:02:46Z] eval_wait tag=fixed500k_real_n16_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed2_resampled +[2026-07-25T18:03:46Z] eval_wait tag=fixed500k_real_n16_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed2_resampled +[2026-07-25T18:04:46Z] eval_wait tag=fixed500k_real_n16_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed2_resampled +[2026-07-25T18:05:47Z] eval_wait tag=fixed500k_real_n16_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed2_resampled +[2026-07-25T18:06:47Z] eval_wait tag=fixed500k_real_n16_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed2_resampled +[2026-07-25T18:07:47Z] eval_wait tag=fixed500k_real_n16_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed2_resampled +[2026-07-25T18:08:47Z] eval_wait tag=fixed500k_real_n16_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed2_resampled +[2026-07-25T18:09:47Z] eval_wait tag=fixed500k_real_n16_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed2_resampled +[2026-07-25T18:10:47Z] eval_wait tag=fixed500k_real_n16_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed2_resampled +[2026-07-25T18:11:47Z] eval_wait tag=fixed500k_real_n16_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed2_resampled +[2026-07-25T18:12:48Z] eval_wait tag=fixed500k_real_n16_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed2_resampled +[2026-07-25T18:13:48Z] eval_wait tag=fixed500k_real_n16_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed2_resampled +[2026-07-25T18:14:48Z] eval_wait tag=fixed500k_real_n16_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed2_resampled +[2026-07-25T18:15:48Z] eval_wait tag=fixed500k_real_n16_seed2_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed2_resampled +[2026-07-25T18:16:48Z] eval_skip tag=fixed500k_real_n16_seed2_resampled reason=complete_after_lock output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed2_resampled +[2026-07-25T18:16:49Z] eval_skip tag=fixed500k_fake_n32_seed2_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed2_resampled +[2026-07-25T18:16:49Z] eval_wait tag=fixed500k_real_n16_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed3_resampled +[2026-07-25T18:17:49Z] eval_wait tag=fixed500k_real_n16_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed3_resampled +[2026-07-25T18:18:49Z] eval_wait tag=fixed500k_real_n16_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed3_resampled +[2026-07-25T18:19:49Z] eval_wait tag=fixed500k_real_n16_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed3_resampled +[2026-07-25T18:20:49Z] eval_wait tag=fixed500k_real_n16_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed3_resampled +[2026-07-25T18:21:50Z] eval_wait tag=fixed500k_real_n16_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed3_resampled +[2026-07-25T18:22:50Z] eval_wait tag=fixed500k_real_n16_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed3_resampled +[2026-07-25T18:23:50Z] eval_wait tag=fixed500k_real_n16_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed3_resampled +[2026-07-25T18:24:50Z] eval_wait tag=fixed500k_real_n16_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed3_resampled +[2026-07-25T18:25:50Z] eval_wait tag=fixed500k_real_n16_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed3_resampled +[2026-07-25T18:26:50Z] eval_wait tag=fixed500k_real_n16_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed3_resampled +[2026-07-25T18:27:51Z] eval_wait tag=fixed500k_real_n16_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed3_resampled +[2026-07-25T18:28:51Z] eval_wait tag=fixed500k_real_n16_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed3_resampled +[2026-07-25T18:29:51Z] eval_wait tag=fixed500k_real_n16_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed3_resampled +[2026-07-25T18:30:51Z] eval_wait tag=fixed500k_real_n16_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed3_resampled +[2026-07-25T18:31:51Z] eval_wait tag=fixed500k_real_n16_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed3_resampled +[2026-07-25T18:32:51Z] eval_wait tag=fixed500k_real_n16_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed3_resampled +[2026-07-25T18:33:52Z] eval_wait tag=fixed500k_real_n16_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed3_resampled +[2026-07-25T18:34:52Z] eval_wait tag=fixed500k_real_n16_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed3_resampled +[2026-07-25T18:35:52Z] eval_wait tag=fixed500k_real_n16_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed3_resampled +[2026-07-25T18:36:52Z] eval_wait tag=fixed500k_real_n16_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed3_resampled +[2026-07-25T18:37:52Z] eval_skip tag=fixed500k_real_n16_seed3_resampled reason=complete_after_lock output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed3_resampled +[2026-07-25T18:37:52Z] eval_wait tag=fixed500k_fake_n32_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed3_resampled +[2026-07-25T18:38:52Z] eval_wait tag=fixed500k_fake_n32_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed3_resampled +[2026-07-25T18:39:53Z] eval_wait tag=fixed500k_fake_n32_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed3_resampled +[2026-07-25T18:40:53Z] eval_wait tag=fixed500k_fake_n32_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed3_resampled +[2026-07-25T18:41:53Z] eval_wait tag=fixed500k_fake_n32_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed3_resampled +[2026-07-25T18:42:53Z] eval_wait tag=fixed500k_fake_n32_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed3_resampled +[2026-07-25T18:43:53Z] eval_wait tag=fixed500k_fake_n32_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed3_resampled +[2026-07-25T18:44:53Z] eval_wait tag=fixed500k_fake_n32_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed3_resampled +[2026-07-25T18:45:54Z] eval_wait tag=fixed500k_fake_n32_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed3_resampled +[2026-07-25T18:46:54Z] eval_wait tag=fixed500k_fake_n32_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed3_resampled +[2026-07-25T18:47:54Z] eval_wait tag=fixed500k_fake_n32_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed3_resampled +[2026-07-25T18:48:54Z] eval_wait tag=fixed500k_fake_n32_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed3_resampled +[2026-07-25T18:49:54Z] eval_wait tag=fixed500k_fake_n32_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed3_resampled +[2026-07-25T18:50:54Z] eval_wait tag=fixed500k_fake_n32_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed3_resampled +[2026-07-25T18:51:55Z] eval_wait tag=fixed500k_fake_n32_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed3_resampled +[2026-07-25T18:52:55Z] eval_wait tag=fixed500k_fake_n32_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed3_resampled +[2026-07-25T18:53:55Z] eval_wait tag=fixed500k_fake_n32_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed3_resampled +[2026-07-25T18:54:55Z] eval_wait tag=fixed500k_fake_n32_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed3_resampled +[2026-07-25T18:55:55Z] eval_wait tag=fixed500k_fake_n32_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed3_resampled +[2026-07-25T18:56:55Z] eval_wait tag=fixed500k_fake_n32_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed3_resampled +[2026-07-25T18:57:56Z] eval_wait tag=fixed500k_fake_n32_seed3_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed3_resampled +[2026-07-25T18:58:56Z] eval_skip tag=fixed500k_fake_n32_seed3_resampled reason=complete_after_lock output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed3_resampled +[2026-07-25T18:58:56Z] eval_wait tag=fixed500k_real_n16_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed4_resampled +[2026-07-25T18:59:56Z] eval_wait tag=fixed500k_real_n16_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed4_resampled +[2026-07-25T19:00:56Z] eval_wait tag=fixed500k_real_n16_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed4_resampled +[2026-07-25T19:01:56Z] eval_wait tag=fixed500k_real_n16_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed4_resampled +[2026-07-25T19:02:57Z] eval_wait tag=fixed500k_real_n16_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed4_resampled +[2026-07-25T19:03:57Z] eval_wait tag=fixed500k_real_n16_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed4_resampled +[2026-07-25T19:04:57Z] eval_wait tag=fixed500k_real_n16_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed4_resampled +[2026-07-25T19:05:57Z] eval_wait tag=fixed500k_real_n16_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed4_resampled +[2026-07-25T19:06:57Z] eval_wait tag=fixed500k_real_n16_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed4_resampled +[2026-07-25T19:07:57Z] eval_wait tag=fixed500k_real_n16_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed4_resampled +[2026-07-25T19:08:58Z] eval_wait tag=fixed500k_real_n16_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed4_resampled +[2026-07-25T19:09:58Z] eval_wait tag=fixed500k_real_n16_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed4_resampled +[2026-07-25T19:10:58Z] eval_wait tag=fixed500k_real_n16_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed4_resampled +[2026-07-25T19:11:58Z] eval_wait tag=fixed500k_real_n16_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed4_resampled +[2026-07-25T19:12:58Z] eval_wait tag=fixed500k_real_n16_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed4_resampled +[2026-07-25T19:13:58Z] eval_wait tag=fixed500k_real_n16_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed4_resampled +[2026-07-25T19:14:59Z] eval_wait tag=fixed500k_real_n16_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed4_resampled +[2026-07-25T19:15:59Z] eval_wait tag=fixed500k_real_n16_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed4_resampled +[2026-07-25T19:16:59Z] eval_wait tag=fixed500k_real_n16_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed4_resampled +[2026-07-25T19:17:59Z] eval_wait tag=fixed500k_real_n16_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed4_resampled +[2026-07-25T19:18:59Z] eval_wait tag=fixed500k_real_n16_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed4_resampled +[2026-07-25T19:19:59Z] eval_skip tag=fixed500k_real_n16_seed4_resampled reason=complete_after_wait output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n16_seed4_resampled +[2026-07-25T19:20:00Z] eval_wait tag=fixed500k_fake_n32_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed4_resampled +[2026-07-25T19:21:00Z] eval_wait tag=fixed500k_fake_n32_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed4_resampled +[2026-07-25T19:22:00Z] eval_wait tag=fixed500k_fake_n32_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed4_resampled +[2026-07-25T19:23:00Z] eval_wait tag=fixed500k_fake_n32_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed4_resampled +[2026-07-25T19:24:00Z] eval_wait tag=fixed500k_fake_n32_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed4_resampled +[2026-07-25T19:25:00Z] eval_wait tag=fixed500k_fake_n32_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed4_resampled +[2026-07-25T19:26:01Z] eval_wait tag=fixed500k_fake_n32_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed4_resampled +[2026-07-25T19:27:01Z] eval_wait tag=fixed500k_fake_n32_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed4_resampled +[2026-07-25T19:28:01Z] eval_wait tag=fixed500k_fake_n32_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed4_resampled +[2026-07-25T19:29:01Z] eval_wait tag=fixed500k_fake_n32_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed4_resampled +[2026-07-25T19:30:01Z] eval_wait tag=fixed500k_fake_n32_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed4_resampled +[2026-07-25T19:31:01Z] eval_wait tag=fixed500k_fake_n32_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed4_resampled +[2026-07-25T19:32:02Z] eval_wait tag=fixed500k_fake_n32_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed4_resampled +[2026-07-25T19:33:02Z] eval_wait tag=fixed500k_fake_n32_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed4_resampled +[2026-07-25T19:34:02Z] eval_skip tag=fixed500k_fake_n32_seed4_resampled reason=complete_after_lock output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n32_seed4_resampled +eval_queue_done_utc=2026-07-25T19:34:02Z diff --git a/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_host_done_20260725_112559/eval_queue_n111-046-081.pid b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_host_done_20260725_112559/eval_queue_n111-046-081.pid new file mode 100644 index 0000000000000000000000000000000000000000..fa8d69d7875090b458e29fff91963b15bb8f5662 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_host_done_20260725_112559/eval_queue_n111-046-081.pid @@ -0,0 +1 @@ +2557701 diff --git a/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_host_done_20260725_112559/eval_queue_n111-046-081.submitted b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_host_done_20260725_112559/eval_queue_n111-046-081.submitted new file mode 100644 index 0000000000000000000000000000000000000000..d17960422dfc154399f7fc2e87189565ff8d1a26 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_host_done_20260725_112559/eval_queue_n111-046-081.submitted @@ -0,0 +1 @@ +2026-07-25T16:41:49Z diff --git a/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_host_done_20260725_112559/eval_queue_n116-040-168.log b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_host_done_20260725_112559/eval_queue_n116-040-168.log new file mode 100644 index 0000000000000000000000000000000000000000..e9b6e308a1ca6819ccc62fb5248a18189fd1469c --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_host_done_20260725_112559/eval_queue_n116-040-168.log @@ -0,0 +1,5164 @@ +eval_queue_started_utc=2026-07-25T16:48:50Z +hostname=n116-040-168 +train_tag=20260725_112559 +eval_tag=20260725_112559 +real_level=5 +fake_level=11 +eval_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559 +batch_per_gpu=256 +[2026-07-25T16:48:51Z] eval_start tag=fixed500k_real_n5_seed1_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed1_resampled_n116-040-168_20260725_112559/best_checkpoint.pth +started_utc=2026-07-25T16:48:51Z +hostname=n116-040-168 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed1_resampled_n116-040-168_20260725_112559/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed1_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0725 16:49:03.952000 2554956 torch/distributed/run.py:766] +W0725 16:49:03.952000 2554956 torch/distributed/run.py:766] ***************************************** +W0725 16:49:03.952000 2554956 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 16:49:03.952000 2554956 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 16:49:33.154054249 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 16:49:34.211996893 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 16:49:34.354823527 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 16:49:34.753891997 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 16:49:34.765112442 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +| distributed init (rank 1): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 16:49:34.918108386 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank2]:[W725 16:49:34.918113303 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W725 16:49:34.918124923 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed1_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed1_resampled_n116-040-168_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:28:33 acc1_pos: 94.531 acc5_pos: 100.000 loss: 0.2625 (0.2625) acc1: 94.5312 (94.5312) acc5: 100.0000 (100.0000) time: 7.3225 data: 5.1065 max mem: 3531 +Test: [ 10/234] eta: 0:02:48 acc1_pos: 65.104 acc5_pos: 100.000 loss: 0.4067 (0.4107) acc1: 82.5521 (82.9072) acc5: 100.0000 (100.0000) time: 0.7500 data: 0.4646 max mem: 3531 +Test: [ 20/234] eta: 0:01:34 acc1_pos: 45.312 acc5_pos: 100.000 loss: 0.6313 (0.7021) acc1: 60.6771 (64.6205) acc5: 100.0000 (100.0000) time: 0.0963 data: 0.0004 max mem: 3531 +Test: [ 30/234] eta: 0:01:18 acc1_pos: 71.094 acc5_pos: 100.000 loss: 1.0838 (0.8310) acc1: 40.8854 (56.8968) acc5: 100.0000 (100.0000) time: 0.1849 data: 0.0889 max mem: 3531 +Test: [ 40/234] eta: 0:01:08 acc1_pos: 17.708 acc5_pos: 100.000 loss: 1.0846 (0.9202) acc1: 39.5833 (51.8356) acc5: 100.0000 (100.0000) time: 0.2599 data: 0.1672 max mem: 3531 +Test: [ 50/234] eta: 0:01:00 acc1_pos: 15.365 acc5_pos: 100.000 loss: 1.5883 (1.0608) acc1: 19.2708 (45.4504) acc5: 100.0000 (100.0000) time: 0.2349 data: 0.1426 max mem: 3531 +Test: [ 60/234] eta: 0:00:51 acc1_pos: 22.069 acc5_pos: 100.000 loss: 0.6347 (0.9133) acc1: 25.2604 (53.8806) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1856 data: 0.0934 max mem: 3531 +Test: [ 70/234] eta: 0:00:45 acc1_pos: 22.069 acc5_pos: 100.000 loss: 0.1162 (0.8087) acc1: 99.4792 (59.8665) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1549 data: 0.0625 max mem: 3531 +Test: [ 80/234] eta: 0:00:41 acc1_pos: 22.069 acc5_pos: 100.000 loss: 0.1950 (0.7384) acc1: 94.5312 (63.9082) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1892 data: 0.0954 max mem: 3531 +Test: [ 90/234] eta: 0:00:37 acc1_pos: 22.069 acc5_pos: 100.000 loss: 0.1702 (0.6697) acc1: 96.0938 (67.7742) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1901 data: 0.0965 max mem: 3531 +Test: [100/234] eta: 0:00:33 acc1_pos: 22.069 acc5_pos: 100.000 loss: 0.1702 (0.6311) acc1: 95.3125 (69.9257) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1599 data: 0.0671 max mem: 3531 +Test: [110/234] eta: 0:00:30 acc1_pos: 22.069 acc5_pos: 100.000 loss: 0.2739 (0.5987) acc1: 90.1042 (71.7084) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1697 data: 0.0764 max mem: 3531 +Test: [120/234] eta: 0:00:27 acc1_pos: 22.069 acc5_pos: 100.000 loss: 0.2988 (0.5874) acc1: 87.5000 (72.1569) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1899 data: 0.0944 max mem: 3531 +Test: [130/234] eta: 0:00:24 acc1_pos: 22.069 acc5_pos: 100.000 loss: 0.4448 (0.5818) acc1: 79.1667 (72.3322) acc5: 100.0000 (100.0000) acc1_neg: 69.271 acc5_neg: 100.000 time: 0.2102 data: 0.1144 max mem: 3531 +Test: [140/234] eta: 0:00:21 acc1_pos: 22.069 acc5_pos: 100.000 loss: 0.3688 (0.5588) acc1: 83.5938 (73.6444) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1949 data: 0.1012 max mem: 3531 +Test: [150/234] eta: 0:00:18 acc1_pos: 22.069 acc5_pos: 100.000 loss: 0.2286 (0.5379) acc1: 90.8854 (74.8551) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1346 data: 0.0398 max mem: 3531 +Test: [160/234] eta: 0:00:16 acc1_pos: 22.069 acc5_pos: 100.000 loss: 0.3439 (0.5417) acc1: 87.2396 (74.4905) acc5: 100.0000 (100.0000) acc1_neg: 64.062 acc5_neg: 100.000 time: 0.1399 data: 0.0458 max mem: 3531 +Test: [170/234] eta: 0:00:14 acc1_pos: 22.069 acc5_pos: 100.000 loss: 0.6315 (0.5520) acc1: 65.6250 (73.7040) acc5: 100.0000 (100.0000) acc1_neg: 49.219 acc5_neg: 100.000 time: 0.1949 data: 0.1003 max mem: 3531 +Test: [180/234] eta: 0:00:11 acc1_pos: 22.069 acc5_pos: 100.000 loss: 0.5534 (0.5382) acc1: 69.0104 (74.4964) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1999 data: 0.1051 max mem: 3531 +Test: [190/234] eta: 0:00:09 acc1_pos: 22.069 acc5_pos: 100.000 loss: 0.3332 (0.5281) acc1: 87.2396 (75.1227) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.1876 data: 0.0916 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 22.069 acc5_pos: 100.000 loss: 0.3385 (0.5123) acc1: 87.5000 (76.0546) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1675 data: 0.0719 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 22.069 acc5_pos: 100.000 loss: 0.1118 (0.4933) acc1: 99.7396 (77.1808) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1443 data: 0.0526 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 22.069 acc5_pos: 100.000 loss: 0.1117 (0.4762) acc1: 99.7396 (78.1969) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1353 data: 0.0450 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 22.069 acc5_pos: 100.000 loss: 0.1371 (0.4636) acc1: 96.3542 (78.9243) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1354 data: 0.0459 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 22.069 acc5_pos: 100.000 loss: 0.1743 (0.4601) acc1: 95.5729 (79.1004) acc5: 100.0000 (100.0000) acc1_neg: 91.829 acc5_neg: 100.000 time: 0.1317 data: 0.0412 max mem: 3532 +Test: Total time: 0:00:47 (0.2038 s / it) +* Overall Acc@1 79.100 Acc@5 100.000 + target == real Acc@1 45.279 Acc@5 100.000 + target == fake Acc@1 88.633 Acc@5 100.000 + loss 0.460 + PR-AUC 0.9036 ROC-AUC 0.7681 EER 28.87% (thr=0.7935) +Accuracy of the network on the 89729 test images: 79.1% +[rank0]:[W725 16:50:31.817015491 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 16:50:51.158000 2555515 torch/distributed/run.py:766] +W0725 16:50:51.158000 2555515 torch/distributed/run.py:766] ***************************************** +W0725 16:50:51.158000 2555515 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 16:50:51.158000 2555515 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 16:51:20.985441483 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 16:51:21.260462588 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 16:51:21.447479909 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 16:51:21.620364985 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 16:51:21.670217463 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 16:51:21.747045573 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 16:51:21.784164907 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W725 16:51:21.787491631 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed1_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed1_resampled_n116-040-168_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 5:02:49 acc1_pos: 52.865 acc5_pos: 100.000 loss: 0.7543 (0.7543) acc1: 52.8646 (52.8646) acc5: 100.0000 (100.0000) time: 7.6120 data: 6.0067 max mem: 3531 +Test: [ 10/2387] eta: 0:30:39 acc1_pos: 51.302 acc5_pos: 100.000 loss: 0.8711 (0.8367) acc1: 52.8646 (52.7699) acc5: 100.0000 (100.0000) time: 0.7737 data: 0.5464 max mem: 3531 +Test: [ 20/2387] eta: 0:17:46 acc1_pos: 69.792 acc5_pos: 100.000 loss: 0.7706 (0.8084) acc1: 57.2917 (55.0595) acc5: 100.0000 (100.0000) time: 0.0926 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:14:57 acc1_pos: 12.760 acc5_pos: 100.000 loss: 0.7398 (0.8352) acc1: 55.4688 (52.8562) acc5: 100.0000 (100.0000) time: 0.1650 data: 0.0683 max mem: 3531 +Test: [ 40/2387] eta: 0:13:20 acc1_pos: 44.271 acc5_pos: 100.000 loss: 0.8901 (0.8718) acc1: 41.9271 (49.6443) acc5: 100.0000 (100.0000) time: 0.2260 data: 0.1306 max mem: 3531 +Test: [ 50/2387] eta: 0:12:34 acc1_pos: 57.031 acc5_pos: 100.000 loss: 0.9119 (0.8744) acc1: 44.2708 (49.5966) acc5: 100.0000 (100.0000) time: 0.2326 data: 0.1364 max mem: 3531 +Test: [ 60/2387] eta: 0:11:56 acc1_pos: 56.250 acc5_pos: 100.000 loss: 0.9463 (0.8985) acc1: 46.8750 (48.8431) acc5: 100.0000 (100.0000) time: 0.2396 data: 0.1436 max mem: 3531 +Test: [ 70/2387] eta: 0:11:10 acc1_pos: 50.000 acc5_pos: 100.000 loss: 1.0385 (0.9128) acc1: 44.0104 (48.2761) acc5: 100.0000 (100.0000) time: 0.2049 data: 0.1136 max mem: 3531 +Test: [ 80/2387] eta: 0:10:45 acc1_pos: 37.500 acc5_pos: 100.000 loss: 0.9076 (0.9065) acc1: 45.8333 (48.3796) acc5: 100.0000 (100.0000) time: 0.1944 data: 0.1040 max mem: 3531 +Test: [ 90/2387] eta: 0:10:32 acc1_pos: 42.969 acc5_pos: 100.000 loss: 0.8300 (0.9009) acc1: 47.9167 (48.6664) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1319 max mem: 3531 +Test: [ 100/2387] eta: 0:10:19 acc1_pos: 62.500 acc5_pos: 100.000 loss: 0.8244 (0.9004) acc1: 48.4375 (48.7546) acc5: 100.0000 (100.0000) time: 0.2350 data: 0.1398 max mem: 3531 +Test: [ 110/2387] eta: 0:10:10 acc1_pos: 64.844 acc5_pos: 100.000 loss: 0.7800 (0.8859) acc1: 55.4688 (49.7255) acc5: 100.0000 (100.0000) time: 0.2354 data: 0.1426 max mem: 3531 +Test: [ 120/2387] eta: 0:10:02 acc1_pos: 70.312 acc5_pos: 100.000 loss: 0.7186 (0.8783) acc1: 60.4167 (50.3702) acc5: 100.0000 (100.0000) time: 0.2402 data: 0.1482 max mem: 3531 +Test: [ 130/2387] eta: 0:09:58 acc1_pos: 62.760 acc5_pos: 100.000 loss: 0.7623 (0.8674) acc1: 58.5938 (50.9820) acc5: 100.0000 (100.0000) time: 0.2499 data: 0.1564 max mem: 3531 +Test: [ 140/2387] eta: 0:09:47 acc1_pos: 70.052 acc5_pos: 100.000 loss: 0.7578 (0.8572) acc1: 57.8125 (51.4628) acc5: 100.0000 (100.0000) time: 0.2354 data: 0.1413 max mem: 3531 +Test: [ 150/2387] eta: 0:09:44 acc1_pos: 39.583 acc5_pos: 100.000 loss: 0.7063 (0.8507) acc1: 57.8125 (51.8747) acc5: 100.0000 (100.0000) time: 0.2345 data: 0.1403 max mem: 3531 +Test: [ 160/2387] eta: 0:09:29 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.6517 (0.8283) acc1: 61.9792 (53.3337) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2139 data: 0.1184 max mem: 3531 +Test: [ 170/2387] eta: 0:09:14 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3031 (0.7977) acc1: 90.3646 (55.5327) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1660 data: 0.0704 max mem: 3531 +Test: [ 180/2387] eta: 0:09:07 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3031 (0.7706) acc1: 90.3646 (57.4730) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1857 data: 0.0923 max mem: 3531 +Test: [ 190/2387] eta: 0:09:00 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3070 (0.7462) acc1: 90.3646 (59.2318) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.2124 data: 0.1163 max mem: 3531 +Test: [ 200/2387] eta: 0:08:50 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3027 (0.7241) acc1: 90.6250 (60.8041) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1945 data: 0.0939 max mem: 3531 +Test: [ 210/2387] eta: 0:08:37 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3050 (0.7046) acc1: 90.3646 (62.2087) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.1583 data: 0.0607 max mem: 3532 +Test: [ 220/2387] eta: 0:08:28 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3141 (0.6867) acc1: 90.3646 (63.4945) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1556 data: 0.0636 max mem: 3532 +Test: [ 230/2387] eta: 0:08:22 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3124 (0.6706) acc1: 90.6250 (64.6634) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1808 data: 0.0889 max mem: 3532 +Test: [ 240/2387] eta: 0:08:16 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3237 (0.6577) acc1: 88.5417 (65.5839) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.1945 data: 0.1017 max mem: 3532 +Test: [ 250/2387] eta: 0:08:09 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3017 (0.6381) acc1: 88.2812 (66.8358) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1871 data: 0.0957 max mem: 3532 +Test: [ 260/2387] eta: 0:08:04 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1421 (0.6186) acc1: 99.7396 (68.0795) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1818 data: 0.0899 max mem: 3532 +Test: [ 270/2387] eta: 0:07:58 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1216 (0.6004) acc1: 100.0000 (69.2439) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1864 data: 0.0944 max mem: 3532 +Test: [ 280/2387] eta: 0:07:53 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1274 (0.5835) acc1: 99.7396 (70.3097) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1875 data: 0.0956 max mem: 3532 +Test: [ 290/2387] eta: 0:07:47 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1276 (0.5681) acc1: 99.7396 (71.2951) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1836 data: 0.0901 max mem: 3532 +Test: [ 300/2387] eta: 0:07:43 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1237 (0.5529) acc1: 100.0000 (72.2462) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1847 data: 0.0897 max mem: 3532 +Test: [ 310/2387] eta: 0:07:38 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1127 (0.5396) acc1: 100.0000 (73.1185) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1847 data: 0.0908 max mem: 3532 +Test: [ 320/2387] eta: 0:07:34 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1734 (0.5283) acc1: 99.4792 (73.9348) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1870 data: 0.0956 max mem: 3532 +Test: [ 330/2387] eta: 0:07:30 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1766 (0.5177) acc1: 99.4792 (74.7065) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1962 data: 0.1050 max mem: 3532 +Test: [ 340/2387] eta: 0:07:27 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1766 (0.5077) acc1: 99.4792 (75.4330) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1978 data: 0.1043 max mem: 3532 +Test: [ 350/2387] eta: 0:07:23 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1672 (0.4966) acc1: 99.7396 (76.1292) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1940 data: 0.0990 max mem: 3532 +Test: [ 360/2387] eta: 0:07:21 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1371 (0.4898) acc1: 99.2188 (76.6209) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2053 data: 0.1120 max mem: 3532 +Test: [ 370/2387] eta: 0:07:16 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2672 (0.4838) acc1: 93.2292 (77.0588) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1914 data: 0.0998 max mem: 3532 +Test: [ 380/2387] eta: 0:07:12 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2690 (0.4782) acc1: 92.4479 (77.4777) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1745 data: 0.0805 max mem: 3532 +Test: [ 390/2387] eta: 0:07:08 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2641 (0.4729) acc1: 92.7083 (77.8672) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1849 data: 0.0892 max mem: 3532 +Test: [ 400/2387] eta: 0:07:04 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2651 (0.4678) acc1: 92.9688 (78.2445) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1799 data: 0.0827 max mem: 3532 +Test: [ 410/2387] eta: 0:07:00 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2626 (0.4628) acc1: 92.9688 (78.6205) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1747 data: 0.0793 max mem: 3532 +Test: [ 420/2387] eta: 0:06:58 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2642 (0.4582) acc1: 92.9688 (78.9496) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1928 data: 0.0997 max mem: 3532 +Test: [ 430/2387] eta: 0:06:55 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2821 (0.4545) acc1: 91.4062 (79.2259) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1990 data: 0.1059 max mem: 3532 +Test: [ 440/2387] eta: 0:06:52 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2931 (0.4489) acc1: 91.4062 (79.6072) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1929 data: 0.0985 max mem: 3533 +Test: [ 450/2387] eta: 0:06:49 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1311 (0.4418) acc1: 100.0000 (80.0582) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1982 data: 0.1005 max mem: 3533 +Test: [ 460/2387] eta: 0:06:47 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1298 (0.4351) acc1: 100.0000 (80.4897) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2007 data: 0.1059 max mem: 3533 +Test: [ 470/2387] eta: 0:06:44 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1297 (0.4286) acc1: 100.0000 (80.9039) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2015 data: 0.1106 max mem: 3533 +Test: [ 480/2387] eta: 0:06:41 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1305 (0.4224) acc1: 100.0000 (81.2998) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1896 data: 0.0979 max mem: 3533 +Test: [ 490/2387] eta: 0:06:37 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1305 (0.4165) acc1: 100.0000 (81.6791) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1766 data: 0.0820 max mem: 3533 +Test: [ 500/2387] eta: 0:06:34 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1304 (0.4108) acc1: 100.0000 (82.0437) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1797 data: 0.0828 max mem: 3533 +Test: [ 510/2387] eta: 0:06:31 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1306 (0.4059) acc1: 100.0000 (82.3910) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1811 data: 0.0842 max mem: 3533 +Test: [ 520/2387] eta: 0:06:27 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2307 (0.4028) acc1: 99.2188 (82.7020) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1683 data: 0.0728 max mem: 3533 +Test: [ 530/2387] eta: 0:06:23 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2592 (0.4003) acc1: 97.9167 (82.9734) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1567 data: 0.0632 max mem: 3533 +Test: [ 540/2387] eta: 0:06:20 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2615 (0.3977) acc1: 97.6562 (83.2486) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1547 data: 0.0635 max mem: 3533 +Test: [ 550/2387] eta: 0:06:16 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2586 (0.3952) acc1: 97.6562 (83.5044) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1551 data: 0.0620 max mem: 3533 +Test: [ 560/2387] eta: 0:06:12 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2626 (0.3929) acc1: 97.1354 (83.7507) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1598 data: 0.0663 max mem: 3533 +Test: [ 570/2387] eta: 0:06:09 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2636 (0.3905) acc1: 97.3958 (83.9905) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1657 data: 0.0740 max mem: 3533 +Test: [ 580/2387] eta: 0:06:06 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2634 (0.3883) acc1: 97.3958 (84.2217) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1645 data: 0.0730 max mem: 3533 +Test: [ 590/2387] eta: 0:06:03 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2627 (0.3862) acc1: 97.3958 (84.4490) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1630 data: 0.0690 max mem: 3533 +Test: [ 600/2387] eta: 0:06:00 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2544 (0.3838) acc1: 98.4375 (84.6870) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1648 data: 0.0696 max mem: 3533 +Test: [ 610/2387] eta: 0:05:57 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2604 (0.3826) acc1: 94.5312 (84.7906) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1789 data: 0.0836 max mem: 3533 +Test: [ 620/2387] eta: 0:05:54 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3136 (0.3816) acc1: 90.3646 (84.8824) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1823 data: 0.0884 max mem: 3533 +Test: [ 630/2387] eta: 0:05:52 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3190 (0.3805) acc1: 90.1042 (84.9821) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1755 data: 0.0843 max mem: 3533 +Test: [ 640/2387] eta: 0:05:48 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3122 (0.3795) acc1: 90.8854 (85.0705) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1597 data: 0.0686 max mem: 3533 +Test: [ 650/2387] eta: 0:05:46 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3132 (0.3785) acc1: 90.8854 (85.1635) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1597 data: 0.0691 max mem: 3533 +Test: [ 660/2387] eta: 0:05:43 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3155 (0.3774) acc1: 91.1458 (85.2587) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1782 data: 0.0874 max mem: 3533 +Test: [ 670/2387] eta: 0:05:40 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3137 (0.3765) acc1: 91.4062 (85.3460) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1769 data: 0.0857 max mem: 3534 +Test: [ 680/2387] eta: 0:05:39 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3045 (0.3752) acc1: 91.6667 (85.4438) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1916 data: 0.1000 max mem: 3534 +Test: [ 690/2387] eta: 0:05:37 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2752 (0.3728) acc1: 92.7083 (85.5897) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2135 data: 0.1223 max mem: 3534 +Test: [ 700/2387] eta: 0:05:35 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1887 (0.3702) acc1: 96.3542 (85.7399) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2035 data: 0.1130 max mem: 3534 +Test: [ 710/2387] eta: 0:05:33 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1851 (0.3676) acc1: 96.3542 (85.8968) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1911 data: 0.0998 max mem: 3534 +Test: [ 720/2387] eta: 0:05:31 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1861 (0.3651) acc1: 96.8750 (86.0408) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2038 data: 0.1118 max mem: 3534 +Test: [ 730/2387] eta: 0:05:30 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1921 (0.3627) acc1: 95.5729 (86.1741) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2155 data: 0.1243 max mem: 3534 +Test: [ 740/2387] eta: 0:05:28 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1891 (0.3603) acc1: 96.6146 (86.3192) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2099 data: 0.1170 max mem: 3534 +Test: [ 750/2387] eta: 0:05:26 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1853 (0.3580) acc1: 97.1354 (86.4590) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2037 data: 0.1083 max mem: 3534 +Test: [ 760/2387] eta: 0:05:24 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1821 (0.3557) acc1: 96.6146 (86.5877) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2012 data: 0.1078 max mem: 3534 +Test: [ 770/2387] eta: 0:05:23 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1821 (0.3536) acc1: 96.3542 (86.7103) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2287 data: 0.1352 max mem: 3534 +Test: [ 780/2387] eta: 0:05:21 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1843 (0.3517) acc1: 95.8333 (86.8288) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2245 data: 0.1284 max mem: 3534 +Test: [ 790/2387] eta: 0:05:18 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1833 (0.3494) acc1: 97.3958 (86.9611) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1833 data: 0.0867 max mem: 3534 +Test: [ 800/2387] eta: 0:05:17 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1834 (0.3479) acc1: 97.3958 (87.0669) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1946 data: 0.1000 max mem: 3534 +Test: [ 810/2387] eta: 0:05:15 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2026 (0.3461) acc1: 97.1354 (87.1946) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2086 data: 0.1167 max mem: 3534 +Test: [ 820/2387] eta: 0:05:13 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2066 (0.3445) acc1: 96.8750 (87.3081) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2007 data: 0.1096 max mem: 3534 +Test: [ 830/2387] eta: 0:05:11 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2216 (0.3432) acc1: 95.3125 (87.3891) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1999 data: 0.1087 max mem: 3534 +Test: [ 840/2387] eta: 0:05:09 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3146 (0.3453) acc1: 88.5417 (87.2340) acc5: 100.0000 (100.0000) acc1_neg: 58.594 acc5_neg: 100.000 time: 0.2002 data: 0.1089 max mem: 3534 +Test: [ 850/2387] eta: 0:05:07 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.4960 (0.3469) acc1: 73.4375 (87.0951) acc5: 100.0000 (100.0000) acc1_neg: 68.490 acc5_neg: 100.000 time: 0.2024 data: 0.1094 max mem: 3534 +Test: [ 860/2387] eta: 0:05:05 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.4920 (0.3484) acc1: 72.1354 (86.9758) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.1961 data: 0.0992 max mem: 3534 +Test: [ 870/2387] eta: 0:05:03 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.4442 (0.3492) acc1: 78.9062 (86.9080) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.1908 data: 0.0955 max mem: 3534 +Test: [ 880/2387] eta: 0:05:00 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3900 (0.3505) acc1: 81.2500 (86.8181) acc5: 100.0000 (100.0000) acc1_neg: 70.312 acc5_neg: 100.000 time: 0.1893 data: 0.0980 max mem: 3534 +Test: [ 890/2387] eta: 0:04:58 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3545 (0.3498) acc1: 86.7188 (86.8757) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1840 data: 0.0922 max mem: 3534 +Test: [ 900/2387] eta: 0:04:56 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2084 (0.3479) acc1: 97.1354 (86.9829) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1852 data: 0.0935 max mem: 3535 +Test: [ 910/2387] eta: 0:04:54 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1426 (0.3456) acc1: 98.6979 (87.1135) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1866 data: 0.0951 max mem: 3535 +Test: [ 920/2387] eta: 0:04:51 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1414 (0.3434) acc1: 99.2188 (87.2458) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1882 data: 0.0962 max mem: 3535 +Test: [ 930/2387] eta: 0:04:49 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1401 (0.3412) acc1: 99.2188 (87.3725) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1834 data: 0.0920 max mem: 3535 +Test: [ 940/2387] eta: 0:04:47 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1401 (0.3391) acc1: 99.2188 (87.4967) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1890 data: 0.0976 max mem: 3535 +Test: [ 950/2387] eta: 0:04:45 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1401 (0.3370) acc1: 99.2188 (87.6186) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2008 data: 0.1098 max mem: 3535 +Test: [ 960/2387] eta: 0:04:43 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1398 (0.3349) acc1: 99.2188 (87.7387) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1961 data: 0.1037 max mem: 3535 +Test: [ 970/2387] eta: 0:04:41 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1404 (0.3329) acc1: 99.2188 (87.8575) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2004 data: 0.1034 max mem: 3535 +Test: [ 980/2387] eta: 0:04:39 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1415 (0.3316) acc1: 98.9583 (87.9407) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1954 data: 0.1009 max mem: 3535 +Test: [ 990/2387] eta: 0:04:37 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2536 (0.3321) acc1: 91.1458 (87.9291) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1883 data: 0.0959 max mem: 3535 +Test: [1000/2387] eta: 0:04:35 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3546 (0.3324) acc1: 87.2396 (87.9176) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1796 data: 0.0869 max mem: 3535 +Test: [1010/2387] eta: 0:04:32 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3401 (0.3323) acc1: 88.2812 (87.9335) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.1619 data: 0.0709 max mem: 3535 +Test: [1020/2387] eta: 0:04:30 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3216 (0.3324) acc1: 88.8021 (87.9321) acc5: 100.0000 (100.0000) acc1_neg: 66.927 acc5_neg: 100.000 time: 0.1599 data: 0.0670 max mem: 3535 +Test: [1030/2387] eta: 0:04:27 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3311 (0.3330) acc1: 86.7188 (87.8963) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1596 data: 0.0667 max mem: 3535 +Test: [1040/2387] eta: 0:04:25 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3289 (0.3326) acc1: 88.5417 (87.9223) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.1587 data: 0.0674 max mem: 3535 +Test: [1050/2387] eta: 0:04:22 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3289 (0.3334) acc1: 88.5417 (87.8702) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1561 data: 0.0651 max mem: 3535 +Test: [1060/2387] eta: 0:04:20 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3802 (0.3338) acc1: 84.8958 (87.8586) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1555 data: 0.0642 max mem: 3535 +Test: [1070/2387] eta: 0:04:17 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3281 (0.3340) acc1: 86.4583 (87.8458) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.1622 data: 0.0680 max mem: 3535 +Test: [1080/2387] eta: 0:04:15 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3285 (0.3338) acc1: 86.4583 (87.8546) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1590 data: 0.0649 max mem: 3535 +Test: [1090/2387] eta: 0:04:12 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3285 (0.3339) acc1: 89.0625 (87.8571) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.1458 data: 0.0528 max mem: 3535 +Test: [1100/2387] eta: 0:04:10 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2829 (0.3336) acc1: 90.6250 (87.8756) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1524 data: 0.0577 max mem: 3535 +Test: [1110/2387] eta: 0:04:08 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2829 (0.3343) acc1: 91.1458 (87.8326) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.1647 data: 0.0709 max mem: 3535 +Test: [1120/2387] eta: 0:04:05 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3340 (0.3344) acc1: 89.0625 (87.8345) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1574 data: 0.0659 max mem: 3536 +Test: [1130/2387] eta: 0:04:03 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3035 (0.3345) acc1: 89.5833 (87.8452) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1590 data: 0.0659 max mem: 3536 +Test: [1140/2387] eta: 0:04:01 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3475 (0.3348) acc1: 86.7188 (87.8273) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.1660 data: 0.0725 max mem: 3536 +Test: [1150/2387] eta: 0:03:58 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3260 (0.3344) acc1: 90.3646 (87.8616) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1608 data: 0.0693 max mem: 3536 +Test: [1160/2387] eta: 0:03:56 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2381 (0.3335) acc1: 95.8333 (87.9262) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1523 data: 0.0602 max mem: 3536 +Test: [1170/2387] eta: 0:03:54 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2148 (0.3326) acc1: 96.3542 (87.9861) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1589 data: 0.0666 max mem: 3536 +Test: [1180/2387] eta: 0:03:52 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2078 (0.3317) acc1: 96.0938 (88.0559) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1741 data: 0.0814 max mem: 3536 +Test: [1190/2387] eta: 0:03:49 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1892 (0.3307) acc1: 98.1771 (88.1321) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1518 data: 0.0594 max mem: 3536 +Test: [1200/2387] eta: 0:03:47 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1881 (0.3297) acc1: 98.4375 (88.1995) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1426 data: 0.0519 max mem: 3536 +Test: [1210/2387] eta: 0:03:45 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1803 (0.3285) acc1: 98.4375 (88.2746) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1553 data: 0.0652 max mem: 3536 +Test: [1220/2387] eta: 0:03:42 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1731 (0.3271) acc1: 98.4375 (88.3563) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1552 data: 0.0642 max mem: 3536 +Test: [1230/2387] eta: 0:03:40 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1548 (0.3256) acc1: 98.9583 (88.4418) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1596 data: 0.0675 max mem: 3536 +Test: [1240/2387] eta: 0:03:38 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1423 (0.3242) acc1: 98.9583 (88.5217) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1664 data: 0.0732 max mem: 3536 +Test: [1250/2387] eta: 0:03:36 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1523 (0.3229) acc1: 98.6979 (88.5995) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1793 data: 0.0868 max mem: 3536 +Test: [1260/2387] eta: 0:03:34 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1457 (0.3214) acc1: 99.4792 (88.6868) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1977 data: 0.1063 max mem: 3536 +Test: [1270/2387] eta: 0:03:32 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1266 (0.3198) acc1: 100.0000 (88.7744) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1951 data: 0.1023 max mem: 3536 +Test: [1280/2387] eta: 0:03:30 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1250 (0.3184) acc1: 99.7396 (88.8541) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1856 data: 0.0918 max mem: 3536 +Test: [1290/2387] eta: 0:03:29 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1408 (0.3171) acc1: 98.6979 (88.9207) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2096 data: 0.1172 max mem: 3536 +Test: [1300/2387] eta: 0:03:27 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1571 (0.3160) acc1: 96.6146 (88.9798) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2241 data: 0.1316 max mem: 3536 +Test: [1310/2387] eta: 0:03:25 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1602 (0.3148) acc1: 96.8750 (89.0452) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1888 data: 0.0946 max mem: 3536 +Test: [1320/2387] eta: 0:03:23 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1504 (0.3136) acc1: 98.9583 (89.1212) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1722 data: 0.0776 max mem: 3536 +Test: [1330/2387] eta: 0:03:21 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1504 (0.3124) acc1: 99.4792 (89.1938) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1875 data: 0.0951 max mem: 3536 +Test: [1340/2387] eta: 0:03:19 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1649 (0.3125) acc1: 98.6979 (89.1814) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1880 data: 0.0973 max mem: 3536 +Test: [1350/2387] eta: 0:03:17 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1130 (0.3109) acc1: 100.0000 (89.2614) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1832 data: 0.0910 max mem: 3537 +Test: [1360/2387] eta: 0:03:15 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.0976 (0.3094) acc1: 100.0000 (89.3403) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1844 data: 0.0865 max mem: 3537 +Test: [1370/2387] eta: 0:03:13 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.0974 (0.3078) acc1: 100.0000 (89.4181) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1831 data: 0.0859 max mem: 3537 +Test: [1380/2387] eta: 0:03:11 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.0994 (0.3063) acc1: 100.0000 (89.4947) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1783 data: 0.0869 max mem: 3537 +Test: [1390/2387] eta: 0:03:09 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1032 (0.3049) acc1: 100.0000 (89.5702) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1752 data: 0.0842 max mem: 3537 +Test: [1400/2387] eta: 0:03:07 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.0968 (0.3034) acc1: 100.0000 (89.6447) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1716 data: 0.0804 max mem: 3537 +Test: [1410/2387] eta: 0:03:05 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.0949 (0.3032) acc1: 100.0000 (89.6511) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.1766 data: 0.0858 max mem: 3537 +Test: [1420/2387] eta: 0:03:03 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2978 (0.3032) acc1: 91.9271 (89.6515) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.1849 data: 0.0942 max mem: 3537 +Test: [1430/2387] eta: 0:03:01 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3136 (0.3040) acc1: 87.7604 (89.6003) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1876 data: 0.0963 max mem: 3537 +Test: [1440/2387] eta: 0:02:59 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3335 (0.3042) acc1: 87.5000 (89.5907) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.1832 data: 0.0920 max mem: 3537 +Test: [1450/2387] eta: 0:02:58 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2898 (0.3041) acc1: 90.1042 (89.5997) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1819 data: 0.0907 max mem: 3537 +Test: [1460/2387] eta: 0:02:56 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2598 (0.3039) acc1: 91.4062 (89.6081) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1857 data: 0.0949 max mem: 3537 +Test: [1470/2387] eta: 0:02:54 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2295 (0.3032) acc1: 93.7500 (89.6480) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1843 data: 0.0933 max mem: 3537 +Test: [1480/2387] eta: 0:02:52 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2295 (0.3028) acc1: 93.7500 (89.6749) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1942 data: 0.1029 max mem: 3537 +Test: [1490/2387] eta: 0:02:50 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2278 (0.3021) acc1: 94.2708 (89.7215) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2028 data: 0.1103 max mem: 3537 +Test: [1500/2387] eta: 0:02:48 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1984 (0.3016) acc1: 97.9167 (89.7603) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1947 data: 0.0952 max mem: 3537 +Test: [1510/2387] eta: 0:02:46 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1862 (0.3008) acc1: 97.9167 (89.8182) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1812 data: 0.0822 max mem: 3537 +Test: [1520/2387] eta: 0:02:44 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2058 (0.3011) acc1: 96.6146 (89.7902) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1899 data: 0.0970 max mem: 3537 +Test: [1530/2387] eta: 0:02:42 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3222 (0.3014) acc1: 87.2396 (89.7713) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.1985 data: 0.1053 max mem: 3537 +Test: [1540/2387] eta: 0:02:41 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3373 (0.3018) acc1: 86.9792 (89.7474) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.2149 data: 0.1198 max mem: 3537 +Test: [1550/2387] eta: 0:02:39 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3242 (0.3018) acc1: 88.8021 (89.7504) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.2198 data: 0.1251 max mem: 3537 +Test: [1560/2387] eta: 0:02:37 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2796 (0.3018) acc1: 91.6667 (89.7510) acc5: 100.0000 (100.0000) acc1_neg: 75.521 acc5_neg: 100.000 time: 0.1699 data: 0.0745 max mem: 3537 +Test: [1570/2387] eta: 0:02:35 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2445 (0.3014) acc1: 96.0938 (89.7856) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1563 data: 0.0607 max mem: 3537 +Test: [1580/2387] eta: 0:02:33 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2270 (0.3011) acc1: 96.3542 (89.8077) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1954 data: 0.1035 max mem: 3538 +Test: [1590/2387] eta: 0:02:31 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2499 (0.3008) acc1: 94.2708 (89.8333) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2085 data: 0.1155 max mem: 3538 +Test: [1600/2387] eta: 0:02:29 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2499 (0.3006) acc1: 94.2708 (89.8481) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.1843 data: 0.0912 max mem: 3538 +Test: [1610/2387] eta: 0:02:27 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2717 (0.3005) acc1: 94.7917 (89.8737) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1799 data: 0.0882 max mem: 3538 +Test: [1620/2387] eta: 0:02:25 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2717 (0.3004) acc1: 94.5312 (89.8939) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1955 data: 0.1037 max mem: 3538 +Test: [1630/2387] eta: 0:02:23 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2602 (0.3005) acc1: 93.2292 (89.8853) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1849 data: 0.0913 max mem: 3538 +Test: [1640/2387] eta: 0:02:22 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2608 (0.3001) acc1: 92.4479 (89.9082) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1805 data: 0.0844 max mem: 3538 +Test: [1650/2387] eta: 0:02:20 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2435 (0.2998) acc1: 93.4896 (89.9245) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1915 data: 0.0959 max mem: 3538 +Test: [1660/2387] eta: 0:02:18 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2540 (0.2998) acc1: 93.2292 (89.9316) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1886 data: 0.0934 max mem: 3538 +Test: [1670/2387] eta: 0:02:16 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2749 (0.2997) acc1: 92.1875 (89.9302) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2133 data: 0.1192 max mem: 3538 +Test: [1680/2387] eta: 0:02:14 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2794 (0.2998) acc1: 91.4062 (89.9245) acc5: 100.0000 (100.0000) acc1_neg: 76.302 acc5_neg: 100.000 time: 0.2437 data: 0.1498 max mem: 3538 +Test: [1690/2387] eta: 0:02:12 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2735 (0.2995) acc1: 92.4479 (89.9435) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2195 data: 0.1238 max mem: 3538 +Test: [1700/2387] eta: 0:02:10 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2565 (0.2993) acc1: 95.0521 (89.9626) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1699 data: 0.0762 max mem: 3538 +Test: [1710/2387] eta: 0:02:08 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1915 (0.2987) acc1: 97.3958 (90.0065) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1650 data: 0.0739 max mem: 3538 +Test: [1720/2387] eta: 0:02:06 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1672 (0.2979) acc1: 98.9583 (90.0532) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1769 data: 0.0855 max mem: 3538 +Test: [1730/2387] eta: 0:02:05 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1611 (0.2972) acc1: 98.9583 (90.0950) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1903 data: 0.0952 max mem: 3538 +Test: [1740/2387] eta: 0:02:03 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1701 (0.2965) acc1: 97.9167 (90.1393) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.2046 data: 0.1095 max mem: 3538 +Test: [1750/2387] eta: 0:02:01 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1699 (0.2957) acc1: 98.9583 (90.1891) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1892 data: 0.0967 max mem: 3538 +Test: [1760/2387] eta: 0:01:59 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1671 (0.2950) acc1: 99.2188 (90.2321) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1840 data: 0.0918 max mem: 3538 +Test: [1770/2387] eta: 0:01:57 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2093 (0.2950) acc1: 94.2708 (90.2340) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1849 data: 0.0933 max mem: 3538 +Test: [1780/2387] eta: 0:01:55 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2776 (0.2949) acc1: 92.1875 (90.2442) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1891 data: 0.0977 max mem: 3538 +Test: [1790/2387] eta: 0:01:53 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2709 (0.2948) acc1: 93.2292 (90.2523) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1863 data: 0.0957 max mem: 3538 +Test: [1800/2387] eta: 0:01:51 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2767 (0.2948) acc1: 92.4479 (90.2558) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1804 data: 0.0890 max mem: 3538 +Test: [1810/2387] eta: 0:01:49 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2515 (0.2944) acc1: 94.2708 (90.2897) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1855 data: 0.0922 max mem: 3539 +Test: [1820/2387] eta: 0:01:47 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2114 (0.2939) acc1: 97.6562 (90.3267) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1832 data: 0.0884 max mem: 3539 +Test: [1830/2387] eta: 0:01:45 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1739 (0.2932) acc1: 99.4792 (90.3710) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1876 data: 0.0939 max mem: 3539 +Test: [1840/2387] eta: 0:01:44 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1656 (0.2925) acc1: 99.2188 (90.4195) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1871 data: 0.0962 max mem: 3539 +Test: [1850/2387] eta: 0:01:42 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1656 (0.2918) acc1: 99.2188 (90.4680) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1748 data: 0.0844 max mem: 3539 +Test: [1860/2387] eta: 0:01:40 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1640 (0.2912) acc1: 99.2188 (90.5153) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1797 data: 0.0849 max mem: 3539 +Test: [1870/2387] eta: 0:01:38 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1640 (0.2905) acc1: 99.2188 (90.5614) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1837 data: 0.0883 max mem: 3539 +Test: [1880/2387] eta: 0:01:36 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1641 (0.2898) acc1: 99.2188 (90.6077) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1782 data: 0.0849 max mem: 3539 +Test: [1890/2387] eta: 0:01:34 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1644 (0.2892) acc1: 99.2188 (90.6530) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1859 data: 0.0933 max mem: 3539 +Test: [1900/2387] eta: 0:01:32 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1677 (0.2885) acc1: 99.2188 (90.6986) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1796 data: 0.0890 max mem: 3539 +Test: [1910/2387] eta: 0:01:30 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1732 (0.2881) acc1: 98.9583 (90.7388) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1687 data: 0.0761 max mem: 3539 +Test: [1920/2387] eta: 0:01:28 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2190 (0.2891) acc1: 96.0938 (90.6624) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.1673 data: 0.0735 max mem: 3539 +Test: [1930/2387] eta: 0:01:26 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3310 (0.2891) acc1: 86.9792 (90.6702) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1649 data: 0.0710 max mem: 3539 +Test: [1940/2387] eta: 0:01:24 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3134 (0.2901) acc1: 92.7083 (90.6039) acc5: 100.0000 (100.0000) acc1_neg: 59.115 acc5_neg: 100.000 time: 0.1681 data: 0.0740 max mem: 3539 +Test: [1950/2387] eta: 0:01:22 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3686 (0.2911) acc1: 85.6771 (90.5436) acc5: 100.0000 (100.0000) acc1_neg: 80.729 acc5_neg: 100.000 time: 0.1738 data: 0.0812 max mem: 3539 +Test: [1960/2387] eta: 0:01:20 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3686 (0.2915) acc1: 85.1562 (90.5322) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.1693 data: 0.0769 max mem: 3539 +Test: [1970/2387] eta: 0:01:18 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3501 (0.2920) acc1: 86.1979 (90.4836) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.1710 data: 0.0786 max mem: 3539 +Test: [1980/2387] eta: 0:01:16 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3189 (0.2917) acc1: 86.9792 (90.4914) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1760 data: 0.0843 max mem: 3539 +Test: [1990/2387] eta: 0:01:15 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1151 (0.2908) acc1: 99.2188 (90.5358) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1771 data: 0.0817 max mem: 3539 +Test: [2000/2387] eta: 0:01:13 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1137 (0.2899) acc1: 99.4792 (90.5808) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1804 data: 0.0851 max mem: 3539 +Test: [2010/2387] eta: 0:01:11 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1130 (0.2890) acc1: 99.4792 (90.6258) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1750 data: 0.0839 max mem: 3539 +Test: [2020/2387] eta: 0:01:09 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1130 (0.2881) acc1: 99.7396 (90.6702) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1750 data: 0.0832 max mem: 3539 +Test: [2030/2387] eta: 0:01:07 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1118 (0.2873) acc1: 99.4792 (90.7135) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1793 data: 0.0872 max mem: 3540 +Test: [2040/2387] eta: 0:01:05 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1124 (0.2864) acc1: 99.4792 (90.7551) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1842 data: 0.0926 max mem: 3540 +Test: [2050/2387] eta: 0:01:03 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1165 (0.2856) acc1: 98.9583 (90.7960) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1808 data: 0.0895 max mem: 3540 +Test: [2060/2387] eta: 0:01:01 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1325 (0.2849) acc1: 98.9583 (90.8356) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1888 data: 0.0973 max mem: 3540 +Test: [2070/2387] eta: 0:00:59 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1385 (0.2842) acc1: 98.9583 (90.8750) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1982 data: 0.1063 max mem: 3540 +Test: [2080/2387] eta: 0:00:57 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1371 (0.2835) acc1: 98.9583 (90.9133) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1927 data: 0.1013 max mem: 3540 +Test: [2090/2387] eta: 0:00:56 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1371 (0.2828) acc1: 98.6979 (90.9506) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1948 data: 0.1038 max mem: 3540 +Test: [2100/2387] eta: 0:00:54 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1386 (0.2821) acc1: 98.6979 (90.9864) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1870 data: 0.0947 max mem: 3540 +Test: [2110/2387] eta: 0:00:52 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1408 (0.2815) acc1: 98.4375 (91.0233) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1790 data: 0.0855 max mem: 3540 +Test: [2120/2387] eta: 0:00:50 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1465 (0.2816) acc1: 98.1771 (91.0146) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.1781 data: 0.0853 max mem: 3540 +Test: [2130/2387] eta: 0:00:48 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2619 (0.2815) acc1: 91.6667 (91.0214) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1761 data: 0.0845 max mem: 3540 +Test: [2140/2387] eta: 0:00:46 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2523 (0.2818) acc1: 91.1458 (91.0074) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1738 data: 0.0831 max mem: 3540 +Test: [2150/2387] eta: 0:00:44 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2933 (0.2820) acc1: 90.1042 (91.0016) acc5: 100.0000 (100.0000) acc1_neg: 75.521 acc5_neg: 100.000 time: 0.1767 data: 0.0862 max mem: 3540 +Test: [2160/2387] eta: 0:00:42 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2692 (0.2819) acc1: 93.2292 (91.0204) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1782 data: 0.0878 max mem: 3540 +Test: [2170/2387] eta: 0:00:40 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2790 (0.2821) acc1: 91.6667 (91.0077) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1692 data: 0.0780 max mem: 3540 +Test: [2180/2387] eta: 0:00:38 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2989 (0.2819) acc1: 89.5833 (91.0208) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1683 data: 0.0764 max mem: 3540 +Test: [2190/2387] eta: 0:00:37 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.1287 (0.2812) acc1: 98.9583 (91.0474) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1744 data: 0.0829 max mem: 3540 +Test: [2200/2387] eta: 0:00:35 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3081 (0.2817) acc1: 91.4062 (91.0113) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1730 data: 0.0821 max mem: 3540 +Test: [2210/2387] eta: 0:00:33 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3354 (0.2819) acc1: 90.3646 (91.0097) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1720 data: 0.0803 max mem: 3540 +Test: [2220/2387] eta: 0:00:31 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3365 (0.2822) acc1: 90.3646 (90.9963) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1736 data: 0.0822 max mem: 3540 +Test: [2230/2387] eta: 0:00:29 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3147 (0.2823) acc1: 91.4062 (91.0049) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1713 data: 0.0781 max mem: 3540 +Test: [2240/2387] eta: 0:00:27 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2862 (0.2823) acc1: 93.4896 (91.0124) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.1748 data: 0.0816 max mem: 3540 +Test: [2250/2387] eta: 0:00:25 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2579 (0.2822) acc1: 93.7500 (91.0210) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1724 data: 0.0808 max mem: 3540 +Test: [2260/2387] eta: 0:00:23 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2299 (0.2822) acc1: 95.0521 (91.0304) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.1753 data: 0.0827 max mem: 3541 +Test: [2270/2387] eta: 0:00:21 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2278 (0.2820) acc1: 94.5312 (91.0450) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1872 data: 0.0935 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2269 (0.2820) acc1: 94.2708 (91.0449) acc5: 100.0000 (100.0000) acc1_neg: 75.521 acc5_neg: 100.000 time: 0.1867 data: 0.0928 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2266 (0.2819) acc1: 94.2708 (91.0582) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1855 data: 0.0933 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2434 (0.2819) acc1: 95.0521 (91.0641) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1850 data: 0.0928 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2481 (0.2818) acc1: 95.0521 (91.0686) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1800 data: 0.0881 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.2630 (0.2819) acc1: 93.2292 (91.0690) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.1784 data: 0.0855 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.3270 (0.2831) acc1: 86.4583 (90.9851) acc5: 100.0000 (100.0000) acc1_neg: 72.396 acc5_neg: 100.000 time: 0.1816 data: 0.0884 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.4465 (0.2834) acc1: 77.6042 (90.9608) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.1721 data: 0.0806 max mem: 3541 +Test: [2350/2387] eta: 0:00:06 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.4465 (0.2848) acc1: 78.3854 (90.8614) acc5: 100.0000 (100.0000) acc1_neg: 75.521 acc5_neg: 100.000 time: 0.1475 data: 0.0568 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.4833 (0.2859) acc1: 75.5208 (90.7823) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1351 data: 0.0449 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.4295 (0.2869) acc1: 82.0312 (90.7139) acc5: 100.0000 (100.0000) acc1_neg: 61.458 acc5_neg: 100.000 time: 0.1345 data: 0.0445 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.4284 (0.2874) acc1: 78.1250 (90.6777) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.1309 data: 0.0411 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 68.037 acc5_pos: 100.000 loss: 0.4167 (0.2873) acc1: 78.1250 (90.6758) acc5: 100.0000 (100.0000) acc1_neg: 97.674 acc5_neg: 100.000 time: 0.1274 data: 0.0402 max mem: 3541 +Test: Total time: 0:07:25 (0.1867 s / it) +* Overall Acc@1 90.676 Acc@5 100.000 + target == real Acc@1 52.115 Acc@5 100.000 + target == fake Acc@1 93.384 Acc@5 100.000 + loss 0.287 + PR-AUC 0.9857 ROC-AUC 0.8525 EER 23.10% (thr=0.7515) +Accuracy of the network on the 916267 test images: 90.7% +[rank0]:[W725 16:59:10.440340149 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 16:59:30.640000 2556135 torch/distributed/run.py:766] +W0725 16:59:30.640000 2556135 torch/distributed/run.py:766] ***************************************** +W0725 16:59:30.640000 2556135 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 16:59:30.640000 2556135 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 16:59:59.879682684 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 16:59:59.958358922 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:00:00.309963550 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:00:00.450677549 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:00:00.497431253 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 17:00:00.505151947 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:00:00.521309377 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank6]:[W725 17:00:00.521320911 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed1_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed1_resampled_n116-040-168_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:01:56 acc1_pos: 52.865 acc5_pos: 100.000 loss: 0.7543 (0.7543) acc1: 52.8646 (52.8646) acc5: 100.0000 (100.0000) time: 7.1879 data: 5.6096 max mem: 3531 +Test: [ 10/517] eta: 0:06:13 acc1_pos: 51.302 acc5_pos: 100.000 loss: 0.8711 (0.8367) acc1: 52.8646 (52.7699) acc5: 100.0000 (100.0000) time: 0.7361 data: 0.5104 max mem: 3531 +Test: [ 20/517] eta: 0:03:35 acc1_pos: 69.792 acc5_pos: 100.000 loss: 0.7706 (0.8084) acc1: 57.2917 (55.0595) acc5: 100.0000 (100.0000) time: 0.0952 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:03:00 acc1_pos: 12.760 acc5_pos: 100.000 loss: 0.7398 (0.8352) acc1: 55.4688 (52.8562) acc5: 100.0000 (100.0000) time: 0.1699 data: 0.0713 max mem: 3531 +Test: [ 40/517] eta: 0:02:42 acc1_pos: 44.271 acc5_pos: 100.000 loss: 0.8901 (0.8718) acc1: 41.9271 (49.6443) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1490 max mem: 3531 +Test: [ 50/517] eta: 0:02:28 acc1_pos: 75.229 acc5_pos: 100.000 loss: 0.8836 (0.8626) acc1: 44.2708 (50.1685) acc5: 100.0000 (100.0000) acc1_neg: 61.198 acc5_neg: 100.000 time: 0.2358 data: 0.1440 max mem: 3531 +Test: [ 60/517] eta: 0:02:20 acc1_pos: 75.229 acc5_pos: 100.000 loss: 0.6153 (0.8072) acc1: 64.5833 (54.3118) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.2349 data: 0.1422 max mem: 3531 +Test: [ 70/517] eta: 0:02:14 acc1_pos: 75.229 acc5_pos: 100.000 loss: 0.6132 (0.7914) acc1: 66.1458 (55.4871) acc5: 100.0000 (100.0000) acc1_neg: 66.146 acc5_neg: 100.000 time: 0.2542 data: 0.1575 max mem: 3531 +Test: [ 80/517] eta: 0:02:08 acc1_pos: 75.229 acc5_pos: 100.000 loss: 0.7965 (0.7977) acc1: 53.1250 (54.9801) acc5: 100.0000 (100.0000) acc1_neg: 46.615 acc5_neg: 100.000 time: 0.2556 data: 0.1609 max mem: 3531 +Test: [ 90/517] eta: 0:02:04 acc1_pos: 75.229 acc5_pos: 100.000 loss: 0.6122 (0.7572) acc1: 62.2396 (58.0729) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.2666 data: 0.1748 max mem: 3531 +Test: [100/517] eta: 0:02:00 acc1_pos: 75.229 acc5_pos: 100.000 loss: 0.5275 (0.7411) acc1: 71.3542 (58.7897) acc5: 100.0000 (100.0000) acc1_neg: 69.531 acc5_neg: 100.000 time: 0.2642 data: 0.1695 max mem: 3531 +Test: [110/517] eta: 0:01:58 acc1_pos: 75.229 acc5_pos: 100.000 loss: 0.5164 (0.7156) acc1: 71.3542 (60.7076) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.2779 data: 0.1798 max mem: 3531 +Test: [120/517] eta: 0:01:53 acc1_pos: 75.229 acc5_pos: 100.000 loss: 0.4395 (0.6941) acc1: 79.1667 (62.2869) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.2716 data: 0.1735 max mem: 3531 +Test: [130/517] eta: 0:01:49 acc1_pos: 75.229 acc5_pos: 100.000 loss: 0.4509 (0.6754) acc1: 79.9479 (63.5993) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.2507 data: 0.1539 max mem: 3531 +Test: [140/517] eta: 0:01:45 acc1_pos: 75.229 acc5_pos: 100.000 loss: 0.4105 (0.6542) acc1: 81.7708 (65.1780) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.2532 data: 0.1565 max mem: 3531 +Test: [150/517] eta: 0:01:42 acc1_pos: 75.229 acc5_pos: 100.000 loss: 0.3896 (0.6371) acc1: 85.4167 (66.5735) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.2491 data: 0.1515 max mem: 3531 +Test: [160/517] eta: 0:01:38 acc1_pos: 75.229 acc5_pos: 100.000 loss: 0.4607 (0.6300) acc1: 75.2604 (66.8333) acc5: 100.0000 (100.0000) acc1_neg: 68.750 acc5_neg: 100.000 time: 0.2514 data: 0.1538 max mem: 3531 +Test: [170/517] eta: 0:01:34 acc1_pos: 75.229 acc5_pos: 100.000 loss: 0.5253 (0.6215) acc1: 69.5312 (67.2164) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.2269 data: 0.1309 max mem: 3531 +Test: [180/517] eta: 0:01:31 acc1_pos: 75.229 acc5_pos: 100.000 loss: 0.4126 (0.6051) acc1: 82.5521 (68.4176) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.2398 data: 0.1458 max mem: 3531 +Test: [190/517] eta: 0:01:27 acc1_pos: 75.229 acc5_pos: 100.000 loss: 0.2422 (0.5847) acc1: 91.4062 (69.7726) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.2282 data: 0.1337 max mem: 3531 +Test: [200/517] eta: 0:01:24 acc1_pos: 75.229 acc5_pos: 100.000 loss: 0.2453 (0.5685) acc1: 94.2708 (71.0277) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2158 data: 0.1197 max mem: 3531 +Test: [210/517] eta: 0:01:21 acc1_pos: 75.229 acc5_pos: 100.000 loss: 0.3250 (0.5683) acc1: 85.9375 (70.9963) acc5: 100.0000 (100.0000) acc1_neg: 78.646 acc5_neg: 100.000 time: 0.2496 data: 0.1543 max mem: 3532 +Test: [220/517] eta: 0:01:19 acc1_pos: 75.229 acc5_pos: 100.000 loss: 0.6270 (0.5719) acc1: 61.4583 (70.5894) acc5: 100.0000 (100.0000) acc1_neg: 55.469 acc5_neg: 100.000 time: 0.2548 data: 0.1586 max mem: 3532 +Test: [230/517] eta: 0:01:16 acc1_pos: 75.229 acc5_pos: 100.000 loss: 0.6034 (0.5734) acc1: 66.4062 (70.4455) acc5: 100.0000 (100.0000) acc1_neg: 33.594 acc5_neg: 100.000 time: 0.2514 data: 0.1552 max mem: 3532 +Test: [240/517] eta: 0:01:13 acc1_pos: 75.229 acc5_pos: 100.000 loss: 0.7014 (0.5925) acc1: 44.2708 (68.8310) acc5: 100.0000 (100.0000) acc1_neg: 17.448 acc5_neg: 100.000 time: 0.2534 data: 0.1586 max mem: 3532 +Test: [250/517] eta: 0:01:10 acc1_pos: 75.229 acc5_pos: 100.000 loss: 0.8512 (0.5930) acc1: 44.0104 (68.8755) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.2558 data: 0.1595 max mem: 3532 +Test: [260/517] eta: 0:01:07 acc1_pos: 75.229 acc5_pos: 100.000 loss: 0.4693 (0.5896) acc1: 82.5521 (69.2239) acc5: 100.0000 (100.0000) acc1_neg: 69.010 acc5_neg: 100.000 time: 0.2541 data: 0.1586 max mem: 3532 +Test: [270/517] eta: 0:01:05 acc1_pos: 75.229 acc5_pos: 100.000 loss: 0.6275 (0.5939) acc1: 67.9688 (68.8701) acc5: 100.0000 (100.0000) acc1_neg: 51.042 acc5_neg: 100.000 time: 0.2533 data: 0.1569 max mem: 3532 +Test: [280/517] eta: 0:01:02 acc1_pos: 75.229 acc5_pos: 100.000 loss: 0.6679 (0.5977) acc1: 64.5833 (68.7213) acc5: 100.0000 (100.0000) acc1_neg: 58.333 acc5_neg: 100.000 time: 0.2551 data: 0.1573 max mem: 3532 +Test: [290/517] eta: 0:00:59 acc1_pos: 75.229 acc5_pos: 100.000 loss: 0.6402 (0.6038) acc1: 62.5000 (68.1835) acc5: 100.0000 (100.0000) acc1_neg: 48.177 acc5_neg: 100.000 time: 0.2492 data: 0.1513 max mem: 3532 +Test: [300/517] eta: 0:00:57 acc1_pos: 75.229 acc5_pos: 100.000 loss: 0.6588 (0.6061) acc1: 58.8542 (67.9739) acc5: 100.0000 (100.0000) acc1_neg: 69.531 acc5_neg: 100.000 time: 0.2497 data: 0.1505 max mem: 3532 +Test: [310/517] eta: 0:00:54 acc1_pos: 75.229 acc5_pos: 100.000 loss: 0.5936 (0.6057) acc1: 63.0208 (67.9604) acc5: 100.0000 (100.0000) acc1_neg: 51.562 acc5_neg: 100.000 time: 0.2567 data: 0.1577 max mem: 3532 +Test: [320/517] eta: 0:00:51 acc1_pos: 75.229 acc5_pos: 100.000 loss: 0.5029 (0.6015) acc1: 75.0000 (68.3233) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.2664 data: 0.1697 max mem: 3532 +Test: [330/517] eta: 0:00:48 acc1_pos: 75.229 acc5_pos: 100.000 loss: 0.4397 (0.5961) acc1: 80.9896 (68.7665) acc5: 100.0000 (100.0000) acc1_neg: 75.521 acc5_neg: 100.000 time: 0.2543 data: 0.1591 max mem: 3532 +Test: [340/517] eta: 0:00:46 acc1_pos: 75.229 acc5_pos: 100.000 loss: 0.4307 (0.5915) acc1: 83.8542 (69.2082) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.2456 data: 0.1498 max mem: 3532 +Test: [350/517] eta: 0:00:43 acc1_pos: 75.229 acc5_pos: 100.000 loss: 0.4921 (0.5953) acc1: 80.9896 (68.9355) acc5: 100.0000 (100.0000) acc1_neg: 35.677 acc5_neg: 100.000 time: 0.2556 data: 0.1605 max mem: 3532 +Test: [360/517] eta: 0:00:41 acc1_pos: 75.229 acc5_pos: 100.000 loss: 0.7506 (0.5956) acc1: 54.6875 (68.9578) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2500 data: 0.1556 max mem: 3532 +Test: [370/517] eta: 0:00:38 acc1_pos: 75.229 acc5_pos: 100.000 loss: 0.2503 (0.5861) acc1: 97.1354 (69.7411) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2533 data: 0.1590 max mem: 3532 +Test: [380/517] eta: 0:00:35 acc1_pos: 75.229 acc5_pos: 100.000 loss: 0.2296 (0.5761) acc1: 98.6979 (70.5073) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2554 data: 0.1595 max mem: 3532 +Test: [390/517] eta: 0:00:33 acc1_pos: 75.229 acc5_pos: 100.000 loss: 0.2400 (0.5690) acc1: 97.1354 (71.0072) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2599 data: 0.1645 max mem: 3532 +Test: [400/517] eta: 0:00:30 acc1_pos: 75.229 acc5_pos: 100.000 loss: 0.2525 (0.5607) acc1: 96.3542 (71.6951) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2539 data: 0.1584 max mem: 3532 +Test: [410/517] eta: 0:00:27 acc1_pos: 75.229 acc5_pos: 100.000 loss: 0.2388 (0.5542) acc1: 97.9167 (72.2507) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.2600 data: 0.1635 max mem: 3532 +Test: [420/517] eta: 0:00:25 acc1_pos: 75.229 acc5_pos: 100.000 loss: 0.4136 (0.5524) acc1: 84.8958 (72.4459) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.2546 data: 0.1582 max mem: 3532 +Test: [430/517] eta: 0:00:22 acc1_pos: 75.229 acc5_pos: 100.000 loss: 0.4695 (0.5505) acc1: 80.7292 (72.6049) acc5: 100.0000 (100.0000) acc1_neg: 80.729 acc5_neg: 100.000 time: 0.2372 data: 0.1405 max mem: 3532 +Test: [440/517] eta: 0:00:20 acc1_pos: 75.229 acc5_pos: 100.000 loss: 0.4247 (0.5475) acc1: 83.3333 (72.8659) acc5: 100.0000 (100.0000) acc1_neg: 72.656 acc5_neg: 100.000 time: 0.2487 data: 0.1544 max mem: 3533 +Test: [450/517] eta: 0:00:17 acc1_pos: 75.229 acc5_pos: 100.000 loss: 0.3740 (0.5433) acc1: 86.9792 (73.2527) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2482 data: 0.1526 max mem: 3533 +Test: [460/517] eta: 0:00:14 acc1_pos: 75.229 acc5_pos: 100.000 loss: 0.3225 (0.5382) acc1: 93.7500 (73.6974) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2447 data: 0.1469 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 75.229 acc5_pos: 100.000 loss: 0.2938 (0.5335) acc1: 93.2292 (74.0855) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2602 data: 0.1622 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 75.229 acc5_pos: 100.000 loss: 0.2900 (0.5285) acc1: 94.2708 (74.5528) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2503 data: 0.1530 max mem: 3533 +Test: [490/517] eta: 0:00:06 acc1_pos: 75.229 acc5_pos: 100.000 loss: 0.2863 (0.5244) acc1: 96.0938 (74.9268) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.2282 data: 0.1312 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 75.229 acc5_pos: 100.000 loss: 0.5250 (0.5289) acc1: 75.0000 (74.5473) acc5: 100.0000 (100.0000) acc1_neg: 52.865 acc5_neg: 100.000 time: 0.2047 data: 0.1107 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 75.229 acc5_pos: 100.000 loss: 0.5671 (0.5269) acc1: 68.4896 (74.7116) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1836 data: 0.0939 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 75.229 acc5_pos: 100.000 loss: 0.5551 (0.5271) acc1: 72.1354 (74.7187) acc5: 100.0000 (100.0000) acc1_neg: 60.071 acc5_neg: 100.000 time: 0.1850 data: 0.0938 max mem: 3533 +Test: Total time: 0:02:11 (0.2537 s / it) +* Overall Acc@1 74.719 Acc@5 100.000 + target == real Acc@1 49.442 Acc@5 100.000 + target == fake Acc@1 77.341 Acc@5 100.000 + loss 0.527 + PR-AUC 0.9401 ROC-AUC 0.6746 EER 37.44% (thr=0.6348) +Accuracy of the network on the 198427 test images: 74.7% +[rank0]:[W725 17:02:22.360176859 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0725 17:02:41.488000 2556708 torch/distributed/run.py:766] +W0725 17:02:41.488000 2556708 torch/distributed/run.py:766] ***************************************** +W0725 17:02:41.488000 2556708 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:02:41.488000 2556708 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +[rank5]:[W725 17:03:11.434186383 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:03:11.444865160 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:03:11.810116375 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:03:11.837050508 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:03:11.859489813 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:03:11.159095993 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:03:12.195486752 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 17:03:12.234410066 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed1_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed1_resampled_n116-040-168_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:40:43 acc1_pos: 41.146 acc5_pos: 100.000 loss: 0.8840 (0.8840) acc1: 41.1458 (41.1458) acc5: 100.0000 (100.0000) time: 6.2810 data: 4.8783 max mem: 3531 +Test: [ 10/389] eta: 0:04:08 acc1_pos: 99.740 acc5_pos: 100.000 loss: 0.2772 (0.4094) acc1: 92.7083 (80.2320) acc5: 100.0000 (100.0000) time: 0.6561 data: 0.4438 max mem: 3531 +Test: [ 20/389] eta: 0:02:24 acc1_pos: 72.917 acc5_pos: 100.000 loss: 0.2772 (0.3472) acc1: 92.9688 (85.9747) acc5: 100.0000 (100.0000) time: 0.0968 data: 0.0004 max mem: 3531 +Test: [ 30/389] eta: 0:02:00 acc1_pos: 84.896 acc5_pos: 100.000 loss: 0.3529 (0.4446) acc1: 90.1042 (77.0749) acc5: 100.0000 (100.0000) time: 0.1614 data: 0.0655 max mem: 3531 +Test: [ 40/389] eta: 0:01:49 acc1_pos: 99.219 acc5_pos: 100.000 loss: 0.3683 (0.4476) acc1: 84.8958 (78.2520) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1373 max mem: 3531 +Test: [ 50/389] eta: 0:01:37 acc1_pos: 89.844 acc5_pos: 100.000 loss: 0.2994 (0.4150) acc1: 89.8438 (81.2092) acc5: 100.0000 (100.0000) time: 0.2134 data: 0.1178 max mem: 3531 +Test: [ 60/389] eta: 0:01:29 acc1_pos: 71.875 acc5_pos: 100.000 loss: 0.3883 (0.4520) acc1: 87.5000 (78.1250) acc5: 100.0000 (100.0000) time: 0.1889 data: 0.0934 max mem: 3531 +Test: [ 70/389] eta: 0:01:24 acc1_pos: 56.250 acc5_pos: 100.000 loss: 0.6012 (0.4880) acc1: 59.8958 (75.5465) acc5: 100.0000 (100.0000) time: 0.2013 data: 0.1097 max mem: 3531 +Test: [ 80/389] eta: 0:01:17 acc1_pos: 99.479 acc5_pos: 100.000 loss: 0.5960 (0.4951) acc1: 70.3125 (75.6109) acc5: 100.0000 (100.0000) time: 0.1878 data: 0.0971 max mem: 3531 +Test: [ 90/389] eta: 0:01:13 acc1_pos: 98.698 acc5_pos: 100.000 loss: 0.2543 (0.4637) acc1: 86.9792 (77.8560) acc5: 100.0000 (100.0000) time: 0.1848 data: 0.0933 max mem: 3531 +Test: [100/389] eta: 0:01:09 acc1_pos: 76.302 acc5_pos: 100.000 loss: 0.5167 (0.4958) acc1: 73.4375 (75.1392) acc5: 100.0000 (100.0000) time: 0.2034 data: 0.1105 max mem: 3531 +Test: [110/389] eta: 0:01:06 acc1_pos: 58.073 acc5_pos: 100.000 loss: 0.6629 (0.4957) acc1: 60.9375 (75.4106) acc5: 100.0000 (100.0000) time: 0.1948 data: 0.1012 max mem: 3531 +Test: [120/389] eta: 0:01:02 acc1_pos: 93.229 acc5_pos: 100.000 loss: 0.4248 (0.5035) acc1: 87.5000 (75.1937) acc5: 100.0000 (100.0000) time: 0.1911 data: 0.0990 max mem: 3531 +Test: [130/389] eta: 0:00:59 acc1_pos: 19.632 acc5_pos: 100.000 loss: 0.8506 (0.5357) acc1: 39.0625 (72.4575) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1841 data: 0.0909 max mem: 3531 +Test: [140/389] eta: 0:00:56 acc1_pos: 19.632 acc5_pos: 100.000 loss: 0.8506 (0.5609) acc1: 39.0625 (70.7613) acc5: 100.0000 (100.0000) acc1_neg: 0.521 acc5_neg: 100.000 time: 0.1884 data: 0.0936 max mem: 3531 +Test: [150/389] eta: 0:00:53 acc1_pos: 19.632 acc5_pos: 100.000 loss: 0.9752 (0.6015) acc1: 27.8646 (67.5186) acc5: 100.0000 (100.0000) acc1_neg: 29.167 acc5_neg: 100.000 time: 0.1816 data: 0.0884 max mem: 3531 +Test: [160/389] eta: 0:00:50 acc1_pos: 19.632 acc5_pos: 100.000 loss: 0.9302 (0.6036) acc1: 29.1667 (67.2571) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.1673 data: 0.0760 max mem: 3531 +Test: [170/389] eta: 0:00:47 acc1_pos: 19.632 acc5_pos: 100.000 loss: 0.4095 (0.5929) acc1: 78.9062 (68.2078) acc5: 100.0000 (100.0000) acc1_neg: 45.312 acc5_neg: 100.000 time: 0.1727 data: 0.0825 max mem: 3531 +Test: [180/389] eta: 0:00:44 acc1_pos: 19.632 acc5_pos: 100.000 loss: 0.3004 (0.5922) acc1: 84.8958 (68.1918) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.1774 data: 0.0853 max mem: 3531 +Test: [190/389] eta: 0:00:42 acc1_pos: 19.632 acc5_pos: 100.000 loss: 0.6657 (0.6008) acc1: 53.9062 (67.2707) acc5: 100.0000 (100.0000) acc1_neg: 30.208 acc5_neg: 100.000 time: 0.1874 data: 0.0950 max mem: 3531 +Test: [200/389] eta: 0:00:40 acc1_pos: 19.632 acc5_pos: 100.000 loss: 0.8193 (0.6197) acc1: 37.5000 (65.4216) acc5: 100.0000 (100.0000) acc1_neg: 8.073 acc5_neg: 100.000 time: 0.1949 data: 0.1040 max mem: 3531 +Test: [210/389] eta: 0:00:37 acc1_pos: 19.632 acc5_pos: 100.000 loss: 0.9175 (0.6283) acc1: 33.0729 (64.6204) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1882 data: 0.0979 max mem: 3532 +Test: [220/389] eta: 0:00:35 acc1_pos: 19.632 acc5_pos: 100.000 loss: 0.4436 (0.6150) acc1: 83.3333 (65.9302) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.1762 data: 0.0837 max mem: 3532 +Test: [230/389] eta: 0:00:33 acc1_pos: 19.632 acc5_pos: 100.000 loss: 0.4408 (0.6216) acc1: 85.6771 (65.3578) acc5: 100.0000 (100.0000) acc1_neg: 25.260 acc5_neg: 100.000 time: 0.1707 data: 0.0777 max mem: 3532 +Test: [240/389] eta: 0:00:30 acc1_pos: 19.632 acc5_pos: 100.000 loss: 0.7342 (0.6244) acc1: 54.6875 (65.2155) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1758 data: 0.0853 max mem: 3532 +Test: [250/389] eta: 0:00:28 acc1_pos: 19.632 acc5_pos: 100.000 loss: 0.4778 (0.6159) acc1: 80.7292 (66.0919) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1811 data: 0.0906 max mem: 3532 +Test: [260/389] eta: 0:00:26 acc1_pos: 19.632 acc5_pos: 100.000 loss: 0.4358 (0.6110) acc1: 86.1979 (66.5729) acc5: 100.0000 (100.0000) acc1_neg: 42.969 acc5_neg: 100.000 time: 0.1827 data: 0.0919 max mem: 3532 +Test: [270/389] eta: 0:00:24 acc1_pos: 19.632 acc5_pos: 100.000 loss: 0.5765 (0.6248) acc1: 52.0833 (65.5904) acc5: 100.0000 (100.0000) acc1_neg: 48.438 acc5_neg: 100.000 time: 0.1772 data: 0.0869 max mem: 3532 +Test: [280/389] eta: 0:00:22 acc1_pos: 19.632 acc5_pos: 100.000 loss: 0.7650 (0.6297) acc1: 38.8021 (64.9364) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1729 data: 0.0820 max mem: 3532 +Test: [290/389] eta: 0:00:20 acc1_pos: 19.632 acc5_pos: 100.000 loss: 0.7269 (0.6262) acc1: 55.2083 (65.1946) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1860 data: 0.0948 max mem: 3532 +Test: [300/389] eta: 0:00:17 acc1_pos: 19.632 acc5_pos: 100.000 loss: 0.6399 (0.6253) acc1: 57.2917 (65.1353) acc5: 100.0000 (100.0000) acc1_neg: 51.302 acc5_neg: 100.000 time: 0.1920 data: 0.1014 max mem: 3532 +Test: [310/389] eta: 0:00:15 acc1_pos: 19.632 acc5_pos: 100.000 loss: 0.6887 (0.6293) acc1: 51.3021 (64.6855) acc5: 100.0000 (100.0000) acc1_neg: 11.719 acc5_neg: 100.000 time: 0.1862 data: 0.0938 max mem: 3532 +Test: [320/389] eta: 0:00:13 acc1_pos: 19.632 acc5_pos: 100.000 loss: 0.6568 (0.6283) acc1: 57.5521 (64.6953) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1915 data: 0.0988 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 19.632 acc5_pos: 100.000 loss: 0.8929 (0.6432) acc1: 23.1771 (63.2820) acc5: 100.0000 (100.0000) acc1_neg: 29.427 acc5_neg: 100.000 time: 0.1997 data: 0.1093 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 19.632 acc5_pos: 100.000 loss: 0.8929 (0.6412) acc1: 29.4271 (63.3897) acc5: 100.0000 (100.0000) acc1_neg: 57.292 acc5_neg: 100.000 time: 0.1738 data: 0.0838 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 19.632 acc5_pos: 100.000 loss: 0.6069 (0.6470) acc1: 57.2917 (63.0869) acc5: 100.0000 (100.0000) acc1_neg: 33.594 acc5_neg: 100.000 time: 0.1464 data: 0.0564 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 19.632 acc5_pos: 100.000 loss: 0.9950 (0.6571) acc1: 27.6042 (62.1133) acc5: 100.0000 (100.0000) acc1_neg: 22.656 acc5_neg: 100.000 time: 0.1385 data: 0.0489 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 19.632 acc5_pos: 100.000 loss: 0.9006 (0.6551) acc1: 37.2396 (62.1785) acc5: 100.0000 (100.0000) acc1_neg: 49.479 acc5_neg: 100.000 time: 0.1252 data: 0.0357 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 19.632 acc5_pos: 100.000 loss: 0.7708 (0.6587) acc1: 44.0104 (61.7133) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1137 data: 0.0244 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 19.632 acc5_pos: 100.000 loss: 0.6542 (0.6544) acc1: 57.0312 (62.1205) acc5: 100.0000 (100.0000) acc1_neg: 83.060 acc5_neg: 100.000 time: 0.1073 data: 0.0145 max mem: 3532 +Test: Total time: 0:01:14 (0.1907 s / it) +* Overall Acc@1 62.121 Acc@5 100.000 + target == real Acc@1 74.371 Acc@5 100.000 + target == fake Acc@1 56.426 Acc@5 100.000 + loss 0.654 + PR-AUC 0.8287 ROC-AUC 0.7410 EER 33.16% (thr=0.4163) +Accuracy of the network on the 149358 test images: 62.1% +[rank0]:[W725 17:04:35.544052970 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 17:04:54.394000 2557316 torch/distributed/run.py:766] +W0725 17:04:54.394000 2557316 torch/distributed/run.py:766] ***************************************** +W0725 17:04:54.394000 2557316 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:04:54.394000 2557316 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:05:23.231674495 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +| distributed init (rank 5): env:// +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:05:23.350115518 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:05:23.372816196 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:05:23.645719675 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:05:23.793050650 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:05:23.865939024 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W725 17:05:23.865976334 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 17:05:23.885527813 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed1_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed1_resampled_n116-040-168_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:35:54 acc1_pos: 17.188 acc5_pos: 100.000 loss: 1.4967 (1.4967) acc1: 17.1875 (17.1875) acc5: 100.0000 (100.0000) time: 6.4695 data: 5.0731 max mem: 3531 +Test: [ 10/333] eta: 0:03:36 acc1_pos: 24.479 acc5_pos: 100.000 loss: 1.3625 (1.3803) acc1: 23.4375 (23.3428) acc5: 100.0000 (100.0000) time: 0.6705 data: 0.4617 max mem: 3531 +Test: [ 20/333] eta: 0:02:04 acc1_pos: 19.271 acc5_pos: 100.000 loss: 1.3711 (1.4101) acc1: 19.7917 (21.8006) acc5: 100.0000 (100.0000) time: 0.0948 data: 0.0005 max mem: 3531 +Test: [ 30/333] eta: 0:01:41 acc1_pos: 70.833 acc5_pos: 100.000 loss: 1.1361 (1.2231) acc1: 32.5521 (31.9136) acc5: 100.0000 (100.0000) time: 0.1512 data: 0.0557 max mem: 3531 +Test: [ 40/333] eta: 0:01:30 acc1_pos: 55.469 acc5_pos: 100.000 loss: 0.8033 (1.1168) acc1: 54.6875 (38.0780) acc5: 100.0000 (100.0000) time: 0.2152 data: 0.1239 max mem: 3531 +Test: [ 50/333] eta: 0:01:22 acc1_pos: 51.562 acc5_pos: 100.000 loss: 0.7987 (1.0489) acc1: 55.4688 (41.8045) acc5: 100.0000 (100.0000) time: 0.2207 data: 0.1282 max mem: 3531 +Test: [ 60/333] eta: 0:01:15 acc1_pos: 66.146 acc5_pos: 100.000 loss: 0.7526 (1.0062) acc1: 55.7292 (44.2538) acc5: 100.0000 (100.0000) time: 0.2151 data: 0.1219 max mem: 3531 +Test: [ 70/333] eta: 0:01:10 acc1_pos: 47.917 acc5_pos: 100.000 loss: 0.8067 (0.9817) acc1: 54.6875 (45.6279) acc5: 100.0000 (100.0000) time: 0.2075 data: 0.1136 max mem: 3531 +Test: [ 80/333] eta: 0:01:05 acc1_pos: 61.198 acc5_pos: 100.000 loss: 0.8104 (0.9593) acc1: 53.3854 (46.8589) acc5: 100.0000 (100.0000) time: 0.1945 data: 0.0991 max mem: 3531 +Test: [ 90/333] eta: 0:01:02 acc1_pos: 69.271 acc5_pos: 100.000 loss: 0.7910 (0.9372) acc1: 57.0312 (48.1370) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1353 max mem: 3531 +Test: [100/333] eta: 0:00:59 acc1_pos: 65.885 acc5_pos: 100.000 loss: 0.7452 (0.9120) acc1: 57.5521 (49.4689) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1519 max mem: 3531 +Test: [110/333] eta: 0:00:54 acc1_pos: 61.719 acc5_pos: 100.000 loss: 0.7156 (0.8973) acc1: 58.5938 (50.2276) acc5: 100.0000 (100.0000) time: 0.1815 data: 0.0870 max mem: 3531 +Test: [120/333] eta: 0:00:50 acc1_pos: 58.854 acc5_pos: 100.000 loss: 0.7156 (0.8784) acc1: 59.8958 (51.3946) acc5: 100.0000 (100.0000) time: 0.1457 data: 0.0507 max mem: 3531 +Test: [130/333] eta: 0:00:47 acc1_pos: 82.292 acc5_pos: 100.000 loss: 0.6710 (0.8627) acc1: 61.4583 (52.2881) acc5: 100.0000 (100.0000) time: 0.1790 data: 0.0839 max mem: 3531 +Test: [140/333] eta: 0:00:44 acc1_pos: 78.385 acc5_pos: 100.000 loss: 0.6619 (0.8490) acc1: 61.4583 (53.0641) acc5: 100.0000 (100.0000) time: 0.2043 data: 0.1094 max mem: 3531 +Test: [150/333] eta: 0:00:42 acc1_pos: 60.938 acc5_pos: 100.000 loss: 0.6856 (0.8427) acc1: 60.9375 (53.3578) acc5: 100.0000 (100.0000) time: 0.2141 data: 0.1218 max mem: 3531 +Test: [160/333] eta: 0:00:39 acc1_pos: 56.579 acc5_pos: 100.000 loss: 0.7701 (0.8376) acc1: 56.2500 (53.5747) acc5: 100.0000 (100.0000) acc1_neg: 51.823 acc5_neg: 100.000 time: 0.2097 data: 0.1159 max mem: 3531 +Test: [170/333] eta: 0:00:37 acc1_pos: 56.579 acc5_pos: 100.000 loss: 0.7176 (0.8291) acc1: 57.0312 (54.0296) acc5: 100.0000 (100.0000) acc1_neg: 68.490 acc5_neg: 100.000 time: 0.2049 data: 0.1056 max mem: 3531 +Test: [180/333] eta: 0:00:34 acc1_pos: 56.579 acc5_pos: 100.000 loss: 0.6126 (0.8156) acc1: 65.8854 (54.8112) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.2063 data: 0.1065 max mem: 3531 +Test: [190/333] eta: 0:00:32 acc1_pos: 56.579 acc5_pos: 100.000 loss: 0.5659 (0.7998) acc1: 69.0104 (55.7960) acc5: 100.0000 (100.0000) acc1_neg: 64.844 acc5_neg: 100.000 time: 0.1825 data: 0.0891 max mem: 3531 +Test: [200/333] eta: 0:00:30 acc1_pos: 56.579 acc5_pos: 100.000 loss: 0.5471 (0.7868) acc1: 71.0938 (56.6633) acc5: 100.0000 (100.0000) acc1_neg: 68.750 acc5_neg: 100.000 time: 0.2158 data: 0.1245 max mem: 3531 +Test: [210/333] eta: 0:00:27 acc1_pos: 56.579 acc5_pos: 100.000 loss: 0.5554 (0.7766) acc1: 71.0938 (57.3262) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.2343 data: 0.1412 max mem: 3532 +Test: [220/333] eta: 0:00:25 acc1_pos: 56.579 acc5_pos: 100.000 loss: 0.5750 (0.7711) acc1: 69.0104 (57.5556) acc5: 100.0000 (100.0000) acc1_neg: 69.531 acc5_neg: 100.000 time: 0.1796 data: 0.0843 max mem: 3532 +Test: [230/333] eta: 0:00:22 acc1_pos: 56.579 acc5_pos: 100.000 loss: 0.5568 (0.7594) acc1: 70.0521 (58.3784) acc5: 100.0000 (100.0000) acc1_neg: 76.302 acc5_neg: 100.000 time: 0.1992 data: 0.1043 max mem: 3532 +Test: [240/333] eta: 0:00:20 acc1_pos: 56.579 acc5_pos: 100.000 loss: 0.5488 (0.7567) acc1: 70.5729 (58.5062) acc5: 100.0000 (100.0000) acc1_neg: 61.458 acc5_neg: 100.000 time: 0.1992 data: 0.1032 max mem: 3532 +Test: [250/333] eta: 0:00:17 acc1_pos: 56.579 acc5_pos: 100.000 loss: 0.6957 (0.7536) acc1: 60.9375 (58.6653) acc5: 100.0000 (100.0000) acc1_neg: 65.625 acc5_neg: 100.000 time: 0.1299 data: 0.0317 max mem: 3532 +Test: [260/333] eta: 0:00:15 acc1_pos: 56.579 acc5_pos: 100.000 loss: 0.6139 (0.7464) acc1: 66.6667 (59.1615) acc5: 100.0000 (100.0000) acc1_neg: 70.573 acc5_neg: 100.000 time: 0.1549 data: 0.0573 max mem: 3532 +Test: [270/333] eta: 0:00:13 acc1_pos: 56.579 acc5_pos: 100.000 loss: 0.5198 (0.7370) acc1: 75.2604 (59.8026) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.1978 data: 0.1038 max mem: 3532 +Test: [280/333] eta: 0:00:11 acc1_pos: 56.579 acc5_pos: 100.000 loss: 0.5200 (0.7347) acc1: 72.3958 (59.9181) acc5: 100.0000 (100.0000) acc1_neg: 66.927 acc5_neg: 100.000 time: 0.2129 data: 0.1226 max mem: 3532 +Test: [290/333] eta: 0:00:09 acc1_pos: 56.579 acc5_pos: 100.000 loss: 0.7790 (0.7368) acc1: 53.9062 (59.7097) acc5: 100.0000 (100.0000) acc1_neg: 53.125 acc5_neg: 100.000 time: 0.1944 data: 0.1045 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 56.579 acc5_pos: 100.000 loss: 0.7798 (0.7377) acc1: 53.9062 (59.5619) acc5: 100.0000 (100.0000) acc1_neg: 54.948 acc5_neg: 100.000 time: 0.1555 data: 0.0658 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 56.579 acc5_pos: 100.000 loss: 0.6625 (0.7323) acc1: 60.6771 (59.9335) acc5: 100.0000 (100.0000) acc1_neg: 65.885 acc5_neg: 100.000 time: 0.1488 data: 0.0592 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 56.579 acc5_pos: 100.000 loss: 0.6190 (0.7295) acc1: 65.8854 (60.0703) acc5: 100.0000 (100.0000) acc1_neg: 69.531 acc5_neg: 100.000 time: 0.1393 data: 0.0493 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 56.579 acc5_pos: 100.000 loss: 0.6133 (0.7261) acc1: 66.6667 (60.2931) acc5: 100.0000 (100.0000) acc1_neg: 67.969 acc5_neg: 100.000 time: 0.1292 data: 0.0395 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 56.579 acc5_pos: 100.000 loss: 0.6123 (0.7254) acc1: 67.1875 (60.3412) acc5: 100.0000 (100.0000) acc1_neg: 69.296 acc5_neg: 100.000 time: 0.1237 data: 0.0309 max mem: 3532 +Test: Total time: 0:01:07 (0.2040 s / it) +* Overall Acc@1 60.341 Acc@5 100.000 + target == real Acc@1 53.636 Acc@5 100.000 + target == fake Acc@1 66.310 Acc@5 100.000 + loss 0.725 + PR-AUC 0.6220 ROC-AUC 0.6340 EER 40.41% (thr=0.5669) +Accuracy of the network on the 127843 test images: 60.3% +[rank0]:[W725 17:06:40.429076906 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 17:07:00.218000 2557884 torch/distributed/run.py:766] +W0725 17:07:00.218000 2557884 torch/distributed/run.py:766] ***************************************** +W0725 17:07:00.218000 2557884 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:07:00.218000 2557884 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:07:29.525110769 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:07:29.715120665 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:07:29.944967066 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:07:29.980080688 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:07:29.982820887 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 17:07:29.154310124 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +[rank7]:[W725 17:07:30.194700706 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:07:30.208470712 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed1_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed1_resampled_n116-040-168_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:20:29 acc1_pos: 42.708 acc5_pos: 100.000 loss: 0.9739 (0.9739) acc1: 42.7083 (42.7083) acc5: 100.0000 (100.0000) time: 5.2302 data: 3.1004 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/235] eta: 0:02:05 acc1_pos: 46.875 acc5_pos: 100.000 loss: 0.9739 (0.9585) acc1: 46.3542 (46.7330) acc5: 100.0000 (100.0000) time: 0.5579 data: 0.2822 max mem: 3531 +Test: [ 20/235] eta: 0:01:16 acc1_pos: 46.094 acc5_pos: 100.000 loss: 0.9390 (0.9503) acc1: 46.8750 (47.1106) acc5: 100.0000 (100.0000) time: 0.1106 data: 0.0194 max mem: 3531 +Test: [ 30/235] eta: 0:01:02 acc1_pos: 46.875 acc5_pos: 100.000 loss: 0.9172 (0.9393) acc1: 48.4375 (47.8075) acc5: 100.0000 (100.0000) time: 0.1692 data: 0.0771 max mem: 3531 +Test: [ 40/235] eta: 0:01:01 acc1_pos: 47.396 acc5_pos: 100.000 loss: 0.9178 (0.9369) acc1: 48.4375 (47.9738) acc5: 100.0000 (100.0000) time: 0.2733 data: 0.1789 max mem: 3531 +Test: [ 50/235] eta: 0:00:54 acc1_pos: 46.354 acc5_pos: 100.000 loss: 0.9405 (0.9379) acc1: 48.4375 (47.9677) acc5: 100.0000 (100.0000) time: 0.2751 data: 0.1803 max mem: 3531 +Test: [ 60/235] eta: 0:00:48 acc1_pos: 48.438 acc5_pos: 100.000 loss: 0.9394 (0.9352) acc1: 48.1771 (48.0234) acc5: 100.0000 (100.0000) time: 0.1949 data: 0.0995 max mem: 3531 +Test: [ 70/235] eta: 0:00:43 acc1_pos: 49.231 acc5_pos: 100.000 loss: 0.8829 (0.8753) acc1: 50.2604 (51.8559) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.1769 data: 0.0816 max mem: 3531 +Test: [ 80/235] eta: 0:00:39 acc1_pos: 49.231 acc5_pos: 100.000 loss: 0.4651 (0.8246) acc1: 77.6042 (55.0604) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.1874 data: 0.0953 max mem: 3531 +Test: [ 90/235] eta: 0:00:35 acc1_pos: 49.231 acc5_pos: 100.000 loss: 0.4689 (0.7860) acc1: 77.3438 (57.6093) acc5: 100.0000 (100.0000) acc1_neg: 76.302 acc5_neg: 100.000 time: 0.1873 data: 0.0951 max mem: 3531 +Test: [100/235] eta: 0:00:31 acc1_pos: 49.231 acc5_pos: 100.000 loss: 0.4699 (0.7540) acc1: 78.1250 (59.6354) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.1674 data: 0.0730 max mem: 3531 +Test: [110/235] eta: 0:00:29 acc1_pos: 49.231 acc5_pos: 100.000 loss: 0.4627 (0.7280) acc1: 78.1250 (61.3035) acc5: 100.0000 (100.0000) acc1_neg: 77.344 acc5_neg: 100.000 time: 0.1947 data: 0.0998 max mem: 3531 +Test: [120/235] eta: 0:00:26 acc1_pos: 49.231 acc5_pos: 100.000 loss: 0.4952 (0.7165) acc1: 75.7812 (62.0610) acc5: 100.0000 (100.0000) acc1_neg: 70.052 acc5_neg: 100.000 time: 0.2008 data: 0.1069 max mem: 3531 +Test: [130/235] eta: 0:00:23 acc1_pos: 49.231 acc5_pos: 100.000 loss: 0.6020 (0.7079) acc1: 69.7917 (62.6411) acc5: 100.0000 (100.0000) acc1_neg: 71.094 acc5_neg: 100.000 time: 0.1652 data: 0.0726 max mem: 3531 +Test: [140/235] eta: 0:00:21 acc1_pos: 49.231 acc5_pos: 100.000 loss: 0.6133 (0.7013) acc1: 68.7500 (63.0596) acc5: 100.0000 (100.0000) acc1_neg: 68.229 acc5_neg: 100.000 time: 0.1840 data: 0.0895 max mem: 3531 +Test: [150/235] eta: 0:00:18 acc1_pos: 49.231 acc5_pos: 100.000 loss: 0.6144 (0.6955) acc1: 68.2292 (63.3882) acc5: 100.0000 (100.0000) acc1_neg: 69.271 acc5_neg: 100.000 time: 0.1896 data: 0.0946 max mem: 3531 +Test: [160/235] eta: 0:00:16 acc1_pos: 49.231 acc5_pos: 100.000 loss: 0.6035 (0.6897) acc1: 69.2708 (63.7859) acc5: 100.0000 (100.0000) acc1_neg: 70.312 acc5_neg: 100.000 time: 0.1799 data: 0.0856 max mem: 3531 +Test: [170/235] eta: 0:00:14 acc1_pos: 49.231 acc5_pos: 100.000 loss: 0.5502 (0.6711) acc1: 72.1354 (65.0204) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.1902 data: 0.0944 max mem: 3531 +Test: [180/235] eta: 0:00:11 acc1_pos: 49.231 acc5_pos: 100.000 loss: 0.3361 (0.6525) acc1: 87.2396 (66.2552) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.1733 data: 0.0788 max mem: 3531 +Test: [190/235] eta: 0:00:09 acc1_pos: 49.231 acc5_pos: 100.000 loss: 0.3316 (0.6355) acc1: 87.7604 (67.3907) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1897 data: 0.0973 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 49.231 acc5_pos: 100.000 loss: 0.3411 (0.6213) acc1: 87.2396 (68.3548) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.1969 data: 0.1060 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 49.231 acc5_pos: 100.000 loss: 0.3457 (0.6083) acc1: 86.7188 (69.2104) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.1496 data: 0.0584 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 49.231 acc5_pos: 100.000 loss: 0.3310 (0.5959) acc1: 87.5000 (70.0403) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1346 data: 0.0439 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 49.231 acc5_pos: 100.000 loss: 0.3301 (0.5848) acc1: 87.2396 (70.7533) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1388 data: 0.0482 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 49.231 acc5_pos: 100.000 loss: 0.3297 (0.5807) acc1: 87.2396 (70.9687) acc5: 100.0000 (100.0000) acc1_neg: 86.957 acc5_neg: 100.000 time: 0.1116 data: 0.0230 max mem: 3532 +Test: Total time: 0:00:47 (0.2010 s / it) +* Overall Acc@1 70.969 Acc@5 100.000 + target == real Acc@1 48.038 Acc@5 100.000 + target == fake Acc@1 79.262 Acc@5 100.000 + loss 0.581 + PR-AUC 0.8522 ROC-AUC 0.7054 EER 35.02% (thr=0.6880) +Accuracy of the network on the 89879 test images: 71.0% +[rank0]:[W725 17:08:27.159961638 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 17:08:47.209000 2558450 torch/distributed/run.py:766] +W0725 17:08:47.209000 2558450 torch/distributed/run.py:766] ***************************************** +W0725 17:08:47.209000 2558450 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:08:47.209000 2558450 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:09:18.005040037 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +[rank1]:[W725 17:09:19.212674219 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:09:19.234909618 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:09:19.728684102 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W725 17:09:19.728691586 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 17:09:19.744234112 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:09:19.757710684 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:09:19.769145321 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed1_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed1_resampled_n116-040-168_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:28:32 acc1_pos: 34.375 acc5_pos: 100.000 loss: 0.9582 (0.9582) acc1: 34.3750 (34.3750) acc5: 100.0000 (100.0000) time: 6.3667 data: 4.9898 max mem: 3531 +Test: [ 10/269] eta: 0:02:51 acc1_pos: 53.906 acc5_pos: 100.000 loss: 0.8056 (0.7514) acc1: 52.6042 (56.2500) acc5: 100.0000 (100.0000) time: 0.6629 data: 0.4539 max mem: 3531 +Test: [ 20/269] eta: 0:01:38 acc1_pos: 60.156 acc5_pos: 100.000 loss: 0.7696 (0.6951) acc1: 53.9062 (59.7346) acc5: 100.0000 (100.0000) time: 0.0962 data: 0.0004 max mem: 3531 +Test: [ 30/269] eta: 0:01:20 acc1_pos: 55.469 acc5_pos: 100.000 loss: 0.6301 (0.6683) acc1: 62.2396 (61.8616) acc5: 100.0000 (100.0000) time: 0.1550 data: 0.0591 max mem: 3531 +Test: [ 40/269] eta: 0:01:11 acc1_pos: 56.897 acc5_pos: 100.000 loss: 0.5657 (0.6383) acc1: 69.5312 (64.8628) acc5: 100.0000 (100.0000) acc1_neg: 65.885 acc5_neg: 100.000 time: 0.2258 data: 0.1346 max mem: 3531 +Test: [ 50/269] eta: 0:01:04 acc1_pos: 56.897 acc5_pos: 100.000 loss: 0.5715 (0.6395) acc1: 69.5312 (64.5323) acc5: 100.0000 (100.0000) acc1_neg: 45.052 acc5_neg: 100.000 time: 0.2348 data: 0.1433 max mem: 3531 +Test: [ 60/269] eta: 0:00:59 acc1_pos: 56.897 acc5_pos: 100.000 loss: 0.6970 (0.6552) acc1: 55.4688 (63.1019) acc5: 100.0000 (100.0000) acc1_neg: 45.573 acc5_neg: 100.000 time: 0.2296 data: 0.1368 max mem: 3531 +Test: [ 70/269] eta: 0:00:53 acc1_pos: 56.897 acc5_pos: 100.000 loss: 0.6036 (0.6422) acc1: 63.0208 (64.3633) acc5: 100.0000 (100.0000) acc1_neg: 73.958 acc5_neg: 100.000 time: 0.1999 data: 0.1061 max mem: 3531 +Test: [ 80/269] eta: 0:00:48 acc1_pos: 56.897 acc5_pos: 100.000 loss: 0.5695 (0.6363) acc1: 70.8333 (64.9820) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.1755 data: 0.0825 max mem: 3531 +Test: [ 90/269] eta: 0:00:44 acc1_pos: 56.897 acc5_pos: 100.000 loss: 0.7287 (0.6527) acc1: 54.1667 (63.4930) acc5: 100.0000 (100.0000) acc1_neg: 72.135 acc5_neg: 100.000 time: 0.1849 data: 0.0909 max mem: 3531 +Test: [100/269] eta: 0:00:41 acc1_pos: 56.897 acc5_pos: 100.000 loss: 0.7170 (0.6487) acc1: 63.5417 (64.3513) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.1841 data: 0.0860 max mem: 3531 +Test: [110/269] eta: 0:00:38 acc1_pos: 56.897 acc5_pos: 100.000 loss: 0.5746 (0.6383) acc1: 72.3958 (65.0572) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.1959 data: 0.1008 max mem: 3531 +Test: [120/269] eta: 0:00:34 acc1_pos: 56.897 acc5_pos: 100.000 loss: 0.5798 (0.6512) acc1: 64.8438 (63.9463) acc5: 100.0000 (100.0000) acc1_neg: 22.396 acc5_neg: 100.000 time: 0.1894 data: 0.0992 max mem: 3531 +Test: [130/269] eta: 0:00:31 acc1_pos: 56.897 acc5_pos: 100.000 loss: 0.7377 (0.6608) acc1: 54.4271 (63.1242) acc5: 100.0000 (100.0000) acc1_neg: 54.948 acc5_neg: 100.000 time: 0.1716 data: 0.0803 max mem: 3531 +Test: [140/269] eta: 0:00:29 acc1_pos: 56.897 acc5_pos: 100.000 loss: 0.7556 (0.6864) acc1: 49.7396 (61.4214) acc5: 100.0000 (100.0000) acc1_neg: 24.740 acc5_neg: 100.000 time: 0.1819 data: 0.0883 max mem: 3531 +Test: [150/269] eta: 0:00:26 acc1_pos: 56.897 acc5_pos: 100.000 loss: 1.1440 (0.7179) acc1: 29.9479 (59.2336) acc5: 100.0000 (100.0000) acc1_neg: 26.042 acc5_neg: 100.000 time: 0.1824 data: 0.0902 max mem: 3531 +Test: [160/269] eta: 0:00:24 acc1_pos: 56.897 acc5_pos: 100.000 loss: 1.0194 (0.7255) acc1: 31.2500 (58.6714) acc5: 100.0000 (100.0000) acc1_neg: 76.042 acc5_neg: 100.000 time: 0.1750 data: 0.0843 max mem: 3531 +Test: [170/269] eta: 0:00:21 acc1_pos: 56.897 acc5_pos: 100.000 loss: 0.6380 (0.7200) acc1: 64.8438 (59.1481) acc5: 100.0000 (100.0000) acc1_neg: 64.844 acc5_neg: 100.000 time: 0.1965 data: 0.1052 max mem: 3531 +Test: [180/269] eta: 0:00:19 acc1_pos: 56.897 acc5_pos: 100.000 loss: 0.5957 (0.7084) acc1: 69.2708 (60.1246) acc5: 100.0000 (100.0000) acc1_neg: 65.625 acc5_neg: 100.000 time: 0.2080 data: 0.1171 max mem: 3531 +Test: [190/269] eta: 0:00:17 acc1_pos: 56.897 acc5_pos: 100.000 loss: 0.5774 (0.7056) acc1: 67.7083 (60.2312) acc5: 100.0000 (100.0000) acc1_neg: 59.375 acc5_neg: 100.000 time: 0.1932 data: 0.1024 max mem: 3531 +Test: [200/269] eta: 0:00:14 acc1_pos: 56.897 acc5_pos: 100.000 loss: 0.6968 (0.7076) acc1: 58.8542 (60.1057) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1775 data: 0.0861 max mem: 3531 +Test: [210/269] eta: 0:00:12 acc1_pos: 56.897 acc5_pos: 100.000 loss: 0.8074 (0.7130) acc1: 49.7396 (59.6650) acc5: 100.0000 (100.0000) acc1_neg: 60.156 acc5_neg: 100.000 time: 0.1701 data: 0.0757 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 56.897 acc5_pos: 100.000 loss: 0.8620 (0.7211) acc1: 42.9688 (58.9897) acc5: 100.0000 (100.0000) acc1_neg: 55.990 acc5_neg: 100.000 time: 0.1733 data: 0.0787 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 56.897 acc5_pos: 100.000 loss: 0.7937 (0.7235) acc1: 52.3438 (58.8497) acc5: 100.0000 (100.0000) acc1_neg: 69.010 acc5_neg: 100.000 time: 0.1821 data: 0.0877 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 56.897 acc5_pos: 100.000 loss: 0.6925 (0.7215) acc1: 59.3750 (59.0573) acc5: 100.0000 (100.0000) acc1_neg: 69.531 acc5_neg: 100.000 time: 0.1733 data: 0.0800 max mem: 3532 +Test: [250/269] eta: 0:00:03 acc1_pos: 56.897 acc5_pos: 100.000 loss: 0.8071 (0.7281) acc1: 54.1667 (58.5191) acc5: 100.0000 (100.0000) acc1_neg: 28.385 acc5_neg: 100.000 time: 0.1440 data: 0.0529 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 56.897 acc5_pos: 100.000 loss: 0.8315 (0.7338) acc1: 47.9167 (58.0829) acc5: 100.0000 (100.0000) acc1_neg: 60.677 acc5_neg: 100.000 time: 0.1320 data: 0.0409 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 56.897 acc5_pos: 100.000 loss: 0.7870 (0.7314) acc1: 53.6458 (58.2489) acc5: 100.0000 (100.0000) acc1_neg: 74.278 acc5_neg: 100.000 time: 0.1243 data: 0.0307 max mem: 3532 +Test: Total time: 0:00:53 (0.2004 s / it) +* Overall Acc@1 58.249 Acc@5 100.000 + target == real Acc@1 61.400 Acc@5 100.000 + target == fake Acc@1 57.819 Acc@5 100.000 + loss 0.731 + PR-AUC 0.9140 ROC-AUC 0.6315 EER 40.18% (thr=0.4839) +Accuracy of the network on the 103293 test images: 58.2% +[rank0]:[W725 17:10:22.548429535 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-25T17:10:29Z] eval_done tag=fixed500k_real_n5_seed1_resampled +[2026-07-25T17:10:29Z] eval_skip tag=fixed500k_fake_n11_seed1_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed1_resampled +[2026-07-25T17:10:29Z] eval_start tag=fixed500k_real_n5_seed2_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed2_resampled_n116-040-168_20260725_112559/best_checkpoint.pth +started_utc=2026-07-25T17:10:29Z +hostname=n116-040-168 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed2_resampled_n116-040-168_20260725_112559/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed2_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0725 17:10:43.802000 2559036 torch/distributed/run.py:766] +W0725 17:10:43.802000 2559036 torch/distributed/run.py:766] ***************************************** +W0725 17:10:43.802000 2559036 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:10:43.802000 2559036 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:11:18.399416345 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 17:11:18.425311876 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:11:18.476177342 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:11:18.125578798 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:11:19.316631038 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:11:19.379221798 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:11:19.452675613 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W725 17:11:19.452671341 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed2_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed2_resampled_n116-040-168_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:23:11 acc1_pos: 26.042 acc5_pos: 100.000 loss: 1.2666 (1.2666) acc1: 26.0417 (26.0417) acc5: 100.0000 (100.0000) time: 5.9486 data: 4.5521 max mem: 3531 +Test: [ 10/234] eta: 0:02:21 acc1_pos: 23.958 acc5_pos: 100.000 loss: 1.2701 (1.2807) acc1: 23.9583 (25.4972) acc5: 100.0000 (100.0000) time: 0.6315 data: 0.4142 max mem: 3531 +Test: [ 20/234] eta: 0:01:21 acc1_pos: 77.865 acc5_pos: 100.000 loss: 1.1898 (1.0672) acc1: 32.8125 (41.1210) acc5: 100.0000 (100.0000) time: 0.1000 data: 0.0035 max mem: 3531 +Test: [ 30/234] eta: 0:01:14 acc1_pos: 42.969 acc5_pos: 100.000 loss: 0.6089 (0.9422) acc1: 67.4479 (49.1599) acc5: 100.0000 (100.0000) time: 0.2199 data: 0.1266 max mem: 3531 +Test: [ 40/234] eta: 0:01:04 acc1_pos: 75.000 acc5_pos: 100.000 loss: 0.6008 (0.9129) acc1: 65.1042 (50.4954) acc5: 100.0000 (100.0000) time: 0.2847 data: 0.1912 max mem: 3531 +Test: [ 50/234] eta: 0:00:55 acc1_pos: 71.615 acc5_pos: 100.000 loss: 0.5968 (0.8448) acc1: 66.9271 (54.2535) acc5: 100.0000 (100.0000) time: 0.2000 data: 0.1072 max mem: 3531 +Test: [ 60/234] eta: 0:00:48 acc1_pos: 64.138 acc5_pos: 100.000 loss: 0.3002 (0.7262) acc1: 77.6042 (61.5181) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1767 data: 0.0850 max mem: 3531 +Test: [ 70/234] eta: 0:00:43 acc1_pos: 64.138 acc5_pos: 100.000 loss: 0.1079 (0.6608) acc1: 99.7396 (65.7314) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1803 data: 0.0884 max mem: 3531 +Test: [ 80/234] eta: 0:00:40 acc1_pos: 64.138 acc5_pos: 100.000 loss: 0.3302 (0.6246) acc1: 86.9792 (68.1745) acc5: 100.0000 (100.0000) acc1_neg: 77.083 acc5_neg: 100.000 time: 0.2031 data: 0.1102 max mem: 3531 +Test: [ 90/234] eta: 0:00:37 acc1_pos: 64.138 acc5_pos: 100.000 loss: 0.3225 (0.5848) acc1: 88.5417 (70.7904) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2304 data: 0.1377 max mem: 3531 +Test: [100/234] eta: 0:00:33 acc1_pos: 64.138 acc5_pos: 100.000 loss: 0.2910 (0.5603) acc1: 90.1042 (72.3159) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.2017 data: 0.1107 max mem: 3531 +Test: [110/234] eta: 0:00:31 acc1_pos: 64.138 acc5_pos: 100.000 loss: 0.3422 (0.5457) acc1: 86.4583 (73.2686) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.2109 data: 0.1186 max mem: 3531 +Test: [120/234] eta: 0:00:27 acc1_pos: 64.138 acc5_pos: 100.000 loss: 0.3890 (0.5388) acc1: 83.3333 (73.7431) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.2050 data: 0.1130 max mem: 3531 +Test: [130/234] eta: 0:00:24 acc1_pos: 64.138 acc5_pos: 100.000 loss: 0.4145 (0.5253) acc1: 80.7292 (74.5766) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1754 data: 0.0851 max mem: 3531 +Test: [140/234] eta: 0:00:22 acc1_pos: 64.138 acc5_pos: 100.000 loss: 0.4209 (0.5198) acc1: 79.4271 (74.9391) acc5: 100.0000 (100.0000) acc1_neg: 75.521 acc5_neg: 100.000 time: 0.1983 data: 0.1083 max mem: 3531 +Test: [150/234] eta: 0:00:19 acc1_pos: 64.138 acc5_pos: 100.000 loss: 0.4497 (0.5117) acc1: 79.1667 (75.4777) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.2142 data: 0.1243 max mem: 3531 +Test: [160/234] eta: 0:00:17 acc1_pos: 64.138 acc5_pos: 100.000 loss: 0.4177 (0.5096) acc1: 81.7708 (75.6033) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.2169 data: 0.1264 max mem: 3531 +Test: [170/234] eta: 0:00:14 acc1_pos: 64.138 acc5_pos: 100.000 loss: 0.4278 (0.5036) acc1: 81.5104 (76.0401) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.2189 data: 0.1287 max mem: 3531 +Test: [180/234] eta: 0:00:12 acc1_pos: 64.138 acc5_pos: 100.000 loss: 0.4080 (0.4965) acc1: 82.8125 (76.4647) acc5: 100.0000 (100.0000) acc1_neg: 72.396 acc5_neg: 100.000 time: 0.2026 data: 0.1121 max mem: 3531 +Test: [190/234] eta: 0:00:09 acc1_pos: 64.138 acc5_pos: 100.000 loss: 0.4899 (0.4976) acc1: 76.0417 (76.3580) acc5: 100.0000 (100.0000) acc1_neg: 76.042 acc5_neg: 100.000 time: 0.1695 data: 0.0790 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 64.138 acc5_pos: 100.000 loss: 0.5041 (0.4861) acc1: 75.5208 (77.0419) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1445 data: 0.0546 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 64.138 acc5_pos: 100.000 loss: 0.1144 (0.4695) acc1: 99.4792 (78.0954) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1311 data: 0.0410 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 64.138 acc5_pos: 100.000 loss: 0.1865 (0.4575) acc1: 98.4375 (78.9404) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1290 data: 0.0392 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 64.138 acc5_pos: 100.000 loss: 0.2524 (0.4524) acc1: 89.5833 (79.2715) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.1194 data: 0.0300 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 64.138 acc5_pos: 100.000 loss: 0.3266 (0.4511) acc1: 87.7604 (79.3333) acc5: 100.0000 (100.0000) acc1_neg: 82.101 acc5_neg: 100.000 time: 0.1201 data: 0.0297 max mem: 3532 +Test: Total time: 0:00:48 (0.2091 s / it) +* Overall Acc@1 79.333 Acc@5 100.000 + target == real Acc@1 54.326 Acc@5 100.000 + target == fake Acc@1 86.381 Acc@5 100.000 + loss 0.451 + PR-AUC 0.9229 ROC-AUC 0.7967 EER 27.54% (thr=0.6973) +Accuracy of the network on the 89729 test images: 79.3% +[rank0]:[W725 17:12:15.448998076 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 17:12:35.569000 2559628 torch/distributed/run.py:766] +W0725 17:12:35.569000 2559628 torch/distributed/run.py:766] ***************************************** +W0725 17:12:35.569000 2559628 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:12:35.569000 2559628 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W725 17:13:09.635710970 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:13:09.943107243 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:13:09.097460208 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:13:10.187266384 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:13:10.385957889 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:13:10.556126630 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W725 17:13:10.560630788 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:13:10.604378125 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed2_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed2_resampled_n116-040-168_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:11:45 acc1_pos: 33.333 acc5_pos: 100.000 loss: 1.0874 (1.0874) acc1: 33.3333 (33.3333) acc5: 100.0000 (100.0000) time: 6.3282 data: 4.4912 max mem: 3531 +Test: [ 10/2387] eta: 0:26:04 acc1_pos: 30.990 acc5_pos: 100.000 loss: 1.1139 (1.1487) acc1: 30.9896 (31.6525) acc5: 100.0000 (100.0000) time: 0.6582 data: 0.4087 max mem: 3531 +Test: [ 20/2387] eta: 0:15:28 acc1_pos: 18.229 acc5_pos: 100.000 loss: 1.1702 (1.1743) acc1: 29.6875 (32.2793) acc5: 100.0000 (100.0000) time: 0.0955 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:14:00 acc1_pos: 14.844 acc5_pos: 100.000 loss: 1.2723 (1.2155) acc1: 25.0000 (30.0823) acc5: 100.0000 (100.0000) time: 0.1908 data: 0.0949 max mem: 3531 +Test: [ 40/2387] eta: 0:13:35 acc1_pos: 18.229 acc5_pos: 100.000 loss: 1.3440 (1.2362) acc1: 18.7500 (28.3791) acc5: 100.0000 (100.0000) time: 0.3005 data: 0.2087 max mem: 3531 +Test: [ 50/2387] eta: 0:12:42 acc1_pos: 22.917 acc5_pos: 100.000 loss: 1.1859 (1.2098) acc1: 22.9167 (29.6211) acc5: 100.0000 (100.0000) time: 0.2791 data: 0.1883 max mem: 3531 +Test: [ 60/2387] eta: 0:11:59 acc1_pos: 30.729 acc5_pos: 100.000 loss: 1.3162 (1.2511) acc1: 26.5625 (28.4708) acc5: 100.0000 (100.0000) time: 0.2304 data: 0.1379 max mem: 3531 +Test: [ 70/2387] eta: 0:11:35 acc1_pos: 27.344 acc5_pos: 100.000 loss: 1.5571 (1.3049) acc1: 19.7917 (27.1347) acc5: 100.0000 (100.0000) time: 0.2332 data: 0.1373 max mem: 3531 +Test: [ 80/2387] eta: 0:11:15 acc1_pos: 12.760 acc5_pos: 100.000 loss: 1.6813 (1.3522) acc1: 14.5833 (25.1575) acc5: 100.0000 (100.0000) time: 0.2424 data: 0.1449 max mem: 3531 +Test: [ 90/2387] eta: 0:10:51 acc1_pos: 10.156 acc5_pos: 100.000 loss: 1.7100 (1.3988) acc1: 7.8125 (23.0540) acc5: 100.0000 (100.0000) time: 0.2250 data: 0.1299 max mem: 3531 +Test: [ 100/2387] eta: 0:10:38 acc1_pos: 8.073 acc5_pos: 100.000 loss: 1.7949 (1.4355) acc1: 5.7292 (21.3928) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1319 max mem: 3531 +Test: [ 110/2387] eta: 0:10:26 acc1_pos: 15.365 acc5_pos: 100.000 loss: 1.7147 (1.4558) acc1: 7.8125 (20.3665) acc5: 100.0000 (100.0000) time: 0.2360 data: 0.1408 max mem: 3531 +Test: [ 120/2387] eta: 0:10:12 acc1_pos: 4.688 acc5_pos: 100.000 loss: 1.6793 (1.4718) acc1: 10.9375 (19.4925) acc5: 100.0000 (100.0000) time: 0.2258 data: 0.1325 max mem: 3531 +Test: [ 130/2387] eta: 0:10:06 acc1_pos: 10.417 acc5_pos: 100.000 loss: 1.6844 (1.4923) acc1: 10.1562 (18.5433) acc5: 100.0000 (100.0000) time: 0.2350 data: 0.1442 max mem: 3531 +Test: [ 140/2387] eta: 0:10:02 acc1_pos: 20.833 acc5_pos: 100.000 loss: 1.6945 (1.5066) acc1: 6.7708 (17.8284) acc5: 100.0000 (100.0000) time: 0.2540 data: 0.1599 max mem: 3531 +Test: [ 150/2387] eta: 0:09:56 acc1_pos: 11.719 acc5_pos: 100.000 loss: 1.6255 (1.5101) acc1: 11.4583 (17.7532) acc5: 100.0000 (100.0000) time: 0.2508 data: 0.1544 max mem: 3531 +Test: [ 160/2387] eta: 0:09:39 acc1_pos: 32.420 acc5_pos: 100.000 loss: 1.5407 (1.4854) acc1: 13.8021 (19.2611) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.2061 data: 0.1116 max mem: 3531 +Test: [ 170/2387] eta: 0:09:31 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.3984 (1.4213) acc1: 81.2500 (22.9365) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1928 data: 0.0963 max mem: 3531 +Test: [ 180/2387] eta: 0:09:17 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.3881 (1.3637) acc1: 82.5521 (26.2935) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.1892 data: 0.0924 max mem: 3531 +Test: [ 190/2387] eta: 0:09:09 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.3757 (1.3119) acc1: 83.3333 (29.3071) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.1856 data: 0.0927 max mem: 3531 +Test: [ 200/2387] eta: 0:08:59 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.3768 (1.2654) acc1: 83.3333 (32.0118) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.1945 data: 0.1019 max mem: 3531 +Test: [ 210/2387] eta: 0:08:53 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.3768 (1.2231) acc1: 83.8542 (34.4787) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1943 data: 0.0994 max mem: 3532 +Test: [ 220/2387] eta: 0:08:48 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.3694 (1.1847) acc1: 83.8542 (36.7199) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.2149 data: 0.1183 max mem: 3532 +Test: [ 230/2387] eta: 0:08:40 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.3799 (1.1498) acc1: 83.8542 (38.7514) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.1999 data: 0.1037 max mem: 3532 +Test: [ 240/2387] eta: 0:08:30 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.3530 (1.1152) acc1: 84.8958 (40.8011) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1704 data: 0.0757 max mem: 3532 +Test: [ 250/2387] eta: 0:08:24 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2972 (1.0777) acc1: 89.5833 (43.0103) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1799 data: 0.0868 max mem: 3532 +Test: [ 260/2387] eta: 0:08:18 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1454 (1.0420) acc1: 98.4375 (45.0980) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1945 data: 0.1006 max mem: 3532 +Test: [ 270/2387] eta: 0:08:11 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1315 (1.0089) acc1: 98.4375 (47.0509) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1799 data: 0.0857 max mem: 3532 +Test: [ 280/2387] eta: 0:08:05 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1364 (0.9783) acc1: 97.9167 (48.8360) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1749 data: 0.0804 max mem: 3532 +Test: [ 290/2387] eta: 0:07:59 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1281 (0.9489) acc1: 98.4375 (50.5566) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1800 data: 0.0834 max mem: 3532 +Test: [ 300/2387] eta: 0:07:54 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1169 (0.9211) acc1: 99.2188 (52.1603) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1903 data: 0.0947 max mem: 3532 +Test: [ 310/2387] eta: 0:07:50 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1169 (0.9063) acc1: 99.4792 (52.8470) acc5: 100.0000 (100.0000) acc1_neg: 42.969 acc5_neg: 100.000 time: 0.1998 data: 0.1048 max mem: 3532 +Test: [ 320/2387] eta: 0:07:44 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.8961 (0.9070) acc1: 38.2812 (52.3486) acc5: 100.0000 (100.0000) acc1_neg: 39.323 acc5_neg: 100.000 time: 0.1845 data: 0.0889 max mem: 3532 +Test: [ 330/2387] eta: 0:07:39 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.9302 (0.9074) acc1: 37.7604 (51.9228) acc5: 100.0000 (100.0000) acc1_neg: 42.448 acc5_neg: 100.000 time: 0.1749 data: 0.0796 max mem: 3532 +Test: [ 340/2387] eta: 0:07:36 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.9289 (0.9082) acc1: 37.7604 (51.4930) acc5: 100.0000 (100.0000) acc1_neg: 34.375 acc5_neg: 100.000 time: 0.1899 data: 0.0931 max mem: 3532 +Test: [ 350/2387] eta: 0:07:32 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.8748 (0.8925) acc1: 40.6250 (52.5567) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1999 data: 0.1039 max mem: 3532 +Test: [ 360/2387] eta: 0:07:26 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2894 (0.8759) acc1: 91.1458 (53.6040) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1799 data: 0.0847 max mem: 3532 +Test: [ 370/2387] eta: 0:07:19 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2917 (0.8605) acc1: 89.3229 (54.5583) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1403 data: 0.0455 max mem: 3532 +Test: [ 380/2387] eta: 0:07:16 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2968 (0.8457) acc1: 89.3229 (55.4783) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1599 data: 0.0638 max mem: 3532 +Test: [ 390/2387] eta: 0:07:11 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2968 (0.8317) acc1: 90.1042 (56.3552) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1895 data: 0.0906 max mem: 3532 +Test: [ 400/2387] eta: 0:07:08 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2987 (0.8185) acc1: 90.1042 (57.1800) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1871 data: 0.0907 max mem: 3532 +Test: [ 410/2387] eta: 0:07:03 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2918 (0.8056) acc1: 90.1042 (57.9810) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1766 data: 0.0839 max mem: 3532 +Test: [ 420/2387] eta: 0:06:58 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2826 (0.7933) acc1: 90.1042 (58.7385) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1490 data: 0.0578 max mem: 3532 +Test: [ 430/2387] eta: 0:06:54 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2799 (0.7809) acc1: 90.6250 (59.5206) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1594 data: 0.0671 max mem: 3532 +Test: [ 440/2387] eta: 0:06:51 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2295 (0.7664) acc1: 93.7500 (60.3842) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1822 data: 0.0883 max mem: 3533 +Test: [ 450/2387] eta: 0:06:46 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.0873 (0.7513) acc1: 100.0000 (61.2626) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1715 data: 0.0783 max mem: 3533 +Test: [ 460/2387] eta: 0:06:43 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.0869 (0.7369) acc1: 100.0000 (62.1023) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1689 data: 0.0769 max mem: 3533 +Test: [ 470/2387] eta: 0:06:39 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.0863 (0.7231) acc1: 100.0000 (62.9069) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1740 data: 0.0824 max mem: 3533 +Test: [ 480/2387] eta: 0:06:36 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.0870 (0.7099) acc1: 100.0000 (63.6776) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1682 data: 0.0762 max mem: 3533 +Test: [ 490/2387] eta: 0:06:32 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.0875 (0.6972) acc1: 100.0000 (64.4173) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1678 data: 0.0751 max mem: 3533 +Test: [ 500/2387] eta: 0:06:28 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.0871 (0.6850) acc1: 100.0000 (65.1276) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1646 data: 0.0696 max mem: 3533 +Test: [ 510/2387] eta: 0:06:24 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.0865 (0.6734) acc1: 100.0000 (65.8100) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1622 data: 0.0676 max mem: 3533 +Test: [ 520/2387] eta: 0:06:21 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1072 (0.6626) acc1: 100.0000 (66.4652) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1623 data: 0.0697 max mem: 3533 +Test: [ 530/2387] eta: 0:06:17 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1096 (0.6522) acc1: 100.0000 (67.0943) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1627 data: 0.0704 max mem: 3533 +Test: [ 540/2387] eta: 0:06:14 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1066 (0.6421) acc1: 100.0000 (67.7002) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1628 data: 0.0697 max mem: 3533 +Test: [ 550/2387] eta: 0:06:11 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1067 (0.6324) acc1: 100.0000 (68.2849) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1626 data: 0.0700 max mem: 3533 +Test: [ 560/2387] eta: 0:06:07 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1067 (0.6230) acc1: 100.0000 (68.8484) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1642 data: 0.0730 max mem: 3533 +Test: [ 570/2387] eta: 0:06:04 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1050 (0.6139) acc1: 100.0000 (69.3917) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1623 data: 0.0693 max mem: 3533 +Test: [ 580/2387] eta: 0:06:01 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1050 (0.6052) acc1: 100.0000 (69.9158) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1598 data: 0.0665 max mem: 3533 +Test: [ 590/2387] eta: 0:05:58 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1056 (0.5967) acc1: 100.0000 (70.4240) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1603 data: 0.0678 max mem: 3533 +Test: [ 600/2387] eta: 0:05:55 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1076 (0.5887) acc1: 100.0000 (70.9148) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1744 data: 0.0809 max mem: 3533 +Test: [ 610/2387] eta: 0:05:52 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1449 (0.5827) acc1: 97.1354 (71.3145) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1644 data: 0.0663 max mem: 3533 +Test: [ 620/2387] eta: 0:05:49 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2172 (0.5768) acc1: 95.3125 (71.7056) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1601 data: 0.0620 max mem: 3533 +Test: [ 630/2387] eta: 0:05:47 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2179 (0.5712) acc1: 95.3125 (72.0764) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1849 data: 0.0922 max mem: 3533 +Test: [ 640/2387] eta: 0:05:44 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2215 (0.5658) acc1: 95.0521 (72.4373) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1816 data: 0.0871 max mem: 3533 +Test: [ 650/2387] eta: 0:05:42 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2221 (0.5606) acc1: 95.5729 (72.7975) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1749 data: 0.0764 max mem: 3533 +Test: [ 660/2387] eta: 0:05:39 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2218 (0.5555) acc1: 96.0938 (73.1412) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1735 data: 0.0787 max mem: 3533 +Test: [ 670/2387] eta: 0:05:37 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2153 (0.5504) acc1: 96.0938 (73.4821) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1749 data: 0.0802 max mem: 3534 +Test: [ 680/2387] eta: 0:05:34 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2116 (0.5450) acc1: 96.6146 (73.8474) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1793 data: 0.0815 max mem: 3534 +Test: [ 690/2387] eta: 0:05:32 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1743 (0.5397) acc1: 96.8750 (74.1780) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1799 data: 0.0853 max mem: 3534 +Test: [ 700/2387] eta: 0:05:29 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1787 (0.5345) acc1: 95.8333 (74.4784) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1764 data: 0.0845 max mem: 3534 +Test: [ 710/2387] eta: 0:05:27 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1718 (0.5294) acc1: 95.5729 (74.7813) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1763 data: 0.0834 max mem: 3534 +Test: [ 720/2387] eta: 0:05:25 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1695 (0.5245) acc1: 95.8333 (75.0704) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1834 data: 0.0902 max mem: 3534 +Test: [ 730/2387] eta: 0:05:23 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1762 (0.5197) acc1: 95.5729 (75.3481) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1837 data: 0.0920 max mem: 3534 +Test: [ 740/2387] eta: 0:05:20 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1742 (0.5151) acc1: 95.5729 (75.6224) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1673 data: 0.0755 max mem: 3534 +Test: [ 750/2387] eta: 0:05:18 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1720 (0.5105) acc1: 95.3125 (75.8874) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1712 data: 0.0794 max mem: 3534 +Test: [ 760/2387] eta: 0:05:16 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1696 (0.5058) acc1: 95.8333 (76.1625) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1982 data: 0.1023 max mem: 3534 +Test: [ 770/2387] eta: 0:05:14 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1173 (0.5011) acc1: 97.1354 (76.4392) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1984 data: 0.1019 max mem: 3534 +Test: [ 780/2387] eta: 0:05:12 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1801 (0.4978) acc1: 93.4896 (76.6389) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1922 data: 0.1005 max mem: 3534 +Test: [ 790/2387] eta: 0:05:10 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2239 (0.4942) acc1: 91.9271 (76.8460) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1898 data: 0.0956 max mem: 3534 +Test: [ 800/2387] eta: 0:05:08 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1820 (0.4905) acc1: 97.6562 (77.0934) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1829 data: 0.0885 max mem: 3534 +Test: [ 810/2387] eta: 0:05:06 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1826 (0.4868) acc1: 97.1354 (77.3357) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1855 data: 0.0940 max mem: 3534 +Test: [ 820/2387] eta: 0:05:04 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1764 (0.4829) acc1: 96.6146 (77.5680) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1989 data: 0.1069 max mem: 3534 +Test: [ 830/2387] eta: 0:05:02 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1474 (0.4794) acc1: 96.3542 (77.7828) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2059 data: 0.1136 max mem: 3534 +Test: [ 840/2387] eta: 0:05:00 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.3189 (0.4793) acc1: 86.9792 (77.7819) acc5: 100.0000 (100.0000) acc1_neg: 70.833 acc5_neg: 100.000 time: 0.1927 data: 0.1012 max mem: 3534 +Test: [ 850/2387] eta: 0:04:59 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.4429 (0.4792) acc1: 77.6042 (77.7761) acc5: 100.0000 (100.0000) acc1_neg: 74.219 acc5_neg: 100.000 time: 0.2083 data: 0.1166 max mem: 3534 +Test: [ 860/2387] eta: 0:04:57 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.4365 (0.4789) acc1: 77.6042 (77.7869) acc5: 100.0000 (100.0000) acc1_neg: 75.521 acc5_neg: 100.000 time: 0.2085 data: 0.1159 max mem: 3534 +Test: [ 870/2387] eta: 0:04:54 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.4365 (0.4789) acc1: 77.0833 (77.7638) acc5: 100.0000 (100.0000) acc1_neg: 77.083 acc5_neg: 100.000 time: 0.1806 data: 0.0878 max mem: 3534 +Test: [ 880/2387] eta: 0:04:52 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.4818 (0.4798) acc1: 75.0000 (77.6931) acc5: 100.0000 (100.0000) acc1_neg: 65.885 acc5_neg: 100.000 time: 0.1589 data: 0.0655 max mem: 3534 +Test: [ 890/2387] eta: 0:04:49 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.4717 (0.4788) acc1: 77.0833 (77.7567) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1592 data: 0.0665 max mem: 3534 +Test: [ 900/2387] eta: 0:04:47 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.4717 (0.4793) acc1: 75.2604 (77.7100) acc5: 100.0000 (100.0000) acc1_neg: 69.271 acc5_neg: 100.000 time: 0.1867 data: 0.0958 max mem: 3535 +Test: [ 910/2387] eta: 0:04:45 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.5569 (0.4802) acc1: 71.8750 (77.6402) acc5: 100.0000 (100.0000) acc1_neg: 69.271 acc5_neg: 100.000 time: 0.1904 data: 0.1000 max mem: 3535 +Test: [ 920/2387] eta: 0:04:43 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.5606 (0.4811) acc1: 70.5729 (77.5620) acc5: 100.0000 (100.0000) acc1_neg: 67.708 acc5_neg: 100.000 time: 0.1676 data: 0.0759 max mem: 3535 +Test: [ 930/2387] eta: 0:04:40 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.5590 (0.4819) acc1: 70.0521 (77.4895) acc5: 100.0000 (100.0000) acc1_neg: 68.750 acc5_neg: 100.000 time: 0.1512 data: 0.0578 max mem: 3535 +Test: [ 940/2387] eta: 0:04:38 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.5642 (0.4827) acc1: 70.0521 (77.4091) acc5: 100.0000 (100.0000) acc1_neg: 69.010 acc5_neg: 100.000 time: 0.1602 data: 0.0665 max mem: 3535 +Test: [ 950/2387] eta: 0:04:36 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.5573 (0.4835) acc1: 69.7917 (77.3347) acc5: 100.0000 (100.0000) acc1_neg: 69.792 acc5_neg: 100.000 time: 0.1888 data: 0.0966 max mem: 3535 +Test: [ 960/2387] eta: 0:04:34 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.5532 (0.4842) acc1: 71.3542 (77.2776) acc5: 100.0000 (100.0000) acc1_neg: 74.479 acc5_neg: 100.000 time: 0.1826 data: 0.0908 max mem: 3535 +Test: [ 970/2387] eta: 0:04:32 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.5535 (0.4850) acc1: 70.8333 (77.2062) acc5: 100.0000 (100.0000) acc1_neg: 72.135 acc5_neg: 100.000 time: 0.1754 data: 0.0819 max mem: 3535 +Test: [ 980/2387] eta: 0:04:30 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.5400 (0.4843) acc1: 72.1354 (77.2378) acc5: 100.0000 (100.0000) acc1_neg: 60.677 acc5_neg: 100.000 time: 0.2076 data: 0.1142 max mem: 3535 +Test: [ 990/2387] eta: 0:04:28 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.3080 (0.4823) acc1: 89.0625 (77.3598) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1987 data: 0.1050 max mem: 3535 +Test: [1000/2387] eta: 0:04:25 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1766 (0.4795) acc1: 98.4375 (77.5488) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1492 data: 0.0553 max mem: 3535 +Test: [1010/2387] eta: 0:04:24 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1707 (0.4764) acc1: 97.6562 (77.7433) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1757 data: 0.0842 max mem: 3535 +Test: [1020/2387] eta: 0:04:22 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1645 (0.4738) acc1: 97.3958 (77.9072) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.2079 data: 0.1173 max mem: 3535 +Test: [1030/2387] eta: 0:04:20 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2412 (0.4718) acc1: 93.7500 (78.0452) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1891 data: 0.0954 max mem: 3535 +Test: [1040/2387] eta: 0:04:18 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2412 (0.4694) acc1: 94.2708 (78.2056) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1931 data: 0.0997 max mem: 3535 +Test: [1050/2387] eta: 0:04:17 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2245 (0.4671) acc1: 94.7917 (78.3606) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2099 data: 0.1175 max mem: 3535 +Test: [1060/2387] eta: 0:04:15 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2145 (0.4648) acc1: 95.3125 (78.5216) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2047 data: 0.1112 max mem: 3535 +Test: [1070/2387] eta: 0:04:13 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2042 (0.4624) acc1: 96.3542 (78.6716) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1872 data: 0.0949 max mem: 3535 +Test: [1080/2387] eta: 0:04:11 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1660 (0.4599) acc1: 94.7917 (78.8166) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1988 data: 0.1039 max mem: 3535 +Test: [1090/2387] eta: 0:04:09 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2066 (0.4581) acc1: 94.0104 (78.9468) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.1826 data: 0.0880 max mem: 3535 +Test: [1100/2387] eta: 0:04:07 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2598 (0.4565) acc1: 91.6667 (79.0415) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1710 data: 0.0801 max mem: 3535 +Test: [1110/2387] eta: 0:04:04 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2910 (0.4554) acc1: 89.8438 (79.1062) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1719 data: 0.0811 max mem: 3535 +Test: [1120/2387] eta: 0:04:02 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2619 (0.4535) acc1: 91.1458 (79.2419) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1615 data: 0.0696 max mem: 3536 +Test: [1130/2387] eta: 0:04:00 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2220 (0.4515) acc1: 95.5729 (79.3797) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1680 data: 0.0752 max mem: 3536 +Test: [1140/2387] eta: 0:03:58 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1989 (0.4490) acc1: 97.1354 (79.5387) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1648 data: 0.0726 max mem: 3536 +Test: [1150/2387] eta: 0:03:56 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2120 (0.4477) acc1: 96.0938 (79.6361) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.1670 data: 0.0738 max mem: 3536 +Test: [1160/2387] eta: 0:03:54 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.3024 (0.4464) acc1: 89.5833 (79.7245) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1822 data: 0.0845 max mem: 3536 +Test: [1170/2387] eta: 0:03:51 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2860 (0.4452) acc1: 92.4479 (79.8116) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1601 data: 0.0638 max mem: 3536 +Test: [1180/2387] eta: 0:03:49 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.3081 (0.4441) acc1: 91.1458 (79.8959) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1605 data: 0.0675 max mem: 3536 +Test: [1190/2387] eta: 0:03:47 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.3081 (0.4428) acc1: 89.5833 (79.9840) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1982 data: 0.1052 max mem: 3536 +Test: [1200/2387] eta: 0:03:45 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2818 (0.4412) acc1: 90.6250 (80.0834) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1793 data: 0.0842 max mem: 3536 +Test: [1210/2387] eta: 0:03:43 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2607 (0.4396) acc1: 93.4896 (80.1849) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1647 data: 0.0687 max mem: 3536 +Test: [1220/2387] eta: 0:03:42 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2468 (0.4381) acc1: 94.0104 (80.2772) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.1999 data: 0.1074 max mem: 3536 +Test: [1230/2387] eta: 0:03:40 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2122 (0.4363) acc1: 94.5312 (80.3894) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1997 data: 0.1077 max mem: 3536 +Test: [1240/2387] eta: 0:03:37 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2275 (0.4346) acc1: 93.7500 (80.4908) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1679 data: 0.0750 max mem: 3536 +Test: [1250/2387] eta: 0:03:35 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2022 (0.4328) acc1: 93.7500 (80.6043) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.1653 data: 0.0735 max mem: 3536 +Test: [1260/2387] eta: 0:03:33 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1915 (0.4310) acc1: 95.5729 (80.7224) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1785 data: 0.0876 max mem: 3536 +Test: [1270/2387] eta: 0:03:31 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1554 (0.4286) acc1: 97.6562 (80.8613) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1701 data: 0.0790 max mem: 3536 +Test: [1280/2387] eta: 0:03:29 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1223 (0.4261) acc1: 99.2188 (81.0061) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1610 data: 0.0689 max mem: 3536 +Test: [1290/2387] eta: 0:03:27 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1231 (0.4240) acc1: 98.6979 (81.1263) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1751 data: 0.0819 max mem: 3536 +Test: [1300/2387] eta: 0:03:25 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1294 (0.4218) acc1: 98.1771 (81.2528) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1773 data: 0.0844 max mem: 3536 +Test: [1310/2387] eta: 0:03:23 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1294 (0.4197) acc1: 98.1771 (81.3751) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1665 data: 0.0743 max mem: 3536 +Test: [1320/2387] eta: 0:03:21 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1344 (0.4176) acc1: 98.6979 (81.4988) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1801 data: 0.0893 max mem: 3536 +Test: [1330/2387] eta: 0:03:19 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1314 (0.4155) acc1: 98.9583 (81.6223) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1773 data: 0.0854 max mem: 3536 +Test: [1340/2387] eta: 0:03:17 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1396 (0.4146) acc1: 96.6146 (81.6737) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1758 data: 0.0836 max mem: 3536 +Test: [1350/2387] eta: 0:03:15 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1071 (0.4123) acc1: 100.0000 (81.8082) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1698 data: 0.0781 max mem: 3537 +Test: [1360/2387] eta: 0:03:13 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.0994 (0.4100) acc1: 100.0000 (81.9419) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1689 data: 0.0752 max mem: 3537 +Test: [1370/2387] eta: 0:03:11 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.0978 (0.4077) acc1: 100.0000 (82.0730) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1818 data: 0.0895 max mem: 3537 +Test: [1380/2387] eta: 0:03:09 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.0933 (0.4055) acc1: 100.0000 (82.2029) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1702 data: 0.0789 max mem: 3537 +Test: [1390/2387] eta: 0:03:07 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.0899 (0.4033) acc1: 100.0000 (82.3308) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1730 data: 0.0806 max mem: 3537 +Test: [1400/2387] eta: 0:03:05 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.0879 (0.4010) acc1: 100.0000 (82.4569) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1764 data: 0.0847 max mem: 3537 +Test: [1410/2387] eta: 0:03:03 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.0879 (0.3994) acc1: 100.0000 (82.5534) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1798 data: 0.0861 max mem: 3537 +Test: [1420/2387] eta: 0:03:01 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2263 (0.3987) acc1: 92.4479 (82.5818) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1781 data: 0.0836 max mem: 3537 +Test: [1430/2387] eta: 0:02:59 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2402 (0.3976) acc1: 92.4479 (82.6589) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1699 data: 0.0771 max mem: 3537 +Test: [1440/2387] eta: 0:02:57 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2088 (0.3961) acc1: 95.3125 (82.7572) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1658 data: 0.0742 max mem: 3537 +Test: [1450/2387] eta: 0:02:55 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1767 (0.3947) acc1: 97.6562 (82.8407) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1767 data: 0.0862 max mem: 3537 +Test: [1460/2387] eta: 0:02:53 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2020 (0.3935) acc1: 94.7917 (82.9182) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1745 data: 0.0831 max mem: 3537 +Test: [1470/2387] eta: 0:02:51 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1820 (0.3920) acc1: 96.6146 (83.0170) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1691 data: 0.0764 max mem: 3537 +Test: [1480/2387] eta: 0:02:49 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1669 (0.3905) acc1: 96.8750 (83.1070) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1857 data: 0.0926 max mem: 3537 +Test: [1490/2387] eta: 0:02:47 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1804 (0.3892) acc1: 96.3542 (83.1898) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1702 data: 0.0775 max mem: 3537 +Test: [1500/2387] eta: 0:02:45 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1830 (0.3878) acc1: 96.8750 (83.2799) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1640 data: 0.0720 max mem: 3537 +Test: [1510/2387] eta: 0:02:43 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1830 (0.3866) acc1: 97.3958 (83.3526) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1651 data: 0.0739 max mem: 3537 +Test: [1520/2387] eta: 0:02:41 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2291 (0.3863) acc1: 90.6250 (83.3760) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.1634 data: 0.0716 max mem: 3537 +Test: [1530/2387] eta: 0:02:39 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.3418 (0.3863) acc1: 87.2396 (83.3735) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.1735 data: 0.0816 max mem: 3537 +Test: [1540/2387] eta: 0:02:37 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.3513 (0.3862) acc1: 83.3333 (83.3820) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.1684 data: 0.0770 max mem: 3537 +Test: [1550/2387] eta: 0:02:35 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.3730 (0.3863) acc1: 83.3333 (83.3676) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1654 data: 0.0747 max mem: 3537 +Test: [1560/2387] eta: 0:02:33 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.4064 (0.3867) acc1: 78.9062 (83.3398) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.1602 data: 0.0696 max mem: 3537 +Test: [1570/2387] eta: 0:02:32 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.3680 (0.3864) acc1: 84.8958 (83.3613) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1683 data: 0.0760 max mem: 3537 +Test: [1580/2387] eta: 0:02:30 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.3007 (0.3856) acc1: 88.2812 (83.4164) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1787 data: 0.0865 max mem: 3538 +Test: [1590/2387] eta: 0:02:28 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2537 (0.3849) acc1: 92.7083 (83.4597) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.1820 data: 0.0904 max mem: 3538 +Test: [1600/2387] eta: 0:02:26 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2716 (0.3843) acc1: 90.6250 (83.4918) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1834 data: 0.0922 max mem: 3538 +Test: [1610/2387] eta: 0:02:24 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2216 (0.3831) acc1: 93.7500 (83.5669) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1727 data: 0.0818 max mem: 3538 +Test: [1620/2387] eta: 0:02:22 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2182 (0.3823) acc1: 95.0521 (83.6264) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1691 data: 0.0775 max mem: 3538 +Test: [1630/2387] eta: 0:02:20 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1949 (0.3812) acc1: 96.0938 (83.6951) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1711 data: 0.0797 max mem: 3538 +Test: [1640/2387] eta: 0:02:18 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1692 (0.3798) acc1: 97.3958 (83.7780) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1753 data: 0.0844 max mem: 3538 +Test: [1650/2387] eta: 0:02:16 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1718 (0.3786) acc1: 97.3958 (83.8557) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1803 data: 0.0900 max mem: 3538 +Test: [1660/2387] eta: 0:02:14 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1561 (0.3773) acc1: 97.6562 (83.9354) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1803 data: 0.0893 max mem: 3538 +Test: [1670/2387] eta: 0:02:13 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1561 (0.3761) acc1: 97.6562 (84.0161) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1783 data: 0.0863 max mem: 3538 +Test: [1680/2387] eta: 0:02:11 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1567 (0.3749) acc1: 97.3958 (84.0875) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1691 data: 0.0767 max mem: 3538 +Test: [1690/2387] eta: 0:02:09 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2089 (0.3742) acc1: 94.2708 (84.1405) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1699 data: 0.0757 max mem: 3538 +Test: [1700/2387] eta: 0:02:07 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2168 (0.3732) acc1: 93.2292 (84.1940) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1669 data: 0.0732 max mem: 3538 +Test: [1710/2387] eta: 0:02:05 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2161 (0.3728) acc1: 92.9688 (84.2121) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1649 data: 0.0732 max mem: 3538 +Test: [1720/2387] eta: 0:02:03 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2511 (0.3722) acc1: 91.9271 (84.2559) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1632 data: 0.0717 max mem: 3538 +Test: [1730/2387] eta: 0:02:01 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2782 (0.3719) acc1: 89.3229 (84.2747) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1627 data: 0.0717 max mem: 3538 +Test: [1740/2387] eta: 0:01:59 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2807 (0.3711) acc1: 89.0625 (84.3197) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1699 data: 0.0794 max mem: 3538 +Test: [1750/2387] eta: 0:01:57 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2292 (0.3705) acc1: 92.4479 (84.3624) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1835 data: 0.0929 max mem: 3538 +Test: [1760/2387] eta: 0:01:55 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1920 (0.3693) acc1: 95.5729 (84.4364) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1742 data: 0.0829 max mem: 3538 +Test: [1770/2387] eta: 0:01:53 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1776 (0.3686) acc1: 96.3542 (84.4772) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1541 data: 0.0631 max mem: 3538 +Test: [1780/2387] eta: 0:01:51 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2016 (0.3677) acc1: 94.7917 (84.5269) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.1624 data: 0.0718 max mem: 3538 +Test: [1790/2387] eta: 0:01:50 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2060 (0.3669) acc1: 95.0521 (84.5789) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1643 data: 0.0739 max mem: 3538 +Test: [1800/2387] eta: 0:01:48 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2104 (0.3661) acc1: 93.4896 (84.6282) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1699 data: 0.0794 max mem: 3538 +Test: [1810/2387] eta: 0:01:46 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2104 (0.3652) acc1: 94.2708 (84.6852) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1701 data: 0.0789 max mem: 3539 +Test: [1820/2387] eta: 0:01:44 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1681 (0.3641) acc1: 96.0938 (84.7495) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1640 data: 0.0734 max mem: 3539 +Test: [1830/2387] eta: 0:01:42 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1218 (0.3628) acc1: 98.6979 (84.8266) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1627 data: 0.0722 max mem: 3539 +Test: [1840/2387] eta: 0:01:40 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1014 (0.3613) acc1: 100.0000 (84.9087) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1665 data: 0.0759 max mem: 3539 +Test: [1850/2387] eta: 0:01:38 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1009 (0.3599) acc1: 100.0000 (84.9900) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1643 data: 0.0732 max mem: 3539 +Test: [1860/2387] eta: 0:01:36 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1003 (0.3585) acc1: 100.0000 (85.0703) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1539 data: 0.0608 max mem: 3539 +Test: [1870/2387] eta: 0:01:34 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.0995 (0.3571) acc1: 100.0000 (85.1500) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1580 data: 0.0658 max mem: 3539 +Test: [1880/2387] eta: 0:01:33 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1000 (0.3558) acc1: 100.0000 (85.2287) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1777 data: 0.0868 max mem: 3539 +Test: [1890/2387] eta: 0:01:31 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1005 (0.3544) acc1: 100.0000 (85.3065) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2048 data: 0.1134 max mem: 3539 +Test: [1900/2387] eta: 0:01:29 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1008 (0.3531) acc1: 100.0000 (85.3838) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1867 data: 0.0959 max mem: 3539 +Test: [1910/2387] eta: 0:01:27 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1128 (0.3520) acc1: 100.0000 (85.4567) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1564 data: 0.0663 max mem: 3539 +Test: [1920/2387] eta: 0:01:25 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1617 (0.3515) acc1: 97.1354 (85.4851) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.1643 data: 0.0744 max mem: 3539 +Test: [1930/2387] eta: 0:01:23 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2161 (0.3509) acc1: 95.0521 (85.5256) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1781 data: 0.0869 max mem: 3539 +Test: [1940/2387] eta: 0:01:21 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2171 (0.3506) acc1: 94.7917 (85.5209) acc5: 100.0000 (100.0000) acc1_neg: 64.583 acc5_neg: 100.000 time: 0.1695 data: 0.0781 max mem: 3539 +Test: [1950/2387] eta: 0:01:19 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2191 (0.3501) acc1: 94.7917 (85.5638) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1499 data: 0.0590 max mem: 3539 +Test: [1960/2387] eta: 0:01:18 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2191 (0.3498) acc1: 91.1458 (85.5904) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1514 data: 0.0606 max mem: 3539 +Test: [1970/2387] eta: 0:01:16 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2682 (0.3493) acc1: 91.4062 (85.6339) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1599 data: 0.0681 max mem: 3539 +Test: [1980/2387] eta: 0:01:14 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2148 (0.3486) acc1: 94.5312 (85.6716) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1590 data: 0.0668 max mem: 3539 +Test: [1990/2387] eta: 0:01:12 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1147 (0.3474) acc1: 99.4792 (85.7421) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1667 data: 0.0749 max mem: 3539 +Test: [2000/2387] eta: 0:01:10 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1106 (0.3462) acc1: 99.7396 (85.8114) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1806 data: 0.0890 max mem: 3539 +Test: [2010/2387] eta: 0:01:08 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1106 (0.3451) acc1: 99.7396 (85.8797) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1830 data: 0.0912 max mem: 3539 +Test: [2020/2387] eta: 0:01:07 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1108 (0.3439) acc1: 99.7396 (85.9472) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1832 data: 0.0889 max mem: 3539 +Test: [2030/2387] eta: 0:01:05 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1087 (0.3428) acc1: 99.7396 (86.0147) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1757 data: 0.0829 max mem: 3540 +Test: [2040/2387] eta: 0:01:03 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1081 (0.3416) acc1: 99.7396 (86.0809) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1844 data: 0.0931 max mem: 3540 +Test: [2050/2387] eta: 0:01:01 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1094 (0.3405) acc1: 99.7396 (86.1473) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1958 data: 0.1040 max mem: 3540 +Test: [2060/2387] eta: 0:00:59 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1092 (0.3393) acc1: 99.7396 (86.2140) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1967 data: 0.1058 max mem: 3540 +Test: [2070/2387] eta: 0:00:57 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1059 (0.3382) acc1: 100.0000 (86.2802) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1931 data: 0.1004 max mem: 3540 +Test: [2080/2387] eta: 0:00:56 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1065 (0.3371) acc1: 100.0000 (86.3458) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1711 data: 0.0784 max mem: 3540 +Test: [2090/2387] eta: 0:00:54 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1062 (0.3360) acc1: 100.0000 (86.4108) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1650 data: 0.0740 max mem: 3540 +Test: [2100/2387] eta: 0:00:52 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1059 (0.3349) acc1: 100.0000 (86.4744) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1718 data: 0.0788 max mem: 3540 +Test: [2110/2387] eta: 0:00:50 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1065 (0.3338) acc1: 100.0000 (86.5379) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1647 data: 0.0717 max mem: 3540 +Test: [2120/2387] eta: 0:00:48 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1131 (0.3333) acc1: 99.2188 (86.5729) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.1614 data: 0.0703 max mem: 3540 +Test: [2130/2387] eta: 0:00:46 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1957 (0.3329) acc1: 96.3542 (86.6020) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1705 data: 0.0793 max mem: 3540 +Test: [2140/2387] eta: 0:00:45 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2797 (0.3326) acc1: 91.4062 (86.6201) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1690 data: 0.0774 max mem: 3540 +Test: [2150/2387] eta: 0:00:43 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2517 (0.3321) acc1: 95.3125 (86.6580) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1627 data: 0.0712 max mem: 3540 +Test: [2160/2387] eta: 0:00:41 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2136 (0.3315) acc1: 95.5729 (86.7038) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1528 data: 0.0611 max mem: 3540 +Test: [2170/2387] eta: 0:00:39 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1859 (0.3308) acc1: 97.3958 (86.7480) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1556 data: 0.0638 max mem: 3540 +Test: [2180/2387] eta: 0:00:37 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2099 (0.3315) acc1: 92.7083 (86.6952) acc5: 100.0000 (100.0000) acc1_neg: 50.521 acc5_neg: 100.000 time: 0.1623 data: 0.0711 max mem: 3540 +Test: [2190/2387] eta: 0:00:35 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2468 (0.3314) acc1: 91.4062 (86.7038) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1643 data: 0.0736 max mem: 3540 +Test: [2200/2387] eta: 0:00:33 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2492 (0.3310) acc1: 91.6667 (86.7306) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1724 data: 0.0786 max mem: 3540 +Test: [2210/2387] eta: 0:00:32 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2521 (0.3307) acc1: 91.6667 (86.7542) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1698 data: 0.0748 max mem: 3540 +Test: [2220/2387] eta: 0:00:30 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2475 (0.3304) acc1: 92.4479 (86.7754) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.1582 data: 0.0641 max mem: 3540 +Test: [2230/2387] eta: 0:00:28 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2459 (0.3301) acc1: 92.7083 (86.7984) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1423 data: 0.0497 max mem: 3540 +Test: [2240/2387] eta: 0:00:26 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2277 (0.3297) acc1: 94.0104 (86.8319) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1636 data: 0.0736 max mem: 3540 +Test: [2250/2387] eta: 0:00:24 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2793 (0.3297) acc1: 91.1458 (86.8393) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1878 data: 0.0978 max mem: 3540 +Test: [2260/2387] eta: 0:00:23 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2051 (0.3291) acc1: 92.4479 (86.8800) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1695 data: 0.0766 max mem: 3541 +Test: [2270/2387] eta: 0:00:21 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1736 (0.3284) acc1: 98.4375 (86.9247) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1546 data: 0.0605 max mem: 3541 +Test: [2280/2387] eta: 0:00:19 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1910 (0.3279) acc1: 96.8750 (86.9522) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1833 data: 0.0901 max mem: 3541 +Test: [2290/2387] eta: 0:00:17 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1410 (0.3271) acc1: 97.9167 (87.0014) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1849 data: 0.0902 max mem: 3541 +Test: [2300/2387] eta: 0:00:15 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1551 (0.3263) acc1: 99.2188 (87.0479) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1413 data: 0.0469 max mem: 3541 +Test: [2310/2387] eta: 0:00:13 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1551 (0.3256) acc1: 98.9583 (87.0916) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1411 data: 0.0480 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1617 (0.3251) acc1: 97.9167 (87.1266) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1879 data: 0.0945 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2538 (0.3250) acc1: 94.0104 (87.1346) acc5: 100.0000 (100.0000) acc1_neg: 77.344 acc5_neg: 100.000 time: 0.1888 data: 0.0940 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2626 (0.3248) acc1: 94.0104 (87.1483) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.1349 data: 0.0387 max mem: 3541 +Test: [2350/2387] eta: 0:00:06 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2214 (0.3248) acc1: 95.3125 (87.1510) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1263 data: 0.0321 max mem: 3541 +Test: [2360/2387] eta: 0:00:04 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2415 (0.3245) acc1: 94.0104 (87.1734) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1557 data: 0.0641 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2421 (0.3243) acc1: 94.5312 (87.2011) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1828 data: 0.0905 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.2214 (0.3238) acc1: 95.8333 (87.2362) acc5: 100.0000 (100.0000) acc1_neg: 78.906 acc5_neg: 100.000 time: 0.1740 data: 0.0788 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 32.420 acc5_pos: 100.000 loss: 0.1843 (0.3234) acc1: 97.6744 (87.2495) acc5: 100.0000 (100.0000) acc1_neg: 97.674 acc5_neg: 100.000 time: 0.1596 data: 0.0663 max mem: 3541 +Test: Total time: 0:07:10 (0.1803 s / it) +* Overall Acc@1 87.249 Acc@5 100.000 + target == real Acc@1 17.443 Acc@5 100.000 + target == fake Acc@1 92.152 Acc@5 100.000 + loss 0.323 + PR-AUC 0.9681 ROC-AUC 0.6906 EER 35.96% (thr=0.8477) +Accuracy of the network on the 916267 test images: 87.2% +[rank0]:[W725 17:20:59.580368904 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 17:21:18.037000 2560342 torch/distributed/run.py:766] +W0725 17:21:18.037000 2560342 torch/distributed/run.py:766] ***************************************** +W0725 17:21:18.037000 2560342 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:21:18.037000 2560342 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:21:45.525711928 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:21:45.983233187 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:21:46.190788898 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:21:46.365316286 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:21:46.394088962 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 17:21:46.517027488 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:21:46.719207888 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:21:46.742186795 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed2_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed2_resampled_n116-040-168_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 0:54:46 acc1_pos: 33.333 acc5_pos: 100.000 loss: 1.0874 (1.0874) acc1: 33.3333 (33.3333) acc5: 100.0000 (100.0000) time: 6.3566 data: 4.0812 max mem: 3531 +Test: [ 10/517] eta: 0:05:34 acc1_pos: 30.990 acc5_pos: 100.000 loss: 1.1139 (1.1487) acc1: 30.9896 (31.6525) acc5: 100.0000 (100.0000) time: 0.6595 data: 0.3715 max mem: 3531 +Test: [ 20/517] eta: 0:03:13 acc1_pos: 18.229 acc5_pos: 100.000 loss: 1.1702 (1.1743) acc1: 29.6875 (32.2793) acc5: 100.0000 (100.0000) time: 0.0910 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:02:44 acc1_pos: 14.844 acc5_pos: 100.000 loss: 1.2723 (1.2155) acc1: 25.0000 (30.0823) acc5: 100.0000 (100.0000) time: 0.1614 data: 0.0704 max mem: 3531 +Test: [ 40/517] eta: 0:02:42 acc1_pos: 18.229 acc5_pos: 100.000 loss: 1.3440 (1.2362) acc1: 18.7500 (28.3791) acc5: 100.0000 (100.0000) time: 0.2909 data: 0.1982 max mem: 3531 +Test: [ 50/517] eta: 0:02:33 acc1_pos: 12.385 acc5_pos: 100.000 loss: 1.0788 (1.1590) acc1: 28.3854 (32.7053) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.3145 data: 0.2206 max mem: 3531 +Test: [ 60/517] eta: 0:02:31 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.3319 (1.0190) acc1: 88.0208 (42.3711) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.3140 data: 0.2185 max mem: 3531 +Test: [ 70/517] eta: 0:02:22 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.2997 (0.9192) acc1: 92.1875 (49.3985) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2949 data: 0.1972 max mem: 3531 +Test: [ 80/517] eta: 0:02:16 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.3006 (0.8486) acc1: 92.1875 (54.3435) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.2549 data: 0.1587 max mem: 3531 +Test: [ 90/517] eta: 0:02:16 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.2527 (0.7824) acc1: 94.7917 (58.8856) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.3249 data: 0.2306 max mem: 3531 +Test: [100/517] eta: 0:02:09 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.2527 (0.7307) acc1: 95.8333 (62.5645) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2999 data: 0.2044 max mem: 3531 +Test: [110/517] eta: 0:02:07 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.2746 (0.6926) acc1: 94.7917 (64.9376) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.2749 data: 0.1753 max mem: 3531 +Test: [120/517] eta: 0:02:01 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.3036 (0.6608) acc1: 90.8854 (67.2069) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2799 data: 0.1813 max mem: 3531 +Test: [130/517] eta: 0:01:56 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.3087 (0.6351) acc1: 90.8854 (68.9289) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2349 data: 0.1377 max mem: 3531 +Test: [140/517] eta: 0:01:51 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.2815 (0.6096) acc1: 92.4479 (70.6800) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2399 data: 0.1411 max mem: 3531 +Test: [150/517] eta: 0:01:47 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.2470 (0.5850) acc1: 95.3125 (72.3613) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2349 data: 0.1382 max mem: 3531 +Test: [160/517] eta: 0:01:43 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.3275 (0.5730) acc1: 90.3646 (73.2240) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.2399 data: 0.1411 max mem: 3531 +Test: [170/517] eta: 0:01:39 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.3857 (0.5605) acc1: 85.6771 (74.0360) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.2549 data: 0.1546 max mem: 3531 +Test: [180/517] eta: 0:01:36 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.1909 (0.5393) acc1: 95.0521 (75.3338) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2499 data: 0.1536 max mem: 3531 +Test: [190/517] eta: 0:01:32 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.1848 (0.5219) acc1: 97.9167 (76.5257) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2404 data: 0.1438 max mem: 3531 +Test: [200/517] eta: 0:01:28 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.1918 (0.5049) acc1: 98.6979 (77.6353) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2358 data: 0.1399 max mem: 3531 +Test: [210/517] eta: 0:01:25 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.2626 (0.4991) acc1: 94.5312 (77.9485) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.2343 data: 0.1396 max mem: 3532 +Test: [220/517] eta: 0:01:22 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.3876 (0.4951) acc1: 82.2917 (78.0720) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.2390 data: 0.1433 max mem: 3532 +Test: [230/517] eta: 0:01:18 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.3426 (0.4865) acc1: 89.3229 (78.6943) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.2357 data: 0.1402 max mem: 3532 +Test: [240/517] eta: 0:01:15 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.2986 (0.4811) acc1: 92.9688 (79.1904) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.2300 data: 0.1356 max mem: 3532 +Test: [250/517] eta: 0:01:12 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.2733 (0.4727) acc1: 95.3125 (79.8006) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2501 data: 0.1566 max mem: 3532 +Test: [260/517] eta: 0:01:09 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.2414 (0.4647) acc1: 95.3125 (80.3351) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.2498 data: 0.1545 max mem: 3532 +Test: [270/517] eta: 0:01:06 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.3309 (0.4634) acc1: 88.0208 (80.4832) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.2288 data: 0.1311 max mem: 3532 +Test: [280/517] eta: 0:01:03 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.3945 (0.4586) acc1: 86.4583 (80.8728) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2360 data: 0.1390 max mem: 3532 +Test: [290/517] eta: 0:01:00 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.2737 (0.4519) acc1: 92.9688 (81.3261) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2399 data: 0.1452 max mem: 3532 +Test: [300/517] eta: 0:00:57 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.2352 (0.4442) acc1: 96.3542 (81.8435) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2438 data: 0.1480 max mem: 3532 +Test: [310/517] eta: 0:00:54 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.1971 (0.4357) acc1: 98.6979 (82.4030) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2403 data: 0.1443 max mem: 3532 +Test: [320/517] eta: 0:00:52 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.1775 (0.4286) acc1: 99.2188 (82.8652) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2402 data: 0.1436 max mem: 3532 +Test: [330/517] eta: 0:00:49 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.2044 (0.4220) acc1: 97.3958 (83.2830) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2445 data: 0.1449 max mem: 3532 +Test: [340/517] eta: 0:00:46 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.1968 (0.4162) acc1: 97.6562 (83.6159) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.2446 data: 0.1453 max mem: 3532 +Test: [350/517] eta: 0:00:43 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.2616 (0.4126) acc1: 96.3542 (83.9454) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2455 data: 0.1495 max mem: 3532 +Test: [360/517] eta: 0:00:41 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.2643 (0.4080) acc1: 96.8750 (84.3216) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2311 data: 0.1367 max mem: 3532 +Test: [370/517] eta: 0:00:38 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.2241 (0.4032) acc1: 96.8750 (84.6214) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2403 data: 0.1471 max mem: 3532 +Test: [380/517] eta: 0:00:35 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.2293 (0.3989) acc1: 96.0938 (84.8767) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.2487 data: 0.1527 max mem: 3532 +Test: [390/517] eta: 0:00:33 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.2528 (0.3962) acc1: 91.1458 (85.0177) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2477 data: 0.1489 max mem: 3532 +Test: [400/517] eta: 0:00:30 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.2823 (0.3934) acc1: 91.1458 (85.2329) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.2488 data: 0.1521 max mem: 3532 +Test: [410/517] eta: 0:00:27 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.2939 (0.3920) acc1: 91.1458 (85.3787) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.2299 data: 0.1339 max mem: 3532 +Test: [420/517] eta: 0:00:25 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.2803 (0.3886) acc1: 93.4896 (85.6455) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2349 data: 0.1373 max mem: 3532 +Test: [430/517] eta: 0:00:22 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.2803 (0.3873) acc1: 95.5729 (85.7786) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.2449 data: 0.1476 max mem: 3532 +Test: [440/517] eta: 0:00:19 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.3036 (0.3841) acc1: 92.9688 (85.9676) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2462 data: 0.1503 max mem: 3533 +Test: [450/517] eta: 0:00:17 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.2476 (0.3820) acc1: 96.0938 (86.1690) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.2649 data: 0.1678 max mem: 3533 +Test: [460/517] eta: 0:00:14 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.2704 (0.3796) acc1: 94.0104 (86.3115) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2487 data: 0.1513 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.1970 (0.3756) acc1: 97.3958 (86.5568) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2302 data: 0.1347 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.2064 (0.3739) acc1: 97.3958 (86.6857) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2249 data: 0.1269 max mem: 3533 +Test: [490/517] eta: 0:00:06 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.3262 (0.3764) acc1: 88.5417 (86.4276) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2297 data: 0.1309 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.2630 (0.3732) acc1: 94.5312 (86.6647) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2351 data: 0.1397 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.2142 (0.3714) acc1: 98.1771 (86.7738) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2035 data: 0.1091 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 12.385 acc5_pos: 100.000 loss: 0.2142 (0.3696) acc1: 98.5866 (86.9010) acc5: 100.0000 (100.0000) acc1_neg: 98.587 acc5_neg: 100.000 time: 0.1496 data: 0.0543 max mem: 3533 +Test: Total time: 0:02:11 (0.2536 s / it) +* Overall Acc@1 86.901 Acc@5 100.000 + target == real Acc@1 29.871 Acc@5 100.000 + target == fake Acc@1 92.817 Acc@5 100.000 + loss 0.370 + PR-AUC 0.9455 ROC-AUC 0.6949 EER 36.15% (thr=0.7690) +Accuracy of the network on the 198427 test images: 86.9% +[rank0]:[W725 17:24:12.171623958 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0725 17:24:32.727000 2560948 torch/distributed/run.py:766] +W0725 17:24:32.727000 2560948 torch/distributed/run.py:766] ***************************************** +W0725 17:24:32.727000 2560948 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:24:32.727000 2560948 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W725 17:25:01.377936441 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:25:01.634629584 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:25:01.074881478 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:25:02.335492064 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:25:02.580858754 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +[rank3]:[W725 17:25:02.680936401 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:25:02.694289001 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:25:02.812523078 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed2_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed2_resampled_n116-040-168_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:34:39 acc1_pos: 3.125 acc5_pos: 100.000 loss: 1.6534 (1.6534) acc1: 3.1250 (3.1250) acc5: 100.0000 (100.0000) time: 5.3454 data: 3.4523 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/389] eta: 0:03:35 acc1_pos: 0.521 acc5_pos: 100.000 loss: 1.6425 (1.6332) acc1: 5.7292 (5.6818) acc5: 100.0000 (100.0000) time: 0.5675 data: 0.3141 max mem: 3531 +Test: [ 20/389] eta: 0:02:06 acc1_pos: 1.302 acc5_pos: 100.000 loss: 1.7373 (1.8115) acc1: 1.3021 (3.3606) acc5: 100.0000 (100.0000) time: 0.0926 data: 0.0018 max mem: 3531 +Test: [ 30/389] eta: 0:01:41 acc1_pos: 4.167 acc5_pos: 100.000 loss: 2.1036 (1.9081) acc1: 0.2604 (2.4194) acc5: 100.0000 (100.0000) time: 0.1252 data: 0.0317 max mem: 3531 +Test: [ 40/389] eta: 0:01:34 acc1_pos: 1.302 acc5_pos: 100.000 loss: 2.1047 (1.9530) acc1: 0.0000 (2.2929) acc5: 100.0000 (100.0000) time: 0.1943 data: 0.1012 max mem: 3531 +Test: [ 50/389] eta: 0:01:28 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.2598 (1.9941) acc1: 0.0000 (2.0782) acc5: 100.0000 (100.0000) time: 0.2270 data: 0.1326 max mem: 3531 +Test: [ 60/389] eta: 0:01:22 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.2932 (2.0097) acc1: 0.0000 (2.2456) acc5: 100.0000 (100.0000) time: 0.2081 data: 0.1135 max mem: 3531 +Test: [ 70/389] eta: 0:01:18 acc1_pos: 1.823 acc5_pos: 100.000 loss: 1.8528 (1.9638) acc1: 2.6042 (2.8426) acc5: 100.0000 (100.0000) time: 0.2078 data: 0.1132 max mem: 3531 +Test: [ 80/389] eta: 0:01:13 acc1_pos: 4.688 acc5_pos: 100.000 loss: 1.6309 (1.9267) acc1: 3.9062 (3.1636) acc5: 100.0000 (100.0000) time: 0.2052 data: 0.1110 max mem: 3531 +Test: [ 90/389] eta: 0:01:11 acc1_pos: 0.260 acc5_pos: 100.000 loss: 1.5681 (1.8862) acc1: 3.9062 (3.8748) acc5: 100.0000 (100.0000) time: 0.2149 data: 0.1231 max mem: 3531 +Test: [100/389] eta: 0:01:08 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.8285 (1.8950) acc1: 2.0833 (3.6819) acc5: 100.0000 (100.0000) time: 0.2296 data: 0.1356 max mem: 3531 +Test: [110/389] eta: 0:01:03 acc1_pos: 4.167 acc5_pos: 100.000 loss: 1.8354 (1.8742) acc1: 2.6042 (4.0986) acc5: 100.0000 (100.0000) time: 0.1816 data: 0.0838 max mem: 3531 +Test: [120/389] eta: 0:01:02 acc1_pos: 0.260 acc5_pos: 100.000 loss: 1.6737 (1.8684) acc1: 0.7812 (3.8180) acc5: 100.0000 (100.0000) time: 0.2073 data: 0.1109 max mem: 3531 +Test: [130/389] eta: 0:00:58 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.6324 (1.7679) acc1: 2.3438 (9.2736) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2153 data: 0.1212 max mem: 3531 +Test: [140/389] eta: 0:00:56 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1440 (1.6573) acc1: 99.2188 (15.4181) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1825 data: 0.0890 max mem: 3531 +Test: [150/389] eta: 0:00:53 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1440 (1.5605) acc1: 99.4792 (20.7575) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2003 data: 0.1077 max mem: 3531 +Test: [160/389] eta: 0:00:50 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1379 (1.4706) acc1: 99.7396 (25.6616) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1958 data: 0.1029 max mem: 3531 +Test: [170/389] eta: 0:00:48 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0975 (1.3928) acc1: 100.0000 (29.9053) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2030 data: 0.1113 max mem: 3531 +Test: [180/389] eta: 0:00:45 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1141 (1.3233) acc1: 100.0000 (33.6973) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1990 data: 0.1079 max mem: 3531 +Test: [190/389] eta: 0:00:43 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1196 (1.2692) acc1: 100.0000 (36.5783) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1844 data: 0.0900 max mem: 3531 +Test: [200/389] eta: 0:00:40 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1506 (1.2144) acc1: 99.7396 (39.7207) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1906 data: 0.0952 max mem: 3531 +Test: [210/389] eta: 0:00:38 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1393 (1.1619) acc1: 99.7396 (42.5652) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1849 data: 0.0921 max mem: 3532 +Test: [220/389] eta: 0:00:36 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0950 (1.1136) acc1: 100.0000 (45.1593) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1915 data: 0.1001 max mem: 3532 +Test: [230/389] eta: 0:00:33 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1226 (1.0732) acc1: 100.0000 (47.3677) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1999 data: 0.1088 max mem: 3532 +Test: [240/389] eta: 0:00:31 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1326 (1.0354) acc1: 99.7396 (49.4446) acc5: 100.0000 (100.0000) acc1_neg: 80.729 acc5_neg: 100.000 time: 0.1959 data: 0.1034 max mem: 3532 +Test: [250/389] eta: 0:00:29 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1560 (1.0040) acc1: 99.7396 (51.2595) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1920 data: 0.0979 max mem: 3532 +Test: [260/389] eta: 0:00:27 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1568 (0.9729) acc1: 99.7396 (53.0821) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1916 data: 0.0975 max mem: 3532 +Test: [270/389] eta: 0:00:25 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1151 (0.9411) acc1: 100.0000 (54.8047) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2048 data: 0.1130 max mem: 3532 +Test: [280/389] eta: 0:00:22 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1078 (0.9117) acc1: 100.0000 (56.4085) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1949 data: 0.1034 max mem: 3532 +Test: [290/389] eta: 0:00:20 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1244 (0.8869) acc1: 99.7396 (57.7973) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2049 data: 0.1138 max mem: 3532 +Test: [300/389] eta: 0:00:18 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1156 (0.8612) acc1: 99.7396 (59.1924) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1976 data: 0.1076 max mem: 3532 +Test: [310/389] eta: 0:00:16 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1084 (0.8373) acc1: 100.0000 (60.5004) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1879 data: 0.0968 max mem: 3532 +Test: [320/389] eta: 0:00:14 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1342 (0.8203) acc1: 99.7396 (61.4778) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2072 data: 0.1149 max mem: 3532 +Test: [330/389] eta: 0:00:12 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1573 (0.7998) acc1: 99.4792 (62.6345) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1901 data: 0.0981 max mem: 3532 +Test: [340/389] eta: 0:00:10 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1358 (0.7796) acc1: 100.0000 (63.7250) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1707 data: 0.0792 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0988 (0.7614) acc1: 100.0000 (64.7391) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1399 data: 0.0492 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0960 (0.7426) acc1: 100.0000 (65.7159) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1226 data: 0.0327 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0814 (0.7258) acc1: 100.0000 (66.6379) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1367 data: 0.0472 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1250 (0.7105) acc1: 100.0000 (67.4903) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1245 data: 0.0351 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1059 (0.6983) acc1: 100.0000 (68.1530) acc5: 100.0000 (100.0000) acc1_neg: 99.727 acc5_neg: 100.000 time: 0.1123 data: 0.0196 max mem: 3532 +Test: Total time: 0:01:16 (0.1962 s / it) +* Overall Acc@1 68.153 Acc@5 100.000 + target == real Acc@1 3.781 Acc@5 100.000 + target == fake Acc@1 98.075 Acc@5 100.000 + loss 0.698 + PR-AUC 0.8033 ROC-AUC 0.6687 EER 37.79% (thr=0.8804) +Accuracy of the network on the 149358 test images: 68.2% +[rank0]:[W725 17:26:26.252622936 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 17:26:45.267000 2561516 torch/distributed/run.py:766] +W0725 17:26:45.267000 2561516 torch/distributed/run.py:766] ***************************************** +W0725 17:26:45.267000 2561516 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:26:45.267000 2561516 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:27:14.353130457 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:27:14.566692995 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:27:14.066438594 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:27:14.142383022 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 17:27:15.244163777 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:27:15.406020505 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:27:15.520480858 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:27:15.599485265 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed2_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed2_resampled_n116-040-168_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:27:04 acc1_pos: 4.427 acc5_pos: 100.000 loss: 2.2213 (2.2213) acc1: 4.4271 (4.4271) acc5: 100.0000 (100.0000) time: 4.8772 data: 2.6552 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/333] eta: 0:02:52 acc1_pos: 7.812 acc5_pos: 100.000 loss: 2.0356 (2.0363) acc1: 5.7292 (7.1496) acc5: 100.0000 (100.0000) time: 0.5337 data: 0.2504 max mem: 3531 +Test: [ 20/333] eta: 0:01:45 acc1_pos: 18.750 acc5_pos: 100.000 loss: 1.9813 (1.9578) acc1: 7.8125 (9.2758) acc5: 100.0000 (100.0000) time: 0.1087 data: 0.0191 max mem: 3531 +Test: [ 30/333] eta: 0:01:26 acc1_pos: 28.906 acc5_pos: 100.000 loss: 1.6274 (1.7727) acc1: 13.5417 (14.3229) acc5: 100.0000 (100.0000) time: 0.1484 data: 0.0569 max mem: 3531 +Test: [ 40/333] eta: 0:01:29 acc1_pos: 34.375 acc5_pos: 100.000 loss: 1.3122 (1.6636) acc1: 27.6042 (17.3590) acc5: 100.0000 (100.0000) time: 0.2723 data: 0.1792 max mem: 3531 +Test: [ 50/333] eta: 0:01:24 acc1_pos: 20.573 acc5_pos: 100.000 loss: 1.3314 (1.5989) acc1: 26.0417 (18.7755) acc5: 100.0000 (100.0000) time: 0.3174 data: 0.2260 max mem: 3531 +Test: [ 60/333] eta: 0:01:14 acc1_pos: 15.625 acc5_pos: 100.000 loss: 1.3314 (1.5600) acc1: 25.0000 (19.5569) acc5: 100.0000 (100.0000) time: 0.2020 data: 0.1109 max mem: 3531 +Test: [ 70/333] eta: 0:01:08 acc1_pos: 24.219 acc5_pos: 100.000 loss: 1.3434 (1.5325) acc1: 24.4792 (20.4922) acc5: 100.0000 (100.0000) time: 0.1683 data: 0.0763 max mem: 3531 +Test: [ 80/333] eta: 0:01:03 acc1_pos: 25.781 acc5_pos: 100.000 loss: 1.3468 (1.5141) acc1: 24.4792 (20.9619) acc5: 100.0000 (100.0000) time: 0.1925 data: 0.1000 max mem: 3531 +Test: [ 90/333] eta: 0:00:59 acc1_pos: 64.583 acc5_pos: 100.000 loss: 1.2929 (1.4489) acc1: 27.3438 (23.8210) acc5: 100.0000 (100.0000) time: 0.1800 data: 0.0884 max mem: 3531 +Test: [100/333] eta: 0:00:55 acc1_pos: 57.812 acc5_pos: 100.000 loss: 0.7749 (1.3804) acc1: 54.1667 (27.0472) acc5: 100.0000 (100.0000) time: 0.1769 data: 0.0851 max mem: 3531 +Test: [110/333] eta: 0:00:52 acc1_pos: 62.760 acc5_pos: 100.000 loss: 0.8070 (1.3310) acc1: 52.6042 (29.3051) acc5: 100.0000 (100.0000) time: 0.1886 data: 0.0960 max mem: 3531 +Test: [120/333] eta: 0:00:49 acc1_pos: 67.448 acc5_pos: 100.000 loss: 0.8070 (1.2853) acc1: 52.6042 (31.3942) acc5: 100.0000 (100.0000) time: 0.1999 data: 0.1077 max mem: 3531 +Test: [130/333] eta: 0:00:46 acc1_pos: 61.719 acc5_pos: 100.000 loss: 0.7295 (1.2417) acc1: 56.5104 (33.6892) acc5: 100.0000 (100.0000) time: 0.2107 data: 0.1187 max mem: 3531 +Test: [140/333] eta: 0:00:43 acc1_pos: 66.667 acc5_pos: 100.000 loss: 0.7036 (1.2049) acc1: 61.7188 (35.6863) acc5: 100.0000 (100.0000) time: 0.1949 data: 0.1036 max mem: 3531 +Test: [150/333] eta: 0:00:40 acc1_pos: 55.990 acc5_pos: 100.000 loss: 0.7429 (1.1756) acc1: 60.4167 (37.1102) acc5: 100.0000 (100.0000) time: 0.1791 data: 0.0862 max mem: 3531 +Test: [160/333] eta: 0:00:38 acc1_pos: 47.697 acc5_pos: 100.000 loss: 0.7420 (1.1456) acc1: 56.7708 (38.6080) acc5: 100.0000 (100.0000) acc1_neg: 66.406 acc5_neg: 100.000 time: 0.1964 data: 0.1042 max mem: 3531 +Test: [170/333] eta: 0:00:36 acc1_pos: 47.697 acc5_pos: 100.000 loss: 0.6024 (1.1098) acc1: 65.8854 (40.4971) acc5: 100.0000 (100.0000) acc1_neg: 74.219 acc5_neg: 100.000 time: 0.2178 data: 0.1268 max mem: 3531 +Test: [180/333] eta: 0:00:33 acc1_pos: 47.697 acc5_pos: 100.000 loss: 0.4851 (1.0744) acc1: 75.0000 (42.4853) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.2100 data: 0.1172 max mem: 3531 +Test: [190/333] eta: 0:00:31 acc1_pos: 47.697 acc5_pos: 100.000 loss: 0.4157 (1.0375) acc1: 79.6875 (44.7058) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.1776 data: 0.0826 max mem: 3531 +Test: [200/333] eta: 0:00:28 acc1_pos: 47.697 acc5_pos: 100.000 loss: 0.4002 (1.0083) acc1: 79.6875 (46.3257) acc5: 100.0000 (100.0000) acc1_neg: 71.354 acc5_neg: 100.000 time: 0.1805 data: 0.0864 max mem: 3531 +Test: [210/333] eta: 0:00:26 acc1_pos: 47.697 acc5_pos: 100.000 loss: 0.4646 (0.9846) acc1: 75.5208 (47.5439) acc5: 100.0000 (100.0000) acc1_neg: 73.698 acc5_neg: 100.000 time: 0.1707 data: 0.0778 max mem: 3532 +Test: [220/333] eta: 0:00:23 acc1_pos: 47.697 acc5_pos: 100.000 loss: 0.4799 (0.9670) acc1: 73.6979 (48.4776) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.1687 data: 0.0762 max mem: 3532 +Test: [230/333] eta: 0:00:21 acc1_pos: 47.697 acc5_pos: 100.000 loss: 0.3830 (0.9398) acc1: 81.5104 (50.1172) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1754 data: 0.0840 max mem: 3532 +Test: [240/333] eta: 0:00:19 acc1_pos: 47.697 acc5_pos: 100.000 loss: 0.3830 (0.9186) acc1: 83.0729 (51.3885) acc5: 100.0000 (100.0000) acc1_neg: 77.344 acc5_neg: 100.000 time: 0.1498 data: 0.0591 max mem: 3532 +Test: [250/333] eta: 0:00:17 acc1_pos: 47.697 acc5_pos: 100.000 loss: 0.4121 (0.8983) acc1: 81.7708 (52.6000) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.1580 data: 0.0672 max mem: 3532 +Test: [260/333] eta: 0:00:14 acc1_pos: 47.697 acc5_pos: 100.000 loss: 0.4187 (0.8802) acc1: 80.7292 (53.6329) acc5: 100.0000 (100.0000) acc1_neg: 77.083 acc5_neg: 100.000 time: 0.1814 data: 0.0900 max mem: 3532 +Test: [270/333] eta: 0:00:12 acc1_pos: 47.697 acc5_pos: 100.000 loss: 0.4268 (0.8635) acc1: 78.3854 (54.5578) acc5: 100.0000 (100.0000) acc1_neg: 72.135 acc5_neg: 100.000 time: 0.1713 data: 0.0799 max mem: 3532 +Test: [280/333] eta: 0:00:10 acc1_pos: 47.697 acc5_pos: 100.000 loss: 0.4424 (0.8541) acc1: 75.2604 (54.9470) acc5: 100.0000 (100.0000) acc1_neg: 63.542 acc5_neg: 100.000 time: 0.1705 data: 0.0789 max mem: 3532 +Test: [290/333] eta: 0:00:08 acc1_pos: 47.697 acc5_pos: 100.000 loss: 0.6757 (0.8480) acc1: 59.8958 (55.0947) acc5: 100.0000 (100.0000) acc1_neg: 63.542 acc5_neg: 100.000 time: 0.1856 data: 0.0936 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 47.697 acc5_pos: 100.000 loss: 0.5761 (0.8377) acc1: 67.7083 (55.6132) acc5: 100.0000 (100.0000) acc1_neg: 67.969 acc5_neg: 100.000 time: 0.1906 data: 0.0986 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 47.697 acc5_pos: 100.000 loss: 0.4898 (0.8244) acc1: 73.4375 (56.3932) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1937 data: 0.1021 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 47.697 acc5_pos: 100.000 loss: 0.4091 (0.8129) acc1: 80.2083 (57.0459) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.1932 data: 0.1024 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 47.697 acc5_pos: 100.000 loss: 0.6061 (0.8082) acc1: 64.8438 (57.2114) acc5: 100.0000 (100.0000) acc1_neg: 64.583 acc5_neg: 100.000 time: 0.1686 data: 0.0777 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 47.697 acc5_pos: 100.000 loss: 0.6377 (0.8073) acc1: 64.0625 (57.2421) acc5: 100.0000 (100.0000) acc1_neg: 63.380 acc5_neg: 100.000 time: 0.1643 data: 0.0702 max mem: 3532 +Test: Total time: 0:01:06 (0.2000 s / it) +* Overall Acc@1 57.242 Acc@5 100.000 + target == real Acc@1 37.950 Acc@5 100.000 + target == fake Acc@1 74.416 Acc@5 100.000 + loss 0.807 + PR-AUC 0.5759 ROC-AUC 0.5817 EER 43.64% (thr=0.6787) +Accuracy of the network on the 127843 test images: 57.2% +[rank0]:[W725 17:28:33.510861009 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 17:28:53.135000 2562072 torch/distributed/run.py:766] +W0725 17:28:53.135000 2562072 torch/distributed/run.py:766] ***************************************** +W0725 17:28:53.135000 2562072 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:28:53.135000 2562072 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:29:20.764028919 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:29:20.966602317 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:29:21.282462521 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 17:29:21.322840953 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +NCCL version 2.26.2+cuda12.2 +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:29:21.504238620 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:29:21.516092132 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:29:21.538022859 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:29:21.583319411 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed2_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed2_resampled_n116-040-168_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:26:04 acc1_pos: 94.010 acc5_pos: 100.000 loss: 0.2316 (0.2316) acc1: 94.0104 (94.0104) acc5: 100.0000 (100.0000) time: 6.6558 data: 4.8326 max mem: 3531 +Test: [ 10/235] eta: 0:02:35 acc1_pos: 95.052 acc5_pos: 100.000 loss: 0.2340 (0.2350) acc1: 95.0521 (95.0994) acc5: 100.0000 (100.0000) time: 0.6932 data: 0.4397 max mem: 3531 +Test: [ 20/235] eta: 0:01:28 acc1_pos: 95.833 acc5_pos: 100.000 loss: 0.2322 (0.2347) acc1: 95.3125 (95.1513) acc5: 100.0000 (100.0000) time: 0.0984 data: 0.0005 max mem: 3531 +Test: [ 30/235] eta: 0:01:24 acc1_pos: 94.531 acc5_pos: 100.000 loss: 0.2276 (0.2334) acc1: 95.3125 (95.2033) acc5: 100.0000 (100.0000) time: 0.2549 data: 0.1583 max mem: 3531 +Test: [ 40/235] eta: 0:01:12 acc1_pos: 94.271 acc5_pos: 100.000 loss: 0.2321 (0.2338) acc1: 95.3125 (95.2172) acc5: 100.0000 (100.0000) time: 0.3286 data: 0.2368 max mem: 3531 +Test: [ 50/235] eta: 0:01:03 acc1_pos: 93.490 acc5_pos: 100.000 loss: 0.2399 (0.2354) acc1: 95.0521 (95.1031) acc5: 100.0000 (100.0000) time: 0.2328 data: 0.1430 max mem: 3531 +Test: [ 60/235] eta: 0:00:55 acc1_pos: 96.354 acc5_pos: 100.000 loss: 0.2432 (0.2363) acc1: 94.5312 (95.1119) acc5: 100.0000 (100.0000) time: 0.2074 data: 0.1150 max mem: 3531 +Test: [ 70/235] eta: 0:00:50 acc1_pos: 96.923 acc5_pos: 100.000 loss: 0.2501 (0.3769) acc1: 94.2708 (85.9485) acc5: 100.0000 (100.0000) acc1_neg: 19.010 acc5_neg: 100.000 time: 0.2220 data: 0.1287 max mem: 3531 +Test: [ 80/235] eta: 0:00:45 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.3534 (0.4991) acc1: 21.8750 (77.9739) acc5: 100.0000 (100.0000) acc1_neg: 19.531 acc5_neg: 100.000 time: 0.2238 data: 0.1292 max mem: 3531 +Test: [ 90/235] eta: 0:00:40 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.3697 (0.5963) acc1: 21.3542 (71.7548) acc5: 100.0000 (100.0000) acc1_neg: 17.188 acc5_neg: 100.000 time: 0.1649 data: 0.0653 max mem: 3531 +Test: [100/235] eta: 0:00:35 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.3660 (0.6731) acc1: 21.6146 (66.7930) acc5: 100.0000 (100.0000) acc1_neg: 16.146 acc5_neg: 100.000 time: 0.1353 data: 0.0388 max mem: 3531 +Test: [110/235] eta: 0:00:31 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.3570 (0.7348) acc1: 21.6146 (62.7510) acc5: 100.0000 (100.0000) acc1_neg: 20.052 acc5_neg: 100.000 time: 0.1553 data: 0.0646 max mem: 3531 +Test: [120/235] eta: 0:00:28 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.3810 (0.8056) acc1: 19.7917 (58.3914) acc5: 100.0000 (100.0000) acc1_neg: 8.854 acc5_neg: 100.000 time: 0.1712 data: 0.0800 max mem: 3531 +Test: [130/235] eta: 0:00:25 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.6073 (0.8679) acc1: 8.8542 (54.5583) acc5: 100.0000 (100.0000) acc1_neg: 7.031 acc5_neg: 100.000 time: 0.1797 data: 0.0857 max mem: 3531 +Test: [140/235] eta: 0:00:22 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.6259 (0.9224) acc1: 8.3333 (51.3169) acc5: 100.0000 (100.0000) acc1_neg: 7.031 acc5_neg: 100.000 time: 0.2033 data: 0.1073 max mem: 3531 +Test: [150/235] eta: 0:00:20 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.6273 (0.9683) acc1: 8.5938 (48.5117) acc5: 100.0000 (100.0000) acc1_neg: 7.812 acc5_neg: 100.000 time: 0.2208 data: 0.1257 max mem: 3531 +Test: [160/235] eta: 0:00:17 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.6213 (1.0086) acc1: 8.8542 (46.0711) acc5: 100.0000 (100.0000) acc1_neg: 8.854 acc5_neg: 100.000 time: 0.1999 data: 0.1056 max mem: 3531 +Test: [170/235] eta: 0:00:15 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.5760 (1.0334) acc1: 10.6771 (44.6942) acc5: 100.0000 (100.0000) acc1_neg: 22.396 acc5_neg: 100.000 time: 0.1961 data: 0.1032 max mem: 3531 +Test: [180/235] eta: 0:00:12 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.4038 (1.0524) acc1: 23.9583 (43.6061) acc5: 100.0000 (100.0000) acc1_neg: 24.219 acc5_neg: 100.000 time: 0.1949 data: 0.1006 max mem: 3531 +Test: [190/235] eta: 0:00:10 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.3588 (1.0687) acc1: 25.2604 (42.6947) acc5: 100.0000 (100.0000) acc1_neg: 26.562 acc5_neg: 100.000 time: 0.1847 data: 0.0913 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.3664 (1.0851) acc1: 25.2604 (41.7664) acc5: 100.0000 (100.0000) acc1_neg: 21.354 acc5_neg: 100.000 time: 0.1813 data: 0.0896 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.3818 (1.0997) acc1: 23.6979 (40.9373) acc5: 100.0000 (100.0000) acc1_neg: 21.875 acc5_neg: 100.000 time: 0.1610 data: 0.0682 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.4068 (1.1134) acc1: 22.6562 (40.1525) acc5: 100.0000 (100.0000) acc1_neg: 20.312 acc5_neg: 100.000 time: 0.1505 data: 0.0594 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.3927 (1.1251) acc1: 24.2188 (39.4988) acc5: 100.0000 (100.0000) acc1_neg: 23.958 acc5_neg: 100.000 time: 0.1293 data: 0.0389 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 96.923 acc5_pos: 100.000 loss: 1.3927 (1.1305) acc1: 24.2188 (39.2917) acc5: 100.0000 (100.0000) acc1_neg: 13.043 acc5_neg: 100.000 time: 0.1203 data: 0.0334 max mem: 3532 +Test: Total time: 0:00:50 (0.2136 s / it) +* Overall Acc@1 39.292 Acc@5 100.000 + target == real Acc@1 95.107 Acc@5 100.000 + target == fake Acc@1 19.104 Acc@5 100.000 + loss 1.131 + PR-AUC 0.8439 ROC-AUC 0.6627 EER 38.67% (thr=0.1720) +Accuracy of the network on the 89879 test images: 39.3% +[rank0]:[W725 17:30:20.085179986 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 17:30:40.626000 2562640 torch/distributed/run.py:766] +W0725 17:30:40.626000 2562640 torch/distributed/run.py:766] ***************************************** +W0725 17:30:40.626000 2562640 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:30:40.626000 2562640 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:31:07.127148621 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:31:08.567784027 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 17:31:08.949504890 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:31:08.067043769 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +NCCL version 2.26.2+cuda12.2 +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:31:08.175891884 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +[rank6]:[W725 17:31:09.328414388 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:31:09.342311751 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:31:09.369808900 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed2_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed2_resampled_n116-040-168_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:24:01 acc1_pos: 33.594 acc5_pos: 100.000 loss: 1.1579 (1.1579) acc1: 33.5938 (33.5938) acc5: 100.0000 (100.0000) time: 5.3600 data: 3.7669 max mem: 3531 +Test: [ 10/269] eta: 0:02:27 acc1_pos: 52.865 acc5_pos: 100.000 loss: 1.1516 (1.1148) acc1: 31.5104 (32.1023) acc5: 100.0000 (100.0000) time: 0.5696 data: 0.3429 max mem: 3531 +Test: [ 20/269] eta: 0:01:26 acc1_pos: 41.927 acc5_pos: 100.000 loss: 1.0028 (1.0526) acc1: 33.5938 (36.1607) acc5: 100.0000 (100.0000) time: 0.0953 data: 0.0004 max mem: 3531 +Test: [ 30/269] eta: 0:01:09 acc1_pos: 40.885 acc5_pos: 100.000 loss: 0.9131 (1.0432) acc1: 40.8854 (37.0884) acc5: 100.0000 (100.0000) time: 0.1375 data: 0.0403 max mem: 3531 +Test: [ 40/269] eta: 0:01:02 acc1_pos: 20.690 acc5_pos: 100.000 loss: 0.6036 (0.8975) acc1: 62.7604 (47.4721) acc5: 100.0000 (100.0000) acc1_neg: 73.958 acc5_neg: 100.000 time: 0.1949 data: 0.0978 max mem: 3531 +Test: [ 50/269] eta: 0:00:56 acc1_pos: 20.690 acc5_pos: 100.000 loss: 0.3591 (0.7947) acc1: 85.9375 (55.2441) acc5: 100.0000 (100.0000) acc1_neg: 79.427 acc5_neg: 100.000 time: 0.2024 data: 0.1057 max mem: 3531 +Test: [ 60/269] eta: 0:00:50 acc1_pos: 20.690 acc5_pos: 100.000 loss: 0.3952 (0.7392) acc1: 84.8958 (59.0719) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.1814 data: 0.0896 max mem: 3531 +Test: [ 70/269] eta: 0:00:48 acc1_pos: 20.690 acc5_pos: 100.000 loss: 0.4618 (0.6979) acc1: 79.1667 (62.1662) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.2174 data: 0.1267 max mem: 3531 +Test: [ 80/269] eta: 0:00:46 acc1_pos: 20.690 acc5_pos: 100.000 loss: 0.4099 (0.6564) acc1: 84.3750 (65.4096) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2634 data: 0.1725 max mem: 3531 +Test: [ 90/269] eta: 0:00:43 acc1_pos: 20.690 acc5_pos: 100.000 loss: 0.3814 (0.6274) acc1: 87.5000 (67.7427) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.2324 data: 0.1377 max mem: 3531 +Test: [100/269] eta: 0:00:39 acc1_pos: 20.690 acc5_pos: 100.000 loss: 0.4082 (0.6074) acc1: 83.8542 (69.0671) acc5: 100.0000 (100.0000) acc1_neg: 77.344 acc5_neg: 100.000 time: 0.1906 data: 0.0936 max mem: 3531 +Test: [110/269] eta: 0:00:36 acc1_pos: 20.690 acc5_pos: 100.000 loss: 0.4357 (0.5927) acc1: 79.9479 (70.1459) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.1709 data: 0.0770 max mem: 3531 +Test: [120/269] eta: 0:00:34 acc1_pos: 20.690 acc5_pos: 100.000 loss: 0.4259 (0.5792) acc1: 80.7292 (70.9797) acc5: 100.0000 (100.0000) acc1_neg: 61.979 acc5_neg: 100.000 time: 0.2005 data: 0.1061 max mem: 3531 +Test: [130/269] eta: 0:00:32 acc1_pos: 20.690 acc5_pos: 100.000 loss: 0.4259 (0.5780) acc1: 79.9479 (71.0122) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.2449 data: 0.1503 max mem: 3531 +Test: [140/269] eta: 0:00:29 acc1_pos: 20.690 acc5_pos: 100.000 loss: 0.5633 (0.5795) acc1: 72.9167 (70.7188) acc5: 100.0000 (100.0000) acc1_neg: 56.771 acc5_neg: 100.000 time: 0.1950 data: 0.1001 max mem: 3531 +Test: [150/269] eta: 0:00:26 acc1_pos: 20.690 acc5_pos: 100.000 loss: 0.5633 (0.5763) acc1: 61.4583 (70.7023) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.1697 data: 0.0737 max mem: 3531 +Test: [160/269] eta: 0:00:24 acc1_pos: 20.690 acc5_pos: 100.000 loss: 0.4960 (0.5703) acc1: 73.1771 (70.9886) acc5: 100.0000 (100.0000) acc1_neg: 75.521 acc5_neg: 100.000 time: 0.2510 data: 0.1580 max mem: 3531 +Test: [170/269] eta: 0:00:22 acc1_pos: 20.690 acc5_pos: 100.000 loss: 0.5037 (0.5669) acc1: 73.1771 (71.1242) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.2412 data: 0.1500 max mem: 3531 +Test: [180/269] eta: 0:00:19 acc1_pos: 20.690 acc5_pos: 100.000 loss: 0.3554 (0.5521) acc1: 80.4688 (72.1498) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.1623 data: 0.0708 max mem: 3531 +Test: [190/269] eta: 0:00:17 acc1_pos: 20.690 acc5_pos: 100.000 loss: 0.3439 (0.5451) acc1: 85.9375 (72.5990) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.1849 data: 0.0944 max mem: 3531 +Test: [200/269] eta: 0:00:15 acc1_pos: 20.690 acc5_pos: 100.000 loss: 0.4713 (0.5454) acc1: 78.1250 (72.4762) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.2017 data: 0.1108 max mem: 3531 +Test: [210/269] eta: 0:00:12 acc1_pos: 20.690 acc5_pos: 100.000 loss: 0.5432 (0.5452) acc1: 71.0938 (72.4711) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.1947 data: 0.1035 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 20.690 acc5_pos: 100.000 loss: 0.5405 (0.5418) acc1: 72.3958 (72.6327) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2132 data: 0.1209 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 20.690 acc5_pos: 100.000 loss: 0.2780 (0.5294) acc1: 90.6250 (73.4815) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1861 data: 0.0941 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 20.690 acc5_pos: 100.000 loss: 0.3512 (0.5299) acc1: 89.5833 (73.3975) acc5: 100.0000 (100.0000) acc1_neg: 55.990 acc5_neg: 100.000 time: 0.1454 data: 0.0548 max mem: 3532 +Test: [250/269] eta: 0:00:03 acc1_pos: 20.690 acc5_pos: 100.000 loss: 0.6111 (0.5358) acc1: 65.8854 (72.8627) acc5: 100.0000 (100.0000) acc1_neg: 43.229 acc5_neg: 100.000 time: 0.1265 data: 0.0360 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 20.690 acc5_pos: 100.000 loss: 0.6456 (0.5404) acc1: 60.6771 (72.4188) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1107 data: 0.0201 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 20.690 acc5_pos: 100.000 loss: 0.5238 (0.5382) acc1: 72.1354 (72.5993) acc5: 100.0000 (100.0000) acc1_neg: 91.339 acc5_neg: 100.000 time: 0.1074 data: 0.0135 max mem: 3532 +Test: Total time: 0:00:54 (0.2027 s / it) +* Overall Acc@1 72.599 Acc@5 100.000 + target == real Acc@1 36.972 Acc@5 100.000 + target == fake Acc@1 77.462 Acc@5 100.000 + loss 0.538 + PR-AUC 0.9110 ROC-AUC 0.6113 EER 42.20% (thr=0.6587) +Accuracy of the network on the 103293 test images: 72.6% +[rank0]:[W725 17:32:13.770842317 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-25T17:32:19Z] eval_done tag=fixed500k_real_n5_seed2_resampled +[2026-07-25T17:32:19Z] eval_skip tag=fixed500k_fake_n11_seed2_resampled reason=complete output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed2_resampled +[2026-07-25T17:32:19Z] eval_start tag=fixed500k_real_n5_seed3_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed3_resampled_n116-040-168_20260725_112559/best_checkpoint.pth +started_utc=2026-07-25T17:32:20Z +hostname=n116-040-168 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed3_resampled_n116-040-168_20260725_112559/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed3_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0725 17:32:32.303000 2563254 torch/distributed/run.py:766] +W0725 17:32:32.303000 2563254 torch/distributed/run.py:766] ***************************************** +W0725 17:32:32.303000 2563254 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:32:32.303000 2563254 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W725 17:33:00.107101671 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:33:01.186320255 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +[rank1]:[W725 17:33:01.247632752 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:33:01.253413837 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:33:01.456831908 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:33:01.568520093 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:33:01.570647328 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:33:01.589692348 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed3_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed3_resampled_n116-040-168_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:22:56 acc1_pos: 67.969 acc5_pos: 100.000 loss: 0.6233 (0.6233) acc1: 67.9688 (67.9688) acc5: 100.0000 (100.0000) time: 5.8822 data: 4.2881 max mem: 3531 +Test: [ 10/234] eta: 0:02:18 acc1_pos: 89.323 acc5_pos: 100.000 loss: 0.4525 (0.4454) acc1: 79.6875 (79.0009) acc5: 100.0000 (100.0000) time: 0.6176 data: 0.3902 max mem: 3531 +Test: [ 20/234] eta: 0:01:19 acc1_pos: 79.948 acc5_pos: 100.000 loss: 0.3986 (0.4279) acc1: 84.8958 (80.8160) acc5: 100.0000 (100.0000) time: 0.0955 data: 0.0041 max mem: 3531 +Test: [ 30/234] eta: 0:01:13 acc1_pos: 65.885 acc5_pos: 100.000 loss: 0.3908 (0.5053) acc1: 84.1146 (77.2681) acc5: 100.0000 (100.0000) time: 0.2176 data: 0.1233 max mem: 3531 +Test: [ 40/234] eta: 0:01:05 acc1_pos: 27.344 acc5_pos: 100.000 loss: 0.7029 (0.6084) acc1: 59.6354 (69.8234) acc5: 100.0000 (100.0000) time: 0.3056 data: 0.2100 max mem: 3531 +Test: [ 50/234] eta: 0:00:57 acc1_pos: 31.510 acc5_pos: 100.000 loss: 1.2195 (0.7363) acc1: 27.3438 (61.2694) acc5: 100.0000 (100.0000) time: 0.2372 data: 0.1411 max mem: 3531 +Test: [ 60/234] eta: 0:00:50 acc1_pos: 20.000 acc5_pos: 100.000 loss: 0.5964 (0.6404) acc1: 38.5417 (67.1064) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1862 data: 0.0896 max mem: 3531 +Test: [ 70/234] eta: 0:00:45 acc1_pos: 20.000 acc5_pos: 100.000 loss: 0.1061 (0.5749) acc1: 99.7396 (71.3028) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1949 data: 0.1009 max mem: 3531 +Test: [ 80/234] eta: 0:00:41 acc1_pos: 20.000 acc5_pos: 100.000 loss: 0.2085 (0.5487) acc1: 95.5729 (73.0549) acc5: 100.0000 (100.0000) acc1_neg: 66.927 acc5_neg: 100.000 time: 0.2152 data: 0.1226 max mem: 3531 +Test: [ 90/234] eta: 0:00:36 acc1_pos: 20.000 acc5_pos: 100.000 loss: 0.2227 (0.5141) acc1: 94.0104 (75.3491) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1717 data: 0.0802 max mem: 3531 +Test: [100/234] eta: 0:00:33 acc1_pos: 20.000 acc5_pos: 100.000 loss: 0.2729 (0.4942) acc1: 90.6250 (76.6063) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.1637 data: 0.0738 max mem: 3531 +Test: [110/234] eta: 0:00:30 acc1_pos: 20.000 acc5_pos: 100.000 loss: 0.3613 (0.4851) acc1: 83.8542 (77.1349) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.1887 data: 0.0986 max mem: 3531 +Test: [120/234] eta: 0:00:27 acc1_pos: 20.000 acc5_pos: 100.000 loss: 0.4292 (0.4935) acc1: 78.3854 (76.4893) acc5: 100.0000 (100.0000) acc1_neg: 68.229 acc5_neg: 100.000 time: 0.1791 data: 0.0885 max mem: 3531 +Test: [130/234] eta: 0:00:24 acc1_pos: 20.000 acc5_pos: 100.000 loss: 0.6031 (0.5127) acc1: 65.3646 (75.1531) acc5: 100.0000 (100.0000) acc1_neg: 69.531 acc5_neg: 100.000 time: 0.1706 data: 0.0776 max mem: 3531 +Test: [140/234] eta: 0:00:21 acc1_pos: 20.000 acc5_pos: 100.000 loss: 0.5883 (0.5098) acc1: 70.0521 (75.3860) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.1764 data: 0.0833 max mem: 3531 +Test: [150/234] eta: 0:00:19 acc1_pos: 20.000 acc5_pos: 100.000 loss: 0.4108 (0.4985) acc1: 81.7708 (76.1141) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1889 data: 0.0981 max mem: 3531 +Test: [160/234] eta: 0:00:16 acc1_pos: 20.000 acc5_pos: 100.000 loss: 0.4421 (0.5052) acc1: 79.6875 (75.6438) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.2002 data: 0.1095 max mem: 3531 +Test: [170/234] eta: 0:00:14 acc1_pos: 20.000 acc5_pos: 100.000 loss: 0.4561 (0.5076) acc1: 76.5625 (75.4843) acc5: 100.0000 (100.0000) acc1_neg: 47.917 acc5_neg: 100.000 time: 0.2157 data: 0.1229 max mem: 3531 +Test: [180/234] eta: 0:00:12 acc1_pos: 20.000 acc5_pos: 100.000 loss: 0.4158 (0.5022) acc1: 81.5104 (75.8187) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.2092 data: 0.1158 max mem: 3531 +Test: [190/234] eta: 0:00:09 acc1_pos: 20.000 acc5_pos: 100.000 loss: 0.4078 (0.4979) acc1: 82.5521 (76.1480) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.2012 data: 0.1075 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 20.000 acc5_pos: 100.000 loss: 0.4024 (0.4845) acc1: 83.0729 (76.9836) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1850 data: 0.0912 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 20.000 acc5_pos: 100.000 loss: 0.1135 (0.4669) acc1: 99.7396 (78.0707) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1625 data: 0.0703 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 20.000 acc5_pos: 100.000 loss: 0.1118 (0.4509) acc1: 99.7396 (79.0535) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1435 data: 0.0522 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 20.000 acc5_pos: 100.000 loss: 0.1177 (0.4370) acc1: 99.4792 (79.9062) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1317 data: 0.0418 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 20.000 acc5_pos: 100.000 loss: 0.1236 (0.4331) acc1: 99.2188 (80.1201) acc5: 100.0000 (100.0000) acc1_neg: 99.222 acc5_neg: 100.000 time: 0.1236 data: 0.0329 max mem: 3532 +Test: Total time: 0:00:48 (0.2087 s / it) +* Overall Acc@1 80.120 Acc@5 100.000 + target == real Acc@1 60.966 Acc@5 100.000 + target == fake Acc@1 85.519 Acc@5 100.000 + loss 0.433 + PR-AUC 0.9435 ROC-AUC 0.8321 EER 25.93% (thr=0.7017) +Accuracy of the network on the 89729 test images: 80.1% +[rank0]:[W725 17:34:00.565908274 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 17:34:18.690000 2563831 torch/distributed/run.py:766] +W0725 17:34:18.690000 2563831 torch/distributed/run.py:766] ***************************************** +W0725 17:34:18.690000 2563831 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:34:18.690000 2563831 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:34:45.350910915 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 17:34:45.688465026 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:34:45.023058039 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:34:45.087717665 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:34:45.160565793 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:34:46.252557061 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:34:46.466372404 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:34:46.567506243 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed3_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed3_resampled_n116-040-168_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:31:29 acc1_pos: 46.615 acc5_pos: 100.000 loss: 0.9287 (0.9287) acc1: 46.6146 (46.6146) acc5: 100.0000 (100.0000) time: 6.8244 data: 5.1174 max mem: 3531 +Test: [ 10/2387] eta: 0:27:54 acc1_pos: 51.823 acc5_pos: 100.000 loss: 0.9668 (1.0143) acc1: 46.6146 (40.2225) acc5: 100.0000 (100.0000) time: 0.7044 data: 0.4657 max mem: 3531 +Test: [ 20/2387] eta: 0:16:23 acc1_pos: 58.333 acc5_pos: 100.000 loss: 0.8298 (0.8960) acc1: 50.7812 (48.2639) acc5: 100.0000 (100.0000) time: 0.0950 data: 0.0005 max mem: 3531 +Test: [ 30/2387] eta: 0:13:58 acc1_pos: 14.062 acc5_pos: 100.000 loss: 0.8477 (0.9764) acc1: 42.9688 (42.8931) acc5: 100.0000 (100.0000) time: 0.1641 data: 0.0671 max mem: 3531 +Test: [ 40/2387] eta: 0:12:38 acc1_pos: 40.885 acc5_pos: 100.000 loss: 1.1808 (1.0244) acc1: 23.4375 (38.6623) acc5: 100.0000 (100.0000) time: 0.2260 data: 0.1313 max mem: 3531 +Test: [ 50/2387] eta: 0:11:37 acc1_pos: 62.500 acc5_pos: 100.000 loss: 1.0492 (0.9975) acc1: 32.0312 (39.8591) acc5: 100.0000 (100.0000) time: 0.2099 data: 0.1187 max mem: 3531 +Test: [ 60/2387] eta: 0:10:56 acc1_pos: 70.573 acc5_pos: 100.000 loss: 0.7569 (0.9544) acc1: 55.2083 (42.7083) acc5: 100.0000 (100.0000) time: 0.1988 data: 0.1081 max mem: 3531 +Test: [ 70/2387] eta: 0:10:40 acc1_pos: 74.740 acc5_pos: 100.000 loss: 0.6850 (0.9215) acc1: 59.6354 (45.1695) acc5: 100.0000 (100.0000) time: 0.2196 data: 0.1254 max mem: 3531 +Test: [ 80/2387] eta: 0:10:18 acc1_pos: 55.729 acc5_pos: 100.000 loss: 0.6626 (0.9052) acc1: 60.6771 (46.1838) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1282 max mem: 3531 +Test: [ 90/2387] eta: 0:10:06 acc1_pos: 45.833 acc5_pos: 100.000 loss: 0.6626 (0.8767) acc1: 63.2812 (48.1141) acc5: 100.0000 (100.0000) time: 0.2199 data: 0.1238 max mem: 3531 +Test: [ 100/2387] eta: 0:09:55 acc1_pos: 78.125 acc5_pos: 100.000 loss: 0.6868 (0.8631) acc1: 60.9375 (49.1543) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1333 max mem: 3531 +Test: [ 110/2387] eta: 0:09:47 acc1_pos: 62.240 acc5_pos: 100.000 loss: 0.7163 (0.8505) acc1: 58.5938 (49.9977) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1339 max mem: 3531 +Test: [ 120/2387] eta: 0:09:37 acc1_pos: 76.562 acc5_pos: 100.000 loss: 0.6470 (0.8329) acc1: 60.4167 (51.2225) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1295 max mem: 3531 +Test: [ 130/2387] eta: 0:09:30 acc1_pos: 66.927 acc5_pos: 100.000 loss: 0.6470 (0.8302) acc1: 57.2917 (51.4671) acc5: 100.0000 (100.0000) time: 0.2249 data: 0.1266 max mem: 3531 +Test: [ 140/2387] eta: 0:09:24 acc1_pos: 70.833 acc5_pos: 100.000 loss: 0.7847 (0.8262) acc1: 51.8229 (51.7564) acc5: 100.0000 (100.0000) time: 0.2299 data: 0.1323 max mem: 3531 +Test: [ 150/2387] eta: 0:09:23 acc1_pos: 36.458 acc5_pos: 100.000 loss: 0.7847 (0.8220) acc1: 55.7292 (52.0920) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1492 max mem: 3531 +Test: [ 160/2387] eta: 0:09:26 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.6528 (0.8062) acc1: 64.5833 (53.3272) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.2760 data: 0.1811 max mem: 3531 +Test: [ 170/2387] eta: 0:09:10 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.3458 (0.7790) acc1: 86.4583 (55.3134) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.2207 data: 0.1267 max mem: 3531 +Test: [ 180/2387] eta: 0:08:57 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.3421 (0.7548) acc1: 86.9792 (57.0629) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1588 data: 0.0639 max mem: 3531 +Test: [ 190/2387] eta: 0:08:46 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.3399 (0.7330) acc1: 86.4583 (58.6456) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.1691 data: 0.0704 max mem: 3531 +Test: [ 200/2387] eta: 0:08:40 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.3352 (0.7132) acc1: 87.2396 (60.0863) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1849 data: 0.0864 max mem: 3531 +Test: [ 210/2387] eta: 0:08:32 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.3354 (0.6955) acc1: 86.9792 (61.3658) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1908 data: 0.0961 max mem: 3532 +Test: [ 220/2387] eta: 0:08:21 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.3343 (0.6791) acc1: 87.7604 (62.5931) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1649 data: 0.0705 max mem: 3532 +Test: [ 230/2387] eta: 0:08:11 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.3343 (0.6645) acc1: 88.2812 (63.6792) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.1517 data: 0.0581 max mem: 3532 +Test: [ 240/2387] eta: 0:08:05 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.3571 (0.6549) acc1: 85.6771 (64.3856) acc5: 100.0000 (100.0000) acc1_neg: 78.646 acc5_neg: 100.000 time: 0.1721 data: 0.0787 max mem: 3532 +Test: [ 250/2387] eta: 0:08:01 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.3385 (0.6360) acc1: 83.8542 (65.6665) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1942 data: 0.1013 max mem: 3532 +Test: [ 260/2387] eta: 0:07:56 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1608 (0.6178) acc1: 98.9583 (66.9241) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1978 data: 0.1070 max mem: 3532 +Test: [ 270/2387] eta: 0:07:52 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1546 (0.6007) acc1: 98.6979 (68.0889) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1992 data: 0.1088 max mem: 3532 +Test: [ 280/2387] eta: 0:07:49 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1465 (0.5848) acc1: 98.6979 (69.1717) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2058 data: 0.1149 max mem: 3532 +Test: [ 290/2387] eta: 0:07:46 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1314 (0.5692) acc1: 99.2188 (70.2132) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2092 data: 0.1184 max mem: 3532 +Test: [ 300/2387] eta: 0:07:43 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1396 (0.5558) acc1: 98.9583 (71.1327) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2088 data: 0.1176 max mem: 3532 +Test: [ 310/2387] eta: 0:07:40 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1670 (0.5520) acc1: 97.6562 (71.2671) acc5: 100.0000 (100.0000) acc1_neg: 47.917 acc5_neg: 100.000 time: 0.2092 data: 0.1172 max mem: 3532 +Test: [ 320/2387] eta: 0:07:36 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.8142 (0.5606) acc1: 45.3125 (70.4180) acc5: 100.0000 (100.0000) acc1_neg: 41.927 acc5_neg: 100.000 time: 0.2050 data: 0.1127 max mem: 3532 +Test: [ 330/2387] eta: 0:07:31 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.8237 (0.5684) acc1: 43.7500 (69.6288) acc5: 100.0000 (100.0000) acc1_neg: 47.917 acc5_neg: 100.000 time: 0.1914 data: 0.0977 max mem: 3532 +Test: [ 340/2387] eta: 0:07:27 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.8195 (0.5760) acc1: 44.0104 (68.8943) acc5: 100.0000 (100.0000) acc1_neg: 39.062 acc5_neg: 100.000 time: 0.1810 data: 0.0863 max mem: 3532 +Test: [ 350/2387] eta: 0:07:23 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.7950 (0.5678) acc1: 46.8750 (69.5320) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1885 data: 0.0958 max mem: 3532 +Test: [ 360/2387] eta: 0:07:20 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2723 (0.5599) acc1: 92.9688 (70.1444) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1948 data: 0.1017 max mem: 3532 +Test: [ 370/2387] eta: 0:07:14 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2839 (0.5525) acc1: 91.6667 (70.7259) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1709 data: 0.0783 max mem: 3532 +Test: [ 380/2387] eta: 0:07:11 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2868 (0.5454) acc1: 91.6667 (71.2797) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1798 data: 0.0871 max mem: 3532 +Test: [ 390/2387] eta: 0:07:07 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2858 (0.5388) acc1: 91.6667 (71.7884) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1938 data: 0.0981 max mem: 3532 +Test: [ 400/2387] eta: 0:07:01 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2851 (0.5325) acc1: 91.6667 (72.2770) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1539 data: 0.0583 max mem: 3532 +Test: [ 410/2387] eta: 0:06:59 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2889 (0.5266) acc1: 90.3646 (72.7298) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1699 data: 0.0750 max mem: 3532 +Test: [ 420/2387] eta: 0:06:57 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2802 (0.5205) acc1: 91.6667 (73.1969) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.2099 data: 0.1153 max mem: 3532 +Test: [ 430/2387] eta: 0:06:52 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2824 (0.5161) acc1: 91.4062 (73.5463) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1862 data: 0.0930 max mem: 3532 +Test: [ 440/2387] eta: 0:06:50 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2909 (0.5092) acc1: 87.7604 (74.0180) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1863 data: 0.0954 max mem: 3533 +Test: [ 450/2387] eta: 0:06:48 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1085 (0.5003) acc1: 100.0000 (74.5935) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2060 data: 0.1134 max mem: 3533 +Test: [ 460/2387] eta: 0:06:45 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1083 (0.4917) acc1: 100.0000 (75.1429) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2037 data: 0.1105 max mem: 3533 +Test: [ 470/2387] eta: 0:06:44 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1082 (0.4836) acc1: 100.0000 (75.6707) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2125 data: 0.1199 max mem: 3533 +Test: [ 480/2387] eta: 0:06:41 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1090 (0.4758) acc1: 100.0000 (76.1743) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2048 data: 0.1126 max mem: 3533 +Test: [ 490/2387] eta: 0:06:38 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1095 (0.4683) acc1: 100.0000 (76.6590) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1915 data: 0.1008 max mem: 3533 +Test: [ 500/2387] eta: 0:06:35 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1083 (0.4612) acc1: 100.0000 (77.1249) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1921 data: 0.0991 max mem: 3533 +Test: [ 510/2387] eta: 0:06:32 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1082 (0.4544) acc1: 100.0000 (77.5705) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1939 data: 0.1011 max mem: 3533 +Test: [ 520/2387] eta: 0:06:30 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1387 (0.4486) acc1: 100.0000 (77.9950) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2008 data: 0.1080 max mem: 3533 +Test: [ 530/2387] eta: 0:06:27 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1650 (0.4433) acc1: 99.4792 (78.4006) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1900 data: 0.0972 max mem: 3533 +Test: [ 540/2387] eta: 0:06:24 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1672 (0.4383) acc1: 99.4792 (78.7864) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1831 data: 0.0924 max mem: 3533 +Test: [ 550/2387] eta: 0:06:22 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1693 (0.4334) acc1: 99.2188 (79.1596) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1980 data: 0.1038 max mem: 3533 +Test: [ 560/2387] eta: 0:06:19 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1704 (0.4287) acc1: 99.2188 (79.5176) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1935 data: 0.0994 max mem: 3533 +Test: [ 570/2387] eta: 0:06:17 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1678 (0.4241) acc1: 99.4792 (79.8708) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1940 data: 0.1019 max mem: 3533 +Test: [ 580/2387] eta: 0:06:15 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1697 (0.4197) acc1: 99.7396 (80.2083) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2051 data: 0.1127 max mem: 3533 +Test: [ 590/2387] eta: 0:06:12 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1698 (0.4155) acc1: 99.4792 (80.5348) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1978 data: 0.1071 max mem: 3533 +Test: [ 600/2387] eta: 0:06:10 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1538 (0.4110) acc1: 99.7396 (80.8548) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1995 data: 0.1088 max mem: 3533 +Test: [ 610/2387] eta: 0:06:08 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1538 (0.4071) acc1: 98.6979 (81.1294) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2027 data: 0.1111 max mem: 3533 +Test: [ 620/2387] eta: 0:06:05 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1739 (0.4034) acc1: 97.9167 (81.4010) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1849 data: 0.0924 max mem: 3533 +Test: [ 630/2387] eta: 0:06:02 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1775 (0.3998) acc1: 97.9167 (81.6582) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1741 data: 0.0827 max mem: 3533 +Test: [ 640/2387] eta: 0:05:59 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1790 (0.3964) acc1: 97.9167 (81.9106) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1839 data: 0.0909 max mem: 3533 +Test: [ 650/2387] eta: 0:05:57 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1790 (0.3931) acc1: 97.6562 (82.1517) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1951 data: 0.1020 max mem: 3533 +Test: [ 660/2387] eta: 0:05:55 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1776 (0.3898) acc1: 97.3958 (82.3882) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2060 data: 0.1145 max mem: 3533 +Test: [ 670/2387] eta: 0:05:53 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1771 (0.3866) acc1: 97.9167 (82.6208) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2096 data: 0.1167 max mem: 3534 +Test: [ 680/2387] eta: 0:05:51 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1790 (0.3840) acc1: 98.4375 (82.8595) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2048 data: 0.1112 max mem: 3534 +Test: [ 690/2387] eta: 0:05:49 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2245 (0.3818) acc1: 98.4375 (83.0149) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.2121 data: 0.1190 max mem: 3534 +Test: [ 700/2387] eta: 0:05:47 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2247 (0.3794) acc1: 92.7083 (83.1576) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2020 data: 0.1111 max mem: 3534 +Test: [ 710/2387] eta: 0:05:46 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2195 (0.3771) acc1: 92.9688 (83.3007) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2126 data: 0.1200 max mem: 3534 +Test: [ 720/2387] eta: 0:05:43 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2112 (0.3747) acc1: 93.2292 (83.4413) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2228 data: 0.1280 max mem: 3534 +Test: [ 730/2387] eta: 0:05:41 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2129 (0.3727) acc1: 92.7083 (83.5617) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1947 data: 0.0992 max mem: 3534 +Test: [ 740/2387] eta: 0:05:39 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2136 (0.3705) acc1: 92.4479 (83.6939) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2077 data: 0.1138 max mem: 3534 +Test: [ 750/2387] eta: 0:05:37 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2074 (0.3684) acc1: 93.2292 (83.8198) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2124 data: 0.1197 max mem: 3534 +Test: [ 760/2387] eta: 0:05:35 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2073 (0.3660) acc1: 92.9688 (83.9568) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2088 data: 0.1158 max mem: 3534 +Test: [ 770/2387] eta: 0:05:33 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1566 (0.3633) acc1: 96.6146 (84.1254) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2179 data: 0.1261 max mem: 3534 +Test: [ 780/2387] eta: 0:05:31 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1526 (0.3607) acc1: 96.6146 (84.2846) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2112 data: 0.1187 max mem: 3534 +Test: [ 790/2387] eta: 0:05:30 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1533 (0.3584) acc1: 96.8750 (84.4445) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2169 data: 0.1245 max mem: 3534 +Test: [ 800/2387] eta: 0:05:27 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1585 (0.3562) acc1: 97.6562 (84.5945) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1919 data: 0.1005 max mem: 3534 +Test: [ 810/2387] eta: 0:05:24 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1372 (0.3537) acc1: 99.2188 (84.7594) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1757 data: 0.0847 max mem: 3534 +Test: [ 820/2387] eta: 0:05:22 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1428 (0.3513) acc1: 98.9583 (84.9139) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1870 data: 0.0946 max mem: 3534 +Test: [ 830/2387] eta: 0:05:19 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1986 (0.3510) acc1: 94.5312 (84.9482) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1773 data: 0.0845 max mem: 3534 +Test: [ 840/2387] eta: 0:05:17 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2999 (0.3506) acc1: 88.8021 (84.9986) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.1735 data: 0.0816 max mem: 3534 +Test: [ 850/2387] eta: 0:05:14 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.3114 (0.3506) acc1: 87.7604 (85.0069) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1788 data: 0.0850 max mem: 3534 +Test: [ 860/2387] eta: 0:05:11 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.3114 (0.3504) acc1: 87.2396 (85.0301) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.1771 data: 0.0839 max mem: 3534 +Test: [ 870/2387] eta: 0:05:09 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2996 (0.3499) acc1: 89.5833 (85.0803) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1769 data: 0.0821 max mem: 3534 +Test: [ 880/2387] eta: 0:05:07 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2996 (0.3499) acc1: 87.7604 (85.1045) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.1840 data: 0.0898 max mem: 3534 +Test: [ 890/2387] eta: 0:05:04 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.3684 (0.3504) acc1: 86.4583 (85.0747) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1822 data: 0.0903 max mem: 3534 +Test: [ 900/2387] eta: 0:05:02 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.4077 (0.3509) acc1: 81.7708 (85.0508) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.1805 data: 0.0864 max mem: 3535 +Test: [ 910/2387] eta: 0:04:59 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.4147 (0.3516) acc1: 82.0312 (85.0273) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.1812 data: 0.0882 max mem: 3535 +Test: [ 920/2387] eta: 0:04:58 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.4211 (0.3523) acc1: 82.0312 (84.9985) acc5: 100.0000 (100.0000) acc1_neg: 80.469 acc5_neg: 100.000 time: 0.2133 data: 0.1215 max mem: 3535 +Test: [ 930/2387] eta: 0:04:56 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.4135 (0.3530) acc1: 82.5521 (84.9705) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.2109 data: 0.1183 max mem: 3535 +Test: [ 940/2387] eta: 0:04:52 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.4138 (0.3537) acc1: 81.5104 (84.9379) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.1381 data: 0.0455 max mem: 3535 +Test: [ 950/2387] eta: 0:04:50 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.4103 (0.3542) acc1: 82.0312 (84.9147) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.1372 data: 0.0434 max mem: 3535 +Test: [ 960/2387] eta: 0:04:47 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.4027 (0.3547) acc1: 82.2917 (84.8988) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.1806 data: 0.0868 max mem: 3535 +Test: [ 970/2387] eta: 0:04:45 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.3985 (0.3553) acc1: 82.5521 (84.8722) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.1839 data: 0.0907 max mem: 3535 +Test: [ 980/2387] eta: 0:04:43 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.3899 (0.3549) acc1: 84.1146 (84.9057) acc5: 100.0000 (100.0000) acc1_neg: 62.240 acc5_neg: 100.000 time: 0.1751 data: 0.0805 max mem: 3535 +Test: [ 990/2387] eta: 0:04:40 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1807 (0.3545) acc1: 93.4896 (84.9318) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1735 data: 0.0796 max mem: 3535 +Test: [1000/2387] eta: 0:04:38 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1801 (0.3529) acc1: 96.3542 (85.0337) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1786 data: 0.0853 max mem: 3535 +Test: [1010/2387] eta: 0:04:35 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2054 (0.3517) acc1: 94.7917 (85.1174) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1764 data: 0.0809 max mem: 3535 +Test: [1020/2387] eta: 0:04:33 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2126 (0.3505) acc1: 94.5312 (85.2070) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1703 data: 0.0765 max mem: 3535 +Test: [1030/2387] eta: 0:04:31 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1579 (0.3487) acc1: 97.6562 (85.3232) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1758 data: 0.0842 max mem: 3535 +Test: [1040/2387] eta: 0:04:29 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1559 (0.3469) acc1: 97.6562 (85.4384) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1861 data: 0.0931 max mem: 3535 +Test: [1050/2387] eta: 0:04:26 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1567 (0.3452) acc1: 97.6562 (85.5557) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1800 data: 0.0882 max mem: 3535 +Test: [1060/2387] eta: 0:04:24 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1480 (0.3435) acc1: 97.9167 (85.6680) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1775 data: 0.0863 max mem: 3535 +Test: [1070/2387] eta: 0:04:22 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1480 (0.3422) acc1: 98.4375 (85.7556) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.1867 data: 0.0952 max mem: 3535 +Test: [1080/2387] eta: 0:04:20 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2905 (0.3421) acc1: 89.3229 (85.7732) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1947 data: 0.1001 max mem: 3535 +Test: [1090/2387] eta: 0:04:17 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2311 (0.3407) acc1: 92.7083 (85.8661) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.1750 data: 0.0794 max mem: 3535 +Test: [1100/2387] eta: 0:04:15 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1869 (0.3395) acc1: 96.0938 (85.9453) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1699 data: 0.0767 max mem: 3535 +Test: [1110/2387] eta: 0:04:13 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2191 (0.3388) acc1: 92.4479 (85.9996) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1714 data: 0.0790 max mem: 3535 +Test: [1120/2387] eta: 0:04:11 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2319 (0.3378) acc1: 91.6667 (86.0706) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1616 data: 0.0705 max mem: 3536 +Test: [1130/2387] eta: 0:04:08 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1812 (0.3363) acc1: 96.8750 (86.1765) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1682 data: 0.0778 max mem: 3536 +Test: [1140/2387] eta: 0:04:06 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1868 (0.3363) acc1: 95.3125 (86.1986) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1736 data: 0.0799 max mem: 3536 +Test: [1150/2387] eta: 0:04:04 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2752 (0.3358) acc1: 92.7083 (86.2540) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1812 data: 0.0868 max mem: 3536 +Test: [1160/2387] eta: 0:04:02 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2293 (0.3346) acc1: 96.3542 (86.3433) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1756 data: 0.0845 max mem: 3536 +Test: [1170/2387] eta: 0:03:59 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2043 (0.3336) acc1: 96.3542 (86.4214) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0844 max mem: 3536 +Test: [1180/2387] eta: 0:03:57 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2043 (0.3326) acc1: 96.0938 (86.5005) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1750 data: 0.0830 max mem: 3536 +Test: [1190/2387] eta: 0:03:55 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1826 (0.3313) acc1: 97.1354 (86.5950) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1736 data: 0.0802 max mem: 3536 +Test: [1200/2387] eta: 0:03:53 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1780 (0.3304) acc1: 97.1354 (86.6691) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1754 data: 0.0810 max mem: 3536 +Test: [1210/2387] eta: 0:03:51 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1816 (0.3295) acc1: 96.0938 (86.7252) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1777 data: 0.0837 max mem: 3536 +Test: [1220/2387] eta: 0:03:49 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1662 (0.3282) acc1: 96.0938 (86.8117) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1814 data: 0.0886 max mem: 3536 +Test: [1230/2387] eta: 0:03:46 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1566 (0.3268) acc1: 97.6562 (86.9011) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1748 data: 0.0829 max mem: 3536 +Test: [1240/2387] eta: 0:03:44 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1566 (0.3255) acc1: 97.9167 (86.9846) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1798 data: 0.0880 max mem: 3536 +Test: [1250/2387] eta: 0:03:42 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1367 (0.3240) acc1: 98.4375 (87.0785) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1800 data: 0.0883 max mem: 3536 +Test: [1260/2387] eta: 0:03:40 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1367 (0.3226) acc1: 98.4375 (87.1659) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1694 data: 0.0791 max mem: 3536 +Test: [1270/2387] eta: 0:03:38 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1524 (0.3215) acc1: 97.9167 (87.2437) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1752 data: 0.0849 max mem: 3536 +Test: [1280/2387] eta: 0:03:36 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1418 (0.3198) acc1: 98.4375 (87.3372) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1842 data: 0.0920 max mem: 3536 +Test: [1290/2387] eta: 0:03:34 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1097 (0.3183) acc1: 98.9583 (87.4215) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1818 data: 0.0882 max mem: 3536 +Test: [1300/2387] eta: 0:03:32 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1266 (0.3169) acc1: 98.1771 (87.5012) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1789 data: 0.0867 max mem: 3536 +Test: [1310/2387] eta: 0:03:30 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1266 (0.3154) acc1: 98.1771 (87.5844) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1854 data: 0.0943 max mem: 3536 +Test: [1320/2387] eta: 0:03:28 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1059 (0.3139) acc1: 99.4792 (87.6717) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2009 data: 0.1092 max mem: 3536 +Test: [1330/2387] eta: 0:03:26 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1195 (0.3132) acc1: 98.9583 (87.7187) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2034 data: 0.1109 max mem: 3536 +Test: [1340/2387] eta: 0:03:24 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2309 (0.3127) acc1: 94.7917 (87.7558) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1738 data: 0.0821 max mem: 3536 +Test: [1350/2387] eta: 0:03:21 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.0869 (0.3110) acc1: 100.0000 (87.8458) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1609 data: 0.0697 max mem: 3537 +Test: [1360/2387] eta: 0:03:19 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.0723 (0.3092) acc1: 100.0000 (87.9351) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1742 data: 0.0800 max mem: 3537 +Test: [1370/2387] eta: 0:03:17 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.0758 (0.3075) acc1: 100.0000 (88.0227) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1745 data: 0.0811 max mem: 3537 +Test: [1380/2387] eta: 0:03:15 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.0740 (0.3058) acc1: 100.0000 (88.1095) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1701 data: 0.0795 max mem: 3537 +Test: [1390/2387] eta: 0:03:13 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.0730 (0.3042) acc1: 100.0000 (88.1949) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1730 data: 0.0820 max mem: 3537 +Test: [1400/2387] eta: 0:03:11 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.0723 (0.3025) acc1: 100.0000 (88.2786) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1780 data: 0.0870 max mem: 3537 +Test: [1410/2387] eta: 0:03:09 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.0743 (0.3015) acc1: 100.0000 (88.3392) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1871 data: 0.0959 max mem: 3537 +Test: [1420/2387] eta: 0:03:07 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2274 (0.3013) acc1: 92.4479 (88.3485) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1816 data: 0.0898 max mem: 3537 +Test: [1430/2387] eta: 0:03:05 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2632 (0.3009) acc1: 91.9271 (88.3752) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1915 data: 0.0995 max mem: 3537 +Test: [1440/2387] eta: 0:03:03 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2237 (0.3005) acc1: 93.2292 (88.4117) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.1931 data: 0.1010 max mem: 3537 +Test: [1450/2387] eta: 0:03:01 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2656 (0.3005) acc1: 91.9271 (88.4198) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1918 data: 0.1001 max mem: 3537 +Test: [1460/2387] eta: 0:02:59 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2161 (0.2997) acc1: 94.5312 (88.4787) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2020 data: 0.1116 max mem: 3537 +Test: [1470/2387] eta: 0:02:58 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2005 (0.2991) acc1: 96.8750 (88.5268) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2048 data: 0.1136 max mem: 3537 +Test: [1480/2387] eta: 0:02:56 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1925 (0.2984) acc1: 95.5729 (88.5751) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1978 data: 0.1055 max mem: 3537 +Test: [1490/2387] eta: 0:02:53 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1803 (0.2976) acc1: 97.1354 (88.6351) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1786 data: 0.0866 max mem: 3537 +Test: [1500/2387] eta: 0:02:51 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1749 (0.2968) acc1: 98.4375 (88.6997) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1818 data: 0.0911 max mem: 3537 +Test: [1510/2387] eta: 0:02:49 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1790 (0.2963) acc1: 98.6979 (88.7456) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1782 data: 0.0885 max mem: 3537 +Test: [1520/2387] eta: 0:02:47 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2109 (0.2958) acc1: 95.8333 (88.7783) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1724 data: 0.0798 max mem: 3537 +Test: [1530/2387] eta: 0:02:45 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2458 (0.2957) acc1: 93.2292 (88.7909) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1798 data: 0.0871 max mem: 3537 +Test: [1540/2387] eta: 0:02:43 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2503 (0.2954) acc1: 93.2292 (88.8212) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.1803 data: 0.0892 max mem: 3537 +Test: [1550/2387] eta: 0:02:41 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2442 (0.2951) acc1: 94.2708 (88.8496) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1813 data: 0.0900 max mem: 3537 +Test: [1560/2387] eta: 0:02:39 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2386 (0.2948) acc1: 94.7917 (88.8818) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1777 data: 0.0867 max mem: 3537 +Test: [1570/2387] eta: 0:02:37 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2812 (0.2950) acc1: 91.6667 (88.8770) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1809 data: 0.0869 max mem: 3537 +Test: [1580/2387] eta: 0:02:35 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2489 (0.2943) acc1: 93.2292 (88.9223) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1952 data: 0.1008 max mem: 3538 +Test: [1590/2387] eta: 0:02:34 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1693 (0.2935) acc1: 96.0938 (88.9697) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1934 data: 0.1016 max mem: 3538 +Test: [1600/2387] eta: 0:02:32 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1771 (0.2930) acc1: 96.0938 (89.0096) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1941 data: 0.1033 max mem: 3538 +Test: [1610/2387] eta: 0:02:30 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1771 (0.2922) acc1: 96.6146 (89.0601) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1895 data: 0.0993 max mem: 3538 +Test: [1620/2387] eta: 0:02:28 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1564 (0.2915) acc1: 97.3958 (89.1051) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1848 data: 0.0948 max mem: 3538 +Test: [1630/2387] eta: 0:02:26 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1837 (0.2911) acc1: 97.1354 (89.1345) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.2153 data: 0.1246 max mem: 3538 +Test: [1640/2387] eta: 0:02:24 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2770 (0.2912) acc1: 91.9271 (89.1431) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.1995 data: 0.1076 max mem: 3538 +Test: [1650/2387] eta: 0:02:22 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2807 (0.2911) acc1: 90.6250 (89.1524) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1736 data: 0.0799 max mem: 3538 +Test: [1660/2387] eta: 0:02:20 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2781 (0.2911) acc1: 90.8854 (89.1581) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1904 data: 0.0973 max mem: 3538 +Test: [1670/2387] eta: 0:02:18 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2774 (0.2910) acc1: 90.8854 (89.1758) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1862 data: 0.0956 max mem: 3538 +Test: [1680/2387] eta: 0:02:16 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2741 (0.2912) acc1: 90.6250 (89.1730) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1853 data: 0.0946 max mem: 3538 +Test: [1690/2387] eta: 0:02:14 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2824 (0.2912) acc1: 90.1042 (89.1743) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1863 data: 0.0955 max mem: 3538 +Test: [1700/2387] eta: 0:02:12 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2813 (0.2911) acc1: 91.4062 (89.1902) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1914 data: 0.0995 max mem: 3538 +Test: [1710/2387] eta: 0:02:10 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2269 (0.2907) acc1: 93.4896 (89.2174) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2188 data: 0.1271 max mem: 3538 +Test: [1720/2387] eta: 0:02:08 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2001 (0.2902) acc1: 94.7917 (89.2485) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1978 data: 0.1073 max mem: 3538 +Test: [1730/2387] eta: 0:02:06 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2106 (0.2898) acc1: 94.5312 (89.2731) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1761 data: 0.0855 max mem: 3538 +Test: [1740/2387] eta: 0:02:05 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1845 (0.2892) acc1: 95.8333 (89.3147) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1976 data: 0.1071 max mem: 3538 +Test: [1750/2387] eta: 0:02:03 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1890 (0.2887) acc1: 96.3542 (89.3495) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1924 data: 0.1023 max mem: 3538 +Test: [1760/2387] eta: 0:02:01 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2177 (0.2884) acc1: 94.2708 (89.3731) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1899 data: 0.0983 max mem: 3538 +Test: [1770/2387] eta: 0:01:59 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2320 (0.2880) acc1: 93.4896 (89.4011) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1866 data: 0.0932 max mem: 3538 +Test: [1780/2387] eta: 0:01:57 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2364 (0.2877) acc1: 91.9271 (89.4161) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1877 data: 0.0957 max mem: 3538 +Test: [1790/2387] eta: 0:01:55 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2167 (0.2873) acc1: 93.4896 (89.4452) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2005 data: 0.1100 max mem: 3538 +Test: [1800/2387] eta: 0:01:53 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2119 (0.2870) acc1: 94.2708 (89.4617) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1856 data: 0.0947 max mem: 3538 +Test: [1810/2387] eta: 0:01:51 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1740 (0.2863) acc1: 96.3542 (89.5045) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1972 data: 0.1051 max mem: 3539 +Test: [1820/2387] eta: 0:01:49 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1759 (0.2859) acc1: 95.3125 (89.5277) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2284 data: 0.1357 max mem: 3539 +Test: [1830/2387] eta: 0:01:47 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1919 (0.2854) acc1: 94.7917 (89.5660) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2052 data: 0.1104 max mem: 3539 +Test: [1840/2387] eta: 0:01:45 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1499 (0.2846) acc1: 98.9583 (89.6188) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1799 data: 0.0859 max mem: 3539 +Test: [1850/2387] eta: 0:01:43 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1466 (0.2839) acc1: 99.4792 (89.6711) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1663 data: 0.0752 max mem: 3539 +Test: [1860/2387] eta: 0:01:41 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1465 (0.2831) acc1: 99.4792 (89.7233) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1775 data: 0.0864 max mem: 3539 +Test: [1870/2387] eta: 0:01:39 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1478 (0.2824) acc1: 99.2188 (89.7729) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1910 data: 0.0986 max mem: 3539 +Test: [1880/2387] eta: 0:01:37 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1481 (0.2817) acc1: 98.9583 (89.8220) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1836 data: 0.0917 max mem: 3539 +Test: [1890/2387] eta: 0:01:35 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1485 (0.2810) acc1: 98.9583 (89.8707) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1806 data: 0.0907 max mem: 3539 +Test: [1900/2387] eta: 0:01:33 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1489 (0.2803) acc1: 98.9583 (89.9181) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1791 data: 0.0888 max mem: 3539 +Test: [1910/2387] eta: 0:01:31 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1823 (0.2800) acc1: 97.9167 (89.9520) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1842 data: 0.0924 max mem: 3539 +Test: [1920/2387] eta: 0:01:30 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2016 (0.2797) acc1: 96.3542 (89.9734) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1830 data: 0.0908 max mem: 3539 +Test: [1930/2387] eta: 0:01:28 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2411 (0.2796) acc1: 94.7917 (89.9902) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.1843 data: 0.0927 max mem: 3539 +Test: [1940/2387] eta: 0:01:26 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2687 (0.2798) acc1: 94.7917 (89.9967) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.1919 data: 0.0995 max mem: 3539 +Test: [1950/2387] eta: 0:01:24 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2916 (0.2801) acc1: 94.0104 (89.9939) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1928 data: 0.0995 max mem: 3539 +Test: [1960/2387] eta: 0:01:22 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2428 (0.2798) acc1: 94.2708 (90.0168) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1943 data: 0.1021 max mem: 3539 +Test: [1970/2387] eta: 0:01:20 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2762 (0.2805) acc1: 86.9792 (89.9722) acc5: 100.0000 (100.0000) acc1_neg: 72.396 acc5_neg: 100.000 time: 0.1976 data: 0.1069 max mem: 3539 +Test: [1980/2387] eta: 0:01:18 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.3608 (0.2805) acc1: 82.5521 (89.9752) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1810 data: 0.0905 max mem: 3539 +Test: [1990/2387] eta: 0:01:16 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1112 (0.2796) acc1: 99.7396 (90.0254) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1702 data: 0.0795 max mem: 3539 +Test: [2000/2387] eta: 0:01:14 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1086 (0.2788) acc1: 100.0000 (90.0749) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1737 data: 0.0807 max mem: 3539 +Test: [2010/2387] eta: 0:01:12 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1095 (0.2779) acc1: 100.0000 (90.1237) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1822 data: 0.0885 max mem: 3539 +Test: [2020/2387] eta: 0:01:10 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1088 (0.2771) acc1: 100.0000 (90.1722) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2064 data: 0.1152 max mem: 3539 +Test: [2030/2387] eta: 0:01:08 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1081 (0.2762) acc1: 100.0000 (90.2201) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1894 data: 0.0980 max mem: 3540 +Test: [2040/2387] eta: 0:01:06 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1079 (0.2754) acc1: 100.0000 (90.2677) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1584 data: 0.0657 max mem: 3540 +Test: [2050/2387] eta: 0:01:04 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1082 (0.2746) acc1: 100.0000 (90.3146) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1676 data: 0.0753 max mem: 3540 +Test: [2060/2387] eta: 0:01:02 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1169 (0.2739) acc1: 99.7396 (90.3608) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1779 data: 0.0868 max mem: 3540 +Test: [2070/2387] eta: 0:01:00 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1178 (0.2731) acc1: 100.0000 (90.4071) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1836 data: 0.0915 max mem: 3540 +Test: [2080/2387] eta: 0:00:58 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1164 (0.2724) acc1: 100.0000 (90.4531) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1823 data: 0.0896 max mem: 3540 +Test: [2090/2387] eta: 0:00:57 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1182 (0.2716) acc1: 100.0000 (90.4983) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1758 data: 0.0849 max mem: 3540 +Test: [2100/2387] eta: 0:00:55 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1174 (0.2709) acc1: 100.0000 (90.5432) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1803 data: 0.0898 max mem: 3540 +Test: [2110/2387] eta: 0:00:53 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1176 (0.2702) acc1: 100.0000 (90.5877) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1755 data: 0.0836 max mem: 3540 +Test: [2120/2387] eta: 0:00:51 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1196 (0.2698) acc1: 99.7396 (90.6117) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.1664 data: 0.0744 max mem: 3540 +Test: [2130/2387] eta: 0:00:49 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1905 (0.2694) acc1: 95.8333 (90.6332) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1682 data: 0.0769 max mem: 3540 +Test: [2140/2387] eta: 0:00:47 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2063 (0.2692) acc1: 94.5312 (90.6533) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1680 data: 0.0767 max mem: 3540 +Test: [2150/2387] eta: 0:00:45 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2063 (0.2689) acc1: 94.5312 (90.6704) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1680 data: 0.0761 max mem: 3540 +Test: [2160/2387] eta: 0:00:43 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1624 (0.2684) acc1: 97.9167 (90.7019) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1653 data: 0.0733 max mem: 3540 +Test: [2170/2387] eta: 0:00:41 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1690 (0.2685) acc1: 97.1354 (90.7019) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1637 data: 0.0724 max mem: 3540 +Test: [2180/2387] eta: 0:00:39 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.3439 (0.2698) acc1: 84.8958 (90.6213) acc5: 100.0000 (100.0000) acc1_neg: 54.948 acc5_neg: 100.000 time: 0.1644 data: 0.0725 max mem: 3540 +Test: [2190/2387] eta: 0:00:37 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.3439 (0.2696) acc1: 84.8958 (90.6288) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1661 data: 0.0729 max mem: 3540 +Test: [2200/2387] eta: 0:00:35 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1833 (0.2693) acc1: 97.1354 (90.6539) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1627 data: 0.0701 max mem: 3540 +Test: [2210/2387] eta: 0:00:33 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2213 (0.2692) acc1: 94.7917 (90.6731) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1616 data: 0.0701 max mem: 3540 +Test: [2220/2387] eta: 0:00:31 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2403 (0.2691) acc1: 95.8333 (90.6917) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1658 data: 0.0741 max mem: 3540 +Test: [2230/2387] eta: 0:00:29 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1947 (0.2689) acc1: 97.3958 (90.7164) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1719 data: 0.0785 max mem: 3540 +Test: [2240/2387] eta: 0:00:27 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1819 (0.2685) acc1: 98.4375 (90.7490) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1772 data: 0.0827 max mem: 3540 +Test: [2250/2387] eta: 0:00:26 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2226 (0.2687) acc1: 95.3125 (90.7399) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.2011 data: 0.1086 max mem: 3540 +Test: [2260/2387] eta: 0:00:24 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2238 (0.2683) acc1: 94.0104 (90.7656) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2116 data: 0.1203 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1452 (0.2678) acc1: 98.6979 (90.7982) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1966 data: 0.1016 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1726 (0.2675) acc1: 97.1354 (90.8205) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2028 data: 0.1069 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1834 (0.2671) acc1: 96.8750 (90.8476) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.2091 data: 0.1171 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1547 (0.2667) acc1: 97.6562 (90.8726) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1858 data: 0.0954 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.1739 (0.2665) acc1: 97.1354 (90.8949) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1594 data: 0.0686 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2062 (0.2664) acc1: 96.3542 (90.9053) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1544 data: 0.0637 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2198 (0.2663) acc1: 95.0521 (90.9136) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1514 data: 0.0613 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2572 (0.2664) acc1: 90.1042 (90.9021) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1468 data: 0.0568 max mem: 3541 +Test: [2350/2387] eta: 0:00:07 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.3246 (0.2668) acc1: 87.7604 (90.8866) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.1414 data: 0.0515 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2647 (0.2668) acc1: 90.6250 (90.8888) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1375 data: 0.0474 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2080 (0.2666) acc1: 94.7917 (90.9076) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1357 data: 0.0453 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2574 (0.2670) acc1: 91.4062 (90.8866) acc5: 100.0000 (100.0000) acc1_neg: 71.615 acc5_neg: 100.000 time: 0.1334 data: 0.0429 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 68.950 acc5_pos: 100.000 loss: 0.2747 (0.2669) acc1: 91.1458 (90.8882) acc5: 100.0000 (100.0000) acc1_neg: 97.674 acc5_neg: 100.000 time: 0.1261 data: 0.0388 max mem: 3541 +Test: Total time: 0:07:31 (0.1890 s / it) +* Overall Acc@1 90.888 Acc@5 100.000 + target == real Acc@1 52.329 Acc@5 100.000 + target == fake Acc@1 93.596 Acc@5 100.000 + loss 0.267 + PR-AUC 0.9904 ROC-AUC 0.8847 EER 20.25% (thr=0.7515) +Accuracy of the network on the 916267 test images: 90.9% +[rank0]:[W725 17:42:34.482122549 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 17:42:53.739000 2564520 torch/distributed/run.py:766] +W0725 17:42:53.739000 2564520 torch/distributed/run.py:766] ***************************************** +W0725 17:42:53.739000 2564520 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:42:53.739000 2564520 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +[rank0]:[W725 17:43:20.969670248 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:43:20.976127057 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:43:21.606662535 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:43:21.718749399 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:43:21.071175976 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:43:21.085684099 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:43:21.104035949 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank4]:[W725 17:43:21.108157318 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed3_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed3_resampled_n116-040-168_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:00:07 acc1_pos: 46.615 acc5_pos: 100.000 loss: 0.9287 (0.9287) acc1: 46.6146 (46.6146) acc5: 100.0000 (100.0000) time: 6.9779 data: 5.3406 max mem: 3531 +Test: [ 10/517] eta: 0:06:03 acc1_pos: 51.823 acc5_pos: 100.000 loss: 0.9668 (1.0143) acc1: 46.6146 (40.2225) acc5: 100.0000 (100.0000) time: 0.7160 data: 0.4858 max mem: 3531 +Test: [ 20/517] eta: 0:03:28 acc1_pos: 58.333 acc5_pos: 100.000 loss: 0.8298 (0.8960) acc1: 50.7812 (48.2639) acc5: 100.0000 (100.0000) time: 0.0913 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:03:06 acc1_pos: 14.062 acc5_pos: 100.000 loss: 0.8477 (0.9764) acc1: 42.9688 (42.8931) acc5: 100.0000 (100.0000) time: 0.2014 data: 0.1068 max mem: 3531 +Test: [ 40/517] eta: 0:02:47 acc1_pos: 40.885 acc5_pos: 100.000 loss: 1.1808 (1.0244) acc1: 23.4375 (38.6623) acc5: 100.0000 (100.0000) time: 0.2815 data: 0.1861 max mem: 3531 +Test: [ 50/517] eta: 0:02:37 acc1_pos: 50.459 acc5_pos: 100.000 loss: 1.0492 (0.9863) acc1: 32.0312 (40.5995) acc5: 100.0000 (100.0000) acc1_neg: 69.271 acc5_neg: 100.000 time: 0.2649 data: 0.1723 max mem: 3531 +Test: [ 60/517] eta: 0:02:28 acc1_pos: 50.459 acc5_pos: 100.000 loss: 0.4751 (0.8888) acc1: 69.2708 (47.3446) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2683 data: 0.1763 max mem: 3531 +Test: [ 70/517] eta: 0:02:20 acc1_pos: 50.459 acc5_pos: 100.000 loss: 0.2628 (0.7972) acc1: 94.2708 (54.3244) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2557 data: 0.1640 max mem: 3531 +Test: [ 80/517] eta: 0:02:14 acc1_pos: 50.459 acc5_pos: 100.000 loss: 0.2339 (0.7291) acc1: 98.4375 (59.6515) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2599 data: 0.1681 max mem: 3531 +Test: [ 90/517] eta: 0:02:09 acc1_pos: 50.459 acc5_pos: 100.000 loss: 0.2674 (0.6806) acc1: 96.0938 (63.6132) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2692 data: 0.1778 max mem: 3531 +Test: [100/517] eta: 0:02:05 acc1_pos: 50.459 acc5_pos: 100.000 loss: 0.2949 (0.6440) acc1: 93.7500 (66.4191) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.2664 data: 0.1759 max mem: 3531 +Test: [110/517] eta: 0:02:00 acc1_pos: 50.459 acc5_pos: 100.000 loss: 0.3208 (0.6160) acc1: 91.6667 (68.5224) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2561 data: 0.1649 max mem: 3531 +Test: [120/517] eta: 0:01:55 acc1_pos: 50.459 acc5_pos: 100.000 loss: 0.2794 (0.5865) acc1: 93.4896 (70.6655) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2484 data: 0.1524 max mem: 3531 +Test: [130/517] eta: 0:01:52 acc1_pos: 50.459 acc5_pos: 100.000 loss: 0.2502 (0.5605) acc1: 95.0521 (72.5907) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2652 data: 0.1696 max mem: 3531 +Test: [140/517] eta: 0:01:48 acc1_pos: 50.459 acc5_pos: 100.000 loss: 0.2916 (0.5455) acc1: 90.8854 (73.6462) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.2599 data: 0.1672 max mem: 3531 +Test: [150/517] eta: 0:01:44 acc1_pos: 50.459 acc5_pos: 100.000 loss: 0.3209 (0.5299) acc1: 88.5417 (74.8120) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.2439 data: 0.1513 max mem: 3531 +Test: [160/517] eta: 0:01:41 acc1_pos: 50.459 acc5_pos: 100.000 loss: 0.3530 (0.5214) acc1: 88.2812 (75.4060) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.2704 data: 0.1784 max mem: 3531 +Test: [170/517] eta: 0:01:37 acc1_pos: 50.459 acc5_pos: 100.000 loss: 0.3957 (0.5138) acc1: 83.3333 (75.9137) acc5: 100.0000 (100.0000) acc1_neg: 79.427 acc5_neg: 100.000 time: 0.2610 data: 0.1683 max mem: 3531 +Test: [180/517] eta: 0:01:34 acc1_pos: 50.459 acc5_pos: 100.000 loss: 0.3941 (0.5052) acc1: 85.1562 (76.5193) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2444 data: 0.1518 max mem: 3531 +Test: [190/517] eta: 0:01:31 acc1_pos: 50.459 acc5_pos: 100.000 loss: 0.2805 (0.4928) acc1: 93.4896 (77.5319) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2534 data: 0.1580 max mem: 3531 +Test: [200/517] eta: 0:01:27 acc1_pos: 50.459 acc5_pos: 100.000 loss: 0.2972 (0.4863) acc1: 93.4896 (77.9993) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.2399 data: 0.1462 max mem: 3531 +Test: [210/517] eta: 0:01:24 acc1_pos: 50.459 acc5_pos: 100.000 loss: 0.3315 (0.4807) acc1: 86.4583 (78.3533) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.2461 data: 0.1551 max mem: 3532 +Test: [220/517] eta: 0:01:21 acc1_pos: 50.459 acc5_pos: 100.000 loss: 0.3534 (0.4755) acc1: 87.5000 (78.7000) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.2556 data: 0.1632 max mem: 3532 +Test: [230/517] eta: 0:01:18 acc1_pos: 50.459 acc5_pos: 100.000 loss: 0.3869 (0.4731) acc1: 84.6354 (78.9344) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.2597 data: 0.1647 max mem: 3532 +Test: [240/517] eta: 0:01:15 acc1_pos: 50.459 acc5_pos: 100.000 loss: 0.4109 (0.4708) acc1: 83.5938 (79.0932) acc5: 100.0000 (100.0000) acc1_neg: 66.927 acc5_neg: 100.000 time: 0.2692 data: 0.1704 max mem: 3532 +Test: [250/517] eta: 0:01:12 acc1_pos: 50.459 acc5_pos: 100.000 loss: 0.3528 (0.4659) acc1: 87.5000 (79.5225) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2540 data: 0.1549 max mem: 3532 +Test: [260/517] eta: 0:01:10 acc1_pos: 50.459 acc5_pos: 100.000 loss: 0.3277 (0.4614) acc1: 89.0625 (79.8631) acc5: 100.0000 (100.0000) acc1_neg: 76.042 acc5_neg: 100.000 time: 0.2599 data: 0.1644 max mem: 3532 +Test: [270/517] eta: 0:01:07 acc1_pos: 50.459 acc5_pos: 100.000 loss: 0.4198 (0.4610) acc1: 81.2500 (79.8182) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.2749 data: 0.1810 max mem: 3532 +Test: [280/517] eta: 0:01:04 acc1_pos: 50.459 acc5_pos: 100.000 loss: 0.4477 (0.4601) acc1: 79.9479 (79.8877) acc5: 100.0000 (100.0000) acc1_neg: 72.917 acc5_neg: 100.000 time: 0.2499 data: 0.1558 max mem: 3532 +Test: [290/517] eta: 0:01:01 acc1_pos: 50.459 acc5_pos: 100.000 loss: 0.4823 (0.4614) acc1: 79.6875 (79.8325) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.2599 data: 0.1658 max mem: 3532 +Test: [300/517] eta: 0:00:58 acc1_pos: 50.459 acc5_pos: 100.000 loss: 0.4090 (0.4584) acc1: 83.0729 (80.1080) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.2658 data: 0.1736 max mem: 3532 +Test: [310/517] eta: 0:00:56 acc1_pos: 50.459 acc5_pos: 100.000 loss: 0.3846 (0.4559) acc1: 86.4583 (80.2402) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.2549 data: 0.1604 max mem: 3532 +Test: [320/517] eta: 0:00:53 acc1_pos: 50.459 acc5_pos: 100.000 loss: 0.3448 (0.4508) acc1: 86.4583 (80.5848) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2490 data: 0.1516 max mem: 3532 +Test: [330/517] eta: 0:00:50 acc1_pos: 50.459 acc5_pos: 100.000 loss: 0.2473 (0.4444) acc1: 94.5312 (81.0486) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2399 data: 0.1420 max mem: 3532 +Test: [340/517] eta: 0:00:47 acc1_pos: 50.459 acc5_pos: 100.000 loss: 0.2200 (0.4379) acc1: 96.8750 (81.5104) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2549 data: 0.1569 max mem: 3532 +Test: [350/517] eta: 0:00:44 acc1_pos: 50.459 acc5_pos: 100.000 loss: 0.2623 (0.4402) acc1: 94.0104 (81.2819) acc5: 100.0000 (100.0000) acc1_neg: 48.438 acc5_neg: 100.000 time: 0.2549 data: 0.1584 max mem: 3532 +Test: [360/517] eta: 0:00:42 acc1_pos: 50.459 acc5_pos: 100.000 loss: 0.4465 (0.4390) acc1: 72.1354 (81.3207) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2549 data: 0.1583 max mem: 3532 +Test: [370/517] eta: 0:00:39 acc1_pos: 50.459 acc5_pos: 100.000 loss: 0.1955 (0.4317) acc1: 98.9583 (81.8087) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2599 data: 0.1637 max mem: 3532 +Test: [380/517] eta: 0:00:36 acc1_pos: 50.459 acc5_pos: 100.000 loss: 0.1634 (0.4245) acc1: 99.4792 (82.2650) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2599 data: 0.1625 max mem: 3532 +Test: [390/517] eta: 0:00:33 acc1_pos: 50.459 acc5_pos: 100.000 loss: 0.1634 (0.4178) acc1: 99.4792 (82.6999) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2599 data: 0.1626 max mem: 3532 +Test: [400/517] eta: 0:00:31 acc1_pos: 50.459 acc5_pos: 100.000 loss: 0.1643 (0.4122) acc1: 99.4792 (83.0677) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2549 data: 0.1579 max mem: 3532 +Test: [410/517] eta: 0:00:28 acc1_pos: 50.459 acc5_pos: 100.000 loss: 0.2305 (0.4083) acc1: 95.5729 (83.3637) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2499 data: 0.1529 max mem: 3532 +Test: [420/517] eta: 0:00:25 acc1_pos: 50.459 acc5_pos: 100.000 loss: 0.2477 (0.4046) acc1: 95.8333 (83.6618) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.2549 data: 0.1592 max mem: 3532 +Test: [430/517] eta: 0:00:23 acc1_pos: 50.459 acc5_pos: 100.000 loss: 0.2726 (0.4026) acc1: 90.3646 (83.7799) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.2549 data: 0.1601 max mem: 3532 +Test: [440/517] eta: 0:00:20 acc1_pos: 50.459 acc5_pos: 100.000 loss: 0.2831 (0.3991) acc1: 91.4062 (84.0514) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2499 data: 0.1547 max mem: 3533 +Test: [450/517] eta: 0:00:17 acc1_pos: 50.459 acc5_pos: 100.000 loss: 0.2446 (0.3961) acc1: 95.8333 (84.2861) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2599 data: 0.1650 max mem: 3533 +Test: [460/517] eta: 0:00:15 acc1_pos: 50.459 acc5_pos: 100.000 loss: 0.2458 (0.3928) acc1: 96.0938 (84.5411) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2599 data: 0.1650 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 50.459 acc5_pos: 100.000 loss: 0.2550 (0.3897) acc1: 95.5729 (84.7759) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2498 data: 0.1543 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 50.459 acc5_pos: 100.000 loss: 0.2700 (0.3889) acc1: 94.7917 (84.8444) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2530 data: 0.1595 max mem: 3533 +Test: [490/517] eta: 0:00:07 acc1_pos: 50.459 acc5_pos: 100.000 loss: 0.2974 (0.3866) acc1: 94.5312 (85.0597) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2155 data: 0.1245 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 50.459 acc5_pos: 100.000 loss: 0.2579 (0.3841) acc1: 95.8333 (85.2587) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1840 data: 0.0947 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 50.459 acc5_pos: 100.000 loss: 0.2828 (0.3842) acc1: 94.0104 (85.2673) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1851 data: 0.0956 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 50.459 acc5_pos: 100.000 loss: 0.2999 (0.3832) acc1: 92.9688 (85.3498) acc5: 100.0000 (100.0000) acc1_neg: 84.099 acc5_neg: 100.000 time: 0.1791 data: 0.0879 max mem: 3533 +Test: Total time: 0:02:14 (0.2602 s / it) +* Overall Acc@1 85.350 Acc@5 100.000 + target == real Acc@1 38.794 Acc@5 100.000 + target == fake Acc@1 90.180 Acc@5 100.000 + loss 0.383 + PR-AUC 0.9576 ROC-AUC 0.7440 EER 31.93% (thr=0.7349) +Accuracy of the network on the 198427 test images: 85.3% +[rank0]:[W725 17:45:44.516619473 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0725 17:46:04.552000 2565113 torch/distributed/run.py:766] +W0725 17:46:04.552000 2565113 torch/distributed/run.py:766] ***************************************** +W0725 17:46:04.552000 2565113 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:46:04.552000 2565113 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:46:32.375607195 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:46:32.512379011 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:46:32.558873671 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:46:32.861605938 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:46:32.011351991 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:46:32.026135990 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 17:46:32.029683658 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:46:32.046544204 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed3_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed3_resampled_n116-040-168_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:32:11 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.1268 (2.1268) acc1: 0.0000 (0.0000) acc5: 100.0000 (100.0000) time: 4.9651 data: 2.8320 max mem: 3531 +Test: [ 10/389] eta: 0:03:22 acc1_pos: 0.781 acc5_pos: 100.000 loss: 1.8441 (1.8853) acc1: 3.3854 (4.6638) acc5: 100.0000 (100.0000) time: 0.5331 data: 0.2579 max mem: 3531 +Test: [ 20/389] eta: 0:02:05 acc1_pos: 1.302 acc5_pos: 100.000 loss: 1.9373 (1.9665) acc1: 1.0417 (2.8770) acc5: 100.0000 (100.0000) time: 0.1080 data: 0.0181 max mem: 3531 +Test: [ 30/389] eta: 0:01:44 acc1_pos: 13.542 acc5_pos: 100.000 loss: 2.2729 (2.0726) acc1: 0.5208 (2.7722) acc5: 100.0000 (100.0000) time: 0.1597 data: 0.0681 max mem: 3531 +Test: [ 40/389] eta: 0:01:42 acc1_pos: 1.302 acc5_pos: 100.000 loss: 2.2928 (2.0999) acc1: 0.0000 (2.4644) acc5: 100.0000 (100.0000) time: 0.2478 data: 0.1557 max mem: 3531 +Test: [ 50/389] eta: 0:01:34 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.1926 (2.1171) acc1: 0.0000 (1.9914) acc5: 100.0000 (100.0000) time: 0.2585 data: 0.1677 max mem: 3531 +Test: [ 60/389] eta: 0:01:25 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.1506 (2.1204) acc1: 0.0000 (2.0833) acc5: 100.0000 (100.0000) time: 0.1901 data: 0.0983 max mem: 3531 +Test: [ 70/389] eta: 0:01:24 acc1_pos: 1.562 acc5_pos: 100.000 loss: 2.1571 (2.1376) acc1: 0.2604 (1.9073) acc5: 100.0000 (100.0000) time: 0.2263 data: 0.1347 max mem: 3531 +Test: [ 80/389] eta: 0:01:18 acc1_pos: 3.646 acc5_pos: 100.000 loss: 2.1991 (2.1159) acc1: 1.0417 (2.1026) acc5: 100.0000 (100.0000) time: 0.2379 data: 0.1480 max mem: 3531 +Test: [ 90/389] eta: 0:01:14 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.9354 (2.0808) acc1: 3.3854 (2.7673) acc5: 100.0000 (100.0000) time: 0.1897 data: 0.0990 max mem: 3531 +Test: [100/389] eta: 0:01:10 acc1_pos: 0.000 acc5_pos: 100.000 loss: 2.0213 (2.0874) acc1: 0.7812 (2.6532) acc5: 100.0000 (100.0000) time: 0.2009 data: 0.1075 max mem: 3531 +Test: [110/389] eta: 0:01:07 acc1_pos: 1.823 acc5_pos: 100.000 loss: 2.0357 (2.0673) acc1: 0.2604 (3.2611) acc5: 100.0000 (100.0000) time: 0.2129 data: 0.1199 max mem: 3531 +Test: [120/389] eta: 0:01:03 acc1_pos: 0.781 acc5_pos: 100.000 loss: 2.0503 (2.0695) acc1: 0.7812 (3.0432) acc5: 100.0000 (100.0000) time: 0.1873 data: 0.0957 max mem: 3531 +Test: [130/389] eta: 0:00:59 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.9520 (1.9603) acc1: 0.7812 (8.5918) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1700 data: 0.0750 max mem: 3531 +Test: [140/389] eta: 0:00:56 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1291 (1.8326) acc1: 100.0000 (15.0174) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1865 data: 0.0918 max mem: 3531 +Test: [150/389] eta: 0:00:54 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1272 (1.7188) acc1: 100.0000 (20.6298) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1957 data: 0.1024 max mem: 3531 +Test: [160/389] eta: 0:00:51 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0975 (1.6179) acc1: 100.0000 (25.5532) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2083 data: 0.1143 max mem: 3531 +Test: [170/389] eta: 0:00:48 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0896 (1.5329) acc1: 100.0000 (29.6982) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1870 data: 0.0940 max mem: 3531 +Test: [180/389] eta: 0:00:46 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0896 (1.4537) acc1: 100.0000 (33.5822) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1911 data: 0.0992 max mem: 3531 +Test: [190/389] eta: 0:00:43 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1048 (1.3883) acc1: 100.0000 (36.7174) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1982 data: 0.1073 max mem: 3531 +Test: [200/389] eta: 0:00:40 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1063 (1.3247) acc1: 100.0000 (39.8606) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1738 data: 0.0810 max mem: 3531 +Test: [210/389] eta: 0:00:38 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1043 (1.2665) acc1: 100.0000 (42.7059) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1807 data: 0.0877 max mem: 3532 +Test: [220/389] eta: 0:00:36 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0869 (1.2132) acc1: 100.0000 (45.2936) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1823 data: 0.0918 max mem: 3532 +Test: [230/389] eta: 0:00:33 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0896 (1.1655) acc1: 100.0000 (47.6281) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1834 data: 0.0928 max mem: 3532 +Test: [240/389] eta: 0:00:31 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0958 (1.1223) acc1: 99.7396 (49.7526) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1825 data: 0.0905 max mem: 3532 +Test: [250/389] eta: 0:00:29 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1077 (1.0842) acc1: 99.7396 (51.6818) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1706 data: 0.0775 max mem: 3532 +Test: [260/389] eta: 0:00:26 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1115 (1.0485) acc1: 99.7396 (53.5271) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1713 data: 0.0790 max mem: 3532 +Test: [270/389] eta: 0:00:24 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1113 (1.0137) acc1: 100.0000 (55.2410) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1840 data: 0.0910 max mem: 3532 +Test: [280/389] eta: 0:00:22 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0982 (0.9811) acc1: 100.0000 (56.8218) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1804 data: 0.0878 max mem: 3532 +Test: [290/389] eta: 0:00:20 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0871 (0.9516) acc1: 100.0000 (58.2403) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1812 data: 0.0900 max mem: 3532 +Test: [300/389] eta: 0:00:18 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0946 (0.9235) acc1: 100.0000 (59.6198) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1806 data: 0.0875 max mem: 3532 +Test: [310/389] eta: 0:00:16 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0985 (0.8971) acc1: 100.0000 (60.9157) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1843 data: 0.0913 max mem: 3532 +Test: [320/389] eta: 0:00:13 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1049 (0.8763) acc1: 100.0000 (61.9329) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1784 data: 0.0878 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.1030 (0.8528) acc1: 100.0000 (63.0806) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1607 data: 0.0699 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0947 (0.8307) acc1: 100.0000 (64.1595) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1589 data: 0.0681 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0853 (0.8097) acc1: 100.0000 (65.1709) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1539 data: 0.0640 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0749 (0.7893) acc1: 100.0000 (66.1357) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1401 data: 0.0506 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0753 (0.7704) acc1: 100.0000 (67.0485) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1335 data: 0.0439 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0957 (0.7536) acc1: 100.0000 (67.9079) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1330 data: 0.0435 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 0.000 acc5_pos: 100.000 loss: 0.0932 (0.7400) acc1: 100.0000 (68.5641) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1312 data: 0.0383 max mem: 3532 +Test: Total time: 0:01:15 (0.1929 s / it) +* Overall Acc@1 68.564 Acc@5 100.000 + target == real Acc@1 2.986 Acc@5 100.000 + target == fake Acc@1 99.047 Acc@5 100.000 + loss 0.740 + PR-AUC 0.8387 ROC-AUC 0.7161 EER 33.12% (thr=0.9014) +Accuracy of the network on the 149358 test images: 68.6% +[rank0]:[W725 17:47:59.399816288 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 17:48:17.540000 2565670 torch/distributed/run.py:766] +W0725 17:48:17.540000 2565670 torch/distributed/run.py:766] ***************************************** +W0725 17:48:17.540000 2565670 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:48:17.540000 2565670 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:48:43.773126411 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:48:43.074867420 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:48:43.155943570 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 17:48:44.287984624 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:48:44.357047750 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:48:44.546941166 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:48:44.671704096 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:48:44.676689452 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed3_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed3_resampled_n116-040-168_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:27:00 acc1_pos: 31.510 acc5_pos: 100.000 loss: 1.0606 (1.0606) acc1: 31.5104 (31.5104) acc5: 100.0000 (100.0000) time: 4.8674 data: 3.0255 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/333] eta: 0:03:16 acc1_pos: 21.875 acc5_pos: 100.000 loss: 1.0606 (1.0700) acc1: 35.9375 (33.5938) acc5: 100.0000 (100.0000) time: 0.6091 data: 0.3603 max mem: 3531 +Test: [ 20/333] eta: 0:01:53 acc1_pos: 29.167 acc5_pos: 100.000 loss: 1.1401 (1.1518) acc1: 27.8646 (29.7247) acc5: 100.0000 (100.0000) time: 0.1378 data: 0.0471 max mem: 3531 +Test: [ 30/333] eta: 0:01:32 acc1_pos: 77.604 acc5_pos: 100.000 loss: 1.0385 (0.9538) acc1: 29.9479 (43.3804) acc5: 100.0000 (100.0000) time: 0.1362 data: 0.0394 max mem: 3531 +Test: [ 40/333] eta: 0:01:22 acc1_pos: 85.417 acc5_pos: 100.000 loss: 0.4627 (0.8421) acc1: 76.0417 (51.3529) acc5: 100.0000 (100.0000) time: 0.1975 data: 0.0994 max mem: 3531 +Test: [ 50/333] eta: 0:01:16 acc1_pos: 72.656 acc5_pos: 100.000 loss: 0.4627 (0.7645) acc1: 78.1250 (56.9547) acc5: 100.0000 (100.0000) time: 0.2203 data: 0.1278 max mem: 3531 +Test: [ 60/333] eta: 0:01:10 acc1_pos: 76.562 acc5_pos: 100.000 loss: 0.4304 (0.7121) acc1: 79.6875 (60.5789) acc5: 100.0000 (100.0000) time: 0.2125 data: 0.1193 max mem: 3531 +Test: [ 70/333] eta: 0:01:06 acc1_pos: 66.927 acc5_pos: 100.000 loss: 0.4304 (0.6742) acc1: 80.4688 (63.2519) acc5: 100.0000 (100.0000) time: 0.2008 data: 0.1071 max mem: 3531 +Test: [ 80/333] eta: 0:01:01 acc1_pos: 76.042 acc5_pos: 100.000 loss: 0.4445 (0.6467) acc1: 78.9062 (65.1620) acc5: 100.0000 (100.0000) time: 0.1971 data: 0.1056 max mem: 3531 +Test: [ 90/333] eta: 0:00:57 acc1_pos: 84.896 acc5_pos: 100.000 loss: 0.4483 (0.6242) acc1: 78.9062 (66.8241) acc5: 100.0000 (100.0000) time: 0.1915 data: 0.0980 max mem: 3531 +Test: [100/333] eta: 0:00:54 acc1_pos: 80.729 acc5_pos: 100.000 loss: 0.4502 (0.6082) acc1: 80.7292 (68.0513) acc5: 100.0000 (100.0000) time: 0.1929 data: 0.0994 max mem: 3531 +Test: [110/333] eta: 0:00:51 acc1_pos: 89.583 acc5_pos: 100.000 loss: 0.4694 (0.5951) acc1: 78.9062 (69.0339) acc5: 100.0000 (100.0000) time: 0.2030 data: 0.1091 max mem: 3531 +Test: [120/333] eta: 0:00:48 acc1_pos: 78.646 acc5_pos: 100.000 loss: 0.4694 (0.5841) acc1: 79.4271 (69.8821) acc5: 100.0000 (100.0000) time: 0.1995 data: 0.1062 max mem: 3531 +Test: [130/333] eta: 0:00:45 acc1_pos: 87.240 acc5_pos: 100.000 loss: 0.4567 (0.5733) acc1: 80.7292 (70.7578) acc5: 100.0000 (100.0000) time: 0.1889 data: 0.0954 max mem: 3531 +Test: [140/333] eta: 0:00:43 acc1_pos: 82.552 acc5_pos: 100.000 loss: 0.4384 (0.5652) acc1: 81.2500 (71.3966) acc5: 100.0000 (100.0000) time: 0.1976 data: 0.1001 max mem: 3531 +Test: [150/333] eta: 0:00:41 acc1_pos: 71.615 acc5_pos: 100.000 loss: 0.4277 (0.5579) acc1: 80.9896 (72.0233) acc5: 100.0000 (100.0000) time: 0.2224 data: 0.1267 max mem: 3531 +Test: [160/333] eta: 0:00:38 acc1_pos: 83.553 acc5_pos: 100.000 loss: 0.4797 (0.5701) acc1: 78.1250 (71.1843) acc5: 100.0000 (100.0000) acc1_neg: 26.302 acc5_neg: 100.000 time: 0.2189 data: 0.1242 max mem: 3531 +Test: [170/333] eta: 0:00:35 acc1_pos: 83.553 acc5_pos: 100.000 loss: 1.0166 (0.5973) acc1: 36.9792 (69.1734) acc5: 100.0000 (100.0000) acc1_neg: 36.979 acc5_neg: 100.000 time: 0.1884 data: 0.0945 max mem: 3531 +Test: [180/333] eta: 0:00:33 acc1_pos: 83.553 acc5_pos: 100.000 loss: 0.9879 (0.6168) acc1: 38.2812 (67.6206) acc5: 100.0000 (100.0000) acc1_neg: 34.115 acc5_neg: 100.000 time: 0.1909 data: 0.0992 max mem: 3531 +Test: [190/333] eta: 0:00:30 acc1_pos: 83.553 acc5_pos: 100.000 loss: 0.8944 (0.6291) acc1: 42.4479 (66.4649) acc5: 100.0000 (100.0000) acc1_neg: 28.646 acc5_neg: 100.000 time: 0.1810 data: 0.0880 max mem: 3531 +Test: [200/333] eta: 0:00:28 acc1_pos: 83.553 acc5_pos: 100.000 loss: 0.8687 (0.6405) acc1: 47.1354 (65.5822) acc5: 100.0000 (100.0000) acc1_neg: 46.354 acc5_neg: 100.000 time: 0.1823 data: 0.0883 max mem: 3531 +Test: [210/333] eta: 0:00:26 acc1_pos: 83.553 acc5_pos: 100.000 loss: 0.8790 (0.6527) acc1: 47.3958 (64.6722) acc5: 100.0000 (100.0000) acc1_neg: 51.302 acc5_neg: 100.000 time: 0.1985 data: 0.1046 max mem: 3532 +Test: [220/333] eta: 0:00:23 acc1_pos: 83.553 acc5_pos: 100.000 loss: 0.8811 (0.6662) acc1: 45.5729 (63.5511) acc5: 100.0000 (100.0000) acc1_neg: 52.083 acc5_neg: 100.000 time: 0.1716 data: 0.0796 max mem: 3532 +Test: [230/333] eta: 0:00:21 acc1_pos: 83.553 acc5_pos: 100.000 loss: 0.8497 (0.6736) acc1: 45.5729 (62.9160) acc5: 100.0000 (100.0000) acc1_neg: 58.594 acc5_neg: 100.000 time: 0.1522 data: 0.0598 max mem: 3532 +Test: [240/333] eta: 0:00:19 acc1_pos: 83.553 acc5_pos: 100.000 loss: 0.9424 (0.6949) acc1: 40.1042 (61.5415) acc5: 100.0000 (100.0000) acc1_neg: 27.604 acc5_neg: 100.000 time: 0.1596 data: 0.0642 max mem: 3532 +Test: [250/333] eta: 0:00:17 acc1_pos: 83.553 acc5_pos: 100.000 loss: 1.1697 (0.7064) acc1: 29.1667 (60.6875) acc5: 100.0000 (100.0000) acc1_neg: 53.646 acc5_neg: 100.000 time: 0.1681 data: 0.0726 max mem: 3532 +Test: [260/333] eta: 0:00:14 acc1_pos: 83.553 acc5_pos: 100.000 loss: 0.8213 (0.7106) acc1: 50.7812 (60.3538) acc5: 100.0000 (100.0000) acc1_neg: 58.073 acc5_neg: 100.000 time: 0.1639 data: 0.0701 max mem: 3532 +Test: [270/333] eta: 0:00:12 acc1_pos: 83.553 acc5_pos: 100.000 loss: 0.8181 (0.7141) acc1: 52.6042 (60.1034) acc5: 100.0000 (100.0000) acc1_neg: 50.000 acc5_neg: 100.000 time: 0.1595 data: 0.0680 max mem: 3532 +Test: [280/333] eta: 0:00:10 acc1_pos: 83.553 acc5_pos: 100.000 loss: 0.8305 (0.7266) acc1: 50.2604 (59.2675) acc5: 100.0000 (100.0000) acc1_neg: 29.167 acc5_neg: 100.000 time: 0.1644 data: 0.0728 max mem: 3532 +Test: [290/333] eta: 0:00:08 acc1_pos: 83.553 acc5_pos: 100.000 loss: 1.1779 (0.7423) acc1: 28.6458 (58.2519) acc5: 100.0000 (100.0000) acc1_neg: 25.521 acc5_neg: 100.000 time: 0.1655 data: 0.0739 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 83.553 acc5_pos: 100.000 loss: 1.1084 (0.7533) acc1: 31.7708 (57.4734) acc5: 100.0000 (100.0000) acc1_neg: 35.938 acc5_neg: 100.000 time: 0.1725 data: 0.0814 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 83.553 acc5_pos: 100.000 loss: 1.0377 (0.7581) acc1: 35.9375 (57.1301) acc5: 100.0000 (100.0000) acc1_neg: 30.469 acc5_neg: 100.000 time: 0.1836 data: 0.0926 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 83.553 acc5_pos: 100.000 loss: 1.0612 (0.7679) acc1: 33.5938 (56.4285) acc5: 100.0000 (100.0000) acc1_neg: 51.042 acc5_neg: 100.000 time: 0.1676 data: 0.0774 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 83.553 acc5_pos: 100.000 loss: 0.9158 (0.7719) acc1: 38.8021 (56.0761) acc5: 100.0000 (100.0000) acc1_neg: 38.802 acc5_neg: 100.000 time: 0.1399 data: 0.0498 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 83.553 acc5_pos: 100.000 loss: 0.9119 (0.7725) acc1: 44.2708 (56.0226) acc5: 100.0000 (100.0000) acc1_neg: 48.169 acc5_neg: 100.000 time: 0.1186 data: 0.0252 max mem: 3532 +Test: Total time: 0:01:05 (0.1959 s / it) +* Overall Acc@1 56.023 Acc@5 100.000 + target == real Acc@1 72.263 Acc@5 100.000 + target == fake Acc@1 41.566 Acc@5 100.000 + loss 0.773 + PR-AUC 0.6155 ROC-AUC 0.6146 EER 41.60% (thr=0.3582) +Accuracy of the network on the 127843 test images: 56.0% +[rank0]:[W725 17:50:02.738012613 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 17:50:21.274000 2566241 torch/distributed/run.py:766] +W0725 17:50:21.274000 2566241 torch/distributed/run.py:766] ***************************************** +W0725 17:50:21.274000 2566241 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:50:21.274000 2566241 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:50:52.786735116 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +[rank0]:[W725 17:50:53.928782659 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:50:53.934448800 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:50:53.971128584 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +NCCL version 2.26.2+cuda12.2 +[rank3]:[W725 17:50:53.059818093 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:50:53.091483924 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:50:53.170700801 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:50:54.190903471 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed3_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed3_resampled_n116-040-168_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:19:58 acc1_pos: 93.490 acc5_pos: 100.000 loss: 0.2117 (0.2117) acc1: 93.4896 (93.4896) acc5: 100.0000 (100.0000) time: 5.0990 data: 2.7245 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/235] eta: 0:02:03 acc1_pos: 92.448 acc5_pos: 100.000 loss: 0.2334 (0.2383) acc1: 92.4479 (92.9451) acc5: 100.0000 (100.0000) time: 0.5473 data: 0.2481 max mem: 3531 +Test: [ 20/235] eta: 0:01:15 acc1_pos: 91.667 acc5_pos: 100.000 loss: 0.2334 (0.2359) acc1: 92.4479 (92.9316) acc5: 100.0000 (100.0000) time: 0.1135 data: 0.0187 max mem: 3531 +Test: [ 30/235] eta: 0:01:01 acc1_pos: 94.010 acc5_pos: 100.000 loss: 0.2289 (0.2339) acc1: 92.9688 (92.9267) acc5: 100.0000 (100.0000) time: 0.1624 data: 0.0648 max mem: 3531 +Test: [ 40/235] eta: 0:00:55 acc1_pos: 92.448 acc5_pos: 100.000 loss: 0.2322 (0.2332) acc1: 92.9688 (92.9370) acc5: 100.0000 (100.0000) time: 0.2162 data: 0.1196 max mem: 3531 +Test: [ 50/235] eta: 0:00:50 acc1_pos: 90.885 acc5_pos: 100.000 loss: 0.2343 (0.2339) acc1: 93.2292 (92.9024) acc5: 100.0000 (100.0000) time: 0.2384 data: 0.1454 max mem: 3531 +Test: [ 60/235] eta: 0:00:45 acc1_pos: 96.094 acc5_pos: 100.000 loss: 0.2339 (0.2343) acc1: 93.2292 (92.9431) acc5: 100.0000 (100.0000) time: 0.2037 data: 0.1096 max mem: 3531 +Test: [ 70/235] eta: 0:00:42 acc1_pos: 92.308 acc5_pos: 100.000 loss: 0.2521 (0.4100) acc1: 90.1042 (83.5717) acc5: 100.0000 (100.0000) acc1_neg: 16.146 acc5_neg: 100.000 time: 0.2044 data: 0.1101 max mem: 3531 +Test: [ 80/235] eta: 0:00:38 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.6489 (0.5644) acc1: 16.9271 (75.2990) acc5: 100.0000 (100.0000) acc1_neg: 14.844 acc5_neg: 100.000 time: 0.2299 data: 0.1351 max mem: 3531 +Test: [ 90/235] eta: 0:00:35 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.6662 (0.6841) acc1: 16.6667 (68.9475) acc5: 100.0000 (100.0000) acc1_neg: 14.062 acc5_neg: 100.000 time: 0.2042 data: 0.1083 max mem: 3531 +Test: [100/235] eta: 0:00:32 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.6570 (0.7795) acc1: 17.4479 (63.8640) acc5: 100.0000 (100.0000) acc1_neg: 13.542 acc5_neg: 100.000 time: 0.2149 data: 0.1228 max mem: 3531 +Test: [110/235] eta: 0:00:29 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.6503 (0.8571) acc1: 17.4479 (59.6847) acc5: 100.0000 (100.0000) acc1_neg: 16.146 acc5_neg: 100.000 time: 0.1996 data: 0.1076 max mem: 3531 +Test: [120/235] eta: 0:00:26 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.6713 (0.9256) acc1: 16.1458 (55.9100) acc5: 100.0000 (100.0000) acc1_neg: 16.146 acc5_neg: 100.000 time: 0.1867 data: 0.0951 max mem: 3531 +Test: [130/235] eta: 0:00:24 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.6791 (0.9824) acc1: 14.5833 (52.7731) acc5: 100.0000 (100.0000) acc1_neg: 14.844 acc5_neg: 100.000 time: 0.2237 data: 0.1332 max mem: 3531 +Test: [140/235] eta: 0:00:21 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.6930 (1.0355) acc1: 14.0625 (49.9686) acc5: 100.0000 (100.0000) acc1_neg: 9.375 acc5_neg: 100.000 time: 0.1989 data: 0.1068 max mem: 3531 +Test: [150/235] eta: 0:00:19 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.6955 (1.0780) acc1: 13.5417 (47.6045) acc5: 100.0000 (100.0000) acc1_neg: 14.062 acc5_neg: 100.000 time: 0.2000 data: 0.1082 max mem: 3531 +Test: [160/235] eta: 0:00:16 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.6696 (1.1157) acc1: 14.0625 (45.5487) acc5: 100.0000 (100.0000) acc1_neg: 14.062 acc5_neg: 100.000 time: 0.2051 data: 0.1140 max mem: 3531 +Test: [170/235] eta: 0:00:14 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.6296 (1.1136) acc1: 15.3646 (45.3932) acc5: 100.0000 (100.0000) acc1_neg: 45.052 acc5_neg: 100.000 time: 0.2040 data: 0.1130 max mem: 3531 +Test: [180/235] eta: 0:00:12 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.0107 (1.1079) acc1: 45.3125 (45.4406) acc5: 100.0000 (100.0000) acc1_neg: 47.396 acc5_neg: 100.000 time: 0.2217 data: 0.1280 max mem: 3531 +Test: [190/235] eta: 0:00:10 acc1_pos: 92.308 acc5_pos: 100.000 loss: 0.9965 (1.1022) acc1: 46.3542 (45.4938) acc5: 100.0000 (100.0000) acc1_neg: 43.229 acc5_neg: 100.000 time: 0.2156 data: 0.1212 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 92.308 acc5_pos: 100.000 loss: 0.9965 (1.0974) acc1: 45.5729 (45.4939) acc5: 100.0000 (100.0000) acc1_neg: 41.406 acc5_neg: 100.000 time: 0.1952 data: 0.1008 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 92.308 acc5_pos: 100.000 loss: 0.9975 (1.0930) acc1: 45.3125 (45.5248) acc5: 100.0000 (100.0000) acc1_neg: 44.792 acc5_neg: 100.000 time: 0.1426 data: 0.0490 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 92.308 acc5_pos: 100.000 loss: 0.9989 (1.0895) acc1: 46.0938 (45.5446) acc5: 100.0000 (100.0000) acc1_neg: 42.969 acc5_neg: 100.000 time: 0.1144 data: 0.0247 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.0143 (1.0866) acc1: 45.5729 (45.5515) acc5: 100.0000 (100.0000) acc1_neg: 44.792 acc5_neg: 100.000 time: 0.1332 data: 0.0437 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 92.308 acc5_pos: 100.000 loss: 1.0241 (1.0864) acc1: 45.5729 (45.5546) acc5: 100.0000 (100.0000) acc1_neg: 34.783 acc5_neg: 100.000 time: 0.1298 data: 0.0437 max mem: 3532 +Test: Total time: 0:00:49 (0.2087 s / it) +* Overall Acc@1 45.555 Acc@5 100.000 + target == real Acc@1 92.921 Acc@5 100.000 + target == fake Acc@1 28.423 Acc@5 100.000 + loss 1.086 + PR-AUC 0.8680 ROC-AUC 0.7111 EER 35.11% (thr=0.1569) +Accuracy of the network on the 89879 test images: 45.6% +[rank0]:[W725 17:51:51.857418333 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 17:52:10.505000 2566809 torch/distributed/run.py:766] +W0725 17:52:10.505000 2566809 torch/distributed/run.py:766] ***************************************** +W0725 17:52:10.505000 2566809 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:52:10.505000 2566809 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:52:36.766963699 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:52:36.110483019 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W725 17:52:36.127123387 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:52:37.183742105 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 17:52:37.323621485 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:52:37.542798834 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:52:37.567577132 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank5]:[W725 17:52:37.570324321 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed3_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed3_resampled_n116-040-168_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:29:39 acc1_pos: 56.771 acc5_pos: 100.000 loss: 0.7008 (0.7008) acc1: 56.7708 (56.7708) acc5: 100.0000 (100.0000) time: 6.6145 data: 4.8605 max mem: 3531 +Test: [ 10/269] eta: 0:02:59 acc1_pos: 76.302 acc5_pos: 100.000 loss: 0.7008 (0.7090) acc1: 56.7708 (58.2623) acc5: 100.0000 (100.0000) time: 0.6912 data: 0.4422 max mem: 3531 +Test: [ 20/269] eta: 0:01:44 acc1_pos: 64.583 acc5_pos: 100.000 loss: 0.6882 (0.7060) acc1: 59.3750 (58.0233) acc5: 100.0000 (100.0000) time: 0.1094 data: 0.0126 max mem: 3531 +Test: [ 30/269] eta: 0:01:21 acc1_pos: 69.271 acc5_pos: 100.000 loss: 0.6882 (0.6882) acc1: 59.6354 (60.4587) acc5: 100.0000 (100.0000) time: 0.1472 data: 0.0517 max mem: 3531 +Test: [ 40/269] eta: 0:01:09 acc1_pos: 18.966 acc5_pos: 100.000 loss: 0.5977 (0.6659) acc1: 69.0104 (62.5064) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.1799 data: 0.0849 max mem: 3531 +Test: [ 50/269] eta: 0:01:02 acc1_pos: 18.966 acc5_pos: 100.000 loss: 0.6073 (0.6634) acc1: 64.3229 (62.7247) acc5: 100.0000 (100.0000) acc1_neg: 51.823 acc5_neg: 100.000 time: 0.1998 data: 0.1061 max mem: 3531 +Test: [ 60/269] eta: 0:00:56 acc1_pos: 18.966 acc5_pos: 100.000 loss: 0.6711 (0.6707) acc1: 60.4167 (61.6376) acc5: 100.0000 (100.0000) acc1_neg: 51.823 acc5_neg: 100.000 time: 0.1999 data: 0.1035 max mem: 3531 +Test: [ 70/269] eta: 0:00:52 acc1_pos: 18.966 acc5_pos: 100.000 loss: 0.7191 (0.6785) acc1: 54.4271 (60.6221) acc5: 100.0000 (100.0000) acc1_neg: 53.906 acc5_neg: 100.000 time: 0.2036 data: 0.1065 max mem: 3531 +Test: [ 80/269] eta: 0:00:48 acc1_pos: 18.966 acc5_pos: 100.000 loss: 0.7025 (0.6703) acc1: 58.0729 (61.4165) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.2099 data: 0.1141 max mem: 3531 +Test: [ 90/269] eta: 0:00:45 acc1_pos: 18.966 acc5_pos: 100.000 loss: 0.5061 (0.6523) acc1: 78.1250 (63.2011) acc5: 100.0000 (100.0000) acc1_neg: 68.750 acc5_neg: 100.000 time: 0.2241 data: 0.1265 max mem: 3531 +Test: [100/269] eta: 0:00:41 acc1_pos: 18.966 acc5_pos: 100.000 loss: 0.5766 (0.6584) acc1: 68.7500 (62.3969) acc5: 100.0000 (100.0000) acc1_neg: 42.969 acc5_neg: 100.000 time: 0.2175 data: 0.1191 max mem: 3531 +Test: [110/269] eta: 0:00:38 acc1_pos: 18.966 acc5_pos: 100.000 loss: 0.6581 (0.6595) acc1: 58.5938 (62.0660) acc5: 100.0000 (100.0000) acc1_neg: 58.594 acc5_neg: 100.000 time: 0.1929 data: 0.0968 max mem: 3531 +Test: [120/269] eta: 0:00:35 acc1_pos: 18.966 acc5_pos: 100.000 loss: 0.6942 (0.6671) acc1: 52.3438 (61.0236) acc5: 100.0000 (100.0000) acc1_neg: 29.688 acc5_neg: 100.000 time: 0.1941 data: 0.1011 max mem: 3531 +Test: [130/269] eta: 0:00:32 acc1_pos: 18.966 acc5_pos: 100.000 loss: 0.7920 (0.6851) acc1: 45.5729 (59.2180) acc5: 100.0000 (100.0000) acc1_neg: 21.615 acc5_neg: 100.000 time: 0.2018 data: 0.1083 max mem: 3531 +Test: [140/269] eta: 0:00:30 acc1_pos: 18.966 acc5_pos: 100.000 loss: 0.8029 (0.6870) acc1: 45.5729 (58.9151) acc5: 100.0000 (100.0000) acc1_neg: 73.698 acc5_neg: 100.000 time: 0.2093 data: 0.1127 max mem: 3531 +Test: [150/269] eta: 0:00:27 acc1_pos: 18.966 acc5_pos: 100.000 loss: 0.6708 (0.6863) acc1: 55.9896 (58.9180) acc5: 100.0000 (100.0000) acc1_neg: 45.573 acc5_neg: 100.000 time: 0.1918 data: 0.0947 max mem: 3531 +Test: [160/269] eta: 0:00:24 acc1_pos: 18.966 acc5_pos: 100.000 loss: 0.6667 (0.6858) acc1: 55.9896 (58.7377) acc5: 100.0000 (100.0000) acc1_neg: 58.854 acc5_neg: 100.000 time: 0.1938 data: 0.0985 max mem: 3531 +Test: [170/269] eta: 0:00:22 acc1_pos: 18.966 acc5_pos: 100.000 loss: 0.6667 (0.6847) acc1: 55.7292 (58.7735) acc5: 100.0000 (100.0000) acc1_neg: 75.260 acc5_neg: 100.000 time: 0.1950 data: 0.0986 max mem: 3531 +Test: [180/269] eta: 0:00:19 acc1_pos: 18.966 acc5_pos: 100.000 loss: 0.5723 (0.6762) acc1: 66.1458 (59.5736) acc5: 100.0000 (100.0000) acc1_neg: 65.104 acc5_neg: 100.000 time: 0.1887 data: 0.0920 max mem: 3531 +Test: [190/269] eta: 0:00:17 acc1_pos: 18.966 acc5_pos: 100.000 loss: 0.5951 (0.6788) acc1: 65.1042 (59.3491) acc5: 100.0000 (100.0000) acc1_neg: 49.479 acc5_neg: 100.000 time: 0.1939 data: 0.0970 max mem: 3531 +Test: [200/269] eta: 0:00:15 acc1_pos: 18.966 acc5_pos: 100.000 loss: 0.7557 (0.6829) acc1: 51.0417 (59.2416) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1722 data: 0.0747 max mem: 3531 +Test: [210/269] eta: 0:00:12 acc1_pos: 18.966 acc5_pos: 100.000 loss: 0.7284 (0.6834) acc1: 53.3854 (59.3059) acc5: 100.0000 (100.0000) acc1_neg: 63.802 acc5_neg: 100.000 time: 0.1687 data: 0.0707 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 18.966 acc5_pos: 100.000 loss: 0.7045 (0.6852) acc1: 59.1146 (59.1818) acc5: 100.0000 (100.0000) acc1_neg: 59.375 acc5_neg: 100.000 time: 0.2105 data: 0.1114 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 18.966 acc5_pos: 100.000 loss: 0.7304 (0.6861) acc1: 53.3854 (58.9613) acc5: 100.0000 (100.0000) acc1_neg: 71.875 acc5_neg: 100.000 time: 0.1854 data: 0.0887 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 18.966 acc5_pos: 100.000 loss: 0.6316 (0.6836) acc1: 61.9792 (59.2140) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.1171 data: 0.0213 max mem: 3532 +Test: [250/269] eta: 0:00:03 acc1_pos: 18.966 acc5_pos: 100.000 loss: 0.6414 (0.6824) acc1: 60.4167 (59.2225) acc5: 100.0000 (100.0000) acc1_neg: 52.344 acc5_neg: 100.000 time: 0.1045 data: 0.0113 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 18.966 acc5_pos: 100.000 loss: 0.7047 (0.6857) acc1: 51.5625 (58.7843) acc5: 100.0000 (100.0000) acc1_neg: 65.104 acc5_neg: 100.000 time: 0.1215 data: 0.0319 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 18.966 acc5_pos: 100.000 loss: 0.6939 (0.6827) acc1: 52.3438 (59.0369) acc5: 100.0000 (100.0000) acc1_neg: 94.226 acc5_neg: 100.000 time: 0.1367 data: 0.0400 max mem: 3532 +Test: Total time: 0:00:54 (0.2021 s / it) +* Overall Acc@1 59.037 Acc@5 100.000 + target == real Acc@1 59.553 Acc@5 100.000 + target == fake Acc@1 58.966 Acc@5 100.000 + loss 0.683 + PR-AUC 0.9188 ROC-AUC 0.6278 EER 40.74% (thr=0.4978) +Accuracy of the network on the 103293 test images: 59.0% +[rank0]:[W725 17:53:40.428779450 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-25T17:53:46Z] eval_done tag=fixed500k_real_n5_seed3_resampled +[2026-07-25T17:53:46Z] eval_start tag=fixed500k_fake_n11_seed3_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed3_resampled_n116-040-168_20260725_112559/best_checkpoint.pth +started_utc=2026-07-25T17:53:46Z +hostname=n116-040-168 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed3_resampled_n116-040-168_20260725_112559/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed3_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0725 17:53:58.673000 2567391 torch/distributed/run.py:766] +W0725 17:53:58.673000 2567391 torch/distributed/run.py:766] ***************************************** +W0725 17:53:58.673000 2567391 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:53:58.673000 2567391 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 17:54:25.659953339 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 17:54:25.836315636 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:54:25.873384433 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:54:25.172874393 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 17:54:26.231566262 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:54:26.271752308 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:54:26.369246390 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:54:26.404522841 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed3_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed3_resampled_n116-040-168_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:18:46 acc1_pos: 80.208 acc5_pos: 100.000 loss: 0.4558 (0.4558) acc1: 80.2083 (80.2083) acc5: 100.0000 (100.0000) time: 4.8135 data: 2.6517 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/234] eta: 0:01:56 acc1_pos: 34.115 acc5_pos: 100.000 loss: 0.8221 (0.8237) acc1: 53.6458 (48.6506) acc5: 100.0000 (100.0000) time: 0.5192 data: 0.2414 max mem: 3531 +Test: [ 20/234] eta: 0:01:08 acc1_pos: 89.062 acc5_pos: 100.000 loss: 0.6546 (0.7034) acc1: 55.9896 (59.3006) acc5: 100.0000 (100.0000) time: 0.0931 data: 0.0036 max mem: 3531 +Test: [ 30/234] eta: 0:00:49 acc1_pos: 54.167 acc5_pos: 100.000 loss: 0.3169 (0.5982) acc1: 89.0625 (67.1623) acc5: 100.0000 (100.0000) time: 0.0933 data: 0.0036 max mem: 3531 +Test: [ 40/234] eta: 0:00:46 acc1_pos: 94.792 acc5_pos: 100.000 loss: 0.3058 (0.5964) acc1: 89.0625 (66.8699) acc5: 100.0000 (100.0000) time: 0.1532 data: 0.0618 max mem: 3531 +Test: [ 50/234] eta: 0:00:42 acc1_pos: 96.875 acc5_pos: 100.000 loss: 0.1951 (0.5163) acc1: 95.5729 (72.6614) acc5: 100.0000 (100.0000) time: 0.2156 data: 0.1180 max mem: 3531 +Test: [ 60/234] eta: 0:00:40 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.1508 (0.4505) acc1: 97.6562 (77.1047) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2167 data: 0.1182 max mem: 3531 +Test: [ 70/234] eta: 0:00:36 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.1124 (0.4390) acc1: 100.0000 (78.0223) acc5: 100.0000 (100.0000) acc1_neg: 70.052 acc5_neg: 100.000 time: 0.1893 data: 0.0940 max mem: 3531 +Test: [ 80/234] eta: 0:00:33 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.4482 (0.4305) acc1: 78.1250 (78.7777) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1649 data: 0.0702 max mem: 3531 +Test: [ 90/234] eta: 0:00:30 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.2560 (0.4056) acc1: 90.1042 (80.4315) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.1710 data: 0.0782 max mem: 3531 +Test: [100/234] eta: 0:00:27 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.2729 (0.4210) acc1: 86.4583 (79.5973) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1799 data: 0.0886 max mem: 3531 +Test: [110/234] eta: 0:00:25 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.4960 (0.4252) acc1: 72.9167 (79.4857) acc5: 100.0000 (100.0000) acc1_neg: 72.917 acc5_neg: 100.000 time: 0.1788 data: 0.0851 max mem: 3531 +Test: [120/234] eta: 0:00:22 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.4502 (0.4258) acc1: 77.8646 (79.6445) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.1699 data: 0.0752 max mem: 3531 +Test: [130/234] eta: 0:00:20 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.3966 (0.4272) acc1: 80.4688 (79.7213) acc5: 100.0000 (100.0000) acc1_neg: 63.281 acc5_neg: 100.000 time: 0.1856 data: 0.0940 max mem: 3531 +Test: [140/234] eta: 0:00:18 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.5218 (0.4602) acc1: 71.6146 (77.7741) acc5: 100.0000 (100.0000) acc1_neg: 28.385 acc5_neg: 100.000 time: 0.1849 data: 0.0924 max mem: 3531 +Test: [150/234] eta: 0:00:16 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.6239 (0.4744) acc1: 64.0625 (77.0368) acc5: 100.0000 (100.0000) acc1_neg: 72.917 acc5_neg: 100.000 time: 0.1726 data: 0.0805 max mem: 3531 +Test: [160/234] eta: 0:00:14 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.5209 (0.4736) acc1: 75.0000 (77.2030) acc5: 100.0000 (100.0000) acc1_neg: 77.083 acc5_neg: 100.000 time: 0.1657 data: 0.0750 max mem: 3531 +Test: [170/234] eta: 0:00:12 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.4857 (0.4726) acc1: 77.8646 (77.3879) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1532 data: 0.0610 max mem: 3531 +Test: [180/234] eta: 0:00:10 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.4718 (0.4736) acc1: 77.8646 (77.3222) acc5: 100.0000 (100.0000) acc1_neg: 51.042 acc5_neg: 100.000 time: 0.1564 data: 0.0645 max mem: 3531 +Test: [190/234] eta: 0:00:08 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.8096 (0.4923) acc1: 56.5104 (76.1903) acc5: 100.0000 (100.0000) acc1_neg: 54.427 acc5_neg: 100.000 time: 0.1607 data: 0.0700 max mem: 3531 +Test: [200/234] eta: 0:00:06 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.8123 (0.4872) acc1: 56.5104 (76.5664) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1549 data: 0.0636 max mem: 3531 +Test: [210/234] eta: 0:00:04 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.1365 (0.4699) acc1: 99.7396 (77.6733) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1573 data: 0.0630 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.0975 (0.4536) acc1: 100.0000 (78.6388) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1667 data: 0.0698 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.2440 (0.4576) acc1: 74.2188 (78.3516) acc5: 100.0000 (100.0000) acc1_neg: 74.219 acc5_neg: 100.000 time: 0.1550 data: 0.0617 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 94.483 acc5_pos: 100.000 loss: 0.5206 (0.4587) acc1: 72.3958 (78.2701) acc5: 100.0000 (100.0000) acc1_neg: 75.486 acc5_neg: 100.000 time: 0.1195 data: 0.0264 max mem: 3532 +Test: Total time: 0:00:42 (0.1828 s / it) +* Overall Acc@1 78.270 Acc@5 100.000 + target == real Acc@1 72.822 Acc@5 100.000 + target == fake Acc@1 79.806 Acc@5 100.000 + loss 0.459 + PR-AUC 0.9545 ROC-AUC 0.8564 EER 23.31% (thr=0.5547) +Accuracy of the network on the 89729 test images: 78.3% +[rank0]:[W725 17:55:25.402114467 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 17:55:44.445000 2567962 torch/distributed/run.py:766] +W0725 17:55:44.445000 2567962 torch/distributed/run.py:766] ***************************************** +W0725 17:55:44.445000 2567962 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 17:55:44.445000 2567962 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 17:56:13.497811507 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 17:56:13.559500993 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +[rank0]:[W725 17:56:13.045491497 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 17:56:13.047030343 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 2): env:// +NCCL version 2.26.2+cuda12.2 +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 4): env:// +[rank2]:[W725 17:56:14.262308321 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 17:56:14.271241495 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W725 17:56:14.271300221 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 17:56:14.381062119 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed3_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed3_resampled_n116-040-168_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:19:13 acc1_pos: 51.562 acc5_pos: 100.000 loss: 0.7815 (0.7815) acc1: 51.5625 (51.5625) acc5: 100.0000 (100.0000) time: 6.5161 data: 4.9273 max mem: 3531 +Test: [ 10/2387] eta: 0:26:42 acc1_pos: 35.677 acc5_pos: 100.000 loss: 1.0925 (1.0379) acc1: 35.6771 (37.0265) acc5: 100.0000 (100.0000) time: 0.6742 data: 0.4483 max mem: 3531 +Test: [ 20/2387] eta: 0:15:37 acc1_pos: 25.260 acc5_pos: 100.000 loss: 1.0925 (1.0975) acc1: 29.9479 (32.6265) acc5: 100.0000 (100.0000) time: 0.0902 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:13:49 acc1_pos: 16.406 acc5_pos: 100.000 loss: 1.1372 (1.1112) acc1: 25.5208 (31.7204) acc5: 100.0000 (100.0000) time: 0.1749 data: 0.0834 max mem: 3531 +Test: [ 40/2387] eta: 0:13:11 acc1_pos: 19.010 acc5_pos: 100.000 loss: 1.2198 (1.1408) acc1: 23.1771 (29.7891) acc5: 100.0000 (100.0000) time: 0.2754 data: 0.1804 max mem: 3531 +Test: [ 50/2387] eta: 0:12:50 acc1_pos: 26.562 acc5_pos: 100.000 loss: 1.2401 (1.1548) acc1: 24.7396 (28.6765) acc5: 100.0000 (100.0000) time: 0.2956 data: 0.2001 max mem: 3531 +Test: [ 60/2387] eta: 0:11:57 acc1_pos: 35.677 acc5_pos: 100.000 loss: 1.2459 (1.1677) acc1: 24.7396 (28.1890) acc5: 100.0000 (100.0000) time: 0.2495 data: 0.1573 max mem: 3531 +Test: [ 70/2387] eta: 0:11:35 acc1_pos: 25.000 acc5_pos: 100.000 loss: 1.2651 (1.1868) acc1: 20.8333 (27.4795) acc5: 100.0000 (100.0000) time: 0.2242 data: 0.1305 max mem: 3531 +Test: [ 80/2387] eta: 0:11:29 acc1_pos: 78.646 acc5_pos: 100.000 loss: 1.0520 (1.1441) acc1: 31.2500 (30.6359) acc5: 100.0000 (100.0000) time: 0.2695 data: 0.1745 max mem: 3531 +Test: [ 90/2387] eta: 0:11:06 acc1_pos: 61.979 acc5_pos: 100.000 loss: 0.6344 (1.0834) acc1: 61.9792 (34.5095) acc5: 100.0000 (100.0000) time: 0.2549 data: 0.1607 max mem: 3531 +Test: [ 100/2387] eta: 0:10:54 acc1_pos: 61.979 acc5_pos: 100.000 loss: 0.5767 (1.0369) acc1: 63.0208 (37.5748) acc5: 100.0000 (100.0000) time: 0.2354 data: 0.1379 max mem: 3531 +Test: [ 110/2387] eta: 0:10:45 acc1_pos: 47.135 acc5_pos: 100.000 loss: 0.6615 (1.0032) acc1: 59.8958 (39.6162) acc5: 100.0000 (100.0000) time: 0.2530 data: 0.1562 max mem: 3531 +Test: [ 120/2387] eta: 0:10:34 acc1_pos: 39.062 acc5_pos: 100.000 loss: 0.7226 (0.9841) acc1: 59.1146 (40.8660) acc5: 100.0000 (100.0000) time: 0.2469 data: 0.1543 max mem: 3531 +Test: [ 130/2387] eta: 0:10:29 acc1_pos: 48.438 acc5_pos: 100.000 loss: 0.7212 (0.9605) acc1: 58.0729 (42.3803) acc5: 100.0000 (100.0000) time: 0.2523 data: 0.1603 max mem: 3531 +Test: [ 140/2387] eta: 0:10:30 acc1_pos: 47.656 acc5_pos: 100.000 loss: 0.7082 (0.9441) acc1: 56.2500 (43.4471) acc5: 100.0000 (100.0000) time: 0.2841 data: 0.1876 max mem: 3531 +Test: [ 150/2387] eta: 0:10:18 acc1_pos: 70.573 acc5_pos: 100.000 loss: 0.6495 (0.9229) acc1: 62.2396 (44.8037) acc5: 100.0000 (100.0000) time: 0.2605 data: 0.1632 max mem: 3531 +Test: [ 160/2387] eta: 0:10:02 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.5296 (0.8957) acc1: 67.1875 (46.8524) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.2014 data: 0.1065 max mem: 3531 +Test: [ 170/2387] eta: 0:09:51 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3839 (0.8657) acc1: 82.5521 (48.9264) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.1962 data: 0.1028 max mem: 3531 +Test: [ 180/2387] eta: 0:09:40 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3809 (0.8386) acc1: 83.3333 (50.8877) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.2018 data: 0.1097 max mem: 3531 +Test: [ 190/2387] eta: 0:09:32 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3687 (0.8142) acc1: 84.1146 (52.6342) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.2042 data: 0.1085 max mem: 3531 +Test: [ 200/2387] eta: 0:09:24 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3693 (0.7923) acc1: 83.8542 (54.1913) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.2127 data: 0.1158 max mem: 3531 +Test: [ 210/2387] eta: 0:09:16 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3776 (0.7726) acc1: 83.8542 (55.6156) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.2103 data: 0.1160 max mem: 3532 +Test: [ 220/2387] eta: 0:09:08 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3776 (0.7549) acc1: 84.3750 (56.8840) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.2047 data: 0.1107 max mem: 3532 +Test: [ 230/2387] eta: 0:08:54 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3762 (0.7383) acc1: 83.5938 (58.0583) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.1631 data: 0.0691 max mem: 3532 +Test: [ 240/2387] eta: 0:08:48 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3903 (0.7320) acc1: 82.0312 (58.5992) acc5: 100.0000 (100.0000) acc1_neg: 68.750 acc5_neg: 100.000 time: 0.1677 data: 0.0719 max mem: 3532 +Test: [ 250/2387] eta: 0:08:46 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3531 (0.7086) acc1: 72.3958 (60.0992) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2310 data: 0.1350 max mem: 3532 +Test: [ 260/2387] eta: 0:08:40 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0854 (0.6847) acc1: 99.7396 (61.6050) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2264 data: 0.1315 max mem: 3532 +Test: [ 270/2387] eta: 0:08:26 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0780 (0.6626) acc1: 99.7396 (63.0170) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1486 data: 0.0528 max mem: 3532 +Test: [ 280/2387] eta: 0:08:17 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0838 (0.6422) acc1: 99.7396 (64.3192) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1281 data: 0.0325 max mem: 3532 +Test: [ 290/2387] eta: 0:08:15 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0896 (0.6231) acc1: 99.7396 (65.5131) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2013 data: 0.1058 max mem: 3532 +Test: [ 300/2387] eta: 0:08:12 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0794 (0.6054) acc1: 100.0000 (66.6416) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2298 data: 0.1341 max mem: 3532 +Test: [ 310/2387] eta: 0:08:08 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0844 (0.5925) acc1: 99.7396 (67.5945) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2161 data: 0.1199 max mem: 3532 +Test: [ 320/2387] eta: 0:08:04 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3190 (0.5842) acc1: 94.2708 (68.4076) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2114 data: 0.1180 max mem: 3532 +Test: [ 330/2387] eta: 0:07:59 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3238 (0.5764) acc1: 93.7500 (69.1906) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2022 data: 0.1093 max mem: 3532 +Test: [ 340/2387] eta: 0:07:54 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3223 (0.5690) acc1: 94.2708 (69.9253) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1885 data: 0.0920 max mem: 3532 +Test: [ 350/2387] eta: 0:07:52 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3009 (0.5589) acc1: 95.0521 (70.7035) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2109 data: 0.1137 max mem: 3532 +Test: [ 360/2387] eta: 0:07:47 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2256 (0.5505) acc1: 94.5312 (71.2849) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.2111 data: 0.1146 max mem: 3532 +Test: [ 370/2387] eta: 0:07:39 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2596 (0.5426) acc1: 91.6667 (71.8329) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1550 data: 0.0600 max mem: 3532 +Test: [ 380/2387] eta: 0:07:36 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2532 (0.5351) acc1: 91.9271 (72.3466) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1685 data: 0.0749 max mem: 3532 +Test: [ 390/2387] eta: 0:07:32 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2532 (0.5279) acc1: 91.6667 (72.8447) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2037 data: 0.1081 max mem: 3532 +Test: [ 400/2387] eta: 0:07:30 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2516 (0.5211) acc1: 91.4062 (73.3063) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.2199 data: 0.1251 max mem: 3532 +Test: [ 410/2387] eta: 0:07:28 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2496 (0.5145) acc1: 91.9271 (73.7651) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.2308 data: 0.1376 max mem: 3532 +Test: [ 420/2387] eta: 0:07:24 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2493 (0.5082) acc1: 91.9271 (74.1884) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.2103 data: 0.1165 max mem: 3532 +Test: [ 430/2387] eta: 0:07:18 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2595 (0.5045) acc1: 90.8854 (74.4586) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.1653 data: 0.0712 max mem: 3532 +Test: [ 440/2387] eta: 0:07:14 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2719 (0.4974) acc1: 85.4167 (74.8854) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1661 data: 0.0725 max mem: 3533 +Test: [ 450/2387] eta: 0:07:10 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0613 (0.4878) acc1: 100.0000 (75.4423) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1909 data: 0.0996 max mem: 3533 +Test: [ 460/2387] eta: 0:07:07 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0605 (0.4785) acc1: 100.0000 (75.9750) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1870 data: 0.0954 max mem: 3533 +Test: [ 470/2387] eta: 0:07:03 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0608 (0.4696) acc1: 100.0000 (76.4851) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1830 data: 0.0919 max mem: 3533 +Test: [ 480/2387] eta: 0:06:59 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0613 (0.4612) acc1: 100.0000 (76.9740) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1801 data: 0.0891 max mem: 3533 +Test: [ 490/2387] eta: 0:06:55 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0611 (0.4530) acc1: 100.0000 (77.4429) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1806 data: 0.0890 max mem: 3533 +Test: [ 500/2387] eta: 0:06:51 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0613 (0.4452) acc1: 100.0000 (77.8932) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1761 data: 0.0838 max mem: 3533 +Test: [ 510/2387] eta: 0:06:48 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0611 (0.4380) acc1: 100.0000 (78.3253) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1788 data: 0.0870 max mem: 3533 +Test: [ 520/2387] eta: 0:06:44 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0724 (0.4316) acc1: 100.0000 (78.7408) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1814 data: 0.0903 max mem: 3533 +Test: [ 530/2387] eta: 0:06:41 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0730 (0.4249) acc1: 100.0000 (79.1412) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1823 data: 0.0902 max mem: 3533 +Test: [ 540/2387] eta: 0:06:38 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0730 (0.4184) acc1: 100.0000 (79.5262) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1811 data: 0.0881 max mem: 3533 +Test: [ 550/2387] eta: 0:06:34 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0725 (0.4121) acc1: 100.0000 (79.8978) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1828 data: 0.0907 max mem: 3533 +Test: [ 560/2387] eta: 0:06:32 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0723 (0.4060) acc1: 100.0000 (80.2561) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1887 data: 0.0979 max mem: 3533 +Test: [ 570/2387] eta: 0:06:29 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0723 (0.4002) acc1: 100.0000 (80.6019) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1946 data: 0.0965 max mem: 3533 +Test: [ 580/2387] eta: 0:06:26 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0716 (0.3946) acc1: 100.0000 (80.9353) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1955 data: 0.0965 max mem: 3533 +Test: [ 590/2387] eta: 0:06:24 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0722 (0.3891) acc1: 100.0000 (81.2579) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2014 data: 0.1097 max mem: 3533 +Test: [ 600/2387] eta: 0:06:20 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0759 (0.3848) acc1: 100.0000 (81.5620) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1828 data: 0.0914 max mem: 3533 +Test: [ 610/2387] eta: 0:06:17 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1987 (0.3847) acc1: 85.1562 (81.5888) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1758 data: 0.0834 max mem: 3533 +Test: [ 620/2387] eta: 0:06:15 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3771 (0.3846) acc1: 83.3333 (81.6060) acc5: 100.0000 (100.0000) acc1_neg: 81.250 acc5_neg: 100.000 time: 0.2004 data: 0.1084 max mem: 3533 +Test: [ 630/2387] eta: 0:06:12 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3724 (0.3845) acc1: 83.3333 (81.6305) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.1976 data: 0.1041 max mem: 3533 +Test: [ 640/2387] eta: 0:06:09 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3755 (0.3845) acc1: 83.0729 (81.6571) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1860 data: 0.0926 max mem: 3533 +Test: [ 650/2387] eta: 0:06:07 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3827 (0.3845) acc1: 83.0729 (81.6792) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.1851 data: 0.0937 max mem: 3533 +Test: [ 660/2387] eta: 0:06:04 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3859 (0.3845) acc1: 82.8125 (81.6916) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.1834 data: 0.0926 max mem: 3533 +Test: [ 670/2387] eta: 0:06:00 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3808 (0.3844) acc1: 82.8125 (81.7107) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.1747 data: 0.0828 max mem: 3534 +Test: [ 680/2387] eta: 0:05:58 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3837 (0.3869) acc1: 81.7708 (81.4947) acc5: 100.0000 (100.0000) acc1_neg: 59.896 acc5_neg: 100.000 time: 0.1748 data: 0.0810 max mem: 3534 +Test: [ 690/2387] eta: 0:05:55 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.5795 (0.3900) acc1: 65.1042 (81.2568) acc5: 100.0000 (100.0000) acc1_neg: 65.104 acc5_neg: 100.000 time: 0.1851 data: 0.0912 max mem: 3534 +Test: [ 700/2387] eta: 0:05:52 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.5686 (0.3925) acc1: 68.4896 (81.0754) acc5: 100.0000 (100.0000) acc1_neg: 68.490 acc5_neg: 100.000 time: 0.1820 data: 0.0884 max mem: 3534 +Test: [ 710/2387] eta: 0:05:50 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.5686 (0.3951) acc1: 67.4479 (80.8771) acc5: 100.0000 (100.0000) acc1_neg: 68.229 acc5_neg: 100.000 time: 0.1825 data: 0.0897 max mem: 3534 +Test: [ 720/2387] eta: 0:05:47 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.5726 (0.3973) acc1: 67.1875 (80.7057) acc5: 100.0000 (100.0000) acc1_neg: 69.792 acc5_neg: 100.000 time: 0.1878 data: 0.0963 max mem: 3534 +Test: [ 730/2387] eta: 0:05:44 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.5734 (0.3998) acc1: 67.7083 (80.5225) acc5: 100.0000 (100.0000) acc1_neg: 67.708 acc5_neg: 100.000 time: 0.1882 data: 0.0968 max mem: 3534 +Test: [ 740/2387] eta: 0:05:42 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.5742 (0.4021) acc1: 67.1875 (80.3507) acc5: 100.0000 (100.0000) acc1_neg: 63.021 acc5_neg: 100.000 time: 0.1812 data: 0.0900 max mem: 3534 +Test: [ 750/2387] eta: 0:05:39 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.5561 (0.4042) acc1: 67.1875 (80.1813) acc5: 100.0000 (100.0000) acc1_neg: 67.708 acc5_neg: 100.000 time: 0.1759 data: 0.0851 max mem: 3534 +Test: [ 760/2387] eta: 0:05:36 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.5573 (0.4053) acc1: 68.2292 (80.0971) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.1786 data: 0.0870 max mem: 3534 +Test: [ 770/2387] eta: 0:05:34 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3499 (0.4036) acc1: 82.8125 (80.2202) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1804 data: 0.0892 max mem: 3534 +Test: [ 780/2387] eta: 0:05:31 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2013 (0.4010) acc1: 94.0104 (80.4147) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1758 data: 0.0853 max mem: 3534 +Test: [ 790/2387] eta: 0:05:28 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1652 (0.3984) acc1: 97.3958 (80.5902) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1725 data: 0.0806 max mem: 3534 +Test: [ 800/2387] eta: 0:05:25 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1587 (0.3952) acc1: 98.1771 (80.8121) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1710 data: 0.0781 max mem: 3534 +Test: [ 810/2387] eta: 0:05:23 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1587 (0.3928) acc1: 97.6562 (80.9992) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1749 data: 0.0831 max mem: 3534 +Test: [ 820/2387] eta: 0:05:20 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1714 (0.3902) acc1: 97.3958 (81.1847) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.1894 data: 0.0978 max mem: 3534 +Test: [ 830/2387] eta: 0:05:18 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2905 (0.3907) acc1: 84.6354 (81.1626) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.1812 data: 0.0897 max mem: 3534 +Test: [ 840/2387] eta: 0:05:15 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.4838 (0.3935) acc1: 75.5208 (80.9791) acc5: 100.0000 (100.0000) acc1_neg: 58.594 acc5_neg: 100.000 time: 0.1742 data: 0.0839 max mem: 3534 +Test: [ 850/2387] eta: 0:05:13 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.6387 (0.3968) acc1: 61.9792 (80.7613) acc5: 100.0000 (100.0000) acc1_neg: 58.594 acc5_neg: 100.000 time: 0.1854 data: 0.0946 max mem: 3534 +Test: [ 860/2387] eta: 0:05:11 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.6387 (0.3994) acc1: 61.9792 (80.6033) acc5: 100.0000 (100.0000) acc1_neg: 72.135 acc5_neg: 100.000 time: 0.1856 data: 0.0937 max mem: 3534 +Test: [ 870/2387] eta: 0:05:08 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.6037 (0.4020) acc1: 64.8438 (80.4254) acc5: 100.0000 (100.0000) acc1_neg: 64.844 acc5_neg: 100.000 time: 0.1831 data: 0.0918 max mem: 3534 +Test: [ 880/2387] eta: 0:05:06 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.6098 (0.4052) acc1: 62.5000 (80.2216) acc5: 100.0000 (100.0000) acc1_neg: 55.729 acc5_neg: 100.000 time: 0.1894 data: 0.0983 max mem: 3534 +Test: [ 890/2387] eta: 0:05:04 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.7253 (0.4091) acc1: 57.8125 (79.9748) acc5: 100.0000 (100.0000) acc1_neg: 71.094 acc5_neg: 100.000 time: 0.1999 data: 0.1080 max mem: 3534 +Test: [ 900/2387] eta: 0:05:02 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.5699 (0.4098) acc1: 68.2292 (79.9392) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.2071 data: 0.1141 max mem: 3535 +Test: [ 910/2387] eta: 0:05:00 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.4115 (0.4098) acc1: 82.5521 (79.9739) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.1997 data: 0.1072 max mem: 3535 +Test: [ 920/2387] eta: 0:04:58 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.4101 (0.4098) acc1: 83.3333 (80.0087) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.1938 data: 0.1019 max mem: 3535 +Test: [ 930/2387] eta: 0:04:56 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.4107 (0.4098) acc1: 83.3333 (80.0357) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.2097 data: 0.1180 max mem: 3535 +Test: [ 940/2387] eta: 0:04:53 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.4107 (0.4099) acc1: 83.3333 (80.0680) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.1856 data: 0.0940 max mem: 3535 +Test: [ 950/2387] eta: 0:04:50 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.4058 (0.4098) acc1: 83.8542 (80.1095) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.1557 data: 0.0645 max mem: 3535 +Test: [ 960/2387] eta: 0:04:48 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.4002 (0.4097) acc1: 83.8542 (80.1487) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.1642 data: 0.0737 max mem: 3535 +Test: [ 970/2387] eta: 0:04:45 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.4052 (0.4098) acc1: 83.5938 (80.1759) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.1580 data: 0.0667 max mem: 3535 +Test: [ 980/2387] eta: 0:04:43 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.4084 (0.4103) acc1: 82.2917 (80.1632) acc5: 100.0000 (100.0000) acc1_neg: 50.781 acc5_neg: 100.000 time: 0.1585 data: 0.0672 max mem: 3535 +Test: [ 990/2387] eta: 0:04:40 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3731 (0.4091) acc1: 83.3333 (80.2430) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1659 data: 0.0701 max mem: 3535 +Test: [1000/2387] eta: 0:04:38 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1474 (0.4067) acc1: 97.9167 (80.4089) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1654 data: 0.0693 max mem: 3535 +Test: [1010/2387] eta: 0:04:35 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1715 (0.4053) acc1: 97.3958 (80.5100) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1654 data: 0.0739 max mem: 3535 +Test: [1020/2387] eta: 0:04:33 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1869 (0.4037) acc1: 92.7083 (80.6088) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1715 data: 0.0802 max mem: 3535 +Test: [1030/2387] eta: 0:04:30 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1458 (0.4012) acc1: 98.6979 (80.7850) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1744 data: 0.0806 max mem: 3535 +Test: [1040/2387] eta: 0:04:28 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1234 (0.3985) acc1: 99.2188 (80.9578) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1769 data: 0.0815 max mem: 3535 +Test: [1050/2387] eta: 0:04:26 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1030 (0.3956) acc1: 99.7396 (81.1365) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1676 data: 0.0746 max mem: 3535 +Test: [1060/2387] eta: 0:04:23 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1030 (0.3932) acc1: 99.7396 (81.2957) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1557 data: 0.0626 max mem: 3535 +Test: [1070/2387] eta: 0:04:21 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1237 (0.3910) acc1: 99.2188 (81.4375) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.1572 data: 0.0636 max mem: 3535 +Test: [1080/2387] eta: 0:04:18 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2846 (0.3908) acc1: 90.3646 (81.4610) acc5: 100.0000 (100.0000) acc1_neg: 74.219 acc5_neg: 100.000 time: 0.1603 data: 0.0666 max mem: 3535 +Test: [1090/2387] eta: 0:04:16 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3658 (0.3908) acc1: 84.6354 (81.4784) acc5: 100.0000 (100.0000) acc1_neg: 73.958 acc5_neg: 100.000 time: 0.1609 data: 0.0683 max mem: 3535 +Test: [1100/2387] eta: 0:04:13 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3732 (0.3907) acc1: 84.6354 (81.4950) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1641 data: 0.0687 max mem: 3535 +Test: [1110/2387] eta: 0:04:11 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3366 (0.3907) acc1: 85.6771 (81.4957) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.1656 data: 0.0686 max mem: 3535 +Test: [1120/2387] eta: 0:04:09 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3735 (0.3908) acc1: 82.2917 (81.5007) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1594 data: 0.0667 max mem: 3536 +Test: [1130/2387] eta: 0:04:06 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3150 (0.3901) acc1: 90.1042 (81.5882) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1533 data: 0.0620 max mem: 3536 +Test: [1140/2387] eta: 0:04:04 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3189 (0.3917) acc1: 89.3229 (81.4990) acc5: 100.0000 (100.0000) acc1_neg: 62.760 acc5_neg: 100.000 time: 0.1599 data: 0.0669 max mem: 3536 +Test: [1150/2387] eta: 0:04:02 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.5946 (0.3928) acc1: 69.7917 (81.4238) acc5: 100.0000 (100.0000) acc1_neg: 69.792 acc5_neg: 100.000 time: 0.1652 data: 0.0719 max mem: 3536 +Test: [1160/2387] eta: 0:03:59 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.5111 (0.3934) acc1: 71.6146 (81.3841) acc5: 100.0000 (100.0000) acc1_neg: 63.281 acc5_neg: 100.000 time: 0.1449 data: 0.0494 max mem: 3536 +Test: [1170/2387] eta: 0:03:57 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.4391 (0.3938) acc1: 77.3438 (81.3685) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.1494 data: 0.0513 max mem: 3536 +Test: [1180/2387] eta: 0:03:54 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.4422 (0.3943) acc1: 78.6458 (81.3397) acc5: 100.0000 (100.0000) acc1_neg: 69.271 acc5_neg: 100.000 time: 0.1654 data: 0.0706 max mem: 3536 +Test: [1190/2387] eta: 0:03:52 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.4228 (0.3946) acc1: 80.2083 (81.3191) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1562 data: 0.0646 max mem: 3536 +Test: [1200/2387] eta: 0:03:50 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3966 (0.3947) acc1: 82.2917 (81.3161) acc5: 100.0000 (100.0000) acc1_neg: 59.375 acc5_neg: 100.000 time: 0.1644 data: 0.0733 max mem: 3536 +Test: [1210/2387] eta: 0:03:48 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3487 (0.3947) acc1: 83.0729 (81.3115) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1687 data: 0.0763 max mem: 3536 +Test: [1220/2387] eta: 0:03:45 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2385 (0.3931) acc1: 91.4062 (81.4202) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.1467 data: 0.0533 max mem: 3536 +Test: [1230/2387] eta: 0:03:43 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1771 (0.3914) acc1: 96.6146 (81.5377) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1720 data: 0.0775 max mem: 3536 +Test: [1240/2387] eta: 0:03:41 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1771 (0.3897) acc1: 96.3542 (81.6552) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2006 data: 0.1071 max mem: 3536 +Test: [1250/2387] eta: 0:03:39 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1712 (0.3881) acc1: 96.0938 (81.7633) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1687 data: 0.0775 max mem: 3536 +Test: [1260/2387] eta: 0:03:38 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1610 (0.3863) acc1: 96.0938 (81.8832) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1974 data: 0.1059 max mem: 3536 +Test: [1270/2387] eta: 0:03:35 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1580 (0.3847) acc1: 96.8750 (81.9888) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1909 data: 0.0972 max mem: 3536 +Test: [1280/2387] eta: 0:03:33 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1583 (0.3828) acc1: 96.8750 (82.1154) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1627 data: 0.0695 max mem: 3536 +Test: [1290/2387] eta: 0:03:31 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1541 (0.3811) acc1: 97.1354 (82.2245) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1987 data: 0.1072 max mem: 3536 +Test: [1300/2387] eta: 0:03:29 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1605 (0.3794) acc1: 96.8750 (82.3353) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1695 data: 0.0769 max mem: 3536 +Test: [1310/2387] eta: 0:03:27 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1589 (0.3778) acc1: 96.8750 (82.4426) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1642 data: 0.0712 max mem: 3536 +Test: [1320/2387] eta: 0:03:25 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1690 (0.3763) acc1: 96.8750 (82.5497) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1826 data: 0.0898 max mem: 3536 +Test: [1330/2387] eta: 0:03:23 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1969 (0.3759) acc1: 92.9688 (82.5732) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.1717 data: 0.0791 max mem: 3536 +Test: [1340/2387] eta: 0:03:21 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3121 (0.3754) acc1: 84.6354 (82.6090) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1781 data: 0.0852 max mem: 3536 +Test: [1350/2387] eta: 0:03:19 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0583 (0.3730) acc1: 100.0000 (82.7377) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1798 data: 0.0870 max mem: 3537 +Test: [1360/2387] eta: 0:03:17 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0465 (0.3706) acc1: 100.0000 (82.8645) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1791 data: 0.0877 max mem: 3537 +Test: [1370/2387] eta: 0:03:15 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0507 (0.3683) acc1: 100.0000 (82.9892) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1737 data: 0.0829 max mem: 3537 +Test: [1380/2387] eta: 0:03:13 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0520 (0.3660) acc1: 100.0000 (83.1123) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1653 data: 0.0743 max mem: 3537 +Test: [1390/2387] eta: 0:03:10 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0486 (0.3638) acc1: 100.0000 (83.2321) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1678 data: 0.0760 max mem: 3537 +Test: [1400/2387] eta: 0:03:08 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0457 (0.3615) acc1: 100.0000 (83.3517) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1688 data: 0.0767 max mem: 3537 +Test: [1410/2387] eta: 0:03:06 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0457 (0.3606) acc1: 100.0000 (83.4145) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.1637 data: 0.0715 max mem: 3537 +Test: [1420/2387] eta: 0:03:04 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2950 (0.3605) acc1: 89.8438 (83.4244) acc5: 100.0000 (100.0000) acc1_neg: 90.365 acc5_neg: 100.000 time: 0.1698 data: 0.0760 max mem: 3537 +Test: [1430/2387] eta: 0:03:02 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3237 (0.3604) acc1: 86.9792 (83.4483) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1784 data: 0.0851 max mem: 3537 +Test: [1440/2387] eta: 0:03:00 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2489 (0.3596) acc1: 92.9688 (83.5122) acc5: 100.0000 (100.0000) acc1_neg: 70.312 acc5_neg: 100.000 time: 0.1759 data: 0.0842 max mem: 3537 +Test: [1450/2387] eta: 0:02:58 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3114 (0.3601) acc1: 83.3333 (83.4805) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1726 data: 0.0812 max mem: 3537 +Test: [1460/2387] eta: 0:02:56 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2194 (0.3590) acc1: 92.1875 (83.5617) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1683 data: 0.0761 max mem: 3537 +Test: [1470/2387] eta: 0:02:54 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1718 (0.3578) acc1: 96.8750 (83.6401) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1671 data: 0.0740 max mem: 3537 +Test: [1480/2387] eta: 0:02:52 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1694 (0.3565) acc1: 97.3958 (83.7284) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1727 data: 0.0793 max mem: 3537 +Test: [1490/2387] eta: 0:02:50 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1736 (0.3554) acc1: 96.8750 (83.8110) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1859 data: 0.0933 max mem: 3537 +Test: [1500/2387] eta: 0:02:48 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1748 (0.3543) acc1: 96.8750 (83.8911) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1819 data: 0.0912 max mem: 3537 +Test: [1510/2387] eta: 0:02:46 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2590 (0.3543) acc1: 90.3646 (83.8935) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1555 data: 0.0647 max mem: 3537 +Test: [1520/2387] eta: 0:02:44 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.4201 (0.3552) acc1: 78.1250 (83.8463) acc5: 100.0000 (100.0000) acc1_neg: 65.885 acc5_neg: 100.000 time: 0.1613 data: 0.0698 max mem: 3537 +Test: [1530/2387] eta: 0:02:42 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.5155 (0.3566) acc1: 73.4375 (83.7496) acc5: 100.0000 (100.0000) acc1_neg: 62.240 acc5_neg: 100.000 time: 0.1786 data: 0.0859 max mem: 3537 +Test: [1540/2387] eta: 0:02:40 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.5528 (0.3578) acc1: 68.7500 (83.6701) acc5: 100.0000 (100.0000) acc1_neg: 68.750 acc5_neg: 100.000 time: 0.1877 data: 0.0946 max mem: 3537 +Test: [1550/2387] eta: 0:02:38 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.5269 (0.3590) acc1: 70.5729 (83.5887) acc5: 100.0000 (100.0000) acc1_neg: 69.531 acc5_neg: 100.000 time: 0.1776 data: 0.0841 max mem: 3537 +Test: [1560/2387] eta: 0:02:36 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.5189 (0.3603) acc1: 72.1354 (83.5080) acc5: 100.0000 (100.0000) acc1_neg: 63.542 acc5_neg: 100.000 time: 0.1849 data: 0.0907 max mem: 3537 +Test: [1570/2387] eta: 0:02:35 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.5853 (0.3619) acc1: 70.0521 (83.3965) acc5: 100.0000 (100.0000) acc1_neg: 74.740 acc5_neg: 100.000 time: 0.2063 data: 0.1120 max mem: 3537 +Test: [1580/2387] eta: 0:02:33 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.4229 (0.3613) acc1: 74.7396 (83.4430) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2069 data: 0.1134 max mem: 3538 +Test: [1590/2387] eta: 0:02:31 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1570 (0.3600) acc1: 97.9167 (83.5342) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1856 data: 0.0943 max mem: 3538 +Test: [1600/2387] eta: 0:02:29 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1430 (0.3588) acc1: 98.4375 (83.5985) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1787 data: 0.0879 max mem: 3538 +Test: [1610/2387] eta: 0:02:27 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1249 (0.3575) acc1: 98.9583 (83.6864) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1858 data: 0.0953 max mem: 3538 +Test: [1620/2387] eta: 0:02:25 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1443 (0.3564) acc1: 98.1771 (83.7628) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1891 data: 0.0974 max mem: 3538 +Test: [1630/2387] eta: 0:02:23 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1719 (0.3560) acc1: 96.6146 (83.7962) acc5: 100.0000 (100.0000) acc1_neg: 78.646 acc5_neg: 100.000 time: 0.2032 data: 0.1104 max mem: 3538 +Test: [1640/2387] eta: 0:02:21 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.4040 (0.3565) acc1: 79.6875 (83.7678) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1877 data: 0.0952 max mem: 3538 +Test: [1650/2387] eta: 0:02:19 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.4581 (0.3572) acc1: 77.6042 (83.7344) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1749 data: 0.0829 max mem: 3538 +Test: [1660/2387] eta: 0:02:17 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3975 (0.3574) acc1: 78.6458 (83.7164) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1726 data: 0.0814 max mem: 3538 +Test: [1670/2387] eta: 0:02:15 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3155 (0.3572) acc1: 84.8958 (83.7452) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1707 data: 0.0800 max mem: 3538 +Test: [1680/2387] eta: 0:02:13 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3018 (0.3570) acc1: 88.0208 (83.7632) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1726 data: 0.0816 max mem: 3538 +Test: [1690/2387] eta: 0:02:11 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3240 (0.3574) acc1: 84.8958 (83.7488) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.1698 data: 0.0786 max mem: 3538 +Test: [1700/2387] eta: 0:02:09 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3650 (0.3574) acc1: 81.7708 (83.7484) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1674 data: 0.0762 max mem: 3538 +Test: [1710/2387] eta: 0:02:07 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2864 (0.3568) acc1: 87.5000 (83.7901) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1656 data: 0.0738 max mem: 3538 +Test: [1720/2387] eta: 0:02:06 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2191 (0.3560) acc1: 92.9688 (83.8436) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1773 data: 0.0852 max mem: 3538 +Test: [1730/2387] eta: 0:02:04 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2335 (0.3554) acc1: 92.9688 (83.8961) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1955 data: 0.1039 max mem: 3538 +Test: [1740/2387] eta: 0:02:02 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2163 (0.3547) acc1: 92.4479 (83.9427) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.1902 data: 0.0988 max mem: 3538 +Test: [1750/2387] eta: 0:02:00 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1984 (0.3540) acc1: 93.7500 (84.0021) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2056 data: 0.1142 max mem: 3538 +Test: [1760/2387] eta: 0:01:58 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2390 (0.3535) acc1: 92.1875 (84.0322) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.2015 data: 0.1110 max mem: 3538 +Test: [1770/2387] eta: 0:01:56 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2598 (0.3531) acc1: 89.5833 (84.0599) acc5: 100.0000 (100.0000) acc1_neg: 89.583 acc5_neg: 100.000 time: 0.1821 data: 0.0909 max mem: 3538 +Test: [1780/2387] eta: 0:01:54 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2453 (0.3526) acc1: 88.5417 (84.0951) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1954 data: 0.1035 max mem: 3538 +Test: [1790/2387] eta: 0:01:52 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2478 (0.3520) acc1: 91.9271 (84.1396) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2004 data: 0.1086 max mem: 3538 +Test: [1800/2387] eta: 0:01:51 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2510 (0.3515) acc1: 91.6667 (84.1824) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1942 data: 0.1017 max mem: 3538 +Test: [1810/2387] eta: 0:01:49 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2199 (0.3506) acc1: 92.1875 (84.2408) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1904 data: 0.0986 max mem: 3539 +Test: [1820/2387] eta: 0:01:47 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2060 (0.3502) acc1: 94.7917 (84.2718) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1961 data: 0.1049 max mem: 3539 +Test: [1830/2387] eta: 0:01:45 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2321 (0.3494) acc1: 90.6250 (84.3214) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2001 data: 0.1087 max mem: 3539 +Test: [1840/2387] eta: 0:01:43 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1230 (0.3482) acc1: 99.4792 (84.4044) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1975 data: 0.1061 max mem: 3539 +Test: [1850/2387] eta: 0:01:41 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1203 (0.3469) acc1: 99.4792 (84.4854) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1947 data: 0.1023 max mem: 3539 +Test: [1860/2387] eta: 0:01:39 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1197 (0.3457) acc1: 99.4792 (84.5664) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1920 data: 0.1002 max mem: 3539 +Test: [1870/2387] eta: 0:01:37 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1219 (0.3445) acc1: 99.4792 (84.6454) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1797 data: 0.0878 max mem: 3539 +Test: [1880/2387] eta: 0:01:35 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1246 (0.3434) acc1: 99.2188 (84.7237) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1605 data: 0.0667 max mem: 3539 +Test: [1890/2387] eta: 0:01:33 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1243 (0.3422) acc1: 99.2188 (84.8011) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1554 data: 0.0621 max mem: 3539 +Test: [1900/2387] eta: 0:01:31 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1259 (0.3411) acc1: 99.2188 (84.8753) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1582 data: 0.0651 max mem: 3539 +Test: [1910/2387] eta: 0:01:29 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2321 (0.3410) acc1: 96.0938 (84.9014) acc5: 100.0000 (100.0000) acc1_neg: 56.771 acc5_neg: 100.000 time: 0.1545 data: 0.0623 max mem: 3539 +Test: [1920/2387] eta: 0:01:27 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2983 (0.3416) acc1: 88.2812 (84.8760) acc5: 100.0000 (100.0000) acc1_neg: 75.781 acc5_neg: 100.000 time: 0.1513 data: 0.0613 max mem: 3539 +Test: [1930/2387] eta: 0:01:26 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.4186 (0.3420) acc1: 84.6354 (84.8535) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1551 data: 0.0649 max mem: 3539 +Test: [1940/2387] eta: 0:01:24 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.4316 (0.3424) acc1: 76.0417 (84.8171) acc5: 100.0000 (100.0000) acc1_neg: 68.750 acc5_neg: 100.000 time: 0.1590 data: 0.0688 max mem: 3539 +Test: [1950/2387] eta: 0:01:22 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.4835 (0.3433) acc1: 73.6979 (84.7773) acc5: 100.0000 (100.0000) acc1_neg: 73.698 acc5_neg: 100.000 time: 0.1587 data: 0.0671 max mem: 3539 +Test: [1960/2387] eta: 0:01:20 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.4104 (0.3436) acc1: 84.1146 (84.7766) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1651 data: 0.0723 max mem: 3539 +Test: [1970/2387] eta: 0:01:18 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.4397 (0.3453) acc1: 79.6875 (84.6645) acc5: 100.0000 (100.0000) acc1_neg: 70.312 acc5_neg: 100.000 time: 0.1712 data: 0.0769 max mem: 3539 +Test: [1980/2387] eta: 0:01:16 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.5280 (0.3461) acc1: 64.5833 (84.6185) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1644 data: 0.0710 max mem: 3539 +Test: [1990/2387] eta: 0:01:14 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1805 (0.3452) acc1: 96.8750 (84.6812) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1536 data: 0.0618 max mem: 3539 +Test: [2000/2387] eta: 0:01:12 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1780 (0.3444) acc1: 97.1354 (84.7436) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1859 data: 0.0925 max mem: 3539 +Test: [2010/2387] eta: 0:01:10 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1844 (0.3436) acc1: 97.1354 (84.8062) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2010 data: 0.1086 max mem: 3539 +Test: [2020/2387] eta: 0:01:08 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1843 (0.3428) acc1: 97.1354 (84.8685) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1854 data: 0.0945 max mem: 3539 +Test: [2030/2387] eta: 0:01:07 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1820 (0.3420) acc1: 97.1354 (84.9299) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1997 data: 0.1087 max mem: 3540 +Test: [2040/2387] eta: 0:01:05 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1820 (0.3413) acc1: 97.1354 (84.9890) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2047 data: 0.1136 max mem: 3540 +Test: [2050/2387] eta: 0:01:03 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1818 (0.3405) acc1: 97.1354 (85.0493) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2052 data: 0.1139 max mem: 3540 +Test: [2060/2387] eta: 0:01:01 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1756 (0.3397) acc1: 97.3958 (85.1085) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1944 data: 0.1028 max mem: 3540 +Test: [2070/2387] eta: 0:00:59 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1727 (0.3388) acc1: 97.3958 (85.1687) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1911 data: 0.0983 max mem: 3540 +Test: [2080/2387] eta: 0:00:57 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1694 (0.3381) acc1: 97.3958 (85.2263) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2040 data: 0.1093 max mem: 3540 +Test: [2090/2387] eta: 0:00:55 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1687 (0.3373) acc1: 97.3958 (85.2859) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2034 data: 0.1083 max mem: 3540 +Test: [2100/2387] eta: 0:00:53 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1678 (0.3365) acc1: 97.9167 (85.3454) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2049 data: 0.1122 max mem: 3540 +Test: [2110/2387] eta: 0:00:52 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1687 (0.3357) acc1: 97.6562 (85.4030) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2115 data: 0.1182 max mem: 3540 +Test: [2120/2387] eta: 0:00:50 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1769 (0.3353) acc1: 96.6146 (85.4355) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.2167 data: 0.1233 max mem: 3540 +Test: [2130/2387] eta: 0:00:48 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2227 (0.3347) acc1: 94.5312 (85.4825) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2178 data: 0.1258 max mem: 3540 +Test: [2140/2387] eta: 0:00:46 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1830 (0.3342) acc1: 96.6146 (85.5167) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2053 data: 0.1113 max mem: 3540 +Test: [2150/2387] eta: 0:00:44 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1830 (0.3336) acc1: 95.5729 (85.5547) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.1953 data: 0.1022 max mem: 3540 +Test: [2160/2387] eta: 0:00:42 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1802 (0.3330) acc1: 96.0938 (85.6016) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1892 data: 0.0978 max mem: 3540 +Test: [2170/2387] eta: 0:00:40 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2200 (0.3334) acc1: 93.7500 (85.5883) acc5: 100.0000 (100.0000) acc1_neg: 76.302 acc5_neg: 100.000 time: 0.1894 data: 0.0971 max mem: 3540 +Test: [2180/2387] eta: 0:00:39 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.4899 (0.3346) acc1: 76.0417 (85.5028) acc5: 100.0000 (100.0000) acc1_neg: 52.083 acc5_neg: 100.000 time: 0.2050 data: 0.1106 max mem: 3540 +Test: [2190/2387] eta: 0:00:37 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.4707 (0.3350) acc1: 76.0417 (85.4857) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.2078 data: 0.1149 max mem: 3540 +Test: [2200/2387] eta: 0:00:35 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3776 (0.3351) acc1: 84.1146 (85.4843) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2006 data: 0.1095 max mem: 3540 +Test: [2210/2387] eta: 0:00:33 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3982 (0.3356) acc1: 80.4688 (85.4527) acc5: 100.0000 (100.0000) acc1_neg: 73.177 acc5_neg: 100.000 time: 0.2080 data: 0.1133 max mem: 3540 +Test: [2220/2387] eta: 0:00:31 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.4108 (0.3360) acc1: 80.4688 (85.4343) acc5: 100.0000 (100.0000) acc1_neg: 83.073 acc5_neg: 100.000 time: 0.2049 data: 0.1091 max mem: 3540 +Test: [2230/2387] eta: 0:00:29 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3712 (0.3362) acc1: 83.5938 (85.4330) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.1944 data: 0.1001 max mem: 3540 +Test: [2240/2387] eta: 0:00:27 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3847 (0.3367) acc1: 81.7708 (85.4010) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.1991 data: 0.1059 max mem: 3540 +Test: [2250/2387] eta: 0:00:25 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.4792 (0.3379) acc1: 71.8750 (85.3220) acc5: 100.0000 (100.0000) acc1_neg: 55.469 acc5_neg: 100.000 time: 0.1970 data: 0.1054 max mem: 3540 +Test: [2260/2387] eta: 0:00:24 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.4110 (0.3375) acc1: 76.0417 (85.3447) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1948 data: 0.1015 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1041 (0.3365) acc1: 100.0000 (85.4089) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1972 data: 0.1040 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0893 (0.3356) acc1: 100.0000 (85.4546) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1990 data: 0.1084 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.0966 (0.3346) acc1: 100.0000 (85.5172) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2024 data: 0.1119 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1016 (0.3336) acc1: 100.0000 (85.5756) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2061 data: 0.1141 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.1323 (0.3330) acc1: 99.4792 (85.6177) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1981 data: 0.1060 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2519 (0.3329) acc1: 91.1458 (85.6275) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1815 data: 0.0899 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2667 (0.3326) acc1: 91.1458 (85.6592) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1679 data: 0.0757 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2649 (0.3327) acc1: 91.4062 (85.6583) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.1397 data: 0.0490 max mem: 3541 +Test: [2350/2387] eta: 0:00:06 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2708 (0.3325) acc1: 92.7083 (85.6707) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1176 data: 0.0275 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2734 (0.3327) acc1: 92.7083 (85.6655) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1113 data: 0.0215 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.2749 (0.3324) acc1: 88.2812 (85.6977) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1109 data: 0.0212 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.3328 (0.3339) acc1: 86.9792 (85.6110) acc5: 100.0000 (100.0000) acc1_neg: 61.979 acc5_neg: 100.000 time: 0.1108 data: 0.0206 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 69.406 acc5_pos: 100.000 loss: 0.5752 (0.3346) acc1: 68.4896 (85.5731) acc5: 100.0000 (100.0000) acc1_neg: 79.070 acc5_neg: 100.000 time: 0.1049 data: 0.0174 max mem: 3541 +Test: Total time: 0:07:27 (0.1875 s / it) +* Overall Acc@1 85.573 Acc@5 100.000 + target == real Acc@1 45.718 Acc@5 100.000 + target == fake Acc@1 88.372 Acc@5 100.000 + loss 0.335 + PR-AUC 0.9838 ROC-AUC 0.8146 EER 26.77% (thr=0.7358) +Accuracy of the network on the 916267 test images: 85.6% +[rank0]:[W725 18:04:02.700246084 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:04:22.133000 2568640 torch/distributed/run.py:766] +W0725 18:04:22.133000 2568640 torch/distributed/run.py:766] ***************************************** +W0725 18:04:22.133000 2568640 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:04:22.133000 2568640 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:04:49.689618280 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:04:49.767531062 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:04:49.123107310 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:04:50.187986035 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:04:50.294589600 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:04:50.435741551 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:04:50.471424591 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:04:50.643395980 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed3_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed3_resampled_n116-040-168_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 0:51:11 acc1_pos: 51.562 acc5_pos: 100.000 loss: 0.7815 (0.7815) acc1: 51.5625 (51.5625) acc5: 100.0000 (100.0000) time: 5.9411 data: 4.0674 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/517] eta: 0:05:20 acc1_pos: 35.677 acc5_pos: 100.000 loss: 1.0925 (1.0379) acc1: 35.6771 (37.0265) acc5: 100.0000 (100.0000) time: 0.6322 data: 0.3805 max mem: 3531 +Test: [ 20/517] eta: 0:03:18 acc1_pos: 25.260 acc5_pos: 100.000 loss: 1.0925 (1.0975) acc1: 29.9479 (32.6265) acc5: 100.0000 (100.0000) time: 0.1229 data: 0.0287 max mem: 3531 +Test: [ 30/517] eta: 0:02:33 acc1_pos: 16.406 acc5_pos: 100.000 loss: 1.1372 (1.1112) acc1: 25.5208 (31.7204) acc5: 100.0000 (100.0000) time: 0.1393 data: 0.0418 max mem: 3531 +Test: [ 40/517] eta: 0:02:35 acc1_pos: 19.010 acc5_pos: 100.000 loss: 1.2198 (1.1408) acc1: 23.1771 (29.7891) acc5: 100.0000 (100.0000) time: 0.2500 data: 0.1539 max mem: 3531 +Test: [ 50/517] eta: 0:02:35 acc1_pos: 23.394 acc5_pos: 100.000 loss: 1.1730 (1.1201) acc1: 24.7396 (30.9487) acc5: 100.0000 (100.0000) acc1_neg: 72.917 acc5_neg: 100.000 time: 0.3636 data: 0.2707 max mem: 3531 +Test: [ 60/517] eta: 0:02:33 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.6779 (1.0159) acc1: 62.7604 (38.6142) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.3548 data: 0.2597 max mem: 3531 +Test: [ 70/517] eta: 0:02:28 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.3848 (0.9166) acc1: 83.0729 (45.8297) acc5: 100.0000 (100.0000) acc1_neg: 79.688 acc5_neg: 100.000 time: 0.3251 data: 0.2295 max mem: 3531 +Test: [ 80/517] eta: 0:02:21 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.3101 (0.8416) acc1: 93.2292 (51.6204) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2899 data: 0.1975 max mem: 3531 +Test: [ 90/517] eta: 0:02:19 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.3461 (0.7872) acc1: 90.8854 (55.8665) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.3098 data: 0.2147 max mem: 3531 +Test: [100/517] eta: 0:02:15 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.3613 (0.7459) acc1: 89.0625 (58.8825) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.3302 data: 0.2367 max mem: 3531 +Test: [110/517] eta: 0:02:11 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2798 (0.6996) acc1: 91.9271 (62.1833) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.3040 data: 0.2091 max mem: 3531 +Test: [120/517] eta: 0:02:08 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2798 (0.6759) acc1: 92.9688 (63.9377) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.3147 data: 0.2188 max mem: 3531 +Test: [130/517] eta: 0:02:03 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.3626 (0.6498) acc1: 86.7188 (65.9490) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2999 data: 0.2048 max mem: 3531 +Test: [140/517] eta: 0:02:01 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.3358 (0.6274) acc1: 90.1042 (67.6271) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.3106 data: 0.2155 max mem: 3531 +Test: [150/517] eta: 0:01:55 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.3205 (0.6052) acc1: 91.4062 (69.2708) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.2949 data: 0.1996 max mem: 3531 +Test: [160/517] eta: 0:01:50 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.3162 (0.5867) acc1: 91.4062 (70.6328) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2291 data: 0.1292 max mem: 3531 +Test: [170/517] eta: 0:01:46 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2601 (0.5659) acc1: 94.7917 (72.1720) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2299 data: 0.1309 max mem: 3531 +Test: [180/517] eta: 0:01:41 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2276 (0.5487) acc1: 96.0938 (73.3684) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2349 data: 0.1382 max mem: 3531 +Test: [190/517] eta: 0:01:38 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.1991 (0.5299) acc1: 97.9167 (74.6782) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2549 data: 0.1573 max mem: 3531 +Test: [200/517] eta: 0:01:35 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.1935 (0.5157) acc1: 98.6979 (75.6737) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2962 data: 0.1998 max mem: 3531 +Test: [210/517] eta: 0:01:31 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.3330 (0.5136) acc1: 88.5417 (75.8343) acc5: 100.0000 (100.0000) acc1_neg: 77.344 acc5_neg: 100.000 time: 0.2699 data: 0.1738 max mem: 3532 +Test: [220/517] eta: 0:01:27 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.4650 (0.5090) acc1: 79.1667 (76.1878) acc5: 100.0000 (100.0000) acc1_neg: 84.635 acc5_neg: 100.000 time: 0.2286 data: 0.1329 max mem: 3532 +Test: [230/517] eta: 0:01:24 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2964 (0.4992) acc1: 92.4479 (76.9785) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.2449 data: 0.1483 max mem: 3532 +Test: [240/517] eta: 0:01:20 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2846 (0.4907) acc1: 92.9688 (77.6690) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.2399 data: 0.1426 max mem: 3532 +Test: [250/517] eta: 0:01:16 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2363 (0.4795) acc1: 95.8333 (78.4435) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2299 data: 0.1349 max mem: 3532 +Test: [260/517] eta: 0:01:13 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.1883 (0.4679) acc1: 98.9583 (79.2186) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.2299 data: 0.1358 max mem: 3532 +Test: [270/517] eta: 0:01:10 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2336 (0.4631) acc1: 95.8333 (79.6837) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2399 data: 0.1435 max mem: 3532 +Test: [280/517] eta: 0:01:07 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.3025 (0.4563) acc1: 93.7500 (80.2176) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2449 data: 0.1468 max mem: 3532 +Test: [290/517] eta: 0:01:03 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2121 (0.4481) acc1: 97.6562 (80.8222) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.2349 data: 0.1351 max mem: 3532 +Test: [300/517] eta: 0:01:00 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2149 (0.4405) acc1: 98.1771 (81.3962) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2399 data: 0.1390 max mem: 3532 +Test: [310/517] eta: 0:00:57 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2366 (0.4362) acc1: 97.1354 (81.7030) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2399 data: 0.1437 max mem: 3532 +Test: [320/517] eta: 0:00:54 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2765 (0.4306) acc1: 94.0104 (82.1140) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2449 data: 0.1516 max mem: 3532 +Test: [330/517] eta: 0:00:51 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.1946 (0.4233) acc1: 97.3958 (82.6048) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2399 data: 0.1450 max mem: 3532 +Test: [340/517] eta: 0:00:48 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.1882 (0.4196) acc1: 97.9167 (82.8469) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.2349 data: 0.1396 max mem: 3532 +Test: [350/517] eta: 0:00:45 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.4378 (0.4211) acc1: 85.4167 (82.7628) acc5: 100.0000 (100.0000) acc1_neg: 74.740 acc5_neg: 100.000 time: 0.2449 data: 0.1483 max mem: 3532 +Test: [360/517] eta: 0:00:42 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.4215 (0.4188) acc1: 85.1562 (82.9561) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2449 data: 0.1441 max mem: 3532 +Test: [370/517] eta: 0:00:40 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2662 (0.4143) acc1: 95.5729 (83.2996) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2399 data: 0.1410 max mem: 3532 +Test: [380/517] eta: 0:00:37 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2662 (0.4116) acc1: 95.3125 (83.5001) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2249 data: 0.1308 max mem: 3532 +Test: [390/517] eta: 0:00:34 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2760 (0.4078) acc1: 92.4479 (83.7669) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.2249 data: 0.1301 max mem: 3532 +Test: [400/517] eta: 0:00:31 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2665 (0.4052) acc1: 93.2292 (83.9496) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2349 data: 0.1388 max mem: 3532 +Test: [410/517] eta: 0:00:28 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2924 (0.4032) acc1: 93.4896 (84.1463) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.2399 data: 0.1449 max mem: 3532 +Test: [420/517] eta: 0:00:26 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2926 (0.4005) acc1: 96.0938 (84.4233) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.2449 data: 0.1497 max mem: 3532 +Test: [430/517] eta: 0:00:23 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2837 (0.3973) acc1: 96.3542 (84.6463) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.2599 data: 0.1620 max mem: 3532 +Test: [440/517] eta: 0:00:20 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2756 (0.3946) acc1: 94.7917 (84.8498) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.2349 data: 0.1377 max mem: 3533 +Test: [450/517] eta: 0:00:17 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2250 (0.3904) acc1: 96.3542 (85.1586) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2249 data: 0.1303 max mem: 3533 +Test: [460/517] eta: 0:00:15 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2141 (0.3870) acc1: 98.4375 (85.4099) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.2399 data: 0.1446 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2257 (0.3836) acc1: 98.1771 (85.6765) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.2205 data: 0.1260 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2085 (0.3797) acc1: 98.1771 (85.9370) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.2249 data: 0.1280 max mem: 3533 +Test: [490/517] eta: 0:00:07 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2085 (0.3770) acc1: 97.9167 (86.1725) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.2194 data: 0.1221 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2702 (0.3756) acc1: 96.6146 (86.3092) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.2049 data: 0.1118 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2672 (0.3729) acc1: 95.0521 (86.4757) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1838 data: 0.0902 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 23.394 acc5_pos: 100.000 loss: 0.2021 (0.3709) acc1: 97.9167 (86.6102) acc5: 100.0000 (100.0000) acc1_neg: 98.587 acc5_neg: 100.000 time: 0.1639 data: 0.0695 max mem: 3533 +Test: Total time: 0:02:14 (0.2593 s / it) +* Overall Acc@1 86.610 Acc@5 100.000 + target == real Acc@1 28.842 Acc@5 100.000 + target == fake Acc@1 92.603 Acc@5 100.000 + loss 0.371 + PR-AUC 0.9543 ROC-AUC 0.7192 EER 33.87% (thr=0.7563) +Accuracy of the network on the 198427 test images: 86.6% +[rank0]:[W725 18:07:17.088731949 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0725 18:07:37.456000 2569229 torch/distributed/run.py:766] +W0725 18:07:37.456000 2569229 torch/distributed/run.py:766] ***************************************** +W0725 18:07:37.456000 2569229 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:07:37.456000 2569229 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +[rank1]:[W725 18:08:03.655793504 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:08:03.662607829 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:08:03.963028777 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:08:04.185775759 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:08:04.270217107 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:08:04.296258543 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:08:04.306789538 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:08:04.309070278 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed3_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed3_resampled_n116-040-168_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:32:43 acc1_pos: 85.417 acc5_pos: 100.000 loss: 0.2725 (0.2725) acc1: 85.4167 (85.4167) acc5: 100.0000 (100.0000) time: 5.0485 data: 3.5201 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/389] eta: 0:03:24 acc1_pos: 64.323 acc5_pos: 100.000 loss: 0.2725 (0.3717) acc1: 87.5000 (80.0426) acc5: 100.0000 (100.0000) time: 0.5407 data: 0.3204 max mem: 3531 +Test: [ 20/389] eta: 0:02:00 acc1_pos: 91.146 acc5_pos: 100.000 loss: 0.2908 (0.4463) acc1: 87.5000 (73.9459) acc5: 100.0000 (100.0000) time: 0.0906 data: 0.0004 max mem: 3531 +Test: [ 30/389] eta: 0:01:48 acc1_pos: 87.760 acc5_pos: 100.000 loss: 0.3688 (0.4212) acc1: 81.5104 (77.6882) acc5: 100.0000 (100.0000) time: 0.1726 data: 0.0796 max mem: 3531 +Test: [ 40/389] eta: 0:01:42 acc1_pos: 88.542 acc5_pos: 100.000 loss: 0.3805 (0.4152) acc1: 81.5104 (78.8173) acc5: 100.0000 (100.0000) time: 0.2613 data: 0.1684 max mem: 3531 +Test: [ 50/389] eta: 0:01:33 acc1_pos: 66.667 acc5_pos: 100.000 loss: 0.3799 (0.4113) acc1: 81.2500 (79.4679) acc5: 100.0000 (100.0000) time: 0.2346 data: 0.1427 max mem: 3531 +Test: [ 60/389] eta: 0:01:25 acc1_pos: 73.958 acc5_pos: 100.000 loss: 0.3799 (0.4124) acc1: 73.9583 (79.2947) acc5: 100.0000 (100.0000) time: 0.1885 data: 0.0968 max mem: 3531 +Test: [ 70/389] eta: 0:01:22 acc1_pos: 92.969 acc5_pos: 100.000 loss: 0.2202 (0.3853) acc1: 92.9688 (81.5691) acc5: 100.0000 (100.0000) time: 0.2100 data: 0.1187 max mem: 3531 +Test: [ 80/389] eta: 0:01:18 acc1_pos: 99.219 acc5_pos: 100.000 loss: 0.2202 (0.3725) acc1: 92.9688 (82.5778) acc5: 100.0000 (100.0000) time: 0.2307 data: 0.1385 max mem: 3531 +Test: [ 90/389] eta: 0:01:13 acc1_pos: 92.708 acc5_pos: 100.000 loss: 0.1874 (0.3517) acc1: 94.2708 (84.0459) acc5: 100.0000 (100.0000) time: 0.2033 data: 0.1108 max mem: 3531 +Test: [100/389] eta: 0:01:10 acc1_pos: 88.802 acc5_pos: 100.000 loss: 0.2832 (0.3637) acc1: 92.7083 (83.4700) acc5: 100.0000 (100.0000) time: 0.2051 data: 0.1125 max mem: 3531 +Test: [110/389] eta: 0:01:07 acc1_pos: 98.698 acc5_pos: 100.000 loss: 0.4490 (0.3724) acc1: 82.5521 (82.9908) acc5: 100.0000 (100.0000) time: 0.2307 data: 0.1377 max mem: 3531 +Test: [120/389] eta: 0:01:03 acc1_pos: 76.042 acc5_pos: 100.000 loss: 0.4514 (0.3759) acc1: 80.7292 (82.9545) acc5: 100.0000 (100.0000) time: 0.2093 data: 0.1166 max mem: 3531 +Test: [130/389] eta: 0:01:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5493 (0.4070) acc1: 70.3125 (80.5940) acc5: 100.0000 (100.0000) acc1_neg: 54.948 acc5_neg: 100.000 time: 0.1794 data: 0.0874 max mem: 3531 +Test: [140/389] eta: 0:00:58 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.8559 (0.4456) acc1: 42.4479 (77.6466) acc5: 100.0000 (100.0000) acc1_neg: 55.469 acc5_neg: 100.000 time: 0.2062 data: 0.1115 max mem: 3531 +Test: [150/389] eta: 0:00:54 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.0874 (0.4887) acc1: 26.3021 (74.9828) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1925 data: 0.0968 max mem: 3531 +Test: [160/389] eta: 0:00:51 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6732 (0.4967) acc1: 51.3021 (74.6619) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1694 data: 0.0762 max mem: 3531 +Test: [170/389] eta: 0:00:48 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5711 (0.5132) acc1: 73.4375 (73.9492) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1819 data: 0.0893 max mem: 3531 +Test: [180/389] eta: 0:00:46 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6153 (0.5189) acc1: 69.2708 (73.7137) acc5: 100.0000 (100.0000) acc1_neg: 44.792 acc5_neg: 100.000 time: 0.1866 data: 0.0923 max mem: 3531 +Test: [190/389] eta: 0:00:43 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.8106 (0.5615) acc1: 54.4271 (71.3228) acc5: 100.0000 (100.0000) acc1_neg: 27.344 acc5_neg: 100.000 time: 0.1872 data: 0.0910 max mem: 3531 +Test: [200/389] eta: 0:00:40 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.9462 (0.5858) acc1: 28.6458 (69.5196) acc5: 100.0000 (100.0000) acc1_neg: 31.250 acc5_neg: 100.000 time: 0.1772 data: 0.0838 max mem: 3531 +Test: [210/389] eta: 0:00:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.7835 (0.5843) acc1: 31.2500 (69.5621) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1942 data: 0.1037 max mem: 3532 +Test: [220/389] eta: 0:00:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3770 (0.5927) acc1: 83.8542 (69.5053) acc5: 100.0000 (100.0000) acc1_neg: 0.781 acc5_neg: 100.000 time: 0.1998 data: 0.1072 max mem: 3532 +Test: [230/389] eta: 0:00:33 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6402 (0.6100) acc1: 52.0833 (68.3148) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.1840 data: 0.0911 max mem: 3532 +Test: [240/389] eta: 0:00:31 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.8293 (0.6158) acc1: 40.3646 (67.8521) acc5: 100.0000 (100.0000) acc1_neg: 4.167 acc5_neg: 100.000 time: 0.1772 data: 0.0858 max mem: 3532 +Test: [250/389] eta: 0:00:29 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.3697 (0.6495) acc1: 11.4583 (65.4880) acc5: 100.0000 (100.0000) acc1_neg: 10.156 acc5_neg: 100.000 time: 0.1809 data: 0.0885 max mem: 3532 +Test: [260/389] eta: 0:00:27 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.3978 (0.6740) acc1: 9.3750 (63.7063) acc5: 100.0000 (100.0000) acc1_neg: 0.260 acc5_neg: 100.000 time: 0.2200 data: 0.1281 max mem: 3532 +Test: [270/389] eta: 0:00:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.2936 (0.7025) acc1: 9.6354 (62.1550) acc5: 100.0000 (100.0000) acc1_neg: 23.698 acc5_neg: 100.000 time: 0.2232 data: 0.1309 max mem: 3532 +Test: [280/389] eta: 0:00:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.5559 (0.7321) acc1: 6.2500 (60.6409) acc5: 100.0000 (100.0000) acc1_neg: 77.083 acc5_neg: 100.000 time: 0.1747 data: 0.0797 max mem: 3532 +Test: [290/389] eta: 0:00:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6542 (0.7263) acc1: 39.3229 (61.0825) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.1603 data: 0.0657 max mem: 3532 +Test: [300/389] eta: 0:00:18 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6042 (0.7283) acc1: 67.9688 (60.9038) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.1789 data: 0.0847 max mem: 3532 +Test: [310/389] eta: 0:00:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.8625 (0.7436) acc1: 42.4479 (59.8138) acc5: 100.0000 (100.0000) acc1_neg: 1.562 acc5_neg: 100.000 time: 0.1909 data: 0.0985 max mem: 3532 +Test: [320/389] eta: 0:00:14 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.1323 (0.7613) acc1: 13.5417 (58.6075) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1712 data: 0.0793 max mem: 3532 +Test: [330/389] eta: 0:00:12 acc1_pos: 100.000 acc5_pos: 100.000 loss: 1.6014 (0.8087) acc1: 1.3021 (57.0446) acc5: 100.0000 (100.0000) acc1_neg: 0.521 acc5_neg: 100.000 time: 0.1510 data: 0.0579 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.9848 (0.8033) acc1: 26.8229 (57.3902) acc5: 100.0000 (100.0000) acc1_neg: 42.448 acc5_neg: 100.000 time: 0.1400 data: 0.0488 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6967 (0.8090) acc1: 53.9062 (56.9170) acc5: 100.0000 (100.0000) acc1_neg: 56.771 acc5_neg: 100.000 time: 0.1270 data: 0.0371 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6812 (0.8006) acc1: 56.7708 (57.4937) acc5: 100.0000 (100.0000) acc1_neg: 86.979 acc5_neg: 100.000 time: 0.1201 data: 0.0300 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4696 (0.8050) acc1: 81.7708 (57.3864) acc5: 100.0000 (100.0000) acc1_neg: 84.896 acc5_neg: 100.000 time: 0.1176 data: 0.0276 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6944 (0.8042) acc1: 55.4688 (57.3361) acc5: 100.0000 (100.0000) acc1_neg: 50.000 acc5_neg: 100.000 time: 0.1119 data: 0.0218 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6944 (0.8062) acc1: 55.4688 (57.1580) acc5: 100.0000 (100.0000) acc1_neg: 51.366 acc5_neg: 100.000 time: 0.1076 data: 0.0143 max mem: 3532 +Test: Total time: 0:01:14 (0.1915 s / it) +* Overall Acc@1 57.158 Acc@5 100.000 + target == real Acc@1 82.614 Acc@5 100.000 + target == fake Acc@1 45.325 Acc@5 100.000 + loss 0.806 + PR-AUC 0.8434 ROC-AUC 0.7053 EER 36.04% (thr=0.3157) +Accuracy of the network on the 149358 test images: 57.2% +[rank0]:[W725 18:09:27.020653975 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:09:47.240000 2569785 torch/distributed/run.py:766] +W0725 18:09:47.240000 2569785 torch/distributed/run.py:766] ***************************************** +W0725 18:09:47.240000 2569785 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:09:47.240000 2569785 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:10:15.582575109 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:10:16.268827917 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank3]:[W725 18:10:16.268828618 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:10:16.457504574 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:10:16.618013536 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:10:16.714551902 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:10:16.762405688 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:10:16.790865799 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed3_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed3_resampled_n116-040-168_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:27:49 acc1_pos: 55.729 acc5_pos: 100.000 loss: 0.8297 (0.8297) acc1: 55.7292 (55.7292) acc5: 100.0000 (100.0000) time: 5.0128 data: 2.9324 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/333] eta: 0:03:01 acc1_pos: 78.906 acc5_pos: 100.000 loss: 0.5728 (0.5660) acc1: 70.8333 (70.5729) acc5: 100.0000 (100.0000) time: 0.5635 data: 0.2929 max mem: 3531 +Test: [ 20/333] eta: 0:01:55 acc1_pos: 89.062 acc5_pos: 100.000 loss: 0.4744 (0.4973) acc1: 75.7812 (75.2232) acc5: 100.0000 (100.0000) time: 0.1356 data: 0.0428 max mem: 3531 +Test: [ 30/333] eta: 0:01:38 acc1_pos: 46.094 acc5_pos: 100.000 loss: 0.5239 (0.6307) acc1: 70.8333 (65.6502) acc5: 100.0000 (100.0000) time: 0.1915 data: 0.0957 max mem: 3531 +Test: [ 40/333] eta: 0:01:34 acc1_pos: 43.229 acc5_pos: 100.000 loss: 0.9187 (0.6927) acc1: 43.2292 (60.7279) acc5: 100.0000 (100.0000) time: 0.2757 data: 0.1804 max mem: 3531 +Test: [ 50/333] eta: 0:01:20 acc1_pos: 42.448 acc5_pos: 100.000 loss: 0.9210 (0.7432) acc1: 41.6667 (56.3011) acc5: 100.0000 (100.0000) time: 0.2247 data: 0.1282 max mem: 3531 +Test: [ 60/333] eta: 0:01:10 acc1_pos: 43.229 acc5_pos: 100.000 loss: 0.9511 (0.7815) acc1: 38.5417 (53.7526) acc5: 100.0000 (100.0000) time: 0.1240 data: 0.0259 max mem: 3531 +Test: [ 70/333] eta: 0:01:05 acc1_pos: 38.802 acc5_pos: 100.000 loss: 0.9364 (0.8003) acc1: 39.8438 (52.4354) acc5: 100.0000 (100.0000) time: 0.1602 data: 0.0635 max mem: 3531 +Test: [ 80/333] eta: 0:01:00 acc1_pos: 43.229 acc5_pos: 100.000 loss: 0.8870 (0.8157) acc1: 43.2292 (51.1928) acc5: 100.0000 (100.0000) time: 0.1811 data: 0.0858 max mem: 3531 +Test: [ 90/333] eta: 0:00:57 acc1_pos: 59.115 acc5_pos: 100.000 loss: 0.8371 (0.8091) acc1: 47.6562 (51.6913) acc5: 100.0000 (100.0000) time: 0.1879 data: 0.0947 max mem: 3531 +Test: [100/333] eta: 0:00:54 acc1_pos: 61.979 acc5_pos: 100.000 loss: 0.7440 (0.8024) acc1: 54.6875 (52.1401) acc5: 100.0000 (100.0000) time: 0.2096 data: 0.1162 max mem: 3531 +Test: [110/333] eta: 0:00:51 acc1_pos: 64.844 acc5_pos: 100.000 loss: 0.7496 (0.7997) acc1: 52.8646 (52.1373) acc5: 100.0000 (100.0000) time: 0.1971 data: 0.1041 max mem: 3531 +Test: [120/333] eta: 0:00:48 acc1_pos: 56.250 acc5_pos: 100.000 loss: 0.7573 (0.7937) acc1: 51.5625 (52.5870) acc5: 100.0000 (100.0000) time: 0.1892 data: 0.0979 max mem: 3531 +Test: [130/333] eta: 0:00:46 acc1_pos: 54.427 acc5_pos: 100.000 loss: 0.6949 (0.7855) acc1: 56.2500 (53.0733) acc5: 100.0000 (100.0000) time: 0.2147 data: 0.1228 max mem: 3531 +Test: [140/333] eta: 0:00:43 acc1_pos: 77.083 acc5_pos: 100.000 loss: 0.6463 (0.7748) acc1: 63.5417 (53.8619) acc5: 100.0000 (100.0000) time: 0.2247 data: 0.1319 max mem: 3531 +Test: [150/333] eta: 0:00:41 acc1_pos: 69.531 acc5_pos: 100.000 loss: 0.6678 (0.7721) acc1: 59.3750 (53.9545) acc5: 100.0000 (100.0000) time: 0.2097 data: 0.1159 max mem: 3531 +Test: [160/333] eta: 0:00:39 acc1_pos: 56.908 acc5_pos: 100.000 loss: 0.7425 (0.7722) acc1: 53.9062 (54.0680) acc5: 100.0000 (100.0000) acc1_neg: 44.271 acc5_neg: 100.000 time: 0.2378 data: 0.1433 max mem: 3531 +Test: [170/333] eta: 0:00:36 acc1_pos: 56.908 acc5_pos: 100.000 loss: 0.7762 (0.7739) acc1: 55.2083 (54.0692) acc5: 100.0000 (100.0000) acc1_neg: 57.812 acc5_neg: 100.000 time: 0.2301 data: 0.1371 max mem: 3531 +Test: [180/333] eta: 0:00:33 acc1_pos: 56.908 acc5_pos: 100.000 loss: 0.7762 (0.7730) acc1: 55.2083 (54.1882) acc5: 100.0000 (100.0000) acc1_neg: 69.531 acc5_neg: 100.000 time: 0.1629 data: 0.0690 max mem: 3531 +Test: [190/333] eta: 0:00:31 acc1_pos: 56.908 acc5_pos: 100.000 loss: 0.6035 (0.7588) acc1: 61.9792 (55.3815) acc5: 100.0000 (100.0000) acc1_neg: 73.698 acc5_neg: 100.000 time: 0.1638 data: 0.0695 max mem: 3531 +Test: [200/333] eta: 0:00:28 acc1_pos: 56.908 acc5_pos: 100.000 loss: 0.5207 (0.7508) acc1: 73.6979 (56.0868) acc5: 100.0000 (100.0000) acc1_neg: 64.844 acc5_neg: 100.000 time: 0.1945 data: 0.1031 max mem: 3531 +Test: [210/333] eta: 0:00:26 acc1_pos: 56.908 acc5_pos: 100.000 loss: 0.6572 (0.7469) acc1: 64.8438 (56.4512) acc5: 100.0000 (100.0000) acc1_neg: 65.365 acc5_neg: 100.000 time: 0.1859 data: 0.0949 max mem: 3532 +Test: [220/333] eta: 0:00:24 acc1_pos: 56.908 acc5_pos: 100.000 loss: 0.6719 (0.7459) acc1: 63.2812 (56.6082) acc5: 100.0000 (100.0000) acc1_neg: 66.667 acc5_neg: 100.000 time: 0.1852 data: 0.0930 max mem: 3532 +Test: [230/333] eta: 0:00:21 acc1_pos: 56.908 acc5_pos: 100.000 loss: 0.6017 (0.7372) acc1: 67.4479 (57.3300) acc5: 100.0000 (100.0000) acc1_neg: 75.781 acc5_neg: 100.000 time: 0.1860 data: 0.0934 max mem: 3532 +Test: [240/333] eta: 0:00:19 acc1_pos: 56.908 acc5_pos: 100.000 loss: 0.5853 (0.7314) acc1: 71.6146 (57.8752) acc5: 100.0000 (100.0000) acc1_neg: 70.312 acc5_neg: 100.000 time: 0.1785 data: 0.0872 max mem: 3532 +Test: [250/333] eta: 0:00:17 acc1_pos: 56.908 acc5_pos: 100.000 loss: 0.5925 (0.7262) acc1: 69.5312 (58.2960) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.1838 data: 0.0928 max mem: 3532 +Test: [260/333] eta: 0:00:15 acc1_pos: 56.908 acc5_pos: 100.000 loss: 0.5843 (0.7222) acc1: 68.4896 (58.6845) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.1863 data: 0.0952 max mem: 3532 +Test: [270/333] eta: 0:00:13 acc1_pos: 56.908 acc5_pos: 100.000 loss: 0.6205 (0.7196) acc1: 65.1042 (58.8974) acc5: 100.0000 (100.0000) acc1_neg: 60.677 acc5_neg: 100.000 time: 0.1795 data: 0.0857 max mem: 3532 +Test: [280/333] eta: 0:00:11 acc1_pos: 56.908 acc5_pos: 100.000 loss: 0.6582 (0.7203) acc1: 63.0208 (58.8505) acc5: 100.0000 (100.0000) acc1_neg: 56.771 acc5_neg: 100.000 time: 0.2097 data: 0.1167 max mem: 3532 +Test: [290/333] eta: 0:00:08 acc1_pos: 56.908 acc5_pos: 100.000 loss: 0.8229 (0.7250) acc1: 52.3438 (58.5642) acc5: 100.0000 (100.0000) acc1_neg: 46.094 acc5_neg: 100.000 time: 0.1871 data: 0.0956 max mem: 3532 +Test: [300/333] eta: 0:00:06 acc1_pos: 56.908 acc5_pos: 100.000 loss: 0.8926 (0.7316) acc1: 46.0938 (58.0989) acc5: 100.0000 (100.0000) acc1_neg: 45.052 acc5_neg: 100.000 time: 0.1728 data: 0.0817 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 56.908 acc5_pos: 100.000 loss: 0.8276 (0.7254) acc1: 52.6042 (58.6088) acc5: 100.0000 (100.0000) acc1_neg: 77.083 acc5_neg: 100.000 time: 0.1678 data: 0.0778 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 56.908 acc5_pos: 100.000 loss: 0.5333 (0.7205) acc1: 73.4375 (59.0343) acc5: 100.0000 (100.0000) acc1_neg: 70.573 acc5_neg: 100.000 time: 0.1126 data: 0.0223 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 56.908 acc5_pos: 100.000 loss: 0.5913 (0.7177) acc1: 69.2708 (59.2790) acc5: 100.0000 (100.0000) acc1_neg: 72.396 acc5_neg: 100.000 time: 0.1078 data: 0.0177 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 56.908 acc5_pos: 100.000 loss: 0.6133 (0.7173) acc1: 67.9688 (59.3165) acc5: 100.0000 (100.0000) acc1_neg: 67.606 acc5_neg: 100.000 time: 0.1184 data: 0.0251 max mem: 3532 +Test: Total time: 0:01:06 (0.1993 s / it) +* Overall Acc@1 59.317 Acc@5 100.000 + target == real Acc@1 54.363 Acc@5 100.000 + target == fake Acc@1 63.726 Acc@5 100.000 + loss 0.717 + PR-AUC 0.6282 ROC-AUC 0.6253 EER 41.18% (thr=0.5386) +Accuracy of the network on the 127843 test images: 59.3% +[rank0]:[W725 18:11:31.739077516 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:11:50.079000 2570354 torch/distributed/run.py:766] +W0725 18:11:50.079000 2570354 torch/distributed/run.py:766] ***************************************** +W0725 18:11:50.079000 2570354 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:11:50.079000 2570354 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:12:18.688234459 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:12:18.115841858 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:12:19.191616419 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:12:19.227331189 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:12:19.421114182 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:12:19.459341318 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:12:19.470140071 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:12:19.475335588 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed3_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed3_resampled_n116-040-168_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:19:12 acc1_pos: 92.969 acc5_pos: 100.000 loss: 0.2515 (0.2515) acc1: 92.9688 (92.9688) acc5: 100.0000 (100.0000) time: 4.9051 data: 2.5734 max mem: 3531 +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 10/235] eta: 0:02:01 acc1_pos: 91.406 acc5_pos: 100.000 loss: 0.2586 (0.2638) acc1: 92.4479 (92.0691) acc5: 100.0000 (100.0000) time: 0.5397 data: 0.2463 max mem: 3531 +Test: [ 20/235] eta: 0:01:14 acc1_pos: 90.885 acc5_pos: 100.000 loss: 0.2575 (0.2590) acc1: 92.4479 (92.3735) acc5: 100.0000 (100.0000) time: 0.1168 data: 0.0239 max mem: 3531 +Test: [ 30/235] eta: 0:01:02 acc1_pos: 93.490 acc5_pos: 100.000 loss: 0.2499 (0.2565) acc1: 92.7083 (92.4059) acc5: 100.0000 (100.0000) time: 0.1755 data: 0.0793 max mem: 3531 +Test: [ 40/235] eta: 0:00:59 acc1_pos: 90.625 acc5_pos: 100.000 loss: 0.2463 (0.2564) acc1: 92.7083 (92.3399) acc5: 100.0000 (100.0000) time: 0.2657 data: 0.1717 max mem: 3531 +Test: [ 50/235] eta: 0:00:53 acc1_pos: 89.323 acc5_pos: 100.000 loss: 0.2544 (0.2587) acc1: 92.1875 (92.1977) acc5: 100.0000 (100.0000) time: 0.2710 data: 0.1791 max mem: 3531 +Test: [ 60/235] eta: 0:00:47 acc1_pos: 92.188 acc5_pos: 100.000 loss: 0.2673 (0.2597) acc1: 91.9271 (92.2345) acc5: 100.0000 (100.0000) time: 0.2050 data: 0.1135 max mem: 3531 +Test: [ 70/235] eta: 0:00:43 acc1_pos: 98.462 acc5_pos: 100.000 loss: 0.2819 (0.4060) acc1: 90.6250 (83.3847) acc5: 100.0000 (100.0000) acc1_neg: 18.750 acc5_neg: 100.000 time: 0.1884 data: 0.0957 max mem: 3531 +Test: [ 80/235] eta: 0:00:39 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4239 (0.5332) acc1: 21.8750 (75.7973) acc5: 100.0000 (100.0000) acc1_neg: 21.354 acc5_neg: 100.000 time: 0.1915 data: 0.0963 max mem: 3531 +Test: [ 90/235] eta: 0:00:35 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4363 (0.6351) acc1: 21.0938 (69.6486) acc5: 100.0000 (100.0000) acc1_neg: 19.010 acc5_neg: 100.000 time: 0.1807 data: 0.0868 max mem: 3531 +Test: [100/235] eta: 0:00:32 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4364 (0.7149) acc1: 20.0521 (64.8670) acc5: 100.0000 (100.0000) acc1_neg: 18.490 acc5_neg: 100.000 time: 0.1737 data: 0.0813 max mem: 3531 +Test: [110/235] eta: 0:00:29 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4479 (0.7807) acc1: 20.3125 (60.8835) acc5: 100.0000 (100.0000) acc1_neg: 22.396 acc5_neg: 100.000 time: 0.1868 data: 0.0934 max mem: 3531 +Test: [120/235] eta: 0:00:26 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4596 (0.8386) acc1: 19.5312 (57.5004) acc5: 100.0000 (100.0000) acc1_neg: 22.396 acc5_neg: 100.000 time: 0.1978 data: 0.1052 max mem: 3531 +Test: [130/235] eta: 0:00:24 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4741 (0.8868) acc1: 20.3125 (54.6875) acc5: 100.0000 (100.0000) acc1_neg: 21.094 acc5_neg: 100.000 time: 0.2102 data: 0.1170 max mem: 3531 +Test: [140/235] eta: 0:00:21 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4938 (0.9309) acc1: 20.0521 (52.2237) acc5: 100.0000 (100.0000) acc1_neg: 17.708 acc5_neg: 100.000 time: 0.1949 data: 0.1017 max mem: 3531 +Test: [150/235] eta: 0:00:18 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4943 (0.9669) acc1: 20.0521 (50.1466) acc5: 100.0000 (100.0000) acc1_neg: 22.135 acc5_neg: 100.000 time: 0.1546 data: 0.0632 max mem: 3531 +Test: [160/235] eta: 0:00:16 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4761 (0.9986) acc1: 21.0938 (48.3291) acc5: 100.0000 (100.0000) acc1_neg: 23.177 acc5_neg: 100.000 time: 0.1834 data: 0.0900 max mem: 3531 +Test: [170/235] eta: 0:00:14 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.4278 (1.0180) acc1: 21.8750 (46.9085) acc5: 100.0000 (100.0000) acc1_neg: 24.219 acc5_neg: 100.000 time: 0.2749 data: 0.1785 max mem: 3531 +Test: [180/235] eta: 0:00:12 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.3114 (1.0345) acc1: 24.2188 (45.6751) acc5: 100.0000 (100.0000) acc1_neg: 24.219 acc5_neg: 100.000 time: 0.2299 data: 0.1332 max mem: 3531 +Test: [190/235] eta: 0:00:09 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.2944 (1.0491) acc1: 25.0000 (44.5953) acc5: 100.0000 (100.0000) acc1_neg: 21.875 acc5_neg: 100.000 time: 0.1567 data: 0.0622 max mem: 3531 +Test: [200/235] eta: 0:00:07 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.3057 (1.0634) acc1: 23.4375 (43.5440) acc5: 100.0000 (100.0000) acc1_neg: 20.052 acc5_neg: 100.000 time: 0.1863 data: 0.0930 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.3267 (1.0756) acc1: 23.4375 (42.6158) acc5: 100.0000 (100.0000) acc1_neg: 24.219 acc5_neg: 100.000 time: 0.1753 data: 0.0828 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.3040 (1.0861) acc1: 24.7396 (41.8128) acc5: 100.0000 (100.0000) acc1_neg: 25.000 acc5_neg: 100.000 time: 0.1521 data: 0.0613 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.3040 (1.0960) acc1: 24.7396 (41.0804) acc5: 100.0000 (100.0000) acc1_neg: 23.698 acc5_neg: 100.000 time: 0.1332 data: 0.0434 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 98.462 acc5_pos: 100.000 loss: 1.3162 (1.1012) acc1: 24.2188 (40.8483) acc5: 100.0000 (100.0000) acc1_neg: 17.391 acc5_neg: 100.000 time: 0.1162 data: 0.0295 max mem: 3532 +Test: Total time: 0:00:48 (0.2068 s / it) +* Overall Acc@1 40.848 Acc@5 100.000 + target == real Acc@1 92.251 Acc@5 100.000 + target == fake Acc@1 22.257 Acc@5 100.000 + loss 1.101 + PR-AUC 0.8426 ROC-AUC 0.6777 EER 37.26% (thr=0.1788) +Accuracy of the network on the 89879 test images: 40.8% +[rank0]:[W725 18:13:16.907439130 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:13:35.377000 2570923 torch/distributed/run.py:766] +W0725 18:13:35.377000 2570923 torch/distributed/run.py:766] ***************************************** +W0725 18:13:35.377000 2570923 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:13:35.377000 2570923 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W725 18:14:03.300749267 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:14:03.781490383 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:14:03.913165322 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:14:03.980841451 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:14:03.127885550 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:14:04.363276807 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:14:04.386151101 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank1]:[W725 18:14:04.386188906 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed3_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_fake_methods_n11_classseed3_resampled_n116-040-168_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:30:58 acc1_pos: 9.375 acc5_pos: 100.000 loss: 1.3855 (1.3855) acc1: 9.3750 (9.3750) acc5: 100.0000 (100.0000) time: 6.9096 data: 5.3195 max mem: 3531 +Test: [ 10/269] eta: 0:03:10 acc1_pos: 41.927 acc5_pos: 100.000 loss: 1.0185 (1.0326) acc1: 31.5104 (33.0256) acc5: 100.0000 (100.0000) time: 0.7355 data: 0.5070 max mem: 3531 +Test: [ 20/269] eta: 0:01:47 acc1_pos: 31.771 acc5_pos: 100.000 loss: 1.0185 (1.0253) acc1: 31.5104 (31.6716) acc5: 100.0000 (100.0000) time: 0.1068 data: 0.0132 max mem: 3531 +Test: [ 30/269] eta: 0:01:31 acc1_pos: 39.583 acc5_pos: 100.000 loss: 1.0318 (1.0228) acc1: 28.9062 (32.4765) acc5: 100.0000 (100.0000) time: 0.1882 data: 0.0957 max mem: 3531 +Test: [ 40/269] eta: 0:01:16 acc1_pos: 9.483 acc5_pos: 100.000 loss: 0.6035 (0.8704) acc1: 55.9896 (45.0838) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.2349 data: 0.1441 max mem: 3531 +Test: [ 50/269] eta: 0:01:06 acc1_pos: 9.483 acc5_pos: 100.000 loss: 0.3239 (0.7922) acc1: 91.1458 (52.0118) acc5: 100.0000 (100.0000) acc1_neg: 51.302 acc5_neg: 100.000 time: 0.1844 data: 0.0920 max mem: 3531 +Test: [ 60/269] eta: 0:00:59 acc1_pos: 9.483 acc5_pos: 100.000 loss: 0.4806 (0.7543) acc1: 78.3854 (55.3279) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.1849 data: 0.0896 max mem: 3531 +Test: [ 70/269] eta: 0:00:55 acc1_pos: 9.483 acc5_pos: 100.000 loss: 0.5216 (0.7197) acc1: 74.4792 (58.3040) acc5: 100.0000 (100.0000) acc1_neg: 74.219 acc5_neg: 100.000 time: 0.2049 data: 0.1065 max mem: 3531 +Test: [ 80/269] eta: 0:00:50 acc1_pos: 9.483 acc5_pos: 100.000 loss: 0.4851 (0.6876) acc1: 79.6875 (61.2590) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1999 data: 0.1032 max mem: 3531 +Test: [ 90/269] eta: 0:00:45 acc1_pos: 9.483 acc5_pos: 100.000 loss: 0.4851 (0.6682) acc1: 77.0833 (62.7719) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1815 data: 0.0857 max mem: 3531 +Test: [100/269] eta: 0:00:42 acc1_pos: 9.483 acc5_pos: 100.000 loss: 0.4974 (0.6502) acc1: 76.3021 (64.2456) acc5: 100.0000 (100.0000) acc1_neg: 74.219 acc5_neg: 100.000 time: 0.1849 data: 0.0879 max mem: 3531 +Test: [110/269] eta: 0:00:38 acc1_pos: 9.483 acc5_pos: 100.000 loss: 0.4478 (0.6275) acc1: 82.2917 (66.1904) acc5: 100.0000 (100.0000) acc1_neg: 82.812 acc5_neg: 100.000 time: 0.1683 data: 0.0735 max mem: 3531 +Test: [120/269] eta: 0:00:35 acc1_pos: 9.483 acc5_pos: 100.000 loss: 0.3511 (0.6037) acc1: 89.0625 (68.1732) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1899 data: 0.0966 max mem: 3531 +Test: [130/269] eta: 0:00:32 acc1_pos: 9.483 acc5_pos: 100.000 loss: 0.3545 (0.5902) acc1: 89.0625 (69.2768) acc5: 100.0000 (100.0000) acc1_neg: 80.729 acc5_neg: 100.000 time: 0.2149 data: 0.1186 max mem: 3531 +Test: [140/269] eta: 0:00:29 acc1_pos: 9.483 acc5_pos: 100.000 loss: 0.4158 (0.5800) acc1: 80.9896 (70.1629) acc5: 100.0000 (100.0000) acc1_neg: 74.740 acc5_neg: 100.000 time: 0.1849 data: 0.0872 max mem: 3531 +Test: [150/269] eta: 0:00:27 acc1_pos: 9.483 acc5_pos: 100.000 loss: 0.4158 (0.5680) acc1: 83.0729 (71.1334) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1849 data: 0.0871 max mem: 3531 +Test: [160/269] eta: 0:00:24 acc1_pos: 9.483 acc5_pos: 100.000 loss: 0.3222 (0.5526) acc1: 91.6667 (72.4055) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1799 data: 0.0823 max mem: 3531 +Test: [170/269] eta: 0:00:22 acc1_pos: 9.483 acc5_pos: 100.000 loss: 0.3511 (0.5435) acc1: 90.3646 (73.2822) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1799 data: 0.0851 max mem: 3531 +Test: [180/269] eta: 0:00:19 acc1_pos: 9.483 acc5_pos: 100.000 loss: 0.3480 (0.5300) acc1: 91.4062 (74.4403) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.1899 data: 0.0961 max mem: 3531 +Test: [190/269] eta: 0:00:17 acc1_pos: 9.483 acc5_pos: 100.000 loss: 0.3165 (0.5224) acc1: 91.9271 (75.1186) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1949 data: 0.1009 max mem: 3531 +Test: [200/269] eta: 0:00:15 acc1_pos: 9.483 acc5_pos: 100.000 loss: 0.3525 (0.5149) acc1: 88.2812 (75.7411) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1949 data: 0.1014 max mem: 3531 +Test: [210/269] eta: 0:00:12 acc1_pos: 9.483 acc5_pos: 100.000 loss: 0.3168 (0.5045) acc1: 90.1042 (76.5588) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1799 data: 0.0857 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 9.483 acc5_pos: 100.000 loss: 0.2733 (0.4955) acc1: 93.7500 (77.1670) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1849 data: 0.0885 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 9.483 acc5_pos: 100.000 loss: 0.2716 (0.4862) acc1: 93.2292 (77.9131) acc5: 100.0000 (100.0000) acc1_neg: 91.146 acc5_neg: 100.000 time: 0.2149 data: 0.1177 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 9.483 acc5_pos: 100.000 loss: 0.3569 (0.4838) acc1: 89.8438 (78.1466) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.1935 data: 0.0992 max mem: 3532 +Test: [250/269] eta: 0:00:03 acc1_pos: 9.483 acc5_pos: 100.000 loss: 0.4340 (0.4864) acc1: 80.4688 (77.9061) acc5: 100.0000 (100.0000) acc1_neg: 58.594 acc5_neg: 100.000 time: 0.1388 data: 0.0483 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 9.483 acc5_pos: 100.000 loss: 0.6299 (0.4929) acc1: 65.3646 (77.3398) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1319 data: 0.0425 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 9.483 acc5_pos: 100.000 loss: 0.5805 (0.4920) acc1: 67.1875 (77.3557) acc5: 100.0000 (100.0000) acc1_neg: 98.950 acc5_neg: 100.000 time: 0.1203 data: 0.0273 max mem: 3532 +Test: Total time: 0:00:55 (0.2050 s / it) +* Overall Acc@1 77.356 Acc@5 100.000 + target == real Acc@1 31.635 Acc@5 100.000 + target == fake Acc@1 83.595 Acc@5 100.000 + loss 0.492 + PR-AUC 0.9247 ROC-AUC 0.6462 EER 39.23% (thr=0.6733) +Accuracy of the network on the 103293 test images: 77.4% +[rank0]:[W725 18:15:08.281666106 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-25T18:15:15Z] eval_done tag=fixed500k_fake_n11_seed3_resampled +[2026-07-25T18:15:15Z] eval_start tag=fixed500k_real_n5_seed4_resampled checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed4_resampled_n116-040-168_20260725_112559/best_checkpoint.pth +started_utc=2026-07-25T18:15:16Z +hostname=n116-040-168 +package_root=/tmp/ScaleDF_repro_20260724 +data_root=/dev/shm/scaledf_bind_storage/ScaleDF +benchmark_root=/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names +checkpoint=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed4_resampled_n116-040-168_20260725_112559/best_checkpoint.pth +model_name=deit_base_patch16_LS +mode=benchmarks +scaledf_main=main_cross_2direct_balance_realscale_perturb_v4.py +benchmark_main=main_cross_2direct_balance_test.py +python_bin=system +output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed4_resampled +nproc_per_node=8 +batch_per_gpu=256 +input_size=224 +skip_missing_benchmarks=1 +W0725 18:15:28.624000 2571498 torch/distributed/run.py:766] +W0725 18:15:28.624000 2571498 torch/distributed/run.py:766] ***************************************** +W0725 18:15:28.624000 2571498 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:15:28.624000 2571498 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:15:56.681415263 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:15:56.922758545 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:15:56.149127638 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank4]:[W725 18:15:56.155463642 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:15:57.406922049 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +[rank7]:[W725 18:15:57.430315892 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:15:57.439754128 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:15:57.527681349 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ScaleDF_v3.1', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed4_resampled/benchmarks/ScaleDF_v3.1', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed4_resampled_n116-040-168_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/234] eta: 0:28:06 acc1_pos: 73.958 acc5_pos: 100.000 loss: 0.5492 (0.5492) acc1: 73.9583 (73.9583) acc5: 100.0000 (100.0000) time: 7.2056 data: 4.9466 max mem: 3531 +Test: [ 10/234] eta: 0:02:45 acc1_pos: 48.438 acc5_pos: 100.000 loss: 0.6776 (0.7694) acc1: 59.6354 (53.1250) acc5: 100.0000 (100.0000) time: 0.7406 data: 0.4502 max mem: 3531 +Test: [ 20/234] eta: 0:01:33 acc1_pos: 30.208 acc5_pos: 100.000 loss: 0.6209 (0.7386) acc1: 63.2812 (58.7178) acc5: 100.0000 (100.0000) time: 0.0970 data: 0.0005 max mem: 3531 +Test: [ 30/234] eta: 0:01:20 acc1_pos: 1.823 acc5_pos: 100.000 loss: 1.2662 (0.9536) acc1: 29.9479 (48.1015) acc5: 100.0000 (100.0000) time: 0.2052 data: 0.1098 max mem: 3531 +Test: [ 40/234] eta: 0:01:07 acc1_pos: 17.188 acc5_pos: 100.000 loss: 1.5664 (1.1800) acc1: 18.2292 (38.5861) acc5: 100.0000 (100.0000) time: 0.2599 data: 0.1653 max mem: 3531 +Test: [ 50/234] eta: 0:01:01 acc1_pos: 18.490 acc5_pos: 100.000 loss: 1.6199 (1.2609) acc1: 16.9271 (34.4159) acc5: 100.0000 (100.0000) time: 0.2353 data: 0.1414 max mem: 3531 +Test: [ 60/234] eta: 0:00:54 acc1_pos: 20.000 acc5_pos: 100.000 loss: 0.6287 (1.0788) acc1: 19.7917 (44.6721) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2453 data: 0.1540 max mem: 3531 +Test: [ 70/234] eta: 0:00:47 acc1_pos: 20.000 acc5_pos: 100.000 loss: 0.0967 (0.9454) acc1: 100.0000 (52.2887) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1843 data: 0.0923 max mem: 3531 +Test: [ 80/234] eta: 0:00:42 acc1_pos: 20.000 acc5_pos: 100.000 loss: 0.1418 (0.8462) acc1: 98.6979 (58.0247) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1498 data: 0.0573 max mem: 3531 +Test: [ 90/234] eta: 0:00:39 acc1_pos: 20.000 acc5_pos: 100.000 loss: 0.1313 (0.7660) acc1: 98.6979 (62.5114) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.2147 data: 0.1229 max mem: 3531 +Test: [100/234] eta: 0:00:36 acc1_pos: 20.000 acc5_pos: 100.000 loss: 0.1458 (0.7124) acc1: 97.3958 (65.5244) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2496 data: 0.1583 max mem: 3531 +Test: [110/234] eta: 0:00:31 acc1_pos: 20.000 acc5_pos: 100.000 loss: 0.1512 (0.6652) acc1: 96.0938 (68.1963) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1849 data: 0.0904 max mem: 3531 +Test: [120/234] eta: 0:00:28 acc1_pos: 20.000 acc5_pos: 100.000 loss: 0.1510 (0.6265) acc1: 97.3958 (70.5342) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1749 data: 0.0782 max mem: 3531 +Test: [130/234] eta: 0:00:25 acc1_pos: 20.000 acc5_pos: 100.000 loss: 0.2311 (0.6153) acc1: 95.0521 (70.8572) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.1999 data: 0.1027 max mem: 3531 +Test: [140/234] eta: 0:00:23 acc1_pos: 20.000 acc5_pos: 100.000 loss: 0.2876 (0.5887) acc1: 89.5833 (72.3884) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1899 data: 0.0924 max mem: 3531 +Test: [150/234] eta: 0:00:20 acc1_pos: 20.000 acc5_pos: 100.000 loss: 0.2257 (0.5645) acc1: 92.4479 (73.7548) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.1699 data: 0.0700 max mem: 3531 +Test: [160/234] eta: 0:00:17 acc1_pos: 20.000 acc5_pos: 100.000 loss: 0.2917 (0.5558) acc1: 87.5000 (74.1815) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.1615 data: 0.0649 max mem: 3531 +Test: [170/234] eta: 0:00:14 acc1_pos: 20.000 acc5_pos: 100.000 loss: 0.4404 (0.5504) acc1: 78.6458 (74.3908) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1756 data: 0.0830 max mem: 3531 +Test: [180/234] eta: 0:00:12 acc1_pos: 20.000 acc5_pos: 100.000 loss: 0.3929 (0.5323) acc1: 80.9896 (75.3885) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1883 data: 0.0920 max mem: 3531 +Test: [190/234] eta: 0:00:09 acc1_pos: 20.000 acc5_pos: 100.000 loss: 0.1961 (0.5152) acc1: 96.0938 (76.4275) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1759 data: 0.0796 max mem: 3531 +Test: [200/234] eta: 0:00:07 acc1_pos: 20.000 acc5_pos: 100.000 loss: 0.1996 (0.4968) acc1: 96.0938 (77.5070) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1549 data: 0.0597 max mem: 3531 +Test: [210/234] eta: 0:00:05 acc1_pos: 20.000 acc5_pos: 100.000 loss: 0.1037 (0.4781) acc1: 99.7396 (78.5631) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1379 data: 0.0453 max mem: 3532 +Test: [220/234] eta: 0:00:02 acc1_pos: 20.000 acc5_pos: 100.000 loss: 0.1036 (0.4615) acc1: 99.7396 (79.5013) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1272 data: 0.0377 max mem: 3532 +Test: [230/234] eta: 0:00:00 acc1_pos: 20.000 acc5_pos: 100.000 loss: 0.1702 (0.4559) acc1: 89.8438 (79.8498) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.1258 data: 0.0365 max mem: 3532 +Test: [233/234] eta: 0:00:00 acc1_pos: 20.000 acc5_pos: 100.000 loss: 0.3067 (0.4542) acc1: 89.5833 (79.9429) acc5: 100.0000 (100.0000) acc1_neg: 91.051 acc5_neg: 100.000 time: 0.1041 data: 0.0138 max mem: 3532 +Test: Total time: 0:00:48 (0.2087 s / it) +* Overall Acc@1 79.943 Acc@5 100.000 + target == real Acc@1 34.310 Acc@5 100.000 + target == fake Acc@1 92.804 Acc@5 100.000 + loss 0.454 + PR-AUC 0.8994 ROC-AUC 0.7626 EER 29.27% (thr=0.8452) +Accuracy of the network on the 89729 test images: 79.9% +[rank0]:[W725 18:16:53.329997153 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:17:12.714000 2572080 torch/distributed/run.py:766] +W0725 18:17:12.714000 2572080 torch/distributed/run.py:766] ***************************************** +W0725 18:17:12.714000 2572080 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:17:12.714000 2572080 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:17:40.856437861 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:17:40.077428960 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:17:40.133261176 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:17:41.288944873 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:17:41.315731707 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:17:41.510852928 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:17:41.529525942 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank7]:[W725 18:17:41.530895081 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DF40_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed4_resampled/benchmarks/DF40_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed4_resampled_n116-040-168_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/2387] eta: 4:06:48 acc1_pos: 87.500 acc5_pos: 100.000 loss: 0.3767 (0.3767) acc1: 87.5000 (87.5000) acc5: 100.0000 (100.0000) time: 6.2036 data: 4.4946 max mem: 3531 +Test: [ 10/2387] eta: 0:25:38 acc1_pos: 48.958 acc5_pos: 100.000 loss: 0.6881 (0.7098) acc1: 61.4583 (61.2689) acc5: 100.0000 (100.0000) time: 0.6471 data: 0.4090 max mem: 3531 +Test: [ 20/2387] eta: 0:15:10 acc1_pos: 50.260 acc5_pos: 100.000 loss: 0.6881 (0.7585) acc1: 58.5938 (57.8497) acc5: 100.0000 (100.0000) time: 0.0936 data: 0.0004 max mem: 3531 +Test: [ 30/2387] eta: 0:12:08 acc1_pos: 46.354 acc5_pos: 100.000 loss: 0.7406 (0.7560) acc1: 51.5625 (57.4177) acc5: 100.0000 (100.0000) time: 0.1229 data: 0.0282 max mem: 3531 +Test: [ 40/2387] eta: 0:11:19 acc1_pos: 61.719 acc5_pos: 100.000 loss: 0.7699 (0.7538) acc1: 51.3021 (57.3552) acc5: 100.0000 (100.0000) time: 0.1899 data: 0.0971 max mem: 3531 +Test: [ 50/2387] eta: 0:10:54 acc1_pos: 47.656 acc5_pos: 100.000 loss: 0.6761 (0.7225) acc1: 63.8021 (60.0746) acc5: 100.0000 (100.0000) time: 0.2348 data: 0.1394 max mem: 3531 +Test: [ 60/2387] eta: 0:10:43 acc1_pos: 39.062 acc5_pos: 100.000 loss: 0.6822 (0.7437) acc1: 61.7188 (58.7474) acc5: 100.0000 (100.0000) time: 0.2499 data: 0.1519 max mem: 3531 +Test: [ 70/2387] eta: 0:10:28 acc1_pos: 29.948 acc5_pos: 100.000 loss: 0.8150 (0.7532) acc1: 50.5208 (58.3297) acc5: 100.0000 (100.0000) time: 0.2499 data: 0.1551 max mem: 3531 +Test: [ 80/2387] eta: 0:10:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5655 (0.7004) acc1: 66.4062 (62.1560) acc5: 100.0000 (100.0000) time: 0.2555 data: 0.1642 max mem: 3531 +Test: [ 90/2387] eta: 0:10:22 acc1_pos: 95.833 acc5_pos: 100.000 loss: 0.1927 (0.6441) acc1: 98.6979 (66.1630) acc5: 100.0000 (100.0000) time: 0.2699 data: 0.1749 max mem: 3531 +Test: [ 100/2387] eta: 0:10:13 acc1_pos: 99.219 acc5_pos: 100.000 loss: 0.1928 (0.6013) acc1: 99.2188 (69.3224) acc5: 100.0000 (100.0000) time: 0.2553 data: 0.1595 max mem: 3531 +Test: [ 110/2387] eta: 0:10:12 acc1_pos: 98.438 acc5_pos: 100.000 loss: 0.2053 (0.5673) acc1: 98.4375 (71.7788) acc5: 100.0000 (100.0000) time: 0.2599 data: 0.1651 max mem: 3531 +Test: [ 120/2387] eta: 0:10:08 acc1_pos: 90.365 acc5_pos: 100.000 loss: 0.2030 (0.5375) acc1: 98.4375 (73.9433) acc5: 100.0000 (100.0000) time: 0.2689 data: 0.1739 max mem: 3531 +Test: [ 130/2387] eta: 0:10:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1982 (0.5119) acc1: 99.7396 (75.8568) acc5: 100.0000 (100.0000) time: 0.2499 data: 0.1550 max mem: 3531 +Test: [ 140/2387] eta: 0:09:55 acc1_pos: 97.656 acc5_pos: 100.000 loss: 0.1862 (0.4883) acc1: 99.7396 (77.4860) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1502 max mem: 3531 +Test: [ 150/2387] eta: 0:09:49 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1827 (0.4688) acc1: 99.7396 (78.8942) acc5: 100.0000 (100.0000) time: 0.2449 data: 0.1524 max mem: 3531 +Test: [ 160/2387] eta: 0:09:33 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1798 (0.4504) acc1: 99.7396 (80.0773) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.2049 data: 0.1104 max mem: 3531 +Test: [ 170/2387] eta: 0:09:19 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2339 (0.4385) acc1: 93.4896 (80.7840) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1699 data: 0.0745 max mem: 3531 +Test: [ 180/2387] eta: 0:09:10 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2468 (0.4284) acc1: 92.4479 (81.3939) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1851 data: 0.0919 max mem: 3531 +Test: [ 190/2387] eta: 0:09:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2466 (0.4188) acc1: 91.9271 (81.9413) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.2063 data: 0.1154 max mem: 3531 +Test: [ 200/2387] eta: 0:08:55 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2433 (0.4104) acc1: 91.9271 (82.4536) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.2003 data: 0.1088 max mem: 3531 +Test: [ 210/2387] eta: 0:08:46 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2467 (0.4031) acc1: 91.9271 (82.9088) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1888 data: 0.0959 max mem: 3532 +Test: [ 220/2387] eta: 0:08:44 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2469 (0.3963) acc1: 91.9271 (83.3109) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.2143 data: 0.1206 max mem: 3532 +Test: [ 230/2387] eta: 0:08:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2469 (0.3898) acc1: 91.9271 (83.7110) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.2146 data: 0.1218 max mem: 3532 +Test: [ 240/2387] eta: 0:08:31 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2588 (0.4014) acc1: 90.8854 (82.9033) acc5: 100.0000 (100.0000) acc1_neg: 56.250 acc5_neg: 100.000 time: 0.1999 data: 0.1047 max mem: 3532 +Test: [ 250/2387] eta: 0:08:23 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2588 (0.3947) acc1: 67.4479 (83.2804) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1900 data: 0.0922 max mem: 3532 +Test: [ 260/2387] eta: 0:08:12 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1565 (0.3853) acc1: 96.8750 (83.8372) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1499 data: 0.0531 max mem: 3532 +Test: [ 270/2387] eta: 0:08:08 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1401 (0.3761) acc1: 98.4375 (84.3731) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1699 data: 0.0757 max mem: 3532 +Test: [ 280/2387] eta: 0:08:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1387 (0.3682) acc1: 98.1771 (84.8439) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2099 data: 0.1155 max mem: 3532 +Test: [ 290/2387] eta: 0:08:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1524 (0.3611) acc1: 97.6562 (85.2529) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2049 data: 0.1109 max mem: 3532 +Test: [ 300/2387] eta: 0:07:56 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1685 (0.3571) acc1: 95.3125 (85.4712) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.2008 data: 0.1099 max mem: 3532 +Test: [ 310/2387] eta: 0:07:53 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2197 (0.3525) acc1: 92.9688 (85.7399) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2114 data: 0.1206 max mem: 3532 +Test: [ 320/2387] eta: 0:07:52 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1350 (0.3457) acc1: 99.4792 (86.1703) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2358 data: 0.1454 max mem: 3532 +Test: [ 330/2387] eta: 0:07:48 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1348 (0.3393) acc1: 99.7396 (86.5763) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2256 data: 0.1352 max mem: 3532 +Test: [ 340/2387] eta: 0:07:46 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1347 (0.3333) acc1: 99.4792 (86.9570) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2138 data: 0.1193 max mem: 3532 +Test: [ 350/2387] eta: 0:07:42 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1323 (0.3275) acc1: 99.7396 (87.3249) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2148 data: 0.1199 max mem: 3532 +Test: [ 360/2387] eta: 0:07:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1358 (0.3244) acc1: 99.2188 (87.5209) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1997 data: 0.1087 max mem: 3532 +Test: [ 370/2387] eta: 0:07:35 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2203 (0.3218) acc1: 93.7500 (87.6797) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.2046 data: 0.1124 max mem: 3532 +Test: [ 380/2387] eta: 0:07:31 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2214 (0.3192) acc1: 93.7500 (87.8506) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2058 data: 0.1107 max mem: 3532 +Test: [ 390/2387] eta: 0:07:28 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2284 (0.3168) acc1: 93.7500 (88.0055) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1997 data: 0.1066 max mem: 3532 +Test: [ 400/2387] eta: 0:07:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2236 (0.3144) acc1: 93.7500 (88.1585) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2209 data: 0.1296 max mem: 3532 +Test: [ 410/2387] eta: 0:07:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2199 (0.3121) acc1: 94.0104 (88.3047) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.2134 data: 0.1209 max mem: 3532 +Test: [ 420/2387] eta: 0:07:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2237 (0.3100) acc1: 93.4896 (88.4260) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1994 data: 0.1075 max mem: 3532 +Test: [ 430/2387] eta: 0:07:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2328 (0.3135) acc1: 92.4479 (88.1701) acc5: 100.0000 (100.0000) acc1_neg: 64.062 acc5_neg: 100.000 time: 0.2049 data: 0.1141 max mem: 3532 +Test: [ 440/2387] eta: 0:07:13 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2381 (0.3136) acc1: 73.1771 (88.1519) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1984 data: 0.1073 max mem: 3533 +Test: [ 450/2387] eta: 0:07:12 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1223 (0.3094) acc1: 99.7396 (88.4094) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2274 data: 0.1344 max mem: 3533 +Test: [ 460/2387] eta: 0:07:08 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1216 (0.3053) acc1: 99.7396 (88.6552) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.2197 data: 0.1273 max mem: 3533 +Test: [ 470/2387] eta: 0:07:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1218 (0.3014) acc1: 99.7396 (88.8917) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1926 data: 0.1012 max mem: 3533 +Test: [ 480/2387] eta: 0:07:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1215 (0.2977) acc1: 99.7396 (89.1183) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2029 data: 0.1109 max mem: 3533 +Test: [ 490/2387] eta: 0:06:59 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1228 (0.2942) acc1: 99.7396 (89.3298) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1965 data: 0.1047 max mem: 3533 +Test: [ 500/2387] eta: 0:06:56 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1233 (0.2908) acc1: 99.7396 (89.5360) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1925 data: 0.0989 max mem: 3533 +Test: [ 510/2387] eta: 0:06:52 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1240 (0.2891) acc1: 99.7396 (89.6572) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.1981 data: 0.1032 max mem: 3533 +Test: [ 520/2387] eta: 0:06:48 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2098 (0.2903) acc1: 87.2396 (89.6163) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1789 data: 0.0858 max mem: 3533 +Test: [ 530/2387] eta: 0:06:45 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2098 (0.2886) acc1: 95.0521 (89.7349) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1799 data: 0.0873 max mem: 3533 +Test: [ 540/2387] eta: 0:06:42 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2053 (0.2871) acc1: 95.5729 (89.8534) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1999 data: 0.1048 max mem: 3533 +Test: [ 550/2387] eta: 0:06:39 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2049 (0.2856) acc1: 95.8333 (89.9652) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1952 data: 0.0978 max mem: 3533 +Test: [ 560/2387] eta: 0:06:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1985 (0.2841) acc1: 96.3542 (90.0810) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1859 data: 0.0918 max mem: 3533 +Test: [ 570/2387] eta: 0:06:35 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1972 (0.2826) acc1: 96.3542 (90.1917) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2101 data: 0.1154 max mem: 3533 +Test: [ 580/2387] eta: 0:06:31 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2012 (0.2813) acc1: 96.3542 (90.2920) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.2044 data: 0.1086 max mem: 3533 +Test: [ 590/2387] eta: 0:06:29 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2088 (0.2801) acc1: 96.0938 (90.3827) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1994 data: 0.1075 max mem: 3533 +Test: [ 600/2387] eta: 0:06:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2386 (0.2822) acc1: 87.5000 (90.2866) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.2152 data: 0.1226 max mem: 3533 +Test: [ 610/2387] eta: 0:06:23 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2055 (0.2807) acc1: 87.5000 (90.4081) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1798 data: 0.0874 max mem: 3533 +Test: [ 620/2387] eta: 0:06:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1885 (0.2793) acc1: 97.3958 (90.5168) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1739 data: 0.0790 max mem: 3533 +Test: [ 630/2387] eta: 0:06:17 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1898 (0.2778) acc1: 97.1354 (90.6279) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1974 data: 0.1005 max mem: 3533 +Test: [ 640/2387] eta: 0:06:15 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1891 (0.2765) acc1: 97.1354 (90.7282) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.2026 data: 0.1087 max mem: 3533 +Test: [ 650/2387] eta: 0:06:12 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1891 (0.2752) acc1: 97.1354 (90.8334) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2046 data: 0.1110 max mem: 3533 +Test: [ 660/2387] eta: 0:06:10 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1878 (0.2739) acc1: 97.3958 (90.9347) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2083 data: 0.1154 max mem: 3533 +Test: [ 670/2387] eta: 0:06:08 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1881 (0.2726) acc1: 97.3958 (91.0337) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.2037 data: 0.1113 max mem: 3534 +Test: [ 680/2387] eta: 0:06:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1946 (0.2742) acc1: 96.8750 (90.9347) acc5: 100.0000 (100.0000) acc1_neg: 75.781 acc5_neg: 100.000 time: 0.2015 data: 0.1087 max mem: 3534 +Test: [ 690/2387] eta: 0:06:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1946 (0.2739) acc1: 83.3333 (90.9374) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2065 data: 0.1141 max mem: 3534 +Test: [ 700/2387] eta: 0:06:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1739 (0.2724) acc1: 96.3542 (91.0232) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1938 data: 0.1025 max mem: 3534 +Test: [ 710/2387] eta: 0:05:58 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1728 (0.2711) acc1: 96.8750 (91.0993) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1922 data: 0.1016 max mem: 3534 +Test: [ 720/2387] eta: 0:05:55 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1712 (0.2697) acc1: 96.6146 (91.1809) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2046 data: 0.1141 max mem: 3534 +Test: [ 730/2387] eta: 0:05:53 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1704 (0.2683) acc1: 96.8750 (91.2588) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1971 data: 0.1065 max mem: 3534 +Test: [ 740/2387] eta: 0:05:50 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1704 (0.2670) acc1: 96.8750 (91.3367) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1986 data: 0.1049 max mem: 3534 +Test: [ 750/2387] eta: 0:05:48 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1728 (0.2658) acc1: 97.1354 (91.4132) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.2028 data: 0.1087 max mem: 3534 +Test: [ 760/2387] eta: 0:05:46 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1777 (0.2659) acc1: 96.6146 (91.3881) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.2059 data: 0.1146 max mem: 3534 +Test: [ 770/2387] eta: 0:05:43 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3538 (0.2673) acc1: 83.5938 (91.2877) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.2052 data: 0.1141 max mem: 3534 +Test: [ 780/2387] eta: 0:05:42 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3106 (0.2671) acc1: 87.7604 (91.3012) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.2252 data: 0.1312 max mem: 3534 +Test: [ 790/2387] eta: 0:05:40 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2393 (0.2669) acc1: 92.7083 (91.3002) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.2284 data: 0.1325 max mem: 3534 +Test: [ 800/2387] eta: 0:05:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2299 (0.2664) acc1: 92.7083 (91.3302) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.1786 data: 0.0843 max mem: 3534 +Test: [ 810/2387] eta: 0:05:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2641 (0.2667) acc1: 92.4479 (91.3070) acc5: 100.0000 (100.0000) acc1_neg: 85.417 acc5_neg: 100.000 time: 0.1790 data: 0.0869 max mem: 3534 +Test: [ 820/2387] eta: 0:05:31 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2707 (0.2677) acc1: 89.0625 (91.2385) acc5: 100.0000 (100.0000) acc1_neg: 46.094 acc5_neg: 100.000 time: 0.1955 data: 0.1024 max mem: 3534 +Test: [ 830/2387] eta: 0:05:29 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.7489 (0.2754) acc1: 55.7292 (90.6554) acc5: 100.0000 (100.0000) acc1_neg: 47.656 acc5_neg: 100.000 time: 0.1967 data: 0.1025 max mem: 3534 +Test: [ 840/2387] eta: 0:05:27 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4637 (0.2756) acc1: 57.5521 (90.6575) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1992 data: 0.1048 max mem: 3534 +Test: [ 850/2387] eta: 0:05:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2330 (0.2750) acc1: 94.0104 (90.7033) acc5: 100.0000 (100.0000) acc1_neg: 87.760 acc5_neg: 100.000 time: 0.2023 data: 0.1075 max mem: 3534 +Test: [ 860/2387] eta: 0:05:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2172 (0.2743) acc1: 94.5312 (90.7514) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2107 data: 0.1165 max mem: 3534 +Test: [ 870/2387] eta: 0:05:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2303 (0.2739) acc1: 94.5312 (90.7933) acc5: 100.0000 (100.0000) acc1_neg: 92.188 acc5_neg: 100.000 time: 0.2125 data: 0.1189 max mem: 3534 +Test: [ 880/2387] eta: 0:05:18 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2307 (0.2733) acc1: 95.3125 (90.8426) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2104 data: 0.1173 max mem: 3534 +Test: [ 890/2387] eta: 0:05:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2917 (0.2747) acc1: 91.4062 (90.7822) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.2060 data: 0.1140 max mem: 3534 +Test: [ 900/2387] eta: 0:05:13 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3009 (0.2741) acc1: 91.1458 (90.8218) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1948 data: 0.1038 max mem: 3535 +Test: [ 910/2387] eta: 0:05:11 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1662 (0.2729) acc1: 97.3958 (90.8994) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1797 data: 0.0890 max mem: 3535 +Test: [ 920/2387] eta: 0:05:08 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1613 (0.2717) acc1: 97.6562 (90.9756) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1890 data: 0.0982 max mem: 3535 +Test: [ 930/2387] eta: 0:05:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1636 (0.2706) acc1: 97.6562 (91.0460) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1890 data: 0.0968 max mem: 3535 +Test: [ 940/2387] eta: 0:05:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1649 (0.2695) acc1: 97.3958 (91.1154) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1804 data: 0.0881 max mem: 3535 +Test: [ 950/2387] eta: 0:05:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1660 (0.2684) acc1: 97.3958 (91.1831) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1807 data: 0.0888 max mem: 3535 +Test: [ 960/2387] eta: 0:04:58 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1598 (0.2672) acc1: 97.9167 (91.2575) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1800 data: 0.0872 max mem: 3535 +Test: [ 970/2387] eta: 0:04:56 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1602 (0.2662) acc1: 98.1771 (91.3277) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1791 data: 0.0865 max mem: 3535 +Test: [ 980/2387] eta: 0:04:54 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1668 (0.2667) acc1: 97.3958 (91.2876) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.1995 data: 0.1073 max mem: 3535 +Test: [ 990/2387] eta: 0:04:51 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3475 (0.2682) acc1: 84.1146 (91.1645) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1930 data: 0.0990 max mem: 3535 +Test: [1000/2387] eta: 0:04:49 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3475 (0.2691) acc1: 83.8542 (91.0982) acc5: 100.0000 (100.0000) acc1_neg: 72.135 acc5_neg: 100.000 time: 0.1814 data: 0.0886 max mem: 3535 +Test: [1010/2387] eta: 0:04:46 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4391 (0.2737) acc1: 77.6042 (90.7721) acc5: 100.0000 (100.0000) acc1_neg: 41.406 acc5_neg: 100.000 time: 0.1768 data: 0.0842 max mem: 3535 +Test: [1020/2387] eta: 0:04:44 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5595 (0.2784) acc1: 53.1250 (90.4378) acc5: 100.0000 (100.0000) acc1_neg: 84.375 acc5_neg: 100.000 time: 0.1751 data: 0.0807 max mem: 3535 +Test: [1030/2387] eta: 0:04:41 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3779 (0.2790) acc1: 81.5104 (90.3921) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.1849 data: 0.0910 max mem: 3535 +Test: [1040/2387] eta: 0:04:39 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2857 (0.2788) acc1: 89.0625 (90.4031) acc5: 100.0000 (100.0000) acc1_neg: 92.708 acc5_neg: 100.000 time: 0.1776 data: 0.0849 max mem: 3535 +Test: [1050/2387] eta: 0:04:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2574 (0.2789) acc1: 90.6250 (90.3995) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.2057 data: 0.1148 max mem: 3535 +Test: [1060/2387] eta: 0:04:35 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3258 (0.2796) acc1: 87.7604 (90.3560) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.2049 data: 0.1128 max mem: 3535 +Test: [1070/2387] eta: 0:04:33 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3645 (0.2820) acc1: 83.8542 (90.1830) acc5: 100.0000 (100.0000) acc1_neg: 28.906 acc5_neg: 100.000 time: 0.1968 data: 0.1016 max mem: 3535 +Test: [1080/2387] eta: 0:04:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.8773 (0.2889) acc1: 40.8854 (89.6848) acc5: 100.0000 (100.0000) acc1_neg: 57.292 acc5_neg: 100.000 time: 0.1766 data: 0.0830 max mem: 3535 +Test: [1090/2387] eta: 0:04:27 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5030 (0.2897) acc1: 57.2917 (89.6201) acc5: 100.0000 (100.0000) acc1_neg: 73.958 acc5_neg: 100.000 time: 0.1498 data: 0.0586 max mem: 3535 +Test: [1100/2387] eta: 0:04:25 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3438 (0.2908) acc1: 83.8542 (89.5356) acc5: 100.0000 (100.0000) acc1_neg: 91.927 acc5_neg: 100.000 time: 0.1737 data: 0.0816 max mem: 3535 +Test: [1110/2387] eta: 0:04:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3716 (0.2914) acc1: 82.0312 (89.4907) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1827 data: 0.0912 max mem: 3535 +Test: [1120/2387] eta: 0:04:21 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3709 (0.2922) acc1: 84.6354 (89.4349) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.2001 data: 0.1066 max mem: 3536 +Test: [1130/2387] eta: 0:04:18 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3832 (0.2931) acc1: 79.9479 (89.3591) acc5: 100.0000 (100.0000) acc1_neg: 78.385 acc5_neg: 100.000 time: 0.2094 data: 0.1130 max mem: 3536 +Test: [1140/2387] eta: 0:04:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5230 (0.2987) acc1: 66.1458 (88.9472) acc5: 100.0000 (100.0000) acc1_neg: 28.906 acc5_neg: 100.000 time: 0.2007 data: 0.1027 max mem: 3536 +Test: [1150/2387] eta: 0:04:14 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.8446 (0.3004) acc1: 39.3229 (88.8331) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1948 data: 0.0975 max mem: 3536 +Test: [1160/2387] eta: 0:04:12 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1433 (0.2989) acc1: 99.2188 (88.9257) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1883 data: 0.0927 max mem: 3536 +Test: [1170/2387] eta: 0:04:10 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1406 (0.2976) acc1: 99.4792 (89.0082) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1908 data: 0.0949 max mem: 3536 +Test: [1180/2387] eta: 0:04:08 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1546 (0.2966) acc1: 98.9583 (89.0680) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1957 data: 0.1024 max mem: 3536 +Test: [1190/2387] eta: 0:04:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1526 (0.2956) acc1: 97.9167 (89.1320) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1898 data: 0.0981 max mem: 3536 +Test: [1200/2387] eta: 0:04:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1522 (0.2946) acc1: 98.4375 (89.1980) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1847 data: 0.0883 max mem: 3536 +Test: [1210/2387] eta: 0:04:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1795 (0.2939) acc1: 98.4375 (89.2537) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1963 data: 0.1008 max mem: 3536 +Test: [1220/2387] eta: 0:03:59 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1660 (0.2929) acc1: 97.1354 (89.3084) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1981 data: 0.1068 max mem: 3536 +Test: [1230/2387] eta: 0:03:57 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1465 (0.2916) acc1: 97.1354 (89.3747) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.2095 data: 0.1178 max mem: 3536 +Test: [1240/2387] eta: 0:03:54 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1365 (0.2903) acc1: 97.3958 (89.4438) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1867 data: 0.0959 max mem: 3536 +Test: [1250/2387] eta: 0:03:52 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1419 (0.2892) acc1: 97.1354 (89.5061) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1634 data: 0.0731 max mem: 3536 +Test: [1260/2387] eta: 0:03:50 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1551 (0.2882) acc1: 96.6146 (89.5604) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1809 data: 0.0879 max mem: 3536 +Test: [1270/2387] eta: 0:03:48 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1818 (0.2883) acc1: 94.2708 (89.5440) acc5: 100.0000 (100.0000) acc1_neg: 83.333 acc5_neg: 100.000 time: 0.1865 data: 0.0926 max mem: 3536 +Test: [1280/2387] eta: 0:03:45 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2725 (0.2881) acc1: 90.3646 (89.5671) acc5: 100.0000 (100.0000) acc1_neg: 88.802 acc5_neg: 100.000 time: 0.1727 data: 0.0786 max mem: 3536 +Test: [1290/2387] eta: 0:03:43 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2470 (0.2877) acc1: 94.0104 (89.5948) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1693 data: 0.0752 max mem: 3536 +Test: [1300/2387] eta: 0:03:41 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2612 (0.2875) acc1: 91.6667 (89.6088) acc5: 100.0000 (100.0000) acc1_neg: 90.104 acc5_neg: 100.000 time: 0.1758 data: 0.0845 max mem: 3536 +Test: [1310/2387] eta: 0:03:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2623 (0.2875) acc1: 90.1042 (89.6056) acc5: 100.0000 (100.0000) acc1_neg: 88.542 acc5_neg: 100.000 time: 0.1635 data: 0.0727 max mem: 3536 +Test: [1320/2387] eta: 0:03:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2663 (0.2873) acc1: 91.4062 (89.6279) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1592 data: 0.0666 max mem: 3536 +Test: [1330/2387] eta: 0:03:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3129 (0.2897) acc1: 87.2396 (89.4407) acc5: 100.0000 (100.0000) acc1_neg: 52.083 acc5_neg: 100.000 time: 0.1641 data: 0.0692 max mem: 3536 +Test: [1340/2387] eta: 0:03:31 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5853 (0.2912) acc1: 65.6250 (89.3412) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1706 data: 0.0771 max mem: 3536 +Test: [1350/2387] eta: 0:03:29 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1090 (0.2897) acc1: 99.2188 (89.4191) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0786 max mem: 3537 +Test: [1360/2387] eta: 0:03:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0895 (0.2882) acc1: 100.0000 (89.4967) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1534 data: 0.0593 max mem: 3537 +Test: [1370/2387] eta: 0:03:24 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0842 (0.2868) acc1: 100.0000 (89.5700) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1499 data: 0.0517 max mem: 3537 +Test: [1380/2387] eta: 0:03:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0907 (0.2854) acc1: 100.0000 (89.6450) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1699 data: 0.0733 max mem: 3537 +Test: [1390/2387] eta: 0:03:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0896 (0.2840) acc1: 100.0000 (89.7193) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1749 data: 0.0813 max mem: 3537 +Test: [1400/2387] eta: 0:03:17 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.0971 (0.2831) acc1: 99.7396 (89.7727) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1599 data: 0.0623 max mem: 3537 +Test: [1410/2387] eta: 0:03:15 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2114 (0.2833) acc1: 94.5312 (89.7609) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1599 data: 0.0608 max mem: 3537 +Test: [1420/2387] eta: 0:03:13 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3109 (0.2838) acc1: 88.5417 (89.7287) acc5: 100.0000 (100.0000) acc1_neg: 89.844 acc5_neg: 100.000 time: 0.1649 data: 0.0709 max mem: 3537 +Test: [1430/2387] eta: 0:03:11 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3233 (0.2843) acc1: 86.4583 (89.6940) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.1649 data: 0.0699 max mem: 3537 +Test: [1440/2387] eta: 0:03:08 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3655 (0.2856) acc1: 85.1562 (89.6003) acc5: 100.0000 (100.0000) acc1_neg: 39.323 acc5_neg: 100.000 time: 0.1649 data: 0.0690 max mem: 3537 +Test: [1450/2387] eta: 0:03:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5033 (0.2887) acc1: 58.0729 (89.3640) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1755 data: 0.0821 max mem: 3537 +Test: [1460/2387] eta: 0:03:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2413 (0.2882) acc1: 92.4479 (89.4015) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1904 data: 0.0983 max mem: 3537 +Test: [1470/2387] eta: 0:03:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1783 (0.2873) acc1: 96.8750 (89.4615) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.1893 data: 0.0931 max mem: 3537 +Test: [1480/2387] eta: 0:03:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1509 (0.2866) acc1: 97.9167 (89.5125) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1645 data: 0.0674 max mem: 3537 +Test: [1490/2387] eta: 0:02:58 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1642 (0.2859) acc1: 97.3958 (89.5605) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1649 data: 0.0689 max mem: 3537 +Test: [1500/2387] eta: 0:02:56 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1845 (0.2852) acc1: 96.6146 (89.6055) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1949 data: 0.0982 max mem: 3537 +Test: [1510/2387] eta: 0:02:54 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2386 (0.2855) acc1: 94.5312 (89.6123) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1799 data: 0.0848 max mem: 3537 +Test: [1520/2387] eta: 0:02:51 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2867 (0.2854) acc1: 91.4062 (89.6205) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1618 data: 0.0666 max mem: 3537 +Test: [1530/2387] eta: 0:02:49 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2290 (0.2850) acc1: 94.0104 (89.6541) acc5: 100.0000 (100.0000) acc1_neg: 94.792 acc5_neg: 100.000 time: 0.1749 data: 0.0776 max mem: 3537 +Test: [1540/2387] eta: 0:02:47 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2133 (0.2847) acc1: 94.0104 (89.6778) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1760 data: 0.0779 max mem: 3537 +Test: [1550/2387] eta: 0:02:45 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2435 (0.2844) acc1: 92.9688 (89.7036) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1750 data: 0.0814 max mem: 3537 +Test: [1560/2387] eta: 0:02:43 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2368 (0.2840) acc1: 94.2708 (89.7345) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1883 data: 0.0969 max mem: 3537 +Test: [1570/2387] eta: 0:02:41 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2570 (0.2843) acc1: 92.4479 (89.7360) acc5: 100.0000 (100.0000) acc1_neg: 85.677 acc5_neg: 100.000 time: 0.1802 data: 0.0877 max mem: 3537 +Test: [1580/2387] eta: 0:02:39 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3120 (0.2844) acc1: 87.7604 (89.7313) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1547 data: 0.0612 max mem: 3538 +Test: [1590/2387] eta: 0:02:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2975 (0.2845) acc1: 87.7604 (89.7256) acc5: 100.0000 (100.0000) acc1_neg: 89.062 acc5_neg: 100.000 time: 0.1468 data: 0.0538 max mem: 3538 +Test: [1600/2387] eta: 0:02:35 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2492 (0.2842) acc1: 90.8854 (89.7471) acc5: 100.0000 (100.0000) acc1_neg: 91.406 acc5_neg: 100.000 time: 0.1612 data: 0.0680 max mem: 3538 +Test: [1610/2387] eta: 0:02:32 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2450 (0.2840) acc1: 93.4896 (89.7595) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1629 data: 0.0702 max mem: 3538 +Test: [1620/2387] eta: 0:02:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2723 (0.2841) acc1: 90.3646 (89.7501) acc5: 100.0000 (100.0000) acc1_neg: 93.229 acc5_neg: 100.000 time: 0.1586 data: 0.0633 max mem: 3538 +Test: [1630/2387] eta: 0:02:28 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3297 (0.2857) acc1: 85.6771 (89.6298) acc5: 100.0000 (100.0000) acc1_neg: 55.208 acc5_neg: 100.000 time: 0.1713 data: 0.0754 max mem: 3538 +Test: [1640/2387] eta: 0:02:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.7839 (0.2894) acc1: 49.2188 (89.3485) acc5: 100.0000 (100.0000) acc1_neg: 41.406 acc5_neg: 100.000 time: 0.1757 data: 0.0840 max mem: 3538 +Test: [1650/2387] eta: 0:02:24 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.8422 (0.2926) acc1: 45.8333 (89.1057) acc5: 100.0000 (100.0000) acc1_neg: 45.573 acc5_neg: 100.000 time: 0.1532 data: 0.0618 max mem: 3538 +Test: [1660/2387] eta: 0:02:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.7825 (0.2956) acc1: 49.4792 (88.8704) acc5: 100.0000 (100.0000) acc1_neg: 60.938 acc5_neg: 100.000 time: 0.1488 data: 0.0558 max mem: 3538 +Test: [1670/2387] eta: 0:02:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.7798 (0.2986) acc1: 48.1771 (88.6377) acc5: 100.0000 (100.0000) acc1_neg: 42.969 acc5_neg: 100.000 time: 0.1699 data: 0.0753 max mem: 3538 +Test: [1680/2387] eta: 0:02:18 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.8046 (0.3017) acc1: 47.9167 (88.3967) acc5: 100.0000 (100.0000) acc1_neg: 38.802 acc5_neg: 100.000 time: 0.1789 data: 0.0857 max mem: 3538 +Test: [1690/2387] eta: 0:02:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.7915 (0.3048) acc1: 47.3958 (88.1587) acc5: 100.0000 (100.0000) acc1_neg: 55.208 acc5_neg: 100.000 time: 0.1926 data: 0.1004 max mem: 3538 +Test: [1700/2387] eta: 0:02:14 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.7820 (0.3070) acc1: 48.6979 (87.9849) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1826 data: 0.0902 max mem: 3538 +Test: [1710/2387] eta: 0:02:12 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1841 (0.3061) acc1: 94.0104 (88.0288) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1685 data: 0.0769 max mem: 3538 +Test: [1720/2387] eta: 0:02:10 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1520 (0.3052) acc1: 96.6146 (88.0835) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1721 data: 0.0799 max mem: 3538 +Test: [1730/2387] eta: 0:02:08 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1355 (0.3042) acc1: 97.3958 (88.1343) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1716 data: 0.0786 max mem: 3538 +Test: [1740/2387] eta: 0:02:06 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1464 (0.3034) acc1: 96.6146 (88.1786) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1772 data: 0.0854 max mem: 3538 +Test: [1750/2387] eta: 0:02:04 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1576 (0.3026) acc1: 95.8333 (88.2225) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.1834 data: 0.0925 max mem: 3538 +Test: [1760/2387] eta: 0:02:02 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2116 (0.3031) acc1: 89.5833 (88.1823) acc5: 100.0000 (100.0000) acc1_neg: 78.125 acc5_neg: 100.000 time: 0.1980 data: 0.1032 max mem: 3538 +Test: [1770/2387] eta: 0:02:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3102 (0.3028) acc1: 84.6354 (88.1901) acc5: 100.0000 (100.0000) acc1_neg: 86.719 acc5_neg: 100.000 time: 0.1932 data: 0.0972 max mem: 3538 +Test: [1780/2387] eta: 0:01:58 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2048 (0.3022) acc1: 94.0104 (88.2197) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1844 data: 0.0912 max mem: 3538 +Test: [1790/2387] eta: 0:01:56 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1658 (0.3014) acc1: 96.0938 (88.2645) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1964 data: 0.1034 max mem: 3538 +Test: [1800/2387] eta: 0:01:54 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1540 (0.3007) acc1: 96.8750 (88.3081) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.2001 data: 0.1082 max mem: 3538 +Test: [1810/2387] eta: 0:01:52 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1756 (0.3000) acc1: 95.5729 (88.3464) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.1965 data: 0.1053 max mem: 3539 +Test: [1820/2387] eta: 0:01:50 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1890 (0.3002) acc1: 94.2708 (88.3277) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.1950 data: 0.1026 max mem: 3539 +Test: [1830/2387] eta: 0:01:48 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3005 (0.3006) acc1: 79.9479 (88.2955) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1819 data: 0.0890 max mem: 3539 +Test: [1840/2387] eta: 0:01:46 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2047 (0.3001) acc1: 95.5729 (88.3391) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.2064 data: 0.1151 max mem: 3539 +Test: [1850/2387] eta: 0:01:44 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1991 (0.2995) acc1: 96.6146 (88.3838) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1960 data: 0.1058 max mem: 3539 +Test: [1860/2387] eta: 0:01:42 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2012 (0.2990) acc1: 96.6146 (88.4272) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1605 data: 0.0683 max mem: 3539 +Test: [1870/2387] eta: 0:01:40 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2026 (0.2985) acc1: 96.3542 (88.4711) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1909 data: 0.0974 max mem: 3539 +Test: [1880/2387] eta: 0:01:38 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1967 (0.2980) acc1: 96.6146 (88.5154) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1982 data: 0.1052 max mem: 3539 +Test: [1890/2387] eta: 0:01:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2028 (0.2975) acc1: 96.6146 (88.5578) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1896 data: 0.0956 max mem: 3539 +Test: [1900/2387] eta: 0:01:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2039 (0.2972) acc1: 96.8750 (88.5882) acc5: 100.0000 (100.0000) acc1_neg: 72.396 acc5_neg: 100.000 time: 0.1954 data: 0.1012 max mem: 3539 +Test: [1910/2387] eta: 0:01:32 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4853 (0.2990) acc1: 72.3958 (88.4607) acc5: 100.0000 (100.0000) acc1_neg: 72.396 acc5_neg: 100.000 time: 0.2088 data: 0.1132 max mem: 3539 +Test: [1920/2387] eta: 0:01:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6299 (0.3001) acc1: 64.8438 (88.3691) acc5: 100.0000 (100.0000) acc1_neg: 50.260 acc5_neg: 100.000 time: 0.1841 data: 0.0899 max mem: 3539 +Test: [1930/2387] eta: 0:01:28 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4008 (0.3004) acc1: 80.2083 (88.3535) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.1662 data: 0.0738 max mem: 3539 +Test: [1940/2387] eta: 0:01:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3628 (0.3008) acc1: 84.6354 (88.3286) acc5: 100.0000 (100.0000) acc1_neg: 80.990 acc5_neg: 100.000 time: 0.1843 data: 0.0918 max mem: 3539 +Test: [1950/2387] eta: 0:01:24 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4023 (0.3015) acc1: 82.2917 (88.2825) acc5: 100.0000 (100.0000) acc1_neg: 79.167 acc5_neg: 100.000 time: 0.1677 data: 0.0756 max mem: 3539 +Test: [1960/2387] eta: 0:01:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4205 (0.3025) acc1: 79.1667 (88.2058) acc5: 100.0000 (100.0000) acc1_neg: 73.698 acc5_neg: 100.000 time: 0.1699 data: 0.0756 max mem: 3539 +Test: [1970/2387] eta: 0:01:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.6996 (0.3057) acc1: 52.6042 (87.9883) acc5: 100.0000 (100.0000) acc1_neg: 38.281 acc5_neg: 100.000 time: 0.1800 data: 0.0869 max mem: 3539 +Test: [1980/2387] eta: 0:01:18 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.8651 (0.3072) acc1: 47.1354 (87.8777) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1700 data: 0.0781 max mem: 3539 +Test: [1990/2387] eta: 0:01:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1514 (0.3065) acc1: 97.9167 (87.9294) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1695 data: 0.0776 max mem: 3539 +Test: [2000/2387] eta: 0:01:14 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1509 (0.3057) acc1: 98.1771 (87.9823) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1748 data: 0.0825 max mem: 3539 +Test: [2010/2387] eta: 0:01:13 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1540 (0.3050) acc1: 98.1771 (88.0311) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1821 data: 0.0890 max mem: 3539 +Test: [2020/2387] eta: 0:01:11 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1547 (0.3042) acc1: 97.9167 (88.0818) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1836 data: 0.0911 max mem: 3539 +Test: [2030/2387] eta: 0:01:09 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1540 (0.3035) acc1: 98.1771 (88.1305) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1776 data: 0.0852 max mem: 3540 +Test: [2040/2387] eta: 0:01:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1540 (0.3027) acc1: 97.9167 (88.1796) acc5: 100.0000 (100.0000) acc1_neg: 98.958 acc5_neg: 100.000 time: 0.1826 data: 0.0911 max mem: 3540 +Test: [2050/2387] eta: 0:01:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1534 (0.3021) acc1: 97.9167 (88.2223) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.1949 data: 0.0960 max mem: 3540 +Test: [2060/2387] eta: 0:01:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2374 (0.3019) acc1: 94.2708 (88.2502) acc5: 100.0000 (100.0000) acc1_neg: 92.969 acc5_neg: 100.000 time: 0.1946 data: 0.0934 max mem: 3540 +Test: [2070/2387] eta: 0:01:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2458 (0.3016) acc1: 94.2708 (88.2799) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1898 data: 0.0969 max mem: 3540 +Test: [2080/2387] eta: 0:00:59 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2413 (0.3013) acc1: 94.2708 (88.3068) acc5: 100.0000 (100.0000) acc1_neg: 92.448 acc5_neg: 100.000 time: 0.1938 data: 0.1010 max mem: 3540 +Test: [2090/2387] eta: 0:00:57 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2454 (0.3011) acc1: 94.2708 (88.3347) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1874 data: 0.0947 max mem: 3540 +Test: [2100/2387] eta: 0:00:55 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2498 (0.3008) acc1: 94.2708 (88.3608) acc5: 100.0000 (100.0000) acc1_neg: 94.531 acc5_neg: 100.000 time: 0.1872 data: 0.0950 max mem: 3540 +Test: [2110/2387] eta: 0:00:53 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2453 (0.3005) acc1: 94.2708 (88.3910) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1927 data: 0.1004 max mem: 3540 +Test: [2120/2387] eta: 0:00:51 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2539 (0.3008) acc1: 93.7500 (88.3695) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1881 data: 0.0959 max mem: 3540 +Test: [2130/2387] eta: 0:00:49 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3938 (0.3013) acc1: 81.2500 (88.3399) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1848 data: 0.0928 max mem: 3540 +Test: [2140/2387] eta: 0:00:47 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3427 (0.3015) acc1: 85.6771 (88.3346) acc5: 100.0000 (100.0000) acc1_neg: 88.281 acc5_neg: 100.000 time: 0.1843 data: 0.0891 max mem: 3540 +Test: [2150/2387] eta: 0:00:45 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3060 (0.3016) acc1: 88.0208 (88.3254) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.1859 data: 0.0883 max mem: 3540 +Test: [2160/2387] eta: 0:00:43 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3623 (0.3021) acc1: 84.1146 (88.2957) acc5: 100.0000 (100.0000) acc1_neg: 83.854 acc5_neg: 100.000 time: 0.1773 data: 0.0843 max mem: 3540 +Test: [2170/2387] eta: 0:00:41 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5142 (0.3045) acc1: 72.3958 (88.1210) acc5: 100.0000 (100.0000) acc1_neg: 42.188 acc5_neg: 100.000 time: 0.1752 data: 0.0844 max mem: 3540 +Test: [2180/2387] eta: 0:00:39 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.9015 (0.3063) acc1: 43.7500 (87.9897) acc5: 100.0000 (100.0000) acc1_neg: 97.396 acc5_neg: 100.000 time: 0.1738 data: 0.0826 max mem: 3540 +Test: [2190/2387] eta: 0:00:37 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1581 (0.3055) acc1: 97.6562 (88.0403) acc5: 100.0000 (100.0000) acc1_neg: 94.010 acc5_neg: 100.000 time: 0.1684 data: 0.0761 max mem: 3540 +Test: [2200/2387] eta: 0:00:36 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1529 (0.3053) acc1: 98.4375 (88.0568) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1736 data: 0.0803 max mem: 3540 +Test: [2210/2387] eta: 0:00:34 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1959 (0.3048) acc1: 95.5729 (88.0934) acc5: 100.0000 (100.0000) acc1_neg: 99.219 acc5_neg: 100.000 time: 0.1705 data: 0.0785 max mem: 3540 +Test: [2220/2387] eta: 0:00:32 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.1989 (0.3045) acc1: 95.5729 (88.1238) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1716 data: 0.0802 max mem: 3540 +Test: [2230/2387] eta: 0:00:30 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2087 (0.3040) acc1: 95.0521 (88.1565) acc5: 100.0000 (100.0000) acc1_neg: 98.698 acc5_neg: 100.000 time: 0.1822 data: 0.0912 max mem: 3540 +Test: [2240/2387] eta: 0:00:28 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2329 (0.3037) acc1: 94.7917 (88.1854) acc5: 100.0000 (100.0000) acc1_neg: 89.323 acc5_neg: 100.000 time: 0.1758 data: 0.0840 max mem: 3540 +Test: [2250/2387] eta: 0:00:26 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2628 (0.3038) acc1: 92.7083 (88.1987) acc5: 100.0000 (100.0000) acc1_neg: 95.052 acc5_neg: 100.000 time: 0.1718 data: 0.0775 max mem: 3540 +Test: [2260/2387] eta: 0:00:24 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3497 (0.3041) acc1: 88.0208 (88.1857) acc5: 100.0000 (100.0000) acc1_neg: 95.312 acc5_neg: 100.000 time: 0.1729 data: 0.0783 max mem: 3541 +Test: [2270/2387] eta: 0:00:22 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3497 (0.3041) acc1: 84.6354 (88.1847) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1703 data: 0.0776 max mem: 3541 +Test: [2280/2387] eta: 0:00:20 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2838 (0.3041) acc1: 89.3229 (88.1887) acc5: 100.0000 (100.0000) acc1_neg: 90.625 acc5_neg: 100.000 time: 0.1634 data: 0.0726 max mem: 3541 +Test: [2290/2387] eta: 0:00:18 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.2651 (0.3040) acc1: 90.1042 (88.1961) acc5: 100.0000 (100.0000) acc1_neg: 82.552 acc5_neg: 100.000 time: 0.1453 data: 0.0554 max mem: 3541 +Test: [2300/2387] eta: 0:00:16 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3563 (0.3044) acc1: 84.3750 (88.1742) acc5: 100.0000 (100.0000) acc1_neg: 81.510 acc5_neg: 100.000 time: 0.1319 data: 0.0420 max mem: 3541 +Test: [2310/2387] eta: 0:00:14 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4155 (0.3059) acc1: 80.9896 (88.0644) acc5: 100.0000 (100.0000) acc1_neg: 28.385 acc5_neg: 100.000 time: 0.1269 data: 0.0370 max mem: 3541 +Test: [2320/2387] eta: 0:00:12 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.7278 (0.3081) acc1: 52.3438 (87.9007) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.1192 data: 0.0294 max mem: 3541 +Test: [2330/2387] eta: 0:00:10 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.5430 (0.3087) acc1: 66.4062 (87.8565) acc5: 100.0000 (100.0000) acc1_neg: 75.000 acc5_neg: 100.000 time: 0.1162 data: 0.0260 max mem: 3541 +Test: [2340/2387] eta: 0:00:08 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4262 (0.3092) acc1: 77.6042 (87.8239) acc5: 100.0000 (100.0000) acc1_neg: 76.823 acc5_neg: 100.000 time: 0.1185 data: 0.0272 max mem: 3541 +Test: [2350/2387] eta: 0:00:07 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.3486 (0.3095) acc1: 83.3333 (87.7990) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.1195 data: 0.0284 max mem: 3541 +Test: [2360/2387] eta: 0:00:05 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4106 (0.3103) acc1: 80.4688 (87.7502) acc5: 100.0000 (100.0000) acc1_neg: 77.604 acc5_neg: 100.000 time: 0.1142 data: 0.0236 max mem: 3541 +Test: [2370/2387] eta: 0:00:03 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.4504 (0.3113) acc1: 76.3021 (87.6795) acc5: 100.0000 (100.0000) acc1_neg: 55.208 acc5_neg: 100.000 time: 0.1109 data: 0.0206 max mem: 3541 +Test: [2380/2387] eta: 0:00:01 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.7500 (0.3142) acc1: 51.3021 (87.4705) acc5: 100.0000 (100.0000) acc1_neg: 34.635 acc5_neg: 100.000 time: 0.1120 data: 0.0223 max mem: 3541 +Test: [2386/2387] eta: 0:00:00 acc1_pos: 100.000 acc5_pos: 100.000 loss: 0.9022 (0.3148) acc1: 39.8438 (87.4224) acc5: 100.0000 (100.0000) acc1_neg: 97.674 acc5_neg: 100.000 time: 0.1065 data: 0.0195 max mem: 3541 +Test: Total time: 0:07:31 (0.1890 s / it) +* Overall Acc@1 87.422 Acc@5 100.000 + target == real Acc@1 79.617 Acc@5 100.000 + target == fake Acc@1 87.971 Acc@5 100.000 + loss 0.315 + PR-AUC 0.9930 ROC-AUC 0.9225 EER 15.33% (thr=0.5840) +Accuracy of the network on the 916267 test images: 87.4% +[rank0]:[W725 18:25:30.024947331 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:25:50.131000 2572751 torch/distributed/run.py:766] +W0725 18:25:50.131000 2572751 torch/distributed/run.py:766] ***************************************** +W0725 18:25:50.131000 2572751 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:25:50.131000 2572751 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:26:18.453862812 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:26:18.672175623 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:26:18.068640250 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:26:18.106039433 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:26:18.124415426 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 3): env:// +[rank6]:[W725 18:26:19.354210940 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:26:19.363667187 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:26:19.375785606 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/CDFv2_ori', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed4_resampled/benchmarks/CDFv2_ori', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed4_resampled_n116-040-168_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/517] eta: 1:07:22 acc1_pos: 87.500 acc5_pos: 100.000 loss: 0.3767 (0.3767) acc1: 87.5000 (87.5000) acc5: 100.0000 (100.0000) time: 7.8194 data: 6.2515 max mem: 3531 +Test: [ 10/517] eta: 0:06:45 acc1_pos: 48.958 acc5_pos: 100.000 loss: 0.6881 (0.7098) acc1: 61.4583 (61.2689) acc5: 100.0000 (100.0000) time: 0.7998 data: 0.5687 max mem: 3531 +Test: [ 20/517] eta: 0:03:51 acc1_pos: 50.260 acc5_pos: 100.000 loss: 0.6881 (0.7585) acc1: 58.5938 (57.8497) acc5: 100.0000 (100.0000) time: 0.0989 data: 0.0004 max mem: 3531 +Test: [ 30/517] eta: 0:03:13 acc1_pos: 46.354 acc5_pos: 100.000 loss: 0.7406 (0.7560) acc1: 51.5625 (57.4177) acc5: 100.0000 (100.0000) time: 0.1772 data: 0.0815 max mem: 3531 +Test: [ 40/517] eta: 0:02:54 acc1_pos: 61.719 acc5_pos: 100.000 loss: 0.7699 (0.7538) acc1: 51.3021 (57.3552) acc5: 100.0000 (100.0000) time: 0.2602 data: 0.1673 max mem: 3531 +Test: [ 50/517] eta: 0:02:41 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.6822 (0.7366) acc1: 61.7188 (58.4661) acc5: 100.0000 (100.0000) acc1_neg: 17.708 acc5_neg: 100.000 time: 0.2636 data: 0.1717 max mem: 3531 +Test: [ 60/517] eta: 0:02:30 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.7232 (0.7758) acc1: 47.3958 (54.4698) acc5: 100.0000 (100.0000) acc1_neg: 51.042 acc5_neg: 100.000 time: 0.2546 data: 0.1583 max mem: 3531 +Test: [ 70/517] eta: 0:02:24 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.6409 (0.7516) acc1: 56.2500 (55.8832) acc5: 100.0000 (100.0000) acc1_neg: 65.625 acc5_neg: 100.000 time: 0.2639 data: 0.1623 max mem: 3531 +Test: [ 80/517] eta: 0:02:18 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.5772 (0.7342) acc1: 65.6250 (56.8930) acc5: 100.0000 (100.0000) acc1_neg: 48.438 acc5_neg: 100.000 time: 0.2799 data: 0.1837 max mem: 3531 +Test: [ 90/517] eta: 0:02:15 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.5317 (0.7061) acc1: 71.0938 (59.3607) acc5: 100.0000 (100.0000) acc1_neg: 69.271 acc5_neg: 100.000 time: 0.2949 data: 0.2000 max mem: 3531 +Test: [100/517] eta: 0:02:06 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.5581 (0.7057) acc1: 69.2708 (59.1687) acc5: 100.0000 (100.0000) acc1_neg: 59.375 acc5_neg: 100.000 time: 0.2499 data: 0.1545 max mem: 3531 +Test: [110/517] eta: 0:02:01 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.6845 (0.7026) acc1: 56.2500 (58.9292) acc5: 100.0000 (100.0000) acc1_neg: 60.417 acc5_neg: 100.000 time: 0.2151 data: 0.1235 max mem: 3531 +Test: [120/517] eta: 0:01:57 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.7205 (0.7093) acc1: 50.2604 (57.8297) acc5: 100.0000 (100.0000) acc1_neg: 43.490 acc5_neg: 100.000 time: 0.2549 data: 0.1623 max mem: 3531 +Test: [130/517] eta: 0:01:53 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.7850 (0.7139) acc1: 44.0104 (57.0452) acc5: 100.0000 (100.0000) acc1_neg: 58.854 acc5_neg: 100.000 time: 0.2697 data: 0.1736 max mem: 3531 +Test: [140/517] eta: 0:01:48 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.8082 (0.7256) acc1: 43.7500 (56.0561) acc5: 100.0000 (100.0000) acc1_neg: 39.844 acc5_neg: 100.000 time: 0.2482 data: 0.1532 max mem: 3531 +Test: [150/517] eta: 0:01:45 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.8474 (0.7240) acc1: 45.0521 (56.2500) acc5: 100.0000 (100.0000) acc1_neg: 74.219 acc5_neg: 100.000 time: 0.2431 data: 0.1456 max mem: 3531 +Test: [160/517] eta: 0:01:41 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.7848 (0.7267) acc1: 52.6042 (56.1707) acc5: 100.0000 (100.0000) acc1_neg: 65.625 acc5_neg: 100.000 time: 0.2549 data: 0.1548 max mem: 3531 +Test: [170/517] eta: 0:01:38 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.7419 (0.7246) acc1: 54.6875 (56.4130) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.2599 data: 0.1626 max mem: 3531 +Test: [180/517] eta: 0:01:35 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.7010 (0.7163) acc1: 61.1979 (57.1161) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.2608 data: 0.1656 max mem: 3531 +Test: [190/517] eta: 0:01:31 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.3788 (0.6964) acc1: 85.9375 (58.8651) acc5: 100.0000 (100.0000) acc1_neg: 87.500 acc5_neg: 100.000 time: 0.2555 data: 0.1638 max mem: 3531 +Test: [200/517] eta: 0:01:28 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.3923 (0.6869) acc1: 84.6354 (59.7300) acc5: 100.0000 (100.0000) acc1_neg: 77.865 acc5_neg: 100.000 time: 0.2640 data: 0.1725 max mem: 3531 +Test: [210/517] eta: 0:01:25 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.5531 (0.6839) acc1: 70.3125 (60.1464) acc5: 100.0000 (100.0000) acc1_neg: 59.635 acc5_neg: 100.000 time: 0.2645 data: 0.1719 max mem: 3532 +Test: [220/517] eta: 0:01:22 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.6405 (0.6820) acc1: 65.8854 (60.3955) acc5: 100.0000 (100.0000) acc1_neg: 46.615 acc5_neg: 100.000 time: 0.2649 data: 0.1699 max mem: 3532 +Test: [230/517] eta: 0:01:19 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.7063 (0.6854) acc1: 53.6458 (59.8564) acc5: 100.0000 (100.0000) acc1_neg: 66.146 acc5_neg: 100.000 time: 0.2647 data: 0.1701 max mem: 3532 +Test: [240/517] eta: 0:01:16 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.6353 (0.6779) acc1: 64.0625 (60.4512) acc5: 100.0000 (100.0000) acc1_neg: 65.885 acc5_neg: 100.000 time: 0.2549 data: 0.1637 max mem: 3532 +Test: [250/517] eta: 0:01:13 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.6353 (0.6813) acc1: 64.0625 (59.8917) acc5: 100.0000 (100.0000) acc1_neg: 38.281 acc5_neg: 100.000 time: 0.2549 data: 0.1614 max mem: 3532 +Test: [260/517] eta: 0:01:11 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.8121 (0.6880) acc1: 40.6250 (59.0647) acc5: 100.0000 (100.0000) acc1_neg: 76.042 acc5_neg: 100.000 time: 0.2703 data: 0.1769 max mem: 3532 +Test: [270/517] eta: 0:01:08 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.7248 (0.6873) acc1: 44.7917 (58.8676) acc5: 100.0000 (100.0000) acc1_neg: 57.292 acc5_neg: 100.000 time: 0.2699 data: 0.1750 max mem: 3532 +Test: [280/517] eta: 0:01:05 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.6362 (0.6834) acc1: 57.2917 (59.1415) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.2546 data: 0.1593 max mem: 3532 +Test: [290/517] eta: 0:01:02 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.5810 (0.6801) acc1: 66.9271 (59.3848) acc5: 100.0000 (100.0000) acc1_neg: 62.240 acc5_neg: 100.000 time: 0.2652 data: 0.1741 max mem: 3532 +Test: [300/517] eta: 0:00:59 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.5098 (0.6738) acc1: 71.3542 (59.8076) acc5: 100.0000 (100.0000) acc1_neg: 52.083 acc5_neg: 100.000 time: 0.2598 data: 0.1671 max mem: 3532 +Test: [310/517] eta: 0:00:56 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.6465 (0.6781) acc1: 52.3438 (59.3465) acc5: 100.0000 (100.0000) acc1_neg: 52.344 acc5_neg: 100.000 time: 0.2547 data: 0.1615 max mem: 3532 +Test: [320/517] eta: 0:00:53 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.8094 (0.6821) acc1: 46.0938 (58.9191) acc5: 100.0000 (100.0000) acc1_neg: 35.677 acc5_neg: 100.000 time: 0.2699 data: 0.1763 max mem: 3532 +Test: [330/517] eta: 0:00:50 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.7634 (0.6828) acc1: 47.9167 (58.8148) acc5: 100.0000 (100.0000) acc1_neg: 78.646 acc5_neg: 100.000 time: 0.2548 data: 0.1605 max mem: 3532 +Test: [340/517] eta: 0:00:48 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.5915 (0.6795) acc1: 64.3229 (59.1169) acc5: 100.0000 (100.0000) acc1_neg: 45.573 acc5_neg: 100.000 time: 0.2499 data: 0.1566 max mem: 3532 +Test: [350/517] eta: 0:00:45 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.6043 (0.6787) acc1: 70.3125 (59.2830) acc5: 100.0000 (100.0000) acc1_neg: 70.312 acc5_neg: 100.000 time: 0.2649 data: 0.1687 max mem: 3532 +Test: [360/517] eta: 0:00:42 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.6784 (0.6801) acc1: 64.0625 (59.2091) acc5: 100.0000 (100.0000) acc1_neg: 37.760 acc5_neg: 100.000 time: 0.2449 data: 0.1459 max mem: 3532 +Test: [370/517] eta: 0:00:39 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.6549 (0.6758) acc1: 65.3646 (59.6389) acc5: 100.0000 (100.0000) acc1_neg: 69.271 acc5_neg: 100.000 time: 0.2405 data: 0.1451 max mem: 3532 +Test: [380/517] eta: 0:00:37 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.4929 (0.6706) acc1: 77.6042 (60.1323) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2658 data: 0.1742 max mem: 3532 +Test: [390/517] eta: 0:00:34 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.4695 (0.6649) acc1: 79.6875 (60.6931) acc5: 100.0000 (100.0000) acc1_neg: 60.156 acc5_neg: 100.000 time: 0.2793 data: 0.1847 max mem: 3532 +Test: [400/517] eta: 0:00:31 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.4701 (0.6636) acc1: 78.9062 (60.8628) acc5: 100.0000 (100.0000) acc1_neg: 75.781 acc5_neg: 100.000 time: 0.2590 data: 0.1635 max mem: 3532 +Test: [410/517] eta: 0:00:28 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.5534 (0.6598) acc1: 67.4479 (61.1574) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.2399 data: 0.1436 max mem: 3532 +Test: [420/517] eta: 0:00:26 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.5769 (0.6608) acc1: 67.1875 (61.0556) acc5: 100.0000 (100.0000) acc1_neg: 41.927 acc5_neg: 100.000 time: 0.2699 data: 0.1730 max mem: 3532 +Test: [430/517] eta: 0:00:23 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.7242 (0.6651) acc1: 46.8750 (60.5707) acc5: 100.0000 (100.0000) acc1_neg: 51.302 acc5_neg: 100.000 time: 0.2658 data: 0.1718 max mem: 3532 +Test: [440/517] eta: 0:00:20 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.6939 (0.6628) acc1: 51.3021 (60.7615) acc5: 100.0000 (100.0000) acc1_neg: 73.438 acc5_neg: 100.000 time: 0.2499 data: 0.1558 max mem: 3533 +Test: [450/517] eta: 0:00:18 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.6502 (0.6635) acc1: 61.4583 (60.6967) acc5: 100.0000 (100.0000) acc1_neg: 48.698 acc5_neg: 100.000 time: 0.2590 data: 0.1640 max mem: 3533 +Test: [460/517] eta: 0:00:15 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.6502 (0.6627) acc1: 61.4583 (60.7601) acc5: 100.0000 (100.0000) acc1_neg: 82.292 acc5_neg: 100.000 time: 0.2449 data: 0.1507 max mem: 3533 +Test: [470/517] eta: 0:00:12 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.4864 (0.6588) acc1: 73.1771 (61.1266) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.2351 data: 0.1420 max mem: 3533 +Test: [480/517] eta: 0:00:09 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.4849 (0.6563) acc1: 79.1667 (61.3820) acc5: 100.0000 (100.0000) acc1_neg: 63.542 acc5_neg: 100.000 time: 0.2244 data: 0.1323 max mem: 3533 +Test: [490/517] eta: 0:00:07 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.6047 (0.6635) acc1: 63.5417 (60.8961) acc5: 100.0000 (100.0000) acc1_neg: 70.052 acc5_neg: 100.000 time: 0.1952 data: 0.1050 max mem: 3533 +Test: [500/517] eta: 0:00:04 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.6789 (0.6632) acc1: 54.4271 (60.8699) acc5: 100.0000 (100.0000) acc1_neg: 56.771 acc5_neg: 100.000 time: 0.1814 data: 0.0918 max mem: 3533 +Test: [510/517] eta: 0:00:01 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.6789 (0.6644) acc1: 54.4271 (60.6857) acc5: 100.0000 (100.0000) acc1_neg: 50.781 acc5_neg: 100.000 time: 0.1778 data: 0.0884 max mem: 3533 +Test: [516/517] eta: 0:00:00 acc1_pos: 71.560 acc5_pos: 100.000 loss: 0.7103 (0.6652) acc1: 52.0833 (60.5623) acc5: 100.0000 (100.0000) acc1_neg: 39.929 acc5_neg: 100.000 time: 0.1668 data: 0.0756 max mem: 3533 +Test: Total time: 0:02:14 (0.2608 s / it) +* Overall Acc@1 60.562 Acc@5 100.000 + target == real Acc@1 60.263 Acc@5 100.000 + target == fake Acc@1 60.593 Acc@5 100.000 + loss 0.665 + PR-AUC 0.9356 ROC-AUC 0.6399 EER 39.62% (thr=0.5015) +Accuracy of the network on the 198427 test images: 60.6% +[rank0]:[W725 18:28:42.245271109 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +Missing benchmark directory: /dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DiFF_processed +Skipping missing benchmark: DiFF_processed +W0725 18:29:03.642000 2573326 torch/distributed/run.py:766] +W0725 18:29:03.642000 2573326 torch/distributed/run.py:766] ***************************************** +W0725 18:29:03.642000 2573326 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:29:03.642000 2573326 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:29:37.388158829 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:29:37.685335556 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:29:37.715548367 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank4]:[W725 18:29:37.722138691 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:29:37.807879493 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 2): env:// +| distributed init (rank 6): env:// +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:29:37.979603101 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:29:37.983669778 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:29:37.986127117 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/Wild_Deepfake_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed4_resampled/benchmarks/Wild_Deepfake_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed4_resampled_n116-040-168_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/389] eta: 0:37:55 acc1_pos: 0.521 acc5_pos: 100.000 loss: 2.0251 (2.0251) acc1: 0.5208 (0.5208) acc5: 100.0000 (100.0000) time: 5.8498 data: 4.1542 max mem: 3531 +Test: [ 10/389] eta: 0:03:58 acc1_pos: 7.552 acc5_pos: 100.000 loss: 1.6935 (1.7374) acc1: 2.3438 (3.9063) acc5: 100.0000 (100.0000) time: 0.6296 data: 0.3936 max mem: 3531 +Test: [ 20/389] eta: 0:02:18 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.7491 (1.8380) acc1: 1.3021 (2.4058) acc5: 100.0000 (100.0000) time: 0.1024 data: 0.0090 max mem: 3531 +Test: [ 30/389] eta: 0:01:53 acc1_pos: 0.260 acc5_pos: 100.000 loss: 2.0987 (1.9143) acc1: 0.0000 (1.6465) acc5: 100.0000 (100.0000) time: 0.1435 data: 0.0487 max mem: 3531 +Test: [ 40/389] eta: 0:01:42 acc1_pos: 1.042 acc5_pos: 100.000 loss: 2.0116 (1.9198) acc1: 0.0000 (1.3021) acc5: 100.0000 (100.0000) time: 0.2049 data: 0.1086 max mem: 3531 +Test: [ 50/389] eta: 0:01:32 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.9741 (1.9335) acc1: 0.0000 (1.0825) acc5: 100.0000 (100.0000) time: 0.2065 data: 0.1114 max mem: 3531 +Test: [ 60/389] eta: 0:01:25 acc1_pos: 0.000 acc5_pos: 100.000 loss: 1.9246 (1.9544) acc1: 0.0000 (0.9264) acc5: 100.0000 (100.0000) time: 0.1949 data: 0.1027 max mem: 3531 +Test: [ 70/389] eta: 0:01:20 acc1_pos: 8.073 acc5_pos: 100.000 loss: 1.8615 (1.9120) acc1: 0.0000 (1.7679) acc5: 100.0000 (100.0000) time: 0.2006 data: 0.1054 max mem: 3531 +Test: [ 80/389] eta: 0:01:15 acc1_pos: 7.031 acc5_pos: 100.000 loss: 1.5245 (1.8648) acc1: 3.6458 (2.8678) acc5: 100.0000 (100.0000) time: 0.1949 data: 0.0988 max mem: 3531 +Test: [ 90/389] eta: 0:01:11 acc1_pos: 5.208 acc5_pos: 100.000 loss: 1.3982 (1.8128) acc1: 7.0312 (3.7002) acc5: 100.0000 (100.0000) time: 0.1927 data: 0.0935 max mem: 3531 +Test: [100/389] eta: 0:01:08 acc1_pos: 80.208 acc5_pos: 100.000 loss: 1.4322 (1.8032) acc1: 4.1667 (4.8860) acc5: 100.0000 (100.0000) time: 0.2016 data: 0.1054 max mem: 3531 +Test: [110/389] eta: 0:01:04 acc1_pos: 2.083 acc5_pos: 100.000 loss: 1.5216 (1.7678) acc1: 2.8646 (6.2828) acc5: 100.0000 (100.0000) time: 0.1916 data: 0.1015 max mem: 3531 +Test: [120/389] eta: 0:01:01 acc1_pos: 0.260 acc5_pos: 100.000 loss: 1.5216 (1.7712) acc1: 1.8229 (5.9573) acc5: 100.0000 (100.0000) time: 0.2006 data: 0.1102 max mem: 3531 +Test: [130/389] eta: 0:00:58 acc1_pos: 1.227 acc5_pos: 100.000 loss: 1.4909 (1.6779) acc1: 4.9479 (11.2377) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.2052 data: 0.1130 max mem: 3531 +Test: [140/389] eta: 0:00:55 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1467 (1.5693) acc1: 99.7396 (17.5236) acc5: 100.0000 (100.0000) acc1_neg: 99.479 acc5_neg: 100.000 time: 0.1842 data: 0.0925 max mem: 3531 +Test: [150/389] eta: 0:00:52 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1584 (1.4785) acc1: 100.0000 (22.9184) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1848 data: 0.0920 max mem: 3531 +Test: [160/389] eta: 0:00:49 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1608 (1.3952) acc1: 99.7396 (27.6883) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1811 data: 0.0878 max mem: 3531 +Test: [170/389] eta: 0:00:47 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1250 (1.3213) acc1: 100.0000 (31.9140) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1880 data: 0.0967 max mem: 3531 +Test: [180/389] eta: 0:00:44 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1250 (1.2560) acc1: 100.0000 (35.6426) acc5: 100.0000 (100.0000) acc1_neg: 98.177 acc5_neg: 100.000 time: 0.1829 data: 0.0910 max mem: 3531 +Test: [190/389] eta: 0:00:42 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1450 (1.2006) acc1: 99.4792 (38.8621) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1718 data: 0.0795 max mem: 3531 +Test: [200/389] eta: 0:00:39 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1450 (1.1481) acc1: 99.7396 (41.9025) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1789 data: 0.0854 max mem: 3531 +Test: [210/389] eta: 0:00:37 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1212 (1.0994) acc1: 100.0000 (44.6522) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1731 data: 0.0803 max mem: 3532 +Test: [220/389] eta: 0:00:34 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1011 (1.0544) acc1: 100.0000 (47.1566) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1668 data: 0.0754 max mem: 3532 +Test: [230/389] eta: 0:00:32 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1182 (1.0154) acc1: 100.0000 (49.4070) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1670 data: 0.0762 max mem: 3532 +Test: [240/389] eta: 0:00:30 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1456 (0.9806) acc1: 100.0000 (51.4566) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1794 data: 0.0893 max mem: 3532 +Test: [250/389] eta: 0:00:28 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1600 (0.9481) acc1: 99.7396 (53.3730) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1840 data: 0.0937 max mem: 3532 +Test: [260/389] eta: 0:00:26 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1440 (0.9175) acc1: 100.0000 (55.1515) acc5: 100.0000 (100.0000) acc1_neg: 98.438 acc5_neg: 100.000 time: 0.1835 data: 0.0925 max mem: 3532 +Test: [270/389] eta: 0:00:24 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1391 (0.8900) acc1: 100.0000 (56.7776) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1863 data: 0.0950 max mem: 3532 +Test: [280/389] eta: 0:00:21 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1342 (0.8633) acc1: 100.0000 (58.3000) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1796 data: 0.0887 max mem: 3532 +Test: [290/389] eta: 0:00:19 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1342 (0.8388) acc1: 100.0000 (59.7231) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1728 data: 0.0822 max mem: 3532 +Test: [300/389] eta: 0:00:17 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1147 (0.8144) acc1: 100.0000 (61.0612) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1841 data: 0.0929 max mem: 3532 +Test: [310/389] eta: 0:00:15 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1165 (0.7928) acc1: 100.0000 (62.2873) acc5: 100.0000 (100.0000) acc1_neg: 99.740 acc5_neg: 100.000 time: 0.1865 data: 0.0956 max mem: 3532 +Test: [320/389] eta: 0:00:13 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1280 (0.7728) acc1: 100.0000 (63.4524) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1672 data: 0.0762 max mem: 3532 +Test: [330/389] eta: 0:00:11 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1578 (0.7541) acc1: 100.0000 (64.5566) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1605 data: 0.0693 max mem: 3532 +Test: [340/389] eta: 0:00:09 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1344 (0.7358) acc1: 100.0000 (65.5960) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1558 data: 0.0657 max mem: 3532 +Test: [350/389] eta: 0:00:07 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1278 (0.7189) acc1: 100.0000 (66.5732) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1411 data: 0.0514 max mem: 3532 +Test: [360/389] eta: 0:00:05 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1282 (0.7024) acc1: 100.0000 (67.4977) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1311 data: 0.0414 max mem: 3532 +Test: [370/389] eta: 0:00:03 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.1265 (0.6882) acc1: 100.0000 (68.3359) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1235 data: 0.0338 max mem: 3532 +Test: [380/389] eta: 0:00:01 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.0975 (0.6730) acc1: 100.0000 (69.1669) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1175 data: 0.0279 max mem: 3532 +Test: [388/389] eta: 0:00:00 acc1_pos: 1.227 acc5_pos: 100.000 loss: 0.0975 (0.6621) acc1: 100.0000 (69.7887) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1195 data: 0.0263 max mem: 3532 +Test: Total time: 0:01:12 (0.1874 s / it) +* Overall Acc@1 69.789 Acc@5 100.000 + target == real Acc@1 5.870 Acc@5 100.000 + target == fake Acc@1 99.500 Acc@5 100.000 + loss 0.662 + PR-AUC 0.8015 ROC-AUC 0.6830 EER 37.09% (thr=0.8730) +Accuracy of the network on the 149358 test images: 69.8% +[rank0]:[W725 18:31:02.930483146 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:31:21.663000 2573913 torch/distributed/run.py:766] +W0725 18:31:21.663000 2573913 torch/distributed/run.py:766] ***************************************** +W0725 18:31:21.663000 2573913 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:31:21.663000 2573913 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank0]:[W725 18:31:50.291510552 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:31:50.616154697 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:31:50.170570424 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:31:51.216914471 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:31:51.283660337 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:31:51.444215914 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 1): env:// +[rank2]:[W725 18:31:51.500417574 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:31:51.513506375 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/ForgeryNet_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed4_resampled/benchmarks/ForgeryNet_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed4_resampled_n116-040-168_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/333] eta: 0:32:31 acc1_pos: 5.469 acc5_pos: 100.000 loss: 1.6815 (1.6815) acc1: 5.4688 (5.4688) acc5: 100.0000 (100.0000) time: 5.8593 data: 4.4542 max mem: 3531 +Test: [ 10/333] eta: 0:03:18 acc1_pos: 15.885 acc5_pos: 100.000 loss: 1.4291 (1.4976) acc1: 14.3229 (15.6960) acc5: 100.0000 (100.0000) time: 0.6144 data: 0.4052 max mem: 3531 +Test: [ 20/333] eta: 0:01:56 acc1_pos: 20.573 acc5_pos: 100.000 loss: 1.5212 (1.5339) acc1: 15.8854 (17.2371) acc5: 100.0000 (100.0000) time: 0.0972 data: 0.0058 max mem: 3531 +Test: [ 30/333] eta: 0:01:55 acc1_pos: 46.875 acc5_pos: 100.000 loss: 1.1816 (1.3265) acc1: 22.6562 (26.7053) acc5: 100.0000 (100.0000) time: 0.2553 data: 0.1637 max mem: 3531 +Test: [ 40/333] eta: 0:01:46 acc1_pos: 48.698 acc5_pos: 100.000 loss: 0.8779 (1.2169) acc1: 46.8750 (32.1265) acc5: 100.0000 (100.0000) time: 0.3573 data: 0.2670 max mem: 3531 +Test: [ 50/333] eta: 0:01:32 acc1_pos: 53.906 acc5_pos: 100.000 loss: 0.8779 (1.1519) acc1: 44.2708 (34.9826) acc5: 100.0000 (100.0000) time: 0.2418 data: 0.1499 max mem: 3531 +Test: [ 60/333] eta: 0:01:19 acc1_pos: 55.990 acc5_pos: 100.000 loss: 0.8294 (1.1041) acc1: 51.0417 (37.6708) acc5: 100.0000 (100.0000) time: 0.1429 data: 0.0486 max mem: 3531 +Test: [ 70/333] eta: 0:01:13 acc1_pos: 54.688 acc5_pos: 100.000 loss: 0.8222 (1.0652) acc1: 52.8646 (39.8401) acc5: 100.0000 (100.0000) time: 0.1599 data: 0.0641 max mem: 3531 +Test: [ 80/333] eta: 0:01:09 acc1_pos: 39.583 acc5_pos: 100.000 loss: 0.8848 (1.0506) acc1: 44.5312 (40.4128) acc5: 100.0000 (100.0000) time: 0.2250 data: 0.1308 max mem: 3531 +Test: [ 90/333] eta: 0:01:04 acc1_pos: 65.365 acc5_pos: 100.000 loss: 0.8783 (1.0159) acc1: 53.9062 (42.6511) acc5: 100.0000 (100.0000) time: 0.2149 data: 0.1218 max mem: 3531 +Test: [100/333] eta: 0:01:00 acc1_pos: 56.510 acc5_pos: 100.000 loss: 0.6757 (0.9827) acc1: 60.4167 (44.6034) acc5: 100.0000 (100.0000) time: 0.1995 data: 0.1053 max mem: 3531 +Test: [110/333] eta: 0:00:56 acc1_pos: 50.000 acc5_pos: 100.000 loss: 0.6679 (0.9561) acc1: 63.5417 (46.3377) acc5: 100.0000 (100.0000) time: 0.2017 data: 0.1094 max mem: 3531 +Test: [120/333] eta: 0:00:53 acc1_pos: 62.240 acc5_pos: 100.000 loss: 0.6770 (0.9350) acc1: 61.7188 (47.5917) acc5: 100.0000 (100.0000) time: 0.1948 data: 0.1036 max mem: 3531 +Test: [130/333] eta: 0:00:49 acc1_pos: 48.438 acc5_pos: 100.000 loss: 0.6697 (0.9149) acc1: 61.7188 (48.6860) acc5: 100.0000 (100.0000) time: 0.1946 data: 0.1029 max mem: 3531 +Test: [140/333] eta: 0:00:46 acc1_pos: 44.531 acc5_pos: 100.000 loss: 0.6504 (0.8934) acc1: 66.6667 (50.0850) acc5: 100.0000 (100.0000) time: 0.1953 data: 0.1031 max mem: 3531 +Test: [150/333] eta: 0:00:43 acc1_pos: 56.510 acc5_pos: 100.000 loss: 0.6415 (0.8793) acc1: 63.0208 (50.7640) acc5: 100.0000 (100.0000) time: 0.2050 data: 0.1128 max mem: 3531 +Test: [160/333] eta: 0:00:41 acc1_pos: 47.697 acc5_pos: 100.000 loss: 0.7029 (0.8679) acc1: 58.5938 (51.3053) acc5: 100.0000 (100.0000) acc1_neg: 48.177 acc5_neg: 100.000 time: 0.2033 data: 0.1105 max mem: 3531 +Test: [170/333] eta: 0:00:38 acc1_pos: 47.697 acc5_pos: 100.000 loss: 0.6852 (0.8571) acc1: 59.8958 (51.8001) acc5: 100.0000 (100.0000) acc1_neg: 63.021 acc5_neg: 100.000 time: 0.1950 data: 0.1019 max mem: 3531 +Test: [180/333] eta: 0:00:35 acc1_pos: 47.697 acc5_pos: 100.000 loss: 0.6354 (0.8429) acc1: 62.5000 (52.5855) acc5: 100.0000 (100.0000) acc1_neg: 67.708 acc5_neg: 100.000 time: 0.2010 data: 0.1084 max mem: 3531 +Test: [190/333] eta: 0:00:33 acc1_pos: 47.697 acc5_pos: 100.000 loss: 0.5879 (0.8277) acc1: 66.9271 (53.5027) acc5: 100.0000 (100.0000) acc1_neg: 69.271 acc5_neg: 100.000 time: 0.2008 data: 0.1055 max mem: 3531 +Test: [200/333] eta: 0:00:30 acc1_pos: 47.697 acc5_pos: 100.000 loss: 0.5865 (0.8166) acc1: 66.9271 (54.0967) acc5: 100.0000 (100.0000) acc1_neg: 62.760 acc5_neg: 100.000 time: 0.1889 data: 0.0949 max mem: 3531 +Test: [210/333] eta: 0:00:28 acc1_pos: 47.697 acc5_pos: 100.000 loss: 0.5737 (0.8042) acc1: 67.9688 (54.8640) acc5: 100.0000 (100.0000) acc1_neg: 88.021 acc5_neg: 100.000 time: 0.2032 data: 0.1107 max mem: 3532 +Test: [220/333] eta: 0:00:25 acc1_pos: 47.697 acc5_pos: 100.000 loss: 0.5505 (0.7917) acc1: 68.4896 (55.6007) acc5: 100.0000 (100.0000) acc1_neg: 68.490 acc5_neg: 100.000 time: 0.2112 data: 0.1184 max mem: 3532 +Test: [230/333] eta: 0:00:23 acc1_pos: 47.697 acc5_pos: 100.000 loss: 0.4835 (0.7781) acc1: 74.7396 (56.5645) acc5: 100.0000 (100.0000) acc1_neg: 82.031 acc5_neg: 100.000 time: 0.1978 data: 0.1053 max mem: 3532 +Test: [240/333] eta: 0:00:20 acc1_pos: 47.697 acc5_pos: 100.000 loss: 0.4959 (0.7731) acc1: 74.2188 (56.7211) acc5: 100.0000 (100.0000) acc1_neg: 50.260 acc5_neg: 100.000 time: 0.1854 data: 0.0919 max mem: 3532 +Test: [250/333] eta: 0:00:18 acc1_pos: 47.697 acc5_pos: 100.000 loss: 0.6626 (0.7682) acc1: 59.1146 (56.8995) acc5: 100.0000 (100.0000) acc1_neg: 69.271 acc5_neg: 100.000 time: 0.1827 data: 0.0912 max mem: 3532 +Test: [260/333] eta: 0:00:16 acc1_pos: 47.697 acc5_pos: 100.000 loss: 0.6549 (0.7629) acc1: 60.6771 (57.1001) acc5: 100.0000 (100.0000) acc1_neg: 62.500 acc5_neg: 100.000 time: 0.2031 data: 0.1123 max mem: 3532 +Test: [270/333] eta: 0:00:13 acc1_pos: 47.697 acc5_pos: 100.000 loss: 0.5991 (0.7558) acc1: 64.3229 (57.5406) acc5: 100.0000 (100.0000) acc1_neg: 72.396 acc5_neg: 100.000 time: 0.2095 data: 0.1187 max mem: 3532 +Test: [280/333] eta: 0:00:11 acc1_pos: 47.697 acc5_pos: 100.000 loss: 0.5927 (0.7523) acc1: 67.9688 (57.7143) acc5: 100.0000 (100.0000) acc1_neg: 58.333 acc5_neg: 100.000 time: 0.2026 data: 0.1099 max mem: 3532 +Test: [290/333] eta: 0:00:09 acc1_pos: 47.697 acc5_pos: 100.000 loss: 0.7096 (0.7521) acc1: 56.5104 (57.5924) acc5: 100.0000 (100.0000) acc1_neg: 50.781 acc5_neg: 100.000 time: 0.2042 data: 0.1112 max mem: 3532 +Test: [300/333] eta: 0:00:07 acc1_pos: 47.697 acc5_pos: 100.000 loss: 0.7292 (0.7509) acc1: 55.2083 (57.5495) acc5: 100.0000 (100.0000) acc1_neg: 52.865 acc5_neg: 100.000 time: 0.1712 data: 0.0811 max mem: 3532 +Test: [310/333] eta: 0:00:04 acc1_pos: 47.697 acc5_pos: 100.000 loss: 0.6846 (0.7447) acc1: 58.5938 (57.9398) acc5: 100.0000 (100.0000) acc1_neg: 67.188 acc5_neg: 100.000 time: 0.1446 data: 0.0543 max mem: 3532 +Test: [320/333] eta: 0:00:02 acc1_pos: 47.697 acc5_pos: 100.000 loss: 0.6464 (0.7423) acc1: 60.6771 (58.0267) acc5: 100.0000 (100.0000) acc1_neg: 68.750 acc5_neg: 100.000 time: 0.1276 data: 0.0375 max mem: 3532 +Test: [330/333] eta: 0:00:00 acc1_pos: 47.697 acc5_pos: 100.000 loss: 0.6196 (0.7378) acc1: 64.3229 (58.2995) acc5: 100.0000 (100.0000) acc1_neg: 67.708 acc5_neg: 100.000 time: 0.1157 data: 0.0263 max mem: 3532 +Test: [332/333] eta: 0:00:00 acc1_pos: 47.697 acc5_pos: 100.000 loss: 0.6081 (0.7369) acc1: 65.8854 (58.3505) acc5: 100.0000 (100.0000) acc1_neg: 67.042 acc5_neg: 100.000 time: 0.1188 data: 0.0263 max mem: 3532 +Test: Total time: 0:01:09 (0.2093 s / it) +* Overall Acc@1 58.350 Acc@5 100.000 + target == real Acc@1 51.320 Acc@5 100.000 + target == fake Acc@1 64.609 Acc@5 100.000 + loss 0.737 + PR-AUC 0.6081 ROC-AUC 0.6103 EER 42.24% (thr=0.5571) +Accuracy of the network on the 127843 test images: 58.4% +[rank0]:[W725 18:33:10.992052084 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:33:31.061000 2574470 torch/distributed/run.py:766] +W0725 18:33:31.061000 2574470 torch/distributed/run.py:766] ***************************************** +W0725 18:33:31.061000 2574470 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:33:31.061000 2574470 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:33:58.803230417 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:33:58.953857703 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:33:59.202257119 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:33:59.294029664 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:33:59.309774187 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:33:59.464037881 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:33:59.532186620 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:33:59.612167946 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeFace_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed4_resampled/benchmarks/DeepFakeFace_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed4_resampled_n116-040-168_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/235] eta: 0:26:24 acc1_pos: 17.708 acc5_pos: 100.000 loss: 1.5918 (1.5918) acc1: 17.7083 (17.7083) acc5: 100.0000 (100.0000) time: 6.7435 data: 4.3976 max mem: 3531 +Test: [ 10/235] eta: 0:02:36 acc1_pos: 19.010 acc5_pos: 100.000 loss: 1.5728 (1.5744) acc1: 19.0104 (19.2472) acc5: 100.0000 (100.0000) time: 0.6968 data: 0.4002 max mem: 3531 +Test: [ 20/235] eta: 0:01:29 acc1_pos: 19.792 acc5_pos: 100.000 loss: 1.5707 (1.5727) acc1: 19.2708 (19.3452) acc5: 100.0000 (100.0000) time: 0.1010 data: 0.0070 max mem: 3531 +Test: [ 30/235] eta: 0:01:17 acc1_pos: 18.490 acc5_pos: 100.000 loss: 1.5712 (1.5738) acc1: 19.2708 (19.3128) acc5: 100.0000 (100.0000) time: 0.2011 data: 0.1066 max mem: 3531 +Test: [ 40/235] eta: 0:01:10 acc1_pos: 18.490 acc5_pos: 100.000 loss: 1.5810 (1.5755) acc1: 19.0104 (19.2200) acc5: 100.0000 (100.0000) time: 0.3006 data: 0.2095 max mem: 3531 +Test: [ 50/235] eta: 0:01:00 acc1_pos: 18.750 acc5_pos: 100.000 loss: 1.5846 (1.5778) acc1: 18.7500 (19.1432) acc5: 100.0000 (100.0000) time: 0.2455 data: 0.1537 max mem: 3531 +Test: [ 60/235] eta: 0:00:52 acc1_pos: 20.052 acc5_pos: 100.000 loss: 1.5796 (1.5763) acc1: 19.0104 (19.1300) acc5: 100.0000 (100.0000) time: 0.1792 data: 0.0834 max mem: 3531 +Test: [ 70/235] eta: 0:00:47 acc1_pos: 27.692 acc5_pos: 100.000 loss: 1.5181 (1.4062) acc1: 21.0938 (28.4624) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.1884 data: 0.0920 max mem: 3531 +Test: [ 80/235] eta: 0:00:43 acc1_pos: 27.692 acc5_pos: 100.000 loss: 0.2111 (1.2587) acc1: 93.7500 (36.4615) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2117 data: 0.1167 max mem: 3531 +Test: [ 90/235] eta: 0:00:39 acc1_pos: 27.692 acc5_pos: 100.000 loss: 0.2106 (1.1434) acc1: 93.7500 (42.7427) acc5: 100.0000 (100.0000) acc1_neg: 93.750 acc5_neg: 100.000 time: 0.2120 data: 0.1190 max mem: 3531 +Test: [100/235] eta: 0:00:35 acc1_pos: 27.692 acc5_pos: 100.000 loss: 0.2055 (1.0502) acc1: 93.7500 (47.8367) acc5: 100.0000 (100.0000) acc1_neg: 95.573 acc5_neg: 100.000 time: 0.2039 data: 0.1107 max mem: 3531 +Test: [110/235] eta: 0:00:32 acc1_pos: 27.692 acc5_pos: 100.000 loss: 0.2038 (0.9744) acc1: 94.0104 (51.9825) acc5: 100.0000 (100.0000) acc1_neg: 94.271 acc5_neg: 100.000 time: 0.2091 data: 0.1144 max mem: 3531 +Test: [120/235] eta: 0:00:29 acc1_pos: 27.692 acc5_pos: 100.000 loss: 0.2193 (0.9218) acc1: 91.9271 (54.8295) acc5: 100.0000 (100.0000) acc1_neg: 86.198 acc5_neg: 100.000 time: 0.2245 data: 0.1325 max mem: 3531 +Test: [130/235] eta: 0:00:26 acc1_pos: 27.692 acc5_pos: 100.000 loss: 0.3534 (0.8787) acc1: 85.1562 (57.1466) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.2246 data: 0.1328 max mem: 3531 +Test: [140/235] eta: 0:00:23 acc1_pos: 27.692 acc5_pos: 100.000 loss: 0.3538 (0.8417) acc1: 84.8958 (59.1312) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.2198 data: 0.1275 max mem: 3531 +Test: [150/235] eta: 0:00:21 acc1_pos: 27.692 acc5_pos: 100.000 loss: 0.3546 (0.8097) acc1: 84.6354 (60.8116) acc5: 100.0000 (100.0000) acc1_neg: 85.156 acc5_neg: 100.000 time: 0.2336 data: 0.1420 max mem: 3531 +Test: [160/235] eta: 0:00:18 acc1_pos: 27.692 acc5_pos: 100.000 loss: 0.3595 (0.7817) acc1: 84.8958 (62.3318) acc5: 100.0000 (100.0000) acc1_neg: 86.458 acc5_neg: 100.000 time: 0.2145 data: 0.1230 max mem: 3531 +Test: [170/235] eta: 0:00:15 acc1_pos: 27.692 acc5_pos: 100.000 loss: 0.3166 (0.7460) acc1: 86.9792 (64.2925) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.2049 data: 0.1135 max mem: 3531 +Test: [180/235] eta: 0:00:13 acc1_pos: 27.692 acc5_pos: 100.000 loss: 0.1529 (0.7132) acc1: 96.8750 (66.0840) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.2099 data: 0.1177 max mem: 3531 +Test: [190/235] eta: 0:00:10 acc1_pos: 27.692 acc5_pos: 100.000 loss: 0.1452 (0.6836) acc1: 97.1354 (67.7233) acc5: 100.0000 (100.0000) acc1_neg: 97.656 acc5_neg: 100.000 time: 0.2049 data: 0.1115 max mem: 3531 +Test: [200/235] eta: 0:00:08 acc1_pos: 27.692 acc5_pos: 100.000 loss: 0.1503 (0.6574) acc1: 97.1354 (69.1555) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.1811 data: 0.0887 max mem: 3531 +Test: [210/235] eta: 0:00:05 acc1_pos: 27.692 acc5_pos: 100.000 loss: 0.1555 (0.6334) acc1: 96.6146 (70.4668) acc5: 100.0000 (100.0000) acc1_neg: 96.354 acc5_neg: 100.000 time: 0.1466 data: 0.0563 max mem: 3532 +Test: [220/235] eta: 0:00:03 acc1_pos: 27.692 acc5_pos: 100.000 loss: 0.1504 (0.6114) acc1: 96.8750 (71.6688) acc5: 100.0000 (100.0000) acc1_neg: 96.875 acc5_neg: 100.000 time: 0.1306 data: 0.0407 max mem: 3532 +Test: [230/235] eta: 0:00:01 acc1_pos: 27.692 acc5_pos: 100.000 loss: 0.1490 (0.5915) acc1: 97.1354 (72.7611) acc5: 100.0000 (100.0000) acc1_neg: 96.094 acc5_neg: 100.000 time: 0.1206 data: 0.0313 max mem: 3532 +Test: [234/235] eta: 0:00:00 acc1_pos: 27.692 acc5_pos: 100.000 loss: 0.1503 (0.5839) acc1: 96.8750 (73.0694) acc5: 100.0000 (100.0000) acc1_neg: 100.000 acc5_neg: 100.000 time: 0.1121 data: 0.0261 max mem: 3532 +Test: Total time: 0:00:52 (0.2213 s / it) +* Overall Acc@1 73.069 Acc@5 100.000 + target == real Acc@1 19.156 Acc@5 100.000 + target == fake Acc@1 92.569 Acc@5 100.000 + loss 0.584 + PR-AUC 0.8252 ROC-AUC 0.6704 EER 36.36% (thr=0.8647) +Accuracy of the network on the 89879 test images: 73.1% +[rank0]:[W725 18:34:58.444530743 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +W0725 18:35:17.364000 2575040 torch/distributed/run.py:766] +W0725 18:35:17.364000 2575040 torch/distributed/run.py:766] ***************************************** +W0725 18:35:17.364000 2575040 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0725 18:35:17.364000 2575040 torch/distributed/run.py:766] ***************************************** +| distributed init (rank 0): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +| distributed init (rank 2): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank2]:[W725 18:35:44.078084161 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 6): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank6]:[W725 18:35:45.361349664 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 4): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank4]:[W725 18:35:45.419721552 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 5): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank5]:[W725 18:35:45.620503212 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +[rank0]:[W725 18:35:45.620556286 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +NCCL version 2.26.2+cuda12.2 +| distributed init (rank 7): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank7]:[W725 18:35:45.963722641 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +| distributed init (rank 1): env:// +| distributed init (rank 3): env:// +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank1]:[W725 18:35:45.061749611 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +/home/tiger/.local/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +[rank3]:[W725 18:35:45.066418088 ProcessGroupNCCL.cpp:4718] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can pecify device_id in init_process_group() to force use of a particular device. +Namespace(batch_size=256, epochs=300, bce_loss=False, unscale_lr=False, select_models='', model='deit_base_patch16_LS', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, two_cls=True, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='', attn_only=False, data_path='/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names/DeepFakeDetection_processed', data_set='ScaleDF', inat_category='name', output_dir='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_real_n5_seed4_resampled/benchmarks/DeepFakeDetection_processed', device='cuda', seed=0, resume='/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs/control_fixed500000_real_domains_n5_classseed4_resampled_n116-040-168_20260725_112559/best_checkpoint.pth', start_epoch=0, eval=True, eval_crop_ratio=1.0, dist_eval=False, num_workers=10, pin_mem=True, distributed=True, world_size=8, dist_url='env://', rank=0, gpu=0, dist_backend='nccl') +Creating model: deit_base_patch16_LS +number of params: 85817858 +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/home/tiger/.local/lib/python3.11/site-packages/timm/utils/cuda.py:40: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead. + self._scaler = torch.cuda.amp.GradScaler() +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +/tmp/ScaleDF_repro_20260724/code/deit/engine_cross_2direct_balance_test.py:144: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. + with torch.cuda.amp.autocast(): +Test: [ 0/269] eta: 0:27:56 acc1_pos: 40.625 acc5_pos: 100.000 loss: 0.9512 (0.9512) acc1: 40.6250 (40.6250) acc5: 100.0000 (100.0000) time: 6.2315 data: 4.5920 max mem: 3531 +Test: [ 10/269] eta: 0:02:47 acc1_pos: 41.406 acc5_pos: 100.000 loss: 1.0851 (1.0645) acc1: 22.6562 (27.8883) acc5: 100.0000 (100.0000) time: 0.6481 data: 0.4178 max mem: 3531 +Test: [ 20/269] eta: 0:01:35 acc1_pos: 39.323 acc5_pos: 100.000 loss: 1.1093 (1.1049) acc1: 22.6562 (28.0134) acc5: 100.0000 (100.0000) time: 0.0925 data: 0.0003 max mem: 3531 +Test: [ 30/269] eta: 0:01:16 acc1_pos: 35.417 acc5_pos: 100.000 loss: 0.9581 (1.0781) acc1: 32.2917 (32.3085) acc5: 100.0000 (100.0000) time: 0.1426 data: 0.0472 max mem: 3531 +Test: [ 40/269] eta: 0:01:07 acc1_pos: 40.517 acc5_pos: 100.000 loss: 0.6636 (0.9485) acc1: 57.0312 (42.4098) acc5: 100.0000 (100.0000) acc1_neg: 69.531 acc5_neg: 100.000 time: 0.1999 data: 0.1048 max mem: 3531 +Test: [ 50/269] eta: 0:01:00 acc1_pos: 40.517 acc5_pos: 100.000 loss: 0.5330 (0.8688) acc1: 72.3958 (48.3915) acc5: 100.0000 (100.0000) acc1_neg: 63.802 acc5_neg: 100.000 time: 0.2002 data: 0.1071 max mem: 3531 +Test: [ 60/269] eta: 0:00:54 acc1_pos: 40.517 acc5_pos: 100.000 loss: 0.5459 (0.8224) acc1: 71.8750 (51.4088) acc5: 100.0000 (100.0000) acc1_neg: 78.646 acc5_neg: 100.000 time: 0.1949 data: 0.1007 max mem: 3531 +Test: [ 70/269] eta: 0:00:49 acc1_pos: 40.517 acc5_pos: 100.000 loss: 0.4717 (0.7656) acc1: 76.0417 (55.9272) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.1806 data: 0.0847 max mem: 3531 +Test: [ 80/269] eta: 0:00:45 acc1_pos: 40.517 acc5_pos: 100.000 loss: 0.3937 (0.7149) acc1: 84.1146 (59.9087) acc5: 100.0000 (100.0000) acc1_neg: 97.917 acc5_neg: 100.000 time: 0.1667 data: 0.0705 max mem: 3531 +Test: [ 90/269] eta: 0:00:42 acc1_pos: 40.517 acc5_pos: 100.000 loss: 0.3490 (0.6769) acc1: 89.0625 (63.2269) acc5: 100.0000 (100.0000) acc1_neg: 91.667 acc5_neg: 100.000 time: 0.1890 data: 0.0942 max mem: 3531 +Test: [100/269] eta: 0:00:39 acc1_pos: 40.517 acc5_pos: 100.000 loss: 0.3502 (0.6442) acc1: 90.3646 (65.8932) acc5: 100.0000 (100.0000) acc1_neg: 87.240 acc5_neg: 100.000 time: 0.2081 data: 0.1155 max mem: 3531 +Test: [110/269] eta: 0:00:36 acc1_pos: 40.517 acc5_pos: 100.000 loss: 0.3678 (0.6254) acc1: 89.5833 (67.6614) acc5: 100.0000 (100.0000) acc1_neg: 90.885 acc5_neg: 100.000 time: 0.2070 data: 0.1122 max mem: 3531 +Test: [120/269] eta: 0:00:33 acc1_pos: 40.517 acc5_pos: 100.000 loss: 0.4405 (0.6105) acc1: 82.2917 (68.7909) acc5: 100.0000 (100.0000) acc1_neg: 74.740 acc5_neg: 100.000 time: 0.1971 data: 0.1003 max mem: 3531 +Test: [130/269] eta: 0:00:31 acc1_pos: 40.517 acc5_pos: 100.000 loss: 0.4405 (0.6019) acc1: 80.2083 (69.3981) acc5: 100.0000 (100.0000) acc1_neg: 80.208 acc5_neg: 100.000 time: 0.1980 data: 0.1042 max mem: 3531 +Test: [140/269] eta: 0:00:29 acc1_pos: 40.517 acc5_pos: 100.000 loss: 0.4290 (0.5878) acc1: 80.2083 (70.3162) acc5: 100.0000 (100.0000) acc1_neg: 81.771 acc5_neg: 100.000 time: 0.2194 data: 0.1262 max mem: 3531 +Test: [150/269] eta: 0:00:26 acc1_pos: 40.517 acc5_pos: 100.000 loss: 0.3346 (0.5691) acc1: 88.2812 (71.6870) acc5: 100.0000 (100.0000) acc1_neg: 96.615 acc5_neg: 100.000 time: 0.2162 data: 0.1227 max mem: 3531 +Test: [160/269] eta: 0:00:24 acc1_pos: 40.517 acc5_pos: 100.000 loss: 0.3117 (0.5545) acc1: 90.3646 (72.7937) acc5: 100.0000 (100.0000) acc1_neg: 85.938 acc5_neg: 100.000 time: 0.1958 data: 0.1044 max mem: 3531 +Test: [170/269] eta: 0:00:21 acc1_pos: 40.517 acc5_pos: 100.000 loss: 0.3216 (0.5437) acc1: 90.3646 (73.6842) acc5: 100.0000 (100.0000) acc1_neg: 95.833 acc5_neg: 100.000 time: 0.2034 data: 0.1104 max mem: 3531 +Test: [180/269] eta: 0:00:19 acc1_pos: 40.517 acc5_pos: 100.000 loss: 0.3678 (0.5343) acc1: 90.6250 (74.5065) acc5: 100.0000 (100.0000) acc1_neg: 79.948 acc5_neg: 100.000 time: 0.2175 data: 0.1225 max mem: 3531 +Test: [190/269] eta: 0:00:17 acc1_pos: 40.517 acc5_pos: 100.000 loss: 0.3989 (0.5321) acc1: 86.4583 (74.5828) acc5: 100.0000 (100.0000) acc1_neg: 93.490 acc5_neg: 100.000 time: 0.2048 data: 0.1108 max mem: 3531 +Test: [200/269] eta: 0:00:15 acc1_pos: 40.517 acc5_pos: 100.000 loss: 0.3926 (0.5255) acc1: 80.2083 (75.0687) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.1998 data: 0.1083 max mem: 3531 +Test: [210/269] eta: 0:00:12 acc1_pos: 40.517 acc5_pos: 100.000 loss: 0.3926 (0.5237) acc1: 83.5938 (75.1086) acc5: 100.0000 (100.0000) acc1_neg: 76.562 acc5_neg: 100.000 time: 0.1923 data: 0.1021 max mem: 3532 +Test: [220/269] eta: 0:00:10 acc1_pos: 40.517 acc5_pos: 100.000 loss: 0.4553 (0.5184) acc1: 76.8229 (75.3064) acc5: 100.0000 (100.0000) acc1_neg: 97.135 acc5_neg: 100.000 time: 0.1800 data: 0.0875 max mem: 3532 +Test: [230/269] eta: 0:00:08 acc1_pos: 40.517 acc5_pos: 100.000 loss: 0.2181 (0.5074) acc1: 94.5312 (76.1048) acc5: 100.0000 (100.0000) acc1_neg: 84.115 acc5_neg: 100.000 time: 0.1851 data: 0.0911 max mem: 3532 +Test: [240/269] eta: 0:00:06 acc1_pos: 40.517 acc5_pos: 100.000 loss: 0.4467 (0.5126) acc1: 80.4688 (75.6030) acc5: 100.0000 (100.0000) acc1_neg: 47.656 acc5_neg: 100.000 time: 0.1802 data: 0.0882 max mem: 3532 +Test: [250/269] eta: 0:00:03 acc1_pos: 40.517 acc5_pos: 100.000 loss: 0.5403 (0.5111) acc1: 66.6667 (75.6817) acc5: 100.0000 (100.0000) acc1_neg: 62.760 acc5_neg: 100.000 time: 0.1566 data: 0.0662 max mem: 3532 +Test: [260/269] eta: 0:00:01 acc1_pos: 40.517 acc5_pos: 100.000 loss: 0.5403 (0.5171) acc1: 68.2292 (75.0918) acc5: 100.0000 (100.0000) acc1_neg: 83.594 acc5_neg: 100.000 time: 0.1313 data: 0.0404 max mem: 3532 +Test: [268/269] eta: 0:00:00 acc1_pos: 40.517 acc5_pos: 100.000 loss: 0.5381 (0.5158) acc1: 68.2292 (75.2432) acc5: 100.0000 (100.0000) acc1_neg: 87.139 acc5_neg: 100.000 time: 0.1263 data: 0.0319 max mem: 3532 +Test: Total time: 0:00:55 (0.2048 s / it) +* Overall Acc@1 75.243 Acc@5 100.000 + target == real Acc@1 32.111 Acc@5 100.000 + target == fake Acc@1 81.130 Acc@5 100.000 + loss 0.516 + PR-AUC 0.9109 ROC-AUC 0.6089 EER 41.84% (thr=0.6689) +Accuracy of the network on the 103293 test images: 75.2% +[rank0]:[W725 18:36:50.640024307 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +[2026-07-25T18:36:57Z] eval_done tag=fixed500k_real_n5_seed4_resampled +[2026-07-25T18:36:57Z] eval_wait tag=fixed500k_fake_n11_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed4_resampled +[2026-07-25T18:37:57Z] eval_wait tag=fixed500k_fake_n11_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed4_resampled +[2026-07-25T18:38:57Z] eval_wait tag=fixed500k_fake_n11_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed4_resampled +[2026-07-25T18:39:57Z] eval_wait tag=fixed500k_fake_n11_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed4_resampled +[2026-07-25T18:40:58Z] eval_wait tag=fixed500k_fake_n11_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed4_resampled +[2026-07-25T18:41:58Z] eval_wait tag=fixed500k_fake_n11_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed4_resampled +[2026-07-25T18:42:58Z] eval_wait tag=fixed500k_fake_n11_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed4_resampled +[2026-07-25T18:43:58Z] eval_wait tag=fixed500k_fake_n11_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed4_resampled +[2026-07-25T18:44:58Z] eval_wait tag=fixed500k_fake_n11_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed4_resampled +[2026-07-25T18:45:58Z] eval_wait tag=fixed500k_fake_n11_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed4_resampled +[2026-07-25T18:46:59Z] eval_wait tag=fixed500k_fake_n11_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed4_resampled +[2026-07-25T18:47:59Z] eval_wait tag=fixed500k_fake_n11_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed4_resampled +[2026-07-25T18:48:59Z] eval_wait tag=fixed500k_fake_n11_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed4_resampled +[2026-07-25T18:49:59Z] eval_wait tag=fixed500k_fake_n11_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed4_resampled +[2026-07-25T18:50:59Z] eval_wait tag=fixed500k_fake_n11_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed4_resampled +[2026-07-25T18:51:59Z] eval_wait tag=fixed500k_fake_n11_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed4_resampled +[2026-07-25T18:53:00Z] eval_wait tag=fixed500k_fake_n11_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed4_resampled +[2026-07-25T18:54:00Z] eval_wait tag=fixed500k_fake_n11_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed4_resampled +[2026-07-25T18:55:00Z] eval_wait tag=fixed500k_fake_n11_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed4_resampled +[2026-07-25T18:56:00Z] eval_wait tag=fixed500k_fake_n11_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed4_resampled +[2026-07-25T18:57:00Z] eval_wait tag=fixed500k_fake_n11_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed4_resampled +[2026-07-25T18:58:00Z] eval_wait tag=fixed500k_fake_n11_seed4_resampled reason=locked output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed4_resampled +[2026-07-25T18:59:01Z] eval_skip tag=fixed500k_fake_n11_seed4_resampled reason=complete_after_wait output_root=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/eval/fixed500k_resampled_seed1_4_benchmark_eval_20260725_112559/fixed500k_fake_n11_seed4_resampled +eval_queue_done_utc=2026-07-25T18:59:01Z diff --git a/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_host_done_20260725_112559/eval_queue_n116-040-168.pid b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_host_done_20260725_112559/eval_queue_n116-040-168.pid new file mode 100644 index 0000000000000000000000000000000000000000..cdadcb6f6b957aa5d2342d89f66e9c9ab783698a --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/fixed500k_resampled_eval_host_done_20260725_112559/eval_queue_n116-040-168.pid @@ -0,0 +1 @@ +2554935 diff --git a/ScaleDF_rebuttal/_host_logs/magface_fake_transfer_20260726_124043.log b/ScaleDF_rebuttal/_host_logs/magface_fake_transfer_20260726_124043.log new file mode 100644 index 0000000000000000000000000000000000000000..807e9491b1e1ff93276eb15319bbb42612a47abf --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/magface_fake_transfer_20260726_124043.log @@ -0,0 +1,2 @@ +tag=magface_fake_transfer_20260726_124043 +source=n116-040-168 src=/dev/shm/scaledf_bind_storage/ScaleDF/train list=/tmp/magface_fake_transfer_20260726_124043_dirs.txt dirs=91 diff --git a/ScaleDF_rebuttal/_host_logs/magface_fake_transfer_20260726_1242_to_n116_041_181.log b/ScaleDF_rebuttal/_host_logs/magface_fake_transfer_20260726_1242_to_n116_041_181.log new file mode 100644 index 0000000000000000000000000000000000000000..d795c4f6c03e4be981281d5fdbfaca3103ee7e1d --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/magface_fake_transfer_20260726_1242_to_n116_041_181.log @@ -0,0 +1,14 @@ +tag=magface_fake_transfer_20260726_1242_to_n116_041_181 +source_host=n116-040-168 +src=/dev/shm/scaledf_bind_storage/ScaleDF/train +dst_host=fdbd:dccd:cdc1:1368:0:1f1:: +dst_port=11062 +dst=/dev/shm/scaledf_bind_storage/ScaleDF/train +dirs=91 +list=/tmp/magface_fake_transfer_20260726_1242_to_n116_041_181_dirs.txt +pid=3244044 +/home/tiger/.profile: line 29: /run/bin/env: No such file or directory +/home/tiger/.profile: line 31: /tmp/m9/bin/env: No such file or directory +started_utc=2026-07-26T12:41:08Z +finished_utc=2026-07-26T12:53:20Z +transfer_status=0 diff --git a/ScaleDF_rebuttal/_host_logs/magface_quality_auto_launch_watch_20260726_1248.log b/ScaleDF_rebuttal/_host_logs/magface_quality_auto_launch_watch_20260726_1248.log new file mode 100644 index 0000000000000000000000000000000000000000..313da0b29253805ce8d2c145e16f9399fe11f1da --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/magface_quality_auto_launch_watch_20260726_1248.log @@ -0,0 +1,7 @@ +watch_started_utc=2026-07-26T12:47:30Z +transfer_pid=3244044 +transfer_log=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/_host_logs/magface_fake_transfer_20260726_1242_to_n116_041_181.log +dst=fdbd:dccd:cdc1:1368:0:1f1:::11062 +tag=20260726_1248_magface_fake_quality_full +transfer_pid_exited_utc=2026-07-26T12:53:31Z +/tmp/watch_transfer_then_launch_magface.sh: line 29: $!: unbound variable diff --git a/ScaleDF_rebuttal/_host_logs/rsync_scaledf_remaining_20260726_1308_n116040168_to_n116041181.log b/ScaleDF_rebuttal/_host_logs/rsync_scaledf_remaining_20260726_1308_n116040168_to_n116041181.log new file mode 100644 index 0000000000000000000000000000000000000000..70c839d5411e2ba4c5c48cf9a587b7610b92e1c7 --- /dev/null +++ b/ScaleDF_rebuttal/_host_logs/rsync_scaledf_remaining_20260726_1308_n116040168_to_n116041181.log @@ -0,0 +1,7 @@ +started_utc=2026-07-26T13:08:10Z +src_host=n116-040-168 +src=/dev/shm/scaledf_bind_storage/ +dest=tiger@fdbd:dccd:cdc1:1368:0:1f1:::/dev/shm/scaledf_bind_storage/ +ssh: Could not resolve hostname fdbd: Name or service not known +rsync: connection unexpectedly closed (0 bytes received so far) [sender] +rsync error: unexplained error (code 255) at io.c(232) [sender=3.2.7] diff --git a/ScaleDF_rebuttal/collect_rebuttal_eval_results.py b/ScaleDF_rebuttal/collect_rebuttal_eval_results.py new file mode 100644 index 0000000000000000000000000000000000000000..8b3efb65dc1a76dd9e38506b53ab2732cef26dd6 --- /dev/null +++ b/ScaleDF_rebuttal/collect_rebuttal_eval_results.py @@ -0,0 +1,143 @@ +#!/usr/bin/env python3 +"""Collect ScaleDF rebuttal eval metrics from persistent eval logs.""" + +from __future__ import annotations + +import argparse +import csv +import re +from pathlib import Path + + +OUTPUT_RE = re.compile(r"output_dir='([^']+)'") +DATA_RE = re.compile(r"data_path='([^']+)'") +CKPT_RE = re.compile(r"^checkpoint=(.+)$") +MODEL_RE = re.compile(r"^model_name=(.+)$") +INPUT_RE = re.compile(r"^input_size=(.+)$") +OVERALL_RE = re.compile(r"\* Overall\s+Acc@1\s+([0-9.]+)\s+Acc@5\s+([0-9.]+)") +PRAUC_RE = re.compile(r"PR-AUC\s+([0-9.]+)\s+ROC-AUC\s+([0-9.]+)\s+EER\s+([0-9.]+)%") +ACC_RE = re.compile(r"Accuracy of the network on the\s+([0-9]+)\s+test images:\s+([0-9.]+)%") + + +def infer_eval_root(path: Path, eval_root: Path) -> tuple[str, str]: + rel = path.relative_to(eval_root) + run = rel.parts[0] if rel.parts else "" + tag = "" + if len(rel.parts) >= 3 and rel.parts[2] == "eval.log": + tag = rel.parts[1] + elif len(rel.parts) >= 2 and rel.parts[1] == "eval.log": + tag = run + return run, tag + + +def normalize_dataset(output_dir: str, data_path: str) -> str: + if output_dir: + p = Path(output_dir) + if p.name == "scaledf_val": + return "ScaleDF_val" + if len(p.parts) >= 2 and p.parts[-2] == "benchmarks": + return p.name + if data_path: + return Path(data_path).name + return "" + + +def collect_log(path: Path, eval_root: Path) -> list[dict[str, str]]: + run, tag = infer_eval_root(path, eval_root) + rows = [] + checkpoint = "" + model = "" + input_size = "" + current_output = "" + current_data = "" + pending: dict[str, str] | None = None + + for raw in path.read_text(errors="replace").splitlines(): + line = raw.strip() + if m := CKPT_RE.match(line): + checkpoint = m.group(1) + if m := MODEL_RE.match(line): + model = m.group(1) + if m := INPUT_RE.match(line): + input_size = m.group(1) + if m := OUTPUT_RE.search(line): + current_output = m.group(1) + if m := DATA_RE.search(line): + current_data = m.group(1) + + if m := OVERALL_RE.search(line): + pending = { + "run": run, + "tag": tag, + "log": str(path), + "dataset": normalize_dataset(current_output, current_data), + "checkpoint": checkpoint, + "model": model, + "input_size": input_size, + "acc1": m.group(1), + "acc5": m.group(2), + "pr_auc": "", + "roc_auc": "", + "eer_percent": "", + "test_images": "", + } + continue + + if pending and (m := PRAUC_RE.search(line)): + pending["pr_auc"] = m.group(1) + pending["roc_auc"] = m.group(2) + pending["eer_percent"] = m.group(3) + continue + + if pending and (m := ACC_RE.search(line)): + pending["test_images"] = m.group(1) + rows.append(pending) + pending = None + + if pending: + rows.append(pending) + return rows + + +def main() -> None: + parser = argparse.ArgumentParser() + parser.add_argument("--eval-root", type=Path, required=True) + parser.add_argument("--out", type=Path, required=True) + args = parser.parse_args() + + rows = [] + seen = set() + for log in sorted(args.eval_root.rglob("eval.log")): + for row in collect_log(log, args.eval_root): + key = (row["run"], row["tag"], row["dataset"], row["checkpoint"], row["model"], row["acc1"]) + if key in seen: + continue + seen.add(key) + rows.append(row) + + args.out.parent.mkdir(parents=True, exist_ok=True) + fields = [ + "run", + "tag", + "dataset", + "test_images", + "acc1", + "acc5", + "pr_auc", + "roc_auc", + "eer_percent", + "model", + "input_size", + "checkpoint", + "log", + ] + with args.out.open("w", newline="", encoding="utf-8") as f: + writer = csv.DictWriter(f, fieldnames=fields, delimiter="\t") + writer.writeheader() + writer.writerows(rows) + print(f"rows={len(rows)}") + print(args.out) + + +if __name__ == "__main__": + main() diff --git a/ScaleDF_rebuttal/datasets_cross_2direct_balance_realscale_perturb_v4_NRealFake_noqcpt.py b/ScaleDF_rebuttal/datasets_cross_2direct_balance_realscale_perturb_v4_NRealFake_noqcpt.py new file mode 100644 index 0000000000000000000000000000000000000000..2bb29bf31544ab5f69dad0aafd78343b1958b310 --- /dev/null +++ b/ScaleDF_rebuttal/datasets_cross_2direct_balance_realscale_perturb_v4_NRealFake_noqcpt.py @@ -0,0 +1,250 @@ +# Copyright (c) 2015-present, Facebook, Inc. +# All rights reserved. +import os +import json +import pickle +import random + +from torchvision import datasets, transforms +from torchvision.datasets.folder import ImageFolder, default_loader + +from timm.data.constants import IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STD +from timm.data import create_transform +from perturb import * + +augmentations = [ + RandomResizeCrop(), Color_jitter(), RandomBlur(), RandomPixelization(), + RandomRotate(), GrayScale(), RandomPad(), RandomAddNoise(), + VertFlip(), HoriFlip(), RandomPerspectiveTransform(), + RandomChangeChannel(), RandomEncodingQuality(), RandomSharpen(), + RandomSkew(), RandomShufflePixels(), Xraylize(), Glass(), + OpticalDistortion(), Solarize(), ZoomBlur(), Elastic(), FancyPCA(), + GridDistortion(), ISONoise(), MultiplicativeNoise(), Posterize(), + RandomGamma(), Spatter(), Binary(), +] +''' +class RandomPairAug: + def __init__(self, aug_list): + self.aug_list = aug_list + + def __call__(self, img): + + score = random.random() + + if score < 0.1: + img = img + + elif score < 0.5: + a1 = random.sample(self.aug_list, 1)[0] + img = a1(img) + + elif score < 0.75: + a1, a2 = random.sample(self.aug_list, 2) + img = a1(img) + img = a2(img) + + else: + a1, a2, a3 = random.sample(self.aug_list, 3) + img = a1(img) + img = a2(img) + img = a3(img) + + return img + +''' +class RandomPairAug: + def __init__(self, aug_list): + self.aug_list = aug_list + + def __call__(self, img): + + score = random.random() + + if score < 0.5: + img = img + elif score < 0.75: + a1, a2 = random.sample(self.aug_list, 2) + img = a1(img) + img = a2(img) + else: + a1 = random.sample(self.aug_list, 1)[0] + img = a1(img) + + return img + +class INatDataset(ImageFolder): + def __init__(self, root, train=True, year=2018, transform=None, target_transform=None, + category='name', loader=default_loader): + self.transform = transform + self.loader = loader + self.target_transform = target_transform + self.year = year + # assert category in ['kingdom','phylum','class','order','supercategory','family','genus','name'] + path_json = os.path.join(root, f'{"train" if train else "val"}{year}.json') + with open(path_json) as json_file: + data = json.load(json_file) + + with open(os.path.join(root, 'categories.json')) as json_file: + data_catg = json.load(json_file) + + path_json_for_targeter = os.path.join(root, f"train{year}.json") + + with open(path_json_for_targeter) as json_file: + data_for_targeter = json.load(json_file) + + targeter = {} + indexer = 0 + for elem in data_for_targeter['annotations']: + king = [] + king.append(data_catg[int(elem['category_id'])][category]) + if king[0] not in targeter.keys(): + targeter[king[0]] = indexer + indexer += 1 + self.nb_classes = len(targeter) + + self.samples = [] + for elem in data['images']: + cut = elem['file_name'].split('/') + target_current = int(cut[2]) + path_current = os.path.join(root, cut[0], cut[2], cut[3]) + + categors = data_catg[target_current] + target_current_true = targeter[categors[category]] + self.samples.append((path_current, target_current_true)) + + # __getitem__ and __len__ inherited from ImageFolder + + +def build_dataset(is_train, args): + transform = build_transform(is_train, args) + + if args.data_set == 'CIFAR': + dataset = datasets.CIFAR100(args.data_path, train=is_train, transform=transform) + nb_classes = 100 + elif args.data_set == 'IMNET': + root = os.path.join(args.data_path, 'train' if is_train else 'val') + dataset = datasets.ImageFolder(root, transform=transform) + nb_classes = 1000 + elif 'ScaleDF' in args.data_set: + + if is_train: + root = os.path.join(args.data_path, 'train') + dataset = datasets.ImageFolder(root, transform=transform) + with open(args.select_models, 'rb') as f: + classes_train = pickle.load(f) + + old2new = {dataset.class_to_idx[c]: i for i, c in enumerate(classes_train)} + + #filtered = [(p, old2new[t]) for p, t in dataset.samples if t in old2new] + with open(args.select_images, 'rb') as f: + filtered = pickle.load(f) + + orig_len = len(filtered) + + num_real_images = len([i for i in filtered if i[1]<46]) + + filtered_real = filtered[:num_real_images] + filtered_fake = filtered[num_real_images:] + + if len(filtered_fake) > len(filtered_real): + filtered_real_extend = random.choices(filtered_real, k=len(filtered_fake)) + filtered_extend = filtered_real_extend + filtered_fake + else: + filtered_fake_extend = random.choices(filtered_fake, k=len(filtered_real)) + filtered_extend = filtered_real + filtered_fake_extend + + + + + + ''' + filtered_1_10 = filtered[:100_0000] + for i in range(88): + filtered_1_10 = filtered_1_10 + random.sample(filtered[100_0000 + i*10_0000 : 100_0000 + (i+1)*10_0000], 1_000) + filtered = filtered_1_10 + ''' + + #filtered = random.sample(filtered[:100_0000], 1_0000) + filtered[100_0000:] + + + + + dataset.samples = filtered_extend + dataset.targets = [t for _, t in filtered_extend] + dataset.classes = classes_train + dataset.class_to_idx = {c: i for i, c in enumerate(classes_train)} + + nb_classes = len(classes_train) + + return orig_len, dataset, nb_classes + + else: + root = os.path.join(args.data_path, 'val') + dataset = datasets.ImageFolder(root, transform=transform) + with open('scaledf_v3_val.pkl', 'rb') as f: + classes_test = pickle.load(f) + old2new = {dataset.class_to_idx[c]: i for i, c in enumerate(classes_test)} + + filtered = [(p, old2new[t]) for p, t in dataset.samples if t in old2new] + + dataset.samples = filtered + dataset.targets = [t for _, t in filtered] + dataset.classes = classes_test + dataset.class_to_idx = {c: i for i, c in enumerate(classes_test)} + + nb_classes = len(classes_test) + + return dataset, nb_classes + + +def build_transform(is_train, args): + resize_im = args.input_size > 32 + if is_train: + base_tf = create_transform( + input_size=args.input_size, + is_training=True, + color_jitter=args.color_jitter, + auto_augment=args.aa, + interpolation=args.train_interpolation, + re_prob=args.reprob, + re_mode=args.remode, + re_count=args.recount, + ) + + if not resize_im: + base_tf.transforms[0] = transforms.RandomCrop( + args.input_size, padding=4 + ) + + + return base_tf + ''' + # this should always dispatch to transforms_imagenet_train + transform = create_transform( + input_size=args.input_size, + is_training=True, + color_jitter=args.color_jitter, + auto_augment=args.aa, + interpolation=args.train_interpolation, + re_prob=args.reprob, + re_mode=args.remode, + re_count=args.recount, + ) + if not resize_im: + # replace RandomResizedCropAndInterpolation with + # RandomCrop + transform.transforms[0] = transforms.RandomCrop( + args.input_size, padding=4) + return transform + ''' + t = [] + if resize_im: + size = int(args.input_size / args.eval_crop_ratio) + t.append( + transforms.Resize(size, interpolation=3), # to maintain same ratio w.r.t. 224 images + ) + t.append(transforms.CenterCrop(args.input_size)) + + t.append(transforms.ToTensor()) + t.append(transforms.Normalize(IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STD)) + return transforms.Compose(t) diff --git a/ScaleDF_rebuttal/distributed_nccl_bandwidth.py b/ScaleDF_rebuttal/distributed_nccl_bandwidth.py new file mode 100644 index 0000000000000000000000000000000000000000..e810f0465645ee014cd57f32819f354b9058e1a9 --- /dev/null +++ b/ScaleDF_rebuttal/distributed_nccl_bandwidth.py @@ -0,0 +1,61 @@ +#!/usr/bin/env python3 +import os +import socket +import sys +import time + +import torch +import torch.distributed as dist + + +def main() -> int: + backend = os.environ.get("BENCH_BACKEND", "nccl") + local_rank = int(os.environ["LOCAL_RANK"]) + rank = int(os.environ["RANK"]) + world_size = int(os.environ["WORLD_SIZE"]) + mb = int(os.environ.get("BENCH_MB", "256")) + warmup = int(os.environ.get("BENCH_WARMUP", "1")) + iters = int(os.environ.get("BENCH_ITERS", "3")) + + torch.cuda.set_device(local_rank) + dist.init_process_group(backend=backend) + device = torch.device("cuda", local_rank) + numel = mb * 1024 * 1024 // 4 + tensor = torch.ones(numel, device=device, dtype=torch.float32) + + for _ in range(warmup): + dist.all_reduce(tensor) + torch.cuda.synchronize() + + times = [] + for _ in range(iters): + dist.barrier() + torch.cuda.synchronize() + start = time.time() + dist.all_reduce(tensor) + torch.cuda.synchronize() + times.append(time.time() - start) + + avg = sum(times) / len(times) + # One all-reduce moves roughly 2*(world_size-1)/world_size tensor bytes per rank. + bytes_per_rank = tensor.numel() * tensor.element_size() + alg_gbps = bytes_per_rank * 8 / avg / 1e9 + bus_gbps = alg_gbps * (2 * (world_size - 1) / world_size) + if rank == 0: + print( + "NCCL_BENCH " + f"host={socket.gethostname()} world={world_size} mb={mb} " + f"iters={iters} avg_s={avg:.6f} " + f"algbw_gbps={alg_gbps:.2f} busbw_gbps={bus_gbps:.2f}", + flush=True, + ) + dist.destroy_process_group() + return 0 + + +if __name__ == "__main__": + try: + raise SystemExit(main()) + except Exception as exc: + print(f"NCCL_BENCH_FAILED: {type(exc).__name__}: {exc}", file=sys.stderr, flush=True) + raise diff --git a/ScaleDF_rebuttal/distributed_nccl_smoke.py b/ScaleDF_rebuttal/distributed_nccl_smoke.py new file mode 100644 index 0000000000000000000000000000000000000000..eeee0d7efdce5cde9c06860e123b17a8afcb91ba --- /dev/null +++ b/ScaleDF_rebuttal/distributed_nccl_smoke.py @@ -0,0 +1,58 @@ +#!/usr/bin/env python3 +"""Tiny multi-node NCCL smoke test for ScaleDF cluster jobs.""" + +from __future__ import annotations + +import datetime as dt +import os +import socket +import sys +import time + +import torch +import torch.distributed as dist + + +def main() -> None: + rank = int(os.environ["RANK"]) + local_rank = int(os.environ["LOCAL_RANK"]) + world_size = int(os.environ["WORLD_SIZE"]) + host = socket.gethostname() + backend = os.environ.get("SMOKE_BACKEND", "nccl") + + if backend == "nccl": + torch.cuda.set_device(local_rank) + device = torch.device("cuda", local_rank) + else: + device = torch.device("cpu") + + started = dt.datetime.utcnow().isoformat(timespec="seconds") + "Z" + print( + f"start utc={started} host={host} rank={rank}/{world_size} local_rank={local_rank} " + f"backend={backend} master={os.environ.get('MASTER_ADDR')}:{os.environ.get('MASTER_PORT')}", + flush=True, + ) + t0 = time.time() + dist.init_process_group(backend=backend, timeout=dt.timedelta(seconds=120)) + tensor = torch.ones(1, device=device) * (rank + 1) + dist.all_reduce(tensor, op=dist.ReduceOp.SUM) + if backend == "nccl": + torch.cuda.synchronize() + expected = world_size * (world_size + 1) / 2 + value = float(tensor.item()) + print( + f"done host={host} rank={rank} value={value:.1f} expected={expected:.1f} " + f"elapsed={time.time() - t0:.3f}s", + flush=True, + ) + if abs(value - expected) > 1e-3: + raise SystemExit(f"bad all_reduce value: {value} expected {expected}") + dist.destroy_process_group() + + +if __name__ == "__main__": + try: + main() + except Exception as exc: + print(f"SMOKE_FAILED: {type(exc).__name__}: {exc}", file=sys.stderr, flush=True) + raise diff --git a/ScaleDF_rebuttal/launch_fixed500k_resampled_eval_completed_watch.sh b/ScaleDF_rebuttal/launch_fixed500k_resampled_eval_completed_watch.sh new file mode 100644 index 0000000000000000000000000000000000000000..8fb9954aaf890c48a7d6771ecb7c6dde1551fbc9 --- /dev/null +++ b/ScaleDF_rebuttal/launch_fixed500k_resampled_eval_completed_watch.sh @@ -0,0 +1,130 @@ +#!/usr/bin/env bash +set -euo pipefail + +if [[ $# -ne 1 ]]; then + echo "Usage: $0 " >&2 + exit 2 +fi + +TRAIN_TAG="$1" +PERSIST_ROOT="${PERSIST_ROOT:-/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal}" +TRAIN_LOG_DIR="${TRAIN_LOG_DIR:-$PERSIST_ROOT/_host_logs/fixed500k_resampled_seed1_4_${TRAIN_TAG}}" +EVAL_ROOT="${EVAL_ROOT:-$PERSIST_ROOT/eval/fixed500k_resampled_seed1_4_benchmark_eval_${TRAIN_TAG}}" +EVAL_LAUNCHER="${EVAL_LAUNCHER:-$PERSIST_ROOT/launch_rebuttal_eval_flex.sh}" +HOST_NAME="$(hostname)" + +export PERSIST_ROOT +export PACKAGE_ROOT="${PACKAGE_ROOT:-/tmp/ScaleDF_repro_20260724}" +export SCALEDF_DATA="${SCALEDF_DATA:-/dev/shm/scaledf_bind_storage/ScaleDF}" +export BENCHMARK_ROOT="${BENCHMARK_ROOT:-/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names}" +export SKIP_MISSING_BENCHMARKS="${SKIP_MISSING_BENCHMARKS:-1}" +export BATCH_PER_GPU="${BATCH_PER_GPU:-256}" +export NPROC_PER_NODE="${NPROC_PER_NODE:-8}" +export PYTHONUNBUFFERED="${PYTHONUNBUFFERED:-1}" + +log_status() { + echo "[$(date -u +%Y-%m-%dT%H:%M:%SZ)] $*" +} + +eval_complete() { + local eval_log="$1" + [[ -f "$eval_log" ]] && [[ "$(grep -c 'Accuracy of the network on' "$eval_log" || true)" -ge 7 ]] +} + +completed_manifest() { + { grep -h 'done family=' "$TRAIN_LOG_DIR"/train_queue_*.log 2>/dev/null || true; } \ + | sed -E 's/.*done family=([^ ]+) level=([^ ]+) seed=([^ ]+) run=([^ ]+).*/\1 \2 \3 \4/' \ + | awk '$1 == "real_domains" || $1 == "fake_methods"' \ + | sort -u +} + +target_complete_count() { + find "$EVAL_ROOT" -mindepth 2 -maxdepth 2 -name eval.log -type f 2>/dev/null \ + | while read -r log; do + if eval_complete "$log"; then + echo complete + fi + done \ + | wc -l +} + +run_eval() { + local family="$1" + local level="$2" + local seed="$3" + local run="$4" + local axis_short + local tag + local checkpoint="$PERSIST_ROOT/runs/$run/best_checkpoint.pth" + + case "$family" in + real_domains) axis_short=real ;; + fake_methods) axis_short=fake ;; + *) log_status "skip unknown family=$family run=$run"; return ;; + esac + + tag="fixed500k_${axis_short}_n${level}_seed${seed}_resampled" + local output_root="$EVAL_ROOT/$tag" + local eval_log="$output_root/eval.log" + local lock_dir="$output_root/.eval.lock" + + if [[ ! -f "$checkpoint" ]]; then + log_status "skip tag=$tag reason=missing_checkpoint checkpoint=$checkpoint" + return + fi + + if eval_complete "$eval_log"; then + log_status "eval_skip tag=$tag reason=complete output_root=$output_root" + return + fi + + mkdir -p "$output_root" + while ! mkdir "$lock_dir" 2>/dev/null; do + if eval_complete "$eval_log"; then + log_status "eval_skip tag=$tag reason=complete_after_wait output_root=$output_root" + return + fi + log_status "eval_wait tag=$tag reason=locked output_root=$output_root" + sleep 60 + done + + if eval_complete "$eval_log"; then + rmdir "$lock_dir" 2>/dev/null || true + log_status "eval_skip tag=$tag reason=complete_after_lock output_root=$output_root" + return + fi + + log_status "eval_start tag=$tag checkpoint=$checkpoint" + if OUTPUT_ROOT="$output_root" RUN_NAME="$tag" "$EVAL_LAUNCHER" "$checkpoint" deit_base_patch16_LS benchmarks; then + rmdir "$lock_dir" 2>/dev/null || true + log_status "eval_done tag=$tag" + else + rmdir "$lock_dir" 2>/dev/null || true + return 1 + fi +} + +log_status "early_eval_watch_started host=$HOST_NAME train_tag=$TRAIN_TAG" +log_status "train_log_dir=$TRAIN_LOG_DIR" +log_status "eval_root=$EVAL_ROOT" +mkdir -p "$EVAL_ROOT" + +while true; do + mapfile -t completed < <(completed_manifest) + queue_done=$({ grep -l 'queue_done_utc' "$TRAIN_LOG_DIR"/train_queue_*.log 2>/dev/null || true; } | wc -l) + eval_done="$(target_complete_count | tr -d ' ')" + log_status "scan completed_train=${#completed[@]}/48 eval_complete=${eval_done}/48 queue_done=${queue_done}/6" + + for row in "${completed[@]}"; do + read -r family level seed run <<<"$row" + run_eval "$family" "$level" "$seed" "$run" + done + + queue_done=$({ grep -l 'queue_done_utc' "$TRAIN_LOG_DIR"/train_queue_*.log 2>/dev/null || true; } | wc -l) + eval_done="$(target_complete_count | tr -d ' ')" + if [[ "$queue_done" -eq 6 && "$eval_done" -eq 48 ]]; then + log_status "early_eval_watch_done eval_complete=$eval_done/48" + break + fi + sleep 60 +done diff --git a/ScaleDF_rebuttal/launch_fixed500k_resampled_eval_host_queue.sh b/ScaleDF_rebuttal/launch_fixed500k_resampled_eval_host_queue.sh new file mode 100644 index 0000000000000000000000000000000000000000..bbd2ba772581ab5b040bc361c8267353a1990d2d --- /dev/null +++ b/ScaleDF_rebuttal/launch_fixed500k_resampled_eval_host_queue.sh @@ -0,0 +1,111 @@ +#!/usr/bin/env bash +set -euo pipefail + +if [[ $# -ne 4 ]]; then + echo "Usage: $0 " >&2 + exit 2 +fi + +TRAIN_TAG="$1" +EVAL_TAG="$2" +REAL_LEVEL="$3" +FAKE_LEVEL="$4" + +PERSIST_ROOT="${PERSIST_ROOT:-/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal}" +EVAL_LAUNCHER="${EVAL_LAUNCHER:-$PERSIST_ROOT/launch_rebuttal_eval_flex.sh}" +SEEDS_RAW="${SEEDS:-1 2 3 4}" +SEED_LABEL="${SEED_LABEL:-1_4}" +EVAL_ROOT="${EVAL_ROOT:-$PERSIST_ROOT/eval/fixed500k_resampled_seed${SEED_LABEL}_benchmark_eval_${EVAL_TAG}}" +HOST_NAME="$(hostname)" + +export PERSIST_ROOT +export PACKAGE_ROOT="${PACKAGE_ROOT:-/tmp/ScaleDF_repro_20260724}" +export SCALEDF_DATA="${SCALEDF_DATA:-/dev/shm/scaledf_bind_storage/ScaleDF}" +export BENCHMARK_ROOT="${BENCHMARK_ROOT:-/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names}" +export SKIP_MISSING_BENCHMARKS="${SKIP_MISSING_BENCHMARKS:-1}" +export BATCH_PER_GPU="${BATCH_PER_GPU:-256}" +export NPROC_PER_NODE="${NPROC_PER_NODE:-8}" +export PYTHONUNBUFFERED="${PYTHONUNBUFFERED:-1}" + +find_checkpoint() { + local family="$1" + local level="$2" + local seed="$3" + local pattern="control_fixed500000_${family}_n${level}_classseed${seed}_resampled_*_${TRAIN_TAG}" + local run_dir + run_dir="$(find "$PERSIST_ROOT/runs" -maxdepth 1 -type d -name "$pattern" | sort | tail -n 1)" + if [[ -z "$run_dir" ]]; then + echo "missing run dir for pattern=$pattern" >&2 + exit 3 + fi + if [[ ! -f "$run_dir/best_checkpoint.pth" ]]; then + echo "missing checkpoint: $run_dir/best_checkpoint.pth" >&2 + exit 4 + fi + printf "%s/best_checkpoint.pth" "$run_dir" +} + +eval_complete() { + local eval_log="$1" + [[ -f "$eval_log" ]] && [[ "$(grep -c 'Accuracy of the network on' "$eval_log" || true)" -ge 7 ]] +} + +run_eval() { + local axis_short="$1" + local family="$2" + local level="$3" + local seed="$4" + local checkpoint + checkpoint="$(find_checkpoint "$family" "$level" "$seed")" + local tag="fixed500k_${axis_short}_n${level}_seed${seed}_resampled" + local output_root="$EVAL_ROOT/$tag" + local eval_log="$output_root/eval.log" + local lock_dir="$output_root/.eval.lock" + + if eval_complete "$eval_log"; then + echo "[$(date -u +%Y-%m-%dT%H:%M:%SZ)] eval_skip tag=$tag reason=complete output_root=$output_root" + return + fi + + mkdir -p "$output_root" + while ! mkdir "$lock_dir" 2>/dev/null; do + if eval_complete "$eval_log"; then + echo "[$(date -u +%Y-%m-%dT%H:%M:%SZ)] eval_skip tag=$tag reason=complete_after_wait output_root=$output_root" + return + fi + echo "[$(date -u +%Y-%m-%dT%H:%M:%SZ)] eval_wait tag=$tag reason=locked output_root=$output_root" + sleep 60 + done + + if eval_complete "$eval_log"; then + rmdir "$lock_dir" 2>/dev/null || true + echo "[$(date -u +%Y-%m-%dT%H:%M:%SZ)] eval_skip tag=$tag reason=complete_after_lock output_root=$output_root" + return + fi + + echo "[$(date -u +%Y-%m-%dT%H:%M:%SZ)] eval_start tag=$tag checkpoint=$checkpoint" + if OUTPUT_ROOT="$output_root" RUN_NAME="$tag" "$EVAL_LAUNCHER" "$checkpoint" deit_base_patch16_LS benchmarks; then + rmdir "$lock_dir" 2>/dev/null || true + else + rmdir "$lock_dir" 2>/dev/null || true + return 1 + fi + echo "[$(date -u +%Y-%m-%dT%H:%M:%SZ)] eval_done tag=$tag" +} + +echo "eval_queue_started_utc=$(date -u +%Y-%m-%dT%H:%M:%SZ)" +echo "hostname=$HOST_NAME" +echo "train_tag=$TRAIN_TAG" +echo "eval_tag=$EVAL_TAG" +echo "real_level=$REAL_LEVEL" +echo "fake_level=$FAKE_LEVEL" +echo "eval_root=$EVAL_ROOT" +echo "batch_per_gpu=$BATCH_PER_GPU" +echo "seeds=$SEEDS_RAW" + +for seed in $SEEDS_RAW; do + run_eval real real_domains "$REAL_LEVEL" "$seed" + run_eval fake fake_methods "$FAKE_LEVEL" "$seed" +done + +echo "eval_queue_done_utc=$(date -u +%Y-%m-%dT%H:%M:%SZ)" diff --git a/ScaleDF_rebuttal/launch_fixed500k_resampled_eval_on_host_train_done.sh b/ScaleDF_rebuttal/launch_fixed500k_resampled_eval_on_host_train_done.sh new file mode 100644 index 0000000000000000000000000000000000000000..ac586b50da51875c49126dfeda06ac71805516e8 --- /dev/null +++ b/ScaleDF_rebuttal/launch_fixed500k_resampled_eval_on_host_train_done.sh @@ -0,0 +1,64 @@ +#!/usr/bin/env bash +set -uo pipefail + +if [[ $# -ne 1 ]]; then + echo "Usage: $0 " >&2 + exit 2 +fi + +TRAIN_TAG="$1" +PERSIST_ROOT="${PERSIST_ROOT:-/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal}" +TRAIN_LOG_DIR="${TRAIN_LOG_DIR:-$PERSIST_ROOT/_host_logs/fixed500k_resampled_seed1_4_${TRAIN_TAG}}" +LOG_DIR="${LOG_DIR:-$PERSIST_ROOT/_host_logs/fixed500k_resampled_eval_host_done_${TRAIN_TAG}}" +REMOTE_SCRIPT="${REMOTE_SCRIPT:-$PERSIST_ROOT/launch_fixed500k_resampled_eval_host_queue.sh}" + +mkdir -p "$LOG_DIR" + +log_status() { + echo "[$(date -u +%Y-%m-%dT%H:%M:%SZ)] $*" +} + +trap 'code=$?; log_status "host_done_eval_watch_exit code=$code"' EXIT + +submit() { + local port="$1" + local host="$2" + local label="$3" + local real_level="$4" + local fake_level="$5" + local train_log="$TRAIN_LOG_DIR/train_queue_${label}.log" + local queue_log="$LOG_DIR/eval_queue_${label}.log" + local pid_file="$LOG_DIR/eval_queue_${label}.pid" + local marker="$LOG_DIR/eval_queue_${label}.submitted" + + [[ -f "$marker" ]] && return 0 + grep -q 'queue_done_utc' "$train_log" 2>/dev/null || return 0 + + log_status "submit_eval label=$label real_level=$real_level fake_level=$fake_level" + if ssh -p "$port" "tiger@$host" \ + "mkdir -p '$LOG_DIR'; nohup bash '$REMOTE_SCRIPT' '$TRAIN_TAG' '$TRAIN_TAG' '$real_level' '$fake_level' > '$queue_log' 2>&1 < /dev/null & echo \$! > '$pid_file'; echo submitted label='$label' pid=\$(cat '$pid_file') log='$queue_log'"; then + date -u +%Y-%m-%dT%H:%M:%SZ > "$marker" + else + log_status "submit_eval_failed label=$label" + fi +} + +log_status "host_done_eval_watch_started train_tag=$TRAIN_TAG train_log_dir=$TRAIN_LOG_DIR log_dir=$LOG_DIR" + +while true; do + submit 10338 'fdbd:dccd:cdc1:1368:0:e4::' n116-040-168 5 11 + submit 11208 'fdbd:dccd:cdc1:1368:0:1d6::' n116-041-153 8 16 + submit 9869 'fdbd:dccd:cdc1:1368:0:1f8::' n116-041-188 11 22 + submit 11986 'fdbd:dccd:cdc1:1302:0:8d::' n111-046-081 16 32 + submit 11693 'fdbd:dccd:cdc1:1368:0:153::' n116-041-023 23 45 + submit 9999 'fdbd:dccd:cdc1:1368:0:111::' n116-040-213 32 64 + + submitted=$(find "$LOG_DIR" -maxdepth 1 -name 'eval_queue_*.submitted' -type f 2>/dev/null | wc -l | tr -d ' ') + done_count=$({ grep -l 'eval_queue_done_utc' "$LOG_DIR"/eval_queue_*.log 2>/dev/null || true; } | wc -l | tr -d ' ') + log_status "scan submitted=$submitted/6 eval_done=$done_count/6" + if [[ "$submitted" -eq 6 && "$done_count" -eq 6 ]]; then + log_status "host_done_eval_watch_done" + break + fi + sleep 60 +done diff --git a/ScaleDF_rebuttal/launch_fixed500k_resampled_host_queue.sh b/ScaleDF_rebuttal/launch_fixed500k_resampled_host_queue.sh new file mode 100644 index 0000000000000000000000000000000000000000..cb0fb4448c04db64e9c80c0fc2dcf23932f3982a --- /dev/null +++ b/ScaleDF_rebuttal/launch_fixed500k_resampled_host_queue.sh @@ -0,0 +1,65 @@ +#!/usr/bin/env bash +set -euo pipefail + +if [[ $# -ne 3 ]]; then + echo "Usage: $0 " >&2 + exit 2 +fi + +REAL_LEVEL="$1" +FAKE_LEVEL="$2" +LAUNCH_TAG="$3" + +PERSIST_ROOT="${PERSIST_ROOT:-/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal}" +SELECTION_DIR="${SELECTION_DIR:-$PERSIST_ROOT/selections/fixed_budget_500k_resampled_20260725}" +TRAIN_LAUNCHER="${TRAIN_LAUNCHER:-$PERSIST_ROOT/launch_rebuttal_diversity_train.sh}" +PACKAGE_ROOT="${PACKAGE_ROOT:-/tmp/ScaleDF_repro_20260724}" +SCALEDF_DATA="${SCALEDF_DATA:-/dev/shm/scaledf_bind_storage/ScaleDF}" +HOST_NAME="$(hostname)" +SEEDS_RAW="${SEEDS:-1 2 3 4}" + +export PERSIST_ROOT PACKAGE_ROOT SCALEDF_DATA +export EPOCHS="${EPOCHS:-10}" +export BATCH_PER_GPU="${BATCH_PER_GPU:-256}" +export NPROC_PER_NODE="${NPROC_PER_NODE:-8}" +export PYTHONUNBUFFERED="${PYTHONUNBUFFERED:-1}" + +if [[ ! -e /raid/ScaleDF_v4/train ]]; then + mkdir -p /raid 2>/dev/null || sudo mkdir -p /raid + ln -sfn "$SCALEDF_DATA" /raid/ScaleDF_v4 2>/dev/null || sudo ln -sfn "$SCALEDF_DATA" /raid/ScaleDF_v4 +fi + +run_one() { + local family="$1" + local level="$2" + local seed="$3" + local selection="$SELECTION_DIR/scaledf_v4_fixed500000_${family}_n${level}_classseed${seed}_imgseed20260725_resampled.pkl" + local run_name="control_fixed500000_${family}_n${level}_classseed${seed}_resampled_${HOST_NAME}_${LAUNCH_TAG}" + + if [[ ! -f "$selection" ]]; then + echo "missing selection: $selection" >&2 + exit 3 + fi + + echo "[$(date -u +%Y-%m-%dT%H:%M:%SZ)] start family=$family level=$level seed=$seed run=$run_name" + SEED="$seed" "$TRAIN_LAUNCHER" nrealfake "$selection" "$run_name" + echo "[$(date -u +%Y-%m-%dT%H:%M:%SZ)] done family=$family level=$level seed=$seed run=$run_name" +} + +echo "queue_started_utc=$(date -u +%Y-%m-%dT%H:%M:%SZ)" +echo "hostname=$HOST_NAME" +echo "real_level=$REAL_LEVEL" +echo "fake_level=$FAKE_LEVEL" +echo "launch_tag=$LAUNCH_TAG" +echo "selection_dir=$SELECTION_DIR" +echo "epochs=$EPOCHS" +echo "batch_per_gpu=$BATCH_PER_GPU" +echo "nproc_per_node=$NPROC_PER_NODE" +echo "seeds=$SEEDS_RAW" + +for seed in $SEEDS_RAW; do + run_one real_domains "$REAL_LEVEL" "$seed" + run_one fake_methods "$FAKE_LEVEL" "$seed" +done + +echo "queue_done_utc=$(date -u +%Y-%m-%dT%H:%M:%SZ)" diff --git a/ScaleDF_rebuttal/launch_fixed500k_resampled_host_queue_resume.sh b/ScaleDF_rebuttal/launch_fixed500k_resampled_host_queue_resume.sh new file mode 100644 index 0000000000000000000000000000000000000000..c45b2ee3e2c994f6c99b2dd481b50d0d074063f6 --- /dev/null +++ b/ScaleDF_rebuttal/launch_fixed500k_resampled_host_queue_resume.sh @@ -0,0 +1,69 @@ +#!/usr/bin/env bash +set -euo pipefail + +if [[ $# -ne 3 ]]; then + echo "Usage: $0 " >&2 + exit 2 +fi + +REAL_LEVEL="$1" +FAKE_LEVEL="$2" +LAUNCH_TAG="$3" + +PERSIST_ROOT="${PERSIST_ROOT:-/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal}" +SELECTION_DIR="${SELECTION_DIR:-$PERSIST_ROOT/selections/fixed_budget_500k_resampled_20260725}" +TRAIN_LAUNCHER="${TRAIN_LAUNCHER:-$PERSIST_ROOT/launch_rebuttal_diversity_train.sh}" +PACKAGE_ROOT="${PACKAGE_ROOT:-/tmp/ScaleDF_repro_20260724}" +SCALEDF_DATA="${SCALEDF_DATA:-/dev/shm/scaledf_bind_storage/ScaleDF}" +HOST_NAME="$(hostname)" + +export PERSIST_ROOT PACKAGE_ROOT SCALEDF_DATA +export EPOCHS="${EPOCHS:-10}" +export BATCH_PER_GPU="${BATCH_PER_GPU:-256}" +export NPROC_PER_NODE="${NPROC_PER_NODE:-8}" +export PYTHONUNBUFFERED="${PYTHONUNBUFFERED:-1}" + +if [[ ! -e /raid/ScaleDF_v4/train ]]; then + mkdir -p /raid 2>/dev/null || sudo mkdir -p /raid + ln -sfn "$SCALEDF_DATA" /raid/ScaleDF_v4 2>/dev/null || sudo ln -sfn "$SCALEDF_DATA" /raid/ScaleDF_v4 +fi + +run_one() { + local family="$1" + local level="$2" + local seed="$3" + local selection="$SELECTION_DIR/scaledf_v4_fixed500000_${family}_n${level}_classseed${seed}_imgseed20260725_resampled.pkl" + local run_name="control_fixed500000_${family}_n${level}_classseed${seed}_resampled_${HOST_NAME}_${LAUNCH_TAG}" + local run_dir="$PERSIST_ROOT/runs/$run_name" + + if [[ ! -f "$selection" ]]; then + echo "missing selection: $selection" >&2 + exit 3 + fi + + if [[ -f "$run_dir/best_checkpoint.pth" ]]; then + echo "[$(date -u +%Y-%m-%dT%H:%M:%SZ)] skip_existing family=$family level=$level seed=$seed run=$run_name" + return + fi + + echo "[$(date -u +%Y-%m-%dT%H:%M:%SZ)] resume_start family=$family level=$level seed=$seed run=$run_name" + SEED="$seed" "$TRAIN_LAUNCHER" nrealfake "$selection" "$run_name" + echo "[$(date -u +%Y-%m-%dT%H:%M:%SZ)] resume_done family=$family level=$level seed=$seed run=$run_name" +} + +echo "queue_resume_started_utc=$(date -u +%Y-%m-%dT%H:%M:%SZ)" +echo "hostname=$HOST_NAME" +echo "real_level=$REAL_LEVEL" +echo "fake_level=$FAKE_LEVEL" +echo "launch_tag=$LAUNCH_TAG" +echo "selection_dir=$SELECTION_DIR" +echo "epochs=$EPOCHS" +echo "batch_per_gpu=$BATCH_PER_GPU" +echo "nproc_per_node=$NPROC_PER_NODE" + +for seed in 1 2 3 4; do + run_one real_domains "$REAL_LEVEL" "$seed" + run_one fake_methods "$FAKE_LEVEL" "$seed" +done + +echo "queue_done_utc=$(date -u +%Y-%m-%dT%H:%M:%SZ)" diff --git a/ScaleDF_rebuttal/launch_fixed_budget_queue.sh b/ScaleDF_rebuttal/launch_fixed_budget_queue.sh new file mode 100644 index 0000000000000000000000000000000000000000..d7a3ba65af5d6226dc5cfe1cef95398fb0b3023a --- /dev/null +++ b/ScaleDF_rebuttal/launch_fixed_budget_queue.sh @@ -0,0 +1,85 @@ +#!/usr/bin/env bash +set -euo pipefail + +if [[ $# -lt 3 ]]; then + cat >&2 <<'USAGE' +Usage: + launch_fixed_budget_queue.sh + +Environment: + PERSIST_ROOT=/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal + PACKAGE_ROOT=/tmp/ScaleDF_repro_20260724 + SCALEDF_DATA=/dev/shm/scaledf_bind_storage/ScaleDF + EPOCHS=20 + BATCH_PER_GPU=256 +USAGE + exit 1 +fi + +MANIFEST="$1" +FAMILY="$2" +CLASS_SEED="$3" + +ROOT="${PERSIST_ROOT:-/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal}" +PACKAGE_ROOT="${PACKAGE_ROOT:-/tmp/ScaleDF_repro_20260724}" +LAUNCHER="$ROOT/launch_rebuttal_diversity_train.sh" +HOST="$(hostname)" +STAMP="$(date -u +%Y%m%d_%H%M%S)" +QUEUE_DIR="$ROOT/queues/fixed500000_${FAMILY}_seed${CLASS_SEED}_queue_${HOST}_${STAMP}" +QUEUE_LOG="$QUEUE_DIR/queue.driver.log" +SUBSET="$QUEUE_DIR/manifest_subset.tsv" + +mkdir -p "$QUEUE_DIR" +echo $$ > "$QUEUE_DIR/driver.pid" + +if [[ ! -f "$MANIFEST" ]]; then + echo "manifest not found: $MANIFEST" >&2 + exit 2 +fi +if [[ ! -x "$LAUNCHER" ]]; then + echo "launcher not executable: $LAUNCHER" >&2 + exit 2 +fi + +awk -F'\t' -v family="$FAMILY" -v seed="$CLASS_SEED" ' + BEGIN { OFS = "\t" } + NR == 1 { + print "family", "level", "class_seed", "real_images", "fake_images", "status", "path" + next + } + $1 == family && $3 == seed && $6 == "ok" { print $1, $2, $3, $4, $5, $6, $8 } +' "$MANIFEST" > "$SUBSET" + +{ + echo "started_utc=$(date -u +%Y-%m-%dT%H:%M:%SZ)" + echo "hostname=$HOST" + echo "manifest=$MANIFEST" + echo "family=$FAMILY" + echo "class_seed=$CLASS_SEED" + echo "subset=$SUBSET" + echo "queue_dir=$QUEUE_DIR" + echo "epochs=${EPOCHS:-20}" + echo "batch_per_gpu=${BATCH_PER_GPU:-256}" +} | tee -a "$QUEUE_LOG" + +tail -n +2 "$SUBSET" | while IFS=$'\t' read -r family level class_seed real_images fake_images status selection; do + [[ -n "${selection:-}" ]] || continue + RUN_NAME="fixed500000_${family}_fs_n${level}_classseed${class_seed}_${HOST}_$(date -u +%Y%m%d_%H%M%S)" + echo "=== RUN $RUN_NAME $(date -u +%Y-%m-%dT%H:%M:%SZ) ===" | tee -a "$QUEUE_LOG" + echo "selection=$selection real_images=$real_images fake_images=$fake_images" | tee -a "$QUEUE_LOG" + if EPOCHS="${EPOCHS:-20}" \ + BATCH_PER_GPU="${BATCH_PER_GPU:-256}" \ + NPROC_PER_NODE="${NPROC_PER_NODE:-8}" \ + SEED="$class_seed" \ + PERSIST_ROOT="$ROOT" \ + PACKAGE_ROOT="$PACKAGE_ROOT" \ + SCALEDF_DATA="${SCALEDF_DATA:-/dev/shm/scaledf_bind_storage/ScaleDF}" \ + "$LAUNCHER" nrealfake "$selection" "$RUN_NAME" 2>&1 | tee -a "$QUEUE_LOG"; then + echo "=== DONE $RUN_NAME $(date -u +%Y-%m-%dT%H:%M:%SZ) ===" | tee -a "$QUEUE_LOG" + else + rc=$? + echo "=== FAILED rc=$rc $RUN_NAME $(date -u +%Y-%m-%dT%H:%M:%SZ) ===" | tee -a "$QUEUE_LOG" + fi +done + +echo "=== QUEUE DONE $(date -u +%Y-%m-%dT%H:%M:%SZ) ===" | tee -a "$QUEUE_LOG" diff --git a/ScaleDF_rebuttal/launch_magface_fake_quality_full.sh b/ScaleDF_rebuttal/launch_magface_fake_quality_full.sh new file mode 100644 index 0000000000000000000000000000000000000000..44ead0f45286ba1aa640e47cad5b507220a2c4a9 --- /dev/null +++ b/ScaleDF_rebuttal/launch_magface_fake_quality_full.sh @@ -0,0 +1,57 @@ +#!/usr/bin/env bash +set -euo pipefail + +TAG="${1:-20260726_magface_fake_quality}" +ROOT="${ROOT:-/dev/shm/scaledf_bind_storage/ScaleDF/train}" +BASE="${BASE:-/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/analysis/magface_fake_quality_${TAG}}" +SCRIPT="${SCRIPT:-/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/magface_quality_eval.py}" +MAGFACE_REPO="${MAGFACE_REPO:-/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/quality_eval/magface/MagFace}" +CHECKPOINT="${CHECKPOINT:-/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/quality_eval/magface/pretrained/magface_iresnet50_official.pth}" +ARCH="${ARCH:-iresnet50}" +BATCH_SIZE="${BATCH_SIZE:-1024}" +WORKERS="${WORKERS:-8}" +NPROC="${NPROC:-8}" + +SHARDS="$BASE/shards" +SCORES="$BASE/scores" +SUMMARY="$BASE/summary" +LOGS="$BASE/logs" +mkdir -p "$SHARDS" "$SCORES" "$SUMMARY" "$LOGS" + +echo "tag=$TAG" +echo "host=$(hostname)" +echo "root=$ROOT" +echo "base=$BASE" +echo "checkpoint=$CHECKPOINT" +echo "arch=$ARCH batch_size=$BATCH_SIZE workers=$WORKERS nproc=$NPROC" + +python3 "$SCRIPT" make-shards \ + --root "$ROOT" \ + --out-dir "$SHARDS" \ + --prefix scaledf_fake \ + --num-shards "$NPROC" \ + --print-freq 500000 \ + 2>&1 | tee "$LOGS/make_shards.log" + +torchrun --standalone --nproc_per_node="$NPROC" "$SCRIPT" score \ + --shard-dir "$SHARDS" \ + --out-dir "$SCORES" \ + --prefix scaledf_fake \ + --out-prefix magface_fake_quality \ + --magface-repo "$MAGFACE_REPO" \ + --checkpoint "$CHECKPOINT" \ + --arch "$ARCH" \ + --batch-size "$BATCH_SIZE" \ + --workers "$WORKERS" \ + --print-freq 200000 \ + --amp \ + 2>&1 | tee "$LOGS/score.log" + +python3 "$SCRIPT" summarize \ + --score-dir "$SCORES" \ + --out-dir "$SUMMARY" \ + --prefix magface_fake_quality \ + 2>&1 | tee "$LOGS/summarize.log" + +echo "done_utc=$(date -u +%Y-%m-%dT%H:%M:%SZ)" +echo "summary=$SUMMARY" diff --git a/ScaleDF_rebuttal/launch_modelscope_scaling_category_stream_queue.sh b/ScaleDF_rebuttal/launch_modelscope_scaling_category_stream_queue.sh new file mode 100644 index 0000000000000000000000000000000000000000..ce3894d52106ec873ddfe664bbf3dbde37bfa5d9 --- /dev/null +++ b/ScaleDF_rebuttal/launch_modelscope_scaling_category_stream_queue.sh @@ -0,0 +1,204 @@ +#!/usr/bin/env bash +set -euo pipefail + +PERSIST_ROOT="${PERSIST_ROOT:-/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal}" +REPO_URL="${REPO_URL:-https://modelscope.cn/datasets/wangwenhao0716/lyh.git}" +BRANCH="${BRANCH:-master}" +MS_SUBDIR="${MS_SUBDIR:-ScaleDF_scaling_law_121_best_checkpoints_20260726}" +LISTING_REPO="${LISTING_REPO:-/tmp/ms_lyh_listing_${MS_SUBDIR}}" +CACHE_ROOT="${CACHE_ROOT:-/tmp/scaledf_ms_ckpt_stream_cache}" +OUTPUT_ROOT="${OUTPUT_ROOT:-$PERSIST_ROOT/per_sample_predictions/modelscope_scaling_law_121_scaledf_val_$(date -u +%Y%m%d_%H%M%S)}" +LAUNCHER="${LAUNCHER:-$PERSIST_ROOT/launch_scaledf_category_predict_8gpu.sh}" +MODEL_NAME="${MODEL_NAME:-deit_base_patch16_LS}" +ARIA2_CONNECTIONS="${ARIA2_CONNECTIONS:-8}" +MAX_TRIES="${MAX_TRIES:-5}" +LIMIT="${LIMIT:-0}" +START_AT="${START_AT:-1}" +SKIP_EXISTING="${SKIP_EXISTING:-1}" + +mkdir -p "$OUTPUT_ROOT" "$CACHE_ROOT" +LOG_DIR="$OUTPUT_ROOT/_queue_logs" +mkdir -p "$LOG_DIR" +QUEUE_LOG="$LOG_DIR/stream_queue.log" +STATUS_TSV="$OUTPUT_ROOT/queue_status.tsv" +MANIFEST_TSV="$OUTPUT_ROOT/manifest.tsv" +LOCK_DIR="$OUTPUT_ROOT/.stream_queue.lock" + +log() { + printf '[%s] %s\n' "$(date -u +%Y-%m-%dT%H:%M:%SZ)" "$*" | tee -a "$QUEUE_LOG" >&2 +} + +url_encode_relpath() { + python3 - "$1" <<'PY' +import sys +from urllib.parse import quote + +rel = sys.argv[1].strip("/") +print("/".join(quote(part) for part in rel.split("/"))) +PY +} + +sanitize_run_name() { + python3 - "$1" <<'PY' +import re +import sys +from pathlib import PurePosixPath + +rel = PurePosixPath(sys.argv[1]) +if rel.name in {"best_checkpoint.pth", "checkpoint.pth"} and rel.parent.name: + name = rel.parent.name +else: + name = rel.with_suffix("").name +print(re.sub(r"[^A-Za-z0-9_.=-]+", "_", name).strip("_")) +PY +} + +prepare_listing_repo() { + if [[ -d "$LISTING_REPO/.git" ]]; then + log "refresh_listing_repo path=$LISTING_REPO" + GIT_LFS_SKIP_SMUDGE=1 git -C "$LISTING_REPO" fetch --depth 1 origin "$BRANCH" >>"$QUEUE_LOG" 2>&1 + GIT_LFS_SKIP_SMUDGE=1 git -C "$LISTING_REPO" reset --hard FETCH_HEAD >>"$QUEUE_LOG" 2>&1 + else + rm -rf "$LISTING_REPO" + log "clone_listing_repo url=$REPO_URL path=$LISTING_REPO" + GIT_LFS_SKIP_SMUDGE=1 git clone --depth 1 --branch "$BRANCH" "$REPO_URL" "$LISTING_REPO" >>"$QUEUE_LOG" 2>&1 + fi +} + +build_manifest() { + local rel run_name idx + : > "$MANIFEST_TSV" + printf 'index\trun_name\trelpath\n' >> "$MANIFEST_TSV" + idx=0 + while IFS= read -r rel; do + idx=$((idx + 1)) + run_name="$(sanitize_run_name "$rel")" + printf '%s\t%s\t%s\n' "$idx" "$run_name" "$rel" >> "$MANIFEST_TSV" + done < <( + find "$LISTING_REPO/$MS_SUBDIR" -type f -name '*.pth' \ + | sed "s#^$LISTING_REPO/##" \ + | sort + ) + log "manifest_ready count=$idx path=$MANIFEST_TSV" +} + +download_checkpoint() { + local rel="$1" + local run_name="$2" + local encoded url dl_dir ckpt_path file_name + + encoded="$(url_encode_relpath "$rel")" + url="https://modelscope.cn/datasets/wangwenhao0716/lyh/resolve/$BRANCH/$encoded" + file_name="$(basename "$rel")" + dl_dir="$CACHE_ROOT/$run_name" + ckpt_path="$dl_dir/$file_name" + mkdir -p "$dl_dir" + + log "download_start run=$run_name rel=$rel" + aria2c \ + --continue=true \ + --max-connection-per-server="$ARIA2_CONNECTIONS" \ + --split="$ARIA2_CONNECTIONS" \ + --min-split-size=64M \ + --file-allocation=none \ + --retry-wait=5 \ + --max-tries="$MAX_TRIES" \ + --timeout=60 \ + --connect-timeout=30 \ + --dir="$dl_dir" \ + --out="$file_name" \ + "$url" >>"$QUEUE_LOG" 2>&1 + + if [[ ! -s "$ckpt_path" ]]; then + log "download_empty run=$run_name path=$ckpt_path" + return 1 + fi + + printf '%s\n' "$ckpt_path" +} + +run_prediction() { + local ckpt_path="$1" + local run_name="$2" + local rel="$3" + local run_out="$OUTPUT_ROOT/$run_name" + + mkdir -p "$run_out" + { + printf 'run_name=%s\n' "$run_name" + printf 'relpath=%s\n' "$rel" + printf 'checkpoint_cache=%s\n' "$ckpt_path" + printf 'model_name=%s\n' "$MODEL_NAME" + printf 'started_utc=%s\n' "$(date -u +%Y-%m-%dT%H:%M:%SZ)" + } > "$run_out/queue_meta.txt" + + log "predict_start run=$run_name out=$run_out" + "$LAUNCHER" "$ckpt_path" "$run_out" "$MODEL_NAME" >>"$QUEUE_LOG" 2>&1 + log "predict_done run=$run_name summary=$run_out/summary.tsv" +} + +cleanup_checkpoint() { + local ckpt_path="$1" + local dl_dir + dl_dir="$(dirname "$ckpt_path")" + rm -f "$ckpt_path" + rmdir "$dl_dir" 2>/dev/null || true +} + +if ! mkdir "$LOCK_DIR" 2>/dev/null; then + echo "Another stream queue appears to be running: $LOCK_DIR" >&2 + exit 2 +fi +trap 'rmdir "$LOCK_DIR" 2>/dev/null || true' EXIT + +if [[ ! -x "$LAUNCHER" ]]; then + echo "Launcher is not executable: $LAUNCHER" >&2 + exit 2 +fi + +if [[ ! -f "$STATUS_TSV" ]]; then + printf 'time_utc\tindex\trun_name\tstatus\trelpath\n' > "$STATUS_TSV" +fi + +log "queue_start output_root=$OUTPUT_ROOT cache_root=$CACHE_ROOT model=$MODEL_NAME" +prepare_listing_repo +build_manifest + +processed=0 +tail -n +2 "$MANIFEST_TSV" | while IFS=$'\t' read -r idx run_name rel; do + if (( idx < START_AT )); then + continue + fi + if (( LIMIT > 0 && processed >= LIMIT )); then + log "limit_reached processed=$processed limit=$LIMIT" + break + fi + + run_out="$OUTPUT_ROOT/$run_name" + if [[ "$SKIP_EXISTING" == "1" && -s "$run_out/summary.tsv" ]]; then + log "skip_existing index=$idx run=$run_name" + printf '%s\t%s\t%s\t%s\t%s\n' "$(date -u +%Y-%m-%dT%H:%M:%SZ)" "$idx" "$run_name" "skipped_existing" "$rel" >> "$STATUS_TSV" + processed=$((processed + 1)) + continue + fi + + ckpt_path="" + if ckpt_path="$(download_checkpoint "$rel" "$run_name")"; then + if run_prediction "$ckpt_path" "$run_name" "$rel"; then + cleanup_checkpoint "$ckpt_path" + printf '%s\t%s\t%s\t%s\t%s\n' "$(date -u +%Y-%m-%dT%H:%M:%SZ)" "$idx" "$run_name" "done" "$rel" >> "$STATUS_TSV" + else + log "predict_failed_keep_checkpoint index=$idx run=$run_name ckpt=$ckpt_path" + printf '%s\t%s\t%s\t%s\t%s\n' "$(date -u +%Y-%m-%dT%H:%M:%SZ)" "$idx" "$run_name" "predict_failed" "$rel" >> "$STATUS_TSV" + exit 1 + fi + else + log "download_failed index=$idx run=$run_name" + printf '%s\t%s\t%s\t%s\t%s\n' "$(date -u +%Y-%m-%dT%H:%M:%SZ)" "$idx" "$run_name" "download_failed" "$rel" >> "$STATUS_TSV" + exit 1 + fi + + processed=$((processed + 1)) +done + +log "queue_done output_root=$OUTPUT_ROOT" diff --git a/ScaleDF_rebuttal/launch_noqcpt_datasize_eval_queue.sh b/ScaleDF_rebuttal/launch_noqcpt_datasize_eval_queue.sh new file mode 100644 index 0000000000000000000000000000000000000000..9d5feb718d7065f5c69be6c0df376e6642592b29 --- /dev/null +++ b/ScaleDF_rebuttal/launch_noqcpt_datasize_eval_queue.sh @@ -0,0 +1,71 @@ +#!/usr/bin/env bash +set -euo pipefail + +if [[ $# -ne 2 ]]; then + echo "Usage: $0 " >&2 + exit 2 +fi + +FRACTION_LABELS="$1" +EVAL_TAG="$2" + +PERSIST_ROOT="${PERSIST_ROOT:-/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal}" +EVAL_LAUNCHER="${EVAL_LAUNCHER:-$PERSIST_ROOT/launch_rebuttal_eval_flex.sh}" +EVAL_ROOT="${EVAL_ROOT:-$PERSIST_ROOT/eval/noqcpt_datasize_benchmark_eval_${EVAL_TAG}}" +HOST_NAME="$(hostname)" + +export PERSIST_ROOT +export PACKAGE_ROOT="${PACKAGE_ROOT:-/tmp/ScaleDF_repro_20260724}" +export SCALEDF_DATA="${SCALEDF_DATA:-/dev/shm/scaledf_bind_storage/ScaleDF}" +export BENCHMARK_ROOT="${BENCHMARK_ROOT:-/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names}" +export SKIP_MISSING_BENCHMARKS="${SKIP_MISSING_BENCHMARKS:-1}" +export BATCH_PER_GPU="${BATCH_PER_GPU:-256}" +export NPROC_PER_NODE="${NPROC_PER_NODE:-8}" +export PYTHONUNBUFFERED="${PYTHONUNBUFFERED:-1}" + +eval_complete() { + local eval_log="$1" + [[ -f "$eval_log" ]] && [[ "$(grep -c 'Accuracy of the network on' "$eval_log" || true)" -ge 7 ]] +} + +find_checkpoint() { + local label="$1" + local run_dir + run_dir="$(find "$PERSIST_ROOT/runs" -maxdepth 1 -type d -name "noqcpt_datasize_${label}_*_${EVAL_TAG}" | sort | tail -n 1)" + if [[ -z "$run_dir" || ! -f "$run_dir/best_checkpoint.pth" ]]; then + echo "missing noqcpt checkpoint for label=$label" >&2 + exit 3 + fi + printf "%s/best_checkpoint.pth" "$run_dir" +} + +run_eval() { + local label="$1" + local checkpoint + checkpoint="$(find_checkpoint "$label")" + local tag="noqcpt_datasize_${label}" + local output_root="$EVAL_ROOT/$tag" + local eval_log="$output_root/eval.log" + + if eval_complete "$eval_log"; then + echo "[$(date -u +%Y-%m-%dT%H:%M:%SZ)] eval_skip tag=$tag reason=complete" + return + fi + + mkdir -p "$output_root" + echo "[$(date -u +%Y-%m-%dT%H:%M:%SZ)] eval_start tag=$tag checkpoint=$checkpoint" + OUTPUT_ROOT="$output_root" RUN_NAME="$tag" "$EVAL_LAUNCHER" "$checkpoint" deit_base_patch16_LS benchmarks + echo "[$(date -u +%Y-%m-%dT%H:%M:%SZ)] eval_done tag=$tag" +} + +echo "eval_queue_started_utc=$(date -u +%Y-%m-%dT%H:%M:%SZ)" +echo "hostname=$HOST_NAME" +echo "eval_tag=$EVAL_TAG" +echo "eval_root=$EVAL_ROOT" +echo "fraction_labels=$FRACTION_LABELS" + +for label in $FRACTION_LABELS; do + run_eval "$label" +done + +echo "eval_queue_done_utc=$(date -u +%Y-%m-%dT%H:%M:%SZ)" diff --git a/ScaleDF_rebuttal/launch_noqcpt_datasize_queue.sh b/ScaleDF_rebuttal/launch_noqcpt_datasize_queue.sh new file mode 100644 index 0000000000000000000000000000000000000000..b1b19bdd4cbd1873671a668dc6665d7c08c7aa8a --- /dev/null +++ b/ScaleDF_rebuttal/launch_noqcpt_datasize_queue.sh @@ -0,0 +1,108 @@ +#!/usr/bin/env bash +set -euo pipefail + +if [[ $# -ne 2 ]]; then + echo "Usage: $0 " >&2 + exit 2 +fi + +FRACTION_LABELS="$1" +LAUNCH_TAG="$2" + +PACKAGE_ROOT="${PACKAGE_ROOT:-/tmp/ScaleDF_repro_20260724}" +DATA_ROOT="${SCALEDF_DATA:-/dev/shm/scaledf_bind_storage/ScaleDF}" +PERSIST_ROOT="${PERSIST_ROOT:-/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal}" +SELECTION_DIR="${SELECTION_DIR:-$PERSIST_ROOT/selections/noqcpt_datasize_20260725}" +MODEL_NAME="${MODEL_NAME:-deit_base_patch16_LS}" +PRETRAIN="${PRETRAIN:-$PACKAGE_ROOT/pretrained/deit_3_base_224_21k.pth}" +HOST_NAME="$(hostname)" + +export PYTHONPATH="/run/tiger/python_stubs:$PACKAGE_ROOT/code/deit:${PYTHONPATH:-}" +export PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION="${PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION:-python}" +export WANDB_DISABLED="${WANDB_DISABLED:-true}" +export NO_ALBUMENTATIONS_UPDATE="${NO_ALBUMENTATIONS_UPDATE:-1}" +export NCCL_DEBUG="${SCALEDF_NCCL_DEBUG:-${NCCL_DEBUG:-WARN}}" +export PYTHONUNBUFFERED="${PYTHONUNBUFFERED:-1}" + +ulimit -n 1048576 >/dev/null 2>&1 || true + +if [[ ! -e /raid/ScaleDF_v4/train ]]; then + mkdir -p /raid 2>/dev/null || sudo mkdir -p /raid + ln -sfn "$DATA_ROOT" /raid/ScaleDF_v4 2>/dev/null || sudo ln -sfn "$DATA_ROOT" /raid/ScaleDF_v4 +fi + +run_one() { + local label="$1" + local selection + selection="$(find "$SELECTION_DIR" -maxdepth 1 -type f -name "scaledf_v4_noqcpt_datasize_N*_${label}_imgseed20260725.pkl" | sort | tail -n 1)" + if [[ -z "$selection" || ! -f "$selection" ]]; then + echo "missing selection for label=$label in $SELECTION_DIR" >&2 + exit 3 + fi + + local run_name="noqcpt_datasize_${label}_${HOST_NAME}_${LAUNCH_TAG}" + local output_root="$PERSIST_ROOT/runs/$run_name" + local log_file="$output_root/train.log" + mkdir -p "$output_root" + + { + echo "started_utc=$(date -u +%Y-%m-%dT%H:%M:%SZ)" + echo "hostname=$HOST_NAME" + echo "package_root=$PACKAGE_ROOT" + echo "data_root=$DATA_ROOT" + echo "selection=$selection" + echo "fraction_label=$label" + echo "model_name=$MODEL_NAME" + echo "output_root=$output_root" + echo "main=main_cross_2direct_balance_realscale_perturb_v4_NRealFake_noqcpt.py" + echo "nproc_per_node=${NPROC_PER_NODE:-8}" + echo "batch_per_gpu=${BATCH_PER_GPU:-256}" + echo "epochs=${EPOCHS:-20}" + } | tee -a "$log_file" + + cd "$PACKAGE_ROOT/code/deit" + torchrun --standalone --nproc_per_node="${NPROC_PER_NODE:-8}" \ + main_cross_2direct_balance_realscale_perturb_v4_NRealFake_noqcpt.py \ + --model "$MODEL_NAME" \ + --data-path "$DATA_ROOT" \ + --batch-size "${BATCH_PER_GPU:-256}" \ + --lr 1e-5 \ + --epochs "${EPOCHS:-20}" \ + --weight-decay 0.1 \ + --sched cosine \ + --input-size 224 \ + --eval-crop-ratio 1.0 \ + --reprob 0.0 \ + --smoothing 0.1 \ + --warmup-epochs 5 \ + --drop 0.0 \ + --data-set ScaleDF \ + --seed "${SEED:-0}" \ + --opt adamw \ + --warmup-lr 1e-6 \ + --mixup 0.8 \ + --drop-path 0.2 \ + --cutmix 1.0 \ + --unscale-lr \ + --aa rand-m9-mstd0.5-inc1 \ + --no-repeated-aug \ + --finetune "$PRETRAIN" \ + --select-models scaledf_v4_train_fix.pkl \ + --select-images "$selection" \ + --output_dir "$output_root" \ + --two-cls 2>&1 | tee -a "$log_file" + + echo "[$(date -u +%Y-%m-%dT%H:%M:%SZ)] done label=$label run=$run_name" +} + +echo "queue_started_utc=$(date -u +%Y-%m-%dT%H:%M:%SZ)" +echo "hostname=$HOST_NAME" +echo "launch_tag=$LAUNCH_TAG" +echo "fraction_labels=$FRACTION_LABELS" +echo "selection_dir=$SELECTION_DIR" + +for label in $FRACTION_LABELS; do + run_one "$label" +done + +echo "queue_done_utc=$(date -u +%Y-%m-%dT%H:%M:%SZ)" diff --git a/ScaleDF_rebuttal/launch_noqcpt_fixed500k_all_hosts.sh b/ScaleDF_rebuttal/launch_noqcpt_fixed500k_all_hosts.sh new file mode 100644 index 0000000000000000000000000000000000000000..a2d8c74fe8d776d31e287a61328db684fcced0e2 --- /dev/null +++ b/ScaleDF_rebuttal/launch_noqcpt_fixed500k_all_hosts.sh @@ -0,0 +1,39 @@ +#!/usr/bin/env bash +set -euo pipefail + +PERSIST_ROOT="${PERSIST_ROOT:-/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal}" +REMOTE_SCRIPT="$PERSIST_ROOT/launch_noqcpt_fixed500k_host_queue.sh" +LAUNCH_TAG="${LAUNCH_TAG:-$(date -u +%Y%m%d_%H%M%S)}" +SEEDS="${SEEDS:-0 1 2 3 4 5 6 7 8 9}" +SEED_LABEL="${SEED_LABEL:-$(printf '%s' "$SEEDS" | tr ' ' '_' | tr -cd '0-9_')}" +LOG_DIR="$PERSIST_ROOT/_host_logs/noqcpt_fixed500k_seed${SEED_LABEL}_${LAUNCH_TAG}" + +cp "$(dirname "$0")/launch_noqcpt_fixed500k_host_queue.sh" "$REMOTE_SCRIPT" +chmod +x "$REMOTE_SCRIPT" +mkdir -p "$LOG_DIR" + +submit() { + local port="$1" + local host="$2" + local label="$3" + local real_level="$4" + local fake_level="$5" + local all_seeds="$6" + local queue_log="$LOG_DIR/train_queue_${label}.log" + local pid_file="$LOG_DIR/train_queue_${label}.pid" + + ssh -p "$port" "tiger@$host" \ + "mkdir -p '$LOG_DIR'; nohup env SEEDS='$SEEDS' bash '$REMOTE_SCRIPT' '$real_level' '$fake_level' '$LAUNCH_TAG' '$all_seeds' > '$queue_log' 2>&1 < /dev/null & echo \$! > '$pid_file'; echo submitted label='$label' pid=\$(cat '$pid_file') log='$queue_log' all_seeds='$all_seeds'" +} + +submit 10338 'fdbd:dccd:cdc1:1368:0:e4::' n116-040-168 5 11 '0' +submit 11208 'fdbd:dccd:cdc1:1368:0:1d6::' n116-041-153 8 16 '' +submit 9869 'fdbd:dccd:cdc1:1368:0:1f8::' n116-041-188 11 22 '' +submit 11986 'fdbd:dccd:cdc1:1302:0:8d::' n111-046-081 16 32 '' +submit 11693 'fdbd:dccd:cdc1:1368:0:153::' n116-041-023 23 45 '' +submit 9999 'fdbd:dccd:cdc1:1368:0:111::' n116-040-213 32 64 '' + +echo "launch_tag=$LAUNCH_TAG" +echo "seeds=$SEEDS" +echo "seed_label=$SEED_LABEL" +echo "log_dir=$LOG_DIR" diff --git a/ScaleDF_rebuttal/launch_noqcpt_fixed500k_eval_all_hosts.sh b/ScaleDF_rebuttal/launch_noqcpt_fixed500k_eval_all_hosts.sh new file mode 100644 index 0000000000000000000000000000000000000000..499f2f92c2ae2054a1fc563b32e88a84048c0381 --- /dev/null +++ b/ScaleDF_rebuttal/launch_noqcpt_fixed500k_eval_all_hosts.sh @@ -0,0 +1,48 @@ +#!/usr/bin/env bash +set -euo pipefail + +if [[ $# -ne 1 ]]; then + echo "Usage: $0 " >&2 + exit 2 +fi + +TRAIN_TAG="$1" +PERSIST_ROOT="${PERSIST_ROOT:-/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal}" +REMOTE_SCRIPT="$PERSIST_ROOT/launch_noqcpt_fixed500k_eval_host_queue.sh" +REMOTE_EVAL_LAUNCHER="$PERSIST_ROOT/launch_rebuttal_eval_flex.sh" +EVAL_TAG="${EVAL_TAG:-$(date -u +%Y%m%d_%H%M%S)}" +SEEDS="${SEEDS:-0 1 2 3 4 5 6 7 8 9}" +SEED_LABEL="${SEED_LABEL:-$(printf '%s' "$SEEDS" | tr ' ' '_' | tr -cd '0-9_')}" +LOG_DIR="$PERSIST_ROOT/_host_logs/noqcpt_fixed500k_eval_${EVAL_TAG}" + +cp "$(dirname "$0")/launch_noqcpt_fixed500k_eval_host_queue.sh" "$REMOTE_SCRIPT" +cp "$(dirname "$0")/launch_rebuttal_eval_flex.sh" "$REMOTE_EVAL_LAUNCHER" +chmod +x "$REMOTE_SCRIPT" "$REMOTE_EVAL_LAUNCHER" +mkdir -p "$LOG_DIR" + +submit() { + local port="$1" + local host="$2" + local label="$3" + local real_level="$4" + local fake_level="$5" + local all_seeds="$6" + local queue_log="$LOG_DIR/eval_queue_${label}.log" + local pid_file="$LOG_DIR/eval_queue_${label}.pid" + + ssh -p "$port" "tiger@$host" \ + "mkdir -p '$LOG_DIR'; nohup env SEEDS='$SEEDS' SEED_LABEL='$SEED_LABEL' bash '$REMOTE_SCRIPT' '$TRAIN_TAG' '$EVAL_TAG' '$real_level' '$fake_level' '$all_seeds' > '$queue_log' 2>&1 < /dev/null & echo \$! > '$pid_file'; echo submitted label='$label' pid=\$(cat '$pid_file') log='$queue_log' all_seeds='$all_seeds'" +} + +submit 10338 'fdbd:dccd:cdc1:1368:0:e4::' n116-040-168 5 11 '0' +submit 11208 'fdbd:dccd:cdc1:1368:0:1d6::' n116-041-153 8 16 '' +submit 9869 'fdbd:dccd:cdc1:1368:0:1f8::' n116-041-188 11 22 '' +submit 11986 'fdbd:dccd:cdc1:1302:0:8d::' n111-046-081 16 32 '' +submit 11693 'fdbd:dccd:cdc1:1368:0:153::' n116-041-023 23 45 '' +submit 9999 'fdbd:dccd:cdc1:1368:0:111::' n116-040-213 32 64 '' + +echo "eval_tag=$EVAL_TAG" +echo "seeds=$SEEDS" +echo "seed_label=$SEED_LABEL" +echo "eval_root=$PERSIST_ROOT/eval/noqcpt_fixed500k_seed${SEED_LABEL}_benchmark_eval_${EVAL_TAG}" +echo "log_dir=$LOG_DIR" diff --git a/ScaleDF_rebuttal/launch_noqcpt_fixed500k_eval_host_queue.sh b/ScaleDF_rebuttal/launch_noqcpt_fixed500k_eval_host_queue.sh new file mode 100644 index 0000000000000000000000000000000000000000..4ce438027b0d33aff1c625bf54c6bb60f37b03d7 --- /dev/null +++ b/ScaleDF_rebuttal/launch_noqcpt_fixed500k_eval_host_queue.sh @@ -0,0 +1,152 @@ +#!/usr/bin/env bash +set -euo pipefail + +if [[ $# -lt 4 || $# -gt 5 ]]; then + echo "Usage: $0 [all-seeds-space-separated]" >&2 + exit 2 +fi + +TRAIN_TAG="$1" +EVAL_TAG="$2" +REAL_LEVEL="$3" +FAKE_LEVEL="$4" +ALL_SEEDS_RAW="${5:-}" + +PERSIST_ROOT="${PERSIST_ROOT:-/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal}" +EVAL_LAUNCHER="${EVAL_LAUNCHER:-$PERSIST_ROOT/launch_rebuttal_eval_flex.sh}" +SEEDS_RAW="${SEEDS:-0 1 2 3 4 5 6 7 8 9}" +SEED_LABEL="${SEED_LABEL:-0_9}" +EVAL_ROOT="${EVAL_ROOT:-$PERSIST_ROOT/eval/noqcpt_fixed500k_seed${SEED_LABEL}_benchmark_eval_${EVAL_TAG}}" +HOST_NAME="$(hostname)" + +export PERSIST_ROOT +export PACKAGE_ROOT="${PACKAGE_ROOT:-/tmp/ScaleDF_repro_20260724}" +export SCALEDF_DATA="${SCALEDF_DATA:-/dev/shm/scaledf_bind_storage/ScaleDF}" +export BENCHMARK_ROOT="${BENCHMARK_ROOT:-/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names}" +export SKIP_MISSING_BENCHMARKS="${SKIP_MISSING_BENCHMARKS:-1}" +export BATCH_PER_GPU="${BATCH_PER_GPU:-256}" +export NPROC_PER_NODE="${NPROC_PER_NODE:-8}" +export PYTHONUNBUFFERED="${PYTHONUNBUFFERED:-1}" + +find_axis_checkpoint() { + local family="$1" + local level="$2" + local seed="$3" + local pattern + if [[ "$seed" == "0" ]]; then + pattern="noqcpt_fixed500000_${family}_n${level}_classseed0_*_${TRAIN_TAG}" + else + pattern="noqcpt_fixed500000_${family}_n${level}_classseed${seed}_resampled_*_${TRAIN_TAG}" + fi + local run_dir + run_dir="$(find "$PERSIST_ROOT/runs" -maxdepth 1 -type d -name "$pattern" | sort | tail -n 1)" + if [[ -z "$run_dir" ]]; then + echo "missing run dir for pattern=$pattern" >&2 + exit 3 + fi + if [[ ! -f "$run_dir/best_checkpoint.pth" ]]; then + echo "missing checkpoint: $run_dir/best_checkpoint.pth" >&2 + exit 4 + fi + printf "%s/best_checkpoint.pth" "$run_dir" +} + +find_all_checkpoint() { + local seed="$1" + local pattern="noqcpt_fixed500000_all_46real_91fake_seed${seed}_*_${TRAIN_TAG}" + local run_dir + run_dir="$(find "$PERSIST_ROOT/runs" -maxdepth 1 -type d -name "$pattern" | sort | tail -n 1)" + if [[ -z "$run_dir" ]]; then + echo "missing all run dir for pattern=$pattern" >&2 + exit 3 + fi + if [[ ! -f "$run_dir/best_checkpoint.pth" ]]; then + echo "missing all checkpoint: $run_dir/best_checkpoint.pth" >&2 + exit 4 + fi + printf "%s/best_checkpoint.pth" "$run_dir" +} + +eval_complete() { + local eval_log="$1" + [[ -f "$eval_log" ]] && [[ "$(grep -c 'Accuracy of the network on' "$eval_log" || true)" -ge 7 ]] +} + +run_one_eval() { + local checkpoint="$1" + local tag="$2" + local output_root="$EVAL_ROOT/$tag" + local eval_log="$output_root/eval.log" + local lock_dir="$output_root/.eval.lock" + + if eval_complete "$eval_log"; then + echo "[$(date -u +%Y-%m-%dT%H:%M:%SZ)] eval_skip tag=$tag reason=complete output_root=$output_root" + return + fi + + mkdir -p "$output_root" + while ! mkdir "$lock_dir" 2>/dev/null; do + if eval_complete "$eval_log"; then + echo "[$(date -u +%Y-%m-%dT%H:%M:%SZ)] eval_skip tag=$tag reason=complete_after_wait output_root=$output_root" + return + fi + echo "[$(date -u +%Y-%m-%dT%H:%M:%SZ)] eval_wait tag=$tag reason=locked output_root=$output_root" + sleep 60 + done + + if eval_complete "$eval_log"; then + rmdir "$lock_dir" 2>/dev/null || true + echo "[$(date -u +%Y-%m-%dT%H:%M:%SZ)] eval_skip tag=$tag reason=complete_after_lock output_root=$output_root" + return + fi + + echo "[$(date -u +%Y-%m-%dT%H:%M:%SZ)] eval_start tag=$tag checkpoint=$checkpoint" + if OUTPUT_ROOT="$output_root" RUN_NAME="$tag" "$EVAL_LAUNCHER" "$checkpoint" deit_base_patch16_LS benchmarks; then + rmdir "$lock_dir" 2>/dev/null || true + else + rmdir "$lock_dir" 2>/dev/null || true + return 1 + fi + echo "[$(date -u +%Y-%m-%dT%H:%M:%SZ)] eval_done tag=$tag" +} + +run_axis_eval() { + local axis_short="$1" + local family="$2" + local level="$3" + local seed="$4" + local checkpoint + checkpoint="$(find_axis_checkpoint "$family" "$level" "$seed")" + local suffix="" + [[ "$seed" != "0" ]] && suffix="_resampled" + run_one_eval "$checkpoint" "fixed500k_${axis_short}_n${level}_seed${seed}${suffix}_noqcpt" +} + +run_all_eval() { + local seed="$1" + local checkpoint + checkpoint="$(find_all_checkpoint "$seed")" + run_one_eval "$checkpoint" "fixed500k_all_46real_91fake_seed${seed}_noqcpt" +} + +echo "eval_queue_started_utc=$(date -u +%Y-%m-%dT%H:%M:%SZ)" +echo "hostname=$HOST_NAME" +echo "train_tag=$TRAIN_TAG" +echo "eval_tag=$EVAL_TAG" +echo "real_level=$REAL_LEVEL" +echo "fake_level=$FAKE_LEVEL" +echo "eval_root=$EVAL_ROOT" +echo "batch_per_gpu=$BATCH_PER_GPU" +echo "seeds=$SEEDS_RAW" +echo "all_seeds=$ALL_SEEDS_RAW" + +for seed in $SEEDS_RAW; do + run_axis_eval real real_domains "$REAL_LEVEL" "$seed" + run_axis_eval fake fake_methods "$FAKE_LEVEL" "$seed" +done + +for seed in $ALL_SEEDS_RAW; do + run_all_eval "$seed" +done + +echo "eval_queue_done_utc=$(date -u +%Y-%m-%dT%H:%M:%SZ)" diff --git a/ScaleDF_rebuttal/launch_noqcpt_fixed500k_eval_on_host_train_done.sh b/ScaleDF_rebuttal/launch_noqcpt_fixed500k_eval_on_host_train_done.sh new file mode 100644 index 0000000000000000000000000000000000000000..bafd40f277b25b4b321780f6c94f11f0d57a70e6 --- /dev/null +++ b/ScaleDF_rebuttal/launch_noqcpt_fixed500k_eval_on_host_train_done.sh @@ -0,0 +1,67 @@ +#!/usr/bin/env bash +set -uo pipefail + +if [[ $# -ne 1 ]]; then + echo "Usage: $0 " >&2 + exit 2 +fi + +TRAIN_TAG="$1" +PERSIST_ROOT="${PERSIST_ROOT:-/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal}" +TRAIN_LOG_DIR="${TRAIN_LOG_DIR:-$PERSIST_ROOT/_host_logs/noqcpt_fixed500k_seed0_1_2_3_4_5_6_7_8_9_${TRAIN_TAG}}" +LOG_DIR="${LOG_DIR:-$PERSIST_ROOT/_host_logs/noqcpt_fixed500k_eval_host_done_${TRAIN_TAG}}" +REMOTE_SCRIPT="${REMOTE_SCRIPT:-$PERSIST_ROOT/launch_noqcpt_fixed500k_eval_host_queue.sh}" +SEEDS="${SEEDS:-0 1 2 3 4 5 6 7 8 9}" +SEED_LABEL="${SEED_LABEL:-0_1_2_3_4_5_6_7_8_9}" + +mkdir -p "$LOG_DIR" + +log_status() { + echo "[$(date -u +%Y-%m-%dT%H:%M:%SZ)] $*" +} + +trap 'code=$?; log_status "noqcpt_host_done_eval_watch_exit code=$code"' EXIT + +submit() { + local port="$1" + local host="$2" + local label="$3" + local real_level="$4" + local fake_level="$5" + local all_seeds="$6" + local train_log="$TRAIN_LOG_DIR/train_queue_${label}.log" + local queue_log="$LOG_DIR/eval_queue_${label}.log" + local pid_file="$LOG_DIR/eval_queue_${label}.pid" + local marker="$LOG_DIR/eval_queue_${label}.submitted" + + [[ -f "$marker" ]] && return 0 + grep -q 'queue_done_utc' "$train_log" 2>/dev/null || return 0 + + log_status "submit_eval label=$label real_level=$real_level fake_level=$fake_level all_seeds=$all_seeds" + if ssh -p "$port" "tiger@$host" \ + "mkdir -p '$LOG_DIR'; nohup env SEEDS='$SEEDS' SEED_LABEL='$SEED_LABEL' bash '$REMOTE_SCRIPT' '$TRAIN_TAG' '$TRAIN_TAG' '$real_level' '$fake_level' '$all_seeds' > '$queue_log' 2>&1 < /dev/null & echo \$! > '$pid_file'; echo submitted label='$label' pid=\$(cat '$pid_file') log='$queue_log' all_seeds='$all_seeds'"; then + date -u +%Y-%m-%dT%H:%M:%SZ > "$marker" + else + log_status "submit_eval_failed label=$label" + fi +} + +log_status "noqcpt_host_done_eval_watch_started train_tag=$TRAIN_TAG train_log_dir=$TRAIN_LOG_DIR log_dir=$LOG_DIR" + +while true; do + submit 10338 'fdbd:dccd:cdc1:1368:0:e4::' n116-040-168 5 11 '0' + submit 11208 'fdbd:dccd:cdc1:1368:0:1d6::' n116-041-153 8 16 '' + submit 9869 'fdbd:dccd:cdc1:1368:0:1f8::' n116-041-188 11 22 '' + submit 11986 'fdbd:dccd:cdc1:1302:0:8d::' n111-046-081 16 32 '' + submit 11693 'fdbd:dccd:cdc1:1368:0:153::' n116-041-023 23 45 '' + submit 9999 'fdbd:dccd:cdc1:1368:0:111::' n116-040-213 32 64 '' + + submitted=$(find "$LOG_DIR" -maxdepth 1 -name 'eval_queue_*.submitted' -type f 2>/dev/null | wc -l | tr -d ' ') + done_count=$({ grep -l 'eval_queue_done_utc' "$LOG_DIR"/eval_queue_*.log 2>/dev/null || true; } | wc -l | tr -d ' ') + log_status "scan submitted=$submitted/6 eval_done=$done_count/6" + if [[ "$submitted" -eq 6 && "$done_count" -eq 6 ]]; then + log_status "noqcpt_host_done_eval_watch_done" + break + fi + sleep 60 +done diff --git a/ScaleDF_rebuttal/launch_noqcpt_fixed500k_host_queue.sh b/ScaleDF_rebuttal/launch_noqcpt_fixed500k_host_queue.sh new file mode 100644 index 0000000000000000000000000000000000000000..400b5a66074287e6da26db6dcbdfd42e6ac31857 --- /dev/null +++ b/ScaleDF_rebuttal/launch_noqcpt_fixed500k_host_queue.sh @@ -0,0 +1,174 @@ +#!/usr/bin/env bash +set -euo pipefail + +if [[ $# -lt 3 || $# -gt 4 ]]; then + echo "Usage: $0 [all-seeds-space-separated]" >&2 + exit 2 +fi + +REAL_LEVEL="$1" +FAKE_LEVEL="$2" +LAUNCH_TAG="$3" +ALL_SEEDS_RAW="${4:-}" + +PACKAGE_ROOT="${PACKAGE_ROOT:-/tmp/ScaleDF_repro_20260724}" +DATA_ROOT="${SCALEDF_DATA:-/dev/shm/scaledf_bind_storage/ScaleDF}" +PERSIST_ROOT="${PERSIST_ROOT:-/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal}" +SELECTION_DIR_SEED0="${SELECTION_DIR_SEED0:-$PERSIST_ROOT/selections/fixed_budget_500k}" +SELECTION_DIR_RESAMPLED="${SELECTION_DIR_RESAMPLED:-$PERSIST_ROOT/selections/fixed_budget_500k_resampled_20260725}" +ALL_SELECTION="${ALL_SELECTION:-$PERSIST_ROOT/selections/fixed_budget_500k_all_46real_91fake_20260725/scaledf_v4_fixed500000_all_46real_91fake_imgseed20260725.pkl}" +MODEL_NAME="${MODEL_NAME:-deit_base_patch16_LS}" +PRETRAIN="${PRETRAIN:-$PACKAGE_ROOT/pretrained/deit_3_base_224_21k.pth}" +HOST_NAME="$(hostname)" +SEEDS_RAW="${SEEDS:-0 1 2 3 4 5 6 7 8 9}" + +export PYTHONPATH="/run/tiger/python_stubs:$PACKAGE_ROOT/code/deit:${PYTHONPATH:-}" +export PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION="${PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION:-python}" +export WANDB_DISABLED="${WANDB_DISABLED:-true}" +export NO_ALBUMENTATIONS_UPDATE="${NO_ALBUMENTATIONS_UPDATE:-1}" +export NCCL_DEBUG="${SCALEDF_NCCL_DEBUG:-${NCCL_DEBUG:-WARN}}" +export PYTHONUNBUFFERED="${PYTHONUNBUFFERED:-1}" + +ulimit -n 1048576 >/dev/null 2>&1 || true + +if [[ ! -e /raid/ScaleDF_v4/train ]]; then + mkdir -p /raid 2>/dev/null || sudo mkdir -p /raid + ln -sfn "$DATA_ROOT" /raid/ScaleDF_v4 2>/dev/null || sudo ln -sfn "$DATA_ROOT" /raid/ScaleDF_v4 +fi + +selection_for() { + local family="$1" + local level="$2" + local seed="$3" + if [[ "$seed" == "0" ]]; then + printf "%s/scaledf_v4_fixed500000_%s_n%s_classseed0_imgseed20260725.pkl" "$SELECTION_DIR_SEED0" "$family" "$level" + else + printf "%s/scaledf_v4_fixed500000_%s_n%s_classseed%s_imgseed20260725_resampled.pkl" "$SELECTION_DIR_RESAMPLED" "$family" "$level" "$seed" + fi +} + +run_name_for() { + local family="$1" + local level="$2" + local seed="$3" + if [[ "$seed" == "0" ]]; then + printf "noqcpt_fixed500000_%s_n%s_classseed0_%s_%s" "$family" "$level" "$HOST_NAME" "$LAUNCH_TAG" + else + printf "noqcpt_fixed500000_%s_n%s_classseed%s_resampled_%s_%s" "$family" "$level" "$seed" "$HOST_NAME" "$LAUNCH_TAG" + fi +} + +run_train() { + local selection="$1" + local run_name="$2" + local seed="$3" + local family="$4" + local level="$5" + local output_root="$PERSIST_ROOT/runs/$run_name" + local log_file="$output_root/train.log" + + if [[ ! -f "$selection" ]]; then + echo "missing selection: $selection" >&2 + exit 3 + fi + if [[ -f "$output_root/best_checkpoint.pth" ]]; then + echo "[$(date -u +%Y-%m-%dT%H:%M:%SZ)] skip_existing family=$family level=$level seed=$seed run=$run_name" + return + fi + + mkdir -p "$output_root" + { + echo "started_utc=$(date -u +%Y-%m-%dT%H:%M:%SZ)" + echo "hostname=$HOST_NAME" + echo "package_root=$PACKAGE_ROOT" + echo "data_root=$DATA_ROOT" + echo "family=$family" + echo "level=$level" + echo "seed=$seed" + echo "selection=$selection" + echo "model_name=$MODEL_NAME" + echo "output_root=$output_root" + echo "main=main_cross_2direct_balance_realscale_perturb_v4_NRealFake_noqcpt.py" + echo "nproc_per_node=${NPROC_PER_NODE:-8}" + echo "batch_per_gpu=${BATCH_PER_GPU:-256}" + echo "epochs=${EPOCHS:-10}" + } | tee -a "$log_file" + + cd "$PACKAGE_ROOT/code/deit" + torchrun --standalone --nproc_per_node="${NPROC_PER_NODE:-8}" \ + main_cross_2direct_balance_realscale_perturb_v4_NRealFake_noqcpt.py \ + --model "$MODEL_NAME" \ + --data-path "$DATA_ROOT" \ + --batch-size "${BATCH_PER_GPU:-256}" \ + --lr 1e-5 \ + --epochs "${EPOCHS:-10}" \ + --weight-decay 0.1 \ + --sched cosine \ + --input-size 224 \ + --eval-crop-ratio 1.0 \ + --reprob 0.0 \ + --smoothing 0.1 \ + --warmup-epochs 5 \ + --drop 0.0 \ + --data-set ScaleDF \ + --seed "$seed" \ + --opt adamw \ + --warmup-lr 1e-6 \ + --mixup 0.8 \ + --drop-path 0.2 \ + --cutmix 1.0 \ + --unscale-lr \ + --aa rand-m9-mstd0.5-inc1 \ + --no-repeated-aug \ + --finetune "$PRETRAIN" \ + --select-models scaledf_v4_train_fix.pkl \ + --select-images "$selection" \ + --output_dir "$output_root" \ + --two-cls 2>&1 | tee -a "$log_file" +} + +run_axis() { + local family="$1" + local level="$2" + local seed="$3" + local selection + local run_name + selection="$(selection_for "$family" "$level" "$seed")" + run_name="$(run_name_for "$family" "$level" "$seed")" + echo "[$(date -u +%Y-%m-%dT%H:%M:%SZ)] start family=$family level=$level seed=$seed run=$run_name" + run_train "$selection" "$run_name" "$seed" "$family" "$level" + echo "[$(date -u +%Y-%m-%dT%H:%M:%SZ)] done family=$family level=$level seed=$seed run=$run_name" +} + +run_all() { + local seed="$1" + local run_name="noqcpt_fixed500000_all_46real_91fake_seed${seed}_${HOST_NAME}_${LAUNCH_TAG}" + echo "[$(date -u +%Y-%m-%dT%H:%M:%SZ)] start family=all_46real_91fake level=all seed=$seed run=$run_name" + run_train "$ALL_SELECTION" "$run_name" "$seed" all_46real_91fake all + echo "[$(date -u +%Y-%m-%dT%H:%M:%SZ)] done family=all_46real_91fake level=all seed=$seed run=$run_name" +} + +echo "queue_started_utc=$(date -u +%Y-%m-%dT%H:%M:%SZ)" +echo "hostname=$HOST_NAME" +echo "real_level=$REAL_LEVEL" +echo "fake_level=$FAKE_LEVEL" +echo "launch_tag=$LAUNCH_TAG" +echo "seed0_selection_dir=$SELECTION_DIR_SEED0" +echo "resampled_selection_dir=$SELECTION_DIR_RESAMPLED" +echo "all_selection=$ALL_SELECTION" +echo "epochs=${EPOCHS:-10}" +echo "batch_per_gpu=${BATCH_PER_GPU:-256}" +echo "nproc_per_node=${NPROC_PER_NODE:-8}" +echo "seeds=$SEEDS_RAW" +echo "all_seeds=$ALL_SEEDS_RAW" + +for seed in $SEEDS_RAW; do + run_axis real_domains "$REAL_LEVEL" "$seed" + run_axis fake_methods "$FAKE_LEVEL" "$seed" +done + +for seed in $ALL_SEEDS_RAW; do + run_all "$seed" +done + +echo "queue_done_utc=$(date -u +%Y-%m-%dT%H:%M:%SZ)" diff --git a/ScaleDF_rebuttal/launch_rebuttal_diversity_train.sh b/ScaleDF_rebuttal/launch_rebuttal_diversity_train.sh new file mode 100644 index 0000000000000000000000000000000000000000..fb7b52877f86cbb21e2acaf129738e8c1dfec520 --- /dev/null +++ b/ScaleDF_rebuttal/launch_rebuttal_diversity_train.sh @@ -0,0 +1,127 @@ +#!/usr/bin/env bash +set -euo pipefail + +if [[ $# -lt 2 ]]; then + cat >&2 <<'USAGE' +Usage: + launch_rebuttal_diversity_train.sh [run-name] + +Examples: + launch_rebuttal_diversity_train.sh standard select_pkl/scaledf_v4_11_0.pkl + launch_rebuttal_diversity_train.sh nreal select_pkl/scaledf_v4_NReal_1_1000_0.pkl + NUMBER_REAL_DOMAINS=8 launch_rebuttal_diversity_train.sh realdomains select_pkl/scaledf_v4_8_real_0.pkl +USAGE + exit 1 +fi + +KIND="$1" +SELECTION="$2" +RUN_NAME="${3:-diversity_${KIND}_$(basename "$SELECTION" .pkl)_$(hostname)_$(date -u +%Y%m%d_%H%M%S)}" + +PACKAGE_ROOT="${PACKAGE_ROOT:-/tmp/ScaleDF_repro_20260724}" +DATA_ROOT="${SCALEDF_DATA:-/dev/shm/scaledf_bind_storage/ScaleDF}" +PERSIST_ROOT="${PERSIST_ROOT:-/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal}" +OUTPUT_ROOT="${OUTPUT_ROOT:-$PERSIST_ROOT/runs/$RUN_NAME}" +LOG_FILE="$OUTPUT_ROOT/train.log" +MODEL_NAME="${MODEL_NAME:-deit_base_patch16_LS}" +FULL_SELECTION="${FULL_SELECTION:-scaledf_v4_train_fix.pkl}" +PRETRAIN="${PRETRAIN:-$PACKAGE_ROOT/pretrained/deit_3_base_224_21k.pth}" + +mkdir -p "$OUTPUT_ROOT" + +export PYTHONPATH="/run/tiger/python_stubs:$PACKAGE_ROOT/code/deit:${PYTHONPATH:-}" +export PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION="${PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION:-python}" +export WANDB_DISABLED="${WANDB_DISABLED:-true}" +export NO_ALBUMENTATIONS_UPDATE="${NO_ALBUMENTATIONS_UPDATE:-1}" +export NCCL_DEBUG="${SCALEDF_NCCL_DEBUG:-${NCCL_DEBUG:-WARN}}" +export PYTHONUNBUFFERED="${PYTHONUNBUFFERED:-1}" + +ulimit -n 1048576 >/dev/null 2>&1 || true + +cd "$PACKAGE_ROOT/code/deit" + +if [[ "$SELECTION" != /* && ! -f "$SELECTION" ]]; then + echo "Selection pkl not found from code/deit: $SELECTION" >&2 + exit 2 +fi +if [[ "$SELECTION" == /* && ! -f "$SELECTION" ]]; then + echo "Selection pkl not found: $SELECTION" >&2 + exit 2 +fi +if [[ "$KIND" == "nreal" || "$KIND" == "nfake" || "$KIND" == "nrealfake" ]]; then + if [[ ! -e /raid/ScaleDF_v4/train ]]; then + echo "Image-level selection files contain /raid/ScaleDF_v4 paths; create that symlink first." >&2 + exit 3 + fi +fi + +case "$KIND" in + standard) + MAIN="main_cross_2direct_balance_realscale_perturb_v4.py" + EXTRA_ARGS=(--select-models "$SELECTION") + ;; + nreal) + MAIN="main_cross_2direct_balance_realscale_perturb_v4_NReal.py" + EXTRA_ARGS=(--select-models "$FULL_SELECTION" --select-images "$SELECTION") + ;; + nfake) + MAIN="main_cross_2direct_balance_realscale_perturb_v4_NFake.py" + EXTRA_ARGS=(--select-models "$FULL_SELECTION" --select-images "$SELECTION") + ;; + nrealfake) + MAIN="main_cross_2direct_balance_realscale_perturb_v4_NRealFake.py" + EXTRA_ARGS=(--select-models "$FULL_SELECTION" --select-images "$SELECTION") + ;; + realdomains) + MAIN="main_cross_2direct_balance_realscale_perturb_v4_realdomains.py" + EXTRA_ARGS=(--select-models "$SELECTION" --number-real-domains "${NUMBER_REAL_DOMAINS:?Set NUMBER_REAL_DOMAINS for realdomains}") + ;; + *) + echo "Unknown kind: $KIND" >&2 + exit 2 + ;; +esac + +{ + echo "started_utc=$(date -u +%Y-%m-%dT%H:%M:%SZ)" + echo "hostname=$(hostname)" + echo "package_root=$PACKAGE_ROOT" + echo "data_root=$DATA_ROOT" + echo "kind=$KIND" + echo "selection=$SELECTION" + echo "model_name=$MODEL_NAME" + echo "output_root=$OUTPUT_ROOT" + echo "main=$MAIN" + echo "nproc_per_node=${NPROC_PER_NODE:-8}" + echo "batch_per_gpu=${BATCH_PER_GPU:-256}" +} | tee -a "$LOG_FILE" + +torchrun --standalone --nproc_per_node="${NPROC_PER_NODE:-8}" \ + "$MAIN" \ + --model "$MODEL_NAME" \ + --data-path "$DATA_ROOT" \ + --batch-size "${BATCH_PER_GPU:-256}" \ + --lr 1e-5 \ + --epochs "${EPOCHS:-20}" \ + --weight-decay 0.1 \ + --sched cosine \ + --input-size 224 \ + --eval-crop-ratio 1.0 \ + --reprob 0.0 \ + --smoothing 0.1 \ + --warmup-epochs 5 \ + --drop 0.0 \ + --data-set ScaleDF \ + --seed "${SEED:-0}" \ + --opt adamw \ + --warmup-lr 1e-6 \ + --mixup 0.8 \ + --drop-path 0.2 \ + --cutmix 1.0 \ + --unscale-lr \ + --aa rand-m9-mstd0.5-inc1 \ + --no-repeated-aug \ + --finetune "$PRETRAIN" \ + --output_dir "$OUTPUT_ROOT" \ + --two-cls \ + "${EXTRA_ARGS[@]}" 2>&1 | tee -a "$LOG_FILE" diff --git a/ScaleDF_rebuttal/launch_rebuttal_eval_flex.sh b/ScaleDF_rebuttal/launch_rebuttal_eval_flex.sh new file mode 100644 index 0000000000000000000000000000000000000000..e2aedf30e61d4f4210e250f04efa5628b41aeebf --- /dev/null +++ b/ScaleDF_rebuttal/launch_rebuttal_eval_flex.sh @@ -0,0 +1,141 @@ +#!/usr/bin/env bash +set -euo pipefail + +if [[ $# -lt 2 ]]; then + cat >&2 <<'USAGE' +Usage: + launch_rebuttal_eval_flex.sh [scaledf|benchmarks|all] + +Useful env: + SCALEDF_MAIN Python file for ScaleDF val eval. + BENCHMARK_MAIN Python file for benchmark eval. + PYTHON_BIN Python executable; if set, uses: python -m torch.distributed.run. + RUN_NAME Output run name under $PERSIST_ROOT/eval. +USAGE + exit 1 +fi + +PACKAGE_ROOT="${PACKAGE_ROOT:-/tmp/ScaleDF_repro_20260724}" +DATA_ROOT="${SCALEDF_DATA:-/dev/shm/scaledf_bind_storage/ScaleDF}" +BENCHMARK_ROOT="${BENCHMARK_ROOT:-/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names}" +PERSIST_ROOT="${PERSIST_ROOT:-/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal}" +CHECKPOINT="$1" +MODEL_NAME="$2" +MODE="${3:-all}" +RUN_NAME="${RUN_NAME:-eval_flex_$(date -u +%Y%m%d_%H%M%S)}" +OUTPUT_ROOT="${OUTPUT_ROOT:-$PERSIST_ROOT/eval/$RUN_NAME}" +LOG_FILE="$OUTPUT_ROOT/eval.log" +SKIP_MISSING_BENCHMARKS="${SKIP_MISSING_BENCHMARKS:-0}" +SCALEDF_MAIN="${SCALEDF_MAIN:-${SCALED_F_MAIN:-main_cross_2direct_balance_realscale_perturb_v4.py}}" +BENCHMARK_MAIN="${BENCHMARK_MAIN:-main_cross_2direct_balance_test.py}" + +mkdir -p "$OUTPUT_ROOT" + +if [[ -d "$CHECKPOINT" ]]; then + CHECKPOINT="$CHECKPOINT/best_checkpoint.pth" +fi +CHECKPOINT="$(realpath "$CHECKPOINT")" + +export PYTHONPATH="/run/tiger/python_stubs:$PACKAGE_ROOT/code/deit:${PYTHONPATH:-}" +export PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION="${PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION:-python}" +export WANDB_DISABLED="${WANDB_DISABLED:-true}" +export NO_ALBUMENTATIONS_UPDATE="${NO_ALBUMENTATIONS_UPDATE:-1}" +export NCCL_DEBUG="${SCALEDF_NCCL_DEBUG:-WARN}" +export PYTHONUNBUFFERED="${PYTHONUNBUFFERED:-1}" + +ulimit -n 1048576 >/dev/null 2>&1 || true + +cd "$PACKAGE_ROOT/code/deit" + +torchrun_prefix() { + if [[ -n "${PYTHON_BIN:-}" ]]; then + "$PYTHON_BIN" -m torch.distributed.run --standalone --nproc_per_node="${NPROC_PER_NODE:-8}" "$@" + else + torchrun --standalone --nproc_per_node="${NPROC_PER_NODE:-8}" "$@" + fi +} + +common_args=( + --model "$MODEL_NAME" + --batch-size "${BATCH_PER_GPU:-256}" + --input-size "${INPUT_SIZE:-224}" + --eval-crop-ratio 1.0 + --data-set ScaleDF + --two-cls + --resume "$CHECKPOINT" + --eval +) + +{ + echo "started_utc=$(date -u +%Y-%m-%dT%H:%M:%SZ)" + echo "hostname=$(hostname)" + echo "package_root=$PACKAGE_ROOT" + echo "data_root=$DATA_ROOT" + echo "benchmark_root=$BENCHMARK_ROOT" + echo "checkpoint=$CHECKPOINT" + echo "model_name=$MODEL_NAME" + echo "mode=$MODE" + echo "scaledf_main=$SCALEDF_MAIN" + echo "benchmark_main=$BENCHMARK_MAIN" + echo "python_bin=${PYTHON_BIN:-system}" + echo "output_root=$OUTPUT_ROOT" + echo "nproc_per_node=${NPROC_PER_NODE:-8}" + echo "batch_per_gpu=${BATCH_PER_GPU:-256}" + echo "input_size=${INPUT_SIZE:-224}" + echo "skip_missing_benchmarks=$SKIP_MISSING_BENCHMARKS" +} | tee -a "$LOG_FILE" + +run_scaledf() { + torchrun_prefix "$SCALEDF_MAIN" \ + --data-path "$DATA_ROOT" \ + --select-models scaledf_v4_train_fix.pkl \ + --output_dir "$OUTPUT_ROOT/scaledf_val" \ + "${common_args[@]}" 2>&1 | tee -a "$LOG_FILE" +} + +run_benchmarks() { + local datasets=( + "ScaleDF_v3.1" + "DF40_ori" + "CDFv2_ori" + "DiFF_processed" + "Wild_Deepfake_processed" + "ForgeryNet_processed" + "DeepFakeFace_processed" + "DeepFakeDetection_processed" + ) + + for name in "${datasets[@]}"; do + local data_path="$BENCHMARK_ROOT/$name" + [[ -d "$data_path" ]] || { + echo "Missing benchmark directory: $data_path" >&2 + if [[ "$SKIP_MISSING_BENCHMARKS" == "1" ]]; then + echo "Skipping missing benchmark: $name" | tee -a "$LOG_FILE" + continue + fi + exit 2 + } + + torchrun_prefix "$BENCHMARK_MAIN" \ + --data-path "$data_path" \ + --output_dir "$OUTPUT_ROOT/benchmarks/$name" \ + "${common_args[@]}" 2>&1 | tee -a "$LOG_FILE" + done +} + +case "$MODE" in + scaledf) + run_scaledf + ;; + benchmarks) + run_benchmarks + ;; + all) + run_scaledf + run_benchmarks + ;; + *) + echo "Unknown mode: $MODE" >&2 + exit 2 + ;; +esac diff --git a/ScaleDF_rebuttal/launch_scaledf_arcface_stage.sh b/ScaleDF_rebuttal/launch_scaledf_arcface_stage.sh new file mode 100644 index 0000000000000000000000000000000000000000..394ca3ea3465c2d43f1916589a9496cd5cdca9f9 --- /dev/null +++ b/ScaleDF_rebuttal/launch_scaledf_arcface_stage.sh @@ -0,0 +1,67 @@ +#!/usr/bin/env bash +set -euo pipefail + +ROOT=${ROOT:-/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal} +OUT=${1:?usage: launch_scaledf_arcface_stage.sh OUT PREFIX STAGE} +PREFIX=${2:?usage: launch_scaledf_arcface_stage.sh OUT PREFIX STAGE} +STAGE=${3:?usage: launch_scaledf_arcface_stage.sh OUT PREFIX STAGE} +NUM_SHARDS=${NUM_SHARDS:-8} +ARCFACE_BATCH=${ARCFACE_BATCH:-4096} +ARCFACE_WORKERS=${ARCFACE_WORKERS:-12} +ARCFACE_THREADS=${ARCFACE_THREADS:-4} + +SCRIPT=$ROOT/scaledf_embedding_audit.py +MODEL=$ROOT/models/insightface/models/buffalo_l/w600k_r50.onnx +MANIFEST_DIR=$OUT/manifests +EMB_DIR=$OUT/embeddings +LOG_DIR=$OUT/logs + +export PYTHONPATH="$ROOT/python_dedup_cuda12_pkgs:$ROOT/python_dedup_pkgs:${PYTHONPATH:-}" +export LD_LIBRARY_PATH="/usr/local/cuda/targets/x86_64-linux/lib:/lib/x86_64-linux-gnu:${LD_LIBRARY_PATH:-}" + +mkdir -p "$EMB_DIR" "$LOG_DIR" + +log() { + printf '%s %s\n' "$(date -u +%Y-%m-%dT%H:%M:%SZ)" "$*" +} + +log "arcface stage start: stage=$STAGE prefix=$PREFIX out=$OUT batch=$ARCFACE_BATCH workers=$ARCFACE_WORKERS" +pids=() +fail=0 +for shard in $(seq 0 $((NUM_SHARDS - 1))); do + shard_name=$(printf "%s_shard%02d_of%02d" "$PREFIX" "$shard" "$NUM_SHARDS") + manifest="$MANIFEST_DIR/${shard_name}.tsv" + out_prefix="$EMB_DIR/${STAGE}_${shard_name}" + log_file="$LOG_DIR/${STAGE}_${shard_name}.log" + if [[ -s "${out_prefix}.embeddings.npy" && -s "${out_prefix}.meta.tsv" ]]; then + log "skip exists: shard=$shard out=$out_prefix" + continue + fi + ( + set -euo pipefail + export CUDA_VISIBLE_DEVICES=$shard + python3 "$SCRIPT" embed-arcface-direct \ + --manifest "$manifest" \ + --model "$MODEL" \ + --out-prefix "$out_prefix" \ + --device cuda:0 \ + --batch-size "$ARCFACE_BATCH" \ + --workers "$ARCFACE_WORKERS" \ + --ort-threads "$ARCFACE_THREADS" \ + --status-every 5 + ) > "$log_file" 2>&1 & + pids+=("$!") + log "launched shard=$shard pid=$!" +done + +for pid in "${pids[@]}"; do + if ! wait "$pid"; then + fail=1 + fi +done + +if [[ "$fail" != 0 ]]; then + log "arcface stage failed: stage=$STAGE prefix=$PREFIX" + exit 1 +fi +log "arcface stage done: stage=$STAGE prefix=$PREFIX" diff --git a/ScaleDF_rebuttal/launch_scaledf_category_predict_8gpu.sh b/ScaleDF_rebuttal/launch_scaledf_category_predict_8gpu.sh new file mode 100644 index 0000000000000000000000000000000000000000..ab868ca16f843bb02f3132d62000c24dc988a3c9 --- /dev/null +++ b/ScaleDF_rebuttal/launch_scaledf_category_predict_8gpu.sh @@ -0,0 +1,96 @@ +#!/usr/bin/env bash +set -euo pipefail + +if [[ $# -lt 2 ]]; then + echo "usage: $0 CHECKPOINT OUTPUT_ROOT [MODEL_NAME]" >&2 + exit 2 +fi + +CHECKPOINT=$1 +OUTPUT_ROOT=$2 +MODEL_NAME=${3:-deit_base_patch16_LS} + +ROOT=${ROOT:-/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal} +PACKAGE_ROOT=${PACKAGE_ROOT:-/tmp/ScaleDF_repro_20260724} +SCALEDF_VAL_ROOT=${SCALEDF_VAL_ROOT:-/dev/shm/scaledf_bind_storage/ScaleDF/val} +NGPU=${NGPU:-8} +BATCH_SIZE=${BATCH_SIZE:-1024} +WORKERS=${WORKERS:-16} +PREFETCH_FACTOR=${PREFETCH_FACTOR:-4} +INPUT_SIZE=${INPUT_SIZE:-224} +MAX_IMAGES_ARG=() +if [[ -n "${MAX_IMAGES:-}" ]]; then + MAX_IMAGES_ARG=(--max-images "$MAX_IMAGES") +fi + +mkdir -p "$OUTPUT_ROOT/logs" "$OUTPUT_ROOT/predictions" +DRIVER_LOG="$OUTPUT_ROOT/logs/category_predict.driver.log" + +if [[ -f "$OUTPUT_ROOT/summary.tsv" ]]; then + echo "$(date -u +%FT%TZ) skip existing: $OUTPUT_ROOT/summary.tsv" + exit 0 +fi + +export PYTHONPATH="$ROOT/python_stubs:/run/tiger/python_stubs:$PACKAGE_ROOT/code/deit:${PYTHONPATH:-}" +export PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION="${PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION:-python}" +export WANDB_DISABLED="${WANDB_DISABLED:-true}" +export NO_ALBUMENTATIONS_UPDATE="${NO_ALBUMENTATIONS_UPDATE:-1}" +export PYTHONUNBUFFERED=1 +export OMP_NUM_THREADS="${OMP_NUM_THREADS:-1}" +export MKL_NUM_THREADS="${MKL_NUM_THREADS:-1}" + +{ + echo "started_utc=$(date -u +%FT%TZ)" + echo "hostname=$(hostname)" + echo "checkpoint=$CHECKPOINT" + echo "output_root=$OUTPUT_ROOT" + echo "scaledf_val_root=$SCALEDF_VAL_ROOT" + echo "model_name=$MODEL_NAME" + echo "ngpu=$NGPU" + echo "batch_size=$BATCH_SIZE" + echo "workers=$WORKERS" + echo "prefetch_factor=$PREFETCH_FACTOR" +} | tee "$DRIVER_LOG" + +pids=() +parts=() +for gpu in $(seq 0 $((NGPU - 1))); do + shard=$(printf '%02d' "$gpu") + out="$OUTPUT_ROOT/predictions/predictions_shard${shard}_of$(printf '%02d' "$NGPU").tsv" + log="$OUTPUT_ROOT/logs/category_predict_shard${shard}.log" + parts+=("$out") + CUDA_VISIBLE_DEVICES="$gpu" python3 "$ROOT/scaledf_predict_scores.py" predict \ + --dataset "ScaleDF:$SCALEDF_VAL_ROOT" \ + --checkpoint "$CHECKPOINT" \ + --model "$MODEL_NAME" \ + --code-root "$PACKAGE_ROOT/code/deit" \ + --out "$out" \ + --device "cuda:0" \ + --input-size "$INPUT_SIZE" \ + --batch-size "$BATCH_SIZE" \ + --workers "$WORKERS" \ + --prefetch-factor "$PREFETCH_FACTOR" \ + --shard-index "$gpu" \ + --num-shards "$NGPU" \ + "${MAX_IMAGES_ARG[@]}" > "$log" 2>&1 & + pids+=("$!") + echo "$(date -u +%FT%TZ) launch gpu=$gpu out=$out" | tee -a "$DRIVER_LOG" +done + +status=0 +for pid in "${pids[@]}"; do + if ! wait "$pid"; then + status=1 + fi +done +if [[ $status -ne 0 ]]; then + echo "$(date -u +%FT%TZ) prediction failed" | tee -a "$DRIVER_LOG" + exit "$status" +fi + +cmd=(python3 "$ROOT/scaledf_predict_scores.py" summarize --out "$OUTPUT_ROOT/summary.tsv") +for part in "${parts[@]}"; do + cmd+=(--prediction-tsv "$part") +done +"${cmd[@]}" >> "$DRIVER_LOG" 2>&1 +echo "$(date -u +%FT%TZ) prediction done summary=$OUTPUT_ROOT/summary.tsv" | tee -a "$DRIVER_LOG" diff --git a/ScaleDF_rebuttal/launch_scaledf_category_predict_queue.sh b/ScaleDF_rebuttal/launch_scaledf_category_predict_queue.sh new file mode 100644 index 0000000000000000000000000000000000000000..5396f2df6807fe834dc65e9c0c0ac4b41da0df9f --- /dev/null +++ b/ScaleDF_rebuttal/launch_scaledf_category_predict_queue.sh @@ -0,0 +1,39 @@ +#!/usr/bin/env bash +set -euo pipefail + +if [[ $# -lt 2 ]]; then + echo "usage: $0 CHECKPOINTS_TSV OUTPUT_ROOT" >&2 + echo "TSV columns: run_namecheckpointmodel_name(optional)" >&2 + exit 2 +fi + +CHECKPOINTS_TSV=$1 +OUTPUT_ROOT=$2 + +ROOT=${ROOT:-/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal} +DEFAULT_MODEL=${DEFAULT_MODEL:-deit_base_patch16_LS} +QUEUE_LOG="$OUTPUT_ROOT/category_predict_queue.log" +mkdir -p "$OUTPUT_ROOT" + +{ + echo "started_utc=$(date -u +%FT%TZ)" + echo "hostname=$(hostname)" + echo "checkpoints_tsv=$CHECKPOINTS_TSV" + echo "output_root=$OUTPUT_ROOT" +} | tee "$QUEUE_LOG" + +while IFS=$'\t' read -r run_name checkpoint model_name extra; do + [[ -z "${run_name:-}" ]] && continue + [[ "$run_name" == \#* ]] && continue + if [[ -z "${checkpoint:-}" ]]; then + echo "skip malformed line: $run_name" | tee -a "$QUEUE_LOG" + continue + fi + model_name=${model_name:-$DEFAULT_MODEL} + run_out="$OUTPUT_ROOT/$run_name" + echo "$(date -u +%FT%TZ) start run=$run_name checkpoint=$checkpoint model=$model_name" | tee -a "$QUEUE_LOG" + bash "$ROOT/launch_scaledf_category_predict_8gpu.sh" "$checkpoint" "$run_out" "$model_name" 2>&1 | tee -a "$QUEUE_LOG" + echo "$(date -u +%FT%TZ) done run=$run_name summary=$run_out/summary.tsv" | tee -a "$QUEUE_LOG" +done < "$CHECKPOINTS_TSV" + +echo "finished_utc=$(date -u +%FT%TZ)" | tee -a "$QUEUE_LOG" diff --git a/ScaleDF_rebuttal/launch_scaledf_embedding_audit_full.sh b/ScaleDF_rebuttal/launch_scaledf_embedding_audit_full.sh new file mode 100644 index 0000000000000000000000000000000000000000..5e5b0161366a26f373dcf6ae9012f70535957e1f --- /dev/null +++ b/ScaleDF_rebuttal/launch_scaledf_embedding_audit_full.sh @@ -0,0 +1,137 @@ +#!/usr/bin/env bash +set -euo pipefail + +ROOT=${ROOT:-/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal} +DATA=${DATA:-/dev/shm/scaledf_bind_storage} +OUT=${1:-$ROOT/analysis/embedding_audit_full_$(date -u +%Y%m%d_%H%M%S)} +NUM_SHARDS=${NUM_SHARDS:-8} +SSCD_BATCH=${SSCD_BATCH:-512} +SSCD_WORKERS=${SSCD_WORKERS:-8} +ARCFACE_BATCH=${ARCFACE_BATCH:-2048} +ARCFACE_THREADS=${ARCFACE_THREADS:-4} + +SCRIPT=$ROOT/scaledf_embedding_audit.py +SSCD_MODEL=$ROOT/models/sscd_disc_mixup.torchscript.pt +ARCFACE_MODEL=$ROOT/models/insightface/models/buffalo_l/w600k_r50.onnx +MANIFEST_DIR=$OUT/manifests +EMB_DIR=$OUT/embeddings +LOG_DIR=$OUT/logs + +export PYTHONPATH="$ROOT/python_dedup_cuda12_pkgs:$ROOT/python_dedup_pkgs:${PYTHONPATH:-}" +export LD_LIBRARY_PATH="/usr/local/cuda/targets/x86_64-linux/lib:/lib/x86_64-linux-gnu:${LD_LIBRARY_PATH:-}" + +mkdir -p "$MANIFEST_DIR" "$EMB_DIR" "$LOG_DIR" +echo "$OUT" > "$ROOT/analysis/latest_embedding_audit_full_dir.txt" + +log() { + printf '%s %s\n' "$(date -u +%Y-%m-%dT%H:%M:%SZ)" "$*" +} + +run_manifest() { + if [[ -s "$OUT/manifest.tsv" ]]; then + log "manifest exists: $OUT/manifest.tsv" + return + fi + log "manifest start" + python3 "$SCRIPT" manifest \ + --root "train=$DATA/ScaleDF/train" \ + --root "scaledf_val=$DATA/ScaleDF/val" \ + --root "benchmark=$DATA/Established_benchmarks" \ + --out "$OUT/manifest.tsv" \ + > "$LOG_DIR/manifest.log" 2>&1 + log "manifest done" +} + +run_split() { + local prefix=$1 + local split=$2 + if [[ -s "$MANIFEST_DIR/${prefix}_split_summary.tsv" ]]; then + log "split exists: $prefix" + return + fi + log "split start: $prefix" + python3 "$SCRIPT" split-manifest \ + --manifest "$OUT/manifest.tsv" \ + --out-dir "$MANIFEST_DIR" \ + --prefix "$prefix" \ + --num-shards "$NUM_SHARDS" \ + --split "$split" \ + > "$LOG_DIR/split_${prefix}.log" 2>&1 + log "split done: $prefix" +} + +run_stage() { + local stage=$1 + local kind=$2 + local prefix=$3 + log "stage start: $stage" + local pids=() + local fail=0 + for shard in $(seq 0 $((NUM_SHARDS - 1))); do + local shard_name + shard_name=$(printf "%s_shard%02d_of%02d" "$prefix" "$shard" "$NUM_SHARDS") + local manifest="$MANIFEST_DIR/${shard_name}.tsv" + local out_prefix="$EMB_DIR/${stage}_${shard_name}" + local log_file="$LOG_DIR/${stage}_${shard_name}.log" + if [[ -s "${out_prefix}.embeddings.npy" && -s "${out_prefix}.meta.tsv" ]]; then + log "skip exists: $stage shard=$shard" + continue + fi + if [[ "$kind" == "sscd" ]]; then + ( + set -euo pipefail + export CUDA_VISIBLE_DEVICES=$shard + python3 "$SCRIPT" embed-sscd \ + --manifest "$manifest" \ + --model "$SSCD_MODEL" \ + --out-prefix "$out_prefix" \ + --device cuda:0 \ + --batch-size "$SSCD_BATCH" \ + --workers "$SSCD_WORKERS" \ + --status-every 50 \ + --overwrite + ) > "$log_file" 2>&1 & + else + ( + set -euo pipefail + export CUDA_VISIBLE_DEVICES=$shard + python3 "$SCRIPT" embed-arcface-direct \ + --manifest "$manifest" \ + --model "$ARCFACE_MODEL" \ + --out-prefix "$out_prefix" \ + --device cuda:0 \ + --batch-size "$ARCFACE_BATCH" \ + --ort-threads "$ARCFACE_THREADS" \ + --status-every 10 \ + --overwrite + ) > "$log_file" 2>&1 & + fi + pids+=("$!") + log "launched: $stage shard=$shard pid=$!" + done + + for pid in "${pids[@]}"; do + if ! wait "$pid"; then + fail=1 + fi + done + if [[ "$fail" != 0 ]]; then + log "stage failed: $stage" + return 1 + fi + log "stage done: $stage" +} + +run_manifest +run_split train train +run_split eval val +run_split benchmark benchmark + +run_stage sscd_train sscd train +run_stage sscd_eval_scaledf sscd eval +run_stage sscd_eval_benchmark sscd benchmark +run_stage arcface_train arcface train +run_stage arcface_eval_scaledf arcface eval +run_stage arcface_eval_benchmark arcface benchmark + +log "all embedding stages done: $OUT" diff --git a/ScaleDF_rebuttal/launch_scaledf_match_exact_torch.sh b/ScaleDF_rebuttal/launch_scaledf_match_exact_torch.sh new file mode 100644 index 0000000000000000000000000000000000000000..9fdfd588d902d060a4ad18500f2c22cad1805839 --- /dev/null +++ b/ScaleDF_rebuttal/launch_scaledf_match_exact_torch.sh @@ -0,0 +1,119 @@ +#!/usr/bin/env bash +set -euo pipefail + +if [[ $# -lt 4 ]]; then + echo "usage: $0 OUT_DIR MODEL_NAME OUTPUT_STEM GALLERY_GLOB QUERY_GLOB..." >&2 + exit 2 +fi + +OUT_DIR=$1 +MODEL_NAME=$2 +OUTPUT_STEM=$3 +GALLERY_GLOB=$4 +shift 4 + +ROOT=${ROOT:-/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal} +NGPU=${NGPU:-8} +QUERY_BATCH=${QUERY_BATCH:-512} +TOPK=${TOPK:-1} +MATCHDIR="$OUT_DIR/matches" +LOGDIR="$OUT_DIR/logs" +mkdir -p "$MATCHDIR" "$LOGDIR" + +MERGED_TSV="$MATCHDIR/${OUTPUT_STEM}.tsv" +SUMMARY_TSV="$MATCHDIR/${OUTPUT_STEM}.summary.tsv" +DRIVER_LOG="$LOGDIR/${OUTPUT_STEM}.driver.log" + +if [[ -f "${MERGED_TSV%.tsv}.metadata.json" && -f "$SUMMARY_TSV" ]]; then + echo "$(date -u +%FT%TZ) skip existing: $MERGED_TSV" + exit 0 +fi + +mapfile -t GALLERY_PREFIXES < <(compgen -G "$GALLERY_GLOB" | sed -E 's/\.(embeddings\.npy|meta\.tsv|metadata\.json)$//' | sort -u) +QUERY_PREFIXES=() +for query_glob in "$@"; do + while IFS= read -r prefix; do + QUERY_PREFIXES+=("$prefix") + done < <(compgen -G "$query_glob" | sed -E 's/\.(embeddings\.npy|meta\.tsv|metadata\.json)$//' | sort -u) +done + +if [[ ${#GALLERY_PREFIXES[@]} -eq 0 ]]; then + echo "no gallery prefixes matched: $GALLERY_GLOB" >&2 + exit 1 +fi +if [[ ${#QUERY_PREFIXES[@]} -eq 0 ]]; then + echo "no query prefixes matched" >&2 + exit 1 +fi + +case "$MODEL_NAME" in + sscd) + SCORE_THRESHOLD=${SCORE_THRESHOLD:-0.75} + L2SQ_THRESHOLD=${L2SQ_THRESHOLD:-0.5} + ;; + arcface) + SCORE_THRESHOLD=${SCORE_THRESHOLD:-0.5} + L2SQ_THRESHOLD=${L2SQ_THRESHOLD:-1.0} + ;; + *) + echo "unknown model name: $MODEL_NAME" >&2 + exit 2 + ;; +esac + +export PYTHONPATH="$ROOT/python_dedup_pkgs:${PYTHONPATH:-}" +echo "$(date -u +%FT%TZ) exact torch start: model=$MODEL_NAME output=$MERGED_TSV gallery=${#GALLERY_PREFIXES[@]} query=${#QUERY_PREFIXES[@]} ngpu=$NGPU batch=$QUERY_BATCH" | tee "$DRIVER_LOG" + +pids=() +parts=() +for gpu in $(seq 0 $((NGPU - 1))); do + assigned_gallery=() + for idx in "${!GALLERY_PREFIXES[@]}"; do + if (( idx % NGPU == gpu )); then + assigned_gallery+=("${GALLERY_PREFIXES[$idx]}") + fi + done + if [[ ${#assigned_gallery[@]} -eq 0 ]]; then + continue + fi + PART_TSV="$MATCHDIR/${OUTPUT_STEM}.galleryshard${gpu}.tsv" + LOG="$LOGDIR/${OUTPUT_STEM}.galleryshard${gpu}.log" + parts+=("$PART_TSV") + CMD=(python3 "$ROOT/scaledf_embedding_audit.py" match + --out "$PART_TSV" + --engine torch + --device "cuda:$gpu" + --query-batch "$QUERY_BATCH" + --topk "$TOPK" + --score-threshold "$SCORE_THRESHOLD" + --l2sq-threshold "$L2SQ_THRESHOLD") + for prefix in "${assigned_gallery[@]}"; do + CMD+=(--gallery-prefix "$prefix") + done + for prefix in "${QUERY_PREFIXES[@]}"; do + CMD+=(--query-prefix "$prefix") + done + echo "$(date -u +%FT%TZ) launch gpu=$gpu gallery=${#assigned_gallery[@]} part=$PART_TSV" | tee -a "$DRIVER_LOG" + "${CMD[@]}" > "$LOG" 2>&1 & + pids+=("$!") +done + +status=0 +for pid in "${pids[@]}"; do + if ! wait "$pid"; then + status=1 + fi +done +if [[ $status -ne 0 ]]; then + echo "$(date -u +%FT%TZ) exact torch failed: model=$MODEL_NAME output=$MERGED_TSV" | tee -a "$DRIVER_LOG" + exit "$status" +fi + +MERGE_CMD=(python3 "$ROOT/scaledf_embedding_audit.py" merge-matches --out "$MERGED_TSV") +SUMMARY_CMD=(python3 "$ROOT/scaledf_embedding_audit.py" summarize-matches --out "$SUMMARY_TSV" --score-threshold "$SCORE_THRESHOLD" --l2sq-threshold "$L2SQ_THRESHOLD") +for part in "${parts[@]}"; do + MERGE_CMD+=(--match-tsv "$part") +done +"${MERGE_CMD[@]}" >> "$DRIVER_LOG" 2>&1 +"${SUMMARY_CMD[@]}" --match-tsv "$MERGED_TSV" >> "$DRIVER_LOG" 2>&1 +echo "$(date -u +%FT%TZ) exact torch done: model=$MODEL_NAME output=$MERGED_TSV summary=$SUMMARY_TSV" | tee -a "$DRIVER_LOG" diff --git a/ScaleDF_rebuttal/launch_scaledf_match_index.sh b/ScaleDF_rebuttal/launch_scaledf_match_index.sh new file mode 100644 index 0000000000000000000000000000000000000000..f5f4a5cb7e516eced170537aa9b1871bbb58d907 --- /dev/null +++ b/ScaleDF_rebuttal/launch_scaledf_match_index.sh @@ -0,0 +1,81 @@ +#!/usr/bin/env bash +set -euo pipefail + +if [[ $# -lt 4 ]]; then + echo "usage: $0 OUT_DIR MODEL_NAME OUTPUT_STEM GALLERY_GLOB QUERY_GLOB..." >&2 + exit 2 +fi + +OUT_DIR=$1 +MODEL_NAME=$2 +OUTPUT_STEM=$3 +GALLERY_GLOB=$4 +shift 4 + +ROOT=${ROOT:-/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal} +MATCHDIR="$OUT_DIR/matches" +LOGDIR="$OUT_DIR/logs" +mkdir -p "$MATCHDIR" "$LOGDIR" + +OUT_TSV="$MATCHDIR/${OUTPUT_STEM}.tsv" +LOG="$LOGDIR/${OUTPUT_STEM}.log" + +if [[ -f "${OUT_TSV%.tsv}.metadata.json" ]]; then + echo "$(date -u +%FT%TZ) skip existing: ${OUT_TSV%.tsv}.metadata.json" + exit 0 +fi + +mapfile -t GALLERY_PREFIXES < <(compgen -G "$GALLERY_GLOB" | sed -E 's/\.(embeddings\.npy|meta\.tsv|metadata\.json)$//' | sort -u) +QUERY_PREFIXES=() +for query_glob in "$@"; do + while IFS= read -r prefix; do + QUERY_PREFIXES+=("$prefix") + done < <(compgen -G "$query_glob" | sed -E 's/\.(embeddings\.npy|meta\.tsv|metadata\.json)$//' | sort -u) +done + +if [[ ${#GALLERY_PREFIXES[@]} -eq 0 ]]; then + echo "no gallery prefixes matched: $GALLERY_GLOB" >&2 + exit 1 +fi +if [[ ${#QUERY_PREFIXES[@]} -eq 0 ]]; then + echo "no query prefixes matched" >&2 + exit 1 +fi + +CMD=(python3 "$ROOT/scaledf_embedding_audit.py" match-index + --out "$OUT_TSV" + --index-factory "${INDEX_FACTORY:-HNSW32}" + --hnsw-ef-search "${HNSW_EF_SEARCH:-256}" + --hnsw-ef-construction "${HNSW_EF_CONSTRUCTION:-200}" + --nprobe "${NPROBE:-64}" + --train-sample "${TRAIN_SAMPLE:-500000}" + --add-batch "${ADD_BATCH:-200000}" + --query-batch "${QUERY_BATCH:-4096}" + --status-every "${STATUS_EVERY:-20}" + --omp-threads "${OMP_THREADS:-128}" + --topk "${TOPK:-1}") + +case "$MODEL_NAME" in + sscd) + CMD+=(--score-threshold "${SCORE_THRESHOLD:-0.75}" --l2sq-threshold "${L2SQ_THRESHOLD:-0.5}") + ;; + arcface) + CMD+=(--score-threshold "${SCORE_THRESHOLD:-0.5}" --l2sq-threshold "${L2SQ_THRESHOLD:-1.0}") + ;; + *) + echo "unknown model name: $MODEL_NAME" >&2 + exit 2 + ;; +esac + +for prefix in "${GALLERY_PREFIXES[@]}"; do + CMD+=(--gallery-prefix "$prefix") +done +for prefix in "${QUERY_PREFIXES[@]}"; do + CMD+=(--query-prefix "$prefix") +done + +export PYTHONPATH="$ROOT/python_dedup_pkgs:${PYTHONPATH:-}" +echo "$(date -u +%FT%TZ) match-index start: model=$MODEL_NAME output=$OUT_TSV gallery=${#GALLERY_PREFIXES[@]} query=${#QUERY_PREFIXES[@]}" | tee "$LOG" +"${CMD[@]}" >> "$LOG" 2>&1 +echo "$(date -u +%FT%TZ) match-index done: model=$MODEL_NAME output=$OUT_TSV" | tee -a "$LOG" diff --git a/ScaleDF_rebuttal/launch_scaledf_predict_scores.sh b/ScaleDF_rebuttal/launch_scaledf_predict_scores.sh new file mode 100644 index 0000000000000000000000000000000000000000..a8ebc37ae0d784627c3a5a107c7c978b2ee7a44e --- /dev/null +++ b/ScaleDF_rebuttal/launch_scaledf_predict_scores.sh @@ -0,0 +1,108 @@ +#!/usr/bin/env bash +set -euo pipefail + +if [[ $# -lt 2 ]]; then + echo "usage: $0 CHECKPOINT OUTPUT_ROOT [scaledf|benchmarks|all] [MODEL_NAME]" >&2 + exit 2 +fi + +CHECKPOINT=$1 +OUTPUT_ROOT=$2 +MODE=${3:-all} +MODEL_NAME=${4:-deit_base_patch16_LS} + +ROOT=${ROOT:-/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal} +PACKAGE_ROOT=${PACKAGE_ROOT:-/tmp/ScaleDF_repro_20260724} +DATA_ROOT=${SCALEDF_DATA:-/dev/shm/scaledf_bind_storage/ScaleDF} +BENCHMARK_ROOT=${BENCHMARK_ROOT:-/dev/shm/scaledf_bind_storage/Established_benchmarks} +NGPU=${NGPU:-8} +BATCH_SIZE=${BATCH_SIZE:-384} +WORKERS=${WORKERS:-12} +MAX_IMAGES_ARG=() +if [[ -n "${MAX_IMAGES:-}" ]]; then + MAX_IMAGES_ARG=(--max-images "$MAX_IMAGES") +fi + +mkdir -p "$OUTPUT_ROOT/logs" "$OUTPUT_ROOT/predictions" +DRIVER_LOG="$OUTPUT_ROOT/logs/predict_scores.driver.log" + +DATASETS=() +case "$MODE" in + scaledf|all) + DATASETS+=(--dataset "ScaleDF_val:$DATA_ROOT/val") + ;;& + benchmarks|all) + for name in CDFv2 DF40 DeepFakeDetection DeepFakeFace ForgeryNet Wild_Deepfake; do + DATASETS+=(--dataset "$name:$BENCHMARK_ROOT/$name/val") + done + ;; + scaledf|benchmarks|all) + ;; + *) + echo "unknown mode: $MODE" >&2 + exit 2 + ;; +esac + +if [[ -f "$OUTPUT_ROOT/summary.tsv" ]]; then + echo "$(date -u +%FT%TZ) skip existing: $OUTPUT_ROOT/summary.tsv" + exit 0 +fi + +export PYTHONPATH="/run/tiger/python_stubs:$PACKAGE_ROOT/code/deit:${PYTHONPATH:-}" +export PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION="${PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION:-python}" +export WANDB_DISABLED="${WANDB_DISABLED:-true}" +export NO_ALBUMENTATIONS_UPDATE="${NO_ALBUMENTATIONS_UPDATE:-1}" +export PYTHONUNBUFFERED=1 + +{ + echo "started_utc=$(date -u +%FT%TZ)" + echo "hostname=$(hostname)" + echo "checkpoint=$CHECKPOINT" + echo "output_root=$OUTPUT_ROOT" + echo "mode=$MODE" + echo "model_name=$MODEL_NAME" + echo "ngpu=$NGPU" + echo "batch_size=$BATCH_SIZE" + echo "workers=$WORKERS" +} | tee "$DRIVER_LOG" + +pids=() +parts=() +for gpu in $(seq 0 $((NGPU - 1))); do + out="$OUTPUT_ROOT/predictions/predictions_shard$(printf '%02d' "$gpu")_of$(printf '%02d' "$NGPU").tsv" + log="$OUTPUT_ROOT/logs/predict_scores_shard$(printf '%02d' "$gpu").log" + parts+=("$out") + python3 "$ROOT/scaledf_predict_scores.py" predict \ + "${DATASETS[@]}" \ + --checkpoint "$CHECKPOINT" \ + --model "$MODEL_NAME" \ + --code-root "$PACKAGE_ROOT/code/deit" \ + --out "$out" \ + --device "cuda:$gpu" \ + --batch-size "$BATCH_SIZE" \ + --workers "$WORKERS" \ + --shard-index "$gpu" \ + --num-shards "$NGPU" \ + "${MAX_IMAGES_ARG[@]}" > "$log" 2>&1 & + pids+=("$!") + echo "$(date -u +%FT%TZ) launch gpu=$gpu out=$out" | tee -a "$DRIVER_LOG" +done + +status=0 +for pid in "${pids[@]}"; do + if ! wait "$pid"; then + status=1 + fi +done +if [[ $status -ne 0 ]]; then + echo "$(date -u +%FT%TZ) prediction failed" | tee -a "$DRIVER_LOG" + exit "$status" +fi + +CMD=(python3 "$ROOT/scaledf_predict_scores.py" summarize --out "$OUTPUT_ROOT/summary.tsv") +for part in "${parts[@]}"; do + CMD+=(--prediction-tsv "$part") +done +"${CMD[@]}" >> "$DRIVER_LOG" 2>&1 +echo "$(date -u +%FT%TZ) prediction done summary=$OUTPUT_ROOT/summary.tsv" | tee -a "$DRIVER_LOG" diff --git a/ScaleDF_rebuttal/launch_scaledf_sscd_stage.sh b/ScaleDF_rebuttal/launch_scaledf_sscd_stage.sh new file mode 100644 index 0000000000000000000000000000000000000000..5f81825e579f47f85b736dad32838e5e61a67d1f --- /dev/null +++ b/ScaleDF_rebuttal/launch_scaledf_sscd_stage.sh @@ -0,0 +1,62 @@ +#!/usr/bin/env bash +set -euo pipefail + +ROOT=${ROOT:-/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal} +OUT=${1:?usage: launch_scaledf_sscd_stage.sh OUT PREFIX STAGE} +PREFIX=${2:?usage: launch_scaledf_sscd_stage.sh OUT PREFIX STAGE} +STAGE=${3:?usage: launch_scaledf_sscd_stage.sh OUT PREFIX STAGE} +NUM_SHARDS=${NUM_SHARDS:-8} +SSCD_BATCH=${SSCD_BATCH:-512} +SSCD_WORKERS=${SSCD_WORKERS:-12} + +SCRIPT=$ROOT/scaledf_embedding_audit.py +MODEL=$ROOT/models/sscd_disc_mixup.torchscript.pt +MANIFEST_DIR=$OUT/manifests +EMB_DIR=$OUT/embeddings +LOG_DIR=$OUT/logs + +mkdir -p "$EMB_DIR" "$LOG_DIR" + +log() { + printf '%s %s\n' "$(date -u +%Y-%m-%dT%H:%M:%SZ)" "$*" +} + +log "sscd stage start: stage=$STAGE prefix=$PREFIX out=$OUT batch=$SSCD_BATCH workers=$SSCD_WORKERS" +pids=() +fail=0 +for shard in $(seq 0 $((NUM_SHARDS - 1))); do + shard_name=$(printf "%s_shard%02d_of%02d" "$PREFIX" "$shard" "$NUM_SHARDS") + manifest="$MANIFEST_DIR/${shard_name}.tsv" + out_prefix="$EMB_DIR/${STAGE}_${shard_name}" + log_file="$LOG_DIR/${STAGE}_${shard_name}.log" + if [[ -s "${out_prefix}.embeddings.npy" && -s "${out_prefix}.meta.tsv" ]]; then + log "skip exists: shard=$shard out=$out_prefix" + continue + fi + ( + set -euo pipefail + export CUDA_VISIBLE_DEVICES=$shard + python3 "$SCRIPT" embed-sscd \ + --manifest "$manifest" \ + --model "$MODEL" \ + --out-prefix "$out_prefix" \ + --device cuda:0 \ + --batch-size "$SSCD_BATCH" \ + --workers "$SSCD_WORKERS" \ + --status-every 25 + ) > "$log_file" 2>&1 & + pids+=("$!") + log "launched shard=$shard pid=$!" +done + +for pid in "${pids[@]}"; do + if ! wait "$pid"; then + fail=1 + fi +done + +if [[ "$fail" != 0 ]]; then + log "sscd stage failed: stage=$STAGE prefix=$PREFIX" + exit 1 +fi +log "sscd stage done: stage=$STAGE prefix=$PREFIX" diff --git a/ScaleDF_rebuttal/launch_vith_16card_resume_node.sh b/ScaleDF_rebuttal/launch_vith_16card_resume_node.sh new file mode 100644 index 0000000000000000000000000000000000000000..9ae8c9e94a5eea598214e7f40da2c3cc318a2362 --- /dev/null +++ b/ScaleDF_rebuttal/launch_vith_16card_resume_node.sh @@ -0,0 +1,110 @@ +#!/usr/bin/env bash +set -euo pipefail + +ROOT="${PERSIST_ROOT:-/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal}" +PACKAGE_ROOT="${PACKAGE_ROOT:-/tmp/ScaleDF_repro_20260724}" +DATA_ROOT="${SCALEDF_DATA:-/dev/shm/scaledf_bind_storage/ScaleDF}" +HOST="$(hostname)" + +RUN_NAME="${RUN_NAME:-modelsize_vith_40epoch_effbs2048_16card_${HOST}_$(date -u +%Y%m%d_%H%M%S)}" +OUTPUT_ROOT="${OUTPUT_ROOT:-$ROOT/runs/$RUN_NAME}" +LOG_FILE="${LOG_FILE:-$OUTPUT_ROOT/train_node${NODE_RANK}_${HOST}.log}" + +MAIN="${MAIN:-main_cross_2direct_balance_realscale_perturb_v4_accum.py}" +MODEL_NAME="${MODEL_NAME:-deit_huge_patch14_LS}" +PRETRAIN="${PRETRAIN:-$PACKAGE_ROOT/pretrained/deit_3_huge_224_21k_v1.pth}" +RESUME="${RESUME:-}" + +NNODES="${NNODES:-2}" +NPROC_PER_NODE="${NPROC_PER_NODE:-8}" +NODE_RANK="${NODE_RANK:?NODE_RANK is required}" +MASTER_ADDR="${MASTER_ADDR:?MASTER_ADDR is required}" +MASTER_PORT="${MASTER_PORT:-29635}" + +BATCH_PER_GPU="${BATCH_PER_GPU:-64}" +ACCUM_ITER="${ACCUM_ITER:-2}" +EVAL_BATCH_PER_GPU="${EVAL_BATCH_PER_GPU:-64}" +EPOCHS="${EPOCHS:-40}" +WORLD_SIZE=$((NNODES * NPROC_PER_NODE)) +EFFECTIVE_BATCH=$((BATCH_PER_GPU * WORLD_SIZE * ACCUM_ITER)) + +mkdir -p "$OUTPUT_ROOT" + +export PYTHONPATH="/run/tiger/python_stubs:$PACKAGE_ROOT/code/deit:${PYTHONPATH:-}" +export PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION="${PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION:-python}" +export WANDB_DISABLED="${WANDB_DISABLED:-true}" +export NO_ALBUMENTATIONS_UPDATE="${NO_ALBUMENTATIONS_UPDATE:-1}" +export NCCL_DEBUG="${SCALEDF_NCCL_DEBUG:-${NCCL_DEBUG:-WARN}}" +export NCCL_SOCKET_IFNAME="${NCCL_SOCKET_IFNAME:-eth0}" +export NCCL_SOCKET_FAMILY="${NCCL_SOCKET_FAMILY:-AF_INET}" +export PYTHONUNBUFFERED="${PYTHONUNBUFFERED:-1}" + +ulimit -n 1048576 >/dev/null 2>&1 || true + +cd "$PACKAGE_ROOT/code/deit" + +{ + echo "started_utc=$(date -u +%Y-%m-%dT%H:%M:%SZ)" + echo "hostname=$HOST" + echo "node_rank=$NODE_RANK" + echo "nnodes=$NNODES" + echo "nproc_per_node=$NPROC_PER_NODE" + echo "world_size=$WORLD_SIZE" + echo "master_addr=$MASTER_ADDR" + echo "master_port=$MASTER_PORT" + echo "package_root=$PACKAGE_ROOT" + echo "data_root=$DATA_ROOT" + echo "main=$MAIN" + echo "model_name=$MODEL_NAME" + echo "pretrain=$PRETRAIN" + echo "resume=$RESUME" + echo "output_root=$OUTPUT_ROOT" + echo "batch_per_gpu=$BATCH_PER_GPU" + echo "accum_iter=$ACCUM_ITER" + echo "effective_batch=$EFFECTIVE_BATCH" + echo "eval_batch_per_gpu=$EVAL_BATCH_PER_GPU" + echo "epochs=$EPOCHS" +} | tee -a "$LOG_FILE" + +extra_checkpoint_args=() +if [[ -n "$RESUME" ]]; then + extra_checkpoint_args=(--resume "$RESUME") +else + extra_checkpoint_args=(--finetune "$PRETRAIN") +fi + +torchrun \ + --nnodes="$NNODES" \ + --nproc_per_node="$NPROC_PER_NODE" \ + --node_rank="$NODE_RANK" \ + --master_addr="$MASTER_ADDR" \ + --master_port="$MASTER_PORT" \ + "$MAIN" \ + --model "$MODEL_NAME" \ + --data-path "$DATA_ROOT" \ + --batch-size "$BATCH_PER_GPU" \ + --accum-iter "$ACCUM_ITER" \ + --eval-batch-size "$EVAL_BATCH_PER_GPU" \ + --lr 1e-5 \ + --epochs "$EPOCHS" \ + --weight-decay 0.1 \ + --sched cosine \ + --input-size 224 \ + --eval-crop-ratio 1.0 \ + --reprob 0.0 \ + --smoothing 0.1 \ + --warmup-epochs 5 \ + --drop 0.0 \ + --data-set ScaleDF \ + --seed "${SEED:-0}" \ + --opt adamw \ + --warmup-lr 1e-6 \ + --mixup 0.8 \ + --drop-path 0.2 \ + --cutmix 1.0 \ + --unscale-lr \ + --aa rand-m9-mstd0.5-inc1 \ + --no-repeated-aug \ + "${extra_checkpoint_args[@]}" \ + --output_dir "$OUTPUT_ROOT" \ + --two-cls 2>&1 | tee -a "$LOG_FILE" diff --git a/ScaleDF_rebuttal/launch_vith_32card_resume_node.sh b/ScaleDF_rebuttal/launch_vith_32card_resume_node.sh new file mode 100644 index 0000000000000000000000000000000000000000..a400b52677bda4ae5fc204654e4b0bcfb737fea2 --- /dev/null +++ b/ScaleDF_rebuttal/launch_vith_32card_resume_node.sh @@ -0,0 +1,103 @@ +#!/usr/bin/env bash +set -euo pipefail + +ROOT="${PERSIST_ROOT:-/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal}" +PACKAGE_ROOT="${PACKAGE_ROOT:-/tmp/ScaleDF_repro_20260724}" +DATA_ROOT="${SCALEDF_DATA:-/dev/shm/scaledf_bind_storage/ScaleDF}" +HOST="$(hostname)" + +RUN_NAME="${RUN_NAME:-modelsize_vith_40epoch_effbs2048_32card_${HOST}_$(date -u +%Y%m%d_%H%M%S)}" +OUTPUT_ROOT="${OUTPUT_ROOT:-$ROOT/runs/$RUN_NAME}" +LOG_FILE="${LOG_FILE:-$OUTPUT_ROOT/train_node${NODE_RANK}_${HOST}.log}" + +MAIN="${MAIN:-main_cross_2direct_balance_realscale_perturb_v4_accum.py}" +MODEL_NAME="${MODEL_NAME:-deit_huge_patch14_LS}" +PRETRAIN="${PRETRAIN:-$PACKAGE_ROOT/pretrained/deit_3_huge_224_21k_v1.pth}" +RESUME="${RESUME:?RESUME checkpoint is required}" + +NNODES="${NNODES:-4}" +NPROC_PER_NODE="${NPROC_PER_NODE:-8}" +NODE_RANK="${NODE_RANK:?NODE_RANK is required}" +MASTER_ADDR="${MASTER_ADDR:?MASTER_ADDR is required}" +MASTER_PORT="${MASTER_PORT:-29647}" + +BATCH_PER_GPU="${BATCH_PER_GPU:-64}" +ACCUM_ITER="${ACCUM_ITER:-1}" +EVAL_BATCH_PER_GPU="${EVAL_BATCH_PER_GPU:-64}" +EPOCHS="${EPOCHS:-40}" +WORLD_SIZE=$((NNODES * NPROC_PER_NODE)) +EFFECTIVE_BATCH=$((BATCH_PER_GPU * WORLD_SIZE * ACCUM_ITER)) + +mkdir -p "$OUTPUT_ROOT" + +export PYTHONPATH="/run/tiger/python_stubs:$PACKAGE_ROOT/code/deit:${PYTHONPATH:-}" +export PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION="${PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION:-python}" +export WANDB_DISABLED="${WANDB_DISABLED:-true}" +export NO_ALBUMENTATIONS_UPDATE="${NO_ALBUMENTATIONS_UPDATE:-1}" +export NCCL_DEBUG="${SCALEDF_NCCL_DEBUG:-${NCCL_DEBUG:-WARN}}" +export NCCL_SOCKET_IFNAME="${NCCL_SOCKET_IFNAME:-eth0}" +export NCCL_SOCKET_FAMILY="${NCCL_SOCKET_FAMILY:-AF_INET}" +export PYTHONUNBUFFERED="${PYTHONUNBUFFERED:-1}" + +ulimit -n 1048576 >/dev/null 2>&1 || true + +cd "$PACKAGE_ROOT/code/deit" + +{ + echo "started_utc=$(date -u +%Y-%m-%dT%H:%M:%SZ)" + echo "hostname=$HOST" + echo "node_rank=$NODE_RANK" + echo "nnodes=$NNODES" + echo "nproc_per_node=$NPROC_PER_NODE" + echo "world_size=$WORLD_SIZE" + echo "master_addr=$MASTER_ADDR" + echo "master_port=$MASTER_PORT" + echo "package_root=$PACKAGE_ROOT" + echo "data_root=$DATA_ROOT" + echo "main=$MAIN" + echo "model_name=$MODEL_NAME" + echo "pretrain=$PRETRAIN" + echo "resume=$RESUME" + echo "output_root=$OUTPUT_ROOT" + echo "batch_per_gpu=$BATCH_PER_GPU" + echo "accum_iter=$ACCUM_ITER" + echo "effective_batch=$EFFECTIVE_BATCH" + echo "eval_batch_per_gpu=$EVAL_BATCH_PER_GPU" + echo "epochs=$EPOCHS" +} | tee -a "$LOG_FILE" + +torchrun \ + --nnodes="$NNODES" \ + --nproc_per_node="$NPROC_PER_NODE" \ + --node_rank="$NODE_RANK" \ + --master_addr="$MASTER_ADDR" \ + --master_port="$MASTER_PORT" \ + "$MAIN" \ + --model "$MODEL_NAME" \ + --data-path "$DATA_ROOT" \ + --batch-size "$BATCH_PER_GPU" \ + --accum-iter "$ACCUM_ITER" \ + --eval-batch-size "$EVAL_BATCH_PER_GPU" \ + --lr 1e-5 \ + --epochs "$EPOCHS" \ + --weight-decay 0.1 \ + --sched cosine \ + --input-size 224 \ + --eval-crop-ratio 1.0 \ + --reprob 0.0 \ + --smoothing 0.1 \ + --warmup-epochs 5 \ + --drop 0.0 \ + --data-set ScaleDF \ + --seed "${SEED:-0}" \ + --opt adamw \ + --warmup-lr 1e-6 \ + --mixup 0.8 \ + --drop-path 0.2 \ + --cutmix 1.0 \ + --unscale-lr \ + --aa rand-m9-mstd0.5-inc1 \ + --no-repeated-aug \ + --resume "$RESUME" \ + --output_dir "$OUTPUT_ROOT" \ + --two-cls 2>&1 | tee -a "$LOG_FILE" diff --git a/ScaleDF_rebuttal/magface_quality_eval.py b/ScaleDF_rebuttal/magface_quality_eval.py new file mode 100644 index 0000000000000000000000000000000000000000..4ccbd8e4ae44e7286e0478e9d30c2ea7bf74b760 --- /dev/null +++ b/ScaleDF_rebuttal/magface_quality_eval.py @@ -0,0 +1,450 @@ +#!/usr/bin/env python3 +"""MagFace face-quality scoring for ScaleDF fake/manipulated faces.""" + +from __future__ import annotations + +import argparse +import csv +import json +import math +import os +import sys +import time +from collections import defaultdict +from pathlib import Path +from typing import Iterable + +import numpy as np +import torch +from torch.utils.data import DataLoader, Dataset + + +IMAGE_EXTS = {".jpg", ".jpeg", ".png", ".bmp", ".webp"} + + +def now() -> str: + return time.strftime("%Y-%m-%dT%H:%M:%SZ", time.gmtime()) + + +def is_fake_method_dir(path: Path) -> bool: + return path.is_dir() and not path.name.startswith("000000") + + +def iter_images(root: Path, method: str) -> Iterable[tuple[str, Path]]: + stack = [root / method] + while stack: + current = stack.pop() + try: + entries = sorted(os.scandir(current), key=lambda e: e.name) + except OSError as exc: + print(f"warn: cannot scan {current}: {exc}", file=sys.stderr, flush=True) + continue + for entry in entries: + try: + if entry.is_dir(follow_symlinks=False): + stack.append(Path(entry.path)) + elif entry.is_file(follow_symlinks=False) and Path(entry.name).suffix.lower() in IMAGE_EXTS: + path = Path(entry.path) + yield method, path + except OSError as exc: + print(f"warn: cannot stat {entry.path}: {exc}", file=sys.stderr, flush=True) + + +def command_make_shards(args: argparse.Namespace) -> int: + root = args.root + out_dir = args.out_dir + out_dir.mkdir(parents=True, exist_ok=True) + methods = sorted([p.name for p in root.iterdir() if is_fake_method_dir(p)]) + writers = [] + files = [] + counts = [0 for _ in range(args.num_shards)] + method_counts: dict[str, int] = defaultdict(int) + fields = ["idx", "method", "relpath", "path"] + try: + for shard in range(args.num_shards): + path = out_dir / f"{args.prefix}_shard{shard:02d}_of{args.num_shards:02d}.tsv" + f = path.open("w", newline="", encoding="utf-8") + files.append(f) + writer = csv.DictWriter(f, fieldnames=fields, delimiter="\t") + writer.writeheader() + writers.append(writer) + idx = 0 + for method in methods: + for method_name, path in iter_images(root, method): + shard = idx % args.num_shards + rel = path.relative_to(root).as_posix() + writers[shard].writerow( + { + "idx": idx, + "method": method_name, + "relpath": rel, + "path": str(path), + } + ) + counts[shard] += 1 + method_counts[method_name] += 1 + idx += 1 + if idx % args.print_freq == 0: + print(f"{now()} shard_rows={idx}", flush=True) + finally: + for f in files: + f.close() + + with (out_dir / f"{args.prefix}_split_summary.tsv").open("w", newline="", encoding="utf-8") as f: + writer = csv.DictWriter(f, fieldnames=["shard", "num_shards", "rows", "path"], delimiter="\t") + writer.writeheader() + for shard, count in enumerate(counts): + writer.writerow( + { + "shard": shard, + "num_shards": args.num_shards, + "rows": count, + "path": str(out_dir / f"{args.prefix}_shard{shard:02d}_of{args.num_shards:02d}.tsv"), + } + ) + with (out_dir / f"{args.prefix}_method_counts.tsv").open("w", newline="", encoding="utf-8") as f: + writer = csv.DictWriter(f, fieldnames=["method", "images"], delimiter="\t") + writer.writeheader() + for method, count in sorted(method_counts.items()): + writer.writerow({"method": method, "images": count}) + metadata = { + "created_utc": now(), + "root": str(root), + "methods": len(methods), + "rows": sum(counts), + "num_shards": args.num_shards, + "image_exts": sorted(IMAGE_EXTS), + } + (out_dir / f"{args.prefix}_metadata.json").write_text(json.dumps(metadata, indent=2) + "\n", encoding="utf-8") + print(f"rows={sum(counts)} methods={len(methods)}") + print(out_dir / f"{args.prefix}_split_summary.tsv") + return 0 + + +class ShardDataset(Dataset): + def __init__(self, shard_path: Path, image_size: int = 112): + self.rows: list[tuple[str, str, str]] = [] + with shard_path.open(newline="", encoding="utf-8") as f: + for row in csv.DictReader(f, delimiter="\t"): + self.rows.append((row["method"], row["relpath"], row["path"])) + self.image_size = image_size + + def __len__(self) -> int: + return len(self.rows) + + def __getitem__(self, index: int): + import cv2 + + method, relpath, path = self.rows[index] + img = cv2.imread(path, cv2.IMREAD_COLOR) + if img is None: + return None, method, relpath, path, "decode_failed" + if img.shape[0] != self.image_size or img.shape[1] != self.image_size: + img = cv2.resize(img, (self.image_size, self.image_size), interpolation=cv2.INTER_AREA) + # Official MagFace inference uses cv2 BGR images with ToTensor and [0,1] scaling. + arr = np.ascontiguousarray(img.transpose(2, 0, 1)) + tensor = torch.from_numpy(arr).float().div_(255.0) + return tensor, method, relpath, path, "" + + +def collate_batch(batch): + good = [item for item in batch if item[0] is not None] + bad = [item for item in batch if item[0] is None] + if good: + images = torch.stack([item[0] for item in good], dim=0) + methods = [item[1] for item in good] + relpaths = [item[2] for item in good] + paths = [item[3] for item in good] + else: + images = torch.empty(0, 3, 112, 112) + methods, relpaths, paths = [], [], [] + bad_rows = [{"method": item[1], "relpath": item[2], "path": item[3], "error": item[4]} for item in bad] + return images, methods, relpaths, paths, bad_rows + + +def load_magface_model(repo: Path, checkpoint: Path, arch: str, device: torch.device) -> torch.nn.Module: + sys.path.insert(0, str(repo)) + from models import iresnet + + if arch == "iresnet18": + model = iresnet.iresnet18(pretrained=False, num_classes=512) + elif arch == "iresnet50": + model = iresnet.iresnet50(pretrained=False, num_classes=512) + elif arch == "iresnet100": + model = iresnet.iresnet100(pretrained=False, num_classes=512) + else: + raise ValueError(f"unsupported arch: {arch}") + + ckpt = torch.load(checkpoint, map_location="cpu") + state = ckpt["state_dict"] if isinstance(ckpt, dict) and "state_dict" in ckpt else ckpt + cleaned = {} + for key, value in state.items(): + if key.startswith("features.module."): + cleaned[key[len("features.module.") :]] = value + elif key.startswith("module.features."): + cleaned[key[len("module.features.") :]] = value + elif key.startswith("features."): + cleaned[key[len("features.") :]] = value + elif key in model.state_dict() and value.shape == model.state_dict()[key].shape: + cleaned[key] = value + missing, unexpected = model.load_state_dict(cleaned, strict=False) + if missing or unexpected: + raise RuntimeError(f"MagFace load mismatch: missing={len(missing)} unexpected={len(unexpected)}") + model.to(device).eval() + return model + + +def command_score(args: argparse.Namespace) -> int: + local_rank = int(os.environ.get("LOCAL_RANK", os.environ.get("RANK", "0"))) + world_size = int(os.environ.get("WORLD_SIZE", "1")) + rank = int(os.environ.get("RANK", str(local_rank))) + torch.cuda.set_device(local_rank) + device = torch.device(f"cuda:{local_rank}") + torch.backends.cudnn.benchmark = True + + shard_path = args.shard_dir / f"{args.prefix}_shard{rank:02d}_of{world_size:02d}.tsv" + if not shard_path.exists(): + raise FileNotFoundError(shard_path) + + out_dir = args.out_dir + out_dir.mkdir(parents=True, exist_ok=True) + score_path = out_dir / f"{args.out_prefix}_shard{rank:02d}_of{world_size:02d}.scores.tsv" + bad_path = out_dir / f"{args.out_prefix}_shard{rank:02d}_of{world_size:02d}.bad.tsv" + meta_path = out_dir / f"{args.out_prefix}_shard{rank:02d}_of{world_size:02d}.metadata.json" + + model = load_magface_model(args.magface_repo, args.checkpoint, args.arch, device) + dataset = ShardDataset(shard_path, image_size=args.image_size) + loader = DataLoader( + dataset, + batch_size=args.batch_size, + shuffle=False, + num_workers=args.workers, + pin_memory=True, + persistent_workers=args.workers > 0, + prefetch_factor=args.prefetch_factor if args.workers > 0 else None, + collate_fn=collate_batch, + ) + + method_stats: dict[str, dict[str, float]] = defaultdict(lambda: {"n": 0, "sum": 0.0, "sum2": 0.0, "min": float("inf"), "max": float("-inf")}) + started = time.time() + processed = 0 + bad = 0 + with score_path.open("w", newline="", encoding="utf-8") as sf, bad_path.open("w", newline="", encoding="utf-8") as bf: + score_writer = csv.DictWriter(sf, fieldnames=["method", "quality"], delimiter="\t") + bad_writer = csv.DictWriter(bf, fieldnames=["method", "relpath", "path", "error"], delimiter="\t") + score_writer.writeheader() + bad_writer.writeheader() + amp_dtype = torch.float16 if args.amp else torch.float32 + with torch.inference_mode(): + for step, (images, methods, _relpaths, _paths, bad_rows) in enumerate(loader): + for row in bad_rows: + bad_writer.writerow(row) + bad += len(bad_rows) + if images.numel() == 0: + continue + images = images.to(device, non_blocking=True) + with torch.autocast("cuda", enabled=args.amp, dtype=amp_dtype): + feats = model(images) + qualities = feats.float().norm(p=2, dim=1).detach().cpu().numpy() + for method, quality in zip(methods, qualities): + q = float(quality) + score_writer.writerow({"method": method, "quality": f"{q:.6f}"}) + st = method_stats[method] + st["n"] += 1 + st["sum"] += q + st["sum2"] += q * q + st["min"] = min(st["min"], q) + st["max"] = max(st["max"], q) + processed += len(methods) + if processed and processed % args.print_freq < args.batch_size: + elapsed = max(time.time() - started, 1e-6) + print( + f"{now()} rank={rank}/{world_size} processed={processed}/{len(dataset)} " + f"bad={bad} ips={processed / elapsed:.1f}", + flush=True, + ) + + stats_path = out_dir / f"{args.out_prefix}_shard{rank:02d}_of{world_size:02d}.method_stats.tsv" + with stats_path.open("w", newline="", encoding="utf-8") as f: + writer = csv.DictWriter(f, fieldnames=["method", "n", "mean", "std", "min", "max"], delimiter="\t") + writer.writeheader() + for method, st in sorted(method_stats.items()): + n = int(st["n"]) + mean = st["sum"] / n if n else float("nan") + var = max(st["sum2"] / n - mean * mean, 0.0) if n else float("nan") + writer.writerow( + { + "method": method, + "n": n, + "mean": f"{mean:.6f}", + "std": f"{math.sqrt(var):.6f}", + "min": f"{st['min']:.6f}", + "max": f"{st['max']:.6f}", + } + ) + metadata = { + "created_utc": now(), + "rank": rank, + "world_size": world_size, + "shard": str(shard_path), + "scores": str(score_path), + "bad": str(bad_path), + "method_stats": str(stats_path), + "processed": processed, + "bad_images": bad, + "elapsed_sec": time.time() - started, + "arch": args.arch, + "checkpoint": str(args.checkpoint), + "image_size": args.image_size, + "batch_size": args.batch_size, + "workers": args.workers, + "amp": args.amp, + } + meta_path.write_text(json.dumps(metadata, indent=2) + "\n", encoding="utf-8") + print(f"{now()} rank={rank} done processed={processed} bad={bad} elapsed={metadata['elapsed_sec']:.1f}s") + return 0 + + +def command_summarize(args: argparse.Namespace) -> int: + import matplotlib + + matplotlib.use("Agg") + import matplotlib.pyplot as plt + + score_files = sorted(args.score_dir.glob(f"{args.prefix}_shard*_of*.scores.tsv")) + if not score_files: + raise FileNotFoundError(f"no scores matching {args.prefix}_shard*_of*.scores.tsv in {args.score_dir}") + + values_by_method: dict[str, list[float]] = defaultdict(list) + all_values: list[float] = [] + for path in score_files: + with path.open(newline="", encoding="utf-8") as f: + for row in csv.DictReader(f, delimiter="\t"): + q = float(row["quality"]) + values_by_method[row["method"]].append(q) + all_values.append(q) + + out_dir = args.out_dir + out_dir.mkdir(parents=True, exist_ok=True) + fields = ["method", "n", "mean", "std", "p01", "p05", "p10", "p25", "p50", "p75", "p90", "p95", "p99"] + rows = [] + for method, values in sorted(values_by_method.items()): + arr = np.asarray(values, dtype=np.float32) + qs = np.percentile(arr, [1, 5, 10, 25, 50, 75, 90, 95, 99]) + rows.append( + { + "method": method, + "n": int(arr.size), + "mean": float(arr.mean()), + "std": float(arr.std()), + "p01": qs[0], + "p05": qs[1], + "p10": qs[2], + "p25": qs[3], + "p50": qs[4], + "p75": qs[5], + "p90": qs[6], + "p95": qs[7], + "p99": qs[8], + } + ) + with (out_dir / f"{args.prefix}_method_quality_summary.tsv").open("w", newline="", encoding="utf-8") as f: + writer = csv.DictWriter(f, fieldnames=fields, delimiter="\t") + writer.writeheader() + for row in rows: + writer.writerow({k: (f"{v:.6f}" if isinstance(v, float) else v) for k, v in row.items()}) + + all_arr = np.asarray(all_values, dtype=np.float32) + global_q = np.percentile(all_arr, [1, 5, 10, 25, 50, 75, 90, 95, 99]) + global_summary = { + "n": int(all_arr.size), + "methods": len(values_by_method), + "mean": float(all_arr.mean()), + "std": float(all_arr.std()), + "p01": float(global_q[0]), + "p05": float(global_q[1]), + "p10": float(global_q[2]), + "p25": float(global_q[3]), + "p50": float(global_q[4]), + "p75": float(global_q[5]), + "p90": float(global_q[6]), + "p95": float(global_q[7]), + "p99": float(global_q[8]), + } + (out_dir / f"{args.prefix}_global_quality_summary.json").write_text(json.dumps(global_summary, indent=2) + "\n", encoding="utf-8") + + plt.figure(figsize=(7.0, 4.0)) + plt.hist(all_arr, bins=120, density=True, color="#3B82F6", alpha=0.78) + plt.axvline(global_summary["mean"], color="#111827", linestyle="--", linewidth=1.4, label=f"mean={global_summary['mean']:.2f}") + plt.axvline(global_summary["p50"], color="#DC2626", linestyle="-", linewidth=1.2, label=f"median={global_summary['p50']:.2f}") + plt.xlabel("MagFace quality score (feature norm)") + plt.ylabel("Density") + plt.title("Manipulated-face quality distribution") + plt.legend(frameon=False) + plt.tight_layout() + plt.savefig(out_dir / f"{args.prefix}_global_quality_hist.png", dpi=220) + plt.savefig(out_dir / f"{args.prefix}_global_quality_hist.pdf") + plt.close() + + rows_sorted = sorted(rows, key=lambda r: r["mean"]) + methods = [r["method"].replace("_faces", "") for r in rows_sorted] + means = [r["mean"] for r in rows_sorted] + plt.figure(figsize=(max(9.0, len(methods) * 0.13), 5.0)) + plt.bar(range(len(methods)), means, color="#10B981", width=0.85) + plt.xticks(range(len(methods)), methods, rotation=90, fontsize=6) + plt.ylabel("Mean MagFace quality score") + plt.title("Mean manipulated-face quality by generation method") + plt.tight_layout() + plt.savefig(out_dir / f"{args.prefix}_method_mean_quality.png", dpi=240) + plt.savefig(out_dir / f"{args.prefix}_method_mean_quality.pdf") + plt.close() + + print(json.dumps(global_summary, indent=2)) + print(out_dir / f"{args.prefix}_method_quality_summary.tsv") + return 0 + + +def build_parser() -> argparse.ArgumentParser: + parser = argparse.ArgumentParser(description=__doc__) + sub = parser.add_subparsers(dest="command", required=True) + + p = sub.add_parser("make-shards") + p.add_argument("--root", type=Path, required=True) + p.add_argument("--out-dir", type=Path, required=True) + p.add_argument("--prefix", default="scaledf_fake") + p.add_argument("--num-shards", type=int, default=8) + p.add_argument("--print-freq", type=int, default=200000) + p.set_defaults(func=command_make_shards) + + p = sub.add_parser("score") + p.add_argument("--shard-dir", type=Path, required=True) + p.add_argument("--out-dir", type=Path, required=True) + p.add_argument("--prefix", default="scaledf_fake") + p.add_argument("--out-prefix", default="magface_fake_quality") + p.add_argument("--magface-repo", type=Path, required=True) + p.add_argument("--checkpoint", type=Path, required=True) + p.add_argument("--arch", default="iresnet50") + p.add_argument("--image-size", type=int, default=112) + p.add_argument("--batch-size", type=int, default=1024) + p.add_argument("--workers", type=int, default=8) + p.add_argument("--prefetch-factor", type=int, default=4) + p.add_argument("--print-freq", type=int, default=100000) + p.add_argument("--amp", action=argparse.BooleanOptionalAction, default=True) + p.set_defaults(func=command_score) + + p = sub.add_parser("summarize") + p.add_argument("--score-dir", type=Path, required=True) + p.add_argument("--out-dir", type=Path, required=True) + p.add_argument("--prefix", default="magface_fake_quality") + p.set_defaults(func=command_summarize) + return parser + + +def main() -> int: + parser = build_parser() + args = parser.parse_args() + return args.func(args) + + +if __name__ == "__main__": + raise SystemExit(main()) diff --git a/ScaleDF_rebuttal/main_cross_2direct_balance_realscale_perturb_v4_NRealFake_noqcpt.py b/ScaleDF_rebuttal/main_cross_2direct_balance_realscale_perturb_v4_NRealFake_noqcpt.py new file mode 100644 index 0000000000000000000000000000000000000000..2ed9865a799f6f4de6e0a767eba944adad4e26f8 --- /dev/null +++ b/ScaleDF_rebuttal/main_cross_2direct_balance_realscale_perturb_v4_NRealFake_noqcpt.py @@ -0,0 +1,549 @@ +# Copyright (c) 2015-present, Facebook, Inc. +# All rights reserved. +import argparse +import datetime +import numpy as np +import time +import torch +import torch.backends.cudnn as cudnn +import json + +from pathlib import Path + +from timm.data import Mixup +from timm.models import create_model +from timm.loss import LabelSmoothingCrossEntropy, SoftTargetCrossEntropy +from timm.scheduler import create_scheduler +from timm.optim import create_optimizer +from timm.utils import NativeScaler, get_state_dict, ModelEma + +from datasets_cross_2direct_balance_realscale_perturb_v4_NRealFake_noqcpt import build_dataset +from engine_cross_2direct_balance_realscale_perturb_v4_NRealFake import train_one_epoch, evaluate +from losses_cross_2direct_balance_realscale_perturb_v4_NRealFake import DistillationLoss +from samplers import RASampler +from augment import new_data_aug_generator + +import models +import models_v2 + +import utils +import math + +import warnings +warnings.filterwarnings("ignore") +warnings.filterwarnings("ignore", category=UserWarning) +warnings.filterwarnings("ignore", category=DeprecationWarning) +warnings.filterwarnings("ignore", category=FutureWarning) + +def get_args_parser(): + parser = argparse.ArgumentParser('DeiT training and evaluation script', add_help=False) + parser.add_argument('--batch-size', default=64, type=int) + parser.add_argument('--epochs', default=300, type=int) + parser.add_argument('--bce-loss', action='store_true') + parser.add_argument('--unscale-lr', action='store_true') + parser.add_argument('--select-models', default='scaledf_v4_train_fix.pkl', type=str) + parser.add_argument('--select-images', default='', type=str) + + + # Model parameters + parser.add_argument('--model', default='deit_base_patch16_224', type=str, metavar='MODEL', + help='Name of model to train') + parser.add_argument('--input-size', default=224, type=int, help='images input size') + + parser.add_argument('--drop', type=float, default=0.0, metavar='PCT', + help='Dropout rate (default: 0.)') + parser.add_argument('--drop-path', type=float, default=0.1, metavar='PCT', + help='Drop path rate (default: 0.1)') + + parser.add_argument('--model-ema', action='store_true') + parser.add_argument('--no-model-ema', action='store_false', dest='model_ema') + parser.set_defaults(model_ema=True) + parser.add_argument('--model-ema-decay', type=float, default=0.99996, help='') + parser.add_argument('--model-ema-force-cpu', action='store_true', default=False, help='') + parser.add_argument('--two-cls', action='store_true') + + # Optimizer parameters + parser.add_argument('--opt', default='adamw', type=str, metavar='OPTIMIZER', + help='Optimizer (default: "adamw"') + parser.add_argument('--opt-eps', default=1e-8, type=float, metavar='EPSILON', + help='Optimizer Epsilon (default: 1e-8)') + parser.add_argument('--opt-betas', default=None, type=float, nargs='+', metavar='BETA', + help='Optimizer Betas (default: None, use opt default)') + parser.add_argument('--clip-grad', type=float, default=None, metavar='NORM', + help='Clip gradient norm (default: None, no clipping)') + parser.add_argument('--momentum', type=float, default=0.9, metavar='M', + help='SGD momentum (default: 0.9)') + parser.add_argument('--weight-decay', type=float, default=0.05, + help='weight decay (default: 0.05)') + # Learning rate schedule parameters + parser.add_argument('--sched', default='cosine', type=str, metavar='SCHEDULER', + help='LR scheduler (default: "cosine"') + parser.add_argument('--lr', type=float, default=5e-4, metavar='LR', + help='learning rate (default: 5e-4)') + parser.add_argument('--lr-noise', type=float, nargs='+', default=None, metavar='pct, pct', + help='learning rate noise on/off epoch percentages') + parser.add_argument('--lr-noise-pct', type=float, default=0.67, metavar='PERCENT', + help='learning rate noise limit percent (default: 0.67)') + parser.add_argument('--lr-noise-std', type=float, default=1.0, metavar='STDDEV', + help='learning rate noise std-dev (default: 1.0)') + parser.add_argument('--warmup-lr', type=float, default=1e-6, metavar='LR', + help='warmup learning rate (default: 1e-6)') + parser.add_argument('--min-lr', type=float, default=1e-6, metavar='LR', + help='lower lr bound for cyclic schedulers that hit 0 (1e-5)') + + parser.add_argument('--decay-epochs', type=float, default=15, metavar='N', + help='epoch interval to decay LR') + parser.add_argument('--warmup-epochs', type=int, default=5, metavar='N', + help='epochs to warmup LR, if scheduler supports') + parser.add_argument('--cooldown-epochs', type=int, default=10, metavar='N', + help='epochs to cooldown LR at min_lr, after cyclic schedule ends') + parser.add_argument('--patience-epochs', type=int, default=10, metavar='N', + help='patience epochs for Plateau LR scheduler (default: 10') + parser.add_argument('--decay-rate', '--dr', type=float, default=0.1, metavar='RATE', + help='LR decay rate (default: 0.1)') + + # Augmentation parameters + parser.add_argument('--color-jitter', type=float, default=0.3, metavar='PCT', + help='Color jitter factor (default: 0.3)') + parser.add_argument('--aa', type=str, default='rand-m9-mstd0.5-inc1', metavar='NAME', + help='Use AutoAugment policy. "v0" or "original". " + \ + "(default: rand-m9-mstd0.5-inc1)'), + parser.add_argument('--smoothing', type=float, default=0.1, help='Label smoothing (default: 0.1)') + parser.add_argument('--train-interpolation', type=str, default='bicubic', + help='Training interpolation (random, bilinear, bicubic default: "bicubic")') + + parser.add_argument('--repeated-aug', action='store_true') + parser.add_argument('--no-repeated-aug', action='store_false', dest='repeated_aug') + parser.set_defaults(repeated_aug=True) + + parser.add_argument('--train-mode', action='store_true') + parser.add_argument('--no-train-mode', action='store_false', dest='train_mode') + parser.set_defaults(train_mode=True) + + parser.add_argument('--ThreeAugment', action='store_true') #3augment + + parser.add_argument('--src', action='store_true') #simple random crop + + # * Random Erase params + parser.add_argument('--reprob', type=float, default=0.25, metavar='PCT', + help='Random erase prob (default: 0.25)') + parser.add_argument('--remode', type=str, default='pixel', + help='Random erase mode (default: "pixel")') + parser.add_argument('--recount', type=int, default=1, + help='Random erase count (default: 1)') + parser.add_argument('--resplit', action='store_true', default=False, + help='Do not random erase first (clean) augmentation split') + + # * Mixup params + parser.add_argument('--mixup', type=float, default=0.8, + help='mixup alpha, mixup enabled if > 0. (default: 0.8)') + parser.add_argument('--cutmix', type=float, default=1.0, + help='cutmix alpha, cutmix enabled if > 0. (default: 1.0)') + parser.add_argument('--cutmix-minmax', type=float, nargs='+', default=None, + help='cutmix min/max ratio, overrides alpha and enables cutmix if set (default: None)') + parser.add_argument('--mixup-prob', type=float, default=1.0, + help='Probability of performing mixup or cutmix when either/both is enabled') + parser.add_argument('--mixup-switch-prob', type=float, default=0.5, + help='Probability of switching to cutmix when both mixup and cutmix enabled') + parser.add_argument('--mixup-mode', type=str, default='batch', + help='How to apply mixup/cutmix params. Per "batch", "pair", or "elem"') + + # Distillation parameters + parser.add_argument('--teacher-model', default='regnety_160', type=str, metavar='MODEL', + help='Name of teacher model to train (default: "regnety_160"') + parser.add_argument('--teacher-path', type=str, default='') + parser.add_argument('--distillation-type', default='none', choices=['none', 'soft', 'hard'], type=str, help="") + parser.add_argument('--distillation-alpha', default=0.5, type=float, help="") + parser.add_argument('--distillation-tau', default=1.0, type=float, help="") + + # * Cosub params + parser.add_argument('--cosub', action='store_true') + + # * Finetuning params + parser.add_argument('--finetune', default='', help='finetune from checkpoint') + parser.add_argument('--attn-only', action='store_true') + + # Dataset parameters + parser.add_argument('--data-path', default='/datasets01/imagenet_full_size/061417/', type=str, + help='dataset path') + parser.add_argument('--data-set', default='IMNET', choices=['CIFAR', 'IMNET', 'INAT', 'INAT19', 'ScaleDF'], + type=str, help='Image Net dataset path') + parser.add_argument('--inat-category', default='name', + choices=['kingdom', 'phylum', 'class', 'order', 'supercategory', 'family', 'genus', 'name'], + type=str, help='semantic granularity') + + parser.add_argument('--output_dir', default='', + help='path where to save, empty for no saving') + parser.add_argument('--device', default='cuda', + help='device to use for training / testing') + parser.add_argument('--seed', default=0, type=int) + parser.add_argument('--resume', default='', help='resume from checkpoint') + parser.add_argument('--start_epoch', default=0, type=int, metavar='N', + help='start epoch') + parser.add_argument('--eval', action='store_true', help='Perform evaluation only') + parser.add_argument('--eval-crop-ratio', default=0.875, type=float, help="Crop ratio for evaluation") + parser.add_argument('--dist-eval', action='store_true', default=False, help='Enabling distributed evaluation') + parser.add_argument('--num_workers', default=10, type=int) + parser.add_argument('--pin-mem', action='store_true', + help='Pin CPU memory in DataLoader for more efficient (sometimes) transfer to GPU.') + parser.add_argument('--no-pin-mem', action='store_false', dest='pin_mem', + help='') + parser.set_defaults(pin_mem=True) + + # distributed training parameters + parser.add_argument('--distributed', action='store_true', default=False, help='Enabling distributed training') + parser.add_argument('--world_size', default=1, type=int, + help='number of distributed processes') + parser.add_argument('--dist_url', default='env://', help='url used to set up distributed training') + return parser + + +class FixedSizeDistributedSampler(torch.utils.data.DistributedSampler): + def __init__(self, dataset, num_samples_per_replica, **kwargs): + super().__init__(dataset, **kwargs) + # 覆盖父类里根据 len(dataset) 推算出的值 + self.num_samples = num_samples_per_replica + self.total_size = self.num_samples * self.num_replicas + + def __iter__(self): + # 跟官方实现一样,只是把 self.num_samples 换成固定值 + g = torch.Generator() + g.manual_seed(self.seed + self.epoch) + if self.shuffle: + indices = torch.randperm(len(self.dataset), generator=g).tolist() + else: + indices = list(range(len(self.dataset))) + + # 够不到 total_size 就补(不会真的用到) + padding_size = self.total_size - len(indices) + if padding_size > 0: + indices += indices[:padding_size] + + # 分片给各 GPU + indices = indices[self.rank:self.total_size:self.num_replicas] + + # 再按固定长度裁掉 + return iter(indices[:self.num_samples]) + + +def main(args): + utils.init_distributed_mode(args) + + print(args) + + if args.distillation_type != 'none' and args.finetune and not args.eval: + raise NotImplementedError("Finetuning with distillation not yet supported") + + device = torch.device(args.device) + + # fix the seed for reproducibility + seed = args.seed + utils.get_rank() + torch.manual_seed(seed) + np.random.seed(seed) + # random.seed(seed) + + cudnn.benchmark = True + + orig_len, dataset_train, args.nb_classes = build_dataset(is_train=True, args=args) + dataset_val, _ = build_dataset(is_train=False, args=args) + + if args.distributed: + num_tasks = utils.get_world_size() + global_rank = utils.get_rank() + if args.repeated_aug: + sampler_train = RASampler( + dataset_train, num_replicas=num_tasks, rank=global_rank, shuffle=True + ) + else: + ''' + sampler_train = torch.utils.data.DistributedSampler( + dataset_train, num_replicas=num_tasks, rank=global_rank, shuffle=True + ) + ''' + sampler_train = FixedSizeDistributedSampler( + dataset_train, + num_samples_per_replica = math.ceil(orig_len / num_tasks), + num_replicas = num_tasks, + rank = global_rank, + shuffle = True + ) + + + + + + + if args.dist_eval: + if len(dataset_val) % num_tasks != 0: + print('Warning: Enabling distributed evaluation with an eval dataset not divisible by process number. ' + 'This will slightly alter validation results as extra duplicate entries are added to achieve ' + 'equal num of samples per-process.') + sampler_val = torch.utils.data.DistributedSampler( + dataset_val, num_replicas=num_tasks, rank=global_rank, shuffle=False) + else: + sampler_val = torch.utils.data.SequentialSampler(dataset_val) + else: + sampler_train = torch.utils.data.RandomSampler(dataset_train) + sampler_val = torch.utils.data.SequentialSampler(dataset_val) + + data_loader_train = torch.utils.data.DataLoader( + dataset_train, sampler=sampler_train, + batch_size=args.batch_size, + num_workers=args.num_workers, + pin_memory=args.pin_mem, + drop_last=True, + ) + if args.ThreeAugment: + data_loader_train.dataset.transform = new_data_aug_generator(args) + + data_loader_val = torch.utils.data.DataLoader( + dataset_val, sampler=sampler_val, + batch_size=int(1.5 * args.batch_size), + num_workers=args.num_workers, + pin_memory=args.pin_mem, + drop_last=False + ) + + mixup_fn = None + mixup_active = args.mixup > 0 or args.cutmix > 0. or args.cutmix_minmax is not None + if mixup_active: + if args.two_cls: + mixup_fn = Mixup( + mixup_alpha=args.mixup, cutmix_alpha=args.cutmix, cutmix_minmax=args.cutmix_minmax, + prob=args.mixup_prob, switch_prob=args.mixup_switch_prob, mode=args.mixup_mode, + label_smoothing=args.smoothing, num_classes=2) + else: + mixup_fn = Mixup( + mixup_alpha=args.mixup, cutmix_alpha=args.cutmix, cutmix_minmax=args.cutmix_minmax, + prob=args.mixup_prob, switch_prob=args.mixup_switch_prob, mode=args.mixup_mode, + label_smoothing=args.smoothing, num_classes=args.nb_classes) + + print(f"Creating model: {args.model}") + model = create_model( + args.model, + pretrained=False, + num_classes=2, + drop_rate=args.drop, + drop_path_rate=args.drop_path, + drop_block_rate=None, + img_size=args.input_size + ) + + + if args.finetune: + if args.finetune.startswith('https'): + checkpoint = torch.hub.load_state_dict_from_url( + args.finetune, map_location='cpu', check_hash=True) + else: + checkpoint = torch.load(args.finetune, map_location='cpu') + + checkpoint_model = checkpoint['model'] + state_dict = model.state_dict() + for k in ['head.weight', 'head.bias', 'head_dist.weight', 'head_dist.bias']: + if k in checkpoint_model and checkpoint_model[k].shape != state_dict[k].shape: + print(f"Removing key {k} from pretrained checkpoint") + del checkpoint_model[k] + + # interpolate position embedding + pos_embed_checkpoint = checkpoint_model['pos_embed'] + embedding_size = pos_embed_checkpoint.shape[-1] + num_patches = model.patch_embed.num_patches + num_extra_tokens = model.pos_embed.shape[-2] - num_patches + # height (== width) for the checkpoint position embedding + orig_size = int((pos_embed_checkpoint.shape[-2] - num_extra_tokens) ** 0.5) + # height (== width) for the new position embedding + new_size = int(num_patches ** 0.5) + # class_token and dist_token are kept unchanged + extra_tokens = pos_embed_checkpoint[:, :num_extra_tokens] + # only the position tokens are interpolated + pos_tokens = pos_embed_checkpoint[:, num_extra_tokens:] + pos_tokens = pos_tokens.reshape(-1, orig_size, orig_size, embedding_size).permute(0, 3, 1, 2) + pos_tokens = torch.nn.functional.interpolate( + pos_tokens, size=(new_size, new_size), mode='bicubic', align_corners=False) + pos_tokens = pos_tokens.permute(0, 2, 3, 1).flatten(1, 2) + new_pos_embed = torch.cat((extra_tokens, pos_tokens), dim=1) + checkpoint_model['pos_embed'] = new_pos_embed + + model.load_state_dict(checkpoint_model, strict=False) + + if args.attn_only: + for name_p,p in model.named_parameters(): + if '.attn.' in name_p: + p.requires_grad = True + else: + p.requires_grad = False + try: + model.head.weight.requires_grad = True + model.head.bias.requires_grad = True + except: + model.fc.weight.requires_grad = True + model.fc.bias.requires_grad = True + try: + model.pos_embed.requires_grad = True + except: + print('no position encoding') + try: + for p in model.patch_embed.parameters(): + p.requires_grad = False + except: + print('no patch embed') + + model.to(device) + + model_ema = None + if args.model_ema: + # Important to create EMA model after cuda(), DP wrapper, and AMP but before SyncBN and DDP wrapper + model_ema = ModelEma( + model, + decay=args.model_ema_decay, + device='cpu' if args.model_ema_force_cpu else '', + resume='') + + model_without_ddp = model + if args.distributed: + model = torch.nn.parallel.DistributedDataParallel(model, device_ids=[args.gpu]) + model_without_ddp = model.module + n_parameters = sum(p.numel() for p in model.parameters() if p.requires_grad) + print('number of params:', n_parameters) + if not args.unscale_lr: + linear_scaled_lr = args.lr * args.batch_size * utils.get_world_size() / 512.0 + args.lr = linear_scaled_lr + optimizer = create_optimizer(args, model_without_ddp) + loss_scaler = NativeScaler() + + lr_scheduler, _ = create_scheduler(args, optimizer) + + criterion = LabelSmoothingCrossEntropy() + + if mixup_active: + # smoothing is handled with mixup label transform + criterion = SoftTargetCrossEntropy() + elif args.smoothing: + criterion = LabelSmoothingCrossEntropy(smoothing=args.smoothing) + else: + criterion = torch.nn.CrossEntropyLoss() + + if args.bce_loss: + criterion = torch.nn.BCEWithLogitsLoss() + + teacher_model = None + if args.distillation_type != 'none': + assert args.teacher_path, 'need to specify teacher-path when using distillation' + print(f"Creating teacher model: {args.teacher_model}") + teacher_model = create_model( + args.teacher_model, + pretrained=False, + num_classes=args.nb_classes, + global_pool='avg', + ) + if args.teacher_path.startswith('https'): + checkpoint = torch.hub.load_state_dict_from_url( + args.teacher_path, map_location='cpu', check_hash=True) + else: + checkpoint = torch.load(args.teacher_path, map_location='cpu') + teacher_model.load_state_dict(checkpoint['model']) + teacher_model.to(device) + teacher_model.eval() + + # wrap the criterion in our custom DistillationLoss, which + # just dispatches to the original criterion if args.distillation_type is 'none' + criterion = DistillationLoss( + criterion, teacher_model, args.distillation_type, args.distillation_alpha, args.distillation_tau + ) + + output_dir = Path(args.output_dir) + if args.resume: + if args.resume.startswith('https'): + checkpoint = torch.hub.load_state_dict_from_url( + args.resume, map_location='cpu', check_hash=True) + else: + checkpoint = torch.load(args.resume, map_location='cpu', weights_only=False)#torch2.6 + model_without_ddp.load_state_dict(checkpoint['model']) + if not args.eval and 'optimizer' in checkpoint and 'lr_scheduler' in checkpoint and 'epoch' in checkpoint: + optimizer.load_state_dict(checkpoint['optimizer']) + lr_scheduler.load_state_dict(checkpoint['lr_scheduler']) + args.start_epoch = checkpoint['epoch'] + 1 + if args.model_ema: + utils._load_checkpoint_for_ema(model_ema, checkpoint['model_ema']) + if 'scaler' in checkpoint: + loss_scaler.load_state_dict(checkpoint['scaler']) + lr_scheduler.step(args.start_epoch) + if args.eval: + test_stats = evaluate(data_loader_val, model, device) + print(f"Accuracy of the network on the {len(dataset_val)} test images: {test_stats['acc1']:.1f}%") + return + + print(f"Start training for {args.epochs} epochs") + start_time = time.time() + max_accuracy = 0.0 + for epoch in range(args.start_epoch, args.epochs): + if args.distributed: + data_loader_train.sampler.set_epoch(epoch) + + train_stats = train_one_epoch( + model, criterion, data_loader_train, + optimizer, device, epoch, loss_scaler, + args.clip_grad, model_ema, mixup_fn, + set_training_mode=args.train_mode, # keep in eval mode for deit finetuning / train mode for training and deit III finetuning + args = args, + ) + + lr_scheduler.step(epoch) + if args.output_dir: + checkpoint_paths = [output_dir / 'checkpoint.pth'] + for checkpoint_path in checkpoint_paths: + utils.save_on_master({ + 'model': model_without_ddp.state_dict(), + 'optimizer': optimizer.state_dict(), + 'lr_scheduler': lr_scheduler.state_dict(), + 'epoch': epoch, + 'model_ema': get_state_dict(model_ema), + 'scaler': loss_scaler.state_dict(), + 'args': args, + }, checkpoint_path) + + + if (epoch+1) % 5 == 0: + test_stats = evaluate(data_loader_val, model, device) + print(f"Accuracy of the network on the {len(dataset_val)} test images: {test_stats['acc1']:.1f}%") + + if max_accuracy < test_stats["acc1"]: + max_accuracy = test_stats["acc1"] + if args.output_dir: + checkpoint_paths = [output_dir / 'best_checkpoint.pth'] + for checkpoint_path in checkpoint_paths: + utils.save_on_master({ + 'model': model_without_ddp.state_dict(), + 'optimizer': optimizer.state_dict(), + 'lr_scheduler': lr_scheduler.state_dict(), + 'epoch': epoch, + 'model_ema': get_state_dict(model_ema), + 'scaler': loss_scaler.state_dict(), + 'args': args, + }, checkpoint_path) + + print(f'Max accuracy: {max_accuracy:.2f}%') + + log_stats = {**{f'train_{k}': v for k, v in train_stats.items()}, + **{f'test_{k}': v for k, v in test_stats.items()}, + 'epoch': epoch, + 'n_parameters': n_parameters} + + + + + if args.output_dir and utils.is_main_process(): + with (output_dir / "log.txt").open("a") as f: + f.write(json.dumps(log_stats) + "\n") + + total_time = time.time() - start_time + total_time_str = str(datetime.timedelta(seconds=int(total_time))) + print('Training time {}'.format(total_time_str)) + + +if __name__ == '__main__': + parser = argparse.ArgumentParser('DeiT training and evaluation script', parents=[get_args_parser()]) + args = parser.parse_args() + if args.output_dir: + Path(args.output_dir).mkdir(parents=True, exist_ok=True) + main(args) diff --git a/ScaleDF_rebuttal/make_data_size_noqcpt_selections.py b/ScaleDF_rebuttal/make_data_size_noqcpt_selections.py new file mode 100644 index 0000000000000000000000000000000000000000..0d34d88f25ba3e5c557e35dff8f2848499743b3f --- /dev/null +++ b/ScaleDF_rebuttal/make_data_size_noqcpt_selections.py @@ -0,0 +1,137 @@ +#!/usr/bin/env python3 +"""Build all-class image-count selections for no-QC/PT data-size scaling.""" + +from __future__ import annotations + +import argparse +import csv +import json +import random +from fractions import Fraction +from pathlib import Path + +from make_fixed_budget_resampled_selections import ( + dump_pickle, + list_image_files, + load_pickle, + sample_from_filesystem, +) + + +def parse_fractions(raw: str) -> list[Fraction]: + out = [] + for part in raw.split(","): + part = part.strip() + if not part: + continue + out.append(Fraction(part)) + return out + + +def fraction_label(frac: Fraction) -> str: + if frac.denominator == 1: + return f"{frac.numerator}" + return f"{frac.numerator}over{frac.denominator}" + + +def main() -> None: + parser = argparse.ArgumentParser() + parser.add_argument("--code-root", default="/tmp/ScaleDF_repro_20260724/code/deit") + parser.add_argument("--out-dir", required=True) + parser.add_argument("--fractions", required=True) + parser.add_argument("--scan-data-root", default="/dev/shm/scaledf_bind_storage/ScaleDF") + parser.add_argument("--out-data-root", default="/dev/shm/scaledf_bind_storage/ScaleDF") + parser.add_argument("--image-seed-base", type=int, default=20260725) + args = parser.parse_args() + + code_root = Path(args.code_root) + out_dir = Path(args.out_dir) + class_names = load_pickle(code_root / "scaledf_v4_train_fix.pkl") + scan_train_root = Path(args.scan_data_root) / "train" + out_train_root = Path(args.out_data_root) / "train" + + real_labels = set(range(46)) + fake_labels = set(range(46, len(class_names))) + file_cache = {} + + def ensure_label(label: int): + if label not in file_cache: + name = class_names[label] + file_cache[label] = list_image_files(scan_train_root, out_train_root, name, label) + print(f"listed label={label} class={name} files={len(file_cache[label])}", flush=True) + return file_cache[label] + + for label in sorted(real_labels | fake_labels): + ensure_label(label) + + real_available = sum(len(file_cache[label]) for label in real_labels) + fake_available = sum(len(file_cache[label]) for label in fake_labels) + full_per_side = min(real_available, fake_available) + full_total = 2 * full_per_side + fractions = parse_fractions(args.fractions) + rows = [] + + for idx, frac in enumerate(fractions): + if frac <= 0 or frac > 1: + raise ValueError(f"fraction must be in (0, 1]: {frac}") + total = int(full_total * frac) + total -= total % 2 + per_side = total // 2 + if per_side <= 0: + raise ValueError(f"fraction too small: {frac}") + + rng = random.Random(args.image_seed_base + idx * 1000003 + frac.denominator * 1009 + frac.numerator) + real_samples, real_quotas = sample_from_filesystem(file_cache, real_labels, per_side, rng) + fake_samples, fake_quotas = sample_from_filesystem(file_cache, fake_labels, per_side, rng) + samples = real_samples + fake_samples + + label = fraction_label(frac) + out_name = f"scaledf_v4_noqcpt_datasize_N{full_total}_{label}_imgseed{args.image_seed_base}.pkl" + out_path = out_dir / out_name + dump_pickle(samples, out_path) + + rows.append( + { + "fraction": str(frac), + "fraction_label": label, + "full_total_images": full_total, + "total_images": len(samples), + "real_images": len(real_samples), + "fake_images": len(fake_samples), + "real_available": real_available, + "fake_available": fake_available, + "path": str(out_path), + "real_quota_json": json.dumps(real_quotas, sort_keys=True), + "fake_quota_json": json.dumps(fake_quotas, sort_keys=True), + } + ) + print( + f"wrote fraction={frac} total={len(samples)} real={len(real_samples)} " + f"fake={len(fake_samples)} path={out_path}", + flush=True, + ) + + manifest = out_dir / f"manifest_noqcpt_datasize_N{full_total}_imgseed{args.image_seed_base}.tsv" + fields = [ + "fraction", + "fraction_label", + "full_total_images", + "total_images", + "real_images", + "fake_images", + "real_available", + "fake_available", + "path", + "real_quota_json", + "fake_quota_json", + ] + out_dir.mkdir(parents=True, exist_ok=True) + with manifest.open("w", newline="", encoding="utf-8") as f: + writer = csv.DictWriter(f, fieldnames=fields, delimiter="\t") + writer.writeheader() + writer.writerows(rows) + print(f"manifest={manifest}") + + +if __name__ == "__main__": + main() diff --git a/ScaleDF_rebuttal/make_fixed_budget_all_selection.py b/ScaleDF_rebuttal/make_fixed_budget_all_selection.py new file mode 100644 index 0000000000000000000000000000000000000000..0fc3183646dcd015ff3b987491ff1a83cc54d603 --- /dev/null +++ b/ScaleDF_rebuttal/make_fixed_budget_all_selection.py @@ -0,0 +1,117 @@ +#!/usr/bin/env python3 +"""Build a fixed-budget selection using all 46 real domains and 91 fake methods.""" + +from __future__ import annotations + +import argparse +import csv +import json +import random +from pathlib import Path + +from make_fixed_budget_resampled_selections import ( + dump_pickle, + list_image_files, + load_pickle, + read_counts, + sample_from_filesystem, +) + + +def main() -> None: + parser = argparse.ArgumentParser() + parser.add_argument("--code-root", default="/tmp/ScaleDF_repro_20260724/code/deit") + parser.add_argument("--out-dir", required=True) + parser.add_argument("--per-side", type=int, default=500000) + parser.add_argument("--image-seed", type=int, default=20260725) + parser.add_argument("--counts-tsv", type=Path) + parser.add_argument("--scan-data-root", default="/dev/shm/scaledf_bind_storage/ScaleDF") + parser.add_argument("--out-data-root", default="/dev/shm/scaledf_bind_storage/ScaleDF") + args = parser.parse_args() + + code_root = Path(args.code_root) + out_dir = Path(args.out_dir) + class_names = load_pickle(code_root / "scaledf_v4_train_fix.pkl") + real_names = class_names[:46] + fake_names = class_names[46:] + counts = read_counts(args.counts_tsv) + scan_train_root = Path(args.scan_data_root) / "train" + out_train_root = Path(args.out_data_root) / "train" + + file_cache = {} + + def ensure_label(label: int): + if label not in file_cache: + name = class_names[label] + file_cache[label] = list_image_files(scan_train_root, out_train_root, name, label) + counts.setdefault(name, len(file_cache[label])) + print(f"listed label={label} class={name} files={len(file_cache[label])}", flush=True) + return file_cache[label] + + real_labels = set(range(46)) + fake_labels = set(range(46, len(class_names))) + for label in sorted(real_labels | fake_labels): + ensure_label(label) + + real_available = sum(len(file_cache[label]) for label in real_labels) + fake_available = sum(len(file_cache[label]) for label in fake_labels) + if real_available < args.per_side or fake_available < args.per_side: + raise RuntimeError( + f"not enough images: real_available={real_available}, fake_available={fake_available}, per_side={args.per_side}" + ) + + rng = random.Random(args.image_seed) + real_samples, real_quotas = sample_from_filesystem(file_cache, real_labels, args.per_side, rng) + fake_samples, fake_quotas = sample_from_filesystem(file_cache, fake_labels, args.per_side, rng) + samples = real_samples + fake_samples + + out_name = f"scaledf_v4_fixed{args.per_side}_all_46real_91fake_imgseed{args.image_seed}.pkl" + out_path = out_dir / out_name + dump_pickle(samples, out_path) + + manifest_path = out_dir / f"manifest_fixed{args.per_side}_all_46real_91fake_imgseed{args.image_seed}.tsv" + fieldnames = [ + "family", + "level", + "class_seed", + "real_images", + "fake_images", + "status", + "error", + "path", + "real_available", + "fake_available", + "real_classes_json", + "fake_classes_json", + "real_quota_json", + "fake_quota_json", + ] + out_dir.mkdir(parents=True, exist_ok=True) + with manifest_path.open("w", newline="", encoding="utf-8") as f: + writer = csv.DictWriter(f, fieldnames=fieldnames, delimiter="\t") + writer.writeheader() + writer.writerow( + { + "family": "all_46real_91fake", + "level": len(class_names), + "class_seed": 0, + "real_images": len(real_samples), + "fake_images": len(fake_samples), + "status": "ok", + "error": "", + "path": str(out_path), + "real_available": real_available, + "fake_available": fake_available, + "real_classes_json": json.dumps(real_names, ensure_ascii=True), + "fake_classes_json": json.dumps(fake_names, ensure_ascii=True), + "real_quota_json": json.dumps(real_quotas, sort_keys=True), + "fake_quota_json": json.dumps(fake_quotas, sort_keys=True), + } + ) + + print(f"wrote all-class selection real={len(real_samples)} fake={len(fake_samples)} path={out_path}") + print(f"manifest={manifest_path}") + + +if __name__ == "__main__": + main() diff --git a/ScaleDF_rebuttal/make_fixed_budget_resampled_selections.py b/ScaleDF_rebuttal/make_fixed_budget_resampled_selections.py new file mode 100644 index 0000000000000000000000000000000000000000..20db1d983f22e34eea864aa9983f6f373f107303 --- /dev/null +++ b/ScaleDF_rebuttal/make_fixed_budget_resampled_selections.py @@ -0,0 +1,259 @@ +#!/usr/bin/env python3 +"""Build fixed-budget selections with resampled domain/method class sets. + +Each repeat independently samples the real-domain or fake-method class set. +If the sampled class set cannot provide the requested image budget, it is +discarded and resampled. +""" + +from __future__ import annotations + +import argparse +import csv +import json +import os +import pickle +import random +from collections import defaultdict +from pathlib import Path + + +IMAGE_EXTENSIONS = {".jpg", ".jpeg", ".png", ".bmp", ".webp"} +REAL_LEVELS = [5, 8, 11, 16, 23, 32] +FAKE_LEVELS = [11, 16, 22, 32, 45, 64] + + +def load_pickle(path: Path): + with path.open("rb") as f: + return pickle.load(f) + + +def dump_pickle(obj, path: Path) -> None: + path.parent.mkdir(parents=True, exist_ok=True) + tmp = path.with_suffix(path.suffix + ".tmp") + with tmp.open("wb") as f: + pickle.dump(obj, f, protocol=pickle.HIGHEST_PROTOCOL) + os.replace(tmp, path) + + +def read_counts(path: Path | None) -> dict[str, int]: + if not path: + return {} + counts: dict[str, int] = {} + with path.open(newline="", encoding="utf-8") as f: + for row in csv.DictReader(f, delimiter="\t"): + if row.get("section") == "train": + counts[row["name"]] = int(row["image_files"]) + return counts + + +def list_image_files(scan_train_root: Path, out_train_root: Path, class_name: str, label: int): + class_root = scan_train_root / class_name + if not class_root.is_dir(): + raise FileNotFoundError(class_root) + samples = [] + for dirpath, _, filenames in os.walk(class_root): + filenames.sort() + rel_dir = Path(dirpath).relative_to(scan_train_root) + for filename in filenames: + if Path(filename).suffix.lower() in IMAGE_EXTENSIONS: + samples.append((str(out_train_root / rel_dir / filename), label)) + if not samples: + raise ValueError(f"no images found under {class_root}") + return samples + + +def sample_classes( + rng: random.Random, + names: list[str], + level: int, + counts: dict[str, int], + min_images: int, + used: set[tuple[str, ...]], + max_attempts: int, +) -> tuple[list[str], int, int]: + for attempt in range(1, max_attempts + 1): + selected = tuple(sorted(rng.sample(names, level))) + total = sum(counts.get(name, 0) for name in selected) + if total >= min_images and selected not in used: + used.add(selected) + return list(selected), total, attempt + raise RuntimeError(f"could not sample {level} classes with >= {min_images} images after {max_attempts} attempts") + + +def sample_from_filesystem(cache, labels, k: int, rng: random.Random): + labels = sorted(labels) + quotas = {label: 0 for label in labels} + remaining = set(labels) + need = k + + while need > 0 and remaining: + order = sorted(remaining) + rng.shuffle(order) + share, extra = divmod(need, len(order)) + next_remaining = set() + for idx, label in enumerate(order): + target = share + (1 if idx < extra else 0) + available = len(cache[label]) - quotas[label] + take = min(target, available) + quotas[label] += take + need -= take + if len(cache[label]) > quotas[label]: + next_remaining.add(label) + remaining = next_remaining + + if need != 0: + available = sum(len(cache[label]) for label in labels) + raise ValueError(f"need {k} samples, only {available} available for labels={labels}") + + samples = [] + for label in labels: + samples.extend(rng.sample(cache[label], quotas[label])) + rng.shuffle(samples) + return samples, quotas + + +def main() -> None: + parser = argparse.ArgumentParser() + parser.add_argument("--code-root", default="/tmp/ScaleDF_repro_20260724/code/deit") + parser.add_argument("--out-dir", required=True) + parser.add_argument("--per-side", type=int, default=500000) + parser.add_argument("--repeats", default="1,2,3,4") + parser.add_argument("--class-seed-base", type=int, default=20260725) + parser.add_argument("--image-seed", type=int, default=20260725) + parser.add_argument("--counts-tsv", type=Path) + parser.add_argument("--scan-data-root", default="/dev/shm/scaledf_bind_storage/ScaleDF") + parser.add_argument("--out-data-root", default="/dev/shm/scaledf_bind_storage/ScaleDF") + parser.add_argument("--max-attempts", type=int, default=100000) + args = parser.parse_args() + + code_root = Path(args.code_root) + out_dir = Path(args.out_dir) + repeats = [int(x) for x in args.repeats.split(",") if x.strip()] + + class_names = load_pickle(code_root / "scaledf_v4_train_fix.pkl") + name_to_label = {name: idx for idx, name in enumerate(class_names)} + real_names = class_names[:46] + fake_names = class_names[46:] + counts = read_counts(args.counts_tsv) + + scan_train_root = Path(args.scan_data_root) / "train" + out_train_root = Path(args.out_data_root) / "train" + + file_cache = {} + + def ensure_label(label: int): + if label not in file_cache: + name = class_names[label] + file_cache[label] = list_image_files(scan_train_root, out_train_root, name, label) + counts.setdefault(name, len(file_cache[label])) + print(f"listed label={label} class={name} files={len(file_cache[label])}", flush=True) + return file_cache[label] + + for name in real_names + fake_names: + if name not in counts: + label = name_to_label[name] + ensure_label(label) + + used_class_sets: dict[tuple[str, int], set[tuple[str, ...]]] = defaultdict(set) + rows = [] + + for repeat in repeats: + repeat_rng = random.Random(args.class_seed_base + repeat * 1000003) + for family, levels in [("real_domains", REAL_LEVELS), ("fake_methods", FAKE_LEVELS)]: + for level in levels: + if family == "real_domains": + variable_names, available, attempts = sample_classes( + repeat_rng, + real_names, + level, + counts, + args.per_side, + used_class_sets[(family, level)], + args.max_attempts, + ) + selected_names = variable_names + fake_names + else: + variable_names, available, attempts = sample_classes( + repeat_rng, + fake_names, + level, + counts, + args.per_side, + used_class_sets[(family, level)], + args.max_attempts, + ) + selected_names = real_names + variable_names + + selected_labels = {name_to_label[name] for name in selected_names} + real_labels = {label for label in selected_labels if label < 46} + fake_labels = {label for label in selected_labels if label >= 46} + + for label in real_labels | fake_labels: + ensure_label(label) + + image_rng = random.Random((args.image_seed * 1000003) + (repeat * 1009) + level) + real_samples, real_quotas = sample_from_filesystem(file_cache, real_labels, args.per_side, image_rng) + fake_samples, fake_quotas = sample_from_filesystem(file_cache, fake_labels, args.per_side, image_rng) + samples = real_samples + fake_samples + + out_name = ( + f"scaledf_v4_fixed{args.per_side}_" + f"{family}_n{level}_classseed{repeat}_imgseed{args.image_seed}_resampled.pkl" + ) + out_path = out_dir / out_name + dump_pickle(samples, out_path) + + rows.append( + { + "family": family, + "level": level, + "class_seed": repeat, + "real_images": len(real_samples), + "fake_images": len(fake_samples), + "status": "ok", + "error": "", + "path": str(out_path), + "variable_pool_images": available, + "resample_attempts": attempts, + "variable_classes_json": json.dumps(variable_names, ensure_ascii=True), + "real_quota_json": json.dumps(real_quotas, sort_keys=True), + "fake_quota_json": json.dumps(fake_quotas, sort_keys=True), + } + ) + print( + f"wrote repeat={repeat} family={family} level={level} " + f"real={len(real_samples)} fake={len(fake_samples)} attempts={attempts} path={out_path}", + flush=True, + ) + + manifest_path = out_dir / ( + f"manifest_fixed{args.per_side}_resampled_repeats_{args.repeats}_" + f"imgseed{args.image_seed}.tsv" + ) + fieldnames = [ + "family", + "level", + "class_seed", + "real_images", + "fake_images", + "status", + "error", + "path", + "variable_pool_images", + "resample_attempts", + "variable_classes_json", + "real_quota_json", + "fake_quota_json", + ] + out_dir.mkdir(parents=True, exist_ok=True) + with manifest_path.open("w", newline="", encoding="utf-8") as f: + writer = csv.DictWriter(f, fieldnames=fieldnames, delimiter="\t") + writer.writeheader() + writer.writerows(rows) + print(f"wrote {len(rows)} selections") + print(f"manifest={manifest_path}") + + +if __name__ == "__main__": + main() diff --git a/ScaleDF_rebuttal/make_fixed_budget_selections.py b/ScaleDF_rebuttal/make_fixed_budget_selections.py new file mode 100644 index 0000000000000000000000000000000000000000..42363bc415df81143ef6e4571a0b23bbb8a70485 --- /dev/null +++ b/ScaleDF_rebuttal/make_fixed_budget_selections.py @@ -0,0 +1,212 @@ +#!/usr/bin/env python3 +"""Build fixed-image-budget ScaleDF selections for rebuttal experiments.""" + +from __future__ import annotations + +import argparse +import os +import pickle +import random +from collections import defaultdict +from pathlib import Path + +IMAGE_EXTENSIONS = {".jpg", ".jpeg", ".png", ".bmp", ".webp"} + + +def load_pickle(path: Path): + with path.open("rb") as f: + return pickle.load(f) + + +def dump_pickle(obj, path: Path) -> None: + path.parent.mkdir(parents=True, exist_ok=True) + tmp = path.with_suffix(path.suffix + ".tmp") + with tmp.open("wb") as f: + pickle.dump(obj, f, protocol=pickle.HIGHEST_PROTOCOL) + os.replace(tmp, path) + + +def sample_from_labels(by_label, labels, k: int, rng: random.Random): + pool = [] + for label in sorted(labels): + pool.extend(by_label[label]) + if len(pool) < k: + raise ValueError(f"need {k} samples, only {len(pool)} available for labels={sorted(labels)[:8]}...") + return rng.sample(pool, k) + + +def list_image_files(scan_train_root: Path, out_train_root: Path, class_name: str, label: int): + class_root = scan_train_root / class_name + if not class_root.is_dir(): + raise FileNotFoundError(class_root) + + samples = [] + for dirpath, _, filenames in os.walk(class_root): + filenames.sort() + rel_dir = Path(dirpath).relative_to(scan_train_root) + for filename in filenames: + if Path(filename).suffix.lower() not in IMAGE_EXTENSIONS: + continue + samples.append((str(out_train_root / rel_dir / filename), label)) + if not samples: + raise ValueError(f"no images found under {class_root}") + return samples + + +def sample_from_filesystem(cache, labels, k: int, rng: random.Random): + labels = sorted(labels) + quotas = {label: 0 for label in labels} + remaining = set(labels) + need = k + + while need > 0 and remaining: + order = sorted(remaining) + rng.shuffle(order) + share, extra = divmod(need, len(order)) + next_remaining = set() + for idx, label in enumerate(order): + target = share + (1 if idx < extra else 0) + available = len(cache[label]) - quotas[label] + take = min(target, available) + quotas[label] += take + need -= take + if len(cache[label]) > quotas[label]: + next_remaining.add(label) + if next_remaining == remaining and share == 0 and extra == 0: + break + remaining = next_remaining + + if need != 0: + available = sum(len(cache[label]) for label in labels) + raise ValueError(f"need {k} samples, only {available} available for labels={labels}") + + samples = [] + for label in labels: + samples.extend(rng.sample(cache[label], quotas[label])) + rng.shuffle(samples) + return samples + + +def main() -> None: + parser = argparse.ArgumentParser() + parser.add_argument("--code-root", default="/tmp/ScaleDF_repro_20260724/code/deit") + parser.add_argument("--out-dir", required=True) + parser.add_argument("--per-side", type=int, default=25000) + parser.add_argument("--class-seeds", default="0,1,2") + parser.add_argument("--image-seed", type=int, default=20260725) + parser.add_argument("--source", choices=["base-pkl", "filesystem"], default="base-pkl") + parser.add_argument( + "--skip-insufficient", + action="store_true", + help="Record selections with too few available images in the manifest and continue.", + ) + parser.add_argument("--base-image-pkl", default="select_pkl/scaledf_v4_NRealFake_1_10_0.pkl") + parser.add_argument("--full-classes-pkl", default="scaledf_v4_train_fix.pkl") + parser.add_argument("--scan-data-root", default="/dev/shm/scaledf_bind_storage/ScaleDF") + parser.add_argument("--out-data-root", default="/dev/shm/scaledf_bind_storage/ScaleDF") + args = parser.parse_args() + + code_root = Path(args.code_root) + out_dir = Path(args.out_dir) + class_seeds = [int(x) for x in args.class_seeds.split(",") if x.strip()] + + full_classes = load_pickle(code_root / args.full_classes_pkl) + name_to_label = {name: idx for idx, name in enumerate(full_classes)} + + by_label = defaultdict(list) + fs_cache = {} + if args.source == "base-pkl": + base_images = load_pickle(code_root / args.base_image_pkl) + for path, label in base_images: + by_label[label].append((path, label)) + else: + scan_train_root = Path(args.scan_data_root) / "train" + out_train_root = Path(args.out_data_root) / "train" + + def ensure_label(label: int): + if label not in fs_cache: + fs_cache[label] = list_image_files( + scan_train_root, + out_train_root, + full_classes[label], + label, + ) + print(f"listed label={label} class={full_classes[label]} files={len(fs_cache[label])}") + return fs_cache[label] + + specs = [ + ("real_domains", [5, 8, 11, 16, 23, 32], "scaledf_v4_{n}_real_{s}.pkl"), + ("fake_methods", [11, 16, 22, 32, 45, 64], "scaledf_v4_{n}_{s}.pkl"), + ] + + manifest = [] + for family, levels, template in specs: + for level in levels: + for class_seed in class_seeds: + selected_names = load_pickle(code_root / "select_pkl" / template.format(n=level, s=class_seed)) + selected_labels = {name_to_label[name] for name in selected_names} + real_labels = {label for label in selected_labels if label < 46} + fake_labels = {label for label in selected_labels if label >= 46} + rng = random.Random((args.image_seed * 1000003) + (class_seed * 1009) + level) + out_name = ( + f"scaledf_v4_fixed{args.per_side}_" + f"{family}_n{level}_classseed{class_seed}_imgseed{args.image_seed}.pkl" + ) + out_path = out_dir / out_name + try: + if args.source == "base-pkl": + real_samples = sample_from_labels(by_label, real_labels, args.per_side, rng) + fake_samples = sample_from_labels(by_label, fake_labels, args.per_side, rng) + else: + for label in real_labels | fake_labels: + ensure_label(label) + real_samples = sample_from_filesystem(fs_cache, real_labels, args.per_side, rng) + fake_samples = sample_from_filesystem(fs_cache, fake_labels, args.per_side, rng) + samples = real_samples + fake_samples + dump_pickle(samples, out_path) + manifest.append( + ( + family, + level, + class_seed, + len(real_samples), + len(fake_samples), + "ok", + "", + str(out_path), + ) + ) + except ValueError as exc: + if not args.skip_insufficient: + raise + manifest.append( + ( + family, + level, + class_seed, + 0, + 0, + "skipped", + str(exc), + str(out_path), + ) + ) + print( + f"skipped family={family} level={level} class_seed={class_seed}: {exc}", + flush=True, + ) + + manifest_path = out_dir / ( + f"manifest_fixed{args.per_side}_seeds_{args.class_seeds}_" + f"imgseed{args.image_seed}_{args.source}.tsv" + ) + with manifest_path.open("w", encoding="utf-8") as f: + f.write("family\tlevel\tclass_seed\treal_images\tfake_images\tstatus\terror\tpath\n") + for row in manifest: + f.write("\t".join(map(str, row)) + "\n") + print(f"wrote {len(manifest)} selections") + print(f"manifest={manifest_path}") + + +if __name__ == "__main__": + main() diff --git a/ScaleDF_rebuttal/monitor_scaledf_rebuttal_jobs.sh b/ScaleDF_rebuttal/monitor_scaledf_rebuttal_jobs.sh new file mode 100644 index 0000000000000000000000000000000000000000..90281f80ce02fd5603256245fcea6450f9295c7d --- /dev/null +++ b/ScaleDF_rebuttal/monitor_scaledf_rebuttal_jobs.sh @@ -0,0 +1,167 @@ +#!/usr/bin/env bash +set -euo pipefail + +PERSIST_ROOT="${PERSIST_ROOT:-/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal}" +TRAIN_TAG="${TRAIN_TAG:-20260725_112559}" +ANCHOR_TAG="${ANCHOR_TAG:-20260725_1215}" +INTERVAL="${INTERVAL:-60}" +LOG_DIR="$PERSIST_ROOT/_host_logs/fixed500k_resampled_seed1_4_${TRAIN_TAG}" +MONITOR_DIR="$PERSIST_ROOT/_host_logs/rebuttal_monitor_${TRAIN_TAG}" +MONITOR_LOG="$MONITOR_DIR/monitor.log" +RESUME_SCRIPT="$PERSIST_ROOT/launch_fixed500k_resampled_host_queue_resume.sh" +WAIT_DRIVER="$PERSIST_ROOT/wait_train_eval_and_summarize_fixed500k_resampled.sh" + +mkdir -p "$MONITOR_DIR" + +labels=(n116-040-168 n116-041-153 n116-041-188 n111-046-081 n116-041-023 n116-040-213) +ports=(10338 11208 9869 11986 11693 9999) +hosts=( + 'fdbd:dccd:cdc1:1368:0:e4::' + 'fdbd:dccd:cdc1:1368:0:1d6::' + 'fdbd:dccd:cdc1:1368:0:1f8::' + 'fdbd:dccd:cdc1:1302:0:8d::' + 'fdbd:dccd:cdc1:1368:0:153::' + 'fdbd:dccd:cdc1:1368:0:111::' +) +real_levels=(5 8 11 16 23 32) +fake_levels=(11 16 22 32 45 64) + +ts() { + date -u +%Y-%m-%dT%H:%M:%SZ +} + +log() { + echo "[$(ts)] $*" | tee -a "$MONITOR_LOG" +} + +count_all_checkpoints() { + find "$PERSIST_ROOT/runs" -maxdepth 2 -type f -name best_checkpoint.pth \ + -path "*/control_fixed500000_*_classseed*_resampled_*_${TRAIN_TAG}/best_checkpoint.pth" | wc -l +} + +count_label_checkpoints() { + local label="$1" + find "$PERSIST_ROOT/runs" -maxdepth 2 -type f -name best_checkpoint.pth \ + -path "*/control_fixed500000_*_classseed*_resampled_${label}_${TRAIN_TAG}/best_checkpoint.pth" | wc -l +} + +queue_alive() { + local port="$1" + local host="$2" + local pid_file="$3" + ssh -o BatchMode=yes -o ConnectTimeout=8 -p "$port" "tiger@$host" \ + "pid=\$(cat '$pid_file' 2>/dev/null || true); [[ -n \"\$pid\" ]] && ps -p \"\$pid\" >/dev/null" >/dev/null 2>&1 +} + +recover_queue_if_needed() { + local idx="$1" + local label="${labels[$idx]}" + local port="${ports[$idx]}" + local host="${hosts[$idx]}" + local real_level="${real_levels[$idx]}" + local fake_level="${fake_levels[$idx]}" + local queue_log="$LOG_DIR/train_queue_${label}.log" + local pid_file="$LOG_DIR/train_queue_${label}.pid" + local ckpts + ckpts="$(count_label_checkpoints "$label")" + + if queue_alive "$port" "$host" "$pid_file"; then + log "train_queue label=$label alive=1 ckpts=$ckpts/8" + return + fi + + if [[ "$ckpts" -ge 8 ]]; then + log "train_queue label=$label alive=0 ckpts=$ckpts/8 action=mark_done_if_needed" + if ! grep -q 'queue_done_utc' "$queue_log" 2>/dev/null; then + echo "queue_done_utc=$(ts) monitor_confirmed_all_checkpoints=1" >> "$queue_log" + fi + return + fi + + log "ALERT train_queue label=$label died_with_missing_ckpts=$ckpts/8 action=relaunch_resume" + ssh -o BatchMode=yes -o ConnectTimeout=12 -p "$port" "tiger@$host" \ + "mkdir -p '$LOG_DIR'; nohup bash '$RESUME_SCRIPT' '$real_level' '$fake_level' '$TRAIN_TAG' >> '$queue_log' 2>&1 < /dev/null & echo \$! > '$pid_file'" +} + +check_train_errors() { + local err_file="$MONITOR_DIR/train_errors_latest.txt" + grep -RniE 'Traceback|RuntimeError|CUDA out of memory|No space left|ChildFailedError|NCCL.*(error|failed)|Connection reset|FileNotFoundError|OSError' "$LOG_DIR"/*.log \ + 2>/dev/null | grep -v 'NCCL INFO.*Failed to find' | tail -n 80 > "$err_file" || true + if [[ -s "$err_file" ]]; then + log "ALERT train_error_lines=$(wc -l < "$err_file") file=$err_file" + fi +} + +ensure_eval_driver() { + local driver_log="$LOG_DIR/auto_eval_driver2.log" + local pid_file="$LOG_DIR/auto_eval_driver2.pid" + if grep -q '\] all_done' "$driver_log" 2>/dev/null; then + log "eval_driver status=all_done" + return + fi + local pid + pid="$(cat "$pid_file" 2>/dev/null || true)" + if [[ -n "$pid" ]] && ps -p "$pid" >/dev/null 2>&1; then + log "eval_driver alive=1 pid=$pid" + return + fi + log "ALERT eval_driver alive=0 action=restart" + nohup bash "$WAIT_DRIVER" "$TRAIN_TAG" >> "$LOG_DIR/auto_eval_driver2_recover.log" 2>&1 < /dev/null & + echo $! > "$pid_file" +} + +check_anchor() { + local port=11451 + local host='fdbd:dccd:cdc1:1368:0:2f6::' + local run="$PERSIST_ROOT/runs/control_fixed500000_all_46real_91fake_imgseed20260725_n116-042-186_${ANCHOR_TAG}" + local train_log="$PERSIST_ROOT/_host_logs/all_46real_91fake_11451_${ANCHOR_TAG}/train_all_46real_91fake_retry1.log" + local best="$run/best_checkpoint.pth" + local status + status="$(ssh -o BatchMode=yes -o ConnectTimeout=8 -p "$port" "tiger@$host" \ + "if pgrep -af 'control_fixed500000_all_46real_91fake|main_cross_2direct_balance_realscale_perturb_v4_NRealFake.py' | grep -v pgrep >/dev/null; then echo alive; else echo dead; fi" 2>/dev/null || echo ssh_failed)" + local last + last="$(grep -E 'Epoch: \[[0-9]+\]|Averaged stats|Accuracy|Traceback|RuntimeError|ChildFailedError|CUDA out of memory|No space' "$train_log" 2>/dev/null | tail -n 1 || true)" + if [[ -f "$best" ]]; then + log "anchor status=$status checkpoint=ready last=${last:-none}" + elif [[ "$status" == alive ]]; then + log "anchor status=alive checkpoint=pending last=${last:-none}" + else + log "ALERT anchor status=$status checkpoint=pending last=${last:-none}" + fi +} + +status_once() { + log "cycle_start" + check_anchor + check_train_errors + for i in "${!labels[@]}"; do + recover_queue_if_needed "$i" + done + local ckpt_count + ckpt_count="$(count_all_checkpoints)" + local queue_done + queue_done="$(grep -l 'queue_done_utc' "$LOG_DIR"/train_queue_*.log 2>/dev/null | wc -l || true)" + local eval_summaries + eval_summaries="$(find "$PERSIST_ROOT/eval" -type f -name summary.txt -path "*${TRAIN_TAG}*" 2>/dev/null | wc -l)" + log "fixed500k ckpts=$ckpt_count/48 queue_done=$queue_done/6 eval_summaries=$eval_summaries" + ensure_eval_driver + log "cycle_end" +} + +if [[ "${ONCE:-0}" == 1 ]]; then + if ! status_once; then + rc=$? + log "ALERT monitor_cycle_failed rc=$rc" + exit "$rc" + fi + exit 0 +fi + +log "monitor_started train_tag=$TRAIN_TAG interval=${INTERVAL}s" +while true; do + if ! status_once; then + rc=$? + log "ALERT monitor_cycle_failed rc=$rc" + fi + sleep "$INTERVAL" +done diff --git a/ScaleDF_rebuttal/rebuttal_no_full_train_audit.py b/ScaleDF_rebuttal/rebuttal_no_full_train_audit.py new file mode 100644 index 0000000000000000000000000000000000000000..2f4a1ebe02458c0fd16a7da0fc48fc5a97e211aa --- /dev/null +++ b/ScaleDF_rebuttal/rebuttal_no_full_train_audit.py @@ -0,0 +1,254 @@ +#!/usr/bin/env python3 +"""Audit reviewer-requested ScaleDF rebuttal items that do not need full training.""" + +from __future__ import annotations + +import argparse +import csv +import json +import math +import re +from pathlib import Path + +import numpy as np +import pandas as pd + + +BENCH_ALIASES = { + "CDF V2": "CDFv2", + "ForgeryN": "ForgeryNet", + "DFF": "DeepFakeFace", + "DFD": "DeepFakeDetection", +} + +WORKBOOKS = { + "method_diversity_original": "Deepfake_scaling_law.xlsx", + "method_diversity_class_balanced": "Deepfake_scaling_law_class.xlsx", + "image_count": "deepfake_scaling_law_numbers.xlsx", + "fake_image_count": "deepfake_scaling_law_numbers_fake.xlsx", + "real_domain_diversity": "deepfake_scaling_law_real_domains.xlsx", + "model_size": "deepfake_scaling_model_size.xlsx", +} + +CATEGORY_LOG_GLOBS = [ + "train_scaledf_v1_cross_2direct_balance_ScaleDF_v4_quality_40_100_perturb_only_*/log.txt", + "train_scaledf_v1_cross_2direct_balance_ScaleDF_v4_quality_40_100_perturb_no_*/log.txt", +] + + +def parse_args() -> argparse.Namespace: + parser = argparse.ArgumentParser() + parser.add_argument("--repro-root", type=Path, default=Path("/tmp/ScaleDF_repro_20260724")) + parser.add_argument("--rebuttal-root", type=Path, default=Path("/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal")) + parser.add_argument("--out-dir", type=Path, required=True) + return parser.parse_args() + + +def canonical_bench(name: object) -> str: + return BENCH_ALIASES.get(str(name), str(name)) + + +def parse_number(value: object) -> float: + if pd.isna(value): + return math.nan + if isinstance(value, (int, float, np.integer, np.floating)): + return float(value) + raw = str(value) + is_percent = "%" in raw + text = raw.replace("%", "").replace("\xa0", "").strip() + try: + parsed = float(text) + except ValueError: + return math.nan + return parsed / 100.0 if is_percent else parsed + + +def parse_level(label: object) -> tuple[str, float, str]: + if pd.isna(label): + return "", math.nan, "" + text = str(label).strip() + if not text: + return "", math.nan, "" + if text.lower() == "all": + return text, math.nan, "all" + match = re.search(r"(\d+(?:\.\d+)?)", text) + level = float(match.group(1)) if match else math.nan + return text, level, text + + +def read_alternating_workbook(path: Path, family: str) -> list[dict[str, object]]: + raw = pd.read_excel(path, sheet_name=0) + bench_cols = [col for col in raw.columns if col != "Unnamed: 0"] + rows: list[dict[str, object]] = [] + for i in range(0, len(raw), 2): + if i + 1 >= len(raw): + break + label, level, replicate = parse_level(raw.iloc[i].get("Unnamed: 0")) + if not label: + continue + for metric, row_idx in [("auc", i), ("eer", i + 1)]: + for col in bench_cols: + value = parse_number(raw.iloc[row_idx].get(col)) + if math.isnan(value): + continue + rows.append( + { + "family": family, + "source_file": str(path), + "label": label, + "level": "" if math.isnan(level) else f"{level:g}", + "replicate": replicate, + "metric": metric, + "benchmark": canonical_bench(col), + "value": value, + } + ) + return rows + + +def summarize_workbook_rows(rows: list[dict[str, object]]) -> pd.DataFrame: + df = pd.DataFrame(rows) + if df.empty: + return pd.DataFrame() + df["level_num"] = pd.to_numeric(df["level"], errors="coerce") + out = ( + df[df["level_num"].notna()] + .groupby(["family", "metric", "benchmark", "level_num"], dropna=False)["value"] + .agg(["count", "mean", "std"]) + .reset_index() + .rename(columns={"level_num": "level", "count": "reps"}) + ) + return out + + +def parse_category_logs(log_root: Path) -> list[dict[str, object]]: + rows: list[dict[str, object]] = [] + for pattern in CATEGORY_LOG_GLOBS: + for path in sorted(log_root.glob(pattern)): + tag = path.parent.name.removeprefix( + "train_scaledf_v1_cross_2direct_balance_ScaleDF_v4_quality_40_100_perturb_" + ) + best = None + best_epoch = None + for line in path.read_text(errors="replace").splitlines(): + try: + record = json.loads(line) + except json.JSONDecodeError: + continue + acc = record.get("test_acc1") + epoch = record.get("epoch") + if acc is None: + continue + if best is None or float(acc) > best: + best = float(acc) + best_epoch = int(epoch) if epoch is not None else None + if best is not None: + rows.append( + { + "tag": tag, + "scope": "only" if tag.startswith("only_") else "without", + "category": tag.replace("only_", "").replace("no_", ""), + "metric": "ScaleDF_val_acc1", + "best_value": best, + "best_epoch": best_epoch, + "source_log": str(path), + } + ) + return rows + + +def latest_eval_tsv(root: Path) -> Path | None: + files = sorted((root / "analysis").glob("eval_results_*.tsv")) + return files[-1] if files else None + + +def write_rows(path: Path, rows: list[dict[str, object]]) -> None: + path.parent.mkdir(parents=True, exist_ok=True) + if not rows: + path.write_text("", encoding="utf-8") + return + fields = list(rows[0].keys()) + with path.open("w", newline="", encoding="utf-8") as f: + writer = csv.DictWriter(f, fieldnames=fields, delimiter="\t") + writer.writeheader() + writer.writerows(rows) + + +def write_status( + path: Path, + *, + workbook_rows: list[dict[str, object]], + workbook_summary: pd.DataFrame, + category_rows: list[dict[str, object]], + eval_path: Path | None, + scaling_dir: Path | None, +) -> None: + families = sorted({str(row["family"]) for row in workbook_rows}) + lines = [ + "# No-Full-Train Reviewer Audit", + "", + "Scope: reviewer-requested additions that can be answered from existing tables/logs/checkpoints, without launching a new full ScaleDF training run.", + "", + "## Completed Or Ready From Existing Artifacts", + "- Power-law exponent uncertainty, WLS, and robust/Theil-Sen fits: ready from the latest `rebuttal_scaling_stats_*` directory.", + "- Per-repetition exponent fits: ready from `per_repetition_powerlaw_fits.csv` and `per_repetition_powerlaw_fit_summary.csv` in the latest scaling stats directory.", + "- Per-benchmark scaling curves/fits for real-domain, fake-method, and image-count axes: ready from `level_summary_long.csv` and `powerlaw_fit_summary.csv`.", + "- Existing workbook AUC/EER tables parsed for: " + (", ".join(families) if families else "none found") + ".", + f"- Category leave-in/leave-out ScaleDF-val logs parsed: {len(category_rows)} rows; these are validation accuracy logs only, not cross-benchmark AUC/EER.", + ] + if eval_path: + lines.append(f"- Latest collected checkpoint eval TSV: `{eval_path}`.") + if scaling_dir: + lines.append(f"- Latest scaling stats directory: `{scaling_dir}`.") + + lines += [ + "", + "## Still Not Fully Done Without More Inputs", + "- Fixed-total-image-count diversity control: running small fixed-budget trainings now; this is not pure eval.", + "- Named forgery-category and generator-architecture cross-benchmark AUC/EER breakdown: not found as a complete ready table. Current artifacts include category/architecture selections and train-val logs, but not complete named-family benchmark metrics.", + "- Stronger/specialized detector or VLM experiments: no matching ready checkpoint found in the current repro bundle; would need external checkpoints/code or new training/inference setup.", + "- Deduplication and identity-overlap quantitative proof: not found as a ready numeric table in the current repro artifacts during this audit; needs either source metadata/face-id outputs or a new offline similarity/identity scan.", + "- DiFF benchmark: referenced in old workbooks but missing from the current downloaded benchmark tree, so current live eval queues skip it.", + "", + "## Output Files", + "- `existing_workbook_auc_eer_long.tsv`: long-form AUC/EER values parsed from xlsx workbooks.", + "- `existing_workbook_level_summary.tsv`: mean/std/repetition summary by family, metric, benchmark, and level.", + "- `category_leave_in_out_val.tsv`: existing JSON-log ScaleDF-val best Acc@1 for only/no category ablations.", + ] + path.write_text("\n".join(lines) + "\n", encoding="utf-8") + + +def main() -> None: + args = parse_args() + args.out_dir.mkdir(parents=True, exist_ok=True) + analysis_dir = args.repro_root / "analysis" + log_root = args.repro_root / "experiment_logs" + + workbook_rows: list[dict[str, object]] = [] + for family, filename in WORKBOOKS.items(): + path = analysis_dir / filename + if path.exists(): + workbook_rows.extend(read_alternating_workbook(path, family)) + + write_rows(args.out_dir / "existing_workbook_auc_eer_long.tsv", workbook_rows) + workbook_summary = summarize_workbook_rows(workbook_rows) + if not workbook_summary.empty: + workbook_summary.to_csv(args.out_dir / "existing_workbook_level_summary.tsv", sep="\t", index=False) + + category_rows = parse_category_logs(log_root) + write_rows(args.out_dir / "category_leave_in_out_val.tsv", category_rows) + + scaling_dirs = sorted((args.rebuttal_root / "analysis").glob("rebuttal_scaling_stats_*")) + write_status( + args.out_dir / "reviewer_no_full_train_status.md", + workbook_rows=workbook_rows, + workbook_summary=workbook_summary, + category_rows=category_rows, + eval_path=latest_eval_tsv(args.rebuttal_root), + scaling_dir=scaling_dirs[-1] if scaling_dirs else None, + ) + print(args.out_dir) + + +if __name__ == "__main__": + main() diff --git a/ScaleDF_rebuttal/rebuttal_scaling_stats.py b/ScaleDF_rebuttal/rebuttal_scaling_stats.py new file mode 100644 index 0000000000000000000000000000000000000000..a14916862acc9804ef4b4c2e5d9c40ab9c2097a6 --- /dev/null +++ b/ScaleDF_rebuttal/rebuttal_scaling_stats.py @@ -0,0 +1,708 @@ +#!/usr/bin/env python3 +"""Summarize ScaleDF scaling-law tables for rebuttal analysis.""" + +from __future__ import annotations + +import argparse +import datetime as dt +import json +import math +import re +from dataclasses import dataclass +from pathlib import Path + +import numpy as np +import pandas as pd + + +BENCH_ALIASES = { + "CDF V2": "CDFv2", + "ForgeryN": "ForgeryNet", + "DFF": "DeepFakeFace", + "DFD": "DeepFakeDetection", +} + + +@dataclass +class Family: + name: str + metric: str + level_col: str + df: pd.DataFrame + bench_cols: list[str] + source: str + fit_kind: str = "one_minus_auc" + + +def parse_args() -> argparse.Namespace: + parser = argparse.ArgumentParser() + parser.add_argument( + "--analysis-dir", + type=Path, + default=Path("/tmp/ScaleDF_repro_20260724/analysis"), + ) + parser.add_argument( + "--out-dir", + type=Path, + required=True, + ) + parser.add_argument("--bootstrap", type=int, default=5000) + parser.add_argument("--seed", type=int, default=20260725) + parser.add_argument( + "--fake-paper-full-n", + type=float, + default=88.0, + help="Paper-facing full fake-method N. Source CSV may label it as 91.", + ) + parser.add_argument( + "--include-diff", + action="store_true", + help="Keep DiFF if present. By default DiFF is excluded to match current paper tables.", + ) + return parser.parse_args() + + +def canonical_bench(name: str) -> str: + return BENCH_ALIASES.get(name, name) + + +def parse_metric_value(value: object) -> float: + if pd.isna(value): + return math.nan + if isinstance(value, (int, float, np.integer, np.floating)): + return float(value) + raw = str(value) + is_percent = "%" in raw + text = raw.replace("%", "").replace("\xa0", "").strip() + try: + parsed = float(text) + except ValueError: + return math.nan + return parsed / 100.0 if is_percent else parsed + + +def parse_level(value: object) -> float: + if pd.isna(value): + return math.nan + if isinstance(value, (pd.Timestamp, dt.datetime, dt.date)): + # Excel may auto-convert labels like "22-1" into 2025-01-22. + return float(value.day) + if isinstance(value, (int, float, np.integer, np.floating)): + return float(value) + text = str(value).strip() + match = re.match(r"^(\d+(?:\.\d+)?)", text) + if not match: + match = re.search(r"\d+(?:\.\d+)?", text) + if not match: + return math.nan + return float(match.group(0)) + + +def numeric_bench_cols( + df: pd.DataFrame, + level_col: str, + *, + include_diff: bool, +) -> list[str]: + exclude = {level_col, "Unnamed: 0", "mean", "replicate_id", "metric"} + cols: list[str] = [] + for col in df.columns: + if col in exclude: + continue + if not include_diff and str(col).strip().lower() == "diff": + continue + series = df[col].map(parse_metric_value) + if series.notna().any(): + df[col] = series + cols.append(col) + return cols + + +def finalize_frame( + df: pd.DataFrame, + level_col: str, + *, + include_diff: bool, + replicate_col: str | None = None, +) -> tuple[pd.DataFrame, list[str]]: + df = df.copy() + df[level_col] = pd.to_numeric(df[level_col], errors="coerce") + df = df[df[level_col].notna()].copy() + if replicate_col and replicate_col in df.columns: + df["replicate_id"] = df[replicate_col].astype(str) + elif "replicate_id" not in df.columns: + df["replicate_id"] = [f"row_{i}" for i in range(len(df))] + + bench_cols = numeric_bench_cols(df, level_col, include_diff=include_diff) + if "mean" in df.columns and pd.to_numeric(df["mean"], errors="coerce").notna().any(): + df["benchmark_mean"] = pd.to_numeric(df["mean"], errors="coerce") + else: + df["benchmark_mean"] = df[bench_cols].mean(axis=1, skipna=True) + return df.reset_index(drop=True), bench_cols + + +def read_csv_family( + analysis_dir: Path, + filename: str, + name: str, + metric: str, + level_col: str, + *, + include_diff: bool, + replicate_col: str | None = None, + source_note: str | None = None, +) -> Family: + path = analysis_dir / filename + df = pd.read_csv(path) + df, bench_cols = finalize_frame( + df, + level_col, + include_diff=include_diff, + replicate_col=replicate_col, + ) + return Family( + name=name, + metric=metric, + level_col=level_col, + df=df, + bench_cols=bench_cols, + source=source_note or str(path), + ) + + +def read_alternating_xlsx_family( + analysis_dir: Path, + filename: str, + name: str, + metric: str, + *, + include_diff: bool, + source_note: str | None = None, +) -> Family: + path = analysis_dir / filename + raw = pd.read_excel(path) + rows = [] + for i in range(0, len(raw), 2): + if i + 1 >= len(raw): + break + label = raw.iloc[i].get("Unnamed: 0") + level = parse_level(label) + if math.isnan(level): + continue + metric_row = raw.iloc[i + 1].copy() + metric_row["level"] = level + metric_row["replicate_id"] = str(label) + rows.append(metric_row) + if not rows: + raise ValueError(f"Could not parse alternating metric rows from {path}") + df = pd.DataFrame(rows) + df, bench_cols = finalize_frame(df, "level", include_diff=include_diff) + return Family( + name=name, + metric=metric, + level_col="level", + df=df, + bench_cols=bench_cols, + source=source_note or str(path), + fit_kind="eer", + ) + + +def clone_with_level_remap(family: Family, name: str, from_level: float, to_level: float, source_note: str) -> Family: + df = family.df.copy() + level_col = family.level_col + df.loc[np.isclose(df[level_col].astype(float), from_level), level_col] = to_level + return Family( + name=name, + metric=family.metric, + level_col=level_col, + df=df, + bench_cols=list(family.bench_cols), + source=source_note, + fit_kind=family.fit_kind, + ) + + +def bootstrap_ci(values: np.ndarray, rng: np.random.Generator, n_boot: int) -> tuple[float, float]: + values = values[np.isfinite(values)] + if len(values) == 0: + return math.nan, math.nan + if len(values) == 1 or n_boot <= 0: + return float(values[0]), float(values[0]) + idx = rng.integers(0, len(values), size=(n_boot, len(values))) + means = values[idx].mean(axis=1) + return tuple(np.percentile(means, [2.5, 97.5]).astype(float)) + + +def summarize_levels(family: Family, rng: np.random.Generator, n_boot: int) -> tuple[pd.DataFrame, pd.DataFrame]: + rows = [] + long_rows = [] + targets = [("benchmark_mean", "Mean")] + [(col, canonical_bench(col)) for col in family.bench_cols] + for level, group in family.df.groupby(family.level_col, sort=True): + for col, target in targets: + vals = pd.to_numeric(group[col], errors="coerce").to_numpy(dtype=float) + vals = vals[np.isfinite(vals)] + lo, hi = bootstrap_ci(vals, rng, n_boot) + row = { + "family": family.name, + "metric": family.metric, + "level": float(level), + "target": target, + "reps": int(len(vals)), + "mean": float(np.mean(vals)) if len(vals) else math.nan, + "std": float(np.std(vals, ddof=1)) if len(vals) > 1 else 0.0, + "se": float(np.std(vals, ddof=1) / math.sqrt(len(vals))) if len(vals) > 1 else 0.0, + "ci95_low": lo, + "ci95_high": hi, + } + long_rows.append(row) + mean_vals = pd.to_numeric(group["benchmark_mean"], errors="coerce").to_numpy(dtype=float) + mean_vals = mean_vals[np.isfinite(mean_vals)] + lo, hi = bootstrap_ci(mean_vals, rng, n_boot) + rows.append( + { + "family": family.name, + "metric": family.metric, + "level": float(level), + "reps": int(len(mean_vals)), + "mean": float(np.mean(mean_vals)) if len(mean_vals) else math.nan, + "std": float(np.std(mean_vals, ddof=1)) if len(mean_vals) > 1 else 0.0, + "se": float(np.std(mean_vals, ddof=1) / math.sqrt(len(mean_vals))) if len(mean_vals) > 1 else 0.0, + "ci95_low": lo, + "ci95_high": hi, + } + ) + return pd.DataFrame(rows), pd.DataFrame(long_rows) + + +def linear_fit(log_x: np.ndarray, log_y: np.ndarray, weights: np.ndarray | None = None) -> tuple[float, float]: + if weights is None: + slope, intercept = np.polyfit(log_x, log_y, 1) + else: + slope, intercept = np.polyfit(log_x, log_y, 1, w=np.sqrt(weights)) + return float(slope), float(intercept) + + +def r2_score(log_x: np.ndarray, log_y: np.ndarray, slope: float, intercept: float) -> float: + pred = slope * log_x + intercept + denom = np.sum((log_y - log_y.mean()) ** 2) + if denom <= 0: + return math.nan + return float(1.0 - np.sum((log_y - pred) ** 2) / denom) + + +def robust_slope(log_x: np.ndarray, log_y: np.ndarray) -> float: + try: + from sklearn.linear_model import TheilSenRegressor + + model = TheilSenRegressor(random_state=0, fit_intercept=True) + model.fit(log_x.reshape(-1, 1), log_y) + return float(model.coef_[0]) + except Exception: + slopes = [] + for i in range(len(log_x)): + for j in range(i + 1, len(log_x)): + if log_x[j] != log_x[i]: + slopes.append((log_y[j] - log_y[i]) / (log_x[j] - log_x[i])) + return float(np.median(slopes)) if slopes else math.nan + + +def fit_powerlaw( + family: Family, + target_col: str, + target_name: str, + level_summary: pd.DataFrame, + rng: np.random.Generator, + n_boot: int, +) -> dict[str, object]: + grouped = family.df.groupby(family.level_col, sort=True) + levels = np.array(list(grouped.groups.keys()), dtype=float) + means = np.array( + [pd.to_numeric(grouped.get_group(level)[target_col], errors="coerce").mean() for level in levels], + dtype=float, + ) + if family.fit_kind == "eer": + errors = np.clip(means, 1e-8, None) + response_name = "eer" + else: + errors = np.clip(1.0 - means, 1e-8, None) + response_name = "1-auc" + + valid = np.isfinite(levels) & (levels > 0) & np.isfinite(errors) & (errors > 0) + levels = levels[valid] + errors = errors[valid] + if len(levels) < 3: + return { + "family": family.name, + "metric": family.metric, + "target": target_name, + "response": response_name, + "n_levels": int(len(levels)), + "alpha_ols": math.nan, + "alpha_ols_ci95_low": math.nan, + "alpha_ols_ci95_high": math.nan, + "r2_ols": math.nan, + "alpha_wls": math.nan, + "r2_wls": math.nan, + "alpha_theilsen": math.nan, + } + + log_x = np.log(levels) + log_y = np.log(errors) + slope, intercept = linear_fit(log_x, log_y) + alpha_ols = -slope + r2_ols = r2_score(log_x, log_y, slope, intercept) + + weight_rows = [] + summary_target = level_summary[level_summary["target"].eq(target_name)] + for level, err in zip(levels, errors): + row = summary_target[np.isclose(summary_target["level"].astype(float), level)] + if row.empty: + weight_rows.append(math.nan) + continue + std = float(row["std"].iloc[0]) + reps = max(int(row["reps"].iloc[0]), 1) + if std > 0 and err > 0: + weight_rows.append(reps / ((std / err) ** 2)) + else: + weight_rows.append(math.nan) + weights = np.array(weight_rows, dtype=float) + finite_weights = weights[np.isfinite(weights) & (weights > 0)] + if len(finite_weights): + fill = float(np.median(finite_weights)) + weights = np.where(np.isfinite(weights) & (weights > 0), weights, fill) + w_slope, w_intercept = linear_fit(log_x, log_y, weights=weights) + alpha_wls = -w_slope + r2_wls = r2_score(log_x, log_y, w_slope, w_intercept) + else: + alpha_wls = math.nan + r2_wls = math.nan + + alpha_theilsen = -robust_slope(log_x, log_y) + + boot_alphas = [] + if n_boot > 0: + group_map = {float(level): grouped.get_group(level) for level in levels} + for _ in range(n_boot): + sampled_means = [] + for level in levels: + values = pd.to_numeric(group_map[float(level)][target_col], errors="coerce").to_numpy(dtype=float) + values = values[np.isfinite(values)] + if len(values) == 0: + sampled_means.append(math.nan) + elif len(values) == 1: + sampled_means.append(float(values[0])) + else: + sampled_means.append(float(values[rng.integers(0, len(values), size=len(values))].mean())) + sampled_means = np.array(sampled_means, dtype=float) + sampled_errors = np.clip(sampled_means if family.fit_kind == "eer" else 1.0 - sampled_means, 1e-8, None) + ok = np.isfinite(sampled_errors) & (sampled_errors > 0) + if ok.sum() >= 3: + b_slope, _ = linear_fit(np.log(levels[ok]), np.log(sampled_errors[ok])) + boot_alphas.append(-b_slope) + if boot_alphas: + ci_low, ci_high = np.percentile(boot_alphas, [2.5, 97.5]).astype(float) + else: + ci_low = ci_high = math.nan + + return { + "family": family.name, + "metric": family.metric, + "target": target_name, + "response": response_name, + "n_levels": int(len(levels)), + "alpha_ols": float(alpha_ols), + "alpha_ols_ci95_low": ci_low, + "alpha_ols_ci95_high": ci_high, + "r2_ols": float(r2_ols), + "alpha_wls": float(alpha_wls), + "r2_wls": float(r2_wls), + "alpha_theilsen": float(alpha_theilsen), + } + + +def fit_powerlaw_from_points( + family_name: str, + metric: str, + target_name: str, + fit_kind: str, + levels: np.ndarray, + means: np.ndarray, +) -> dict[str, object]: + if fit_kind == "eer": + errors = np.clip(means, 1e-8, None) + response_name = "eer" + else: + errors = np.clip(1.0 - means, 1e-8, None) + response_name = "1-auc" + + valid = np.isfinite(levels) & (levels > 0) & np.isfinite(errors) & (errors > 0) + levels = levels[valid] + errors = errors[valid] + if len(levels) < 3: + return { + "family": family_name, + "metric": metric, + "target": target_name, + "response": response_name, + "n_levels": int(len(levels)), + "alpha_ols": math.nan, + "r2_ols": math.nan, + "alpha_theilsen": math.nan, + } + + log_x = np.log(levels) + log_y = np.log(errors) + slope, intercept = linear_fit(log_x, log_y) + return { + "family": family_name, + "metric": metric, + "target": target_name, + "response": response_name, + "n_levels": int(len(levels)), + "alpha_ols": float(-slope), + "r2_ols": float(r2_score(log_x, log_y, slope, intercept)), + "alpha_theilsen": float(-robust_slope(log_x, log_y)), + } + + +def per_repetition_fits(family: Family) -> pd.DataFrame: + df = family.df.copy().reset_index(drop=True) + level_col = family.level_col + targets = [("benchmark_mean", "Mean")] + [(col, canonical_bench(col)) for col in family.bench_cols] + + max_level = float(pd.to_numeric(df[level_col], errors="coerce").max()) + full_rows = df[np.isclose(pd.to_numeric(df[level_col], errors="coerce"), max_level)].copy() + repeated_rows = df[~np.isclose(pd.to_numeric(df[level_col], errors="coerce"), max_level)].copy() + if repeated_rows.empty: + return pd.DataFrame() + + repeated_rows["_rep_slot"] = repeated_rows.groupby(level_col, sort=False).cumcount() + slots = sorted(repeated_rows["_rep_slot"].dropna().astype(int).unique()) + if not slots: + return pd.DataFrame() + + out_rows = [] + for slot in slots: + slot_rows = repeated_rows[repeated_rows["_rep_slot"].eq(slot)].copy() + if not full_rows.empty: + slot_rows = pd.concat([slot_rows, full_rows.iloc[:1]], ignore_index=True) + for target_col, target_name in targets: + values = pd.to_numeric(slot_rows[target_col], errors="coerce").to_numpy(dtype=float) + levels = pd.to_numeric(slot_rows[level_col], errors="coerce").to_numpy(dtype=float) + fit = fit_powerlaw_from_points( + family.name, + family.metric, + target_name, + family.fit_kind, + levels, + values, + ) + fit["replicate_slot"] = int(slot) + 1 + fit["levels"] = ",".join(f"{x:g}" for x in levels[np.isfinite(levels)]) + out_rows.append(fit) + return pd.DataFrame(out_rows) + + +def summarize_per_repetition_fits(per_rep: pd.DataFrame) -> pd.DataFrame: + if per_rep.empty: + return pd.DataFrame() + rows = [] + for keys, group in per_rep.groupby(["family", "metric", "target", "response"], sort=True): + alphas = pd.to_numeric(group["alpha_ols"], errors="coerce").dropna() + robust = pd.to_numeric(group["alpha_theilsen"], errors="coerce").dropna() + rows.append( + { + "family": keys[0], + "metric": keys[1], + "target": keys[2], + "response": keys[3], + "replicate_fits": int(len(alphas)), + "alpha_ols_mean": float(alphas.mean()) if len(alphas) else math.nan, + "alpha_ols_std": float(alphas.std(ddof=1)) if len(alphas) > 1 else 0.0, + "alpha_ols_min": float(alphas.min()) if len(alphas) else math.nan, + "alpha_ols_max": float(alphas.max()) if len(alphas) else math.nan, + "alpha_theilsen_mean": float(robust.mean()) if len(robust) else math.nan, + "r2_ols_mean": float(pd.to_numeric(group["r2_ols"], errors="coerce").mean()), + } + ) + return pd.DataFrame(rows) + + +def write_markdown(out_dir: Path, notes: dict[str, object], level_summary: pd.DataFrame, fits: pd.DataFrame) -> None: + lines = [ + "# ScaleDF Rebuttal Scaling Statistics", + "", + "This report is generated from the reproduced analysis files, excluding DiFF by default to match the current paper tables.", + "", + "## Sources", + ] + for name, source in notes["sources"].items(): + lines.append(f"- `{name}`: `{source}`") + if notes.get("fake_full_note"): + lines += ["", "## Fake-Method N Note", str(notes["fake_full_note"])] + + lines += ["", "## Mean AUC/EER By Level"] + for family in level_summary["family"].drop_duplicates(): + sub = level_summary[level_summary["family"].eq(family)].copy() + lines.append("") + lines.append(f"### {family}") + for _, row in sub.sort_values("level").iterrows(): + lines.append( + f"- level={row['level']:g}, reps={int(row['reps'])}, mean={row['mean']:.4f}, " + f"95% CI [{row['ci95_low']:.4f}, {row['ci95_high']:.4f}]" + ) + + mean_fits = fits[fits["target"].eq("Mean")].copy() + lines += ["", "## Mean Power-Law Fits"] + for _, row in mean_fits.sort_values(["family", "metric"]).iterrows(): + lines.append( + f"- `{row['family']}` ({row['response']}): " + f"OLS alpha={row['alpha_ols']:.4f} " + f"[{row['alpha_ols_ci95_low']:.4f}, {row['alpha_ols_ci95_high']:.4f}], " + f"WLS alpha={row['alpha_wls']:.4f}, Theil-Sen alpha={row['alpha_theilsen']:.4f}, " + f"OLS R2={row['r2_ols']:.4f}" + ) + + lines += [ + "", + "## Files", + "- `level_summary_mean.csv`: mean across benchmarks by level.", + "- `level_summary_long.csv`: per-benchmark rows by level.", + "- `powerlaw_fit_summary.csv`: OLS, WLS, robust, and bootstrap-CI exponents.", + "- `per_repetition_powerlaw_fits.csv`: OLS and Theil-Sen exponents fitted separately for each replicate slot.", + "- `per_repetition_powerlaw_fit_summary.csv`: summary statistics across per-repetition exponent fits.", + "- `metadata.json`: run metadata and source notes.", + ] + (out_dir / "README.md").write_text("\n".join(lines) + "\n", encoding="utf-8") + + +def main() -> None: + args = parse_args() + args.out_dir.mkdir(parents=True, exist_ok=True) + rng = np.random.default_rng(args.seed) + + families: list[Family] = [] + sources: dict[str, str] = {} + + real_auc = read_csv_family( + args.analysis_dir, + "replicates_6_real.csv", + "real_domains_auc", + "auc", + "N", + include_diff=args.include_diff, + replicate_col="Unnamed: 0", + ) + families.append(real_auc) + + fake_auc_src = read_csv_family( + args.analysis_dir, + "scaling_law_1_auc.csv", + "fake_methods_auc_sourceN", + "auc", + "N", + include_diff=args.include_diff, + ) + families.append(fake_auc_src) + + max_fake_n = float(fake_auc_src.df[fake_auc_src.level_col].max()) + fake_full_note = "" + if args.fake_paper_full_n and not np.isclose(max_fake_n, args.fake_paper_full_n): + fake_auc_paper = clone_with_level_remap( + fake_auc_src, + "fake_methods_auc_paperN", + max_fake_n, + args.fake_paper_full_n, + f"{fake_auc_src.source}; full row remapped from N={max_fake_n:g} to N={args.fake_paper_full_n:g}", + ) + families.append(fake_auc_paper) + fake_full_note = ( + f"Source fake-method CSV labels the full-method row as N={max_fake_n:g}; " + f"the paper-facing sensitivity table also reports a version with full N={args.fake_paper_full_n:g}." + ) + + image_auc = read_csv_family( + args.analysis_dir, + "fake_real.csv", + "image_count_auc", + "auc", + "size", + include_diff=args.include_diff, + replicate_col="Unnamed: 0", + ) + families.append(image_auc) + + for filename, name in [ + ("realdomains_6.xlsx", "real_domains_eer"), + ("scaling_deepfake_methods_6.xlsx", "fake_methods_eer_sourceN"), + ]: + try: + fam = read_alternating_xlsx_family( + args.analysis_dir, + filename, + name, + "eer", + include_diff=args.include_diff, + ) + families.append(fam) + if name.startswith("fake_methods"): + max_n = float(fam.df[fam.level_col].max()) + if args.fake_paper_full_n and not np.isclose(max_n, args.fake_paper_full_n): + families.append( + clone_with_level_remap( + fam, + "fake_methods_eer_paperN", + max_n, + args.fake_paper_full_n, + f"{fam.source}; full row remapped from N={max_n:g} to N={args.fake_paper_full_n:g}", + ) + ) + except Exception as exc: + print(f"Warning: skipped {filename}: {exc}") + + level_frames = [] + long_frames = [] + fit_rows = [] + per_rep_frames = [] + for family in families: + sources[family.name] = family.source + level_mean, level_long = summarize_levels(family, rng, args.bootstrap) + level_frames.append(level_mean) + long_frames.append(level_long) + per_rep = per_repetition_fits(family) + if not per_rep.empty: + per_rep_frames.append(per_rep) + for target_col, target_name in [("benchmark_mean", "Mean")] + [ + (col, canonical_bench(col)) for col in family.bench_cols + ]: + fit_rows.append(fit_powerlaw(family, target_col, target_name, level_long, rng, args.bootstrap)) + + level_summary = pd.concat(level_frames, ignore_index=True) + level_long = pd.concat(long_frames, ignore_index=True) + fits = pd.DataFrame(fit_rows) + per_rep_fits = pd.concat(per_rep_frames, ignore_index=True) if per_rep_frames else pd.DataFrame() + per_rep_summary = summarize_per_repetition_fits(per_rep_fits) + + level_summary.to_csv(args.out_dir / "level_summary_mean.csv", index=False) + level_long.to_csv(args.out_dir / "level_summary_long.csv", index=False) + fits.to_csv(args.out_dir / "powerlaw_fit_summary.csv", index=False) + per_rep_fits.to_csv(args.out_dir / "per_repetition_powerlaw_fits.csv", index=False) + per_rep_summary.to_csv(args.out_dir / "per_repetition_powerlaw_fit_summary.csv", index=False) + + metadata = { + "analysis_dir": str(args.analysis_dir), + "bootstrap": args.bootstrap, + "seed": args.seed, + "include_diff": args.include_diff, + "fake_paper_full_n": args.fake_paper_full_n, + "fake_full_note": fake_full_note, + "sources": sources, + } + (args.out_dir / "metadata.json").write_text(json.dumps(metadata, indent=2) + "\n", encoding="utf-8") + write_markdown(args.out_dir, metadata, level_summary, fits) + print(args.out_dir) + + +if __name__ == "__main__": + main() diff --git a/ScaleDF_rebuttal/scaledf_dataset_stats_n116-041-246.tsv b/ScaleDF_rebuttal/scaledf_dataset_stats_n116-041-246.tsv new file mode 100644 index 0000000000000000000000000000000000000000..70865c7eaed8fa064f906f0b2dc5cb3fea387bdb --- /dev/null +++ b/ScaleDF_rebuttal/scaledf_dataset_stats_n116-041-246.tsv @@ -0,0 +1,137 @@ +section class name image_files total_files subdirs errors path exts_json +scaledf_val fake fake 70000 70000 0 0 /dev/shm/scaledf_bind_storage/Established_benchmarks/ScaleDF/val/fake "{"".jpg"": 70000}" +scaledf_val real 000real_faces 19729 19729 0 0 /dev/shm/scaledf_bind_storage/Established_benchmarks/ScaleDF/val/000real_faces "{"".jpg"": 19729}" +train fake AMatrix_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/AMatrix_faces "{"".jpg"": 100000}" +train fake AniPortrait_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/AniPortrait_faces "{"".jpg"": 100000}" +train fake AniTalker_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/AniTalker_faces "{"".jpg"": 100000}" +train fake AnySD_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/AnySD_faces "{"".jpg"": 100000}" +train fake Audio2Head_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/Audio2Head_faces "{"".jpg"": 100000}" +train fake blendface_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/blendface_faces "{"".jpg"": 100000}" +train fake cogview3_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/cogview3_faces "{"".jpg"": 100000}" +train fake cogview4_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/cogview4_faces "{"".jpg"": 100000}" +train fake CSCS_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/CSCS_faces "{"".jpg"": 100000}" +train fake CVTHead_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/CVTHead_faces "{"".jpg"": 100000}" +train fake DaGAN_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/DaGAN_faces "{"".jpg"": 100000}" +train fake DAWN_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/DAWN_faces "{"".jpg"": 100000}" +train fake diff2lip_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/diff2lip_faces "{"".jpg"": 100000}" +train fake DiffusionCLIP_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/DiffusionCLIP_faces "{"".jpg"": 100000}" +train fake ditto_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/ditto_faces "{"".jpg"": 100000}" +train fake dreamtalk_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/dreamtalk_faces "{"".jpg"": 100000}" +train fake E4E_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/E4E_faces "{"".jpg"": 100000}" +train fake e4s_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/e4s_faces "{"".jpg"": 100000}" +train fake echomimic_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/echomimic_faces "{"".jpg"": 100000}" +train fake EDICT_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/EDICT_faces "{"".jpg"": 100000}" +train fake emostyle_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/emostyle_faces "{"".jpg"": 100000}" +train fake faceadapter_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/faceadapter_faces "{"".jpg"": 100000}" +train fake FaceDancer_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/FaceDancer_faces "{"".jpg"": 100000}" +train fake facefusion_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/facefusion_faces "{"".jpg"": 100000}" +train fake FaceID_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/FaceID_faces "{"".jpg"": 100000}" +train fake faceswap_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/faceswap_faces "{"".jpg"": 100000}" +train fake facevid2vid_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/facevid2vid_faces "{"".jpg"": 100000}" +train fake FFpp_Deepfakes 101877 101877 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/FFpp_Deepfakes "{"".jpg"": 101877}" +train fake FFpp_Face2Face 102174 102174 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/FFpp_Face2Face "{"".jpg"": 102174}" +train fake FFpp_FaceSwap 81580 81580 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/FFpp_FaceSwap "{"".jpg"": 81580}" +train fake FFpp_NeuralTextures 81567 81567 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/FFpp_NeuralTextures "{"".jpg"": 81567}" +train fake flux_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/flux_faces "{"".jpg"": 100000}" +train fake FluxEdit_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/FluxEdit_faces "{"".jpg"": 100000}" +train fake followyouremoji_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/followyouremoji_faces "{"".jpg"": 100000}" +train fake fomm_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/fomm_faces "{"".jpg"": 100000}" +train fake fsgan_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/fsgan_faces "{"".jpg"": 100000}" +train fake fsrt_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/fsrt_faces "{"".jpg"": 100000}" +train fake ghost_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/ghost_faces "{"".jpg"": 100000}" +train fake hififace_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/hififace_faces "{"".jpg"": 100000}" +train fake hunyuan_dit_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/hunyuan_dit_faces "{"".jpg"": 100000}" +train fake hunyuanvideo_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/hunyuanvideo_faces "{"".jpg"": 100000}" +train fake HyperReenact_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/HyperReenact_faces "{"".jpg"": 100000}" +train fake infoswap_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/infoswap_faces "{"".jpg"": 100000}" +train fake instantid_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/instantid_faces "{"".jpg"": 100000}" +train fake InstructP2P_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/InstructP2P_faces "{"".jpg"": 100000}" +train fake inswapper_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/inswapper_faces "{"".jpg"": 100000}" +train fake IP_LAP_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/IP_LAP_faces "{"".jpg"": 100000}" +train fake IPAdapter_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/IPAdapter_faces "{"".jpg"": 100000}" +train fake janus_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/janus_faces "{"".jpg"": 100000}" +train fake JoyHallo_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/JoyHallo_faces "{"".jpg"": 100000}" +train fake JoyVASA_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/JoyVASA_faces "{"".jpg"": 100000}" +train fake kolors_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/kolors_faces "{"".jpg"": 100000}" +train fake LIA_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/LIA_faces "{"".jpg"": 100000}" +train fake LivePortrait_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/LivePortrait_faces "{"".jpg"": 100000}" +train fake ltx_video_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/ltx_video_faces "{"".jpg"": 100000}" +train fake MagicFace_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/MagicFace_faces "{"".jpg"": 100000}" +train fake MakeItTalk_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/MakeItTalk_faces "{"".jpg"": 100000}" +train fake MaskFaceGAN_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/MaskFaceGAN_faces "{"".jpg"": 100000}" +train fake MCNET_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/MCNET_faces "{"".jpg"": 100000}" +train fake megactor_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/megactor_faces "{"".jpg"": 100000}" +train fake midjourney_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/midjourney_faces "{"".jpg"": 100000}" +train fake MobileFaceSwap_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/MobileFaceSwap_faces "{"".jpg"": 100000}" +train fake MRFA_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/MRFA_faces "{"".jpg"": 100000}" +train fake OneShot_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/OneShot_faces "{"".jpg"": 100000}" +train fake pika_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/pika_faces "{"".jpg"": 100000}" +train fake pixart_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/pixart_faces "{"".jpg"": 100000}" +train fake pyramidal_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/pyramidal_faces "{"".jpg"": 100000}" +train fake Real3DPortrait_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/Real3DPortrait_faces "{"".jpg"": 100000}" +train fake REFace_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/REFace_faces "{"".jpg"": 100000}" +train fake RFInversion_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/RFInversion_faces "{"".jpg"": 100000}" +train fake RigFace_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/RigFace_faces "{"".jpg"": 100000}" +train fake SadTalker_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/SadTalker_faces "{"".jpg"": 100000}" +train fake sd15_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/sd15_faces "{"".jpg"": 100000}" +train fake sd2_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/sd2_faces "{"".jpg"": 100000}" +train fake sd35_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/sd35_faces "{"".jpg"": 100000}" +train fake SDEdit_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/SDEdit_faces "{"".jpg"": 100000}" +train fake SDFlow_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/SDFlow_faces "{"".jpg"": 100000}" +train fake sdxl_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/sdxl_faces "{"".jpg"": 100000}" +train fake simswap_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/simswap_faces "{"".jpg"": 100000}" +train fake stylegan2_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/stylegan2_faces "{"".jpg"": 100000}" +train fake stylegan3_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/stylegan3_faces "{"".jpg"": 100000}" +train fake styleganxl_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/styleganxl_faces "{"".jpg"": 100000}" +train fake StyleMask_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/StyleMask_faces "{"".jpg"": 100000}" +train fake uniface_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/uniface_faces "{"".jpg"": 100000}" +train fake VecGAN_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/VecGAN_faces "{"".jpg"": 100000}" +train fake video-retalking_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/video-retalking_faces "{"".jpg"": 100000}" +train fake wan_t2v_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/wan_t2v_faces "{"".jpg"": 100000}" +train fake Wav2Lip_faces 100000 100000 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/Wav2Lip_faces "{"".jpg"": 100000}" +train real 000000Adience 18165 18165 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/000000Adience "{"".jpg"": 18165}" +train real 000000AFAD 129901 129901 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/000000AFAD "{"".jpg"": 129901}" +train real 000000afew-va 22644 22644 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/000000afew-va "{"".jpg"": 22644}" +train real 000000Aff2_wild 184669 184669 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/000000Aff2_wild "{"".jpg"": 184669}" +train real 000000AffectNet 30255 30255 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/000000AffectNet "{"".jpg"": 30255}" +train real 000000AgeDB 15157 15157 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/000000AgeDB "{"".jpg"": 15157}" +train real 000000Asian-Celeb 282618 282618 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/000000Asian-Celeb "{"".jpg"": 282618}" +train real 000000AVA 127021 127021 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/000000AVA "{"".jpg"": 127021}" +train real 000000AVS 270594 270594 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/000000AVS "{"".jpg"": 270594}" +train real 000000BUPT-CBFace 273658 273658 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/000000BUPT-CBFace "{"".jpg"": 273658}" +train real 000000CACD 161552 161552 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/000000CACD "{"".jpg"": 161552}" +train real 000000CASIA-WebFace 245880 245880 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/000000CASIA-WebFace "{"".jpg"": 245880}" +train real 000000celeba 194716 194716 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/000000celeba "{"".jpg"": 194716}" +train real 000000celebs500k 289137 289137 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/000000celebs500k "{"".jpg"": 289137}" +train real 000000CelebV-HQ 205724 205724 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/000000CelebV-HQ "{"".jpg"": 205724}" +train real 000000crema-d 19099 19099 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/000000crema-d "{"".jpg"": 19099}" +train real 000000DFEW 37020 37020 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/000000DFEW "{"".jpg"": 37020}" +train real 000000ExpW 67399 67399 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/000000ExpW "{"".jpg"": 67399}" +train real 000000facescrub 41293 41293 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/000000facescrub "{"".bmp"": 22, "".jpeg"": 40391, "".jpg"": 880}" +train real 000000FaceVid 48361 48361 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/000000FaceVid "{"".jpg"": 48361}" +train real 000000FairFace 79486 79486 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/000000FairFace "{"".jpg"": 79486}" +train real 000000ffhq 51913 51913 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/000000ffhq "{"".jpg"": 51913}" +train real 000000FFpp 102193 102193 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/000000FFpp "{"".jpg"": 102193}" +train real 000000glint_360k 258545 258545 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/000000glint_360k "{"".jpg"": 258545}" +train real 000000GRID_Corpus 98627 98627 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/000000GRID_Corpus "{"".jpg"": 98627}" +train real 000000ijb 223649 223649 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/000000ijb "{"".jpg"": 223649}" +train real 000000IMDB 243426 243426 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/000000IMDB "{"".jpg"": 243426}" +train real 000000IMFDB 10116 10116 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/000000IMFDB "{"".jpg"": 10116}" +train real 000000LFW 13166 13166 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/000000LFW "{"".jpg"": 13166}" +train real 000000MEAD 289071 289071 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/000000MEAD "{"".jpg"": 289071}" +train real 000000megaage 40998 40998 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/000000megaage "{"".jpg"": 40998}" +train real 000000MMA 79535 79535 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/000000MMA "{"".jpg"": 79535}" +train real 000000MORPH-2 49999 49999 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/000000MORPH-2 "{"".jpg"": 49999}" +train real 000000Multi_Pie 151707 151707 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/000000Multi_Pie "{"".jpg"": 151707}" +train real 000000RAF-DB 9829 9829 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/000000RAF-DB "{"".jpg"": 9829}" +train real 000000RAVDESS 20185 20185 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/000000RAVDESS "{"".jpg"": 20185}" +train real 000000RMFD 22784 22784 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/000000RMFD "{"".jpeg"": 2, "".jpg"": 22782}" +train real 000000samm_v3 11722 11722 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/000000samm_v3 "{"".jpg"": 11722}" +train real 000000Speaking_Faces 219820 219820 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/000000Speaking_Faces "{"".jpg"": 219820}" +train real 000000UMD 261370 261370 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/000000UMD "{"".jpg"": 261370}" +train real 000000UTKface 23621 23621 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/000000UTKface "{"".jpg"": 23621}" +train real 000000VGG 269674 269674 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/000000VGG "{"".jpg"": 269674}" +train real 000000vox2 274227 274227 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/000000vox2 "{"".jpg"": 274227}" +train real 000000wider_face 20434 20434 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/000000wider_face "{"".jpg"": 20434}" +train real 000000wiki 39229 39229 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/000000wiki "{"".jpg"": 39229}" +train real 000000YouTubeFaces 291102 291102 0 0 /dev/shm/scaledf_bind_storage/ScaleDF/train/000000YouTubeFaces "{"".jpg"": 291102}" diff --git a/ScaleDF_rebuttal/scaledf_embedding_audit.py b/ScaleDF_rebuttal/scaledf_embedding_audit.py new file mode 100644 index 0000000000000000000000000000000000000000..1d8efecfe35571be9ba626c57db69a972ee1b421 --- /dev/null +++ b/ScaleDF_rebuttal/scaledf_embedding_audit.py @@ -0,0 +1,1305 @@ +#!/usr/bin/env python3 +"""Embedding-based overlap audits for ScaleDF rebuttal experiments.""" + +from __future__ import annotations + +import argparse +import csv +import hashlib +import json +import os +import sys +import time +from collections import Counter, defaultdict +from dataclasses import dataclass +from pathlib import Path +from typing import Iterable + +import numpy as np + + +IMAGE_EXTS = {".jpg", ".jpeg", ".png", ".bmp", ".webp"} + + +def sha1_text(value: str) -> str: + return hashlib.sha1(value.encode("utf-8")).hexdigest() + + +def now() -> str: + return time.strftime("%Y-%m-%dT%H:%M:%SZ", time.gmtime()) + + +def write_tsv(path: Path, rows: Iterable[dict[str, object]], fields: list[str] | None = None) -> int: + path.parent.mkdir(parents=True, exist_ok=True) + rows = list(rows) + if fields is None: + fields = list(rows[0].keys()) if rows else [] + with path.open("w", newline="", encoding="utf-8") as f: + writer = csv.DictWriter(f, fieldnames=fields, delimiter="\t") + writer.writeheader() + writer.writerows(rows) + return len(rows) + + +def read_tsv(path: Path) -> list[dict[str, str]]: + with path.open(newline="", encoding="utf-8") as f: + return list(csv.DictReader(f, delimiter="\t")) + + +def is_image(path: str) -> bool: + return Path(path).suffix.lower() in IMAGE_EXTS + + +def infer_binary_label(name: str) -> str: + low = name.lower() + if name.startswith("000000") or low.startswith("000real") or low in {"real", "0_real", "original"} or low.startswith("real"): + return "real" + if "fake" in low or low in {"1_fake", "manipulated"}: + return "fake" + return "fake" + + +def image_walk(root: Path, limit: int | None = None) -> Iterable[Path]: + seen = 0 + stack = [root] + while stack: + current = stack.pop() + try: + entries = sorted(os.scandir(current), key=lambda e: e.name) + except OSError as exc: + print(f"warn: cannot scan {current}: {exc}", file=sys.stderr, flush=True) + continue + for entry in entries: + try: + if entry.is_dir(follow_symlinks=False): + stack.append(Path(entry.path)) + elif entry.is_file(follow_symlinks=False) and is_image(entry.name): + yield Path(entry.path) + seen += 1 + if limit is not None and seen >= limit: + return + except OSError as exc: + print(f"warn: cannot stat {entry.path}: {exc}", file=sys.stderr, flush=True) + + +def top_dirs(root: Path) -> list[Path]: + if not root.exists(): + return [] + return sorted([p for p in root.iterdir() if p.is_dir()], key=lambda p: p.name.lower()) + + +@dataclass +class RootSpec: + role: str + path: Path + + +def parse_root_spec(value: str) -> RootSpec: + if "=" not in value: + raise argparse.ArgumentTypeError("root specs must look like role=/path") + role, path = value.split("=", 1) + role = role.strip() + if role not in {"train", "scaledf_val", "benchmark"}: + raise argparse.ArgumentTypeError("role must be train, scaledf_val, or benchmark") + return RootSpec(role=role, path=Path(path)) + + +def emit_manifest_rows(spec: RootSpec, max_per_group: int | None) -> Iterable[dict[str, str]]: + root = spec.path + if spec.role in {"train", "scaledf_val"}: + split = "train" if spec.role == "train" else "val" + dataset = "ScaleDF_train" if spec.role == "train" else "ScaleDF_val" + for cls_dir in top_dirs(root): + class_name = cls_dir.name + label = infer_binary_label(class_name) + for path in image_walk(cls_dir, max_per_group): + rel = path.relative_to(root).as_posix() + yield { + "image_id": sha1_text(str(path.resolve())), + "split": split, + "dataset": dataset, + "class_name": class_name, + "label": label, + "relpath": rel, + "path": str(path), + } + return + + for dataset_dir in top_dirs(root): + val_dir = dataset_dir / "val" + base = val_dir if val_dir.is_dir() else dataset_dir + class_dirs = top_dirs(base) + if not class_dirs: + class_dirs = [base] + for cls_dir in class_dirs: + class_name = cls_dir.name if cls_dir != base else "" + label = infer_binary_label(class_name or dataset_dir.name) + for path in image_walk(cls_dir, max_per_group): + rel = path.relative_to(root).as_posix() + yield { + "image_id": sha1_text(str(path.resolve())), + "split": "benchmark", + "dataset": dataset_dir.name, + "class_name": class_name, + "label": label, + "relpath": rel, + "path": str(path), + } + + +def command_manifest(args: argparse.Namespace) -> int: + counts = Counter() + fields = ["image_id", "split", "dataset", "class_name", "label", "relpath", "path"] + args.out.parent.mkdir(parents=True, exist_ok=True) + n = 0 + with args.out.open("w", newline="", encoding="utf-8") as f: + writer = csv.DictWriter(f, fieldnames=fields, delimiter="\t") + writer.writeheader() + for spec in args.root: + if not spec.path.exists(): + print(f"warn: missing root {spec.role}={spec.path}", file=sys.stderr) + continue + for row in emit_manifest_rows(spec, args.max_per_group): + key = (row["split"], row["dataset"], row["class_name"], row["label"]) + counts[key] += 1 + writer.writerow(row) + n += 1 + if n % 100000 == 0: + print(f"{now()} manifest_rows={n}", file=sys.stderr, flush=True) + if args.max_total and n >= args.max_total: + break + if args.max_total and n >= args.max_total: + break + summary_rows = [ + { + "split": key[0], + "dataset": key[1], + "class_name": key[2], + "label": key[3], + "images": value, + } + for key, value in sorted(counts.items()) + ] + write_tsv(args.out.with_suffix(".summary.tsv"), summary_rows) + metadata = { + "created_utc": now(), + "rows": n, + "roots": [{"role": spec.role, "path": str(spec.path)} for spec in args.root], + "max_per_group": args.max_per_group, + "max_total": args.max_total, + "image_exts": sorted(IMAGE_EXTS), + } + args.out.with_suffix(".metadata.json").write_text(json.dumps(metadata, indent=2) + "\n", encoding="utf-8") + print(f"rows={n}") + print(args.out) + return 0 + + +def command_split_manifest(args: argparse.Namespace) -> int: + args.out_dir.mkdir(parents=True, exist_ok=True) + selected_splits = set(args.split) if args.split else None + selected_datasets = set(args.dataset) if args.dataset else None + inputs = [args.manifest] + fields: list[str] | None = None + writers = [] + files = [] + counts = [0 for _ in range(args.num_shards)] + total = 0 + try: + for shard in range(args.num_shards): + path = args.out_dir / f"{args.prefix}_shard{shard:02d}_of{args.num_shards:02d}.tsv" + f = path.open("w", newline="", encoding="utf-8") + files.append(f) + writers.append((path, f, None)) + selected_index = 0 + for manifest in inputs: + with manifest.open(newline="", encoding="utf-8") as f: + reader = csv.DictReader(f, delimiter="\t") + if fields is None: + fields = reader.fieldnames or [] + for idx, (path, fh, _) in enumerate(writers): + writer = csv.DictWriter(fh, fieldnames=fields, delimiter="\t") + writer.writeheader() + writers[idx] = (path, fh, writer) + for row in reader: + if selected_splits and row["split"] not in selected_splits: + continue + if selected_datasets and row["dataset"] not in selected_datasets: + continue + shard = selected_index % args.num_shards + writers[shard][2].writerow(row) + counts[shard] += 1 + selected_index += 1 + total += 1 + if total % 100000 == 0: + print(f"{now()} split_rows={total}", flush=True) + finally: + for f in files: + f.close() + + rows = [ + { + "prefix": args.prefix, + "shard": shard, + "num_shards": args.num_shards, + "rows": count, + "path": str(args.out_dir / f"{args.prefix}_shard{shard:02d}_of{args.num_shards:02d}.tsv"), + } + for shard, count in enumerate(counts) + ] + write_tsv(args.out_dir / f"{args.prefix}_split_summary.tsv", rows) + print(f"rows={total}") + print(args.out_dir / f"{args.prefix}_split_summary.tsv") + return 0 + + +def filter_rows(rows: list[dict[str, str]], splits: set[str] | None, datasets: set[str] | None) -> list[dict[str, str]]: + out = [] + for idx, row in enumerate(rows): + row = dict(row) + row["_manifest_index"] = str(idx) + if splits and row["split"] not in splits: + continue + if datasets and row["dataset"] not in datasets: + continue + out.append(row) + return out + + +def shard_rows(rows: list[dict[str, str]], shard_index: int, num_shards: int) -> list[dict[str, str]]: + if num_shards <= 1: + return rows + return [row for i, row in enumerate(rows) if i % num_shards == shard_index] + + +def save_embedding_outputs(prefix: Path, rows: list[dict[str, str]], embeddings: np.ndarray, extra_fields: dict[str, list[object]] | None, metadata: dict[str, object]) -> None: + prefix.parent.mkdir(parents=True, exist_ok=True) + np.save(str(prefix) + ".embeddings.npy", embeddings.astype("float32", copy=False)) + fields = ["row_index", "image_id", "split", "dataset", "class_name", "label", "relpath", "path"] + meta_rows = [] + extra_fields = extra_fields or {} + for out_idx, row in enumerate(rows): + meta = { + "row_index": row.get("_manifest_index", str(out_idx)), + "image_id": row["image_id"], + "split": row["split"], + "dataset": row["dataset"], + "class_name": row["class_name"], + "label": row["label"], + "relpath": row["relpath"], + "path": row["path"], + } + for key, values in extra_fields.items(): + meta[key] = values[out_idx] + meta_rows.append(meta) + if extra_fields: + fields.extend(sorted(extra_fields)) + write_tsv(Path(str(prefix) + ".meta.tsv"), meta_rows, fields) + metadata = dict(metadata) + metadata["created_utc"] = now() + metadata["rows"] = len(rows) + metadata["embedding_shape"] = list(embeddings.shape) + Path(str(prefix) + ".metadata.json").write_text(json.dumps(metadata, indent=2) + "\n", encoding="utf-8") + + +def load_prefix(prefixes: list[Path]) -> tuple[np.ndarray, list[dict[str, str]]]: + emb_parts = [] + meta_rows: list[dict[str, str]] = [] + for prefix in prefixes: + emb_path = Path(str(prefix) + ".embeddings.npy") + meta_path = Path(str(prefix) + ".meta.tsv") + if not emb_path.exists() or not meta_path.exists(): + raise FileNotFoundError(f"missing embedding outputs for {prefix}") + emb = np.load(emb_path, mmap_mode=None).astype("float32", copy=False) + rows = read_tsv(meta_path) + if emb.shape[0] != len(rows): + raise ValueError(f"row mismatch for {prefix}: {emb.shape[0]} embeddings vs {len(rows)} metadata rows") + emb_parts.append(emb) + meta_rows.extend(rows) + if not emb_parts: + raise ValueError("no embedding prefixes provided") + return np.concatenate(emb_parts, axis=0), meta_rows + + +def normalize_matrix(values: np.ndarray) -> np.ndarray: + values = values.astype("float32", copy=False) + norms = np.linalg.norm(values, axis=1, keepdims=True) + return values / np.maximum(norms, 1e-12) + + +META_KEYS = ["image_id", "split", "dataset", "class_name", "label", "path"] + + +def compact_meta(prefix: Path) -> list[tuple[str, str, str, str, str, str]]: + rows = [] + with Path(str(prefix) + ".meta.tsv").open(newline="", encoding="utf-8") as f: + reader = csv.DictReader(f, delimiter="\t") + for row in reader: + rows.append(tuple(row.get(key, "") for key in META_KEYS)) + return rows + + +def compact_get(row: tuple[str, str, str, str, str, str], key: str) -> str: + return row[META_KEYS.index(key)] + + +def command_embed_sscd(args: argparse.Namespace) -> int: + import torch + from PIL import Image + from torch.utils.data import DataLoader, Dataset + from torchvision import transforms + + class ImageDataset(Dataset): + def __init__(self, rows: list[dict[str, str]]): + self.rows = rows + self.transform = transforms.Compose( + [ + transforms.Resize([args.image_size, args.image_size]), + transforms.ToTensor(), + transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]), + ] + ) + + def __len__(self) -> int: + return len(self.rows) + + def __getitem__(self, idx: int): + row = self.rows[idx] + try: + image = Image.open(row["path"]).convert("RGB") + return idx, self.transform(image), "" + except Exception as exc: + return idx, None, f"{type(exc).__name__}: {exc}" + + def collate(batch): + good_idx = [] + tensors = [] + bad = [] + for idx, tensor, error in batch: + if tensor is None: + bad.append((idx, error)) + else: + good_idx.append(idx) + tensors.append(tensor) + if tensors: + return good_idx, torch.stack(tensors, dim=0), bad + return good_idx, None, bad + + rows = read_tsv(args.manifest) + rows = filter_rows(rows, set(args.split) if args.split else None, set(args.dataset) if args.dataset else None) + rows = shard_rows(rows, args.shard_index, args.num_shards) + if args.max_images: + rows = rows[: args.max_images] + if not rows: + raise ValueError("no manifest rows selected") + if Path(str(args.out_prefix) + ".embeddings.npy").exists() and not args.overwrite: + print(f"exists={args.out_prefix}") + return 0 + + device = torch.device(args.device) + model = torch.jit.load(str(args.model), map_location=device) + model.eval() + dataset = ImageDataset(rows) + loader = DataLoader( + dataset, + batch_size=args.batch_size, + shuffle=False, + num_workers=args.workers, + pin_memory=device.type == "cuda", + collate_fn=collate, + prefetch_factor=args.prefetch_factor if args.workers > 0 else None, + persistent_workers=args.workers > 0, + ) + vectors = [] + kept_rows = [] + bad_rows = [] + start = time.time() + with torch.inference_mode(): + for step, (batch_indices, batch, bad) in enumerate(loader, start=1): + for idx, error in bad: + bad_rows.append({"path": rows[idx]["path"], "error": error}) + if batch is None: + continue + batch = batch.to(device, non_blocking=True) + emb = model(batch) + emb = torch.nn.functional.normalize(emb.float(), dim=1) + vectors.append(emb.cpu().numpy()) + kept_rows.extend(rows[idx] for idx in batch_indices) + if step % args.status_every == 0: + done = min(step * args.batch_size, len(rows)) + rate = done / max(time.time() - start, 1e-6) + print(f"{now()} processed={done}/{len(rows)} rate={rate:.1f}/s", flush=True) + + embeddings = np.concatenate(vectors, axis=0) if vectors else np.zeros((0, 512), dtype="float32") + save_embedding_outputs( + args.out_prefix, + kept_rows, + embeddings, + None, + { + "kind": "sscd", + "model": str(args.model), + "image_size": args.image_size, + "source_manifest": str(args.manifest), + "splits": args.split, + "datasets": args.dataset, + "shard_index": args.shard_index, + "num_shards": args.num_shards, + "max_images": args.max_images, + }, + ) + if bad_rows: + write_tsv(Path(str(args.out_prefix) + ".bad_images.tsv"), bad_rows, ["path", "error"]) + print(f"rows={len(kept_rows)}") + print(args.out_prefix) + return 0 + + +def command_embed_arcface(args: argparse.Namespace) -> int: + import cv2 + import onnxruntime as ort + from insightface.app import FaceAnalysis + + rows = read_tsv(args.manifest) + rows = filter_rows(rows, set(args.split) if args.split else None, set(args.dataset) if args.dataset else None) + rows = shard_rows(rows, args.shard_index, args.num_shards) + if args.max_images: + rows = rows[: args.max_images] + if not rows: + raise ValueError("no manifest rows selected") + if Path(str(args.out_prefix) + ".embeddings.npy").exists() and not args.overwrite: + print(f"exists={args.out_prefix}") + return 0 + + providers = ["CPUExecutionProvider"] + ctx_id = -1 + if args.device.startswith("cuda"): + providers = ["CUDAExecutionProvider", "CPUExecutionProvider"] + if ":" in args.device: + ctx_id = int(args.device.split(":", 1)[1]) + else: + ctx_id = 0 + print(f"onnxruntime_providers={ort.get_available_providers()}", flush=True) + app = FaceAnalysis(name=args.model_pack, root=str(args.model_root), providers=providers) + app.prepare(ctx_id=ctx_id, det_size=(args.det_size, args.det_size)) + + kept_rows = [] + vectors = [] + det_scores: list[object] = [] + bboxes: list[object] = [] + bad_rows = [] + start = time.time() + for idx, row in enumerate(rows, start=1): + image = cv2.imread(row["path"]) + if image is None: + bad_rows.append({"path": row["path"], "error": "cv2.imread returned None"}) + continue + try: + faces = app.get(image) + except Exception as exc: + bad_rows.append({"path": row["path"], "error": f"{type(exc).__name__}: {exc}"}) + continue + if not faces: + bad_rows.append({"path": row["path"], "error": "no_face_detected"}) + continue + face = max(faces, key=lambda f: max(float(f.bbox[2] - f.bbox[0]), 0.0) * max(float(f.bbox[3] - f.bbox[1]), 0.0)) + emb = np.asarray(face.normed_embedding, dtype="float32") + norm = float(np.linalg.norm(emb)) + if not np.isfinite(norm) or norm <= 0: + bad_rows.append({"path": row["path"], "error": "bad_embedding"}) + continue + vectors.append(emb / norm) + kept_rows.append(row) + det_scores.append(float(getattr(face, "det_score", 0.0))) + bboxes.append(",".join(f"{float(v):.2f}" for v in face.bbox)) + if idx % args.status_every == 0: + rate = idx / max(time.time() - start, 1e-6) + print(f"{now()} processed={idx}/{len(rows)} kept={len(kept_rows)} rate={rate:.1f}/s", flush=True) + + dim = len(vectors[0]) if vectors else 512 + embeddings = np.stack(vectors, axis=0).astype("float32") if vectors else np.zeros((0, dim), dtype="float32") + save_embedding_outputs( + args.out_prefix, + kept_rows, + embeddings, + {"det_score": det_scores, "bbox_xyxy": bboxes}, + { + "kind": "arcface", + "model_pack": args.model_pack, + "model_root": str(args.model_root), + "det_size": args.det_size, + "source_manifest": str(args.manifest), + "splits": args.split, + "datasets": args.dataset, + "shard_index": args.shard_index, + "num_shards": args.num_shards, + "max_images": args.max_images, + }, + ) + if bad_rows: + write_tsv(Path(str(args.out_prefix) + ".bad_images.tsv"), bad_rows, ["path", "error"]) + print(f"rows={len(kept_rows)} bad={len(bad_rows)}") + print(args.out_prefix) + return 0 + + +def command_embed_arcface_direct(args: argparse.Namespace) -> int: + import cv2 + import onnxruntime as ort + import torch + from torch.utils.data import DataLoader, Dataset + + rows = read_tsv(args.manifest) + rows = filter_rows(rows, set(args.split) if args.split else None, set(args.dataset) if args.dataset else None) + rows = shard_rows(rows, args.shard_index, args.num_shards) + if args.max_images: + rows = rows[: args.max_images] + if not rows: + raise ValueError("no manifest rows selected") + if Path(str(args.out_prefix) + ".embeddings.npy").exists() and not args.overwrite: + print(f"exists={args.out_prefix}") + return 0 + + providers = ["CPUExecutionProvider"] + if args.device.startswith("cuda"): + providers = ["CUDAExecutionProvider", "CPUExecutionProvider"] + session_options = ort.SessionOptions() + session_options.log_severity_level = 3 + session_options.intra_op_num_threads = args.ort_threads + session_options.inter_op_num_threads = 1 + try: + session = ort.InferenceSession(str(args.model), sess_options=session_options, providers=providers) + except Exception as exc: + if not args.allow_cpu_fallback: + raise + print(f"warn: CUDA ArcFace session failed ({type(exc).__name__}: {exc}); falling back to CPU", file=sys.stderr) + session = ort.InferenceSession(str(args.model), sess_options=session_options, providers=["CPUExecutionProvider"]) + input_cfg = session.get_inputs()[0] + input_name = input_cfg.name + output_names = [out.name for out in session.get_outputs()] + input_shape = input_cfg.shape + if len(input_shape) >= 4 and isinstance(input_shape[2], int) and isinstance(input_shape[3], int): + input_size = (int(input_shape[3]), int(input_shape[2])) + else: + input_size = (args.image_size, args.image_size) + print(f"onnxruntime_available={ort.get_available_providers()}", flush=True) + print(f"onnxruntime_applied={session.get_providers()}", flush=True) + print(f"arcface_input={input_name} shape={input_shape} size={input_size}", flush=True) + + class CroppedFaceDataset(Dataset): + def __init__(self, items: list[dict[str, str]]): + self.items = items + + def __len__(self) -> int: + return len(self.items) + + def __getitem__(self, idx: int): + row = self.items[idx] + img = cv2.imread(row["path"]) + if img is None: + return idx, None, "", "cv2.imread returned None" + shape = f"{img.shape[0]}x{img.shape[1]}" + if (img.shape[1], img.shape[0]) != input_size: + img = cv2.resize(img, input_size, interpolation=cv2.INTER_LINEAR) + img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB) + blob = (img.astype("float32") - args.input_mean) / args.input_std + blob = np.transpose(blob, (2, 0, 1)) + return idx, blob, shape, "" + + def collate(batch): + indices = [] + tensors = [] + local_shapes = [] + bad = [] + for idx, tensor, shape, error in batch: + if tensor is None: + bad.append((idx, error)) + else: + indices.append(idx) + tensors.append(tensor) + local_shapes.append(shape) + if tensors: + return indices, torch.as_tensor(np.stack(tensors, axis=0), dtype=torch.float32), local_shapes, bad + return indices, None, local_shapes, bad + + loader = DataLoader( + CroppedFaceDataset(rows), + batch_size=args.batch_size, + shuffle=False, + num_workers=args.workers, + pin_memory=False, + collate_fn=collate, + prefetch_factor=args.prefetch_factor if args.workers > 0 else None, + persistent_workers=args.workers > 0, + ) + kept_rows = [] + vectors = [] + shapes: list[object] = [] + bad_rows = [] + start = time.time() + processed = 0 + for step, (batch_indices, batch, local_shapes, bad) in enumerate(loader, start=1): + for idx, error in bad: + bad_rows.append({"path": rows[idx]["path"], "error": error}) + processed += len(batch_indices) + len(bad) + if batch is None: + continue + blob = np.ascontiguousarray(batch.numpy()) + net_out = session.run(output_names, {input_name: blob})[0].astype("float32", copy=False) + norms = np.linalg.norm(net_out, axis=1, keepdims=True) + net_out = net_out / np.maximum(norms, 1e-12) + vectors.append(net_out) + kept_rows.extend(rows[idx] for idx in batch_indices) + shapes.extend(local_shapes) + if step % args.status_every == 0 or processed == len(rows): + rate = processed / max(time.time() - start, 1e-6) + print(f"{now()} processed={processed}/{len(rows)} kept={len(kept_rows)} rate={rate:.1f}/s", flush=True) + + dim = vectors[0].shape[1] if vectors else 512 + embeddings = np.concatenate(vectors, axis=0).astype("float32", copy=False) if vectors else np.zeros((0, dim), dtype="float32") + save_embedding_outputs( + args.out_prefix, + kept_rows, + embeddings, + {"orig_hw": shapes}, + { + "kind": "arcface_direct", + "model": str(args.model), + "input_size": list(input_size), + "input_mean": args.input_mean, + "input_std": args.input_std, + "workers": args.workers, + "prefetch_factor": args.prefetch_factor, + "onnxruntime_available_providers": ort.get_available_providers(), + "onnxruntime_applied_providers": session.get_providers(), + "source_manifest": str(args.manifest), + "splits": args.split, + "datasets": args.dataset, + "shard_index": args.shard_index, + "num_shards": args.num_shards, + "max_images": args.max_images, + }, + ) + if bad_rows: + write_tsv(Path(str(args.out_prefix) + ".bad_images.tsv"), bad_rows, ["path", "error"]) + print(f"rows={len(kept_rows)} bad={len(bad_rows)}") + print(args.out_prefix) + return 0 + + +def topk_torch(gallery: np.ndarray, query: np.ndarray, k: int, device: str, query_batch: int) -> tuple[np.ndarray, np.ndarray]: + import torch + + dev = torch.device(device) + torch.backends.cuda.matmul.allow_tf32 = True + torch.backends.cudnn.allow_tf32 = True + gallery_t = torch.from_numpy(gallery.astype("float32", copy=False)).to(dev) + values = [] + indices = [] + start_time = time.time() + with torch.inference_mode(): + for step, start in enumerate(range(0, query.shape[0], query_batch), start=1): + end = min(start + query_batch, query.shape[0]) + batch = torch.from_numpy(query[start:end].astype("float32", copy=False)).to(dev) + scores = batch @ gallery_t.T + val, idx = torch.topk(scores, k=min(k, gallery.shape[0]), dim=1) + values.append(val.cpu().numpy()) + indices.append(idx.cpu().numpy()) + if step % 10 == 0 or end == query.shape[0]: + elapsed = max(time.time() - start_time, 1e-6) + print(f"{now()} topk_torch device={device} searched={end}/{query.shape[0]} rate={end/elapsed:.1f}/s", flush=True) + return np.concatenate(values, axis=0), np.concatenate(indices, axis=0) + + +def topk_faiss(gallery: np.ndarray, query: np.ndarray, k: int) -> tuple[np.ndarray, np.ndarray]: + import faiss + + index = faiss.IndexFlatIP(gallery.shape[1]) + index.add(np.ascontiguousarray(gallery.astype("float32", copy=False))) + return index.search(np.ascontiguousarray(query.astype("float32", copy=False)), min(k, gallery.shape[0])) + + +def quantile(values: list[float], q: float) -> float: + if not values: + return float("nan") + return float(np.quantile(np.asarray(values, dtype="float64"), q)) + + +def command_match(args: argparse.Namespace) -> int: + gallery, gallery_rows = load_prefix(args.gallery_prefix) + query, query_rows = load_prefix(args.query_prefix) + if gallery.shape[1] != query.shape[1]: + raise ValueError(f"dimension mismatch: gallery {gallery.shape} query {query.shape}") + gallery = normalize_matrix(gallery) + query = normalize_matrix(query) + k_search = args.topk + 1 if args.exclude_same_image else args.topk + if args.engine == "faiss": + values, indices = topk_faiss(gallery, query, k_search) + elif args.engine == "torch": + values, indices = topk_torch(gallery, query, k_search, args.device, args.query_batch) + else: + try: + values, indices = topk_faiss(gallery, query, k_search) + except Exception as exc: + print(f"warn: faiss unavailable ({type(exc).__name__}: {exc}); falling back to torch", file=sys.stderr) + values, indices = topk_torch(gallery, query, k_search, args.device, args.query_batch) + + match_rows = [] + summary_scores: dict[tuple[str, str, str], list[float]] = defaultdict(list) + summary_l2sq: dict[tuple[str, str, str], list[float]] = defaultdict(list) + threshold_hits = Counter() + l2_hits = Counter() + for q_idx, q_row in enumerate(query_rows): + candidates = [] + for rank, (score, g_idx) in enumerate(zip(values[q_idx], indices[q_idx]), start=1): + if g_idx < 0: + continue + g_row = gallery_rows[int(g_idx)] + if args.exclude_same_image and q_row.get("image_id") == g_row.get("image_id"): + continue + candidates.append((rank, float(score), int(g_idx), g_row)) + if len(candidates) >= args.topk: + break + if not candidates: + continue + rank, score, g_idx, g_row = candidates[0] + l2sq = max(0.0, 2.0 - 2.0 * score) + key = (q_row["split"], q_row["dataset"], q_row.get("label", "")) + summary_scores[key].append(score) + summary_l2sq[key].append(l2sq) + if score > args.score_threshold: + threshold_hits[key] += 1 + if l2sq < args.l2sq_threshold: + l2_hits[key] += 1 + match_rows.append( + { + "query_split": q_row["split"], + "query_dataset": q_row["dataset"], + "query_label": q_row.get("label", ""), + "query_class": q_row.get("class_name", ""), + "query_path": q_row["path"], + "gallery_split": g_row["split"], + "gallery_dataset": g_row["dataset"], + "gallery_label": g_row.get("label", ""), + "gallery_class": g_row.get("class_name", ""), + "gallery_path": g_row["path"], + "rank": rank, + "score_cosine": f"{score:.8f}", + "score_gt_threshold": int(score > args.score_threshold), + "l2sq": f"{l2sq:.8f}", + "l2sq_lt_threshold": int(l2sq < args.l2sq_threshold), + } + ) + + fields = [ + "query_split", + "query_dataset", + "query_label", + "query_class", + "query_path", + "gallery_split", + "gallery_dataset", + "gallery_label", + "gallery_class", + "gallery_path", + "rank", + "score_cosine", + "score_gt_threshold", + "l2sq", + "l2sq_lt_threshold", + ] + write_tsv(args.out, match_rows, fields) + summary_rows = [] + for key in sorted(summary_scores): + scores = summary_scores[key] + l2s = summary_l2sq[key] + count = len(scores) + summary_rows.append( + { + "query_split": key[0], + "query_dataset": key[1], + "query_label": key[2], + "queries": count, + "max_score": f"{max(scores):.8f}", + "p99_score": f"{quantile(scores, 0.99):.8f}", + "p95_score": f"{quantile(scores, 0.95):.8f}", + "median_score": f"{quantile(scores, 0.50):.8f}", + "score_threshold": args.score_threshold, + "score_gt_threshold": threshold_hits[key], + "score_gt_threshold_fraction": f"{threshold_hits[key] / count:.8f}", + "min_l2sq": f"{min(l2s):.8f}", + "p01_l2sq": f"{quantile(l2s, 0.01):.8f}", + "l2sq_threshold": args.l2sq_threshold, + "l2sq_lt_threshold": l2_hits[key], + "l2sq_lt_threshold_fraction": f"{l2_hits[key] / count:.8f}", + } + ) + write_tsv(args.out.with_suffix(".summary.tsv"), summary_rows) + args.out.with_suffix(".metadata.json").write_text( + json.dumps( + { + "created_utc": now(), + "gallery_prefixes": [str(p) for p in args.gallery_prefix], + "query_prefixes": [str(p) for p in args.query_prefix], + "gallery_rows": len(gallery_rows), + "query_rows": len(query_rows), + "topk": args.topk, + "score_threshold": args.score_threshold, + "l2sq_threshold": args.l2sq_threshold, + "engine": args.engine, + }, + indent=2, + ) + + "\n", + encoding="utf-8", + ) + print(f"matches={len(match_rows)}") + print(args.out) + return 0 + + +def sample_embeddings(prefixes: list[Path], sample_size: int, seed: int) -> np.ndarray: + rng = np.random.default_rng(seed) + samples = [] + per_prefix = max(1, sample_size // max(len(prefixes), 1)) + for prefix in prefixes: + emb = np.load(str(prefix) + ".embeddings.npy", mmap_mode="r") + take = min(per_prefix, emb.shape[0]) + if take <= 0: + continue + idx = rng.choice(emb.shape[0], size=take, replace=False) + idx.sort() + samples.append(np.asarray(emb[idx], dtype="float32")) + if not samples: + raise ValueError("no embeddings available for index training") + sample = np.concatenate(samples, axis=0) + if sample.shape[0] > sample_size: + idx = rng.choice(sample.shape[0], size=sample_size, replace=False) + sample = sample[idx] + return normalize_matrix(sample) + + +def command_match_index(args: argparse.Namespace) -> int: + import faiss + + if args.omp_threads: + faiss.omp_set_num_threads(args.omp_threads) + first = np.load(str(args.gallery_prefix[0]) + ".embeddings.npy", mmap_mode="r") + dim = int(first.shape[1]) + del first + index = faiss.index_factory(dim, args.index_factory, faiss.METRIC_INNER_PRODUCT) + if not index.is_trained: + print(f"{now()} train_index factory={args.index_factory} sample={args.train_sample}", flush=True) + index.train(sample_embeddings(args.gallery_prefix, args.train_sample, args.seed)) + if hasattr(index, "hnsw"): + index.hnsw.efSearch = args.hnsw_ef_search + index.hnsw.efConstruction = args.hnsw_ef_construction + if hasattr(index, "nprobe"): + index.nprobe = args.nprobe + + gallery_meta: list[tuple[str, str, str, str, str, str]] = [] + start = time.time() + for prefix in args.gallery_prefix: + emb_path = Path(str(prefix) + ".embeddings.npy") + emb = np.load(emb_path, mmap_mode="r") + meta = compact_meta(prefix) + if emb.shape[0] != len(meta): + raise ValueError(f"row mismatch for {prefix}: {emb.shape[0]} vs {len(meta)}") + for offset in range(0, emb.shape[0], args.add_batch): + batch = normalize_matrix(np.asarray(emb[offset : offset + args.add_batch], dtype="float32")) + index.add(batch) + rate = index.ntotal / max(time.time() - start, 1e-6) + print( + f"{now()} adding latest={prefix.name} " + f"prefix_rows={min(offset + args.add_batch, emb.shape[0])}/{emb.shape[0]} " + f"added_total={index.ntotal} rate={rate:.1f}/s", + flush=True, + ) + gallery_meta.extend(meta) + rate = index.ntotal / max(time.time() - start, 1e-6) + print(f"{now()} added={index.ntotal} latest={prefix.name} rate={rate:.1f}/s", flush=True) + + args.out.parent.mkdir(parents=True, exist_ok=True) + fields = [ + "query_split", + "query_dataset", + "query_label", + "query_class", + "query_path", + "gallery_split", + "gallery_dataset", + "gallery_label", + "gallery_class", + "gallery_path", + "rank", + "score_cosine", + "score_gt_threshold", + "l2sq", + "l2sq_lt_threshold", + ] + summary_scores: dict[tuple[str, str, str], list[float]] = defaultdict(list) + summary_l2sq: dict[tuple[str, str, str], list[float]] = defaultdict(list) + threshold_hits = Counter() + l2_hits = Counter() + query_total = 0 + written = 0 + with args.out.open("w", newline="", encoding="utf-8") as f: + writer = csv.DictWriter(f, fieldnames=fields, delimiter="\t") + writer.writeheader() + for prefix in args.query_prefix: + query = np.load(str(prefix) + ".embeddings.npy", mmap_mode="r") + query_meta = compact_meta(prefix) + if query.shape[0] != len(query_meta): + raise ValueError(f"row mismatch for {prefix}: {query.shape[0]} vs {len(query_meta)}") + for offset in range(0, query.shape[0], args.query_batch): + q_batch = normalize_matrix(np.asarray(query[offset : offset + args.query_batch], dtype="float32")) + values, indices = index.search(q_batch, args.topk) + for local_idx, (score_row, index_row) in enumerate(zip(values, indices)): + q_row = query_meta[offset + local_idx] + for rank, (score, g_idx) in enumerate(zip(score_row, index_row), start=1): + if g_idx < 0: + continue + g_row = gallery_meta[int(g_idx)] + score = float(score) + l2sq = max(0.0, 2.0 - 2.0 * score) + key = ( + compact_get(q_row, "split"), + compact_get(q_row, "dataset"), + compact_get(q_row, "label"), + ) + if rank == 1: + summary_scores[key].append(score) + summary_l2sq[key].append(l2sq) + if score > args.score_threshold: + threshold_hits[key] += 1 + if l2sq < args.l2sq_threshold: + l2_hits[key] += 1 + if args.write_all or score > args.score_threshold or l2sq < args.l2sq_threshold or rank == 1: + writer.writerow( + { + "query_split": compact_get(q_row, "split"), + "query_dataset": compact_get(q_row, "dataset"), + "query_label": compact_get(q_row, "label"), + "query_class": compact_get(q_row, "class_name"), + "query_path": compact_get(q_row, "path"), + "gallery_split": compact_get(g_row, "split"), + "gallery_dataset": compact_get(g_row, "dataset"), + "gallery_label": compact_get(g_row, "label"), + "gallery_class": compact_get(g_row, "class_name"), + "gallery_path": compact_get(g_row, "path"), + "rank": rank, + "score_cosine": f"{score:.8f}", + "score_gt_threshold": int(score > args.score_threshold), + "l2sq": f"{l2sq:.8f}", + "l2sq_lt_threshold": int(l2sq < args.l2sq_threshold), + } + ) + written += 1 + query_total += q_batch.shape[0] + if query_total % (args.query_batch * args.status_every) == 0: + elapsed = max(time.time() - start, 1e-6) + print(f"{now()} searched={query_total} written={written} rate={query_total/elapsed:.1f}/s", flush=True) + + summary_rows = [] + for key in sorted(summary_scores): + scores = summary_scores[key] + l2s = summary_l2sq[key] + count = len(scores) + summary_rows.append( + { + "query_split": key[0], + "query_dataset": key[1], + "query_label": key[2], + "queries": count, + "max_score": f"{max(scores):.8f}", + "p99_score": f"{quantile(scores, 0.99):.8f}", + "p95_score": f"{quantile(scores, 0.95):.8f}", + "median_score": f"{quantile(scores, 0.50):.8f}", + "score_threshold": args.score_threshold, + "score_gt_threshold": threshold_hits[key], + "score_gt_threshold_fraction": f"{threshold_hits[key] / count:.8f}", + "min_l2sq": f"{min(l2s):.8f}", + "p01_l2sq": f"{quantile(l2s, 0.01):.8f}", + "l2sq_threshold": args.l2sq_threshold, + "l2sq_lt_threshold": l2_hits[key], + "l2sq_lt_threshold_fraction": f"{l2_hits[key] / count:.8f}", + } + ) + write_tsv(args.out.with_suffix(".summary.tsv"), summary_rows) + args.out.with_suffix(".metadata.json").write_text( + json.dumps( + { + "created_utc": now(), + "gallery_prefixes": [str(p) for p in args.gallery_prefix], + "query_prefixes": [str(p) for p in args.query_prefix], + "gallery_rows": len(gallery_meta), + "query_rows": query_total, + "written_rows": written, + "topk": args.topk, + "score_threshold": args.score_threshold, + "l2sq_threshold": args.l2sq_threshold, + "index_factory": args.index_factory, + "hnsw_ef_search": args.hnsw_ef_search, + "hnsw_ef_construction": args.hnsw_ef_construction, + "nprobe": args.nprobe, + "faiss_version": getattr(faiss, "__version__", "unknown"), + "omp_threads": args.omp_threads, + }, + indent=2, + ) + + "\n", + encoding="utf-8", + ) + print(f"queries={query_total} written={written}") + print(args.out) + return 0 + + +def command_summarize_matches(args: argparse.Namespace) -> int: + summary_scores: dict[tuple[str, str, str], list[float]] = defaultdict(list) + summary_l2sq: dict[tuple[str, str, str], list[float]] = defaultdict(list) + threshold_hits = Counter() + l2_hits = Counter() + total_rows = 0 + for path in args.match_tsv: + with path.open(newline="", encoding="utf-8") as f: + reader = csv.DictReader(f, delimiter="\t") + for row in reader: + if row.get("rank", "1") != "1": + continue + key = (row["query_split"], row["query_dataset"], row.get("query_label", "")) + score = float(row["score_cosine"]) + l2sq = float(row["l2sq"]) + summary_scores[key].append(score) + summary_l2sq[key].append(l2sq) + if score > args.score_threshold: + threshold_hits[key] += 1 + if l2sq < args.l2sq_threshold: + l2_hits[key] += 1 + total_rows += 1 + + summary_rows = [] + for key in sorted(summary_scores): + scores = summary_scores[key] + l2s = summary_l2sq[key] + count = len(scores) + summary_rows.append( + { + "query_split": key[0], + "query_dataset": key[1], + "query_label": key[2], + "queries": count, + "max_score": f"{max(scores):.8f}", + "p99_score": f"{quantile(scores, 0.99):.8f}", + "p95_score": f"{quantile(scores, 0.95):.8f}", + "median_score": f"{quantile(scores, 0.50):.8f}", + "score_threshold": args.score_threshold, + "score_gt_threshold": threshold_hits[key], + "score_gt_threshold_fraction": f"{threshold_hits[key] / count:.8f}", + "min_l2sq": f"{min(l2s):.8f}", + "p01_l2sq": f"{quantile(l2s, 0.01):.8f}", + "l2sq_threshold": args.l2sq_threshold, + "l2sq_lt_threshold": l2_hits[key], + "l2sq_lt_threshold_fraction": f"{l2_hits[key] / count:.8f}", + } + ) + write_tsv(args.out, summary_rows) + args.out.with_suffix(".metadata.json").write_text( + json.dumps( + { + "created_utc": now(), + "match_tsv": [str(p) for p in args.match_tsv], + "top1_rows": total_rows, + "score_threshold": args.score_threshold, + "l2sq_threshold": args.l2sq_threshold, + }, + indent=2, + ) + + "\n", + encoding="utf-8", + ) + print(f"top1_rows={total_rows}") + print(args.out) + return 0 + + +def command_merge_matches(args: argparse.Namespace) -> int: + best_rows: dict[str, dict[str, str]] = {} + best_scores: dict[str, float] = {} + fields: list[str] | None = None + read_rows = 0 + for path in args.match_tsv: + with path.open(newline="", encoding="utf-8") as f: + reader = csv.DictReader(f, delimiter="\t") + if fields is None: + fields = list(reader.fieldnames or []) + for row in reader: + if row.get("rank", "1") != "1": + continue + key = row["query_path"] + score = float(row["score_cosine"]) + if key not in best_scores or score > best_scores[key]: + best_scores[key] = score + best_rows[key] = row + read_rows += 1 + if read_rows % args.status_every == 0: + print(f"{now()} merge read_rows={read_rows} unique_queries={len(best_rows)}", flush=True) + + if fields is None: + fields = [] + args.out.parent.mkdir(parents=True, exist_ok=True) + with args.out.open("w", newline="", encoding="utf-8") as f: + writer = csv.DictWriter(f, fieldnames=fields, delimiter="\t") + writer.writeheader() + for key in sorted(best_rows): + writer.writerow(best_rows[key]) + args.out.with_suffix(".metadata.json").write_text( + json.dumps( + { + "created_utc": now(), + "match_tsv": [str(p) for p in args.match_tsv], + "candidate_rows": read_rows, + "merged_rows": len(best_rows), + }, + indent=2, + ) + + "\n", + encoding="utf-8", + ) + print(f"candidate_rows={read_rows} merged_rows={len(best_rows)}") + print(args.out) + return 0 + + +def build_parser() -> argparse.ArgumentParser: + parser = argparse.ArgumentParser(description=__doc__) + sub = parser.add_subparsers(dest="command", required=True) + + p = sub.add_parser("manifest") + p.add_argument("--root", type=parse_root_spec, action="append", required=True) + p.add_argument("--out", type=Path, required=True) + p.add_argument("--max-per-group", type=int) + p.add_argument("--max-total", type=int) + p.set_defaults(func=command_manifest) + + p = sub.add_parser("split-manifest") + p.add_argument("--manifest", type=Path, required=True) + p.add_argument("--out-dir", type=Path, required=True) + p.add_argument("--prefix", required=True) + p.add_argument("--num-shards", type=int, required=True) + p.add_argument("--split", action="append") + p.add_argument("--dataset", action="append") + p.set_defaults(func=command_split_manifest) + + p = sub.add_parser("embed-sscd") + p.add_argument("--manifest", type=Path, required=True) + p.add_argument("--model", type=Path, required=True) + p.add_argument("--out-prefix", type=Path, required=True) + p.add_argument("--split", action="append") + p.add_argument("--dataset", action="append") + p.add_argument("--device", default="cuda:0") + p.add_argument("--batch-size", type=int, default=256) + p.add_argument("--workers", type=int, default=8) + p.add_argument("--prefetch-factor", type=int, default=4) + p.add_argument("--image-size", type=int, default=320) + p.add_argument("--shard-index", type=int, default=0) + p.add_argument("--num-shards", type=int, default=1) + p.add_argument("--max-images", type=int) + p.add_argument("--status-every", type=int, default=20) + p.add_argument("--overwrite", action="store_true") + p.set_defaults(func=command_embed_sscd) + + p = sub.add_parser("embed-arcface") + p.add_argument("--manifest", type=Path, required=True) + p.add_argument("--out-prefix", type=Path, required=True) + p.add_argument("--split", action="append") + p.add_argument("--dataset", action="append") + p.add_argument("--device", default="cuda:0") + p.add_argument("--model-pack", default="buffalo_l") + p.add_argument("--model-root", type=Path, default=Path("/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/models/insightface")) + p.add_argument("--det-size", type=int, default=640) + p.add_argument("--shard-index", type=int, default=0) + p.add_argument("--num-shards", type=int, default=1) + p.add_argument("--max-images", type=int) + p.add_argument("--status-every", type=int, default=200) + p.add_argument("--overwrite", action="store_true") + p.set_defaults(func=command_embed_arcface) + + p = sub.add_parser("embed-arcface-direct") + p.add_argument("--manifest", type=Path, required=True) + p.add_argument("--model", type=Path, required=True) + p.add_argument("--out-prefix", type=Path, required=True) + p.add_argument("--split", action="append") + p.add_argument("--dataset", action="append") + p.add_argument("--device", default="cuda:0") + p.add_argument("--batch-size", type=int, default=512) + p.add_argument("--workers", type=int, default=8) + p.add_argument("--prefetch-factor", type=int, default=4) + p.add_argument("--image-size", type=int, default=112) + p.add_argument("--input-mean", type=float, default=127.5) + p.add_argument("--input-std", type=float, default=127.5) + p.add_argument("--ort-threads", type=int, default=4) + p.add_argument("--shard-index", type=int, default=0) + p.add_argument("--num-shards", type=int, default=1) + p.add_argument("--max-images", type=int) + p.add_argument("--status-every", type=int, default=10) + p.add_argument("--allow-cpu-fallback", action="store_true") + p.add_argument("--overwrite", action="store_true") + p.set_defaults(func=command_embed_arcface_direct) + + p = sub.add_parser("match") + p.add_argument("--gallery-prefix", type=Path, action="append", required=True) + p.add_argument("--query-prefix", type=Path, action="append", required=True) + p.add_argument("--out", type=Path, required=True) + p.add_argument("--topk", type=int, default=1) + p.add_argument("--score-threshold", type=float, default=0.75) + p.add_argument("--l2sq-threshold", type=float, default=0.5) + p.add_argument("--engine", choices=["auto", "faiss", "torch"], default="auto") + p.add_argument("--device", default="cuda:0") + p.add_argument("--query-batch", type=int, default=64) + p.add_argument("--exclude-same-image", action="store_true") + p.set_defaults(func=command_match) + + p = sub.add_parser("match-index") + p.add_argument("--gallery-prefix", type=Path, action="append", required=True) + p.add_argument("--query-prefix", type=Path, action="append", required=True) + p.add_argument("--out", type=Path, required=True) + p.add_argument("--topk", type=int, default=1) + p.add_argument("--score-threshold", type=float, default=0.75) + p.add_argument("--l2sq-threshold", type=float, default=0.5) + p.add_argument("--index-factory", default="HNSW32") + p.add_argument("--hnsw-ef-search", type=int, default=256) + p.add_argument("--hnsw-ef-construction", type=int, default=200) + p.add_argument("--nprobe", type=int, default=64) + p.add_argument("--train-sample", type=int, default=500000) + p.add_argument("--seed", type=int, default=123) + p.add_argument("--add-batch", type=int, default=200000) + p.add_argument("--query-batch", type=int, default=4096) + p.add_argument("--status-every", type=int, default=20) + p.add_argument("--write-all", action="store_true") + p.add_argument("--omp-threads", type=int, default=96) + p.set_defaults(func=command_match_index) + + p = sub.add_parser("summarize-matches") + p.add_argument("--match-tsv", type=Path, action="append", required=True) + p.add_argument("--out", type=Path, required=True) + p.add_argument("--score-threshold", type=float, default=0.75) + p.add_argument("--l2sq-threshold", type=float, default=0.5) + p.set_defaults(func=command_summarize_matches) + + p = sub.add_parser("merge-matches") + p.add_argument("--match-tsv", type=Path, action="append", required=True) + p.add_argument("--out", type=Path, required=True) + p.add_argument("--status-every", type=int, default=1000000) + p.set_defaults(func=command_merge_matches) + return parser + + +def main() -> int: + parser = build_parser() + args = parser.parse_args() + return args.func(args) + + +if __name__ == "__main__": + raise SystemExit(main()) diff --git a/ScaleDF_rebuttal/scaledf_overlap_audit.py b/ScaleDF_rebuttal/scaledf_overlap_audit.py new file mode 100644 index 0000000000000000000000000000000000000000..83e094c422753d4a22c0075e327e5c19d29c0bb0 --- /dev/null +++ b/ScaleDF_rebuttal/scaledf_overlap_audit.py @@ -0,0 +1,150 @@ +#!/usr/bin/env python3 +"""Lightweight path/class-name overlap audit for ScaleDF rebuttal.""" + +from __future__ import annotations + +import argparse +import csv +import pickle +import re +from pathlib import Path + + +def parse_args() -> argparse.Namespace: + parser = argparse.ArgumentParser() + parser.add_argument("--data-root", type=Path, default=Path("/dev/shm/scaledf_bind_storage")) + parser.add_argument("--repro-root", type=Path, default=Path("/tmp/ScaleDF_repro_20260724")) + parser.add_argument("--out-dir", type=Path, required=True) + return parser.parse_args() + + +def norm_name(value: str) -> str: + return re.sub(r"[^a-z0-9]+", "", value.lower()) + + +def load_pkl_list(path: Path) -> set[str]: + if not path.exists(): + return set() + with path.open("rb") as f: + data = pickle.load(f) + if isinstance(data, list): + return {str(x) for x in data} + if isinstance(data, dict): + return {str(x) for x in data} + return set() + + +def top_dirs(path: Path) -> set[str]: + if not path.exists(): + return set() + return {p.name for p in path.iterdir() if p.is_dir()} + + +def compare_sets(left_name: str, left: set[str], right_name: str, right: set[str]) -> tuple[dict[str, object], list[dict[str, str]]]: + exact = left & right + left_norm = {} + right_norm = {} + for item in left: + left_norm.setdefault(norm_name(item), set()).add(item) + for item in right: + right_norm.setdefault(norm_name(item), set()).add(item) + norm_overlap_keys = set(left_norm) & set(right_norm) + norm_details = [] + for key in sorted(norm_overlap_keys): + for l_item in sorted(left_norm[key]): + for r_item in sorted(right_norm[key]): + norm_details.append( + { + "left_set": left_name, + "right_set": right_name, + "match_type": "exact" if l_item == r_item else "normalized", + "left_name": l_item, + "right_name": r_item, + "normalized": key, + } + ) + summary = { + "left_set": left_name, + "right_set": right_name, + "left_count": len(left), + "right_count": len(right), + "exact_overlap_count": len(exact), + "normalized_overlap_count": len(norm_overlap_keys), + } + return summary, norm_details + + +def write_tsv(path: Path, rows: list[dict[str, object]]) -> None: + path.parent.mkdir(parents=True, exist_ok=True) + if not rows: + path.write_text("", encoding="utf-8") + return + fields = list(rows[0].keys()) + with path.open("w", newline="", encoding="utf-8") as f: + writer = csv.DictWriter(f, fieldnames=fields, delimiter="\t") + writer.writeheader() + writer.writerows(rows) + + +def main() -> None: + args = parse_args() + args.out_dir.mkdir(parents=True, exist_ok=True) + + pkl_root = args.repro_root / "code" / "deit" + sets: dict[str, set[str]] = { + "train_pkl_scaledf_v4_train_fix": load_pkl_list(pkl_root / "scaledf_v4_train_fix.pkl"), + "val_pkl_scaledf_v3_val": load_pkl_list(pkl_root / "scaledf_v3_val.pkl"), + "test_pkl_scaledf_test": load_pkl_list(pkl_root / "scaledf_test.pkl"), + "train_dirs": top_dirs(args.data_root / "ScaleDF" / "train"), + "val_dirs": top_dirs(args.data_root / "ScaleDF" / "val"), + "benchmark_names": top_dirs(args.data_root / "Established_benchmarks_eval_names"), + } + + summaries = [] + details = [] + comparisons = [ + ("train_pkl_scaledf_v4_train_fix", "val_pkl_scaledf_v3_val"), + ("train_pkl_scaledf_v4_train_fix", "test_pkl_scaledf_test"), + ("train_dirs", "val_dirs"), + ("train_dirs", "benchmark_names"), + ("val_dirs", "benchmark_names"), + ] + for left_name, right_name in comparisons: + summary, detail = compare_sets(left_name, sets[left_name], right_name, sets[right_name]) + summaries.append(summary) + details.extend(detail) + + inventory_rows = [ + {"set_name": name, "count": len(values), "examples": ",".join(sorted(values)[:20])} + for name, values in sorted(sets.items()) + ] + write_tsv(args.out_dir / "class_dir_inventory.tsv", inventory_rows) + write_tsv(args.out_dir / "class_dir_overlap_summary.tsv", summaries) + write_tsv(args.out_dir / "class_dir_overlap_details.tsv", details) + + lines = [ + "# ScaleDF Lightweight Overlap Audit", + "", + "This is a top-level path/class-name overlap audit only. It does not prove identity-level non-overlap; that requires face-recognition embeddings or source identity metadata.", + "", + "## Summary", + ] + for row in summaries: + lines.append( + f"- `{row['left_set']}` vs `{row['right_set']}`: " + f"exact={row['exact_overlap_count']}, normalized={row['normalized_overlap_count']} " + f"(counts {row['left_count']} vs {row['right_count']})." + ) + lines += [ + "", + "## Files", + "- `class_dir_inventory.tsv`: set sizes and examples.", + "- `class_dir_overlap_summary.tsv`: pairwise top-level overlap counts.", + "- `class_dir_overlap_details.tsv`: exact/normalized overlap names, if any.", + ] + (args.out_dir / "README.md").write_text("\n".join(lines) + "\n", encoding="utf-8") + print(args.out_dir) + + +if __name__ == "__main__": + main() diff --git a/ScaleDF_rebuttal/scaledf_predict_scores.py b/ScaleDF_rebuttal/scaledf_predict_scores.py new file mode 100644 index 0000000000000000000000000000000000000000..da9087359d7e272fade2a3c71dd196dc40cc1652 --- /dev/null +++ b/ScaleDF_rebuttal/scaledf_predict_scores.py @@ -0,0 +1,424 @@ +#!/usr/bin/env python3 +"""Per-sample ScaleDF prediction export for rebuttal breakdowns.""" + +from __future__ import annotations + +import argparse +import csv +import math +import sys +import time +from collections import defaultdict +from pathlib import Path + +import numpy as np +import torch +from PIL import Image, ImageFile +from torch.utils.data import DataLoader, Dataset +from torchvision import transforms + +ImageFile.LOAD_TRUNCATED_IMAGES = True + +IMAGENET_DEFAULT_MEAN = (0.485, 0.456, 0.406) +IMAGENET_DEFAULT_STD = (0.229, 0.224, 0.225) +IMAGE_EXTS = {".jpg", ".jpeg", ".png", ".bmp", ".webp"} + + +# Table 7 in the NeurIPS submission. The Established_benchmarks/ScaleDF split +# flattens fake methods into one "fake" folder, so we recover method labels +# from filename tokens such as "raid_ScaleDF_v3_val_vqgan_faces_...jpg". +SCALEDF_METHOD_META = { + "3dSwap_faces": ("3DSwap", "FS", "3D"), + "DiffFace_faces": ("DiffFace", "FS", "Diff."), + "G3FA_faces": ("G3FA", "FR", "3D"), + "TPSMM_faces": ("TPSMM", "FR", "Affine"), + "SkyReelsA1_faces": ("SkyReels-A1", "FR", "Diff."), + "vqgan_faces": ("VQGAN", "FF", "GAN"), + "gpt-image-1_faces": ("GPT-Image-1", "FF", "N/A"), + "cogvideox_5b_faces": ("CogVideoX", "FF", "Diff."), + "simplear_faces": ("SimpleAR", "FF", "AR"), + "triplane_faces": ("Triplane", "FE", "GAN"), + "Step1XEdit_faces": ("Step1X-Edit", "FE", "LLM"), + "EDTalk_faces": ("EDTalk", "TF", "3D"), + "KDTalker_faces": ("KDTalker", "TF", "Diff."), + "float_faces": ("FLOAT", "TF", "Flow"), +} + +SCALEDF_REAL_DOMAINS = [ + "000000300VW", + "000000GENKI-4K", + "000000MAFA", + "000000Tufts", + "000000cfp", +] + + +def now() -> str: + return time.strftime("%Y-%m-%dT%H:%M:%SZ", time.gmtime()) + + +def infer_label(class_name: str) -> int: + low = class_name.lower() + if class_name.startswith("000") or low.startswith("real") or low.startswith("0_real") or low == "original": + return 0 + return 1 + + +def arch_group(arch: str) -> str: + if arch == "Diff.": + return "diffusion" + if arch in {"GAN", "AR"}: + return arch + if arch == "real": + return "real" + return "other" + + +def infer_scaledf_method(path: str, class_name: str, label: int) -> tuple[str, str, str, str, str]: + if label == 0: + stem = Path(path).stem + domain = "real" + for token in SCALEDF_REAL_DOMAINS: + if f"_{token}_" in stem: + domain = token + break + return domain, "real", "real", "real", "real" + + if class_name in SCALEDF_METHOD_META: + method_name, category, arch = SCALEDF_METHOD_META[class_name] + return class_name, method_name, category, arch, arch_group(arch) + + stem = Path(path).stem + for token in sorted(SCALEDF_METHOD_META, key=len, reverse=True): + if f"_{token}_" in stem or stem.endswith(f"_{token}"): + method_name, category, arch = SCALEDF_METHOD_META[token] + return token, method_name, category, arch, arch_group(arch) + if class_name != "fake": + token = class_name + else: + token = "unknown" + return token, token, "unknown", "unknown", "unknown" + + +def iter_images(root: Path) -> list[tuple[str, str, int]]: + samples = [] + for class_dir in sorted(p for p in root.iterdir() if p.is_dir()): + label = infer_label(class_dir.name) + for path in sorted(class_dir.rglob("*")): + if path.is_file() and path.suffix.lower() in IMAGE_EXTS: + samples.append((str(path), class_dir.name, label)) + return samples + + +def parse_dataset_spec(spec: str) -> tuple[str, Path]: + if ":" not in spec: + raise ValueError(f"dataset spec must be NAME:ROOT, got {spec}") + name, root = spec.split(":", 1) + return name, Path(root) + + +class PredictionDataset(Dataset): + def __init__(self, samples: list[tuple[str, str, str, int]], image_size: int): + self.samples = samples + self.transform = transforms.Compose( + [ + transforms.Resize(image_size, interpolation=transforms.InterpolationMode.BICUBIC), + transforms.CenterCrop(image_size), + transforms.ToTensor(), + transforms.Normalize(IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STD), + ] + ) + + def __len__(self) -> int: + return len(self.samples) + + def __getitem__(self, index: int): + dataset, path, class_name, label = self.samples[index] + with Image.open(path) as img: + image = self.transform(img.convert("RGB")) + return image, label, index + + +def collate(batch): + images, labels, indices = zip(*batch) + return torch.stack(images, 0), torch.tensor(labels, dtype=torch.long), list(indices) + + +def load_model(args): + sys.path.insert(0, str(args.code_root)) + import models # noqa: F401 + import models_v2 # noqa: F401 + from timm.models import create_model + + model = create_model(args.model, pretrained=False, num_classes=2) + checkpoint = torch.load(args.checkpoint, map_location="cpu", weights_only=False) + state = checkpoint.get("model", checkpoint) + missing, unexpected = model.load_state_dict(state, strict=False) + print(f"{now()} loaded model missing={len(missing)} unexpected={len(unexpected)}", flush=True) + return model + + +def command_predict(args: argparse.Namespace) -> int: + all_samples: list[tuple[str, str, str, int]] = [] + for spec in args.dataset: + name, root = parse_dataset_spec(spec) + samples = iter_images(root) + print(f"{now()} dataset={name} root={root} images={len(samples)}", flush=True) + all_samples.extend((name, path, class_name, label) for path, class_name, label in samples) + + if args.max_images: + all_samples = all_samples[: args.max_images] + shard_samples = all_samples[args.shard_index :: args.num_shards] + print( + f"{now()} shard={args.shard_index}/{args.num_shards} total={len(all_samples)} shard_images={len(shard_samples)}", + flush=True, + ) + + dataset = PredictionDataset(shard_samples, args.input_size) + loader = DataLoader( + dataset, + batch_size=args.batch_size, + shuffle=False, + num_workers=args.workers, + pin_memory=True, + drop_last=False, + collate_fn=collate, + prefetch_factor=args.prefetch_factor if args.workers > 0 else None, + persistent_workers=args.workers > 0, + ) + + device = torch.device(args.device) + model = load_model(args).to(device) + model.eval() + + args.out.parent.mkdir(parents=True, exist_ok=True) + fields = [ + "dataset", + "path", + "class_name", + "label", + "method_key", + "method_name", + "category", + "architecture", + "arch_group", + "logit_real", + "logit_fake", + "prob_fake", + "pred", + ] + done = 0 + start = time.time() + with args.out.open("w", newline="", encoding="utf-8") as f: + writer = csv.DictWriter(f, fieldnames=fields, delimiter="\t") + writer.writeheader() + with torch.inference_mode(): + for step, (images, labels, indices) in enumerate(loader, start=1): + images = images.to(device, non_blocking=True) + with torch.amp.autocast("cuda"): + logits = model(images) + probs = torch.softmax(logits, dim=1) + logits_cpu = logits.float().cpu().numpy() + probs_cpu = probs.float().cpu().numpy() + preds = probs_cpu[:, 1] >= args.threshold + for local_i, sample_index in enumerate(indices): + dataset_name, path, class_name, label = shard_samples[sample_index] + method_key, method_name, category, architecture, group = infer_scaledf_method(path, class_name, label) + writer.writerow( + { + "dataset": dataset_name, + "path": path, + "class_name": class_name, + "label": label, + "method_key": method_key, + "method_name": method_name, + "category": category, + "architecture": architecture, + "arch_group": group, + "logit_real": f"{logits_cpu[local_i, 0]:.8f}", + "logit_fake": f"{logits_cpu[local_i, 1]:.8f}", + "prob_fake": f"{probs_cpu[local_i, 1]:.8f}", + "pred": int(preds[local_i]), + } + ) + done += len(indices) + if step % args.status_every == 0 or done == len(shard_samples): + elapsed = max(time.time() - start, 1e-6) + print(f"{now()} predicted={done}/{len(shard_samples)} rate={done/elapsed:.1f}/s", flush=True) + print(args.out) + return 0 + + +def binary_auc(labels: list[int], scores: list[float]) -> float: + n_pos = sum(labels) + n_neg = len(labels) - n_pos + if n_pos == 0 or n_neg == 0: + return float("nan") + order = sorted(range(len(scores)), key=lambda i: scores[i]) + ranks = [0.0] * len(scores) + i = 0 + while i < len(order): + j = i + 1 + while j < len(order) and scores[order[j]] == scores[order[i]]: + j += 1 + avg_rank = (i + 1 + j) / 2.0 + for k in range(i, j): + ranks[order[k]] = avg_rank + i = j + rank_sum_pos = sum(ranks[i] for i, y in enumerate(labels) if y == 1) + return (rank_sum_pos - n_pos * (n_pos + 1) / 2.0) / (n_pos * n_neg) + + +def eer_percent(labels: list[int], scores: list[float]) -> float: + n_pos = sum(labels) + n_neg = len(labels) - n_pos + if n_pos == 0 or n_neg == 0: + return float("nan") + order = sorted(range(len(scores)), key=lambda i: scores[i], reverse=True) + fp = tp = 0 + best = (1.0, 100.0) + i = 0 + while i < len(order): + score = scores[order[i]] + while i < len(order) and scores[order[i]] == score: + if labels[order[i]] == 1: + tp += 1 + else: + fp += 1 + i += 1 + fpr = fp / n_neg + fnr = (n_pos - tp) / n_pos + gap = abs(fpr - fnr) + if gap < best[0]: + best = (gap, (fpr + fnr) * 50.0) + return best[1] + + +def summarize_group(rows: list[dict[str, str]], group_type: str, dataset: str, group_name: str) -> dict[str, object]: + labels = [int(r["label"]) for r in rows] + scores = [float(r["prob_fake"]) for r in rows] + preds = [int(r["pred"]) for r in rows] + n_real = sum(1 for y in labels if y == 0) + n_fake = sum(1 for y in labels if y == 1) + correct = sum(int(p == y) for p, y in zip(preds, labels)) + real_scores = [s for s, y in zip(scores, labels) if y == 0] + fake_scores = [s for s, y in zip(scores, labels) if y == 1] + return { + "group_type": group_type, + "dataset": dataset, + "group_name": group_name, + "images": len(rows), + "n_real": n_real, + "n_fake": n_fake, + "accuracy": correct / len(rows) if rows else float("nan"), + "auc": binary_auc(labels, scores), + "eer_percent": eer_percent(labels, scores), + "mean_real_score": float(np.mean(real_scores)) if real_scores else float("nan"), + "mean_fake_score": float(np.mean(fake_scores)) if fake_scores else float("nan"), + } + + +def command_summarize(args: argparse.Namespace) -> int: + rows = [] + for path in args.prediction_tsv: + with path.open(newline="", encoding="utf-8") as f: + rows.extend(csv.DictReader(f, delimiter="\t")) + + summaries = [] + by_dataset = defaultdict(list) + for row in rows: + by_dataset[row["dataset"]].append(row) + for dataset, dataset_rows in sorted(by_dataset.items()): + summaries.append(summarize_group(dataset_rows, "dataset", dataset, "ALL")) + real_rows = [r for r in dataset_rows if int(r["label"]) == 0] + fake_rows = [r for r in dataset_rows if int(r["label"]) == 1] + + for method in sorted({r.get("method_name", r["class_name"]) for r in fake_rows}): + group_rows = [r for r in fake_rows if r.get("method_name", r["class_name"]) == method] + summaries.append(summarize_group(real_rows + group_rows, "fake_method_vs_real", dataset, method)) + + for category in sorted({r.get("category", "unknown") for r in fake_rows}): + group_rows = [r for r in fake_rows if r.get("category", "unknown") == category] + summaries.append(summarize_group(real_rows + group_rows, "forgery_category_vs_real", dataset, category)) + + for architecture in sorted({r.get("architecture", "unknown") for r in fake_rows}): + group_rows = [r for r in fake_rows if r.get("architecture", "unknown") == architecture] + summaries.append(summarize_group(real_rows + group_rows, "architecture_vs_real", dataset, architecture)) + + for group in sorted({r.get("arch_group", "unknown") for r in fake_rows}): + group_rows = [r for r in fake_rows if r.get("arch_group", "unknown") == group] + summaries.append(summarize_group(real_rows + group_rows, "architecture_group_vs_real", dataset, group)) + + category_arch = sorted({(r.get("category", "unknown"), r.get("arch_group", "unknown")) for r in fake_rows}) + for category, group in category_arch: + group_rows = [ + r + for r in fake_rows + if r.get("category", "unknown") == category and r.get("arch_group", "unknown") == group + ] + summaries.append(summarize_group(real_rows + group_rows, "category_arch_group_vs_real", dataset, f"{category}:{group}")) + + args.out.parent.mkdir(parents=True, exist_ok=True) + fields = [ + "group_type", + "dataset", + "group_name", + "images", + "n_real", + "n_fake", + "accuracy", + "auc", + "eer_percent", + "mean_real_score", + "mean_fake_score", + ] + with args.out.open("w", newline="", encoding="utf-8") as f: + writer = csv.DictWriter(f, fieldnames=fields, delimiter="\t") + writer.writeheader() + for row in summaries: + formatted = row.copy() + for key in ["accuracy", "auc", "eer_percent", "mean_real_score", "mean_fake_score"]: + val = formatted[key] + formatted[key] = "nan" if not math.isfinite(float(val)) else f"{float(val):.8f}" + writer.writerow(formatted) + print(f"rows={len(summaries)}") + print(args.out) + return 0 + + +def build_parser() -> argparse.ArgumentParser: + parser = argparse.ArgumentParser(description=__doc__) + sub = parser.add_subparsers(dest="command", required=True) + + p = sub.add_parser("predict") + p.add_argument("--dataset", action="append", required=True, help="NAME:ROOT") + p.add_argument("--checkpoint", type=Path, required=True) + p.add_argument("--model", default="deit_base_patch16_LS") + p.add_argument("--code-root", type=Path, default=Path("/tmp/ScaleDF_repro_20260724/code/deit")) + p.add_argument("--out", type=Path, required=True) + p.add_argument("--device", default="cuda:0") + p.add_argument("--input-size", type=int, default=224) + p.add_argument("--batch-size", type=int, default=384) + p.add_argument("--workers", type=int, default=12) + p.add_argument("--prefetch-factor", type=int, default=4) + p.add_argument("--threshold", type=float, default=0.5) + p.add_argument("--shard-index", type=int, default=0) + p.add_argument("--num-shards", type=int, default=1) + p.add_argument("--max-images", type=int) + p.add_argument("--status-every", type=int, default=20) + p.set_defaults(func=command_predict) + + p = sub.add_parser("summarize") + p.add_argument("--prediction-tsv", type=Path, action="append", required=True) + p.add_argument("--out", type=Path, required=True) + p.set_defaults(func=command_summarize) + return parser + + +def main() -> int: + args = build_parser().parse_args() + return args.func(args) + + +if __name__ == "__main__": + raise SystemExit(main()) diff --git a/ScaleDF_rebuttal/summarize_fixed_budget_eval.py b/ScaleDF_rebuttal/summarize_fixed_budget_eval.py new file mode 100644 index 0000000000000000000000000000000000000000..485a28d48f096b1edd5797bf42e873250e68404e --- /dev/null +++ b/ScaleDF_rebuttal/summarize_fixed_budget_eval.py @@ -0,0 +1,287 @@ +#!/usr/bin/env python3 +"""Summarize fixed-budget ScaleDF benchmark evals for rebuttal plots/tables.""" + +from __future__ import annotations + +import argparse +import csv +import math +import re +import statistics +from collections import defaultdict +from pathlib import Path + + +TAG_RE = re.compile(r"^fixed500k_(?Preal|fake)_n(?P\d+)_seed(?P\d+)(?:_.*)?$") +METRICS = ("acc1", "roc_auc", "pr_auc", "eer_percent") + + +def to_float(value: str) -> float: + try: + return float(value) + except (TypeError, ValueError): + return math.nan + + +def metric_error(metric: str, value: float) -> float: + if math.isnan(value): + return math.nan + if metric in {"roc_auc", "pr_auc"}: + return max(1.0 - value, 1e-8) + if metric == "acc1": + return max(1.0 - value / 100.0, 1e-8) + if metric == "eer_percent": + return max(value / 100.0, 1e-8) + raise ValueError(metric) + + +def mean(values: list[float]) -> float: + vals = [v for v in values if not math.isnan(v)] + return statistics.fmean(vals) if vals else math.nan + + +def std(values: list[float]) -> float: + vals = [v for v in values if not math.isnan(v)] + if len(vals) < 2: + return 0.0 if vals else math.nan + return statistics.stdev(vals) + + +def linreg(xs: list[float], ys: list[float]) -> dict[str, float]: + n = len(xs) + if n < 2: + return {"alpha_ols": math.nan, "intercept": math.nan, "r2": math.nan} + xbar = statistics.fmean(xs) + ybar = statistics.fmean(ys) + sxx = sum((x - xbar) ** 2 for x in xs) + if sxx <= 0: + return {"alpha_ols": math.nan, "intercept": math.nan, "r2": math.nan} + slope = sum((x - xbar) * (y - ybar) for x, y in zip(xs, ys)) / sxx + intercept = ybar - slope * xbar + sst = sum((y - ybar) ** 2 for y in ys) + sse = sum((y - (intercept + slope * x)) ** 2 for x, y in zip(xs, ys)) + r2 = 1.0 - sse / sst if sst > 0 else math.nan + return {"alpha_ols": -slope, "intercept": intercept, "r2": r2} + + +def theil_sen(xs: list[float], ys: list[float]) -> float: + slopes = [] + for i in range(len(xs)): + for j in range(i + 1, len(xs)): + dx = xs[j] - xs[i] + if abs(dx) > 1e-12: + slopes.append((ys[j] - ys[i]) / dx) + return -statistics.median(slopes) if slopes else math.nan + + +def read_eval(path: Path) -> list[dict[str, object]]: + rows: list[dict[str, object]] = [] + with path.open(newline="", encoding="utf-8") as f: + for row in csv.DictReader(f, delimiter="\t"): + m = TAG_RE.match(row.get("tag", "")) + if not m: + continue + axis = "real_domains" if m.group("axis") == "real" else "fake_methods" + out: dict[str, object] = { + "axis": axis, + "level": int(m.group("level")), + "seed": int(m.group("seed")), + "tag": row["tag"], + "dataset": row.get("dataset", ""), + "test_images": int(float(row.get("test_images") or 0)), + "checkpoint": row.get("checkpoint", ""), + } + for metric in METRICS: + value = to_float(row.get(metric, "")) + out[metric] = value + out[f"{metric}_error"] = metric_error(metric, value) + rows.append(out) + return rows + + +def write_tsv(path: Path, rows: list[dict[str, object]], fields: list[str]) -> None: + path.parent.mkdir(parents=True, exist_ok=True) + with path.open("w", newline="", encoding="utf-8") as f: + writer = csv.DictWriter(f, fieldnames=fields, delimiter="\t", extrasaction="ignore") + writer.writeheader() + writer.writerows(rows) + + +def build_metric_points(rows: list[dict[str, object]]) -> list[dict[str, object]]: + points = [] + for row in rows: + for metric in METRICS: + points.append( + { + "axis": row["axis"], + "level": row["level"], + "seed": row["seed"], + "tag": row["tag"], + "dataset": row["dataset"], + "metric": metric, + "value": row[metric], + "error": row[f"{metric}_error"], + "test_images": row["test_images"], + "checkpoint": row["checkpoint"], + } + ) + return points + + +def aggregate_points(points: list[dict[str, object]]) -> list[dict[str, object]]: + groups: dict[tuple[object, ...], list[dict[str, object]]] = defaultdict(list) + for point in points: + groups[(point["axis"], point["level"], point["metric"])].append(point) + + rows = [] + for (axis, level, metric), group in sorted(groups.items()): + values = [float(g["value"]) for g in group] + errors = [float(g["error"]) for g in group] + weights = [int(g["test_images"]) for g in group] + weighted_value = math.nan + weighted_error = math.nan + if sum(weights) > 0: + weighted_value = sum(v * w for v, w in zip(values, weights) if not math.isnan(v)) / sum(weights) + weighted_error = sum(e * w for e, w in zip(errors, weights) if not math.isnan(e)) / sum(weights) + rows.append( + { + "axis": axis, + "level": level, + "dataset": "ALL_BENCHMARKS_MEAN", + "metric": metric, + "n_datasets": len(group), + "value_mean": mean(values), + "value_std": std(values), + "error_mean": mean(errors), + "error_std": std(errors), + "value_weighted_by_images": weighted_value, + "error_weighted_by_images": weighted_error, + "total_test_images": sum(weights), + } + ) + return rows + + +def fit_rows(points: list[dict[str, object]], aggregates: list[dict[str, object]]) -> list[dict[str, object]]: + fit_input: dict[tuple[str, str, str], list[tuple[int, float]]] = defaultdict(list) + for point in points: + fit_input[(str(point["axis"]), str(point["dataset"]), str(point["metric"]))].append( + (int(point["level"]), float(point["error"])) + ) + for row in aggregates: + fit_input[(str(row["axis"]), str(row["dataset"]), str(row["metric"]))].append( + (int(row["level"]), float(row["error_mean"])) + ) + fit_input[(str(row["axis"]), "ALL_BENCHMARKS_WEIGHTED", str(row["metric"]))].append( + (int(row["level"]), float(row["error_weighted_by_images"])) + ) + + rows = [] + for (axis, dataset, metric), values in sorted(fit_input.items()): + dedup = {} + for level, err in values: + if not math.isnan(err) and err > 0: + dedup[level] = err + if len(dedup) < 2: + continue + levels = sorted(dedup) + xs = [math.log(x) for x in levels] + ys = [math.log(dedup[x]) for x in levels] + reg = linreg(xs, ys) + rows.append( + { + "axis": axis, + "dataset": dataset, + "metric": metric, + "n_levels": len(levels), + "levels": ",".join(str(x) for x in levels), + "alpha_ols": reg["alpha_ols"], + "alpha_theilsen": theil_sen(xs, ys), + "r2": reg["r2"], + "response": "eer" if metric == "eer_percent" else f"1-{metric}", + } + ) + return rows + + +def main() -> None: + parser = argparse.ArgumentParser() + parser.add_argument("--eval-tsv", type=Path, required=True) + parser.add_argument("--out-dir", type=Path, required=True) + args = parser.parse_args() + + rows = read_eval(args.eval_tsv) + points = build_metric_points(rows) + aggregates = aggregate_points(points) + fits = fit_rows(points, aggregates) + + write_tsv( + args.out_dir / "fixed_budget_eval_rows.tsv", + rows, + [ + "axis", + "level", + "seed", + "tag", + "dataset", + "test_images", + "acc1", + "roc_auc", + "pr_auc", + "eer_percent", + "checkpoint", + ], + ) + write_tsv( + args.out_dir / "fixed_budget_eval_metric_points.tsv", + points, + ["axis", "level", "seed", "tag", "dataset", "metric", "value", "error", "test_images", "checkpoint"], + ) + write_tsv( + args.out_dir / "fixed_budget_eval_level_summary.tsv", + aggregates, + [ + "axis", + "level", + "dataset", + "metric", + "n_datasets", + "value_mean", + "value_std", + "error_mean", + "error_std", + "value_weighted_by_images", + "error_weighted_by_images", + "total_test_images", + ], + ) + write_tsv( + args.out_dir / "fixed_budget_eval_powerlaw_fit.tsv", + fits, + ["axis", "dataset", "metric", "n_levels", "levels", "alpha_ols", "alpha_theilsen", "r2", "response"], + ) + + readme = [ + "# Fixed-Budget Benchmark Eval Summary", + "", + f"Input: `{args.eval_tsv}`", + "", + "The fixed-budget condition uses 500k real and 500k fake training images for every level.", + "Fits use log(error) vs log(level), where error is `1-AUC`, `1-Acc`, or EER.", + "", + "Outputs:", + "- `fixed_budget_eval_rows.tsv`: one row per checkpoint and benchmark.", + "- `fixed_budget_eval_metric_points.tsv`: long metric table.", + "- `fixed_budget_eval_level_summary.tsv`: unweighted and image-weighted benchmark means by level.", + "- `fixed_budget_eval_powerlaw_fit.tsv`: OLS and Theil-Sen exponents.", + ] + args.out_dir.mkdir(parents=True, exist_ok=True) + (args.out_dir / "README.md").write_text("\n".join(readme) + "\n", encoding="utf-8") + print(f"rows={len(rows)}") + print(f"points={len(points)}") + print(f"fits={len(fits)}") + print(args.out_dir) + + +if __name__ == "__main__": + main() diff --git a/ScaleDF_rebuttal/summarize_fixed_budget_runs.py b/ScaleDF_rebuttal/summarize_fixed_budget_runs.py new file mode 100644 index 0000000000000000000000000000000000000000..daff32544f9867dc54b58701e13b03cbac4d18e5 --- /dev/null +++ b/ScaleDF_rebuttal/summarize_fixed_budget_runs.py @@ -0,0 +1,228 @@ +#!/usr/bin/env python3 +"""Summarize fixed-image-budget ScaleDF rebuttal runs.""" + +from __future__ import annotations + +import argparse +import json +import math +import re +from pathlib import Path + +import numpy as np +import pandas as pd + + +RUN_RE = re.compile( + r"^(?:control_)?fixed(?P\d+)_(?Pfake_methods|real_domains)" + r"(?:_fs)?_n(?P\d+)_classseed(?P\d+)_" +) + + +def parse_args() -> argparse.Namespace: + parser = argparse.ArgumentParser() + parser.add_argument( + "--runs-root", + type=Path, + default=Path("/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal/runs"), + ) + parser.add_argument("--out-dir", type=Path, required=True) + parser.add_argument("--per-side-images", type=int, default=None) + parser.add_argument("--name-regex", default=None) + return parser.parse_args() + + +def load_jsonl(path: Path) -> list[dict[str, object]]: + rows: list[dict[str, object]] = [] + if not path.is_file(): + return rows + with path.open("r", encoding="utf-8", errors="replace") as f: + for line in f: + line = line.strip() + if not line or not line.startswith("{"): + continue + try: + rows.append(json.loads(line)) + except json.JSONDecodeError: + continue + return rows + + +def parse_train_header(path: Path) -> dict[str, str]: + out: dict[str, str] = {} + if not path.is_file(): + return out + with path.open("r", encoding="utf-8", errors="replace") as f: + for line in f: + text = line.rstrip("\n") + if not text or "=" not in text: + continue + key, value = text.split("=", 1) + if key in { + "started_utc", + "hostname", + "package_root", + "data_root", + "kind", + "selection", + "model_name", + "output_root", + "main", + "nproc_per_node", + "batch_per_gpu", + }: + out[key] = value + if len(out) >= 11: + break + return out + + +def summarize_run(run_dir: Path) -> dict[str, object] | None: + match = RUN_RE.match(run_dir.name) + if not match: + return None + eval_rows = load_jsonl(run_dir / "log.txt") + if not eval_rows: + return None + + frame = pd.DataFrame(eval_rows) + frame["epoch"] = pd.to_numeric(frame["epoch"], errors="coerce") + frame["test_acc1"] = pd.to_numeric(frame["test_acc1"], errors="coerce") + frame = frame[frame["test_acc1"].notna()].copy() + if frame.empty: + return None + + best = frame.loc[frame["test_acc1"].idxmax()].to_dict() + last = frame.sort_values("epoch").iloc[-1].to_dict() + header = parse_train_header(run_dir / "train.log") + + row: dict[str, object] = { + "run_name": run_dir.name, + "run_dir": str(run_dir), + "family": match.group("family"), + "level": int(match.group("level")), + "class_seed": int(match.group("class_seed")), + "per_side_images": int(match.group("per_side")), + "total_train_images": int(match.group("per_side")) * 2, + "eval_points": int(len(frame)), + "best_epoch": int(best.get("epoch", -1)), + "best_acc1": float(best.get("test_acc1", math.nan)), + "best_acc1_real": float(best.get("test_acc1_pos", math.nan)), + "best_acc1_fake": float(best.get("test_acc1_neg", math.nan)), + "best_loss": float(best.get("test_loss", math.nan)), + "last_epoch": int(last.get("epoch", -1)), + "last_acc1": float(last.get("test_acc1", math.nan)), + "last_acc1_real": float(last.get("test_acc1_pos", math.nan)), + "last_acc1_fake": float(last.get("test_acc1_neg", math.nan)), + "last_loss": float(last.get("test_loss", math.nan)), + } + row.update({f"header_{k}": v for k, v in header.items()}) + return row + + +def level_summary(run_df: pd.DataFrame, metric: str) -> pd.DataFrame: + rows = [] + for keys, group in run_df.groupby(["family", "level"], sort=True): + values = pd.to_numeric(group[metric], errors="coerce").dropna() + rows.append( + { + "family": keys[0], + "level": int(keys[1]), + "metric": metric, + "runs": int(len(values)), + "mean": float(values.mean()) if len(values) else math.nan, + "std": float(values.std(ddof=1)) if len(values) > 1 else 0.0, + "min": float(values.min()) if len(values) else math.nan, + "max": float(values.max()) if len(values) else math.nan, + } + ) + return pd.DataFrame(rows) + + +def fit_power(run_df: pd.DataFrame, metric: str) -> pd.DataFrame: + rows = [] + grouped = run_df.groupby(["family", "level"], sort=True)[metric].mean().reset_index() + for family, group in grouped.groupby("family", sort=True): + levels = pd.to_numeric(group["level"], errors="coerce").to_numpy(dtype=float) + acc = pd.to_numeric(group[metric], errors="coerce").to_numpy(dtype=float) / 100.0 + err = np.clip(1.0 - acc, 1e-8, None) + valid = np.isfinite(levels) & (levels > 0) & np.isfinite(err) & (err > 0) + levels = levels[valid] + err = err[valid] + if len(levels) < 3: + alpha = r2 = math.nan + else: + slope, intercept = np.polyfit(np.log(levels), np.log(err), 1) + pred = slope * np.log(levels) + intercept + denom = float(np.sum((np.log(err) - np.log(err).mean()) ** 2)) + alpha = float(-slope) + r2 = float(1.0 - np.sum((np.log(err) - pred) ** 2) / denom) if denom > 0 else math.nan + rows.append( + { + "family": family, + "metric": metric, + "response": "1-acc", + "levels": ",".join(f"{x:g}" for x in levels), + "alpha_ols": alpha, + "r2_ols": r2, + } + ) + return pd.DataFrame(rows) + + +def main() -> None: + args = parse_args() + args.out_dir.mkdir(parents=True, exist_ok=True) + name_re = re.compile(args.name_regex) if args.name_regex else None + + rows = [] + for run_dir in sorted(args.runs_root.iterdir()): + if not run_dir.is_dir(): + continue + if name_re and not name_re.search(run_dir.name): + continue + row = summarize_run(run_dir) + if row is not None: + if args.per_side_images is not None and row["per_side_images"] != args.per_side_images: + continue + rows.append(row) + if not rows: + raise SystemExit(f"No fixed-budget runs found under {args.runs_root}") + + run_df = pd.DataFrame(rows).sort_values(["family", "level", "class_seed"]) + metrics = [ + "best_acc1", + "best_acc1_real", + "best_acc1_fake", + "last_acc1", + "last_acc1_real", + "last_acc1_fake", + ] + level_df = pd.concat([level_summary(run_df, metric) for metric in metrics], ignore_index=True) + fit_df = pd.concat([fit_power(run_df, metric) for metric in metrics], ignore_index=True) + + run_df.to_csv(args.out_dir / "fixed_budget_run_summary.tsv", sep="\t", index=False) + level_df.to_csv(args.out_dir / "fixed_budget_level_summary.tsv", sep="\t", index=False) + fit_df.to_csv(args.out_dir / "fixed_budget_powerlaw_fit.tsv", sep="\t", index=False) + (args.out_dir / "README.md").write_text( + "\n".join( + [ + "# Fixed-Budget Rebuttal Runs", + "", + "All included runs use a fixed number of real and fake training images; the swept level is the number of real domains or fake methods.", + "The main table reports ScaleDF validation accuracy from each run's `log.txt`.", + "", + "- `fixed_budget_run_summary.tsv`: one row per run.", + "- `fixed_budget_level_summary.tsv`: mean/std across class seeds.", + "- `fixed_budget_powerlaw_fit.tsv`: log-log fit of `1-accuracy` against level.", + "", + ] + ), + encoding="utf-8", + ) + print(args.out_dir) + print(f"runs={len(run_df)}") + + +if __name__ == "__main__": + main() diff --git a/ScaleDF_rebuttal/torch_distributed_smoke.py b/ScaleDF_rebuttal/torch_distributed_smoke.py new file mode 100644 index 0000000000000000000000000000000000000000..46ff9239b8fe5a92ad05ea44c8a07e1c25d4cf76 --- /dev/null +++ b/ScaleDF_rebuttal/torch_distributed_smoke.py @@ -0,0 +1,34 @@ +#!/usr/bin/env python3 +"""Tiny CUDA distributed smoke test for multi-node torchrun.""" + +from __future__ import annotations + +import os +import socket + +import torch +import torch.distributed as dist + + +def main() -> None: + dist.init_process_group(backend="nccl") + rank = dist.get_rank() + world_size = dist.get_world_size() + local_rank = int(os.environ["LOCAL_RANK"]) + torch.cuda.set_device(local_rank) + value = torch.tensor([rank + 1.0], device=f"cuda:{local_rank}") + dist.all_reduce(value, op=dist.ReduceOp.SUM) + expected = world_size * (world_size + 1) / 2 + if rank == 0: + print( + f"smoke_ok host={socket.gethostname()} world_size={world_size} " + f"sum={value.item():.1f} expected={expected:.1f}", + flush=True, + ) + assert abs(value.item() - expected) < 1e-3 + dist.barrier() + dist.destroy_process_group() + + +if __name__ == "__main__": + main() diff --git a/ScaleDF_rebuttal/wait_all46_anchor_train_complete_then_eval.sh b/ScaleDF_rebuttal/wait_all46_anchor_train_complete_then_eval.sh new file mode 100644 index 0000000000000000000000000000000000000000..07bb0ebffcceb096e0ae6d2112f25151ef97f6c3 --- /dev/null +++ b/ScaleDF_rebuttal/wait_all46_anchor_train_complete_then_eval.sh @@ -0,0 +1,68 @@ +#!/usr/bin/env bash +set -euo pipefail + +PERSIST_ROOT="${PERSIST_ROOT:-/mnt/bn/search-aipf-haoyu/wenhaowang/ScaleDF_rebuttal}" +PACKAGE_ROOT="${PACKAGE_ROOT:-/tmp/ScaleDF_repro_20260724}" +SCALEDF_DATA="${SCALEDF_DATA:-/dev/shm/scaledf_bind_storage/ScaleDF}" +BENCHMARK_ROOT="${BENCHMARK_ROOT:-/dev/shm/scaledf_bind_storage/Established_benchmarks_eval_names}" +RUN_TAG="${RUN_TAG:-20260725_1215}" +RUN_DIR="$PERSIST_ROOT/runs/control_fixed500000_all_46real_91fake_imgseed20260725_n116-042-186_${RUN_TAG}" +TRAIN_LOG="$PERSIST_ROOT/_host_logs/all_46real_91fake_11451_${RUN_TAG}/train_all_46real_91fake_retry1.log" +WAIT_LOG="$PERSIST_ROOT/_host_logs/all_46real_91fake_11451_${RUN_TAG}/eval_after_train_complete.log" +CHECKPOINT="$RUN_DIR/best_checkpoint.pth" +EVAL_ROOT="$PERSIST_ROOT/eval/fixed500k_all_46real_91fake_benchmark_eval_${RUN_TAG}/fixed500k_all_46real_91fake_seed0" + +export PERSIST_ROOT PACKAGE_ROOT SCALEDF_DATA BENCHMARK_ROOT +export SKIP_MISSING_BENCHMARKS="${SKIP_MISSING_BENCHMARKS:-1}" +export BATCH_PER_GPU="${BATCH_PER_GPU:-256}" +export NPROC_PER_NODE="${NPROC_PER_NODE:-8}" +export PYTHONUNBUFFERED="${PYTHONUNBUFFERED:-1}" + +log() { + echo "[$(date -u +%Y-%m-%dT%H:%M:%SZ)] $*" | tee -a "$WAIT_LOG" +} + +train_alive() { + pgrep -af 'control_fixed500000_all_46real_91fake|main_cross_2direct_balance_realscale_perturb_v4_NRealFake.py' \ + | grep -v pgrep >/dev/null +} + +eval_alive() { + pgrep -af "fixed500k_all_46real_91fake_benchmark_eval_${RUN_TAG}|launch_rebuttal_eval_flex.sh" \ + | grep -v pgrep >/dev/null +} + +log "waiter_started run_dir=$RUN_DIR" + +while true; do + if grep -E 'Traceback|RuntimeError|CUDA out of memory|ChildFailedError|No space left' "$TRAIN_LOG" >/tmp/all46_anchor_train_errors.$$ 2>/dev/null; then + cat /tmp/all46_anchor_train_errors.$$ >> "$WAIT_LOG" + rm -f /tmp/all46_anchor_train_errors.$$ + log "train_error_detected; not starting eval" + exit 20 + fi + rm -f /tmp/all46_anchor_train_errors.$$ + + if grep -q 'Training time' "$TRAIN_LOG" 2>/dev/null && ! train_alive; then + break + fi + + last="$(grep -E 'Epoch: \[[0-9]+\]|Averaged stats|Training time' "$TRAIN_LOG" 2>/dev/null | tail -n 1 || true)" + log "waiting_train_complete last=${last:-none}" + sleep 60 +done + +if [[ ! -f "$CHECKPOINT" ]]; then + log "missing_checkpoint=$CHECKPOINT" + exit 21 +fi + +if eval_alive; then + log "eval_already_running; leaving it alone" + exit 0 +fi + +log "starting_eval checkpoint=$CHECKPOINT output_root=$EVAL_ROOT" +OUTPUT_ROOT="$EVAL_ROOT" RUN_NAME=fixed500k_all_46real_91fake_seed0 \ + "$PERSIST_ROOT/launch_rebuttal_eval_flex.sh" "$CHECKPOINT" deit_base_patch16_LS benchmarks +log "eval_done output_root=$EVAL_ROOT"